Import Mesa 18.3.2

307 files changed, 44062 insertions, 6306 deletions

diff --git a/lib/mesa/bin/git_sha1_gen.py b/lib/mesa/bin/git_sha1_gen.py
index c75dba101..c6fbf4903 100755
--- a/lib/mesa/bin/git_sha1_gen.py
+++ b/lib/mesa/bin/git_sha1_gen.py
@@ -1,11 +1,10 @@
-#!/usr/bin/env python
-
 """
 Generate the contents of the git_sha1.h file.
 The output of this script goes to stdout.
 """
 
 
+import argparse
 import os
 import os.path
 import subprocess
@@ -27,10 +26,25 @@ def get_git_sha1():
         git_sha1 = ''
     return git_sha1
 
+def write_if_different(contents):
+    """
+    Avoid touching the output file if it doesn't need modifications
+    Useful to avoid triggering rebuilds when nothing has changed.
+    """
+    if os.path.isfile(args.output):
+        with open(args.output, 'r') as file:
+            if file.read() == contents:
+                return
+    with open(args.output, 'w') as file:
+        file.write(contents)
+
+parser = argparse.ArgumentParser()
+parser.add_argument('--output', help='File to write the #define in',
+                    required=True)
+args = parser.parse_args()
 
 git_sha1 = os.environ.get('MESA_GIT_SHA1_OVERRIDE', get_git_sha1())[:10]
 if git_sha1:
-    git_sha1_h_in_path = os.path.join(os.path.dirname(sys.argv[0]),
-            '..', 'src', 'git_sha1.h.in')
-    with open(git_sha1_h_in_path , 'r') as git_sha1_h_in:
-        sys.stdout.write(git_sha1_h_in.read().replace('@VCS_TAG@', git_sha1))
+    write_if_different('#define MESA_GIT_SHA1 " (git-' + git_sha1 + ')"')
+else:
+    write_if_different('#define MESA_GIT_SHA1 ""')
diff --git a/lib/mesa/bin/install_megadrivers.py b/lib/mesa/bin/install_megadrivers.py
new file mode 100644
index 000000000..d29b19112
--- /dev/null
+++ b/lib/mesa/bin/install_megadrivers.py
@@ -0,0 +1,74 @@
+# encoding=utf-8
+# Copyright © 2017-2018 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+"""Script to install megadriver symlinks for meson."""
+
+from __future__ import print_function
+import argparse
+import os
+import shutil
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument('megadriver')
+    parser.add_argument('libdir')
+    parser.add_argument('drivers', nargs='+')
+    args = parser.parse_args()
+
+    if os.path.isabs(args.libdir):
+        to = os.path.join(os.environ.get('DESTDIR', '/'), args.libdir[1:])
+    else:
+        to = os.path.join(os.environ['MESON_INSTALL_DESTDIR_PREFIX'], args.libdir)
+
+    master = os.path.join(to, os.path.basename(args.megadriver))
+
+    if not os.path.exists(to):
+        if os.path.lexists(to):
+            os.unlink(to)
+        os.makedirs(to)
+    shutil.copy(args.megadriver, master)
+
+    for driver in args.drivers:
+        abs_driver = os.path.join(to, driver)
+
+        if os.path.lexists(abs_driver):
+            os.unlink(abs_driver)
+        print('installing {} to {}'.format(args.megadriver, abs_driver))
+        os.link(master, abs_driver)
+
+        try:
+            ret = os.getcwd()
+            os.chdir(to)
+
+            name, ext = os.path.splitext(driver)
+            while ext != '.so':
+                if os.path.lexists(name):
+                    os.unlink(name)
+                os.symlink(driver, name)
+                name, ext = os.path.splitext(name)
+        finally:
+            os.chdir(ret)
+    os.unlink(master)
+
+
+if __name__ == '__main__':
+    main()
diff --git a/lib/mesa/bin/meson.build b/lib/mesa/bin/meson.build
new file mode 100644
index 000000000..b8b44baf7
--- /dev/null
+++ b/lib/mesa/bin/meson.build
@@ -0,0 +1,21 @@
+# Copyright © 2017 Eric Engestrom
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+git_sha1_gen_py = files('git_sha1_gen.py')
diff --git a/lib/mesa/bin/meson_get_version.py b/lib/mesa/bin/meson_get_version.py
new file mode 100644
index 000000000..a221e26f2
--- /dev/null
+++ b/lib/mesa/bin/meson_get_version.py
@@ -0,0 +1,35 @@
+#!/usr/bin/env python
+# encoding=utf-8
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+from __future__ import print_function
+import os
+
+
+def main():
+    filename = os.path.join(os.environ['MESON_SOURCE_ROOT'], 'VERSION')
+    with open(filename) as f:
+        version = f.read().strip()
+    print(version, end='')
+
+
+if __name__ == '__main__':
+    main()
diff --git a/lib/mesa/meson.build b/lib/mesa/meson.build
new file mode 100644
index 000000000..5a20e1ea3
--- /dev/null
+++ b/lib/mesa/meson.build
@@ -0,0 +1,1468 @@
+# Copyright © 2017-2018 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+project(
+  'mesa',
+  ['c', 'cpp'],
+  version : run_command(
+    [find_program('python', 'python2', 'python3'), 'bin/meson_get_version.py']
+  ).stdout(),
+  license : 'MIT',
+  meson_version : '>= 0.45',
+  default_options : ['buildtype=debugoptimized', 'b_ndebug=if-release', 'c_std=c99', 'cpp_std=c++11']
+)
+
+cc = meson.get_compiler('c')
+cpp = meson.get_compiler('cpp')
+
+null_dep = dependency('', required : false)
+
+system_has_kms_drm = ['openbsd', 'netbsd', 'freebsd', 'dragonfly', 'linux'].contains(host_machine.system())
+
+# Arguments for the preprocessor, put these in a separate array from the C and
+# C++ (cpp in meson terminology) arguments since they need to be added to the
+# default arguments for both C and C++.
+pre_args = [
+  '-D__STDC_CONSTANT_MACROS',
+  '-D__STDC_FORMAT_MACROS',
+  '-D__STDC_LIMIT_MACROS',
+  '-DVERSION="@0@"'.format(meson.project_version()),
+  '-DPACKAGE_VERSION=VERSION',
+  '-DPACKAGE_BUGREPORT="https://bugs.freedesktop.org/enter_bug.cgi?product=Mesa"',
+]
+
+with_vulkan_icd_dir = get_option('vulkan-icd-dir')
+with_tests = get_option('build-tests')
+with_valgrind = get_option('valgrind')
+with_libunwind = get_option('libunwind')
+with_asm = get_option('asm')
+with_glx_read_only_text = get_option('glx-read-only-text')
+with_glx_direct = get_option('glx-direct')
+with_osmesa = get_option('osmesa')
+with_swr_arches = get_option('swr-arches')
+with_tools = get_option('tools')
+if with_tools.contains('all')
+  with_tools = ['freedreno', 'glsl', 'intel', 'nir', 'nouveau', 'xvmc']
+endif
+
+dri_drivers_path = get_option('dri-drivers-path')
+if dri_drivers_path == ''
+  dri_drivers_path = join_paths(get_option('libdir'), 'dri')
+endif
+dri_search_path = get_option('dri-search-path')
+if dri_search_path == ''
+  dri_search_path = join_paths(get_option('prefix'), dri_drivers_path)
+endif
+
+with_gles1 = get_option('gles1')
+with_gles2 = get_option('gles2')
+with_opengl = get_option('opengl')
+with_any_opengl = with_opengl or with_gles1 or with_gles2
+# Only build shared_glapi if at least one OpenGL API is enabled
+with_shared_glapi = get_option('shared-glapi') and with_any_opengl
+
+
+# shared-glapi is required if at least two OpenGL APIs are being built
+if not with_shared_glapi
+  if ((with_gles1 and with_gles2) or (with_gles1 and with_opengl)
+      or (with_gles2 and with_opengl))
+    error('shared-glapi required for building two or more of OpenGL, OpenGL ES 1.x, OpenGL ES 2.x')
+  endif
+endif
+
+# We require OpenGL for OpenGL ES
+if (with_gles1 or with_gles2) and not with_opengl
+  error('building OpenGL ES without OpenGL is not supported.')
+endif
+
+system_has_kms_drm = ['openbsd', 'netbsd', 'freebsd', 'dragonfly', 'linux'].contains(host_machine.system())
+
+_drivers = get_option('dri-drivers')
+if _drivers.contains('auto')
+  if system_has_kms_drm
+    # TODO: PPC, Sparc
+    if ['x86', 'x86_64'].contains(host_machine.cpu_family())
+      _drivers = ['i915', 'i965', 'r100', 'r200', 'nouveau']
+    elif ['arm', 'aarch64'].contains(host_machine.cpu_family())
+      _drivers = []
+    else
+      error('Unknown architecture @0@. Please pass -Ddri-drivers to set driver options. Patches gladly accepted to fix this.'.format(
+            host_machine.cpu_family()))
+    endif
+  elif ['darwin', 'windows', 'cygwin', 'haiku'].contains(host_machine.system())
+    # only swrast would make sense here, but gallium swrast is a much better default
+    _drivers = []
+  else
+    error('Unknown OS @0@. Please pass -Ddri-drivers to set driver options. Patches gladly accepted to fix this.'.format(
+          host_machine.system()))
+  endif
+endif
+
+with_dri_i915 = _drivers.contains('i915')
+with_dri_i965 = _drivers.contains('i965')
+with_dri_r100 = _drivers.contains('r100')
+with_dri_r200 = _drivers.contains('r200')
+with_dri_nouveau = _drivers.contains('nouveau')
+with_dri_swrast = _drivers.contains('swrast')
+
+with_dri = _drivers.length() != 0 and _drivers != ['']
+
+_drivers = get_option('gallium-drivers')
+if _drivers.contains('auto')
+  if system_has_kms_drm
+    # TODO: PPC, Sparc
+    if ['x86', 'x86_64'].contains(host_machine.cpu_family())
+      _drivers = [
+        'r300', 'r600', 'radeonsi', 'nouveau', 'virgl', 'svga', 'swrast'
+      ]
+    elif ['arm', 'aarch64'].contains(host_machine.cpu_family())
+      _drivers = [
+        'pl111', 'v3d', 'vc4', 'freedreno', 'etnaviv', 'imx', 'nouveau',
+        'tegra', 'virgl', 'swrast',
+      ]
+    else
+      error('Unknown architecture @0@. Please pass -Dgallium-drivers to set driver options. Patches gladly accepted to fix this.'.format(
+            host_machine.cpu_family()))
+    endif
+  elif ['darwin', 'windows', 'cygwin', 'haiku'].contains(host_machine.system())
+    _drivers = ['swrast']
+  else
+    error('Unknown OS @0@. Please pass -Dgallium-drivers to set driver options. Patches gladly accepted to fix this.'.format(
+          host_machine.system()))
+  endif
+endif
+with_gallium_pl111 = _drivers.contains('pl111')
+with_gallium_radeonsi = _drivers.contains('radeonsi')
+with_gallium_r300 = _drivers.contains('r300')
+with_gallium_r600 = _drivers.contains('r600')
+with_gallium_nouveau = _drivers.contains('nouveau')
+with_gallium_freedreno = _drivers.contains('freedreno')
+with_gallium_softpipe = _drivers.contains('swrast')
+with_gallium_vc4 = _drivers.contains('vc4')
+with_gallium_v3d = _drivers.contains('v3d')
+with_gallium_etnaviv = _drivers.contains('etnaviv')
+with_gallium_imx = _drivers.contains('imx')
+with_gallium_tegra = _drivers.contains('tegra')
+with_gallium_i915 = _drivers.contains('i915')
+with_gallium_svga = _drivers.contains('svga')
+with_gallium_virgl = _drivers.contains('virgl')
+with_gallium_swr = _drivers.contains('swr')
+
+with_gallium = _drivers.length() != 0 and _drivers != ['']
+
+if with_gallium and system_has_kms_drm
+  _glx = get_option('glx')
+  _egl = get_option('egl')
+  if _glx == 'dri' or _egl == 'true' or (_glx == 'disabled' and _egl != 'false')
+    with_dri = true
+  endif
+endif
+
+_vulkan_drivers = get_option('vulkan-drivers')
+if _vulkan_drivers.contains('auto')
+  if system_has_kms_drm
+    if host_machine.cpu_family().startswith('x86')
+      _vulkan_drivers = ['amd', 'intel']
+    elif ['arm', 'aarch64'].contains(host_machine.cpu_family())
+      _vulkan_drivers = []
+    else
+      error('Unknown architecture @0@. Please pass -Dvulkan-drivers to set driver options. Patches gladly accepted to fix this.'.format(
+            host_machine.cpu_family()))
+    endif
+  elif ['darwin', 'windows', 'cygwin', 'haiku'].contains(host_machine.system())
+    # No vulkan driver supports windows or macOS currently
+    _vulkan_drivers = []
+  else
+    error('Unknown OS @0@. Please pass -Dvulkan-drivers to set driver options. Patches gladly accepted to fix this.'.format(
+          host_machine.system()))
+  endif
+endif
+
+with_intel_vk = _vulkan_drivers.contains('intel')
+with_amd_vk = _vulkan_drivers.contains('amd')
+with_any_vk = _vulkan_drivers.length() != 0 and _vulkan_drivers != ['']
+
+if with_dri_swrast and (with_gallium_softpipe or with_gallium_swr)
+  error('Only one swrast provider can be built')
+endif
+if with_dri_i915 and with_gallium_i915
+  error('Only one i915 provider can be built')
+endif
+if with_gallium_imx and not with_gallium_etnaviv
+  error('IMX driver requires etnaviv driver')
+endif
+if with_gallium_pl111 and not with_gallium_vc4
+  error('pl111 driver requires vc4 driver')
+endif
+if with_gallium_tegra and not with_gallium_nouveau
+  error('tegra driver requires nouveau driver')
+endif
+
+if host_machine.system() == 'darwin'
+  with_dri_platform = 'apple'
+elif ['windows', 'cygwin'].contains(host_machine.system())
+  with_dri_platform = 'windows'
+elif system_has_kms_drm
+  with_dri_platform = 'drm'
+else
+  # FIXME: haiku doesn't use dri, and xlib doesn't use dri, probably should
+  # assert here that one of those cases has been met.
+  # FIXME: illumos ends up here as well
+  with_dri_platform = 'none'
+endif
+
+_platforms = get_option('platforms')
+if _platforms.contains('auto')
+  if system_has_kms_drm
+    _platforms = ['x11', 'wayland', 'drm', 'surfaceless']
+  elif ['darwin', 'windows', 'cygwin'].contains(host_machine.system())
+    _platforms = ['x11', 'surfaceless']
+  elif ['haiku'].contains(host_machine.system())
+    _platforms = ['haiku']
+  else
+    error('Unknown OS @0@. Please pass -Dplatforms to set platforms. Patches gladly accepted to fix this.'.format(
+          host_machine.system()))
+  endif
+endif
+
+with_platform_android = _platforms.contains('android')
+with_platform_x11 = _platforms.contains('x11')
+with_platform_wayland = _platforms.contains('wayland')
+with_platform_drm = _platforms.contains('drm')
+with_platform_haiku = _platforms.contains('haiku')
+with_platform_surfaceless = _platforms.contains('surfaceless')
+
+with_platforms = false
+if _platforms.length() != 0 and _platforms != ['']
+  with_platforms = true
+  egl_native_platform = _platforms[0]
+endif
+
+_xlib_lease = get_option('xlib-lease')
+if _xlib_lease == 'auto'
+  with_xlib_lease = with_platform_x11 and with_platform_drm
+else
+  with_xlib_lease = _xlib_lease == 'true'
+endif
+
+with_glx = get_option('glx')
+if with_glx == 'auto'
+  if with_dri
+    with_glx = 'dri'
+  elif with_platform_haiku
+    with_glx = 'disabled'
+  elif with_gallium
+    # Even when building just gallium drivers the user probably wants dri
+    with_glx = 'dri'
+  elif with_platform_x11 and with_any_opengl and not with_any_vk
+    # The automatic behavior should not be to turn on xlib based glx when
+    # building only vulkan drivers
+    with_glx = 'xlib'
+  else
+    with_glx = 'disabled'
+  endif
+endif
+if with_glx == 'dri'
+   if with_gallium
+      with_dri = true
+   endif
+endif
+
+if not (with_dri or with_gallium or with_glx == 'xlib' or with_glx == 'gallium-xlib')
+  with_gles1 = false
+  with_gles2 = false
+  with_opengl = false
+  with_any_opengl = false
+  with_shared_glapi = false
+endif
+
+_gbm = get_option('gbm')
+if _gbm == 'auto'
+  with_gbm = system_has_kms_drm and with_dri
+else
+  with_gbm = _gbm == 'true'
+endif
+if with_gbm and not system_has_kms_drm
+  error('GBM only supports DRM/KMS platforms')
+endif
+
+_egl = get_option('egl')
+if _egl == 'auto'
+  with_egl = (
+    not ['darwin', 'windows'].contains(host_machine.system()) and
+    with_dri and with_shared_glapi and with_platforms
+  )
+elif _egl == 'true'
+  if not with_dri
+    error('EGL requires dri')
+  elif not with_shared_glapi
+    error('EGL requires shared-glapi')
+  elif not with_platforms
+    error('No platforms specified, consider -Dplatforms=drm,x11,surfaceless at least')
+  elif not ['disabled', 'dri'].contains(with_glx)
+    error('EGL requires dri, but a GLX is being built without dri')
+  elif ['darwin', 'windows'].contains(host_machine.system())
+    error('EGL is not available on Windows or MacOS')
+  endif
+  with_egl = true
+else
+  with_egl = false
+endif
+
+if with_egl and not (with_platform_drm or with_platform_surfaceless)
+  if with_gallium_radeonsi
+    error('RadeonSI requires drm or surfaceless platform when using EGL')
+  endif
+  if with_gallium_virgl
+    error('Virgl requires drm or surfaceless platform when using EGL')
+  endif
+endif
+
+pre_args += '-DGLX_USE_TLS'
+if with_glx != 'disabled'
+  if not (with_platform_x11 and with_any_opengl)
+    error('Cannot build GLX support without X11 platform support and at least one OpenGL API')
+  elif with_glx == 'gallium-xlib' 
+    if not with_gallium
+      error('Gallium-xlib based GLX requires at least one gallium driver')
+    elif not with_gallium_softpipe
+      error('Gallium-xlib based GLX requires softpipe or llvmpipe.')
+    elif with_dri
+      error('gallium-xlib conflicts with any dri driver')
+    endif
+  elif with_glx == 'xlib' 
+    if with_dri
+      error('xlib conflicts with any dri driver')
+    endif
+  elif with_glx == 'dri'
+    if not with_dri
+      error('dri based GLX requires at least one DRI driver')
+    elif not with_shared_glapi
+      error('dri based GLX requires shared-glapi')
+    endif
+  endif
+endif
+
+with_glvnd = get_option('glvnd')
+if with_glvnd
+  if with_glx == 'xlib' or with_glx == 'gallium-xlib'
+    error('Cannot build glvnd support for GLX that is not DRI based.')
+  elif with_glx == 'disabled' and not with_egl
+    error('glvnd requires DRI based GLX and/or EGL')
+  endif
+endif
+
+if with_vulkan_icd_dir == ''
+  with_vulkan_icd_dir = join_paths(get_option('datadir'), 'vulkan/icd.d')
+endif
+
+with_dri2 = (with_dri or with_any_vk) and with_dri_platform == 'drm'
+_dri3 = get_option('dri3')
+if _dri3 == 'auto'
+  with_dri3 = system_has_kms_drm and with_dri2
+else
+  with_dri3 = _dri3 == 'true'
+endif
+
+if with_any_vk and (with_platform_x11 and not with_dri3)
+  error('Vulkan drivers require dri3 for X11 support')
+endif
+if with_dri
+  if with_glx == 'disabled' and not with_egl and not with_gbm
+    error('building dri drivers require at least one windowing system')
+  endif
+endif
+
+prog_pkgconfig = find_program('pkg-config')
+
+_vdpau = get_option('gallium-vdpau')
+if not system_has_kms_drm
+  if _vdpau == 'true'
+    error('VDPAU state tracker can only be build on unix-like OSes.')
+  else
+    _vdpau = 'false'
+  endif
+elif not with_platform_x11
+  if _vdpau == 'true'
+    error('VDPAU state tracker requires X11 support.')
+  else
+    _vdpau = 'false'
+  endif
+elif not (with_gallium_r300 or with_gallium_r600 or with_gallium_radeonsi or
+          with_gallium_nouveau)
+  if _vdpau == 'true'
+    error('VDPAU state tracker requires at least one of the following gallium drivers: r300, r600, radeonsi, nouveau.')
+  else
+    _vdpau = 'false'
+  endif
+endif
+dep_vdpau = null_dep
+with_gallium_vdpau = false
+if _vdpau != 'false'
+  dep_vdpau = dependency('vdpau', version : '>= 1.1', required : _vdpau == 'true')
+  if dep_vdpau.found()
+    dep_vdpau = declare_dependency(
+      compile_args : run_command(prog_pkgconfig, ['vdpau', '--cflags']).stdout().split()
+    )
+    with_gallium_vdpau = true
+  endif
+endif
+
+if with_gallium_vdpau
+  pre_args += '-DHAVE_ST_VDPAU'
+endif
+vdpau_drivers_path = get_option('vdpau-libs-path')
+if vdpau_drivers_path == ''
+  vdpau_drivers_path = join_paths(get_option('libdir'), 'vdpau')
+endif
+
+_xvmc = get_option('gallium-xvmc')
+if not system_has_kms_drm
+  if _xvmc == 'true'
+    error('XVMC state tracker can only be build on unix-like OSes.')
+  else
+    _xvmc = 'false'
+  endif
+elif not with_platform_x11
+  if _xvmc == 'true'
+    error('XVMC state tracker requires X11 support.')
+  else
+    _xvmc = 'false'
+  endif
+elif not (with_gallium_r600 or with_gallium_nouveau)
+  if _xvmc == 'true'
+    error('XVMC state tracker requires at least one of the following gallium drivers: r600, nouveau.')
+  else
+    _xvmc = 'false'
+  endif
+endif
+dep_xvmc = null_dep
+with_gallium_xvmc = false
+if _xvmc != 'false'
+  dep_xvmc = dependency('xvmc', version : '>= 1.0.6', required : _xvmc == 'true')
+  with_gallium_xvmc = dep_xvmc.found()
+endif
+
+xvmc_drivers_path = get_option('xvmc-libs-path')
+if xvmc_drivers_path == ''
+  xvmc_drivers_path = get_option('libdir')
+endif
+
+_omx = get_option('gallium-omx')
+if not system_has_kms_drm
+  if ['auto', 'disabled'].contains(_omx)
+    _omx = 'disabled'
+  else
+    error('OMX state tracker can only be built on unix-like OSes.')
+  endif
+elif not (with_platform_x11 or with_platform_drm)
+  if ['auto', 'disabled'].contains(_omx)
+    _omx = 'disabled'
+  else
+    error('OMX state tracker requires X11 or drm platform support.')
+  endif
+elif not (with_gallium_r600 or with_gallium_radeonsi or with_gallium_nouveau)
+  if ['auto', 'disabled'].contains(_omx)
+    _omx = 'disabled'
+  else
+    error('OMX state tracker requires at least one of the following gallium drivers: r600, radeonsi, nouveau.')
+  endif
+endif
+with_gallium_omx = _omx
+dep_omx = null_dep
+dep_omx_other = []
+if ['auto', 'bellagio'].contains(_omx)
+  dep_omx = dependency(
+    'libomxil-bellagio', required : _omx == 'bellagio'
+  )
+  if dep_omx.found()
+    with_gallium_omx = 'bellagio'
+  endif
+endif
+if ['auto', 'tizonia'].contains(_omx)
+  if with_dri and with_egl
+    dep_omx = dependency(
+      'libtizonia', version : '>= 0.10.0',
+      required : _omx == 'tizonia',
+    )
+    dep_omx_other = [
+      dependency('libtizplatform', required : _omx == 'tizonia'),
+      dependency('tizilheaders', required : _omx == 'tizonia'),
+    ]
+    if dep_omx.found() and dep_omx_other[0].found() and dep_omx_other[1].found()
+      with_gallium_omx = 'tizonia'
+    endif
+  elif _omx == 'tizonia'
+    error('OMX-Tizonia state tracker requires dri and egl')
+  endif
+endif
+if _omx == 'auto'
+  with_gallium_omx = 'disabled'
+else
+  with_gallium_omx = _omx
+endif
+
+pre_args += [
+  '-DENABLE_ST_OMX_BELLAGIO=' + (with_gallium_omx == 'bellagio' ? '1' : '0'),
+  '-DENABLE_ST_OMX_TIZONIA=' + (with_gallium_omx == 'tizonia' ? '1' : '0'),
+]
+
+
+omx_drivers_path = get_option('omx-libs-path')
+
+if with_gallium_omx != 'disabled'
+  # Figure out where to put the omx driver.
+  # FIXME: this could all be vastly simplified by adding a 'defined_variable'
+  # argument to meson's get_pkgconfig_variable method.
+  if omx_drivers_path == ''
+    _omx_libdir = dep_omx.get_pkgconfig_variable('libdir')
+    _omx_drivers_dir = dep_omx.get_pkgconfig_variable('pluginsdir')
+    if _omx_libdir == get_option('libdir')
+      omx_drivers_path = _omx_drivers_dir
+    else
+      _omx_base_dir = []
+      # This will fail on windows. Does OMX run on windows?
+      _omx_libdir = _omx_libdir.split('/')
+      _omx_drivers_dir = _omx_drivers_dir.split('/')
+      foreach o : _omx_drivers_dir
+        if not _omx_libdir.contains(o)
+          _omx_base_dir += o
+        endif
+      endforeach
+      omx_drivers_path = join_paths(get_option('libdir'), _omx_base_dir)
+    endif
+  endif
+endif
+
+_va = get_option('gallium-va')
+if not system_has_kms_drm
+  if _va == 'true'
+    error('VA state tracker can only be built on unix-like OSes.')
+  else
+    _va = 'false'
+  endif
+elif not (with_platform_x11 or with_platform_drm)
+  if _va == 'true'
+    error('VA state tracker requires X11 or drm or wayland platform support.')
+  else
+    _va = 'false'
+  endif
+elif not (with_gallium_r600 or with_gallium_radeonsi or with_gallium_nouveau)
+  if _va == 'true'
+    error('VA state tracker requires at least one of the following gallium drivers: r600, radeonsi, nouveau.')
+  else
+    _va = 'false'
+  endif
+endif
+with_gallium_va = false
+dep_va = null_dep
+if _va != 'false'
+  dep_va = dependency('libva', version : '>= 0.38.0', required : _va == 'true')
+  if dep_va.found()
+    dep_va_headers = declare_dependency(
+      compile_args : run_command(prog_pkgconfig, ['libva', '--cflags']).stdout().split()
+    )
+    with_gallium_va = true
+  endif
+endif
+
+va_drivers_path = get_option('va-libs-path')
+if va_drivers_path == ''
+  va_drivers_path = join_paths(get_option('libdir'), 'dri')
+endif
+
+_xa = get_option('gallium-xa')
+if not system_has_kms_drm
+  if _xa == 'true'
+    error('XA state tracker can only be built on unix-like OSes.')
+  else
+    _xa = 'false'
+  endif
+elif not (with_gallium_nouveau or with_gallium_freedreno or with_gallium_i915
+          or with_gallium_svga)
+  if _xa == 'true'
+    error('XA state tracker requires at least one of the following gallium drivers: nouveau, freedreno, i915, svga.')
+  else
+    _xa = 'false'
+  endif
+endif
+with_gallium_xa = _xa != 'false'
+
+d3d_drivers_path = get_option('d3d-drivers-path')
+if d3d_drivers_path == ''
+  d3d_drivers_path = join_paths(get_option('libdir'), 'd3d')
+endif
+
+with_gallium_st_nine =  get_option('gallium-nine')
+if with_gallium_st_nine
+  if not with_gallium_softpipe
+    error('The nine state tracker requires gallium softpipe/llvmpipe.')
+  elif not (with_gallium_radeonsi or with_gallium_nouveau or with_gallium_r600
+            or with_gallium_r300 or with_gallium_svga or with_gallium_i915)
+    error('The nine state tracker requires at least one non-swrast gallium driver.')
+  endif
+  if not with_dri3
+    error('Using nine with wine requires dri3')
+  endif
+endif
+
+if get_option('power8') != 'false'
+  # on old versions of meson the cpu family would return as ppc64le on little
+  # endian power8, this was changed in 0.48 such that the family would always
+  # be ppc64 regardless of endianness, and the the machine.endian() value
+  # should be checked. Since we support versions < 0.48 we need to use
+  # startswith.
+  if host_machine.cpu_family().startswith('ppc64') and host_machine.endian() == 'little'
+    if cc.get_id() == 'gcc' and cc.version().version_compare('< 4.8')
+      error('Altivec is not supported with gcc version < 4.8.')
+    endif
+    if cc.compiles('''
+        #include <altivec.h>
+        int main() {
+          vector unsigned char r;
+          vector unsigned int v = vec_splat_u32 (1);
+          r = __builtin_vec_vgbbd ((vector unsigned char) v);
+          return 0;
+        }''',
+        args : '-mpower8-vector',
+        name : 'POWER8 intrinsics')
+      pre_args += ['-D_ARCH_PWR8', '-mpower8-vector']
+    elif get_option('power8') == 'true'
+      error('POWER8 intrinsic support required but not found.')
+    endif
+  endif
+endif
+
+_opencl = get_option('gallium-opencl')
+clover_cpp_std = []
+if _opencl != 'disabled'
+  if not with_gallium
+    error('OpenCL Clover implementation requires at least one gallium driver.')
+  endif
+
+  dep_clc = dependency('libclc')
+  with_gallium_opencl = true
+  with_opencl_icd = _opencl == 'icd'
+
+  if host_machine.cpu_family().startswith('ppc') and cpp.compiles('''
+      #if !defined(__VEC__) || !defined(__ALTIVEC__)
+      #error "AltiVec not enabled"
+      #endif''',
+      name : 'Altivec')
+    clover_cpp_std += ['cpp_std=gnu++11']
+  endif
+else
+  dep_clc = null_dep
+  with_gallium_opencl = false
+  with_gallium_icd = false
+endif
+
+gl_pkgconfig_c_flags = []
+if with_platform_x11
+  if with_any_vk or with_egl or (with_glx == 'dri' and with_dri_platform == 'drm')
+    pre_args += '-DHAVE_X11_PLATFORM'
+  endif
+  if with_glx == 'xlib' or with_glx == 'gallium-xlib'
+    pre_args += '-DUSE_XSHM'
+  else
+    pre_args += '-DGLX_INDIRECT_RENDERING'
+    if with_glx_direct
+      pre_args += '-DGLX_DIRECT_RENDERING'
+    endif
+    if with_dri_platform == 'drm'
+      pre_args += '-DGLX_USE_DRM'
+    elif with_dri_platform == 'apple'
+      pre_args += '-DGLX_USE_APPLEGL'
+    elif with_dri_platform == 'windows'
+      pre_args += '-DGLX_USE_WINDOWSGL'
+    endif
+  endif
+else
+  pre_args += '-DMESA_EGL_NO_X11_HEADERS'
+  gl_pkgconfig_c_flags += '-DMESA_EGL_NO_X11_HEADERS'
+endif
+if with_platform_drm
+  if with_egl and not with_gbm
+    error('EGL drm platform requires gbm')
+  endif
+  pre_args += '-DHAVE_DRM_PLATFORM'
+endif
+if with_platform_surfaceless
+  pre_args += '-DHAVE_SURFACELESS_PLATFORM'
+endif
+if with_platform_android
+  dep_android = [
+    dependency('cutils'),
+    dependency('hardware'),
+    dependency('sync'),
+  ]
+  pre_args += '-DHAVE_ANDROID_PLATFORM'
+endif
+if with_platform_haiku
+  pre_args += '-DHAVE_HAIKU_PLATFORM'
+endif
+
+prog_python = import('python3').find_python()
+has_mako = run_command(
+  prog_python, '-c',
+  '''
+from distutils.version import StrictVersion
+import mako
+assert StrictVersion(mako.__version__) > StrictVersion("0.8.0")
+  ''')
+if has_mako.returncode() != 0
+  error('Python (3.x) mako module >= 0.8.0 required to build mesa.')
+endif
+
+if cc.get_id() == 'gcc' and cc.version().version_compare('< 4.4.6')
+  error('When using GCC, version 4.4.6 or later is required.')
+endif
+
+# Define DEBUG for debug builds only (debugoptimized is not included on this one)
+if get_option('buildtype') == 'debug'
+  pre_args += '-DDEBUG'
+endif
+
+if get_option('shader-cache')
+  pre_args += '-DENABLE_SHADER_CACHE'
+elif with_amd_vk
+  error('Radv requires shader cache support')
+endif
+
+# Check for GCC style builtins
+foreach b : ['bswap32', 'bswap64', 'clz', 'clzll', 'ctz', 'expect', 'ffs',
+             'ffsll', 'popcount', 'popcountll', 'unreachable']
+  if cc.has_function(b)
+    pre_args += '-DHAVE___BUILTIN_@0@'.format(b.to_upper())
+  endif
+endforeach
+
+# check for GCC __attribute__
+foreach a : ['const', 'flatten', 'malloc', 'pure', 'unused',
+             'warn_unused_result', 'weak',]
+  if cc.compiles('int foo(void) __attribute__((@0@));'.format(a),
+                 name : '__attribute__((@0@))'.format(a))
+    pre_args += '-DHAVE_FUNC_ATTRIBUTE_@0@'.format(a.to_upper())
+  endif
+endforeach
+if cc.compiles('int foo(const char *p, ...) __attribute__((format(printf, 1, 2)));',
+               name : '__attribute__((format(...)))')
+  pre_args += '-DHAVE_FUNC_ATTRIBUTE_FORMAT'
+endif
+if cc.compiles('struct __attribute__((packed)) foo { int bar; };',
+               name : '__attribute__((packed))')
+  pre_args += '-DHAVE_FUNC_ATTRIBUTE_PACKED'
+endif
+if cc.compiles('int *foo(void) __attribute__((returns_nonnull));',
+               name : '__attribute__((returns_nonnull))')
+  pre_args += '-DHAVE_FUNC_ATTRIBUTE_RETURNS_NONNULL'
+endif
+if cc.compiles('''int foo_def(void) __attribute__((visibility("default")));
+                  int foo_hid(void) __attribute__((visibility("hidden")));
+                  int foo_int(void) __attribute__((visibility("internal")));
+                  int foo_pro(void) __attribute__((visibility("protected")));''',
+               name : '__attribute__((visibility(...)))')
+  pre_args += '-DHAVE_FUNC_ATTRIBUTE_VISIBILITY'
+endif
+if cc.compiles('int foo(void) { return 0; } int bar(void) __attribute__((alias("foo")));',
+               name : '__attribute__((alias(...)))')
+  pre_args += '-DHAVE_FUNC_ATTRIBUTE_ALIAS'
+endif
+if cc.compiles('int foo(void) __attribute__((__noreturn__));',
+               name : '__attribute__((__noreturn__))')
+  pre_args += '-DHAVE_FUNC_ATTRIBUTE_NORETURN'
+endif
+
+# TODO: this is very incomplete
+if ['linux', 'cygwin', 'gnu'].contains(host_machine.system())
+  pre_args += '-D_GNU_SOURCE'
+endif
+
+# Check for generic C arguments
+c_args = []
+foreach a : ['-Werror=implicit-function-declaration',
+             '-Werror=missing-prototypes', '-Werror=return-type',
+             '-fno-math-errno',
+             '-fno-trapping-math', '-Qunused-arguments']
+  if cc.has_argument(a)
+    c_args += a
+  endif
+endforeach
+
+foreach a : ['missing-field-initializers', 'format-truncation']
+  if cc.has_argument('-W' + a)
+    c_args += '-Wno-' + a
+  endif
+endforeach
+
+c_vis_args = []
+if cc.has_argument('-fvisibility=hidden')
+  c_vis_args += '-fvisibility=hidden'
+endif
+
+# Check for generic C++ arguments
+cpp_args = []
+foreach a : ['-Werror=return-type',
+             '-fno-math-errno', '-fno-trapping-math',
+             '-Qunused-arguments']
+  if cpp.has_argument(a)
+    cpp_args += a
+  endif
+endforeach
+
+# For some reason, the test for -Wno-foo always succeeds with gcc, even if the
+# option is not supported. Hence, check for -Wfoo instead.
+
+foreach a : ['non-virtual-dtor', 'missing-field-initializers', 'format-truncation']
+  if cpp.has_argument('-W' + a)
+    cpp_args += '-Wno-' + a
+  endif
+endforeach
+
+no_override_init_args = []
+foreach a : ['override-init', 'initializer-overrides']
+  if cc.has_argument('-W' + a)
+    no_override_init_args += '-Wno-' + a
+  endif
+endforeach
+
+cpp_vis_args = []
+if cpp.has_argument('-fvisibility=hidden')
+  cpp_vis_args += '-fvisibility=hidden'
+endif
+
+# Check for C and C++ arguments for MSVC2013 compatibility. These are only used
+# in parts of the mesa code base that need to compile with old versions of
+# MSVC, mainly common code
+c_msvc_compat_args = []
+cpp_msvc_compat_args = []
+foreach a : ['-Werror=pointer-arith', '-Werror=vla']
+  if cc.has_argument(a)
+    c_msvc_compat_args += a
+  endif
+  if cpp.has_argument(a)
+    cpp_msvc_compat_args += a
+  endif
+endforeach
+
+if host_machine.cpu_family().startswith('x86')
+  pre_args += '-DUSE_SSE41'
+  with_sse41 = true
+  sse41_args = ['-msse4.1']
+
+  # GCC on x86 (not x86_64) with -msse* assumes a 16 byte aligned stack, but
+  # that's not guaranteed
+  if host_machine.cpu_family() == 'x86'
+    sse41_args += '-mstackrealign'
+  endif
+else
+  with_sse41 = false
+  sse41_args = []
+endif
+
+# Check for GCC style atomics
+dep_atomic = null_dep
+
+if cc.compiles('''#include <stdint.h>
+                  int main() {
+                    struct {
+                      uint64_t *v;
+                    } x;
+                    return (int)__atomic_load_n(x.v, __ATOMIC_ACQUIRE) &
+                           (int)__atomic_add_fetch(x.v, (uint64_t)1, __ATOMIC_ACQ_REL);
+
+                  }''',
+               name : 'GCC atomic builtins')
+  pre_args += '-DUSE_GCC_ATOMIC_BUILTINS'
+
+  # Not all atomic calls can be turned into lock-free instructions, in which
+  # GCC will make calls into the libatomic library. Check whether we need to
+  # link with -latomic.
+  #
+  # This can happen for 64-bit atomic operations on 32-bit architectures such
+  # as ARM.
+  if not cc.links('''#include <stdint.h>
+                     int main() {
+                       struct {
+                         uint64_t *v;
+                       } x;
+                       return (int)__atomic_load_n(x.v, __ATOMIC_ACQUIRE) &
+                              (int)__atomic_add_fetch(x.v, (uint64_t)1, __ATOMIC_ACQ_REL);
+                     }''',
+                  name : 'GCC atomic builtins required -latomic')
+    dep_atomic = cc.find_library('atomic')
+  endif
+endif
+if not cc.links('''#include <stdint.h>
+                   uint64_t v;
+                   int main() {
+                     return __sync_add_and_fetch(&v, (uint64_t)1);
+                   }''',
+                dependencies : dep_atomic,
+                name : 'GCC 64bit atomics')
+  pre_args += '-DMISSING_64BIT_ATOMICS'
+endif
+
+# TODO: shared/static? Is this even worth doing?
+
+# When cross compiling we generally need to turn off the use of assembly,
+# because mesa's assembly relies on building an executable for the host system,
+# and running it to get information about struct sizes. There is at least one
+# case of cross compiling where we can use asm, and that's x86_64 -> x86 when
+# host OS == build OS, since in that case the build machine can run the host's
+# binaries.
+if meson.is_cross_build() 
+  if build_machine.system() != host_machine.system()
+    # TODO: It may be possible to do this with an exe_wrapper (like wine).
+    message('Cross compiling from one OS to another, disabling assembly.')
+    with_asm = false
+  elif not (build_machine.cpu_family().startswith('x86') and host_machine.cpu_family() == 'x86')
+    # FIXME: Gentoo always sets -m32 for x86_64 -> x86 builds, resulting in an
+    # x86 -> x86 cross compile. We use startswith rather than == to handle this
+    # case.
+    # TODO: There may be other cases where the 64 bit version of the
+    # architecture can run 32 bit binaries (aarch64 and armv7 for example)
+    message('''
+      Cross compiling to different architectures, and the host cannot run
+      the build machine's binaries. Disabling assembly.
+    ''')
+    with_asm = false
+  endif
+endif
+
+with_asm_arch = ''
+if with_asm
+  if host_machine.cpu_family() == 'x86'
+    if system_has_kms_drm or host_machine.system() == 'gnu'
+      with_asm_arch = 'x86'
+      pre_args += ['-DUSE_X86_ASM', '-DUSE_MMX_ASM', '-DUSE_3DNOW_ASM',
+                   '-DUSE_SSE_ASM']
+
+      if with_glx_read_only_text
+         pre_args += ['-DGLX_X86_READONLY_TEXT']
+      endif
+    endif
+  elif host_machine.cpu_family() == 'x86_64'
+    if system_has_kms_drm
+      with_asm_arch = 'x86_64'
+      pre_args += ['-DUSE_X86_64_ASM']
+    endif
+  elif host_machine.cpu_family() == 'arm'
+    if system_has_kms_drm
+      with_asm_arch = 'arm'
+      pre_args += ['-DUSE_ARM_ASM']
+    endif
+  elif host_machine.cpu_family() == 'aarch64'
+    if system_has_kms_drm
+      with_asm_arch = 'aarch64'
+      pre_args += ['-DUSE_AARCH64_ASM']
+    endif
+  elif host_machine.cpu_family() == 'sparc64'
+    if system_has_kms_drm
+      with_asm_arch = 'sparc'
+      pre_args += ['-DUSE_SPARC_ASM']
+    endif
+  elif host_machine.cpu_family().startswith('ppc64') and host_machine.endian() == 'little'
+    if system_has_kms_drm
+      with_asm_arch = 'ppc64le'
+      pre_args += ['-DUSE_PPC64LE_ASM']
+    endif
+  endif
+endif
+
+# Check for standard headers and functions
+if cc.has_header_symbol('sys/sysmacros.h', 'major')
+  pre_args += '-DMAJOR_IN_SYSMACROS'
+elif cc.has_header_symbol('sys/mkdev.h', 'major')
+  pre_args += '-DMAJOR_IN_MKDEV'
+endif
+
+foreach h : ['xlocale.h', 'sys/sysctl.h', 'linux/futex.h', 'endian.h', 'dlfcn.h']
+  if cc.compiles('#include <@0@>'.format(h), name : '@0@'.format(h))
+    pre_args += '-DHAVE_@0@'.format(h.to_upper().underscorify())
+  endif
+endforeach
+
+foreach f : ['strtof', 'mkostemp', 'posix_memalign', 'timespec_get', 'memfd_create']
+  if cc.has_function(f)
+    pre_args += '-DHAVE_@0@'.format(f.to_upper())
+  endif
+endforeach
+
+# strtod locale support
+if cc.links('''
+    #define _GNU_SOURCE
+    #include <stdlib.h>
+    #include <locale.h>
+    #ifdef HAVE_XLOCALE_H
+    #include <xlocale.h>
+    #endif
+    int main() {
+      locale_t loc = newlocale(LC_CTYPE_MASK, "C", NULL);
+      const char *s = "1.0";
+      char *end;
+      double d = strtod_l(s, end, loc);
+      float f = strtof_l(s, end, loc);
+      freelocale(loc);
+      return 0;
+    }''',
+    args : pre_args,
+    name : 'strtod has locale support')
+  pre_args += '-DHAVE_STRTOD_L'
+endif
+
+# Check for some linker flags
+ld_args_bsymbolic = []
+if cc.links('int main() { return 0; }', args : '-Wl,-Bsymbolic', name : 'Bsymbolic')
+  ld_args_bsymbolic += '-Wl,-Bsymbolic'
+endif
+ld_args_gc_sections = []
+if cc.links('static char unused() { return 5; } int main() { return 0; }',
+            args : '-Wl,--gc-sections', name : 'gc-sections')
+  ld_args_gc_sections += '-Wl,--gc-sections'
+endif
+with_ld_version_script = false
+if cc.links('int main() { return 0; }',
+            args : '-Wl,--version-script=@0@'.format(
+              join_paths(meson.source_root(), 'build-support/conftest.map')),
+            name : 'version-script')
+  with_ld_version_script = true
+endif
+with_ld_dynamic_list = false
+if cc.links('int main() { return 0; }',
+            args : '-Wl,--dynamic-list=@0@'.format(
+              join_paths(meson.source_root(), 'build-support/conftest.dyn')),
+            name : 'dynamic-list')
+  with_ld_dynamic_list = true
+endif
+ld_args_build_id = []
+if build_machine.system() != 'darwin'
+   ld_args_build_id += '-Wl,--build-id=sha1'
+endif
+
+# check for dl support
+if cc.has_function('dlopen')
+  dep_dl = null_dep
+else
+  dep_dl = cc.find_library('dl')
+endif
+if cc.has_function('dladdr', dependencies : dep_dl)
+  # This is really only required for megadrivers
+  pre_args += '-DHAVE_DLADDR'
+endif
+
+if cc.has_function('dl_iterate_phdr')
+  pre_args += '-DHAVE_DL_ITERATE_PHDR'
+elif with_intel_vk
+  error('Intel "Anvil" Vulkan driver requires the dl_iterate_phdr function')
+elif with_dri_i965 and get_option('shader-cache')
+  error('Intel i965 GL driver requires dl_iterate_phdr when built with shader caching.')
+endif
+
+# Determine whether or not the rt library is needed for time functions
+if cc.has_function('clock_gettime')
+  dep_clock = null_dep
+else
+  dep_clock = cc.find_library('rt')
+endif
+
+# TODO: some of these may be conditional
+dep_zlib = dependency('zlib', version : '>= 1.2.3')
+pre_args += '-DHAVE_ZLIB'
+dep_thread = dependency('threads')
+if dep_thread.found() and host_machine.system() != 'windows'
+  pre_args += '-DHAVE_PTHREAD'
+  if cc.has_function(
+      'pthread_setaffinity_np',
+      dependencies : dep_thread,
+      prefix : '#include <pthread.h>',
+      args : '-D_GNU_SOURCE')
+    pre_args += '-DHAVE_PTHREAD_SETAFFINITY'
+  endif
+endif
+dep_expat = dependency('expat')
+# this only exists on linux so either this is linux and it will be found, or
+# its not linux and and wont
+dep_m = cc.find_library('m', required : false)
+
+# Check for libdrm. various drivers have different libdrm version requirements,
+# but we always want to use the same version for all libdrm modules. That means
+# even if driver foo requires 2.4.0 and driver bar requires 2.4.3, if foo and
+# bar are both on use 2.4.3 for both of them
+dep_libdrm_amdgpu = null_dep
+dep_libdrm_radeon = null_dep
+dep_libdrm_nouveau = null_dep
+dep_libdrm_etnaviv = null_dep
+dep_libdrm_intel = null_dep
+
+_drm_amdgpu_ver = '2.4.95'
+_drm_radeon_ver = '2.4.71'
+_drm_nouveau_ver = '2.4.66'
+_drm_etnaviv_ver = '2.4.89'
+_drm_intel_ver = '2.4.75'
+_drm_ver = '2.4.75'
+
+_libdrm_checks = [
+  ['intel', with_dri_i915 or with_gallium_i915],
+  ['amdgpu', with_amd_vk or with_gallium_radeonsi],
+  ['radeon', (with_gallium_radeonsi or with_dri_r100 or with_dri_r200 or
+              with_gallium_r300 or with_gallium_r600)],
+  ['nouveau', (with_gallium_nouveau or with_dri_nouveau)],
+  ['etnaviv', with_gallium_etnaviv],
+]
+
+# VC4 only needs core libdrm support of this version, not a libdrm_vc4
+# library.
+if with_gallium_vc4
+  _drm_ver = '2.4.89'
+endif
+
+# Loop over the enables versions and get the highest libdrm requirement for all
+# active drivers.
+_drm_blame = ''
+foreach d : _libdrm_checks
+  ver = get_variable('_drm_@0@_ver'.format(d[0]))
+  if d[1] and ver.version_compare('>' + _drm_ver)
+    _drm_ver = ver
+    _drm_blame = d[0]
+  endif
+endforeach
+if _drm_blame != ''
+  message('libdrm @0@ needed because @1@ has the highest requirement'.format(_drm_ver, _drm_blame))
+endif
+
+# Then get each libdrm module
+foreach d : _libdrm_checks
+  if d[1]
+    set_variable(
+      'dep_libdrm_' + d[0],
+      dependency('libdrm_' + d[0], version : '>=' + _drm_ver)
+    )
+  endif
+endforeach
+
+with_gallium_drisw_kms = false
+dep_libdrm = dependency(
+  'libdrm', version : '>=' + _drm_ver,
+  required : with_dri2 or with_dri3
+)
+if dep_libdrm.found()
+  pre_args += '-DHAVE_LIBDRM'
+  if with_dri_platform == 'drm' and with_dri
+    with_gallium_drisw_kms = true
+  endif
+endif
+
+llvm_modules = ['bitwriter', 'engine', 'mcdisassembler', 'mcjit']
+llvm_optional_modules = []
+if with_amd_vk or with_gallium_radeonsi or with_gallium_r600
+  llvm_modules += ['amdgpu', 'native', 'bitreader', 'ipo']
+  if with_gallium_r600
+    llvm_modules += 'asmparser'
+  endif
+endif
+if with_gallium_opencl
+  llvm_modules += [
+    'all-targets', 'linker', 'coverage', 'instrumentation', 'ipo', 'irreader',
+    'lto', 'option', 'objcarcopts', 'profiledata',
+  ]
+  llvm_optional_modules += ['coroutines']
+endif
+
+if with_amd_vk or with_gallium_radeonsi
+  _llvm_version = '>= 6.0.0'
+elif with_gallium_swr
+  _llvm_version = '>= 6.0.0'
+elif with_gallium_opencl or with_gallium_r600
+  _llvm_version = '>= 3.9.0'
+else
+  _llvm_version = '>= 3.3.0'
+endif
+
+_shared_llvm = get_option('shared-llvm')
+
+_llvm = get_option('llvm')
+dep_llvm = null_dep
+with_llvm = false
+if _llvm != 'false'
+  dep_llvm = dependency(
+    'llvm',
+    version : _llvm_version,
+    modules : llvm_modules,
+    optional_modules : llvm_optional_modules,
+    required : (
+      with_amd_vk or with_gallium_radeonsi or with_gallium_swr or
+      with_gallium_opencl or _llvm == 'true'
+    ),
+    static : not _shared_llvm,
+  )
+  with_llvm = dep_llvm.found()
+endif
+if with_llvm
+  _llvm_version = dep_llvm.version().split('.')
+
+  # 3 digits versions in LLVM only started from 3.4.1 on
+  if dep_llvm.version().version_compare('>= 3.4.1')
+    _llvm_patch = _llvm_version[2]
+  else
+    _llvm_patch = '0'
+  endif
+
+  pre_args += [
+    '-DHAVE_LLVM=0x0@0@0@1@'.format(_llvm_version[0], _llvm_version[1]),
+    '-DMESA_LLVM_VERSION_PATCH=@0@'.format(_llvm_patch),
+  ]
+
+  # LLVM can be built without rtti, turning off rtti changes the ABI of C++
+  # programs, so we need to build all C++ code in mesa without rtti as well to
+  # ensure that linking works.
+  if dep_llvm.get_configtool_variable('has-rtti') == 'NO'
+    if with_gallium_nouveau
+      error('The Nouveau driver requires rtti. You either need to turn off nouveau or use an LLVM built with LLVM_ENABLE_RTTI.')
+    endif
+    cpp_args += '-fno-rtti'
+  endif
+elif with_amd_vk or with_gallium_radeonsi or with_gallium_swr
+  error('The following drivers require LLVM: Radv, RadeonSI, SWR. One of these is enabled, but LLVM is disabled.')
+endif
+
+if (with_amd_vk or with_gallium_radeonsi or with_gallium_opencl or
+    (with_gallium_r600 and with_llvm))
+  dep_elf = dependency('libelf', required : false)
+  if not dep_elf.found()
+    dep_elf = cc.find_library('elf')
+  endif
+else
+  dep_elf = null_dep
+endif
+
+dep_glvnd = null_dep
+if with_glvnd
+  dep_glvnd = dependency('libglvnd', version : '>= 0.2.0')
+  pre_args += '-DUSE_LIBGLVND=1'
+endif
+
+if with_valgrind != 'false'
+  dep_valgrind = dependency('valgrind', required : with_valgrind == 'true')
+  if dep_valgrind.found()
+    pre_args += '-DHAVE_VALGRIND'
+  endif
+else
+  dep_valgrind = null_dep
+endif
+
+# pthread stubs. Lets not and say we didn't
+
+prog_bison = find_program('bison', required : with_any_opengl)
+prog_flex = find_program('flex', required : with_any_opengl)
+
+dep_selinux = null_dep
+if get_option('selinux')
+  dep_selinux = dependency('libselinux')
+  pre_args += '-DMESA_SELINUX'
+endif
+
+if with_libunwind != 'false'
+  dep_unwind = dependency('libunwind', required : with_libunwind == 'true')
+  if dep_unwind.found()
+    pre_args += '-DHAVE_LIBUNWIND'
+  endif
+else
+  dep_unwind = null_dep
+endif
+
+if with_osmesa != 'none'
+  if with_osmesa == 'classic' and not with_dri_swrast
+    error('OSMesa classic requires dri (classic) swrast.')
+  endif
+  if with_osmesa == 'gallium' and not with_gallium_softpipe
+    error('OSMesa gallium requires gallium softpipe or llvmpipe.')
+  endif
+  osmesa_lib_name = 'OSMesa'
+  osmesa_bits = get_option('osmesa-bits')
+  if osmesa_bits != '8'
+    if with_dri or with_glx != 'disabled'
+      error('OSMesa bits must be 8 if building glx or dir based drivers')
+    endif
+    osmesa_lib_name = osmesa_lib_name + osmesa_bits
+    pre_args += [
+      '-DCHAN_BITS=@0@'.format(osmesa_bits), '-DDEFAULT_SOFTWARE_DEPTH_BITS=31'
+    ]
+  endif
+endif
+
+# TODO: symbol mangling
+
+if with_platform_wayland
+  dep_wl_scanner = dependency('wayland-scanner', native: true)
+  prog_wl_scanner = find_program(dep_wl_scanner.get_pkgconfig_variable('wayland_scanner'))
+  if dep_wl_scanner.version().version_compare('>= 1.15')
+    wl_scanner_arg = 'private-code'
+  else
+    wl_scanner_arg = 'code'
+  endif
+  dep_wl_protocols = dependency('wayland-protocols', version : '>= 1.8')
+  dep_wayland_client = dependency('wayland-client', version : '>=1.11')
+  dep_wayland_server = dependency('wayland-server', version : '>=1.11')
+  if with_egl
+    dep_wayland_egl = dependency('wayland-egl-backend', version : '>= 3')
+    dep_wayland_egl_headers = declare_dependency(
+      compile_args : run_command(prog_pkgconfig, ['wayland-egl-backend', '--cflags']).stdout().split())
+  endif
+  wayland_dmabuf_xml = join_paths(
+    dep_wl_protocols.get_pkgconfig_variable('pkgdatadir'), 'unstable',
+    'linux-dmabuf', 'linux-dmabuf-unstable-v1.xml'
+  )
+  pre_args += ['-DHAVE_WAYLAND_PLATFORM', '-DWL_HIDE_DEPRECATED']
+endif
+
+dep_x11 = null_dep
+dep_xext = null_dep
+dep_xdamage = null_dep
+dep_xfixes = null_dep
+dep_x11_xcb = null_dep
+dep_xcb = null_dep
+dep_xcb_glx = null_dep
+dep_xcb_dri2 = null_dep
+dep_xcb_dri3 = null_dep
+dep_dri2proto = null_dep
+dep_glproto = null_dep
+dep_xxf86vm = null_dep
+dep_xcb_dri3 = null_dep
+dep_xcb_present = null_dep
+dep_xcb_sync = null_dep
+dep_xcb_xfixes = null_dep
+dep_xshmfence = null_dep
+dep_xcb_xrandr = null_dep
+dep_xlib_xrandr = null_dep
+if with_platform_x11
+  if with_glx == 'xlib' or with_glx == 'gallium-xlib'
+    dep_x11 = dependency('x11')
+    dep_xext = dependency('xext')
+    dep_xcb = dependency('xcb')
+  elif with_glx == 'dri'
+    dep_x11 = dependency('x11')
+    dep_xext = dependency('xext')
+    dep_xdamage = dependency('xdamage', version : '>= 1.1')
+    dep_xfixes = dependency('xfixes')
+    dep_xcb_glx = dependency('xcb-glx', version : '>= 1.8.1')
+  endif
+  if (with_any_vk or with_glx == 'dri' or
+       (with_gallium_vdpau or with_gallium_xvmc or with_gallium_va or
+        with_gallium_omx != 'disabled'))
+    dep_xcb = dependency('xcb')
+    dep_x11_xcb = dependency('x11-xcb')
+  endif
+  if with_any_vk or with_egl or (with_glx == 'dri' and with_dri_platform == 'drm')
+    dep_xcb_dri2 = dependency('xcb-dri2', version : '>= 1.8')
+
+    if with_dri3
+      pre_args += '-DHAVE_DRI3'
+      dep_xcb_dri3 = dependency('xcb-dri3')
+      dep_xcb_present = dependency('xcb-present')
+      # until xcb-dri3 has been around long enough to make a hard-dependency:
+      if (dep_xcb_dri3.version().version_compare('>= 1.13') and
+          dep_xcb_present.version().version_compare('>= 1.13'))
+        pre_args += '-DHAVE_DRI3_MODIFIERS'
+      endif
+      dep_xcb_sync = dependency('xcb-sync')
+      dep_xshmfence = dependency('xshmfence', version : '>= 1.1')
+    endif
+  endif
+  if with_glx == 'dri'
+    if with_dri_platform == 'drm'
+      dep_dri2proto = dependency('dri2proto', version : '>= 2.8')
+      dep_xxf86vm = dependency('xxf86vm')
+    endif
+    dep_glproto = dependency('glproto', version : '>= 1.4.14')
+  endif
+  if (with_egl or (
+      with_gallium_vdpau or with_gallium_xvmc or with_gallium_xa or
+      with_gallium_omx != 'disabled'))
+    dep_xcb_xfixes = dependency('xcb-xfixes')
+  endif
+  if with_xlib_lease
+    dep_xcb_xrandr = dependency('xcb-randr', version : '>= 1.12')
+    dep_xlib_xrandr = dependency('xrandr', version : '>= 1.3')
+  endif
+endif
+
+if get_option('gallium-extra-hud')
+  pre_args += '-DHAVE_GALLIUM_EXTRA_HUD=1'
+endif
+
+_sensors = get_option('lmsensors')
+if _sensors != 'false'
+  dep_lmsensors = cc.find_library('sensors', required : _sensors == 'true')
+  if dep_lmsensors.found()
+    pre_args += '-DHAVE_LIBSENSORS=1'
+  endif
+else
+  dep_lmsensors = null_dep
+endif
+
+foreach a : pre_args
+  add_project_arguments(a, language : ['c', 'cpp'])
+endforeach
+foreach a : c_args
+  add_project_arguments(a, language : ['c'])
+endforeach
+foreach a : cpp_args
+  add_project_arguments(a, language : ['cpp'])
+endforeach
+
+inc_include = include_directories('include')
+
+gl_priv_reqs = []
+
+if with_glx == 'xlib' or with_glx == 'gallium-xlib'
+  gl_priv_reqs += ['x11', 'xext', 'xcb']
+elif with_glx == 'dri'
+  gl_priv_reqs += [
+    'x11', 'xext', 'xdamage >= 1.1', 'xfixes', 'x11-xcb', 'xcb',
+    'xcb-glx >= 1.8.1']
+  if with_dri_platform == 'drm'
+    gl_priv_reqs += 'xcb-dri2 >= 1.8'
+    gl_priv_reqs += 'xxf86vm'
+  endif
+endif
+if dep_libdrm.found()
+  gl_priv_reqs += 'libdrm >= 2.4.75'
+endif
+
+gl_priv_libs = []
+if dep_thread.found()
+  gl_priv_libs += ['-lpthread', '-pthread']
+endif
+if dep_m.found()
+  gl_priv_libs += '-lm'
+endif
+if dep_dl.found()
+  gl_priv_libs += '-ldl'
+endif
+
+pkg = import('pkgconfig')
+
+env_test = environment()
+env_test.set('NM', find_program('nm').path())
+
+subdir('include')
+subdir('bin')
+subdir('src')
diff --git a/lib/mesa/meson_options.txt b/lib/mesa/meson_options.txt
new file mode 100644
index 000000000..589d10bb3
--- /dev/null
+++ b/lib/mesa/meson_options.txt
@@ -0,0 +1,326 @@
+# Copyright © 2017-2018 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+option(
+  'platforms',
+  type : 'array',
+  value : ['auto'],
+  choices : [
+    '', 'auto', 'x11', 'wayland', 'drm', 'surfaceless', 'haiku', 'android',
+  ],
+  description : 'window systems to support. If this is set to `auto`, all platforms applicable will be enabled.'
+)
+option(
+  'dri3',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'enable support for dri3'
+)
+option(
+  'dri-drivers',
+  type : 'array',
+  value : ['auto'],
+  choices : ['', 'auto', 'i915', 'i965', 'r100', 'r200', 'nouveau', 'swrast'],
+  description : 'List of dri drivers to build. If this is set to auto all drivers applicable to the target OS/architecture will be built'
+)
+option(
+  'dri-drivers-path',
+  type : 'string',
+  value : '',
+  description : 'Location to install dri drivers. Default: $libdir/dri.'
+)
+option(
+  'dri-search-path',
+  type : 'string',
+  value : '',
+  description : 'Locations to search for dri drivers, passed as colon separated list. Default: dri-drivers-path.'
+)
+option(
+  'gallium-drivers',
+  type : 'array',
+  value : ['auto'],
+  choices : [
+    '', 'auto', 'pl111', 'radeonsi', 'r300', 'r600', 'nouveau', 'freedreno',
+    'swrast', 'v3d', 'vc4', 'etnaviv', 'imx', 'tegra', 'i915', 'svga', 'virgl',
+    'swr',
+  ],
+  description : 'List of gallium drivers to build. If this is set to auto all drivers applicable to the target OS/architecture will be built'
+)
+option(
+  'gallium-extra-hud',
+  type : 'boolean',
+  value : false,
+  description : 'Enable HUD block/NIC I/O HUD status support',
+)
+option(
+  'gallium-vdpau',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'enable gallium vdpau state tracker.',
+)
+option(
+  'vdpau-libs-path',
+  type : 'string',
+  value : '',
+  description : 'path to put vdpau libraries. defaults to $libdir/vdpau.'
+)
+option(
+  'gallium-xvmc',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'enable gallium xvmc state tracker.',
+)
+option(
+  'xvmc-libs-path',
+  type : 'string',
+  value : '',
+  description : 'path to put xvmc libraries. defaults to $libdir.'
+)
+option(
+  'gallium-omx',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'disabled', 'bellagio', 'tizonia'],
+  description : 'enable gallium omx state tracker.',
+)
+option(
+  'omx-libs-path',
+  type : 'string',
+  value : '',
+  description : 'path to put omx libraries. defaults to omx-bellagio pkg-config pluginsdir.'
+)
+option(
+  'gallium-va',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'enable gallium va state tracker.',
+)
+option(
+  'va-libs-path',
+  type : 'string',
+  value : '',
+  description : 'path to put va libraries. defaults to $libdir/dri.'
+)
+option(
+  'gallium-xa',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'enable gallium xa state tracker.',
+)
+option(
+  'gallium-nine',
+  type : 'boolean',
+  value : false,
+  description : 'build gallium "nine" Direct3D 9.x state tracker.',
+)
+option(
+  'gallium-opencl',
+  type : 'combo',
+  choices : ['icd', 'standalone', 'disabled'],
+  value : 'disabled',
+  description : 'build gallium "clover" OpenCL state tracker.',
+)
+option(
+  'd3d-drivers-path',
+  type : 'string',
+  value : '',
+  description : 'Location of D3D drivers. Default: $libdir/d3d',
+)
+option(
+  'vulkan-drivers',
+  type : 'array',
+  value : ['auto'],
+  choices : ['', 'auto', 'amd', 'intel'],
+  description : 'List of vulkan drivers to build. If this is set to auto all drivers applicable to the target OS/architecture will be built'
+)
+option(
+  'shader-cache',
+  type : 'boolean',
+  value : true,
+  description : 'Build with on-disk shader cache support'
+)
+option(
+  'vulkan-icd-dir',
+  type : 'string',
+  value : '',
+  description : 'Location relative to prefix to put vulkan icds on install. Default: $datadir/vulkan/icd.d'
+)
+option(
+  'shared-glapi',
+  type : 'boolean',
+  value : true,
+  description : 'Whether to build a shared or static glapi'
+)
+option(
+  'gles1',
+  type : 'boolean',
+  value : true,
+  description : 'Build support for OpenGL ES 1.x'
+)
+option(
+  'gles2',
+  type : 'boolean',
+  value : true,
+  description : 'Build support for OpenGL ES 2.x and 3.x'
+)
+option(
+  'opengl',
+  type : 'boolean',
+  value : true,
+  description : 'Build support for OpenGL (all versions)'
+)
+option(
+  'gbm',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'Build support for gbm platform'
+)
+option(
+  'glx',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'disabled', 'dri', 'xlib', 'gallium-xlib'],
+  description : 'Build support for GLX platform'
+)
+option(
+  'egl',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'Build support for EGL platform'
+)
+option(
+  'glvnd',
+  type : 'boolean',
+  value : false,
+  description : 'Enable GLVND support.'
+)
+option(
+  'asm',
+  type : 'boolean',
+  value : true,
+  description : 'Build assembly code if possible'
+)
+option(
+   'glx-read-only-text',
+   type : 'boolean',
+   value : false,
+   description : 'Disable writable .text section on x86 (decreases performance)'
+)
+option(
+  'llvm',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'Build with LLVM support.'
+)
+option(
+  'shared-llvm',
+  type : 'boolean',
+  value : true,
+  description : 'Whether to link llvm shared or statically.'
+)
+option(
+  'valgrind',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'Build with valgrind support'
+)
+option(
+  'libunwind',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'Use libunwind for stack-traces'
+)
+option(
+  'lmsensors',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'Enable HUD lmsensors support.'
+)
+option(
+  'build-tests',
+  type : 'boolean',
+  value : false,
+  description : 'Build unit tests. Currently this will build *all* unit tests, which may build more than expected.'
+)
+option(
+  'selinux',
+  type : 'boolean',
+  value : false,
+  description : 'Build an SELinux-aware Mesa'
+)
+option(
+  'osmesa',
+  type : 'combo',
+  value : 'none',
+  choices : ['none', 'classic', 'gallium'],
+  description : 'Build OSmesa.'
+)
+option(
+  'osmesa-bits',
+  type : 'combo',
+  value : '8',
+  choices : ['8', '16', '32'],
+  description : 'Number of channel bits for OSMesa.'
+)
+option(
+  'swr-arches',
+  type : 'array',
+  value : ['avx', 'avx2'],
+  choices : ['avx', 'avx2', 'knl', 'skx'],
+  description : 'Architectures to build SWR support for.',
+)
+option(
+  'tools',
+  type : 'array',
+  value : [],
+  choices : ['freedreno', 'glsl', 'intel', 'intel-ui', 'nir', 'nouveau', 'xvmc', 'all'],
+  description : 'List of tools to build. (Note: `intel-ui` selects `intel`)',
+)
+option(
+  'power8',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'Enable power8 optimizations.',
+)
+option(
+  'xlib-lease',
+  type : 'combo',
+  value : 'auto',
+  choices : ['auto', 'true', 'false'],
+  description : 'Enable VK_EXT_acquire_xlib_display.'
+)
+option(
+  'glx-direct',
+  type : 'boolean',
+  value : true,
+  description : 'Enable direct rendering in GLX and EGL for DRI',
+)
diff --git a/lib/mesa/src/broadcom/Makefile.v3d.am b/lib/mesa/src/broadcom/Makefile.v3d.am
new file mode 100644
index 000000000..97ef2d745
--- /dev/null
+++ b/lib/mesa/src/broadcom/Makefile.v3d.am
@@ -0,0 +1,32 @@
+noinst_LTLIBRARIES += libbroadcom.la
+noinst_LTLIBRARIES += libbroadcom_v33.la
+noinst_LTLIBRARIES += libbroadcom_v41.la
+noinst_LTLIBRARIES += libbroadcom_v42.la
+
+if USE_V3D_SIMULATOR
+AM_CFLAGS += $(V3D_SIMULATOR_CFLAGS)
+libbroadcom_la_LDFLAGS = $(V3D_SIMULATOR_LIBS)
+endif
+
+libbroadcom_v33_la_SOURCES = $(BROADCOM_PER_VERSION_SOURCES)
+libbroadcom_v33_la_CFLAGS = -DV3D_VERSION=33
+
+libbroadcom_v41_la_SOURCES = $(BROADCOM_PER_VERSION_SOURCES)
+libbroadcom_v41_la_CFLAGS = -DV3D_VERSION=41
+
+libbroadcom_v42_la_SOURCES = $(BROADCOM_PER_VERSION_SOURCES)
+libbroadcom_v42_la_CFLAGS = -DV3D_VERSION=42
+
+libbroadcom_la_SOURCES = $(BROADCOM_FILES)
+
+check_PROGRAMS += \
+	qpu/tests/qpu_disasm \
+	$(NULL)
+
+LDADD = \
+	libbroadcom.la \
+	$(top_builddir)/src/compiler/nir/libnir.la \
+	$(top_builddir)/src/util/libmesautil.la \
+	$(NULL)
+
+TESTS += $(check_PROGRAMS)
diff --git a/lib/mesa/src/broadcom/cle/meson.build b/lib/mesa/src/broadcom/cle/meson.build
new file mode 100644
index 000000000..afaf5a1b4
--- /dev/null
+++ b/lib/mesa/src/broadcom/cle/meson.build
@@ -0,0 +1,63 @@
+# Copyright © 2017 Broadcom
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+# [version, cle XML version]
+v3d_versions = [
+  [21, 21],
+  [33, 33],
+  [41, 33],
+  [42, 33]
+]
+
+v3d_xml_files = []
+v3d_xml_pack = []
+foreach _v : v3d_versions
+  v = _v[0]
+  xmlver = _v[1]
+  f = 'v3d_packet_v@0@.xml'.format(xmlver)
+  _name = 'v3d_packet_v@0@_pack.h'.format(v)
+  if not v3d_xml_files.contains(f)
+    v3d_xml_files += f
+  endif
+  v3d_xml_pack += custom_target(
+    _name,
+    input : ['gen_pack_header.py', f],
+    output : _name,
+    command : [prog_python, '@INPUT@', '@0@'.format(v)],
+    capture : true,
+  )
+endforeach
+
+v3d_xml_h = custom_target(
+  'v3d_xml.h',
+  input : ['../../intel/genxml/gen_zipped_file.py', v3d_xml_files],
+  output : 'v3d_xml.h',
+  command : [prog_python, '@INPUT@'],
+  capture : true,
+)
+
+libbroadcom_cle = static_library(
+  ['broadcom_cle', v3d_xml_h],
+  'v3d_decoder.c',
+  include_directories : [inc_common, inc_broadcom],
+  c_args : [c_vis_args, no_override_init_args],
+  dependencies : [dep_libdrm, dep_valgrind],
+  build_by_default : false,
+)
diff --git a/lib/mesa/src/broadcom/cle/v3d_decoder.c b/lib/mesa/src/broadcom/cle/v3d_decoder.c
index 4ac40af05..373a1d996 100644
--- a/lib/mesa/src/broadcom/cle/v3d_decoder.c
+++ b/lib/mesa/src/broadcom/cle/v3d_decoder.c
@@ -37,6 +37,7 @@
 #include "v3d_decoder.h"
 #include "v3d_packet_helpers.h"
 #include "v3d_xml.h"
+#include "broadcom/clif/clif_private.h"
 
 struct v3d_spec {
         uint32_t ver;
@@ -58,6 +59,7 @@ struct location {
 
 struct parser_context {
         XML_Parser parser;
+        const struct v3d_device_info *devinfo;
         int foo;
         struct location loc;
 
@@ -68,6 +70,9 @@ struct parser_context {
         struct v3d_value *values[256];
 
         struct v3d_spec *spec;
+
+        int parse_depth;
+        int parse_skip_depth;
 };
 
 const char *
@@ -311,6 +316,8 @@ string_to_type(struct parser_context *ctx, const char *s)
                 return (struct v3d_type) { .kind = V3D_TYPE_BOOL };
         else if (strcmp(s, "float") == 0)
                 return (struct v3d_type) { .kind = V3D_TYPE_FLOAT };
+        else if (strcmp(s, "f187") == 0)
+                return (struct v3d_type) { .kind = V3D_TYPE_F187 };
         else if (strcmp(s, "address") == 0)
                 return (struct v3d_type) { .kind = V3D_TYPE_ADDRESS };
         else if (strcmp(s, "offset") == 0)
@@ -359,6 +366,9 @@ create_field(struct parser_context *ctx, const char **atts)
                 else if (strcmp(atts[i], "default") == 0) {
                         field->has_default = true;
                         field->default_value = strtoul(atts[i + 1], &p, 0);
+                } else if (strcmp(atts[i], "minus_one") == 0) {
+                        assert(strcmp(atts[i + 1], "true") == 0);
+                        field->minus_one = true;
                 }
         }
 
@@ -411,6 +421,25 @@ set_group_opcode(struct v3d_group *group, const char **atts)
         return;
 }
 
+static bool
+ver_in_range(int ver, int min_ver, int max_ver)
+{
+        return ((min_ver == 0 || ver >= min_ver) &&
+                (max_ver == 0 || ver <= max_ver));
+}
+
+static bool
+skip_if_ver_mismatch(struct parser_context *ctx, int min_ver, int max_ver)
+{
+        if (!ctx->parse_skip_depth && !ver_in_range(ctx->devinfo->ver,
+                                                    min_ver, max_ver)) {
+                assert(ctx->parse_depth != 0);
+                ctx->parse_skip_depth = ctx->parse_depth;
+        }
+
+        return ctx->parse_skip_depth;
+}
+
 static void
 start_element(void *data, const char *element_name, const char **atts)
 {
@@ -418,20 +447,35 @@ start_element(void *data, const char *element_name, const char **atts)
         int i;
         const char *name = NULL;
         const char *ver = NULL;
+        int min_ver = 0;
+        int max_ver = 0;
 
         ctx->loc.line_number = XML_GetCurrentLineNumber(ctx->parser);
 
         for (i = 0; atts[i]; i += 2) {
-                if (strcmp(atts[i], "name") == 0)
+                if (strcmp(atts[i], "shortname") == 0)
+                        name = atts[i + 1];
+                else if (strcmp(atts[i], "name") == 0 && !name)
                         name = atts[i + 1];
                 else if (strcmp(atts[i], "gen") == 0)
                         ver = atts[i + 1];
+                else if (strcmp(atts[i], "min_ver") == 0)
+                        min_ver = strtoul(atts[i + 1], NULL, 0);
+                else if (strcmp(atts[i], "max_ver") == 0)
+                        max_ver = strtoul(atts[i + 1], NULL, 0);
         }
 
+        if (skip_if_ver_mismatch(ctx, min_ver, max_ver))
+                goto skip;
+
         if (strcmp(element_name, "vcxml") == 0) {
                 if (ver == NULL)
                         fail(&ctx->loc, "no ver given");
 
+                /* Make sure that we picked an XML that matched our version.
+                 */
+                assert(ver_in_range(ctx->devinfo->ver, min_ver, max_ver));
+
                 int major, minor;
                 int n = sscanf(ver, "%d.%d", &major, &minor);
                 if (n == 0)
@@ -467,6 +511,15 @@ start_element(void *data, const char *element_name, const char **atts)
                 assert(ctx->nvalues < ARRAY_SIZE(ctx->values));
         }
 
+skip:
+        ctx->parse_depth++;
+}
+
+static int
+field_offset_compare(const void *a, const void *b)
+{
+        return ((*(const struct v3d_field **)a)->start -
+                (*(const struct v3d_field **)b)->start);
 }
 
 static void
@@ -475,6 +528,14 @@ end_element(void *data, const char *name)
         struct parser_context *ctx = data;
         struct v3d_spec *spec = ctx->spec;
 
+        ctx->parse_depth--;
+
+        if (ctx->parse_skip_depth) {
+                if (ctx->parse_skip_depth == ctx->parse_depth)
+                        ctx->parse_skip_depth = 0;
+                return;
+        }
+
         if (strcmp(name, "packet") == 0 ||
             strcmp(name, "struct") == 0 ||
             strcmp(name, "register") == 0) {
@@ -499,6 +560,13 @@ end_element(void *data, const char *name)
                 else if (strcmp(name, "register") == 0)
                         spec->registers[spec->nregisters++] = group;
 
+                /* Sort the fields in increasing offset order.  The XML might
+                 * be specified in any order, but we'll want to iterate from
+                 * the bottom.
+                 */
+                qsort(group->fields, group->nfields, sizeof(*group->fields),
+                      field_offset_compare);
+
                 assert(spec->ncommands < ARRAY_SIZE(spec->commands));
                 assert(spec->nstructs < ARRAY_SIZE(spec->structs));
                 assert(spec->nregisters < ARRAY_SIZE(spec->registers));
@@ -586,10 +654,14 @@ v3d_spec_load(const struct v3d_device_info *devinfo)
         uint32_t text_offset = 0, text_length = 0, total_length;
 
         for (int i = 0; i < ARRAY_SIZE(genxml_files_table); i++) {
-                if (genxml_files_table[i].gen_10 == devinfo->ver) {
+                if (i != 0) {
+                        assert(genxml_files_table[i - 1].gen_10 <
+                               genxml_files_table[i].gen_10);
+                }
+
+                if (genxml_files_table[i].gen_10 <= devinfo->ver) {
                         text_offset = genxml_files_table[i].offset;
                         text_length = genxml_files_table[i].length;
-                        break;
                 }
         }
 
@@ -600,6 +672,7 @@ v3d_spec_load(const struct v3d_device_info *devinfo)
 
         memset(&ctx, 0, sizeof ctx);
         ctx.parser = XML_ParserCreate(NULL);
+        ctx.devinfo = devinfo;
         XML_SetUserData(ctx.parser, &ctx);
         if (ctx.parser == NULL) {
                 fprintf(stderr, "failed to create parser\n");
@@ -687,14 +760,12 @@ v3d_group_get_length(struct v3d_group *group)
 void
 v3d_field_iterator_init(struct v3d_field_iterator *iter,
                         struct v3d_group *group,
-                        const uint8_t *p,
-                        bool print_colors)
+                        const uint8_t *p)
 {
         memset(iter, 0, sizeof(*iter));
 
         iter->group = group;
         iter->p = p;
-        iter->print_colors = print_colors;
 }
 
 static const char *
@@ -763,7 +834,7 @@ iter_advance_field(struct v3d_field_iterator *iter)
 
         iter->field = iter->group->fields[iter->field_iter++];
         if (iter->field->name)
-                strncpy(iter->name, iter->field->name, sizeof(iter->name));
+                snprintf(iter->name, sizeof(iter->name), "%s", iter->field->name);
         else
                 memset(iter->name, 0, sizeof(iter->name));
         iter->offset = iter_group_offset_bits(iter, iter->group_iter) / 8 +
@@ -774,43 +845,76 @@ iter_advance_field(struct v3d_field_iterator *iter)
 }
 
 bool
-v3d_field_iterator_next(struct v3d_field_iterator *iter)
+v3d_field_iterator_next(struct clif_dump *clif, struct v3d_field_iterator *iter)
 {
         if (!iter_advance_field(iter))
                 return false;
 
         const char *enum_name = NULL;
 
-        int s = iter->field->start;
-        int e = iter->field->end;
+        int group_member_offset =
+                iter_group_offset_bits(iter, iter->group_iter);
+        int s = group_member_offset + iter->field->start;
+        int e = group_member_offset + iter->field->end;
+
+        assert(!iter->field->minus_one ||
+               iter->field->type.kind == V3D_TYPE_INT ||
+               iter->field->type.kind == V3D_TYPE_UINT);
 
         switch (iter->field->type.kind) {
         case V3D_TYPE_UNKNOWN:
         case V3D_TYPE_INT: {
                 uint32_t value = __gen_unpack_sint(iter->p, s, e);
+                if (iter->field->minus_one)
+                        value++;
                 snprintf(iter->value, sizeof(iter->value), "%d", value);
                 enum_name = v3d_get_enum_name(&iter->field->inline_enum, value);
                 break;
         }
         case V3D_TYPE_UINT: {
                 uint32_t value = __gen_unpack_uint(iter->p, s, e);
+                if (iter->field->minus_one)
+                        value++;
+                if (strcmp(iter->field->name, "Vec size") == 0 && value == 0)
+                        value = 1 << (e - s);
                 snprintf(iter->value, sizeof(iter->value), "%u", value);
                 enum_name = v3d_get_enum_name(&iter->field->inline_enum, value);
                 break;
         }
-        case V3D_TYPE_BOOL: {
-                const char *true_string =
-                        iter->print_colors ? "\e[0;35mtrue\e[0m" : "true";
+        case V3D_TYPE_BOOL:
                 snprintf(iter->value, sizeof(iter->value), "%s",
                          __gen_unpack_uint(iter->p, s, e) ?
-                         true_string : "false");
+                         "1 /* true */" : "0 /* false */");
                 break;
-        }
         case V3D_TYPE_FLOAT:
                 snprintf(iter->value, sizeof(iter->value), "%f",
                          __gen_unpack_float(iter->p, s, e));
                 break;
-        case V3D_TYPE_ADDRESS:
+
+        case V3D_TYPE_F187:
+                snprintf(iter->value, sizeof(iter->value), "%f",
+                         __gen_unpack_f187(iter->p, s, e));
+                break;
+
+        case V3D_TYPE_ADDRESS: {
+                uint32_t addr =
+                        __gen_unpack_uint(iter->p, s, e) << (31 - (e - s));
+                struct clif_bo *bo = clif_lookup_bo(clif, addr);
+                if (bo) {
+                        snprintf(iter->value, sizeof(iter->value),
+                                 "[%s+0x%08x] /* 0x%08x */",
+                                 bo->name, addr - bo->offset, addr);
+                } else if (addr) {
+                        snprintf(iter->value, sizeof(iter->value),
+                                 "/* XXX: BO unknown */ 0x%08x", addr);
+                } else {
+                        snprintf(iter->value, sizeof(iter->value),
+                                 "[null]");
+                }
+
+                break;
+        }
+
         case V3D_TYPE_OFFSET:
                 snprintf(iter->value, sizeof(iter->value), "0x%08"PRIx64,
                          __gen_unpack_uint(iter->p, s, e) << (31 - (e - s)));
@@ -823,14 +927,24 @@ v3d_field_iterator_next(struct v3d_field_iterator *iter)
                                              iter->field->type.v3d_struct->name);
                 break;
         case V3D_TYPE_SFIXED:
-                snprintf(iter->value, sizeof(iter->value), "%f",
-                         __gen_unpack_sfixed(iter->p, s, e,
-                                             iter->field->type.f));
+                if (clif->pretty) {
+                        snprintf(iter->value, sizeof(iter->value), "%f",
+                                 __gen_unpack_sfixed(iter->p, s, e,
+                                                     iter->field->type.f));
+                } else {
+                        snprintf(iter->value, sizeof(iter->value), "%u",
+                                 (unsigned)__gen_unpack_uint(iter->p, s, e));
+                }
                 break;
         case V3D_TYPE_UFIXED:
-                snprintf(iter->value, sizeof(iter->value), "%f",
-                         __gen_unpack_ufixed(iter->p, s, e,
-                                             iter->field->type.f));
+                if (clif->pretty) {
+                        snprintf(iter->value, sizeof(iter->value), "%f",
+                                 __gen_unpack_ufixed(iter->p, s, e,
+                                                     iter->field->type.f));
+                } else {
+                        snprintf(iter->value, sizeof(iter->value), "%u",
+                                 (unsigned)__gen_unpack_uint(iter->p, s, e));
+                }
                 break;
         case V3D_TYPE_MBO:
                 break;
@@ -851,26 +965,40 @@ v3d_field_iterator_next(struct v3d_field_iterator *iter)
         if (enum_name) {
                 int length = strlen(iter->value);
                 snprintf(iter->value + length, sizeof(iter->value) - length,
-                         " (%s)", enum_name);
+                         " /* %s */", enum_name);
         }
 
         return true;
 }
 
 void
-v3d_print_group(FILE *outfile, struct v3d_group *group,
-                uint64_t offset, const uint8_t *p, bool color)
+v3d_print_group(struct clif_dump *clif, struct v3d_group *group,
+                uint64_t offset, const uint8_t *p)
 {
         struct v3d_field_iterator iter;
 
-        v3d_field_iterator_init(&iter, group, p, color);
-        while (v3d_field_iterator_next(&iter)) {
-                fprintf(outfile, "    %s: %s\n", iter.name, iter.value);
+        v3d_field_iterator_init(&iter, group, p);
+        while (v3d_field_iterator_next(clif, &iter)) {
+                /* Clif parsing uses the packet name, and expects no
+                 * sub-id.
+                 */
+                if (strcmp(iter.field->name, "sub-id") == 0 ||
+                    strcmp(iter.field->name, "unused") == 0 ||
+                    strcmp(iter.field->name, "Pad") == 0)
+                        continue;
+
+                if (clif->pretty) {
+                        fprintf(clif->out, "    %s: %s\n",
+                                iter.name, iter.value);
+                } else {
+                        fprintf(clif->out, "  /* %30s: */ %s\n",
+                                iter.name, iter.value);
+                }
                 if (iter.struct_desc) {
                         uint64_t struct_offset = offset + iter.offset;
-                        v3d_print_group(outfile, iter.struct_desc,
+                        v3d_print_group(clif, iter.struct_desc,
                                         struct_offset,
-                                        &p[iter.offset], color);
+                                        &p[iter.offset]);
                 }
         }
 }
diff --git a/lib/mesa/src/broadcom/cle/v3d_decoder.h b/lib/mesa/src/broadcom/cle/v3d_decoder.h
index 541d877a9..b5ead383b 100644
--- a/lib/mesa/src/broadcom/cle/v3d_decoder.h
+++ b/lib/mesa/src/broadcom/cle/v3d_decoder.h
@@ -34,6 +34,7 @@
 struct v3d_spec;
 struct v3d_group;
 struct v3d_field;
+struct clif_dump;
 
 struct v3d_group *v3d_spec_find_struct(struct v3d_spec *spec, const char *name);
 struct v3d_spec *v3d_spec_load(const struct v3d_device_info *devinfo);
@@ -57,7 +58,6 @@ struct v3d_field_iterator {
         int group_iter;
 
         struct v3d_field *field;
-        bool print_colors;
 };
 
 struct v3d_group {
@@ -99,6 +99,7 @@ struct v3d_type {
                 V3D_TYPE_UINT,
                 V3D_TYPE_BOOL,
                 V3D_TYPE_FLOAT,
+                V3D_TYPE_F187,
                 V3D_TYPE_ADDRESS,
                 V3D_TYPE_OFFSET,
                 V3D_TYPE_STRUCT,
@@ -125,6 +126,7 @@ struct v3d_field {
         char *name;
         int start, end;
         struct v3d_type type;
+        bool minus_one;
         bool has_default;
         uint32_t default_value;
 
@@ -133,14 +135,13 @@ struct v3d_field {
 
 void v3d_field_iterator_init(struct v3d_field_iterator *iter,
                              struct v3d_group *group,
-                             const uint8_t *p,
-                             bool print_colors);
+                             const uint8_t *p);
 
-bool v3d_field_iterator_next(struct v3d_field_iterator *iter);
+bool v3d_field_iterator_next(struct clif_dump *clif,
+                             struct v3d_field_iterator *iter);
 
-void v3d_print_group(FILE *out,
+void v3d_print_group(struct clif_dump *clif,
                      struct v3d_group *group,
-                     uint64_t offset, const uint8_t *p,
-                     bool color);
+                     uint64_t offset, const uint8_t *p);
 
 #endif /* V3D_DECODER_H */
diff --git a/lib/mesa/src/broadcom/cle/v3d_packet_v33.xml b/lib/mesa/src/broadcom/cle/v3d_packet_v33.xml
index e6562a262..f471d542c 100644
--- a/lib/mesa/src/broadcom/cle/v3d_packet_v33.xml
+++ b/lib/mesa/src/broadcom/cle/v3d_packet_v33.xml
@@ -1,4 +1,4 @@
-<vcxml gen="3.3">
+<vcxml gen="3.3" min_ver="33" max_ver="42">
 
   <enum name="Compare Function" prefix="V3D_COMPARE_FUNC">
     <value name="NEVER" value="0"/>
@@ -69,24 +69,195 @@
     <value name="TRIANGLE_FAN_TF" value="22"/>
   </enum>
 
+  <enum name="TMU Filter" prefix="V3D_TMU_FILTER" max_ver="33">
+    <!-- Names are mip filter, min filter, mag filter -->
+    <value name="MIN_LIN_MIP_NONE_MAG_LIN" value="0"/>
+    <value name="MIN_LIN_MIP_NONE_MAG_NEAR" value="1"/>
+    <value name="MIN_NEAR_MIP_NONE_MAG_LIN" value="2"/>
+    <value name="MIN_NEAR_MIP_NONE_MAG_NEAR" value="3"/>
+
+    <value name="MIN_NEAR_MIP_NEAR_MAG_LIN" value="4"/>
+    <value name="MIN_NEAR_MIP_NEAR_MAG_NEAR" value="5"/>
+    <value name="MIN_NEAR_MIP_LIN_MAG_LIN" value="6"/>
+    <value name="MIN_NEAR_MIP_LIN_MAG_NEAR" value="7"/>
+
+    <value name="MIN_LIN_MIP_NEAR_MAG_LIN" value="8"/>
+    <value name="MIN_LIN_MIP_NEAR_MAG_NEAR" value="9"/>
+    <value name="MIN_LIN_MIP_LIN_MAG_LIN" value="10"/>
+    <value name="MIN_LIN_MIP_LIN_MAG_NEAR" value="11"/>
+
+    <value name="ANISOTROPIC_2_1" value="12"/>
+    <value name="ANISOTROPIC_4_1" value="13"/>
+    <value name="ANISOTROPIC_8_1" value="14"/>
+    <value name="ANISOTROPIC_16_1" value="15"/>
+  </enum>
+
+  <enum name="Border Color Mode" prefix="V3D_BORDER_COLOR" min_ver="41">
+    <value name="0000" value="0"/>
+    <value name="0001" value="1"/>
+    <value name="1111" value="2"/>
+    <value name="Follows" value="7"/>
+  </enum>
+
+  <enum name="Wrap Mode" prefix="V3D_WRAP_MODE" min_ver="41">
+      <value name="Wrap mode REPEAT" value="0"/>
+      <value name="Wrap mode CLAMP" value="1"/>
+      <value name="Wrap mode MIRROR" value="2"/>
+      <value name="Wrap mode BORDER" value="3"/>
+      <value name="Wrap mode MIRROR_ONCE" value="4"/>
+  </enum>
+
+  <enum name="TMU Op" prefix="V3D_TMU_OP" min_ver="41">
+    <value name="Write ADD, Read Prefetch" value="0"/>
+    <value name="Write SUB, Read Clear" value="1"/>
+    <value name="Write XCHG, Read Flush" value="2"/>
+    <value name="Write CMPXCHG, Read Flush" value="3"/>
+    <value name="Write UMIN, Full L1 Clear" value="4"/>
+    <value name="Write UMAX" value="5"/>
+    <value name="Write SMIN" value="6"/>
+    <value name="Write SMAX" value="7"/>
+    <value name="Write AND, Read INC" value="8"/>
+    <value name="Write OR, Read DEC" value="9"/>
+    <value name="Write XOR, Read NOT" value="10"/>
+    <value name="Regular" value="15"/>
+  </enum>
+
+  <enum name="Varying Flags Action" prefix="V3D_VARYING_FLAGS_ACTION">
+    <value name="unchanged" value="0"/>
+    <value name="zeroed" value="1"/>
+    <value name="set" value="2"/>
+  </enum>
+
+  <enum name="Memory Format" prefix="V3D_MEMORY_FORMAT">
+    <value name="Raster" value="0"/>
+    <value name="Lineartile" value="1"/>
+    <value name="UB-linear (1 UIF block wide)" value="2"/>
+    <value name="UB-linear (2 UIF blocks wide)" value="3"/>
+    <value name="UIF (No XOR)" value="4"/>
+    <value name="UIF (XOR)" value="5"/>
+  </enum>
+
+  <enum name="Decimate Mode" prefix="V3D_DECIMATE_MODE">
+    <value name="sample 0" value="0"/>
+    <value name="4x" value="1"/>
+    <value name="all samples" value="3"/>
+  </enum>
+
+  <enum name="Internal Type" prefix="V3D_INTERNAL_TYPE">
+    <value name="8i" value="0"/>
+    <value name="8ui" value="1"/>
+    <value name="8" value="2"/>
+    <value name="16i" value="4"/>
+    <value name="16ui" value="5"/>
+    <value name="16f" value="6"/>
+    <value name="32i" value="8"/>
+    <value name="32ui" value="9"/>
+    <value name="32f" value="10"/>
+  </enum>
+
+  <enum name="Internal BPP" prefix="V3D_INTERNAL_BPP">
+    <value name="32" value="0"/>
+    <value name="64" value="1"/>
+    <value name="128" value="2"/>
+  </enum>
+
+  <enum name="Internal Depth Type" prefix="V3D_INTERNAL_TYPE">
+    <value name="depth_32f" value="0"/>
+    <value name="depth_24" value="1"/>
+    <value name="depth_16" value="2"/>
+  </enum>
+
+  <enum name="Render Target Clamp" prefix="V3D_RENDER_TARGET_CLAMP" min_ver="41">
+    <value name="none" value="0"/> <!-- no clamping -->
+    <value name="norm" value="1"/> <!-- [0,1] for f16 -->
+    <value name="pos" value="2"/> <!-- [0, for f16 -->
+    <value name="int" value="3" min_ver="42"/> <!-- clamp to integer RT's range -->
+  </enum>
+
+  <enum name="Output Image Format" prefix="V3D_OUTPUT_IMAGE_FORMAT">
+    <!--
+	Formats appear with their channels named from the low bits to
+	the high bits.
+    -->
+    <value name="srgb8_alpha8" value="0"/>
+    <value name="srgb" value="1"/>
+    <value name="rgb10_a2ui" value="2"/>
+    <value name="rgb10_a2" value="3"/>
+    <value name="abgr1555" value="4"/>
+    <value name="alpha-masked abgr1555" value="5"/>
+    <value name="abgr4444" value="6"/>
+    <value name="bgr565" value="7"/>
+    <value name="r11f_g11f_b10f" value="8"/>
+    <value name="rgba32f"  value="9"/>
+    <value name="rg32f"    value="10"/>
+    <value name="r32f"     value="11"/>
+    <value name="rgba32i"  value="12"/>
+    <value name="rg32i"    value="13"/>
+    <value name="r32i"     value="14"/>
+    <value name="rgba32ui" value="15"/>
+    <value name="rg32ui"   value="16"/>
+    <value name="r32ui"    value="17"/>
+    <value name="rgba16f"  value="18"/>
+    <value name="rg16f"    value="19"/>
+    <value name="r16f"     value="20"/>
+    <value name="rgba16i"  value="21"/>
+    <value name="rg16i"    value="22"/>
+    <value name="r16i"     value="23"/>
+    <value name="rgba16ui" value="24"/>
+    <value name="rg16ui"   value="25"/>
+    <value name="r16ui"    value="26"/>
+    <value name="rgba8"    value="27"/>
+    <value name="rgb8"     value="28"/>
+    <value name="rg8"      value="29"/>
+    <value name="r8"       value="30"/>
+    <value name="rgba8i"   value="31"/>
+    <value name="rg8i"     value="32"/>
+    <value name="r8i"      value="33"/>
+    <value name="rgba8ui"  value="34"/>
+    <value name="rg8ui"    value="35"/>
+    <value name="r8ui"     value="36"/>
+    <value name="srgbx8"   value="37" max_ver="33"/>
+    <value name="rgbx8"    value="38" max_ver="33"/>
+    <value name="bstc"     value="39" min_ver="41"/>
+    <value name="d32f"     value="40" min_ver="41"/>
+    <value name="d24"      value="41" min_ver="41"/>
+    <value name="d16"      value="42" min_ver="41"/>
+    <value name="d24s8"    value="43" min_ver="41"/>
+    <value name="s8"       value="44" min_ver="41"/>
+  </enum>
+
+  <enum name="Z/S Output Image Format" prefix="V3D_OUTPUT_IMAGE_FORMAT_ZS" max_ver="33">
+    <value name="depth_component32f" value="0"/>
+    <value name="depth_component24" value="1"/> <!-- depth low, pad high -->
+    <value name="depth_component16" value="2"/>
+    <value name="depth24_stencil8" value="3"/> <!-- stencil low, depth high -->
+  </enum>
+
+  <enum name="Dither Mode" prefix="V3D_DITHER_MODE">
+    <value name="None" value="0"/>
+    <value name="RGB" value="1"/>
+    <value name="A" value="2"/>
+    <value name="RGBA" value="3"/>
+  </enum>
+
   <packet code="0" name="Halt"/>
   <packet code="1" name="NOP"/>
   <packet code="4" name="Flush"/>
   <packet code="5" name="Flush All State"/>
   <packet code="6" name="Start Tile Binning"/>
-  <packet code="7" name="Increment Semaphore"/>
-  <packet code="8" name="Wait on Semaphore"/>
-  <packet code="9" name="Wait for previous frame"/>
-  <packet code="10" name="Enable Z-only rendering" cl="R"/>
-  <packet code="11" name="Disable Z-only rendering" cl="R"/>
-  <packet code="12" name="End of Z-only rendering in frame"/>
-  <packet code="13" name="End of rendering"/>
-
-  <packet code="14" name="Wait for transform feedback" cl="B">
+  <packet code="7" shortname="incr_semaphore" name="Increment Semaphore"/>
+  <packet code="8" shortname="wait_semaphore" name="Wait on Semaphore"/>
+  <packet code="9" shortname="wait_prev_frame" name="Wait for previous frame"/>
+  <packet code="10" shortname="enable_z_only" name="Enable Z-only rendering" cl="R"/>
+  <packet code="11" shortname="disable_z_only" name="Disable Z-only rendering" cl="R"/>
+  <packet code="12" shortname="end_z_only" name="End of Z-only rendering in frame"/>
+  <packet code="13" shortname="end_render" name="End of rendering"/>
+
+  <packet code="14" shortname="wait_transform_feedback" name="Wait for transform feedback" cl="B">
     <field name="Block count" size="8" start="0" type="uint"/>
   </packet>
 
-  <packet code="15" name="Branch to auto-chained sub-list">
+  <packet code="15" shortname="branch_sub_autochain" name="Branch to auto-chained sub-list">
     <field name="address" size="32" start="0" type="address"/>
   </packet>
 
@@ -94,62 +265,69 @@
     <field name="address" size="32" start="0" type="address"/>
   </packet>
 
-  <packet code="17" name="Branch to Sub-list">
+  <packet code="17" shortname="branch_sub" name="Branch to Sub-list">
     <field name="address" size="32" start="0" type="address"/>
   </packet>
 
-  <packet code="18" name="Return from sub-list"/>
-  <packet code="19" name="Flush VCD cache"/>
+  <packet code="18" shortname="return" name="Return from sub-list"/>
+  <packet code="19" shortname="clear_vcd_cache" name="Flush VCD cache"/>
 
-  <packet code="20" name="Start Address of Generic Tile List">
+  <packet code="20" shortname="generic_tile_list" name="Start Address of Generic Tile List">
     <field name="start" size="32" start="0" type="address"/>
     <field name="end" size="32" start="32" type="address"/>
   </packet>
 
-  <packet code="21" name="Branch to Implicit Tile List">
+  <packet code="21" shortname="branch_implicit_tile" name="Branch to Implicit Tile List">
     <field name="tile list set number" size="8" start="0" type="uint"/>
   </packet>
 
-  <packet code="22" name="Branch to Explicit Supertile">
+  <packet code="22" shortname="branch_explicit_supertile" name="Branch to Explicit Supertile">
     <field name="Absolute address of explicit supertile render list" size="32" start="24" type="address"/>
     <field name="explicit supertile number" size="8" start="16" type="uint"/>
     <field name="row number" size="8" start="8" type="uint"/>
     <field name="column number" size="8" start="0" type="uint"/>
   </packet>
 
-  <packet code="23" name="Supertile Coordinates">
+  <packet code="23" shortname="supertile_coords" name="Supertile Coordinates">
     <field name="row number in supertiles" size="8" start="8" type="uint"/>
     <field name="column number in supertiles" size="8" start="0" type="uint"/>
   </packet>
 
-  <packet code="24" name="Store Multi-Sample Resolved Tile Color Buffer" cl="R"/>
+  <packet code="24" shortname="store_subsample" name="Store Multi-Sample Resolved Tile Color Buffer" cl="R" max_ver="33"/>
 
-  <packet code="25" name="Store Multi-Sample Resolved Tile Color Buffer (extended)" cl="R">
+  <packet code="25" shortname="store_subsample_ex" name="Store Multi-Sample Resolved Tile Color Buffer (extended)" cl="R" max_ver="33">
     <field name="Disable Color Buffer write" size="8" start="8" type="uint"/>
     <field name="Enable Z write" size="1" start="7" type="bool"/>
     <field name="Enable Stencil write" size="1" start="6" type="bool"/>
     <!-- bit 5 unused -->
-    <field name="Disable Colour buffer(s) clear on write" size="1" start="4" type="bool"/>
+    <field name="Disable Color buffer(s) clear on write" size="1" start="4" type="bool"/>
     <field name="Disable Stencil buffer clear on write" size="1" start="3" type="bool"/>
     <field name="Disable Z buffer clear on write" size="1" start="2" type="bool"/>
     <field name="Disable fast opportunistic write out in multisample mode" size="1" start="1" type="bool"/>
     <field name="Last Tile of Frame" size="1" start="0" type="bool"/>
   </packet>
 
-  <packet code="26" name="Reload Tile Colour Buffer" cl="R">
-    <field name="Disable Colour Buffer load" size="8" start="8" type="uint"/>
+  <packet code="25" shortname="clear" name="Clear Tile Buffers" cl="R" min_ver="41">
+    <field name="Clear Z/Stencil Buffer" size="1" start="1" type="bool"/>
+    <field name="Clear all Render Targets" size="1" start="0" type="bool"/>
+  </packet>
+
+  <packet code="26" shortname="load" name="Reload Tile Color Buffer" cl="R" max_ver="33">
+    <field name="Disable Color Buffer load" size="8" start="8" type="uint"/>
     <field name="Enable Z load" size="1" start="7" type="bool"/>
     <field name="Enable Stencil load" size="1" start="6" type="bool"/>
   </packet>
 
-  <packet code="27" name="End of Tile Marker" cl="R"/>
+  <packet code="26" shortname="end_loads" name="End of Loads" cl="R" min_ver="41"/>
 
-  <packet code="29" name="Store Tile Buffer General" cl="R">
-    <field name="Address" size="32" start="16" type="address"/>
+  <packet code="27" shortname="end_tile" name="End of Tile Marker" cl="R"/>
+
+  <packet code="29" shortname="store_general" name="Store Tile Buffer General" cl="R" max_ver="33">
+    <field name="Address" size="24" start="24" type="address"/>
     <field name="Padded height of output image in UIF blocks" size="13" start="11" type="uint"/>
     <field name="XOR UIF" size="1" start="10" type="bool"/>
     <field name="Last Tile of Frame" size="1" start="8" type="bool"/>
-    <field name="Disable Colour buffer(s) clear on write" size="1" start="7" type="bool"/>
+    <field name="Disable Color buffer(s) clear on write" size="1" start="7" type="bool"/>
     <field name="Disable Stencil buffer clear on write" size="1" start="6" type="bool"/>
     <field name="Disable Z buffer clear on write" size="1" start="5" type="bool"/>
     <field name="Raw Mode" size="1" start="4" type="bool"/>
@@ -165,7 +343,40 @@
     </field>
   </packet>
 
-  <packet code="30" name="Load Tile Buffer General" cl="R">
+  <packet code="29" shortname="store" name="Store Tile Buffer General" cl="R" min_ver="41">
+    <field name="Address" size="32" start="64" type="address"/>
+
+    <!-- used for y flip -->
+    <field name="Height" size="16" start="48" type="uint"/>
+
+    <!-- height in ub for UIF, byte stride for raster -->
+    <field name="Height in UB or Stride" size="20" start="28" type="uint"/>
+
+    <field name="R/B swap" size="1" start="20" type="bool"/>
+    <field name="Channel Reverse" size="1" start="19" type="bool"/>
+    <field name="Clear buffer being stored" size="1" start="18" type="bool"/>
+    <field name="Output Image Format" size="6" start="12" type="Output Image Format"/>
+
+    <field name="Decimate mode" size="2" start="10" type="Decimate Mode"/>
+
+    <field name="Dither Mode" size="2" start="8" type="Dither Mode"/>
+
+    <field name="Flip Y" size="1" start="7" type="bool"/>
+
+    <field name="Memory Format" size="3" start="4" type="Memory Format"/>
+    <field name="Buffer to Store" size="4" start="0" type="uint">
+      <value name="Render target 0" value="0"/>
+      <value name="Render target 1" value="1"/>
+      <value name="Render target 2" value="2"/>
+      <value name="Render target 3" value="3"/>
+      <value name="None" value="8"/>
+      <value name="Z" value="9"/>
+      <value name="Stencil" value="10"/>
+      <value name="Z+Stencil" value="11"/>
+    </field>
+  </packet>
+
+  <packet code="30" shortname="load_general" name="Load Tile Buffer General" cl="R" max_ver="33">
     <field name="Address" size="24" start="24" type="address"/>
     <field name="Padded height of output image in UIF blocks" size="13" start="11" type="uint"/>
     <field name="XOR UIF" size="1" start="10" type="bool"/>
@@ -182,7 +393,40 @@
     </field>
   </packet>
 
-  <packet code="32" name="Indexed Primitive List" cl="B">
+  <packet code="30" shortname="load" name="Load Tile Buffer General" cl="R" min_ver="41">
+    <field name="Address" size="32" start="64" type="address"/>
+
+    <!-- used for y flip -->
+    <field name="Height" size="16" start="48" type="uint"/>
+
+    <!-- height in ub for UIF, byte stride for raster -->
+    <field name="Height in UB or Stride" size="20" start="28" type="uint"/>
+
+    <field name="R/B swap" size="1" start="20" type="bool"/>
+    <field name="Channel Reverse" size="1" start="19" type="bool"/>
+
+    <field name="Input Image Format" size="6" start="12" type="Output Image Format"/>
+
+    <field name="Decimate mode" size="2" start="10" type="Decimate Mode"/>
+
+    <field name="Flip Y" size="1" start="7" type="bool"/>
+
+    <field name="Memory Format" size="3" start="4" type="Memory Format"/>
+    <field name="Buffer to Load" size="4" start="0" type="uint">
+      <value name="Render target 0" value="0"/>
+      <value name="Render target 1" value="1"/>
+      <value name="Render target 2" value="2"/>
+      <value name="Render target 3" value="3"/>
+      <value name="None" value="8"/>
+      <value name="Z" value="9"/>
+      <value name="Stencil" value="10"/>
+      <value name="Z+Stencil" value="11"/>
+    </field>
+  </packet>
+
+  <packet code="31" shortname="tf_draw_flush_and_count" name="Transform Feedback Flush and Count"/>
+
+  <packet code="32" name="Indexed Prim List" cl="B" max_ver="33">
     <field name="Minimum index" size="32" start="104" type="uint"/>
     <field name="Enable Primitive Restarts" size="1" start="103" type="bool"/>
     <field name="Maximum index" size="31" start="72" type="uint"/>
@@ -195,11 +439,25 @@
       <value name="Index type 32-bit" value="2"/>
     </field>
 
-    <field name="mode" size="5" start="0" type="uint">
+    <field name="mode" size="5" start="0" type="Primitive"/>
+  </packet>
+
+  <packet code="32" name="Indexed Prim List" cl="B" min_ver="41">
+    <field name="Index Offset" size="32" start="40" type="uint"/>
+
+    <field name="Enable Primitive Restarts" size="1" start="39" type="bool"/>
+    <field name="Length" size="31" start="8" type="uint"/>
+
+    <field name="Index type" size="2" start="6" type="uint">
+      <value name="Index type 8-bit" value="0"/>
+      <value name="Index type 16-bit" value="1"/>
+      <value name="Index type 32-bit" value="2"/>
     </field>
+
+    <field name="mode" size="6" start="0" type="Primitive"/>
   </packet>
 
-  <packet code="34" name="Indexed Instanced Primitive List" cl="B">
+  <packet code="34" name="Indexed Instanced Prim List" cl="B" max_ver="33">
     <field name="Enable Primitive Restarts" size="1" start="135" type="bool"/>
     <field name="Maximum index" size="31" start="104" type="uint"/>
     <field name="Address of Indices List" size="32" start="72" type="address"/>
@@ -215,14 +473,29 @@
     <field name="mode" size="5" start="0" type="Primitive"/>
   </packet>
 
-  <packet code="36" name="Vertex Array Primitives" cl="B">
+  <packet code="34" name="Indexed Instanced Prim List" cl="B" min_ver="41">
+    <field name="Index Offset" size="32" start="72" type="uint"/>
+    <field name="Number of Instances" size="32" start="40" type="uint"/>
+    <field name="Enable Primitive Restarts" size="1" start="39" type="bool"/>
+    <field name="Instance Length" size="31" start="8" type="uint"/>
+
+    <field name="Index type" size="2" start="6" type="uint">
+      <value name="Index type 8-bit" value="0"/>
+      <value name="Index type 16-bit" value="1"/>
+      <value name="Index type 32-bit" value="2"/>
+    </field>
+
+    <field name="mode" size="6" start="0" type="Primitive"/>
+  </packet>
+
+  <packet code="36" name="Vertex Array Prims" cl="B">
     <field name="Index of First Vertex" size="32" start="40" type="uint"/>
     <field name="Length" size="32" start="8" type="uint"/>
 
     <field name="mode" size="8" start="0" type="Primitive"/>
   </packet>
 
-  <packet code="38" name="Vertex Array Instanced Primitives" cl="B">
+  <packet code="38" name="Vertex Array Instanced Prims" cl="B">
     <field name="Index of First Vertex" size="32" start="72" type="uint"/>
     <field name="Number of Instances" size="32" start="40" type="uint"/>
     <field name="Instance Length" size="32" start="8" type="uint"/>
@@ -236,11 +509,13 @@
     <field name="Base Vertex" size="32" start="0" type="uint"/>
   </packet>
 
-  <packet code="56" name="Primitive List Format">
-    <field name="data type" size="1" start="6" type="uint">
-      <value name="List Indexed" value="0"/>
-      <value name="List 32-bit X/Y" value="1"/>
-    </field>
+  <packet code="44" name="Index Buffer Setup" cl="B" min_ver="41">
+    <field name="Address" size="32" start="0" type="address"/>
+    <field name="Size" size="32" start="32" type="uint"/>
+  </packet>
+
+  <packet code="56" name="Prim List Format">
+    <field name="tri strip or fan" size="1" start="7" type="bool"/>
     <field name="primitive type" size="6" start="0" type="uint">
       <value name="List Points" value="0"/>
       <value name="List Lines" value="1"/>
@@ -248,29 +523,57 @@
     </field>
   </packet>
 
-  <packet code="64" name="GL Shader State">
+  <packet code="64" shortname="gl_shader" name="GL Shader State">
     <field name="address" size="27" start="5" type="address"/>
     <field name="number of attribute arrays" size="5" start="0" type="uint"/>
   </packet>
 
-  <packet code="74" name="Transform Feedback Enable">
+  <packet code="71" name="VCM Cache Size" min_ver="41">
+    <field name="Number of 16-vertex batches for rendering" size="4" start="4" type="uint"/>
+    <field name="Number of 16-vertex batches for binning" size="4" start="0" type="uint"/>
+  </packet>
+
+  <packet code="73" name="VCM Cache Size" max_ver="33">
+    <field name="Number of 16-vertex batches for rendering" size="4" start="4" type="uint"/>
+    <field name="Number of 16-vertex batches for binning" size="4" start="0" type="uint"/>
+  </packet>
+
+  <packet code="73" name="Transform Feedback Buffer" min_ver="41">
+    <field name="Buffer Address" size="32" start="32" type="address"/>
+    <field name="Buffer Size in 32-bit words" size="30" start="2" type="uint"/>
+    <field name="Buffer Number" size="2" start="0" type="uint"/>
+  </packet>
+
+  <packet code="74" name="Transform Feedback Enable" max_ver="33">
     <field name="number of 32-bit Output Buffer Address following" size="3" start="8" type="uint"/>
     <field name="number of 16-bit Output Data Specs following" size="5" start="11" type="uint"/>
   </packet>
 
+  <packet code="74" name="Transform Feedback Specs" min_ver="41">
+    <field name="Enable" size="1" start="7" type="bool"/>
+    <field name="Number of 16-bit Output Data Specs following" size="5" start="0" type="uint"/>
+  </packet>
+
   <packet code="75" name="Flush Transform Feedback Data"/>
 
-  <struct name="Transform Feedback Output Data Spec">
+  <struct name="Transform Feedback Output Data Spec" max_ver="33">
+    <field name="First Shaded Vertex Value to output" size="8" start="0" type="uint"/>
+    <field name="Number of consecutive Vertex Values to output as 32-bit values" size="4" start="8" type="uint" minus_one="true"/>
+    <field name="Output Buffer to write to" size="2" start="12" type="uint"/>
+  </struct>
+
+  <struct name="Transform Feedback Output Data Spec" min_ver="41">
     <field name="First Shaded Vertex Value to output" size="8" start="0" type="uint"/>
-    <field name="Number of consecutive Vertex Values to output as 32-bit values minus 1" size="4" start="8" type="uint"/>
+    <field name="Number of consecutive Vertex Values to output as 32-bit values" size="4" start="8" type="uint" minus_one="true"/>
     <field name="Output Buffer to write to" size="2" start="12" type="uint"/>
+    <field name="Stream number" size="2" start="14" type="uint"/>
   </struct>
 
   <struct name="Transform Feedback Output Address">
     <field name="address" size="32" start="0" type="address"/>
   </struct>
 
-  <packet code="80" name="Stencil Config">
+  <packet code="80" name="Stencil Cfg">
     <field name="Stencil Write Mask" size="8" start="32" type="uint"/>
     <field name="Back Config" size="1" start="29" type="bool"/>
     <field name="Front Config" size="1" start="28" type="bool"/>
@@ -282,32 +585,59 @@
     <field name="Stencil Ref Value" size="8" start="0" type="uint"/>
   </packet>
 
-  <packet code="84" name="Blend Config">
-    <field name="VG Coverage Modes" size="2" start="28" type="uint"/>
-    <field name="Colour blend dst factor" size="4" start="20" type="Blend Factor"/>
-    <field name="Colour blend src factor" size="4" start="16" type="Blend Factor"/>
-    <field name="Colour blend mode" size="4" start="12" type="Blend Mode"/>
+  <packet code="83" name="Blend Enables" min_ver="41">
+    <field name="Mask" size="8" start="0" type="uint"/>
+  </packet>
+
+  <packet code="84" name="Blend Cfg" max_ver="33">
+    <field name="Color blend dst factor" size="4" start="20" type="Blend Factor"/>
+    <field name="Color blend src factor" size="4" start="16" type="Blend Factor"/>
+    <field name="Color blend mode" size="4" start="12" type="Blend Mode"/>
+    <field name="Alpha blend dst factor" size="4" start="8" type="Blend Factor"/>
+    <field name="Alpha blend src factor" size="4" start="4" type="Blend Factor"/>
+    <field name="Alpha blend mode" size="4" start="0" type="Blend Mode"/>
+  </packet>
+
+  <packet code="84" name="Blend Cfg" min_ver="41">
+    <field name="Render Target Mask" size="4" start="24" type="uint"/>
+    <field name="Color blend dst factor" size="4" start="20" type="Blend Factor"/>
+    <field name="Color blend src factor" size="4" start="16" type="Blend Factor"/>
+    <field name="Color blend mode" size="4" start="12" type="Blend Mode"/>
     <field name="Alpha blend dst factor" size="4" start="8" type="Blend Factor"/>
     <field name="Alpha blend src factor" size="4" start="4" type="Blend Factor"/>
     <field name="Alpha blend mode" size="4" start="0" type="Blend Mode"/>
   </packet>
 
-  <packet code="86" name="Blend Constant Colour">
+  <packet code="86" shortname="blend_ccolor" name="Blend Constant Color">
     <field name="Alpha (F16)" size="16" start="48" type="uint"/>
     <field name="Blue (F16)" size="16" start="32" type="uint"/>
     <field name="Green (F16)" size="16" start="16" type="uint"/>
     <field name="Red (F16)" size="16" start="0" type="uint"/>
   </packet>
 
-  <packet code="87" name="Colour Write Masks">
-    <field name="Reserved" size="16" start="16" type="uint"/>
-    <field name="Render Target 3 per colour component write masks" size="4" start="12" type="uint"/>
-    <field name="Render Target 2 per colour component write masks" size="4" start="8" type="uint"/>
-    <field name="Render Target 1 per colour component write masks" size="4" start="4" type="uint"/>
-    <field name="Render Target 0 per colour component write masks" size="4" start="0" type="uint"/>
+  <packet code="87" shortname="color_wmasks" name="Color Write Masks">
+    <field name="Mask" size="32" start="0" type="uint"/>
   </packet>
 
-  <packet code="96" name="Configuration Bits">
+  <packet code="88" name="Zero All Centroid Flags" min_ver="41"/>
+
+  <packet code="89" name="Centroid Flags" min_ver="41">
+    <field name="Centroid Flags for varyings V0*24" size="24" start="8" type="uint"/>
+    <field name="Action for Centroid Flags of higher numbered varyings" size="2" start="6" type="Varying Flags Action"/>
+    <field name="Action for Centroid Flags of lower numbered varyings" size="2" start="4" type="Varying Flags Action"/>
+    <field name="Varying offset V0" size="4" start="0" type="uint"/>
+  </packet>
+
+  <packet code="91" name="Sample State" min_ver="41">
+    <field name="Coverage" size="16" start="16" type="f187"/>
+    <field name="Mask" size="4" start="0" type="uint"/>
+  </packet>
+
+  <packet code="92" shortname="occlusion_query_counter_enable" name="Occlusion Query Counter">
+    <field name="address" size="32" start="0" type="address"/>
+  </packet>
+
+  <packet code="96" name="Cfg Bits">
     <field name="Direct3D Provoking Vertex" size="1" start="21" type="bool"/>
     <field name="Direct3D 'Point-fill' mode" size="1" start="20" type="bool"/>
     <field name="Blend enable" size="1" start="19" type="bool"/>
@@ -317,8 +647,6 @@
     <field name="Z updates enable" size="1" start="15" type="bool"/>
     <field name="Depth-Test Function" size="3" start="12" type="Compare Function"/>
     <field name="Direct3D Wireframe triangles mode" size="1" start="11" type="bool"/>
-    <field name="Coverage Update Mode" size="2" start="9" type="uint"/>
-    <field name="Coverage Pipe Select" size="1" start="8" type="bool"/>
     <field name="Rasterizer Oversample Mode" size="2" start="6" type="uint"/>
     <field name="Line Rasterization" size="2" start="4" type="uint"/>
     <field name="Enable Depth Offset" size="1" start="3" type="bool"/>
@@ -327,12 +655,21 @@
     <field name="Enable Forward Facing Primitive" size="1" start="0" type="bool"/>
   </packet>
 
-  <packet code="97" name="Zero All Flat Shade Flags"/>
+  <packet code="97" shortname="zero_all_flatshade_flags" name="Zero All Flat Shade Flags"/>
 
-  <packet code="98" name="Flat Shade Flags">
+  <packet code="98" shortname="flatshade_flags" name="Flat Shade Flags">
     <field name="Flat Shade Flags for varyings V0*24" size="24" start="8" type="uint"/>
-    <field name="Action for Flat Shade Flags of higher numbered varyings" size="2" start="6" type="uint"/>
-    <field name="Action for Flat Shade Flags of lower numbered varyings" size="2" start="4" type="uint"/>
+    <field name="Action for Flat Shade Flags of higher numbered varyings" size="2" start="6" type="Varying Flags Action"/>
+    <field name="Action for Flat Shade Flags of lower numbered varyings" size="2" start="4" type="Varying Flags Action"/>
+    <field name="Varying offset V0" size="4" start="0" type="uint"/>
+  </packet>
+
+  <packet code="99" shortname="zero_all_noperspective_flags" name="Zero All Non-perspective Flags" min_ver="41"/>
+
+  <packet code="100" shortname="noperspective_flags" name="Non-perspective Flags" min_ver="41">
+    <field name="Non-perspective Flags for varyings V0*24" size="24" start="8" type="uint"/>
+    <field name="Action for Non-perspective Flags of higher numbered varyings" size="2" start="6" type="Varying Flags Action"/>
+    <field name="Action for Non-perspective Flags of lower numbered varyings" size="2" start="4" type="Varying Flags Action"/>
     <field name="Varying offset V0" size="4" start="0" type="uint"/>
   </packet>
 
@@ -344,54 +681,66 @@
     <field name="Line width" size="32" start="0" type="float"/>
   </packet>
 
-  <packet name="Depth Offset" code="106">
-    <!-- these fields are both float-1-8-7 encoded (top 16 bits of a float32) -->
-    <field name="Depth Offset Units" size="16" start="16" type="uint"/>
-    <field name="Depth Offset Factor" size="16" start="0" type="uint"/>
+  <packet name="Depth Offset" code="106" max_ver="33">
+    <field name="Depth Offset Units" size="16" start="16" type="f187"/>
+    <field name="Depth Offset Factor" size="16" start="0" type="f187"/>
+  </packet>
+
+  <packet name="Depth Offset" code="106" min_ver="41">
+    <field name="Limit" size="32" start="32" type="float"/>
+    <field name="Depth Offset Units" size="16" start="16" type="f187"/>
+    <field name="Depth Offset Factor" size="16" start="0" type="f187"/>
   </packet>
 
-  <packet name="Clip Window" code="107">
+  <packet shortname="clip" name="clip_window" code="107">
     <field name="Clip Window Height in pixels" size="16" start="48" type="uint"/>
     <field name="Clip Window Width in pixels" size="16" start="32" type="uint"/>
     <field name="Clip Window Bottom Pixel Coordinate" size="16" start="16" type="uint"/>
     <field name="Clip Window Left Pixel Coordinate" size="16" start="0" type="uint"/>
   </packet>
 
-  <packet name="Viewport Offset" code="108">
+  <packet name="Viewport Offset" code="108" max_ver="33">
     <field name="Viewport Centre Y-coordinate" size="32" start="32" type="s24.8"/>
     <field name="Viewport Centre X-coordinate" size="32" start="0" type="s24.8"/>
   </packet>
 
-  <packet name="Clipper Z min/max clipping planes" code="109">
+  <packet name="Viewport Offset" code="108" min_ver="41">
+    <field name="Coarse Y" size="10" start="54" type="uint"/>
+    <field name="Viewport Centre Y-coordinate" size="22" start="32" type="s14.8"/>
+    <field name="Coarse X" size="10" start="22" type="uint"/>
+    <field name="Viewport Centre X-coordinate" size="22" start="0" type="s14.8"/>
+  </packet>
+
+  <packet shortname="clipz" name="Clipper Z min/max clipping planes" code="109">
     <field name="Maximum Zw" size="32" start="32" type="float"/>
     <field name="Minimum Zw" size="32" start="0" type="float"/>
   </packet>
 
-  <packet name="Clipper XY Scaling" code="110" cl="B">
+  <packet shortname="clipper_xy" name="Clipper XY Scaling" code="110" cl="B">
     <field name="Viewport Half-Height in 1/256th of pixel" size="32" start="32" type="float"/>
     <field name="Viewport Half-Width in 1/256th of pixel" size="32" start="0" type="float"/>
   </packet>
 
-  <packet name="Clipper Z Scale and Offset" code="111" cl="B">
+  <packet shortname="clipper_z" name="Clipper Z Scale and Offset" code="111" cl="B">
     <field name="Viewport Z Offset (Zc to Zs)" size="32" start="32" type="float"/>
     <field name="Viewport Z Scale (Zc to Zs)" size="32" start="0" type="float"/>
   </packet>
 
-  <packet code="120" name="Tile Binning Mode Configuration (Part1)">
+  <packet name="Number of Layers" code="119" min_ver="41">
+    <field name="Number of Layers" size="8" start="0" type="uint" minus_one="true"/>
+  </packet>
+
+  <packet code="120" name="Tile Binning Mode Cfg (Part1)" max_ver="33">
     <field name="Double-buffer in non-ms mode" size="1" start="63" type="bool"/>
     <field name="Multisample Mode (4x)" size="1" start="62" type="bool"/>
 
-    <field name="Maximum BPP of all render targets" size="2" start="60" type="uint">
-      <value name="Render target maximum 32bpp" value="0"/>
-      <value name="Render target maximum 64bpp" value="1"/>
-      <value name="Render target maximum 128bpp" value="2"/>
-    </field>
+    <field name="Maximum BPP of all render targets" size="2" start="60" type="Internal BPP"/>
 
     <field name="Number of Render Targets" size="4" start="56" type="uint"/>
     <field name="Height (in tiles)" size="12" start="44" type="uint"/>
     <field name="Width (in tiles)" size="12" start="32" type="uint"/>
 
-    <field name="Tile State Data Array Base Address" size="32" start="0" type="address"/>
+    <field name="Tile State Data Array Base Address" size="26" start="6" type="address"/>
 
     <field name="tile allocation block size" size="2" start="4" type="uint">
       <value name="tile allocation block size 64b" value="0"/>
@@ -407,14 +756,42 @@
     <field name="sub-id" size="1" start="0" type="uint" default="0"/>
   </packet>
 
-  <packet code="120" name="Tile Binning Mode Configuration (Part2)" cl="B">
+  <packet code="120" name="Tile Binning Mode Cfg" min_ver="41">
+
+    <field name="Height (in pixels)" size="12" start="48" type="uint" minus_one="true"/>
+    <field name="Width (in pixels)" size="12" start="32" type="uint" minus_one="true"/>
+
+    <field name="Double-buffer in non-ms mode" size="1" start="15" type="bool"/>
+    <field name="Multisample Mode (4x)" size="1" start="14" type="bool"/>
+
+    <field name="Maximum BPP of all render targets" size="2" start="12" type="uint">
+      <value name="Render target maximum 32bpp" value="0"/>
+      <value name="Render target maximum 64bpp" value="1"/>
+      <value name="Render target maximum 128bpp" value="2"/>
+    </field>
+
+    <field name="Number of Render Targets" size="4" start="8" type="uint" minus_one="true"/>
+
+    <field name="tile allocation block size" size="2" start="4" type="uint">
+      <value name="tile allocation block size 64b" value="0"/>
+      <value name="tile allocation block size 128b" value="1"/>
+      <value name="tile allocation block size 256b" value="2"/>
+    </field>
+    <field name="tile allocation initial block size" size="2" start="2" type="uint">
+      <value name="tile allocation initial block size 64b" value="0"/>
+      <value name="tile allocation initial block size 128b" value="1"/>
+      <value name="tile allocation initial block size 256b" value="2"/>
+    </field>
+  </packet>
+
+  <packet code="120" name="Tile Binning Mode Cfg (Part2)" cl="B" max_ver="33">
     <field name="Tile Allocation Memory Address" size="32" start="32" type="address"/>
     <field name="Tile Allocation Memory Size" size="32" start="0" type="uint"/>
 
     <field name="sub-id" size="1" start="0" type="uint" default="1"/>
   </packet>
 
-  <packet code="121" name="Tile Rendering Mode Configuration (Common Configuration)" cl="R">
+  <packet code="121" name="Tile Rendering Mode Cfg (Common)" cl="R" max_ver="33">
     <field name="Disable Render Target Stores" size="8" start="56" type="uint"/>
     <field name="Enable Z Store" size="1" start="55" type="bool"/>
     <field name="Enable Stencil Store" size="1" start="54" type="bool"/>
@@ -426,7 +803,6 @@
       <value name="Early-Z direction GT/GE" value="1"/>
     </field>
 
-    <field name="Select Coverage Mode" size="1" start="44" type="bool"/>
     <field name="Double-buffer in non-ms mode" size="1" start="43" type="bool"/>
     <field name="Multisample Mode (4x)" size="1" start="42" type="bool"/>
 
@@ -438,123 +814,92 @@
 
     <field name="Image Height (pixels)" size="16" start="24" type="uint"/>
     <field name="Image Width (pixels)" size="16" start="8" type="uint"/>
-    <field name="Number of Render Targets Minus 1" size="4" start="4" type="uint"/>
+    <field name="Number of Render Targets" size="4" start="4" type="uint" minus_one="true"/>
+
+    <field name="sub-id" size="4" start="0" type="uint" default="0"/>
+  </packet>
+
+  <packet code="121" name="Tile Rendering Mode Cfg (Common)" cl="R" min_ver="41">
+    <field name="Pad" size="12" start="52" type="uint"/>
+
+    <field name="Early Depth/Stencil Clear" size="1" start="51" type="bool"/>
+    <field name="Internal Depth Type" size="4" start="47" type="Internal Depth Type"/>
+
+    <field name="Early-Z disable" size="1" start="46" type="bool"/>
+
+    <field name="Early-Z Test and Update Direction" size="1" start="45" type="uint">
+      <value name="Early-Z direction LT/LE" value="0"/>
+      <value name="Early-Z direction GT/GE" value="1"/>
+    </field>
+
+    <field name="Double-buffer in non-ms mode" size="1" start="43" type="bool"/>
+    <field name="Multisample Mode (4x)" size="1" start="42" type="bool"/>
+
+    <field name="Maximum BPP of all render targets" size="2" start="40" type="Internal BPP"/>
+
+    <field name="Image Height (pixels)" size="16" start="24" type="uint"/>
+    <field name="Image Width (pixels)" size="16" start="8" type="uint"/>
+    <field name="Number of Render Targets" size="4" start="4" type="uint" minus_one="true"/>
 
     <field name="sub-id" size="4" start="0" type="uint" default="0"/>
   </packet>
 
-  <packet code="121" name="Tile Rendering Mode Configuration (Render Target config)" cl="R">
+  <packet code="121" name="Tile Rendering Mode Cfg (Color)" cl="R" max_ver="33">
     <field name="Address" size="32" start="32" type="address"/>
 
     <field name="Pad" size="4" start="28" type="uint"/>
 
     <field name="Flip Y" size="1" start="27" type="bool"/>
 
-    <field name="Memory Format" size="3" start="24" type="uint" prefix="Memory Format">
-      <value name="Raster" value="0"/>
-      <value name="Lineartile" value="1"/>
-      <value name="UB-linear (1 UIF block wide)" value="2"/>
-      <value name="UB-linear (2 UIF blocks wide)" value="3"/>
-      <value name="UIF (No XOR)" value="4"/>
-      <value name="UIF (XOR)" value="5"/>
-    </field>
+    <field name="Memory Format" size="3" start="24" type="Memory Format"/>
 
-    <field name="A dithered" size="1" start="23" type="bool"/>
-    <field name="BGR dithered" size="1" start="22" type="bool"/>
-
-    <field name="Output image format" size="6" start="16" type="uint" prefix="Output Image Format">
-      <value name="srgb8_alpha8" value="0"/>
-      <value name="srgb" value="1"/>
-      <value name="rgb10_a2ui" value="2"/>
-      <value name="rgb10_a2" value="3"/>
-      <value name="abgr1555" value="4"/>
-      <value name="alpha-masked abgr1555" value="5"/>
-      <value name="abgr4444" value="6"/>
-      <value name="bgr565" value="7"/>
-      <value name="r11f_g11f_b10f" value="8"/>
-      <value name="rgba32f"  value="9"/>
-      <value name="rg32f"    value="10"/>
-      <value name="r32f"     value="11"/>
-      <value name="rgba32i"  value="12"/>
-      <value name="rg32i"    value="13"/>
-      <value name="r32i"     value="14"/>
-      <value name="rgba32ui" value="15"/>
-      <value name="rg32ui"   value="16"/>
-      <value name="r32ui"    value="17"/>
-      <value name="rgba16f"  value="18"/>
-      <value name="rg16f"    value="19"/>
-      <value name="r16f"     value="20"/>
-      <value name="rgba16i"  value="21"/>
-      <value name="rg16i"    value="22"/>
-      <value name="r16i"     value="23"/>
-      <value name="rgba16ui" value="24"/>
-      <value name="rg16ui"   value="25"/>
-      <value name="r16ui"    value="26"/>
-      <value name="rgba8"    value="27"/>
-      <value name="rgb8"     value="28"/>
-      <value name="rg8"      value="29"/>
-      <value name="r8"       value="30"/>
-      <value name="rgba8i"   value="31"/>
-      <value name="rg8i"     value="32"/>
-      <value name="r8i"      value="33"/>
-      <value name="rgba8ui"  value="34"/>
-      <value name="rg8ui"    value="35"/>
-      <value name="r8ui"     value="36"/>
-      <value name="srgbx8"   value="37"/>
-      <value name="rgbx8"    value="38"/>
-    </field>
+    <field name="Dither Mode" size="2" start="22" type="Dither Mode"/>
 
-    <field name="Decimate mode" size="2" start="14" type="uint"/>
+    <field name="Output image format" size="6" start="16" type="Output Image Format"/>
 
-    <field name="Internal Type" size="4" start="10" type="uint" prefix="Internal Type">
-      <value name="8i" value="0"/>
-      <value name="8ui" value="1"/>
-      <value name="8" value="2"/>
-      <value name="16i" value="4"/>
-      <value name="16ui" value="5"/>
-      <value name="16f" value="6"/>
-      <value name="32i" value="8"/>
-      <value name="32ui" value="9"/>
-      <value name="32f" value="10"/>
-    </field>
+    <field name="Decimate mode" size="2" start="14" type="Decimate Mode"/>
 
-    <field name="Internal BPP" size="2" start="8" type="uint" prefix="Internal Bpp">
-      <value name="32" value="0"/>
-      <value name="64" value="1"/>
-      <value name="128" value="2"/>
-    </field>
+    <field name="Internal Type" size="4" start="10" type="Internal Type"/>
+    <field name="Internal BPP" size="2" start="8" type="Internal BPP"/>
     <field name="Render Target Number" size="4" start="4" type="uint"/>
     <field name="sub-id" size="4" start="0" type="uint" default="2"/>
   </packet>
 
-  <packet code="121" name="Tile Rendering Mode Configuration (Z/Stencil config)" cl="R">
-    <field name="Address" size="32" start="32" type="address"/>
+  <packet code="121" name="Tile Rendering Mode Cfg (Color)" cl="R" min_ver="41">
+
+    <field name="Pad" size="28" start="34" type="uint"/>
+
+    <field name="Render Target 3 Clamp" size="2" start="32" type="Render Target Clamp"/>
+    <field name="Render Target 3 Internal Type" size="4" start="30" type="Internal Type"/>
+    <field name="Render Target 3 Internal BPP" size="2" start="28" type="Internal BPP"/>
+
+    <field name="Render Target 2 Clamp" size="2" start="26" type="Render Target Clamp"/>
+    <field name="Render Target 2 Internal Type" size="4" start="22" type="Internal Type"/>
+    <field name="Render Target 2 Internal BPP" size="2" start="20" type="Internal BPP"/>
+
+    <field name="Render Target 1 Clamp" size="2" start="18" type="Render Target Clamp"/>
+    <field name="Render Target 1 Internal Type" size="4" start="14" type="Internal Type"/>
+    <field name="Render Target 1 Internal BPP" size="2" start="12" type="Internal BPP"/>
+
+    <field name="Render Target 0 Clamp" size="2" start="10" type="Render Target Clamp"/>
+    <field name="Render Target 0 Internal Type" size="4" start="6" type="Internal Type"/>
+    <field name="Render Target 0 Internal BPP" size="2" start="4" type="Internal BPP"/>
+
+    <field name="sub-id" size="4" start="0" type="uint" default="1"/>
+  </packet>
+
+  <packet code="121" name="Tile Rendering Mode Cfg (Z/Stencil)" cl="R" max_ver="33">
+    <field name="Address" size="26" start="38" type="address"/>
 
     <field name="Padded height of output image in UIF blocks" size="13" start="25" type="uint"/>
 
-    <field name="Memory Format" size="3" start="22" type="uint" prefix="Memory Format">
-      <value name="Raster" value="0"/>
-      <value name="Lineartile" value="1"/>
-      <value name="UB-linear (1 UIF block wide)" value="2"/>
-      <value name="UB-linear (2 UIF blocks wide)" value="3"/>
-      <value name="UIF (No XOR)" value="4"/>
-      <value name="UIF (XOR)" value="5"/>
-    </field>
+    <field name="Memory Format" size="3" start="22" type="Memory Format"/>
 
-    <field name="Output image format" size="6" start="16" type="uint" prefix="Output Image Format">
-      <value name="depth_component32f" value="0"/>
-      <value name="depth_component24" value="1"/>
-      <value name="depth_component16" value="2"/>
-      <value name="depth24_stencil8" value="3"/>
-    </field>
+    <field name="Output image format" size="6" start="16" type="Z/S Output Image Format"/>
 
     <field name="Decimate mode" size="2" start="14" type="uint"/>
 
-    <field name="Internal Type" size="4" start="10" type="uint" prefix="Internal Type">
-      <value name="depth_32f" value="0"/>
-      <value name="depth_24" value="1"/>
-      <value name="depth_16" value="2"/>
-    </field>
+    <field name="Internal Type" size="4" start="10" type="Internal Depth Type"/>
 
     <field name="Internal BPP (ignored)" size="2" start="8" type="uint"/>
     <!-- selects between Z/Stencil config packet and Separate Stencil packet. -->
@@ -562,16 +907,25 @@
     <field name="sub-id" size="4" start="0" type="uint" default="1"/>
   </packet>
 
-  <packet code="121" name="Tile Rendering Mode Configuration (Z Stencil Clear Values)" cl="R">
+  <packet code="121" name="Tile Rendering Mode Cfg (ZS Clear Values)" cl="R" max_ver="33">
     <field name="unused" size="16" start="48" type="uint"/>
 
     <field name="Z Clear Value" size="32" start="16" type="float"/>
 
-    <field name="Stencil/VG Mask Clear Value" size="8" start="8" type="uint"/>
+    <field name="Stencil Clear Value" size="8" start="8" type="uint"/>
     <field name="sub-id" size="4" start="0" type="uint" default="3"/>
   </packet>
 
-  <packet code="121" name="Tile Rendering Mode Configuration (Clear Colors Part1)" cl="R">
+  <packet code="121" name="Tile Rendering Mode Cfg (ZS Clear Values)" cl="R" min_ver="41">
+    <field name="unused" size="16" start="48" type="uint"/>
+
+    <field name="Z Clear Value" size="32" start="16" type="float"/>
+
+    <field name="Stencil Clear Value" size="8" start="8" type="uint"/>
+    <field name="sub-id" size="4" start="0" type="uint" default="2"/>
+  </packet>
+
+  <packet code="121" name="Tile Rendering Mode Cfg (Clear Colors Part1)" cl="R" max_ver="33">
     <!-- Express this as a 56-bit field? -->
     <field name="Clear Color next 24 bits" size="24" start="40" type="uint"/>
     <field name="Clear Color low 32 bits" size="32" start="8" type="uint"/>
@@ -580,7 +934,16 @@
     <field name="sub-id" size="4" start="0" type="uint" default="4"/>
   </packet>
 
-  <packet code="121" name="Tile Rendering Mode Configuration (Clear Colors Part2)" cl="R">
+  <packet code="121" name="Tile Rendering Mode Cfg (Clear Colors Part1)" cl="R" min_ver="41">
+    <!-- Express this as a 56-bit field? -->
+    <field name="Clear Color next 24 bits" size="24" start="40" type="uint"/>
+    <field name="Clear Color low 32 bits" size="32" start="8" type="uint"/>
+
+    <field name="Render Target number" size="4" start="4" type="uint"/>
+    <field name="sub-id" size="4" start="0" type="uint" default="3"/>
+  </packet>
+
+  <packet code="121" name="Tile Rendering Mode Cfg (Clear Colors Part2)" cl="R" max_ver="33">
     <!-- Express this as a 56-bit field? -->
     <field name="Clear Color mid-high 24 bits" size="24" start="40" type="uint"/>
     <field name="Clear Color mid-low 32 bits" size="32" start="8" type="uint"/>
@@ -589,7 +952,16 @@
     <field name="sub-id" size="4" start="0" type="uint" default="5"/>
   </packet>
 
-  <packet code="121" name="Tile Rendering Mode Configuration (Clear Colors Part3)" cl="R">
+  <packet code="121" name="Tile Rendering Mode Cfg (Clear Colors Part2)" cl="R" min_ver="41">
+    <!-- Express this as a 56-bit field? -->
+    <field name="Clear Color mid-high 24 bits" size="24" start="40" type="uint"/>
+    <field name="Clear Color mid-low 32 bits" size="32" start="8" type="uint"/>
+
+    <field name="Render Target number" size="4" start="4" type="uint"/>
+    <field name="sub-id" size="4" start="0" type="uint" default="4"/>
+  </packet>
+
+  <packet code="121" name="Tile Rendering Mode Cfg (Clear Colors Part3)" cl="R" max_ver="33">
     <field name="pad" size="11" start="53" type="uint"/>
     <field name="UIF padded height in UIF blocks" size="13" start="40" type="uint"/>
     <!-- image height is for Y flipping -->
@@ -600,12 +972,24 @@
     <field name="sub-id" size="4" start="0" type="uint" default="6"/>
   </packet>
 
-  <packet code="124" name="Tile Coordinates">
+  <packet code="121" name="Tile Rendering Mode Cfg (Clear Colors Part3)" cl="R" min_ver="41">
+    <field name="pad" size="11" start="53" type="uint"/>
+    <field name="UIF padded height in UIF blocks" size="13" start="40" type="uint"/>
+    <!-- image height is for Y flipping -->
+    <field name="Raster Row Stride or Image Height in Pixels" size="16" start="24" type="uint"/>
+    <field name="Clear Color high 16 bits" size="16" start="8" type="uint"/>
+
+    <field name="Render Target number" size="4" start="4" type="uint"/>
+    <field name="sub-id" size="4" start="0" type="uint" default="5"/>
+  </packet>
+
+  <packet code="124" shortname="tile_coords" name="Tile Coordinates">
     <field name="tile row number" size="12" start="12" type="uint"/>
     <field name="tile column number" size="12" start="0" type="uint"/>
   </packet>
 
-  <packet code="122" name="Multicore Rendering Supertile Configuration" cl="R">
+  <packet code="122" name="Multicore Rendering Supertile Cfg" cl="R">
+    <field name="Number of Bin Tile Lists" size="3" start="61" type="uint" minus_one="true"/>
     <field name="Supertile Raster Order" size="1" start="60" type="bool"/>
     <field name="Multicore Enable" size="1" start="56" type="bool"/>
 
@@ -615,17 +999,17 @@
     <field name="Total Frame Height in Supertiles" size="8" start="24" type="uint"/>
     <field name="Total Frame Width in Supertiles" size="8" start="16" type="uint"/>
 
-    <field name="Supertile Height in Tiles minus 1" size="8" start="8" type="uint"/>
-    <field name="Supertile Width in Tiles minus 1" size="8" start="0" type="uint"/>
+    <field name="Supertile Height in Tiles" size="8" start="8" type="uint" minus_one="true"/>
+    <field name="Supertile Width in Tiles" size="8" start="0" type="uint" minus_one="true"/>
   </packet>
 
-  <packet code="123" name="Multicore Rendering Tile List Set Base" cl="R">
+  <packet code="123" shortname="multicore_rendering_tile_list_base" name="Multicore Rendering Tile List Set Base" cl="R">
     <field name="address" size="26" start="6" type="address"/>
     <field name="Tile List Set Number" size="4" start="0" type="uint"/>
   </packet>
 
   <!-- add fields -->
-  <packet code="125" name="Tile Coordinates Implicit"/>
+  <packet code="125" shortname="implicit_tile_coords" name="Tile Coordinates Implicit"/>
 
   <packet code="126" name="Tile List Initial Block Size">
     <field name="Use auto-chained tile lists" size="1" start="2" type="bool"/>
@@ -637,7 +1021,7 @@
     </field>
   </packet>
 
-  <struct name="GL Shader State Record">
+  <struct name="GL Shader State Record" max_ver="33">
     <field name="Point size in shaded vertex data" size="1" start="0" type="bool"/>
     <field name="Enable clipping" size="1" start="1" type="bool"/>
     <field name="Vertex ID read by coordinate shader" size="1" start="2" type="bool"/>
@@ -657,17 +1041,128 @@
     <field name="Vertex Shader input VPM segment size" size="8" start="7b" type="uint"/>
     <field name="Address of default attribute values" size="32" start="8b" type="address"/>
     <field name="Fragment Shader Code Address" size="29" start="99" type="address"/>
-    <field name="2-way threadable" size="1" start="96" type="bool"/>
-    <field name="4-way threadable" size="1" start="97" type="bool"/>
-    <field name="Propagate NaNs" size="1" start="98" type="bool"/>
+    <field name="Fragment Shader 2-way threadable" size="1" start="96" type="bool"/>
+    <field name="Fragment Shader 4-way threadable" size="1" start="97" type="bool"/>
+    <field name="Fragment Shader Propagate NaNs" size="1" start="98" type="bool"/>
     <field name="Fragment Shader Uniforms Address" size="32" start="16b" type="address"/>
     <field name="Vertex Shader Code Address" size="32" start="20b" type="address"/>
+    <field name="Vertex Shader 2-way threadable" size="1" start="160" type="bool"/>
+    <field name="Vertex Shader 4-way threadable" size="1" start="161" type="bool"/>
+    <field name="Vertex Shader Propagate NaNs" size="1" start="162" type="bool"/>
     <field name="Vertex Shader Uniforms Address" size="32" start="24b" type="address"/>
     <field name="Coordinate Shader Code Address" size="32" start="28b" type="address"/>
+    <field name="Coordinate Shader 2-way threadable" size="1" start="224" type="bool"/>
+    <field name="Coordinate Shader 4-way threadable" size="1" start="225" type="bool"/>
+    <field name="Coordinate Shader Propagate NaNs" size="1" start="226" type="bool"/>
     <field name="Coordinate Shader Uniforms Address" size="32" start="32b" type="address"/>
   </struct>
 
-  <struct name="GL Shader State Attribute Record">
+  <struct name="GL Shader State Record" min_ver="41">
+    <field name="Point size in shaded vertex data" size="1" start="0" type="bool"/>
+    <field name="Enable clipping" size="1" start="1" type="bool"/>
+
+    <field name="Vertex ID read by coordinate shader" size="1" start="2" type="bool"/>
+    <field name="Instance ID read by coordinate shader" size="1" start="3" type="bool"/>
+    <field name="Base Instance ID read by coordinate shader" size="1" start="4" type="bool"/>
+    <field name="Vertex ID read by vertex shader" size="1" start="5" type="bool"/>
+    <field name="Instance ID read by vertex shader" size="1" start="6" type="bool"/>
+    <field name="Base Instance ID read by vertex shader" size="1" start="7" type="bool"/>
+
+    <field name="Fragment shader does Z writes" size="1" start="8" type="bool"/>
+    <field name="Turn off early-z test" size="1" start="9" type="bool"/>
+    <field name="Coordinate shader has separate input and output VPM blocks" size="1" start="10" type="bool"/>
+    <field name="Vertex shader has separate input and output VPM blocks" size="1" start="11" type="bool"/>
+    <field name="Fragment shader uses real pixel centre W in addition to centroid W2" size="1" start="12" type="bool"/>
+    <field name="Enable Sample Rate Shading" size="1" start="13" type="bool"/>
+    <field name="Any shader reads hardware-written Primitive ID" size="1" start="14" type="bool"/>
+    <field name="Insert Primitive ID as first varying to fragment shader" size="1" start="15" type="bool"/>
+    <field name="Turn off scoreboard" size="1" start="16" type="bool"/>
+    <field name="Do scoreboard wait on first thread switch" size="1" start="17" type="bool"/>
+    <field name="Disable implicit point/line varyings" size="1" start="18" type="bool"/>
+    <field name="No prim pack" size="1" start="19" type="bool"/>
+
+    <field name="Number of varyings in Fragment Shader" size="8" start="3b" type="uint"/>
+
+    <field name="Coordinate Shader output VPM segment size" size="4" start="4b" type="uint"/>
+    <field name="Min Coord Shader output segments required in play in addition to VCM cache size" size="4" start="36" type="uint"/>
+
+    <field name="Coordinate Shader input VPM segment size" size="4" start="5b" type="uint"/>
+    <field name="Min Coord Shader input segments required in play" size="4" start="44" type="uint" minus_one="true"/>
+
+    <field name="Vertex Shader output VPM segment size" size="4" start="6b" type="uint"/>
+    <field name="Min Vertex Shader output segments required in play in addition to VCM cache size" size="4" start="52" type="uint"/>
+
+    <field name="Vertex Shader input VPM segment size" size="4" start="7b" type="uint"/>
+    <field name="Min Vertex Shader input segments required in play" size="4" start="60" type="uint" minus_one="true"/>
+
+    <field name="Address of default attribute values" size="32" start="8b" type="address"/>
+
+    <field name="Fragment Shader Code Address" size="29" start="99" type="address"/>
+    <field name="Fragment Shader 4-way threadable" size="1" start="96" type="bool"/>
+    <field name="Fragment Shader start in final thread section" size="1" start="97" type="bool"/>
+    <field name="Fragment Shader Propagate NaNs" size="1" start="98" type="bool"/>
+    <field name="Fragment Shader Uniforms Address" size="32" start="16b" type="address"/>
+
+    <field name="Vertex Shader Code Address" size="29" start="163" type="address"/>
+    <field name="Vertex Shader 4-way threadable" size="1" start="160" type="bool"/>
+    <field name="Vertex Shader start in final thread section" size="1" start="161" type="bool"/>
+    <field name="Vertex Shader Propagate NaNs" size="1" start="162" type="bool"/>
+    <field name="Vertex Shader Uniforms Address" size="32" start="24b" type="address"/>
+
+    <field name="Coordinate Shader Code Address" size="29" start="227" type="address"/>
+    <field name="Coordinate Shader 4-way threadable" size="1" start="224" type="bool"/>
+    <field name="Coordinate Shader start in final thread section" size="1" start="225" type="bool"/>
+    <field name="Coordinate Shader Propagate NaNs" size="1" start="226" type="bool"/>
+    <field name="Coordinate Shader Uniforms Address" size="32" start="32b" type="address"/>
+  </struct>
+
+  <struct name="Geometry Shader State Record" min_ver="41">
+    <field name="Geometry Bin Mode Shader Code Address" size="32" start="0b" type="address"/>
+    <field name="4-way threadable" size="1" start="0" type="bool"/>
+    <field name="Start in final thread section" size="1" start="1" type="bool"/>
+    <field name="Propagate NaNs" size="1" start="2" type="bool"/>
+    <field name="Geometry Bin Mode Shader Uniforms Address" size="32" start="4b" type="address"/>
+    <field name="Geometry Render Mode Shader Code Address" size="32" start="8b" type="address"/>
+    <field name="Geometry Render Mode Shader Uniforms Address" size="32" start="12b" type="address"/>
+  </struct>
+
+  <struct name="Tessellation Shader State Record" min_ver="41">
+    <field name="Tessellation Bin Mode Control Shader Code Address" size="32" start="0b" type="address"/>
+    <field name="Tessellation Bin Mode Control Shader Uniforms Address" size="32" start="4b" type="address"/>
+    <field name="Tessellation Render Mode Control Shader Code Address" size="32" start="8b" type="address"/>
+    <field name="Tessellation Render Mode Control Shader Uniforms Address" size="32" start="12b" type="address"/>
+
+    <field name="Tessellation Bin Mode Evaluation Shader Code Address" size="32" start="16b" type="address"/>
+    <field name="Tessellation Bin Mode Evaluation Shader Uniforms Address" size="32" start="20b" type="address"/>
+    <field name="Tessellation Render Mode Evaluation Shader Code Address" size="32" start="24b" type="address"/>
+    <field name="Tessellation Render Mode Evaluation Shader Uniforms Address" size="32" start="28b" type="address"/>
+  </struct>
+
+  <struct name="GL Shader State Attribute Record" max_ver="33">
+    <field name="Address" size="32" start="0" type="address"/>
+
+    <field name="Vec size" size="2" start="32" type="uint"/>
+    <field name="Type" size="3" start="34" type="uint">
+      <value name="Attribute half-float" value="1"/>
+      <value name="Attribute float" value="2"/>
+      <value name="Attribute fixed" value="3"/>
+      <value name="Attribute byte" value="4"/>
+      <value name="Attribute short" value="5"/>
+      <value name="Attribute int" value="6"/>
+      <value name="Attribute int2_10_10_10" value="7"/>
+    </field>
+    <field name="Signed int type" size="1" start="37" type="bool"/>
+    <field name="Normalized int type" size="1" start="38" type="bool"/>
+    <field name="Read as int/uint" size="1" start="39" type="bool"/>
+
+    <field name="Number of values read by Coordinate shader" size="4" start="40" type="uint"/>
+    <field name="Number of values read by Vertex shader" size="4" start="44" type="uint"/>
+
+    <field name="Instance Divisor" size="16" start="6b" type="uint"/>
+    <field name="Stride" size="32" start="8b" type="uint"/>
+  </struct>
+
+  <struct name="GL Shader State Attribute Record" min_ver="41">
     <field name="Address" size="32" start="0" type="address"/>
 
     <field name="Vec size" size="2" start="32" type="uint"/>
@@ -689,6 +1184,7 @@
 
     <field name="Instance Divisor" size="16" start="6b" type="uint"/>
     <field name="Stride" size="32" start="8b" type="uint"/>
+    <field name="Maximum Index" size="32" start="12b" type="uint"/>
   </struct>
 
   <struct name="VPM generic block write setup">
@@ -727,7 +1223,7 @@
     <field name="addr" size="13" start="0" type="uint"/>
   </struct>
 
-  <struct name="Texture Uniform Parameter 0 CFG_MODE=1">
+  <struct name="Texture Uniform Parameter 0 CFG_MODE=1" max_ver="33">
     <field name="Per-pixel mask enable" size="1" start="31" type="bool"/>
 
     <field name="Texel offset for r coordinate" size="4" start="27" type="int"/>
@@ -767,7 +1263,7 @@
     <field name="Gather sample mode" size="1" start="4" type="bool"/>
     <field name="Fetch sample mode" size="1" start="3" type="bool"/>
 
-    <field name="Lookup Type" size="3" start="0" type="bool">
+    <field name="Lookup Type" size="3" start="0" type="uint">
       <value name="Texture 2D" value="0"/>
       <value name="Texture 2D array" value="1"/>
       <value name="Texture 3D" value="2"/>
@@ -778,18 +1274,59 @@
     </field>
   </struct>
 
-  <struct name="Texture Uniform Parameter 1 CFG_MODE=1">
-    <field name="Texture state record base address" size="32" start="0" type="address"/>
-    <field name="Return word 3 of texture data" size="1" start="3" type="bool"/>
-    <field name="Return word 2 of texture data" size="1" start="2" type="bool"/>
-    <field name="Return word 1 of texture data" size="1" start="1" type="bool"/>
-    <field name="Return word 0 of texture data" size="1" start="0" type="bool"/>
+  <struct name="Texture Uniform Parameter 1 CFG_MODE=1" max_ver="33">
+    <field name="Texture state record base address" size="28" start="4" type="address"/>
+    <field name="Return words of texture data" size="4" start="0" type="uint"/>
+  </struct>
+
+  <struct name="TMU Config Parameter 0" min_ver="41">
+    <field name="Texture state address" size="32" start="0" type="address"/>
+    <field name="Return words of texture data" size="4" start="0" type="uint"/>
+  </struct>
+
+  <struct name="TMU Config Parameter 1" min_ver="41">
+    <field name="Sampler state address" size="32" start="0" type="address"/>
+    <field name="Per-pixel mask enable" size="1" start="2" type="bool"/>
+    <field name="Unnormalized coordinates" size="1" start="1" type="bool"/>
+    <field name="Output Type 32-bit" size="1" start="0" type="bool"/>
   </struct>
 
-  <struct name="Texture Shader State">
+  <struct name="TMU Config Parameter 2" min_ver="41" max_ver="41">
+    <field name="Pad" size="24" start="8" type="uint"/>
+    <field name="Op" size="4" start="20" type="TMU Op"/>
+    <field name="Offset R" size="4" start="16" type="int"/>
+    <field name="Offset T" size="4" start="12" type="int"/>
+    <field name="Offset S" size="4" start="8" type="int"/>
+    <field name="Gather Mode" size="1" start="7" type="bool"/>
+    <field name="Gather Component" size="2" start="5" type="uint"/>
+    <field name="Coefficient Mode" size="1" start="4" type="bool"/>
+    <field name="Sample Number" size="2" start="2" type="uint"/>
+    <field name="Disable AutoLOD" size="1" start="1" type="bool"/>
+    <field name="Offset Format 8" size="1" start="0" type="bool"/>
+  </struct>
+
+  <struct name="TMU Config Parameter 2" min_ver="42">
+    <field name="Pad" size="23" start="9" type="uint"/>
+    <field name="LOD Query" size="1" start="8" type="bool"/>
+    <field name="Op" size="4" start="20" type="TMU Op"/>
+    <field name="Offset R" size="4" start="16" type="int"/>
+    <field name="Offset T" size="4" start="12" type="int"/>
+    <field name="Offset S" size="4" start="8" type="int"/>
+    <field name="Gather Mode" size="1" start="7" type="bool"/>
+    <field name="Gather Component" size="2" start="5" type="uint"/>
+    <field name="Coefficient Mode" size="1" start="4" type="bool"/>
+    <field name="Sample Number" size="2" start="2" type="uint"/>
+    <field name="Disable AutoLOD" size="1" start="1" type="bool"/>
+    <field name="Offset Format 8" size="1" start="0" type="bool"/>
+  </struct>
+
+  <struct name="Texture Shader State" max_ver="33">
+    <field name="UIF XOR disable" size="1" start="255" type="bool"/>
     <field name="Level 0 is strictly UIF" size="1" start="254" type="bool"/>
     <field name="Level 0 XOR enable" size="1" start="252" type="bool"/>
     <field name="Level 0 UB_PAD" size="4" start="248" type="uint"/>
+    <field name="Output 32-bit" size="1" start="246" type="bool"/>
+    <field name="Sample Number" size="2" start="244" type="uint"/>
 
     <field name="Base Level" size="4" start="240" type="uint"/>
     <field name="Fixed Bias" size="16" start="224" type="s8.8"/>
@@ -831,23 +1368,94 @@
 
     <field name="Array Stride (64-byte aligned)" size="26" start="32" type="uint"/>
 
+    <field name="Texture base pointer" size="30" start="2" type="address"/>
+
+    <field name="Filter" size="4" start="0" type="TMU Filter"/>
+  </struct>
+
+  <struct name="Texture Shader State" min_ver="41">
+    <field name="Pad" size="56" start="136" type="uint"/>
+    <field name="UIF XOR disable" size="1" start="135" type="bool"/>
+    <field name="Level 0 is strictly UIF" size="1" start="134" type="bool"/>
+    <field name="Level 0 XOR enable" size="1" start="132" type="bool"/>
+    <field name="Level 0 UB_PAD" size="4" start="128" type="uint"/>
+
+    <field name="Base Level" size="4" start="124" type="uint"/>
+    <field name="Max Level" size="4" start="120" type="uint"/>
+
+    <field name="Swizzle A" size="3" start="117" type="uint">
+      <value name="Swizzle Zero" value="0"/>
+      <value name="Swizzle One" value="1"/>
+      <value name="Swizzle Red" value="2"/>
+      <value name="Swizzle Green" value="3"/>
+      <value name="Swizzle Blue" value="4"/>
+      <value name="Swizzle Alpha" value="5"/>
+    </field>
+
+    <field name="Swizzle B" size="3" start="114" type="uint"/>
+    <field name="Swizzle G" size="3" start="111" type="uint"/>
+    <field name="Swizzle R" size="3" start="108" type="uint"/>
+    <field name="Extended" size="1" start="107" type="bool"/>
+
+    <field name="Texture type" size="7" start="100" type="uint"/>
+    <field name="Image Depth" size="14" start="86" type="uint"/>
+    <field name="Image Height" size="14" start="72" type="uint"/>
+    <field name="Image Width" size="14" start="58" type="uint"/>
+
+    <field name="Array Stride (64-byte aligned)" size="26" start="32" type="uint"/>
+
     <field name="Texture base pointer" size="32" start="0" type="address"/>
 
-    <field name="Minification Filter" size="3" start="1" type="uint"/>
-    <field name="Magnification Filter" size="1" start="0" type="uint"/>
+    <field name="Reverse Standard Border Color" size="1" start="5" type="bool"/>
+    <field name="AHDR" size="1" start="4" type="bool"/>
+    <field name="sRGB" size="1" start="3" type="bool"/>
+    <field name="Flip S and T on incoming request" size="1" start="2" type="bool"/>
+    <field name="Flip texture Y Axis" size="1" start="1" type="bool"/>
+    <field name="Flip texture X Axis" size="1" start="0" type="bool"/>
+  </struct>
+
+  <struct name="Sampler State" min_ver="41">
+    <field name="Border color Alpha" size="32" start="160" type="uint"/>
+    <field name="Border color Blue" size="32" start="128" type="uint"/>
+    <field name="Border color Green" size="32" start="96" type="uint"/>
+    <field name="Border color Red" size="32" start="64" type="uint"/>
+
+    <field name="Maximum Anisotropy" size="2" start="61" type="uint"/>
+    <field name="Border Color Mode" size="3" start="58" type="Border Color Mode"/>
+    <field name="Wrap I Border" size="1" start="57" type="bool"/>
+    <field name="Wrap R" size="3" start="54" type="Wrap Mode"/>
+    <field name="Wrap T" size="3" start="51" type="Wrap Mode"/>
+    <field name="Wrap S" size="3" start="48" type="Wrap Mode"/>
+
+    <field name="Fixed Bias" size="16" start="32" type="s8.8"/>
+    <field name="Max Level-of-Detail" size="12" start="20" type="u4.8"/>
+    <field name="Min Level-of-Detail" size="12" start="8" type="u4.8"/>
+
+    <field name="sRGB Disable" size="1" start="7" type="bool"/>
+
+    <field name="Depth Compare Function" size="3" start="4" type="Compare Function"/>
+
+    <field name="Anisotropy Enable" size="1" start="3" type="bool"/>
+    <field name="Mip filter Nearest" size="1" start="2" type="bool"/>
+    <field name="Min filter Nearest" size="1" start="1" type="bool"/>
+    <field name="Mag filter Nearest" size="1" start="0" type="bool"/>
   </struct>
 
   <enum name="Texture Data Formats">
+    <!--
+	most formats here have R in the low bits, A in the high bits.
+	Exceptions noted.
+    -->
     <value name="Texture Data Format R8" value="0"/>
     <value name="Texture Data Format R8 SNORM" value="1"/>
     <value name="Texture Data Format RG8" value="2"/>
     <value name="Texture Data Format RG8 SNORM" value="3"/>
     <value name="Texture Data Format RGBA8" value="4"/>
     <value name="Texture Data Format RGBA8 SNORM" value="5"/>
-    <value name="Texture Data Format RGB565" value="6"/>
-    <value name="Texture Data Format RGBA4" value="7"/>
-    <value name="Texture Data Format RGB5_A1" value="8"/>
-    <value name="Texture Data Format RGB10_A2" value="9"/>
+    <value name="Texture Data Format RGB565" value="6"/> <!-- B in low bits -->
+    <value name="Texture Data Format RGBA4" value="7"/> <!-- A low, R high -->
+    <value name="Texture Data Format RGB5_A1" value="8"/> <!-- A low, R high -->
+    <value name="Texture Data Format RGB10_A2" value="9"/> <!-- R low, A high -->
     <value name="Texture Data Format R16" value="10"/>
     <value name="Texture Data Format R16 SNORM" value="11"/>
     <value name="Texture Data Format RG16" value="12"/>
@@ -862,7 +1470,7 @@
     <value name="Texture Data Format DEPTH COMP16" value="21"/>
     <value name="Texture Data Format DEPTH COMP24" value="22"/>
     <value name="Texture Data Format DEPTH COMP32F" value="23"/>
-    <value name="Texture Data Format DEPTH24_X8" value="24"/>
+    <value name="Texture Data Format DEPTH24_X8" value="24"/> <!-- X low, D high -->
     <value name="Texture Data Format R4" value="25"/>
     <value name="Texture Data Format R1" value="26"/>
     <!-- generic unfiltered 8-bit sample -->
@@ -906,5 +1514,28 @@
     <value name="Texture Data Format ASTC_10x10" value="75"/>
     <value name="Texture Data Format ASTC_12x10" value="76"/>
     <value name="Texture Data Format ASTC_12x12" value="77"/>
+
+    <value name="Texture Data Format R8I" value="96"/>
+    <value name="Texture Data Format R8UI" value="97"/>
+    <value name="Texture Data Format RG8I" value="98"/>
+    <value name="Texture Data Format RG8UI" value="99"/>
+    <value name="Texture Data Format RGBA8I" value="100"/>
+    <value name="Texture Data Format RGBA8UI" value="101"/>
+
+    <value name="Texture Data Format R16I" value="102"/>
+    <value name="Texture Data Format R16UI" value="103"/>
+    <value name="Texture Data Format RG16I" value="104"/>
+    <value name="Texture Data Format RG16UI" value="105"/>
+    <value name="Texture Data Format RGBA16I" value="106"/>
+    <value name="Texture Data Format RGBA16UI" value="107"/>
+
+    <value name="Texture Data Format R32I" value="108"/>
+    <value name="Texture Data Format R32UI" value="109"/>
+    <value name="Texture Data Format RG32I" value="110"/>
+    <value name="Texture Data Format RG32UI" value="111"/>
+    <value name="Texture Data Format RGBA32I" value="112"/>
+    <value name="Texture Data Format RGBA32UI" value="113"/>
+    <value name="Texture Data Format RGB10_A2UI" value="114"/>
+
   </enum>
 </vcxml>
diff --git a/lib/mesa/src/broadcom/cle/v3d_packet_v33_pack.h b/lib/mesa/src/broadcom/cle/v3d_packet_v33_pack.h
index 934b0de4f..ba3bf8de8 100644
--- a/lib/mesa/src/broadcom/cle/v3d_packet_v33_pack.h
+++ b/lib/mesa/src/broadcom/cle/v3d_packet_v33_pack.h
@@ -9,7 +9,7 @@
 #ifndef V3D33_PACK_H
 #define V3D33_PACK_H
 
-#include "v3d_packet_helpers.h"
+#include "cle/v3d_packet_helpers.h"
 
 
 enum V3D33_Compare_Function {
@@ -81,6 +81,126 @@ enum V3D33_Primitive {
         V3D_PRIM_TRIANGLE_FAN_TF             =     22,
 };
 
+enum V3D33_TMU_Filter {
+        V3D_TMU_FILTER_MIN_LIN_MIP_NONE_MAG_LIN =      0,
+        V3D_TMU_FILTER_MIN_LIN_MIP_NONE_MAG_NEAR =      1,
+        V3D_TMU_FILTER_MIN_NEAR_MIP_NONE_MAG_LIN =      2,
+        V3D_TMU_FILTER_MIN_NEAR_MIP_NONE_MAG_NEAR =      3,
+        V3D_TMU_FILTER_MIN_NEAR_MIP_NEAR_MAG_LIN =      4,
+        V3D_TMU_FILTER_MIN_NEAR_MIP_NEAR_MAG_NEAR =      5,
+        V3D_TMU_FILTER_MIN_NEAR_MIP_LIN_MAG_LIN =      6,
+        V3D_TMU_FILTER_MIN_NEAR_MIP_LIN_MAG_NEAR =      7,
+        V3D_TMU_FILTER_MIN_LIN_MIP_NEAR_MAG_LIN =      8,
+        V3D_TMU_FILTER_MIN_LIN_MIP_NEAR_MAG_NEAR =      9,
+        V3D_TMU_FILTER_MIN_LIN_MIP_LIN_MAG_LIN =     10,
+        V3D_TMU_FILTER_MIN_LIN_MIP_LIN_MAG_NEAR =     11,
+        V3D_TMU_FILTER_ANISOTROPIC_2_1       =     12,
+        V3D_TMU_FILTER_ANISOTROPIC_4_1       =     13,
+        V3D_TMU_FILTER_ANISOTROPIC_8_1       =     14,
+        V3D_TMU_FILTER_ANISOTROPIC_16_1      =     15,
+};
+
+enum V3D33_Varying_Flags_Action {
+        V3D_VARYING_FLAGS_ACTION_UNCHANGED   =      0,
+        V3D_VARYING_FLAGS_ACTION_ZEROED      =      1,
+        V3D_VARYING_FLAGS_ACTION_SET         =      2,
+};
+
+enum V3D33_Memory_Format {
+        V3D_MEMORY_FORMAT_RASTER             =      0,
+        V3D_MEMORY_FORMAT_LINEARTILE         =      1,
+        V3D_MEMORY_FORMAT_UB_LINEAR_1_UIF_BLOCK_WIDE =      2,
+        V3D_MEMORY_FORMAT_UB_LINEAR_2_UIF_BLOCKS_WIDE =      3,
+        V3D_MEMORY_FORMAT_UIF_NO_XOR         =      4,
+        V3D_MEMORY_FORMAT_UIF_XOR            =      5,
+};
+
+enum V3D33_Decimate_Mode {
+        V3D_DECIMATE_MODE_SAMPLE_0           =      0,
+        V3D_DECIMATE_MODE_4X                 =      1,
+        V3D_DECIMATE_MODE_ALL_SAMPLES        =      3,
+};
+
+enum V3D33_Internal_Type {
+        V3D_INTERNAL_TYPE_8I                 =      0,
+        V3D_INTERNAL_TYPE_8UI                =      1,
+        V3D_INTERNAL_TYPE_8                  =      2,
+        V3D_INTERNAL_TYPE_16I                =      4,
+        V3D_INTERNAL_TYPE_16UI               =      5,
+        V3D_INTERNAL_TYPE_16F                =      6,
+        V3D_INTERNAL_TYPE_32I                =      8,
+        V3D_INTERNAL_TYPE_32UI               =      9,
+        V3D_INTERNAL_TYPE_32F                =     10,
+};
+
+enum V3D33_Internal_BPP {
+        V3D_INTERNAL_BPP_32                  =      0,
+        V3D_INTERNAL_BPP_64                  =      1,
+        V3D_INTERNAL_BPP_128                 =      2,
+};
+
+enum V3D33_Internal_Depth_Type {
+        V3D_INTERNAL_TYPE_DEPTH_32F          =      0,
+        V3D_INTERNAL_TYPE_DEPTH_24           =      1,
+        V3D_INTERNAL_TYPE_DEPTH_16           =      2,
+};
+
+enum V3D33_Output_Image_Format {
+        V3D_OUTPUT_IMAGE_FORMAT_SRGB8_ALPHA8 =      0,
+        V3D_OUTPUT_IMAGE_FORMAT_SRGB         =      1,
+        V3D_OUTPUT_IMAGE_FORMAT_RGB10_A2UI   =      2,
+        V3D_OUTPUT_IMAGE_FORMAT_RGB10_A2     =      3,
+        V3D_OUTPUT_IMAGE_FORMAT_ABGR1555     =      4,
+        V3D_OUTPUT_IMAGE_FORMAT_ALPHA_MASKED_ABGR1555 =      5,
+        V3D_OUTPUT_IMAGE_FORMAT_ABGR4444     =      6,
+        V3D_OUTPUT_IMAGE_FORMAT_BGR565       =      7,
+        V3D_OUTPUT_IMAGE_FORMAT_R11F_G11F_B10F =      8,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA32F      =      9,
+        V3D_OUTPUT_IMAGE_FORMAT_RG32F        =     10,
+        V3D_OUTPUT_IMAGE_FORMAT_R32F         =     11,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA32I      =     12,
+        V3D_OUTPUT_IMAGE_FORMAT_RG32I        =     13,
+        V3D_OUTPUT_IMAGE_FORMAT_R32I         =     14,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA32UI     =     15,
+        V3D_OUTPUT_IMAGE_FORMAT_RG32UI       =     16,
+        V3D_OUTPUT_IMAGE_FORMAT_R32UI        =     17,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA16F      =     18,
+        V3D_OUTPUT_IMAGE_FORMAT_RG16F        =     19,
+        V3D_OUTPUT_IMAGE_FORMAT_R16F         =     20,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA16I      =     21,
+        V3D_OUTPUT_IMAGE_FORMAT_RG16I        =     22,
+        V3D_OUTPUT_IMAGE_FORMAT_R16I         =     23,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA16UI     =     24,
+        V3D_OUTPUT_IMAGE_FORMAT_RG16UI       =     25,
+        V3D_OUTPUT_IMAGE_FORMAT_R16UI        =     26,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA8        =     27,
+        V3D_OUTPUT_IMAGE_FORMAT_RGB8         =     28,
+        V3D_OUTPUT_IMAGE_FORMAT_RG8          =     29,
+        V3D_OUTPUT_IMAGE_FORMAT_R8           =     30,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA8I       =     31,
+        V3D_OUTPUT_IMAGE_FORMAT_RG8I         =     32,
+        V3D_OUTPUT_IMAGE_FORMAT_R8I          =     33,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA8UI      =     34,
+        V3D_OUTPUT_IMAGE_FORMAT_RG8UI        =     35,
+        V3D_OUTPUT_IMAGE_FORMAT_R8UI         =     36,
+        V3D_OUTPUT_IMAGE_FORMAT_SRGBX8       =     37,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBX8        =     38,
+};
+
+enum V3D33_Z_S_Output_Image_Format {
+        V3D_OUTPUT_IMAGE_FORMAT_ZS_DEPTH_COMPONENT32F =      0,
+        V3D_OUTPUT_IMAGE_FORMAT_ZS_DEPTH_COMPONENT24 =      1,
+        V3D_OUTPUT_IMAGE_FORMAT_ZS_DEPTH_COMPONENT16 =      2,
+        V3D_OUTPUT_IMAGE_FORMAT_ZS_DEPTH24_STENCIL8 =      3,
+};
+
+enum V3D33_Dither_Mode {
+        V3D_DITHER_MODE_NONE                 =      0,
+        V3D_DITHER_MODE_RGB                  =      1,
+        V3D_DITHER_MODE_A                    =      2,
+        V3D_DITHER_MODE_RGBA                 =      3,
+};
+
 #define V3D33_HALT_opcode                      0
 #define V3D33_HALT_header                       \
    .opcode                              =      0
@@ -805,7 +925,7 @@ struct V3D33_STORE_MULTI_SAMPLE_RESOLVED_TILE_COLOR_BUFFER_EXTENDED {
    uint32_t                             disable_color_buffer_write;
    bool                                 enable_z_write;
    bool                                 enable_stencil_write;
-   bool                                 disable_colour_buffers_clear_on_write;
+   bool                                 disable_color_buffers_clear_on_write;
    bool                                 disable_stencil_buffer_clear_on_write;
    bool                                 disable_z_buffer_clear_on_write;
    bool                                 disable_fast_opportunistic_write_out_in_multisample_mode;
@@ -820,7 +940,7 @@ V3D33_STORE_MULTI_SAMPLE_RESOLVED_TILE_COLOR_BUFFER_EXTENDED_pack(__gen_user_dat
 
    cl[ 1] = __gen_uint(values->enable_z_write, 7, 7) |
             __gen_uint(values->enable_stencil_write, 6, 6) |
-            __gen_uint(values->disable_colour_buffers_clear_on_write, 4, 4) |
+            __gen_uint(values->disable_color_buffers_clear_on_write, 4, 4) |
             __gen_uint(values->disable_stencil_buffer_clear_on_write, 3, 3) |
             __gen_uint(values->disable_z_buffer_clear_on_write, 2, 2) |
             __gen_uint(values->disable_fast_opportunistic_write_out_in_multisample_mode, 1, 1) |
@@ -840,7 +960,7 @@ V3D33_STORE_MULTI_SAMPLE_RESOLVED_TILE_COLOR_BUFFER_EXTENDED_unpack(const uint8_
    values->disable_color_buffer_write = __gen_unpack_uint(cl, 16, 23);
    values->enable_z_write = __gen_unpack_uint(cl, 15, 15);
    values->enable_stencil_write = __gen_unpack_uint(cl, 14, 14);
-   values->disable_colour_buffers_clear_on_write = __gen_unpack_uint(cl, 12, 12);
+   values->disable_color_buffers_clear_on_write = __gen_unpack_uint(cl, 12, 12);
    values->disable_stencil_buffer_clear_on_write = __gen_unpack_uint(cl, 11, 11);
    values->disable_z_buffer_clear_on_write = __gen_unpack_uint(cl, 10, 10);
    values->disable_fast_opportunistic_write_out_in_multisample_mode = __gen_unpack_uint(cl, 9, 9);
@@ -849,38 +969,38 @@ V3D33_STORE_MULTI_SAMPLE_RESOLVED_TILE_COLOR_BUFFER_EXTENDED_unpack(const uint8_
 #endif
 
 
-#define V3D33_RELOAD_TILE_COLOUR_BUFFER_opcode     26
-#define V3D33_RELOAD_TILE_COLOUR_BUFFER_header  \
+#define V3D33_RELOAD_TILE_COLOR_BUFFER_opcode     26
+#define V3D33_RELOAD_TILE_COLOR_BUFFER_header   \
    .opcode                              =     26
 
-struct V3D33_RELOAD_TILE_COLOUR_BUFFER {
+struct V3D33_RELOAD_TILE_COLOR_BUFFER {
    uint32_t                             opcode;
-   uint32_t                             disable_colour_buffer_load;
+   uint32_t                             disable_color_buffer_load;
    bool                                 enable_z_load;
    bool                                 enable_stencil_load;
 };
 
 static inline void
-V3D33_RELOAD_TILE_COLOUR_BUFFER_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                     const struct V3D33_RELOAD_TILE_COLOUR_BUFFER * restrict values)
+V3D33_RELOAD_TILE_COLOR_BUFFER_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                    const struct V3D33_RELOAD_TILE_COLOR_BUFFER * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
    cl[ 1] = __gen_uint(values->enable_z_load, 7, 7) |
             __gen_uint(values->enable_stencil_load, 6, 6);
 
-   cl[ 2] = __gen_uint(values->disable_colour_buffer_load, 0, 7);
+   cl[ 2] = __gen_uint(values->disable_color_buffer_load, 0, 7);
 
 }
 
-#define V3D33_RELOAD_TILE_COLOUR_BUFFER_length      3
+#define V3D33_RELOAD_TILE_COLOR_BUFFER_length      3
 #ifdef __gen_unpack_address
 static inline void
-V3D33_RELOAD_TILE_COLOUR_BUFFER_unpack(const uint8_t * restrict cl,
-                                       struct V3D33_RELOAD_TILE_COLOUR_BUFFER * restrict values)
+V3D33_RELOAD_TILE_COLOR_BUFFER_unpack(const uint8_t * restrict cl,
+                                      struct V3D33_RELOAD_TILE_COLOR_BUFFER * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
-   values->disable_colour_buffer_load = __gen_unpack_uint(cl, 16, 23);
+   values->disable_color_buffer_load = __gen_unpack_uint(cl, 16, 23);
    values->enable_z_load = __gen_unpack_uint(cl, 15, 15);
    values->enable_stencil_load = __gen_unpack_uint(cl, 14, 14);
 }
@@ -924,7 +1044,7 @@ struct V3D33_STORE_TILE_BUFFER_GENERAL {
    uint32_t                             padded_height_of_output_image_in_uif_blocks;
    bool                                 xor_uif;
    bool                                 last_tile_of_frame;
-   bool                                 disable_colour_buffers_clear_on_write;
+   bool                                 disable_color_buffers_clear_on_write;
    bool                                 disable_stencil_buffer_clear_on_write;
    bool                                 disable_z_buffer_clear_on_write;
    bool                                 raw_mode;
@@ -945,7 +1065,7 @@ V3D33_STORE_TILE_BUFFER_GENERAL_pack(__gen_user_data *data, uint8_t * restrict c
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
-   cl[ 1] = __gen_uint(values->disable_colour_buffers_clear_on_write, 7, 7) |
+   cl[ 1] = __gen_uint(values->disable_color_buffers_clear_on_write, 7, 7) |
             __gen_uint(values->disable_stencil_buffer_clear_on_write, 6, 6) |
             __gen_uint(values->disable_z_buffer_clear_on_write, 5, 5) |
             __gen_uint(values->raw_mode, 4, 4) |
@@ -955,10 +1075,9 @@ V3D33_STORE_TILE_BUFFER_GENERAL_pack(__gen_user_data *data, uint8_t * restrict c
             __gen_uint(values->xor_uif, 2, 2) |
             __gen_uint(values->last_tile_of_frame, 0, 0);
 
-   __gen_emit_reloc(data, &values->address);
-   cl[ 3] = __gen_address_offset(&values->address) |
-            __gen_uint(values->padded_height_of_output_image_in_uif_blocks, 3, 15) >> 8;
+   cl[ 3] = __gen_uint(values->padded_height_of_output_image_in_uif_blocks, 3, 15) >> 8;
 
+   __gen_emit_reloc(data, &values->address);
    cl[ 4] = __gen_address_offset(&values->address) >> 8;
 
    cl[ 5] = __gen_address_offset(&values->address) >> 16;
@@ -974,11 +1093,11 @@ V3D33_STORE_TILE_BUFFER_GENERAL_unpack(const uint8_t * restrict cl,
                                        struct V3D33_STORE_TILE_BUFFER_GENERAL * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
-   values->address = __gen_unpack_address(cl, 24, 55);
+   values->address = __gen_unpack_address(cl, 32, 55);
    values->padded_height_of_output_image_in_uif_blocks = __gen_unpack_uint(cl, 19, 31);
    values->xor_uif = __gen_unpack_uint(cl, 18, 18);
    values->last_tile_of_frame = __gen_unpack_uint(cl, 16, 16);
-   values->disable_colour_buffers_clear_on_write = __gen_unpack_uint(cl, 15, 15);
+   values->disable_color_buffers_clear_on_write = __gen_unpack_uint(cl, 15, 15);
    values->disable_stencil_buffer_clear_on_write = __gen_unpack_uint(cl, 14, 14);
    values->disable_z_buffer_clear_on_write = __gen_unpack_uint(cl, 13, 13);
    values->raw_mode = __gen_unpack_uint(cl, 12, 12);
@@ -1023,11 +1142,11 @@ V3D33_LOAD_TILE_BUFFER_GENERAL_pack(__gen_user_data *data, uint8_t * restrict cl
    cl[ 3] = __gen_uint(values->padded_height_of_output_image_in_uif_blocks, 3, 15) >> 8;
 
    __gen_emit_reloc(data, &values->address);
-   cl[ 4] = __gen_address_offset(&values->address);
+   cl[ 4] = __gen_address_offset(&values->address) >> 8;
 
-   cl[ 5] = __gen_address_offset(&values->address) >> 8;
+   cl[ 5] = __gen_address_offset(&values->address) >> 16;
 
-   cl[ 6] = __gen_address_offset(&values->address) >> 16;
+   cl[ 6] = __gen_address_offset(&values->address) >> 24;
 
 }
 
@@ -1047,11 +1166,38 @@ V3D33_LOAD_TILE_BUFFER_GENERAL_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_INDEXED_PRIMITIVE_LIST_opcode     32
-#define V3D33_INDEXED_PRIMITIVE_LIST_header     \
+#define V3D33_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_opcode     31
+#define V3D33_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_header\
+   .opcode                              =     31
+
+struct V3D33_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D33_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                              const struct V3D33_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D33_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D33_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_unpack(const uint8_t * restrict cl,
+                                                struct V3D33_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D33_INDEXED_PRIM_LIST_opcode        32
+#define V3D33_INDEXED_PRIM_LIST_header          \
    .opcode                              =     32
 
-struct V3D33_INDEXED_PRIMITIVE_LIST {
+struct V3D33_INDEXED_PRIM_LIST {
    uint32_t                             opcode;
    uint32_t                             minimum_index;
    bool                                 enable_primitive_restarts;
@@ -1062,12 +1208,12 @@ struct V3D33_INDEXED_PRIMITIVE_LIST {
 #define INDEX_TYPE_8_BIT                         0
 #define INDEX_TYPE_16_BIT                        1
 #define INDEX_TYPE_32_BIT                        2
-   uint32_t                             mode;
+   enum V3D33_Primitive                 mode;
 };
 
 static inline void
-V3D33_INDEXED_PRIMITIVE_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                  const struct V3D33_INDEXED_PRIMITIVE_LIST * restrict values)
+V3D33_INDEXED_PRIM_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                             const struct V3D33_INDEXED_PRIM_LIST * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -1098,11 +1244,11 @@ V3D33_INDEXED_PRIMITIVE_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
    memcpy(&cl[14], &values->minimum_index, sizeof(values->minimum_index));
 }
 
-#define V3D33_INDEXED_PRIMITIVE_LIST_length     18
+#define V3D33_INDEXED_PRIM_LIST_length        18
 #ifdef __gen_unpack_address
 static inline void
-V3D33_INDEXED_PRIMITIVE_LIST_unpack(const uint8_t * restrict cl,
-                                    struct V3D33_INDEXED_PRIMITIVE_LIST * restrict values)
+V3D33_INDEXED_PRIM_LIST_unpack(const uint8_t * restrict cl,
+                               struct V3D33_INDEXED_PRIM_LIST * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->minimum_index = __gen_unpack_uint(cl, 112, 143);
@@ -1116,11 +1262,11 @@ V3D33_INDEXED_PRIMITIVE_LIST_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_INDEXED_INSTANCED_PRIMITIVE_LIST_opcode     34
-#define V3D33_INDEXED_INSTANCED_PRIMITIVE_LIST_header\
+#define V3D33_INDEXED_INSTANCED_PRIM_LIST_opcode     34
+#define V3D33_INDEXED_INSTANCED_PRIM_LIST_header\
    .opcode                              =     34
 
-struct V3D33_INDEXED_INSTANCED_PRIMITIVE_LIST {
+struct V3D33_INDEXED_INSTANCED_PRIM_LIST {
    uint32_t                             opcode;
    bool                                 enable_primitive_restarts;
    uint32_t                             maximum_index;
@@ -1135,8 +1281,8 @@ struct V3D33_INDEXED_INSTANCED_PRIMITIVE_LIST {
 };
 
 static inline void
-V3D33_INDEXED_INSTANCED_PRIMITIVE_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                            const struct V3D33_INDEXED_INSTANCED_PRIMITIVE_LIST * restrict values)
+V3D33_INDEXED_INSTANCED_PRIM_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                       const struct V3D33_INDEXED_INSTANCED_PRIM_LIST * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -1167,11 +1313,11 @@ V3D33_INDEXED_INSTANCED_PRIMITIVE_LIST_pack(__gen_user_data *data, uint8_t * res
 
 }
 
-#define V3D33_INDEXED_INSTANCED_PRIMITIVE_LIST_length     18
+#define V3D33_INDEXED_INSTANCED_PRIM_LIST_length     18
 #ifdef __gen_unpack_address
 static inline void
-V3D33_INDEXED_INSTANCED_PRIMITIVE_LIST_unpack(const uint8_t * restrict cl,
-                                              struct V3D33_INDEXED_INSTANCED_PRIMITIVE_LIST * restrict values)
+V3D33_INDEXED_INSTANCED_PRIM_LIST_unpack(const uint8_t * restrict cl,
+                                         struct V3D33_INDEXED_INSTANCED_PRIM_LIST * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->enable_primitive_restarts = __gen_unpack_uint(cl, 143, 143);
@@ -1185,11 +1331,11 @@ V3D33_INDEXED_INSTANCED_PRIMITIVE_LIST_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_VERTEX_ARRAY_PRIMITIVES_opcode     36
-#define V3D33_VERTEX_ARRAY_PRIMITIVES_header    \
+#define V3D33_VERTEX_ARRAY_PRIMS_opcode       36
+#define V3D33_VERTEX_ARRAY_PRIMS_header         \
    .opcode                              =     36
 
-struct V3D33_VERTEX_ARRAY_PRIMITIVES {
+struct V3D33_VERTEX_ARRAY_PRIMS {
    uint32_t                             opcode;
    uint32_t                             index_of_first_vertex;
    uint32_t                             length;
@@ -1197,8 +1343,8 @@ struct V3D33_VERTEX_ARRAY_PRIMITIVES {
 };
 
 static inline void
-V3D33_VERTEX_ARRAY_PRIMITIVES_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                   const struct V3D33_VERTEX_ARRAY_PRIMITIVES * restrict values)
+V3D33_VERTEX_ARRAY_PRIMS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D33_VERTEX_ARRAY_PRIMS * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -1210,11 +1356,11 @@ V3D33_VERTEX_ARRAY_PRIMITIVES_pack(__gen_user_data *data, uint8_t * restrict cl,
    memcpy(&cl[6], &values->index_of_first_vertex, sizeof(values->index_of_first_vertex));
 }
 
-#define V3D33_VERTEX_ARRAY_PRIMITIVES_length     10
+#define V3D33_VERTEX_ARRAY_PRIMS_length       10
 #ifdef __gen_unpack_address
 static inline void
-V3D33_VERTEX_ARRAY_PRIMITIVES_unpack(const uint8_t * restrict cl,
-                                     struct V3D33_VERTEX_ARRAY_PRIMITIVES * restrict values)
+V3D33_VERTEX_ARRAY_PRIMS_unpack(const uint8_t * restrict cl,
+                                struct V3D33_VERTEX_ARRAY_PRIMS * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->index_of_first_vertex = __gen_unpack_uint(cl, 48, 79);
@@ -1224,11 +1370,11 @@ V3D33_VERTEX_ARRAY_PRIMITIVES_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_VERTEX_ARRAY_INSTANCED_PRIMITIVES_opcode     38
-#define V3D33_VERTEX_ARRAY_INSTANCED_PRIMITIVES_header\
+#define V3D33_VERTEX_ARRAY_INSTANCED_PRIMS_opcode     38
+#define V3D33_VERTEX_ARRAY_INSTANCED_PRIMS_header\
    .opcode                              =     38
 
-struct V3D33_VERTEX_ARRAY_INSTANCED_PRIMITIVES {
+struct V3D33_VERTEX_ARRAY_INSTANCED_PRIMS {
    uint32_t                             opcode;
    uint32_t                             index_of_first_vertex;
    uint32_t                             number_of_instances;
@@ -1237,8 +1383,8 @@ struct V3D33_VERTEX_ARRAY_INSTANCED_PRIMITIVES {
 };
 
 static inline void
-V3D33_VERTEX_ARRAY_INSTANCED_PRIMITIVES_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                             const struct V3D33_VERTEX_ARRAY_INSTANCED_PRIMITIVES * restrict values)
+V3D33_VERTEX_ARRAY_INSTANCED_PRIMS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D33_VERTEX_ARRAY_INSTANCED_PRIMS * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -1252,11 +1398,11 @@ V3D33_VERTEX_ARRAY_INSTANCED_PRIMITIVES_pack(__gen_user_data *data, uint8_t * re
    memcpy(&cl[10], &values->index_of_first_vertex, sizeof(values->index_of_first_vertex));
 }
 
-#define V3D33_VERTEX_ARRAY_INSTANCED_PRIMITIVES_length     14
+#define V3D33_VERTEX_ARRAY_INSTANCED_PRIMS_length     14
 #ifdef __gen_unpack_address
 static inline void
-V3D33_VERTEX_ARRAY_INSTANCED_PRIMITIVES_unpack(const uint8_t * restrict cl,
-                                               struct V3D33_VERTEX_ARRAY_INSTANCED_PRIMITIVES * restrict values)
+V3D33_VERTEX_ARRAY_INSTANCED_PRIMS_unpack(const uint8_t * restrict cl,
+                                          struct V3D33_VERTEX_ARRAY_INSTANCED_PRIMS * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->index_of_first_vertex = __gen_unpack_uint(cl, 80, 111);
@@ -1302,15 +1448,13 @@ V3D33_BASE_VERTEX_BASE_INSTANCE_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_PRIMITIVE_LIST_FORMAT_opcode     56
-#define V3D33_PRIMITIVE_LIST_FORMAT_header      \
+#define V3D33_PRIM_LIST_FORMAT_opcode         56
+#define V3D33_PRIM_LIST_FORMAT_header           \
    .opcode                              =     56
 
-struct V3D33_PRIMITIVE_LIST_FORMAT {
+struct V3D33_PRIM_LIST_FORMAT {
    uint32_t                             opcode;
-   uint32_t                             data_type;
-#define LIST_INDEXED                             0
-#define LIST_32_BIT_X_Y                          1
+   bool                                 tri_strip_or_fan;
    uint32_t                             primitive_type;
 #define LIST_POINTS                              0
 #define LIST_LINES                               1
@@ -1318,24 +1462,24 @@ struct V3D33_PRIMITIVE_LIST_FORMAT {
 };
 
 static inline void
-V3D33_PRIMITIVE_LIST_FORMAT_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                 const struct V3D33_PRIMITIVE_LIST_FORMAT * restrict values)
+V3D33_PRIM_LIST_FORMAT_pack(__gen_user_data *data, uint8_t * restrict cl,
+                            const struct V3D33_PRIM_LIST_FORMAT * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
-   cl[ 1] = __gen_uint(values->data_type, 6, 6) |
+   cl[ 1] = __gen_uint(values->tri_strip_or_fan, 7, 7) |
             __gen_uint(values->primitive_type, 0, 5);
 
 }
 
-#define V3D33_PRIMITIVE_LIST_FORMAT_length      2
+#define V3D33_PRIM_LIST_FORMAT_length          2
 #ifdef __gen_unpack_address
 static inline void
-V3D33_PRIMITIVE_LIST_FORMAT_unpack(const uint8_t * restrict cl,
-                                   struct V3D33_PRIMITIVE_LIST_FORMAT * restrict values)
+V3D33_PRIM_LIST_FORMAT_unpack(const uint8_t * restrict cl,
+                              struct V3D33_PRIM_LIST_FORMAT * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
-   values->data_type = __gen_unpack_uint(cl, 14, 14);
+   values->tri_strip_or_fan = __gen_unpack_uint(cl, 15, 15);
    values->primitive_type = __gen_unpack_uint(cl, 8, 13);
 }
 #endif
@@ -1382,6 +1526,40 @@ V3D33_GL_SHADER_STATE_unpack(const uint8_t * restrict cl,
 #endif
 
 
+#define V3D33_VCM_CACHE_SIZE_opcode           73
+#define V3D33_VCM_CACHE_SIZE_header             \
+   .opcode                              =     73
+
+struct V3D33_VCM_CACHE_SIZE {
+   uint32_t                             opcode;
+   uint32_t                             number_of_16_vertex_batches_for_rendering;
+   uint32_t                             number_of_16_vertex_batches_for_binning;
+};
+
+static inline void
+V3D33_VCM_CACHE_SIZE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                          const struct V3D33_VCM_CACHE_SIZE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->number_of_16_vertex_batches_for_rendering, 4, 7) |
+            __gen_uint(values->number_of_16_vertex_batches_for_binning, 0, 3);
+
+}
+
+#define V3D33_VCM_CACHE_SIZE_length            2
+#ifdef __gen_unpack_address
+static inline void
+V3D33_VCM_CACHE_SIZE_unpack(const uint8_t * restrict cl,
+                            struct V3D33_VCM_CACHE_SIZE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->number_of_16_vertex_batches_for_rendering = __gen_unpack_uint(cl, 12, 15);
+   values->number_of_16_vertex_batches_for_binning = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
 #define V3D33_TRANSFORM_FEEDBACK_ENABLE_opcode     74
 #define V3D33_TRANSFORM_FEEDBACK_ENABLE_header  \
    .opcode                              =     74
@@ -1449,7 +1627,7 @@ V3D33_FLUSH_TRANSFORM_FEEDBACK_DATA_unpack(const uint8_t * restrict cl,
 
 struct V3D33_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC {
    uint32_t                             first_shaded_vertex_value_to_output;
-   uint32_t                             number_of_consecutive_vertex_values_to_output_as_32_bit_values_minus_1;
+   uint32_t                             number_of_consecutive_vertex_values_to_output_as_32_bit_values;
    uint32_t                             output_buffer_to_write_to;
 };
 
@@ -1457,9 +1635,10 @@ static inline void
 V3D33_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC_pack(__gen_user_data *data, uint8_t * restrict cl,
                                                const struct V3D33_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC * restrict values)
 {
+   assert(values->number_of_consecutive_vertex_values_to_output_as_32_bit_values >= 1);
    cl[ 0] = __gen_uint(values->first_shaded_vertex_value_to_output, 0, 7);
 
-   cl[ 1] = __gen_uint(values->number_of_consecutive_vertex_values_to_output_as_32_bit_values_minus_1, 0, 3) |
+   cl[ 1] = __gen_uint(values->number_of_consecutive_vertex_values_to_output_as_32_bit_values - 1, 0, 3) |
             __gen_uint(values->output_buffer_to_write_to, 4, 5);
 
 }
@@ -1471,7 +1650,7 @@ V3D33_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC_unpack(const uint8_t * restrict cl,
                                                  struct V3D33_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC * restrict values)
 {
    values->first_shaded_vertex_value_to_output = __gen_unpack_uint(cl, 0, 7);
-   values->number_of_consecutive_vertex_values_to_output_as_32_bit_values_minus_1 = __gen_unpack_uint(cl, 8, 11);
+   values->number_of_consecutive_vertex_values_to_output_as_32_bit_values = __gen_unpack_uint(cl, 8, 11) + 1;
    values->output_buffer_to_write_to = __gen_unpack_uint(cl, 12, 13);
 }
 #endif
@@ -1510,11 +1689,11 @@ V3D33_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_STENCIL_CONFIG_opcode           80
-#define V3D33_STENCIL_CONFIG_header             \
+#define V3D33_STENCIL_CFG_opcode              80
+#define V3D33_STENCIL_CFG_header                \
    .opcode                              =     80
 
-struct V3D33_STENCIL_CONFIG {
+struct V3D33_STENCIL_CFG {
    uint32_t                             opcode;
    uint32_t                             stencil_write_mask;
    bool                                 back_config;
@@ -1528,8 +1707,8 @@ struct V3D33_STENCIL_CONFIG {
 };
 
 static inline void
-V3D33_STENCIL_CONFIG_pack(__gen_user_data *data, uint8_t * restrict cl,
-                          const struct V3D33_STENCIL_CONFIG * restrict values)
+V3D33_STENCIL_CFG_pack(__gen_user_data *data, uint8_t * restrict cl,
+                       const struct V3D33_STENCIL_CFG * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -1550,11 +1729,11 @@ V3D33_STENCIL_CONFIG_pack(__gen_user_data *data, uint8_t * restrict cl,
 
 }
 
-#define V3D33_STENCIL_CONFIG_length            6
+#define V3D33_STENCIL_CFG_length               6
 #ifdef __gen_unpack_address
 static inline void
-V3D33_STENCIL_CONFIG_unpack(const uint8_t * restrict cl,
-                            struct V3D33_STENCIL_CONFIG * restrict values)
+V3D33_STENCIL_CFG_unpack(const uint8_t * restrict cl,
+                         struct V3D33_STENCIL_CFG * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->stencil_write_mask = __gen_unpack_uint(cl, 40, 47);
@@ -1570,51 +1749,47 @@ V3D33_STENCIL_CONFIG_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_BLEND_CONFIG_opcode             84
-#define V3D33_BLEND_CONFIG_header               \
+#define V3D33_BLEND_CFG_opcode                84
+#define V3D33_BLEND_CFG_header                  \
    .opcode                              =     84
 
-struct V3D33_BLEND_CONFIG {
+struct V3D33_BLEND_CFG {
    uint32_t                             opcode;
-   uint32_t                             vg_coverage_modes;
-   enum V3D33_Blend_Factor              colour_blend_dst_factor;
-   enum V3D33_Blend_Factor              colour_blend_src_factor;
-   enum V3D33_Blend_Mode                colour_blend_mode;
+   enum V3D33_Blend_Factor              color_blend_dst_factor;
+   enum V3D33_Blend_Factor              color_blend_src_factor;
+   enum V3D33_Blend_Mode                color_blend_mode;
    enum V3D33_Blend_Factor              alpha_blend_dst_factor;
    enum V3D33_Blend_Factor              alpha_blend_src_factor;
    enum V3D33_Blend_Mode                alpha_blend_mode;
 };
 
 static inline void
-V3D33_BLEND_CONFIG_pack(__gen_user_data *data, uint8_t * restrict cl,
-                        const struct V3D33_BLEND_CONFIG * restrict values)
+V3D33_BLEND_CFG_pack(__gen_user_data *data, uint8_t * restrict cl,
+                     const struct V3D33_BLEND_CFG * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
    cl[ 1] = __gen_uint(values->alpha_blend_src_factor, 4, 7) |
             __gen_uint(values->alpha_blend_mode, 0, 3);
 
-   cl[ 2] = __gen_uint(values->colour_blend_mode, 4, 7) |
+   cl[ 2] = __gen_uint(values->color_blend_mode, 4, 7) |
             __gen_uint(values->alpha_blend_dst_factor, 0, 3);
 
-   cl[ 3] = __gen_uint(values->colour_blend_dst_factor, 4, 7) |
-            __gen_uint(values->colour_blend_src_factor, 0, 3);
-
-   cl[ 4] = __gen_uint(values->vg_coverage_modes, 4, 5);
+   cl[ 3] = __gen_uint(values->color_blend_dst_factor, 4, 7) |
+            __gen_uint(values->color_blend_src_factor, 0, 3);
 
 }
 
-#define V3D33_BLEND_CONFIG_length              5
+#define V3D33_BLEND_CFG_length                 4
 #ifdef __gen_unpack_address
 static inline void
-V3D33_BLEND_CONFIG_unpack(const uint8_t * restrict cl,
-                          struct V3D33_BLEND_CONFIG * restrict values)
+V3D33_BLEND_CFG_unpack(const uint8_t * restrict cl,
+                       struct V3D33_BLEND_CFG * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
-   values->vg_coverage_modes = __gen_unpack_uint(cl, 36, 37);
-   values->colour_blend_dst_factor = __gen_unpack_uint(cl, 28, 31);
-   values->colour_blend_src_factor = __gen_unpack_uint(cl, 24, 27);
-   values->colour_blend_mode = __gen_unpack_uint(cl, 20, 23);
+   values->color_blend_dst_factor = __gen_unpack_uint(cl, 28, 31);
+   values->color_blend_src_factor = __gen_unpack_uint(cl, 24, 27);
+   values->color_blend_mode = __gen_unpack_uint(cl, 20, 23);
    values->alpha_blend_dst_factor = __gen_unpack_uint(cl, 16, 19);
    values->alpha_blend_src_factor = __gen_unpack_uint(cl, 12, 15);
    values->alpha_blend_mode = __gen_unpack_uint(cl, 8, 11);
@@ -1622,11 +1797,11 @@ V3D33_BLEND_CONFIG_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_BLEND_CONSTANT_COLOUR_opcode     86
-#define V3D33_BLEND_CONSTANT_COLOUR_header      \
+#define V3D33_BLEND_CONSTANT_COLOR_opcode     86
+#define V3D33_BLEND_CONSTANT_COLOR_header       \
    .opcode                              =     86
 
-struct V3D33_BLEND_CONSTANT_COLOUR {
+struct V3D33_BLEND_CONSTANT_COLOR {
    uint32_t                             opcode;
    uint32_t                             alpha_f16;
    uint32_t                             blue_f16;
@@ -1635,8 +1810,8 @@ struct V3D33_BLEND_CONSTANT_COLOUR {
 };
 
 static inline void
-V3D33_BLEND_CONSTANT_COLOUR_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                 const struct V3D33_BLEND_CONSTANT_COLOUR * restrict values)
+V3D33_BLEND_CONSTANT_COLOR_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                const struct V3D33_BLEND_CONSTANT_COLOR * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -1658,11 +1833,11 @@ V3D33_BLEND_CONSTANT_COLOUR_pack(__gen_user_data *data, uint8_t * restrict cl,
 
 }
 
-#define V3D33_BLEND_CONSTANT_COLOUR_length      9
+#define V3D33_BLEND_CONSTANT_COLOR_length      9
 #ifdef __gen_unpack_address
 static inline void
-V3D33_BLEND_CONSTANT_COLOUR_unpack(const uint8_t * restrict cl,
-                                   struct V3D33_BLEND_CONSTANT_COLOUR * restrict values)
+V3D33_BLEND_CONSTANT_COLOR_unpack(const uint8_t * restrict cl,
+                                  struct V3D33_BLEND_CONSTANT_COLOR * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->alpha_f16 = __gen_unpack_uint(cl, 56, 71);
@@ -1673,58 +1848,80 @@ V3D33_BLEND_CONSTANT_COLOUR_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_COLOUR_WRITE_MASKS_opcode       87
-#define V3D33_COLOUR_WRITE_MASKS_header         \
+#define V3D33_COLOR_WRITE_MASKS_opcode        87
+#define V3D33_COLOR_WRITE_MASKS_header          \
    .opcode                              =     87
 
-struct V3D33_COLOUR_WRITE_MASKS {
+struct V3D33_COLOR_WRITE_MASKS {
+   uint32_t                             opcode;
+   uint32_t                             mask;
+};
+
+static inline void
+V3D33_COLOR_WRITE_MASKS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                             const struct V3D33_COLOR_WRITE_MASKS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->mask, sizeof(values->mask));
+}
+
+#define V3D33_COLOR_WRITE_MASKS_length         5
+#ifdef __gen_unpack_address
+static inline void
+V3D33_COLOR_WRITE_MASKS_unpack(const uint8_t * restrict cl,
+                               struct V3D33_COLOR_WRITE_MASKS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->mask = __gen_unpack_uint(cl, 8, 39);
+}
+#endif
+
+
+#define V3D33_OCCLUSION_QUERY_COUNTER_opcode     92
+#define V3D33_OCCLUSION_QUERY_COUNTER_header    \
+   .opcode                              =     92
+
+struct V3D33_OCCLUSION_QUERY_COUNTER {
    uint32_t                             opcode;
-   uint32_t                             reserved;
-   uint32_t                             render_target_3_per_colour_component_write_masks;
-   uint32_t                             render_target_2_per_colour_component_write_masks;
-   uint32_t                             render_target_1_per_colour_component_write_masks;
-   uint32_t                             render_target_0_per_colour_component_write_masks;
+   __gen_address_type                   address;
 };
 
 static inline void
-V3D33_COLOUR_WRITE_MASKS_pack(__gen_user_data *data, uint8_t * restrict cl,
-                              const struct V3D33_COLOUR_WRITE_MASKS * restrict values)
+V3D33_OCCLUSION_QUERY_COUNTER_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                   const struct V3D33_OCCLUSION_QUERY_COUNTER * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
-   cl[ 1] = __gen_uint(values->render_target_1_per_colour_component_write_masks, 4, 7) |
-            __gen_uint(values->render_target_0_per_colour_component_write_masks, 0, 3);
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address);
 
-   cl[ 2] = __gen_uint(values->render_target_3_per_colour_component_write_masks, 4, 7) |
-            __gen_uint(values->render_target_2_per_colour_component_write_masks, 0, 3);
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
 
-   cl[ 3] = __gen_uint(values->reserved, 0, 15);
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
 
-   cl[ 4] = __gen_uint(values->reserved, 0, 15) >> 8;
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
 
 }
 
-#define V3D33_COLOUR_WRITE_MASKS_length        5
+#define V3D33_OCCLUSION_QUERY_COUNTER_length      5
 #ifdef __gen_unpack_address
 static inline void
-V3D33_COLOUR_WRITE_MASKS_unpack(const uint8_t * restrict cl,
-                                struct V3D33_COLOUR_WRITE_MASKS * restrict values)
+V3D33_OCCLUSION_QUERY_COUNTER_unpack(const uint8_t * restrict cl,
+                                     struct V3D33_OCCLUSION_QUERY_COUNTER * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
-   values->reserved = __gen_unpack_uint(cl, 24, 39);
-   values->render_target_3_per_colour_component_write_masks = __gen_unpack_uint(cl, 20, 23);
-   values->render_target_2_per_colour_component_write_masks = __gen_unpack_uint(cl, 16, 19);
-   values->render_target_1_per_colour_component_write_masks = __gen_unpack_uint(cl, 12, 15);
-   values->render_target_0_per_colour_component_write_masks = __gen_unpack_uint(cl, 8, 11);
+   values->address = __gen_unpack_address(cl, 8, 39);
 }
 #endif
 
 
-#define V3D33_CONFIGURATION_BITS_opcode       96
-#define V3D33_CONFIGURATION_BITS_header         \
+#define V3D33_CFG_BITS_opcode                 96
+#define V3D33_CFG_BITS_header                   \
    .opcode                              =     96
 
-struct V3D33_CONFIGURATION_BITS {
+struct V3D33_CFG_BITS {
    uint32_t                             opcode;
    bool                                 direct3d_provoking_vertex;
    bool                                 direct3d_point_fill_mode;
@@ -1735,8 +1932,6 @@ struct V3D33_CONFIGURATION_BITS {
    bool                                 z_updates_enable;
    enum V3D33_Compare_Function          depth_test_function;
    bool                                 direct3d_wireframe_triangles_mode;
-   uint32_t                             coverage_update_mode;
-   bool                                 coverage_pipe_select;
    uint32_t                             rasterizer_oversample_mode;
    uint32_t                             line_rasterization;
    bool                                 enable_depth_offset;
@@ -1746,8 +1941,8 @@ struct V3D33_CONFIGURATION_BITS {
 };
 
 static inline void
-V3D33_CONFIGURATION_BITS_pack(__gen_user_data *data, uint8_t * restrict cl,
-                              const struct V3D33_CONFIGURATION_BITS * restrict values)
+V3D33_CFG_BITS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                    const struct V3D33_CFG_BITS * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -1760,9 +1955,7 @@ V3D33_CONFIGURATION_BITS_pack(__gen_user_data *data, uint8_t * restrict cl,
 
    cl[ 2] = __gen_uint(values->z_updates_enable, 7, 7) |
             __gen_uint(values->depth_test_function, 4, 6) |
-            __gen_uint(values->direct3d_wireframe_triangles_mode, 3, 3) |
-            __gen_uint(values->coverage_update_mode, 1, 2) |
-            __gen_uint(values->coverage_pipe_select, 0, 0);
+            __gen_uint(values->direct3d_wireframe_triangles_mode, 3, 3);
 
    cl[ 3] = __gen_uint(values->direct3d_provoking_vertex, 5, 5) |
             __gen_uint(values->direct3d_point_fill_mode, 4, 4) |
@@ -1773,11 +1966,11 @@ V3D33_CONFIGURATION_BITS_pack(__gen_user_data *data, uint8_t * restrict cl,
 
 }
 
-#define V3D33_CONFIGURATION_BITS_length        4
+#define V3D33_CFG_BITS_length                  4
 #ifdef __gen_unpack_address
 static inline void
-V3D33_CONFIGURATION_BITS_unpack(const uint8_t * restrict cl,
-                                struct V3D33_CONFIGURATION_BITS * restrict values)
+V3D33_CFG_BITS_unpack(const uint8_t * restrict cl,
+                      struct V3D33_CFG_BITS * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->direct3d_provoking_vertex = __gen_unpack_uint(cl, 29, 29);
@@ -1789,8 +1982,6 @@ V3D33_CONFIGURATION_BITS_unpack(const uint8_t * restrict cl,
    values->z_updates_enable = __gen_unpack_uint(cl, 23, 23);
    values->depth_test_function = __gen_unpack_uint(cl, 20, 22);
    values->direct3d_wireframe_triangles_mode = __gen_unpack_uint(cl, 19, 19);
-   values->coverage_update_mode = __gen_unpack_uint(cl, 17, 18);
-   values->coverage_pipe_select = __gen_unpack_uint(cl, 16, 16);
    values->rasterizer_oversample_mode = __gen_unpack_uint(cl, 14, 15);
    values->line_rasterization = __gen_unpack_uint(cl, 12, 13);
    values->enable_depth_offset = __gen_unpack_uint(cl, 11, 11);
@@ -1835,8 +2026,8 @@ V3D33_ZERO_ALL_FLAT_SHADE_FLAGS_unpack(const uint8_t * restrict cl,
 struct V3D33_FLAT_SHADE_FLAGS {
    uint32_t                             opcode;
    uint32_t                             flat_shade_flags_for_varyings_v024;
-   uint32_t                             action_for_flat_shade_flags_of_higher_numbered_varyings;
-   uint32_t                             action_for_flat_shade_flags_of_lower_numbered_varyings;
+   enum V3D33_Varying_Flags_Action      action_for_flat_shade_flags_of_higher_numbered_varyings;
+   enum V3D33_Varying_Flags_Action      action_for_flat_shade_flags_of_lower_numbered_varyings;
    uint32_t                             varying_offset_v0;
 };
 
@@ -1941,8 +2132,8 @@ V3D33_LINE_WIDTH_unpack(const uint8_t * restrict cl,
 
 struct V3D33_DEPTH_OFFSET {
    uint32_t                             opcode;
-   uint32_t                             depth_offset_units;
-   uint32_t                             depth_offset_factor;
+   float                                depth_offset_units;
+   float                                depth_offset_factor;
 };
 
 static inline void
@@ -1951,13 +2142,13 @@ V3D33_DEPTH_OFFSET_pack(__gen_user_data *data, uint8_t * restrict cl,
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
-   cl[ 1] = __gen_uint(values->depth_offset_factor, 0, 15);
+   cl[ 1] = __gen_uint(fui(values->depth_offset_factor) >> 16, 0, 15);
 
-   cl[ 2] = __gen_uint(values->depth_offset_factor, 0, 15) >> 8;
+   cl[ 2] = __gen_uint(fui(values->depth_offset_factor) >> 16, 0, 15) >> 8;
 
-   cl[ 3] = __gen_uint(values->depth_offset_units, 0, 15);
+   cl[ 3] = __gen_uint(fui(values->depth_offset_units) >> 16, 0, 15);
 
-   cl[ 4] = __gen_uint(values->depth_offset_units, 0, 15) >> 8;
+   cl[ 4] = __gen_uint(fui(values->depth_offset_units) >> 16, 0, 15) >> 8;
 
 }
 
@@ -1968,8 +2159,8 @@ V3D33_DEPTH_OFFSET_unpack(const uint8_t * restrict cl,
                           struct V3D33_DEPTH_OFFSET * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
-   values->depth_offset_units = __gen_unpack_uint(cl, 24, 39);
-   values->depth_offset_factor = __gen_unpack_uint(cl, 8, 23);
+   values->depth_offset_units = __gen_unpack_f187(cl, 24, 39);
+   values->depth_offset_factor = __gen_unpack_f187(cl, 8, 23);
 }
 #endif
 
@@ -2177,20 +2368,17 @@ V3D33_CLIPPER_Z_SCALE_AND_OFFSET_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1_opcode    120
-#define V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1_header\
+#define V3D33_TILE_BINNING_MODE_CFG_PART1_opcode    120
+#define V3D33_TILE_BINNING_MODE_CFG_PART1_header\
    .opcode                              =    120,  \
    .auto_initialize_tile_state_data_array =      1,  \
    .sub_id                              =      0
 
-struct V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1 {
+struct V3D33_TILE_BINNING_MODE_CFG_PART1 {
    uint32_t                             opcode;
    bool                                 double_buffer_in_non_ms_mode;
    bool                                 multisample_mode_4x;
-   uint32_t                             maximum_bpp_of_all_render_targets;
-#define RENDER_TARGET_MAXIMUM_32BPP              0
-#define RENDER_TARGET_MAXIMUM_64BPP              1
-#define RENDER_TARGET_MAXIMUM_128BPP             2
+   enum V3D33_Internal_BPP              maximum_bpp_of_all_render_targets;
    uint32_t                             number_of_render_targets;
    uint32_t                             height_in_tiles;
    uint32_t                             width_in_tiles;
@@ -2208,8 +2396,8 @@ struct V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1 {
 };
 
 static inline void
-V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                                 const struct V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1 * restrict values)
+V3D33_TILE_BINNING_MODE_CFG_PART1_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                       const struct V3D33_TILE_BINNING_MODE_CFG_PART1 * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -2240,11 +2428,11 @@ V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1_pack(__gen_user_data *data, uint8_t
 
 }
 
-#define V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1_length      9
+#define V3D33_TILE_BINNING_MODE_CFG_PART1_length      9
 #ifdef __gen_unpack_address
 static inline void
-V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1_unpack(const uint8_t * restrict cl,
-                                                   struct V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1 * restrict values)
+V3D33_TILE_BINNING_MODE_CFG_PART1_unpack(const uint8_t * restrict cl,
+                                         struct V3D33_TILE_BINNING_MODE_CFG_PART1 * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->double_buffer_in_non_ms_mode = __gen_unpack_uint(cl, 71, 71);
@@ -2253,7 +2441,7 @@ V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1_unpack(const uint8_t * restrict cl,
    values->number_of_render_targets = __gen_unpack_uint(cl, 64, 67);
    values->height_in_tiles = __gen_unpack_uint(cl, 52, 63);
    values->width_in_tiles = __gen_unpack_uint(cl, 40, 51);
-   values->tile_state_data_array_base_address = __gen_unpack_address(cl, 8, 39);
+   values->tile_state_data_array_base_address = __gen_unpack_address(cl, 14, 39);
    values->tile_allocation_block_size = __gen_unpack_uint(cl, 12, 13);
    values->tile_allocation_initial_block_size = __gen_unpack_uint(cl, 10, 11);
    values->auto_initialize_tile_state_data_array = __gen_unpack_uint(cl, 9, 9);
@@ -2262,12 +2450,12 @@ V3D33_TILE_BINNING_MODE_CONFIGURATION_PART1_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_TILE_BINNING_MODE_CONFIGURATION_PART2_opcode    120
-#define V3D33_TILE_BINNING_MODE_CONFIGURATION_PART2_header\
+#define V3D33_TILE_BINNING_MODE_CFG_PART2_opcode    120
+#define V3D33_TILE_BINNING_MODE_CFG_PART2_header\
    .opcode                              =    120,  \
    .sub_id                              =      1
 
-struct V3D33_TILE_BINNING_MODE_CONFIGURATION_PART2 {
+struct V3D33_TILE_BINNING_MODE_CFG_PART2 {
    uint32_t                             opcode;
    __gen_address_type                   tile_allocation_memory_address;
    uint32_t                             tile_allocation_memory_size;
@@ -2275,8 +2463,8 @@ struct V3D33_TILE_BINNING_MODE_CONFIGURATION_PART2 {
 };
 
 static inline void
-V3D33_TILE_BINNING_MODE_CONFIGURATION_PART2_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                                 const struct V3D33_TILE_BINNING_MODE_CONFIGURATION_PART2 * restrict values)
+V3D33_TILE_BINNING_MODE_CFG_PART2_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                       const struct V3D33_TILE_BINNING_MODE_CFG_PART2 * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -2300,11 +2488,11 @@ V3D33_TILE_BINNING_MODE_CONFIGURATION_PART2_pack(__gen_user_data *data, uint8_t
 
 }
 
-#define V3D33_TILE_BINNING_MODE_CONFIGURATION_PART2_length      9
+#define V3D33_TILE_BINNING_MODE_CFG_PART2_length      9
 #ifdef __gen_unpack_address
 static inline void
-V3D33_TILE_BINNING_MODE_CONFIGURATION_PART2_unpack(const uint8_t * restrict cl,
-                                                   struct V3D33_TILE_BINNING_MODE_CONFIGURATION_PART2 * restrict values)
+V3D33_TILE_BINNING_MODE_CFG_PART2_unpack(const uint8_t * restrict cl,
+                                         struct V3D33_TILE_BINNING_MODE_CFG_PART2 * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->tile_allocation_memory_address = __gen_unpack_address(cl, 40, 71);
@@ -2314,12 +2502,12 @@ V3D33_TILE_BINNING_MODE_CONFIGURATION_PART2_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION_opcode    121
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION_header\
+#define V3D33_TILE_RENDERING_MODE_CFG_COMMON_opcode    121
+#define V3D33_TILE_RENDERING_MODE_CFG_COMMON_header\
    .opcode                              =    121,  \
    .sub_id                              =      0
 
-struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION {
+struct V3D33_TILE_RENDERING_MODE_CFG_COMMON {
    uint32_t                             opcode;
    uint32_t                             disable_render_target_stores;
    bool                                 enable_z_store;
@@ -2328,7 +2516,6 @@ struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION {
    uint32_t                             early_z_test_and_update_direction;
 #define EARLY_Z_DIRECTION_LT_LE                  0
 #define EARLY_Z_DIRECTION_GT_GE                  1
-   bool                                 select_coverage_mode;
    bool                                 double_buffer_in_non_ms_mode;
    bool                                 multisample_mode_4x;
    uint32_t                             maximum_bpp_of_all_render_targets;
@@ -2337,17 +2524,18 @@ struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION {
 #define RENDER_TARGET_MAXIMUM_128BPP             2
    uint32_t                             image_height_pixels;
    uint32_t                             image_width_pixels;
-   uint32_t                             number_of_render_targets_minus_1;
+   uint32_t                             number_of_render_targets;
    uint32_t                             sub_id;
 };
 
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                                                  const struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_COMMON_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                          const struct V3D33_TILE_RENDERING_MODE_CFG_COMMON * restrict values)
 {
+   assert(values->number_of_render_targets >= 1);
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
-   cl[ 1] = __gen_uint(values->number_of_render_targets_minus_1, 4, 7) |
+   cl[ 1] = __gen_uint(values->number_of_render_targets - 1, 4, 7) |
             __gen_uint(values->sub_id, 0, 3);
 
    cl[ 2] = __gen_uint(values->image_width_pixels, 0, 15);
@@ -2360,7 +2548,6 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION_pack(__gen_user_dat
 
    cl[ 6] = __gen_uint(values->early_z_disable, 6, 6) |
             __gen_uint(values->early_z_test_and_update_direction, 5, 5) |
-            __gen_uint(values->select_coverage_mode, 4, 4) |
             __gen_uint(values->double_buffer_in_non_ms_mode, 3, 3) |
             __gen_uint(values->multisample_mode_4x, 2, 2) |
             __gen_uint(values->maximum_bpp_of_all_render_targets, 0, 1);
@@ -2372,11 +2559,11 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION_pack(__gen_user_dat
 
 }
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION_length      9
+#define V3D33_TILE_RENDERING_MODE_CFG_COMMON_length      9
 #ifdef __gen_unpack_address
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION_unpack(const uint8_t * restrict cl,
-                                                                    struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_COMMON_unpack(const uint8_t * restrict cl,
+                                            struct V3D33_TILE_RENDERING_MODE_CFG_COMMON * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->disable_render_target_stores = __gen_unpack_uint(cl, 64, 71);
@@ -2384,99 +2571,40 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_COMMON_CONFIGURATION_unpack(const uint8_
    values->enable_stencil_store = __gen_unpack_uint(cl, 62, 62);
    values->early_z_disable = __gen_unpack_uint(cl, 54, 54);
    values->early_z_test_and_update_direction = __gen_unpack_uint(cl, 53, 53);
-   values->select_coverage_mode = __gen_unpack_uint(cl, 52, 52);
    values->double_buffer_in_non_ms_mode = __gen_unpack_uint(cl, 51, 51);
    values->multisample_mode_4x = __gen_unpack_uint(cl, 50, 50);
    values->maximum_bpp_of_all_render_targets = __gen_unpack_uint(cl, 48, 49);
    values->image_height_pixels = __gen_unpack_uint(cl, 32, 47);
    values->image_width_pixels = __gen_unpack_uint(cl, 16, 31);
-   values->number_of_render_targets_minus_1 = __gen_unpack_uint(cl, 12, 15);
+   values->number_of_render_targets = __gen_unpack_uint(cl, 12, 15) + 1;
    values->sub_id = __gen_unpack_uint(cl, 8, 11);
 }
 #endif
 
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_RENDER_TARGET_CONFIG_opcode    121
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_RENDER_TARGET_CONFIG_header\
+#define V3D33_TILE_RENDERING_MODE_CFG_COLOR_opcode    121
+#define V3D33_TILE_RENDERING_MODE_CFG_COLOR_header\
    .opcode                              =    121,  \
    .sub_id                              =      2
 
-struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_RENDER_TARGET_CONFIG {
+struct V3D33_TILE_RENDERING_MODE_CFG_COLOR {
    uint32_t                             opcode;
    __gen_address_type                   address;
    uint32_t                             pad;
    bool                                 flip_y;
-   uint32_t                             memory_format;
-#define MEMORY_FORMAT_RASTER                     0
-#define MEMORY_FORMAT_LINEARTILE                 1
-#define MEMORY_FORMAT_UB_LINEAR_1_UIF_BLOCK_WIDE 2
-#define MEMORY_FORMAT_UB_LINEAR_2_UIF_BLOCKS_WIDE 3
-#define MEMORY_FORMAT_UIF_NO_XOR                 4
-#define MEMORY_FORMAT_UIF_XOR                    5
-   bool                                 a_dithered;
-   bool                                 bgr_dithered;
-   uint32_t                             output_image_format;
-#define OUTPUT_IMAGE_FORMAT_SRGB8_ALPHA8         0
-#define OUTPUT_IMAGE_FORMAT_SRGB                 1
-#define OUTPUT_IMAGE_FORMAT_RGB10_A2UI           2
-#define OUTPUT_IMAGE_FORMAT_RGB10_A2             3
-#define OUTPUT_IMAGE_FORMAT_ABGR1555             4
-#define OUTPUT_IMAGE_FORMAT_ALPHA_MASKED_ABGR1555 5
-#define OUTPUT_IMAGE_FORMAT_ABGR4444             6
-#define OUTPUT_IMAGE_FORMAT_BGR565               7
-#define OUTPUT_IMAGE_FORMAT_R11F_G11F_B10F       8
-#define OUTPUT_IMAGE_FORMAT_RGBA32F              9
-#define OUTPUT_IMAGE_FORMAT_RG32F                10
-#define OUTPUT_IMAGE_FORMAT_R32F                 11
-#define OUTPUT_IMAGE_FORMAT_RGBA32I              12
-#define OUTPUT_IMAGE_FORMAT_RG32I                13
-#define OUTPUT_IMAGE_FORMAT_R32I                 14
-#define OUTPUT_IMAGE_FORMAT_RGBA32UI             15
-#define OUTPUT_IMAGE_FORMAT_RG32UI               16
-#define OUTPUT_IMAGE_FORMAT_R32UI                17
-#define OUTPUT_IMAGE_FORMAT_RGBA16F              18
-#define OUTPUT_IMAGE_FORMAT_RG16F                19
-#define OUTPUT_IMAGE_FORMAT_R16F                 20
-#define OUTPUT_IMAGE_FORMAT_RGBA16I              21
-#define OUTPUT_IMAGE_FORMAT_RG16I                22
-#define OUTPUT_IMAGE_FORMAT_R16I                 23
-#define OUTPUT_IMAGE_FORMAT_RGBA16UI             24
-#define OUTPUT_IMAGE_FORMAT_RG16UI               25
-#define OUTPUT_IMAGE_FORMAT_R16UI                26
-#define OUTPUT_IMAGE_FORMAT_RGBA8                27
-#define OUTPUT_IMAGE_FORMAT_RGB8                 28
-#define OUTPUT_IMAGE_FORMAT_RG8                  29
-#define OUTPUT_IMAGE_FORMAT_R8                   30
-#define OUTPUT_IMAGE_FORMAT_RGBA8I               31
-#define OUTPUT_IMAGE_FORMAT_RG8I                 32
-#define OUTPUT_IMAGE_FORMAT_R8I                  33
-#define OUTPUT_IMAGE_FORMAT_RGBA8UI              34
-#define OUTPUT_IMAGE_FORMAT_RG8UI                35
-#define OUTPUT_IMAGE_FORMAT_R8UI                 36
-#define OUTPUT_IMAGE_FORMAT_SRGBX8               37
-#define OUTPUT_IMAGE_FORMAT_RGBX8                38
-   uint32_t                             decimate_mode;
-   uint32_t                             internal_type;
-#define INTERNAL_TYPE_8I                         0
-#define INTERNAL_TYPE_8UI                        1
-#define INTERNAL_TYPE_8                          2
-#define INTERNAL_TYPE_16I                        4
-#define INTERNAL_TYPE_16UI                       5
-#define INTERNAL_TYPE_16F                        6
-#define INTERNAL_TYPE_32I                        8
-#define INTERNAL_TYPE_32UI                       9
-#define INTERNAL_TYPE_32F                        10
-   uint32_t                             internal_bpp;
-#define INTERNAL_BPP_32                          0
-#define INTERNAL_BPP_64                          1
-#define INTERNAL_BPP_128                         2
+   enum V3D33_Memory_Format             memory_format;
+   enum V3D33_Dither_Mode               dither_mode;
+   enum V3D33_Output_Image_Format       output_image_format;
+   enum V3D33_Decimate_Mode             decimate_mode;
+   enum V3D33_Internal_Type             internal_type;
+   enum V3D33_Internal_BPP              internal_bpp;
    uint32_t                             render_target_number;
    uint32_t                             sub_id;
 };
 
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_RENDER_TARGET_CONFIG_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                                                  const struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_RENDER_TARGET_CONFIG * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_COLOR_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                         const struct V3D33_TILE_RENDERING_MODE_CFG_COLOR * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -2487,8 +2615,7 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_RENDER_TARGET_CONFIG_pack(__gen_user_dat
             __gen_uint(values->internal_type, 2, 5) |
             __gen_uint(values->internal_bpp, 0, 1);
 
-   cl[ 3] = __gen_uint(values->a_dithered, 7, 7) |
-            __gen_uint(values->bgr_dithered, 6, 6) |
+   cl[ 3] = __gen_uint(values->dither_mode, 6, 7) |
             __gen_uint(values->output_image_format, 0, 5);
 
    cl[ 4] = __gen_uint(values->pad, 4, 7) |
@@ -2506,19 +2633,18 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_RENDER_TARGET_CONFIG_pack(__gen_user_dat
 
 }
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_RENDER_TARGET_CONFIG_length      9
+#define V3D33_TILE_RENDERING_MODE_CFG_COLOR_length      9
 #ifdef __gen_unpack_address
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_RENDER_TARGET_CONFIG_unpack(const uint8_t * restrict cl,
-                                                                    struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_RENDER_TARGET_CONFIG * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_COLOR_unpack(const uint8_t * restrict cl,
+                                           struct V3D33_TILE_RENDERING_MODE_CFG_COLOR * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->address = __gen_unpack_address(cl, 40, 71);
    values->pad = __gen_unpack_uint(cl, 36, 39);
    values->flip_y = __gen_unpack_uint(cl, 35, 35);
    values->memory_format = __gen_unpack_uint(cl, 32, 34);
-   values->a_dithered = __gen_unpack_uint(cl, 31, 31);
-   values->bgr_dithered = __gen_unpack_uint(cl, 30, 30);
+   values->dither_mode = __gen_unpack_uint(cl, 30, 31);
    values->output_image_format = __gen_unpack_uint(cl, 24, 29);
    values->decimate_mode = __gen_unpack_uint(cl, 22, 23);
    values->internal_type = __gen_unpack_uint(cl, 18, 21);
@@ -2529,41 +2655,28 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_RENDER_TARGET_CONFIG_unpack(const uint8_
 #endif
 
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CONFIG_opcode    121
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CONFIG_header\
+#define V3D33_TILE_RENDERING_MODE_CFG_Z_STENCIL_opcode    121
+#define V3D33_TILE_RENDERING_MODE_CFG_Z_STENCIL_header\
    .opcode                              =    121,  \
    .z_stencil_id                        =      0,  \
    .sub_id                              =      1
 
-struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CONFIG {
+struct V3D33_TILE_RENDERING_MODE_CFG_Z_STENCIL {
    uint32_t                             opcode;
    __gen_address_type                   address;
    uint32_t                             padded_height_of_output_image_in_uif_blocks;
-   uint32_t                             memory_format;
-#define MEMORY_FORMAT_RASTER                     0
-#define MEMORY_FORMAT_LINEARTILE                 1
-#define MEMORY_FORMAT_UB_LINEAR_1_UIF_BLOCK_WIDE 2
-#define MEMORY_FORMAT_UB_LINEAR_2_UIF_BLOCKS_WIDE 3
-#define MEMORY_FORMAT_UIF_NO_XOR                 4
-#define MEMORY_FORMAT_UIF_XOR                    5
-   uint32_t                             output_image_format;
-#define OUTPUT_IMAGE_FORMAT_DEPTH_COMPONENT32F   0
-#define OUTPUT_IMAGE_FORMAT_DEPTH_COMPONENT24    1
-#define OUTPUT_IMAGE_FORMAT_DEPTH_COMPONENT16    2
-#define OUTPUT_IMAGE_FORMAT_DEPTH24_STENCIL8     3
+   enum V3D33_Memory_Format             memory_format;
+   enum V3D33_Z_S_Output_Image_Format   output_image_format;
    uint32_t                             decimate_mode;
-   uint32_t                             internal_type;
-#define INTERNAL_TYPE_DEPTH_32F                  0
-#define INTERNAL_TYPE_DEPTH_24                   1
-#define INTERNAL_TYPE_DEPTH_16                   2
+   enum V3D33_Internal_Depth_Type       internal_type;
    uint32_t                             internal_bpp_ignored;
    uint32_t                             z_stencil_id;
    uint32_t                             sub_id;
 };
 
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CONFIG_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                                              const struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CONFIG * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_Z_STENCIL_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                             const struct V3D33_TILE_RENDERING_MODE_CFG_Z_STENCIL * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -2592,14 +2705,14 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CONFIG_pack(__gen_user_data *d
 
 }
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CONFIG_length      9
+#define V3D33_TILE_RENDERING_MODE_CFG_Z_STENCIL_length      9
 #ifdef __gen_unpack_address
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CONFIG_unpack(const uint8_t * restrict cl,
-                                                                struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CONFIG * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_Z_STENCIL_unpack(const uint8_t * restrict cl,
+                                               struct V3D33_TILE_RENDERING_MODE_CFG_Z_STENCIL * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
-   values->address = __gen_unpack_address(cl, 40, 71);
+   values->address = __gen_unpack_address(cl, 46, 71);
    values->padded_height_of_output_image_in_uif_blocks = __gen_unpack_uint(cl, 33, 45);
    values->memory_format = __gen_unpack_uint(cl, 30, 32);
    values->output_image_format = __gen_unpack_uint(cl, 24, 29);
@@ -2612,28 +2725,28 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CONFIG_unpack(const uint8_t *
 #endif
 
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CLEAR_VALUES_opcode    121
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CLEAR_VALUES_header\
+#define V3D33_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_opcode    121
+#define V3D33_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_header\
    .opcode                              =    121,  \
    .sub_id                              =      3
 
-struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CLEAR_VALUES {
+struct V3D33_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES {
    uint32_t                             opcode;
    uint32_t                             unused;
    float                                z_clear_value;
-   uint32_t                             stencil_vg_mask_clear_value;
+   uint32_t                             stencil_clear_value;
    uint32_t                             sub_id;
 };
 
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CLEAR_VALUES_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                                                    const struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CLEAR_VALUES * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                   const struct V3D33_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
    cl[ 1] = __gen_uint(values->sub_id, 0, 3);
 
-   cl[ 2] = __gen_uint(values->stencil_vg_mask_clear_value, 0, 7);
+   cl[ 2] = __gen_uint(values->stencil_clear_value, 0, 7);
 
 
    memcpy(&cl[3], &values->z_clear_value, sizeof(values->z_clear_value));
@@ -2643,27 +2756,27 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CLEAR_VALUES_pack(__gen_user_d
 
 }
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CLEAR_VALUES_length      9
+#define V3D33_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_length      9
 #ifdef __gen_unpack_address
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CLEAR_VALUES_unpack(const uint8_t * restrict cl,
-                                                                      struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_Z_STENCIL_CLEAR_VALUES * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_unpack(const uint8_t * restrict cl,
+                                                     struct V3D33_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->unused = __gen_unpack_uint(cl, 56, 71);
    values->z_clear_value = __gen_unpack_float(cl, 24, 55);
-   values->stencil_vg_mask_clear_value = __gen_unpack_uint(cl, 16, 23);
+   values->stencil_clear_value = __gen_unpack_uint(cl, 16, 23);
    values->sub_id = __gen_unpack_uint(cl, 8, 11);
 }
 #endif
 
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART1_opcode    121
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART1_header\
+#define V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_opcode    121
+#define V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_header\
    .opcode                              =    121,  \
    .sub_id                              =      4
 
-struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART1 {
+struct V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1 {
    uint32_t                             opcode;
    uint32_t                             clear_color_next_24_bits;
    uint32_t                             clear_color_low_32_bits;
@@ -2672,8 +2785,8 @@ struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART1 {
 };
 
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART1_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                                                const struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART1 * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                      const struct V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1 * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -2690,11 +2803,11 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART1_pack(__gen_user_data
 
 }
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART1_length      9
+#define V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_length      9
 #ifdef __gen_unpack_address
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART1_unpack(const uint8_t * restrict cl,
-                                                                  struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART1 * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_unpack(const uint8_t * restrict cl,
+                                                        struct V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1 * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->clear_color_next_24_bits = __gen_unpack_uint(cl, 48, 71);
@@ -2705,12 +2818,12 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART1_unpack(const uint8_t
 #endif
 
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART2_opcode    121
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART2_header\
+#define V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_opcode    121
+#define V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_header\
    .opcode                              =    121,  \
    .sub_id                              =      5
 
-struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART2 {
+struct V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2 {
    uint32_t                             opcode;
    uint32_t                             clear_color_mid_high_24_bits;
    uint32_t                             clear_color_mid_low_32_bits;
@@ -2719,8 +2832,8 @@ struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART2 {
 };
 
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART2_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                                                const struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART2 * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                      const struct V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2 * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -2737,11 +2850,11 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART2_pack(__gen_user_data
 
 }
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART2_length      9
+#define V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_length      9
 #ifdef __gen_unpack_address
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART2_unpack(const uint8_t * restrict cl,
-                                                                  struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART2 * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_unpack(const uint8_t * restrict cl,
+                                                        struct V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2 * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->clear_color_mid_high_24_bits = __gen_unpack_uint(cl, 48, 71);
@@ -2752,12 +2865,12 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART2_unpack(const uint8_t
 #endif
 
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART3_opcode    121
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART3_header\
+#define V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_opcode    121
+#define V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_header\
    .opcode                              =    121,  \
    .sub_id                              =      6
 
-struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART3 {
+struct V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3 {
    uint32_t                             opcode;
    uint32_t                             pad;
    uint32_t                             uif_padded_height_in_uif_blocks;
@@ -2768,8 +2881,8 @@ struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART3 {
 };
 
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART3_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                                                const struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART3 * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                      const struct V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3 * restrict values)
 {
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
@@ -2793,11 +2906,11 @@ V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART3_pack(__gen_user_data
 
 }
 
-#define V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART3_length      9
+#define V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_length      9
 #ifdef __gen_unpack_address
 static inline void
-V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART3_unpack(const uint8_t * restrict cl,
-                                                                  struct V3D33_TILE_RENDERING_MODE_CONFIGURATION_CLEAR_COLORS_PART3 * restrict values)
+V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_unpack(const uint8_t * restrict cl,
+                                                        struct V3D33_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3 * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
    values->pad = __gen_unpack_uint(cl, 61, 71);
@@ -2848,31 +2961,35 @@ V3D33_TILE_COORDINATES_unpack(const uint8_t * restrict cl,
 #endif
 
 
-#define V3D33_MULTICORE_RENDERING_SUPERTILE_CONFIGURATION_opcode    122
-#define V3D33_MULTICORE_RENDERING_SUPERTILE_CONFIGURATION_header\
+#define V3D33_MULTICORE_RENDERING_SUPERTILE_CFG_opcode    122
+#define V3D33_MULTICORE_RENDERING_SUPERTILE_CFG_header\
    .opcode                              =    122
 
-struct V3D33_MULTICORE_RENDERING_SUPERTILE_CONFIGURATION {
+struct V3D33_MULTICORE_RENDERING_SUPERTILE_CFG {
    uint32_t                             opcode;
+   uint32_t                             number_of_bin_tile_lists;
    bool                                 supertile_raster_order;
    bool                                 multicore_enable;
    uint32_t                             total_frame_height_in_tiles;
    uint32_t                             total_frame_width_in_tiles;
    uint32_t                             total_frame_height_in_supertiles;
    uint32_t                             total_frame_width_in_supertiles;
-   uint32_t                             supertile_height_in_tiles_minus_1;
-   uint32_t                             supertile_width_in_tiles_minus_1;
+   uint32_t                             supertile_height_in_tiles;
+   uint32_t                             supertile_width_in_tiles;
 };
 
 static inline void
-V3D33_MULTICORE_RENDERING_SUPERTILE_CONFIGURATION_pack(__gen_user_data *data, uint8_t * restrict cl,
-                                                       const struct V3D33_MULTICORE_RENDERING_SUPERTILE_CONFIGURATION * restrict values)
+V3D33_MULTICORE_RENDERING_SUPERTILE_CFG_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                             const struct V3D33_MULTICORE_RENDERING_SUPERTILE_CFG * restrict values)
 {
+   assert(values->number_of_bin_tile_lists >= 1);
+   assert(values->supertile_height_in_tiles >= 1);
+   assert(values->supertile_width_in_tiles >= 1);
    cl[ 0] = __gen_uint(values->opcode, 0, 7);
 
-   cl[ 1] = __gen_uint(values->supertile_width_in_tiles_minus_1, 0, 7);
+   cl[ 1] = __gen_uint(values->supertile_width_in_tiles - 1, 0, 7);
 
-   cl[ 2] = __gen_uint(values->supertile_height_in_tiles_minus_1, 0, 7);
+   cl[ 2] = __gen_uint(values->supertile_height_in_tiles - 1, 0, 7);
 
    cl[ 3] = __gen_uint(values->total_frame_width_in_supertiles, 0, 7);
 
@@ -2885,26 +3002,28 @@ V3D33_MULTICORE_RENDERING_SUPERTILE_CONFIGURATION_pack(__gen_user_data *data, ui
 
    cl[ 7] = __gen_uint(values->total_frame_height_in_tiles, 4, 15) >> 8;
 
-   cl[ 8] = __gen_uint(values->supertile_raster_order, 4, 4) |
+   cl[ 8] = __gen_uint(values->number_of_bin_tile_lists - 1, 5, 7) |
+            __gen_uint(values->supertile_raster_order, 4, 4) |
             __gen_uint(values->multicore_enable, 0, 0);
 
 }
 
-#define V3D33_MULTICORE_RENDERING_SUPERTILE_CONFIGURATION_length      9
+#define V3D33_MULTICORE_RENDERING_SUPERTILE_CFG_length      9
 #ifdef __gen_unpack_address
 static inline void
-V3D33_MULTICORE_RENDERING_SUPERTILE_CONFIGURATION_unpack(const uint8_t * restrict cl,
-                                                         struct V3D33_MULTICORE_RENDERING_SUPERTILE_CONFIGURATION * restrict values)
+V3D33_MULTICORE_RENDERING_SUPERTILE_CFG_unpack(const uint8_t * restrict cl,
+                                               struct V3D33_MULTICORE_RENDERING_SUPERTILE_CFG * restrict values)
 {
    values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->number_of_bin_tile_lists = __gen_unpack_uint(cl, 69, 71) + 1;
    values->supertile_raster_order = __gen_unpack_uint(cl, 68, 68);
    values->multicore_enable = __gen_unpack_uint(cl, 64, 64);
    values->total_frame_height_in_tiles = __gen_unpack_uint(cl, 52, 63);
    values->total_frame_width_in_tiles = __gen_unpack_uint(cl, 40, 51);
    values->total_frame_height_in_supertiles = __gen_unpack_uint(cl, 32, 39);
    values->total_frame_width_in_supertiles = __gen_unpack_uint(cl, 24, 31);
-   values->supertile_height_in_tiles_minus_1 = __gen_unpack_uint(cl, 16, 23);
-   values->supertile_width_in_tiles_minus_1 = __gen_unpack_uint(cl, 8, 15);
+   values->supertile_height_in_tiles = __gen_unpack_uint(cl, 16, 23) + 1;
+   values->supertile_width_in_tiles = __gen_unpack_uint(cl, 8, 15) + 1;
 }
 #endif
 
@@ -3036,13 +3155,19 @@ struct V3D33_GL_SHADER_STATE_RECORD {
    uint32_t                             vertex_shader_input_vpm_segment_size;
    __gen_address_type                   address_of_default_attribute_values;
    __gen_address_type                   fragment_shader_code_address;
-   bool                                 _2_way_threadable;
-   bool                                 _4_way_threadable;
-   bool                                 propagate_nans;
+   bool                                 fragment_shader_2_way_threadable;
+   bool                                 fragment_shader_4_way_threadable;
+   bool                                 fragment_shader_propagate_nans;
    __gen_address_type                   fragment_shader_uniforms_address;
    __gen_address_type                   vertex_shader_code_address;
+   bool                                 vertex_shader_2_way_threadable;
+   bool                                 vertex_shader_4_way_threadable;
+   bool                                 vertex_shader_propagate_nans;
    __gen_address_type                   vertex_shader_uniforms_address;
    __gen_address_type                   coordinate_shader_code_address;
+   bool                                 coordinate_shader_2_way_threadable;
+   bool                                 coordinate_shader_4_way_threadable;
+   bool                                 coordinate_shader_propagate_nans;
    __gen_address_type                   coordinate_shader_uniforms_address;
 };
 
@@ -3085,9 +3210,9 @@ V3D33_GL_SHADER_STATE_RECORD_pack(__gen_user_data *data, uint8_t * restrict cl,
 
    __gen_emit_reloc(data, &values->fragment_shader_code_address);
    cl[12] = __gen_address_offset(&values->fragment_shader_code_address) |
-            __gen_uint(values->_2_way_threadable, 0, 0) |
-            __gen_uint(values->_4_way_threadable, 1, 1) |
-            __gen_uint(values->propagate_nans, 2, 2);
+            __gen_uint(values->fragment_shader_2_way_threadable, 0, 0) |
+            __gen_uint(values->fragment_shader_4_way_threadable, 1, 1) |
+            __gen_uint(values->fragment_shader_propagate_nans, 2, 2);
 
    cl[13] = __gen_address_offset(&values->fragment_shader_code_address) >> 8;
 
@@ -3105,7 +3230,10 @@ V3D33_GL_SHADER_STATE_RECORD_pack(__gen_user_data *data, uint8_t * restrict cl,
    cl[19] = __gen_address_offset(&values->fragment_shader_uniforms_address) >> 24;
 
    __gen_emit_reloc(data, &values->vertex_shader_code_address);
-   cl[20] = __gen_address_offset(&values->vertex_shader_code_address);
+   cl[20] = __gen_address_offset(&values->vertex_shader_code_address) |
+            __gen_uint(values->vertex_shader_2_way_threadable, 0, 0) |
+            __gen_uint(values->vertex_shader_4_way_threadable, 1, 1) |
+            __gen_uint(values->vertex_shader_propagate_nans, 2, 2);
 
    cl[21] = __gen_address_offset(&values->vertex_shader_code_address) >> 8;
 
@@ -3123,7 +3251,10 @@ V3D33_GL_SHADER_STATE_RECORD_pack(__gen_user_data *data, uint8_t * restrict cl,
    cl[27] = __gen_address_offset(&values->vertex_shader_uniforms_address) >> 24;
 
    __gen_emit_reloc(data, &values->coordinate_shader_code_address);
-   cl[28] = __gen_address_offset(&values->coordinate_shader_code_address);
+   cl[28] = __gen_address_offset(&values->coordinate_shader_code_address) |
+            __gen_uint(values->coordinate_shader_2_way_threadable, 0, 0) |
+            __gen_uint(values->coordinate_shader_4_way_threadable, 1, 1) |
+            __gen_uint(values->coordinate_shader_propagate_nans, 2, 2);
 
    cl[29] = __gen_address_offset(&values->coordinate_shader_code_address) >> 8;
 
@@ -3166,13 +3297,19 @@ V3D33_GL_SHADER_STATE_RECORD_unpack(const uint8_t * restrict cl,
    values->vertex_shader_input_vpm_segment_size = __gen_unpack_uint(cl, 56, 63);
    values->address_of_default_attribute_values = __gen_unpack_address(cl, 64, 95);
    values->fragment_shader_code_address = __gen_unpack_address(cl, 99, 127);
-   values->_2_way_threadable = __gen_unpack_uint(cl, 96, 96);
-   values->_4_way_threadable = __gen_unpack_uint(cl, 97, 97);
-   values->propagate_nans = __gen_unpack_uint(cl, 98, 98);
+   values->fragment_shader_2_way_threadable = __gen_unpack_uint(cl, 96, 96);
+   values->fragment_shader_4_way_threadable = __gen_unpack_uint(cl, 97, 97);
+   values->fragment_shader_propagate_nans = __gen_unpack_uint(cl, 98, 98);
    values->fragment_shader_uniforms_address = __gen_unpack_address(cl, 128, 159);
    values->vertex_shader_code_address = __gen_unpack_address(cl, 160, 191);
+   values->vertex_shader_2_way_threadable = __gen_unpack_uint(cl, 160, 160);
+   values->vertex_shader_4_way_threadable = __gen_unpack_uint(cl, 161, 161);
+   values->vertex_shader_propagate_nans = __gen_unpack_uint(cl, 162, 162);
    values->vertex_shader_uniforms_address = __gen_unpack_address(cl, 192, 223);
    values->coordinate_shader_code_address = __gen_unpack_address(cl, 224, 255);
+   values->coordinate_shader_2_way_threadable = __gen_unpack_uint(cl, 224, 224);
+   values->coordinate_shader_4_way_threadable = __gen_unpack_uint(cl, 225, 225);
+   values->coordinate_shader_propagate_nans = __gen_unpack_uint(cl, 226, 226);
    values->coordinate_shader_uniforms_address = __gen_unpack_address(cl, 256, 287);
 }
 #endif
@@ -3396,7 +3533,7 @@ struct V3D33_TEXTURE_UNIFORM_PARAMETER_0_CFG_MODE1 {
    bool                                 bias_supplied;
    bool                                 gather_sample_mode;
    bool                                 fetch_sample_mode;
-   bool                                 lookup_type;
+   uint32_t                             lookup_type;
 #define TEXTURE_2D                               0
 #define TEXTURE_2D_ARRAY                         1
 #define TEXTURE_3D                               2
@@ -3462,10 +3599,7 @@ V3D33_TEXTURE_UNIFORM_PARAMETER_0_CFG_MODE1_unpack(const uint8_t * restrict cl,
 
 struct V3D33_TEXTURE_UNIFORM_PARAMETER_1_CFG_MODE1 {
    __gen_address_type                   texture_state_record_base_address;
-   bool                                 return_word_3_of_texture_data;
-   bool                                 return_word_2_of_texture_data;
-   bool                                 return_word_1_of_texture_data;
-   bool                                 return_word_0_of_texture_data;
+   uint32_t                             return_words_of_texture_data;
 };
 
 static inline void
@@ -3474,10 +3608,7 @@ V3D33_TEXTURE_UNIFORM_PARAMETER_1_CFG_MODE1_pack(__gen_user_data *data, uint8_t
 {
    __gen_emit_reloc(data, &values->texture_state_record_base_address);
    cl[ 0] = __gen_address_offset(&values->texture_state_record_base_address) |
-            __gen_uint(values->return_word_3_of_texture_data, 3, 3) |
-            __gen_uint(values->return_word_2_of_texture_data, 2, 2) |
-            __gen_uint(values->return_word_1_of_texture_data, 1, 1) |
-            __gen_uint(values->return_word_0_of_texture_data, 0, 0);
+            __gen_uint(values->return_words_of_texture_data, 0, 3);
 
    cl[ 1] = __gen_address_offset(&values->texture_state_record_base_address) >> 8;
 
@@ -3493,11 +3624,8 @@ static inline void
 V3D33_TEXTURE_UNIFORM_PARAMETER_1_CFG_MODE1_unpack(const uint8_t * restrict cl,
                                                    struct V3D33_TEXTURE_UNIFORM_PARAMETER_1_CFG_MODE1 * restrict values)
 {
-   values->texture_state_record_base_address = __gen_unpack_address(cl, 0, 31);
-   values->return_word_3_of_texture_data = __gen_unpack_uint(cl, 3, 3);
-   values->return_word_2_of_texture_data = __gen_unpack_uint(cl, 2, 2);
-   values->return_word_1_of_texture_data = __gen_unpack_uint(cl, 1, 1);
-   values->return_word_0_of_texture_data = __gen_unpack_uint(cl, 0, 0);
+   values->texture_state_record_base_address = __gen_unpack_address(cl, 4, 31);
+   values->return_words_of_texture_data = __gen_unpack_uint(cl, 0, 3);
 }
 #endif
 
@@ -3506,9 +3634,12 @@ V3D33_TEXTURE_UNIFORM_PARAMETER_1_CFG_MODE1_unpack(const uint8_t * restrict cl,
    .flip_etc_y                          =      1
 
 struct V3D33_TEXTURE_SHADER_STATE {
+   bool                                 uif_xor_disable;
    bool                                 level_0_is_strictly_uif;
    bool                                 level_0_xor_enable;
    uint32_t                             level_0_ub_pad;
+   bool                                 output_32_bit;
+   uint32_t                             sample_number;
    uint32_t                             base_level;
    float                                fixed_bias;
    float                                max_level_of_detail;
@@ -3539,8 +3670,7 @@ struct V3D33_TEXTURE_SHADER_STATE {
    uint32_t                             image_width;
    uint32_t                             array_stride_64_byte_aligned;
    __gen_address_type                   texture_base_pointer;
-   uint32_t                             minification_filter;
-   uint32_t                             magnification_filter;
+   enum V3D33_TMU_Filter                filter;
 };
 
 static inline void
@@ -3549,8 +3679,7 @@ V3D33_TEXTURE_SHADER_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
 {
    __gen_emit_reloc(data, &values->texture_base_pointer);
    cl[ 0] = __gen_address_offset(&values->texture_base_pointer) |
-            __gen_uint(values->minification_filter, 1, 3) |
-            __gen_uint(values->magnification_filter, 0, 0);
+            __gen_uint(values->filter, 0, 3);
 
    cl[ 1] = __gen_address_offset(&values->texture_base_pointer) >> 8;
 
@@ -3622,9 +3751,12 @@ V3D33_TEXTURE_SHADER_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
 
    cl[29] = __gen_sfixed(values->fixed_bias, 0, 15, 8) >> 8;
 
-   cl[30] = __gen_uint(values->base_level, 0, 3);
+   cl[30] = __gen_uint(values->output_32_bit, 6, 6) |
+            __gen_uint(values->sample_number, 4, 5) |
+            __gen_uint(values->base_level, 0, 3);
 
-   cl[31] = __gen_uint(values->level_0_is_strictly_uif, 6, 6) |
+   cl[31] = __gen_uint(values->uif_xor_disable, 7, 7) |
+            __gen_uint(values->level_0_is_strictly_uif, 6, 6) |
             __gen_uint(values->level_0_xor_enable, 4, 4) |
             __gen_uint(values->level_0_ub_pad, 0, 3);
 
@@ -3636,9 +3768,12 @@ static inline void
 V3D33_TEXTURE_SHADER_STATE_unpack(const uint8_t * restrict cl,
                                   struct V3D33_TEXTURE_SHADER_STATE * restrict values)
 {
+   values->uif_xor_disable = __gen_unpack_uint(cl, 255, 255);
    values->level_0_is_strictly_uif = __gen_unpack_uint(cl, 254, 254);
    values->level_0_xor_enable = __gen_unpack_uint(cl, 252, 252);
    values->level_0_ub_pad = __gen_unpack_uint(cl, 248, 251);
+   values->output_32_bit = __gen_unpack_uint(cl, 246, 246);
+   values->sample_number = __gen_unpack_uint(cl, 244, 245);
    values->base_level = __gen_unpack_uint(cl, 240, 243);
    values->fixed_bias = __gen_unpack_sfixed(cl, 224, 239, 8);
    values->max_level_of_detail = __gen_unpack_sfixed(cl, 208, 223, 8);
@@ -3662,9 +3797,8 @@ V3D33_TEXTURE_SHADER_STATE_unpack(const uint8_t * restrict cl,
    values->image_height = __gen_unpack_uint(cl, 72, 85);
    values->image_width = __gen_unpack_uint(cl, 58, 71);
    values->array_stride_64_byte_aligned = __gen_unpack_uint(cl, 32, 57);
-   values->texture_base_pointer = __gen_unpack_address(cl, 0, 31);
-   values->minification_filter = __gen_unpack_uint(cl, 1, 3);
-   values->magnification_filter = __gen_unpack_uint(cl, 0, 0);
+   values->texture_base_pointer = __gen_unpack_address(cl, 2, 31);
+   values->filter = __gen_unpack_uint(cl, 0, 3);
 }
 #endif
 
@@ -3728,6 +3862,25 @@ enum V3D33_Texture_Data_Formats {
         TEXTURE_DATA_FORMAT_ASTC_10X10       =     75,
         TEXTURE_DATA_FORMAT_ASTC_12X10       =     76,
         TEXTURE_DATA_FORMAT_ASTC_12X12       =     77,
+        TEXTURE_DATA_FORMAT_R8I              =     96,
+        TEXTURE_DATA_FORMAT_R8UI             =     97,
+        TEXTURE_DATA_FORMAT_RG8I             =     98,
+        TEXTURE_DATA_FORMAT_RG8UI            =     99,
+        TEXTURE_DATA_FORMAT_RGBA8I           =    100,
+        TEXTURE_DATA_FORMAT_RGBA8UI          =    101,
+        TEXTURE_DATA_FORMAT_R16I             =    102,
+        TEXTURE_DATA_FORMAT_R16UI            =    103,
+        TEXTURE_DATA_FORMAT_RG16I            =    104,
+        TEXTURE_DATA_FORMAT_RG16UI           =    105,
+        TEXTURE_DATA_FORMAT_RGBA16I          =    106,
+        TEXTURE_DATA_FORMAT_RGBA16UI         =    107,
+        TEXTURE_DATA_FORMAT_R32I             =    108,
+        TEXTURE_DATA_FORMAT_R32UI            =    109,
+        TEXTURE_DATA_FORMAT_RG32I            =    110,
+        TEXTURE_DATA_FORMAT_RG32UI           =    111,
+        TEXTURE_DATA_FORMAT_RGBA32I          =    112,
+        TEXTURE_DATA_FORMAT_RGBA32UI         =    113,
+        TEXTURE_DATA_FORMAT_RGB10_A2UI       =    114,
 };
 
 #endif /* V3D33_PACK_H */
diff --git a/lib/mesa/src/broadcom/cle/v3d_packet_v41_pack.h b/lib/mesa/src/broadcom/cle/v3d_packet_v41_pack.h
new file mode 100644
index 000000000..03094eb22
--- /dev/null
+++ b/lib/mesa/src/broadcom/cle/v3d_packet_v41_pack.h
@@ -0,0 +1,4399 @@
+/* Generated code, see packets.xml and gen_packet_header.py */
+
+
+/* Packets, enums and structures for V3D 4.1.
+ *
+ * This file has been generated, do not hand edit.
+ */
+
+#ifndef V3D41_PACK_H
+#define V3D41_PACK_H
+
+#include "cle/v3d_packet_helpers.h"
+
+
+enum V3D41_Compare_Function {
+        V3D_COMPARE_FUNC_NEVER               =      0,
+        V3D_COMPARE_FUNC_LESS                =      1,
+        V3D_COMPARE_FUNC_EQUAL               =      2,
+        V3D_COMPARE_FUNC_LEQUAL              =      3,
+        V3D_COMPARE_FUNC_GREATER             =      4,
+        V3D_COMPARE_FUNC_NOTEQUAL            =      5,
+        V3D_COMPARE_FUNC_GEQUAL              =      6,
+        V3D_COMPARE_FUNC_ALWAYS              =      7,
+};
+
+enum V3D41_Blend_Factor {
+        V3D_BLEND_FACTOR_ZERO                =      0,
+        V3D_BLEND_FACTOR_ONE                 =      1,
+        V3D_BLEND_FACTOR_SRC_COLOR           =      2,
+        V3D_BLEND_FACTOR_INV_SRC_COLOR       =      3,
+        V3D_BLEND_FACTOR_DST_COLOR           =      4,
+        V3D_BLEND_FACTOR_INV_DST_COLOR       =      5,
+        V3D_BLEND_FACTOR_SRC_ALPHA           =      6,
+        V3D_BLEND_FACTOR_INV_SRC_ALPHA       =      7,
+        V3D_BLEND_FACTOR_DST_ALPHA           =      8,
+        V3D_BLEND_FACTOR_INV_DST_ALPHA       =      9,
+        V3D_BLEND_FACTOR_CONST_COLOR         =     10,
+        V3D_BLEND_FACTOR_INV_CONST_COLOR     =     11,
+        V3D_BLEND_FACTOR_CONST_ALPHA         =     12,
+        V3D_BLEND_FACTOR_INV_CONST_ALPHA     =     13,
+        V3D_BLEND_FACTOR_SRC_ALPHA_SATURATE  =     14,
+};
+
+enum V3D41_Blend_Mode {
+        V3D_BLEND_MODE_ADD                   =      0,
+        V3D_BLEND_MODE_SUB                   =      1,
+        V3D_BLEND_MODE_RSUB                  =      2,
+        V3D_BLEND_MODE_MIN                   =      3,
+        V3D_BLEND_MODE_MAX                   =      4,
+        V3D_BLEND_MODE_MUL                   =      5,
+        V3D_BLEND_MODE_SCREEN                =      6,
+        V3D_BLEND_MODE_DARKEN                =      7,
+        V3D_BLEND_MODE_LIGHTEN               =      8,
+};
+
+enum V3D41_Stencil_Op {
+        V3D_STENCIL_OP_ZERO                  =      0,
+        V3D_STENCIL_OP_KEEP                  =      1,
+        V3D_STENCIL_OP_REPLACE               =      2,
+        V3D_STENCIL_OP_INCR                  =      3,
+        V3D_STENCIL_OP_DECR                  =      4,
+        V3D_STENCIL_OP_INVERT                =      5,
+        V3D_STENCIL_OP_INCWRAP               =      6,
+        V3D_STENCIL_OP_DECWRAP               =      7,
+};
+
+enum V3D41_Primitive {
+        V3D_PRIM_POINTS                      =      0,
+        V3D_PRIM_LINES                       =      1,
+        V3D_PRIM_LINE_LOOP                   =      2,
+        V3D_PRIM_LINE_STRIP                  =      3,
+        V3D_PRIM_TRIANGLES                   =      4,
+        V3D_PRIM_TRIANGLE_STRIP              =      5,
+        V3D_PRIM_TRIANGLE_FAN                =      6,
+        V3D_PRIM_POINTS_TF                   =     16,
+        V3D_PRIM_LINES_TF                    =     17,
+        V3D_PRIM_LINE_LOOP_TF                =     18,
+        V3D_PRIM_LINE_STRIP_TF               =     19,
+        V3D_PRIM_TRIANGLES_TF                =     20,
+        V3D_PRIM_TRIANGLE_STRIP_TF           =     21,
+        V3D_PRIM_TRIANGLE_FAN_TF             =     22,
+};
+
+enum V3D41_Border_Color_Mode {
+        V3D_BORDER_COLOR_0000                =      0,
+        V3D_BORDER_COLOR_0001                =      1,
+        V3D_BORDER_COLOR_1111                =      2,
+        V3D_BORDER_COLOR_FOLLOWS             =      7,
+};
+
+enum V3D41_Wrap_Mode {
+        V3D_WRAP_MODE_WRAP_MODE_REPEAT       =      0,
+        V3D_WRAP_MODE_WRAP_MODE_CLAMP        =      1,
+        V3D_WRAP_MODE_WRAP_MODE_MIRROR       =      2,
+        V3D_WRAP_MODE_WRAP_MODE_BORDER       =      3,
+        V3D_WRAP_MODE_WRAP_MODE_MIRROR_ONCE  =      4,
+};
+
+enum V3D41_TMU_Op {
+        V3D_TMU_OP_WRITE_ADD_READ_PREFETCH   =      0,
+        V3D_TMU_OP_WRITE_SUB_READ_CLEAR      =      1,
+        V3D_TMU_OP_WRITE_XCHG_READ_FLUSH     =      2,
+        V3D_TMU_OP_WRITE_CMPXCHG_READ_FLUSH  =      3,
+        V3D_TMU_OP_WRITE_UMIN_FULL_L1_CLEAR  =      4,
+        V3D_TMU_OP_WRITE_UMAX                =      5,
+        V3D_TMU_OP_WRITE_SMIN                =      6,
+        V3D_TMU_OP_WRITE_SMAX                =      7,
+        V3D_TMU_OP_WRITE_AND_READ_INC        =      8,
+        V3D_TMU_OP_WRITE_OR_READ_DEC         =      9,
+        V3D_TMU_OP_WRITE_XOR_READ_NOT        =     10,
+        V3D_TMU_OP_REGULAR                   =     15,
+};
+
+enum V3D41_Varying_Flags_Action {
+        V3D_VARYING_FLAGS_ACTION_UNCHANGED   =      0,
+        V3D_VARYING_FLAGS_ACTION_ZEROED      =      1,
+        V3D_VARYING_FLAGS_ACTION_SET         =      2,
+};
+
+enum V3D41_Memory_Format {
+        V3D_MEMORY_FORMAT_RASTER             =      0,
+        V3D_MEMORY_FORMAT_LINEARTILE         =      1,
+        V3D_MEMORY_FORMAT_UB_LINEAR_1_UIF_BLOCK_WIDE =      2,
+        V3D_MEMORY_FORMAT_UB_LINEAR_2_UIF_BLOCKS_WIDE =      3,
+        V3D_MEMORY_FORMAT_UIF_NO_XOR         =      4,
+        V3D_MEMORY_FORMAT_UIF_XOR            =      5,
+};
+
+enum V3D41_Decimate_Mode {
+        V3D_DECIMATE_MODE_SAMPLE_0           =      0,
+        V3D_DECIMATE_MODE_4X                 =      1,
+        V3D_DECIMATE_MODE_ALL_SAMPLES        =      3,
+};
+
+enum V3D41_Internal_Type {
+        V3D_INTERNAL_TYPE_8I                 =      0,
+        V3D_INTERNAL_TYPE_8UI                =      1,
+        V3D_INTERNAL_TYPE_8                  =      2,
+        V3D_INTERNAL_TYPE_16I                =      4,
+        V3D_INTERNAL_TYPE_16UI               =      5,
+        V3D_INTERNAL_TYPE_16F                =      6,
+        V3D_INTERNAL_TYPE_32I                =      8,
+        V3D_INTERNAL_TYPE_32UI               =      9,
+        V3D_INTERNAL_TYPE_32F                =     10,
+};
+
+enum V3D41_Internal_BPP {
+        V3D_INTERNAL_BPP_32                  =      0,
+        V3D_INTERNAL_BPP_64                  =      1,
+        V3D_INTERNAL_BPP_128                 =      2,
+};
+
+enum V3D41_Internal_Depth_Type {
+        V3D_INTERNAL_TYPE_DEPTH_32F          =      0,
+        V3D_INTERNAL_TYPE_DEPTH_24           =      1,
+        V3D_INTERNAL_TYPE_DEPTH_16           =      2,
+};
+
+enum V3D41_Render_Target_Clamp {
+        V3D_RENDER_TARGET_CLAMP_NONE         =      0,
+        V3D_RENDER_TARGET_CLAMP_NORM         =      1,
+        V3D_RENDER_TARGET_CLAMP_POS          =      2,
+};
+
+enum V3D41_Output_Image_Format {
+        V3D_OUTPUT_IMAGE_FORMAT_SRGB8_ALPHA8 =      0,
+        V3D_OUTPUT_IMAGE_FORMAT_SRGB         =      1,
+        V3D_OUTPUT_IMAGE_FORMAT_RGB10_A2UI   =      2,
+        V3D_OUTPUT_IMAGE_FORMAT_RGB10_A2     =      3,
+        V3D_OUTPUT_IMAGE_FORMAT_ABGR1555     =      4,
+        V3D_OUTPUT_IMAGE_FORMAT_ALPHA_MASKED_ABGR1555 =      5,
+        V3D_OUTPUT_IMAGE_FORMAT_ABGR4444     =      6,
+        V3D_OUTPUT_IMAGE_FORMAT_BGR565       =      7,
+        V3D_OUTPUT_IMAGE_FORMAT_R11F_G11F_B10F =      8,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA32F      =      9,
+        V3D_OUTPUT_IMAGE_FORMAT_RG32F        =     10,
+        V3D_OUTPUT_IMAGE_FORMAT_R32F         =     11,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA32I      =     12,
+        V3D_OUTPUT_IMAGE_FORMAT_RG32I        =     13,
+        V3D_OUTPUT_IMAGE_FORMAT_R32I         =     14,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA32UI     =     15,
+        V3D_OUTPUT_IMAGE_FORMAT_RG32UI       =     16,
+        V3D_OUTPUT_IMAGE_FORMAT_R32UI        =     17,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA16F      =     18,
+        V3D_OUTPUT_IMAGE_FORMAT_RG16F        =     19,
+        V3D_OUTPUT_IMAGE_FORMAT_R16F         =     20,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA16I      =     21,
+        V3D_OUTPUT_IMAGE_FORMAT_RG16I        =     22,
+        V3D_OUTPUT_IMAGE_FORMAT_R16I         =     23,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA16UI     =     24,
+        V3D_OUTPUT_IMAGE_FORMAT_RG16UI       =     25,
+        V3D_OUTPUT_IMAGE_FORMAT_R16UI        =     26,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA8        =     27,
+        V3D_OUTPUT_IMAGE_FORMAT_RGB8         =     28,
+        V3D_OUTPUT_IMAGE_FORMAT_RG8          =     29,
+        V3D_OUTPUT_IMAGE_FORMAT_R8           =     30,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA8I       =     31,
+        V3D_OUTPUT_IMAGE_FORMAT_RG8I         =     32,
+        V3D_OUTPUT_IMAGE_FORMAT_R8I          =     33,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA8UI      =     34,
+        V3D_OUTPUT_IMAGE_FORMAT_RG8UI        =     35,
+        V3D_OUTPUT_IMAGE_FORMAT_R8UI         =     36,
+        V3D_OUTPUT_IMAGE_FORMAT_BSTC         =     39,
+        V3D_OUTPUT_IMAGE_FORMAT_D32F         =     40,
+        V3D_OUTPUT_IMAGE_FORMAT_D24          =     41,
+        V3D_OUTPUT_IMAGE_FORMAT_D16          =     42,
+        V3D_OUTPUT_IMAGE_FORMAT_D24S8        =     43,
+        V3D_OUTPUT_IMAGE_FORMAT_S8           =     44,
+};
+
+enum V3D41_Dither_Mode {
+        V3D_DITHER_MODE_NONE                 =      0,
+        V3D_DITHER_MODE_RGB                  =      1,
+        V3D_DITHER_MODE_A                    =      2,
+        V3D_DITHER_MODE_RGBA                 =      3,
+};
+
+#define V3D41_HALT_opcode                      0
+#define V3D41_HALT_header                       \
+   .opcode                              =      0
+
+struct V3D41_HALT {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_HALT_pack(__gen_user_data *data, uint8_t * restrict cl,
+                const struct V3D41_HALT * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_HALT_length                      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_HALT_unpack(const uint8_t * restrict cl,
+                  struct V3D41_HALT * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_NOP_opcode                       1
+#define V3D41_NOP_header                        \
+   .opcode                              =      1
+
+struct V3D41_NOP {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_NOP_pack(__gen_user_data *data, uint8_t * restrict cl,
+               const struct V3D41_NOP * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_NOP_length                       1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_NOP_unpack(const uint8_t * restrict cl,
+                 struct V3D41_NOP * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_FLUSH_opcode                     4
+#define V3D41_FLUSH_header                      \
+   .opcode                              =      4
+
+struct V3D41_FLUSH {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_FLUSH_pack(__gen_user_data *data, uint8_t * restrict cl,
+                 const struct V3D41_FLUSH * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_FLUSH_length                     1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_FLUSH_unpack(const uint8_t * restrict cl,
+                   struct V3D41_FLUSH * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_FLUSH_ALL_STATE_opcode           5
+#define V3D41_FLUSH_ALL_STATE_header            \
+   .opcode                              =      5
+
+struct V3D41_FLUSH_ALL_STATE {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_FLUSH_ALL_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                           const struct V3D41_FLUSH_ALL_STATE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_FLUSH_ALL_STATE_length           1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_FLUSH_ALL_STATE_unpack(const uint8_t * restrict cl,
+                             struct V3D41_FLUSH_ALL_STATE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_START_TILE_BINNING_opcode        6
+#define V3D41_START_TILE_BINNING_header         \
+   .opcode                              =      6
+
+struct V3D41_START_TILE_BINNING {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_START_TILE_BINNING_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D41_START_TILE_BINNING * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_START_TILE_BINNING_length        1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_START_TILE_BINNING_unpack(const uint8_t * restrict cl,
+                                struct V3D41_START_TILE_BINNING * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_INCREMENT_SEMAPHORE_opcode       7
+#define V3D41_INCREMENT_SEMAPHORE_header        \
+   .opcode                              =      7
+
+struct V3D41_INCREMENT_SEMAPHORE {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_INCREMENT_SEMAPHORE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                               const struct V3D41_INCREMENT_SEMAPHORE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_INCREMENT_SEMAPHORE_length       1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_INCREMENT_SEMAPHORE_unpack(const uint8_t * restrict cl,
+                                 struct V3D41_INCREMENT_SEMAPHORE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_WAIT_ON_SEMAPHORE_opcode         8
+#define V3D41_WAIT_ON_SEMAPHORE_header          \
+   .opcode                              =      8
+
+struct V3D41_WAIT_ON_SEMAPHORE {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_WAIT_ON_SEMAPHORE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                             const struct V3D41_WAIT_ON_SEMAPHORE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_WAIT_ON_SEMAPHORE_length         1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_WAIT_ON_SEMAPHORE_unpack(const uint8_t * restrict cl,
+                               struct V3D41_WAIT_ON_SEMAPHORE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_WAIT_FOR_PREVIOUS_FRAME_opcode      9
+#define V3D41_WAIT_FOR_PREVIOUS_FRAME_header    \
+   .opcode                              =      9
+
+struct V3D41_WAIT_FOR_PREVIOUS_FRAME {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_WAIT_FOR_PREVIOUS_FRAME_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                   const struct V3D41_WAIT_FOR_PREVIOUS_FRAME * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_WAIT_FOR_PREVIOUS_FRAME_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_WAIT_FOR_PREVIOUS_FRAME_unpack(const uint8_t * restrict cl,
+                                     struct V3D41_WAIT_FOR_PREVIOUS_FRAME * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_ENABLE_Z_ONLY_RENDERING_opcode     10
+#define V3D41_ENABLE_Z_ONLY_RENDERING_header    \
+   .opcode                              =     10
+
+struct V3D41_ENABLE_Z_ONLY_RENDERING {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_ENABLE_Z_ONLY_RENDERING_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                   const struct V3D41_ENABLE_Z_ONLY_RENDERING * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_ENABLE_Z_ONLY_RENDERING_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_ENABLE_Z_ONLY_RENDERING_unpack(const uint8_t * restrict cl,
+                                     struct V3D41_ENABLE_Z_ONLY_RENDERING * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_DISABLE_Z_ONLY_RENDERING_opcode     11
+#define V3D41_DISABLE_Z_ONLY_RENDERING_header   \
+   .opcode                              =     11
+
+struct V3D41_DISABLE_Z_ONLY_RENDERING {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_DISABLE_Z_ONLY_RENDERING_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                    const struct V3D41_DISABLE_Z_ONLY_RENDERING * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_DISABLE_Z_ONLY_RENDERING_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_DISABLE_Z_ONLY_RENDERING_unpack(const uint8_t * restrict cl,
+                                      struct V3D41_DISABLE_Z_ONLY_RENDERING * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_END_OF_Z_ONLY_RENDERING_IN_FRAME_opcode     12
+#define V3D41_END_OF_Z_ONLY_RENDERING_IN_FRAME_header\
+   .opcode                              =     12
+
+struct V3D41_END_OF_Z_ONLY_RENDERING_IN_FRAME {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_END_OF_Z_ONLY_RENDERING_IN_FRAME_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                            const struct V3D41_END_OF_Z_ONLY_RENDERING_IN_FRAME * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_END_OF_Z_ONLY_RENDERING_IN_FRAME_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_END_OF_Z_ONLY_RENDERING_IN_FRAME_unpack(const uint8_t * restrict cl,
+                                              struct V3D41_END_OF_Z_ONLY_RENDERING_IN_FRAME * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_END_OF_RENDERING_opcode         13
+#define V3D41_END_OF_RENDERING_header           \
+   .opcode                              =     13
+
+struct V3D41_END_OF_RENDERING {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_END_OF_RENDERING_pack(__gen_user_data *data, uint8_t * restrict cl,
+                            const struct V3D41_END_OF_RENDERING * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_END_OF_RENDERING_length          1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_END_OF_RENDERING_unpack(const uint8_t * restrict cl,
+                              struct V3D41_END_OF_RENDERING * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_WAIT_FOR_TRANSFORM_FEEDBACK_opcode     14
+#define V3D41_WAIT_FOR_TRANSFORM_FEEDBACK_header\
+   .opcode                              =     14
+
+struct V3D41_WAIT_FOR_TRANSFORM_FEEDBACK {
+   uint32_t                             opcode;
+   uint32_t                             block_count;
+};
+
+static inline void
+V3D41_WAIT_FOR_TRANSFORM_FEEDBACK_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                       const struct V3D41_WAIT_FOR_TRANSFORM_FEEDBACK * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->block_count, 0, 7);
+
+}
+
+#define V3D41_WAIT_FOR_TRANSFORM_FEEDBACK_length      2
+#ifdef __gen_unpack_address
+static inline void
+V3D41_WAIT_FOR_TRANSFORM_FEEDBACK_unpack(const uint8_t * restrict cl,
+                                         struct V3D41_WAIT_FOR_TRANSFORM_FEEDBACK * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->block_count = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D41_BRANCH_TO_AUTO_CHAINED_SUB_LIST_opcode     15
+#define V3D41_BRANCH_TO_AUTO_CHAINED_SUB_LIST_header\
+   .opcode                              =     15
+
+struct V3D41_BRANCH_TO_AUTO_CHAINED_SUB_LIST {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+};
+
+static inline void
+V3D41_BRANCH_TO_AUTO_CHAINED_SUB_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                           const struct V3D41_BRANCH_TO_AUTO_CHAINED_SUB_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D41_BRANCH_TO_AUTO_CHAINED_SUB_LIST_length      5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_BRANCH_TO_AUTO_CHAINED_SUB_LIST_unpack(const uint8_t * restrict cl,
+                                             struct V3D41_BRANCH_TO_AUTO_CHAINED_SUB_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 8, 39);
+}
+#endif
+
+
+#define V3D41_BRANCH_opcode                   16
+#define V3D41_BRANCH_header                     \
+   .opcode                              =     16
+
+struct V3D41_BRANCH {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+};
+
+static inline void
+V3D41_BRANCH_pack(__gen_user_data *data, uint8_t * restrict cl,
+                  const struct V3D41_BRANCH * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D41_BRANCH_length                    5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_BRANCH_unpack(const uint8_t * restrict cl,
+                    struct V3D41_BRANCH * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 8, 39);
+}
+#endif
+
+
+#define V3D41_BRANCH_TO_SUB_LIST_opcode       17
+#define V3D41_BRANCH_TO_SUB_LIST_header         \
+   .opcode                              =     17
+
+struct V3D41_BRANCH_TO_SUB_LIST {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+};
+
+static inline void
+V3D41_BRANCH_TO_SUB_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D41_BRANCH_TO_SUB_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D41_BRANCH_TO_SUB_LIST_length        5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_BRANCH_TO_SUB_LIST_unpack(const uint8_t * restrict cl,
+                                struct V3D41_BRANCH_TO_SUB_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 8, 39);
+}
+#endif
+
+
+#define V3D41_RETURN_FROM_SUB_LIST_opcode     18
+#define V3D41_RETURN_FROM_SUB_LIST_header       \
+   .opcode                              =     18
+
+struct V3D41_RETURN_FROM_SUB_LIST {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_RETURN_FROM_SUB_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                const struct V3D41_RETURN_FROM_SUB_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_RETURN_FROM_SUB_LIST_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_RETURN_FROM_SUB_LIST_unpack(const uint8_t * restrict cl,
+                                  struct V3D41_RETURN_FROM_SUB_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_FLUSH_VCD_CACHE_opcode          19
+#define V3D41_FLUSH_VCD_CACHE_header            \
+   .opcode                              =     19
+
+struct V3D41_FLUSH_VCD_CACHE {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_FLUSH_VCD_CACHE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                           const struct V3D41_FLUSH_VCD_CACHE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_FLUSH_VCD_CACHE_length           1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_FLUSH_VCD_CACHE_unpack(const uint8_t * restrict cl,
+                             struct V3D41_FLUSH_VCD_CACHE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_START_ADDRESS_OF_GENERIC_TILE_LIST_opcode     20
+#define V3D41_START_ADDRESS_OF_GENERIC_TILE_LIST_header\
+   .opcode                              =     20
+
+struct V3D41_START_ADDRESS_OF_GENERIC_TILE_LIST {
+   uint32_t                             opcode;
+   __gen_address_type                   start;
+   __gen_address_type                   end;
+};
+
+static inline void
+V3D41_START_ADDRESS_OF_GENERIC_TILE_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                              const struct V3D41_START_ADDRESS_OF_GENERIC_TILE_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->start);
+   cl[ 1] = __gen_address_offset(&values->start);
+
+   cl[ 2] = __gen_address_offset(&values->start) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->start) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->start) >> 24;
+
+   __gen_emit_reloc(data, &values->end);
+   cl[ 5] = __gen_address_offset(&values->end);
+
+   cl[ 6] = __gen_address_offset(&values->end) >> 8;
+
+   cl[ 7] = __gen_address_offset(&values->end) >> 16;
+
+   cl[ 8] = __gen_address_offset(&values->end) >> 24;
+
+}
+
+#define V3D41_START_ADDRESS_OF_GENERIC_TILE_LIST_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_START_ADDRESS_OF_GENERIC_TILE_LIST_unpack(const uint8_t * restrict cl,
+                                                struct V3D41_START_ADDRESS_OF_GENERIC_TILE_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->start = __gen_unpack_address(cl, 8, 39);
+   values->end = __gen_unpack_address(cl, 40, 71);
+}
+#endif
+
+
+#define V3D41_BRANCH_TO_IMPLICIT_TILE_LIST_opcode     21
+#define V3D41_BRANCH_TO_IMPLICIT_TILE_LIST_header\
+   .opcode                              =     21
+
+struct V3D41_BRANCH_TO_IMPLICIT_TILE_LIST {
+   uint32_t                             opcode;
+   uint32_t                             tile_list_set_number;
+};
+
+static inline void
+V3D41_BRANCH_TO_IMPLICIT_TILE_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D41_BRANCH_TO_IMPLICIT_TILE_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->tile_list_set_number, 0, 7);
+
+}
+
+#define V3D41_BRANCH_TO_IMPLICIT_TILE_LIST_length      2
+#ifdef __gen_unpack_address
+static inline void
+V3D41_BRANCH_TO_IMPLICIT_TILE_LIST_unpack(const uint8_t * restrict cl,
+                                          struct V3D41_BRANCH_TO_IMPLICIT_TILE_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->tile_list_set_number = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D41_BRANCH_TO_EXPLICIT_SUPERTILE_opcode     22
+#define V3D41_BRANCH_TO_EXPLICIT_SUPERTILE_header\
+   .opcode                              =     22
+
+struct V3D41_BRANCH_TO_EXPLICIT_SUPERTILE {
+   uint32_t                             opcode;
+   __gen_address_type                   absolute_address_of_explicit_supertile_render_list;
+   uint32_t                             explicit_supertile_number;
+   uint32_t                             row_number;
+   uint32_t                             column_number;
+};
+
+static inline void
+V3D41_BRANCH_TO_EXPLICIT_SUPERTILE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D41_BRANCH_TO_EXPLICIT_SUPERTILE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->column_number, 0, 7);
+
+   cl[ 2] = __gen_uint(values->row_number, 0, 7);
+
+   cl[ 3] = __gen_uint(values->explicit_supertile_number, 0, 7);
+
+   __gen_emit_reloc(data, &values->absolute_address_of_explicit_supertile_render_list);
+   cl[ 4] = __gen_address_offset(&values->absolute_address_of_explicit_supertile_render_list);
+
+   cl[ 5] = __gen_address_offset(&values->absolute_address_of_explicit_supertile_render_list) >> 8;
+
+   cl[ 6] = __gen_address_offset(&values->absolute_address_of_explicit_supertile_render_list) >> 16;
+
+   cl[ 7] = __gen_address_offset(&values->absolute_address_of_explicit_supertile_render_list) >> 24;
+
+}
+
+#define V3D41_BRANCH_TO_EXPLICIT_SUPERTILE_length      8
+#ifdef __gen_unpack_address
+static inline void
+V3D41_BRANCH_TO_EXPLICIT_SUPERTILE_unpack(const uint8_t * restrict cl,
+                                          struct V3D41_BRANCH_TO_EXPLICIT_SUPERTILE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->absolute_address_of_explicit_supertile_render_list = __gen_unpack_address(cl, 32, 63);
+   values->explicit_supertile_number = __gen_unpack_uint(cl, 24, 31);
+   values->row_number = __gen_unpack_uint(cl, 16, 23);
+   values->column_number = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D41_SUPERTILE_COORDINATES_opcode     23
+#define V3D41_SUPERTILE_COORDINATES_header      \
+   .opcode                              =     23
+
+struct V3D41_SUPERTILE_COORDINATES {
+   uint32_t                             opcode;
+   uint32_t                             row_number_in_supertiles;
+   uint32_t                             column_number_in_supertiles;
+};
+
+static inline void
+V3D41_SUPERTILE_COORDINATES_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                 const struct V3D41_SUPERTILE_COORDINATES * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->column_number_in_supertiles, 0, 7);
+
+   cl[ 2] = __gen_uint(values->row_number_in_supertiles, 0, 7);
+
+}
+
+#define V3D41_SUPERTILE_COORDINATES_length      3
+#ifdef __gen_unpack_address
+static inline void
+V3D41_SUPERTILE_COORDINATES_unpack(const uint8_t * restrict cl,
+                                   struct V3D41_SUPERTILE_COORDINATES * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->row_number_in_supertiles = __gen_unpack_uint(cl, 16, 23);
+   values->column_number_in_supertiles = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D41_CLEAR_TILE_BUFFERS_opcode       25
+#define V3D41_CLEAR_TILE_BUFFERS_header         \
+   .opcode                              =     25
+
+struct V3D41_CLEAR_TILE_BUFFERS {
+   uint32_t                             opcode;
+   bool                                 clear_z_stencil_buffer;
+   bool                                 clear_all_render_targets;
+};
+
+static inline void
+V3D41_CLEAR_TILE_BUFFERS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D41_CLEAR_TILE_BUFFERS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->clear_z_stencil_buffer, 1, 1) |
+            __gen_uint(values->clear_all_render_targets, 0, 0);
+
+}
+
+#define V3D41_CLEAR_TILE_BUFFERS_length        2
+#ifdef __gen_unpack_address
+static inline void
+V3D41_CLEAR_TILE_BUFFERS_unpack(const uint8_t * restrict cl,
+                                struct V3D41_CLEAR_TILE_BUFFERS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->clear_z_stencil_buffer = __gen_unpack_uint(cl, 9, 9);
+   values->clear_all_render_targets = __gen_unpack_uint(cl, 8, 8);
+}
+#endif
+
+
+#define V3D41_END_OF_LOADS_opcode             26
+#define V3D41_END_OF_LOADS_header               \
+   .opcode                              =     26
+
+struct V3D41_END_OF_LOADS {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_END_OF_LOADS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                        const struct V3D41_END_OF_LOADS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_END_OF_LOADS_length              1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_END_OF_LOADS_unpack(const uint8_t * restrict cl,
+                          struct V3D41_END_OF_LOADS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_END_OF_TILE_MARKER_opcode       27
+#define V3D41_END_OF_TILE_MARKER_header         \
+   .opcode                              =     27
+
+struct V3D41_END_OF_TILE_MARKER {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_END_OF_TILE_MARKER_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D41_END_OF_TILE_MARKER * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_END_OF_TILE_MARKER_length        1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_END_OF_TILE_MARKER_unpack(const uint8_t * restrict cl,
+                                struct V3D41_END_OF_TILE_MARKER * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_STORE_TILE_BUFFER_GENERAL_opcode     29
+#define V3D41_STORE_TILE_BUFFER_GENERAL_header  \
+   .opcode                              =     29
+
+struct V3D41_STORE_TILE_BUFFER_GENERAL {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+   uint32_t                             height;
+   uint32_t                             height_in_ub_or_stride;
+   bool                                 r_b_swap;
+   bool                                 channel_reverse;
+   bool                                 clear_buffer_being_stored;
+   enum V3D41_Output_Image_Format       output_image_format;
+   enum V3D41_Decimate_Mode             decimate_mode;
+   enum V3D41_Dither_Mode               dither_mode;
+   bool                                 flip_y;
+   enum V3D41_Memory_Format             memory_format;
+   uint32_t                             buffer_to_store;
+#define RENDER_TARGET_0                          0
+#define RENDER_TARGET_1                          1
+#define RENDER_TARGET_2                          2
+#define RENDER_TARGET_3                          3
+#define NONE                                     8
+#define Z                                        9
+#define STENCIL                                  10
+#define ZSTENCIL                                 11
+};
+
+static inline void
+V3D41_STORE_TILE_BUFFER_GENERAL_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                     const struct V3D41_STORE_TILE_BUFFER_GENERAL * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->flip_y, 7, 7) |
+            __gen_uint(values->memory_format, 4, 6) |
+            __gen_uint(values->buffer_to_store, 0, 3);
+
+   cl[ 2] = __gen_uint(values->output_image_format, 4, 9) |
+            __gen_uint(values->decimate_mode, 2, 3) |
+            __gen_uint(values->dither_mode, 0, 1);
+
+   cl[ 3] = __gen_uint(values->r_b_swap, 4, 4) |
+            __gen_uint(values->channel_reverse, 3, 3) |
+            __gen_uint(values->clear_buffer_being_stored, 2, 2) |
+            __gen_uint(values->output_image_format, 4, 9) >> 8;
+
+   cl[ 4] = __gen_uint(values->height_in_ub_or_stride, 4, 23);
+
+   cl[ 5] = __gen_uint(values->height_in_ub_or_stride, 4, 23) >> 8;
+
+   cl[ 6] = __gen_uint(values->height_in_ub_or_stride, 4, 23) >> 16;
+
+   cl[ 7] = __gen_uint(values->height, 0, 15);
+
+   cl[ 8] = __gen_uint(values->height, 0, 15) >> 8;
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 9] = __gen_address_offset(&values->address);
+
+   cl[10] = __gen_address_offset(&values->address) >> 8;
+
+   cl[11] = __gen_address_offset(&values->address) >> 16;
+
+   cl[12] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D41_STORE_TILE_BUFFER_GENERAL_length     13
+#ifdef __gen_unpack_address
+static inline void
+V3D41_STORE_TILE_BUFFER_GENERAL_unpack(const uint8_t * restrict cl,
+                                       struct V3D41_STORE_TILE_BUFFER_GENERAL * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 72, 103);
+   values->height = __gen_unpack_uint(cl, 56, 71);
+   values->height_in_ub_or_stride = __gen_unpack_uint(cl, 36, 55);
+   values->r_b_swap = __gen_unpack_uint(cl, 28, 28);
+   values->channel_reverse = __gen_unpack_uint(cl, 27, 27);
+   values->clear_buffer_being_stored = __gen_unpack_uint(cl, 26, 26);
+   values->output_image_format = __gen_unpack_uint(cl, 20, 25);
+   values->decimate_mode = __gen_unpack_uint(cl, 18, 19);
+   values->dither_mode = __gen_unpack_uint(cl, 16, 17);
+   values->flip_y = __gen_unpack_uint(cl, 15, 15);
+   values->memory_format = __gen_unpack_uint(cl, 12, 14);
+   values->buffer_to_store = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_LOAD_TILE_BUFFER_GENERAL_opcode     30
+#define V3D41_LOAD_TILE_BUFFER_GENERAL_header   \
+   .opcode                              =     30
+
+struct V3D41_LOAD_TILE_BUFFER_GENERAL {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+   uint32_t                             height;
+   uint32_t                             height_in_ub_or_stride;
+   bool                                 r_b_swap;
+   bool                                 channel_reverse;
+   enum V3D41_Output_Image_Format       input_image_format;
+   enum V3D41_Decimate_Mode             decimate_mode;
+   bool                                 flip_y;
+   enum V3D41_Memory_Format             memory_format;
+   uint32_t                             buffer_to_load;
+#define RENDER_TARGET_0                          0
+#define RENDER_TARGET_1                          1
+#define RENDER_TARGET_2                          2
+#define RENDER_TARGET_3                          3
+#define NONE                                     8
+#define Z                                        9
+#define STENCIL                                  10
+#define ZSTENCIL                                 11
+};
+
+static inline void
+V3D41_LOAD_TILE_BUFFER_GENERAL_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                    const struct V3D41_LOAD_TILE_BUFFER_GENERAL * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->flip_y, 7, 7) |
+            __gen_uint(values->memory_format, 4, 6) |
+            __gen_uint(values->buffer_to_load, 0, 3);
+
+   cl[ 2] = __gen_uint(values->input_image_format, 4, 9) |
+            __gen_uint(values->decimate_mode, 2, 3);
+
+   cl[ 3] = __gen_uint(values->r_b_swap, 4, 4) |
+            __gen_uint(values->channel_reverse, 3, 3) |
+            __gen_uint(values->input_image_format, 4, 9) >> 8;
+
+   cl[ 4] = __gen_uint(values->height_in_ub_or_stride, 4, 23);
+
+   cl[ 5] = __gen_uint(values->height_in_ub_or_stride, 4, 23) >> 8;
+
+   cl[ 6] = __gen_uint(values->height_in_ub_or_stride, 4, 23) >> 16;
+
+   cl[ 7] = __gen_uint(values->height, 0, 15);
+
+   cl[ 8] = __gen_uint(values->height, 0, 15) >> 8;
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 9] = __gen_address_offset(&values->address);
+
+   cl[10] = __gen_address_offset(&values->address) >> 8;
+
+   cl[11] = __gen_address_offset(&values->address) >> 16;
+
+   cl[12] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D41_LOAD_TILE_BUFFER_GENERAL_length     13
+#ifdef __gen_unpack_address
+static inline void
+V3D41_LOAD_TILE_BUFFER_GENERAL_unpack(const uint8_t * restrict cl,
+                                      struct V3D41_LOAD_TILE_BUFFER_GENERAL * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 72, 103);
+   values->height = __gen_unpack_uint(cl, 56, 71);
+   values->height_in_ub_or_stride = __gen_unpack_uint(cl, 36, 55);
+   values->r_b_swap = __gen_unpack_uint(cl, 28, 28);
+   values->channel_reverse = __gen_unpack_uint(cl, 27, 27);
+   values->input_image_format = __gen_unpack_uint(cl, 20, 25);
+   values->decimate_mode = __gen_unpack_uint(cl, 18, 19);
+   values->flip_y = __gen_unpack_uint(cl, 15, 15);
+   values->memory_format = __gen_unpack_uint(cl, 12, 14);
+   values->buffer_to_load = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_opcode     31
+#define V3D41_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_header\
+   .opcode                              =     31
+
+struct V3D41_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                              const struct V3D41_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_unpack(const uint8_t * restrict cl,
+                                                struct V3D41_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_INDEXED_PRIM_LIST_opcode        32
+#define V3D41_INDEXED_PRIM_LIST_header          \
+   .opcode                              =     32
+
+struct V3D41_INDEXED_PRIM_LIST {
+   uint32_t                             opcode;
+   uint32_t                             index_offset;
+   bool                                 enable_primitive_restarts;
+   uint32_t                             length;
+   uint32_t                             index_type;
+#define INDEX_TYPE_8_BIT                         0
+#define INDEX_TYPE_16_BIT                        1
+#define INDEX_TYPE_32_BIT                        2
+   enum V3D41_Primitive                 mode;
+};
+
+static inline void
+V3D41_INDEXED_PRIM_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                             const struct V3D41_INDEXED_PRIM_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->index_type, 6, 7) |
+            __gen_uint(values->mode, 0, 5);
+
+   cl[ 2] = __gen_uint(values->length, 0, 30);
+
+   cl[ 3] = __gen_uint(values->length, 0, 30) >> 8;
+
+   cl[ 4] = __gen_uint(values->length, 0, 30) >> 16;
+
+   cl[ 5] = __gen_uint(values->enable_primitive_restarts, 7, 7) |
+            __gen_uint(values->length, 0, 30) >> 24;
+
+
+   memcpy(&cl[6], &values->index_offset, sizeof(values->index_offset));
+}
+
+#define V3D41_INDEXED_PRIM_LIST_length        10
+#ifdef __gen_unpack_address
+static inline void
+V3D41_INDEXED_PRIM_LIST_unpack(const uint8_t * restrict cl,
+                               struct V3D41_INDEXED_PRIM_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->index_offset = __gen_unpack_uint(cl, 48, 79);
+   values->enable_primitive_restarts = __gen_unpack_uint(cl, 47, 47);
+   values->length = __gen_unpack_uint(cl, 16, 46);
+   values->index_type = __gen_unpack_uint(cl, 14, 15);
+   values->mode = __gen_unpack_uint(cl, 8, 13);
+}
+#endif
+
+
+#define V3D41_INDEXED_INSTANCED_PRIM_LIST_opcode     34
+#define V3D41_INDEXED_INSTANCED_PRIM_LIST_header\
+   .opcode                              =     34
+
+struct V3D41_INDEXED_INSTANCED_PRIM_LIST {
+   uint32_t                             opcode;
+   uint32_t                             index_offset;
+   uint32_t                             number_of_instances;
+   bool                                 enable_primitive_restarts;
+   uint32_t                             instance_length;
+   uint32_t                             index_type;
+#define INDEX_TYPE_8_BIT                         0
+#define INDEX_TYPE_16_BIT                        1
+#define INDEX_TYPE_32_BIT                        2
+   enum V3D41_Primitive                 mode;
+};
+
+static inline void
+V3D41_INDEXED_INSTANCED_PRIM_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                       const struct V3D41_INDEXED_INSTANCED_PRIM_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->index_type, 6, 7) |
+            __gen_uint(values->mode, 0, 5);
+
+   cl[ 2] = __gen_uint(values->instance_length, 0, 30);
+
+   cl[ 3] = __gen_uint(values->instance_length, 0, 30) >> 8;
+
+   cl[ 4] = __gen_uint(values->instance_length, 0, 30) >> 16;
+
+   cl[ 5] = __gen_uint(values->enable_primitive_restarts, 7, 7) |
+            __gen_uint(values->instance_length, 0, 30) >> 24;
+
+
+   memcpy(&cl[6], &values->number_of_instances, sizeof(values->number_of_instances));
+
+   memcpy(&cl[10], &values->index_offset, sizeof(values->index_offset));
+}
+
+#define V3D41_INDEXED_INSTANCED_PRIM_LIST_length     14
+#ifdef __gen_unpack_address
+static inline void
+V3D41_INDEXED_INSTANCED_PRIM_LIST_unpack(const uint8_t * restrict cl,
+                                         struct V3D41_INDEXED_INSTANCED_PRIM_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->index_offset = __gen_unpack_uint(cl, 80, 111);
+   values->number_of_instances = __gen_unpack_uint(cl, 48, 79);
+   values->enable_primitive_restarts = __gen_unpack_uint(cl, 47, 47);
+   values->instance_length = __gen_unpack_uint(cl, 16, 46);
+   values->index_type = __gen_unpack_uint(cl, 14, 15);
+   values->mode = __gen_unpack_uint(cl, 8, 13);
+}
+#endif
+
+
+#define V3D41_VERTEX_ARRAY_PRIMS_opcode       36
+#define V3D41_VERTEX_ARRAY_PRIMS_header         \
+   .opcode                              =     36
+
+struct V3D41_VERTEX_ARRAY_PRIMS {
+   uint32_t                             opcode;
+   uint32_t                             index_of_first_vertex;
+   uint32_t                             length;
+   enum V3D41_Primitive                 mode;
+};
+
+static inline void
+V3D41_VERTEX_ARRAY_PRIMS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D41_VERTEX_ARRAY_PRIMS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->mode, 0, 7);
+
+
+   memcpy(&cl[2], &values->length, sizeof(values->length));
+
+   memcpy(&cl[6], &values->index_of_first_vertex, sizeof(values->index_of_first_vertex));
+}
+
+#define V3D41_VERTEX_ARRAY_PRIMS_length       10
+#ifdef __gen_unpack_address
+static inline void
+V3D41_VERTEX_ARRAY_PRIMS_unpack(const uint8_t * restrict cl,
+                                struct V3D41_VERTEX_ARRAY_PRIMS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->index_of_first_vertex = __gen_unpack_uint(cl, 48, 79);
+   values->length = __gen_unpack_uint(cl, 16, 47);
+   values->mode = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D41_VERTEX_ARRAY_INSTANCED_PRIMS_opcode     38
+#define V3D41_VERTEX_ARRAY_INSTANCED_PRIMS_header\
+   .opcode                              =     38
+
+struct V3D41_VERTEX_ARRAY_INSTANCED_PRIMS {
+   uint32_t                             opcode;
+   uint32_t                             index_of_first_vertex;
+   uint32_t                             number_of_instances;
+   uint32_t                             instance_length;
+   enum V3D41_Primitive                 mode;
+};
+
+static inline void
+V3D41_VERTEX_ARRAY_INSTANCED_PRIMS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D41_VERTEX_ARRAY_INSTANCED_PRIMS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->mode, 0, 7);
+
+
+   memcpy(&cl[2], &values->instance_length, sizeof(values->instance_length));
+
+   memcpy(&cl[6], &values->number_of_instances, sizeof(values->number_of_instances));
+
+   memcpy(&cl[10], &values->index_of_first_vertex, sizeof(values->index_of_first_vertex));
+}
+
+#define V3D41_VERTEX_ARRAY_INSTANCED_PRIMS_length     14
+#ifdef __gen_unpack_address
+static inline void
+V3D41_VERTEX_ARRAY_INSTANCED_PRIMS_unpack(const uint8_t * restrict cl,
+                                          struct V3D41_VERTEX_ARRAY_INSTANCED_PRIMS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->index_of_first_vertex = __gen_unpack_uint(cl, 80, 111);
+   values->number_of_instances = __gen_unpack_uint(cl, 48, 79);
+   values->instance_length = __gen_unpack_uint(cl, 16, 47);
+   values->mode = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D41_BASE_VERTEX_BASE_INSTANCE_opcode     43
+#define V3D41_BASE_VERTEX_BASE_INSTANCE_header  \
+   .opcode                              =     43
+
+struct V3D41_BASE_VERTEX_BASE_INSTANCE {
+   uint32_t                             opcode;
+   uint32_t                             base_instance;
+   uint32_t                             base_vertex;
+};
+
+static inline void
+V3D41_BASE_VERTEX_BASE_INSTANCE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                     const struct V3D41_BASE_VERTEX_BASE_INSTANCE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->base_vertex, sizeof(values->base_vertex));
+
+   memcpy(&cl[5], &values->base_instance, sizeof(values->base_instance));
+}
+
+#define V3D41_BASE_VERTEX_BASE_INSTANCE_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_BASE_VERTEX_BASE_INSTANCE_unpack(const uint8_t * restrict cl,
+                                       struct V3D41_BASE_VERTEX_BASE_INSTANCE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->base_instance = __gen_unpack_uint(cl, 40, 71);
+   values->base_vertex = __gen_unpack_uint(cl, 8, 39);
+}
+#endif
+
+
+#define V3D41_INDEX_BUFFER_SETUP_opcode       44
+#define V3D41_INDEX_BUFFER_SETUP_header         \
+   .opcode                              =     44
+
+struct V3D41_INDEX_BUFFER_SETUP {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+   uint32_t                             size;
+};
+
+static inline void
+V3D41_INDEX_BUFFER_SETUP_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D41_INDEX_BUFFER_SETUP * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+
+   memcpy(&cl[5], &values->size, sizeof(values->size));
+}
+
+#define V3D41_INDEX_BUFFER_SETUP_length        9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_INDEX_BUFFER_SETUP_unpack(const uint8_t * restrict cl,
+                                struct V3D41_INDEX_BUFFER_SETUP * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 8, 39);
+   values->size = __gen_unpack_uint(cl, 40, 71);
+}
+#endif
+
+
+#define V3D41_PRIM_LIST_FORMAT_opcode         56
+#define V3D41_PRIM_LIST_FORMAT_header           \
+   .opcode                              =     56
+
+struct V3D41_PRIM_LIST_FORMAT {
+   uint32_t                             opcode;
+   bool                                 tri_strip_or_fan;
+   uint32_t                             primitive_type;
+#define LIST_POINTS                              0
+#define LIST_LINES                               1
+#define LIST_TRIANGLES                           2
+};
+
+static inline void
+V3D41_PRIM_LIST_FORMAT_pack(__gen_user_data *data, uint8_t * restrict cl,
+                            const struct V3D41_PRIM_LIST_FORMAT * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->tri_strip_or_fan, 7, 7) |
+            __gen_uint(values->primitive_type, 0, 5);
+
+}
+
+#define V3D41_PRIM_LIST_FORMAT_length          2
+#ifdef __gen_unpack_address
+static inline void
+V3D41_PRIM_LIST_FORMAT_unpack(const uint8_t * restrict cl,
+                              struct V3D41_PRIM_LIST_FORMAT * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->tri_strip_or_fan = __gen_unpack_uint(cl, 15, 15);
+   values->primitive_type = __gen_unpack_uint(cl, 8, 13);
+}
+#endif
+
+
+#define V3D41_GL_SHADER_STATE_opcode          64
+#define V3D41_GL_SHADER_STATE_header            \
+   .opcode                              =     64
+
+struct V3D41_GL_SHADER_STATE {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+   uint32_t                             number_of_attribute_arrays;
+};
+
+static inline void
+V3D41_GL_SHADER_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                           const struct V3D41_GL_SHADER_STATE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address) |
+            __gen_uint(values->number_of_attribute_arrays, 0, 4);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D41_GL_SHADER_STATE_length           5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_GL_SHADER_STATE_unpack(const uint8_t * restrict cl,
+                             struct V3D41_GL_SHADER_STATE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 13, 39);
+   values->number_of_attribute_arrays = __gen_unpack_uint(cl, 8, 12);
+}
+#endif
+
+
+#define V3D41_VCM_CACHE_SIZE_opcode           71
+#define V3D41_VCM_CACHE_SIZE_header             \
+   .opcode                              =     71
+
+struct V3D41_VCM_CACHE_SIZE {
+   uint32_t                             opcode;
+   uint32_t                             number_of_16_vertex_batches_for_rendering;
+   uint32_t                             number_of_16_vertex_batches_for_binning;
+};
+
+static inline void
+V3D41_VCM_CACHE_SIZE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                          const struct V3D41_VCM_CACHE_SIZE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->number_of_16_vertex_batches_for_rendering, 4, 7) |
+            __gen_uint(values->number_of_16_vertex_batches_for_binning, 0, 3);
+
+}
+
+#define V3D41_VCM_CACHE_SIZE_length            2
+#ifdef __gen_unpack_address
+static inline void
+V3D41_VCM_CACHE_SIZE_unpack(const uint8_t * restrict cl,
+                            struct V3D41_VCM_CACHE_SIZE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->number_of_16_vertex_batches_for_rendering = __gen_unpack_uint(cl, 12, 15);
+   values->number_of_16_vertex_batches_for_binning = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_TRANSFORM_FEEDBACK_BUFFER_opcode     73
+#define V3D41_TRANSFORM_FEEDBACK_BUFFER_header  \
+   .opcode                              =     73
+
+struct V3D41_TRANSFORM_FEEDBACK_BUFFER {
+   uint32_t                             opcode;
+   __gen_address_type                   buffer_address;
+   uint32_t                             buffer_size_in_32_bit_words;
+   uint32_t                             buffer_number;
+};
+
+static inline void
+V3D41_TRANSFORM_FEEDBACK_BUFFER_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                     const struct V3D41_TRANSFORM_FEEDBACK_BUFFER * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->buffer_size_in_32_bit_words, 2, 31) |
+            __gen_uint(values->buffer_number, 0, 1);
+
+   cl[ 2] = __gen_uint(values->buffer_size_in_32_bit_words, 2, 31) >> 8;
+
+   cl[ 3] = __gen_uint(values->buffer_size_in_32_bit_words, 2, 31) >> 16;
+
+   cl[ 4] = __gen_uint(values->buffer_size_in_32_bit_words, 2, 31) >> 24;
+
+   __gen_emit_reloc(data, &values->buffer_address);
+   cl[ 5] = __gen_address_offset(&values->buffer_address);
+
+   cl[ 6] = __gen_address_offset(&values->buffer_address) >> 8;
+
+   cl[ 7] = __gen_address_offset(&values->buffer_address) >> 16;
+
+   cl[ 8] = __gen_address_offset(&values->buffer_address) >> 24;
+
+}
+
+#define V3D41_TRANSFORM_FEEDBACK_BUFFER_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TRANSFORM_FEEDBACK_BUFFER_unpack(const uint8_t * restrict cl,
+                                       struct V3D41_TRANSFORM_FEEDBACK_BUFFER * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->buffer_address = __gen_unpack_address(cl, 40, 71);
+   values->buffer_size_in_32_bit_words = __gen_unpack_uint(cl, 10, 39);
+   values->buffer_number = __gen_unpack_uint(cl, 8, 9);
+}
+#endif
+
+
+#define V3D41_TRANSFORM_FEEDBACK_SPECS_opcode     74
+#define V3D41_TRANSFORM_FEEDBACK_SPECS_header   \
+   .opcode                              =     74
+
+struct V3D41_TRANSFORM_FEEDBACK_SPECS {
+   uint32_t                             opcode;
+   bool                                 enable;
+   uint32_t                             number_of_16_bit_output_data_specs_following;
+};
+
+static inline void
+V3D41_TRANSFORM_FEEDBACK_SPECS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                    const struct V3D41_TRANSFORM_FEEDBACK_SPECS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->enable, 7, 7) |
+            __gen_uint(values->number_of_16_bit_output_data_specs_following, 0, 4);
+
+}
+
+#define V3D41_TRANSFORM_FEEDBACK_SPECS_length      2
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TRANSFORM_FEEDBACK_SPECS_unpack(const uint8_t * restrict cl,
+                                      struct V3D41_TRANSFORM_FEEDBACK_SPECS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->enable = __gen_unpack_uint(cl, 15, 15);
+   values->number_of_16_bit_output_data_specs_following = __gen_unpack_uint(cl, 8, 12);
+}
+#endif
+
+
+#define V3D41_FLUSH_TRANSFORM_FEEDBACK_DATA_opcode     75
+#define V3D41_FLUSH_TRANSFORM_FEEDBACK_DATA_header\
+   .opcode                              =     75
+
+struct V3D41_FLUSH_TRANSFORM_FEEDBACK_DATA {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_FLUSH_TRANSFORM_FEEDBACK_DATA_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                         const struct V3D41_FLUSH_TRANSFORM_FEEDBACK_DATA * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_FLUSH_TRANSFORM_FEEDBACK_DATA_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_FLUSH_TRANSFORM_FEEDBACK_DATA_unpack(const uint8_t * restrict cl,
+                                           struct V3D41_FLUSH_TRANSFORM_FEEDBACK_DATA * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC_header\
+
+
+struct V3D41_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC {
+   uint32_t                             first_shaded_vertex_value_to_output;
+   uint32_t                             number_of_consecutive_vertex_values_to_output_as_32_bit_values;
+   uint32_t                             output_buffer_to_write_to;
+   uint32_t                             stream_number;
+};
+
+static inline void
+V3D41_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                               const struct V3D41_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC * restrict values)
+{
+   assert(values->number_of_consecutive_vertex_values_to_output_as_32_bit_values >= 1);
+   cl[ 0] = __gen_uint(values->first_shaded_vertex_value_to_output, 0, 7);
+
+   cl[ 1] = __gen_uint(values->number_of_consecutive_vertex_values_to_output_as_32_bit_values - 1, 0, 3) |
+            __gen_uint(values->output_buffer_to_write_to, 4, 5) |
+            __gen_uint(values->stream_number, 6, 7);
+
+}
+
+#define V3D41_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC_length      2
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC_unpack(const uint8_t * restrict cl,
+                                                 struct V3D41_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC * restrict values)
+{
+   values->first_shaded_vertex_value_to_output = __gen_unpack_uint(cl, 0, 7);
+   values->number_of_consecutive_vertex_values_to_output_as_32_bit_values = __gen_unpack_uint(cl, 8, 11) + 1;
+   values->output_buffer_to_write_to = __gen_unpack_uint(cl, 12, 13);
+   values->stream_number = __gen_unpack_uint(cl, 14, 15);
+}
+#endif
+
+
+#define V3D41_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS_header\
+
+
+struct V3D41_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS {
+   __gen_address_type                   address;
+};
+
+static inline void
+V3D41_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                             const struct V3D41_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS * restrict values)
+{
+   __gen_emit_reloc(data, &values->address);
+   cl[ 0] = __gen_address_offset(&values->address);
+
+   cl[ 1] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D41_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS_unpack(const uint8_t * restrict cl,
+                                               struct V3D41_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS * restrict values)
+{
+   values->address = __gen_unpack_address(cl, 0, 31);
+}
+#endif
+
+
+#define V3D41_STENCIL_CFG_opcode              80
+#define V3D41_STENCIL_CFG_header                \
+   .opcode                              =     80
+
+struct V3D41_STENCIL_CFG {
+   uint32_t                             opcode;
+   uint32_t                             stencil_write_mask;
+   bool                                 back_config;
+   bool                                 front_config;
+   enum V3D41_Stencil_Op                stencil_pass_op;
+   enum V3D41_Stencil_Op                depth_test_fail_op;
+   enum V3D41_Stencil_Op                stencil_test_fail_op;
+   enum V3D41_Compare_Function          stencil_test_function;
+   uint32_t                             stencil_test_mask;
+   uint32_t                             stencil_ref_value;
+};
+
+static inline void
+V3D41_STENCIL_CFG_pack(__gen_user_data *data, uint8_t * restrict cl,
+                       const struct V3D41_STENCIL_CFG * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->stencil_ref_value, 0, 7);
+
+   cl[ 2] = __gen_uint(values->stencil_test_mask, 0, 7);
+
+   cl[ 3] = __gen_uint(values->depth_test_fail_op, 6, 8) |
+            __gen_uint(values->stencil_test_fail_op, 3, 5) |
+            __gen_uint(values->stencil_test_function, 0, 2);
+
+   cl[ 4] = __gen_uint(values->back_config, 5, 5) |
+            __gen_uint(values->front_config, 4, 4) |
+            __gen_uint(values->stencil_pass_op, 1, 3) |
+            __gen_uint(values->depth_test_fail_op, 6, 8) >> 8;
+
+   cl[ 5] = __gen_uint(values->stencil_write_mask, 0, 7);
+
+}
+
+#define V3D41_STENCIL_CFG_length               6
+#ifdef __gen_unpack_address
+static inline void
+V3D41_STENCIL_CFG_unpack(const uint8_t * restrict cl,
+                         struct V3D41_STENCIL_CFG * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->stencil_write_mask = __gen_unpack_uint(cl, 40, 47);
+   values->back_config = __gen_unpack_uint(cl, 37, 37);
+   values->front_config = __gen_unpack_uint(cl, 36, 36);
+   values->stencil_pass_op = __gen_unpack_uint(cl, 33, 35);
+   values->depth_test_fail_op = __gen_unpack_uint(cl, 30, 32);
+   values->stencil_test_fail_op = __gen_unpack_uint(cl, 27, 29);
+   values->stencil_test_function = __gen_unpack_uint(cl, 24, 26);
+   values->stencil_test_mask = __gen_unpack_uint(cl, 16, 23);
+   values->stencil_ref_value = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D41_BLEND_ENABLES_opcode            83
+#define V3D41_BLEND_ENABLES_header              \
+   .opcode                              =     83
+
+struct V3D41_BLEND_ENABLES {
+   uint32_t                             opcode;
+   uint32_t                             mask;
+};
+
+static inline void
+V3D41_BLEND_ENABLES_pack(__gen_user_data *data, uint8_t * restrict cl,
+                         const struct V3D41_BLEND_ENABLES * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->mask, 0, 7);
+
+}
+
+#define V3D41_BLEND_ENABLES_length             2
+#ifdef __gen_unpack_address
+static inline void
+V3D41_BLEND_ENABLES_unpack(const uint8_t * restrict cl,
+                           struct V3D41_BLEND_ENABLES * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->mask = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D41_BLEND_CFG_opcode                84
+#define V3D41_BLEND_CFG_header                  \
+   .opcode                              =     84
+
+struct V3D41_BLEND_CFG {
+   uint32_t                             opcode;
+   uint32_t                             render_target_mask;
+   enum V3D41_Blend_Factor              color_blend_dst_factor;
+   enum V3D41_Blend_Factor              color_blend_src_factor;
+   enum V3D41_Blend_Mode                color_blend_mode;
+   enum V3D41_Blend_Factor              alpha_blend_dst_factor;
+   enum V3D41_Blend_Factor              alpha_blend_src_factor;
+   enum V3D41_Blend_Mode                alpha_blend_mode;
+};
+
+static inline void
+V3D41_BLEND_CFG_pack(__gen_user_data *data, uint8_t * restrict cl,
+                     const struct V3D41_BLEND_CFG * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->alpha_blend_src_factor, 4, 7) |
+            __gen_uint(values->alpha_blend_mode, 0, 3);
+
+   cl[ 2] = __gen_uint(values->color_blend_mode, 4, 7) |
+            __gen_uint(values->alpha_blend_dst_factor, 0, 3);
+
+   cl[ 3] = __gen_uint(values->color_blend_dst_factor, 4, 7) |
+            __gen_uint(values->color_blend_src_factor, 0, 3);
+
+   cl[ 4] = __gen_uint(values->render_target_mask, 0, 3);
+
+}
+
+#define V3D41_BLEND_CFG_length                 5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_BLEND_CFG_unpack(const uint8_t * restrict cl,
+                       struct V3D41_BLEND_CFG * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->render_target_mask = __gen_unpack_uint(cl, 32, 35);
+   values->color_blend_dst_factor = __gen_unpack_uint(cl, 28, 31);
+   values->color_blend_src_factor = __gen_unpack_uint(cl, 24, 27);
+   values->color_blend_mode = __gen_unpack_uint(cl, 20, 23);
+   values->alpha_blend_dst_factor = __gen_unpack_uint(cl, 16, 19);
+   values->alpha_blend_src_factor = __gen_unpack_uint(cl, 12, 15);
+   values->alpha_blend_mode = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_BLEND_CONSTANT_COLOR_opcode     86
+#define V3D41_BLEND_CONSTANT_COLOR_header       \
+   .opcode                              =     86
+
+struct V3D41_BLEND_CONSTANT_COLOR {
+   uint32_t                             opcode;
+   uint32_t                             alpha_f16;
+   uint32_t                             blue_f16;
+   uint32_t                             green_f16;
+   uint32_t                             red_f16;
+};
+
+static inline void
+V3D41_BLEND_CONSTANT_COLOR_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                const struct V3D41_BLEND_CONSTANT_COLOR * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->red_f16, 0, 15);
+
+   cl[ 2] = __gen_uint(values->red_f16, 0, 15) >> 8;
+
+   cl[ 3] = __gen_uint(values->green_f16, 0, 15);
+
+   cl[ 4] = __gen_uint(values->green_f16, 0, 15) >> 8;
+
+   cl[ 5] = __gen_uint(values->blue_f16, 0, 15);
+
+   cl[ 6] = __gen_uint(values->blue_f16, 0, 15) >> 8;
+
+   cl[ 7] = __gen_uint(values->alpha_f16, 0, 15);
+
+   cl[ 8] = __gen_uint(values->alpha_f16, 0, 15) >> 8;
+
+}
+
+#define V3D41_BLEND_CONSTANT_COLOR_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_BLEND_CONSTANT_COLOR_unpack(const uint8_t * restrict cl,
+                                  struct V3D41_BLEND_CONSTANT_COLOR * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->alpha_f16 = __gen_unpack_uint(cl, 56, 71);
+   values->blue_f16 = __gen_unpack_uint(cl, 40, 55);
+   values->green_f16 = __gen_unpack_uint(cl, 24, 39);
+   values->red_f16 = __gen_unpack_uint(cl, 8, 23);
+}
+#endif
+
+
+#define V3D41_COLOR_WRITE_MASKS_opcode        87
+#define V3D41_COLOR_WRITE_MASKS_header          \
+   .opcode                              =     87
+
+struct V3D41_COLOR_WRITE_MASKS {
+   uint32_t                             opcode;
+   uint32_t                             mask;
+};
+
+static inline void
+V3D41_COLOR_WRITE_MASKS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                             const struct V3D41_COLOR_WRITE_MASKS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->mask, sizeof(values->mask));
+}
+
+#define V3D41_COLOR_WRITE_MASKS_length         5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_COLOR_WRITE_MASKS_unpack(const uint8_t * restrict cl,
+                               struct V3D41_COLOR_WRITE_MASKS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->mask = __gen_unpack_uint(cl, 8, 39);
+}
+#endif
+
+
+#define V3D41_ZERO_ALL_CENTROID_FLAGS_opcode     88
+#define V3D41_ZERO_ALL_CENTROID_FLAGS_header    \
+   .opcode                              =     88
+
+struct V3D41_ZERO_ALL_CENTROID_FLAGS {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_ZERO_ALL_CENTROID_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                   const struct V3D41_ZERO_ALL_CENTROID_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_ZERO_ALL_CENTROID_FLAGS_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_ZERO_ALL_CENTROID_FLAGS_unpack(const uint8_t * restrict cl,
+                                     struct V3D41_ZERO_ALL_CENTROID_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_CENTROID_FLAGS_opcode           89
+#define V3D41_CENTROID_FLAGS_header             \
+   .opcode                              =     89
+
+struct V3D41_CENTROID_FLAGS {
+   uint32_t                             opcode;
+   uint32_t                             centroid_flags_for_varyings_v024;
+   enum V3D41_Varying_Flags_Action      action_for_centroid_flags_of_higher_numbered_varyings;
+   enum V3D41_Varying_Flags_Action      action_for_centroid_flags_of_lower_numbered_varyings;
+   uint32_t                             varying_offset_v0;
+};
+
+static inline void
+V3D41_CENTROID_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                          const struct V3D41_CENTROID_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->action_for_centroid_flags_of_higher_numbered_varyings, 6, 7) |
+            __gen_uint(values->action_for_centroid_flags_of_lower_numbered_varyings, 4, 5) |
+            __gen_uint(values->varying_offset_v0, 0, 3);
+
+   cl[ 2] = __gen_uint(values->centroid_flags_for_varyings_v024, 0, 23);
+
+   cl[ 3] = __gen_uint(values->centroid_flags_for_varyings_v024, 0, 23) >> 8;
+
+   cl[ 4] = __gen_uint(values->centroid_flags_for_varyings_v024, 0, 23) >> 16;
+
+}
+
+#define V3D41_CENTROID_FLAGS_length            5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_CENTROID_FLAGS_unpack(const uint8_t * restrict cl,
+                            struct V3D41_CENTROID_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->centroid_flags_for_varyings_v024 = __gen_unpack_uint(cl, 16, 39);
+   values->action_for_centroid_flags_of_higher_numbered_varyings = __gen_unpack_uint(cl, 14, 15);
+   values->action_for_centroid_flags_of_lower_numbered_varyings = __gen_unpack_uint(cl, 12, 13);
+   values->varying_offset_v0 = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_SAMPLE_STATE_opcode             91
+#define V3D41_SAMPLE_STATE_header               \
+   .opcode                              =     91
+
+struct V3D41_SAMPLE_STATE {
+   uint32_t                             opcode;
+   float                                coverage;
+   uint32_t                             mask;
+};
+
+static inline void
+V3D41_SAMPLE_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                        const struct V3D41_SAMPLE_STATE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->mask, 0, 3);
+
+   cl[ 2] = 0;
+   cl[ 3] = __gen_uint(fui(values->coverage) >> 16, 0, 15);
+
+   cl[ 4] = __gen_uint(fui(values->coverage) >> 16, 0, 15) >> 8;
+
+}
+
+#define V3D41_SAMPLE_STATE_length              5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_SAMPLE_STATE_unpack(const uint8_t * restrict cl,
+                          struct V3D41_SAMPLE_STATE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->coverage = __gen_unpack_f187(cl, 24, 39);
+   values->mask = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_OCCLUSION_QUERY_COUNTER_opcode     92
+#define V3D41_OCCLUSION_QUERY_COUNTER_header    \
+   .opcode                              =     92
+
+struct V3D41_OCCLUSION_QUERY_COUNTER {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+};
+
+static inline void
+V3D41_OCCLUSION_QUERY_COUNTER_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                   const struct V3D41_OCCLUSION_QUERY_COUNTER * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D41_OCCLUSION_QUERY_COUNTER_length      5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_OCCLUSION_QUERY_COUNTER_unpack(const uint8_t * restrict cl,
+                                     struct V3D41_OCCLUSION_QUERY_COUNTER * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 8, 39);
+}
+#endif
+
+
+#define V3D41_CFG_BITS_opcode                 96
+#define V3D41_CFG_BITS_header                   \
+   .opcode                              =     96
+
+struct V3D41_CFG_BITS {
+   uint32_t                             opcode;
+   bool                                 direct3d_provoking_vertex;
+   bool                                 direct3d_point_fill_mode;
+   bool                                 blend_enable;
+   bool                                 stencil_enable;
+   bool                                 early_z_updates_enable;
+   bool                                 early_z_enable;
+   bool                                 z_updates_enable;
+   enum V3D41_Compare_Function          depth_test_function;
+   bool                                 direct3d_wireframe_triangles_mode;
+   uint32_t                             rasterizer_oversample_mode;
+   uint32_t                             line_rasterization;
+   bool                                 enable_depth_offset;
+   bool                                 clockwise_primitives;
+   bool                                 enable_reverse_facing_primitive;
+   bool                                 enable_forward_facing_primitive;
+};
+
+static inline void
+V3D41_CFG_BITS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                    const struct V3D41_CFG_BITS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->rasterizer_oversample_mode, 6, 7) |
+            __gen_uint(values->line_rasterization, 4, 5) |
+            __gen_uint(values->enable_depth_offset, 3, 3) |
+            __gen_uint(values->clockwise_primitives, 2, 2) |
+            __gen_uint(values->enable_reverse_facing_primitive, 1, 1) |
+            __gen_uint(values->enable_forward_facing_primitive, 0, 0);
+
+   cl[ 2] = __gen_uint(values->z_updates_enable, 7, 7) |
+            __gen_uint(values->depth_test_function, 4, 6) |
+            __gen_uint(values->direct3d_wireframe_triangles_mode, 3, 3);
+
+   cl[ 3] = __gen_uint(values->direct3d_provoking_vertex, 5, 5) |
+            __gen_uint(values->direct3d_point_fill_mode, 4, 4) |
+            __gen_uint(values->blend_enable, 3, 3) |
+            __gen_uint(values->stencil_enable, 2, 2) |
+            __gen_uint(values->early_z_updates_enable, 1, 1) |
+            __gen_uint(values->early_z_enable, 0, 0);
+
+}
+
+#define V3D41_CFG_BITS_length                  4
+#ifdef __gen_unpack_address
+static inline void
+V3D41_CFG_BITS_unpack(const uint8_t * restrict cl,
+                      struct V3D41_CFG_BITS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->direct3d_provoking_vertex = __gen_unpack_uint(cl, 29, 29);
+   values->direct3d_point_fill_mode = __gen_unpack_uint(cl, 28, 28);
+   values->blend_enable = __gen_unpack_uint(cl, 27, 27);
+   values->stencil_enable = __gen_unpack_uint(cl, 26, 26);
+   values->early_z_updates_enable = __gen_unpack_uint(cl, 25, 25);
+   values->early_z_enable = __gen_unpack_uint(cl, 24, 24);
+   values->z_updates_enable = __gen_unpack_uint(cl, 23, 23);
+   values->depth_test_function = __gen_unpack_uint(cl, 20, 22);
+   values->direct3d_wireframe_triangles_mode = __gen_unpack_uint(cl, 19, 19);
+   values->rasterizer_oversample_mode = __gen_unpack_uint(cl, 14, 15);
+   values->line_rasterization = __gen_unpack_uint(cl, 12, 13);
+   values->enable_depth_offset = __gen_unpack_uint(cl, 11, 11);
+   values->clockwise_primitives = __gen_unpack_uint(cl, 10, 10);
+   values->enable_reverse_facing_primitive = __gen_unpack_uint(cl, 9, 9);
+   values->enable_forward_facing_primitive = __gen_unpack_uint(cl, 8, 8);
+}
+#endif
+
+
+#define V3D41_ZERO_ALL_FLAT_SHADE_FLAGS_opcode     97
+#define V3D41_ZERO_ALL_FLAT_SHADE_FLAGS_header  \
+   .opcode                              =     97
+
+struct V3D41_ZERO_ALL_FLAT_SHADE_FLAGS {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_ZERO_ALL_FLAT_SHADE_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                     const struct V3D41_ZERO_ALL_FLAT_SHADE_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_ZERO_ALL_FLAT_SHADE_FLAGS_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_ZERO_ALL_FLAT_SHADE_FLAGS_unpack(const uint8_t * restrict cl,
+                                       struct V3D41_ZERO_ALL_FLAT_SHADE_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_FLAT_SHADE_FLAGS_opcode         98
+#define V3D41_FLAT_SHADE_FLAGS_header           \
+   .opcode                              =     98
+
+struct V3D41_FLAT_SHADE_FLAGS {
+   uint32_t                             opcode;
+   uint32_t                             flat_shade_flags_for_varyings_v024;
+   enum V3D41_Varying_Flags_Action      action_for_flat_shade_flags_of_higher_numbered_varyings;
+   enum V3D41_Varying_Flags_Action      action_for_flat_shade_flags_of_lower_numbered_varyings;
+   uint32_t                             varying_offset_v0;
+};
+
+static inline void
+V3D41_FLAT_SHADE_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                            const struct V3D41_FLAT_SHADE_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->action_for_flat_shade_flags_of_higher_numbered_varyings, 6, 7) |
+            __gen_uint(values->action_for_flat_shade_flags_of_lower_numbered_varyings, 4, 5) |
+            __gen_uint(values->varying_offset_v0, 0, 3);
+
+   cl[ 2] = __gen_uint(values->flat_shade_flags_for_varyings_v024, 0, 23);
+
+   cl[ 3] = __gen_uint(values->flat_shade_flags_for_varyings_v024, 0, 23) >> 8;
+
+   cl[ 4] = __gen_uint(values->flat_shade_flags_for_varyings_v024, 0, 23) >> 16;
+
+}
+
+#define V3D41_FLAT_SHADE_FLAGS_length          5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_FLAT_SHADE_FLAGS_unpack(const uint8_t * restrict cl,
+                              struct V3D41_FLAT_SHADE_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->flat_shade_flags_for_varyings_v024 = __gen_unpack_uint(cl, 16, 39);
+   values->action_for_flat_shade_flags_of_higher_numbered_varyings = __gen_unpack_uint(cl, 14, 15);
+   values->action_for_flat_shade_flags_of_lower_numbered_varyings = __gen_unpack_uint(cl, 12, 13);
+   values->varying_offset_v0 = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_ZERO_ALL_NON_PERSPECTIVE_FLAGS_opcode     99
+#define V3D41_ZERO_ALL_NON_PERSPECTIVE_FLAGS_header\
+   .opcode                              =     99
+
+struct V3D41_ZERO_ALL_NON_PERSPECTIVE_FLAGS {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_ZERO_ALL_NON_PERSPECTIVE_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                          const struct V3D41_ZERO_ALL_NON_PERSPECTIVE_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_ZERO_ALL_NON_PERSPECTIVE_FLAGS_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_ZERO_ALL_NON_PERSPECTIVE_FLAGS_unpack(const uint8_t * restrict cl,
+                                            struct V3D41_ZERO_ALL_NON_PERSPECTIVE_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_NON_PERSPECTIVE_FLAGS_opcode    100
+#define V3D41_NON_PERSPECTIVE_FLAGS_header      \
+   .opcode                              =    100
+
+struct V3D41_NON_PERSPECTIVE_FLAGS {
+   uint32_t                             opcode;
+   uint32_t                             non_perspective_flags_for_varyings_v024;
+   enum V3D41_Varying_Flags_Action      action_for_non_perspective_flags_of_higher_numbered_varyings;
+   enum V3D41_Varying_Flags_Action      action_for_non_perspective_flags_of_lower_numbered_varyings;
+   uint32_t                             varying_offset_v0;
+};
+
+static inline void
+V3D41_NON_PERSPECTIVE_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                 const struct V3D41_NON_PERSPECTIVE_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->action_for_non_perspective_flags_of_higher_numbered_varyings, 6, 7) |
+            __gen_uint(values->action_for_non_perspective_flags_of_lower_numbered_varyings, 4, 5) |
+            __gen_uint(values->varying_offset_v0, 0, 3);
+
+   cl[ 2] = __gen_uint(values->non_perspective_flags_for_varyings_v024, 0, 23);
+
+   cl[ 3] = __gen_uint(values->non_perspective_flags_for_varyings_v024, 0, 23) >> 8;
+
+   cl[ 4] = __gen_uint(values->non_perspective_flags_for_varyings_v024, 0, 23) >> 16;
+
+}
+
+#define V3D41_NON_PERSPECTIVE_FLAGS_length      5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_NON_PERSPECTIVE_FLAGS_unpack(const uint8_t * restrict cl,
+                                   struct V3D41_NON_PERSPECTIVE_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->non_perspective_flags_for_varyings_v024 = __gen_unpack_uint(cl, 16, 39);
+   values->action_for_non_perspective_flags_of_higher_numbered_varyings = __gen_unpack_uint(cl, 14, 15);
+   values->action_for_non_perspective_flags_of_lower_numbered_varyings = __gen_unpack_uint(cl, 12, 13);
+   values->varying_offset_v0 = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_POINT_SIZE_opcode              104
+#define V3D41_POINT_SIZE_header                 \
+   .opcode                              =    104
+
+struct V3D41_POINT_SIZE {
+   uint32_t                             opcode;
+   float                                point_size;
+};
+
+static inline void
+V3D41_POINT_SIZE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                      const struct V3D41_POINT_SIZE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->point_size, sizeof(values->point_size));
+}
+
+#define V3D41_POINT_SIZE_length                5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_POINT_SIZE_unpack(const uint8_t * restrict cl,
+                        struct V3D41_POINT_SIZE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->point_size = __gen_unpack_float(cl, 8, 39);
+}
+#endif
+
+
+#define V3D41_LINE_WIDTH_opcode              105
+#define V3D41_LINE_WIDTH_header                 \
+   .opcode                              =    105
+
+struct V3D41_LINE_WIDTH {
+   uint32_t                             opcode;
+   float                                line_width;
+};
+
+static inline void
+V3D41_LINE_WIDTH_pack(__gen_user_data *data, uint8_t * restrict cl,
+                      const struct V3D41_LINE_WIDTH * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->line_width, sizeof(values->line_width));
+}
+
+#define V3D41_LINE_WIDTH_length                5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_LINE_WIDTH_unpack(const uint8_t * restrict cl,
+                        struct V3D41_LINE_WIDTH * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->line_width = __gen_unpack_float(cl, 8, 39);
+}
+#endif
+
+
+#define V3D41_DEPTH_OFFSET_opcode            106
+#define V3D41_DEPTH_OFFSET_header               \
+   .opcode                              =    106
+
+struct V3D41_DEPTH_OFFSET {
+   uint32_t                             opcode;
+   float                                limit;
+   float                                depth_offset_units;
+   float                                depth_offset_factor;
+};
+
+static inline void
+V3D41_DEPTH_OFFSET_pack(__gen_user_data *data, uint8_t * restrict cl,
+                        const struct V3D41_DEPTH_OFFSET * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(fui(values->depth_offset_factor) >> 16, 0, 15);
+
+   cl[ 2] = __gen_uint(fui(values->depth_offset_factor) >> 16, 0, 15) >> 8;
+
+   cl[ 3] = __gen_uint(fui(values->depth_offset_units) >> 16, 0, 15);
+
+   cl[ 4] = __gen_uint(fui(values->depth_offset_units) >> 16, 0, 15) >> 8;
+
+
+   memcpy(&cl[5], &values->limit, sizeof(values->limit));
+}
+
+#define V3D41_DEPTH_OFFSET_length              9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_DEPTH_OFFSET_unpack(const uint8_t * restrict cl,
+                          struct V3D41_DEPTH_OFFSET * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->limit = __gen_unpack_float(cl, 40, 71);
+   values->depth_offset_units = __gen_unpack_f187(cl, 24, 39);
+   values->depth_offset_factor = __gen_unpack_f187(cl, 8, 23);
+}
+#endif
+
+
+#define V3D41_CLIP_WINDOW_opcode             107
+#define V3D41_CLIP_WINDOW_header                \
+   .opcode                              =    107
+
+struct V3D41_CLIP_WINDOW {
+   uint32_t                             opcode;
+   uint32_t                             clip_window_height_in_pixels;
+   uint32_t                             clip_window_width_in_pixels;
+   uint32_t                             clip_window_bottom_pixel_coordinate;
+   uint32_t                             clip_window_left_pixel_coordinate;
+};
+
+static inline void
+V3D41_CLIP_WINDOW_pack(__gen_user_data *data, uint8_t * restrict cl,
+                       const struct V3D41_CLIP_WINDOW * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->clip_window_left_pixel_coordinate, 0, 15);
+
+   cl[ 2] = __gen_uint(values->clip_window_left_pixel_coordinate, 0, 15) >> 8;
+
+   cl[ 3] = __gen_uint(values->clip_window_bottom_pixel_coordinate, 0, 15);
+
+   cl[ 4] = __gen_uint(values->clip_window_bottom_pixel_coordinate, 0, 15) >> 8;
+
+   cl[ 5] = __gen_uint(values->clip_window_width_in_pixels, 0, 15);
+
+   cl[ 6] = __gen_uint(values->clip_window_width_in_pixels, 0, 15) >> 8;
+
+   cl[ 7] = __gen_uint(values->clip_window_height_in_pixels, 0, 15);
+
+   cl[ 8] = __gen_uint(values->clip_window_height_in_pixels, 0, 15) >> 8;
+
+}
+
+#define V3D41_CLIP_WINDOW_length               9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_CLIP_WINDOW_unpack(const uint8_t * restrict cl,
+                         struct V3D41_CLIP_WINDOW * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->clip_window_height_in_pixels = __gen_unpack_uint(cl, 56, 71);
+   values->clip_window_width_in_pixels = __gen_unpack_uint(cl, 40, 55);
+   values->clip_window_bottom_pixel_coordinate = __gen_unpack_uint(cl, 24, 39);
+   values->clip_window_left_pixel_coordinate = __gen_unpack_uint(cl, 8, 23);
+}
+#endif
+
+
+#define V3D41_VIEWPORT_OFFSET_opcode         108
+#define V3D41_VIEWPORT_OFFSET_header            \
+   .opcode                              =    108
+
+struct V3D41_VIEWPORT_OFFSET {
+   uint32_t                             opcode;
+   uint32_t                             coarse_y;
+   float                                viewport_centre_y_coordinate;
+   uint32_t                             coarse_x;
+   float                                viewport_centre_x_coordinate;
+};
+
+static inline void
+V3D41_VIEWPORT_OFFSET_pack(__gen_user_data *data, uint8_t * restrict cl,
+                           const struct V3D41_VIEWPORT_OFFSET * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_sfixed(values->viewport_centre_x_coordinate, 0, 21, 8);
+
+   cl[ 2] = __gen_sfixed(values->viewport_centre_x_coordinate, 0, 21, 8) >> 8;
+
+   cl[ 3] = __gen_uint(values->coarse_x, 6, 15) |
+            __gen_sfixed(values->viewport_centre_x_coordinate, 0, 21, 8) >> 16;
+
+   cl[ 4] = __gen_uint(values->coarse_x, 6, 15) >> 8;
+
+   cl[ 5] = __gen_sfixed(values->viewport_centre_y_coordinate, 0, 21, 8);
+
+   cl[ 6] = __gen_sfixed(values->viewport_centre_y_coordinate, 0, 21, 8) >> 8;
+
+   cl[ 7] = __gen_uint(values->coarse_y, 6, 15) |
+            __gen_sfixed(values->viewport_centre_y_coordinate, 0, 21, 8) >> 16;
+
+   cl[ 8] = __gen_uint(values->coarse_y, 6, 15) >> 8;
+
+}
+
+#define V3D41_VIEWPORT_OFFSET_length           9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_VIEWPORT_OFFSET_unpack(const uint8_t * restrict cl,
+                             struct V3D41_VIEWPORT_OFFSET * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->coarse_y = __gen_unpack_uint(cl, 62, 71);
+   values->viewport_centre_y_coordinate = __gen_unpack_sfixed(cl, 40, 61, 8);
+   values->coarse_x = __gen_unpack_uint(cl, 30, 39);
+   values->viewport_centre_x_coordinate = __gen_unpack_sfixed(cl, 8, 29, 8);
+}
+#endif
+
+
+#define V3D41_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES_opcode    109
+#define V3D41_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES_header\
+   .opcode                              =    109
+
+struct V3D41_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES {
+   uint32_t                             opcode;
+   float                                maximum_zw;
+   float                                minimum_zw;
+};
+
+static inline void
+V3D41_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                             const struct V3D41_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->minimum_zw, sizeof(values->minimum_zw));
+
+   memcpy(&cl[5], &values->maximum_zw, sizeof(values->maximum_zw));
+}
+
+#define V3D41_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES_unpack(const uint8_t * restrict cl,
+                                               struct V3D41_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->maximum_zw = __gen_unpack_float(cl, 40, 71);
+   values->minimum_zw = __gen_unpack_float(cl, 8, 39);
+}
+#endif
+
+
+#define V3D41_CLIPPER_XY_SCALING_opcode      110
+#define V3D41_CLIPPER_XY_SCALING_header         \
+   .opcode                              =    110
+
+struct V3D41_CLIPPER_XY_SCALING {
+   uint32_t                             opcode;
+   float                                viewport_half_height_in_1_256th_of_pixel;
+   float                                viewport_half_width_in_1_256th_of_pixel;
+};
+
+static inline void
+V3D41_CLIPPER_XY_SCALING_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D41_CLIPPER_XY_SCALING * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->viewport_half_width_in_1_256th_of_pixel, sizeof(values->viewport_half_width_in_1_256th_of_pixel));
+
+   memcpy(&cl[5], &values->viewport_half_height_in_1_256th_of_pixel, sizeof(values->viewport_half_height_in_1_256th_of_pixel));
+}
+
+#define V3D41_CLIPPER_XY_SCALING_length        9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_CLIPPER_XY_SCALING_unpack(const uint8_t * restrict cl,
+                                struct V3D41_CLIPPER_XY_SCALING * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->viewport_half_height_in_1_256th_of_pixel = __gen_unpack_float(cl, 40, 71);
+   values->viewport_half_width_in_1_256th_of_pixel = __gen_unpack_float(cl, 8, 39);
+}
+#endif
+
+
+#define V3D41_CLIPPER_Z_SCALE_AND_OFFSET_opcode    111
+#define V3D41_CLIPPER_Z_SCALE_AND_OFFSET_header \
+   .opcode                              =    111
+
+struct V3D41_CLIPPER_Z_SCALE_AND_OFFSET {
+   uint32_t                             opcode;
+   float                                viewport_z_offset_zc_to_zs;
+   float                                viewport_z_scale_zc_to_zs;
+};
+
+static inline void
+V3D41_CLIPPER_Z_SCALE_AND_OFFSET_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                      const struct V3D41_CLIPPER_Z_SCALE_AND_OFFSET * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->viewport_z_scale_zc_to_zs, sizeof(values->viewport_z_scale_zc_to_zs));
+
+   memcpy(&cl[5], &values->viewport_z_offset_zc_to_zs, sizeof(values->viewport_z_offset_zc_to_zs));
+}
+
+#define V3D41_CLIPPER_Z_SCALE_AND_OFFSET_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_CLIPPER_Z_SCALE_AND_OFFSET_unpack(const uint8_t * restrict cl,
+                                        struct V3D41_CLIPPER_Z_SCALE_AND_OFFSET * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->viewport_z_offset_zc_to_zs = __gen_unpack_float(cl, 40, 71);
+   values->viewport_z_scale_zc_to_zs = __gen_unpack_float(cl, 8, 39);
+}
+#endif
+
+
+#define V3D41_NUMBER_OF_LAYERS_opcode        119
+#define V3D41_NUMBER_OF_LAYERS_header           \
+   .opcode                              =    119
+
+struct V3D41_NUMBER_OF_LAYERS {
+   uint32_t                             opcode;
+   uint32_t                             number_of_layers;
+};
+
+static inline void
+V3D41_NUMBER_OF_LAYERS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                            const struct V3D41_NUMBER_OF_LAYERS * restrict values)
+{
+   assert(values->number_of_layers >= 1);
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->number_of_layers - 1, 0, 7);
+
+}
+
+#define V3D41_NUMBER_OF_LAYERS_length          2
+#ifdef __gen_unpack_address
+static inline void
+V3D41_NUMBER_OF_LAYERS_unpack(const uint8_t * restrict cl,
+                              struct V3D41_NUMBER_OF_LAYERS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->number_of_layers = __gen_unpack_uint(cl, 8, 15) + 1;
+}
+#endif
+
+
+#define V3D41_TILE_BINNING_MODE_CFG_opcode    120
+#define V3D41_TILE_BINNING_MODE_CFG_header      \
+   .opcode                              =    120
+
+struct V3D41_TILE_BINNING_MODE_CFG {
+   uint32_t                             opcode;
+   uint32_t                             height_in_pixels;
+   uint32_t                             width_in_pixels;
+   bool                                 double_buffer_in_non_ms_mode;
+   bool                                 multisample_mode_4x;
+   uint32_t                             maximum_bpp_of_all_render_targets;
+#define RENDER_TARGET_MAXIMUM_32BPP              0
+#define RENDER_TARGET_MAXIMUM_64BPP              1
+#define RENDER_TARGET_MAXIMUM_128BPP             2
+   uint32_t                             number_of_render_targets;
+   uint32_t                             tile_allocation_block_size;
+#define TILE_ALLOCATION_BLOCK_SIZE_64B           0
+#define TILE_ALLOCATION_BLOCK_SIZE_128B          1
+#define TILE_ALLOCATION_BLOCK_SIZE_256B          2
+   uint32_t                             tile_allocation_initial_block_size;
+#define TILE_ALLOCATION_INITIAL_BLOCK_SIZE_64B   0
+#define TILE_ALLOCATION_INITIAL_BLOCK_SIZE_128B  1
+#define TILE_ALLOCATION_INITIAL_BLOCK_SIZE_256B  2
+};
+
+static inline void
+V3D41_TILE_BINNING_MODE_CFG_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                 const struct V3D41_TILE_BINNING_MODE_CFG * restrict values)
+{
+   assert(values->height_in_pixels >= 1);
+   assert(values->width_in_pixels >= 1);
+   assert(values->number_of_render_targets >= 1);
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->tile_allocation_block_size, 4, 5) |
+            __gen_uint(values->tile_allocation_initial_block_size, 2, 3);
+
+   cl[ 2] = __gen_uint(values->double_buffer_in_non_ms_mode, 7, 7) |
+            __gen_uint(values->multisample_mode_4x, 6, 6) |
+            __gen_uint(values->maximum_bpp_of_all_render_targets, 4, 5) |
+            __gen_uint(values->number_of_render_targets - 1, 0, 3);
+
+   cl[ 3] = 0;
+   cl[ 4] = 0;
+   cl[ 5] = __gen_uint(values->width_in_pixels - 1, 0, 11);
+
+   cl[ 6] = __gen_uint(values->width_in_pixels - 1, 0, 11) >> 8;
+
+   cl[ 7] = __gen_uint(values->height_in_pixels - 1, 0, 11);
+
+   cl[ 8] = __gen_uint(values->height_in_pixels - 1, 0, 11) >> 8;
+
+}
+
+#define V3D41_TILE_BINNING_MODE_CFG_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TILE_BINNING_MODE_CFG_unpack(const uint8_t * restrict cl,
+                                   struct V3D41_TILE_BINNING_MODE_CFG * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->height_in_pixels = __gen_unpack_uint(cl, 56, 67) + 1;
+   values->width_in_pixels = __gen_unpack_uint(cl, 40, 51) + 1;
+   values->double_buffer_in_non_ms_mode = __gen_unpack_uint(cl, 23, 23);
+   values->multisample_mode_4x = __gen_unpack_uint(cl, 22, 22);
+   values->maximum_bpp_of_all_render_targets = __gen_unpack_uint(cl, 20, 21);
+   values->number_of_render_targets = __gen_unpack_uint(cl, 16, 19) + 1;
+   values->tile_allocation_block_size = __gen_unpack_uint(cl, 12, 13);
+   values->tile_allocation_initial_block_size = __gen_unpack_uint(cl, 10, 11);
+}
+#endif
+
+
+#define V3D41_TILE_RENDERING_MODE_CFG_COMMON_opcode    121
+#define V3D41_TILE_RENDERING_MODE_CFG_COMMON_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      0
+
+struct V3D41_TILE_RENDERING_MODE_CFG_COMMON {
+   uint32_t                             opcode;
+   uint32_t                             pad;
+   bool                                 early_depth_stencil_clear;
+   enum V3D41_Internal_Depth_Type       internal_depth_type;
+   bool                                 early_z_disable;
+   uint32_t                             early_z_test_and_update_direction;
+#define EARLY_Z_DIRECTION_LT_LE                  0
+#define EARLY_Z_DIRECTION_GT_GE                  1
+   bool                                 double_buffer_in_non_ms_mode;
+   bool                                 multisample_mode_4x;
+   enum V3D41_Internal_BPP              maximum_bpp_of_all_render_targets;
+   uint32_t                             image_height_pixels;
+   uint32_t                             image_width_pixels;
+   uint32_t                             number_of_render_targets;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_COMMON_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                          const struct V3D41_TILE_RENDERING_MODE_CFG_COMMON * restrict values)
+{
+   assert(values->number_of_render_targets >= 1);
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->number_of_render_targets - 1, 4, 7) |
+            __gen_uint(values->sub_id, 0, 3);
+
+   cl[ 2] = __gen_uint(values->image_width_pixels, 0, 15);
+
+   cl[ 3] = __gen_uint(values->image_width_pixels, 0, 15) >> 8;
+
+   cl[ 4] = __gen_uint(values->image_height_pixels, 0, 15);
+
+   cl[ 5] = __gen_uint(values->image_height_pixels, 0, 15) >> 8;
+
+   cl[ 6] = __gen_uint(values->internal_depth_type, 7, 10) |
+            __gen_uint(values->early_z_disable, 6, 6) |
+            __gen_uint(values->early_z_test_and_update_direction, 5, 5) |
+            __gen_uint(values->double_buffer_in_non_ms_mode, 3, 3) |
+            __gen_uint(values->multisample_mode_4x, 2, 2) |
+            __gen_uint(values->maximum_bpp_of_all_render_targets, 0, 1);
+
+   cl[ 7] = __gen_uint(values->pad, 4, 15) |
+            __gen_uint(values->early_depth_stencil_clear, 3, 3) |
+            __gen_uint(values->internal_depth_type, 7, 10) >> 8;
+
+   cl[ 8] = __gen_uint(values->pad, 4, 15) >> 8;
+
+}
+
+#define V3D41_TILE_RENDERING_MODE_CFG_COMMON_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_COMMON_unpack(const uint8_t * restrict cl,
+                                            struct V3D41_TILE_RENDERING_MODE_CFG_COMMON * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->pad = __gen_unpack_uint(cl, 60, 71);
+   values->early_depth_stencil_clear = __gen_unpack_uint(cl, 59, 59);
+   values->internal_depth_type = __gen_unpack_uint(cl, 55, 58);
+   values->early_z_disable = __gen_unpack_uint(cl, 54, 54);
+   values->early_z_test_and_update_direction = __gen_unpack_uint(cl, 53, 53);
+   values->double_buffer_in_non_ms_mode = __gen_unpack_uint(cl, 51, 51);
+   values->multisample_mode_4x = __gen_unpack_uint(cl, 50, 50);
+   values->maximum_bpp_of_all_render_targets = __gen_unpack_uint(cl, 48, 49);
+   values->image_height_pixels = __gen_unpack_uint(cl, 32, 47);
+   values->image_width_pixels = __gen_unpack_uint(cl, 16, 31);
+   values->number_of_render_targets = __gen_unpack_uint(cl, 12, 15) + 1;
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_TILE_RENDERING_MODE_CFG_COLOR_opcode    121
+#define V3D41_TILE_RENDERING_MODE_CFG_COLOR_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      1
+
+struct V3D41_TILE_RENDERING_MODE_CFG_COLOR {
+   uint32_t                             opcode;
+   uint32_t                             pad;
+   enum V3D41_Render_Target_Clamp       render_target_3_clamp;
+   enum V3D41_Internal_Type             render_target_3_internal_type;
+   enum V3D41_Internal_BPP              render_target_3_internal_bpp;
+   enum V3D41_Render_Target_Clamp       render_target_2_clamp;
+   enum V3D41_Internal_Type             render_target_2_internal_type;
+   enum V3D41_Internal_BPP              render_target_2_internal_bpp;
+   enum V3D41_Render_Target_Clamp       render_target_1_clamp;
+   enum V3D41_Internal_Type             render_target_1_internal_type;
+   enum V3D41_Internal_BPP              render_target_1_internal_bpp;
+   enum V3D41_Render_Target_Clamp       render_target_0_clamp;
+   enum V3D41_Internal_Type             render_target_0_internal_type;
+   enum V3D41_Internal_BPP              render_target_0_internal_bpp;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_COLOR_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                         const struct V3D41_TILE_RENDERING_MODE_CFG_COLOR * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->render_target_0_internal_type, 6, 9) |
+            __gen_uint(values->render_target_0_internal_bpp, 4, 5) |
+            __gen_uint(values->sub_id, 0, 3);
+
+   cl[ 2] = __gen_uint(values->render_target_1_internal_type, 6, 9) |
+            __gen_uint(values->render_target_1_internal_bpp, 4, 5) |
+            __gen_uint(values->render_target_0_clamp, 2, 3) |
+            __gen_uint(values->render_target_0_internal_type, 6, 9) >> 8;
+
+   cl[ 3] = __gen_uint(values->render_target_2_internal_type, 6, 9) |
+            __gen_uint(values->render_target_2_internal_bpp, 4, 5) |
+            __gen_uint(values->render_target_1_clamp, 2, 3) |
+            __gen_uint(values->render_target_1_internal_type, 6, 9) >> 8;
+
+   cl[ 4] = __gen_uint(values->render_target_3_internal_type, 6, 9) |
+            __gen_uint(values->render_target_3_internal_bpp, 4, 5) |
+            __gen_uint(values->render_target_2_clamp, 2, 3) |
+            __gen_uint(values->render_target_2_internal_type, 6, 9) >> 8;
+
+   cl[ 5] = __gen_uint(values->pad, 2, 29) |
+            __gen_uint(values->render_target_3_clamp, 0, 1) |
+            __gen_uint(values->render_target_3_internal_type, 6, 9) >> 8;
+
+   cl[ 6] = __gen_uint(values->pad, 2, 29) >> 8;
+
+   cl[ 7] = __gen_uint(values->pad, 2, 29) >> 16;
+
+   cl[ 8] = __gen_uint(values->pad, 2, 29) >> 24;
+
+}
+
+#define V3D41_TILE_RENDERING_MODE_CFG_COLOR_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_COLOR_unpack(const uint8_t * restrict cl,
+                                           struct V3D41_TILE_RENDERING_MODE_CFG_COLOR * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->pad = __gen_unpack_uint(cl, 42, 69);
+   values->render_target_3_clamp = __gen_unpack_uint(cl, 40, 41);
+   values->render_target_3_internal_type = __gen_unpack_uint(cl, 38, 41);
+   values->render_target_3_internal_bpp = __gen_unpack_uint(cl, 36, 37);
+   values->render_target_2_clamp = __gen_unpack_uint(cl, 34, 35);
+   values->render_target_2_internal_type = __gen_unpack_uint(cl, 30, 33);
+   values->render_target_2_internal_bpp = __gen_unpack_uint(cl, 28, 29);
+   values->render_target_1_clamp = __gen_unpack_uint(cl, 26, 27);
+   values->render_target_1_internal_type = __gen_unpack_uint(cl, 22, 25);
+   values->render_target_1_internal_bpp = __gen_unpack_uint(cl, 20, 21);
+   values->render_target_0_clamp = __gen_unpack_uint(cl, 18, 19);
+   values->render_target_0_internal_type = __gen_unpack_uint(cl, 14, 17);
+   values->render_target_0_internal_bpp = __gen_unpack_uint(cl, 12, 13);
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_opcode    121
+#define V3D41_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      2
+
+struct V3D41_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES {
+   uint32_t                             opcode;
+   uint32_t                             unused;
+   float                                z_clear_value;
+   uint32_t                             stencil_clear_value;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                   const struct V3D41_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->sub_id, 0, 3);
+
+   cl[ 2] = __gen_uint(values->stencil_clear_value, 0, 7);
+
+
+   memcpy(&cl[3], &values->z_clear_value, sizeof(values->z_clear_value));
+   cl[ 7] = __gen_uint(values->unused, 0, 15);
+
+   cl[ 8] = __gen_uint(values->unused, 0, 15) >> 8;
+
+}
+
+#define V3D41_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_unpack(const uint8_t * restrict cl,
+                                                     struct V3D41_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->unused = __gen_unpack_uint(cl, 56, 71);
+   values->z_clear_value = __gen_unpack_float(cl, 24, 55);
+   values->stencil_clear_value = __gen_unpack_uint(cl, 16, 23);
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_opcode    121
+#define V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      3
+
+struct V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1 {
+   uint32_t                             opcode;
+   uint32_t                             clear_color_next_24_bits;
+   uint32_t                             clear_color_low_32_bits;
+   uint32_t                             render_target_number;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                      const struct V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1 * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->render_target_number, 4, 7) |
+            __gen_uint(values->sub_id, 0, 3);
+
+
+   memcpy(&cl[2], &values->clear_color_low_32_bits, sizeof(values->clear_color_low_32_bits));
+   cl[ 6] = __gen_uint(values->clear_color_next_24_bits, 0, 23);
+
+   cl[ 7] = __gen_uint(values->clear_color_next_24_bits, 0, 23) >> 8;
+
+   cl[ 8] = __gen_uint(values->clear_color_next_24_bits, 0, 23) >> 16;
+
+}
+
+#define V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_unpack(const uint8_t * restrict cl,
+                                                        struct V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1 * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->clear_color_next_24_bits = __gen_unpack_uint(cl, 48, 71);
+   values->clear_color_low_32_bits = __gen_unpack_uint(cl, 16, 47);
+   values->render_target_number = __gen_unpack_uint(cl, 12, 15);
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_opcode    121
+#define V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      4
+
+struct V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2 {
+   uint32_t                             opcode;
+   uint32_t                             clear_color_mid_high_24_bits;
+   uint32_t                             clear_color_mid_low_32_bits;
+   uint32_t                             render_target_number;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                      const struct V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2 * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->render_target_number, 4, 7) |
+            __gen_uint(values->sub_id, 0, 3);
+
+
+   memcpy(&cl[2], &values->clear_color_mid_low_32_bits, sizeof(values->clear_color_mid_low_32_bits));
+   cl[ 6] = __gen_uint(values->clear_color_mid_high_24_bits, 0, 23);
+
+   cl[ 7] = __gen_uint(values->clear_color_mid_high_24_bits, 0, 23) >> 8;
+
+   cl[ 8] = __gen_uint(values->clear_color_mid_high_24_bits, 0, 23) >> 16;
+
+}
+
+#define V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_unpack(const uint8_t * restrict cl,
+                                                        struct V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2 * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->clear_color_mid_high_24_bits = __gen_unpack_uint(cl, 48, 71);
+   values->clear_color_mid_low_32_bits = __gen_unpack_uint(cl, 16, 47);
+   values->render_target_number = __gen_unpack_uint(cl, 12, 15);
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_opcode    121
+#define V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      5
+
+struct V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3 {
+   uint32_t                             opcode;
+   uint32_t                             pad;
+   uint32_t                             uif_padded_height_in_uif_blocks;
+   uint32_t                             raster_row_stride_or_image_height_in_pixels;
+   uint32_t                             clear_color_high_16_bits;
+   uint32_t                             render_target_number;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                      const struct V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3 * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->render_target_number, 4, 7) |
+            __gen_uint(values->sub_id, 0, 3);
+
+   cl[ 2] = __gen_uint(values->clear_color_high_16_bits, 0, 15);
+
+   cl[ 3] = __gen_uint(values->clear_color_high_16_bits, 0, 15) >> 8;
+
+   cl[ 4] = __gen_uint(values->raster_row_stride_or_image_height_in_pixels, 0, 15);
+
+   cl[ 5] = __gen_uint(values->raster_row_stride_or_image_height_in_pixels, 0, 15) >> 8;
+
+   cl[ 6] = __gen_uint(values->uif_padded_height_in_uif_blocks, 0, 12);
+
+   cl[ 7] = __gen_uint(values->pad, 5, 15) |
+            __gen_uint(values->uif_padded_height_in_uif_blocks, 0, 12) >> 8;
+
+   cl[ 8] = __gen_uint(values->pad, 5, 15) >> 8;
+
+}
+
+#define V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_unpack(const uint8_t * restrict cl,
+                                                        struct V3D41_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3 * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->pad = __gen_unpack_uint(cl, 61, 71);
+   values->uif_padded_height_in_uif_blocks = __gen_unpack_uint(cl, 48, 60);
+   values->raster_row_stride_or_image_height_in_pixels = __gen_unpack_uint(cl, 32, 47);
+   values->clear_color_high_16_bits = __gen_unpack_uint(cl, 16, 31);
+   values->render_target_number = __gen_unpack_uint(cl, 12, 15);
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_TILE_COORDINATES_opcode        124
+#define V3D41_TILE_COORDINATES_header           \
+   .opcode                              =    124
+
+struct V3D41_TILE_COORDINATES {
+   uint32_t                             opcode;
+   uint32_t                             tile_row_number;
+   uint32_t                             tile_column_number;
+};
+
+static inline void
+V3D41_TILE_COORDINATES_pack(__gen_user_data *data, uint8_t * restrict cl,
+                            const struct V3D41_TILE_COORDINATES * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->tile_column_number, 0, 11);
+
+   cl[ 2] = __gen_uint(values->tile_row_number, 4, 15) |
+            __gen_uint(values->tile_column_number, 0, 11) >> 8;
+
+   cl[ 3] = __gen_uint(values->tile_row_number, 4, 15) >> 8;
+
+}
+
+#define V3D41_TILE_COORDINATES_length          4
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TILE_COORDINATES_unpack(const uint8_t * restrict cl,
+                              struct V3D41_TILE_COORDINATES * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->tile_row_number = __gen_unpack_uint(cl, 20, 31);
+   values->tile_column_number = __gen_unpack_uint(cl, 8, 19);
+}
+#endif
+
+
+#define V3D41_MULTICORE_RENDERING_SUPERTILE_CFG_opcode    122
+#define V3D41_MULTICORE_RENDERING_SUPERTILE_CFG_header\
+   .opcode                              =    122
+
+struct V3D41_MULTICORE_RENDERING_SUPERTILE_CFG {
+   uint32_t                             opcode;
+   uint32_t                             number_of_bin_tile_lists;
+   bool                                 supertile_raster_order;
+   bool                                 multicore_enable;
+   uint32_t                             total_frame_height_in_tiles;
+   uint32_t                             total_frame_width_in_tiles;
+   uint32_t                             total_frame_height_in_supertiles;
+   uint32_t                             total_frame_width_in_supertiles;
+   uint32_t                             supertile_height_in_tiles;
+   uint32_t                             supertile_width_in_tiles;
+};
+
+static inline void
+V3D41_MULTICORE_RENDERING_SUPERTILE_CFG_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                             const struct V3D41_MULTICORE_RENDERING_SUPERTILE_CFG * restrict values)
+{
+   assert(values->number_of_bin_tile_lists >= 1);
+   assert(values->supertile_height_in_tiles >= 1);
+   assert(values->supertile_width_in_tiles >= 1);
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->supertile_width_in_tiles - 1, 0, 7);
+
+   cl[ 2] = __gen_uint(values->supertile_height_in_tiles - 1, 0, 7);
+
+   cl[ 3] = __gen_uint(values->total_frame_width_in_supertiles, 0, 7);
+
+   cl[ 4] = __gen_uint(values->total_frame_height_in_supertiles, 0, 7);
+
+   cl[ 5] = __gen_uint(values->total_frame_width_in_tiles, 0, 11);
+
+   cl[ 6] = __gen_uint(values->total_frame_height_in_tiles, 4, 15) |
+            __gen_uint(values->total_frame_width_in_tiles, 0, 11) >> 8;
+
+   cl[ 7] = __gen_uint(values->total_frame_height_in_tiles, 4, 15) >> 8;
+
+   cl[ 8] = __gen_uint(values->number_of_bin_tile_lists - 1, 5, 7) |
+            __gen_uint(values->supertile_raster_order, 4, 4) |
+            __gen_uint(values->multicore_enable, 0, 0);
+
+}
+
+#define V3D41_MULTICORE_RENDERING_SUPERTILE_CFG_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D41_MULTICORE_RENDERING_SUPERTILE_CFG_unpack(const uint8_t * restrict cl,
+                                               struct V3D41_MULTICORE_RENDERING_SUPERTILE_CFG * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->number_of_bin_tile_lists = __gen_unpack_uint(cl, 69, 71) + 1;
+   values->supertile_raster_order = __gen_unpack_uint(cl, 68, 68);
+   values->multicore_enable = __gen_unpack_uint(cl, 64, 64);
+   values->total_frame_height_in_tiles = __gen_unpack_uint(cl, 52, 63);
+   values->total_frame_width_in_tiles = __gen_unpack_uint(cl, 40, 51);
+   values->total_frame_height_in_supertiles = __gen_unpack_uint(cl, 32, 39);
+   values->total_frame_width_in_supertiles = __gen_unpack_uint(cl, 24, 31);
+   values->supertile_height_in_tiles = __gen_unpack_uint(cl, 16, 23) + 1;
+   values->supertile_width_in_tiles = __gen_unpack_uint(cl, 8, 15) + 1;
+}
+#endif
+
+
+#define V3D41_MULTICORE_RENDERING_TILE_LIST_SET_BASE_opcode    123
+#define V3D41_MULTICORE_RENDERING_TILE_LIST_SET_BASE_header\
+   .opcode                              =    123
+
+struct V3D41_MULTICORE_RENDERING_TILE_LIST_SET_BASE {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+   uint32_t                             tile_list_set_number;
+};
+
+static inline void
+V3D41_MULTICORE_RENDERING_TILE_LIST_SET_BASE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                  const struct V3D41_MULTICORE_RENDERING_TILE_LIST_SET_BASE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address) |
+            __gen_uint(values->tile_list_set_number, 0, 3);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D41_MULTICORE_RENDERING_TILE_LIST_SET_BASE_length      5
+#ifdef __gen_unpack_address
+static inline void
+V3D41_MULTICORE_RENDERING_TILE_LIST_SET_BASE_unpack(const uint8_t * restrict cl,
+                                                    struct V3D41_MULTICORE_RENDERING_TILE_LIST_SET_BASE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 14, 39);
+   values->tile_list_set_number = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D41_TILE_COORDINATES_IMPLICIT_opcode    125
+#define V3D41_TILE_COORDINATES_IMPLICIT_header  \
+   .opcode                              =    125
+
+struct V3D41_TILE_COORDINATES_IMPLICIT {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D41_TILE_COORDINATES_IMPLICIT_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                     const struct V3D41_TILE_COORDINATES_IMPLICIT * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D41_TILE_COORDINATES_IMPLICIT_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TILE_COORDINATES_IMPLICIT_unpack(const uint8_t * restrict cl,
+                                       struct V3D41_TILE_COORDINATES_IMPLICIT * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D41_TILE_LIST_INITIAL_BLOCK_SIZE_opcode    126
+#define V3D41_TILE_LIST_INITIAL_BLOCK_SIZE_header\
+   .opcode                              =    126
+
+struct V3D41_TILE_LIST_INITIAL_BLOCK_SIZE {
+   uint32_t                             opcode;
+   bool                                 use_auto_chained_tile_lists;
+   uint32_t                             size_of_first_block_in_chained_tile_lists;
+#define TILE_ALLOCATION_BLOCK_SIZE_64B           0
+#define TILE_ALLOCATION_BLOCK_SIZE_128B          1
+#define TILE_ALLOCATION_BLOCK_SIZE_256B          2
+};
+
+static inline void
+V3D41_TILE_LIST_INITIAL_BLOCK_SIZE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D41_TILE_LIST_INITIAL_BLOCK_SIZE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->use_auto_chained_tile_lists, 2, 2) |
+            __gen_uint(values->size_of_first_block_in_chained_tile_lists, 0, 1);
+
+}
+
+#define V3D41_TILE_LIST_INITIAL_BLOCK_SIZE_length      2
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TILE_LIST_INITIAL_BLOCK_SIZE_unpack(const uint8_t * restrict cl,
+                                          struct V3D41_TILE_LIST_INITIAL_BLOCK_SIZE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->use_auto_chained_tile_lists = __gen_unpack_uint(cl, 10, 10);
+   values->size_of_first_block_in_chained_tile_lists = __gen_unpack_uint(cl, 8, 9);
+}
+#endif
+
+
+#define V3D41_GL_SHADER_STATE_RECORD_header     \
+
+
+struct V3D41_GL_SHADER_STATE_RECORD {
+   bool                                 point_size_in_shaded_vertex_data;
+   bool                                 enable_clipping;
+   bool                                 vertex_id_read_by_coordinate_shader;
+   bool                                 instance_id_read_by_coordinate_shader;
+   bool                                 base_instance_id_read_by_coordinate_shader;
+   bool                                 vertex_id_read_by_vertex_shader;
+   bool                                 instance_id_read_by_vertex_shader;
+   bool                                 base_instance_id_read_by_vertex_shader;
+   bool                                 fragment_shader_does_z_writes;
+   bool                                 turn_off_early_z_test;
+   bool                                 coordinate_shader_has_separate_input_and_output_vpm_blocks;
+   bool                                 vertex_shader_has_separate_input_and_output_vpm_blocks;
+   bool                                 fragment_shader_uses_real_pixel_centre_w_in_addition_to_centroid_w2;
+   bool                                 enable_sample_rate_shading;
+   bool                                 any_shader_reads_hardware_written_primitive_id;
+   bool                                 insert_primitive_id_as_first_varying_to_fragment_shader;
+   bool                                 turn_off_scoreboard;
+   bool                                 do_scoreboard_wait_on_first_thread_switch;
+   bool                                 disable_implicit_point_line_varyings;
+   bool                                 no_prim_pack;
+   uint32_t                             number_of_varyings_in_fragment_shader;
+   uint32_t                             coordinate_shader_output_vpm_segment_size;
+   uint32_t                             min_coord_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size;
+   uint32_t                             coordinate_shader_input_vpm_segment_size;
+   uint32_t                             min_coord_shader_input_segments_required_in_play;
+   uint32_t                             vertex_shader_output_vpm_segment_size;
+   uint32_t                             min_vertex_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size;
+   uint32_t                             vertex_shader_input_vpm_segment_size;
+   uint32_t                             min_vertex_shader_input_segments_required_in_play;
+   __gen_address_type                   address_of_default_attribute_values;
+   __gen_address_type                   fragment_shader_code_address;
+   bool                                 fragment_shader_4_way_threadable;
+   bool                                 fragment_shader_start_in_final_thread_section;
+   bool                                 fragment_shader_propagate_nans;
+   __gen_address_type                   fragment_shader_uniforms_address;
+   __gen_address_type                   vertex_shader_code_address;
+   bool                                 vertex_shader_4_way_threadable;
+   bool                                 vertex_shader_start_in_final_thread_section;
+   bool                                 vertex_shader_propagate_nans;
+   __gen_address_type                   vertex_shader_uniforms_address;
+   __gen_address_type                   coordinate_shader_code_address;
+   bool                                 coordinate_shader_4_way_threadable;
+   bool                                 coordinate_shader_start_in_final_thread_section;
+   bool                                 coordinate_shader_propagate_nans;
+   __gen_address_type                   coordinate_shader_uniforms_address;
+};
+
+static inline void
+V3D41_GL_SHADER_STATE_RECORD_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                  const struct V3D41_GL_SHADER_STATE_RECORD * restrict values)
+{
+   assert(values->min_coord_shader_input_segments_required_in_play >= 1);
+   assert(values->min_vertex_shader_input_segments_required_in_play >= 1);
+   cl[ 0] = __gen_uint(values->point_size_in_shaded_vertex_data, 0, 0) |
+            __gen_uint(values->enable_clipping, 1, 1) |
+            __gen_uint(values->vertex_id_read_by_coordinate_shader, 2, 2) |
+            __gen_uint(values->instance_id_read_by_coordinate_shader, 3, 3) |
+            __gen_uint(values->base_instance_id_read_by_coordinate_shader, 4, 4) |
+            __gen_uint(values->vertex_id_read_by_vertex_shader, 5, 5) |
+            __gen_uint(values->instance_id_read_by_vertex_shader, 6, 6) |
+            __gen_uint(values->base_instance_id_read_by_vertex_shader, 7, 7);
+
+   cl[ 1] = __gen_uint(values->fragment_shader_does_z_writes, 0, 0) |
+            __gen_uint(values->turn_off_early_z_test, 1, 1) |
+            __gen_uint(values->coordinate_shader_has_separate_input_and_output_vpm_blocks, 2, 2) |
+            __gen_uint(values->vertex_shader_has_separate_input_and_output_vpm_blocks, 3, 3) |
+            __gen_uint(values->fragment_shader_uses_real_pixel_centre_w_in_addition_to_centroid_w2, 4, 4) |
+            __gen_uint(values->enable_sample_rate_shading, 5, 5) |
+            __gen_uint(values->any_shader_reads_hardware_written_primitive_id, 6, 6) |
+            __gen_uint(values->insert_primitive_id_as_first_varying_to_fragment_shader, 7, 7);
+
+   cl[ 2] = __gen_uint(values->turn_off_scoreboard, 0, 0) |
+            __gen_uint(values->do_scoreboard_wait_on_first_thread_switch, 1, 1) |
+            __gen_uint(values->disable_implicit_point_line_varyings, 2, 2) |
+            __gen_uint(values->no_prim_pack, 3, 3);
+
+   cl[ 3] = __gen_uint(values->number_of_varyings_in_fragment_shader, 0, 7);
+
+   cl[ 4] = __gen_uint(values->coordinate_shader_output_vpm_segment_size, 0, 3) |
+            __gen_uint(values->min_coord_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size, 4, 7);
+
+   cl[ 5] = __gen_uint(values->coordinate_shader_input_vpm_segment_size, 0, 3) |
+            __gen_uint(values->min_coord_shader_input_segments_required_in_play - 1, 4, 7);
+
+   cl[ 6] = __gen_uint(values->vertex_shader_output_vpm_segment_size, 0, 3) |
+            __gen_uint(values->min_vertex_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size, 4, 7);
+
+   cl[ 7] = __gen_uint(values->vertex_shader_input_vpm_segment_size, 0, 3) |
+            __gen_uint(values->min_vertex_shader_input_segments_required_in_play - 1, 4, 7);
+
+   __gen_emit_reloc(data, &values->address_of_default_attribute_values);
+   cl[ 8] = __gen_address_offset(&values->address_of_default_attribute_values);
+
+   cl[ 9] = __gen_address_offset(&values->address_of_default_attribute_values) >> 8;
+
+   cl[10] = __gen_address_offset(&values->address_of_default_attribute_values) >> 16;
+
+   cl[11] = __gen_address_offset(&values->address_of_default_attribute_values) >> 24;
+
+   __gen_emit_reloc(data, &values->fragment_shader_code_address);
+   cl[12] = __gen_address_offset(&values->fragment_shader_code_address) |
+            __gen_uint(values->fragment_shader_4_way_threadable, 0, 0) |
+            __gen_uint(values->fragment_shader_start_in_final_thread_section, 1, 1) |
+            __gen_uint(values->fragment_shader_propagate_nans, 2, 2);
+
+   cl[13] = __gen_address_offset(&values->fragment_shader_code_address) >> 8;
+
+   cl[14] = __gen_address_offset(&values->fragment_shader_code_address) >> 16;
+
+   cl[15] = __gen_address_offset(&values->fragment_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->fragment_shader_uniforms_address);
+   cl[16] = __gen_address_offset(&values->fragment_shader_uniforms_address);
+
+   cl[17] = __gen_address_offset(&values->fragment_shader_uniforms_address) >> 8;
+
+   cl[18] = __gen_address_offset(&values->fragment_shader_uniforms_address) >> 16;
+
+   cl[19] = __gen_address_offset(&values->fragment_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->vertex_shader_code_address);
+   cl[20] = __gen_address_offset(&values->vertex_shader_code_address) |
+            __gen_uint(values->vertex_shader_4_way_threadable, 0, 0) |
+            __gen_uint(values->vertex_shader_start_in_final_thread_section, 1, 1) |
+            __gen_uint(values->vertex_shader_propagate_nans, 2, 2);
+
+   cl[21] = __gen_address_offset(&values->vertex_shader_code_address) >> 8;
+
+   cl[22] = __gen_address_offset(&values->vertex_shader_code_address) >> 16;
+
+   cl[23] = __gen_address_offset(&values->vertex_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->vertex_shader_uniforms_address);
+   cl[24] = __gen_address_offset(&values->vertex_shader_uniforms_address);
+
+   cl[25] = __gen_address_offset(&values->vertex_shader_uniforms_address) >> 8;
+
+   cl[26] = __gen_address_offset(&values->vertex_shader_uniforms_address) >> 16;
+
+   cl[27] = __gen_address_offset(&values->vertex_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->coordinate_shader_code_address);
+   cl[28] = __gen_address_offset(&values->coordinate_shader_code_address) |
+            __gen_uint(values->coordinate_shader_4_way_threadable, 0, 0) |
+            __gen_uint(values->coordinate_shader_start_in_final_thread_section, 1, 1) |
+            __gen_uint(values->coordinate_shader_propagate_nans, 2, 2);
+
+   cl[29] = __gen_address_offset(&values->coordinate_shader_code_address) >> 8;
+
+   cl[30] = __gen_address_offset(&values->coordinate_shader_code_address) >> 16;
+
+   cl[31] = __gen_address_offset(&values->coordinate_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->coordinate_shader_uniforms_address);
+   cl[32] = __gen_address_offset(&values->coordinate_shader_uniforms_address);
+
+   cl[33] = __gen_address_offset(&values->coordinate_shader_uniforms_address) >> 8;
+
+   cl[34] = __gen_address_offset(&values->coordinate_shader_uniforms_address) >> 16;
+
+   cl[35] = __gen_address_offset(&values->coordinate_shader_uniforms_address) >> 24;
+
+}
+
+#define V3D41_GL_SHADER_STATE_RECORD_length     36
+#ifdef __gen_unpack_address
+static inline void
+V3D41_GL_SHADER_STATE_RECORD_unpack(const uint8_t * restrict cl,
+                                    struct V3D41_GL_SHADER_STATE_RECORD * restrict values)
+{
+   values->point_size_in_shaded_vertex_data = __gen_unpack_uint(cl, 0, 0);
+   values->enable_clipping = __gen_unpack_uint(cl, 1, 1);
+   values->vertex_id_read_by_coordinate_shader = __gen_unpack_uint(cl, 2, 2);
+   values->instance_id_read_by_coordinate_shader = __gen_unpack_uint(cl, 3, 3);
+   values->base_instance_id_read_by_coordinate_shader = __gen_unpack_uint(cl, 4, 4);
+   values->vertex_id_read_by_vertex_shader = __gen_unpack_uint(cl, 5, 5);
+   values->instance_id_read_by_vertex_shader = __gen_unpack_uint(cl, 6, 6);
+   values->base_instance_id_read_by_vertex_shader = __gen_unpack_uint(cl, 7, 7);
+   values->fragment_shader_does_z_writes = __gen_unpack_uint(cl, 8, 8);
+   values->turn_off_early_z_test = __gen_unpack_uint(cl, 9, 9);
+   values->coordinate_shader_has_separate_input_and_output_vpm_blocks = __gen_unpack_uint(cl, 10, 10);
+   values->vertex_shader_has_separate_input_and_output_vpm_blocks = __gen_unpack_uint(cl, 11, 11);
+   values->fragment_shader_uses_real_pixel_centre_w_in_addition_to_centroid_w2 = __gen_unpack_uint(cl, 12, 12);
+   values->enable_sample_rate_shading = __gen_unpack_uint(cl, 13, 13);
+   values->any_shader_reads_hardware_written_primitive_id = __gen_unpack_uint(cl, 14, 14);
+   values->insert_primitive_id_as_first_varying_to_fragment_shader = __gen_unpack_uint(cl, 15, 15);
+   values->turn_off_scoreboard = __gen_unpack_uint(cl, 16, 16);
+   values->do_scoreboard_wait_on_first_thread_switch = __gen_unpack_uint(cl, 17, 17);
+   values->disable_implicit_point_line_varyings = __gen_unpack_uint(cl, 18, 18);
+   values->no_prim_pack = __gen_unpack_uint(cl, 19, 19);
+   values->number_of_varyings_in_fragment_shader = __gen_unpack_uint(cl, 24, 31);
+   values->coordinate_shader_output_vpm_segment_size = __gen_unpack_uint(cl, 32, 35);
+   values->min_coord_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size = __gen_unpack_uint(cl, 36, 39);
+   values->coordinate_shader_input_vpm_segment_size = __gen_unpack_uint(cl, 40, 43);
+   values->min_coord_shader_input_segments_required_in_play = __gen_unpack_uint(cl, 44, 47) + 1;
+   values->vertex_shader_output_vpm_segment_size = __gen_unpack_uint(cl, 48, 51);
+   values->min_vertex_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size = __gen_unpack_uint(cl, 52, 55);
+   values->vertex_shader_input_vpm_segment_size = __gen_unpack_uint(cl, 56, 59);
+   values->min_vertex_shader_input_segments_required_in_play = __gen_unpack_uint(cl, 60, 63) + 1;
+   values->address_of_default_attribute_values = __gen_unpack_address(cl, 64, 95);
+   values->fragment_shader_code_address = __gen_unpack_address(cl, 99, 127);
+   values->fragment_shader_4_way_threadable = __gen_unpack_uint(cl, 96, 96);
+   values->fragment_shader_start_in_final_thread_section = __gen_unpack_uint(cl, 97, 97);
+   values->fragment_shader_propagate_nans = __gen_unpack_uint(cl, 98, 98);
+   values->fragment_shader_uniforms_address = __gen_unpack_address(cl, 128, 159);
+   values->vertex_shader_code_address = __gen_unpack_address(cl, 163, 191);
+   values->vertex_shader_4_way_threadable = __gen_unpack_uint(cl, 160, 160);
+   values->vertex_shader_start_in_final_thread_section = __gen_unpack_uint(cl, 161, 161);
+   values->vertex_shader_propagate_nans = __gen_unpack_uint(cl, 162, 162);
+   values->vertex_shader_uniforms_address = __gen_unpack_address(cl, 192, 223);
+   values->coordinate_shader_code_address = __gen_unpack_address(cl, 227, 255);
+   values->coordinate_shader_4_way_threadable = __gen_unpack_uint(cl, 224, 224);
+   values->coordinate_shader_start_in_final_thread_section = __gen_unpack_uint(cl, 225, 225);
+   values->coordinate_shader_propagate_nans = __gen_unpack_uint(cl, 226, 226);
+   values->coordinate_shader_uniforms_address = __gen_unpack_address(cl, 256, 287);
+}
+#endif
+
+
+#define V3D41_GEOMETRY_SHADER_STATE_RECORD_header\
+
+
+struct V3D41_GEOMETRY_SHADER_STATE_RECORD {
+   __gen_address_type                   geometry_bin_mode_shader_code_address;
+   bool                                 _4_way_threadable;
+   bool                                 start_in_final_thread_section;
+   bool                                 propagate_nans;
+   __gen_address_type                   geometry_bin_mode_shader_uniforms_address;
+   __gen_address_type                   geometry_render_mode_shader_code_address;
+   __gen_address_type                   geometry_render_mode_shader_uniforms_address;
+};
+
+static inline void
+V3D41_GEOMETRY_SHADER_STATE_RECORD_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D41_GEOMETRY_SHADER_STATE_RECORD * restrict values)
+{
+   __gen_emit_reloc(data, &values->geometry_bin_mode_shader_code_address);
+   cl[ 0] = __gen_address_offset(&values->geometry_bin_mode_shader_code_address) |
+            __gen_uint(values->_4_way_threadable, 0, 0) |
+            __gen_uint(values->start_in_final_thread_section, 1, 1) |
+            __gen_uint(values->propagate_nans, 2, 2);
+
+   cl[ 1] = __gen_address_offset(&values->geometry_bin_mode_shader_code_address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->geometry_bin_mode_shader_code_address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->geometry_bin_mode_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->geometry_bin_mode_shader_uniforms_address);
+   cl[ 4] = __gen_address_offset(&values->geometry_bin_mode_shader_uniforms_address);
+
+   cl[ 5] = __gen_address_offset(&values->geometry_bin_mode_shader_uniforms_address) >> 8;
+
+   cl[ 6] = __gen_address_offset(&values->geometry_bin_mode_shader_uniforms_address) >> 16;
+
+   cl[ 7] = __gen_address_offset(&values->geometry_bin_mode_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->geometry_render_mode_shader_code_address);
+   cl[ 8] = __gen_address_offset(&values->geometry_render_mode_shader_code_address);
+
+   cl[ 9] = __gen_address_offset(&values->geometry_render_mode_shader_code_address) >> 8;
+
+   cl[10] = __gen_address_offset(&values->geometry_render_mode_shader_code_address) >> 16;
+
+   cl[11] = __gen_address_offset(&values->geometry_render_mode_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->geometry_render_mode_shader_uniforms_address);
+   cl[12] = __gen_address_offset(&values->geometry_render_mode_shader_uniforms_address);
+
+   cl[13] = __gen_address_offset(&values->geometry_render_mode_shader_uniforms_address) >> 8;
+
+   cl[14] = __gen_address_offset(&values->geometry_render_mode_shader_uniforms_address) >> 16;
+
+   cl[15] = __gen_address_offset(&values->geometry_render_mode_shader_uniforms_address) >> 24;
+
+}
+
+#define V3D41_GEOMETRY_SHADER_STATE_RECORD_length     16
+#ifdef __gen_unpack_address
+static inline void
+V3D41_GEOMETRY_SHADER_STATE_RECORD_unpack(const uint8_t * restrict cl,
+                                          struct V3D41_GEOMETRY_SHADER_STATE_RECORD * restrict values)
+{
+   values->geometry_bin_mode_shader_code_address = __gen_unpack_address(cl, 0, 31);
+   values->_4_way_threadable = __gen_unpack_uint(cl, 0, 0);
+   values->start_in_final_thread_section = __gen_unpack_uint(cl, 1, 1);
+   values->propagate_nans = __gen_unpack_uint(cl, 2, 2);
+   values->geometry_bin_mode_shader_uniforms_address = __gen_unpack_address(cl, 32, 63);
+   values->geometry_render_mode_shader_code_address = __gen_unpack_address(cl, 64, 95);
+   values->geometry_render_mode_shader_uniforms_address = __gen_unpack_address(cl, 96, 127);
+}
+#endif
+
+
+#define V3D41_TESSELLATION_SHADER_STATE_RECORD_header\
+
+
+struct V3D41_TESSELLATION_SHADER_STATE_RECORD {
+   __gen_address_type                   tessellation_bin_mode_control_shader_code_address;
+   __gen_address_type                   tessellation_bin_mode_control_shader_uniforms_address;
+   __gen_address_type                   tessellation_render_mode_control_shader_code_address;
+   __gen_address_type                   tessellation_render_mode_control_shader_uniforms_address;
+   __gen_address_type                   tessellation_bin_mode_evaluation_shader_code_address;
+   __gen_address_type                   tessellation_bin_mode_evaluation_shader_uniforms_address;
+   __gen_address_type                   tessellation_render_mode_evaluation_shader_code_address;
+   __gen_address_type                   tessellation_render_mode_evaluation_shader_uniforms_address;
+};
+
+static inline void
+V3D41_TESSELLATION_SHADER_STATE_RECORD_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                            const struct V3D41_TESSELLATION_SHADER_STATE_RECORD * restrict values)
+{
+   __gen_emit_reloc(data, &values->tessellation_bin_mode_control_shader_code_address);
+   cl[ 0] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_code_address);
+
+   cl[ 1] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_code_address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_code_address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_bin_mode_control_shader_uniforms_address);
+   cl[ 4] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_uniforms_address);
+
+   cl[ 5] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_uniforms_address) >> 8;
+
+   cl[ 6] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_uniforms_address) >> 16;
+
+   cl[ 7] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_render_mode_control_shader_code_address);
+   cl[ 8] = __gen_address_offset(&values->tessellation_render_mode_control_shader_code_address);
+
+   cl[ 9] = __gen_address_offset(&values->tessellation_render_mode_control_shader_code_address) >> 8;
+
+   cl[10] = __gen_address_offset(&values->tessellation_render_mode_control_shader_code_address) >> 16;
+
+   cl[11] = __gen_address_offset(&values->tessellation_render_mode_control_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_render_mode_control_shader_uniforms_address);
+   cl[12] = __gen_address_offset(&values->tessellation_render_mode_control_shader_uniforms_address);
+
+   cl[13] = __gen_address_offset(&values->tessellation_render_mode_control_shader_uniforms_address) >> 8;
+
+   cl[14] = __gen_address_offset(&values->tessellation_render_mode_control_shader_uniforms_address) >> 16;
+
+   cl[15] = __gen_address_offset(&values->tessellation_render_mode_control_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_bin_mode_evaluation_shader_code_address);
+   cl[16] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_code_address);
+
+   cl[17] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_code_address) >> 8;
+
+   cl[18] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_code_address) >> 16;
+
+   cl[19] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_bin_mode_evaluation_shader_uniforms_address);
+   cl[20] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_uniforms_address);
+
+   cl[21] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_uniforms_address) >> 8;
+
+   cl[22] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_uniforms_address) >> 16;
+
+   cl[23] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_render_mode_evaluation_shader_code_address);
+   cl[24] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_code_address);
+
+   cl[25] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_code_address) >> 8;
+
+   cl[26] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_code_address) >> 16;
+
+   cl[27] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_render_mode_evaluation_shader_uniforms_address);
+   cl[28] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_uniforms_address);
+
+   cl[29] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_uniforms_address) >> 8;
+
+   cl[30] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_uniforms_address) >> 16;
+
+   cl[31] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_uniforms_address) >> 24;
+
+}
+
+#define V3D41_TESSELLATION_SHADER_STATE_RECORD_length     32
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TESSELLATION_SHADER_STATE_RECORD_unpack(const uint8_t * restrict cl,
+                                              struct V3D41_TESSELLATION_SHADER_STATE_RECORD * restrict values)
+{
+   values->tessellation_bin_mode_control_shader_code_address = __gen_unpack_address(cl, 0, 31);
+   values->tessellation_bin_mode_control_shader_uniforms_address = __gen_unpack_address(cl, 32, 63);
+   values->tessellation_render_mode_control_shader_code_address = __gen_unpack_address(cl, 64, 95);
+   values->tessellation_render_mode_control_shader_uniforms_address = __gen_unpack_address(cl, 96, 127);
+   values->tessellation_bin_mode_evaluation_shader_code_address = __gen_unpack_address(cl, 128, 159);
+   values->tessellation_bin_mode_evaluation_shader_uniforms_address = __gen_unpack_address(cl, 160, 191);
+   values->tessellation_render_mode_evaluation_shader_code_address = __gen_unpack_address(cl, 192, 223);
+   values->tessellation_render_mode_evaluation_shader_uniforms_address = __gen_unpack_address(cl, 224, 255);
+}
+#endif
+
+
+#define V3D41_GL_SHADER_STATE_ATTRIBUTE_RECORD_header\
+
+
+struct V3D41_GL_SHADER_STATE_ATTRIBUTE_RECORD {
+   __gen_address_type                   address;
+   uint32_t                             vec_size;
+   uint32_t                             type;
+#define ATTRIBUTE_HALF_FLOAT                     1
+#define ATTRIBUTE_FLOAT                          2
+#define ATTRIBUTE_FIXED                          3
+#define ATTRIBUTE_BYTE                           4
+#define ATTRIBUTE_SHORT                          5
+#define ATTRIBUTE_INT                            6
+#define ATTRIBUTE_INT2_10_10_10                  7
+   bool                                 signed_int_type;
+   bool                                 normalized_int_type;
+   bool                                 read_as_int_uint;
+   uint32_t                             number_of_values_read_by_coordinate_shader;
+   uint32_t                             number_of_values_read_by_vertex_shader;
+   uint32_t                             instance_divisor;
+   uint32_t                             stride;
+   uint32_t                             maximum_index;
+};
+
+static inline void
+V3D41_GL_SHADER_STATE_ATTRIBUTE_RECORD_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                            const struct V3D41_GL_SHADER_STATE_ATTRIBUTE_RECORD * restrict values)
+{
+   __gen_emit_reloc(data, &values->address);
+   cl[ 0] = __gen_address_offset(&values->address);
+
+   cl[ 1] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 24;
+
+   cl[ 4] = __gen_uint(values->vec_size, 0, 1) |
+            __gen_uint(values->type, 2, 4) |
+            __gen_uint(values->signed_int_type, 5, 5) |
+            __gen_uint(values->normalized_int_type, 6, 6) |
+            __gen_uint(values->read_as_int_uint, 7, 7);
+
+   cl[ 5] = __gen_uint(values->number_of_values_read_by_coordinate_shader, 0, 3) |
+            __gen_uint(values->number_of_values_read_by_vertex_shader, 4, 7);
+
+   cl[ 6] = __gen_uint(values->instance_divisor, 0, 15);
+
+   cl[ 7] = __gen_uint(values->instance_divisor, 0, 15) >> 8;
+
+
+   memcpy(&cl[8], &values->stride, sizeof(values->stride));
+
+   memcpy(&cl[12], &values->maximum_index, sizeof(values->maximum_index));
+}
+
+#define V3D41_GL_SHADER_STATE_ATTRIBUTE_RECORD_length     16
+#ifdef __gen_unpack_address
+static inline void
+V3D41_GL_SHADER_STATE_ATTRIBUTE_RECORD_unpack(const uint8_t * restrict cl,
+                                              struct V3D41_GL_SHADER_STATE_ATTRIBUTE_RECORD * restrict values)
+{
+   values->address = __gen_unpack_address(cl, 0, 31);
+   values->vec_size = __gen_unpack_uint(cl, 32, 33);
+   values->type = __gen_unpack_uint(cl, 34, 36);
+   values->signed_int_type = __gen_unpack_uint(cl, 37, 37);
+   values->normalized_int_type = __gen_unpack_uint(cl, 38, 38);
+   values->read_as_int_uint = __gen_unpack_uint(cl, 39, 39);
+   values->number_of_values_read_by_coordinate_shader = __gen_unpack_uint(cl, 40, 43);
+   values->number_of_values_read_by_vertex_shader = __gen_unpack_uint(cl, 44, 47);
+   values->instance_divisor = __gen_unpack_uint(cl, 48, 63);
+   values->stride = __gen_unpack_uint(cl, 64, 95);
+   values->maximum_index = __gen_unpack_uint(cl, 96, 127);
+}
+#endif
+
+
+#define V3D41_VPM_GENERIC_BLOCK_WRITE_SETUP_header\
+   .id                                  =      0,  \
+   .id0                                 =      0
+
+struct V3D41_VPM_GENERIC_BLOCK_WRITE_SETUP {
+   uint32_t                             id;
+   uint32_t                             id0;
+   bool                                 horiz;
+   bool                                 laned;
+   bool                                 segs;
+   int32_t                              stride;
+   uint32_t                             size;
+#define VPM_SETUP_SIZE_8_BIT                     0
+#define VPM_SETUP_SIZE_16_BIT                    1
+#define VPM_SETUP_SIZE_32_BIT                    2
+   uint32_t                             addr;
+};
+
+static inline void
+V3D41_VPM_GENERIC_BLOCK_WRITE_SETUP_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                         const struct V3D41_VPM_GENERIC_BLOCK_WRITE_SETUP * restrict values)
+{
+   cl[ 0] = __gen_uint(values->addr, 0, 12);
+
+   cl[ 1] = __gen_sint(values->stride, 7, 13) |
+            __gen_uint(values->size, 5, 6) |
+            __gen_uint(values->addr, 0, 12) >> 8;
+
+   cl[ 2] = __gen_uint(values->laned, 7, 7) |
+            __gen_uint(values->segs, 6, 6) |
+            __gen_sint(values->stride, 7, 13) >> 8;
+
+   cl[ 3] = __gen_uint(values->id, 6, 7) |
+            __gen_uint(values->id0, 3, 5) |
+            __gen_uint(values->horiz, 0, 0);
+
+}
+
+#define V3D41_VPM_GENERIC_BLOCK_WRITE_SETUP_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D41_VPM_GENERIC_BLOCK_WRITE_SETUP_unpack(const uint8_t * restrict cl,
+                                           struct V3D41_VPM_GENERIC_BLOCK_WRITE_SETUP * restrict values)
+{
+   values->id = __gen_unpack_uint(cl, 30, 31);
+   values->id0 = __gen_unpack_uint(cl, 27, 29);
+   values->horiz = __gen_unpack_uint(cl, 24, 24);
+   values->laned = __gen_unpack_uint(cl, 23, 23);
+   values->segs = __gen_unpack_uint(cl, 22, 22);
+   values->stride = __gen_unpack_sint(cl, 15, 21);
+   values->size = __gen_unpack_uint(cl, 13, 14);
+   values->addr = __gen_unpack_uint(cl, 0, 12);
+}
+#endif
+
+
+#define V3D41_VPM_GENERIC_BLOCK_READ_SETUP_header\
+   .id                                  =      1
+
+struct V3D41_VPM_GENERIC_BLOCK_READ_SETUP {
+   uint32_t                             id;
+   bool                                 horiz;
+   bool                                 laned;
+   bool                                 segs;
+   uint32_t                             num;
+   int32_t                              stride;
+   uint32_t                             size;
+#define VPM_SETUP_SIZE_8_BIT                     0
+#define VPM_SETUP_SIZE_16_BIT                    1
+#define VPM_SETUP_SIZE_32_BIT                    2
+   uint32_t                             addr;
+};
+
+static inline void
+V3D41_VPM_GENERIC_BLOCK_READ_SETUP_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D41_VPM_GENERIC_BLOCK_READ_SETUP * restrict values)
+{
+   cl[ 0] = __gen_uint(values->addr, 0, 12);
+
+   cl[ 1] = __gen_sint(values->stride, 7, 13) |
+            __gen_uint(values->size, 5, 6) |
+            __gen_uint(values->addr, 0, 12) >> 8;
+
+   cl[ 2] = __gen_uint(values->num, 6, 10) |
+            __gen_sint(values->stride, 7, 13) >> 8;
+
+   cl[ 3] = __gen_uint(values->id, 6, 7) |
+            __gen_uint(values->horiz, 5, 5) |
+            __gen_uint(values->laned, 4, 4) |
+            __gen_uint(values->segs, 3, 3) |
+            __gen_uint(values->num, 6, 10) >> 8;
+
+}
+
+#define V3D41_VPM_GENERIC_BLOCK_READ_SETUP_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D41_VPM_GENERIC_BLOCK_READ_SETUP_unpack(const uint8_t * restrict cl,
+                                          struct V3D41_VPM_GENERIC_BLOCK_READ_SETUP * restrict values)
+{
+   values->id = __gen_unpack_uint(cl, 30, 31);
+   values->horiz = __gen_unpack_uint(cl, 29, 29);
+   values->laned = __gen_unpack_uint(cl, 28, 28);
+   values->segs = __gen_unpack_uint(cl, 27, 27);
+   values->num = __gen_unpack_uint(cl, 22, 26);
+   values->stride = __gen_unpack_sint(cl, 15, 21);
+   values->size = __gen_unpack_uint(cl, 13, 14);
+   values->addr = __gen_unpack_uint(cl, 0, 12);
+}
+#endif
+
+
+#define V3D41_TMU_CONFIG_PARAMETER_0_header     \
+
+
+struct V3D41_TMU_CONFIG_PARAMETER_0 {
+   __gen_address_type                   texture_state_address;
+   uint32_t                             return_words_of_texture_data;
+};
+
+static inline void
+V3D41_TMU_CONFIG_PARAMETER_0_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                  const struct V3D41_TMU_CONFIG_PARAMETER_0 * restrict values)
+{
+   __gen_emit_reloc(data, &values->texture_state_address);
+   cl[ 0] = __gen_address_offset(&values->texture_state_address) |
+            __gen_uint(values->return_words_of_texture_data, 0, 3);
+
+   cl[ 1] = __gen_address_offset(&values->texture_state_address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->texture_state_address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->texture_state_address) >> 24;
+
+}
+
+#define V3D41_TMU_CONFIG_PARAMETER_0_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TMU_CONFIG_PARAMETER_0_unpack(const uint8_t * restrict cl,
+                                    struct V3D41_TMU_CONFIG_PARAMETER_0 * restrict values)
+{
+   values->texture_state_address = __gen_unpack_address(cl, 0, 31);
+   values->return_words_of_texture_data = __gen_unpack_uint(cl, 0, 3);
+}
+#endif
+
+
+#define V3D41_TMU_CONFIG_PARAMETER_1_header     \
+
+
+struct V3D41_TMU_CONFIG_PARAMETER_1 {
+   __gen_address_type                   sampler_state_address;
+   bool                                 per_pixel_mask_enable;
+   bool                                 unnormalized_coordinates;
+   bool                                 output_type_32_bit;
+};
+
+static inline void
+V3D41_TMU_CONFIG_PARAMETER_1_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                  const struct V3D41_TMU_CONFIG_PARAMETER_1 * restrict values)
+{
+   __gen_emit_reloc(data, &values->sampler_state_address);
+   cl[ 0] = __gen_address_offset(&values->sampler_state_address) |
+            __gen_uint(values->per_pixel_mask_enable, 2, 2) |
+            __gen_uint(values->unnormalized_coordinates, 1, 1) |
+            __gen_uint(values->output_type_32_bit, 0, 0);
+
+   cl[ 1] = __gen_address_offset(&values->sampler_state_address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->sampler_state_address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->sampler_state_address) >> 24;
+
+}
+
+#define V3D41_TMU_CONFIG_PARAMETER_1_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TMU_CONFIG_PARAMETER_1_unpack(const uint8_t * restrict cl,
+                                    struct V3D41_TMU_CONFIG_PARAMETER_1 * restrict values)
+{
+   values->sampler_state_address = __gen_unpack_address(cl, 0, 31);
+   values->per_pixel_mask_enable = __gen_unpack_uint(cl, 2, 2);
+   values->unnormalized_coordinates = __gen_unpack_uint(cl, 1, 1);
+   values->output_type_32_bit = __gen_unpack_uint(cl, 0, 0);
+}
+#endif
+
+
+#define V3D41_TMU_CONFIG_PARAMETER_2_header     \
+
+
+struct V3D41_TMU_CONFIG_PARAMETER_2 {
+   uint32_t                             pad;
+   enum V3D41_TMU_Op                    op;
+   int32_t                              offset_r;
+   int32_t                              offset_t;
+   int32_t                              offset_s;
+   bool                                 gather_mode;
+   uint32_t                             gather_component;
+   bool                                 coefficient_mode;
+   uint32_t                             sample_number;
+   bool                                 disable_autolod;
+   bool                                 offset_format_8;
+};
+
+static inline void
+V3D41_TMU_CONFIG_PARAMETER_2_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                  const struct V3D41_TMU_CONFIG_PARAMETER_2 * restrict values)
+{
+   cl[ 0] = __gen_uint(values->gather_mode, 7, 7) |
+            __gen_uint(values->gather_component, 5, 6) |
+            __gen_uint(values->coefficient_mode, 4, 4) |
+            __gen_uint(values->sample_number, 2, 3) |
+            __gen_uint(values->disable_autolod, 1, 1) |
+            __gen_uint(values->offset_format_8, 0, 0);
+
+   cl[ 1] = __gen_uint(values->pad, 0, 23) |
+            __gen_sint(values->offset_t, 4, 7) |
+            __gen_sint(values->offset_s, 0, 3);
+
+   cl[ 2] = __gen_uint(values->pad, 0, 23) >> 8 |
+            __gen_uint(values->op, 4, 7) |
+            __gen_sint(values->offset_r, 0, 3);
+
+   cl[ 3] = __gen_uint(values->pad, 0, 23) >> 16;
+
+}
+
+#define V3D41_TMU_CONFIG_PARAMETER_2_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TMU_CONFIG_PARAMETER_2_unpack(const uint8_t * restrict cl,
+                                    struct V3D41_TMU_CONFIG_PARAMETER_2 * restrict values)
+{
+   values->pad = __gen_unpack_uint(cl, 8, 31);
+   values->op = __gen_unpack_uint(cl, 20, 23);
+   values->offset_r = __gen_unpack_sint(cl, 16, 19);
+   values->offset_t = __gen_unpack_sint(cl, 12, 15);
+   values->offset_s = __gen_unpack_sint(cl, 8, 11);
+   values->gather_mode = __gen_unpack_uint(cl, 7, 7);
+   values->gather_component = __gen_unpack_uint(cl, 5, 6);
+   values->coefficient_mode = __gen_unpack_uint(cl, 4, 4);
+   values->sample_number = __gen_unpack_uint(cl, 2, 3);
+   values->disable_autolod = __gen_unpack_uint(cl, 1, 1);
+   values->offset_format_8 = __gen_unpack_uint(cl, 0, 0);
+}
+#endif
+
+
+#define V3D41_TEXTURE_SHADER_STATE_header       \
+
+
+struct V3D41_TEXTURE_SHADER_STATE {
+   uint64_t                             pad;
+   bool                                 uif_xor_disable;
+   bool                                 level_0_is_strictly_uif;
+   bool                                 level_0_xor_enable;
+   uint32_t                             level_0_ub_pad;
+   uint32_t                             base_level;
+   uint32_t                             max_level;
+   uint32_t                             swizzle_a;
+#define SWIZZLE_ZERO                             0
+#define SWIZZLE_ONE                              1
+#define SWIZZLE_RED                              2
+#define SWIZZLE_GREEN                            3
+#define SWIZZLE_BLUE                             4
+#define SWIZZLE_ALPHA                            5
+   uint32_t                             swizzle_b;
+   uint32_t                             swizzle_g;
+   uint32_t                             swizzle_r;
+   bool                                 extended;
+   uint32_t                             texture_type;
+   uint32_t                             image_depth;
+   uint32_t                             image_height;
+   uint32_t                             image_width;
+   uint32_t                             array_stride_64_byte_aligned;
+   __gen_address_type                   texture_base_pointer;
+   bool                                 reverse_standard_border_color;
+   bool                                 ahdr;
+   bool                                 srgb;
+   bool                                 flip_s_and_t_on_incoming_request;
+   bool                                 flip_texture_y_axis;
+   bool                                 flip_texture_x_axis;
+};
+
+static inline void
+V3D41_TEXTURE_SHADER_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                const struct V3D41_TEXTURE_SHADER_STATE * restrict values)
+{
+   __gen_emit_reloc(data, &values->texture_base_pointer);
+   cl[ 0] = __gen_address_offset(&values->texture_base_pointer) |
+            __gen_uint(values->reverse_standard_border_color, 5, 5) |
+            __gen_uint(values->ahdr, 4, 4) |
+            __gen_uint(values->srgb, 3, 3) |
+            __gen_uint(values->flip_s_and_t_on_incoming_request, 2, 2) |
+            __gen_uint(values->flip_texture_y_axis, 1, 1) |
+            __gen_uint(values->flip_texture_x_axis, 0, 0);
+
+   cl[ 1] = __gen_address_offset(&values->texture_base_pointer) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->texture_base_pointer) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->texture_base_pointer) >> 24;
+
+   cl[ 4] = __gen_uint(values->array_stride_64_byte_aligned, 0, 25);
+
+   cl[ 5] = __gen_uint(values->array_stride_64_byte_aligned, 0, 25) >> 8;
+
+   cl[ 6] = __gen_uint(values->array_stride_64_byte_aligned, 0, 25) >> 16;
+
+   cl[ 7] = __gen_uint(values->image_width, 2, 15) |
+            __gen_uint(values->array_stride_64_byte_aligned, 0, 25) >> 24;
+
+   cl[ 8] = __gen_uint(values->image_width, 2, 15) >> 8;
+
+   cl[ 9] = __gen_uint(values->image_height, 0, 13);
+
+   cl[10] = __gen_uint(values->image_depth, 6, 19) |
+            __gen_uint(values->image_height, 0, 13) >> 8;
+
+   cl[11] = __gen_uint(values->image_depth, 6, 19) >> 8;
+
+   cl[12] = __gen_uint(values->texture_type, 4, 10) |
+            __gen_uint(values->image_depth, 6, 19) >> 16;
+
+   cl[13] = __gen_uint(values->swizzle_g, 7, 9) |
+            __gen_uint(values->swizzle_r, 4, 6) |
+            __gen_uint(values->extended, 3, 3) |
+            __gen_uint(values->texture_type, 4, 10) >> 8;
+
+   cl[14] = __gen_uint(values->swizzle_a, 5, 7) |
+            __gen_uint(values->swizzle_b, 2, 4) |
+            __gen_uint(values->swizzle_g, 7, 9) >> 8;
+
+   cl[15] = __gen_uint(values->base_level, 4, 7) |
+            __gen_uint(values->max_level, 0, 3);
+
+   cl[16] = __gen_uint(values->uif_xor_disable, 7, 7) |
+            __gen_uint(values->level_0_is_strictly_uif, 6, 6) |
+            __gen_uint(values->level_0_xor_enable, 4, 4) |
+            __gen_uint(values->level_0_ub_pad, 0, 3);
+
+   cl[17] = __gen_uint(values->pad, 0, 55);
+
+   cl[18] = __gen_uint(values->pad, 0, 55) >> 8;
+
+   cl[19] = __gen_uint(values->pad, 0, 55) >> 16;
+
+   cl[20] = __gen_uint(values->pad, 0, 55) >> 24;
+
+   cl[21] = __gen_uint(values->pad, 0, 55) >> 32;
+
+   cl[22] = __gen_uint(values->pad, 0, 55) >> 40;
+
+   cl[23] = __gen_uint(values->pad, 0, 55) >> 48;
+
+}
+
+#define V3D41_TEXTURE_SHADER_STATE_length     24
+#ifdef __gen_unpack_address
+static inline void
+V3D41_TEXTURE_SHADER_STATE_unpack(const uint8_t * restrict cl,
+                                  struct V3D41_TEXTURE_SHADER_STATE * restrict values)
+{
+   values->pad = __gen_unpack_uint(cl, 136, 191);
+   values->uif_xor_disable = __gen_unpack_uint(cl, 135, 135);
+   values->level_0_is_strictly_uif = __gen_unpack_uint(cl, 134, 134);
+   values->level_0_xor_enable = __gen_unpack_uint(cl, 132, 132);
+   values->level_0_ub_pad = __gen_unpack_uint(cl, 128, 131);
+   values->base_level = __gen_unpack_uint(cl, 124, 127);
+   values->max_level = __gen_unpack_uint(cl, 120, 123);
+   values->swizzle_a = __gen_unpack_uint(cl, 117, 119);
+   values->swizzle_b = __gen_unpack_uint(cl, 114, 116);
+   values->swizzle_g = __gen_unpack_uint(cl, 111, 113);
+   values->swizzle_r = __gen_unpack_uint(cl, 108, 110);
+   values->extended = __gen_unpack_uint(cl, 107, 107);
+   values->texture_type = __gen_unpack_uint(cl, 100, 106);
+   values->image_depth = __gen_unpack_uint(cl, 86, 99);
+   values->image_height = __gen_unpack_uint(cl, 72, 85);
+   values->image_width = __gen_unpack_uint(cl, 58, 71);
+   values->array_stride_64_byte_aligned = __gen_unpack_uint(cl, 32, 57);
+   values->texture_base_pointer = __gen_unpack_address(cl, 0, 31);
+   values->reverse_standard_border_color = __gen_unpack_uint(cl, 5, 5);
+   values->ahdr = __gen_unpack_uint(cl, 4, 4);
+   values->srgb = __gen_unpack_uint(cl, 3, 3);
+   values->flip_s_and_t_on_incoming_request = __gen_unpack_uint(cl, 2, 2);
+   values->flip_texture_y_axis = __gen_unpack_uint(cl, 1, 1);
+   values->flip_texture_x_axis = __gen_unpack_uint(cl, 0, 0);
+}
+#endif
+
+
+#define V3D41_SAMPLER_STATE_header              \
+
+
+struct V3D41_SAMPLER_STATE {
+   uint32_t                             border_color_alpha;
+   uint32_t                             border_color_blue;
+   uint32_t                             border_color_green;
+   uint32_t                             border_color_red;
+   uint32_t                             maximum_anisotropy;
+   enum V3D41_Border_Color_Mode         border_color_mode;
+   bool                                 wrap_i_border;
+   enum V3D41_Wrap_Mode                 wrap_r;
+   enum V3D41_Wrap_Mode                 wrap_t;
+   enum V3D41_Wrap_Mode                 wrap_s;
+   float                                fixed_bias;
+   float                                max_level_of_detail;
+   float                                min_level_of_detail;
+   bool                                 srgb_disable;
+   enum V3D41_Compare_Function          depth_compare_function;
+   bool                                 anisotropy_enable;
+   bool                                 mip_filter_nearest;
+   bool                                 min_filter_nearest;
+   bool                                 mag_filter_nearest;
+};
+
+static inline void
+V3D41_SAMPLER_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                         const struct V3D41_SAMPLER_STATE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->srgb_disable, 7, 7) |
+            __gen_uint(values->depth_compare_function, 4, 6) |
+            __gen_uint(values->anisotropy_enable, 3, 3) |
+            __gen_uint(values->mip_filter_nearest, 2, 2) |
+            __gen_uint(values->min_filter_nearest, 1, 1) |
+            __gen_uint(values->mag_filter_nearest, 0, 0);
+
+   cl[ 1] = __gen_ufixed(values->min_level_of_detail, 0, 11, 8);
+
+   cl[ 2] = __gen_ufixed(values->max_level_of_detail, 4, 15, 8) |
+            __gen_ufixed(values->min_level_of_detail, 0, 11, 8) >> 8;
+
+   cl[ 3] = __gen_ufixed(values->max_level_of_detail, 4, 15, 8) >> 8;
+
+   cl[ 4] = __gen_sfixed(values->fixed_bias, 0, 15, 8);
+
+   cl[ 5] = __gen_sfixed(values->fixed_bias, 0, 15, 8) >> 8;
+
+   cl[ 6] = __gen_uint(values->wrap_r, 6, 8) |
+            __gen_uint(values->wrap_t, 3, 5) |
+            __gen_uint(values->wrap_s, 0, 2);
+
+   cl[ 7] = __gen_uint(values->maximum_anisotropy, 5, 6) |
+            __gen_uint(values->border_color_mode, 2, 4) |
+            __gen_uint(values->wrap_i_border, 1, 1) |
+            __gen_uint(values->wrap_r, 6, 8) >> 8;
+
+
+   memcpy(&cl[8], &values->border_color_red, sizeof(values->border_color_red));
+
+   memcpy(&cl[12], &values->border_color_green, sizeof(values->border_color_green));
+
+   memcpy(&cl[16], &values->border_color_blue, sizeof(values->border_color_blue));
+
+   memcpy(&cl[20], &values->border_color_alpha, sizeof(values->border_color_alpha));
+}
+
+#define V3D41_SAMPLER_STATE_length            24
+#ifdef __gen_unpack_address
+static inline void
+V3D41_SAMPLER_STATE_unpack(const uint8_t * restrict cl,
+                           struct V3D41_SAMPLER_STATE * restrict values)
+{
+   values->border_color_alpha = __gen_unpack_uint(cl, 160, 191);
+   values->border_color_blue = __gen_unpack_uint(cl, 128, 159);
+   values->border_color_green = __gen_unpack_uint(cl, 96, 127);
+   values->border_color_red = __gen_unpack_uint(cl, 64, 95);
+   values->maximum_anisotropy = __gen_unpack_uint(cl, 61, 62);
+   values->border_color_mode = __gen_unpack_uint(cl, 58, 60);
+   values->wrap_i_border = __gen_unpack_uint(cl, 57, 57);
+   values->wrap_r = __gen_unpack_uint(cl, 54, 56);
+   values->wrap_t = __gen_unpack_uint(cl, 51, 53);
+   values->wrap_s = __gen_unpack_uint(cl, 48, 50);
+   values->fixed_bias = __gen_unpack_sfixed(cl, 32, 47, 8);
+   values->max_level_of_detail = __gen_unpack_ufixed(cl, 20, 31, 8);
+   values->min_level_of_detail = __gen_unpack_ufixed(cl, 8, 19, 8);
+   values->srgb_disable = __gen_unpack_uint(cl, 7, 7);
+   values->depth_compare_function = __gen_unpack_uint(cl, 4, 6);
+   values->anisotropy_enable = __gen_unpack_uint(cl, 3, 3);
+   values->mip_filter_nearest = __gen_unpack_uint(cl, 2, 2);
+   values->min_filter_nearest = __gen_unpack_uint(cl, 1, 1);
+   values->mag_filter_nearest = __gen_unpack_uint(cl, 0, 0);
+}
+#endif
+
+
+enum V3D41_Texture_Data_Formats {
+        TEXTURE_DATA_FORMAT_R8               =      0,
+        TEXTURE_DATA_FORMAT_R8_SNORM         =      1,
+        TEXTURE_DATA_FORMAT_RG8              =      2,
+        TEXTURE_DATA_FORMAT_RG8_SNORM        =      3,
+        TEXTURE_DATA_FORMAT_RGBA8            =      4,
+        TEXTURE_DATA_FORMAT_RGBA8_SNORM      =      5,
+        TEXTURE_DATA_FORMAT_RGB565           =      6,
+        TEXTURE_DATA_FORMAT_RGBA4            =      7,
+        TEXTURE_DATA_FORMAT_RGB5_A1          =      8,
+        TEXTURE_DATA_FORMAT_RGB10_A2         =      9,
+        TEXTURE_DATA_FORMAT_R16              =     10,
+        TEXTURE_DATA_FORMAT_R16_SNORM        =     11,
+        TEXTURE_DATA_FORMAT_RG16             =     12,
+        TEXTURE_DATA_FORMAT_RG16_SNORM       =     13,
+        TEXTURE_DATA_FORMAT_RGBA16           =     14,
+        TEXTURE_DATA_FORMAT_RGBA16_SNORM     =     15,
+        TEXTURE_DATA_FORMAT_R16F             =     16,
+        TEXTURE_DATA_FORMAT_RG16F            =     17,
+        TEXTURE_DATA_FORMAT_RGBA16F          =     18,
+        TEXTURE_DATA_FORMAT_R11F_G11F_B10F   =     19,
+        TEXTURE_DATA_FORMAT_RGB9_E5          =     20,
+        TEXTURE_DATA_FORMAT_DEPTH_COMP16     =     21,
+        TEXTURE_DATA_FORMAT_DEPTH_COMP24     =     22,
+        TEXTURE_DATA_FORMAT_DEPTH_COMP32F    =     23,
+        TEXTURE_DATA_FORMAT_DEPTH24_X8       =     24,
+        TEXTURE_DATA_FORMAT_R4               =     25,
+        TEXTURE_DATA_FORMAT_R1               =     26,
+        TEXTURE_DATA_FORMAT_S8               =     27,
+        TEXTURE_DATA_FORMAT_S16              =     28,
+        TEXTURE_DATA_FORMAT_R32F             =     29,
+        TEXTURE_DATA_FORMAT_RG32F            =     30,
+        TEXTURE_DATA_FORMAT_RGBA32F          =     31,
+        TEXTURE_DATA_FORMAT_RGB8_ETC2        =     32,
+        TEXTURE_DATA_FORMAT_RGB8_PUNCHTHROUGH_ALPHA1 =     33,
+        TEXTURE_DATA_FORMAT_R11_EAC          =     34,
+        TEXTURE_DATA_FORMAT_SIGNED_R11_EAC   =     35,
+        TEXTURE_DATA_FORMAT_RG11_EAC         =     36,
+        TEXTURE_DATA_FORMAT_SIGNED_RG11_EAC  =     37,
+        TEXTURE_DATA_FORMAT_RGBA8_ETC2_EAC   =     38,
+        TEXTURE_DATA_FORMAT_YCBCR_LUMA       =     39,
+        TEXTURE_DATA_FORMAT_YCBCR_420_CHROMA =     40,
+        TEXTURE_DATA_FORMAT_BC1              =     48,
+        TEXTURE_DATA_FORMAT_BC2              =     49,
+        TEXTURE_DATA_FORMAT_BC3              =     50,
+        TEXTURE_DATA_FORMAT_ASTC_4X4         =     64,
+        TEXTURE_DATA_FORMAT_ASTC_5X4         =     65,
+        TEXTURE_DATA_FORMAT_ASTC_5X5         =     66,
+        TEXTURE_DATA_FORMAT_ASTC_6X5         =     67,
+        TEXTURE_DATA_FORMAT_ASTC_6X6         =     68,
+        TEXTURE_DATA_FORMAT_ASTC_8X5         =     69,
+        TEXTURE_DATA_FORMAT_ASTC_8X6         =     70,
+        TEXTURE_DATA_FORMAT_ASTC_8X8         =     71,
+        TEXTURE_DATA_FORMAT_ASTC_10X5        =     72,
+        TEXTURE_DATA_FORMAT_ASTC_10X6        =     73,
+        TEXTURE_DATA_FORMAT_ASTC_10X8        =     74,
+        TEXTURE_DATA_FORMAT_ASTC_10X10       =     75,
+        TEXTURE_DATA_FORMAT_ASTC_12X10       =     76,
+        TEXTURE_DATA_FORMAT_ASTC_12X12       =     77,
+        TEXTURE_DATA_FORMAT_R8I              =     96,
+        TEXTURE_DATA_FORMAT_R8UI             =     97,
+        TEXTURE_DATA_FORMAT_RG8I             =     98,
+        TEXTURE_DATA_FORMAT_RG8UI            =     99,
+        TEXTURE_DATA_FORMAT_RGBA8I           =    100,
+        TEXTURE_DATA_FORMAT_RGBA8UI          =    101,
+        TEXTURE_DATA_FORMAT_R16I             =    102,
+        TEXTURE_DATA_FORMAT_R16UI            =    103,
+        TEXTURE_DATA_FORMAT_RG16I            =    104,
+        TEXTURE_DATA_FORMAT_RG16UI           =    105,
+        TEXTURE_DATA_FORMAT_RGBA16I          =    106,
+        TEXTURE_DATA_FORMAT_RGBA16UI         =    107,
+        TEXTURE_DATA_FORMAT_R32I             =    108,
+        TEXTURE_DATA_FORMAT_R32UI            =    109,
+        TEXTURE_DATA_FORMAT_RG32I            =    110,
+        TEXTURE_DATA_FORMAT_RG32UI           =    111,
+        TEXTURE_DATA_FORMAT_RGBA32I          =    112,
+        TEXTURE_DATA_FORMAT_RGBA32UI         =    113,
+        TEXTURE_DATA_FORMAT_RGB10_A2UI       =    114,
+};
+
+#endif /* V3D41_PACK_H */
diff --git a/lib/mesa/src/broadcom/cle/v3d_packet_v42_pack.h b/lib/mesa/src/broadcom/cle/v3d_packet_v42_pack.h
new file mode 100644
index 000000000..48870087f
--- /dev/null
+++ b/lib/mesa/src/broadcom/cle/v3d_packet_v42_pack.h
@@ -0,0 +1,4403 @@
+/* Generated code, see packets.xml and gen_packet_header.py */
+
+
+/* Packets, enums and structures for V3D 4.2.
+ *
+ * This file has been generated, do not hand edit.
+ */
+
+#ifndef V3D42_PACK_H
+#define V3D42_PACK_H
+
+#include "cle/v3d_packet_helpers.h"
+
+
+enum V3D42_Compare_Function {
+        V3D_COMPARE_FUNC_NEVER               =      0,
+        V3D_COMPARE_FUNC_LESS                =      1,
+        V3D_COMPARE_FUNC_EQUAL               =      2,
+        V3D_COMPARE_FUNC_LEQUAL              =      3,
+        V3D_COMPARE_FUNC_GREATER             =      4,
+        V3D_COMPARE_FUNC_NOTEQUAL            =      5,
+        V3D_COMPARE_FUNC_GEQUAL              =      6,
+        V3D_COMPARE_FUNC_ALWAYS              =      7,
+};
+
+enum V3D42_Blend_Factor {
+        V3D_BLEND_FACTOR_ZERO                =      0,
+        V3D_BLEND_FACTOR_ONE                 =      1,
+        V3D_BLEND_FACTOR_SRC_COLOR           =      2,
+        V3D_BLEND_FACTOR_INV_SRC_COLOR       =      3,
+        V3D_BLEND_FACTOR_DST_COLOR           =      4,
+        V3D_BLEND_FACTOR_INV_DST_COLOR       =      5,
+        V3D_BLEND_FACTOR_SRC_ALPHA           =      6,
+        V3D_BLEND_FACTOR_INV_SRC_ALPHA       =      7,
+        V3D_BLEND_FACTOR_DST_ALPHA           =      8,
+        V3D_BLEND_FACTOR_INV_DST_ALPHA       =      9,
+        V3D_BLEND_FACTOR_CONST_COLOR         =     10,
+        V3D_BLEND_FACTOR_INV_CONST_COLOR     =     11,
+        V3D_BLEND_FACTOR_CONST_ALPHA         =     12,
+        V3D_BLEND_FACTOR_INV_CONST_ALPHA     =     13,
+        V3D_BLEND_FACTOR_SRC_ALPHA_SATURATE  =     14,
+};
+
+enum V3D42_Blend_Mode {
+        V3D_BLEND_MODE_ADD                   =      0,
+        V3D_BLEND_MODE_SUB                   =      1,
+        V3D_BLEND_MODE_RSUB                  =      2,
+        V3D_BLEND_MODE_MIN                   =      3,
+        V3D_BLEND_MODE_MAX                   =      4,
+        V3D_BLEND_MODE_MUL                   =      5,
+        V3D_BLEND_MODE_SCREEN                =      6,
+        V3D_BLEND_MODE_DARKEN                =      7,
+        V3D_BLEND_MODE_LIGHTEN               =      8,
+};
+
+enum V3D42_Stencil_Op {
+        V3D_STENCIL_OP_ZERO                  =      0,
+        V3D_STENCIL_OP_KEEP                  =      1,
+        V3D_STENCIL_OP_REPLACE               =      2,
+        V3D_STENCIL_OP_INCR                  =      3,
+        V3D_STENCIL_OP_DECR                  =      4,
+        V3D_STENCIL_OP_INVERT                =      5,
+        V3D_STENCIL_OP_INCWRAP               =      6,
+        V3D_STENCIL_OP_DECWRAP               =      7,
+};
+
+enum V3D42_Primitive {
+        V3D_PRIM_POINTS                      =      0,
+        V3D_PRIM_LINES                       =      1,
+        V3D_PRIM_LINE_LOOP                   =      2,
+        V3D_PRIM_LINE_STRIP                  =      3,
+        V3D_PRIM_TRIANGLES                   =      4,
+        V3D_PRIM_TRIANGLE_STRIP              =      5,
+        V3D_PRIM_TRIANGLE_FAN                =      6,
+        V3D_PRIM_POINTS_TF                   =     16,
+        V3D_PRIM_LINES_TF                    =     17,
+        V3D_PRIM_LINE_LOOP_TF                =     18,
+        V3D_PRIM_LINE_STRIP_TF               =     19,
+        V3D_PRIM_TRIANGLES_TF                =     20,
+        V3D_PRIM_TRIANGLE_STRIP_TF           =     21,
+        V3D_PRIM_TRIANGLE_FAN_TF             =     22,
+};
+
+enum V3D42_Border_Color_Mode {
+        V3D_BORDER_COLOR_0000                =      0,
+        V3D_BORDER_COLOR_0001                =      1,
+        V3D_BORDER_COLOR_1111                =      2,
+        V3D_BORDER_COLOR_FOLLOWS             =      7,
+};
+
+enum V3D42_Wrap_Mode {
+        V3D_WRAP_MODE_WRAP_MODE_REPEAT       =      0,
+        V3D_WRAP_MODE_WRAP_MODE_CLAMP        =      1,
+        V3D_WRAP_MODE_WRAP_MODE_MIRROR       =      2,
+        V3D_WRAP_MODE_WRAP_MODE_BORDER       =      3,
+        V3D_WRAP_MODE_WRAP_MODE_MIRROR_ONCE  =      4,
+};
+
+enum V3D42_TMU_Op {
+        V3D_TMU_OP_WRITE_ADD_READ_PREFETCH   =      0,
+        V3D_TMU_OP_WRITE_SUB_READ_CLEAR      =      1,
+        V3D_TMU_OP_WRITE_XCHG_READ_FLUSH     =      2,
+        V3D_TMU_OP_WRITE_CMPXCHG_READ_FLUSH  =      3,
+        V3D_TMU_OP_WRITE_UMIN_FULL_L1_CLEAR  =      4,
+        V3D_TMU_OP_WRITE_UMAX                =      5,
+        V3D_TMU_OP_WRITE_SMIN                =      6,
+        V3D_TMU_OP_WRITE_SMAX                =      7,
+        V3D_TMU_OP_WRITE_AND_READ_INC        =      8,
+        V3D_TMU_OP_WRITE_OR_READ_DEC         =      9,
+        V3D_TMU_OP_WRITE_XOR_READ_NOT        =     10,
+        V3D_TMU_OP_REGULAR                   =     15,
+};
+
+enum V3D42_Varying_Flags_Action {
+        V3D_VARYING_FLAGS_ACTION_UNCHANGED   =      0,
+        V3D_VARYING_FLAGS_ACTION_ZEROED      =      1,
+        V3D_VARYING_FLAGS_ACTION_SET         =      2,
+};
+
+enum V3D42_Memory_Format {
+        V3D_MEMORY_FORMAT_RASTER             =      0,
+        V3D_MEMORY_FORMAT_LINEARTILE         =      1,
+        V3D_MEMORY_FORMAT_UB_LINEAR_1_UIF_BLOCK_WIDE =      2,
+        V3D_MEMORY_FORMAT_UB_LINEAR_2_UIF_BLOCKS_WIDE =      3,
+        V3D_MEMORY_FORMAT_UIF_NO_XOR         =      4,
+        V3D_MEMORY_FORMAT_UIF_XOR            =      5,
+};
+
+enum V3D42_Decimate_Mode {
+        V3D_DECIMATE_MODE_SAMPLE_0           =      0,
+        V3D_DECIMATE_MODE_4X                 =      1,
+        V3D_DECIMATE_MODE_ALL_SAMPLES        =      3,
+};
+
+enum V3D42_Internal_Type {
+        V3D_INTERNAL_TYPE_8I                 =      0,
+        V3D_INTERNAL_TYPE_8UI                =      1,
+        V3D_INTERNAL_TYPE_8                  =      2,
+        V3D_INTERNAL_TYPE_16I                =      4,
+        V3D_INTERNAL_TYPE_16UI               =      5,
+        V3D_INTERNAL_TYPE_16F                =      6,
+        V3D_INTERNAL_TYPE_32I                =      8,
+        V3D_INTERNAL_TYPE_32UI               =      9,
+        V3D_INTERNAL_TYPE_32F                =     10,
+};
+
+enum V3D42_Internal_BPP {
+        V3D_INTERNAL_BPP_32                  =      0,
+        V3D_INTERNAL_BPP_64                  =      1,
+        V3D_INTERNAL_BPP_128                 =      2,
+};
+
+enum V3D42_Internal_Depth_Type {
+        V3D_INTERNAL_TYPE_DEPTH_32F          =      0,
+        V3D_INTERNAL_TYPE_DEPTH_24           =      1,
+        V3D_INTERNAL_TYPE_DEPTH_16           =      2,
+};
+
+enum V3D42_Render_Target_Clamp {
+        V3D_RENDER_TARGET_CLAMP_NONE         =      0,
+        V3D_RENDER_TARGET_CLAMP_NORM         =      1,
+        V3D_RENDER_TARGET_CLAMP_POS          =      2,
+        V3D_RENDER_TARGET_CLAMP_INT          =      3,
+};
+
+enum V3D42_Output_Image_Format {
+        V3D_OUTPUT_IMAGE_FORMAT_SRGB8_ALPHA8 =      0,
+        V3D_OUTPUT_IMAGE_FORMAT_SRGB         =      1,
+        V3D_OUTPUT_IMAGE_FORMAT_RGB10_A2UI   =      2,
+        V3D_OUTPUT_IMAGE_FORMAT_RGB10_A2     =      3,
+        V3D_OUTPUT_IMAGE_FORMAT_ABGR1555     =      4,
+        V3D_OUTPUT_IMAGE_FORMAT_ALPHA_MASKED_ABGR1555 =      5,
+        V3D_OUTPUT_IMAGE_FORMAT_ABGR4444     =      6,
+        V3D_OUTPUT_IMAGE_FORMAT_BGR565       =      7,
+        V3D_OUTPUT_IMAGE_FORMAT_R11F_G11F_B10F =      8,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA32F      =      9,
+        V3D_OUTPUT_IMAGE_FORMAT_RG32F        =     10,
+        V3D_OUTPUT_IMAGE_FORMAT_R32F         =     11,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA32I      =     12,
+        V3D_OUTPUT_IMAGE_FORMAT_RG32I        =     13,
+        V3D_OUTPUT_IMAGE_FORMAT_R32I         =     14,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA32UI     =     15,
+        V3D_OUTPUT_IMAGE_FORMAT_RG32UI       =     16,
+        V3D_OUTPUT_IMAGE_FORMAT_R32UI        =     17,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA16F      =     18,
+        V3D_OUTPUT_IMAGE_FORMAT_RG16F        =     19,
+        V3D_OUTPUT_IMAGE_FORMAT_R16F         =     20,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA16I      =     21,
+        V3D_OUTPUT_IMAGE_FORMAT_RG16I        =     22,
+        V3D_OUTPUT_IMAGE_FORMAT_R16I         =     23,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA16UI     =     24,
+        V3D_OUTPUT_IMAGE_FORMAT_RG16UI       =     25,
+        V3D_OUTPUT_IMAGE_FORMAT_R16UI        =     26,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA8        =     27,
+        V3D_OUTPUT_IMAGE_FORMAT_RGB8         =     28,
+        V3D_OUTPUT_IMAGE_FORMAT_RG8          =     29,
+        V3D_OUTPUT_IMAGE_FORMAT_R8           =     30,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA8I       =     31,
+        V3D_OUTPUT_IMAGE_FORMAT_RG8I         =     32,
+        V3D_OUTPUT_IMAGE_FORMAT_R8I          =     33,
+        V3D_OUTPUT_IMAGE_FORMAT_RGBA8UI      =     34,
+        V3D_OUTPUT_IMAGE_FORMAT_RG8UI        =     35,
+        V3D_OUTPUT_IMAGE_FORMAT_R8UI         =     36,
+        V3D_OUTPUT_IMAGE_FORMAT_BSTC         =     39,
+        V3D_OUTPUT_IMAGE_FORMAT_D32F         =     40,
+        V3D_OUTPUT_IMAGE_FORMAT_D24          =     41,
+        V3D_OUTPUT_IMAGE_FORMAT_D16          =     42,
+        V3D_OUTPUT_IMAGE_FORMAT_D24S8        =     43,
+        V3D_OUTPUT_IMAGE_FORMAT_S8           =     44,
+};
+
+enum V3D42_Dither_Mode {
+        V3D_DITHER_MODE_NONE                 =      0,
+        V3D_DITHER_MODE_RGB                  =      1,
+        V3D_DITHER_MODE_A                    =      2,
+        V3D_DITHER_MODE_RGBA                 =      3,
+};
+
+#define V3D42_HALT_opcode                      0
+#define V3D42_HALT_header                       \
+   .opcode                              =      0
+
+struct V3D42_HALT {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_HALT_pack(__gen_user_data *data, uint8_t * restrict cl,
+                const struct V3D42_HALT * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_HALT_length                      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_HALT_unpack(const uint8_t * restrict cl,
+                  struct V3D42_HALT * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_NOP_opcode                       1
+#define V3D42_NOP_header                        \
+   .opcode                              =      1
+
+struct V3D42_NOP {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_NOP_pack(__gen_user_data *data, uint8_t * restrict cl,
+               const struct V3D42_NOP * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_NOP_length                       1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_NOP_unpack(const uint8_t * restrict cl,
+                 struct V3D42_NOP * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_FLUSH_opcode                     4
+#define V3D42_FLUSH_header                      \
+   .opcode                              =      4
+
+struct V3D42_FLUSH {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_FLUSH_pack(__gen_user_data *data, uint8_t * restrict cl,
+                 const struct V3D42_FLUSH * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_FLUSH_length                     1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_FLUSH_unpack(const uint8_t * restrict cl,
+                   struct V3D42_FLUSH * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_FLUSH_ALL_STATE_opcode           5
+#define V3D42_FLUSH_ALL_STATE_header            \
+   .opcode                              =      5
+
+struct V3D42_FLUSH_ALL_STATE {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_FLUSH_ALL_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                           const struct V3D42_FLUSH_ALL_STATE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_FLUSH_ALL_STATE_length           1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_FLUSH_ALL_STATE_unpack(const uint8_t * restrict cl,
+                             struct V3D42_FLUSH_ALL_STATE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_START_TILE_BINNING_opcode        6
+#define V3D42_START_TILE_BINNING_header         \
+   .opcode                              =      6
+
+struct V3D42_START_TILE_BINNING {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_START_TILE_BINNING_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D42_START_TILE_BINNING * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_START_TILE_BINNING_length        1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_START_TILE_BINNING_unpack(const uint8_t * restrict cl,
+                                struct V3D42_START_TILE_BINNING * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_INCREMENT_SEMAPHORE_opcode       7
+#define V3D42_INCREMENT_SEMAPHORE_header        \
+   .opcode                              =      7
+
+struct V3D42_INCREMENT_SEMAPHORE {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_INCREMENT_SEMAPHORE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                               const struct V3D42_INCREMENT_SEMAPHORE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_INCREMENT_SEMAPHORE_length       1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_INCREMENT_SEMAPHORE_unpack(const uint8_t * restrict cl,
+                                 struct V3D42_INCREMENT_SEMAPHORE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_WAIT_ON_SEMAPHORE_opcode         8
+#define V3D42_WAIT_ON_SEMAPHORE_header          \
+   .opcode                              =      8
+
+struct V3D42_WAIT_ON_SEMAPHORE {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_WAIT_ON_SEMAPHORE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                             const struct V3D42_WAIT_ON_SEMAPHORE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_WAIT_ON_SEMAPHORE_length         1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_WAIT_ON_SEMAPHORE_unpack(const uint8_t * restrict cl,
+                               struct V3D42_WAIT_ON_SEMAPHORE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_WAIT_FOR_PREVIOUS_FRAME_opcode      9
+#define V3D42_WAIT_FOR_PREVIOUS_FRAME_header    \
+   .opcode                              =      9
+
+struct V3D42_WAIT_FOR_PREVIOUS_FRAME {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_WAIT_FOR_PREVIOUS_FRAME_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                   const struct V3D42_WAIT_FOR_PREVIOUS_FRAME * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_WAIT_FOR_PREVIOUS_FRAME_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_WAIT_FOR_PREVIOUS_FRAME_unpack(const uint8_t * restrict cl,
+                                     struct V3D42_WAIT_FOR_PREVIOUS_FRAME * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_ENABLE_Z_ONLY_RENDERING_opcode     10
+#define V3D42_ENABLE_Z_ONLY_RENDERING_header    \
+   .opcode                              =     10
+
+struct V3D42_ENABLE_Z_ONLY_RENDERING {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_ENABLE_Z_ONLY_RENDERING_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                   const struct V3D42_ENABLE_Z_ONLY_RENDERING * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_ENABLE_Z_ONLY_RENDERING_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_ENABLE_Z_ONLY_RENDERING_unpack(const uint8_t * restrict cl,
+                                     struct V3D42_ENABLE_Z_ONLY_RENDERING * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_DISABLE_Z_ONLY_RENDERING_opcode     11
+#define V3D42_DISABLE_Z_ONLY_RENDERING_header   \
+   .opcode                              =     11
+
+struct V3D42_DISABLE_Z_ONLY_RENDERING {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_DISABLE_Z_ONLY_RENDERING_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                    const struct V3D42_DISABLE_Z_ONLY_RENDERING * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_DISABLE_Z_ONLY_RENDERING_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_DISABLE_Z_ONLY_RENDERING_unpack(const uint8_t * restrict cl,
+                                      struct V3D42_DISABLE_Z_ONLY_RENDERING * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_END_OF_Z_ONLY_RENDERING_IN_FRAME_opcode     12
+#define V3D42_END_OF_Z_ONLY_RENDERING_IN_FRAME_header\
+   .opcode                              =     12
+
+struct V3D42_END_OF_Z_ONLY_RENDERING_IN_FRAME {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_END_OF_Z_ONLY_RENDERING_IN_FRAME_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                            const struct V3D42_END_OF_Z_ONLY_RENDERING_IN_FRAME * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_END_OF_Z_ONLY_RENDERING_IN_FRAME_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_END_OF_Z_ONLY_RENDERING_IN_FRAME_unpack(const uint8_t * restrict cl,
+                                              struct V3D42_END_OF_Z_ONLY_RENDERING_IN_FRAME * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_END_OF_RENDERING_opcode         13
+#define V3D42_END_OF_RENDERING_header           \
+   .opcode                              =     13
+
+struct V3D42_END_OF_RENDERING {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_END_OF_RENDERING_pack(__gen_user_data *data, uint8_t * restrict cl,
+                            const struct V3D42_END_OF_RENDERING * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_END_OF_RENDERING_length          1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_END_OF_RENDERING_unpack(const uint8_t * restrict cl,
+                              struct V3D42_END_OF_RENDERING * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_WAIT_FOR_TRANSFORM_FEEDBACK_opcode     14
+#define V3D42_WAIT_FOR_TRANSFORM_FEEDBACK_header\
+   .opcode                              =     14
+
+struct V3D42_WAIT_FOR_TRANSFORM_FEEDBACK {
+   uint32_t                             opcode;
+   uint32_t                             block_count;
+};
+
+static inline void
+V3D42_WAIT_FOR_TRANSFORM_FEEDBACK_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                       const struct V3D42_WAIT_FOR_TRANSFORM_FEEDBACK * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->block_count, 0, 7);
+
+}
+
+#define V3D42_WAIT_FOR_TRANSFORM_FEEDBACK_length      2
+#ifdef __gen_unpack_address
+static inline void
+V3D42_WAIT_FOR_TRANSFORM_FEEDBACK_unpack(const uint8_t * restrict cl,
+                                         struct V3D42_WAIT_FOR_TRANSFORM_FEEDBACK * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->block_count = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D42_BRANCH_TO_AUTO_CHAINED_SUB_LIST_opcode     15
+#define V3D42_BRANCH_TO_AUTO_CHAINED_SUB_LIST_header\
+   .opcode                              =     15
+
+struct V3D42_BRANCH_TO_AUTO_CHAINED_SUB_LIST {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+};
+
+static inline void
+V3D42_BRANCH_TO_AUTO_CHAINED_SUB_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                           const struct V3D42_BRANCH_TO_AUTO_CHAINED_SUB_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D42_BRANCH_TO_AUTO_CHAINED_SUB_LIST_length      5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_BRANCH_TO_AUTO_CHAINED_SUB_LIST_unpack(const uint8_t * restrict cl,
+                                             struct V3D42_BRANCH_TO_AUTO_CHAINED_SUB_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 8, 39);
+}
+#endif
+
+
+#define V3D42_BRANCH_opcode                   16
+#define V3D42_BRANCH_header                     \
+   .opcode                              =     16
+
+struct V3D42_BRANCH {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+};
+
+static inline void
+V3D42_BRANCH_pack(__gen_user_data *data, uint8_t * restrict cl,
+                  const struct V3D42_BRANCH * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D42_BRANCH_length                    5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_BRANCH_unpack(const uint8_t * restrict cl,
+                    struct V3D42_BRANCH * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 8, 39);
+}
+#endif
+
+
+#define V3D42_BRANCH_TO_SUB_LIST_opcode       17
+#define V3D42_BRANCH_TO_SUB_LIST_header         \
+   .opcode                              =     17
+
+struct V3D42_BRANCH_TO_SUB_LIST {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+};
+
+static inline void
+V3D42_BRANCH_TO_SUB_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D42_BRANCH_TO_SUB_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D42_BRANCH_TO_SUB_LIST_length        5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_BRANCH_TO_SUB_LIST_unpack(const uint8_t * restrict cl,
+                                struct V3D42_BRANCH_TO_SUB_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 8, 39);
+}
+#endif
+
+
+#define V3D42_RETURN_FROM_SUB_LIST_opcode     18
+#define V3D42_RETURN_FROM_SUB_LIST_header       \
+   .opcode                              =     18
+
+struct V3D42_RETURN_FROM_SUB_LIST {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_RETURN_FROM_SUB_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                const struct V3D42_RETURN_FROM_SUB_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_RETURN_FROM_SUB_LIST_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_RETURN_FROM_SUB_LIST_unpack(const uint8_t * restrict cl,
+                                  struct V3D42_RETURN_FROM_SUB_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_FLUSH_VCD_CACHE_opcode          19
+#define V3D42_FLUSH_VCD_CACHE_header            \
+   .opcode                              =     19
+
+struct V3D42_FLUSH_VCD_CACHE {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_FLUSH_VCD_CACHE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                           const struct V3D42_FLUSH_VCD_CACHE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_FLUSH_VCD_CACHE_length           1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_FLUSH_VCD_CACHE_unpack(const uint8_t * restrict cl,
+                             struct V3D42_FLUSH_VCD_CACHE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_START_ADDRESS_OF_GENERIC_TILE_LIST_opcode     20
+#define V3D42_START_ADDRESS_OF_GENERIC_TILE_LIST_header\
+   .opcode                              =     20
+
+struct V3D42_START_ADDRESS_OF_GENERIC_TILE_LIST {
+   uint32_t                             opcode;
+   __gen_address_type                   start;
+   __gen_address_type                   end;
+};
+
+static inline void
+V3D42_START_ADDRESS_OF_GENERIC_TILE_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                              const struct V3D42_START_ADDRESS_OF_GENERIC_TILE_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->start);
+   cl[ 1] = __gen_address_offset(&values->start);
+
+   cl[ 2] = __gen_address_offset(&values->start) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->start) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->start) >> 24;
+
+   __gen_emit_reloc(data, &values->end);
+   cl[ 5] = __gen_address_offset(&values->end);
+
+   cl[ 6] = __gen_address_offset(&values->end) >> 8;
+
+   cl[ 7] = __gen_address_offset(&values->end) >> 16;
+
+   cl[ 8] = __gen_address_offset(&values->end) >> 24;
+
+}
+
+#define V3D42_START_ADDRESS_OF_GENERIC_TILE_LIST_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_START_ADDRESS_OF_GENERIC_TILE_LIST_unpack(const uint8_t * restrict cl,
+                                                struct V3D42_START_ADDRESS_OF_GENERIC_TILE_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->start = __gen_unpack_address(cl, 8, 39);
+   values->end = __gen_unpack_address(cl, 40, 71);
+}
+#endif
+
+
+#define V3D42_BRANCH_TO_IMPLICIT_TILE_LIST_opcode     21
+#define V3D42_BRANCH_TO_IMPLICIT_TILE_LIST_header\
+   .opcode                              =     21
+
+struct V3D42_BRANCH_TO_IMPLICIT_TILE_LIST {
+   uint32_t                             opcode;
+   uint32_t                             tile_list_set_number;
+};
+
+static inline void
+V3D42_BRANCH_TO_IMPLICIT_TILE_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D42_BRANCH_TO_IMPLICIT_TILE_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->tile_list_set_number, 0, 7);
+
+}
+
+#define V3D42_BRANCH_TO_IMPLICIT_TILE_LIST_length      2
+#ifdef __gen_unpack_address
+static inline void
+V3D42_BRANCH_TO_IMPLICIT_TILE_LIST_unpack(const uint8_t * restrict cl,
+                                          struct V3D42_BRANCH_TO_IMPLICIT_TILE_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->tile_list_set_number = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D42_BRANCH_TO_EXPLICIT_SUPERTILE_opcode     22
+#define V3D42_BRANCH_TO_EXPLICIT_SUPERTILE_header\
+   .opcode                              =     22
+
+struct V3D42_BRANCH_TO_EXPLICIT_SUPERTILE {
+   uint32_t                             opcode;
+   __gen_address_type                   absolute_address_of_explicit_supertile_render_list;
+   uint32_t                             explicit_supertile_number;
+   uint32_t                             row_number;
+   uint32_t                             column_number;
+};
+
+static inline void
+V3D42_BRANCH_TO_EXPLICIT_SUPERTILE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D42_BRANCH_TO_EXPLICIT_SUPERTILE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->column_number, 0, 7);
+
+   cl[ 2] = __gen_uint(values->row_number, 0, 7);
+
+   cl[ 3] = __gen_uint(values->explicit_supertile_number, 0, 7);
+
+   __gen_emit_reloc(data, &values->absolute_address_of_explicit_supertile_render_list);
+   cl[ 4] = __gen_address_offset(&values->absolute_address_of_explicit_supertile_render_list);
+
+   cl[ 5] = __gen_address_offset(&values->absolute_address_of_explicit_supertile_render_list) >> 8;
+
+   cl[ 6] = __gen_address_offset(&values->absolute_address_of_explicit_supertile_render_list) >> 16;
+
+   cl[ 7] = __gen_address_offset(&values->absolute_address_of_explicit_supertile_render_list) >> 24;
+
+}
+
+#define V3D42_BRANCH_TO_EXPLICIT_SUPERTILE_length      8
+#ifdef __gen_unpack_address
+static inline void
+V3D42_BRANCH_TO_EXPLICIT_SUPERTILE_unpack(const uint8_t * restrict cl,
+                                          struct V3D42_BRANCH_TO_EXPLICIT_SUPERTILE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->absolute_address_of_explicit_supertile_render_list = __gen_unpack_address(cl, 32, 63);
+   values->explicit_supertile_number = __gen_unpack_uint(cl, 24, 31);
+   values->row_number = __gen_unpack_uint(cl, 16, 23);
+   values->column_number = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D42_SUPERTILE_COORDINATES_opcode     23
+#define V3D42_SUPERTILE_COORDINATES_header      \
+   .opcode                              =     23
+
+struct V3D42_SUPERTILE_COORDINATES {
+   uint32_t                             opcode;
+   uint32_t                             row_number_in_supertiles;
+   uint32_t                             column_number_in_supertiles;
+};
+
+static inline void
+V3D42_SUPERTILE_COORDINATES_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                 const struct V3D42_SUPERTILE_COORDINATES * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->column_number_in_supertiles, 0, 7);
+
+   cl[ 2] = __gen_uint(values->row_number_in_supertiles, 0, 7);
+
+}
+
+#define V3D42_SUPERTILE_COORDINATES_length      3
+#ifdef __gen_unpack_address
+static inline void
+V3D42_SUPERTILE_COORDINATES_unpack(const uint8_t * restrict cl,
+                                   struct V3D42_SUPERTILE_COORDINATES * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->row_number_in_supertiles = __gen_unpack_uint(cl, 16, 23);
+   values->column_number_in_supertiles = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D42_CLEAR_TILE_BUFFERS_opcode       25
+#define V3D42_CLEAR_TILE_BUFFERS_header         \
+   .opcode                              =     25
+
+struct V3D42_CLEAR_TILE_BUFFERS {
+   uint32_t                             opcode;
+   bool                                 clear_z_stencil_buffer;
+   bool                                 clear_all_render_targets;
+};
+
+static inline void
+V3D42_CLEAR_TILE_BUFFERS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D42_CLEAR_TILE_BUFFERS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->clear_z_stencil_buffer, 1, 1) |
+            __gen_uint(values->clear_all_render_targets, 0, 0);
+
+}
+
+#define V3D42_CLEAR_TILE_BUFFERS_length        2
+#ifdef __gen_unpack_address
+static inline void
+V3D42_CLEAR_TILE_BUFFERS_unpack(const uint8_t * restrict cl,
+                                struct V3D42_CLEAR_TILE_BUFFERS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->clear_z_stencil_buffer = __gen_unpack_uint(cl, 9, 9);
+   values->clear_all_render_targets = __gen_unpack_uint(cl, 8, 8);
+}
+#endif
+
+
+#define V3D42_END_OF_LOADS_opcode             26
+#define V3D42_END_OF_LOADS_header               \
+   .opcode                              =     26
+
+struct V3D42_END_OF_LOADS {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_END_OF_LOADS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                        const struct V3D42_END_OF_LOADS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_END_OF_LOADS_length              1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_END_OF_LOADS_unpack(const uint8_t * restrict cl,
+                          struct V3D42_END_OF_LOADS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_END_OF_TILE_MARKER_opcode       27
+#define V3D42_END_OF_TILE_MARKER_header         \
+   .opcode                              =     27
+
+struct V3D42_END_OF_TILE_MARKER {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_END_OF_TILE_MARKER_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D42_END_OF_TILE_MARKER * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_END_OF_TILE_MARKER_length        1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_END_OF_TILE_MARKER_unpack(const uint8_t * restrict cl,
+                                struct V3D42_END_OF_TILE_MARKER * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_STORE_TILE_BUFFER_GENERAL_opcode     29
+#define V3D42_STORE_TILE_BUFFER_GENERAL_header  \
+   .opcode                              =     29
+
+struct V3D42_STORE_TILE_BUFFER_GENERAL {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+   uint32_t                             height;
+   uint32_t                             height_in_ub_or_stride;
+   bool                                 r_b_swap;
+   bool                                 channel_reverse;
+   bool                                 clear_buffer_being_stored;
+   enum V3D42_Output_Image_Format       output_image_format;
+   enum V3D42_Decimate_Mode             decimate_mode;
+   enum V3D42_Dither_Mode               dither_mode;
+   bool                                 flip_y;
+   enum V3D42_Memory_Format             memory_format;
+   uint32_t                             buffer_to_store;
+#define RENDER_TARGET_0                          0
+#define RENDER_TARGET_1                          1
+#define RENDER_TARGET_2                          2
+#define RENDER_TARGET_3                          3
+#define NONE                                     8
+#define Z                                        9
+#define STENCIL                                  10
+#define ZSTENCIL                                 11
+};
+
+static inline void
+V3D42_STORE_TILE_BUFFER_GENERAL_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                     const struct V3D42_STORE_TILE_BUFFER_GENERAL * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->flip_y, 7, 7) |
+            __gen_uint(values->memory_format, 4, 6) |
+            __gen_uint(values->buffer_to_store, 0, 3);
+
+   cl[ 2] = __gen_uint(values->output_image_format, 4, 9) |
+            __gen_uint(values->decimate_mode, 2, 3) |
+            __gen_uint(values->dither_mode, 0, 1);
+
+   cl[ 3] = __gen_uint(values->r_b_swap, 4, 4) |
+            __gen_uint(values->channel_reverse, 3, 3) |
+            __gen_uint(values->clear_buffer_being_stored, 2, 2) |
+            __gen_uint(values->output_image_format, 4, 9) >> 8;
+
+   cl[ 4] = __gen_uint(values->height_in_ub_or_stride, 4, 23);
+
+   cl[ 5] = __gen_uint(values->height_in_ub_or_stride, 4, 23) >> 8;
+
+   cl[ 6] = __gen_uint(values->height_in_ub_or_stride, 4, 23) >> 16;
+
+   cl[ 7] = __gen_uint(values->height, 0, 15);
+
+   cl[ 8] = __gen_uint(values->height, 0, 15) >> 8;
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 9] = __gen_address_offset(&values->address);
+
+   cl[10] = __gen_address_offset(&values->address) >> 8;
+
+   cl[11] = __gen_address_offset(&values->address) >> 16;
+
+   cl[12] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D42_STORE_TILE_BUFFER_GENERAL_length     13
+#ifdef __gen_unpack_address
+static inline void
+V3D42_STORE_TILE_BUFFER_GENERAL_unpack(const uint8_t * restrict cl,
+                                       struct V3D42_STORE_TILE_BUFFER_GENERAL * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 72, 103);
+   values->height = __gen_unpack_uint(cl, 56, 71);
+   values->height_in_ub_or_stride = __gen_unpack_uint(cl, 36, 55);
+   values->r_b_swap = __gen_unpack_uint(cl, 28, 28);
+   values->channel_reverse = __gen_unpack_uint(cl, 27, 27);
+   values->clear_buffer_being_stored = __gen_unpack_uint(cl, 26, 26);
+   values->output_image_format = __gen_unpack_uint(cl, 20, 25);
+   values->decimate_mode = __gen_unpack_uint(cl, 18, 19);
+   values->dither_mode = __gen_unpack_uint(cl, 16, 17);
+   values->flip_y = __gen_unpack_uint(cl, 15, 15);
+   values->memory_format = __gen_unpack_uint(cl, 12, 14);
+   values->buffer_to_store = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_LOAD_TILE_BUFFER_GENERAL_opcode     30
+#define V3D42_LOAD_TILE_BUFFER_GENERAL_header   \
+   .opcode                              =     30
+
+struct V3D42_LOAD_TILE_BUFFER_GENERAL {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+   uint32_t                             height;
+   uint32_t                             height_in_ub_or_stride;
+   bool                                 r_b_swap;
+   bool                                 channel_reverse;
+   enum V3D42_Output_Image_Format       input_image_format;
+   enum V3D42_Decimate_Mode             decimate_mode;
+   bool                                 flip_y;
+   enum V3D42_Memory_Format             memory_format;
+   uint32_t                             buffer_to_load;
+#define RENDER_TARGET_0                          0
+#define RENDER_TARGET_1                          1
+#define RENDER_TARGET_2                          2
+#define RENDER_TARGET_3                          3
+#define NONE                                     8
+#define Z                                        9
+#define STENCIL                                  10
+#define ZSTENCIL                                 11
+};
+
+static inline void
+V3D42_LOAD_TILE_BUFFER_GENERAL_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                    const struct V3D42_LOAD_TILE_BUFFER_GENERAL * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->flip_y, 7, 7) |
+            __gen_uint(values->memory_format, 4, 6) |
+            __gen_uint(values->buffer_to_load, 0, 3);
+
+   cl[ 2] = __gen_uint(values->input_image_format, 4, 9) |
+            __gen_uint(values->decimate_mode, 2, 3);
+
+   cl[ 3] = __gen_uint(values->r_b_swap, 4, 4) |
+            __gen_uint(values->channel_reverse, 3, 3) |
+            __gen_uint(values->input_image_format, 4, 9) >> 8;
+
+   cl[ 4] = __gen_uint(values->height_in_ub_or_stride, 4, 23);
+
+   cl[ 5] = __gen_uint(values->height_in_ub_or_stride, 4, 23) >> 8;
+
+   cl[ 6] = __gen_uint(values->height_in_ub_or_stride, 4, 23) >> 16;
+
+   cl[ 7] = __gen_uint(values->height, 0, 15);
+
+   cl[ 8] = __gen_uint(values->height, 0, 15) >> 8;
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 9] = __gen_address_offset(&values->address);
+
+   cl[10] = __gen_address_offset(&values->address) >> 8;
+
+   cl[11] = __gen_address_offset(&values->address) >> 16;
+
+   cl[12] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D42_LOAD_TILE_BUFFER_GENERAL_length     13
+#ifdef __gen_unpack_address
+static inline void
+V3D42_LOAD_TILE_BUFFER_GENERAL_unpack(const uint8_t * restrict cl,
+                                      struct V3D42_LOAD_TILE_BUFFER_GENERAL * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 72, 103);
+   values->height = __gen_unpack_uint(cl, 56, 71);
+   values->height_in_ub_or_stride = __gen_unpack_uint(cl, 36, 55);
+   values->r_b_swap = __gen_unpack_uint(cl, 28, 28);
+   values->channel_reverse = __gen_unpack_uint(cl, 27, 27);
+   values->input_image_format = __gen_unpack_uint(cl, 20, 25);
+   values->decimate_mode = __gen_unpack_uint(cl, 18, 19);
+   values->flip_y = __gen_unpack_uint(cl, 15, 15);
+   values->memory_format = __gen_unpack_uint(cl, 12, 14);
+   values->buffer_to_load = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_opcode     31
+#define V3D42_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_header\
+   .opcode                              =     31
+
+struct V3D42_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                              const struct V3D42_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT_unpack(const uint8_t * restrict cl,
+                                                struct V3D42_TRANSFORM_FEEDBACK_FLUSH_AND_COUNT * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_INDEXED_PRIM_LIST_opcode        32
+#define V3D42_INDEXED_PRIM_LIST_header          \
+   .opcode                              =     32
+
+struct V3D42_INDEXED_PRIM_LIST {
+   uint32_t                             opcode;
+   uint32_t                             index_offset;
+   bool                                 enable_primitive_restarts;
+   uint32_t                             length;
+   uint32_t                             index_type;
+#define INDEX_TYPE_8_BIT                         0
+#define INDEX_TYPE_16_BIT                        1
+#define INDEX_TYPE_32_BIT                        2
+   enum V3D42_Primitive                 mode;
+};
+
+static inline void
+V3D42_INDEXED_PRIM_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                             const struct V3D42_INDEXED_PRIM_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->index_type, 6, 7) |
+            __gen_uint(values->mode, 0, 5);
+
+   cl[ 2] = __gen_uint(values->length, 0, 30);
+
+   cl[ 3] = __gen_uint(values->length, 0, 30) >> 8;
+
+   cl[ 4] = __gen_uint(values->length, 0, 30) >> 16;
+
+   cl[ 5] = __gen_uint(values->enable_primitive_restarts, 7, 7) |
+            __gen_uint(values->length, 0, 30) >> 24;
+
+
+   memcpy(&cl[6], &values->index_offset, sizeof(values->index_offset));
+}
+
+#define V3D42_INDEXED_PRIM_LIST_length        10
+#ifdef __gen_unpack_address
+static inline void
+V3D42_INDEXED_PRIM_LIST_unpack(const uint8_t * restrict cl,
+                               struct V3D42_INDEXED_PRIM_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->index_offset = __gen_unpack_uint(cl, 48, 79);
+   values->enable_primitive_restarts = __gen_unpack_uint(cl, 47, 47);
+   values->length = __gen_unpack_uint(cl, 16, 46);
+   values->index_type = __gen_unpack_uint(cl, 14, 15);
+   values->mode = __gen_unpack_uint(cl, 8, 13);
+}
+#endif
+
+
+#define V3D42_INDEXED_INSTANCED_PRIM_LIST_opcode     34
+#define V3D42_INDEXED_INSTANCED_PRIM_LIST_header\
+   .opcode                              =     34
+
+struct V3D42_INDEXED_INSTANCED_PRIM_LIST {
+   uint32_t                             opcode;
+   uint32_t                             index_offset;
+   uint32_t                             number_of_instances;
+   bool                                 enable_primitive_restarts;
+   uint32_t                             instance_length;
+   uint32_t                             index_type;
+#define INDEX_TYPE_8_BIT                         0
+#define INDEX_TYPE_16_BIT                        1
+#define INDEX_TYPE_32_BIT                        2
+   enum V3D42_Primitive                 mode;
+};
+
+static inline void
+V3D42_INDEXED_INSTANCED_PRIM_LIST_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                       const struct V3D42_INDEXED_INSTANCED_PRIM_LIST * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->index_type, 6, 7) |
+            __gen_uint(values->mode, 0, 5);
+
+   cl[ 2] = __gen_uint(values->instance_length, 0, 30);
+
+   cl[ 3] = __gen_uint(values->instance_length, 0, 30) >> 8;
+
+   cl[ 4] = __gen_uint(values->instance_length, 0, 30) >> 16;
+
+   cl[ 5] = __gen_uint(values->enable_primitive_restarts, 7, 7) |
+            __gen_uint(values->instance_length, 0, 30) >> 24;
+
+
+   memcpy(&cl[6], &values->number_of_instances, sizeof(values->number_of_instances));
+
+   memcpy(&cl[10], &values->index_offset, sizeof(values->index_offset));
+}
+
+#define V3D42_INDEXED_INSTANCED_PRIM_LIST_length     14
+#ifdef __gen_unpack_address
+static inline void
+V3D42_INDEXED_INSTANCED_PRIM_LIST_unpack(const uint8_t * restrict cl,
+                                         struct V3D42_INDEXED_INSTANCED_PRIM_LIST * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->index_offset = __gen_unpack_uint(cl, 80, 111);
+   values->number_of_instances = __gen_unpack_uint(cl, 48, 79);
+   values->enable_primitive_restarts = __gen_unpack_uint(cl, 47, 47);
+   values->instance_length = __gen_unpack_uint(cl, 16, 46);
+   values->index_type = __gen_unpack_uint(cl, 14, 15);
+   values->mode = __gen_unpack_uint(cl, 8, 13);
+}
+#endif
+
+
+#define V3D42_VERTEX_ARRAY_PRIMS_opcode       36
+#define V3D42_VERTEX_ARRAY_PRIMS_header         \
+   .opcode                              =     36
+
+struct V3D42_VERTEX_ARRAY_PRIMS {
+   uint32_t                             opcode;
+   uint32_t                             index_of_first_vertex;
+   uint32_t                             length;
+   enum V3D42_Primitive                 mode;
+};
+
+static inline void
+V3D42_VERTEX_ARRAY_PRIMS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D42_VERTEX_ARRAY_PRIMS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->mode, 0, 7);
+
+
+   memcpy(&cl[2], &values->length, sizeof(values->length));
+
+   memcpy(&cl[6], &values->index_of_first_vertex, sizeof(values->index_of_first_vertex));
+}
+
+#define V3D42_VERTEX_ARRAY_PRIMS_length       10
+#ifdef __gen_unpack_address
+static inline void
+V3D42_VERTEX_ARRAY_PRIMS_unpack(const uint8_t * restrict cl,
+                                struct V3D42_VERTEX_ARRAY_PRIMS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->index_of_first_vertex = __gen_unpack_uint(cl, 48, 79);
+   values->length = __gen_unpack_uint(cl, 16, 47);
+   values->mode = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D42_VERTEX_ARRAY_INSTANCED_PRIMS_opcode     38
+#define V3D42_VERTEX_ARRAY_INSTANCED_PRIMS_header\
+   .opcode                              =     38
+
+struct V3D42_VERTEX_ARRAY_INSTANCED_PRIMS {
+   uint32_t                             opcode;
+   uint32_t                             index_of_first_vertex;
+   uint32_t                             number_of_instances;
+   uint32_t                             instance_length;
+   enum V3D42_Primitive                 mode;
+};
+
+static inline void
+V3D42_VERTEX_ARRAY_INSTANCED_PRIMS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D42_VERTEX_ARRAY_INSTANCED_PRIMS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->mode, 0, 7);
+
+
+   memcpy(&cl[2], &values->instance_length, sizeof(values->instance_length));
+
+   memcpy(&cl[6], &values->number_of_instances, sizeof(values->number_of_instances));
+
+   memcpy(&cl[10], &values->index_of_first_vertex, sizeof(values->index_of_first_vertex));
+}
+
+#define V3D42_VERTEX_ARRAY_INSTANCED_PRIMS_length     14
+#ifdef __gen_unpack_address
+static inline void
+V3D42_VERTEX_ARRAY_INSTANCED_PRIMS_unpack(const uint8_t * restrict cl,
+                                          struct V3D42_VERTEX_ARRAY_INSTANCED_PRIMS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->index_of_first_vertex = __gen_unpack_uint(cl, 80, 111);
+   values->number_of_instances = __gen_unpack_uint(cl, 48, 79);
+   values->instance_length = __gen_unpack_uint(cl, 16, 47);
+   values->mode = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D42_BASE_VERTEX_BASE_INSTANCE_opcode     43
+#define V3D42_BASE_VERTEX_BASE_INSTANCE_header  \
+   .opcode                              =     43
+
+struct V3D42_BASE_VERTEX_BASE_INSTANCE {
+   uint32_t                             opcode;
+   uint32_t                             base_instance;
+   uint32_t                             base_vertex;
+};
+
+static inline void
+V3D42_BASE_VERTEX_BASE_INSTANCE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                     const struct V3D42_BASE_VERTEX_BASE_INSTANCE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->base_vertex, sizeof(values->base_vertex));
+
+   memcpy(&cl[5], &values->base_instance, sizeof(values->base_instance));
+}
+
+#define V3D42_BASE_VERTEX_BASE_INSTANCE_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_BASE_VERTEX_BASE_INSTANCE_unpack(const uint8_t * restrict cl,
+                                       struct V3D42_BASE_VERTEX_BASE_INSTANCE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->base_instance = __gen_unpack_uint(cl, 40, 71);
+   values->base_vertex = __gen_unpack_uint(cl, 8, 39);
+}
+#endif
+
+
+#define V3D42_INDEX_BUFFER_SETUP_opcode       44
+#define V3D42_INDEX_BUFFER_SETUP_header         \
+   .opcode                              =     44
+
+struct V3D42_INDEX_BUFFER_SETUP {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+   uint32_t                             size;
+};
+
+static inline void
+V3D42_INDEX_BUFFER_SETUP_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D42_INDEX_BUFFER_SETUP * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+
+   memcpy(&cl[5], &values->size, sizeof(values->size));
+}
+
+#define V3D42_INDEX_BUFFER_SETUP_length        9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_INDEX_BUFFER_SETUP_unpack(const uint8_t * restrict cl,
+                                struct V3D42_INDEX_BUFFER_SETUP * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 8, 39);
+   values->size = __gen_unpack_uint(cl, 40, 71);
+}
+#endif
+
+
+#define V3D42_PRIM_LIST_FORMAT_opcode         56
+#define V3D42_PRIM_LIST_FORMAT_header           \
+   .opcode                              =     56
+
+struct V3D42_PRIM_LIST_FORMAT {
+   uint32_t                             opcode;
+   bool                                 tri_strip_or_fan;
+   uint32_t                             primitive_type;
+#define LIST_POINTS                              0
+#define LIST_LINES                               1
+#define LIST_TRIANGLES                           2
+};
+
+static inline void
+V3D42_PRIM_LIST_FORMAT_pack(__gen_user_data *data, uint8_t * restrict cl,
+                            const struct V3D42_PRIM_LIST_FORMAT * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->tri_strip_or_fan, 7, 7) |
+            __gen_uint(values->primitive_type, 0, 5);
+
+}
+
+#define V3D42_PRIM_LIST_FORMAT_length          2
+#ifdef __gen_unpack_address
+static inline void
+V3D42_PRIM_LIST_FORMAT_unpack(const uint8_t * restrict cl,
+                              struct V3D42_PRIM_LIST_FORMAT * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->tri_strip_or_fan = __gen_unpack_uint(cl, 15, 15);
+   values->primitive_type = __gen_unpack_uint(cl, 8, 13);
+}
+#endif
+
+
+#define V3D42_GL_SHADER_STATE_opcode          64
+#define V3D42_GL_SHADER_STATE_header            \
+   .opcode                              =     64
+
+struct V3D42_GL_SHADER_STATE {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+   uint32_t                             number_of_attribute_arrays;
+};
+
+static inline void
+V3D42_GL_SHADER_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                           const struct V3D42_GL_SHADER_STATE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address) |
+            __gen_uint(values->number_of_attribute_arrays, 0, 4);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D42_GL_SHADER_STATE_length           5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_GL_SHADER_STATE_unpack(const uint8_t * restrict cl,
+                             struct V3D42_GL_SHADER_STATE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 13, 39);
+   values->number_of_attribute_arrays = __gen_unpack_uint(cl, 8, 12);
+}
+#endif
+
+
+#define V3D42_VCM_CACHE_SIZE_opcode           71
+#define V3D42_VCM_CACHE_SIZE_header             \
+   .opcode                              =     71
+
+struct V3D42_VCM_CACHE_SIZE {
+   uint32_t                             opcode;
+   uint32_t                             number_of_16_vertex_batches_for_rendering;
+   uint32_t                             number_of_16_vertex_batches_for_binning;
+};
+
+static inline void
+V3D42_VCM_CACHE_SIZE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                          const struct V3D42_VCM_CACHE_SIZE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->number_of_16_vertex_batches_for_rendering, 4, 7) |
+            __gen_uint(values->number_of_16_vertex_batches_for_binning, 0, 3);
+
+}
+
+#define V3D42_VCM_CACHE_SIZE_length            2
+#ifdef __gen_unpack_address
+static inline void
+V3D42_VCM_CACHE_SIZE_unpack(const uint8_t * restrict cl,
+                            struct V3D42_VCM_CACHE_SIZE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->number_of_16_vertex_batches_for_rendering = __gen_unpack_uint(cl, 12, 15);
+   values->number_of_16_vertex_batches_for_binning = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_TRANSFORM_FEEDBACK_BUFFER_opcode     73
+#define V3D42_TRANSFORM_FEEDBACK_BUFFER_header  \
+   .opcode                              =     73
+
+struct V3D42_TRANSFORM_FEEDBACK_BUFFER {
+   uint32_t                             opcode;
+   __gen_address_type                   buffer_address;
+   uint32_t                             buffer_size_in_32_bit_words;
+   uint32_t                             buffer_number;
+};
+
+static inline void
+V3D42_TRANSFORM_FEEDBACK_BUFFER_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                     const struct V3D42_TRANSFORM_FEEDBACK_BUFFER * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->buffer_size_in_32_bit_words, 2, 31) |
+            __gen_uint(values->buffer_number, 0, 1);
+
+   cl[ 2] = __gen_uint(values->buffer_size_in_32_bit_words, 2, 31) >> 8;
+
+   cl[ 3] = __gen_uint(values->buffer_size_in_32_bit_words, 2, 31) >> 16;
+
+   cl[ 4] = __gen_uint(values->buffer_size_in_32_bit_words, 2, 31) >> 24;
+
+   __gen_emit_reloc(data, &values->buffer_address);
+   cl[ 5] = __gen_address_offset(&values->buffer_address);
+
+   cl[ 6] = __gen_address_offset(&values->buffer_address) >> 8;
+
+   cl[ 7] = __gen_address_offset(&values->buffer_address) >> 16;
+
+   cl[ 8] = __gen_address_offset(&values->buffer_address) >> 24;
+
+}
+
+#define V3D42_TRANSFORM_FEEDBACK_BUFFER_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TRANSFORM_FEEDBACK_BUFFER_unpack(const uint8_t * restrict cl,
+                                       struct V3D42_TRANSFORM_FEEDBACK_BUFFER * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->buffer_address = __gen_unpack_address(cl, 40, 71);
+   values->buffer_size_in_32_bit_words = __gen_unpack_uint(cl, 10, 39);
+   values->buffer_number = __gen_unpack_uint(cl, 8, 9);
+}
+#endif
+
+
+#define V3D42_TRANSFORM_FEEDBACK_SPECS_opcode     74
+#define V3D42_TRANSFORM_FEEDBACK_SPECS_header   \
+   .opcode                              =     74
+
+struct V3D42_TRANSFORM_FEEDBACK_SPECS {
+   uint32_t                             opcode;
+   bool                                 enable;
+   uint32_t                             number_of_16_bit_output_data_specs_following;
+};
+
+static inline void
+V3D42_TRANSFORM_FEEDBACK_SPECS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                    const struct V3D42_TRANSFORM_FEEDBACK_SPECS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->enable, 7, 7) |
+            __gen_uint(values->number_of_16_bit_output_data_specs_following, 0, 4);
+
+}
+
+#define V3D42_TRANSFORM_FEEDBACK_SPECS_length      2
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TRANSFORM_FEEDBACK_SPECS_unpack(const uint8_t * restrict cl,
+                                      struct V3D42_TRANSFORM_FEEDBACK_SPECS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->enable = __gen_unpack_uint(cl, 15, 15);
+   values->number_of_16_bit_output_data_specs_following = __gen_unpack_uint(cl, 8, 12);
+}
+#endif
+
+
+#define V3D42_FLUSH_TRANSFORM_FEEDBACK_DATA_opcode     75
+#define V3D42_FLUSH_TRANSFORM_FEEDBACK_DATA_header\
+   .opcode                              =     75
+
+struct V3D42_FLUSH_TRANSFORM_FEEDBACK_DATA {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_FLUSH_TRANSFORM_FEEDBACK_DATA_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                         const struct V3D42_FLUSH_TRANSFORM_FEEDBACK_DATA * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_FLUSH_TRANSFORM_FEEDBACK_DATA_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_FLUSH_TRANSFORM_FEEDBACK_DATA_unpack(const uint8_t * restrict cl,
+                                           struct V3D42_FLUSH_TRANSFORM_FEEDBACK_DATA * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC_header\
+
+
+struct V3D42_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC {
+   uint32_t                             first_shaded_vertex_value_to_output;
+   uint32_t                             number_of_consecutive_vertex_values_to_output_as_32_bit_values;
+   uint32_t                             output_buffer_to_write_to;
+   uint32_t                             stream_number;
+};
+
+static inline void
+V3D42_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                               const struct V3D42_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC * restrict values)
+{
+   assert(values->number_of_consecutive_vertex_values_to_output_as_32_bit_values >= 1);
+   cl[ 0] = __gen_uint(values->first_shaded_vertex_value_to_output, 0, 7);
+
+   cl[ 1] = __gen_uint(values->number_of_consecutive_vertex_values_to_output_as_32_bit_values - 1, 0, 3) |
+            __gen_uint(values->output_buffer_to_write_to, 4, 5) |
+            __gen_uint(values->stream_number, 6, 7);
+
+}
+
+#define V3D42_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC_length      2
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC_unpack(const uint8_t * restrict cl,
+                                                 struct V3D42_TRANSFORM_FEEDBACK_OUTPUT_DATA_SPEC * restrict values)
+{
+   values->first_shaded_vertex_value_to_output = __gen_unpack_uint(cl, 0, 7);
+   values->number_of_consecutive_vertex_values_to_output_as_32_bit_values = __gen_unpack_uint(cl, 8, 11) + 1;
+   values->output_buffer_to_write_to = __gen_unpack_uint(cl, 12, 13);
+   values->stream_number = __gen_unpack_uint(cl, 14, 15);
+}
+#endif
+
+
+#define V3D42_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS_header\
+
+
+struct V3D42_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS {
+   __gen_address_type                   address;
+};
+
+static inline void
+V3D42_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                             const struct V3D42_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS * restrict values)
+{
+   __gen_emit_reloc(data, &values->address);
+   cl[ 0] = __gen_address_offset(&values->address);
+
+   cl[ 1] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D42_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS_unpack(const uint8_t * restrict cl,
+                                               struct V3D42_TRANSFORM_FEEDBACK_OUTPUT_ADDRESS * restrict values)
+{
+   values->address = __gen_unpack_address(cl, 0, 31);
+}
+#endif
+
+
+#define V3D42_STENCIL_CFG_opcode              80
+#define V3D42_STENCIL_CFG_header                \
+   .opcode                              =     80
+
+struct V3D42_STENCIL_CFG {
+   uint32_t                             opcode;
+   uint32_t                             stencil_write_mask;
+   bool                                 back_config;
+   bool                                 front_config;
+   enum V3D42_Stencil_Op                stencil_pass_op;
+   enum V3D42_Stencil_Op                depth_test_fail_op;
+   enum V3D42_Stencil_Op                stencil_test_fail_op;
+   enum V3D42_Compare_Function          stencil_test_function;
+   uint32_t                             stencil_test_mask;
+   uint32_t                             stencil_ref_value;
+};
+
+static inline void
+V3D42_STENCIL_CFG_pack(__gen_user_data *data, uint8_t * restrict cl,
+                       const struct V3D42_STENCIL_CFG * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->stencil_ref_value, 0, 7);
+
+   cl[ 2] = __gen_uint(values->stencil_test_mask, 0, 7);
+
+   cl[ 3] = __gen_uint(values->depth_test_fail_op, 6, 8) |
+            __gen_uint(values->stencil_test_fail_op, 3, 5) |
+            __gen_uint(values->stencil_test_function, 0, 2);
+
+   cl[ 4] = __gen_uint(values->back_config, 5, 5) |
+            __gen_uint(values->front_config, 4, 4) |
+            __gen_uint(values->stencil_pass_op, 1, 3) |
+            __gen_uint(values->depth_test_fail_op, 6, 8) >> 8;
+
+   cl[ 5] = __gen_uint(values->stencil_write_mask, 0, 7);
+
+}
+
+#define V3D42_STENCIL_CFG_length               6
+#ifdef __gen_unpack_address
+static inline void
+V3D42_STENCIL_CFG_unpack(const uint8_t * restrict cl,
+                         struct V3D42_STENCIL_CFG * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->stencil_write_mask = __gen_unpack_uint(cl, 40, 47);
+   values->back_config = __gen_unpack_uint(cl, 37, 37);
+   values->front_config = __gen_unpack_uint(cl, 36, 36);
+   values->stencil_pass_op = __gen_unpack_uint(cl, 33, 35);
+   values->depth_test_fail_op = __gen_unpack_uint(cl, 30, 32);
+   values->stencil_test_fail_op = __gen_unpack_uint(cl, 27, 29);
+   values->stencil_test_function = __gen_unpack_uint(cl, 24, 26);
+   values->stencil_test_mask = __gen_unpack_uint(cl, 16, 23);
+   values->stencil_ref_value = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D42_BLEND_ENABLES_opcode            83
+#define V3D42_BLEND_ENABLES_header              \
+   .opcode                              =     83
+
+struct V3D42_BLEND_ENABLES {
+   uint32_t                             opcode;
+   uint32_t                             mask;
+};
+
+static inline void
+V3D42_BLEND_ENABLES_pack(__gen_user_data *data, uint8_t * restrict cl,
+                         const struct V3D42_BLEND_ENABLES * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->mask, 0, 7);
+
+}
+
+#define V3D42_BLEND_ENABLES_length             2
+#ifdef __gen_unpack_address
+static inline void
+V3D42_BLEND_ENABLES_unpack(const uint8_t * restrict cl,
+                           struct V3D42_BLEND_ENABLES * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->mask = __gen_unpack_uint(cl, 8, 15);
+}
+#endif
+
+
+#define V3D42_BLEND_CFG_opcode                84
+#define V3D42_BLEND_CFG_header                  \
+   .opcode                              =     84
+
+struct V3D42_BLEND_CFG {
+   uint32_t                             opcode;
+   uint32_t                             render_target_mask;
+   enum V3D42_Blend_Factor              color_blend_dst_factor;
+   enum V3D42_Blend_Factor              color_blend_src_factor;
+   enum V3D42_Blend_Mode                color_blend_mode;
+   enum V3D42_Blend_Factor              alpha_blend_dst_factor;
+   enum V3D42_Blend_Factor              alpha_blend_src_factor;
+   enum V3D42_Blend_Mode                alpha_blend_mode;
+};
+
+static inline void
+V3D42_BLEND_CFG_pack(__gen_user_data *data, uint8_t * restrict cl,
+                     const struct V3D42_BLEND_CFG * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->alpha_blend_src_factor, 4, 7) |
+            __gen_uint(values->alpha_blend_mode, 0, 3);
+
+   cl[ 2] = __gen_uint(values->color_blend_mode, 4, 7) |
+            __gen_uint(values->alpha_blend_dst_factor, 0, 3);
+
+   cl[ 3] = __gen_uint(values->color_blend_dst_factor, 4, 7) |
+            __gen_uint(values->color_blend_src_factor, 0, 3);
+
+   cl[ 4] = __gen_uint(values->render_target_mask, 0, 3);
+
+}
+
+#define V3D42_BLEND_CFG_length                 5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_BLEND_CFG_unpack(const uint8_t * restrict cl,
+                       struct V3D42_BLEND_CFG * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->render_target_mask = __gen_unpack_uint(cl, 32, 35);
+   values->color_blend_dst_factor = __gen_unpack_uint(cl, 28, 31);
+   values->color_blend_src_factor = __gen_unpack_uint(cl, 24, 27);
+   values->color_blend_mode = __gen_unpack_uint(cl, 20, 23);
+   values->alpha_blend_dst_factor = __gen_unpack_uint(cl, 16, 19);
+   values->alpha_blend_src_factor = __gen_unpack_uint(cl, 12, 15);
+   values->alpha_blend_mode = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_BLEND_CONSTANT_COLOR_opcode     86
+#define V3D42_BLEND_CONSTANT_COLOR_header       \
+   .opcode                              =     86
+
+struct V3D42_BLEND_CONSTANT_COLOR {
+   uint32_t                             opcode;
+   uint32_t                             alpha_f16;
+   uint32_t                             blue_f16;
+   uint32_t                             green_f16;
+   uint32_t                             red_f16;
+};
+
+static inline void
+V3D42_BLEND_CONSTANT_COLOR_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                const struct V3D42_BLEND_CONSTANT_COLOR * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->red_f16, 0, 15);
+
+   cl[ 2] = __gen_uint(values->red_f16, 0, 15) >> 8;
+
+   cl[ 3] = __gen_uint(values->green_f16, 0, 15);
+
+   cl[ 4] = __gen_uint(values->green_f16, 0, 15) >> 8;
+
+   cl[ 5] = __gen_uint(values->blue_f16, 0, 15);
+
+   cl[ 6] = __gen_uint(values->blue_f16, 0, 15) >> 8;
+
+   cl[ 7] = __gen_uint(values->alpha_f16, 0, 15);
+
+   cl[ 8] = __gen_uint(values->alpha_f16, 0, 15) >> 8;
+
+}
+
+#define V3D42_BLEND_CONSTANT_COLOR_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_BLEND_CONSTANT_COLOR_unpack(const uint8_t * restrict cl,
+                                  struct V3D42_BLEND_CONSTANT_COLOR * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->alpha_f16 = __gen_unpack_uint(cl, 56, 71);
+   values->blue_f16 = __gen_unpack_uint(cl, 40, 55);
+   values->green_f16 = __gen_unpack_uint(cl, 24, 39);
+   values->red_f16 = __gen_unpack_uint(cl, 8, 23);
+}
+#endif
+
+
+#define V3D42_COLOR_WRITE_MASKS_opcode        87
+#define V3D42_COLOR_WRITE_MASKS_header          \
+   .opcode                              =     87
+
+struct V3D42_COLOR_WRITE_MASKS {
+   uint32_t                             opcode;
+   uint32_t                             mask;
+};
+
+static inline void
+V3D42_COLOR_WRITE_MASKS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                             const struct V3D42_COLOR_WRITE_MASKS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->mask, sizeof(values->mask));
+}
+
+#define V3D42_COLOR_WRITE_MASKS_length         5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_COLOR_WRITE_MASKS_unpack(const uint8_t * restrict cl,
+                               struct V3D42_COLOR_WRITE_MASKS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->mask = __gen_unpack_uint(cl, 8, 39);
+}
+#endif
+
+
+#define V3D42_ZERO_ALL_CENTROID_FLAGS_opcode     88
+#define V3D42_ZERO_ALL_CENTROID_FLAGS_header    \
+   .opcode                              =     88
+
+struct V3D42_ZERO_ALL_CENTROID_FLAGS {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_ZERO_ALL_CENTROID_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                   const struct V3D42_ZERO_ALL_CENTROID_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_ZERO_ALL_CENTROID_FLAGS_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_ZERO_ALL_CENTROID_FLAGS_unpack(const uint8_t * restrict cl,
+                                     struct V3D42_ZERO_ALL_CENTROID_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_CENTROID_FLAGS_opcode           89
+#define V3D42_CENTROID_FLAGS_header             \
+   .opcode                              =     89
+
+struct V3D42_CENTROID_FLAGS {
+   uint32_t                             opcode;
+   uint32_t                             centroid_flags_for_varyings_v024;
+   enum V3D42_Varying_Flags_Action      action_for_centroid_flags_of_higher_numbered_varyings;
+   enum V3D42_Varying_Flags_Action      action_for_centroid_flags_of_lower_numbered_varyings;
+   uint32_t                             varying_offset_v0;
+};
+
+static inline void
+V3D42_CENTROID_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                          const struct V3D42_CENTROID_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->action_for_centroid_flags_of_higher_numbered_varyings, 6, 7) |
+            __gen_uint(values->action_for_centroid_flags_of_lower_numbered_varyings, 4, 5) |
+            __gen_uint(values->varying_offset_v0, 0, 3);
+
+   cl[ 2] = __gen_uint(values->centroid_flags_for_varyings_v024, 0, 23);
+
+   cl[ 3] = __gen_uint(values->centroid_flags_for_varyings_v024, 0, 23) >> 8;
+
+   cl[ 4] = __gen_uint(values->centroid_flags_for_varyings_v024, 0, 23) >> 16;
+
+}
+
+#define V3D42_CENTROID_FLAGS_length            5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_CENTROID_FLAGS_unpack(const uint8_t * restrict cl,
+                            struct V3D42_CENTROID_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->centroid_flags_for_varyings_v024 = __gen_unpack_uint(cl, 16, 39);
+   values->action_for_centroid_flags_of_higher_numbered_varyings = __gen_unpack_uint(cl, 14, 15);
+   values->action_for_centroid_flags_of_lower_numbered_varyings = __gen_unpack_uint(cl, 12, 13);
+   values->varying_offset_v0 = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_SAMPLE_STATE_opcode             91
+#define V3D42_SAMPLE_STATE_header               \
+   .opcode                              =     91
+
+struct V3D42_SAMPLE_STATE {
+   uint32_t                             opcode;
+   float                                coverage;
+   uint32_t                             mask;
+};
+
+static inline void
+V3D42_SAMPLE_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                        const struct V3D42_SAMPLE_STATE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->mask, 0, 3);
+
+   cl[ 2] = 0;
+   cl[ 3] = __gen_uint(fui(values->coverage) >> 16, 0, 15);
+
+   cl[ 4] = __gen_uint(fui(values->coverage) >> 16, 0, 15) >> 8;
+
+}
+
+#define V3D42_SAMPLE_STATE_length              5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_SAMPLE_STATE_unpack(const uint8_t * restrict cl,
+                          struct V3D42_SAMPLE_STATE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->coverage = __gen_unpack_f187(cl, 24, 39);
+   values->mask = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_OCCLUSION_QUERY_COUNTER_opcode     92
+#define V3D42_OCCLUSION_QUERY_COUNTER_header    \
+   .opcode                              =     92
+
+struct V3D42_OCCLUSION_QUERY_COUNTER {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+};
+
+static inline void
+V3D42_OCCLUSION_QUERY_COUNTER_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                   const struct V3D42_OCCLUSION_QUERY_COUNTER * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D42_OCCLUSION_QUERY_COUNTER_length      5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_OCCLUSION_QUERY_COUNTER_unpack(const uint8_t * restrict cl,
+                                     struct V3D42_OCCLUSION_QUERY_COUNTER * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 8, 39);
+}
+#endif
+
+
+#define V3D42_CFG_BITS_opcode                 96
+#define V3D42_CFG_BITS_header                   \
+   .opcode                              =     96
+
+struct V3D42_CFG_BITS {
+   uint32_t                             opcode;
+   bool                                 direct3d_provoking_vertex;
+   bool                                 direct3d_point_fill_mode;
+   bool                                 blend_enable;
+   bool                                 stencil_enable;
+   bool                                 early_z_updates_enable;
+   bool                                 early_z_enable;
+   bool                                 z_updates_enable;
+   enum V3D42_Compare_Function          depth_test_function;
+   bool                                 direct3d_wireframe_triangles_mode;
+   uint32_t                             rasterizer_oversample_mode;
+   uint32_t                             line_rasterization;
+   bool                                 enable_depth_offset;
+   bool                                 clockwise_primitives;
+   bool                                 enable_reverse_facing_primitive;
+   bool                                 enable_forward_facing_primitive;
+};
+
+static inline void
+V3D42_CFG_BITS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                    const struct V3D42_CFG_BITS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->rasterizer_oversample_mode, 6, 7) |
+            __gen_uint(values->line_rasterization, 4, 5) |
+            __gen_uint(values->enable_depth_offset, 3, 3) |
+            __gen_uint(values->clockwise_primitives, 2, 2) |
+            __gen_uint(values->enable_reverse_facing_primitive, 1, 1) |
+            __gen_uint(values->enable_forward_facing_primitive, 0, 0);
+
+   cl[ 2] = __gen_uint(values->z_updates_enable, 7, 7) |
+            __gen_uint(values->depth_test_function, 4, 6) |
+            __gen_uint(values->direct3d_wireframe_triangles_mode, 3, 3);
+
+   cl[ 3] = __gen_uint(values->direct3d_provoking_vertex, 5, 5) |
+            __gen_uint(values->direct3d_point_fill_mode, 4, 4) |
+            __gen_uint(values->blend_enable, 3, 3) |
+            __gen_uint(values->stencil_enable, 2, 2) |
+            __gen_uint(values->early_z_updates_enable, 1, 1) |
+            __gen_uint(values->early_z_enable, 0, 0);
+
+}
+
+#define V3D42_CFG_BITS_length                  4
+#ifdef __gen_unpack_address
+static inline void
+V3D42_CFG_BITS_unpack(const uint8_t * restrict cl,
+                      struct V3D42_CFG_BITS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->direct3d_provoking_vertex = __gen_unpack_uint(cl, 29, 29);
+   values->direct3d_point_fill_mode = __gen_unpack_uint(cl, 28, 28);
+   values->blend_enable = __gen_unpack_uint(cl, 27, 27);
+   values->stencil_enable = __gen_unpack_uint(cl, 26, 26);
+   values->early_z_updates_enable = __gen_unpack_uint(cl, 25, 25);
+   values->early_z_enable = __gen_unpack_uint(cl, 24, 24);
+   values->z_updates_enable = __gen_unpack_uint(cl, 23, 23);
+   values->depth_test_function = __gen_unpack_uint(cl, 20, 22);
+   values->direct3d_wireframe_triangles_mode = __gen_unpack_uint(cl, 19, 19);
+   values->rasterizer_oversample_mode = __gen_unpack_uint(cl, 14, 15);
+   values->line_rasterization = __gen_unpack_uint(cl, 12, 13);
+   values->enable_depth_offset = __gen_unpack_uint(cl, 11, 11);
+   values->clockwise_primitives = __gen_unpack_uint(cl, 10, 10);
+   values->enable_reverse_facing_primitive = __gen_unpack_uint(cl, 9, 9);
+   values->enable_forward_facing_primitive = __gen_unpack_uint(cl, 8, 8);
+}
+#endif
+
+
+#define V3D42_ZERO_ALL_FLAT_SHADE_FLAGS_opcode     97
+#define V3D42_ZERO_ALL_FLAT_SHADE_FLAGS_header  \
+   .opcode                              =     97
+
+struct V3D42_ZERO_ALL_FLAT_SHADE_FLAGS {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_ZERO_ALL_FLAT_SHADE_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                     const struct V3D42_ZERO_ALL_FLAT_SHADE_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_ZERO_ALL_FLAT_SHADE_FLAGS_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_ZERO_ALL_FLAT_SHADE_FLAGS_unpack(const uint8_t * restrict cl,
+                                       struct V3D42_ZERO_ALL_FLAT_SHADE_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_FLAT_SHADE_FLAGS_opcode         98
+#define V3D42_FLAT_SHADE_FLAGS_header           \
+   .opcode                              =     98
+
+struct V3D42_FLAT_SHADE_FLAGS {
+   uint32_t                             opcode;
+   uint32_t                             flat_shade_flags_for_varyings_v024;
+   enum V3D42_Varying_Flags_Action      action_for_flat_shade_flags_of_higher_numbered_varyings;
+   enum V3D42_Varying_Flags_Action      action_for_flat_shade_flags_of_lower_numbered_varyings;
+   uint32_t                             varying_offset_v0;
+};
+
+static inline void
+V3D42_FLAT_SHADE_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                            const struct V3D42_FLAT_SHADE_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->action_for_flat_shade_flags_of_higher_numbered_varyings, 6, 7) |
+            __gen_uint(values->action_for_flat_shade_flags_of_lower_numbered_varyings, 4, 5) |
+            __gen_uint(values->varying_offset_v0, 0, 3);
+
+   cl[ 2] = __gen_uint(values->flat_shade_flags_for_varyings_v024, 0, 23);
+
+   cl[ 3] = __gen_uint(values->flat_shade_flags_for_varyings_v024, 0, 23) >> 8;
+
+   cl[ 4] = __gen_uint(values->flat_shade_flags_for_varyings_v024, 0, 23) >> 16;
+
+}
+
+#define V3D42_FLAT_SHADE_FLAGS_length          5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_FLAT_SHADE_FLAGS_unpack(const uint8_t * restrict cl,
+                              struct V3D42_FLAT_SHADE_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->flat_shade_flags_for_varyings_v024 = __gen_unpack_uint(cl, 16, 39);
+   values->action_for_flat_shade_flags_of_higher_numbered_varyings = __gen_unpack_uint(cl, 14, 15);
+   values->action_for_flat_shade_flags_of_lower_numbered_varyings = __gen_unpack_uint(cl, 12, 13);
+   values->varying_offset_v0 = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_ZERO_ALL_NON_PERSPECTIVE_FLAGS_opcode     99
+#define V3D42_ZERO_ALL_NON_PERSPECTIVE_FLAGS_header\
+   .opcode                              =     99
+
+struct V3D42_ZERO_ALL_NON_PERSPECTIVE_FLAGS {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_ZERO_ALL_NON_PERSPECTIVE_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                          const struct V3D42_ZERO_ALL_NON_PERSPECTIVE_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_ZERO_ALL_NON_PERSPECTIVE_FLAGS_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_ZERO_ALL_NON_PERSPECTIVE_FLAGS_unpack(const uint8_t * restrict cl,
+                                            struct V3D42_ZERO_ALL_NON_PERSPECTIVE_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_NON_PERSPECTIVE_FLAGS_opcode    100
+#define V3D42_NON_PERSPECTIVE_FLAGS_header      \
+   .opcode                              =    100
+
+struct V3D42_NON_PERSPECTIVE_FLAGS {
+   uint32_t                             opcode;
+   uint32_t                             non_perspective_flags_for_varyings_v024;
+   enum V3D42_Varying_Flags_Action      action_for_non_perspective_flags_of_higher_numbered_varyings;
+   enum V3D42_Varying_Flags_Action      action_for_non_perspective_flags_of_lower_numbered_varyings;
+   uint32_t                             varying_offset_v0;
+};
+
+static inline void
+V3D42_NON_PERSPECTIVE_FLAGS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                 const struct V3D42_NON_PERSPECTIVE_FLAGS * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->action_for_non_perspective_flags_of_higher_numbered_varyings, 6, 7) |
+            __gen_uint(values->action_for_non_perspective_flags_of_lower_numbered_varyings, 4, 5) |
+            __gen_uint(values->varying_offset_v0, 0, 3);
+
+   cl[ 2] = __gen_uint(values->non_perspective_flags_for_varyings_v024, 0, 23);
+
+   cl[ 3] = __gen_uint(values->non_perspective_flags_for_varyings_v024, 0, 23) >> 8;
+
+   cl[ 4] = __gen_uint(values->non_perspective_flags_for_varyings_v024, 0, 23) >> 16;
+
+}
+
+#define V3D42_NON_PERSPECTIVE_FLAGS_length      5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_NON_PERSPECTIVE_FLAGS_unpack(const uint8_t * restrict cl,
+                                   struct V3D42_NON_PERSPECTIVE_FLAGS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->non_perspective_flags_for_varyings_v024 = __gen_unpack_uint(cl, 16, 39);
+   values->action_for_non_perspective_flags_of_higher_numbered_varyings = __gen_unpack_uint(cl, 14, 15);
+   values->action_for_non_perspective_flags_of_lower_numbered_varyings = __gen_unpack_uint(cl, 12, 13);
+   values->varying_offset_v0 = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_POINT_SIZE_opcode              104
+#define V3D42_POINT_SIZE_header                 \
+   .opcode                              =    104
+
+struct V3D42_POINT_SIZE {
+   uint32_t                             opcode;
+   float                                point_size;
+};
+
+static inline void
+V3D42_POINT_SIZE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                      const struct V3D42_POINT_SIZE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->point_size, sizeof(values->point_size));
+}
+
+#define V3D42_POINT_SIZE_length                5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_POINT_SIZE_unpack(const uint8_t * restrict cl,
+                        struct V3D42_POINT_SIZE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->point_size = __gen_unpack_float(cl, 8, 39);
+}
+#endif
+
+
+#define V3D42_LINE_WIDTH_opcode              105
+#define V3D42_LINE_WIDTH_header                 \
+   .opcode                              =    105
+
+struct V3D42_LINE_WIDTH {
+   uint32_t                             opcode;
+   float                                line_width;
+};
+
+static inline void
+V3D42_LINE_WIDTH_pack(__gen_user_data *data, uint8_t * restrict cl,
+                      const struct V3D42_LINE_WIDTH * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->line_width, sizeof(values->line_width));
+}
+
+#define V3D42_LINE_WIDTH_length                5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_LINE_WIDTH_unpack(const uint8_t * restrict cl,
+                        struct V3D42_LINE_WIDTH * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->line_width = __gen_unpack_float(cl, 8, 39);
+}
+#endif
+
+
+#define V3D42_DEPTH_OFFSET_opcode            106
+#define V3D42_DEPTH_OFFSET_header               \
+   .opcode                              =    106
+
+struct V3D42_DEPTH_OFFSET {
+   uint32_t                             opcode;
+   float                                limit;
+   float                                depth_offset_units;
+   float                                depth_offset_factor;
+};
+
+static inline void
+V3D42_DEPTH_OFFSET_pack(__gen_user_data *data, uint8_t * restrict cl,
+                        const struct V3D42_DEPTH_OFFSET * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(fui(values->depth_offset_factor) >> 16, 0, 15);
+
+   cl[ 2] = __gen_uint(fui(values->depth_offset_factor) >> 16, 0, 15) >> 8;
+
+   cl[ 3] = __gen_uint(fui(values->depth_offset_units) >> 16, 0, 15);
+
+   cl[ 4] = __gen_uint(fui(values->depth_offset_units) >> 16, 0, 15) >> 8;
+
+
+   memcpy(&cl[5], &values->limit, sizeof(values->limit));
+}
+
+#define V3D42_DEPTH_OFFSET_length              9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_DEPTH_OFFSET_unpack(const uint8_t * restrict cl,
+                          struct V3D42_DEPTH_OFFSET * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->limit = __gen_unpack_float(cl, 40, 71);
+   values->depth_offset_units = __gen_unpack_f187(cl, 24, 39);
+   values->depth_offset_factor = __gen_unpack_f187(cl, 8, 23);
+}
+#endif
+
+
+#define V3D42_CLIP_WINDOW_opcode             107
+#define V3D42_CLIP_WINDOW_header                \
+   .opcode                              =    107
+
+struct V3D42_CLIP_WINDOW {
+   uint32_t                             opcode;
+   uint32_t                             clip_window_height_in_pixels;
+   uint32_t                             clip_window_width_in_pixels;
+   uint32_t                             clip_window_bottom_pixel_coordinate;
+   uint32_t                             clip_window_left_pixel_coordinate;
+};
+
+static inline void
+V3D42_CLIP_WINDOW_pack(__gen_user_data *data, uint8_t * restrict cl,
+                       const struct V3D42_CLIP_WINDOW * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->clip_window_left_pixel_coordinate, 0, 15);
+
+   cl[ 2] = __gen_uint(values->clip_window_left_pixel_coordinate, 0, 15) >> 8;
+
+   cl[ 3] = __gen_uint(values->clip_window_bottom_pixel_coordinate, 0, 15);
+
+   cl[ 4] = __gen_uint(values->clip_window_bottom_pixel_coordinate, 0, 15) >> 8;
+
+   cl[ 5] = __gen_uint(values->clip_window_width_in_pixels, 0, 15);
+
+   cl[ 6] = __gen_uint(values->clip_window_width_in_pixels, 0, 15) >> 8;
+
+   cl[ 7] = __gen_uint(values->clip_window_height_in_pixels, 0, 15);
+
+   cl[ 8] = __gen_uint(values->clip_window_height_in_pixels, 0, 15) >> 8;
+
+}
+
+#define V3D42_CLIP_WINDOW_length               9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_CLIP_WINDOW_unpack(const uint8_t * restrict cl,
+                         struct V3D42_CLIP_WINDOW * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->clip_window_height_in_pixels = __gen_unpack_uint(cl, 56, 71);
+   values->clip_window_width_in_pixels = __gen_unpack_uint(cl, 40, 55);
+   values->clip_window_bottom_pixel_coordinate = __gen_unpack_uint(cl, 24, 39);
+   values->clip_window_left_pixel_coordinate = __gen_unpack_uint(cl, 8, 23);
+}
+#endif
+
+
+#define V3D42_VIEWPORT_OFFSET_opcode         108
+#define V3D42_VIEWPORT_OFFSET_header            \
+   .opcode                              =    108
+
+struct V3D42_VIEWPORT_OFFSET {
+   uint32_t                             opcode;
+   uint32_t                             coarse_y;
+   float                                viewport_centre_y_coordinate;
+   uint32_t                             coarse_x;
+   float                                viewport_centre_x_coordinate;
+};
+
+static inline void
+V3D42_VIEWPORT_OFFSET_pack(__gen_user_data *data, uint8_t * restrict cl,
+                           const struct V3D42_VIEWPORT_OFFSET * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_sfixed(values->viewport_centre_x_coordinate, 0, 21, 8);
+
+   cl[ 2] = __gen_sfixed(values->viewport_centre_x_coordinate, 0, 21, 8) >> 8;
+
+   cl[ 3] = __gen_uint(values->coarse_x, 6, 15) |
+            __gen_sfixed(values->viewport_centre_x_coordinate, 0, 21, 8) >> 16;
+
+   cl[ 4] = __gen_uint(values->coarse_x, 6, 15) >> 8;
+
+   cl[ 5] = __gen_sfixed(values->viewport_centre_y_coordinate, 0, 21, 8);
+
+   cl[ 6] = __gen_sfixed(values->viewport_centre_y_coordinate, 0, 21, 8) >> 8;
+
+   cl[ 7] = __gen_uint(values->coarse_y, 6, 15) |
+            __gen_sfixed(values->viewport_centre_y_coordinate, 0, 21, 8) >> 16;
+
+   cl[ 8] = __gen_uint(values->coarse_y, 6, 15) >> 8;
+
+}
+
+#define V3D42_VIEWPORT_OFFSET_length           9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_VIEWPORT_OFFSET_unpack(const uint8_t * restrict cl,
+                             struct V3D42_VIEWPORT_OFFSET * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->coarse_y = __gen_unpack_uint(cl, 62, 71);
+   values->viewport_centre_y_coordinate = __gen_unpack_sfixed(cl, 40, 61, 8);
+   values->coarse_x = __gen_unpack_uint(cl, 30, 39);
+   values->viewport_centre_x_coordinate = __gen_unpack_sfixed(cl, 8, 29, 8);
+}
+#endif
+
+
+#define V3D42_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES_opcode    109
+#define V3D42_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES_header\
+   .opcode                              =    109
+
+struct V3D42_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES {
+   uint32_t                             opcode;
+   float                                maximum_zw;
+   float                                minimum_zw;
+};
+
+static inline void
+V3D42_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                             const struct V3D42_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->minimum_zw, sizeof(values->minimum_zw));
+
+   memcpy(&cl[5], &values->maximum_zw, sizeof(values->maximum_zw));
+}
+
+#define V3D42_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES_unpack(const uint8_t * restrict cl,
+                                               struct V3D42_CLIPPER_Z_MIN_MAX_CLIPPING_PLANES * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->maximum_zw = __gen_unpack_float(cl, 40, 71);
+   values->minimum_zw = __gen_unpack_float(cl, 8, 39);
+}
+#endif
+
+
+#define V3D42_CLIPPER_XY_SCALING_opcode      110
+#define V3D42_CLIPPER_XY_SCALING_header         \
+   .opcode                              =    110
+
+struct V3D42_CLIPPER_XY_SCALING {
+   uint32_t                             opcode;
+   float                                viewport_half_height_in_1_256th_of_pixel;
+   float                                viewport_half_width_in_1_256th_of_pixel;
+};
+
+static inline void
+V3D42_CLIPPER_XY_SCALING_pack(__gen_user_data *data, uint8_t * restrict cl,
+                              const struct V3D42_CLIPPER_XY_SCALING * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->viewport_half_width_in_1_256th_of_pixel, sizeof(values->viewport_half_width_in_1_256th_of_pixel));
+
+   memcpy(&cl[5], &values->viewport_half_height_in_1_256th_of_pixel, sizeof(values->viewport_half_height_in_1_256th_of_pixel));
+}
+
+#define V3D42_CLIPPER_XY_SCALING_length        9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_CLIPPER_XY_SCALING_unpack(const uint8_t * restrict cl,
+                                struct V3D42_CLIPPER_XY_SCALING * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->viewport_half_height_in_1_256th_of_pixel = __gen_unpack_float(cl, 40, 71);
+   values->viewport_half_width_in_1_256th_of_pixel = __gen_unpack_float(cl, 8, 39);
+}
+#endif
+
+
+#define V3D42_CLIPPER_Z_SCALE_AND_OFFSET_opcode    111
+#define V3D42_CLIPPER_Z_SCALE_AND_OFFSET_header \
+   .opcode                              =    111
+
+struct V3D42_CLIPPER_Z_SCALE_AND_OFFSET {
+   uint32_t                             opcode;
+   float                                viewport_z_offset_zc_to_zs;
+   float                                viewport_z_scale_zc_to_zs;
+};
+
+static inline void
+V3D42_CLIPPER_Z_SCALE_AND_OFFSET_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                      const struct V3D42_CLIPPER_Z_SCALE_AND_OFFSET * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+
+   memcpy(&cl[1], &values->viewport_z_scale_zc_to_zs, sizeof(values->viewport_z_scale_zc_to_zs));
+
+   memcpy(&cl[5], &values->viewport_z_offset_zc_to_zs, sizeof(values->viewport_z_offset_zc_to_zs));
+}
+
+#define V3D42_CLIPPER_Z_SCALE_AND_OFFSET_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_CLIPPER_Z_SCALE_AND_OFFSET_unpack(const uint8_t * restrict cl,
+                                        struct V3D42_CLIPPER_Z_SCALE_AND_OFFSET * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->viewport_z_offset_zc_to_zs = __gen_unpack_float(cl, 40, 71);
+   values->viewport_z_scale_zc_to_zs = __gen_unpack_float(cl, 8, 39);
+}
+#endif
+
+
+#define V3D42_NUMBER_OF_LAYERS_opcode        119
+#define V3D42_NUMBER_OF_LAYERS_header           \
+   .opcode                              =    119
+
+struct V3D42_NUMBER_OF_LAYERS {
+   uint32_t                             opcode;
+   uint32_t                             number_of_layers;
+};
+
+static inline void
+V3D42_NUMBER_OF_LAYERS_pack(__gen_user_data *data, uint8_t * restrict cl,
+                            const struct V3D42_NUMBER_OF_LAYERS * restrict values)
+{
+   assert(values->number_of_layers >= 1);
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->number_of_layers - 1, 0, 7);
+
+}
+
+#define V3D42_NUMBER_OF_LAYERS_length          2
+#ifdef __gen_unpack_address
+static inline void
+V3D42_NUMBER_OF_LAYERS_unpack(const uint8_t * restrict cl,
+                              struct V3D42_NUMBER_OF_LAYERS * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->number_of_layers = __gen_unpack_uint(cl, 8, 15) + 1;
+}
+#endif
+
+
+#define V3D42_TILE_BINNING_MODE_CFG_opcode    120
+#define V3D42_TILE_BINNING_MODE_CFG_header      \
+   .opcode                              =    120
+
+struct V3D42_TILE_BINNING_MODE_CFG {
+   uint32_t                             opcode;
+   uint32_t                             height_in_pixels;
+   uint32_t                             width_in_pixels;
+   bool                                 double_buffer_in_non_ms_mode;
+   bool                                 multisample_mode_4x;
+   uint32_t                             maximum_bpp_of_all_render_targets;
+#define RENDER_TARGET_MAXIMUM_32BPP              0
+#define RENDER_TARGET_MAXIMUM_64BPP              1
+#define RENDER_TARGET_MAXIMUM_128BPP             2
+   uint32_t                             number_of_render_targets;
+   uint32_t                             tile_allocation_block_size;
+#define TILE_ALLOCATION_BLOCK_SIZE_64B           0
+#define TILE_ALLOCATION_BLOCK_SIZE_128B          1
+#define TILE_ALLOCATION_BLOCK_SIZE_256B          2
+   uint32_t                             tile_allocation_initial_block_size;
+#define TILE_ALLOCATION_INITIAL_BLOCK_SIZE_64B   0
+#define TILE_ALLOCATION_INITIAL_BLOCK_SIZE_128B  1
+#define TILE_ALLOCATION_INITIAL_BLOCK_SIZE_256B  2
+};
+
+static inline void
+V3D42_TILE_BINNING_MODE_CFG_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                 const struct V3D42_TILE_BINNING_MODE_CFG * restrict values)
+{
+   assert(values->height_in_pixels >= 1);
+   assert(values->width_in_pixels >= 1);
+   assert(values->number_of_render_targets >= 1);
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->tile_allocation_block_size, 4, 5) |
+            __gen_uint(values->tile_allocation_initial_block_size, 2, 3);
+
+   cl[ 2] = __gen_uint(values->double_buffer_in_non_ms_mode, 7, 7) |
+            __gen_uint(values->multisample_mode_4x, 6, 6) |
+            __gen_uint(values->maximum_bpp_of_all_render_targets, 4, 5) |
+            __gen_uint(values->number_of_render_targets - 1, 0, 3);
+
+   cl[ 3] = 0;
+   cl[ 4] = 0;
+   cl[ 5] = __gen_uint(values->width_in_pixels - 1, 0, 11);
+
+   cl[ 6] = __gen_uint(values->width_in_pixels - 1, 0, 11) >> 8;
+
+   cl[ 7] = __gen_uint(values->height_in_pixels - 1, 0, 11);
+
+   cl[ 8] = __gen_uint(values->height_in_pixels - 1, 0, 11) >> 8;
+
+}
+
+#define V3D42_TILE_BINNING_MODE_CFG_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TILE_BINNING_MODE_CFG_unpack(const uint8_t * restrict cl,
+                                   struct V3D42_TILE_BINNING_MODE_CFG * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->height_in_pixels = __gen_unpack_uint(cl, 56, 67) + 1;
+   values->width_in_pixels = __gen_unpack_uint(cl, 40, 51) + 1;
+   values->double_buffer_in_non_ms_mode = __gen_unpack_uint(cl, 23, 23);
+   values->multisample_mode_4x = __gen_unpack_uint(cl, 22, 22);
+   values->maximum_bpp_of_all_render_targets = __gen_unpack_uint(cl, 20, 21);
+   values->number_of_render_targets = __gen_unpack_uint(cl, 16, 19) + 1;
+   values->tile_allocation_block_size = __gen_unpack_uint(cl, 12, 13);
+   values->tile_allocation_initial_block_size = __gen_unpack_uint(cl, 10, 11);
+}
+#endif
+
+
+#define V3D42_TILE_RENDERING_MODE_CFG_COMMON_opcode    121
+#define V3D42_TILE_RENDERING_MODE_CFG_COMMON_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      0
+
+struct V3D42_TILE_RENDERING_MODE_CFG_COMMON {
+   uint32_t                             opcode;
+   uint32_t                             pad;
+   bool                                 early_depth_stencil_clear;
+   enum V3D42_Internal_Depth_Type       internal_depth_type;
+   bool                                 early_z_disable;
+   uint32_t                             early_z_test_and_update_direction;
+#define EARLY_Z_DIRECTION_LT_LE                  0
+#define EARLY_Z_DIRECTION_GT_GE                  1
+   bool                                 double_buffer_in_non_ms_mode;
+   bool                                 multisample_mode_4x;
+   enum V3D42_Internal_BPP              maximum_bpp_of_all_render_targets;
+   uint32_t                             image_height_pixels;
+   uint32_t                             image_width_pixels;
+   uint32_t                             number_of_render_targets;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_COMMON_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                          const struct V3D42_TILE_RENDERING_MODE_CFG_COMMON * restrict values)
+{
+   assert(values->number_of_render_targets >= 1);
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->number_of_render_targets - 1, 4, 7) |
+            __gen_uint(values->sub_id, 0, 3);
+
+   cl[ 2] = __gen_uint(values->image_width_pixels, 0, 15);
+
+   cl[ 3] = __gen_uint(values->image_width_pixels, 0, 15) >> 8;
+
+   cl[ 4] = __gen_uint(values->image_height_pixels, 0, 15);
+
+   cl[ 5] = __gen_uint(values->image_height_pixels, 0, 15) >> 8;
+
+   cl[ 6] = __gen_uint(values->internal_depth_type, 7, 10) |
+            __gen_uint(values->early_z_disable, 6, 6) |
+            __gen_uint(values->early_z_test_and_update_direction, 5, 5) |
+            __gen_uint(values->double_buffer_in_non_ms_mode, 3, 3) |
+            __gen_uint(values->multisample_mode_4x, 2, 2) |
+            __gen_uint(values->maximum_bpp_of_all_render_targets, 0, 1);
+
+   cl[ 7] = __gen_uint(values->pad, 4, 15) |
+            __gen_uint(values->early_depth_stencil_clear, 3, 3) |
+            __gen_uint(values->internal_depth_type, 7, 10) >> 8;
+
+   cl[ 8] = __gen_uint(values->pad, 4, 15) >> 8;
+
+}
+
+#define V3D42_TILE_RENDERING_MODE_CFG_COMMON_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_COMMON_unpack(const uint8_t * restrict cl,
+                                            struct V3D42_TILE_RENDERING_MODE_CFG_COMMON * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->pad = __gen_unpack_uint(cl, 60, 71);
+   values->early_depth_stencil_clear = __gen_unpack_uint(cl, 59, 59);
+   values->internal_depth_type = __gen_unpack_uint(cl, 55, 58);
+   values->early_z_disable = __gen_unpack_uint(cl, 54, 54);
+   values->early_z_test_and_update_direction = __gen_unpack_uint(cl, 53, 53);
+   values->double_buffer_in_non_ms_mode = __gen_unpack_uint(cl, 51, 51);
+   values->multisample_mode_4x = __gen_unpack_uint(cl, 50, 50);
+   values->maximum_bpp_of_all_render_targets = __gen_unpack_uint(cl, 48, 49);
+   values->image_height_pixels = __gen_unpack_uint(cl, 32, 47);
+   values->image_width_pixels = __gen_unpack_uint(cl, 16, 31);
+   values->number_of_render_targets = __gen_unpack_uint(cl, 12, 15) + 1;
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_TILE_RENDERING_MODE_CFG_COLOR_opcode    121
+#define V3D42_TILE_RENDERING_MODE_CFG_COLOR_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      1
+
+struct V3D42_TILE_RENDERING_MODE_CFG_COLOR {
+   uint32_t                             opcode;
+   uint32_t                             pad;
+   enum V3D42_Render_Target_Clamp       render_target_3_clamp;
+   enum V3D42_Internal_Type             render_target_3_internal_type;
+   enum V3D42_Internal_BPP              render_target_3_internal_bpp;
+   enum V3D42_Render_Target_Clamp       render_target_2_clamp;
+   enum V3D42_Internal_Type             render_target_2_internal_type;
+   enum V3D42_Internal_BPP              render_target_2_internal_bpp;
+   enum V3D42_Render_Target_Clamp       render_target_1_clamp;
+   enum V3D42_Internal_Type             render_target_1_internal_type;
+   enum V3D42_Internal_BPP              render_target_1_internal_bpp;
+   enum V3D42_Render_Target_Clamp       render_target_0_clamp;
+   enum V3D42_Internal_Type             render_target_0_internal_type;
+   enum V3D42_Internal_BPP              render_target_0_internal_bpp;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_COLOR_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                         const struct V3D42_TILE_RENDERING_MODE_CFG_COLOR * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->render_target_0_internal_type, 6, 9) |
+            __gen_uint(values->render_target_0_internal_bpp, 4, 5) |
+            __gen_uint(values->sub_id, 0, 3);
+
+   cl[ 2] = __gen_uint(values->render_target_1_internal_type, 6, 9) |
+            __gen_uint(values->render_target_1_internal_bpp, 4, 5) |
+            __gen_uint(values->render_target_0_clamp, 2, 3) |
+            __gen_uint(values->render_target_0_internal_type, 6, 9) >> 8;
+
+   cl[ 3] = __gen_uint(values->render_target_2_internal_type, 6, 9) |
+            __gen_uint(values->render_target_2_internal_bpp, 4, 5) |
+            __gen_uint(values->render_target_1_clamp, 2, 3) |
+            __gen_uint(values->render_target_1_internal_type, 6, 9) >> 8;
+
+   cl[ 4] = __gen_uint(values->render_target_3_internal_type, 6, 9) |
+            __gen_uint(values->render_target_3_internal_bpp, 4, 5) |
+            __gen_uint(values->render_target_2_clamp, 2, 3) |
+            __gen_uint(values->render_target_2_internal_type, 6, 9) >> 8;
+
+   cl[ 5] = __gen_uint(values->pad, 2, 29) |
+            __gen_uint(values->render_target_3_clamp, 0, 1) |
+            __gen_uint(values->render_target_3_internal_type, 6, 9) >> 8;
+
+   cl[ 6] = __gen_uint(values->pad, 2, 29) >> 8;
+
+   cl[ 7] = __gen_uint(values->pad, 2, 29) >> 16;
+
+   cl[ 8] = __gen_uint(values->pad, 2, 29) >> 24;
+
+}
+
+#define V3D42_TILE_RENDERING_MODE_CFG_COLOR_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_COLOR_unpack(const uint8_t * restrict cl,
+                                           struct V3D42_TILE_RENDERING_MODE_CFG_COLOR * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->pad = __gen_unpack_uint(cl, 42, 69);
+   values->render_target_3_clamp = __gen_unpack_uint(cl, 40, 41);
+   values->render_target_3_internal_type = __gen_unpack_uint(cl, 38, 41);
+   values->render_target_3_internal_bpp = __gen_unpack_uint(cl, 36, 37);
+   values->render_target_2_clamp = __gen_unpack_uint(cl, 34, 35);
+   values->render_target_2_internal_type = __gen_unpack_uint(cl, 30, 33);
+   values->render_target_2_internal_bpp = __gen_unpack_uint(cl, 28, 29);
+   values->render_target_1_clamp = __gen_unpack_uint(cl, 26, 27);
+   values->render_target_1_internal_type = __gen_unpack_uint(cl, 22, 25);
+   values->render_target_1_internal_bpp = __gen_unpack_uint(cl, 20, 21);
+   values->render_target_0_clamp = __gen_unpack_uint(cl, 18, 19);
+   values->render_target_0_internal_type = __gen_unpack_uint(cl, 14, 17);
+   values->render_target_0_internal_bpp = __gen_unpack_uint(cl, 12, 13);
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_opcode    121
+#define V3D42_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      2
+
+struct V3D42_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES {
+   uint32_t                             opcode;
+   uint32_t                             unused;
+   float                                z_clear_value;
+   uint32_t                             stencil_clear_value;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                   const struct V3D42_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->sub_id, 0, 3);
+
+   cl[ 2] = __gen_uint(values->stencil_clear_value, 0, 7);
+
+
+   memcpy(&cl[3], &values->z_clear_value, sizeof(values->z_clear_value));
+   cl[ 7] = __gen_uint(values->unused, 0, 15);
+
+   cl[ 8] = __gen_uint(values->unused, 0, 15) >> 8;
+
+}
+
+#define V3D42_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES_unpack(const uint8_t * restrict cl,
+                                                     struct V3D42_TILE_RENDERING_MODE_CFG_ZS_CLEAR_VALUES * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->unused = __gen_unpack_uint(cl, 56, 71);
+   values->z_clear_value = __gen_unpack_float(cl, 24, 55);
+   values->stencil_clear_value = __gen_unpack_uint(cl, 16, 23);
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_opcode    121
+#define V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      3
+
+struct V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1 {
+   uint32_t                             opcode;
+   uint32_t                             clear_color_next_24_bits;
+   uint32_t                             clear_color_low_32_bits;
+   uint32_t                             render_target_number;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                      const struct V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1 * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->render_target_number, 4, 7) |
+            __gen_uint(values->sub_id, 0, 3);
+
+
+   memcpy(&cl[2], &values->clear_color_low_32_bits, sizeof(values->clear_color_low_32_bits));
+   cl[ 6] = __gen_uint(values->clear_color_next_24_bits, 0, 23);
+
+   cl[ 7] = __gen_uint(values->clear_color_next_24_bits, 0, 23) >> 8;
+
+   cl[ 8] = __gen_uint(values->clear_color_next_24_bits, 0, 23) >> 16;
+
+}
+
+#define V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1_unpack(const uint8_t * restrict cl,
+                                                        struct V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART1 * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->clear_color_next_24_bits = __gen_unpack_uint(cl, 48, 71);
+   values->clear_color_low_32_bits = __gen_unpack_uint(cl, 16, 47);
+   values->render_target_number = __gen_unpack_uint(cl, 12, 15);
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_opcode    121
+#define V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      4
+
+struct V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2 {
+   uint32_t                             opcode;
+   uint32_t                             clear_color_mid_high_24_bits;
+   uint32_t                             clear_color_mid_low_32_bits;
+   uint32_t                             render_target_number;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                      const struct V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2 * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->render_target_number, 4, 7) |
+            __gen_uint(values->sub_id, 0, 3);
+
+
+   memcpy(&cl[2], &values->clear_color_mid_low_32_bits, sizeof(values->clear_color_mid_low_32_bits));
+   cl[ 6] = __gen_uint(values->clear_color_mid_high_24_bits, 0, 23);
+
+   cl[ 7] = __gen_uint(values->clear_color_mid_high_24_bits, 0, 23) >> 8;
+
+   cl[ 8] = __gen_uint(values->clear_color_mid_high_24_bits, 0, 23) >> 16;
+
+}
+
+#define V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2_unpack(const uint8_t * restrict cl,
+                                                        struct V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART2 * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->clear_color_mid_high_24_bits = __gen_unpack_uint(cl, 48, 71);
+   values->clear_color_mid_low_32_bits = __gen_unpack_uint(cl, 16, 47);
+   values->render_target_number = __gen_unpack_uint(cl, 12, 15);
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_opcode    121
+#define V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_header\
+   .opcode                              =    121,  \
+   .sub_id                              =      5
+
+struct V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3 {
+   uint32_t                             opcode;
+   uint32_t                             pad;
+   uint32_t                             uif_padded_height_in_uif_blocks;
+   uint32_t                             raster_row_stride_or_image_height_in_pixels;
+   uint32_t                             clear_color_high_16_bits;
+   uint32_t                             render_target_number;
+   uint32_t                             sub_id;
+};
+
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                      const struct V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3 * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->render_target_number, 4, 7) |
+            __gen_uint(values->sub_id, 0, 3);
+
+   cl[ 2] = __gen_uint(values->clear_color_high_16_bits, 0, 15);
+
+   cl[ 3] = __gen_uint(values->clear_color_high_16_bits, 0, 15) >> 8;
+
+   cl[ 4] = __gen_uint(values->raster_row_stride_or_image_height_in_pixels, 0, 15);
+
+   cl[ 5] = __gen_uint(values->raster_row_stride_or_image_height_in_pixels, 0, 15) >> 8;
+
+   cl[ 6] = __gen_uint(values->uif_padded_height_in_uif_blocks, 0, 12);
+
+   cl[ 7] = __gen_uint(values->pad, 5, 15) |
+            __gen_uint(values->uif_padded_height_in_uif_blocks, 0, 12) >> 8;
+
+   cl[ 8] = __gen_uint(values->pad, 5, 15) >> 8;
+
+}
+
+#define V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3_unpack(const uint8_t * restrict cl,
+                                                        struct V3D42_TILE_RENDERING_MODE_CFG_CLEAR_COLORS_PART3 * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->pad = __gen_unpack_uint(cl, 61, 71);
+   values->uif_padded_height_in_uif_blocks = __gen_unpack_uint(cl, 48, 60);
+   values->raster_row_stride_or_image_height_in_pixels = __gen_unpack_uint(cl, 32, 47);
+   values->clear_color_high_16_bits = __gen_unpack_uint(cl, 16, 31);
+   values->render_target_number = __gen_unpack_uint(cl, 12, 15);
+   values->sub_id = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_TILE_COORDINATES_opcode        124
+#define V3D42_TILE_COORDINATES_header           \
+   .opcode                              =    124
+
+struct V3D42_TILE_COORDINATES {
+   uint32_t                             opcode;
+   uint32_t                             tile_row_number;
+   uint32_t                             tile_column_number;
+};
+
+static inline void
+V3D42_TILE_COORDINATES_pack(__gen_user_data *data, uint8_t * restrict cl,
+                            const struct V3D42_TILE_COORDINATES * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->tile_column_number, 0, 11);
+
+   cl[ 2] = __gen_uint(values->tile_row_number, 4, 15) |
+            __gen_uint(values->tile_column_number, 0, 11) >> 8;
+
+   cl[ 3] = __gen_uint(values->tile_row_number, 4, 15) >> 8;
+
+}
+
+#define V3D42_TILE_COORDINATES_length          4
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TILE_COORDINATES_unpack(const uint8_t * restrict cl,
+                              struct V3D42_TILE_COORDINATES * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->tile_row_number = __gen_unpack_uint(cl, 20, 31);
+   values->tile_column_number = __gen_unpack_uint(cl, 8, 19);
+}
+#endif
+
+
+#define V3D42_MULTICORE_RENDERING_SUPERTILE_CFG_opcode    122
+#define V3D42_MULTICORE_RENDERING_SUPERTILE_CFG_header\
+   .opcode                              =    122
+
+struct V3D42_MULTICORE_RENDERING_SUPERTILE_CFG {
+   uint32_t                             opcode;
+   uint32_t                             number_of_bin_tile_lists;
+   bool                                 supertile_raster_order;
+   bool                                 multicore_enable;
+   uint32_t                             total_frame_height_in_tiles;
+   uint32_t                             total_frame_width_in_tiles;
+   uint32_t                             total_frame_height_in_supertiles;
+   uint32_t                             total_frame_width_in_supertiles;
+   uint32_t                             supertile_height_in_tiles;
+   uint32_t                             supertile_width_in_tiles;
+};
+
+static inline void
+V3D42_MULTICORE_RENDERING_SUPERTILE_CFG_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                             const struct V3D42_MULTICORE_RENDERING_SUPERTILE_CFG * restrict values)
+{
+   assert(values->number_of_bin_tile_lists >= 1);
+   assert(values->supertile_height_in_tiles >= 1);
+   assert(values->supertile_width_in_tiles >= 1);
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->supertile_width_in_tiles - 1, 0, 7);
+
+   cl[ 2] = __gen_uint(values->supertile_height_in_tiles - 1, 0, 7);
+
+   cl[ 3] = __gen_uint(values->total_frame_width_in_supertiles, 0, 7);
+
+   cl[ 4] = __gen_uint(values->total_frame_height_in_supertiles, 0, 7);
+
+   cl[ 5] = __gen_uint(values->total_frame_width_in_tiles, 0, 11);
+
+   cl[ 6] = __gen_uint(values->total_frame_height_in_tiles, 4, 15) |
+            __gen_uint(values->total_frame_width_in_tiles, 0, 11) >> 8;
+
+   cl[ 7] = __gen_uint(values->total_frame_height_in_tiles, 4, 15) >> 8;
+
+   cl[ 8] = __gen_uint(values->number_of_bin_tile_lists - 1, 5, 7) |
+            __gen_uint(values->supertile_raster_order, 4, 4) |
+            __gen_uint(values->multicore_enable, 0, 0);
+
+}
+
+#define V3D42_MULTICORE_RENDERING_SUPERTILE_CFG_length      9
+#ifdef __gen_unpack_address
+static inline void
+V3D42_MULTICORE_RENDERING_SUPERTILE_CFG_unpack(const uint8_t * restrict cl,
+                                               struct V3D42_MULTICORE_RENDERING_SUPERTILE_CFG * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->number_of_bin_tile_lists = __gen_unpack_uint(cl, 69, 71) + 1;
+   values->supertile_raster_order = __gen_unpack_uint(cl, 68, 68);
+   values->multicore_enable = __gen_unpack_uint(cl, 64, 64);
+   values->total_frame_height_in_tiles = __gen_unpack_uint(cl, 52, 63);
+   values->total_frame_width_in_tiles = __gen_unpack_uint(cl, 40, 51);
+   values->total_frame_height_in_supertiles = __gen_unpack_uint(cl, 32, 39);
+   values->total_frame_width_in_supertiles = __gen_unpack_uint(cl, 24, 31);
+   values->supertile_height_in_tiles = __gen_unpack_uint(cl, 16, 23) + 1;
+   values->supertile_width_in_tiles = __gen_unpack_uint(cl, 8, 15) + 1;
+}
+#endif
+
+
+#define V3D42_MULTICORE_RENDERING_TILE_LIST_SET_BASE_opcode    123
+#define V3D42_MULTICORE_RENDERING_TILE_LIST_SET_BASE_header\
+   .opcode                              =    123
+
+struct V3D42_MULTICORE_RENDERING_TILE_LIST_SET_BASE {
+   uint32_t                             opcode;
+   __gen_address_type                   address;
+   uint32_t                             tile_list_set_number;
+};
+
+static inline void
+V3D42_MULTICORE_RENDERING_TILE_LIST_SET_BASE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                                  const struct V3D42_MULTICORE_RENDERING_TILE_LIST_SET_BASE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   __gen_emit_reloc(data, &values->address);
+   cl[ 1] = __gen_address_offset(&values->address) |
+            __gen_uint(values->tile_list_set_number, 0, 3);
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 4] = __gen_address_offset(&values->address) >> 24;
+
+}
+
+#define V3D42_MULTICORE_RENDERING_TILE_LIST_SET_BASE_length      5
+#ifdef __gen_unpack_address
+static inline void
+V3D42_MULTICORE_RENDERING_TILE_LIST_SET_BASE_unpack(const uint8_t * restrict cl,
+                                                    struct V3D42_MULTICORE_RENDERING_TILE_LIST_SET_BASE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->address = __gen_unpack_address(cl, 14, 39);
+   values->tile_list_set_number = __gen_unpack_uint(cl, 8, 11);
+}
+#endif
+
+
+#define V3D42_TILE_COORDINATES_IMPLICIT_opcode    125
+#define V3D42_TILE_COORDINATES_IMPLICIT_header  \
+   .opcode                              =    125
+
+struct V3D42_TILE_COORDINATES_IMPLICIT {
+   uint32_t                             opcode;
+};
+
+static inline void
+V3D42_TILE_COORDINATES_IMPLICIT_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                     const struct V3D42_TILE_COORDINATES_IMPLICIT * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+}
+
+#define V3D42_TILE_COORDINATES_IMPLICIT_length      1
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TILE_COORDINATES_IMPLICIT_unpack(const uint8_t * restrict cl,
+                                       struct V3D42_TILE_COORDINATES_IMPLICIT * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+}
+#endif
+
+
+#define V3D42_TILE_LIST_INITIAL_BLOCK_SIZE_opcode    126
+#define V3D42_TILE_LIST_INITIAL_BLOCK_SIZE_header\
+   .opcode                              =    126
+
+struct V3D42_TILE_LIST_INITIAL_BLOCK_SIZE {
+   uint32_t                             opcode;
+   bool                                 use_auto_chained_tile_lists;
+   uint32_t                             size_of_first_block_in_chained_tile_lists;
+#define TILE_ALLOCATION_BLOCK_SIZE_64B           0
+#define TILE_ALLOCATION_BLOCK_SIZE_128B          1
+#define TILE_ALLOCATION_BLOCK_SIZE_256B          2
+};
+
+static inline void
+V3D42_TILE_LIST_INITIAL_BLOCK_SIZE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D42_TILE_LIST_INITIAL_BLOCK_SIZE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->opcode, 0, 7);
+
+   cl[ 1] = __gen_uint(values->use_auto_chained_tile_lists, 2, 2) |
+            __gen_uint(values->size_of_first_block_in_chained_tile_lists, 0, 1);
+
+}
+
+#define V3D42_TILE_LIST_INITIAL_BLOCK_SIZE_length      2
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TILE_LIST_INITIAL_BLOCK_SIZE_unpack(const uint8_t * restrict cl,
+                                          struct V3D42_TILE_LIST_INITIAL_BLOCK_SIZE * restrict values)
+{
+   values->opcode = __gen_unpack_uint(cl, 0, 7);
+   values->use_auto_chained_tile_lists = __gen_unpack_uint(cl, 10, 10);
+   values->size_of_first_block_in_chained_tile_lists = __gen_unpack_uint(cl, 8, 9);
+}
+#endif
+
+
+#define V3D42_GL_SHADER_STATE_RECORD_header     \
+
+
+struct V3D42_GL_SHADER_STATE_RECORD {
+   bool                                 point_size_in_shaded_vertex_data;
+   bool                                 enable_clipping;
+   bool                                 vertex_id_read_by_coordinate_shader;
+   bool                                 instance_id_read_by_coordinate_shader;
+   bool                                 base_instance_id_read_by_coordinate_shader;
+   bool                                 vertex_id_read_by_vertex_shader;
+   bool                                 instance_id_read_by_vertex_shader;
+   bool                                 base_instance_id_read_by_vertex_shader;
+   bool                                 fragment_shader_does_z_writes;
+   bool                                 turn_off_early_z_test;
+   bool                                 coordinate_shader_has_separate_input_and_output_vpm_blocks;
+   bool                                 vertex_shader_has_separate_input_and_output_vpm_blocks;
+   bool                                 fragment_shader_uses_real_pixel_centre_w_in_addition_to_centroid_w2;
+   bool                                 enable_sample_rate_shading;
+   bool                                 any_shader_reads_hardware_written_primitive_id;
+   bool                                 insert_primitive_id_as_first_varying_to_fragment_shader;
+   bool                                 turn_off_scoreboard;
+   bool                                 do_scoreboard_wait_on_first_thread_switch;
+   bool                                 disable_implicit_point_line_varyings;
+   bool                                 no_prim_pack;
+   uint32_t                             number_of_varyings_in_fragment_shader;
+   uint32_t                             coordinate_shader_output_vpm_segment_size;
+   uint32_t                             min_coord_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size;
+   uint32_t                             coordinate_shader_input_vpm_segment_size;
+   uint32_t                             min_coord_shader_input_segments_required_in_play;
+   uint32_t                             vertex_shader_output_vpm_segment_size;
+   uint32_t                             min_vertex_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size;
+   uint32_t                             vertex_shader_input_vpm_segment_size;
+   uint32_t                             min_vertex_shader_input_segments_required_in_play;
+   __gen_address_type                   address_of_default_attribute_values;
+   __gen_address_type                   fragment_shader_code_address;
+   bool                                 fragment_shader_4_way_threadable;
+   bool                                 fragment_shader_start_in_final_thread_section;
+   bool                                 fragment_shader_propagate_nans;
+   __gen_address_type                   fragment_shader_uniforms_address;
+   __gen_address_type                   vertex_shader_code_address;
+   bool                                 vertex_shader_4_way_threadable;
+   bool                                 vertex_shader_start_in_final_thread_section;
+   bool                                 vertex_shader_propagate_nans;
+   __gen_address_type                   vertex_shader_uniforms_address;
+   __gen_address_type                   coordinate_shader_code_address;
+   bool                                 coordinate_shader_4_way_threadable;
+   bool                                 coordinate_shader_start_in_final_thread_section;
+   bool                                 coordinate_shader_propagate_nans;
+   __gen_address_type                   coordinate_shader_uniforms_address;
+};
+
+static inline void
+V3D42_GL_SHADER_STATE_RECORD_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                  const struct V3D42_GL_SHADER_STATE_RECORD * restrict values)
+{
+   assert(values->min_coord_shader_input_segments_required_in_play >= 1);
+   assert(values->min_vertex_shader_input_segments_required_in_play >= 1);
+   cl[ 0] = __gen_uint(values->point_size_in_shaded_vertex_data, 0, 0) |
+            __gen_uint(values->enable_clipping, 1, 1) |
+            __gen_uint(values->vertex_id_read_by_coordinate_shader, 2, 2) |
+            __gen_uint(values->instance_id_read_by_coordinate_shader, 3, 3) |
+            __gen_uint(values->base_instance_id_read_by_coordinate_shader, 4, 4) |
+            __gen_uint(values->vertex_id_read_by_vertex_shader, 5, 5) |
+            __gen_uint(values->instance_id_read_by_vertex_shader, 6, 6) |
+            __gen_uint(values->base_instance_id_read_by_vertex_shader, 7, 7);
+
+   cl[ 1] = __gen_uint(values->fragment_shader_does_z_writes, 0, 0) |
+            __gen_uint(values->turn_off_early_z_test, 1, 1) |
+            __gen_uint(values->coordinate_shader_has_separate_input_and_output_vpm_blocks, 2, 2) |
+            __gen_uint(values->vertex_shader_has_separate_input_and_output_vpm_blocks, 3, 3) |
+            __gen_uint(values->fragment_shader_uses_real_pixel_centre_w_in_addition_to_centroid_w2, 4, 4) |
+            __gen_uint(values->enable_sample_rate_shading, 5, 5) |
+            __gen_uint(values->any_shader_reads_hardware_written_primitive_id, 6, 6) |
+            __gen_uint(values->insert_primitive_id_as_first_varying_to_fragment_shader, 7, 7);
+
+   cl[ 2] = __gen_uint(values->turn_off_scoreboard, 0, 0) |
+            __gen_uint(values->do_scoreboard_wait_on_first_thread_switch, 1, 1) |
+            __gen_uint(values->disable_implicit_point_line_varyings, 2, 2) |
+            __gen_uint(values->no_prim_pack, 3, 3);
+
+   cl[ 3] = __gen_uint(values->number_of_varyings_in_fragment_shader, 0, 7);
+
+   cl[ 4] = __gen_uint(values->coordinate_shader_output_vpm_segment_size, 0, 3) |
+            __gen_uint(values->min_coord_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size, 4, 7);
+
+   cl[ 5] = __gen_uint(values->coordinate_shader_input_vpm_segment_size, 0, 3) |
+            __gen_uint(values->min_coord_shader_input_segments_required_in_play - 1, 4, 7);
+
+   cl[ 6] = __gen_uint(values->vertex_shader_output_vpm_segment_size, 0, 3) |
+            __gen_uint(values->min_vertex_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size, 4, 7);
+
+   cl[ 7] = __gen_uint(values->vertex_shader_input_vpm_segment_size, 0, 3) |
+            __gen_uint(values->min_vertex_shader_input_segments_required_in_play - 1, 4, 7);
+
+   __gen_emit_reloc(data, &values->address_of_default_attribute_values);
+   cl[ 8] = __gen_address_offset(&values->address_of_default_attribute_values);
+
+   cl[ 9] = __gen_address_offset(&values->address_of_default_attribute_values) >> 8;
+
+   cl[10] = __gen_address_offset(&values->address_of_default_attribute_values) >> 16;
+
+   cl[11] = __gen_address_offset(&values->address_of_default_attribute_values) >> 24;
+
+   __gen_emit_reloc(data, &values->fragment_shader_code_address);
+   cl[12] = __gen_address_offset(&values->fragment_shader_code_address) |
+            __gen_uint(values->fragment_shader_4_way_threadable, 0, 0) |
+            __gen_uint(values->fragment_shader_start_in_final_thread_section, 1, 1) |
+            __gen_uint(values->fragment_shader_propagate_nans, 2, 2);
+
+   cl[13] = __gen_address_offset(&values->fragment_shader_code_address) >> 8;
+
+   cl[14] = __gen_address_offset(&values->fragment_shader_code_address) >> 16;
+
+   cl[15] = __gen_address_offset(&values->fragment_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->fragment_shader_uniforms_address);
+   cl[16] = __gen_address_offset(&values->fragment_shader_uniforms_address);
+
+   cl[17] = __gen_address_offset(&values->fragment_shader_uniforms_address) >> 8;
+
+   cl[18] = __gen_address_offset(&values->fragment_shader_uniforms_address) >> 16;
+
+   cl[19] = __gen_address_offset(&values->fragment_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->vertex_shader_code_address);
+   cl[20] = __gen_address_offset(&values->vertex_shader_code_address) |
+            __gen_uint(values->vertex_shader_4_way_threadable, 0, 0) |
+            __gen_uint(values->vertex_shader_start_in_final_thread_section, 1, 1) |
+            __gen_uint(values->vertex_shader_propagate_nans, 2, 2);
+
+   cl[21] = __gen_address_offset(&values->vertex_shader_code_address) >> 8;
+
+   cl[22] = __gen_address_offset(&values->vertex_shader_code_address) >> 16;
+
+   cl[23] = __gen_address_offset(&values->vertex_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->vertex_shader_uniforms_address);
+   cl[24] = __gen_address_offset(&values->vertex_shader_uniforms_address);
+
+   cl[25] = __gen_address_offset(&values->vertex_shader_uniforms_address) >> 8;
+
+   cl[26] = __gen_address_offset(&values->vertex_shader_uniforms_address) >> 16;
+
+   cl[27] = __gen_address_offset(&values->vertex_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->coordinate_shader_code_address);
+   cl[28] = __gen_address_offset(&values->coordinate_shader_code_address) |
+            __gen_uint(values->coordinate_shader_4_way_threadable, 0, 0) |
+            __gen_uint(values->coordinate_shader_start_in_final_thread_section, 1, 1) |
+            __gen_uint(values->coordinate_shader_propagate_nans, 2, 2);
+
+   cl[29] = __gen_address_offset(&values->coordinate_shader_code_address) >> 8;
+
+   cl[30] = __gen_address_offset(&values->coordinate_shader_code_address) >> 16;
+
+   cl[31] = __gen_address_offset(&values->coordinate_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->coordinate_shader_uniforms_address);
+   cl[32] = __gen_address_offset(&values->coordinate_shader_uniforms_address);
+
+   cl[33] = __gen_address_offset(&values->coordinate_shader_uniforms_address) >> 8;
+
+   cl[34] = __gen_address_offset(&values->coordinate_shader_uniforms_address) >> 16;
+
+   cl[35] = __gen_address_offset(&values->coordinate_shader_uniforms_address) >> 24;
+
+}
+
+#define V3D42_GL_SHADER_STATE_RECORD_length     36
+#ifdef __gen_unpack_address
+static inline void
+V3D42_GL_SHADER_STATE_RECORD_unpack(const uint8_t * restrict cl,
+                                    struct V3D42_GL_SHADER_STATE_RECORD * restrict values)
+{
+   values->point_size_in_shaded_vertex_data = __gen_unpack_uint(cl, 0, 0);
+   values->enable_clipping = __gen_unpack_uint(cl, 1, 1);
+   values->vertex_id_read_by_coordinate_shader = __gen_unpack_uint(cl, 2, 2);
+   values->instance_id_read_by_coordinate_shader = __gen_unpack_uint(cl, 3, 3);
+   values->base_instance_id_read_by_coordinate_shader = __gen_unpack_uint(cl, 4, 4);
+   values->vertex_id_read_by_vertex_shader = __gen_unpack_uint(cl, 5, 5);
+   values->instance_id_read_by_vertex_shader = __gen_unpack_uint(cl, 6, 6);
+   values->base_instance_id_read_by_vertex_shader = __gen_unpack_uint(cl, 7, 7);
+   values->fragment_shader_does_z_writes = __gen_unpack_uint(cl, 8, 8);
+   values->turn_off_early_z_test = __gen_unpack_uint(cl, 9, 9);
+   values->coordinate_shader_has_separate_input_and_output_vpm_blocks = __gen_unpack_uint(cl, 10, 10);
+   values->vertex_shader_has_separate_input_and_output_vpm_blocks = __gen_unpack_uint(cl, 11, 11);
+   values->fragment_shader_uses_real_pixel_centre_w_in_addition_to_centroid_w2 = __gen_unpack_uint(cl, 12, 12);
+   values->enable_sample_rate_shading = __gen_unpack_uint(cl, 13, 13);
+   values->any_shader_reads_hardware_written_primitive_id = __gen_unpack_uint(cl, 14, 14);
+   values->insert_primitive_id_as_first_varying_to_fragment_shader = __gen_unpack_uint(cl, 15, 15);
+   values->turn_off_scoreboard = __gen_unpack_uint(cl, 16, 16);
+   values->do_scoreboard_wait_on_first_thread_switch = __gen_unpack_uint(cl, 17, 17);
+   values->disable_implicit_point_line_varyings = __gen_unpack_uint(cl, 18, 18);
+   values->no_prim_pack = __gen_unpack_uint(cl, 19, 19);
+   values->number_of_varyings_in_fragment_shader = __gen_unpack_uint(cl, 24, 31);
+   values->coordinate_shader_output_vpm_segment_size = __gen_unpack_uint(cl, 32, 35);
+   values->min_coord_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size = __gen_unpack_uint(cl, 36, 39);
+   values->coordinate_shader_input_vpm_segment_size = __gen_unpack_uint(cl, 40, 43);
+   values->min_coord_shader_input_segments_required_in_play = __gen_unpack_uint(cl, 44, 47) + 1;
+   values->vertex_shader_output_vpm_segment_size = __gen_unpack_uint(cl, 48, 51);
+   values->min_vertex_shader_output_segments_required_in_play_in_addition_to_vcm_cache_size = __gen_unpack_uint(cl, 52, 55);
+   values->vertex_shader_input_vpm_segment_size = __gen_unpack_uint(cl, 56, 59);
+   values->min_vertex_shader_input_segments_required_in_play = __gen_unpack_uint(cl, 60, 63) + 1;
+   values->address_of_default_attribute_values = __gen_unpack_address(cl, 64, 95);
+   values->fragment_shader_code_address = __gen_unpack_address(cl, 99, 127);
+   values->fragment_shader_4_way_threadable = __gen_unpack_uint(cl, 96, 96);
+   values->fragment_shader_start_in_final_thread_section = __gen_unpack_uint(cl, 97, 97);
+   values->fragment_shader_propagate_nans = __gen_unpack_uint(cl, 98, 98);
+   values->fragment_shader_uniforms_address = __gen_unpack_address(cl, 128, 159);
+   values->vertex_shader_code_address = __gen_unpack_address(cl, 163, 191);
+   values->vertex_shader_4_way_threadable = __gen_unpack_uint(cl, 160, 160);
+   values->vertex_shader_start_in_final_thread_section = __gen_unpack_uint(cl, 161, 161);
+   values->vertex_shader_propagate_nans = __gen_unpack_uint(cl, 162, 162);
+   values->vertex_shader_uniforms_address = __gen_unpack_address(cl, 192, 223);
+   values->coordinate_shader_code_address = __gen_unpack_address(cl, 227, 255);
+   values->coordinate_shader_4_way_threadable = __gen_unpack_uint(cl, 224, 224);
+   values->coordinate_shader_start_in_final_thread_section = __gen_unpack_uint(cl, 225, 225);
+   values->coordinate_shader_propagate_nans = __gen_unpack_uint(cl, 226, 226);
+   values->coordinate_shader_uniforms_address = __gen_unpack_address(cl, 256, 287);
+}
+#endif
+
+
+#define V3D42_GEOMETRY_SHADER_STATE_RECORD_header\
+
+
+struct V3D42_GEOMETRY_SHADER_STATE_RECORD {
+   __gen_address_type                   geometry_bin_mode_shader_code_address;
+   bool                                 _4_way_threadable;
+   bool                                 start_in_final_thread_section;
+   bool                                 propagate_nans;
+   __gen_address_type                   geometry_bin_mode_shader_uniforms_address;
+   __gen_address_type                   geometry_render_mode_shader_code_address;
+   __gen_address_type                   geometry_render_mode_shader_uniforms_address;
+};
+
+static inline void
+V3D42_GEOMETRY_SHADER_STATE_RECORD_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D42_GEOMETRY_SHADER_STATE_RECORD * restrict values)
+{
+   __gen_emit_reloc(data, &values->geometry_bin_mode_shader_code_address);
+   cl[ 0] = __gen_address_offset(&values->geometry_bin_mode_shader_code_address) |
+            __gen_uint(values->_4_way_threadable, 0, 0) |
+            __gen_uint(values->start_in_final_thread_section, 1, 1) |
+            __gen_uint(values->propagate_nans, 2, 2);
+
+   cl[ 1] = __gen_address_offset(&values->geometry_bin_mode_shader_code_address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->geometry_bin_mode_shader_code_address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->geometry_bin_mode_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->geometry_bin_mode_shader_uniforms_address);
+   cl[ 4] = __gen_address_offset(&values->geometry_bin_mode_shader_uniforms_address);
+
+   cl[ 5] = __gen_address_offset(&values->geometry_bin_mode_shader_uniforms_address) >> 8;
+
+   cl[ 6] = __gen_address_offset(&values->geometry_bin_mode_shader_uniforms_address) >> 16;
+
+   cl[ 7] = __gen_address_offset(&values->geometry_bin_mode_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->geometry_render_mode_shader_code_address);
+   cl[ 8] = __gen_address_offset(&values->geometry_render_mode_shader_code_address);
+
+   cl[ 9] = __gen_address_offset(&values->geometry_render_mode_shader_code_address) >> 8;
+
+   cl[10] = __gen_address_offset(&values->geometry_render_mode_shader_code_address) >> 16;
+
+   cl[11] = __gen_address_offset(&values->geometry_render_mode_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->geometry_render_mode_shader_uniforms_address);
+   cl[12] = __gen_address_offset(&values->geometry_render_mode_shader_uniforms_address);
+
+   cl[13] = __gen_address_offset(&values->geometry_render_mode_shader_uniforms_address) >> 8;
+
+   cl[14] = __gen_address_offset(&values->geometry_render_mode_shader_uniforms_address) >> 16;
+
+   cl[15] = __gen_address_offset(&values->geometry_render_mode_shader_uniforms_address) >> 24;
+
+}
+
+#define V3D42_GEOMETRY_SHADER_STATE_RECORD_length     16
+#ifdef __gen_unpack_address
+static inline void
+V3D42_GEOMETRY_SHADER_STATE_RECORD_unpack(const uint8_t * restrict cl,
+                                          struct V3D42_GEOMETRY_SHADER_STATE_RECORD * restrict values)
+{
+   values->geometry_bin_mode_shader_code_address = __gen_unpack_address(cl, 0, 31);
+   values->_4_way_threadable = __gen_unpack_uint(cl, 0, 0);
+   values->start_in_final_thread_section = __gen_unpack_uint(cl, 1, 1);
+   values->propagate_nans = __gen_unpack_uint(cl, 2, 2);
+   values->geometry_bin_mode_shader_uniforms_address = __gen_unpack_address(cl, 32, 63);
+   values->geometry_render_mode_shader_code_address = __gen_unpack_address(cl, 64, 95);
+   values->geometry_render_mode_shader_uniforms_address = __gen_unpack_address(cl, 96, 127);
+}
+#endif
+
+
+#define V3D42_TESSELLATION_SHADER_STATE_RECORD_header\
+
+
+struct V3D42_TESSELLATION_SHADER_STATE_RECORD {
+   __gen_address_type                   tessellation_bin_mode_control_shader_code_address;
+   __gen_address_type                   tessellation_bin_mode_control_shader_uniforms_address;
+   __gen_address_type                   tessellation_render_mode_control_shader_code_address;
+   __gen_address_type                   tessellation_render_mode_control_shader_uniforms_address;
+   __gen_address_type                   tessellation_bin_mode_evaluation_shader_code_address;
+   __gen_address_type                   tessellation_bin_mode_evaluation_shader_uniforms_address;
+   __gen_address_type                   tessellation_render_mode_evaluation_shader_code_address;
+   __gen_address_type                   tessellation_render_mode_evaluation_shader_uniforms_address;
+};
+
+static inline void
+V3D42_TESSELLATION_SHADER_STATE_RECORD_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                            const struct V3D42_TESSELLATION_SHADER_STATE_RECORD * restrict values)
+{
+   __gen_emit_reloc(data, &values->tessellation_bin_mode_control_shader_code_address);
+   cl[ 0] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_code_address);
+
+   cl[ 1] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_code_address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_code_address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_bin_mode_control_shader_uniforms_address);
+   cl[ 4] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_uniforms_address);
+
+   cl[ 5] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_uniforms_address) >> 8;
+
+   cl[ 6] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_uniforms_address) >> 16;
+
+   cl[ 7] = __gen_address_offset(&values->tessellation_bin_mode_control_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_render_mode_control_shader_code_address);
+   cl[ 8] = __gen_address_offset(&values->tessellation_render_mode_control_shader_code_address);
+
+   cl[ 9] = __gen_address_offset(&values->tessellation_render_mode_control_shader_code_address) >> 8;
+
+   cl[10] = __gen_address_offset(&values->tessellation_render_mode_control_shader_code_address) >> 16;
+
+   cl[11] = __gen_address_offset(&values->tessellation_render_mode_control_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_render_mode_control_shader_uniforms_address);
+   cl[12] = __gen_address_offset(&values->tessellation_render_mode_control_shader_uniforms_address);
+
+   cl[13] = __gen_address_offset(&values->tessellation_render_mode_control_shader_uniforms_address) >> 8;
+
+   cl[14] = __gen_address_offset(&values->tessellation_render_mode_control_shader_uniforms_address) >> 16;
+
+   cl[15] = __gen_address_offset(&values->tessellation_render_mode_control_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_bin_mode_evaluation_shader_code_address);
+   cl[16] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_code_address);
+
+   cl[17] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_code_address) >> 8;
+
+   cl[18] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_code_address) >> 16;
+
+   cl[19] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_bin_mode_evaluation_shader_uniforms_address);
+   cl[20] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_uniforms_address);
+
+   cl[21] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_uniforms_address) >> 8;
+
+   cl[22] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_uniforms_address) >> 16;
+
+   cl[23] = __gen_address_offset(&values->tessellation_bin_mode_evaluation_shader_uniforms_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_render_mode_evaluation_shader_code_address);
+   cl[24] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_code_address);
+
+   cl[25] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_code_address) >> 8;
+
+   cl[26] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_code_address) >> 16;
+
+   cl[27] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_code_address) >> 24;
+
+   __gen_emit_reloc(data, &values->tessellation_render_mode_evaluation_shader_uniforms_address);
+   cl[28] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_uniforms_address);
+
+   cl[29] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_uniforms_address) >> 8;
+
+   cl[30] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_uniforms_address) >> 16;
+
+   cl[31] = __gen_address_offset(&values->tessellation_render_mode_evaluation_shader_uniforms_address) >> 24;
+
+}
+
+#define V3D42_TESSELLATION_SHADER_STATE_RECORD_length     32
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TESSELLATION_SHADER_STATE_RECORD_unpack(const uint8_t * restrict cl,
+                                              struct V3D42_TESSELLATION_SHADER_STATE_RECORD * restrict values)
+{
+   values->tessellation_bin_mode_control_shader_code_address = __gen_unpack_address(cl, 0, 31);
+   values->tessellation_bin_mode_control_shader_uniforms_address = __gen_unpack_address(cl, 32, 63);
+   values->tessellation_render_mode_control_shader_code_address = __gen_unpack_address(cl, 64, 95);
+   values->tessellation_render_mode_control_shader_uniforms_address = __gen_unpack_address(cl, 96, 127);
+   values->tessellation_bin_mode_evaluation_shader_code_address = __gen_unpack_address(cl, 128, 159);
+   values->tessellation_bin_mode_evaluation_shader_uniforms_address = __gen_unpack_address(cl, 160, 191);
+   values->tessellation_render_mode_evaluation_shader_code_address = __gen_unpack_address(cl, 192, 223);
+   values->tessellation_render_mode_evaluation_shader_uniforms_address = __gen_unpack_address(cl, 224, 255);
+}
+#endif
+
+
+#define V3D42_GL_SHADER_STATE_ATTRIBUTE_RECORD_header\
+
+
+struct V3D42_GL_SHADER_STATE_ATTRIBUTE_RECORD {
+   __gen_address_type                   address;
+   uint32_t                             vec_size;
+   uint32_t                             type;
+#define ATTRIBUTE_HALF_FLOAT                     1
+#define ATTRIBUTE_FLOAT                          2
+#define ATTRIBUTE_FIXED                          3
+#define ATTRIBUTE_BYTE                           4
+#define ATTRIBUTE_SHORT                          5
+#define ATTRIBUTE_INT                            6
+#define ATTRIBUTE_INT2_10_10_10                  7
+   bool                                 signed_int_type;
+   bool                                 normalized_int_type;
+   bool                                 read_as_int_uint;
+   uint32_t                             number_of_values_read_by_coordinate_shader;
+   uint32_t                             number_of_values_read_by_vertex_shader;
+   uint32_t                             instance_divisor;
+   uint32_t                             stride;
+   uint32_t                             maximum_index;
+};
+
+static inline void
+V3D42_GL_SHADER_STATE_ATTRIBUTE_RECORD_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                            const struct V3D42_GL_SHADER_STATE_ATTRIBUTE_RECORD * restrict values)
+{
+   __gen_emit_reloc(data, &values->address);
+   cl[ 0] = __gen_address_offset(&values->address);
+
+   cl[ 1] = __gen_address_offset(&values->address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->address) >> 24;
+
+   cl[ 4] = __gen_uint(values->vec_size, 0, 1) |
+            __gen_uint(values->type, 2, 4) |
+            __gen_uint(values->signed_int_type, 5, 5) |
+            __gen_uint(values->normalized_int_type, 6, 6) |
+            __gen_uint(values->read_as_int_uint, 7, 7);
+
+   cl[ 5] = __gen_uint(values->number_of_values_read_by_coordinate_shader, 0, 3) |
+            __gen_uint(values->number_of_values_read_by_vertex_shader, 4, 7);
+
+   cl[ 6] = __gen_uint(values->instance_divisor, 0, 15);
+
+   cl[ 7] = __gen_uint(values->instance_divisor, 0, 15) >> 8;
+
+
+   memcpy(&cl[8], &values->stride, sizeof(values->stride));
+
+   memcpy(&cl[12], &values->maximum_index, sizeof(values->maximum_index));
+}
+
+#define V3D42_GL_SHADER_STATE_ATTRIBUTE_RECORD_length     16
+#ifdef __gen_unpack_address
+static inline void
+V3D42_GL_SHADER_STATE_ATTRIBUTE_RECORD_unpack(const uint8_t * restrict cl,
+                                              struct V3D42_GL_SHADER_STATE_ATTRIBUTE_RECORD * restrict values)
+{
+   values->address = __gen_unpack_address(cl, 0, 31);
+   values->vec_size = __gen_unpack_uint(cl, 32, 33);
+   values->type = __gen_unpack_uint(cl, 34, 36);
+   values->signed_int_type = __gen_unpack_uint(cl, 37, 37);
+   values->normalized_int_type = __gen_unpack_uint(cl, 38, 38);
+   values->read_as_int_uint = __gen_unpack_uint(cl, 39, 39);
+   values->number_of_values_read_by_coordinate_shader = __gen_unpack_uint(cl, 40, 43);
+   values->number_of_values_read_by_vertex_shader = __gen_unpack_uint(cl, 44, 47);
+   values->instance_divisor = __gen_unpack_uint(cl, 48, 63);
+   values->stride = __gen_unpack_uint(cl, 64, 95);
+   values->maximum_index = __gen_unpack_uint(cl, 96, 127);
+}
+#endif
+
+
+#define V3D42_VPM_GENERIC_BLOCK_WRITE_SETUP_header\
+   .id                                  =      0,  \
+   .id0                                 =      0
+
+struct V3D42_VPM_GENERIC_BLOCK_WRITE_SETUP {
+   uint32_t                             id;
+   uint32_t                             id0;
+   bool                                 horiz;
+   bool                                 laned;
+   bool                                 segs;
+   int32_t                              stride;
+   uint32_t                             size;
+#define VPM_SETUP_SIZE_8_BIT                     0
+#define VPM_SETUP_SIZE_16_BIT                    1
+#define VPM_SETUP_SIZE_32_BIT                    2
+   uint32_t                             addr;
+};
+
+static inline void
+V3D42_VPM_GENERIC_BLOCK_WRITE_SETUP_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                         const struct V3D42_VPM_GENERIC_BLOCK_WRITE_SETUP * restrict values)
+{
+   cl[ 0] = __gen_uint(values->addr, 0, 12);
+
+   cl[ 1] = __gen_sint(values->stride, 7, 13) |
+            __gen_uint(values->size, 5, 6) |
+            __gen_uint(values->addr, 0, 12) >> 8;
+
+   cl[ 2] = __gen_uint(values->laned, 7, 7) |
+            __gen_uint(values->segs, 6, 6) |
+            __gen_sint(values->stride, 7, 13) >> 8;
+
+   cl[ 3] = __gen_uint(values->id, 6, 7) |
+            __gen_uint(values->id0, 3, 5) |
+            __gen_uint(values->horiz, 0, 0);
+
+}
+
+#define V3D42_VPM_GENERIC_BLOCK_WRITE_SETUP_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D42_VPM_GENERIC_BLOCK_WRITE_SETUP_unpack(const uint8_t * restrict cl,
+                                           struct V3D42_VPM_GENERIC_BLOCK_WRITE_SETUP * restrict values)
+{
+   values->id = __gen_unpack_uint(cl, 30, 31);
+   values->id0 = __gen_unpack_uint(cl, 27, 29);
+   values->horiz = __gen_unpack_uint(cl, 24, 24);
+   values->laned = __gen_unpack_uint(cl, 23, 23);
+   values->segs = __gen_unpack_uint(cl, 22, 22);
+   values->stride = __gen_unpack_sint(cl, 15, 21);
+   values->size = __gen_unpack_uint(cl, 13, 14);
+   values->addr = __gen_unpack_uint(cl, 0, 12);
+}
+#endif
+
+
+#define V3D42_VPM_GENERIC_BLOCK_READ_SETUP_header\
+   .id                                  =      1
+
+struct V3D42_VPM_GENERIC_BLOCK_READ_SETUP {
+   uint32_t                             id;
+   bool                                 horiz;
+   bool                                 laned;
+   bool                                 segs;
+   uint32_t                             num;
+   int32_t                              stride;
+   uint32_t                             size;
+#define VPM_SETUP_SIZE_8_BIT                     0
+#define VPM_SETUP_SIZE_16_BIT                    1
+#define VPM_SETUP_SIZE_32_BIT                    2
+   uint32_t                             addr;
+};
+
+static inline void
+V3D42_VPM_GENERIC_BLOCK_READ_SETUP_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                        const struct V3D42_VPM_GENERIC_BLOCK_READ_SETUP * restrict values)
+{
+   cl[ 0] = __gen_uint(values->addr, 0, 12);
+
+   cl[ 1] = __gen_sint(values->stride, 7, 13) |
+            __gen_uint(values->size, 5, 6) |
+            __gen_uint(values->addr, 0, 12) >> 8;
+
+   cl[ 2] = __gen_uint(values->num, 6, 10) |
+            __gen_sint(values->stride, 7, 13) >> 8;
+
+   cl[ 3] = __gen_uint(values->id, 6, 7) |
+            __gen_uint(values->horiz, 5, 5) |
+            __gen_uint(values->laned, 4, 4) |
+            __gen_uint(values->segs, 3, 3) |
+            __gen_uint(values->num, 6, 10) >> 8;
+
+}
+
+#define V3D42_VPM_GENERIC_BLOCK_READ_SETUP_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D42_VPM_GENERIC_BLOCK_READ_SETUP_unpack(const uint8_t * restrict cl,
+                                          struct V3D42_VPM_GENERIC_BLOCK_READ_SETUP * restrict values)
+{
+   values->id = __gen_unpack_uint(cl, 30, 31);
+   values->horiz = __gen_unpack_uint(cl, 29, 29);
+   values->laned = __gen_unpack_uint(cl, 28, 28);
+   values->segs = __gen_unpack_uint(cl, 27, 27);
+   values->num = __gen_unpack_uint(cl, 22, 26);
+   values->stride = __gen_unpack_sint(cl, 15, 21);
+   values->size = __gen_unpack_uint(cl, 13, 14);
+   values->addr = __gen_unpack_uint(cl, 0, 12);
+}
+#endif
+
+
+#define V3D42_TMU_CONFIG_PARAMETER_0_header     \
+
+
+struct V3D42_TMU_CONFIG_PARAMETER_0 {
+   __gen_address_type                   texture_state_address;
+   uint32_t                             return_words_of_texture_data;
+};
+
+static inline void
+V3D42_TMU_CONFIG_PARAMETER_0_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                  const struct V3D42_TMU_CONFIG_PARAMETER_0 * restrict values)
+{
+   __gen_emit_reloc(data, &values->texture_state_address);
+   cl[ 0] = __gen_address_offset(&values->texture_state_address) |
+            __gen_uint(values->return_words_of_texture_data, 0, 3);
+
+   cl[ 1] = __gen_address_offset(&values->texture_state_address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->texture_state_address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->texture_state_address) >> 24;
+
+}
+
+#define V3D42_TMU_CONFIG_PARAMETER_0_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TMU_CONFIG_PARAMETER_0_unpack(const uint8_t * restrict cl,
+                                    struct V3D42_TMU_CONFIG_PARAMETER_0 * restrict values)
+{
+   values->texture_state_address = __gen_unpack_address(cl, 0, 31);
+   values->return_words_of_texture_data = __gen_unpack_uint(cl, 0, 3);
+}
+#endif
+
+
+#define V3D42_TMU_CONFIG_PARAMETER_1_header     \
+
+
+struct V3D42_TMU_CONFIG_PARAMETER_1 {
+   __gen_address_type                   sampler_state_address;
+   bool                                 per_pixel_mask_enable;
+   bool                                 unnormalized_coordinates;
+   bool                                 output_type_32_bit;
+};
+
+static inline void
+V3D42_TMU_CONFIG_PARAMETER_1_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                  const struct V3D42_TMU_CONFIG_PARAMETER_1 * restrict values)
+{
+   __gen_emit_reloc(data, &values->sampler_state_address);
+   cl[ 0] = __gen_address_offset(&values->sampler_state_address) |
+            __gen_uint(values->per_pixel_mask_enable, 2, 2) |
+            __gen_uint(values->unnormalized_coordinates, 1, 1) |
+            __gen_uint(values->output_type_32_bit, 0, 0);
+
+   cl[ 1] = __gen_address_offset(&values->sampler_state_address) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->sampler_state_address) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->sampler_state_address) >> 24;
+
+}
+
+#define V3D42_TMU_CONFIG_PARAMETER_1_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TMU_CONFIG_PARAMETER_1_unpack(const uint8_t * restrict cl,
+                                    struct V3D42_TMU_CONFIG_PARAMETER_1 * restrict values)
+{
+   values->sampler_state_address = __gen_unpack_address(cl, 0, 31);
+   values->per_pixel_mask_enable = __gen_unpack_uint(cl, 2, 2);
+   values->unnormalized_coordinates = __gen_unpack_uint(cl, 1, 1);
+   values->output_type_32_bit = __gen_unpack_uint(cl, 0, 0);
+}
+#endif
+
+
+#define V3D42_TMU_CONFIG_PARAMETER_2_header     \
+
+
+struct V3D42_TMU_CONFIG_PARAMETER_2 {
+   uint32_t                             pad;
+   bool                                 lod_query;
+   enum V3D42_TMU_Op                    op;
+   int32_t                              offset_r;
+   int32_t                              offset_t;
+   int32_t                              offset_s;
+   bool                                 gather_mode;
+   uint32_t                             gather_component;
+   bool                                 coefficient_mode;
+   uint32_t                             sample_number;
+   bool                                 disable_autolod;
+   bool                                 offset_format_8;
+};
+
+static inline void
+V3D42_TMU_CONFIG_PARAMETER_2_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                  const struct V3D42_TMU_CONFIG_PARAMETER_2 * restrict values)
+{
+   cl[ 0] = __gen_uint(values->gather_mode, 7, 7) |
+            __gen_uint(values->gather_component, 5, 6) |
+            __gen_uint(values->coefficient_mode, 4, 4) |
+            __gen_uint(values->sample_number, 2, 3) |
+            __gen_uint(values->disable_autolod, 1, 1) |
+            __gen_uint(values->offset_format_8, 0, 0);
+
+   cl[ 1] = __gen_uint(values->pad, 1, 23) |
+            __gen_uint(values->lod_query, 0, 0) |
+            __gen_sint(values->offset_t, 4, 7) |
+            __gen_sint(values->offset_s, 0, 3);
+
+   cl[ 2] = __gen_uint(values->pad, 1, 23) >> 8 |
+            __gen_uint(values->op, 4, 7) |
+            __gen_sint(values->offset_r, 0, 3);
+
+   cl[ 3] = __gen_uint(values->pad, 1, 23) >> 16;
+
+}
+
+#define V3D42_TMU_CONFIG_PARAMETER_2_length      4
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TMU_CONFIG_PARAMETER_2_unpack(const uint8_t * restrict cl,
+                                    struct V3D42_TMU_CONFIG_PARAMETER_2 * restrict values)
+{
+   values->pad = __gen_unpack_uint(cl, 9, 31);
+   values->lod_query = __gen_unpack_uint(cl, 8, 8);
+   values->op = __gen_unpack_uint(cl, 20, 23);
+   values->offset_r = __gen_unpack_sint(cl, 16, 19);
+   values->offset_t = __gen_unpack_sint(cl, 12, 15);
+   values->offset_s = __gen_unpack_sint(cl, 8, 11);
+   values->gather_mode = __gen_unpack_uint(cl, 7, 7);
+   values->gather_component = __gen_unpack_uint(cl, 5, 6);
+   values->coefficient_mode = __gen_unpack_uint(cl, 4, 4);
+   values->sample_number = __gen_unpack_uint(cl, 2, 3);
+   values->disable_autolod = __gen_unpack_uint(cl, 1, 1);
+   values->offset_format_8 = __gen_unpack_uint(cl, 0, 0);
+}
+#endif
+
+
+#define V3D42_TEXTURE_SHADER_STATE_header       \
+
+
+struct V3D42_TEXTURE_SHADER_STATE {
+   uint64_t                             pad;
+   bool                                 uif_xor_disable;
+   bool                                 level_0_is_strictly_uif;
+   bool                                 level_0_xor_enable;
+   uint32_t                             level_0_ub_pad;
+   uint32_t                             base_level;
+   uint32_t                             max_level;
+   uint32_t                             swizzle_a;
+#define SWIZZLE_ZERO                             0
+#define SWIZZLE_ONE                              1
+#define SWIZZLE_RED                              2
+#define SWIZZLE_GREEN                            3
+#define SWIZZLE_BLUE                             4
+#define SWIZZLE_ALPHA                            5
+   uint32_t                             swizzle_b;
+   uint32_t                             swizzle_g;
+   uint32_t                             swizzle_r;
+   bool                                 extended;
+   uint32_t                             texture_type;
+   uint32_t                             image_depth;
+   uint32_t                             image_height;
+   uint32_t                             image_width;
+   uint32_t                             array_stride_64_byte_aligned;
+   __gen_address_type                   texture_base_pointer;
+   bool                                 reverse_standard_border_color;
+   bool                                 ahdr;
+   bool                                 srgb;
+   bool                                 flip_s_and_t_on_incoming_request;
+   bool                                 flip_texture_y_axis;
+   bool                                 flip_texture_x_axis;
+};
+
+static inline void
+V3D42_TEXTURE_SHADER_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                                const struct V3D42_TEXTURE_SHADER_STATE * restrict values)
+{
+   __gen_emit_reloc(data, &values->texture_base_pointer);
+   cl[ 0] = __gen_address_offset(&values->texture_base_pointer) |
+            __gen_uint(values->reverse_standard_border_color, 5, 5) |
+            __gen_uint(values->ahdr, 4, 4) |
+            __gen_uint(values->srgb, 3, 3) |
+            __gen_uint(values->flip_s_and_t_on_incoming_request, 2, 2) |
+            __gen_uint(values->flip_texture_y_axis, 1, 1) |
+            __gen_uint(values->flip_texture_x_axis, 0, 0);
+
+   cl[ 1] = __gen_address_offset(&values->texture_base_pointer) >> 8;
+
+   cl[ 2] = __gen_address_offset(&values->texture_base_pointer) >> 16;
+
+   cl[ 3] = __gen_address_offset(&values->texture_base_pointer) >> 24;
+
+   cl[ 4] = __gen_uint(values->array_stride_64_byte_aligned, 0, 25);
+
+   cl[ 5] = __gen_uint(values->array_stride_64_byte_aligned, 0, 25) >> 8;
+
+   cl[ 6] = __gen_uint(values->array_stride_64_byte_aligned, 0, 25) >> 16;
+
+   cl[ 7] = __gen_uint(values->image_width, 2, 15) |
+            __gen_uint(values->array_stride_64_byte_aligned, 0, 25) >> 24;
+
+   cl[ 8] = __gen_uint(values->image_width, 2, 15) >> 8;
+
+   cl[ 9] = __gen_uint(values->image_height, 0, 13);
+
+   cl[10] = __gen_uint(values->image_depth, 6, 19) |
+            __gen_uint(values->image_height, 0, 13) >> 8;
+
+   cl[11] = __gen_uint(values->image_depth, 6, 19) >> 8;
+
+   cl[12] = __gen_uint(values->texture_type, 4, 10) |
+            __gen_uint(values->image_depth, 6, 19) >> 16;
+
+   cl[13] = __gen_uint(values->swizzle_g, 7, 9) |
+            __gen_uint(values->swizzle_r, 4, 6) |
+            __gen_uint(values->extended, 3, 3) |
+            __gen_uint(values->texture_type, 4, 10) >> 8;
+
+   cl[14] = __gen_uint(values->swizzle_a, 5, 7) |
+            __gen_uint(values->swizzle_b, 2, 4) |
+            __gen_uint(values->swizzle_g, 7, 9) >> 8;
+
+   cl[15] = __gen_uint(values->base_level, 4, 7) |
+            __gen_uint(values->max_level, 0, 3);
+
+   cl[16] = __gen_uint(values->uif_xor_disable, 7, 7) |
+            __gen_uint(values->level_0_is_strictly_uif, 6, 6) |
+            __gen_uint(values->level_0_xor_enable, 4, 4) |
+            __gen_uint(values->level_0_ub_pad, 0, 3);
+
+   cl[17] = __gen_uint(values->pad, 0, 55);
+
+   cl[18] = __gen_uint(values->pad, 0, 55) >> 8;
+
+   cl[19] = __gen_uint(values->pad, 0, 55) >> 16;
+
+   cl[20] = __gen_uint(values->pad, 0, 55) >> 24;
+
+   cl[21] = __gen_uint(values->pad, 0, 55) >> 32;
+
+   cl[22] = __gen_uint(values->pad, 0, 55) >> 40;
+
+   cl[23] = __gen_uint(values->pad, 0, 55) >> 48;
+
+}
+
+#define V3D42_TEXTURE_SHADER_STATE_length     24
+#ifdef __gen_unpack_address
+static inline void
+V3D42_TEXTURE_SHADER_STATE_unpack(const uint8_t * restrict cl,
+                                  struct V3D42_TEXTURE_SHADER_STATE * restrict values)
+{
+   values->pad = __gen_unpack_uint(cl, 136, 191);
+   values->uif_xor_disable = __gen_unpack_uint(cl, 135, 135);
+   values->level_0_is_strictly_uif = __gen_unpack_uint(cl, 134, 134);
+   values->level_0_xor_enable = __gen_unpack_uint(cl, 132, 132);
+   values->level_0_ub_pad = __gen_unpack_uint(cl, 128, 131);
+   values->base_level = __gen_unpack_uint(cl, 124, 127);
+   values->max_level = __gen_unpack_uint(cl, 120, 123);
+   values->swizzle_a = __gen_unpack_uint(cl, 117, 119);
+   values->swizzle_b = __gen_unpack_uint(cl, 114, 116);
+   values->swizzle_g = __gen_unpack_uint(cl, 111, 113);
+   values->swizzle_r = __gen_unpack_uint(cl, 108, 110);
+   values->extended = __gen_unpack_uint(cl, 107, 107);
+   values->texture_type = __gen_unpack_uint(cl, 100, 106);
+   values->image_depth = __gen_unpack_uint(cl, 86, 99);
+   values->image_height = __gen_unpack_uint(cl, 72, 85);
+   values->image_width = __gen_unpack_uint(cl, 58, 71);
+   values->array_stride_64_byte_aligned = __gen_unpack_uint(cl, 32, 57);
+   values->texture_base_pointer = __gen_unpack_address(cl, 0, 31);
+   values->reverse_standard_border_color = __gen_unpack_uint(cl, 5, 5);
+   values->ahdr = __gen_unpack_uint(cl, 4, 4);
+   values->srgb = __gen_unpack_uint(cl, 3, 3);
+   values->flip_s_and_t_on_incoming_request = __gen_unpack_uint(cl, 2, 2);
+   values->flip_texture_y_axis = __gen_unpack_uint(cl, 1, 1);
+   values->flip_texture_x_axis = __gen_unpack_uint(cl, 0, 0);
+}
+#endif
+
+
+#define V3D42_SAMPLER_STATE_header              \
+
+
+struct V3D42_SAMPLER_STATE {
+   uint32_t                             border_color_alpha;
+   uint32_t                             border_color_blue;
+   uint32_t                             border_color_green;
+   uint32_t                             border_color_red;
+   uint32_t                             maximum_anisotropy;
+   enum V3D42_Border_Color_Mode         border_color_mode;
+   bool                                 wrap_i_border;
+   enum V3D42_Wrap_Mode                 wrap_r;
+   enum V3D42_Wrap_Mode                 wrap_t;
+   enum V3D42_Wrap_Mode                 wrap_s;
+   float                                fixed_bias;
+   float                                max_level_of_detail;
+   float                                min_level_of_detail;
+   bool                                 srgb_disable;
+   enum V3D42_Compare_Function          depth_compare_function;
+   bool                                 anisotropy_enable;
+   bool                                 mip_filter_nearest;
+   bool                                 min_filter_nearest;
+   bool                                 mag_filter_nearest;
+};
+
+static inline void
+V3D42_SAMPLER_STATE_pack(__gen_user_data *data, uint8_t * restrict cl,
+                         const struct V3D42_SAMPLER_STATE * restrict values)
+{
+   cl[ 0] = __gen_uint(values->srgb_disable, 7, 7) |
+            __gen_uint(values->depth_compare_function, 4, 6) |
+            __gen_uint(values->anisotropy_enable, 3, 3) |
+            __gen_uint(values->mip_filter_nearest, 2, 2) |
+            __gen_uint(values->min_filter_nearest, 1, 1) |
+            __gen_uint(values->mag_filter_nearest, 0, 0);
+
+   cl[ 1] = __gen_ufixed(values->min_level_of_detail, 0, 11, 8);
+
+   cl[ 2] = __gen_ufixed(values->max_level_of_detail, 4, 15, 8) |
+            __gen_ufixed(values->min_level_of_detail, 0, 11, 8) >> 8;
+
+   cl[ 3] = __gen_ufixed(values->max_level_of_detail, 4, 15, 8) >> 8;
+
+   cl[ 4] = __gen_sfixed(values->fixed_bias, 0, 15, 8);
+
+   cl[ 5] = __gen_sfixed(values->fixed_bias, 0, 15, 8) >> 8;
+
+   cl[ 6] = __gen_uint(values->wrap_r, 6, 8) |
+            __gen_uint(values->wrap_t, 3, 5) |
+            __gen_uint(values->wrap_s, 0, 2);
+
+   cl[ 7] = __gen_uint(values->maximum_anisotropy, 5, 6) |
+            __gen_uint(values->border_color_mode, 2, 4) |
+            __gen_uint(values->wrap_i_border, 1, 1) |
+            __gen_uint(values->wrap_r, 6, 8) >> 8;
+
+
+   memcpy(&cl[8], &values->border_color_red, sizeof(values->border_color_red));
+
+   memcpy(&cl[12], &values->border_color_green, sizeof(values->border_color_green));
+
+   memcpy(&cl[16], &values->border_color_blue, sizeof(values->border_color_blue));
+
+   memcpy(&cl[20], &values->border_color_alpha, sizeof(values->border_color_alpha));
+}
+
+#define V3D42_SAMPLER_STATE_length            24
+#ifdef __gen_unpack_address
+static inline void
+V3D42_SAMPLER_STATE_unpack(const uint8_t * restrict cl,
+                           struct V3D42_SAMPLER_STATE * restrict values)
+{
+   values->border_color_alpha = __gen_unpack_uint(cl, 160, 191);
+   values->border_color_blue = __gen_unpack_uint(cl, 128, 159);
+   values->border_color_green = __gen_unpack_uint(cl, 96, 127);
+   values->border_color_red = __gen_unpack_uint(cl, 64, 95);
+   values->maximum_anisotropy = __gen_unpack_uint(cl, 61, 62);
+   values->border_color_mode = __gen_unpack_uint(cl, 58, 60);
+   values->wrap_i_border = __gen_unpack_uint(cl, 57, 57);
+   values->wrap_r = __gen_unpack_uint(cl, 54, 56);
+   values->wrap_t = __gen_unpack_uint(cl, 51, 53);
+   values->wrap_s = __gen_unpack_uint(cl, 48, 50);
+   values->fixed_bias = __gen_unpack_sfixed(cl, 32, 47, 8);
+   values->max_level_of_detail = __gen_unpack_ufixed(cl, 20, 31, 8);
+   values->min_level_of_detail = __gen_unpack_ufixed(cl, 8, 19, 8);
+   values->srgb_disable = __gen_unpack_uint(cl, 7, 7);
+   values->depth_compare_function = __gen_unpack_uint(cl, 4, 6);
+   values->anisotropy_enable = __gen_unpack_uint(cl, 3, 3);
+   values->mip_filter_nearest = __gen_unpack_uint(cl, 2, 2);
+   values->min_filter_nearest = __gen_unpack_uint(cl, 1, 1);
+   values->mag_filter_nearest = __gen_unpack_uint(cl, 0, 0);
+}
+#endif
+
+
+enum V3D42_Texture_Data_Formats {
+        TEXTURE_DATA_FORMAT_R8               =      0,
+        TEXTURE_DATA_FORMAT_R8_SNORM         =      1,
+        TEXTURE_DATA_FORMAT_RG8              =      2,
+        TEXTURE_DATA_FORMAT_RG8_SNORM        =      3,
+        TEXTURE_DATA_FORMAT_RGBA8            =      4,
+        TEXTURE_DATA_FORMAT_RGBA8_SNORM      =      5,
+        TEXTURE_DATA_FORMAT_RGB565           =      6,
+        TEXTURE_DATA_FORMAT_RGBA4            =      7,
+        TEXTURE_DATA_FORMAT_RGB5_A1          =      8,
+        TEXTURE_DATA_FORMAT_RGB10_A2         =      9,
+        TEXTURE_DATA_FORMAT_R16              =     10,
+        TEXTURE_DATA_FORMAT_R16_SNORM        =     11,
+        TEXTURE_DATA_FORMAT_RG16             =     12,
+        TEXTURE_DATA_FORMAT_RG16_SNORM       =     13,
+        TEXTURE_DATA_FORMAT_RGBA16           =     14,
+        TEXTURE_DATA_FORMAT_RGBA16_SNORM     =     15,
+        TEXTURE_DATA_FORMAT_R16F             =     16,
+        TEXTURE_DATA_FORMAT_RG16F            =     17,
+        TEXTURE_DATA_FORMAT_RGBA16F          =     18,
+        TEXTURE_DATA_FORMAT_R11F_G11F_B10F   =     19,
+        TEXTURE_DATA_FORMAT_RGB9_E5          =     20,
+        TEXTURE_DATA_FORMAT_DEPTH_COMP16     =     21,
+        TEXTURE_DATA_FORMAT_DEPTH_COMP24     =     22,
+        TEXTURE_DATA_FORMAT_DEPTH_COMP32F    =     23,
+        TEXTURE_DATA_FORMAT_DEPTH24_X8       =     24,
+        TEXTURE_DATA_FORMAT_R4               =     25,
+        TEXTURE_DATA_FORMAT_R1               =     26,
+        TEXTURE_DATA_FORMAT_S8               =     27,
+        TEXTURE_DATA_FORMAT_S16              =     28,
+        TEXTURE_DATA_FORMAT_R32F             =     29,
+        TEXTURE_DATA_FORMAT_RG32F            =     30,
+        TEXTURE_DATA_FORMAT_RGBA32F          =     31,
+        TEXTURE_DATA_FORMAT_RGB8_ETC2        =     32,
+        TEXTURE_DATA_FORMAT_RGB8_PUNCHTHROUGH_ALPHA1 =     33,
+        TEXTURE_DATA_FORMAT_R11_EAC          =     34,
+        TEXTURE_DATA_FORMAT_SIGNED_R11_EAC   =     35,
+        TEXTURE_DATA_FORMAT_RG11_EAC         =     36,
+        TEXTURE_DATA_FORMAT_SIGNED_RG11_EAC  =     37,
+        TEXTURE_DATA_FORMAT_RGBA8_ETC2_EAC   =     38,
+        TEXTURE_DATA_FORMAT_YCBCR_LUMA       =     39,
+        TEXTURE_DATA_FORMAT_YCBCR_420_CHROMA =     40,
+        TEXTURE_DATA_FORMAT_BC1              =     48,
+        TEXTURE_DATA_FORMAT_BC2              =     49,
+        TEXTURE_DATA_FORMAT_BC3              =     50,
+        TEXTURE_DATA_FORMAT_ASTC_4X4         =     64,
+        TEXTURE_DATA_FORMAT_ASTC_5X4         =     65,
+        TEXTURE_DATA_FORMAT_ASTC_5X5         =     66,
+        TEXTURE_DATA_FORMAT_ASTC_6X5         =     67,
+        TEXTURE_DATA_FORMAT_ASTC_6X6         =     68,
+        TEXTURE_DATA_FORMAT_ASTC_8X5         =     69,
+        TEXTURE_DATA_FORMAT_ASTC_8X6         =     70,
+        TEXTURE_DATA_FORMAT_ASTC_8X8         =     71,
+        TEXTURE_DATA_FORMAT_ASTC_10X5        =     72,
+        TEXTURE_DATA_FORMAT_ASTC_10X6        =     73,
+        TEXTURE_DATA_FORMAT_ASTC_10X8        =     74,
+        TEXTURE_DATA_FORMAT_ASTC_10X10       =     75,
+        TEXTURE_DATA_FORMAT_ASTC_12X10       =     76,
+        TEXTURE_DATA_FORMAT_ASTC_12X12       =     77,
+        TEXTURE_DATA_FORMAT_R8I              =     96,
+        TEXTURE_DATA_FORMAT_R8UI             =     97,
+        TEXTURE_DATA_FORMAT_RG8I             =     98,
+        TEXTURE_DATA_FORMAT_RG8UI            =     99,
+        TEXTURE_DATA_FORMAT_RGBA8I           =    100,
+        TEXTURE_DATA_FORMAT_RGBA8UI          =    101,
+        TEXTURE_DATA_FORMAT_R16I             =    102,
+        TEXTURE_DATA_FORMAT_R16UI            =    103,
+        TEXTURE_DATA_FORMAT_RG16I            =    104,
+        TEXTURE_DATA_FORMAT_RG16UI           =    105,
+        TEXTURE_DATA_FORMAT_RGBA16I          =    106,
+        TEXTURE_DATA_FORMAT_RGBA16UI         =    107,
+        TEXTURE_DATA_FORMAT_R32I             =    108,
+        TEXTURE_DATA_FORMAT_R32UI            =    109,
+        TEXTURE_DATA_FORMAT_RG32I            =    110,
+        TEXTURE_DATA_FORMAT_RG32UI           =    111,
+        TEXTURE_DATA_FORMAT_RGBA32I          =    112,
+        TEXTURE_DATA_FORMAT_RGBA32UI         =    113,
+        TEXTURE_DATA_FORMAT_RGB10_A2UI       =    114,
+};
+
+#endif /* V3D42_PACK_H */
diff --git a/lib/mesa/src/broadcom/cle/v3d_xml.h b/lib/mesa/src/broadcom/cle/v3d_xml.h
index a940429c1..1c807324b 100644
--- a/lib/mesa/src/broadcom/cle/v3d_xml.h
+++ b/lib/mesa/src/broadcom/cle/v3d_xml.h
@@ -3,717 +3,1051 @@ static const struct {
    uint32_t offset;
    uint32_t length;
 } genxml_files_table[] = {
-   { 21, 0, 15431 },
-   { 33, 15431, 42910 },
+   { 21, 0, 15538 },
+   { 33, 15538, 75716 },
 };
 
 static const uint8_t compress_genxmls[] = {
-   0x78, 0x9c, 0xed, 0x5d, 0x5b, 0x73, 0xe3, 0xb6, 0x92, 0x7e, 0xcf, 0xaf, 
-   0xc0, 0xce, 0xcb, 0xf1, 0xec, 0xae, 0x62, 0x91, 0xba, 0x58, 0xae, 0x4a, 
-   0xb2, 0xa5, 0x9b, 0x3d, 0xae, 0x23, 0x5b, 0x5a, 0x49, 0xf6, 0xcc, 0xf8, 
-   0x45, 0x45, 0x49, 0x94, 0xcd, 0x8a, 0x44, 0xea, 0x90, 0x94, 0x2f, 0xf9, 
-   0xf5, 0x8b, 0x0b, 0x49, 0x00, 0x64, 0x83, 0x00, 0x35, 0x9e, 0x49, 0xb2, 
-   0xc7, 0xa7, 0x4e, 0x25, 0x8e, 0xdd, 0x68, 0x00, 0xdd, 0x8d, 0x46, 0xa3, 
-   0xd1, 0xf8, 0xf8, 0xcb, 0xd3, 0xea, 0x65, 0xb7, 0x45, 0x0f, 0xae, 0xff, 
-   0xeb, 0x07, 0xfb, 0x67, 0xeb, 0xc3, 0x6f, 0x3f, 0x21, 0xf4, 0xcb, 0xde, 
-   0x59, 0xfd, 0xee, 0xc6, 0xc8, 0x77, 0x76, 0xee, 0xaf, 0x1f, 0x3e, 0x39, 
-   0xdb, 0xf8, 0x03, 0x5a, 0x05, 0x6b, 0xfc, 0x73, 0xfd, 0xc3, 0x69, 0x91, 
-   0xe0, 0x66, 0x3c, 0x49, 0xff, 0x6e, 0x41, 0x7f, 0xbf, 0xd8, 0x1e, 0xa2, 
-   0xc7, 0x94, 0xa2, 0x89, 0x7f, 0xd8, 0xfe, 0xfa, 0xa1, 0xa7, 0x24, 0x44, 
-   0xdd, 0xed, 0x16, 0xcd, 0x62, 0x27, 0x76, 0xd3, 0x26, 0xad, 0x92, 0x26, 
-   0x98, 0x30, 0x8c, 0xd1, 0xdc, 0xdb, 0xba, 0xa8, 0xe7, 0xf9, 0xbe, 0xe7, 
-   0x3f, 0xa4, 0xad, 0xda, 0x25, 0xad, 0xae, 0xfc, 0x55, 0xe8, 0xee, 0x5c, 
-   0x3f, 0x46, 0x33, 0x77, 0xe7, 0xec, 0x1f, 0x83, 0x30, 0xeb, 0xec, 0x0c, 
-   0xa2, 0xff, 0xec, 0x78, 0x31, 0x0a, 0xfc, 0x22, 0x75, 0x07, 0xa2, 0xee, 
-   0x85, 0x8e, 0xbf, 0xca, 0x26, 0x6c, 0xb5, 0xa9, 0x4c, 0x31, 0xd1, 0xc6, 
-   0x73, 0xb7, 0xeb, 0x84, 0xa6, 0xbb, 0x5e, 0x87, 0x6e, 0x14, 0x7d, 0x40, 
-   0x91, 0xf7, 0x07, 0xfe, 0xcf, 0x86, 0x8d, 0x7f, 0x22, 0x53, 0x21, 0x22, 
-   0x46, 0xf1, 0xeb, 0x1e, 0xff, 0xce, 0x49, 0x48, 0x58, 0x0f, 0xa7, 0xac, 
-   0x0b, 0x55, 0x6f, 0x28, 0x0e, 0x50, 0x74, 0x58, 0xd6, 0xb6, 0x5e, 0x94, 
-   0x29, 0xcb, 0x3a, 0xfb, 0xbe, 0x3d, 0x4f, 0xdd, 0xf8, 0x10, 0xfa, 0x68, 
-   0x13, 0x06, 0xbb, 0x62, 0xdf, 0x54, 0x32, 0x80, 0xba, 0xb0, 0xf0, 0xd0, 
-   0xf5, 0x61, 0x1b, 0x7b, 0xb5, 0xc8, 0xd9, 0xed, 0xb1, 0xda, 0xa6, 0x6e, 
-   0x14, 0x6c, 0x9f, 0xdc, 0x35, 0x53, 0x62, 0x3f, 0xd8, 0x06, 0x21, 0xea, 
-   0x1d, 0x36, 0x1b, 0x37, 0x4c, 0x79, 0xd9, 0x89, 0xcd, 0x4c, 0x61, 0x03, 
-   0xa8, 0xc0, 0x11, 0x39, 0xfe, 0x1a, 0x0d, 0xc7, 0x17, 0x19, 0xe7, 0x96, 
-   0xc0, 0x59, 0xc1, 0xfa, 0xe2, 0x80, 0xed, 0x91, 0xb2, 0x3c, 0xc4, 0x1e, 
-   0xb6, 0x01, 0x66, 0x6b, 0xe9, 0x00, 0x69, 0xe3, 0x94, 0x9b, 0x89, 0xa6, 
-   0xed, 0x4e, 0x26, 0xef, 0x26, 0x28, 0xef, 0x5c, 0xf3, 0x91, 0x13, 0x31, 
-   0xfb, 0x4e, 0x19, 0x58, 0x59, 0xfb, 0x46, 0xda, 0x7e, 0x19, 0x04, 0x5b, 
-   0xb0, 0xf1, 0xc0, 0x8b, 0x9c, 0x25, 0x91, 0xc1, 0xd6, 0x75, 0x42, 0x62, 
-   0xc0, 0x9f, 0x43, 0x2f, 0x06, 0x38, 0xd9, 0xc6, 0x9c, 0xee, 0x4f, 0x67, 
-   0xb1, 0xeb, 0xaf, 0xbc, 0x6d, 0x2a, 0xd1, 0x67, 0x98, 0xa3, 0x65, 0x3e, 
-   0x36, 0x51, 0x3f, 0x0a, 0x6e, 0xf5, 0x22, 0x37, 0x6e, 0x96, 0x80, 0x5d, 
-   0xd6, 0xb6, 0x81, 0xb3, 0xae, 0xa2, 0x39, 0x93, 0x95, 0x52, 0x55, 0x73, 
-   0x4a, 0x91, 0x85, 0xae, 0xb3, 0x7e, 0x33, 0x89, 0xc1, 0xcc, 0xaa, 0x09, 
-   0x8c, 0x19, 0xba, 0x20, 0x1e, 0x74, 0xe9, 0xfa, 0x6e, 0xe8, 0x6c, 0x33, 
-   0xf9, 0x74, 0x52, 0x81, 0x01, 0xc3, 0xba, 0x76, 0x77, 0x41, 0xf8, 0x8a, 
-   0x96, 0x4e, 0xe4, 0xa2, 0x44, 0x1e, 0x28, 0xd8, 0x60, 0xb7, 0x80, 0xff, 
-   0x7a, 0x1a, 0x13, 0xa6, 0xeb, 0xc3, 0x6e, 0x8f, 0x96, 0x89, 0xe0, 0x0b, 
-   0xc2, 0xb4, 0x61, 0xbf, 0xa3, 0x5a, 0x07, 0x84, 0xf7, 0x05, 0xe1, 0x0d, 
-   0x88, 0xf0, 0xdc, 0x58, 0x86, 0x77, 0x97, 0xb5, 0x6b, 0x27, 0xfa, 0x3d, 
-   0x19, 0x15, 0x1d, 0x21, 0xc0, 0xaf, 0x73, 0xc4, 0xba, 0x28, 0xe7, 0x78, 
-   0x56, 0x51, 0xcb, 0xe5, 0xdc, 0xda, 0xc7, 0xce, 0x77, 0x95, 0x3a, 0x84, 
-   0x88, 0xa8, 0xfe, 0x54, 0xc1, 0xbe, 0x75, 0xfc, 0xf4, 0xcd, 0x3a, 0x68, 
-   0x1e, 0x27, 0x0d, 0x33, 0xe6, 0x05, 0xf7, 0x58, 0xe4, 0x3e, 0xf1, 0x5e, 
-   0xdc, 0x6d, 0xc2, 0xfb, 0x22, 0x08, 0x77, 0x4e, 0x9c, 0xd9, 0x67, 0xc6, 
-   0x27, 0xb3, 0x81, 0x83, 0xe7, 0xc7, 0x89, 0xfd, 0x63, 0x36, 0x4f, 0xce, 
-   0xf6, 0xe0, 0x26, 0x6c, 0xc2, 0x87, 0xa5, 0xd3, 0xc1, 0xff, 0xfb, 0x80, 
-   0xe8, 0x6f, 0xb3, 0x30, 0xa9, 0x40, 0xb8, 0x7c, 0x08, 0x5b, 0xed, 0x16, 
-   0x5a, 0x7b, 0xf1, 0xa3, 0x1b, 0xba, 0xeb, 0x8c, 0xde, 0x2a, 0xa7, 0xf7, 
-   0x83, 0xa4, 0x49, 0xd6, 0xc0, 0xce, 0x84, 0x75, 0x4a, 0xe7, 0x03, 0x4d, 
-   0xee, 0x1a, 0x2f, 0xdd, 0xe2, 0x74, 0xda, 0xfa, 0xe9, 0xcc, 0xd8, 0x36, 
-   0x5a, 0xd7, 0x4e, 0x67, 0xe0, 0xae, 0x3c, 0x2c, 0x33, 0x17, 0xbd, 0x34, 
-   0xb5, 0x53, 0xe1, 0xb4, 0xc4, 0x6a, 0x2b, 0x4c, 0x43, 0xa5, 0x97, 0xa6, 
-   0x7e, 0x22, 0x53, 0xec, 0x33, 0x04, 0xa1, 0xa9, 0xa6, 0x31, 0xd7, 0x0e, 
-   0x7e, 0x34, 0xaf, 0x34, 0xe4, 0xc4, 0x89, 0xe2, 0x90, 0x8c, 0x7a, 0xd6, 
-   0x2c, 0xd4, 0x2a, 0xba, 0x67, 0xf5, 0xd8, 0x6f, 0x02, 0xdf, 0xd5, 0x8e, 
-   0x9c, 0x5a, 0xae, 0x76, 0xf4, 0xf7, 0xa7, 0x11, 0x5b, 0x9c, 0xc5, 0x49, 
-   0xe4, 0x29, 0x33, 0x8a, 0x86, 0x82, 0x22, 0xf1, 0x23, 0x19, 0x5d, 0xb3, 
-   0x20, 0x8e, 0xd2, 0x9d, 0x66, 0x44, 0xf6, 0xe5, 0xb2, 0x8d, 0xe6, 0xfc, 
-   0x4f, 0xdf, 0x68, 0x14, 0x1e, 0x93, 0x44, 0x14, 0x6f, 0xbb, 0x43, 0x28, 
-   0x38, 0x1e, 0xb9, 0x43, 0x28, 0xb8, 0x15, 0x76, 0x88, 0x77, 0x27, 0xf8, 
-   0xee, 0x04, 0xdf, 0x9d, 0xe0, 0x9f, 0xea, 0x04, 0xaf, 0xfc, 0xb5, 0xfb, 
-   0x82, 0x8f, 0xa7, 0x93, 0xd0, 0xdb, 0x79, 0xb1, 0xf7, 0xe4, 0xa2, 0x91, 
-   0x70, 0x72, 0xc6, 0x07, 0x72, 0xc8, 0xf7, 0x39, 0x2f, 0xde, 0xee, 0xb0, 
-   0x43, 0xb4, 0x2d, 0x70, 0x76, 0x3f, 0xb3, 0x25, 0x69, 0x42, 0x6e, 0xa3, 
-   0xcb, 0x5d, 0x26, 0xe6, 0xe2, 0xad, 0xdc, 0x28, 0xe9, 0xb7, 0xc0, 0xac, 
-   0x59, 0xd7, 0x32, 0x1b, 0xb9, 0xfe, 0x43, 0xfc, 0x08, 0xb4, 0xed, 0x68, 
-   0x9b, 0xd2, 0x29, 0x50, 0xa2, 0xb4, 0x79, 0xb3, 0x8a, 0x3d, 0x77, 0x6a, 
-   0x4b, 0x2f, 0xd6, 0x1a, 0x85, 0xd5, 0x96, 0xc8, 0x2c, 0x40, 0x43, 0x79, 
-   0x37, 0x98, 0x69, 0x63, 0x27, 0xf8, 0x8c, 0x66, 0x15, 0x73, 0xdd, 0x07, 
-   0xf8, 0x4f, 0x91, 0x76, 0x6c, 0x5b, 0xcf, 0x77, 0x23, 0xad, 0xc1, 0x12, 
-   0x2a, 0xec, 0xd5, 0x83, 0xbd, 0xd6, 0x60, 0x29, 0x65, 0x14, 0x87, 0xde, 
-   0x5e, 0x6b, 0xb9, 0x98, 0xc8, 0xf1, 0x1f, 0xb6, 0x42, 0xf7, 0x4d, 0x1d, 
-   0x65, 0x8e, 0x73, 0x4b, 0x4b, 0xbf, 0x71, 0xfc, 0x8c, 0xba, 0x5d, 0x6d, 
-   0x65, 0xdc, 0xb9, 0x61, 0x8c, 0x6d, 0xa3, 0x1b, 0x86, 0xce, 0x2b, 0x5f, 
-   0x1e, 0x51, 0xb6, 0x34, 0x1a, 0xd0, 0xd2, 0x60, 0xf6, 0x44, 0x4e, 0x85, 
-   0x5e, 0x88, 0xcf, 0x88, 0x8c, 0xc7, 0x0f, 0xb7, 0xea, 0x77, 0xf3, 0xf9, 
-   0xf3, 0xcd, 0x47, 0x76, 0xa8, 0x59, 0x30, 0x23, 0xa5, 0x7a, 0x5a, 0x60, 
-   0x92, 0x6e, 0xe0, 0xc4, 0x0e, 0x9a, 0x1f, 0xeb, 0x92, 0x98, 0xb3, 0x41, 
-   0x1e, 0xf3, 0xcd, 0x1a, 0xc5, 0x34, 0x6c, 0x4a, 0xfc, 0x72, 0xfa, 0x5a, 
-   0x14, 0x77, 0xc9, 0x76, 0xca, 0xe7, 0x06, 0x0f, 0xd3, 0xc0, 0xbe, 0x26, 
-   0xd4, 0xbe, 0x12, 0xa7, 0xaf, 0x31, 0xb2, 0x11, 0x31, 0x32, 0x99, 0x54, 
-   0x35, 0xa1, 0x79, 0xa6, 0x3a, 0x89, 0x5c, 0x65, 0x6e, 0xd3, 0x4f, 0x5f, 
-   0x65, 0xc2, 0xe2, 0xec, 0x4b, 0x95, 0x7c, 0x39, 0x42, 0xb3, 0x47, 0x67, 
-   0x8d, 0x43, 0x0b, 0xe9, 0x92, 0xa0, 0xdd, 0x4c, 0xf5, 0xfa, 0x1f, 0xb5, 
-   0x1a, 0x9a, 0x3f, 0xf2, 0x23, 0x02, 0x93, 0xe2, 0xb3, 0xb7, 0xc5, 0x91, 
-   0xb7, 0x8b, 0xff, 0x6b, 0xbb, 0xc5, 0x7b, 0xaf, 0xe7, 0xa3, 0xe5, 0x2b, 
-   0xfa, 0xdd, 0x0d, 0x7d, 0x1c, 0xfb, 0xe2, 0x81, 0x78, 0x6b, 0x87, 0xe6, 
-   0x08, 0x09, 0xb3, 0x9f, 0x51, 0xad, 0x56, 0x29, 0x1d, 0xa8, 0x77, 0x2c, 
-   0xc3, 0x17, 0x1c, 0x7c, 0xac, 0x71, 0xbf, 0x11, 0x1b, 0x7a, 0xe8, 0xae, 
-   0x82, 0x10, 0x88, 0xd7, 0xf5, 0x29, 0xdd, 0x9b, 0xc3, 0x6e, 0x89, 0x19, 
-   0x60, 0x77, 0xe7, 0xc4, 0x78, 0xd1, 0x2c, 0x0f, 0x38, 0x9a, 0x74, 0x88, 
-   0xbf, 0x8c, 0x74, 0x11, 0x96, 0x2e, 0x0b, 0xc8, 0x92, 0xc4, 0x34, 0x9a, 
-   0x8a, 0x72, 0x8b, 0xc6, 0xb2, 0x9a, 0xd0, 0xaa, 0x61, 0x2d, 0x66, 0x69, 
-   0x5c, 0xc5, 0xba, 0xe7, 0x62, 0x39, 0x6f, 0x6b, 0xe5, 0xc2, 0x38, 0xdc, 
-   0x5d, 0x22, 0x16, 0x4f, 0xe5, 0x39, 0x74, 0xf4, 0x7e, 0x97, 0x71, 0xb8, 
-   0x9f, 0x65, 0x6a, 0xe1, 0x4b, 0xa2, 0xdd, 0x34, 0x6c, 0x9d, 0x44, 0x90, 
-   0x8c, 0x41, 0x5b, 0xb5, 0xa6, 0xb4, 0xf2, 0x0b, 0xfc, 0x8d, 0xf7, 0x70, 
-   0x08, 0x99, 0x21, 0xf5, 0xbc, 0x38, 0xdb, 0xb7, 0xce, 0x41, 0x9f, 0x33, 
-   0x74, 0xc2, 0xed, 0x2b, 0xba, 0x47, 0x87, 0x3d, 0xb6, 0x3d, 0xbc, 0x7a, 
-   0x5c, 0x9f, 0x9c, 0xea, 0x8e, 0x3a, 0x11, 0xa6, 0xac, 0x94, 0x2c, 0xf4, 
-   0x89, 0x42, 0x83, 0x71, 0x18, 0xa4, 0x03, 0xdd, 0x7d, 0xfc, 0x58, 0x9b, 
-   0xbb, 0xc4, 0xf9, 0x1e, 0xfc, 0x15, 0x11, 0x44, 0xd1, 0x2a, 0x2d, 0x38, 
-   0x54, 0x25, 0xcb, 0x16, 0x2f, 0x59, 0xe6, 0x15, 0x22, 0x78, 0x09, 0xf6, 
-   0x83, 0x27, 0x37, 0x74, 0x1e, 0xc8, 0x15, 0x8f, 0xb3, 0x46, 0xe2, 0xd9, 
-   0x4e, 0x18, 0xa6, 0xf5, 0xed, 0xec, 0x27, 0xde, 0xde, 0x45, 0x33, 0x77, 
-   0xeb, 0xae, 0xe2, 0x62, 0x07, 0xfa, 0x23, 0x3f, 0x3b, 0x72, 0x79, 0x11, 
-   0x5e, 0xa7, 0x63, 0xcc, 0x31, 0xb9, 0x96, 0xd2, 0x1c, 0x45, 0x25, 0x6e, 
-   0xd5, 0x85, 0x21, 0xee, 0x0a, 0x7c, 0xac, 0xad, 0xa3, 0x65, 0xd1, 0xf5, 
-   0x63, 0x0f, 0xbb, 0x45, 0x27, 0x22, 0x07, 0x15, 0xb6, 0x71, 0x90, 0xcb, 
-   0xb3, 0x11, 0x0b, 0x3e, 0xf2, 0xfd, 0xe8, 0x13, 0xb9, 0x43, 0x6a, 0x55, 
-   0x88, 0x5a, 0x08, 0x1a, 0x6f, 0x36, 0x91, 0x0b, 0x88, 0x56, 0xef, 0xfe, 
-   0xfa, 0xdb, 0x60, 0xf5, 0xfb, 0x33, 0x16, 0xad, 0x14, 0x1e, 0x56, 0xbf, 
-   0xcf, 0x4a, 0x86, 0x33, 0x75, 0x89, 0x7e, 0x5c, 0x74, 0xe1, 0xac, 0x3c, 
-   0xff, 0x81, 0xf3, 0x3c, 0xe6, 0x7a, 0x26, 0x61, 0x89, 0xa3, 0x8e, 0x67, 
-   0x27, 0x5c, 0x1b, 0xb0, 0xac, 0x76, 0x49, 0x73, 0xb1, 0x75, 0x62, 0xb6, 
-   0xf3, 0x21, 0xfc, 0xe3, 0x03, 0x77, 0x2e, 0xe0, 0xdd, 0x15, 0xa1, 0xae, 
-   0x91, 0xcd, 0x86, 0x8c, 0x21, 0xa1, 0x57, 0x5f, 0xf8, 0x1a, 0xf9, 0x37, 
-   0x6a, 0x05, 0x94, 0x49, 0xd6, 0x75, 0x07, 0xea, 0x9a, 0xd1, 0xcd, 0x28, 
-   0x9d, 0xba, 0xcb, 0xcd, 0x36, 0x70, 0xf4, 0x7d, 0x12, 0x73, 0xc3, 0x5b, 
-   0xf7, 0x3a, 0xce, 0xae, 0xd3, 0xcf, 0xcf, 0xa1, 0x3e, 0x45, 0xba, 0x6f, 
-   0xed, 0x73, 0xfa, 0x69, 0x8e, 0xbe, 0xa0, 0x65, 0x70, 0xf0, 0xd7, 0x4e, 
-   0xf8, 0x9a, 0x6d, 0x81, 0xf5, 0x3a, 0xd4, 0x31, 0x21, 0xde, 0x67, 0xa1, 
-   0x99, 0xd8, 0x2c, 0xd1, 0x76, 0xbb, 0x38, 0x0c, 0x13, 0x61, 0xcb, 0xeb, 
-   0x24, 0x1d, 0x82, 0x25, 0xc6, 0x38, 0xf1, 0xa3, 0x1b, 0xb9, 0x2c, 0xba, 
-   0xc1, 0x6b, 0x33, 0x74, 0x71, 0xdf, 0xb8, 0x09, 0x9d, 0x63, 0xcd, 0xaa, 
-   0x75, 0x6a, 0x67, 0xd8, 0x93, 0x93, 0x86, 0x6b, 0x74, 0x12, 0x07, 0x7b, 
-   0x64, 0xb5, 0x11, 0x8e, 0x3b, 0xe9, 0xf1, 0xdf, 0x61, 0x54, 0x0d, 0xfb, 
-   0x23, 0xbc, 0xf6, 0xc5, 0xce, 0xd1, 0xad, 0x4f, 0x77, 0xb2, 0xc2, 0x74, 
-   0x2c, 0xfd, 0xee, 0x2e, 0xf1, 0xc1, 0xeb, 0x21, 0xe6, 0xbb, 0x2c, 0x24, 
-   0x17, 0xc3, 0x28, 0xc5, 0xdb, 0xa3, 0xcf, 0x38, 0xd4, 0x0e, 0x9e, 0xb9, 
-   0x5c, 0xc0, 0x94, 0x89, 0x40, 0x89, 0x3e, 0xb9, 0xde, 0xc3, 0x23, 0x89, 
-   0xd0, 0xd1, 0x9e, 0x64, 0x3b, 0xa1, 0xf9, 0x34, 0x4d, 0x62, 0x0d, 0xce, 
-   0xf1, 0x33, 0x31, 0xb6, 0x52, 0x86, 0x0d, 0x7d, 0x4a, 0x46, 0x64, 0xd8, 
-   0x0b, 0xe2, 0x38, 0xd8, 0xa1, 0x34, 0x1b, 0x8b, 0xc3, 0x43, 0xcf, 0x77, 
-   0x84, 0xcb, 0xf0, 0x4a, 0x92, 0x17, 0x19, 0x8f, 0xdc, 0x4d, 0x6c, 0xc2, 
-   0xb6, 0x9a, 0x1e, 0xee, 0x3c, 0xf7, 0x79, 0x1f, 0x84, 0x71, 0xc1, 0x46, 
-   0xc1, 0x33, 0x7a, 0x46, 0xdd, 0x77, 0xfd, 0x18, 0x9b, 0xea, 0xd7, 0xda, 
-   0xca, 0x6c, 0x86, 0x91, 0x65, 0xff, 0xdc, 0x04, 0xa7, 0x98, 0x67, 0xf9, 
-   0xa5, 0x9c, 0x65, 0x1d, 0xe0, 0x58, 0x32, 0xbd, 0x7b, 0xb4, 0xc3, 0xba, 
-   0x25, 0x3b, 0xde, 0xce, 0x79, 0xc1, 0xc1, 0xb0, 0xb7, 0xdf, 0x13, 0x4f, 
-   0xba, 0xdf, 0x3a, 0x3e, 0xcf, 0x47, 0x58, 0x75, 0x30, 0x2c, 0x4e, 0x73, 
-   0x75, 0xf7, 0xcf, 0x80, 0x33, 0xe2, 0x56, 0x21, 0x78, 0xa3, 0x3c, 0x03, 
-   0xcf, 0x57, 0x33, 0xa8, 0xe8, 0xcd, 0x88, 0x25, 0xec, 0x71, 0x20, 0xf2, 
-   0xe5, 0x2b, 0x9a, 0xad, 0xf0, 0x86, 0xce, 0x0b, 0xa4, 0xac, 0x7a, 0x56, 
-   0x58, 0x55, 0x26, 0xde, 0x4f, 0xce, 0x76, 0x53, 0xe3, 0xeb, 0xc7, 0x3a, 
-   0xb5, 0xda, 0xd8, 0xf0, 0xb1, 0x17, 0xa1, 0x86, 0x7f, 0xe4, 0x0c, 0x65, 
-   0xee, 0xd9, 0x5a, 0xd2, 0x33, 0x3f, 0x72, 0xf6, 0xf7, 0x74, 0xf2, 0x2e, 
-   0x55, 0x68, 0xde, 0x60, 0xdb, 0x26, 0x52, 0xb8, 0x4f, 0xdd, 0xd8, 0xc9, 
-   0xfd, 0x8a, 0xe4, 0xb4, 0xef, 0xa3, 0x8f, 0xdf, 0x3a, 0xf5, 0x74, 0x4c, 
-   0xa5, 0x1c, 0x2b, 0xce, 0x57, 0x2c, 0x84, 0xa3, 0x01, 0x27, 0x92, 0x4e, 
-   0x25, 0xfc, 0x3c, 0x67, 0x97, 0x4c, 0x7a, 0x10, 0x1c, 0x70, 0x20, 0x53, 
-   0x4b, 0x2e, 0x96, 0xb0, 0x5a, 0xfc, 0xc0, 0xaf, 0xed, 0x22, 0x29, 0xb1, 
-   0x25, 0xc6, 0xdb, 0x85, 0xaa, 0x8b, 0x22, 0x4b, 0x3a, 0xae, 0xee, 0x16, 
-   0x07, 0x6f, 0xc9, 0xe9, 0x88, 0x84, 0x71, 0x52, 0x8c, 0x60, 0x0b, 0x0c, 
-   0xcf, 0x24, 0x4f, 0x04, 0x5f, 0x0a, 0x51, 0x06, 0xa4, 0x29, 0x22, 0xd2, 
-   0xd2, 0x5d, 0x39, 0x71, 0xea, 0xb6, 0xfe, 0x96, 0x46, 0xa0, 0xb6, 0xec, 
-   0x8e, 0x36, 0xa1, 0x21, 0x90, 0xdb, 0xad, 0x76, 0xa5, 0xa4, 0x4e, 0x5e, 
-   0x2c, 0x57, 0x78, 0xaf, 0xc5, 0x51, 0xb7, 0x46, 0x3c, 0xad, 0x7f, 0x17, 
-   0xf1, 0x74, 0x0f, 0x71, 0x50, 0xf3, 0x98, 0x4c, 0x48, 0xc8, 0x4f, 0xc5, 
-   0x45, 0xb3, 0x3e, 0x88, 0x26, 0xec, 0x68, 0x9e, 0x18, 0xb2, 0x48, 0xfd, 
-   0x71, 0x44, 0xbc, 0x83, 0x6e, 0x37, 0x69, 0x46, 0x8e, 0xdd, 0x80, 0xd2, 
-   0x98, 0x05, 0xe2, 0xa9, 0x3f, 0x9b, 0xd0, 0x22, 0x44, 0xe1, 0xb0, 0x87, 
-   0x4e, 0x9a, 0x2f, 0x1f, 0x21, 0x56, 0x85, 0xe3, 0x49, 0x91, 0x57, 0xe2, 
-   0x70, 0x4f, 0xf0, 0xea, 0x23, 0x37, 0xdb, 0xdc, 0x37, 0xf0, 0xe4, 0x08, 
-   0xd9, 0x7d, 0x74, 0x71, 0x00, 0xf3, 0xac, 0x65, 0x5c, 0x8a, 0x49, 0x1a, 
-   0x85, 0xb0, 0xf2, 0x72, 0x47, 0xea, 0x42, 0x52, 0x83, 0xc4, 0x4b, 0xde, 
-   0xf2, 0x77, 0xec, 0x6e, 0x3f, 0x82, 0x4f, 0x0d, 0x06, 0xa1, 0x94, 0x82, 
-   0xa1, 0xa3, 0xaf, 0x76, 0x15, 0xe6, 0xad, 0xf3, 0xac, 0x53, 0x92, 0xc4, 
-   0x0b, 0x75, 0xbe, 0xb5, 0x51, 0x52, 0xc3, 0x50, 0xcd, 0xb7, 0x9e, 0xe9, 
-   0xf3, 0x36, 0x34, 0xf5, 0x53, 0xbb, 0x3f, 0x65, 0xff, 0xee, 0x07, 0x4f, 
-   0x68, 0xcd, 0xca, 0x03, 0x00, 0x6e, 0xfa, 0x04, 0x4e, 0xc2, 0x0d, 0xdd, 
-   0xd2, 0x5c, 0x10, 0x1a, 0x78, 0xa1, 0x4b, 0xb3, 0x38, 0xe8, 0x72, 0x7e, 
-   0x7a, 0x39, 0x04, 0x58, 0xea, 0x57, 0x1a, 0x4b, 0xa3, 0xa0, 0x2c, 0x5b, 
-   0x01, 0x67, 0x6d, 0xce, 0xf4, 0xcb, 0x2b, 0x39, 0x5f, 0x27, 0xc9, 0xc2, 
-   0xac, 0x90, 0xb3, 0xc0, 0x49, 0x7f, 0xf8, 0x4f, 0x0a, 0x49, 0x54, 0x97, 
-   0xea, 0x67, 0x06, 0xb9, 0xf4, 0x1f, 0x73, 0xab, 0x0e, 0x9c, 0xa8, 0x92, 
-   0x02, 0x02, 0x45, 0x32, 0xa9, 0x70, 0x86, 0x01, 0x12, 0xc8, 0xd8, 0xe0, 
-   0x3e, 0x0d, 0xa6, 0xac, 0x86, 0x32, 0x75, 0x81, 0xe5, 0xd5, 0x1f, 0x6d, 
-   0x83, 0x7a, 0x89, 0x30, 0x5b, 0x1a, 0x2b, 0xba, 0x2a, 0x90, 0xaa, 0xcc, 
-   0x43, 0x25, 0xaa, 0x02, 0x83, 0x42, 0x41, 0x89, 0x4a, 0x84, 0xaa, 0xae, 
-   0x8f, 0xab, 0x18, 0xa9, 0xba, 0x33, 0xb4, 0xf5, 0xab, 0xca, 0x70, 0x63, 
-   0x68, 0xeb, 0x17, 0x53, 0xba, 0x2d, 0xb0, 0x43, 0x27, 0xe7, 0x51, 0xe9, 
-   0x18, 0x9b, 0x6c, 0x0a, 0x6a, 0x1e, 0x06, 0xee, 0x36, 0x59, 0x42, 0xc7, 
-   0x3f, 0x26, 0x80, 0x1d, 0x2c, 0x3f, 0xa0, 0xf2, 0x43, 0x96, 0xd5, 0x2a, 
-   0xf1, 0xa7, 0xcc, 0x2d, 0xe3, 0x53, 0x2e, 0xbb, 0x17, 0x01, 0xae, 0x10, 
-   0xcc, 0x76, 0x0e, 0xac, 0xe2, 0xc3, 0xce, 0x57, 0x72, 0xa9, 0x76, 0x36, 
-   0xbe, 0x74, 0x77, 0x78, 0xab, 0x48, 0xb7, 0xa2, 0x6c, 0x26, 0x76, 0xab, 
-   0x59, 0x12, 0x75, 0x27, 0x5b, 0x82, 0x75, 0xdc, 0xf6, 0x97, 0xb4, 0xae, 
-   0x1f, 0x93, 0xe5, 0x8b, 0xe2, 0xf0, 0xb0, 0xca, 0x6a, 0xc1, 0xd9, 0x05, 
-   0xd5, 0x94, 0x5d, 0x50, 0x41, 0x99, 0x45, 0xec, 0xc5, 0xd9, 0xd3, 0x19, 
-   0x46, 0xe9, 0x45, 0xb8, 0x4b, 0x72, 0xff, 0x87, 0xe2, 0x47, 0x52, 0x8c, 
-   0xee, 0x9a, 0x95, 0xa3, 0x2b, 0x93, 0x86, 0x78, 0x53, 0x5c, 0x6d, 0x0f, 
-   0x6b, 0x76, 0x47, 0x47, 0xaf, 0xcb, 0xd6, 0xe8, 0x89, 0xd5, 0x05, 0xe0, 
-   0x3d, 0xc9, 0xf9, 0x86, 0xcc, 0x6c, 0x3f, 0x39, 0xc5, 0x1b, 0xe4, 0x8b, 
-   0xf5, 0xd3, 0xe6, 0xf7, 0x70, 0xb7, 0xbe, 0xb7, 0xc1, 0x8e, 0x33, 0x42, 
-   0x27, 0x7e, 0x10, 0xa3, 0x03, 0x49, 0x98, 0xaf, 0x0e, 0x21, 0x76, 0x4a, 
-   0xf1, 0xf6, 0x15, 0x5a, 0x5d, 0xf6, 0x52, 0xab, 0x4d, 0x75, 0x67, 0x77, 
-   0x4e, 0xf8, 0x8a, 0xa7, 0x10, 0x15, 0x4d, 0xb4, 0x51, 0x9d, 0x6d, 0x9f, 
-   0xf8, 0x22, 0xf5, 0x12, 0x6e, 0x2e, 0xc1, 0x35, 0x9c, 0xa4, 0x1f, 0xc9, 
-   0x61, 0xf8, 0xb0, 0x27, 0xb7, 0xa8, 0xd8, 0xc5, 0xa6, 0x37, 0xa9, 0x60, 
-   0x3a, 0x31, 0xdf, 0x6d, 0x26, 0x30, 0x75, 0xd7, 0x9d, 0xc2, 0x64, 0x80, 
-   0x83, 0x34, 0x33, 0x8a, 0x6f, 0xd6, 0x87, 0x65, 0xa0, 0x10, 0xb9, 0xaf, 
-   0x6e, 0x76, 0xf3, 0xca, 0x22, 0xe1, 0x88, 0x85, 0x38, 0x4b, 0x21, 0x51, 
-   0x2a, 0x84, 0xe9, 0xcd, 0xaa, 0xec, 0xe7, 0x41, 0x8c, 0xcf, 0x80, 0x59, 
-   0x27, 0x91, 0x74, 0x10, 0x14, 0x18, 0xb7, 0xaa, 0x32, 0xd6, 0xe8, 0xdb, 
-   0x6a, 0xbf, 0x91, 0xc2, 0xe5, 0x5e, 0x0d, 0xd4, 0x2d, 0xf4, 0xac, 0xd6, 
-   0x37, 0xdf, 0x1b, 0xde, 0x60, 0x0d, 0x1a, 0x28, 0xa5, 0xd8, 0x5f, 0x25, 
-   0xbd, 0xdb, 0xc5, 0x39, 0x19, 0x74, 0x61, 0xa8, 0x7b, 0xfb, 0xec, 0x18, 
-   0xe6, 0x1a, 0xfd, 0xdb, 0x9d, 0x37, 0xd2, 0x7f, 0xb1, 0x67, 0x03, 0x1b, 
-   0x68, 0x40, 0xcb, 0xf0, 0x97, 0x53, 0xb6, 0x39, 0x01, 0x1b, 0x15, 0x57, 
-   0x86, 0x7a, 0xaf, 0x2a, 0x09, 0x4f, 0x94, 0x53, 0x85, 0x8b, 0x2d, 0x7a, 
-   0xaf, 0x44, 0x19, 0x3b, 0xcf, 0x3f, 0x44, 0x7c, 0x8b, 0xee, 0x00, 0xae, 
-   0x52, 0xa9, 0x8e, 0x19, 0x1e, 0xee, 0x1a, 0x50, 0x65, 0xe5, 0x55, 0x7c, 
-   0x37, 0xb9, 0xc6, 0x23, 0x12, 0x2f, 0x50, 0x39, 0xb3, 0xa3, 0x6c, 0xae, 
-   0x8c, 0x21, 0x64, 0x67, 0x82, 0x52, 0x7e, 0x39, 0xa5, 0xcf, 0x87, 0x7f, 
-   0xfb, 0xe9, 0x17, 0xe1, 0x19, 0x71, 0xe3, 0x67, 0x72, 0x07, 0x40, 0x08, 
-   0x5d, 0xff, 0xb0, 0xcb, 0xba, 0xdd, 0xed, 0x1d, 0xfa, 0x9c, 0x32, 0xad, 
-   0x0b, 0xd8, 0x87, 0xee, 0xc6, 0x7b, 0xc1, 0xb3, 0x6b, 0x0c, 0x16, 0xfd, 
-   0xf1, 0xf5, 0xa4, 0x3b, 0x1d, 0x2e, 0x2e, 0x6e, 0x6f, 0xfa, 0xa9, 0x22, 
-   0xa5, 0x42, 0xe0, 0xe1, 0xdd, 0x70, 0x5a, 0x3c, 0x42, 0xc8, 0x27, 0xa9, 
-   0xe1, 0x6c, 0x06, 0x14, 0x7c, 0x8a, 0x24, 0xc3, 0xff, 0xbd, 0xed, 0x8e, 
-   0x80, 0xd3, 0x80, 0xcc, 0x46, 0x22, 0x6a, 0x80, 0x44, 0x97, 0xd3, 0x61, 
-   0x77, 0x2e, 0x8c, 0xa8, 0x09, 0x52, 0xdd, 0x8c, 0xe7, 0x32, 0xb3, 0x16, 
-   0xcc, 0x4c, 0x26, 0x6a, 0x83, 0x44, 0xdd, 0xd1, 0xe7, 0xee, 0x57, 0x3e, 
-   0xbf, 0xf4, 0x99, 0xf2, 0x29, 0x11, 0x71, 0x41, 0xd8, 0xbd, 0x2d, 0x3e, 
-   0x12, 0x65, 0x37, 0x6e, 0xa2, 0xa0, 0x7b, 0xa3, 0xe1, 0xcd, 0x60, 0x71, 
-   0xd1, 0xed, 0xcf, 0xc7, 0x53, 0x48, 0xd0, 0xf7, 0xc3, 0xe9, 0x58, 0x23, 
-   0xe7, 0xf1, 0xcd, 0x50, 0x23, 0xe6, 0xd9, 0xb4, 0x8f, 0x35, 0x3a, 0x1a, 
-   0x4f, 0x35, 0xa2, 0xbe, 0xba, 0xb9, 0x5b, 0x14, 0x69, 0x61, 0x89, 0x0f, 
-   0x66, 0xf3, 0x1c, 0x1d, 0x2c, 0x73, 0xc2, 0xb3, 0x48, 0x0b, 0x0b, 0x9e, 
-   0xf4, 0xdd, 0x1d, 0x4d, 0x3e, 0x75, 0x35, 0xb2, 0x4f, 0xc7, 0x29, 0xd3, 
-   0x9e, 0x29, 0xc7, 0x29, 0xd3, 0x75, 0x4a, 0xc7, 0x29, 0xd3, 0x9e, 0x83, 
-   0xb4, 0xfd, 0xf1, 0x4d, 0x61, 0x46, 0x16, 0xac, 0x1c, 0xc2, 0x16, 0x24, 
-   0x87, 0x35, 0xc5, 0x48, 0xe5, 0x31, 0x58, 0x6a, 0x65, 0x81, 0xe4, 0xb0, 
-   0xbe, 0x32, 0x79, 0x2d, 0x66, 0xdd, 0xf9, 0xed, 0x14, 0xaf, 0x16, 0xde, 
-   0xa2, 0x69, 0x60, 0xbc, 0x2c, 0xc1, 0x51, 0x34, 0xdd, 0xeb, 0xf1, 0x60, 
-   0x08, 0x19, 0x6e, 0x77, 0x30, 0xd0, 0xd8, 0xed, 0xec, 0xb6, 0xa7, 0xb1, 
-   0xdb, 0xa9, 0x48, 0x02, 0x4b, 0xe1, 0xfa, 0xea, 0x46, 0x63, 0xa8, 0xd7, 
-   0xdd, 0x2f, 0x1a, 0x13, 0xbd, 0xbe, 0xd5, 0x79, 0x84, 0x59, 0x7f, 0x3a, 
-   0x1c, 0xde, 0x68, 0xac, 0x72, 0xd0, 0x9d, 0xfe, 0x53, 0x20, 0x82, 0xcd, 
-   0x71, 0x74, 0x75, 0xf9, 0x69, 0x2e, 0x50, 0x75, 0x4a, 0x45, 0x9f, 0xbe, 
-   0x50, 0x1a, 0xef, 0x65, 0xd1, 0xcf, 0x30, 0x8b, 0xfe, 0xd5, 0x68, 0x31, 
-   0x9e, 0x1c, 0xe9, 0x33, 0xfe, 0x39, 0x1c, 0x4e, 0x74, 0xc2, 0x1f, 0x4e, 
-   0x46, 0xdd, 0xfe, 0x50, 0xeb, 0x32, 0xfa, 0x5a, 0x4f, 0x31, 0xec, 0x1b, 
-   0x38, 0x89, 0xe1, 0x74, 0xae, 0x51, 0x02, 0xee, 0xea, 0xf3, 0xb4, 0x3b, 
-   0xd1, 0x69, 0x61, 0x28, 0x53, 0x95, 0x3b, 0x66, 0xa1, 0x1a, 0x48, 0x94, 
-   0xef, 0x64, 0x7a, 0x75, 0x0d, 0x49, 0x76, 0x32, 0xbe, 0xba, 0x99, 0xcf, 
-   0x34, 0xb2, 0x1d, 0x5d, 0xdd, 0x0c, 0x75, 0x1b, 0x1f, 0xa1, 0x59, 0x8c, 
-   0xc6, 0xe3, 0x89, 0x46, 0xbc, 0x94, 0x6e, 0x36, 0x9f, 0x5e, 0x4d, 0x34, 
-   0x42, 0xc6, 0x24, 0xdd, 0x9b, 0xcb, 0x91, 0xd0, 0x31, 0x2c, 0xe9, 0x94, 
-   0x2e, 0xc7, 0x14, 0x96, 0x78, 0x46, 0x7c, 0xd1, 0xd5, 0x19, 0x3f, 0x93, 
-   0xcc, 0x62, 0x7e, 0xc1, 0x27, 0x0e, 0x13, 0x52, 0xe9, 0x48, 0x74, 0xaa, 
-   0x85, 0x92, 0x48, 0x48, 0xa2, 0x85, 0x7d, 0x37, 0x97, 0x92, 0x44, 0x0c, 
-   0x3b, 0xef, 0x4c, 0x52, 0x22, 0xad, 0x0d, 0xab, 0x52, 0x96, 0x96, 0xd4, 
-   0x00, 0xd6, 0xab, 0x28, 0x31, 0x89, 0xdc, 0x2e, 0x9a, 0x61, 0x92, 0xa2, 
-   0x4a, 0x81, 0x5c, 0xd2, 0x8c, 0x22, 0x41, 0x77, 0x61, 0xb4, 0x12, 0x81, 
-   0x95, 0x12, 0x10, 0x74, 0x17, 0xe0, 0xef, 0xcd, 0xf4, 0xef, 0x0c, 0xdd, 
-   0x05, 0xa0, 0x68, 0x49, 0x14, 0x02, 0xac, 0x0b, 0x40, 0xdb, 0xfe, 0x90, 
-   0x79, 0x20, 0x27, 0x8f, 0xe7, 0x02, 0x90, 0x9f, 0xa5, 0xe4, 0x10, 0x90, 
-   0x0b, 0x40, 0xdf, 0x49, 0xe9, 0x8b, 0x40, 0x2e, 0x00, 0xf5, 0xb9, 0x44, 
-   0x8d, 0xcf, 0x2d, 0x64, 0xad, 0x3e, 0x79, 0x01, 0x0e, 0xfd, 0xe9, 0x5b, 
-   0x53, 0x50, 0x5c, 0x99, 0x40, 0x93, 0x64, 0xd3, 0x7d, 0x2d, 0xf0, 0xb7, 
-   0xaf, 0x28, 0xcb, 0x4d, 0x83, 0x10, 0x26, 0x69, 0x86, 0x33, 0x6d, 0x9b, 
-   0xbd, 0x15, 0xad, 0xd0, 0xd8, 0xe6, 0x1d, 0xaf, 0xc9, 0x49, 0x25, 0xdf, 
-   0x96, 0xa4, 0xd1, 0xd4, 0xc3, 0x6e, 0xe4, 0x5a, 0xf3, 0x2e, 0x4f, 0x01, 
-   0xab, 0xb1, 0x9a, 0x05, 0xd1, 0xc4, 0xa1, 0xe3, 0x47, 0xe4, 0x6c, 0x87, 
-   0x36, 0xae, 0xbb, 0x5e, 0x62, 0xea, 0x92, 0x1c, 0x27, 0xbb, 0xd7, 0x5e, 
-   0x05, 0x07, 0x1f, 0x38, 0x6a, 0x98, 0xe6, 0x58, 0x93, 0xa1, 0x64, 0xe6, 
-   0xc5, 0x01, 0x70, 0x1c, 0x72, 0x5f, 0xbc, 0x7a, 0x74, 0x3c, 0x9f, 0x54, 
-   0xda, 0xa7, 0x88, 0x34, 0xc0, 0x38, 0x0c, 0x6e, 0x05, 0xf5, 0x69, 0xeb, 
-   0x0c, 0xde, 0x47, 0x1a, 0xc7, 0xf7, 0xee, 0xee, 0xac, 0x38, 0xed, 0xd9, 
-   0x0f, 0x9a, 0x69, 0xb6, 0x8a, 0x40, 0xe0, 0x1f, 0xc8, 0xba, 0xce, 0x65, 
-   0x17, 0x70, 0xd7, 0x1f, 0xa0, 0x95, 0xb3, 0x7a, 0x74, 0x41, 0xe3, 0xb2, 
-   0xeb, 0xb2, 0x13, 0x10, 0xde, 0x2b, 0xd2, 0x57, 0xe2, 0xde, 0x8a, 0xf9, 
-   0x85, 0x91, 0x62, 0xaa, 0x74, 0x5e, 0xc6, 0x13, 0xcd, 0x35, 0xc6, 0x66, 
-   0x5f, 0x9a, 0x78, 0x37, 0x15, 0x92, 0x6d, 0x15, 0xf5, 0x73, 0xb5, 0xdb, 
-   0x6f, 0xbd, 0x95, 0x17, 0x97, 0x0f, 0x9f, 0x3e, 0x60, 0x27, 0x92, 0xa4, 
-   0x89, 0x16, 0xff, 0x1b, 0x2f, 0x22, 0x92, 0xd1, 0xd8, 0xc5, 0xd1, 0x0c, 
-   0x5f, 0x92, 0xd1, 0xcc, 0x0e, 0x7b, 0x37, 0x24, 0xbd, 0x82, 0x99, 0x93, 
-   0x25, 0x85, 0xcd, 0x91, 0x1e, 0xda, 0xbb, 0x69, 0xcb, 0x28, 0x6d, 0x99, 
-   0xb8, 0x0b, 0xb4, 0x85, 0xdf, 0x91, 0xda, 0x46, 0x38, 0x39, 0x00, 0x5b, 
-   0xd5, 0xec, 0x0d, 0x2a, 0x1f, 0xc3, 0xe0, 0x59, 0xd9, 0x5c, 0x7f, 0x17, 
-   0xb4, 0x62, 0xd7, 0x40, 0x6f, 0x24, 0xfd, 0xcc, 0xb9, 0x66, 0xa2, 0x96, 
-   0x2e, 0xb8, 0x4a, 0x47, 0x4f, 0x6f, 0x3f, 0xd2, 0x66, 0x40, 0x7a, 0xb1, 
-   0xe2, 0x5c, 0x74, 0xec, 0x2a, 0xce, 0xac, 0xc9, 0x97, 0x2a, 0x87, 0xdf, 
-   0x9a, 0x19, 0x00, 0x7a, 0x01, 0x60, 0x5b, 0x1c, 0x8a, 0xeb, 0x08, 0x96, 
-   0xe8, 0xc4, 0x4d, 0x9e, 0x56, 0x7d, 0x2c, 0xab, 0xba, 0xd0, 0x62, 0x4d, 
-   0x55, 0x11, 0x6c, 0xba, 0xc3, 0xab, 0xd0, 0xaa, 0x0c, 0xde, 0xeb, 0x30, 
-   0x0e, 0xe9, 0x71, 0x4b, 0xc1, 0x47, 0xfd, 0x34, 0x84, 0x5c, 0x4a, 0xb7, 
-   0xd0, 0xc1, 0xa7, 0x89, 0x6e, 0xb8, 0x84, 0x5b, 0x98, 0xf1, 0x21, 0x05, 
-   0x89, 0x38, 0x89, 0x3e, 0x72, 0xec, 0x1c, 0x45, 0xa7, 0xe6, 0x90, 0x3c, 
-   0x2a, 0xc4, 0x1f, 0x05, 0x63, 0x73, 0xb4, 0xb2, 0x7b, 0x53, 0x96, 0xe6, 
-   0xb0, 0x65, 0x1b, 0x82, 0x1e, 0x15, 0xec, 0x49, 0xdd, 0xe5, 0xc1, 0xc7, 
-   0xce, 0x0a, 0xef, 0x25, 0x94, 0x27, 0x0a, 0x0e, 0xb4, 0xaa, 0x75, 0x27, 
-   0xdc, 0xd0, 0x2b, 0xca, 0x1c, 0xb5, 0x7d, 0x99, 0x20, 0x54, 0x99, 0x3e, 
-   0xf9, 0xc8, 0x00, 0xe5, 0xb2, 0x1d, 0x77, 0x9b, 0x21, 0xa7, 0x24, 0x2a, 
-   0x95, 0x97, 0x94, 0xde, 0x04, 0x7a, 0x45, 0x9c, 0x90, 0xa3, 0x8c, 0x1e, 
-   0x06, 0x1a, 0xa9, 0x6c, 0xf3, 0x30, 0x1b, 0xc0, 0xe4, 0xcb, 0xe4, 0x73, 
-   0x96, 0x8b, 0x5e, 0xa9, 0x7c, 0xae, 0x9d, 0xf0, 0x77, 0x8d, 0xaf, 0x39, 
-   0x97, 0x7d, 0x0d, 0x0c, 0x48, 0xa3, 0x92, 0x6b, 0xd9, 0x35, 0x96, 0xc9, 
-   0x7e, 0x37, 0xc1, 0x7f, 0xc4, 0xcb, 0xf6, 0x91, 0x95, 0x72, 0xe0, 0x51, 
-   0x63, 0x1b, 0xdc, 0x13, 0x33, 0xdc, 0x91, 0xf2, 0x28, 0x6c, 0x8c, 0xb7, 
-   0x57, 0x17, 0x88, 0x56, 0x4d, 0xf2, 0x57, 0x05, 0xc2, 0xeb, 0x39, 0xf8, 
-   0x79, 0x9b, 0xd8, 0xc1, 0x97, 0xf1, 0x94, 0xf0, 0x00, 0x6c, 0x58, 0x7f, 
-   0x0f, 0x6f, 0x62, 0xc4, 0xe6, 0x98, 0x37, 0x95, 0xbd, 0x8f, 0x39, 0xf8, 
-   0x4d, 0x45, 0xef, 0x63, 0x8e, 0x94, 0x66, 0xec, 0x7d, 0xf4, 0x75, 0x3f, 
-   0x53, 0xe7, 0x59, 0x51, 0xee, 0xa6, 0xf7, 0xb2, 0x0a, 0xf8, 0x94, 0x4a, 
-   0x0f, 0xbe, 0x59, 0xb9, 0x22, 0xc2, 0xf4, 0x0f, 0xd8, 0xf6, 0xf5, 0x50, 
-   0x36, 0x32, 0xbd, 0xa5, 0x2d, 0xbc, 0x92, 0xe9, 0xed, 0x62, 0x6e, 0xa9, 
-   0x94, 0xbe, 0xa1, 0x85, 0x1b, 0x90, 0xe0, 0x5f, 0x3a, 0x0a, 0x22, 0x0e, 
-   0xd7, 0x72, 0xae, 0xa0, 0x98, 0xe5, 0x80, 0x5f, 0x2c, 0xd5, 0xfc, 0xef, 
-   0xff, 0xab, 0x40, 0x0a, 0x01, 0x87, 0x94, 0x78, 0xa4, 0x46, 0x76, 0x86, 
-   0x29, 0x41, 0xba, 0x32, 0x74, 0x2c, 0xc2, 0x4b, 0x66, 0xb3, 0x40, 0xfa, 
-   0x2f, 0xed, 0x58, 0xde, 0x66, 0x2d, 0x8c, 0x84, 0x5d, 0xe3, 0x7d, 0x29, 
-   0xfc, 0xb5, 0x97, 0x82, 0xcd, 0x93, 0x74, 0x0a, 0xbc, 0x23, 0x55, 0x9a, 
-   0x28, 0x7d, 0xff, 0xe4, 0x29, 0xc0, 0x8e, 0x4c, 0x2a, 0xdf, 0x93, 0x70, 
-   0x83, 0xf7, 0x89, 0xcf, 0x10, 0xa4, 0x35, 0xf0, 0x5a, 0x98, 0xbc, 0x59, 
-   0xd3, 0x96, 0x70, 0x26, 0x6f, 0xba, 0xe4, 0x21, 0x41, 0x55, 0xc7, 0x6f, 
-   0x8a, 0xbf, 0x54, 0x8a, 0x88, 0x6a, 0x0c, 0x56, 0x53, 0x6c, 0x5b, 0xc4, 
-   0x60, 0xaa, 0x04, 0x8e, 0xc6, 0x9b, 0x23, 0x33, 0x38, 0x26, 0xa1, 0x81, 
-   0x0a, 0x99, 0x49, 0xdd, 0x82, 0x21, 0xa6, 0x54, 0x2a, 0xe2, 0x15, 0x63, 
-   0xf8, 0x56, 0xa9, 0x9f, 0x30, 0x34, 0xe6, 0x66, 0xde, 0x98, 0xaf, 0x7c, 
-   0xcc, 0xd4, 0x5f, 0x55, 0x31, 0xeb, 0x2a, 0x26, 0xd9, 0x30, 0xa8, 0x2a, 
-   0xd6, 0x98, 0xa4, 0xc9, 0x2a, 0x31, 0xb7, 0xc9, 0x33, 0x38, 0x15, 0x96, 
-   0xe3, 0xc7, 0x0b, 0x66, 0x52, 0xf9, 0x80, 0x85, 0x85, 0x7a, 0xc0, 0x94, 
-   0xb4, 0x39, 0x7a, 0xb7, 0xf2, 0xa3, 0xad, 0x9c, 0x5f, 0x36, 0xea, 0x0e, 
-   0x52, 0x8d, 0xec, 0xa0, 0xa9, 0x46, 0xe0, 0x52, 0x59, 0xf5, 0x9f, 0x02, 
-   0xbf, 0x55, 0x2e, 0x0b, 0x20, 0xad, 0x55, 0x41, 0x16, 0x1d, 0x50, 0x16, 
-   0xc0, 0x7a, 0x7f, 0x0b, 0xa9, 0x18, 0xec, 0x1b, 0x7f, 0xad, 0x35, 0xf5, 
-   0x66, 0x72, 0x6e, 0x66, 0xd9, 0xd1, 0x1e, 0x01, 0x74, 0x4d, 0x84, 0x4d, 
-   0x7f, 0x4e, 0x07, 0x5a, 0x76, 0x93, 0x24, 0xd3, 0x19, 0x94, 0xcc, 0x2b, 
-   0x78, 0x28, 0x35, 0x53, 0x2d, 0x23, 0xda, 0xca, 0x56, 0x10, 0x0c, 0x42, 
-   0x06, 0x4c, 0x81, 0x54, 0xb3, 0x4b, 0xce, 0x09, 0x38, 0xad, 0xaa, 0x9d, 
-   0x13, 0x65, 0x9e, 0x6c, 0x46, 0x06, 0x68, 0x5e, 0x98, 0x38, 0x01, 0x1e, 
-   0xfb, 0x72, 0xfa, 0xb5, 0x0a, 0x2c, 0x22, 0x07, 0xb7, 0x10, 0x87, 0xaa, 
-   0x42, 0x0d, 0x50, 0xf7, 0x3e, 0x31, 0x03, 0xb7, 0xa3, 0xb4, 0x23, 0x23, 
-   0x84, 0x3b, 0x4a, 0x3a, 0x2f, 0x20, 0xd2, 0x41, 0xaf, 0xa8, 0x4a, 0x14, 
-   0xd7, 0xce, 0x76, 0xf6, 0x3c, 0xb0, 0x98, 0xfe, 0x36, 0x8d, 0xe4, 0x9f, 
-   0xf2, 0xc9, 0x80, 0xb2, 0xdd, 0xd1, 0xd7, 0x62, 0x77, 0xa9, 0x62, 0x15, 
-   0x9d, 0xf9, 0x9d, 0x65, 0xb3, 0x98, 0x67, 0xf7, 0xb1, 0x17, 0xc9, 0x7d, 
-   0x2c, 0x62, 0x71, 0x07, 0x34, 0x1f, 0x3e, 0xa0, 0xc4, 0x38, 0xc6, 0xec, 
-   0xc4, 0x98, 0x1c, 0xb8, 0xd2, 0xe8, 0x60, 0x13, 0x6c, 0xb7, 0xc1, 0xb3, 
-   0xf0, 0x72, 0xa2, 0xa1, 0xf6, 0x18, 0x25, 0x7d, 0x24, 0x30, 0x79, 0x49, 
-   0x1f, 0xf4, 0x25, 0xe9, 0x6c, 0xef, 0xae, 0x00, 0xfe, 0x5c, 0x0c, 0xd0, 
-   0xa9, 0xb4, 0x4c, 0x0e, 0xb9, 0xa2, 0x03, 0x40, 0x1a, 0xa4, 0xdf, 0x34, 
-   0x2b, 0x28, 0x15, 0x14, 0x03, 0xb4, 0xf9, 0xa1, 0x82, 0xef, 0x61, 0xa8, 
-   0x83, 0x9f, 0xb1, 0x87, 0x2a, 0x89, 0x17, 0xbb, 0xa3, 0x66, 0x8a, 0xcf, 
-   0xab, 0xec, 0x04, 0x6e, 0x74, 0xdd, 0xa2, 0xf4, 0xfa, 0xab, 0xc0, 0x8f, 
-   0xdc, 0xd5, 0x81, 0xae, 0x41, 0x91, 0x7f, 0xc4, 0x3b, 0x40, 0x4e, 0x94, 
-   0x6a, 0xf0, 0x89, 0xfd, 0x29, 0x57, 0xa9, 0xdc, 0xac, 0xa0, 0x30, 0xd9, 
-   0x06, 0x70, 0x1f, 0x2c, 0x53, 0x1e, 0x07, 0xc5, 0x20, 0x0a, 0x82, 0xf2, 
-   0x2a, 0x29, 0xd9, 0x56, 0x4b, 0x38, 0x4d, 0x7d, 0xbc, 0xc9, 0x0d, 0xb6, 
-   0xd8, 0xbf, 0x5c, 0x06, 0x22, 0xdc, 0x30, 0xb3, 0xec, 0x21, 0x7b, 0xc1, 
-   0x0b, 0x75, 0x9b, 0x52, 0xd0, 0x2f, 0xa6, 0x28, 0xe0, 0xe9, 0x0c, 0xde, 
-   0x65, 0xf5, 0xc8, 0x24, 0x93, 0x5e, 0x8a, 0xf7, 0x17, 0xfa, 0xef, 0x55, 
-   0x5c, 0x84, 0x81, 0x1f, 0xab, 0x19, 0xe8, 0x53, 0xb1, 0xe9, 0x3c, 0x26, 
-   0x0e, 0x5e, 0xc8, 0xe3, 0x7d, 0x71, 0x05, 0xdb, 0x99, 0xeb, 0x12, 0xaa, 
-   0xff, 0xc0, 0x04, 0x29, 0x45, 0xf3, 0x61, 0xe0, 0x6e, 0x4e, 0x52, 0x24, 
-   0x58, 0x60, 0x66, 0x1b, 0x32, 0x4b, 0xff, 0x5c, 0xce, 0x8e, 0x43, 0x4b, 
-   0x54, 0x62, 0xa7, 0x86, 0x9e, 0xcb, 0x36, 0xd6, 0x42, 0x39, 0xba, 0x96, 
-   0xab, 0x02, 0xa1, 0xd0, 0xa0, 0xee, 0x9f, 0x31, 0x99, 0xba, 0x1b, 0xb6, 
-   0x6e, 0x8d, 0xbc, 0x41, 0x89, 0x8f, 0xeb, 0x64, 0xbe, 0x9e, 0x55, 0xca, 
-   0xaa, 0x6d, 0xf8, 0xee, 0x52, 0x7e, 0x76, 0x1d, 0x15, 0x57, 0xb0, 0xad, 
-   0x9f, 0x40, 0x9a, 0xc4, 0xa7, 0x9d, 0xad, 0xb1, 0x24, 0x36, 0x12, 0x92, 
-   0x93, 0x90, 0xa5, 0xcc, 0xe6, 0x21, 0xd5, 0x9f, 0x6b, 0x79, 0x46, 0xe1, 
-   0x4a, 0xc9, 0x93, 0x6b, 0xac, 0x1a, 0x4f, 0x18, 0x84, 0x97, 0xfb, 0x2b, 
-   0xa1, 0xc8, 0x18, 0x3c, 0x17, 0x6f, 0xf7, 0x8f, 0x8e, 0xc1, 0x94, 0x3b, 
-   0xc6, 0xa3, 0x13, 0x39, 0x96, 0x4d, 0xb8, 0x79, 0x14, 0x47, 0x0d, 0xe6, 
-   0x70, 0x7e, 0xb6, 0x65, 0xd6, 0xd5, 0x2e, 0x58, 0x17, 0x09, 0xb2, 0xe3, 
-   0xe4, 0x2e, 0x07, 0xcc, 0x5b, 0xd3, 0x81, 0x9c, 0x5c, 0x58, 0xed, 0x23, 
-   0x1f, 0x24, 0xf7, 0xc8, 0x6e, 0xa9, 0x6a, 0x6f, 0xe0, 0x64, 0x2f, 0x43, 
-   0xd7, 0xf5, 0x95, 0x0c, 0x0c, 0x8a, 0x46, 0xa6, 0x04, 0x24, 0x4d, 0xd1, 
-   0xbc, 0xe2, 0xe2, 0xcc, 0xae, 0x24, 0x13, 0x73, 0xe4, 0xfb, 0x07, 0xb8, 
-   0xbb, 0x4d, 0xdd, 0xc8, 0x0d, 0x9f, 0x84, 0x37, 0xb3, 0x15, 0xc7, 0x4d, 
-   0xb3, 0xc8, 0xf3, 0x24, 0x8b, 0x8c, 0x08, 0xd4, 0xd1, 0x8a, 0xf5, 0xbb, 
-   0xc2, 0x0e, 0x2e, 0xf0, 0x49, 0x65, 0x24, 0xdb, 0xbe, 0x77, 0x74, 0x04, 
-   0x25, 0x2b, 0xc2, 0xb0, 0x13, 0xfb, 0x88, 0x4e, 0xf4, 0x16, 0x20, 0xf7, 
-   0x61, 0x1d, 0xd1, 0x87, 0x3e, 0xd7, 0x25, 0xf7, 0x51, 0x3f, 0xa2, 0x8f, 
-   0x6a, 0xa6, 0x70, 0xde, 0xe6, 0xa6, 0x50, 0x40, 0x86, 0x05, 0x2f, 0x20, 
-   0x09, 0xd8, 0x46, 0x63, 0x80, 0x26, 0x61, 0xf0, 0x14, 0xfc, 0x4e, 0x4a, 
-   0x37, 0xe5, 0xc3, 0xa9, 0xb0, 0xef, 0x9b, 0x7c, 0x2c, 0x2c, 0xe1, 0xf6, 
-   0x0f, 0x7a, 0xf2, 0xaa, 0x11, 0xdc, 0xe3, 0x7f, 0x28, 0x6a, 0x1b, 0x6c, 
-   0xfd, 0xf5, 0x0d, 0x73, 0x06, 0x4a, 0x68, 0x58, 0x7d, 0x20, 0x93, 0x6e, 
-   0x83, 0x4a, 0x16, 0xfa, 0x50, 0xe6, 0x1d, 0x30, 0xd7, 0x28, 0x6a, 0xc9, 
-   0x34, 0xff, 0x19, 0xff, 0x40, 0x2b, 0x7f, 0x11, 0x47, 0x57, 0x57, 0x61, 
-   0x78, 0x69, 0x47, 0x9a, 0x05, 0x12, 0x09, 0x34, 0x0c, 0x0c, 0x40, 0x72, 
-   0x6e, 0x10, 0x4d, 0xbc, 0x29, 0xc6, 0xee, 0x1f, 0x95, 0x30, 0x76, 0xd5, 
-   0x99, 0x47, 0x82, 0x1f, 0x9a, 0xb2, 0x74, 0x44, 0x35, 0xa8, 0xbe, 0xb5, 
-   0x52, 0x72, 0xe7, 0xf4, 0x8e, 0x71, 0x7b, 0x6c, 0xc1, 0xd3, 0x79, 0xb6, 
-   0x7b, 0xde, 0xbb, 0x61, 0x40, 0x9f, 0x0c, 0x14, 0x60, 0x6f, 0xa1, 0x82, 
-   0x9e, 0xf3, 0x0e, 0x3f, 0xf6, 0xe7, 0xc8, 0xa1, 0x33, 0x55, 0x8e, 0x86, 
-   0xd6, 0xb1, 0x3f, 0x25, 0xc0, 0x0a, 0xe8, 0xae, 0xfe, 0x9f, 0xf4, 0xe6, 
-   0x3d, 0x7f, 0x07, 0xaf, 0xdf, 0xc4, 0xba, 0x0c, 0x2a, 0x89, 0x70, 0x2b, 
-   0x74, 0x81, 0xcf, 0xf1, 0x8f, 0xde, 0xc3, 0x23, 0xb6, 0x56, 0x96, 0x0f, 
-   0x21, 0xa0, 0x16, 0x39, 0x28, 0x87, 0x2a, 0xf6, 0xaa, 0xe9, 0x6a, 0x1b, 
-   0x3c, 0x9b, 0xf5, 0xa4, 0xb7, 0xe9, 0x04, 0x70, 0x22, 0x79, 0x20, 0x8c, 
-   0xa5, 0xf3, 0xad, 0xdb, 0x22, 0xbd, 0x96, 0x4a, 0x52, 0xa5, 0x1c, 0x51, 
-   0x18, 0x2a, 0x67, 0x78, 0x13, 0x1c, 0x61, 0x01, 0xeb, 0x52, 0x58, 0xea, 
-   0x0c, 0x2a, 0x58, 0xe5, 0x08, 0xde, 0x08, 0x48, 0x18, 0xc6, 0xf0, 0xcd, 
-   0xb0, 0xe0, 0xdf, 0x31, 0x7c, 0x25, 0x0c, 0x5f, 0x10, 0xc6, 0xfa, 0x1d, 
-   0xc3, 0xf7, 0x4f, 0xc1, 0xf0, 0x05, 0x71, 0xbd, 0x4d, 0x30, 0x7c, 0xc1, 
-   0x1b, 0x98, 0xc8, 0x6e, 0xfe, 0xdc, 0x81, 0x9d, 0x8b, 0x01, 0x86, 0x2f, 
-   0xb4, 0xfe, 0x04, 0x8e, 0x46, 0xe0, 0xaf, 0x3b, 0xcf, 0x3f, 0x2d, 0xc5, 
-   0xf1, 0x05, 0x41, 0xc5, 0xff, 0x2e, 0x38, 0xbe, 0xa4, 0xfa, 0xea, 0x08, 
-   0x1c, 0x5f, 0xbb, 0xf5, 0x3d, 0x81, 0x7c, 0xb5, 0xdc, 0xdf, 0x1e, 0xc9, 
-   0x97, 0x44, 0xb1, 0xff, 0x0f, 0x90, 0x7c, 0xd3, 0x77, 0x7d, 0x59, 0xb2, 
-   0x5a, 0x03, 0xea, 0x8b, 0x4e, 0x26, 0xb8, 0x03, 0xeb, 0xe3, 0xb7, 0xc3, 
-   0x4d, 0xb6, 0xdf, 0x0c, 0xe3, 0xb4, 0x6d, 0x80, 0xa8, 0x95, 0x2e, 0xaf, 
-   0xde, 0x64, 0x42, 0x37, 0x34, 0x1c, 0xec, 0x85, 0x62, 0x55, 0x1c, 0x14, 
-   0x18, 0x55, 0xae, 0xf4, 0xdb, 0x25, 0x9d, 0x34, 0xec, 0xe5, 0x7e, 0x5f, 
-   0xb1, 0xea, 0x2f, 0x6d, 0xdb, 0x6e, 0x8a, 0x6d, 0xcd, 0x2a, 0x00, 0xd3, 
-   0xb6, 0x96, 0xdd, 0x11, 0x1b, 0x9b, 0x54, 0x75, 0xf0, 0x9b, 0x1f, 0x29, 
-   0x63, 0x01, 0xe4, 0x23, 0x5a, 0xfa, 0x4d, 0x44, 0x89, 0x23, 0x6b, 0x09, 
-   0x41, 0xe0, 0xf1, 0x38, 0xb2, 0x16, 0xb4, 0x62, 0xaa, 0x02, 0xc9, 0xd2, 
-   0x7b, 0xf8, 0x6a, 0x48, 0x82, 0x45, 0xd6, 0xf4, 0x75, 0x95, 0xc3, 0xf1, 
-   0x5f, 0x39, 0xec, 0xb0, 0x2e, 0xea, 0x05, 0xb5, 0xa9, 0x66, 0x47, 0xcc, 
-   0x41, 0x6b, 0x48, 0x25, 0xcd, 0x2d, 0x0a, 0xb3, 0xa4, 0x31, 0xa6, 0x92, 
-   0xf6, 0xd8, 0xb1, 0x2e, 0x4b, 0x2f, 0xbb, 0x35, 0x82, 0x49, 0xf0, 0x94, 
-   0x4b, 0x05, 0x64, 0x57, 0x17, 0x50, 0x91, 0xed, 0x51, 0x82, 0x02, 0xd8, 
-   0x1c, 0x25, 0x30, 0x80, 0x4f, 0x55, 0xc1, 0x39, 0x22, 0xfa, 0x34, 0x6e, 
-   0x4f, 0xbb, 0x88, 0xa8, 0xf1, 0xd2, 0xd2, 0x0d, 0x47, 0x81, 0x3e, 0x2d, 
-   0x79, 0x3e, 0xb4, 0x76, 0x37, 0x0e, 0xf6, 0x99, 0x62, 0xad, 0x85, 0xd8, 
-   0x09, 0x79, 0x3e, 0xeb, 0x95, 0x21, 0x26, 0x52, 0x0d, 0x70, 0x2e, 0xf5, 
-   0x37, 0xdf, 0x38, 0xec, 0x8f, 0x25, 0xbb, 0x66, 0x1e, 0x55, 0x59, 0x0b, 
-   0xfb, 0xa9, 0x78, 0x31, 0x6b, 0xc6, 0x56, 0x77, 0xd6, 0x53, 0x3b, 0x96, 
-   0xaa, 0x62, 0xb4, 0x0c, 0xc4, 0x68, 0x49, 0x62, 0x2c, 0x83, 0x7e, 0x46, 
-   0x27, 0xfd, 0x60, 0xb7, 0xc3, 0xff, 0x96, 0x7e, 0x6b, 0xf2, 0x20, 0x51, 
-   0xce, 0x49, 0xd3, 0x97, 0x1e, 0xc0, 0xe3, 0x4c, 0x03, 0x3f, 0x9f, 0x3d, 
-   0xcf, 0x92, 0x5e, 0x8b, 0x08, 0x8f, 0x55, 0x0c, 0xb0, 0x9f, 0xe5, 0x07, 
-   0x5a, 0x2a, 0x46, 0x85, 0x72, 0x7d, 0x45, 0x76, 0xb5, 0x76, 0xaf, 0x46, 
-   0xa2, 0x6e, 0x1a, 0x7c, 0x96, 0x3a, 0x65, 0x42, 0x93, 0xa3, 0x24, 0xda, 
-   0xcb, 0x63, 0x52, 0x03, 0x6c, 0x0d, 0x90, 0xf1, 0xf9, 0xd8, 0x52, 0x68, 
-   0xeb, 0xd1, 0xfc, 0x74, 0x34, 0xd4, 0x3a, 0xaa, 0x62, 0xbb, 0x04, 0x12, 
-   0x5b, 0x59, 0x4a, 0x05, 0xe4, 0xc0, 0x8d, 0xa0, 0xb0, 0x9b, 0x05, 0x09, 
-   0x7f, 0x63, 0x58, 0xd7, 0x7c, 0xb3, 0xb0, 0xae, 0xf9, 0x7d, 0xc2, 0xba, 
-   0xe6, 0xbf, 0x43, 0x58, 0x77, 0x45, 0x5f, 0xe1, 0xe8, 0x01, 0x9c, 0x6d, 
-   0x7d, 0x34, 0xc6, 0x58, 0x69, 0x61, 0x9c, 0xf5, 0xc9, 0x10, 0x55, 0xa8, 
-   0x89, 0xae, 0x55, 0xe5, 0x43, 0x85, 0xc1, 0xe9, 0x7c, 0xb1, 0x2a, 0x4d, 
-   0x58, 0x6d, 0x4b, 0xab, 0xe2, 0x8b, 0x65, 0x9f, 0xca, 0x20, 0xc1, 0xcb, 
-   0x7c, 0x71, 0xc5, 0xcd, 0x0c, 0x48, 0x54, 0x42, 0x6f, 0x92, 0x8a, 0x85, 
-   0x15, 0x50, 0x22, 0xda, 0xdb, 0xa3, 0xaf, 0x40, 0x6a, 0xbf, 0x70, 0x81, 
-   0x05, 0x2c, 0x33, 0x08, 0xc3, 0xbe, 0xa1, 0xb2, 0xa2, 0x0c, 0x6d, 0x4a, 
-   0x6e, 0x06, 0x5b, 0xbc, 0x19, 0xb4, 0x3d, 0xc9, 0x94, 0x3a, 0x0c, 0x3f, 
-   0x42, 0xb7, 0x86, 0x6e, 0x7b, 0xb5, 0x2d, 0xa5, 0x46, 0x27, 0x16, 0x7f, 
-   0x84, 0x46, 0xd2, 0xac, 0xee, 0x47, 0xed, 0x4b, 0x29, 0xa1, 0xb1, 0x2d, 
-   0xbc, 0x60, 0xcb, 0xb5, 0x56, 0xbd, 0x9b, 0x22, 0x0d, 0x4e, 0x6e, 0x02, 
-   0xf4, 0x65, 0x3c, 0xe5, 0xc4, 0xcd, 0x32, 0x62, 0x89, 0xb2, 0x65, 0xb0, 
-   0xb0, 0xbb, 0x02, 0xc6, 0x7d, 0x41, 0x93, 0x7a, 0xd7, 0xdb, 0xbb, 0x9c, 
-   0x96, 0x31, 0x30, 0xf0, 0xb8, 0x63, 0xf1, 0x99, 0xdf, 0x46, 0x32, 0x08, 
-   0x65, 0xc2, 0x31, 0x33, 0x88, 0xa4, 0x31, 0x73, 0x29, 0x65, 0x66, 0x11, 
-   0x85, 0x0f, 0xcb, 0xce, 0xc2, 0x21, 0xb5, 0x1e, 0x1d, 0xad, 0x71, 0x10, 
-   0x62, 0x3d, 0x6e, 0xff, 0xc3, 0xd2, 0xaa, 0x2f, 0x1c, 0xfb, 0xe0, 0x69, 
-   0x8d, 0x20, 0x25, 0xd5, 0xea, 0xdb, 0x59, 0x3e, 0x84, 0x56, 0x8b, 0x44, 
-   0x3b, 0x1a, 0x5d, 0xd3, 0x89, 0xd4, 0xc8, 0x45, 0xbf, 0xbb, 0x46, 0x85, 
-   0x56, 0xaa, 0xcf, 0x54, 0x13, 0xc2, 0x66, 0xb3, 0xc9, 0x3f, 0x92, 0xd3, 
-   0x56, 0x10, 0xb2, 0x4f, 0x10, 0x14, 0xc1, 0xf9, 0xf2, 0xf3, 0xb2, 0xac, 
-   0xcd, 0xe2, 0x81, 0xfc, 0x03, 0xcf, 0x6f, 0xa3, 0x7d, 0xe0, 0x47, 0x3e, 
-   0x89, 0xd0, 0xb0, 0x31, 0x9d, 0xee, 0x9d, 0x5f, 0xf8, 0xc0, 0xc8, 0x32, 
-   0x42, 0xe5, 0x43, 0xbf, 0x30, 0x25, 0x2c, 0xbe, 0xf3, 0x03, 0x3b, 0xf7, 
-   0x78, 0xe7, 0x96, 0x5a, 0x5b, 0x8c, 0x8e, 0x53, 0xaa, 0xd4, 0x15, 0xa6, 
-   0x84, 0x79, 0x6c, 0x48, 0x45, 0xef, 0x82, 0xa9, 0x58, 0x2a, 0x1d, 0x91, 
-   0xde, 0x0f, 0x94, 0x6b, 0x4a, 0xa9, 0x52, 0x52, 0x98, 0x12, 0x16, 0xb1, 
-   0xe1, 0x80, 0xde, 0xad, 0xb6, 0x20, 0x78, 0x4b, 0xad, 0x22, 0x46, 0xc7, 
-   0x29, 0x95, 0x3a, 0x4a, 0x09, 0x8b, 0xa0, 0x70, 0x60, 0xef, 0x82, 0xe4, 
-   0x6d, 0xb5, 0x8e, 0x18, 0x1d, 0xa7, 0x54, 0xea, 0x28, 0x25, 0xcc, 0x28, 
-   0x95, 0x3a, 0xa2, 0xbd, 0x8b, 0x8b, 0x54, 0xad, 0x23, 0x46, 0xc7, 0x79, 
-   0x2a, 0x75, 0x94, 0x12, 0x66, 0x94, 0x4a, 0x1d, 0x91, 0xaf, 0x80, 0x48, 
-   0x94, 0x25, 0x3a, 0xea, 0xc8, 0x33, 0x52, 0xeb, 0x28, 0x21, 0xcc, 0x28, 
-   0x95, 0x3a, 0x4a, 0x09, 0x33, 0xbf, 0x53, 0xa6, 0xa3, 0x8e, 0x38, 0xf7, 
-   0x86, 0x5a, 0x47, 0x1d, 0x59, 0xf2, 0x0d, 0xa5, 0x8e, 0x52, 0xc2, 0x8c, 
-   0xb2, 0x4c, 0x47, 0x9d, 0x83, 0x60, 0x21, 0x0d, 0xb5, 0x8e, 0x3a, 0xb2, 
-   0xe4, 0x1b, 0x4a, 0x1d, 0xa5, 0x84, 0x19, 0xa5, 0x4a, 0x47, 0xc4, 0xdb, 
-   0xbf, 0x74, 0xc4, 0xb9, 0x97, 0xe8, 0xe8, 0x45, 0xd2, 0x66, 0x83, 0x5f, 
-   0x0c, 0xa9, 0x77, 0xd8, 0xec, 0xab, 0x37, 0x3b, 0xb0, 0xbc, 0xc3, 0x32, 
-   0x08, 0x4a, 0xaf, 0x7c, 0x1c, 0xd3, 0xf8, 0xce, 0x56, 0xfa, 0x4e, 0xb2, 
-   0x50, 0x06, 0x57, 0x87, 0x77, 0x47, 0xb9, 0x19, 0x38, 0xa5, 0x8e, 0xa7, 
-   0xdd, 0x0d, 0x3b, 0xa2, 0xdb, 0x52, 0xd1, 0x68, 0xf7, 0x40, 0xba, 0x60, 
-   0x35, 0xbb, 0x9a, 0xb4, 0x4e, 0x55, 0x8a, 0xa5, 0x7e, 0x47, 0xb3, 0x7f, 
-   0x51, 0xbf, 0xac, 0xd9, 0x8d, 0x24, 0x77, 0xac, 0x5a, 0x41, 0x74, 0x7b, 
-   0x29, 0x6c, 0x42, 0x65, 0xe7, 0xa4, 0x54, 0xe4, 0xf8, 0x00, 0x59, 0xd4, 
-   0x75, 0x47, 0xa3, 0xa7, 0x1e, 0x3e, 0x95, 0x29, 0xc6, 0xa1, 0x55, 0x93, 
-   0xc1, 0xb7, 0xef, 0xc0, 0x0f, 0xde, 0xa9, 0x53, 0x88, 0xf2, 0x51, 0x44, 
-   0xfe, 0xb8, 0x4c, 0x95, 0xd2, 0xc5, 0xaa, 0x47, 0xaa, 0x0c, 0xfd, 0xf3, 
-   0x8d, 0x8e, 0x54, 0xf7, 0xa7, 0x69, 0x56, 0xe8, 0xc7, 0x1c, 0xa7, 0xbe, 
-   0x01, 0xc6, 0xc2, 0x2e, 0x7c, 0xf2, 0xbc, 0xfa, 0xd9, 0xc9, 0x7e, 0x3f, 
-   0x3b, 0xfd, 0xa0, 0xb3, 0xd3, 0x8f, 0x39, 0xba, 0xac, 0x49, 0x0d, 0xcb, 
-   0x22, 0x2b, 0xf2, 0x15, 0x5d, 0x92, 0x4a, 0x43, 0xb9, 0x26, 0xb6, 0xde, 
-   0x35, 0xe4, 0x5a, 0x58, 0x6d, 0xad, 0x76, 0x68, 0x0b, 0xbb, 0xb9, 0x88, 
-   0xd8, 0xda, 0xea, 0x14, 0x15, 0xf2, 0x77, 0xdd, 0x13, 0x99, 0x2c, 0xcc, 
-   0xe5, 0x6c, 0x2c, 0x5e, 0x48, 0xaa, 0x86, 0x7b, 0x09, 0x3a, 0xf1, 0x1e, 
-   0xfc, 0x20, 0xa4, 0x40, 0x81, 0xe5, 0xbb, 0x4a, 0xca, 0x97, 0x7d, 0xf4, 
-   0x84, 0xa4, 0x6c, 0x23, 0xb4, 0x74, 0xe3, 0x67, 0xf2, 0xc8, 0x20, 0xef, 
-   0x09, 0x51, 0xe2, 0x55, 0x49, 0x82, 0x7a, 0xe6, 0xee, 0x9d, 0x90, 0x7e, 
-   0x57, 0x23, 0x21, 0x61, 0x7f, 0xfb, 0x19, 0x2e, 0xd0, 0xe2, 0x9c, 0xae, 
-   0x06, 0xba, 0x3d, 0xa1, 0x90, 0x2c, 0xfb, 0xc6, 0xfd, 0xe1, 0xad, 0xaf, 
-   0x3f, 0xee, 0xb3, 0x29, 0xf7, 0x29, 0x68, 0xd6, 0x1d, 0x51, 0x51, 0x54, 
-   0xb6, 0x4b, 0x30, 0xec, 0x42, 0xa3, 0x12, 0x2e, 0x40, 0x74, 0x62, 0x37, 
-   0xc0, 0x46, 0xc3, 0x1d, 0x46, 0x56, 0x6b, 0x01, 0x24, 0xe3, 0xd9, 0x88, 
-   0x4f, 0xd3, 0xef, 0x49, 0x02, 0x2c, 0xab, 0x3c, 0xf3, 0xaa, 0xaa, 0x82, 
-   0xc6, 0x5b, 0xdf, 0x40, 0xd1, 0xe1, 0x53, 0x80, 0xcb, 0x08, 0x25, 0x05, 
-   0x21, 0xb2, 0xf8, 0x89, 0x39, 0x0f, 0x5f, 0xf6, 0xf4, 0x45, 0x6f, 0xfc, 
-   0xe8, 0x45, 0xe4, 0xd1, 0xa8, 0x83, 0x5a, 0xec, 0x51, 0x2e, 0x9d, 0xca, 
-   0xff, 0x28, 0x3e, 0xe7, 0xc3, 0x59, 0x23, 0xdf, 0x7d, 0x89, 0x91, 0xdd, 
-   0x94, 0xbe, 0x79, 0x24, 0x14, 0xc9, 0x1a, 0xa0, 0x10, 0x88, 0xdc, 0xb6, 
-   0xc1, 0x33, 0x6a, 0xd8, 0x12, 0xb3, 0x6a, 0x68, 0x04, 0x72, 0x80, 0xe5, 
-   0xff, 0xb0, 0x00, 0xab, 0xf9, 0xbd, 0xb5, 0x67, 0x7f, 0x17, 0xed, 0xed, 
-   0xbc, 0x75, 0x8d, 0x94, 0x21, 0xbf, 0x9d, 0x06, 0x09, 0xc7, 0xbf, 0xad, 
-   0x16, 0x5b, 0xdf, 0x5b, 0x8b, 0x8d, 0x32, 0x17, 0xb8, 0x17, 0x00, 0x30, 
-   0x85, 0xfb, 0xd5, 0x86, 0x76, 0x9a, 0x24, 0xd4, 0xda, 0x4b, 0x41, 0xb2, 
-   0x2e, 0x28, 0x56, 0x28, 0x95, 0xd8, 0x14, 0x0b, 0xbd, 0x52, 0x3e, 0xac, 
-   0x16, 0xfe, 0x2b, 0xda, 0xa4, 0x85, 0x92, 0xa0, 0x49, 0xb1, 0x28, 0x97, 
-   0x7e, 0x76, 0x93, 0x7d, 0xd5, 0x0a, 0xe1, 0x46, 0xd2, 0xfd, 0x16, 0x1e, 
-   0xd0, 0x44, 0x55, 0x53, 0x6b, 0x70, 0xc1, 0x25, 0x5a, 0x18, 0xb5, 0xd7, 
-   0xa4, 0xf0, 0xd9, 0xe4, 0x92, 0xeb, 0xaf, 0x63, 0x5e, 0x6d, 0x03, 0xf3, 
-   0x6a, 0x4a, 0xe6, 0xa5, 0x41, 0x87, 0x66, 0x90, 0xdb, 0x05, 0x80, 0x6b, 
-   0xa1, 0x08, 0xcb, 0xe0, 0x39, 0x1f, 0x65, 0x02, 0xe2, 0x5c, 0x5b, 0x25, 
-   0x25, 0x17, 0x9a, 0x69, 0x64, 0x58, 0x6f, 0xf4, 0x22, 0x79, 0x45, 0x40, 
-   0x55, 0xf9, 0x52, 0x11, 0xd1, 0xaf, 0xe5, 0x0f, 0x66, 0xab, 0x96, 0x06, 
-   0x6f, 0x91, 0x98, 0xda, 0x38, 0x5c, 0x43, 0xdf, 0x7a, 0xe6, 0xf5, 0x80, 
-   0xe5, 0x37, 0xdb, 0x74, 0x40, 0x43, 0xc5, 0x33, 0xad, 0x96, 0x41, 0x31, 
-   0x02, 0xfb, 0x2c, 0x1f, 0xfb, 0x64, 0x32, 0x37, 0xf1, 0xb9, 0x88, 0xa5, 
-   0x5d, 0xad, 0xa0, 0x4e, 0x64, 0x98, 0x55, 0xcd, 0xaa, 0xf8, 0x19, 0x95, 
-   0xd6, 0x81, 0x23, 0x9c, 0x95, 0x40, 0x7e, 0x1b, 0xac, 0x43, 0x70, 0x94, 
-   0x65, 0x3c, 0x8b, 0x65, 0xed, 0x65, 0xca, 0xcd, 0x49, 0x52, 0xfd, 0x79, 
-   0x3d, 0x83, 0x57, 0xf6, 0x19, 0x53, 0x59, 0x9a, 0x6a, 0x9e, 0x55, 0x4d, 
-   0xbc, 0x51, 0x66, 0xe2, 0x19, 0xb2, 0x3f, 0x3e, 0x04, 0xc4, 0xb4, 0xa0, 
-   0xb1, 0xc4, 0xba, 0xf3, 0x28, 0x32, 0xdc, 0x99, 0x19, 0xe1, 0x07, 0xcb, 
-   0x7d, 0xa9, 0x92, 0x4a, 0xda, 0xe9, 0x11, 0xe7, 0x8f, 0xbb, 0x49, 0x1f, 
-   0x9f, 0x30, 0x47, 0x9f, 0x9f, 0x74, 0x4b, 0xe5, 0x9e, 0xb2, 0x8f, 0x1a, 
-   0x24, 0x3a, 0xce, 0x37, 0x6c, 0x7f, 0x28, 0x8c, 0xf6, 0x2a, 0x29, 0xc9, 
-   0x63, 0x1f, 0x01, 0x99, 0x29, 0x5e, 0xfd, 0xdc, 0x46, 0xae, 0xfc, 0x1d, 
-   0x8f, 0xec, 0xbb, 0x08, 0x26, 0x6f, 0xe0, 0x00, 0xc3, 0x20, 0xa5, 0x7b, 
-   0xc1, 0x06, 0x4f, 0x93, 0xc0, 0xb4, 0xb0, 0xe4, 0x09, 0x36, 0x0e, 0x35, 
-   0x77, 0x95, 0x13, 0x04, 0x4f, 0xa5, 0x7f, 0xe3, 0x32, 0xd1, 0x92, 0x6f, 
-   0x40, 0xe7, 0x90, 0x90, 0x4a, 0x3e, 0xb0, 0xc9, 0x5f, 0x70, 0x99, 0x7e, 
-   0xad, 0x59, 0xef, 0xb1, 0x93, 0x3a, 0xb4, 0x95, 0xf2, 0x6b, 0xcd, 0xfa, 
-   0x67, 0x8a, 0x09, 0x4c, 0xce, 0xd5, 0x00, 0x91, 0x2f, 0x52, 0x93, 0x4f, 
-   0x96, 0xf2, 0x77, 0x25, 0x6c, 0x94, 0xc0, 0x66, 0xa2, 0x7f, 0x50, 0x99, 
-   0x01, 0xa8, 0x55, 0x63, 0xac, 0xaf, 0x01, 0x28, 0x8e, 0x37, 0x91, 0xa1, 
-   0x8a, 0xa5, 0xbe, 0x44, 0x0c, 0x1a, 0xab, 0x86, 0xa9, 0xbe, 0x44, 0x30, 
-   0xfb, 0x92, 0x33, 0x63, 0x81, 0xd6, 0x01, 0xf6, 0x04, 0xc9, 0x77, 0x10, 
-   0x80, 0xe5, 0xa9, 0x7f, 0x44, 0x3d, 0x27, 0xdf, 0x91, 0x09, 0x36, 0x1b, 
-   0xe4, 0xd2, 0xf2, 0xba, 0x3f, 0x10, 0xe6, 0x03, 0xbc, 0x99, 0xd5, 0x3f, 
-   0xe8, 0xee, 0xe7, 0xd5, 0x80, 0x1e, 0xf1, 0x01, 0x29, 0x4a, 0x33, 0x32, 
-   0x9e, 0x4f, 0x51, 0x92, 0x08, 0x4a, 0x3c, 0x4b, 0x1a, 0x92, 0x6f, 0x9e, 
-   0xe6, 0xc2, 0xe5, 0xe2, 0x6e, 0xa3, 0xd3, 0xd7, 0x37, 0xf7, 0x64, 0x82, 
-   0xff, 0x23, 0x4b, 0xfc, 0x10, 0x61, 0x89, 0x63, 0x7d, 0x6e, 0xd9, 0xbb, 
-   0x1a, 0xb4, 0x62, 0x0f, 0xa7, 0x3e, 0x93, 0xe5, 0x87, 0xbd, 0xb9, 0x47, 
-   0xdd, 0x40, 0x1c, 0xb0, 0xdf, 0x07, 0xde, 0x1a, 0x7d, 0xb6, 0x81, 0x05, 
-   0x54, 0x58, 0x84, 0xc5, 0x8e, 0x79, 0x69, 0x58, 0xf6, 0x32, 0x16, 0x77, 
-   0x91, 0xfb, 0x94, 0x37, 0x10, 0x54, 0x1c, 0xf3, 0x01, 0x5a, 0x41, 0x52, 
-   0x91, 0x9b, 0x4c, 0xd7, 0x83, 0x3e, 0x7b, 0x7c, 0xd4, 0x57, 0x9b, 0x99, 
-   0x46, 0xf4, 0xcc, 0x2b, 0x7f, 0x88, 0xd7, 0x78, 0xd8, 0x06, 0x5f, 0xe5, 
-   0x95, 0x39, 0x9b, 0x0e, 0xd9, 0xe0, 0x2b, 0xd0, 0x02, 0xc2, 0x6b, 0x72, 
-   0x30, 0x11, 0xc0, 0xe7, 0xe8, 0xf6, 0x50, 0xfe, 0xfd, 0xf5, 0xb2, 0x48, 
-   0xc4, 0xe4, 0x7b, 0xf2, 0x04, 0xe3, 0x2a, 0xb5, 0xf6, 0x73, 0x13, 0xa6, 
-   0x76, 0xed, 0xd9, 0x21, 0x9f, 0x98, 0x26, 0x4e, 0x0b, 0x0e, 0xd8, 0xcf, 
-   0xf5, 0x7e, 0xa5, 0xa9, 0x67, 0xa2, 0xf7, 0x29, 0x93, 0x30, 0xd8, 0x3b, 
-   0x0f, 0xc4, 0x92, 0x6e, 0x9c, 0x1b, 0x68, 0xf5, 0xea, 0x1d, 0x45, 0x5e, 
-   0x42, 0xd5, 0xbe, 0x85, 0x5e, 0x26, 0xa6, 0x4a, 0x5f, 0x76, 0xb7, 0xcd, 
-   0x3e, 0x77, 0x5d, 0xf9, 0xbb, 0xed, 0xc2, 0x67, 0xd4, 0xcb, 0xf8, 0xbe, 
-   0xe5, 0x97, 0xc8, 0xcb, 0xf9, 0x56, 0xfb, 0xce, 0x78, 0x09, 0x6e, 0x5c, 
-   0x69, 0x3c, 0xf4, 0x8d, 0x9f, 0x1e, 0x37, 0xba, 0x05, 0xbd, 0x73, 0x57, 
-   0x8a, 0xc7, 0x39, 0x06, 0x6f, 0x92, 0xc5, 0x0b, 0x1d, 0x9e, 0x05, 0x6a, 
-   0x18, 0xbc, 0x7b, 0xe2, 0x53, 0x7b, 0x24, 0x0f, 0x3a, 0x59, 0xfe, 0x5c, 
-   0x17, 0x86, 0xf2, 0x46, 0x32, 0xbd, 0xea, 0xe2, 0x4b, 0xa0, 0xf7, 0x44, 
-   0xf8, 0x52, 0xd5, 0x45, 0x24, 0xa7, 0x5f, 0xbe, 0xc6, 0xae, 0xf6, 0x2a, 
-   0x92, 0x93, 0x47, 0x8f, 0x41, 0x18, 0x6b, 0xeb, 0x21, 0x38, 0x3d, 0xcd, 
-   0xdf, 0x68, 0x2a, 0x23, 0x24, 0x6a, 0x7b, 0x61, 0xd5, 0xd9, 0xff, 0x81, 
-   0xef, 0xf0, 0x2a, 0xcb, 0x02, 0x66, 0xde, 0x03, 0x39, 0x7b, 0x90, 0xd0, 
-   0x39, 0x7e, 0x85, 0xf0, 0x5f, 0x1b, 0x7a, 0xe7, 0x74, 0x43, 0xee, 0x3b, 
-   0xc9, 0xbb, 0xa4, 0x52, 0x3e, 0x06, 0x38, 0x2a, 0x24, 0x36, 0x74, 0xc8, 
-   0xe6, 0x1e, 0x9f, 0xb2, 0xfc, 0x55, 0x81, 0x49, 0x21, 0x4a, 0x29, 0x8f, 
-   0x16, 0xc8, 0x96, 0x92, 0x85, 0x9c, 0x85, 0xb8, 0x0c, 0x48, 0xbc, 0x55, 
-   0x01, 0xc3, 0xcc, 0xb1, 0xbf, 0x83, 0x22, 0x5a, 0x81, 0xb5, 0xd1, 0x2d, 
-   0x68, 0x12, 0x28, 0x0f, 0xbc, 0x27, 0x2f, 0x02, 0xd1, 0x0d, 0x0c, 0xf6, 
-   0x6f, 0x96, 0x07, 0x2d, 0xdd, 0x49, 0x8d, 0x20, 0x32, 0xc9, 0x9e, 0xff, 
-   0x90, 0x7c, 0xed, 0x30, 0xb9, 0xe9, 0xa7, 0x88, 0x54, 0x91, 0x1b, 0x1f, 
-   0xf6, 0x90, 0x9f, 0xe1, 0x09, 0x49, 0xc1, 0x43, 0x68, 0x4a, 0xed, 0x0b, 
-   0x2c, 0xea, 0x45, 0x67, 0xc1, 0x6b, 0xd0, 0x21, 0x1e, 0x45, 0x26, 0x78, 
-   0xa5, 0x79, 0x7f, 0x00, 0x87, 0x2a, 0xfd, 0x49, 0x85, 0x3c, 0xc5, 0x3f, 
-   0xae, 0x74, 0x1a, 0x07, 0x47, 0x50, 0x52, 0x40, 0x7f, 0x90, 0x8b, 0x24, 
-   0x6d, 0x71, 0x14, 0x5f, 0x8e, 0xd8, 0xa4, 0x7e, 0xdd, 0x00, 0xfa, 0x64, 
-   0xab, 0xa1, 0x77, 0xad, 0x2c, 0x9c, 0xc3, 0x6a, 0x64, 0xe7, 0x65, 0x33, 
-   0x48, 0xf9, 0x5c, 0x23, 0x43, 0x58, 0xf9, 0x5c, 0xab, 0x23, 0xa0, 0xe5, 
-   0xc9, 0xde, 0x04, 0x5c, 0x24, 0x80, 0x59, 0xa5, 0x0a, 0xc6, 0x4c, 0xd7, 
-   0xed, 0xdb, 0xd8, 0xb2, 0x55, 0xcd, 0x0e, 0xf5, 0x47, 0x2d, 0x95, 0x1d, 
-   0xea, 0x7d, 0xa8, 0xc2, 0x0e, 0xf5, 0x4e, 0xdc, 0x3f, 0xec, 0xd2, 0x76, 
-   0x2d, 0xc0, 0x7e, 0xd5, 0x37, 0x11, 0xef, 0xf6, 0xfb, 0x23, 0xec, 0x77, 
-   0xee, 0xbe, 0xc4, 0x07, 0x7c, 0xba, 0x4e, 0xa2, 0x4a, 0x72, 0xa5, 0x87, 
-   0x7f, 0x4f, 0xee, 0x23, 0xea, 0xa8, 0x7f, 0x71, 0xb9, 0xb8, 0x1e, 0x0f, 
-   0x86, 0xbf, 0x5a, 0x60, 0x5e, 0xcc, 0x0d, 0x6b, 0xec, 0x80, 0x4e, 0x5e, 
-   0x18, 0x28, 0x11, 0xe2, 0x1a, 0x85, 0x54, 0x16, 0x10, 0xcb, 0xb9, 0x84, 
-   0x4d, 0x02, 0xbf, 0x44, 0x6e, 0xe7, 0x42, 0x54, 0x04, 0x49, 0x69, 0x02, 
-   0x56, 0xa7, 0x0c, 0x0e, 0xf3, 0x0c, 0xe3, 0x72, 0x86, 0x8d, 0xca, 0x0c, 
-   0xa3, 0x52, 0x86, 0x1c, 0x2c, 0x50, 0x7d, 0x63, 0x87, 0x3e, 0x87, 0xce, 
-   0x5e, 0x7a, 0x1f, 0x09, 0x81, 0xf8, 0xaa, 0x6d, 0x95, 0x36, 0x27, 0x95, 
-   0x51, 0x68, 0x3a, 0x9c, 0x0c, 0xbb, 0x73, 0xad, 0x9d, 0xf2, 0x06, 0xfd, 
-   0x51, 0xf7, 0x7a, 0xa2, 0xb5, 0x50, 0x4e, 0x7f, 0x7d, 0x35, 0x9d, 0x8e, 
-   0xa7, 0xda, 0x98, 0x97, 0x37, 0xe8, 0x8d, 0xa7, 0x83, 0xe1, 0x54, 0x1b, 
-   0xf4, 0xe6, 0x7b, 0x58, 0x8c, 0x6f, 0xfa, 0xc3, 0x62, 0xec, 0x5b, 0xb2, 
-   0x06, 0xe6, 0xa5, 0x52, 0x34, 0x58, 0xf1, 0xef, 0x52, 0x24, 0x91, 0x5d, 
-   0xa9, 0x14, 0x0d, 0xee, 0x08, 0xde, 0xa5, 0x48, 0x22, 0x78, 0xf7, 0x39, 
-   0xa9, 0x98, 0x4b, 0xab, 0x25, 0xe0, 0xb7, 0xca, 0xf2, 0x36, 0xad, 0xdd, 
-   0xed, 0xc9, 0xb1, 0x3c, 0x78, 0x2e, 0xb2, 0xd1, 0xef, 0xd9, 0xfd, 0xc0, 
-   0xdd, 0x6c, 0xbc, 0x95, 0x47, 0x92, 0x33, 0xdb, 0x20, 0xf8, 0xfd, 0xb0, 
-   0x57, 0x8c, 0xc8, 0xfc, 0x93, 0x94, 0xdd, 0x43, 0x1c, 0x8c, 0xc6, 0x83, 
-   0xff, 0x26, 0x49, 0x5a, 0xb4, 0xf4, 0xf0, 0xa1, 0x2a, 0xf0, 0xb7, 0x00, 
-   0x42, 0x84, 0x3e, 0x7d, 0xd5, 0x23, 0x8d, 0xa3, 0xc3, 0x7e, 0xbf, 0xf5, 
-   0xa0, 0x90, 0x44, 0x9f, 0xa8, 0xbf, 0x74, 0xc8, 0x83, 0x42, 0x54, 0xfa, 
-   0x45, 0x5b, 0x7d, 0x64, 0x7e, 0xe1, 0xc6, 0xab, 0xc7, 0x72, 0x26, 0x85, 
-   0x20, 0xbd, 0xc8, 0x65, 0xc4, 0xa4, 0x0b, 0xe7, 0x24, 0xe4, 0x4c, 0x34, 
-   0x68, 0x7c, 0xe9, 0x2e, 0x6c, 0x0f, 0xb4, 0x2b, 0x87, 0x93, 0xa6, 0xe8, 
-   0x1c, 0x9a, 0xa5, 0x93, 0x36, 0x68, 0x0c, 0xb4, 0x8b, 0x26, 0x25, 0xed, 
-   0x1f, 0x96, 0x04, 0xcf, 0x79, 0xaf, 0x5d, 0x34, 0x69, 0x03, 0x6b, 0xa0, 
-   0xcd, 0x58, 0x70, 0x52, 0x06, 0x89, 0xa3, 0x4d, 0x59, 0x64, 0x83, 0x79, 
-   0xf4, 0xb0, 0x9c, 0x69, 0xe9, 0x4d, 0x31, 0x71, 0x21, 0x5d, 0xf6, 0x1d, 
-   0x11, 0xe4, 0x58, 0x9a, 0x20, 0x27, 0x6d, 0x19, 0xd1, 0xa4, 0x58, 0x48, 
-   0x53, 0x61, 0x68, 0x49, 0xc0, 0x7c, 0xaa, 0x7e, 0xb3, 0xa1, 0x90, 0x93, 
-   0x88, 0xc9, 0xbd, 0xd0, 0x33, 0xe1, 0xd7, 0x20, 0xa7, 0xfe, 0x38, 0xe9, 
-   0x09, 0xbe, 0x52, 0xd4, 0x1f, 0x14, 0x45, 0x86, 0xb6, 0x9e, 0xa1, 0xfe, 
-   0x00, 0x29, 0x32, 0xb4, 0xf4, 0x0c, 0xf5, 0x37, 0x96, 0x22, 0xc3, 0xba, 
-   0x9e, 0x21, 0x08, 0xab, 0xaa, 0x55, 0xb1, 0xee, 0xf3, 0x36, 0x23, 0xf7, 
-   0x09, 0x07, 0x72, 0x75, 0x52, 0x06, 0x46, 0x4e, 0x18, 0xab, 0x78, 0xfb, 
-   0x0a, 0x7f, 0xef, 0xd3, 0x2e, 0xa2, 0x80, 0x28, 0x99, 0x91, 0x8f, 0x86, 
-   0xaa, 0xe2, 0x5f, 0xbb, 0xa5, 0x17, 0x75, 0xca, 0xe7, 0xb6, 0xb7, 0x98, 
-   0x74, 0x81, 0x4a, 0x69, 0xdb, 0xa4, 0x62, 0x98, 0x42, 0x4c, 0x51, 0x4e, 
-   0x10, 0x03, 0x7d, 0x02, 0xea, 0x82, 0xe4, 0x29, 0x11, 0x71, 0xcb, 0x40, 
-   0x7a, 0xc8, 0xe6, 0x49, 0x8e, 0xa8, 0xa3, 0x40, 0x3a, 0xbc, 0x76, 0x5e, 
-   0x58, 0xff, 0xb5, 0x60, 0x53, 0x1b, 0xb8, 0xb1, 0xe3, 0x6d, 0x21, 0x4e, 
-   0xf5, 0x8e, 0x9e, 0x93, 0xe7, 0x1b, 0x70, 0xb2, 0xce, 0xed, 0x3c, 0x27, 
-   0x40, 0x2a, 0x41, 0xc8, 0x92, 0xf1, 0xa4, 0xce, 0x8e, 0x3e, 0x88, 0x86, 
-   0x38, 0x9d, 0xe9, 0xd1, 0x67, 0x24, 0x46, 0x4b, 0xa1, 0x82, 0x5a, 0x42, 
-   0xbc, 0xd4, 0x8b, 0x59, 0xe2, 0xf3, 0x40, 0x50, 0xfd, 0x21, 0x46, 0x06, 
-   0xe5, 0x55, 0x12, 0xa3, 0x10, 0x86, 0xd7, 0x37, 0x46, 0x68, 0x98, 0xd1, 
-   0xeb, 0xdc, 0x39, 0xa2, 0xb0, 0x52, 0xab, 0x60, 0x47, 0x0a, 0x7e, 0x42, 
-   0xf7, 0x5f, 0x07, 0xf0, 0xaa, 0xda, 0x32, 0x38, 0xf6, 0x53, 0xae, 0xc3, 
-   0x79, 0x1f, 0xc2, 0x7e, 0xa0, 0xa5, 0x33, 0xea, 0x10, 0x08, 0xe6, 0x95, 
-   0x3a, 0x8b, 0xaf, 0xa8, 0xfb, 0xe2, 0x41, 0x9f, 0x93, 0xb4, 0x0d, 0xee, 
-   0xf6, 0x45, 0x46, 0x5f, 0x94, 0x8c, 0x0c, 0xe0, 0x7f, 0x66, 0xcf, 0xde, 
-   0x1f, 0x7f, 0x90, 0x78, 0x08, 0x88, 0x97, 0x39, 0x72, 0xbd, 0x3a, 0x60, 
-   0x4e, 0xdb, 0x13, 0x0c, 0x68, 0xed, 0x96, 0x9f, 0x12, 0x8f, 0x85, 0xef, 
-   0xe3, 0xaa, 0x76, 0xfb, 0x94, 0x76, 0x2a, 0xdc, 0x39, 0xa8, 0xb6, 0xfb, 
-   0x94, 0xf6, 0x92, 0x59, 0xa1, 0x66, 0xaf, 0x4f, 0xa9, 0x7b, 0xd4, 0xf6, 
-   0x35, 0xbb, 0x7d, 0x26, 0x1f, 0xb6, 0xe2, 0x2a, 0x3c, 0x97, 0xca, 0xba, 
-   0x01, 0x24, 0x6b, 0x19, 0x14, 0xc9, 0xa5, 0x73, 0x82, 0x9a, 0x17, 0xde, 
-   0xcd, 0x29, 0x9b, 0x4f, 0xa1, 0xe6, 0x06, 0x25, 0x8a, 0x0c, 0x4a, 0xb8, 
-   0x80, 0x61, 0x0f, 0x1c, 0xaa, 0xb2, 0x53, 0x00, 0x04, 0x78, 0x0f, 0xe4, 
-   0xb4, 0xa6, 0x97, 0x3d, 0xc0, 0x56, 0xeb, 0x06, 0x30, 0x2a, 0xe9, 0xd6, 
-   0x28, 0xde, 0x9f, 0x08, 0xf9, 0xb4, 0xba, 0x01, 0xf6, 0x18, 0x2b, 0x7c, 
-   0xa6, 0x93, 0xcb, 0x06, 0x21, 0xc0, 0x86, 0xeb, 0xbd, 0xa7, 0x58, 0x39, 
-   0x0d, 0x70, 0x30, 0xf8, 0x44, 0xa4, 0x00, 0x08, 0x04, 0x30, 0x68, 0x19, 
-   0x38, 0x3a, 0x86, 0xc4, 0x98, 0x54, 0x73, 0x9f, 0xb4, 0x9b, 0x35, 0x72, 
-   0xd3, 0x86, 0xf7, 0x04, 0x7a, 0x53, 0xc5, 0x5f, 0x4c, 0x95, 0x21, 0x23, 
-   0xab, 0x85, 0x4b, 0x63, 0xc2, 0x3d, 0x29, 0x15, 0xe3, 0x97, 0x34, 0x95, 
-   0x2f, 0x44, 0x09, 0x98, 0x2e, 0x3e, 0xb0, 0xb1, 0xc3, 0xe3, 0x85, 0xb7, 
-   0x15, 0x79, 0x15, 0x03, 0x2d, 0xa5, 0xa8, 0xae, 0x9d, 0x07, 0x25, 0x1f, 
-   0x15, 0x6e, 0x1c, 0x10, 0x5f, 0xb9, 0xfe, 0x61, 0x97, 0x9b, 0x25, 0x45, 
-   0xb4, 0x64, 0xcf, 0x10, 0xb3, 0xea, 0x6d, 0x28, 0x66, 0x17, 0xe8, 0xd0, 
-   0x14, 0x80, 0x52, 0x31, 0x68, 0x84, 0x66, 0x37, 0xe3, 0xe9, 0x35, 0x80, 
-   0x3d, 0xa6, 0x6d, 0x7a, 0x09, 0xbd, 0x5f, 0x36, 0x68, 0x95, 0xeb, 0xb1, 
-   0x51, 0xa1, 0x6d, 0xaf, 0xdb, 0x01, 0x52, 0x05, 0x46, 0xed, 0x72, 0xbd, 
-   0xb6, 0xaa, 0xb4, 0x16, 0xe1, 0x57, 0xda, 0x95, 0xba, 0xe5, 0x8f, 0x11, 
-   0xcf, 0x2a, 0x75, 0xb8, 0xe8, 0xf2, 0xef, 0xc4, 0x77, 0xaa, 0xb4, 0xb4, 
-   0xea, 0x8b, 0x2e, 0x7f, 0x9e, 0x7e, 0x6e, 0xde, 0x54, 0x78, 0x0e, 0x69, 
-   0x55, 0xb0, 0x1f, 0xab, 0x9d, 0x37, 0xa0, 0x2a, 0x16, 0x24, 0x76, 0x5a, 
-   0xc5, 0x86, 0x8a, 0xbd, 0x56, 0xb3, 0x22, 0xb1, 0xdf, 0x6a, 0x76, 0x54, 
-   0xec, 0xb9, 0x82, 0x25, 0x59, 0xed, 0x0b, 0xde, 0xae, 0x8a, 0x21, 0x49, 
-   0x0d, 0x2b, 0x59, 0x52, 0x57, 0x6a, 0x5a, 0xc1, 0x94, 0x2c, 0xeb, 0x62, 
-   0x71, 0x49, 0xfe, 0x81, 0x4d, 0x4a, 0xe0, 0x50, 0xc1, 0xa2, 0x2e, 0x7b, 
-   0xe7, 0x8b, 0x21, 0x5f, 0x38, 0xb6, 0xb9, 0x55, 0x0d, 0x86, 0x93, 0xf9, 
-   0x27, 0xd4, 0x1f, 0x5f, 0x4f, 0xc4, 0x47, 0xba, 0xe6, 0x86, 0xc5, 0xdb, 
-   0x0b, 0xaf, 0x81, 0x6d, 0x73, 0x03, 0xe3, 0xed, 0x1b, 0x36, 0x9f, 0xbb, 
-   0x6d, 0x6e, 0x63, 0x94, 0x81, 0xdd, 0x5c, 0x7c, 0x11, 0x5c, 0x64, 0x05, 
-   0x3b, 0x13, 0x06, 0x5d, 0xc5, 0xba, 0x78, 0x2b, 0x6e, 0x5b, 0xa4, 0x34, 
-   0x3f, 0xbd, 0x37, 0x3d, 0xf8, 0x1b, 0xba, 0x47, 0xe1, 0xe3, 0x27, 0xbd, 
-   0x71, 0x4b, 0xd3, 0x71, 0xd9, 0xc3, 0x2c, 0x5d, 0x17, 0x33, 0x61, 0x3a, 
-   0x67, 0xba, 0x2e, 0x92, 0x4f, 0xb4, 0x56, 0xee, 0x43, 0x54, 0x79, 0x47, 
-   0xd7, 0x49, 0xf2, 0xa5, 0xd2, 0xaa, 0x9d, 0x4c, 0x25, 0xbd, 0x9e, 0xeb, 
-   0x7a, 0x21, 0x31, 0xcc, 0x31, 0xbd, 0x5c, 0x8a, 0xdd, 0x70, 0xb8, 0x1f, 
-   0xa5, 0xc4, 0xec, 0xe3, 0xd4, 0x42, 0x96, 0xb9, 0xd4, 0x93, 0x90, 0xe1, 
-   0x36, 0x68, 0xdb, 0x59, 0xe0, 0xf3, 0xa3, 0x00, 0x87, 0x56, 0xc5, 0x11, 
-   0xe3, 0xc6, 0x93, 0xdb, 0x9b, 0xfe, 0xa7, 0xf9, 0xa7, 0xe9, 0xf8, 0xf6, 
-   0xf2, 0xd3, 0xa2, 0x3b, 0x9a, 0x7c, 0x12, 0xb6, 0x2e, 0x06, 0x32, 0x64, 
-   0xec, 0x71, 0x16, 0xc3, 0x6e, 0x9f, 0xb7, 0x35, 0x5f, 0x30, 0xb3, 0xab, 
-   0xcb, 0x9b, 0xe1, 0x60, 0x51, 0xe0, 0x50, 0x65, 0x93, 0xcf, 0xb7, 0x35, 
-   0x77, 0xcf, 0x69, 0xef, 0x05, 0x16, 0x67, 0x95, 0xd4, 0xd0, 0x65, 0x7a, 
-   0x90, 0x59, 0x98, 0x3b, 0xec, 0xaf, 0xfd, 0x5e, 0x7f, 0xba, 0x18, 0xdd, 
-   0x5e, 0x77, 0x79, 0x6b, 0x73, 0x67, 0xcd, 0x5a, 0x37, 0xed, 0xfa, 0xa2, 
-   0x8f, 0x35, 0x29, 0xf0, 0x68, 0xd6, 0xcd, 0x27, 0xd1, 0xeb, 0x73, 0xcd, 
-   0x37, 0xcd, 0x47, 0xde, 0x13, 0x8c, 0xaf, 0x69, 0x3e, 0xe4, 0x5e, 0xbf, 
-   0xc1, 0xa3, 0xb9, 0x0a, 0xa3, 0xec, 0xce, 0xe6, 0xfd, 0x45, 0xf3, 0x45, 
-   0xc0, 0xdd, 0x33, 0x37, 0x34, 0xda, 0xb6, 0x25, 0xb6, 0x35, 0x37, 0xb1, 
-   0xa4, 0xad, 0x10, 0x49, 0x9a, 0x9b, 0x18, 0x6d, 0xdb, 0x16, 0xdb, 0x9a, 
-   0x07, 0x01, 0x49, 0x5b, 0xee, 0x55, 0xdb, 0xe6, 0xaa, 0xa1, 0x6d, 0x3b, 
-   0x62, 0xbf, 0xe6, 0xfa, 0x49, 0xda, 0xf2, 0x7e, 0xcf, 0xcc, 0x03, 0x80, 
-   0xa4, 0x2d, 0xdf, 0x6e, 0xce, 0xcc, 0x37, 0x7f, 0xda, 0xd6, 0xaa, 0x0b, 
-   0x83, 0x3e, 0x33, 0xf7, 0x68, 0x69, 0x63, 0x61, 0xd4, 0xe6, 0xbb, 0x7e, 
-   0xda, 0x58, 0x18, 0x76, 0x45, 0xd3, 0xc2, 0x8d, 0xc5, 0xf2, 0xcf, 0x8a, 
-   0xc6, 0x65, 0xd9, 0x52, 0xeb, 0x8a, 0xe6, 0x45, 0x5a, 0xf3, 0x85, 0x78, 
-   0x76, 0x96, 0x1e, 0x55, 0xc9, 0xf1, 0xf4, 0xb7, 0x9f, 0x7e, 0x39, 0x7d, 
-   0x5a, 0xbd, 0xec, 0xb6, 0xbf, 0xfd, 0xf4, 0x7f, 0xf9, 0x0b, 0x6a, 0x39, 
-   
+   0x78, 0x9c, 0xed, 0x7d, 0x6b, 0x53, 0xe3, 0x48, 0xb2, 0xe8, 0xe7, 0x9d, 
+   0x5f, 0x51, 0xa7, 0xbf, 0x6c, 0xf7, 0xbd, 0xc3, 0x80, 0xe4, 0x07, 0x26, 
+   0x62, 0x66, 0x6f, 0x18, 0x63, 0x68, 0xe2, 0x00, 0x66, 0x6d, 0xd3, 0x0f, 
+   0x6e, 0xdc, 0x70, 0x08, 0x5b, 0x06, 0xc5, 0xd8, 0x92, 0x8f, 0x24, 0x37, 
+   0x30, 0xbf, 0xfe, 0xd6, 0x43, 0x52, 0x55, 0x49, 0x59, 0xca, 0x92, 0xa1, 
+   0xe7, 0xb5, 0x4c, 0x9c, 0xd8, 0x43, 0x43, 0x56, 0x96, 0x2a, 0x33, 0x2b, 
+   0x5f, 0x95, 0x95, 0xf5, 0xf3, 0xb7, 0xf9, 0xd3, 0x7a, 0x45, 0xee, 0xfd, 
+   0xf0, 0x97, 0x77, 0xee, 0x4f, 0xce, 0x3b, 0xb2, 0x0e, 0xc2, 0xd9, 0x37, 
+   0x3f, 0xa6, 0xff, 0x62, 0xff, 0xf0, 0x9e, 0x8a, 0x7f, 0xfc, 0xeb, 0x87, 
+   0x1f, 0x08, 0xf9, 0xd9, 0x0f, 0xb7, 0x6b, 0x12, 0x7a, 0x6b, 0xff, 0x97, 
+   0x77, 0x83, 0x68, 0xbd, 0xf1, 0x62, 0x9f, 0x9c, 0x6e, 0xc3, 0x79, 0x1a, 
+   0x44, 0xe1, 0x3b, 0xb2, 0x89, 0xfd, 0x65, 0xf0, 0xf4, 0xcb, 0xbb, 0x4f, 
+   0xad, 0x93, 0xd9, 0x60, 0x74, 0x79, 0xdd, 0x1f, 0x0f, 0x67, 0xa7, 0x37, 
+   0x57, 0x03, 0x3a, 0x96, 0xd0, 0xff, 0x7e, 0xfe, 0xe6, 0xad, 0xb6, 0x7e, 
+   0x36, 0xfa, 0x6a, 0xf8, 0x69, 0x38, 0x7e, 0x47, 0xf8, 0xaf, 0x7e, 0x79, 
+   0x77, 0xf0, 0x6e, 0x1f, 0x80, 0xb9, 0x18, 0x4e, 0x26, 0x05, 0x88, 0x03, 
+   0x82, 0x0c, 0xff, 0x7d, 0xd3, 0xbf, 0x28, 0x60, 0x5c, 0x03, 0x1a, 0x0d, 
+   0xa8, 0x05, 0x02, 0x9d, 0x8d, 0x87, 0xfd, 0xa9, 0xf2, 0x45, 0x6d, 0x10, 
+   0xea, 0x6a, 0x34, 0xd5, 0x91, 0x75, 0x60, 0x64, 0x3a, 0x50, 0x17, 0x04, 
+   0xea, 0x5f, 0x7c, 0xee, 0x7f, 0x95, 0xeb, 0x3b, 0x14, 0x40, 0x3f, 0xef, 
+   0x33, 0x12, 0x57, 0x88, 0x7d, 0x1d, 0x07, 0xeb, 0x20, 0x0d, 0xbe, 0xf9, 
+   0x3a, 0x95, 0xaf, 0xc7, 0xe7, 0x97, 0x10, 0x75, 0xaf, 0x47, 0xe7, 0x57, 
+   0xd3, 0x09, 0x46, 0xde, 0xf3, 0xab, 0x21, 0x46, 0x5f, 0x06, 0x33, 0xbb, 
+   0x18, 0x8d, 0xae, 0x31, 0x1a, 0x33, 0xb8, 0xc9, 0x74, 0x7c, 0x7e, 0x8d, 
+   0xd0, 0x99, 0x82, 0xf4, 0xaf, 0xce, 0x2e, 0x94, 0x89, 0x61, 0x4a, 0xe7, 
+   0x70, 0x25, 0xa4, 0x30, 0xbd, 0x0b, 0xe0, 0xd3, 0xfe, 0x55, 0x99, 0xea, 
+   0x2a, 0x41, 0x37, 0xde, 0xfc, 0x57, 0x3f, 0xcd, 0x06, 0x7d, 0xf4, 0x56, 
+   0xe9, 0x3b, 0x32, 0x8f, 0x16, 0x92, 0x40, 0x3a, 0xc0, 0x15, 0x5b, 0xb5, 
+   0xf8, 0xbb, 0x03, 0xfd, 0xfd, 0x74, 0xb5, 0x4d, 0x1e, 0x72, 0x88, 0x36, 
+   0xfd, 0x61, 0xf5, 0xcb, 0xbb, 0x63, 0x23, 0x20, 0xe9, 0xaf, 0x56, 0x64, 
+   0x92, 0x7a, 0xa9, 0x9f, 0x0f, 0xe9, 0xd4, 0x0c, 0xa1, 0x80, 0x71, 0x4a, 
+   0xa6, 0xc1, 0xca, 0x27, 0xc7, 0x41, 0x18, 0x06, 0xe1, 0x7d, 0x3e, 0xaa, 
+   0x5b, 0x33, 0xea, 0x3c, 0x9c, 0xc7, 0xfe, 0xda, 0x0f, 0x53, 0x32, 0xf1, 
+   0xd7, 0xde, 0xe6, 0x21, 0x8a, 0x8b, 0xc9, 0x0e, 0x21, 0xf8, 0xcf, 0x5e, 
+   0x90, 0x92, 0x28, 0xac, 0x42, 0xf7, 0x20, 0xe8, 0xe3, 0xd8, 0x0b, 0xe7, 
+   0xc5, 0x82, 0x9d, 0x6e, 0x2e, 0x78, 0xcb, 0xc0, 0x5f, 0x2d, 0x72, 0xa1, 
+   0x5e, 0x2c, 0x62, 0x3f, 0x49, 0xde, 0x91, 0x24, 0xf8, 0x8d, 0xc9, 0x80, 
+   0x4b, 0x7f, 0x62, 0x4b, 0x61, 0x24, 0x26, 0xe9, 0xf3, 0x86, 0xfe, 0xce, 
+   0xcb, 0x40, 0x32, 0xee, 0x88, 0x29, 0x4c, 0xb3, 0x91, 0x34, 0x22, 0xc9, 
+   0xf6, 0x6e, 0x6f, 0x15, 0x24, 0x05, 0xb3, 0x9c, 0xc3, 0xef, 0x3b, 0xf3, 
+   0xd8, 0x4f, 0xb7, 0x71, 0x48, 0x96, 0x71, 0xb4, 0xae, 0xce, 0xcd, 0x29, 
+   0x03, 0xb0, 0x8b, 0x12, 0x8f, 0x5c, 0x6e, 0x57, 0x69, 0xb0, 0x97, 0x78, 
+   0xeb, 0x0d, 0x65, 0xdb, 0xd8, 0x4f, 0xa2, 0xd5, 0x37, 0x7f, 0x21, 0x98, 
+   0x38, 0x88, 0x56, 0x51, 0x4c, 0x8e, 0xb7, 0xcb, 0xa5, 0x1f, 0xe7, 0xb8, 
+   0xdc, 0x4c, 0x66, 0xc6, 0xb0, 0x00, 0x34, 0xc0, 0x48, 0xbc, 0x70, 0x41, 
+   0x86, 0xa3, 0xd3, 0x02, 0x73, 0x47, 0xc1, 0x6c, 0x40, 0x7d, 0xba, 0xa5, 
+   0xf2, 0xc8, 0x51, 0x6e, 0x99, 0xf6, 0xce, 0x64, 0x2d, 0xff, 0x40, 0x3e, 
+   0x38, 0xc7, 0x66, 0xc3, 0x69, 0xb7, 0x57, 0xd0, 0xbb, 0x0d, 0xd2, 0xbb, 
+   0x34, 0xfc, 0xc2, 0x4b, 0x84, 0x7c, 0xe7, 0x08, 0x9c, 0x62, 0x7c, 0x2b, 
+   0x1f, 0x7f, 0x17, 0x45, 0x2b, 0x70, 0xf0, 0x49, 0x90, 0x78, 0x77, 0x8c, 
+   0x06, 0x2b, 0xdf, 0x8b, 0x99, 0x00, 0x7f, 0x8e, 0x83, 0x14, 0xc0, 0xe4, 
+   0x5a, 0x63, 0xba, 0xdd, 0x9f, 0xa4, 0x7e, 0x38, 0x0f, 0x56, 0x39, 0x45, 
+   0x1f, 0x61, 0x8c, 0x8e, 0xfd, 0xb7, 0xa9, 0xfc, 0x31, 0x60, 0x3b, 0xa8, 
+   0x62, 0x93, 0x62, 0x09, 0xc8, 0xe5, 0xde, 0x2a, 0xf2, 0x16, 0x4d, 0x38, 
+   0x67, 0xb3, 0x53, 0x9a, 0x72, 0xce, 0x48, 0xb2, 0xd8, 0xf7, 0x16, 0xaf, 
+   0x46, 0x31, 0x18, 0x59, 0x33, 0x82, 0x09, 0x41, 0x57, 0xc8, 0x43, 0xce, 
+   0xfc, 0xd0, 0x8f, 0xbd, 0x55, 0x41, 0x9f, 0x5e, 0x4e, 0x30, 0xe0, 0xb3, 
+   0x2e, 0xfd, 0x75, 0x14, 0x3f, 0x93, 0x3b, 0x2f, 0xf1, 0x49, 0x46, 0x0f, 
+   0x12, 0x2d, 0xa9, 0x5a, 0xa0, 0x7f, 0xdd, 0x4f, 0x19, 0xd2, 0xc5, 0x76, 
+   0xbd, 0x21, 0x77, 0x19, 0xe1, 0x2b, 0xc4, 0x74, 0x61, 0xbd, 0x63, 0xda, 
+   0x07, 0x0c, 0xf7, 0x29, 0xc3, 0x0d, 0x90, 0xf0, 0xc8, 0x9a, 0x86, 0x9f, 
+   0xce, 0xf6, 0x2e, 0xbd, 0xe4, 0xd7, 0xec, 0xab, 0xf8, 0x17, 0x02, 0xf8, 
+   0x7a, 0x3b, 0xec, 0x8b, 0x7a, 0x8c, 0x87, 0x0d, 0xb9, 0x5c, 0x8f, 0xad, 
+   0xbb, 0xeb, 0x7a, 0xe7, 0xb9, 0x42, 0x48, 0x18, 0xeb, 0xf7, 0x0d, 0xe8, 
+   0x3b, 0xbb, 0x2f, 0xdf, 0x6e, 0x82, 0xf6, 0x6e, 0xd4, 0xb0, 0x43, 0x5e, 
+   0x51, 0x8f, 0x55, 0xec, 0xd7, 0xc1, 0x93, 0xbf, 0xca, 0x70, 0x9f, 0x46, 
+   0xf1, 0xda, 0x4b, 0x0b, 0xf9, 0x2c, 0xf0, 0x14, 0x32, 0xb0, 0x0d, 0xc2, 
+   0x34, 0x93, 0xff, 0x92, 0x5b, 0x15, 0xdf, 0xdf, 0x79, 0x3d, 0xfa, 0x5f, 
+   0xd5, 0x8f, 0x2c, 0x01, 0xde, 0xdd, 0xc7, 0x9d, 0x6e, 0x87, 0x2c, 0x82, 
+   0xf4, 0xc1, 0x8f, 0xfd, 0x45, 0xd5, 0xa7, 0x84, 0xe1, 0xc3, 0x28, 0x1b, 
+   0x02, 0x38, 0x97, 0xfb, 0x7c, 0x3d, 0xd0, 0xe2, 0x2e, 0xe9, 0xd6, 0xad, 
+   0x2e, 0xa7, 0x8b, 0x2f, 0x67, 0x22, 0xcc, 0xe8, 0x01, 0xba, 0x9c, 0x13, 
+   0x7f, 0x1e, 0x50, 0x9a, 0xf9, 0xe4, 0xa9, 0x8d, 0x2e, 0x45, 0xc2, 0x32, 
+   0xa9, 0x6d, 0xb0, 0x0c, 0x13, 0x5f, 0xda, 0xf8, 0x42, 0xc6, 0x54, 0x67, 
+   0x28, 0x44, 0x33, 0x2d, 0x63, 0x8a, 0x7e, 0xfc, 0xc5, 0xb4, 0xd1, 0x27, 
+   0x67, 0x4a, 0x94, 0xba, 0x64, 0x5c, 0xb3, 0x16, 0xae, 0x56, 0x55, 0x3d, 
+   0x9b, 0xbf, 0xfd, 0x2a, 0x0a, 0x7d, 0xf4, 0xcb, 0xb9, 0xe4, 0xa2, 0x5f, 
+   0x7f, 0xbb, 0x9f, 0x88, 0xcd, 0x59, 0x5d, 0x44, 0x19, 0xb2, 0x1a, 0x94, 
+   0x94, 0x20, 0x32, 0x3d, 0x02, 0x04, 0x25, 0x39, 0x39, 0x6a, 0x2d, 0xcd, 
+   0x05, 0xb3, 0xcb, 0x75, 0x86, 0xe6, 0xe8, 0x0f, 0x37, 0x34, 0x06, 0x8d, 
+   0xc9, 0x3c, 0x8a, 0xd7, 0xb5, 0x10, 0x06, 0x8c, 0x3b, 0x5a, 0x08, 0x03, 
+   0xb6, 0x8a, 0x85, 0x78, 0x53, 0x82, 0x6f, 0x4a, 0xf0, 0x4d, 0x09, 0xfe, 
+   0xa1, 0x4a, 0xf0, 0x3c, 0x5c, 0xf8, 0x4f, 0x34, 0x3c, 0x2d, 0x72, 0x55, 
+   0xe4, 0x42, 0x89, 0x9c, 0x69, 0x40, 0x0e, 0xe9, 0x3e, 0xef, 0x29, 0x58, 
+   0x6f, 0xd7, 0x84, 0x8f, 0x05, 0x62, 0xf7, 0x43, 0x57, 0xa3, 0x26, 0xa4, 
+   0x36, 0xfa, 0x52, 0x65, 0x52, 0x2c, 0xc1, 0xdc, 0x4f, 0xb2, 0x79, 0x2b, 
+   0xc8, 0xda, 0x07, 0x28, 0xb2, 0x0b, 0x3f, 0xbc, 0x4f, 0x1f, 0x80, 0xb1, 
+   0x3d, 0x74, 0x28, 0x5f, 0x02, 0x07, 0xca, 0x87, 0xb7, 0x9b, 0xc8, 0x73, 
+   0x6f, 0xef, 0x2e, 0x48, 0x51, 0xa1, 0x70, 0xba, 0x1a, 0x98, 0x03, 0x70, 
+   0xa8, 0xac, 0x06, 0x0b, 0x6e, 0xac, 0x15, 0x9d, 0xd1, 0xae, 0x8a, 0xab, 
+   0x4c, 0x31, 0x62, 0x71, 0xd5, 0x27, 0x3f, 0x4e, 0xe9, 0x52, 0xfb, 0x71, 
+   0xec, 0x3d, 0x4b, 0x6e, 0x27, 0x05, 0xa7, 0x5b, 0x10, 0xa7, 0x05, 0x79, 
+   0x58, 0x90, 0x13, 0xc4, 0x34, 0xe4, 0x11, 0x38, 0x7e, 0x77, 0x26, 0x7d, 
+   0x07, 0x6a, 0xe8, 0xe2, 0x5e, 0x98, 0x1a, 0x2d, 0x10, 0xef, 0x80, 0x29, 
+   0x94, 0x13, 0x2f, 0xf5, 0xc8, 0x74, 0x57, 0x81, 0x11, 0xa2, 0x40, 0x02, 
+   0xb1, 0x73, 0x10, 0x35, 0xd1, 0x72, 0x39, 0xf0, 0xd3, 0xfe, 0x33, 0x90, 
+   0x9f, 0x35, 0x2b, 0x3b, 0xb9, 0x36, 0xf8, 0x33, 0x2d, 0x74, 0xdd, 0x75, 
+   0x44, 0xff, 0x94, 0x6f, 0x49, 0x44, 0xba, 0x2f, 0x82, 0xd0, 0x2f, 0x81, 
+   0x9a, 0x16, 0x34, 0x8d, 0x03, 0x2f, 0xbc, 0x5f, 0x95, 0xc1, 0x4d, 0xca, 
+   0x6f, 0xfc, 0xf1, 0xab, 0x0e, 0x58, 0x5d, 0x7d, 0x2d, 0x93, 0xcf, 0x2e, 
+   0xc8, 0xe4, 0xc1, 0x5b, 0x50, 0xc5, 0xaf, 0xa5, 0x70, 0xbb, 0xed, 0x9c, 
+   0xaf, 0xff, 0xb5, 0xb7, 0x47, 0xa6, 0x0f, 0xd2, 0x81, 0x13, 0x54, 0x7c, 
+   0x0c, 0x56, 0xd4, 0x2f, 0xf2, 0xe9, 0xbf, 0x56, 0x2b, 0xaa, 0x19, 0x83, 
+   0x90, 0xdc, 0x3d, 0x93, 0x5f, 0xfd, 0x38, 0xa4, 0x9e, 0x09, 0xfd, 0x90, 
+   0x60, 0xe1, 0xf1, 0x0c, 0x0e, 0x43, 0xf6, 0x13, 0xd9, 0xdb, 0x6b, 0x94, 
+   0xac, 0xc1, 0xf7, 0xc9, 0xf0, 0x89, 0x9a, 0x86, 0x05, 0x9d, 0x37, 0x11, 
+   0x9f, 0x1e, 0xfb, 0xf3, 0x28, 0x06, 0xbc, 0x29, 0x3c, 0xe1, 0x76, 0xb5, 
+   0x5d, 0xdf, 0x51, 0x04, 0x74, 0xf7, 0x7a, 0x69, 0x1a, 0x07, 0x77, 0x5b, 
+   0x6a, 0xeb, 0x3d, 0xb6, 0xfd, 0x13, 0xcc, 0xfe, 0x61, 0xbb, 0x47, 0xa4, 
+   0xf0, 0xb8, 0xad, 0x4b, 0x4a, 0x9b, 0xc6, 0x71, 0xda, 0xd0, 0xae, 0x11, 
+   0x23, 0x26, 0xb9, 0xd5, 0x13, 0xd3, 0x4b, 0xb2, 0x1c, 0x75, 0x51, 0xba, 
+   0x08, 0x0c, 0x9f, 0xce, 0x88, 0xb0, 0x76, 0x65, 0x0c, 0x3d, 0x5c, 0x8d, 
+   0x08, 0x0c, 0xb7, 0x93, 0x82, 0x2d, 0x72, 0x4b, 0x74, 0xdb, 0x96, 0xa3, 
+   0x33, 0xfb, 0x2e, 0x10, 0x74, 0x4d, 0x7b, 0x0a, 0xa5, 0x5f, 0x14, 0x2e, 
+   0x83, 0xfb, 0x6d, 0x2c, 0x04, 0xe9, 0x38, 0x48, 0x0b, 0x35, 0x7c, 0x04, 
+   0xea, 0x9c, 0xa1, 0x17, 0xaf, 0x9e, 0xc9, 0x2d, 0xd9, 0x6e, 0xa8, 0xec, 
+   0xd1, 0xdd, 0xe3, 0x87, 0xcc, 0xe7, 0xde, 0xc9, 0x5f, 0xcf, 0x51, 0x19, 
+   0x51, 0xe0, 0x69, 0x1c, 0x8b, 0xef, 0xb0, 0x48, 0xd6, 0xf8, 0x9b, 0xf4, 
+   0x61, 0x6f, 0xea, 0x33, 0xe5, 0x5b, 0x9c, 0x45, 0x96, 0xa5, 0xd2, 0x29, 
+   0x1c, 0x89, 0xca, 0xb9, 0x25, 0xc8, 0xa2, 0xe8, 0x1b, 0x8d, 0xdf, 0xee, 
+   0x59, 0xd2, 0x9d, 0x46, 0x76, 0xaa, 0xb7, 0xad, 0x7c, 0x9a, 0x03, 0x72, 
+   0x9a, 0x69, 0x02, 0xaa, 0x05, 0x84, 0xa2, 0x49, 0xe0, 0x5d, 0x5d, 0xa0, 
+   0xbf, 0x0e, 0x36, 0x3e, 0x99, 0xf8, 0x2b, 0x7f, 0x9e, 0x56, 0x27, 0xc0, 
+   0x83, 0x30, 0xe1, 0x04, 0x07, 0x09, 0xdd, 0x9b, 0x23, 0x8a, 0x31, 0x3b, 
+   0x28, 0xb0, 0x09, 0x0e, 0x76, 0xfa, 0x5a, 0x4e, 0x0c, 0xd5, 0x12, 0xc8, 
+   0x6f, 0xed, 0xec, 0x8c, 0xbd, 0x1f, 0xa6, 0x01, 0x55, 0x85, 0x34, 0xfa, 
+   0xa5, 0xae, 0xa3, 0x30, 0x16, 0xec, 0x38, 0x83, 0xdb, 0x82, 0xea, 0x3c, 
+   0x78, 0x6a, 0x6d, 0xc8, 0x25, 0x89, 0x70, 0xa9, 0x20, 0xa3, 0xe5, 0x32, 
+   0xf1, 0x01, 0xd2, 0xe2, 0x2a, 0x6f, 0xb0, 0x8a, 0xe6, 0xbf, 0x3e, 0x52, 
+   0xd2, 0x6a, 0x1e, 0x4e, 0xf3, 0x13, 0x86, 0xec, 0x73, 0xc6, 0x3e, 0xe3, 
+   0x0f, 0x95, 0x3a, 0x6f, 0x1e, 0x84, 0xf7, 0x44, 0x39, 0xcf, 0x6d, 0x9e, 
+   0x30, 0xcf, 0x50, 0x52, 0x4f, 0xe3, 0xd1, 0x8b, 0x17, 0x16, 0x28, 0x9b, 
+   0xa5, 0xcd, 0x4f, 0x57, 0x5e, 0x2a, 0xac, 0x1d, 0xa1, 0x3f, 0xde, 0x4b, 
+   0x85, 0x02, 0x9e, 0x26, 0x30, 0xe8, 0x3d, 0x66, 0x60, 0xd8, 0x37, 0x64, 
+   0xf0, 0xe6, 0x23, 0x38, 0x2b, 0x9d, 0xc6, 0xa5, 0x80, 0x23, 0x29, 0xa6, 
+   0xee, 0x41, 0x53, 0x0b, 0xb8, 0x09, 0x87, 0x33, 0x4f, 0xb9, 0x5c, 0x45, 
+   0x1e, 0x3e, 0x27, 0x13, 0x37, 0x6a, 0xae, 0x17, 0x69, 0x71, 0xc0, 0x79, 
+   0x74, 0x04, 0xcd, 0xa9, 0xc2, 0xbd, 0x74, 0xce, 0xf1, 0xc7, 0x29, 0xf9, 
+   0x42, 0xee, 0xa2, 0x6d, 0xb8, 0xf0, 0xe2, 0xe7, 0xc2, 0xec, 0x1d, 0x1c, 
+   0x40, 0x13, 0x33, 0xe0, 0x4d, 0xe1, 0x8e, 0xa9, 0xc3, 0x32, 0x6e, 0x77, 
+   0xab, 0x9f, 0x61, 0x43, 0x6c, 0x7d, 0x9f, 0xe4, 0x9f, 0xe0, 0xa8, 0x7e, 
+   0x4d, 0xfa, 0xe0, 0x27, 0xbe, 0xf0, 0x68, 0xe8, 0xde, 0xa4, 0xaa, 0xf3, 
+   0x2e, 0xa2, 0x43, 0xf8, 0x1a, 0xf7, 0x9c, 0xbd, 0xde, 0xde, 0x21, 0xd5, 
+   0xde, 0x6c, 0xe0, 0x82, 0xbc, 0x4f, 0xa3, 0x0d, 0x71, 0xba, 0x84, 0xfa, 
+   0x9a, 0x3c, 0x20, 0xf3, 0x04, 0x54, 0xcb, 0xfd, 0x00, 0xef, 0x7d, 0x75, 
+   0x72, 0x72, 0x13, 0x72, 0xeb, 0x55, 0x59, 0x8e, 0x83, 0x5b, 0x74, 0x0d, 
+   0x0f, 0xdd, 0x0f, 0xa9, 0xb4, 0xac, 0x10, 0x5d, 0x2c, 0x3d, 0x93, 0x60, 
+   0x43, 0x3e, 0x53, 0xf7, 0x3a, 0x7a, 0x94, 0x74, 0x01, 0x83, 0x58, 0x05, 
+   0x92, 0x7c, 0xf4, 0x83, 0xfb, 0x07, 0xe6, 0x95, 0x93, 0x0d, 0xcb, 0x3f, 
+   0x41, 0xeb, 0x69, 0xdb, 0xf8, 0x17, 0x12, 0xe3, 0x67, 0x26, 0x6c, 0xb5, 
+   0x08, 0x5b, 0x78, 0x90, 0xac, 0x22, 0x3c, 0x8e, 0xd2, 0x34, 0x5a, 0x93, 
+   0x3c, 0x3f, 0x46, 0x5d, 0xc2, 0x20, 0xf4, 0x94, 0xe3, 0xc9, 0x46, 0x94, 
+   0x57, 0x11, 0x5f, 0xf8, 0xcb, 0xd4, 0x06, 0x6d, 0x33, 0x3e, 0x7c, 0x0a, 
+   0xfc, 0xc7, 0x4d, 0x14, 0xa7, 0x15, 0x19, 0x05, 0xc3, 0xcc, 0x02, 0x7a, 
+   0xe0, 0x87, 0x29, 0x15, 0xd5, 0xaf, 0x7b, 0x73, 0xbb, 0x15, 0x26, 0x8e, 
+   0xfb, 0x53, 0x1b, 0x5c, 0x62, 0x19, 0xe5, 0x97, 0x7a, 0x94, 0x07, 0x00, 
+   0xc6, 0x9a, 0xe5, 0xdd, 0xb2, 0x92, 0x2b, 0x6e, 0xf1, 0xd6, 0xde, 0x13, 
+   0x75, 0x80, 0x83, 0xcd, 0x86, 0x69, 0xd2, 0xcd, 0xca, 0x0b, 0x65, 0x48, 
+   0xed, 0x1c, 0x80, 0xae, 0x70, 0x9e, 0x3d, 0xb9, 0x7d, 0x04, 0x94, 0x91, 
+   0x94, 0x0a, 0x45, 0x1b, 0x95, 0x11, 0x04, 0xa1, 0x19, 0x41, 0x43, 0x6d, 
+   0xc6, 0x24, 0x61, 0x43, 0x1d, 0x91, 0x2f, 0x5f, 0xc9, 0x64, 0x4e, 0x0d, 
+   0xba, 0x2c, 0x59, 0x71, 0x0e, 0x8a, 0x52, 0x97, 0x3a, 0xf2, 0x7e, 0xf4, 
+   0x56, 0xcb, 0x3d, 0xb9, 0x7f, 0x9c, 0x7d, 0xa7, 0x4b, 0x05, 0x9f, 0x6a, 
+   0x11, 0x2e, 0xf8, 0x3b, 0xae, 0x50, 0xc7, 0x5e, 0xec, 0x25, 0x1c, 0xf9, 
+   0x8e, 0xab, 0xbf, 0xe5, 0x8b, 0xf7, 0x39, 0x43, 0xcb, 0x02, 0xdb, 0xb5, 
+   0xa1, 0xc2, 0x6d, 0xae, 0xc6, 0xde, 0xdf, 0xce, 0x59, 0x96, 0xf1, 0x36, 
+   0xf9, 0xf0, 0xd2, 0xa5, 0xe7, 0xdf, 0x54, 0x8b, 0xb1, 0xe1, 0x7a, 0xd5, 
+   0xd2, 0x24, 0xee, 0x70, 0x12, 0x2d, 0x12, 0x91, 0x31, 0x9c, 0x5b, 0xb3, 
+   0xe8, 0x93, 0x68, 0x4b, 0x1d, 0x99, 0xbd, 0x2c, 0xd5, 0x4f, 0xd9, 0x12, 
+   0x46, 0xe1, 0xde, 0x3a, 0xd1, 0x72, 0x33, 0xaa, 0xbf, 0x5d, 0x39, 0x07, 
+   0xaf, 0xa2, 0xe4, 0xdf, 0xd5, 0x5f, 0x51, 0xe7, 0x2d, 0x8b, 0x88, 0x98, 
+   0x1b, 0xa7, 0xf9, 0x08, 0xae, 0x82, 0xf0, 0x50, 0xd3, 0x44, 0x70, 0x9a, 
+   0x9e, 0x23, 0x60, 0x43, 0x09, 0xa3, 0x16, 0x76, 0x08, 0x20, 0xa1, 0xbb, 
+   0x78, 0xde, 0x5c, 0x81, 0x76, 0xdc, 0x1e, 0x9a, 0xc4, 0x50, 0xc0, 0xdd, 
+   0x4e, 0xb7, 0x51, 0x22, 0xa7, 0x4c, 0x96, 0x73, 0x6a, 0x6b, 0xa9, 0xd7, 
+   0x8d, 0x90, 0xa7, 0xf3, 0x9f, 0x42, 0x9e, 0xfe, 0x36, 0x8d, 0xf6, 0x02, 
+   0x41, 0x13, 0xe6, 0xf2, 0x73, 0x72, 0xf1, 0x4c, 0x0f, 0xe1, 0x49, 0x3a, 
+   0x9e, 0xea, 0x84, 0x24, 0x12, 0x0f, 0x47, 0xd4, 0x53, 0xc1, 0x6e, 0x9b, 
+   0x67, 0xe1, 0xc4, 0x99, 0x14, 0xf7, 0x59, 0x20, 0x9c, 0x78, 0x6c, 0xc2, 
+   0xcb, 0xc2, 0x94, 0x60, 0x8f, 0xbc, 0x6f, 0x3f, 0x7d, 0x80, 0x50, 0x55, 
+   0xc2, 0x93, 0x2a, 0xae, 0x4c, 0xe1, 0xbe, 0xa7, 0xbb, 0x8f, 0x9d, 0x35, 
+   0x4a, 0xdd, 0x20, 0x13, 0x22, 0xcc, 0xfa, 0x60, 0x7e, 0x80, 0xd0, 0xac, 
+   0x75, 0x58, 0xaa, 0x89, 0x19, 0x03, 0xb1, 0xca, 0x74, 0x27, 0xe6, 0xd2, 
+   0x3e, 0x8b, 0x64, 0x4b, 0x59, 0xf2, 0xd7, 0xe2, 0xb4, 0x35, 0x81, 0xa3, 
+   0x06, 0x0b, 0x57, 0xca, 0x80, 0xd0, 0xc3, 0xeb, 0x0f, 0x95, 0x75, 0x63, 
+   0x9a, 0x75, 0xcc, 0x12, 0x77, 0x31, 0xa6, 0x5b, 0x5b, 0x35, 0xa7, 0xca, 
+   0xcd, 0x74, 0xeb, 0x21, 0x9e, 0xab, 0xe1, 0xe9, 0x9e, 0xbd, 0xdb, 0x7d, 
+   0xf1, 0xff, 0x07, 0xd1, 0x37, 0xb2, 0x10, 0x07, 0xb6, 0x00, 0x36, 0x3c, 
+   0x69, 0x93, 0x61, 0x23, 0x37, 0x3c, 0xff, 0x43, 0x4e, 0x82, 0xd8, 0xe7, 
+   0xd9, 0x18, 0x72, 0x36, 0xdd, 0x3f, 0x1b, 0x02, 0x28, 0xf1, 0x9d, 0x26, 
+   0xd2, 0x28, 0xa4, 0xc8, 0x56, 0xc0, 0x59, 0x9b, 0x43, 0x7c, 0x7b, 0x65, 
+   0xf1, 0x75, 0x96, 0x20, 0x2c, 0x4a, 0xeb, 0x2a, 0x98, 0xf0, 0xe0, 0x3f, 
+   0x3b, 0xda, 0x37, 0x1d, 0x73, 0x1e, 0x5a, 0xe4, 0xcf, 0x7f, 0x9f, 0x73, 
+   0x4e, 0x20, 0xa2, 0xca, 0x8e, 0x74, 0x0d, 0xc9, 0xa4, 0x4a, 0x0c, 0x03, 
+   0x24, 0x8d, 0xa9, 0xc0, 0x7d, 0x3c, 0x19, 0x8b, 0xaa, 0xb6, 0x5c, 0x05, 
+   0xd6, 0x9f, 0xc7, 0x77, 0x2d, 0x4e, 0xb0, 0xe3, 0x62, 0x6b, 0xcc, 0xf9, 
+   0xae, 0x20, 0xa6, 0x83, 0x77, 0x13, 0xa9, 0x2a, 0x08, 0x2a, 0x47, 0xfc, 
+   0x26, 0x12, 0x9a, 0xa6, 0xde, 0xed, 0x0c, 0xbf, 0xa9, 0x65, 0xe8, 0xe2, 
+   0xbb, 0xca, 0xd2, 0x30, 0x74, 0xf1, 0xcd, 0x94, 0x9b, 0x05, 0x11, 0x74, 
+   0x4a, 0x1c, 0x8d, 0xc2, 0xd8, 0xcc, 0x28, 0x98, 0x71, 0x58, 0xa8, 0xdb, 
+   0x6c, 0x0b, 0xed, 0x5e, 0xde, 0x0d, 0x2b, 0x58, 0x19, 0xa0, 0xca, 0x20, 
+   0xcb, 0xe9, 0xd4, 0xe8, 0x53, 0xa1, 0x96, 0x69, 0x94, 0x2b, 0xce, 0x42, 
+   0x80, 0x63, 0x03, 0x3b, 0xcb, 0x41, 0x59, 0xbc, 0x5d, 0x87, 0x46, 0x2c, 
+   0xcd, 0x62, 0xe3, 0x33, 0x7f, 0x4d, 0x4d, 0x45, 0x6e, 0x8a, 0x8a, 0x95, 
+   0xb8, 0x9d, 0x76, 0x8d, 0xd7, 0x9d, 0x99, 0x04, 0x67, 0x37, 0xf3, 0x97, 
+   0x8d, 0x3e, 0xd8, 0x25, 0xcb, 0x97, 0xa4, 0xf1, 0x76, 0x5e, 0x54, 0xe7, 
+   0x8a, 0x43, 0xa9, 0xb1, 0x38, 0x94, 0x82, 0x32, 0x8b, 0x54, 0x8b, 0x8b, 
+   0xcb, 0x0c, 0x02, 0x32, 0x48, 0xe8, 0x94, 0xec, 0xcc, 0x8f, 0xa4, 0x0f, 
+   0xac, 0x3c, 0xd8, 0xb7, 0x2b, 0x10, 0x36, 0x26, 0x0d, 0xa9, 0x51, 0x9c, 
+   0xaf, 0xb6, 0x0b, 0x71, 0x2e, 0xc7, 0x8f, 0xc8, 0x16, 0xe4, 0x9b, 0x38, 
+   0xda, 0xa6, 0x36, 0xc9, 0x7b, 0x41, 0x66, 0x76, 0x90, 0x45, 0xf1, 0x16, 
+   0xf9, 0x62, 0x7c, 0xd9, 0xf2, 0xec, 0xed, 0x26, 0x0c, 0x96, 0x54, 0x71, 
+   0x26, 0xe4, 0x7d, 0x18, 0xa5, 0x64, 0xcb, 0x12, 0xe6, 0xf3, 0x6d, 0x4c, 
+   0x95, 0x52, 0xba, 0x7a, 0x86, 0x76, 0x97, 0x7b, 0x87, 0x72, 0xd3, 0x3c, 
+   0xd9, 0x27, 0x2f, 0x7e, 0xa6, 0x4b, 0x48, 0xaa, 0x22, 0xda, 0x6a, 0x8e, 
+   0x76, 0xc0, 0x74, 0x91, 0x79, 0x0b, 0xb7, 0xef, 0xc0, 0x3d, 0x9c, 0xa5, 
+   0x1f, 0x59, 0x30, 0xbc, 0xdd, 0xb0, 0x93, 0x53, 0xaa, 0x62, 0xf3, 0xd3, 
+   0x53, 0x30, 0x9d, 0x58, 0x9e, 0xb6, 0x20, 0x98, 0x79, 0xea, 0x5e, 0x65, 
+   0x31, 0x40, 0x20, 0x2d, 0x84, 0xe2, 0xc5, 0xfc, 0x70, 0x2c, 0x18, 0xa2, 
+   0xcf, 0xd5, 0x2f, 0x4e, 0x5b, 0x85, 0x27, 0x9c, 0x08, 0x17, 0xe7, 0x4e, 
+   0x49, 0x94, 0x2a, 0x6e, 0x7a, 0xbb, 0x29, 0xfa, 0x69, 0x94, 0xd2, 0x18, 
+   0xb0, 0x98, 0x24, 0xd1, 0x02, 0x41, 0x05, 0x71, 0xa7, 0x29, 0x62, 0x84, 
+   0xdf, 0x4e, 0xf7, 0x95, 0x18, 0xae, 0xcf, 0x6a, 0xc1, 0x6e, 0x65, 0x66, 
+   0x33, 0xbf, 0xa5, 0x6d, 0x78, 0x85, 0x3d, 0x68, 0xc1, 0x94, 0xea, 0x7c, 
+   0x8d, 0xf8, 0xee, 0x56, 0xd7, 0x64, 0x31, 0x85, 0x25, 0xef, 0xdd, 0xc3, 
+   0x5d, 0x90, 0x23, 0xfc, 0x77, 0x7b, 0xaf, 0xc4, 0xff, 0xea, 0xcc, 0x16, 
+   0x32, 0xd0, 0x82, 0xb6, 0xe1, 0xcf, 0xfb, 0xc2, 0x38, 0x01, 0x86, 0x4a, 
+   0x32, 0xc3, 0x6c, 0xab, 0x6a, 0xdc, 0x13, 0xe3, 0x52, 0xe1, 0x02, 0x8b, 
+   0xe3, 0x67, 0xc6, 0x8c, 0x75, 0x10, 0x6e, 0x13, 0x69, 0xa2, 0x7b, 0x80, 
+   0xaa, 0x34, 0xb2, 0x63, 0x42, 0x3f, 0x77, 0x01, 0xb0, 0xb2, 0xf1, 0x2e, 
+   0xfe, 0x74, 0x7d, 0x49, 0xbf, 0x48, 0x3d, 0x40, 0x95, 0xc8, 0x76, 0x92, 
+   0xb9, 0x3a, 0x84, 0x90, 0x9c, 0x29, 0x4c, 0xf9, 0x79, 0x9f, 0xdf, 0x54, 
+   0xfe, 0xd7, 0x0f, 0x3f, 0x2b, 0x37, 0x96, 0x5b, 0x3f, 0xb5, 0x94, 0x1b, 
+   0xcb, 0xad, 0x96, 0x72, 0x63, 0xb9, 0xed, 0xbe, 0xdd, 0x58, 0xe6, 0x5f, 
+   0xfd, 0x47, 0xdd, 0x58, 0x3e, 0x5e, 0xd1, 0x60, 0xa9, 0x38, 0x8b, 0x53, 
+   0x09, 0x7d, 0x7c, 0x31, 0xbc, 0x3a, 0x99, 0x9d, 0xf6, 0x07, 0xd3, 0xd1, 
+   0x18, 0x22, 0xf4, 0xed, 0x70, 0x3c, 0x42, 0xe8, 0x3c, 0xba, 0x1a, 0x22, 
+   0x64, 0x9e, 0x8c, 0x07, 0x94, 0xa3, 0x17, 0xa3, 0x31, 0x42, 0xea, 0xf3, 
+   0xab, 0x4f, 0xb3, 0x2a, 0x2c, 0x4c, 0xf1, 0x93, 0xc9, 0xb4, 0x04, 0x07, 
+   0xd3, 0x9c, 0xe1, 0xac, 0xc2, 0xc2, 0x84, 0x67, 0x73, 0xf7, 0x2f, 0xae, 
+   0x3f, 0xf6, 0x11, 0xda, 0xe7, 0xdf, 0xa9, 0xc3, 0x1e, 0x1a, 0xbf, 0x53, 
+   0x87, 0xeb, 0xd5, 0x7e, 0xa7, 0x0e, 0x7b, 0x04, 0xc2, 0x0e, 0x46, 0x57, 
+   0x95, 0x15, 0x39, 0x30, 0x73, 0x18, 0x5a, 0x10, 0x1c, 0xe6, 0x94, 0x00, 
+   0xd5, 0xbf, 0xc1, 0x31, 0x33, 0x0b, 0x04, 0x87, 0xf9, 0x55, 0xd0, 0x6b, 
+   0x36, 0xe9, 0x4f, 0x6f, 0xc6, 0x74, 0xb7, 0xc8, 0x11, 0x6d, 0x0b, 0xe1, 
+   0x15, 0xa9, 0x8f, 0xaa, 0xe8, 0x5e, 0x8e, 0x4e, 0x86, 0x90, 0xe0, 0xf6, 
+   0x4f, 0x4e, 0x10, 0xb9, 0x9d, 0xdc, 0x1c, 0x23, 0x72, 0x3b, 0x56, 0x41, 
+   0x60, 0x2a, 0x5c, 0x9e, 0x5f, 0x21, 0x82, 0x7a, 0xd9, 0xff, 0x82, 0x88, 
+   0xe8, 0xe5, 0x0d, 0xa6, 0x11, 0x26, 0x83, 0xf1, 0x70, 0x58, 0xb9, 0x4d, 
+   0x5f, 0x96, 0xb4, 0xfe, 0xf8, 0xbf, 0x15, 0x20, 0x58, 0x1c, 0x2f, 0xce, 
+   0xcf, 0x3e, 0x4e, 0x15, 0xa8, 0x5e, 0x2d, 0xe9, 0xf3, 0xdb, 0x24, 0xa3, 
+   0x8d, 0x4e, 0xfa, 0x09, 0x45, 0x31, 0x38, 0xbf, 0x98, 0x8d, 0xae, 0x77, 
+   0xd4, 0x19, 0xff, 0x3d, 0x1c, 0x5e, 0x63, 0xc4, 0x1f, 0x5e, 0x5f, 0xf4, 
+   0x07, 0x43, 0x54, 0x65, 0x0c, 0x50, 0x4d, 0x31, 0x1c, 0x58, 0x28, 0x89, 
+   0xe1, 0x78, 0x8a, 0x30, 0x81, 0x4e, 0xf5, 0x79, 0xdc, 0xbf, 0xc6, 0xb8, 
+   0x30, 0xd4, 0xa1, 0xde, 0x5a, 0x49, 0x7c, 0x9f, 0x56, 0x12, 0x20, 0x65, 
+   0x66, 0xd3, 0x53, 0xb9, 0x70, 0x18, 0x90, 0x53, 0x47, 0x83, 0x33, 0x6d, 
+   0x94, 0x8c, 0x42, 0x1a, 0x2c, 0xac, 0xbb, 0x25, 0x95, 0x34, 0x60, 0x58, 
+   0x79, 0x17, 0x94, 0x52, 0x61, 0x5d, 0x98, 0x95, 0x3a, 0xb5, 0xb4, 0x01, 
+   0x30, 0x5f, 0x55, 0x8a, 0x69, 0xe0, 0x6e, 0xad, 0x18, 0x4e, 0x2f, 0x6f, 
+   0xc8, 0x69, 0xb0, 0xe2, 0xb9, 0x6d, 0x55, 0x0e, 0xe9, 0xef, 0x67, 0xa7, 
+   0xe7, 0x17, 0xdc, 0xb3, 0x29, 0xbc, 0x39, 0x79, 0xa5, 0x80, 0xc5, 0x07, 
+   0x57, 0x14, 0x81, 0x28, 0x44, 0x5a, 0x07, 0x1b, 0x56, 0x5b, 0x4d, 0x91, 
+   0xfc, 0xc8, 0xcb, 0x28, 0x8a, 0x9f, 0xbd, 0xfb, 0xec, 0x67, 0x19, 0x37, 
+   0x94, 0x54, 0xe8, 0x8c, 0x52, 0x70, 0x76, 0x49, 0x57, 0x78, 0x45, 0x9d, 
+   0x89, 0xd9, 0x65, 0xff, 0x8c, 0xfd, 0x02, 0x11, 0x73, 0x70, 0xd8, 0xd5, 
+   0xb0, 0x3f, 0x46, 0x44, 0x9f, 0x8d, 0x63, 0x60, 0xe6, 0xf9, 0x8c, 0x9a, 
+   0x1e, 0x18, 0xa7, 0x4d, 0xd8, 0x92, 0x21, 0xac, 0x79, 0x24, 0xff, 0xa1, 
+   0x34, 0xa3, 0xc1, 0x2e, 0x80, 0xe3, 0xb4, 0x19, 0xe1, 0x7d, 0xa3, 0x0d, 
+   0xe4, 0x34, 0x2a, 0xcd, 0x07, 0x6f, 0x0d, 0x70, 0x98, 0x36, 0xdd, 0xa1, 
+   0x79, 0x81, 0x05, 0x2b, 0xa0, 0xf5, 0xc1, 0xdb, 0x06, 0x1c, 0xa6, 0xcd, 
+   0x07, 0xef, 0x20, 0x75, 0x1c, 0xb4, 0x3a, 0x83, 0x2b, 0x04, 0x0d, 0xd3, 
+   0xe5, 0xc5, 0x81, 0x97, 0xd7, 0xbf, 0x3a, 0x9f, 0x8c, 0xa6, 0xe3, 0xd1, 
+   0xf5, 0xf9, 0x60, 0xe6, 0xce, 0x1c, 0xcc, 0x31, 0x52, 0xc1, 0xdb, 0x2a, 
+   0x38, 0xac, 0x39, 0x55, 0xf0, 0x9e, 0x0a, 0x0e, 0x0b, 0x85, 0x0a, 0xee, 
+   0x74, 0x55, 0xf8, 0x4e, 0xbd, 0x17, 0x45, 0xe3, 0xbf, 0xe2, 0xc8, 0x07, 
+   0x70, 0xa6, 0x46, 0xe3, 0x93, 0xe1, 0x38, 0x77, 0x12, 0x8b, 0x30, 0xae, 
+   0xed, 0x40, 0x66, 0xe8, 0x80, 0xfe, 0x87, 0xec, 0x4e, 0x0a, 0xe1, 0x20, 
+   0x1b, 0xd1, 0x71, 0x1c, 0x07, 0xd9, 0x73, 0xa7, 0xd1, 0x6a, 0x15, 0x3d, 
+   0x26, 0x96, 0xc6, 0xf4, 0x73, 0xec, 0x6d, 0x80, 0xa5, 0x31, 0x73, 0x2c, 
+   0xdc, 0x44, 0x60, 0x5d, 0xa5, 0x09, 0x39, 0x0a, 0x76, 0x2a, 0x4b, 0xa8, 
+   0xfb, 0x41, 0x43, 0x3a, 0xf4, 0xfc, 0x4a, 0x0e, 0x18, 0x5c, 0xf4, 0x2f, 
+   0x01, 0x9f, 0xc6, 0x08, 0x7f, 0x79, 0x3e, 0x1e, 0x43, 0x11, 0x91, 0x71, 
+   0x80, 0x60, 0x11, 0x7a, 0x71, 0xb0, 0x3c, 0xc3, 0x6c, 0x74, 0xa5, 0x78, 
+   0x51, 0xf5, 0xae, 0x36, 0xb3, 0x03, 0x65, 0x5f, 0x8f, 0xd9, 0x00, 0xe6, 
+   0x27, 0x20, 0x32, 0xc1, 0xbb, 0xbe, 0x10, 0xea, 0x75, 0xff, 0x28, 0x2a, 
+   0xd3, 0xaf, 0x29, 0x0a, 0x3f, 0x65, 0x2d, 0x8a, 0x6a, 0xe5, 0x44, 0x0c, 
+   0xa3, 0x7e, 0x76, 0x36, 0x8c, 0x5f, 0xc7, 0x40, 0x04, 0x47, 0x8c, 0xf9, 
+   0x32, 0xf8, 0x78, 0x96, 0x0d, 0xca, 0x9a, 0x3f, 0xd5, 0x8a, 0x92, 0x18, 
+   0x34, 0xb8, 0xbc, 0x36, 0x8e, 0x83, 0xb7, 0xa7, 0x18, 0x77, 0x43, 0x95, 
+   0xc7, 0x8f, 0xa2, 0xd9, 0xcb, 0x85, 0x53, 0xfa, 0x48, 0x78, 0x9f, 0xe6, 
+   0xe3, 0x14, 0xef, 0x1f, 0x56, 0xd6, 0x19, 0x05, 0x2e, 0x51, 0xf5, 0x9c, 
+   0x03, 0x2a, 0x18, 0x61, 0xd7, 0x25, 0x63, 0xc5, 0x55, 0xce, 0x0a, 0xea, 
+   0xb8, 0x22, 0x9a, 0x58, 0x8c, 0x18, 0x8d, 0xb3, 0x01, 0xd4, 0x87, 0x45, 
+   0x54, 0x70, 0xc6, 0x82, 0x62, 0xc4, 0xd5, 0x68, 0x8a, 0xa9, 0xdf, 0xb1, 
+   0x7f, 0xbf, 0x5d, 0xa9, 0xbc, 0xad, 0xd7, 0x57, 0xd9, 0x91, 0x87, 0xa8, 
+   0x62, 0x27, 0x7d, 0x20, 0x47, 0xf4, 0xa9, 0x3f, 0xfe, 0x7a, 0x7e, 0x75, 
+   0x36, 0x3b, 0xbd, 0xe8, 0x9f, 0x4d, 0x66, 0xfd, 0xc1, 0xf4, 0x7c, 0x74, 
+   0x05, 0x49, 0xe6, 0x36, 0x9c, 0x3f, 0x78, 0xe1, 0x3d, 0x74, 0x18, 0xad, 
+   0xc1, 0xfd, 0xe6, 0xc7, 0x11, 0x74, 0x55, 0x5c, 0x03, 0xe2, 0xe9, 0x32, 
+   0x5d, 0xd0, 0x0c, 0x0b, 0x28, 0x95, 0x1b, 0xa8, 0x5f, 0x7e, 0x39, 0xbc, 
+   0x1c, 0x8d, 0xbf, 0xce, 0x4e, 0x47, 0xe3, 0x4b, 0xaa, 0x62, 0x20, 0x5a, 
+   0x19, 0xea, 0x0c, 0x2a, 0x17, 0xf4, 0xbc, 0x38, 0xe5, 0x6d, 0x9b, 0x6a, 
+   0xbf, 0xf9, 0xe6, 0x78, 0x6f, 0xc5, 0x61, 0xc9, 0x7b, 0x87, 0xdc, 0x9c, 
+   0x9f, 0x12, 0x51, 0x9c, 0xf5, 0x18, 0x2c, 0xfc, 0x0f, 0xc8, 0xae, 0x51, 
+   0x86, 0xba, 0x72, 0x68, 0x52, 0x1a, 0x0b, 0xef, 0x1c, 0x06, 0xfe, 0xfe, 
+   0x2a, 0x62, 0x52, 0xf2, 0x01, 0xd9, 0x2c, 0x1c, 0x54, 0x83, 0xab, 0x17, 
+   0x8e, 0x52, 0x41, 0x84, 0x4a, 0x5b, 0x2a, 0xbb, 0xe7, 0x94, 0xaa, 0x43, 
+   0x63, 0x62, 0x20, 0x31, 0x5e, 0xba, 0xd7, 0xc0, 0xda, 0xc0, 0x3d, 0x4e, 
+   0x0d, 0xc0, 0xa3, 0xba, 0x40, 0xe0, 0x4a, 0xca, 0x84, 0x30, 0x7c, 0xf5, 
+   0x79, 0x48, 0x99, 0x1a, 0x7a, 0xab, 0xec, 0xd6, 0x8e, 0xfa, 0xd5, 0x34, 
+   0x78, 0x19, 0x8e, 0xaf, 0xfa, 0x17, 0xb3, 0xe9, 0xd7, 0x6b, 0xf0, 0xab, 
+   0x7b, 0x01, 0xf2, 0xbd, 0xbd, 0x6d, 0x80, 0x7c, 0x30, 0x50, 0xb4, 0x57, 
+   0xba, 0xc3, 0x1a, 0x20, 0x6c, 0x72, 0xba, 0xca, 0x24, 0xb0, 0x36, 0x73, 
+   0xba, 0x4b, 0x44, 0x8d, 0xb5, 0xdc, 0x00, 0xd1, 0x46, 0x2d, 0x57, 0x99, 
+   0x06, 0xd6, 0x3f, 0x2d, 0x77, 0x59, 0xd1, 0x37, 0x18, 0xd5, 0x8f, 0xaf, 
+   0xaf, 0x0d, 0x44, 0x67, 0x7f, 0x01, 0x27, 0x41, 0x68, 0x0e, 0x95, 0x4d, 
+   0xea, 0xd4, 0xa8, 0x56, 0x4a, 0x62, 0x5f, 0x29, 0x6e, 0x4f, 0xec, 0x20, 
+   0x21, 0x0b, 0x36, 0x70, 0xa6, 0x12, 0x06, 0xfe, 0x68, 0x01, 0xe7, 0x62, 
+   0x9f, 0x2e, 0xc0, 0xaa, 0x2d, 0x24, 0x0c, 0xdf, 0x2f, 0x6a, 0xe4, 0xc8, 
+   0xd4, 0x8b, 0xef, 0xfd, 0x94, 0x1a, 0x48, 0x6f, 0x5d, 0x72, 0x23, 0xc6, 
+   0xc3, 0x2b, 0xe6, 0x60, 0x4e, 0xfb, 0xe3, 0xb3, 0xe1, 0x74, 0x96, 0x39, 
+   0x4b, 0x88, 0x4f, 0x11, 0x96, 0x5b, 0x33, 0x88, 0x90, 0x33, 0x8c, 0xc8, 
+   0x9c, 0x4d, 0xc0, 0xcc, 0x03, 0x18, 0x51, 0x86, 0x54, 0xe7, 0x6a, 0xcb, 
+   0x13, 0xe3, 0xfe, 0xef, 0xc1, 0x8f, 0xce, 0xff, 0x23, 0x4b, 0xea, 0xff, 
+   0x2e, 0x9d, 0x2e, 0x3c, 0x72, 0x13, 0x25, 0xda, 0x82, 0x8b, 0x81, 0xf5, 
+   0xc3, 0xd8, 0x21, 0x89, 0xd4, 0x01, 0xca, 0xb2, 0x24, 0x0e, 0xfe, 0xc5, 
+   0xac, 0xe2, 0x9b, 0xc2, 0xfa, 0xf7, 0xac, 0xd6, 0x62, 0xfa, 0xcf, 0x84, 
+   0xc4, 0xcc, 0x30, 0x65, 0x38, 0x0d, 0x94, 0x1d, 0x6d, 0xd3, 0xcd, 0x36, 
+   0x25, 0xe7, 0x6b, 0x56, 0x46, 0x07, 0x59, 0x93, 0xd1, 0xcd, 0xf4, 0xfa, 
+   0x66, 0x3a, 0xa3, 0x5a, 0xef, 0x6c, 0x58, 0xb2, 0x29, 0x74, 0xe2, 0x1f, 
+   0xfe, 0x21, 0xc6, 0xd0, 0x18, 0x7d, 0xb3, 0x61, 0x8a, 0xfc, 0x31, 0xa0, 
+   0xf2, 0x95, 0x3e, 0xf8, 0x41, 0x4c, 0x98, 0x5d, 0x0c, 0xfd, 0x55, 0xc2, 
+   0x67, 0x5a, 0x88, 0x26, 0x8b, 0xec, 0x94, 0x8f, 0xfa, 0xdc, 0xe2, 0xd6, 
+   0x50, 0x1a, 0xfd, 0xf0, 0x0f, 0xf6, 0x8b, 0x87, 0xe0, 0xfe, 0x81, 0xff, 
+   0xe6, 0x27, 0x8e, 0x17, 0xa4, 0x41, 0x12, 0xdf, 0xdf, 0xf5, 0x66, 0xde, 
+   0x6a, 0xf3, 0xe0, 0x01, 0xfd, 0x59, 0x2a, 0xa0, 0x88, 0xf8, 0x51, 0x08, 
+   0xe7, 0x60, 0xe6, 0xa9, 0x9a, 0x00, 0xd6, 0x5a, 0x39, 0x20, 0x62, 0x8c, 
+   0xbc, 0xbb, 0xfb, 0xd8, 0xe9, 0x74, 0x3a, 0x88, 0x86, 0xe3, 0x9f, 0xbf, 
+   0xb7, 0xf6, 0x92, 0x5f, 0x29, 0x3d, 0x2a, 0x63, 0x60, 0x95, 0xc7, 0xc0, 
+   0xda, 0xf4, 0x3f, 0x44, 0xef, 0x89, 0xf2, 0x34, 0xc4, 0x75, 0x8b, 0x1d, 
+   0x67, 0x39, 0xbb, 0x67, 0xff, 0x43, 0x57, 0xb5, 0x44, 0xf4, 0x24, 0x2b, 
+   0x95, 0xe3, 0xfb, 0xbd, 0x5e, 0x57, 0xc6, 0xf7, 0x02, 0x88, 0x94, 0x15, 
+   0x66, 0x19, 0x2e, 0x07, 0x43, 0xce, 0x08, 0xc4, 0xb4, 0x81, 0x9c, 0xd6, 
+   0x10, 0x07, 0xb3, 0x79, 0x03, 0x6d, 0x5e, 0x98, 0x35, 0x71, 0x0e, 0x86, 
+   0x44, 0xbe, 0x62, 0x5e, 0xd5, 0xcc, 0xc1, 0xfc, 0x60, 0xf3, 0x6e, 0x39, 
+   0xc6, 0x1c, 0x0e, 0x66, 0x48, 0x9c, 0x83, 0x21, 0x99, 0x40, 0x36, 0x2f, 
+   0x37, 0x6b, 0x48, 0x16, 0x30, 0xbe, 0x17, 0x50, 0x12, 0xce, 0xc0, 0x8f, 
+   0x1c, 0x0c, 0x49, 0xfe, 0x89, 0x79, 0x15, 0x3a, 0x1b, 0x72, 0x7e, 0x6c, 
+   0x5e, 0x6d, 0x1d, 0xae, 0x81, 0x1f, 0x39, 0x58, 0x01, 0x67, 0xe0, 0x07, 
+   0x9f, 0x57, 0xdd, 0x78, 0x26, 0x7e, 0x08, 0x28, 0x89, 0xcf, 0xc0, 0x8f, 
+   0x1c, 0xac, 0x80, 0x33, 0xf0, 0x83, 0x55, 0x7e, 0x6a, 0x70, 0x46, 0x7e, 
+   0xf4, 0xf4, 0x75, 0x98, 0xf8, 0x91, 0x81, 0x15, 0x70, 0x06, 0x7e, 0xe4, 
+   0x60, 0x85, 0x06, 0x31, 0xf3, 0xa3, 0xa7, 0xae, 0xb7, 0x65, 0xe2, 0x47, 
+   0x4f, 0xa7, 0x73, 0xcb, 0xc0, 0x8f, 0x1c, 0xac, 0x80, 0x33, 0xf3, 0x83, 
+   0x7b, 0x77, 0x05, 0x9c, 0x89, 0x1f, 0x3d, 0x9d, 0xce, 0x2d, 0x03, 0x3f, 
+   0x72, 0xb0, 0x02, 0x0e, 0xe6, 0x07, 0xd3, 0xd2, 0x4f, 0x3d, 0x75, 0xbd, 
+   0x87, 0x7a, 0x16, 0xd8, 0xf0, 0xb1, 0x4f, 0x1a, 0x13, 0x5b, 0x3d, 0x7c, 
+   0xd0, 0x5d, 0x92, 0xce, 0xf5, 0x2f, 0x3a, 0xd2, 0x7d, 0x03, 0xd0, 0x41, 
+   0x29, 0xab, 0x2b, 0xd6, 0xac, 0x06, 0x1d, 0xc4, 0xfc, 0x1e, 0x6d, 0x90, 
+   0x63, 0x31, 0x88, 0x79, 0x41, 0xda, 0x20, 0xd7, 0x6a, 0xa6, 0x44, 0x23, 
+   0x44, 0xbb, 0x85, 0x0f, 0x4a, 0xca, 0xa2, 0xc8, 0xac, 0x4a, 0x65, 0x90, 
+   0xc1, 0x51, 0xb8, 0xdd, 0x9f, 0x90, 0x5d, 0x9c, 0x85, 0x19, 0x6b, 0x9d, 
+   0x01, 0x64, 0xf7, 0xab, 0xde, 0xe0, 0x3c, 0x5a, 0x6f, 0xa8, 0x4b, 0x16, 
+   0xa6, 0xb6, 0x5e, 0x66, 0x31, 0xa0, 0xe4, 0x6e, 0x0a, 0x97, 0x88, 0xc3, 
+   0x30, 0x67, 0xe3, 0x47, 0xb2, 0xf1, 0x16, 0xc2, 0xc9, 0x00, 0x5d, 0x8b, 
+   0x12, 0x32, 0xb0, 0xaf, 0x59, 0x05, 0xde, 0x6d, 0xcf, 0xb2, 0x6e, 0x5c, 
+   0x3d, 0xcd, 0x37, 0xc8, 0xca, 0x9a, 0xb2, 0x93, 0x4d, 0x3e, 0xbb, 0xf8, 
+   0x10, 0x65, 0x7e, 0x53, 0xd8, 0xc9, 0xeb, 0xca, 0xa1, 0xa0, 0xf3, 0x7c, 
+   0xfa, 0x91, 0x3a, 0xb7, 0xa6, 0x90, 0x13, 0x6e, 0x31, 0xa6, 0x47, 0xfc, 
+   0x67, 0xd8, 0x61, 0x74, 0x1f, 0x59, 0x34, 0xc5, 0xd0, 0x2f, 0x3b, 0x41, 
+   0x40, 0xbf, 0xf5, 0xbc, 0xc7, 0x7a, 0x36, 0x8a, 0x37, 0x5e, 0x17, 0xb0, 
+   0x1a, 0x80, 0x93, 0x03, 0xb0, 0xc6, 0xeb, 0xc0, 0xdf, 0xdb, 0xf9, 0xdf, 
+   0x45, 0x0e, 0x0d, 0x80, 0xe8, 0x68, 0x10, 0x4a, 0xc7, 0x75, 0x00, 0xb6, 
+   0x9b, 0xc3, 0x4e, 0x58, 0xfd, 0x91, 0xde, 0x6a, 0x1d, 0x00, 0xa7, 0x7a, 
+   0x28, 0x79, 0x88, 0xe2, 0x34, 0x77, 0xbe, 0xe7, 0xf1, 0x2c, 0x91, 0x5d, 
+   0xd3, 0xf3, 0x80, 0xaa, 0xda, 0x7e, 0x1d, 0x40, 0xd5, 0xd3, 0x50, 0x3d, 
+   0x7a, 0x41, 0x5a, 0x45, 0x55, 0xed, 0xcc, 0x0e, 0x20, 0x3a, 0xaa, 0x22, 
+   0xa2, 0x12, 0xf2, 0x6d, 0xb6, 0x14, 0x6d, 0x84, 0x15, 0x4c, 0x2c, 0x92, 
+   0x60, 0x7f, 0x0a, 0xa2, 0x6d, 0x22, 0x1a, 0x4b, 0x82, 0x0c, 0x38, 0xd0, 
+   0x10, 0x8a, 0x96, 0x2c, 0xb3, 0xdf, 0x66, 0x51, 0xb8, 0x7a, 0xce, 0xd1, 
+   0x65, 0x15, 0xce, 0xb7, 0x7b, 0xec, 0x97, 0xa4, 0xb8, 0x10, 0x01, 0x76, 
+   0x32, 0xcf, 0xcb, 0xea, 0x35, 0xac, 0xd9, 0xa5, 0xa1, 0x12, 0xda, 0xa2, 
+   0x9b, 0x64, 0x03, 0xbc, 0x6e, 0xe9, 0x6b, 0x17, 0x95, 0x4f, 0x5d, 0xb0, 
+   0x82, 0xba, 0x32, 0x4a, 0x56, 0xed, 0x6d, 0x26, 0x41, 0xab, 0x82, 0x54, 
+   0x0c, 0x2c, 0x21, 0x95, 0x1f, 0xb8, 0x0f, 0x88, 0x3a, 0xbb, 0x16, 0x58, 
+   0xe6, 0x4c, 0x4a, 0xe3, 0xae, 0x84, 0x95, 0x26, 0xce, 0x96, 0xbe, 0xbf, 
+   0xb8, 0xa3, 0xe0, 0x15, 0x0e, 0x15, 0x20, 0x44, 0x82, 0x18, 0xab, 0xf8, 
+   0xc5, 0xcd, 0xcd, 0x79, 0xb4, 0x0d, 0x81, 0x62, 0x3a, 0xdb, 0x5b, 0x04, 
+   0xd9, 0xe7, 0x76, 0xb4, 0xcf, 0xbd, 0xe3, 0x9d, 0xf7, 0x67, 0xc9, 0xf6, 
+   0x6e, 0xe6, 0x6d, 0xd3, 0x88, 0x46, 0x6c, 0x41, 0x98, 0x7f, 0xab, 0xec, 
+   0xca, 0xcf, 0xfe, 0xb4, 0xc7, 0xff, 0xc6, 0x1a, 0x4c, 0xe5, 0x6d, 0xf2, 
+   0x81, 0x0f, 0xb5, 0xb8, 0x18, 0x87, 0xdf, 0xdc, 0x28, 0xde, 0x1c, 0xd0, 
+   0xbe, 0xe3, 0x7b, 0x4f, 0x77, 0x68, 0xa0, 0x4b, 0x95, 0x1a, 0x93, 0xdf, 
+   0x89, 0x00, 0xba, 0xee, 0x88, 0xf9, 0x4b, 0x05, 0xf9, 0xd7, 0x80, 0xef, 
+   0x16, 0x40, 0x32, 0xae, 0xeb, 0x0d, 0xde, 0x75, 0x7a, 0xf6, 0x6d, 0xbe, 
+   0x98, 0xcd, 0xbd, 0xf9, 0x83, 0xaf, 0xab, 0xd0, 0x4f, 0x83, 0x13, 0x22, 
+   0x7e, 0x0d, 0xc9, 0xb9, 0xab, 0x2b, 0x8c, 0x7b, 0xd6, 0xf4, 0x36, 0x98, 
+   0xcf, 0x58, 0x7e, 0x78, 0x26, 0x5e, 0x4d, 0x50, 0x55, 0xac, 0xd2, 0xa8, 
+   0xf1, 0x4c, 0x40, 0x0a, 0xad, 0x7b, 0x61, 0xa0, 0x1b, 0x27, 0x92, 0x35, 
+   0xd5, 0x4a, 0x83, 0xe9, 0xfe, 0xac, 0xbd, 0xdf, 0x62, 0x4b, 0x71, 0xd7, 
+   0x81, 0x64, 0x20, 0x58, 0x6f, 0x56, 0xc1, 0x9c, 0xed, 0x6a, 0x9e, 0x0a, 
+   0x2f, 0x4b, 0xc3, 0x79, 0xf6, 0xe7, 0xfa, 0xf5, 0xf1, 0xd6, 0xbe, 0x8c, 
+   0x4c, 0xbc, 0xe0, 0x39, 0x7c, 0xe1, 0x85, 0xa0, 0xec, 0x73, 0x5d, 0xe8, 
+   0x73, 0xfd, 0xa7, 0xec, 0x73, 0x93, 0xed, 0xc6, 0x8f, 0xe1, 0x6f, 0x1e, 
+   0x66, 0x30, 0x54, 0x94, 0x73, 0x18, 0xe0, 0x9b, 0xfb, 0x77, 0xfc, 0xd9, 
+   0x01, 0xad, 0x51, 0x71, 0x8e, 0x9d, 0x14, 0xd8, 0x33, 0xf5, 0x48, 0x56, 
+   0x70, 0x1f, 0x4e, 0xd7, 0xea, 0x9d, 0x01, 0x00, 0xad, 0x89, 0x46, 0x16, 
+   0x7d, 0x4a, 0xe2, 0xe8, 0xd1, 0x38, 0x1c, 0xbf, 0xb9, 0x35, 0x17, 0x97, 
+   0xb6, 0x5e, 0x89, 0x47, 0xba, 0x8d, 0x29, 0x96, 0x37, 0xe3, 0x6d, 0x45, 
+   0x92, 0x62, 0xd3, 0x14, 0xcb, 0x56, 0xef, 0xaa, 0xd5, 0x2e, 0x8d, 0x5f, 
+   0x64, 0xca, 0x87, 0x01, 0x37, 0x05, 0x1a, 0x2e, 0x14, 0x43, 0xd7, 0x70, 
+   0xd9, 0xba, 0x51, 0xe4, 0x3d, 0xee, 0x99, 0x32, 0x15, 0x07, 0x1a, 0x52, 
+   0x55, 0xc8, 0x77, 0x4f, 0x26, 0x16, 0x2f, 0xa9, 0x88, 0x86, 0x81, 0xa5, 
+   0x68, 0xaf, 0x3a, 0x77, 0xa7, 0x6e, 0xee, 0x19, 0x6b, 0xa0, 0xb9, 0xc3, 
+   0xf4, 0xe4, 0xbd, 0x9f, 0x75, 0x58, 0xfc, 0x00, 0x7e, 0x09, 0x40, 0x5f, 
+   0xfc, 0x75, 0x90, 0x26, 0xdc, 0xca, 0x9d, 0x31, 0xd3, 0xfb, 0x22, 0x16, 
+   0x3d, 0xfc, 0x04, 0x86, 0xbc, 0xe8, 0xd2, 0x80, 0x07, 0xbe, 0x1c, 0xce, 
+   0x62, 0x1a, 0x76, 0x69, 0xb5, 0x43, 0xb6, 0x21, 0xbf, 0x08, 0x03, 0xb7, 
+   0x78, 0x02, 0xba, 0x7a, 0xbf, 0x4f, 0x3e, 0xc8, 0xc7, 0x0e, 0x0c, 0x73, 
+   0xda, 0xbf, 0xa1, 0x60, 0x7a, 0xa2, 0xc1, 0x80, 0xd8, 0xfe, 0x79, 0x99, 
+   0x5b, 0x5b, 0x94, 0xf6, 0xef, 0xcc, 0x2c, 0xd9, 0x73, 0x1f, 0xd1, 0x86, 
+   0xb5, 0x65, 0xd9, 0x86, 0x54, 0x3b, 0x52, 0x1b, 0xc8, 0x71, 0x92, 0x68, 
+   0xcb, 0x9b, 0xde, 0xac, 0x95, 0x0b, 0xbc, 0x86, 0x2e, 0x28, 0xe8, 0x5c, 
+   0x36, 0x4f, 0x8a, 0xd8, 0x76, 0x84, 0x03, 0xb7, 0xcf, 0x5c, 0xd4, 0x22, 
+   0x88, 0x7f, 0x88, 0x66, 0x96, 0xca, 0x5d, 0x66, 0xd9, 0xc6, 0x13, 0x38, 
+   0x6a, 0x01, 0x9a, 0x68, 0x96, 0x9e, 0x91, 0xd9, 0x65, 0xc5, 0x02, 0x93, 
+   0xc7, 0x9f, 0xc5, 0x51, 0xce, 0x84, 0x80, 0x7e, 0x7d, 0x0d, 0xd7, 0xdd, 
+   0xd5, 0xd6, 0x2d, 0x7a, 0xd1, 0xe7, 0xce, 0xd6, 0xaa, 0xe8, 0xf9, 0x8f, 
+   0xaa, 0x24, 0x5b, 0x45, 0xa0, 0x36, 0xbb, 0xdf, 0x49, 0x0f, 0xc0, 0xdd, 
+   0xf2, 0x1b, 0xab, 0x01, 0x18, 0x0d, 0xa0, 0x05, 0xec, 0x69, 0xc7, 0x22, 
+   0x29, 0x86, 0x36, 0x29, 0x05, 0x52, 0x17, 0xe2, 0x77, 0x80, 0xc4, 0x80, 
+   0x7a, 0xfc, 0xb0, 0x82, 0x54, 0xf5, 0x66, 0x32, 0x9c, 0x9c, 0x29, 0x97, 
+   0x5e, 0xfc, 0xab, 0x64, 0x07, 0x88, 0xec, 0x08, 0x30, 0x0a, 0xf7, 0xf9, 
+   0xc3, 0x0d, 0xaa, 0x3d, 0x80, 0x5f, 0x76, 0xb0, 0xe2, 0x73, 0xb9, 0x7f, 
+   0xaf, 0xec, 0xf3, 0x6a, 0xe7, 0x05, 0x5d, 0xd3, 0x3f, 0x52, 0xd5, 0xfa, 
+   0x20, 0xae, 0xe3, 0xd3, 0xc5, 0x45, 0x22, 0xdd, 0x16, 0xf0, 0x74, 0x1b, 
+   0xd5, 0x18, 0xb2, 0x42, 0xa2, 0xe0, 0x98, 0xd2, 0xf5, 0x14, 0x6e, 0x51, 
+   0xaa, 0x4e, 0xf0, 0x65, 0x34, 0x66, 0x38, 0x80, 0x6d, 0x87, 0xdf, 0xa5, 
+   0xb6, 0xd1, 0x34, 0xf6, 0x2f, 0x49, 0x34, 0xb5, 0x10, 0xf6, 0x2f, 0x4a, 
+   0x34, 0xb4, 0x10, 0xf6, 0xcf, 0x0f, 0x59, 0x5b, 0x08, 0xbc, 0x75, 0xc3, 
+   0xd8, 0x7b, 0x34, 0x74, 0x2c, 0xc1, 0x2d, 0xa1, 0xe1, 0x4d, 0x82, 0x46, 
+   0x7d, 0xba, 0x33, 0xcd, 0x99, 0x8a, 0xd3, 0x74, 0xfc, 0x7d, 0x08, 0x1d, 
+   0x1e, 0x28, 0xbc, 0xac, 0x85, 0x77, 0xab, 0x99, 0xc5, 0x5a, 0xf8, 0x16, 
+   0x5a, 0x83, 0xa8, 0x25, 0x3c, 0x7b, 0x06, 0x20, 0xf9, 0x06, 0xc2, 0x91, 
+   0x01, 0x62, 0x52, 0x7a, 0x4d, 0xc1, 0x31, 0xad, 0xff, 0xf6, 0x7f, 0x57, 
+   0x40, 0xa1, 0x6e, 0xfc, 0x75, 0x1a, 0x12, 0xd0, 0x3f, 0x0d, 0xf4, 0x4e, 
+   0xbd, 0x71, 0xb5, 0xe9, 0x99, 0xa4, 0xe8, 0x1d, 0xe9, 0xd1, 0x71, 0x4f, 
+   0x8e, 0x65, 0xa3, 0x9e, 0xc9, 0x92, 0xf5, 0x76, 0x04, 0x9d, 0x3a, 0xd1, 
+   0x1d, 0x04, 0xb8, 0xee, 0x5c, 0x6d, 0x0a, 0x22, 0x31, 0x3f, 0x14, 0xad, 
+   0xfd, 0xb6, 0x77, 0x7c, 0x06, 0xaa, 0x76, 0x7e, 0x24, 0x77, 0xcf, 0xd4, 
+   0xfb, 0x49, 0xf8, 0x95, 0x55, 0xfe, 0xcb, 0x98, 0xd7, 0xa1, 0xd5, 0x4d, 
+   0xcb, 0xb5, 0xde, 0x31, 0x89, 0x58, 0x13, 0x70, 0xf5, 0xa6, 0x2b, 0xcf, 
+   0x46, 0xe4, 0x17, 0x8c, 0xf1, 0x5b, 0xde, 0xe3, 0xfd, 0x63, 0x92, 0x3c, 
+   0x7a, 0xc0, 0x13, 0x5b, 0x2e, 0xae, 0xf9, 0x06, 0xa2, 0x8c, 0x21, 0x6f, 
+   0xeb, 0x0b, 0x68, 0x4f, 0xfc, 0xcd, 0x36, 0xe1, 0xb5, 0x64, 0x0a, 0xe4, 
+   0xce, 0x67, 0xa9, 0x49, 0x2e, 0x03, 0xbb, 0xbd, 0xc7, 0x03, 0x1e, 0xc3, 
+   0x08, 0x3c, 0x6a, 0x07, 0x82, 0x1c, 0x0f, 0x04, 0x0e, 0x92, 0xa9, 0xa8, 
+   0x7b, 0x53, 0xdd, 0x51, 0x17, 0xb0, 0x12, 0x7a, 0x81, 0x1c, 0x8c, 0x4b, 
+   0x3d, 0xcc, 0x28, 0x63, 0x2a, 0x96, 0xa8, 0x42, 0x81, 0x68, 0x4e, 0x99, 
+   0x5c, 0x7e, 0xb5, 0x30, 0x08, 0xd5, 0xa1, 0x60, 0x43, 0xa6, 0x56, 0x55, 
+   0xd5, 0xea, 0x70, 0x6f, 0x3a, 0xf7, 0xef, 0xa2, 0x73, 0x5b, 0x7a, 0xc6, 
+   0x92, 0x79, 0xa4, 0x65, 0x97, 0xaf, 0xe6, 0x2d, 0xaf, 0x37, 0x8f, 0xef, 
+   0x75, 0xbc, 0x94, 0x0b, 0x25, 0xbe, 0x78, 0xdb, 0x30, 0x7f, 0xb5, 0x0d, 
+   0x63, 0xbf, 0x51, 0xde, 0x5c, 0x94, 0xbf, 0x96, 0x8b, 0x52, 0x45, 0x72, 
+   0x1e, 0xfe, 0xe9, 0x9d, 0x8a, 0x3f, 0x99, 0x37, 0xf0, 0xa6, 0xdb, 0xfe, 
+   0x3a, 0xba, 0x4d, 0x3f, 0xdb, 0x4b, 0x97, 0xb3, 0x45, 0xec, 0x3d, 0xce, 
+   0x96, 0xec, 0xe4, 0x73, 0xe6, 0x85, 0x8b, 0x59, 0x76, 0xba, 0x2e, 0xfe, 
+   0x3c, 0x2d, 0x4e, 0xe7, 0x4f, 0xb3, 0xd3, 0x79, 0x71, 0x99, 0x8b, 0xb7, 
+   0xfa, 0x1e, 0x70, 0x48, 0x28, 0xcb, 0xc4, 0xd4, 0x5a, 0xbe, 0x97, 0xe4, 
+   0xfb, 0x78, 0xf9, 0xd3, 0x78, 0xfc, 0x74, 0x1f, 0xf3, 0x28, 0xf2, 0x2e, 
+   0xed, 0x81, 0xe1, 0x91, 0x3c, 0x9b, 0xfe, 0xbc, 0x59, 0x86, 0x4f, 0x3e, 
+   0xe8, 0x35, 0xf6, 0xf9, 0x68, 0x20, 0x47, 0xca, 0x3a, 0xeb, 0xa3, 0x8d, 
+   0x26, 0xb3, 0xce, 0xf3, 0xfa, 0x27, 0x41, 0xbd, 0x51, 0x5f, 0xf5, 0xdd, 
+   0xbe, 0xda, 0x97, 0xb4, 0xad, 0x5f, 0x85, 0x83, 0xd4, 0x5c, 0xf9, 0xed, 
+   0x3e, 0xd3, 0xbb, 0x39, 0xa0, 0x28, 0xca, 0xe1, 0xc4, 0xee, 0x19, 0x3f, 
+   0x65, 0x80, 0xe9, 0x45, 0x3f, 0xf3, 0x08, 0xf1, 0x96, 0x1b, 0x50, 0x9a, 
+   0x65, 0x6e, 0x35, 0xaa, 0xaa, 0xd9, 0x4e, 0x55, 0x29, 0xd9, 0x3c, 0x76, 
+   0xd7, 0x40, 0x9a, 0xeb, 0xcd, 0xbe, 0x58, 0x89, 0xfe, 0x30, 0x4f, 0xed, 
+   0xfb, 0x7f, 0x2f, 0x12, 0xe6, 0x16, 0x1e, 0x87, 0x97, 0x04, 0x07, 0x48, 
+   0x5d, 0xbe, 0x09, 0x8e, 0x6e, 0xf1, 0x77, 0x11, 0x9c, 0x76, 0x59, 0x70, 
+   0xce, 0x43, 0x8a, 0x2d, 0x9c, 0xef, 0xa2, 0x10, 0x9b, 0x28, 0xb3, 0x96, 
+   0x45, 0xd7, 0x5c, 0x44, 0x99, 0xd9, 0xe8, 0x57, 0x7b, 0x6d, 0x26, 0x55, 
+   0xa3, 0x5d, 0x43, 0xb8, 0x9c, 0x4e, 0x60, 0xe3, 0x4c, 0xfc, 0x11, 0xc0, 
+   0x7c, 0x38, 0x79, 0xd3, 0x8f, 0xbf, 0x87, 0x7e, 0x6c, 0x26, 0xe6, 0x2f, 
+   0xd2, 0x94, 0x16, 0x36, 0xf6, 0xd5, 0xa4, 0xe8, 0x75, 0x35, 0xae, 0x49, 
+   0x26, 0xdf, 0x54, 0xef, 0x77, 0x51, 0xbd, 0x45, 0xd5, 0x64, 0xe5, 0xe1, 
+   0xde, 0xa4, 0xa6, 0xc0, 0xf4, 0x0f, 0x79, 0xb0, 0xb7, 0x9e, 0x0c, 0x40, 
+   0xe9, 0x51, 0x03, 0x32, 0xf4, 0x40, 0x32, 0xe8, 0xfb, 0xf3, 0x35, 0x08, 
+   0xf2, 0x7b, 0xee, 0xcb, 0x57, 0xd1, 0xee, 0xaf, 0x46, 0x62, 0x76, 0xdf, 
+   0x24, 0x8b, 0x87, 0xbd, 0xc4, 0xcf, 0x28, 0x44, 0xf8, 0xcf, 0xf9, 0x87, 
+   0xd6, 0x55, 0x34, 0xeb, 0x70, 0x16, 0xcd, 0xc9, 0x0d, 0x38, 0x8c, 0x9c, 
+   0x69, 0x56, 0xb0, 0xd6, 0xd6, 0x75, 0x79, 0x9e, 0xe5, 0x9a, 0xf8, 0xe9, 
+   0x76, 0x63, 0xab, 0xc2, 0x9b, 0xb6, 0xad, 0x2f, 0x0d, 0x57, 0x7b, 0xf2, 
+   0x22, 0x6d, 0xda, 0x6b, 0x16, 0xd2, 0x29, 0x14, 0x40, 0x61, 0x82, 0xf2, 
+   0x5c, 0x06, 0x30, 0x67, 0x1a, 0x07, 0x3c, 0xd9, 0xb5, 0x61, 0xe9, 0xab, 
+   0xa5, 0x17, 0x56, 0x35, 0x3c, 0x7e, 0xf0, 0x2e, 0x1f, 0x2a, 0x54, 0xf5, 
+   0xb3, 0xe9, 0x05, 0x38, 0x50, 0x79, 0xf2, 0xaf, 0x14, 0xef, 0x81, 0xa2, 
+   0xaa, 0x99, 0xc3, 0x66, 0x0f, 0x86, 0x22, 0x3a, 0x99, 0x83, 0x16, 0x8f, 
+   0x47, 0xd7, 0x28, 0xe4, 0x5a, 0x8a, 0x76, 0xf5, 0x5a, 0xc6, 0xfb, 0xd5, 
+   0x4c, 0x3c, 0xb1, 0x9c, 0xd3, 0xb9, 0xfc, 0x5c, 0x34, 0x40, 0xa2, 0x52, 
+   0x55, 0x38, 0xaf, 0x6c, 0x29, 0xd5, 0x0a, 0xd4, 0x49, 0x46, 0x88, 0xbe, 
+   0xc8, 0x0c, 0xf8, 0x31, 0x56, 0xd2, 0x72, 0x58, 0xdc, 0xc3, 0xf9, 0x34, 
+   0xb8, 0x24, 0x03, 0x56, 0xfc, 0x9d, 0xf5, 0x86, 0x46, 0x64, 0x5d, 0x2a, 
+   0x35, 0x6a, 0x1f, 0xb3, 0x46, 0xfa, 0x77, 0x5e, 0x4a, 0xc7, 0x27, 0x22, 
+   0x6d, 0x2a, 0x2f, 0x75, 0x94, 0xd3, 0x63, 0xb8, 0x7f, 0x8d, 0x21, 0xbf, 
+   0xcb, 0xae, 0xee, 0x20, 0x99, 0x37, 0x74, 0xf1, 0x2d, 0xe3, 0xe2, 0xeb, 
+   0x43, 0x92, 0xbf, 0xd7, 0xe2, 0x81, 0x5c, 0x57, 0x5e, 0xde, 0x86, 0x08, 
+   0x41, 0xa6, 0x24, 0xeb, 0xba, 0x6f, 0xdb, 0x88, 0x77, 0xae, 0x6b, 0x03, 
+   0xfe, 0x4c, 0x43, 0xe6, 0x3e, 0x91, 0x47, 0x51, 0x1f, 0x9d, 0xe1, 0x54, 
+   0x72, 0xe9, 0x28, 0x05, 0x33, 0x7c, 0xfa, 0xab, 0x1b, 0xbb, 0x1a, 0x86, 
+   0xc3, 0x76, 0x0d, 0x9d, 0x86, 0xd9, 0x43, 0xd6, 0x88, 0xbc, 0xc8, 0xfd, 
+   0x9b, 0xad, 0x2d, 0xcb, 0x9d, 0xeb, 0xf4, 0xa3, 0xdc, 0x65, 0xdd, 0xc8, 
+   0x14, 0xde, 0xca, 0xec, 0x34, 0x5e, 0x16, 0x18, 0xaa, 0x62, 0xa3, 0xcc, 
+   0xc1, 0x9f, 0xd6, 0x9a, 0x6c, 0xfc, 0x39, 0x80, 0x5f, 0x6a, 0x0d, 0xe8, 
+   0x24, 0x71, 0x57, 0xa2, 0xf0, 0xc9, 0x50, 0xd9, 0x19, 0x1a, 0xde, 0x00, 
+   0xc7, 0xed, 0xcd, 0xd5, 0x8b, 0x56, 0xda, 0x90, 0xfb, 0xa5, 0x5b, 0x86, 
+   0xc0, 0x72, 0xd9, 0xb4, 0x79, 0x32, 0x58, 0x6b, 0x25, 0x0f, 0xc0, 0x96, 
+   0xbf, 0x14, 0x95, 0x1c, 0xe1, 0x7d, 0x4e, 0xc4, 0x7b, 0x25, 0x99, 0x8b, 
+   0xf5, 0x89, 0x5b, 0xb8, 0x34, 0xca, 0xce, 0x8a, 0x6b, 0x9c, 0x39, 0x0b, 
+   0x1d, 0x33, 0x8f, 0xc2, 0xc4, 0x9f, 0x6f, 0xb9, 0xf9, 0x56, 0xf1, 0x27, 
+   0x72, 0x02, 0xe2, 0x25, 0xb9, 0xdc, 0x72, 0x03, 0x9a, 0x54, 0x55, 0x8f, 
+   0x2e, 0x9e, 0xa2, 0xa1, 0xfd, 0x2c, 0x0a, 0xb9, 0x83, 0xb1, 0xf5, 0xeb, 
+   0x4a, 0x58, 0xe4, 0x59, 0x8a, 0x28, 0x8d, 0x4e, 0x23, 0xe0, 0x84, 0x08, 
+   0xf4, 0x80, 0x8c, 0x2d, 0xfc, 0xed, 0xe8, 0x5e, 0x2f, 0x9d, 0x6f, 0x74, 
+   0x37, 0xd0, 0xbd, 0xec, 0xb3, 0xa6, 0xb1, 0xef, 0xad, 0x4b, 0x77, 0x65, 
+   0x14, 0x0c, 0x80, 0xb9, 0xdb, 0x89, 0x73, 0xb9, 0x89, 0xc1, 0x7d, 0x2b, 
+   0x8b, 0x1b, 0x77, 0xea, 0xfc, 0xfa, 0x65, 0xdd, 0x03, 0x59, 0x92, 0x27, 
+   0x8a, 0x48, 0x07, 0xcb, 0x7b, 0x68, 0xce, 0xfc, 0xcf, 0xa2, 0x4d, 0x1d, 
+   0x7b, 0x38, 0xae, 0x2a, 0x0d, 0x16, 0xcf, 0x2b, 0x1d, 0xb3, 0x15, 0x8a, 
+   0xe7, 0xfe, 0xaa, 0x8a, 0xd0, 0xc5, 0x53, 0x2b, 0xa7, 0x71, 0x14, 0xa6, 
+   0x66, 0x04, 0x78, 0x1d, 0x59, 0xbe, 0x8e, 0x6b, 0x8f, 0x9a, 0x9f, 0xd1, 
+   0xa6, 0x6a, 0x77, 0xdc, 0xc2, 0x3f, 0x55, 0x5a, 0x75, 0x43, 0x98, 0xb2, 
+   0xb6, 0x52, 0x3e, 0x8b, 0x36, 0xbc, 0xac, 0xa3, 0x77, 0x05, 0x99, 0x6b, 
+   0x89, 0x2c, 0xff, 0x73, 0x3d, 0x3a, 0x79, 0xa4, 0xdd, 0x08, 0x5d, 0xf1, 
+   0x1e, 0x44, 0x05, 0x5f, 0x91, 0x48, 0xaa, 0xbc, 0x1d, 0x81, 0x62, 0x85, 
+   0x45, 0x00, 0xb7, 0xda, 0x39, 0x92, 0xb1, 0xbf, 0x14, 0xfb, 0xdf, 0x4a, 
+   0xab, 0xd4, 0x18, 0xac, 0x9e, 0x8c, 0xca, 0x79, 0x5b, 0x7b, 0x61, 0x66, 
+   0x71, 0x73, 0x0c, 0x2f, 0xa0, 0xe1, 0xdc, 0x6d, 0x7d, 0x6e, 0xb6, 0x79, 
+   0x30, 0x03, 0x97, 0xd5, 0x81, 0x73, 0xf8, 0x05, 0x25, 0xe4, 0x52, 0x7b, 
+   0xcf, 0x5d, 0x29, 0xa7, 0x2a, 0x3e, 0x45, 0x7b, 0x6b, 0x02, 0xac, 0x85, 
+   0x50, 0x50, 0x26, 0xf1, 0xdc, 0x88, 0x52, 0xf2, 0xbb, 0x11, 0x4a, 0x35, 
+   0x81, 0xd2, 0x06, 0x74, 0xa5, 0xf2, 0x9c, 0x00, 0x78, 0x76, 0xc0, 0x3a, 
+   0x40, 0x59, 0x2c, 0xb8, 0x67, 0xfd, 0x71, 0x2a, 0xc6, 0xba, 0xf5, 0xb6, 
+   0x77, 0xc2, 0x08, 0x2f, 0xf7, 0xc0, 0xb4, 0xda, 0xc6, 0xe2, 0x51, 0x2f, 
+   0x96, 0x7a, 0xdb, 0x37, 0x55, 0x48, 0xa1, 0x4a, 0xbb, 0x9a, 0xd7, 0x69, 
+   0xde, 0xa4, 0xec, 0x3f, 0x45, 0xca, 0xf4, 0xab, 0x48, 0x1c, 0xfd, 0x6c, 
+   0x3e, 0x67, 0xa4, 0x2d, 0x89, 0x5f, 0xc4, 0xb3, 0x8f, 0xd9, 0x3b, 0x9c, 
+   0x60, 0x3e, 0x8f, 0x7f, 0xe0, 0xfb, 0x53, 0xa7, 0xbb, 0xe3, 0x93, 0x98, 
+   0xc7, 0xcc, 0x61, 0x34, 0x8d, 0xb7, 0xf0, 0x0f, 0xce, 0x62, 0xdf, 0x0f, 
+   0x8d, 0x08, 0x2c, 0x2e, 0x42, 0x8f, 0xa9, 0xf7, 0x6a, 0x1a, 0xde, 0x50, 
+   0xb7, 0xeb, 0x97, 0xb1, 0x38, 0x3d, 0x67, 0x8f, 0xac, 0x91, 0x5d, 0x71, 
+   0xc9, 0x4b, 0x88, 0xaf, 0x74, 0x89, 0x40, 0x6f, 0x4d, 0xdd, 0xc2, 0x3b, 
+   0x27, 0x6b, 0x7b, 0x45, 0x76, 0xff, 0xd6, 0x8f, 0x23, 0xde, 0xfa, 0x65, 
+   0xe0, 0x87, 0x69, 0x1c, 0x05, 0x0b, 0xd1, 0xcf, 0x17, 0xbf, 0x5c, 0xd6, 
+   0x3b, 0x2a, 0x3e, 0xbb, 0x66, 0x24, 0xb4, 0x49, 0x35, 0x70, 0x9e, 0x91, 
+   0xf9, 0x96, 0xbd, 0x9e, 0x48, 0x3e, 0x1d, 0xfc, 0x2f, 0x5e, 0x01, 0x5c, 
+   0xae, 0x05, 0xc6, 0x25, 0x45, 0xf4, 0x1f, 0xe6, 0xd8, 0x4a, 0x13, 0xd0, 
+   0x58, 0x81, 0x35, 0x0e, 0xa2, 0x4a, 0x50, 0x38, 0xd9, 0xec, 0xdd, 0xca, 
+   0xd2, 0x6b, 0x8d, 0xc0, 0x51, 0x18, 0xd8, 0xdd, 0xb8, 0xf1, 0xc4, 0x34, 
+   0x86, 0xb6, 0x9b, 0xb7, 0xdd, 0x78, 0xde, 0x1c, 0x50, 0xbc, 0x10, 0x46, 
+   0x29, 0xf7, 0xd2, 0x7c, 0xd6, 0x51, 0x91, 0xc9, 0xcc, 0xae, 0x72, 0x8b, 
+   0x64, 0x2c, 0xca, 0xce, 0xec, 0x01, 0xe9, 0xda, 0xfd, 0xb5, 0x74, 0x7a, 
+   0x87, 0xe0, 0x2a, 0x60, 0x83, 0xd4, 0xf0, 0xc3, 0xf5, 0x2e, 0x0e, 0xd1, 
+   0x7c, 0xbe, 0xda, 0x26, 0x94, 0x6c, 0xb3, 0xff, 0xd9, 0xfa, 0xf1, 0xb3, 
+   0x28, 0x4b, 0xf3, 0xe3, 0x99, 0x9f, 0xe5, 0x4b, 0xb2, 0x28, 0x2e, 0x87, 
+   0x22, 0xff, 0x66, 0x50, 0xa2, 0x24, 0xcd, 0x07, 0xb5, 0xd8, 0x2b, 0x76, 
+   0x24, 0x39, 0x2a, 0xce, 0x16, 0xa8, 0xed, 0x26, 0xc7, 0xec, 0x81, 0x42, 
+   0x28, 0x14, 0xe0, 0xcf, 0x7b, 0xb7, 0x4e, 0xc8, 0x75, 0x1c, 0x7d, 0x8b, 
+   0x7e, 0x65, 0x5c, 0xd6, 0x0f, 0x69, 0x94, 0x10, 0x05, 0xbf, 0xee, 0x5b, 
+   0x60, 0xfb, 0x27, 0x3f, 0x1f, 0xd8, 0x5b, 0x06, 0xab, 0xd5, 0x3f, 0x0d, 
+   0xd7, 0xa5, 0x2d, 0x0a, 0x65, 0x85, 0xee, 0xf7, 0x0d, 0xc9, 0x27, 0x8b, 
+   0x8b, 0x3c, 0xb9, 0xc7, 0x6e, 0x44, 0x81, 0x47, 0x5d, 0xfc, 0x39, 0x74, 
+   0x72, 0x4b, 0xb6, 0xfc, 0x39, 0xf4, 0xc4, 0x8c, 0xca, 0xe2, 0x2e, 0x6f, 
+   0x86, 0xca, 0x88, 0x02, 0xbf, 0xde, 0x68, 0xf1, 0x1d, 0x78, 0x55, 0x0d, 
+   0x0f, 0xff, 0xf6, 0xb0, 0x00, 0xab, 0xb0, 0x78, 0x56, 0x01, 0x56, 0xc1, 
+   0xf9, 0xcf, 0xf4, 0x07, 0xde, 0xc4, 0x89, 0xa4, 0xf9, 0x59, 0x8e, 0xe9, 
+   0xbe, 0x3c, 0x2e, 0x4f, 0xa2, 0xf5, 0x39, 0x1d, 0x17, 0x93, 0x11, 0xab, 
+   0x95, 0x96, 0xef, 0x67, 0xd7, 0xa8, 0x53, 0xf3, 0x61, 0x77, 0x10, 0xfa, 
+   0x24, 0x47, 0xe9, 0xa9, 0xcb, 0x06, 0x94, 0x23, 0x56, 0x70, 0x21, 0x62, 
+   0x68, 0xbd, 0x04, 0xa4, 0x49, 0x2b, 0x83, 0x01, 0xbb, 0x27, 0xf2, 0x18, 
+   0x24, 0x4a, 0xe9, 0x06, 0x50, 0xb2, 0x81, 0xf7, 0x2f, 0xc8, 0x3e, 0x27, 
+   0xab, 0x25, 0x67, 0x2e, 0x1c, 0xdb, 0xc3, 0xca, 0x53, 0x52, 0x15, 0xb2, 
+   0xdb, 0xa2, 0x3c, 0x8d, 0xe2, 0x47, 0x2f, 0x5e, 0x58, 0xa0, 0x6c, 0x76, 
+   0x77, 0xff, 0x48, 0xf7, 0x4e, 0x58, 0xa7, 0xfd, 0x99, 0xb7, 0x5a, 0xcd, 
+   0x96, 0x2b, 0x2f, 0xe5, 0x87, 0x75, 0xec, 0xa7, 0xfb, 0xa4, 0xe2, 0x35, 
+   0x50, 0x33, 0x95, 0x65, 0xf9, 0x32, 0xeb, 0x0f, 0xb9, 0x0a, 0x47, 0x7a, 
+   0x1f, 0x26, 0x03, 0xce, 0x0a, 0x2a, 0x28, 0x5f, 0x53, 0x82, 0xf9, 0x0e, 
+   0xde, 0x43, 0x65, 0x8a, 0xdf, 0xcf, 0x7f, 0x80, 0xa6, 0xfe, 0x4b, 0x79, 
+   0x10, 0x47, 0xb0, 0x10, 0x85, 0xd1, 0x86, 0xee, 0x83, 0x0d, 0xd5, 0x45, 
+   0x54, 0x52, 0x0d, 0x82, 0x74, 0x15, 0x85, 0x7b, 0x0a, 0x94, 0xad, 0x17, 
+   0xea, 0x1c, 0xe8, 0x37, 0x6e, 0x6a, 0xa6, 0xb2, 0x98, 0x01, 0x4a, 0x38, 
+   0x43, 0xa3, 0xbe, 0x83, 0xd8, 0xc1, 0xf3, 0xfc, 0x7e, 0xb2, 0x67, 0x9c, 
+   0xff, 0xaf, 0x24, 0x80, 0xbc, 0xf4, 0x34, 0xab, 0xdd, 0xe0, 0x0f, 0xd7, 
+   0x33, 0x74, 0x10, 0x57, 0xe5, 0xb3, 0xf6, 0x75, 0x5e, 0xdd, 0x72, 0x15, 
+   0x79, 0x36, 0x93, 0x16, 0x47, 0x5c, 0xdc, 0x98, 0x3d, 0x06, 0x8b, 0x14, 
+   0xec, 0xb5, 0xa7, 0xfc, 0xf5, 0x05, 0x93, 0xaa, 0xd9, 0xe2, 0xdc, 0xd2, 
+   0xe5, 0x1f, 0xd2, 0xc5, 0x1b, 0xa9, 0x28, 0xe3, 0xd8, 0xeb, 0xcf, 0x4a, 
+   0x49, 0x62, 0x23, 0xbf, 0x5d, 0xc3, 0x73, 0xaa, 0x65, 0x32, 0xa0, 0x08, 
+   0x59, 0xe2, 0xd9, 0x69, 0x4d, 0xf5, 0x5b, 0xf4, 0x82, 0xd9, 0x40, 0x80, 
+   0xa4, 0x32, 0x3f, 0xa0, 0xd0, 0xf4, 0x2f, 0x41, 0x0f, 0xad, 0x7b, 0x50, 
+   0xb0, 0xc9, 0xc5, 0x8b, 0xfd, 0x3c, 0x7b, 0x0c, 0xc2, 0x45, 0xf4, 0x28, 
+   0xe9, 0x73, 0x08, 0x37, 0x0b, 0x0a, 0x36, 0xd4, 0xeb, 0x63, 0x90, 0x44, 
+   0xde, 0xb1, 0xdb, 0x04, 0x4f, 0xfe, 0x0a, 0x5a, 0xa0, 0x45, 0x26, 0x46, 
+   0xc5, 0xf8, 0x99, 0xc9, 0x70, 0x2d, 0x42, 0x8b, 0xd4, 0x8c, 0x8a, 0xf0, 
+   0x38, 0x4a, 0xd3, 0x68, 0x4d, 0xae, 0x19, 0x3a, 0xa5, 0x7f, 0xdb, 0x6e, 
+   0x29, 0x1b, 0x15, 0xf1, 0x85, 0xbf, 0x4c, 0x6d, 0xd0, 0xda, 0xaa, 0x99, 
+   0x4c, 0x6f, 0x05, 0xfe, 0x23, 0xeb, 0x49, 0x55, 0x91, 0xd5, 0x1e, 0xba, 
+   0xff, 0x8a, 0xa1, 0x3c, 0x57, 0xe0, 0x93, 0xaf, 0x7b, 0xf3, 0xca, 0x77, 
+   0x81, 0x22, 0x9c, 0xb8, 0xed, 0x9f, 0x7a, 0xb0, 0x16, 0x2d, 0xa1, 0xfc, 
+   0x52, 0x8f, 0xf2, 0x00, 0xc0, 0xb8, 0xeb, 0x5a, 0xb1, 0xcc, 0x80, 0xc7, 
+   0xbc, 0x5f, 0x79, 0xd9, 0x50, 0x96, 0xa0, 0x74, 0x70, 0x97, 0xde, 0x86, 
+   0x52, 0x2e, 0x48, 0x29, 0xc7, 0x44, 0xa9, 0xec, 0x83, 0xbe, 0x00, 0x1f, 
+   0xe4, 0xe2, 0x02, 0x6b, 0x43, 0x67, 0x17, 0xa2, 0xb3, 0x53, 0x4f, 0xe7, 
+   0xd2, 0x66, 0xff, 0xed, 0x9d, 0x22, 0xc7, 0xd4, 0x22, 0xd3, 0xd8, 0x94, 
+   0xd2, 0x79, 0x9f, 0xca, 0x15, 0x61, 0x7f, 0xe6, 0x6f, 0xaa, 0x6c, 0x56, 
+   0x1e, 0xaf, 0xde, 0xcb, 0x79, 0x71, 0x04, 0x27, 0x0a, 0xc5, 0x3d, 0x8d, 
+   0xdb, 0xc7, 0x1d, 0x75, 0x63, 0x7e, 0x91, 0x0e, 0x44, 0x60, 0x6d, 0xaf, 
+   0x4a, 0xcb, 0xa3, 0x2b, 0x9a, 0x3d, 0x3d, 0x97, 0xd7, 0xf8, 0xe5, 0x2b, 
+   0x99, 0xcc, 0xbd, 0x95, 0x68, 0x61, 0x9c, 0xf5, 0x41, 0x3e, 0xa8, 0x29, 
+   0x7e, 0x2d, 0x78, 0xf1, 0xd1, 0x5b, 0x2d, 0xf7, 0xa4, 0x86, 0x73, 0xf6, 
+   0xdd, 0x4e, 0x97, 0xea, 0x26, 0xea, 0xb7, 0x70, 0xdd, 0xb4, 0xe3, 0xca, 
+   0x75, 0xf4, 0x85, 0xba, 0xb3, 0xc0, 0xfe, 0x12, 0xb2, 0x00, 0x9c, 0x67, 
+   0x44, 0xf1, 0xf9, 0x95, 0xc9, 0xd2, 0xf6, 0x63, 0x71, 0x3a, 0x4e, 0x9d, 
+   0xdb, 0xdc, 0x22, 0xbd, 0xbf, 0x9d, 0xb3, 0x4a, 0x85, 0xdb, 0xe4, 0xc3, 
+   0x4b, 0x29, 0x92, 0x7f, 0x53, 0x2d, 0xc6, 0x86, 0xce, 0x8c, 0x2c, 0xea, 
+   0xb8, 0xf0, 0x9e, 0x45, 0x4b, 0xbc, 0x6c, 0x95, 0x47, 0x0d, 0xca, 0x28, 
+   0xf3, 0xb1, 0xf5, 0x47, 0xb0, 0x70, 0x89, 0x47, 0x9d, 0x7b, 0xe7, 0x16, 
+   0x35, 0x0d, 0x6a, 0xd7, 0x73, 0x9e, 0xf0, 0x60, 0x27, 0x70, 0xe4, 0xfd, 
+   0x35, 0x9d, 0xc6, 0xf9, 0x80, 0xbb, 0x5e, 0xd1, 0x96, 0x46, 0xf0, 0x7b, 
+   0x59, 0x07, 0x1b, 0x2a, 0x4e, 0x21, 0x75, 0xb9, 0xd7, 0xa6, 0x44, 0x4c, 
+   0xd7, 0xe2, 0x56, 0xa9, 0xd2, 0xfd, 0x90, 0x7f, 0xcd, 0xfb, 0xf6, 0xd3, 
+   0x07, 0x00, 0x53, 0x25, 0x5f, 0x61, 0xd6, 0x15, 0xc7, 0xd7, 0xd7, 0xbc, 
+   0x48, 0x96, 0x06, 0x64, 0xb1, 0x7a, 0xa1, 0x19, 0x8a, 0x36, 0x0a, 0xba, 
+   0x6a, 0xef, 0x77, 0x81, 0xe8, 0x25, 0x93, 0xe0, 0x8e, 0x83, 0xd2, 0xdf, 
+   0xef, 0xe0, 0xf6, 0x3d, 0xdb, 0xed, 0xef, 0x29, 0x05, 0x79, 0xa7, 0x55, 
+   0xb9, 0x64, 0x25, 0x16, 0xc1, 0xed, 0x8b, 0xd8, 0xd5, 0xb5, 0x58, 0x6c, 
+   0x4a, 0xe8, 0xb9, 0x4c, 0xf0, 0x4c, 0xb9, 0x28, 0x8e, 0x12, 0xd7, 0x24, 
+   0x78, 0x61, 0x7d, 0xb9, 0xb5, 0x89, 0x74, 0x37, 0x8a, 0x45, 0x56, 0x1a, 
+   0x36, 0x68, 0xa5, 0xe5, 0x0c, 0x35, 0x65, 0x45, 0x34, 0xe7, 0x79, 0xb1, 
+   0xec, 0x51, 0xc0, 0x44, 0x09, 0x5a, 0x4c, 0x29, 0x32, 0xb1, 0x5c, 0xa2, 
+   0xd7, 0x72, 0x9b, 0xd1, 0x91, 0x6e, 0xfb, 0x0e, 0xad, 0x1a, 0xaf, 0x19, 
+   0xee, 0xb8, 0x3d, 0xe0, 0xb5, 0x28, 0xfb, 0xf1, 0x54, 0xab, 0xde, 0xd5, 
+   0xd6, 0x96, 0x23, 0x84, 0xa1, 0x76, 0x2a, 0x0d, 0xa8, 0xfc, 0xd5, 0x11, 
+   0xc8, 0x6d, 0x4e, 0xa0, 0x2a, 0xda, 0x9d, 0x08, 0x05, 0xa0, 0xd9, 0x89, 
+   0x60, 0x00, 0x9e, 0xa6, 0x84, 0xe3, 0x5d, 0xe8, 0x33, 0x3c, 0x6c, 0x3c, 
+   0x9f, 0x22, 0xe1, 0xc2, 0xbb, 0x60, 0xc2, 0xcb, 0x4b, 0xe2, 0xb1, 0xec, 
+   0x24, 0x59, 0xf8, 0x4b, 0x8f, 0xaa, 0x1e, 0xe5, 0xcb, 0xb5, 0x49, 0x58, 
+   0x33, 0xf5, 0x00, 0xe8, 0xca, 0x55, 0xd2, 0xc3, 0x05, 0x96, 0x83, 0x17, 
+   0xab, 0xe0, 0xb2, 0x91, 0xa8, 0xd5, 0x19, 0x22, 0x66, 0x01, 0x95, 0xc6, 
+   0x0e, 0xc5, 0x80, 0x52, 0x8b, 0x98, 0xf1, 0x96, 0xd4, 0x08, 0x84, 0xf7, 
+   0xa5, 0xd6, 0xc2, 0xe2, 0x80, 0xc1, 0xd2, 0x5a, 0xc8, 0x82, 0xc3, 0x57, 
+   0xb5, 0x16, 0x8e, 0xc5, 0xfe, 0xd3, 0x5b, 0x68, 0xac, 0xb3, 0x49, 0x5a, 
+   0xee, 0xdd, 0x66, 0xd3, 0xb0, 0xbd, 0x47, 0x3e, 0x96, 0x6e, 0x57, 0x65, 
+   0xac, 0x5d, 0xab, 0x8f, 0x7c, 0x2c, 0xdb, 0xa3, 0xca, 0x60, 0x9b, 0x3b, 
+   0x88, 0xf6, 0x66, 0x0e, 0x97, 0xb4, 0x37, 0x73, 0xf0, 0x66, 0x0e, 0x5e, 
+   0xdd, 0x1c, 0xbc, 0xd8, 0xcd, 0x75, 0x3f, 0xd8, 0x36, 0x1d, 0xe0, 0x48, 
+   0xfa, 0xf2, 0x73, 0xb3, 0xa6, 0x40, 0x2f, 0xbd, 0xe9, 0x62, 0x40, 0x8b, 
+   0x65, 0x72, 0xcd, 0xfe, 0x5b, 0x53, 0x6b, 0xe5, 0x58, 0x58, 0x2b, 0x47, 
+   0xa3, 0xe4, 0xb8, 0x78, 0x4a, 0x47, 0xd2, 0x72, 0x10, 0xad, 0xd7, 0x51, 
+   0xd8, 0xb8, 0x0f, 0xbe, 0x5e, 0xcc, 0xc7, 0x3b, 0x29, 0x02, 0x01, 0x8f, 
+   0x85, 0x0f, 0x5d, 0xb4, 0xc0, 0xd6, 0xba, 0x31, 0x2a, 0x0d, 0x78, 0x71, 
+   0x7b, 0x52, 0x6a, 0x82, 0x6d, 0x42, 0x64, 0x61, 0x4e, 0xf8, 0x11, 0xfc, 
+   0xde, 0x2d, 0xc9, 0xde, 0x3a, 0xaa, 0x22, 0x69, 0x57, 0xce, 0xe0, 0xcd, 
+   0x48, 0xf8, 0x09, 0x3a, 0x0b, 0x98, 0x6f, 0xf8, 0x99, 0x3c, 0x11, 0x67, 
+   0xe0, 0xca, 0xb1, 0xb2, 0x82, 0xb6, 0x83, 0x2b, 0x01, 0xf9, 0x6d, 0x19, 
+   0x1a, 0x72, 0x31, 0xdd, 0xbf, 0x18, 0xa2, 0xbb, 0xbe, 0x3a, 0xee, 0x6c, 
+   0xba, 0x7f, 0x36, 0xac, 0xee, 0xf2, 0x1a, 0x73, 0xd2, 0xcc, 0x05, 0x68, 
+   0xbf, 0x5a, 0xc0, 0xd8, 0xfe, 0x3e, 0x01, 0x63, 0xe9, 0x7e, 0xf6, 0xdf, 
+   0xd3, 0x05, 0x10, 0xcd, 0xdc, 0x72, 0x77, 0xb3, 0xec, 0x13, 0xca, 0x08, 
+   0xd0, 0xa2, 0xe4, 0x56, 0xa0, 0xca, 0x1c, 0x4c, 0x33, 0x26, 0x3c, 0x75, 
+   0x6f, 0xef, 0x96, 0xb4, 0x77, 0x70, 0x4b, 0x74, 0x05, 0x6a, 0x3a, 0xb9, 
+   0x6b, 0xe6, 0xee, 0xef, 0xa0, 0x40, 0xeb, 0xb3, 0x44, 0xd7, 0x4a, 0x87, 
+   0x7e, 0x29, 0x91, 0x1d, 0x8b, 0xf8, 0x5e, 0xd4, 0x07, 0xf1, 0xe3, 0x9d, 
+   0xe2, 0xe9, 0x85, 0x81, 0x78, 0xd1, 0xa1, 0xa2, 0xeb, 0xf0, 0x2a, 0x0e, 
+   0xf0, 0x11, 0xf1, 0x0a, 0x17, 0x8a, 0xaa, 0x25, 0x08, 0xfc, 0x4d, 0x85, 
+   0xbe, 0xa9, 0x50, 0x9b, 0x9c, 0xdb, 0x9b, 0x26, 0xfa, 0xbd, 0x34, 0xd1, 
+   0x2a, 0x8a, 0xad, 0x3d, 0xb9, 0x86, 0xfe, 0x6f, 0xad, 0x26, 0x53, 0x6e, 
+   0x5d, 0x58, 0xb4, 0x2d, 0x31, 0x35, 0xfe, 0x74, 0x5f, 0xde, 0xf9, 0x53, 
+   0x8a, 0x50, 0xa5, 0xf5, 0x27, 0xe4, 0xcb, 0xd6, 0x74, 0x35, 0x97, 0xc7, 
+   0x70, 0x68, 0x5b, 0xf3, 0x91, 0xda, 0xe7, 0x79, 0x69, 0x6a, 0xb6, 0x5a, 
+   0xa8, 0xbd, 0x57, 0x6b, 0xb6, 0x5a, 0xac, 0xd5, 0xa2, 0xd9, 0x6a, 0xb1, 
+   0x4b, 0x61, 0x45, 0xef, 0x54, 0xb7, 0x73, 0xae, 0xe2, 0xcd, 0xa8, 0xd8, 
+   0x86, 0xaf, 0x7c, 0x55, 0xcf, 0xa4, 0x17, 0x2a, 0x78, 0xf4, 0x10, 0x42, 
+   0xbf, 0xe1, 0xdf, 0xa4, 0xc9, 0x42, 0xd3, 0xcd, 0xe6, 0x7e, 0x97, 0xcd, 
+   0x86, 0xa4, 0xfd, 0x94, 0xcd, 0xe2, 0x2a, 0xf7, 0x4a, 0x2b, 0xab, 0xc3, 
+   0xc8, 0xd4, 0xa2, 0x26, 0x9f, 0xea, 0xfc, 0x2a, 0xdd, 0xe5, 0xa6, 0xd5, 
+   0x07, 0x08, 0x70, 0x9c, 0xfe, 0x2d, 0x82, 0x88, 0x48, 0xab, 0x81, 0x88, 
+   0x18, 0x51, 0x83, 0x22, 0xe3, 0x1a, 0x65, 0x06, 0xc3, 0xec, 0x9a, 0xa8, 
+   0xe1, 0x76, 0x5f, 0x46, 0x0d, 0x17, 0xa3, 0x86, 0x54, 0x0f, 0x4d, 0xa9, 
+   0xe1, 0x62, 0xd4, 0x68, 0x60, 0x59, 0x75, 0xcc, 0x8e, 0x89, 0x1a, 0xb2, 
+   0x8e, 0x7d, 0x37, 0x6a, 0x38, 0x18, 0x35, 0xa4, 0x2a, 0x6a, 0x4a, 0x0d, 
+   0x07, 0xa1, 0x86, 0x53, 0xa5, 0xb3, 0x25, 0x35, 0x0e, 0x8c, 0xd4, 0x38, 
+   0x78, 0x19, 0x35, 0x0e, 0x30, 0x6a, 0x74, 0x77, 0x25, 0xc6, 0x01, 0x42, 
+   0x8c, 0x2a, 0x95, 0x8d, 0xb4, 0x68, 0xaa, 0x14, 0x5f, 0x27, 0x99, 0x54, 
+   0xbc, 0x09, 0xb7, 0xab, 0x17, 0xa2, 0x9c, 0x4c, 0xb6, 0x7a, 0xb6, 0x5e, 
+   0xc8, 0x0b, 0x5e, 0x5d, 0x70, 0x3b, 0xb8, 0x02, 0xc6, 0x5c, 0x0e, 0xb7, 
+   0x81, 0xcb, 0xd1, 0xd0, 0x55, 0xa0, 0xf4, 0x24, 0xaf, 0xef, 0x2e, 0xd4, 
+   0xb5, 0x66, 0x6c, 0xea, 0x25, 0x60, 0xe1, 0xa0, 0x2a, 0xa9, 0xe4, 0x7d, 
+   0x70, 0x1f, 0xb2, 0x77, 0x72, 0x3e, 0x54, 0xbf, 0x0e, 0xf6, 0xd8, 0x59, 
+   0x6b, 0xff, 0xc4, 0x5f, 0xd1, 0xa0, 0x2b, 0x21, 0x77, 0x7e, 0xfa, 0xc8, 
+   0xae, 0x50, 0xca, 0x77, 0x07, 0xe7, 0xbc, 0x59, 0x02, 0xc9, 0xa4, 0x94, 
+   0x85, 0x89, 0x13, 0x7f, 0xe3, 0xc5, 0x8c, 0x04, 0x39, 0x88, 0xf8, 0xdb, 
+   0x4f, 0x70, 0xcb, 0x7f, 0x89, 0xe9, 0xfc, 0x04, 0xf5, 0xfb, 0xcb, 0xbe, 
+   0xfa, 0x9f, 0x63, 0xbf, 0x4d, 0x44, 0x06, 0x20, 0xeb, 0x37, 0x62, 0xbd, 
+   0xeb, 0xc4, 0xbb, 0x9f, 0x40, 0xe0, 0x64, 0x7a, 0x63, 0x41, 0x23, 0x9b, 
+   0x3a, 0x27, 0x10, 0x3e, 0x28, 0x35, 0xaa, 0x79, 0xcd, 0x4d, 0x15, 0x89, 
+   0x96, 0xc0, 0x30, 0xb5, 0x4b, 0xc0, 0xc3, 0xb8, 0xa6, 0x64, 0x6f, 0x7d, 
+   0x67, 0xb2, 0xd7, 0xe7, 0x7e, 0xfe, 0x63, 0xc9, 0xfe, 0x4a, 0x2e, 0x37, 
+   0xff, 0x68, 0xee, 0x78, 0x27, 0x24, 0x2f, 0x75, 0x32, 0x0b, 0x3c, 0x53, 
+   0x1e, 0xc3, 0xa7, 0x0d, 0x6f, 0xc7, 0x95, 0x3e, 0x04, 0x09, 0xeb, 0xc1, 
+   0xe3, 0x91, 0x8e, 0xe8, 0x33, 0xc5, 0x57, 0xf3, 0x7f, 0x60, 0xbd, 0xa0, 
+   0xcc, 0x43, 0x42, 0xff, 0x89, 0xba, 0x6c, 0x6d, 0xf6, 0x5e, 0x2e, 0xf4, 
+   0x32, 0x90, 0x45, 0xa7, 0x4f, 0x15, 0xdb, 0x2a, 0x7a, 0x24, 0x2d, 0x57, 
+   0x43, 0xd6, 0xac, 0xe3, 0xa7, 0xee, 0x2f, 0x84, 0xbf, 0x5b, 0xcc, 0xd4, 
+   0xfe, 0xfe, 0x0c, 0xac, 0x6e, 0x9d, 0x37, 0x06, 0xfe, 0xd9, 0x14, 0x5f, 
+   0x85, 0x81, 0xee, 0xf7, 0xdf, 0x81, 0xeb, 0x60, 0xb1, 0xc7, 0x6e, 0x1e, 
+   0xbd, 0x1e, 0x13, 0x19, 0xc6, 0xbf, 0x2c, 0x23, 0x3b, 0xdf, 0x9f, 0x91, 
+   0xdf, 0x69, 0x27, 0xbe, 0x31, 0xf2, 0xf7, 0x50, 0xa9, 0x2d, 0x6b, 0x27, 
+   0x70, 0xa3, 0x9c, 0x44, 0x29, 0x47, 0x47, 0x2d, 0x74, 0xa5, 0x2c, 0xa2, 
+   0xda, 0x68, 0x61, 0x17, 0x16, 0x66, 0x19, 0xf8, 0xca, 0xc4, 0x4a, 0x84, 
+   0x42, 0x39, 0x1e, 0x71, 0x9f, 0xf1, 0x2b, 0x7f, 0x1e, 0x8c, 0xdf, 0x2f, 
+   0x00, 0xa5, 0x4a, 0x5c, 0xd2, 0x26, 0x63, 0xca, 0x7a, 0xf1, 0x4c, 0x17, 
+   0xeb, 0x78, 0xac, 0x9d, 0x32, 0xd0, 0x0f, 0xba, 0x36, 0xdd, 0x08, 0xb2, 
+   0xe9, 0x31, 0xa4, 0x08, 0x19, 0x17, 0x59, 0xa7, 0xab, 0x49, 0x58, 0xdd, 
+   0x51, 0xc3, 0x9f, 0x47, 0xc2, 0xba, 0xdf, 0x5f, 0xc2, 0xea, 0xfd, 0xdd, 
+   0x37, 0x09, 0xfb, 0x9b, 0x4b, 0x98, 0x8d, 0x31, 0x62, 0x58, 0x94, 0xb7, 
+   0xba, 0xa8, 0x98, 0xcd, 0xf8, 0x6d, 0xa5, 0x44, 0x93, 0x3c, 0x79, 0x27, 
+   0x0e, 0xbc, 0x6d, 0xcf, 0x4b, 0xce, 0x62, 0xca, 0x0d, 0x7d, 0x6d, 0x0e, 
+   0x94, 0x28, 0x34, 0xae, 0x8e, 0x23, 0x99, 0x47, 0xab, 0xed, 0x3a, 0x34, 
+   0xe3, 0x69, 0x78, 0xcd, 0x97, 0x65, 0x7f, 0x94, 0xf3, 0xd8, 0x39, 0x7b, 
+   0xfd, 0x59, 0xee, 0xa2, 0xc9, 0x76, 0xe3, 0xc7, 0x7c, 0x5a, 0x5e, 0x1b, 
+   0x2c, 0x76, 0x4d, 0xed, 0x69, 0xe4, 0x31, 0x95, 0x2b, 0x4e, 0x10, 0xd6, 
+   0xef, 0x5c, 0x5a, 0x34, 0x99, 0x5b, 0xd4, 0x9b, 0xfb, 0x5a, 0x55, 0x06, 
+   0xcb, 0xaf, 0xc8, 0xc4, 0x7a, 0x14, 0x2f, 0x94, 0xb5, 0x4b, 0xdc, 0x78, 
+   0xcb, 0x15, 0xb9, 0x48, 0x53, 0xcf, 0xdf, 0x8e, 0xc5, 0xf1, 0xfe, 0x34, 
+   0x4a, 0xbd, 0x95, 0x78, 0x77, 0x5e, 0xd9, 0x4e, 0x6c, 0xd9, 0x09, 0xc0, 
+   0x11, 0x8b, 0x1b, 0x14, 0x2a, 0xc2, 0xe2, 0x8e, 0x94, 0x09, 0x9f, 0xd5, 
+   0x5d, 0x0a, 0xf0, 0x0b, 0x0b, 0x42, 0x02, 0xa5, 0x6e, 0x16, 0x7b, 0x1e, 
+   0xfc, 0xca, 0x3a, 0x9c, 0xd5, 0x2b, 0xa6, 0x75, 0xcc, 0x35, 0x50, 0xd2, 
+   0x14, 0xd7, 0x37, 0x14, 0x1d, 0x98, 0xac, 0xaf, 0x78, 0xb7, 0xa9, 0xa5, 
+   0xe9, 0x89, 0x75, 0x2e, 0x68, 0xb3, 0xa2, 0x01, 0xfb, 0x8c, 0xeb, 0x8e, 
+   0x15, 0xdd, 0x17, 0xb3, 0x3b, 0x2f, 0xf1, 0xeb, 0xf6, 0x5d, 0xb1, 0x83, 
+   0xd8, 0xa3, 0x0f, 0xfc, 0x0a, 0x4c, 0xcd, 0xe6, 0x2b, 0x37, 0xf4, 0xaf, 
+   0xbf, 0x1c, 0x53, 0xe5, 0xaa, 0x36, 0x97, 0xe9, 0x18, 0x13, 0x55, 0x2b, 
+   0xcc, 0x52, 0xd1, 0x69, 0x84, 0x37, 0x9d, 0x64, 0x56, 0xa9, 0x4c, 0xa3, 
+   0x8e, 0x46, 0xa3, 0x60, 0xbd, 0x59, 0x05, 0xf3, 0x20, 0x9d, 0x59, 0x28, 
+   0x55, 0x6a, 0xbc, 0x04, 0x70, 0x26, 0x45, 0x65, 0xcc, 0xdd, 0x77, 0x95, 
+   0xe5, 0x9c, 0x67, 0x65, 0xbd, 0xc7, 0xbc, 0xac, 0x97, 0xd7, 0xb5, 0x42, 
+   0xa6, 0x3a, 0xf1, 0x09, 0xbf, 0xcc, 0x31, 0x7f, 0xf0, 0x82, 0x90, 0x1a, 
+   0x6c, 0x2e, 0x2d, 0x2b, 0x55, 0x7b, 0xd5, 0xf4, 0xab, 0x01, 0x24, 0x8e, 
+   0x95, 0x0f, 0x53, 0x45, 0xb8, 0xe4, 0x5d, 0x9b, 0x45, 0x49, 0x31, 0x95, 
+   0x3a, 0x33, 0x76, 0x93, 0xea, 0xfe, 0xbb, 0x95, 0x9a, 0xeb, 0xe2, 0xa2, 
+   0xf5, 0x5a, 0x2e, 0xbd, 0x5a, 0x41, 0x37, 0x02, 0xdd, 0x0f, 0x0b, 0x34, 
+   0x00, 0x90, 0x8d, 0x29, 0x18, 0x85, 0x13, 0xd1, 0x20, 0x3b, 0x7b, 0xa5, 
+   0x80, 0x5d, 0xc2, 0xa9, 0xf2, 0x0f, 0x37, 0x10, 0x59, 0x2d, 0x6e, 0x7e, 
+   0x3f, 0xb8, 0x8a, 0x02, 0xaf, 0x4c, 0xcb, 0x5a, 0x68, 0x9f, 0x9f, 0x90, 
+   0xd8, 0xf7, 0x16, 0xe4, 0xee, 0x99, 0xc8, 0xdb, 0xcd, 0x24, 0x7f, 0xae, 
+   0x03, 0x95, 0xac, 0x0a, 0xda, 0xe2, 0xc5, 0x9d, 0x66, 0x88, 0xf1, 0x42, 
+   0xac, 0xea, 0xf7, 0x66, 0x34, 0x34, 0xa1, 0xac, 0x14, 0x22, 0x5b, 0x7d, 
+   0x2b, 0x82, 0x14, 0x2f, 0x93, 0xa6, 0x86, 0xe7, 0x7e, 0xed, 0x33, 0x8e, 
+   0x0b, 0x59, 0x59, 0x44, 0x54, 0x2d, 0xdc, 0x8a, 0x9e, 0xe1, 0xc0, 0x5e, 
+   0xc5, 0xbb, 0x8d, 0x4d, 0xb7, 0x71, 0xc8, 0x3a, 0xa6, 0x10, 0x9f, 0xd7, 
+   0xc7, 0xfd, 0x46, 0x28, 0x1e, 0xa0, 0xd9, 0x15, 0xde, 0xf9, 0x6c, 0x50, 
+   0x66, 0x03, 0x79, 0xf0, 0x12, 0x92, 0xe4, 0x87, 0x39, 0x01, 0x7f, 0xcb, 
+   0x98, 0x9d, 0xef, 0x64, 0x67, 0x7b, 0x9f, 0xae, 0x2f, 0xcb, 0x91, 0x40, 
+   0xd5, 0xc8, 0x61, 0xfc, 0x7a, 0xf1, 0x4c, 0x36, 0x3d, 0xbd, 0x75, 0x8a, 
+   0x6f, 0x13, 0x4a, 0x71, 0xca, 0xcf, 0x95, 0xb8, 0xa2, 0x45, 0xe6, 0xe2, 
+   0xfa, 0xfe, 0x67, 0xb6, 0xfd, 0xa8, 0xee, 0x0f, 0xb8, 0x4e, 0x48, 0x23, 
+   0xf1, 0x7b, 0xd6, 0xbd, 0xf1, 0xb3, 0x0b, 0x6c, 0xa0, 0xca, 0x26, 0xac, 
+   0xf3, 0x27, 0x8b, 0xd6, 0x40, 0x74, 0x8a, 0xe2, 0x73, 0x26, 0x9a, 0x0c, 
+   0x29, 0x3e, 0xcc, 0x1d, 0x1e, 0xb7, 0x48, 0x56, 0x65, 0x2a, 0x47, 0xa1, 
+   0x54, 0xe2, 0x67, 0xcb, 0x0d, 0x7e, 0x03, 0x8e, 0x15, 0xda, 0xbb, 0x60, 
+   0x17, 0x1c, 0xc1, 0x91, 0x77, 0x70, 0xe4, 0x19, 0xdf, 0x9b, 0x7e, 0x76, 
+   0xb7, 0x29, 0x66, 0xdb, 0x4f, 0x3e, 0xc4, 0x11, 0x2b, 0x8f, 0x53, 0x66, 
+   0x31, 0x97, 0xf2, 0x6a, 0x90, 0xfe, 0x90, 0x80, 0x9a, 0xf4, 0xba, 0xb3, 
+   0xf1, 0x5b, 0x4a, 0xd2, 0x40, 0x9d, 0x85, 0x45, 0xf5, 0xa2, 0xf0, 0x91, 
+   0x94, 0xf6, 0xa3, 0x5d, 0x90, 0xba, 0x7b, 0x8f, 0xde, 0x33, 0x49, 0x1f, 
+   0x98, 0x12, 0x83, 0xe3, 0x85, 0x23, 0x5c, 0xcf, 0x94, 0x91, 0xb6, 0x71, 
+   0xa4, 0xb8, 0xce, 0x29, 0x23, 0xbd, 0x8e, 0xa3, 0x8d, 0x77, 0xcf, 0x24, 
+   0xef, 0xca, 0xbb, 0x82, 0x76, 0x3b, 0xae, 0x58, 0xca, 0x28, 0x6f, 0xc2, 
+   0x80, 0x1d, 0xf1, 0x27, 0x35, 0x57, 0x8d, 0x9c, 0xae, 0x15, 0xaf, 0x74, 
+   0x01, 0x83, 0x38, 0xa5, 0xe0, 0x74, 0x0f, 0x76, 0xc0, 0x89, 0x33, 0xca, 
+   0xb1, 0x08, 0x0f, 0x75, 0x9c, 0x38, 0x9f, 0x9c, 0xae, 0xb5, 0x3b, 0x60, 
+   0xc9, 0x26, 0xa7, 0xda, 0x63, 0x00, 0xc1, 0x68, 0xc1, 0x25, 0xb7, 0x6d, 
+   0x45, 0xd1, 0xaa, 0xf6, 0xc2, 0x38, 0x65, 0xb7, 0x53, 0xab, 0x78, 0x71, 
+   0x6e, 0xb9, 0x2e, 0xee, 0x65, 0x54, 0xf1, 0xe2, 0x1c, 0x73, 0x5d, 0xdc, 
+   0xd1, 0xa8, 0xe2, 0xc5, 0xb8, 0xe6, 0xba, 0xb8, 0x16, 0xa8, 0x62, 0xb5, 
+   0xe0, 0x5c, 0xcb, 0x85, 0x29, 0x5c, 0xf3, 0x70, 0x89, 0xd1, 0x99, 0x46, 
+   0xee, 0x75, 0xfc, 0x29, 0x9c, 0xe9, 0xbf, 0xbe, 0x37, 0xad, 0x3d, 0x26, 
+   0xd9, 0x10, 0x3b, 0x2e, 0xf2, 0x8d, 0x7d, 0x75, 0x5c, 0xda, 0x77, 0xf0, 
+   0xd5, 0x71, 0x61, 0x37, 0x52, 0x01, 0xc1, 0x6c, 0x51, 0x41, 0xdf, 0x30, 
+   0x0c, 0xc0, 0xcd, 0x9e, 0x65, 0x18, 0x80, 0xbb, 0xcb, 0xaf, 0x19, 0x06, 
+   0x54, 0xbd, 0xe4, 0xef, 0x16, 0x07, 0x58, 0x34, 0x2a, 0xfe, 0x3e, 0x81, 
+   0x00, 0xbe, 0x4b, 0xf3, 0x9b, 0xb1, 0xe2, 0x96, 0xd0, 0x38, 0xd7, 0x9d, 
+   0xb0, 0x2e, 0xc1, 0xf7, 0x66, 0x3f, 0x7c, 0xce, 0x57, 0xc0, 0xe4, 0x31, 
+   0xa1, 0x44, 0x8b, 0x17, 0x8f, 0x5e, 0xec, 0xef, 0x31, 0xe9, 0x49, 0xfd, 
+   0x50, 0x79, 0x67, 0x5a, 0x96, 0xd4, 0xd5, 0x75, 0x74, 0x80, 0x76, 0x13, 
+   0x65, 0x8b, 0x86, 0x87, 0x1d, 0x36, 0x8b, 0x6c, 0x50, 0x16, 0xcc, 0x30, 
+   0xca, 0x2c, 0x75, 0x8a, 0x02, 0x73, 0xe1, 0x3b, 0xb7, 0x10, 0xdc, 0x84, 
+   0xa5, 0x0f, 0xef, 0x22, 0x2f, 0x06, 0x6e, 0x5f, 0x5b, 0x74, 0xdd, 0x3e, 
+   0x89, 0x14, 0x0c, 0xe4, 0xd1, 0x0b, 0x52, 0xc2, 0x5a, 0x99, 0xf2, 0x4f, 
+   0x16, 0xb6, 0x94, 0x24, 0x8f, 0x41, 0x3a, 0x7f, 0x00, 0xb0, 0xe3, 0x8e, 
+   0x6a, 0x7e, 0xe3, 0x3a, 0xcf, 0xf6, 0x91, 0x0d, 0xb3, 0x34, 0xfb, 0x2b, 
+   0xd6, 0xc2, 0xb3, 0xdc, 0x02, 0xb5, 0x51, 0xeb, 0xf2, 0xab, 0x88, 0xb0, 
+   0xe7, 0x5e, 0x79, 0x8d, 0x24, 0x80, 0xa0, 0xb2, 0x5f, 0x5f, 0x2d, 0xcc, 
+   0x6c, 0x55, 0x02, 0x9f, 0x57, 0x88, 0x33, 0x95, 0x63, 0x2e, 0x3c, 0xae, 
+   0xba, 0xa4, 0x9f, 0xc9, 0x67, 0x28, 0x21, 0xcf, 0x10, 0xb3, 0xbd, 0xf9, 
+   0x3f, 0xdb, 0x80, 0x35, 0x99, 0x65, 0xed, 0x54, 0x56, 0xd4, 0x27, 0x2a, 
+   0x6d, 0x4c, 0xf6, 0xc8, 0xe8, 0x9c, 0x3f, 0x32, 0x0a, 0x7f, 0x43, 0xcb, 
+   0x22, 0x73, 0xdf, 0x34, 0xd8, 0x55, 0x5a, 0x53, 0xed, 0xb0, 0x42, 0x81, 
+   0xdb, 0xb8, 0x40, 0x80, 0x8c, 0xbb, 0x5c, 0x7c, 0x6b, 0x14, 0x63, 0x2b, 
+   0x15, 0xf3, 0x76, 0x0b, 0x02, 0xb1, 0xbf, 0x1a, 0xcf, 0x6c, 0xee, 0xe0, 
+   0x36, 0x89, 0xf4, 0x25, 0x66, 0x8b, 0x48, 0xbf, 0xba, 0xbc, 0xc6, 0x0c, 
+   0xeb, 0xe2, 0x47, 0x30, 0xdf, 0x35, 0xc3, 0xf0, 0x07, 0xa5, 0x18, 0x2c, 
+   0xb2, 0x01, 0xcd, 0x53, 0x0c, 0x7c, 0x24, 0x23, 0xf5, 0x32, 0x60, 0xc5, 
+   0xeb, 0xb9, 0x16, 0x37, 0x5d, 0x3e, 0xfe, 0x8b, 0xe7, 0x1b, 0x76, 0x49, 
+   0x38, 0x28, 0x7c, 0x73, 0x64, 0x87, 0x40, 0xfb, 0x84, 0x83, 0x4d, 0x72, 
+   0xa0, 0x69, 0xc2, 0xa1, 0x21, 0xd7, 0xfe, 0xd2, 0xd9, 0x87, 0x5d, 0xd3, 
+   0x0f, 0x0a, 0xdf, 0x5c, 0x79, 0xe1, 0xb6, 0x59, 0xfa, 0xc1, 0x26, 0x4d, 
+   0xb0, 0x4b, 0xfa, 0xa1, 0x21, 0xff, 0xfe, 0x03, 0x72, 0x11, 0x7e, 0xb4, 
+   0xf6, 0x53, 0xd6, 0x55, 0x68, 0x87, 0x8c, 0x44, 0x31, 0x98, 0xd5, 0x94, 
+   0xf0, 0x7a, 0x2e, 0xbb, 0xac, 0x94, 0x5d, 0xfa, 0x10, 0x97, 0x01, 0x7c, 
+   0xf7, 0x4e, 0x1a, 0xee, 0x57, 0x14, 0x21, 0xca, 0x5b, 0x14, 0x83, 0x91, 
+   0x68, 0x16, 0x0c, 0xb6, 0xcb, 0x12, 0x16, 0x33, 0x64, 0x45, 0x5b, 0x0d, 
+   0x38, 0x63, 0x97, 0x2e, 0xac, 0x9b, 0xc0, 0xc6, 0x44, 0x34, 0x17, 0xd3, 
+   0x29, 0x05, 0xf2, 0x57, 0x2b, 0x71, 0xa6, 0xbd, 0x8b, 0xa8, 0x6a, 0x08, 
+   0x0a, 0xca, 0x0f, 0x22, 0x16, 0xfd, 0xae, 0x5e, 0x55, 0x6c, 0xad, 0x66, 
+   0x7a, 0x35, 0x5e, 0x6b, 0xb3, 0xa9, 0xec, 0x68, 0xb6, 0x34, 0x3b, 0xbe, 
+   0xdb, 0x4e, 0xb6, 0xb3, 0x0c, 0xd8, 0x52, 0x73, 0xc8, 0xfc, 0x46, 0x4d, 
+   0x18, 0x90, 0xf5, 0x59, 0x9e, 0x82, 0xd8, 0xce, 0x66, 0x63, 0x53, 0x77, 
+   0x90, 0x16, 0x95, 0xa4, 0x8d, 0x97, 0x68, 0x79, 0x84, 0xd0, 0x60, 0x42, 
+   0x9b, 0x55, 0x1a, 0x04, 0xa7, 0x41, 0xf6, 0xbb, 0x5f, 0x04, 0x03, 0x96, 
+   0x45, 0x25, 0x35, 0x1b, 0xd4, 0xd2, 0xf3, 0x9c, 0x6b, 0x51, 0x54, 0x5d, 
+   0xdd, 0x5e, 0x95, 0x7c, 0xcf, 0xf2, 0xda, 0xaa, 0xac, 0x99, 0x2c, 0x75, 
+   0x59, 0x00, 0x2b, 0x73, 0xe4, 0x3a, 0x1f, 0x58, 0xa3, 0x75, 0x71, 0x85, 
+   0x0e, 0x2b, 0xe7, 0x91, 0x83, 0x74, 0x78, 0x17, 0x87, 0x0f, 0x9e, 0xd8, 
+   0x45, 0xc0, 0x0c, 0xbe, 0x85, 0xc2, 0xdf, 0x3d, 0xb3, 0xce, 0xfe, 0x19, 
+   0x78, 0x1b, 0x05, 0xe7, 0x45, 0x62, 0x05, 0x7c, 0x07, 0x85, 0xe7, 0xf1, 
+   0x62, 0x06, 0xdd, 0xb5, 0x81, 0x76, 0x67, 0xce, 0x81, 0xf8, 0xbf, 0x62, 
+   0x9c, 0x7c, 0x8d, 0xc4, 0xd8, 0x1e, 0x73, 0x12, 0xdc, 0x87, 0x3c, 0x82, 
+   0x4d, 0x39, 0x3f, 0x80, 0xe3, 0x01, 0x3c, 0x8e, 0xba, 0x62, 0x37, 0xa0, 
+   0x59, 0xcb, 0xe0, 0x5a, 0x3c, 0x78, 0xf0, 0x34, 0x66, 0xae, 0x86, 0xc7, 
+   0x12, 0x56, 0xe9, 0xbe, 0x88, 0x97, 0x2b, 0x48, 0x1a, 0x66, 0xc0, 0x58, 
+   0xac, 0x5c, 0xe4, 0xe9, 0x2b, 0xb9, 0x6c, 0x20, 0xc1, 0x82, 0x5f, 0x76, 
+   0x31, 0xa2, 0xff, 0x04, 0x1d, 0x03, 0x18, 0x73, 0x37, 0xe0, 0xa7, 0x17, 
+   0xa7, 0x0b, 0x27, 0xc1, 0xb7, 0x20, 0x01, 0x1f, 0x89, 0xb1, 0x48, 0xcb, 
+   0x88, 0xea, 0xf8, 0x5a, 0x9b, 0xa5, 0x56, 0x3c, 0xbe, 0x44, 0xef, 0xd4, 
+   0xbb, 0x10, 0x6f, 0x7a, 0xe7, 0x4d, 0xef, 0xbc, 0xe9, 0x9d, 0x37, 0xbd, 
+   0x03, 0xe9, 0x9d, 0x72, 0xae, 0x35, 0x6b, 0x77, 0x77, 0x4e, 0xdd, 0xab, 
+   0x27, 0x00, 0x85, 0x03, 0xd5, 0xc1, 0xd5, 0xe8, 0x2e, 0x96, 0x00, 0xbe, 
+   0xf7, 0x43, 0x3f, 0x0e, 0xe6, 0x59, 0x45, 0x2f, 0x3f, 0x4c, 0xa5, 0x81, 
+   0x6c, 0xba, 0xdd, 0x40, 0xba, 0x4a, 0x5e, 0xb1, 0x51, 0xb4, 0x0c, 0xd2, 
+   0x1b, 0xae, 0x82, 0xe2, 0xa0, 0xaa, 0x70, 0x64, 0x3e, 0x07, 0xc2, 0x51, 
+   0x45, 0x42, 0x77, 0x6b, 0xf0, 0x1b, 0x10, 0x25, 0xe3, 0xb9, 0x1b, 0xf6, 
+   0xea, 0x0f, 0x70, 0x58, 0xe6, 0xe2, 0x67, 0x88, 0x89, 0x0f, 0x1d, 0x57, 
+   0xb9, 0x78, 0x64, 0x9e, 0x68, 0x1c, 0x3f, 0x94, 0xdc, 0x2a, 0x92, 0x40, 
+   0x46, 0x91, 0x83, 0xf5, 0xba, 0x3c, 0xf0, 0x34, 0xab, 0x67, 0x91, 0xdb, 
+   0xa7, 0x6c, 0x14, 0x95, 0x1d, 0x3d, 0x76, 0xe1, 0x15, 0xad, 0x12, 0x2f, 
+   0x0d, 0x72, 0xba, 0xda, 0x28, 0x93, 0x52, 0x2f, 0x8d, 0x6a, 0xb9, 0xda, 
+   0x28, 0x9b, 0x8e, 0xaf, 0xcc, 0xbe, 0x15, 0xa4, 0x95, 0x52, 0x01, 0x5e, 
+   0x3d, 0x68, 0x20, 0xcc, 0x59, 0x52, 0xe6, 0x35, 0x64, 0xd9, 0x69, 0x26, 
+   0x87, 0x78, 0xc9, 0x80, 0x49, 0x0e, 0x71, 0x3d, 0x6c, 0x90, 0x43, 0xdc, 
+   0x10, 0x84, 0xdb, 0x75, 0x3e, 0xae, 0x03, 0xc8, 0xaf, 0xf9, 0x6e, 0xdd, 
+   0x9b, 0xfc, 0xfe, 0x1e, 0xf2, 0x3b, 0xf5, 0x9f, 0xd2, 0x6d, 0xec, 0xe7, 
+   0xe1, 0x31, 0xbb, 0xb1, 0x4a, 0x7f, 0xcf, 0x6e, 0xbd, 0x1d, 0x90, 0xc1, 
+   0xe9, 0xd9, 0xec, 0x72, 0x74, 0x32, 0xfc, 0xc5, 0xc1, 0xef, 0x46, 0xf8, 
+   0xf1, 0x9e, 0xa8, 0xcd, 0x58, 0x7b, 0xc9, 0xaf, 0xc6, 0xe7, 0xb4, 0x5b, 
+   0x16, 0x15, 0x58, 0xf4, 0x8b, 0x28, 0x9a, 0xec, 0x65, 0x51, 0x76, 0xf9, 
+   0x34, 0x26, 0xd5, 0xe7, 0xda, 0xda, 0x80, 0x08, 0x1a, 0xbd, 0xcd, 0x32, 
+   0xc2, 0xb4, 0x1e, 0x61, 0xab, 0x31, 0xc2, 0xa4, 0x16, 0xa1, 0x2c, 0x0d, 
+   0x30, 0x5f, 0x48, 0x25, 0x9f, 0x63, 0x6f, 0xa3, 0xb5, 0xd6, 0x94, 0x0c, 
+   0x2d, 0x5d, 0x6b, 0x03, 0xa5, 0x89, 0x0f, 0x67, 0x8d, 0xac, 0xc8, 0x78, 
+   0x78, 0x3d, 0xec, 0x4f, 0x51, 0xa1, 0x95, 0x03, 0x06, 0x17, 0xfd, 0xcb, 
+   0x6b, 0x54, 0x5c, 0x25, 0xfc, 0xe5, 0xf9, 0x78, 0x3c, 0x1a, 0xa3, 0x4e, 
+   0xb4, 0x1c, 0x70, 0x3c, 0x1a, 0x9f, 0x0c, 0xc7, 0xa8, 0x17, 0x5d, 0x9e, 
+   0x61, 0x36, 0xba, 0x1a, 0x0c, 0xab, 0xce, 0x74, 0xcd, 0x86, 0x98, 0xd6, 
+   0x52, 0xd1, 0x62, 0xfb, 0xbf, 0x51, 0x91, 0xb9, 0x8a, 0xb5, 0x54, 0xb4, 
+   0xb8, 0x34, 0xf6, 0x46, 0x45, 0x16, 0x12, 0xf8, 0x8f, 0x59, 0xc3, 0xb5, 
+   0x6d, 0x2c, 0xf2, 0x90, 0x70, 0xe7, 0x6a, 0xdd, 0x66, 0xa3, 0xa6, 0x9f, 
+   0x05, 0xfd, 0xd1, 0x63, 0x15, 0x0d, 0x6e, 0xc0, 0x07, 0x91, 0xbf, 0x5c, 
+   0x06, 0xf3, 0x80, 0x1d, 0x80, 0xaf, 0xa2, 0xe8, 0xd7, 0xed, 0xc6, 0xf0, 
+   0x45, 0xf6, 0x25, 0x56, 0xfd, 0x6d, 0x1a, 0x5d, 0x8c, 0x4e, 0x7e, 0x64, 
+   0xf5, 0x79, 0xe4, 0x2e, 0xa0, 0x51, 0x5a, 0x14, 0xae, 0x80, 0x57, 0xa8, 
+   0x2c, 0xea, 0x37, 0xd9, 0xe0, 0x64, 0xbb, 0xd9, 0xac, 0x02, 0xc8, 0x3f, 
+   0xc1, 0xcf, 0x2d, 0xcf, 0x3c, 0xde, 0x74, 0x38, 0xeb, 0xef, 0x6d, 0x68, 
+   0x12, 0x8e, 0x57, 0x08, 0xf8, 0xe9, 0xfc, 0xa1, 0x1e, 0x49, 0xc5, 0x63, 
+   0xaf, 0x62, 0xb9, 0x10, 0xd4, 0x85, 0x93, 0x1c, 0x16, 0xfb, 0x27, 0x37, 
+   0xc9, 0xee, 0x09, 0xba, 0x73, 0x24, 0x68, 0xfe, 0x02, 0x18, 0xb2, 0x75, 
+   0xf2, 0x01, 0xad, 0x13, 0x74, 0xd3, 0xe4, 0xa0, 0x83, 0xed, 0x1d, 0xdd, 
+   0x06, 0xde, 0x06, 0xdd, 0x34, 0xf9, 0x00, 0xe7, 0x04, 0x4d, 0x81, 0x48, 
+   0x50, 0xf1, 0xec, 0x1e, 0x9a, 0x03, 0x29, 0x3e, 0xe6, 0x21, 0xa0, 0x74, 
+   0xe6, 0x9d, 0x25, 0xaa, 0x99, 0x10, 0xed, 0xf6, 0xe7, 0x0e, 0x1e, 0x8f, 
+   0xd3, 0xc4, 0xe3, 0xc9, 0xd1, 0x24, 0x3c, 0xff, 0x16, 0xf3, 0xac, 0x1b, 
+   0x61, 0x17, 0xad, 0x49, 0xf9, 0x82, 0x74, 0xaf, 0x2a, 0x81, 0x75, 0x67, 
+   0x1a, 0x63, 0xea, 0x14, 0xc6, 0x21, 0x79, 0x64, 0xd7, 0x93, 0x59, 0x4a, 
+   0x21, 0xcd, 0x26, 0x52, 0xab, 0xe2, 0x91, 0x8b, 0xd2, 0xe6, 0xb5, 0x5f, 
+   0xde, 0xb0, 0xe3, 0x2d, 0xd6, 0x06, 0x52, 0x71, 0xf4, 0x2c, 0xce, 0x1b, 
+   0xd5, 0xb5, 0x7a, 0x8d, 0x52, 0x87, 0x7f, 0xfc, 0xfa, 0x1c, 0x74, 0x7d, 
+   0xa2, 0xe0, 0x37, 0x7e, 0x9d, 0xf5, 0x59, 0xba, 0xc2, 0x78, 0x04, 0x7f, 
+   0x13, 0x86, 0x32, 0x89, 0x36, 0x57, 0x1a, 0x7e, 0xec, 0x70, 0xd2, 0x9f, 
+   0x75, 0x45, 0x65, 0x3a, 0xa9, 0x08, 0x35, 0x2c, 0x0a, 0x10, 0x9a, 0x92, 
+   0xda, 0xd5, 0x49, 0x2d, 0xf7, 0x10, 0xfa, 0xb8, 0x87, 0xd2, 0xd5, 0x0a, 
+   0xef, 0xa7, 0x38, 0xda, 0x54, 0x05, 0x45, 0xf6, 0x81, 0x66, 0x5f, 0x36, 
+   0x82, 0x3b, 0x13, 0x67, 0x8f, 0xea, 0x8e, 0x01, 0x27, 0xbd, 0x8b, 0x79, 
+   0xfd, 0xd9, 0xd8, 0x29, 0x30, 0xd6, 0xb5, 0x1c, 0x3b, 0xa9, 0x8e, 0xed, 
+   0x61, 0x43, 0x33, 0xcb, 0x76, 0xb9, 0xa3, 0xad, 0xce, 0x86, 0x0f, 0xa2, 
+   0xf5, 0x26, 0x0a, 0x7d, 0x99, 0x3f, 0x95, 0x42, 0x5d, 0x69, 0xe9, 0x5b, 
+   0xeb, 0x3a, 0xc0, 0xdf, 0x81, 0x9b, 0xd6, 0xac, 0xa6, 0x5e, 0xef, 0xf7, 
+   0xe0, 0x56, 0x77, 0x83, 0xf1, 0x23, 0x4a, 0x4e, 0xc7, 0x4e, 0x9b, 0x40, 
+   0x70, 0x41, 0x34, 0x05, 0x26, 0xbd, 0xef, 0xbf, 0x03, 0x5c, 0x4c, 0xe8, 
+   0xa5, 0x5f, 0x70, 0x84, 0x12, 0x80, 0x2e, 0x9a, 0xfc, 0x7b, 0xeb, 0xc7, 
+   0x80, 0x83, 0x85, 0x7b, 0x7f, 0x6f, 0x3b, 0xe6, 0x6d, 0xc7, 0xfc, 0x09, 
+   0x76, 0x4c, 0x66, 0xea, 0xd5, 0xa3, 0x4b, 0xd4, 0xd1, 0x62, 0x7d, 0xcc, 
+   0xbe, 0x8c, 0xc6, 0xe6, 0x17, 0x92, 0x5c, 0x8b, 0x37, 0xef, 0x2e, 0xfc, 
+   0x6f, 0xd4, 0x20, 0x1f, 0xb0, 0x7e, 0x66, 0x2c, 0xb1, 0x38, 0x4f, 0x57, 
+   0xcf, 0xac, 0x41, 0x1a, 0x84, 0x0c, 0xe7, 0x4c, 0x8e, 0x8c, 0x7d, 0x95, 
+   0xd1, 0xbc, 0x77, 0x70, 0x03, 0x9f, 0xe3, 0xb9, 0x39, 0x9e, 0x5d, 0xf7, 
+   0x81, 0x96, 0xda, 0x6e, 0xb5, 0xcf, 0xb1, 0xc9, 0xb0, 0x9b, 0x6c, 0xba, 
+   0x5b, 0x7d, 0x3e, 0xaa, 0xb1, 0xa0, 0xd9, 0x9c, 0x52, 0xf1, 0x0b, 0x78, 
+   0x7c, 0x3d, 0xd0, 0x32, 0xf0, 0x13, 0xb4, 0x53, 0x76, 0xd0, 0x4a, 0x58, 
+   0x18, 0x58, 0xac, 0x41, 0xe9, 0x0f, 0x27, 0x4f, 0x58, 0x92, 0xde, 0x4f, 
+   0x3d, 0xd3, 0x21, 0x95, 0x98, 0x7f, 0x2f, 0x5a, 0xee, 0x9d, 0xf8, 0xa9, 
+   0x17, 0xac, 0x20, 0x4c, 0x07, 0x3d, 0x1c, 0x53, 0x10, 0x5a, 0x60, 0x72, 
+   0x8e, 0xdc, 0x32, 0x26, 0x80, 0x2a, 0x51, 0x2c, 0x6a, 0x9d, 0x58, 0xdb, 
+   0x3a, 0x6f, 0xb5, 0x79, 0xf0, 0x20, 0x4c, 0x87, 0xf8, 0xab, 0x8f, 0x1a, 
+   0xa2, 0x3b, 0xa5, 0x0b, 0xb5, 0x8a, 0xa7, 0x8b, 0x93, 0x59, 0xc3, 0x73, 
+   0x1f, 0xfb, 0x7e, 0x08, 0x21, 0xb2, 0xe8, 0x20, 0xa6, 0x21, 0x8a, 0xc1, 
+   0x26, 0xdd, 0x8e, 0xf5, 0xdb, 0x46, 0x13, 0x7e, 0x75, 0x70, 0x4a, 0xf8, 
+   0xdb, 0xa8, 0xf3, 0x68, 0xcd, 0xee, 0xac, 0xb1, 0xfb, 0x1b, 0xe0, 0xbd, 
+   0x48, 0xc7, 0xe2, 0xcc, 0x81, 0x63, 0x1d, 0x4e, 0x07, 0xd0, 0xab, 0x49, 
+   0x4e, 0xa9, 0x5e, 0x19, 0x7b, 0x14, 0x9b, 0xe3, 0xca, 0x63, 0x94, 0xaf, 
+   0xa4, 0xff, 0x14, 0x40, 0x0e, 0xb8, 0x45, 0x69, 0xb5, 0x86, 0xe8, 0x8b, 
+   0x11, 0x51, 0x45, 0xfd, 0x00, 0x1b, 0xf6, 0x31, 0xf8, 0xed, 0x37, 0xa6, 
+   0xd8, 0x81, 0xfc, 0x9c, 0xab, 0x37, 0xbc, 0x03, 0x83, 0xe9, 0x7c, 0xfc, 
+   0xad, 0x1f, 0x47, 0x68, 0x8a, 0x21, 0x07, 0x1e, 0x85, 0x3e, 0x9a, 0x5d, 
+   0xc8, 0x61, 0xc7, 0x4a, 0xd1, 0x84, 0x29, 0xbd, 0x90, 0xc3, 0x9e, 0x09, 
+   0x29, 0x44, 0x72, 0x0b, 0x39, 0xf4, 0x31, 0x97, 0x7d, 0x24, 0xbb, 0x50, 
+   0xd0, 0x47, 0xec, 0xb8, 0x72, 0x6a, 0xa1, 0x2e, 0xf7, 0x99, 0x4f, 0x03, 
+   0x50, 0xd6, 0xc1, 0x95, 0x71, 0xb1, 0x26, 0x68, 0x38, 0xee, 0x35, 0x14, 
+   0xe4, 0x83, 0x86, 0x5b, 0xdc, 0xb6, 0x12, 0x8f, 0x4b, 0x30, 0xc7, 0xc5, 
+   0xa3, 0x22, 0x76, 0xba, 0x0d, 0xb5, 0xc2, 0x70, 0x35, 0x89, 0x5b, 0x78, 
+   0x9b, 0x15, 0x60, 0xf8, 0x40, 0x6d, 0x7c, 0x76, 0x0c, 0xc8, 0xea, 0x81, 
+   0xc5, 0xf5, 0xe9, 0xdc, 0xe4, 0xab, 0x05, 0x20, 0xca, 0x61, 0xde, 0x81, 
+   0xc5, 0x43, 0xbf, 0xa2, 0x8f, 0x28, 0x5f, 0x5c, 0xf1, 0x11, 0x8a, 0x3b, 
+   0x88, 0x6b, 0x4f, 0xb5, 0x11, 0x29, 0x80, 0xe1, 0x10, 0x77, 0xa5, 0x94, 
+   0x77, 0xf4, 0x00, 0x04, 0x1d, 0x0b, 0x45, 0xc7, 0xd3, 0x5c, 0x79, 0x73, 
+   0xd4, 0xf7, 0xdd, 0xf6, 0x1e, 0x2b, 0x15, 0xa2, 0x36, 0x81, 0x97, 0xda, 
+   0xc8, 0x07, 0x3e, 0x94, 0xf7, 0x5c, 0x6c, 0xda, 0x2e, 0x66, 0xc4, 0xe5, 
+   0x69, 0x27, 0x7e, 0xcf, 0x55, 0x1a, 0x70, 0x76, 0xa8, 0x5c, 0x76, 0x15, 
+   0xeb, 0x6f, 0x9f, 0x05, 0xab, 0xb4, 0xb6, 0x15, 0x1f, 0x8b, 0x12, 0x32, 
+   0xa0, 0x1d, 0x1d, 0x3d, 0xeb, 0xa7, 0x3e, 0x3b, 0x8a, 0x19, 0xa9, 0xde, 
+   0x0d, 0x6d, 0xee, 0x22, 0x3a, 0xad, 0x57, 0x74, 0x11, 0x9d, 0xd6, 0xeb, 
+   0xb8, 0x88, 0x4e, 0xeb, 0xe5, 0x2e, 0xa2, 0x95, 0x91, 0xad, 0x73, 0xce, 
+   0x1c, 0x8b, 0x46, 0x9c, 0x85, 0x6f, 0x05, 0x0d, 0xb7, 0xd8, 0xc0, 0x75, 
+   0xf6, 0xca, 0x39, 0x7c, 0xb3, 0x57, 0xdf, 0xc9, 0x5e, 0xe1, 0x8c, 0xad, 
+   0xb5, 0x57, 0xce, 0x8b, 0xec, 0xd5, 0x01, 0x6e, 0x2d, 0x87, 0x4f, 0x29, 
+   0xab, 0xd8, 0x87, 0x9a, 0x08, 0x7c, 0x17, 0xcb, 0xf2, 0x66, 0x58, 0x5e, 
+   0xc7, 0xb0, 0x34, 0x2d, 0x15, 0x1e, 0x53, 0xd5, 0x11, 0x27, 0xec, 0x91, 
+   0x29, 0xea, 0xe6, 0xb3, 0x96, 0x0f, 0x6a, 0xf0, 0x50, 0x65, 0x3e, 0xae, 
+   0xa9, 0xfb, 0x1f, 0x4f, 0xc6, 0xd5, 0x71, 0xb8, 0x52, 0x86, 0xdd, 0x18, 
+   0xbc, 0x06, 0xaf, 0x79, 0x90, 0x82, 0x2b, 0x76, 0xbb, 0xb0, 0xe2, 0x75, 
+   0x82, 0x8a, 0x66, 0x49, 0x9a, 0xfc, 0x30, 0xc4, 0xce, 0x68, 0x67, 0xbc, 
+   0x9c, 0xf3, 0x40, 0xb0, 0xaf, 0x86, 0xb8, 0xda, 0x85, 0x25, 0xeb, 0xd0, 
+   0x54, 0x20, 0x3a, 0x36, 0xbc, 0xd9, 0x54, 0xb5, 0x76, 0xf5, 0x78, 0xce, 
+   0xd4, 0x10, 0x57, 0x41, 0x74, 0x64, 0x1d, 0x72, 0x0b, 0x3c, 0x63, 0xa9, 
+   0xa5, 0x14, 0x2c, 0x5d, 0x8b, 0xc4, 0x48, 0x5e, 0x3c, 0xdb, 0x0f, 0x83, 
+   0x24, 0x4a, 0xe3, 0x68, 0x53, 0xa4, 0x6f, 0x15, 0x3c, 0xb8, 0xaa, 0xd5, 
+   0xe2, 0x6d, 0xb8, 0x34, 0x43, 0x2a, 0x8b, 0x2a, 0x30, 0x84, 0x92, 0xd7, 
+   0x38, 0x9c, 0x67, 0x9b, 0x11, 0xd8, 0x86, 0x78, 0x98, 0xcd, 0x31, 0x00, 
+   0xca, 0x5f, 0xa6, 0xd0, 0x64, 0x25, 0x89, 0x71, 0xfc, 0x14, 0x18, 0xef, 
+   0x34, 0x18, 0x3f, 0xa9, 0x8e, 0x97, 0x69, 0x33, 0x6d, 0x7c, 0xc3, 0xac, 
+   0x53, 0xab, 0x92, 0xe0, 0x31, 0x3b, 0x46, 0x40, 0xaa, 0x48, 0xbd, 0x3c, 
+   0x57, 0xb0, 0xb7, 0xbd, 0x43, 0xce, 0x49, 0xa9, 0xd5, 0x2e, 0xe3, 0x81, 
+   0x55, 0x1c, 0x39, 0x31, 0xb9, 0xc1, 0x16, 0x56, 0xd5, 0x36, 0x86, 0x2c, 
+   0x58, 0x6c, 0x17, 0x41, 0x4a, 0xf1, 0x37, 0x76, 0xc2, 0xc7, 0xf5, 0xf0, 
+   0x25, 0xd5, 0x75, 0x4b, 0x1e, 0x80, 0x90, 0x2b, 0x9f, 0xce, 0xb9, 0x9b, 
+   0xe6, 0xbd, 0xe4, 0x17, 0xa8, 0xeb, 0xb1, 0xe0, 0x8a, 0xf7, 0xd2, 0xbb, 
+   0x47, 0xb1, 0xa0, 0x7a, 0xd7, 0x0f, 0xa9, 0x66, 0xd0, 0x2d, 0xee, 0x89, 
+   0x97, 0x7a, 0x59, 0xbe, 0xbd, 0x78, 0xd9, 0xe1, 0xbf, 0xf6, 0xf6, 0x7e, 
+   0xf8, 0xc7, 0x3a, 0x4a, 0xd2, 0xec, 0x09, 0xcc, 0x84, 0x3c, 0xf8, 0x31, 
+   0xbb, 0x01, 0xf3, 0x8d, 0xfa, 0x5e, 0xec, 0x3e, 0x78, 0xfa, 0xe0, 0xf3, 
+   0xa7, 0xb6, 0xd8, 0x8b, 0x18, 0x3f, 0x92, 0x7e, 0xfe, 0x2b, 0xfe, 0x50, 
+   0x06, 0xfb, 0xdd, 0x4f, 0x3f, 0xfc, 0x63, 0xf8, 0x34, 0xa7, 0x9c, 0xa5, 
+   0xfc, 0x49, 0x48, 0x18, 0xa5, 0xfe, 0xe2, 0x27, 0x8e, 0xba, 0x78, 0xcc, 
+   0x03, 0xaa, 0xb5, 0x50, 0x3e, 0x85, 0x8c, 0x7b, 0x55, 0xc7, 0xdb, 0x62, 
+   0x10, 0x99, 0x5c, 0x8d, 0xc6, 0x97, 0x55, 0x3f, 0x1c, 0x1f, 0x7a, 0xd6, 
+   0xab, 0x7a, 0xe4, 0x36, 0xa3, 0x4a, 0x33, 0xb6, 0x1a, 0x8c, 0x3d, 0xee, 
+   0xf7, 0x80, 0x12, 0x2f, 0xab, 0x71, 0xa5, 0x59, 0x3b, 0x4d, 0x46, 0x77, 
+   0xba, 0x1d, 0xad, 0x94, 0x45, 0xf4, 0xb6, 0x3f, 0x66, 0x4c, 0xcc, 0x79, 
+   0xda, 0x80, 0x4f, 0xf4, 0x6b, 0xda, 0xda, 0x5d, 0x1f, 0x81, 0xae, 0xcf, 
+   0x70, 0xfd, 0x48, 0xa5, 0x85, 0xcb, 0x44, 0x13, 0x74, 0x9d, 0x59, 0xdf, 
+   0x29, 0x10, 0xf6, 0x5e, 0x01, 0xa1, 0x73, 0x30, 0xeb, 0xbb, 0x05, 0xc6, 
+   0xa3, 0x02, 0xe3, 0x58, 0x60, 0xec, 0x37, 0xc5, 0xc8, 0x54, 0x76, 0x2e, 
+   0x5f, 0x0d, 0x64, 0xd3, 0xe9, 0x96, 0x85, 0xb3, 0x89, 0x74, 0xaa, 0x93, 
+   0x36, 0x91, 0xcf, 0xea, 0xac, 0xcd, 0x24, 0x54, 0x9d, 0xb7, 0x99, 0x8c, 
+   0x56, 0x67, 0x6e, 0x20, 0xa5, 0x4e, 0xf7, 0x54, 0x8e, 0xeb, 0x36, 0x5a, 
+   0xaf, 0x32, 0xf0, 0xb0, 0xe1, 0x07, 0x2b, 0x43, 0x7b, 0x0d, 0xbe, 0xd5, 
+   0x39, 0x9d, 0x9d, 0xb1, 0xff, 0xa1, 0x92, 0xa6, 0x60, 0x38, 0x6a, 0x32, 
+   0xf9, 0xd1, 0x6c, 0x28, 0x37, 0xa5, 0x6b, 0x2f, 0x55, 0x27, 0xc3, 0xeb, 
+   0xe9, 0x47, 0x32, 0x18, 0x5d, 0x5e, 0x2b, 0x8c, 0x72, 0xed, 0x05, 0x4b, 
+   0x8e, 0x77, 0xe5, 0x2e, 0x76, 0xed, 0x05, 0x4c, 0x8e, 0x6f, 0xb9, 0x72, 
+   0xed, 0xae, 0xbd, 0x8c, 0x71, 0x04, 0x6e, 0x7b, 0xf6, 0x45, 0x51, 0xbf, 
+   0xed, 0x62, 0x8b, 0x7e, 0x11, 0x5b, 0xf4, 0xa4, 0xe9, 0x16, 0x55, 0xd6, 
+   0xd2, 0x44, 0xe8, 0xe4, 0x28, 0x29, 0x72, 0xec, 0x33, 0xf2, 0xfb, 0x36, 
+   0xdb, 0x50, 0x58, 0x60, 0xea, 0xc3, 0xf1, 0x9b, 0x1a, 0x79, 0xe5, 0xa6, 
+   0xf5, 0x87, 0x4d, 0x94, 0x55, 0x1e, 0x62, 0x53, 0x88, 0x7b, 0x1d, 0xcd, 
+   0xe7, 0x50, 0x25, 0xa1, 0x87, 0x4d, 0x22, 0xce, 0x71, 0x1b, 0x4f, 0x32, 
+   0xd6, 0xd8, 0x7d, 0x84, 0xcd, 0xc2, 0xb2, 0x04, 0xbb, 0xcc, 0x72, 0xa6, 
+   0x4e, 0xd3, 0x3a, 0x40, 0x29, 0xe6, 0xee, 0xc6, 0x16, 0xb6, 0xfb, 0xb5, 
+   0x99, 0x94, 0x62, 0x68, 0x8b, 0xb1, 0xbd, 0xd9, 0x70, 0x3a, 0x90, 0x16, 
+   0xa6, 0xd5, 0x44, 0x3f, 0xd3, 0xc1, 0xd7, 0x37, 0x57, 0x83, 0x8f, 0xd3, 
+   0x8f, 0xe3, 0xd1, 0xcd, 0xd9, 0xc7, 0x59, 0xff, 0xe2, 0xfa, 0xa3, 0x62, 
+   0xff, 0x5a, 0xad, 0x06, 0x5f, 0xe2, 0x38, 0xb3, 0x61, 0x7f, 0x20, 0xc7, 
+   0xda, 0xeb, 0xeb, 0xc9, 0xf9, 0xd9, 0xd5, 0xf0, 0x64, 0x56, 0xc1, 0xd0, 
+   0xc4, 0xaf, 0x28, 0x8f, 0xb5, 0xd7, 0xda, 0xf9, 0xec, 0x15, 0x14, 0x87, 
+   0x8d, 0xd8, 0xd0, 0x17, 0x7c, 0xd0, 0x51, 0xd8, 0xeb, 0xf1, 0xaf, 0x83, 
+   0xe3, 0xc1, 0x78, 0x76, 0x71, 0x73, 0xd9, 0x97, 0xa3, 0xed, 0x75, 0xb8, 
+   0x18, 0xdd, 0x76, 0x0f, 0x66, 0x03, 0xca, 0x49, 0x05, 0x47, 0xfb, 0xc0, 
+   0x7e, 0x11, 0xc7, 0x03, 0xc9, 0xf9, 0xb6, 0xfd, 0x97, 0x1f, 0x2b, 0xc2, 
+   0xd7, 0xb6, 0xff, 0xe4, 0xe3, 0x41, 0x4b, 0x3a, 0x90, 0x0d, 0xbe, 0xb2, 
+   0x3f, 0x99, 0x0e, 0x66, 0xed, 0x27, 0xa9, 0x63, 0xbb, 0xf6, 0x82, 0xc6, 
+   0xc7, 0x76, 0xd4, 0xb1, 0xf6, 0x22, 0x96, 0x8d, 0x55, 0x9c, 0x57, 0x7b, 
+   0x11, 0xe3, 0x63, 0xbb, 0xea, 0x58, 0x7b, 0xdf, 0x20, 0x1b, 0x2b, 0xb5, 
+   0x6a, 0xd7, 0x9e, 0x35, 0x7c, 0x6c, 0x4f, 0x9d, 0xd7, 0x9e, 0x3f, 0xd9, 
+   0x58, 0x39, 0xef, 0xa1, 0xbd, 0x5f, 0x90, 0x8d, 0x95, 0xe6, 0xe6, 0xd0, 
+   0xde, 0x27, 0xe0, 0x63, 0x9d, 0x03, 0xe5, 0xa3, 0x0f, 0xed, 0x35, 0x5a, 
+   0x3e, 0x58, 0xf9, 0x6a, 0x7b, 0x67, 0x20, 0x1f, 0xac, 0x7c, 0x76, 0x43, 
+   0xd1, 0xa2, 0x83, 0xd5, 0xd6, 0x03, 0x0d, 0x85, 0xcb, 0x71, 0xb5, 0xd1, 
+   0x0d, 0xc5, 0x8b, 0x8d, 0x96, 0x1b, 0xf1, 0xb0, 0x89, 0xf2, 0xea, 0x9d, 
+   0xcb, 0xf8, 0xa4, 0x81, 0xb7, 0xdb, 0xbb, 0x51, 0xc6, 0x35, 0x71, 0x76, 
+   0xd5, 0xf9, 0x1a, 0x78, 0xba, 0x67, 0xda, 0x84, 0x8d, 0x1c, 0xdc, 0xbe, 
+   0x32, 0x25, 0x3b, 0x2a, 0x69, 0x34, 0xf4, 0x46, 0x1d, 0xdb, 0xc4, 0x36, 
+   0x3b, 0x5d, 0x75, 0x64, 0x03, 0xbb, 0xec, 0x74, 0xb5, 0x39, 0x9b, 0xc4, 
+   0x4c, 0xfa, 0x9c, 0x4d, 0x62, 0xa6, 0xd2, 0xa4, 0x8d, 0x82, 0xfa, 0xbe, 
+   0x3e, 0x6d, 0x93, 0x90, 0x89, 0x8d, 0xd5, 0x26, 0x6e, 0x22, 0xb9, 0x2d, 
+   0x57, 0x1d, 0xd9, 0x40, 0x96, 0x5a, 0xae, 0x36, 0x67, 0x13, 0x61, 0xd2, 
+   0xe6, 0x6c, 0x12, 0x82, 0x9f, 0xe9, 0x93, 0x36, 0x0a, 0xc0, 0x99, 0x87, 
+   0xa8, 0x8e, 0x6d, 0xe4, 0xe4, 0xf5, 0x4b, 0x13, 0x37, 0x8a, 0xc1, 0x79, 
+   0xfe, 0x42, 0x1b, 0xde, 0xce, 0x18, 0xf4, 0xf3, 0x3e, 0x4b, 0xe8, 0xfd, 
+   0xeb, 0x87, 0x9f, 0xf7, 0xbf, 0xcd, 0x9f, 0xd6, 0xab, 0x7f, 0xfd, 0xf0, 
+   0xff, 0x01, 0x87, 0x8b, 0x46, 0xb8, 
 };
diff --git a/lib/mesa/src/broadcom/cle/v3dx_pack.h b/lib/mesa/src/broadcom/cle/v3dx_pack.h
new file mode 100644
index 000000000..5762e5aaa
--- /dev/null
+++ b/lib/mesa/src/broadcom/cle/v3dx_pack.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright © 2015 Intel Corporation
+ * Copyright © 2015 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef V3DX_PACK_H
+#define V3DX_PACK_H
+
+#ifndef V3D_VERSION
+#  error "The V3D_VERSION macro must be defined"
+#endif
+
+#if (V3D_VERSION == 21)
+#  include "cle/v3d_packet_v21_pack.h"
+#elif (V3D_VERSION == 33)
+#  include "cle/v3d_packet_v33_pack.h"
+#elif (V3D_VERSION == 41)
+#  include "cle/v3d_packet_v41_pack.h"
+#elif (V3D_VERSION == 42)
+#  include "cle/v3d_packet_v42_pack.h"
+#else
+#  error "Need to add a pack header include for this v3d version"
+#endif
+
+#endif /* V3DX_PACK_H */
diff --git a/lib/mesa/src/broadcom/clif/clif_dump.c b/lib/mesa/src/broadcom/clif/clif_dump.c
index 339fc3835..2bc73e60f 100644
--- a/lib/mesa/src/broadcom/clif/clif_dump.c
+++ b/lib/mesa/src/broadcom/clif/clif_dump.c
@@ -24,61 +24,15 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include "drm-uapi/v3d_drm.h"
 #include "clif_dump.h"
+#include "clif_private.h"
 #include "util/list.h"
 #include "util/ralloc.h"
 
 #include "broadcom/cle/v3d_decoder.h"
 
-#define __gen_user_data void
-#define __gen_address_type uint32_t
-#define __gen_address_offset(reloc) (*reloc)
-#define __gen_emit_reloc(cl, reloc)
-#define __gen_unpack_address(cl, s, e) (__gen_unpack_uint(cl, s, e) << (31 - (e - s)))
-
-enum reloc_worklist_type {
-        reloc_gl_shader_state,
-};
-
-struct reloc_worklist_entry {
-        struct list_head link;
-
-        enum reloc_worklist_type type;
-        uint32_t addr;
-
-        union {
-                struct {
-                        uint32_t num_attrs;
-                } shader_state;
-        };
-};
-
-struct clif_dump {
-        const struct v3d_device_info *devinfo;
-        bool (*lookup_vaddr)(void *data, uint32_t addr, void **vaddr);
-        FILE *out;
-        /* Opaque data from the caller that is passed to the callbacks. */
-        void *data;
-
-        struct v3d_spec *spec;
-
-        /* List of struct reloc_worklist_entry */
-        struct list_head worklist;
-};
-
-static void
-out(struct clif_dump *clif, const char *fmt, ...)
-{
-        va_list args;
-
-        va_start(args, fmt);
-        vfprintf(clif->out, fmt, args);
-        va_end(args);
-}
-
-#include "broadcom/cle/v3d_packet_v33_pack.h"
-
-static struct reloc_worklist_entry *
+struct reloc_worklist_entry *
 clif_dump_add_address_to_worklist(struct clif_dump *clif,
                                   enum reloc_worklist_type type,
                                   uint32_t addr)
@@ -98,17 +52,14 @@ clif_dump_add_address_to_worklist(struct clif_dump *clif,
 
 struct clif_dump *
 clif_dump_init(const struct v3d_device_info *devinfo,
-               FILE *out,
-               bool (*lookup_vaddr)(void *data, uint32_t addr, void **vaddr),
-               void *data)
+               FILE *out, bool pretty)
 {
         struct clif_dump *clif = rzalloc(NULL, struct clif_dump);
 
         clif->devinfo = devinfo;
-        clif->lookup_vaddr = lookup_vaddr;
         clif->out = out;
-        clif->data = data;
         clif->spec = v3d_spec_load(devinfo);
+        clif->pretty = pretty;
 
         list_inithead(&clif->worklist);
 
@@ -121,85 +72,92 @@ clif_dump_destroy(struct clif_dump *clif)
         ralloc_free(clif);
 }
 
-#define out_uint(_clif, field) out(_clif, "    /* %s = */ %u\n",        \
-                            #field,  values-> field);
+struct clif_bo *
+clif_lookup_bo(struct clif_dump *clif, uint32_t addr)
+{
+        for (int i = 0; i < clif->bo_count; i++) {
+                struct clif_bo *bo = &clif->bo[i];
+
+                if (addr >= bo->offset &&
+                    addr < bo->offset + bo->size) {
+                        return bo;
+                }
+        }
+
+        return NULL;
+}
 
 static bool
-clif_dump_packet(struct clif_dump *clif, uint32_t offset, const uint8_t *cl,
-                 uint32_t *size)
+clif_lookup_vaddr(struct clif_dump *clif, uint32_t addr, void **vaddr)
 {
-        struct v3d_group *inst = v3d_spec_find_instruction(clif->spec, cl);
-        if (!inst) {
-                out(clif, "0x%08x: Unknown packet %d!\n", offset, *cl);
+        struct clif_bo *bo = clif_lookup_bo(clif, addr);
+        if (!bo)
                 return false;
-        }
 
-        *size = v3d_group_get_length(inst);
+        *vaddr = bo->vaddr + addr - bo->offset;
+        return true;
+}
 
-        out(clif, "%s\n", v3d_group_get_name(inst));
-        v3d_print_group(clif->out, inst, 0, cl, "");
+#define out_uint(_clif, field) out(_clif, "    /* %s = */ %u\n",        \
+                            #field,  values-> field);
 
-        switch (*cl) {
-        case V3D33_GL_SHADER_STATE_opcode: {
-                struct V3D33_GL_SHADER_STATE values;
-                V3D33_GL_SHADER_STATE_unpack(cl, &values);
+static bool
+clif_dump_packet(struct clif_dump *clif, uint32_t offset, const uint8_t *cl,
+                 uint32_t *size, bool reloc_mode)
+{
+        if (clif->devinfo->ver >= 41)
+                return v3d41_clif_dump_packet(clif, offset, cl, size, reloc_mode);
+        else
+                return v3d33_clif_dump_packet(clif, offset, cl, size, reloc_mode);
+}
 
-                struct reloc_worklist_entry *reloc =
-                        clif_dump_add_address_to_worklist(clif,
-                                                          reloc_gl_shader_state,
-                                                          values.address);
-                if (reloc) {
-                        reloc->shader_state.num_attrs =
-                                values.number_of_attribute_arrays;
-                }
-                return true;
+static uint32_t
+clif_dump_cl(struct clif_dump *clif, uint32_t start, uint32_t end,
+             bool reloc_mode)
+{
+        struct clif_bo *bo = clif_lookup_bo(clif, start);
+        if (!bo) {
+                out(clif, "Failed to look up address 0x%08x\n",
+                    start);
+                return 0;
         }
 
-        case V3D33_STORE_MULTI_SAMPLE_RESOLVED_TILE_COLOR_BUFFER_EXTENDED_opcode: {
-                struct V3D33_STORE_MULTI_SAMPLE_RESOLVED_TILE_COLOR_BUFFER_EXTENDED values;
-                V3D33_STORE_MULTI_SAMPLE_RESOLVED_TILE_COLOR_BUFFER_EXTENDED_unpack(cl, &values);
+        void *start_vaddr = bo->vaddr + start - bo->offset;
 
-                if (values.last_tile_of_frame)
-                        return false;
-                break;
+        /* The end address is optional (for example, a BRANCH instruction
+         * won't set an end), but is used for BCL/RCL termination.
+         */
+        void *end_vaddr = NULL;
+        if (end && !clif_lookup_vaddr(clif, end, &end_vaddr)) {
+                out(clif, "Failed to look up address 0x%08x\n",
+                    end);
+                return 0;
         }
 
-        case V3D33_TRANSFORM_FEEDBACK_ENABLE_opcode: {
-                struct V3D33_TRANSFORM_FEEDBACK_ENABLE values;
-                V3D33_TRANSFORM_FEEDBACK_ENABLE_unpack(cl, &values);
-                struct v3d_group *spec = v3d_spec_find_struct(clif->spec,
-                                                              "Transform Feedback Output Data Spec");
-                struct v3d_group *addr = v3d_spec_find_struct(clif->spec,
-                                                              "Transform Feedback Output Address");
-                assert(spec);
-                assert(addr);
-
-                cl += *size;
-
-                for (int i = 0; i < values.number_of_16_bit_output_data_specs_following; i++) {
-                        v3d_print_group(clif->out, spec, 0, cl, "");
-                        cl += v3d_group_get_length(spec);
-                        *size += v3d_group_get_length(spec);
-                }
+        if (!reloc_mode)
+                out(clif, "@format ctrllist  /* [%s+0x%08x] */\n",
+                    bo->name, start - bo->offset);
 
-                for (int i = 0; i < values.number_of_32_bit_output_buffer_address_following; i++) {
-                        v3d_print_group(clif->out, addr, 0, cl, "");
-                        cl += v3d_group_get_length(addr);
-                        *size += v3d_group_get_length(addr);
-                }
-                break;
-        }
+        uint32_t size;
+        uint8_t *cl = start_vaddr;
+        while (clif_dump_packet(clif, start, cl, &size, reloc_mode)) {
+                cl += size;
+                start += size;
 
-        case V3D33_HALT_opcode:
-                return false;
+                if (cl == end_vaddr)
+                        break;
         }
 
-        return true;
+        return (void *)cl - bo->vaddr;
 }
 
-static void
+/* Walks the worklist, parsing the relocs for any memory regions that might
+ * themselves have additional relocations.
+ */
+static uint32_t
 clif_dump_gl_shader_state_record(struct clif_dump *clif,
-                                 struct reloc_worklist_entry *reloc, void *vaddr)
+                                 struct reloc_worklist_entry *reloc,
+                                 void *vaddr)
 {
         struct v3d_group *state = v3d_spec_find_struct(clif->spec,
                                                        "GL Shader State Record");
@@ -207,75 +165,280 @@ clif_dump_gl_shader_state_record(struct clif_dump *clif,
                                                       "GL Shader State Attribute Record");
         assert(state);
         assert(attr);
+        uint32_t offset = 0;
 
-        out(clif, "GL Shader State Record at 0x%08x\n", reloc->addr);
-        v3d_print_group(clif->out, state, 0, vaddr, "");
-        vaddr += v3d_group_get_length(state);
+        out(clif, "@format shadrec_gl_main\n");
+        v3d_print_group(clif, state, 0, vaddr + offset);
+        offset += v3d_group_get_length(state);
 
         for (int i = 0; i < reloc->shader_state.num_attrs; i++) {
-                out(clif, "  Attribute %d\n", i);
-                v3d_print_group(clif->out, attr, 0, vaddr, "");
-                vaddr += v3d_group_get_length(attr);
+                out(clif, "@format shadrec_gl_attr /* %d */\n", i);
+                v3d_print_group(clif, attr, 0, vaddr + offset);
+                offset += v3d_group_get_length(attr);
         }
+
+        return offset;
 }
 
 static void
 clif_process_worklist(struct clif_dump *clif)
 {
-        while (!list_empty(&clif->worklist)) {
-                struct reloc_worklist_entry *reloc =
-                        list_first_entry(&clif->worklist,
-                                         struct reloc_worklist_entry, link);
-                list_del(&reloc->link);
-
+        list_for_each_entry_safe(struct reloc_worklist_entry, reloc,
+                                 &clif->worklist, link) {
                 void *vaddr;
-                if (!clif->lookup_vaddr(clif->data, reloc->addr, &vaddr)) {
+                if (!clif_lookup_vaddr(clif, reloc->addr, &vaddr)) {
                         out(clif, "Failed to look up address 0x%08x\n",
                             reloc->addr);
                         continue;
                 }
 
                 switch (reloc->type) {
+                case reloc_cl:
+                        clif_dump_cl(clif, reloc->addr, reloc->cl.end, true);
+                        break;
+
                 case reloc_gl_shader_state:
-                        clif_dump_gl_shader_state_record(clif, reloc, vaddr);
+                        break;
+                case reloc_generic_tile_list:
+                        clif_dump_cl(clif, reloc->addr,
+                                     reloc->generic_tile_list.end, true);
                         break;
                 }
-                out(clif, "\n");
         }
 }
 
-void
-clif_dump_add_cl(struct clif_dump *clif, uint32_t start, uint32_t end)
+static int
+worklist_entry_compare(const void *a, const void *b)
 {
-        uint32_t size;
+        return ((*(struct reloc_worklist_entry **)a)->addr -
+                (*(struct reloc_worklist_entry **)b)->addr);
+}
 
-        void *start_vaddr;
-        if (!clif->lookup_vaddr(clif->data, start, &start_vaddr)) {
-                out(clif, "Failed to look up address 0x%08x\n",
-                    start);
-                return;
+static bool
+clif_dump_if_blank(struct clif_dump *clif, struct clif_bo *bo,
+                   uint32_t start, uint32_t end)
+{
+        for (int i = start; i < end; i++) {
+                if (((uint8_t *)bo->vaddr)[i] != 0)
+                        return false;
         }
 
-        /* The end address is optional (for example, a BRANCH instruction
-         * won't set an end), but is used for BCL/RCL termination.
-         */
-        void *end_vaddr = NULL;
-        if (end && !clif->lookup_vaddr(clif->data, end, &end_vaddr)) {
-                out(clif, "Failed to look up address 0x%08x\n",
-                    end);
+        out(clif, "\n");
+        out(clif, "@format blank %d /* [%s+0x%08x..0x%08x] */\n", end - start,
+            bo->name, start, end - 1);
+        return true;
+}
+
+/* Dumps the binary data in the BO from start to end (relative to the start of
+ * the BO).
+ */
+static void
+clif_dump_binary(struct clif_dump *clif, struct clif_bo *bo,
+                 uint32_t start, uint32_t end)
+{
+        if (start == end)
                 return;
+
+        if (clif_dump_if_blank(clif, bo, start, end))
+                return;
+
+        out(clif, "@format binary /* [%s+0x%08x] */\n",
+            bo->name, start);
+
+        uint32_t offset = start;
+        int dumped_in_line = 0;
+        while (offset < end) {
+                if (clif_dump_if_blank(clif, bo, offset, end))
+                        return;
+
+                if (end - offset >= 4) {
+                        out(clif, "0x%08x ", *(uint32_t *)(bo->vaddr + offset));
+                        offset += 4;
+                } else {
+                        out(clif, "0x%02x ", *(uint8_t *)(bo->vaddr + offset));
+                        offset++;
+                }
+
+                if (++dumped_in_line == 8) {
+                        out(clif, "\n");
+                        dumped_in_line = 0;
+                }
         }
+        if (dumped_in_line)
+                out(clif, "\n");
+}
 
-        uint8_t *cl = start_vaddr;
-        while (clif_dump_packet(clif, start, cl, &size)) {
-                cl += size;
-                start += size;
+/* Walks the list of relocations, dumping each buffer's contents (using our
+ * codegenned dump routines for pretty printing, and most importantly proper
+ * address references so that the CLIF parser can relocate buffers).
+ */
+static void
+clif_dump_buffers(struct clif_dump *clif)
+{
+        int num_relocs = 0;
+        list_for_each_entry(struct reloc_worklist_entry, reloc,
+                            &clif->worklist, link) {
+                num_relocs++;
+        }
+        struct reloc_worklist_entry **relocs =
+                ralloc_array(clif, struct reloc_worklist_entry *, num_relocs);
+        int i = 0;
+        list_for_each_entry(struct reloc_worklist_entry, reloc,
+                            &clif->worklist, link) {
+                relocs[i++] = reloc;
+        }
+        qsort(relocs, num_relocs, sizeof(*relocs), worklist_entry_compare);
 
-                if (cl == end_vaddr)
+        struct clif_bo *bo = NULL;
+        uint32_t offset = 0;
+
+        for (i = 0; i < num_relocs; i++) {
+                struct reloc_worklist_entry *reloc = relocs[i];
+                struct clif_bo *new_bo = clif_lookup_bo(clif, reloc->addr);
+
+                if (!new_bo) {
+                        out(clif, "Failed to look up address 0x%08x\n",
+                            reloc->addr);
+                        continue;
+                }
+
+                if (new_bo != bo) {
+                        if (bo) {
+                                /* Finish out the last of the last BO. */
+                                clif_dump_binary(clif, bo,
+                                                 offset,
+                                                 bo->size);
+                        }
+
+                        out(clif, "\n");
+                        out(clif, "@buffer %s\n", new_bo->name);
+                        bo = new_bo;
+                        offset = 0;
+                        bo->dumped = true;
+                }
+
+                int reloc_offset = reloc->addr - bo->offset;
+                if (offset != reloc_offset)
+                        clif_dump_binary(clif, bo, offset, reloc_offset);
+                offset = reloc_offset;
+
+                switch (reloc->type) {
+                case reloc_cl:
+                        offset = clif_dump_cl(clif, reloc->addr, reloc->cl.end,
+                                              false);
+                        out(clif, "\n");
+                        break;
+
+                case reloc_gl_shader_state:
+                        offset += clif_dump_gl_shader_state_record(clif,
+                                                                   reloc,
+                                                                   bo->vaddr +
+                                                                   offset);
+                        break;
+                case reloc_generic_tile_list:
+                        offset = clif_dump_cl(clif, reloc->addr,
+                                              reloc->generic_tile_list.end,
+                                              false);
                         break;
+                }
+                out(clif, "\n");
         }
 
-        out(clif, "\n");
+        if (bo) {
+                clif_dump_binary(clif, bo, offset, bo->size);
+        }
+
+        /* For any BOs that didn't have relocations, just dump them raw. */
+        for (int i = 0; i < clif->bo_count; i++) {
+                bo = &clif->bo[i];
+                if (bo->dumped)
+                        continue;
+                out(clif, "@buffer %s\n", bo->name);
+                clif_dump_binary(clif, bo, 0, bo->size);
+                out(clif, "\n");
+        }
+}
+
+void
+clif_dump_add_cl(struct clif_dump *clif, uint32_t start, uint32_t end)
+{
+        struct reloc_worklist_entry *entry =
+                clif_dump_add_address_to_worklist(clif, reloc_cl, start);
+
+        entry->cl.end = end;
+}
+
+static int
+clif_bo_offset_compare(const void *a, const void *b)
+{
+        return ((struct clif_bo *)a)->offset - ((struct clif_bo *)b)->offset;
+}
 
+void
+clif_dump(struct clif_dump *clif, const struct drm_v3d_submit_cl *submit)
+{
+        clif_dump_add_cl(clif, submit->bcl_start, submit->bcl_end);
+        clif_dump_add_cl(clif, submit->rcl_start, submit->rcl_end);
+
+        qsort(clif->bo, clif->bo_count, sizeof(clif->bo[0]),
+              clif_bo_offset_compare);
+
+        /* A buffer needs to be defined before we can emit a CLIF address
+         * referencing it, so emit them all now.
+         */
+        for (int i = 0; i < clif->bo_count; i++) {
+                out(clif, "@createbuf_aligned 4096 %s\n", clif->bo[i].name);
+        }
+
+        /* Walk the worklist figuring out the locations of structs based on
+         * the CL contents.
+         */
         clif_process_worklist(clif);
+
+        /* Dump the contents of the buffers using the relocations we found to
+         * pretty-print structures.
+         */
+        clif_dump_buffers(clif);
+
+        out(clif, "@add_bin 0\n  ");
+        out_address(clif, submit->bcl_start);
+        out(clif, "\n  ");
+        out_address(clif, submit->bcl_end);
+        out(clif, "\n  ");
+        out_address(clif, submit->qma);
+        out(clif, "\n  %d\n  ", submit->qms);
+        out_address(clif, submit->qts);
+        out(clif, "\n");
+        out(clif, "@wait_bin_all_cores\n");
+
+        out(clif, "@add_render 0\n  ");
+        out_address(clif, submit->rcl_start);
+        out(clif, "\n  ");
+        out_address(clif, submit->rcl_end);
+        out(clif, "\n  ");
+        out_address(clif, submit->qma);
+        out(clif, "\n");
+        out(clif, "@wait_render_all_cores\n");
+}
+
+void
+clif_dump_add_bo(struct clif_dump *clif, const char *name,
+                 uint32_t offset, uint32_t size, void *vaddr)
+{
+        if (clif->bo_count >= clif->bo_array_size) {
+                clif->bo_array_size = MAX2(4, clif->bo_array_size * 2);
+                clif->bo = reralloc(clif, clif->bo, struct clif_bo,
+                                    clif->bo_array_size);
+        }
+
+        /* CLIF relocs use the buffer name, so make sure they're unique. */
+        for (int i = 0; i < clif->bo_count; i++)
+                assert(strcmp(clif->bo[i].name, name) != 0);
+
+        clif->bo[clif->bo_count].name = ralloc_strdup(clif, name);
+        clif->bo[clif->bo_count].offset = offset;
+        clif->bo[clif->bo_count].size = size;
+        clif->bo[clif->bo_count].vaddr = vaddr;
+        clif->bo[clif->bo_count].dumped = false;
+        clif->bo_count++;
 }
diff --git a/lib/mesa/src/broadcom/clif/clif_dump.h b/lib/mesa/src/broadcom/clif/clif_dump.h
index d46cc8471..8de3a2cbe 100644
--- a/lib/mesa/src/broadcom/clif/clif_dump.h
+++ b/lib/mesa/src/broadcom/clif/clif_dump.h
@@ -29,14 +29,15 @@
 
 struct v3d_device_info;
 struct clif_dump;
+struct drm_v3d_submit_cl;
 
 struct clif_dump *clif_dump_init(const struct v3d_device_info *devinfo,
-                                 FILE *output,
-                                 bool (*lookup_vaddr)(void *data, uint32_t addr,
-                                                      void **vaddr),
-                                 void *data);
+                                 FILE *output, bool pretty);
+void clif_dump(struct clif_dump *clif, const struct drm_v3d_submit_cl *submit);
 void clif_dump_destroy(struct clif_dump *clif);
 
+void clif_dump_add_bo(struct clif_dump *clif, const char *name,
+                      uint32_t offset, uint32_t size, void *vaddr);
 void clif_dump_add_cl(struct clif_dump *clif, uint32_t start, uint32_t end);
 
 #endif
diff --git a/lib/mesa/src/broadcom/clif/clif_private.h b/lib/mesa/src/broadcom/clif/clif_private.h
new file mode 100644
index 000000000..597d0b506
--- /dev/null
+++ b/lib/mesa/src/broadcom/clif/clif_private.h
@@ -0,0 +1,123 @@
+/*
+ * Copyright © 2016-2018 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef CLIF_PRIVATE_H
+#define CLIF_PRIVATE_H
+
+#include <stdint.h>
+#include <stdarg.h>
+#include "util/list.h"
+
+struct clif_bo {
+        const char *name;
+        uint32_t offset;
+        uint32_t size;
+        void *vaddr;
+        bool dumped;
+};
+
+struct clif_dump {
+        const struct v3d_device_info *devinfo;
+        FILE *out;
+
+        struct v3d_spec *spec;
+
+        /* List of struct reloc_worklist_entry */
+        struct list_head worklist;
+
+        struct clif_bo *bo;
+        int bo_count;
+        int bo_array_size;
+
+        /**
+         * Flag to switch from CLIF ABI to slightly more human-readable
+         * output.
+         */
+        bool pretty;
+};
+
+enum reloc_worklist_type {
+        reloc_cl,
+        reloc_gl_shader_state,
+        reloc_generic_tile_list,
+};
+
+struct reloc_worklist_entry {
+        struct list_head link;
+
+        enum reloc_worklist_type type;
+        uint32_t addr;
+
+        union {
+                struct {
+                        uint32_t end;
+                } cl;
+                struct {
+                        uint32_t num_attrs;
+                } shader_state;
+                struct {
+                        uint32_t end;
+                } generic_tile_list;
+        };
+};
+
+struct clif_bo *
+clif_lookup_bo(struct clif_dump *clif, uint32_t addr);
+
+struct reloc_worklist_entry *
+clif_dump_add_address_to_worklist(struct clif_dump *clif,
+                                  enum reloc_worklist_type type,
+                                  uint32_t addr);
+
+bool v3d33_clif_dump_packet(struct clif_dump *clif, uint32_t offset,
+                            const uint8_t *cl, uint32_t *size, bool reloc_mode);
+bool v3d41_clif_dump_packet(struct clif_dump *clif, uint32_t offset,
+                            const uint8_t *cl, uint32_t *size, bool reloc_mode);
+bool v3d42_clif_dump_packet(struct clif_dump *clif, uint32_t offset,
+                            const uint8_t *cl, uint32_t *size, bool reloc_mode);
+
+static inline void
+out(struct clif_dump *clif, const char *fmt, ...)
+{
+        va_list args;
+
+        va_start(args, fmt);
+        vfprintf(clif->out, fmt, args);
+        va_end(args);
+}
+
+static inline void
+out_address(struct clif_dump *clif, uint32_t addr)
+{
+        struct clif_bo *bo = clif_lookup_bo(clif, addr);
+        if (bo) {
+                out(clif, "[%s+0x%08x] /* 0x%08x */",
+                    bo->name, addr - bo->offset, addr);
+        } else if (addr) {
+                out(clif, "/* XXX: BO unknown */ 0x%08x", addr);
+        } else {
+                out(clif, "[null]");
+        }
+}
+
+#endif /* CLIF_PRIVATE_H */
diff --git a/lib/mesa/src/broadcom/clif/v3dx_dump.c b/lib/mesa/src/broadcom/clif/v3dx_dump.c
new file mode 100644
index 000000000..9cf59f889
--- /dev/null
+++ b/lib/mesa/src/broadcom/clif/v3dx_dump.c
@@ -0,0 +1,174 @@
+/*
+ * Copyright © 2016-2018 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <ctype.h>
+#include <stdlib.h>
+#include <string.h>
+#include "util/macros.h"
+#include "broadcom/cle/v3d_decoder.h"
+#include "clif_dump.h"
+#include "clif_private.h"
+
+#define __gen_user_data void
+#define __gen_address_type uint32_t
+#define __gen_address_offset(reloc) (*reloc)
+#define __gen_emit_reloc(cl, reloc)
+#define __gen_unpack_address(cl, s, e) (__gen_unpack_uint(cl, s, e) << (31 - (e - s)))
+#include "broadcom/cle/v3dx_pack.h"
+#include "broadcom/common/v3d_macros.h"
+
+static char *
+clif_name(const char *xml_name)
+{
+        char *name = malloc(strlen(xml_name) + 1);
+
+        int j = 0;
+        for (int i = 0; i < strlen(xml_name); i++) {
+                if (xml_name[i] == ' ') {
+                        name[j++] = '_';
+                } else if (xml_name[i] == '(' || xml_name[i] == ')') {
+                        /* skip */
+                } else {
+                        name[j++] = toupper(xml_name[i]);
+                }
+        }
+        name[j++] = 0;
+
+        return name;
+}
+
+bool
+v3dX(clif_dump_packet)(struct clif_dump *clif, uint32_t offset,
+                       const uint8_t *cl, uint32_t *size, bool reloc_mode)
+{
+        struct v3d_group *inst = v3d_spec_find_instruction(clif->spec, cl);
+        if (!inst) {
+                out(clif, "0x%08x: Unknown packet %d!\n", offset, *cl);
+                return false;
+        }
+
+        *size = v3d_group_get_length(inst);
+
+        if (!reloc_mode) {
+                char *name = clif_name(v3d_group_get_name(inst));
+                out(clif, "%s\n", name);
+                free(name);
+                v3d_print_group(clif, inst, 0, cl);
+        }
+
+        switch (*cl) {
+        case V3DX(GL_SHADER_STATE_opcode): {
+                struct V3DX(GL_SHADER_STATE) values;
+                V3DX(GL_SHADER_STATE_unpack)(cl, &values);
+
+                if (reloc_mode) {
+                        struct reloc_worklist_entry *reloc =
+                                clif_dump_add_address_to_worklist(clif,
+                                                                  reloc_gl_shader_state,
+                                                                  values.address);
+                        if (reloc) {
+                                reloc->shader_state.num_attrs =
+                                        values.number_of_attribute_arrays;
+                        }
+                }
+                return true;
+        }
+
+#if V3D_VERSION < 40
+        case V3DX(STORE_MULTI_SAMPLE_RESOLVED_TILE_COLOR_BUFFER_EXTENDED_opcode): {
+                struct V3DX(STORE_MULTI_SAMPLE_RESOLVED_TILE_COLOR_BUFFER_EXTENDED) values;
+                V3DX(STORE_MULTI_SAMPLE_RESOLVED_TILE_COLOR_BUFFER_EXTENDED_unpack)(cl, &values);
+
+                if (values.last_tile_of_frame)
+                        return false;
+                break;
+        }
+#endif /* V3D_VERSION < 40 */
+
+#if V3D_VERSION > 40
+        case V3DX(TRANSFORM_FEEDBACK_SPECS_opcode): {
+                struct V3DX(TRANSFORM_FEEDBACK_SPECS) values;
+                V3DX(TRANSFORM_FEEDBACK_SPECS_unpack)(cl, &values);
+                struct v3d_group *spec = v3d_spec_find_struct(clif->spec,
+                                                              "Transform Feedback Output Data Spec");
+                assert(spec);
+
+                cl += *size;
+
+                for (int i = 0; i < values.number_of_16_bit_output_data_specs_following; i++) {
+                        if (!reloc_mode)
+                                v3d_print_group(clif, spec, 0, cl);
+                        cl += v3d_group_get_length(spec);
+                        *size += v3d_group_get_length(spec);
+                }
+                if (!reloc_mode)
+                        out(clif, "@format ctrllist\n");
+                break;
+        }
+#else /* V3D_VERSION < 40 */
+        case V3DX(TRANSFORM_FEEDBACK_ENABLE_opcode): {
+                struct V3DX(TRANSFORM_FEEDBACK_ENABLE) values;
+                V3DX(TRANSFORM_FEEDBACK_ENABLE_unpack)(cl, &values);
+                struct v3d_group *spec = v3d_spec_find_struct(clif->spec,
+                                                              "Transform Feedback Output Data Spec");
+                struct v3d_group *addr = v3d_spec_find_struct(clif->spec,
+                                                              "Transform Feedback Output Address");
+                assert(spec);
+                assert(addr);
+
+                cl += *size;
+
+                for (int i = 0; i < values.number_of_16_bit_output_data_specs_following; i++) {
+                        if (!reloc_mode)
+                                v3d_print_group(clif, spec, 0, cl);
+                        cl += v3d_group_get_length(spec);
+                        *size += v3d_group_get_length(spec);
+                }
+
+                for (int i = 0; i < values.number_of_32_bit_output_buffer_address_following; i++) {
+                        if (!reloc_mode)
+                                v3d_print_group(clif, addr, 0, cl);
+                        cl += v3d_group_get_length(addr);
+                        *size += v3d_group_get_length(addr);
+                }
+                break;
+        }
+#endif /* V3D_VERSION < 40 */
+
+        case V3DX(START_ADDRESS_OF_GENERIC_TILE_LIST_opcode): {
+                struct V3DX(START_ADDRESS_OF_GENERIC_TILE_LIST) values;
+                V3DX(START_ADDRESS_OF_GENERIC_TILE_LIST_unpack)(cl, &values);
+                struct reloc_worklist_entry *reloc =
+                        clif_dump_add_address_to_worklist(clif,
+                                                          reloc_generic_tile_list,
+                                                          values.start);
+                reloc->generic_tile_list.end = values.end;
+                break;
+        }
+
+        case V3DX(HALT_opcode):
+                return false;
+        }
+
+        return true;
+}
diff --git a/lib/mesa/src/broadcom/common/v3d_debug.c b/lib/mesa/src/broadcom/common/v3d_debug.c
index 630bfe0fc..97404448e 100644
--- a/lib/mesa/src/broadcom/common/v3d_debug.c
+++ b/lib/mesa/src/broadcom/common/v3d_debug.c
@@ -41,6 +41,7 @@ uint32_t V3D_DEBUG = 0;
 
 static const struct debug_control debug_control[] = {
         { "cl",          V3D_DEBUG_CL},
+        { "clif",        V3D_DEBUG_CLIF},
         { "qpu",         V3D_DEBUG_QPU},
         { "vir",         V3D_DEBUG_VIR},
         { "nir",         V3D_DEBUG_NIR},
@@ -52,6 +53,7 @@ static const struct debug_control debug_control[] = {
         { "fs",          V3D_DEBUG_FS},
         { "vs",          V3D_DEBUG_VS},
         { "cs",          V3D_DEBUG_CS},
+        { "always_flush", V3D_DEBUG_ALWAYS_FLUSH},
         { NULL,    0 }
 };
 
diff --git a/lib/mesa/src/broadcom/common/v3d_debug.h b/lib/mesa/src/broadcom/common/v3d_debug.h
index bdb951854..d9f5255e2 100644
--- a/lib/mesa/src/broadcom/common/v3d_debug.h
+++ b/lib/mesa/src/broadcom/common/v3d_debug.h
@@ -54,10 +54,15 @@ extern uint32_t V3D_DEBUG;
 #define V3D_DEBUG_PERF			(1 << 10)
 #define V3D_DEBUG_NORAST		(1 << 11)
 #define V3D_DEBUG_ALWAYS_FLUSH		(1 << 12)
+#define V3D_DEBUG_CLIF			(1 << 13)
 
 #ifdef HAVE_ANDROID_PLATFORM
 #define LOG_TAG "BROADCOM-MESA"
+#if ANDROID_API_LEVEL >= 26
+#include <log/log.h>
+#else
 #include <cutils/log.h>
+#endif /* use log/log.h start from android 8 major version */
 #ifndef ALOGW
 #define ALOGW LOGW
 #endif
diff --git a/lib/mesa/src/broadcom/common/v3d_macros.h b/lib/mesa/src/broadcom/common/v3d_macros.h
new file mode 100644
index 000000000..fe8939820
--- /dev/null
+++ b/lib/mesa/src/broadcom/common/v3d_macros.h
@@ -0,0 +1,48 @@
+/*
+ * Copyright © 2015 Intel Corporation
+ * Copyright © 2015 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef V3DX_MACROS_H
+#define V3DX_MACROS_H
+
+#ifndef V3D_VERSION
+#  error "The V3D_VERSION macro must be defined"
+#endif
+
+#if (V3D_VERSION == 21)
+#  define V3DX(x) V3D21_##x
+#  define v3dX(x) v3d21_##x
+#elif (V3D_VERSION == 33)
+#  define V3DX(x) V3D33_##x
+#  define v3dX(x) v3d33_##x
+#elif (V3D_VERSION == 41)
+#  define V3DX(x) V3D41_##x
+#  define v3dX(x) v3d41_##x
+#elif (V3D_VERSION == 42)
+#  define V3DX(x) V3D42_##x
+#  define v3dX(x) v3d42_##x
+#else
+#  error "Need to add prefixing macros for this v3d version"
+#endif
+
+#endif /* V3DX_MACROS_H */
diff --git a/lib/mesa/src/broadcom/compiler/meson.build b/lib/mesa/src/broadcom/compiler/meson.build
new file mode 100644
index 000000000..86ef365aa
--- /dev/null
+++ b/lib/mesa/src/broadcom/compiler/meson.build
@@ -0,0 +1,51 @@
+# Copyright © 2017 Broadcom
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+libbroadcom_compiler_files = files(
+  'nir_to_vir.c',
+  'vir.c',
+  'vir_dump.c',
+  'vir_live_variables.c',
+  'vir_lower_uniforms.c',
+  'vir_opt_copy_propagate.c',
+  'vir_opt_dead_code.c',
+  'vir_opt_small_immediates.c',
+  'vir_register_allocate.c',
+  'vir_to_qpu.c',
+  'qpu_schedule.c',
+  'qpu_validate.c',
+  'v3d33_tex.c',
+  'v3d40_tex.c',
+  'v3d33_vpm_setup.c',
+  'v3d_compiler.h',
+  'v3d_nir_lower_io.c',
+  'v3d_nir_lower_txf_ms.c',
+)
+
+libbroadcom_compiler = static_library(
+  ['broadcom_compiler', v3d_xml_pack],
+  libbroadcom_compiler_files,
+  include_directories : [inc_common, inc_broadcom],
+  c_args : [c_vis_args, no_override_init_args],
+  dependencies : [dep_libdrm, dep_valgrind, idep_nir_headers],
+  build_by_default : false,
+)
+
+v3d_libs += libbroadcom_compiler
diff --git a/lib/mesa/src/broadcom/compiler/nir_to_vir.c b/lib/mesa/src/broadcom/compiler/nir_to_vir.c
index 3b032b704..bb41c1be9 100644
--- a/lib/mesa/src/broadcom/compiler/nir_to_vir.c
+++ b/lib/mesa/src/broadcom/compiler/nir_to_vir.c
@@ -29,17 +29,9 @@
 #include "util/hash_table.h"
 #include "compiler/nir/nir.h"
 #include "compiler/nir/nir_builder.h"
+#include "common/v3d_device_info.h"
 #include "v3d_compiler.h"
 
-/* We don't do any address packing. */
-#define __gen_user_data void
-#define __gen_address_type uint32_t
-#define __gen_address_offset(reloc) (*reloc)
-#define __gen_emit_reloc(cl, reloc)
-#include "cle/v3d_packet_v33_pack.h"
-
-static struct qreg
-ntq_get_src(struct v3d_compile *c, nir_src src, int i);
 static void
 ntq_emit_cf_list(struct v3d_compile *c, struct exec_list *list);
 
@@ -64,18 +56,21 @@ resize_qreg_array(struct v3d_compile *c,
                 (*regs)[i] = c->undef;
 }
 
-static struct qreg
-vir_SFU(struct v3d_compile *c, int waddr, struct qreg src)
+void
+vir_emit_thrsw(struct v3d_compile *c)
 {
-        vir_FMOV_dest(c, vir_reg(QFILE_MAGIC, waddr), src);
-        return vir_FMOV(c, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_R4));
-}
+        if (c->threads == 1)
+                return;
 
-static struct qreg
-vir_LDTMU(struct v3d_compile *c)
-{
-        vir_NOP(c)->qpu.sig.ldtmu = true;
-        return vir_MOV(c, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_R4));
+        /* Always thread switch after each texture operation for now.
+         *
+         * We could do better by batching a bunch of texture fetches up and
+         * then doing one thread switch and collecting all their results
+         * afterward.
+         */
+        c->last_thrsw = vir_NOP(c);
+        c->last_thrsw->qpu.sig.thrsw = true;
+        c->last_thrsw_at_top_level = (c->execute.file == QFILE_NULL);
 }
 
 static struct qreg
@@ -117,6 +112,7 @@ indirect_uniform_load(struct v3d_compile *c, nir_intrinsic_instr *intr)
                      vir_uniform(c, QUNIFORM_UBO_ADDR, 0),
                      indirect_offset);
 
+        vir_emit_thrsw(c);
         return vir_LDTMU(c);
 }
 
@@ -144,7 +140,7 @@ ntq_init_ssa_def(struct v3d_compile *c, nir_ssa_def *def)
  * (knowing that the previous instruction doesn't depend on flags) and rewrite
  * its destination to be the NIR reg's destination
  */
-static void
+void
 ntq_store_dest(struct v3d_compile *c, nir_dest *dest, int chan,
                struct qreg result)
 {
@@ -195,14 +191,11 @@ ntq_store_dest(struct v3d_compile *c, nir_dest *dest, int chan,
                 if (c->execute.file != QFILE_NULL) {
                         last_inst->dst.index = qregs[chan].index;
 
-                        /* Set the flags to the current exec mask.  To insert
-                         * the flags push, we temporarily remove our SSA
-                         * instruction.
+                        /* Set the flags to the current exec mask.
                          */
-                        list_del(&last_inst->link);
+                        c->cursor = vir_before_inst(last_inst);
                         vir_PF(c, c->execute, V3D_QPU_PF_PUSHZ);
-                        list_addtail(&last_inst->link,
-                                     &c->cur_block->instructions);
+                        c->cursor = vir_after_inst(last_inst);
 
                         vir_set_cond(last_inst, V3D_QPU_COND_IFA);
                         last_inst->cond_is_exec_mask = true;
@@ -210,7 +203,7 @@ ntq_store_dest(struct v3d_compile *c, nir_dest *dest, int chan,
         }
 }
 
-static struct qreg
+struct qreg
 ntq_get_src(struct v3d_compile *c, nir_src src, int i)
 {
         struct hash_entry *entry;
@@ -233,7 +226,7 @@ static struct qreg
 ntq_get_alu_src(struct v3d_compile *c, nir_alu_instr *instr,
                 unsigned src)
 {
-        assert(util_is_power_of_two(instr->dest.write_mask));
+        assert(util_is_power_of_two_or_zero(instr->dest.write_mask));
         unsigned chan = ffs(instr->dest.write_mask) - 1;
         struct qreg r = ntq_get_src(c, instr->src[src].src,
                                     instr->src[src].swizzle[chan]);
@@ -244,21 +237,6 @@ ntq_get_alu_src(struct v3d_compile *c, nir_alu_instr *instr,
         return r;
 };
 
-static inline struct qreg
-vir_SAT(struct v3d_compile *c, struct qreg val)
-{
-        return vir_FMAX(c,
-                        vir_FMIN(c, val, vir_uniform_f(c, 1.0)),
-                        vir_uniform_f(c, 0.0));
-}
-
-static struct qreg
-ntq_umul(struct v3d_compile *c, struct qreg src0, struct qreg src1)
-{
-        vir_MULTOP(c, src0, src1);
-        return vir_UMUL24(c, src0, src1);
-}
-
 static struct qreg
 ntq_minify(struct v3d_compile *c, struct qreg size, struct qreg level)
 {
@@ -331,165 +309,10 @@ ntq_emit_tex(struct v3d_compile *c, nir_tex_instr *instr)
                 break;
         }
 
-        struct V3D33_TEXTURE_UNIFORM_PARAMETER_0_CFG_MODE1 p0_unpacked = {
-                V3D33_TEXTURE_UNIFORM_PARAMETER_0_CFG_MODE1_header,
-
-                .fetch_sample_mode = instr->op == nir_texop_txf,
-        };
-
-        switch (instr->sampler_dim) {
-        case GLSL_SAMPLER_DIM_1D:
-                if (instr->is_array)
-                        p0_unpacked.lookup_type = TEXTURE_1D_ARRAY;
-                else
-                        p0_unpacked.lookup_type = TEXTURE_1D;
-                break;
-        case GLSL_SAMPLER_DIM_2D:
-        case GLSL_SAMPLER_DIM_RECT:
-                if (instr->is_array)
-                        p0_unpacked.lookup_type = TEXTURE_2D_ARRAY;
-                else
-                        p0_unpacked.lookup_type = TEXTURE_2D;
-                break;
-        case GLSL_SAMPLER_DIM_3D:
-                p0_unpacked.lookup_type = TEXTURE_3D;
-                break;
-        case GLSL_SAMPLER_DIM_CUBE:
-                p0_unpacked.lookup_type = TEXTURE_CUBE_MAP;
-                break;
-        default:
-                unreachable("Bad sampler type");
-        }
-
-        struct qreg coords[5];
-        int next_coord = 0;
-        for (unsigned i = 0; i < instr->num_srcs; i++) {
-                switch (instr->src[i].src_type) {
-                case nir_tex_src_coord:
-                        for (int j = 0; j < instr->coord_components; j++) {
-                                coords[next_coord++] =
-                                        ntq_get_src(c, instr->src[i].src, j);
-                        }
-                        if (instr->coord_components < 2)
-                                coords[next_coord++] = vir_uniform_f(c, 0.5);
-                        break;
-                case nir_tex_src_bias:
-                        coords[next_coord++] =
-                                ntq_get_src(c, instr->src[i].src, 0);
-
-                        p0_unpacked.bias_supplied = true;
-                        break;
-                case nir_tex_src_lod:
-                        /* XXX: Needs base level addition */
-                        coords[next_coord++] =
-                                ntq_get_src(c, instr->src[i].src, 0);
-
-                        if (instr->op != nir_texop_txf &&
-                            instr->op != nir_texop_tg4) {
-                                p0_unpacked.disable_autolod_use_bias_only = true;
-                        }
-                        break;
-                case nir_tex_src_comparator:
-                        coords[next_coord++] =
-                                ntq_get_src(c, instr->src[i].src, 0);
-
-                        p0_unpacked.shadow = true;
-                        break;
-
-                case nir_tex_src_offset: {
-                        nir_const_value *offset =
-                                nir_src_as_const_value(instr->src[i].src);
-                        p0_unpacked.texel_offset_for_s_coordinate =
-                                offset->i32[0];
-
-                        if (instr->coord_components >= 2)
-                                p0_unpacked.texel_offset_for_t_coordinate =
-                                        offset->i32[1];
-
-                        if (instr->coord_components >= 3)
-                                p0_unpacked.texel_offset_for_r_coordinate =
-                                        offset->i32[2];
-                        break;
-                }
-
-                default:
-                        unreachable("unknown texture source");
-                }
-        }
-
-        uint32_t p0_packed;
-        V3D33_TEXTURE_UNIFORM_PARAMETER_0_CFG_MODE1_pack(NULL,
-                                                         (uint8_t *)&p0_packed,
-                                                         &p0_unpacked);
-
-        /* There is no native support for GL texture rectangle coordinates, so
-         * we have to rescale from ([0, width], [0, height]) to ([0, 1], [0,
-         * 1]).
-         */
-        if (instr->sampler_dim == GLSL_SAMPLER_DIM_RECT) {
-                coords[0] = vir_FMUL(c, coords[0],
-                                     vir_uniform(c, QUNIFORM_TEXRECT_SCALE_X,
-                                                 unit));
-                coords[1] = vir_FMUL(c, coords[1],
-                                     vir_uniform(c, QUNIFORM_TEXRECT_SCALE_Y,
-                                                 unit));
-        }
-
-        struct qreg texture_u[] = {
-                vir_uniform(c, QUNIFORM_TEXTURE_CONFIG_P0_0 + unit, p0_packed),
-                vir_uniform(c, QUNIFORM_TEXTURE_CONFIG_P1, unit),
-        };
-        uint32_t next_texture_u = 0;
-
-        for (int i = 0; i < next_coord; i++) {
-                struct qreg dst;
-
-                if (i == next_coord - 1)
-                        dst = vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_TMUL);
-                else
-                        dst = vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_TMU);
-
-                struct qinst *tmu = vir_MOV_dest(c, dst, coords[i]);
-
-                if (i < 2) {
-                        tmu->has_implicit_uniform = true;
-                        tmu->src[vir_get_implicit_uniform_src(tmu)] =
-                                texture_u[next_texture_u++];
-                }
-        }
-
-        bool return_16 = (c->key->tex[unit].return_size == 16 ||
-                          p0_unpacked.shadow);
-
-        struct qreg return_values[4];
-        for (int i = 0; i < c->key->tex[unit].return_channels; i++)
-                return_values[i] = vir_LDTMU(c);
-        /* Swizzling .zw of an RG texture should give undefined results, not
-         * crash the compiler.
-         */
-        for (int i = c->key->tex[unit].return_channels; i < 4; i++)
-                return_values[i] = c->undef;
-
-        for (int i = 0; i < nir_tex_instr_dest_size(instr); i++) {
-                struct qreg chan;
-
-                if (return_16) {
-                        STATIC_ASSERT(PIPE_SWIZZLE_X == 0);
-                        chan = return_values[i / 2];
-
-                        enum v3d_qpu_input_unpack unpack;
-                        if (i & 1)
-                                unpack = V3D_QPU_UNPACK_H;
-                        else
-                                unpack = V3D_QPU_UNPACK_L;
-
-                        chan = vir_FMOV(c, chan);
-                        vir_set_unpack(c->defs[chan.index], 0, unpack);
-                } else {
-                        chan = vir_MOV(c, return_values[i]);
-                }
-                ntq_store_dest(c, &instr->dest, i, chan);
-        }
+        if (c->devinfo->ver >= 40)
+                v3d40_vir_emit_tex(c, instr);
+        else
+                v3d33_vir_emit_tex(c, instr);
 }
 
 static struct qreg
@@ -500,8 +323,7 @@ ntq_fsincos(struct v3d_compile *c, struct qreg src, bool is_cos)
                 input = vir_FADD(c, input, vir_uniform_f(c, 0.5));
 
         struct qreg periods = vir_FROUND(c, input);
-        struct qreg sin_output = vir_SFU(c, V3D_QPU_WADDR_SIN,
-                                         vir_FSUB(c, input, periods));
+        struct qreg sin_output = vir_SIN(c, vir_FSUB(c, input, periods));
         return vir_XOR(c, sin_output, vir_SHL(c,
                                               vir_FTOIN(c, periods),
                                               vir_uniform_ui(c, -1)));
@@ -539,17 +361,27 @@ emit_fragcoord_input(struct v3d_compile *c, int attr)
         c->inputs[attr * 4 + 0] = vir_FXCD(c);
         c->inputs[attr * 4 + 1] = vir_FYCD(c);
         c->inputs[attr * 4 + 2] = c->payload_z;
-        c->inputs[attr * 4 + 3] = vir_SFU(c, V3D_QPU_WADDR_RECIP,
-                                          c->payload_w);
+        c->inputs[attr * 4 + 3] = vir_RECIP(c, c->payload_w);
 }
 
 static struct qreg
 emit_fragment_varying(struct v3d_compile *c, nir_variable *var,
                       uint8_t swizzle)
 {
-        struct qreg vary = vir_reg(QFILE_VARY, ~0);
+        struct qreg r3 = vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_R3);
         struct qreg r5 = vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_R5);
 
+        struct qreg vary;
+        if (c->devinfo->ver >= 41) {
+                struct qinst *ldvary = vir_add_inst(V3D_QPU_A_NOP, c->undef,
+                                                    c->undef, c->undef);
+                ldvary->qpu.sig.ldvary = true;
+                vary = vir_emit_def(c, ldvary);
+        } else {
+                vir_NOP(c)->qpu.sig.ldvary = true;
+                vary = r3;
+        }
+
         /* For gl_PointCoord input or distance along a line, we'll be called
          * with no nir_variable, and we don't count toward VPM size so we
          * don't track an input slot.
@@ -565,31 +397,36 @@ emit_fragment_varying(struct v3d_compile *c, nir_variable *var,
         switch (var->data.interpolation) {
         case INTERP_MODE_NONE:
                 /* If a gl_FrontColor or gl_BackColor input has no interp
-                 * qualifier, then flag it for glShadeModel() handling by the
-                 * driver.
+                 * qualifier, then if we're using glShadeModel(GL_FLAT) it
+                 * needs to be flat shaded.
                  */
                 switch (var->data.location) {
                 case VARYING_SLOT_COL0:
                 case VARYING_SLOT_COL1:
                 case VARYING_SLOT_BFC0:
                 case VARYING_SLOT_BFC1:
-                        BITSET_SET(c->shade_model_flags, i);
-                        break;
+                        if (c->fs_key->shade_model_flat) {
+                                BITSET_SET(c->flat_shade_flags, i);
+                                vir_MOV_dest(c, c->undef, vary);
+                                return vir_MOV(c, r5);
+                        } else {
+                                return vir_FADD(c, vir_FMUL(c, vary,
+                                                            c->payload_w), r5);
+                        }
                 default:
                         break;
                 }
                 /* FALLTHROUGH */
         case INTERP_MODE_SMOOTH:
                 if (var->data.centroid) {
+                        BITSET_SET(c->centroid_flags, i);
                         return vir_FADD(c, vir_FMUL(c, vary,
                                                     c->payload_w_centroid), r5);
                 } else {
                         return vir_FADD(c, vir_FMUL(c, vary, c->payload_w), r5);
                 }
         case INTERP_MODE_NOPERSPECTIVE:
-                /* C appears after the mov from the varying.
-                   XXX: improve ldvary setup.
-                */
+                BITSET_SET(c->noperspective_flags, i);
                 return vir_FADD(c, vir_MOV(c, vary), r5);
         case INTERP_MODE_FLAT:
                 BITSET_SET(c->flat_shade_flags, i);
@@ -604,8 +441,9 @@ static void
 emit_fragment_input(struct v3d_compile *c, int attr, nir_variable *var)
 {
         for (int i = 0; i < glsl_get_vector_elements(var->type); i++) {
-                c->inputs[attr * 4 + i] =
-                        emit_fragment_varying(c, var, i);
+                int chan = var->data.location_frac + i;
+                c->inputs[attr * 4 + chan] =
+                        emit_fragment_varying(c, var, chan);
         }
 }
 
@@ -662,7 +500,9 @@ ntq_emit_comparison(struct v3d_compile *c, struct qreg *dest,
                     nir_alu_instr *sel_instr)
 {
         struct qreg src0 = ntq_get_alu_src(c, compare_instr, 0);
-        struct qreg src1 = ntq_get_alu_src(c, compare_instr, 1);
+        struct qreg src1;
+        if (nir_op_infos[compare_instr->op].num_inputs > 1)
+                src1 = ntq_get_alu_src(c, compare_instr, 1);
         bool cond_invert = false;
 
         switch (compare_instr->op) {
@@ -806,6 +646,14 @@ ntq_emit_alu(struct v3d_compile *c, nir_alu_instr *instr)
         case nir_op_imov:
                 result = vir_MOV(c, src[0]);
                 break;
+
+        case nir_op_fneg:
+                result = vir_XOR(c, src[0], vir_uniform_ui(c, 1 << 31));
+                break;
+        case nir_op_ineg:
+                result = vir_NEG(c, src[0]);
+                break;
+
         case nir_op_fmul:
                 result = vir_FMUL(c, src[0], src[1]);
                 break;
@@ -888,8 +736,12 @@ ntq_emit_alu(struct v3d_compile *c, nir_alu_instr *instr)
                 result = vir_NOT(c, src[0]);
                 break;
 
+        case nir_op_ufind_msb:
+                result = vir_SUB(c, vir_uniform_ui(c, 31), vir_CLZ(c, src[0]));
+                break;
+
         case nir_op_imul:
-                result = ntq_umul(c, src[0], src[1]);
+                result = vir_UMUL(c, src[0], src[1]);
                 break;
 
         case nir_op_seq:
@@ -921,16 +773,16 @@ ntq_emit_alu(struct v3d_compile *c, nir_alu_instr *instr)
                 break;
 
         case nir_op_frcp:
-                result = vir_SFU(c, V3D_QPU_WADDR_RECIP, src[0]);
+                result = vir_RECIP(c, src[0]);
                 break;
         case nir_op_frsq:
-                result = vir_SFU(c, V3D_QPU_WADDR_RSQRT, src[0]);
+                result = vir_RSQRT(c, src[0]);
                 break;
         case nir_op_fexp2:
-                result = vir_SFU(c, V3D_QPU_WADDR_EXP, src[0]);
+                result = vir_EXP(c, src[0]);
                 break;
         case nir_op_flog2:
-                result = vir_SFU(c, V3D_QPU_WADDR_LOG, src[0]);
+                result = vir_LOG(c, src[0]);
                 break;
 
         case nir_op_fceil:
@@ -986,6 +838,27 @@ ntq_emit_alu(struct v3d_compile *c, nir_alu_instr *instr)
                 result = vir_FDY(c, src[0]);
                 break;
 
+        case nir_op_uadd_carry:
+                vir_PF(c, vir_ADD(c, src[0], src[1]), V3D_QPU_PF_PUSHC);
+                result = vir_MOV(c, vir_SEL(c, V3D_QPU_COND_IFA,
+                                            vir_uniform_ui(c, ~0),
+                                            vir_uniform_ui(c, 0)));
+                break;
+
+        case nir_op_pack_half_2x16_split:
+                result = vir_VFPACK(c, src[0], src[1]);
+                break;
+
+        case nir_op_unpack_half_2x16_split_x:
+                result = vir_FMOV(c, src[0]);
+                vir_set_unpack(c->defs[result.index], 0, V3D_QPU_UNPACK_L);
+                break;
+
+        case nir_op_unpack_half_2x16_split_y:
+                result = vir_FMOV(c, src[0]);
+                vir_set_unpack(c->defs[result.index], 0, V3D_QPU_UNPACK_H);
+                break;
+
         default:
                 fprintf(stderr, "unknown NIR ALU inst: ");
                 nir_print_instr(&instr->instr, stderr);
@@ -996,7 +869,7 @@ ntq_emit_alu(struct v3d_compile *c, nir_alu_instr *instr)
         /* We have a scalar result, so the instruction should only have a
          * single channel written to.
          */
-        assert(util_is_power_of_two(instr->dest.write_mask));
+        assert(util_is_power_of_two_or_zero(instr->dest.write_mask));
         ntq_store_dest(c, &instr->dest.dest,
                        ffs(instr->dest.write_mask) - 1, result);
 }
@@ -1036,6 +909,22 @@ emit_frag_end(struct v3d_compile *c)
         }
         */
 
+        bool has_any_tlb_color_write = false;
+        for (int rt = 0; rt < c->fs_key->nr_cbufs; rt++) {
+                if (c->output_color_var[rt])
+                        has_any_tlb_color_write = true;
+        }
+
+        if (c->fs_key->sample_alpha_to_coverage && c->output_color_var[0]) {
+                struct nir_variable *var = c->output_color_var[0];
+                struct qreg *color = &c->outputs[var->data.driver_location * 4];
+
+                vir_SETMSF_dest(c, vir_reg(QFILE_NULL, 0),
+                                vir_AND(c,
+                                        vir_MSF(c),
+                                        vir_FTOC(c, color[3])));
+        }
+
         if (c->output_position_index != -1) {
                 struct qinst *inst = vir_MOV_dest(c,
                                                   vir_reg(QFILE_TLBU, 0),
@@ -1046,7 +935,19 @@ emit_frag_end(struct v3d_compile *c)
                                        TLB_TYPE_DEPTH |
                                        TLB_DEPTH_TYPE_PER_PIXEL |
                                        0xffffff00);
-        } else if (c->s->info.fs.uses_discard) {
+        } else if (c->s->info.fs.uses_discard ||
+                   c->fs_key->sample_alpha_to_coverage ||
+                   !has_any_tlb_color_write) {
+                /* Emit passthrough Z if it needed to be delayed until shader
+                 * end due to potential discards.
+                 *
+                 * Since (single-threaded) fragment shaders always need a TLB
+                 * write, emit passthrouh Z if we didn't have any color
+                 * buffers and flag us as potentially discarding, so that we
+                 * can use Z as the TLB write.
+                 */
+                c->s->info.fs.uses_discard = true;
+
                 struct qinst *inst = vir_MOV_dest(c,
                                                   vir_reg(QFILE_TLBU, 0),
                                                   vir_reg(QFILE_NULL, 0));
@@ -1075,11 +976,18 @@ emit_frag_end(struct v3d_compile *c)
                 conf |= TLB_SAMPLE_MODE_PER_PIXEL;
                 conf |= (7 - rt) << TLB_RENDER_TARGET_SHIFT;
 
+                if (c->fs_key->swap_color_rb & (1 << rt))
+                        num_components = MAX2(num_components, 3);
+
                 assert(num_components != 0);
                 switch (glsl_get_base_type(var->type)) {
                 case GLSL_TYPE_UINT:
                 case GLSL_TYPE_INT:
-                        conf |= TLB_TYPE_I32_COLOR;
+                        /* The F32 vs I32 distinction was dropped in 4.2. */
+                        if (c->devinfo->ver < 42)
+                                conf |= TLB_TYPE_I32_COLOR;
+                        else
+                                conf |= TLB_TYPE_F32_COLOR;
                         conf |= ((num_components - 1) <<
                                  TLB_VEC_SIZE_MINUS_1_SHIFT);
 
@@ -1099,7 +1007,7 @@ emit_frag_end(struct v3d_compile *c)
                         struct qreg b = color[2];
                         struct qreg a = color[3];
 
-                        if (c->fs_key->f32_color_rb) {
+                        if (c->fs_key->f32_color_rb & (1 << rt)) {
                                 conf |= TLB_TYPE_F32_COLOR;
                                 conf |= ((num_components - 1) <<
                                          TLB_VEC_SIZE_MINUS_1_SHIFT);
@@ -1117,15 +1025,20 @@ emit_frag_end(struct v3d_compile *c)
                                 b = color[0];
                         }
 
+                        if (c->fs_key->sample_alpha_to_one)
+                                a = vir_uniform_f(c, 1.0);
+
                         if (c->fs_key->f32_color_rb & (1 << rt)) {
-                                inst = vir_MOV_dest(c, vir_reg(QFILE_TLBU, 0), color[0]);
+                                inst = vir_MOV_dest(c, vir_reg(QFILE_TLBU, 0), r);
                                 inst->src[vir_get_implicit_uniform_src(inst)] =
                                         vir_uniform_ui(c, conf);
 
-                                for (int i = 1; i < num_components; i++) {
-                                        inst = vir_MOV_dest(c, vir_reg(QFILE_TLB, 0),
-                                                            color[i]);
-                                }
+                                if (num_components >= 2)
+                                        vir_MOV_dest(c, vir_reg(QFILE_TLB, 0), g);
+                                if (num_components >= 3)
+                                        vir_MOV_dest(c, vir_reg(QFILE_TLB, 0), b);
+                                if (num_components >= 4)
+                                        vir_MOV_dest(c, vir_reg(QFILE_TLB, 0), a);
                         } else {
                                 inst = vir_VFPACK_dest(c, vir_reg(QFILE_TLB, 0), r, g);
                                 if (conf != ~0) {
@@ -1134,7 +1047,8 @@ emit_frag_end(struct v3d_compile *c)
                                                 vir_uniform_ui(c, conf);
                                 }
 
-                                inst = vir_VFPACK_dest(c, vir_reg(QFILE_TLB, 0), b, a);
+                                if (num_components >= 3)
+                                        inst = vir_VFPACK_dest(c, vir_reg(QFILE_TLB, 0), b, a);
                         }
                         break;
                 }
@@ -1143,7 +1057,21 @@ emit_frag_end(struct v3d_compile *c)
 }
 
 static void
-emit_scaled_viewport_write(struct v3d_compile *c, struct qreg rcp_w)
+vir_VPM_WRITE(struct v3d_compile *c, struct qreg val, uint32_t *vpm_index)
+{
+        if (c->devinfo->ver >= 40) {
+                vir_STVPMV(c, vir_uniform_ui(c, *vpm_index), val);
+                *vpm_index = *vpm_index + 1;
+        } else {
+                vir_MOV_dest(c, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_VPM), val);
+        }
+
+        c->num_vpm_writes++;
+}
+
+static void
+emit_scaled_viewport_write(struct v3d_compile *c, struct qreg rcp_w,
+                           uint32_t *vpm_index)
 {
         for (int i = 0; i < 2; i++) {
                 struct qreg coord = c->outputs[c->output_position_index + i];
@@ -1151,34 +1079,32 @@ emit_scaled_viewport_write(struct v3d_compile *c, struct qreg rcp_w)
                                  vir_uniform(c, QUNIFORM_VIEWPORT_X_SCALE + i,
                                              0));
                 coord = vir_FMUL(c, coord, rcp_w);
-                vir_FTOIN_dest(c, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_VPM),
-                               coord);
+                vir_VPM_WRITE(c, vir_FTOIN(c, coord), vpm_index);
         }
 
 }
 
 static void
-emit_zs_write(struct v3d_compile *c, struct qreg rcp_w)
+emit_zs_write(struct v3d_compile *c, struct qreg rcp_w, uint32_t *vpm_index)
 {
         struct qreg zscale = vir_uniform(c, QUNIFORM_VIEWPORT_Z_SCALE, 0);
         struct qreg zoffset = vir_uniform(c, QUNIFORM_VIEWPORT_Z_OFFSET, 0);
 
-        vir_FADD_dest(c, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_VPM),
-                      vir_FMUL(c, vir_FMUL(c,
-                                           c->outputs[c->output_position_index + 2],
-                                           zscale),
-                               rcp_w),
-                      zoffset);
+        struct qreg z = c->outputs[c->output_position_index + 2];
+        z = vir_FMUL(c, z, zscale);
+        z = vir_FMUL(c, z, rcp_w);
+        z = vir_FADD(c, z, zoffset);
+        vir_VPM_WRITE(c, z, vpm_index);
 }
 
 static void
-emit_rcp_wc_write(struct v3d_compile *c, struct qreg rcp_w)
+emit_rcp_wc_write(struct v3d_compile *c, struct qreg rcp_w, uint32_t *vpm_index)
 {
-        vir_VPM_WRITE(c, rcp_w);
+        vir_VPM_WRITE(c, rcp_w, vpm_index);
 }
 
 static void
-emit_point_size_write(struct v3d_compile *c)
+emit_point_size_write(struct v3d_compile *c, uint32_t *vpm_index)
 {
         struct qreg point_size;
 
@@ -1192,55 +1118,67 @@ emit_point_size_write(struct v3d_compile *c)
          */
         point_size = vir_FMAX(c, point_size, vir_uniform_f(c, .125));
 
-        vir_VPM_WRITE(c, point_size);
+        vir_VPM_WRITE(c, point_size, vpm_index);
 }
 
 static void
 emit_vpm_write_setup(struct v3d_compile *c)
 {
-        uint32_t packed;
-        struct V3D33_VPM_GENERIC_BLOCK_WRITE_SETUP unpacked = {
-                V3D33_VPM_GENERIC_BLOCK_WRITE_SETUP_header,
-
-                .horiz = true,
-                .laned = false,
-                .segs = true,
-                .stride = 1,
-                .size = VPM_SETUP_SIZE_32_BIT,
-                .addr = 0,
-        };
-
-        V3D33_VPM_GENERIC_BLOCK_WRITE_SETUP_pack(NULL,
-                                                (uint8_t *)&packed,
-                                                &unpacked);
-        vir_VPMSETUP(c, vir_uniform_ui(c, packed));
+        if (c->devinfo->ver >= 40)
+                return;
+
+        v3d33_vir_vpm_write_setup(c);
+}
+
+/**
+ * Sets up c->outputs[c->output_position_index] for the vertex shader
+ * epilogue, if an output vertex position wasn't specified in the user's
+ * shader.  This may be the case for transform feedback with rasterizer
+ * discard enabled.
+ */
+static void
+setup_default_position(struct v3d_compile *c)
+{
+        if (c->output_position_index != -1)
+                return;
+
+        c->output_position_index = c->outputs_array_size;
+        for (int i = 0; i < 4; i++) {
+                add_output(c,
+                           c->output_position_index + i,
+                           VARYING_SLOT_POS, i);
+        }
 }
 
 static void
 emit_vert_end(struct v3d_compile *c)
 {
-        struct qreg rcp_w = vir_SFU(c, V3D_QPU_WADDR_RECIP,
-                                    c->outputs[c->output_position_index + 3]);
+        setup_default_position(c);
+
+        uint32_t vpm_index = 0;
+        struct qreg rcp_w = vir_RECIP(c,
+                                      c->outputs[c->output_position_index + 3]);
 
         emit_vpm_write_setup(c);
 
         if (c->vs_key->is_coord) {
                 for (int i = 0; i < 4; i++)
-                        vir_VPM_WRITE(c, c->outputs[c->output_position_index + i]);
-                emit_scaled_viewport_write(c, rcp_w);
+                        vir_VPM_WRITE(c, c->outputs[c->output_position_index + i],
+                                      &vpm_index);
+                emit_scaled_viewport_write(c, rcp_w, &vpm_index);
                 if (c->vs_key->per_vertex_point_size) {
-                        emit_point_size_write(c);
+                        emit_point_size_write(c, &vpm_index);
                         /* emit_rcp_wc_write(c, rcp_w); */
                 }
                 /* XXX: Z-only rendering */
                 if (0)
-                        emit_zs_write(c, rcp_w);
+                        emit_zs_write(c, rcp_w, &vpm_index);
         } else {
-                emit_scaled_viewport_write(c, rcp_w);
-                emit_zs_write(c, rcp_w);
-                emit_rcp_wc_write(c, rcp_w);
+                emit_scaled_viewport_write(c, rcp_w, &vpm_index);
+                emit_zs_write(c, rcp_w, &vpm_index);
+                emit_rcp_wc_write(c, rcp_w, &vpm_index);
                 if (c->vs_key->per_vertex_point_size)
-                        emit_point_size_write(c);
+                        emit_point_size_write(c, &vpm_index);
         }
 
         for (int i = 0; i < c->vs_key->num_fs_inputs; i++) {
@@ -1251,7 +1189,8 @@ emit_vert_end(struct v3d_compile *c)
                         struct v3d_varying_slot output = c->output_slots[j];
 
                         if (!memcmp(&input, &output, sizeof(input))) {
-                                vir_VPM_WRITE(c, c->outputs[j]);
+                                vir_VPM_WRITE(c, c->outputs[j],
+                                              &vpm_index);
                                 break;
                         }
                 }
@@ -1259,8 +1198,14 @@ emit_vert_end(struct v3d_compile *c)
                  * this FS input.
                  */
                 if (j == c->num_outputs)
-                        vir_VPM_WRITE(c, vir_uniform_f(c, 0.0));
+                        vir_VPM_WRITE(c, vir_uniform_f(c, 0.0),
+                                      &vpm_index);
         }
+
+        /* GFXH-1684: VPM writes need to be complete by the end of the shader.
+         */
+        if (c->devinfo->ver >= 40 && c->devinfo->ver <= 42)
+                vir_VPMWT(c);
 }
 
 void
@@ -1284,6 +1229,8 @@ v3d_optimize_nir(struct nir_shader *s)
                 NIR_PASS(progress, s, nir_opt_constant_folding);
                 NIR_PASS(progress, s, nir_opt_undef);
         } while (progress);
+
+        NIR_PASS(progress, s, nir_opt_move_load_ubo);
 }
 
 static int
@@ -1303,6 +1250,12 @@ ntq_emit_vpm_read(struct v3d_compile *c,
 {
         struct qreg vpm = vir_reg(QFILE_VPM, vpm_index);
 
+        if (c->devinfo->ver >= 40 ) {
+                return vir_LDVPMV_IN(c,
+                                     vir_uniform_ui(c,
+                                                    (*num_components_queued)++));
+        }
+
         if (*num_components_queued != 0) {
                 (*num_components_queued)--;
                 c->num_inputs++;
@@ -1311,24 +1264,7 @@ ntq_emit_vpm_read(struct v3d_compile *c,
 
         uint32_t num_components = MIN2(*remaining, 32);
 
-        struct V3D33_VPM_GENERIC_BLOCK_READ_SETUP unpacked = {
-                V3D33_VPM_GENERIC_BLOCK_READ_SETUP_header,
-
-                .horiz = true,
-                .laned = false,
-                /* If the field is 0, that means a read count of 32. */
-                .num = num_components & 31,
-                .segs = true,
-                .stride = 1,
-                .size = VPM_SETUP_SIZE_32_BIT,
-                .addr = c->num_inputs,
-        };
-
-        uint32_t packed;
-        V3D33_VPM_GENERIC_BLOCK_READ_SETUP_pack(NULL,
-                                                (uint8_t *)&packed,
-                                                &unpacked);
-        vir_VPMSETUP(c, vir_uniform_ui(c, packed));
+        v3d33_vir_vpm_read_setup(c, num_components);
 
         *num_components_queued = num_components - 1;
         *remaining -= num_components;
@@ -1338,7 +1274,72 @@ ntq_emit_vpm_read(struct v3d_compile *c,
 }
 
 static void
-ntq_setup_inputs(struct v3d_compile *c)
+ntq_setup_vpm_inputs(struct v3d_compile *c)
+{
+        /* Figure out how many components of each vertex attribute the shader
+         * uses.  Each variable should have been split to individual
+         * components and unused ones DCEed.  The vertex fetcher will load
+         * from the start of the attribute to the number of components we
+         * declare we need in c->vattr_sizes[].
+         */
+        nir_foreach_variable(var, &c->s->inputs) {
+                /* No VS attribute array support. */
+                assert(MAX2(glsl_get_length(var->type), 1) == 1);
+
+                unsigned loc = var->data.driver_location;
+                int start_component = var->data.location_frac;
+                int num_components = glsl_get_components(var->type);
+
+                c->vattr_sizes[loc] = MAX2(c->vattr_sizes[loc],
+                                           start_component + num_components);
+        }
+
+        unsigned num_components = 0;
+        uint32_t vpm_components_queued = 0;
+        bool uses_iid = c->s->info.system_values_read &
+                (1ull << SYSTEM_VALUE_INSTANCE_ID);
+        bool uses_vid = c->s->info.system_values_read &
+                (1ull << SYSTEM_VALUE_VERTEX_ID);
+        num_components += uses_iid;
+        num_components += uses_vid;
+
+        for (int i = 0; i < ARRAY_SIZE(c->vattr_sizes); i++)
+                num_components += c->vattr_sizes[i];
+
+        if (uses_iid) {
+                c->iid = ntq_emit_vpm_read(c, &vpm_components_queued,
+                                           &num_components, ~0);
+        }
+
+        if (uses_vid) {
+                c->vid = ntq_emit_vpm_read(c, &vpm_components_queued,
+                                           &num_components, ~0);
+        }
+
+        for (int loc = 0; loc < ARRAY_SIZE(c->vattr_sizes); loc++) {
+                resize_qreg_array(c, &c->inputs, &c->inputs_array_size,
+                                  (loc + 1) * 4);
+
+                for (int i = 0; i < c->vattr_sizes[loc]; i++) {
+                        c->inputs[loc * 4 + i] =
+                                ntq_emit_vpm_read(c,
+                                                  &vpm_components_queued,
+                                                  &num_components,
+                                                  loc * 4 + i);
+
+                }
+        }
+
+        if (c->devinfo->ver >= 40) {
+                assert(vpm_components_queued == num_components);
+        } else {
+                assert(vpm_components_queued == 0);
+                assert(num_components == 0);
+        }
+}
+
+static void
+ntq_setup_fs_inputs(struct v3d_compile *c)
 {
         unsigned num_entries = 0;
         unsigned num_components = 0;
@@ -1360,27 +1361,6 @@ ntq_setup_inputs(struct v3d_compile *c)
          */
         qsort(&vars, num_entries, sizeof(*vars), driver_location_compare);
 
-        uint32_t vpm_components_queued = 0;
-        if (c->s->info.stage == MESA_SHADER_VERTEX) {
-                bool uses_iid = c->s->info.system_values_read &
-                        (1ull << SYSTEM_VALUE_INSTANCE_ID);
-                bool uses_vid = c->s->info.system_values_read &
-                        (1ull << SYSTEM_VALUE_VERTEX_ID);
-
-                num_components += uses_iid;
-                num_components += uses_vid;
-
-                if (uses_iid) {
-                        c->iid = ntq_emit_vpm_read(c, &vpm_components_queued,
-                                                   &num_components, ~0);
-                }
-
-                if (uses_vid) {
-                        c->vid = ntq_emit_vpm_read(c, &vpm_components_queued,
-                                                   &num_components, ~0);
-                }
-        }
-
         for (unsigned i = 0; i < num_entries; i++) {
                 nir_variable *var = vars[i];
                 unsigned array_len = MAX2(glsl_get_length(var->type), 1);
@@ -1391,38 +1371,19 @@ ntq_setup_inputs(struct v3d_compile *c)
                 resize_qreg_array(c, &c->inputs, &c->inputs_array_size,
                                   (loc + 1) * 4);
 
-                if (c->s->info.stage == MESA_SHADER_FRAGMENT) {
-                        if (var->data.location == VARYING_SLOT_POS) {
-                                emit_fragcoord_input(c, loc);
-                        } else if (var->data.location == VARYING_SLOT_PNTC ||
-                                   (var->data.location >= VARYING_SLOT_VAR0 &&
-                                    (c->fs_key->point_sprite_mask &
-                                     (1 << (var->data.location -
-                                            VARYING_SLOT_VAR0))))) {
-                                c->inputs[loc * 4 + 0] = c->point_x;
-                                c->inputs[loc * 4 + 1] = c->point_y;
-                        } else {
-                                emit_fragment_input(c, loc, var);
-                        }
+                if (var->data.location == VARYING_SLOT_POS) {
+                        emit_fragcoord_input(c, loc);
+                } else if (var->data.location == VARYING_SLOT_PNTC ||
+                           (var->data.location >= VARYING_SLOT_VAR0 &&
+                            (c->fs_key->point_sprite_mask &
+                             (1 << (var->data.location -
+                                    VARYING_SLOT_VAR0))))) {
+                        c->inputs[loc * 4 + 0] = c->point_x;
+                        c->inputs[loc * 4 + 1] = c->point_y;
                 } else {
-                        int var_components = glsl_get_components(var->type);
-
-                        for (int i = 0; i < var_components; i++) {
-                                c->inputs[loc * 4 + i] =
-                                        ntq_emit_vpm_read(c,
-                                                          &vpm_components_queued,
-                                                          &num_components,
-                                                          loc * 4 + i);
-
-                        }
-                        c->vattr_sizes[loc] = var_components;
+                        emit_fragment_input(c, loc, var);
                 }
         }
-
-        if (c->s->info.stage == MESA_SHADER_VERTEX) {
-                assert(vpm_components_queued == 0);
-                assert(num_components == 0);
-        }
 }
 
 static void
@@ -1435,8 +1396,11 @@ ntq_setup_outputs(struct v3d_compile *c)
                 assert(array_len == 1);
                 (void)array_len;
 
-                for (int i = 0; i < 4; i++)
-                        add_output(c, loc + i, var->data.location, i);
+                for (int i = 0; i < 4 - var->data.location_frac; i++) {
+                        add_output(c, loc + var->data.location_frac + i,
+                                   var->data.location,
+                                   var->data.location_frac + i);
+                }
 
                 if (c->s->info.stage == MESA_SHADER_FRAGMENT) {
                         switch (var->data.location) {
@@ -1566,6 +1530,8 @@ ntq_emit_intrinsic(struct v3d_compile *c, nir_intrinsic_instr *instr)
                                              ntq_get_src(c, instr->src[1], 0),
                                              vir_uniform_ui(c, i * 4)));
 
+                        vir_emit_thrsw(c);
+
                         ntq_store_dest(c, &instr->dest, i, vir_LDTMU(c));
                 }
                 break;
@@ -1669,7 +1635,7 @@ ntq_emit_intrinsic(struct v3d_compile *c, nir_intrinsic_instr *instr)
                          * the condition so that we can use zero as "executing
                          * and discarding."
                          */
-                        vir_PF(c, vir_AND(c, c->execute, vir_NOT(c, cond)),
+                        vir_PF(c, vir_OR(c, c->execute, vir_NOT(c, cond)),
                                V3D_QPU_PF_PUSHZ);
                         vir_set_cond(vir_SETMSF_dest(c, vir_reg(QFILE_NULL, 0),
                                                      vir_uniform_ui(c, 0)),
@@ -1698,7 +1664,7 @@ ntq_emit_intrinsic(struct v3d_compile *c, nir_intrinsic_instr *instr)
 static void
 ntq_activate_execute_for_block(struct v3d_compile *c)
 {
-        vir_PF(c, vir_SUB(c, c->execute, vir_uniform_ui(c, c->cur_block->index)),
+        vir_PF(c, vir_XOR(c, c->execute, vir_uniform_ui(c, c->cur_block->index)),
                V3D_QPU_PF_PUSHZ);
 
         vir_MOV_cond(c, V3D_QPU_COND_IFA, c->execute, vir_uniform_ui(c, 0));
@@ -1760,7 +1726,7 @@ ntq_emit_if(struct v3d_compile *c, nir_if *if_stmt)
                              vir_uniform_ui(c, after_block->index));
 
                 /* If everything points at ENDIF, then jump there immediately. */
-                vir_PF(c, vir_SUB(c, c->execute,
+                vir_PF(c, vir_XOR(c, c->execute,
                                   vir_uniform_ui(c, after_block->index)),
                        V3D_QPU_PF_PUSHZ);
                 vir_BRANCH(c, V3D_QPU_BRANCH_COND_ALLA);
@@ -1874,7 +1840,7 @@ ntq_emit_loop(struct v3d_compile *c, nir_loop *loop)
          *
          * XXX: Use the .ORZ flags update, instead.
          */
-        vir_PF(c, vir_SUB(c,
+        vir_PF(c, vir_XOR(c,
                           c->execute,
                           vir_uniform_ui(c, c->loop_cont_block->index)),
                V3D_QPU_PF_PUSHZ);
@@ -1882,7 +1848,11 @@ ntq_emit_loop(struct v3d_compile *c, nir_loop *loop)
 
         vir_PF(c, c->execute, V3D_QPU_PF_PUSHZ);
 
-        vir_BRANCH(c, V3D_QPU_BRANCH_COND_ANYA);
+        struct qinst *branch = vir_BRANCH(c, V3D_QPU_BRANCH_COND_ANYA);
+        /* Pixels that were not dispatched or have been discarded should not
+         * contribute to looping again.
+         */
+        branch->qpu.branch.msfign = V3D_QPU_MSFIGN_P;
         vir_link_blocks(c->cur_block, c->loop_cont_block);
         vir_link_blocks(c->cur_block, c->loop_break_block);
 
@@ -1954,7 +1924,11 @@ nir_to_vir(struct v3d_compile *c)
                 }
         }
 
-        ntq_setup_inputs(c);
+        if (c->s->info.stage == MESA_SHADER_FRAGMENT)
+                ntq_setup_fs_inputs(c);
+        else
+                ntq_setup_vpm_inputs(c);
+
         ntq_setup_outputs(c);
         ntq_setup_uniforms(c);
         ntq_setup_registers(c, &c->s->registers);
@@ -1968,20 +1942,33 @@ nir_to_vir(struct v3d_compile *c)
 }
 
 const nir_shader_compiler_options v3d_nir_options = {
+        .lower_all_io_to_temps = true,
         .lower_extract_byte = true,
         .lower_extract_word = true,
-        .lower_bitfield_insert = true,
-        .lower_bitfield_extract = true,
+        .lower_bfm = true,
+        .lower_bitfield_insert_to_shifts = true,
+        .lower_bitfield_extract_to_shifts = true,
+        .lower_bitfield_reverse = true,
+        .lower_bit_count = true,
         .lower_pack_unorm_2x16 = true,
         .lower_pack_snorm_2x16 = true,
         .lower_pack_unorm_4x8 = true,
         .lower_pack_snorm_4x8 = true,
+        .lower_unpack_unorm_4x8 = true,
+        .lower_unpack_snorm_4x8 = true,
+        .lower_pack_half_2x16 = true,
+        .lower_unpack_half_2x16 = true,
+        .lower_fdiv = true,
+        .lower_find_lsb = true,
         .lower_ffma = true,
         .lower_flrp32 = true,
         .lower_fpow = true,
         .lower_fsat = true,
         .lower_fsqrt = true,
-        .lower_negate = true,
+        .lower_ifind_msb = true,
+        .lower_ldexp = true,
+        .lower_mul_high = true,
+        .lower_wpos_pntc = true,
         .native_integers = true,
 };
 
@@ -2003,6 +1990,85 @@ count_nir_instrs(nir_shader *nir)
 }
 #endif
 
+/**
+ * When demoting a shader down to single-threaded, removes the THRSW
+ * instructions (one will still be inserted at v3d_vir_to_qpu() for the
+ * program end).
+ */
+static void
+vir_remove_thrsw(struct v3d_compile *c)
+{
+        vir_for_each_block(block, c) {
+                vir_for_each_inst_safe(inst, block) {
+                        if (inst->qpu.sig.thrsw)
+                                vir_remove_instruction(c, inst);
+                }
+        }
+
+        c->last_thrsw = NULL;
+}
+
+void
+vir_emit_last_thrsw(struct v3d_compile *c)
+{
+        /* On V3D before 4.1, we need a TMU op to be outstanding when thread
+         * switching, so disable threads if we didn't do any TMU ops (each of
+         * which would have emitted a THRSW).
+         */
+        if (!c->last_thrsw_at_top_level && c->devinfo->ver < 41) {
+                c->threads = 1;
+                if (c->last_thrsw)
+                        vir_remove_thrsw(c);
+                return;
+        }
+
+        /* If we're threaded and the last THRSW was in conditional code, then
+         * we need to emit another one so that we can flag it as the last
+         * thrsw.
+         */
+        if (c->last_thrsw && !c->last_thrsw_at_top_level) {
+                assert(c->devinfo->ver >= 41);
+                vir_emit_thrsw(c);
+        }
+
+        /* If we're threaded, then we need to mark the last THRSW instruction
+         * so we can emit a pair of them at QPU emit time.
+         *
+         * For V3D 4.x, we can spawn the non-fragment shaders already in the
+         * post-last-THRSW state, so we can skip this.
+         */
+        if (!c->last_thrsw && c->s->info.stage == MESA_SHADER_FRAGMENT) {
+                assert(c->devinfo->ver >= 41);
+                vir_emit_thrsw(c);
+        }
+
+        if (c->last_thrsw)
+                c->last_thrsw->is_last_thrsw = true;
+}
+
+/* There's a flag in the shader for "center W is needed for reasons other than
+ * non-centroid varyings", so we just walk the program after VIR optimization
+ * to see if it's used.  It should be harmless to set even if we only use
+ * center W for varyings.
+ */
+static void
+vir_check_payload_w(struct v3d_compile *c)
+{
+        if (c->s->info.stage != MESA_SHADER_FRAGMENT)
+                return;
+
+        vir_for_each_inst_inorder(inst, c) {
+                for (int i = 0; i < vir_get_nsrc(inst); i++) {
+                        if (inst->src[i].file == QFILE_REG &&
+                            inst->src[i].index == 0) {
+                                c->uses_center_w = true;
+                                return;
+                        }
+                }
+        }
+
+}
+
 void
 v3d_nir_to_vir(struct v3d_compile *c)
 {
@@ -2016,6 +2082,9 @@ v3d_nir_to_vir(struct v3d_compile *c)
 
         nir_to_vir(c);
 
+        /* Emit the last THRSW before STVPM and TLB writes. */
+        vir_emit_last_thrsw(c);
+
         switch (c->s->info.stage) {
         case MESA_SHADER_FRAGMENT:
                 emit_frag_end(c);
@@ -2039,6 +2108,8 @@ v3d_nir_to_vir(struct v3d_compile *c)
         vir_optimize(c);
         vir_lower_uniforms(c);
 
+        vir_check_payload_w(c);
+
         /* XXX: vir_schedule_instructions(c); */
 
         if (V3D_DEBUG & (V3D_DEBUG_VIR |
@@ -2050,5 +2121,33 @@ v3d_nir_to_vir(struct v3d_compile *c)
                 fprintf(stderr, "\n");
         }
 
-        v3d_vir_to_qpu(c);
+        /* Attempt to allocate registers for the temporaries.  If we fail,
+         * reduce thread count and try again.
+         */
+        int min_threads = (c->devinfo->ver >= 41) ? 2 : 1;
+        struct qpu_reg *temp_registers;
+        while (true) {
+                bool spilled;
+                temp_registers = v3d_register_allocate(c, &spilled);
+                if (spilled)
+                        continue;
+
+                if (temp_registers)
+                        break;
+
+                if (c->threads == min_threads) {
+                        fprintf(stderr, "Failed to register allocate at %d threads:\n",
+                                c->threads);
+                        vir_dump(c);
+                        c->failed = true;
+                        return;
+                }
+
+                c->threads /= 2;
+
+                if (c->threads == 1)
+                        vir_remove_thrsw(c);
+        }
+
+        v3d_vir_to_qpu(c, temp_registers);
 }
diff --git a/lib/mesa/src/broadcom/compiler/qpu_schedule.c b/lib/mesa/src/broadcom/compiler/qpu_schedule.c
index dd221e027..544831959 100644
--- a/lib/mesa/src/broadcom/compiler/qpu_schedule.c
+++ b/lib/mesa/src/broadcom/compiler/qpu_schedule.c
@@ -78,11 +78,13 @@ struct schedule_node_child {
 enum direction { F, R };
 
 struct schedule_state {
+        const struct v3d_device_info *devinfo;
         struct schedule_node *last_r[6];
         struct schedule_node *last_rf[64];
         struct schedule_node *last_sf;
         struct schedule_node *last_vpm_read;
         struct schedule_node *last_tmu_write;
+        struct schedule_node *last_tmu_config;
         struct schedule_node *last_tlb;
         struct schedule_node *last_vpm;
         struct schedule_node *last_unif;
@@ -194,6 +196,16 @@ process_waddr_deps(struct schedule_state *state, struct schedule_node *n,
                 add_write_dep(state, &state->last_rf[waddr], n);
         } else if (v3d_qpu_magic_waddr_is_tmu(waddr)) {
                 add_write_dep(state, &state->last_tmu_write, n);
+                switch (waddr) {
+                case V3D_QPU_WADDR_TMUS:
+                case V3D_QPU_WADDR_TMUSCM:
+                case V3D_QPU_WADDR_TMUSF:
+                case V3D_QPU_WADDR_TMUSLOD:
+                        add_write_dep(state, &state->last_tmu_config, n);
+                        break;
+                default:
+                        break;
+                }
         } else if (v3d_qpu_magic_waddr_is_sfu(waddr)) {
                 /* Handled by v3d_qpu_writes_r4() check. */
         } else {
@@ -201,13 +213,15 @@ process_waddr_deps(struct schedule_state *state, struct schedule_node *n,
                 case V3D_QPU_WADDR_R0:
                 case V3D_QPU_WADDR_R1:
                 case V3D_QPU_WADDR_R2:
-                case V3D_QPU_WADDR_R3:
-                case V3D_QPU_WADDR_R4:
-                case V3D_QPU_WADDR_R5:
                         add_write_dep(state,
                                       &state->last_r[waddr - V3D_QPU_WADDR_R0],
                                       n);
                         break;
+                case V3D_QPU_WADDR_R3:
+                case V3D_QPU_WADDR_R4:
+                case V3D_QPU_WADDR_R5:
+                        /* Handled by v3d_qpu_writes_r*() checks below. */
+                        break;
 
                 case V3D_QPU_WADDR_VPM:
                 case V3D_QPU_WADDR_VPMU:
@@ -263,6 +277,7 @@ process_uf_deps(struct schedule_state *state, struct schedule_node *n,
 static void
 calculate_deps(struct schedule_state *state, struct schedule_node *n)
 {
+        const struct v3d_device_info *devinfo = state->devinfo;
         struct qinst *qinst = n->inst;
         struct v3d_qpu_instr *inst = &qinst->qpu;
 
@@ -306,6 +321,10 @@ calculate_deps(struct schedule_state *state, struct schedule_node *n)
                 add_write_dep(state, &state->last_vpm, n);
                 break;
 
+        case V3D_QPU_A_VPMWT:
+                add_read_dep(state, state->last_vpm, n);
+                break;
+
         case V3D_QPU_A_MSF:
                 add_read_dep(state, state->last_tlb, n);
                 break;
@@ -324,7 +343,7 @@ calculate_deps(struct schedule_state *state, struct schedule_node *n)
                 add_read_dep(state, state->last_sf, n);
                 break;
 
-        case V3D_QPU_A_FLBPOP:
+        case V3D_QPU_A_FLPOP:
                 add_write_dep(state, &state->last_sf, n);
                 break;
 
@@ -354,12 +373,16 @@ calculate_deps(struct schedule_state *state, struct schedule_node *n)
                 process_waddr_deps(state, n, inst->alu.mul.waddr,
                                    inst->alu.mul.magic_write);
         }
+        if (v3d_qpu_sig_writes_address(devinfo, &inst->sig)) {
+                process_waddr_deps(state, n, inst->sig_addr,
+                                   inst->sig_magic);
+        }
 
-        if (v3d_qpu_writes_r3(inst))
+        if (v3d_qpu_writes_r3(devinfo, inst))
                 add_write_dep(state, &state->last_r[3], n);
-        if (v3d_qpu_writes_r4(inst))
+        if (v3d_qpu_writes_r4(devinfo, inst))
                 add_write_dep(state, &state->last_r[4], n);
-        if (v3d_qpu_writes_r5(inst))
+        if (v3d_qpu_writes_r5(devinfo, inst))
                 add_write_dep(state, &state->last_r[5], n);
 
         if (inst->sig.thrsw) {
@@ -369,6 +392,7 @@ calculate_deps(struct schedule_state *state, struct schedule_node *n)
                 for (int i = 0; i < ARRAY_SIZE(state->last_r); i++)
                         add_write_dep(state, &state->last_r[i], n);
                 add_write_dep(state, &state->last_sf, n);
+                add_write_dep(state, &state->last_rtop, n);
 
                 /* Scoreboard-locking operations have to stay after the last
                  * thread switch.
@@ -376,14 +400,18 @@ calculate_deps(struct schedule_state *state, struct schedule_node *n)
                 add_write_dep(state, &state->last_tlb, n);
 
                 add_write_dep(state, &state->last_tmu_write, n);
+                add_write_dep(state, &state->last_tmu_config, n);
         }
 
-        if (inst->sig.ldtmu) {
+        if (v3d_qpu_waits_on_tmu(inst)) {
                 /* TMU loads are coming from a FIFO, so ordering is important.
                  */
                 add_write_dep(state, &state->last_tmu_write, n);
         }
 
+        if (inst->sig.wrtmuc)
+                add_write_dep(state, &state->last_tmu_config, n);
+
         if (inst->sig.ldtlb | inst->sig.ldtlbu)
                 add_read_dep(state, state->last_tlb, n);
 
@@ -408,6 +436,7 @@ calculate_forward_deps(struct v3d_compile *c, struct list_head *schedule_list)
         struct schedule_state state;
 
         memset(&state, 0, sizeof(state));
+        state.devinfo = c->devinfo;
         state.dir = F;
 
         list_for_each_entry(struct schedule_node, node, schedule_list, link)
@@ -421,6 +450,7 @@ calculate_reverse_deps(struct v3d_compile *c, struct list_head *schedule_list)
         struct schedule_state state;
 
         memset(&state, 0, sizeof(state));
+        state.devinfo = c->devinfo;
         state.dir = R;
 
         for (node = schedule_list->prev; schedule_list != node; node = node->prev) {
@@ -430,10 +460,10 @@ calculate_reverse_deps(struct v3d_compile *c, struct list_head *schedule_list)
 
 struct choose_scoreboard {
         int tick;
-        int last_sfu_write_tick;
+        int last_magic_sfu_write_tick;
         int last_ldvary_tick;
         int last_uniforms_reset_tick;
-        uint32_t last_waddr_add, last_waddr_mul;
+        int last_thrsw_tick;
         bool tlb_locked;
 };
 
@@ -442,22 +472,8 @@ mux_reads_too_soon(struct choose_scoreboard *scoreboard,
                    const struct v3d_qpu_instr *inst, enum v3d_qpu_mux mux)
 {
         switch (mux) {
-        case V3D_QPU_MUX_A:
-                if (scoreboard->last_waddr_add == inst->raddr_a ||
-                    scoreboard->last_waddr_mul == inst->raddr_a) {
-                        return true;
-                }
-                break;
-
-        case V3D_QPU_MUX_B:
-                if (scoreboard->last_waddr_add == inst->raddr_b ||
-                    scoreboard->last_waddr_mul == inst->raddr_b) {
-                        return true;
-                }
-                break;
-
         case V3D_QPU_MUX_R4:
-                if (scoreboard->tick - scoreboard->last_sfu_write_tick <= 2)
+                if (scoreboard->tick - scoreboard->last_magic_sfu_write_tick <= 2)
                         return true;
                 break;
 
@@ -512,7 +528,8 @@ reads_too_soon_after_write(struct choose_scoreboard *scoreboard,
 }
 
 static bool
-writes_too_soon_after_write(struct choose_scoreboard *scoreboard,
+writes_too_soon_after_write(const struct v3d_device_info *devinfo,
+                            struct choose_scoreboard *scoreboard,
                             struct qinst *qinst)
 {
         const struct v3d_qpu_instr *inst = &qinst->qpu;
@@ -521,8 +538,8 @@ writes_too_soon_after_write(struct choose_scoreboard *scoreboard,
          * This would normally be prevented by dependency tracking, but might
          * occur if a dead SFU computation makes it to scheduling.
          */
-        if (scoreboard->tick - scoreboard->last_sfu_write_tick < 2 &&
-            v3d_qpu_writes_r4(inst))
+        if (scoreboard->tick - scoreboard->last_magic_sfu_write_tick < 2 &&
+            v3d_qpu_writes_r4(devinfo, inst))
                 return true;
 
         return false;
@@ -549,7 +566,7 @@ get_instruction_priority(const struct v3d_qpu_instr *inst)
         next_score++;
 
         /* Schedule texture read results collection late to hide latency. */
-        if (inst->sig.ldtmu)
+        if (v3d_qpu_waits_on_tmu(inst))
                 return next_score;
         next_score++;
 
@@ -558,13 +575,8 @@ get_instruction_priority(const struct v3d_qpu_instr *inst)
         next_score++;
 
         /* Schedule texture read setup early to hide their latency better. */
-        if (inst->type == V3D_QPU_INSTR_TYPE_ALU &&
-            ((inst->alu.add.magic_write &&
-              v3d_qpu_magic_waddr_is_tmu(inst->alu.add.waddr)) ||
-             (inst->alu.mul.magic_write &&
-              v3d_qpu_magic_waddr_is_tmu(inst->alu.mul.waddr)))) {
+        if (v3d_qpu_writes_tmu(inst))
                 return next_score;
-        }
         next_score++;
 
         return baseline_score;
@@ -583,6 +595,11 @@ qpu_magic_waddr_is_periph(enum v3d_qpu_waddr waddr)
 static bool
 qpu_accesses_peripheral(const struct v3d_qpu_instr *inst)
 {
+        if (v3d_qpu_uses_vpm(inst))
+                return true;
+        if (v3d_qpu_uses_sfu(inst))
+                return true;
+
         if (inst->type == V3D_QPU_INSTR_TYPE_ALU) {
                 if (inst->alu.add.op != V3D_QPU_A_NOP &&
                     inst->alu.add.magic_write &&
@@ -590,7 +607,7 @@ qpu_accesses_peripheral(const struct v3d_qpu_instr *inst)
                         return true;
                 }
 
-                if (inst->alu.add.op == V3D_QPU_A_VPMSETUP)
+                if (inst->alu.add.op == V3D_QPU_A_TMUWT)
                         return true;
 
                 if (inst->alu.mul.op != V3D_QPU_M_NOP &&
@@ -603,7 +620,8 @@ qpu_accesses_peripheral(const struct v3d_qpu_instr *inst)
         return (inst->sig.ldvpm ||
                 inst->sig.ldtmu ||
                 inst->sig.ldtlb ||
-                inst->sig.ldtlbu);
+                inst->sig.ldtlbu ||
+                inst->sig.wrtmuc);
 }
 
 static bool
@@ -617,7 +635,11 @@ qpu_merge_inst(const struct v3d_device_info *devinfo,
                 return false;
         }
 
-        /* Can't do more than one peripheral access in an instruction. */
+        /* Can't do more than one peripheral access in an instruction.
+         *
+         * XXX: V3D 4.1 allows TMU read along with a VPM read or write, and
+         * WRTMUC with a TMU magic register write (other than tmuc).
+         */
         if (qpu_accesses_peripheral(a) && qpu_accesses_peripheral(b))
                 return false;
 
@@ -653,7 +675,8 @@ qpu_merge_inst(const struct v3d_device_info *devinfo,
 
         if (v3d_qpu_uses_mux(b, V3D_QPU_MUX_B)) {
                 if (v3d_qpu_uses_mux(a, V3D_QPU_MUX_B) &&
-                    a->raddr_b != b->raddr_b) {
+                    (a->raddr_b != b->raddr_b ||
+                     a->sig.small_imm != b->sig.small_imm)) {
                         return false;
                 }
                 merge.raddr_b = b->raddr_b;
@@ -661,6 +684,9 @@ qpu_merge_inst(const struct v3d_device_info *devinfo,
 
         merge.sig.thrsw |= b->sig.thrsw;
         merge.sig.ldunif |= b->sig.ldunif;
+        merge.sig.ldunifrf |= b->sig.ldunifrf;
+        merge.sig.ldunifa |= b->sig.ldunifa;
+        merge.sig.ldunifarf |= b->sig.ldunifarf;
         merge.sig.ldtmu |= b->sig.ldtmu;
         merge.sig.ldvary |= b->sig.ldvary;
         merge.sig.ldvpm |= b->sig.ldvpm;
@@ -671,6 +697,12 @@ qpu_merge_inst(const struct v3d_device_info *devinfo,
         merge.sig.rotate |= b->sig.rotate;
         merge.sig.wrtmuc |= b->sig.wrtmuc;
 
+        if (v3d_qpu_sig_writes_address(devinfo, &a->sig) &&
+            v3d_qpu_sig_writes_address(devinfo, &b->sig))
+                return false;
+        merge.sig_addr |= b->sig_addr;
+        merge.sig_magic |= b->sig_magic;
+
         uint64_t packed;
         bool ok = v3d_qpu_instr_pack(devinfo, &merge, &packed);
 
@@ -717,7 +749,7 @@ choose_instruction_to_schedule(const struct v3d_device_info *devinfo,
                 if (reads_too_soon_after_write(scoreboard, n->inst))
                         continue;
 
-                if (writes_too_soon_after_write(scoreboard, n->inst))
+                if (writes_too_soon_after_write(devinfo, scoreboard, n->inst))
                         continue;
 
                 /* "A scoreboard wait must not occur in the first two
@@ -733,7 +765,7 @@ choose_instruction_to_schedule(const struct v3d_device_info *devinfo,
                  * otherwise get scheduled so ldunif and ldvary try to update
                  * r5 in the same tick.
                  */
-                if (inst->sig.ldunif &&
+                if ((inst->sig.ldunif || inst->sig.ldunifa) &&
                     scoreboard->tick == scoreboard->last_ldvary_tick + 1) {
                         continue;
                 }
@@ -801,16 +833,13 @@ update_scoreboard_for_magic_waddr(struct choose_scoreboard *scoreboard,
                                   enum v3d_qpu_waddr waddr)
 {
         if (v3d_qpu_magic_waddr_is_sfu(waddr))
-                scoreboard->last_sfu_write_tick = scoreboard->tick;
+                scoreboard->last_magic_sfu_write_tick = scoreboard->tick;
 }
 
 static void
 update_scoreboard_for_chosen(struct choose_scoreboard *scoreboard,
                              const struct v3d_qpu_instr *inst)
 {
-        scoreboard->last_waddr_add = ~0;
-        scoreboard->last_waddr_mul = ~0;
-
         if (inst->type == V3D_QPU_INSTR_TYPE_BRANCH)
                 return;
 
@@ -820,8 +849,6 @@ update_scoreboard_for_chosen(struct choose_scoreboard *scoreboard,
                 if (inst->alu.add.magic_write) {
                         update_scoreboard_for_magic_waddr(scoreboard,
                                                           inst->alu.add.waddr);
-                } else {
-                        scoreboard->last_waddr_add = inst->alu.add.waddr;
                 }
         }
 
@@ -829,8 +856,6 @@ update_scoreboard_for_chosen(struct choose_scoreboard *scoreboard,
                 if (inst->alu.mul.magic_write) {
                         update_scoreboard_for_magic_waddr(scoreboard,
                                                           inst->alu.mul.waddr);
-                } else {
-                        scoreboard->last_waddr_mul = inst->alu.mul.waddr;
                 }
         }
 
@@ -890,7 +915,7 @@ static uint32_t magic_waddr_latency(enum v3d_qpu_waddr waddr,
          *
          * because we associate the first load_tmu0 with the *second* tmu0_s.
          */
-        if (v3d_qpu_magic_waddr_is_tmu(waddr) && after->sig.ldtmu)
+        if (v3d_qpu_magic_waddr_is_tmu(waddr) && v3d_qpu_waits_on_tmu(after))
                 return 100;
 
         /* Assume that anything depending on us is consuming the SFU result. */
@@ -983,6 +1008,19 @@ mark_instruction_scheduled(struct list_head *schedule_list,
         }
 }
 
+static void
+insert_scheduled_instruction(struct v3d_compile *c,
+                             struct qblock *block,
+                             struct choose_scoreboard *scoreboard,
+                             struct qinst *inst)
+{
+        list_addtail(&inst->link, &block->instructions);
+
+        update_scoreboard_for_chosen(scoreboard, &inst->qpu);
+        c->qpu_inst_count++;
+        scoreboard->tick++;
+}
+
 static struct qinst *
 vir_nop()
 {
@@ -992,61 +1030,190 @@ vir_nop()
         return qinst;
 }
 
-#if 0
-static struct qinst *
-nop_after(struct qinst *inst)
+static void
+emit_nop(struct v3d_compile *c, struct qblock *block,
+         struct choose_scoreboard *scoreboard)
+{
+        insert_scheduled_instruction(c, block, scoreboard, vir_nop());
+}
+
+static bool
+qpu_instruction_valid_in_thrend_slot(struct v3d_compile *c,
+                                     const struct qinst *qinst, int slot)
+{
+        const struct v3d_qpu_instr *inst = &qinst->qpu;
+
+        /* Only TLB Z writes are prohibited in the last slot, but we don't
+         * have those flagged so prohibit all TLB ops for now.
+         */
+        if (slot == 2 && qpu_inst_is_tlb(inst))
+                return false;
+
+        if (slot > 0 && qinst->uniform != ~0)
+                return false;
+
+        if (v3d_qpu_uses_vpm(inst))
+                return false;
+
+        if (inst->sig.ldvary)
+                return false;
+
+        if (inst->type == V3D_QPU_INSTR_TYPE_ALU) {
+                /* GFXH-1625: TMUWT not allowed in the final instruction. */
+                if (slot == 2 && inst->alu.add.op == V3D_QPU_A_TMUWT)
+                        return false;
+
+                /* No writing physical registers at the end. */
+                if (!inst->alu.add.magic_write ||
+                    !inst->alu.mul.magic_write) {
+                        return false;
+                }
+
+                if (c->devinfo->ver < 40 && inst->alu.add.op == V3D_QPU_A_SETMSF)
+                        return false;
+
+                /* RF0-2 might be overwritten during the delay slots by
+                 * fragment shader setup.
+                 */
+                if (inst->raddr_a < 3 &&
+                    (inst->alu.add.a == V3D_QPU_MUX_A ||
+                     inst->alu.add.b == V3D_QPU_MUX_A ||
+                     inst->alu.mul.a == V3D_QPU_MUX_A ||
+                     inst->alu.mul.b == V3D_QPU_MUX_A)) {
+                        return false;
+                }
+
+                if (inst->raddr_b < 3 &&
+                    !inst->sig.small_imm &&
+                    (inst->alu.add.a == V3D_QPU_MUX_B ||
+                     inst->alu.add.b == V3D_QPU_MUX_B ||
+                     inst->alu.mul.a == V3D_QPU_MUX_B ||
+                     inst->alu.mul.b == V3D_QPU_MUX_B)) {
+                        return false;
+                }
+        }
+
+        return true;
+}
+
+static bool
+valid_thrsw_sequence(struct v3d_compile *c, struct choose_scoreboard *scoreboard,
+                     struct qinst *qinst, int instructions_in_sequence,
+                     bool is_thrend)
 {
-        struct qinst *q = vir_nop();
+        /* No emitting our thrsw while the previous thrsw hasn't happened yet. */
+        if (scoreboard->last_thrsw_tick + 3 >
+            scoreboard->tick - instructions_in_sequence) {
+                return false;
+        }
+
+        for (int slot = 0; slot < instructions_in_sequence; slot++) {
+                /* No scheduling SFU when the result would land in the other
+                 * thread.  The simulator complains for safety, though it
+                 * would only occur for dead code in our case.
+                 */
+                if (slot > 0 &&
+                    qinst->qpu.type == V3D_QPU_INSTR_TYPE_ALU &&
+                    (v3d_qpu_magic_waddr_is_sfu(qinst->qpu.alu.add.waddr) ||
+                     v3d_qpu_magic_waddr_is_sfu(qinst->qpu.alu.mul.waddr))) {
+                        return false;
+                }
+
+                if (slot > 0 && qinst->qpu.sig.ldvary)
+                        return false;
+
+                if (is_thrend &&
+                    !qpu_instruction_valid_in_thrend_slot(c, qinst, slot)) {
+                        return false;
+                }
 
-        list_add(&q->link, &inst->link);
+                /* Note that the list is circular, so we can only do this up
+                 * to instructions_in_sequence.
+                 */
+                qinst = (struct qinst *)qinst->link.next;
+        }
 
-        return q;
+        return true;
 }
 
 /**
- * Emits a THRSW/LTHRSW signal in the stream, trying to move it up to pair
- * with another instruction.
+ * Emits a THRSW signal in the stream, trying to move it up to pair with
+ * another instruction.
  */
-static void
+static int
 emit_thrsw(struct v3d_compile *c,
+           struct qblock *block,
            struct choose_scoreboard *scoreboard,
-           const struct v3d_qpu_instr *inst)
+           struct qinst *inst,
+           bool is_thrend)
 {
+        int time = 0;
+
         /* There should be nothing in a thrsw inst being scheduled other than
          * the signal bits.
          */
-        assert(inst->type == V3D_QPU_INSTR_TYPE_ALU);
-        assert(inst->alu.add.op == V3D_QPU_A_NOP);
-        assert(inst->alu.mul.op == V3D_QPU_M_NOP);
+        assert(inst->qpu.type == V3D_QPU_INSTR_TYPE_ALU);
+        assert(inst->qpu.alu.add.op == V3D_QPU_A_NOP);
+        assert(inst->qpu.alu.mul.op == V3D_QPU_M_NOP);
+
+        /* Find how far back into previous instructions we can put the THRSW. */
+        int slots_filled = 0;
+        struct qinst *merge_inst = NULL;
+        vir_for_each_inst_rev(prev_inst, block) {
+                struct v3d_qpu_sig sig = prev_inst->qpu.sig;
+                sig.thrsw = true;
+                uint32_t packed_sig;
+
+                if (!v3d_qpu_sig_pack(c->devinfo, &sig, &packed_sig))
+                        break;
 
-        /* Try to find an earlier scheduled instruction that we can merge the
-         * thrsw into.
-         */
-        int thrsw_ip = c->qpu_inst_count;
-        for (int i = 1; i <= MIN2(c->qpu_inst_count, 3); i++) {
-                uint64_t prev_instr = c->qpu_insts[c->qpu_inst_count - i];
-                uint32_t prev_sig = QPU_GET_FIELD(prev_instr, QPU_SIG);
+                if (!valid_thrsw_sequence(c, scoreboard,
+                                          prev_inst, slots_filled + 1,
+                                          is_thrend)) {
+                        break;
+                }
 
-                if (prev_sig == QPU_SIG_NONE)
-                        thrsw_ip = c->qpu_inst_count - i;
+                merge_inst = prev_inst;
+                if (++slots_filled == 3)
+                        break;
         }
 
-        if (thrsw_ip != c->qpu_inst_count) {
-                /* Merge the thrsw into the existing instruction. */
-                c->qpu_insts[thrsw_ip] =
-                        QPU_UPDATE_FIELD(c->qpu_insts[thrsw_ip], sig, QPU_SIG);
+        bool needs_free = false;
+        if (merge_inst) {
+                merge_inst->qpu.sig.thrsw = true;
+                needs_free = true;
+                scoreboard->last_thrsw_tick = scoreboard->tick - slots_filled;
         } else {
-                qpu_serialize_one_inst(c, inst);
-                update_scoreboard_for_chosen(scoreboard, inst);
+                scoreboard->last_thrsw_tick = scoreboard->tick;
+                insert_scheduled_instruction(c, block, scoreboard, inst);
+                time++;
+                slots_filled++;
+                merge_inst = inst;
+        }
+
+        /* Insert any extra delay slot NOPs we need. */
+        for (int i = 0; i < 3 - slots_filled; i++) {
+                emit_nop(c, block, scoreboard);
+                time++;
         }
 
-        /* Fill the delay slots. */
-        while (c->qpu_inst_count < thrsw_ip + 3) {
-                update_scoreboard_for_chosen(scoreboard, v3d_qpu_nop());
-                qpu_serialize_one_inst(c, v3d_qpu_nop());
+        /* If we're emitting the last THRSW (other than program end), then
+         * signal that to the HW by emitting two THRSWs in a row.
+         */
+        if (inst->is_last_thrsw) {
+                struct qinst *second_inst =
+                        (struct qinst *)merge_inst->link.next;
+                second_inst->qpu.sig.thrsw = true;
         }
+
+        /* If we put our THRSW into another instruction, free up the
+         * instruction that didn't end up scheduled into the list.
+         */
+        if (needs_free)
+                free(inst);
+
+        return time;
 }
-#endif
 
 static uint32_t
 schedule_instructions(struct v3d_compile *c,
@@ -1095,6 +1262,13 @@ schedule_instructions(struct v3d_compile *c,
                         fprintf(stderr, "\n");
                 }
 
+                /* We can't mark_instruction_scheduled() the chosen inst until
+                 * we're done identifying instructions to merge, so put the
+                 * merged instructions on a list for a moment.
+                 */
+                struct list_head merged_list;
+                list_inithead(&merged_list);
+
                 /* Schedule this instruction onto the QPU list. Also try to
                  * find an instruction to pair with it.
                  */
@@ -1104,13 +1278,14 @@ schedule_instructions(struct v3d_compile *c,
                         mark_instruction_scheduled(schedule_list, time,
                                                    chosen, true);
 
-                        merge = choose_instruction_to_schedule(devinfo,
+                        while ((merge =
+                                choose_instruction_to_schedule(devinfo,
                                                                scoreboard,
                                                                schedule_list,
-                                                               chosen);
-                        if (merge) {
+                                                               chosen))) {
                                 time = MAX2(merge->unblocked_time, time);
                                 list_del(&merge->link);
+                                list_addtail(&merge->link, &merged_list);
                                 (void)qpu_merge_inst(devinfo, inst,
                                                      inst, &merge->inst->qpu);
                                 if (merge->inst->uniform != -1) {
@@ -1156,8 +1331,8 @@ schedule_instructions(struct v3d_compile *c,
                  * DAG edge as we do so.
                  */
                 mark_instruction_scheduled(schedule_list, time, chosen, false);
-
-                if (merge) {
+                list_for_each_entry(struct schedule_node, merge, &merged_list,
+                                    link) {
                         mark_instruction_scheduled(schedule_list, time, merge,
                                                    false);
 
@@ -1167,40 +1342,24 @@ schedule_instructions(struct v3d_compile *c,
                         free(merge->inst);
                 }
 
-                if (0 && inst->sig.thrsw) {
-                        /* XXX emit_thrsw(c, scoreboard, qinst); */
+                if (inst->sig.thrsw) {
+                        time += emit_thrsw(c, block, scoreboard, qinst, false);
                 } else {
-                        c->qpu_inst_count++;
-                        list_addtail(&qinst->link, &block->instructions);
-                        update_scoreboard_for_chosen(scoreboard, inst);
-                }
-
-                scoreboard->tick++;
-                time++;
-
-                if (inst->type == V3D_QPU_INSTR_TYPE_BRANCH ||
-                    inst->sig.thrsw /* XXX */) {
-                        block->branch_qpu_ip = c->qpu_inst_count - 1;
-                        /* Fill the delay slots.
-                         *
-                         * We should fill these with actual instructions,
-                         * instead, but that will probably need to be done
-                         * after this, once we know what the leading
-                         * instructions of the successors are (so we can
-                         * handle A/B register file write latency)
-                        */
-                        /* XXX: scoreboard */
-                        int slots = (inst->type == V3D_QPU_INSTR_TYPE_BRANCH ?
-                                     3 : 2);
-                        for (int i = 0; i < slots; i++) {
-                                struct qinst *nop = vir_nop();
-                                list_addtail(&nop->link, &block->instructions);
-
-                                update_scoreboard_for_chosen(scoreboard,
-                                                             &nop->qpu);
-                                c->qpu_inst_count++;
-                                scoreboard->tick++;
-                                time++;
+                        insert_scheduled_instruction(c, block,
+                                                     scoreboard, qinst);
+
+                        if (inst->type == V3D_QPU_INSTR_TYPE_BRANCH) {
+                                block->branch_qpu_ip = c->qpu_inst_count - 1;
+                                /* Fill the delay slots.
+                                 *
+                                 * We should fill these with actual instructions,
+                                 * instead, but that will probably need to be done
+                                 * after this, once we know what the leading
+                                 * instructions of the successors are (so we can
+                                 * handle A/B register file write latency)
+                                 */
+                                for (int i = 0; i < 3; i++)
+                                        emit_nop(c, block, scoreboard);
                         }
                 }
         }
@@ -1308,6 +1467,8 @@ uint32_t
 v3d_qpu_schedule_instructions(struct v3d_compile *c)
 {
         const struct v3d_device_info *devinfo = c->devinfo;
+        struct qblock *end_block = list_last_entry(&c->blocks,
+                                                   struct qblock, link);
 
         /* We reorder the uniforms as we schedule instructions, so save the
          * old data off and replace it.
@@ -1322,11 +1483,10 @@ v3d_qpu_schedule_instructions(struct v3d_compile *c)
 
         struct choose_scoreboard scoreboard;
         memset(&scoreboard, 0, sizeof(scoreboard));
-        scoreboard.last_waddr_add = ~0;
-        scoreboard.last_waddr_mul = ~0;
         scoreboard.last_ldvary_tick = -10;
-        scoreboard.last_sfu_write_tick = -10;
+        scoreboard.last_magic_sfu_write_tick = -10;
         scoreboard.last_uniforms_reset_tick = -10;
+        scoreboard.last_thrsw_tick = -10;
 
         if (debug) {
                 fprintf(stderr, "Pre-schedule instructions\n");
@@ -1357,6 +1517,11 @@ v3d_qpu_schedule_instructions(struct v3d_compile *c)
                 block->end_qpu_ip = c->qpu_inst_count - 1;
         }
 
+        /* Emit the program-end THRSW instruction. */;
+        struct qinst *thrsw = vir_nop();
+        thrsw->qpu.sig.thrsw = true;
+        emit_thrsw(c, end_block, &scoreboard, thrsw, true);
+
         qpu_set_branch_targets(c);
 
         assert(next_uniform == c->num_uniforms);
diff --git a/lib/mesa/src/broadcom/compiler/qpu_validate.c b/lib/mesa/src/broadcom/compiler/qpu_validate.c
index d99d76a8b..fb2ed123a 100644
--- a/lib/mesa/src/broadcom/compiler/qpu_validate.c
+++ b/lib/mesa/src/broadcom/compiler/qpu_validate.c
@@ -39,6 +39,18 @@ struct v3d_qpu_validate_state {
         const struct v3d_qpu_instr *last;
         int ip;
         int last_sfu_write;
+        int last_branch_ip;
+        int last_thrsw_ip;
+
+        /* Set when we've found the last-THRSW signal, or if we were started
+         * in single-segment mode.
+         */
+        bool last_thrsw_found;
+
+        /* Set when we've found the THRSW after the last THRSW */
+        bool thrend_found;
+
+        int thrsw_count;
 };
 
 static void
@@ -63,6 +75,18 @@ fail_instr(struct v3d_qpu_validate_state *state, const char *msg)
 }
 
 static bool
+in_branch_delay_slots(struct v3d_qpu_validate_state *state)
+{
+        return (state->ip - state->last_branch_ip) < 3;
+}
+
+static bool
+in_thrsw_delay_slots(struct v3d_qpu_validate_state *state)
+{
+        return (state->ip - state->last_thrsw_ip) < 3;
+}
+
+static bool
 qpu_magic_waddr_matches(const struct v3d_qpu_instr *inst,
                         bool (*predicate)(enum v3d_qpu_waddr waddr))
 {
@@ -85,6 +109,7 @@ qpu_magic_waddr_matches(const struct v3d_qpu_instr *inst,
 static void
 qpu_validate_inst(struct v3d_qpu_validate_state *state, struct qinst *qinst)
 {
+        const struct v3d_device_info *devinfo = state->c->devinfo;
         const struct v3d_qpu_instr *inst = &qinst->qpu;
 
         if (inst->type != V3D_QPU_INSTR_TYPE_ALU)
@@ -94,10 +119,24 @@ qpu_validate_inst(struct v3d_qpu_validate_state *state, struct qinst *qinst)
          * r5 one instruction later, which is illegal to have
          * together.
          */
-        if (state->last && state->last->sig.ldvary && inst->sig.ldunif) {
+        if (state->last && state->last->sig.ldvary &&
+            (inst->sig.ldunif || inst->sig.ldunifa)) {
                 fail_instr(state, "LDUNIF after a LDVARY");
         }
 
+        /* GFXH-1633 */
+        bool last_reads_ldunif = (state->last && (state->last->sig.ldunif ||
+                                                  state->last->sig.ldunifrf));
+        bool last_reads_ldunifa = (state->last && (state->last->sig.ldunifa ||
+                                                   state->last->sig.ldunifarf));
+        bool reads_ldunif = inst->sig.ldunif || inst->sig.ldunifrf;
+        bool reads_ldunifa = inst->sig.ldunifa || inst->sig.ldunifarf;
+        if ((last_reads_ldunif && reads_ldunifa) ||
+            (last_reads_ldunifa && reads_ldunif)) {
+                fail_instr(state,
+                           "LDUNIF and LDUNIFA can't be next to each other");
+        }
+
         int tmu_writes = 0;
         int sfu_writes = 0;
         int vpm_writes = 0;
@@ -134,6 +173,19 @@ qpu_validate_inst(struct v3d_qpu_validate_state *state, struct qinst *qinst)
                 }
         }
 
+        if (in_thrsw_delay_slots(state)) {
+                /* There's no way you want to start SFU during the THRSW delay
+                 * slots, since the result would land in the other thread.
+                 */
+                if (sfu_writes) {
+                        fail_instr(state,
+                                   "SFU write started during THRSW delay slots ");
+                }
+
+                if (inst->sig.ldvary)
+                        fail_instr(state, "LDVARY during THRSW delay slots");
+        }
+
         (void)qpu_magic_waddr_matches; /* XXX */
 
         /* SFU r4 results come back two instructions later.  No doing
@@ -143,7 +195,7 @@ qpu_validate_inst(struct v3d_qpu_validate_state *state, struct qinst *qinst)
                 if (v3d_qpu_uses_mux(inst, V3D_QPU_MUX_R4))
                         fail_instr(state, "R4 read too soon after SFU");
 
-                if (v3d_qpu_writes_r4(inst))
+                if (v3d_qpu_writes_r4(devinfo, inst))
                         fail_instr(state, "R4 write too soon after SFU");
 
                 if (sfu_writes)
@@ -168,6 +220,60 @@ qpu_validate_inst(struct v3d_qpu_validate_state *state, struct qinst *qinst)
 
         if (sfu_writes)
                 state->last_sfu_write = state->ip;
+
+        if (inst->sig.thrsw) {
+                if (in_branch_delay_slots(state))
+                        fail_instr(state, "THRSW in a branch delay slot.");
+
+                if (state->last_thrsw_found)
+                        state->thrend_found = true;
+
+                if (state->last_thrsw_ip == state->ip - 1) {
+                        /* If it's the second THRSW in a row, then it's just a
+                         * last-thrsw signal.
+                         */
+                        if (state->last_thrsw_found)
+                                fail_instr(state, "Two last-THRSW signals");
+                        state->last_thrsw_found = true;
+                } else {
+                        if (in_thrsw_delay_slots(state)) {
+                                fail_instr(state,
+                                           "THRSW too close to another THRSW.");
+                        }
+                        state->thrsw_count++;
+                        state->last_thrsw_ip = state->ip;
+                }
+        }
+
+        if (state->thrend_found &&
+            state->last_thrsw_ip - state->ip <= 2 &&
+            inst->type == V3D_QPU_INSTR_TYPE_ALU) {
+                if ((inst->alu.add.op != V3D_QPU_A_NOP &&
+                     !inst->alu.add.magic_write)) {
+                        fail_instr(state, "RF write after THREND");
+                }
+
+                if ((inst->alu.mul.op != V3D_QPU_M_NOP &&
+                     !inst->alu.mul.magic_write)) {
+                        fail_instr(state, "RF write after THREND");
+                }
+
+                if (v3d_qpu_sig_writes_address(devinfo, &inst->sig))
+                        fail_instr(state, "RF write after THREND");
+
+                /* GFXH-1625: No TMUWT in the last instruction */
+                if (state->last_thrsw_ip - state->ip == 2 &&
+                    inst->alu.add.op == V3D_QPU_A_TMUWT)
+                        fail_instr(state, "TMUWT in last instruction");
+        }
+
+        if (inst->type == V3D_QPU_INSTR_TYPE_BRANCH) {
+                if (in_branch_delay_slots(state))
+                        fail_instr(state, "branch in a branch delay slot.");
+                if (in_thrsw_delay_slots(state))
+                        fail_instr(state, "branch in a THRSW delay slot.");
+                state->last_branch_ip = state->ip;
+        }
 }
 
 static void
@@ -199,10 +305,22 @@ qpu_validate(struct v3d_compile *c)
         struct v3d_qpu_validate_state state = {
                 .c = c,
                 .last_sfu_write = -10,
+                .last_thrsw_ip = -10,
+                .last_branch_ip = -10,
                 .ip = 0,
+
+                .last_thrsw_found = !c->last_thrsw,
         };
 
         vir_for_each_block(block, c) {
                 qpu_validate_block(&state, block);
         }
+
+        if (state.thrsw_count > 1 && !state.last_thrsw_found) {
+                fail_instr(&state,
+                           "thread switch found without last-THRSW in program");
+        }
+
+        if (!state.thrend_found)
+                fail_instr(&state, "No program-end THRSW found");
 }
diff --git a/lib/mesa/src/broadcom/compiler/v3d33_tex.c b/lib/mesa/src/broadcom/compiler/v3d33_tex.c
new file mode 100644
index 000000000..9af9285e0
--- /dev/null
+++ b/lib/mesa/src/broadcom/compiler/v3d33_tex.c
@@ -0,0 +1,248 @@
+/*
+ * Copyright © 2016-2018 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "v3d_compiler.h"
+
+/* We don't do any address packing. */
+#define __gen_user_data void
+#define __gen_address_type uint32_t
+#define __gen_address_offset(reloc) (*reloc)
+#define __gen_emit_reloc(cl, reloc)
+#include "cle/v3d_packet_v33_pack.h"
+
+void
+v3d33_vir_emit_tex(struct v3d_compile *c, nir_tex_instr *instr)
+{
+        unsigned unit = instr->texture_index;
+
+        struct V3D33_TEXTURE_UNIFORM_PARAMETER_0_CFG_MODE1 p0_unpacked = {
+                V3D33_TEXTURE_UNIFORM_PARAMETER_0_CFG_MODE1_header,
+
+                .fetch_sample_mode = instr->op == nir_texop_txf,
+        };
+
+        struct V3D33_TEXTURE_UNIFORM_PARAMETER_1_CFG_MODE1 p1_unpacked = {
+        };
+
+        switch (instr->sampler_dim) {
+        case GLSL_SAMPLER_DIM_1D:
+                if (instr->is_array)
+                        p0_unpacked.lookup_type = TEXTURE_1D_ARRAY;
+                else
+                        p0_unpacked.lookup_type = TEXTURE_1D;
+                break;
+        case GLSL_SAMPLER_DIM_2D:
+        case GLSL_SAMPLER_DIM_RECT:
+                if (instr->is_array)
+                        p0_unpacked.lookup_type = TEXTURE_2D_ARRAY;
+                else
+                        p0_unpacked.lookup_type = TEXTURE_2D;
+                break;
+        case GLSL_SAMPLER_DIM_3D:
+                p0_unpacked.lookup_type = TEXTURE_3D;
+                break;
+        case GLSL_SAMPLER_DIM_CUBE:
+                p0_unpacked.lookup_type = TEXTURE_CUBE_MAP;
+                break;
+        default:
+                unreachable("Bad sampler type");
+        }
+
+        struct qreg coords[5];
+        int next_coord = 0;
+        for (unsigned i = 0; i < instr->num_srcs; i++) {
+                switch (instr->src[i].src_type) {
+                case nir_tex_src_coord:
+                        for (int j = 0; j < instr->coord_components; j++) {
+                                coords[next_coord++] =
+                                        ntq_get_src(c, instr->src[i].src, j);
+                        }
+                        if (instr->coord_components < 2)
+                                coords[next_coord++] = vir_uniform_f(c, 0.5);
+                        break;
+                case nir_tex_src_bias:
+                        coords[next_coord++] =
+                                ntq_get_src(c, instr->src[i].src, 0);
+
+                        p0_unpacked.bias_supplied = true;
+                        break;
+                case nir_tex_src_lod:
+                        coords[next_coord++] =
+                                vir_FADD(c,
+                                         ntq_get_src(c, instr->src[i].src, 0),
+                                         vir_uniform(c, QUNIFORM_TEXTURE_FIRST_LEVEL,
+                                                     unit));
+
+                        if (instr->op != nir_texop_txf &&
+                            instr->op != nir_texop_tg4) {
+                                p0_unpacked.disable_autolod_use_bias_only = true;
+                        }
+                        break;
+                case nir_tex_src_comparator:
+                        coords[next_coord++] =
+                                ntq_get_src(c, instr->src[i].src, 0);
+
+                        p0_unpacked.shadow = true;
+                        break;
+
+                case nir_tex_src_offset: {
+                        nir_const_value *offset =
+                                nir_src_as_const_value(instr->src[i].src);
+                        p0_unpacked.texel_offset_for_s_coordinate =
+                                offset->i32[0];
+
+                        if (instr->coord_components >= 2)
+                                p0_unpacked.texel_offset_for_t_coordinate =
+                                        offset->i32[1];
+
+                        if (instr->coord_components >= 3)
+                                p0_unpacked.texel_offset_for_r_coordinate =
+                                        offset->i32[2];
+                        break;
+                }
+
+                default:
+                        unreachable("unknown texture source");
+                }
+        }
+
+        bool return_16 = (c->key->tex[unit].return_size == 16 ||
+                          p0_unpacked.shadow);
+
+        /* Limit the number of channels returned to both how many the NIR
+         * instruction writes and how many the instruction could produce.
+         */
+        uint32_t instr_return_channels = nir_tex_instr_dest_size(instr);
+        if (return_16)
+                instr_return_channels = (instr_return_channels + 1) / 2;
+
+        p1_unpacked.return_words_of_texture_data =
+                (1 << MIN2(instr_return_channels,
+                           c->key->tex[unit].return_channels)) - 1;
+
+        uint32_t p0_packed;
+        V3D33_TEXTURE_UNIFORM_PARAMETER_0_CFG_MODE1_pack(NULL,
+                                                         (uint8_t *)&p0_packed,
+                                                         &p0_unpacked);
+
+        uint32_t p1_packed;
+        V3D33_TEXTURE_UNIFORM_PARAMETER_1_CFG_MODE1_pack(NULL,
+                                                         (uint8_t *)&p1_packed,
+                                                         &p1_unpacked);
+        /* Load unit number into the address field, which will be be used by
+         * the driver to decide which texture to put in the actual address
+         * field.
+         */
+        p1_packed |= unit << 5;
+
+        /* There is no native support for GL texture rectangle coordinates, so
+         * we have to rescale from ([0, width], [0, height]) to ([0, 1], [0,
+         * 1]).
+         */
+        if (instr->sampler_dim == GLSL_SAMPLER_DIM_RECT) {
+                coords[0] = vir_FMUL(c, coords[0],
+                                     vir_uniform(c, QUNIFORM_TEXRECT_SCALE_X,
+                                                 unit));
+                coords[1] = vir_FMUL(c, coords[1],
+                                     vir_uniform(c, QUNIFORM_TEXRECT_SCALE_Y,
+                                                 unit));
+        }
+
+        struct qreg texture_u[] = {
+                vir_uniform(c, QUNIFORM_TEXTURE_CONFIG_P0_0 + unit, p0_packed),
+                vir_uniform(c, QUNIFORM_TEXTURE_CONFIG_P1, p1_packed),
+        };
+        uint32_t next_texture_u = 0;
+
+        for (int i = 0; i < next_coord; i++) {
+                struct qreg dst;
+
+                if (i == next_coord - 1)
+                        dst = vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_TMUL);
+                else
+                        dst = vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_TMU);
+
+                struct qinst *tmu = vir_MOV_dest(c, dst, coords[i]);
+
+                if (i < 2) {
+                        tmu->has_implicit_uniform = true;
+                        tmu->src[vir_get_implicit_uniform_src(tmu)] =
+                                texture_u[next_texture_u++];
+                }
+        }
+
+        vir_emit_thrsw(c);
+
+        struct qreg return_values[4];
+        for (int i = 0; i < 4; i++) {
+                /* Swizzling .zw of an RG texture should give undefined
+                 * results, not crash the compiler.
+                 */
+                if (p1_unpacked.return_words_of_texture_data & (1 << i))
+                        return_values[i] = vir_LDTMU(c);
+                else
+                        return_values[i] = c->undef;
+        }
+
+        for (int i = 0; i < nir_tex_instr_dest_size(instr); i++) {
+                struct qreg chan;
+
+                if (return_16) {
+                        STATIC_ASSERT(PIPE_SWIZZLE_X == 0);
+                        chan = return_values[i / 2];
+
+                        if (nir_alu_type_get_base_type(instr->dest_type) ==
+                            nir_type_float) {
+                                enum v3d_qpu_input_unpack unpack;
+                                if (i & 1)
+                                        unpack = V3D_QPU_UNPACK_H;
+                                else
+                                        unpack = V3D_QPU_UNPACK_L;
+
+                                chan = vir_FMOV(c, chan);
+                                vir_set_unpack(c->defs[chan.index], 0, unpack);
+                        } else {
+                                /* If we're unpacking the low field, shift it
+                                 * up to the top first.
+                                 */
+                                if ((i & 1) == 0) {
+                                        chan = vir_SHL(c, chan,
+                                                       vir_uniform_ui(c, 16));
+                                }
+
+                                /* Do proper sign extension to a 32-bit int. */
+                                if (nir_alu_type_get_base_type(instr->dest_type) ==
+                                    nir_type_int) {
+                                        chan = vir_ASR(c, chan,
+                                                       vir_uniform_ui(c, 16));
+                                } else {
+                                        chan = vir_SHR(c, chan,
+                                                       vir_uniform_ui(c, 16));
+                                }
+                        }
+                } else {
+                        chan = vir_MOV(c, return_values[i]);
+                }
+                ntq_store_dest(c, &instr->dest, i, chan);
+        }
+}
diff --git a/lib/mesa/src/broadcom/compiler/v3d33_vpm_setup.c b/lib/mesa/src/broadcom/compiler/v3d33_vpm_setup.c
new file mode 100644
index 000000000..8bce67dfa
--- /dev/null
+++ b/lib/mesa/src/broadcom/compiler/v3d33_vpm_setup.c
@@ -0,0 +1,75 @@
+/*
+ * Copyright © 2016-2018 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "v3d_compiler.h"
+
+/* We don't do any address packing. */
+#define __gen_user_data void
+#define __gen_address_type uint32_t
+#define __gen_address_offset(reloc) (*reloc)
+#define __gen_emit_reloc(cl, reloc)
+#include "broadcom/cle/v3d_packet_v33_pack.h"
+
+void
+v3d33_vir_vpm_read_setup(struct v3d_compile *c, int num_components)
+{
+        struct V3D33_VPM_GENERIC_BLOCK_READ_SETUP unpacked = {
+                V3D33_VPM_GENERIC_BLOCK_READ_SETUP_header,
+
+                .horiz = true,
+                .laned = false,
+                /* If the field is 0, that means a read count of 32. */
+                .num = num_components & 31,
+                .segs = true,
+                .stride = 1,
+                .size = VPM_SETUP_SIZE_32_BIT,
+                .addr = c->num_inputs,
+        };
+
+        uint32_t packed;
+        V3D33_VPM_GENERIC_BLOCK_READ_SETUP_pack(NULL,
+                                                (uint8_t *)&packed,
+                                                &unpacked);
+        vir_VPMSETUP(c, vir_uniform_ui(c, packed));
+}
+
+void
+v3d33_vir_vpm_write_setup(struct v3d_compile *c)
+{
+        uint32_t packed;
+        struct V3D33_VPM_GENERIC_BLOCK_WRITE_SETUP unpacked = {
+                V3D33_VPM_GENERIC_BLOCK_WRITE_SETUP_header,
+
+                .horiz = true,
+                .laned = false,
+                .segs = true,
+                .stride = 1,
+                .size = VPM_SETUP_SIZE_32_BIT,
+                .addr = 0,
+        };
+
+        V3D33_VPM_GENERIC_BLOCK_WRITE_SETUP_pack(NULL,
+                                                (uint8_t *)&packed,
+                                                &unpacked);
+        vir_VPMSETUP(c, vir_uniform_ui(c, packed));
+}
diff --git a/lib/mesa/src/broadcom/compiler/v3d40_tex.c b/lib/mesa/src/broadcom/compiler/v3d40_tex.c
new file mode 100644
index 000000000..9f1fd9a0d
--- /dev/null
+++ b/lib/mesa/src/broadcom/compiler/v3d40_tex.c
@@ -0,0 +1,263 @@
+/*
+ * Copyright © 2016-2018 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "v3d_compiler.h"
+
+/* We don't do any address packing. */
+#define __gen_user_data void
+#define __gen_address_type uint32_t
+#define __gen_address_offset(reloc) (*reloc)
+#define __gen_emit_reloc(cl, reloc)
+#include "cle/v3d_packet_v41_pack.h"
+
+static void
+vir_TMU_WRITE(struct v3d_compile *c, enum v3d_qpu_waddr waddr, struct qreg val,
+              int *tmu_writes)
+{
+        vir_MOV_dest(c, vir_reg(QFILE_MAGIC, waddr), val);
+
+        (*tmu_writes)++;
+}
+
+static void
+vir_WRTMUC(struct v3d_compile *c, enum quniform_contents contents, uint32_t data)
+{
+        struct qinst *inst = vir_NOP(c);
+        inst->qpu.sig.wrtmuc = true;
+        inst->has_implicit_uniform = true;
+        inst->src[0] = vir_uniform(c, contents, data);
+}
+
+void
+v3d40_vir_emit_tex(struct v3d_compile *c, nir_tex_instr *instr)
+{
+        unsigned unit = instr->texture_index;
+        int tmu_writes = 0;
+        static const struct V3D41_TMU_CONFIG_PARAMETER_2 p2_unpacked_default = {
+                .op = V3D_TMU_OP_REGULAR,
+        };
+
+        struct V3D41_TMU_CONFIG_PARAMETER_0 p0_unpacked = {
+        };
+
+        struct V3D41_TMU_CONFIG_PARAMETER_1 p1_unpacked = {
+                .output_type_32_bit = (c->key->tex[unit].return_size == 32 &&
+                                       !instr->is_shadow),
+
+                .unnormalized_coordinates = (instr->sampler_dim ==
+                                             GLSL_SAMPLER_DIM_RECT),
+        };
+
+        struct V3D41_TMU_CONFIG_PARAMETER_2 p2_unpacked = {
+                .op = V3D_TMU_OP_REGULAR,
+
+                .gather_mode = instr->op == nir_texop_tg4,
+                .gather_component = instr->component,
+
+                .coefficient_mode = instr->op == nir_texop_txd,
+        };
+
+        int non_array_components = instr->coord_components - instr->is_array;
+        struct qreg s;
+
+        for (unsigned i = 0; i < instr->num_srcs; i++) {
+                switch (instr->src[i].src_type) {
+                case nir_tex_src_coord:
+                        /* S triggers the lookup, so save it for the end. */
+                        s = ntq_get_src(c, instr->src[i].src, 0);
+
+                        if (non_array_components > 1) {
+                                vir_TMU_WRITE(c, V3D_QPU_WADDR_TMUT,
+                                              ntq_get_src(c, instr->src[i].src,
+                                                          1), &tmu_writes);
+                        }
+                        if (non_array_components > 2) {
+                                vir_TMU_WRITE(c, V3D_QPU_WADDR_TMUR,
+                                              ntq_get_src(c, instr->src[i].src,
+                                                          2), &tmu_writes);
+                        }
+
+                        if (instr->is_array) {
+                                vir_TMU_WRITE(c, V3D_QPU_WADDR_TMUI,
+                                              ntq_get_src(c, instr->src[i].src,
+                                                          instr->coord_components - 1),
+                                              &tmu_writes);
+                        }
+                        break;
+
+                case nir_tex_src_bias:
+                        vir_TMU_WRITE(c, V3D_QPU_WADDR_TMUB,
+                                      ntq_get_src(c, instr->src[i].src, 0),
+                                      &tmu_writes);
+                        break;
+
+                case nir_tex_src_lod:
+                        vir_TMU_WRITE(c, V3D_QPU_WADDR_TMUB,
+                                      ntq_get_src(c, instr->src[i].src, 0),
+                                      &tmu_writes);
+
+                        if (instr->op != nir_texop_txf &&
+                            instr->op != nir_texop_tg4) {
+                                p2_unpacked.disable_autolod = true;
+                        }
+                        break;
+
+                case nir_tex_src_comparator:
+                        vir_TMU_WRITE(c, V3D_QPU_WADDR_TMUDREF,
+                                      ntq_get_src(c, instr->src[i].src, 0),
+                                      &tmu_writes);
+                        break;
+
+                case nir_tex_src_offset: {
+                        nir_const_value *offset =
+                                nir_src_as_const_value(instr->src[i].src);
+
+                        p2_unpacked.offset_s = offset->i32[0];
+                        if (instr->coord_components >= 2)
+                                p2_unpacked.offset_t = offset->i32[1];
+                        if (instr->coord_components >= 3)
+                                p2_unpacked.offset_r = offset->i32[2];
+                        break;
+                }
+
+                default:
+                        unreachable("unknown texture source");
+                }
+        }
+
+        /* Limit the number of channels returned to both how many the NIR
+         * instruction writes and how many the instruction could produce.
+         */
+        uint32_t instr_return_channels = nir_tex_instr_dest_size(instr);
+        if (!p1_unpacked.output_type_32_bit)
+                instr_return_channels = (instr_return_channels + 1) / 2;
+
+        p0_unpacked.return_words_of_texture_data =
+                (1 << MIN2(instr_return_channels,
+                           c->key->tex[unit].return_channels)) - 1;
+
+        /* Word enables can't ask for more channels than the output type could
+         * provide (2 for f16, 4 for 32-bit).
+         */
+        assert(!p1_unpacked.output_type_32_bit ||
+               p0_unpacked.return_words_of_texture_data < (1 << 4));
+        assert(p1_unpacked.output_type_32_bit ||
+               p0_unpacked.return_words_of_texture_data < (1 << 2));
+
+        uint32_t p0_packed;
+        V3D41_TMU_CONFIG_PARAMETER_0_pack(NULL,
+                                          (uint8_t *)&p0_packed,
+                                          &p0_unpacked);
+
+        uint32_t p1_packed;
+        V3D41_TMU_CONFIG_PARAMETER_1_pack(NULL,
+                                          (uint8_t *)&p1_packed,
+                                          &p1_unpacked);
+
+        uint32_t p2_packed;
+        V3D41_TMU_CONFIG_PARAMETER_2_pack(NULL,
+                                          (uint8_t *)&p2_packed,
+                                          &p2_unpacked);
+
+        /* Load unit number into the high bits of the texture or sampler
+         * address field, which will be be used by the driver to decide which
+         * texture to put in the actual address field.
+         */
+        p0_packed |= unit << 24;
+        p1_packed |= unit << 24;
+
+        vir_WRTMUC(c, QUNIFORM_TMU_CONFIG_P0, p0_packed);
+        vir_WRTMUC(c, QUNIFORM_TMU_CONFIG_P1, p1_packed);
+        if (memcmp(&p2_unpacked, &p2_unpacked_default, sizeof(p2_unpacked)) != 0)
+                vir_WRTMUC(c, QUNIFORM_CONSTANT, p2_packed);
+
+        if (instr->op == nir_texop_txf) {
+                assert(instr->sampler_dim != GLSL_SAMPLER_DIM_CUBE);
+                vir_TMU_WRITE(c, V3D_QPU_WADDR_TMUSF, s, &tmu_writes);
+        } else if (instr->sampler_dim == GLSL_SAMPLER_DIM_CUBE) {
+                vir_TMU_WRITE(c, V3D_QPU_WADDR_TMUSCM, s, &tmu_writes);
+        } else {
+                vir_TMU_WRITE(c, V3D_QPU_WADDR_TMUS, s, &tmu_writes);
+        }
+
+        vir_emit_thrsw(c);
+
+        /* The input FIFO has 16 slots across all threads, so make sure we
+         * don't overfill our allocation.
+         */
+        while (tmu_writes > 16 / c->threads)
+                c->threads /= 2;
+
+        struct qreg return_values[4];
+        for (int i = 0; i < 4; i++) {
+                /* Swizzling .zw of an RG texture should give undefined
+                 * results, not crash the compiler.
+                 */
+                if (p0_unpacked.return_words_of_texture_data & (1 << i))
+                        return_values[i] = vir_LDTMU(c);
+                else
+                        return_values[i] = c->undef;
+        }
+
+        for (int i = 0; i < nir_tex_instr_dest_size(instr); i++) {
+                struct qreg chan;
+
+                if (!p1_unpacked.output_type_32_bit) {
+                        STATIC_ASSERT(PIPE_SWIZZLE_X == 0);
+                        chan = return_values[i / 2];
+
+                        if (nir_alu_type_get_base_type(instr->dest_type) ==
+                            nir_type_float) {
+                                enum v3d_qpu_input_unpack unpack;
+                                if (i & 1)
+                                        unpack = V3D_QPU_UNPACK_H;
+                                else
+                                        unpack = V3D_QPU_UNPACK_L;
+
+                                chan = vir_FMOV(c, chan);
+                                vir_set_unpack(c->defs[chan.index], 0, unpack);
+                        } else {
+                                /* If we're unpacking the low field, shift it
+                                 * up to the top first.
+                                 */
+                                if ((i & 1) == 0) {
+                                        chan = vir_SHL(c, chan,
+                                                       vir_uniform_ui(c, 16));
+                                }
+
+                                /* Do proper sign extension to a 32-bit int. */
+                                if (nir_alu_type_get_base_type(instr->dest_type) ==
+                                    nir_type_int) {
+                                        chan = vir_ASR(c, chan,
+                                                       vir_uniform_ui(c, 16));
+                                } else {
+                                        chan = vir_SHR(c, chan,
+                                                       vir_uniform_ui(c, 16));
+                                }
+                        }
+                } else {
+                        chan = vir_MOV(c, return_values[i]);
+                }
+                ntq_store_dest(c, &instr->dest, i, chan);
+        }
+}
diff --git a/lib/mesa/src/broadcom/compiler/v3d_compiler.h b/lib/mesa/src/broadcom/compiler/v3d_compiler.h
index 021c88f7b..070e6a3aa 100644
--- a/lib/mesa/src/broadcom/compiler/v3d_compiler.h
+++ b/lib/mesa/src/broadcom/compiler/v3d_compiler.h
@@ -33,6 +33,7 @@
 
 #include "util/macros.h"
 #include "common/v3d_debug.h"
+#include "common/v3d_device_info.h"
 #include "compiler/nir/nir.h"
 #include "util/list.h"
 #include "util/u_math.h"
@@ -72,7 +73,6 @@ enum qfile {
          * or physical registers later.
          */
         QFILE_TEMP,
-        QFILE_VARY,
         QFILE_UNIF,
         QFILE_TLB,
         QFILE_TLBU,
@@ -115,6 +115,7 @@ static inline struct qreg vir_reg(enum qfile file, uint32_t index)
  */
 struct qpu_reg {
         bool magic;
+        bool smimm;
         int index;
 };
 
@@ -134,6 +135,7 @@ struct qinst {
         struct qreg src[3];
         bool cond_is_exec_mask;
         bool has_implicit_uniform;
+        bool is_last_thrsw;
 
         /* After vir_to_qpu.c: If instr reads a uniform, which uniform from
          * the uncompiled stream it is.
@@ -171,7 +173,7 @@ enum quniform_contents {
         QUNIFORM_USER_CLIP_PLANE,
 
         /**
-         * A reference to a texture config parameter 0 uniform.
+         * A reference to a V3D 3.x texture config parameter 0 uniform.
          *
          * This is a uniform implicitly loaded with a QPU_W_TMU* write, which
          * defines texture type, miplevels, and such.  It will be found as a
@@ -212,15 +214,22 @@ enum quniform_contents {
         QUNIFORM_TEXTURE_CONFIG_P0_32,
 
         /**
-         * A reference to a texture config parameter 1 uniform.
+         * A reference to a V3D 3.x texture config parameter 1 uniform.
          *
          * This is a uniform implicitly loaded with a QPU_W_TMU* write, which
-         * defines texture width, height, filters, and wrap modes.  It will be
-         * found as a parameter to the second QOP_TEX_[STRB] instruction in a
-         * sequence.
+         * has the pointer to the indirect texture state.  Our data[] field
+         * will have a packed p1 value, but the address field will be just
+         * which texture unit's texture should be referenced.
          */
         QUNIFORM_TEXTURE_CONFIG_P1,
 
+        /* A a V3D 4.x texture config parameter.  The high 8 bits will be
+         * which texture or sampler is being sampled, and the driver must
+         * replace the address field with the appropriate address.
+         */
+        QUNIFORM_TMU_CONFIG_P0,
+        QUNIFORM_TMU_CONFIG_P1,
+
         QUNIFORM_TEXTURE_FIRST_LEVEL,
 
         QUNIFORM_TEXTURE_WIDTH,
@@ -229,8 +238,6 @@ enum quniform_contents {
         QUNIFORM_TEXTURE_ARRAY_SIZE,
         QUNIFORM_TEXTURE_LEVELS,
 
-        QUNIFORM_TEXTURE_MSAA_ADDR,
-
         QUNIFORM_UBO_ADDR,
 
         QUNIFORM_TEXRECT_SCALE_X,
@@ -238,10 +245,14 @@ enum quniform_contents {
 
         QUNIFORM_TEXTURE_BORDER_COLOR,
 
-        QUNIFORM_STENCIL,
-
         QUNIFORM_ALPHA_REF,
         QUNIFORM_SAMPLE_MASK,
+
+        /**
+         * Returns the the offset of the scratch buffer for register spilling.
+         */
+        QUNIFORM_SPILL_OFFSET,
+        QUNIFORM_SPILL_SIZE_PER_THREAD,
 };
 
 struct v3d_varying_slot {
@@ -290,17 +301,11 @@ struct v3d_key {
                 uint8_t swizzle[4];
                 uint8_t return_size;
                 uint8_t return_channels;
-                union {
-                        struct {
-                                unsigned compare_mode:1;
-                                unsigned compare_func:3;
-                                unsigned wrap_s:3;
-                                unsigned wrap_t:3;
-                        };
-                        struct {
-                                uint16_t msaa_width, msaa_height;
-                        };
-                };
+                unsigned compare_mode:1;
+                unsigned compare_func:3;
+                bool clamp_s:1;
+                bool clamp_t:1;
+                bool clamp_r:1;
         } tex[V3D_MAX_TEXTURE_SAMPLERS];
         uint8_t ucp_enables;
 };
@@ -318,10 +323,16 @@ struct v3d_fs_key {
         bool sample_alpha_to_coverage;
         bool sample_alpha_to_one;
         bool clamp_color;
+        bool shade_model_flat;
         uint8_t nr_cbufs;
         uint8_t swap_color_rb;
         /* Mask of which render targets need to be written as 32-bit floats */
         uint8_t f32_color_rb;
+        /* Masks of which render targets need to be written as ints/uints.
+         * Used by gallium to work around lost information in TGSI.
+         */
+        uint8_t int_color_rb;
+        uint8_t uint_color_rb;
         uint8_t alpha_test_func;
         uint8_t logicop_func;
         uint32_t point_sprite_mask;
@@ -376,6 +387,48 @@ struct qblock {
         /** @} */
 };
 
+/** Which util/list.h add mode we should use when inserting an instruction. */
+enum vir_cursor_mode {
+        vir_cursor_add,
+        vir_cursor_addtail,
+};
+
+/**
+ * Tracking structure for where new instructions should be inserted.  Create
+ * with one of the vir_after_inst()-style helper functions.
+ *
+ * This does not protect against removal of the block or instruction, so we
+ * have an assert in instruction removal to try to catch it.
+ */
+struct vir_cursor {
+        enum vir_cursor_mode mode;
+        struct list_head *link;
+};
+
+static inline struct vir_cursor
+vir_before_inst(struct qinst *inst)
+{
+        return (struct vir_cursor){ vir_cursor_addtail, &inst->link };
+}
+
+static inline struct vir_cursor
+vir_after_inst(struct qinst *inst)
+{
+        return (struct vir_cursor){ vir_cursor_add, &inst->link };
+}
+
+static inline struct vir_cursor
+vir_before_block(struct qblock *block)
+{
+        return (struct vir_cursor){ vir_cursor_add, &block->instructions };
+}
+
+static inline struct vir_cursor
+vir_after_block(struct qblock *block)
+{
+        return (struct vir_cursor){ vir_cursor_addtail, &block->instructions };
+}
+
 /**
  * Compiler state saved across compiler invocations, for any expensive global
  * setup.
@@ -383,7 +436,8 @@ struct qblock {
 struct v3d_compiler {
         const struct v3d_device_info *devinfo;
         struct ra_regs *regs;
-        unsigned int reg_class[3];
+        unsigned int reg_class_phys[3];
+        unsigned int reg_class_phys_or_acc[3];
 };
 
 struct v3d_compile {
@@ -418,16 +472,16 @@ struct v3d_compile {
         uint32_t uniforms_array_size;
 
         /* Booleans for whether the corresponding QFILE_VARY[i] is
-         * flat-shaded.  This doesn't count gl_FragColor flat-shading, which is
-         * controlled by shader->color_inputs and rasterizer->flatshade in the
-         * gallium driver.
+         * flat-shaded.  This includes gl_FragColor flat-shading, which is
+         * customized based on the shademodel_flat shader key.
          */
-        BITSET_WORD flat_shade_flags[BITSET_WORDS(V3D_MAX_FS_INPUTS)];
+        uint32_t flat_shade_flags[BITSET_WORDS(V3D_MAX_FS_INPUTS)];
 
-        /* Booleans for whether the corresponding QFILE_VARY[i] uses the
-         * default glShadeModel() behavior.
-         */
-        BITSET_WORD shade_model_flags[BITSET_WORDS(V3D_MAX_FS_INPUTS)];
+        uint32_t noperspective_flags[BITSET_WORDS(V3D_MAX_FS_INPUTS)];
+
+        uint32_t centroid_flags[BITSET_WORDS(V3D_MAX_FS_INPUTS)];
+
+        bool uses_center_w;
 
         struct v3d_ubo_range *ubo_ranges;
         bool *ubo_range_used;
@@ -461,6 +515,20 @@ struct v3d_compile {
         uint8_t vattr_sizes[V3D_MAX_VS_INPUTS];
         uint32_t num_vpm_writes;
 
+        /* Size in bytes of registers that have been spilled. This is how much
+         * space needs to be available in the spill BO per thread per QPU.
+         */
+        uint32_t spill_size;
+        /* Shader-db stats for register spilling. */
+        uint32_t spills, fills;
+        /**
+         * Register spilling's per-thread base address, shared between each
+         * spill/fill's addressing calculations.
+         */
+        struct qreg spill_base;
+        /* Bit vector of which temps may be spilled */
+        BITSET_WORD *spillable;
+
         /**
          * Array of the VARYING_SLOT_* of all FS QFILE_VARY reads.
          *
@@ -483,6 +551,7 @@ struct v3d_compile {
 
         /* Live ranges of temps. */
         int *temp_start, *temp_end;
+        bool live_intervals_valid;
 
         uint32_t *uniform_data;
         enum quniform_contents *uniform_contents;
@@ -497,6 +566,7 @@ struct v3d_compile {
         struct qreg undef;
         uint32_t num_temps;
 
+        struct vir_cursor cursor;
         struct list_head blocks;
         int next_block_index;
         struct qblock *cur_block;
@@ -527,12 +597,16 @@ struct v3d_compile {
         uint32_t program_id;
         uint32_t variant_id;
 
-        /* Set to compile program in threaded FS mode, where SIG_THREAD_SWITCH
-         * is used to hide texturing latency at the cost of limiting ourselves
-         * to the bottom half of physical reg space.
+        /* Set to compile program in in 1x, 2x, or 4x threaded mode, where
+         * SIG_THREAD_SWITCH is used to hide texturing latency at the cost of
+         * limiting ourselves to the part of the physical reg space.
+         *
+         * On V3D 3.x, 2x or 4x divide the physical reg space by 2x or 4x.  On
+         * V3D 4.x, all shaders are 2x threaded, and 4x only divides the
+         * physical reg space in half.
          */
-        bool fs_threaded;
-
+        uint8_t threads;
+        struct qinst *last_thrsw;
         bool last_thrsw_at_top_level;
 
         bool failed;
@@ -550,9 +624,15 @@ struct v3d_prog_data {
         struct v3d_ubo_range *ubo_ranges;
         uint32_t num_ubo_ranges;
         uint32_t ubo_size;
+        uint32_t spill_size;
 
         uint8_t num_inputs;
+        uint8_t threads;
 
+        /* For threads > 1, whether the program should be dispatched in the
+         * after-final-THRSW state.
+         */
+        bool single_seg;
 };
 
 struct v3d_vs_prog_data {
@@ -568,6 +648,9 @@ struct v3d_vs_prog_data {
 
         /* Total number of components written, for the shader state record. */
         uint32_t vpm_output_size;
+
+        /* Value to be programmed in VCM_CACHE_SIZE. */
+        uint8_t vcm_cache_size;
 };
 
 struct v3d_fs_prog_data {
@@ -575,17 +658,20 @@ struct v3d_fs_prog_data {
 
         struct v3d_varying_slot input_slots[V3D_MAX_FS_INPUTS];
 
-        /* Bitmask for whether the corresponding input is flat-shaded,
-         * independent of rasterizer (gl_FragColor) flat-shading.
-         */
-        BITSET_WORD flat_shade_flags[BITSET_WORDS(V3D_MAX_FS_INPUTS)];
-        /* Bitmask for whether the corresponding input uses the default
-         * glShadeModel() behavior.
+        /* Array of flat shade flags.
+         *
+         * Each entry is only 24 bits (high 8 bits 0), to match the hardware
+         * packet layout.
          */
-        BITSET_WORD shade_model_flags[BITSET_WORDS(V3D_MAX_FS_INPUTS)];
+        uint32_t flat_shade_flags[((V3D_MAX_FS_INPUTS - 1) / 24) + 1];
+
+        uint32_t noperspective_flags[((V3D_MAX_FS_INPUTS - 1) / 24) + 1];
+
+        uint32_t centroid_flags[((V3D_MAX_FS_INPUTS - 1) / 24) + 1];
 
         bool writes_z;
         bool discard;
+        bool uses_center_w;
 };
 
 /* Special nir_load_input intrinsic index for loading the current TLB
@@ -644,6 +730,7 @@ void vir_set_unpack(struct qinst *inst, int src,
                     enum v3d_qpu_input_unpack unpack);
 
 struct qreg vir_get_temp(struct v3d_compile *c);
+void vir_emit_last_thrsw(struct v3d_compile *c);
 void vir_calculate_live_intervals(struct v3d_compile *c);
 bool vir_has_implicit_uniform(struct qinst *inst);
 int vir_get_implicit_uniform_src(struct qinst *inst);
@@ -658,10 +745,14 @@ bool vir_is_add(struct qinst *inst);
 bool vir_is_mul(struct qinst *inst);
 bool vir_is_float_input(struct qinst *inst);
 bool vir_depends_on_flags(struct qinst *inst);
-bool vir_writes_r3(struct qinst *inst);
-bool vir_writes_r4(struct qinst *inst);
+bool vir_writes_r3(const struct v3d_device_info *devinfo, struct qinst *inst);
+bool vir_writes_r4(const struct v3d_device_info *devinfo, struct qinst *inst);
 struct qreg vir_follow_movs(struct v3d_compile *c, struct qreg reg);
 uint8_t vir_channels_written(struct qinst *inst);
+struct qreg ntq_get_src(struct v3d_compile *c, nir_src src, int i);
+void ntq_store_dest(struct v3d_compile *c, nir_dest *dest, int chan,
+                    struct qreg result);
+void vir_emit_thrsw(struct v3d_compile *c);
 
 void vir_dump(struct v3d_compile *c);
 void vir_dump_inst(struct v3d_compile *c, struct qinst *inst);
@@ -681,10 +772,15 @@ void v3d_nir_lower_io(nir_shader *s, struct v3d_compile *c);
 void v3d_nir_lower_txf_ms(nir_shader *s, struct v3d_compile *c);
 void vir_lower_uniforms(struct v3d_compile *c);
 
-void v3d_vir_to_qpu(struct v3d_compile *c);
+void v3d33_vir_vpm_read_setup(struct v3d_compile *c, int num_components);
+void v3d33_vir_vpm_write_setup(struct v3d_compile *c);
+void v3d33_vir_emit_tex(struct v3d_compile *c, nir_tex_instr *instr);
+void v3d40_vir_emit_tex(struct v3d_compile *c, nir_tex_instr *instr);
+
+void v3d_vir_to_qpu(struct v3d_compile *c, struct qpu_reg *temp_registers);
 uint32_t v3d_qpu_schedule_instructions(struct v3d_compile *c);
 void qpu_validate(struct v3d_compile *c);
-struct qpu_reg *v3d_register_allocate(struct v3d_compile *c);
+struct qpu_reg *v3d_register_allocate(struct v3d_compile *c, bool *spilled);
 bool vir_init_reg_sets(struct v3d_compiler *compiler);
 
 void vir_PF(struct v3d_compile *c, struct qreg src, enum v3d_qpu_pf pf);
@@ -751,6 +847,14 @@ vir_##name##_dest(struct v3d_compile *c, struct qreg dest,               \
         return vir_emit_nondef(c, vir_inst(op, dest, a, b));     \
 }
 
+#define VIR_NODST_0(name, vir_inst, op)                                 \
+static inline struct qinst *                                            \
+vir_##name(struct v3d_compile *c)                                       \
+{                                                                       \
+        return vir_emit_nondef(c, vir_inst(op, c->undef,                \
+                                           c->undef, c->undef));        \
+}
+
 #define VIR_NODST_1(name, vir_inst, op)                                               \
 static inline struct qinst *                                            \
 vir_##name(struct v3d_compile *c, struct qreg a)                        \
@@ -767,6 +871,33 @@ vir_##name(struct v3d_compile *c, struct qreg a, struct qreg b)         \
                                            a, b));                      \
 }
 
+#define VIR_SFU(name)                                                      \
+static inline struct qreg                                                \
+vir_##name(struct v3d_compile *c, struct qreg a)                         \
+{                                                                        \
+        if (c->devinfo->ver >= 41) {                                     \
+                return vir_emit_def(c, vir_add_inst(V3D_QPU_A_##name,    \
+                                                    c->undef,            \
+                                                    a, c->undef));       \
+        } else {                                                         \
+                vir_FMOV_dest(c, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_##name), a); \
+                return vir_FMOV(c, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_R4)); \
+        }                                                                \
+}                                                                        \
+static inline struct qinst *                                             \
+vir_##name##_dest(struct v3d_compile *c, struct qreg dest,               \
+                  struct qreg a)                                         \
+{                                                                        \
+        if (c->devinfo->ver >= 41) {                                     \
+                return vir_emit_nondef(c, vir_add_inst(V3D_QPU_A_##name, \
+                                                       dest,             \
+                                                       a, c->undef));    \
+        } else {                                                         \
+                vir_FMOV_dest(c, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_##name), a); \
+                return vir_FMOV_dest(c, dest, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_R4)); \
+        }                                                                \
+}
+
 #define VIR_A_ALU2(name) VIR_ALU2(name, vir_add_inst, V3D_QPU_A_##name)
 #define VIR_M_ALU2(name) VIR_ALU2(name, vir_mul_inst, V3D_QPU_M_##name)
 #define VIR_A_ALU1(name) VIR_ALU1(name, vir_add_inst, V3D_QPU_A_##name)
@@ -777,6 +908,7 @@ vir_##name(struct v3d_compile *c, struct qreg a, struct qreg b)         \
 #define VIR_M_NODST_2(name) VIR_NODST_2(name, vir_mul_inst, V3D_QPU_M_##name)
 #define VIR_A_NODST_1(name) VIR_NODST_1(name, vir_add_inst, V3D_QPU_A_##name)
 #define VIR_M_NODST_1(name) VIR_NODST_1(name, vir_mul_inst, V3D_QPU_M_##name)
+#define VIR_A_NODST_0(name) VIR_NODST_0(name, vir_add_inst, V3D_QPU_A_##name)
 
 VIR_A_ALU2(FADD)
 VIR_A_ALU2(VFPACK)
@@ -799,15 +931,19 @@ VIR_A_ALU2(OR)
 VIR_A_ALU2(XOR)
 VIR_A_ALU2(VADD)
 VIR_A_ALU2(VSUB)
+VIR_A_NODST_2(STVPMV)
 VIR_A_ALU1(NOT)
 VIR_A_ALU1(NEG)
 VIR_A_ALU1(FLAPUSH)
 VIR_A_ALU1(FLBPUSH)
-VIR_A_ALU1(FLBPOP)
+VIR_A_ALU1(FLPOP)
 VIR_A_ALU1(SETMSF)
 VIR_A_ALU1(SETREVF)
-VIR_A_ALU1(TIDX)
-VIR_A_ALU1(EIDX)
+VIR_A_ALU0(TIDX)
+VIR_A_ALU0(EIDX)
+VIR_A_ALU1(LDVPMV_IN)
+VIR_A_ALU1(LDVPMV_OUT)
+VIR_A_ALU0(TMUWT)
 
 VIR_A_ALU0(FXCD)
 VIR_A_ALU0(XCD)
@@ -816,6 +952,7 @@ VIR_A_ALU0(YCD)
 VIR_A_ALU0(MSF)
 VIR_A_ALU0(REVF)
 VIR_A_NODST_1(VPMSETUP)
+VIR_A_NODST_0(VPMWT)
 VIR_A_ALU2(FCMP)
 VIR_A_ALU2(VFMAX)
 
@@ -843,6 +980,13 @@ VIR_M_NODST_2(MULTOP)
 VIR_M_ALU1(MOV)
 VIR_M_ALU1(FMOV)
 
+VIR_SFU(RECIP)
+VIR_SFU(RSQRT)
+VIR_SFU(EXP)
+VIR_SFU(LOG)
+VIR_SFU(SIN)
+VIR_SFU(RSQRT2)
+
 static inline struct qinst *
 vir_MOV_cond(struct v3d_compile *c, enum v3d_qpu_cond cond,
              struct qreg dest, struct qreg src)
@@ -862,18 +1006,35 @@ vir_SEL(struct v3d_compile *c, enum v3d_qpu_cond cond,
         return t;
 }
 
-static inline void
-vir_VPM_WRITE(struct v3d_compile *c, struct qreg val)
-{
-        vir_MOV_dest(c, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_VPM), val);
-}
-
 static inline struct qinst *
 vir_NOP(struct v3d_compile *c)
 {
         return vir_emit_nondef(c, vir_add_inst(V3D_QPU_A_NOP,
                                                c->undef, c->undef, c->undef));
 }
+
+static inline struct qreg
+vir_LDTMU(struct v3d_compile *c)
+{
+        if (c->devinfo->ver >= 41) {
+                struct qinst *ldtmu = vir_add_inst(V3D_QPU_A_NOP, c->undef,
+                                                   c->undef, c->undef);
+                ldtmu->qpu.sig.ldtmu = true;
+
+                return vir_emit_def(c, ldtmu);
+        } else {
+                vir_NOP(c)->qpu.sig.ldtmu = true;
+                return vir_MOV(c, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_R4));
+        }
+}
+
+static inline struct qreg
+vir_UMUL(struct v3d_compile *c, struct qreg src0, struct qreg src1)
+{
+        vir_MULTOP(c, src0, src1);
+        return vir_UMUL24(c, src0, src1);
+}
+
 /*
 static inline struct qreg
 vir_LOAD_IMM(struct v3d_compile *c, uint32_t val)
diff --git a/lib/mesa/src/broadcom/compiler/v3d_nir_lower_io.c b/lib/mesa/src/broadcom/compiler/v3d_nir_lower_io.c
index 9cdcc0219..10bc25811 100644
--- a/lib/mesa/src/broadcom/compiler/v3d_nir_lower_io.c
+++ b/lib/mesa/src/broadcom/compiler/v3d_nir_lower_io.c
@@ -52,44 +52,6 @@ replace_intrinsic_with_vec(nir_builder *b, nir_intrinsic_instr *intr,
 }
 
 static void
-v3d_nir_lower_output(struct v3d_compile *c, nir_builder *b,
-                     nir_intrinsic_instr *intr)
-{
-        nir_variable *output_var = NULL;
-        nir_foreach_variable(var, &c->s->outputs) {
-                if (var->data.driver_location == nir_intrinsic_base(intr)) {
-                        output_var = var;
-                        break;
-                }
-        }
-        assert(output_var);
-
-        if (c->vs_key) {
-                int slot = output_var->data.location;
-                bool used = false;
-
-                switch (slot) {
-                case VARYING_SLOT_PSIZ:
-                case VARYING_SLOT_POS:
-                        used = true;
-                        break;
-
-                default:
-                        for (int i = 0; i < c->vs_key->num_fs_inputs; i++) {
-                                if (v3d_slot_get_slot(c->vs_key->fs_inputs[i]) == slot) {
-                                        used = true;
-                                        break;
-                                }
-                        }
-                        break;
-                }
-
-                if (!used)
-                        nir_instr_remove(&intr->instr);
-        }
-}
-
-static void
 v3d_nir_lower_uniform(struct v3d_compile *c, nir_builder *b,
                       nir_intrinsic_instr *intr)
 {
@@ -135,10 +97,6 @@ v3d_nir_lower_io_instr(struct v3d_compile *c, nir_builder *b,
         case nir_intrinsic_load_input:
                 break;
 
-        case nir_intrinsic_store_output:
-                v3d_nir_lower_output(c, b, intr);
-                break;
-
         case nir_intrinsic_load_uniform:
                 v3d_nir_lower_uniform(c, b, intr);
                 break;
diff --git a/lib/mesa/src/broadcom/compiler/v3d_nir_lower_txf_ms.c b/lib/mesa/src/broadcom/compiler/v3d_nir_lower_txf_ms.c
new file mode 100644
index 000000000..ffaee0d76
--- /dev/null
+++ b/lib/mesa/src/broadcom/compiler/v3d_nir_lower_txf_ms.c
@@ -0,0 +1,93 @@
+/*
+ * Copyright © 2015 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "v3d_compiler.h"
+#include "compiler/nir/nir_builder.h"
+
+/** @file v3d_nir_lower_txf_ms.c
+ *
+ * V3D's MSAA surfaces are laid out in UIF textures where each pixel is a 2x2
+ * quad in the texture.  This pass lowers the txf_ms with a ms_index source to
+ * a plain txf with the sample_index pulling out the correct texel from the
+ * 2x2 quad.
+ */
+
+#define V3D_MAX_SAMPLES 4
+
+static void
+vc4_nir_lower_txf_ms_instr(struct v3d_compile *c, nir_builder *b,
+                           nir_tex_instr *instr)
+{
+        if (instr->op != nir_texop_txf_ms)
+                return;
+
+        b->cursor = nir_before_instr(&instr->instr);
+
+        int coord_index = nir_tex_instr_src_index(instr, nir_tex_src_coord);
+        int sample_index = nir_tex_instr_src_index(instr, nir_tex_src_ms_index);
+        nir_ssa_def *coord = instr->src[coord_index].src.ssa;
+        nir_ssa_def *sample = instr->src[sample_index].src.ssa;
+
+        nir_ssa_def *one = nir_imm_int(b, 1);
+        coord = nir_ishl(b, coord, nir_imm_int(b, 1));
+        coord = nir_vec2(b,
+                         nir_iadd(b,
+                                  nir_channel(b, coord, 0),
+                                  nir_iand(b, sample, one)),
+                         nir_iadd(b,
+                                  nir_channel(b, coord, 1),
+                                  nir_iand(b, nir_ushr(b, sample, one), one)));
+
+        nir_instr_rewrite_src(&instr->instr,
+                              &instr->src[nir_tex_src_coord].src,
+                              nir_src_for_ssa(coord));
+        nir_tex_instr_remove_src(instr, sample_index);
+        instr->op = nir_texop_txf;
+        instr->sampler_dim = GLSL_SAMPLER_DIM_2D;
+}
+
+void
+v3d_nir_lower_txf_ms(nir_shader *s, struct v3d_compile *c)
+{
+        nir_foreach_function(function, s) {
+                if (!function->impl)
+                        continue;
+
+                nir_builder b;
+                nir_builder_init(&b, function->impl);
+
+                nir_foreach_block(block, function->impl) {
+                        nir_foreach_instr_safe(instr, block) {
+                                if (instr->type != nir_instr_type_tex)
+                                        continue;
+
+                                vc4_nir_lower_txf_ms_instr(c, &b,
+                                                           nir_instr_as_tex(instr));
+                        }
+                }
+
+                nir_metadata_preserve(function->impl,
+                                      nir_metadata_block_index |
+                                      nir_metadata_dominance);
+        }
+}
diff --git a/lib/mesa/src/broadcom/compiler/vir.c b/lib/mesa/src/broadcom/compiler/vir.c
index 99b31841b..3f5a28d1b 100644
--- a/lib/mesa/src/broadcom/compiler/vir.c
+++ b/lib/mesa/src/broadcom/compiler/vir.c
@@ -21,6 +21,7 @@
  * IN THE SOFTWARE.
  */
 
+#include "broadcom/common/v3d_device_info.h"
 #include "v3d_compiler.h"
 
 int
@@ -73,6 +74,8 @@ vir_has_implicit_uniform(struct qinst *inst)
 int
 vir_get_implicit_uniform_src(struct qinst *inst)
 {
+        if (!vir_has_implicit_uniform(inst))
+                return -1;
         return vir_get_nsrc(inst) - 1;
 }
 
@@ -91,6 +94,11 @@ vir_has_side_effects(struct v3d_compile *c, struct qinst *inst)
                 case V3D_QPU_A_SETREVF:
                 case V3D_QPU_A_SETMSF:
                 case V3D_QPU_A_VPMSETUP:
+                case V3D_QPU_A_STVPMV:
+                case V3D_QPU_A_STVPMD:
+                case V3D_QPU_A_STVPMP:
+                case V3D_QPU_A_VPMWT:
+                case V3D_QPU_A_TMUWT:
                         return true;
                 default:
                         break;
@@ -104,8 +112,12 @@ vir_has_side_effects(struct v3d_compile *c, struct qinst *inst)
                 }
         }
 
-        if (inst->qpu.sig.ldtmu)
+        if (inst->qpu.sig.ldtmu ||
+            inst->qpu.sig.ldvary ||
+            inst->qpu.sig.wrtmuc ||
+            inst->qpu.sig.thrsw) {
                 return true;
+        }
 
         return false;
 }
@@ -183,6 +195,11 @@ vir_is_tex(struct qinst *inst)
         if (inst->dst.file == QFILE_MAGIC)
                 return v3d_qpu_magic_waddr_is_tmu(inst->dst.index);
 
+        if (inst->qpu.type == V3D_QPU_INSTR_TYPE_ALU &&
+            inst->qpu.alu.add.op == V3D_QPU_A_TMUWT) {
+                return true;
+        }
+
         return false;
 }
 
@@ -198,11 +215,10 @@ vir_depends_on_flags(struct qinst *inst)
 }
 
 bool
-vir_writes_r3(struct qinst *inst)
+vir_writes_r3(const struct v3d_device_info *devinfo, struct qinst *inst)
 {
         for (int i = 0; i < vir_get_nsrc(inst); i++) {
                 switch (inst->src[i].file) {
-                case QFILE_VARY:
                 case QFILE_VPM:
                         return true;
                 default:
@@ -210,11 +226,18 @@ vir_writes_r3(struct qinst *inst)
                 }
         }
 
+        if (devinfo->ver < 41 && (inst->qpu.sig.ldvary ||
+                                  inst->qpu.sig.ldtlb ||
+                                  inst->qpu.sig.ldtlbu ||
+                                  inst->qpu.sig.ldvpm)) {
+                return true;
+        }
+
         return false;
 }
 
 bool
-vir_writes_r4(struct qinst *inst)
+vir_writes_r4(const struct v3d_device_info *devinfo, struct qinst *inst)
 {
         switch (inst->dst.file) {
         case QFILE_MAGIC:
@@ -231,7 +254,7 @@ vir_writes_r4(struct qinst *inst)
                 break;
         }
 
-        if (inst->qpu.sig.ldtmu)
+        if (devinfo->ver < 41 && inst->qpu.sig.ldtmu)
                 return true;
 
         return false;
@@ -339,10 +362,17 @@ vir_get_temp(struct v3d_compile *c)
         if (c->num_temps > c->defs_array_size) {
                 uint32_t old_size = c->defs_array_size;
                 c->defs_array_size = MAX2(old_size * 2, 16);
+
                 c->defs = reralloc(c, c->defs, struct qinst *,
                                    c->defs_array_size);
                 memset(&c->defs[old_size], 0,
                        sizeof(c->defs[0]) * (c->defs_array_size - old_size));
+
+                c->spillable = reralloc(c, c->spillable,
+                                        BITSET_WORD,
+                                        BITSET_WORDS(c->defs_array_size));
+                for (int i = old_size; i < c->defs_array_size; i++)
+                        BITSET_SET(c->spillable, i);
         }
 
         return reg;
@@ -403,11 +433,17 @@ vir_branch_inst(enum v3d_qpu_branch_cond cond, struct qreg src)
 static void
 vir_emit(struct v3d_compile *c, struct qinst *inst)
 {
-        list_addtail(&inst->link, &c->cur_block->instructions);
+        switch (c->cursor.mode) {
+        case vir_cursor_add:
+                list_add(&inst->link, c->cursor.link);
+                break;
+        case vir_cursor_addtail:
+                list_addtail(&inst->link, c->cursor.link);
+                break;
+        }
 
-        if (inst->dst.file == QFILE_MAGIC &&
-            inst->dst.index == V3D_QPU_WADDR_VPM)
-                c->num_vpm_writes++;
+        c->cursor = vir_after_inst(inst);
+        c->live_intervals_valid = false;
 }
 
 /* Updates inst to write to a new temporary, emits it, and notes the def. */
@@ -416,6 +452,16 @@ vir_emit_def(struct v3d_compile *c, struct qinst *inst)
 {
         assert(inst->dst.file == QFILE_NULL);
 
+        /* If we're emitting an instruction that's a def, it had better be
+         * writing a register.
+         */
+        if (inst->qpu.type == V3D_QPU_INSTR_TYPE_ALU) {
+                assert(inst->qpu.alu.add.op == V3D_QPU_A_NOP ||
+                       v3d_qpu_add_op_has_dst(inst->qpu.alu.add.op));
+                assert(inst->qpu.alu.mul.op == V3D_QPU_M_NOP ||
+                       v3d_qpu_mul_op_has_dst(inst->qpu.alu.mul.op));
+        }
+
         inst->dst = vir_get_temp(c);
 
         if (inst->dst.file == QFILE_TEMP)
@@ -457,6 +503,7 @@ void
 vir_set_emit_block(struct v3d_compile *c, struct qblock *block)
 {
         c->cur_block = block;
+        c->cursor = vir_after_block(block);
         list_addtail(&block->link, &c->blocks);
 }
 
@@ -520,6 +567,7 @@ vir_compile_init(const struct v3d_compiler *compiler,
         c->key = key;
         c->program_id = program_id;
         c->variant_id = variant_id;
+        c->threads = 4;
 
         s = nir_shader_clone(c, s);
         c->s = s;
@@ -537,11 +585,18 @@ vir_compile_init(const struct v3d_compiler *compiler,
         return c;
 }
 
+static int
+type_size_vec4(const struct glsl_type *type)
+{
+        return glsl_count_attribute_slots(type, false);
+}
+
 static void
 v3d_lower_nir(struct v3d_compile *c)
 {
         struct nir_lower_tex_options tex_options = {
-                .lower_rect = false, /* XXX */
+                .lower_txd = true,
+                .lower_rect = false, /* XXX: Use this on V3D 3.x */
                 .lower_txp = ~0,
                 /* Apply swizzles to all samplers. */
                 .swizzle_result = ~0,
@@ -553,6 +608,13 @@ v3d_lower_nir(struct v3d_compile *c)
         for (int i = 0; i < ARRAY_SIZE(c->key->tex); i++) {
                 for (int j = 0; j < 4; j++)
                         tex_options.swizzles[i][j] = c->key->tex[i].swizzle[j];
+
+                if (c->key->tex[i].clamp_s)
+                        tex_options.saturate_s |= 1 << i;
+                if (c->key->tex[i].clamp_t)
+                        tex_options.saturate_t |= 1 << i;
+                if (c->key->tex[i].clamp_r)
+                        tex_options.saturate_r |= 1 << i;
         }
 
         NIR_PASS_V(c->s, nir_lower_tex, &tex_options);
@@ -562,6 +624,7 @@ static void
 v3d_lower_nir_late(struct v3d_compile *c)
 {
         NIR_PASS_V(c->s, v3d_nir_lower_io, c);
+        NIR_PASS_V(c->s, v3d_nir_lower_txf_ms, c);
         NIR_PASS_V(c->s, nir_lower_idiv);
 }
 
@@ -620,6 +683,10 @@ static void
 v3d_set_prog_data(struct v3d_compile *c,
                   struct v3d_prog_data *prog_data)
 {
+        prog_data->threads = c->threads;
+        prog_data->single_seg = !c->last_thrsw;
+        prog_data->spill_size = c->spill_size;
+
         v3d_set_prog_data_uniforms(c, prog_data);
         v3d_set_prog_data_ubo(c, prog_data);
 }
@@ -652,6 +719,26 @@ uint64_t *v3d_compile_vs(const struct v3d_compiler *compiler,
 
         c->vs_key = key;
 
+        /* Split our I/O vars and dead code eliminate the unused
+         * components.
+         */
+        NIR_PASS_V(c->s, nir_lower_io_to_scalar_early,
+                   nir_var_shader_in | nir_var_shader_out);
+        uint64_t used_outputs[4] = {0};
+        for (int i = 0; i < c->vs_key->num_fs_inputs; i++) {
+                int slot = v3d_slot_get_slot(c->vs_key->fs_inputs[i]);
+                int comp = v3d_slot_get_component(c->vs_key->fs_inputs[i]);
+                used_outputs[comp] |= 1ull << slot;
+        }
+        NIR_PASS_V(c->s, nir_remove_unused_io_vars,
+                   &c->s->outputs, used_outputs, NULL); /* demotes to globals */
+        NIR_PASS_V(c->s, nir_lower_global_vars_to_local);
+        v3d_optimize_nir(c->s);
+        NIR_PASS_V(c->s, nir_remove_dead_variables, nir_var_shader_in);
+        NIR_PASS_V(c->s, nir_lower_io, nir_var_shader_in | nir_var_shader_out,
+                   type_size_vec4,
+                   (nir_lower_io_options)0);
+
         v3d_lower_nir(c);
 
         if (key->clamp_color)
@@ -685,15 +772,39 @@ uint64_t *v3d_compile_vs(const struct v3d_compiler *compiler,
                 prog_data->vpm_input_size += c->vattr_sizes[i];
         }
 
-        /* Input/output segment size are in 8x32-bit multiples. */
-        prog_data->vpm_input_size = align(prog_data->vpm_input_size, 8) / 8;
-        prog_data->vpm_output_size = align(c->num_vpm_writes, 8) / 8;
-
         prog_data->uses_vid = (s->info.system_values_read &
                                (1ull << SYSTEM_VALUE_VERTEX_ID));
         prog_data->uses_iid = (s->info.system_values_read &
                                (1ull << SYSTEM_VALUE_INSTANCE_ID));
 
+        if (prog_data->uses_vid)
+                prog_data->vpm_input_size++;
+        if (prog_data->uses_iid)
+                prog_data->vpm_input_size++;
+
+        /* Input/output segment size are in sectors (8 rows of 32 bits per
+         * channel).
+         */
+        prog_data->vpm_input_size = align(prog_data->vpm_input_size, 8) / 8;
+        prog_data->vpm_output_size = align(c->num_vpm_writes, 8) / 8;
+
+        /* Compute VCM cache size.  We set up our program to take up less than
+         * half of the VPM, so that any set of bin and render programs won't
+         * run out of space.  We need space for at least one input segment,
+         * and then allocate the rest to output segments (one for the current
+         * program, the rest to VCM).  The valid range of the VCM cache size
+         * field is 1-4 16-vertex batches, but GFXH-1744 limits us to 2-4
+         * batches.
+         */
+        assert(c->devinfo->vpm_size);
+        int sector_size = 16 * sizeof(uint32_t) * 8;
+        int vpm_size_in_sectors = c->devinfo->vpm_size / sector_size;
+        int half_vpm = vpm_size_in_sectors / 2;
+        int vpm_output_sectors = half_vpm - prog_data->vpm_input_size;
+        int vpm_output_batches = vpm_output_sectors / prog_data->vpm_output_size;
+        assert(vpm_output_batches >= 2);
+        prog_data->vcm_cache_size = CLAMP(vpm_output_batches - 1, 2, 4);
+
         return v3d_return_qpu_insts(c, final_assembly_size);
 }
 
@@ -705,10 +816,48 @@ v3d_set_fs_prog_data_inputs(struct v3d_compile *c,
         memcpy(prog_data->input_slots, c->input_slots,
                c->num_inputs * sizeof(*c->input_slots));
 
-        memcpy(prog_data->flat_shade_flags, c->flat_shade_flags,
-               sizeof(c->flat_shade_flags));
-        memcpy(prog_data->shade_model_flags, c->shade_model_flags,
-               sizeof(c->shade_model_flags));
+        STATIC_ASSERT(ARRAY_SIZE(prog_data->flat_shade_flags) >
+                      (V3D_MAX_FS_INPUTS - 1) / 24);
+        for (int i = 0; i < V3D_MAX_FS_INPUTS; i++) {
+                if (BITSET_TEST(c->flat_shade_flags, i))
+                        prog_data->flat_shade_flags[i / 24] |= 1 << (i % 24);
+
+                if (BITSET_TEST(c->noperspective_flags, i))
+                        prog_data->noperspective_flags[i / 24] |= 1 << (i % 24);
+
+                if (BITSET_TEST(c->centroid_flags, i))
+                        prog_data->centroid_flags[i / 24] |= 1 << (i % 24);
+        }
+}
+
+static void
+v3d_fixup_fs_output_types(struct v3d_compile *c)
+{
+        nir_foreach_variable(var, &c->s->outputs) {
+                uint32_t mask = 0;
+
+                switch (var->data.location) {
+                case FRAG_RESULT_COLOR:
+                        mask = ~0;
+                        break;
+                case FRAG_RESULT_DATA0:
+                case FRAG_RESULT_DATA1:
+                case FRAG_RESULT_DATA2:
+                case FRAG_RESULT_DATA3:
+                        mask = 1 << (var->data.location - FRAG_RESULT_DATA0);
+                        break;
+                }
+
+                if (c->fs_key->int_color_rb & mask) {
+                        var->type =
+                                glsl_vector_type(GLSL_TYPE_INT,
+                                                 glsl_get_components(var->type));
+                } else if (c->fs_key->uint_color_rb & mask) {
+                        var->type =
+                                glsl_vector_type(GLSL_TYPE_UINT,
+                                                 glsl_get_components(var->type));
+                }
+        }
 }
 
 uint64_t *v3d_compile_fs(const struct v3d_compiler *compiler,
@@ -723,6 +872,9 @@ uint64_t *v3d_compile_fs(const struct v3d_compiler *compiler,
 
         c->fs_key = key;
 
+        if (key->int_color_rb || key->uint_color_rb)
+                v3d_fixup_fs_output_types(c);
+
         v3d_lower_nir(c);
 
         if (key->light_twoside)
@@ -754,7 +906,9 @@ uint64_t *v3d_compile_fs(const struct v3d_compiler *compiler,
         v3d_set_fs_prog_data_inputs(c, prog_data);
         prog_data->writes_z = (c->s->info.outputs_written &
                                (1 << FRAG_RESULT_DEPTH));
-        prog_data->discard = c->s->info.fs.uses_discard;
+        prog_data->discard = (c->s->info.fs.uses_discard ||
+                              c->fs_key->sample_alpha_to_coverage);
+        prog_data->uses_center_w = c->uses_center_w;
 
         return v3d_return_qpu_insts(c, final_assembly_size);
 }
@@ -765,8 +919,12 @@ vir_remove_instruction(struct v3d_compile *c, struct qinst *qinst)
         if (qinst->dst.file == QFILE_TEMP)
                 c->defs[qinst->dst.index] = NULL;
 
+        assert(&qinst->link != c->cursor.link);
+
         list_del(&qinst->link);
         free(qinst);
+
+        c->live_intervals_valid = false;
 }
 
 struct qreg
@@ -792,6 +950,10 @@ vir_follow_movs(struct v3d_compile *c, struct qreg reg)
 void
 vir_compile_destroy(struct v3d_compile *c)
 {
+        /* Defuse the assert that we aren't removing the cursor's instruction.
+         */
+        c->cursor.link = NULL;
+
         vir_for_each_block(block, c) {
                 while (!list_empty(&block->instructions)) {
                         struct qinst *qinst =
@@ -836,20 +998,40 @@ vir_uniform(struct v3d_compile *c,
         return vir_reg(QFILE_UNIF, uniform);
 }
 
+static bool
+vir_can_set_flags(struct v3d_compile *c, struct qinst *inst)
+{
+        if (c->devinfo->ver >= 40 && (v3d_qpu_reads_vpm(&inst->qpu) ||
+                                      v3d_qpu_uses_sfu(&inst->qpu))) {
+                return false;
+        }
+
+        return true;
+}
+
 void
 vir_PF(struct v3d_compile *c, struct qreg src, enum v3d_qpu_pf pf)
 {
         struct qinst *last_inst = NULL;
 
-        if (!list_empty(&c->cur_block->instructions))
+        if (!list_empty(&c->cur_block->instructions)) {
                 last_inst = (struct qinst *)c->cur_block->instructions.prev;
 
+                /* Can't stuff the PF into the last last inst if our cursor
+                 * isn't pointing after it.
+                 */
+                struct vir_cursor after_inst = vir_after_inst(last_inst);
+                if (c->cursor.mode != after_inst.mode ||
+                    c->cursor.link != after_inst.link)
+                        last_inst = NULL;
+        }
+
         if (src.file != QFILE_TEMP ||
             !c->defs[src.index] ||
-            last_inst != c->defs[src.index]) {
+            last_inst != c->defs[src.index] ||
+            !vir_can_set_flags(c, last_inst)) {
                 /* XXX: Make the MOV be the appropriate type */
                 last_inst = vir_MOV_dest(c, vir_reg(QFILE_NULL, 0), src);
-                last_inst = (struct qinst *)c->cur_block->instructions.prev;
         }
 
         vir_set_pf(last_inst, pf);
@@ -880,6 +1062,7 @@ vir_optimize(struct v3d_compile *c)
 
                 OPTPASS(vir_opt_copy_propagate);
                 OPTPASS(vir_opt_dead_code);
+                OPTPASS(vir_opt_small_immediates);
 
                 if (!progress)
                         break;
diff --git a/lib/mesa/src/broadcom/compiler/vir_dump.c b/lib/mesa/src/broadcom/compiler/vir_dump.c
index ad5c061a1..c43578302 100644
--- a/lib/mesa/src/broadcom/compiler/vir_dump.c
+++ b/lib/mesa/src/broadcom/compiler/vir_dump.c
@@ -21,14 +21,15 @@
  * IN THE SOFTWARE.
  */
 
+#include "broadcom/common/v3d_device_info.h"
 #include "v3d_compiler.h"
 
 static void
-vir_print_reg(struct v3d_compile *c, struct qreg reg)
+vir_print_reg(struct v3d_compile *c, const struct qinst *inst,
+              struct qreg reg)
 {
         static const char *files[] = {
                 [QFILE_TEMP] = "t",
-                [QFILE_VARY] = "v",
                 [QFILE_UNIF] = "u",
                 [QFILE_TLB] = "tlb",
                 [QFILE_TLBU] = "tlbu",
@@ -58,12 +59,20 @@ vir_print_reg(struct v3d_compile *c, struct qreg reg)
                 fprintf(stderr, "%s", v3d_qpu_magic_waddr_name(reg.index));
                 break;
 
-        case QFILE_SMALL_IMM:
-                if ((int)reg.index >= -16 && (int)reg.index <= 15)
-                        fprintf(stderr, "%d", reg.index);
+        case QFILE_SMALL_IMM: {
+                uint32_t unpacked;
+                bool ok = v3d_qpu_small_imm_unpack(c->devinfo,
+                                                   inst->qpu.raddr_b,
+                                                   &unpacked);
+                assert(ok); (void) ok;
+
+                if ((int)inst->qpu.raddr_b >= -16 &&
+                    (int)inst->qpu.raddr_b <= 15)
+                        fprintf(stderr, "%d", unpacked);
                 else
-                        fprintf(stderr, "%f", uif(reg.index));
+                        fprintf(stderr, "%f", uif(unpacked));
                 break;
+        }
 
         case QFILE_VPM:
                 fprintf(stderr, "vpm%d.%d",
@@ -71,6 +80,7 @@ vir_print_reg(struct v3d_compile *c, struct qreg reg)
                 break;
 
         case QFILE_TLB:
+        case QFILE_TLBU:
                 fprintf(stderr, "%s", files[reg.file]);
                 break;
 
@@ -146,20 +156,60 @@ vir_print_reg(struct v3d_compile *c, struct qreg reg)
 }
 
 static void
+vir_dump_sig_addr(const struct v3d_device_info *devinfo,
+                  const struct v3d_qpu_instr *instr)
+{
+        if (devinfo->ver < 41)
+                return;
+
+        if (!instr->sig_magic)
+                fprintf(stderr, ".rf%d", instr->sig_addr);
+        else {
+                const char *name = v3d_qpu_magic_waddr_name(instr->sig_addr);
+                if (name)
+                        fprintf(stderr, ".%s", name);
+                else
+                        fprintf(stderr, ".UNKNOWN%d", instr->sig_addr);
+        }
+}
+
+static void
 vir_dump_sig(struct v3d_compile *c, struct qinst *inst)
 {
         struct v3d_qpu_sig *sig = &inst->qpu.sig;
 
         if (sig->thrsw)
                 fprintf(stderr, "; thrsw");
-        if (sig->ldvary)
+        if (sig->ldvary) {
                 fprintf(stderr, "; ldvary");
+                vir_dump_sig_addr(c->devinfo, &inst->qpu);
+        }
         if (sig->ldvpm)
                 fprintf(stderr, "; ldvpm");
-        if (sig->ldtmu)
+        if (sig->ldtmu) {
                 fprintf(stderr, "; ldtmu");
+                vir_dump_sig_addr(c->devinfo, &inst->qpu);
+        }
+        if (sig->ldtlb) {
+                fprintf(stderr, "; ldtlb");
+                vir_dump_sig_addr(c->devinfo, &inst->qpu);
+        }
+        if (sig->ldtlbu) {
+                fprintf(stderr, "; ldtlbu");
+                vir_dump_sig_addr(c->devinfo, &inst->qpu);
+        }
         if (sig->ldunif)
                 fprintf(stderr, "; ldunif");
+        if (sig->ldunifrf) {
+                fprintf(stderr, "; ldunifrf");
+                vir_dump_sig_addr(c->devinfo, &inst->qpu);
+        }
+        if (sig->ldunifa)
+                fprintf(stderr, "; ldunifa");
+        if (sig->ldunifarf) {
+                fprintf(stderr, "; ldunifarf");
+                vir_dump_sig_addr(c->devinfo, &inst->qpu);
+        }
         if (sig->wrtmuc)
                 fprintf(stderr, "; wrtmuc");
 }
@@ -179,7 +229,7 @@ vir_dump_alu(struct v3d_compile *c, struct qinst *inst)
                 fprintf(stderr, "%s", v3d_qpu_uf_name(instr->flags.auf));
                 fprintf(stderr, " ");
 
-                vir_print_reg(c, inst->dst);
+                vir_print_reg(c, inst, inst->dst);
                 fprintf(stderr, "%s", v3d_qpu_pack_name(instr->alu.add.output_pack));
 
                 unpack[0] = instr->alu.add.a_unpack;
@@ -191,7 +241,7 @@ vir_dump_alu(struct v3d_compile *c, struct qinst *inst)
                 fprintf(stderr, "%s", v3d_qpu_uf_name(instr->flags.muf));
                 fprintf(stderr, " ");
 
-                vir_print_reg(c, inst->dst);
+                vir_print_reg(c, inst, inst->dst);
                 fprintf(stderr, "%s", v3d_qpu_pack_name(instr->alu.mul.output_pack));
 
                 unpack[0] = instr->alu.mul.a_unpack;
@@ -200,7 +250,7 @@ vir_dump_alu(struct v3d_compile *c, struct qinst *inst)
 
         for (int i = 0; i < sideband_nsrc; i++) {
                 fprintf(stderr, ", ");
-                vir_print_reg(c, inst->src[i]);
+                vir_print_reg(c, inst, inst->src[i]);
                 if (i < nsrc)
                         fprintf(stderr, "%s", v3d_qpu_unpack_name(unpack[i]));
         }
@@ -266,7 +316,7 @@ vir_dump_inst(struct v3d_compile *c, struct qinst *inst)
 
                 if (vir_has_implicit_uniform(inst)) {
                         fprintf(stderr, " ");
-                        vir_print_reg(c, inst->src[vir_get_implicit_uniform_src(inst)]);
+                        vir_print_reg(c, inst, inst->src[vir_get_implicit_uniform_src(inst)]);
                 }
 
                 break;
@@ -281,7 +331,7 @@ vir_dump(struct v3d_compile *c)
         vir_for_each_block(block, c) {
                 fprintf(stderr, "BLOCK %d:\n", block->index);
                 vir_for_each_inst(inst, block) {
-                        if (c->temp_start) {
+                        if (c->live_intervals_valid) {
                                 bool first = true;
 
                                 for (int i = 0; i < c->num_temps; i++) {
@@ -302,7 +352,7 @@ vir_dump(struct v3d_compile *c)
                                         fprintf(stderr, " ");
                         }
 
-                        if (c->temp_end) {
+                        if (c->live_intervals_valid) {
                                 bool first = true;
 
                                 for (int i = 0; i < c->num_temps; i++) {
diff --git a/lib/mesa/src/broadcom/compiler/vir_live_variables.c b/lib/mesa/src/broadcom/compiler/vir_live_variables.c
index 217b716fd..2879e23b4 100644
--- a/lib/mesa/src/broadcom/compiler/vir_live_variables.c
+++ b/lib/mesa/src/broadcom/compiler/vir_live_variables.c
@@ -180,8 +180,6 @@ vir_setup_def(struct v3d_compile *c, struct qblock *block, int ip,
 static void
 sf_state_clear(struct hash_table *partial_update_ht)
 {
-        struct hash_entry *entry;
-
         hash_table_foreach(partial_update_ht, entry) {
                 struct partial_update_state *state = entry->data;
 
@@ -311,10 +309,20 @@ vir_calculate_live_intervals(struct v3d_compile *c)
 {
         int bitset_words = BITSET_WORDS(c->num_temps);
 
-        /* If we called this function more than once, then we should be
-         * freeing the previous arrays.
+        /* We may be called more than once if we've rearranged the program to
+         * try to get register allocation to succeed.
          */
-        assert(!c->temp_start);
+        if (c->temp_start) {
+                ralloc_free(c->temp_start);
+                ralloc_free(c->temp_end);
+
+                vir_for_each_block(block, c) {
+                        ralloc_free(block->def);
+                        ralloc_free(block->use);
+                        ralloc_free(block->live_in);
+                        ralloc_free(block->live_out);
+                }
+        }
 
         c->temp_start = rzalloc_array(c, int, c->num_temps);
         c->temp_end = rzalloc_array(c, int, c->num_temps);
@@ -337,4 +345,6 @@ vir_calculate_live_intervals(struct v3d_compile *c)
                 ;
 
         vir_compute_start_end(c, c->num_temps);
+
+        c->live_intervals_valid = true;
 }
diff --git a/lib/mesa/src/broadcom/compiler/vir_lower_uniforms.c b/lib/mesa/src/broadcom/compiler/vir_lower_uniforms.c
index 7f3bb8460..570e601a6 100644
--- a/lib/mesa/src/broadcom/compiler/vir_lower_uniforms.c
+++ b/lib/mesa/src/broadcom/compiler/vir_lower_uniforms.c
@@ -76,9 +76,7 @@ is_lowerable_uniform(struct qinst *inst, int i)
 {
         if (inst->src[i].file != QFILE_UNIF)
                 return false;
-        if (vir_has_implicit_uniform(inst))
-                return i != vir_get_implicit_uniform_src(inst);
-        return true;
+        return i != vir_get_implicit_uniform_src(inst);
 }
 
 /* Returns the number of different uniform values referenced by the
@@ -136,7 +134,6 @@ vir_lower_uniforms(struct v3d_compile *c)
                  */
                 uint32_t max_count = 0;
                 uint32_t max_index = 0;
-                struct hash_entry *entry;
                 hash_table_foreach(ht, entry) {
                         uint32_t count = (uintptr_t)entry->data;
                         uint32_t index = (uintptr_t)entry->key - 1;
@@ -152,7 +149,7 @@ vir_lower_uniforms(struct v3d_compile *c)
                  * reference a temp instead.
                  */
                 vir_for_each_block(block, c) {
-                        struct qinst *mov = NULL;
+                        struct qreg temp = c->undef;
 
                         vir_for_each_inst(inst, block) {
                                 uint32_t nsrc = vir_get_nsrc(inst);
@@ -162,29 +159,27 @@ vir_lower_uniforms(struct v3d_compile *c)
                                 if (count <= 1)
                                         continue;
 
-                                /* If the block doesn't have a load of the
-                                 * uniform yet, add it.  We could potentially
-                                 * do better and CSE MOVs from multiple blocks
-                                 * into dominating blocks, except that may
-                                 * cause troubles for register allocation.
-                                 */
-                                if (!mov) {
-                                        mov = vir_mul_inst(V3D_QPU_M_MOV,
-                                                           vir_get_temp(c),
-                                                           unif, c->undef);
-                                        list_add(&mov->link,
-                                                 &block->instructions);
-                                        c->defs[mov->dst.index] = mov;
-                                }
-
                                 bool removed = false;
                                 for (int i = 0; i < nsrc; i++) {
                                         if (is_lowerable_uniform(inst, i) &&
                                             inst->src[i].index == max_index) {
-                                                inst->src[i].file =
-                                                        mov->dst.file;
-                                                inst->src[i].index =
-                                                        mov->dst.index;
+                                                /* If the block doesn't have a
+                                                 * load of the uniform yet,
+                                                 * add it now.  We could
+                                                 * potentially do better and
+                                                 * CSE MOVs from multiple
+                                                 * blocks into dominating
+                                                 * blocks, except that may
+                                                 * cause troubles for register
+                                                 * allocation.
+                                                 */
+                                                if (temp.file == QFILE_NULL) {
+                                                        c->cursor =
+                                                                vir_before_inst(inst);
+                                                        temp = vir_MOV(c, unif);
+                                                }
+
+                                                inst->src[i] = temp;
                                                 remove_uniform(ht, unif);
                                                 removed = true;
                                         }
diff --git a/lib/mesa/src/broadcom/compiler/vir_opt_dead_code.c b/lib/mesa/src/broadcom/compiler/vir_opt_dead_code.c
index 9e0ef20b6..362fc9e52 100644
--- a/lib/mesa/src/broadcom/compiler/vir_opt_dead_code.c
+++ b/lib/mesa/src/broadcom/compiler/vir_opt_dead_code.c
@@ -49,6 +49,8 @@ dce(struct v3d_compile *c, struct qinst *inst)
         }
         assert(inst->qpu.flags.apf == V3D_QPU_PF_NONE);
         assert(inst->qpu.flags.mpf == V3D_QPU_PF_NONE);
+        assert(inst->qpu.flags.auf == V3D_QPU_UF_NONE);
+        assert(inst->qpu.flags.muf == V3D_QPU_UF_NONE);
         vir_remove_instruction(c, inst);
 }
 
@@ -78,17 +80,21 @@ has_nonremovable_reads(struct v3d_compile *c, struct qinst *inst)
                         if (total_size == 1)
                                 return true;
                 }
-
-                /* Dead code removal of varyings is tricky, so just assert
-                 * that it all happened at the NIR level.
-                 */
-                if (inst->src[i].file == QFILE_VARY)
-                        return true;
         }
 
         return false;
 }
 
+static bool
+can_write_to_null(struct v3d_compile *c, struct qinst *inst)
+{
+        /* The SFU instructions must write to a physical register. */
+        if (c->devinfo->ver >= 41 && v3d_qpu_uses_sfu(&inst->qpu))
+                return false;
+
+        return true;
+}
+
 bool
 vir_opt_dead_code(struct v3d_compile *c)
 {
@@ -114,7 +120,9 @@ vir_opt_dead_code(struct v3d_compile *c)
                                 continue;
 
                         if (inst->qpu.flags.apf != V3D_QPU_PF_NONE ||
-                            inst->qpu.flags.mpf != V3D_QPU_PF_NONE||
+                            inst->qpu.flags.mpf != V3D_QPU_PF_NONE ||
+                            inst->qpu.flags.auf != V3D_QPU_UF_NONE ||
+                            inst->qpu.flags.muf != V3D_QPU_UF_NONE ||
                             has_nonremovable_reads(c, inst)) {
                                 /* If we can't remove the instruction, but we
                                  * don't need its destination value, just
@@ -124,7 +132,8 @@ vir_opt_dead_code(struct v3d_compile *c)
                                  * it's nicer to read the VIR code without
                                  * unused destination regs.
                                  */
-                                if (inst->dst.file == QFILE_TEMP) {
+                                if (inst->dst.file == QFILE_TEMP &&
+                                    can_write_to_null(c, inst)) {
                                         if (debug) {
                                                 fprintf(stderr,
                                                         "Removing dst from: ");
diff --git a/lib/mesa/src/broadcom/compiler/vir_opt_small_immediates.c b/lib/mesa/src/broadcom/compiler/vir_opt_small_immediates.c
new file mode 100644
index 000000000..5491f9c24
--- /dev/null
+++ b/lib/mesa/src/broadcom/compiler/vir_opt_small_immediates.c
@@ -0,0 +1,112 @@
+/*
+ * Copyright © 2014 Broadcom
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+/**
+ * @file v3d_opt_small_immediates.c
+ *
+ * Turns references to small constant uniform values into small immediates
+ * fields.
+ */
+
+#include "v3d_compiler.h"
+
+static bool debug;
+
+bool
+vir_opt_small_immediates(struct v3d_compile *c)
+{
+        bool progress = false;
+
+        vir_for_each_inst_inorder(inst, c) {
+                if (inst->qpu.type != V3D_QPU_INSTR_TYPE_ALU)
+                        continue;
+
+                /* The small immediate value sits in the raddr B field, so we
+                 * can't have 2 small immediates in one instruction (unless
+                 * they're the same value, but that should be optimized away
+                 * elsewhere).
+                 */
+                bool uses_small_imm = false;
+                for (int i = 0; i < vir_get_nsrc(inst); i++) {
+                        if (inst->src[i].file == QFILE_SMALL_IMM)
+                                uses_small_imm = true;
+                }
+                if (uses_small_imm)
+                        continue;
+
+                for (int i = 0; i < vir_get_nsrc(inst); i++) {
+                        struct qreg src = vir_follow_movs(c, inst->src[i]);
+
+                        if (src.file != QFILE_UNIF ||
+                            c->uniform_contents[src.index] !=
+                            QUNIFORM_CONSTANT) {
+                                continue;
+                        }
+
+                        if (vir_has_implicit_uniform(inst) &&
+                            i == vir_get_implicit_uniform_src(inst)) {
+                                /* No turning the implicit uniform read into
+                                 * an immediate.
+                                 */
+                                continue;
+                        }
+
+                        /* Check if the uniform is suitable as a small
+                         * immediate.
+                         */
+                        uint32_t imm = c->uniform_data[src.index];
+                        uint32_t packed;
+                        if (!v3d_qpu_small_imm_pack(c->devinfo, imm, &packed))
+                                continue;
+
+                        /* Check that we don't have any other signals already
+                         * that would be incompatible with small_imm.
+                         */
+                        struct v3d_qpu_sig new_sig = inst->qpu.sig;
+                        uint32_t sig_packed;
+                        new_sig.small_imm = true;
+                        if (!v3d_qpu_sig_pack(c->devinfo, &new_sig, &sig_packed))
+                                continue;
+
+                        if (debug) {
+                                fprintf(stderr, "opt_small_immediate() from: ");
+                                vir_dump_inst(c, inst);
+                                fprintf(stderr, "\n");
+                        }
+                        inst->qpu.sig.small_imm = true;
+                        inst->qpu.raddr_b = packed;
+
+                        inst->src[i].file = QFILE_SMALL_IMM;
+                        inst->src[i].index = imm;
+                        if (debug) {
+                                fprintf(stderr, "to: ");
+                                vir_dump_inst(c, inst);
+                                fprintf(stderr, "\n");
+                        }
+                        progress = true;
+                        break;
+                }
+        }
+
+        return progress;
+}
diff --git a/lib/mesa/src/broadcom/compiler/vir_register_allocate.c b/lib/mesa/src/broadcom/compiler/vir_register_allocate.c
index 9ebf2cd69..accc07a3a 100644
--- a/lib/mesa/src/broadcom/compiler/vir_register_allocate.c
+++ b/lib/mesa/src/broadcom/compiler/vir_register_allocate.c
@@ -23,6 +23,7 @@
 
 #include "util/ralloc.h"
 #include "util/register_allocate.h"
+#include "common/v3d_device_info.h"
 #include "v3d_compiler.h"
 
 #define QPU_R(i) { .magic = false, .index = i }
@@ -32,30 +33,292 @@
 #define PHYS_INDEX    (ACC_INDEX + ACC_COUNT)
 #define PHYS_COUNT    64
 
+static bool
+is_last_ldtmu(struct qinst *inst, struct qblock *block)
+{
+        list_for_each_entry_from(struct qinst, scan_inst, inst,
+                                 &block->instructions, link) {
+                if (inst->qpu.sig.ldtmu)
+                        return false;
+                if (v3d_qpu_writes_tmu(&inst->qpu))
+                        return true;
+        }
+
+        return true;
+}
+
+static int
+v3d_choose_spill_node(struct v3d_compile *c, struct ra_graph *g,
+                      uint32_t *temp_to_node)
+{
+        float block_scale = 1.0;
+        float spill_costs[c->num_temps];
+        bool in_tmu_operation = false;
+        bool started_last_seg = false;
+
+        for (unsigned i = 0; i < c->num_temps; i++)
+                spill_costs[i] = 0.0;
+
+        /* XXX: Scale the cost up when inside of a loop. */
+        vir_for_each_block(block, c) {
+                vir_for_each_inst(inst, block) {
+                        /* We can't insert a new TMU operation while currently
+                         * in a TMU operation, and we can't insert new thread
+                         * switches after starting output writes.
+                         */
+                        bool no_spilling =
+                                (in_tmu_operation ||
+                                 (c->threads > 1 && started_last_seg));
+
+                        for (int i = 0; i < vir_get_nsrc(inst); i++) {
+                                if (inst->src[i].file != QFILE_TEMP)
+                                        continue;
+
+                                int temp = inst->src[i].index;
+                                if (no_spilling) {
+                                        BITSET_CLEAR(c->spillable,
+                                                     temp);
+                                } else {
+                                        spill_costs[temp] += block_scale;
+                                }
+                        }
+
+                        if (inst->dst.file == QFILE_TEMP) {
+                                int temp = inst->dst.index;
+
+                                if (no_spilling) {
+                                        BITSET_CLEAR(c->spillable,
+                                                     temp);
+                                } else {
+                                        spill_costs[temp] += block_scale;
+                                }
+                        }
+
+                        /* Refuse to spill a ldvary's dst, because that means
+                         * that ldvary's r5 would end up being used across a
+                         * thrsw.
+                         */
+                        if (inst->qpu.sig.ldvary) {
+                                assert(inst->dst.file == QFILE_TEMP);
+                                BITSET_CLEAR(c->spillable, inst->dst.index);
+                        }
+
+                        if (inst->is_last_thrsw)
+                                started_last_seg = true;
+
+                        if (v3d_qpu_writes_vpm(&inst->qpu) ||
+                            v3d_qpu_uses_tlb(&inst->qpu))
+                                started_last_seg = true;
+
+                        /* Track when we're in between a TMU setup and the
+                         * final LDTMU or TMUWT from that TMU setup.  We can't
+                         * spill/fill any temps during that time, because that
+                         * involves inserting a new TMU setup/LDTMU sequence.
+                         */
+                        if (inst->qpu.sig.ldtmu &&
+                            is_last_ldtmu(inst, block))
+                                in_tmu_operation = false;
+
+                        if (inst->qpu.type == V3D_QPU_INSTR_TYPE_ALU &&
+                            inst->qpu.alu.add.op == V3D_QPU_A_TMUWT)
+                                in_tmu_operation = false;
+
+                        if (v3d_qpu_writes_tmu(&inst->qpu))
+                                in_tmu_operation = true;
+                }
+        }
+
+        for (unsigned i = 0; i < c->num_temps; i++) {
+                int node = temp_to_node[i];
+
+                if (BITSET_TEST(c->spillable, i))
+                        ra_set_node_spill_cost(g, node, spill_costs[i]);
+        }
+
+        return ra_get_best_spill_node(g);
+}
+
+/* The spill offset for this thread takes a bit of setup, so do it once at
+ * program start.
+ */
+static void
+v3d_setup_spill_base(struct v3d_compile *c)
+{
+        c->cursor = vir_before_block(vir_entry_block(c));
+
+        int start_num_temps = c->num_temps;
+
+        /* Each thread wants to be in a separate region of the scratch space
+         * so that the QPUs aren't fighting over cache lines.  We have the
+         * driver keep a single global spill BO rather than
+         * per-spilling-program BOs, so we need a uniform from the driver for
+         * what the per-thread scale is.
+         */
+        struct qreg thread_offset =
+                vir_UMUL(c,
+                         vir_TIDX(c),
+                         vir_uniform(c, QUNIFORM_SPILL_SIZE_PER_THREAD, 0));
+
+        /* Each channel in a reg is 4 bytes, so scale them up by that. */
+        struct qreg element_offset = vir_SHL(c, vir_EIDX(c),
+                                             vir_uniform_ui(c, 2));
+
+        c->spill_base = vir_ADD(c,
+                                vir_ADD(c, thread_offset, element_offset),
+                                vir_uniform(c, QUNIFORM_SPILL_OFFSET, 0));
+
+        /* Make sure that we don't spill the spilling setup instructions. */
+        for (int i = start_num_temps; i < c->num_temps; i++)
+                BITSET_CLEAR(c->spillable, i);
+}
+
+static void
+v3d_emit_spill_tmua(struct v3d_compile *c, uint32_t spill_offset)
+{
+        vir_ADD_dest(c, vir_reg(QFILE_MAGIC,
+                                V3D_QPU_WADDR_TMUA),
+                     c->spill_base,
+                     vir_uniform_ui(c, spill_offset));
+}
+
+static void
+v3d_spill_reg(struct v3d_compile *c, int spill_temp)
+{
+        uint32_t spill_offset = c->spill_size;
+        c->spill_size += 16 * sizeof(uint32_t);
+
+        if (spill_offset == 0)
+                v3d_setup_spill_base(c);
+
+        struct qinst *last_thrsw = c->last_thrsw;
+        assert(!last_thrsw || last_thrsw->is_last_thrsw);
+
+        int start_num_temps = c->num_temps;
+
+        vir_for_each_inst_inorder(inst, c) {
+                for (int i = 0; i < vir_get_nsrc(inst); i++) {
+                        if (inst->src[i].file != QFILE_TEMP ||
+                            inst->src[i].index != spill_temp) {
+                                continue;
+                        }
+
+                        c->cursor = vir_before_inst(inst);
+
+                        v3d_emit_spill_tmua(c, spill_offset);
+                        vir_emit_thrsw(c);
+                        inst->src[i] = vir_LDTMU(c);
+                        c->fills++;
+                }
+
+                if (inst->dst.file == QFILE_TEMP &&
+                    inst->dst.index == spill_temp) {
+                        c->cursor = vir_after_inst(inst);
+
+                        inst->dst.index = c->num_temps++;
+                        vir_MOV_dest(c, vir_reg(QFILE_MAGIC, V3D_QPU_WADDR_TMUD),
+                                     inst->dst);
+                        v3d_emit_spill_tmua(c, spill_offset);
+                        vir_emit_thrsw(c);
+                        vir_TMUWT(c);
+                        c->spills++;
+                }
+
+                /* If we didn't have a last-thrsw inserted by nir_to_vir and
+                 * we've been inserting thrsws, then insert a new last_thrsw
+                 * right before we start the vpm/tlb sequence for the last
+                 * thread segment.
+                 */
+                if (!last_thrsw && c->last_thrsw &&
+                    (v3d_qpu_writes_vpm(&inst->qpu) ||
+                     v3d_qpu_uses_tlb(&inst->qpu))) {
+                        c->cursor = vir_before_inst(inst);
+                        vir_emit_thrsw(c);
+
+                        last_thrsw = c->last_thrsw;
+                        last_thrsw->is_last_thrsw = true;
+                }
+        }
+
+        /* Make sure c->last_thrsw is the actual last thrsw, not just one we
+         * inserted in our most recent unspill.
+         */
+        if (last_thrsw)
+                c->last_thrsw = last_thrsw;
+
+        /* Don't allow spilling of our spilling instructions.  There's no way
+         * they can help get things colored.
+         */
+        for (int i = start_num_temps; i < c->num_temps; i++)
+                BITSET_CLEAR(c->spillable, i);
+}
+
+struct v3d_ra_select_callback_data {
+        uint32_t next_acc;
+        uint32_t next_phys;
+};
+
+static unsigned int
+v3d_ra_select_callback(struct ra_graph *g, BITSET_WORD *regs, void *data)
+{
+        struct v3d_ra_select_callback_data *v3d_ra = data;
+
+        /* Choose an accumulator if possible (I think it's lower power than
+         * phys regs), but round-robin through them to give post-RA
+         * instruction selection more options.
+         */
+        for (int i = 0; i < ACC_COUNT; i++) {
+                int acc_off = (v3d_ra->next_acc + i) % ACC_COUNT;
+                int acc = ACC_INDEX + acc_off;
+
+                if (BITSET_TEST(regs, acc)) {
+                        v3d_ra->next_acc = acc_off + 1;
+                        return acc;
+                }
+        }
+
+        for (int i = 0; i < PHYS_COUNT; i++) {
+                int phys_off = (v3d_ra->next_phys + i) % PHYS_COUNT;
+                int phys = PHYS_INDEX + phys_off;
+
+                if (BITSET_TEST(regs, phys)) {
+                        v3d_ra->next_phys = phys_off + 1;
+                        return phys;
+                }
+        }
+
+        unreachable("RA must pass us at least one possible reg.");
+}
+
 bool
 vir_init_reg_sets(struct v3d_compiler *compiler)
 {
+        /* Allocate up to 3 regfile classes, for the ways the physical
+         * register file can be divided up for fragment shader threading.
+         */
+        int max_thread_index = (compiler->devinfo->ver >= 40 ? 2 : 3);
+
         compiler->regs = ra_alloc_reg_set(compiler, PHYS_INDEX + PHYS_COUNT,
                                           true);
         if (!compiler->regs)
                 return false;
 
-        /* Allocate 3 regfile classes, for the ways the physical register file
-         * can be divided up for fragment shader threading.
-         */
-        for (int threads = 0; threads < 3; threads++) {
-                compiler->reg_class[threads] =
+        for (int threads = 0; threads < max_thread_index; threads++) {
+                compiler->reg_class_phys_or_acc[threads] =
+                        ra_alloc_reg_class(compiler->regs);
+                compiler->reg_class_phys[threads] =
                         ra_alloc_reg_class(compiler->regs);
 
                 for (int i = PHYS_INDEX;
                      i < PHYS_INDEX + (PHYS_COUNT >> threads); i++) {
                         ra_class_add_reg(compiler->regs,
-                                         compiler->reg_class[threads], i);
+                                         compiler->reg_class_phys_or_acc[threads], i);
+                        ra_class_add_reg(compiler->regs,
+                                         compiler->reg_class_phys[threads], i);
                 }
 
                 for (int i = ACC_INDEX + 0; i < ACC_INDEX + ACC_COUNT; i++) {
                         ra_class_add_reg(compiler->regs,
-                                         compiler->reg_class[threads], i);
+                                         compiler->reg_class_phys_or_acc[threads], i);
                 }
         }
 
@@ -89,7 +352,7 @@ node_to_temp_priority(const void *in_a, const void *in_b)
  * The return value should be freed by the caller.
  */
 struct qpu_reg *
-v3d_register_allocate(struct v3d_compile *c)
+v3d_register_allocate(struct v3d_compile *c, bool *spilled)
 {
         struct node_to_temp_map map[c->num_temps];
         uint32_t temp_to_node[c->num_temps];
@@ -97,10 +360,33 @@ v3d_register_allocate(struct v3d_compile *c)
         struct qpu_reg *temp_registers = calloc(c->num_temps,
                                                 sizeof(*temp_registers));
         int acc_nodes[ACC_COUNT];
+        struct v3d_ra_select_callback_data callback_data = {
+                .next_acc = 0,
+                /* Start at RF3, to try to keep the TLB writes from using
+                 * RF0-2.
+                 */
+                .next_phys = 3,
+        };
+
+        *spilled = false;
+
+        vir_calculate_live_intervals(c);
+
+        /* Convert 1, 2, 4 threads to 0, 1, 2 index.
+         *
+         * V3D 4.x has double the physical register space, so 64 physical regs
+         * are available at both 1x and 2x threading, and 4x has 32.
+         */
+        int thread_index = ffs(c->threads) - 1;
+        if (c->devinfo->ver >= 40) {
+                if (thread_index >= 1)
+                        thread_index--;
+        }
 
         struct ra_graph *g = ra_alloc_interference_graph(c->compiler->regs,
                                                          c->num_temps +
                                                          ARRAY_SIZE(acc_nodes));
+        ra_set_select_reg_callback(g, v3d_ra_select_callback, &callback_data);
 
         /* Make some fixed nodes for the accumulators, which we will need to
          * interfere with when ops have implied r3/r4 writes or for the thread
@@ -113,9 +399,6 @@ v3d_register_allocate(struct v3d_compile *c)
                 ra_set_node_reg(g, acc_nodes[i], ACC_INDEX + i);
         }
 
-        /* Compute the live ranges so we can figure out interference. */
-        vir_calculate_live_intervals(c);
-
         for (uint32_t i = 0; i < c->num_temps; i++) {
                 map[i].temp = i;
                 map[i].priority = c->temp_end[i] - c->temp_start[i];
@@ -139,7 +422,7 @@ v3d_register_allocate(struct v3d_compile *c)
                  * result to a temp), nothing else can be stored in r3/r4 across
                  * it.
                  */
-                if (vir_writes_r3(inst)) {
+                if (vir_writes_r3(c->devinfo, inst)) {
                         for (int i = 0; i < c->num_temps; i++) {
                                 if (c->temp_start[i] < ip &&
                                     c->temp_end[i] > ip) {
@@ -149,7 +432,7 @@ v3d_register_allocate(struct v3d_compile *c)
                                 }
                         }
                 }
-                if (vir_writes_r4(inst)) {
+                if (vir_writes_r4(c->devinfo, inst)) {
                         for (int i = 0; i < c->num_temps; i++) {
                                 if (c->temp_start[i] < ip &&
                                     c->temp_end[i] > ip) {
@@ -160,6 +443,40 @@ v3d_register_allocate(struct v3d_compile *c)
                         }
                 }
 
+                if (inst->qpu.type == V3D_QPU_INSTR_TYPE_ALU) {
+                        switch (inst->qpu.alu.add.op) {
+                        case V3D_QPU_A_LDVPMV_IN:
+                        case V3D_QPU_A_LDVPMV_OUT:
+                        case V3D_QPU_A_LDVPMD_IN:
+                        case V3D_QPU_A_LDVPMD_OUT:
+                        case V3D_QPU_A_LDVPMP:
+                        case V3D_QPU_A_LDVPMG_IN:
+                        case V3D_QPU_A_LDVPMG_OUT:
+                                /* LDVPMs only store to temps (the MA flag
+                                 * decides whether the LDVPM is in or out)
+                                 */
+                                assert(inst->dst.file == QFILE_TEMP);
+                                class_bits[inst->dst.index] &= CLASS_BIT_PHYS;
+                                break;
+
+                        case V3D_QPU_A_RECIP:
+                        case V3D_QPU_A_RSQRT:
+                        case V3D_QPU_A_EXP:
+                        case V3D_QPU_A_LOG:
+                        case V3D_QPU_A_SIN:
+                        case V3D_QPU_A_RSQRT2:
+                                /* The SFU instructions write directly to the
+                                 * phys regfile.
+                                 */
+                                assert(inst->dst.file == QFILE_TEMP);
+                                class_bits[inst->dst.index] &= CLASS_BIT_PHYS;
+                                break;
+
+                        default:
+                                break;
+                        }
+                }
+
                 if (inst->src[0].file == QFILE_REG) {
                         switch (inst->src[0].index) {
                         case 0:
@@ -179,30 +496,31 @@ v3d_register_allocate(struct v3d_compile *c)
                         }
                 }
 
-#if 0
-                switch (inst->op) {
-                case QOP_THRSW:
+                if (inst->qpu.sig.thrsw) {
                         /* All accumulators are invalidated across a thread
                          * switch.
                          */
                         for (int i = 0; i < c->num_temps; i++) {
                                 if (c->temp_start[i] < ip && c->temp_end[i] > ip)
-                                        class_bits[i] &= ~(CLASS_BIT_R0_R3 |
-                                                           CLASS_BIT_R4);
+                                        class_bits[i] &= CLASS_BIT_PHYS;
                         }
-                        break;
-
-                default:
-                        break;
                 }
-#endif
 
                 ip++;
         }
 
         for (uint32_t i = 0; i < c->num_temps; i++) {
-                ra_set_node_class(g, temp_to_node[i],
-                                  c->compiler->reg_class[c->fs_threaded]);
+                if (class_bits[i] == CLASS_BIT_PHYS) {
+                        ra_set_node_class(g, temp_to_node[i],
+                                          c->compiler->reg_class_phys[thread_index]);
+                } else {
+                        assert(class_bits[i] == (CLASS_BIT_PHYS |
+                                                 CLASS_BIT_R0_R2 |
+                                                 CLASS_BIT_R3 |
+                                                 CLASS_BIT_R4));
+                        ra_set_node_class(g, temp_to_node[i],
+                                          c->compiler->reg_class_phys_or_acc[thread_index]);
+                }
         }
 
         for (uint32_t i = 0; i < c->num_temps; i++) {
@@ -216,14 +534,36 @@ v3d_register_allocate(struct v3d_compile *c)
                 }
         }
 
+        /* Debug code to force a bit of register spilling, for running across
+         * conformance tests to make sure that spilling works.
+         */
+        int force_register_spills = 0;
+        if (c->spill_size < 16 * sizeof(uint32_t) * force_register_spills) {
+                int node = v3d_choose_spill_node(c, g, temp_to_node);
+                if (node != -1) {
+                        v3d_spill_reg(c, map[node].temp);
+                        ralloc_free(g);
+                        *spilled = true;
+                        return NULL;
+                }
+        }
+
         bool ok = ra_allocate(g);
         if (!ok) {
-                if (!c->fs_threaded) {
-                        fprintf(stderr, "Failed to register allocate:\n");
-                        vir_dump(c);
+                /* Try to spill, if we can't reduce threading first. */
+                if (thread_index == 0) {
+                        int node = v3d_choose_spill_node(c, g, temp_to_node);
+
+                        if (node != -1) {
+                                v3d_spill_reg(c, map[node].temp);
+                                ralloc_free(g);
+
+                                /* Ask the outer loop to call back in. */
+                                *spilled = true;
+                                return NULL;
+                        }
                 }
 
-                c->failed = true;
                 free(temp_registers);
                 return NULL;
         }
@@ -250,5 +590,17 @@ v3d_register_allocate(struct v3d_compile *c)
 
         ralloc_free(g);
 
+        if (V3D_DEBUG & V3D_DEBUG_SHADERDB) {
+                fprintf(stderr, "SHADER-DB: %s prog %d/%d: %d spills\n",
+                        vir_get_stage_name(c),
+                        c->program_id, c->variant_id,
+                        c->spills);
+
+                fprintf(stderr, "SHADER-DB: %s prog %d/%d: %d fills\n",
+                        vir_get_stage_name(c),
+                        c->program_id, c->variant_id,
+                        c->fills);
+        }
+
         return temp_registers;
 }
diff --git a/lib/mesa/src/broadcom/compiler/vir_to_qpu.c b/lib/mesa/src/broadcom/compiler/vir_to_qpu.c
index eeb7b0bc2..4baadce29 100644
--- a/lib/mesa/src/broadcom/compiler/vir_to_qpu.c
+++ b/lib/mesa/src/broadcom/compiler/vir_to_qpu.c
@@ -109,6 +109,12 @@ new_ldunif_instr(struct qinst *inst, int i)
 static void
 set_src(struct v3d_qpu_instr *instr, enum v3d_qpu_mux *mux, struct qpu_reg src)
 {
+        if (src.smimm) {
+                assert(instr->sig.small_imm);
+                *mux = V3D_QPU_MUX_B;
+                return;
+        }
+
         if (src.magic) {
                 assert(src.index >= V3D_QPU_WADDR_R0 &&
                        src.index <= V3D_QPU_WADDR_R5);
@@ -138,6 +144,60 @@ set_src(struct v3d_qpu_instr *instr, enum v3d_qpu_mux *mux, struct qpu_reg src)
         }
 }
 
+static bool
+is_no_op_mov(struct qinst *qinst)
+{
+        static const struct v3d_qpu_sig no_sig = {0};
+
+        /* Make sure it's just a lone MOV. */
+        if (qinst->qpu.type != V3D_QPU_INSTR_TYPE_ALU ||
+            qinst->qpu.alu.mul.op != V3D_QPU_M_MOV ||
+            qinst->qpu.alu.add.op != V3D_QPU_A_NOP ||
+            memcmp(&qinst->qpu.sig, &no_sig, sizeof(no_sig)) != 0) {
+                return false;
+        }
+
+        /* Check if it's a MOV from a register to itself. */
+        enum v3d_qpu_waddr waddr = qinst->qpu.alu.mul.waddr;
+        if (qinst->qpu.alu.mul.magic_write) {
+                if (waddr < V3D_QPU_WADDR_R0 || waddr > V3D_QPU_WADDR_R4)
+                        return false;
+
+                if (qinst->qpu.alu.mul.a !=
+                    V3D_QPU_MUX_R0 + (waddr - V3D_QPU_WADDR_R0)) {
+                        return false;
+                }
+        } else {
+                int raddr;
+
+                switch (qinst->qpu.alu.mul.a) {
+                case V3D_QPU_MUX_A:
+                        raddr = qinst->qpu.raddr_a;
+                        break;
+                case V3D_QPU_MUX_B:
+                        raddr = qinst->qpu.raddr_b;
+                        break;
+                default:
+                        return false;
+                }
+                if (raddr != waddr)
+                        return false;
+        }
+
+        /* No packing or flags updates, or we need to execute the
+         * instruction.
+         */
+        if (qinst->qpu.alu.mul.a_unpack != V3D_QPU_UNPACK_NONE ||
+            qinst->qpu.alu.mul.output_pack != V3D_QPU_PACK_NONE ||
+            qinst->qpu.flags.mc != V3D_QPU_COND_NONE ||
+            qinst->qpu.flags.mpf != V3D_QPU_PF_NONE ||
+            qinst->qpu.flags.muf != V3D_QPU_UF_NONE) {
+                return false;
+        }
+
+        return true;
+}
+
 static void
 v3d_generate_code_block(struct v3d_compile *c,
                         struct qblock *block,
@@ -145,7 +205,7 @@ v3d_generate_code_block(struct v3d_compile *c,
 {
         int last_vpm_read_index = -1;
 
-        vir_for_each_inst(qinst, block) {
+        vir_for_each_inst_safe(qinst, block) {
 #if 0
                 fprintf(stderr, "translating qinst to qpu: ");
                 vir_dump_inst(c, qinst);
@@ -189,22 +249,8 @@ v3d_generate_code_block(struct v3d_compile *c,
 
                                 src[i] = qpu_acc(5);
                                 break;
-                        case QFILE_VARY:
-                                temp = new_qpu_nop_before(qinst);
-                                temp->qpu.sig.ldvary = true;
-
-                                src[i] = qpu_acc(3);
-                                break;
                         case QFILE_SMALL_IMM:
-                                abort(); /* XXX */
-#if 0
-                                src[i].mux = QPU_MUX_SMALL_IMM;
-                                src[i].addr = qpu_encode_small_immediate(qinst->src[i].index);
-                                /* This should only have returned a valid
-                                 * small immediate field, not ~0 for failure.
-                                 */
-                                assert(src[i].addr <= 47);
-#endif
+                                src[i].smimm = true;
                                 break;
 
                         case QFILE_VPM:
@@ -255,7 +301,6 @@ v3d_generate_code_block(struct v3d_compile *c,
                         dst = qpu_magic(V3D_QPU_WADDR_TLBU);
                         break;
 
-                case QFILE_VARY:
                 case QFILE_UNIF:
                 case QFILE_SMALL_IMM:
                 case QFILE_LOAD_IMM:
@@ -264,7 +309,14 @@ v3d_generate_code_block(struct v3d_compile *c,
                 }
 
                 if (qinst->qpu.type == V3D_QPU_INSTR_TYPE_ALU) {
-                        if (qinst->qpu.alu.add.op != V3D_QPU_A_NOP) {
+                        if (v3d_qpu_sig_writes_address(c->devinfo,
+                                                       &qinst->qpu.sig)) {
+                                assert(qinst->qpu.alu.add.op == V3D_QPU_A_NOP);
+                                assert(qinst->qpu.alu.mul.op == V3D_QPU_M_NOP);
+
+                                qinst->qpu.sig_addr = dst.index;
+                                qinst->qpu.sig_magic = dst.magic;
+                        } else if (qinst->qpu.alu.add.op != V3D_QPU_A_NOP) {
                                 assert(qinst->qpu.alu.mul.op == V3D_QPU_M_NOP);
                                 if (nsrc >= 1) {
                                         set_src(&qinst->qpu,
@@ -289,6 +341,11 @@ v3d_generate_code_block(struct v3d_compile *c,
 
                                 qinst->qpu.alu.mul.waddr = dst.index;
                                 qinst->qpu.alu.mul.magic_write = dst.magic;
+
+                                if (is_no_op_mov(qinst)) {
+                                        vir_remove_instruction(c, qinst);
+                                        continue;
+                                }
                         }
                 } else {
                         assert(qinst->qpu.type == V3D_QPU_INSTR_TYPE_BRANCH);
@@ -307,17 +364,14 @@ v3d_dump_qpu(struct v3d_compile *c)
         for (int i = 0; i < c->qpu_inst_count; i++) {
                 const char *str = v3d_qpu_disasm(c->devinfo, c->qpu_insts[i]);
                 fprintf(stderr, "0x%016"PRIx64" %s\n", c->qpu_insts[i], str);
+                ralloc_free((void *)str);
         }
         fprintf(stderr, "\n");
 }
 
 void
-v3d_vir_to_qpu(struct v3d_compile *c)
+v3d_vir_to_qpu(struct v3d_compile *c, struct qpu_reg *temp_registers)
 {
-        struct qpu_reg *temp_registers = v3d_register_allocate(c);
-        struct qblock *end_block = list_last_entry(&c->blocks,
-                                                   struct qblock, link);
-
         /* Reset the uniform count to how many will be actually loaded by the
          * generated QPU code.
          */
@@ -326,10 +380,6 @@ v3d_vir_to_qpu(struct v3d_compile *c)
         vir_for_each_block(block, c)
                 v3d_generate_code_block(c, block, temp_registers);
 
-        struct qinst *thrsw = vir_nop();
-        list_addtail(&thrsw->link, &end_block->instructions);
-        thrsw->qpu.sig.thrsw = true;
-
         uint32_t cycles = v3d_qpu_schedule_instructions(c);
 
         c->qpu_insts = rzalloc_array(c, uint64_t, c->qpu_inst_count);
@@ -337,11 +387,27 @@ v3d_vir_to_qpu(struct v3d_compile *c)
         vir_for_each_inst_inorder(inst, c) {
                 bool ok = v3d_qpu_instr_pack(c->devinfo, &inst->qpu,
                                              &c->qpu_insts[i++]);
-                assert(ok); (void) ok;
+                if (!ok) {
+                        fprintf(stderr, "Failed to pack instruction:\n");
+                        vir_dump_inst(c, inst);
+                        fprintf(stderr, "\n");
+                        c->failed = true;
+                        return;
+                }
         }
         assert(i == c->qpu_inst_count);
 
         if (V3D_DEBUG & V3D_DEBUG_SHADERDB) {
+                fprintf(stderr, "SHADER-DB: %s prog %d/%d: %d instructions\n",
+                        vir_get_stage_name(c),
+                        c->program_id, c->variant_id,
+                        c->qpu_inst_count);
+        }
+
+        /* The QPU cycle estimates are pretty broken (see waddr_latency()), so
+         * don't report them for now.
+         */
+        if (false) {
                 fprintf(stderr, "SHADER-DB: %s prog %d/%d: %d estimated cycles\n",
                         vir_get_stage_name(c),
                         c->program_id, c->variant_id,
diff --git a/lib/mesa/src/broadcom/meson.build b/lib/mesa/src/broadcom/meson.build
new file mode 100644
index 000000000..d3ea362f2
--- /dev/null
+++ b/lib/mesa/src/broadcom/meson.build
@@ -0,0 +1,58 @@
+# Copyright © 2017 Broadcom
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+inc_broadcom = include_directories('.', 'cle')
+
+subdir('cle')
+
+v3d_versions = ['33', '41', '42']
+v3d_libs = []
+
+if with_gallium_v3d
+  subdir('compiler')
+  subdir('qpu')
+endif
+
+per_version_libs = []
+foreach ver : v3d_versions
+  per_version_libs += static_library(
+    'libbroadcom-v' + ver,
+    [
+      files('clif/v3dx_dump.c'),
+      v3d_xml_pack
+    ],
+    include_directories : [inc_common, inc_broadcom, inc_src],
+    c_args : [c_vis_args, no_override_init_args, '-DV3D_VERSION=' + ver],
+    dependencies: [dep_valgrind, dep_thread],
+  )
+endforeach
+
+libbroadcom_v3d = static_library(
+  'libbroadcom_v3d',
+  [
+    files('common/v3d_debug.c', 'clif/clif_dump.c'),
+    v3d_xml_pack,
+  ],
+  include_directories : [inc_common, inc_broadcom, inc_src],
+  c_args : [c_vis_args, no_override_init_args],
+  link_whole : v3d_libs + per_version_libs,
+  build_by_default : false,
+  dependencies: [dep_valgrind, dep_thread],
+)
diff --git a/lib/mesa/src/broadcom/qpu/meson.build b/lib/mesa/src/broadcom/qpu/meson.build
new file mode 100644
index 000000000..8a4001606
--- /dev/null
+++ b/lib/mesa/src/broadcom/qpu/meson.build
@@ -0,0 +1,45 @@
+# Copyright © 2017 Broadcom
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+libbroadcom_qpu_files = files(
+  'qpu_disasm.c',
+  'qpu_instr.c',
+  'qpu_pack.c',
+)
+
+libbroadcom_qpu = static_library(
+  ['broadcom_qpu', v3d_xml_pack],
+  libbroadcom_qpu_files,
+  include_directories : [inc_common, inc_broadcom],
+  c_args : [c_vis_args, no_override_init_args],
+  dependencies : [dep_libdrm, dep_valgrind],
+  build_by_default : false,
+)
+
+v3d_libs += libbroadcom_qpu
+
+test(
+  'qpu_disasm',
+  executable(
+    'qpu_disasm', 'tests/qpu_disasm.c',
+    link_with: [libbroadcom_qpu, libmesa_util],
+    include_directories: inc_common
+  )
+)
diff --git a/lib/mesa/src/broadcom/qpu/qpu_disasm.c b/lib/mesa/src/broadcom/qpu/qpu_disasm.c
index 5ee834852..32e7ba12a 100644
--- a/lib/mesa/src/broadcom/qpu/qpu_disasm.c
+++ b/lib/mesa/src/broadcom/qpu/qpu_disasm.c
@@ -62,7 +62,21 @@ v3d_qpu_disasm_raddr(struct disasm_state *disasm,
         if (mux == V3D_QPU_MUX_A) {
                 append(disasm, "rf%d", instr->raddr_a);
         } else if (mux == V3D_QPU_MUX_B) {
-                append(disasm, "rf%d", instr->raddr_b);
+                if (instr->sig.small_imm) {
+                        uint32_t val;
+                        MAYBE_UNUSED bool ok =
+                                v3d_qpu_small_imm_unpack(disasm->devinfo,
+                                                         instr->raddr_b,
+                                                         &val);
+
+                        if ((int)val >= -16 && (int)val <= 15)
+                                append(disasm, "%d", val);
+                        else
+                                append(disasm, "0x%08x", val);
+                        assert(ok);
+                } else {
+                        append(disasm, "rf%d", instr->raddr_b);
+                }
         } else {
                 append(disasm, "r%d", mux);
         }
@@ -91,7 +105,8 @@ v3d_qpu_disasm_add(struct disasm_state *disasm,
         int num_src = v3d_qpu_add_op_num_src(instr->alu.add.op);
 
         append(disasm, "%s", v3d_qpu_add_op_name(instr->alu.add.op));
-        append(disasm, "%s", v3d_qpu_cond_name(instr->flags.ac));
+        if (!v3d_qpu_sig_writes_address(disasm->devinfo, &instr->sig))
+                append(disasm, "%s", v3d_qpu_cond_name(instr->flags.ac));
         append(disasm, "%s", v3d_qpu_pf_name(instr->flags.apf));
         append(disasm, "%s", v3d_qpu_uf_name(instr->flags.auf));
 
@@ -130,7 +145,8 @@ v3d_qpu_disasm_mul(struct disasm_state *disasm,
         append(disasm, "; ");
 
         append(disasm, "%s", v3d_qpu_mul_op_name(instr->alu.mul.op));
-        append(disasm, "%s", v3d_qpu_cond_name(instr->flags.mc));
+        if (!v3d_qpu_sig_writes_address(disasm->devinfo, &instr->sig))
+                append(disasm, "%s", v3d_qpu_cond_name(instr->flags.mc));
         append(disasm, "%s", v3d_qpu_pf_name(instr->flags.mpf));
         append(disasm, "%s", v3d_qpu_uf_name(instr->flags.muf));
 
@@ -162,6 +178,24 @@ v3d_qpu_disasm_mul(struct disasm_state *disasm,
 }
 
 static void
+v3d_qpu_disasm_sig_addr(struct disasm_state *disasm,
+                        const struct v3d_qpu_instr *instr)
+{
+        if (disasm->devinfo->ver < 41)
+                return;
+
+        if (!instr->sig_magic)
+                append(disasm, ".rf%d", instr->sig_addr);
+        else {
+                const char *name = v3d_qpu_magic_waddr_name(instr->sig_addr);
+                if (name)
+                        append(disasm, ".%s", name);
+                else
+                        append(disasm, ".UNKNOWN%d", instr->sig_addr);
+        }
+}
+
+static void
 v3d_qpu_disasm_sig(struct disasm_state *disasm,
                    const struct v3d_qpu_instr *instr)
 {
@@ -172,6 +206,9 @@ v3d_qpu_disasm_sig(struct disasm_state *disasm,
             !sig->ldvpm &&
             !sig->ldtmu &&
             !sig->ldunif &&
+            !sig->ldunifrf &&
+            !sig->ldunifa &&
+            !sig->ldunifarf &&
             !sig->wrtmuc) {
                 return;
         }
@@ -180,14 +217,36 @@ v3d_qpu_disasm_sig(struct disasm_state *disasm,
 
         if (sig->thrsw)
                 append(disasm, "; thrsw");
-        if (sig->ldvary)
+        if (sig->ldvary) {
                 append(disasm, "; ldvary");
+                v3d_qpu_disasm_sig_addr(disasm, instr);
+        }
         if (sig->ldvpm)
                 append(disasm, "; ldvpm");
-        if (sig->ldtmu)
+        if (sig->ldtmu) {
                 append(disasm, "; ldtmu");
+                v3d_qpu_disasm_sig_addr(disasm, instr);
+        }
+        if (sig->ldtlb) {
+                append(disasm, "; ldtlb");
+                v3d_qpu_disasm_sig_addr(disasm, instr);
+        }
+        if (sig->ldtlbu) {
+                append(disasm, "; ldtlbu");
+                v3d_qpu_disasm_sig_addr(disasm, instr);
+        }
         if (sig->ldunif)
                 append(disasm, "; ldunif");
+        if (sig->ldunifrf) {
+                append(disasm, "; ldunifrf");
+                v3d_qpu_disasm_sig_addr(disasm, instr);
+        }
+        if (sig->ldunifa)
+                append(disasm, "; ldunifa");
+        if (sig->ldunifarf) {
+                append(disasm, "; ldunifarf");
+                v3d_qpu_disasm_sig_addr(disasm, instr);
+        }
         if (sig->wrtmuc)
                 append(disasm, "; wrtmuc");
 }
diff --git a/lib/mesa/src/broadcom/qpu/qpu_instr.c b/lib/mesa/src/broadcom/qpu/qpu_instr.c
index 7499170de..147017a65 100644
--- a/lib/mesa/src/broadcom/qpu/qpu_instr.c
+++ b/lib/mesa/src/broadcom/qpu/qpu_instr.c
@@ -23,92 +23,9 @@
 
 #include <stdlib.h>
 #include "util/macros.h"
+#include "broadcom/common/v3d_device_info.h"
 #include "qpu_instr.h"
 
-#ifndef QPU_MASK
-#define QPU_MASK(high, low) ((((uint64_t)1<<((high)-(low)+1))-1)<<(low))
-/* Using the GNU statement expression extension */
-#define QPU_SET_FIELD(value, field)                                       \
-        ({                                                                \
-                uint64_t fieldval = (uint64_t)(value) << field ## _SHIFT; \
-                assert((fieldval & ~ field ## _MASK) == 0);               \
-                fieldval & field ## _MASK;                                \
-         })
-
-#define QPU_GET_FIELD(word, field) ((uint32_t)(((word)  & field ## _MASK) >> field ## _SHIFT))
-
-#define QPU_UPDATE_FIELD(inst, value, field)                              \
-        (((inst) & ~(field ## _MASK)) | QPU_SET_FIELD(value, field))
-#endif /* QPU_MASK */
-
-#define VC5_QPU_OP_MUL_SHIFT                58
-#define VC5_QPU_OP_MUL_MASK                 QPU_MASK(63, 58)
-
-#define VC5_QPU_SIG_SHIFT                   53
-#define VC5_QPU_SIG_MASK                    QPU_MASK(57, 53)
-# define VC5_QPU_SIG_THRSW_BIT              0x1
-# define VC5_QPU_SIG_LDUNIF_BIT             0x2
-# define VC5_QPU_SIG_LDTMU_BIT              0x4
-# define VC5_QPU_SIG_LDVARY_BIT             0x8
-
-#define VC5_QPU_COND_SHIFT                  46
-#define VC5_QPU_COND_MASK                   QPU_MASK(52, 46)
-
-#define VC5_QPU_COND_IFA                    0
-#define VC5_QPU_COND_IFB                    1
-#define VC5_QPU_COND_IFNA                   2
-#define VC5_QPU_COND_IFNB                   3
-
-#define VC5_QPU_MM                          QPU_MASK(45, 45)
-#define VC5_QPU_MA                          QPU_MASK(44, 44)
-
-#define V3D_QPU_WADDR_M_SHIFT               38
-#define V3D_QPU_WADDR_M_MASK                QPU_MASK(43, 38)
-
-#define VC5_QPU_BRANCH_ADDR_LOW_SHIFT       35
-#define VC5_QPU_BRANCH_ADDR_LOW_MASK        QPU_MASK(55, 35)
-
-#define V3D_QPU_WADDR_A_SHIFT               32
-#define V3D_QPU_WADDR_A_MASK                QPU_MASK(37, 32)
-
-#define VC5_QPU_BRANCH_COND_SHIFT           32
-#define VC5_QPU_BRANCH_COND_MASK            QPU_MASK(34, 32)
-
-#define VC5_QPU_BRANCH_ADDR_HIGH_SHIFT      24
-#define VC5_QPU_BRANCH_ADDR_HIGH_MASK       QPU_MASK(31, 24)
-
-#define VC5_QPU_OP_ADD_SHIFT                24
-#define VC5_QPU_OP_ADD_MASK                 QPU_MASK(31, 24)
-
-#define VC5_QPU_MUL_B_SHIFT                 21
-#define VC5_QPU_MUL_B_MASK                  QPU_MASK(23, 21)
-
-#define VC5_QPU_BRANCH_MSFIGN_SHIFT         21
-#define VC5_QPU_BRANCH_MSFIGN_MASK          QPU_MASK(22, 21)
-
-#define VC5_QPU_MUL_A_SHIFT                 18
-#define VC5_QPU_MUL_A_MASK                  QPU_MASK(20, 18)
-
-#define VC5_QPU_ADD_B_SHIFT                 15
-#define VC5_QPU_ADD_B_MASK                  QPU_MASK(17, 15)
-
-#define VC5_QPU_BRANCH_BDU_SHIFT            15
-#define VC5_QPU_BRANCH_BDU_MASK             QPU_MASK(17, 15)
-
-#define VC5_QPU_BRANCH_UB                   QPU_MASK(14, 14)
-
-#define VC5_QPU_ADD_A_SHIFT                 12
-#define VC5_QPU_ADD_A_MASK                  QPU_MASK(14, 12)
-
-#define VC5_QPU_BRANCH_BDI_SHIFT            12
-#define VC5_QPU_BRANCH_BDI_MASK             QPU_MASK(13, 12)
-
-#define VC5_QPU_RADDR_A_SHIFT               6
-#define VC5_QPU_RADDR_A_MASK                QPU_MASK(11, 6)
-
-#define VC5_QPU_RADDR_B_SHIFT               0
-#define VC5_QPU_RADDR_B_MASK                QPU_MASK(5, 0)
-
 const char *
 v3d_qpu_magic_waddr_name(enum v3d_qpu_waddr waddr)
 {
@@ -131,12 +48,29 @@ v3d_qpu_magic_waddr_name(enum v3d_qpu_waddr waddr)
                 [V3D_QPU_WADDR_VPMU] = "vpmu",
                 [V3D_QPU_WADDR_SYNC] = "sync",
                 [V3D_QPU_WADDR_SYNCU] = "syncu",
+                [V3D_QPU_WADDR_SYNCB] = "syncb",
                 [V3D_QPU_WADDR_RECIP] = "recip",
                 [V3D_QPU_WADDR_RSQRT] = "rsqrt",
                 [V3D_QPU_WADDR_EXP] = "exp",
                 [V3D_QPU_WADDR_LOG] = "log",
                 [V3D_QPU_WADDR_SIN] = "sin",
                 [V3D_QPU_WADDR_RSQRT2] = "rsqrt2",
+                [V3D_QPU_WADDR_TMUC] = "tmuc",
+                [V3D_QPU_WADDR_TMUS] = "tmus",
+                [V3D_QPU_WADDR_TMUT] = "tmut",
+                [V3D_QPU_WADDR_TMUR] = "tmur",
+                [V3D_QPU_WADDR_TMUI] = "tmui",
+                [V3D_QPU_WADDR_TMUB] = "tmub",
+                [V3D_QPU_WADDR_TMUDREF] = "tmudref",
+                [V3D_QPU_WADDR_TMUOFF] = "tmuoff",
+                [V3D_QPU_WADDR_TMUSCM] = "tmuscm",
+                [V3D_QPU_WADDR_TMUSF] = "tmusf",
+                [V3D_QPU_WADDR_TMUSLOD] = "tmuslod",
+                [V3D_QPU_WADDR_TMUHS] = "tmuhs",
+                [V3D_QPU_WADDR_TMUHSCM] = "tmuscm",
+                [V3D_QPU_WADDR_TMUHSF] = "tmuhsf",
+                [V3D_QPU_WADDR_TMUHSLOD] = "tmuhslod",
+                [V3D_QPU_WADDR_R5REP] = "r5rep",
         };
 
         return waddr_magic[waddr];
@@ -172,7 +106,8 @@ v3d_qpu_add_op_name(enum v3d_qpu_add_op op)
                 [V3D_QPU_A_NEG] = "neg",
                 [V3D_QPU_A_FLAPUSH] = "flapush",
                 [V3D_QPU_A_FLBPUSH] = "flbpush",
-                [V3D_QPU_A_FLBPOP] = "flbpop",
+                [V3D_QPU_A_FLPOP] = "flpop",
+                [V3D_QPU_A_RECIP] = "recip",
                 [V3D_QPU_A_SETMSF] = "setmsf",
                 [V3D_QPU_A_SETREVF] = "setrevf",
                 [V3D_QPU_A_NOP] = "nop",
@@ -192,14 +127,22 @@ v3d_qpu_add_op_name(enum v3d_qpu_add_op op)
                 [V3D_QPU_A_VDWWT] = "vdwwt",
                 [V3D_QPU_A_IID] = "iid",
                 [V3D_QPU_A_SAMPID] = "sampid",
-                [V3D_QPU_A_PATCHID] = "patchid",
+                [V3D_QPU_A_BARRIERID] = "barrierid",
                 [V3D_QPU_A_TMUWT] = "tmuwt",
                 [V3D_QPU_A_VPMSETUP] = "vpmsetup",
                 [V3D_QPU_A_VPMWT] = "vpmwt",
-                [V3D_QPU_A_LDVPMV] = "ldvpmv",
-                [V3D_QPU_A_LDVPMD] = "ldvpmd",
+                [V3D_QPU_A_LDVPMV_IN] = "ldvpmv_in",
+                [V3D_QPU_A_LDVPMV_OUT] = "ldvpmv_out",
+                [V3D_QPU_A_LDVPMD_IN] = "ldvpmd_in",
+                [V3D_QPU_A_LDVPMD_OUT] = "ldvpmd_out",
                 [V3D_QPU_A_LDVPMP] = "ldvpmp",
-                [V3D_QPU_A_LDVPMG] = "ldvpmg",
+                [V3D_QPU_A_RSQRT] = "rsqrt",
+                [V3D_QPU_A_EXP] = "exp",
+                [V3D_QPU_A_LOG] = "log",
+                [V3D_QPU_A_SIN] = "sin",
+                [V3D_QPU_A_RSQRT2] = "rsqrt2",
+                [V3D_QPU_A_LDVPMG_IN] = "ldvpmg_in",
+                [V3D_QPU_A_LDVPMG_OUT] = "ldvpmg_out",
                 [V3D_QPU_A_FCMP] = "fcmp",
                 [V3D_QPU_A_VFMAX] = "vfmax",
                 [V3D_QPU_A_FROUND] = "fround",
@@ -431,7 +374,8 @@ static const uint8_t add_op_args[] = {
         [V3D_QPU_A_NEG] = D | A,
         [V3D_QPU_A_FLAPUSH] = D | A,
         [V3D_QPU_A_FLBPUSH] = D | A,
-        [V3D_QPU_A_FLBPOP] = D | A,
+        [V3D_QPU_A_FLPOP] = D | A,
+        [V3D_QPU_A_RECIP] = D | A,
         [V3D_QPU_A_SETMSF] = D | A,
         [V3D_QPU_A_SETREVF] = D | A,
         [V3D_QPU_A_NOP] = 0,
@@ -453,16 +397,24 @@ static const uint8_t add_op_args[] = {
         [V3D_QPU_A_VDWWT] = D,
         [V3D_QPU_A_IID] = D,
         [V3D_QPU_A_SAMPID] = D,
-        [V3D_QPU_A_PATCHID] = D,
+        [V3D_QPU_A_BARRIERID] = D,
         [V3D_QPU_A_TMUWT] = D,
         [V3D_QPU_A_VPMWT] = D,
 
         [V3D_QPU_A_VPMSETUP] = D | A,
 
-        [V3D_QPU_A_LDVPMV] = D | A,
-        [V3D_QPU_A_LDVPMD] = D | A,
+        [V3D_QPU_A_LDVPMV_IN] = D | A,
+        [V3D_QPU_A_LDVPMV_OUT] = D | A,
+        [V3D_QPU_A_LDVPMD_IN] = D | A,
+        [V3D_QPU_A_LDVPMD_OUT] = D | A,
         [V3D_QPU_A_LDVPMP] = D | A,
-        [V3D_QPU_A_LDVPMG] = D | A | B,
+        [V3D_QPU_A_RSQRT] = D | A,
+        [V3D_QPU_A_EXP] = D | A,
+        [V3D_QPU_A_LOG] = D | A,
+        [V3D_QPU_A_SIN] = D | A,
+        [V3D_QPU_A_RSQRT2] = D | A,
+        [V3D_QPU_A_LDVPMG_IN] = D | A | B,
+        [V3D_QPU_A_LDVPMG_OUT] = D | A | B,
 
         /* FIXME: MOVABSNEG */
 
@@ -566,16 +518,19 @@ v3d_qpu_magic_waddr_is_sfu(enum v3d_qpu_waddr waddr)
 bool
 v3d_qpu_magic_waddr_is_tmu(enum v3d_qpu_waddr waddr)
 {
-        switch (waddr) {
-        case V3D_QPU_WADDR_TMU:
-        case V3D_QPU_WADDR_TMUL:
-        case V3D_QPU_WADDR_TMUD:
-        case V3D_QPU_WADDR_TMUA:
-        case V3D_QPU_WADDR_TMUAU:
-                return true;
-        default:
-                return false;
-        }
+        /* XXX: WADDR_TMU changed to UNIFA on 4.x */
+        return ((waddr >= V3D_QPU_WADDR_TMU &&
+                 waddr <= V3D_QPU_WADDR_TMUAU) ||
+                (waddr >= V3D_QPU_WADDR_TMUC &&
+                 waddr <= V3D_QPU_WADDR_TMUHSLOD));
+}
+
+bool
+v3d_qpu_waits_on_tmu(const struct v3d_qpu_instr *inst)
+{
+        return (inst->sig.ldtmu ||
+                (inst->type == V3D_QPU_INSTR_TYPE_ALU &&
+                 inst->alu.add.op == V3D_QPU_A_TMUWT));
 }
 
 bool
@@ -596,40 +551,222 @@ bool
 v3d_qpu_magic_waddr_is_tsy(enum v3d_qpu_waddr waddr)
 {
         return (waddr == V3D_QPU_WADDR_SYNC ||
+                waddr == V3D_QPU_WADDR_SYNCB ||
                 waddr == V3D_QPU_WADDR_SYNCU);
 }
 
+static bool
+v3d_qpu_add_op_reads_vpm(enum  v3d_qpu_add_op op)
+{
+        switch (op) {
+        case V3D_QPU_A_VPMSETUP:
+        case V3D_QPU_A_VPMWT:
+        case V3D_QPU_A_LDVPMV_IN:
+        case V3D_QPU_A_LDVPMV_OUT:
+        case V3D_QPU_A_LDVPMD_IN:
+        case V3D_QPU_A_LDVPMD_OUT:
+        case V3D_QPU_A_LDVPMP:
+        case V3D_QPU_A_LDVPMG_IN:
+        case V3D_QPU_A_LDVPMG_OUT:
+                return true;
+        default:
+                return false;
+        }
+}
+
+static bool
+v3d_qpu_add_op_writes_vpm(enum  v3d_qpu_add_op op)
+{
+        switch (op) {
+        case V3D_QPU_A_VPMSETUP:
+        case V3D_QPU_A_VPMWT:
+        case V3D_QPU_A_STVPMV:
+        case V3D_QPU_A_STVPMD:
+        case V3D_QPU_A_STVPMP:
+                return true;
+        default:
+                return false;
+        }
+}
+
 bool
-v3d_qpu_writes_r3(const struct v3d_qpu_instr *inst)
+v3d_qpu_uses_tlb(const struct v3d_qpu_instr *inst)
 {
+        if (inst->sig.ldtlb ||
+            inst->sig.ldtlbu)
+                return true;
+
+        if (inst->type == V3D_QPU_INSTR_TYPE_ALU) {
+                if (inst->alu.add.magic_write &&
+                    v3d_qpu_magic_waddr_is_tlb(inst->alu.add.waddr)) {
+                        return true;
+                }
+
+                if (inst->alu.mul.magic_write &&
+                    v3d_qpu_magic_waddr_is_tlb(inst->alu.mul.waddr)) {
+                        return true;
+                }
+        }
+
+        return false;
+}
+
+bool
+v3d_qpu_uses_sfu(const struct v3d_qpu_instr *inst)
+{
+        if (inst->type == V3D_QPU_INSTR_TYPE_ALU) {
+                switch (inst->alu.add.op) {
+                case V3D_QPU_A_RECIP:
+                case V3D_QPU_A_RSQRT:
+                case V3D_QPU_A_EXP:
+                case V3D_QPU_A_LOG:
+                case V3D_QPU_A_SIN:
+                case V3D_QPU_A_RSQRT2:
+                        return true;
+                default:
+                        break;
+                }
+
+                if (inst->alu.add.magic_write &&
+                    v3d_qpu_magic_waddr_is_sfu(inst->alu.add.waddr)) {
+                        return true;
+                }
+
+                if (inst->alu.mul.magic_write &&
+                    v3d_qpu_magic_waddr_is_sfu(inst->alu.mul.waddr)) {
+                        return true;
+                }
+        }
+
+        return false;
+}
+
+bool
+v3d_qpu_writes_tmu(const struct v3d_qpu_instr *inst)
+{
+        return (inst->type == V3D_QPU_INSTR_TYPE_ALU &&
+                ((inst->alu.add.magic_write &&
+                  v3d_qpu_magic_waddr_is_tmu(inst->alu.add.waddr)) ||
+                 (inst->alu.mul.magic_write &&
+                  v3d_qpu_magic_waddr_is_tmu(inst->alu.mul.waddr))));
+}
+
+bool
+v3d_qpu_reads_vpm(const struct v3d_qpu_instr *inst)
+{
+        if (inst->sig.ldvpm)
+                return true;
+
+        if (inst->type == V3D_QPU_INSTR_TYPE_ALU) {
+                if (v3d_qpu_add_op_reads_vpm(inst->alu.add.op))
+                        return true;
+        }
+
+        return false;
+}
+
+bool
+v3d_qpu_writes_vpm(const struct v3d_qpu_instr *inst)
+{
+        if (inst->type == V3D_QPU_INSTR_TYPE_ALU) {
+                if (v3d_qpu_add_op_writes_vpm(inst->alu.add.op))
+                        return true;
+
+                if (inst->alu.add.magic_write &&
+                    v3d_qpu_magic_waddr_is_vpm(inst->alu.add.waddr)) {
+                        return true;
+                }
+
+                if (inst->alu.mul.magic_write &&
+                    v3d_qpu_magic_waddr_is_vpm(inst->alu.mul.waddr)) {
+                        return true;
+                }
+        }
+
+        return false;
+}
+
+bool
+v3d_qpu_uses_vpm(const struct v3d_qpu_instr *inst)
+{
+        return v3d_qpu_reads_vpm(inst) || v3d_qpu_writes_vpm(inst);
+}
+
+bool
+v3d_qpu_writes_r3(const struct v3d_device_info *devinfo,
+                  const struct v3d_qpu_instr *inst)
+{
+        if (inst->type == V3D_QPU_INSTR_TYPE_ALU) {
+                if (inst->alu.add.magic_write &&
+                    inst->alu.add.waddr == V3D_QPU_WADDR_R3) {
+                        return true;
+                }
+
+                if (inst->alu.mul.magic_write &&
+                    inst->alu.mul.waddr == V3D_QPU_WADDR_R3) {
+                        return true;
+                }
+        }
+
+        if (v3d_qpu_sig_writes_address(devinfo, &inst->sig) &&
+            inst->sig_magic && inst->sig_addr == V3D_QPU_WADDR_R3) {
+                return true;
+        }
+
         return inst->sig.ldvary || inst->sig.ldvpm;
 }
 
 bool
-v3d_qpu_writes_r4(const struct v3d_qpu_instr *inst)
+v3d_qpu_writes_r4(const struct v3d_device_info *devinfo,
+                  const struct v3d_qpu_instr *inst)
 {
         if (inst->sig.ldtmu)
                 return true;
 
         if (inst->type == V3D_QPU_INSTR_TYPE_ALU) {
                 if (inst->alu.add.magic_write &&
-                    v3d_qpu_magic_waddr_is_sfu(inst->alu.add.waddr)) {
+                    (inst->alu.add.waddr == V3D_QPU_WADDR_R4 ||
+                     v3d_qpu_magic_waddr_is_sfu(inst->alu.add.waddr))) {
                         return true;
                 }
 
                 if (inst->alu.mul.magic_write &&
-                    v3d_qpu_magic_waddr_is_sfu(inst->alu.mul.waddr)) {
+                    (inst->alu.mul.waddr == V3D_QPU_WADDR_R4 ||
+                     v3d_qpu_magic_waddr_is_sfu(inst->alu.mul.waddr))) {
                         return true;
                 }
         }
 
+        if (v3d_qpu_sig_writes_address(devinfo, &inst->sig) &&
+            inst->sig_magic && inst->sig_addr == V3D_QPU_WADDR_R4) {
+                return true;
+        }
+
         return false;
 }
 
 bool
-v3d_qpu_writes_r5(const struct v3d_qpu_instr *inst)
+v3d_qpu_writes_r5(const struct v3d_device_info *devinfo,
+                  const struct v3d_qpu_instr *inst)
 {
-        return inst->sig.ldvary || inst->sig.ldunif;
+        if (inst->type == V3D_QPU_INSTR_TYPE_ALU) {
+                if (inst->alu.add.magic_write &&
+                    inst->alu.add.waddr == V3D_QPU_WADDR_R5) {
+                        return true;
+                }
+
+                if (inst->alu.mul.magic_write &&
+                    inst->alu.mul.waddr == V3D_QPU_WADDR_R5) {
+                        return true;
+                }
+        }
+
+        if (v3d_qpu_sig_writes_address(devinfo, &inst->sig) &&
+            inst->sig_magic && inst->sig_addr == V3D_QPU_WADDR_R5) {
+                return true;
+        }
+
+        return inst->sig.ldvary || inst->sig.ldunif || inst->sig.ldunifa;
 }
 
 bool
@@ -643,3 +780,18 @@ v3d_qpu_uses_mux(const struct v3d_qpu_instr *inst, enum v3d_qpu_mux mux)
                 (mul_nsrc > 0 && inst->alu.mul.a == mux) ||
                 (mul_nsrc > 1 && inst->alu.mul.b == mux));
 }
+
+bool
+v3d_qpu_sig_writes_address(const struct v3d_device_info *devinfo,
+                           const struct v3d_qpu_sig *sig)
+{
+        if (devinfo->ver < 41)
+                return false;
+
+        return (sig->ldunifrf ||
+                sig->ldunifarf ||
+                sig->ldvary ||
+                sig->ldtmu ||
+                sig->ldtlb ||
+                sig->ldtlbu);
+}
diff --git a/lib/mesa/src/broadcom/qpu/qpu_instr.h b/lib/mesa/src/broadcom/qpu/qpu_instr.h
index a425fae8b..c2b4ebd19 100644
--- a/lib/mesa/src/broadcom/qpu/qpu_instr.h
+++ b/lib/mesa/src/broadcom/qpu/qpu_instr.h
@@ -42,6 +42,9 @@ struct v3d_device_info;
 struct v3d_qpu_sig {
         bool thrsw:1;
         bool ldunif:1;
+        bool ldunifa:1;
+        bool ldunifrf:1;
+        bool ldunifarf:1;
         bool ldtmu:1;
         bool ldvary:1;
         bool ldvpm:1;
@@ -104,13 +107,29 @@ enum v3d_qpu_waddr {
         V3D_QPU_WADDR_VPMU = 15,
         V3D_QPU_WADDR_SYNC = 16,
         V3D_QPU_WADDR_SYNCU = 17,
-        /* reserved */
+        V3D_QPU_WADDR_SYNCB = 18,
         V3D_QPU_WADDR_RECIP = 19,
         V3D_QPU_WADDR_RSQRT = 20,
         V3D_QPU_WADDR_EXP = 21,
         V3D_QPU_WADDR_LOG = 22,
         V3D_QPU_WADDR_SIN = 23,
         V3D_QPU_WADDR_RSQRT2 = 24,
+        V3D_QPU_WADDR_TMUC = 32,
+        V3D_QPU_WADDR_TMUS = 33,
+        V3D_QPU_WADDR_TMUT = 34,
+        V3D_QPU_WADDR_TMUR = 35,
+        V3D_QPU_WADDR_TMUI = 36,
+        V3D_QPU_WADDR_TMUB = 37,
+        V3D_QPU_WADDR_TMUDREF = 38,
+        V3D_QPU_WADDR_TMUOFF = 39,
+        V3D_QPU_WADDR_TMUSCM = 40,
+        V3D_QPU_WADDR_TMUSF = 41,
+        V3D_QPU_WADDR_TMUSLOD = 42,
+        V3D_QPU_WADDR_TMUHS = 43,
+        V3D_QPU_WADDR_TMUHSCM = 44,
+        V3D_QPU_WADDR_TMUHSF = 45,
+        V3D_QPU_WADDR_TMUHSLOD = 46,
+        V3D_QPU_WADDR_R5REP = 55,
 };
 
 struct v3d_qpu_flags {
@@ -146,7 +165,8 @@ enum v3d_qpu_add_op {
         V3D_QPU_A_NEG,
         V3D_QPU_A_FLAPUSH,
         V3D_QPU_A_FLBPUSH,
-        V3D_QPU_A_FLBPOP,
+        V3D_QPU_A_FLPOP,
+        V3D_QPU_A_RECIP,
         V3D_QPU_A_SETMSF,
         V3D_QPU_A_SETREVF,
         V3D_QPU_A_NOP,
@@ -166,14 +186,22 @@ enum v3d_qpu_add_op {
         V3D_QPU_A_VDWWT,
         V3D_QPU_A_IID,
         V3D_QPU_A_SAMPID,
-        V3D_QPU_A_PATCHID,
+        V3D_QPU_A_BARRIERID,
         V3D_QPU_A_TMUWT,
         V3D_QPU_A_VPMSETUP,
         V3D_QPU_A_VPMWT,
-        V3D_QPU_A_LDVPMV,
-        V3D_QPU_A_LDVPMD,
+        V3D_QPU_A_LDVPMV_IN,
+        V3D_QPU_A_LDVPMV_OUT,
+        V3D_QPU_A_LDVPMD_IN,
+        V3D_QPU_A_LDVPMD_OUT,
         V3D_QPU_A_LDVPMP,
-        V3D_QPU_A_LDVPMG,
+        V3D_QPU_A_RSQRT,
+        V3D_QPU_A_EXP,
+        V3D_QPU_A_LOG,
+        V3D_QPU_A_SIN,
+        V3D_QPU_A_RSQRT2,
+        V3D_QPU_A_LDVPMG_IN,
+        V3D_QPU_A_LDVPMG_OUT,
         V3D_QPU_A_FCMP,
         V3D_QPU_A_VFMAX,
         V3D_QPU_A_FROUND,
@@ -347,6 +375,8 @@ struct v3d_qpu_instr {
         enum v3d_qpu_instr_type type;
 
         struct v3d_qpu_sig sig;
+        uint8_t sig_addr;
+        bool sig_magic; /* If the signal writes to a magic address */
         uint8_t raddr_a;
         uint8_t raddr_b;
         struct v3d_qpu_flags flags;
@@ -390,6 +420,16 @@ v3d_qpu_flags_unpack(const struct v3d_device_info *devinfo,
                      struct v3d_qpu_flags *cond);
 
 bool
+v3d_qpu_small_imm_pack(const struct v3d_device_info *devinfo,
+                       uint32_t value,
+                       uint32_t *packed_small_immediate);
+
+bool
+v3d_qpu_small_imm_unpack(const struct v3d_device_info *devinfo,
+                         uint32_t packed_small_immediate,
+                         uint32_t *small_immediate);
+
+bool
 v3d_qpu_instr_pack(const struct v3d_device_info *devinfo,
                    const struct v3d_qpu_instr *instr,
                    uint64_t *packed_instr);
@@ -403,9 +443,21 @@ bool v3d_qpu_magic_waddr_is_tmu(enum v3d_qpu_waddr waddr) ATTRIBUTE_CONST;
 bool v3d_qpu_magic_waddr_is_tlb(enum v3d_qpu_waddr waddr) ATTRIBUTE_CONST;
 bool v3d_qpu_magic_waddr_is_vpm(enum v3d_qpu_waddr waddr) ATTRIBUTE_CONST;
 bool v3d_qpu_magic_waddr_is_tsy(enum v3d_qpu_waddr waddr) ATTRIBUTE_CONST;
-bool v3d_qpu_writes_r3(const struct v3d_qpu_instr *instr) ATTRIBUTE_CONST;
-bool v3d_qpu_writes_r4(const struct v3d_qpu_instr *instr) ATTRIBUTE_CONST;
-bool v3d_qpu_writes_r5(const struct v3d_qpu_instr *instr) ATTRIBUTE_CONST;
+bool v3d_qpu_uses_tlb(const struct v3d_qpu_instr *inst) ATTRIBUTE_CONST;
+bool v3d_qpu_uses_sfu(const struct v3d_qpu_instr *inst) ATTRIBUTE_CONST;
+bool v3d_qpu_writes_tmu(const struct v3d_qpu_instr *inst) ATTRIBUTE_CONST;
+bool v3d_qpu_writes_r3(const struct v3d_device_info *devinfo,
+                       const struct v3d_qpu_instr *instr) ATTRIBUTE_CONST;
+bool v3d_qpu_writes_r4(const struct v3d_device_info *devinfo,
+                       const struct v3d_qpu_instr *instr) ATTRIBUTE_CONST;
+bool v3d_qpu_writes_r5(const struct v3d_device_info *devinfo,
+                       const struct v3d_qpu_instr *instr) ATTRIBUTE_CONST;
+bool v3d_qpu_waits_on_tmu(const struct v3d_qpu_instr *inst) ATTRIBUTE_CONST;
 bool v3d_qpu_uses_mux(const struct v3d_qpu_instr *inst, enum v3d_qpu_mux mux);
+bool v3d_qpu_uses_vpm(const struct v3d_qpu_instr *inst) ATTRIBUTE_CONST;
+bool v3d_qpu_reads_vpm(const struct v3d_qpu_instr *inst) ATTRIBUTE_CONST;
+bool v3d_qpu_writes_vpm(const struct v3d_qpu_instr *inst) ATTRIBUTE_CONST;
+bool v3d_qpu_sig_writes_address(const struct v3d_device_info *devinfo,
+                                const struct v3d_qpu_sig *sig) ATTRIBUTE_CONST;
 
 #endif
diff --git a/lib/mesa/src/broadcom/qpu/qpu_pack.c b/lib/mesa/src/broadcom/qpu/qpu_pack.c
index 0ecce8666..70f31d734 100644
--- a/lib/mesa/src/broadcom/qpu/qpu_pack.c
+++ b/lib/mesa/src/broadcom/qpu/qpu_pack.c
@@ -48,18 +48,10 @@
 
 #define VC5_QPU_SIG_SHIFT                   53
 #define VC5_QPU_SIG_MASK                    QPU_MASK(57, 53)
-# define VC5_QPU_SIG_THRSW_BIT              0x1
-# define VC5_QPU_SIG_LDUNIF_BIT             0x2
-# define VC5_QPU_SIG_LDTMU_BIT              0x4
-# define VC5_QPU_SIG_LDVARY_BIT             0x8
 
 #define VC5_QPU_COND_SHIFT                  46
 #define VC5_QPU_COND_MASK                   QPU_MASK(52, 46)
-
-#define VC5_QPU_COND_IFA                    0
-#define VC5_QPU_COND_IFB                    1
-#define VC5_QPU_COND_IFNA                   2
-#define VC5_QPU_COND_IFNB                   3
+#define VC5_QPU_COND_SIG_MAGIC_ADDR         (1 << 6)
 
 #define VC5_QPU_MM                          QPU_MASK(45, 45)
 #define VC5_QPU_MA                          QPU_MASK(44, 44)
@@ -113,6 +105,9 @@
 
 #define THRSW .thrsw = true
 #define LDUNIF .ldunif = true
+#define LDUNIFRF .ldunifrf = true
+#define LDUNIFA .ldunifa = true
+#define LDUNIFARF .ldunifarf = true
 #define LDTMU .ldtmu = true
 #define LDVARY .ldvary = true
 #define LDVPM .ldvpm = true
@@ -156,6 +151,67 @@ static const struct v3d_qpu_sig v33_sig_map[] = {
         [31] = { SMIMM,                        },
 };
 
+static const struct v3d_qpu_sig v40_sig_map[] = {
+        /*      MISC    R3      R4      R5 */
+        [0]  = {                               },
+        [1]  = { THRSW,                        },
+        [2]  = {                        LDUNIF },
+        [3]  = { THRSW,                 LDUNIF },
+        [4]  = {                LDTMU,         },
+        [5]  = { THRSW,         LDTMU,         },
+        [6]  = {                LDTMU,  LDUNIF },
+        [7]  = { THRSW,         LDTMU,  LDUNIF },
+        [8]  = {        LDVARY,                },
+        [9]  = { THRSW, LDVARY,                },
+        [10] = {        LDVARY,         LDUNIF },
+        [11] = { THRSW, LDVARY,         LDUNIF },
+        /* 12-13 reserved */
+        [14] = { SMIMM, LDVARY,                },
+        [15] = { SMIMM,                        },
+        [16] = {        LDTLB,                 },
+        [17] = {        LDTLBU,                },
+        [18] = {                        WRTMUC },
+        [19] = { THRSW,                 WRTMUC },
+        [20] = {        LDVARY,         WRTMUC },
+        [21] = { THRSW, LDVARY,         WRTMUC },
+        [22] = { UCB,                          },
+        [23] = { ROT,                          },
+        /* 24-30 reserved */
+        [31] = { SMIMM,         LDTMU,         },
+};
+
+static const struct v3d_qpu_sig v41_sig_map[] = {
+        /*      MISC       phys    R5 */
+        [0]  = {                          },
+        [1]  = { THRSW,                   },
+        [2]  = {                   LDUNIF },
+        [3]  = { THRSW,            LDUNIF },
+        [4]  = {           LDTMU,         },
+        [5]  = { THRSW,    LDTMU,         },
+        [6]  = {           LDTMU,  LDUNIF },
+        [7]  = { THRSW,    LDTMU,  LDUNIF },
+        [8]  = {           LDVARY,        },
+        [9]  = { THRSW,    LDVARY,        },
+        [10] = {           LDVARY, LDUNIF },
+        [11] = { THRSW,    LDVARY, LDUNIF },
+        [12] = { LDUNIFRF                 },
+        [13] = { THRSW,    LDUNIFRF       },
+        [14] = { SMIMM,    LDVARY,        },
+        [15] = { SMIMM,                   },
+        [16] = {           LDTLB,         },
+        [17] = {           LDTLBU,        },
+        [18] = {                          WRTMUC },
+        [19] = { THRSW,                   WRTMUC },
+        [20] = {           LDVARY,        WRTMUC },
+        [21] = { THRSW,    LDVARY,        WRTMUC },
+        [22] = { UCB,                     },
+        [23] = { ROT,                     },
+        /* 24-30 reserved */
+        [24] = {                   LDUNIFA},
+        [25] = { LDUNIFARF                },
+        [31] = { SMIMM,            LDTMU, },
+};
+
 bool
 v3d_qpu_sig_unpack(const struct v3d_device_info *devinfo,
                    uint32_t packed_sig,
@@ -164,11 +220,16 @@ v3d_qpu_sig_unpack(const struct v3d_device_info *devinfo,
         if (packed_sig >= ARRAY_SIZE(v33_sig_map))
                 return false;
 
-        *sig = v33_sig_map[packed_sig];
+        if (devinfo->ver >= 41)
+                *sig = v41_sig_map[packed_sig];
+        else if (devinfo->ver == 40)
+                *sig = v40_sig_map[packed_sig];
+        else
+                *sig = v33_sig_map[packed_sig];
 
         /* Signals with zeroed unpacked contents after element 0 are reserved. */
         return (packed_sig == 0 ||
-                memcmp(sig, &v33_sig_map[0], sizeof(*sig) != 0));
+                memcmp(sig, &v33_sig_map[0], sizeof(*sig)) != 0);
 }
 
 bool
@@ -178,7 +239,12 @@ v3d_qpu_sig_pack(const struct v3d_device_info *devinfo,
 {
         static const struct v3d_qpu_sig *map;
 
-        map = v33_sig_map;
+        if (devinfo->ver >= 41)
+                map = v41_sig_map;
+        else if (devinfo->ver == 40)
+                map = v40_sig_map;
+        else
+                map = v33_sig_map;
 
         for (int i = 0; i < ARRAY_SIZE(v33_sig_map); i++) {
                 if (memcmp(&map[i], sig, sizeof(*sig)) == 0) {
@@ -189,6 +255,69 @@ v3d_qpu_sig_pack(const struct v3d_device_info *devinfo,
 
         return false;
 }
+static inline unsigned
+fui( float f )
+{
+        union {float f; unsigned ui;} fi;
+   fi.f = f;
+   return fi.ui;
+}
+
+static const uint32_t small_immediates[] = {
+        0, 1, 2, 3,
+        4, 5, 6, 7,
+        8, 9, 10, 11,
+        12, 13, 14, 15,
+        -16, -15, -14, -13,
+        -12, -11, -10, -9,
+        -8, -7, -6, -5,
+        -4, -3, -2, -1,
+        0x3b800000, /* 2.0^-8 */
+        0x3c000000, /* 2.0^-7 */
+        0x3c800000, /* 2.0^-6 */
+        0x3d000000, /* 2.0^-5 */
+        0x3d800000, /* 2.0^-4 */
+        0x3e000000, /* 2.0^-3 */
+        0x3e800000, /* 2.0^-2 */
+        0x3f000000, /* 2.0^-1 */
+        0x3f800000, /* 2.0^0 */
+        0x40000000, /* 2.0^1 */
+        0x40800000, /* 2.0^2 */
+        0x41000000, /* 2.0^3 */
+        0x41800000, /* 2.0^4 */
+        0x42000000, /* 2.0^5 */
+        0x42800000, /* 2.0^6 */
+        0x43000000, /* 2.0^7 */
+};
+
+bool
+v3d_qpu_small_imm_unpack(const struct v3d_device_info *devinfo,
+                         uint32_t packed_small_immediate,
+                         uint32_t *small_immediate)
+{
+        if (packed_small_immediate >= ARRAY_SIZE(small_immediates))
+                return false;
+
+        *small_immediate = small_immediates[packed_small_immediate];
+        return true;
+}
+
+bool
+v3d_qpu_small_imm_pack(const struct v3d_device_info *devinfo,
+                       uint32_t value,
+                       uint32_t *packed_small_immediate)
+{
+        STATIC_ASSERT(ARRAY_SIZE(small_immediates) == 48);
+
+        for (int i = 0; i < ARRAY_SIZE(small_immediates); i++) {
+                if (small_immediates[i] == value) {
+                        *packed_small_immediate = i;
+                        return true;
+                }
+        }
+
+        return false;
+}
 
 bool
 v3d_qpu_flags_unpack(const struct v3d_device_info *devinfo,
@@ -363,7 +492,8 @@ static const struct opcode_desc add_ops[] = {
         { 186, 186, 1 << 1, ANYMUX, V3D_QPU_A_NEG },
         { 186, 186, 1 << 2, ANYMUX, V3D_QPU_A_FLAPUSH },
         { 186, 186, 1 << 3, ANYMUX, V3D_QPU_A_FLBPUSH },
-        { 186, 186, 1 << 4, ANYMUX, V3D_QPU_A_FLBPOP },
+        { 186, 186, 1 << 4, ANYMUX, V3D_QPU_A_FLPOP },
+        { 186, 186, 1 << 5, ANYMUX, V3D_QPU_A_RECIP },
         { 186, 186, 1 << 6, ANYMUX, V3D_QPU_A_SETMSF },
         { 186, 186, 1 << 7, ANYMUX, V3D_QPU_A_SETREVF },
         { 187, 187, 1 << 0, 1 << 0, V3D_QPU_A_NOP, 0 },
@@ -382,11 +512,23 @@ static const struct opcode_desc add_ops[] = {
 
         { 187, 187, 1 << 2, 1 << 0, V3D_QPU_A_MSF },
         { 187, 187, 1 << 2, 1 << 1, V3D_QPU_A_REVF },
-        { 187, 187, 1 << 2, 1 << 2, V3D_QPU_A_VDWWT },
+        { 187, 187, 1 << 2, 1 << 2, V3D_QPU_A_VDWWT, 33 },
+        { 187, 187, 1 << 2, 1 << 2, V3D_QPU_A_IID, 40 },
+        { 187, 187, 1 << 2, 1 << 3, V3D_QPU_A_SAMPID, 40 },
+        { 187, 187, 1 << 2, 1 << 4, V3D_QPU_A_BARRIERID, 40 },
         { 187, 187, 1 << 2, 1 << 5, V3D_QPU_A_TMUWT },
         { 187, 187, 1 << 2, 1 << 6, V3D_QPU_A_VPMWT },
 
-        { 187, 187, 1 << 3, ANYMUX, V3D_QPU_A_VPMSETUP },
+        { 187, 187, 1 << 3, ANYMUX, V3D_QPU_A_VPMSETUP, 33 },
+        { 188, 188, 1 << 0, ANYMUX, V3D_QPU_A_LDVPMV_IN, 40 },
+        { 188, 188, 1 << 1, ANYMUX, V3D_QPU_A_LDVPMD_IN, 40 },
+        { 188, 188, 1 << 2, ANYMUX, V3D_QPU_A_LDVPMP, 40 },
+        { 188, 188, 1 << 3, ANYMUX, V3D_QPU_A_RSQRT, 41 },
+        { 188, 188, 1 << 4, ANYMUX, V3D_QPU_A_EXP, 41 },
+        { 188, 188, 1 << 5, ANYMUX, V3D_QPU_A_LOG, 41 },
+        { 188, 188, 1 << 6, ANYMUX, V3D_QPU_A_SIN, 41 },
+        { 188, 188, 1 << 7, ANYMUX, V3D_QPU_A_RSQRT2, 41 },
+        { 189, 189, ANYMUX, ANYMUX, V3D_QPU_A_LDVPMG_IN, 40 },
 
         /* FIXME: MORE COMPLICATED */
         /* { 190, 191, ANYMUX, ANYMUX, V3D_QPU_A_VFMOVABSNEGNAB }, */
@@ -694,7 +836,24 @@ v3d_qpu_add_unpack(const struct v3d_device_info *devinfo, uint64_t packed_inst,
         instr->alu.add.a = mux_a;
         instr->alu.add.b = mux_b;
         instr->alu.add.waddr = QPU_GET_FIELD(packed_inst, V3D_QPU_WADDR_A);
-        instr->alu.add.magic_write = packed_inst & VC5_QPU_MA;
+
+        instr->alu.add.magic_write = false;
+        if (packed_inst & VC5_QPU_MA) {
+                switch (instr->alu.add.op) {
+                case V3D_QPU_A_LDVPMV_IN:
+                        instr->alu.add.op = V3D_QPU_A_LDVPMV_OUT;
+                        break;
+                case V3D_QPU_A_LDVPMD_IN:
+                        instr->alu.add.op = V3D_QPU_A_LDVPMD_OUT;
+                        break;
+                case V3D_QPU_A_LDVPMG_IN:
+                        instr->alu.add.op = V3D_QPU_A_LDVPMG_OUT;
+                        break;
+                default:
+                        instr->alu.add.magic_write = true;
+                        break;
+                }
+        }
 
         return true;
 }
@@ -743,6 +902,19 @@ v3d_qpu_mul_unpack(const struct v3d_device_info *devinfo, uint64_t packed_inst,
                 }
 
                 break;
+
+        case V3D_QPU_M_VFMUL:
+                instr->alu.mul.output_pack = V3D_QPU_PACK_NONE;
+
+                if (!v3d_qpu_float16_unpack_unpack(((op & 0x7) - 4) & 7,
+                                                   &instr->alu.mul.a_unpack)) {
+                        return false;
+                }
+
+                instr->alu.mul.b_unpack = V3D_QPU_UNPACK_NONE;
+
+                break;
+
         default:
                 instr->alu.mul.output_pack = V3D_QPU_PACK_NONE;
                 instr->alu.mul.a_unpack = V3D_QPU_UNPACK_NONE;
@@ -788,16 +960,36 @@ v3d_qpu_add_pack(const struct v3d_device_info *devinfo,
         if (nsrc < 1)
                 mux_a = ffs(desc->mux_a_mask) - 1;
 
+        bool no_magic_write = false;
+
         switch (instr->alu.add.op) {
         case V3D_QPU_A_STVPMV:
                 waddr = 0;
+                no_magic_write = true;
                 break;
         case V3D_QPU_A_STVPMD:
                 waddr = 1;
+                no_magic_write = true;
                 break;
         case V3D_QPU_A_STVPMP:
                 waddr = 2;
+                no_magic_write = true;
                 break;
+
+        case V3D_QPU_A_LDVPMV_IN:
+        case V3D_QPU_A_LDVPMD_IN:
+        case V3D_QPU_A_LDVPMP:
+        case V3D_QPU_A_LDVPMG_IN:
+                assert(!instr->alu.add.magic_write);
+                break;
+
+        case V3D_QPU_A_LDVPMV_OUT:
+        case V3D_QPU_A_LDVPMD_OUT:
+        case V3D_QPU_A_LDVPMG_OUT:
+                assert(!instr->alu.add.magic_write);
+                *packed_instr |= VC5_QPU_MA;
+                break;
+
         default:
                 break;
         }
@@ -923,7 +1115,7 @@ v3d_qpu_add_pack(const struct v3d_device_info *devinfo,
         *packed_instr |= QPU_SET_FIELD(mux_b, VC5_QPU_ADD_B);
         *packed_instr |= QPU_SET_FIELD(opcode, VC5_QPU_OP_ADD);
         *packed_instr |= QPU_SET_FIELD(waddr, V3D_QPU_WADDR_A);
-        if (instr->alu.add.magic_write)
+        if (instr->alu.add.magic_write && !no_magic_write)
                 *packed_instr |= VC5_QPU_MA;
 
         return true;
@@ -1002,6 +1194,27 @@ v3d_qpu_mul_pack(const struct v3d_device_info *devinfo,
                 break;
         }
 
+        case V3D_QPU_M_VFMUL: {
+                uint32_t packed;
+
+                if (instr->alu.mul.output_pack != V3D_QPU_PACK_NONE)
+                        return false;
+
+                if (!v3d_qpu_float16_unpack_pack(instr->alu.mul.a_unpack,
+                                                 &packed)) {
+                        return false;
+                }
+                if (instr->alu.mul.a_unpack == V3D_QPU_UNPACK_SWAP_16)
+                        opcode = 8;
+                else
+                        opcode |= (packed + 4) & 7;
+
+                if (instr->alu.mul.b_unpack != V3D_QPU_UNPACK_NONE)
+                        return false;
+
+                break;
+        }
+
         default:
                 break;
         }
@@ -1029,10 +1242,21 @@ v3d_qpu_instr_unpack_alu(const struct v3d_device_info *devinfo,
                                 &instr->sig))
                 return false;
 
-        if (!v3d_qpu_flags_unpack(devinfo,
-                                  QPU_GET_FIELD(packed_instr, VC5_QPU_COND),
-                                  &instr->flags))
-                return false;
+        uint32_t packed_cond = QPU_GET_FIELD(packed_instr, VC5_QPU_COND);
+        if (v3d_qpu_sig_writes_address(devinfo, &instr->sig)) {
+                instr->sig_addr = packed_cond & ~VC5_QPU_COND_SIG_MAGIC_ADDR;
+                instr->sig_magic = packed_cond & VC5_QPU_COND_SIG_MAGIC_ADDR;
+
+                instr->flags.ac = V3D_QPU_COND_NONE;
+                instr->flags.mc = V3D_QPU_COND_NONE;
+                instr->flags.apf = V3D_QPU_PF_NONE;
+                instr->flags.mpf = V3D_QPU_PF_NONE;
+                instr->flags.auf = V3D_QPU_UF_NONE;
+                instr->flags.muf = V3D_QPU_UF_NONE;
+        } else {
+                if (!v3d_qpu_flags_unpack(devinfo, packed_cond, &instr->flags))
+                        return false;
+        }
 
         instr->raddr_a = QPU_GET_FIELD(packed_instr, VC5_QPU_RADDR_A);
         instr->raddr_b = QPU_GET_FIELD(packed_instr, VC5_QPU_RADDR_B);
@@ -1130,9 +1354,28 @@ v3d_qpu_instr_pack_alu(const struct v3d_device_info *devinfo,
                         return false;
 
                 uint32_t flags;
-                if (!v3d_qpu_flags_pack(devinfo, &instr->flags, &flags))
-                        return false;
+                if (v3d_qpu_sig_writes_address(devinfo, &instr->sig)) {
+                        if (instr->flags.ac != V3D_QPU_COND_NONE ||
+                            instr->flags.mc != V3D_QPU_COND_NONE ||
+                            instr->flags.apf != V3D_QPU_PF_NONE ||
+                            instr->flags.mpf != V3D_QPU_PF_NONE ||
+                            instr->flags.auf != V3D_QPU_UF_NONE ||
+                            instr->flags.muf != V3D_QPU_UF_NONE) {
+                                return false;
+                        }
+
+                        flags = instr->sig_addr;
+                        if (instr->sig_magic)
+                                flags |= VC5_QPU_COND_SIG_MAGIC_ADDR;
+                } else {
+                        if (!v3d_qpu_flags_pack(devinfo, &instr->flags, &flags))
+                                return false;
+                }
+
                 *packed_instr |= QPU_SET_FIELD(flags, VC5_QPU_COND);
+        } else {
+                if (v3d_qpu_sig_writes_address(devinfo, &instr->sig))
+                        return false;
         }
 
         return true;
diff --git a/lib/mesa/src/broadcom/qpu/tests/qpu_disasm.c b/lib/mesa/src/broadcom/qpu/tests/qpu_disasm.c
index c7f6476de..2e8d98058 100644
--- a/lib/mesa/src/broadcom/qpu/tests/qpu_disasm.c
+++ b/lib/mesa/src/broadcom/qpu/tests/qpu_disasm.c
@@ -39,6 +39,20 @@ static const struct {
         { 33, 0x3f003186bb800000ull, "nop                  ; nop               ; ldvpm" },
         { 33, 0x3c002380b6edb000ull, "or  rf0, r3, r3      ; mov  vpm, r3" },
         { 33, 0x57403006bbb80000ull, "nop                  ; fmul  r0, rf0, r5 ; ldvpm; ldunif" },
+        { 33, 0x9c094adef634b000ull, "ffloor.ifb  rf30.l, r3; fmul.pushz  rf43.l, r5, r1.h" },
+        { 33, 0xb0044c56ba326840ull, "flpop  rf22, rf33    ; fmul.pushz  rf49.l, r4.h, r1.abs" },
+
+        /* vfmul input packing */
+        { 33, 0x101e8b6e8aad4000ull, "fmax.nornn  rf46, r4.l, r2.l; vfmul.ifnb  rf45, r3, r5" },
+        { 33, 0x1857d3c219825000ull, "faddnf.norc  r2.l, r5.l, r4; vfmul.ifb  rf15, r0.ll, r4; ldunif" },
+        { 33, 0x1c0a0dfde2294000ull, "fcmp.ifna  rf61.h, r4.abs, r2.l; vfmul  rf55, r2.hh, r1" },
+        { 33, 0x2011c89b402cc000ull, "fsub.norz  rf27, r4.abs, r1.abs; vfmul.ifa  rf34, r3.swp, r1" },
+
+        /* small immediates */
+        { 33, 0x5de24398bbdc6218ull, "vflb.andnn  rf24     ; fmul  rf14, -8, rf8.h" },
+        { 33, 0x25ef83d8b166f00full, "vfmin.pushn  rf24, 15.ff, r5; smul24.ifnb  rf15, r1, r3" },
+        { 33, 0xadedcdf70839f990ull, "faddnf.pushc  rf55, -16.l, r3.abs; fmul.ifb  rf55.l, rf38.l, r1.h" },
+        { 33, 0x7dff89fa6a01f020ull, "fsub.nornc  rf58.h, 0x3b800000.l, r3.l; fmul.ifnb  rf39, r0.h, r0.h" },
 
         /* branch conditions */
         { 33, 0x02000006002034c0ull, "b.anyap  rf19" },
@@ -57,6 +71,29 @@ static const struct {
         { 33, 0x041618d57c453000ull, "shl.andn  exp, r3, r2; add.ifb  rf35, r1, r2" },
         { 33, 0x7048e5da49272800ull, "fsub.ifa  rf26, r2.l, rf32; fmul.pushc  sin, r1.h, r1.abs; ldunif" },
 
+        /* v4.1 signals */
+        { 41, 0x1f010520cf60a000ull, "fcmp.andz  rf32, r2.h, r1.h; vfmul  rf20, r0.hh, r3; ldunifa" },
+        { 41, 0x932045e6c16ea000ull, "fcmp  rf38, r2.abs, r5; fmul  rf23.l, r3, r3.abs; ldunifarf.rf1" },
+        { 41, 0xd72f0434e43ae5c0ull, "fcmp  rf52.h, rf23, r5.abs; fmul  rf16.h, rf23, r1; ldunifarf.rf60" },
+        { 41, 0xdb3048eb9d533780ull, "fmax  rf43.l, r3.h, rf30; fmul  rf35.h, r4, r2.l; ldunifarf.r1" },
+        { 41, 0x733620471e6ce700ull, "faddnf  rf7.l, rf28.h, r1.l; fmul  r1, r3.h, r3.abs; ldunifarf.rsqrt2" },
+        { 41, 0x9c094adef634b000ull, "ffloor.ifb  rf30.l, r3; fmul.pushz  rf43.l, r5, r1.h" },
+
+        /* v4.1 opcodes */
+        { 41, 0x3de020c7bdfd200dull, "ldvpmg_in  rf7, r2, r2; mov  r3, 13" },
+        { 41, 0x3de02040f8ff7201ull, "stvpmv  1, rf8       ; mov  r1, 1" },
+        { 41, 0xd8000e50bb2d3000ull, "sampid  rf16         ; fmul  rf57.h, r3, r1.l" },
+
+        /* v4.1 SFU instructions. */
+        { 41, 0xe98d60c1ba2aef80ull, "recip  rf1, rf62     ; fmul  r3.h, r2.l, r1.l; ldunifrf.rf53" },
+        { 41, 0x7d87c2debc51c000ull, "rsqrt  rf30, r4      ; fmul  rf11, r4.h, r2.h; ldunifrf.rf31" },
+        { 41, 0xb182475abc2bb000ull, "rsqrt2  rf26, r3     ; fmul  rf29.l, r2.h, r1.abs; ldunifrf.rf9" },
+        { 41, 0x79880808bc0b6900ull, "sin  rf8, rf36       ; fmul  rf32, r2.h, r0.l; ldunifrf.rf32" },
+        { 41, 0x04092094bc5a28c0ull, "exp.ifb  rf20, r2    ; add  r2, rf35, r2" },
+        { 41, 0xe00648bfbc32a000ull, "log  rf63, r2        ; fmul.andnn  rf34.h, r4.l, r1.abs" },
+
+        /* v4.2 changes */
+        { 42, 0x3c203192bb814000ull, "barrierid  syncb     ; nop               ; thrsw" },
 };
 
 static void
@@ -84,9 +121,10 @@ main(int argc, char **argv)
         for (int i = 0; i < ARRAY_SIZE(tests); i++) {
                 devinfo.ver = tests[i].ver;
 
-                printf("Testing v%d.%d 0x%016llx... ",
+                printf("Testing v%d.%d 0x%016llx (\"%s\")... ",
                        devinfo.ver / 10, devinfo.ver % 10,
-                       (long long)tests[i].inst);
+                       (long long)tests[i].inst,
+                        tests[i].expected);
 
                 const char *disasm_output = v3d_qpu_disasm(&devinfo,
                                                            tests[i].inst);
diff --git a/lib/mesa/src/glx/apple/apple_glx_log.c b/lib/mesa/src/glx/apple/apple_glx_log.c
index a3f446c26..ea39d3095 100644
--- a/lib/mesa/src/glx/apple/apple_glx_log.c
+++ b/lib/mesa/src/glx/apple/apple_glx_log.c
@@ -97,6 +97,7 @@ void _apple_glx_vlog(int level, const char *file, const char *function,
         fprintf(stderr, "%-9s %24s:%-4d %s(%"PRIu64"): ",
                 _asl_level_string(level), file, line, function, thread);
         vfprintf(stderr, fmt, args2);
+        va_end(args2);
     }
 
     msg = asl_new(ASL_TYPE_MSG);
diff --git a/lib/mesa/src/glx/apple/meson.build b/lib/mesa/src/glx/apple/meson.build
new file mode 100644
index 000000000..c6b1d617c
--- /dev/null
+++ b/lib/mesa/src/glx/apple/meson.build
@@ -0,0 +1,61 @@
+# Copyright © 2017 Jon Turney
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+# library for native GL on macos
+files_libappleglx = files(
+  'apple_cgl.c',
+  'apple_cgl.h',
+  'appledri.c',
+  'appledri.h',
+  'appledristr.h',
+  'apple_glapi.c',
+  'apple_glx.c',
+  'apple_glx_context.c',
+  'apple_glx_context.h',
+  'apple_glx_drawable.c',
+  'apple_glx_drawable.h',
+  'apple_glx.h',
+  'apple_glx_log.c',
+  'apple_glx_log.h',
+  'apple_glx_pbuffer.c',
+  'apple_glx_pixmap.c',
+  'apple_glx_surface.c',
+  'apple_visual.c',
+  'apple_visual.h',
+  'apple_xgl_api.h',
+  'apple_xgl_api_read.c',
+  'apple_xgl_api_stereo.c',
+  'apple_xgl_api_viewport.c',
+  'glx_empty.c',
+)
+
+dep_xplugin = null_dep
+if with_dri_platform == 'apple'
+  dep_xplugin = meson.get_compiler('c').find_library('Xplugin')
+endif
+
+libappleglx = static_library(
+  'glxapple',
+  [files_libappleglx, glapitable_h],
+  include_directories: [inc_mesa, inc_glx, inc_src, inc_include, inc_glapi],
+  dependencies: [dep_xext, dep_xplugin],
+  c_args: [c_vis_args],
+  build_by_default: false,
+)
diff --git a/lib/mesa/src/glx/dri2_priv.h b/lib/mesa/src/glx/dri2_priv.h
index 30ab2cdd8..a93551b14 100644
--- a/lib/mesa/src/glx/dri2_priv.h
+++ b/lib/mesa/src/glx/dri2_priv.h
@@ -30,6 +30,9 @@
  *   Kristian Høgsberg (krh@redhat.com)
  */
 
+#ifndef DRI2_PRIV_H
+#define DRI2_PRIV_H
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -82,3 +85,5 @@ dri2_interop_export_object(struct glx_context *ctx,
 #ifdef __cplusplus
 }
 #endif
+
+#endif
diff --git a/lib/mesa/src/glx/dri_common.h b/lib/mesa/src/glx/dri_common.h
index 947d33177..4d97ff82b 100644
--- a/lib/mesa/src/glx/dri_common.h
+++ b/lib/mesa/src/glx/dri_common.h
@@ -78,6 +78,6 @@ extern bool
 dri2_convert_glx_attribs(unsigned num_attribs, const uint32_t *attribs,
                          unsigned *major_ver, unsigned *minor_ver,
                          uint32_t *render_type, uint32_t *flags, unsigned *api,
-                         int *reset, unsigned *error);
+                         int *reset, int *release, unsigned *error);
 
 #endif /* _DRI_COMMON_H */
diff --git a/lib/mesa/src/glx/drisw_priv.h b/lib/mesa/src/glx/drisw_priv.h
index 5d4790031..259fc864f 100644
--- a/lib/mesa/src/glx/drisw_priv.h
+++ b/lib/mesa/src/glx/drisw_priv.h
@@ -23,6 +23,11 @@
  * SOFTWARE.
  */
 
+#ifndef DRISW_PRIV_H
+#define DRISW_PRIV_H
+
+#include <X11/extensions/XShm.h>
+
 struct drisw_display
 {
    __GLXDRIdisplay base;
@@ -62,6 +67,7 @@ struct drisw_drawable
    __DRIdrawable *driDrawable;
    XVisualInfo *visinfo;
    XImage *ximage;
+   XShmSegmentInfo shminfo;
 };
 
 _X_HIDDEN int
@@ -70,3 +76,5 @@ drisw_query_renderer_integer(struct glx_screen *base, int attribute,
 _X_HIDDEN int
 drisw_query_renderer_string(struct glx_screen *base, int attribute,
                             const char **value);
+
+#endif
diff --git a/lib/mesa/src/glx/glx_error.h b/lib/mesa/src/glx/glx_error.h
index 5d3992672..6a2f32aa5 100644
--- a/lib/mesa/src/glx/glx_error.h
+++ b/lib/mesa/src/glx/glx_error.h
@@ -26,6 +26,10 @@
  promote the sale, use or other dealings in this Software without
  prior written authorization.
 */
+
+#ifndef GLX_ERROR_H
+#define GLX_ERROR_H
+
 #include <stdbool.h>
 #include <stdint.h>
 #include <X11/Xlib.h>
@@ -45,3 +49,5 @@ _X_HIDDEN void __glXSendErrorForXcb(Display * dpy,
 #ifdef __cplusplus
 }
 #endif
+
+#endif
diff --git a/lib/mesa/src/glx/glxcurrent.c b/lib/mesa/src/glx/glxcurrent.c
index d1193265f..2b9c708c3 100644
--- a/lib/mesa/src/glx/glxcurrent.c
+++ b/lib/mesa/src/glx/glxcurrent.c
@@ -36,8 +36,8 @@
 #include <pthread.h>
 
 #include "glxclient.h"
-
 #include "glapi.h"
+#include "glx_error.h"
 
 /*
 ** We setup some dummy structures here so that the API can be used
@@ -165,21 +165,6 @@ glXGetCurrentDrawable(void)
    return gc->currentDrawable;
 }
 
-static void
-__glXGenerateError(Display * dpy, XID resource,
-                   BYTE errorCode, CARD16 minorCode)
-{
-   xError error;
-
-   error.errorCode = errorCode;
-   error.resourceID = resource;
-   error.sequenceNumber = dpy->request;
-   error.type = X_Error;
-   error.majorCode = __glXSetupForCommand(dpy);
-   error.minorCode = minorCode;
-   _XError(dpy, &error);
-}
-
 /**
  * Make a particular context current.
  *
@@ -209,6 +194,13 @@ MakeContextCurrent(Display * dpy, GLXDrawable draw,
       return True;
    }
 
+   /* can't have only one be 0 */
+   if (!!draw != !!read) {
+      __glXUnlock();
+      __glXSendError(dpy, BadMatch, None, X_GLXMakeContextCurrent, True);
+      return False;
+   }
+
    if (oldGC != &dummyContext) {
       if (--oldGC->thread_refcount == 0) {
 	 oldGC->vtable->unbind(oldGC, gc);
@@ -228,7 +220,8 @@ MakeContextCurrent(Display * dpy, GLXDrawable draw,
       if (gc->vtable->bind(gc, oldGC, draw, read) != Success) {
          __glXSetCurrentContextNull();
          __glXUnlock();
-         __glXGenerateError(dpy, None, GLXBadContext, X_GLXMakeContextCurrent);
+         __glXSendError(dpy, GLXBadContext, None, X_GLXMakeContextCurrent,
+                        False);
          return GL_FALSE;
       }
 
diff --git a/lib/mesa/src/glx/glxext.c b/lib/mesa/src/glx/glxext.c
index 5f23d3717..cef819203 100644
--- a/lib/mesa/src/glx/glxext.c
+++ b/lib/mesa/src/glx/glxext.c
@@ -343,9 +343,12 @@ static GLint
 convert_from_x_visual_type(int visualType)
 {
    static const int glx_visual_types[] = {
-      GLX_STATIC_GRAY, GLX_GRAY_SCALE,
-      GLX_STATIC_COLOR, GLX_PSEUDO_COLOR,
-      GLX_TRUE_COLOR, GLX_DIRECT_COLOR
+      [StaticGray]  = GLX_STATIC_GRAY,
+      [GrayScale]   = GLX_GRAY_SCALE,
+      [StaticColor] = GLX_STATIC_COLOR,
+      [PseudoColor] = GLX_PSEUDO_COLOR,
+      [TrueColor]   = GLX_TRUE_COLOR,
+      [DirectColor] = GLX_DIRECT_COLOR,
    };
 
    if (visualType < ARRAY_SIZE(glx_visual_types))
@@ -402,8 +405,6 @@ __glXInitializeVisualConfigFromTags(struct glx_config * config, int count,
 #endif
    }
 
-   config->sRGBCapable = GL_FALSE;
-
    /*
     ** Additional properties may be in a list at the end
     ** of the reply.  They are in pairs of property type
@@ -660,6 +661,8 @@ createConfigsFromProperties(Display * dpy, int nvisuals, int nprops,
        */
       m->drawableType = GLX_WINDOW_BIT | GLX_PIXMAP_BIT | GLX_PBUFFER_BIT;
 #endif
+      /* Older X servers don't send this so we default it here. */
+      m->sRGBCapable = GL_FALSE;
        __glXInitializeVisualConfigFromTags(m, nprops, props,
                                           tagged_only, GL_TRUE);
       m->screen = screen;
diff --git a/lib/mesa/src/glx/glxextensions.c b/lib/mesa/src/glx/glxextensions.c
index 6882e442f..e85a8c928 100644
--- a/lib/mesa/src/glx/glxextensions.c
+++ b/lib/mesa/src/glx/glxextensions.c
@@ -132,6 +132,7 @@ struct extension_info
 
 /* *INDENT-OFF* */
 static const struct extension_info known_glx_extensions[] = {
+   { GLX(ARB_context_flush_control),   VER(0,0), Y, N, N, N },
    { GLX(ARB_create_context),          VER(0,0), Y, N, N, N },
    { GLX(ARB_create_context_profile),  VER(0,0), Y, N, N, N },
    { GLX(ARB_create_context_robustness), VER(0,0), Y, N, N, N },
@@ -151,7 +152,7 @@ static const struct extension_info known_glx_extensions[] = {
    { GLX(ATI_pixel_format_float),      VER(0,0), N, N, N, N },
    { GLX(INTEL_swap_event),            VER(0,0), Y, N, N, N },
    { GLX(MESA_copy_sub_buffer),        VER(0,0), Y, N, N, N },
-   { GLX(MESA_multithread_makecurrent),VER(0,0), Y, N, Y, N },
+   { GLX(MESA_multithread_makecurrent),VER(0,0), Y, N, N, Y },
    { GLX(MESA_query_renderer),         VER(0,0), Y, N, N, Y },
    { GLX(MESA_swap_control),           VER(0,0), Y, N, N, Y },
    { GLX(NV_float_buffer),             VER(0,0), N, N, N, N },
@@ -160,8 +161,6 @@ static const struct extension_info known_glx_extensions[] = {
    { GLX(SGIS_multisample),            VER(0,0), Y, Y, N, N },
    { GLX(SGIX_fbconfig),               VER(1,3), Y, Y, N, N },
    { GLX(SGIX_pbuffer),                VER(1,3), Y, Y, N, N },
-   { GLX(SGIX_swap_barrier),           VER(0,0), N, N, N, N },
-   { GLX(SGIX_swap_group),             VER(0,0), N, N, N, N },
    { GLX(SGIX_visual_select_group),    VER(0,0), Y, Y, N, N },
    { GLX(SGI_make_current_read),       VER(1,3), Y, N, N, N },
    { GLX(SGI_swap_control),            VER(0,0), Y, N, N, N },
diff --git a/lib/mesa/src/glx/indirect_glx.c b/lib/mesa/src/glx/indirect_glx.c
index 4302a8ff2..cfae12f6c 100644
--- a/lib/mesa/src/glx/indirect_glx.c
+++ b/lib/mesa/src/glx/indirect_glx.c
@@ -62,13 +62,13 @@ indirect_destroy_context(struct glx_context *gc)
 }
 
 static Bool
-SendMakeCurrentRequest(Display * dpy, CARD8 opcode,
-                       GLXContextID gc_id, GLXContextTag gc_tag,
-                       GLXDrawable draw, GLXDrawable read,
-                       GLXContextTag *out_tag)
+SendMakeCurrentRequest(Display * dpy, GLXContextID gc_id,
+                       GLXContextTag gc_tag, GLXDrawable draw,
+                       GLXDrawable read, GLXContextTag *out_tag)
 {
    xGLXMakeCurrentReply reply;
    Bool ret;
+   int opcode = __glXSetupForCommand(dpy);
 
    LockDisplay(dpy);
 
@@ -136,7 +136,6 @@ indirect_bind_context(struct glx_context *gc, struct glx_context *old,
 {
    GLXContextTag tag;
    Display *dpy = gc->psc->dpy;
-   int opcode = __glXSetupForCommand(dpy);
    Bool sent;
 
    if (old != &dummyContext && !old->isDirect && old->psc->dpy == dpy) {
@@ -146,7 +145,7 @@ indirect_bind_context(struct glx_context *gc, struct glx_context *old,
       tag = 0;
    }
 
-   sent = SendMakeCurrentRequest(dpy, opcode, gc->xid, tag, draw, read,
+   sent = SendMakeCurrentRequest(dpy, gc->xid, tag, draw, read,
 				 &gc->currentContextTag);
 
    if (!IndirectAPI)
@@ -160,7 +159,6 @@ static void
 indirect_unbind_context(struct glx_context *gc, struct glx_context *new)
 {
    Display *dpy = gc->psc->dpy;
-   int opcode = __glXSetupForCommand(dpy);
 
    if (gc == new)
       return;
@@ -170,8 +168,8 @@ indirect_unbind_context(struct glx_context *gc, struct glx_context *new)
     * to send a request to the dpy to unbind the previous context.
     */
    if (!new || new->isDirect || new->psc->dpy != dpy) {
-      SendMakeCurrentRequest(dpy, opcode, None,
-			     gc->currentContextTag, None, None, NULL);
+      SendMakeCurrentRequest(dpy, None, gc->currentContextTag, None, None,
+                             NULL);
       gc->currentContextTag = 0;
    }
 }
diff --git a/lib/mesa/src/glx/meson.build b/lib/mesa/src/glx/meson.build
new file mode 100644
index 000000000..a61f959e8
--- /dev/null
+++ b/lib/mesa/src/glx/meson.build
@@ -0,0 +1,172 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+inc_glx = include_directories('.')
+
+subdir('apple')
+if with_dri_platform == 'windows'
+  subdir('windows')
+endif
+
+files_libglx = files(
+  'clientattrib.c',
+  'clientinfo.c',
+  'compsize.c',
+  'create_context.c',
+  'eval.c',
+  'glxclient.h',
+  'glxcmds.c',
+  'glxconfig.c',
+  'glxconfig.h',
+  'glxcurrent.c',
+  'glx_error.c',
+  'glx_error.h',
+  'glxext.c',
+  'glxextensions.c',
+  'glxextensions.h',
+  'glxhash.c',
+  'glxhash.h',
+  'glx_pbuffer.c',
+  'glx_query.c',
+  'indirect_glx.c',
+  'indirect_init.h',
+  'indirect_texture_compression.c',
+  'indirect_transpose_matrix.c',
+  'indirect_vertex_array.c',
+  'indirect_vertex_array.h',
+  'indirect_vertex_array_priv.h',
+  'indirect_vertex_program.c',
+  'indirect_window_pos.c',
+  'packrender.h',
+  'packsingle.h',
+  'pixel.c',
+  'pixelstore.c',
+  'query_renderer.c',
+  'render2.c',
+  'renderpix.c',
+  'single2.c',
+  'singlepix.c',
+  'vertarr.c',
+)
+
+extra_libs_libglx = []
+extra_deps_libgl = []
+extra_ld_args_libgl = []
+
+if with_dri
+  files_libglx += files(
+    'dri_common.c',
+    'dri_common.h',
+    'dri_common_query_renderer.c',
+    'dri_common_interop.c',
+    'xfont.c',
+    'drisw_glx.c',
+    'drisw_priv.h',
+  )
+endif
+
+# dri2
+if with_dri and with_dri_platform == 'drm' and dep_libdrm.found()
+  files_libglx += files(
+    'dri2.c',
+    'dri2_glx.c',
+    'dri2.h',
+    'dri2_priv.h',
+    'dri_glx.c',
+    'dri_sarea.h',
+    'XF86dri.c',
+    'xf86dri.h',
+    'xf86dristr.h',
+  )
+endif
+
+if with_dri3
+  files_libglx += files('dri3_glx.c', 'dri3_priv.h')
+endif
+
+if with_dri_platform == 'apple'
+  files_libglx += files('applegl_glx.c')
+  extra_libs_libglx += libappleglx
+elif with_dri_platform == 'windows'
+  files_libglx += files('driwindows_glx.c')
+  extra_libs_libglx += [
+    libwindowsdri,
+    libwindowsglx,
+  ]
+  extra_deps_libgl = [
+    meson.get_compiler('c').find_library('gdi32'),
+    meson.get_compiler('c').find_library('opengl32')
+  ]
+  extra_ld_args_libgl = '-Wl,--disable-stdcall-fixup'
+endif
+
+if not with_glvnd
+  gl_lib_name = 'GL'
+  gl_lib_version = '1.2.0'
+else
+  gl_lib_name = 'GLX_mesa'
+  gl_lib_version = '0.0.0'
+  files_libglx += files(
+    'g_glxglvnddispatchfuncs.c',
+    'g_glxglvnddispatchindices.h',
+    'glxglvnd.c',
+    'glxglvnd.h',
+    'glxglvnddispatchfuncs.h',
+  )
+endif
+
+gl_lib_cargs = [
+  '-D_REENTRANT',
+  '-DDEFAULT_DRIVER_DIR="@0@"'.format(dri_search_path),
+]
+
+libglx = static_library(
+  'glx',
+  [files_libglx, glx_generated],
+  include_directories : [inc_common, inc_glapi, inc_loader, inc_gl_internal],
+  c_args : [
+    c_vis_args, gl_lib_cargs,
+    '-DGL_LIB_NAME="lib@0@.so.@1@"'.format(gl_lib_name, gl_lib_version.split('.')[0]),
+  ],
+  link_with : [
+    libloader, libloader_dri3_helper, libmesa_util, libxmlconfig,
+    extra_libs_libglx,
+  ],
+  dependencies : [dep_libdrm, dep_dri2proto, dep_glproto, dep_x11, dep_glvnd],
+)
+
+libgl = shared_library(
+  gl_lib_name,
+  [],
+  link_with : [libglapi_static, libglapi],
+  link_whole : libglx,
+  link_args : [ld_args_bsymbolic, ld_args_gc_sections, extra_ld_args_libgl],
+  dependencies : [
+    dep_libdrm, dep_dl, dep_m, dep_thread, dep_x11, dep_xcb_glx, dep_xcb,
+    dep_x11_xcb, dep_xcb_dri2, dep_xext, dep_xfixes, dep_xdamage, dep_xxf86vm,
+    extra_deps_libgl,
+  ],
+  version : gl_lib_version,
+  install : true,
+)
+
+if with_tests
+  subdir('tests')
+endif
diff --git a/lib/mesa/src/glx/tests/fake_glx_screen.cpp b/lib/mesa/src/glx/tests/fake_glx_screen.cpp
index 801f54a6f..ca0ca3680 100644
--- a/lib/mesa/src/glx/tests/fake_glx_screen.cpp
+++ b/lib/mesa/src/glx/tests/fake_glx_screen.cpp
@@ -75,6 +75,18 @@ indirect_create_context_attribs(struct glx_screen *base,
    return indirect_create_context(base, config_base, shareList, 0);
 }
 
+#ifdef GLX_USE_APPLEGL
+#warning Indirect GLX tests are not built
+extern "C" struct glx_context *
+applegl_create_context(struct glx_screen *base,
+		       struct glx_config *config_base,
+		       struct glx_context *shareList,
+		       int renderType)
+{
+   return indirect_create_context(base, config_base, shareList, renderType);
+}
+#endif
+
 /* This is necessary so that we don't have to link with glxcurrent.c
  * which would require us to link with X libraries and what not.
  */
diff --git a/lib/mesa/src/glx/tests/indirect_api.cpp b/lib/mesa/src/glx/tests/indirect_api.cpp
index 34304a185..b9a4ca065 100644
--- a/lib/mesa/src/glx/tests/indirect_api.cpp
+++ b/lib/mesa/src/glx/tests/indirect_api.cpp
@@ -705,6 +705,8 @@ void __indirect_glFramebufferTextureLayer(void) { }
 }
 /*@}*/
 
+#ifndef GLX_USE_APPLEGL
+
 class IndirectAPI : public ::testing::Test {
 public:
    virtual void SetUp();
@@ -1518,3 +1520,5 @@ TEST_F(IndirectAPI, EXT_texture_array)
 {
    EXPECT_EQ((_glapi_proc) __indirect_glFramebufferTextureLayer, table[_glapi_get_proc_offset("glFramebufferTextureLayerEXT")]);
 }
+
+#endif
diff --git a/lib/mesa/src/glx/tests/meson.build b/lib/mesa/src/glx/tests/meson.build
new file mode 100644
index 000000000..e59b42d19
--- /dev/null
+++ b/lib/mesa/src/glx/tests/meson.build
@@ -0,0 +1,54 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+if with_shared_glapi
+  files_glx_test = files(
+    'clientinfo_unittest.cpp',
+    'create_context_unittest.cpp',
+    'enum_sizes.cpp',
+    'fake_glx_screen.cpp',
+    'fake_glx_screen.h',
+    'indirect_api.cpp',
+    'mock_xdisplay.h',
+    'query_renderer_unittest.cpp',
+  )
+  if with_dri2
+    files_glx_test += files('query_renderer_implementation_unittest.cpp')
+  endif
+
+  test(
+    'dispatch-index-check',
+    files('dispatch-index-check'),
+    suite : ['glx'],
+  )
+  test(
+    'glx-test',
+    executable(
+      'glx-test',
+      [files_glx_test, glx_indirect_size_h, main_dispatch_h],
+      link_with : [libglx, libglapi],
+      include_directories : [
+        inc_src, inc_include, inc_mesa, inc_mapi, inc_gl_internal,
+        inc_glx,
+      ],
+      dependencies : [dep_libdrm, dep_glproto, dep_thread, idep_gtest]
+    ),
+  )
+endif
diff --git a/lib/mesa/src/glx/windows/meson.build b/lib/mesa/src/glx/windows/meson.build
new file mode 100644
index 000000000..153f41ab7
--- /dev/null
+++ b/lib/mesa/src/glx/windows/meson.build
@@ -0,0 +1,68 @@
+# Copyright © 2017 Jon Turney
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+# protocol defines for the Windows-DRI server extension
+
+files_windowsdriproto = files(
+ 'windowsdriconst.h',
+ 'windowsdristr.h',
+ )
+
+install_headers(
+  files_windowsdriproto,
+  subdir: 'X11/extensions',
+)
+
+pkg.generate(
+  name : 'windowsdriproto',
+  description : 'Windows-DRI extension headers',
+  version : '1.0.0',
+)
+
+# library for using the Windows-DRI server extension
+files_libwindowsdri = files(
+  'xwindowsdri.c',
+  'xwindowsdri.h',
+)
+
+libwindowsdri = static_library(
+  'driwindows',
+  [files_libwindowsdri, files_windowsdriproto],
+  dependencies: dep_xext,
+  build_by_default: false,
+)
+
+# library for native GL on windows
+files_libwindowsglx = files(
+  'windowsgl.c',
+  'windowsgl.h',
+  'windowsgl_internal.h',
+  'windows_drawable.c',
+  'wgl.c',
+  'wgl.h',
+)
+
+libwindowsglx = static_library(
+  'glxwindows',
+  [files_libwindowsglx, files_windowsdriproto],
+  include_directories: [inc_include, inc_src, inc_glapi],
+  c_args : [c_vis_args],
+  build_by_default: false,
+)
diff --git a/lib/mesa/src/gtest/Makefile.am b/lib/mesa/src/gtest/Makefile.am
index 29d6c6d19..23b755361 100644
--- a/lib/mesa/src/gtest/Makefile.am
+++ b/lib/mesa/src/gtest/Makefile.am
@@ -37,4 +37,5 @@ EXTRA_DIST = \
 	src/gtest-port.cc \
 	src/gtest-printers.cc \
 	src/gtest-test-part.cc \
-	src/gtest-typed-test.cc
+	src/gtest-typed-test.cc \
+	meson.build
diff --git a/lib/mesa/src/gtest/meson.build b/lib/mesa/src/gtest/meson.build
new file mode 100644
index 000000000..23c21423d
--- /dev/null
+++ b/lib/mesa/src/gtest/meson.build
@@ -0,0 +1,32 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+libgtest = static_library(
+  'gtest',
+  files('src/gtest-all.cc', 'src/gtest_main.cc'),
+  include_directories : include_directories('include'),
+  cpp_args : '-w',
+  build_by_default : false,
+)
+
+idep_gtest = declare_dependency(
+  link_with : libgtest,
+  include_directories : include_directories('include', is_system : true),
+)
diff --git a/lib/mesa/src/hgl/meson.build b/lib/mesa/src/hgl/meson.build
new file mode 100644
index 000000000..ec06272ed
--- /dev/null
+++ b/lib/mesa/src/hgl/meson.build
@@ -0,0 +1,36 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+libgl = shared_library(
+  'GL',
+  files(
+    'GLView.cpp', 'GLRenderer.cpp', 'GLRendererRoster.cpp', 'GLDispatcher.cpp',
+  ),
+  link_args : [ld_args_bsymbolic, ld_args_gc_sections],
+  include_directories : [
+    inc_src, inc_mapi, inc_mesa, inc_include, inc_glapi, inc_haikugl,
+    inc_gl_internal, include_directories('/system/develop/headers/private')
+  ],
+  link_with : [libglapi_static, libglapi],
+  dependencies : cpp.find_library('be'),
+  install : true,
+)
+
+# TODO: We need some tests here
diff --git a/lib/mesa/src/mapi/es1api/meson.build b/lib/mesa/src/mapi/es1api/meson.build
new file mode 100644
index 000000000..d8a77a41d
--- /dev/null
+++ b/lib/mesa/src/mapi/es1api/meson.build
@@ -0,0 +1,61 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+es1_glapi_mapi_tmp_h = custom_target(
+  'es1_glapi_mapi_tmp.h',
+  input : [mapi_abi_py, gl_and_es_api_files],
+  output : 'glapi_mapi_tmp.h',
+  command : [prog_python, '@INPUT0@', '--printer', 'es1api', '@INPUT1@'],
+  depend_files : api_xml_files,
+  capture : true,
+)
+
+libglesv1_cm = shared_library(
+  'GLESv1_CM',
+  ['../entry.c', es1_glapi_mapi_tmp_h],
+  c_args : [
+    c_msvc_compat_args, c_vis_args, '-DMAPI_MODE_BRIDGE',
+    '-DMAPI_ABI_HEADER="@0@"'.format(es1_glapi_mapi_tmp_h.full_path()),
+  ],
+  link_args : [ld_args_gc_sections],
+  include_directories : [inc_src, inc_include, inc_mapi],
+  link_with : libglapi,
+  dependencies : [dep_thread, dep_libdrm, dep_m, dep_dl],
+  version : '1.0.0',
+  install : true,
+)
+
+pkg.generate(
+  name : 'glesv1_cm',
+  filebase : 'glesv1_cm',
+  description : 'Mesa OpenGL ES 1.1 CM library',
+  version : meson.project_version(),
+  libraries : libglesv1_cm,
+  libraries_private : gl_priv_libs,
+)
+
+if with_tests
+  test(
+    'es1-ABI-check',
+    find_program('ABI-check'),
+    env : env_test,
+    args : libglesv1_cm
+  )
+endif
diff --git a/lib/mesa/src/mapi/es2api/meson.build b/lib/mesa/src/mapi/es2api/meson.build
new file mode 100644
index 000000000..891e6f7b2
--- /dev/null
+++ b/lib/mesa/src/mapi/es2api/meson.build
@@ -0,0 +1,61 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+es2_glapi_mapi_tmp_h = custom_target(
+  'es2_glapi_mapi_tmp.h',
+  input : [mapi_abi_py, gl_and_es_api_files],
+  output : 'glapi_mapi_tmp.h',
+  command : [prog_python, '@INPUT0@', '--printer', 'es2api', '@INPUT1@'],
+  depend_files : api_xml_files,
+  capture : true,
+)
+
+libgles2 = shared_library(
+  'GLESv2',
+  ['../entry.c', es2_glapi_mapi_tmp_h],
+  c_args : [
+    c_msvc_compat_args, c_vis_args, '-DMAPI_MODE_BRIDGE',
+    '-DMAPI_ABI_HEADER="@0@"'.format(es2_glapi_mapi_tmp_h.full_path()),
+  ],
+  link_args : [ld_args_gc_sections],
+  include_directories : [inc_src, inc_include, inc_mapi],
+  link_with : libglapi,
+  dependencies : [dep_thread, dep_libdrm, dep_m, dep_dl],
+  version : '2.0.0',
+  install : true,
+)
+
+pkg.generate(
+  name : 'glesv2',
+  filebase : 'glesv2',
+  description : 'Mesa OpenGL ES 2.0 library',
+  version : meson.project_version(),
+  libraries : libgles2,
+  libraries_private : gl_priv_libs,
+)
+
+if with_tests
+  test(
+    'es2-ABI-check',
+    find_program('ABI-check'),
+    env : env_test,
+    args : libgles2
+  )
+endif
diff --git a/lib/mesa/src/mapi/glapi/SConscript b/lib/mesa/src/mapi/glapi/SConscript
index 994778a8f..118da1d53 100644
--- a/lib/mesa/src/mapi/glapi/SConscript
+++ b/lib/mesa/src/mapi/glapi/SConscript
@@ -19,17 +19,14 @@ if env['platform'] == 'windows':
         '_GDI32_', # prevent gl* being declared __declspec(dllimport) in MS headers
         'BUILD_GL32', # declare gl* as __declspec(dllexport) in Mesa headers
         'KHRONOS_DLL_EXPORTS', # declare gl* as __declspec(dllexport) in Khronos headers
+        '_GLAPI_NO_EXPORTS', # prevent _glapi_* from being declared __declspec(dllimport)
     ])
-    if env['gles']:
-        env.Append(CPPDEFINES = ['_GLAPI_DLL_EXPORTS'])
-    else:
-        # prevent _glapi_* from being declared __declspec(dllimport)
-        env.Append(CPPDEFINES = ['_GLAPI_NO_EXPORTS'])
 
 env.Append(CPPPATH = [
     '#/src',
     '#/src/mapi',
     '#/src/mesa',
+    Dir('.'), # src/mapi/glapi build path
     Dir('..'), # src/mapi build path
 ])
 
diff --git a/lib/mesa/src/mapi/glapi/gen/AMD_depth_clamp_separate.xml b/lib/mesa/src/mapi/glapi/gen/AMD_depth_clamp_separate.xml
new file mode 100644
index 000000000..b7142f1b7
--- /dev/null
+++ b/lib/mesa/src/mapi/glapi/gen/AMD_depth_clamp_separate.xml
@@ -0,0 +1,15 @@
+<?xml version="1.0"?>
+<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd">
+
+<OpenGLAPI>
+
+<category name="GL_AMD_depth_clamp_separate" number="401">
+    <enum name="DEPTH_CLAMP_NEAR_AMD" count="1"  value="0x901E">
+        <size name="Get" mode="get"/>
+    </enum>
+    <enum name="DEPTH_CLAMP_FAR_AMD" count="1"  value="0x901F">
+        <size name="Get" mode="get"/>
+    </enum>
+</category>
+
+</OpenGLAPI>
diff --git a/lib/mesa/src/mapi/glapi/gen/AMD_gpu_shader_int64.xml b/lib/mesa/src/mapi/glapi/gen/AMD_gpu_shader_int64.xml
new file mode 100644
index 000000000..42eb750d1
--- /dev/null
+++ b/lib/mesa/src/mapi/glapi/gen/AMD_gpu_shader_int64.xml
@@ -0,0 +1,239 @@
+<?xml version="1.0"?>
+<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd">
+
+<OpenGLAPI>
+
+<category name="GL_AMD_gpu_shader_int64" number="451">
+
+    <function name="Uniform1i64NV" alias="Uniform1i64ARB">
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint64"/>
+    </function>
+
+    <function name="Uniform2i64NV" alias="Uniform2i64ARB">
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint64"/>
+        <param name="y" type="GLint64"/>
+    </function>
+
+    <function name="Uniform3i64NV" alias="Uniform3i64ARB">
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint64"/>
+        <param name="y" type="GLint64"/>
+        <param name="z" type="GLint64"/>
+    </function>
+
+    <function name="Uniform4i64NV" alias="Uniform4i64ARB">
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint64"/>
+        <param name="y" type="GLint64"/>
+        <param name="z" type="GLint64"/>
+        <param name="w" type="GLint64"/>
+    </function>
+
+    <function name="Uniform1i64vNV" alias="Uniform1i64vARB">
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint64 *" count="count"/>
+    </function>
+
+    <function name="Uniform2i64vNV" alias="Uniform2i64vARB">
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint64 *" count="count" count_scale="2"/>
+    </function>
+
+    <function name="Uniform3i64vNV" alias="Uniform3i64vARB">
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint64 *" count="count" count_scale="3"/>
+    </function>
+
+    <function name="Uniform4i64vNV" alias="Uniform4i64vARB">
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint64 *" count="count" count_scale="4"/>
+    </function>
+
+    <function name="Uniform1ui64NV" alias="Uniform1ui64ARB">
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint64"/>
+    </function>
+
+    <function name="Uniform2ui64NV" alias="Uniform2ui64ARB">
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint64"/>
+        <param name="y" type="GLuint64"/>
+    </function>
+
+    <function name="Uniform3ui64NV" alias="Uniform3ui64ARB">
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint64"/>
+        <param name="y" type="GLuint64"/>
+        <param name="z" type="GLuint64"/>
+    </function>
+
+    <function name="Uniform4ui64NV" alias="Uniform4ui64ARB">
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint64"/>
+        <param name="y" type="GLuint64"/>
+        <param name="z" type="GLuint64"/>
+        <param name="w" type="GLuint64"/>
+    </function>
+
+    <function name="Uniform1ui64vNV" alias="Uniform1ui64vARB">
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint64 *"  count="count"/>
+    </function>
+
+    <function name="Uniform2ui64vNV" alias="Uniform2ui64vARB">
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint64 *" count="count" count_scale="2"/>
+    </function>
+
+    <function name="Uniform3ui64vNV" alias="Uniform3ui64vARB">
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint64 *" count="count" count_scale="3"/>
+    </function>
+
+    <function name="Uniform4ui64vNV" alias="Uniform4ui64vARB">
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint64 *" count="count" count_scale="4"/>
+    </function>
+
+    <function name="GetUniformi64vNV" alias="GetUniformi64vARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="params" type="GLint64 *"/>
+    </function>
+
+    <function name="GetUniformui64vNV" alias="GetUniformui64vARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="params" type="GLuint64 *"/>
+    </function>
+
+    <function name="ProgramUniform1i64NV" alias="ProgramUniform1i64ARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint64"/>
+    </function>
+
+    <function name="ProgramUniform2i64NV" alias="ProgramUniform2i64ARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint64"/>
+        <param name="y" type="GLint64"/>
+    </function>
+
+    <function name="ProgramUniform3i64NV" alias="ProgramUniform3i64ARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint64"/>
+        <param name="y" type="GLint64"/>
+        <param name="z" type="GLint64"/>
+    </function>
+
+    <function name="ProgramUniform4i64NV" alias="ProgramUniform4i64ARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLint64"/>
+        <param name="y" type="GLint64"/>
+        <param name="z" type="GLint64"/>
+        <param name="w" type="GLint64"/>
+    </function>
+
+    <function name="ProgramUniform1i64vNV" alias="ProgramUniform1i64vARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint64 *" count="count"/>
+    </function>
+
+    <function name="ProgramUniform2i64vNV" alias="ProgramUniform2i64vARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint64 *" count="count" count_scale="2"/>
+    </function>
+
+    <function name="ProgramUniform3i64vNV" alias="ProgramUniform3i64vARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint64 *" count="count" count_scale="3"/>
+    </function>
+
+    <function name="ProgramUniform4i64vNV" alias="ProgramUniform4i64vARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLint64 *" count="count" count_scale="4"/>
+    </function>
+
+    <function name="ProgramUniform1ui64NV" alias="ProgramUniform1ui64ARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint64"/>
+    </function>
+
+    <function name="ProgramUniform2ui64NV" alias="ProgramUniform2ui64ARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint64"/>
+        <param name="y" type="GLuint64"/>
+    </function>
+
+    <function name="ProgramUniform3ui64NV" alias="ProgramUniform3ui64ARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint64"/>
+        <param name="y" type="GLuint64"/>
+        <param name="z" type="GLuint64"/>
+    </function>
+
+    <function name="ProgramUniform4ui64NV" alias="ProgramUniform4ui64ARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="x" type="GLuint64"/>
+        <param name="y" type="GLuint64"/>
+        <param name="z" type="GLuint64"/>
+        <param name="w" type="GLuint64"/>
+    </function>
+
+    <function name="ProgramUniform1ui64vNV" alias="ProgramUniform1ui64vARB">
+        <param name="program" type="GLuint"/>
+	<param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint64 *" count="count"/>
+    </function>
+
+    <function name="ProgramUniform2ui64vNV" alias="ProgramUniform2ui64vARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint64 *" count="count" count_scale="2"/>
+    </function>
+
+    <function name="ProgramUniform3ui64vNV" alias="ProgramUniform3ui64vARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint64 *" count="count" count_scale="3"/>
+    </function>
+
+    <function name="ProgramUniform4ui64vNV" alias="ProgramUniform4ui64vARB">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="count" type="GLsizei"/>
+        <param name="value" type="const GLuint64 *" count="count" count_scale="4"/>
+    </function>
+
+</category>
+
+</OpenGLAPI>
+
diff --git a/lib/mesa/src/mapi/glapi/gen/AMD_performance_monitor.xml b/lib/mesa/src/mapi/glapi/gen/AMD_performance_monitor.xml
index b29dc5d90..785ea076c 100644
--- a/lib/mesa/src/mapi/glapi/gen/AMD_performance_monitor.xml
+++ b/lib/mesa/src/mapi/glapi/gen/AMD_performance_monitor.xml
@@ -5,13 +5,13 @@
 
 <category name="GL_AMD_performance_monitor" number="360">
 
-    <function name="GetPerfMonitorGroupsAMD">
+    <function name="GetPerfMonitorGroupsAMD" es2="2.0">
         <param name="numGroups" type="GLint *"/>
         <param name="groupsSize" type="GLsizei"/>
         <param name="groups" type="GLuint *"/>
     </function>
 
-    <function name="GetPerfMonitorCountersAMD">
+    <function name="GetPerfMonitorCountersAMD" es2="2.0">
         <param name="group" type="GLuint"/>
         <param name="numCounters" type="GLint *"/>
         <param name="maxActiveCounters" type="GLint *"/>
@@ -19,14 +19,14 @@
         <param name="counters" type="GLuint *"/>
     </function>
 
-    <function name="GetPerfMonitorGroupStringAMD">
+    <function name="GetPerfMonitorGroupStringAMD" es2="2.0">
         <param name="group" type="GLuint"/>
         <param name="bufSize" type="GLsizei"/>
         <param name="length" type="GLsizei *"/>
         <param name="groupString" type="GLchar *"/>
     </function>
 
-    <function name="GetPerfMonitorCounterStringAMD">
+    <function name="GetPerfMonitorCounterStringAMD" es2="2.0">
         <param name="group" type="GLuint"/>
         <param name="counter" type="GLuint"/>
         <param name="bufSize" type="GLsizei"/>
@@ -34,24 +34,24 @@
         <param name="counterString" type="GLchar *"/>
     </function>
 
-    <function name="GetPerfMonitorCounterInfoAMD">
+    <function name="GetPerfMonitorCounterInfoAMD" es2="2.0">
         <param name="group" type="GLuint"/>
         <param name="counter" type="GLuint"/>
         <param name="pname" type="GLenum"/>
         <param name="data" type="GLvoid *"/>
     </function>
 
-    <function name="GenPerfMonitorsAMD">
+    <function name="GenPerfMonitorsAMD" es2="2.0">
         <param name="n" type="GLsizei"/>
         <param name="monitors" type="GLuint *"/>
     </function>
 
-    <function name="DeletePerfMonitorsAMD">
+    <function name="DeletePerfMonitorsAMD" es2="2.0">
         <param name="n" type="GLsizei"/>
         <param name="monitors" type="GLuint *"/>
     </function>
 
-    <function name="SelectPerfMonitorCountersAMD">
+    <function name="SelectPerfMonitorCountersAMD" es2="2.0">
         <param name="monitor" type="GLuint"/>
         <param name="enable" type="GLboolean"/>
         <param name="group" type="GLuint"/>
@@ -59,15 +59,15 @@
         <param name="counterList" type="GLuint *"/>
     </function>
 
-    <function name="BeginPerfMonitorAMD">
+    <function name="BeginPerfMonitorAMD" es2="2.0">
         <param name="monitor" type="GLuint"/>
     </function>
 
-    <function name="EndPerfMonitorAMD">
+    <function name="EndPerfMonitorAMD" es2="2.0">
         <param name="monitor" type="GLuint"/>
     </function>
 
-    <function name="GetPerfMonitorCounterDataAMD">
+    <function name="GetPerfMonitorCounterDataAMD" es2="2.0">
         <param name="monitor" type="GLuint"/>
         <param name="pname" type="GLenum"/>
         <param name="dataSize" type="GLsizei"/>
diff --git a/lib/mesa/src/mapi/glapi/gen/ARB_draw_indirect.xml b/lib/mesa/src/mapi/glapi/gen/ARB_draw_indirect.xml
index 3b29d6b86..b4399f1e8 100644
--- a/lib/mesa/src/mapi/glapi/gen/ARB_draw_indirect.xml
+++ b/lib/mesa/src/mapi/glapi/gen/ARB_draw_indirect.xml
@@ -42,4 +42,24 @@
 </category>
 
 
+<category name="GL_AMD_multi_draw_indirect" number="408">
+
+    <function name="MultiDrawArraysIndirectAMD" exec="dynamic" alias="MultiDrawArraysIndirect">
+        <param name="mode" type="GLenum"/>
+        <param name="indirect" type="const GLvoid *"/>
+        <param name="primcount" type="GLsizei"/>
+        <param name="stride" type="GLsizei"/>
+    </function>
+
+    <function name="MultiDrawElementsIndirectAMD" exec="dynamic" alias="MultiDrawElementsIndirect">
+        <param name="mode" type="GLenum"/>
+        <param name="type" type="GLenum"/>
+        <param name="indirect" type="const GLvoid *"/>
+        <param name="primcount" type="GLsizei"/>
+        <param name="stride" type="GLsizei"/>
+    </function>
+
+</category>
+
+
 </OpenGLAPI>
diff --git a/lib/mesa/src/mapi/glapi/gen/ARB_gl_spirv.xml b/lib/mesa/src/mapi/glapi/gen/ARB_gl_spirv.xml
new file mode 100644
index 000000000..0dd615480
--- /dev/null
+++ b/lib/mesa/src/mapi/glapi/gen/ARB_gl_spirv.xml
@@ -0,0 +1,21 @@
+<?xml version="1.0"?>
+<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd">
+
+<OpenGLAPI>
+
+<category name="GL_ARB_gl_spirv" number="190">
+
+    <enum name="SHADER_BINARY_FORMAT_SPIR_V_ARB" value="0x9551"/>
+    <enum name="SPIR_V_BINARY_ARB" value="0x9552"/>
+
+    <function name="SpecializeShaderARB">
+        <param name="shader" type="GLuint"/>
+        <param name="pEntryPoint" type="const GLchar *"/>
+        <param name="numSpecializationConstants" type="GLuint"/>
+        <param name="pConstantIndex" type="const GLuint *"/>
+        <param name="pConstantValue" type="const GLuint *"/>
+    </function>
+
+</category>
+
+</OpenGLAPI>
diff --git a/lib/mesa/src/mapi/glapi/gen/EXT_framebuffer_object.xml b/lib/mesa/src/mapi/glapi/gen/EXT_framebuffer_object.xml
index 310e8ee95..6c0e54af1 100644
--- a/lib/mesa/src/mapi/glapi/gen/EXT_framebuffer_object.xml
+++ b/lib/mesa/src/mapi/glapi/gen/EXT_framebuffer_object.xml
@@ -75,7 +75,7 @@
 	<return type="GLboolean"/>
     </function>
 
-    <function name="BindRenderbufferEXT" deprecated="3.1">
+    <function name="BindRenderbufferEXT">
         <param name="target" type="GLenum"/>
         <param name="renderbuffer" type="GLuint"/>
         <glx rop="4316"/>
@@ -109,7 +109,7 @@
 	<return type="GLboolean"/>
     </function>
 
-    <function name="BindFramebufferEXT" deprecated="3.1">
+    <function name="BindFramebufferEXT">
         <param name="target" type="GLenum"/>
         <param name="framebuffer" type="GLuint"/>
         <glx rop="4319"/>
diff --git a/lib/mesa/src/mapi/glapi/gen/EXT_vertex_attrib_64bit.xml b/lib/mesa/src/mapi/glapi/gen/EXT_vertex_attrib_64bit.xml
new file mode 100644
index 000000000..6b63b122f
--- /dev/null
+++ b/lib/mesa/src/mapi/glapi/gen/EXT_vertex_attrib_64bit.xml
@@ -0,0 +1,70 @@
+<?xml version="1.0"?>
+<!DOCTYPE OpenGLAPI SYSTEM "gl_API.dtd">
+
+<OpenGLAPI>
+
+<category name="GL_EXT_vertex_attrib_64bit" number="387">
+
+    <function name="VertexAttribL1dEXT" alias="VertexAttribL1d">
+        <param name="index" type="GLuint"/>
+        <param name="x" type="GLdouble"/>
+    </function>
+
+    <function name="VertexAttribL2dEXT" alias="VertexAttribL2d">
+        <param name="index" type="GLuint"/>
+        <param name="x" type="GLdouble"/>
+        <param name="y" type="GLdouble"/>
+    </function>
+
+    <function name="VertexAttribL3dEXT" alias="VertexAttribL3d">
+        <param name="index" type="GLuint"/>
+        <param name="x" type="GLdouble"/>
+        <param name="y" type="GLdouble"/>
+        <param name="z" type="GLdouble"/>
+    </function>
+
+    <function name="VertexAttribL4dEXT" alias="VertexAttribL4d">
+        <param name="index" type="GLuint"/>
+        <param name="x" type="GLdouble"/>
+        <param name="y" type="GLdouble"/>
+        <param name="z" type="GLdouble"/>
+        <param name="w" type="GLdouble"/>
+    </function>
+
+    <function name="VertexAttribL1dvEXT" alias="VertexAttribL1dv">
+        <param name="index" type="GLuint"/>
+        <param name="v" type="const GLdouble *"/>
+    </function>
+
+    <function name="VertexAttribL2dvEXT" alias="VertexAttribL2dv">
+        <param name="index" type="GLuint"/>
+        <param name="v" type="const GLdouble *"/>
+    </function>
+
+    <function name="VertexAttribL3dvEXT" alias="VertexAttribL3dv">
+        <param name="index" type="GLuint"/>
+        <param name="v" type="const GLdouble *"/>
+    </function>
+
+    <function name="VertexAttribL4dvEXT" alias="VertexAttribL4dv">
+        <param name="index" type="GLuint"/>
+        <param name="v" type="const GLdouble *"/>
+    </function>
+
+    <function name="VertexAttribLPointerEXT" alias="VertexAttribLPointer">
+        <param name="index" type="GLuint"/>
+        <param name="size" type="GLint"/>
+        <param name="type" type="GLenum"/>
+        <param name="stride" type="GLsizei"/>
+        <param name="pointer" type="const GLvoid *"/>
+    </function>
+
+    <function name="GetVertexAttribLdvEXT" alias="GetVertexAttribLdv">
+        <param name="index" type="GLuint"/>
+        <param name="pname" type="GLenum"/>
+        <param name="params" type="GLdouble *"/>
+    </function>
+</category>
+
+</OpenGLAPI>
+
diff --git a/lib/mesa/src/mapi/glapi/gen/GL4x.xml b/lib/mesa/src/mapi/glapi/gen/GL4x.xml
index 512addc2b..2116286b3 100644
--- a/lib/mesa/src/mapi/glapi/gen/GL4x.xml
+++ b/lib/mesa/src/mapi/glapi/gen/GL4x.xml
@@ -68,12 +68,50 @@
 </category>
 
 <category name="4.6">
+  <enum name="PARAMETER_BUFFER"                          value="0x80EE"/>
+  <enum name="PARAMETER_BUFFER_BINDING"                  value="0x80EF"/>
+  <enum name="POLYGON_OFFSET_CLAMP"                      value="0x8E1B"/>
+  <enum name="SHADER_BINARY_FORMAT_SPIR_V"               value="0x9551"/>
+  <enum name="SPIR_V_BINARY"                             value="0x9552"/>
+
+  <!-- This function aliases one from GL_EXT_polygon_offset_clamp -->
+
   <function name="PolygonOffsetClamp" alias="PolygonOffsetClampEXT">
     <param name="factor" type="GLfloat"/>
     <param name="units" type="GLfloat"/>
     <param name="clamp" type="GLfloat"/>
   </function>
-  <enum name="POLYGON_OFFSET_CLAMP" value="0x8E1B"/>
+
+  <!-- This function aliases one from GL_ARB_gl_spirv -->
+
+  <function name="SpecializeShader" alias="SpecializeShaderARB">
+    <param name="shader" type="GLuint"/>
+    <param name="pEntryPoint" type="const GLchar *"/>
+    <param name="numSpecializationConstants" type="GLuint"/>
+    <param name="pConstantIndex" type="const GLuint *"/>
+    <param name="pConstantValue" type="const GLuint *"/>
+  </function>
+
+  <!-- These functions alias ones from GL_ARB_indirect_parameters -->
+
+  <function name="MultiDrawArraysIndirectCount"
+            alias="MultiDrawArraysIndirectCountARB">
+    <param name="mode" type="GLenum"/>
+    <param name="indirect" type="GLintptr"/>
+    <param name="drawcount" type="GLintptr"/>
+    <param name="maxdrawcount" type="GLsizei"/>
+    <param name="stride" type="GLsizei"/>
+  </function>
+
+  <function name="MultiDrawElementsIndirectCount"
+            alias="MultiDrawElementsIndirectCountARB">
+    <param name="mode" type="GLenum"/>
+    <param name="type" type="GLenum"/>
+    <param name="indirect" type="GLintptr"/>
+    <param name="drawcount" type="GLintptr"/>
+    <param name="maxdrawcount" type="GLsizei"/>
+    <param name="stride" type="GLsizei"/>
+  </function>
 </category>
 
 </OpenGLAPI>
diff --git a/lib/mesa/src/mapi/glapi/gen/KHR_robustness_es.xml b/lib/mesa/src/mapi/glapi/gen/KHR_robustness_es.xml
index 84f6fd2cd..82b7edf31 100644
--- a/lib/mesa/src/mapi/glapi/gen/KHR_robustness_es.xml
+++ b/lib/mesa/src/mapi/glapi/gen/KHR_robustness_es.xml
@@ -60,4 +60,36 @@
 
 </category>
 
+<category name="GL_EXT_robustness" number="107">
+    <function name="GetGraphicsResetStatusEXT"
+              alias="GetGraphicsResetStatusARB" es2="2.0">
+        <return type="GLenum"/>
+    </function>
+
+    <function name="ReadnPixelsEXT" alias="ReadnPixelsARB" es2="2.0">
+        <param name="x" type="GLint"/>
+        <param name="y" type="GLint"/>
+        <param name="width" type="GLsizei"/>
+        <param name="height" type="GLsizei"/>
+        <param name="format" type="GLenum"/>
+        <param name="type" type="GLenum"/>
+        <param name="bufSize" type="GLsizei"/>
+        <param name="data" type="GLvoid *" output="true"/>
+    </function>
+
+    <function name="GetnUniformfvEXT" alias="GetnUniformfvARB" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="bufSize" type="GLsizei"/>
+        <param name="params" type="GLfloat *" output="true"/>
+    </function>
+
+    <function name="GetnUniformivEXT" alias="GetnUniformivARB" es2="2.0">
+        <param name="program" type="GLuint"/>
+        <param name="location" type="GLint"/>
+        <param name="bufSize" type="GLsizei"/>
+        <param name="params" type="GLint *" output="true"/>
+    </function>
+</category>
+
 </OpenGLAPI>
diff --git a/lib/mesa/src/mapi/glapi/gen/meson.build b/lib/mesa/src/mapi/glapi/gen/meson.build
new file mode 100644
index 000000000..f494e9707
--- /dev/null
+++ b/lib/mesa/src/mapi/glapi/gen/meson.build
@@ -0,0 +1,273 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+gl_and_es_api_files = files('gl_and_es_API.xml')
+
+api_xml_files = files(
+  'gl_API.xml',
+  'es_EXT.xml',
+  'gl_and_es_API.xml',
+  'gl_and_glX_API.xml',
+  'ARB_base_instance.xml',
+  'ARB_blend_func_extended.xml',
+  'ARB_bindless_texture.xml',
+  'ARB_clear_buffer_object.xml',
+  'ARB_clear_texture.xml',
+  'ARB_clip_control.xml',
+  'ARB_color_buffer_float.xml',
+  'ARB_compressed_texture_pixel_storage.xml',
+  'ARB_compute_shader.xml',
+  'ARB_compute_variable_group_size.xml',
+  'ARB_copy_buffer.xml',
+  'ARB_copy_image.xml',
+  'ARB_debug_output.xml',
+  'ARB_depth_buffer_float.xml',
+  'ARB_depth_clamp.xml',
+  'ARB_direct_state_access.xml',
+  'ARB_draw_buffers.xml',
+  'ARB_draw_buffers_blend.xml',
+  'ARB_draw_elements_base_vertex.xml',
+  'ARB_draw_indirect.xml',
+  'ARB_draw_instanced.xml',
+  'ARB_ES2_compatibility.xml',
+  'ARB_ES3_compatibility.xml',
+  'ARB_framebuffer_no_attachments.xml',
+  'ARB_framebuffer_object.xml',
+  'ARB_get_program_binary.xml',
+  'ARB_get_texture_sub_image.xml',
+  'ARB_gl_spirv.xml',
+  'ARB_gpu_shader_fp64.xml',
+  'ARB_gpu_shader_int64.xml',
+  'ARB_gpu_shader5.xml',
+  'ARB_indirect_parameters.xml',
+  'ARB_instanced_arrays.xml',
+  'ARB_internalformat_query.xml',
+  'ARB_internalformat_query2.xml',
+  'ARB_invalidate_subdata.xml',
+  'ARB_map_buffer_range.xml',
+  'ARB_multi_bind.xml',
+  'ARB_pipeline_statistics_query.xml',
+  'ARB_program_interface_query.xml',
+  'ARB_robustness.xml',
+  'ARB_sample_shading.xml',
+  'ARB_sampler_objects.xml',
+  'ARB_seamless_cube_map.xml',
+  'ARB_separate_shader_objects.xml',
+  'ARB_shader_atomic_counters.xml',
+  'ARB_shader_image_load_store.xml',
+  'ARB_shader_subroutine.xml',
+  'ARB_shader_storage_buffer_object.xml',
+  'ARB_sparse_buffer.xml',
+  'ARB_sync.xml',
+  'ARB_tessellation_shader.xml',
+  'ARB_texture_barrier.xml',
+  'ARB_texture_buffer_object.xml',
+  'ARB_texture_buffer_range.xml',
+  'ARB_texture_compression_rgtc.xml',
+  'ARB_texture_cube_map_array.xml',
+  'ARB_texture_float.xml',
+  'ARB_texture_gather.xml',
+  'ARB_texture_multisample.xml',
+  'ARB_texture_rgb10_a2ui.xml',
+  'ARB_texture_rg.xml',
+  'ARB_texture_storage_multisample.xml',
+  'ARB_texture_storage.xml',
+  'ARB_texture_view.xml',
+  'ARB_uniform_buffer_object.xml',
+  'ARB_vertex_array_object.xml',
+  'ARB_vertex_attrib_64bit.xml',
+  'ARB_vertex_attrib_binding.xml',
+  'ARB_viewport_array.xml',
+  'AMD_depth_clamp_separate.xml',
+  'AMD_draw_buffers_blend.xml',
+  'AMD_gpu_shader_int64.xml',
+  'AMD_performance_monitor.xml',
+  'ARB_vertex_type_2_10_10_10_rev.xml',
+  'APPLE_object_purgeable.xml',
+  'APPLE_vertex_array_object.xml',
+  'EXT_draw_buffers2.xml',
+  'EXT_external_objects.xml',
+  'EXT_external_objects_fd.xml',
+  'EXT_framebuffer_object.xml',
+  'EXT_gpu_shader4.xml',
+  'EXT_packed_depth_stencil.xml',
+  'EXT_provoking_vertex.xml',
+  'EXT_separate_shader_objects.xml',
+  'EXT_texture_array.xml',
+  'EXT_texture_integer.xml',
+  'EXT_transform_feedback.xml',
+  'EXT_vertex_attrib_64bit.xml',
+  'EXT_window_rectangles.xml',
+  'GREMEDY_string_marker.xml',
+  'INTEL_performance_query.xml',
+  'KHR_debug.xml',
+  'KHR_context_flush_control.xml',
+  'KHR_robustness.xml',
+  'KHR_robustness_es.xml',
+  'KHR_texture_compression_astc.xml',
+  'NV_conditional_render.xml',
+  'NV_primitive_restart.xml',
+  'NV_texture_barrier.xml',
+  'NV_vdpau_interop.xml',
+  'OES_EGL_image.xml',
+  'OES_fixed_point.xml',
+  'OES_single_precision.xml',
+  'OES_texture_compression_astc.xml',
+  'GL3x.xml',
+  'GL4x.xml',
+)
+
+glapi_gen_depends = files(
+  'gl_XML.py',
+  'glX_XML.py',
+  'license.py',
+  'static_data.py',
+  'typeexpr.py',
+) + api_xml_files
+
+glx_gen_depends = files(
+  'glX_API.xml',
+  'glX_XML.py',
+  'glX_proto_common.py',
+) + api_xml_files
+
+glapi_mapi_tmp_h = custom_target(
+  'glapi_mapi_tmp.h',
+  input : [mapi_abi_py, 'gl_and_es_API.xml'],
+  output : 'glapi_mapi_tmp.h',
+  command : [prog_python, '@INPUT0@', '--printer', 'glapi', '@INPUT1@'],
+  depend_files : glapi_gen_depends,
+  capture : true,
+)
+
+glprocs_h = custom_target(
+  'glprocs.h',
+  input : ['gl_procs.py', 'gl_and_es_API.xml'],
+  output : 'glprocs.h',
+  command : [prog_python, '@INPUT0@', '-c', '-f', '@INPUT1@'],
+  depend_files : glapi_gen_depends,
+  capture : true,
+)
+
+glapitemp_h = custom_target(
+  'glapitemp.h',
+  input : ['gl_apitemp.py', 'gl_and_es_API.xml'],
+  output : 'glapitemp.h',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  depend_files : glapi_gen_depends,
+  capture : true,
+)
+
+glapitable_h = custom_target(
+  'glapitable.h',
+  input : ['gl_table.py', 'gl_and_es_API.xml'],
+  output : 'glapitable.h',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  depend_files : glapi_gen_depends,
+  capture : true,
+)
+
+glapi_gentable_c = custom_target(
+  'glapi_gentable.c',
+  input : ['gl_gentable.py', 'gl_and_es_API.xml'],
+  output : 'glapi_gentable.c',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  depend_files : glapi_gen_depends,
+  capture : true,
+)
+
+main_enums_c = custom_target(
+  'enums.c',
+  input : ['gl_enums.py', files('../registry/gl.xml')],
+  output : 'enums.c',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  capture : true,
+)
+
+main_api_exec_c = custom_target(
+  'api_exec.c',
+  input : ['gl_genexec.py', 'gl_and_es_API.xml'],
+  output : 'api_exec.c',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  depend_files : files('apiexec.py') + glapi_gen_depends,
+  capture : true,
+)
+
+main_marshal_generated_c = custom_target(
+  'marshal_generated.c',
+  input : ['gl_marshal.py', 'gl_and_es_API.xml'],
+  output : 'marshal_generated.c',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  depend_files : files('marshal_XML.py') + glapi_gen_depends,
+  capture : true,
+)
+
+glx_generated = []
+
+foreach x : [['indirect.c', 'proto'], ['indirect.h', 'init_h'], ['indirect_init.c', 'init_c']]
+  glx_generated += custom_target(
+    x[0],
+    input : ['glX_proto_send.py', 'gl_API.xml'],
+    output : x[0],
+    command : [prog_python, '@INPUT0@', '-f', '@INPUT1@', '-m', x[1]],
+    depend_files : glx_gen_depends,
+    capture : true,
+  )
+endforeach
+
+foreach x : [['indirect_size.h', ['-m', 'size_h', '--header-tag', '_INDIRECT_SIZE_H_']],
+             ['indirect_size.c', ['-m', 'size_c']]]
+  glx_generated += custom_target(
+    x[0],
+    input : ['glX_proto_size.py', 'gl_API.xml'],
+    output : x[0],
+    command : [prog_python, '@INPUT0@', '-f', '@INPUT1@', '--only-set', x[1]],
+    depend_files : glx_gen_depends,
+    capture : true,
+  )
+endforeach
+glx_indirect_size_h = glx_generated[3]
+
+glapi_x86_s = custom_target(
+  'glapi_x86.S',
+  input : ['gl_x86_asm.py', gl_and_es_api_files],
+  output : 'glapi_x86.S',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  depend_files : glapi_gen_depends,
+  capture : true,
+)
+
+glapi_x86_64_s = custom_target(
+  'glapi_x86-64.S',
+  input : ['gl_x86-64_asm.py', gl_and_es_api_files],
+  output : 'glapi_x86-64.S',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  depend_files : glapi_gen_depends,
+  capture : true,
+)
+
+glapi_sparc_s = custom_target(
+  'glapi_sparc.S',
+  input : ['gl_SPARC_asm.py', gl_and_es_api_files],
+  output : 'glapi_sparc.S',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  depend_files : glapi_gen_depends,
+  capture : true,
+)
diff --git a/lib/mesa/src/mapi/glapi/glapi_dispatch.c b/lib/mesa/src/mapi/glapi/glapi_dispatch.c
index df907ff9d..ca15449ed 100644
--- a/lib/mesa/src/mapi/glapi/glapi_dispatch.c
+++ b/lib/mesa/src/mapi/glapi/glapi_dispatch.c
@@ -38,7 +38,7 @@
  */
 
 #include "glapi/glapi_priv.h"
-#include "glapi/glapitable.h"
+#include "glapitable.h"
 
 
 #if !(defined(USE_X86_ASM) || defined(USE_X86_64_ASM) || defined(USE_SPARC_ASM))
@@ -97,6 +97,13 @@
  */
 #include <GLES/glplatform.h>
 
+
+/* Redefine GL_API to avoid MSVC/MinGW warnings about different dllimport
+ * attributes for these prototypes vs those in the GLES/gl.h header.
+ */
+#undef GL_API
+#define GL_API KEYWORD1
+
 GL_API void GL_APIENTRY glClearDepthf (GLclampf depth);
 GL_API void GL_APIENTRY glClipPlanef (GLenum plane, const GLfloat *equation);
 GL_API void GL_APIENTRY glFrustumf (GLfloat left, GLfloat right, GLfloat bottom, GLfloat top, GLfloat zNear, GLfloat zFar);
@@ -143,6 +150,8 @@ GL_API void GL_APIENTRY glTexParameterx (GLenum target, GLenum pname, GLfixed pa
 GL_API void GL_APIENTRY glTexParameterxv (GLenum target, GLenum pname, const GLfixed *params);
 GL_API void GL_APIENTRY glTranslatex (GLfixed x, GLfixed y, GLfixed z);
 GL_API void GL_APIENTRY glPointSizePointerOES (GLenum type, GLsizei stride, const GLvoid *pointer);
+GL_API void GL_APIENTRY glBlendBarrier (void);
+GL_API void GL_APIENTRY glPrimitiveBoundingBox (GLfloat minX, GLfloat minY, GLfloat minZ, GLfloat minW, GLfloat maxX, GLfloat maxY, GLfloat maxZ, GLfloat maxW);
 
 /* Enable frame pointer elimination on Windows, otherwise forgetting to add
  * APIENTRY to _mesa_* entrypoints will not cause crashes on debug builds, as
diff --git a/lib/mesa/src/mapi/glapi/glapi_getproc.c b/lib/mesa/src/mapi/glapi/glapi_getproc.c
index a6b2455f1..d3fd3f82d 100644
--- a/lib/mesa/src/mapi/glapi/glapi_getproc.c
+++ b/lib/mesa/src/mapi/glapi/glapi_getproc.c
@@ -34,7 +34,7 @@
 #include <string.h>
 #include <stdlib.h>
 #include "glapi/glapi_priv.h"
-#include "glapi/glapitable.h"
+#include "glapitable.h"
 
 
 #define FIRST_DYNAMIC_OFFSET (sizeof(struct _glapi_table) / sizeof(void *))
@@ -49,7 +49,7 @@
 #if !defined(DISPATCH_FUNCTION_SIZE) 
 # define NEED_FUNCTION_POINTER
 #endif
-#include "glapi/glprocs.h"
+#include "glprocs.h"
 
 
 /**
diff --git a/lib/mesa/src/mapi/glapi/glapi_nop.c b/lib/mesa/src/mapi/glapi/glapi_nop.c
index 13db310b3..2e130e63e 100644
--- a/lib/mesa/src/mapi/glapi/glapi_nop.c
+++ b/lib/mesa/src/mapi/glapi/glapi_nop.c
@@ -116,7 +116,7 @@ NoOpUnused(void)
 #define DISPATCH_TABLE_NAME __glapi_noop_table
 #define UNUSED_TABLE_NAME __unused_noop_functions
 
-#include "glapi/glapitemp.h"
+#include "glapitemp.h"
 
 
 /** Return pointer to new dispatch table filled with no-op functions */
diff --git a/lib/mesa/src/mapi/glapi/glapi_priv.h b/lib/mesa/src/mapi/glapi/glapi_priv.h
index 8f2c6775d..ddfc594a9 100644
--- a/lib/mesa/src/mapi/glapi/glapi_priv.h
+++ b/lib/mesa/src/mapi/glapi/glapi_priv.h
@@ -27,25 +27,12 @@
 #define _GLAPI_PRIV_H
 
 
-#ifdef HAVE_DIX_CONFIG_H
-#include <dix-config.h>
-#include "glapi/mesa.h"
-#else /* HAVE_DIX_CONFIG_H */
 #define GL_GLEXT_PROTOTYPES
 #include "GL/gl.h"
 #include "GL/glext.h"
 
-#ifndef GL_OES_fixed_point
-typedef int GLfixed;
-#endif
 typedef int GLclampx;
 
-#ifndef GL_OES_EGL_image
-typedef void *GLeglImageOES;
-#endif
-
-#endif /* HAVE_DIX_CONFIG_H */
-
 #include "glapi/glapi.h"
 
 
diff --git a/lib/mesa/src/mapi/glapi/meson.build b/lib/mesa/src/mapi/glapi/meson.build
new file mode 100644
index 000000000..048bee8a1
--- /dev/null
+++ b/lib/mesa/src/mapi/glapi/meson.build
@@ -0,0 +1,92 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+subdir('gen')
+
+inc_glapi = include_directories('.')
+
+static_glapi_files = []
+static_glapi_args = []
+
+if ['apple', 'windows'].contains(with_dri_platform)
+  static_glapi_files += [glapi_gentable_c, glapitable_h]
+endif
+
+if with_shared_glapi
+  static_glapi_files += files(
+    '../entry.c',
+    '../entry.h',
+    '../entry_x86-64_tls.h',
+    '../entry_x86_tls.h',
+    '../entry_x86_tsd.h',
+    '../entry_ppc64le_tls.h',
+    '../entry_ppc64le_tsd.h',
+    '../mapi_tmp.h',
+  )
+  static_glapi_files += glapi_mapi_tmp_h
+  static_glapi_args += [
+    '-DMAPI_MODE_BRIDGE',
+    '-DMAPI_ABI_HEADER="@0@"'.format(glapi_mapi_tmp_h.full_path()),
+  ]
+else
+  static_glapi_args += '-DMAPI_MODE_UTIL'
+  static_glapi_files += files(
+    'glapi_dispatch.c',
+    'glapi_entrypoint.c',
+    'glapi_getproc.c',
+    'glapi_nop.c',
+    'glapi.c',
+    'glapi.h',
+    'glapi_priv.h',
+  )
+  static_glapi_files += files_mapi_util
+  static_glapi_files += [
+    glapitable_h, glapi_mapi_tmp_h, glprocs_h, glapitemp_h,
+  ]
+  if with_asm_arch == 'x86'
+    static_glapi_files += glapi_x86_s
+  elif with_asm_arch == 'x86_64'
+    static_glapi_files += glapi_x86_64_s
+  elif with_asm_arch == 'sparc'
+    static_glapi_files += glapi_sparc_s
+  endif
+endif
+
+libglapi_static = static_library(
+  'glapi_static',
+  static_glapi_files,
+  include_directories : [inc_mesa, inc_include, inc_src, inc_mapi],
+  c_args : [c_msvc_compat_args, static_glapi_args],
+  dependencies : [dep_thread, dep_selinux],
+  build_by_default : false,
+)
+
+if with_any_opengl and not with_shared_glapi and with_tests
+  test(
+    'glapi_static_check_table',
+    executable(
+      'glapi_static_check_table',
+      ['tests/check_table.cpp', glapitable_h],
+      include_directories : [inc_include, inc_src, inc_mesa, inc_mapi],
+      link_with : [libglapi_static],
+      dependencies : [idep_gtest, dep_thread],
+    )
+  )
+endif
diff --git a/lib/mesa/src/mapi/glapi/tests/check_table.cpp b/lib/mesa/src/mapi/glapi/tests/check_table.cpp
index f23f00be5..761f2a24e 100644
--- a/lib/mesa/src/mapi/glapi/tests/check_table.cpp
+++ b/lib/mesa/src/mapi/glapi/tests/check_table.cpp
@@ -22,10 +22,10 @@
  */
 
 #include <gtest/gtest.h>
-#include "../mesa/main/glheader.h"
+#include "main/glheader.h"
 
 #include "glapi/glapi.h"
-#include "glapi/glapitable.h"
+#include "glapitable.h"
 
 struct name_offset {
    const char *name;
@@ -1260,7 +1260,6 @@ const struct name_offset known_dispatch[] = {
    { "glTextureStorage1DEXT", _O(TextureStorage1DEXT) },
    { "glTextureStorage2DEXT", _O(TextureStorage2DEXT) },
    { "glTextureStorage3DEXT", _O(TextureStorage3DEXT) },
-   { "glPolygonOffsetEXT", _O(PolygonOffsetEXT) },
    { "glSampleMaskSGIS", _O(SampleMaskSGIS) },
    { "glSamplePatternSGIS", _O(SamplePatternSGIS) },
    { "glColorPointerEXT", _O(ColorPointerEXT) },
diff --git a/lib/mesa/src/mapi/meson.build b/lib/mesa/src/mapi/meson.build
new file mode 100644
index 000000000..798586bfb
--- /dev/null
+++ b/lib/mesa/src/mapi/meson.build
@@ -0,0 +1,39 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+files_mapi_util = files(
+  'u_current.c',
+  'u_current.h',
+  'u_execmem.c',
+  'u_execmem.h',
+)
+
+mapi_abi_py = files('mapi_abi.py')
+
+subdir('glapi')
+if with_shared_glapi
+  subdir('shared-glapi')
+endif
+if with_gles1
+  subdir('es1api')
+endif
+if with_gles2
+  subdir('es2api')
+endif
diff --git a/lib/mesa/src/mapi/shared-glapi/meson.build b/lib/mesa/src/mapi/shared-glapi/meson.build
new file mode 100644
index 000000000..3f041471f
--- /dev/null
+++ b/lib/mesa/src/mapi/shared-glapi/meson.build
@@ -0,0 +1,65 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+files_mapi_glapi = files(
+  '../entry.c',
+  '../mapi_glapi.c',
+  '../stub.c',
+  '../stub.h',
+  '../table.c',
+  '../table.h',
+)
+
+shared_glapi_mapi_tmp_h = custom_target(
+  'shared_glapi_mapi_tmp.h',
+  input : [mapi_abi_py, gl_and_es_api_files],
+  output : 'glapi_mapi_tmp.h',
+  command : [prog_python, '@INPUT0@', '--printer', 'shared-glapi', '@INPUT1@'],
+  depend_files : api_xml_files,
+  capture : true,
+)
+
+libglapi = shared_library(
+  'glapi',
+  [files_mapi_glapi, files_mapi_util, shared_glapi_mapi_tmp_h],
+  c_args : [
+    c_msvc_compat_args, c_vis_args, '-DMAPI_MODE_GLAPI',
+    '-DMAPI_ABI_HEADER="@0@"'.format(shared_glapi_mapi_tmp_h.full_path()),
+  ],
+  link_args : [ld_args_gc_sections],
+  include_directories : [inc_src, inc_include, inc_mapi],
+  dependencies : [dep_thread, dep_selinux],
+  version : '0.0.0',
+  install : true,
+)
+
+if with_any_opengl and with_tests
+  test(
+    'shared-glapi-test',
+    executable(
+      ['shared-glapi-test', glapitable_h],
+      'tests/check_table.cpp',
+      cpp_args : [cpp_msvc_compat_args],
+      include_directories : [inc_src, inc_include, inc_mapi],
+      link_with : [libglapi],
+      dependencies : [dep_thread, idep_gtest],
+    )
+  )
+endif
diff --git a/lib/mesa/src/mesa/drivers/common/meta_generate_mipmap.c b/lib/mesa/src/mesa/drivers/common/meta_generate_mipmap.c
index 55093e955..99d093169 100644
--- a/lib/mesa/src/mesa/drivers/common/meta_generate_mipmap.c
+++ b/lib/mesa/src/mesa/drivers/common/meta_generate_mipmap.c
@@ -201,10 +201,10 @@ _mesa_meta_GenerateMipmap(struct gl_context *ctx, GLenum target,
 
    /* We may have been called from glGenerateTextureMipmap with CurrentUnit
     * still set to 0, so we don't know when we can skip binding the texture.
-    * Assume that _mesa_BindTexture will be fast if we're rebinding the same
+    * Assume that _mesa_bind_texture will be fast if we're rebinding the same
     * texture.
     */
-   _mesa_BindTexture(target, texObj->Name);
+   _mesa_bind_texture(ctx, target, texObj);
 
    if (mipmap->samp_obj == NULL) {
       mipmap->samp_obj =  ctx->Driver.NewSamplerObject(ctx, 0xDEADBEEF);
diff --git a/lib/mesa/src/mesa/drivers/dri/common/meson.build b/lib/mesa/src/mesa/drivers/dri/common/meson.build
new file mode 100644
index 000000000..d9a9abdfb
--- /dev/null
+++ b/lib/mesa/src/mesa/drivers/dri/common/meson.build
@@ -0,0 +1,39 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+inc_dri_common = include_directories('.')
+
+libdricommon = static_library(
+  'dricommon',
+  ['utils.c', 'utils.h', 'dri_util.c', 'dri_util.h', xmlpool_options_h],
+  include_directories : [inc_common, inc_util],
+  c_args : c_vis_args,
+  dependencies : dep_libdrm,
+  build_by_default : false,
+)
+
+libmegadriver_stub = static_library(
+  'megadriver_stub',
+  'megadriver_stub.c',
+  include_directories : inc_common,
+  c_args : c_vis_args,
+  dependencies : dep_libdrm,
+  build_by_default : false,
+)
diff --git a/lib/mesa/src/mesa/drivers/dri/common/utils.c b/lib/mesa/src/mesa/drivers/dri/common/utils.c
index e944754a4..5a66bcf8e 100644
--- a/lib/mesa/src/mesa/drivers/dri/common/utils.c
+++ b/lib/mesa/src/mesa/drivers/dri/common/utils.c
@@ -41,6 +41,20 @@
 #include "utils.h"
 #include "dri_util.h"
 
+/* WARNING: HACK: Local defines to avoid pulling glx.h.
+ *
+ * Any parts of this file that use the following defines are either partial or
+ * entirely broken wrt EGL.
+ *
+ * For example any getConfigAttrib() or indexConfigAttrib() query from EGL for
+ * SLOW or NON_CONFORMANT_CONFIG will not work as expected since the EGL tokens
+ * are different from the GLX ones.
+ */
+#define GLX_NONE                                                0x8000
+#define GLX_SLOW_CONFIG                                         0x8001
+#define GLX_NON_CONFORMANT_CONFIG                               0x800D
+#define GLX_DONT_CARE                                           0xFFFFFFFF
+
 /**
  * Create the \c GL_RENDERER string for DRI drivers.
  * 
@@ -134,11 +148,12 @@ driGetRendererString( char * buffer, const char * hardware_name,
  * \param num_depth_stencil_bits  Number of entries in both \c depth_bits and
  *                      \c stencil_bits.
  * \param db_modes      Array of buffer swap modes.  If an element has a
- *                      value of \c GLX_NONE, then it represents a
- *                      single-buffered mode.  Other valid values are
- *                      \c GLX_SWAP_EXCHANGE_OML, \c GLX_SWAP_COPY_OML, and
- *                      \c GLX_SWAP_UNDEFINED_OML.  See the
- *                      GLX_OML_swap_method extension spec for more details.
+ *                      value of \c __DRI_ATTRIB_SWAP_NONE, then it
+ *                      represents a single-buffered mode.  Other valid
+ *                      values are \c __DRI_ATTRIB_SWAP_EXCHANGE,
+ *                      \c __DRI_ATTRIB_SWAP_COPY, and \c __DRI_ATTRIB_SWAP_UNDEFINED.
+ *                      They represent the respective GLX values as in
+ *                      the GLX_OML_swap_method extension spec.
  * \param num_db_modes  Number of entries in \c db_modes.
  * \param msaa_samples  Array of msaa sample count. 0 represents a visual
  *                      without a multisample buffer.
@@ -147,7 +162,10 @@ driGetRendererString( char * buffer, const char * hardware_name,
  * \param color_depth_match Whether the color depth must match the zs depth
  *                          This forces 32-bit color to have 24-bit depth, and
  *                          16-bit color to have 16-bit depth.
- * 
+ * \param mutable_render_buffer Enable __DRI_ATTRIB_MUTABLE_RENDER_BUFFER,
+ *                              which translates to
+ *                              EGL_MUTABLE_RENDER_BUFFER_BIT_KHR.
+ *
  * \returns
  * Pointer to any array of pointers to the \c __DRIconfig structures created
  * for the specified formats.  If there is an error, \c NULL is returned.
@@ -160,7 +178,8 @@ driCreateConfigs(mesa_format format,
 		 unsigned num_depth_stencil_bits,
 		 const GLenum * db_modes, unsigned num_db_modes,
 		 const uint8_t * msaa_samples, unsigned num_msaa_modes,
-		 GLboolean enable_accum, GLboolean color_depth_match)
+		 GLboolean enable_accum, GLboolean color_depth_match,
+		 GLboolean mutable_render_buffer)
 {
    static const uint32_t masks_table[][4] = {
       /* MESA_FORMAT_B5G6R5_UNORM */
@@ -177,6 +196,10 @@ driCreateConfigs(mesa_format format,
       { 0x000000FF, 0x0000FF00, 0x00FF0000, 0xFF000000 },
       /* MESA_FORMAT_R8G8B8X8_UNORM */
       { 0x000000FF, 0x0000FF00, 0x00FF0000, 0x00000000 },
+      /* MESA_FORMAT_R10G10B10X2_UNORM */
+      { 0x000003FF, 0x000FFC00, 0x3FF00000, 0x00000000 },
+      /* MESA_FORMAT_R10G10B10A2_UNORM */
+      { 0x000003FF, 0x000FFC00, 0x3FF00000, 0xC0000000 },
    };
 
    const uint32_t * masks;
@@ -204,6 +227,7 @@ driCreateConfigs(mesa_format format,
       masks = masks_table[2];
       break;
    case MESA_FORMAT_R8G8B8A8_UNORM:
+   case MESA_FORMAT_R8G8B8A8_SRGB:
       masks = masks_table[5];
       break;
    case MESA_FORMAT_R8G8B8X8_UNORM:
@@ -215,6 +239,12 @@ driCreateConfigs(mesa_format format,
    case MESA_FORMAT_B10G10R10A2_UNORM:
       masks = masks_table[4];
       break;
+   case MESA_FORMAT_R10G10B10X2_UNORM:
+      masks = masks_table[7];
+      break;
+   case MESA_FORMAT_R10G10B10A2_UNORM:
+      masks = masks_table[8];
+      break;
    default:
       fprintf(stderr, "[%s:%u] Unknown framebuffer type %s (%d).\n",
               __func__, __LINE__,
@@ -314,6 +344,7 @@ driCreateConfigs(mesa_format format,
 
 		    modes->yInverted = GL_TRUE;
 		    modes->sRGBCapable = is_srgb;
+		    modes->mutableRenderBuffer = mutable_render_buffer;
 		}
 	    }
 	}
@@ -398,6 +429,7 @@ static const struct { unsigned int attrib, offset; } attribMap[] = {
     __ATTRIB(__DRI_ATTRIB_BIND_TO_TEXTURE_TARGETS,	bindToTextureTargets),
     __ATTRIB(__DRI_ATTRIB_YINVERTED,			yInverted),
     __ATTRIB(__DRI_ATTRIB_FRAMEBUFFER_SRGB_CAPABLE,	sRGBCapable),
+    __ATTRIB(__DRI_ATTRIB_MUTABLE_RENDER_BUFFER,	mutableRenderBuffer),
 
     /* The struct field doesn't matter here, these are handled by the
      * switch in driGetConfigAttribIndex.  We need them in the array
diff --git a/lib/mesa/src/mesa/drivers/dri/common/utils.h b/lib/mesa/src/mesa/drivers/dri/common/utils.h
index 7be0465c2..7c9719f9f 100644
--- a/lib/mesa/src/mesa/drivers/dri/common/utils.h
+++ b/lib/mesa/src/mesa/drivers/dri/common/utils.h
@@ -45,7 +45,8 @@ driCreateConfigs(mesa_format format,
 		 unsigned num_depth_stencil_bits,
 		 const GLenum * db_modes, unsigned num_db_modes,
 		 const uint8_t * msaa_samples, unsigned num_msaa_modes,
-		 GLboolean enable_accum, GLboolean color_depth_match);
+		 GLboolean enable_accum, GLboolean color_depth_match,
+		 GLboolean mutable_render_buffer);
 
 __DRIconfig **driConcatConfigs(__DRIconfig **a,
 			       __DRIconfig **b);
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/i830_context.c b/lib/mesa/src/mesa/drivers/dri/i915/i830_context.c
index 299e54d50..ca0048306 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/i830_context.c
+++ b/lib/mesa/src/mesa/drivers/dri/i915/i830_context.c
@@ -27,6 +27,7 @@
 
 #include "i830_context.h"
 #include "main/api_exec.h"
+#include "main/extensions.h"
 #include "main/imports.h"
 #include "main/version.h"
 #include "main/vtxfmt.h"
@@ -124,6 +125,7 @@ i830CreateContext(int api,
    _tnl_allow_vertex_fog(ctx, 1);
    _tnl_allow_pixel_fog(ctx, 0);
 
+   _mesa_override_extensions(ctx);
    _mesa_compute_version(ctx);
 
    _mesa_initialize_dispatch_tables(ctx);
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/i830_state.c b/lib/mesa/src/mesa/drivers/dri/i915/i830_state.c
index 7ce5ef7e4..c8f48fbfa 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/i830_state.c
+++ b/lib/mesa/src/mesa/drivers/dri/i915/i830_state.c
@@ -573,16 +573,16 @@ i830Scissor(struct gl_context * ctx)
 }
 
 static void
-i830LogicOp(struct gl_context * ctx, GLenum opcode)
+i830LogicOp(struct gl_context * ctx, enum gl_logicop_mode opcode)
 {
    struct i830_context *i830 = i830_context(ctx);
-   int tmp = intel_translate_logic_op(opcode);
 
    DBG("%s\n", __func__);
-   
+
+   assert((unsigned)opcode <= 15);
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE4] &= ~LOGICOP_MASK;
-   i830->state.Ctx[I830_CTXREG_STATE4] |= LOGIC_OP_FUNC(tmp);
+   i830->state.Ctx[I830_CTXREG_STATE4] |= opcode;
 }
 
 
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/intel_blit.c b/lib/mesa/src/mesa/drivers/dri/i915/intel_blit.c
index 279db28b7..98cffaf24 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/intel_blit.c
+++ b/lib/mesa/src/mesa/drivers/dri/i915/intel_blit.c
@@ -48,27 +48,9 @@ intel_miptree_set_alpha_to_one(struct intel_context *intel,
                                struct intel_mipmap_tree *mt,
                                int x, int y, int width, int height);
 
-static GLuint translate_raster_op(GLenum logicop)
+static GLuint translate_raster_op(enum gl_logicop_mode logicop)
 {
-   switch(logicop) {
-   case GL_CLEAR: return 0x00;
-   case GL_AND: return 0x88;
-   case GL_AND_REVERSE: return 0x44;
-   case GL_COPY: return 0xCC;
-   case GL_AND_INVERTED: return 0x22;
-   case GL_NOOP: return 0xAA;
-   case GL_XOR: return 0x66;
-   case GL_OR: return 0xEE;
-   case GL_NOR: return 0x11;
-   case GL_EQUIV: return 0x99;
-   case GL_INVERT: return 0x55;
-   case GL_OR_REVERSE: return 0xDD;
-   case GL_COPY_INVERTED: return 0x33;
-   case GL_OR_INVERTED: return 0xBB;
-   case GL_NAND: return 0x77;
-   case GL_SET: return 0xFF;
-   default: return 0;
-   }
+   return logicop | (logicop << 4);
 }
 
 static uint32_t
@@ -90,144 +72,23 @@ br13_for_cpp(int cpp)
    }
 }
 
-/**
- * Implements a rectangular block transfer (blit) of pixels between two
- * miptrees.
- *
- * Our blitter can operate on 1, 2, or 4-byte-per-pixel data, with generous,
- * but limited, pitches and sizes allowed.
- *
- * The src/dst coordinates are relative to the given level/slice of the
- * miptree.
- *
- * If @src_flip or @dst_flip is set, then the rectangle within that miptree
- * will be inverted (including scanline order) when copying.  This is common
- * in GL when copying between window system and user-created
- * renderbuffers/textures.
- */
-bool
-intel_miptree_blit(struct intel_context *intel,
-                   struct intel_mipmap_tree *src_mt,
-                   int src_level, int src_slice,
-                   uint32_t src_x, uint32_t src_y, bool src_flip,
-                   struct intel_mipmap_tree *dst_mt,
-                   int dst_level, int dst_slice,
-                   uint32_t dst_x, uint32_t dst_y, bool dst_flip,
-                   uint32_t width, uint32_t height,
-                   GLenum logicop)
-{
-   /* No sRGB decode or encode is done by the hardware blitter, which is
-    * consistent with what we want in the callers (glCopyTexSubImage(),
-    * glBlitFramebuffer(), texture validation, etc.).
-    */
-   mesa_format src_format = _mesa_get_srgb_format_linear(src_mt->format);
-   mesa_format dst_format = _mesa_get_srgb_format_linear(dst_mt->format);
-
-   /* The blitter doesn't support doing any format conversions.  We do also
-    * support blitting ARGB8888 to XRGB8888 (trivial, the values dropped into
-    * the X channel don't matter), and XRGB8888 to ARGB8888 by setting the A
-    * channel to 1.0 at the end.
-    */
-   if (src_format != dst_format &&
-      ((src_format != MESA_FORMAT_B8G8R8A8_UNORM &&
-        src_format != MESA_FORMAT_B8G8R8X8_UNORM) ||
-       (dst_format != MESA_FORMAT_B8G8R8A8_UNORM &&
-        dst_format != MESA_FORMAT_B8G8R8X8_UNORM))) {
-      perf_debug("%s: Can't use hardware blitter from %s to %s, "
-                 "falling back.\n", __func__,
-                 _mesa_get_format_name(src_format),
-                 _mesa_get_format_name(dst_format));
-      return false;
-   }
-
-   /* According to the Ivy Bridge PRM, Vol1 Part4, section 1.2.1.2 (Graphics
-    * Data Size Limitations):
-    *
-    *    The BLT engine is capable of transferring very large quantities of
-    *    graphics data. Any graphics data read from and written to the
-    *    destination is permitted to represent a number of pixels that
-    *    occupies up to 65,536 scan lines and up to 32,768 bytes per scan line
-    *    at the destination. The maximum number of pixels that may be
-    *    represented per scan line’s worth of graphics data depends on the
-    *    color depth.
-    *
-    * Furthermore, intelEmitCopyBlit (which is called below) uses a signed
-    * 16-bit integer to represent buffer pitch, so it can only handle buffer
-    * pitches < 32k.
-    *
-    * As a result of these two limitations, we can only use the blitter to do
-    * this copy when the region's pitch is less than 32k.
-    */
-   if (src_mt->region->pitch > 32768 ||
-       dst_mt->region->pitch > 32768) {
-      perf_debug("Falling back due to >32k pitch\n");
-      return false;
-   }
-
-   if (src_flip)
-      src_y = src_mt->level[src_level].height - src_y - height;
-
-   if (dst_flip)
-      dst_y = dst_mt->level[dst_level].height - dst_y - height;
-
-   int src_pitch = src_mt->region->pitch;
-   if (src_flip != dst_flip)
-      src_pitch = -src_pitch;
-
-   uint32_t src_image_x, src_image_y;
-   intel_miptree_get_image_offset(src_mt, src_level, src_slice,
-                                  &src_image_x, &src_image_y);
-   src_x += src_image_x;
-   src_y += src_image_y;
-
-   uint32_t dst_image_x, dst_image_y;
-   intel_miptree_get_image_offset(dst_mt, dst_level, dst_slice,
-                                  &dst_image_x, &dst_image_y);
-   dst_x += dst_image_x;
-   dst_y += dst_image_y;
-
-   if (!intelEmitCopyBlit(intel,
-                          src_mt->cpp,
-                          src_pitch,
-                          src_mt->region->bo, src_mt->offset,
-                          src_mt->region->tiling,
-                          dst_mt->region->pitch,
-                          dst_mt->region->bo, dst_mt->offset,
-                          dst_mt->region->tiling,
-                          src_x, src_y,
-                          dst_x, dst_y,
-                          width, height,
-                          logicop)) {
-      return false;
-   }
-
-   if (src_mt->format == MESA_FORMAT_B8G8R8X8_UNORM &&
-       dst_mt->format == MESA_FORMAT_B8G8R8A8_UNORM) {
-      intel_miptree_set_alpha_to_one(intel, dst_mt,
-                                     dst_x, dst_y,
-                                     width, height);
-   }
-
-   return true;
-}
-
 /* Copy BitBlt
  */
-bool
-intelEmitCopyBlit(struct intel_context *intel,
-		  GLuint cpp,
-		  GLshort src_pitch,
-		  drm_intel_bo *src_buffer,
-		  GLuint src_offset,
-		  uint32_t src_tiling,
-		  GLshort dst_pitch,
-		  drm_intel_bo *dst_buffer,
-		  GLuint dst_offset,
-		  uint32_t dst_tiling,
-		  GLshort src_x, GLshort src_y,
-		  GLshort dst_x, GLshort dst_y,
-		  GLshort w, GLshort h,
-		  GLenum logic_op)
+static bool
+emit_copy_blit(struct intel_context *intel,
+               GLuint cpp,
+               GLshort src_pitch,
+               drm_intel_bo *src_buffer,
+               GLuint src_offset,
+               uint32_t src_tiling,
+               GLshort dst_pitch,
+               drm_intel_bo *dst_buffer,
+               GLuint dst_offset,
+               uint32_t dst_tiling,
+               GLshort src_x, GLshort src_y,
+               GLshort dst_x, GLshort dst_y,
+               GLshort w, GLshort h,
+               enum gl_logicop_mode logic_op)
 {
    GLuint CMD, BR13, pass = 0;
    int dst_y2 = dst_y + h;
@@ -338,6 +199,126 @@ intelEmitCopyBlit(struct intel_context *intel,
    return true;
 }
 
+/**
+ * Implements a rectangular block transfer (blit) of pixels between two
+ * miptrees.
+ *
+ * Our blitter can operate on 1, 2, or 4-byte-per-pixel data, with generous,
+ * but limited, pitches and sizes allowed.
+ *
+ * The src/dst coordinates are relative to the given level/slice of the
+ * miptree.
+ *
+ * If @src_flip or @dst_flip is set, then the rectangle within that miptree
+ * will be inverted (including scanline order) when copying.  This is common
+ * in GL when copying between window system and user-created
+ * renderbuffers/textures.
+ */
+bool
+intel_miptree_blit(struct intel_context *intel,
+                   struct intel_mipmap_tree *src_mt,
+                   int src_level, int src_slice,
+                   uint32_t src_x, uint32_t src_y, bool src_flip,
+                   struct intel_mipmap_tree *dst_mt,
+                   int dst_level, int dst_slice,
+                   uint32_t dst_x, uint32_t dst_y, bool dst_flip,
+                   uint32_t width, uint32_t height,
+                   enum gl_logicop_mode logicop)
+{
+   /* No sRGB decode or encode is done by the hardware blitter, which is
+    * consistent with what we want in the callers (glCopyTexSubImage(),
+    * glBlitFramebuffer(), texture validation, etc.).
+    */
+   mesa_format src_format = _mesa_get_srgb_format_linear(src_mt->format);
+   mesa_format dst_format = _mesa_get_srgb_format_linear(dst_mt->format);
+
+   /* The blitter doesn't support doing any format conversions.  We do also
+    * support blitting ARGB8888 to XRGB8888 (trivial, the values dropped into
+    * the X channel don't matter), and XRGB8888 to ARGB8888 by setting the A
+    * channel to 1.0 at the end.
+    */
+   if (src_format != dst_format &&
+      ((src_format != MESA_FORMAT_B8G8R8A8_UNORM &&
+        src_format != MESA_FORMAT_B8G8R8X8_UNORM) ||
+       (dst_format != MESA_FORMAT_B8G8R8A8_UNORM &&
+        dst_format != MESA_FORMAT_B8G8R8X8_UNORM))) {
+      perf_debug("%s: Can't use hardware blitter from %s to %s, "
+                 "falling back.\n", __func__,
+                 _mesa_get_format_name(src_format),
+                 _mesa_get_format_name(dst_format));
+      return false;
+   }
+
+   /* According to the Ivy Bridge PRM, Vol1 Part4, section 1.2.1.2 (Graphics
+    * Data Size Limitations):
+    *
+    *    The BLT engine is capable of transferring very large quantities of
+    *    graphics data. Any graphics data read from and written to the
+    *    destination is permitted to represent a number of pixels that
+    *    occupies up to 65,536 scan lines and up to 32,768 bytes per scan line
+    *    at the destination. The maximum number of pixels that may be
+    *    represented per scan line’s worth of graphics data depends on the
+    *    color depth.
+    *
+    * Furthermore, emit_copy_blit (which is called below) uses a signed
+    * 16-bit integer to represent buffer pitch, so it can only handle buffer
+    * pitches < 32k.
+    *
+    * As a result of these two limitations, we can only use the blitter to do
+    * this copy when the region's pitch is less than 32k.
+    */
+   if (src_mt->region->pitch > 32768 ||
+       dst_mt->region->pitch > 32768) {
+      perf_debug("Falling back due to >32k pitch\n");
+      return false;
+   }
+
+   if (src_flip)
+      src_y = src_mt->level[src_level].height - src_y - height;
+
+   if (dst_flip)
+      dst_y = dst_mt->level[dst_level].height - dst_y - height;
+
+   int src_pitch = src_mt->region->pitch;
+   if (src_flip != dst_flip)
+      src_pitch = -src_pitch;
+
+   uint32_t src_image_x, src_image_y;
+   intel_miptree_get_image_offset(src_mt, src_level, src_slice,
+                                  &src_image_x, &src_image_y);
+   src_x += src_image_x;
+   src_y += src_image_y;
+
+   uint32_t dst_image_x, dst_image_y;
+   intel_miptree_get_image_offset(dst_mt, dst_level, dst_slice,
+                                  &dst_image_x, &dst_image_y);
+   dst_x += dst_image_x;
+   dst_y += dst_image_y;
+
+   if (!emit_copy_blit(intel,
+                       src_mt->cpp,
+                       src_pitch,
+                       src_mt->region->bo, src_mt->offset,
+                       src_mt->region->tiling,
+                       dst_mt->region->pitch,
+                       dst_mt->region->bo, dst_mt->offset,
+                       dst_mt->region->tiling,
+                       src_x, src_y,
+                       dst_x, dst_y,
+                       width, height,
+                       logicop)) {
+      return false;
+   }
+
+   if (src_mt->format == MESA_FORMAT_B8G8R8X8_UNORM &&
+       dst_mt->format == MESA_FORMAT_B8G8R8A8_UNORM) {
+      intel_miptree_set_alpha_to_one(intel, dst_mt,
+                                     dst_x, dst_y,
+                                     width, height);
+   }
+
+   return true;
+}
 
 /**
  * Use blitting to clear the renderbuffers named by 'flags'.
@@ -523,7 +504,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
 				  uint32_t dst_tiling,
 				  GLshort x, GLshort y,
 				  GLshort w, GLshort h,
-				  GLenum logic_op)
+				  enum gl_logicop_mode logic_op)
 {
    int dwords = ALIGN(src_size, 8) / 4;
    uint32_t opcode, br13, blit_cmd;
@@ -535,7 +516,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
 	 return false;
    }
 
-   assert((logic_op >= GL_CLEAR) && (logic_op <= (GL_CLEAR + 0x0f)));
+   assert((unsigned)logic_op <= 0x0f);
    assert(dst_pitch > 0);
 
    if (w < 0 || h < 0)
@@ -607,13 +588,13 @@ intel_emit_linear_blit(struct intel_context *intel,
     */
    pitch = ROUND_DOWN_TO(MIN2(size, (1 << 15) - 1), 4);
    height = (pitch == 0) ? 1 : size / pitch;
-   ok = intelEmitCopyBlit(intel, 1,
-			  pitch, src_bo, src_offset, I915_TILING_NONE,
-			  pitch, dst_bo, dst_offset, I915_TILING_NONE,
-			  0, 0, /* src x/y */
-			  0, 0, /* dst x/y */
-			  pitch, height, /* w, h */
-			  GL_COPY);
+   ok = emit_copy_blit(intel, 1,
+                       pitch, src_bo, src_offset, I915_TILING_NONE,
+                       pitch, dst_bo, dst_offset, I915_TILING_NONE,
+                       0, 0, /* src x/y */
+                       0, 0, /* dst x/y */
+                       pitch, height, /* w, h */
+                       COLOR_LOGICOP_COPY);
    if (!ok)
       _mesa_problem(ctx, "Failed to linear blit %dx%d\n", pitch, height);
 
@@ -623,13 +604,13 @@ intel_emit_linear_blit(struct intel_context *intel,
    assert (size < (1 << 15));
    pitch = ALIGN(size, 4);
    if (size != 0) {
-      ok = intelEmitCopyBlit(intel, 1,
-			     pitch, src_bo, src_offset, I915_TILING_NONE,
-			     pitch, dst_bo, dst_offset, I915_TILING_NONE,
-			     0, 0, /* src x/y */
-			     0, 0, /* dst x/y */
-			     size, 1, /* w, h */
-			     GL_COPY);
+      ok = emit_copy_blit(intel, 1,
+                          pitch, src_bo, src_offset, I915_TILING_NONE,
+                          pitch, dst_bo, dst_offset, I915_TILING_NONE,
+                          0, 0, /* src x/y */
+                          0, 0, /* dst x/y */
+                          size, 1, /* w, h */
+                          COLOR_LOGICOP_COPY);
       if (!ok)
          _mesa_problem(ctx, "Failed to linear blit %dx%d\n", size, 1);
    }
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/intel_blit.h b/lib/mesa/src/mesa/drivers/dri/i915/intel_blit.h
index 1e76126aa..61de411b3 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/intel_blit.h
+++ b/lib/mesa/src/mesa/drivers/dri/i915/intel_blit.h
@@ -35,22 +35,6 @@ extern void intelCopyBuffer(const __DRIdrawable * dpriv,
 
 extern GLbitfield intelClearWithBlit(struct gl_context * ctx, GLbitfield mask);
 
-bool
-intelEmitCopyBlit(struct intel_context *intel,
-                              GLuint cpp,
-                              GLshort src_pitch,
-                              drm_intel_bo *src_buffer,
-                              GLuint src_offset,
-			      uint32_t src_tiling,
-                              GLshort dst_pitch,
-                              drm_intel_bo *dst_buffer,
-                              GLuint dst_offset,
-			      uint32_t dst_tiling,
-                              GLshort srcx, GLshort srcy,
-                              GLshort dstx, GLshort dsty,
-                              GLshort w, GLshort h,
-			      GLenum logicop );
-
 bool intel_miptree_blit(struct intel_context *intel,
                         struct intel_mipmap_tree *src_mt,
                         int src_level, int src_slice,
@@ -59,7 +43,7 @@ bool intel_miptree_blit(struct intel_context *intel,
                         int dst_level, int dst_slice,
                         uint32_t dst_x, uint32_t dst_y, bool dst_flip,
                         uint32_t width, uint32_t height,
-                        GLenum logicop);
+                        enum gl_logicop_mode logicop);
 
 bool
 intelEmitImmediateColorExpandBlit(struct intel_context *intel,
@@ -72,7 +56,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
 				  uint32_t dst_tiling,
 				  GLshort x, GLshort y,
 				  GLshort w, GLshort h,
-				  GLenum logic_op);
+                                  enum gl_logicop_mode logic_op);
 void intel_emit_linear_blit(struct intel_context *intel,
 			    drm_intel_bo *dst_bo,
 			    unsigned int dst_offset,
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/intel_buffers.c b/lib/mesa/src/mesa/drivers/dri/i915/intel_buffers.c
index 386e03244..83d59edb6 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/intel_buffers.c
+++ b/lib/mesa/src/mesa/drivers/dri/i915/intel_buffers.c
@@ -53,7 +53,7 @@ intel_check_front_buffer_rendering(struct intel_context *intel)
 }
 
 static void
-intelDrawBuffer(struct gl_context * ctx, GLenum mode)
+intelDrawBuffer(struct gl_context * ctx)
 {
    if (_mesa_is_front_buffer_drawing(ctx->DrawBuffer)) {
       struct intel_context *const intel = intel_context(ctx);
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/intel_fbo.c b/lib/mesa/src/mesa/drivers/dri/i915/intel_fbo.c
index 8750c601b..78e2c1e66 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/intel_fbo.c
+++ b/lib/mesa/src/mesa/drivers/dri/i915/intel_fbo.c
@@ -86,7 +86,8 @@ intel_map_renderbuffer(struct gl_context *ctx,
 		       GLuint x, GLuint y, GLuint w, GLuint h,
 		       GLbitfield mode,
 		       GLubyte **out_map,
-		       GLint *out_stride)
+		       GLint *out_stride,
+		       bool flip_y)
 {
    struct intel_context *intel = intel_context(ctx);
    struct swrast_renderbuffer *srb = (struct swrast_renderbuffer *)rb;
@@ -94,6 +95,9 @@ intel_map_renderbuffer(struct gl_context *ctx,
    void *map;
    int stride;
 
+   /* driver does not support GL_FRAMEBUFFER_FLIP_Y_MESA */
+   assert((rb->Name == 0) == flip_y);
+
    if (srb->Buffer) {
       /* this is a malloc'd renderbuffer (accum buffer), not an irb */
       GLint bpp = _mesa_get_format_bytes(rb->Format);
@@ -287,7 +291,7 @@ intel_image_target_renderbuffer_storage(struct gl_context *ctx,
  * intel_process_dri2_buffer().
  */
 static GLboolean
-intel_alloc_window_storage(struct gl_context * ctx, struct gl_renderbuffer *rb,
+intel_alloc_window_storage(UNUSED struct gl_context *ctx, struct gl_renderbuffer *rb,
                            GLenum internalFormat, GLuint width, GLuint height)
 {
    assert(rb->Name == 0);
@@ -300,8 +304,10 @@ intel_alloc_window_storage(struct gl_context * ctx, struct gl_renderbuffer *rb,
 
 /** Dummy function for gl_renderbuffer::AllocStorage() */
 static GLboolean
-intel_nop_alloc_storage(struct gl_context * ctx, struct gl_renderbuffer *rb,
-                        GLenum internalFormat, GLuint width, GLuint height)
+intel_nop_alloc_storage(UNUSED struct gl_context *ctx,
+                        UNUSED struct gl_renderbuffer *rb,
+                        UNUSED GLenum internalFormat,
+                        UNUSED GLuint width, UNUSED GLuint height)
 {
    _mesa_problem(ctx, "intel_op_alloc_storage should never be called.");
    return false;
@@ -393,7 +399,8 @@ intel_new_renderbuffer(struct gl_context * ctx, GLuint name)
  */
 static void
 intel_bind_framebuffer(struct gl_context * ctx, GLenum target,
-                       struct gl_framebuffer *fb, struct gl_framebuffer *fbread)
+                       UNUSED struct gl_framebuffer *fb,
+                       UNUSED struct gl_framebuffer *fbread)
 {
    if (target == GL_FRAMEBUFFER_EXT || target == GL_DRAW_FRAMEBUFFER_EXT) {
       intel_draw_buffer(ctx);
@@ -419,8 +426,7 @@ intel_framebuffer_renderbuffer(struct gl_context * ctx,
 }
 
 static bool
-intel_renderbuffer_update_wrapper(struct intel_context *intel,
-                                  struct intel_renderbuffer *irb,
+intel_renderbuffer_update_wrapper(struct intel_renderbuffer *irb,
 				  struct gl_texture_image *image,
                                   uint32_t layer)
 {
@@ -468,7 +474,6 @@ intel_render_texture(struct gl_context * ctx,
                      struct gl_framebuffer *fb,
                      struct gl_renderbuffer_attachment *att)
 {
-   struct intel_context *intel = intel_context(ctx);
    struct gl_renderbuffer *rb = att->Renderbuffer;
    struct intel_renderbuffer *irb = intel_renderbuffer(rb);
    struct gl_texture_image *image = rb->TexImage;
@@ -495,7 +500,7 @@ intel_render_texture(struct gl_context * ctx,
 
    intel_miptree_check_level_layer(mt, att->TextureLevel, layer);
 
-   if (!intel_renderbuffer_update_wrapper(intel, irb, image, layer)) {
+   if (!intel_renderbuffer_update_wrapper(irb, image, layer)) {
        _swrast_render_texture(ctx, fb, att);
        return;
    }
@@ -641,7 +646,7 @@ intel_blit_framebuffer_with_blitter(struct gl_context *ctx,
                                     GLint srcX1, GLint srcY1,
                                     GLint dstX0, GLint dstY0,
                                     GLint dstX1, GLint dstY1,
-                                    GLbitfield mask, GLenum filter)
+                                    GLbitfield mask)
 {
    struct intel_context *intel = intel_context(ctx);
 
@@ -714,7 +719,7 @@ intel_blit_framebuffer_with_blitter(struct gl_context *ctx,
                                  dst_irb->mt,
                                  dst_irb->mt_level, dst_irb->mt_layer,
                                  dstX0, dstY0, dst_rb->Name == 0,
-                                 dstX1 - dstX0, dstY1 - dstY0, GL_COPY)) {
+                                 dstX1 - dstX0, dstY1 - dstY0, COLOR_LOGICOP_COPY)) {
             perf_debug("glBlitFramebuffer(): unknown blit failure.  "
                        "Falling back to software rendering.\n");
             return mask;
@@ -739,7 +744,7 @@ intel_blit_framebuffer(struct gl_context *ctx,
    mask = intel_blit_framebuffer_with_blitter(ctx, readFb, drawFb,
                                               srcX0, srcY0, srcX1, srcY1,
                                               dstX0, dstY0, dstX1, dstY1,
-                                              mask, filter);
+                                              mask);
    if (mask == 0x0)
       return;
 
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/intel_pixel_bitmap.c b/lib/mesa/src/mesa/drivers/dri/i915/intel_pixel_bitmap.c
index a41b6929c..8dee0bea9 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/intel_pixel_bitmap.c
+++ b/lib/mesa/src/mesa/drivers/dri/i915/intel_pixel_bitmap.c
@@ -262,8 +262,8 @@ do_blit_bitmap( struct gl_context *ctx,
 	 int h = MIN2(DY, height - py);
 	 int w = MIN2(DX, width - px);
 	 GLuint sz = ALIGN(ALIGN(w,8) * h, 64)/8;
-	 GLenum logic_op = ctx->Color.ColorLogicOpEnabled ?
-	    ctx->Color.LogicOp : GL_COPY;
+         const enum gl_logicop_mode logic_op = ctx->Color.ColorLogicOpEnabled ?
+            ctx->Color._LogicOp : COLOR_LOGICOP_COPY;
 
 	 assert(sz <= sizeof(stipple));
 	 memset(stipple, 0, sz);
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/intel_pixel_read.c b/lib/mesa/src/mesa/drivers/dri/i915/intel_pixel_read.c
index e6fa8f2dc..3560963c8 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/intel_pixel_read.c
+++ b/lib/mesa/src/mesa/drivers/dri/i915/intel_pixel_read.c
@@ -141,7 +141,7 @@ do_blit_readpixels(struct gl_context * ctx,
                            x, y, _mesa_is_winsys_fbo(ctx->ReadBuffer),
                            pbo_mt, 0, 0,
                            0, 0, dst_flip,
-                           width, height, GL_COPY)) {
+                           width, height, COLOR_LOGICOP_COPY)) {
       intel_miptree_release(&pbo_mt);
       return false;
    }
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/intel_render.c b/lib/mesa/src/mesa/drivers/dri/i915/intel_render.c
index c1603565c..a39528262 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/intel_render.c
+++ b/lib/mesa/src/mesa/drivers/dri/i915/intel_render.c
@@ -62,6 +62,8 @@
 #define HAVE_TRI_FANS    1
 #define HAVE_POLYGONS    1
 
+#define HAVE_QUADS       0
+#define HAVE_QUAD_STRIPS 0
 #define HAVE_ELTS        0
 
 static const uint32_t hw_prim[GL_POLYGON + 1] = {
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/intel_screen.h b/lib/mesa/src/mesa/drivers/dri/i915/intel_screen.h
index a22888b39..ac70d3655 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/intel_screen.h
+++ b/lib/mesa/src/mesa/drivers/dri/i915/intel_screen.h
@@ -117,7 +117,6 @@ struct intel_screen
 #define intel_check_front_buffer_rendering  old_intel_check_front_buffer_rendering
 #define intelInitBufferFuncs                old_intelInitBufferFuncs
 #define intelClearWithBlit                  old_intelClearWithBlit
-#define intelEmitCopyBlit                   old_intelEmitCopyBlit
 #define intelEmitImmediateColorExpandBlit   old_intelEmitImmediateColorExpandBlit
 #define intel_emit_linear_blit              old_intel_emit_linear_blit
 #define intel_miptree_blit                  old_intel_miptree_blit
@@ -139,7 +138,6 @@ struct intel_screen
 #define get_time                            old_get_time
 #define intel_translate_blend_factor        old_intel_translate_blend_factor
 #define intel_translate_compare_func        old_intel_translate_compare_func
-#define intel_translate_logic_op            old_intel_translate_logic_op
 #define intel_translate_shadow_compare_func old_intel_translate_shadow_compare_func
 #define intel_translate_stencil_op          old_intel_translate_stencil_op
 #define intel_init_syncobj_functions        old_intel_init_syncobj_functions
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/intel_state.c b/lib/mesa/src/mesa/drivers/dri/i915/intel_state.c
index 3de9d50a4..4f47013dd 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/intel_state.c
+++ b/lib/mesa/src/mesa/drivers/dri/i915/intel_state.c
@@ -151,44 +151,3 @@ intel_translate_blend_factor(GLenum factor)
    fprintf(stderr, "Unknown value in %s: %x\n", __func__, factor);
    return BLENDFACT_ZERO;
 }
-
-int
-intel_translate_logic_op(GLenum opcode)
-{
-   switch (opcode) {
-   case GL_CLEAR:
-      return LOGICOP_CLEAR;
-   case GL_AND:
-      return LOGICOP_AND;
-   case GL_AND_REVERSE:
-      return LOGICOP_AND_RVRSE;
-   case GL_COPY:
-      return LOGICOP_COPY;
-   case GL_COPY_INVERTED:
-      return LOGICOP_COPY_INV;
-   case GL_AND_INVERTED:
-      return LOGICOP_AND_INV;
-   case GL_NOOP:
-      return LOGICOP_NOOP;
-   case GL_XOR:
-      return LOGICOP_XOR;
-   case GL_OR:
-      return LOGICOP_OR;
-   case GL_OR_INVERTED:
-      return LOGICOP_OR_INV;
-   case GL_NOR:
-      return LOGICOP_NOR;
-   case GL_EQUIV:
-      return LOGICOP_EQUIV;
-   case GL_INVERT:
-      return LOGICOP_INV;
-   case GL_OR_REVERSE:
-      return LOGICOP_OR_RVRSE;
-   case GL_NAND:
-      return LOGICOP_NAND;
-   case GL_SET:
-      return LOGICOP_SET;
-   default:
-      return LOGICOP_SET;
-   }
-}
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/intel_tex_copy.c b/lib/mesa/src/mesa/drivers/dri/i915/intel_tex_copy.c
index a5d00af3b..4b087b616 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/intel_tex_copy.c
+++ b/lib/mesa/src/mesa/drivers/dri/i915/intel_tex_copy.c
@@ -70,7 +70,7 @@ intel_copy_texsubimage(struct intel_context *intel,
                            intelImage->mt, intelImage->base.Base.Level,
                            intelImage->base.Base.Face + slice,
                            dstx, dsty, false,
-                           width, height, GL_COPY)) {
+                           width, height, COLOR_LOGICOP_COPY)) {
       return false;
    }
 
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/intel_tex_subimage.c b/lib/mesa/src/mesa/drivers/dri/i915/intel_tex_subimage.c
index 4083d696b..b372e114f 100644
--- a/lib/mesa/src/mesa/drivers/dri/i915/intel_tex_subimage.c
+++ b/lib/mesa/src/mesa/drivers/dri/i915/intel_tex_subimage.c
@@ -111,7 +111,7 @@ intel_blit_texsubimage(struct gl_context * ctx,
                             0, 0, false,
                             intelImage->mt, texImage->Level, texImage->Face,
                             xoffset, yoffset, false,
-                            width, height, GL_COPY);
+                            width, height, COLOR_LOGICOP_COPY);
    assert(ret);
 
    intel_miptree_release(&temp_mt);
diff --git a/lib/mesa/src/mesa/drivers/dri/i915/meson.build b/lib/mesa/src/mesa/drivers/dri/i915/meson.build
new file mode 100644
index 000000000..1971419a6
--- /dev/null
+++ b/lib/mesa/src/mesa/drivers/dri/i915/meson.build
@@ -0,0 +1,97 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+files_i915 = files(
+  'i830_context.c',
+  'i830_context.h',
+  'i830_reg.h',
+  'i830_state.c',
+  'i830_texblend.c',
+  'i830_texstate.c',
+  'i830_vtbl.c',
+  'i915_context.c',
+  'i915_context.h',
+  'i915_debug_fp.c',
+  'i915_debug.h',
+  'i915_fragprog.c',
+  'i915_program.c',
+  'i915_program.h',
+  'i915_reg.h',
+  'i915_state.c',
+  'i915_tex_layout.c',
+  'i915_texstate.c',
+  'i915_vtbl.c',
+  'intel_batchbuffer.c',
+  'intel_batchbuffer.h',
+  'intel_blit.c',
+  'intel_blit.h',
+  'intel_buffer_objects.c',
+  'intel_buffer_objects.h',
+  'intel_buffers.c',
+  'intel_buffers.h',
+  'intel_chipset.h',
+  'intel_clear.c',
+  'intel_clear.h',
+  'intel_context.c',
+  'intel_context.h',
+  'intel_extensions.c',
+  'intel_extensions.h',
+  'intel_fbo.c',
+  'intel_fbo.h',
+  'intel_mipmap_tree.c',
+  'intel_mipmap_tree.h',
+  'intel_pixel_bitmap.c',
+  'intel_pixel.c',
+  'intel_pixel_copy.c',
+  'intel_pixel_draw.c',
+  'intel_pixel.h',
+  'intel_pixel_read.c',
+  'intel_reg.h',
+  'intel_regions.c',
+  'intel_regions.h',
+  'intel_render.c',
+  'intel_screen.c',
+  'intel_screen.h',
+  'intel_state.c',
+  'intel_syncobj.c',
+  'intel_tex.c',
+  'intel_tex_copy.c',
+  'intel_tex.h',
+  'intel_tex_image.c',
+  'intel_tex_layout.c',
+  'intel_tex_layout.h',
+  'intel_tex_obj.h',
+  'intel_tex_subimage.c',
+  'intel_tex_validate.c',
+  'intel_tris.c',
+  'intel_tris.h',
+)
+
+libi915 = static_library(
+  'i915',
+  [files_i915, xmlpool_options_h],
+  include_directories : [inc_common, inc_dri_common, inc_util],
+  c_args : [c_vis_args, no_override_init_args],
+  cpp_args : [cpp_vis_args],
+  dependencies : [dep_libdrm, dep_libdrm_intel],
+)
+
+dri_drivers += libi915
+dri_link += 'i915_dri.so'
diff --git a/lib/mesa/src/mesa/drivers/dri/i965/brw_disk_cache.c b/lib/mesa/src/mesa/drivers/dri/i965/brw_disk_cache.c
new file mode 100644
index 000000000..65fcab24b
--- /dev/null
+++ b/lib/mesa/src/mesa/drivers/dri/i965/brw_disk_cache.c
@@ -0,0 +1,403 @@
+/*
+ * Copyright © 2014 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "compiler/blob.h"
+#include "compiler/glsl/ir_uniform.h"
+#include "compiler/glsl/shader_cache.h"
+#include "main/mtypes.h"
+#include "util/build_id.h"
+#include "util/debug.h"
+#include "util/disk_cache.h"
+#include "util/macros.h"
+#include "util/mesa-sha1.h"
+
+#include "compiler/brw_eu.h"
+#include "common/gen_debug.h"
+
+#include "brw_context.h"
+#include "brw_program.h"
+#include "brw_cs.h"
+#include "brw_gs.h"
+#include "brw_state.h"
+#include "brw_vs.h"
+#include "brw_wm.h"
+
+static bool
+debug_enabled_for_stage(gl_shader_stage stage)
+{
+   static const uint64_t stage_debug_flags[] = {
+      DEBUG_VS, DEBUG_TCS, DEBUG_TES, DEBUG_GS, DEBUG_WM, DEBUG_CS,
+   };
+   assert((int)stage >= 0 && stage < ARRAY_SIZE(stage_debug_flags));
+   return (INTEL_DEBUG & stage_debug_flags[stage]) != 0;
+}
+
+static void
+gen_shader_sha1(struct gl_program *prog, gl_shader_stage stage,
+                void *key, unsigned char *out_sha1)
+{
+   char sha1_buf[41];
+   unsigned char sha1[20];
+   char manifest[256];
+   int offset = 0;
+
+   _mesa_sha1_format(sha1_buf, prog->sh.data->sha1);
+   offset += snprintf(manifest, sizeof(manifest), "program: %s\n", sha1_buf);
+
+   _mesa_sha1_compute(key, brw_prog_key_size(stage), sha1);
+   _mesa_sha1_format(sha1_buf, sha1);
+   offset += snprintf(manifest + offset, sizeof(manifest) - offset,
+                      "%s_key: %s\n", _mesa_shader_stage_to_abbrev(stage),
+                      sha1_buf);
+
+   _mesa_sha1_compute(manifest, strlen(manifest), out_sha1);
+}
+
+static bool
+read_blob_program_data(struct blob_reader *binary, struct gl_program *prog,
+                       gl_shader_stage stage, const uint8_t **program,
+                       struct brw_stage_prog_data *prog_data)
+{
+   return
+      brw_read_blob_program_data(binary, prog, stage, program, prog_data) &&
+      (binary->current == binary->end);
+}
+
+static bool
+read_and_upload(struct brw_context *brw, struct disk_cache *cache,
+                struct gl_program *prog, gl_shader_stage stage)
+{
+   unsigned char binary_sha1[20];
+
+   union brw_any_prog_key prog_key;
+
+   switch (stage) {
+   case MESA_SHADER_VERTEX:
+      brw_vs_populate_key(brw, &prog_key.vs);
+      break;
+   case MESA_SHADER_TESS_CTRL:
+      brw_tcs_populate_key(brw, &prog_key.tcs);
+      break;
+   case MESA_SHADER_TESS_EVAL:
+      brw_tes_populate_key(brw, &prog_key.tes);
+      break;
+   case MESA_SHADER_GEOMETRY:
+      brw_gs_populate_key(brw, &prog_key.gs);
+      break;
+   case MESA_SHADER_FRAGMENT:
+      brw_wm_populate_key(brw, &prog_key.wm);
+      break;
+   case MESA_SHADER_COMPUTE:
+      brw_cs_populate_key(brw, &prog_key.cs);
+      break;
+   default:
+      unreachable("Unsupported stage!");
+   }
+
+   /* We don't care what instance of the program it is for the disk cache hash
+    * lookup, so set the id to 0 for the sha1 hashing. program_string_id will
+    * be set below.
+    */
+   brw_prog_key_set_id(&prog_key, stage, 0);
+
+   gen_shader_sha1(prog, stage, &prog_key, binary_sha1);
+
+   size_t buffer_size;
+   uint8_t *buffer = disk_cache_get(cache, binary_sha1, &buffer_size);
+   if (buffer == NULL) {
+      if (brw->ctx._Shader->Flags & GLSL_CACHE_INFO) {
+         char sha1_buf[41];
+         _mesa_sha1_format(sha1_buf, binary_sha1);
+         fprintf(stderr, "No cached %s binary found for: %s\n",
+                 _mesa_shader_stage_to_abbrev(stage), sha1_buf);
+      }
+      return false;
+   }
+
+   if (brw->ctx._Shader->Flags & GLSL_CACHE_INFO) {
+      char sha1_buf[41];
+      _mesa_sha1_format(sha1_buf, binary_sha1);
+      fprintf(stderr, "attempting to populate bo cache with binary: %s\n",
+              sha1_buf);
+   }
+
+   struct blob_reader binary;
+   blob_reader_init(&binary, buffer, buffer_size);
+
+   const uint8_t *program;
+   struct brw_stage_prog_data *prog_data =
+      ralloc_size(NULL, sizeof(union brw_any_prog_data));
+   if (!read_blob_program_data(&binary, prog, stage, &program, prog_data)) {
+      /* Something very bad has gone wrong discard the item from the cache and
+       * rebuild from source.
+       */
+      if (brw->ctx._Shader->Flags & GLSL_CACHE_INFO) {
+         fprintf(stderr, "Error reading program from cache (invalid i965 "
+                 "cache item)\n");
+      }
+
+      disk_cache_remove(cache, binary_sha1);
+      ralloc_free(prog_data);
+      free(buffer);
+      return false;
+   }
+
+   enum brw_cache_id cache_id;
+   struct brw_stage_state *stage_state;
+
+   switch (stage) {
+   case MESA_SHADER_VERTEX:
+      cache_id = BRW_CACHE_VS_PROG;
+      stage_state = &brw->vs.base;
+      break;
+   case MESA_SHADER_TESS_CTRL:
+      cache_id = BRW_CACHE_TCS_PROG;
+      stage_state = &brw->tcs.base;
+      break;
+   case MESA_SHADER_TESS_EVAL:
+      cache_id = BRW_CACHE_TES_PROG;
+      stage_state = &brw->tes.base;
+      break;
+   case MESA_SHADER_GEOMETRY:
+      cache_id = BRW_CACHE_GS_PROG;
+      stage_state = &brw->gs.base;
+      break;
+   case MESA_SHADER_FRAGMENT:
+      cache_id = BRW_CACHE_FS_PROG;
+      stage_state = &brw->wm.base;
+      break;
+   case MESA_SHADER_COMPUTE:
+      cache_id = BRW_CACHE_CS_PROG;
+      stage_state = &brw->cs.base;
+      break;
+   default:
+      unreachable("Unsupported stage!");
+   }
+
+   brw_prog_key_set_id(&prog_key, stage, brw_program(prog)->id);
+
+   brw_alloc_stage_scratch(brw, stage_state, prog_data->total_scratch);
+
+   if (unlikely(debug_enabled_for_stage(stage))) {
+      fprintf(stderr, "NIR for %s program %d loaded from disk shader cache:\n",
+              _mesa_shader_stage_to_abbrev(stage), brw_program(prog)->id);
+      brw_program_deserialize_driver_blob(&brw->ctx, prog, stage);
+      nir_shader *nir = prog->nir;
+      nir_print_shader(nir, stderr);
+      fprintf(stderr, "Native code for %s %s shader %s from disk cache:\n",
+              nir->info.label ? nir->info.label : "unnamed",
+              _mesa_shader_stage_to_string(nir->info.stage), nir->info.name);
+      brw_disassemble(&brw->screen->devinfo, program, 0,
+                      prog_data->program_size, stderr);
+   }
+
+   brw_upload_cache(&brw->cache, cache_id, &prog_key, brw_prog_key_size(stage),
+                    program, prog_data->program_size, prog_data,
+                    brw_prog_data_size(stage), &stage_state->prog_offset,
+                    &stage_state->prog_data);
+
+   prog->program_written_to_cache = true;
+
+   ralloc_free(prog_data);
+   free(buffer);
+
+   return true;
+}
+
+bool
+brw_disk_cache_upload_program(struct brw_context *brw, gl_shader_stage stage)
+{
+   struct disk_cache *cache = brw->ctx.Cache;
+   if (cache == NULL)
+      return false;
+
+   struct gl_program *prog = brw->ctx._Shader->CurrentProgram[stage];
+   if (prog == NULL)
+      return false;
+
+   if (brw->ctx._Shader->Flags & GLSL_CACHE_FALLBACK)
+      goto fail;
+
+   if (!read_and_upload(brw, cache, prog, stage))
+      goto fail;
+
+   if (brw->ctx._Shader->Flags & GLSL_CACHE_INFO) {
+      fprintf(stderr, "read gen program from cache\n");
+   }
+
+   return true;
+
+fail:
+   prog->program_written_to_cache = false;
+   if (brw->ctx._Shader->Flags & GLSL_CACHE_INFO) {
+      fprintf(stderr, "falling back to nir %s.\n",
+              _mesa_shader_stage_to_abbrev(prog->info.stage));
+   }
+
+   brw_program_deserialize_driver_blob(&brw->ctx, prog, stage);
+
+   return false;
+}
+
+static void
+write_program_data(struct brw_context *brw, struct gl_program *prog,
+                   void *key, struct brw_stage_prog_data *prog_data,
+                   uint32_t prog_offset, struct disk_cache *cache,
+                   gl_shader_stage stage)
+{
+   struct blob binary;
+   blob_init(&binary);
+
+   const void *program_map = brw->cache.map + prog_offset;
+   /* TODO: Improve perf for non-LLC. It would be best to save it at program
+    * generation time when the program is in normal memory accessible with
+    * cache to the CPU. Another easier change would be to use
+    * _mesa_streaming_load_memcpy to read from the program mapped memory. */
+   brw_write_blob_program_data(&binary, stage, program_map, prog_data);
+
+   unsigned char sha1[20];
+   char buf[41];
+   gen_shader_sha1(prog, stage, key, sha1);
+   _mesa_sha1_format(buf, sha1);
+   if (brw->ctx._Shader->Flags & GLSL_CACHE_INFO) {
+      fprintf(stderr, "putting binary in cache: %s\n", buf);
+   }
+
+   disk_cache_put(cache, sha1, binary.data, binary.size, NULL);
+
+   prog->program_written_to_cache = true;
+   blob_finish(&binary);
+}
+
+void
+brw_disk_cache_write_render_programs(struct brw_context *brw)
+{
+   struct disk_cache *cache = brw->ctx.Cache;
+   if (cache == NULL)
+      return;
+
+   struct gl_program *prog =
+      brw->ctx._Shader->CurrentProgram[MESA_SHADER_VERTEX];
+   if (prog && !prog->program_written_to_cache) {
+      struct brw_vs_prog_key vs_key;
+      brw_vs_populate_key(brw, &vs_key);
+      vs_key.program_string_id = 0;
+
+      write_program_data(brw, prog, &vs_key, brw->vs.base.prog_data,
+                         brw->vs.base.prog_offset, cache,
+                         MESA_SHADER_VERTEX);
+   }
+
+   prog = brw->ctx._Shader->CurrentProgram[MESA_SHADER_TESS_CTRL];
+   if (prog && !prog->program_written_to_cache) {
+      struct brw_tcs_prog_key tcs_key;
+      brw_tcs_populate_key(brw, &tcs_key);
+      tcs_key.program_string_id = 0;
+
+      write_program_data(brw, prog, &tcs_key, brw->tcs.base.prog_data,
+                         brw->tcs.base.prog_offset, cache,
+                         MESA_SHADER_TESS_CTRL);
+   }
+
+   prog = brw->ctx._Shader->CurrentProgram[MESA_SHADER_TESS_EVAL];
+   if (prog && !prog->program_written_to_cache) {
+      struct brw_tes_prog_key tes_key;
+      brw_tes_populate_key(brw, &tes_key);
+      tes_key.program_string_id = 0;
+
+      write_program_data(brw, prog, &tes_key, brw->tes.base.prog_data,
+                         brw->tes.base.prog_offset, cache,
+                         MESA_SHADER_TESS_EVAL);
+   }
+
+   prog = brw->ctx._Shader->CurrentProgram[MESA_SHADER_GEOMETRY];
+   if (prog && !prog->program_written_to_cache) {
+      struct brw_gs_prog_key gs_key;
+      brw_gs_populate_key(brw, &gs_key);
+      gs_key.program_string_id = 0;
+
+      write_program_data(brw, prog, &gs_key, brw->gs.base.prog_data,
+                         brw->gs.base.prog_offset, cache,
+                         MESA_SHADER_GEOMETRY);
+   }
+
+   prog = brw->ctx._Shader->CurrentProgram[MESA_SHADER_FRAGMENT];
+   if (prog && !prog->program_written_to_cache) {
+      struct brw_wm_prog_key wm_key;
+      brw_wm_populate_key(brw, &wm_key);
+      wm_key.program_string_id = 0;
+
+      write_program_data(brw, prog, &wm_key, brw->wm.base.prog_data,
+                         brw->wm.base.prog_offset, cache,
+                         MESA_SHADER_FRAGMENT);
+   }
+}
+
+void
+brw_disk_cache_write_compute_program(struct brw_context *brw)
+{
+   struct disk_cache *cache = brw->ctx.Cache;
+   if (cache == NULL)
+      return;
+
+   struct gl_program *prog =
+      brw->ctx._Shader->CurrentProgram[MESA_SHADER_COMPUTE];
+   if (prog && !prog->program_written_to_cache) {
+      struct brw_cs_prog_key cs_key;
+      brw_cs_populate_key(brw, &cs_key);
+      cs_key.program_string_id = 0;
+
+      write_program_data(brw, prog, &cs_key, brw->cs.base.prog_data,
+                         brw->cs.base.prog_offset, cache,
+                         MESA_SHADER_COMPUTE);
+   }
+}
+
+void
+brw_disk_cache_init(struct intel_screen *screen)
+{
+#ifdef ENABLE_SHADER_CACHE
+   if (INTEL_DEBUG & DEBUG_DISK_CACHE_DISABLE_MASK)
+      return;
+
+   /* array length: print length + null char + 1 extra to verify it is unused */
+   char renderer[11];
+   MAYBE_UNUSED int len = snprintf(renderer, sizeof(renderer), "i965_%04x",
+                                   screen->deviceID);
+   assert(len == sizeof(renderer) - 2);
+
+   const struct build_id_note *note =
+      build_id_find_nhdr_for_addr(brw_disk_cache_init);
+   assert(note && build_id_length(note) == 20 /* sha1 */);
+
+   const uint8_t *id_sha1 = build_id_data(note);
+   assert(id_sha1);
+
+   char timestamp[41];
+   _mesa_sha1_format(timestamp, id_sha1);
+
+   const uint64_t driver_flags =
+      brw_get_compiler_config_value(screen->compiler);
+   screen->disk_cache = disk_cache_create(renderer, timestamp, driver_flags);
+#endif
+}
diff --git a/lib/mesa/src/mesa/drivers/dri/i965/brw_generate_mipmap.c b/lib/mesa/src/mesa/drivers/dri/i965/brw_generate_mipmap.c
new file mode 100644
index 000000000..4125ae6e1
--- /dev/null
+++ b/lib/mesa/src/mesa/drivers/dri/i965/brw_generate_mipmap.c
@@ -0,0 +1,144 @@
+/*
+ * Copyright © 2016 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include "main/mipmap.h"
+#include "main/teximage.h"
+#include "brw_blorp.h"
+#include "brw_context.h"
+#include "intel_tex.h"
+#include "drivers/common/meta.h"
+
+#define FILE_DEBUG_FLAG DEBUG_BLORP
+
+
+/**
+ * The GenerateMipmap() driver hook.
+ */
+void
+brw_generate_mipmap(struct gl_context *ctx, GLenum target,
+                    struct gl_texture_object *tex_obj)
+{
+   struct brw_context *brw = brw_context(ctx);
+   struct gen_device_info *devinfo = &brw->screen->devinfo;
+   struct intel_texture_object *intel_obj = intel_texture_object(tex_obj);
+   const unsigned base_level = tex_obj->BaseLevel;
+   unsigned last_level, first_layer, last_layer;
+
+   /* Blorp doesn't handle combined depth/stencil surfaces on Gen4-5 yet. */
+   if (devinfo->gen <= 5 &&
+       (tex_obj->Image[0][base_level]->_BaseFormat == GL_DEPTH_COMPONENT ||
+        tex_obj->Image[0][base_level]->_BaseFormat == GL_DEPTH_STENCIL)) {
+      _mesa_meta_GenerateMipmap(ctx, target, tex_obj);
+      return;
+   }
+
+   /* find expected last mipmap level to generate */
+   last_level = _mesa_compute_num_levels(ctx, tex_obj, target) - 1;
+
+   if (last_level == 0)
+      return;
+
+   /* The texture isn't in a "complete" state yet so set the expected
+    * last_level here; we're not going through normal texture validation.
+    */
+   intel_obj->_MaxLevel = last_level;
+
+   if (!tex_obj->Immutable) {
+      _mesa_prepare_mipmap_levels(ctx, tex_obj, base_level, last_level);
+
+      /* At this point, memory for all the texture levels has been
+       * allocated.  However, the base level image may be in one resource
+       * while the subsequent/smaller levels may be in another resource.
+       * Finalizing the texture will copy the base images from the former
+       * resource to the latter.
+       *
+       * After this, we'll have all mipmap levels in one resource.
+       */
+      intel_finalize_mipmap_tree(brw, tex_obj);
+   }
+
+   struct intel_mipmap_tree *mt = intel_obj->mt;
+   if (!mt) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "mipmap generation");
+      return;
+   }
+
+   const mesa_format format = intel_obj->_Format;
+
+   /* Fall back to the CPU for non-renderable cases.
+    *
+    * TODO: 3D textures require blending data from multiple slices,
+    * which means we need custom shaders.  For now, fall back.
+    */
+   if (!brw->mesa_format_supports_render[format] || target == GL_TEXTURE_3D) {
+      _mesa_generate_mipmap(ctx, target, tex_obj);
+      return;
+   }
+
+   const struct isl_extent4d *base_size = &mt->surf.logical_level0_px;
+
+   if (mt->target == GL_TEXTURE_CUBE_MAP) {
+      first_layer = _mesa_tex_target_to_face(target);
+      last_layer = first_layer;
+   } else {
+      first_layer = 0;
+      last_layer = base_size->array_len - 1;
+   }
+
+   /* The GL_EXT_texture_sRGB_decode extension's issues section says:
+    *
+    *    "10) How is mipmap generation of sRGB textures affected by the
+    *     TEXTURE_SRGB_DECODE_EXT parameter?
+    *
+    *     RESOLVED:  When the TEXTURE_SRGB_DECODE parameter is DECODE_EXT
+    *     for an sRGB texture, mipmap generation should decode sRGB texels
+    *     to a linear RGB color space, perform downsampling, then encode
+    *     back to an sRGB color space.  (Issue 24 in the EXT_texture_sRGB
+    *     specification provides a rationale for why.)  When the parameter
+    *     is SKIP_DECODE_EXT instead, mipmap generation skips the encode
+    *     and decode steps during mipmap generation.  By skipping the
+    *     encode and decode steps, sRGB mipmap generation should match
+    *     the mipmap generation for a non-sRGB texture."
+    */
+   bool do_srgb = tex_obj->Sampler.sRGBDecode == GL_DECODE_EXT;
+
+   for (unsigned dst_level = base_level + 1;
+        dst_level <= last_level;
+        dst_level++) {
+
+      const unsigned src_level = dst_level - 1;
+
+      for (unsigned layer = first_layer; layer <= last_layer; layer++) {
+         brw_blorp_blit_miptrees(brw, mt, src_level, layer, format,
+                                 SWIZZLE_XYZW, mt, dst_level, layer, format,
+                                 0, 0,
+                                 minify(base_size->width, src_level),
+                                 minify(base_size->height, src_level),
+                                 0, 0,
+                                 minify(base_size->width, dst_level),
+                                 minify(base_size->height, dst_level),
+                                 GL_LINEAR, false, false,
+                                 do_srgb, do_srgb);
+      }
+   }
+}
diff --git a/lib/mesa/src/mesa/drivers/dri/i965/brw_pipe_control.h b/lib/mesa/src/mesa/drivers/dri/i965/brw_pipe_control.h
index 6e9a40487..4c58e1666 100644
--- a/lib/mesa/src/mesa/drivers/dri/i965/brw_pipe_control.h
+++ b/lib/mesa/src/mesa/drivers/dri/i965/brw_pipe_control.h
@@ -34,6 +34,7 @@ struct brw_bo;
  * additional flushing control.
  */
 #define _3DSTATE_PIPE_CONTROL		(CMD_3D | (3 << 27) | (2 << 24))
+#define PIPE_CONTROL_LRI_WRITE_IMMEDIATE (1 << 23) /* Gen7+ */
 #define PIPE_CONTROL_CS_STALL		(1 << 20)
 #define PIPE_CONTROL_GLOBAL_SNAPSHOT_COUNT_RESET	(1 << 19)
 #define PIPE_CONTROL_TLB_INVALIDATE	(1 << 18)
@@ -85,5 +86,6 @@ void brw_emit_post_sync_nonzero_flush(struct brw_context *brw);
 void brw_emit_depth_stall_flushes(struct brw_context *brw);
 void gen7_emit_vs_workaround_flush(struct brw_context *brw);
 void gen7_emit_cs_stall_flush(struct brw_context *brw);
+void gen10_emit_isp_disable(struct brw_context *brw);
 
 #endif
diff --git a/lib/mesa/src/mesa/drivers/dri/i965/intel_buffers.c b/lib/mesa/src/mesa/drivers/dri/i965/intel_buffers.c
index fd522cc4f..dae56e336 100644
--- a/lib/mesa/src/mesa/drivers/dri/i965/intel_buffers.c
+++ b/lib/mesa/src/mesa/drivers/dri/i965/intel_buffers.c
@@ -33,7 +33,7 @@
 #include "main/renderbuffer.h"
 
 static void
-intelDrawBuffer(struct gl_context * ctx, GLenum mode)
+intelDrawBuffer(struct gl_context *ctx)
 {
    if (_mesa_is_front_buffer_drawing(ctx->DrawBuffer)) {
       struct brw_context *const brw = brw_context(ctx);
diff --git a/lib/mesa/src/mesa/drivers/dri/meson.build b/lib/mesa/src/mesa/drivers/dri/meson.build
new file mode 100644
index 000000000..d98c823f5
--- /dev/null
+++ b/lib/mesa/src/mesa/drivers/dri/meson.build
@@ -0,0 +1,84 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+subdir('common')
+
+dri_drivers = []
+dri_link = []
+if with_dri_swrast
+  subdir('swrast')
+endif
+if with_dri_i915
+  subdir('i915')
+endif
+if with_dri_i965
+  subdir('i965')
+endif
+if with_dri_r100
+  subdir('radeon')
+endif
+if with_dri_r200
+  subdir('r200')
+endif
+if with_dri_nouveau
+  subdir('nouveau')
+endif
+
+if dri_drivers != []
+  libmesa_dri_drivers = shared_library(
+    'mesa_dri_drivers',
+    [],
+    link_whole : dri_drivers,
+    link_with : [
+      libmegadriver_stub, libdricommon, libxmlconfig, libglapi, libmesa_util,
+      libmesa_classic,
+    ],
+    dependencies : [
+      dep_selinux, dep_libdrm, dep_expat, dep_m, dep_thread, dep_dl, idep_nir,
+    ],
+    link_args : [ld_args_build_id, ld_args_bsymbolic, ld_args_gc_sections],
+  )
+
+  meson.add_install_script(
+    prog_python.path(),
+    join_paths(meson.source_root(), 'bin/install_megadrivers.py'),
+    libmesa_dri_drivers.full_path(),
+    dri_drivers_path,
+    dri_link,
+  )
+endif
+
+# This needs to be installed if any dri drivers (including gallium dri drivers)
+# are built.
+if with_dri
+  dri_req_private = []
+  if dep_libdrm.found()
+    dri_req_private = ['libdrm >= ' + dep_libdrm.version()]
+  endif
+
+  pkg.generate(
+    name : 'dri',
+    filebase : 'dri',
+    description : 'Direct Rendering Infrastructure',
+    version : meson.project_version(),
+    variables : ['dridriverdir=${prefix}/' + dri_drivers_path],
+    requires_private : dri_req_private,
+  )
+endif
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/Makefile.am b/lib/mesa/src/mesa/drivers/dri/nouveau/Makefile.am
index 01e34a8e3..59fd7fce7 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/Makefile.am
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/Makefile.am
@@ -26,7 +26,8 @@ include Makefile.sources
 EXTRA_DIST = \
 	nouveau_render_t.c \
 	nouveau_swtnl_t.c \
-	nouveau_vbo_t.c
+	nouveau_vbo_t.c \
+	meson.build
 
 AM_CFLAGS = \
 	-I$(top_srcdir)/include \
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/meson.build b/lib/mesa/src/mesa/drivers/dri/nouveau/meson.build
new file mode 100644
index 000000000..9ad6779d8
--- /dev/null
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/meson.build
@@ -0,0 +1,92 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+files_nouveau_vieux = files(
+  'nouveau_array.c',
+  'nouveau_array.h',
+  'nouveau_bufferobj.c',
+  'nouveau_bufferobj.h',
+  'nouveau_context.c',
+  'nouveau_context.h',
+  'nouveau_driver.c',
+  'nouveau_driver.h',
+  'nouveau_fbo.c',
+  'nouveau_fbo.h',
+  'nouveau_gldefs.h',
+  'nouveau_local.h',
+  'nouveau_render.h',
+  'nouveau_scratch.c',
+  'nouveau_scratch.h',
+  'nouveau_screen.c',
+  'nouveau_screen.h',
+  'nouveau_span.c',
+  'nouveau_state.c',
+  'nouveau_state.h',
+  'nouveau_surface.c',
+  'nouveau_surface.h',
+  'nouveau_texture.c',
+  'nouveau_texture.h',
+  'nouveau_util.h',
+  'nv01_2d.xml.h',
+  'nv04_3d.xml.h',
+  'nv04_context.c',
+  'nv04_context.h',
+  'nv04_driver.h',
+  'nv04_render.c',
+  'nv04_state_fb.c',
+  'nv04_state_frag.c',
+  'nv04_state_raster.c',
+  'nv04_state_tex.c',
+  'nv04_surface.c',
+  'nv10_3d.xml.h',
+  'nv10_context.c',
+  'nv10_driver.h',
+  'nv10_render.c',
+  'nv10_state_fb.c',
+  'nv10_state_frag.c',
+  'nv10_state_polygon.c',
+  'nv10_state_raster.c',
+  'nv10_state_tex.c',
+  'nv10_state_tnl.c',
+  'nv20_3d.xml.h',
+  'nv20_context.c',
+  'nv20_driver.h',
+  'nv20_render.c',
+  'nv20_state_fb.c',
+  'nv20_state_frag.c',
+  'nv20_state_polygon.c',
+  'nv20_state_raster.c',
+  'nv20_state_tex.c',
+  'nv20_state_tnl.c',
+  'nv_m2mf.xml.h',
+  'nv_object.xml.h',
+)
+
+libnouveau_vieux = static_library(
+  'nouveau_vieux',
+  [files_nouveau_vieux, xmlpool_options_h],
+  include_directories : [inc_common, inc_dri_common, inc_util],
+  c_args : [c_vis_args],
+  cpp_args : [cpp_vis_args],
+  dependencies : [dep_libdrm, dep_libdrm_nouveau],
+)
+
+dri_drivers += libnouveau_vieux
+dri_link += 'nouveau_vieux_dri.so'
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_context.h b/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_context.h
index 6ab865c7b..dcb7bbb23 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_context.h
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_context.h
@@ -110,8 +110,7 @@ struct nouveau_context {
 GLboolean
 nouveau_context_create(gl_api api,
 		       const struct gl_config *visual, __DRIcontext *dri_ctx,
-		       unsigned major_version, unsigned minor_version,
-		       uint32_t flags, bool notify_reset, unsigned priority,
+		       const struct __DriverContextConfig *ctx_config,
 		       unsigned *error, void *share_ctx);
 
 GLboolean
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_driver.c b/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_driver.c
index 998e751fc..1698df6ab 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_driver.c
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_driver.c
@@ -132,7 +132,13 @@ nouveau_clear(struct gl_context *ctx, GLbitfield buffers)
 			else
 				value = pack_rgba_clamp_f(s->format, color);
 
-			mask = pack_rgba_i(s->format, ctx->Color.ColorMask[0]);
+			const uint8_t colormask[4] = {
+				GET_COLORMASK_BIT(ctx->Color.ColorMask, 0, 0) ? 0xff : 0,
+				GET_COLORMASK_BIT(ctx->Color.ColorMask, 0, 1) ? 0xff : 0,
+				GET_COLORMASK_BIT(ctx->Color.ColorMask, 0, 2) ? 0xff : 0,
+				GET_COLORMASK_BIT(ctx->Color.ColorMask, 0, 3) ? 0xff : 0,
+			};
+			mask = pack_rgba_i(s->format, colormask);
 
 			if (mask)
 				context_drv(ctx)->surface_fill(
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_fbo.c b/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_fbo.c
index c78d4baa1..77e7be112 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_fbo.c
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_fbo.c
@@ -133,13 +133,17 @@ nouveau_renderbuffer_map(struct gl_context *ctx,
 			 GLuint x, GLuint y, GLuint w, GLuint h,
 			 GLbitfield mode,
 			 GLubyte **out_map,
-			 GLint *out_stride)
+			 GLint *out_stride,
+			 bool flip_y)
 {
 	struct nouveau_surface *s = &to_nouveau_renderbuffer(rb)->surface;
 	GLubyte *map;
 	int stride;
 	int flags = 0;
 
+	/* driver does not support GL_FRAMEBUFFER_FLIP_Y_MESA */
+	assert((rb->Name == 0) == flip_y);
+
 	if (mode & GL_MAP_READ_BIT)
 		flags |= NOUVEAU_BO_RD;
 	if (mode & GL_MAP_WRITE_BIT)
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_render.h b/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_render.h
index 0539c3775..4d45d5c2e 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_render.h
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_render.h
@@ -27,7 +27,7 @@
 #ifndef __NOUVEAU_RENDER_H__
 #define __NOUVEAU_RENDER_H__
 
-#include "vbo/vbo_context.h"
+#include "vbo/vbo.h"
 #include "nouveau_array.h"
 
 typedef void (*dispatch_t)(struct gl_context *, unsigned int, int, unsigned int);
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_util.h b/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_util.h
index 8087445b6..088a25637 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_util.h
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nouveau_util.h
@@ -193,7 +193,7 @@ is_texture_source(int s)
 }
 
 static inline struct gl_texgen *
-get_texgen_coord(struct gl_texture_unit *u, int i)
+get_texgen_coord(struct gl_fixedfunc_texture_unit *u, int i)
 {
 	return ((struct gl_texgen *[])
 		{ &u->GenS, &u->GenT, &u->GenR, &u->GenQ }) [i];
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nv04_context.c b/lib/mesa/src/mesa/drivers/dri/nouveau/nv04_context.c
index 3cc219bea..324595e24 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nv04_context.c
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nv04_context.c
@@ -32,17 +32,18 @@
 #include "nv04_driver.h"
 
 static GLboolean
-texunit_needs_combiners(struct gl_texture_unit *u)
+texunit_needs_combiners(struct gl_texture_unit *u,
+                        struct gl_fixedfunc_texture_unit *f)
 {
 	struct gl_texture_object *t = u->_Current;
 	struct gl_texture_image *ti = t->Image[0][t->BaseLevel];
 
 	return ti->TexFormat == MESA_FORMAT_A_UNORM8 ||
 		ti->TexFormat == MESA_FORMAT_L_UNORM8 ||
-		u->EnvMode == GL_COMBINE ||
-		u->EnvMode == GL_COMBINE4_NV ||
-		u->EnvMode == GL_BLEND ||
-		u->EnvMode == GL_ADD;
+		f->EnvMode == GL_COMBINE ||
+		f->EnvMode == GL_COMBINE4_NV ||
+		f->EnvMode == GL_BLEND ||
+		f->EnvMode == GL_ADD;
 }
 
 struct nouveau_object *
@@ -54,13 +55,11 @@ nv04_context_engine(struct gl_context *ctx)
 	struct nouveau_object *fahrenheit;
 
 	if ((ctx->Texture.Unit[0]._Current &&
-	     texunit_needs_combiners(&ctx->Texture.Unit[0])) ||
+	     texunit_needs_combiners(&ctx->Texture.Unit[0],
+                                     &ctx->Texture.FixedFuncUnit[0])) ||
 	    ctx->Texture.Unit[1]._Current ||
 	    ctx->Stencil.Enabled ||
-	    !(ctx->Color.ColorMask[0][RCOMP] &&
-	      ctx->Color.ColorMask[0][GCOMP] &&
-	      ctx->Color.ColorMask[0][BCOMP] &&
-	      ctx->Color.ColorMask[0][ACOMP]))
+	    GET_COLORMASK(ctx->Color.ColorMask, 0) != 0xf)
 		fahrenheit = hw->eng3dm;
 	else
 		fahrenheit = hw->eng3d;
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nv04_state_frag.c b/lib/mesa/src/mesa/drivers/dri/nouveau/nv04_state_frag.c
index 248a7d2b5..26e2e91a6 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nv04_state_frag.c
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nv04_state_frag.c
@@ -49,8 +49,8 @@ struct combiner_state {
 
 	/* GL state */
 	GLenum mode;
-	GLenum *source;
-	GLenum *operand;
+	GLenum16 *source;
+	GLenum16 *operand;
 	GLuint logscale;
 
 	/* Derived HW state */
@@ -64,7 +64,7 @@ struct combiner_state {
  * context. */
 #define INIT_COMBINER(chan, ctx, rc, i) do {			\
 		struct gl_tex_env_combine_state *c =		\
-			ctx->Texture.Unit[i]._CurrentCombine;	\
+			ctx->Texture.FixedFuncUnit[i]._CurrentCombine;	\
 		(rc)->ctx = ctx;				\
 		(rc)->unit = i;					\
 		(rc)->alpha = __INIT_COMBINER_ALPHA_##chan;	\
@@ -287,7 +287,7 @@ nv04_emit_tex_env(struct gl_context *ctx, int emit)
 	/* calculate non-multitex state */
 	nv04->blend &= ~NV04_TEXTURED_TRIANGLE_BLEND_TEXTURE_MAP__MASK;
 	if (ctx->Texture._MaxEnabledTexImageUnit != -1)
-		nv04->blend |= get_texenv_mode(ctx->Texture.Unit[0].EnvMode);
+		nv04->blend |= get_texenv_mode(ctx->Texture.FixedFuncUnit[0].EnvMode);
 	else
 		nv04->blend |= get_texenv_mode(GL_MODULATE);
 
@@ -295,5 +295,5 @@ nv04_emit_tex_env(struct gl_context *ctx, int emit)
 	nv04->alpha[i] = rc_a.hw;
 	nv04->color[i] = rc_c.hw;
 	nv04->factor   = pack_rgba_f(MESA_FORMAT_B8G8R8A8_UNORM,
-				     ctx->Texture.Unit[0].EnvColor);
+				     ctx->Texture.FixedFuncUnit[0].EnvColor);
 }
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_render.c b/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_render.c
index 701ba9a87..9b2cf229f 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_render.c
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_render.c
@@ -72,19 +72,19 @@ static struct nouveau_attr_info nv10_vertex_attrs[VERT_ATTRIB_MAX] = {
 		.imm_method = NV10_3D_VERTEX_FOG_1F,
 		.imm_fields = 1,
 	},
-	[VERT_ATTRIB_GENERIC0] = {
+	[VERT_ATTRIB_MAT(0)] = {
 		.emit = nv10_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC2] = {
+	[VERT_ATTRIB_MAT(2)] = {
 		.emit = nv10_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC4] = {
+	[VERT_ATTRIB_MAT(4)] = {
 		.emit = nv10_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC6] = {
+	[VERT_ATTRIB_MAT(6)] = {
 		.emit = nv10_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC8] = {
+	[VERT_ATTRIB_MAT(8)] = {
 		.emit = nv10_emit_material,
 	},
 };
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_state_frag.c b/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_state_frag.c
index c6e4bb0d5..27400d3eb 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_state_frag.c
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_state_frag.c
@@ -67,8 +67,8 @@ struct combiner_state {
 
 	/* GL state */
 	GLenum mode;
-	GLenum *source;
-	GLenum *operand;
+	GLenum16 *source;
+	GLenum16 *operand;
 	GLuint logscale;
 
 	/* Derived HW state */
@@ -80,7 +80,7 @@ struct combiner_state {
  * context. */
 #define INIT_COMBINER(chan, ctx, rc, i) do {			\
 		struct gl_tex_env_combine_state *c =		\
-			ctx->Texture.Unit[i]._CurrentCombine;	\
+			ctx->Texture.FixedFuncUnit[i]._CurrentCombine;	\
 		(rc)->ctx = ctx;				\
 		(rc)->unit = i;					\
 		(rc)->premodulate = c->_NumArgs##chan == 4;	\
@@ -332,7 +332,7 @@ nv10_get_general_combiner(struct gl_context *ctx, int i,
 	}
 
 	*k = pack_rgba_f(MESA_FORMAT_B8G8R8A8_UNORM,
-			 ctx->Texture.Unit[i].EnvColor);
+			 ctx->Texture.FixedFuncUnit[i].EnvColor);
 	*a_in = rc_a.in;
 	*a_out = rc_a.out;
 	*c_in = rc_c.in;
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_state_tex.c b/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_state_tex.c
index d85adfa9c..d741d4166 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_state_tex.c
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nv10_state_tex.c
@@ -39,7 +39,8 @@ nv10_emit_tex_gen(struct gl_context *ctx, int emit)
 	const int i = emit - NOUVEAU_STATE_TEX_GEN0;
 	struct nouveau_context *nctx = to_nouveau_context(ctx);
 	struct nouveau_pushbuf *push = context_push(ctx);
-	struct gl_texture_unit *unit = &ctx->Texture.Unit[i];
+	struct gl_fixedfunc_texture_unit *unit =
+           &ctx->Texture.FixedFuncUnit[i];
 	int j;
 
 	for (j = 0; j < 4; j++) {
@@ -73,7 +74,7 @@ nv10_emit_tex_mat(struct gl_context *ctx, int emit)
 
 	if (nctx->fallback == HWTNL &&
 	    ((ctx->Texture._TexMatEnabled & 1 << i) ||
-	     ctx->Texture.Unit[i]._GenFlags)) {
+	     ctx->Texture.FixedFuncUnit[i]._GenFlags)) {
 		BEGIN_NV04(push, NV10_3D(TEX_MATRIX_ENABLE(i)), 1);
 		PUSH_DATA (push, 1);
 
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_context.c b/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_context.c
index 0ab2db0b0..887aa5f48 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_context.c
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_context.c
@@ -55,13 +55,13 @@ nv20_clear(struct gl_context *ctx, GLbitfield buffers)
 		struct nouveau_surface *s = &to_nouveau_renderbuffer(
 			fb->_ColorDrawBuffers[0])->surface;
 
-		if (ctx->Color.ColorMask[0][RCOMP])
+		if (GET_COLORMASK_BIT(ctx->Color.ColorMask, 0, 0))
 			clear |= NV20_3D_CLEAR_BUFFERS_COLOR_R;
-		if (ctx->Color.ColorMask[0][GCOMP])
+		if (GET_COLORMASK_BIT(ctx->Color.ColorMask, 0, 1))
 			clear |= NV20_3D_CLEAR_BUFFERS_COLOR_G;
-		if (ctx->Color.ColorMask[0][BCOMP])
+		if (GET_COLORMASK_BIT(ctx->Color.ColorMask, 0, 2))
 			clear |= NV20_3D_CLEAR_BUFFERS_COLOR_B;
-		if (ctx->Color.ColorMask[0][ACOMP])
+		if (GET_COLORMASK_BIT(ctx->Color.ColorMask, 0, 3))
 			clear |= NV20_3D_CLEAR_BUFFERS_COLOR_A;
 
 		BEGIN_NV04(push, NV20_3D(CLEAR_VALUE), 1);
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_render.c b/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_render.c
index 2627bfa35..9a426ccec 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_render.c
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_render.c
@@ -82,34 +82,34 @@ static struct nouveau_attr_info nv20_vertex_attrs[VERT_ATTRIB_MAX] = {
 		.imm_method = NV20_3D_VERTEX_TX3_4F_S,
 		.imm_fields = 4,
 	},
-	[VERT_ATTRIB_GENERIC0] = {
+	[VERT_ATTRIB_MAT(0)] = {
 		.emit = nv20_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC1] = {
+	[VERT_ATTRIB_MAT(1)] = {
 		.emit = nv20_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC2] = {
+	[VERT_ATTRIB_MAT(2)] = {
 		.emit = nv20_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC3] = {
+	[VERT_ATTRIB_MAT(3)] = {
 		.emit = nv20_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC4] = {
+	[VERT_ATTRIB_MAT(4)] = {
 		.emit = nv20_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC5] = {
+	[VERT_ATTRIB_MAT(5)] = {
 		.emit = nv20_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC6] = {
+	[VERT_ATTRIB_MAT(6)] = {
 		.emit = nv20_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC7] = {
+	[VERT_ATTRIB_MAT(7)] = {
 		.emit = nv20_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC8] = {
+	[VERT_ATTRIB_MAT(8)] = {
 		.emit = nv20_emit_material,
 	},
-	[VERT_ATTRIB_GENERIC9] = {
+	[VERT_ATTRIB_MAT(9)] = {
 		.emit = nv20_emit_material,
 	},
 };
diff --git a/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_state_raster.c b/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_state_raster.c
index 4856053aa..c24c5bbae 100644
--- a/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_state_raster.c
+++ b/lib/mesa/src/mesa/drivers/dri/nouveau/nv20_state_raster.c
@@ -38,5 +38,5 @@ nv20_emit_logic_opcode(struct gl_context *ctx, int emit)
 
 	BEGIN_NV04(push, NV20_3D(COLOR_LOGIC_OP_ENABLE), 2);
 	PUSH_DATAb(push, ctx->Color.ColorLogicOpEnabled);
-	PUSH_DATA (push, nvgl_logicop_func(ctx->Color.LogicOp));
+	PUSH_DATA (push, ctx->Color.LogicOp);
 }
diff --git a/lib/mesa/src/mesa/drivers/dri/r200/Makefile.am b/lib/mesa/src/mesa/drivers/dri/r200/Makefile.am
index 110c02ed2..3d15c0199 100644
--- a/lib/mesa/src/mesa/drivers/dri/r200/Makefile.am
+++ b/lib/mesa/src/mesa/drivers/dri/r200/Makefile.am
@@ -24,7 +24,8 @@
 include Makefile.sources
 
 EXTRA_DIST = \
-	r200_maos_arrays.c
+	r200_maos_arrays.c \
+	meson.build
 
 AM_CFLAGS = \
 	-DRADEON_R200 \
diff --git a/lib/mesa/src/mesa/drivers/dri/r200/meson.build b/lib/mesa/src/mesa/drivers/dri/r200/meson.build
new file mode 100644
index 000000000..8389b4f63
--- /dev/null
+++ b/lib/mesa/src/mesa/drivers/dri/r200/meson.build
@@ -0,0 +1,91 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+files_r200 = files(
+  'radeon_buffer_objects.c',
+  'radeon_buffer_objects.h',
+  'radeon_cmdbuf.h',
+  'radeon_common.c',
+  'radeon_common.h',
+  'radeon_common_context.c',
+  'radeon_common_context.h',
+  'radeon_debug.c',
+  'radeon_debug.h',
+  'radeon_dma.c',
+  'radeon_dma.h',
+  'radeon_fbo.c',
+  'radeon_fog.c',
+  'radeon_fog.h',
+  'radeon_mipmap_tree.c',
+  'radeon_mipmap_tree.h',
+  'radeon_pixel_read.c',
+  'radeon_queryobj.c',
+  'radeon_queryobj.h',
+  'radeon_span.c',
+  'radeon_span.h',
+  'radeon_tex_copy.c',
+  'radeon_texture.c',
+  'radeon_texture.h',
+  'radeon_tile.c',
+  'radeon_tile.h',
+  'r200_blit.c',
+  'r200_blit.h',
+  'r200_cmdbuf.c',
+  'r200_context.c',
+  'r200_context.h',
+  'r200_fragshader.c',
+  'r200_ioctl.c',
+  'r200_ioctl.h',
+  'r200_maos.c',
+  'r200_maos.h',
+  'r200_reg.h',
+  'r200_sanity.c',
+  'r200_sanity.h',
+  'r200_state.c',
+  'r200_state.h',
+  'r200_state_init.c',
+  'r200_swtcl.c',
+  'r200_swtcl.h',
+  'r200_tcl.c',
+  'r200_tcl.h',
+  'r200_tex.c',
+  'r200_tex.h',
+  'r200_texstate.c',
+  'r200_vertprog.c',
+  'r200_vertprog.h',
+  'radeon_chipset.h',
+  'radeon_screen.c',
+  'radeon_screen.h',
+  'server/radeon_reg.h',
+)
+
+libr200 = static_library(
+  'r200',
+  [files_r200, xmlpool_options_h],
+  include_directories : [
+    inc_common, inc_dri_common, inc_util, include_directories('server'),
+  ],
+  c_args : [c_vis_args, '-DRADEON_R200'],
+  cpp_args : [cpp_vis_args],
+  dependencies : [dep_libdrm, dep_libdrm_radeon],
+)
+
+dri_drivers += libr200
+dri_link += 'r200_dri.so'
diff --git a/lib/mesa/src/mesa/drivers/dri/r200/radeon_common.h b/lib/mesa/src/mesa/drivers/dri/r200/radeon_common.h
index fedaf5063..a39b9360e 100644
--- a/lib/mesa/src/mesa/drivers/dri/r200/radeon_common.h
+++ b/lib/mesa/src/mesa/drivers/dri/r200/radeon_common.h
@@ -21,7 +21,7 @@ void radeon_clear_tris(struct gl_context *ctx, GLbitfield mask);
 
 void radeon_window_moved(radeonContextPtr radeon);
 void radeon_draw_buffer(struct gl_context *ctx, struct gl_framebuffer *fb);
-void radeonDrawBuffer( struct gl_context *ctx, GLenum mode );
+void radeonDrawBuffer(struct gl_context *ctx);
 void radeonReadBuffer( struct gl_context *ctx, GLenum mode );
 void radeon_viewport(struct gl_context *ctx);
 void radeon_fbo_init(struct radeon_context *radeon);
@@ -37,7 +37,6 @@ radeonReadPixels(struct gl_context * ctx,
 				GLenum format, GLenum type,
 				const struct gl_pixelstore_attrib *pack, GLvoid * pixels);
 
-void radeon_check_front_buffer_rendering(struct gl_context *ctx);
 static inline struct radeon_renderbuffer *radeon_renderbuffer(struct gl_renderbuffer *rb)
 {
 	struct radeon_renderbuffer *rrb = (struct radeon_renderbuffer *)rb;
diff --git a/lib/mesa/src/mesa/drivers/dri/r200/radeon_debug.c b/lib/mesa/src/mesa/drivers/dri/r200/radeon_debug.c
index 383a5df67..91f86a96b 100644
--- a/lib/mesa/src/mesa/drivers/dri/r200/radeon_debug.c
+++ b/lib/mesa/src/mesa/drivers/dri/r200/radeon_debug.c
@@ -75,7 +75,7 @@ void _radeon_debug_add_indent(void)
 	if (radeon->debug.indent_depth < length - 1) {
 		radeon->debug.indent[radeon->debug.indent_depth] = '\t';
 		++radeon->debug.indent_depth;
-	};
+	}
 }
 
 void _radeon_debug_remove_indent(void)
diff --git a/lib/mesa/src/mesa/drivers/dri/r200/radeon_screen.h b/lib/mesa/src/mesa/drivers/dri/r200/radeon_screen.h
index e70e334ab..af1b9454e 100644
--- a/lib/mesa/src/mesa/drivers/dri/r200/radeon_screen.h
+++ b/lib/mesa/src/mesa/drivers/dri/r200/radeon_screen.h
@@ -47,6 +47,68 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_reg.h"
 #include "util/xmlconfig.h"
 
+#define DRI_CONF_COLOR_REDUCTION_ROUND 0
+#define DRI_CONF_COLOR_REDUCTION_DITHER 1
+#define DRI_CONF_COLOR_REDUCTION(def) \
+DRI_CONF_OPT_BEGIN_V(color_reduction,enum,def,"0:1") \
+        DRI_CONF_DESC_BEGIN(en,"Initial color reduction method") \
+                DRI_CONF_ENUM(0,"Round colors") \
+                DRI_CONF_ENUM(1,"Dither colors") \
+        DRI_CONF_DESC_END \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_DITHER_XERRORDIFF 0
+#define DRI_CONF_DITHER_XERRORDIFFRESET 1
+#define DRI_CONF_DITHER_ORDERED 2
+#define DRI_CONF_DITHER_MODE(def) \
+DRI_CONF_OPT_BEGIN_V(dither_mode,enum,def,"0:2") \
+       DRI_CONF_DESC_BEGIN(en,"Color dithering method") \
+                DRI_CONF_ENUM(0,"Horizontal error diffusion") \
+                DRI_CONF_ENUM(1,"Horizontal error diffusion, reset error at line start") \
+                DRI_CONF_ENUM(2,"Ordered 2D color dithering") \
+        DRI_CONF_DESC_END \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_ROUND_TRUNC 0
+#define DRI_CONF_ROUND_ROUND 1
+#define DRI_CONF_ROUND_MODE(def) \
+DRI_CONF_OPT_BEGIN_V(round_mode,enum,def,"0:1") \
+       DRI_CONF_DESC_BEGIN(en,"Color rounding method") \
+                DRI_CONF_ENUM(0,"Round color components downward") \
+                DRI_CONF_ENUM(1,"Round to nearest color") \
+        DRI_CONF_DESC_END \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_FTHROTTLE_BUSY 0
+#define DRI_CONF_FTHROTTLE_USLEEPS 1
+#define DRI_CONF_FTHROTTLE_IRQS 2
+#define DRI_CONF_FTHROTTLE_MODE(def) \
+DRI_CONF_OPT_BEGIN_V(fthrottle_mode,enum,def,"0:2") \
+        DRI_CONF_DESC_BEGIN(en,"Method to limit rendering latency") \
+                DRI_CONF_ENUM(0,"Busy waiting for the graphics hardware") \
+                DRI_CONF_ENUM(1,"Sleep for brief intervals while waiting for the graphics hardware") \
+                DRI_CONF_ENUM(2,"Let the graphics hardware emit a software interrupt and sleep") \
+        DRI_CONF_DESC_END \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_TEXTURE_DEPTH_FB       0
+#define DRI_CONF_TEXTURE_DEPTH_32       1
+#define DRI_CONF_TEXTURE_DEPTH_16       2
+#define DRI_CONF_TEXTURE_DEPTH_FORCE_16 3
+#define DRI_CONF_TEXTURE_DEPTH(def) \
+DRI_CONF_OPT_BEGIN_V(texture_depth,enum,def,"0:3") \
+	DRI_CONF_DESC_BEGIN(en,"Texture color depth") \
+                DRI_CONF_ENUM(0,"Prefer frame buffer color depth") \
+                DRI_CONF_ENUM(1,"Prefer 32 bits per texel") \
+                DRI_CONF_ENUM(2,"Prefer 16 bits per texel") \
+                DRI_CONF_ENUM(3,"Force 16 bits per texel") \
+        DRI_CONF_DESC_END \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_TCL_SW 0
+#define DRI_CONF_TCL_PIPELINED 1
+#define DRI_CONF_TCL_VTXFMT 2
+#define DRI_CONF_TCL_CODEGEN 3
 
 typedef struct {
    drm_handle_t handle;			/* Handle to the DRM region */
@@ -126,7 +188,6 @@ struct __DRIimageRec {
 #define radeon_prepare_render               r200_radeon_prepare_render
 #define radeonUnbindContext                 r200_radeonUnbindContext
 #define radeon_update_renderbuffers         r200_radeon_update_renderbuffers
-#define radeon_check_front_buffer_rendering r200_radeon_check_front_buffer_rendering
 #define radeonCountStateEmitSize            r200_radeonCountStateEmitSize
 #define radeon_draw_buffer                  r200_radeon_draw_buffer
 #define radeonDrawBuffer                    r200_radeonDrawBuffer
diff --git a/lib/mesa/src/mesa/drivers/dri/r200/radeon_span.c b/lib/mesa/src/mesa/drivers/dri/r200/radeon_span.c
index 42f62a0cb..fa5b2d98f 100644
--- a/lib/mesa/src/mesa/drivers/dri/r200/radeon_span.c
+++ b/lib/mesa/src/mesa/drivers/dri/r200/radeon_span.c
@@ -44,6 +44,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/texformat.h"
 #include "main/renderbuffer.h"
 #include "main/samplerobj.h"
+#include "main/framebuffer.h"
 #include "swrast/swrast.h"
 #include "swrast/s_renderbuffer.h"
 
@@ -52,7 +53,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 
 static void
-radeon_renderbuffer_map(struct gl_context *ctx, struct gl_renderbuffer *rb)
+radeon_renderbuffer_map(struct gl_context *ctx,
+			struct gl_renderbuffer *rb,
+			bool flip_y)
 {
 	struct radeon_renderbuffer *rrb = radeon_renderbuffer(rb);
 	GLubyte *map;
@@ -63,7 +66,7 @@ radeon_renderbuffer_map(struct gl_context *ctx, struct gl_renderbuffer *rb)
 
 	ctx->Driver.MapRenderbuffer(ctx, rb, 0, 0, rb->Width, rb->Height,
 				    GL_MAP_READ_BIT | GL_MAP_WRITE_BIT,
-				    &map, &stride);
+				    &map, &stride, flip_y);
 
 	rrb->base.Map = map;
 	rrb->base.RowStride = stride;
@@ -95,9 +98,11 @@ radeon_map_framebuffer(struct gl_context *ctx, struct gl_framebuffer *fb)
 
 	/* check for render to textures */
 	for (i = 0; i < BUFFER_COUNT; i++)
-		radeon_renderbuffer_map(ctx, fb->Attachment[i].Renderbuffer);
+		radeon_renderbuffer_map(ctx, fb->Attachment[i].Renderbuffer,
+			fb->FlipY);
 
-	radeon_check_front_buffer_rendering(ctx);
+        if (_mesa_is_front_buffer_drawing(fb))
+		RADEON_CONTEXT(ctx)->front_buffer_dirty = true;
 }
 
 static void
@@ -113,7 +118,8 @@ radeon_unmap_framebuffer(struct gl_context *ctx, struct gl_framebuffer *fb)
 	for (i = 0; i < BUFFER_COUNT; i++)
 		radeon_renderbuffer_unmap(ctx, fb->Attachment[i].Renderbuffer);
 
-	radeon_check_front_buffer_rendering(ctx);
+        if (_mesa_is_front_buffer_drawing(fb))
+		RADEON_CONTEXT(ctx)->front_buffer_dirty = true;
 }
 
 static void radeonSpanRenderStart(struct gl_context * ctx)
diff --git a/lib/mesa/src/mesa/drivers/dri/radeon/Makefile.am b/lib/mesa/src/mesa/drivers/dri/radeon/Makefile.am
index aa898645c..9d43d2627 100644
--- a/lib/mesa/src/mesa/drivers/dri/radeon/Makefile.am
+++ b/lib/mesa/src/mesa/drivers/dri/radeon/Makefile.am
@@ -25,7 +25,8 @@ include Makefile.sources
 
 EXTRA_DIST = \
 	radeon_maos_verts.c \
-	radeon_maos_arrays.c
+	radeon_maos_arrays.c \
+	meson.build
 
 AM_CFLAGS = \
 	-DRADEON_R100 \
diff --git a/lib/mesa/src/mesa/drivers/dri/radeon/meson.build b/lib/mesa/src/mesa/drivers/dri/radeon/meson.build
new file mode 100644
index 000000000..bed608258
--- /dev/null
+++ b/lib/mesa/src/mesa/drivers/dri/radeon/meson.build
@@ -0,0 +1,87 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+files_r100 = files(
+  'radeon_buffer_objects.c',
+  'radeon_buffer_objects.h',
+  'radeon_cmdbuf.h',
+  'radeon_common.c',
+  'radeon_common_context.c',
+  'radeon_common_context.h',
+  'radeon_common.h',
+  'radeon_debug.c',
+  'radeon_debug.h',
+  'radeon_dma.c',
+  'radeon_dma.h',
+  'radeon_fbo.c',
+  'radeon_fog.c',
+  'radeon_fog.h',
+  'radeon_mipmap_tree.c',
+  'radeon_mipmap_tree.h',
+  'radeon_pixel_read.c',
+  'radeon_queryobj.c',
+  'radeon_queryobj.h',
+  'radeon_span.c',
+  'radeon_span.h',
+  'radeon_tex_copy.c',
+  'radeon_texture.c',
+  'radeon_texture.h',
+  'radeon_tile.c',
+  'radeon_tile.h',
+  'radeon_blit.c',
+  'radeon_blit.h',
+  'radeon_context.c',
+  'radeon_context.h',
+  'radeon_chipset.h',
+  'radeon_ioctl.c',
+  'radeon_ioctl.h',
+  'radeon_maos.c',
+  'radeon_maos.h',
+  'radeon_maos_vbtmp.h',
+  'radeon_sanity.c',
+  'radeon_sanity.h',
+  'radeon_screen.c',
+  'radeon_screen.h',
+  'radeon_state.c',
+  'radeon_state.h',
+  'radeon_state_init.c',
+  'radeon_swtcl.c',
+  'radeon_swtcl.h',
+  'radeon_tcl.c',
+  'radeon_tcl.h',
+  'radeon_tex.c',
+  'radeon_tex.h',
+  'radeon_texstate.c',
+  'server/radeon_reg.h',
+)
+
+libr100 = static_library(
+  'r100',
+  [files_r100, xmlpool_options_h],
+  include_directories : [
+    inc_common, inc_dri_common, inc_util, include_directories('server'),
+  ],
+  c_args : [c_vis_args, '-DRADEON_R100'],
+  cpp_args : [cpp_vis_args],
+  dependencies : [dep_libdrm, dep_libdrm_radeon],
+)
+
+dri_drivers += libr100
+dri_link += 'radeon_dri.so'
diff --git a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_common.h b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_common.h
index fedaf5063..a39b9360e 100644
--- a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_common.h
+++ b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_common.h
@@ -21,7 +21,7 @@ void radeon_clear_tris(struct gl_context *ctx, GLbitfield mask);
 
 void radeon_window_moved(radeonContextPtr radeon);
 void radeon_draw_buffer(struct gl_context *ctx, struct gl_framebuffer *fb);
-void radeonDrawBuffer( struct gl_context *ctx, GLenum mode );
+void radeonDrawBuffer(struct gl_context *ctx);
 void radeonReadBuffer( struct gl_context *ctx, GLenum mode );
 void radeon_viewport(struct gl_context *ctx);
 void radeon_fbo_init(struct radeon_context *radeon);
@@ -37,7 +37,6 @@ radeonReadPixels(struct gl_context * ctx,
 				GLenum format, GLenum type,
 				const struct gl_pixelstore_attrib *pack, GLvoid * pixels);
 
-void radeon_check_front_buffer_rendering(struct gl_context *ctx);
 static inline struct radeon_renderbuffer *radeon_renderbuffer(struct gl_renderbuffer *rb)
 {
 	struct radeon_renderbuffer *rrb = (struct radeon_renderbuffer *)rb;
diff --git a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_context.h b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_context.h
index 4124f50db..94917cf30 100644
--- a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_context.h
+++ b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_context.h
@@ -452,11 +452,8 @@ R100_CONTEXT(struct gl_context *ctx)
 extern GLboolean r100CreateContext( gl_api api,
 				    const struct gl_config *glVisual,
 				    __DRIcontext *driContextPriv,
-				    unsigned major_version,
-				    unsigned minor_version,
-				    uint32_t flags,
-                                    bool notify_reset,
-                                    unsigned priority,
+				    const struct __DriverContextConfig *
+				       ctx_config,
 				    unsigned *error,
 				    void *sharedContextPrivate);
 
diff --git a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_debug.c b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_debug.c
index 383a5df67..91f86a96b 100644
--- a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_debug.c
+++ b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_debug.c
@@ -75,7 +75,7 @@ void _radeon_debug_add_indent(void)
 	if (radeon->debug.indent_depth < length - 1) {
 		radeon->debug.indent[radeon->debug.indent_depth] = '\t';
 		++radeon->debug.indent_depth;
-	};
+	}
 }
 
 void _radeon_debug_remove_indent(void)
diff --git a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_maos_verts.c b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
index 9a7785062..c9c91f864 100644
--- a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
+++ b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
@@ -352,7 +352,7 @@ void radeonEmitArrays( struct gl_context *ctx, GLuint inputs )
 	    req |= RADEON_Q_BIT(unit);
 	    vtx |= RADEON_Q_BIT(unit);
 	 }
-	 if ( (ctx->Texture.Unit[unit].TexGenEnabled & (R_BIT | Q_BIT)) )
+	 if ( (ctx->Texture.FixedFuncUnit[unit].TexGenEnabled & (R_BIT | Q_BIT)) )
 	    vtx |= RADEON_Q_BIT(unit);
 	 else if ((VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size >= 3) &&
 	          (!ctx->Texture.Unit[unit]._Current ||
diff --git a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_screen.h b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_screen.h
index e70e334ab..af1b9454e 100644
--- a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_screen.h
+++ b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_screen.h
@@ -47,6 +47,68 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_reg.h"
 #include "util/xmlconfig.h"
 
+#define DRI_CONF_COLOR_REDUCTION_ROUND 0
+#define DRI_CONF_COLOR_REDUCTION_DITHER 1
+#define DRI_CONF_COLOR_REDUCTION(def) \
+DRI_CONF_OPT_BEGIN_V(color_reduction,enum,def,"0:1") \
+        DRI_CONF_DESC_BEGIN(en,"Initial color reduction method") \
+                DRI_CONF_ENUM(0,"Round colors") \
+                DRI_CONF_ENUM(1,"Dither colors") \
+        DRI_CONF_DESC_END \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_DITHER_XERRORDIFF 0
+#define DRI_CONF_DITHER_XERRORDIFFRESET 1
+#define DRI_CONF_DITHER_ORDERED 2
+#define DRI_CONF_DITHER_MODE(def) \
+DRI_CONF_OPT_BEGIN_V(dither_mode,enum,def,"0:2") \
+       DRI_CONF_DESC_BEGIN(en,"Color dithering method") \
+                DRI_CONF_ENUM(0,"Horizontal error diffusion") \
+                DRI_CONF_ENUM(1,"Horizontal error diffusion, reset error at line start") \
+                DRI_CONF_ENUM(2,"Ordered 2D color dithering") \
+        DRI_CONF_DESC_END \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_ROUND_TRUNC 0
+#define DRI_CONF_ROUND_ROUND 1
+#define DRI_CONF_ROUND_MODE(def) \
+DRI_CONF_OPT_BEGIN_V(round_mode,enum,def,"0:1") \
+       DRI_CONF_DESC_BEGIN(en,"Color rounding method") \
+                DRI_CONF_ENUM(0,"Round color components downward") \
+                DRI_CONF_ENUM(1,"Round to nearest color") \
+        DRI_CONF_DESC_END \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_FTHROTTLE_BUSY 0
+#define DRI_CONF_FTHROTTLE_USLEEPS 1
+#define DRI_CONF_FTHROTTLE_IRQS 2
+#define DRI_CONF_FTHROTTLE_MODE(def) \
+DRI_CONF_OPT_BEGIN_V(fthrottle_mode,enum,def,"0:2") \
+        DRI_CONF_DESC_BEGIN(en,"Method to limit rendering latency") \
+                DRI_CONF_ENUM(0,"Busy waiting for the graphics hardware") \
+                DRI_CONF_ENUM(1,"Sleep for brief intervals while waiting for the graphics hardware") \
+                DRI_CONF_ENUM(2,"Let the graphics hardware emit a software interrupt and sleep") \
+        DRI_CONF_DESC_END \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_TEXTURE_DEPTH_FB       0
+#define DRI_CONF_TEXTURE_DEPTH_32       1
+#define DRI_CONF_TEXTURE_DEPTH_16       2
+#define DRI_CONF_TEXTURE_DEPTH_FORCE_16 3
+#define DRI_CONF_TEXTURE_DEPTH(def) \
+DRI_CONF_OPT_BEGIN_V(texture_depth,enum,def,"0:3") \
+	DRI_CONF_DESC_BEGIN(en,"Texture color depth") \
+                DRI_CONF_ENUM(0,"Prefer frame buffer color depth") \
+                DRI_CONF_ENUM(1,"Prefer 32 bits per texel") \
+                DRI_CONF_ENUM(2,"Prefer 16 bits per texel") \
+                DRI_CONF_ENUM(3,"Force 16 bits per texel") \
+        DRI_CONF_DESC_END \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_TCL_SW 0
+#define DRI_CONF_TCL_PIPELINED 1
+#define DRI_CONF_TCL_VTXFMT 2
+#define DRI_CONF_TCL_CODEGEN 3
 
 typedef struct {
    drm_handle_t handle;			/* Handle to the DRM region */
@@ -126,7 +188,6 @@ struct __DRIimageRec {
 #define radeon_prepare_render               r200_radeon_prepare_render
 #define radeonUnbindContext                 r200_radeonUnbindContext
 #define radeon_update_renderbuffers         r200_radeon_update_renderbuffers
-#define radeon_check_front_buffer_rendering r200_radeon_check_front_buffer_rendering
 #define radeonCountStateEmitSize            r200_radeonCountStateEmitSize
 #define radeon_draw_buffer                  r200_radeon_draw_buffer
 #define radeonDrawBuffer                    r200_radeonDrawBuffer
diff --git a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_span.c b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_span.c
index 42f62a0cb..fa5b2d98f 100644
--- a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_span.c
+++ b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_span.c
@@ -44,6 +44,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/texformat.h"
 #include "main/renderbuffer.h"
 #include "main/samplerobj.h"
+#include "main/framebuffer.h"
 #include "swrast/swrast.h"
 #include "swrast/s_renderbuffer.h"
 
@@ -52,7 +53,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 
 static void
-radeon_renderbuffer_map(struct gl_context *ctx, struct gl_renderbuffer *rb)
+radeon_renderbuffer_map(struct gl_context *ctx,
+			struct gl_renderbuffer *rb,
+			bool flip_y)
 {
 	struct radeon_renderbuffer *rrb = radeon_renderbuffer(rb);
 	GLubyte *map;
@@ -63,7 +66,7 @@ radeon_renderbuffer_map(struct gl_context *ctx, struct gl_renderbuffer *rb)
 
 	ctx->Driver.MapRenderbuffer(ctx, rb, 0, 0, rb->Width, rb->Height,
 				    GL_MAP_READ_BIT | GL_MAP_WRITE_BIT,
-				    &map, &stride);
+				    &map, &stride, flip_y);
 
 	rrb->base.Map = map;
 	rrb->base.RowStride = stride;
@@ -95,9 +98,11 @@ radeon_map_framebuffer(struct gl_context *ctx, struct gl_framebuffer *fb)
 
 	/* check for render to textures */
 	for (i = 0; i < BUFFER_COUNT; i++)
-		radeon_renderbuffer_map(ctx, fb->Attachment[i].Renderbuffer);
+		radeon_renderbuffer_map(ctx, fb->Attachment[i].Renderbuffer,
+			fb->FlipY);
 
-	radeon_check_front_buffer_rendering(ctx);
+        if (_mesa_is_front_buffer_drawing(fb))
+		RADEON_CONTEXT(ctx)->front_buffer_dirty = true;
 }
 
 static void
@@ -113,7 +118,8 @@ radeon_unmap_framebuffer(struct gl_context *ctx, struct gl_framebuffer *fb)
 	for (i = 0; i < BUFFER_COUNT; i++)
 		radeon_renderbuffer_unmap(ctx, fb->Attachment[i].Renderbuffer);
 
-	radeon_check_front_buffer_rendering(ctx);
+        if (_mesa_is_front_buffer_drawing(fb))
+		RADEON_CONTEXT(ctx)->front_buffer_dirty = true;
 }
 
 static void radeonSpanRenderStart(struct gl_context * ctx)
diff --git a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_state_init.c b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_state_init.c
index 99c535a49..6a8cf606d 100644
--- a/lib/mesa/src/mesa/drivers/dri/radeon/radeon_state_init.c
+++ b/lib/mesa/src/mesa/drivers/dri/radeon/radeon_state_init.c
@@ -27,6 +27,7 @@
  *    Keith Whitwell <keithw@vmware.com>
  */
 
+#include "main/errors.h"
 #include "main/glheader.h"
 #include "main/imports.h"
 #include "main/api_arrayelt.h"
@@ -440,7 +441,7 @@ static void cube_emit_cs(struct gl_context *ctx, struct radeon_state_atom *atom)
 	case 2: base_reg = RADEON_PP_CUBIC_OFFSET_T2_0; break;
 	default:
 	case 0: base_reg = RADEON_PP_CUBIC_OFFSET_T0_0; break;
-   };
+   }
    BEGIN_BATCH(dwords);
    OUT_BATCH_TABLE(atom->cmd, 2);
    lvl = &t->mt->levels[0];
diff --git a/lib/mesa/src/mesa/drivers/dri/swrast/Makefile.am b/lib/mesa/src/mesa/drivers/dri/swrast/Makefile.am
index 7e1ea59e9..110c05605 100644
--- a/lib/mesa/src/mesa/drivers/dri/swrast/Makefile.am
+++ b/lib/mesa/src/mesa/drivers/dri/swrast/Makefile.am
@@ -38,3 +38,5 @@ AM_CFLAGS = \
 
 noinst_LTLIBRARIES = libswrast_dri.la
 libswrast_dri_la_SOURCES = $(SWRAST_C_FILES)
+
+EXTRA_DIST = meson.build
diff --git a/lib/mesa/src/mesa/drivers/dri/swrast/meson.build b/lib/mesa/src/mesa/drivers/dri/swrast/meson.build
new file mode 100644
index 000000000..5ceb1e33e
--- /dev/null
+++ b/lib/mesa/src/mesa/drivers/dri/swrast/meson.build
@@ -0,0 +1,30 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+libswrast_dri = static_library(
+  'swrast_dri',
+  files('swrast.c', 'swrast_priv.h'),
+  c_args : [c_vis_args],
+  include_directories : [inc_common, inc_dri_common],
+  dependencies : dep_libdrm,
+)
+
+dri_drivers += libswrast_dri
+dri_link += 'swrast_dri.so'
diff --git a/lib/mesa/src/mesa/drivers/osmesa/Makefile.am b/lib/mesa/src/mesa/drivers/osmesa/Makefile.am
index 2c8d4668b..f78169959 100644
--- a/lib/mesa/src/mesa/drivers/osmesa/Makefile.am
+++ b/lib/mesa/src/mesa/drivers/osmesa/Makefile.am
@@ -21,7 +21,7 @@
 # FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
 # IN THE SOFTWARE.
 
-EXTRA_DIST = osmesa.def SConscript
+EXTRA_DIST = osmesa.def SConscript meson.build
 
 AM_CPPFLAGS = \
 	-I$(top_srcdir)/include \
@@ -30,6 +30,8 @@ AM_CPPFLAGS = \
 	-I$(top_srcdir)/src/gallium/auxiliary \
 	-I$(top_builddir)/src/mapi \
 	-I$(top_srcdir)/src/mapi \
+	-I$(top_builddir)/src/mapi/glapi \
+	-I$(top_srcdir)/src/mapi/glapi \
 	-I$(top_srcdir)/src/mesa/ \
 	$(DEFINES)
 AM_CFLAGS = $(PTHREAD_CFLAGS) \
diff --git a/lib/mesa/src/mesa/drivers/osmesa/SConscript b/lib/mesa/src/mesa/drivers/osmesa/SConscript
index 728031446..b08a79eb8 100644
--- a/lib/mesa/src/mesa/drivers/osmesa/SConscript
+++ b/lib/mesa/src/mesa/drivers/osmesa/SConscript
@@ -7,6 +7,7 @@ env.Prepend(CPPPATH = [
     '#src/mapi',
     '#src/mesa',
     Dir('../../../mapi'), # src/mapi build path for python-generated GL API files/headers
+    Dir('../../../mapi/glapi'), # src/mapi/glapi build path
 ])
 
 env.Prepend(LIBS = [
@@ -14,6 +15,8 @@ env.Prepend(LIBS = [
     glapi,
     compiler,
     mesa,
+    spirv,
+    nir,
     glsl,
 ])
 
@@ -25,10 +28,8 @@ if env['platform'] == 'windows':
     env.AppendUnique(CPPDEFINES = [
         '_GDI32_', # prevent wgl* being declared __declspec(dllimport)
         'BUILD_GL32', # declare gl* as __declspec(dllexport) in Mesa headers
+        '_GLAPI_NO_EXPORTS', # prevent _glapi_* from being declared __declspec(dllimport)
     ])
-    if not env['gles']:
-        # prevent _glapi_* from being declared __declspec(dllimport)
-        env.Append(CPPDEFINES = ['_GLAPI_NO_EXPORTS'])
 
     sources += ['osmesa.def']
 
diff --git a/lib/mesa/src/mesa/drivers/osmesa/meson.build b/lib/mesa/src/mesa/drivers/osmesa/meson.build
new file mode 100644
index 000000000..a406bb3c2
--- /dev/null
+++ b/lib/mesa/src/mesa/drivers/osmesa/meson.build
@@ -0,0 +1,48 @@
+# Copyright © 2017 Dylan Baker
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+osmesa_link_with = []
+
+if with_shared_glapi
+  osmesa_link_with += libglapi
+endif
+
+libosmesa = shared_library(
+  osmesa_lib_name,
+  ['osmesa.c', glapitable_h, glapitemp_h],
+  c_args : c_vis_args,
+  cpp_args : cpp_vis_args,
+  link_args : ld_args_gc_sections,
+  include_directories : [
+    inc_include, inc_src, inc_mapi, inc_mesa, inc_gallium, inc_gallium_aux,
+  ],
+  link_with : [libmesa_classic, libglapi_static, osmesa_link_with],
+  dependencies : [dep_thread, dep_selinux],
+  version : '8.0.0',
+  install : true,
+)
+
+pkg.generate(
+  name : 'osmesa',
+  description : 'Mesa Off-screen Rendering Library',
+  version : '8.0.0',
+  libraries : libosmesa,
+  libraries_private : gl_priv_libs,
+)
diff --git a/lib/mesa/src/mesa/drivers/x11/SConscript b/lib/mesa/src/mesa/drivers/x11/SConscript
index 59c8df4b3..b097dcc59 100644
--- a/lib/mesa/src/mesa/drivers/x11/SConscript
+++ b/lib/mesa/src/mesa/drivers/x11/SConscript
@@ -21,6 +21,8 @@ env.Prepend(LIBS = [
     compiler,
     glsl,
     mesa,
+    spirv,
+    nir,
 ])
 
 sources = [
diff --git a/lib/mesa/src/mesa/drivers/x11/meson.build b/lib/mesa/src/mesa/drivers/x11/meson.build
new file mode 100644
index 000000000..cb092d25d
--- /dev/null
+++ b/lib/mesa/src/mesa/drivers/x11/meson.build
@@ -0,0 +1,39 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+gl_link_with = []
+if with_shared_glapi
+  gl_link_with += libglapi
+endif
+
+libgl = shared_library(
+  'GL',
+  files(
+    'fakeglx.c', 'glxapi.c', 'xfonts.c', 'xm_api.c', 'xm_buffer.c', 'xm_dd.c',
+    'xm_line.c', 'xm_tri.c',
+  ),
+  include_directories : [
+    inc_include, inc_src, inc_mesa, inc_mapi, inc_gallium, inc_gallium_aux
+  ],
+  link_with : [libmesa_classic, libglapi_static, libmesa_util, gl_link_with],
+  dependencies : [dep_x11, dep_xext, dep_xcb, dep_thread],
+  version : '1.6.0',
+  install : true,
+)
diff --git a/lib/mesa/src/mesa/drivers/x11/xm_buffer.c b/lib/mesa/src/mesa/drivers/x11/xm_buffer.c
index 4d52169d7..d945d8af5 100644
--- a/lib/mesa/src/mesa/drivers/x11/xm_buffer.c
+++ b/lib/mesa/src/mesa/drivers/x11/xm_buffer.c
@@ -31,6 +31,7 @@
 
 #include "glxheader.h"
 #include "xmesaP.h"
+#include "main/errors.h"
 #include "main/imports.h"
 #include "main/formats.h"
 #include "main/framebuffer.h"
@@ -422,7 +423,8 @@ xmesa_MapRenderbuffer(struct gl_context *ctx,
                       struct gl_renderbuffer *rb,
                       GLuint x, GLuint y, GLuint w, GLuint h,
                       GLbitfield mode,
-                      GLubyte **mapOut, GLint *rowStrideOut)
+                      GLubyte **mapOut, GLint *rowStrideOut,
+                      bool flip_y)
 {
    struct xmesa_renderbuffer *xrb = xmesa_renderbuffer(rb);
 
@@ -505,7 +507,7 @@ xmesa_MapRenderbuffer(struct gl_context *ctx,
 
    /* otherwise, this is an ordinary malloc-based renderbuffer */
    _swrast_map_soft_renderbuffer(ctx, rb, x, y, w, h, mode,
-                                 mapOut, rowStrideOut);
+                                 mapOut, rowStrideOut, false);
 }
 
 
diff --git a/lib/mesa/src/mesa/gl.pc.in b/lib/mesa/src/mesa/gl.pc.in
index 181724b97..680f74277 100644
--- a/lib/mesa/src/mesa/gl.pc.in
+++ b/lib/mesa/src/mesa/gl.pc.in
@@ -7,7 +7,7 @@ Name: gl
 Description: Mesa OpenGL library
 Requires.private: @GL_PC_REQ_PRIV@
 Version: @PACKAGE_VERSION@
-Libs: -L${libdir} -l@GL_LIB@
+Libs: -L${libdir} -l@GL_PKGCONF_LIB@
 Libs.private: @GL_PC_LIB_PRIV@
 Cflags: -I${includedir} @GL_PC_CFLAGS@
 glx_tls: @GLX_TLS@
diff --git a/lib/mesa/src/mesa/main/barrier.h b/lib/mesa/src/mesa/main/barrier.h
index 53ecf863f..acc15c677 100644
--- a/lib/mesa/src/mesa/main/barrier.h
+++ b/lib/mesa/src/mesa/main/barrier.h
@@ -53,4 +53,7 @@ _mesa_MemoryBarrierByRegion(GLbitfield barriers);
 void GLAPIENTRY
 _mesa_BlendBarrier(void);
 
+void GLAPIENTRY
+_mesa_FramebufferFetchBarrierEXT(void);
+
 #endif /* BARRIER_H */
diff --git a/lib/mesa/src/mesa/main/colormac.h b/lib/mesa/src/mesa/main/colormac.h
index 33ca5af07..8cd305cb3 100644
--- a/lib/mesa/src/mesa/main/colormac.h
+++ b/lib/mesa/src/mesa/main/colormac.h
@@ -35,7 +35,6 @@
 
 #include "config.h"
 #include "macros.h"
-#include "mtypes.h"
 
 
 /**
diff --git a/lib/mesa/src/mesa/main/colortab.c b/lib/mesa/src/mesa/main/colortab.c
index a8edb03dd..e8df73a0b 100644
--- a/lib/mesa/src/mesa/main/colortab.c
+++ b/lib/mesa/src/mesa/main/colortab.c
@@ -35,7 +35,6 @@
 #include "state.h"
 #include "teximage.h"
 #include "texstate.h"
-#include "main/dispatch.h"
 
 
 void GLAPIENTRY
diff --git a/lib/mesa/src/mesa/main/conservativeraster.c b/lib/mesa/src/mesa/main/conservativeraster.c
new file mode 100644
index 000000000..9068a00b4
--- /dev/null
+++ b/lib/mesa/src/mesa/main/conservativeraster.c
@@ -0,0 +1,128 @@
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 2018 Rhys Perry
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+/**
+ * \file conservativeraster.c
+ * glConservativeRasterParameteriNV and glConservativeRasterParameterfNV functions
+ */
+
+#include "conservativeraster.h"
+#include "context.h"
+#include "enums.h"
+
+static ALWAYS_INLINE void
+conservative_raster_parameter(GLenum pname, GLfloat param,
+                              bool no_error, const char *func)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (!no_error && !ctx->Extensions.NV_conservative_raster_dilate &&
+       !ctx->Extensions.NV_conservative_raster_pre_snap_triangles) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s not supported", func);
+      return;
+   }
+
+   if (MESA_VERBOSE & VERBOSE_API)
+      _mesa_debug(ctx, "%s(%s, %g)\n",
+                  func, _mesa_enum_to_string(pname), param);
+
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   switch (pname) {
+   case GL_CONSERVATIVE_RASTER_DILATE_NV:
+      if (!no_error && !ctx->Extensions.NV_conservative_raster_dilate)
+         goto invalid_pname_enum;
+
+      if (!no_error && param<0.0) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "%s(param=%g)", func, param);
+         return;
+      }
+      ctx->ConservativeRasterDilate =
+         CLAMP(param,
+               ctx->Const.ConservativeRasterDilateRange[0],
+               ctx->Const.ConservativeRasterDilateRange[1]);
+      break;
+   case GL_CONSERVATIVE_RASTER_MODE_NV:
+      if (!no_error && !ctx->Extensions.NV_conservative_raster_pre_snap_triangles)
+         goto invalid_pname_enum;
+
+      if (!no_error && param != GL_CONSERVATIVE_RASTER_MODE_POST_SNAP_NV &&
+          param != GL_CONSERVATIVE_RASTER_MODE_PRE_SNAP_TRIANGLES_NV) {
+         _mesa_error(ctx, GL_INVALID_ENUM,
+                     "%s(pname=%s)", func, _mesa_enum_to_string(param));
+         return;
+      }
+      ctx->ConservativeRasterMode = param;
+      break;
+   default:
+      goto invalid_pname_enum;
+      break;
+   }
+
+   FLUSH_VERTICES(ctx, 0);
+   ctx->NewDriverState |=
+      ctx->DriverFlags.NewNvConservativeRasterizationParams;
+
+   return;
+invalid_pname_enum:
+   if (!no_error)
+      _mesa_error(ctx, GL_INVALID_ENUM, "%s(pname=%s)",
+                  func, _mesa_enum_to_string(pname));
+}
+
+void GLAPIENTRY
+_mesa_ConservativeRasterParameteriNV_no_error(GLenum pname, GLint param)
+{
+   conservative_raster_parameter(pname, param, true, 
+                                 "glConservativeRasterParameteriNV");
+}
+
+void GLAPIENTRY
+_mesa_ConservativeRasterParameteriNV(GLenum pname, GLint param)
+{
+   conservative_raster_parameter(pname, param, false,
+                                 "glConservativeRasterParameteriNV");
+}
+
+void GLAPIENTRY
+_mesa_ConservativeRasterParameterfNV_no_error(GLenum pname, GLfloat param)
+{
+   conservative_raster_parameter(pname, param, true, 
+                                 "glConservativeRasterParameterfNV");
+}
+
+void GLAPIENTRY
+_mesa_ConservativeRasterParameterfNV(GLenum pname, GLfloat param)
+{
+   conservative_raster_parameter(pname, param, false, 
+                                 "glConservativeRasterParameterfNV");
+}
+
+void
+_mesa_init_conservative_raster(struct gl_context *ctx)
+{
+   ctx->ConservativeRasterDilate = 0.0;
+   ctx->ConservativeRasterMode = GL_CONSERVATIVE_RASTER_MODE_POST_SNAP_NV;
+}
diff --git a/lib/mesa/src/mesa/main/conservativeraster.h b/lib/mesa/src/mesa/main/conservativeraster.h
new file mode 100644
index 000000000..1865cfc2a
--- /dev/null
+++ b/lib/mesa/src/mesa/main/conservativeraster.h
@@ -0,0 +1,48 @@
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 2018 Rhys Perry
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+#ifndef CONSERVATIVERASTER_H
+#define CONSERVATIVERASTER_H
+
+#include "glheader.h"
+
+struct gl_context;
+
+extern void GLAPIENTRY
+_mesa_ConservativeRasterParameteriNV_no_error(GLenum pname, GLint param);
+
+extern void GLAPIENTRY
+_mesa_ConservativeRasterParameteriNV(GLenum pname, GLint param);
+
+extern void GLAPIENTRY
+_mesa_ConservativeRasterParameterfNV_no_error(GLenum pname, GLfloat param);
+
+extern void GLAPIENTRY
+_mesa_ConservativeRasterParameterfNV(GLenum pname, GLfloat param);
+
+extern void
+_mesa_init_conservative_raster(struct gl_context *ctx);
+
+#endif
diff --git a/lib/mesa/src/mesa/main/convolve.c b/lib/mesa/src/mesa/main/convolve.c
index 83d590f4a..e2c355c4f 100644
--- a/lib/mesa/src/mesa/main/convolve.c
+++ b/lib/mesa/src/mesa/main/convolve.c
@@ -34,7 +34,6 @@
 #include "glheader.h"
 #include "context.h"
 #include "convolve.h"
-#include "main/dispatch.h"
 
 
 void GLAPIENTRY
diff --git a/lib/mesa/src/mesa/main/debug_output.h b/lib/mesa/src/mesa/main/debug_output.h
index 9d8be4f22..8a5eedc21 100644
--- a/lib/mesa/src/mesa/main/debug_output.h
+++ b/lib/mesa/src/mesa/main/debug_output.h
@@ -29,15 +29,15 @@
 
 #include <stdio.h>
 #include <stdarg.h>
-#include "compiler.h"
 #include "glheader.h"
-#include "mtypes.h"
+#include "menums.h"
 
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
+struct gl_context;
 
 void
 _mesa_init_debug_output(struct gl_context *ctx);
diff --git a/lib/mesa/src/mesa/main/dlist.h b/lib/mesa/src/mesa/main/dlist.h
index 22b696f50..1ac17d5bb 100644
--- a/lib/mesa/src/mesa/main/dlist.h
+++ b/lib/mesa/src/mesa/main/dlist.h
@@ -33,8 +33,8 @@
 #define DLIST_H
 
 #include <stdio.h>
-#include "main/mtypes.h"
 
+struct gl_context;
 
 /**
  * Describes the location and size of a glBitmap image in a texture atlas.
diff --git a/lib/mesa/src/mesa/main/draw.c b/lib/mesa/src/mesa/main/draw.c
new file mode 100644
index 000000000..cceadf261
--- /dev/null
+++ b/lib/mesa/src/mesa/main/draw.c
@@ -0,0 +1,2262 @@
+/**************************************************************************
+ *
+ * Copyright 2003 VMware, Inc.
+ * Copyright 2009 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+#include <stdio.h>
+#include "arrayobj.h"
+#include "glheader.h"
+#include "c99_alloca.h"
+#include "context.h"
+#include "state.h"
+#include "draw.h"
+#include "draw_validate.h"
+#include "dispatch.h"
+#include "varray.h"
+#include "bufferobj.h"
+#include "enums.h"
+#include "macros.h"
+#include "transformfeedback.h"
+
+typedef struct {
+   GLuint count;
+   GLuint primCount;
+   GLuint first;
+   GLuint baseInstance;
+} DrawArraysIndirectCommand;
+
+typedef struct {
+   GLuint count;
+   GLuint primCount;
+   GLuint firstIndex;
+   GLint  baseVertex;
+   GLuint baseInstance;
+} DrawElementsIndirectCommand;
+
+
+/**
+ * Check that element 'j' of the array has reasonable data.
+ * Map VBO if needed.
+ * For debugging purposes; not normally used.
+ */
+static void
+check_array_data(struct gl_context *ctx, struct gl_vertex_array_object *vao,
+                 GLuint attrib, GLuint j)
+{
+   const struct gl_array_attributes *array = &vao->VertexAttrib[attrib];
+   if (array->Enabled) {
+      const struct gl_vertex_buffer_binding *binding =
+         &vao->BufferBinding[array->BufferBindingIndex];
+      struct gl_buffer_object *bo = binding->BufferObj;
+      const void *data = array->Ptr;
+      if (_mesa_is_bufferobj(bo)) {
+         if (!bo->Mappings[MAP_INTERNAL].Pointer) {
+            /* need to map now */
+            bo->Mappings[MAP_INTERNAL].Pointer =
+               ctx->Driver.MapBufferRange(ctx, 0, bo->Size,
+                                          GL_MAP_READ_BIT, bo, MAP_INTERNAL);
+         }
+         data = ADD_POINTERS(_mesa_vertex_attrib_address(array, binding),
+                             bo->Mappings[MAP_INTERNAL].Pointer);
+      }
+      switch (array->Type) {
+      case GL_FLOAT:
+         {
+            GLfloat *f = (GLfloat *) ((GLubyte *) data + binding->Stride * j);
+            GLint k;
+            for (k = 0; k < array->Size; k++) {
+               if (IS_INF_OR_NAN(f[k]) || f[k] >= 1.0e20F || f[k] <= -1.0e10F) {
+                  printf("Bad array data:\n");
+                  printf("  Element[%u].%u = %f\n", j, k, f[k]);
+                  printf("  Array %u at %p\n", attrib, (void *) array);
+                  printf("  Type 0x%x, Size %d, Stride %d\n",
+                         array->Type, array->Size, binding->Stride);
+                  printf("  Address/offset %p in Buffer Object %u\n",
+                         array->Ptr, bo->Name);
+                  f[k] = 1.0F;  /* XXX replace the bad value! */
+               }
+               /*assert(!IS_INF_OR_NAN(f[k])); */
+            }
+         }
+         break;
+      default:
+         ;
+      }
+   }
+}
+
+
+/**
+ * Unmap the buffer object referenced by given array, if mapped.
+ */
+static void
+unmap_array_buffer(struct gl_context *ctx, struct gl_vertex_array_object *vao,
+                   GLuint attrib)
+{
+   const struct gl_array_attributes *array = &vao->VertexAttrib[attrib];
+   if (array->Enabled) {
+      const struct gl_vertex_buffer_binding *binding =
+         &vao->BufferBinding[array->BufferBindingIndex];
+      struct gl_buffer_object *bo = binding->BufferObj;
+      if (_mesa_is_bufferobj(bo) && _mesa_bufferobj_mapped(bo, MAP_INTERNAL)) {
+         ctx->Driver.UnmapBuffer(ctx, bo, MAP_INTERNAL);
+      }
+   }
+}
+
+
+static inline int
+sizeof_ib_type(GLenum type)
+{
+   switch (type) {
+   case GL_UNSIGNED_INT:
+      return sizeof(GLuint);
+   case GL_UNSIGNED_SHORT:
+      return sizeof(GLushort);
+   case GL_UNSIGNED_BYTE:
+      return sizeof(GLubyte);
+   default:
+      assert(!"unsupported index data type");
+      /* In case assert is turned off */
+      return 0;
+   }
+}
+
+/**
+ * Examine the array's data for NaNs, etc.
+ * For debug purposes; not normally used.
+ */
+static void
+check_draw_elements_data(struct gl_context *ctx, GLsizei count,
+                         GLenum elemType, const void *elements,
+                         GLint basevertex)
+{
+   struct gl_vertex_array_object *vao = ctx->Array.VAO;
+   const void *elemMap;
+   GLint i;
+   GLuint k;
+
+   if (_mesa_is_bufferobj(vao->IndexBufferObj)) {
+      elemMap = ctx->Driver.MapBufferRange(ctx, 0,
+                                           vao->IndexBufferObj->Size,
+                                           GL_MAP_READ_BIT,
+                                           vao->IndexBufferObj, MAP_INTERNAL);
+      elements = ADD_POINTERS(elements, elemMap);
+   }
+
+   for (i = 0; i < count; i++) {
+      GLuint j;
+
+      /* j = element[i] */
+      switch (elemType) {
+      case GL_UNSIGNED_BYTE:
+         j = ((const GLubyte *) elements)[i];
+         break;
+      case GL_UNSIGNED_SHORT:
+         j = ((const GLushort *) elements)[i];
+         break;
+      case GL_UNSIGNED_INT:
+         j = ((const GLuint *) elements)[i];
+         break;
+      default:
+         unreachable("Unexpected index buffer type");
+      }
+
+      /* check element j of each enabled array */
+      for (k = 0; k < VERT_ATTRIB_MAX; k++) {
+         check_array_data(ctx, vao, k, j);
+      }
+   }
+
+   if (_mesa_is_bufferobj(vao->IndexBufferObj)) {
+      ctx->Driver.UnmapBuffer(ctx, vao->IndexBufferObj, MAP_INTERNAL);
+   }
+
+   for (k = 0; k < VERT_ATTRIB_MAX; k++) {
+      unmap_array_buffer(ctx, vao, k);
+   }
+}
+
+
+/**
+ * Check array data, looking for NaNs, etc.
+ */
+static void
+check_draw_arrays_data(struct gl_context *ctx, GLint start, GLsizei count)
+{
+   /* TO DO */
+}
+
+
+/**
+ * Check if we should skip the draw call even after validation was successful.
+ */
+static bool
+skip_validated_draw(struct gl_context *ctx)
+{
+   switch (ctx->API) {
+   case API_OPENGLES2:
+      /* For ES2, we can draw if we have a vertex program/shader). */
+      return ctx->VertexProgram._Current == NULL;
+
+   case API_OPENGLES:
+      /* For OpenGL ES, only draw if we have vertex positions
+       */
+      if (!ctx->Array.VAO->VertexAttrib[VERT_ATTRIB_POS].Enabled)
+         return true;
+      break;
+
+   case API_OPENGL_CORE:
+      /* Section 7.3 (Program Objects) of the OpenGL 4.5 Core Profile spec
+       * says:
+       *
+       *     "If there is no active program for the vertex or fragment shader
+       *     stages, the results of vertex and/or fragment processing will be
+       *     undefined. However, this is not an error."
+       *
+       * The fragment shader is not tested here because other state (e.g.,
+       * GL_RASTERIZER_DISCARD) affects whether or not we actually care.
+       */
+      return ctx->VertexProgram._Current == NULL;
+
+   case API_OPENGL_COMPAT:
+      if (ctx->VertexProgram._Current != NULL) {
+         /* Draw regardless of whether or not we have any vertex arrays.
+          * (Ex: could draw a point using a constant vertex pos)
+          */
+         return false;
+      } else {
+         /* Draw if we have vertex positions (GL_VERTEX_ARRAY or generic
+          * array [0]).
+          */
+         return (!ctx->Array.VAO->VertexAttrib[VERT_ATTRIB_POS].Enabled &&
+                 !ctx->Array.VAO->VertexAttrib[VERT_ATTRIB_GENERIC0].Enabled);
+      }
+      break;
+
+   default:
+      unreachable("Invalid API value in check_valid_to_render()");
+   }
+
+   return false;
+}
+
+
+/**
+ * Print info/data for glDrawArrays(), for debugging.
+ */
+static void
+print_draw_arrays(struct gl_context *ctx,
+                  GLenum mode, GLint start, GLsizei count)
+{
+   const struct gl_vertex_array_object *vao = ctx->Array.VAO;
+
+   printf("_mesa_exec_DrawArrays(mode 0x%x, start %d, count %d):\n",
+          mode, start, count);
+
+   unsigned i;
+   for (i = 0; i < VERT_ATTRIB_MAX; ++i) {
+      const struct gl_array_attributes *array = &vao->VertexAttrib[i];
+      if (!array->Enabled)
+         continue;
+
+      const struct gl_vertex_buffer_binding *binding =
+         &vao->BufferBinding[array->BufferBindingIndex];
+      struct gl_buffer_object *bufObj = binding->BufferObj;
+
+      printf("attr %s: size %d stride %d  enabled %d  "
+             "ptr %p  Bufobj %u\n",
+             gl_vert_attrib_name((gl_vert_attrib) i),
+             array->Size, binding->Stride, array->Enabled,
+             array->Ptr, bufObj->Name);
+
+      if (_mesa_is_bufferobj(bufObj)) {
+         GLubyte *p = ctx->Driver.MapBufferRange(ctx, 0, bufObj->Size,
+                                                 GL_MAP_READ_BIT, bufObj,
+                                                 MAP_INTERNAL);
+         int offset = (int) (GLintptr)
+            _mesa_vertex_attrib_address(array, binding);
+
+         unsigned multiplier;
+         switch (array->Type) {
+         case GL_DOUBLE:
+         case GL_INT64_ARB:
+         case GL_UNSIGNED_INT64_ARB:
+            multiplier = 2;
+            break;
+         default:
+            multiplier = 1;
+         }
+
+         float *f = (float *) (p + offset);
+         int *k = (int *) f;
+         int i = 0;
+         int n = (count - 1) * (binding->Stride / (4 * multiplier))
+           + array->Size;
+         if (n > 32)
+            n = 32;
+         printf("  Data at offset %d:\n", offset);
+         do {
+            if (multiplier == 2)
+               printf("    double[%d] = 0x%016llx %lf\n", i,
+                      ((unsigned long long *) k)[i], ((double *) f)[i]);
+            else
+               printf("    float[%d] = 0x%08x %f\n", i, k[i], f[i]);
+            i++;
+         } while (i < n);
+         ctx->Driver.UnmapBuffer(ctx, bufObj, MAP_INTERNAL);
+      }
+   }
+}
+
+
+/**
+ * Return a filter mask for the net enabled vao arrays.
+ * This is to mask out arrays that would otherwise supersed required current
+ * values for the fixed function shaders for example.
+ */
+static GLbitfield
+enabled_filter(const struct gl_context *ctx)
+{
+   switch (ctx->VertexProgram._VPMode) {
+   case VP_MODE_FF:
+      /* When no vertex program is active (or the vertex program is generated
+       * from fixed-function state).  We put the material values into the
+       * generic slots.  Since the vao has no material arrays, mute these
+       * slots from the enabled arrays so that the current material values
+       * are pulled instead of the vao arrays.
+       */
+      return VERT_BIT_FF_ALL;
+
+   case VP_MODE_SHADER:
+      /* There are no shaders in OpenGL ES 1.x, so this code path should be
+       * impossible to reach.  The meta code is careful to not use shaders in
+       * ES1.
+       */
+      assert(ctx->API != API_OPENGLES);
+
+      /* Other parts of the code assume that inputs[VERT_ATTRIB_POS] through
+       * inputs[VERT_ATTRIB_FF_MAX] will be non-NULL.  However, in OpenGL
+       * ES 2.0+ or OpenGL core profile, none of these arrays should ever
+       * be enabled.
+       */
+      if (ctx->API != API_OPENGL_COMPAT)
+         return VERT_BIT_GENERIC_ALL;
+
+      return VERT_BIT_ALL;
+
+   default:
+      assert(0);
+      return 0;
+   }
+}
+
+
+/**
+ * Helper function called by the other DrawArrays() functions below.
+ * This is where we handle primitive restart for drawing non-indexed
+ * arrays.  If primitive restart is enabled, it typically means
+ * splitting one DrawArrays() into two.
+ */
+static void
+_mesa_draw_arrays(struct gl_context *ctx, GLenum mode, GLint start,
+                  GLsizei count, GLuint numInstances, GLuint baseInstance,
+                  GLuint drawID)
+{
+   struct _mesa_prim prim;
+
+   if (skip_validated_draw(ctx))
+      return;
+
+   /* OpenGL 4.5 says that primitive restart is ignored with non-indexed
+    * draws.
+    */
+   memset(&prim, 0, sizeof(prim));
+   prim.begin = 1;
+   prim.end = 1;
+   prim.mode = mode;
+   prim.num_instances = numInstances;
+   prim.base_instance = baseInstance;
+   prim.draw_id = drawID;
+   prim.is_indirect = 0;
+   prim.start = start;
+   prim.count = count;
+
+   ctx->Driver.Draw(ctx, &prim, 1, NULL,
+                    GL_TRUE, start, start + count - 1, NULL, 0, NULL);
+
+   if (MESA_DEBUG_FLAGS & DEBUG_ALWAYS_FLUSH) {
+      _mesa_flush(ctx);
+   }
+}
+
+
+/**
+ * Execute a glRectf() function.
+ */
+static void GLAPIENTRY
+_mesa_exec_Rectf(GLfloat x1, GLfloat y1, GLfloat x2, GLfloat y2)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   CALL_Begin(GET_DISPATCH(), (GL_QUADS));
+   CALL_Vertex2f(GET_DISPATCH(), (x1, y1));
+   CALL_Vertex2f(GET_DISPATCH(), (x2, y1));
+   CALL_Vertex2f(GET_DISPATCH(), (x2, y2));
+   CALL_Vertex2f(GET_DISPATCH(), (x1, y2));
+   CALL_End(GET_DISPATCH(), ());
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_EvalMesh1(GLenum mode, GLint i1, GLint i2)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLint i;
+   GLfloat u, du;
+   GLenum prim;
+
+   switch (mode) {
+   case GL_POINT:
+      prim = GL_POINTS;
+      break;
+   case GL_LINE:
+      prim = GL_LINE_STRIP;
+      break;
+   default:
+      _mesa_error(ctx, GL_INVALID_ENUM, "glEvalMesh1(mode)");
+      return;
+   }
+
+   /* No effect if vertex maps disabled.
+    */
+   if (!ctx->Eval.Map1Vertex4 && !ctx->Eval.Map1Vertex3)
+      return;
+
+   du = ctx->Eval.MapGrid1du;
+   u = ctx->Eval.MapGrid1u1 + i1 * du;
+
+   CALL_Begin(GET_DISPATCH(), (prim));
+   for (i = i1; i <= i2; i++, u += du) {
+      CALL_EvalCoord1f(GET_DISPATCH(), (u));
+   }
+   CALL_End(GET_DISPATCH(), ());
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_EvalMesh2(GLenum mode, GLint i1, GLint i2, GLint j1, GLint j2)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLfloat u, du, v, dv, v1, u1;
+   GLint i, j;
+
+   switch (mode) {
+   case GL_POINT:
+   case GL_LINE:
+   case GL_FILL:
+      break;
+   default:
+      _mesa_error(ctx, GL_INVALID_ENUM, "glEvalMesh2(mode)");
+      return;
+   }
+
+   /* No effect if vertex maps disabled.
+    */
+   if (!ctx->Eval.Map2Vertex4 && !ctx->Eval.Map2Vertex3)
+      return;
+
+   du = ctx->Eval.MapGrid2du;
+   dv = ctx->Eval.MapGrid2dv;
+   v1 = ctx->Eval.MapGrid2v1 + j1 * dv;
+   u1 = ctx->Eval.MapGrid2u1 + i1 * du;
+
+   switch (mode) {
+   case GL_POINT:
+      CALL_Begin(GET_DISPATCH(), (GL_POINTS));
+      for (v = v1, j = j1; j <= j2; j++, v += dv) {
+         for (u = u1, i = i1; i <= i2; i++, u += du) {
+            CALL_EvalCoord2f(GET_DISPATCH(), (u, v));
+         }
+      }
+      CALL_End(GET_DISPATCH(), ());
+      break;
+   case GL_LINE:
+      for (v = v1, j = j1; j <= j2; j++, v += dv) {
+         CALL_Begin(GET_DISPATCH(), (GL_LINE_STRIP));
+         for (u = u1, i = i1; i <= i2; i++, u += du) {
+            CALL_EvalCoord2f(GET_DISPATCH(), (u, v));
+         }
+         CALL_End(GET_DISPATCH(), ());
+      }
+      for (u = u1, i = i1; i <= i2; i++, u += du) {
+         CALL_Begin(GET_DISPATCH(), (GL_LINE_STRIP));
+         for (v = v1, j = j1; j <= j2; j++, v += dv) {
+            CALL_EvalCoord2f(GET_DISPATCH(), (u, v));
+         }
+         CALL_End(GET_DISPATCH(), ());
+      }
+      break;
+   case GL_FILL:
+      for (v = v1, j = j1; j < j2; j++, v += dv) {
+         CALL_Begin(GET_DISPATCH(), (GL_TRIANGLE_STRIP));
+         for (u = u1, i = i1; i <= i2; i++, u += du) {
+            CALL_EvalCoord2f(GET_DISPATCH(), (u, v));
+            CALL_EvalCoord2f(GET_DISPATCH(), (u, v + dv));
+         }
+         CALL_End(GET_DISPATCH(), ());
+      }
+      break;
+   }
+}
+
+
+/**
+ * Called from glDrawArrays when in immediate mode (not display list mode).
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawArrays(GLenum mode, GLint start, GLsizei count)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawArrays(%s, %d, %d)\n",
+                  _mesa_enum_to_string(mode), start, count);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawArrays(ctx, mode, count))
+         return;
+   }
+
+   if (0)
+      check_draw_arrays_data(ctx, start, count);
+
+   _mesa_draw_arrays(ctx, mode, start, count, 1, 0, 0);
+
+   if (0)
+      print_draw_arrays(ctx, mode, start, count);
+}
+
+
+/**
+ * Called from glDrawArraysInstanced when in immediate mode (not
+ * display list mode).
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawArraysInstanced(GLenum mode, GLint start, GLsizei count,
+                               GLsizei numInstances)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawArraysInstanced(%s, %d, %d, %d)\n",
+                  _mesa_enum_to_string(mode), start, count, numInstances);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawArraysInstanced(ctx, mode, start, count,
+                                              numInstances))
+         return;
+   }
+
+   if (0)
+      check_draw_arrays_data(ctx, start, count);
+
+   _mesa_draw_arrays(ctx, mode, start, count, numInstances, 0, 0);
+
+   if (0)
+      print_draw_arrays(ctx, mode, start, count);
+}
+
+
+/**
+ * Called from glDrawArraysInstancedBaseInstance when in immediate mode.
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawArraysInstancedBaseInstance(GLenum mode, GLint first,
+                                           GLsizei count, GLsizei numInstances,
+                                           GLuint baseInstance)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx,
+                  "glDrawArraysInstancedBaseInstance(%s, %d, %d, %d, %d)\n",
+                  _mesa_enum_to_string(mode), first, count,
+                  numInstances, baseInstance);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawArraysInstanced(ctx, mode, first, count,
+                                              numInstances))
+         return;
+   }
+
+   if (0)
+      check_draw_arrays_data(ctx, first, count);
+
+   _mesa_draw_arrays(ctx, mode, first, count, numInstances, baseInstance, 0);
+
+   if (0)
+      print_draw_arrays(ctx, mode, first, count);
+}
+
+
+/**
+ * Called from glMultiDrawArrays when in immediate mode.
+ */
+static void GLAPIENTRY
+_mesa_exec_MultiDrawArrays(GLenum mode, const GLint *first,
+                           const GLsizei *count, GLsizei primcount)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLint i;
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx,
+                  "glMultiDrawArrays(%s, %p, %p, %d)\n",
+                  _mesa_enum_to_string(mode), first, count, primcount);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_MultiDrawArrays(ctx, mode, count, primcount))
+         return;
+   }
+
+   for (i = 0; i < primcount; i++) {
+      if (count[i] > 0) {
+         if (0)
+            check_draw_arrays_data(ctx, first[i], count[i]);
+
+         /* The GL_ARB_shader_draw_parameters spec adds the following after the
+          * pseudo-code describing glMultiDrawArrays:
+          *
+          *    "The index of the draw (<i> in the above pseudo-code) may be
+          *     read by a vertex shader as <gl_DrawIDARB>, as described in
+          *     Section 11.1.3.9."
+          */
+         _mesa_draw_arrays(ctx, mode, first[i], count[i], 1, 0, i);
+
+         if (0)
+            print_draw_arrays(ctx, mode, first[i], count[i]);
+      }
+   }
+}
+
+
+
+/**
+ * Map GL_ELEMENT_ARRAY_BUFFER and print contents.
+ * For debugging.
+ */
+#if 0
+static void
+dump_element_buffer(struct gl_context *ctx, GLenum type)
+{
+   const GLvoid *map =
+      ctx->Driver.MapBufferRange(ctx, 0,
+                                 ctx->Array.VAO->IndexBufferObj->Size,
+                                 GL_MAP_READ_BIT,
+                                 ctx->Array.VAO->IndexBufferObj,
+                                 MAP_INTERNAL);
+   switch (type) {
+   case GL_UNSIGNED_BYTE:
+      {
+         const GLubyte *us = (const GLubyte *) map;
+         GLint i;
+         for (i = 0; i < ctx->Array.VAO->IndexBufferObj->Size; i++) {
+            printf("%02x ", us[i]);
+            if (i % 32 == 31)
+               printf("\n");
+         }
+         printf("\n");
+      }
+      break;
+   case GL_UNSIGNED_SHORT:
+      {
+         const GLushort *us = (const GLushort *) map;
+         GLint i;
+         for (i = 0; i < ctx->Array.VAO->IndexBufferObj->Size / 2; i++) {
+            printf("%04x ", us[i]);
+            if (i % 16 == 15)
+               printf("\n");
+         }
+         printf("\n");
+      }
+      break;
+   case GL_UNSIGNED_INT:
+      {
+         const GLuint *us = (const GLuint *) map;
+         GLint i;
+         for (i = 0; i < ctx->Array.VAO->IndexBufferObj->Size / 4; i++) {
+            printf("%08x ", us[i]);
+            if (i % 8 == 7)
+               printf("\n");
+         }
+         printf("\n");
+      }
+      break;
+   default:
+      ;
+   }
+
+   ctx->Driver.UnmapBuffer(ctx, ctx->Array.VAO->IndexBufferObj, MAP_INTERNAL);
+}
+#endif
+
+
+static bool
+skip_draw_elements(struct gl_context *ctx, GLsizei count,
+                   const GLvoid *indices)
+{
+   if (count == 0)
+      return true;
+
+   /* Not using a VBO for indices, so avoid NULL pointer derefs later.
+    */
+   if (!_mesa_is_bufferobj(ctx->Array.VAO->IndexBufferObj) && indices == NULL)
+      return true;
+
+   if (skip_validated_draw(ctx))
+      return true;
+
+   return false;
+}
+
+
+/**
+ * Inner support for both _mesa_DrawElements and _mesa_DrawRangeElements.
+ * Do the rendering for a glDrawElements or glDrawRangeElements call after
+ * we've validated buffer bounds, etc.
+ */
+static void
+_mesa_validated_drawrangeelements(struct gl_context *ctx, GLenum mode,
+                                  GLboolean index_bounds_valid,
+                                  GLuint start, GLuint end,
+                                  GLsizei count, GLenum type,
+                                  const GLvoid * indices,
+                                  GLint basevertex, GLuint numInstances,
+                                  GLuint baseInstance)
+{
+   struct _mesa_index_buffer ib;
+   struct _mesa_prim prim;
+
+   if (!index_bounds_valid) {
+      assert(start == 0u);
+      assert(end == ~0u);
+   }
+
+   if (skip_draw_elements(ctx, count, indices))
+      return;
+
+   ib.count = count;
+   ib.index_size = sizeof_ib_type(type);
+   ib.obj = ctx->Array.VAO->IndexBufferObj;
+   ib.ptr = indices;
+
+   prim.begin = 1;
+   prim.end = 1;
+   prim.pad = 0;
+   prim.mode = mode;
+   prim.start = 0;
+   prim.count = count;
+   prim.indexed = 1;
+   prim.is_indirect = 0;
+   prim.basevertex = basevertex;
+   prim.num_instances = numInstances;
+   prim.base_instance = baseInstance;
+   prim.draw_id = 0;
+
+   /* Need to give special consideration to rendering a range of
+    * indices starting somewhere above zero.  Typically the
+    * application is issuing multiple DrawRangeElements() to draw
+    * successive primitives layed out linearly in the vertex arrays.
+    * Unless the vertex arrays are all in a VBO (or locked as with
+    * CVA), the OpenGL semantics imply that we need to re-read or
+    * re-upload the vertex data on each draw call.
+    *
+    * In the case of hardware tnl, we want to avoid starting the
+    * upload at zero, as it will mean every draw call uploads an
+    * increasing amount of not-used vertex data.  Worse - in the
+    * software tnl module, all those vertices might be transformed and
+    * lit but never rendered.
+    *
+    * If we just upload or transform the vertices in start..end,
+    * however, the indices will be incorrect.
+    *
+    * At this level, we don't know exactly what the requirements of
+    * the backend are going to be, though it will likely boil down to
+    * either:
+    *
+    * 1) Do nothing, everything is in a VBO and is processed once
+    *       only.
+    *
+    * 2) Adjust the indices and vertex arrays so that start becomes
+    *    zero.
+    *
+    * Rather than doing anything here, I'll provide a helper function
+    * for the latter case elsewhere.
+    */
+
+   ctx->Driver.Draw(ctx, &prim, 1, &ib,
+                    index_bounds_valid, start, end, NULL, 0, NULL);
+
+   if (MESA_DEBUG_FLAGS & DEBUG_ALWAYS_FLUSH) {
+      _mesa_flush(ctx);
+   }
+}
+
+
+/**
+ * Called by glDrawRangeElementsBaseVertex() in immediate mode.
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawRangeElementsBaseVertex(GLenum mode, GLuint start, GLuint end,
+                                       GLsizei count, GLenum type,
+                                       const GLvoid * indices, GLint basevertex)
+{
+   static GLuint warnCount = 0;
+   GLboolean index_bounds_valid = GL_TRUE;
+
+   /* This is only useful to catch invalid values in the "end" parameter
+    * like ~0.
+    */
+   GLuint max_element = 2 * 1000 * 1000 * 1000; /* just a big number */
+
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx,
+                  "glDrawRangeElementsBaseVertex(%s, %u, %u, %d, %s, %p, %d)\n",
+                  _mesa_enum_to_string(mode), start, end, count,
+                  _mesa_enum_to_string(type), indices, basevertex);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawRangeElements(ctx, mode, start, end, count,
+                                            type, indices))
+         return;
+   }
+
+   if ((int) end + basevertex < 0 || start + basevertex >= max_element) {
+      /* The application requested we draw using a range of indices that's
+       * outside the bounds of the current VBO.  This is invalid and appears
+       * to give undefined results.  The safest thing to do is to simply
+       * ignore the range, in case the application botched their range tracking
+       * but did provide valid indices.  Also issue a warning indicating that
+       * the application is broken.
+       */
+      if (warnCount++ < 10) {
+         _mesa_warning(ctx, "glDrawRangeElements(start %u, end %u, "
+                       "basevertex %d, count %d, type 0x%x, indices=%p):\n"
+                       "\trange is outside VBO bounds (max=%u); ignoring.\n"
+                       "\tThis should be fixed in the application.",
+                       start, end, basevertex, count, type, indices,
+                       max_element - 1);
+      }
+      index_bounds_valid = GL_FALSE;
+   }
+
+   /* NOTE: It's important that 'end' is a reasonable value.
+    * in _tnl_draw_prims(), we use end to determine how many vertices
+    * to transform.  If it's too large, we can unnecessarily split prims
+    * or we can read/write out of memory in several different places!
+    */
+
+   /* Catch/fix some potential user errors */
+   if (type == GL_UNSIGNED_BYTE) {
+      start = MIN2(start, 0xff);
+      end = MIN2(end, 0xff);
+   }
+   else if (type == GL_UNSIGNED_SHORT) {
+      start = MIN2(start, 0xffff);
+      end = MIN2(end, 0xffff);
+   }
+
+   if (0) {
+      printf("glDraw[Range]Elements{,BaseVertex}"
+             "(start %u, end %u, type 0x%x, count %d) ElemBuf %u, "
+             "base %d\n",
+             start, end, type, count,
+             ctx->Array.VAO->IndexBufferObj->Name, basevertex);
+   }
+
+   if ((int) start + basevertex < 0 || end + basevertex >= max_element)
+      index_bounds_valid = GL_FALSE;
+
+#if 0
+   check_draw_elements_data(ctx, count, type, indices, basevertex);
+#else
+   (void) check_draw_elements_data;
+#endif
+
+   if (!index_bounds_valid) {
+      start = 0;
+      end = ~0;
+   }
+
+   _mesa_validated_drawrangeelements(ctx, mode, index_bounds_valid, start, end,
+                                     count, type, indices, basevertex, 1, 0);
+}
+
+
+/**
+ * Called by glDrawRangeElements() in immediate mode.
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawRangeElements(GLenum mode, GLuint start, GLuint end,
+                             GLsizei count, GLenum type, const GLvoid * indices)
+{
+   if (MESA_VERBOSE & VERBOSE_DRAW) {
+      GET_CURRENT_CONTEXT(ctx);
+      _mesa_debug(ctx,
+                  "glDrawRangeElements(%s, %u, %u, %d, %s, %p)\n",
+                  _mesa_enum_to_string(mode), start, end, count,
+                  _mesa_enum_to_string(type), indices);
+   }
+
+   _mesa_exec_DrawRangeElementsBaseVertex(mode, start, end, count, type,
+                                          indices, 0);
+}
+
+
+/**
+ * Called by glDrawElements() in immediate mode.
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawElements(GLenum mode, GLsizei count, GLenum type,
+                        const GLvoid * indices)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawElements(%s, %u, %s, %p)\n",
+                  _mesa_enum_to_string(mode), count,
+                  _mesa_enum_to_string(type), indices);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawElements(ctx, mode, count, type, indices))
+         return;
+   }
+
+   _mesa_validated_drawrangeelements(ctx, mode, GL_FALSE, 0, ~0,
+                                     count, type, indices, 0, 1, 0);
+}
+
+
+/**
+ * Called by glDrawElementsBaseVertex() in immediate mode.
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawElementsBaseVertex(GLenum mode, GLsizei count, GLenum type,
+                                  const GLvoid * indices, GLint basevertex)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawElements(%s, %u, %s, %p)\n",
+                  _mesa_enum_to_string(mode), count,
+                  _mesa_enum_to_string(type), indices);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawElements(ctx, mode, count, type, indices))
+         return;
+   }
+
+   _mesa_validated_drawrangeelements(ctx, mode, GL_FALSE, 0, ~0,
+                                     count, type, indices, basevertex, 1, 0);
+}
+
+
+/**
+ * Called by glDrawElementsInstanced() in immediate mode.
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawElementsInstanced(GLenum mode, GLsizei count, GLenum type,
+                                 const GLvoid * indices, GLsizei numInstances)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawElements(%s, %u, %s, %p)\n",
+                  _mesa_enum_to_string(mode), count,
+                  _mesa_enum_to_string(type), indices);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawElementsInstanced(ctx, mode, count, type,
+                                                indices, numInstances))
+         return;
+   }
+
+   _mesa_validated_drawrangeelements(ctx, mode, GL_FALSE, 0, ~0,
+                                     count, type, indices, 0, numInstances, 0);
+}
+
+
+/**
+ * Called by glDrawElementsInstancedBaseVertex() in immediate mode.
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawElementsInstancedBaseVertex(GLenum mode, GLsizei count,
+                                           GLenum type, const GLvoid * indices,
+                                           GLsizei numInstances,
+                                           GLint basevertex)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx,
+                  "glDrawElementsInstancedBaseVertex"
+                  "(%s, %d, %s, %p, %d; %d)\n",
+                  _mesa_enum_to_string(mode), count,
+                  _mesa_enum_to_string(type), indices,
+                  numInstances, basevertex);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawElementsInstanced(ctx, mode, count, type,
+                                                indices, numInstances))
+         return;
+   }
+
+   _mesa_validated_drawrangeelements(ctx, mode, GL_FALSE, 0, ~0,
+                                     count, type, indices,
+                                     basevertex, numInstances, 0);
+}
+
+
+/**
+ * Called by glDrawElementsInstancedBaseInstance() in immediate mode.
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawElementsInstancedBaseInstance(GLenum mode, GLsizei count,
+                                             GLenum type,
+                                             const GLvoid *indices,
+                                             GLsizei numInstances,
+                                             GLuint baseInstance)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx,
+                  "glDrawElementsInstancedBaseInstance"
+                  "(%s, %d, %s, %p, %d, %d)\n",
+                  _mesa_enum_to_string(mode), count,
+                  _mesa_enum_to_string(type), indices,
+                  numInstances, baseInstance);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawElementsInstanced(ctx, mode, count, type,
+                                                indices, numInstances))
+         return;
+   }
+
+   _mesa_validated_drawrangeelements(ctx, mode, GL_FALSE, 0, ~0,
+                                     count, type, indices, 0, numInstances,
+                                     baseInstance);
+}
+
+
+/**
+ * Called by glDrawElementsInstancedBaseVertexBaseInstance() in immediate mode.
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawElementsInstancedBaseVertexBaseInstance(GLenum mode,
+                                                       GLsizei count,
+                                                       GLenum type,
+                                                       const GLvoid *indices,
+                                                       GLsizei numInstances,
+                                                       GLint basevertex,
+                                                       GLuint baseInstance)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx,
+                  "glDrawElementsInstancedBaseVertexBaseInstance"
+                  "(%s, %d, %s, %p, %d, %d, %d)\n",
+                  _mesa_enum_to_string(mode), count,
+                  _mesa_enum_to_string(type), indices,
+                  numInstances, basevertex, baseInstance);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawElementsInstanced(ctx, mode, count, type,
+                                                indices, numInstances))
+         return;
+   }
+
+   _mesa_validated_drawrangeelements(ctx, mode, GL_FALSE, 0, ~0,
+                                     count, type, indices, basevertex,
+                                     numInstances, baseInstance);
+}
+
+
+/**
+ * Inner support for both _mesa_MultiDrawElements() and
+ * _mesa_MultiDrawRangeElements().
+ * This does the actual rendering after we've checked array indexes, etc.
+ */
+static void
+_mesa_validated_multidrawelements(struct gl_context *ctx, GLenum mode,
+                                  const GLsizei *count, GLenum type,
+                                  const GLvoid * const *indices,
+                                  GLsizei primcount, const GLint *basevertex)
+{
+   struct _mesa_index_buffer ib;
+   struct _mesa_prim *prim;
+   unsigned int index_type_size = sizeof_ib_type(type);
+   uintptr_t min_index_ptr, max_index_ptr;
+   GLboolean fallback = GL_FALSE;
+   int i;
+
+   if (primcount == 0)
+      return;
+
+   prim = calloc(primcount, sizeof(*prim));
+   if (prim == NULL) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMultiDrawElements");
+      return;
+   }
+
+   min_index_ptr = (uintptr_t) indices[0];
+   max_index_ptr = 0;
+   for (i = 0; i < primcount; i++) {
+      min_index_ptr = MIN2(min_index_ptr, (uintptr_t) indices[i]);
+      max_index_ptr = MAX2(max_index_ptr, (uintptr_t) indices[i] +
+                           index_type_size * count[i]);
+   }
+
+   /* Check if we can handle this thing as a bunch of index offsets from the
+    * same index pointer.  If we can't, then we have to fall back to doing
+    * a draw_prims per primitive.
+    * Check that the difference between each prim's indexes is a multiple of
+    * the index/element size.
+    */
+   if (index_type_size != 1) {
+      for (i = 0; i < primcount; i++) {
+         if ((((uintptr_t) indices[i] - min_index_ptr) % index_type_size) !=
+             0) {
+            fallback = GL_TRUE;
+            break;
+         }
+      }
+   }
+
+   /* Draw primitives individually if one count is zero, so we can easily skip
+    * that primitive.
+    */
+   for (i = 0; i < primcount; i++) {
+      if (count[i] == 0) {
+         fallback = GL_TRUE;
+         break;
+      }
+   }
+
+   /* If the index buffer isn't in a VBO, then treating the application's
+    * subranges of the index buffer as one large index buffer may lead to
+    * us reading unmapped memory.
+    */
+   if (!_mesa_is_bufferobj(ctx->Array.VAO->IndexBufferObj))
+      fallback = GL_TRUE;
+
+   if (!fallback) {
+      ib.count = (max_index_ptr - min_index_ptr) / index_type_size;
+      ib.index_size = sizeof_ib_type(type);
+      ib.obj = ctx->Array.VAO->IndexBufferObj;
+      ib.ptr = (void *) min_index_ptr;
+
+      for (i = 0; i < primcount; i++) {
+         prim[i].begin = (i == 0);
+         prim[i].end = (i == primcount - 1);
+         prim[i].pad = 0;
+         prim[i].mode = mode;
+         prim[i].start =
+            ((uintptr_t) indices[i] - min_index_ptr) / index_type_size;
+         prim[i].count = count[i];
+         prim[i].indexed = 1;
+         prim[i].num_instances = 1;
+         prim[i].base_instance = 0;
+         prim[i].draw_id = i;
+         prim[i].is_indirect = 0;
+         if (basevertex != NULL)
+            prim[i].basevertex = basevertex[i];
+         else
+            prim[i].basevertex = 0;
+      }
+
+      ctx->Driver.Draw(ctx, prim, primcount, &ib,
+                       false, 0, ~0, NULL, 0, NULL);
+   }
+   else {
+      /* render one prim at a time */
+      for (i = 0; i < primcount; i++) {
+         if (count[i] == 0)
+            continue;
+         ib.count = count[i];
+         ib.index_size = sizeof_ib_type(type);
+         ib.obj = ctx->Array.VAO->IndexBufferObj;
+         ib.ptr = indices[i];
+
+         prim[0].begin = 1;
+         prim[0].end = 1;
+         prim[0].pad = 0;
+         prim[0].mode = mode;
+         prim[0].start = 0;
+         prim[0].count = count[i];
+         prim[0].indexed = 1;
+         prim[0].num_instances = 1;
+         prim[0].base_instance = 0;
+         prim[0].draw_id = i;
+         prim[0].is_indirect = 0;
+         if (basevertex != NULL)
+            prim[0].basevertex = basevertex[i];
+         else
+            prim[0].basevertex = 0;
+
+         ctx->Driver.Draw(ctx, prim, 1, &ib, false, 0, ~0, NULL, 0, NULL);
+      }
+   }
+
+   free(prim);
+
+   if (MESA_DEBUG_FLAGS & DEBUG_ALWAYS_FLUSH) {
+      _mesa_flush(ctx);
+   }
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_MultiDrawElements(GLenum mode,
+                             const GLsizei *count, GLenum type,
+                             const GLvoid * const *indices, GLsizei primcount)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (!_mesa_validate_MultiDrawElements(ctx, mode, count, type, indices,
+                                         primcount))
+      return;
+
+   if (skip_validated_draw(ctx))
+      return;
+
+   _mesa_validated_multidrawelements(ctx, mode, count, type, indices, primcount,
+                                     NULL);
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_MultiDrawElementsBaseVertex(GLenum mode,
+                                       const GLsizei *count, GLenum type,
+                                       const GLvoid * const *indices,
+                                       GLsizei primcount,
+                                       const GLsizei *basevertex)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_MultiDrawElements(ctx, mode, count, type, indices,
+                                            primcount))
+         return;
+   }
+
+   if (skip_validated_draw(ctx))
+      return;
+
+   _mesa_validated_multidrawelements(ctx, mode, count, type, indices, primcount,
+                                     basevertex);
+}
+
+
+/**
+ * Draw a GL primitive using a vertex count obtained from transform feedback.
+ * \param mode  the type of GL primitive to draw
+ * \param obj  the transform feedback object to use
+ * \param stream  index of the transform feedback stream from which to
+ *                get the primitive count.
+ * \param numInstances  number of instances to draw
+ */
+static void
+_mesa_draw_transform_feedback(struct gl_context *ctx, GLenum mode,
+                              struct gl_transform_feedback_object *obj,
+                              GLuint stream, GLuint numInstances)
+{
+   struct _mesa_prim prim;
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawTransformFeedback(ctx, mode, obj, stream,
+                                                numInstances)) {
+         return;
+      }
+   }
+
+   if (ctx->Driver.GetTransformFeedbackVertexCount &&
+       (ctx->Const.AlwaysUseGetTransformFeedbackVertexCount ||
+        !_mesa_all_varyings_in_vbos(ctx->Array.VAO))) {
+      GLsizei n =
+         ctx->Driver.GetTransformFeedbackVertexCount(ctx, obj, stream);
+      _mesa_draw_arrays(ctx, mode, 0, n, numInstances, 0, 0);
+      return;
+   }
+
+   if (skip_validated_draw(ctx))
+      return;
+
+   /* init most fields to zero */
+   memset(&prim, 0, sizeof(prim));
+   prim.begin = 1;
+   prim.end = 1;
+   prim.mode = mode;
+   prim.num_instances = numInstances;
+   prim.base_instance = 0;
+   prim.is_indirect = 0;
+
+   /* Maybe we should do some primitive splitting for primitive restart
+    * (like in DrawArrays), but we have no way to know how many vertices
+    * will be rendered. */
+
+   ctx->Driver.Draw(ctx, &prim, 1, NULL, GL_FALSE, 0, ~0, obj, stream, NULL);
+
+   if (MESA_DEBUG_FLAGS & DEBUG_ALWAYS_FLUSH) {
+      _mesa_flush(ctx);
+   }
+}
+
+
+/**
+ * Like DrawArrays, but take the count from a transform feedback object.
+ * \param mode  GL_POINTS, GL_LINES, GL_TRIANGLE_STRIP, etc.
+ * \param name  the transform feedback object
+ * User still has to setup of the vertex attribute info with
+ * glVertexPointer, glColorPointer, etc.
+ * Part of GL_ARB_transform_feedback2.
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawTransformFeedback(GLenum mode, GLuint name)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_transform_feedback_object *obj =
+      _mesa_lookup_transform_feedback_object(ctx, name);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawTransformFeedback(%s, %d)\n",
+                  _mesa_enum_to_string(mode), name);
+
+   _mesa_draw_transform_feedback(ctx, mode, obj, 0, 1);
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_DrawTransformFeedbackStream(GLenum mode, GLuint name, GLuint stream)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_transform_feedback_object *obj =
+      _mesa_lookup_transform_feedback_object(ctx, name);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawTransformFeedbackStream(%s, %u, %u)\n",
+                  _mesa_enum_to_string(mode), name, stream);
+
+   _mesa_draw_transform_feedback(ctx, mode, obj, stream, 1);
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_DrawTransformFeedbackInstanced(GLenum mode, GLuint name,
+                                          GLsizei primcount)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_transform_feedback_object *obj =
+      _mesa_lookup_transform_feedback_object(ctx, name);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawTransformFeedbackInstanced(%s, %d)\n",
+                  _mesa_enum_to_string(mode), name);
+
+   _mesa_draw_transform_feedback(ctx, mode, obj, 0, primcount);
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_DrawTransformFeedbackStreamInstanced(GLenum mode, GLuint name,
+                                                GLuint stream,
+                                                GLsizei primcount)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_transform_feedback_object *obj =
+      _mesa_lookup_transform_feedback_object(ctx, name);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawTransformFeedbackStreamInstanced"
+                  "(%s, %u, %u, %i)\n",
+                  _mesa_enum_to_string(mode), name, stream, primcount);
+
+   _mesa_draw_transform_feedback(ctx, mode, obj, stream, primcount);
+}
+
+
+static void
+_mesa_validated_drawarraysindirect(struct gl_context *ctx,
+                                   GLenum mode, const GLvoid *indirect)
+{
+   ctx->Driver.DrawIndirect(ctx, mode,
+                            ctx->DrawIndirectBuffer, (GLsizeiptr) indirect,
+                            1 /* draw_count */ , 16 /* stride */ ,
+                            NULL, 0, NULL);
+
+   if (MESA_DEBUG_FLAGS & DEBUG_ALWAYS_FLUSH)
+      _mesa_flush(ctx);
+}
+
+
+static void
+_mesa_validated_multidrawarraysindirect(struct gl_context *ctx,
+                                        GLenum mode,
+                                        const GLvoid *indirect,
+                                        GLsizei primcount, GLsizei stride)
+{
+   GLsizeiptr offset = (GLsizeiptr) indirect;
+
+   if (primcount == 0)
+      return;
+
+   ctx->Driver.DrawIndirect(ctx, mode, ctx->DrawIndirectBuffer, offset,
+                            primcount, stride, NULL, 0, NULL);
+
+   if (MESA_DEBUG_FLAGS & DEBUG_ALWAYS_FLUSH)
+      _mesa_flush(ctx);
+}
+
+
+static void
+_mesa_validated_drawelementsindirect(struct gl_context *ctx,
+                                     GLenum mode, GLenum type,
+                                     const GLvoid *indirect)
+{
+   struct _mesa_index_buffer ib;
+
+   ib.count = 0;                /* unknown */
+   ib.index_size = sizeof_ib_type(type);
+   ib.obj = ctx->Array.VAO->IndexBufferObj;
+   ib.ptr = NULL;
+
+   ctx->Driver.DrawIndirect(ctx, mode,
+                            ctx->DrawIndirectBuffer, (GLsizeiptr) indirect,
+                            1 /* draw_count */ , 20 /* stride */ ,
+                            NULL, 0, &ib);
+
+   if (MESA_DEBUG_FLAGS & DEBUG_ALWAYS_FLUSH)
+      _mesa_flush(ctx);
+}
+
+
+static void
+_mesa_validated_multidrawelementsindirect(struct gl_context *ctx,
+                                          GLenum mode, GLenum type,
+                                          const GLvoid *indirect,
+                                          GLsizei primcount, GLsizei stride)
+{
+   struct _mesa_index_buffer ib;
+   GLsizeiptr offset = (GLsizeiptr) indirect;
+
+   if (primcount == 0)
+      return;
+
+   /* NOTE: IndexBufferObj is guaranteed to be a VBO. */
+
+   ib.count = 0;                /* unknown */
+   ib.index_size = sizeof_ib_type(type);
+   ib.obj = ctx->Array.VAO->IndexBufferObj;
+   ib.ptr = NULL;
+
+   ctx->Driver.DrawIndirect(ctx, mode,
+                            ctx->DrawIndirectBuffer, offset,
+                            primcount, stride, NULL, 0, &ib);
+
+   if (MESA_DEBUG_FLAGS & DEBUG_ALWAYS_FLUSH)
+      _mesa_flush(ctx);
+}
+
+
+/**
+ * Like [Multi]DrawArrays/Elements, but they take most arguments from
+ * a buffer object.
+ */
+static void GLAPIENTRY
+_mesa_exec_DrawArraysIndirect(GLenum mode, const GLvoid *indirect)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawArraysIndirect(%s, %p)\n",
+                  _mesa_enum_to_string(mode), indirect);
+
+   /* From the ARB_draw_indirect spec:
+    *
+    *    "Initially zero is bound to DRAW_INDIRECT_BUFFER. In the
+    *    compatibility profile, this indicates that DrawArraysIndirect and
+    *    DrawElementsIndirect are to source their arguments directly from the
+    *    pointer passed as their <indirect> parameters."
+    */
+   if (ctx->API == API_OPENGL_COMPAT &&
+       !_mesa_is_bufferobj(ctx->DrawIndirectBuffer)) {
+      DrawArraysIndirectCommand *cmd = (DrawArraysIndirectCommand *) indirect;
+
+      _mesa_exec_DrawArraysInstancedBaseInstance(mode, cmd->first, cmd->count,
+                                                 cmd->primCount,
+                                                 cmd->baseInstance);
+      return;
+   }
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawArraysIndirect(ctx, mode, indirect))
+         return;
+   }
+
+   if (skip_validated_draw(ctx))
+      return;
+
+   _mesa_validated_drawarraysindirect(ctx, mode, indirect);
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_DrawElementsIndirect(GLenum mode, GLenum type, const GLvoid *indirect)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glDrawElementsIndirect(%s, %s, %p)\n",
+                  _mesa_enum_to_string(mode),
+                  _mesa_enum_to_string(type), indirect);
+
+   /* From the ARB_draw_indirect spec:
+    *
+    *    "Initially zero is bound to DRAW_INDIRECT_BUFFER. In the
+    *    compatibility profile, this indicates that DrawArraysIndirect and
+    *    DrawElementsIndirect are to source their arguments directly from the
+    *    pointer passed as their <indirect> parameters."
+    */
+   if (ctx->API == API_OPENGL_COMPAT &&
+       !_mesa_is_bufferobj(ctx->DrawIndirectBuffer)) {
+      /*
+       * Unlike regular DrawElementsInstancedBaseVertex commands, the indices
+       * may not come from a client array and must come from an index buffer.
+       * If no element array buffer is bound, an INVALID_OPERATION error is
+       * generated.
+       */
+      if (!_mesa_is_bufferobj(ctx->Array.VAO->IndexBufferObj)) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "glDrawElementsIndirect(no buffer bound "
+                     "to GL_ELEMENT_ARRAY_BUFFER)");
+      } else {
+         DrawElementsIndirectCommand *cmd =
+            (DrawElementsIndirectCommand *) indirect;
+
+         /* Convert offset to pointer */
+         void *offset = (void *)
+            ((cmd->firstIndex * _mesa_sizeof_type(type)) & 0xffffffffUL);
+
+         _mesa_exec_DrawElementsInstancedBaseVertexBaseInstance(mode, cmd->count,
+                                                                type, offset,
+                                                                cmd->primCount,
+                                                                cmd->baseVertex,
+                                                                cmd->baseInstance);
+      }
+
+      return;
+   }
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_DrawElementsIndirect(ctx, mode, type, indirect))
+         return;
+   }
+
+   if (skip_validated_draw(ctx))
+      return;
+
+   _mesa_validated_drawelementsindirect(ctx, mode, type, indirect);
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_MultiDrawArraysIndirect(GLenum mode, const GLvoid *indirect,
+                                   GLsizei primcount, GLsizei stride)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glMultiDrawArraysIndirect(%s, %p, %i, %i)\n",
+                  _mesa_enum_to_string(mode), indirect, primcount, stride);
+
+   /* If <stride> is zero, the array elements are treated as tightly packed. */
+   if (stride == 0)
+      stride = sizeof(DrawArraysIndirectCommand);
+
+   /* From the ARB_draw_indirect spec:
+    *
+    *    "Initially zero is bound to DRAW_INDIRECT_BUFFER. In the
+    *    compatibility profile, this indicates that DrawArraysIndirect and
+    *    DrawElementsIndirect are to source their arguments directly from the
+    *    pointer passed as their <indirect> parameters."
+    */
+   if (ctx->API == API_OPENGL_COMPAT &&
+       !_mesa_is_bufferobj(ctx->DrawIndirectBuffer)) {
+
+      if (!_mesa_valid_draw_indirect_multi(ctx, primcount, stride,
+                                           "glMultiDrawArraysIndirect"))
+         return;
+
+      const ubyte *ptr = (const ubyte *) indirect;
+      for (unsigned i = 0; i < primcount; i++) {
+         DrawArraysIndirectCommand *cmd = (DrawArraysIndirectCommand *) ptr;
+         _mesa_exec_DrawArraysInstancedBaseInstance(mode, cmd->first,
+                                                    cmd->count, cmd->primCount,
+                                                    cmd->baseInstance);
+
+         if (stride == 0) {
+            ptr += sizeof(DrawArraysIndirectCommand);
+         } else {
+            ptr += stride;
+         }
+      }
+
+      return;
+   }
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_MultiDrawArraysIndirect(ctx, mode, indirect,
+                                                  primcount, stride))
+         return;
+   }
+
+   if (skip_validated_draw(ctx))
+      return;
+
+   _mesa_validated_multidrawarraysindirect(ctx, mode, indirect,
+                                           primcount, stride);
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_MultiDrawElementsIndirect(GLenum mode, GLenum type,
+                                     const GLvoid *indirect,
+                                     GLsizei primcount, GLsizei stride)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glMultiDrawElementsIndirect(%s, %s, %p, %i, %i)\n",
+                  _mesa_enum_to_string(mode),
+                  _mesa_enum_to_string(type), indirect, primcount, stride);
+
+   /* If <stride> is zero, the array elements are treated as tightly packed. */
+   if (stride == 0)
+      stride = sizeof(DrawElementsIndirectCommand);
+
+
+   /* From the ARB_draw_indirect spec:
+    *
+    *    "Initially zero is bound to DRAW_INDIRECT_BUFFER. In the
+    *    compatibility profile, this indicates that DrawArraysIndirect and
+    *    DrawElementsIndirect are to source their arguments directly from the
+    *    pointer passed as their <indirect> parameters."
+    */
+   if (ctx->API == API_OPENGL_COMPAT &&
+       !_mesa_is_bufferobj(ctx->DrawIndirectBuffer)) {
+      /*
+       * Unlike regular DrawElementsInstancedBaseVertex commands, the indices
+       * may not come from a client array and must come from an index buffer.
+       * If no element array buffer is bound, an INVALID_OPERATION error is
+       * generated.
+       */
+      if (!_mesa_is_bufferobj(ctx->Array.VAO->IndexBufferObj)) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "glMultiDrawElementsIndirect(no buffer bound "
+                     "to GL_ELEMENT_ARRAY_BUFFER)");
+
+         return;
+      }
+
+      if (!_mesa_valid_draw_indirect_multi(ctx, primcount, stride,
+                                           "glMultiDrawArraysIndirect"))
+         return;
+
+      const ubyte *ptr = (const ubyte *) indirect;
+      for (unsigned i = 0; i < primcount; i++) {
+         _mesa_exec_DrawElementsIndirect(mode, type, ptr);
+
+         if (stride == 0) {
+            ptr += sizeof(DrawElementsIndirectCommand);
+         } else {
+            ptr += stride;
+         }
+      }
+
+      return;
+   }
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_MultiDrawElementsIndirect(ctx, mode, type, indirect,
+                                                    primcount, stride))
+         return;
+   }
+
+   if (skip_validated_draw(ctx))
+      return;
+
+   _mesa_validated_multidrawelementsindirect(ctx, mode, type, indirect,
+                                             primcount, stride);
+}
+
+
+static void
+_mesa_validated_multidrawarraysindirectcount(struct gl_context *ctx,
+                                             GLenum mode,
+                                             GLintptr indirect,
+                                             GLintptr drawcount_offset,
+                                             GLsizei maxdrawcount,
+                                             GLsizei stride)
+{
+   GLsizeiptr offset = indirect;
+
+   if (maxdrawcount == 0)
+      return;
+
+   ctx->Driver.DrawIndirect(ctx, mode,
+                            ctx->DrawIndirectBuffer, offset,
+                            maxdrawcount, stride,
+                            ctx->ParameterBuffer, drawcount_offset, NULL);
+
+   if (MESA_DEBUG_FLAGS & DEBUG_ALWAYS_FLUSH)
+      _mesa_flush(ctx);
+}
+
+
+static void
+_mesa_validated_multidrawelementsindirectcount(struct gl_context *ctx,
+                                               GLenum mode, GLenum type,
+                                               GLintptr indirect,
+                                               GLintptr drawcount_offset,
+                                               GLsizei maxdrawcount,
+                                               GLsizei stride)
+{
+   struct _mesa_index_buffer ib;
+   GLsizeiptr offset = (GLsizeiptr) indirect;
+
+   if (maxdrawcount == 0)
+      return;
+
+   /* NOTE: IndexBufferObj is guaranteed to be a VBO. */
+
+   ib.count = 0;                /* unknown */
+   ib.index_size = sizeof_ib_type(type);
+   ib.obj = ctx->Array.VAO->IndexBufferObj;
+   ib.ptr = NULL;
+
+   ctx->Driver.DrawIndirect(ctx, mode,
+                            ctx->DrawIndirectBuffer, offset,
+                            maxdrawcount, stride,
+                            ctx->ParameterBuffer, drawcount_offset, &ib);
+
+   if (MESA_DEBUG_FLAGS & DEBUG_ALWAYS_FLUSH)
+      _mesa_flush(ctx);
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_MultiDrawArraysIndirectCount(GLenum mode, GLintptr indirect,
+                                        GLintptr drawcount_offset,
+                                        GLsizei maxdrawcount, GLsizei stride)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glMultiDrawArraysIndirectCountARB"
+                  "(%s, %lx, %lx, %i, %i)\n",
+                  _mesa_enum_to_string(mode),
+                  (unsigned long) indirect, (unsigned long) drawcount_offset,
+                  maxdrawcount, stride);
+
+   /* If <stride> is zero, the array elements are treated as tightly packed. */
+   if (stride == 0)
+      stride = 4 * sizeof(GLuint);      /* sizeof(DrawArraysIndirectCommand) */
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_MultiDrawArraysIndirectCount(ctx, mode,
+                                                       indirect,
+                                                       drawcount_offset,
+                                                       maxdrawcount, stride))
+         return;
+   }
+
+   if (skip_validated_draw(ctx))
+      return;
+
+   _mesa_validated_multidrawarraysindirectcount(ctx, mode, indirect,
+                                                drawcount_offset,
+                                                maxdrawcount, stride);
+}
+
+
+static void GLAPIENTRY
+_mesa_exec_MultiDrawElementsIndirectCount(GLenum mode, GLenum type,
+                                          GLintptr indirect,
+                                          GLintptr drawcount_offset,
+                                          GLsizei maxdrawcount, GLsizei stride)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (MESA_VERBOSE & VERBOSE_DRAW)
+      _mesa_debug(ctx, "glMultiDrawElementsIndirectCountARB"
+                  "(%s, %s, %lx, %lx, %i, %i)\n",
+                  _mesa_enum_to_string(mode), _mesa_enum_to_string(type),
+                  (unsigned long) indirect, (unsigned long) drawcount_offset,
+                  maxdrawcount, stride);
+
+   /* If <stride> is zero, the array elements are treated as tightly packed. */
+   if (stride == 0)
+      stride = 5 * sizeof(GLuint);      /* sizeof(DrawElementsIndirectCommand) */
+
+   FLUSH_FOR_DRAW(ctx);
+
+   _mesa_set_draw_vao(ctx, ctx->Array.VAO, enabled_filter(ctx));
+
+   if (_mesa_is_no_error_enabled(ctx)) {
+      if (ctx->NewState)
+         _mesa_update_state(ctx);
+   } else {
+      if (!_mesa_validate_MultiDrawElementsIndirectCount(ctx, mode, type,
+                                                         indirect,
+                                                         drawcount_offset,
+                                                         maxdrawcount, stride))
+         return;
+   }
+
+   if (skip_validated_draw(ctx))
+      return;
+
+   _mesa_validated_multidrawelementsindirectcount(ctx, mode, type, indirect,
+                                                  drawcount_offset, maxdrawcount,
+                                                  stride);
+}
+
+
+/**
+ * Initialize the dispatch table with the VBO functions for drawing.
+ */
+void
+_mesa_initialize_exec_dispatch(const struct gl_context *ctx,
+                               struct _glapi_table *exec)
+{
+   SET_DrawArrays(exec, _mesa_exec_DrawArrays);
+   SET_DrawElements(exec, _mesa_exec_DrawElements);
+
+   if (_mesa_is_desktop_gl(ctx) || _mesa_is_gles3(ctx)) {
+      SET_DrawRangeElements(exec, _mesa_exec_DrawRangeElements);
+   }
+
+   SET_MultiDrawArrays(exec, _mesa_exec_MultiDrawArrays);
+   SET_MultiDrawElementsEXT(exec, _mesa_exec_MultiDrawElements);
+
+   if (ctx->API == API_OPENGL_COMPAT) {
+      SET_Rectf(exec, _mesa_exec_Rectf);
+      SET_EvalMesh1(exec, _mesa_exec_EvalMesh1);
+      SET_EvalMesh2(exec, _mesa_exec_EvalMesh2);
+   }
+
+   if (ctx->API != API_OPENGLES &&
+       ctx->Extensions.ARB_draw_elements_base_vertex) {
+      SET_DrawElementsBaseVertex(exec, _mesa_exec_DrawElementsBaseVertex);
+      SET_MultiDrawElementsBaseVertex(exec,
+                                      _mesa_exec_MultiDrawElementsBaseVertex);
+
+      if (_mesa_is_desktop_gl(ctx) || _mesa_is_gles3(ctx)) {
+         SET_DrawRangeElementsBaseVertex(exec,
+                                         _mesa_exec_DrawRangeElementsBaseVertex);
+         SET_DrawElementsInstancedBaseVertex(exec,
+                                             _mesa_exec_DrawElementsInstancedBaseVertex);
+      }
+   }
+
+   if (_mesa_is_desktop_gl(ctx) || _mesa_is_gles3(ctx)) {
+      SET_DrawArraysInstancedBaseInstance(exec,
+                                          _mesa_exec_DrawArraysInstancedBaseInstance);
+      SET_DrawElementsInstancedBaseInstance(exec,
+                                            _mesa_exec_DrawElementsInstancedBaseInstance);
+      SET_DrawElementsInstancedBaseVertexBaseInstance(exec,
+                                                      _mesa_exec_DrawElementsInstancedBaseVertexBaseInstance);
+   }
+
+   if (_mesa_is_desktop_gl(ctx) || _mesa_is_gles31(ctx)) {
+      SET_DrawArraysIndirect(exec, _mesa_exec_DrawArraysIndirect);
+      SET_DrawElementsIndirect(exec, _mesa_exec_DrawElementsIndirect);
+   }
+
+   if (_mesa_is_desktop_gl(ctx) || _mesa_is_gles3(ctx)) {
+      SET_DrawArraysInstancedARB(exec, _mesa_exec_DrawArraysInstanced);
+      SET_DrawElementsInstancedARB(exec, _mesa_exec_DrawElementsInstanced);
+   }
+
+   if (_mesa_is_desktop_gl(ctx)) {
+      SET_DrawTransformFeedback(exec, _mesa_exec_DrawTransformFeedback);
+      SET_DrawTransformFeedbackStream(exec,
+                                      _mesa_exec_DrawTransformFeedbackStream);
+      SET_DrawTransformFeedbackInstanced(exec,
+                                         _mesa_exec_DrawTransformFeedbackInstanced);
+      SET_DrawTransformFeedbackStreamInstanced(exec,
+                                               _mesa_exec_DrawTransformFeedbackStreamInstanced);
+      SET_MultiDrawArraysIndirect(exec, _mesa_exec_MultiDrawArraysIndirect);
+      SET_MultiDrawElementsIndirect(exec, _mesa_exec_MultiDrawElementsIndirect);
+      SET_MultiDrawArraysIndirectCountARB(exec,
+                                          _mesa_exec_MultiDrawArraysIndirectCount);
+      SET_MultiDrawElementsIndirectCountARB(exec,
+                                            _mesa_exec_MultiDrawElementsIndirectCount);
+   }
+}
+
+
+
+/**
+ * The following functions are only used for OpenGL ES 1/2 support.
+ * And some aren't even supported (yet) in ES 1/2.
+ */
+
+
+void GLAPIENTRY
+_mesa_DrawArrays(GLenum mode, GLint first, GLsizei count)
+{
+   _mesa_exec_DrawArrays(mode, first, count);
+}
+
+
+void GLAPIENTRY
+_mesa_DrawArraysInstanced(GLenum mode, GLint first, GLsizei count,
+                          GLsizei primcount)
+{
+   _mesa_exec_DrawArraysInstanced(mode, first, count, primcount);
+}
+
+
+void GLAPIENTRY
+_mesa_DrawElements(GLenum mode, GLsizei count, GLenum type,
+                   const GLvoid *indices)
+{
+   _mesa_exec_DrawElements(mode, count, type, indices);
+}
+
+
+void GLAPIENTRY
+_mesa_DrawElementsBaseVertex(GLenum mode, GLsizei count, GLenum type,
+                             const GLvoid *indices, GLint basevertex)
+{
+   _mesa_exec_DrawElementsBaseVertex(mode, count, type, indices, basevertex);
+}
+
+
+void GLAPIENTRY
+_mesa_DrawRangeElements(GLenum mode, GLuint start, GLuint end, GLsizei count,
+                        GLenum type, const GLvoid * indices)
+{
+   _mesa_exec_DrawRangeElements(mode, start, end, count, type, indices);
+}
+
+
+void GLAPIENTRY
+_mesa_DrawRangeElementsBaseVertex(GLenum mode, GLuint start, GLuint end,
+                                  GLsizei count, GLenum type,
+                                  const GLvoid *indices, GLint basevertex)
+{
+   _mesa_exec_DrawRangeElementsBaseVertex(mode, start, end, count, type,
+                                          indices, basevertex);
+}
+
+
+void GLAPIENTRY
+_mesa_MultiDrawElementsEXT(GLenum mode, const GLsizei *count, GLenum type,
+                           const GLvoid ** indices, GLsizei primcount)
+{
+   _mesa_exec_MultiDrawElements(mode, count, type, indices, primcount);
+}
+
+
+void GLAPIENTRY
+_mesa_MultiDrawElementsBaseVertex(GLenum mode,
+                                  const GLsizei *count, GLenum type,
+                                  const GLvoid **indices, GLsizei primcount,
+                                  const GLint *basevertex)
+{
+   _mesa_exec_MultiDrawElementsBaseVertex(mode, count, type, indices,
+                                          primcount, basevertex);
+}
+
+
+void GLAPIENTRY
+_mesa_DrawTransformFeedback(GLenum mode, GLuint name)
+{
+   _mesa_exec_DrawTransformFeedback(mode, name);
+}
+
+
+/* GL_IBM_multimode_draw_arrays */
+void GLAPIENTRY
+_mesa_MultiModeDrawArraysIBM( const GLenum * mode, const GLint * first,
+                              const GLsizei * count,
+                              GLsizei primcount, GLint modestride )
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLint i;
+
+   FLUSH_VERTICES(ctx, 0);
+
+   for ( i = 0 ; i < primcount ; i++ ) {
+      if ( count[i] > 0 ) {
+         GLenum m = *((GLenum *) ((GLubyte *) mode + i * modestride));
+         CALL_DrawArrays(ctx->CurrentServerDispatch, ( m, first[i], count[i] ));
+      }
+   }
+}
+
+
+/* GL_IBM_multimode_draw_arrays */
+void GLAPIENTRY
+_mesa_MultiModeDrawElementsIBM( const GLenum * mode, const GLsizei * count,
+                                GLenum type, const GLvoid * const * indices,
+                                GLsizei primcount, GLint modestride )
+{
+   GET_CURRENT_CONTEXT(ctx);
+   GLint i;
+
+   FLUSH_VERTICES(ctx, 0);
+
+   /* XXX not sure about ARB_vertex_buffer_object handling here */
+
+   for ( i = 0 ; i < primcount ; i++ ) {
+      if ( count[i] > 0 ) {
+         GLenum m = *((GLenum *) ((GLubyte *) mode + i * modestride));
+         CALL_DrawElements(ctx->CurrentServerDispatch, ( m, count[i], type,
+                                                         indices[i] ));
+      }
+   }
+}
+
+
+/*
+ * Helper function for _mesa_draw_indirect below that additionally takes a zero
+ * initialized array of _mesa_prim scratch space memory as the last argument.
+ */
+static void
+draw_indirect(struct gl_context *ctx, GLuint mode,
+              struct gl_buffer_object *indirect_data,
+              GLsizeiptr indirect_offset, unsigned draw_count,
+              unsigned stride,
+              struct gl_buffer_object *indirect_draw_count_buffer,
+              GLsizeiptr indirect_draw_count_offset,
+              const struct _mesa_index_buffer *ib,
+              struct _mesa_prim *prim)
+{
+   prim[0].begin = 1;
+   prim[draw_count - 1].end = 1;
+   for (unsigned i = 0; i < draw_count; ++i, indirect_offset += stride) {
+      prim[i].mode = mode;
+      prim[i].indexed = !!ib;
+      prim[i].indirect_offset = indirect_offset;
+      prim[i].is_indirect = 1;
+      prim[i].draw_id = i;
+   }
+
+   /* This should always be true at this time */
+   assert(indirect_data == ctx->DrawIndirectBuffer);
+
+   ctx->Driver.Draw(ctx, prim, draw_count, ib, false, 0u, ~0u,
+                    NULL, 0, indirect_data);
+}
+
+
+/*
+ * Function to be put into dd_function_table::DrawIndirect as fallback.
+ * Calls into dd_function_table::Draw past adapting call arguments.
+ * See dd_function_table::DrawIndirect for call argument documentation.
+ */
+void
+_mesa_draw_indirect(struct gl_context *ctx, GLuint mode,
+                    struct gl_buffer_object *indirect_data,
+                    GLsizeiptr indirect_offset, unsigned draw_count,
+                    unsigned stride,
+                    struct gl_buffer_object *indirect_draw_count_buffer,
+                    GLsizeiptr indirect_draw_count_offset,
+                    const struct _mesa_index_buffer *ib)
+{
+   /* Use alloca for the prim space if we are somehow in bounds. */
+   if (draw_count*sizeof(struct _mesa_prim) < 1024) {
+      struct _mesa_prim *space = alloca(draw_count*sizeof(struct _mesa_prim));
+      memset(space, 0, draw_count*sizeof(struct _mesa_prim));
+
+      draw_indirect(ctx, mode, indirect_data, indirect_offset, draw_count,
+                    stride, indirect_draw_count_buffer,
+                    indirect_draw_count_offset, ib, space);
+   } else {
+      struct _mesa_prim *space = calloc(draw_count, sizeof(struct _mesa_prim));
+      if (space == NULL) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "gl%sDraw%sIndirect%s",
+                     (draw_count > 1) ? "Multi" : "",
+                     ib ? "Elements" : "Arrays",
+                     indirect_data ? "CountARB" : "");
+         return;
+      }
+
+      draw_indirect(ctx, mode, indirect_data, indirect_offset, draw_count,
+                    stride, indirect_draw_count_buffer,
+                    indirect_draw_count_offset, ib, space);
+
+      free(space);
+   }
+}
diff --git a/lib/mesa/src/mesa/main/draw.h b/lib/mesa/src/mesa/main/draw.h
new file mode 100644
index 000000000..76d54f482
--- /dev/null
+++ b/lib/mesa/src/mesa/main/draw.h
@@ -0,0 +1,159 @@
+/*
+ * mesa 3-D graphics library
+ *
+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/**
+ * \brief Array type draw functions, the main workhorse of any OpenGL API
+ * \author Keith Whitwell
+ */
+
+
+#ifndef DRAW_H
+#define DRAW_H
+
+#include <stdbool.h>
+#include "main/glheader.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct gl_context;
+
+struct _mesa_prim
+{
+   GLuint mode:8;    /**< GL_POINTS, GL_LINES, GL_QUAD_STRIP, etc */
+   GLuint indexed:1;
+   GLuint begin:1;
+   GLuint end:1;
+   GLuint is_indirect:1;
+   GLuint pad:20;
+
+   GLuint start;
+   GLuint count;
+   GLint basevertex;
+   GLuint num_instances;
+   GLuint base_instance;
+   GLuint draw_id;
+
+   GLsizeiptr indirect_offset;
+};
+
+/* Would like to call this a "vbo_index_buffer", but this would be
+ * confusing as the indices are not neccessarily yet in a non-null
+ * buffer object.
+ */
+struct _mesa_index_buffer
+{
+   GLuint count;
+   unsigned index_size;
+   struct gl_buffer_object *obj;
+   const void *ptr;
+};
+
+
+void
+_mesa_initialize_exec_dispatch(const struct gl_context *ctx,
+                               struct _glapi_table *exec);
+
+
+void
+_mesa_draw_indirect(struct gl_context *ctx, GLuint mode,
+                    struct gl_buffer_object *indirect_data,
+                    GLsizeiptr indirect_offset, unsigned draw_count,
+                    unsigned stride,
+                    struct gl_buffer_object *indirect_draw_count_buffer,
+                    GLsizeiptr indirect_draw_count_offset,
+                    const struct _mesa_index_buffer *ib);
+
+
+void GLAPIENTRY
+_mesa_DrawArrays(GLenum mode, GLint first, GLsizei count);
+
+
+void GLAPIENTRY
+_mesa_DrawArraysInstanced(GLenum mode, GLint first, GLsizei count,
+                          GLsizei primcount);
+
+
+void GLAPIENTRY
+_mesa_DrawElements(GLenum mode, GLsizei count, GLenum type,
+                   const GLvoid *indices);
+
+
+void GLAPIENTRY
+_mesa_DrawRangeElements(GLenum mode, GLuint start, GLuint end, GLsizei count,
+                        GLenum type, const GLvoid *indices);
+
+
+void GLAPIENTRY
+_mesa_DrawElementsBaseVertex(GLenum mode, GLsizei count, GLenum type,
+                             const GLvoid *indices, GLint basevertex);
+
+
+void GLAPIENTRY
+_mesa_DrawRangeElementsBaseVertex(GLenum mode, GLuint start, GLuint end,
+                                  GLsizei count, GLenum type,
+                                  const GLvoid *indices,
+                                  GLint basevertex);
+
+
+void GLAPIENTRY
+_mesa_DrawTransformFeedback(GLenum mode, GLuint name);
+
+
+
+void GLAPIENTRY
+_mesa_MultiDrawArrays(GLenum mode, const GLint *first,
+                      const GLsizei *count, GLsizei primcount);
+
+
+void GLAPIENTRY
+_mesa_MultiDrawElementsEXT(GLenum mode, const GLsizei *count, GLenum type,
+                           const GLvoid **indices, GLsizei primcount);
+
+
+void GLAPIENTRY
+_mesa_MultiDrawElementsBaseVertex(GLenum mode,
+                                  const GLsizei *count, GLenum type,
+                                  const GLvoid **indices, GLsizei primcount,
+                                  const GLint *basevertex);
+
+
+void GLAPIENTRY
+_mesa_MultiModeDrawArraysIBM(const GLenum * mode, const GLint * first,
+                             const GLsizei * count,
+                             GLsizei primcount, GLint modestride);
+
+
+void GLAPIENTRY
+_mesa_MultiModeDrawElementsIBM(const GLenum * mode, const GLsizei * count,
+                               GLenum type, const GLvoid * const * indices,
+                               GLsizei primcount, GLint modestride);
+
+
+#ifdef __cplusplus
+} // extern "C"
+#endif
+
+#endif
diff --git a/lib/mesa/src/mesa/main/draw_validate.c b/lib/mesa/src/mesa/main/draw_validate.c
new file mode 100644
index 000000000..29304bd51
--- /dev/null
+++ b/lib/mesa/src/mesa/main/draw_validate.c
@@ -0,0 +1,1417 @@
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include <stdbool.h>
+#include "glheader.h"
+#include "draw_validate.h"
+#include "arrayobj.h"
+#include "bufferobj.h"
+#include "context.h"
+#include "imports.h"
+#include "mtypes.h"
+#include "pipelineobj.h"
+#include "enums.h"
+#include "state.h"
+#include "transformfeedback.h"
+#include "uniforms.h"
+#include "program/prog_print.h"
+
+
+static bool
+check_blend_func_error(struct gl_context *ctx)
+{
+   /* The ARB_blend_func_extended spec's ERRORS section says:
+    *
+    *    "The error INVALID_OPERATION is generated by Begin or any procedure
+    *     that implicitly calls Begin if any draw buffer has a blend function
+    *     requiring the second color input (SRC1_COLOR, ONE_MINUS_SRC1_COLOR,
+    *     SRC1_ALPHA or ONE_MINUS_SRC1_ALPHA), and a framebuffer is bound that
+    *     has more than the value of MAX_DUAL_SOURCE_DRAW_BUFFERS-1 active
+    *     color attachements."
+    */
+   for (unsigned i = ctx->Const.MaxDualSourceDrawBuffers;
+	i < ctx->DrawBuffer->_NumColorDrawBuffers;
+	i++) {
+      if (ctx->Color.Blend[i]._UsesDualSrc) {
+	 _mesa_error(ctx, GL_INVALID_OPERATION,
+		     "dual source blend on illegal attachment");
+	 return false;
+      }
+   }
+
+   if (ctx->Color.BlendEnabled && ctx->Color._AdvancedBlendMode) {
+      /* The KHR_blend_equation_advanced spec says:
+       *
+       *    "If any non-NONE draw buffer uses a blend equation found in table
+       *     X.1 or X.2, the error INVALID_OPERATION is generated by Begin or
+       *     any operation that implicitly calls Begin (such as DrawElements)
+       *     if:
+       *
+       *       * the draw buffer for color output zero selects multiple color
+       *         buffers (e.g., FRONT_AND_BACK in the default framebuffer); or
+       *
+       *       * the draw buffer for any other color output is not NONE."
+       */
+      if (ctx->DrawBuffer->ColorDrawBuffer[0] == GL_FRONT_AND_BACK) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "advanced blending is active and draw buffer for color "
+                     "output zero selects multiple color buffers");
+         return false;
+      }
+
+      for (unsigned i = 1; i < ctx->DrawBuffer->_NumColorDrawBuffers; i++) {
+         if (ctx->DrawBuffer->ColorDrawBuffer[i] != GL_NONE) {
+            _mesa_error(ctx, GL_INVALID_OPERATION,
+                        "advanced blending is active with multiple color "
+                        "draw buffers");
+            return false;
+         }
+      }
+
+      /* The KHR_blend_equation_advanced spec says:
+       *
+       *    "Advanced blending equations require the use of a fragment shader
+       *     with a matching "blend_support" layout qualifier.  If the current
+       *     blend equation is found in table X.1 or X.2, and the active
+       *     fragment shader does not include the layout qualifier matching
+       *     the blend equation or "blend_support_all_equations", the error
+       *     INVALID_OPERATION is generated [...]"
+       */
+      const struct gl_program *prog = ctx->FragmentProgram._Current;
+      const GLbitfield blend_support = !prog ? 0 : prog->sh.fs.BlendSupport;
+
+      if ((blend_support & ctx->Color._AdvancedBlendMode) == 0) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "fragment shader does not allow advanced blending mode "
+                     "(%s)",
+                      _mesa_enum_to_string(ctx->Color.Blend[0].EquationRGB));
+      }
+   }
+
+   return true;
+}
+
+
+/**
+ * Prior to drawing anything with glBegin, glDrawArrays, etc. this function
+ * is called to see if it's valid to render.  This involves checking that
+ * the current shader is valid and the framebuffer is complete.
+ * It also check the current pipeline object is valid if any.
+ * If an error is detected it'll be recorded here.
+ * \return GL_TRUE if OK to render, GL_FALSE if not
+ */
+GLboolean
+_mesa_valid_to_render(struct gl_context *ctx, const char *where)
+{
+   /* This depends on having up to date derived state (shaders) */
+   if (ctx->NewState)
+      _mesa_update_state(ctx);
+
+   if (ctx->API == API_OPENGL_COMPAT) {
+      /* Any shader stages that are not supplied by the GLSL shader and have
+       * assembly shaders enabled must now be validated.
+       */
+      if (!ctx->_Shader->CurrentProgram[MESA_SHADER_VERTEX] &&
+          ctx->VertexProgram.Enabled &&
+          !_mesa_arb_vertex_program_enabled(ctx)) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "%s(vertex program not valid)", where);
+         return GL_FALSE;
+      }
+
+      if (!ctx->_Shader->CurrentProgram[MESA_SHADER_FRAGMENT]) {
+         if (ctx->FragmentProgram.Enabled &&
+             !_mesa_arb_fragment_program_enabled(ctx)) {
+            _mesa_error(ctx, GL_INVALID_OPERATION,
+                        "%s(fragment program not valid)", where);
+            return GL_FALSE;
+         }
+
+         /* If drawing to integer-valued color buffers, there must be an
+          * active fragment shader (GL_EXT_texture_integer).
+          */
+         if (ctx->DrawBuffer && ctx->DrawBuffer->_IntegerBuffers) {
+            _mesa_error(ctx, GL_INVALID_OPERATION,
+                        "%s(integer format but no fragment shader)", where);
+            return GL_FALSE;
+         }
+      }
+   }
+
+   /* A pipeline object is bound */
+   if (ctx->_Shader->Name && !ctx->_Shader->Validated) {
+      if (!_mesa_validate_program_pipeline(ctx, ctx->_Shader)) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "glValidateProgramPipeline failed to validate the "
+                     "pipeline");
+         return GL_FALSE;
+      }
+   }
+
+   /* If a program is active and SSO not in use, check if validation of
+    * samplers succeeded for the active program. */
+   if (ctx->_Shader->ActiveProgram && ctx->_Shader != ctx->Pipeline.Current) {
+      char errMsg[100];
+      if (!_mesa_sampler_uniforms_are_valid(ctx->_Shader->ActiveProgram,
+                                            errMsg, 100)) {
+         _mesa_error(ctx, GL_INVALID_OPERATION, "%s", errMsg);
+         return GL_FALSE;
+      }
+   }
+
+   if (ctx->DrawBuffer->_Status != GL_FRAMEBUFFER_COMPLETE_EXT) {
+      _mesa_error(ctx, GL_INVALID_FRAMEBUFFER_OPERATION_EXT,
+                  "%s(incomplete framebuffer)", where);
+      return GL_FALSE;
+   }
+
+   if (!check_blend_func_error(ctx)) {
+      return GL_FALSE;
+   }
+
+   /* From the GL_NV_fill_rectangle spec:
+    *
+    * "An INVALID_OPERATION error is generated by Begin or any Draw command if
+    *  only one of the front and back polygon mode is FILL_RECTANGLE_NV."
+    */
+   if ((ctx->Polygon.FrontMode == GL_FILL_RECTANGLE_NV) !=
+       (ctx->Polygon.BackMode == GL_FILL_RECTANGLE_NV)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "GL_FILL_RECTANGLE_NV must be used as both front/back "
+                  "polygon mode or neither");
+      return GL_FALSE;
+   }
+
+#ifdef DEBUG
+   if (ctx->_Shader->Flags & GLSL_LOG) {
+      struct gl_program **prog = ctx->_Shader->CurrentProgram;
+
+      for (unsigned i = 0; i < MESA_SHADER_STAGES; i++) {
+	 if (prog[i] == NULL || prog[i]->_Used)
+	    continue;
+
+	 /* This is the first time this shader is being used.
+	  * Append shader's constants/uniforms to log file.
+	  *
+	  * Only log data for the program target that matches the shader
+	  * target.  It's possible to have a program bound to the vertex
+	  * shader target that also supplied a fragment shader.  If that
+	  * program isn't also bound to the fragment shader target we don't
+	  * want to log its fragment data.
+	  */
+	 _mesa_append_uniforms_to_file(prog[i]);
+      }
+
+      for (unsigned i = 0; i < MESA_SHADER_STAGES; i++) {
+	 if (prog[i] != NULL)
+	    prog[i]->_Used = GL_TRUE;
+      }
+   }
+#endif
+
+   return GL_TRUE;
+}
+
+
+/**
+ * Check if OK to draw arrays/elements.
+ */
+static bool
+check_valid_to_render(struct gl_context *ctx, const char *function)
+{
+   if (!_mesa_valid_to_render(ctx, function)) {
+      return false;
+   }
+
+   /* Section 6.3.2 from the GL 4.5:
+    * "Any GL command which attempts to read from, write to, or change
+    *  the state of a buffer object may generate an INVALID_OPERATION error if
+    *  all or part of the buffer object is mapped ... However, only commands
+    *  which explicitly describe this error are required to do so. If an error
+    *  is not generated, such commands will have undefined results and may
+    *  result in GL interruption or termination."
+    *
+    * Only some buffer API functions require INVALID_OPERATION with mapped
+    * buffers. No other functions list such an error, thus it's not required
+    * to report INVALID_OPERATION for draw calls with mapped buffers.
+    */
+   if (!ctx->Const.AllowMappedBuffersDuringExecution &&
+       !_mesa_all_buffers_are_unmapped(ctx->Array.VAO)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(vertex buffers are mapped)", function);
+      return false;
+   }
+
+   /* Section 11.2 (Tessellation) of the ES 3.2 spec says:
+    *
+    * "An INVALID_OPERATION error is generated by any command that
+    *  transfers vertices to the GL if the current program state has
+    *  one but not both of a tessellation control shader and tessellation
+    *  evaluation shader."
+    *
+    * The OpenGL spec argues that this is allowed because a tess ctrl shader
+    * without a tess eval shader can be used with transform feedback.
+    * However, glBeginTransformFeedback doesn't allow GL_PATCHES and
+    * therefore doesn't allow tessellation.
+    *
+    * Further investigation showed that this is indeed a spec bug and
+    * a tess ctrl shader without a tess eval shader shouldn't have been
+    * allowed, because there is no API in GL 4.0 that can make use this
+    * to produce something useful.
+    *
+    * Also, all vendors except one don't support a tess ctrl shader without
+    * a tess eval shader anyway.
+    */
+   if (ctx->TessCtrlProgram._Current && !ctx->TessEvalProgram._Current) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(tess eval shader is missing)", function);
+      return false;
+   }
+
+   switch (ctx->API) {
+   case API_OPENGLES2:
+      /* Section 11.2 (Tessellation) of the ES 3.2 spec says:
+       *
+       * "An INVALID_OPERATION error is generated by any command that
+       *  transfers vertices to the GL if the current program state has
+       *  one but not both of a tessellation control shader and tessellation
+       *  evaluation shader."
+       */
+      if (_mesa_is_gles3(ctx) &&
+          ctx->TessEvalProgram._Current && !ctx->TessCtrlProgram._Current) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "%s(tess ctrl shader is missing)", function);
+         return false;
+      }
+      break;
+
+   case API_OPENGL_CORE:
+      /* Section 10.4 (Drawing Commands Using Vertex Arrays) of the OpenGL 4.5
+       * Core Profile spec says:
+       *
+       *     "An INVALID_OPERATION error is generated if no vertex array
+       *     object is bound (see section 10.3.1)."
+       */
+      if (ctx->Array.VAO == ctx->Array.DefaultVAO) {
+         _mesa_error(ctx, GL_INVALID_OPERATION, "%s(no VAO bound)", function);
+         return false;
+      }
+      break;
+
+   case API_OPENGLES:
+   case API_OPENGL_COMPAT:
+      break;
+
+   default:
+      unreachable("Invalid API value in check_valid_to_render()");
+   }
+
+   return true;
+}
+
+
+/**
+ * Is 'mode' a valid value for glBegin(), glDrawArrays(), glDrawElements(),
+ * etc?  The set of legal values depends on whether geometry shaders/programs
+ * are supported.
+ * Note: This may be called during display list compilation.
+ */
+bool
+_mesa_is_valid_prim_mode(const struct gl_context *ctx, GLenum mode)
+{
+   /* The overwhelmingly common case is (mode <= GL_TRIANGLE_FAN).  Test that
+    * first and exit.  You would think that a switch-statement would be the
+    * right approach, but at least GCC 4.7.2 generates some pretty dire code
+    * for the common case.
+    */
+   if (likely(mode <= GL_TRIANGLE_FAN))
+      return true;
+
+   if (mode <= GL_POLYGON)
+      return (ctx->API == API_OPENGL_COMPAT);
+
+   if (mode <= GL_TRIANGLE_STRIP_ADJACENCY)
+      return _mesa_has_geometry_shaders(ctx);
+
+   if (mode == GL_PATCHES)
+      return _mesa_has_tessellation(ctx);
+
+   return false;
+}
+
+
+/**
+ * Is 'mode' a valid value for glBegin(), glDrawArrays(), glDrawElements(),
+ * etc?  Also, do additional checking related to transformation feedback.
+ * Note: this function cannot be called during glNewList(GL_COMPILE) because
+ * this code depends on current transform feedback state.
+ * Also, do additional checking related to tessellation shaders.
+ */
+GLboolean
+_mesa_valid_prim_mode(struct gl_context *ctx, GLenum mode, const char *name)
+{
+   bool valid_enum = _mesa_is_valid_prim_mode(ctx, mode);
+
+   if (!valid_enum) {
+      _mesa_error(ctx, GL_INVALID_ENUM, "%s(mode=%x)", name, mode);
+      return GL_FALSE;
+   }
+
+   /* From the OpenGL 4.5 specification, section 11.3.1:
+    *
+    * The error INVALID_OPERATION is generated if Begin, or any command that
+    * implicitly calls Begin, is called when a geometry shader is active and:
+    *
+    * * the input primitive type of the current geometry shader is
+    *   POINTS and <mode> is not POINTS,
+    *
+    * * the input primitive type of the current geometry shader is
+    *   LINES and <mode> is not LINES, LINE_STRIP, or LINE_LOOP,
+    *
+    * * the input primitive type of the current geometry shader is
+    *   TRIANGLES and <mode> is not TRIANGLES, TRIANGLE_STRIP or
+    *   TRIANGLE_FAN,
+    *
+    * * the input primitive type of the current geometry shader is
+    *   LINES_ADJACENCY_ARB and <mode> is not LINES_ADJACENCY_ARB or
+    *   LINE_STRIP_ADJACENCY_ARB, or
+    *
+    * * the input primitive type of the current geometry shader is
+    *   TRIANGLES_ADJACENCY_ARB and <mode> is not
+    *   TRIANGLES_ADJACENCY_ARB or TRIANGLE_STRIP_ADJACENCY_ARB.
+    *
+    * The GL spec doesn't mention any interaction with tessellation, which
+    * is clearly a spec bug. The same rule should apply, but instead of
+    * the draw primitive mode, the tessellation evaluation shader primitive
+    * mode should be used for the checking.
+   */
+   if (ctx->_Shader->CurrentProgram[MESA_SHADER_GEOMETRY]) {
+      const GLenum geom_mode =
+         ctx->_Shader->CurrentProgram[MESA_SHADER_GEOMETRY]->
+            info.gs.input_primitive;
+      struct gl_program *tes =
+         ctx->_Shader->CurrentProgram[MESA_SHADER_TESS_EVAL];
+      GLenum mode_before_gs = mode;
+
+      if (tes) {
+         if (tes->info.tess.point_mode)
+            mode_before_gs = GL_POINTS;
+         else if (tes->info.tess.primitive_mode == GL_ISOLINES)
+            mode_before_gs = GL_LINES;
+         else
+            /* the GL_QUADS mode generates triangles too */
+            mode_before_gs = GL_TRIANGLES;
+      }
+
+      switch (mode_before_gs) {
+      case GL_POINTS:
+         valid_enum = (geom_mode == GL_POINTS);
+         break;
+      case GL_LINES:
+      case GL_LINE_LOOP:
+      case GL_LINE_STRIP:
+         valid_enum = (geom_mode == GL_LINES);
+         break;
+      case GL_TRIANGLES:
+      case GL_TRIANGLE_STRIP:
+      case GL_TRIANGLE_FAN:
+         valid_enum = (geom_mode == GL_TRIANGLES);
+         break;
+      case GL_QUADS:
+      case GL_QUAD_STRIP:
+      case GL_POLYGON:
+         valid_enum = false;
+         break;
+      case GL_LINES_ADJACENCY:
+      case GL_LINE_STRIP_ADJACENCY:
+         valid_enum = (geom_mode == GL_LINES_ADJACENCY);
+         break;
+      case GL_TRIANGLES_ADJACENCY:
+      case GL_TRIANGLE_STRIP_ADJACENCY:
+         valid_enum = (geom_mode == GL_TRIANGLES_ADJACENCY);
+         break;
+      default:
+         valid_enum = false;
+         break;
+      }
+      if (!valid_enum) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "%s(mode=%s vs geometry shader input %s)",
+                     name,
+                     _mesa_lookup_prim_by_nr(mode_before_gs),
+                     _mesa_lookup_prim_by_nr(geom_mode));
+         return GL_FALSE;
+      }
+   }
+
+   /* From the OpenGL 4.0 (Core Profile) spec (section 2.12):
+    *
+    *     "Tessellation operates only on patch primitives. If tessellation is
+    *      active, any command that transfers vertices to the GL will
+    *      generate an INVALID_OPERATION error if the primitive mode is not
+    *      PATCHES.
+    *      Patch primitives are not supported by pipeline stages below the
+    *      tessellation evaluation shader. If there is no active program
+    *      object or the active program object does not contain a tessellation
+    *      evaluation shader, the error INVALID_OPERATION is generated by any
+    *      command that transfers vertices to the GL if the primitive mode is
+    *      PATCHES."
+    *
+    */
+   if (ctx->_Shader->CurrentProgram[MESA_SHADER_TESS_EVAL] ||
+       ctx->_Shader->CurrentProgram[MESA_SHADER_TESS_CTRL]) {
+      if (mode != GL_PATCHES) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "only GL_PATCHES valid with tessellation");
+         return GL_FALSE;
+      }
+   }
+   else {
+      if (mode == GL_PATCHES) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "GL_PATCHES only valid with tessellation");
+         return GL_FALSE;
+      }
+   }
+
+   /* From the GL_EXT_transform_feedback spec:
+    *
+    *     "The error INVALID_OPERATION is generated if Begin, or any command
+    *      that performs an explicit Begin, is called when:
+    *
+    *      * a geometry shader is not active and <mode> does not match the
+    *        allowed begin modes for the current transform feedback state as
+    *        given by table X.1.
+    *
+    *      * a geometry shader is active and the output primitive type of the
+    *        geometry shader does not match the allowed begin modes for the
+    *        current transform feedback state as given by table X.1.
+    *
+    */
+   if (_mesa_is_xfb_active_and_unpaused(ctx)) {
+      GLboolean pass = GL_TRUE;
+
+      if(ctx->_Shader->CurrentProgram[MESA_SHADER_GEOMETRY]) {
+         switch (ctx->_Shader->CurrentProgram[MESA_SHADER_GEOMETRY]->
+                    info.gs.output_primitive) {
+         case GL_POINTS:
+            pass = ctx->TransformFeedback.Mode == GL_POINTS;
+            break;
+         case GL_LINE_STRIP:
+            pass = ctx->TransformFeedback.Mode == GL_LINES;
+            break;
+         case GL_TRIANGLE_STRIP:
+            pass = ctx->TransformFeedback.Mode == GL_TRIANGLES;
+            break;
+         default:
+            pass = GL_FALSE;
+         }
+      }
+      else if (ctx->_Shader->CurrentProgram[MESA_SHADER_TESS_EVAL]) {
+         struct gl_program *tes =
+            ctx->_Shader->CurrentProgram[MESA_SHADER_TESS_EVAL];
+         if (tes->info.tess.point_mode)
+            pass = ctx->TransformFeedback.Mode == GL_POINTS;
+         else if (tes->info.tess.primitive_mode == GL_ISOLINES)
+            pass = ctx->TransformFeedback.Mode == GL_LINES;
+         else
+            pass = ctx->TransformFeedback.Mode == GL_TRIANGLES;
+      }
+      else {
+         switch (mode) {
+         case GL_POINTS:
+            pass = ctx->TransformFeedback.Mode == GL_POINTS;
+            break;
+         case GL_LINES:
+         case GL_LINE_STRIP:
+         case GL_LINE_LOOP:
+            pass = ctx->TransformFeedback.Mode == GL_LINES;
+            break;
+         default:
+            pass = ctx->TransformFeedback.Mode == GL_TRIANGLES;
+            break;
+         }
+      }
+      if (!pass) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                         "%s(mode=%s vs transform feedback %s)",
+                         name,
+                         _mesa_lookup_prim_by_nr(mode),
+                         _mesa_lookup_prim_by_nr(ctx->TransformFeedback.Mode));
+         return GL_FALSE;
+      }
+   }
+
+   /* From GL_INTEL_conservative_rasterization spec:
+    *
+    * The conservative rasterization option applies only to polygons with
+    * PolygonMode state set to FILL. Draw requests for polygons with different
+    * PolygonMode setting or for other primitive types (points/lines) generate
+    * INVALID_OPERATION error.
+    */
+   if (ctx->IntelConservativeRasterization) {
+      GLboolean pass = GL_TRUE;
+
+      switch (mode) {
+      case GL_POINTS:
+      case GL_LINES:
+      case GL_LINE_LOOP:
+      case GL_LINE_STRIP:
+      case GL_LINES_ADJACENCY:
+      case GL_LINE_STRIP_ADJACENCY:
+         pass = GL_FALSE;
+         break;
+      case GL_TRIANGLES:
+      case GL_TRIANGLE_STRIP:
+      case GL_TRIANGLE_FAN:
+      case GL_QUADS:
+      case GL_QUAD_STRIP:
+      case GL_POLYGON:
+      case GL_TRIANGLES_ADJACENCY:
+      case GL_TRIANGLE_STRIP_ADJACENCY:
+         if (ctx->Polygon.FrontMode != GL_FILL ||
+             ctx->Polygon.BackMode != GL_FILL)
+            pass = GL_FALSE;
+         break;
+      default:
+         pass = GL_FALSE;
+      }
+      if (!pass) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "mode=%s invalid with GL_INTEL_conservative_rasterization",
+                     _mesa_lookup_prim_by_nr(mode));
+         return GL_FALSE;
+      }
+   }
+
+   return GL_TRUE;
+}
+
+/**
+ * Verify that the element type is valid.
+ *
+ * Generates \c GL_INVALID_ENUM and returns \c false if it is not.
+ */
+static bool
+valid_elements_type(struct gl_context *ctx, GLenum type, const char *name)
+{
+   switch (type) {
+   case GL_UNSIGNED_BYTE:
+   case GL_UNSIGNED_SHORT:
+   case GL_UNSIGNED_INT:
+      return true;
+
+   default:
+      _mesa_error(ctx, GL_INVALID_ENUM, "%s(type = %s)", name,
+                  _mesa_enum_to_string(type));
+      return false;
+   }
+}
+
+static bool
+validate_DrawElements_common(struct gl_context *ctx,
+                             GLenum mode, GLsizei count, GLenum type,
+                             const GLvoid *indices,
+                             const char *caller)
+{
+   /* Section 2.14.2 (Transform Feedback Primitive Capture) of the OpenGL ES
+    * 3.1 spec says:
+    *
+    *   The error INVALID_OPERATION is also generated by DrawElements,
+    *   DrawElementsInstanced, and DrawRangeElements while transform feedback
+    *   is active and not paused, regardless of mode.
+    *
+    * The OES_geometry_shader_spec says:
+    *
+    *    Issues:
+    *
+    *    ...
+    *
+    *    (13) Does this extension change how transform feedback operates
+    *    compared to unextended OpenGL ES 3.0 or 3.1?
+    *
+    *    RESOLVED: Yes... Since we no longer require being able to predict how
+    *    much geometry will be generated, we also lift the restriction that
+    *    only DrawArray* commands are supported and also support the
+    *    DrawElements* commands for transform feedback.
+    *
+    * This should also be reflected in the body of the spec, but that appears
+    * to have been overlooked.  The body of the spec only explicitly allows
+    * the indirect versions.
+    */
+   if (_mesa_is_gles3(ctx) &&
+       !_mesa_has_OES_geometry_shader(ctx) &&
+       _mesa_is_xfb_active_and_unpaused(ctx)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(transform feedback active)", caller);
+      return false;
+   }
+
+   if (count < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "%s(count)", caller);
+      return false;
+   }
+
+   if (!_mesa_valid_prim_mode(ctx, mode, caller)) {
+      return false;
+   }
+
+   if (!valid_elements_type(ctx, type, caller))
+      return false;
+
+   if (!check_valid_to_render(ctx, caller))
+      return false;
+
+   return true;
+}
+
+/**
+ * Error checking for glDrawElements().  Includes parameter checking
+ * and VBO bounds checking.
+ * \return GL_TRUE if OK to render, GL_FALSE if error found
+ */
+GLboolean
+_mesa_validate_DrawElements(struct gl_context *ctx,
+                            GLenum mode, GLsizei count, GLenum type,
+                            const GLvoid *indices)
+{
+   return validate_DrawElements_common(ctx, mode, count, type, indices,
+                                       "glDrawElements");
+}
+
+
+/**
+ * Error checking for glMultiDrawElements().  Includes parameter checking
+ * and VBO bounds checking.
+ * \return GL_TRUE if OK to render, GL_FALSE if error found
+ */
+GLboolean
+_mesa_validate_MultiDrawElements(struct gl_context *ctx,
+                                 GLenum mode, const GLsizei *count,
+                                 GLenum type, const GLvoid * const *indices,
+                                 GLsizei primcount)
+{
+   GLsizei i;
+
+   /*
+    * Section 2.3.1 (Errors) of the OpenGL 4.5 (Core Profile) spec says:
+    *
+    *    "If a negative number is provided where an argument of type sizei or
+    *     sizeiptr is specified, an INVALID_VALUE error is generated."
+    *
+    * and in the same section:
+    *
+    *    "In other cases, there are no side effects unless otherwise noted;
+    *     the command which generates the error is ignored so that it has no
+    *     effect on GL state or framebuffer contents."
+    *
+    * Hence, check both primcount and all the count[i].
+    */
+   if (primcount < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glMultiDrawElements(primcount=%d)", primcount);
+      return GL_FALSE;
+   }
+
+   for (i = 0; i < primcount; i++) {
+      if (count[i] < 0) {
+         _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glMultiDrawElements(count)" );
+         return GL_FALSE;
+      }
+   }
+
+   if (!_mesa_valid_prim_mode(ctx, mode, "glMultiDrawElements")) {
+      return GL_FALSE;
+   }
+
+   if (!valid_elements_type(ctx, type, "glMultiDrawElements"))
+      return GL_FALSE;
+
+   if (!check_valid_to_render(ctx, "glMultiDrawElements"))
+      return GL_FALSE;
+
+   /* Not using a VBO for indices, so avoid NULL pointer derefs later.
+    */
+   if (!_mesa_is_bufferobj(ctx->Array.VAO->IndexBufferObj)) {
+      for (i = 0; i < primcount; i++) {
+         if (!indices[i])
+            return GL_FALSE;
+      }
+   }
+
+   return GL_TRUE;
+}
+
+
+/**
+ * Error checking for glDrawRangeElements().  Includes parameter checking
+ * and VBO bounds checking.
+ * \return GL_TRUE if OK to render, GL_FALSE if error found
+ */
+GLboolean
+_mesa_validate_DrawRangeElements(struct gl_context *ctx, GLenum mode,
+                                 GLuint start, GLuint end,
+                                 GLsizei count, GLenum type,
+                                 const GLvoid *indices)
+{
+   if (end < start) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glDrawRangeElements(end<start)");
+      return GL_FALSE;
+   }
+
+   return validate_DrawElements_common(ctx, mode, count, type, indices,
+                                       "glDrawRangeElements");
+}
+
+
+static bool
+need_xfb_remaining_prims_check(const struct gl_context *ctx)
+{
+   /* From the GLES3 specification, section 2.14.2 (Transform Feedback
+    * Primitive Capture):
+    *
+    *   The error INVALID_OPERATION is generated by DrawArrays and
+    *   DrawArraysInstanced if recording the vertices of a primitive to the
+    *   buffer objects being used for transform feedback purposes would result
+    *   in either exceeding the limits of any buffer object’s size, or in
+    *   exceeding the end position offset + size − 1, as set by
+    *   BindBufferRange.
+    *
+    * This is in contrast to the behaviour of desktop GL, where the extra
+    * primitives are silently dropped from the transform feedback buffer.
+    *
+    * This text is removed in ES 3.2, presumably because it's not really
+    * implementable with geometry and tessellation shaders.  In fact,
+    * the OES_geometry_shader spec says:
+    *
+    *    "(13) Does this extension change how transform feedback operates
+    *     compared to unextended OpenGL ES 3.0 or 3.1?
+    *
+    *     RESOLVED: Yes. Because dynamic geometry amplification in a geometry
+    *     shader can make it difficult if not impossible to predict the amount
+    *     of geometry that may be generated in advance of executing the shader,
+    *     the draw-time error for transform feedback buffer overflow conditions
+    *     is removed and replaced with the GL behavior (primitives are not
+    *     written and the corresponding counter is not updated)..."
+    */
+   return _mesa_is_gles3(ctx) && _mesa_is_xfb_active_and_unpaused(ctx) &&
+          !_mesa_has_OES_geometry_shader(ctx) &&
+          !_mesa_has_OES_tessellation_shader(ctx);
+}
+
+
+/**
+ * Figure out the number of transform feedback primitives that will be output
+ * considering the drawing mode, number of vertices, and instance count,
+ * assuming that no geometry shading is done and primitive restart is not
+ * used.
+ *
+ * This is used by driver back-ends in implementing the PRIMITIVES_GENERATED
+ * and TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN queries.  It is also used to
+ * pre-validate draw calls in GLES3 (where draw calls only succeed if there is
+ * enough room in the transform feedback buffer for the result).
+ */
+static size_t
+count_tessellated_primitives(GLenum mode, GLuint count, GLuint num_instances)
+{
+   size_t num_primitives;
+   switch (mode) {
+   case GL_POINTS:
+      num_primitives = count;
+      break;
+   case GL_LINE_STRIP:
+      num_primitives = count >= 2 ? count - 1 : 0;
+      break;
+   case GL_LINE_LOOP:
+      num_primitives = count >= 2 ? count : 0;
+      break;
+   case GL_LINES:
+      num_primitives = count / 2;
+      break;
+   case GL_TRIANGLE_STRIP:
+   case GL_TRIANGLE_FAN:
+   case GL_POLYGON:
+      num_primitives = count >= 3 ? count - 2 : 0;
+      break;
+   case GL_TRIANGLES:
+      num_primitives = count / 3;
+      break;
+   case GL_QUAD_STRIP:
+      num_primitives = count >= 4 ? ((count / 2) - 1) * 2 : 0;
+      break;
+   case GL_QUADS:
+      num_primitives = (count / 4) * 2;
+      break;
+   case GL_LINES_ADJACENCY:
+      num_primitives = count / 4;
+      break;
+   case GL_LINE_STRIP_ADJACENCY:
+      num_primitives = count >= 4 ? count - 3 : 0;
+      break;
+   case GL_TRIANGLES_ADJACENCY:
+      num_primitives = count / 6;
+      break;
+   case GL_TRIANGLE_STRIP_ADJACENCY:
+      num_primitives = count >= 6 ? (count - 4) / 2 : 0;
+      break;
+   default:
+      assert(!"Unexpected primitive type in count_tessellated_primitives");
+      num_primitives = 0;
+      break;
+   }
+   return num_primitives * num_instances;
+}
+
+
+static bool
+validate_draw_arrays(struct gl_context *ctx, const char *func,
+                     GLenum mode, GLsizei count, GLsizei numInstances)
+{
+   if (count < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "%s(count)", func);
+      return false;
+   }
+
+   if (!_mesa_valid_prim_mode(ctx, mode, func))
+      return false;
+
+   if (!check_valid_to_render(ctx, func))
+      return false;
+
+   if (need_xfb_remaining_prims_check(ctx)) {
+      struct gl_transform_feedback_object *xfb_obj
+         = ctx->TransformFeedback.CurrentObject;
+      size_t prim_count = count_tessellated_primitives(mode, count, numInstances);
+      if (xfb_obj->GlesRemainingPrims < prim_count) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "%s(exceeds transform feedback size)", func);
+         return false;
+      }
+      xfb_obj->GlesRemainingPrims -= prim_count;
+   }
+
+   if (count == 0)
+      return false;
+
+   return true;
+}
+
+/**
+ * Called from the tnl module to error check the function parameters and
+ * verify that we really can draw something.
+ * \return GL_TRUE if OK to render, GL_FALSE if error found
+ */
+GLboolean
+_mesa_validate_DrawArrays(struct gl_context *ctx, GLenum mode, GLsizei count)
+{
+   return validate_draw_arrays(ctx, "glDrawArrays", mode, count, 1);
+}
+
+
+GLboolean
+_mesa_validate_DrawArraysInstanced(struct gl_context *ctx, GLenum mode, GLint first,
+                                   GLsizei count, GLsizei numInstances)
+{
+   if (first < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glDrawArraysInstanced(start=%d)", first);
+      return GL_FALSE;
+   }
+
+   if (numInstances <= 0) {
+      if (numInstances < 0)
+         _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glDrawArraysInstanced(numInstances=%d)", numInstances);
+      return GL_FALSE;
+   }
+
+   return validate_draw_arrays(ctx, "glDrawArraysInstanced", mode, count, 1);
+}
+
+
+/**
+ * Called to error check the function parameters.
+ *
+ * Note that glMultiDrawArrays is not part of GLES, so there's limited scope
+ * for sharing code with the validation of glDrawArrays.
+ */
+bool
+_mesa_validate_MultiDrawArrays(struct gl_context *ctx, GLenum mode,
+                               const GLsizei *count, GLsizei primcount)
+{
+   int i;
+
+   if (!_mesa_valid_prim_mode(ctx, mode, "glMultiDrawArrays"))
+      return false;
+
+   if (!check_valid_to_render(ctx, "glMultiDrawArrays"))
+      return false;
+
+   if (primcount < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glMultiDrawArrays(primcount=%d)",
+                  primcount);
+      return false;
+   }
+
+   for (i = 0; i < primcount; ++i) {
+      if (count[i] < 0) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "glMultiDrawArrays(count[%d]=%d)",
+                     i, count[i]);
+         return false;
+      }
+   }
+
+   if (need_xfb_remaining_prims_check(ctx)) {
+      struct gl_transform_feedback_object *xfb_obj
+         = ctx->TransformFeedback.CurrentObject;
+      size_t xfb_prim_count = 0;
+
+      for (i = 0; i < primcount; ++i)
+         xfb_prim_count += count_tessellated_primitives(mode, count[i], 1);
+
+      if (xfb_obj->GlesRemainingPrims < xfb_prim_count) {
+         _mesa_error(ctx, GL_INVALID_OPERATION,
+                     "glMultiDrawArrays(exceeds transform feedback size)");
+         return false;
+      }
+      xfb_obj->GlesRemainingPrims -= xfb_prim_count;
+   }
+
+   return true;
+}
+
+
+GLboolean
+_mesa_validate_DrawElementsInstanced(struct gl_context *ctx,
+                                     GLenum mode, GLsizei count, GLenum type,
+                                     const GLvoid *indices, GLsizei numInstances)
+{
+   if (numInstances < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glDrawElementsInstanced(numInstances=%d)", numInstances);
+      return GL_FALSE;
+   }
+
+   return validate_DrawElements_common(ctx, mode, count, type, indices,
+                                       "glDrawElementsInstanced")
+      && (numInstances > 0);
+}
+
+
+GLboolean
+_mesa_validate_DrawTransformFeedback(struct gl_context *ctx,
+                                     GLenum mode,
+                                     struct gl_transform_feedback_object *obj,
+                                     GLuint stream,
+                                     GLsizei numInstances)
+{
+   if (!_mesa_valid_prim_mode(ctx, mode, "glDrawTransformFeedback*(mode)")) {
+      return GL_FALSE;
+   }
+
+   if (!obj) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glDrawTransformFeedback*(name)");
+      return GL_FALSE;
+   }
+
+   /* From the GL 4.5 specification, page 429:
+    * "An INVALID_VALUE error is generated if id is not the name of a
+    *  transform feedback object."
+    */
+   if (!obj->EverBound) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glDrawTransformFeedback*(name)");
+      return GL_FALSE;
+   }
+
+   if (stream >= ctx->Const.MaxVertexStreams) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glDrawTransformFeedbackStream*(index>=MaxVertexStream)");
+      return GL_FALSE;
+   }
+
+   if (!obj->EndedAnytime) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawTransformFeedback*");
+      return GL_FALSE;
+   }
+
+   if (numInstances <= 0) {
+      if (numInstances < 0)
+         _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glDrawTransformFeedback*Instanced(numInstances=%d)",
+                     numInstances);
+      return GL_FALSE;
+   }
+
+   if (!check_valid_to_render(ctx, "glDrawTransformFeedback*")) {
+      return GL_FALSE;
+   }
+
+   return GL_TRUE;
+}
+
+static GLboolean
+valid_draw_indirect(struct gl_context *ctx,
+                    GLenum mode, const GLvoid *indirect,
+                    GLsizei size, const char *name)
+{
+   const uint64_t end = (uint64_t) (uintptr_t) indirect + size;
+
+   /* OpenGL ES 3.1 spec. section 10.5:
+    *
+    *      "DrawArraysIndirect requires that all data sourced for the
+    *      command, including the DrawArraysIndirectCommand
+    *      structure,  be in buffer objects,  and may not be called when
+    *      the default vertex array object is bound."
+    */
+   if (ctx->API != API_OPENGL_COMPAT &&
+       ctx->Array.VAO == ctx->Array.DefaultVAO) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "(no VAO bound)");
+      return GL_FALSE;
+   }
+
+   /* From OpenGL ES 3.1 spec. section 10.5:
+    *     "An INVALID_OPERATION error is generated if zero is bound to
+    *     VERTEX_ARRAY_BINDING, DRAW_INDIRECT_BUFFER or to any enabled
+    *     vertex array."
+    *
+    * Here we check that for each enabled vertex array we have a vertex
+    * buffer bound.
+    */
+   if (_mesa_is_gles31(ctx) &&
+       ctx->Array.VAO->_Enabled & ~ctx->Array.VAO->VertexAttribBufferMask) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(No VBO bound)", name);
+      return GL_FALSE;
+   }
+
+   if (!_mesa_valid_prim_mode(ctx, mode, name))
+      return GL_FALSE;
+
+   /* OpenGL ES 3.1 specification, section 10.5:
+    *
+    *      "An INVALID_OPERATION error is generated if
+    *      transform feedback is active and not paused."
+    *
+    * The OES_geometry_shader spec says:
+    *
+    *    On p. 250 in the errors section for the DrawArraysIndirect command,
+    *    and on p. 254 in the errors section for the DrawElementsIndirect
+    *    command, delete the errors which state:
+    *
+    *    "An INVALID_OPERATION error is generated if transform feedback is
+    *    active and not paused."
+    *
+    *    (thus allowing transform feedback to work with indirect draw commands).
+    */
+   if (_mesa_is_gles31(ctx) && !ctx->Extensions.OES_geometry_shader &&
+       _mesa_is_xfb_active_and_unpaused(ctx)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(TransformFeedback is active and not paused)", name);
+   }
+
+   /* From OpenGL version 4.4. section 10.5
+    * and OpenGL ES 3.1, section 10.6:
+    *
+    *      "An INVALID_VALUE error is generated if indirect is not a
+    *       multiple of the size, in basic machine units, of uint."
+    */
+   if ((GLsizeiptr)indirect & (sizeof(GLuint) - 1)) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "%s(indirect is not aligned)", name);
+      return GL_FALSE;
+   }
+
+   if (!_mesa_is_bufferobj(ctx->DrawIndirectBuffer)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s: no buffer bound to DRAW_INDIRECT_BUFFER", name);
+      return GL_FALSE;
+   }
+
+   if (_mesa_check_disallowed_mapping(ctx->DrawIndirectBuffer)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(DRAW_INDIRECT_BUFFER is mapped)", name);
+      return GL_FALSE;
+   }
+
+   /* From the ARB_draw_indirect specification:
+    * "An INVALID_OPERATION error is generated if the commands source data
+    *  beyond the end of the buffer object [...]"
+    */
+   if (ctx->DrawIndirectBuffer->Size < end) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(DRAW_INDIRECT_BUFFER too small)", name);
+      return GL_FALSE;
+   }
+
+   if (!check_valid_to_render(ctx, name))
+      return GL_FALSE;
+
+   return GL_TRUE;
+}
+
+static inline GLboolean
+valid_draw_indirect_elements(struct gl_context *ctx,
+                             GLenum mode, GLenum type, const GLvoid *indirect,
+                             GLsizeiptr size, const char *name)
+{
+   if (!valid_elements_type(ctx, type, name))
+      return GL_FALSE;
+
+   /*
+    * Unlike regular DrawElementsInstancedBaseVertex commands, the indices
+    * may not come from a client array and must come from an index buffer.
+    * If no element array buffer is bound, an INVALID_OPERATION error is
+    * generated.
+    */
+   if (!_mesa_is_bufferobj(ctx->Array.VAO->IndexBufferObj)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(no buffer bound to GL_ELEMENT_ARRAY_BUFFER)", name);
+      return GL_FALSE;
+   }
+
+   return valid_draw_indirect(ctx, mode, indirect, size, name);
+}
+
+GLboolean
+_mesa_valid_draw_indirect_multi(struct gl_context *ctx,
+                                GLsizei primcount, GLsizei stride,
+                                const char *name)
+{
+
+   /* From the ARB_multi_draw_indirect specification:
+    * "INVALID_VALUE is generated by MultiDrawArraysIndirect or
+    *  MultiDrawElementsIndirect if <primcount> is negative."
+    *
+    * "<primcount> must be positive, otherwise an INVALID_VALUE error will
+    *  be generated."
+    */
+   if (primcount < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "%s(primcount < 0)", name);
+      return GL_FALSE;
+   }
+
+
+   /* From the ARB_multi_draw_indirect specification:
+    * "<stride> must be a multiple of four, otherwise an INVALID_VALUE
+    *  error is generated."
+    */
+   if (stride % 4) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "%s(stride %% 4)", name);
+      return GL_FALSE;
+   }
+
+   return GL_TRUE;
+}
+
+GLboolean
+_mesa_validate_DrawArraysIndirect(struct gl_context *ctx,
+                                  GLenum mode,
+                                  const GLvoid *indirect)
+{
+   const unsigned drawArraysNumParams = 4;
+
+   return valid_draw_indirect(ctx, mode,
+                              indirect, drawArraysNumParams * sizeof(GLuint),
+                              "glDrawArraysIndirect");
+}
+
+GLboolean
+_mesa_validate_DrawElementsIndirect(struct gl_context *ctx,
+                                    GLenum mode, GLenum type,
+                                    const GLvoid *indirect)
+{
+   const unsigned drawElementsNumParams = 5;
+
+   return valid_draw_indirect_elements(ctx, mode, type,
+                                       indirect, drawElementsNumParams * sizeof(GLuint),
+                                       "glDrawElementsIndirect");
+}
+
+GLboolean
+_mesa_validate_MultiDrawArraysIndirect(struct gl_context *ctx,
+                                       GLenum mode,
+                                       const GLvoid *indirect,
+                                       GLsizei primcount, GLsizei stride)
+{
+   GLsizeiptr size = 0;
+   const unsigned drawArraysNumParams = 4;
+
+   /* caller has converted stride==0 to drawArraysNumParams * sizeof(GLuint) */
+   assert(stride != 0);
+
+   if (!_mesa_valid_draw_indirect_multi(ctx, primcount, stride,
+                                        "glMultiDrawArraysIndirect"))
+      return GL_FALSE;
+
+   /* number of bytes of the indirect buffer which will be read */
+   size = primcount
+      ? (primcount - 1) * stride + drawArraysNumParams * sizeof(GLuint)
+      : 0;
+
+   if (!valid_draw_indirect(ctx, mode, indirect, size,
+                            "glMultiDrawArraysIndirect"))
+      return GL_FALSE;
+
+   return GL_TRUE;
+}
+
+GLboolean
+_mesa_validate_MultiDrawElementsIndirect(struct gl_context *ctx,
+                                         GLenum mode, GLenum type,
+                                         const GLvoid *indirect,
+                                         GLsizei primcount, GLsizei stride)
+{
+   GLsizeiptr size = 0;
+   const unsigned drawElementsNumParams = 5;
+
+   /* caller has converted stride==0 to drawElementsNumParams * sizeof(GLuint) */
+   assert(stride != 0);
+
+   if (!_mesa_valid_draw_indirect_multi(ctx, primcount, stride,
+                                        "glMultiDrawElementsIndirect"))
+      return GL_FALSE;
+
+   /* number of bytes of the indirect buffer which will be read */
+   size = primcount
+      ? (primcount - 1) * stride + drawElementsNumParams * sizeof(GLuint)
+      : 0;
+
+   if (!valid_draw_indirect_elements(ctx, mode, type,
+                                     indirect, size,
+                                     "glMultiDrawElementsIndirect"))
+      return GL_FALSE;
+
+   return GL_TRUE;
+}
+
+static GLboolean
+valid_draw_indirect_parameters(struct gl_context *ctx,
+                               const char *name,
+                               GLintptr drawcount)
+{
+   /* From the ARB_indirect_parameters specification:
+    * "INVALID_VALUE is generated by MultiDrawArraysIndirectCountARB or
+    *  MultiDrawElementsIndirectCountARB if <drawcount> is not a multiple of
+    *  four."
+    */
+   if (drawcount & 3) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "%s(drawcount is not a multiple of 4)", name);
+      return GL_FALSE;
+   }
+
+   /* From the ARB_indirect_parameters specification:
+    * "INVALID_OPERATION is generated by MultiDrawArraysIndirectCountARB or
+    *  MultiDrawElementsIndirectCountARB if no buffer is bound to the
+    *  PARAMETER_BUFFER_ARB binding point."
+    */
+   if (!_mesa_is_bufferobj(ctx->ParameterBuffer)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s: no buffer bound to PARAMETER_BUFFER", name);
+      return GL_FALSE;
+   }
+
+   if (_mesa_check_disallowed_mapping(ctx->ParameterBuffer)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(PARAMETER_BUFFER is mapped)", name);
+      return GL_FALSE;
+   }
+
+   /* From the ARB_indirect_parameters specification:
+    * "INVALID_OPERATION is generated by MultiDrawArraysIndirectCountARB or
+    *  MultiDrawElementsIndirectCountARB if reading a <sizei> typed value
+    *  from the buffer bound to the PARAMETER_BUFFER_ARB target at the offset
+    *  specified by <drawcount> would result in an out-of-bounds access."
+    */
+   if (ctx->ParameterBuffer->Size < drawcount + sizeof(GLsizei)) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "%s(PARAMETER_BUFFER too small)", name);
+      return GL_FALSE;
+   }
+
+   return GL_TRUE;
+}
+
+GLboolean
+_mesa_validate_MultiDrawArraysIndirectCount(struct gl_context *ctx,
+                                            GLenum mode,
+                                            GLintptr indirect,
+                                            GLintptr drawcount,
+                                            GLsizei maxdrawcount,
+                                            GLsizei stride)
+{
+   GLsizeiptr size = 0;
+   const unsigned drawArraysNumParams = 4;
+
+   /* caller has converted stride==0 to drawArraysNumParams * sizeof(GLuint) */
+   assert(stride != 0);
+
+   if (!_mesa_valid_draw_indirect_multi(ctx, maxdrawcount, stride,
+                                        "glMultiDrawArraysIndirectCountARB"))
+      return GL_FALSE;
+
+   /* number of bytes of the indirect buffer which will be read */
+   size = maxdrawcount
+      ? (maxdrawcount - 1) * stride + drawArraysNumParams * sizeof(GLuint)
+      : 0;
+
+   if (!valid_draw_indirect(ctx, mode, (void *)indirect, size,
+                            "glMultiDrawArraysIndirectCountARB"))
+      return GL_FALSE;
+
+   return valid_draw_indirect_parameters(
+         ctx, "glMultiDrawArraysIndirectCountARB", drawcount);
+}
+
+GLboolean
+_mesa_validate_MultiDrawElementsIndirectCount(struct gl_context *ctx,
+                                              GLenum mode, GLenum type,
+                                              GLintptr indirect,
+                                              GLintptr drawcount,
+                                              GLsizei maxdrawcount,
+                                              GLsizei stride)
+{
+   GLsizeiptr size = 0;
+   const unsigned drawElementsNumParams = 5;
+
+   /* caller has converted stride==0 to drawElementsNumParams * sizeof(GLuint) */
+   assert(stride != 0);
+
+   if (!_mesa_valid_draw_indirect_multi(ctx, maxdrawcount, stride,
+                                        "glMultiDrawElementsIndirectCountARB"))
+      return GL_FALSE;
+
+   /* number of bytes of the indirect buffer which will be read */
+   size = maxdrawcount
+      ? (maxdrawcount - 1) * stride + drawElementsNumParams * sizeof(GLuint)
+      : 0;
+
+   if (!valid_draw_indirect_elements(ctx, mode, type,
+                                     (void *)indirect, size,
+                                     "glMultiDrawElementsIndirectCountARB"))
+      return GL_FALSE;
+
+   return valid_draw_indirect_parameters(
+         ctx, "glMultiDrawElementsIndirectCountARB", drawcount);
+}
diff --git a/lib/mesa/src/mesa/main/draw_validate.h b/lib/mesa/src/mesa/main/draw_validate.h
new file mode 100644
index 000000000..d015c7e83
--- /dev/null
+++ b/lib/mesa/src/mesa/main/draw_validate.h
@@ -0,0 +1,134 @@
+
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 1999-2001  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+#ifndef API_VALIDATE_H
+#define API_VALIDATE_H
+
+#include <stdbool.h>
+#include "glheader.h"
+
+struct gl_buffer_object;
+struct gl_context;
+struct gl_transform_feedback_object;
+
+
+extern GLboolean
+_mesa_valid_to_render(struct gl_context *ctx, const char *where);
+
+extern bool
+_mesa_is_valid_prim_mode(const struct gl_context *ctx, GLenum mode);
+
+extern GLboolean
+_mesa_valid_prim_mode(struct gl_context *ctx, GLenum mode, const char *name);
+
+extern GLboolean
+_mesa_valid_draw_indirect_multi(struct gl_context *ctx, GLsizei primcount,
+                                GLsizei stride, const char *name);
+
+extern GLboolean
+_mesa_validate_DrawArrays(struct gl_context *ctx, GLenum mode, GLsizei count);
+
+extern bool
+_mesa_validate_MultiDrawArrays(struct gl_context *ctx, GLenum mode,
+                               const GLsizei *count, GLsizei primcount);
+
+extern GLboolean
+_mesa_validate_DrawElements(struct gl_context *ctx,
+			    GLenum mode, GLsizei count, GLenum type,
+			    const GLvoid *indices);
+
+extern GLboolean
+_mesa_validate_MultiDrawElements(struct gl_context *ctx,
+                                 GLenum mode, const GLsizei *count,
+                                 GLenum type, const GLvoid * const *indices,
+                                 GLsizei primcount);
+
+extern GLboolean
+_mesa_validate_DrawRangeElements(struct gl_context *ctx, GLenum mode,
+				 GLuint start, GLuint end,
+				 GLsizei count, GLenum type,
+				 const GLvoid *indices);
+
+
+extern GLboolean
+_mesa_validate_DrawArraysInstanced(struct gl_context *ctx, GLenum mode, GLint first,
+                                   GLsizei count, GLsizei primcount);
+
+extern GLboolean
+_mesa_validate_DrawElementsInstanced(struct gl_context *ctx,
+                                     GLenum mode, GLsizei count, GLenum type,
+                                     const GLvoid *indices, GLsizei primcount);
+
+extern GLboolean
+_mesa_validate_DrawTransformFeedback(struct gl_context *ctx,
+                                     GLenum mode,
+                                     struct gl_transform_feedback_object *obj,
+                                     GLuint stream,
+                                     GLsizei numInstances);
+
+extern GLboolean
+_mesa_validate_DrawArraysIndirect(struct gl_context *ctx,
+                                  GLenum mode,
+                                  const GLvoid *indirect);
+
+extern GLboolean
+_mesa_validate_DrawElementsIndirect(struct gl_context *ctx,
+                                    GLenum mode,
+                                    GLenum type,
+                                    const GLvoid *indirect);
+
+extern GLboolean
+_mesa_validate_MultiDrawArraysIndirect(struct gl_context *ctx,
+                                       GLenum mode,
+                                       const GLvoid *indirect,
+                                       GLsizei primcount,
+                                       GLsizei stride);
+
+extern GLboolean
+_mesa_validate_MultiDrawElementsIndirect(struct gl_context *ctx,
+                                         GLenum mode,
+                                         GLenum type,
+                                         const GLvoid *indirect,
+                                         GLsizei primcount,
+                                         GLsizei stride);
+
+extern GLboolean
+_mesa_validate_MultiDrawArraysIndirectCount(struct gl_context *ctx,
+                                            GLenum mode,
+                                            GLintptr indirect,
+                                            GLintptr drawcount,
+                                            GLsizei maxdrawcount,
+                                            GLsizei stride);
+
+extern GLboolean
+_mesa_validate_MultiDrawElementsIndirectCount(struct gl_context *ctx,
+                                              GLenum mode, GLenum type,
+                                              GLintptr indirect,
+                                              GLintptr drawcount,
+                                              GLsizei maxdrawcount,
+                                              GLsizei stride);
+
+#endif
diff --git a/lib/mesa/src/mesa/main/drawpix.c b/lib/mesa/src/mesa/main/drawpix.c
index ec1d2618c..2f55dde7b 100644
--- a/lib/mesa/src/mesa/main/drawpix.c
+++ b/lib/mesa/src/mesa/main/drawpix.c
@@ -24,7 +24,7 @@
 
 #include "glheader.h"
 #include "imports.h"
-#include "api_validate.h"
+#include "draw_validate.h"
 #include "bufferobj.h"
 #include "context.h"
 #include "drawpix.h"
@@ -34,7 +34,6 @@
 #include "image.h"
 #include "pbo.h"
 #include "state.h"
-#include "dispatch.h"
 #include "glformats.h"
 #include "fbobject.h"
 
diff --git a/lib/mesa/src/mesa/main/drawtex.c b/lib/mesa/src/mesa/main/drawtex.c
index 9c4fdf90e..6114b3277 100644
--- a/lib/mesa/src/mesa/main/drawtex.c
+++ b/lib/mesa/src/mesa/main/drawtex.c
@@ -21,6 +21,7 @@
  * DEALINGS IN THE SOFTWARE.
  */
 
+#include "main/errors.h"
 #include "main/drawtex.h"
 #include "main/state.h"
 #include "main/imports.h"
diff --git a/lib/mesa/src/mesa/main/errors.h b/lib/mesa/src/mesa/main/errors.h
index 5ad5254ff..5911da295 100644
--- a/lib/mesa/src/mesa/main/errors.h
+++ b/lib/mesa/src/mesa/main/errors.h
@@ -38,15 +38,16 @@
 
 #include <stdio.h>
 #include <stdarg.h>
-#include "compiler.h"
 #include "glheader.h"
-#include "mtypes.h"
+#include "menums.h"
 
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
+struct gl_context;
+
 extern void
 _mesa_warning( struct gl_context *gc, const char *fmtString, ... ) PRINTFLIKE(2, 3);
 
diff --git a/lib/mesa/src/mesa/main/eval.h b/lib/mesa/src/mesa/main/eval.h
index 5b2fce119..938e3579b 100644
--- a/lib/mesa/src/mesa/main/eval.h
+++ b/lib/mesa/src/mesa/main/eval.h
@@ -37,7 +37,7 @@
 #define EVAL_H
 
 
-#include "main/mtypes.h"
+#include "dd.h"
 #include <stdbool.h>
 
 
diff --git a/lib/mesa/src/mesa/main/extensions.c b/lib/mesa/src/mesa/main/extensions.c
index f185aa5aa..4d95a0727 100644
--- a/lib/mesa/src/mesa/main/extensions.c
+++ b/lib/mesa/src/mesa/main/extensions.c
@@ -39,7 +39,7 @@
 
 struct gl_extensions _mesa_extension_override_enables;
 struct gl_extensions _mesa_extension_override_disables;
-static char *extra_extensions = NULL;
+static char *unrecognized_extensions = NULL;
 
 
 /**
@@ -48,8 +48,6 @@ static char *extra_extensions = NULL;
  */
 #define o(x) offsetof(struct gl_extensions, x)
 
-static bool disabled_extensions[MESA_EXTENSION_COUNT];
-
 /**
  * Given an extension name, lookup up the corresponding member of struct
  * gl_extensions and return that member's index.  If the name is
@@ -78,8 +76,8 @@ name_to_index(const char* name)
  * Overrides extensions in \c ctx based on the values in
  * _mesa_extension_override_enables and _mesa_extension_override_disables.
  */
-static void
-override_extensions_in_context(struct gl_context *ctx)
+void
+_mesa_override_extensions(struct gl_context *ctx)
 {
    unsigned i;
    const GLboolean *enables =
@@ -131,9 +129,7 @@ _mesa_enable_sw_extensions(struct gl_context *ctx)
    ctx->Extensions.ARB_texture_env_crossbar = GL_TRUE;
    ctx->Extensions.ARB_texture_env_dot3 = GL_TRUE;
    ctx->Extensions.ARB_texture_filter_anisotropic = GL_TRUE;
-#ifdef TEXTURE_FLOAT_ENABLED
    ctx->Extensions.ARB_texture_float = GL_TRUE;
-#endif
    ctx->Extensions.ARB_texture_mirror_clamp_to_edge = GL_TRUE;
    ctx->Extensions.ARB_texture_non_power_of_two = GL_TRUE;
    ctx->Extensions.ARB_texture_rg = GL_TRUE;
@@ -146,7 +142,6 @@ _mesa_enable_sw_extensions(struct gl_context *ctx)
    ctx->Extensions.ATI_texture_compression_3dc = GL_TRUE;
    ctx->Extensions.ATI_texture_env_combine3 = GL_TRUE;
    ctx->Extensions.ATI_texture_mirror_once = GL_TRUE;
-   ctx->Extensions.ATI_separate_stencil = GL_TRUE;
    ctx->Extensions.EXT_blend_color = GL_TRUE;
    ctx->Extensions.EXT_blend_equation_separate = GL_TRUE;
    ctx->Extensions.EXT_blend_func_separate = GL_TRUE;
@@ -198,66 +193,40 @@ set_extension(struct gl_extensions *ext, int i, GLboolean state)
    return offset;
 }
 
-/**
- * \brief Apply the \c MESA_EXTENSION_OVERRIDE environment variable.
- *
- * \c MESA_EXTENSION_OVERRIDE is a space-separated list of extensions to
- * enable or disable. The list is processed thus:
- *    - Enable recognized extension names that are prefixed with '+'.
- *    - Disable recognized extension names that are prefixed with '-'.
- *    - Enable recognized extension names that are not prefixed.
- *    - Collect unrecognized extension names in a new string.
- *
- * \c MESA_EXTENSION_OVERRIDE was previously parsed during
- * _mesa_one_time_init_extension_overrides. We just use the results of that
- * parsing in this function.
- *
- * \return Space-separated list of unrecognized extension names (which must
- *    be freed). Does not return \c NULL.
- */
-static char *
-get_extension_override( struct gl_context *ctx )
-{
-   override_extensions_in_context(ctx);
-
-   if (extra_extensions == NULL) {
-      return calloc(1, sizeof(char));
-   } else {
-      _mesa_problem(ctx, "Trying to enable unknown extensions: %s",
-                    extra_extensions);
-      return strdup(extra_extensions);
-   }
-}
-
 
 /**
- * \brief Free extra_extensions string
+ * \brief Free string pointed by unrecognized_extensions
  *
- * These strings are allocated early during the first context creation by
+ * This string is allocated early during the first context creation by
  * _mesa_one_time_init_extension_overrides.
  */
 static void
 free_unknown_extensions_strings(void)
 {
-   free(extra_extensions);
+   free(unrecognized_extensions);
 }
 
 
 /**
- * \brief Initialize extension override tables.
+ * \brief Initialize extension override tables based on \c MESA_EXTENSION_OVERRIDE
  *
  * This should be called one time early during first context initialization.
+
+ * \c MESA_EXTENSION_OVERRIDE is a space-separated list of extensions to
+ * enable or disable. The list is processed thus:
+ *    - Enable recognized extension names that are prefixed with '+'.
+ *    - Disable recognized extension names that are prefixed with '-'.
+ *    - Enable recognized extension names that are not prefixed.
+ *    - Collect unrecognized extension names in a new string.
  */
 void
-_mesa_one_time_init_extension_overrides(void)
+_mesa_one_time_init_extension_overrides(struct gl_context *ctx)
 {
    const char *env_const = getenv("MESA_EXTENSION_OVERRIDE");
    char *env;
    char *ext;
-   int len;
    size_t offset;
-
-   atexit(free_unknown_extensions_strings);
+   unsigned unknown_ext = 0;
 
    memset(&_mesa_extension_override_enables, 0, sizeof(struct gl_extensions));
    memset(&_mesa_extension_override_disables, 0, sizeof(struct gl_extensions));
@@ -266,18 +235,11 @@ _mesa_one_time_init_extension_overrides(void)
       return;
    }
 
-   /* extra_exts: List of unrecognized extensions. */
-   extra_extensions = calloc(ALIGN(strlen(env_const) + 2, 4), sizeof(char));
-
    /* Copy env_const because strtok() is destructive. */
    env = strdup(env_const);
 
-   if (env == NULL ||
-       extra_extensions == NULL) {
-      free(env);
-      free(extra_extensions);
+   if (env == NULL)
       return;
-   }
 
    for (ext = strtok(env, " "); ext != NULL; ext = strtok(NULL, " ")) {
       int enable;
@@ -299,31 +261,36 @@ _mesa_one_time_init_extension_overrides(void)
 
       i = name_to_index(ext);
       offset = set_extension(&_mesa_extension_override_enables, i, enable);
-      if (offset != 0 && (offset != o(dummy_true) || enable != GL_FALSE)) {
-         ((GLboolean *) &_mesa_extension_override_disables)[offset] = !enable;
+      offset = set_extension(&_mesa_extension_override_disables, i, !enable);
+      if (offset != 0)
          recognized = true;
-      } else {
+      else
          recognized = false;
-      }
-
-      if (i >= 0)
-         disabled_extensions[i] = !enable;
 
       if (!recognized && enable) {
-         strcat(extra_extensions, ext);
-         strcat(extra_extensions, " ");
+         if (unknown_ext >= MAX_UNRECOGNIZED_EXTENSIONS) {
+            static bool warned;
+
+            if (!warned) {
+               warned = true;
+               _mesa_problem(ctx, "Trying to enable too many unknown extension. "
+                                  "Only the first %d will be honoured",
+                                  MAX_UNRECOGNIZED_EXTENSIONS);
+            }
+         } else {
+            ctx->Extensions.unrecognized_extensions[unknown_ext] = ext;
+            unknown_ext++;
+
+            _mesa_problem(ctx, "Trying to enable unknown extension: %s", ext);
+         }
       }
    }
 
-   free(env);
-
-   /* Remove trailing space, and free if unused. */
-   len = strlen(extra_extensions);
-   if (len == 0) {
-      free(extra_extensions);
-      extra_extensions = NULL;
-   } else if (extra_extensions[len - 1] == ' ') {
-      extra_extensions[len - 1] = '\0';
+   if (!unknown_ext) {
+      free(env);
+   } else {
+      unrecognized_extensions = env;
+      atexit(free_unknown_extensions_strings);
    }
 }
 
@@ -365,8 +332,7 @@ _mesa_extension_supported(const struct gl_context *ctx, extension_index i)
    const bool *base = (bool *) &ctx->Extensions;
    const struct mesa_extension *ext = _mesa_extension_table + i;
 
-   return !disabled_extensions[i] &&
-          (ctx->Version >= ext->version[ctx->API]) && base[ext->offset];
+   return (ctx->Version >= ext->version[ctx->API]) && base[ext->offset];
 }
 
 /**
@@ -408,9 +374,7 @@ _mesa_make_extension_string(struct gl_context *ctx)
    /* Number of extensions */
    unsigned count;
    /* Indices of the extensions sorted by year */
-   extension_index *extension_indices;
-   /* String of extra extensions. */
-   char *extra_extensions = get_extension_override(ctx);
+   extension_index extension_indices[MESA_EXTENSION_COUNT];
    unsigned k;
    unsigned j;
    unsigned maxYear = ~0;
@@ -436,26 +400,21 @@ _mesa_make_extension_string(struct gl_context *ctx)
 	 ++count;
       }
    }
-   if (extra_extensions != NULL)
-      length += 1 + strlen(extra_extensions); /* +1 for space */
+   for (k = 0; k < MAX_UNRECOGNIZED_EXTENSIONS; k++)
+      if (ctx->Extensions.unrecognized_extensions[k])
+         length += 1 + strlen(ctx->Extensions.unrecognized_extensions[k]); /* +1 for space */
 
    exts = calloc(ALIGN(length + 1, 4), sizeof(char));
    if (exts == NULL) {
-      free(extra_extensions);
       return NULL;
    }
 
-   extension_indices = malloc(count * sizeof(extension_index));
-   if (extension_indices == NULL) {
-      free(exts);
-      free(extra_extensions);
-      return NULL;
-   }
-
-   /* Sort extensions in chronological order because certain old applications
-    * (e.g., Quake3 demo) store the extension list in a static size buffer so
-    * chronologically order ensure that the extensions that such applications
-    * expect will fit into that buffer.
+   /* Sort extensions in chronological order because idTech 2/3 games
+    * (e.g., Quake3 demo) store the extension list in a fixed size buffer.
+    * Some cases truncate, while others overflow the buffer. Resulting in
+    * misrendering and crashes, respectively.
+    * Address the former here, while the latter will be addressed by setting
+    * the MESA_EXTENSION_MAX_YEAR environment variable.
     */
    j = 0;
    for (k = 0; k < MESA_EXTENSION_COUNT; ++k) {
@@ -475,10 +434,11 @@ _mesa_make_extension_string(struct gl_context *ctx)
       strcat(exts, i->name);
       strcat(exts, " ");
    }
-   free(extension_indices);
-   if (extra_extensions != 0) {
-      strcat(exts, extra_extensions);
-      free(extra_extensions);
+   for (j = 0; j < MAX_UNRECOGNIZED_EXTENSIONS; j++) {
+      if (ctx->Extensions.unrecognized_extensions[j]) {
+         strcat(exts, ctx->Extensions.unrecognized_extensions[j]);
+         strcat(exts, " ");
+      }
    }
 
    return (GLubyte *) exts;
@@ -500,6 +460,11 @@ _mesa_get_extension_count(struct gl_context *ctx)
       if (_mesa_extension_supported(ctx, k))
 	 ctx->Extensions.Count++;
    }
+
+   for (k = 0; k < MAX_UNRECOGNIZED_EXTENSIONS; ++k) {
+      if (ctx->Extensions.unrecognized_extensions[k])
+	 ctx->Extensions.Count++;
+   }
    return ctx->Extensions.Count;
 }
 
@@ -521,5 +486,13 @@ _mesa_get_enabled_extension(struct gl_context *ctx, GLuint index)
       }
    }
 
+   for (i = 0; i < MAX_UNRECOGNIZED_EXTENSIONS; ++i) {
+      if (ctx->Extensions.unrecognized_extensions[i]) {
+         if (n == index)
+            return (const GLubyte*) ctx->Extensions.unrecognized_extensions[i];
+         else
+            ++n;
+      }
+   }
    return NULL;
 }
diff --git a/lib/mesa/src/mesa/main/extensions.h b/lib/mesa/src/mesa/main/extensions.h
index efef1be47..705d65dcd 100644
--- a/lib/mesa/src/mesa/main/extensions.h
+++ b/lib/mesa/src/mesa/main/extensions.h
@@ -36,7 +36,7 @@
 #ifndef _EXTENSIONS_H_
 #define _EXTENSIONS_H_
 
-#include "glheader.h"
+#include "mtypes.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -47,12 +47,14 @@ struct gl_extensions;
 
 extern void _mesa_enable_sw_extensions(struct gl_context *ctx);
 
-extern void _mesa_one_time_init_extension_overrides(void);
+extern void _mesa_one_time_init_extension_overrides(struct gl_context *ctx);
 
 extern void _mesa_init_extensions(struct gl_extensions *extentions);
 
 extern GLubyte *_mesa_make_extension_string(struct gl_context *ctx);
 
+extern void _mesa_override_extensions(struct gl_context *ctx);
+
 extern GLuint
 _mesa_get_extension_count(struct gl_context *ctx);
 
@@ -104,6 +106,15 @@ _mesa_has_##name_str(const struct gl_context *ctx) \
 #include "extensions_table.h"
 #undef EXT
 
+/* Sometimes the driver wants to query the extension override status before
+ * a context is created. These variables are filled with extension override
+ * information before context creation.
+ *
+ * This can be useful during extension bring-up when an extension is
+ * partially implemented, but cannot yet be advertised as supported.
+ *
+ * Use it with care and keep access read-only.
+ */
 extern struct gl_extensions _mesa_extension_override_enables;
 extern struct gl_extensions _mesa_extension_override_disables;
 
diff --git a/lib/mesa/src/mesa/main/extensions_table.c b/lib/mesa/src/mesa/main/extensions_table.c
index 1e37fbcc4..6241705de 100644
--- a/lib/mesa/src/mesa/main/extensions_table.c
+++ b/lib/mesa/src/mesa/main/extensions_table.c
@@ -23,7 +23,6 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
-#include "main/mtypes.h"
 #include "main/extensions.h"
 
 /**
diff --git a/lib/mesa/src/mesa/main/externalobjects.c b/lib/mesa/src/mesa/main/externalobjects.c
index e70280c96..56bf817a6 100644
--- a/lib/mesa/src/mesa/main/externalobjects.c
+++ b/lib/mesa/src/mesa/main/externalobjects.c
@@ -23,6 +23,8 @@
 
 #include "macros.h"
 #include "mtypes.h"
+#include "bufferobj.h"
+#include "context.h"
 #include "externalobjects.h"
 #include "teximage.h"
 #include "texobj.h"
@@ -148,8 +150,7 @@ _mesa_CreateMemoryObjectsEXT(GLsizei n, GLuint *memoryObjects)
       _mesa_debug(ctx, "%s(%d, %p)", func, n, memoryObjects);
 
    if (!ctx->Extensions.EXT_memory_object) {
-      _mesa_error(ctx, GL_INVALID_OPERATION,
-                  "glCreateMemoryObjectsEXT(unsupported)");
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(unsupported)", func);
       return;
    }
 
@@ -195,9 +196,10 @@ _mesa_MemoryObjectParameterivEXT(GLuint memoryObject,
    GET_CURRENT_CONTEXT(ctx);
    struct gl_memory_object *memObj;
 
+   const char *func = "glMemoryObjectParameterivEXT";
+
    if (!ctx->Extensions.EXT_memory_object) {
-      _mesa_error(ctx, GL_INVALID_OPERATION,
-                  "glMemoryObjectParameterivEXT(unsupported)");
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(unsupported)", func);
       return;
    }
 
@@ -206,8 +208,7 @@ _mesa_MemoryObjectParameterivEXT(GLuint memoryObject,
       return;
 
    if (memObj->Immutable) {
-      _mesa_error(ctx, GL_INVALID_OPERATION,
-                  "glMemoryObjectParameterivEXT(memoryObject is immutable");
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(memoryObject is immutable", func);
       return;
    }
 
@@ -224,8 +225,7 @@ _mesa_MemoryObjectParameterivEXT(GLuint memoryObject,
    return;
 
 invalid_pname:
-   _mesa_error(ctx, GL_INVALID_ENUM,
-               "glMemoryObjectParameterivEXT(pname=0x%x)", pname);
+   _mesa_error(ctx, GL_INVALID_ENUM, "%s(pname=0x%x)", func, pname);
 }
 
 void GLAPIENTRY
@@ -236,9 +236,10 @@ _mesa_GetMemoryObjectParameterivEXT(GLuint memoryObject,
    GET_CURRENT_CONTEXT(ctx);
    struct gl_memory_object *memObj;
 
+   const char *func = "glMemoryObjectParameterivEXT";
+
    if (!ctx->Extensions.EXT_memory_object) {
-      _mesa_error(ctx, GL_INVALID_OPERATION,
-                  "glGetMemoryObjectParameterivEXT(unsupported)");
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(unsupported)", func);
       return;
    }
 
@@ -259,8 +260,7 @@ _mesa_GetMemoryObjectParameterivEXT(GLuint memoryObject,
    return;
 
 invalid_pname:
-   _mesa_error(ctx, GL_INVALID_ENUM,
-               "glGetMemoryObjectParameterivEXT(pname=0x%x)", pname);
+   _mesa_error(ctx, GL_INVALID_ENUM, "%s(pname=0x%x)", func, pname);
 }
 
 static struct gl_memory_object *
@@ -547,22 +547,144 @@ _mesa_TextureStorageMem1DEXT(GLuint texture,
                          memory, offset, "glTextureStorageMem1DEXT");
 }
 
+/**
+ * Used as a placeholder for semaphore objects between glGenSemaphoresEXT()
+ * and glImportSemaphoreFdEXT(), so that glIsSemaphoreEXT() can work correctly.
+ */
+static struct gl_semaphore_object DummySemaphoreObject;
+
+/**
+ * Delete a semaphore object.  Called via ctx->Driver.DeleteSemaphore().
+ * Not removed from hash table here.
+ */
+void
+_mesa_delete_semaphore_object(struct gl_context *ctx,
+                              struct gl_semaphore_object *semObj)
+{
+   if (semObj != &DummySemaphoreObject)
+      free(semObj);
+}
+
+/**
+ * Initialize a semaphore object to default values.
+ */
+void
+_mesa_initialize_semaphore_object(struct gl_context *ctx,
+                                  struct gl_semaphore_object *obj,
+                                  GLuint name)
+{
+   memset(obj, 0, sizeof(struct gl_semaphore_object));
+   obj->Name = name;
+}
+
 void GLAPIENTRY
 _mesa_GenSemaphoresEXT(GLsizei n, GLuint *semaphores)
 {
+   GET_CURRENT_CONTEXT(ctx);
+
+   const char *func = "glGenSemaphoresEXT";
+
+   if (MESA_VERBOSE & (VERBOSE_API))
+      _mesa_debug(ctx, "%s(%d, %p)", func, n, semaphores);
+
+   if (!ctx->Extensions.EXT_semaphore) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(unsupported)", func);
+      return;
+   }
+
+   if (n < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "%s(n < 0)", func);
+      return;
+   }
+
+   if (!semaphores)
+      return;
+
+   _mesa_HashLockMutex(ctx->Shared->SemaphoreObjects);
+   GLuint first = _mesa_HashFindFreeKeyBlock(ctx->Shared->SemaphoreObjects, n);
+   if (first) {
+      for (GLsizei i = 0; i < n; i++) {
+         semaphores[i] = first + i;
+         _mesa_HashInsertLocked(ctx->Shared->SemaphoreObjects,
+                                semaphores[i], &DummySemaphoreObject);
+      }
+   }
 
+   _mesa_HashUnlockMutex(ctx->Shared->SemaphoreObjects);
 }
 
 void GLAPIENTRY
 _mesa_DeleteSemaphoresEXT(GLsizei n, const GLuint *semaphores)
 {
+   GET_CURRENT_CONTEXT(ctx);
+
+   const char *func = "glDeleteSemaphoresEXT";
+
+   if (MESA_VERBOSE & (VERBOSE_API)) {
+      _mesa_debug(ctx, "%s(%d, %p)\n", func, n, semaphores);
+   }
+
+   if (!ctx->Extensions.EXT_semaphore) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(unsupported)", func);
+      return;
+   }
+
+   if (n < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "%s(n < 0)", func);
+      return;
+   }
+
+   if (!semaphores)
+      return;
+
+   _mesa_HashLockMutex(ctx->Shared->SemaphoreObjects);
+   for (GLint i = 0; i < n; i++) {
+      if (semaphores[i] > 0) {
+         struct gl_semaphore_object *delObj
+            = _mesa_lookup_semaphore_object_locked(ctx, semaphores[i]);
 
+         if (delObj) {
+            _mesa_HashRemoveLocked(ctx->Shared->SemaphoreObjects,
+                                   semaphores[i]);
+            ctx->Driver.DeleteSemaphoreObject(ctx, delObj);
+         }
+      }
+   }
+   _mesa_HashUnlockMutex(ctx->Shared->SemaphoreObjects);
 }
 
 GLboolean GLAPIENTRY
 _mesa_IsSemaphoreEXT(GLuint semaphore)
 {
-   return GL_FALSE;
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (!ctx->Extensions.EXT_semaphore) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glIsSemaphoreEXT(unsupported)");
+      return GL_FALSE;
+   }
+
+   struct gl_semaphore_object *obj =
+      _mesa_lookup_semaphore_object(ctx, semaphore);
+
+   return obj ? GL_TRUE : GL_FALSE;
+}
+
+/**
+ * Helper that outputs the correct error status for parameter
+ * calls where no pnames are defined
+ */
+static void
+semaphore_parameter_stub(const char* func, GLenum pname)
+{
+   GET_CURRENT_CONTEXT(ctx);
+
+   if (!ctx->Extensions.EXT_semaphore) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(unsupported)", func);
+      return;
+   }
+
+   /* EXT_semaphore and EXT_semaphore_fd define no parameters */
+   _mesa_error(ctx, GL_INVALID_ENUM, "%s(pname=0x%x)", func, pname);
 }
 
 void GLAPIENTRY
@@ -570,7 +692,9 @@ _mesa_SemaphoreParameterui64vEXT(GLuint semaphore,
                                  GLenum pname,
                                  const GLuint64 *params)
 {
+   const char *func = "glSemaphoreParameterui64vEXT";
 
+   semaphore_parameter_stub(func, pname);
 }
 
 void GLAPIENTRY
@@ -578,7 +702,9 @@ _mesa_GetSemaphoreParameterui64vEXT(GLuint semaphore,
                                     GLenum pname,
                                     GLuint64 *params)
 {
+   const char *func = "glGetSemaphoreParameterui64vEXT";
 
+   semaphore_parameter_stub(func, pname);
 }
 
 void GLAPIENTRY
@@ -589,7 +715,57 @@ _mesa_WaitSemaphoreEXT(GLuint semaphore,
                        const GLuint *textures,
                        const GLenum *srcLayouts)
 {
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_semaphore_object *semObj = NULL;
+   struct gl_buffer_object **bufObjs = NULL;
+   struct gl_texture_object **texObjs = NULL;
+
+   const char *func = "glWaitSemaphoreEXT";
+
+   if (!ctx->Extensions.EXT_semaphore) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(unsupported)", func);
+      return;
+   }
+
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   semObj = _mesa_lookup_semaphore_object(ctx, semaphore);
+   if (!semObj)
+      return;
+
+   FLUSH_VERTICES(ctx, 0);
+   FLUSH_CURRENT(ctx, 0);
+
+   bufObjs = malloc(sizeof(struct gl_buffer_object *) * numBufferBarriers);
+   if (!bufObjs) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "%s(numBufferBarriers=%u)",
+                  func, numBufferBarriers);
+      goto end;
+   }
 
+   for (unsigned i = 0; i < numBufferBarriers; i++) {
+      bufObjs[i] = _mesa_lookup_bufferobj(ctx, buffers[i]);
+   }
+
+   texObjs = malloc(sizeof(struct gl_texture_object *) * numTextureBarriers);
+   if (!texObjs) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "%s(numTextureBarriers=%u)",
+                  func, numTextureBarriers);
+      goto end;
+   }
+
+   for (unsigned i = 0; i < numTextureBarriers; i++) {
+      texObjs[i] = _mesa_lookup_texture(ctx, textures[i]);
+   }
+
+   ctx->Driver.ServerWaitSemaphoreObject(ctx, semObj,
+                                         numBufferBarriers, bufObjs,
+                                         numTextureBarriers, texObjs,
+                                         srcLayouts);
+
+end:
+   free(bufObjs);
+   free(texObjs);
 }
 
 void GLAPIENTRY
@@ -600,7 +776,57 @@ _mesa_SignalSemaphoreEXT(GLuint semaphore,
                          const GLuint *textures,
                          const GLenum *dstLayouts)
 {
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_semaphore_object *semObj = NULL;
+   struct gl_buffer_object **bufObjs = NULL;
+   struct gl_texture_object **texObjs = NULL;
+
+   const char *func = "glSignalSemaphoreEXT";
 
+   if (!ctx->Extensions.EXT_semaphore) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(unsupported)", func);
+      return;
+   }
+
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   semObj = _mesa_lookup_semaphore_object(ctx, semaphore);
+   if (!semObj)
+      return;
+
+   FLUSH_VERTICES(ctx, 0);
+   FLUSH_CURRENT(ctx, 0);
+
+   bufObjs = malloc(sizeof(struct gl_buffer_object *) * numBufferBarriers);
+   if (!bufObjs) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "%s(numBufferBarriers=%u)",
+                  func, numBufferBarriers);
+      goto end;
+   }
+
+   for (unsigned i = 0; i < numBufferBarriers; i++) {
+      bufObjs[i] = _mesa_lookup_bufferobj(ctx, buffers[i]);
+   }
+
+   texObjs = malloc(sizeof(struct gl_texture_object *) * numTextureBarriers);
+   if (!texObjs) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "%s(numTextureBarriers=%u)",
+                  func, numTextureBarriers);
+      goto end;
+   }
+
+   for (unsigned i = 0; i < numTextureBarriers; i++) {
+      texObjs[i] = _mesa_lookup_texture(ctx, textures[i]);
+   }
+
+   ctx->Driver.ServerSignalSemaphoreObject(ctx, semObj,
+                                           numBufferBarriers, bufObjs,
+                                           numTextureBarriers, texObjs,
+                                           dstLayouts);
+
+end:
+   free(bufObjs);
+   free(texObjs);
 }
 
 void GLAPIENTRY
@@ -611,15 +837,15 @@ _mesa_ImportMemoryFdEXT(GLuint memory,
 {
    GET_CURRENT_CONTEXT(ctx);
 
+   const char *func = "glImportMemoryFdEXT";
+
    if (!ctx->Extensions.EXT_memory_object_fd) {
-      _mesa_error(ctx, GL_INVALID_OPERATION,
-                  "glImportMemoryFdEXT(unsupported)");
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(unsupported)", func);
       return;
    }
 
    if (handleType != GL_HANDLE_TYPE_OPAQUE_FD_EXT) {
-      _mesa_error(ctx, GL_INVALID_VALUE, "glImportMemoryFdEXT(handleType=%u)",
-                  handleType);
+      _mesa_error(ctx, GL_INVALID_ENUM, "%s(handleType=%u)", func, handleType);
       return;
    }
 
@@ -636,5 +862,33 @@ _mesa_ImportSemaphoreFdEXT(GLuint semaphore,
                            GLenum handleType,
                            GLint fd)
 {
+   GET_CURRENT_CONTEXT(ctx);
+
+   const char *func = "glImportSemaphoreFdEXT";
+
+   if (!ctx->Extensions.EXT_semaphore_fd) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "%s(unsupported)", func);
+      return;
+   }
+
+   if (handleType != GL_HANDLE_TYPE_OPAQUE_FD_EXT) {
+      _mesa_error(ctx, GL_INVALID_ENUM, "%s(handleType=%u)", func, handleType);
+      return;
+   }
+
+   struct gl_semaphore_object *semObj = _mesa_lookup_semaphore_object(ctx,
+                                                                      semaphore);
+   if (!semObj)
+      return;
+
+   if (semObj == &DummySemaphoreObject) {
+      semObj = ctx->Driver.NewSemaphoreObject(ctx, semaphore);
+      if (!semObj) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "%s", func);
+         return;
+      }
+      _mesa_HashInsert(ctx->Shared->SemaphoreObjects, semaphore, semObj);
+   }
 
+   ctx->Driver.ImportSemaphoreFd(ctx, semObj, fd);
 }
diff --git a/lib/mesa/src/mesa/main/externalobjects.h b/lib/mesa/src/mesa/main/externalobjects.h
index a9a12b821..75b0bc2ed 100644
--- a/lib/mesa/src/mesa/main/externalobjects.h
+++ b/lib/mesa/src/mesa/main/externalobjects.h
@@ -57,6 +57,26 @@ _mesa_lookup_memory_object_locked(struct gl_context *ctx, GLuint memory)
       _mesa_HashLookupLocked(ctx->Shared->MemoryObjects, memory);
 }
 
+static inline struct gl_semaphore_object *
+_mesa_lookup_semaphore_object(struct gl_context *ctx, GLuint semaphore)
+{
+   if (!semaphore)
+      return NULL;
+
+   return (struct gl_semaphore_object *)
+      _mesa_HashLookup(ctx->Shared->SemaphoreObjects, semaphore);
+}
+
+static inline struct gl_semaphore_object *
+_mesa_lookup_semaphore_object_locked(struct gl_context *ctx, GLuint semaphore)
+{
+   if (!semaphore)
+      return NULL;
+
+   return (struct gl_semaphore_object *)
+      _mesa_HashLookupLocked(ctx->Shared->SemaphoreObjects, semaphore);
+}
+
 extern void
 _mesa_init_memory_object_functions(struct dd_function_table *driver);
 
@@ -65,7 +85,16 @@ _mesa_initialize_memory_object(struct gl_context *ctx,
                                struct gl_memory_object *obj,
                                GLuint name);
 extern void
-_mesa_delete_memory_object(struct gl_context *ctx, struct gl_memory_object *mo);
+_mesa_delete_memory_object(struct gl_context *ctx,
+                           struct gl_memory_object *semObj);
+
+extern void
+_mesa_initialize_semaphore_object(struct gl_context *ctx,
+                                  struct gl_semaphore_object *obj,
+                                  GLuint name);
+extern void
+_mesa_delete_semaphore_object(struct gl_context *ctx,
+                              struct gl_semaphore_object *semObj);
 
 extern void GLAPIENTRY
 _mesa_DeleteMemoryObjectsEXT(GLsizei n, const GLuint *memoryObjects);
diff --git a/lib/mesa/src/mesa/main/feedback.c b/lib/mesa/src/mesa/main/feedback.c
index 699e2a855..65f4c35d9 100644
--- a/lib/mesa/src/mesa/main/feedback.c
+++ b/lib/mesa/src/mesa/main/feedback.c
@@ -35,7 +35,6 @@
 #include "feedback.h"
 #include "macros.h"
 #include "mtypes.h"
-#include "main/dispatch.h"
 
 
 #define FB_3D		0x01
diff --git a/lib/mesa/src/mesa/main/format_pack.c b/lib/mesa/src/mesa/main/format_pack.c
index f9e194d31..6b183d55f 100644
--- a/lib/mesa/src/mesa/main/format_pack.c
+++ b/lib/mesa/src/mesa/main/format_pack.c
@@ -38,6 +38,8 @@
 
 #include <stdint.h>
 
+#include "config.h"
+#include "errors.h"
 #include "format_pack.h"
 #include "format_utils.h"
 #include "macros.h"
@@ -10752,6 +10754,10 @@ pack_float_Z_UNORM32(const GLfloat *src, void *dst)
    *d = (GLuint) (*src * scale);
 }
 
+/**
+ ** Pack float to Z_FLOAT32 or Z_FLOAT32_X24S8.
+ **/
+
 static void
 pack_float_Z_FLOAT32(const GLfloat *src, void *dst)
 {
@@ -10824,18 +10830,12 @@ pack_uint_Z_UNORM32(const GLuint *src, void *dst)
    *d = *src;
 }
 
-static void
-pack_uint_Z_FLOAT32(const GLuint *src, void *dst)
-{
-   GLuint *d = ((GLuint *) dst);
-   const GLdouble scale = 1.0 / (GLdouble) 0xffffffff;
-   *d = (GLuint) (*src * scale);
-   assert(*d >= 0.0f);
-   assert(*d <= 1.0f);
-}
+/**
+ ** Pack uint to Z_FLOAT32 or Z_FLOAT32_X24S8.
+ **/
 
 static void
-pack_uint_Z_FLOAT32_X24S8(const GLuint *src, void *dst)
+pack_uint_Z_FLOAT32(const GLuint *src, void *dst)
 {
    GLfloat *d = ((GLfloat *) dst);
    const GLdouble scale = 1.0 / (GLdouble) 0xffffffff;
@@ -10859,9 +10859,8 @@ _mesa_get_pack_uint_z_func(mesa_format format)
    case MESA_FORMAT_Z_UNORM32:
       return pack_uint_Z_UNORM32;
    case MESA_FORMAT_Z_FLOAT32:
-      return pack_uint_Z_FLOAT32;
    case MESA_FORMAT_Z32_FLOAT_S8X24_UINT:
-      return pack_uint_Z_FLOAT32_X24S8;
+      return pack_uint_Z_FLOAT32;
    default:
       _mesa_problem(NULL, "unexpected format in _mesa_get_pack_uint_z_func()");
       return NULL;
diff --git a/lib/mesa/src/mesa/main/format_unpack.c b/lib/mesa/src/mesa/main/format_unpack.c
index 21c55416b..b3289acd3 100644
--- a/lib/mesa/src/mesa/main/format_unpack.c
+++ b/lib/mesa/src/mesa/main/format_unpack.c
@@ -38,6 +38,7 @@
 
 #include <stdint.h>
 
+#include "errors.h"
 #include "format_unpack.h"
 #include "format_utils.h"
 #include "macros.h"
diff --git a/lib/mesa/src/mesa/main/format_utils.c b/lib/mesa/src/mesa/main/format_utils.c
index 31580750b..6959bf062 100644
--- a/lib/mesa/src/mesa/main/format_utils.c
+++ b/lib/mesa/src/mesa/main/format_utils.c
@@ -22,6 +22,7 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
+#include "errors.h"
 #include "format_utils.h"
 #include "glformats.h"
 #include "format_pack.h"
diff --git a/lib/mesa/src/mesa/main/format_utils.h b/lib/mesa/src/mesa/main/format_utils.h
index 378997b38..78365cab5 100644
--- a/lib/mesa/src/mesa/main/format_utils.h
+++ b/lib/mesa/src/mesa/main/format_utils.h
@@ -31,6 +31,7 @@
 #ifndef FORMAT_UTILS_H
 #define FORMAT_UTILS_H
 
+#include "formats.h"
 #include "imports.h"
 #include "macros.h"
 #include "util/rounding.h"
diff --git a/lib/mesa/src/mesa/main/glheader.h b/lib/mesa/src/mesa/main/glheader.h
index 3f2a92378..e95fc9fb4 100644
--- a/lib/mesa/src/mesa/main/glheader.h
+++ b/lib/mesa/src/mesa/main/glheader.h
@@ -43,13 +43,13 @@ extern "C" {
 #endif
 
 
-typedef int GLclampx;
-
-
-#ifndef GL_OES_EGL_image
-typedef void *GLeglImageOES;
-#endif
+/* Custom Mesa types to save space. */
+typedef unsigned short GLenum16;
+typedef unsigned char GLbitfield8;
+typedef unsigned short GLbitfield16;
+typedef GLuint64 GLbitfield64;
 
+/* Common GLES 1.0 and 2.0 tokens */
 
 #ifndef GL_OES_EGL_image_external
 #define GL_TEXTURE_EXTERNAL_OES                                 0x8D65
@@ -58,6 +58,14 @@ typedef void *GLeglImageOES;
 #define GL_REQUIRED_TEXTURE_IMAGE_UNITS_OES                     0x8D68
 #endif
 
+#ifndef GL_OES_compressed_ETC1_RGB8_texture
+#define GL_ETC1_RGB8_OES                                        0x8D64
+#endif
+
+
+/* GLES 1.0 only tokens */
+
+typedef int GLclampx;
 
 #ifndef GL_OES_point_size_array
 #define GL_POINT_SIZE_ARRAY_OES                                 0x8B9C
@@ -72,31 +80,15 @@ typedef void *GLeglImageOES;
 #define GL_TEXTURE_CROP_RECT_OES                                0x8B9D
 #endif
 
-
-#ifndef GL_PROGRAM_BINARY_LENGTH_OES
-#define GL_PROGRAM_BINARY_LENGTH_OES                            0x8741
-#endif
-
-/* GLES 2.0 tokens */
-#ifndef GL_RGB565
-#define GL_RGB565                                               0x8D62
-#endif
-
 #ifndef GL_TEXTURE_GEN_STR_OES
 #define GL_TEXTURE_GEN_STR_OES                                  0x8D60
 #endif
 
-#ifndef GL_OES_compressed_paletted_texture
-#define GL_PALETTE4_RGB8_OES                                    0x8B90
-#define GL_PALETTE4_RGBA8_OES                                   0x8B91
-#define GL_PALETTE4_R5_G6_B5_OES                                0x8B92
-#define GL_PALETTE4_RGBA4_OES                                   0x8B93
-#define GL_PALETTE4_RGB5_A1_OES                                 0x8B94
-#define GL_PALETTE8_RGB8_OES                                    0x8B95
-#define GL_PALETTE8_RGBA8_OES                                   0x8B96
-#define GL_PALETTE8_R5_G6_B5_OES                                0x8B97
-#define GL_PALETTE8_RGBA4_OES                                   0x8B98
-#define GL_PALETTE8_RGB5_A1_OES                                 0x8B99
+
+/* GLES 2.0 only tokens */
+
+#ifndef GL_PROGRAM_BINARY_LENGTH_OES
+#define GL_PROGRAM_BINARY_LENGTH_OES                            0x8741
 #endif
 
 #ifndef GL_OES_texture_compression_astc
@@ -122,34 +114,30 @@ typedef void *GLeglImageOES;
 #define GL_COMPRESSED_SRGB8_ALPHA8_ASTC_6x6x6_OES               0x93E9
 #endif
 
-#ifndef GL_ES_VERSION_2_0
-#define GL_SHADER_BINARY_FORMATS                                0x8DF8
-#define GL_NUM_SHADER_BINARY_FORMATS                            0x8DF9
-#define GL_SHADER_COMPILER                                      0x8DFA
-#define GL_MAX_VERTEX_UNIFORM_VECTORS                           0x8DFB
-#define GL_MAX_VARYING_VECTORS                                  0x8DFC
-#define GL_MAX_FRAGMENT_UNIFORM_VECTORS                         0x8DFD
-#endif
-
-#ifndef GL_ATI_texture_compression_3dc
-#define GL_ATI_texture_compression_3dc                          1
-#define GL_COMPRESSED_LUMINANCE_ALPHA_3DC_ATI                   0x8837
-#endif
-
-#ifndef GL_OES_compressed_ETC1_RGB8_texture
-#define GL_ETC1_RGB8_OES                                        0x8D64
-#endif
-
 #ifndef GL_EXT_shader_framebuffer_fetch
 #define GL_FRAGMENT_SHADER_DISCARDS_SAMPLES_EXT                 0x8A52
 #endif
 
+#ifndef GL_EXT_disjoint_timer_query
+#define GL_GPU_DISJOINT_EXT                                     0x8FBB
+#endif
+
 /* Inexplicably, GL_HALF_FLOAT_OES has a different value than GL_HALF_FLOAT.
  */
 #ifndef GL_HALF_FLOAT_OES
 #define GL_HALF_FLOAT_OES                                       0x8D61
 #endif
 
+#ifndef GL_MESA_framebuffer_flip_y
+#define GL_FRAMEBUFFER_FLIP_Y_MESA                              0x8BBB
+#endif
+
+/* There is no formal spec for the following extension. */
+#ifndef GL_ATI_texture_compression_3dc
+#define GL_ATI_texture_compression_3dc                          1
+#define GL_COMPRESSED_LUMINANCE_ALPHA_3DC_ATI                   0x8837
+#endif
+
 
 /**
  * Internal token to represent a GLSL shader program (a collection of
@@ -161,28 +149,6 @@ typedef void *GLeglImageOES;
 #define GL_SHADER_PROGRAM_MESA                                  0x9999
 
 
-/* Several fields of struct gl_config can take these as values.  Since
- * GLX header files may not be available everywhere they need to be used,
- * redefine them here.
- */
-#define GLX_NONE                                                0x8000
-#define GLX_SLOW_CONFIG                                         0x8001
-#define GLX_TRUE_COLOR                                          0x8002
-#define GLX_DIRECT_COLOR                                        0x8003
-#define GLX_PSEUDO_COLOR                                        0x8004
-#define GLX_STATIC_COLOR                                        0x8005
-#define GLX_GRAY_SCALE                                          0x8006
-#define GLX_STATIC_GRAY                                         0x8007
-#define GLX_TRANSPARENT_RGB                                     0x8008
-#define GLX_TRANSPARENT_INDEX                                   0x8009
-#define GLX_NON_CONFORMANT_CONFIG                               0x800D
-#define GLX_SWAP_EXCHANGE_OML                                   0x8061
-#define GLX_SWAP_COPY_OML                                       0x8062
-#define GLX_SWAP_UNDEFINED_OML                                  0x8063
-
-#define GLX_DONT_CARE                                           0xFFFFFFFF
-
-
 #ifdef __cplusplus
 }
 #endif
diff --git a/lib/mesa/src/mesa/main/glspirv.c b/lib/mesa/src/mesa/main/glspirv.c
new file mode 100644
index 000000000..04e46ba57
--- /dev/null
+++ b/lib/mesa/src/mesa/main/glspirv.c
@@ -0,0 +1,378 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "glspirv.h"
+#include "errors.h"
+#include "shaderobj.h"
+#include "mtypes.h"
+
+#include "compiler/nir/nir.h"
+#include "compiler/spirv/nir_spirv.h"
+
+#include "program/program.h"
+
+#include "util/u_atomic.h"
+
+void
+_mesa_spirv_module_reference(struct gl_spirv_module **dest,
+                             struct gl_spirv_module *src)
+{
+   struct gl_spirv_module *old = *dest;
+
+   if (old && p_atomic_dec_zero(&old->RefCount))
+      free(old);
+
+   *dest = src;
+
+   if (src)
+      p_atomic_inc(&src->RefCount);
+}
+
+void
+_mesa_shader_spirv_data_reference(struct gl_shader_spirv_data **dest,
+                                  struct gl_shader_spirv_data *src)
+{
+   struct gl_shader_spirv_data *old = *dest;
+
+   if (old && p_atomic_dec_zero(&old->RefCount)) {
+      _mesa_spirv_module_reference(&(*dest)->SpirVModule, NULL);
+      ralloc_free(old);
+   }
+
+   *dest = src;
+
+   if (src)
+      p_atomic_inc(&src->RefCount);
+}
+
+void
+_mesa_spirv_shader_binary(struct gl_context *ctx,
+                          unsigned n, struct gl_shader **shaders,
+                          const void* binary, size_t length)
+{
+   struct gl_spirv_module *module;
+   struct gl_shader_spirv_data *spirv_data;
+
+   module = malloc(sizeof(*module) + length);
+   if (!module) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glShaderBinary");
+      return;
+   }
+
+   p_atomic_set(&module->RefCount, 0);
+   module->Length = length;
+   memcpy(&module->Binary[0], binary, length);
+
+   for (int i = 0; i < n; ++i) {
+      struct gl_shader *sh = shaders[i];
+
+      spirv_data = rzalloc(NULL, struct gl_shader_spirv_data);
+      _mesa_shader_spirv_data_reference(&sh->spirv_data, spirv_data);
+      _mesa_spirv_module_reference(&spirv_data->SpirVModule, module);
+
+      sh->CompileStatus = COMPILE_FAILURE;
+
+      free((void *)sh->Source);
+      sh->Source = NULL;
+      free((void *)sh->FallbackSource);
+      sh->FallbackSource = NULL;
+
+      ralloc_free(sh->ir);
+      sh->ir = NULL;
+      ralloc_free(sh->symbols);
+      sh->symbols = NULL;
+   }
+}
+
+/**
+ * This is the equivalent to compiler/glsl/linker.cpp::link_shaders()
+ * but for SPIR-V programs.
+ *
+ * This method just creates the gl_linked_shader structs with a reference to
+ * the SPIR-V data collected during previous steps.
+ *
+ * The real linking happens later in the driver-specifc call LinkShader().
+ * This is so backends can implement different linking strategies for
+ * SPIR-V programs.
+ */
+void
+_mesa_spirv_link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
+{
+   prog->data->LinkStatus = LINKING_SUCCESS;
+   prog->data->Validated = false;
+
+   for (unsigned i = 0; i < prog->NumShaders; i++) {
+      struct gl_shader *shader = prog->Shaders[i];
+      gl_shader_stage shader_type = shader->Stage;
+
+      /* We only support one shader per stage. The gl_spirv spec doesn't seem
+       * to prevent this, but the way the API is designed, requiring all shaders
+       * to be specialized with an entry point, makes supporting this quite
+       * undefined.
+       *
+       * TODO: Turn this into a proper error once the spec bug
+       * <https://gitlab.khronos.org/opengl/API/issues/58> is resolved.
+       */
+      if (prog->_LinkedShaders[shader_type]) {
+         ralloc_strcat(&prog->data->InfoLog,
+                       "\nError trying to link more than one SPIR-V shader "
+                       "per stage.\n");
+         prog->data->LinkStatus = LINKING_FAILURE;
+         return;
+      }
+
+      assert(shader->spirv_data);
+
+      struct gl_linked_shader *linked = rzalloc(NULL, struct gl_linked_shader);
+      linked->Stage = shader_type;
+
+      /* Create program and attach it to the linked shader */
+      struct gl_program *gl_prog =
+         ctx->Driver.NewProgram(ctx,
+                                _mesa_shader_stage_to_program(shader_type),
+                                prog->Name, false);
+      if (!gl_prog) {
+         prog->data->LinkStatus = LINKING_FAILURE;
+         _mesa_delete_linked_shader(ctx, linked);
+         return;
+      }
+
+      _mesa_reference_shader_program_data(ctx,
+                                          &gl_prog->sh.data,
+                                          prog->data);
+
+      /* Don't use _mesa_reference_program() just take ownership */
+      linked->Program = gl_prog;
+
+      /* Reference the SPIR-V data from shader to the linked shader */
+      _mesa_shader_spirv_data_reference(&linked->spirv_data,
+                                        shader->spirv_data);
+
+      prog->_LinkedShaders[shader_type] = linked;
+      prog->data->linked_stages |= 1 << shader_type;
+   }
+
+   int last_vert_stage =
+      util_last_bit(prog->data->linked_stages &
+                    ((1 << (MESA_SHADER_GEOMETRY + 1)) - 1));
+
+   if (last_vert_stage)
+      prog->last_vert_prog = prog->_LinkedShaders[last_vert_stage - 1]->Program;
+}
+
+nir_shader *
+_mesa_spirv_to_nir(struct gl_context *ctx,
+                   const struct gl_shader_program *prog,
+                   gl_shader_stage stage,
+                   const nir_shader_compiler_options *options)
+{
+   nir_shader *nir = NULL;
+
+   struct gl_linked_shader *linked_shader = prog->_LinkedShaders[stage];
+   assert (linked_shader);
+
+   struct gl_shader_spirv_data *spirv_data = linked_shader->spirv_data;
+   assert(spirv_data);
+
+   struct gl_spirv_module *spirv_module = spirv_data->SpirVModule;
+   assert (spirv_module != NULL);
+
+   const char *entry_point_name = spirv_data->SpirVEntryPoint;
+   assert(entry_point_name);
+
+   struct nir_spirv_specialization *spec_entries =
+      calloc(sizeof(*spec_entries),
+             spirv_data->NumSpecializationConstants);
+
+   for (unsigned i = 0; i < spirv_data->NumSpecializationConstants; ++i) {
+      spec_entries[i].id = spirv_data->SpecializationConstantsIndex[i];
+      spec_entries[i].data32 = spirv_data->SpecializationConstantsValue[i];
+      spec_entries[i].defined_on_module = false;
+   }
+
+   const struct spirv_to_nir_options spirv_options = {
+      .lower_workgroup_access_to_offsets = true,
+      .caps = ctx->Const.SpirVCapabilities
+   };
+
+   nir_function *entry_point =
+      spirv_to_nir((const uint32_t *) &spirv_module->Binary[0],
+                   spirv_module->Length / 4,
+                   spec_entries, spirv_data->NumSpecializationConstants,
+                   stage, entry_point_name,
+                   &spirv_options,
+                   options);
+   free(spec_entries);
+
+   assert (entry_point);
+   nir = entry_point->shader;
+   assert(nir->info.stage == stage);
+
+   nir->options = options;
+
+   nir->info.name =
+      ralloc_asprintf(nir, "SPIRV:%s:%d",
+                      _mesa_shader_stage_to_abbrev(nir->info.stage),
+                      prog->Name);
+   nir_validate_shader(nir, "after spirv_to_nir");
+
+   nir->info.separate_shader = linked_shader->Program->info.separate_shader;
+
+   /* We have to lower away local constant initializers right before we
+    * inline functions.  That way they get properly initialized at the top
+    * of the function and not at the top of its caller.
+    */
+   NIR_PASS_V(nir, nir_lower_constant_initializers, nir_var_local);
+   NIR_PASS_V(nir, nir_lower_returns);
+   NIR_PASS_V(nir, nir_inline_functions);
+   NIR_PASS_V(nir, nir_copy_prop);
+
+   /* Pick off the single entrypoint that we want */
+   foreach_list_typed_safe(nir_function, func, node, &nir->functions) {
+      if (func != entry_point)
+         exec_node_remove(&func->node);
+   }
+   assert(exec_list_length(&nir->functions) == 1);
+
+   /* Split member structs.  We do this before lower_io_to_temporaries so that
+    * it doesn't lower system values to temporaries by accident.
+    */
+   NIR_PASS_V(nir, nir_split_var_copies);
+   NIR_PASS_V(nir, nir_split_per_member_structs);
+
+   if (nir->info.stage == MESA_SHADER_VERTEX)
+      nir_remap_dual_slot_attributes(nir, &linked_shader->Program->DualSlotInputs);
+
+   return nir;
+}
+
+void GLAPIENTRY
+_mesa_SpecializeShaderARB(GLuint shader,
+                          const GLchar *pEntryPoint,
+                          GLuint numSpecializationConstants,
+                          const GLuint *pConstantIndex,
+                          const GLuint *pConstantValue)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   struct gl_shader *sh;
+   bool has_entry_point;
+   struct nir_spirv_specialization *spec_entries = NULL;
+
+   if (!ctx->Extensions.ARB_gl_spirv) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glSpecializeShaderARB");
+      return;
+   }
+
+   sh = _mesa_lookup_shader_err(ctx, shader, "glSpecializeShaderARB");
+   if (!sh)
+      return;
+
+   if (!sh->spirv_data) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glSpecializeShaderARB(not SPIR-V)");
+      return;
+   }
+
+   if (sh->CompileStatus) {
+      _mesa_error(ctx, GL_INVALID_OPERATION,
+                  "glSpecializeShaderARB(already specialized)");
+      return;
+   }
+
+   struct gl_shader_spirv_data *spirv_data = sh->spirv_data;
+
+   /* From the GL_ARB_gl_spirv spec:
+    *
+    *    "The OpenGL API expects the SPIR-V module to have already been
+    *     validated, and can return an error if it discovers anything invalid
+    *     in the module. An invalid SPIR-V module is allowed to result in
+    *     undefined behavior."
+    *
+    * However, the following errors still need to be detected (from the same
+    * spec):
+    *
+    *    "INVALID_VALUE is generated if <pEntryPoint> does not name a valid
+    *     entry point for <shader>.
+    *
+    *     INVALID_VALUE is generated if any element of <pConstantIndex>
+    *     refers to a specialization constant that does not exist in the
+    *     shader module contained in <shader>."
+    *
+    * We cannot flag those errors a-priori because detecting them requires
+    * parsing the module. However, flagging them during specialization is okay,
+    * since it makes no difference in terms of application-visible state.
+    */
+   spec_entries = calloc(sizeof(*spec_entries), numSpecializationConstants);
+
+   for (unsigned i = 0; i < numSpecializationConstants; ++i) {
+      spec_entries[i].id = pConstantIndex[i];
+      spec_entries[i].data32 = pConstantValue[i];
+      spec_entries[i].defined_on_module = false;
+   }
+
+   has_entry_point =
+      gl_spirv_validation((uint32_t *)&spirv_data->SpirVModule->Binary[0],
+                          spirv_data->SpirVModule->Length / 4,
+                          spec_entries, numSpecializationConstants,
+                          sh->Stage, pEntryPoint);
+
+   /* See previous spec comment */
+   if (!has_entry_point) {
+      _mesa_error(ctx, GL_INVALID_VALUE,
+                  "glSpecializeShaderARB(\"%s\" is not a valid entry point"
+                  " for shader)", pEntryPoint);
+      goto end;
+   }
+
+   for (unsigned i = 0; i < numSpecializationConstants; ++i) {
+      if (spec_entries[i].defined_on_module == false) {
+         _mesa_error(ctx, GL_INVALID_VALUE,
+                     "glSpecializeShaderARB(constant \"%i\" does not exist "
+                     "in shader)", spec_entries[i].id);
+         goto end;
+      }
+   }
+
+   spirv_data->SpirVEntryPoint = ralloc_strdup(spirv_data, pEntryPoint);
+
+   /* Note that we didn't make a real compilation of the module (spirv_to_nir),
+    * but just checked some error conditions. Real "compilation" will be done
+    * later, upon linking.
+    */
+   sh->CompileStatus = COMPILE_SUCCESS;
+
+   spirv_data->NumSpecializationConstants = numSpecializationConstants;
+   spirv_data->SpecializationConstantsIndex =
+      rzalloc_array_size(spirv_data, sizeof(GLuint),
+                         numSpecializationConstants);
+   spirv_data->SpecializationConstantsValue =
+      rzalloc_array_size(spirv_data, sizeof(GLuint),
+                         numSpecializationConstants);
+   for (unsigned i = 0; i < numSpecializationConstants; ++i) {
+      spirv_data->SpecializationConstantsIndex[i] = pConstantIndex[i];
+      spirv_data->SpecializationConstantsValue[i] = pConstantValue[i];
+   }
+
+ end:
+   free(spec_entries);
+}
diff --git a/lib/mesa/src/mesa/main/glspirv.h b/lib/mesa/src/mesa/main/glspirv.h
new file mode 100644
index 000000000..8025c17c0
--- /dev/null
+++ b/lib/mesa/src/mesa/main/glspirv.h
@@ -0,0 +1,111 @@
+/*
+ * Copyright 2017 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHOR(S) AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#ifndef GLSPIRV_H
+#define GLSPIRV_H
+
+#include "compiler/nir/nir.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct gl_shader_program;
+struct gl_context;
+struct gl_shader;
+
+/**
+ * A SPIR-V module contains the raw SPIR-V binary as set by ShaderBinary.
+ *
+ * It is reference-counted, because the same module can be attached to multiple
+ * shader objects simultaneously.
+ */
+struct gl_spirv_module {
+   unsigned RefCount;
+   GLint Length;
+   char Binary[0];
+};
+
+/**
+ * SPIR-V data needed to compile and link a SPIR-V shader.
+ *
+ * It includes a SPIR-V binary that is potentially shared among different
+ * shaders; and shader-specific specialization constants and entry point.
+ *
+ * It is reference-counted because it is shared between gl_shader and its
+ * corresponding gl_linked_shader.
+ */
+struct gl_shader_spirv_data {
+   GLint RefCount;
+
+   struct gl_spirv_module *SpirVModule;
+
+   GLchar *SpirVEntryPoint;
+
+   GLuint NumSpecializationConstants;
+   GLuint *SpecializationConstantsIndex;
+   GLuint *SpecializationConstantsValue;
+};
+
+void
+_mesa_spirv_module_reference(struct gl_spirv_module **dest,
+                             struct gl_spirv_module *src);
+
+void
+_mesa_shader_spirv_data_reference(struct gl_shader_spirv_data **dest,
+                                  struct gl_shader_spirv_data *src);
+
+void
+_mesa_spirv_shader_binary(struct gl_context *ctx,
+                          unsigned n, struct gl_shader **shaders,
+                          const void* binary, size_t length);
+
+void
+_mesa_spirv_link_shaders(struct gl_context *ctx,
+                         struct gl_shader_program *prog);
+
+nir_shader *
+_mesa_spirv_to_nir(struct gl_context *ctx,
+                   const struct gl_shader_program *prog,
+                   gl_shader_stage stage,
+                   const nir_shader_compiler_options *options);
+
+/**
+ * \name API functions
+ */
+/*@{*/
+
+void GLAPIENTRY
+_mesa_SpecializeShaderARB(GLuint shader,
+                          const GLchar *pEntryPoint,
+                          GLuint numSpecializationConstants,
+                          const GLuint *pConstantIndex,
+                          const GLuint *pConstantValue);
+
+/*@}*/
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* GLSPIRV_H */
diff --git a/lib/mesa/src/mesa/main/histogram.c b/lib/mesa/src/mesa/main/histogram.c
index 5759763d4..5696f548c 100644
--- a/lib/mesa/src/mesa/main/histogram.c
+++ b/lib/mesa/src/mesa/main/histogram.c
@@ -26,7 +26,6 @@
 #include "glheader.h"
 #include "context.h"
 #include "histogram.h"
-#include "main/dispatch.h"
 
 
 /**********************************************************************
diff --git a/lib/mesa/src/mesa/main/image.h b/lib/mesa/src/mesa/main/image.h
index b5075be4b..7955bbb7c 100644
--- a/lib/mesa/src/mesa/main/image.h
+++ b/lib/mesa/src/mesa/main/image.h
@@ -28,7 +28,6 @@
 
 
 #include "glheader.h"
-#include "compiler.h"
 
 struct gl_context;
 struct gl_pixelstore_attrib;
diff --git a/lib/mesa/src/mesa/main/matrix.c b/lib/mesa/src/mesa/main/matrix.c
index 83f081e88..8065a8370 100644
--- a/lib/mesa/src/mesa/main/matrix.c
+++ b/lib/mesa/src/mesa/main/matrix.c
@@ -657,20 +657,16 @@ void _mesa_update_modelview_project( struct gl_context *ctx, GLuint new_state )
  * _math_matrix_ctr() for each element to initialize it.
  */
 static void
-init_matrix_stack( struct gl_matrix_stack *stack,
-                   GLuint maxDepth, GLuint dirtyFlag )
+init_matrix_stack(struct gl_matrix_stack *stack,
+                  GLuint maxDepth, GLuint dirtyFlag)
 {
-   GLuint i;
-
    stack->Depth = 0;
    stack->MaxDepth = maxDepth;
    stack->DirtyFlag = dirtyFlag;
    /* The stack will be dynamically resized at glPushMatrix() time */
    stack->Stack = calloc(1, sizeof(GLmatrix));
    stack->StackSize = 1;
-   for (i = 0; i < stack->StackSize; i++) {
-      _math_matrix_ctr(&stack->Stack[i]);
-   }
+   _math_matrix_ctr(&stack->Stack[0]);
    stack->Top = stack->Stack;
 }
 
diff --git a/lib/mesa/src/mesa/main/menums.h b/lib/mesa/src/mesa/main/menums.h
new file mode 100644
index 000000000..40b16cb9c
--- /dev/null
+++ b/lib/mesa/src/mesa/main/menums.h
@@ -0,0 +1,213 @@
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2009  VMware, Inc.  All Rights Reserved.
+ * Copyright (C) 2018 Advanced Micro Devices, Inc.  All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/**
+ * \file menums.h
+ * Often used definitions and enums.
+ */
+
+#ifndef MENUMS_H
+#define MENUMS_H
+
+#include "util/macros.h"
+
+/**
+ * Enum for the OpenGL APIs we know about and may support.
+ *
+ * NOTE: This must match the api_enum table in
+ * src/mesa/main/get_hash_generator.py
+ */
+typedef enum
+{
+   API_OPENGL_COMPAT,      /* legacy / compatibility contexts */
+   API_OPENGLES,
+   API_OPENGLES2,
+   API_OPENGL_CORE,
+   API_OPENGL_LAST = API_OPENGL_CORE
+} gl_api;
+
+/**
+ * An index for each type of texture object.  These correspond to the GL
+ * texture target enums, such as GL_TEXTURE_2D, GL_TEXTURE_CUBE_MAP, etc.
+ * Note: the order is from highest priority to lowest priority.
+ */
+typedef enum
+{
+   TEXTURE_2D_MULTISAMPLE_INDEX,
+   TEXTURE_2D_MULTISAMPLE_ARRAY_INDEX,
+   TEXTURE_CUBE_ARRAY_INDEX,
+   TEXTURE_BUFFER_INDEX,
+   TEXTURE_2D_ARRAY_INDEX,
+   TEXTURE_1D_ARRAY_INDEX,
+   TEXTURE_EXTERNAL_INDEX,
+   TEXTURE_CUBE_INDEX,
+   TEXTURE_3D_INDEX,
+   TEXTURE_RECT_INDEX,
+   TEXTURE_2D_INDEX,
+   TEXTURE_1D_INDEX,
+   NUM_TEXTURE_TARGETS
+} gl_texture_index;
+
+/**
+ * Remapped color logical operations
+ *
+ * With the exception of NVIDIA hardware, which consumes the OpenGL enumerants
+ * directly, everything wants this mapping of color logical operations.
+ *
+ * Fun fact: These values are just the bit-reverse of the low-nibble of the GL
+ * enumerant values (i.e., `GL_NOOP & 0x0f` is `b0101' while
+ * \c COLOR_LOGICOP_NOOP is `b1010`).
+ *
+ * Fun fact #2: These values are just an encoding of the operation as a table
+ * of bit values. The result of the logic op is:
+ *
+ *    result_bit = (logic_op >> (2 * src_bit + dst_bit)) & 1
+ *
+ * For the GL enums, the result is:
+ *
+ *    result_bit = logic_op & (1 << (2 * src_bit + dst_bit))
+ */
+enum PACKED gl_logicop_mode {
+   COLOR_LOGICOP_CLEAR = 0,
+   COLOR_LOGICOP_NOR = 1,
+   COLOR_LOGICOP_AND_INVERTED = 2,
+   COLOR_LOGICOP_COPY_INVERTED = 3,
+   COLOR_LOGICOP_AND_REVERSE = 4,
+   COLOR_LOGICOP_INVERT = 5,
+   COLOR_LOGICOP_XOR = 6,
+   COLOR_LOGICOP_NAND = 7,
+   COLOR_LOGICOP_AND = 8,
+   COLOR_LOGICOP_EQUIV = 9,
+   COLOR_LOGICOP_NOOP = 10,
+   COLOR_LOGICOP_OR_INVERTED = 11,
+   COLOR_LOGICOP_COPY = 12,
+   COLOR_LOGICOP_OR_REVERSE = 13,
+   COLOR_LOGICOP_OR = 14,
+   COLOR_LOGICOP_SET = 15
+};
+
+/**
+ * Indexes for all renderbuffers
+ */
+typedef enum
+{
+   /* the four standard color buffers */
+   BUFFER_FRONT_LEFT,
+   BUFFER_BACK_LEFT,
+   BUFFER_FRONT_RIGHT,
+   BUFFER_BACK_RIGHT,
+   BUFFER_DEPTH,
+   BUFFER_STENCIL,
+   BUFFER_ACCUM,
+   /* optional aux buffer */
+   BUFFER_AUX0,
+   /* generic renderbuffers */
+   BUFFER_COLOR0,
+   BUFFER_COLOR1,
+   BUFFER_COLOR2,
+   BUFFER_COLOR3,
+   BUFFER_COLOR4,
+   BUFFER_COLOR5,
+   BUFFER_COLOR6,
+   BUFFER_COLOR7,
+   BUFFER_COUNT,
+   BUFFER_NONE = -1,
+} gl_buffer_index;
+
+typedef enum
+{
+   MAP_USER,
+   MAP_INTERNAL,
+   MAP_COUNT
+} gl_map_buffer_index;
+
+/** @{
+ *
+ * These are a mapping of the GL_ARB_debug_output/GL_KHR_debug enums
+ * to small enums suitable for use as an array index.
+ */
+
+enum mesa_debug_source
+{
+   MESA_DEBUG_SOURCE_API,
+   MESA_DEBUG_SOURCE_WINDOW_SYSTEM,
+   MESA_DEBUG_SOURCE_SHADER_COMPILER,
+   MESA_DEBUG_SOURCE_THIRD_PARTY,
+   MESA_DEBUG_SOURCE_APPLICATION,
+   MESA_DEBUG_SOURCE_OTHER,
+   MESA_DEBUG_SOURCE_COUNT
+};
+
+enum mesa_debug_type
+{
+   MESA_DEBUG_TYPE_ERROR,
+   MESA_DEBUG_TYPE_DEPRECATED,
+   MESA_DEBUG_TYPE_UNDEFINED,
+   MESA_DEBUG_TYPE_PORTABILITY,
+   MESA_DEBUG_TYPE_PERFORMANCE,
+   MESA_DEBUG_TYPE_OTHER,
+   MESA_DEBUG_TYPE_MARKER,
+   MESA_DEBUG_TYPE_PUSH_GROUP,
+   MESA_DEBUG_TYPE_POP_GROUP,
+   MESA_DEBUG_TYPE_COUNT
+};
+
+enum mesa_debug_severity
+{
+   MESA_DEBUG_SEVERITY_LOW,
+   MESA_DEBUG_SEVERITY_MEDIUM,
+   MESA_DEBUG_SEVERITY_HIGH,
+   MESA_DEBUG_SEVERITY_NOTIFICATION,
+   MESA_DEBUG_SEVERITY_COUNT
+};
+
+/** @} */
+
+/** Set a single bit */
+#define BITFIELD_BIT(b)      (1u << (b))
+/** Set all bits up to excluding bit b */
+#define BITFIELD_MASK(b)      \
+   ((b) == 32 ? (~0u) : BITFIELD_BIT((b) % 32) - 1)
+/** Set count bits starting from bit b  */
+#define BITFIELD_RANGE(b, count) \
+   (BITFIELD_MASK((b) + (count)) & ~BITFIELD_MASK(b))
+
+
+/**
+ * \name 64-bit extension of GLbitfield.
+ */
+/*@{*/
+
+/** Set a single bit */
+#define BITFIELD64_BIT(b)      (1ull << (b))
+/** Set all bits up to excluding bit b */
+#define BITFIELD64_MASK(b)      \
+   ((b) == 64 ? (~0ull) : BITFIELD64_BIT(b) - 1)
+/** Set count bits starting from bit b  */
+#define BITFIELD64_RANGE(b, count) \
+   (BITFIELD64_MASK((b) + (count)) & ~BITFIELD64_MASK(b))
+
+#endif
diff --git a/lib/mesa/src/mesa/main/meson.build b/lib/mesa/src/mesa/main/meson.build
new file mode 100644
index 000000000..a5f0e02c6
--- /dev/null
+++ b/lib/mesa/src/mesa/main/meson.build
@@ -0,0 +1,46 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+main_dispatch_h = custom_target(
+  'dispatch.h',
+  input : [files('../../mapi/glapi/gen/gl_table.py'), gl_and_es_api_files],
+  output : 'dispatch.h',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@', '-m', 'remap_table'],
+  depend_files : glapi_gen_depends,
+  capture : true,
+)
+
+main_marshal_generated_h = custom_target(
+  'marshal_generated.h',
+  input : [files('../../mapi/glapi/gen/gl_marshal_h.py'), gl_and_es_api_files],
+  output : 'marshal_generated.h',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  depend_files : files('../../mapi/glapi/gen/marshal_XML.py') + glapi_gen_depends,
+  capture : true,
+)
+
+main_remap_helper_h = custom_target(
+  'remap_helper.h',
+  input : [files('../../mapi/glapi/gen/remap_helper.py'), gl_and_es_api_files],
+  output : 'remap_helper.h',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  depend_files : glapi_gen_depends,
+  capture : true,
+)
diff --git a/lib/mesa/src/mesa/main/mipmap.h b/lib/mesa/src/mesa/main/mipmap.h
index d11c7fada..3ec997217 100644
--- a/lib/mesa/src/mesa/main/mipmap.h
+++ b/lib/mesa/src/mesa/main/mipmap.h
@@ -26,8 +26,15 @@
 #ifndef MIPMAP_H
 #define MIPMAP_H
 
-#include "mtypes.h"
+#include "glheader.h"
 
+struct gl_context;
+struct gl_texture_object;
+
+unsigned
+_mesa_compute_num_levels(struct gl_context *ctx,
+                         struct gl_texture_object *texObj,
+                         GLenum target);
 
 extern void
 _mesa_generate_mipmap_level(GLenum target,
diff --git a/lib/mesa/src/mesa/main/mm.c b/lib/mesa/src/mesa/main/mm.c
index 473e90fc2..5f0bfe969 100644
--- a/lib/mesa/src/mesa/main/mm.c
+++ b/lib/mesa/src/mesa/main/mm.c
@@ -26,7 +26,6 @@
 #include <stdio.h>
 #include <stdlib.h>
 
-#include "compiler.h"
 #include "mm.h"
 
 
diff --git a/lib/mesa/src/mesa/main/multisample.h b/lib/mesa/src/mesa/main/multisample.h
index a7cd2918d..49683cacd 100644
--- a/lib/mesa/src/mesa/main/multisample.h
+++ b/lib/mesa/src/mesa/main/multisample.h
@@ -55,6 +55,7 @@ _mesa_MinSampleShading(GLclampf value);
 
 extern GLenum
 _mesa_check_sample_count(struct gl_context *ctx, GLenum target,
-                   GLenum internalFormat, GLsizei samples);
+                         GLenum internalFormat, GLsizei samples,
+                         GLsizei storageSamples);
 
 #endif
diff --git a/lib/mesa/src/mesa/main/pack.h b/lib/mesa/src/mesa/main/pack.h
index ac0a099e3..8625a145f 100644
--- a/lib/mesa/src/mesa/main/pack.h
+++ b/lib/mesa/src/mesa/main/pack.h
@@ -28,8 +28,10 @@
 #define PACK_H
 
 
-#include "mtypes.h"
+#include "glheader.h"
 
+struct gl_context;
+struct gl_pixelstore_attrib;
 
 extern void
 _mesa_unpack_polygon_stipple(const GLubyte *pattern, GLuint dest[32],
diff --git a/lib/mesa/src/mesa/main/pbo.c b/lib/mesa/src/mesa/main/pbo.c
index 7762324a4..cea55f2a0 100644
--- a/lib/mesa/src/mesa/main/pbo.c
+++ b/lib/mesa/src/mesa/main/pbo.c
@@ -31,6 +31,7 @@
 
 
 
+#include "errors.h"
 #include "glheader.h"
 #include "bufferobj.h"
 #include "glformats.h"
diff --git a/lib/mesa/src/mesa/main/pbo.h b/lib/mesa/src/mesa/main/pbo.h
index b3f24e62b..bc764175f 100644
--- a/lib/mesa/src/mesa/main/pbo.h
+++ b/lib/mesa/src/mesa/main/pbo.h
@@ -28,8 +28,10 @@
 #define PBO_H
 
 
-#include "mtypes.h"
+#include "glheader.h"
 
+struct gl_context;
+struct gl_pixelstore_attrib;
 
 extern GLboolean
 _mesa_validate_pbo_access(GLuint dimensions,
diff --git a/lib/mesa/src/mesa/main/pixeltransfer.h b/lib/mesa/src/mesa/main/pixeltransfer.h
index b0a301f26..caa2911fc 100644
--- a/lib/mesa/src/mesa/main/pixeltransfer.h
+++ b/lib/mesa/src/mesa/main/pixeltransfer.h
@@ -28,7 +28,9 @@
 #define PIXELTRANSFER_H
 
 
-#include "mtypes.h"
+#include "glheader.h"
+
+struct gl_context;
 
 
 extern void
diff --git a/lib/mesa/src/mesa/main/program_binary.c b/lib/mesa/src/mesa/main/program_binary.c
new file mode 100644
index 000000000..7390fef58
--- /dev/null
+++ b/lib/mesa/src/mesa/main/program_binary.c
@@ -0,0 +1,321 @@
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (c) 2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/**
+ * \file program_binary.c
+ *
+ * Helper functions for serializing a binary program.
+ */
+
+
+#include "compiler/blob.h"
+#include "compiler/glsl/serialize.h"
+#include "main/errors.h"
+#include "main/mtypes.h"
+#include "main/shaderapi.h"
+#include "util/bitscan.h"
+#include "util/crc32.h"
+#include "program_binary.h"
+#include "program/prog_parameter.h"
+
+/**
+ * Mesa supports one binary format, but it must differentiate between formats
+ * produced by different drivers and different Mesa versions.
+ *
+ * Mesa uses a uint32_t value to specify an internal format. The only format
+ * defined has one uint32_t value of 0, followed by 20 bytes specifying a sha1
+ * that uniquely identifies the Mesa driver type and version.
+ */
+
+struct program_binary_header {
+   /* If internal_format is 0, it must be followed by the 20 byte sha1 that
+    * identifies the Mesa driver and version supported. If we want to support
+    * something besides a sha1, then a new internal_format value can be added.
+    */
+   uint32_t internal_format;
+   uint8_t sha1[20];
+   /* Fields following sha1 can be changed since the sha1 will guarantee that
+    * the binary only works with the same Mesa version.
+    */
+   uint32_t size;
+   uint32_t crc32;
+};
+
+/**
+ * Returns the header size needed for a binary
+ */
+static unsigned
+get_program_binary_header_size(void)
+{
+   return sizeof(struct program_binary_header);
+}
+
+static bool
+write_program_binary(const void *payload, unsigned payload_size,
+                     const void *sha1, void *binary, unsigned binary_size,
+                     GLenum *binary_format)
+{
+   struct program_binary_header *hdr = binary;
+
+   if (binary_size < sizeof(*hdr))
+      return false;
+
+   /* binary_size is the size of the buffer provided by the application.
+    * Make sure our program (payload) will fit in the buffer.
+    */
+   if (payload_size > binary_size - sizeof(*hdr))
+      return false;
+
+   hdr->internal_format = 0;
+   memcpy(hdr->sha1, sha1, sizeof(hdr->sha1));
+   memcpy(hdr + 1, payload, payload_size);
+   hdr->size = payload_size;
+
+   hdr->crc32 = util_hash_crc32(hdr + 1, payload_size);
+   *binary_format = GL_PROGRAM_BINARY_FORMAT_MESA;
+
+   return true;
+}
+
+static bool
+simple_header_checks(const struct program_binary_header *hdr, unsigned length)
+{
+   if (hdr == NULL || length < sizeof(*hdr))
+      return false;
+
+   if (hdr->internal_format != 0)
+      return false;
+
+   return true;
+}
+
+static bool
+check_crc32(const struct program_binary_header *hdr, unsigned length)
+{
+   uint32_t crc32;
+   unsigned crc32_len;
+
+   crc32_len = hdr->size;
+   if (crc32_len > length - sizeof(*hdr))
+      return false;
+
+   crc32 = util_hash_crc32(hdr + 1, crc32_len);
+   if (hdr->crc32 != crc32)
+      return false;
+
+   return true;
+}
+
+static bool
+is_program_binary_valid(GLenum binary_format, const void *sha1,
+                        const struct program_binary_header *hdr,
+                        unsigned length)
+{
+   if (binary_format != GL_PROGRAM_BINARY_FORMAT_MESA)
+      return false;
+
+   if (!simple_header_checks(hdr, length))
+      return false;
+
+   if (memcmp(hdr->sha1, sha1, sizeof(hdr->sha1)) != 0)
+      return false;
+
+   if (!check_crc32(hdr, length))
+      return false;
+
+   return true;
+}
+
+/**
+ * Returns the payload within the binary.
+ *
+ * If NULL is returned, then the binary not supported. If non-NULL is
+ * returned, it will be a pointer contained within the specified `binary`
+ * buffer.
+ *
+ * This can be used to access the payload of `binary` during the
+ * glProgramBinary call.
+ */
+static const void*
+get_program_binary_payload(GLenum binary_format, const void *sha1,
+                           const void *binary, unsigned length)
+{
+   const struct program_binary_header *hdr = binary;
+   if (!is_program_binary_valid(binary_format, sha1, hdr, length))
+      return NULL;
+   return (const uint8_t*)binary + sizeof(*hdr);
+}
+
+static void
+write_program_payload(struct gl_context *ctx, struct blob *blob,
+                      struct gl_shader_program *sh_prog)
+{
+   for (unsigned stage = 0; stage < MESA_SHADER_STAGES; stage++) {
+      struct gl_linked_shader *shader = sh_prog->_LinkedShaders[stage];
+      if (shader)
+         ctx->Driver.ProgramBinarySerializeDriverBlob(ctx, sh_prog,
+                                                      shader->Program);
+   }
+
+   serialize_glsl_program(blob, ctx, sh_prog);
+
+   for (unsigned stage = 0; stage < MESA_SHADER_STAGES; stage++) {
+      struct gl_linked_shader *shader = sh_prog->_LinkedShaders[stage];
+      if (shader) {
+         struct gl_program *prog = sh_prog->_LinkedShaders[stage]->Program;
+         ralloc_free(prog->driver_cache_blob);
+         prog->driver_cache_blob = NULL;
+         prog->driver_cache_blob_size = 0;
+      }
+   }
+}
+
+static bool
+read_program_payload(struct gl_context *ctx, struct blob_reader *blob,
+                     GLenum binary_format, struct gl_shader_program *sh_prog)
+{
+   if (!deserialize_glsl_program(blob, ctx, sh_prog))
+      return false;
+
+   unsigned int stage;
+   for (stage = 0; stage < ARRAY_SIZE(sh_prog->_LinkedShaders); stage++) {
+      struct gl_linked_shader *shader = sh_prog->_LinkedShaders[stage];
+      if (!shader)
+         continue;
+
+      ctx->Driver.ProgramBinaryDeserializeDriverBlob(ctx, sh_prog,
+                                                     shader->Program);
+   }
+
+   return true;
+}
+
+void
+_mesa_get_program_binary_length(struct gl_context *ctx,
+                                struct gl_shader_program *sh_prog,
+                                GLint *length)
+{
+   struct blob blob;
+   blob_init_fixed(&blob, NULL, SIZE_MAX);
+   write_program_payload(ctx, &blob, sh_prog);
+   *length = get_program_binary_header_size() + blob.size;
+   blob_finish(&blob);
+}
+
+void
+_mesa_get_program_binary(struct gl_context *ctx,
+                       struct gl_shader_program *sh_prog,
+                       GLsizei buf_size, GLsizei *length,
+                       GLenum *binary_format, GLvoid *binary)
+{
+   struct blob blob;
+   uint8_t driver_sha1[20];
+   unsigned header_size = get_program_binary_header_size();
+
+   ctx->Driver.GetProgramBinaryDriverSHA1(ctx, driver_sha1);
+
+   blob_init(&blob);
+
+   if (buf_size < header_size)
+      goto fail;
+
+   write_program_payload(ctx, &blob, sh_prog);
+   if (blob.size + header_size > buf_size ||
+       blob.out_of_memory)
+      goto fail;
+
+   bool written = write_program_binary(blob.data, blob.size, driver_sha1,
+                                      binary, buf_size, binary_format);
+   if (!written || blob.out_of_memory)
+      goto fail;
+
+   *length = header_size + blob.size;
+
+   blob_finish(&blob);
+   return;
+
+fail:
+   _mesa_error(ctx, GL_INVALID_OPERATION,
+               "glGetProgramBinary(buffer too small)");
+   *length = 0;
+   blob_finish(&blob);
+}
+
+void
+_mesa_program_binary(struct gl_context *ctx, struct gl_shader_program *sh_prog,
+                     GLenum binary_format, const GLvoid *binary,
+                     GLsizei length)
+{
+   uint8_t driver_sha1[20];
+   unsigned header_size = get_program_binary_header_size();
+
+   ctx->Driver.GetProgramBinaryDriverSHA1(ctx, driver_sha1);
+
+   const void *payload = get_program_binary_payload(binary_format, driver_sha1,
+                                                    binary, length);
+
+   if (payload == NULL) {
+      sh_prog->data->LinkStatus = LINKING_FAILURE;
+      return;
+   }
+
+   struct blob_reader blob;
+   blob_reader_init(&blob, payload, length - header_size);
+
+   unsigned programs_in_use = 0;
+   if (ctx->_Shader)
+      for (unsigned stage = 0; stage < MESA_SHADER_STAGES; stage++) {
+         if (ctx->_Shader->CurrentProgram[stage] &&
+             ctx->_Shader->CurrentProgram[stage]->Id == sh_prog->Name) {
+            programs_in_use |= 1 << stage;
+         }
+   }
+
+   if (!read_program_payload(ctx, &blob, binary_format, sh_prog)) {
+      sh_prog->data->LinkStatus = LINKING_FAILURE;
+      return;
+   }
+
+   /* From section 7.3 (Program Objects) of the OpenGL 4.5 spec:
+    *
+    *    "If LinkProgram or ProgramBinary successfully re-links a program
+    *     object that is active for any shader stage, then the newly generated
+    *     executable code will be installed as part of the current rendering
+    *     state for all shader stages where the program is active.
+    *     Additionally, the newly generated executable code is made part of
+    *     the state of any program pipeline for all stages where the program
+    *     is attached."
+    */
+   while (programs_in_use) {
+      const int stage = u_bit_scan(&programs_in_use);
+
+      struct gl_program *prog = NULL;
+      if (sh_prog->_LinkedShaders[stage])
+         prog = sh_prog->_LinkedShaders[stage]->Program;
+
+      _mesa_use_program(ctx, stage, sh_prog, prog, ctx->_Shader);
+   }
+
+   sh_prog->data->LinkStatus = LINKING_SKIPPED;
+}
diff --git a/lib/mesa/src/mesa/main/program_binary.h b/lib/mesa/src/mesa/main/program_binary.h
new file mode 100644
index 000000000..9e16b9ec5
--- /dev/null
+++ b/lib/mesa/src/mesa/main/program_binary.h
@@ -0,0 +1,56 @@
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 2004-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2010  VMware, Inc.  All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+#ifndef PROGRAM_BINARY_H
+#define PROGRAM_BINARY_H
+
+#include <stdbool.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void
+_mesa_get_program_binary_length(struct gl_context *ctx,
+                                struct gl_shader_program *sh_prog,
+                                GLint *length);
+
+void
+_mesa_get_program_binary(struct gl_context *ctx,
+                       struct gl_shader_program *sh_prog,
+                       GLsizei buf_size, GLsizei *length,
+                       GLenum *binary_format, GLvoid *binary);
+
+void
+_mesa_program_binary(struct gl_context *ctx, struct gl_shader_program *sh_prog,
+                     GLenum binary_format, const GLvoid *binary,
+                     GLsizei length);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* PROGRAM_BINARY_H */
diff --git a/lib/mesa/src/mesa/main/querymatrix.c b/lib/mesa/src/mesa/main/querymatrix.c
index 18361c929..b80bae6a6 100644
--- a/lib/mesa/src/mesa/main/querymatrix.c
+++ b/lib/mesa/src/mesa/main/querymatrix.c
@@ -17,6 +17,7 @@
 #include "glheader.h"
 #include "querymatrix.h"
 #include "main/get.h"
+#include "util/macros.h"
 
 
 /**
diff --git a/lib/mesa/src/mesa/main/querymatrix.h b/lib/mesa/src/mesa/main/querymatrix.h
index e08d843a9..649218910 100644
--- a/lib/mesa/src/mesa/main/querymatrix.h
+++ b/lib/mesa/src/mesa/main/querymatrix.h
@@ -25,7 +25,7 @@
 #define QUERYMATRIX_H
 
 
-#include "mtypes.h"
+#include "glheader.h"
 
 
 /*
diff --git a/lib/mesa/src/mesa/main/remap.c b/lib/mesa/src/mesa/main/remap.c
index 6dc4235cb..fa412c0c4 100644
--- a/lib/mesa/src/mesa/main/remap.c
+++ b/lib/mesa/src/mesa/main/remap.c
@@ -44,6 +44,7 @@
 
 #define need_MESA_remap_table
 #include "main/remap_helper.h"
+#include "errors.h"
 
 
 /* this is global for quick access */
diff --git a/lib/mesa/src/mesa/main/tests/meson.build b/lib/mesa/src/mesa/main/tests/meson.build
new file mode 100644
index 000000000..d1bd81a56
--- /dev/null
+++ b/lib/mesa/src/mesa/main/tests/meson.build
@@ -0,0 +1,45 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+files_main_test = files('enum_strings.cpp')
+link_main_test = []
+
+if with_shared_glapi
+  files_main_test += files(
+    'dispatch_sanity.cpp',
+    'mesa_formats.cpp',
+    'mesa_extensions.cpp',
+    'program_state_string.cpp',
+  )
+  link_main_test += libglapi
+else
+  files_main_test += files('stubs.cpp')
+endif
+
+test(
+  'main-test',
+  executable(
+    'main_test',
+    [files_main_test, main_dispatch_h],
+    include_directories : [inc_include, inc_src, inc_mapi, inc_mesa],
+    dependencies : [idep_gtest, dep_clock, dep_dl, dep_thread],
+    link_with : [libmesa_classic, link_main_test],
+  )
+)
diff --git a/lib/mesa/src/mesa/main/tests/program_state_string.cpp b/lib/mesa/src/mesa/main/tests/program_state_string.cpp
index ddfdb49ee..38805f92c 100644
--- a/lib/mesa/src/mesa/main/tests/program_state_string.cpp
+++ b/lib/mesa/src/mesa/main/tests/program_state_string.cpp
@@ -25,13 +25,12 @@
 
 #include "GL/gl.h"
 #include "GL/glext.h"
-#include "main/compiler.h"
 
 #include "program/prog_statevars.h"
 
 TEST(program_state_string, depth_range)
 {
-   const gl_state_index state[STATE_LENGTH] = {
+   const gl_state_index16 state[STATE_LENGTH] = {
       STATE_DEPTH_RANGE
    };
 
diff --git a/lib/mesa/src/mesa/main/texcompress_astc.cpp b/lib/mesa/src/mesa/main/texcompress_astc.cpp
new file mode 100644
index 000000000..23540c470
--- /dev/null
+++ b/lib/mesa/src/mesa/main/texcompress_astc.cpp
@@ -0,0 +1,1871 @@
+/*
+ * Copyright 2015 Philip Taylor <philip@zaynar.co.uk>
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ */
+
+/**
+ * \file texcompress_astc.c
+ *
+ * Decompression code for GL_KHR_texture_compression_astc_ldr, which is just
+ * ASTC 2D LDR.
+ *
+ * The ASTC 2D LDR decoder (without the sRGB part) was copied from the OASTC
+ * library written by Philip Taylor. I added sRGB support and adjusted it for
+ * Mesa. - Marek
+ */
+
+#include "texcompress_astc.h"
+#include "macros.h"
+#include "util/half_float.h"
+#include <stdio.h>
+
+static bool VERBOSE_DECODE = false;
+static bool VERBOSE_WRITE = false;
+
+static inline uint8_t
+uint16_div_64k_to_half_to_unorm8(uint16_t v)
+{
+   return _mesa_half_to_unorm8(_mesa_uint16_div_64k_to_half(v));
+}
+
+class decode_error
+{
+public:
+   enum type {
+      ok,
+      unsupported_hdr_void_extent,
+      reserved_block_mode_1,
+      reserved_block_mode_2,
+      dual_plane_and_too_many_partitions,
+      invalid_range_in_void_extent,
+      weight_grid_exceeds_block_size,
+      invalid_colour_endpoints_size,
+      invalid_colour_endpoints_count,
+      invalid_weight_bits,
+      invalid_num_weights,
+   };
+};
+
+
+struct cem_range {
+   uint8_t max;
+   uint8_t t, q, b;
+};
+
+/* Based on the Color Unquantization Parameters table,
+ * plus the bit-only representations, sorted by increasing size
+ */
+static cem_range cem_ranges[] = {
+   { 5, 1, 0, 1 },
+   { 7, 0, 0, 3 },
+   { 9, 0, 1, 1 },
+   { 11, 1, 0, 2 },
+   { 15, 0, 0, 4 },
+   { 19, 0, 1, 2 },
+   { 23, 1, 0, 3 },
+   { 31, 0, 0, 5 },
+   { 39, 0, 1, 3 },
+   { 47, 1, 0, 4 },
+   { 63, 0, 0, 6 },
+   { 79, 0, 1, 4 },
+   { 95, 1, 0, 5 },
+   { 127, 0, 0, 7 },
+   { 159, 0, 1, 5 },
+   { 191, 1, 0, 6 },
+   { 255, 0, 0, 8 },
+};
+
+#define CAT_BITS_2(a, b)          ( ((a) << 1) | (b) )
+#define CAT_BITS_3(a, b, c)       ( ((a) << 2) | ((b) << 1) | (c) )
+#define CAT_BITS_4(a, b, c, d)    ( ((a) << 3) | ((b) << 2) | ((c) << 1) | (d) )
+#define CAT_BITS_5(a, b, c, d, e) ( ((a) << 4) | ((b) << 3) | ((c) << 2) | ((d) << 1) | (e) )
+
+/**
+ * Unpack 5n+8 bits from 'in' into 5 output values.
+ * If n <= 4 then T should be uint32_t, else it must be uint64_t.
+ */
+template <typename T>
+static void unpack_trit_block(int n, T in, uint8_t *out)
+{
+   assert(n <= 6); /* else output will overflow uint8_t */
+
+   uint8_t T0 = (in >> (n)) & 0x1;
+   uint8_t T1 = (in >> (n+1)) & 0x1;
+   uint8_t T2 = (in >> (2*n+2)) & 0x1;
+   uint8_t T3 = (in >> (2*n+3)) & 0x1;
+   uint8_t T4 = (in >> (3*n+4)) & 0x1;
+   uint8_t T5 = (in >> (4*n+5)) & 0x1;
+   uint8_t T6 = (in >> (4*n+6)) & 0x1;
+   uint8_t T7 = (in >> (5*n+7)) & 0x1;
+   uint8_t mmask = (1 << n) - 1;
+   uint8_t m0 = (in >> (0)) & mmask;
+   uint8_t m1 = (in >> (n+2)) & mmask;
+   uint8_t m2 = (in >> (2*n+4)) & mmask;
+   uint8_t m3 = (in >> (3*n+5)) & mmask;
+   uint8_t m4 = (in >> (4*n+7)) & mmask;
+
+   uint8_t C;
+   uint8_t t4, t3, t2, t1, t0;
+   if (CAT_BITS_3(T4, T3, T2) == 0x7) {
+      C = CAT_BITS_5(T7, T6, T5, T1, T0);
+      t4 = t3 = 2;
+   } else {
+      C = CAT_BITS_5(T4, T3, T2, T1, T0);
+      if (CAT_BITS_2(T6, T5) == 0x3) {
+         t4 = 2;
+         t3 = T7;
+      } else {
+         t4 = T7;
+         t3 = CAT_BITS_2(T6, T5);
+      }
+   }
+
+   if ((C & 0x3) == 0x3) {
+      t2 = 2;
+      t1 = (C >> 4) & 0x1;
+      uint8_t C3 = (C >> 3) & 0x1;
+      uint8_t C2 = (C >> 2) & 0x1;
+      t0 = (C3 << 1) | (C2 & ~C3);
+   } else if (((C >> 2) & 0x3) == 0x3) {
+      t2 = 2;
+      t1 = 2;
+      t0 = C & 0x3;
+   } else {
+      t2 = (C >> 4) & 0x1;
+      t1 = (C >> 2) & 0x3;
+      uint8_t C1 = (C >> 1) & 0x1;
+      uint8_t C0 = (C >> 0) & 0x1;
+      t0 = (C1 << 1) | (C0 & ~C1);
+   }
+
+   out[0] = (t0 << n) | m0;
+   out[1] = (t1 << n) | m1;
+   out[2] = (t2 << n) | m2;
+   out[3] = (t3 << n) | m3;
+   out[4] = (t4 << n) | m4;
+}
+
+/**
+ * Unpack 3n+7 bits from 'in' into 3 output values
+ */
+static void unpack_quint_block(int n, uint32_t in, uint8_t *out)
+{
+   assert(n <= 5); /* else output will overflow uint8_t */
+
+   uint8_t Q0 = (in >> (n)) & 0x1;
+   uint8_t Q1 = (in >> (n+1)) & 0x1;
+   uint8_t Q2 = (in >> (n+2)) & 0x1;
+   uint8_t Q3 = (in >> (2*n+3)) & 0x1;
+   uint8_t Q4 = (in >> (2*n+4)) & 0x1;
+   uint8_t Q5 = (in >> (3*n+5)) & 0x1;
+   uint8_t Q6 = (in >> (3*n+6)) & 0x1;
+   uint8_t mmask = (1 << n) - 1;
+   uint8_t m0 = (in >> (0)) & mmask;
+   uint8_t m1 = (in >> (n+3)) & mmask;
+   uint8_t m2 = (in >> (2*n+5)) & mmask;
+
+   uint8_t C;
+   uint8_t q2, q1, q0;
+   if (CAT_BITS_4(Q6, Q5, Q2, Q1) == 0x3) {
+      q2 = CAT_BITS_3(Q0, Q4 & ~Q0, Q3 & ~Q0);
+      q1 = 4;
+      q0 = 4;
+   } else {
+      if (CAT_BITS_2(Q2, Q1) == 0x3) {
+         q2 = 4;
+         C = CAT_BITS_5(Q4, Q3, 0x1 & ~Q6, 0x1 & ~Q5, Q0);
+      } else {
+         q2 = CAT_BITS_2(Q6, Q5);
+         C = CAT_BITS_5(Q4, Q3, Q2, Q1, Q0);
+      }
+      if ((C & 0x7) == 0x5) {
+         q1 = 4;
+         q0 = (C >> 3) & 0x3;
+      } else {
+         q1 = (C >> 3) & 0x3;
+         q0 = C & 0x7;
+      }
+   }
+   out[0] = (q0 << n) | m0;
+   out[1] = (q1 << n) | m1;
+   out[2] = (q2 << n) | m2;
+}
+
+
+struct uint8x4_t
+{
+   uint8_t v[4];
+
+   uint8x4_t() { }
+
+   uint8x4_t(int a, int b, int c, int d)
+   {
+      assert(0 <= a && a <= 255);
+      assert(0 <= b && b <= 255);
+      assert(0 <= c && c <= 255);
+      assert(0 <= d && d <= 255);
+      v[0] = a;
+      v[1] = b;
+      v[2] = c;
+      v[3] = d;
+   }
+
+   static uint8x4_t clamped(int a, int b, int c, int d)
+   {
+      uint8x4_t r;
+      r.v[0] = MAX2(0, MIN2(255, a));
+      r.v[1] = MAX2(0, MIN2(255, b));
+      r.v[2] = MAX2(0, MIN2(255, c));
+      r.v[3] = MAX2(0, MIN2(255, d));
+      return r;
+   }
+};
+
+static uint8x4_t blue_contract(int r, int g, int b, int a)
+{
+   return uint8x4_t((r+b) >> 1, (g+b) >> 1, b, a);
+}
+
+static uint8x4_t blue_contract_clamped(int r, int g, int b, int a)
+{
+   return uint8x4_t::clamped((r+b) >> 1, (g+b) >> 1, b, a);
+}
+
+static void bit_transfer_signed(int &a, int &b)
+{
+   b >>= 1;
+   b |= a & 0x80;
+   a >>= 1;
+   a &= 0x3f;
+   if (a & 0x20)
+      a -= 0x40;
+}
+
+static uint32_t hash52(uint32_t p)
+{
+   p ^= p >> 15;
+   p -= p << 17;
+   p += p << 7;
+   p += p << 4;
+   p ^= p >> 5;
+   p += p << 16;
+   p ^= p >> 7;
+   p ^= p >> 3;
+   p ^= p << 6;
+   p ^= p >> 17;
+   return p;
+}
+
+static int select_partition(int seed, int x, int y, int z, int partitioncount,
+                            int small_block)
+{
+   if (small_block) {
+      x <<= 1;
+      y <<= 1;
+      z <<= 1;
+   }
+   seed += (partitioncount - 1) * 1024;
+   uint32_t rnum = hash52(seed);
+   uint8_t seed1 = rnum & 0xF;
+   uint8_t seed2 = (rnum >> 4) & 0xF;
+   uint8_t seed3 = (rnum >> 8) & 0xF;
+   uint8_t seed4 = (rnum >> 12) & 0xF;
+   uint8_t seed5 = (rnum >> 16) & 0xF;
+   uint8_t seed6 = (rnum >> 20) & 0xF;
+   uint8_t seed7 = (rnum >> 24) & 0xF;
+   uint8_t seed8 = (rnum >> 28) & 0xF;
+   uint8_t seed9 = (rnum >> 18) & 0xF;
+   uint8_t seed10 = (rnum >> 22) & 0xF;
+   uint8_t seed11 = (rnum >> 26) & 0xF;
+   uint8_t seed12 = ((rnum >> 30) | (rnum << 2)) & 0xF;
+
+   seed1 *= seed1;
+   seed2 *= seed2;
+   seed3 *= seed3;
+   seed4 *= seed4;
+   seed5 *= seed5;
+   seed6 *= seed6;
+   seed7 *= seed7;
+   seed8 *= seed8;
+   seed9 *= seed9;
+   seed10 *= seed10;
+   seed11 *= seed11;
+   seed12 *= seed12;
+
+   int sh1, sh2, sh3;
+   if (seed & 1) {
+      sh1 = (seed & 2 ? 4 : 5);
+      sh2 = (partitioncount == 3 ? 6 : 5);
+   } else {
+      sh1 = (partitioncount == 3 ? 6 : 5);
+      sh2 = (seed & 2 ? 4 : 5);
+   }
+   sh3 = (seed & 0x10) ? sh1 : sh2;
+
+   seed1 >>= sh1;
+   seed2 >>= sh2;
+   seed3 >>= sh1;
+   seed4 >>= sh2;
+   seed5 >>= sh1;
+   seed6 >>= sh2;
+   seed7 >>= sh1;
+   seed8 >>= sh2;
+   seed9 >>= sh3;
+   seed10 >>= sh3;
+   seed11 >>= sh3;
+   seed12 >>= sh3;
+
+   int a = seed1 * x + seed2 * y + seed11 * z + (rnum >> 14);
+   int b = seed3 * x + seed4 * y + seed12 * z + (rnum >> 10);
+   int c = seed5 * x + seed6 * y + seed9 * z + (rnum >> 6);
+   int d = seed7 * x + seed8 * y + seed10 * z + (rnum >> 2);
+
+   a &= 0x3F;
+   b &= 0x3F;
+   c &= 0x3F;
+   d &= 0x3F;
+
+   if (partitioncount < 4)
+      d = 0;
+   if (partitioncount < 3)
+      c = 0;
+
+   if (a >= b && a >= c && a >= d)
+      return 0;
+   else if (b >= c && b >= d)
+      return 1;
+   else if (c >= d)
+      return 2;
+   else
+      return 3;
+}
+
+
+struct InputBitVector
+{
+   uint32_t data[4];
+
+   void printf_bits(int offset, int count, const char *fmt = "", ...)
+   {
+      char out[129];
+      memset(out, '.', 128);
+      out[128] = '\0';
+      int idx = offset;
+      for (int i = 0; i < count; ++i) {
+         out[127 - idx] = ((data[idx >> 5] >> (idx & 31)) & 1) ? '1' : '0';
+         ++idx;
+      }
+      printf("%s ", out);
+      va_list ap;
+      va_start(ap, fmt);
+      vprintf(fmt, ap);
+      va_end(ap);
+      printf("\n");
+   }
+
+   uint32_t get_bits(int offset, int count)
+   {
+      assert(count >= 0 && count < 32);
+
+      uint32_t out = 0;
+      if (offset < 32)
+         out |= data[0] >> offset;
+
+      if (0 < offset && offset <= 32)
+         out |= data[1] << (32 - offset);
+      if (32 < offset && offset < 64)
+         out |= data[1] >> (offset - 32);
+
+      if (32 < offset && offset <= 64)
+         out |= data[2] << (64 - offset);
+      if (64 < offset && offset < 96)
+         out |= data[2] >> (offset - 64);
+
+      if (64 < offset && offset <= 96)
+         out |= data[3] << (96 - offset);
+      if (96 < offset && offset < 128)
+         out |= data[3] >> (offset - 96);
+
+      out &= (1 << count) - 1;
+      return out;
+   }
+
+   uint64_t get_bits64(int offset, int count)
+   {
+      assert(count >= 0 && count < 64);
+
+      uint64_t out = 0;
+      if (offset < 32)
+         out |= data[0] >> offset;
+
+      if (offset <= 32)
+         out |= (uint64_t)data[1] << (32 - offset);
+      if (32 < offset && offset < 64)
+         out |= data[1] >> (offset - 32);
+
+      if (0 < offset && offset <= 64)
+         out |= (uint64_t)data[2] << (64 - offset);
+      if (64 < offset && offset < 96)
+         out |= data[2] >> (offset - 64);
+
+      if (32 < offset && offset <= 96)
+         out |= (uint64_t)data[3] << (96 - offset);
+      if (96 < offset && offset < 128)
+         out |= data[3] >> (offset - 96);
+
+      out &= ((uint64_t)1 << count) - 1;
+      return out;
+   }
+
+   uint32_t get_bits_rev(int offset, int count)
+   {
+      assert(offset >= count);
+      uint32_t tmp = get_bits(offset - count, count);
+      uint32_t out = 0;
+      for (int i = 0; i < count; ++i)
+         out |= ((tmp >> i) & 1) << (count - 1 - i);
+      return out;
+   }
+};
+
+struct OutputBitVector
+{
+   uint32_t data[4];
+   int offset;
+
+   OutputBitVector()
+      : offset(0)
+   {
+      memset(data, 0, sizeof(data));
+   }
+
+   void append(uint32_t value, int size)
+   {
+      if (VERBOSE_WRITE)
+         printf("append offset=%d size=%d values=0x%x\n", offset, size, value);
+
+      assert(offset + size <= 128);
+
+      assert(size <= 32);
+      if (size < 32)
+         assert((value >> size) == 0);
+
+      while (size) {
+         int c = MIN2(size, 32 - (offset & 31));
+         data[offset >> 5] |= (value << (offset & 31));
+         offset += c;
+         size -= c;
+         value >>= c;
+      }
+   }
+
+   void append64(uint64_t value, int size)
+   {
+      if (VERBOSE_WRITE)
+         printf("append offset=%d size=%d values=0x%llx\n", offset, size, (unsigned long long)value);
+
+      assert(offset + size <= 128);
+
+      assert(size <= 64);
+      if (size < 64)
+         assert((value >> size) == 0);
+
+      while (size) {
+         int c = MIN2(size, 32 - (offset & 31));
+         data[offset >> 5] |= (value << (offset & 31));
+         offset += c;
+         size -= c;
+         value >>= c;
+      }
+   }
+
+   void append(OutputBitVector &v, int size)
+   {
+      if (VERBOSE_WRITE)
+         printf("append vector offset=%d size=%d\n", offset, size);
+
+      assert(offset + size <= 128);
+      int i = 0;
+      while (size >= 32) {
+         append(v.data[i++], 32);
+         size -= 32;
+      }
+      if (size > 0)
+         append(v.data[i] & ((1 << size) - 1), size);
+   }
+
+   void append_end(OutputBitVector &v, int size)
+   {
+      for (int i = 0; i < size; ++i)
+         data[(127 - i) >> 5] |= ((v.data[i >> 5] >> (i & 31)) & 1) << ((127 - i) & 31);
+   }
+
+   /* Insert the given number of '1' bits. (We could use 0s instead, but 1s are
+    * more likely to flush out bugs where we accidentally read undefined bits.)
+    */
+   void skip(int size)
+   {
+      if (VERBOSE_WRITE)
+         printf("skip offset=%d size=%d\n", offset, size);
+
+      assert(offset + size <= 128);
+      while (size >= 32) {
+         append(0xffffffff, 32);
+         size -= 32;
+      }
+      if (size > 0)
+         append(0xffffffff >> (32 - size), size);
+   }
+};
+
+
+class Decoder
+{
+public:
+   Decoder(int block_w, int block_h, int block_d, bool srgb, bool output_unorm8)
+      : block_w(block_w), block_h(block_h), block_d(block_d), srgb(srgb),
+        output_unorm8(output_unorm8) {}
+
+   decode_error::type decode(const uint8_t *in, uint16_t *output) const;
+
+   int block_w, block_h, block_d;
+   bool srgb, output_unorm8;
+};
+
+struct Block
+{
+   bool is_error;
+   bool bogus_colour_endpoints;
+   bool bogus_weights;
+
+   int high_prec;
+   int dual_plane;
+   int colour_component_selector;
+   int wt_range;
+   int wt_w, wt_h, wt_d;
+   int num_parts;
+   int partition_index;
+
+   bool is_void_extent;
+   int void_extent_d;
+   int void_extent_min_s;
+   int void_extent_max_s;
+   int void_extent_min_t;
+   int void_extent_max_t;
+   uint16_t void_extent_colour_r;
+   uint16_t void_extent_colour_g;
+   uint16_t void_extent_colour_b;
+   uint16_t void_extent_colour_a;
+
+   bool is_multi_cem;
+   int num_extra_cem_bits;
+   int colour_endpoint_data_offset;
+   int extra_cem_bits;
+   int cem_base_class;
+   int cems[4];
+
+   int num_cem_values;
+
+   /* Calculated by unpack_weights(): */
+   uint8_t weights_quant[64 + 4]; /* max 64 values, plus padding for overflows in trit parsing */
+
+   /* Calculated by unquantise_weights(): */
+   uint8_t weights[64 + 18]; /* max 64 values, plus padding for the infill interpolation */
+
+   /* Calculated by unpack_colour_endpoints(): */
+   uint8_t colour_endpoints_quant[18 + 4]; /* max 18 values, plus padding for overflows in trit parsing */
+
+   /* Calculated by unquantise_colour_endpoints(): */
+   uint8_t colour_endpoints[18];
+
+   /* Calculated by calculate_from_weights(): */
+   int wt_trits;
+   int wt_quints;
+   int wt_bits;
+   int wt_max;
+   int num_weights;
+   int weight_bits;
+
+   /* Calculated by calculate_remaining_bits(): */
+   int remaining_bits;
+
+   /* Calculated by calculate_colour_endpoints_size(): */
+   int colour_endpoint_bits;
+   int ce_max;
+   int ce_trits;
+   int ce_quints;
+   int ce_bits;
+
+   /* Calculated by compute_infill_weights(); */
+   uint8_t infill_weights[2][216]; /* large enough for 6x6x6 */
+
+   /* Calculated by decode_colour_endpoints(); */
+   uint8x4_t endpoints_decoded[2][4];
+
+   void calculate_from_weights();
+   void calculate_remaining_bits();
+   decode_error::type calculate_colour_endpoints_size();
+
+   void unquantise_weights();
+   void unquantise_colour_endpoints();
+
+   decode_error::type decode(const Decoder &decoder, InputBitVector in);
+
+   decode_error::type decode_block_mode(InputBitVector in);
+   decode_error::type decode_void_extent(InputBitVector in);
+   void decode_cem(InputBitVector in);
+   void unpack_colour_endpoints(InputBitVector in);
+   void decode_colour_endpoints();
+   void unpack_weights(InputBitVector in);
+   void compute_infill_weights(int block_w, int block_h, int block_d);
+
+   void write_decoded(const Decoder &decoder, uint16_t *output);
+};
+
+
+decode_error::type Decoder::decode(const uint8_t *in, uint16_t *output) const
+{
+   Block blk;
+   InputBitVector in_vec;
+   memcpy(&in_vec.data, in, 16);
+   decode_error::type err = blk.decode(*this, in_vec);
+   if (err == decode_error::ok) {
+      blk.write_decoded(*this, output);
+   } else {
+      /* Fill output with the error colour */
+      for (int i = 0; i < block_w * block_h * block_d; ++i) {
+         if (output_unorm8) {
+            output[i*4+0] = 0xff;
+            output[i*4+1] = 0;
+            output[i*4+2] = 0xff;
+            output[i*4+3] = 0xff;
+         } else {
+            assert(!srgb); /* srgb must use unorm8 */
+
+            output[i*4+0] = FP16_ONE;
+            output[i*4+1] = FP16_ZERO;
+            output[i*4+2] = FP16_ONE;
+            output[i*4+3] = FP16_ONE;
+         }
+      }
+   }
+   return err;
+}
+
+
+decode_error::type Block::decode_void_extent(InputBitVector block)
+{
+   /* TODO: 3D */
+
+   is_void_extent = true;
+   void_extent_d = block.get_bits(9, 1);
+   void_extent_min_s = block.get_bits(12, 13);
+   void_extent_max_s = block.get_bits(25, 13);
+   void_extent_min_t = block.get_bits(38, 13);
+   void_extent_max_t = block.get_bits(51, 13);
+   void_extent_colour_r = block.get_bits(64, 16);
+   void_extent_colour_g = block.get_bits(80, 16);
+   void_extent_colour_b = block.get_bits(96, 16);
+   void_extent_colour_a = block.get_bits(112, 16);
+
+   /* TODO: maybe we should do something useful with the extent coordinates? */
+
+   if (void_extent_d) {
+      return decode_error::unsupported_hdr_void_extent;
+   }
+
+   if (void_extent_min_s == 0x1fff && void_extent_max_s == 0x1fff
+       && void_extent_min_t == 0x1fff && void_extent_max_t == 0x1fff) {
+
+      /* No extents */
+
+   } else {
+
+      /* Check for illegal encoding */
+      if (void_extent_min_s >= void_extent_max_s || void_extent_min_t >= void_extent_max_t) {
+         return decode_error::invalid_range_in_void_extent;
+      }
+   }
+
+   return decode_error::ok;
+}
+
+decode_error::type Block::decode_block_mode(InputBitVector in)
+{
+   dual_plane = in.get_bits(10, 1);
+   high_prec = in.get_bits(9, 1);
+
+   if (in.get_bits(0, 2) != 0x0) {
+      wt_range = (in.get_bits(0, 2) << 1) | in.get_bits(4, 1);
+      int a = in.get_bits(5, 2);
+      int b = in.get_bits(7, 2);
+      switch (in.get_bits(2, 2)) {
+      case 0x0:
+         if (VERBOSE_DECODE)
+            in.printf_bits(0, 11, "DHBBAAR00RR");
+         wt_w = b + 4;
+         wt_h = a + 2;
+         break;
+      case 0x1:
+         if (VERBOSE_DECODE)
+            in.printf_bits(0, 11, "DHBBAAR01RR");
+         wt_w = b + 8;
+         wt_h = a + 2;
+         break;
+      case 0x2:
+         if (VERBOSE_DECODE)
+            in.printf_bits(0, 11, "DHBBAAR10RR");
+         wt_w = a + 2;
+         wt_h = b + 8;
+         break;
+      case 0x3:
+         if ((b & 0x2) == 0) {
+            if (VERBOSE_DECODE)
+               in.printf_bits(0, 11, "DH0BAAR11RR");
+            wt_w = a + 2;
+            wt_h = b + 6;
+         } else {
+            if (VERBOSE_DECODE)
+               in.printf_bits(0, 11, "DH1BAAR11RR");
+            wt_w = (b & 0x1) + 2;
+            wt_h = a + 2;
+         }
+         break;
+      }
+   } else {
+      if (in.get_bits(6, 3) == 0x7) {
+         if (in.get_bits(0, 9) == 0x1fc) {
+            if (VERBOSE_DECODE)
+               in.printf_bits(0, 11, "xx111111100 (void extent)");
+            return decode_void_extent(in);
+         } else {
+            if (VERBOSE_DECODE)
+               in.printf_bits(0, 11, "xx111xxxx00");
+            return decode_error::reserved_block_mode_1;
+         }
+      }
+      if (in.get_bits(0, 4) == 0x0) {
+         if (VERBOSE_DECODE)
+            in.printf_bits(0, 11, "xxxxxxx0000");
+         return decode_error::reserved_block_mode_2;
+      }
+
+      wt_range = in.get_bits(1, 3) | in.get_bits(4, 1);
+      int a = in.get_bits(5, 2);
+      int b;
+
+      switch (in.get_bits(7, 2)) {
+      case 0x0:
+         if (VERBOSE_DECODE)
+            in.printf_bits(0, 11, "DH00AARRR00");
+         wt_w = 12;
+         wt_h = a + 2;
+         break;
+      case 0x1:
+         if (VERBOSE_DECODE)
+            in.printf_bits(0, 11, "DH01AARRR00");
+         wt_w = a + 2;
+         wt_h = 12;
+         break;
+      case 0x3:
+         if (in.get_bits(5, 1) == 0) {
+            if (VERBOSE_DECODE)
+               in.printf_bits(0, 11, "DH1100RRR00");
+            wt_w = 6;
+            wt_h = 10;
+         } else {
+            if (VERBOSE_DECODE)
+               in.printf_bits(0, 11, "DH1101RRR00");
+            wt_w = 10;
+            wt_h = 6;
+         }
+         break;
+      case 0x2:
+         if (VERBOSE_DECODE)
+            in.printf_bits(0, 11, "BB10AARRR00");
+         b = in.get_bits(9, 2);
+         wt_w = a + 6;
+         wt_h = b + 6;
+         dual_plane = 0;
+         high_prec = 0;
+         break;
+      }
+   }
+   return decode_error::ok;
+}
+
+void Block::decode_cem(InputBitVector in)
+{
+   cems[0] = cems[1] = cems[2] = cems[3] = -1;
+
+   num_extra_cem_bits = 0;
+   extra_cem_bits = 0;
+
+   if (num_parts > 1) {
+
+      partition_index = in.get_bits(13, 10);
+      if (VERBOSE_DECODE)
+         in.printf_bits(13, 10, "partition ID (%d)", partition_index);
+
+      uint32_t cem = in.get_bits(23, 6);
+
+      if ((cem & 0x3) == 0x0) {
+         cem >>= 2;
+         cem_base_class = cem >> 2;
+         is_multi_cem = false;
+
+         for (int i = 0; i < num_parts; ++i)
+            cems[i] = cem;
+
+         if (VERBOSE_DECODE)
+            in.printf_bits(23, 6, "CEM (single, %d)", cem);
+      } else {
+
+         cem_base_class = (cem & 0x3) - 1;
+         is_multi_cem = true;
+
+         if (VERBOSE_DECODE)
+            in.printf_bits(23, 6, "CEM (multi, base class %d)", cem_base_class);
+
+         int offset = 128 - weight_bits;
+
+         if (num_parts == 2) {
+            if (VERBOSE_DECODE) {
+               in.printf_bits(25, 4, "M0M0 C1 C0");
+               in.printf_bits(offset - 2, 2, "M1M1");
+            }
+
+            uint32_t c0 = in.get_bits(25, 1);
+            uint32_t c1 = in.get_bits(26, 1);
+
+            extra_cem_bits = c0 + c1;
+
+            num_extra_cem_bits = 2;
+
+            uint32_t m0 = in.get_bits(27, 2);
+            uint32_t m1 = in.get_bits(offset - 2, 2);
+
+            cems[0] = ((cem_base_class + c0) << 2) | m0;
+            cems[1] = ((cem_base_class + c1) << 2) | m1;
+
+         } else if (num_parts == 3) {
+            if (VERBOSE_DECODE) {
+               in.printf_bits(25, 4, "M0 C2 C1 C0");
+               in.printf_bits(offset - 5, 5, "M2M2 M1M1 M0");
+            }
+
+            uint32_t c0 = in.get_bits(25, 1);
+            uint32_t c1 = in.get_bits(26, 1);
+            uint32_t c2 = in.get_bits(27, 1);
+
+            extra_cem_bits = c0 + c1 + c2;
+
+            num_extra_cem_bits = 5;
+
+            uint32_t m0 = in.get_bits(28, 1) | (in.get_bits(128 - weight_bits - 5, 1) << 1);
+            uint32_t m1 = in.get_bits(offset - 4, 2);
+            uint32_t m2 = in.get_bits(offset - 2, 2);
+
+            cems[0] = ((cem_base_class + c0) << 2) | m0;
+            cems[1] = ((cem_base_class + c1) << 2) | m1;
+            cems[2] = ((cem_base_class + c2) << 2) | m2;
+
+         } else if (num_parts == 4) {
+            if (VERBOSE_DECODE) {
+               in.printf_bits(25, 4, "C3 C2 C1 C0");
+               in.printf_bits(offset - 8, 8, "M3M3 M2M2 M1M1 M0M0");
+            }
+
+            uint32_t c0 = in.get_bits(25, 1);
+            uint32_t c1 = in.get_bits(26, 1);
+            uint32_t c2 = in.get_bits(27, 1);
+            uint32_t c3 = in.get_bits(28, 1);
+
+            extra_cem_bits = c0 + c1 + c2 + c3;
+
+            num_extra_cem_bits = 8;
+
+            uint32_t m0 = in.get_bits(offset - 8, 2);
+            uint32_t m1 = in.get_bits(offset - 6, 2);
+            uint32_t m2 = in.get_bits(offset - 4, 2);
+            uint32_t m3 = in.get_bits(offset - 2, 2);
+
+            cems[0] = ((cem_base_class + c0) << 2) | m0;
+            cems[1] = ((cem_base_class + c1) << 2) | m1;
+            cems[2] = ((cem_base_class + c2) << 2) | m2;
+            cems[3] = ((cem_base_class + c3) << 2) | m3;
+         } else {
+            unreachable("");
+         }
+      }
+
+      colour_endpoint_data_offset = 29;
+
+   } else {
+      uint32_t cem = in.get_bits(13, 4);
+
+      cem_base_class = cem >> 2;
+      is_multi_cem = false;
+
+      cems[0] = cem;
+
+      partition_index = -1;
+
+      if (VERBOSE_DECODE)
+         in.printf_bits(13, 4, "CEM = %d (class %d)", cem, cem_base_class);
+
+      colour_endpoint_data_offset = 17;
+   }
+}
+
+void Block::unpack_colour_endpoints(InputBitVector in)
+{
+   if (ce_trits) {
+      int offset = colour_endpoint_data_offset;
+      int bits_left = colour_endpoint_bits;
+      for (int i = 0; i < num_cem_values; i += 5) {
+         int bits_to_read = MIN2(bits_left, 8 + ce_bits * 5);
+         /* If ce_trits then ce_bits <= 6, so bits_to_read <= 38 and we have to use uint64_t */
+         uint64_t raw = in.get_bits64(offset, bits_to_read);
+         unpack_trit_block(ce_bits, raw, &colour_endpoints_quant[i]);
+
+         if (VERBOSE_DECODE)
+            in.printf_bits(offset, bits_to_read,
+                           "trits [%d,%d,%d,%d,%d]",
+                           colour_endpoints_quant[i+0], colour_endpoints_quant[i+1],
+                  colour_endpoints_quant[i+2], colour_endpoints_quant[i+3],
+                  colour_endpoints_quant[i+4]);
+
+         offset += 8 + ce_bits * 5;
+         bits_left -= 8 + ce_bits * 5;
+      }
+   } else if (ce_quints) {
+      int offset = colour_endpoint_data_offset;
+      int bits_left = colour_endpoint_bits;
+      for (int i = 0; i < num_cem_values; i += 3) {
+         int bits_to_read = MIN2(bits_left, 7 + ce_bits * 3);
+         /* If ce_quints then ce_bits <= 5, so bits_to_read <= 22 and we can use uint32_t */
+         uint32_t raw = in.get_bits(offset, bits_to_read);
+         unpack_quint_block(ce_bits, raw, &colour_endpoints_quant[i]);
+
+         if (VERBOSE_DECODE)
+            in.printf_bits(offset, bits_to_read,
+                           "quints [%d,%d,%d]",
+                           colour_endpoints_quant[i], colour_endpoints_quant[i+1], colour_endpoints_quant[i+2]);
+
+         offset += 7 + ce_bits * 3;
+         bits_left -= 7 + ce_bits * 3;
+      }
+   } else {
+      assert((colour_endpoint_bits % ce_bits) == 0);
+      int offset = colour_endpoint_data_offset;
+      for (int i = 0; i < num_cem_values; i++) {
+         colour_endpoints_quant[i] = in.get_bits(offset, ce_bits);
+
+         if (VERBOSE_DECODE)
+            in.printf_bits(offset, ce_bits, "bits [%d]", colour_endpoints_quant[i]);
+
+         offset += ce_bits;
+      }
+   }
+}
+
+void Block::decode_colour_endpoints()
+{
+   int cem_values_idx = 0;
+   for (int part = 0; part < num_parts; ++part) {
+      uint8_t *v = &colour_endpoints[cem_values_idx];
+      int v0 = v[0];
+      int v1 = v[1];
+      int v2 = v[2];
+      int v3 = v[3];
+      int v4 = v[4];
+      int v5 = v[5];
+      int v6 = v[6];
+      int v7 = v[7];
+      cem_values_idx += ((cems[part] >> 2) + 1) * 2;
+
+      uint8x4_t e0, e1;
+      int s0, s1, L0, L1;
+
+      switch (cems[part])
+      {
+      case 0:
+         e0 = uint8x4_t(v0, v0, v0, 0xff);
+         e1 = uint8x4_t(v1, v1, v1, 0xff);
+         break;
+      case 1:
+         L0 = (v0 >> 2) | (v1 & 0xc0);
+         L1 = L0 + (v1 & 0x3f);
+         if (L1 > 0xff)
+            L1 = 0xff;
+         e0 = uint8x4_t(L0, L0, L0, 0xff);
+         e1 = uint8x4_t(L1, L1, L1, 0xff);
+         break;
+      case 4:
+         e0 = uint8x4_t(v0, v0, v0, v2);
+         e1 = uint8x4_t(v1, v1, v1, v3);
+         break;
+      case 5:
+         bit_transfer_signed(v1, v0);
+         bit_transfer_signed(v3, v2);
+         e0 = uint8x4_t(v0, v0, v0, v2);
+         e1 = uint8x4_t::clamped(v0+v1, v0+v1, v0+v1, v2+v3);
+         break;
+      case 6:
+         e0 = uint8x4_t(v0*v3 >> 8, v1*v3 >> 8, v2*v3 >> 8, 0xff);
+         e1 = uint8x4_t(v0, v1, v2, 0xff);
+         break;
+      case 8:
+         s0 = v0 + v2 + v4;
+         s1 = v1 + v3 + v5;
+         if (s1 >= s0) {
+            e0 = uint8x4_t(v0, v2, v4, 0xff);
+            e1 = uint8x4_t(v1, v3, v5, 0xff);
+         } else {
+            e0 = blue_contract(v1, v3, v5, 0xff);
+            e1 = blue_contract(v0, v2, v4, 0xff);
+         }
+         break;
+      case 9:
+         bit_transfer_signed(v1, v0);
+         bit_transfer_signed(v3, v2);
+         bit_transfer_signed(v5, v4);
+         if (v1 + v3 + v5 >= 0) {
+            e0 = uint8x4_t(v0, v2, v4, 0xff);
+            e1 = uint8x4_t::clamped(v0+v1, v2+v3, v4+v5, 0xff);
+         } else {
+            e0 = blue_contract_clamped(v0+v1, v2+v3, v4+v5, 0xff);
+            e1 = blue_contract(v0, v2, v4, 0xff);
+         }
+         break;
+      case 10:
+         e0 = uint8x4_t(v0*v3 >> 8, v1*v3 >> 8, v2*v3 >> 8, v4);
+         e1 = uint8x4_t(v0, v1, v2, v5);
+         break;
+      case 12:
+         s0 = v0 + v2 + v4;
+         s1 = v1 + v3 + v5;
+         if (s1 >= s0) {
+            e0 = uint8x4_t(v0, v2, v4, v6);
+            e1 = uint8x4_t(v1, v3, v5, v7);
+         } else {
+            e0 = blue_contract(v1, v3, v5, v7);
+            e1 = blue_contract(v0, v2, v4, v6);
+         }
+         break;
+      case 13:
+         bit_transfer_signed(v1, v0);
+         bit_transfer_signed(v3, v2);
+         bit_transfer_signed(v5, v4);
+         bit_transfer_signed(v7, v6);
+         if (v1 + v3 + v5 >= 0) {
+            e0 = uint8x4_t(v0, v2, v4, v6);
+            e1 = uint8x4_t::clamped(v0+v1, v2+v3, v4+v5, v6+v7);
+         } else {
+            e0 = blue_contract_clamped(v0+v1, v2+v3, v4+v5, v6+v7);
+            e1 = blue_contract(v0, v2, v4, v6);
+         }
+         break;
+      default:
+         /* HDR endpoints not supported; return error colour */
+         e0 = uint8x4_t(255, 0, 255, 255);
+         e1 = uint8x4_t(255, 0, 255, 255);
+         break;
+      }
+
+      endpoints_decoded[0][part] = e0;
+      endpoints_decoded[1][part] = e1;
+
+      if (VERBOSE_DECODE) {
+         printf("cems[%d]=%d v=[", part, cems[part]);
+         for (int i = 0; i < (cems[part] >> 2) + 1; ++i) {
+            if (i)
+               printf(", ");
+            printf("%3d", v[i]);
+         }
+         printf("] e0=[%3d,%4d,%4d,%4d] e1=[%3d,%4d,%4d,%4d]\n",
+                e0.v[0], e0.v[1], e0.v[2], e0.v[3],
+               e1.v[0], e1.v[1], e1.v[2], e1.v[3]);
+      }
+   }
+}
+
+void Block::unpack_weights(InputBitVector in)
+{
+   if (wt_trits) {
+      int offset = 128;
+      int bits_left = weight_bits;
+      for (int i = 0; i < num_weights; i += 5) {
+         int bits_to_read = MIN2(bits_left, 8 + 5*wt_bits);
+         /* If wt_trits then wt_bits <= 3, so bits_to_read <= 23 and we can use uint32_t */
+         uint32_t raw = in.get_bits_rev(offset, bits_to_read);
+         unpack_trit_block(wt_bits, raw, &weights_quant[i]);
+
+         if (VERBOSE_DECODE)
+            in.printf_bits(offset - bits_to_read, bits_to_read, "weight trits [%d,%d,%d,%d,%d]",
+                           weights_quant[i+0], weights_quant[i+1],
+                  weights_quant[i+2], weights_quant[i+3],
+                  weights_quant[i+4]);
+
+         offset -= 8 + wt_bits * 5;
+         bits_left -= 8 + wt_bits * 5;
+      }
+
+   } else if (wt_quints) {
+
+      int offset = 128;
+      int bits_left = weight_bits;
+      for (int i = 0; i < num_weights; i += 3) {
+         int bits_to_read = MIN2(bits_left, 7 + 3*wt_bits);
+         /* If wt_quints then wt_bits <= 2, so bits_to_read <= 13 and we can use uint32_t */
+         uint32_t raw = in.get_bits_rev(offset, bits_to_read);
+         unpack_quint_block(wt_bits, raw, &weights_quant[i]);
+
+         if (VERBOSE_DECODE)
+            in.printf_bits(offset - bits_to_read, bits_to_read, "weight quints [%d,%d,%d]",
+                           weights_quant[i], weights_quant[i+1], weights_quant[i+2]);
+
+         offset -= 7 + wt_bits * 3;
+         bits_left -= 7 + wt_bits * 3;
+      }
+
+   } else {
+      int offset = 128;
+      assert((weight_bits % wt_bits) == 0);
+      for (int i = 0; i < num_weights; ++i) {
+         weights_quant[i] = in.get_bits_rev(offset, wt_bits);
+
+         if (VERBOSE_DECODE)
+            in.printf_bits(offset - wt_bits, wt_bits, "weight bits [%d]", weights_quant[i]);
+
+         offset -= wt_bits;
+      }
+   }
+}
+
+void Block::unquantise_weights()
+{
+   assert(num_weights <= (int)ARRAY_SIZE(weights_quant));
+   assert(num_weights <= (int)ARRAY_SIZE(weights));
+
+   memset(weights, 0, sizeof(weights));
+
+   for (int i = 0; i < num_weights; ++i) {
+
+      uint8_t v = weights_quant[i];
+      uint8_t w;
+
+      if (wt_trits) {
+
+         if (wt_bits == 0) {
+            w = v * 32;
+         } else {
+            uint8_t A, B, C, D;
+            A = (v & 0x1) ? 0x7F : 0x00;
+            switch (wt_bits) {
+            case 1:
+               B = 0;
+               C = 50;
+               D = v >> 1;
+               break;
+            case 2:
+               B = (v & 0x2) ? 0x45 : 0x00;
+               C = 23;
+               D = v >> 2;
+               break;
+            case 3:
+               B = ((v & 0x6) >> 1) | ((v & 0x6) << 4);
+               C = 11;
+               D = v >> 3;
+               break;
+            default:
+               unreachable("");
+            }
+            uint16_t T = D * C + B;
+            T = T ^ A;
+            T = (A & 0x20) | (T >> 2);
+            assert(T < 64);
+            if (T > 32)
+               T++;
+            w = T;
+         }
+
+      } else if (wt_quints) {
+
+         if (wt_bits == 0) {
+            w = v * 16;
+         } else {
+            uint8_t A, B, C, D;
+            A = (v & 0x1) ? 0x7F : 0x00;
+            switch (wt_bits) {
+            case 1:
+               B = 0;
+               C = 28;
+               D = v >> 1;
+               break;
+            case 2:
+               B = (v & 0x2) ? 0x42 : 0x00;
+               C = 13;
+               D = v >> 2;
+               break;
+            default:
+               unreachable("");
+            }
+            uint16_t T = D * C + B;
+            T = T ^ A;
+            T = (A & 0x20) | (T >> 2);
+            assert(T < 64);
+            if (T > 32)
+               T++;
+            w = T;
+         }
+         weights[i] = w;
+
+      } else {
+
+         switch (wt_bits) {
+         case 1: w = v ? 0x3F : 0x00; break;
+         case 2: w = v | (v << 2) | (v << 4); break;
+         case 3: w = v | (v << 3); break;
+         case 4: w = (v >> 2) | (v << 2); break;
+         case 5: w = (v >> 4) | (v << 1); break;
+         default: unreachable("");
+         }
+         assert(w < 64);
+         if (w > 32)
+            w++;
+      }
+      weights[i] = w;
+   }
+}
+
+void Block::compute_infill_weights(int block_w, int block_h, int block_d)
+{
+   int Ds = block_w <= 1 ? 0 : (1024 + block_w / 2) / (block_w - 1);
+   int Dt = block_h <= 1 ? 0 : (1024 + block_h / 2) / (block_h - 1);
+   int Dr = block_d <= 1 ? 0 : (1024 + block_d / 2) / (block_d - 1);
+   for (int r = 0; r < block_d; ++r) {
+      for (int t = 0; t < block_h; ++t) {
+         for (int s = 0; s < block_w; ++s) {
+            int cs = Ds * s;
+            int ct = Dt * t;
+            int cr = Dr * r;
+            int gs = (cs * (wt_w - 1) + 32) >> 6;
+            int gt = (ct * (wt_h - 1) + 32) >> 6;
+            int gr = (cr * (wt_d - 1) + 32) >> 6;
+            assert(gs >= 0 && gs <= 176);
+            assert(gt >= 0 && gt <= 176);
+            assert(gr >= 0 && gr <= 176);
+            int js = gs >> 4;
+            int fs = gs & 0xf;
+            int jt = gt >> 4;
+            int ft = gt & 0xf;
+            int jr = gr >> 4;
+            int fr = gr & 0xf;
+
+            /* TODO: 3D */
+            (void)jr;
+            (void)fr;
+
+            int w11 = (fs * ft + 8) >> 4;
+            int w10 = ft - w11;
+            int w01 = fs - w11;
+            int w00 = 16 - fs - ft + w11;
+
+            if (dual_plane) {
+               int p00, p01, p10, p11, i0, i1;
+               int v0 = js + jt * wt_w;
+               p00 = weights[(v0) * 2];
+               p01 = weights[(v0 + 1) * 2];
+               p10 = weights[(v0 + wt_w) * 2];
+               p11 = weights[(v0 + wt_w + 1) * 2];
+               i0 = (p00*w00 + p01*w01 + p10*w10 + p11*w11 + 8) >> 4;
+               p00 = weights[(v0) * 2 + 1];
+               p01 = weights[(v0 + 1) * 2 + 1];
+               p10 = weights[(v0 + wt_w) * 2 + 1];
+               p11 = weights[(v0 + wt_w + 1) * 2 + 1];
+               assert((v0 + wt_w + 1) * 2 + 1 < (int)ARRAY_SIZE(weights));
+               i1 = (p00*w00 + p01*w01 + p10*w10 + p11*w11 + 8) >> 4;
+               assert(0 <= i0 && i0 <= 64);
+               infill_weights[0][s + t*block_w + r*block_w*block_h] = i0;
+               infill_weights[1][s + t*block_w + r*block_w*block_h] = i1;
+            } else {
+               int p00, p01, p10, p11, i;
+               int v0 = js + jt * wt_w;
+               p00 = weights[v0];
+               p01 = weights[v0 + 1];
+               p10 = weights[v0 + wt_w];
+               p11 = weights[v0 + wt_w + 1];
+               assert(v0 + wt_w + 1 < (int)ARRAY_SIZE(weights));
+               i = (p00*w00 + p01*w01 + p10*w10 + p11*w11 + 8) >> 4;
+               assert(0 <= i && i <= 64);
+               infill_weights[0][s + t*block_w + r*block_w*block_h] = i;
+            }
+         }
+      }
+   }
+}
+
+void Block::unquantise_colour_endpoints()
+{
+   assert(num_cem_values <= (int)ARRAY_SIZE(colour_endpoints_quant));
+   assert(num_cem_values <= (int)ARRAY_SIZE(colour_endpoints));
+
+   for (int i = 0; i < num_cem_values; ++i) {
+      uint8_t v = colour_endpoints_quant[i];
+
+      if (ce_trits) {
+         uint16_t A, B, C, D;
+         uint16_t t;
+         A = (v & 0x1) ? 0x1FF : 0x000;
+         switch (ce_bits) {
+         case 1:
+            B = 0;
+            C = 204;
+            D = v >> 1;
+            break;
+         case 2:
+            B = (v & 0x2) ? 0x116 : 0x000;
+            C = 93;
+            D = v >> 2;
+            break;
+         case 3:
+            t = ((v >> 1) & 0x3);
+            B = t | (t << 2) | (t << 7);
+            C = 44;
+            D = v >> 3;
+            break;
+         case 4:
+            t = ((v >> 1) & 0x7);
+            B = t | (t << 6);
+            C = 22;
+            D = v >> 4;
+            break;
+         case 5:
+            t = ((v >> 1) & 0xF);
+            B = (t >> 2) | (t << 5);
+            C = 11;
+            D = v >> 5;
+            break;
+         case 6:
+            B = ((v & 0x3E) << 3) | ((v >> 5) & 0x1);
+            C = 5;
+            D = v >> 6;
+            break;
+         default:
+            unreachable("");
+         }
+         uint16_t T = D * C + B;
+         T = T ^ A;
+         T = (A & 0x80) | (T >> 2);
+         assert(T < 256);
+         colour_endpoints[i] = T;
+      } else if (ce_quints) {
+         uint16_t A, B, C, D;
+         uint16_t t;
+         A = (v & 0x1) ? 0x1FF : 0x000;
+         switch (ce_bits) {
+         case 1:
+            B = 0;
+            C = 113;
+            D = v >> 1;
+            break;
+         case 2:
+            B = (v & 0x2) ? 0x10C : 0x000;
+            C = 54;
+            D = v >> 2;
+            break;
+         case 3:
+            t = ((v >> 1) & 0x3);
+            B = (t >> 1) | (t << 1) | (t << 7);
+            C = 26;
+            D = v >> 3;
+            break;
+         case 4:
+            t = ((v >> 1) & 0x7);
+            B = (t >> 1) | (t << 6);
+            C = 13;
+            D = v >> 4;
+            break;
+         case 5:
+            t = ((v >> 1) & 0xF);
+            B = (t >> 4) | (t << 5);
+            C = 6;
+            D = v >> 5;
+            break;
+         default:
+            unreachable("");
+         }
+         uint16_t T = D * C + B;
+         T = T ^ A;
+         T = (A & 0x80) | (T >> 2);
+         assert(T < 256);
+         colour_endpoints[i] = T;
+      } else {
+         switch (ce_bits) {
+         case 1: v = v ? 0xFF : 0x00; break;
+         case 2: v = (v << 6) | (v << 4) | (v << 2) | v; break;
+         case 3: v = (v << 5) | (v << 2) | (v >> 1); break;
+         case 4: v = (v << 4) | v; break;
+         case 5: v = (v << 3) | (v >> 2); break;
+         case 6: v = (v << 2) | (v >> 4); break;
+         case 7: v = (v << 1) | (v >> 6); break;
+         case 8: break;
+         default: unreachable("");
+         }
+         colour_endpoints[i] = v;
+      }
+   }
+}
+
+decode_error::type Block::decode(const Decoder &decoder, InputBitVector in)
+{
+   decode_error::type err;
+
+   is_error = false;
+   bogus_colour_endpoints = false;
+   bogus_weights = false;
+   is_void_extent = false;
+
+   wt_d = 1;
+   /* TODO: 3D */
+
+   /* TODO: test for all the illegal encodings */
+
+   if (VERBOSE_DECODE)
+      in.printf_bits(0, 128);
+
+   err = decode_block_mode(in);
+   if (err != decode_error::ok)
+      return err;
+
+   if (is_void_extent)
+      return decode_error::ok;
+
+   /* TODO: 3D */
+
+   calculate_from_weights();
+
+   if (VERBOSE_DECODE)
+      printf("weights_grid=%dx%dx%d dual_plane=%d num_weights=%d high_prec=%d r=%d range=0..%d (%dt %dq %db) weight_bits=%d\n",
+             wt_w, wt_h, wt_d, dual_plane, num_weights, high_prec, wt_range, wt_max, wt_trits, wt_quints, wt_bits, weight_bits);
+
+   if (wt_w > decoder.block_w || wt_h > decoder.block_h || wt_d > decoder.block_d)
+      return decode_error::weight_grid_exceeds_block_size;
+
+   num_parts = in.get_bits(11, 2) + 1;
+
+   if (VERBOSE_DECODE)
+      in.printf_bits(11, 2, "partitions = %d", num_parts);
+
+   if (dual_plane && num_parts > 3)
+      return decode_error::dual_plane_and_too_many_partitions;
+
+   decode_cem(in);
+
+   if (VERBOSE_DECODE)
+      printf("cem=[%d,%d,%d,%d] base_cem_class=%d\n", cems[0], cems[1], cems[2], cems[3], cem_base_class);
+
+   int num_cem_pairs = (cem_base_class + 1) * num_parts + extra_cem_bits;
+   num_cem_values = num_cem_pairs * 2;
+
+   calculate_remaining_bits();
+   err = calculate_colour_endpoints_size();
+   if (err != decode_error::ok)
+      return err;
+
+   if (VERBOSE_DECODE)
+      in.printf_bits(colour_endpoint_data_offset, colour_endpoint_bits,
+                     "endpoint data (%d bits, %d vals, %dt %dq %db)",
+                     colour_endpoint_bits, num_cem_values, ce_trits, ce_quints, ce_bits);
+
+   unpack_colour_endpoints(in);
+
+   if (VERBOSE_DECODE) {
+      printf("cem values raw =[");
+      for (int i = 0; i < num_cem_values; i++) {
+         if (i)
+            printf(", ");
+         printf("%3d", colour_endpoints_quant[i]);
+      }
+      printf("]\n");
+   }
+
+   if (num_cem_values > 18)
+      return decode_error::invalid_colour_endpoints_count;
+
+   unquantise_colour_endpoints();
+
+   if (VERBOSE_DECODE) {
+      printf("cem values norm=[");
+      for (int i = 0; i < num_cem_values; i++) {
+         if (i)
+            printf(", ");
+         printf("%3d", colour_endpoints[i]);
+      }
+      printf("]\n");
+   }
+
+   decode_colour_endpoints();
+
+   if (dual_plane) {
+      int ccs_offset = 128 - weight_bits - num_extra_cem_bits - 2;
+      colour_component_selector = in.get_bits(ccs_offset, 2);
+
+      if (VERBOSE_DECODE)
+         in.printf_bits(ccs_offset, 2, "colour component selector = %d", colour_component_selector);
+   } else {
+      colour_component_selector = 0;
+   }
+
+
+   if (VERBOSE_DECODE)
+      in.printf_bits(128 - weight_bits, weight_bits, "weights (%d bits)", weight_bits);
+
+   if (num_weights > 64)
+      return decode_error::invalid_num_weights;
+
+   if (weight_bits < 24 || weight_bits > 96)
+      return decode_error::invalid_weight_bits;
+
+   unpack_weights(in);
+
+   unquantise_weights();
+
+   if (VERBOSE_DECODE) {
+      printf("weights=[");
+      for (int i = 0; i < num_weights; ++i) {
+         if (i)
+            printf(", ");
+         printf("%d", weights[i]);
+      }
+      printf("]\n");
+
+      for (int plane = 0; plane <= dual_plane; ++plane) {
+         printf("weights (plane %d):\n", plane);
+         int i = 0;
+         (void)i;
+
+         for (int r = 0; r < wt_d; ++r) {
+            for (int t = 0; t < wt_h; ++t) {
+               for (int s = 0; s < wt_w; ++s) {
+                  printf("%3d", weights[i++ * (1 + dual_plane) + plane]);
+               }
+               printf("\n");
+            }
+            if (r < wt_d - 1)
+               printf("\n");
+         }
+      }
+   }
+
+   compute_infill_weights(decoder.block_w, decoder.block_h, decoder.block_d);
+
+   if (VERBOSE_DECODE) {
+      for (int plane = 0; plane <= dual_plane; ++plane) {
+         printf("infilled weights (plane %d):\n", plane);
+         int i = 0;
+         (void)i;
+
+         for (int r = 0; r < decoder.block_d; ++r) {
+            for (int t = 0; t < decoder.block_h; ++t) {
+               for (int s = 0; s < decoder.block_w; ++s) {
+                  printf("%3d", infill_weights[plane][i++]);
+               }
+               printf("\n");
+            }
+            if (r < decoder.block_d - 1)
+               printf("\n");
+         }
+      }
+   }
+   if (VERBOSE_DECODE)
+      printf("\n");
+
+   return decode_error::ok;
+}
+
+void Block::write_decoded(const Decoder &decoder, uint16_t *output)
+{
+   /* sRGB can only be stored as unorm8. */
+   assert(!decoder.srgb || decoder.output_unorm8);
+
+   if (is_void_extent) {
+      for (int idx = 0; idx < decoder.block_w*decoder.block_h*decoder.block_d; ++idx) {
+         if (decoder.output_unorm8) {
+            if (decoder.srgb) {
+               output[idx*4+0] = void_extent_colour_r >> 8;
+               output[idx*4+1] = void_extent_colour_g >> 8;
+               output[idx*4+2] = void_extent_colour_b >> 8;
+            } else {
+               output[idx*4+0] = uint16_div_64k_to_half_to_unorm8(void_extent_colour_r);
+               output[idx*4+1] = uint16_div_64k_to_half_to_unorm8(void_extent_colour_g);
+               output[idx*4+2] = uint16_div_64k_to_half_to_unorm8(void_extent_colour_b);
+            }
+            output[idx*4+3] = uint16_div_64k_to_half_to_unorm8(void_extent_colour_a);
+         } else {
+            /* Store the color as FP16. */
+            output[idx*4+0] = _mesa_uint16_div_64k_to_half(void_extent_colour_r);
+            output[idx*4+1] = _mesa_uint16_div_64k_to_half(void_extent_colour_g);
+            output[idx*4+2] = _mesa_uint16_div_64k_to_half(void_extent_colour_b);
+            output[idx*4+3] = _mesa_uint16_div_64k_to_half(void_extent_colour_a);
+         }
+      }
+      return;
+   }
+
+   int small_block = (decoder.block_w * decoder.block_h * decoder.block_d) < 31;
+
+   int idx = 0;
+   for (int z = 0; z < decoder.block_d; ++z) {
+      for (int y = 0; y < decoder.block_h; ++y) {
+         for (int x = 0; x < decoder.block_w; ++x) {
+
+            int partition;
+            if (num_parts > 1) {
+               partition = select_partition(partition_index, x, y, z, num_parts, small_block);
+               assert(partition < num_parts);
+            } else {
+               partition = 0;
+            }
+
+            /* TODO: HDR */
+
+            uint8x4_t e0 = endpoints_decoded[0][partition];
+            uint8x4_t e1 = endpoints_decoded[1][partition];
+            uint16_t c0[4], c1[4];
+
+            /* Expand to 16 bits. */
+            if (decoder.srgb) {
+               c0[0] = (uint16_t)((e0.v[0] << 8) | 0x80);
+               c0[1] = (uint16_t)((e0.v[1] << 8) | 0x80);
+               c0[2] = (uint16_t)((e0.v[2] << 8) | 0x80);
+               c0[3] = (uint16_t)((e0.v[3] << 8) | 0x80);
+
+               c1[0] = (uint16_t)((e1.v[0] << 8) | 0x80);
+               c1[1] = (uint16_t)((e1.v[1] << 8) | 0x80);
+               c1[2] = (uint16_t)((e1.v[2] << 8) | 0x80);
+               c1[3] = (uint16_t)((e1.v[3] << 8) | 0x80);
+            } else {
+               c0[0] = (uint16_t)((e0.v[0] << 8) | e0.v[0]);
+               c0[1] = (uint16_t)((e0.v[1] << 8) | e0.v[1]);
+               c0[2] = (uint16_t)((e0.v[2] << 8) | e0.v[2]);
+               c0[3] = (uint16_t)((e0.v[3] << 8) | e0.v[3]);
+
+               c1[0] = (uint16_t)((e1.v[0] << 8) | e1.v[0]);
+               c1[1] = (uint16_t)((e1.v[1] << 8) | e1.v[1]);
+               c1[2] = (uint16_t)((e1.v[2] << 8) | e1.v[2]);
+               c1[3] = (uint16_t)((e1.v[3] << 8) | e1.v[3]);
+            }
+
+            int w[4];
+            if (dual_plane) {
+               int w0 = infill_weights[0][idx];
+               int w1 = infill_weights[1][idx];
+               w[0] = w[1] = w[2] = w[3] = w0;
+               w[colour_component_selector] = w1;
+            } else {
+               int w0 = infill_weights[0][idx];
+               w[0] = w[1] = w[2] = w[3] = w0;
+            }
+
+            /* Interpolate to produce UNORM16, applying weights. */
+            uint16_t c[4] = {
+               (uint16_t)((c0[0] * (64 - w[0]) + c1[0] * w[0] + 32) >> 6),
+               (uint16_t)((c0[1] * (64 - w[1]) + c1[1] * w[1] + 32) >> 6),
+               (uint16_t)((c0[2] * (64 - w[2]) + c1[2] * w[2] + 32) >> 6),
+               (uint16_t)((c0[3] * (64 - w[3]) + c1[3] * w[3] + 32) >> 6),
+            };
+
+            if (decoder.output_unorm8) {
+               if (decoder.srgb) {
+                  output[idx*4+0] = c[0] >> 8;
+                  output[idx*4+1] = c[1] >> 8;
+                  output[idx*4+2] = c[2] >> 8;
+               } else {
+                  output[idx*4+0] = c[0] == 65535 ? 0xff : uint16_div_64k_to_half_to_unorm8(c[0]);
+                  output[idx*4+1] = c[1] == 65535 ? 0xff : uint16_div_64k_to_half_to_unorm8(c[1]);
+                  output[idx*4+2] = c[2] == 65535 ? 0xff : uint16_div_64k_to_half_to_unorm8(c[2]);
+               }
+               output[idx*4+3] = c[3] == 65535 ? 0xff : uint16_div_64k_to_half_to_unorm8(c[3]);
+            } else {
+               /* Store the color as FP16. */
+               output[idx*4+0] = c[0] == 65535 ? FP16_ONE : _mesa_uint16_div_64k_to_half(c[0]);
+               output[idx*4+1] = c[1] == 65535 ? FP16_ONE : _mesa_uint16_div_64k_to_half(c[1]);
+               output[idx*4+2] = c[2] == 65535 ? FP16_ONE : _mesa_uint16_div_64k_to_half(c[2]);
+               output[idx*4+3] = c[3] == 65535 ? FP16_ONE : _mesa_uint16_div_64k_to_half(c[3]);
+            }
+
+            idx++;
+         }
+      }
+   }
+}
+
+void Block::calculate_from_weights()
+{
+   wt_trits = 0;
+   wt_quints = 0;
+   wt_bits = 0;
+   switch (high_prec) {
+   case 0:
+      switch (wt_range) {
+      case 0x2: wt_max = 1; wt_bits = 1; break;
+      case 0x3: wt_max = 2; wt_trits = 1; break;
+      case 0x4: wt_max = 3; wt_bits = 2; break;
+      case 0x5: wt_max = 4; wt_quints = 1; break;
+      case 0x6: wt_max = 5; wt_trits = 1; wt_bits = 1; break;
+      case 0x7: wt_max = 7; wt_bits = 3; break;
+      default: abort();
+      }
+      break;
+   case 1:
+      switch (wt_range) {
+      case 0x2: wt_max = 9; wt_quints = 1; wt_bits = 1; break;
+      case 0x3: wt_max = 11; wt_trits = 1; wt_bits = 2; break;
+      case 0x4: wt_max = 15; wt_bits = 4; break;
+      case 0x5: wt_max = 19; wt_quints = 1; wt_bits = 2; break;
+      case 0x6: wt_max = 23; wt_trits = 1; wt_bits = 3; break;
+      case 0x7: wt_max = 31; wt_bits = 5; break;
+      default: abort();
+      }
+      break;
+   }
+
+   assert(wt_trits || wt_quints || wt_bits);
+
+   num_weights = wt_w * wt_h * wt_d;
+
+   if (dual_plane)
+      num_weights *= 2;
+
+   weight_bits =
+         (num_weights * 8 * wt_trits + 4) / 5
+         + (num_weights * 7 * wt_quints + 2) / 3
+         +  num_weights * wt_bits;
+}
+
+void Block::calculate_remaining_bits()
+{
+   int config_bits;
+   if (num_parts > 1) {
+      if (!is_multi_cem)
+         config_bits = 29;
+      else
+         config_bits = 25 + 3 * num_parts;
+   } else {
+      config_bits = 17;
+   }
+
+   if (dual_plane)
+      config_bits += 2;
+
+   remaining_bits = 128 - config_bits - weight_bits;
+}
+
+decode_error::type Block::calculate_colour_endpoints_size()
+{
+   /* Specified as illegal */
+   if (remaining_bits < (13 * num_cem_values + 4) / 5) {
+      colour_endpoint_bits = ce_max = ce_trits = ce_quints = ce_bits = 0;
+      return decode_error::invalid_colour_endpoints_size;
+   }
+
+   /* Find the largest cem_ranges that fits within remaining_bits */
+   for (int i = ARRAY_SIZE(cem_ranges)-1; i >= 0; --i) {
+      int cem_bits;
+      cem_bits = (num_cem_values * 8 * cem_ranges[i].t + 4) / 5
+                 + (num_cem_values * 7 * cem_ranges[i].q + 2) / 3
+                 +  num_cem_values * cem_ranges[i].b;
+
+      if (cem_bits <= remaining_bits)
+      {
+         colour_endpoint_bits = cem_bits;
+         ce_max = cem_ranges[i].max;
+         ce_trits = cem_ranges[i].t;
+         ce_quints = cem_ranges[i].q;
+         ce_bits = cem_ranges[i].b;
+         return decode_error::ok;
+      }
+   }
+
+   assert(0);
+   return decode_error::invalid_colour_endpoints_size;
+}
+
+/**
+ * Decode ASTC 2D LDR texture data.
+ *
+ * \param src_width in pixels
+ * \param src_height in pixels
+ * \param dst_stride in bytes
+ */
+extern "C" void
+_mesa_unpack_astc_2d_ldr(uint8_t *dst_row,
+                         unsigned dst_stride,
+                         const uint8_t *src_row,
+                         unsigned src_stride,
+                         unsigned src_width,
+                         unsigned src_height,
+                         mesa_format format)
+{
+   assert(_mesa_is_format_astc_2d(format));
+   bool srgb = _mesa_get_format_color_encoding(format) == GL_SRGB;
+
+   unsigned blk_w, blk_h;
+   _mesa_get_format_block_size(format, &blk_w, &blk_h);
+
+   const unsigned block_size = 16;
+   unsigned x_blocks = (src_width + blk_w - 1) / blk_w;
+   unsigned y_blocks = (src_height + blk_h - 1) / blk_h;
+
+   Decoder dec(blk_w, blk_h, 1, srgb, true);
+
+   for (unsigned y = 0; y < y_blocks; ++y) {
+      for (unsigned x = 0; x < x_blocks; ++x) {
+         /* Same size as the largest block. */
+         uint16_t block_out[12 * 12 * 4];
+
+         dec.decode(src_row + x * block_size, block_out);
+
+         /* This can be smaller with NPOT dimensions. */
+         unsigned dst_blk_w = MIN2(blk_w, src_width  - x*blk_w);
+         unsigned dst_blk_h = MIN2(blk_h, src_height - y*blk_h);
+
+         for (unsigned sub_y = 0; sub_y < dst_blk_h; ++sub_y) {
+            for (unsigned sub_x = 0; sub_x < dst_blk_w; ++sub_x) {
+               uint8_t *dst = dst_row + sub_y * dst_stride +
+                              (x * blk_w + sub_x) * 4;
+               const uint16_t *src = &block_out[(sub_y * blk_w + sub_x) * 4];
+
+               dst[0] = src[0];
+               dst[1] = src[1];
+               dst[2] = src[2];
+               dst[3] = src[3];
+            }
+         }
+      }
+      src_row += src_stride;
+      dst_row += dst_stride * blk_h;
+   }
+}
diff --git a/lib/mesa/src/mesa/main/texcompress_astc.h b/lib/mesa/src/mesa/main/texcompress_astc.h
new file mode 100644
index 000000000..9f9c5281d
--- /dev/null
+++ b/lib/mesa/src/mesa/main/texcompress_astc.h
@@ -0,0 +1,47 @@
+/*
+ * Copyright 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ */
+
+#ifndef TEXCOMPRESS_ASTC_H
+#define TEXCOMPRESS_ASTC_H
+
+#include <inttypes.h>
+#include "texcompress.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void
+_mesa_unpack_astc_2d_ldr(uint8_t *dst_row,
+                         unsigned dst_stride,
+                         const uint8_t *src_row,
+                         unsigned src_stride,
+                         unsigned src_width,
+                         unsigned src_height,
+                         mesa_format format);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/lib/mesa/src/mesa/main/texcompress_bptc.c b/lib/mesa/src/mesa/main/texcompress_bptc.c
index 26e591580..46279f144 100644
--- a/lib/mesa/src/mesa/main/texcompress_bptc.c
+++ b/lib/mesa/src/mesa/main/texcompress_bptc.c
@@ -29,631 +29,38 @@
 #include <stdbool.h>
 #include "texcompress.h"
 #include "texcompress_bptc.h"
-#include "util/format_srgb.h"
-#include "util/half_float.h"
+#include "texcompress_bptc_tmp.h"
 #include "texstore.h"
-#include "macros.h"
 #include "image.h"
+#include "mtypes.h"
 
-#define BLOCK_SIZE 4
-#define N_PARTITIONS 64
-#define BLOCK_BYTES 16
-
-struct bptc_unorm_mode {
-   int n_subsets;
-   int n_partition_bits;
-   bool has_rotation_bits;
-   bool has_index_selection_bit;
-   int n_color_bits;
-   int n_alpha_bits;
-   bool has_endpoint_pbits;
-   bool has_shared_pbits;
-   int n_index_bits;
-   int n_secondary_index_bits;
-};
-
-struct bptc_float_bitfield {
-   int8_t endpoint;
-   uint8_t component;
-   uint8_t offset;
-   uint8_t n_bits;
-   bool reverse;
-};
-
-struct bptc_float_mode {
-   bool reserved;
-   bool transformed_endpoints;
-   int n_partition_bits;
-   int n_endpoint_bits;
-   int n_index_bits;
-   int n_delta_bits[3];
-   struct bptc_float_bitfield bitfields[24];
-};
-
-struct bit_writer {
-   uint8_t buf;
-   int pos;
-   uint8_t *dst;
-};
-
-static const struct bptc_unorm_mode
-bptc_unorm_modes[] = {
-   /* 0 */ { 3, 4, false, false, 4, 0, true,  false, 3, 0 },
-   /* 1 */ { 2, 6, false, false, 6, 0, false, true,  3, 0 },
-   /* 2 */ { 3, 6, false, false, 5, 0, false, false, 2, 0 },
-   /* 3 */ { 2, 6, false, false, 7, 0, true,  false, 2, 0 },
-   /* 4 */ { 1, 0, true,  true,  5, 6, false, false, 2, 3 },
-   /* 5 */ { 1, 0, true,  false, 7, 8, false, false, 2, 2 },
-   /* 6 */ { 1, 0, false, false, 7, 7, true,  false, 4, 0 },
-   /* 7 */ { 2, 6, false, false, 5, 5, true,  false, 2, 0 }
-};
-
-static const struct bptc_float_mode
-bptc_float_modes[] = {
-   /* 00 */
-   { false, true, 5, 10, 3, { 5, 5, 5 },
-     { { 2, 1, 4, 1, false }, { 2, 2, 4, 1, false }, { 3, 2, 4, 1, false },
-       { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
-       { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false },
-       { 1, 1, 0, 5, false }, { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false },
-       { 1, 2, 0, 5, false }, { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false },
-       { 2, 0, 0, 5, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false },
-       { 3, 2, 3, 1, false },
-       { -1 } }
-   },
-   /* 01 */
-   { false, true, 5, 7, 3, { 6, 6, 6 },
-     { { 2, 1, 5, 1, false }, { 3, 1, 4, 1, false }, { 3, 1, 5, 1, false },
-       { 0, 0, 0, 7, false }, { 3, 2, 0, 1, false }, { 3, 2, 1, 1, false },
-       { 2, 2, 4, 1, false }, { 0, 1, 0, 7, false }, { 2, 2, 5, 1, false },
-       { 3, 2, 2, 1, false }, { 2, 1, 4, 1, false }, { 0, 2, 0, 7, false },
-       { 3, 2, 3, 1, false }, { 3, 2, 5, 1, false }, { 3, 2, 4, 1, false },
-       { 1, 0, 0, 6, false }, { 2, 1, 0, 4, false }, { 1, 1, 0, 6, false },
-       { 3, 1, 0, 4, false }, { 1, 2, 0, 6, false }, { 2, 2, 0, 4, false },
-       { 2, 0, 0, 6, false },
-       { 3, 0, 0, 6, false },
-       { -1 } }
-   },
-   /* 00010 */
-   { false, true, 5, 11, 3, { 5, 4, 4 },
-     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
-       { 1, 0, 0, 5, false }, { 0, 0, 10, 1, false }, { 2, 1, 0, 4, false },
-       { 1, 1, 0, 4, false }, { 0, 1, 10, 1, false }, { 3, 2, 0, 1, false },
-       { 3, 1, 0, 4, false }, { 1, 2, 0, 4, false }, { 0, 2, 10, 1, false },
-       { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 5, false },
-       { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false }, { 3, 2, 3, 1, false },
-       { -1 } }
-   },
-   /* 00011 */
-   { false, false, 0, 10, 4, { 10, 10, 10 },
-     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
-       { 1, 0, 0, 10, false }, { 1, 1, 0, 10, false }, { 1, 2, 0, 10, false },
-       { -1 } }
-   },
-   /* 00110 */
-   { false, true, 5, 11, 3, { 4, 5, 4 },
-     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
-       { 1, 0, 0, 4, false }, { 0, 0, 10, 1, false }, { 3, 1, 4, 1, false },
-       { 2, 1, 0, 4, false }, { 1, 1, 0, 5, false }, { 0, 1, 10, 1, false },
-       { 3, 1, 0, 4, false }, { 1, 2, 0, 4, false }, { 0, 2, 10, 1, false },
-       { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 4, false },
-       { 3, 2, 0, 1, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 4, false },
-       { 2, 1, 4, 1, false }, { 3, 2, 3, 1, false },
-       { -1 } }
-   },
-   /* 00111 */
-   { false, true, 0, 11, 4, { 9, 9, 9 },
-     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
-       { 1, 0, 0, 9, false }, { 0, 0, 10, 1, false }, { 1, 1, 0, 9, false },
-       { 0, 1, 10, 1, false }, { 1, 2, 0, 9, false }, { 0, 2, 10, 1, false },
-       { -1 } }
-   },
-   /* 01010 */
-   { false, true, 5, 11, 3, { 4, 4, 5 },
-     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
-       { 1, 0, 0, 4, false }, { 0, 0, 10, 1, false }, { 2, 2, 4, 1, false },
-       { 2, 1, 0, 4, false }, { 1, 1, 0, 4, false }, { 0, 1, 10, 1, false },
-       { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false }, { 1, 2, 0, 5, false },
-       { 0, 2, 10, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 4, false },
-       { 3, 2, 1, 1, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 4, false },
-       { 3, 2, 4, 1, false }, { 3, 2, 3, 1, false },
-       { -1 } }
-   },
-   /* 01011 */
-   { false, true, 0, 12, 4, { 8, 8, 8 },
-     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
-       { 1, 0, 0, 8, false }, { 0, 0, 10, 2, true }, { 1, 1, 0, 8, false },
-       { 0, 1, 10, 2, true }, { 1, 2, 0, 8, false }, { 0, 2, 10, 2, true },
-       { -1 } }
-   },
-   /* 01110 */
-   { false, true, 5, 9, 3, { 5, 5, 5 },
-     { { 0, 0, 0, 9, false }, { 2, 2, 4, 1, false }, { 0, 1, 0, 9, false },
-       { 2, 1, 4, 1, false }, { 0, 2, 0, 9, false }, { 3, 2, 4, 1, false },
-       { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false },
-       { 1, 1, 0, 5, false }, { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false },
-       { 1, 2, 0, 5, false }, { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false },
-       { 2, 0, 0, 5, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false },
-       { 3, 2, 3, 1, false },
-       { -1 } }
-   },
-   /* 01111 */
-   { false, true, 0, 16, 4, { 4, 4, 4 },
-     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
-       { 1, 0, 0, 4, false }, { 0, 0, 10, 6, true }, { 1, 1, 0, 4, false },
-       { 0, 1, 10, 6, true }, { 1, 2, 0, 4, false }, { 0, 2, 10, 6, true },
-       { -1 } }
-   },
-   /* 10010 */
-   { false, true, 5, 8, 3, { 6, 5, 5 },
-     { { 0, 0, 0, 8, false }, { 3, 1, 4, 1, false }, { 2, 2, 4, 1, false },
-       { 0, 1, 0, 8, false }, { 3, 2, 2, 1, false }, { 2, 1, 4, 1, false },
-       { 0, 2, 0, 8, false }, { 3, 2, 3, 1, false }, { 3, 2, 4, 1, false },
-       { 1, 0, 0, 6, false }, { 2, 1, 0, 4, false }, { 1, 1, 0, 5, false },
-       { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false }, { 1, 2, 0, 5, false },
-       { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 6, false },
-       { 3, 0, 0, 6, false },
-       { -1 } }
-   },
-   /* 10011 */
-   { true /* reserved */ },
-   /* 10110 */
-   { false, true, 5, 8, 3, { 5, 6, 5 },
-     { { 0, 0, 0, 8, false }, { 3, 2, 0, 1, false }, { 2, 2, 4, 1, false },
-       { 0, 1, 0, 8, false }, { 2, 1, 5, 1, false }, { 2, 1, 4, 1, false },
-       { 0, 2, 0, 8, false }, { 3, 1, 5, 1, false }, { 3, 2, 4, 1, false },
-       { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false },
-       { 1, 1, 0, 6, false }, { 3, 1, 0, 4, false }, { 1, 2, 0, 5, false },
-       { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 5, false },
-       { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false }, { 3, 2, 3, 1, false },
-       { -1 } }
-   },
-   /* 10111 */
-   { true /* reserved */ },
-   /* 11010 */
-   { false, true, 5, 8, 3, { 5, 5, 6 },
-     { { 0, 0, 0, 8, false }, { 3, 2, 1, 1, false }, { 2, 2, 4, 1, false },
-       { 0, 1, 0, 8, false }, { 2, 2, 5, 1, false }, { 2, 1, 4, 1, false },
-       { 0, 2, 0, 8, false }, { 3, 2, 5, 1, false }, { 3, 2, 4, 1, false },
-       { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false },
-       { 1, 1, 0, 5, false }, { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false },
-       { 1, 2, 0, 6, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 5, false },
-       { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false }, { 3, 2, 3, 1, false },
-       { -1 } }
-   },
-   /* 11011 */
-   { true /* reserved */ },
-   /* 11110 */
-   { false, false, 5, 6, 3, { 6, 6, 6 },
-     { { 0, 0, 0, 6, false }, { 3, 1, 4, 1, false }, { 3, 2, 0, 1, false },
-       { 3, 2, 1, 1, false }, { 2, 2, 4, 1, false }, { 0, 1, 0, 6, false },
-       { 2, 1, 5, 1, false }, { 2, 2, 5, 1, false }, { 3, 2, 2, 1, false },
-       { 2, 1, 4, 1, false }, { 0, 2, 0, 6, false }, { 3, 1, 5, 1, false },
-       { 3, 2, 3, 1, false }, { 3, 2, 5, 1, false }, { 3, 2, 4, 1, false },
-       { 1, 0, 0, 6, false }, { 2, 1, 0, 4, false }, { 1, 1, 0, 6, false },
-       { 3, 1, 0, 4, false }, { 1, 2, 0, 6, false }, { 2, 2, 0, 4, false },
-       { 2, 0, 0, 6, false }, { 3, 0, 0, 6, false },
-       { -1 } }
-   },
-   /* 11111 */
-   { true /* reserved */ },
-};
-
-/* This partition table is used when the mode has two subsets. Each
- * partition is represented by a 32-bit value which gives 2 bits per texel
- * within the block. The value of the two bits represents which subset to use
- * (0 or 1).
- */
-static const uint32_t
-partition_table1[N_PARTITIONS] = {
-   0x50505050U, 0x40404040U, 0x54545454U, 0x54505040U,
-   0x50404000U, 0x55545450U, 0x55545040U, 0x54504000U,
-   0x50400000U, 0x55555450U, 0x55544000U, 0x54400000U,
-   0x55555440U, 0x55550000U, 0x55555500U, 0x55000000U,
-   0x55150100U, 0x00004054U, 0x15010000U, 0x00405054U,
-   0x00004050U, 0x15050100U, 0x05010000U, 0x40505054U,
-   0x00404050U, 0x05010100U, 0x14141414U, 0x05141450U,
-   0x01155440U, 0x00555500U, 0x15014054U, 0x05414150U,
-   0x44444444U, 0x55005500U, 0x11441144U, 0x05055050U,
-   0x05500550U, 0x11114444U, 0x41144114U, 0x44111144U,
-   0x15055054U, 0x01055040U, 0x05041050U, 0x05455150U,
-   0x14414114U, 0x50050550U, 0x41411414U, 0x00141400U,
-   0x00041504U, 0x00105410U, 0x10541000U, 0x04150400U,
-   0x50410514U, 0x41051450U, 0x05415014U, 0x14054150U,
-   0x41050514U, 0x41505014U, 0x40011554U, 0x54150140U,
-   0x50505500U, 0x00555050U, 0x15151010U, 0x54540404U,
-};
-
-/* This partition table is used when the mode has three subsets. In this case
- * the values can be 0, 1 or 2.
- */
-static const uint32_t
-partition_table2[N_PARTITIONS] = {
-   0xaa685050U, 0x6a5a5040U, 0x5a5a4200U, 0x5450a0a8U,
-   0xa5a50000U, 0xa0a05050U, 0x5555a0a0U, 0x5a5a5050U,
-   0xaa550000U, 0xaa555500U, 0xaaaa5500U, 0x90909090U,
-   0x94949494U, 0xa4a4a4a4U, 0xa9a59450U, 0x2a0a4250U,
-   0xa5945040U, 0x0a425054U, 0xa5a5a500U, 0x55a0a0a0U,
-   0xa8a85454U, 0x6a6a4040U, 0xa4a45000U, 0x1a1a0500U,
-   0x0050a4a4U, 0xaaa59090U, 0x14696914U, 0x69691400U,
-   0xa08585a0U, 0xaa821414U, 0x50a4a450U, 0x6a5a0200U,
-   0xa9a58000U, 0x5090a0a8U, 0xa8a09050U, 0x24242424U,
-   0x00aa5500U, 0x24924924U, 0x24499224U, 0x50a50a50U,
-   0x500aa550U, 0xaaaa4444U, 0x66660000U, 0xa5a0a5a0U,
-   0x50a050a0U, 0x69286928U, 0x44aaaa44U, 0x66666600U,
-   0xaa444444U, 0x54a854a8U, 0x95809580U, 0x96969600U,
-   0xa85454a8U, 0x80959580U, 0xaa141414U, 0x96960000U,
-   0xaaaa1414U, 0xa05050a0U, 0xa0a5a5a0U, 0x96000000U,
-   0x40804080U, 0xa9a8a9a8U, 0xaaaaaa44U, 0x2a4a5254U
-};
-
-static const uint8_t
-anchor_indices[][N_PARTITIONS] = {
-   /* Anchor index values for the second subset of two-subset partitioning */
-   {
-      0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,
-      0xf,0x2,0x8,0x2,0x2,0x8,0x8,0xf,0x2,0x8,0x2,0x2,0x8,0x8,0x2,0x2,
-      0xf,0xf,0x6,0x8,0x2,0x8,0xf,0xf,0x2,0x8,0x2,0x2,0x2,0xf,0xf,0x6,
-      0x6,0x2,0x6,0x8,0xf,0xf,0x2,0x2,0xf,0xf,0xf,0xf,0xf,0x2,0x2,0xf
-   },
-
-   /* Anchor index values for the second subset of three-subset partitioning */
-   {
-      0x3,0x3,0xf,0xf,0x8,0x3,0xf,0xf,0x8,0x8,0x6,0x6,0x6,0x5,0x3,0x3,
-      0x3,0x3,0x8,0xf,0x3,0x3,0x6,0xa,0x5,0x8,0x8,0x6,0x8,0x5,0xf,0xf,
-      0x8,0xf,0x3,0x5,0x6,0xa,0x8,0xf,0xf,0x3,0xf,0x5,0xf,0xf,0xf,0xf,
-      0x3,0xf,0x5,0x5,0x5,0x8,0x5,0xa,0x5,0xa,0x8,0xd,0xf,0xc,0x3,0x3
-   },
-
-   /* Anchor index values for the third subset of three-subset
-    * partitioning
-    */
-   {
-      0xf,0x8,0x8,0x3,0xf,0xf,0x3,0x8,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0x8,
-      0xf,0x8,0xf,0x3,0xf,0x8,0xf,0x8,0x3,0xf,0x6,0xa,0xf,0xf,0xa,0x8,
-      0xf,0x3,0xf,0xa,0xa,0x8,0x9,0xa,0x6,0xf,0x8,0xf,0x3,0x6,0x6,0x8,
-      0xf,0x3,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0x3,0xf,0xf,0x8
-   }
-};
-
-static int
-extract_bits(const uint8_t *block,
-             int offset,
-             int n_bits)
-{
-   int byte_index = offset / 8;
-   int bit_index = offset % 8;
-   int n_bits_in_byte = MIN2(n_bits, 8 - bit_index);
-   int result = 0;
-   int bit = 0;
-
-   while (true) {
-      result |= ((block[byte_index] >> bit_index) &
-                 ((1 << n_bits_in_byte) - 1)) << bit;
-
-      n_bits -= n_bits_in_byte;
-
-      if (n_bits <= 0)
-         return result;
-
-      bit += n_bits_in_byte;
-      byte_index++;
-      bit_index = 0;
-      n_bits_in_byte = MIN2(n_bits, 8);
-   }
-}
-
-static uint8_t
-expand_component(uint8_t byte,
-                 int n_bits)
-{
-   /* Expands a n-bit quantity into a byte by copying the most-significant
-    * bits into the unused least-significant bits.
-    */
-   return byte << (8 - n_bits) | (byte >> (2 * n_bits - 8));
-}
-
-static int
-extract_unorm_endpoints(const struct bptc_unorm_mode *mode,
-                        const uint8_t *block,
-                        int bit_offset,
-                        uint8_t endpoints[][4])
-{
-   int component;
-   int subset;
-   int endpoint;
-   int pbit;
-   int n_components;
-
-   /* Extract each color component */
-   for (component = 0; component < 3; component++) {
-      for (subset = 0; subset < mode->n_subsets; subset++) {
-         for (endpoint = 0; endpoint < 2; endpoint++) {
-            endpoints[subset * 2 + endpoint][component] =
-               extract_bits(block, bit_offset, mode->n_color_bits);
-            bit_offset += mode->n_color_bits;
-         }
-      }
-   }
-
-   /* Extract the alpha values */
-   if (mode->n_alpha_bits > 0) {
-      for (subset = 0; subset < mode->n_subsets; subset++) {
-         for (endpoint = 0; endpoint < 2; endpoint++) {
-            endpoints[subset * 2 + endpoint][3] =
-               extract_bits(block, bit_offset, mode->n_alpha_bits);
-            bit_offset += mode->n_alpha_bits;
-         }
-      }
-
-      n_components = 4;
-   } else {
-      for (subset = 0; subset < mode->n_subsets; subset++)
-         for (endpoint = 0; endpoint < 2; endpoint++)
-            endpoints[subset * 2 + endpoint][3] = 255;
-
-      n_components = 3;
-   }
-
-   /* Add in the p-bits */
-   if (mode->has_endpoint_pbits) {
-      for (subset = 0; subset < mode->n_subsets; subset++) {
-         for (endpoint = 0; endpoint < 2; endpoint++) {
-            pbit = extract_bits(block, bit_offset, 1);
-            bit_offset += 1;
-
-            for (component = 0; component < n_components; component++) {
-               endpoints[subset * 2 + endpoint][component] <<= 1;
-               endpoints[subset * 2 + endpoint][component] |= pbit;
-            }
-         }
-      }
-   } else if (mode->has_shared_pbits) {
-      for (subset = 0; subset < mode->n_subsets; subset++) {
-         pbit = extract_bits(block, bit_offset, 1);
-         bit_offset += 1;
-
-         for (endpoint = 0; endpoint < 2; endpoint++) {
-            for (component = 0; component < n_components; component++) {
-               endpoints[subset * 2 + endpoint][component] <<= 1;
-               endpoints[subset * 2 + endpoint][component] |= pbit;
-            }
-         }
-      }
-   }
-
-   /* Expand the n-bit values to a byte */
-   for (subset = 0; subset < mode->n_subsets; subset++) {
-      for (endpoint = 0; endpoint < 2; endpoint++) {
-         for (component = 0; component < 3; component++) {
-            endpoints[subset * 2 + endpoint][component] =
-               expand_component(endpoints[subset * 2 + endpoint][component],
-                                mode->n_color_bits +
-                                mode->has_endpoint_pbits +
-                                mode->has_shared_pbits);
-         }
-
-         if (mode->n_alpha_bits > 0) {
-            endpoints[subset * 2 + endpoint][3] =
-               expand_component(endpoints[subset * 2 + endpoint][3],
-                                mode->n_alpha_bits +
-                                mode->has_endpoint_pbits +
-                                mode->has_shared_pbits);
-         }
-      }
-   }
-
-   return bit_offset;
-}
-
-static bool
-is_anchor(int n_subsets,
-          int partition_num,
-          int texel)
-{
-   if (texel == 0)
-      return true;
-
-   switch (n_subsets) {
-   case 1:
-      return false;
-   case 2:
-      return anchor_indices[0][partition_num] == texel;
-   case 3:
-      return (anchor_indices[1][partition_num] == texel ||
-              anchor_indices[2][partition_num] == texel);
-   default:
-      assert(false);
-      return false;
-   }
-}
-
-static int
-count_anchors_before_texel(int n_subsets,
-                           int partition_num,
-                           int texel)
-{
-   int count = 1;
-
-   if (texel == 0)
-      return 0;
-
-   switch (n_subsets) {
-   case 1:
-      break;
-   case 2:
-      if (texel > anchor_indices[0][partition_num])
-         count++;
-      break;
-   case 3:
-      if (texel > anchor_indices[1][partition_num])
-         count++;
-      if (texel > anchor_indices[2][partition_num])
-         count++;
-      break;
-   default:
-      assert(false);
-      return 0;
-   }
-
-   return count;
-}
-
-static int32_t
-interpolate(int32_t a, int32_t b,
-            int index,
-            int index_bits)
+static void
+fetch_bptc_rgb_float(const GLubyte *map,
+                     GLint rowStride, GLint i, GLint j,
+                     GLfloat *texel,
+                     bool is_signed)
 {
-   static const uint8_t weights2[] = { 0, 21, 43, 64 };
-   static const uint8_t weights3[] = { 0, 9, 18, 27, 37, 46, 55, 64 };
-   static const uint8_t weights4[] =
-      { 0, 4, 9, 13, 17, 21, 26, 30, 34, 38, 43, 47, 51, 55, 60, 64 };
-   static const uint8_t *weights[] = {
-      NULL, NULL, weights2, weights3, weights4
-   };
-   int weight;
+   const GLubyte *block;
 
-   weight = weights[index_bits][index];
+   block = map + (((rowStride + 3) / 4) * (j / 4) + (i / 4)) * 16;
 
-   return ((64 - weight) * a + weight * b + 32) >> 6;
+   fetch_rgb_float_from_block(block, texel, (i % 4) + (j % 4) * 4, is_signed);
 }
 
 static void
-apply_rotation(int rotation,
-               uint8_t *result)
+fetch_bptc_rgb_signed_float(const GLubyte *map,
+                            GLint rowStride, GLint i, GLint j,
+                            GLfloat *texel)
 {
-   uint8_t t;
-
-   if (rotation == 0)
-      return;
-
-   rotation--;
-
-   t = result[rotation];
-   result[rotation] = result[3];
-   result[3] = t;
+   fetch_bptc_rgb_float(map, rowStride, i, j, texel, true);
 }
 
 static void
-fetch_rgba_unorm_from_block(const uint8_t *block,
-                            uint8_t *result,
-                            int texel)
+fetch_bptc_rgb_unsigned_float(const GLubyte *map,
+                              GLint rowStride, GLint i, GLint j,
+                              GLfloat *texel)
 {
-   int mode_num = ffs(block[0]);
-   const struct bptc_unorm_mode *mode;
-   int bit_offset, secondary_bit_offset;
-   int partition_num;
-   int subset_num;
-   int rotation;
-   int index_selection;
-   int index_bits;
-   int indices[2];
-   int index;
-   int anchors_before_texel;
-   bool anchor;
-   uint8_t endpoints[3 * 2][4];
-   uint32_t subsets;
-   int component;
-
-   if (mode_num == 0) {
-      /* According to the spec this mode is reserved and shouldn't be used. */
-      memset(result, 0, 3);
-      result[3] = 0xff;
-      return;
-   }
-
-   mode = bptc_unorm_modes + mode_num - 1;
-   bit_offset = mode_num;
-
-   partition_num = extract_bits(block, bit_offset, mode->n_partition_bits);
-   bit_offset += mode->n_partition_bits;
-
-   switch (mode->n_subsets) {
-   case 1:
-      subsets = 0;
-      break;
-   case 2:
-      subsets = partition_table1[partition_num];
-      break;
-   case 3:
-      subsets = partition_table2[partition_num];
-      break;
-   default:
-      assert(false);
-      return;
-   }
-
-   if (mode->has_rotation_bits) {
-      rotation = extract_bits(block, bit_offset, 2);
-      bit_offset += 2;
-   } else {
-      rotation = 0;
-   }
-
-   if (mode->has_index_selection_bit) {
-      index_selection = extract_bits(block, bit_offset, 1);
-      bit_offset++;
-   } else {
-      index_selection = 0;
-   }
-
-   bit_offset = extract_unorm_endpoints(mode, block, bit_offset, endpoints);
-
-   anchors_before_texel = count_anchors_before_texel(mode->n_subsets,
-                                                     partition_num, texel);
-
-   /* Calculate the offset to the secondary index */
-   secondary_bit_offset = (bit_offset +
-                           BLOCK_SIZE * BLOCK_SIZE * mode->n_index_bits -
-                           mode->n_subsets +
-                           mode->n_secondary_index_bits * texel -
-                           anchors_before_texel);
-
-   /* Calculate the offset to the primary index for this texel */
-   bit_offset += mode->n_index_bits * texel - anchors_before_texel;
-
-   subset_num = (subsets >> (texel * 2)) & 3;
-
-   anchor = is_anchor(mode->n_subsets, partition_num, texel);
-
-   index_bits = mode->n_index_bits;
-   if (anchor)
-      index_bits--;
-   indices[0] = extract_bits(block, bit_offset, index_bits);
-
-   if (mode->n_secondary_index_bits) {
-      index_bits = mode->n_secondary_index_bits;
-      if (anchor)
-         index_bits--;
-      indices[1] = extract_bits(block, secondary_bit_offset, index_bits);
-   }
-
-   index = indices[index_selection];
-   index_bits = (index_selection ?
-                 mode->n_secondary_index_bits :
-                 mode->n_index_bits);
-
-   for (component = 0; component < 3; component++)
-      result[component] = interpolate(endpoints[subset_num * 2][component],
-                                      endpoints[subset_num * 2 + 1][component],
-                                      index,
-                                      index_bits);
-
-   /* Alpha uses the opposite index from the color components */
-   if (mode->n_secondary_index_bits && !index_selection) {
-      index = indices[1];
-      index_bits = mode->n_secondary_index_bits;
-   } else {
-      index = indices[0];
-      index_bits = mode->n_index_bits;
-   }
-
-   result[3] = interpolate(endpoints[subset_num * 2][3],
-                           endpoints[subset_num * 2 + 1][3],
-                           index,
-                           index_bits);
-
-   apply_rotation(rotation, result);
+   fetch_bptc_rgb_float(map, rowStride, i, j, texel, false);
 }
 
 static void
@@ -698,257 +105,6 @@ fetch_bptc_srgb_alpha_unorm(const GLubyte *map,
    texel[ACOMP] = UBYTE_TO_FLOAT(texel_bytes[3]);
 }
 
-static int32_t
-sign_extend(int32_t value,
-            int n_bits)
-{
-   if ((value & (1 << (n_bits - 1)))) {
-      value |= (~(int32_t) 0) << n_bits;
-   }
-
-   return value;
-}
-
-static int
-signed_unquantize(int value, int n_endpoint_bits)
-{
-   bool sign;
-
-   if (n_endpoint_bits >= 16)
-      return value;
-
-   if (value == 0)
-      return 0;
-
-   sign = false;
-
-   if (value < 0) {
-      sign = true;
-      value = -value;
-   }
-
-   if (value >= (1 << (n_endpoint_bits - 1)) - 1)
-      value = 0x7fff;
-   else
-      value = ((value << 15) + 0x4000) >> (n_endpoint_bits - 1);
-
-   if (sign)
-      value = -value;
-
-   return value;
-}
-
-static int
-unsigned_unquantize(int value, int n_endpoint_bits)
-{
-   if (n_endpoint_bits >= 15)
-      return value;
-
-   if (value == 0)
-      return 0;
-
-   if (value == (1 << n_endpoint_bits) - 1)
-      return 0xffff;
-
-   return ((value << 15) + 0x4000) >> (n_endpoint_bits - 1);
-}
-
-static int
-extract_float_endpoints(const struct bptc_float_mode *mode,
-                        const uint8_t *block,
-                        int bit_offset,
-                        int32_t endpoints[][3],
-                        bool is_signed)
-{
-   const struct bptc_float_bitfield *bitfield;
-   int endpoint, component;
-   int n_endpoints;
-   int value;
-   int i;
-
-   if (mode->n_partition_bits)
-      n_endpoints = 4;
-   else
-      n_endpoints = 2;
-
-   memset(endpoints, 0, sizeof endpoints[0][0] * n_endpoints * 3);
-
-   for (bitfield = mode->bitfields; bitfield->endpoint != -1; bitfield++) {
-      value = extract_bits(block, bit_offset, bitfield->n_bits);
-      bit_offset += bitfield->n_bits;
-
-      if (bitfield->reverse) {
-         for (i = 0; i < bitfield->n_bits; i++) {
-            if (value & (1 << i))
-               endpoints[bitfield->endpoint][bitfield->component] |=
-                  1 << ((bitfield->n_bits - 1 - i) + bitfield->offset);
-         }
-      } else {
-         endpoints[bitfield->endpoint][bitfield->component] |=
-            value << bitfield->offset;
-      }
-   }
-
-   if (mode->transformed_endpoints) {
-      /* The endpoints are specified as signed offsets from e0 */
-      for (endpoint = 1; endpoint < n_endpoints; endpoint++) {
-         for (component = 0; component < 3; component++) {
-            value = sign_extend(endpoints[endpoint][component],
-                                mode->n_delta_bits[component]);
-            endpoints[endpoint][component] =
-               ((endpoints[0][component] + value) &
-                ((1 << mode->n_endpoint_bits) - 1));
-         }
-      }
-   }
-
-   if (is_signed) {
-      for (endpoint = 0; endpoint < n_endpoints; endpoint++) {
-         for (component = 0; component < 3; component++) {
-            value = sign_extend(endpoints[endpoint][component],
-                                mode->n_endpoint_bits);
-            endpoints[endpoint][component] =
-               signed_unquantize(value, mode->n_endpoint_bits);
-         }
-      }
-   } else {
-      for (endpoint = 0; endpoint < n_endpoints; endpoint++) {
-         for (component = 0; component < 3; component++) {
-            endpoints[endpoint][component] =
-               unsigned_unquantize(endpoints[endpoint][component],
-                                   mode->n_endpoint_bits);
-         }
-      }
-   }
-
-   return bit_offset;
-}
-
-static int32_t
-finish_unsigned_unquantize(int32_t value)
-{
-   return value * 31 / 64;
-}
-
-static int32_t
-finish_signed_unquantize(int32_t value)
-{
-   if (value < 0)
-      return (-value * 31 / 32) | 0x8000;
-   else
-      return value * 31 / 32;
-}
-
-static void
-fetch_rgb_float_from_block(const uint8_t *block,
-                           float *result,
-                           int texel,
-                           bool is_signed)
-{
-   int mode_num;
-   const struct bptc_float_mode *mode;
-   int bit_offset;
-   int partition_num;
-   int subset_num;
-   int index_bits;
-   int index;
-   int anchors_before_texel;
-   int32_t endpoints[2 * 2][3];
-   uint32_t subsets;
-   int n_subsets;
-   int component;
-   int32_t value;
-
-   if (block[0] & 0x2) {
-      mode_num = (((block[0] >> 1) & 0xe) | (block[0] & 1)) + 2;
-      bit_offset = 5;
-   } else {
-      mode_num = block[0] & 3;
-      bit_offset = 2;
-   }
-
-   mode = bptc_float_modes + mode_num;
-
-   if (mode->reserved) {
-      memset(result, 0, sizeof result[0] * 3);
-      result[3] = 1.0f;
-      return;
-   }
-
-   bit_offset = extract_float_endpoints(mode, block, bit_offset,
-                                        endpoints, is_signed);
-
-   if (mode->n_partition_bits) {
-      partition_num = extract_bits(block, bit_offset, mode->n_partition_bits);
-      bit_offset += mode->n_partition_bits;
-
-      subsets = partition_table1[partition_num];
-      n_subsets = 2;
-   } else {
-      partition_num = 0;
-      subsets = 0;
-      n_subsets = 1;
-   }
-
-   anchors_before_texel =
-      count_anchors_before_texel(n_subsets, partition_num, texel);
-
-   /* Calculate the offset to the primary index for this texel */
-   bit_offset += mode->n_index_bits * texel - anchors_before_texel;
-
-   subset_num = (subsets >> (texel * 2)) & 3;
-
-   index_bits = mode->n_index_bits;
-   if (is_anchor(n_subsets, partition_num, texel))
-      index_bits--;
-   index = extract_bits(block, bit_offset, index_bits);
-
-   for (component = 0; component < 3; component++) {
-      value = interpolate(endpoints[subset_num * 2][component],
-                          endpoints[subset_num * 2 + 1][component],
-                          index,
-                          mode->n_index_bits);
-
-      if (is_signed)
-         value = finish_signed_unquantize(value);
-      else
-         value = finish_unsigned_unquantize(value);
-
-      result[component] = _mesa_half_to_float(value);
-   }
-
-   result[3] = 1.0f;
-}
-
-static void
-fetch_bptc_rgb_float(const GLubyte *map,
-                     GLint rowStride, GLint i, GLint j,
-                     GLfloat *texel,
-                     bool is_signed)
-{
-   const GLubyte *block;
-
-   block = map + (((rowStride + 3) / 4) * (j / 4) + (i / 4)) * 16;
-
-   fetch_rgb_float_from_block(block, texel, (i % 4) + (j % 4) * 4, is_signed);
-}
-
-static void
-fetch_bptc_rgb_signed_float(const GLubyte *map,
-                            GLint rowStride, GLint i, GLint j,
-                            GLfloat *texel)
-{
-   fetch_bptc_rgb_float(map, rowStride, i, j, texel, true);
-}
-
-static void
-fetch_bptc_rgb_unsigned_float(const GLubyte *map,
-                              GLint rowStride, GLint i, GLint j,
-                              GLfloat *texel)
-{
-   fetch_bptc_rgb_float(map, rowStride, i, j, texel, false);
-}
-
 compressed_fetch_func
 _mesa_get_bptc_fetch_func(mesa_format format)
 {
@@ -966,312 +122,6 @@ _mesa_get_bptc_fetch_func(mesa_format format)
    }
 }
 
-static void
-write_bits(struct bit_writer *writer, int n_bits, int value)
-{
-   do {
-      if (n_bits + writer->pos >= 8) {
-         *(writer->dst++) = writer->buf | (value << writer->pos);
-         writer->buf = 0;
-         value >>= (8 - writer->pos);
-         n_bits -= (8 - writer->pos);
-         writer->pos = 0;
-      } else {
-         writer->buf |= value << writer->pos;
-         writer->pos += n_bits;
-         break;
-      }
-   } while (n_bits > 0);
-}
-
-static void
-get_average_luminance_alpha_unorm(int width, int height,
-                                  const uint8_t *src, int src_rowstride,
-                                  int *average_luminance, int *average_alpha)
-{
-   int luminance_sum = 0, alpha_sum = 0;
-   int y, x;
-
-   for (y = 0; y < height; y++) {
-      for (x = 0; x < width; x++) {
-         luminance_sum += src[0] + src[1] + src[2];
-         alpha_sum += src[3];
-         src += 4;
-      }
-      src += src_rowstride - width * 4;
-   }
-
-   *average_luminance = luminance_sum / (width * height);
-   *average_alpha = alpha_sum / (width * height);
-}
-
-static void
-get_rgba_endpoints_unorm(int width, int height,
-                         const uint8_t *src, int src_rowstride,
-                         int average_luminance, int average_alpha,
-                         uint8_t endpoints[][4])
-{
-   int endpoint_luminances[2];
-   int midpoint;
-   int sums[2][4];
-   int endpoint;
-   int luminance;
-   uint8_t temp[3];
-   const uint8_t *p = src;
-   int rgb_left_endpoint_count = 0;
-   int alpha_left_endpoint_count = 0;
-   int y, x, i;
-
-   memset(sums, 0, sizeof sums);
-
-   for (y = 0; y < height; y++) {
-      for (x = 0; x < width; x++) {
-         luminance = p[0] + p[1] + p[2];
-         if (luminance < average_luminance) {
-            endpoint = 0;
-            rgb_left_endpoint_count++;
-         } else {
-            endpoint = 1;
-         }
-         for (i = 0; i < 3; i++)
-            sums[endpoint][i] += p[i];
-
-         if (p[2] < average_alpha) {
-            endpoint = 0;
-            alpha_left_endpoint_count++;
-         } else {
-            endpoint = 1;
-         }
-         sums[endpoint][3] += p[3];
-
-         p += 4;
-      }
-
-      p += src_rowstride - width * 4;
-   }
-
-   if (rgb_left_endpoint_count == 0 ||
-       rgb_left_endpoint_count == width * height) {
-      for (i = 0; i < 3; i++)
-         endpoints[0][i] = endpoints[1][i] =
-            (sums[0][i] + sums[1][i]) / (width * height);
-   } else {
-      for (i = 0; i < 3; i++) {
-         endpoints[0][i] = sums[0][i] / rgb_left_endpoint_count;
-         endpoints[1][i] = (sums[1][i] /
-                            (width * height - rgb_left_endpoint_count));
-      }
-   }
-
-   if (alpha_left_endpoint_count == 0 ||
-       alpha_left_endpoint_count == width * height) {
-      endpoints[0][3] = endpoints[1][3] =
-         (sums[0][3] + sums[1][3]) / (width * height);
-   } else {
-         endpoints[0][3] = sums[0][3] / alpha_left_endpoint_count;
-         endpoints[1][3] = (sums[1][3] /
-                            (width * height - alpha_left_endpoint_count));
-   }
-
-   /* We may need to swap the endpoints to ensure the most-significant bit of
-    * the first index is zero */
-
-   for (endpoint = 0; endpoint < 2; endpoint++) {
-      endpoint_luminances[endpoint] =
-         endpoints[endpoint][0] +
-         endpoints[endpoint][1] +
-         endpoints[endpoint][2];
-   }
-   midpoint = (endpoint_luminances[0] + endpoint_luminances[1]) / 2;
-
-   if ((src[0] + src[1] + src[2] <= midpoint) !=
-       (endpoint_luminances[0] <= midpoint)) {
-      memcpy(temp, endpoints[0], 3);
-      memcpy(endpoints[0], endpoints[1], 3);
-      memcpy(endpoints[1], temp, 3);
-   }
-
-   /* Same for the alpha endpoints */
-
-   midpoint = (endpoints[0][3] + endpoints[1][3]) / 2;
-
-   if ((src[3] <= midpoint) != (endpoints[0][3] <= midpoint)) {
-      temp[0] = endpoints[0][3];
-      endpoints[0][3] = endpoints[1][3];
-      endpoints[1][3] = temp[0];
-   }
-}
-
-static void
-write_rgb_indices_unorm(struct bit_writer *writer,
-                        int src_width, int src_height,
-                        const uint8_t *src, int src_rowstride,
-                        uint8_t endpoints[][4])
-{
-   int luminance;
-   int endpoint_luminances[2];
-   int endpoint;
-   int index;
-   int y, x;
-
-   for (endpoint = 0; endpoint < 2; endpoint++) {
-      endpoint_luminances[endpoint] =
-         endpoints[endpoint][0] +
-         endpoints[endpoint][1] +
-         endpoints[endpoint][2];
-   }
-
-   /* If the endpoints have the same luminance then we'll just use index 0 for
-    * all of the texels */
-   if (endpoint_luminances[0] == endpoint_luminances[1]) {
-      write_bits(writer, BLOCK_SIZE * BLOCK_SIZE * 2 - 1, 0);
-      return;
-   }
-
-   for (y = 0; y < src_height; y++) {
-      for (x = 0; x < src_width; x++) {
-         luminance = src[0] + src[1] + src[2];
-
-         index = ((luminance - endpoint_luminances[0]) * 3 /
-                  (endpoint_luminances[1] - endpoint_luminances[0]));
-         if (index < 0)
-            index = 0;
-         else if (index > 3)
-            index = 3;
-
-         assert(x != 0 || y != 0 || index < 2);
-
-         write_bits(writer, (x == 0 && y == 0) ? 1 : 2, index);
-
-         src += 4;
-      }
-
-      /* Pad the indices out to the block size */
-      if (src_width < BLOCK_SIZE)
-         write_bits(writer, 2 * (BLOCK_SIZE - src_width), 0);
-
-      src += src_rowstride - src_width * 4;
-   }
-
-   /* Pad the indices out to the block size */
-   if (src_height < BLOCK_SIZE)
-      write_bits(writer, 2 * BLOCK_SIZE * (BLOCK_SIZE - src_height), 0);
-}
-
-static void
-write_alpha_indices_unorm(struct bit_writer *writer,
-                          int src_width, int src_height,
-                          const uint8_t *src, int src_rowstride,
-                          uint8_t endpoints[][4])
-{
-   int index;
-   int y, x;
-
-   /* If the endpoints have the same alpha then we'll just use index 0 for
-    * all of the texels */
-   if (endpoints[0][3] == endpoints[1][3]) {
-      write_bits(writer, BLOCK_SIZE * BLOCK_SIZE * 3 - 1, 0);
-      return;
-   }
-
-   for (y = 0; y < src_height; y++) {
-      for (x = 0; x < src_width; x++) {
-         index = (((int) src[3] - (int) endpoints[0][3]) * 7 /
-                  ((int) endpoints[1][3] - endpoints[0][3]));
-         if (index < 0)
-            index = 0;
-         else if (index > 7)
-            index = 7;
-
-         assert(x != 0 || y != 0 || index < 4);
-
-         /* The first index has one less bit */
-         write_bits(writer, (x == 0 && y == 0) ? 2 : 3, index);
-
-         src += 4;
-      }
-
-      /* Pad the indices out to the block size */
-      if (src_width < BLOCK_SIZE)
-         write_bits(writer, 3 * (BLOCK_SIZE - src_width), 0);
-
-      src += src_rowstride - src_width * 4;
-   }
-
-   /* Pad the indices out to the block size */
-   if (src_height < BLOCK_SIZE)
-      write_bits(writer, 3 * BLOCK_SIZE * (BLOCK_SIZE - src_height), 0);
-}
-
-static void
-compress_rgba_unorm_block(int src_width, int src_height,
-                          const uint8_t *src, int src_rowstride,
-                          uint8_t *dst)
-{
-   int average_luminance, average_alpha;
-   uint8_t endpoints[2][4];
-   struct bit_writer writer;
-   int component, endpoint;
-
-   get_average_luminance_alpha_unorm(src_width, src_height, src, src_rowstride,
-                                     &average_luminance, &average_alpha);
-   get_rgba_endpoints_unorm(src_width, src_height, src, src_rowstride,
-                            average_luminance, average_alpha,
-                            endpoints);
-
-   writer.dst = dst;
-   writer.pos = 0;
-   writer.buf = 0;
-
-   write_bits(&writer, 5, 0x10); /* mode 4 */
-   write_bits(&writer, 2, 0); /* rotation 0 */
-   write_bits(&writer, 1, 0); /* index selection bit */
-
-   /* Write the color endpoints */
-   for (component = 0; component < 3; component++)
-      for (endpoint = 0; endpoint < 2; endpoint++)
-         write_bits(&writer, 5, endpoints[endpoint][component] >> 3);
-
-   /* Write the alpha endpoints */
-   for (endpoint = 0; endpoint < 2; endpoint++)
-      write_bits(&writer, 6, endpoints[endpoint][3] >> 2);
-
-   write_rgb_indices_unorm(&writer,
-                           src_width, src_height,
-                           src, src_rowstride,
-                           endpoints);
-   write_alpha_indices_unorm(&writer,
-                             src_width, src_height,
-                             src, src_rowstride,
-                             endpoints);
-}
-
-static void
-compress_rgba_unorm(int width, int height,
-                    const uint8_t *src, int src_rowstride,
-                    uint8_t *dst, int dst_rowstride)
-{
-   int dst_row_diff;
-   int y, x;
-
-   if (dst_rowstride >= width * 4)
-      dst_row_diff = dst_rowstride - ((width + 3) & ~3) * 4;
-   else
-      dst_row_diff = 0;
-
-   for (y = 0; y < height; y += BLOCK_SIZE) {
-      for (x = 0; x < width; x += BLOCK_SIZE) {
-         compress_rgba_unorm_block(MIN2(width - x, BLOCK_SIZE),
-                                   MIN2(height - y, BLOCK_SIZE),
-                                   src + x * 4 + y * src_rowstride,
-                                   src_rowstride,
-                                   dst);
-         dst += BLOCK_BYTES;
-      }
-      dst += dst_row_diff;
-   }
-}
-
 GLboolean
 _mesa_texstore_bptc_rgba_unorm(TEXSTORE_PARAMS)
 {
@@ -1317,272 +167,6 @@ _mesa_texstore_bptc_rgba_unorm(TEXSTORE_PARAMS)
    return GL_TRUE;
 }
 
-static float
-get_average_luminance_float(int width, int height,
-                            const float *src, int src_rowstride)
-{
-   float luminance_sum = 0;
-   int y, x;
-
-   for (y = 0; y < height; y++) {
-      for (x = 0; x < width; x++) {
-         luminance_sum += src[0] + src[1] + src[2];
-         src += 3;
-      }
-      src += (src_rowstride - width * 3 * sizeof (float)) / sizeof (float);
-   }
-
-   return luminance_sum / (width * height);
-}
-
-static float
-clamp_value(float value, bool is_signed)
-{
-   if (value > 65504.0f)
-      return 65504.0f;
-
-   if (is_signed) {
-      if (value < -65504.0f)
-         return -65504.0f;
-      else
-         return value;
-   }
-
-   if (value < 0.0f)
-      return 0.0f;
-
-   return value;
-}
-
-static void
-get_endpoints_float(int width, int height,
-                    const float *src, int src_rowstride,
-                    float average_luminance, float endpoints[][3],
-                    bool is_signed)
-{
-   float endpoint_luminances[2];
-   float midpoint;
-   float sums[2][3];
-   int endpoint, component;
-   float luminance;
-   float temp[3];
-   const float *p = src;
-   int left_endpoint_count = 0;
-   int y, x, i;
-
-   memset(sums, 0, sizeof sums);
-
-   for (y = 0; y < height; y++) {
-      for (x = 0; x < width; x++) {
-         luminance = p[0] + p[1] + p[2];
-         if (luminance < average_luminance) {
-            endpoint = 0;
-            left_endpoint_count++;
-         } else {
-            endpoint = 1;
-         }
-         for (i = 0; i < 3; i++)
-            sums[endpoint][i] += p[i];
-
-         p += 3;
-      }
-
-      p += (src_rowstride - width * 3 * sizeof (float)) / sizeof (float);
-   }
-
-   if (left_endpoint_count == 0 ||
-       left_endpoint_count == width * height) {
-      for (i = 0; i < 3; i++)
-         endpoints[0][i] = endpoints[1][i] =
-            (sums[0][i] + sums[1][i]) / (width * height);
-   } else {
-      for (i = 0; i < 3; i++) {
-         endpoints[0][i] = sums[0][i] / left_endpoint_count;
-         endpoints[1][i] = sums[1][i] / (width * height - left_endpoint_count);
-      }
-   }
-
-   /* Clamp the endpoints to the range of a half float and strip out
-    * infinities */
-   for (endpoint = 0; endpoint < 2; endpoint++) {
-      for (component = 0; component < 3; component++) {
-         endpoints[endpoint][component] =
-            clamp_value(endpoints[endpoint][component], is_signed);
-      }
-   }
-
-   /* We may need to swap the endpoints to ensure the most-significant bit of
-    * the first index is zero */
-
-   for (endpoint = 0; endpoint < 2; endpoint++) {
-      endpoint_luminances[endpoint] =
-         endpoints[endpoint][0] +
-         endpoints[endpoint][1] +
-         endpoints[endpoint][2];
-   }
-   midpoint = (endpoint_luminances[0] + endpoint_luminances[1]) / 2.0f;
-
-   if ((src[0] + src[1] + src[2] <= midpoint) !=
-       (endpoint_luminances[0] <= midpoint)) {
-      memcpy(temp, endpoints[0], sizeof temp);
-      memcpy(endpoints[0], endpoints[1], sizeof temp);
-      memcpy(endpoints[1], temp, sizeof temp);
-   }
-}
-
-static void
-write_rgb_indices_float(struct bit_writer *writer,
-                        int src_width, int src_height,
-                        const float *src, int src_rowstride,
-                        float endpoints[][3])
-{
-   float luminance;
-   float endpoint_luminances[2];
-   int endpoint;
-   int index;
-   int y, x;
-
-   for (endpoint = 0; endpoint < 2; endpoint++) {
-      endpoint_luminances[endpoint] =
-         endpoints[endpoint][0] +
-         endpoints[endpoint][1] +
-         endpoints[endpoint][2];
-   }
-
-   /* If the endpoints have the same luminance then we'll just use index 0 for
-    * all of the texels */
-   if (endpoint_luminances[0] == endpoint_luminances[1]) {
-      write_bits(writer, BLOCK_SIZE * BLOCK_SIZE * 4 - 1, 0);
-      return;
-   }
-
-   for (y = 0; y < src_height; y++) {
-      for (x = 0; x < src_width; x++) {
-         luminance = src[0] + src[1] + src[2];
-
-         index = ((luminance - endpoint_luminances[0]) * 15 /
-                  (endpoint_luminances[1] - endpoint_luminances[0]));
-         if (index < 0)
-            index = 0;
-         else if (index > 15)
-            index = 15;
-
-         assert(x != 0 || y != 0 || index < 8);
-
-         write_bits(writer, (x == 0 && y == 0) ? 3 : 4, index);
-
-         src += 3;
-      }
-
-      /* Pad the indices out to the block size */
-      if (src_width < BLOCK_SIZE)
-         write_bits(writer, 4 * (BLOCK_SIZE - src_width), 0);
-
-      src += (src_rowstride - src_width * 3 * sizeof (float)) / sizeof (float);
-   }
-
-   /* Pad the indices out to the block size */
-   if (src_height < BLOCK_SIZE)
-      write_bits(writer, 4 * BLOCK_SIZE * (BLOCK_SIZE - src_height), 0);
-}
-
-static int
-get_endpoint_value(float value, bool is_signed)
-{
-   bool sign = false;
-   int half;
-
-   if (is_signed) {
-      half = _mesa_float_to_half(value);
-
-      if (half & 0x8000) {
-         half &= 0x7fff;
-         sign = true;
-      }
-
-      half = (32 * half / 31) >> 6;
-
-      if (sign)
-         half = -half & ((1 << 10) - 1);
-
-      return half;
-   } else {
-      if (value <= 0.0f)
-         return 0;
-
-      half = _mesa_float_to_half(value);
-
-      return (64 * half / 31) >> 6;
-   }
-}
-
-static void
-compress_rgb_float_block(int src_width, int src_height,
-                         const float *src, int src_rowstride,
-                         uint8_t *dst,
-                         bool is_signed)
-{
-   float average_luminance;
-   float endpoints[2][3];
-   struct bit_writer writer;
-   int component, endpoint;
-   int endpoint_value;
-
-   average_luminance =
-      get_average_luminance_float(src_width, src_height, src, src_rowstride);
-   get_endpoints_float(src_width, src_height, src, src_rowstride,
-                       average_luminance, endpoints, is_signed);
-
-   writer.dst = dst;
-   writer.pos = 0;
-   writer.buf = 0;
-
-   write_bits(&writer, 5, 3); /* mode 3 */
-
-   /* Write the endpoints */
-   for (endpoint = 0; endpoint < 2; endpoint++) {
-      for (component = 0; component < 3; component++) {
-         endpoint_value =
-            get_endpoint_value(endpoints[endpoint][component], is_signed);
-         write_bits(&writer, 10, endpoint_value);
-      }
-   }
-
-   write_rgb_indices_float(&writer,
-                           src_width, src_height,
-                           src, src_rowstride,
-                           endpoints);
-}
-
-static void
-compress_rgb_float(int width, int height,
-                   const float *src, int src_rowstride,
-                   uint8_t *dst, int dst_rowstride,
-                   bool is_signed)
-{
-   int dst_row_diff;
-   int y, x;
-
-   if (dst_rowstride >= width * 4)
-      dst_row_diff = dst_rowstride - ((width + 3) & ~3) * 4;
-   else
-      dst_row_diff = 0;
-
-   for (y = 0; y < height; y += BLOCK_SIZE) {
-      for (x = 0; x < width; x += BLOCK_SIZE) {
-         compress_rgb_float_block(MIN2(width - x, BLOCK_SIZE),
-                                  MIN2(height - y, BLOCK_SIZE),
-                                  src + x * 3 +
-                                  y * src_rowstride / sizeof (float),
-                                  src_rowstride,
-                                  dst,
-                                  is_signed);
-         dst += BLOCK_BYTES;
-      }
-      dst += dst_row_diff;
-   }
-}
-
 static GLboolean
 texstore_bptc_rgb_float(TEXSTORE_PARAMS,
                         bool is_signed)
diff --git a/lib/mesa/src/mesa/main/texcompress_bptc_tmp.h b/lib/mesa/src/mesa/main/texcompress_bptc_tmp.h
new file mode 100644
index 000000000..90837d6af
--- /dev/null
+++ b/lib/mesa/src/mesa/main/texcompress_bptc_tmp.h
@@ -0,0 +1,1748 @@
+/*
+ * Copyright (C) 2014 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ */
+
+/*
+ * Included by texcompress_bptc and gallium to define BPTC decoding routines.
+ */
+
+#ifndef TEXCOMPRESS_BPTC_TMP_H
+#define TEXCOMPRESS_BPTC_TMP_H
+
+#include "util/format_srgb.h"
+#include "util/half_float.h"
+#include "macros.h"
+
+#define BLOCK_SIZE 4
+#define N_PARTITIONS 64
+#define BLOCK_BYTES 16
+
+struct bptc_unorm_mode {
+   int n_subsets;
+   int n_partition_bits;
+   bool has_rotation_bits;
+   bool has_index_selection_bit;
+   int n_color_bits;
+   int n_alpha_bits;
+   bool has_endpoint_pbits;
+   bool has_shared_pbits;
+   int n_index_bits;
+   int n_secondary_index_bits;
+};
+
+struct bptc_float_bitfield {
+   int8_t endpoint;
+   uint8_t component;
+   uint8_t offset;
+   uint8_t n_bits;
+   bool reverse;
+};
+
+struct bptc_float_mode {
+   bool reserved;
+   bool transformed_endpoints;
+   int n_partition_bits;
+   int n_endpoint_bits;
+   int n_index_bits;
+   int n_delta_bits[3];
+   struct bptc_float_bitfield bitfields[24];
+};
+
+struct bit_writer {
+   uint8_t buf;
+   int pos;
+   uint8_t *dst;
+};
+
+static const struct bptc_unorm_mode
+bptc_unorm_modes[] = {
+   /* 0 */ { 3, 4, false, false, 4, 0, true,  false, 3, 0 },
+   /* 1 */ { 2, 6, false, false, 6, 0, false, true,  3, 0 },
+   /* 2 */ { 3, 6, false, false, 5, 0, false, false, 2, 0 },
+   /* 3 */ { 2, 6, false, false, 7, 0, true,  false, 2, 0 },
+   /* 4 */ { 1, 0, true,  true,  5, 6, false, false, 2, 3 },
+   /* 5 */ { 1, 0, true,  false, 7, 8, false, false, 2, 2 },
+   /* 6 */ { 1, 0, false, false, 7, 7, true,  false, 4, 0 },
+   /* 7 */ { 2, 6, false, false, 5, 5, true,  false, 2, 0 }
+};
+
+static const struct bptc_float_mode
+bptc_float_modes[] = {
+   /* 00 */
+   { false, true, 5, 10, 3, { 5, 5, 5 },
+     { { 2, 1, 4, 1, false }, { 2, 2, 4, 1, false }, { 3, 2, 4, 1, false },
+       { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
+       { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false },
+       { 1, 1, 0, 5, false }, { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false },
+       { 1, 2, 0, 5, false }, { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false },
+       { 2, 0, 0, 5, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false },
+       { 3, 2, 3, 1, false },
+       { -1 } }
+   },
+   /* 01 */
+   { false, true, 5, 7, 3, { 6, 6, 6 },
+     { { 2, 1, 5, 1, false }, { 3, 1, 4, 1, false }, { 3, 1, 5, 1, false },
+       { 0, 0, 0, 7, false }, { 3, 2, 0, 1, false }, { 3, 2, 1, 1, false },
+       { 2, 2, 4, 1, false }, { 0, 1, 0, 7, false }, { 2, 2, 5, 1, false },
+       { 3, 2, 2, 1, false }, { 2, 1, 4, 1, false }, { 0, 2, 0, 7, false },
+       { 3, 2, 3, 1, false }, { 3, 2, 5, 1, false }, { 3, 2, 4, 1, false },
+       { 1, 0, 0, 6, false }, { 2, 1, 0, 4, false }, { 1, 1, 0, 6, false },
+       { 3, 1, 0, 4, false }, { 1, 2, 0, 6, false }, { 2, 2, 0, 4, false },
+       { 2, 0, 0, 6, false },
+       { 3, 0, 0, 6, false },
+       { -1 } }
+   },
+   /* 00010 */
+   { false, true, 5, 11, 3, { 5, 4, 4 },
+     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
+       { 1, 0, 0, 5, false }, { 0, 0, 10, 1, false }, { 2, 1, 0, 4, false },
+       { 1, 1, 0, 4, false }, { 0, 1, 10, 1, false }, { 3, 2, 0, 1, false },
+       { 3, 1, 0, 4, false }, { 1, 2, 0, 4, false }, { 0, 2, 10, 1, false },
+       { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 5, false },
+       { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false }, { 3, 2, 3, 1, false },
+       { -1 } }
+   },
+   /* 00011 */
+   { false, false, 0, 10, 4, { 10, 10, 10 },
+     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
+       { 1, 0, 0, 10, false }, { 1, 1, 0, 10, false }, { 1, 2, 0, 10, false },
+       { -1 } }
+   },
+   /* 00110 */
+   { false, true, 5, 11, 3, { 4, 5, 4 },
+     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
+       { 1, 0, 0, 4, false }, { 0, 0, 10, 1, false }, { 3, 1, 4, 1, false },
+       { 2, 1, 0, 4, false }, { 1, 1, 0, 5, false }, { 0, 1, 10, 1, false },
+       { 3, 1, 0, 4, false }, { 1, 2, 0, 4, false }, { 0, 2, 10, 1, false },
+       { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 4, false },
+       { 3, 2, 0, 1, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 4, false },
+       { 2, 1, 4, 1, false }, { 3, 2, 3, 1, false },
+       { -1 } }
+   },
+   /* 00111 */
+   { false, true, 0, 11, 4, { 9, 9, 9 },
+     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
+       { 1, 0, 0, 9, false }, { 0, 0, 10, 1, false }, { 1, 1, 0, 9, false },
+       { 0, 1, 10, 1, false }, { 1, 2, 0, 9, false }, { 0, 2, 10, 1, false },
+       { -1 } }
+   },
+   /* 01010 */
+   { false, true, 5, 11, 3, { 4, 4, 5 },
+     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
+       { 1, 0, 0, 4, false }, { 0, 0, 10, 1, false }, { 2, 2, 4, 1, false },
+       { 2, 1, 0, 4, false }, { 1, 1, 0, 4, false }, { 0, 1, 10, 1, false },
+       { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false }, { 1, 2, 0, 5, false },
+       { 0, 2, 10, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 4, false },
+       { 3, 2, 1, 1, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 4, false },
+       { 3, 2, 4, 1, false }, { 3, 2, 3, 1, false },
+       { -1 } }
+   },
+   /* 01011 */
+   { false, true, 0, 12, 4, { 8, 8, 8 },
+     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
+       { 1, 0, 0, 8, false }, { 0, 0, 10, 2, true }, { 1, 1, 0, 8, false },
+       { 0, 1, 10, 2, true }, { 1, 2, 0, 8, false }, { 0, 2, 10, 2, true },
+       { -1 } }
+   },
+   /* 01110 */
+   { false, true, 5, 9, 3, { 5, 5, 5 },
+     { { 0, 0, 0, 9, false }, { 2, 2, 4, 1, false }, { 0, 1, 0, 9, false },
+       { 2, 1, 4, 1, false }, { 0, 2, 0, 9, false }, { 3, 2, 4, 1, false },
+       { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false },
+       { 1, 1, 0, 5, false }, { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false },
+       { 1, 2, 0, 5, false }, { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false },
+       { 2, 0, 0, 5, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false },
+       { 3, 2, 3, 1, false },
+       { -1 } }
+   },
+   /* 01111 */
+   { false, true, 0, 16, 4, { 4, 4, 4 },
+     { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false },
+       { 1, 0, 0, 4, false }, { 0, 0, 10, 6, true }, { 1, 1, 0, 4, false },
+       { 0, 1, 10, 6, true }, { 1, 2, 0, 4, false }, { 0, 2, 10, 6, true },
+       { -1 } }
+   },
+   /* 10010 */
+   { false, true, 5, 8, 3, { 6, 5, 5 },
+     { { 0, 0, 0, 8, false }, { 3, 1, 4, 1, false }, { 2, 2, 4, 1, false },
+       { 0, 1, 0, 8, false }, { 3, 2, 2, 1, false }, { 2, 1, 4, 1, false },
+       { 0, 2, 0, 8, false }, { 3, 2, 3, 1, false }, { 3, 2, 4, 1, false },
+       { 1, 0, 0, 6, false }, { 2, 1, 0, 4, false }, { 1, 1, 0, 5, false },
+       { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false }, { 1, 2, 0, 5, false },
+       { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 6, false },
+       { 3, 0, 0, 6, false },
+       { -1 } }
+   },
+   /* 10011 */
+   { true /* reserved */ },
+   /* 10110 */
+   { false, true, 5, 8, 3, { 5, 6, 5 },
+     { { 0, 0, 0, 8, false }, { 3, 2, 0, 1, false }, { 2, 2, 4, 1, false },
+       { 0, 1, 0, 8, false }, { 2, 1, 5, 1, false }, { 2, 1, 4, 1, false },
+       { 0, 2, 0, 8, false }, { 3, 1, 5, 1, false }, { 3, 2, 4, 1, false },
+       { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false },
+       { 1, 1, 0, 6, false }, { 3, 1, 0, 4, false }, { 1, 2, 0, 5, false },
+       { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 5, false },
+       { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false }, { 3, 2, 3, 1, false },
+       { -1 } }
+   },
+   /* 10111 */
+   { true /* reserved */ },
+   /* 11010 */
+   { false, true, 5, 8, 3, { 5, 5, 6 },
+     { { 0, 0, 0, 8, false }, { 3, 2, 1, 1, false }, { 2, 2, 4, 1, false },
+       { 0, 1, 0, 8, false }, { 2, 2, 5, 1, false }, { 2, 1, 4, 1, false },
+       { 0, 2, 0, 8, false }, { 3, 2, 5, 1, false }, { 3, 2, 4, 1, false },
+       { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false },
+       { 1, 1, 0, 5, false }, { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false },
+       { 1, 2, 0, 6, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 5, false },
+       { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false }, { 3, 2, 3, 1, false },
+       { -1 } }
+   },
+   /* 11011 */
+   { true /* reserved */ },
+   /* 11110 */
+   { false, false, 5, 6, 3, { 6, 6, 6 },
+     { { 0, 0, 0, 6, false }, { 3, 1, 4, 1, false }, { 3, 2, 0, 1, false },
+       { 3, 2, 1, 1, false }, { 2, 2, 4, 1, false }, { 0, 1, 0, 6, false },
+       { 2, 1, 5, 1, false }, { 2, 2, 5, 1, false }, { 3, 2, 2, 1, false },
+       { 2, 1, 4, 1, false }, { 0, 2, 0, 6, false }, { 3, 1, 5, 1, false },
+       { 3, 2, 3, 1, false }, { 3, 2, 5, 1, false }, { 3, 2, 4, 1, false },
+       { 1, 0, 0, 6, false }, { 2, 1, 0, 4, false }, { 1, 1, 0, 6, false },
+       { 3, 1, 0, 4, false }, { 1, 2, 0, 6, false }, { 2, 2, 0, 4, false },
+       { 2, 0, 0, 6, false }, { 3, 0, 0, 6, false },
+       { -1 } }
+   },
+   /* 11111 */
+   { true /* reserved */ },
+};
+
+/* This partition table is used when the mode has two subsets. Each
+ * partition is represented by a 32-bit value which gives 2 bits per texel
+ * within the block. The value of the two bits represents which subset to use
+ * (0 or 1).
+ */
+static const uint32_t
+partition_table1[N_PARTITIONS] = {
+   0x50505050U, 0x40404040U, 0x54545454U, 0x54505040U,
+   0x50404000U, 0x55545450U, 0x55545040U, 0x54504000U,
+   0x50400000U, 0x55555450U, 0x55544000U, 0x54400000U,
+   0x55555440U, 0x55550000U, 0x55555500U, 0x55000000U,
+   0x55150100U, 0x00004054U, 0x15010000U, 0x00405054U,
+   0x00004050U, 0x15050100U, 0x05010000U, 0x40505054U,
+   0x00404050U, 0x05010100U, 0x14141414U, 0x05141450U,
+   0x01155440U, 0x00555500U, 0x15014054U, 0x05414150U,
+   0x44444444U, 0x55005500U, 0x11441144U, 0x05055050U,
+   0x05500550U, 0x11114444U, 0x41144114U, 0x44111144U,
+   0x15055054U, 0x01055040U, 0x05041050U, 0x05455150U,
+   0x14414114U, 0x50050550U, 0x41411414U, 0x00141400U,
+   0x00041504U, 0x00105410U, 0x10541000U, 0x04150400U,
+   0x50410514U, 0x41051450U, 0x05415014U, 0x14054150U,
+   0x41050514U, 0x41505014U, 0x40011554U, 0x54150140U,
+   0x50505500U, 0x00555050U, 0x15151010U, 0x54540404U,
+};
+
+/* This partition table is used when the mode has three subsets. In this case
+ * the values can be 0, 1 or 2.
+ */
+static const uint32_t
+partition_table2[N_PARTITIONS] = {
+   0xaa685050U, 0x6a5a5040U, 0x5a5a4200U, 0x5450a0a8U,
+   0xa5a50000U, 0xa0a05050U, 0x5555a0a0U, 0x5a5a5050U,
+   0xaa550000U, 0xaa555500U, 0xaaaa5500U, 0x90909090U,
+   0x94949494U, 0xa4a4a4a4U, 0xa9a59450U, 0x2a0a4250U,
+   0xa5945040U, 0x0a425054U, 0xa5a5a500U, 0x55a0a0a0U,
+   0xa8a85454U, 0x6a6a4040U, 0xa4a45000U, 0x1a1a0500U,
+   0x0050a4a4U, 0xaaa59090U, 0x14696914U, 0x69691400U,
+   0xa08585a0U, 0xaa821414U, 0x50a4a450U, 0x6a5a0200U,
+   0xa9a58000U, 0x5090a0a8U, 0xa8a09050U, 0x24242424U,
+   0x00aa5500U, 0x24924924U, 0x24499224U, 0x50a50a50U,
+   0x500aa550U, 0xaaaa4444U, 0x66660000U, 0xa5a0a5a0U,
+   0x50a050a0U, 0x69286928U, 0x44aaaa44U, 0x66666600U,
+   0xaa444444U, 0x54a854a8U, 0x95809580U, 0x96969600U,
+   0xa85454a8U, 0x80959580U, 0xaa141414U, 0x96960000U,
+   0xaaaa1414U, 0xa05050a0U, 0xa0a5a5a0U, 0x96000000U,
+   0x40804080U, 0xa9a8a9a8U, 0xaaaaaa44U, 0x2a4a5254U
+};
+
+static const uint8_t
+anchor_indices[][N_PARTITIONS] = {
+   /* Anchor index values for the second subset of two-subset partitioning */
+   {
+      0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,
+      0xf,0x2,0x8,0x2,0x2,0x8,0x8,0xf,0x2,0x8,0x2,0x2,0x8,0x8,0x2,0x2,
+      0xf,0xf,0x6,0x8,0x2,0x8,0xf,0xf,0x2,0x8,0x2,0x2,0x2,0xf,0xf,0x6,
+      0x6,0x2,0x6,0x8,0xf,0xf,0x2,0x2,0xf,0xf,0xf,0xf,0xf,0x2,0x2,0xf
+   },
+
+   /* Anchor index values for the second subset of three-subset partitioning */
+   {
+      0x3,0x3,0xf,0xf,0x8,0x3,0xf,0xf,0x8,0x8,0x6,0x6,0x6,0x5,0x3,0x3,
+      0x3,0x3,0x8,0xf,0x3,0x3,0x6,0xa,0x5,0x8,0x8,0x6,0x8,0x5,0xf,0xf,
+      0x8,0xf,0x3,0x5,0x6,0xa,0x8,0xf,0xf,0x3,0xf,0x5,0xf,0xf,0xf,0xf,
+      0x3,0xf,0x5,0x5,0x5,0x8,0x5,0xa,0x5,0xa,0x8,0xd,0xf,0xc,0x3,0x3
+   },
+
+   /* Anchor index values for the third subset of three-subset
+    * partitioning
+    */
+   {
+      0xf,0x8,0x8,0x3,0xf,0xf,0x3,0x8,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0x8,
+      0xf,0x8,0xf,0x3,0xf,0x8,0xf,0x8,0x3,0xf,0x6,0xa,0xf,0xf,0xa,0x8,
+      0xf,0x3,0xf,0xa,0xa,0x8,0x9,0xa,0x6,0xf,0x8,0xf,0x3,0x6,0x6,0x8,
+      0xf,0x3,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0x3,0xf,0xf,0x8
+   }
+};
+
+static int
+extract_bits(const uint8_t *block,
+             int offset,
+             int n_bits)
+{
+   int byte_index = offset / 8;
+   int bit_index = offset % 8;
+   int n_bits_in_byte = MIN2(n_bits, 8 - bit_index);
+   int result = 0;
+   int bit = 0;
+
+   while (true) {
+      result |= ((block[byte_index] >> bit_index) &
+                 ((1 << n_bits_in_byte) - 1)) << bit;
+
+      n_bits -= n_bits_in_byte;
+
+      if (n_bits <= 0)
+         return result;
+
+      bit += n_bits_in_byte;
+      byte_index++;
+      bit_index = 0;
+      n_bits_in_byte = MIN2(n_bits, 8);
+   }
+}
+
+static uint8_t
+expand_component(uint8_t byte,
+                 int n_bits)
+{
+   /* Expands a n-bit quantity into a byte by copying the most-significant
+    * bits into the unused least-significant bits.
+    */
+   return byte << (8 - n_bits) | (byte >> (2 * n_bits - 8));
+}
+
+static int
+extract_unorm_endpoints(const struct bptc_unorm_mode *mode,
+                        const uint8_t *block,
+                        int bit_offset,
+                        uint8_t endpoints[][4])
+{
+   int component;
+   int subset;
+   int endpoint;
+   int pbit;
+   int n_components;
+
+   /* Extract each color component */
+   for (component = 0; component < 3; component++) {
+      for (subset = 0; subset < mode->n_subsets; subset++) {
+         for (endpoint = 0; endpoint < 2; endpoint++) {
+            endpoints[subset * 2 + endpoint][component] =
+               extract_bits(block, bit_offset, mode->n_color_bits);
+            bit_offset += mode->n_color_bits;
+         }
+      }
+   }
+
+   /* Extract the alpha values */
+   if (mode->n_alpha_bits > 0) {
+      for (subset = 0; subset < mode->n_subsets; subset++) {
+         for (endpoint = 0; endpoint < 2; endpoint++) {
+            endpoints[subset * 2 + endpoint][3] =
+               extract_bits(block, bit_offset, mode->n_alpha_bits);
+            bit_offset += mode->n_alpha_bits;
+         }
+      }
+
+      n_components = 4;
+   } else {
+      for (subset = 0; subset < mode->n_subsets; subset++)
+         for (endpoint = 0; endpoint < 2; endpoint++)
+            endpoints[subset * 2 + endpoint][3] = 255;
+
+      n_components = 3;
+   }
+
+   /* Add in the p-bits */
+   if (mode->has_endpoint_pbits) {
+      for (subset = 0; subset < mode->n_subsets; subset++) {
+         for (endpoint = 0; endpoint < 2; endpoint++) {
+            pbit = extract_bits(block, bit_offset, 1);
+            bit_offset += 1;
+
+            for (component = 0; component < n_components; component++) {
+               endpoints[subset * 2 + endpoint][component] <<= 1;
+               endpoints[subset * 2 + endpoint][component] |= pbit;
+            }
+         }
+      }
+   } else if (mode->has_shared_pbits) {
+      for (subset = 0; subset < mode->n_subsets; subset++) {
+         pbit = extract_bits(block, bit_offset, 1);
+         bit_offset += 1;
+
+         for (endpoint = 0; endpoint < 2; endpoint++) {
+            for (component = 0; component < n_components; component++) {
+               endpoints[subset * 2 + endpoint][component] <<= 1;
+               endpoints[subset * 2 + endpoint][component] |= pbit;
+            }
+         }
+      }
+   }
+
+   /* Expand the n-bit values to a byte */
+   for (subset = 0; subset < mode->n_subsets; subset++) {
+      for (endpoint = 0; endpoint < 2; endpoint++) {
+         for (component = 0; component < 3; component++) {
+            endpoints[subset * 2 + endpoint][component] =
+               expand_component(endpoints[subset * 2 + endpoint][component],
+                                mode->n_color_bits +
+                                mode->has_endpoint_pbits +
+                                mode->has_shared_pbits);
+         }
+
+         if (mode->n_alpha_bits > 0) {
+            endpoints[subset * 2 + endpoint][3] =
+               expand_component(endpoints[subset * 2 + endpoint][3],
+                                mode->n_alpha_bits +
+                                mode->has_endpoint_pbits +
+                                mode->has_shared_pbits);
+         }
+      }
+   }
+
+   return bit_offset;
+}
+
+static bool
+is_anchor(int n_subsets,
+          int partition_num,
+          int texel)
+{
+   if (texel == 0)
+      return true;
+
+   switch (n_subsets) {
+   case 1:
+      return false;
+   case 2:
+      return anchor_indices[0][partition_num] == texel;
+   case 3:
+      return (anchor_indices[1][partition_num] == texel ||
+              anchor_indices[2][partition_num] == texel);
+   default:
+      assert(false);
+      return false;
+   }
+}
+
+static int
+count_anchors_before_texel(int n_subsets,
+                           int partition_num,
+                           int texel)
+{
+   int count = 1;
+
+   if (texel == 0)
+      return 0;
+
+   switch (n_subsets) {
+   case 1:
+      break;
+   case 2:
+      if (texel > anchor_indices[0][partition_num])
+         count++;
+      break;
+   case 3:
+      if (texel > anchor_indices[1][partition_num])
+         count++;
+      if (texel > anchor_indices[2][partition_num])
+         count++;
+      break;
+   default:
+      assert(false);
+      return 0;
+   }
+
+   return count;
+}
+
+static int32_t
+interpolate(int32_t a, int32_t b,
+            int index,
+            int index_bits)
+{
+   static const uint8_t weights2[] = { 0, 21, 43, 64 };
+   static const uint8_t weights3[] = { 0, 9, 18, 27, 37, 46, 55, 64 };
+   static const uint8_t weights4[] =
+      { 0, 4, 9, 13, 17, 21, 26, 30, 34, 38, 43, 47, 51, 55, 60, 64 };
+   static const uint8_t *weights[] = {
+      NULL, NULL, weights2, weights3, weights4
+   };
+   int weight;
+
+   weight = weights[index_bits][index];
+
+   return ((64 - weight) * a + weight * b + 32) >> 6;
+}
+
+static void
+apply_rotation(int rotation,
+               uint8_t *result)
+{
+   uint8_t t;
+
+   if (rotation == 0)
+      return;
+
+   rotation--;
+
+   t = result[rotation];
+   result[rotation] = result[3];
+   result[3] = t;
+}
+
+static void
+fetch_rgba_unorm_from_block(const uint8_t *block,
+                            uint8_t *result,
+                            int texel)
+{
+   int mode_num = ffs(block[0]);
+   const struct bptc_unorm_mode *mode;
+   int bit_offset, secondary_bit_offset;
+   int partition_num;
+   int subset_num;
+   int rotation;
+   int index_selection;
+   int index_bits;
+   int indices[2];
+   int index;
+   int anchors_before_texel;
+   bool anchor;
+   uint8_t endpoints[3 * 2][4];
+   uint32_t subsets;
+   int component;
+
+   if (mode_num == 0) {
+      /* According to the spec this mode is reserved and shouldn't be used. */
+      memset(result, 0, 3);
+      result[3] = 0xff;
+      return;
+   }
+
+   mode = bptc_unorm_modes + mode_num - 1;
+   bit_offset = mode_num;
+
+   partition_num = extract_bits(block, bit_offset, mode->n_partition_bits);
+   bit_offset += mode->n_partition_bits;
+
+   switch (mode->n_subsets) {
+   case 1:
+      subsets = 0;
+      break;
+   case 2:
+      subsets = partition_table1[partition_num];
+      break;
+   case 3:
+      subsets = partition_table2[partition_num];
+      break;
+   default:
+      assert(false);
+      return;
+   }
+
+   if (mode->has_rotation_bits) {
+      rotation = extract_bits(block, bit_offset, 2);
+      bit_offset += 2;
+   } else {
+      rotation = 0;
+   }
+
+   if (mode->has_index_selection_bit) {
+      index_selection = extract_bits(block, bit_offset, 1);
+      bit_offset++;
+   } else {
+      index_selection = 0;
+   }
+
+   bit_offset = extract_unorm_endpoints(mode, block, bit_offset, endpoints);
+
+   anchors_before_texel = count_anchors_before_texel(mode->n_subsets,
+                                                     partition_num, texel);
+
+   /* Calculate the offset to the secondary index */
+   secondary_bit_offset = (bit_offset +
+                           BLOCK_SIZE * BLOCK_SIZE * mode->n_index_bits -
+                           mode->n_subsets +
+                           mode->n_secondary_index_bits * texel -
+                           anchors_before_texel);
+
+   /* Calculate the offset to the primary index for this texel */
+   bit_offset += mode->n_index_bits * texel - anchors_before_texel;
+
+   subset_num = (subsets >> (texel * 2)) & 3;
+
+   anchor = is_anchor(mode->n_subsets, partition_num, texel);
+
+   index_bits = mode->n_index_bits;
+   if (anchor)
+      index_bits--;
+   indices[0] = extract_bits(block, bit_offset, index_bits);
+
+   if (mode->n_secondary_index_bits) {
+      index_bits = mode->n_secondary_index_bits;
+      if (anchor)
+         index_bits--;
+      indices[1] = extract_bits(block, secondary_bit_offset, index_bits);
+   }
+
+   index = indices[index_selection];
+   index_bits = (index_selection ?
+                 mode->n_secondary_index_bits :
+                 mode->n_index_bits);
+
+   for (component = 0; component < 3; component++)
+      result[component] = interpolate(endpoints[subset_num * 2][component],
+                                      endpoints[subset_num * 2 + 1][component],
+                                      index,
+                                      index_bits);
+
+   /* Alpha uses the opposite index from the color components */
+   if (mode->n_secondary_index_bits && !index_selection) {
+      index = indices[1];
+      index_bits = mode->n_secondary_index_bits;
+   } else {
+      index = indices[0];
+      index_bits = mode->n_index_bits;
+   }
+
+   result[3] = interpolate(endpoints[subset_num * 2][3],
+                           endpoints[subset_num * 2 + 1][3],
+                           index,
+                           index_bits);
+
+   apply_rotation(rotation, result);
+}
+
+#ifdef BPTC_BLOCK_DECODE
+static void
+decompress_rgba_unorm_block(int src_width, int src_height,
+                            const uint8_t *block,
+                            uint8_t *dst_row, int dst_rowstride)
+{
+   int mode_num = ffs(block[0]);
+   const struct bptc_unorm_mode *mode;
+   int bit_offset, secondary_bit_offset;
+   int partition_num;
+   int subset_num;
+   int rotation;
+   int index_selection;
+   int index_bits;
+   int indices[2];
+   int index;
+   int anchors_before_texel;
+   bool anchor;
+   uint8_t endpoints[3 * 2][4];
+   uint32_t subsets;
+   int component;
+   unsigned x, y;
+
+   if (mode_num == 0) {
+      /* According to the spec this mode is reserved and shouldn't be used. */
+      for(y = 0; y < src_height; y += 1) {
+         uint8_t *result = dst_row;
+         memset(result, 0, 4 * src_width);
+         for(x = 0; x < src_width; x += 1) {
+            result[3] = 0xff;
+            result += 4;
+         }
+         dst_row += dst_rowstride;
+      }
+      return;
+   }
+
+   mode = bptc_unorm_modes + mode_num - 1;
+   bit_offset = mode_num;
+
+   partition_num = extract_bits(block, bit_offset, mode->n_partition_bits);
+   bit_offset += mode->n_partition_bits;
+
+   switch (mode->n_subsets) {
+   case 1:
+      subsets = 0;
+      break;
+   case 2:
+      subsets = partition_table1[partition_num];
+      break;
+   case 3:
+      subsets = partition_table2[partition_num];
+      break;
+   default:
+      assert(false);
+      return;
+   }
+
+   if (mode->has_rotation_bits) {
+      rotation = extract_bits(block, bit_offset, 2);
+      bit_offset += 2;
+   } else {
+      rotation = 0;
+   }
+
+   if (mode->has_index_selection_bit) {
+      index_selection = extract_bits(block, bit_offset, 1);
+      bit_offset++;
+   } else {
+      index_selection = 0;
+   }
+
+   bit_offset = extract_unorm_endpoints(mode, block, bit_offset, endpoints);
+
+   for(y = 0; y < src_height; y += 1) {
+      uint8_t *result = dst_row;
+      for(x = 0; x < src_width; x += 1) {
+         int texel;
+         texel = x + y * 4;
+
+         anchors_before_texel = count_anchors_before_texel(mode->n_subsets,
+                                                           partition_num,
+                                                           texel);
+
+         /* Calculate the offset to the secondary index */
+         secondary_bit_offset = (bit_offset +
+                                 BLOCK_SIZE * BLOCK_SIZE * mode->n_index_bits -
+                                 mode->n_subsets +
+                                 mode->n_secondary_index_bits * texel -
+                                 anchors_before_texel);
+
+         /* Calculate the offset to the primary index for this texel */
+         bit_offset += mode->n_index_bits * texel - anchors_before_texel;
+
+         subset_num = (subsets >> (texel * 2)) & 3;
+
+         anchor = is_anchor(mode->n_subsets, partition_num, texel);
+
+         index_bits = mode->n_index_bits;
+         if (anchor)
+            index_bits--;
+         indices[0] = extract_bits(block, bit_offset, index_bits);
+
+         if (mode->n_secondary_index_bits) {
+            index_bits = mode->n_secondary_index_bits;
+            if (anchor)
+               index_bits--;
+            indices[1] = extract_bits(block, secondary_bit_offset, index_bits);
+         }
+
+         index = indices[index_selection];
+         index_bits = (index_selection ?
+                       mode->n_secondary_index_bits :
+                       mode->n_index_bits);
+
+         for (component = 0; component < 3; component++)
+            result[component] = interpolate(endpoints[subset_num * 2][component],
+                                            endpoints[subset_num * 2 + 1][component],
+                                            index,
+                                            index_bits);
+
+         /* Alpha uses the opposite index from the color components */
+         if (mode->n_secondary_index_bits && !index_selection) {
+            index = indices[1];
+            index_bits = mode->n_secondary_index_bits;
+         } else {
+            index = indices[0];
+            index_bits = mode->n_index_bits;
+         }
+
+         result[3] = interpolate(endpoints[subset_num * 2][3],
+                                 endpoints[subset_num * 2 + 1][3],
+                                 index,
+                                 index_bits);
+
+         apply_rotation(rotation, result);
+         result += 4;
+      }
+      dst_row += dst_rowstride;
+   }
+}
+
+static void
+decompress_rgba_unorm(int width, int height,
+                      const uint8_t *src, int src_rowstride,
+                      uint8_t *dst, int dst_rowstride)
+{
+   int src_row_diff;
+   int y, x;
+
+   if (src_rowstride >= width * 4)
+      src_row_diff = src_rowstride - ((width + 3) & ~3) * 4;
+   else
+      src_row_diff = 0;
+
+   for (y = 0; y < height; y += BLOCK_SIZE) {
+      for (x = 0; x < width; x += BLOCK_SIZE) {
+         decompress_rgba_unorm_block(MIN2(width - x, BLOCK_SIZE),
+                                     MIN2(height - y, BLOCK_SIZE),
+                                     src,
+                                     dst + x * 4 + y * dst_rowstride,
+                                     dst_rowstride);
+         src += BLOCK_BYTES;
+      }
+      src += src_row_diff;
+   }
+}
+#endif // BPTC_BLOCK_DECODE
+
+static int32_t
+sign_extend(int32_t value,
+            int n_bits)
+{
+   if ((value & (1 << (n_bits - 1)))) {
+      value |= (~(int32_t) 0) << n_bits;
+   }
+
+   return value;
+}
+
+static int
+signed_unquantize(int value, int n_endpoint_bits)
+{
+   bool sign;
+
+   if (n_endpoint_bits >= 16)
+      return value;
+
+   if (value == 0)
+      return 0;
+
+   sign = false;
+
+   if (value < 0) {
+      sign = true;
+      value = -value;
+   }
+
+   if (value >= (1 << (n_endpoint_bits - 1)) - 1)
+      value = 0x7fff;
+   else
+      value = ((value << 15) + 0x4000) >> (n_endpoint_bits - 1);
+
+   if (sign)
+      value = -value;
+
+   return value;
+}
+
+static int
+unsigned_unquantize(int value, int n_endpoint_bits)
+{
+   if (n_endpoint_bits >= 15)
+      return value;
+
+   if (value == 0)
+      return 0;
+
+   if (value == (1 << n_endpoint_bits) - 1)
+      return 0xffff;
+
+   return ((value << 15) + 0x4000) >> (n_endpoint_bits - 1);
+}
+
+static int
+extract_float_endpoints(const struct bptc_float_mode *mode,
+                        const uint8_t *block,
+                        int bit_offset,
+                        int32_t endpoints[][3],
+                        bool is_signed)
+{
+   const struct bptc_float_bitfield *bitfield;
+   int endpoint, component;
+   int n_endpoints;
+   int value;
+   int i;
+
+   if (mode->n_partition_bits)
+      n_endpoints = 4;
+   else
+      n_endpoints = 2;
+
+   memset(endpoints, 0, sizeof endpoints[0][0] * n_endpoints * 3);
+
+   for (bitfield = mode->bitfields; bitfield->endpoint != -1; bitfield++) {
+      value = extract_bits(block, bit_offset, bitfield->n_bits);
+      bit_offset += bitfield->n_bits;
+
+      if (bitfield->reverse) {
+         for (i = 0; i < bitfield->n_bits; i++) {
+            if (value & (1 << i))
+               endpoints[bitfield->endpoint][bitfield->component] |=
+                  1 << ((bitfield->n_bits - 1 - i) + bitfield->offset);
+         }
+      } else {
+         endpoints[bitfield->endpoint][bitfield->component] |=
+            value << bitfield->offset;
+      }
+   }
+
+   if (mode->transformed_endpoints) {
+      /* The endpoints are specified as signed offsets from e0 */
+      for (endpoint = 1; endpoint < n_endpoints; endpoint++) {
+         for (component = 0; component < 3; component++) {
+            value = sign_extend(endpoints[endpoint][component],
+                                mode->n_delta_bits[component]);
+            endpoints[endpoint][component] =
+               ((endpoints[0][component] + value) &
+                ((1 << mode->n_endpoint_bits) - 1));
+         }
+      }
+   }
+
+   if (is_signed) {
+      for (endpoint = 0; endpoint < n_endpoints; endpoint++) {
+         for (component = 0; component < 3; component++) {
+            value = sign_extend(endpoints[endpoint][component],
+                                mode->n_endpoint_bits);
+            endpoints[endpoint][component] =
+               signed_unquantize(value, mode->n_endpoint_bits);
+         }
+      }
+   } else {
+      for (endpoint = 0; endpoint < n_endpoints; endpoint++) {
+         for (component = 0; component < 3; component++) {
+            endpoints[endpoint][component] =
+               unsigned_unquantize(endpoints[endpoint][component],
+                                   mode->n_endpoint_bits);
+         }
+      }
+   }
+
+   return bit_offset;
+}
+
+static int32_t
+finish_unsigned_unquantize(int32_t value)
+{
+   return value * 31 / 64;
+}
+
+static int32_t
+finish_signed_unquantize(int32_t value)
+{
+   if (value < 0)
+      return (-value * 31 / 32) | 0x8000;
+   else
+      return value * 31 / 32;
+}
+
+static void
+fetch_rgb_float_from_block(const uint8_t *block,
+                           float *result,
+                           int texel,
+                           bool is_signed)
+{
+   int mode_num;
+   const struct bptc_float_mode *mode;
+   int bit_offset;
+   int partition_num;
+   int subset_num;
+   int index_bits;
+   int index;
+   int anchors_before_texel;
+   int32_t endpoints[2 * 2][3];
+   uint32_t subsets;
+   int n_subsets;
+   int component;
+   int32_t value;
+
+   if (block[0] & 0x2) {
+      mode_num = (((block[0] >> 1) & 0xe) | (block[0] & 1)) + 2;
+      bit_offset = 5;
+   } else {
+      mode_num = block[0] & 3;
+      bit_offset = 2;
+   }
+
+   mode = bptc_float_modes + mode_num;
+
+   if (mode->reserved) {
+      memset(result, 0, sizeof result[0] * 3);
+      result[3] = 1.0f;
+      return;
+   }
+
+   bit_offset = extract_float_endpoints(mode, block, bit_offset,
+                                        endpoints, is_signed);
+
+   if (mode->n_partition_bits) {
+      partition_num = extract_bits(block, bit_offset, mode->n_partition_bits);
+      bit_offset += mode->n_partition_bits;
+
+      subsets = partition_table1[partition_num];
+      n_subsets = 2;
+   } else {
+      partition_num = 0;
+      subsets = 0;
+      n_subsets = 1;
+   }
+
+   anchors_before_texel =
+      count_anchors_before_texel(n_subsets, partition_num, texel);
+
+   /* Calculate the offset to the primary index for this texel */
+   bit_offset += mode->n_index_bits * texel - anchors_before_texel;
+
+   subset_num = (subsets >> (texel * 2)) & 3;
+
+   index_bits = mode->n_index_bits;
+   if (is_anchor(n_subsets, partition_num, texel))
+      index_bits--;
+   index = extract_bits(block, bit_offset, index_bits);
+
+   for (component = 0; component < 3; component++) {
+      value = interpolate(endpoints[subset_num * 2][component],
+                          endpoints[subset_num * 2 + 1][component],
+                          index,
+                          mode->n_index_bits);
+
+      if (is_signed)
+         value = finish_signed_unquantize(value);
+      else
+         value = finish_unsigned_unquantize(value);
+
+      result[component] = _mesa_half_to_float(value);
+   }
+
+   result[3] = 1.0f;
+}
+
+#ifdef BPTC_BLOCK_DECODE
+static void
+decompress_rgb_float_block(unsigned src_width, unsigned src_height,
+                           const uint8_t *block,
+                           float *dst_row, unsigned dst_rowstride,
+                           bool is_signed)
+{
+   int mode_num;
+   const struct bptc_float_mode *mode;
+   int bit_offset;
+   int partition_num;
+   int subset_num;
+   int index_bits;
+   int index;
+   int anchors_before_texel;
+   int32_t endpoints[2 * 2][3];
+   uint32_t subsets;
+   int n_subsets;
+   int component;
+   int32_t value;
+   unsigned x, y;
+
+   if (block[0] & 0x2) {
+      mode_num = (((block[0] >> 1) & 0xe) | (block[0] & 1)) + 2;
+      bit_offset = 5;
+   } else {
+      mode_num = block[0] & 3;
+      bit_offset = 2;
+   }
+
+   mode = bptc_float_modes + mode_num;
+
+   if (mode->reserved) {
+      for(y = 0; y < src_height; y += 1) {
+         float *result = dst_row;
+         memset(result, 0, sizeof result[0] * 4 * src_width);
+         for(x = 0; x < src_width; x += 1) {
+            result[3] = 1.0f;
+            result += 4;
+         }
+         dst_row += dst_rowstride / sizeof dst_row[0];
+      }
+      return;
+   }
+
+   bit_offset = extract_float_endpoints(mode, block, bit_offset,
+                                        endpoints, is_signed);
+
+   if (mode->n_partition_bits) {
+      partition_num = extract_bits(block, bit_offset, mode->n_partition_bits);
+      bit_offset += mode->n_partition_bits;
+
+      subsets = partition_table1[partition_num];
+      n_subsets = 2;
+   } else {
+      partition_num = 0;
+      subsets = 0;
+      n_subsets = 1;
+   }
+
+   for(y = 0; y < src_height; y += 1) {
+      float *result = dst_row;
+      for(x = 0; x < src_width; x += 1) {
+         int texel;
+
+         texel = x + y * 4;
+
+         anchors_before_texel =
+            count_anchors_before_texel(n_subsets, partition_num, texel);
+
+         /* Calculate the offset to the primary index for this texel */
+         bit_offset += mode->n_index_bits * texel - anchors_before_texel;
+
+         subset_num = (subsets >> (texel * 2)) & 3;
+
+         index_bits = mode->n_index_bits;
+         if (is_anchor(n_subsets, partition_num, texel))
+            index_bits--;
+         index = extract_bits(block, bit_offset, index_bits);
+
+         for (component = 0; component < 3; component++) {
+            value = interpolate(endpoints[subset_num * 2][component],
+                                endpoints[subset_num * 2 + 1][component],
+                                index,
+                                mode->n_index_bits);
+
+            if (is_signed)
+               value = finish_signed_unquantize(value);
+            else
+               value = finish_unsigned_unquantize(value);
+
+            result[component] = _mesa_half_to_float(value);
+         }
+
+         result[3] = 1.0f;
+         result += 4;
+      }
+      dst_row += dst_rowstride / sizeof dst_row[0];
+   }
+}
+
+static void
+decompress_rgb_float(int width, int height,
+                      const uint8_t *src, int src_rowstride,
+                      float *dst, int dst_rowstride, bool is_signed)
+{
+   int src_row_diff;
+   int y, x;
+
+   if (src_rowstride >= width * 4)
+      src_row_diff = src_rowstride - ((width + 3) & ~3) * 4;
+   else
+      src_row_diff = 0;
+
+   for (y = 0; y < height; y += BLOCK_SIZE) {
+      for (x = 0; x < width; x += BLOCK_SIZE) {
+         decompress_rgb_float_block(MIN2(width - x, BLOCK_SIZE),
+                                    MIN2(height - y, BLOCK_SIZE),
+                                    src,
+                                    (dst + x * 4 +
+                                     (y * dst_rowstride / sizeof dst[0])),
+                                    dst_rowstride, is_signed);
+         src += BLOCK_BYTES;
+      }
+      src += src_row_diff;
+   }
+}
+#endif // BPTC_BLOCK_DECODE
+
+static void
+write_bits(struct bit_writer *writer, int n_bits, int value)
+{
+   do {
+      if (n_bits + writer->pos >= 8) {
+         *(writer->dst++) = writer->buf | (value << writer->pos);
+         writer->buf = 0;
+         value >>= (8 - writer->pos);
+         n_bits -= (8 - writer->pos);
+         writer->pos = 0;
+      } else {
+         writer->buf |= value << writer->pos;
+         writer->pos += n_bits;
+         break;
+      }
+   } while (n_bits > 0);
+}
+
+static void
+get_average_luminance_alpha_unorm(int width, int height,
+                                  const uint8_t *src, int src_rowstride,
+                                  int *average_luminance, int *average_alpha)
+{
+   int luminance_sum = 0, alpha_sum = 0;
+   int y, x;
+
+   for (y = 0; y < height; y++) {
+      for (x = 0; x < width; x++) {
+         luminance_sum += src[0] + src[1] + src[2];
+         alpha_sum += src[3];
+         src += 4;
+      }
+      src += src_rowstride - width * 4;
+   }
+
+   *average_luminance = luminance_sum / (width * height);
+   *average_alpha = alpha_sum / (width * height);
+}
+
+static void
+get_rgba_endpoints_unorm(int width, int height,
+                         const uint8_t *src, int src_rowstride,
+                         int average_luminance, int average_alpha,
+                         uint8_t endpoints[][4])
+{
+   int endpoint_luminances[2];
+   int midpoint;
+   int sums[2][4];
+   int endpoint;
+   int luminance;
+   uint8_t temp[3];
+   const uint8_t *p = src;
+   int rgb_left_endpoint_count = 0;
+   int alpha_left_endpoint_count = 0;
+   int y, x, i;
+
+   memset(sums, 0, sizeof sums);
+
+   for (y = 0; y < height; y++) {
+      for (x = 0; x < width; x++) {
+         luminance = p[0] + p[1] + p[2];
+         if (luminance < average_luminance) {
+            endpoint = 0;
+            rgb_left_endpoint_count++;
+         } else {
+            endpoint = 1;
+         }
+         for (i = 0; i < 3; i++)
+            sums[endpoint][i] += p[i];
+
+         if (p[2] < average_alpha) {
+            endpoint = 0;
+            alpha_left_endpoint_count++;
+         } else {
+            endpoint = 1;
+         }
+         sums[endpoint][3] += p[3];
+
+         p += 4;
+      }
+
+      p += src_rowstride - width * 4;
+   }
+
+   if (rgb_left_endpoint_count == 0 ||
+       rgb_left_endpoint_count == width * height) {
+      for (i = 0; i < 3; i++)
+         endpoints[0][i] = endpoints[1][i] =
+            (sums[0][i] + sums[1][i]) / (width * height);
+   } else {
+      for (i = 0; i < 3; i++) {
+         endpoints[0][i] = sums[0][i] / rgb_left_endpoint_count;
+         endpoints[1][i] = (sums[1][i] /
+                            (width * height - rgb_left_endpoint_count));
+      }
+   }
+
+   if (alpha_left_endpoint_count == 0 ||
+       alpha_left_endpoint_count == width * height) {
+      endpoints[0][3] = endpoints[1][3] =
+         (sums[0][3] + sums[1][3]) / (width * height);
+   } else {
+         endpoints[0][3] = sums[0][3] / alpha_left_endpoint_count;
+         endpoints[1][3] = (sums[1][3] /
+                            (width * height - alpha_left_endpoint_count));
+   }
+
+   /* We may need to swap the endpoints to ensure the most-significant bit of
+    * the first index is zero */
+
+   for (endpoint = 0; endpoint < 2; endpoint++) {
+      endpoint_luminances[endpoint] =
+         endpoints[endpoint][0] +
+         endpoints[endpoint][1] +
+         endpoints[endpoint][2];
+   }
+   midpoint = (endpoint_luminances[0] + endpoint_luminances[1]) / 2;
+
+   if ((src[0] + src[1] + src[2] <= midpoint) !=
+       (endpoint_luminances[0] <= midpoint)) {
+      memcpy(temp, endpoints[0], 3);
+      memcpy(endpoints[0], endpoints[1], 3);
+      memcpy(endpoints[1], temp, 3);
+   }
+
+   /* Same for the alpha endpoints */
+
+   midpoint = (endpoints[0][3] + endpoints[1][3]) / 2;
+
+   if ((src[3] <= midpoint) != (endpoints[0][3] <= midpoint)) {
+      temp[0] = endpoints[0][3];
+      endpoints[0][3] = endpoints[1][3];
+      endpoints[1][3] = temp[0];
+   }
+}
+
+static void
+write_rgb_indices_unorm(struct bit_writer *writer,
+                        int src_width, int src_height,
+                        const uint8_t *src, int src_rowstride,
+                        uint8_t endpoints[][4])
+{
+   int luminance;
+   int endpoint_luminances[2];
+   int endpoint;
+   int index;
+   int y, x;
+
+   for (endpoint = 0; endpoint < 2; endpoint++) {
+      endpoint_luminances[endpoint] =
+         endpoints[endpoint][0] +
+         endpoints[endpoint][1] +
+         endpoints[endpoint][2];
+   }
+
+   /* If the endpoints have the same luminance then we'll just use index 0 for
+    * all of the texels */
+   if (endpoint_luminances[0] == endpoint_luminances[1]) {
+      write_bits(writer, BLOCK_SIZE * BLOCK_SIZE * 2 - 1, 0);
+      return;
+   }
+
+   for (y = 0; y < src_height; y++) {
+      for (x = 0; x < src_width; x++) {
+         luminance = src[0] + src[1] + src[2];
+
+         index = ((luminance - endpoint_luminances[0]) * 3 /
+                  (endpoint_luminances[1] - endpoint_luminances[0]));
+         if (index < 0)
+            index = 0;
+         else if (index > 3)
+            index = 3;
+
+         assert(x != 0 || y != 0 || index < 2);
+
+         write_bits(writer, (x == 0 && y == 0) ? 1 : 2, index);
+
+         src += 4;
+      }
+
+      /* Pad the indices out to the block size */
+      if (src_width < BLOCK_SIZE)
+         write_bits(writer, 2 * (BLOCK_SIZE - src_width), 0);
+
+      src += src_rowstride - src_width * 4;
+   }
+
+   /* Pad the indices out to the block size */
+   if (src_height < BLOCK_SIZE)
+      write_bits(writer, 2 * BLOCK_SIZE * (BLOCK_SIZE - src_height), 0);
+}
+
+static void
+write_alpha_indices_unorm(struct bit_writer *writer,
+                          int src_width, int src_height,
+                          const uint8_t *src, int src_rowstride,
+                          uint8_t endpoints[][4])
+{
+   int index;
+   int y, x;
+
+   /* If the endpoints have the same alpha then we'll just use index 0 for
+    * all of the texels */
+   if (endpoints[0][3] == endpoints[1][3]) {
+      write_bits(writer, BLOCK_SIZE * BLOCK_SIZE * 3 - 1, 0);
+      return;
+   }
+
+   for (y = 0; y < src_height; y++) {
+      for (x = 0; x < src_width; x++) {
+         index = (((int) src[3] - (int) endpoints[0][3]) * 7 /
+                  ((int) endpoints[1][3] - endpoints[0][3]));
+         if (index < 0)
+            index = 0;
+         else if (index > 7)
+            index = 7;
+
+         assert(x != 0 || y != 0 || index < 4);
+
+         /* The first index has one less bit */
+         write_bits(writer, (x == 0 && y == 0) ? 2 : 3, index);
+
+         src += 4;
+      }
+
+      /* Pad the indices out to the block size */
+      if (src_width < BLOCK_SIZE)
+         write_bits(writer, 3 * (BLOCK_SIZE - src_width), 0);
+
+      src += src_rowstride - src_width * 4;
+   }
+
+   /* Pad the indices out to the block size */
+   if (src_height < BLOCK_SIZE)
+      write_bits(writer, 3 * BLOCK_SIZE * (BLOCK_SIZE - src_height), 0);
+}
+
+static void
+compress_rgba_unorm_block(int src_width, int src_height,
+                          const uint8_t *src, int src_rowstride,
+                          uint8_t *dst)
+{
+   int average_luminance, average_alpha;
+   uint8_t endpoints[2][4];
+   struct bit_writer writer;
+   int component, endpoint;
+
+   get_average_luminance_alpha_unorm(src_width, src_height, src, src_rowstride,
+                                     &average_luminance, &average_alpha);
+   get_rgba_endpoints_unorm(src_width, src_height, src, src_rowstride,
+                            average_luminance, average_alpha,
+                            endpoints);
+
+   writer.dst = dst;
+   writer.pos = 0;
+   writer.buf = 0;
+
+   write_bits(&writer, 5, 0x10); /* mode 4 */
+   write_bits(&writer, 2, 0); /* rotation 0 */
+   write_bits(&writer, 1, 0); /* index selection bit */
+
+   /* Write the color endpoints */
+   for (component = 0; component < 3; component++)
+      for (endpoint = 0; endpoint < 2; endpoint++)
+         write_bits(&writer, 5, endpoints[endpoint][component] >> 3);
+
+   /* Write the alpha endpoints */
+   for (endpoint = 0; endpoint < 2; endpoint++)
+      write_bits(&writer, 6, endpoints[endpoint][3] >> 2);
+
+   write_rgb_indices_unorm(&writer,
+                           src_width, src_height,
+                           src, src_rowstride,
+                           endpoints);
+   write_alpha_indices_unorm(&writer,
+                             src_width, src_height,
+                             src, src_rowstride,
+                             endpoints);
+}
+
+static void
+compress_rgba_unorm(int width, int height,
+                    const uint8_t *src, int src_rowstride,
+                    uint8_t *dst, int dst_rowstride)
+{
+   int dst_row_diff;
+   int y, x;
+
+   if (dst_rowstride >= width * 4)
+      dst_row_diff = dst_rowstride - ((width + 3) & ~3) * 4;
+   else
+      dst_row_diff = 0;
+
+   for (y = 0; y < height; y += BLOCK_SIZE) {
+      for (x = 0; x < width; x += BLOCK_SIZE) {
+         compress_rgba_unorm_block(MIN2(width - x, BLOCK_SIZE),
+                                   MIN2(height - y, BLOCK_SIZE),
+                                   src + x * 4 + y * src_rowstride,
+                                   src_rowstride,
+                                   dst);
+         dst += BLOCK_BYTES;
+      }
+      dst += dst_row_diff;
+   }
+}
+
+static float
+get_average_luminance_float(int width, int height,
+                            const float *src, int src_rowstride)
+{
+   float luminance_sum = 0;
+   int y, x;
+
+   for (y = 0; y < height; y++) {
+      for (x = 0; x < width; x++) {
+         luminance_sum += src[0] + src[1] + src[2];
+         src += 3;
+      }
+      src += (src_rowstride - width * 3 * sizeof (float)) / sizeof (float);
+   }
+
+   return luminance_sum / (width * height);
+}
+
+static float
+clamp_value(float value, bool is_signed)
+{
+   if (value > 65504.0f)
+      return 65504.0f;
+
+   if (is_signed) {
+      if (value < -65504.0f)
+         return -65504.0f;
+      else
+         return value;
+   }
+
+   if (value < 0.0f)
+      return 0.0f;
+
+   return value;
+}
+
+static void
+get_endpoints_float(int width, int height,
+                    const float *src, int src_rowstride,
+                    float average_luminance, float endpoints[][3],
+                    bool is_signed)
+{
+   float endpoint_luminances[2];
+   float midpoint;
+   float sums[2][3];
+   int endpoint, component;
+   float luminance;
+   float temp[3];
+   const float *p = src;
+   int left_endpoint_count = 0;
+   int y, x, i;
+
+   memset(sums, 0, sizeof sums);
+
+   for (y = 0; y < height; y++) {
+      for (x = 0; x < width; x++) {
+         luminance = p[0] + p[1] + p[2];
+         if (luminance < average_luminance) {
+            endpoint = 0;
+            left_endpoint_count++;
+         } else {
+            endpoint = 1;
+         }
+         for (i = 0; i < 3; i++)
+            sums[endpoint][i] += p[i];
+
+         p += 3;
+      }
+
+      p += (src_rowstride - width * 3 * sizeof (float)) / sizeof (float);
+   }
+
+   if (left_endpoint_count == 0 ||
+       left_endpoint_count == width * height) {
+      for (i = 0; i < 3; i++)
+         endpoints[0][i] = endpoints[1][i] =
+            (sums[0][i] + sums[1][i]) / (width * height);
+   } else {
+      for (i = 0; i < 3; i++) {
+         endpoints[0][i] = sums[0][i] / left_endpoint_count;
+         endpoints[1][i] = sums[1][i] / (width * height - left_endpoint_count);
+      }
+   }
+
+   /* Clamp the endpoints to the range of a half float and strip out
+    * infinities */
+   for (endpoint = 0; endpoint < 2; endpoint++) {
+      for (component = 0; component < 3; component++) {
+         endpoints[endpoint][component] =
+            clamp_value(endpoints[endpoint][component], is_signed);
+      }
+   }
+
+   /* We may need to swap the endpoints to ensure the most-significant bit of
+    * the first index is zero */
+
+   for (endpoint = 0; endpoint < 2; endpoint++) {
+      endpoint_luminances[endpoint] =
+         endpoints[endpoint][0] +
+         endpoints[endpoint][1] +
+         endpoints[endpoint][2];
+   }
+   midpoint = (endpoint_luminances[0] + endpoint_luminances[1]) / 2.0f;
+
+   if ((src[0] + src[1] + src[2] <= midpoint) !=
+       (endpoint_luminances[0] <= midpoint)) {
+      memcpy(temp, endpoints[0], sizeof temp);
+      memcpy(endpoints[0], endpoints[1], sizeof temp);
+      memcpy(endpoints[1], temp, sizeof temp);
+   }
+}
+
+static void
+write_rgb_indices_float(struct bit_writer *writer,
+                        int src_width, int src_height,
+                        const float *src, int src_rowstride,
+                        float endpoints[][3])
+{
+   float luminance;
+   float endpoint_luminances[2];
+   int endpoint;
+   int index;
+   int y, x;
+
+   for (endpoint = 0; endpoint < 2; endpoint++) {
+      endpoint_luminances[endpoint] =
+         endpoints[endpoint][0] +
+         endpoints[endpoint][1] +
+         endpoints[endpoint][2];
+   }
+
+   /* If the endpoints have the same luminance then we'll just use index 0 for
+    * all of the texels */
+   if (endpoint_luminances[0] == endpoint_luminances[1]) {
+      write_bits(writer, BLOCK_SIZE * BLOCK_SIZE * 4 - 1, 0);
+      return;
+   }
+
+   for (y = 0; y < src_height; y++) {
+      for (x = 0; x < src_width; x++) {
+         luminance = src[0] + src[1] + src[2];
+
+         index = ((luminance - endpoint_luminances[0]) * 15 /
+                  (endpoint_luminances[1] - endpoint_luminances[0]));
+         if (index < 0)
+            index = 0;
+         else if (index > 15)
+            index = 15;
+
+         assert(x != 0 || y != 0 || index < 8);
+
+         write_bits(writer, (x == 0 && y == 0) ? 3 : 4, index);
+
+         src += 3;
+      }
+
+      /* Pad the indices out to the block size */
+      if (src_width < BLOCK_SIZE)
+         write_bits(writer, 4 * (BLOCK_SIZE - src_width), 0);
+
+      src += (src_rowstride - src_width * 3 * sizeof (float)) / sizeof (float);
+   }
+
+   /* Pad the indices out to the block size */
+   if (src_height < BLOCK_SIZE)
+      write_bits(writer, 4 * BLOCK_SIZE * (BLOCK_SIZE - src_height), 0);
+}
+
+static int
+get_endpoint_value(float value, bool is_signed)
+{
+   bool sign = false;
+   int half;
+
+   if (is_signed) {
+      half = _mesa_float_to_half(value);
+
+      if (half & 0x8000) {
+         half &= 0x7fff;
+         sign = true;
+      }
+
+      half = (32 * half / 31) >> 6;
+
+      if (sign)
+         half = -half & ((1 << 10) - 1);
+
+      return half;
+   } else {
+      if (value <= 0.0f)
+         return 0;
+
+      half = _mesa_float_to_half(value);
+
+      return (64 * half / 31) >> 6;
+   }
+}
+
+static void
+compress_rgb_float_block(int src_width, int src_height,
+                         const float *src, int src_rowstride,
+                         uint8_t *dst,
+                         bool is_signed)
+{
+   float average_luminance;
+   float endpoints[2][3];
+   struct bit_writer writer;
+   int component, endpoint;
+   int endpoint_value;
+
+   average_luminance =
+      get_average_luminance_float(src_width, src_height, src, src_rowstride);
+   get_endpoints_float(src_width, src_height, src, src_rowstride,
+                       average_luminance, endpoints, is_signed);
+
+   writer.dst = dst;
+   writer.pos = 0;
+   writer.buf = 0;
+
+   write_bits(&writer, 5, 3); /* mode 3 */
+
+   /* Write the endpoints */
+   for (endpoint = 0; endpoint < 2; endpoint++) {
+      for (component = 0; component < 3; component++) {
+         endpoint_value =
+            get_endpoint_value(endpoints[endpoint][component], is_signed);
+         write_bits(&writer, 10, endpoint_value);
+      }
+   }
+
+   write_rgb_indices_float(&writer,
+                           src_width, src_height,
+                           src, src_rowstride,
+                           endpoints);
+}
+
+static void
+compress_rgb_float(int width, int height,
+                   const float *src, int src_rowstride,
+                   uint8_t *dst, int dst_rowstride,
+                   bool is_signed)
+{
+   int dst_row_diff;
+   int y, x;
+
+   if (dst_rowstride >= width * 4)
+      dst_row_diff = dst_rowstride - ((width + 3) & ~3) * 4;
+   else
+      dst_row_diff = 0;
+
+   for (y = 0; y < height; y += BLOCK_SIZE) {
+      for (x = 0; x < width; x += BLOCK_SIZE) {
+         compress_rgb_float_block(MIN2(width - x, BLOCK_SIZE),
+                                  MIN2(height - y, BLOCK_SIZE),
+                                  src + x * 3 +
+                                  y * src_rowstride / sizeof (float),
+                                  src_rowstride,
+                                  dst,
+                                  is_signed);
+         dst += BLOCK_BYTES;
+      }
+      dst += dst_row_diff;
+   }
+}
+
+#endif
diff --git a/lib/mesa/src/mesa/main/texcompress_etc.c b/lib/mesa/src/mesa/main/texcompress_etc.c
index d46501098..b39ab33d3 100644
--- a/lib/mesa/src/mesa/main/texcompress_etc.c
+++ b/lib/mesa/src/mesa/main/texcompress_etc.c
@@ -41,6 +41,7 @@
 #include "texcompress.h"
 #include "texcompress_etc.h"
 #include "texstore.h"
+#include "config.h"
 #include "macros.h"
 #include "format_unpack.h"
 #include "util/format_srgb.h"
@@ -105,7 +106,7 @@ static const int etc2_modifier_tables_non_opaque[8][4] = {
 #undef UINT8_TYPE
 
 GLboolean
-_mesa_texstore_etc1_rgb8(TEXSTORE_PARAMS)
+_mesa_texstore_etc1_rgb8(UNUSED_TEXSTORE_PARAMS)
 {
    /* GL_ETC1_RGB8_OES is only valid in glCompressedTexImage2D */
    assert(0);
@@ -718,7 +719,8 @@ etc2_unpack_srgb8(uint8_t *dst_row,
                   const uint8_t *src_row,
                   unsigned src_stride,
                   unsigned width,
-                  unsigned height)
+		  unsigned height,
+		  bool bgra)
 {
    const unsigned bw = 4, bh = 4, bs = 8, comps = 4;
    struct etc2_block block;
@@ -740,11 +742,14 @@ etc2_unpack_srgb8(uint8_t *dst_row,
             for (i = 0; i < w; i++) {
                etc2_rgb8_fetch_texel(&block, i, j, dst,
                                      false /* punchthrough_alpha */);
-               /* Convert to MESA_FORMAT_B8G8R8A8_SRGB */
-               tmp = dst[0];
-               dst[0] = dst[2];
-               dst[2] = tmp;
-               dst[3] = 255;
+
+	       if (bgra) {
+		  /* Convert to MESA_FORMAT_B8G8R8A8_SRGB */
+		  tmp = dst[0];
+		  dst[0] = dst[2];
+		  dst[2] = tmp;
+		  dst[3] = 255;
+	       }
 
                dst += comps;
             }
@@ -800,7 +805,8 @@ etc2_unpack_srgb8_alpha8(uint8_t *dst_row,
                          const uint8_t *src_row,
                          unsigned src_stride,
                          unsigned width,
-                         unsigned height)
+			 unsigned height,
+			 bool bgra)
 {
    /* If internalformat is COMPRESSED_SRGB8_ALPHA8_ETC2_EAC, each 4 × 4 block
     * of RGBA8888 information is compressed to 128 bits. To decode a block, the
@@ -824,11 +830,13 @@ etc2_unpack_srgb8_alpha8(uint8_t *dst_row,
             for (i = 0; i < w; i++) {
                etc2_rgba8_fetch_texel(&block, i, j, dst);
 
-               /* Convert to MESA_FORMAT_B8G8R8A8_SRGB */
-               tmp = dst[0];
-               dst[0] = dst[2];
-               dst[2] = tmp;
-               dst[3] = dst[3];
+	       if (bgra) {
+		  /* Convert to MESA_FORMAT_B8G8R8A8_SRGB */
+		  tmp = dst[0];
+		  dst[0] = dst[2];
+		  dst[2] = tmp;
+		  dst[3] = dst[3];
+	       }
 
                dst += comps;
             }
@@ -1057,7 +1065,8 @@ etc2_unpack_srgb8_punchthrough_alpha1(uint8_t *dst_row,
                                      const uint8_t *src_row,
                                      unsigned src_stride,
                                      unsigned width,
-                                     unsigned height)
+				     unsigned height,
+				     bool bgra)
 {
    const unsigned bw = 4, bh = 4, bs = 8, comps = 4;
    struct etc2_block block;
@@ -1077,11 +1086,14 @@ etc2_unpack_srgb8_punchthrough_alpha1(uint8_t *dst_row,
             for (i = 0; i < w; i++) {
                etc2_rgb8_fetch_texel(&block, i, j, dst,
                                      true /* punchthrough_alpha */);
-               /* Convert to MESA_FORMAT_B8G8R8A8_SRGB */
-               tmp = dst[0];
-               dst[0] = dst[2];
-               dst[2] = tmp;
-               dst[3] = dst[3];
+
+	       if (bgra) {
+		  /* Convert to MESA_FORMAT_B8G8R8A8_SRGB */
+		  tmp = dst[0];
+		  dst[0] = dst[2];
+		  dst[2] = tmp;
+		  dst[3] = dst[3];
+	       }
 
                dst += comps;
             }
@@ -1097,7 +1109,7 @@ etc2_unpack_srgb8_punchthrough_alpha1(uint8_t *dst_row,
 /* ETC2 texture formats are valid in glCompressedTexImage2D and
  * glCompressedTexSubImage2D functions */
 GLboolean
-_mesa_texstore_etc2_rgb8(TEXSTORE_PARAMS)
+_mesa_texstore_etc2_rgb8(UNUSED_TEXSTORE_PARAMS)
 {
    assert(0);
 
@@ -1105,7 +1117,7 @@ _mesa_texstore_etc2_rgb8(TEXSTORE_PARAMS)
 }
 
 GLboolean
-_mesa_texstore_etc2_srgb8(TEXSTORE_PARAMS)
+_mesa_texstore_etc2_srgb8(UNUSED_TEXSTORE_PARAMS)
 {
    assert(0);
 
@@ -1113,7 +1125,7 @@ _mesa_texstore_etc2_srgb8(TEXSTORE_PARAMS)
 }
 
 GLboolean
-_mesa_texstore_etc2_rgba8_eac(TEXSTORE_PARAMS)
+_mesa_texstore_etc2_rgba8_eac(UNUSED_TEXSTORE_PARAMS)
 {
    assert(0);
 
@@ -1121,7 +1133,7 @@ _mesa_texstore_etc2_rgba8_eac(TEXSTORE_PARAMS)
 }
 
 GLboolean
-_mesa_texstore_etc2_srgb8_alpha8_eac(TEXSTORE_PARAMS)
+_mesa_texstore_etc2_srgb8_alpha8_eac(UNUSED_TEXSTORE_PARAMS)
 {
    assert(0);
 
@@ -1129,7 +1141,7 @@ _mesa_texstore_etc2_srgb8_alpha8_eac(TEXSTORE_PARAMS)
 }
 
 GLboolean
-_mesa_texstore_etc2_r11_eac(TEXSTORE_PARAMS)
+_mesa_texstore_etc2_r11_eac(UNUSED_TEXSTORE_PARAMS)
 {
    assert(0);
 
@@ -1137,7 +1149,7 @@ _mesa_texstore_etc2_r11_eac(TEXSTORE_PARAMS)
 }
 
 GLboolean
-_mesa_texstore_etc2_signed_r11_eac(TEXSTORE_PARAMS)
+_mesa_texstore_etc2_signed_r11_eac(UNUSED_TEXSTORE_PARAMS)
 {
    assert(0);
 
@@ -1145,7 +1157,7 @@ _mesa_texstore_etc2_signed_r11_eac(TEXSTORE_PARAMS)
 }
 
 GLboolean
-_mesa_texstore_etc2_rg11_eac(TEXSTORE_PARAMS)
+_mesa_texstore_etc2_rg11_eac(UNUSED_TEXSTORE_PARAMS)
 {
    assert(0);
 
@@ -1153,7 +1165,7 @@ _mesa_texstore_etc2_rg11_eac(TEXSTORE_PARAMS)
 }
 
 GLboolean
-_mesa_texstore_etc2_signed_rg11_eac(TEXSTORE_PARAMS)
+_mesa_texstore_etc2_signed_rg11_eac(UNUSED_TEXSTORE_PARAMS)
 {
    assert(0);
 
@@ -1161,7 +1173,7 @@ _mesa_texstore_etc2_signed_rg11_eac(TEXSTORE_PARAMS)
 }
 
 GLboolean
-_mesa_texstore_etc2_rgb8_punchthrough_alpha1(TEXSTORE_PARAMS)
+_mesa_texstore_etc2_rgb8_punchthrough_alpha1(UNUSED_TEXSTORE_PARAMS)
 {
    assert(0);
 
@@ -1169,7 +1181,7 @@ _mesa_texstore_etc2_rgb8_punchthrough_alpha1(TEXSTORE_PARAMS)
 }
 
 GLboolean
-_mesa_texstore_etc2_srgb8_punchthrough_alpha1(TEXSTORE_PARAMS)
+_mesa_texstore_etc2_srgb8_punchthrough_alpha1(UNUSED_TEXSTORE_PARAMS)
 {
    assert(0);
 
@@ -1205,7 +1217,8 @@ _mesa_unpack_etc2_format(uint8_t *dst_row,
                          unsigned src_stride,
                          unsigned src_width,
                          unsigned src_height,
-                         mesa_format format)
+			 mesa_format format,
+			 bool bgra)
 {
    if (format == MESA_FORMAT_ETC2_RGB8)
       etc2_unpack_rgb8(dst_row, dst_stride,
@@ -1214,7 +1227,7 @@ _mesa_unpack_etc2_format(uint8_t *dst_row,
    else if (format == MESA_FORMAT_ETC2_SRGB8)
       etc2_unpack_srgb8(dst_row, dst_stride,
                         src_row, src_stride,
-                        src_width, src_height);
+			src_width, src_height, bgra);
    else if (format == MESA_FORMAT_ETC2_RGBA8_EAC)
       etc2_unpack_rgba8(dst_row, dst_stride,
                         src_row, src_stride,
@@ -1222,7 +1235,7 @@ _mesa_unpack_etc2_format(uint8_t *dst_row,
    else if (format == MESA_FORMAT_ETC2_SRGB8_ALPHA8_EAC)
       etc2_unpack_srgb8_alpha8(dst_row, dst_stride,
                                src_row, src_stride,
-                               src_width, src_height);
+			       src_width, src_height, bgra);
    else if (format == MESA_FORMAT_ETC2_R11_EAC)
       etc2_unpack_r11(dst_row, dst_stride,
                       src_row, src_stride,
@@ -1246,7 +1259,7 @@ _mesa_unpack_etc2_format(uint8_t *dst_row,
    else if (format == MESA_FORMAT_ETC2_SRGB8_PUNCHTHROUGH_ALPHA1)
       etc2_unpack_srgb8_punchthrough_alpha1(dst_row, dst_stride,
                                             src_row, src_stride,
-                                            src_width, src_height);
+					    src_width, src_height, bgra);
 }
 
 
diff --git a/lib/mesa/src/mesa/main/texcompress_etc.h b/lib/mesa/src/mesa/main/texcompress_etc.h
index 319b7bea7..2c764b88b 100644
--- a/lib/mesa/src/mesa/main/texcompress_etc.h
+++ b/lib/mesa/src/mesa/main/texcompress_etc.h
@@ -77,7 +77,8 @@ _mesa_unpack_etc2_format(uint8_t *dst_row,
                          unsigned src_stride,
                          unsigned src_width,
                          unsigned src_height,
-                         mesa_format format);
+			 mesa_format format,
+			 bool bgra);
 
 compressed_fetch_func
 _mesa_get_etc_fetch_func(mesa_format format);
diff --git a/lib/mesa/src/mesa/main/texcompress_fxt1.c b/lib/mesa/src/mesa/main/texcompress_fxt1.c
index c5646fbd7..19df6baf3 100644
--- a/lib/mesa/src/mesa/main/texcompress_fxt1.c
+++ b/lib/mesa/src/mesa/main/texcompress_fxt1.c
@@ -29,6 +29,7 @@
  */
 
 
+#include "errors.h"
 #include "glheader.h"
 #include "imports.h"
 #include "image.h"
@@ -37,6 +38,7 @@
 #include "texcompress.h"
 #include "texcompress_fxt1.h"
 #include "texstore.h"
+#include "mtypes.h"
 
 
 static void
diff --git a/lib/mesa/src/mesa/main/texcompress_rgtc.c b/lib/mesa/src/mesa/main/texcompress_rgtc.c
index 8cab7a56b..843b42e54 100644
--- a/lib/mesa/src/mesa/main/texcompress_rgtc.c
+++ b/lib/mesa/src/mesa/main/texcompress_rgtc.c
@@ -33,6 +33,7 @@
  */
 
 
+#include "config.h"
 #include "glheader.h"
 #include "imports.h"
 #include "image.h"
diff --git a/lib/mesa/src/mesa/main/texcompress_s3tc_tmp.h b/lib/mesa/src/mesa/main/texcompress_s3tc_tmp.h
index 61630f247..dbf8c871b 100644
--- a/lib/mesa/src/mesa/main/texcompress_s3tc_tmp.h
+++ b/lib/mesa/src/mesa/main/texcompress_s3tc_tmp.h
@@ -22,6 +22,9 @@
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
+#ifndef TEXCOMPRESS_S3TC_TMP_H
+#define TEXCOMPRESS_S3TC_TMP_H
+
 #ifdef __APPLE__
 #include <OpenGL/gl.h>
 #else
@@ -182,8 +185,8 @@ static void fetch_2d_texel_rgba_dxt5(GLint srcRowStride, const GLubyte *pixdata,
 
 #define ALPHACUT 127
 
-static void fancybasecolorsearch( GLubyte *blkaddr, GLubyte srccolors[4][4][4], GLubyte *bestcolor[2],
-                           GLint numxpixels, GLint numypixels, GLint type, GLboolean haveAlpha)
+static void fancybasecolorsearch( UNUSED GLubyte *blkaddr, GLubyte srccolors[4][4][4], GLubyte *bestcolor[2],
+                           GLint numxpixels, GLint numypixels, UNUSED GLint type, UNUSED GLboolean haveAlpha)
 {
    /* use same luminance-weighted distance metric to determine encoding as for finding the base colors */
 
@@ -987,3 +990,5 @@ static void tx_compress_dxtn(GLint srccomps, GLint width, GLint height, const GL
       return;
    }
 }
+
+#endif
diff --git a/lib/mesa/src/mesa/main/texgetimage.h b/lib/mesa/src/mesa/main/texgetimage.h
index 63c75eb93..3928e7abb 100644
--- a/lib/mesa/src/mesa/main/texgetimage.h
+++ b/lib/mesa/src/mesa/main/texgetimage.h
@@ -44,14 +44,6 @@ _mesa_GetTexSubImage_sw(struct gl_context *ctx,
                         struct gl_texture_image *texImage);
 
 extern void
-_mesa_GetCompressedTexSubImage_sw(struct gl_context *ctx,
-                                  struct gl_texture_image *texImage,
-                                  GLint xoffset, GLint yoffset,
-                                  GLint zoffset, GLsizei width,
-                                  GLint height, GLint depth,
-                                  GLvoid *data);
-
-extern void
 _mesa_get_compressed_texture_image( struct gl_context *ctx,
                                     struct gl_texture_object *texObj,
                                     struct gl_texture_image *texImage,
diff --git a/lib/mesa/src/mesa/main/texstore.h b/lib/mesa/src/mesa/main/texstore.h
index f08dc08ed..2fef7ba7d 100644
--- a/lib/mesa/src/mesa/main/texstore.h
+++ b/lib/mesa/src/mesa/main/texstore.h
@@ -36,9 +36,13 @@
 #define TEXSTORE_H
 
 
-#include "mtypes.h"
+#include "glheader.h"
 #include "formats.h"
+#include "util/macros.h"
 
+struct gl_context;
+struct gl_pixelstore_attrib;
+struct gl_texture_image;
 
 /**
  * This macro defines the (many) parameters to the texstore functions.
@@ -56,8 +60,8 @@
  */
 #define TEXSTORE_PARAMS \
 	struct gl_context *ctx, GLuint dims, \
-	GLenum baseInternalFormat, \
-	mesa_format dstFormat, \
+        MAYBE_UNUSED GLenum baseInternalFormat, \
+        MAYBE_UNUSED mesa_format dstFormat, \
         GLint dstRowStride, \
         GLubyte **dstSlices, \
 	GLint srcWidth, GLint srcHeight, GLint srcDepth, \
@@ -65,6 +69,20 @@
 	const GLvoid *srcAddr, \
 	const struct gl_pixelstore_attrib *srcPacking
 
+/* This macro must be kept in sync with TEXSTORE_PARAMS.  It is used in the
+ * few places where none of the parameters are used (i.e., the ETC texstore
+ * functions).
+ */
+#define UNUSED_TEXSTORE_PARAMS                                          \
+        UNUSED struct gl_context *ctx, UNUSED GLuint dims,              \
+        UNUSED GLenum baseInternalFormat,                               \
+        UNUSED mesa_format dstFormat,                                   \
+        UNUSED GLint dstRowStride,                                      \
+        UNUSED GLubyte **dstSlices,                                     \
+        UNUSED GLint srcWidth, UNUSED GLint srcHeight, UNUSED GLint srcDepth, \
+        UNUSED GLenum srcFormat, UNUSED GLenum srcType,                 \
+        UNUSED const GLvoid *srcAddr,                                   \
+        UNUSED const struct gl_pixelstore_attrib *srcPacking
 
 extern GLboolean
 _mesa_texstore(TEXSTORE_PARAMS);
diff --git a/lib/mesa/src/mesa/main/version.h b/lib/mesa/src/mesa/main/version.h
index adfec6f82..4469509c0 100644
--- a/lib/mesa/src/mesa/main/version.h
+++ b/lib/mesa/src/mesa/main/version.h
@@ -27,8 +27,13 @@
 #ifndef VERSION_H
 #define VERSION_H
 
-#include "mtypes.h"
+#include <stdbool.h>
+#include "glheader.h"
+#include "menums.h"
 
+struct gl_context;
+struct gl_constants;
+struct gl_extensions;
 
 extern GLuint
 _mesa_get_version(const struct gl_extensions *extensions,
diff --git a/lib/mesa/src/mesa/main/vtxfmt.h b/lib/mesa/src/mesa/main/vtxfmt.h
index 4f8bc9cdf..20fc46670 100644
--- a/lib/mesa/src/mesa/main/vtxfmt.h
+++ b/lib/mesa/src/mesa/main/vtxfmt.h
@@ -33,12 +33,14 @@
 #ifndef _VTXFMT_H_
 #define _VTXFMT_H_
 
-#include "mtypes.h"
+#include "dd.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
+struct gl_context;
+
 extern void _mesa_install_exec_vtxfmt( struct gl_context *ctx, const GLvertexformat *vfmt );
 extern void _mesa_install_save_vtxfmt( struct gl_context *ctx, const GLvertexformat *vfmt );
 extern void _mesa_initialize_vbo_vtxfmt(struct gl_context *ctx);
diff --git a/lib/mesa/src/mesa/meson.build b/lib/mesa/src/mesa/meson.build
new file mode 100644
index 000000000..27c8c401c
--- /dev/null
+++ b/lib/mesa/src/mesa/meson.build
@@ -0,0 +1,754 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+subdir('program')
+subdir('main')
+
+# files shared between classic mesa and gallium mesa
+files_libmesa_common = files(
+  'program/arbprogparse.c',
+  'program/arbprogparse.h',
+  'program/ir_to_mesa.cpp',
+  'program/ir_to_mesa.h',
+  'program/prog_cache.c',
+  'program/prog_cache.h',
+  'program/prog_execute.c',
+  'program/prog_execute.h',
+  'program/prog_instruction.c',
+  'program/prog_instruction.h',
+  'program/prog_noise.c',
+  'program/prog_noise.h',
+  'program/prog_opt_constant_fold.c',
+  'program/prog_optimize.c',
+  'program/prog_optimize.h',
+  'program/prog_parameter.c',
+  'program/prog_parameter.h',
+  'program/prog_parameter_layout.c',
+  'program/prog_parameter_layout.h',
+  'program/prog_print.c',
+  'program/prog_print.h',
+  'program/program.c',
+  'program/program.h',
+  'program/programopt.c',
+  'program/programopt.h',
+  'program/program_parse_extra.c',
+  'program/program_parser.h',
+  'program/prog_statevars.c',
+  'program/prog_statevars.h',
+  'program/symbol_table.c',
+  'program/symbol_table.h',
+  'program/prog_to_nir.c',
+  'program/prog_to_nir.h',
+  'main/accum.c',
+  'main/accum.h',
+  'main/api_arrayelt.c',
+  'main/api_arrayelt.h',
+  'main/api_exec.h',
+  'main/api_loopback.c',
+  'main/api_loopback.h',
+  'main/arbprogram.c',
+  'main/arbprogram.h',
+  'main/arrayobj.c',
+  'main/arrayobj.h',
+  'main/atifragshader.c',
+  'main/atifragshader.h',
+  'main/attrib.c',
+  'main/attrib.h',
+  'main/barrier.c',
+  'main/barrier.h',
+  'main/bbox.c',
+  'main/bbox.h',
+  'main/blend.c',
+  'main/blend.h',
+  'main/blit.c',
+  'main/blit.h',
+  'main/bufferobj.c',
+  'main/bufferobj.h',
+  'main/buffers.c',
+  'main/buffers.h',
+  'main/clear.c',
+  'main/clear.h',
+  'main/clip.c',
+  'main/clip.h',
+  'main/colormac.h',
+  'main/colortab.c',
+  'main/colortab.h',
+  'main/compute.c',
+  'main/compute.h',
+  'main/compiler.h',
+  'main/condrender.c',
+  'main/condrender.h',
+  'main/config.h',
+  'main/conservativeraster.c',
+  'main/conservativeraster.h',
+  'main/context.c',
+  'main/context.h',
+  'main/convolve.c',
+  'main/convolve.h',
+  'main/copyimage.c',
+  'main/copyimage.h',
+  'main/cpuinfo.c',
+  'main/cpuinfo.h',
+  'main/dd.h',
+  'main/debug.c',
+  'main/debug.h',
+  'main/debug_output.c',
+  'main/debug_output.h',
+  'main/depth.c',
+  'main/depth.h',
+  'main/dlist.c',
+  'main/dlist.h',
+  'main/draw.c',
+  'main/draw.h',
+  'main/drawpix.c',
+  'main/drawpix.h',
+  'main/drawtex.c',
+  'main/drawtex.h',
+  'main/draw_validate.c',
+  'main/draw_validate.h',
+  'main/enable.c',
+  'main/enable.h',
+  'main/enums.h',
+  'main/errors.c',
+  'main/errors.h',
+  'main/eval.c',
+  'main/eval.h',
+  'main/execmem.c',
+  'main/extensions.c',
+  'main/extensions.h',
+  'main/extensions_table.c',
+  'main/extensions_table.h',
+  'main/externalobjects.c',
+  'main/externalobjects.h',
+  'main/fbobject.c',
+  'main/fbobject.h',
+  'main/feedback.c',
+  'main/feedback.h',
+  'main/ff_fragment_shader.cpp',
+  'main/ffvertex_prog.c',
+  'main/ffvertex_prog.h',
+  'main/fog.c',
+  'main/fog.h',
+  'main/format_pack.h',
+  'main/format_unpack.h',
+  'main/formatquery.c',
+  'main/formatquery.h',
+  'main/formats.c',
+  'main/formats.h',
+  'main/format_utils.c',
+  'main/format_utils.h',
+  'main/framebuffer.c',
+  'main/framebuffer.h',
+  'main/get.c',
+  'main/get.h',
+  'main/genmipmap.c',
+  'main/genmipmap.h',
+  'main/getstring.c',
+  'main/glformats.c',
+  'main/glformats.h',
+  'main/glspirv.c',
+  'main/glspirv.h',
+  'main/glthread.c',
+  'main/glthread.h',
+  'main/glheader.h',
+  'main/hash.c',
+  'main/hash.h',
+  'main/hint.c',
+  'main/hint.h',
+  'main/histogram.c',
+  'main/histogram.h',
+  'main/image.c',
+  'main/image.h',
+  'main/imports.c',
+  'main/imports.h',
+  'main/light.c',
+  'main/light.h',
+  'main/lines.c',
+  'main/lines.h',
+  'main/macros.h',
+  'main/marshal.c',
+  'main/marshal.h',
+  'main/matrix.c',
+  'main/matrix.h',
+  'main/mipmap.c',
+  'main/mipmap.h',
+  'main/mm.c',
+  'main/mm.h',
+  'main/menums.h',
+  'main/mtypes.h',
+  'main/multisample.c',
+  'main/multisample.h',
+  'main/objectlabel.c',
+  'main/objectlabel.h',
+  'main/objectpurge.c',
+  'main/objectpurge.h',
+  'main/pack.c',
+  'main/pack.h',
+  'main/pbo.c',
+  'main/pbo.h',
+  'main/performance_monitor.c',
+  'main/performance_monitor.h',
+  'main/performance_query.c',
+  'main/performance_query.h',
+  'main/pipelineobj.c',
+  'main/pipelineobj.h',
+  'main/pixel.c',
+  'main/pixel.h',
+  'main/pixelstore.c',
+  'main/pixelstore.h',
+  'main/pixeltransfer.c',
+  'main/pixeltransfer.h',
+  'main/points.c',
+  'main/points.h',
+  'main/polygon.c',
+  'main/polygon.h',
+  'main/program_binary.c',
+  'main/program_binary.h',
+  'main/program_resource.c',
+  'main/program_resource.h',
+  'main/querymatrix.c',
+  'main/querymatrix.h',
+  'main/queryobj.c',
+  'main/queryobj.h',
+  'main/rastpos.c',
+  'main/rastpos.h',
+  'main/readpix.c',
+  'main/readpix.h',
+  'main/remap.c',
+  'main/remap.h',
+  'main/renderbuffer.c',
+  'main/renderbuffer.h',
+  'main/robustness.c',
+  'main/samplerobj.c',
+  'main/samplerobj.h',
+  'main/scissor.c',
+  'main/scissor.h',
+  'main/shaderapi.c',
+  'main/shaderapi.h',
+  'main/shaderimage.c',
+  'main/shaderimage.h',
+  'main/shaderobj.c',
+  'main/shaderobj.h',
+  'main/shader_query.cpp',
+  'main/shared.c',
+  'main/shared.h',
+  'main/state.c',
+  'main/state.h',
+  'main/stencil.c',
+  'main/stencil.h',
+  'main/syncobj.c',
+  'main/syncobj.h',
+  'main/texcompress.c',
+  'main/texcompress_astc.cpp',
+  'main/texcompress_astc.h',
+  'main/texcompress_bptc.c',
+  'main/texcompress_bptc.h',
+  'main/texcompress_cpal.c',
+  'main/texcompress_cpal.h',
+  'main/texcompress_etc.c',
+  'main/texcompress_etc.h',
+  'main/texcompress_etc_tmp.h',
+  'main/texcompress_fxt1.c',
+  'main/texcompress_fxt1.h',
+  'main/texcompress.h',
+  'main/texcompress_rgtc.c',
+  'main/texcompress_rgtc.h',
+  'main/texcompress_s3tc.c',
+  'main/texcompress_s3tc.h',
+  'main/texenv.c',
+  'main/texenv.h',
+  'main/texenvprogram.h',
+  'main/texformat.c',
+  'main/texformat.h',
+  'main/texgen.c',
+  'main/texgen.h',
+  'main/texgetimage.c',
+  'main/texgetimage.h',
+  'main/teximage.c',
+  'main/teximage.h',
+  'main/texobj.c',
+  'main/texobj.h',
+  'main/texparam.c',
+  'main/texparam.h',
+  'main/texstate.c',
+  'main/texstate.h',
+  'main/texstorage.c',
+  'main/texstorage.h',
+  'main/texstore.c',
+  'main/texstore.h',
+  'main/texturebindless.c',
+  'main/texturebindless.h',
+  'main/textureview.c',
+  'main/textureview.h',
+  'main/transformfeedback.c',
+  'main/transformfeedback.h',
+  'main/uniform_query.cpp',
+  'main/uniforms.c',
+  'main/uniforms.h',
+  'main/varray.c',
+  'main/varray.h',
+  'main/vdpau.c',
+  'main/vdpau.h',
+  'main/version.c',
+  'main/version.h',
+  'main/viewport.c',
+  'main/viewport.h',
+  'main/vtxfmt.c',
+  'main/vtxfmt.h',
+  'main/es1_conversion.c',
+  'main/es1_conversion.h',
+  'math/m_debug.h',
+  'math/m_debug_clip.c',
+  'math/m_debug_norm.c',
+  'math/m_debug_util.h',
+  'math/m_debug_xform.c',
+  'math/m_eval.c',
+  'math/m_eval.h',
+  'math/m_matrix.c',
+  'math/m_matrix.h',
+  'math/m_trans_tmp.h',
+  'math/m_translate.c',
+  'math/m_translate.h',
+  'math/m_vector.c',
+  'math/m_vector.h',
+  'vbo/vbo_attrib.h',
+  'vbo/vbo_attrib_tmp.h',
+  'vbo/vbo_context.c',
+  'vbo/vbo_exec_api.c',
+  'vbo/vbo_exec.c',
+  'vbo/vbo_exec_draw.c',
+  'vbo/vbo_exec_eval.c',
+  'vbo/vbo_exec.h',
+  'vbo/vbo.h',
+  'vbo/vbo_minmax_index.c',
+  'vbo/vbo_noop.c',
+  'vbo/vbo_noop.h',
+  'vbo/vbo_primitive_restart.c',
+  'vbo/vbo_save_api.c',
+  'vbo/vbo_save.c',
+  'vbo/vbo_save_draw.c',
+  'vbo/vbo_save.h',
+  'vbo/vbo_save_loopback.c',
+  'x86/common_x86.c',
+)
+
+# mesa files
+files_libmesa_classic = files(
+  'math/m_clip_tmp.h',
+  'math/m_copy_tmp.h',
+  'math/m_dotprod_tmp.h',
+  'math/m_norm_tmp.h',
+  'math/m_xform.c',
+  'math/m_xform.h',
+  'math/m_xform_tmp.h',
+  'tnl/t_context.c',
+  'tnl/t_context.h',
+  'tnl/t_draw.c',
+  'tnl/tnl.h',
+  'tnl/t_pipeline.c',
+  'tnl/t_pipeline.h',
+  'tnl/t_rebase.c',
+  'tnl/t_split.c',
+  'tnl/t_split_copy.c',
+  'tnl/t_split.h',
+  'tnl/t_split_inplace.c',
+  'tnl/t_vb_cliptmp.h',
+  'tnl/t_vb_fog.c',
+  'tnl/t_vb_light.c',
+  'tnl/t_vb_lighttmp.h',
+  'tnl/t_vb_normals.c',
+  'tnl/t_vb_points.c',
+  'tnl/t_vb_program.c',
+  'tnl/t_vb_render.c',
+  'tnl/t_vb_rendertmp.h',
+  'tnl/t_vb_texgen.c',
+  'tnl/t_vb_texmat.c',
+  'tnl/t_vb_vertex.c',
+  'tnl/t_vertex.c',
+  'tnl/t_vertex_generic.c',
+  'tnl/t_vertex.h',
+  'tnl/t_vertex_sse.c',
+  'tnl/t_vp_build.c',
+  'tnl/t_vp_build.h',
+  'swrast/s_aaline.c',
+  'swrast/s_aaline.h',
+  'swrast/s_aalinetemp.h',
+  'swrast/s_aatriangle.c',
+  'swrast/s_aatriangle.h',
+  'swrast/s_aatritemp.h',
+  'swrast/s_alpha.c',
+  'swrast/s_alpha.h',
+  'swrast/s_atifragshader.c',
+  'swrast/s_atifragshader.h',
+  'swrast/s_bitmap.c',
+  'swrast/s_blend.c',
+  'swrast/s_blend.h',
+  'swrast/s_blit.c',
+  'swrast/s_chan.h',
+  'swrast/s_clear.c',
+  'swrast/s_context.c',
+  'swrast/s_context.h',
+  'swrast/s_copypix.c',
+  'swrast/s_depth.c',
+  'swrast/s_depth.h',
+  'swrast/s_drawpix.c',
+  'swrast_setup/ss_tritmp.h',
+  'swrast_setup/ss_vb.h',
+  'swrast_setup/swrast_setup.h',
+  'swrast/s_feedback.c',
+  'swrast/s_feedback.h',
+  'swrast/s_fog.c',
+  'swrast/s_fog.h',
+  'swrast/s_fragprog.c',
+  'swrast/s_fragprog.h',
+  'swrast/s_lines.c',
+  'swrast/s_lines.h',
+  'swrast/s_linetemp.h',
+  'swrast/s_logic.c',
+  'swrast/s_logic.h',
+  'swrast/s_masking.c',
+  'swrast/s_masking.h',
+  'swrast/s_points.c',
+  'swrast/s_points.h',
+  'swrast/s_renderbuffer.c',
+  'swrast/s_renderbuffer.h',
+  'swrast/s_span.c',
+  'swrast/s_span.h',
+  'swrast/s_stencil.c',
+  'swrast/s_stencil.h',
+  'swrast/s_texcombine.c',
+  'swrast/s_texcombine.h',
+  'swrast/s_texfetch.c',
+  'swrast/s_texfetch.h',
+  'swrast/s_texfetch_tmp.h',
+  'swrast/s_texfilter.c',
+  'swrast/s_texfilter.h',
+  'swrast/s_texrender.c',
+  'swrast/s_texture.c',
+  'swrast/s_triangle.c',
+  'swrast/s_triangle.h',
+  'swrast/s_tritemp.h',
+  'swrast/swrast.h',
+  'swrast/s_zoom.c',
+  'swrast/s_zoom.h',
+  'swrast_setup/ss_context.c',
+  'swrast_setup/ss_context.h',
+  'swrast_setup/ss_triangle.c',
+  'swrast_setup/ss_triangle.h',
+  'drivers/common/driverfuncs.c',
+  'drivers/common/driverfuncs.h',
+  'drivers/common/meta_blit.c',
+  'drivers/common/meta_generate_mipmap.c',
+  'drivers/common/meta.c',
+  'drivers/common/meta.h',
+  'x86/x86_xform.c',
+  'x86/3dnow.c',
+  'x86/sse.c',
+  'x86/rtasm/x86sse.c',
+  'x86/rtasm/x86sse.h',
+  'sparc/sparc.c',
+  'x86-64/x86-64.c',
+)
+
+files_libmesa_gallium = files(
+  'state_tracker/st_atifs_to_tgsi.c',
+  'state_tracker/st_atifs_to_tgsi.h',
+  'state_tracker/st_atom_array.c',
+  'state_tracker/st_atom_atomicbuf.c',
+  'state_tracker/st_atom_blend.c',
+  'state_tracker/st_atom.c',
+  'state_tracker/st_atom_clip.c',
+  'state_tracker/st_atom_constbuf.c',
+  'state_tracker/st_atom_constbuf.h',
+  'state_tracker/st_atom_depth.c',
+  'state_tracker/st_atom_framebuffer.c',
+  'state_tracker/st_atom.h',
+  'state_tracker/st_atom_list.h',
+  'state_tracker/st_atom_image.c',
+  'state_tracker/st_atom_msaa.c',
+  'state_tracker/st_atom_pixeltransfer.c',
+  'state_tracker/st_atom_rasterizer.c',
+  'state_tracker/st_atom_sampler.c',
+  'state_tracker/st_atom_scissor.c',
+  'state_tracker/st_atom_shader.c',
+  'state_tracker/st_atom_shader.h',
+  'state_tracker/st_atom_stipple.c',
+  'state_tracker/st_atom_storagebuf.c',
+  'state_tracker/st_atom_tess.c',
+  'state_tracker/st_atom_texture.c',
+  'state_tracker/st_atom_viewport.c',
+  'state_tracker/st_cb_bitmap.c',
+  'state_tracker/st_cb_bitmap.h',
+  'state_tracker/st_cb_bitmap_shader.c',
+  'state_tracker/st_cb_blit.c',
+  'state_tracker/st_cb_blit.h',
+  'state_tracker/st_cb_bufferobjects.c',
+  'state_tracker/st_cb_bufferobjects.h',
+  'state_tracker/st_cb_clear.c',
+  'state_tracker/st_cb_clear.h',
+  'state_tracker/st_cb_compute.c',
+  'state_tracker/st_cb_compute.h',
+  'state_tracker/st_cb_condrender.c',
+  'state_tracker/st_cb_condrender.h',
+  'state_tracker/st_cb_copyimage.c',
+  'state_tracker/st_cb_copyimage.h',
+  'state_tracker/st_cb_drawpixels.c',
+  'state_tracker/st_cb_drawpixels.h',
+  'state_tracker/st_cb_drawpixels_shader.c',
+  'state_tracker/st_cb_drawtex.c',
+  'state_tracker/st_cb_drawtex.h',
+  'state_tracker/st_cb_eglimage.c',
+  'state_tracker/st_cb_eglimage.h',
+  'state_tracker/st_cb_fbo.c',
+  'state_tracker/st_cb_fbo.h',
+  'state_tracker/st_cb_feedback.c',
+  'state_tracker/st_cb_feedback.h',
+  'state_tracker/st_cb_flush.c',
+  'state_tracker/st_cb_flush.h',
+  'state_tracker/st_cb_memoryobjects.c',
+  'state_tracker/st_cb_memoryobjects.h',
+  'state_tracker/st_cb_msaa.c',
+  'state_tracker/st_cb_msaa.h',
+  'state_tracker/st_cb_perfmon.c',
+  'state_tracker/st_cb_perfmon.h',
+  'state_tracker/st_cb_program.c',
+  'state_tracker/st_cb_program.h',
+  'state_tracker/st_cb_queryobj.c',
+  'state_tracker/st_cb_queryobj.h',
+  'state_tracker/st_cb_rasterpos.c',
+  'state_tracker/st_cb_rasterpos.h',
+  'state_tracker/st_cb_readpixels.c',
+  'state_tracker/st_cb_readpixels.h',
+  'state_tracker/st_cb_strings.c',
+  'state_tracker/st_cb_strings.h',
+  'state_tracker/st_cb_semaphoreobjects.c',
+  'state_tracker/st_cb_semaphoreobjects.h',
+  'state_tracker/st_cb_syncobj.c',
+  'state_tracker/st_cb_syncobj.h',
+  'state_tracker/st_cb_texturebarrier.c',
+  'state_tracker/st_cb_texturebarrier.h',
+  'state_tracker/st_cb_texture.c',
+  'state_tracker/st_cb_texture.h',
+  'state_tracker/st_cb_viewport.c',
+  'state_tracker/st_cb_viewport.h',
+  'state_tracker/st_cb_xformfb.c',
+  'state_tracker/st_cb_xformfb.h',
+  'state_tracker/st_context.c',
+  'state_tracker/st_context.h',
+  'state_tracker/st_copytex.c',
+  'state_tracker/st_copytex.h',
+  'state_tracker/st_debug.c',
+  'state_tracker/st_debug.h',
+  'state_tracker/st_draw.c',
+  'state_tracker/st_draw_feedback.c',
+  'state_tracker/st_draw.h',
+  'state_tracker/st_extensions.c',
+  'state_tracker/st_extensions.h',
+  'state_tracker/st_format.c',
+  'state_tracker/st_format.h',
+  'state_tracker/st_gen_mipmap.c',
+  'state_tracker/st_gen_mipmap.h',
+  'state_tracker/st_gl_api.h',
+  'state_tracker/st_glsl_to_nir.cpp',
+  'state_tracker/st_glsl_to_tgsi.cpp',
+  'state_tracker/st_glsl_to_tgsi.h',
+  'state_tracker/st_glsl_to_tgsi_array_merge.cpp',
+  'state_tracker/st_glsl_to_tgsi_array_merge.h',
+  'state_tracker/st_glsl_to_tgsi_private.cpp',
+  'state_tracker/st_glsl_to_tgsi_private.h',
+  'state_tracker/st_glsl_to_tgsi_temprename.cpp',
+  'state_tracker/st_glsl_to_tgsi_temprename.h',
+  'state_tracker/st_glsl_types.cpp',
+  'state_tracker/st_glsl_types.h',
+  'state_tracker/st_manager.c',
+  'state_tracker/st_manager.h',
+  'state_tracker/st_mesa_to_tgsi.c',
+  'state_tracker/st_mesa_to_tgsi.h',
+  'state_tracker/st_nir.h',
+  'state_tracker/st_nir_lower_builtin.c',
+  'state_tracker/st_nir_lower_tex_src_plane.c',
+  'state_tracker/st_nir_lower_uniforms_to_ubo.c',
+  'state_tracker/st_pbo.c',
+  'state_tracker/st_pbo.h',
+  'state_tracker/st_program.c',
+  'state_tracker/st_program.h',
+  'state_tracker/st_sampler_view.c',
+  'state_tracker/st_sampler_view.h',
+  'state_tracker/st_scissor.c',
+  'state_tracker/st_scissor.h',
+  'state_tracker/st_shader_cache.c',
+  'state_tracker/st_shader_cache.h',
+  'state_tracker/st_texture.c',
+  'state_tracker/st_texture.h',
+  'state_tracker/st_tgsi_lower_yuv.c',
+  'state_tracker/st_tgsi_lower_yuv.h',
+  'state_tracker/st_vdpau.c',
+  'state_tracker/st_vdpau.h',
+)
+
+matypes_h = []
+if with_asm_arch == 'x86' or with_asm_arch == 'x86_64'
+  gen_matypes = executable(
+    'gen_matypes',
+    'x86/gen_matypes.c',
+    c_args : [c_vis_args, c_msvc_compat_args],
+    include_directories : inc_common,
+  )
+  matypes_h = custom_target(
+    'matypes.h',
+    output : 'matypes.h',
+    command : [gen_matypes],
+    capture : true,
+  )
+endif
+
+inc_libmesa_asm = []
+if with_asm_arch == 'x86'
+  files_libmesa_common += files(
+    'x86/assyntax.h',
+    'x86/clip_args.h',
+    'x86/norm_args.h',
+    'x86/xform_args.h',
+    'x86/common_x86_asm.S',
+    'x86/common_x86_asm.h',
+    'x86/common_x86_features.h',
+    'x86/x86_xform.h',
+    'x86/x86_xform2.S',
+    'x86/x86_xform3.S',
+    'x86/x86_xform4.S',
+    'x86/x86_cliptest.S',
+    'x86/mmx.h',
+    'x86/mmx_blend.S',
+    'x86/mmx_blendtmp.h',
+    'x86/3dnow.h',
+    'x86/3dnow_xform1.S',
+    'x86/3dnow_xform2.S',
+    'x86/3dnow_xform3.S',
+    'x86/3dnow_xform4.S',
+    'x86/sse.h',
+    'x86/sse_xform1.S',
+    'x86/sse_xform2.S',
+    'x86/sse_xform3.S',
+    'x86/sse_xform4.S',
+    'x86/sse_normal.S',
+    'x86/read_rgba_span_x86.S',
+  )
+  inc_libmesa_asm = include_directories('x86')
+elif with_asm_arch == 'x86_64'
+  files_libmesa_common += files('x86-64/x86-64.h', 'x86-64/xform4.S')
+  inc_libmesa_asm = include_directories('x86-64')
+elif with_asm_arch == 'sparc'
+  files_libmesa_common += files(
+    'sparc/sparc_clip.S',
+    'sparc/norm.S',
+    'sparc/xform.S',
+  )
+  inc_libmesa_asm = include_directories('sparc')
+endif
+
+format_fallback_c = custom_target(
+  'format_fallback.c',
+  input : ['main/format_fallback.py', 'main/formats.csv'],
+  output : 'format_fallback.c',
+  command : [prog_python, '@INPUT0@', '@INPUT1@', '@OUTPUT@'],
+  depend_files : files('main/format_parser.py'),
+)
+
+get_hash_h = custom_target(
+  'get_hash.h',
+  input : ['main/get_hash_generator.py', gl_and_es_api_files],
+  output : 'get_hash.h',
+  command : [prog_python, '@INPUT0@', '-f', '@INPUT1@'],
+  depend_files : files('main/get_hash_params.py'),
+  capture : true,
+)
+
+foreach x : [['format_info.h', 'format_info.py'],
+             ['format_pack.c', 'format_pack.py'],
+             ['format_unpack.c', 'format_unpack.py']]
+  files_libmesa_common += custom_target(
+    x[0],
+    input : ['main/@0@'.format(x[1]), 'main/formats.csv'],
+    output : x[0],
+    command : [prog_python, '@INPUT0@', '@INPUT1@'],
+    depend_files : files('main/format_parser.py'),
+    capture : true,
+  )
+endforeach
+
+files_libmesa_common += [
+  mesa_lex,
+  program_parse_tab,
+  main_api_exec_c,
+  main_enums_c,
+  format_fallback_c,
+  get_hash_h,
+  main_marshal_generated_c,
+  main_marshal_generated_h,
+  main_dispatch_h,
+  ir_expression_operation_h,
+  main_remap_helper_h,
+  matypes_h,
+  sha1_h,
+]
+
+if with_sse41
+  libmesa_sse41 = static_library(
+    'mesa_sse41',
+    files('main/streaming-load-memcpy.c', 'main/sse_minmax.c'),
+    c_args : [c_vis_args, c_msvc_compat_args, sse41_args],
+    include_directories : inc_common,
+  )
+else
+  libmesa_sse41 = []
+endif
+
+libmesa_classic = static_library(
+  'mesa_classic',
+  [files_libmesa_common, files_libmesa_classic],
+  c_args : [c_vis_args, c_msvc_compat_args],
+  cpp_args : [cpp_vis_args, cpp_msvc_compat_args],
+  include_directories : [inc_common, inc_libmesa_asm, include_directories('main')],
+  link_with : [libglsl, libmesa_sse41],
+  dependencies : idep_nir_headers,
+  build_by_default : false,
+)
+
+libmesa_gallium = static_library(
+  'mesa_gallium',
+  [files_libmesa_common, files_libmesa_gallium],
+  c_args : [c_vis_args, c_msvc_compat_args],
+  cpp_args : [cpp_vis_args, cpp_msvc_compat_args],
+  include_directories : [inc_common, inc_libmesa_asm, include_directories('main')],
+  link_with : [libglsl, libmesa_sse41],
+  dependencies : [idep_nir_headers, dep_vdpau],
+  build_by_default : false,
+)
+
+subdir('drivers/dri')
+if with_osmesa == 'classic'
+  subdir('drivers/osmesa')
+endif
+if with_glx == 'xlib'
+  subdir('drivers/x11')
+endif
+if with_tests and dri_drivers != []
+  subdir('main/tests')
+endif
diff --git a/lib/mesa/src/mesa/swrast/s_aalinetemp.h b/lib/mesa/src/mesa/swrast/s_aalinetemp.h
index bebb131a5..64767a3a5 100644
--- a/lib/mesa/src/mesa/swrast/s_aalinetemp.h
+++ b/lib/mesa/src/mesa/swrast/s_aalinetemp.h
@@ -179,10 +179,12 @@ NAME(line)(struct gl_context *ctx, const SWvertex *v0, const SWvertex *v1)
          if (attr >= VARYING_SLOT_TEX0 && attr < VARYING_SLOT_VAR0) {
             const GLuint u = attr - VARYING_SLOT_TEX0;
             const struct gl_texture_object *obj = ctx->Texture.Unit[u]._Current;
-            const struct gl_texture_image *texImage =
-               _mesa_base_tex_image(obj);
-            line.texWidth[attr]  = (GLfloat) texImage->Width;
-            line.texHeight[attr] = (GLfloat) texImage->Height;
+            if (obj) {
+               const struct gl_texture_image *texImage =
+                  _mesa_base_tex_image(obj);
+               line.texWidth[attr]  = (GLfloat) texImage->Width;
+               line.texHeight[attr] = (GLfloat) texImage->Height;
+            }
          }
       ATTRIB_LOOP_END
    }
diff --git a/lib/mesa/src/mesa/swrast/s_blit.c b/lib/mesa/src/mesa/swrast/s_blit.c
index 3e838a41d..107e41307 100644
--- a/lib/mesa/src/mesa/swrast/s_blit.c
+++ b/lib/mesa/src/mesa/swrast/s_blit.c
@@ -198,8 +198,8 @@ blit_nearest(struct gl_context *ctx,
    /* Blit to all the draw buffers */
    for (i = 0; i < numDrawBuffers; i++) {
       if (buffer == GL_COLOR_BUFFER_BIT) {
-         int idx = drawFb->_ColorDrawBufferIndexes[i];
-         if (idx == -1)
+         gl_buffer_index idx = drawFb->_ColorDrawBufferIndexes[i];
+         if (idx == BUFFER_NONE)
             continue;
          drawAtt = &drawFb->Attachment[idx];
          drawRb = drawAtt->Renderbuffer;
@@ -253,7 +253,7 @@ blit_nearest(struct gl_context *ctx,
          ctx->Driver.MapRenderbuffer(ctx, readRb, 0, 0,
                                      readRb->Width, readRb->Height,
                                      GL_MAP_READ_BIT | GL_MAP_WRITE_BIT,
-                                     &map, &rowStride);
+                                     &map, &rowStride, readFb->FlipY);
          if (!map) {
             goto fail_no_memory;
          }
@@ -280,14 +280,16 @@ blit_nearest(struct gl_context *ctx,
          ctx->Driver.MapRenderbuffer(ctx, readRb,
                                      srcXpos, srcYpos,
                                      srcWidth, srcHeight,
-                                     GL_MAP_READ_BIT, &srcMap, &srcRowStride);
+                                     GL_MAP_READ_BIT, &srcMap, &srcRowStride,
+                                     readFb->FlipY);
          if (!srcMap) {
             goto fail_no_memory;
          }
          ctx->Driver.MapRenderbuffer(ctx, drawRb,
                                      dstXpos, dstYpos,
                                      dstWidth, dstHeight,
-                                     GL_MAP_WRITE_BIT, &dstMap, &dstRowStride);
+                                     GL_MAP_WRITE_BIT, &dstMap, &dstRowStride,
+                                     drawFb->FlipY);
          if (!dstMap) {
             ctx->Driver.UnmapRenderbuffer(ctx, readRb);
             goto fail_no_memory;
@@ -569,12 +571,12 @@ blit_linear(struct gl_context *ctx,
    }
 
    for (i = 0; i < drawFb->_NumColorDrawBuffers; i++) {
-      GLint idx = drawFb->_ColorDrawBufferIndexes[i];
+      gl_buffer_index idx = drawFb->_ColorDrawBufferIndexes[i];
       struct gl_renderbuffer_attachment *drawAtt;
       struct gl_renderbuffer *drawRb;
       mesa_format drawFormat;
 
-      if (idx == -1)
+      if (idx == BUFFER_NONE)
          continue;
 
       drawAtt = &drawFb->Attachment[idx];
@@ -594,7 +596,8 @@ blit_linear(struct gl_context *ctx,
          ctx->Driver.MapRenderbuffer(ctx, readRb,
                                      0, 0, readRb->Width, readRb->Height,
                                      GL_MAP_READ_BIT | GL_MAP_WRITE_BIT,
-                                     &srcMap, &srcRowStride);
+                                     &srcMap, &srcRowStride,
+                                     readFb->FlipY);
          if (!srcMap) {
             goto fail_no_memory;
          }
@@ -609,13 +612,15 @@ blit_linear(struct gl_context *ctx,
           */
          ctx->Driver.MapRenderbuffer(ctx, readRb,
                                      0, 0, readRb->Width, readRb->Height,
-                                     GL_MAP_READ_BIT, &srcMap, &srcRowStride);
+                                     GL_MAP_READ_BIT, &srcMap, &srcRowStride,
+                                     readFb->FlipY);
          if (!srcMap) {
             goto fail_no_memory;
          }
          ctx->Driver.MapRenderbuffer(ctx, drawRb,
                                      0, 0, drawRb->Width, drawRb->Height,
-                                     GL_MAP_WRITE_BIT, &dstMap, &dstRowStride);
+                                     GL_MAP_WRITE_BIT, &dstMap, &dstRowStride,
+                                     drawFb->FlipY);
          if (!dstMap) {
             ctx->Driver.UnmapRenderbuffer(ctx, readRb);
             goto fail_no_memory;
diff --git a/lib/mesa/src/mesa/swrast/s_clear.c b/lib/mesa/src/mesa/swrast/s_clear.c
index c26b4be0d..ef0f6df9d 100644
--- a/lib/mesa/src/mesa/swrast/s_clear.c
+++ b/lib/mesa/src/mesa/swrast/s_clear.c
@@ -66,7 +66,8 @@ clear_rgba_buffer(struct gl_context *ctx, struct gl_renderbuffer *rb,
 
    /* map dest buffer */
    ctx->Driver.MapRenderbuffer(ctx, rb, x, y, width, height,
-                               mapMode, &map, &rowStride);
+                               mapMode, &map, &rowStride,
+                               ctx->DrawBuffer->FlipY);
    if (!map) {
       _mesa_error(ctx, GL_OUT_OF_MEMORY, "glClear(color)");
       return;
@@ -187,7 +188,13 @@ clear_color_buffers(struct gl_context *ctx)
       if (rb == NULL)
 	 continue;
 
-      clear_rgba_buffer(ctx, rb, ctx->Color.ColorMask[buf]);
+      const GLubyte colormask[4] = {
+         GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 0) ? 0xff : 0,
+         GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 1) ? 0xff : 0,
+         GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 2) ? 0xff : 0,
+         GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 3) ? 0xff : 0,
+      };
+      clear_rgba_buffer(ctx, rb, colormask);
    }
 }
 
diff --git a/lib/mesa/src/mesa/swrast/s_context.h b/lib/mesa/src/mesa/swrast/s_context.h
index 7cf0e30dc..49c30a3de 100644
--- a/lib/mesa/src/mesa/swrast/s_context.h
+++ b/lib/mesa/src/mesa/swrast/s_context.h
@@ -43,7 +43,6 @@
 #ifndef S_CONTEXT_H
 #define S_CONTEXT_H
 
-#include "main/compiler.h"
 #include "main/mtypes.h"
 #include "main/texcompress.h"
 #include "program/prog_execute.h"
diff --git a/lib/mesa/src/mesa/swrast/s_copypix.c b/lib/mesa/src/mesa/swrast/s_copypix.c
index 0dbccc0f6..d0703fa07 100644
--- a/lib/mesa/src/mesa/swrast/s_copypix.c
+++ b/lib/mesa/src/mesa/swrast/s_copypix.c
@@ -503,7 +503,7 @@ swrast_fast_copy_pixels(struct gl_context *ctx,
       ctx->Driver.MapRenderbuffer(ctx, srcRb, 0, 0,
                                   srcRb->Width, srcRb->Height,
                                   GL_MAP_READ_BIT | GL_MAP_WRITE_BIT,
-                                  &map, &rowStride);
+                                  &map, &rowStride, srcFb->FlipY);
       if (!map) {
          _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyPixels");
          return GL_TRUE; /* don't retry with slow path */
@@ -530,14 +530,16 @@ swrast_fast_copy_pixels(struct gl_context *ctx,
       /* different src/dst buffers */
       ctx->Driver.MapRenderbuffer(ctx, srcRb, srcX, srcY,
                                   width, height,
-                                  GL_MAP_READ_BIT, &srcMap, &srcRowStride);
+                                  GL_MAP_READ_BIT, &srcMap, &srcRowStride,
+                                  srcFb->FlipY);
       if (!srcMap) {
          _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyPixels");
          return GL_TRUE; /* don't retry with slow path */
       }
       ctx->Driver.MapRenderbuffer(ctx, dstRb, dstX, dstY,
                                   width, height,
-                                  GL_MAP_WRITE_BIT, &dstMap, &dstRowStride);
+                                  GL_MAP_WRITE_BIT, &dstMap, &dstRowStride,
+                                  dstFb->FlipY);
       if (!dstMap) {
          ctx->Driver.UnmapRenderbuffer(ctx, srcRb);
          _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCopyPixels");
@@ -598,7 +600,8 @@ map_readbuffer(struct gl_context *ctx, GLenum type)
    ctx->Driver.MapRenderbuffer(ctx, rb,
                                0, 0, rb->Width, rb->Height,
                                GL_MAP_READ_BIT,
-                               &srb->Map, &srb->RowStride);
+                               &srb->Map, &srb->RowStride,
+                               fb->FlipY);
 
    return rb;
 }
diff --git a/lib/mesa/src/mesa/swrast/s_depth.c b/lib/mesa/src/mesa/swrast/s_depth.c
index ffadc05a7..de7f14a4f 100644
--- a/lib/mesa/src/mesa/swrast/s_depth.c
+++ b/lib/mesa/src/mesa/swrast/s_depth.c
@@ -310,12 +310,6 @@ _swrast_depth_test_span(struct gl_context *ctx, SWspan *span)
       zBufferVals = zStart;
    }
    else {
-      if (_mesa_get_format_datatype(rb->Format) != GL_UNSIGNED_NORMALIZED) {
-         _mesa_problem(ctx, "Incorrectly writing swrast's integer depth "
-                       "values to %s depth buffer",
-                       _mesa_get_format_name(rb->Format));
-      }
-
       /* copy Z buffer values into temp buffer (32-bit Z values) */
       zBufferTemp = malloc(count * sizeof(GLuint));
       if (!zBufferTemp)
@@ -570,7 +564,8 @@ _swrast_clear_depth_buffer(struct gl_context *ctx)
    }
 
    ctx->Driver.MapRenderbuffer(ctx, rb, x, y, width, height,
-                               mapMode, &map, &rowStride);
+                               mapMode, &map, &rowStride,
+                               ctx->DrawBuffer->FlipY);
    if (!map) {
       _mesa_error(ctx, GL_OUT_OF_MEMORY, "glClear(depth)");
       return;
@@ -695,7 +690,8 @@ _swrast_clear_depth_stencil_buffer(struct gl_context *ctx)
    }
 
    ctx->Driver.MapRenderbuffer(ctx, rb, x, y, width, height,
-                               mapMode, &map, &rowStride);
+                               mapMode, &map, &rowStride,
+                               ctx->DrawBuffer->FlipY);
    if (!map) {
       _mesa_error(ctx, GL_OUT_OF_MEMORY, "glClear(depth+stencil)");
       return;
diff --git a/lib/mesa/src/mesa/swrast/s_drawpix.c b/lib/mesa/src/mesa/swrast/s_drawpix.c
index f05528d0d..7ee401b2d 100644
--- a/lib/mesa/src/mesa/swrast/s_drawpix.c
+++ b/lib/mesa/src/mesa/swrast/s_drawpix.c
@@ -55,7 +55,8 @@ fast_draw_rgb_ubyte_pixels(struct gl_context *ctx,
                            GLint x, GLint y,
                            GLsizei width, GLsizei height,
                            const struct gl_pixelstore_attrib *unpack,
-                           const GLvoid *pixels)
+                           const GLvoid *pixels,
+                           bool flip_y)
 {
    const GLubyte *src = (const GLubyte *)
       _mesa_image_address2d(unpack, pixels, width,
@@ -67,7 +68,8 @@ fast_draw_rgb_ubyte_pixels(struct gl_context *ctx,
    GLint dstRowStride;
 
    ctx->Driver.MapRenderbuffer(ctx, rb, x, y, width, height,
-                               GL_MAP_WRITE_BIT, &dst, &dstRowStride);
+                               GL_MAP_WRITE_BIT, &dst, &dstRowStride,
+                               flip_y);
 
    if (!dst) {
       _mesa_error(ctx, GL_OUT_OF_MEMORY, "glDrawPixels");
@@ -102,7 +104,8 @@ fast_draw_rgba_ubyte_pixels(struct gl_context *ctx,
                            GLint x, GLint y,
                            GLsizei width, GLsizei height,
                            const struct gl_pixelstore_attrib *unpack,
-                           const GLvoid *pixels)
+                           const GLvoid *pixels,
+                           bool flip_y)
 {
    const GLubyte *src = (const GLubyte *)
       _mesa_image_address2d(unpack, pixels, width,
@@ -114,7 +117,8 @@ fast_draw_rgba_ubyte_pixels(struct gl_context *ctx,
    GLint dstRowStride;
 
    ctx->Driver.MapRenderbuffer(ctx, rb, x, y, width, height,
-                               GL_MAP_WRITE_BIT, &dst, &dstRowStride);
+                               GL_MAP_WRITE_BIT, &dst, &dstRowStride,
+                               flip_y);
 
    if (!dst) {
       _mesa_error(ctx, GL_OUT_OF_MEMORY, "glDrawPixels");
@@ -151,7 +155,8 @@ fast_draw_generic_pixels(struct gl_context *ctx,
                          GLsizei width, GLsizei height,
                          GLenum format, GLenum type,
                          const struct gl_pixelstore_attrib *unpack,
-                         const GLvoid *pixels)
+                         const GLvoid *pixels,
+                         bool flip_y)
 {
    const GLubyte *src = (const GLubyte *)
       _mesa_image_address2d(unpack, pixels, width,
@@ -164,7 +169,8 @@ fast_draw_generic_pixels(struct gl_context *ctx,
    GLint dstRowStride;
 
    ctx->Driver.MapRenderbuffer(ctx, rb, x, y, width, height,
-                               GL_MAP_WRITE_BIT, &dst, &dstRowStride);
+                               GL_MAP_WRITE_BIT, &dst, &dstRowStride,
+                               flip_y);
 
    if (!dst) {
       _mesa_error(ctx, GL_OUT_OF_MEMORY, "glDrawPixels");
@@ -197,6 +203,7 @@ fast_draw_rgba_pixels(struct gl_context *ctx, GLint x, GLint y,
                       const struct gl_pixelstore_attrib *userUnpack,
                       const GLvoid *pixels)
 {
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
    struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[0];
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
    struct gl_pixelstore_attrib unpack;
@@ -228,7 +235,7 @@ fast_draw_rgba_pixels(struct gl_context *ctx, GLint x, GLint y,
        (rb->Format == MESA_FORMAT_B8G8R8X8_UNORM ||
         rb->Format == MESA_FORMAT_B8G8R8A8_UNORM)) {
       fast_draw_rgb_ubyte_pixels(ctx, rb, x, y, width, height,
-                                 &unpack, pixels);
+                                 &unpack, pixels, fb->FlipY);
       return GL_TRUE;
    }
 
@@ -237,14 +244,15 @@ fast_draw_rgba_pixels(struct gl_context *ctx, GLint x, GLint y,
        (rb->Format == MESA_FORMAT_B8G8R8X8_UNORM ||
         rb->Format == MESA_FORMAT_B8G8R8A8_UNORM)) {
       fast_draw_rgba_ubyte_pixels(ctx, rb, x, y, width, height,
-                                  &unpack, pixels);
+                                  &unpack, pixels, fb->FlipY);
       return GL_TRUE;
    }
 
    if (_mesa_format_matches_format_and_type(rb->Format, format, type,
                                             ctx->Unpack.SwapBytes, NULL)) {
       fast_draw_generic_pixels(ctx, rb, x, y, width, height,
-                               format, type, &unpack, pixels);
+                               format, type, &unpack, pixels,
+                               fb->FlipY);
       return GL_TRUE;
    }
 
diff --git a/lib/mesa/src/mesa/swrast/s_fog.c b/lib/mesa/src/mesa/swrast/s_fog.c
index 8b0bdf8fa..33da09b61 100644
--- a/lib/mesa/src/mesa/swrast/s_fog.c
+++ b/lib/mesa/src/mesa/swrast/s_fog.c
@@ -24,6 +24,7 @@
 
 
 #include "c99_math.h"
+#include "main/errors.h"
 #include "main/glheader.h"
 #include "main/macros.h"
 
diff --git a/lib/mesa/src/mesa/swrast/s_masking.c b/lib/mesa/src/mesa/swrast/s_masking.c
index c10bf1ac2..8941375a9 100644
--- a/lib/mesa/src/mesa/swrast/s_masking.c
+++ b/lib/mesa/src/mesa/swrast/s_masking.c
@@ -56,8 +56,14 @@ _swrast_mask_rgba_span(struct gl_context *ctx, struct gl_renderbuffer *rb,
     * Note that we're not using span->array->mask[] here.  We could...
     */
    if (span->array->ChanType == GL_UNSIGNED_BYTE) {
+      const GLubyte colormask[4] = {
+         GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 0) ? 0xff : 0,
+         GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 1) ? 0xff : 0,
+         GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 2) ? 0xff : 0,
+         GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 3) ? 0xff : 0,
+      };
       GLuint srcMask;
-      memcpy(&srcMask, ctx->Color.ColorMask[buf], sizeof(srcMask));
+      memcpy(&srcMask, colormask, sizeof(srcMask));
       const GLuint dstMask = ~srcMask;
       const GLuint *dst = (const GLuint *) rbPixels;
       GLuint *src = (GLuint *) span->array->rgba8;
@@ -69,10 +75,10 @@ _swrast_mask_rgba_span(struct gl_context *ctx, struct gl_renderbuffer *rb,
    else if (span->array->ChanType == GL_UNSIGNED_SHORT) {
       /* 2-byte components */
       /* XXX try to use 64-bit arithmetic someday */
-      const GLushort rMask = ctx->Color.ColorMask[buf][RCOMP] ? 0xffff : 0x0;
-      const GLushort gMask = ctx->Color.ColorMask[buf][GCOMP] ? 0xffff : 0x0;
-      const GLushort bMask = ctx->Color.ColorMask[buf][BCOMP] ? 0xffff : 0x0;
-      const GLushort aMask = ctx->Color.ColorMask[buf][ACOMP] ? 0xffff : 0x0;
+      const GLushort rMask = GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 0) ? 0xffff : 0x0;
+      const GLushort gMask = GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 1) ? 0xffff : 0x0;
+      const GLushort bMask = GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 2) ? 0xffff : 0x0;
+      const GLushort aMask = GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 3) ? 0xffff : 0x0;
       const GLushort (*dst)[4] = (const GLushort (*)[4]) rbPixels;
       GLushort (*src)[4] = span->array->rgba16;
       GLuint i;
@@ -85,10 +91,10 @@ _swrast_mask_rgba_span(struct gl_context *ctx, struct gl_renderbuffer *rb,
    }
    else {
       /* 4-byte components */
-      const GLuint rMask = ctx->Color.ColorMask[buf][RCOMP] ? ~0x0 : 0x0;
-      const GLuint gMask = ctx->Color.ColorMask[buf][GCOMP] ? ~0x0 : 0x0;
-      const GLuint bMask = ctx->Color.ColorMask[buf][BCOMP] ? ~0x0 : 0x0;
-      const GLuint aMask = ctx->Color.ColorMask[buf][ACOMP] ? ~0x0 : 0x0;
+      const GLuint rMask = GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 0) ? ~0x0 : 0x0;
+      const GLuint gMask = GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 1) ? ~0x0 : 0x0;
+      const GLuint bMask = GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 2) ? ~0x0 : 0x0;
+      const GLuint aMask = GET_COLORMASK_BIT(ctx->Color.ColorMask, buf, 3) ? ~0x0 : 0x0;
       const GLuint (*dst)[4] = (const GLuint (*)[4]) rbPixels;
       GLuint (*src)[4] = (GLuint (*)[4]) span->array->attribs[VARYING_SLOT_COL0];
       GLuint i;
diff --git a/lib/mesa/src/mesa/swrast/s_renderbuffer.h b/lib/mesa/src/mesa/swrast/s_renderbuffer.h
index 2595d7c17..9238d8afa 100644
--- a/lib/mesa/src/mesa/swrast/s_renderbuffer.h
+++ b/lib/mesa/src/mesa/swrast/s_renderbuffer.h
@@ -43,7 +43,8 @@ _swrast_map_soft_renderbuffer(struct gl_context *ctx,
                               GLuint x, GLuint y, GLuint w, GLuint h,
                               GLbitfield mode,
                               GLubyte **out_map,
-                              GLint *out_stride);
+                              GLint *out_stride,
+                              bool flip_y);
 
 extern void
 _swrast_unmap_soft_renderbuffer(struct gl_context *ctx,
diff --git a/lib/mesa/src/mesa/swrast/s_stencil.c b/lib/mesa/src/mesa/swrast/s_stencil.c
index 294b593a2..8ccd5a164 100644
--- a/lib/mesa/src/mesa/swrast/s_stencil.c
+++ b/lib/mesa/src/mesa/swrast/s_stencil.c
@@ -28,7 +28,6 @@
 #include "main/imports.h"
 #include "main/format_pack.h"
 #include "main/format_unpack.h"
-#include "main/core.h"
 #include "main/stencil.h"
 
 #include "s_context.h"
@@ -580,7 +579,8 @@ _swrast_clear_stencil_buffer(struct gl_context *ctx)
    }
 
    ctx->Driver.MapRenderbuffer(ctx, rb, x, y, width, height,
-                               mapMode, &map, &rowStride);
+                               mapMode, &map, &rowStride,
+                               ctx->DrawBuffer->FlipY);
    if (!map) {
       _mesa_error(ctx, GL_OUT_OF_MEMORY, "glClear(stencil)");
       return;
diff --git a/lib/mesa/src/mesa/swrast/s_texcombine.c b/lib/mesa/src/mesa/swrast/s_texcombine.c
index da4a01363..743ee4015 100644
--- a/lib/mesa/src/mesa/swrast/s_texcombine.c
+++ b/lib/mesa/src/mesa/swrast/s_texcombine.c
@@ -84,7 +84,8 @@ texture_combine( struct gl_context *ctx, GLuint unit,
                  SWspan *span )
 {
    SWcontext *swrast = SWRAST_CONTEXT(ctx);
-   const struct gl_texture_unit *textureUnit = &(ctx->Texture.Unit[unit]);
+   const struct gl_fixedfunc_texture_unit *textureUnit =
+      &ctx->Texture.FixedFuncUnit[unit];
    const struct gl_tex_env_combine_state *combine = textureUnit->_CurrentCombine;
    float4_array argRGB[MAX_COMBINER_TERMS];
    float4_array argA[MAX_COMBINER_TERMS];
diff --git a/lib/mesa/src/mesa/swrast/s_texfetch.c b/lib/mesa/src/mesa/swrast/s_texfetch.c
index e2c3c085b..fec8728a7 100644
--- a/lib/mesa/src/mesa/swrast/s_texfetch.c
+++ b/lib/mesa/src/mesa/swrast/s_texfetch.c
@@ -33,6 +33,7 @@
  */
 
 
+#include "main/errors.h"
 #include "main/macros.h"
 #include "main/texcompress.h"
 #include "main/texcompress_fxt1.h"
diff --git a/lib/mesa/src/mesa/swrast/s_zoom.c b/lib/mesa/src/mesa/swrast/s_zoom.c
index 34b8eb196..2472aa730 100644
--- a/lib/mesa/src/mesa/swrast/s_zoom.c
+++ b/lib/mesa/src/mesa/swrast/s_zoom.c
@@ -22,6 +22,7 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
+#include "main/errors.h"
 #include "main/glheader.h"
 #include "main/macros.h"
 #include "main/imports.h"
diff --git a/lib/mesa/src/mesa/tnl/t_context.h b/lib/mesa/src/mesa/tnl/t_context.h
index e7adb5f53..eca9f6603 100644
--- a/lib/mesa/src/mesa/tnl/t_context.h
+++ b/lib/mesa/src/mesa/tnl/t_context.h
@@ -57,6 +57,8 @@
 
 #include "vbo/vbo.h"
 
+#include "tnl.h"
+
 #define MAX_PIPELINE_STAGES     30
 
 /*
@@ -76,39 +78,45 @@
  * attribs want (16).
  */
 enum {
-	_TNL_ATTRIB_POS = 0,
-	_TNL_ATTRIB_WEIGHT = 1,
-	_TNL_ATTRIB_NORMAL = 2,
-	_TNL_ATTRIB_COLOR0 = 3,
-	_TNL_ATTRIB_COLOR1 = 4,
-	_TNL_ATTRIB_FOG = 5,
-	_TNL_ATTRIB_COLOR_INDEX = 6,
-	_TNL_ATTRIB_EDGEFLAG = 7,
-	_TNL_ATTRIB_TEX0 = 8,
-	_TNL_ATTRIB_TEX1 = 9,
-	_TNL_ATTRIB_TEX2 = 10,
-	_TNL_ATTRIB_TEX3 = 11,
-	_TNL_ATTRIB_TEX4 = 12,
-	_TNL_ATTRIB_TEX5 = 13,
-	_TNL_ATTRIB_TEX6 = 14,
-	_TNL_ATTRIB_TEX7 = 15,
-
-	_TNL_ATTRIB_GENERIC0 = 17, /* doesn't really exist! */
-	_TNL_ATTRIB_GENERIC1 = 18,
-	_TNL_ATTRIB_GENERIC2 = 19,
-	_TNL_ATTRIB_GENERIC3 = 20,
-	_TNL_ATTRIB_GENERIC4 = 21,
-	_TNL_ATTRIB_GENERIC5 = 22,
-	_TNL_ATTRIB_GENERIC6 = 23,
-	_TNL_ATTRIB_GENERIC7 = 24,
-	_TNL_ATTRIB_GENERIC8 = 25,
-	_TNL_ATTRIB_GENERIC9 = 26,
-	_TNL_ATTRIB_GENERIC10 = 27,
-	_TNL_ATTRIB_GENERIC11 = 28,
-	_TNL_ATTRIB_GENERIC12 = 29,
-	_TNL_ATTRIB_GENERIC13 = 30,
-	_TNL_ATTRIB_GENERIC14 = 31,
-	_TNL_ATTRIB_GENERIC15 = 32,
+	_TNL_ATTRIB_POS,
+	_TNL_ATTRIB_NORMAL,
+	_TNL_ATTRIB_COLOR0,
+	_TNL_ATTRIB_COLOR1,
+	_TNL_ATTRIB_FOG,
+	_TNL_ATTRIB_COLOR_INDEX,
+	_TNL_ATTRIB_EDGEFLAG,
+	_TNL_ATTRIB_TEX0,
+	_TNL_ATTRIB_TEX1,
+	_TNL_ATTRIB_TEX2,
+	_TNL_ATTRIB_TEX3,
+	_TNL_ATTRIB_TEX4,
+	_TNL_ATTRIB_TEX5,
+	_TNL_ATTRIB_TEX6,
+	_TNL_ATTRIB_TEX7,
+
+	/* This is really a VARYING_SLOT, not an attrib.  Need to fix
+	 * tnl to understand the difference.
+	 */
+	_TNL_ATTRIB_POINTSIZE,
+
+	_TNL_ATTRIB_GENERIC0, /* doesn't really exist! */
+	_TNL_ATTRIB_GENERIC1,
+	_TNL_ATTRIB_GENERIC2,
+	_TNL_ATTRIB_GENERIC3,
+	_TNL_ATTRIB_GENERIC4,
+	_TNL_ATTRIB_GENERIC5,
+	_TNL_ATTRIB_GENERIC6,
+	_TNL_ATTRIB_GENERIC7,
+	_TNL_ATTRIB_GENERIC8,
+	_TNL_ATTRIB_GENERIC9,
+	_TNL_ATTRIB_GENERIC10,
+	_TNL_ATTRIB_GENERIC11,
+	_TNL_ATTRIB_GENERIC12,
+	_TNL_ATTRIB_GENERIC13,
+	_TNL_ATTRIB_GENERIC14,
+	_TNL_ATTRIB_GENERIC15,
+
+	_TNL_ATTRIB_MAX,
 
 	/* These alias with the generics, but they are not active
 	 * concurrently, so it's not a problem.  The TNL module
@@ -120,26 +128,19 @@ enum {
 	 * generic attribute in order to pick up per-vertex material
 	 * data.
 	 */
-	_TNL_ATTRIB_MAT_FRONT_AMBIENT = 17,
-	_TNL_ATTRIB_MAT_BACK_AMBIENT = 18,
-	_TNL_ATTRIB_MAT_FRONT_DIFFUSE = 19,
-	_TNL_ATTRIB_MAT_BACK_DIFFUSE = 20,
-	_TNL_ATTRIB_MAT_FRONT_SPECULAR = 21,
-	_TNL_ATTRIB_MAT_BACK_SPECULAR = 22,
-	_TNL_ATTRIB_MAT_FRONT_EMISSION = 23,
-	_TNL_ATTRIB_MAT_BACK_EMISSION = 24,
-	_TNL_ATTRIB_MAT_FRONT_SHININESS = 25,
-	_TNL_ATTRIB_MAT_BACK_SHININESS = 26,
-	_TNL_ATTRIB_MAT_FRONT_INDEXES = 27,
-	_TNL_ATTRIB_MAT_BACK_INDEXES = 28,
-
-	/* This is really a VARYING_SLOT, not an attrib.  Need to fix
-	 * tnl to understand the difference.
-	 */
-	_TNL_ATTRIB_POINTSIZE = 16,
-
-	_TNL_ATTRIB_MAX = 33
-} ;
+	_TNL_ATTRIB_MAT_FRONT_AMBIENT=VERT_ATTRIB_MAT(MAT_ATTRIB_FRONT_AMBIENT),
+	_TNL_ATTRIB_MAT_BACK_AMBIENT,
+	_TNL_ATTRIB_MAT_FRONT_DIFFUSE,
+	_TNL_ATTRIB_MAT_BACK_DIFFUSE,
+	_TNL_ATTRIB_MAT_FRONT_SPECULAR,
+	_TNL_ATTRIB_MAT_BACK_SPECULAR,
+	_TNL_ATTRIB_MAT_FRONT_EMISSION,
+	_TNL_ATTRIB_MAT_BACK_EMISSION,
+	_TNL_ATTRIB_MAT_FRONT_SHININESS,
+	_TNL_ATTRIB_MAT_BACK_SHININESS,
+	_TNL_ATTRIB_MAT_FRONT_INDEXES,
+	_TNL_ATTRIB_MAT_BACK_INDEXES,
+};
 
 #define _TNL_ATTRIB_TEX(u)       (_TNL_ATTRIB_TEX0 + (u))
 #define _TNL_ATTRIB_GENERIC(n) (_TNL_ATTRIB_GENERIC0 + (n))
@@ -150,7 +151,7 @@ enum {
 /**
  * Handy attribute ranges:
  */
-#define _TNL_FIRST_PROG      _TNL_ATTRIB_WEIGHT
+#define _TNL_FIRST_PROG      _TNL_ATTRIB_NORMAL
 #define _TNL_LAST_PROG       _TNL_ATTRIB_TEX7
 
 #define _TNL_FIRST_TEX       _TNL_ATTRIB_TEX0
@@ -159,8 +160,8 @@ enum {
 #define _TNL_FIRST_GENERIC _TNL_ATTRIB_GENERIC0
 #define _TNL_LAST_GENERIC  _TNL_ATTRIB_GENERIC15
 
-#define _TNL_FIRST_MAT       _TNL_ATTRIB_MAT_FRONT_AMBIENT /* GENERIC0 */
-#define _TNL_LAST_MAT        _TNL_ATTRIB_MAT_BACK_INDEXES  /* GENERIC11 */
+#define _TNL_FIRST_MAT       _TNL_ATTRIB_MAT_FRONT_AMBIENT /* GENERIC4 */
+#define _TNL_LAST_MAT        _TNL_ATTRIB_MAT_BACK_INDEXES  /* GENERIC15 */
 
 /* Number of available texture attributes */
 #define _TNL_NUM_TEX 8
@@ -498,6 +499,41 @@ struct tnl_device_driver
 
 
 /**
+ * Utility that tracks and updates the current array entries.
+ */
+struct tnl_inputs
+{
+   /**
+    * Array of inputs to be set to the _DrawArrays pointer.
+    * The array contains pointers into the _DrawVAO and to the vbo modules
+    * current values. The array of pointers is updated incrementally
+    * based on the current and vertex_processing_mode values below.
+    */
+   struct tnl_vertex_array inputs[VERT_ATTRIB_MAX];
+   /** Those VERT_BIT_'s where the inputs array point to current values. */
+   GLbitfield current;
+   /** Store which aliasing current values - generics or materials - are set. */
+   gl_vertex_processing_mode vertex_processing_mode;
+};
+
+
+/**
+ * Initialize inputs.
+ */
+void
+_tnl_init_inputs(struct tnl_inputs *inputs);
+
+
+/**
+ * Update the tnl_vertex_array array inside the tnl_inputs structure
+ * provided the current _VPMode, the provided vao and
+ * the vao's enabled arrays filtered by the filter bitmask.
+ */
+void
+_tnl_update_inputs(struct gl_context *ctx, struct tnl_inputs *inputs);
+
+
+/**
  * Context state for T&L context.
  */
 typedef struct
@@ -537,6 +573,9 @@ typedef struct
    struct tnl_shine_tab *_ShineTable[2]; /**< Active shine tables */
    struct tnl_shine_tab *_ShineTabList;  /**< MRU list of inactive shine tables */
    /**@}*/
+
+   /* The list of tnl_vertex_array inputs. */
+   struct tnl_inputs draw_arrays;
 } TNLcontext;
 
 
diff --git a/lib/mesa/src/mesa/tnl/t_rebase.c b/lib/mesa/src/mesa/tnl/t_rebase.c
new file mode 100644
index 000000000..b6950e04f
--- /dev/null
+++ b/lib/mesa/src/mesa/tnl/t_rebase.c
@@ -0,0 +1,253 @@
+
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Keith Whitwell <keithw@vmware.com>
+ */
+
+/* Helper for drivers which find themselves rendering a range of
+ * indices starting somewhere above zero.  Typically the application
+ * is issuing multiple DrawArrays() or DrawElements() to draw
+ * successive primitives layed out linearly in the vertex arrays.
+ * Unless the vertex arrays are all in a VBO, the OpenGL semantics
+ * imply that we need to re-upload the vertex data on each draw call.
+ * In that case, we want to avoid starting the upload at zero, as it
+ * will mean every draw call uploads an increasing amount of not-used
+ * vertex data.  Worse - in the software tnl module, all those
+ * vertices will be transformed and lit.
+ *
+ * If we just upload the new data, however, the indices will be
+ * incorrect as we tend to upload each set of vertex data to a new
+ * region.  
+ *
+ * This file provides a helper to adjust the arrays, primitives and
+ * indices of a draw call so that it can be re-issued with a min_index
+ * of zero.
+ */
+
+#include <stdio.h>
+#include "main/bufferobj.h"
+#include "main/errors.h"
+#include "main/glheader.h"
+#include "main/imports.h"
+#include "main/mtypes.h"
+#include "vbo/vbo.h"
+
+#include "t_rebase.h"
+
+
+#define REBASE(TYPE) 						\
+static void *rebase_##TYPE( const void *ptr,			\
+			  GLuint count, 			\
+			  TYPE min_index )			\
+{								\
+   GLuint i;							\
+   const TYPE *in = (TYPE *)ptr;				\
+   TYPE *tmp_indices = malloc(count * sizeof(TYPE));		\
+								\
+   if (tmp_indices == NULL) {                                   \
+      _mesa_error_no_memory(__func__);                          \
+      return NULL;                                              \
+   }                                                            \
+								\
+   for (i = 0; i < count; i++)  				\
+      tmp_indices[i] = in[i] - min_index;			\
+								\
+   return (void *)tmp_indices;					\
+}
+
+
+REBASE(GLuint)
+REBASE(GLushort)
+REBASE(GLubyte)
+
+
+
+/* Adjust primitives, indices and vertex definitions so that min_index
+ * becomes zero. There are lots of reasons for wanting to do this, eg:
+ *
+ * Software tnl:
+ *    - any time min_index != 0, otherwise unused vertices lower than
+ *      min_index will be transformed.
+ *
+ * Hardware tnl:
+ *    - if ib != NULL and min_index != 0, otherwise vertices lower than 
+ *      min_index will be uploaded.  Requires adjusting index values.
+ *
+ *    - if ib == NULL and min_index != 0, just for convenience so this doesn't
+ *      have to be handled within the driver.
+ *
+ * Hardware tnl with VBO support:
+ *    - as above, but only when vertices are not (all?) in VBO's.
+ *    - can't save time by trying to upload half a vbo - typically it is
+ *      all or nothing.
+ */
+void t_rebase_prims( struct gl_context *ctx,
+                     const struct tnl_vertex_array *arrays,
+                     const struct _mesa_prim *prim,
+                     GLuint nr_prims,
+                     const struct _mesa_index_buffer *ib,
+                     GLuint min_index,
+                     GLuint max_index,
+                     tnl_draw_func draw )
+{
+   struct gl_array_attributes tmp_attribs[VERT_ATTRIB_MAX];
+   struct tnl_vertex_array tmp_arrays[VERT_ATTRIB_MAX];
+
+   struct _mesa_index_buffer tmp_ib;
+   struct _mesa_prim *tmp_prims = NULL;
+   void *tmp_indices = NULL;
+   GLuint i;
+
+   assert(min_index != 0);
+
+   if (0)
+      printf("%s %d..%d\n", __func__, min_index, max_index);
+
+
+   /* XXX this path is disabled for now.
+    * There's rendering corruption in some apps when it's enabled.
+    */
+   if (0 && ib && ctx->Extensions.ARB_draw_elements_base_vertex) {
+      /* If we can just tell the hardware or the TNL to interpret our
+       * indices with a different base, do so.
+       */
+      tmp_prims = malloc(sizeof(*prim) * nr_prims);
+
+      if (tmp_prims == NULL) {
+         _mesa_error_no_memory(__func__);
+         return;
+      }
+
+      for (i = 0; i < nr_prims; i++) {
+	 tmp_prims[i] = prim[i];
+	 tmp_prims[i].basevertex -= min_index;
+      }
+
+      prim = tmp_prims;
+   } else if (ib) {
+      /* Unfortunately need to adjust each index individually.
+       */
+      GLboolean map_ib = ib->obj->Name &&
+                         !ib->obj->Mappings[MAP_INTERNAL].Pointer;
+      void *ptr;
+
+      if (map_ib) 
+	 ctx->Driver.MapBufferRange(ctx, 0, ib->obj->Size, GL_MAP_READ_BIT,
+				    ib->obj, MAP_INTERNAL);
+
+
+      ptr = ADD_POINTERS(ib->obj->Mappings[MAP_INTERNAL].Pointer, ib->ptr);
+
+      /* Some users might prefer it if we translated elements to
+       * GLuints here.  Others wouldn't...
+       */
+      switch (ib->index_size) {
+      case 4:
+	 tmp_indices = rebase_GLuint( ptr, ib->count, min_index );
+	 break;
+      case 2:
+	 tmp_indices = rebase_GLushort( ptr, ib->count, min_index );
+	 break;
+      case 1:
+	 tmp_indices = rebase_GLubyte( ptr, ib->count, min_index );
+	 break;
+      }      
+
+      if (map_ib) 
+	 ctx->Driver.UnmapBuffer(ctx, ib->obj, MAP_INTERNAL);
+
+      if (tmp_indices == NULL) {
+         return;
+      }
+
+      tmp_ib.obj = ctx->Shared->NullBufferObj;
+      tmp_ib.ptr = tmp_indices;
+      tmp_ib.count = ib->count;
+      tmp_ib.index_size = ib->index_size;
+
+      ib = &tmp_ib;
+   }
+   else {
+      /* Otherwise the primitives need adjustment.
+       */
+      tmp_prims = malloc(sizeof(*prim) * nr_prims);
+
+      if (tmp_prims == NULL) {
+         _mesa_error_no_memory(__func__);
+         return;
+      }
+
+      for (i = 0; i < nr_prims; i++) {
+	 /* If this fails, it could indicate an application error:
+	  */
+	 assert(prim[i].start >= min_index);
+
+	 tmp_prims[i] = prim[i];
+	 tmp_prims[i].start -= min_index;
+      }
+
+      prim = tmp_prims;
+   }
+
+   /* Just need to adjust the pointer values on each incoming array.
+    * This works for VBO and non-vbo rendering and shouldn't pesimize
+    * VBO-based upload schemes.  However this may still not be a fast
+    * path for hardware tnl for VBO based rendering as most machines
+    * will be happier if you just specify a starting vertex value in
+    * each primitive.
+    *
+    * For drivers with hardware tnl, you only want to do this if you
+    * are forced to, eg non-VBO indexed rendering with start != 0.
+    */
+   for (i = 0; i < VERT_ATTRIB_MAX; i++) {
+      tmp_attribs[i] = *(arrays[i].VertexAttrib);
+      tmp_arrays[i].BufferBinding = arrays[i].BufferBinding;
+      tmp_arrays[i].VertexAttrib = &tmp_attribs[i];
+      if (_mesa_is_bufferobj(arrays[i].BufferBinding->BufferObj))
+         tmp_attribs[i].RelativeOffset +=
+            min_index * arrays[i].BufferBinding->Stride;
+      else
+         tmp_attribs[i].Ptr += min_index * arrays[i].BufferBinding->Stride;
+   }
+   
+   /* Re-issue the draw call.
+    */
+   draw( ctx,
+         tmp_arrays,
+	 prim,
+	 nr_prims, 
+	 ib, 
+	 GL_TRUE,
+	 0, 
+	 max_index - min_index,
+	 NULL, 0, NULL );
+
+   free(tmp_indices);
+   
+   free(tmp_prims);
+}
+
+
+
diff --git a/lib/mesa/src/mesa/tnl/t_rebase.h b/lib/mesa/src/mesa/tnl/t_rebase.h
new file mode 100644
index 000000000..d0aa9e189
--- /dev/null
+++ b/lib/mesa/src/mesa/tnl/t_rebase.h
@@ -0,0 +1,39 @@
+/*
+ * mesa 3-D graphics library
+ *
+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#ifndef _T_REBASE_H_
+#define _T_REBASE_H_
+
+#include "tnl.h"
+
+void t_rebase_prims( struct gl_context *ctx,
+                     const struct tnl_vertex_array *arrays,
+                     const struct _mesa_prim *prim,
+                     GLuint nr_prims,
+                     const struct _mesa_index_buffer *ib,
+                     GLuint min_index,
+                     GLuint max_index,
+                     tnl_draw_func draw );
+
+#endif
diff --git a/lib/mesa/src/mesa/tnl/t_split.c b/lib/mesa/src/mesa/tnl/t_split.c
new file mode 100644
index 000000000..d7aac10e4
--- /dev/null
+++ b/lib/mesa/src/mesa/tnl/t_split.c
@@ -0,0 +1,160 @@
+
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Keith Whitwell <keithw@vmware.com>
+ */
+
+/* Deal with hardware and/or swtnl maximums:
+ * - maximum number of vertices in buffer
+ * - maximum number of elements (maybe zero)
+ *
+ * The maximums may vary with opengl state (eg if a larger hardware
+ * vertex is required in this state, the maximum number of vertices
+ * may be smaller than in another state).
+ *
+ * We want buffer splitting to be a convenience function for the code
+ * actually drawing the primitives rather than a system-wide maximum,
+ * otherwise it is hard to avoid pessimism.
+ *
+ * For instance, if a driver has no hardware limits on vertex buffer
+ * dimensions, it would not ordinarily want to split vbos.  But if
+ * there is an unexpected fallback, eg memory manager fails to upload
+ * textures, it will want to pass the drawing commands onto swtnl,
+ * which does have limitations.  A convenience function allows swtnl
+ * to split the drawing and vbos internally without imposing its
+ * limitations on drivers which want to use it as a fallback path.
+ */
+
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "vbo/vbo.h"
+
+#include "t_split.h"
+
+
+/* True if a primitive can be split without copying of vertices, false
+ * otherwise.
+ */
+GLboolean
+_tnl_split_prim_inplace(GLenum mode, GLuint *first, GLuint *incr)
+{
+   switch (mode) {
+   case GL_POINTS:
+      *first = 1;
+      *incr = 1;
+      return GL_TRUE;
+   case GL_LINES:
+      *first = 2;
+      *incr = 2;
+      return GL_TRUE;
+   case GL_LINE_STRIP:
+      *first = 2;
+      *incr = 1;
+      return GL_TRUE;
+   case GL_TRIANGLES:
+      *first = 3;
+      *incr = 3;
+      return GL_TRUE;
+   case GL_TRIANGLE_STRIP:
+      *first = 3;
+      *incr = 1;
+      return GL_TRUE;
+   case GL_QUADS:
+      *first = 4;
+      *incr = 4;
+      return GL_TRUE;
+   case GL_QUAD_STRIP:
+      *first = 4;
+      *incr = 2;
+      return GL_TRUE;
+   default:
+      *first = 0;
+      *incr = 1;		/* so that count % incr works */
+      return GL_FALSE;
+   }
+}
+
+
+
+void
+_tnl_split_prims(struct gl_context *ctx,
+                 const struct tnl_vertex_array arrays[],
+                 const struct _mesa_prim *prim,
+                 GLuint nr_prims,
+                 const struct _mesa_index_buffer *ib,
+                 GLuint min_index,
+                 GLuint max_index,
+                 tnl_draw_func draw,
+                 const struct split_limits *limits)
+{
+   if (ib) {
+      if (limits->max_indices == 0) {
+         /* Could traverse the indices, re-emitting vertices in turn.
+          * But it's hard to see why this case would be needed - for
+          * software tnl, it is better to convert to non-indexed
+          * rendering after transformation is complete.  Are there any devices
+          * with hardware tnl that cannot do indexed rendering?
+          *
+          * For now, this path is disabled.
+          */
+         assert(0);
+      }
+      else if (max_index - min_index >= limits->max_verts) {
+         /* The vertex buffers are too large for hardware (or the
+          * swtnl module).  Traverse the indices, re-emitting vertices
+          * in turn.  Use a vertex cache to preserve some of the
+          * sharing from the original index list.
+          */
+         _tnl_split_copy(ctx, arrays, prim, nr_prims, ib, draw, limits);
+      }
+      else if (ib->count > limits->max_indices) {
+         /* The index buffer is too large for hardware.  Try to split
+          * on whole-primitive boundaries, otherwise try to split the
+          * individual primitives.
+          */
+         _tnl_split_inplace(ctx, arrays, prim, nr_prims, ib,
+                            min_index, max_index, draw, limits);
+      }
+      else {
+         /* Why were we called? */
+         assert(0);
+      }
+   }
+   else {
+      if (max_index - min_index >= limits->max_verts) {
+         /* The vertex buffer is too large for hardware (or the swtnl
+          * module).  Try to split on whole-primitive boundaries,
+          * otherwise try to split the individual primitives.
+          */
+         _tnl_split_inplace(ctx, arrays, prim, nr_prims, ib,
+                            min_index, max_index, draw, limits);
+      }
+      else {
+         /* Why were we called? */
+         assert(0);
+      }
+   }
+}
+
diff --git a/lib/mesa/src/mesa/tnl/t_split.h b/lib/mesa/src/mesa/tnl/t_split.h
new file mode 100644
index 000000000..49017e5df
--- /dev/null
+++ b/lib/mesa/src/mesa/tnl/t_split.h
@@ -0,0 +1,74 @@
+/*
+ * mesa 3-D graphics library
+ *
+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/**
+ * \brief VBO builder module datatypes and definitions.
+ * \author Keith Whitwell
+ */
+
+
+/**
+ * \mainpage The TNL splitter
+ *
+ * This is the private data used internally to the _tnl_split_prims()
+ * helper function.  Nobody outside the _tnl_split* files needs to
+ * include or know about this structure.
+ */
+
+
+#ifndef _TNL_SPLIT_H
+#define _TNL_SPLIT_H
+
+#include "tnl.h"
+
+
+/* True if a primitive can be split without copying of vertices, false
+ * otherwise.
+ */
+GLboolean
+_tnl_split_prim_inplace(GLenum mode, GLuint *first, GLuint *incr);
+
+void
+_tnl_split_inplace(struct gl_context *ctx,
+                   const struct tnl_vertex_array arrays[],
+                   const struct _mesa_prim *prim,
+                   GLuint nr_prims,
+                   const struct _mesa_index_buffer *ib,
+                   GLuint min_index,
+                   GLuint max_index,
+                   tnl_draw_func draw,
+                   const struct split_limits *limits);
+
+/* Requires ib != NULL:
+ */
+void
+_tnl_split_copy(struct gl_context *ctx,
+                const struct tnl_vertex_array arrays[],
+                const struct _mesa_prim *prim,
+                GLuint nr_prims,
+                const struct _mesa_index_buffer *ib,
+                tnl_draw_func draw,
+                const struct split_limits *limits);
+
+#endif
diff --git a/lib/mesa/src/mesa/tnl/t_split_copy.c b/lib/mesa/src/mesa/tnl/t_split_copy.c
new file mode 100644
index 000000000..085ae9a28
--- /dev/null
+++ b/lib/mesa/src/mesa/tnl/t_split_copy.c
@@ -0,0 +1,650 @@
+
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Keith Whitwell <keithw@vmware.com>
+ */
+
+/* Split indexed primitives with per-vertex copying.
+ */
+
+#include <stdio.h>
+
+#include "main/glheader.h"
+#include "main/bufferobj.h"
+#include "main/imports.h"
+#include "main/glformats.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/varray.h"
+#include "vbo/vbo.h"
+
+#include "t_split.h"
+#include "tnl.h"
+
+
+#define ELT_TABLE_SIZE 16
+
+/**
+ * Used for vertex-level splitting of indexed buffers.  Note that
+ * non-indexed primitives may be converted to indexed in some cases
+ * (eg loops, fans) in order to use this splitting path.
+ */
+struct copy_context {
+   struct gl_context *ctx;
+   const struct tnl_vertex_array *array;
+   const struct _mesa_prim *prim;
+   GLuint nr_prims;
+   const struct _mesa_index_buffer *ib;
+   tnl_draw_func draw;
+
+   const struct split_limits *limits;
+
+   struct {
+      GLuint attr;
+      GLuint size;
+      const struct tnl_vertex_array *array;
+      const GLubyte *src_ptr;
+
+      struct gl_vertex_buffer_binding dstbinding;
+      struct gl_array_attributes dstattribs;
+
+   } varying[VERT_ATTRIB_MAX];
+   GLuint nr_varying;
+
+   struct tnl_vertex_array dstarray[VERT_ATTRIB_MAX];
+   struct _mesa_index_buffer dstib;
+
+   GLuint *translated_elt_buf;
+   const GLuint *srcelt;
+
+   /** A baby hash table to avoid re-emitting (some) duplicate
+    * vertices when splitting indexed primitives.
+    */
+   struct {
+      GLuint in;
+      GLuint out;
+   } vert_cache[ELT_TABLE_SIZE];
+
+   GLuint vertex_size;
+   GLubyte *dstbuf;
+   GLubyte *dstptr;     /**< dstptr == dstbuf + dstelt_max * vertsize */
+   GLuint dstbuf_size;  /**< in vertices */
+   GLuint dstbuf_nr;    /**< count of emitted vertices, also the largest value
+                         * in dstelt.  Our MaxIndex.
+                         */
+
+   GLuint *dstelt;
+   GLuint dstelt_nr;
+   GLuint dstelt_size;
+
+#define MAX_PRIM 32
+   struct _mesa_prim dstprim[MAX_PRIM];
+   GLuint dstprim_nr;
+};
+
+
+static GLuint
+attr_size(const struct gl_array_attributes *attrib)
+{
+   return attrib->Size * _mesa_sizeof_type(attrib->Type);
+}
+
+
+/**
+ * Shallow copy one vertex array to another.
+ */
+static inline void
+copy_vertex_array(struct tnl_vertex_array *dst,
+                  const struct tnl_vertex_array *src)
+{
+   dst->VertexAttrib = src->VertexAttrib;
+   dst->BufferBinding = src->BufferBinding;
+}
+
+
+/**
+ * Starts returning true slightly before the buffer fills, to ensure
+ * that there is sufficient room for any remaining vertices to finish
+ * off the prim:
+ */
+static GLboolean
+check_flush(struct copy_context *copy)
+{
+   GLenum mode = copy->dstprim[copy->dstprim_nr].mode;
+
+   if (GL_TRIANGLE_STRIP == mode &&
+       copy->dstelt_nr & 1) { /* see bug9962 */
+       return GL_FALSE;
+   }
+
+   if (copy->dstbuf_nr + 4 > copy->dstbuf_size)
+      return GL_TRUE;
+
+   if (copy->dstelt_nr + 4 > copy->dstelt_size)
+      return GL_TRUE;
+
+   return GL_FALSE;
+}
+
+
+/**
+ * Dump the parameters/info for a vbo->draw() call.
+ */
+static void
+dump_draw_info(struct gl_context *ctx,
+               const struct tnl_vertex_array *arrays,
+               const struct _mesa_prim *prims,
+               GLuint nr_prims,
+               const struct _mesa_index_buffer *ib,
+               GLuint min_index,
+               GLuint max_index)
+{
+   GLuint i, j;
+
+   printf("VBO Draw:\n");
+   for (i = 0; i < nr_prims; i++) {
+      printf("Prim %u of %u\n", i, nr_prims);
+      printf("  Prim mode 0x%x\n", prims[i].mode);
+      printf("  IB: %p\n", (void*) ib);
+      for (j = 0; j < VERT_ATTRIB_MAX; j++) {
+         const struct tnl_vertex_array *array = &arrays[j];
+         const struct gl_vertex_buffer_binding *binding
+            = array->BufferBinding;
+         const struct gl_array_attributes *attrib = array->VertexAttrib;
+         const GLubyte *ptr = _mesa_vertex_attrib_address(attrib, binding);
+         printf("    array %d at %p:\n", j, (void*) &arrays[j]);
+         printf("      ptr %p, size %d, type 0x%x, stride %d\n",
+                ptr, attrib->Size, attrib->Type, binding->Stride);
+         if (0) {
+            GLint k = prims[i].start + prims[i].count - 1;
+            GLfloat *last = (GLfloat *) (ptr + binding->Stride * k);
+            printf("        last: %f %f %f\n",
+                   last[0], last[1], last[2]);
+         }
+      }
+   }
+}
+
+
+static void
+flush(struct copy_context *copy)
+{
+   struct gl_context *ctx = copy->ctx;
+   GLuint i;
+
+   /* Set some counters:
+    */
+   copy->dstib.count = copy->dstelt_nr;
+
+#if 0
+   dump_draw_info(copy->ctx,
+                  copy->dstarray,
+                  copy->dstprim,
+                  copy->dstprim_nr,
+                  &copy->dstib,
+                  0,
+                  copy->dstbuf_nr);
+#else
+   (void) dump_draw_info;
+#endif
+
+   copy->draw(ctx,
+              copy->dstarray,
+              copy->dstprim,
+              copy->dstprim_nr,
+              &copy->dstib,
+              GL_TRUE,
+              0,
+              copy->dstbuf_nr - 1,
+              NULL, 0, NULL);
+
+   /* Reset all pointers:
+    */
+   copy->dstprim_nr = 0;
+   copy->dstelt_nr = 0;
+   copy->dstbuf_nr = 0;
+   copy->dstptr = copy->dstbuf;
+
+   /* Clear the vertex cache:
+    */
+   for (i = 0; i < ELT_TABLE_SIZE; i++)
+      copy->vert_cache[i].in = ~0;
+}
+
+
+/**
+ * Called at begin of each primitive during replay.
+ */
+static void
+begin(struct copy_context *copy, GLenum mode, GLboolean begin_flag)
+{
+   struct _mesa_prim *prim = &copy->dstprim[copy->dstprim_nr];
+
+   prim->mode = mode;
+   prim->begin = begin_flag;
+   prim->num_instances = 1;
+}
+
+
+/**
+ * Use a hashtable to attempt to identify recently-emitted vertices
+ * and avoid re-emitting them.
+ */
+static GLuint
+elt(struct copy_context *copy, GLuint elt_idx)
+{
+   GLuint elt = copy->srcelt[elt_idx] + copy->prim->basevertex;
+   GLuint slot = elt & (ELT_TABLE_SIZE-1);
+
+   /* Look up the incoming element in the vertex cache.  Re-emit if
+    * necessary.
+    */
+   if (copy->vert_cache[slot].in != elt) {
+      GLubyte *csr = copy->dstptr;
+      GLuint i;
+
+      for (i = 0; i < copy->nr_varying; i++) {
+         const struct tnl_vertex_array *srcarray = copy->varying[i].array;
+         const struct gl_vertex_buffer_binding* srcbinding
+            = srcarray->BufferBinding;
+         const GLubyte *srcptr
+            = copy->varying[i].src_ptr + elt * srcbinding->Stride;
+
+         memcpy(csr, srcptr, copy->varying[i].size);
+         csr += copy->varying[i].size;
+
+#ifdef NAN_CHECK
+         if (srcarray->Type == GL_FLOAT) {
+            GLuint k;
+            GLfloat *f = (GLfloat *) srcptr;
+            for (k = 0; k < srcarray->Size; k++) {
+               assert(!IS_INF_OR_NAN(f[k]));
+               assert(f[k] <= 1.0e20 && f[k] >= -1.0e20);
+            }
+         }
+#endif
+
+         if (0) {
+            const GLuint *f = (const GLuint *)srcptr;
+            GLuint j;
+            printf("  varying %d: ", i);
+            for (j = 0; j < copy->varying[i].size / 4; j++)
+               printf("%x ", f[j]);
+            printf("\n");
+         }
+      }
+
+      copy->vert_cache[slot].in = elt;
+      copy->vert_cache[slot].out = copy->dstbuf_nr++;
+      copy->dstptr += copy->vertex_size;
+
+      assert(csr == copy->dstptr);
+      assert(copy->dstptr == (copy->dstbuf +
+                              copy->dstbuf_nr * copy->vertex_size));
+   }
+
+   copy->dstelt[copy->dstelt_nr++] = copy->vert_cache[slot].out;
+   return check_flush(copy);
+}
+
+
+/**
+ * Called at end of each primitive during replay.
+ */
+static void
+end(struct copy_context *copy, GLboolean end_flag)
+{
+   struct _mesa_prim *prim = &copy->dstprim[copy->dstprim_nr];
+
+   prim->end = end_flag;
+   prim->count = copy->dstelt_nr - prim->start;
+
+   if (++copy->dstprim_nr == MAX_PRIM || check_flush(copy)) {
+      flush(copy);
+   }
+}
+
+
+static void
+replay_elts(struct copy_context *copy)
+{
+   GLuint i, j, k;
+   GLboolean split;
+
+   for (i = 0; i < copy->nr_prims; i++) {
+      const struct _mesa_prim *prim = &copy->prim[i];
+      const GLuint start = prim->start;
+      GLuint first, incr;
+
+      switch (prim->mode) {
+      case GL_LINE_LOOP:
+         /* Convert to linestrip and emit the final vertex explicitly,
+          * but only in the resultant strip that requires it.
+          */
+         j = 0;
+         while (j != prim->count) {
+            begin(copy, GL_LINE_STRIP, prim->begin && j == 0);
+
+            for (split = GL_FALSE; j != prim->count && !split; j++)
+               split = elt(copy, start + j);
+
+            if (j == prim->count) {
+               /* Done, emit final line.  Split doesn't matter as
+                * it is always raised a bit early so we can emit
+                * the last verts if necessary!
+                */
+               if (prim->end)
+                  (void)elt(copy, start + 0);
+
+               end(copy, prim->end);
+            }
+            else {
+               /* Wrap
+                */
+               assert(split);
+               end(copy, 0);
+               j--;
+            }
+         }
+         break;
+
+      case GL_TRIANGLE_FAN:
+      case GL_POLYGON:
+         j = 2;
+         while (j != prim->count) {
+            begin(copy, prim->mode, prim->begin && j == 0);
+
+            split = elt(copy, start+0);
+            assert(!split);
+
+            split = elt(copy, start+j-1);
+            assert(!split);
+
+            for (; j != prim->count && !split; j++)
+               split = elt(copy, start+j);
+
+            end(copy, prim->end && j == prim->count);
+
+            if (j != prim->count) {
+               /* Wrapped the primitive, need to repeat some vertices:
+                */
+               j -= 1;
+            }
+         }
+         break;
+
+      default:
+         (void)_tnl_split_prim_inplace(prim->mode, &first, &incr);
+
+         j = 0;
+         while (j != prim->count) {
+
+            begin(copy, prim->mode, prim->begin && j == 0);
+
+            split = 0;
+            for (k = 0; k < first; k++, j++)
+               split |= elt(copy, start+j);
+
+            assert(!split);
+
+            for (; j != prim->count && !split;)
+               for (k = 0; k < incr; k++, j++)
+                  split |= elt(copy, start+j);
+
+            end(copy, prim->end && j == prim->count);
+
+            if (j != prim->count) {
+               /* Wrapped the primitive, need to repeat some vertices:
+                */
+               assert(j > first - incr);
+               j -= (first - incr);
+            }
+         }
+         break;
+      }
+   }
+
+   if (copy->dstprim_nr)
+      flush(copy);
+}
+
+
+static void
+replay_init(struct copy_context *copy)
+{
+   struct gl_context *ctx = copy->ctx;
+   GLuint i;
+   GLuint offset;
+   const GLvoid *srcptr;
+
+   /* Make a list of varying attributes and their vbo's.  Also
+    * calculate vertex size.
+    */
+   copy->vertex_size = 0;
+   for (i = 0; i < VERT_ATTRIB_MAX; i++) {
+      const struct tnl_vertex_array *array = &copy->array[i];
+      const struct gl_vertex_buffer_binding *binding = array->BufferBinding;
+
+      if (binding->Stride == 0) {
+         copy_vertex_array(&copy->dstarray[i], array);
+      }
+      else {
+         const struct gl_array_attributes *attrib = array->VertexAttrib;
+         struct gl_buffer_object *vbo = binding->BufferObj;
+         const GLubyte *ptr = _mesa_vertex_attrib_address(attrib, binding);
+         GLuint j = copy->nr_varying++;
+
+         copy->varying[j].attr = i;
+         copy->varying[j].array = &copy->array[i];
+         copy->varying[j].size = attr_size(attrib);
+         copy->vertex_size += attr_size(attrib);
+
+         if (_mesa_is_bufferobj(vbo) &&
+             !_mesa_bufferobj_mapped(vbo, MAP_INTERNAL))
+            ctx->Driver.MapBufferRange(ctx, 0, vbo->Size, GL_MAP_READ_BIT, vbo,
+                                       MAP_INTERNAL);
+
+         copy->varying[j].src_ptr =
+               ADD_POINTERS(vbo->Mappings[MAP_INTERNAL].Pointer, ptr);
+
+         copy->dstarray[i].VertexAttrib = &copy->varying[j].dstattribs;
+         copy->dstarray[i].BufferBinding = &copy->varying[j].dstbinding;
+      }
+   }
+
+   /* There must always be an index buffer.  Currently require the
+    * caller convert non-indexed prims to indexed.  Could alternately
+    * do it internally.
+    */
+   if (_mesa_is_bufferobj(copy->ib->obj) &&
+       !_mesa_bufferobj_mapped(copy->ib->obj, MAP_INTERNAL))
+      ctx->Driver.MapBufferRange(ctx, 0, copy->ib->obj->Size, GL_MAP_READ_BIT,
+                                 copy->ib->obj, MAP_INTERNAL);
+
+   srcptr = (const GLubyte *)
+            ADD_POINTERS(copy->ib->obj->Mappings[MAP_INTERNAL].Pointer,
+                         copy->ib->ptr);
+
+   switch (copy->ib->index_size) {
+   case 1:
+      copy->translated_elt_buf = malloc(sizeof(GLuint) * copy->ib->count);
+      copy->srcelt = copy->translated_elt_buf;
+
+      for (i = 0; i < copy->ib->count; i++)
+         copy->translated_elt_buf[i] = ((const GLubyte *)srcptr)[i];
+      break;
+
+   case 2:
+      copy->translated_elt_buf = malloc(sizeof(GLuint) * copy->ib->count);
+      copy->srcelt = copy->translated_elt_buf;
+
+      for (i = 0; i < copy->ib->count; i++)
+         copy->translated_elt_buf[i] = ((const GLushort *)srcptr)[i];
+      break;
+
+   case 4:
+      copy->translated_elt_buf = NULL;
+      copy->srcelt = (const GLuint *)srcptr;
+      break;
+   }
+
+   /* Figure out the maximum allowed vertex buffer size:
+    */
+   if (copy->vertex_size * copy->limits->max_verts <= copy->limits->max_vb_size) {
+      copy->dstbuf_size = copy->limits->max_verts;
+   }
+   else {
+      copy->dstbuf_size = copy->limits->max_vb_size / copy->vertex_size;
+   }
+
+   /* Allocate an output vertex buffer:
+    *
+    * XXX:  This should be a VBO!
+    */
+   copy->dstbuf = malloc(copy->dstbuf_size * copy->vertex_size);
+   copy->dstptr = copy->dstbuf;
+
+   /* Setup new vertex arrays to point into the output buffer:
+    */
+   for (offset = 0, i = 0; i < copy->nr_varying; i++) {
+      const struct tnl_vertex_array *src = copy->varying[i].array;
+      const struct gl_array_attributes *srcattr = src->VertexAttrib;
+      struct tnl_vertex_array *dst = &copy->dstarray[copy->varying[i].attr];
+      struct gl_vertex_buffer_binding *dstbind = &copy->varying[i].dstbinding;
+      struct gl_array_attributes *dstattr = &copy->varying[i].dstattribs;
+
+      dstattr->Size = srcattr->Size;
+      dstattr->Type = srcattr->Type;
+      dstattr->Format = GL_RGBA;
+      dstbind->Stride = copy->vertex_size;
+      dstattr->Ptr = copy->dstbuf + offset;
+      dstattr->Normalized = srcattr->Normalized;
+      dstattr->Integer = srcattr->Integer;
+      dstattr->Doubles = srcattr->Doubles;
+      dstbind->BufferObj = ctx->Shared->NullBufferObj;
+      dstattr->_ElementSize = srcattr->_ElementSize;
+      dst->BufferBinding = dstbind;
+      dst->VertexAttrib = dstattr;
+
+      offset += copy->varying[i].size;
+   }
+
+   /* Allocate an output element list:
+    */
+   copy->dstelt_size = MIN2(65536, copy->ib->count * 2 + 3);
+   copy->dstelt_size = MIN2(copy->dstelt_size, copy->limits->max_indices);
+   copy->dstelt = malloc(sizeof(GLuint) * copy->dstelt_size);
+   copy->dstelt_nr = 0;
+
+   /* Setup the new index buffer to point to the allocated element
+    * list:
+    */
+   copy->dstib.count = 0;        /* duplicates dstelt_nr */
+   copy->dstib.index_size = 4;
+   copy->dstib.obj = ctx->Shared->NullBufferObj;
+   copy->dstib.ptr = copy->dstelt;
+}
+
+
+/**
+ * Free up everything allocated during split/replay.
+ */
+static void
+replay_finish(struct copy_context *copy)
+{
+   struct gl_context *ctx = copy->ctx;
+   GLuint i;
+
+   /* Free our vertex and index buffers */
+   free(copy->translated_elt_buf);
+   free(copy->dstbuf);
+   free(copy->dstelt);
+
+   /* Unmap VBO's */
+   for (i = 0; i < copy->nr_varying; i++) {
+      struct gl_buffer_object *vbo =
+         copy->varying[i].array->BufferBinding->BufferObj;
+      if (_mesa_is_bufferobj(vbo) && _mesa_bufferobj_mapped(vbo, MAP_INTERNAL))
+         ctx->Driver.UnmapBuffer(ctx, vbo, MAP_INTERNAL);
+   }
+
+   /* Unmap index buffer */
+   if (_mesa_is_bufferobj(copy->ib->obj) &&
+       _mesa_bufferobj_mapped(copy->ib->obj, MAP_INTERNAL)) {
+      ctx->Driver.UnmapBuffer(ctx, copy->ib->obj, MAP_INTERNAL);
+   }
+}
+
+
+/**
+ * Split VBO into smaller pieces, draw the pieces.
+ */
+void
+_tnl_split_copy(struct gl_context *ctx,
+                const struct tnl_vertex_array *arrays,
+                const struct _mesa_prim *prim,
+                GLuint nr_prims,
+                const struct _mesa_index_buffer *ib,
+                tnl_draw_func draw,
+                const struct split_limits *limits)
+{
+   struct copy_context copy;
+   GLuint i, this_nr_prims;
+
+   for (i = 0; i < nr_prims;) {
+      /* Our SW TNL pipeline doesn't handle basevertex yet, so bind_indices
+       * will rebase the elements to the basevertex, and we'll only
+       * emit strings of prims with the same basevertex in one draw call.
+       */
+      for (this_nr_prims = 1; i + this_nr_prims < nr_prims;
+           this_nr_prims++) {
+         if (prim[i].basevertex != prim[i + this_nr_prims].basevertex)
+            break;
+      }
+
+      memset(&copy, 0, sizeof(copy));
+
+      /* Require indexed primitives:
+       */
+      assert(ib);
+
+      copy.ctx = ctx;
+      copy.array = arrays;
+      copy.prim = &prim[i];
+      copy.nr_prims = this_nr_prims;
+      copy.ib = ib;
+      copy.draw = draw;
+      copy.limits = limits;
+
+      /* Clear the vertex cache:
+       */
+      for (i = 0; i < ELT_TABLE_SIZE; i++)
+         copy.vert_cache[i].in = ~0;
+
+      replay_init(&copy);
+      replay_elts(&copy);
+      replay_finish(&copy);
+   }
+}
diff --git a/lib/mesa/src/mesa/tnl/t_split_inplace.c b/lib/mesa/src/mesa/tnl/t_split_inplace.c
new file mode 100644
index 000000000..8e9ecb704
--- /dev/null
+++ b/lib/mesa/src/mesa/tnl/t_split_inplace.c
@@ -0,0 +1,298 @@
+
+/*
+ * Mesa 3-D graphics library
+ *
+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Keith Whitwell <keithw@vmware.com>
+ */
+
+
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "vbo/vbo.h"
+
+#include "t_split.h"
+
+
+#define MAX_PRIM 32
+
+/* Used for splitting without copying. No attempt is made to handle
+ * too large indexed vertex buffers: In general you need to copy to do
+ * that.
+ */
+struct split_context {
+   struct gl_context *ctx;
+   const struct tnl_vertex_array *array;
+   const struct _mesa_prim *prim;
+   GLuint nr_prims;
+   const struct _mesa_index_buffer *ib;
+   GLuint min_index;
+   GLuint max_index;
+   tnl_draw_func draw;
+
+   const struct split_limits *limits;
+   GLuint limit;
+
+   struct _mesa_prim dstprim[MAX_PRIM];
+   GLuint dstprim_nr;
+};
+
+
+
+
+static void
+flush_vertex( struct split_context *split)
+{
+   struct gl_context *ctx = split->ctx;
+   struct _mesa_index_buffer ib;
+   GLuint i;
+
+   if (!split->dstprim_nr)
+      return;
+
+   if (split->ib) {
+      ib = *split->ib;
+
+      ib.count = split->max_index - split->min_index + 1;
+      ib.ptr = (const void *)((const char *)ib.ptr +
+                              split->min_index * ib.index_size);
+
+      /* Rebase the primitives to save index buffer entries. */
+      for (i = 0; i < split->dstprim_nr; i++)
+         split->dstprim[i].start -= split->min_index;
+   }
+
+   assert(split->max_index >= split->min_index);
+
+   split->draw(ctx,
+               split->array,
+               split->dstprim,
+               split->dstprim_nr,
+               split->ib ? &ib : NULL,
+               !split->ib,
+               split->min_index,
+               split->max_index,
+               NULL, 0, NULL);
+
+   split->dstprim_nr = 0;
+   split->min_index = ~0;
+   split->max_index = 0;
+}
+
+
+static struct _mesa_prim *
+next_outprim(struct split_context *split)
+{
+   if (split->dstprim_nr == MAX_PRIM-1) {
+      flush_vertex(split);
+   }
+
+   {
+      struct _mesa_prim *prim = &split->dstprim[split->dstprim_nr++];
+      memset(prim, 0, sizeof(*prim));
+      return prim;
+   }
+}
+
+
+static void
+update_index_bounds(struct split_context *split,
+                    const struct _mesa_prim *prim)
+{
+   split->min_index = MIN2(split->min_index, prim->start);
+   split->max_index = MAX2(split->max_index, prim->start + prim->count - 1);
+}
+
+
+/* Return the maximum amount of vertices that can be emitted for a
+ * primitive starting at 'prim->start', depending on the previous
+ * index bounds.
+ */
+static GLuint
+get_max_vertices(struct split_context *split,
+                 const struct _mesa_prim *prim)
+{
+   if ((prim->start > split->min_index &&
+        prim->start - split->min_index >= split->limit) ||
+       (prim->start < split->max_index &&
+        split->max_index - prim->start >= split->limit))
+      /* "prim" starts too far away from the old range. */
+      return 0;
+
+   return MIN2(split->min_index, prim->start) + split->limit - prim->start;
+}
+
+
+/* Break large primitives into smaller ones.  If not possible, convert
+ * the primitive to indexed and pass to split_elts().
+ */
+static void
+split_prims(struct split_context *split)
+{
+   GLuint i;
+
+   for (i = 0; i < split->nr_prims; i++) {
+      const struct _mesa_prim *prim = &split->prim[i];
+      GLuint first, incr;
+      GLboolean split_inplace =
+         _tnl_split_prim_inplace(prim->mode, &first, &incr);
+      GLuint available = get_max_vertices(split, prim);
+      GLuint count = prim->count - (prim->count - first) % incr;
+
+      if (prim->count < first)
+         continue;
+
+      if ((available < count && !split_inplace) ||
+          (available < first && split_inplace)) {
+         flush_vertex(split);
+         available = get_max_vertices(split, prim);
+      }
+
+      if (available >= count) {
+         struct _mesa_prim *outprim = next_outprim(split);
+
+         *outprim = *prim;
+         update_index_bounds(split, outprim);
+      }
+      else if (split_inplace) {
+         GLuint j, nr;
+
+         for (j = 0 ; j < count ;) {
+            GLuint remaining = count - j;
+            struct _mesa_prim *outprim = next_outprim(split);
+
+            nr = MIN2(available, remaining);
+            nr -= (nr - first) % incr;
+
+            outprim->mode = prim->mode;
+            outprim->begin = (j == 0 && prim->begin);
+            outprim->end = (nr == remaining && prim->end);
+            outprim->start = prim->start + j;
+            outprim->count = nr;
+            outprim->num_instances = prim->num_instances;
+            outprim->base_instance = prim->base_instance;
+
+            update_index_bounds(split, outprim);
+
+            if (nr == remaining) {
+               /* Finished */
+               j += nr;
+            }
+            else {
+               /* Wrapped the primitive */
+               j += nr - (first - incr);
+               flush_vertex(split);
+               available = get_max_vertices(split, prim);
+            }
+         }
+      }
+      else if (split->ib == NULL) {
+         /* XXX: could at least send the first max_verts off from the
+          * inplace buffers.
+          */
+
+         /* else convert to indexed primitive and pass to split_elts,
+          * which will do the necessary copying and turn it back into a
+          * vertex primitive for rendering...
+          */
+         struct _mesa_index_buffer ib;
+         struct _mesa_prim tmpprim;
+         GLuint *elts = malloc(count * sizeof(GLuint));
+         GLuint j;
+
+         for (j = 0; j < count; j++)
+            elts[j] = prim->start + j;
+
+         ib.count = count;
+         ib.index_size = 4;
+         ib.obj = split->ctx->Shared->NullBufferObj;
+         ib.ptr = elts;
+
+         tmpprim = *prim;
+         tmpprim.indexed = 1;
+         tmpprim.start = 0;
+         tmpprim.count = count;
+         tmpprim.num_instances = 1;
+         tmpprim.base_instance = 0;
+
+         flush_vertex(split);
+
+         _tnl_split_copy(split->ctx,
+                         split->array,
+                         &tmpprim, 1,
+                         &ib,
+                         split->draw,
+                         split->limits);
+
+         free(elts);
+      }
+      else {
+         flush_vertex(split);
+
+         _tnl_split_copy(split->ctx,
+                         split->array,
+                         prim, 1,
+                         split->ib,
+                         split->draw,
+                         split->limits);
+      }
+   }
+
+   flush_vertex(split);
+}
+
+
+void
+_tnl_split_inplace(struct gl_context *ctx,
+                   const struct tnl_vertex_array *arrays,
+                   const struct _mesa_prim *prim,
+                   GLuint nr_prims,
+                   const struct _mesa_index_buffer *ib,
+                   GLuint min_index,
+                   GLuint max_index,
+                   tnl_draw_func draw,
+                   const struct split_limits *limits)
+{
+   struct split_context split;
+
+   memset(&split, 0, sizeof(split));
+
+   split.ctx = ctx;
+   split.array = arrays;
+   split.prim = prim;
+   split.nr_prims = nr_prims;
+   split.ib = ib;
+
+   /* Empty interval, makes calculations simpler. */
+   split.min_index = ~0;
+   split.max_index = 0;
+
+   split.draw = draw;
+   split.limits = limits;
+   split.limit = ib ? limits->max_indices : limits->max_verts;
+
+   split_prims(&split);
+}
+
+
diff --git a/lib/mesa/src/mesa/tnl/t_vb_fog.c b/lib/mesa/src/mesa/tnl/t_vb_fog.c
index 5489ed685..c224a4192 100644
--- a/lib/mesa/src/mesa/tnl/t_vb_fog.c
+++ b/lib/mesa/src/mesa/tnl/t_vb_fog.c
@@ -27,6 +27,7 @@
 
 
 #include "c99_math.h"
+#include "main/errors.h"
 #include "main/glheader.h"
 #include "main/macros.h"
 #include "main/imports.h"
diff --git a/lib/mesa/src/mesa/tnl/t_vb_texgen.c b/lib/mesa/src/mesa/tnl/t_vb_texgen.c
index 94066f4f6..bd585324e 100644
--- a/lib/mesa/src/mesa/tnl/t_vb_texgen.c
+++ b/lib/mesa/src/mesa/tnl/t_vb_texgen.c
@@ -34,6 +34,7 @@
  * including any use thereof or modifications thereto.
  */
 
+#include "main/errors.h"
 #include "main/glheader.h"
 #include "main/macros.h"
 #include "main/imports.h"
@@ -337,7 +338,8 @@ static void texgen( struct gl_context *ctx,
    struct vertex_buffer *VB = &tnl->vb;
    GLvector4f *in = VB->AttribPtr[VERT_ATTRIB_TEX0 + unit];
    GLvector4f *out = &store->texcoord[unit];
-   const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
+   const struct gl_fixedfunc_texture_unit *texUnit =
+      &ctx->Texture.FixedFuncUnit[unit];
    const GLvector4f *obj = VB->AttribPtr[_TNL_ATTRIB_POS];
    const GLvector4f *eye = VB->EyePtr;
    const GLvector4f *normal = VB->AttribPtr[_TNL_ATTRIB_NORMAL];
@@ -489,10 +491,10 @@ static GLboolean run_texgen_stage( struct gl_context *ctx,
       return GL_TRUE;
 
    for (i = 0 ; i < ctx->Const.MaxTextureCoordUnits ; i++) {
-      struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
+      struct gl_fixedfunc_texture_unit *texUnit =
+         &ctx->Texture.FixedFuncUnit[i];
 
       if (texUnit->TexGenEnabled) {
-
 	 store->TexgenFunc[i]( ctx, store, i );
 
          VB->AttribPtr[VERT_ATTRIB_TEX0 + i] = &store->texcoord[i];
@@ -513,7 +515,8 @@ static void validate_texgen_stage( struct gl_context *ctx,
       return;
 
    for (i = 0 ; i < ctx->Const.MaxTextureCoordUnits ; i++) {
-      struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
+      struct gl_fixedfunc_texture_unit *texUnit =
+         &ctx->Texture.FixedFuncUnit[i];
 
       if (texUnit->TexGenEnabled) {
 	 GLuint sz;
diff --git a/lib/mesa/src/mesa/tnl/t_vb_vertex.c b/lib/mesa/src/mesa/tnl/t_vb_vertex.c
index 71a32b495..4fb3659a3 100644
--- a/lib/mesa/src/mesa/tnl/t_vb_vertex.c
+++ b/lib/mesa/src/mesa/tnl/t_vb_vertex.c
@@ -123,7 +123,7 @@ tnl_clip_prepare(struct gl_context *ctx)
    /* Neither the x86 nor sparc asm cliptest functions have been updated
     * for ARB_depth_clamp, so force the C paths.
     */
-   if (ctx->Transform.DepthClamp) {
+   if (ctx->Transform.DepthClampNear && ctx->Transform.DepthClampFar) {
       static GLboolean c_funcs_installed = GL_FALSE;
       if (!c_funcs_installed) {
          init_c_cliptest();
@@ -191,7 +191,8 @@ static GLboolean run_vertex_stage( struct gl_context *ctx,
 					    store->clipmask,
 					    &store->ormask,
 					    &store->andmask,
-					    !ctx->Transform.DepthClamp );
+					    !(ctx->Transform.DepthClampNear &&
+                                              ctx->Transform.DepthClampFar) );
    }
    else {
       VB->NdcPtr = NULL;
@@ -200,7 +201,8 @@ static GLboolean run_vertex_stage( struct gl_context *ctx,
 					    store->clipmask,
 					    &store->ormask,
 					    &store->andmask,
-					    !ctx->Transform.DepthClamp );
+					    !(ctx->Transform.DepthClampNear &&
+                                              ctx->Transform.DepthClampFar) );
    }
 
    if (store->andmask)
diff --git a/lib/mesa/src/mesa/vbo/vbo_noop.h b/lib/mesa/src/mesa/vbo/vbo_noop.h
index f61cd5136..0ca1bfa8e 100644
--- a/lib/mesa/src/mesa/vbo/vbo_noop.h
+++ b/lib/mesa/src/mesa/vbo/vbo_noop.h
@@ -23,12 +23,12 @@
  * OTHER DEALINGS IN THE SOFTWARE.
  */
 
-#ifndef _API_NOOP_H
-#define _API_NOOP_H
+#ifndef VBO_NOOP_H
+#define VBO_NOOP_H
 
+#include "main/dd.h"
 
-#include "main/mtypes.h"
-
+struct _glapi_table;
 
 extern void
 _mesa_noop_vtxfmt_init(GLvertexformat *vfmt);
@@ -37,4 +37,4 @@ extern GLboolean
 _mesa_using_noop_vtxfmt(const struct _glapi_table *dispatch);
 
 
-#endif /* _API_NOOP_H */
+#endif /* VBO_NOOP_H */
diff --git a/lib/mesa/src/mesa/vbo/vbo_private.h b/lib/mesa/src/mesa/vbo/vbo_private.h
new file mode 100644
index 000000000..86f6b41b7
--- /dev/null
+++ b/lib/mesa/src/mesa/vbo/vbo_private.h
@@ -0,0 +1,232 @@
+/*
+ * mesa 3-D graphics library
+ *
+ * Copyright (C) 1999-2006  Brian Paul   All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+/**
+ * Types, functions, etc which are private to the VBO module.
+ */
+
+
+#ifndef VBO_PRIVATE_H
+#define VBO_PRIVATE_H
+
+
+#include "vbo/vbo_attrib.h"
+#include "vbo/vbo_exec.h"
+#include "vbo/vbo_save.h"
+#include "main/varray.h"
+
+
+struct _glapi_table;
+struct _mesa_prim;
+
+
+struct vbo_context {
+   struct gl_vertex_buffer_binding binding;
+   struct gl_array_attributes current[VBO_ATTRIB_MAX];
+
+   struct gl_vertex_array_object *VAO;
+
+   struct vbo_exec_context exec;
+   struct vbo_save_context save;
+};
+
+
+static inline struct vbo_context *
+vbo_context(struct gl_context *ctx)
+{
+   return ctx->vbo_context;
+}
+
+
+static inline const struct vbo_context *
+vbo_context_const(const struct gl_context *ctx)
+{
+   return ctx->vbo_context;
+}
+
+
+/**
+ * Array to apply the fixed function material aliasing map to
+ * an attribute value used in vbo processing inputs to an attribute
+ * as they appear in the vao.
+ */
+extern const GLubyte
+_vbo_attribute_alias_map[VP_MODE_MAX][VERT_ATTRIB_MAX];
+
+
+/**
+ * Return if format is integer. The immediate mode commands only emit floats
+ * for non-integer types, thus everything else is integer.
+ */
+static inline GLboolean
+vbo_attrtype_to_integer_flag(GLenum format)
+{
+   switch (format) {
+   case GL_FLOAT:
+   case GL_DOUBLE:
+      return GL_FALSE;
+   case GL_INT:
+   case GL_UNSIGNED_INT:
+   case GL_UNSIGNED_INT64_ARB:
+      return GL_TRUE;
+   default:
+      unreachable("Bad vertex attribute type");
+      return GL_FALSE;
+   }
+}
+
+static inline GLboolean
+vbo_attrtype_to_double_flag(GLenum format)
+{
+   switch (format) {
+   case GL_FLOAT:
+   case GL_INT:
+   case GL_UNSIGNED_INT:
+      return GL_FALSE;
+   case GL_UNSIGNED_INT64_ARB:
+   case GL_DOUBLE:
+      return GL_TRUE;
+   default:
+      unreachable("Bad vertex attribute type");
+      return GL_FALSE;
+   }
+}
+
+
+/**
+ * Return default component values for the given format.
+ * The return type is an array of fi_types, because that's how we declare
+ * the vertex storage : floats , integers or unsigned integers.
+ */
+static inline const fi_type *
+vbo_get_default_vals_as_union(GLenum format)
+{
+   static const GLfloat default_float[4] = { 0, 0, 0, 1 };
+   static const GLint default_int[4] = { 0, 0, 0, 1 };
+
+   switch (format) {
+   case GL_FLOAT:
+      return (fi_type *)default_float;
+   case GL_INT:
+   case GL_UNSIGNED_INT:
+      return (fi_type *)default_int;
+   default:
+      unreachable("Bad vertex format");
+      return NULL;
+   }
+}
+
+
+/**
+ * Compute the max number of vertices which can be stored in
+ * a vertex buffer, given the current vertex size, and the amount
+ * of space already used.
+ */
+static inline unsigned
+vbo_compute_max_verts(const struct vbo_exec_context *exec)
+{
+   unsigned n = (VBO_VERT_BUFFER_SIZE - exec->vtx.buffer_used) /
+      (exec->vtx.vertex_size * sizeof(GLfloat));
+   if (n == 0)
+      return 0;
+   /* Subtract one so we're always sure to have room for an extra
+    * vertex for GL_LINE_LOOP -> GL_LINE_STRIP conversion.
+    */
+   n--;
+   return n;
+}
+
+
+void
+vbo_try_prim_conversion(struct _mesa_prim *p);
+
+bool
+vbo_can_merge_prims(const struct _mesa_prim *p0, const struct _mesa_prim *p1);
+
+void
+vbo_merge_prims(struct _mesa_prim *p0, const struct _mesa_prim *p1);
+
+
+/**
+ * Get the filter mask for vbo draws depending on the vertex_processing_mode.
+ */
+static inline GLbitfield
+_vbo_get_vao_filter(gl_vertex_processing_mode vertex_processing_mode)
+{
+   if (vertex_processing_mode == VP_MODE_FF) {
+      /* The materials mapped into the generic arrays */
+      return VERT_BIT_FF_ALL | VERT_BIT_MAT_ALL;
+   } else {
+      return VERT_BIT_ALL;
+   }
+}
+
+
+/**
+ * Translate the bitmask of VBO_ATTRIB_BITs to VERT_ATTRIB_BITS.
+ * Note that position/generic0 attribute aliasing is done
+ * generically in the VAO.
+ */
+static inline GLbitfield
+_vbo_get_vao_enabled_from_vbo(gl_vertex_processing_mode vertex_processing_mode,
+                              GLbitfield64 enabled)
+{
+   if (vertex_processing_mode == VP_MODE_FF) {
+      /* The materials mapped into the generic arrays */
+      return (((GLbitfield)enabled) & VERT_BIT_FF_ALL)
+         | (((GLbitfield)(enabled >> VBO_MATERIAL_SHIFT)) & VERT_BIT_MAT_ALL);
+   } else {
+      return ((GLbitfield)enabled) & VERT_BIT_ALL;
+   }
+}
+
+
+/**
+ * Set the vertex attrib for vbo draw use.
+ */
+static inline void
+_vbo_set_attrib_format(struct gl_context *ctx,
+                       struct gl_vertex_array_object *vao,
+                       gl_vert_attrib attr, GLintptr buffer_offset,
+                       GLubyte size, GLenum16 type, GLuint offset)
+{
+   const GLboolean integer = vbo_attrtype_to_integer_flag(type);
+   const GLboolean doubles = vbo_attrtype_to_double_flag(type);
+
+   if (doubles)
+      size /= 2;
+   _mesa_update_array_format(ctx, vao, attr, size, type, GL_RGBA,
+                             GL_FALSE, integer, doubles, offset);
+   /* Ptr for userspace arrays.
+    * For updating the pointer we would need to add the vao->NewArrays flag
+    * to the VAO. But but that is done already unconditionally in
+    * _mesa_update_array_format called above.
+    */
+   assert((vao->NewArrays | ~vao->_Enabled) & VERT_BIT(attr));
+   vao->VertexAttrib[attr].Ptr = ADD_POINTERS(buffer_offset, offset);
+}
+
+
+#endif /* VBO_PRIVATE_H */
diff --git a/lib/mesa/src/meson.build b/lib/mesa/src/meson.build
new file mode 100644
index 000000000..3b91c6a88
--- /dev/null
+++ b/lib/mesa/src/meson.build
@@ -0,0 +1,116 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+inc_common = include_directories(
+  '../include', '.', 'mapi', 'mesa', 'gallium/include', 'gallium/auxiliary')
+inc_mesa = include_directories('mesa')
+inc_mapi = include_directories('mapi')
+inc_src = include_directories('.')
+inc_gallium = include_directories('gallium/include')
+inc_gallium_aux = include_directories('gallium/auxiliary')
+inc_amd_common = include_directories('amd/common')
+
+libglsl_util = static_library(
+  'glsl_util',
+  files(
+    'mesa/main/extensions_table.c', 'mesa/main/imports.c',
+    'mesa/program/prog_parameter.c', 'mesa/program/symbol_table.c',
+    'mesa/program/dummy_errors.c',
+  ),
+  include_directories : [inc_common],
+  c_args : [c_vis_args],
+  build_by_default : false,
+)
+
+sha1_h = custom_target(
+  'git_sha1.h',
+  output : 'git_sha1.h',
+  command : [prog_python, git_sha1_gen_py, '--output', '@OUTPUT@'],
+  build_always : true, # commit sha1 can change without having touched these files
+)
+
+subdir('gtest')
+subdir('util')
+subdir('mapi')
+# TODO: opengl
+subdir('compiler')
+if with_platform_wayland
+  subdir('egl/wayland/wayland-drm')
+endif
+if with_any_vk
+  subdir('vulkan')
+endif
+if with_gallium_radeonsi or with_amd_vk
+  subdir('amd')
+endif
+if with_gallium_vc4 or with_gallium_v3d
+  subdir('broadcom')
+endif
+if with_dri_i965 or with_intel_vk
+  subdir('intel')
+endif
+subdir('mesa')
+subdir('loader')
+if with_platform_haiku
+  subdir('hgl')
+endif
+if with_glx == 'dri'
+  subdir('glx')
+endif
+if with_gbm
+  subdir('gbm')
+else
+  inc_gbm = []
+endif
+if with_egl
+  subdir('egl')
+endif
+if with_gallium
+  subdir('gallium')
+  # This has to be here since it requires libgallium, and subdir cannot
+  # contain ..
+  if with_tests and with_shared_glapi
+    subdir('mesa/state_tracker/tests')
+  endif
+endif
+
+# This must be after at least mesa, glx, and gallium, since libgl will be
+# defined in one of those subdirs depending on the glx provider.
+if with_glx != 'disabled'
+  # If using glvnd the pkg-config header should not point to GL_mesa, it should
+  # point to GL. glvnd is only available on unix like platforms so adding -l
+  # should be safe here
+  # TODO: in the glvnd case glvnd itself should really be providing this.
+  if with_glvnd
+    _gl = '-L${libdir} -lGL'
+  else
+    _gl = libgl
+  endif
+
+  pkg.generate(
+    name : 'gl',
+    description : 'Mesa OpenGL Library',
+    version : meson.project_version(),
+    libraries : _gl,
+    libraries_private : gl_priv_libs,
+    requires_private : gl_priv_reqs,
+    variables : ['glx_tls=yes'],
+  )
+endif
diff --git a/lib/mesa/src/util/00-mesa-defaults.conf b/lib/mesa/src/util/00-mesa-defaults.conf
new file mode 100644
index 000000000..a937c46d0
--- /dev/null
+++ b/lib/mesa/src/util/00-mesa-defaults.conf
@@ -0,0 +1,338 @@
+<!--
+
+============================================
+Application bugs worked around in this file:
+============================================
+
+* Unigine Heaven 3.0 and older contain too many bugs and can't be supported
+  by drivers that want to be compliant.
+
+* Various Unigine products don't use the #version and #extension GLSL
+  directives, meaning they only get GLSL 1.10 and no extensions for their
+  shaders.
+  Enabling all extensions for Unigine fixes most issues, but the GLSL version
+  is still 1.10.
+
+* If ARB_sample_shading is supported, Unigine Heaven 4.0 and Valley 1.0 uses
+  an #extension directive in the middle of its shaders, which is illegal
+  in GLSL.
+
+* Dying Light and Dead Island Definitive Edition redeclare vertex shader
+  built-ins (specifically gl_VertexID), which causes the vertex shaders to fail
+  to compile.
+
+TODO: document the other workarounds.
+
+-->
+
+<driconf>
+    <!-- Please always enable app-specific workarounds for all drivers and
+         screens. -->
+    <device>
+        <application name="Unigine Sanctuary" executable="Sanctuary">
+            <option name="force_glsl_extensions_warn" value="true" />
+            <option name="disable_blend_func_extended" value="true" />
+        </application>
+
+        <application name="Unigine Tropics" executable="Tropics">
+            <option name="force_glsl_extensions_warn" value="true" />
+            <option name="disable_blend_func_extended" value="true" />
+        </application>
+
+        <application name="Unigine Heaven (32-bit)" executable="heaven_x86">
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+            <!-- remove dual_color_blend_by_location if 4.1 ever comes out -->
+            <option name="dual_color_blend_by_location" value="true" />
+        </application>
+
+        <application name="Unigine Heaven (64-bit)" executable="heaven_x64">
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+            <!-- remove dual_color_blend_by_location if 4.1 ever comes out -->
+            <option name="dual_color_blend_by_location" value="true" />
+        </application>
+
+        <application name="Unigine Valley (32-bit)" executable="valley_x86">
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+            <!-- remove dual_color_blend_by_location if 1.1 ever comes out -->
+            <option name="dual_color_blend_by_location" value="true" />
+        </application>
+
+        <application name="Unigine Valley (64-bit)" executable="valley_x64">
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+            <!-- remove dual_color_blend_by_location if 1.1 ever comes out -->
+            <option name="dual_color_blend_by_location" value="true" />
+        </application>
+
+        <application name="Unigine OilRush (32-bit)" executable="OilRush_x86">
+            <option name="disable_blend_func_extended" value="true" />
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+        </application>
+
+        <application name="Unigine OilRush (64-bit)" executable="OilRush_x64">
+            <option name="disable_blend_func_extended" value="true" />
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+        </application>
+
+        <application name="Savage 2" executable="savage2.bin">
+            <option name="disable_glsl_line_continuations" value="true" />
+        </application>
+
+        <application name="Topogun (32-bit)" executable="topogun32">
+            <option name="always_have_depth_buffer" value="true" />
+        </application>
+
+        <application name="Topogun (64-bit)" executable="topogun64">
+            <option name="always_have_depth_buffer" value="true" />
+        </application>
+
+        <application name="Dead Island (incl. Definitive Edition)" executable="DeadIslandGame">
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+
+            <!-- For the Definitive Edition which shares the same executable name -->
+            <option name="allow_glsl_builtin_variable_redeclaration" value="true" />
+        </application>
+
+        <application name="Dead Island Riptide Definitive Edition" executable="DeadIslandRiptideGame">
+            <option name="allow_glsl_builtin_variable_redeclaration" value="true" />
+        </application>
+
+        <application name="Dying Light" executable="DyingLightGame">
+            <option name="allow_glsl_builtin_variable_redeclaration" value="true" />
+        </application>
+
+        <application name="RAGE (64-bit)" executable="Rage64.exe">
+            <option name="allow_glsl_builtin_variable_redeclaration" value="true" />
+        </application>
+
+        <application name="RAGE (32-bit)" executable="Rage.exe">
+            <option name="allow_glsl_builtin_variable_redeclaration" value="true" />
+        </application>
+
+        <application name="Second Life" executable="do-not-directly-run-secondlife-bin">
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+        </application>
+
+        <application name="Warsow (32-bit)" executable="warsow.i386">
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+        </application>
+
+        <application name="Warsow (64-bit)" executable="warsow.x86_64">
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+        </application>
+
+        <application name="Rust" executable="rust">
+            <option name="glsl_zero_init" value="true"/>
+        </application>
+
+        <application name="Divinity: Original Sin Enhanced Edition" executable="EoCApp">
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+        </application>
+
+        <application name="Metro 2033 Redux / Metro Last Night Redux" executable="metro">
+            <option name="allow_glsl_extension_directive_midshader" value="true" />
+        </application>
+
+        <application name="Worms W.M.D" executable="Worms W.M.Dx64">
+            <option name="allow_higher_compat_version" value="true" />
+        </application>
+
+        <application name="Crookz - The Big Heist" executable="Crookz">
+            <option name="allow_higher_compat_version" value="true" />
+        </application>
+
+        <application name="Tropico 5" executable="Tropico5">
+            <option name="allow_higher_compat_version" value="true" />
+        </application>
+
+        <application name="The Culling" executable="Victory">
+            <option name="force_glsl_version" value="440" />
+        </application>
+
+        <application name="Spec Ops: The Line (32-bit)" executable="specops.i386">
+            <option name="force_glsl_abs_sqrt" value="true" />
+        </application>
+
+        <application name="Spec Ops: The Line (64-bit)" executable="specops">
+            <option name="force_glsl_abs_sqrt" value="true" />
+        </application>
+
+        <application name="Kerbal Space Program (32-bit)" executable="KSP.x86">
+            <option name="glsl_zero_init" value="true"/>
+        </application>
+
+        <application name="Kerbal Space Program (64-bit)" executable="KSP.x86_64">
+            <option name="glsl_zero_init" value="true"/>
+        </application>
+
+        <application name="Rocket League" executable="RocketLeague">
+            <option name="glsl_correct_derivatives_after_discard" value="true"/>
+        </application>
+
+        <application name="The Witcher 2" executable="witcher2">
+            <option name="glsl_correct_derivatives_after_discard" value="true"/>
+        </application>
+
+        <application name="Unreal 4 Editor" executable="UE4Editor">
+            <option name="allow_glsl_cross_stage_interpolation_mismatch" value="true"/>
+        </application>
+
+        <application name="Observer" executable="TheObserver-Linux-Shipping">
+            <option name="allow_glsl_cross_stage_interpolation_mismatch" value="true"/>
+        </application>
+
+        <application name="Steamroll" executable="Steamroll-Linux-Shipping">
+            <option name="allow_glsl_cross_stage_interpolation_mismatch" value="true"/>
+        </application>
+
+        <application name="Refunct" executable="Refunct-Linux-Shipping">
+            <option name="allow_glsl_cross_stage_interpolation_mismatch" value="true"/>
+        </application>
+
+        <application name="Google Earth VR" executable="Earth.exe">
+            <option name="allow_glsl_builtin_const_expression" value="true"/>
+            <option name="allow_glsl_relaxed_es" value="true"/>
+        </application>
+
+        <application name="No Mans Sky" executable="NMS.exe">
+            <option name="force_glsl_extensions_warn" value="true" />
+            <option name="allow_glsl_layout_qualifier_on_function_parameters" value="true" />
+        </application>
+
+        <application name="Wolfenstein The Old Blood" executable="WolfOldBlood_x64.exe">
+            <option name="force_compat_profile" value="true" />
+        </application>
+
+        <application name="ARMA 3" executable="arma3.x86_64">
+            <option name="glsl_correct_derivatives_after_discard" value="true"/>
+        </application>
+
+        <!-- The GL thread whitelist is below, workarounds are above.
+             Keep it that way. -->
+
+        <application name="Alien Isolation" executable="AlienIsolation">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="BioShock Infinite" executable="bioshock.i386">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="Borderlands 2" executable="Borderlands2">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="Civilization 5" executable="Civ5XP">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+        <application name="Civilization 6" executable="Civ6">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="Dreamfall Chapters" executable="Dreamfall Chapters">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="Hitman" executable="HitmanPro">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="Renowned Explorers: International Society" executable="abbeycore_steam">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="Saints Row 2" executable="saintsrow2.i386">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+        <application name="Saints Row: The Third" executable="SaintsRow3.i386">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+        <application name="Saints Row IV" executable="SaintsRow4.i386">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+        <application name="Saints Row: Gat out of Hell" executable="SaintsRow4GooH.i386">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="Sid Meier's: Civilization Beyond Earth" executable="CivBE">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="The Witcher 2" executable="witcher2">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="American Truck Simulator" executable="amtrucks">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+        <application name="Euro Truck Simulator 2" executable="eurotrucks2">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="Overlord" executable="overlord.i386">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+        <application name="Overlord 2" executable="overlord2.i386">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="Oil Rush" executable="OilRush_x86">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="War Thunder" executable="aces">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+        <application name="War Thunder (Wine)" executable="aces.exe">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="Outlast" executable="OLGame.x86_64">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <application name="Spec Ops: The Line (32-bit)" executable="specops.i386">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+        <application name="Spec Ops: The Line (64-bit)" executable="specops">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+        <application name="Mount and Blade Warband" executable="mb_warband_linux">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <!-- around 18% performance increase in min and avg fps, max fps capped at 60fps. -->
+        <application name="Medieval II: Total War" executable="Medieval2">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <!-- min fps ~21 ===> ~27 while standing still in game, also higher gpu load. -->
+        <application name="Carnivores: Dinosaur Hunter Reborn (wine)" executable="Carnivores-master.exe">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+
+        <!-- around 30% increase in avg fps -->
+        <application name="Far Cry 2 (wine)" executable="farcry2.exe">
+            <option name="mesa_glthread" value="true"/>
+        </application>
+    </device>
+    <!-- vmwgfx doesn't like full buffer swaps and can't sync to vertical retraces.-->
+    <device driver="vmwgfx">
+        <application name="gnome-shell" executable="gnome-shell">
+            <option name="glx_disable_ext_buffer_age" value="true" />
+            <option name="glx_disable_oml_sync_control" value="true" />
+            <option name="glx_disable_sgi_video_sync" value="true" />
+        </application>
+        <application name="Compiz" executable="Compiz">
+            <option name="glx_disable_ext_buffer_age" value="true" />
+            <option name="glx_disable_oml_sync_control" value="true" />
+        </application>
+    </device>
+    <device driver="radeonsi">
+        <application name="ARK: Survival Evolved (and unintentionally the UE4 demo template)" executable="ShooterGame">
+            <option name="radeonsi_clear_db_cache_before_clear" value="true" />
+        </application>
+        <application name="No Mans Sky" executable="NMS.exe">
+            <option name="radeonsi_zerovram" value="true" />
+        </application>
+    </device>
+</driconf>
diff --git a/lib/mesa/src/util/bigmath.h b/lib/mesa/src/util/bigmath.h
new file mode 100644
index 000000000..6339bb6f6
--- /dev/null
+++ b/lib/mesa/src/util/bigmath.h
@@ -0,0 +1,112 @@
+/*
+ * Copyright © 2018 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef UTIL_BIGMATH_H
+#define UTIL_BIGMATH_H
+
+#include "macros.h"
+
+#include <assert.h>
+#include <stdint.h>
+#include <string.h>
+
+static inline bool
+_ubm_add_u32arr(uint32_t *dst, unsigned dst_len,
+                uint32_t *a, unsigned a_len,
+                uint32_t *b, unsigned b_len)
+{
+   uint32_t carry = 0;
+   for (unsigned i = 0; i < dst_len; i++) {
+      uint64_t sum = carry;
+      if (i < a_len)
+         sum += a[i];
+      if (i < b_len)
+         sum += b[i];
+      dst[i] = sum;
+      carry = sum >> 32;
+   }
+
+   /* Now compute overflow */
+
+   for (unsigned i = dst_len; i < a_len; i++) {
+      if (a[i])
+         return true;
+   }
+
+   for (unsigned i = dst_len; i < b_len; i++) {
+      if (b[i])
+         return true;
+   }
+
+   return carry;
+}
+#define ubm_add_u32arr(dst, a, b) \
+   _ubm_add_u32arr(dst, ARRAY_SIZE(dst), a, ARRAY_SIZE(a), b, ARRAY_SIZE(b))
+
+static inline bool
+_ubm_mul_u32arr(uint32_t *dst, unsigned dst_len,
+                uint32_t *a, unsigned a_len,
+                uint32_t *b, unsigned b_len)
+{
+   memset(dst, 0, dst_len * sizeof(*dst));
+
+   bool overflow = false;
+
+   for (unsigned i = 0; i < a_len; i++) {
+      uint32_t carry = 0;
+      for (unsigned j = 0; j < b_len; j++) {
+         /* The maximum values of a[i] and b[i] are UINT32_MAX so the maximum
+          * value of tmp is UINT32_MAX * UINT32_MAX.  The maximum value that
+          * will fit in tmp is
+          *
+          *    UINT64_MAX = UINT32_MAX << 32 + UINT32_MAX
+          *               = UINT32_MAX * (UINT32_MAX + 1) + UINT32_MAX
+          *               = UINT32_MAX * UINT32_MAX + 2 * UINT32_MAX
+          *
+          * so we're guaranteed that we can add in two more 32-bit values
+          * without overflowing tmp.
+          */
+         uint64_t tmp = (uint64_t)a[i] * (uint64_t)b[j];
+         tmp += carry;
+         if (i + j < dst_len) {
+            tmp += dst[i + j];
+            dst[i + j] = tmp;
+            carry = tmp >> 32;
+         } else {
+            /* We're trying to write a value that doesn't fit */
+            overflow = overflow || tmp > 0;
+            break;
+         }
+      }
+      if (i + b_len < dst_len)
+         dst[i + b_len] = carry;
+      else
+         overflow = overflow || carry > 0;
+   }
+
+   return overflow;
+}
+#define ubm_mul_u32arr(dst, a, b) \
+   _ubm_mul_u32arr(dst, ARRAY_SIZE(dst), a, ARRAY_SIZE(a), b, ARRAY_SIZE(b))
+
+#endif /* UTIL_BIGMATH_H */
diff --git a/lib/mesa/src/util/bitset.h b/lib/mesa/src/util/bitset.h
index 2404ce7f6..3b18abac7 100644
--- a/lib/mesa/src/util/bitset.h
+++ b/lib/mesa/src/util/bitset.h
@@ -31,7 +31,8 @@
 #ifndef BITSET_H
 #define BITSET_H
 
-#include "util/u_math.h"
+#include "util/bitscan.h"
+#include "util/macros.h"
 
 /****************************************************************************
  * generic bitset implementation
@@ -53,7 +54,7 @@
 #define BITSET_ONES(x) memset( (x), 0xff, sizeof (x) )
 
 #define BITSET_BITWORD(b) ((b) / BITSET_WORDBITS)
-#define BITSET_BIT(b) (1 << ((b) % BITSET_WORDBITS))
+#define BITSET_BIT(b) (1u << ((b) % BITSET_WORDBITS))
 
 /* single bit operations
  */
@@ -132,4 +133,121 @@ __bitset_next_set(unsigned i, BITSET_WORD *tmp,
    for (__tmp = *(__set), __i = 0; \
         (__i = __bitset_next_set(__i, &__tmp, __set, __size)) < __size;)
 
+#ifdef __cplusplus
+
+/**
+ * Simple C++ wrapper of a bitset type of static size, with value semantics
+ * and basic bitwise arithmetic operators.  The operators defined below are
+ * expected to have the same semantics as the same operator applied to other
+ * fundamental integer types.  T is the name of the struct to instantiate
+ * it as, and N is the number of bits in the bitset.
+ */
+#define DECLARE_BITSET_T(T, N) struct T {                       \
+      EXPLICIT_CONVERSION                                       \
+      operator bool() const                                     \
+      {                                                         \
+         for (unsigned i = 0; i < BITSET_WORDS(N); i++)         \
+            if (words[i])                                       \
+               return true;                                     \
+         return false;                                          \
+      }                                                         \
+                                                                \
+      T &                                                       \
+      operator=(int x)                                          \
+      {                                                         \
+         const T c = {{ (BITSET_WORD)x }};                      \
+         return *this = c;                                      \
+      }                                                         \
+                                                                \
+      friend bool                                               \
+      operator==(const T &b, const T &c)                        \
+      {                                                         \
+         return BITSET_EQUAL(b.words, c.words);                 \
+      }                                                         \
+                                                                \
+      friend bool                                               \
+      operator!=(const T &b, const T &c)                        \
+      {                                                         \
+         return !(b == c);                                      \
+      }                                                         \
+                                                                \
+      friend bool                                               \
+      operator==(const T &b, int x)                             \
+      {                                                         \
+         const T c = {{ (BITSET_WORD)x }};                      \
+         return b == c;                                         \
+      }                                                         \
+                                                                \
+      friend bool                                               \
+      operator!=(const T &b, int x)                             \
+      {                                                         \
+         return !(b == x);                                      \
+      }                                                         \
+                                                                \
+      friend T                                                  \
+      operator~(const T &b)                                     \
+      {                                                         \
+         T c;                                                   \
+         for (unsigned i = 0; i < BITSET_WORDS(N); i++)         \
+            c.words[i] = ~b.words[i];                           \
+         return c;                                              \
+      }                                                         \
+                                                                \
+      T &                                                       \
+      operator|=(const T &b)                                    \
+      {                                                         \
+         for (unsigned i = 0; i < BITSET_WORDS(N); i++)         \
+            words[i] |= b.words[i];                             \
+         return *this;                                          \
+      }                                                         \
+                                                                \
+      friend T                                                  \
+      operator|(const T &b, const T &c)                         \
+      {                                                         \
+         T d = b;                                               \
+         d |= c;                                                \
+         return d;                                              \
+      }                                                         \
+                                                                \
+      T &                                                       \
+      operator&=(const T &b)                                    \
+      {                                                         \
+         for (unsigned i = 0; i < BITSET_WORDS(N); i++)         \
+            words[i] &= b.words[i];                             \
+         return *this;                                          \
+      }                                                         \
+                                                                \
+      friend T                                                  \
+      operator&(const T &b, const T &c)                         \
+      {                                                         \
+         T d = b;                                               \
+         d &= c;                                                \
+         return d;                                              \
+      }                                                         \
+                                                                \
+      bool                                                      \
+      test(unsigned i) const                                    \
+      {                                                         \
+         return BITSET_TEST(words, i);                          \
+      }                                                         \
+                                                                \
+      T &                                                       \
+      set(unsigned i)                                           \
+      {                                                         \
+         BITSET_SET(words, i);                                  \
+         return *this;                                          \
+      }                                                         \
+                                                                \
+      T &                                                       \
+      clear(unsigned i)                                         \
+      {                                                         \
+         BITSET_CLEAR(words, i);                                \
+         return *this;                                          \
+      }                                                         \
+                                                                \
+      BITSET_WORD words[BITSET_WORDS(N)];                       \
+   }
+
+#endif
+
 #endif
diff --git a/lib/mesa/src/util/fast_idiv_by_const.c b/lib/mesa/src/util/fast_idiv_by_const.c
new file mode 100644
index 000000000..7b9331626
--- /dev/null
+++ b/lib/mesa/src/util/fast_idiv_by_const.c
@@ -0,0 +1,243 @@
+/*
+ * Copyright © 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+/* Imported from:
+ *   https://raw.githubusercontent.com/ridiculousfish/libdivide/master/divide_by_constants_codegen_reference.c
+ * Paper:
+ *   http://ridiculousfish.com/files/faster_unsigned_division_by_constants.pdf
+ *
+ * The author, ridiculous_fish, wrote:
+ *
+ *  ''Reference implementations of computing and using the "magic number"
+ *    approach to dividing by constants, including codegen instructions.
+ *    The unsigned division incorporates the "round down" optimization per
+ *    ridiculous_fish.
+ *
+ *    This is free and unencumbered software. Any copyright is dedicated
+ *    to the Public Domain.''
+ */
+
+#include "fast_idiv_by_const.h"
+#include "u_math.h"
+#include <limits.h>
+#include <assert.h>
+
+struct util_fast_udiv_info
+util_compute_fast_udiv_info(uint64_t D, unsigned num_bits, unsigned UINT_BITS)
+{
+   /* The numerator must fit in a uint64_t */
+   assert(num_bits > 0 && num_bits <= UINT_BITS);
+   assert(D != 0);
+
+   /* The eventual result */
+   struct util_fast_udiv_info result;
+
+   if (util_is_power_of_two_or_zero64(D)) {
+      unsigned div_shift = util_logbase2_64(D);
+
+      if (div_shift) {
+         /* Dividing by a power of two. */
+         result.multiplier = 1ull << (UINT_BITS - div_shift);
+         result.pre_shift = 0;
+         result.post_shift = 0;
+         result.increment = 0;
+         return result;
+      } else {
+         /* Dividing by 1. */
+         /* Assuming: floor((num + 1) * (2^32 - 1) / 2^32) = num */
+         result.multiplier = UINT_BITS == 64 ? UINT64_MAX :
+                                               (1ull << UINT_BITS) - 1;
+         result.pre_shift = 0;
+         result.post_shift = 0;
+         result.increment = 1;
+         return result;
+      }
+   }
+
+   /* The extra shift implicit in the difference between UINT_BITS and num_bits
+    */
+   const unsigned extra_shift = UINT_BITS - num_bits;
+
+   /* The initial power of 2 is one less than the first one that can possibly
+    * work.
+    */
+   const uint64_t initial_power_of_2 = (uint64_t)1 << (UINT_BITS-1);
+
+   /* The remainder and quotient of our power of 2 divided by d */
+   uint64_t quotient = initial_power_of_2 / D;
+   uint64_t remainder = initial_power_of_2 % D;
+
+   /* ceil(log_2 D) */
+   unsigned ceil_log_2_D;
+
+   /* The magic info for the variant "round down" algorithm */
+   uint64_t down_multiplier = 0;
+   unsigned down_exponent = 0;
+   int has_magic_down = 0;
+
+   /* Compute ceil(log_2 D) */
+   ceil_log_2_D = 0;
+   uint64_t tmp;
+   for (tmp = D; tmp > 0; tmp >>= 1)
+      ceil_log_2_D += 1;
+
+
+   /* Begin a loop that increments the exponent, until we find a power of 2
+    * that works.
+    */
+   unsigned exponent;
+   for (exponent = 0; ; exponent++) {
+      /* Quotient and remainder is from previous exponent; compute it for this
+       * exponent.
+       */
+      if (remainder >= D - remainder) {
+         /* Doubling remainder will wrap around D */
+         quotient = quotient * 2 + 1;
+         remainder = remainder * 2 - D;
+      } else {
+         /* Remainder will not wrap */
+         quotient = quotient * 2;
+         remainder = remainder * 2;
+      }
+
+      /* We're done if this exponent works for the round_up algorithm.
+       * Note that exponent may be larger than the maximum shift supported,
+       * so the check for >= ceil_log_2_D is critical.
+       */
+      if ((exponent + extra_shift >= ceil_log_2_D) ||
+          (D - remainder) <= ((uint64_t)1 << (exponent + extra_shift)))
+         break;
+
+      /* Set magic_down if we have not set it yet and this exponent works for
+       * the round_down algorithm
+       */
+      if (!has_magic_down &&
+          remainder <= ((uint64_t)1 << (exponent + extra_shift))) {
+         has_magic_down = 1;
+         down_multiplier = quotient;
+         down_exponent = exponent;
+      }
+   }
+
+   if (exponent < ceil_log_2_D) {
+      /* magic_up is efficient */
+      result.multiplier = quotient + 1;
+      result.pre_shift = 0;
+      result.post_shift = exponent;
+      result.increment = 0;
+   } else if (D & 1) {
+      /* Odd divisor, so use magic_down, which must have been set */
+      assert(has_magic_down);
+      result.multiplier = down_multiplier;
+      result.pre_shift = 0;
+      result.post_shift = down_exponent;
+      result.increment = 1;
+   } else {
+      /* Even divisor, so use a prefix-shifted dividend */
+      unsigned pre_shift = 0;
+      uint64_t shifted_D = D;
+      while ((shifted_D & 1) == 0) {
+         shifted_D >>= 1;
+         pre_shift += 1;
+      }
+      result = util_compute_fast_udiv_info(shifted_D, num_bits - pre_shift,
+                                           UINT_BITS);
+      /* expect no increment or pre_shift in this path */
+      assert(result.increment == 0 && result.pre_shift == 0);
+      result.pre_shift = pre_shift;
+   }
+   return result;
+}
+
+static inline int64_t
+sign_extend(int64_t x, unsigned SINT_BITS)
+{
+   return (x << (64 - SINT_BITS)) >> (64 - SINT_BITS);
+}
+
+struct util_fast_sdiv_info
+util_compute_fast_sdiv_info(int64_t D, unsigned SINT_BITS)
+{
+   /* D must not be zero. */
+   assert(D != 0);
+   /* The result is not correct for these divisors. */
+   assert(D != 1 && D != -1);
+
+   /* Our result */
+   struct util_fast_sdiv_info result;
+
+   /* Absolute value of D (we know D is not the most negative value since
+    * that's a power of 2)
+    */
+   const uint64_t abs_d = (D < 0 ? -D : D);
+
+   /* The initial power of 2 is one less than the first one that can possibly
+    * work */
+   /* "two31" in Warren */
+   unsigned exponent = SINT_BITS - 1;
+   const uint64_t initial_power_of_2 = (uint64_t)1 << exponent;
+
+   /* Compute the absolute value of our "test numerator,"
+    * which is the largest dividend whose remainder with d is d-1.
+    * This is called anc in Warren.
+    */
+   const uint64_t tmp = initial_power_of_2 + (D < 0);
+   const uint64_t abs_test_numer = tmp - 1 - tmp % abs_d;
+
+   /* Initialize our quotients and remainders (q1, r1, q2, r2 in Warren) */
+   uint64_t quotient1 = initial_power_of_2 / abs_test_numer;
+   uint64_t remainder1 = initial_power_of_2 % abs_test_numer;
+   uint64_t quotient2 = initial_power_of_2 / abs_d;
+   uint64_t remainder2 = initial_power_of_2 % abs_d;
+   uint64_t delta;
+
+   /* Begin our loop */
+   do {
+      /* Update the exponent */
+      exponent++;
+
+      /* Update quotient1 and remainder1 */
+      quotient1 *= 2;
+      remainder1 *= 2;
+      if (remainder1 >= abs_test_numer) {
+         quotient1 += 1;
+         remainder1 -= abs_test_numer;
+      }
+
+      /* Update quotient2 and remainder2 */
+      quotient2 *= 2;
+      remainder2 *= 2;
+      if (remainder2 >= abs_d) {
+         quotient2 += 1;
+         remainder2 -= abs_d;
+      }
+
+      /* Keep going as long as (2**exponent) / abs_d <= delta */
+      delta = abs_d - remainder2;
+   } while (quotient1 < delta || (quotient1 == delta && remainder1 == 0));
+
+   result.multiplier = sign_extend(quotient2 + 1, SINT_BITS);
+   if (D < 0) result.multiplier = -result.multiplier;
+   result.shift = exponent - SINT_BITS;
+   return result;
+}
diff --git a/lib/mesa/src/util/fast_idiv_by_const.h b/lib/mesa/src/util/fast_idiv_by_const.h
new file mode 100644
index 000000000..638b52a3f
--- /dev/null
+++ b/lib/mesa/src/util/fast_idiv_by_const.h
@@ -0,0 +1,183 @@
+/*
+ * Copyright © 2018 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef FAST_IDIV_BY_CONST_H
+#define FAST_IDIV_BY_CONST_H
+
+/* Imported from:
+ *   https://raw.githubusercontent.com/ridiculousfish/libdivide/master/divide_by_constants_codegen_reference.c
+ */
+
+#include <inttypes.h>
+#include <limits.h>
+#include <assert.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/* Computes "magic info" for performing signed division by a fixed integer D.
+ * The type 'sint_t' is assumed to be defined as a signed integer type large
+ * enough to hold both the dividend and the divisor.
+ * Here >> is arithmetic (signed) shift, and >>> is logical shift.
+ *
+ * To emit code for n/d, rounding towards zero, use the following sequence:
+ *
+ *   m = compute_signed_magic_info(D)
+ *   emit("result = (m.multiplier * n) >> SINT_BITS");
+ *   if d > 0 and m.multiplier < 0: emit("result += n")
+ *   if d < 0 and m.multiplier > 0: emit("result -= n")
+ *   if m.post_shift > 0: emit("result >>= m.shift")
+ *   emit("result += (result < 0)")
+ *
+ * The shifts by SINT_BITS may be "free" if the high half of the full multiply
+ * is put in a separate register.
+ *
+ * The final add can of course be implemented via the sign bit, e.g.
+ *    result += (result >>> (SINT_BITS - 1))
+ * or
+ *    result -= (result >> (SINT_BITS - 1))
+ *
+ * This code is heavily indebted to Hacker's Delight by Henry Warren.
+ * See http://www.hackersdelight.org/HDcode/magic.c.txt
+ * Used with permission from http://www.hackersdelight.org/permissions.htm
+ */
+
+struct util_fast_sdiv_info {
+   int64_t multiplier; /* the "magic number" multiplier */
+   unsigned shift; /* shift for the dividend after multiplying */
+};
+
+struct util_fast_sdiv_info
+util_compute_fast_sdiv_info(int64_t D, unsigned SINT_BITS);
+
+/* Computes "magic info" for performing unsigned division by a fixed positive
+ * integer D.  UINT_BITS is the bit size at which the final "magic"
+ * calculation will be performed; it is assumed to be large enough to hold
+ * both the dividand and the divisor.  num_bits can be set appropriately if n
+ * is known to be smaller than calc_bits; if this is not known then UINT_BITS
+ * for num_bits.
+ *
+ * Assume we have a hardware register of width UINT_BITS, a known constant D
+ * which is not zero and not a power of 2, and a variable n of width num_bits
+ * (which may be up to UINT_BITS). To emit code for n/d, use one of the two
+ * following sequences (here >>> refers to a logical bitshift):
+ *
+ *   m = compute_unsigned_magic_info(D, num_bits)
+ *   if m.pre_shift > 0: emit("n >>>= m.pre_shift")
+ *   if m.increment: emit("n = saturated_increment(n)")
+ *   emit("result = (m.multiplier * n) >>> UINT_BITS")
+ *   if m.post_shift > 0: emit("result >>>= m.post_shift")
+ *
+ * or
+ *
+ *   m = compute_unsigned_magic_info(D, num_bits)
+ *   if m.pre_shift > 0: emit("n >>>= m.pre_shift")
+ *   emit("result = m.multiplier * n")
+ *   if m.increment: emit("result = result + m.multiplier")
+ *   emit("result >>>= UINT_BITS")
+ *   if m.post_shift > 0: emit("result >>>= m.post_shift")
+ *
+ * This second version works even if D is 1.  The shifts by UINT_BITS may be
+ * "free" if the high half of the full multiply is put in a separate register.
+ *
+ * saturated_increment(n) means "increment n unless it would wrap to 0," i.e.
+ *   if n == (1 << UINT_BITS)-1: result = n
+ *   else: result = n+1
+ * A common way to implement this is with the carry bit. For example, on x86:
+ *   add 1
+ *   sbb 0
+ *
+ * Some invariants:
+ *   1: At least one of pre_shift and increment is zero
+ *   2: multiplier is never zero
+ *
+ * This code incorporates the "round down" optimization per ridiculous_fish.
+ */
+
+struct util_fast_udiv_info {
+   uint64_t multiplier; /* the "magic number" multiplier */
+   unsigned pre_shift; /* shift for the dividend before multiplying */
+   unsigned post_shift; /* shift for the dividend after multiplying */
+   int increment; /* 0 or 1; if set then increment the numerator, using one of
+                     the two strategies */
+};
+
+struct util_fast_udiv_info
+util_compute_fast_udiv_info(uint64_t D, unsigned num_bits, unsigned UINT_BITS);
+
+/* Below are possible options for dividing by a uniform in a shader where
+ * the divisor is constant but not known at compile time.
+ */
+
+/* Full version. */
+static inline uint32_t
+util_fast_udiv32(uint32_t n, struct util_fast_udiv_info info)
+{
+   n = n >> info.pre_shift;
+   /* If the divisor is not 1, you can instead use a 32-bit ADD that clamps
+    * to UINT_MAX. Dividing by 1 needs the full 64-bit ADD.
+    *
+    * If you have unsigned 64-bit MAD with 32-bit inputs, you can do:
+    *    increment = increment ? multiplier : 0; // on the CPU
+    *    (n * multiplier + increment) // on the GPU using unsigned 64-bit MAD
+    */
+   n = (((uint64_t)n + info.increment) * info.multiplier) >> 32;
+   n = n >> info.post_shift;
+   return n;
+}
+
+/* A little more efficient version if n != UINT_MAX, i.e. no unsigned
+ * wraparound in the computation.
+ */
+static inline uint32_t
+util_fast_udiv32_nuw(uint32_t n, struct util_fast_udiv_info info)
+{
+   assert(n != UINT32_MAX);
+   n = n >> info.pre_shift;
+   n = n + info.increment;
+   n = ((uint64_t)n * info.multiplier) >> 32;
+   n = n >> info.post_shift;
+   return n;
+}
+
+/* Even faster version but both operands must be 31-bit unsigned integers
+ * and the divisor must be greater than 1.
+ *
+ * info must be computed with num_bits == 31.
+ */
+static inline uint32_t
+util_fast_udiv32_u31_d_not_one(uint32_t n, struct util_fast_udiv_info info)
+{
+   assert(info.pre_shift == 0);
+   assert(info.increment == 0);
+   n = ((uint64_t)n * info.multiplier) >> 32;
+   n = n >> info.post_shift;
+   return n;
+}
+
+#ifdef __cplusplus
+} /* extern C */
+#endif
+
+#endif /* FAST_IDIV_BY_CONST_H */
diff --git a/lib/mesa/src/util/format_srgb.h b/lib/mesa/src/util/format_srgb.h
index 34b50afe3..596af56f4 100644
--- a/lib/mesa/src/util/format_srgb.h
+++ b/lib/mesa/src/util/format_srgb.h
@@ -55,6 +55,20 @@ util_format_linear_to_srgb_helper_table[104];
 
 
 static inline float
+util_format_srgb_to_linear_float(float cs)
+{
+   if (cs <= 0.0f)
+      return 0.0f;
+   else if (cs <= 0.04045f)
+      return cs / 12.92f;
+   else if (cs < 1.0f)
+      return powf((cs + 0.055) / 1.055f, 2.4f);
+   else
+      return 1.0f;
+}
+
+
+static inline float
 util_format_linear_to_srgb_float(float cl)
 {
    if (cl <= 0.0f)
diff --git a/lib/mesa/src/util/futex.h b/lib/mesa/src/util/futex.h
new file mode 100644
index 000000000..440289306
--- /dev/null
+++ b/lib/mesa/src/util/futex.h
@@ -0,0 +1,56 @@
+/*
+ * Copyright © 2015 Intel
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef UTIL_FUTEX_H
+#define UTIL_FUTEX_H
+
+#if defined(HAVE_LINUX_FUTEX_H)
+
+#include <limits.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <linux/futex.h>
+#include <sys/syscall.h>
+#include <sys/time.h>
+
+static inline long sys_futex(void *addr1, int op, int val1, const struct timespec *timeout, void *addr2, int val3)
+{
+   return syscall(SYS_futex, addr1, op, val1, timeout, addr2, val3);
+}
+
+static inline int futex_wake(uint32_t *addr, int count)
+{
+   return sys_futex(addr, FUTEX_WAKE, count, NULL, NULL, 0);
+}
+
+static inline int futex_wait(uint32_t *addr, int32_t value, const struct timespec *timeout)
+{
+   /* FUTEX_WAIT_BITSET with FUTEX_BITSET_MATCH_ANY is equivalent to
+    * FUTEX_WAIT, except that it treats the timeout as absolute. */
+   return sys_futex(addr, FUTEX_WAIT_BITSET, value, timeout, NULL,
+                    FUTEX_BITSET_MATCH_ANY);
+}
+
+#endif
+
+#endif /* UTIL_FUTEX_H */
diff --git a/lib/mesa/src/util/half_float.c b/lib/mesa/src/util/half_float.c
index 4df64c2cc..63aec5c5c 100644
--- a/lib/mesa/src/util/half_float.c
+++ b/lib/mesa/src/util/half_float.c
@@ -2,6 +2,8 @@
  * Mesa 3-D graphics library
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright 2015 Philip Taylor <philip@zaynar.co.uk>
+ * Copyright 2018 Advanced Micro Devices, Inc.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -26,6 +28,7 @@
 #include <assert.h>
 #include "half_float.h"
 #include "rounding.h"
+#include "macros.h"
 
 typedef union { float f; int32_t i; uint32_t u; } fi_type;
 
@@ -175,3 +178,70 @@ _mesa_half_to_float(uint16_t val)
    result = fi.f;
    return result;
 }
+
+/**
+  * Convert 0.0 to 0x00, 1.0 to 0xff.
+  * Values outside the range [0.0, 1.0] will give undefined results.
+  */
+uint8_t _mesa_half_to_unorm8(uint16_t val)
+{
+   const int m = val & 0x3ff;
+   const int e = (val >> 10) & 0x1f;
+   MAYBE_UNUSED const int s = (val >> 15) & 0x1;
+
+   /* v = round_to_nearest(1.mmmmmmmmmm * 2^(e-15) * 255)
+    *   = round_to_nearest((1.mmmmmmmmmm * 255) * 2^(e-15))
+    *   = round_to_nearest((1mmmmmmmmmm * 255) * 2^(e-25))
+    *   = round_to_zero((1mmmmmmmmmm * 255) * 2^(e-25) + 0.5)
+    *   = round_to_zero(((1mmmmmmmmmm * 255) * 2^(e-24) + 1) / 2)
+    *
+    * This happens to give the correct answer for zero/subnormals too
+    */
+   assert(s == 0 && val <= FP16_ONE); /* check 0 <= this <= 1 */
+   /* (implies e <= 15, which means the bit-shifts below are safe) */
+
+   uint32_t v = ((1 << 10) | m) * 255;
+   v = ((v >> (24 - e)) + 1) >> 1;
+   return v;
+}
+
+/**
+  * Takes a uint16_t, divides by 65536, converts the infinite-precision
+  * result to fp16 with round-to-zero. Used by the ASTC decoder.
+  */
+uint16_t _mesa_uint16_div_64k_to_half(uint16_t v)
+{
+   /* Zero or subnormal. Set the mantissa to (v << 8) and return. */
+   if (v < 4)
+      return v << 8;
+
+   /* Count the leading 0s in the uint16_t */
+#ifdef HAVE___BUILTIN_CLZ
+   int n = __builtin_clz(v) - 16;
+#else
+   int n = 16;
+   for (int i = 15; i >= 0; i--) {
+      if (v & (1 << i)) {
+         n = 15 - i;
+         break;
+      }
+   }
+#endif
+
+   /* Shift the mantissa up so bit 16 is the hidden 1 bit,
+    * mask it off, then shift back down to 10 bits
+    */
+   int m = ( ((uint32_t)v << (n + 1)) & 0xffff ) >> 6;
+
+   /*  (0{n} 1 X{15-n}) * 2^-16
+    * = 1.X * 2^(15-n-16)
+    * = 1.X * 2^(14-n - 15)
+    * which is the FP16 form with e = 14 - n
+    */
+   int e = 14 - n;
+
+   assert(e >= 1 && e <= 30);
+   assert(m >= 0 && m < 0x400);
+
+   return (e << 10) | m;
+}
diff --git a/lib/mesa/src/util/half_float.h b/lib/mesa/src/util/half_float.h
index b3bc3f687..015574247 100644
--- a/lib/mesa/src/util/half_float.h
+++ b/lib/mesa/src/util/half_float.h
@@ -32,8 +32,13 @@
 extern "C" {
 #endif
 
+#define FP16_ONE     0x3C00
+#define FP16_ZERO    0
+
 uint16_t _mesa_float_to_half(float val);
 float _mesa_half_to_float(uint16_t val);
+uint8_t _mesa_half_to_unorm8(uint16_t v);
+uint16_t _mesa_uint16_div_64k_to_half(uint16_t v);
 
 static inline bool
 _mesa_half_is_negative(uint16_t h)
diff --git a/lib/mesa/src/util/meson.build b/lib/mesa/src/util/meson.build
new file mode 100644
index 000000000..156621aff
--- /dev/null
+++ b/lib/mesa/src/util/meson.build
@@ -0,0 +1,184 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+inc_util = include_directories('.')
+
+subdir('xmlpool')
+
+files_mesa_util = files(
+  'bigmath.h',
+  'bitscan.c',
+  'bitscan.h',
+  'bitset.h',
+  'build_id.c',
+  'build_id.h',
+  'crc32.c',
+  'crc32.h',
+  'debug.c',
+  'debug.h',
+  'disk_cache.c',
+  'disk_cache.h',
+  'fast_idiv_by_const.c',
+  'fast_idiv_by_const.h',
+  'format_r11g11b10f.h',
+  'format_rgb9e5.h',
+  'format_srgb.h',
+  'futex.h',
+  'half_float.c',
+  'half_float.h',
+  'hash_table.c',
+  'hash_table.h',
+  'list.h',
+  'macros.h',
+  'mesa-sha1.c',
+  'mesa-sha1.h',
+  'os_time.c',
+  'os_time.h',
+  'os_misc.c',
+  'os_misc.h',
+  'u_process.c',
+  'u_process.h',
+  'sha1/sha1.c',
+  'sha1/sha1.h',
+  'ralloc.c',
+  'ralloc.h',
+  'rand_xor.c',
+  'rand_xor.h',
+  'rb_tree.c',
+  'rb_tree.h',
+  'register_allocate.c',
+  'register_allocate.h',
+  'rgtc.c',
+  'rgtc.h',
+  'rounding.h',
+  'set.c',
+  'set.h',
+  'simple_list.h',
+  'simple_mtx.h',
+  'slab.c',
+  'slab.h',
+  'string_buffer.c',
+  'string_buffer.h',
+  'strndup.h',
+  'strtod.c',
+  'strtod.h',
+  'texcompress_rgtc_tmp.h',
+  'u_atomic.c',
+  'u_atomic.h',
+  'u_dynarray.h',
+  'u_endian.h',
+  'u_queue.c',
+  'u_queue.h',
+  'u_string.h',
+  'u_thread.h',
+  'u_vector.c',
+  'u_vector.h',
+  'u_math.c',
+  'u_math.h',
+  'u_debug.c',
+  'u_debug.h',
+  'u_cpu_detect.c',
+  'u_cpu_detect.h',
+  'vma.c',
+  'vma.h',
+)
+
+install_data('00-mesa-defaults.conf', install_dir : join_paths(get_option('datadir'), 'drirc.d'))
+
+files_xmlconfig = files(
+  'xmlconfig.c',
+  'xmlconfig.h',
+)
+
+format_srgb = custom_target(
+  'format_srgb',
+  input : ['format_srgb.py'],
+  output : 'format_srgb.c',
+  command : [prog_python, '@INPUT0@'],
+  capture : true,
+)
+
+libmesa_util = static_library(
+  'mesa_util',
+  [files_mesa_util, format_srgb],
+  include_directories : inc_common,
+  dependencies : [dep_zlib, dep_clock, dep_thread, dep_atomic, dep_m],
+  c_args : [c_msvc_compat_args, c_vis_args],
+  build_by_default : false
+)
+
+libxmlconfig = static_library(
+  'xmlconfig',
+  files_xmlconfig,
+  include_directories : inc_common,
+  link_with : libmesa_util,
+  dependencies : [dep_expat, dep_m],
+  c_args : [
+    c_msvc_compat_args, c_vis_args,
+    '-DSYSCONFDIR="@0@"'.format(
+      join_paths(get_option('prefix'), get_option('sysconfdir'))
+    ),
+    '-DDATADIR="@0@"'.format(
+      join_paths(get_option('prefix'), get_option('datadir'))
+    ),
+  ],
+  build_by_default : false,
+)
+
+if with_tests
+  test(
+    'u_atomic',
+    executable(
+      'u_atomic_test',
+      files('u_atomic_test.c'),
+      include_directories : inc_common,
+      link_with : libmesa_util,
+      c_args : [c_msvc_compat_args],
+    )
+  )
+
+  test(
+    'roundeven',
+    executable(
+      'roundeven_test',
+      files('roundeven_test.c'),
+      include_directories : inc_common,
+      c_args : [c_msvc_compat_args],
+      dependencies : [dep_m],
+    )
+  )
+
+  test(
+    'mesa-sha1',
+    executable(
+      'mesa-sha1_test',
+      files('mesa-sha1_test.c'),
+      include_directories : inc_common,
+      link_with : libmesa_util,
+      c_args : [c_msvc_compat_args],
+    )
+  )
+
+  subdir('tests/fast_idiv_by_const')
+  subdir('tests/hash_table')
+  subdir('tests/string_buffer')
+  subdir('tests/vma')
+  subdir('tests/set')
+endif
diff --git a/lib/mesa/src/util/os_misc.c b/lib/mesa/src/util/os_misc.c
new file mode 100644
index 000000000..09d4400e0
--- /dev/null
+++ b/lib/mesa/src/util/os_misc.c
@@ -0,0 +1,176 @@
+/**************************************************************************
+ *
+ * Copyright 2008-2010 Vmware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+
+#include "os_misc.h"
+
+#include <stdarg.h>
+
+
+#if defined(PIPE_SUBSYSTEM_WINDOWS_USER)
+
+#ifndef WIN32_LEAN_AND_MEAN
+#define WIN32_LEAN_AND_MEAN      // Exclude rarely-used stuff from Windows headers
+#endif
+#include <windows.h>
+#include <stdio.h>
+
+#else
+
+#include <stdio.h>
+#include <stdlib.h>
+
+#endif
+
+
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_CYGWIN) || defined(PIPE_OS_SOLARIS)
+#  include <unistd.h>
+#elif defined(PIPE_OS_APPLE) || defined(PIPE_OS_BSD)
+#  include <sys/sysctl.h>
+#elif defined(PIPE_OS_HAIKU)
+#  include <kernel/OS.h>
+#elif defined(PIPE_OS_WINDOWS)
+#  include <windows.h>
+#else
+#error unexpected platform in os_sysinfo.c
+#endif
+
+
+void
+os_log_message(const char *message)
+{
+   /* If the GALLIUM_LOG_FILE environment variable is set to a valid filename,
+    * write all messages to that file.
+    */
+   static FILE *fout = NULL;
+
+   if (!fout) {
+#ifdef DEBUG
+      /* one-time init */
+      const char *filename = os_get_option("GALLIUM_LOG_FILE");
+      if (filename) {
+         const char *mode = "w";
+         if (filename[0] == '+') {
+            /* If the filename is prefixed with '+' then open the file for
+             * appending instead of normal writing.
+             */
+            mode = "a";
+            filename++; /* skip the '+' */
+         }
+         fout = fopen(filename, mode);
+      }
+#endif
+      if (!fout)
+         fout = stderr;
+   }
+
+#if defined(PIPE_SUBSYSTEM_WINDOWS_USER)
+   OutputDebugStringA(message);
+   if(GetConsoleWindow() && !IsDebuggerPresent()) {
+      fflush(stdout);
+      fputs(message, fout);
+      fflush(fout);
+   }
+   else if (fout != stderr) {
+      fputs(message, fout);
+      fflush(fout);
+   }
+#else /* !PIPE_SUBSYSTEM_WINDOWS */
+   fflush(stdout);
+   fputs(message, fout);
+   fflush(fout);
+#endif
+}
+
+
+#if !defined(PIPE_SUBSYSTEM_EMBEDDED)
+const char *
+os_get_option(const char *name)
+{
+   return getenv(name);
+}
+#endif /* !PIPE_SUBSYSTEM_EMBEDDED */
+
+
+/**
+ * Return the size of the total physical memory.
+ * \param size returns the size of the total physical memory
+ * \return true for success, or false on failure
+ */
+bool
+os_get_total_physical_memory(uint64_t *size)
+{
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_CYGWIN) || defined(PIPE_OS_SOLARIS)
+   const long phys_pages = sysconf(_SC_PHYS_PAGES);
+   const long page_size = sysconf(_SC_PAGE_SIZE);
+
+   if (phys_pages <= 0 || page_size <= 0)
+      return false;
+
+   *size = (uint64_t)phys_pages * (uint64_t)page_size;
+   return true;
+#elif defined(PIPE_OS_APPLE) || defined(PIPE_OS_BSD)
+   size_t len = sizeof(*size);
+   int mib[2];
+
+   mib[0] = CTL_HW;
+#if defined(PIPE_OS_APPLE)
+   mib[1] = HW_MEMSIZE;
+#elif defined(PIPE_OS_NETBSD) || defined(PIPE_OS_OPENBSD)
+   mib[1] = HW_PHYSMEM64;
+#elif defined(PIPE_OS_FREEBSD)
+   mib[1] = HW_REALMEM;
+#elif defined(PIPE_OS_DRAGONFLY)
+   mib[1] = HW_PHYSMEM;
+#else
+#error Unsupported *BSD
+#endif
+
+   return (sysctl(mib, 2, size, &len, NULL, 0) == 0);
+#elif defined(PIPE_OS_HAIKU)
+   system_info info;
+   status_t ret;
+
+   ret = get_system_info(&info);
+   if (ret != B_OK || info.max_pages <= 0)
+      return false;
+
+   *size = (uint64_t)info.max_pages * (uint64_t)B_PAGE_SIZE;
+   return true;
+#elif defined(PIPE_OS_WINDOWS)
+   MEMORYSTATUSEX status;
+   BOOL ret;
+
+   status.dwLength = sizeof(status);
+   ret = GlobalMemoryStatusEx(&status);
+   *size = status.ullTotalPhys;
+   return (ret == TRUE);
+#else
+#error unexpected platform in os_sysinfo.c
+   return false;
+#endif
+}
diff --git a/lib/mesa/src/util/os_misc.h b/lib/mesa/src/util/os_misc.h
new file mode 100644
index 000000000..403c8ee6e
--- /dev/null
+++ b/lib/mesa/src/util/os_misc.h
@@ -0,0 +1,102 @@
+/**************************************************************************
+ *
+ * Copyright 2010 Vmware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+
+/*
+ * Miscellaneous OS services.
+ */
+
+
+#ifndef _OS_MISC_H_
+#define _OS_MISC_H_
+
+
+#include "pipe/p_compiler.h"
+
+
+#if defined(PIPE_OS_UNIX)
+#  include <signal.h> /* for kill() */
+#  include <unistd.h> /* getpid() */
+#endif
+
+
+#ifdef  __cplusplus
+extern "C" {
+#endif
+
+
+/*
+ * Trap into the debugger.
+ */
+#if (defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)) && defined(PIPE_CC_GCC)
+#  define os_break() __asm("int3")
+#elif defined(PIPE_CC_MSVC)
+#  define os_break()  __debugbreak()
+#elif defined(PIPE_OS_UNIX)
+#  define os_break() kill(getpid(), SIGTRAP)
+#else
+#  define os_break() abort()
+#endif
+
+
+/*
+ * Abort the program.
+ */
+#if defined(DEBUG)
+#  define os_abort() do { os_break(); abort(); } while(0)
+#else
+#  define os_abort() abort()
+#endif
+
+
+/*
+ * Output a message. Message should preferably end in a newline.
+ */
+void
+os_log_message(const char *message);
+
+
+/*
+ * Get an option. Should return NULL if specified option is not set.
+ */
+const char *
+os_get_option(const char *name);
+
+
+/*
+ * Get the total amount of physical memory available on the system.
+ */
+bool
+os_get_total_physical_memory(uint64_t *size);
+
+
+#ifdef	__cplusplus
+}
+#endif
+
+
+#endif /* _OS_MISC_H_ */
diff --git a/lib/mesa/src/util/os_time.c b/lib/mesa/src/util/os_time.c
new file mode 100644
index 000000000..ac488b228
--- /dev/null
+++ b/lib/mesa/src/util/os_time.c
@@ -0,0 +1,192 @@
+/**************************************************************************
+ *
+ * Copyright 2008-2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * OS independent time-manipulation functions.
+ * 
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+#include "os_time.h"
+
+/* TODO: fix this dependency */
+#include "gallium/include/pipe/p_config.h"
+
+#include "util/u_atomic.h"
+
+#if defined(PIPE_OS_UNIX)
+#  include <unistd.h> /* usleep */
+#  include <time.h> /* timeval */
+#  include <sys/time.h> /* timeval */
+#  include <sched.h> /* sched_yield */
+#  include <errno.h>
+#elif defined(PIPE_SUBSYSTEM_WINDOWS_USER)
+#  include <windows.h>
+#else
+#  error Unsupported OS
+#endif
+
+
+int64_t
+os_time_get_nano(void)
+{
+#if defined(PIPE_OS_LINUX) || defined(PIPE_OS_BSD)
+
+   struct timespec tv;
+   clock_gettime(CLOCK_MONOTONIC, &tv);
+   return tv.tv_nsec + tv.tv_sec*INT64_C(1000000000);
+
+#elif defined(PIPE_OS_UNIX)
+
+   struct timeval tv;
+   gettimeofday(&tv, NULL);
+   return tv.tv_usec*INT64_C(1000) + tv.tv_sec*INT64_C(1000000000);
+
+#elif defined(PIPE_SUBSYSTEM_WINDOWS_USER)
+
+   static LARGE_INTEGER frequency;
+   LARGE_INTEGER counter;
+   int64_t secs, nanosecs;
+   if(!frequency.QuadPart)
+      QueryPerformanceFrequency(&frequency);
+   QueryPerformanceCounter(&counter);
+   /* Compute seconds and nanoseconds parts separately to
+    * reduce severity of precision loss.
+    */
+   secs = counter.QuadPart / frequency.QuadPart;
+   nanosecs = (counter.QuadPart % frequency.QuadPart) * INT64_C(1000000000)
+      / frequency.QuadPart;
+   return secs*INT64_C(1000000000) + nanosecs;
+
+#else
+
+#error Unsupported OS
+
+#endif
+}
+
+
+
+void
+os_time_sleep(int64_t usecs)
+{
+#if defined(PIPE_OS_LINUX)
+   struct timespec time;
+   time.tv_sec = usecs / 1000000;
+   time.tv_nsec = (usecs % 1000000) * 1000;
+   while (clock_nanosleep(CLOCK_MONOTONIC, 0, &time, &time) == EINTR);
+
+#elif defined(PIPE_OS_UNIX)
+   usleep(usecs);
+
+#elif defined(PIPE_SUBSYSTEM_WINDOWS_USER)
+   DWORD dwMilliseconds = (DWORD) ((usecs + 999) / 1000);
+   /* Avoid Sleep(O) as that would cause to sleep for an undetermined duration */
+   if (dwMilliseconds) {
+      Sleep(dwMilliseconds);
+   }
+#else
+#  error Unsupported OS
+#endif
+}
+
+
+
+int64_t
+os_time_get_absolute_timeout(uint64_t timeout)
+{
+   int64_t time, abs_timeout;
+
+   /* Also check for the type upper bound. */
+   if (timeout == OS_TIMEOUT_INFINITE || timeout > INT64_MAX)
+      return OS_TIMEOUT_INFINITE;
+
+   time = os_time_get_nano();
+   abs_timeout = time + (int64_t)timeout;
+
+   /* Check for overflow. */
+   if (abs_timeout < time)
+      return OS_TIMEOUT_INFINITE;
+
+   return abs_timeout;
+}
+
+
+bool
+os_wait_until_zero(volatile int *var, uint64_t timeout)
+{
+   if (!p_atomic_read(var))
+      return true;
+
+   if (!timeout)
+      return false;
+
+   if (timeout == OS_TIMEOUT_INFINITE) {
+      while (p_atomic_read(var)) {
+#if defined(PIPE_OS_UNIX)
+         sched_yield();
+#endif
+      }
+      return true;
+   }
+   else {
+      int64_t start_time = os_time_get_nano();
+      int64_t end_time = start_time + timeout;
+
+      while (p_atomic_read(var)) {
+         if (os_time_timeout(start_time, end_time, os_time_get_nano()))
+            return false;
+
+#if defined(PIPE_OS_UNIX)
+         sched_yield();
+#endif
+      }
+      return true;
+   }
+}
+
+
+bool
+os_wait_until_zero_abs_timeout(volatile int *var, int64_t timeout)
+{
+   if (!p_atomic_read(var))
+      return true;
+
+   if (timeout == OS_TIMEOUT_INFINITE)
+      return os_wait_until_zero(var, OS_TIMEOUT_INFINITE);
+
+   while (p_atomic_read(var)) {
+      if (os_time_get_nano() >= timeout)
+         return false;
+
+#if defined(PIPE_OS_UNIX)
+      sched_yield();
+#endif
+   }
+   return true;
+}
diff --git a/lib/mesa/src/util/os_time.h b/lib/mesa/src/util/os_time.h
new file mode 100644
index 000000000..049ab118d
--- /dev/null
+++ b/lib/mesa/src/util/os_time.h
@@ -0,0 +1,130 @@
+/**************************************************************************
+ *
+ * Copyright 2008-2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+/**
+ * @file
+ * OS independent time-manipulation functions.
+ * 
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+#ifndef _OS_TIME_H_
+#define _OS_TIME_H_
+
+#include <stdbool.h>
+#include <stdint.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/* must be equal to PIPE_TIMEOUT_INFINITE */
+#define OS_TIMEOUT_INFINITE 0xffffffffffffffffull
+
+/*
+ * Get the current time in nanoseconds from an unknown base.
+ */
+int64_t
+os_time_get_nano(void);
+
+
+/*
+ * Get the current time in microseconds from an unknown base.
+ */
+static inline int64_t
+os_time_get(void)
+{
+   return os_time_get_nano() / 1000;
+}
+
+
+/*
+ * Sleep.
+ */
+void
+os_time_sleep(int64_t usecs);
+
+
+/*
+ * Helper function for detecting time outs, taking in account overflow.
+ *
+ * Returns true if the current time has elapsed beyond the specified interval.
+ */
+static inline bool
+os_time_timeout(int64_t start,
+                int64_t end,
+                int64_t curr)
+{
+   if (start <= end)
+      return !(start <= curr && curr < end);
+   else
+      return !((start <= curr) || (curr < end));
+}
+
+
+/**
+ * Convert a relative timeout in nanoseconds into an absolute timeout,
+ * in other words, it returns current time + timeout.
+ * os_time_get_nano() must be monotonic.
+ * OS_TIMEOUT_INFINITE is passed through unchanged. If the calculation
+ * overflows, OS_TIMEOUT_INFINITE is returned.
+ */
+int64_t
+os_time_get_absolute_timeout(uint64_t timeout);
+
+
+/**
+ * Wait until the variable at the given memory location is zero.
+ *
+ * \param var           variable
+ * \param timeout       timeout in ns, can be anything from 0 (no wait) to
+ *                      OS_TIMEOUT_INFINITE (wait forever)
+ * \return     true if the variable is zero
+ */
+bool
+os_wait_until_zero(volatile int *var, uint64_t timeout);
+
+
+/**
+ * Wait until the variable at the given memory location is zero.
+ * The timeout is the absolute time when the waiting should stop. If it is
+ * less than or equal to the current time, it only returns the status and
+ * doesn't wait. OS_TIMEOUT_INFINITE waits forever. This requires that
+ * os_time_get_nano is monotonic.
+ *
+ * \param var       variable
+ * \param timeout   the time in ns when the waiting should stop
+ * \return     true if the variable is zero
+ */
+bool
+os_wait_until_zero_abs_timeout(volatile int *var, int64_t timeout);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _OS_TIME_H_ */
diff --git a/lib/mesa/src/util/rb_tree.c b/lib/mesa/src/util/rb_tree.c
new file mode 100644
index 000000000..a86fa31a8
--- /dev/null
+++ b/lib/mesa/src/util/rb_tree.c
@@ -0,0 +1,421 @@
+/*
+ * Copyright © 2017 Jason Ekstrand
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ */
+
+#include "rb_tree.h"
+
+/** \file rb_tree.c
+ *
+ * An implementation of a red-black tree
+ *
+ * This file implements the guts of a red-black tree.  The implementation
+ * is mostly based on the one in "Introduction to Algorithms", third
+ * edition, by Cormen, Leiserson, Rivest, and Stein.  The primary
+ * divergence in our algorithms from those presented in CLRS is that we use
+ * NULL for the leaves instead of a sentinel.  This means we have to do a
+ * tiny bit more tracking in our implementation of delete but it makes the
+ * algorithms far more explicit than stashing stuff in the sentinel.
+ */
+
+#include <stdlib.h>
+#include <string.h>
+#include <assert.h>
+
+static bool
+rb_node_is_black(struct rb_node *n)
+{
+    /* NULL nodes are leaves and therefore black */
+    return (n == NULL) || (n->parent & 1);
+}
+
+static bool
+rb_node_is_red(struct rb_node *n)
+{
+    return !rb_node_is_black(n);
+}
+
+static void
+rb_node_set_black(struct rb_node *n)
+{
+    n->parent |= 1;
+}
+
+static void
+rb_node_set_red(struct rb_node *n)
+{
+    n->parent &= ~1ull;
+}
+
+static void
+rb_node_copy_color(struct rb_node *dst, struct rb_node *src)
+{
+    dst->parent = (dst->parent & ~1ull) | (src->parent & 1);
+}
+
+static void
+rb_node_set_parent(struct rb_node *n, struct rb_node *p)
+{
+    n->parent = (n->parent & 1) | (uintptr_t)p;
+}
+
+static struct rb_node *
+rb_node_minimum(struct rb_node *node)
+{
+    while (node->left)
+        node = node->left;
+    return node;
+}
+
+static struct rb_node *
+rb_node_maximum(struct rb_node *node)
+{
+    while (node->right)
+        node = node->right;
+    return node;
+}
+
+void
+rb_tree_init(struct rb_tree *T)
+{
+    T->root = NULL;
+}
+
+/**
+ * Replace the subtree of T rooted at u with the subtree rooted at v
+ *
+ * This is called RB-transplant in CLRS.
+ *
+ * The node to be replaced is assumed to be a non-leaf.
+ */
+static void
+rb_tree_splice(struct rb_tree *T, struct rb_node *u, struct rb_node *v)
+{
+    assert(u);
+    struct rb_node *p = rb_node_parent(u);
+    if (p == NULL) {
+        assert(T->root == u);
+        T->root = v;
+    } else if (u == p->left) {
+        p->left = v;
+    } else {
+        assert(u == p->right);
+        p->right = v;
+    }
+    if (v)
+        rb_node_set_parent(v, p);
+}
+
+static void
+rb_tree_rotate_left(struct rb_tree *T, struct rb_node *x)
+{
+    assert(x && x->right);
+
+    struct rb_node *y = x->right;
+    x->right = y->left;
+    if (y->left)
+        rb_node_set_parent(y->left, x);
+    rb_tree_splice(T, x, y);
+    y->left = x;
+    rb_node_set_parent(x, y);
+}
+
+static void
+rb_tree_rotate_right(struct rb_tree *T, struct rb_node *y)
+{
+    assert(y && y->left);
+
+    struct rb_node *x = y->left;
+    y->left = x->right;
+    if (x->right)
+        rb_node_set_parent(x->right, y);
+    rb_tree_splice(T, y, x);
+    x->right = y;
+    rb_node_set_parent(y, x);
+}
+
+void
+rb_tree_insert_at(struct rb_tree *T, struct rb_node *parent,
+                  struct rb_node *node, bool insert_left)
+{
+    /* This sets null children, parent, and a color of red */
+    memset(node, 0, sizeof(*node));
+
+    if (parent == NULL) {
+        assert(T->root == NULL);
+        T->root = node;
+        rb_node_set_black(node);
+        return;
+    }
+
+    if (insert_left) {
+        assert(parent->left == NULL);
+        parent->left = node;
+    } else {
+        assert(parent->right == NULL);
+        parent->right = node;
+    }
+    rb_node_set_parent(node, parent);
+
+    /* Now we do the insertion fixup */
+    struct rb_node *z = node;
+    while (rb_node_is_red(rb_node_parent(z))) {
+        struct rb_node *z_p = rb_node_parent(z);
+        assert(z == z_p->left || z == z_p->right);
+        struct rb_node *z_p_p = rb_node_parent(z_p);
+        assert(z_p_p != NULL);
+        if (z_p == z_p_p->left) {
+            struct rb_node *y = z_p_p->right;
+            if (rb_node_is_red(y)) {
+                rb_node_set_black(z_p);
+                rb_node_set_black(y);
+                rb_node_set_red(z_p_p);
+                z = z_p_p;
+            } else {
+                if (z == z_p->right) {
+                    z = z_p;
+                    rb_tree_rotate_left(T, z);
+                    /* We changed z */
+                    z_p = rb_node_parent(z);
+                    assert(z == z_p->left || z == z_p->right);
+                    z_p_p = rb_node_parent(z_p);
+                }
+                rb_node_set_black(z_p);
+                rb_node_set_red(z_p_p);
+                rb_tree_rotate_right(T, z_p_p);
+            }
+        } else {
+            struct rb_node *y = z_p_p->left;
+            if (rb_node_is_red(y)) {
+                rb_node_set_black(z_p);
+                rb_node_set_black(y);
+                rb_node_set_red(z_p_p);
+                z = z_p_p;
+            } else {
+                if (z == z_p->left) {
+                    z = z_p;
+                    rb_tree_rotate_right(T, z);
+                    /* We changed z */
+                    z_p = rb_node_parent(z);
+                    assert(z == z_p->left || z == z_p->right);
+                    z_p_p = rb_node_parent(z_p);
+                }
+                rb_node_set_black(z_p);
+                rb_node_set_red(z_p_p);
+                rb_tree_rotate_left(T, z_p_p);
+            }
+        }
+    }
+    rb_node_set_black(T->root);
+}
+
+void
+rb_tree_remove(struct rb_tree *T, struct rb_node *z)
+{
+    /* x_p is always the parent node of X.  We have to track this
+     * separately because x may be NULL.
+     */
+    struct rb_node *x, *x_p;
+    struct rb_node *y = z;
+    bool y_was_black = rb_node_is_black(y);
+    if (z->left == NULL) {
+        x = z->right;
+        x_p = rb_node_parent(z);
+        rb_tree_splice(T, z, x);
+    } else if (z->right == NULL) {
+        x = z->left;
+        x_p = rb_node_parent(z);
+        rb_tree_splice(T, z, x);
+    } else {
+        /* Find the minimum sub-node of z->right */
+        y = rb_node_minimum(z->right);
+        y_was_black = rb_node_is_black(y);
+
+        x = y->right;
+        if (rb_node_parent(y) == z) {
+            x_p = y;
+        } else {
+            x_p = rb_node_parent(y);
+            rb_tree_splice(T, y, x);
+            y->right = z->right;
+            rb_node_set_parent(y->right, y);
+        }
+        assert(y->left == NULL);
+        rb_tree_splice(T, z, y);
+        y->left = z->left;
+        rb_node_set_parent(y->left, y);
+        rb_node_copy_color(y, z);
+    }
+
+    assert(x_p == NULL || x == x_p->left || x == x_p->right);
+
+    if (!y_was_black)
+        return;
+
+    /* Fixup RB tree after the delete */
+    while (x != T->root && rb_node_is_black(x)) {
+        if (x == x_p->left) {
+            struct rb_node *w = x_p->right;
+            if (rb_node_is_red(w)) {
+                rb_node_set_black(w);
+                rb_node_set_red(x_p);
+                rb_tree_rotate_left(T, x_p);
+                assert(x == x_p->left);
+                w = x_p->right;
+            }
+            if (rb_node_is_black(w->left) && rb_node_is_black(w->right)) {
+                rb_node_set_red(w);
+                x = x_p;
+            } else {
+                if (rb_node_is_black(w->right)) {
+                    rb_node_set_black(w->left);
+                    rb_node_set_red(w);
+                    rb_tree_rotate_right(T, w);
+                    w = x_p->right;
+                }
+                rb_node_copy_color(w, x_p);
+                rb_node_set_black(x_p);
+                rb_node_set_black(w->right);
+                rb_tree_rotate_left(T, x_p);
+                x = T->root;
+            }
+        } else {
+            struct rb_node *w = x_p->left;
+            if (rb_node_is_red(w)) {
+                rb_node_set_black(w);
+                rb_node_set_red(x_p);
+                rb_tree_rotate_right(T, x_p);
+                assert(x == x_p->right);
+                w = x_p->left;
+            }
+            if (rb_node_is_black(w->right) && rb_node_is_black(w->left)) {
+                rb_node_set_red(w);
+                x = x_p;
+            } else {
+                if (rb_node_is_black(w->left)) {
+                    rb_node_set_black(w->right);
+                    rb_node_set_red(w);
+                    rb_tree_rotate_left(T, w);
+                    w = x_p->left;
+                }
+                rb_node_copy_color(w, x_p);
+                rb_node_set_black(x_p);
+                rb_node_set_black(w->left);
+                rb_tree_rotate_right(T, x_p);
+                x = T->root;
+            }
+        }
+        x_p = rb_node_parent(x);
+    }
+    if (x)
+        rb_node_set_black(x);
+}
+
+struct rb_node *
+rb_tree_first(struct rb_tree *T)
+{
+    return T->root ? rb_node_minimum(T->root) : NULL;
+}
+
+struct rb_node *
+rb_tree_last(struct rb_tree *T)
+{
+    return T->root ? rb_node_maximum(T->root) : NULL;
+}
+
+struct rb_node *
+rb_node_next(struct rb_node *node)
+{
+    if (node->right) {
+        /* If we have a right child, then the next thing (compared to this
+         * node) is the left-most child of our right child.
+         */
+        return rb_node_minimum(node->right);
+    } else {
+        /* If node doesn't have a right child, crawl back up the to the
+         * left until we hit a parent to the right.
+         */
+        struct rb_node *p = rb_node_parent(node);
+        while (p && node == p->right) {
+            node = p;
+            p = rb_node_parent(node);
+        }
+        assert(p == NULL || node == p->left);
+        return p;
+    }
+}
+
+struct rb_node *
+rb_node_prev(struct rb_node *node)
+{
+    if (node->left) {
+        /* If we have a left child, then the previous thing (compared to
+         * this node) is the right-most child of our left child.
+         */
+        return rb_node_maximum(node->left);
+    } else {
+        /* If node doesn't have a left child, crawl back up the to the
+         * right until we hit a parent to the left.
+         */
+        struct rb_node *p = rb_node_parent(node);
+        while (p && node == p->left) {
+            node = p;
+            p = rb_node_parent(node);
+        }
+        assert(p == NULL || node == p->right);
+        return p;
+    }
+}
+
+static void
+validate_rb_node(struct rb_node *n, int black_depth)
+{
+    if (n == NULL) {
+        assert(black_depth == 0);
+        return;
+    }
+
+    if (rb_node_is_black(n)) {
+        black_depth--;
+    } else {
+        assert(rb_node_is_black(n->left));
+        assert(rb_node_is_black(n->right));
+    }
+
+    validate_rb_node(n->left, black_depth);
+    validate_rb_node(n->right, black_depth);
+}
+
+void
+rb_tree_validate(struct rb_tree *T)
+{
+    if (T->root == NULL)
+        return;
+
+    assert(rb_node_is_black(T->root));
+
+    unsigned black_depth = 0;
+    for (struct rb_node *n = T->root; n; n = n->left) {
+        if (rb_node_is_black(n))
+            black_depth++;
+    }
+
+    validate_rb_node(T->root, black_depth);
+}
diff --git a/lib/mesa/src/util/rb_tree.h b/lib/mesa/src/util/rb_tree.h
new file mode 100644
index 000000000..1e8aeb4a7
--- /dev/null
+++ b/lib/mesa/src/util/rb_tree.h
@@ -0,0 +1,327 @@
+/*
+ * Copyright © 2017 Jason Ekstrand
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ */
+
+#ifndef RB_TREE_H
+#define RB_TREE_H
+
+#include <stdbool.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <stdlib.h>
+
+/** A red-black tree node
+ *
+ * This struct represents a node in the red-black tree.  This struct should
+ * be embedded as a member in whatever structure you wish to put in the
+ * tree.
+ */
+struct rb_node {
+    /** Parent and color of this node
+     *
+     * The least significant bit represents the color and is est to 1 for
+     * black and 0 for red.  The other bits are the pointer to the parent
+     * and that pointer can be retrieved by masking off the bottom bit and
+     * casting to a pointer.
+     */
+    uintptr_t parent;
+
+    /** Left child of this node, null for a leaf */
+    struct rb_node *left;
+
+    /** Right child of this node, null for a leaf */
+    struct rb_node *right;
+};
+
+/** Return the parent node of the given node or NULL if it is the root */
+static inline struct rb_node *
+rb_node_parent(struct rb_node *n)
+{
+    return (struct rb_node *)(n->parent & ~(uintptr_t)1);
+}
+
+/** A red-black tree
+ *
+ * This struct represents the red-black tree itself.  It is just a pointer
+ * to the root node with no other metadata.
+ */
+struct rb_tree {
+    struct rb_node *root;
+};
+
+/** Initialize a red-black tree */
+void rb_tree_init(struct rb_tree *T);
+
+/** Returns true if the red-black tree is empty */
+static inline bool
+rb_tree_is_empty(const struct rb_tree *T)
+{
+    return T->root == NULL;
+}
+
+/** Retrieve the data structure containing a node
+ *
+ * \param   type    The type of the containing data structure
+ *
+ * \param   node    A pointer to a rb_node
+ *
+ * \param   field   The rb_node field in the containing data structure
+ */
+#define rb_node_data(type, node, field) \
+    ((type *)(((char *)(node)) - offsetof(type, field)))
+
+/** Insert a node into a tree at a particular location
+ *
+ * This function should probably not be used directly as it relies on the
+ * caller to ensure that the parent node is correct.  Use rb_tree_insert
+ * instead.
+ *
+ * \param   T           The red-black tree into which to insert the new node
+ *
+ * \param   parent      The node in the tree that will be the parent of the
+ *                      newly inserted node
+ *
+ * \param   node        The node to insert
+ *
+ * \param   insert_left If true, the new node will be the left child of
+ *                      \p parent, otherwise it will be the right child
+ */
+void rb_tree_insert_at(struct rb_tree *T, struct rb_node *parent,
+                       struct rb_node *node, bool insert_left);
+
+/** Insert a node into a tree
+ *
+ * \param   T       The red-black tree into which to insert the new node
+ *
+ * \param   node    The node to insert
+ *
+ * \param   cmp     A comparison function to use to order the nodes.
+ */
+static inline void
+rb_tree_insert(struct rb_tree *T, struct rb_node *node,
+               int (*cmp)(const struct rb_node *, const struct rb_node *))
+{
+    /* This function is declared inline in the hopes that the compiler can
+     * optimize away the comparison function pointer call.
+     */
+    struct rb_node *y = NULL;
+    struct rb_node *x = T->root;
+    bool left = false;
+    while (x != NULL) {
+        y = x;
+        left = cmp(node, x) < 0;
+        if (left)
+            x = x->left;
+        else
+            x = x->right;
+    }
+
+    rb_tree_insert_at(T, y, node, left);
+}
+
+/** Remove a node from a tree
+ *
+ * \param   T       The red-black tree from which to remove the node
+ *
+ * \param   node    The node to remove
+ */
+void rb_tree_remove(struct rb_tree *T, struct rb_node *z);
+
+/** Search the tree for a node
+ *
+ * If a node with a matching key exists, the first matching node found will
+ * be returned.  If no matching node exists, NULL is returned.
+ *
+ * \param   T       The red-black tree to search
+ *
+ * \param   key     The key to search for
+ *
+ * \param   cmp     A comparison function to use to order the nodes
+ */
+static inline struct rb_node *
+rb_tree_search(struct rb_tree *T, const void *key,
+               int (*cmp)(const struct rb_node *, const void *))
+{
+    /* This function is declared inline in the hopes that the compiler can
+     * optimize away the comparison function pointer call.
+     */
+    struct rb_node *x = T->root;
+    while (x != NULL) {
+        int c = cmp(x, key);
+        if (c < 0)
+            x = x->right;
+        else if (c > 0)
+            x = x->left;
+        else
+            return x;
+    }
+
+    return x;
+}
+
+/** Sloppily search the tree for a node
+ *
+ * This function searches the tree for a given node.  If a node with a
+ * matching key exists, that first matching node found will be returned.
+ * If no node with an exactly matching key exists, the node returned will
+ * be either the right-most node comparing less than \p key or the
+ * right-most node comparing greater than \p key.  If the tree is empty,
+ * NULL is returned.
+ *
+ * \param   T       The red-black tree to search
+ *
+ * \param   key     The key to search for
+ *
+ * \param   cmp     A comparison function to use to order the nodes
+ */
+static inline struct rb_node *
+rb_tree_search_sloppy(struct rb_tree *T, const void *key,
+                      int (*cmp)(const struct rb_node *, const void *))
+{
+    /* This function is declared inline in the hopes that the compiler can
+     * optimize away the comparison function pointer call.
+     */
+    struct rb_node *y = NULL;
+    struct rb_node *x = T->root;
+    while (x != NULL) {
+        y = x;
+        int c = cmp(x, key);
+        if (c < 0)
+            x = x->right;
+        else if (c > 0)
+            x = x->left;
+        else
+            return x;
+    }
+
+    return y;
+}
+
+/** Get the first (left-most) node in the tree or NULL */
+struct rb_node *rb_tree_first(struct rb_tree *T);
+
+/** Get the last (right-most) node in the tree or NULL */
+struct rb_node *rb_tree_last(struct rb_tree *T);
+
+/** Get the next node (to the right) in the tree or NULL */
+struct rb_node *rb_node_next(struct rb_node *node);
+
+/** Get the next previous (to the left) in the tree or NULL */
+struct rb_node *rb_node_prev(struct rb_node *node);
+
+/** Get the next node if available or the same node again.
+ *
+ * \param   type    The type of the containing data structure
+ *
+ * \param   node    The variable name for current node in the iteration;
+ *                  this will be declared as a pointer to \p type
+ *
+ * \param   field   The rb_node field in containing data structure
+ */
+#define rb_tree_node_next_if_available(type, node, field) \
+   (&node->field != NULL) ? rb_node_data(type, rb_node_next(&node->field), field) : node
+
+/** Get the previous node if available or the same node again.
+ *
+ * \param   type    The type of the containing data structure
+ *
+ * \param   node    The variable name for current node in the iteration;
+ *                  this will be declared as a pointer to \p type
+ *
+ * \param   field   The rb_node field in containing data structure
+ */
+#define rb_tree_node_prev_if_available(type, node, field) \
+   (&node->field != NULL) ? rb_node_data(type, rb_node_prev(&node->field), field) : node
+
+/** Iterate over the nodes in the tree
+ *
+ * \param   type    The type of the containing data structure
+ *
+ * \param   node    The variable name for current node in the iteration;
+ *                  this will be declared as a pointer to \p type
+ *
+ * \param   T       The red-black tree
+ *
+ * \param   field   The rb_node field in containing data structure
+ */
+#define rb_tree_foreach(type, node, T, field) \
+   for (type *node = rb_node_data(type, rb_tree_first(T), field); \
+        &node->field != NULL; \
+        node = rb_node_data(type, rb_node_next(&node->field), field))
+
+/** Iterate over the nodes in the tree, allowing the current node to be freed
+ *
+ * \param   type    The type of the containing data structure
+ *
+ * \param   node    The variable name for current node in the iteration;
+ *                  this will be declared as a pointer to \p type
+ *
+ * \param   T       The red-black tree
+ *
+ * \param   field   The rb_node field in containing data structure
+ */
+#define rb_tree_foreach_safe(type, node, T, field) \
+   for (type *node = rb_node_data(type, rb_tree_first(T), field), \
+           *__next = rb_tree_node_next_if_available(type, node, field); \
+        &node->field != NULL; \
+        node = __next, __next = rb_tree_node_next_if_available(type, node, field))
+
+/** Iterate over the nodes in the tree in reverse
+ *
+ * \param   type    The type of the containing data structure
+ *
+ * \param   node    The variable name for current node in the iteration;
+ *                  this will be declared as a pointer to \p type
+ *
+ * \param   T       The red-black tree
+ *
+ * \param   field   The rb_node field in containing data structure
+ */
+#define rb_tree_foreach_rev(type, node, T, field) \
+   for (type *node = rb_node_data(type, rb_tree_last(T), field); \
+        &node->field != NULL; \
+        node = rb_node_data(type, rb_node_prev(&node->field), field))
+
+/** Iterate over the nodes in the tree in reverse, allowing the current node to be freed
+ *
+ * \param   type    The type of the containing data structure
+ *
+ * \param   node    The variable name for current node in the iteration;
+ *                  this will be declared as a pointer to \p type
+ *
+ * \param   T       The red-black tree
+ *
+ * \param   field   The rb_node field in containing data structure
+ */
+#define rb_tree_foreach_rev_safe(type, node, T, field) \
+   for (type *node = rb_node_data(type, rb_tree_last(T), field), \
+           *__prev = rb_tree_node_prev_if_available(type, node, field);  \
+        &node->field != NULL; \
+        node = __prev, __prev = rb_tree_node_prev_if_available(type, node, field))
+
+/** Validate a red-black tree
+ *
+ * This function walks the tree and validates that this is a valid red-
+ * black tree.  If anything is wrong, it will assert-fail.
+ */
+void rb_tree_validate(struct rb_tree *T);
+
+#endif /* RB_TREE_H */
diff --git a/lib/mesa/src/util/register_allocate.c b/lib/mesa/src/util/register_allocate.c
index de8978bb9..500540bfd 100644
--- a/lib/mesa/src/util/register_allocate.c
+++ b/lib/mesa/src/util/register_allocate.c
@@ -75,7 +75,6 @@
 #include "ralloc.h"
 #include "main/imports.h"
 #include "main/macros.h"
-#include "main/mtypes.h"
 #include "util/bitset.h"
 #include "register_allocate.h"
 
diff --git a/lib/mesa/src/util/set.h b/lib/mesa/src/util/set.h
index 9acd2c28c..54719e4c8 100644
--- a/lib/mesa/src/util/set.h
+++ b/lib/mesa/src/util/set.h
@@ -58,9 +58,15 @@ _mesa_set_create(void *mem_ctx,
                  uint32_t (*key_hash_function)(const void *key),
                  bool (*key_equals_function)(const void *a,
                                              const void *b));
+struct set *
+_mesa_set_clone(struct set *set, void *dst_mem_ctx);
+
 void
 _mesa_set_destroy(struct set *set,
                   void (*delete_function)(struct set_entry *entry));
+void
+_mesa_set_clear(struct set *set,
+                void (*delete_function)(struct set_entry *entry));
 
 struct set_entry *
 _mesa_set_add(struct set *set, const void *key);
@@ -75,6 +81,8 @@ _mesa_set_search_pre_hashed(const struct set *set, uint32_t hash,
 
 void
 _mesa_set_remove(struct set *set, struct set_entry *entry);
+void
+_mesa_set_remove_key(struct set *set, const void *key);
 
 struct set_entry *
 _mesa_set_next_entry(const struct set *set, struct set_entry *entry);
@@ -88,9 +96,9 @@ _mesa_set_random_entry(struct set *set,
  * insertion (which may rehash the set, making entry a dangling
  * pointer).
  */
-#define set_foreach(set, entry)                          \
-   for (entry = _mesa_set_next_entry(set, NULL);  \
-        entry != NULL;                                   \
+#define set_foreach(set, entry)                                     \
+   for (struct set_entry *entry = _mesa_set_next_entry(set, NULL);  \
+        entry != NULL;                                              \
         entry = _mesa_set_next_entry(set, entry))
 
 #ifdef __cplusplus
diff --git a/lib/mesa/src/util/simple_mtx.h b/lib/mesa/src/util/simple_mtx.h
new file mode 100644
index 000000000..cfb82ba56
--- /dev/null
+++ b/lib/mesa/src/util/simple_mtx.h
@@ -0,0 +1,136 @@
+/*
+ * Copyright © 2015 Intel
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef _SIMPLE_MTX_H
+#define _SIMPLE_MTX_H
+
+#include "util/futex.h"
+
+#include "c11/threads.h"
+
+#if defined(__GNUC__) && defined(HAVE_LINUX_FUTEX_H)
+
+/* mtx_t - Fast, simple mutex
+ *
+ * While modern pthread mutexes are very fast (implemented using futex), they
+ * still incur a call to an external DSO and overhead of the generality and
+ * features of pthread mutexes.  Most mutexes in mesa only needs lock/unlock,
+ * and the idea here is that we can inline the atomic operation and make the
+ * fast case just two intructions.  Mutexes are subtle and finicky to
+ * implement, so we carefully copy the implementation from Ulrich Dreppers
+ * well-written and well-reviewed paper:
+ *
+ *   "Futexes Are Tricky"
+ *   http://www.akkadia.org/drepper/futex.pdf
+ *
+ * We implement "mutex3", which gives us a mutex that has no syscalls on
+ * uncontended lock or unlock.  Further, the uncontended case boils down to a
+ * locked cmpxchg and an untaken branch, the uncontended unlock is just a
+ * locked decr and an untaken branch.  We use __builtin_expect() to indicate
+ * that contention is unlikely so that gcc will put the contention code out of
+ * the main code flow.
+ *
+ * A fast mutex only supports lock/unlock, can't be recursive or used with
+ * condition variables.
+ */
+
+typedef struct {
+   uint32_t val;
+} simple_mtx_t;
+
+#define _SIMPLE_MTX_INITIALIZER_NP { 0 }
+
+static inline void
+simple_mtx_init(simple_mtx_t *mtx, MAYBE_UNUSED int type)
+{
+   assert(type == mtx_plain);
+
+   mtx->val = 0;
+}
+
+static inline void
+simple_mtx_destroy(UNUSED simple_mtx_t *mtx)
+{
+}
+
+static inline void
+simple_mtx_lock(simple_mtx_t *mtx)
+{
+   uint32_t c;
+
+   c = __sync_val_compare_and_swap(&mtx->val, 0, 1);
+   if (__builtin_expect(c != 0, 0)) {
+      if (c != 2)
+         c = __sync_lock_test_and_set(&mtx->val, 2);
+      while (c != 0) {
+         futex_wait(&mtx->val, 2, NULL);
+         c = __sync_lock_test_and_set(&mtx->val, 2);
+      }
+   }
+}
+
+static inline void
+simple_mtx_unlock(simple_mtx_t *mtx)
+{
+   uint32_t c;
+
+   c = __sync_fetch_and_sub(&mtx->val, 1);
+   if (__builtin_expect(c != 1, 0)) {
+      mtx->val = 0;
+      futex_wake(&mtx->val, 1);
+   }
+}
+
+#else
+
+typedef mtx_t simple_mtx_t;
+
+#define _SIMPLE_MTX_INITIALIZER_NP _MTX_INITIALIZER_NP
+
+static inline void
+simple_mtx_init(simple_mtx_t *mtx, int type)
+{
+   mtx_init(mtx, type);
+}
+
+static inline void
+simple_mtx_destroy(simple_mtx_t *mtx)
+{
+   mtx_destroy(mtx);
+}
+
+static inline void
+simple_mtx_lock(simple_mtx_t *mtx)
+{
+   mtx_lock(mtx);
+}
+
+static inline void
+simple_mtx_unlock(simple_mtx_t *mtx)
+{
+   mtx_unlock(mtx);
+}
+
+#endif
+
+#endif
diff --git a/lib/mesa/src/util/tests/fast_idiv_by_const/Makefile.am b/lib/mesa/src/util/tests/fast_idiv_by_const/Makefile.am
new file mode 100644
index 000000000..1ebee09f5
--- /dev/null
+++ b/lib/mesa/src/util/tests/fast_idiv_by_const/Makefile.am
@@ -0,0 +1,43 @@
+# Copyright © 2018 Intel
+#
+#  Permission is hereby granted, free of charge, to any person obtaining a
+#  copy of this software and associated documentation files (the "Software"),
+#  to deal in the Software without restriction, including without limitation
+#  the rights to use, copy, modify, merge, publish, distribute, sublicense,
+#  and/or sell copies of the Software, and to permit persons to whom the
+#  Software is furnished to do so, subject to the following conditions:
+#
+#  The above copyright notice and this permission notice (including the next
+#  paragraph) shall be included in all copies or substantial portions of the
+#  Software.
+#
+#  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+#  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+#  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+#  THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+#  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+#  FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+#  IN THE SOFTWARE.
+
+AM_CPPFLAGS = \
+	-I$(top_srcdir)/src \
+	-I$(top_srcdir)/include \
+	-I$(top_srcdir)/src/gallium/include \
+	-I$(top_srcdir)/src/gtest/include \
+	$(PTHREAD_CFLAGS) \
+	$(DEFINES)
+
+TESTS = fast_idiv_by_const_test
+
+check_PROGRAMS = $(TESTS)
+
+fast_idiv_by_const_test_SOURCES = \
+	fast_idiv_by_const_test.cpp
+
+fast_idiv_by_const_test_LDADD = \
+	$(top_builddir)/src/gtest/libgtest.la \
+	$(top_builddir)/src/util/libmesautil.la \
+	$(PTHREAD_LIBS) \
+	$(DLOPEN_LIBS)
+
+EXTRA_DIST = meson.build
diff --git a/lib/mesa/src/util/tests/fast_idiv_by_const/Makefile.in b/lib/mesa/src/util/tests/fast_idiv_by_const/Makefile.in
new file mode 100644
index 000000000..0242034e8
--- /dev/null
+++ b/lib/mesa/src/util/tests/fast_idiv_by_const/Makefile.in
@@ -0,0 +1,1224 @@
+# Makefile.in generated by automake 1.16.1 from Makefile.am.
+# @configure_input@
+
+# Copyright (C) 1994-2018 Free Software Foundation, Inc.
+
+# This Makefile.in is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
+# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
+# PARTICULAR PURPOSE.
+
+@SET_MAKE@
+
+# Copyright © 2018 Intel
+#
+#  Permission is hereby granted, free of charge, to any person obtaining a
+#  copy of this software and associated documentation files (the "Software"),
+#  to deal in the Software without restriction, including without limitation
+#  the rights to use, copy, modify, merge, publish, distribute, sublicense,
+#  and/or sell copies of the Software, and to permit persons to whom the
+#  Software is furnished to do so, subject to the following conditions:
+#
+#  The above copyright notice and this permission notice (including the next
+#  paragraph) shall be included in all copies or substantial portions of the
+#  Software.
+#
+#  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+#  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+#  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+#  THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+#  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+#  FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+#  IN THE SOFTWARE.
+VPATH = @srcdir@
+am__is_gnu_make = { \
+  if test -z '$(MAKELEVEL)'; then \
+    false; \
+  elif test -n '$(MAKE_HOST)'; then \
+    true; \
+  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
+    true; \
+  else \
+    false; \
+  fi; \
+}
+am__make_running_with_option = \
+  case $${target_option-} in \
+      ?) ;; \
+      *) echo "am__make_running_with_option: internal error: invalid" \
+              "target option '$${target_option-}' specified" >&2; \
+         exit 1;; \
+  esac; \
+  has_opt=no; \
+  sane_makeflags=$$MAKEFLAGS; \
+  if $(am__is_gnu_make); then \
+    sane_makeflags=$$MFLAGS; \
+  else \
+    case $$MAKEFLAGS in \
+      *\\[\ \	]*) \
+        bs=\\; \
+        sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
+          | sed "s/$$bs$$bs[$$bs $$bs	]*//g"`;; \
+    esac; \
+  fi; \
+  skip_next=no; \
+  strip_trailopt () \
+  { \
+    flg=`printf '%s\n' "$$flg" | sed "s/$$1.*$$//"`; \
+  }; \
+  for flg in $$sane_makeflags; do \
+    test $$skip_next = yes && { skip_next=no; continue; }; \
+    case $$flg in \
+      *=*|--*) continue;; \
+        -*I) strip_trailopt 'I'; skip_next=yes;; \
+      -*I?*) strip_trailopt 'I';; \
+        -*O) strip_trailopt 'O'; skip_next=yes;; \
+      -*O?*) strip_trailopt 'O';; \
+        -*l) strip_trailopt 'l'; skip_next=yes;; \
+      -*l?*) strip_trailopt 'l';; \
+      -[dEDm]) skip_next=yes;; \
+      -[JT]) skip_next=yes;; \
+    esac; \
+    case $$flg in \
+      *$$target_option*) has_opt=yes; break;; \
+    esac; \
+  done; \
+  test $$has_opt = yes
+am__make_dryrun = (target_option=n; $(am__make_running_with_option))
+am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
+pkgdatadir = $(datadir)/@PACKAGE@
+pkgincludedir = $(includedir)/@PACKAGE@
+pkglibdir = $(libdir)/@PACKAGE@
+pkglibexecdir = $(libexecdir)/@PACKAGE@
+am__cd = CDPATH="$${ZSH_VERSION+.}$(PATH_SEPARATOR)" && cd
+install_sh_DATA = $(install_sh) -c -m 644
+install_sh_PROGRAM = $(install_sh) -c
+install_sh_SCRIPT = $(install_sh) -c
+INSTALL_HEADER = $(INSTALL_DATA)
+transform = $(program_transform_name)
+NORMAL_INSTALL = :
+PRE_INSTALL = :
+POST_INSTALL = :
+NORMAL_UNINSTALL = :
+PRE_UNINSTALL = :
+POST_UNINSTALL = :
+build_triplet = @build@
+host_triplet = @host@
+target_triplet = @target@
+TESTS = fast_idiv_by_const_test$(EXEEXT)
+check_PROGRAMS = $(am__EXEEXT_1)
+subdir = src/util/tests/fast_idiv_by_const
+ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
+am__aclocal_m4_deps = $(top_srcdir)/m4/ax_check_compile_flag.m4 \
+	$(top_srcdir)/m4/ax_check_gnu_make.m4 \
+	$(top_srcdir)/m4/ax_check_python_mako_module.m4 \
+	$(top_srcdir)/m4/ax_gcc_builtin.m4 \
+	$(top_srcdir)/m4/ax_gcc_func_attribute.m4 \
+	$(top_srcdir)/m4/ax_prog_bison.m4 \
+	$(top_srcdir)/m4/ax_prog_flex.m4 \
+	$(top_srcdir)/m4/ax_pthread.m4 $(top_srcdir)/m4/libtool.m4 \
+	$(top_srcdir)/m4/ltoptions.m4 $(top_srcdir)/m4/ltsugar.m4 \
+	$(top_srcdir)/m4/ltversion.m4 $(top_srcdir)/m4/lt~obsolete.m4 \
+	$(top_srcdir)/VERSION $(top_srcdir)/configure.ac
+am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
+	$(ACLOCAL_M4)
+DIST_COMMON = $(srcdir)/Makefile.am $(am__DIST_COMMON)
+mkinstalldirs = $(install_sh) -d
+CONFIG_CLEAN_FILES =
+CONFIG_CLEAN_VPATH_FILES =
+am__EXEEXT_1 = fast_idiv_by_const_test$(EXEEXT)
+am_fast_idiv_by_const_test_OBJECTS =  \
+	fast_idiv_by_const_test.$(OBJEXT)
+fast_idiv_by_const_test_OBJECTS =  \
+	$(am_fast_idiv_by_const_test_OBJECTS)
+am__DEPENDENCIES_1 =
+fast_idiv_by_const_test_DEPENDENCIES =  \
+	$(top_builddir)/src/gtest/libgtest.la \
+	$(top_builddir)/src/util/libmesautil.la $(am__DEPENDENCIES_1) \
+	$(am__DEPENDENCIES_1)
+AM_V_lt = $(am__v_lt_@AM_V@)
+am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@)
+am__v_lt_0 = --silent
+am__v_lt_1 = 
+AM_V_P = $(am__v_P_@AM_V@)
+am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
+am__v_P_0 = false
+am__v_P_1 = :
+AM_V_GEN = $(am__v_GEN_@AM_V@)
+am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
+am__v_GEN_0 = @echo "  GEN     " $@;
+am__v_GEN_1 = 
+AM_V_at = $(am__v_at_@AM_V@)
+am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
+am__v_at_0 = @
+am__v_at_1 = 
+DEFAULT_INCLUDES = -I.@am__isrc@
+depcomp = $(SHELL) $(top_srcdir)/bin/depcomp
+am__maybe_remake_depfiles = depfiles
+am__depfiles_remade = ./$(DEPDIR)/fast_idiv_by_const_test.Po
+am__mv = mv -f
+CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
+	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
+LTCXXCOMPILE = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
+	$(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) \
+	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
+	$(AM_CXXFLAGS) $(CXXFLAGS)
+AM_V_CXX = $(am__v_CXX_@AM_V@)
+am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
+am__v_CXX_0 = @echo "  CXX     " $@;
+am__v_CXX_1 = 
+CXXLD = $(CXX)
+CXXLINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
+	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(AM_CXXFLAGS) \
+	$(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) -o $@
+AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
+am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
+am__v_CXXLD_0 = @echo "  CXXLD   " $@;
+am__v_CXXLD_1 = 
+SOURCES = $(fast_idiv_by_const_test_SOURCES)
+DIST_SOURCES = $(fast_idiv_by_const_test_SOURCES)
+am__can_run_installinfo = \
+  case $$AM_UPDATE_INFO_DIR in \
+    n|no|NO) false;; \
+    *) (install-info --version) >/dev/null 2>&1;; \
+  esac
+am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
+# Read a list of newline-separated strings from the standard input,
+# and print each of them once, without duplicates.  Input order is
+# *not* preserved.
+am__uniquify_input = $(AWK) '\
+  BEGIN { nonempty = 0; } \
+  { items[$$0] = 1; nonempty = 1; } \
+  END { if (nonempty) { for (i in items) print i; }; } \
+'
+# Make sure the list of sources is unique.  This is necessary because,
+# e.g., the same source file might be shared among _SOURCES variables
+# for different programs/libraries.
+am__define_uniq_tagged_files = \
+  list='$(am__tagged_files)'; \
+  unique=`for i in $$list; do \
+    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+  done | $(am__uniquify_input)`
+ETAGS = etags
+CTAGS = ctags
+am__tty_colors_dummy = \
+  mgn= red= grn= lgn= blu= brg= std=; \
+  am__color_tests=no
+am__tty_colors = { \
+  $(am__tty_colors_dummy); \
+  if test "X$(AM_COLOR_TESTS)" = Xno; then \
+    am__color_tests=no; \
+  elif test "X$(AM_COLOR_TESTS)" = Xalways; then \
+    am__color_tests=yes; \
+  elif test "X$$TERM" != Xdumb && { test -t 1; } 2>/dev/null; then \
+    am__color_tests=yes; \
+  fi; \
+  if test $$am__color_tests = yes; then \
+    red=''; \
+    grn=''; \
+    lgn=''; \
+    blu=''; \
+    mgn=''; \
+    brg=''; \
+    std=''; \
+  fi; \
+}
+am__vpath_adj_setup = srcdirstrip=`echo "$(srcdir)" | sed 's|.|.|g'`;
+am__vpath_adj = case $$p in \
+    $(srcdir)/*) f=`echo "$$p" | sed "s|^$$srcdirstrip/||"`;; \
+    *) f=$$p;; \
+  esac;
+am__strip_dir = f=`echo $$p | sed -e 's|^.*/||'`;
+am__install_max = 40
+am__nobase_strip_setup = \
+  srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*|]/\\\\&/g'`
+am__nobase_strip = \
+  for p in $$list; do echo "$$p"; done | sed -e "s|$$srcdirstrip/||"
+am__nobase_list = $(am__nobase_strip_setup); \
+  for p in $$list; do echo "$$p $$p"; done | \
+  sed "s| $$srcdirstrip/| |;"' / .*\//!s/ .*/ ./; s,\( .*\)/[^/]*$$,\1,' | \
+  $(AWK) 'BEGIN { files["."] = "" } { files[$$2] = files[$$2] " " $$1; \
+    if (++n[$$2] == $(am__install_max)) \
+      { print $$2, files[$$2]; n[$$2] = 0; files[$$2] = "" } } \
+    END { for (dir in files) print dir, files[dir] }'
+am__base_list = \
+  sed '$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;s/\n/ /g' | \
+  sed '$$!N;$$!N;$$!N;$$!N;s/\n/ /g'
+am__uninstall_files_from_dir = { \
+  test -z "$$files" \
+    || { test ! -d "$$dir" && test ! -f "$$dir" && test ! -r "$$dir"; } \
+    || { echo " ( cd '$$dir' && rm -f" $$files ")"; \
+         $(am__cd) "$$dir" && rm -f $$files; }; \
+  }
+am__recheck_rx = ^[ 	]*:recheck:[ 	]*
+am__global_test_result_rx = ^[ 	]*:global-test-result:[ 	]*
+am__copy_in_global_log_rx = ^[ 	]*:copy-in-global-log:[ 	]*
+# A command that, given a newline-separated list of test names on the
+# standard input, print the name of the tests that are to be re-run
+# upon "make recheck".
+am__list_recheck_tests = $(AWK) '{ \
+  recheck = 1; \
+  while ((rc = (getline line < ($$0 ".trs"))) != 0) \
+    { \
+      if (rc < 0) \
+        { \
+          if ((getline line2 < ($$0 ".log")) < 0) \
+	    recheck = 0; \
+          break; \
+        } \
+      else if (line ~ /$(am__recheck_rx)[nN][Oo]/) \
+        { \
+          recheck = 0; \
+          break; \
+        } \
+      else if (line ~ /$(am__recheck_rx)[yY][eE][sS]/) \
+        { \
+          break; \
+        } \
+    }; \
+  if (recheck) \
+    print $$0; \
+  close ($$0 ".trs"); \
+  close ($$0 ".log"); \
+}'
+# A command that, given a newline-separated list of test names on the
+# standard input, create the global log from their .trs and .log files.
+am__create_global_log = $(AWK) ' \
+function fatal(msg) \
+{ \
+  print "fatal: making $@: " msg | "cat >&2"; \
+  exit 1; \
+} \
+function rst_section(header) \
+{ \
+  print header; \
+  len = length(header); \
+  for (i = 1; i <= len; i = i + 1) \
+    printf "="; \
+  printf "\n\n"; \
+} \
+{ \
+  copy_in_global_log = 1; \
+  global_test_result = "RUN"; \
+  while ((rc = (getline line < ($$0 ".trs"))) != 0) \
+    { \
+      if (rc < 0) \
+         fatal("failed to read from " $$0 ".trs"); \
+      if (line ~ /$(am__global_test_result_rx)/) \
+        { \
+          sub("$(am__global_test_result_rx)", "", line); \
+          sub("[ 	]*$$", "", line); \
+          global_test_result = line; \
+        } \
+      else if (line ~ /$(am__copy_in_global_log_rx)[nN][oO]/) \
+        copy_in_global_log = 0; \
+    }; \
+  if (copy_in_global_log) \
+    { \
+      rst_section(global_test_result ": " $$0); \
+      while ((rc = (getline line < ($$0 ".log"))) != 0) \
+      { \
+        if (rc < 0) \
+          fatal("failed to read from " $$0 ".log"); \
+        print line; \
+      }; \
+      printf "\n"; \
+    }; \
+  close ($$0 ".trs"); \
+  close ($$0 ".log"); \
+}'
+# Restructured Text title.
+am__rst_title = { sed 's/.*/   &   /;h;s/./=/g;p;x;s/ *$$//;p;g' && echo; }
+# Solaris 10 'make', and several other traditional 'make' implementations,
+# pass "-e" to $(SHELL), and POSIX 2008 even requires this.  Work around it
+# by disabling -e (using the XSI extension "set +e") if it's set.
+am__sh_e_setup = case $$- in *e*) set +e;; esac
+# Default flags passed to test drivers.
+am__common_driver_flags = \
+  --color-tests "$$am__color_tests" \
+  --enable-hard-errors "$$am__enable_hard_errors" \
+  --expect-failure "$$am__expect_failure"
+# To be inserted before the command running the test.  Creates the
+# directory for the log if needed.  Stores in $dir the directory
+# containing $f, in $tst the test, in $log the log.  Executes the
+# developer- defined test setup AM_TESTS_ENVIRONMENT (if any), and
+# passes TESTS_ENVIRONMENT.  Set up options for the wrapper that
+# will run the test scripts (or their associated LOG_COMPILER, if
+# thy have one).
+am__check_pre = \
+$(am__sh_e_setup);					\
+$(am__vpath_adj_setup) $(am__vpath_adj)			\
+$(am__tty_colors);					\
+srcdir=$(srcdir); export srcdir;			\
+case "$@" in						\
+  */*) am__odir=`echo "./$@" | sed 's|/[^/]*$$||'`;;	\
+    *) am__odir=.;; 					\
+esac;							\
+test "x$$am__odir" = x"." || test -d "$$am__odir" 	\
+  || $(MKDIR_P) "$$am__odir" || exit $$?;		\
+if test -f "./$$f"; then dir=./;			\
+elif test -f "$$f"; then dir=;				\
+else dir="$(srcdir)/"; fi;				\
+tst=$$dir$$f; log='$@'; 				\
+if test -n '$(DISABLE_HARD_ERRORS)'; then		\
+  am__enable_hard_errors=no; 				\
+else							\
+  am__enable_hard_errors=yes; 				\
+fi; 							\
+case " $(XFAIL_TESTS) " in				\
+  *[\ \	]$$f[\ \	]* | *[\ \	]$$dir$$f[\ \	]*) \
+    am__expect_failure=yes;;				\
+  *)							\
+    am__expect_failure=no;;				\
+esac; 							\
+$(AM_TESTS_ENVIRONMENT) $(TESTS_ENVIRONMENT)
+# A shell command to get the names of the tests scripts with any registered
+# extension removed (i.e., equivalently, the names of the test logs, with
+# the '.log' extension removed).  The result is saved in the shell variable
+# '$bases'.  This honors runtime overriding of TESTS and TEST_LOGS.  Sadly,
+# we cannot use something simpler, involving e.g., "$(TEST_LOGS:.log=)",
+# since that might cause problem with VPATH rewrites for suffix-less tests.
+# See also 'test-harness-vpath-rewrite.sh' and 'test-trs-basic.sh'.
+am__set_TESTS_bases = \
+  bases='$(TEST_LOGS)'; \
+  bases=`for i in $$bases; do echo $$i; done | sed 's/\.log$$//'`; \
+  bases=`echo $$bases`
+RECHECK_LOGS = $(TEST_LOGS)
+AM_RECURSIVE_TARGETS = check recheck
+TEST_SUITE_LOG = test-suite.log
+TEST_EXTENSIONS = @EXEEXT@ .test
+LOG_DRIVER = $(SHELL) $(top_srcdir)/bin/test-driver
+LOG_COMPILE = $(LOG_COMPILER) $(AM_LOG_FLAGS) $(LOG_FLAGS)
+am__set_b = \
+  case '$@' in \
+    */*) \
+      case '$*' in \
+        */*) b='$*';; \
+          *) b=`echo '$@' | sed 's/\.log$$//'`; \
+       esac;; \
+    *) \
+      b='$*';; \
+  esac
+am__test_logs1 = $(TESTS:=.log)
+am__test_logs2 = $(am__test_logs1:@EXEEXT@.log=.log)
+TEST_LOGS = $(am__test_logs2:.test.log=.log)
+TEST_LOG_DRIVER = $(SHELL) $(top_srcdir)/bin/test-driver
+TEST_LOG_COMPILE = $(TEST_LOG_COMPILER) $(AM_TEST_LOG_FLAGS) \
+	$(TEST_LOG_FLAGS)
+am__DIST_COMMON = $(srcdir)/Makefile.in $(top_srcdir)/bin/depcomp \
+	$(top_srcdir)/bin/test-driver
+DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
+ACLOCAL = @ACLOCAL@
+AMDGPU_CFLAGS = @AMDGPU_CFLAGS@
+AMDGPU_LIBS = @AMDGPU_LIBS@
+AMTAR = @AMTAR@
+AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
+ANDROID_CFLAGS = @ANDROID_CFLAGS@
+ANDROID_LIBS = @ANDROID_LIBS@
+AR = @AR@
+AUTOCONF = @AUTOCONF@
+AUTOHEADER = @AUTOHEADER@
+AUTOMAKE = @AUTOMAKE@
+AWK = @AWK@
+BACKTRACE_CFLAGS = @BACKTRACE_CFLAGS@
+BACKTRACE_LIBS = @BACKTRACE_LIBS@
+BSYMBOLIC = @BSYMBOLIC@
+CC = @CC@
+CCAS = @CCAS@
+CCASDEPMODE = @CCASDEPMODE@
+CCASFLAGS = @CCASFLAGS@
+CCDEPMODE = @CCDEPMODE@
+CFLAGS = @CFLAGS@
+CLANG_RESOURCE_DIR = @CLANG_RESOURCE_DIR@
+CLOCK_LIB = @CLOCK_LIB@
+CLOVER_STD_OVERRIDE = @CLOVER_STD_OVERRIDE@
+CPP = @CPP@
+CPPFLAGS = @CPPFLAGS@
+CXX = @CXX@
+CXX11_CXXFLAGS = @CXX11_CXXFLAGS@
+CXXCPP = @CXXCPP@
+CXXDEPMODE = @CXXDEPMODE@
+CXXFLAGS = @CXXFLAGS@
+CYGPATH_W = @CYGPATH_W@
+D3D_DRIVER_INSTALL_DIR = @D3D_DRIVER_INSTALL_DIR@
+DEFINES = @DEFINES@
+DEFS = @DEFS@
+DEPDIR = @DEPDIR@
+DLLTOOL = @DLLTOOL@
+DLOPEN_LIBS = @DLOPEN_LIBS@
+DRI2PROTO_CFLAGS = @DRI2PROTO_CFLAGS@
+DRI2PROTO_LIBS = @DRI2PROTO_LIBS@
+DRIGL_CFLAGS = @DRIGL_CFLAGS@
+DRIGL_LIBS = @DRIGL_LIBS@
+DRI_DRIVER_INSTALL_DIR = @DRI_DRIVER_INSTALL_DIR@
+DRI_DRIVER_SEARCH_DIR = @DRI_DRIVER_SEARCH_DIR@
+DRI_LIB_DEPS = @DRI_LIB_DEPS@
+DRI_PC_REQ_PRIV = @DRI_PC_REQ_PRIV@
+DSYMUTIL = @DSYMUTIL@
+DUMPBIN = @DUMPBIN@
+ECHO_C = @ECHO_C@
+ECHO_N = @ECHO_N@
+ECHO_T = @ECHO_T@
+EGL_CFLAGS = @EGL_CFLAGS@
+EGL_LIB_DEPS = @EGL_LIB_DEPS@
+EGL_NATIVE_PLATFORM = @EGL_NATIVE_PLATFORM@
+EGREP = @EGREP@
+ETNAVIV_CFLAGS = @ETNAVIV_CFLAGS@
+ETNAVIV_LIBS = @ETNAVIV_LIBS@
+EXEEXT = @EXEEXT@
+EXPAT_CFLAGS = @EXPAT_CFLAGS@
+EXPAT_LIBS = @EXPAT_LIBS@
+FGREP = @FGREP@
+GALLIUM_PIPE_LOADER_DEFINES = @GALLIUM_PIPE_LOADER_DEFINES@
+GBM_PC_LIB_PRIV = @GBM_PC_LIB_PRIV@
+GBM_PC_REQ_PRIV = @GBM_PC_REQ_PRIV@
+GC_SECTIONS = @GC_SECTIONS@
+GLESv1_CM_LIB_DEPS = @GLESv1_CM_LIB_DEPS@
+GLESv1_CM_PC_LIB_PRIV = @GLESv1_CM_PC_LIB_PRIV@
+GLESv2_LIB_DEPS = @GLESv2_LIB_DEPS@
+GLESv2_PC_LIB_PRIV = @GLESv2_PC_LIB_PRIV@
+GLPROTO_CFLAGS = @GLPROTO_CFLAGS@
+GLPROTO_LIBS = @GLPROTO_LIBS@
+GLVND_CFLAGS = @GLVND_CFLAGS@
+GLVND_LIBS = @GLVND_LIBS@
+GLX_TLS = @GLX_TLS@
+GL_LIB = @GL_LIB@
+GL_LIB_DEPS = @GL_LIB_DEPS@
+GL_PC_CFLAGS = @GL_PC_CFLAGS@
+GL_PC_LIB_PRIV = @GL_PC_LIB_PRIV@
+GL_PC_REQ_PRIV = @GL_PC_REQ_PRIV@
+GL_PKGCONF_LIB = @GL_PKGCONF_LIB@
+GREP = @GREP@
+I915_CFLAGS = @I915_CFLAGS@
+I915_LIBS = @I915_LIBS@
+INDENT = @INDENT@
+INDENT_FLAGS = @INDENT_FLAGS@
+INSTALL = @INSTALL@
+INSTALL_DATA = @INSTALL_DATA@
+INSTALL_PROGRAM = @INSTALL_PROGRAM@
+INSTALL_SCRIPT = @INSTALL_SCRIPT@
+INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@
+LD = @LD@
+LDFLAGS = @LDFLAGS@
+LD_BUILD_ID = @LD_BUILD_ID@
+LD_NO_UNDEFINED = @LD_NO_UNDEFINED@
+LEX = @LEX@
+LEXLIB = @LEXLIB@
+LEX_OUTPUT_ROOT = @LEX_OUTPUT_ROOT@
+LIBATOMIC_LIBS = @LIBATOMIC_LIBS@
+LIBCLC_INCLUDEDIR = @LIBCLC_INCLUDEDIR@
+LIBCLC_LIBEXECDIR = @LIBCLC_LIBEXECDIR@
+LIBDRM_CFLAGS = @LIBDRM_CFLAGS@
+LIBDRM_LIBS = @LIBDRM_LIBS@
+LIBELF_CFLAGS = @LIBELF_CFLAGS@
+LIBELF_LIBS = @LIBELF_LIBS@
+LIBGLVND_DATADIR = @LIBGLVND_DATADIR@
+LIBOBJS = @LIBOBJS@
+LIBS = @LIBS@
+LIBSENSORS_LIBS = @LIBSENSORS_LIBS@
+LIBTOOL = @LIBTOOL@
+LIBUNWIND_CFLAGS = @LIBUNWIND_CFLAGS@
+LIBUNWIND_LIBS = @LIBUNWIND_LIBS@
+LIB_DIR = @LIB_DIR@
+LIB_EXT = @LIB_EXT@
+LIPO = @LIPO@
+LLVM_CFLAGS = @LLVM_CFLAGS@
+LLVM_CONFIG = @LLVM_CONFIG@
+LLVM_CXXFLAGS = @LLVM_CXXFLAGS@
+LLVM_INCLUDEDIR = @LLVM_INCLUDEDIR@
+LLVM_LDFLAGS = @LLVM_LDFLAGS@
+LLVM_LIBS = @LLVM_LIBS@
+LN_S = @LN_S@
+LTLIBOBJS = @LTLIBOBJS@
+LT_SYS_LIBRARY_PATH = @LT_SYS_LIBRARY_PATH@
+MAKEINFO = @MAKEINFO@
+MANIFEST_TOOL = @MANIFEST_TOOL@
+MKDIR_P = @MKDIR_P@
+MSVC2013_COMPAT_CFLAGS = @MSVC2013_COMPAT_CFLAGS@
+MSVC2013_COMPAT_CXXFLAGS = @MSVC2013_COMPAT_CXXFLAGS@
+NINE_MAJOR = @NINE_MAJOR@
+NINE_MINOR = @NINE_MINOR@
+NINE_PATCH = @NINE_PATCH@
+NINE_VERSION = @NINE_VERSION@
+NM = @NM@
+NMEDIT = @NMEDIT@
+NOUVEAU_CFLAGS = @NOUVEAU_CFLAGS@
+NOUVEAU_LIBS = @NOUVEAU_LIBS@
+NVVIEUX_CFLAGS = @NVVIEUX_CFLAGS@
+NVVIEUX_LIBS = @NVVIEUX_LIBS@
+OBJDUMP = @OBJDUMP@
+OBJEXT = @OBJEXT@
+OMX_BELLAGIO_CFLAGS = @OMX_BELLAGIO_CFLAGS@
+OMX_BELLAGIO_LIBS = @OMX_BELLAGIO_LIBS@
+OMX_BELLAGIO_LIB_INSTALL_DIR = @OMX_BELLAGIO_LIB_INSTALL_DIR@
+OMX_TIZONIA_CFLAGS = @OMX_TIZONIA_CFLAGS@
+OMX_TIZONIA_LIBS = @OMX_TIZONIA_LIBS@
+OMX_TIZONIA_LIB_INSTALL_DIR = @OMX_TIZONIA_LIB_INSTALL_DIR@
+OPENCL_LIBNAME = @OPENCL_LIBNAME@
+OPENCL_VERSION = @OPENCL_VERSION@
+OSMESA_LIB = @OSMESA_LIB@
+OSMESA_LIB_DEPS = @OSMESA_LIB_DEPS@
+OSMESA_PC_LIB_PRIV = @OSMESA_PC_LIB_PRIV@
+OSMESA_PC_REQ = @OSMESA_PC_REQ@
+OSMESA_VERSION = @OSMESA_VERSION@
+OTOOL = @OTOOL@
+OTOOL64 = @OTOOL64@
+PACKAGE = @PACKAGE@
+PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
+PACKAGE_NAME = @PACKAGE_NAME@
+PACKAGE_STRING = @PACKAGE_STRING@
+PACKAGE_TARNAME = @PACKAGE_TARNAME@
+PACKAGE_URL = @PACKAGE_URL@
+PACKAGE_VERSION = @PACKAGE_VERSION@
+PATH_SEPARATOR = @PATH_SEPARATOR@
+PKG_CONFIG = @PKG_CONFIG@
+PKG_CONFIG_LIBDIR = @PKG_CONFIG_LIBDIR@
+PKG_CONFIG_PATH = @PKG_CONFIG_PATH@
+POSIX_SHELL = @POSIX_SHELL@
+PTHREADSTUBS_CFLAGS = @PTHREADSTUBS_CFLAGS@
+PTHREADSTUBS_LIBS = @PTHREADSTUBS_LIBS@
+PTHREAD_CC = @PTHREAD_CC@
+PTHREAD_CFLAGS = @PTHREAD_CFLAGS@
+PTHREAD_LIBS = @PTHREAD_LIBS@
+PWR8_CFLAGS = @PWR8_CFLAGS@
+PYTHON = @PYTHON@
+PYTHON_EXEC_PREFIX = @PYTHON_EXEC_PREFIX@
+PYTHON_PLATFORM = @PYTHON_PLATFORM@
+PYTHON_PREFIX = @PYTHON_PREFIX@
+PYTHON_VERSION = @PYTHON_VERSION@
+RADEON_CFLAGS = @RADEON_CFLAGS@
+RADEON_LIBS = @RADEON_LIBS@
+RANLIB = @RANLIB@
+RM = @RM@
+SCANNER_ARG = @SCANNER_ARG@
+SED = @SED@
+SELINUX_CFLAGS = @SELINUX_CFLAGS@
+SELINUX_LIBS = @SELINUX_LIBS@
+SET_MAKE = @SET_MAKE@
+SHELL = @SHELL@
+SIMPENROSE_CFLAGS = @SIMPENROSE_CFLAGS@
+SIMPENROSE_LIBS = @SIMPENROSE_LIBS@
+SSE41_CFLAGS = @SSE41_CFLAGS@
+STRIP = @STRIP@
+SWR_AVX2_CXXFLAGS = @SWR_AVX2_CXXFLAGS@
+SWR_AVX_CXXFLAGS = @SWR_AVX_CXXFLAGS@
+SWR_KNL_CXXFLAGS = @SWR_KNL_CXXFLAGS@
+SWR_SKX_CXXFLAGS = @SWR_SKX_CXXFLAGS@
+V3D_SIMULATOR_CFLAGS = @V3D_SIMULATOR_CFLAGS@
+V3D_SIMULATOR_LIBS = @V3D_SIMULATOR_LIBS@
+VALGRIND_CFLAGS = @VALGRIND_CFLAGS@
+VALGRIND_LIBS = @VALGRIND_LIBS@
+VA_CFLAGS = @VA_CFLAGS@
+VA_LIBS = @VA_LIBS@
+VA_LIB_INSTALL_DIR = @VA_LIB_INSTALL_DIR@
+VA_MAJOR = @VA_MAJOR@
+VA_MINOR = @VA_MINOR@
+VC4_CFLAGS = @VC4_CFLAGS@
+VC4_LIBS = @VC4_LIBS@
+VDPAU_CFLAGS = @VDPAU_CFLAGS@
+VDPAU_LIBS = @VDPAU_LIBS@
+VDPAU_LIB_INSTALL_DIR = @VDPAU_LIB_INSTALL_DIR@
+VDPAU_MAJOR = @VDPAU_MAJOR@
+VDPAU_MINOR = @VDPAU_MINOR@
+VERSION = @VERSION@
+VISIBILITY_CFLAGS = @VISIBILITY_CFLAGS@
+VISIBILITY_CXXFLAGS = @VISIBILITY_CXXFLAGS@
+VL_CFLAGS = @VL_CFLAGS@
+VL_LIBS = @VL_LIBS@
+VULKAN_ICD_INSTALL_DIR = @VULKAN_ICD_INSTALL_DIR@
+WAYLAND_CLIENT_CFLAGS = @WAYLAND_CLIENT_CFLAGS@
+WAYLAND_CLIENT_LIBS = @WAYLAND_CLIENT_LIBS@
+WAYLAND_EGL_CFLAGS = @WAYLAND_EGL_CFLAGS@
+WAYLAND_EGL_LIBS = @WAYLAND_EGL_LIBS@
+WAYLAND_PROTOCOLS_CFLAGS = @WAYLAND_PROTOCOLS_CFLAGS@
+WAYLAND_PROTOCOLS_DATADIR = @WAYLAND_PROTOCOLS_DATADIR@
+WAYLAND_PROTOCOLS_LIBS = @WAYLAND_PROTOCOLS_LIBS@
+WAYLAND_SCANNER = @WAYLAND_SCANNER@
+WAYLAND_SCANNER_CFLAGS = @WAYLAND_SCANNER_CFLAGS@
+WAYLAND_SCANNER_LIBS = @WAYLAND_SCANNER_LIBS@
+WAYLAND_SERVER_CFLAGS = @WAYLAND_SERVER_CFLAGS@
+WAYLAND_SERVER_LIBS = @WAYLAND_SERVER_LIBS@
+WNO_OVERRIDE_INIT = @WNO_OVERRIDE_INIT@
+X11_INCLUDES = @X11_INCLUDES@
+XA_MAJOR = @XA_MAJOR@
+XA_MINOR = @XA_MINOR@
+XA_PATCH = @XA_PATCH@
+XA_VERSION = @XA_VERSION@
+XCB_DRI2_CFLAGS = @XCB_DRI2_CFLAGS@
+XCB_DRI2_LIBS = @XCB_DRI2_LIBS@
+XCB_DRI3_CFLAGS = @XCB_DRI3_CFLAGS@
+XCB_DRI3_LIBS = @XCB_DRI3_LIBS@
+XCB_DRI3_MODIFIERS_CFLAGS = @XCB_DRI3_MODIFIERS_CFLAGS@
+XCB_DRI3_MODIFIERS_LIBS = @XCB_DRI3_MODIFIERS_LIBS@
+XCB_RANDR_CFLAGS = @XCB_RANDR_CFLAGS@
+XCB_RANDR_LIBS = @XCB_RANDR_LIBS@
+XLIBGL_CFLAGS = @XLIBGL_CFLAGS@
+XLIBGL_LIBS = @XLIBGL_LIBS@
+XLIB_RANDR_CFLAGS = @XLIB_RANDR_CFLAGS@
+XLIB_RANDR_LIBS = @XLIB_RANDR_LIBS@
+XVMC_CFLAGS = @XVMC_CFLAGS@
+XVMC_LIBS = @XVMC_LIBS@
+XVMC_LIB_INSTALL_DIR = @XVMC_LIB_INSTALL_DIR@
+XVMC_MAJOR = @XVMC_MAJOR@
+XVMC_MINOR = @XVMC_MINOR@
+YACC = @YACC@
+YFLAGS = @YFLAGS@
+ZLIB_CFLAGS = @ZLIB_CFLAGS@
+ZLIB_LIBS = @ZLIB_LIBS@
+abs_builddir = @abs_builddir@
+abs_srcdir = @abs_srcdir@
+abs_top_builddir = @abs_top_builddir@
+abs_top_srcdir = @abs_top_srcdir@
+ac_ct_AR = @ac_ct_AR@
+ac_ct_CC = @ac_ct_CC@
+ac_ct_CXX = @ac_ct_CXX@
+ac_ct_DUMPBIN = @ac_ct_DUMPBIN@
+acv_mako_found = @acv_mako_found@
+am__include = @am__include@
+am__leading_dot = @am__leading_dot@
+am__quote = @am__quote@
+am__tar = @am__tar@
+am__untar = @am__untar@
+ax_pthread_config = @ax_pthread_config@
+bindir = @bindir@
+build = @build@
+build_alias = @build_alias@
+build_cpu = @build_cpu@
+build_os = @build_os@
+build_vendor = @build_vendor@
+builddir = @builddir@
+datadir = @datadir@
+datarootdir = @datarootdir@
+docdir = @docdir@
+dvidir = @dvidir@
+exec_prefix = @exec_prefix@
+host = @host@
+host_alias = @host_alias@
+host_cpu = @host_cpu@
+host_os = @host_os@
+host_vendor = @host_vendor@
+htmldir = @htmldir@
+ifGNUmake = @ifGNUmake@
+includedir = @includedir@
+infodir = @infodir@
+install_sh = @install_sh@
+libdir = @libdir@
+libexecdir = @libexecdir@
+localedir = @localedir@
+localstatedir = @localstatedir@
+mandir = @mandir@
+mkdir_p = @mkdir_p@
+oldincludedir = @oldincludedir@
+pdfdir = @pdfdir@
+pkgpyexecdir = @pkgpyexecdir@
+pkgpythondir = @pkgpythondir@
+prefix = @prefix@
+program_transform_name = @program_transform_name@
+psdir = @psdir@
+pyexecdir = @pyexecdir@
+pythondir = @pythondir@
+sbindir = @sbindir@
+sharedstatedir = @sharedstatedir@
+srcdir = @srcdir@
+sysconfdir = @sysconfdir@
+target = @target@
+target_alias = @target_alias@
+target_cpu = @target_cpu@
+target_os = @target_os@
+target_vendor = @target_vendor@
+top_build_prefix = @top_build_prefix@
+top_builddir = @top_builddir@
+top_srcdir = @top_srcdir@
+AM_CPPFLAGS = \
+	-I$(top_srcdir)/src \
+	-I$(top_srcdir)/include \
+	-I$(top_srcdir)/src/gallium/include \
+	-I$(top_srcdir)/src/gtest/include \
+	$(PTHREAD_CFLAGS) \
+	$(DEFINES)
+
+fast_idiv_by_const_test_SOURCES = \
+	fast_idiv_by_const_test.cpp
+
+fast_idiv_by_const_test_LDADD = \
+	$(top_builddir)/src/gtest/libgtest.la \
+	$(top_builddir)/src/util/libmesautil.la \
+	$(PTHREAD_LIBS) \
+	$(DLOPEN_LIBS)
+
+EXTRA_DIST = meson.build
+all: all-am
+
+.SUFFIXES:
+.SUFFIXES: .cpp .lo .log .o .obj .test .test$(EXEEXT) .trs
+$(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
+	@for dep in $?; do \
+	  case '$(am__configure_deps)' in \
+	    *$$dep*) \
+	      ( cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh ) \
+	        && { if test -f $@; then exit 0; else break; fi; }; \
+	      exit 1;; \
+	  esac; \
+	done; \
+	echo ' cd $(top_srcdir) && $(AUTOMAKE) --foreign src/util/tests/fast_idiv_by_const/Makefile'; \
+	$(am__cd) $(top_srcdir) && \
+	  $(AUTOMAKE) --foreign src/util/tests/fast_idiv_by_const/Makefile
+Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
+	@case '$?' in \
+	  *config.status*) \
+	    cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \
+	  *) \
+	    echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__maybe_remake_depfiles)'; \
+	    cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__maybe_remake_depfiles);; \
+	esac;
+
+$(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+
+$(top_srcdir)/configure:  $(am__configure_deps)
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+$(ACLOCAL_M4):  $(am__aclocal_m4_deps)
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+$(am__aclocal_m4_deps):
+
+clean-checkPROGRAMS:
+	@list='$(check_PROGRAMS)'; test -n "$$list" || exit 0; \
+	echo " rm -f" $$list; \
+	rm -f $$list || exit $$?; \
+	test -n "$(EXEEXT)" || exit 0; \
+	list=`for p in $$list; do echo "$$p"; done | sed 's/$(EXEEXT)$$//'`; \
+	echo " rm -f" $$list; \
+	rm -f $$list
+
+fast_idiv_by_const_test$(EXEEXT): $(fast_idiv_by_const_test_OBJECTS) $(fast_idiv_by_const_test_DEPENDENCIES) $(EXTRA_fast_idiv_by_const_test_DEPENDENCIES) 
+	@rm -f fast_idiv_by_const_test$(EXEEXT)
+	$(AM_V_CXXLD)$(CXXLINK) $(fast_idiv_by_const_test_OBJECTS) $(fast_idiv_by_const_test_LDADD) $(LIBS)
+
+mostlyclean-compile:
+	-rm -f *.$(OBJEXT)
+
+distclean-compile:
+	-rm -f *.tab.c
+
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/fast_idiv_by_const_test.Po@am__quote@ # am--include-marker
+
+$(am__depfiles_remade):
+	@$(MKDIR_P) $(@D)
+	@echo '# dummy' >$@-t && $(am__mv) $@-t $@
+
+am--depfiles: $(am__depfiles_remade)
+
+.cpp.o:
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)depbase=`echo $@ | sed 's|[^/]*$$|$(DEPDIR)/&|;s|\.o$$||'`;\
+@am__fastdepCXX_TRUE@	$(CXXCOMPILE) -MT $@ -MD -MP -MF $$depbase.Tpo -c -o $@ $< &&\
+@am__fastdepCXX_TRUE@	$(am__mv) $$depbase.Tpo $$depbase.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
+
+.cpp.obj:
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)depbase=`echo $@ | sed 's|[^/]*$$|$(DEPDIR)/&|;s|\.obj$$||'`;\
+@am__fastdepCXX_TRUE@	$(CXXCOMPILE) -MT $@ -MD -MP -MF $$depbase.Tpo -c -o $@ `$(CYGPATH_W) '$<'` &&\
+@am__fastdepCXX_TRUE@	$(am__mv) $$depbase.Tpo $$depbase.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
+
+.cpp.lo:
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)depbase=`echo $@ | sed 's|[^/]*$$|$(DEPDIR)/&|;s|\.lo$$||'`;\
+@am__fastdepCXX_TRUE@	$(LTCXXCOMPILE) -MT $@ -MD -MP -MF $$depbase.Tpo -c -o $@ $< &&\
+@am__fastdepCXX_TRUE@	$(am__mv) $$depbase.Tpo $$depbase.Plo
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(LTCXXCOMPILE) -c -o $@ $<
+
+mostlyclean-libtool:
+	-rm -f *.lo
+
+clean-libtool:
+	-rm -rf .libs _libs
+
+ID: $(am__tagged_files)
+	$(am__define_uniq_tagged_files); mkid -fID $$unique
+tags: tags-am
+TAGS: tags
+
+tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+	set x; \
+	here=`pwd`; \
+	$(am__define_uniq_tagged_files); \
+	shift; \
+	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
+	  test -n "$$unique" || unique=$$empty_fix; \
+	  if test $$# -gt 0; then \
+	    $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
+	      "$$@" $$unique; \
+	  else \
+	    $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
+	      $$unique; \
+	  fi; \
+	fi
+ctags: ctags-am
+
+CTAGS: ctags
+ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+	$(am__define_uniq_tagged_files); \
+	test -z "$(CTAGS_ARGS)$$unique" \
+	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
+	     $$unique
+
+GTAGS:
+	here=`$(am__cd) $(top_builddir) && pwd` \
+	  && $(am__cd) $(top_srcdir) \
+	  && gtags -i $(GTAGS_ARGS) "$$here"
+cscopelist: cscopelist-am
+
+cscopelist-am: $(am__tagged_files)
+	list='$(am__tagged_files)'; \
+	case "$(srcdir)" in \
+	  [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
+	  *) sdir=$(subdir)/$(srcdir) ;; \
+	esac; \
+	for i in $$list; do \
+	  if test -f "$$i"; then \
+	    echo "$(subdir)/$$i"; \
+	  else \
+	    echo "$$sdir/$$i"; \
+	  fi; \
+	done >> $(top_builddir)/cscope.files
+
+distclean-tags:
+	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
+
+# Recover from deleted '.trs' file; this should ensure that
+# "rm -f foo.log; make foo.trs" re-run 'foo.test', and re-create
+# both 'foo.log' and 'foo.trs'.  Break the recipe in two subshells
+# to avoid problems with "make -n".
+.log.trs:
+	rm -f $< $@
+	$(MAKE) $(AM_MAKEFLAGS) $<
+
+# Leading 'am--fnord' is there to ensure the list of targets does not
+# expand to empty, as could happen e.g. with make check TESTS=''.
+am--fnord $(TEST_LOGS) $(TEST_LOGS:.log=.trs): $(am__force_recheck)
+am--force-recheck:
+	@:
+
+$(TEST_SUITE_LOG): $(TEST_LOGS)
+	@$(am__set_TESTS_bases); \
+	am__f_ok () { test -f "$$1" && test -r "$$1"; }; \
+	redo_bases=`for i in $$bases; do \
+	              am__f_ok $$i.trs && am__f_ok $$i.log || echo $$i; \
+	            done`; \
+	if test -n "$$redo_bases"; then \
+	  redo_logs=`for i in $$redo_bases; do echo $$i.log; done`; \
+	  redo_results=`for i in $$redo_bases; do echo $$i.trs; done`; \
+	  if $(am__make_dryrun); then :; else \
+	    rm -f $$redo_logs && rm -f $$redo_results || exit 1; \
+	  fi; \
+	fi; \
+	if test -n "$$am__remaking_logs"; then \
+	  echo "fatal: making $(TEST_SUITE_LOG): possible infinite" \
+	       "recursion detected" >&2; \
+	elif test -n "$$redo_logs"; then \
+	  am__remaking_logs=yes $(MAKE) $(AM_MAKEFLAGS) $$redo_logs; \
+	fi; \
+	if $(am__make_dryrun); then :; else \
+	  st=0;  \
+	  errmsg="fatal: making $(TEST_SUITE_LOG): failed to create"; \
+	  for i in $$redo_bases; do \
+	    test -f $$i.trs && test -r $$i.trs \
+	      || { echo "$$errmsg $$i.trs" >&2; st=1; }; \
+	    test -f $$i.log && test -r $$i.log \
+	      || { echo "$$errmsg $$i.log" >&2; st=1; }; \
+	  done; \
+	  test $$st -eq 0 || exit 1; \
+	fi
+	@$(am__sh_e_setup); $(am__tty_colors); $(am__set_TESTS_bases); \
+	ws='[ 	]'; \
+	results=`for b in $$bases; do echo $$b.trs; done`; \
+	test -n "$$results" || results=/dev/null; \
+	all=`  grep "^$$ws*:test-result:"           $$results | wc -l`; \
+	pass=` grep "^$$ws*:test-result:$$ws*PASS"  $$results | wc -l`; \
+	fail=` grep "^$$ws*:test-result:$$ws*FAIL"  $$results | wc -l`; \
+	skip=` grep "^$$ws*:test-result:$$ws*SKIP"  $$results | wc -l`; \
+	xfail=`grep "^$$ws*:test-result:$$ws*XFAIL" $$results | wc -l`; \
+	xpass=`grep "^$$ws*:test-result:$$ws*XPASS" $$results | wc -l`; \
+	error=`grep "^$$ws*:test-result:$$ws*ERROR" $$results | wc -l`; \
+	if test `expr $$fail + $$xpass + $$error` -eq 0; then \
+	  success=true; \
+	else \
+	  success=false; \
+	fi; \
+	br='==================='; br=$$br$$br$$br$$br; \
+	result_count () \
+	{ \
+	    if test x"$$1" = x"--maybe-color"; then \
+	      maybe_colorize=yes; \
+	    elif test x"$$1" = x"--no-color"; then \
+	      maybe_colorize=no; \
+	    else \
+	      echo "$@: invalid 'result_count' usage" >&2; exit 4; \
+	    fi; \
+	    shift; \
+	    desc=$$1 count=$$2; \
+	    if test $$maybe_colorize = yes && test $$count -gt 0; then \
+	      color_start=$$3 color_end=$$std; \
+	    else \
+	      color_start= color_end=; \
+	    fi; \
+	    echo "$${color_start}# $$desc $$count$${color_end}"; \
+	}; \
+	create_testsuite_report () \
+	{ \
+	  result_count $$1 "TOTAL:" $$all   "$$brg"; \
+	  result_count $$1 "PASS: " $$pass  "$$grn"; \
+	  result_count $$1 "SKIP: " $$skip  "$$blu"; \
+	  result_count $$1 "XFAIL:" $$xfail "$$lgn"; \
+	  result_count $$1 "FAIL: " $$fail  "$$red"; \
+	  result_count $$1 "XPASS:" $$xpass "$$red"; \
+	  result_count $$1 "ERROR:" $$error "$$mgn"; \
+	}; \
+	{								\
+	  echo "$(PACKAGE_STRING): $(subdir)/$(TEST_SUITE_LOG)" |	\
+	    $(am__rst_title);						\
+	  create_testsuite_report --no-color;				\
+	  echo;								\
+	  echo ".. contents:: :depth: 2";				\
+	  echo;								\
+	  for b in $$bases; do echo $$b; done				\
+	    | $(am__create_global_log);					\
+	} >$(TEST_SUITE_LOG).tmp || exit 1;				\
+	mv $(TEST_SUITE_LOG).tmp $(TEST_SUITE_LOG);			\
+	if $$success; then						\
+	  col="$$grn";							\
+	 else								\
+	  col="$$red";							\
+	  test x"$$VERBOSE" = x || cat $(TEST_SUITE_LOG);		\
+	fi;								\
+	echo "$${col}$$br$${std}"; 					\
+	echo "$${col}Testsuite summary for $(PACKAGE_STRING)$${std}";	\
+	echo "$${col}$$br$${std}"; 					\
+	create_testsuite_report --maybe-color;				\
+	echo "$$col$$br$$std";						\
+	if $$success; then :; else					\
+	  echo "$${col}See $(subdir)/$(TEST_SUITE_LOG)$${std}";		\
+	  if test -n "$(PACKAGE_BUGREPORT)"; then			\
+	    echo "$${col}Please report to $(PACKAGE_BUGREPORT)$${std}";	\
+	  fi;								\
+	  echo "$$col$$br$$std";					\
+	fi;								\
+	$$success || exit 1
+
+check-TESTS: $(check_PROGRAMS)
+	@list='$(RECHECK_LOGS)';           test -z "$$list" || rm -f $$list
+	@list='$(RECHECK_LOGS:.log=.trs)'; test -z "$$list" || rm -f $$list
+	@test -z "$(TEST_SUITE_LOG)" || rm -f $(TEST_SUITE_LOG)
+	@set +e; $(am__set_TESTS_bases); \
+	log_list=`for i in $$bases; do echo $$i.log; done`; \
+	trs_list=`for i in $$bases; do echo $$i.trs; done`; \
+	log_list=`echo $$log_list`; trs_list=`echo $$trs_list`; \
+	$(MAKE) $(AM_MAKEFLAGS) $(TEST_SUITE_LOG) TEST_LOGS="$$log_list"; \
+	exit $$?;
+recheck: all $(check_PROGRAMS)
+	@test -z "$(TEST_SUITE_LOG)" || rm -f $(TEST_SUITE_LOG)
+	@set +e; $(am__set_TESTS_bases); \
+	bases=`for i in $$bases; do echo $$i; done \
+	         | $(am__list_recheck_tests)` || exit 1; \
+	log_list=`for i in $$bases; do echo $$i.log; done`; \
+	log_list=`echo $$log_list`; \
+	$(MAKE) $(AM_MAKEFLAGS) $(TEST_SUITE_LOG) \
+	        am__force_recheck=am--force-recheck \
+	        TEST_LOGS="$$log_list"; \
+	exit $$?
+fast_idiv_by_const_test.log: fast_idiv_by_const_test$(EXEEXT)
+	@p='fast_idiv_by_const_test$(EXEEXT)'; \
+	b='fast_idiv_by_const_test'; \
+	$(am__check_pre) $(LOG_DRIVER) --test-name "$$f" \
+	--log-file $$b.log --trs-file $$b.trs \
+	$(am__common_driver_flags) $(AM_LOG_DRIVER_FLAGS) $(LOG_DRIVER_FLAGS) -- $(LOG_COMPILE) \
+	"$$tst" $(AM_TESTS_FD_REDIRECT)
+.test.log:
+	@p='$<'; \
+	$(am__set_b); \
+	$(am__check_pre) $(TEST_LOG_DRIVER) --test-name "$$f" \
+	--log-file $$b.log --trs-file $$b.trs \
+	$(am__common_driver_flags) $(AM_TEST_LOG_DRIVER_FLAGS) $(TEST_LOG_DRIVER_FLAGS) -- $(TEST_LOG_COMPILE) \
+	"$$tst" $(AM_TESTS_FD_REDIRECT)
+@am__EXEEXT_TRUE@.test$(EXEEXT).log:
+@am__EXEEXT_TRUE@	@p='$<'; \
+@am__EXEEXT_TRUE@	$(am__set_b); \
+@am__EXEEXT_TRUE@	$(am__check_pre) $(TEST_LOG_DRIVER) --test-name "$$f" \
+@am__EXEEXT_TRUE@	--log-file $$b.log --trs-file $$b.trs \
+@am__EXEEXT_TRUE@	$(am__common_driver_flags) $(AM_TEST_LOG_DRIVER_FLAGS) $(TEST_LOG_DRIVER_FLAGS) -- $(TEST_LOG_COMPILE) \
+@am__EXEEXT_TRUE@	"$$tst" $(AM_TESTS_FD_REDIRECT)
+
+distdir: $(BUILT_SOURCES)
+	$(MAKE) $(AM_MAKEFLAGS) distdir-am
+
+distdir-am: $(DISTFILES)
+	@srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
+	topsrcdirstrip=`echo "$(top_srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
+	list='$(DISTFILES)'; \
+	  dist_files=`for file in $$list; do echo $$file; done | \
+	  sed -e "s|^$$srcdirstrip/||;t" \
+	      -e "s|^$$topsrcdirstrip/|$(top_builddir)/|;t"`; \
+	case $$dist_files in \
+	  */*) $(MKDIR_P) `echo "$$dist_files" | \
+			   sed '/\//!d;s|^|$(distdir)/|;s,/[^/]*$$,,' | \
+			   sort -u` ;; \
+	esac; \
+	for file in $$dist_files; do \
+	  if test -f $$file || test -d $$file; then d=.; else d=$(srcdir); fi; \
+	  if test -d $$d/$$file; then \
+	    dir=`echo "/$$file" | sed -e 's,/[^/]*$$,,'`; \
+	    if test -d "$(distdir)/$$file"; then \
+	      find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
+	    fi; \
+	    if test -d $(srcdir)/$$file && test $$d != $(srcdir); then \
+	      cp -fpR $(srcdir)/$$file "$(distdir)$$dir" || exit 1; \
+	      find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
+	    fi; \
+	    cp -fpR $$d/$$file "$(distdir)$$dir" || exit 1; \
+	  else \
+	    test -f "$(distdir)/$$file" \
+	    || cp -p $$d/$$file "$(distdir)/$$file" \
+	    || exit 1; \
+	  fi; \
+	done
+check-am: all-am
+	$(MAKE) $(AM_MAKEFLAGS) $(check_PROGRAMS)
+	$(MAKE) $(AM_MAKEFLAGS) check-TESTS
+check: check-am
+all-am: Makefile
+installdirs:
+install: install-am
+install-exec: install-exec-am
+install-data: install-data-am
+uninstall: uninstall-am
+
+install-am: all-am
+	@$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am
+
+installcheck: installcheck-am
+install-strip:
+	if test -z '$(STRIP)'; then \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	      install; \
+	else \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	    "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
+	fi
+mostlyclean-generic:
+	-test -z "$(TEST_LOGS)" || rm -f $(TEST_LOGS)
+	-test -z "$(TEST_LOGS:.log=.trs)" || rm -f $(TEST_LOGS:.log=.trs)
+	-test -z "$(TEST_SUITE_LOG)" || rm -f $(TEST_SUITE_LOG)
+
+clean-generic:
+
+distclean-generic:
+	-test -z "$(CONFIG_CLEAN_FILES)" || rm -f $(CONFIG_CLEAN_FILES)
+	-test . = "$(srcdir)" || test -z "$(CONFIG_CLEAN_VPATH_FILES)" || rm -f $(CONFIG_CLEAN_VPATH_FILES)
+
+maintainer-clean-generic:
+	@echo "This command is intended for maintainers to use"
+	@echo "it deletes files that may require special tools to rebuild."
+clean: clean-am
+
+clean-am: clean-checkPROGRAMS clean-generic clean-libtool \
+	mostlyclean-am
+
+distclean: distclean-am
+		-rm -f ./$(DEPDIR)/fast_idiv_by_const_test.Po
+	-rm -f Makefile
+distclean-am: clean-am distclean-compile distclean-generic \
+	distclean-tags
+
+dvi: dvi-am
+
+dvi-am:
+
+html: html-am
+
+html-am:
+
+info: info-am
+
+info-am:
+
+install-data-am:
+
+install-dvi: install-dvi-am
+
+install-dvi-am:
+
+install-exec-am:
+
+install-html: install-html-am
+
+install-html-am:
+
+install-info: install-info-am
+
+install-info-am:
+
+install-man:
+
+install-pdf: install-pdf-am
+
+install-pdf-am:
+
+install-ps: install-ps-am
+
+install-ps-am:
+
+installcheck-am:
+
+maintainer-clean: maintainer-clean-am
+		-rm -f ./$(DEPDIR)/fast_idiv_by_const_test.Po
+	-rm -f Makefile
+maintainer-clean-am: distclean-am maintainer-clean-generic
+
+mostlyclean: mostlyclean-am
+
+mostlyclean-am: mostlyclean-compile mostlyclean-generic \
+	mostlyclean-libtool
+
+pdf: pdf-am
+
+pdf-am:
+
+ps: ps-am
+
+ps-am:
+
+uninstall-am:
+
+.MAKE: check-am install-am install-strip
+
+.PHONY: CTAGS GTAGS TAGS all all-am am--depfiles check check-TESTS \
+	check-am clean clean-checkPROGRAMS clean-generic clean-libtool \
+	cscopelist-am ctags ctags-am distclean distclean-compile \
+	distclean-generic distclean-libtool distclean-tags distdir dvi \
+	dvi-am html html-am info info-am install install-am \
+	install-data install-data-am install-dvi install-dvi-am \
+	install-exec install-exec-am install-html install-html-am \
+	install-info install-info-am install-man install-pdf \
+	install-pdf-am install-ps install-ps-am install-strip \
+	installcheck installcheck-am installdirs maintainer-clean \
+	maintainer-clean-generic mostlyclean mostlyclean-compile \
+	mostlyclean-generic mostlyclean-libtool pdf pdf-am ps ps-am \
+	recheck tags tags-am uninstall uninstall-am
+
+.PRECIOUS: Makefile
+
+
+# Tell versions [3.59,3.63) of GNU make to not export all variables.
+# Otherwise a system limit (for SysV at least) may be exceeded.
+.NOEXPORT:
diff --git a/lib/mesa/src/util/tests/fast_idiv_by_const/fast_idiv_by_const_test.cpp b/lib/mesa/src/util/tests/fast_idiv_by_const/fast_idiv_by_const_test.cpp
new file mode 100644
index 000000000..3983a39ed
--- /dev/null
+++ b/lib/mesa/src/util/tests/fast_idiv_by_const/fast_idiv_by_const_test.cpp
@@ -0,0 +1,472 @@
+/*
+ * Copyright © 2018 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <gtest/gtest.h>
+#include "util/bigmath.h"
+#include "util/fast_idiv_by_const.h"
+#include "util/u_math.h"
+
+#define RAND_TEST_ITERATIONS 100000
+
+#define MAX_UINT(bits) \
+   (((bits) == 64) ? UINT64_MAX : ((1ull << (bits)) - 1))
+
+static inline uint64_t
+utrunc(uint64_t x, unsigned num_bits)
+{
+   if (num_bits == 64)
+      return x;
+
+   return (x << (64 - num_bits)) >> (64 - num_bits);
+}
+
+static inline int64_t
+strunc(int64_t x, unsigned num_bits)
+{
+   if (num_bits == 64)
+      return x;
+
+   return (x << (64 - num_bits)) >> (64 - num_bits);
+}
+
+static inline bool
+uint_is_in_range(uint64_t x, unsigned num_bits)
+{
+   if (num_bits == 64)
+      return true;
+
+   return x < (1ull << num_bits);
+}
+
+static inline bool
+sint_is_in_range(int64_t x, unsigned num_bits)
+{
+   if (num_bits == 64)
+      return true;
+
+   return x >= -(1ll << (num_bits - 1)) &&
+          x < (1ll << (num_bits - 1));
+}
+
+static inline uint64_t
+uadd_sat(uint64_t a, uint64_t b, unsigned num_bits)
+{
+   assert(uint_is_in_range(a, num_bits));
+   assert(uint_is_in_range(b, num_bits));
+
+   uint64_t sum = a + b;
+   if (num_bits == 64) {
+      /* Standard overflow check */
+      return sum < a ? UINT64_MAX : sum;
+   } else {
+      /* Check if sum is more than num_bits */
+      return (sum >> num_bits) ? MAX_UINT(num_bits) : sum;
+   }
+}
+
+static inline uint64_t
+umul_add_high(uint64_t a, uint64_t b, uint64_t c, unsigned num_bits)
+{
+   assert(uint_is_in_range(a, num_bits));
+   assert(uint_is_in_range(b, num_bits));
+   assert(uint_is_in_range(c, num_bits));
+
+   if (num_bits == 64) {
+      uint32_t a32[2] = { (uint32_t)a, (uint32_t)(a >> 32) };
+      uint32_t b32[2] = { (uint32_t)b, (uint32_t)(b >> 32) };
+      uint32_t c32[2] = { (uint32_t)c, (uint32_t)(c >> 32) };
+
+      uint32_t ab32[4];
+      ubm_mul_u32arr(ab32, a32, b32);
+
+      uint32_t abc32[4];
+      ubm_add_u32arr(abc32, ab32, c32);
+
+      return abc32[2] | ((uint64_t)abc32[3] << 32);
+   } else {
+      assert(num_bits <= 32);
+      return utrunc(((a * b) + c) >> num_bits, num_bits);
+   }
+}
+
+static inline int64_t
+smul_high(int64_t a, int64_t b, unsigned num_bits)
+{
+   assert(sint_is_in_range(a, num_bits));
+   assert(sint_is_in_range(b, num_bits));
+
+   if (num_bits == 64) {
+      uint32_t a32[4] = {
+         (uint32_t)a,
+         (uint32_t)(a >> 32),
+         (uint32_t)(a >> 63), /* sign extend */
+         (uint32_t)(a >> 63), /* sign extend */
+      };
+      uint32_t b32[4] = {
+         (uint32_t)b,
+         (uint32_t)(b >> 32),
+         (uint32_t)(b >> 63), /* sign extend */
+         (uint32_t)(b >> 63), /* sign extend */
+      };
+
+      uint32_t ab32[4];
+      ubm_mul_u32arr(ab32, a32, b32);
+
+      return ab32[2] | ((uint64_t)ab32[3] << 32);
+   } else {
+      assert(num_bits <= 32);
+      return strunc((a * b) >> num_bits, num_bits);
+   }
+}
+
+static inline uint64_t
+fast_udiv_add_sat(uint64_t n, struct util_fast_udiv_info m, unsigned num_bits)
+{
+   assert(uint_is_in_range(n, num_bits));
+   assert(uint_is_in_range(m.multiplier, num_bits));
+
+   n = n >> m.pre_shift;
+   n = uadd_sat(n, m.increment, num_bits);
+   n = umul_add_high(n, m.multiplier, 0, num_bits);
+   n = n >> m.post_shift;
+
+   return n;
+}
+
+static inline uint64_t
+fast_udiv_mul_add(uint64_t n, struct util_fast_udiv_info m, unsigned num_bits)
+{
+   assert(uint_is_in_range(n, num_bits));
+   assert(uint_is_in_range(m.multiplier, num_bits));
+
+   n = n >> m.pre_shift;
+   n = umul_add_high(n, m.multiplier,
+                     m.increment ? m.multiplier : 0,
+                     num_bits);
+   n = n >> m.post_shift;
+
+   return n;
+}
+
+static inline uint64_t
+fast_sdiv(int64_t n, int64_t d, struct util_fast_sdiv_info m, unsigned num_bits)
+{
+   assert(sint_is_in_range(n, num_bits));
+   assert(sint_is_in_range(d, num_bits));
+   assert(sint_is_in_range(m.multiplier, num_bits));
+
+   int64_t res;
+   res = smul_high(n, m.multiplier, num_bits);
+   if (d > 0 && m.multiplier < 0)
+      res = strunc(res + n, num_bits);
+   if (d < 0 && m.multiplier > 0)
+      res = strunc(res - n, num_bits);
+   res = res >> m.shift;
+   res = res - (res >> (num_bits - 1));
+
+   return res;
+}
+
+static uint64_t
+rand_uint(unsigned bits, unsigned min)
+{
+   assert(bits >= 4);
+
+   /* Make sure we get some small and large numbers and powers of two every
+    * once in a while
+    */
+   int k = rand() % 64;
+   if (k == 17) {
+      return min + (rand() % 16);
+   } else if (k == 42) {
+      return MAX_UINT(bits) - (rand() % 16);
+   } else if (k == 9) {
+      uint64_t r;
+      do {
+         r = 1ull << (rand() % bits);
+      } while (r < min);
+      return r;
+   }
+
+   if (min == 0) {
+      assert(bits <= 64);
+      uint64_t r = 0;
+      for (unsigned i = 0; i < 8; i++)
+         r |= ((uint64_t)rand() & 0xf) << i * 8;
+      return r >> (63 - (rand() % bits));
+   } else {
+      uint64_t r;
+      do {
+         r = rand_uint(bits, 0);
+      } while (r < min);
+      return r;
+   }
+}
+
+static int64_t
+rand_sint(unsigned bits, unsigned min_abs)
+{
+   /* Make sure we hit MIN_INT every once in a while */
+   if (rand() % 64 == 37)
+      return -(1 << (bits - 1));
+
+   int64_t s = rand_uint(bits - 1, min_abs);
+   return rand() & 1 ? s : -s;
+}
+
+static uint64_t
+udiv(uint64_t a, uint64_t b, unsigned bit_size)
+{
+   switch (bit_size) {
+   case 64: return (uint64_t)a / (uint64_t)b;
+   case 32: return (uint32_t)a / (uint32_t)b;
+   case 16: return (uint16_t)a / (uint16_t)b;
+   case 8:  return  (uint8_t)a / (uint8_t)b;
+   default:
+      assert(!"Invalid bit size");
+      return 0;
+   }
+}
+
+static int64_t
+sdiv(int64_t a, int64_t b, unsigned bit_size)
+{
+   switch (bit_size) {
+   case 64: return (int64_t)a / (int64_t)b;
+   case 32: return (int32_t)a / (int32_t)b;
+   case 16: return (int16_t)a / (int16_t)b;
+   case 8:  return  (int8_t)a / (int8_t)b;
+   default:
+      assert(!"Invalid bit size");
+      return 0;
+   }
+}
+
+static void
+random_udiv_add_sat_test(unsigned bits, bool bounded)
+{
+   for (unsigned i = 0; i < RAND_TEST_ITERATIONS; i++) {
+      uint64_t n = rand_uint(bits, 0);
+      uint64_t d = rand_uint(bits, 2);
+      assert(uint_is_in_range(n, bits));
+      assert(uint_is_in_range(d, bits) && d >= 2);
+
+      unsigned n_bits = bounded ? util_logbase2_64(MAX2(n, 1)) + 1 : bits;
+
+      struct util_fast_udiv_info m =
+         util_compute_fast_udiv_info(d, n_bits, bits);
+      EXPECT_EQ(fast_udiv_add_sat(n, m, bits), udiv(n, d, bits));
+   }
+}
+
+static void
+random_udiv_mul_add_test(unsigned bits, bool bounded)
+{
+   for (unsigned i = 0; i < RAND_TEST_ITERATIONS; i++) {
+      uint64_t n = rand_uint(bits, 0);
+      uint64_t d = rand_uint(bits, 1);
+      assert(uint_is_in_range(n, bits));
+      assert(uint_is_in_range(d, bits) && d >= 1);
+
+      unsigned n_bits = bounded ? util_logbase2_64(MAX2(n, 1)) + 1: bits;
+
+      struct util_fast_udiv_info m =
+         util_compute_fast_udiv_info(d, n_bits, bits);
+      EXPECT_EQ(fast_udiv_mul_add(n, m, bits), udiv(n, d, bits));
+   }
+}
+
+static void
+random_sdiv_test(unsigned bits)
+{
+   for (unsigned i = 0; i < RAND_TEST_ITERATIONS; i++) {
+      int64_t n = rand_sint(bits, 0);
+      int64_t d;
+      do {
+         d = rand_sint(bits, 2);
+      } while (util_is_power_of_two_or_zero64(llabs(d)));
+
+      assert(sint_is_in_range(n, bits));
+      assert(sint_is_in_range(d, bits) && llabs(d) >= 2);
+
+      struct util_fast_sdiv_info m =
+         util_compute_fast_sdiv_info(d, bits);
+      EXPECT_EQ(fast_sdiv(n, d, m, bits), sdiv(n, d, bits));
+   }
+}
+
+TEST(fast_idiv_by_const, uint8_add_sat)
+{
+   /* 8-bit is small enough we can brute-force the entire space */
+   for (unsigned d = 2; d < 256; d++) {
+      for (unsigned n_bits = 1; n_bits <= 8; n_bits++) {
+         struct util_fast_udiv_info m =
+            util_compute_fast_udiv_info(d, n_bits, 8);
+
+         for (unsigned n = 0; n < (1u << n_bits); n++)
+            EXPECT_EQ(fast_udiv_add_sat(n, m, 8), udiv(n, d, 8));
+      }
+   }
+}
+
+TEST(fast_idiv_by_const, uint8_mul_add)
+{
+   /* 8-bit is small enough we can brute-force the entire space */
+   for (unsigned d = 2; d < 256; d++) {
+      for (unsigned n_bits = 1; n_bits <= 8; n_bits++) {
+         struct util_fast_udiv_info m =
+            util_compute_fast_udiv_info(d, n_bits, 8);
+
+         for (unsigned n = 0; n < (1u << n_bits); n++)
+            EXPECT_EQ(fast_udiv_mul_add(n, m, 8), udiv(n, d, 8));
+      }
+   }
+}
+
+TEST(fast_idiv_by_const, int8)
+{
+   /* 8-bit is small enough we can brute-force the entire space */
+   for (int n = -128; n < 128; n++) {
+      for (int d = -128; d < 128; d++) {
+         if (util_is_power_of_two_or_zero(abs(d)))
+            continue;
+
+         struct util_fast_sdiv_info m =
+            util_compute_fast_sdiv_info(d, 8);
+         EXPECT_EQ(fast_sdiv(n, d, m, 8), sdiv(n, d, 8));
+      }
+   }
+}
+
+TEST(fast_idiv_by_const, uint16_add_sat_bounded)
+{
+   random_udiv_add_sat_test(16, true);
+}
+
+TEST(fast_idiv_by_const, uint16_add_sat_full)
+{
+   random_udiv_add_sat_test(16, false);
+}
+
+TEST(fast_idiv_by_const, uint16_mul_add_bounded)
+{
+   random_udiv_mul_add_test(16, true);
+}
+
+TEST(fast_idiv_by_const, uint16_mul_add_full)
+{
+   random_udiv_mul_add_test(16, false);
+}
+
+TEST(fast_idiv_by_const, int16)
+{
+   random_sdiv_test(16);
+}
+
+TEST(fast_idiv_by_const, uint32_add_sat_bounded)
+{
+   random_udiv_add_sat_test(32, true);
+}
+
+TEST(fast_idiv_by_const, uint32_add_sat_full)
+{
+   random_udiv_add_sat_test(32, false);
+}
+
+TEST(fast_idiv_by_const, uint32_mul_add_bounded)
+{
+   random_udiv_mul_add_test(32, true);
+}
+
+TEST(fast_idiv_by_const, uint32_mul_add_full)
+{
+   random_udiv_mul_add_test(32, false);
+}
+
+TEST(fast_idiv_by_const, int32)
+{
+   random_sdiv_test(32);
+}
+
+TEST(fast_idiv_by_const, util_fast_udiv32)
+{
+   for (unsigned i = 0; i < RAND_TEST_ITERATIONS; i++) {
+      uint32_t n = rand_uint(32, 0);
+      uint32_t d = rand_uint(32, 1);
+
+      struct util_fast_udiv_info m =
+         util_compute_fast_udiv_info(d, 32, 32);
+      EXPECT_EQ(util_fast_udiv32(n, m), n / d);
+   }
+}
+
+TEST(fast_idiv_by_const, util_fast_udiv32_nuw)
+{
+   for (unsigned i = 0; i < RAND_TEST_ITERATIONS; i++) {
+      uint32_t n = rand_uint(32, 0);
+      if (n == UINT32_MAX)
+         continue;
+      uint32_t d = rand_uint(32, 1);
+
+      struct util_fast_udiv_info m =
+         util_compute_fast_udiv_info(d, 32, 32);
+      EXPECT_EQ(util_fast_udiv32_nuw(n, m), n / d);
+   }
+}
+
+TEST(fast_idiv_by_const, util_fast_udiv32_u31_d_not_one)
+{
+   for (unsigned i = 0; i < RAND_TEST_ITERATIONS; i++) {
+      uint32_t n = rand_uint(31, 0);
+      uint32_t d = rand_uint(31, 2);
+
+      struct util_fast_udiv_info m =
+         util_compute_fast_udiv_info(d, 31, 32);
+      EXPECT_EQ(util_fast_udiv32_u31_d_not_one(n, m), n / d);
+   }
+}
+
+TEST(fast_idiv_by_const, uint64_add_sat_bounded)
+{
+   random_udiv_add_sat_test(64, true);
+}
+
+TEST(fast_idiv_by_const, uint64_add_sat_full)
+{
+   random_udiv_add_sat_test(64, false);
+}
+
+TEST(fast_idiv_by_const, uint64_mul_add_bounded)
+{
+   random_udiv_mul_add_test(64, true);
+}
+
+TEST(fast_idiv_by_const, uint64_mul_add_full)
+{
+   random_udiv_mul_add_test(64, false);
+}
+
+TEST(fast_idiv_by_const, int64)
+{
+   random_sdiv_test(64);
+}
diff --git a/lib/mesa/src/util/tests/fast_idiv_by_const/meson.build b/lib/mesa/src/util/tests/fast_idiv_by_const/meson.build
new file mode 100644
index 000000000..8c3b79493
--- /dev/null
+++ b/lib/mesa/src/util/tests/fast_idiv_by_const/meson.build
@@ -0,0 +1,30 @@
+# Copyright © 2018 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+test(
+  'fast_idiv_by_const',
+  executable(
+    'fast_idiv_by_const_test',
+    'fast_idiv_by_const_test.cpp',
+    dependencies : [dep_thread, dep_dl, idep_gtest],
+    include_directories : inc_common,
+    link_with : [libmesa_util],
+  )
+)
diff --git a/lib/mesa/src/util/tests/hash_table/Makefile.am b/lib/mesa/src/util/tests/hash_table/Makefile.am
index 8f12240ce..526454cb1 100644
--- a/lib/mesa/src/util/tests/hash_table/Makefile.am
+++ b/lib/mesa/src/util/tests/hash_table/Makefile.am
@@ -38,8 +38,11 @@ TESTS = \
 	insert_many \
 	null_destroy \
 	random_entry \
+	remove_key \
 	remove_null \
 	replacement \
 	$()
 
 check_PROGRAMS = $(TESTS)
+
+EXTRA_DIST = meson.build
diff --git a/lib/mesa/src/util/tests/hash_table/clear.c b/lib/mesa/src/util/tests/hash_table/clear.c
index 526700bfb..3d6bf80c0 100644
--- a/lib/mesa/src/util/tests/hash_table/clear.c
+++ b/lib/mesa/src/util/tests/hash_table/clear.c
@@ -53,7 +53,6 @@ static void delete_function(struct hash_entry *entry)
 int main()
 {
    struct hash_table *ht;
-   struct hash_entry *entry;
    const uint32_t size = 1000;
    bool flags[size];
    uint32_t i;
diff --git a/lib/mesa/src/util/tests/hash_table/collision.c b/lib/mesa/src/util/tests/hash_table/collision.c
index 69a4c29ec..ba8913281 100644
--- a/lib/mesa/src/util/tests/hash_table/collision.c
+++ b/lib/mesa/src/util/tests/hash_table/collision.c
@@ -37,7 +37,7 @@ main(int argc, char **argv)
    const char *str1 = "test1";
    const char *str2 = "test2";
    const char *str3 = "test3";
-   struct hash_entry *entry1, *entry2, *search_entry;
+   struct hash_entry *entry1, *entry2;
    uint32_t bad_hash = 5;
    int i;
 
diff --git a/lib/mesa/src/util/tests/hash_table/meson.build b/lib/mesa/src/util/tests/hash_table/meson.build
new file mode 100644
index 000000000..c7b03f19c
--- /dev/null
+++ b/lib/mesa/src/util/tests/hash_table/meson.build
@@ -0,0 +1,35 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+foreach t : ['clear', 'collision', 'delete_and_lookup', 'delete_management',
+             'destroy_callback', 'insert_and_lookup', 'insert_many',
+             'null_destroy', 'random_entry', 'remove_key', 'remove_null',
+             'replacement']
+  test(
+    t,
+    executable(
+      '@0@_test'.format(t),
+      files('@0@.c'.format(t)),
+      dependencies : [dep_thread, dep_dl],
+      include_directories : [inc_include, inc_util],
+      link_with : libmesa_util,
+    )
+  )
+endforeach
diff --git a/lib/mesa/src/util/tests/hash_table/remove_key.c b/lib/mesa/src/util/tests/hash_table/remove_key.c
new file mode 100644
index 000000000..906de6799
--- /dev/null
+++ b/lib/mesa/src/util/tests/hash_table/remove_key.c
@@ -0,0 +1,63 @@
+/*
+ * Copyright © 2018 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <assert.h>
+#include "hash_table.h"
+
+int
+main(int argc, char **argv)
+{
+   struct hash_table *ht;
+   const char *str1 = "test1";
+   const char *str2 = "test2";
+   struct hash_entry *entry;
+
+   (void) argc;
+   (void) argv;
+
+   ht = _mesa_hash_table_create(NULL, _mesa_key_hash_string, _mesa_key_string_equal);
+
+   _mesa_hash_table_insert(ht, str1, NULL);
+   _mesa_hash_table_insert(ht, str2, NULL);
+
+   entry = _mesa_hash_table_search(ht, str2);
+   assert(strcmp(entry->key, str2) == 0);
+
+   entry = _mesa_hash_table_search(ht, str1);
+   assert(strcmp(entry->key, str1) == 0);
+
+   _mesa_hash_table_remove_key(ht, str1);
+
+   entry = _mesa_hash_table_search(ht, str1);
+   assert(entry == NULL);
+
+   entry = _mesa_hash_table_search(ht, str2);
+   assert(strcmp(entry->key, str2) == 0);
+
+   _mesa_hash_table_destroy(ht, NULL);
+
+   return 0;
+}
diff --git a/lib/mesa/src/util/tests/set/Makefile.am b/lib/mesa/src/util/tests/set/Makefile.am
new file mode 100644
index 000000000..5529f4c14
--- /dev/null
+++ b/lib/mesa/src/util/tests/set/Makefile.am
@@ -0,0 +1,42 @@
+# Copyright © 2018 Intel
+#
+#  Permission is hereby granted, free of charge, to any person obtaining a
+#  copy of this software and associated documentation files (the "Software"),
+#  to deal in the Software without restriction, including without limitation
+#  the rights to use, copy, modify, merge, publish, distribute, sublicense,
+#  and/or sell copies of the Software, and to permit persons to whom the
+#  Software is furnished to do so, subject to the following conditions:
+#
+#  The above copyright notice and this permission notice (including the next
+#  paragraph) shall be included in all copies or substantial portions of the
+#  Software.
+#
+#  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+#  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+#  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+#  THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+#  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+#  FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+#  IN THE SOFTWARE.
+
+AM_CPPFLAGS = \
+	-I$(top_srcdir)/src \
+	-I$(top_srcdir)/include \
+	-I$(top_srcdir)/src/gtest/include \
+	$(PTHREAD_CFLAGS) \
+	$(DEFINES)
+
+TESTS = set_test
+
+check_PROGRAMS = $(TESTS)
+
+set_test_SOURCES = \
+	set_test.cpp
+
+set_test_LDADD = \
+	$(top_builddir)/src/gtest/libgtest.la \
+	$(top_builddir)/src/util/libmesautil.la \
+	$(PTHREAD_LIBS) \
+	$(DLOPEN_LIBS)
+
+EXTRA_DIST = meson.build
diff --git a/lib/mesa/src/util/tests/set/Makefile.in b/lib/mesa/src/util/tests/set/Makefile.in
new file mode 100644
index 000000000..0d8f2b6de
--- /dev/null
+++ b/lib/mesa/src/util/tests/set/Makefile.in
@@ -0,0 +1,1220 @@
+# Makefile.in generated by automake 1.16.1 from Makefile.am.
+# @configure_input@
+
+# Copyright (C) 1994-2018 Free Software Foundation, Inc.
+
+# This Makefile.in is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
+# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
+# PARTICULAR PURPOSE.
+
+@SET_MAKE@
+
+# Copyright © 2018 Intel
+#
+#  Permission is hereby granted, free of charge, to any person obtaining a
+#  copy of this software and associated documentation files (the "Software"),
+#  to deal in the Software without restriction, including without limitation
+#  the rights to use, copy, modify, merge, publish, distribute, sublicense,
+#  and/or sell copies of the Software, and to permit persons to whom the
+#  Software is furnished to do so, subject to the following conditions:
+#
+#  The above copyright notice and this permission notice (including the next
+#  paragraph) shall be included in all copies or substantial portions of the
+#  Software.
+#
+#  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+#  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+#  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+#  THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+#  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+#  FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+#  IN THE SOFTWARE.
+VPATH = @srcdir@
+am__is_gnu_make = { \
+  if test -z '$(MAKELEVEL)'; then \
+    false; \
+  elif test -n '$(MAKE_HOST)'; then \
+    true; \
+  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
+    true; \
+  else \
+    false; \
+  fi; \
+}
+am__make_running_with_option = \
+  case $${target_option-} in \
+      ?) ;; \
+      *) echo "am__make_running_with_option: internal error: invalid" \
+              "target option '$${target_option-}' specified" >&2; \
+         exit 1;; \
+  esac; \
+  has_opt=no; \
+  sane_makeflags=$$MAKEFLAGS; \
+  if $(am__is_gnu_make); then \
+    sane_makeflags=$$MFLAGS; \
+  else \
+    case $$MAKEFLAGS in \
+      *\\[\ \	]*) \
+        bs=\\; \
+        sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
+          | sed "s/$$bs$$bs[$$bs $$bs	]*//g"`;; \
+    esac; \
+  fi; \
+  skip_next=no; \
+  strip_trailopt () \
+  { \
+    flg=`printf '%s\n' "$$flg" | sed "s/$$1.*$$//"`; \
+  }; \
+  for flg in $$sane_makeflags; do \
+    test $$skip_next = yes && { skip_next=no; continue; }; \
+    case $$flg in \
+      *=*|--*) continue;; \
+        -*I) strip_trailopt 'I'; skip_next=yes;; \
+      -*I?*) strip_trailopt 'I';; \
+        -*O) strip_trailopt 'O'; skip_next=yes;; \
+      -*O?*) strip_trailopt 'O';; \
+        -*l) strip_trailopt 'l'; skip_next=yes;; \
+      -*l?*) strip_trailopt 'l';; \
+      -[dEDm]) skip_next=yes;; \
+      -[JT]) skip_next=yes;; \
+    esac; \
+    case $$flg in \
+      *$$target_option*) has_opt=yes; break;; \
+    esac; \
+  done; \
+  test $$has_opt = yes
+am__make_dryrun = (target_option=n; $(am__make_running_with_option))
+am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
+pkgdatadir = $(datadir)/@PACKAGE@
+pkgincludedir = $(includedir)/@PACKAGE@
+pkglibdir = $(libdir)/@PACKAGE@
+pkglibexecdir = $(libexecdir)/@PACKAGE@
+am__cd = CDPATH="$${ZSH_VERSION+.}$(PATH_SEPARATOR)" && cd
+install_sh_DATA = $(install_sh) -c -m 644
+install_sh_PROGRAM = $(install_sh) -c
+install_sh_SCRIPT = $(install_sh) -c
+INSTALL_HEADER = $(INSTALL_DATA)
+transform = $(program_transform_name)
+NORMAL_INSTALL = :
+PRE_INSTALL = :
+POST_INSTALL = :
+NORMAL_UNINSTALL = :
+PRE_UNINSTALL = :
+POST_UNINSTALL = :
+build_triplet = @build@
+host_triplet = @host@
+target_triplet = @target@
+TESTS = set_test$(EXEEXT)
+check_PROGRAMS = $(am__EXEEXT_1)
+subdir = src/util/tests/set
+ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
+am__aclocal_m4_deps = $(top_srcdir)/m4/ax_check_compile_flag.m4 \
+	$(top_srcdir)/m4/ax_check_gnu_make.m4 \
+	$(top_srcdir)/m4/ax_check_python_mako_module.m4 \
+	$(top_srcdir)/m4/ax_gcc_builtin.m4 \
+	$(top_srcdir)/m4/ax_gcc_func_attribute.m4 \
+	$(top_srcdir)/m4/ax_prog_bison.m4 \
+	$(top_srcdir)/m4/ax_prog_flex.m4 \
+	$(top_srcdir)/m4/ax_pthread.m4 $(top_srcdir)/m4/libtool.m4 \
+	$(top_srcdir)/m4/ltoptions.m4 $(top_srcdir)/m4/ltsugar.m4 \
+	$(top_srcdir)/m4/ltversion.m4 $(top_srcdir)/m4/lt~obsolete.m4 \
+	$(top_srcdir)/VERSION $(top_srcdir)/configure.ac
+am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
+	$(ACLOCAL_M4)
+DIST_COMMON = $(srcdir)/Makefile.am $(am__DIST_COMMON)
+mkinstalldirs = $(install_sh) -d
+CONFIG_CLEAN_FILES =
+CONFIG_CLEAN_VPATH_FILES =
+am__EXEEXT_1 = set_test$(EXEEXT)
+am_set_test_OBJECTS = set_test.$(OBJEXT)
+set_test_OBJECTS = $(am_set_test_OBJECTS)
+am__DEPENDENCIES_1 =
+set_test_DEPENDENCIES = $(top_builddir)/src/gtest/libgtest.la \
+	$(top_builddir)/src/util/libmesautil.la $(am__DEPENDENCIES_1) \
+	$(am__DEPENDENCIES_1)
+AM_V_lt = $(am__v_lt_@AM_V@)
+am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@)
+am__v_lt_0 = --silent
+am__v_lt_1 = 
+AM_V_P = $(am__v_P_@AM_V@)
+am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
+am__v_P_0 = false
+am__v_P_1 = :
+AM_V_GEN = $(am__v_GEN_@AM_V@)
+am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
+am__v_GEN_0 = @echo "  GEN     " $@;
+am__v_GEN_1 = 
+AM_V_at = $(am__v_at_@AM_V@)
+am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
+am__v_at_0 = @
+am__v_at_1 = 
+DEFAULT_INCLUDES = -I.@am__isrc@
+depcomp = $(SHELL) $(top_srcdir)/bin/depcomp
+am__maybe_remake_depfiles = depfiles
+am__depfiles_remade = ./$(DEPDIR)/set_test.Po
+am__mv = mv -f
+CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
+	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
+LTCXXCOMPILE = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
+	$(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) \
+	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
+	$(AM_CXXFLAGS) $(CXXFLAGS)
+AM_V_CXX = $(am__v_CXX_@AM_V@)
+am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
+am__v_CXX_0 = @echo "  CXX     " $@;
+am__v_CXX_1 = 
+CXXLD = $(CXX)
+CXXLINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
+	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(AM_CXXFLAGS) \
+	$(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) -o $@
+AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
+am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
+am__v_CXXLD_0 = @echo "  CXXLD   " $@;
+am__v_CXXLD_1 = 
+SOURCES = $(set_test_SOURCES)
+DIST_SOURCES = $(set_test_SOURCES)
+am__can_run_installinfo = \
+  case $$AM_UPDATE_INFO_DIR in \
+    n|no|NO) false;; \
+    *) (install-info --version) >/dev/null 2>&1;; \
+  esac
+am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
+# Read a list of newline-separated strings from the standard input,
+# and print each of them once, without duplicates.  Input order is
+# *not* preserved.
+am__uniquify_input = $(AWK) '\
+  BEGIN { nonempty = 0; } \
+  { items[$$0] = 1; nonempty = 1; } \
+  END { if (nonempty) { for (i in items) print i; }; } \
+'
+# Make sure the list of sources is unique.  This is necessary because,
+# e.g., the same source file might be shared among _SOURCES variables
+# for different programs/libraries.
+am__define_uniq_tagged_files = \
+  list='$(am__tagged_files)'; \
+  unique=`for i in $$list; do \
+    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+  done | $(am__uniquify_input)`
+ETAGS = etags
+CTAGS = ctags
+am__tty_colors_dummy = \
+  mgn= red= grn= lgn= blu= brg= std=; \
+  am__color_tests=no
+am__tty_colors = { \
+  $(am__tty_colors_dummy); \
+  if test "X$(AM_COLOR_TESTS)" = Xno; then \
+    am__color_tests=no; \
+  elif test "X$(AM_COLOR_TESTS)" = Xalways; then \
+    am__color_tests=yes; \
+  elif test "X$$TERM" != Xdumb && { test -t 1; } 2>/dev/null; then \
+    am__color_tests=yes; \
+  fi; \
+  if test $$am__color_tests = yes; then \
+    red=''; \
+    grn=''; \
+    lgn=''; \
+    blu=''; \
+    mgn=''; \
+    brg=''; \
+    std=''; \
+  fi; \
+}
+am__vpath_adj_setup = srcdirstrip=`echo "$(srcdir)" | sed 's|.|.|g'`;
+am__vpath_adj = case $$p in \
+    $(srcdir)/*) f=`echo "$$p" | sed "s|^$$srcdirstrip/||"`;; \
+    *) f=$$p;; \
+  esac;
+am__strip_dir = f=`echo $$p | sed -e 's|^.*/||'`;
+am__install_max = 40
+am__nobase_strip_setup = \
+  srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*|]/\\\\&/g'`
+am__nobase_strip = \
+  for p in $$list; do echo "$$p"; done | sed -e "s|$$srcdirstrip/||"
+am__nobase_list = $(am__nobase_strip_setup); \
+  for p in $$list; do echo "$$p $$p"; done | \
+  sed "s| $$srcdirstrip/| |;"' / .*\//!s/ .*/ ./; s,\( .*\)/[^/]*$$,\1,' | \
+  $(AWK) 'BEGIN { files["."] = "" } { files[$$2] = files[$$2] " " $$1; \
+    if (++n[$$2] == $(am__install_max)) \
+      { print $$2, files[$$2]; n[$$2] = 0; files[$$2] = "" } } \
+    END { for (dir in files) print dir, files[dir] }'
+am__base_list = \
+  sed '$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;s/\n/ /g' | \
+  sed '$$!N;$$!N;$$!N;$$!N;s/\n/ /g'
+am__uninstall_files_from_dir = { \
+  test -z "$$files" \
+    || { test ! -d "$$dir" && test ! -f "$$dir" && test ! -r "$$dir"; } \
+    || { echo " ( cd '$$dir' && rm -f" $$files ")"; \
+         $(am__cd) "$$dir" && rm -f $$files; }; \
+  }
+am__recheck_rx = ^[ 	]*:recheck:[ 	]*
+am__global_test_result_rx = ^[ 	]*:global-test-result:[ 	]*
+am__copy_in_global_log_rx = ^[ 	]*:copy-in-global-log:[ 	]*
+# A command that, given a newline-separated list of test names on the
+# standard input, print the name of the tests that are to be re-run
+# upon "make recheck".
+am__list_recheck_tests = $(AWK) '{ \
+  recheck = 1; \
+  while ((rc = (getline line < ($$0 ".trs"))) != 0) \
+    { \
+      if (rc < 0) \
+        { \
+          if ((getline line2 < ($$0 ".log")) < 0) \
+	    recheck = 0; \
+          break; \
+        } \
+      else if (line ~ /$(am__recheck_rx)[nN][Oo]/) \
+        { \
+          recheck = 0; \
+          break; \
+        } \
+      else if (line ~ /$(am__recheck_rx)[yY][eE][sS]/) \
+        { \
+          break; \
+        } \
+    }; \
+  if (recheck) \
+    print $$0; \
+  close ($$0 ".trs"); \
+  close ($$0 ".log"); \
+}'
+# A command that, given a newline-separated list of test names on the
+# standard input, create the global log from their .trs and .log files.
+am__create_global_log = $(AWK) ' \
+function fatal(msg) \
+{ \
+  print "fatal: making $@: " msg | "cat >&2"; \
+  exit 1; \
+} \
+function rst_section(header) \
+{ \
+  print header; \
+  len = length(header); \
+  for (i = 1; i <= len; i = i + 1) \
+    printf "="; \
+  printf "\n\n"; \
+} \
+{ \
+  copy_in_global_log = 1; \
+  global_test_result = "RUN"; \
+  while ((rc = (getline line < ($$0 ".trs"))) != 0) \
+    { \
+      if (rc < 0) \
+         fatal("failed to read from " $$0 ".trs"); \
+      if (line ~ /$(am__global_test_result_rx)/) \
+        { \
+          sub("$(am__global_test_result_rx)", "", line); \
+          sub("[ 	]*$$", "", line); \
+          global_test_result = line; \
+        } \
+      else if (line ~ /$(am__copy_in_global_log_rx)[nN][oO]/) \
+        copy_in_global_log = 0; \
+    }; \
+  if (copy_in_global_log) \
+    { \
+      rst_section(global_test_result ": " $$0); \
+      while ((rc = (getline line < ($$0 ".log"))) != 0) \
+      { \
+        if (rc < 0) \
+          fatal("failed to read from " $$0 ".log"); \
+        print line; \
+      }; \
+      printf "\n"; \
+    }; \
+  close ($$0 ".trs"); \
+  close ($$0 ".log"); \
+}'
+# Restructured Text title.
+am__rst_title = { sed 's/.*/   &   /;h;s/./=/g;p;x;s/ *$$//;p;g' && echo; }
+# Solaris 10 'make', and several other traditional 'make' implementations,
+# pass "-e" to $(SHELL), and POSIX 2008 even requires this.  Work around it
+# by disabling -e (using the XSI extension "set +e") if it's set.
+am__sh_e_setup = case $$- in *e*) set +e;; esac
+# Default flags passed to test drivers.
+am__common_driver_flags = \
+  --color-tests "$$am__color_tests" \
+  --enable-hard-errors "$$am__enable_hard_errors" \
+  --expect-failure "$$am__expect_failure"
+# To be inserted before the command running the test.  Creates the
+# directory for the log if needed.  Stores in $dir the directory
+# containing $f, in $tst the test, in $log the log.  Executes the
+# developer- defined test setup AM_TESTS_ENVIRONMENT (if any), and
+# passes TESTS_ENVIRONMENT.  Set up options for the wrapper that
+# will run the test scripts (or their associated LOG_COMPILER, if
+# thy have one).
+am__check_pre = \
+$(am__sh_e_setup);					\
+$(am__vpath_adj_setup) $(am__vpath_adj)			\
+$(am__tty_colors);					\
+srcdir=$(srcdir); export srcdir;			\
+case "$@" in						\
+  */*) am__odir=`echo "./$@" | sed 's|/[^/]*$$||'`;;	\
+    *) am__odir=.;; 					\
+esac;							\
+test "x$$am__odir" = x"." || test -d "$$am__odir" 	\
+  || $(MKDIR_P) "$$am__odir" || exit $$?;		\
+if test -f "./$$f"; then dir=./;			\
+elif test -f "$$f"; then dir=;				\
+else dir="$(srcdir)/"; fi;				\
+tst=$$dir$$f; log='$@'; 				\
+if test -n '$(DISABLE_HARD_ERRORS)'; then		\
+  am__enable_hard_errors=no; 				\
+else							\
+  am__enable_hard_errors=yes; 				\
+fi; 							\
+case " $(XFAIL_TESTS) " in				\
+  *[\ \	]$$f[\ \	]* | *[\ \	]$$dir$$f[\ \	]*) \
+    am__expect_failure=yes;;				\
+  *)							\
+    am__expect_failure=no;;				\
+esac; 							\
+$(AM_TESTS_ENVIRONMENT) $(TESTS_ENVIRONMENT)
+# A shell command to get the names of the tests scripts with any registered
+# extension removed (i.e., equivalently, the names of the test logs, with
+# the '.log' extension removed).  The result is saved in the shell variable
+# '$bases'.  This honors runtime overriding of TESTS and TEST_LOGS.  Sadly,
+# we cannot use something simpler, involving e.g., "$(TEST_LOGS:.log=)",
+# since that might cause problem with VPATH rewrites for suffix-less tests.
+# See also 'test-harness-vpath-rewrite.sh' and 'test-trs-basic.sh'.
+am__set_TESTS_bases = \
+  bases='$(TEST_LOGS)'; \
+  bases=`for i in $$bases; do echo $$i; done | sed 's/\.log$$//'`; \
+  bases=`echo $$bases`
+RECHECK_LOGS = $(TEST_LOGS)
+AM_RECURSIVE_TARGETS = check recheck
+TEST_SUITE_LOG = test-suite.log
+TEST_EXTENSIONS = @EXEEXT@ .test
+LOG_DRIVER = $(SHELL) $(top_srcdir)/bin/test-driver
+LOG_COMPILE = $(LOG_COMPILER) $(AM_LOG_FLAGS) $(LOG_FLAGS)
+am__set_b = \
+  case '$@' in \
+    */*) \
+      case '$*' in \
+        */*) b='$*';; \
+          *) b=`echo '$@' | sed 's/\.log$$//'`; \
+       esac;; \
+    *) \
+      b='$*';; \
+  esac
+am__test_logs1 = $(TESTS:=.log)
+am__test_logs2 = $(am__test_logs1:@EXEEXT@.log=.log)
+TEST_LOGS = $(am__test_logs2:.test.log=.log)
+TEST_LOG_DRIVER = $(SHELL) $(top_srcdir)/bin/test-driver
+TEST_LOG_COMPILE = $(TEST_LOG_COMPILER) $(AM_TEST_LOG_FLAGS) \
+	$(TEST_LOG_FLAGS)
+am__DIST_COMMON = $(srcdir)/Makefile.in $(top_srcdir)/bin/depcomp \
+	$(top_srcdir)/bin/test-driver
+DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
+ACLOCAL = @ACLOCAL@
+AMDGPU_CFLAGS = @AMDGPU_CFLAGS@
+AMDGPU_LIBS = @AMDGPU_LIBS@
+AMTAR = @AMTAR@
+AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
+ANDROID_CFLAGS = @ANDROID_CFLAGS@
+ANDROID_LIBS = @ANDROID_LIBS@
+AR = @AR@
+AUTOCONF = @AUTOCONF@
+AUTOHEADER = @AUTOHEADER@
+AUTOMAKE = @AUTOMAKE@
+AWK = @AWK@
+BACKTRACE_CFLAGS = @BACKTRACE_CFLAGS@
+BACKTRACE_LIBS = @BACKTRACE_LIBS@
+BSYMBOLIC = @BSYMBOLIC@
+CC = @CC@
+CCAS = @CCAS@
+CCASDEPMODE = @CCASDEPMODE@
+CCASFLAGS = @CCASFLAGS@
+CCDEPMODE = @CCDEPMODE@
+CFLAGS = @CFLAGS@
+CLANG_RESOURCE_DIR = @CLANG_RESOURCE_DIR@
+CLOCK_LIB = @CLOCK_LIB@
+CLOVER_STD_OVERRIDE = @CLOVER_STD_OVERRIDE@
+CPP = @CPP@
+CPPFLAGS = @CPPFLAGS@
+CXX = @CXX@
+CXX11_CXXFLAGS = @CXX11_CXXFLAGS@
+CXXCPP = @CXXCPP@
+CXXDEPMODE = @CXXDEPMODE@
+CXXFLAGS = @CXXFLAGS@
+CYGPATH_W = @CYGPATH_W@
+D3D_DRIVER_INSTALL_DIR = @D3D_DRIVER_INSTALL_DIR@
+DEFINES = @DEFINES@
+DEFS = @DEFS@
+DEPDIR = @DEPDIR@
+DLLTOOL = @DLLTOOL@
+DLOPEN_LIBS = @DLOPEN_LIBS@
+DRI2PROTO_CFLAGS = @DRI2PROTO_CFLAGS@
+DRI2PROTO_LIBS = @DRI2PROTO_LIBS@
+DRIGL_CFLAGS = @DRIGL_CFLAGS@
+DRIGL_LIBS = @DRIGL_LIBS@
+DRI_DRIVER_INSTALL_DIR = @DRI_DRIVER_INSTALL_DIR@
+DRI_DRIVER_SEARCH_DIR = @DRI_DRIVER_SEARCH_DIR@
+DRI_LIB_DEPS = @DRI_LIB_DEPS@
+DRI_PC_REQ_PRIV = @DRI_PC_REQ_PRIV@
+DSYMUTIL = @DSYMUTIL@
+DUMPBIN = @DUMPBIN@
+ECHO_C = @ECHO_C@
+ECHO_N = @ECHO_N@
+ECHO_T = @ECHO_T@
+EGL_CFLAGS = @EGL_CFLAGS@
+EGL_LIB_DEPS = @EGL_LIB_DEPS@
+EGL_NATIVE_PLATFORM = @EGL_NATIVE_PLATFORM@
+EGREP = @EGREP@
+ETNAVIV_CFLAGS = @ETNAVIV_CFLAGS@
+ETNAVIV_LIBS = @ETNAVIV_LIBS@
+EXEEXT = @EXEEXT@
+EXPAT_CFLAGS = @EXPAT_CFLAGS@
+EXPAT_LIBS = @EXPAT_LIBS@
+FGREP = @FGREP@
+GALLIUM_PIPE_LOADER_DEFINES = @GALLIUM_PIPE_LOADER_DEFINES@
+GBM_PC_LIB_PRIV = @GBM_PC_LIB_PRIV@
+GBM_PC_REQ_PRIV = @GBM_PC_REQ_PRIV@
+GC_SECTIONS = @GC_SECTIONS@
+GLESv1_CM_LIB_DEPS = @GLESv1_CM_LIB_DEPS@
+GLESv1_CM_PC_LIB_PRIV = @GLESv1_CM_PC_LIB_PRIV@
+GLESv2_LIB_DEPS = @GLESv2_LIB_DEPS@
+GLESv2_PC_LIB_PRIV = @GLESv2_PC_LIB_PRIV@
+GLPROTO_CFLAGS = @GLPROTO_CFLAGS@
+GLPROTO_LIBS = @GLPROTO_LIBS@
+GLVND_CFLAGS = @GLVND_CFLAGS@
+GLVND_LIBS = @GLVND_LIBS@
+GLX_TLS = @GLX_TLS@
+GL_LIB = @GL_LIB@
+GL_LIB_DEPS = @GL_LIB_DEPS@
+GL_PC_CFLAGS = @GL_PC_CFLAGS@
+GL_PC_LIB_PRIV = @GL_PC_LIB_PRIV@
+GL_PC_REQ_PRIV = @GL_PC_REQ_PRIV@
+GL_PKGCONF_LIB = @GL_PKGCONF_LIB@
+GREP = @GREP@
+I915_CFLAGS = @I915_CFLAGS@
+I915_LIBS = @I915_LIBS@
+INDENT = @INDENT@
+INDENT_FLAGS = @INDENT_FLAGS@
+INSTALL = @INSTALL@
+INSTALL_DATA = @INSTALL_DATA@
+INSTALL_PROGRAM = @INSTALL_PROGRAM@
+INSTALL_SCRIPT = @INSTALL_SCRIPT@
+INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@
+LD = @LD@
+LDFLAGS = @LDFLAGS@
+LD_BUILD_ID = @LD_BUILD_ID@
+LD_NO_UNDEFINED = @LD_NO_UNDEFINED@
+LEX = @LEX@
+LEXLIB = @LEXLIB@
+LEX_OUTPUT_ROOT = @LEX_OUTPUT_ROOT@
+LIBATOMIC_LIBS = @LIBATOMIC_LIBS@
+LIBCLC_INCLUDEDIR = @LIBCLC_INCLUDEDIR@
+LIBCLC_LIBEXECDIR = @LIBCLC_LIBEXECDIR@
+LIBDRM_CFLAGS = @LIBDRM_CFLAGS@
+LIBDRM_LIBS = @LIBDRM_LIBS@
+LIBELF_CFLAGS = @LIBELF_CFLAGS@
+LIBELF_LIBS = @LIBELF_LIBS@
+LIBGLVND_DATADIR = @LIBGLVND_DATADIR@
+LIBOBJS = @LIBOBJS@
+LIBS = @LIBS@
+LIBSENSORS_LIBS = @LIBSENSORS_LIBS@
+LIBTOOL = @LIBTOOL@
+LIBUNWIND_CFLAGS = @LIBUNWIND_CFLAGS@
+LIBUNWIND_LIBS = @LIBUNWIND_LIBS@
+LIB_DIR = @LIB_DIR@
+LIB_EXT = @LIB_EXT@
+LIPO = @LIPO@
+LLVM_CFLAGS = @LLVM_CFLAGS@
+LLVM_CONFIG = @LLVM_CONFIG@
+LLVM_CXXFLAGS = @LLVM_CXXFLAGS@
+LLVM_INCLUDEDIR = @LLVM_INCLUDEDIR@
+LLVM_LDFLAGS = @LLVM_LDFLAGS@
+LLVM_LIBS = @LLVM_LIBS@
+LN_S = @LN_S@
+LTLIBOBJS = @LTLIBOBJS@
+LT_SYS_LIBRARY_PATH = @LT_SYS_LIBRARY_PATH@
+MAKEINFO = @MAKEINFO@
+MANIFEST_TOOL = @MANIFEST_TOOL@
+MKDIR_P = @MKDIR_P@
+MSVC2013_COMPAT_CFLAGS = @MSVC2013_COMPAT_CFLAGS@
+MSVC2013_COMPAT_CXXFLAGS = @MSVC2013_COMPAT_CXXFLAGS@
+NINE_MAJOR = @NINE_MAJOR@
+NINE_MINOR = @NINE_MINOR@
+NINE_PATCH = @NINE_PATCH@
+NINE_VERSION = @NINE_VERSION@
+NM = @NM@
+NMEDIT = @NMEDIT@
+NOUVEAU_CFLAGS = @NOUVEAU_CFLAGS@
+NOUVEAU_LIBS = @NOUVEAU_LIBS@
+NVVIEUX_CFLAGS = @NVVIEUX_CFLAGS@
+NVVIEUX_LIBS = @NVVIEUX_LIBS@
+OBJDUMP = @OBJDUMP@
+OBJEXT = @OBJEXT@
+OMX_BELLAGIO_CFLAGS = @OMX_BELLAGIO_CFLAGS@
+OMX_BELLAGIO_LIBS = @OMX_BELLAGIO_LIBS@
+OMX_BELLAGIO_LIB_INSTALL_DIR = @OMX_BELLAGIO_LIB_INSTALL_DIR@
+OMX_TIZONIA_CFLAGS = @OMX_TIZONIA_CFLAGS@
+OMX_TIZONIA_LIBS = @OMX_TIZONIA_LIBS@
+OMX_TIZONIA_LIB_INSTALL_DIR = @OMX_TIZONIA_LIB_INSTALL_DIR@
+OPENCL_LIBNAME = @OPENCL_LIBNAME@
+OPENCL_VERSION = @OPENCL_VERSION@
+OSMESA_LIB = @OSMESA_LIB@
+OSMESA_LIB_DEPS = @OSMESA_LIB_DEPS@
+OSMESA_PC_LIB_PRIV = @OSMESA_PC_LIB_PRIV@
+OSMESA_PC_REQ = @OSMESA_PC_REQ@
+OSMESA_VERSION = @OSMESA_VERSION@
+OTOOL = @OTOOL@
+OTOOL64 = @OTOOL64@
+PACKAGE = @PACKAGE@
+PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
+PACKAGE_NAME = @PACKAGE_NAME@
+PACKAGE_STRING = @PACKAGE_STRING@
+PACKAGE_TARNAME = @PACKAGE_TARNAME@
+PACKAGE_URL = @PACKAGE_URL@
+PACKAGE_VERSION = @PACKAGE_VERSION@
+PATH_SEPARATOR = @PATH_SEPARATOR@
+PKG_CONFIG = @PKG_CONFIG@
+PKG_CONFIG_LIBDIR = @PKG_CONFIG_LIBDIR@
+PKG_CONFIG_PATH = @PKG_CONFIG_PATH@
+POSIX_SHELL = @POSIX_SHELL@
+PTHREADSTUBS_CFLAGS = @PTHREADSTUBS_CFLAGS@
+PTHREADSTUBS_LIBS = @PTHREADSTUBS_LIBS@
+PTHREAD_CC = @PTHREAD_CC@
+PTHREAD_CFLAGS = @PTHREAD_CFLAGS@
+PTHREAD_LIBS = @PTHREAD_LIBS@
+PWR8_CFLAGS = @PWR8_CFLAGS@
+PYTHON = @PYTHON@
+PYTHON_EXEC_PREFIX = @PYTHON_EXEC_PREFIX@
+PYTHON_PLATFORM = @PYTHON_PLATFORM@
+PYTHON_PREFIX = @PYTHON_PREFIX@
+PYTHON_VERSION = @PYTHON_VERSION@
+RADEON_CFLAGS = @RADEON_CFLAGS@
+RADEON_LIBS = @RADEON_LIBS@
+RANLIB = @RANLIB@
+RM = @RM@
+SCANNER_ARG = @SCANNER_ARG@
+SED = @SED@
+SELINUX_CFLAGS = @SELINUX_CFLAGS@
+SELINUX_LIBS = @SELINUX_LIBS@
+SET_MAKE = @SET_MAKE@
+SHELL = @SHELL@
+SIMPENROSE_CFLAGS = @SIMPENROSE_CFLAGS@
+SIMPENROSE_LIBS = @SIMPENROSE_LIBS@
+SSE41_CFLAGS = @SSE41_CFLAGS@
+STRIP = @STRIP@
+SWR_AVX2_CXXFLAGS = @SWR_AVX2_CXXFLAGS@
+SWR_AVX_CXXFLAGS = @SWR_AVX_CXXFLAGS@
+SWR_KNL_CXXFLAGS = @SWR_KNL_CXXFLAGS@
+SWR_SKX_CXXFLAGS = @SWR_SKX_CXXFLAGS@
+V3D_SIMULATOR_CFLAGS = @V3D_SIMULATOR_CFLAGS@
+V3D_SIMULATOR_LIBS = @V3D_SIMULATOR_LIBS@
+VALGRIND_CFLAGS = @VALGRIND_CFLAGS@
+VALGRIND_LIBS = @VALGRIND_LIBS@
+VA_CFLAGS = @VA_CFLAGS@
+VA_LIBS = @VA_LIBS@
+VA_LIB_INSTALL_DIR = @VA_LIB_INSTALL_DIR@
+VA_MAJOR = @VA_MAJOR@
+VA_MINOR = @VA_MINOR@
+VC4_CFLAGS = @VC4_CFLAGS@
+VC4_LIBS = @VC4_LIBS@
+VDPAU_CFLAGS = @VDPAU_CFLAGS@
+VDPAU_LIBS = @VDPAU_LIBS@
+VDPAU_LIB_INSTALL_DIR = @VDPAU_LIB_INSTALL_DIR@
+VDPAU_MAJOR = @VDPAU_MAJOR@
+VDPAU_MINOR = @VDPAU_MINOR@
+VERSION = @VERSION@
+VISIBILITY_CFLAGS = @VISIBILITY_CFLAGS@
+VISIBILITY_CXXFLAGS = @VISIBILITY_CXXFLAGS@
+VL_CFLAGS = @VL_CFLAGS@
+VL_LIBS = @VL_LIBS@
+VULKAN_ICD_INSTALL_DIR = @VULKAN_ICD_INSTALL_DIR@
+WAYLAND_CLIENT_CFLAGS = @WAYLAND_CLIENT_CFLAGS@
+WAYLAND_CLIENT_LIBS = @WAYLAND_CLIENT_LIBS@
+WAYLAND_EGL_CFLAGS = @WAYLAND_EGL_CFLAGS@
+WAYLAND_EGL_LIBS = @WAYLAND_EGL_LIBS@
+WAYLAND_PROTOCOLS_CFLAGS = @WAYLAND_PROTOCOLS_CFLAGS@
+WAYLAND_PROTOCOLS_DATADIR = @WAYLAND_PROTOCOLS_DATADIR@
+WAYLAND_PROTOCOLS_LIBS = @WAYLAND_PROTOCOLS_LIBS@
+WAYLAND_SCANNER = @WAYLAND_SCANNER@
+WAYLAND_SCANNER_CFLAGS = @WAYLAND_SCANNER_CFLAGS@
+WAYLAND_SCANNER_LIBS = @WAYLAND_SCANNER_LIBS@
+WAYLAND_SERVER_CFLAGS = @WAYLAND_SERVER_CFLAGS@
+WAYLAND_SERVER_LIBS = @WAYLAND_SERVER_LIBS@
+WNO_OVERRIDE_INIT = @WNO_OVERRIDE_INIT@
+X11_INCLUDES = @X11_INCLUDES@
+XA_MAJOR = @XA_MAJOR@
+XA_MINOR = @XA_MINOR@
+XA_PATCH = @XA_PATCH@
+XA_VERSION = @XA_VERSION@
+XCB_DRI2_CFLAGS = @XCB_DRI2_CFLAGS@
+XCB_DRI2_LIBS = @XCB_DRI2_LIBS@
+XCB_DRI3_CFLAGS = @XCB_DRI3_CFLAGS@
+XCB_DRI3_LIBS = @XCB_DRI3_LIBS@
+XCB_DRI3_MODIFIERS_CFLAGS = @XCB_DRI3_MODIFIERS_CFLAGS@
+XCB_DRI3_MODIFIERS_LIBS = @XCB_DRI3_MODIFIERS_LIBS@
+XCB_RANDR_CFLAGS = @XCB_RANDR_CFLAGS@
+XCB_RANDR_LIBS = @XCB_RANDR_LIBS@
+XLIBGL_CFLAGS = @XLIBGL_CFLAGS@
+XLIBGL_LIBS = @XLIBGL_LIBS@
+XLIB_RANDR_CFLAGS = @XLIB_RANDR_CFLAGS@
+XLIB_RANDR_LIBS = @XLIB_RANDR_LIBS@
+XVMC_CFLAGS = @XVMC_CFLAGS@
+XVMC_LIBS = @XVMC_LIBS@
+XVMC_LIB_INSTALL_DIR = @XVMC_LIB_INSTALL_DIR@
+XVMC_MAJOR = @XVMC_MAJOR@
+XVMC_MINOR = @XVMC_MINOR@
+YACC = @YACC@
+YFLAGS = @YFLAGS@
+ZLIB_CFLAGS = @ZLIB_CFLAGS@
+ZLIB_LIBS = @ZLIB_LIBS@
+abs_builddir = @abs_builddir@
+abs_srcdir = @abs_srcdir@
+abs_top_builddir = @abs_top_builddir@
+abs_top_srcdir = @abs_top_srcdir@
+ac_ct_AR = @ac_ct_AR@
+ac_ct_CC = @ac_ct_CC@
+ac_ct_CXX = @ac_ct_CXX@
+ac_ct_DUMPBIN = @ac_ct_DUMPBIN@
+acv_mako_found = @acv_mako_found@
+am__include = @am__include@
+am__leading_dot = @am__leading_dot@
+am__quote = @am__quote@
+am__tar = @am__tar@
+am__untar = @am__untar@
+ax_pthread_config = @ax_pthread_config@
+bindir = @bindir@
+build = @build@
+build_alias = @build_alias@
+build_cpu = @build_cpu@
+build_os = @build_os@
+build_vendor = @build_vendor@
+builddir = @builddir@
+datadir = @datadir@
+datarootdir = @datarootdir@
+docdir = @docdir@
+dvidir = @dvidir@
+exec_prefix = @exec_prefix@
+host = @host@
+host_alias = @host_alias@
+host_cpu = @host_cpu@
+host_os = @host_os@
+host_vendor = @host_vendor@
+htmldir = @htmldir@
+ifGNUmake = @ifGNUmake@
+includedir = @includedir@
+infodir = @infodir@
+install_sh = @install_sh@
+libdir = @libdir@
+libexecdir = @libexecdir@
+localedir = @localedir@
+localstatedir = @localstatedir@
+mandir = @mandir@
+mkdir_p = @mkdir_p@
+oldincludedir = @oldincludedir@
+pdfdir = @pdfdir@
+pkgpyexecdir = @pkgpyexecdir@
+pkgpythondir = @pkgpythondir@
+prefix = @prefix@
+program_transform_name = @program_transform_name@
+psdir = @psdir@
+pyexecdir = @pyexecdir@
+pythondir = @pythondir@
+sbindir = @sbindir@
+sharedstatedir = @sharedstatedir@
+srcdir = @srcdir@
+sysconfdir = @sysconfdir@
+target = @target@
+target_alias = @target_alias@
+target_cpu = @target_cpu@
+target_os = @target_os@
+target_vendor = @target_vendor@
+top_build_prefix = @top_build_prefix@
+top_builddir = @top_builddir@
+top_srcdir = @top_srcdir@
+AM_CPPFLAGS = \
+	-I$(top_srcdir)/src \
+	-I$(top_srcdir)/include \
+	-I$(top_srcdir)/src/gtest/include \
+	$(PTHREAD_CFLAGS) \
+	$(DEFINES)
+
+set_test_SOURCES = \
+	set_test.cpp
+
+set_test_LDADD = \
+	$(top_builddir)/src/gtest/libgtest.la \
+	$(top_builddir)/src/util/libmesautil.la \
+	$(PTHREAD_LIBS) \
+	$(DLOPEN_LIBS)
+
+EXTRA_DIST = meson.build
+all: all-am
+
+.SUFFIXES:
+.SUFFIXES: .cpp .lo .log .o .obj .test .test$(EXEEXT) .trs
+$(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
+	@for dep in $?; do \
+	  case '$(am__configure_deps)' in \
+	    *$$dep*) \
+	      ( cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh ) \
+	        && { if test -f $@; then exit 0; else break; fi; }; \
+	      exit 1;; \
+	  esac; \
+	done; \
+	echo ' cd $(top_srcdir) && $(AUTOMAKE) --foreign src/util/tests/set/Makefile'; \
+	$(am__cd) $(top_srcdir) && \
+	  $(AUTOMAKE) --foreign src/util/tests/set/Makefile
+Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
+	@case '$?' in \
+	  *config.status*) \
+	    cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \
+	  *) \
+	    echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__maybe_remake_depfiles)'; \
+	    cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__maybe_remake_depfiles);; \
+	esac;
+
+$(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+
+$(top_srcdir)/configure:  $(am__configure_deps)
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+$(ACLOCAL_M4):  $(am__aclocal_m4_deps)
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+$(am__aclocal_m4_deps):
+
+clean-checkPROGRAMS:
+	@list='$(check_PROGRAMS)'; test -n "$$list" || exit 0; \
+	echo " rm -f" $$list; \
+	rm -f $$list || exit $$?; \
+	test -n "$(EXEEXT)" || exit 0; \
+	list=`for p in $$list; do echo "$$p"; done | sed 's/$(EXEEXT)$$//'`; \
+	echo " rm -f" $$list; \
+	rm -f $$list
+
+set_test$(EXEEXT): $(set_test_OBJECTS) $(set_test_DEPENDENCIES) $(EXTRA_set_test_DEPENDENCIES) 
+	@rm -f set_test$(EXEEXT)
+	$(AM_V_CXXLD)$(CXXLINK) $(set_test_OBJECTS) $(set_test_LDADD) $(LIBS)
+
+mostlyclean-compile:
+	-rm -f *.$(OBJEXT)
+
+distclean-compile:
+	-rm -f *.tab.c
+
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/set_test.Po@am__quote@ # am--include-marker
+
+$(am__depfiles_remade):
+	@$(MKDIR_P) $(@D)
+	@echo '# dummy' >$@-t && $(am__mv) $@-t $@
+
+am--depfiles: $(am__depfiles_remade)
+
+.cpp.o:
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)depbase=`echo $@ | sed 's|[^/]*$$|$(DEPDIR)/&|;s|\.o$$||'`;\
+@am__fastdepCXX_TRUE@	$(CXXCOMPILE) -MT $@ -MD -MP -MF $$depbase.Tpo -c -o $@ $< &&\
+@am__fastdepCXX_TRUE@	$(am__mv) $$depbase.Tpo $$depbase.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
+
+.cpp.obj:
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)depbase=`echo $@ | sed 's|[^/]*$$|$(DEPDIR)/&|;s|\.obj$$||'`;\
+@am__fastdepCXX_TRUE@	$(CXXCOMPILE) -MT $@ -MD -MP -MF $$depbase.Tpo -c -o $@ `$(CYGPATH_W) '$<'` &&\
+@am__fastdepCXX_TRUE@	$(am__mv) $$depbase.Tpo $$depbase.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
+
+.cpp.lo:
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)depbase=`echo $@ | sed 's|[^/]*$$|$(DEPDIR)/&|;s|\.lo$$||'`;\
+@am__fastdepCXX_TRUE@	$(LTCXXCOMPILE) -MT $@ -MD -MP -MF $$depbase.Tpo -c -o $@ $< &&\
+@am__fastdepCXX_TRUE@	$(am__mv) $$depbase.Tpo $$depbase.Plo
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(LTCXXCOMPILE) -c -o $@ $<
+
+mostlyclean-libtool:
+	-rm -f *.lo
+
+clean-libtool:
+	-rm -rf .libs _libs
+
+ID: $(am__tagged_files)
+	$(am__define_uniq_tagged_files); mkid -fID $$unique
+tags: tags-am
+TAGS: tags
+
+tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+	set x; \
+	here=`pwd`; \
+	$(am__define_uniq_tagged_files); \
+	shift; \
+	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
+	  test -n "$$unique" || unique=$$empty_fix; \
+	  if test $$# -gt 0; then \
+	    $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
+	      "$$@" $$unique; \
+	  else \
+	    $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
+	      $$unique; \
+	  fi; \
+	fi
+ctags: ctags-am
+
+CTAGS: ctags
+ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+	$(am__define_uniq_tagged_files); \
+	test -z "$(CTAGS_ARGS)$$unique" \
+	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
+	     $$unique
+
+GTAGS:
+	here=`$(am__cd) $(top_builddir) && pwd` \
+	  && $(am__cd) $(top_srcdir) \
+	  && gtags -i $(GTAGS_ARGS) "$$here"
+cscopelist: cscopelist-am
+
+cscopelist-am: $(am__tagged_files)
+	list='$(am__tagged_files)'; \
+	case "$(srcdir)" in \
+	  [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
+	  *) sdir=$(subdir)/$(srcdir) ;; \
+	esac; \
+	for i in $$list; do \
+	  if test -f "$$i"; then \
+	    echo "$(subdir)/$$i"; \
+	  else \
+	    echo "$$sdir/$$i"; \
+	  fi; \
+	done >> $(top_builddir)/cscope.files
+
+distclean-tags:
+	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
+
+# Recover from deleted '.trs' file; this should ensure that
+# "rm -f foo.log; make foo.trs" re-run 'foo.test', and re-create
+# both 'foo.log' and 'foo.trs'.  Break the recipe in two subshells
+# to avoid problems with "make -n".
+.log.trs:
+	rm -f $< $@
+	$(MAKE) $(AM_MAKEFLAGS) $<
+
+# Leading 'am--fnord' is there to ensure the list of targets does not
+# expand to empty, as could happen e.g. with make check TESTS=''.
+am--fnord $(TEST_LOGS) $(TEST_LOGS:.log=.trs): $(am__force_recheck)
+am--force-recheck:
+	@:
+
+$(TEST_SUITE_LOG): $(TEST_LOGS)
+	@$(am__set_TESTS_bases); \
+	am__f_ok () { test -f "$$1" && test -r "$$1"; }; \
+	redo_bases=`for i in $$bases; do \
+	              am__f_ok $$i.trs && am__f_ok $$i.log || echo $$i; \
+	            done`; \
+	if test -n "$$redo_bases"; then \
+	  redo_logs=`for i in $$redo_bases; do echo $$i.log; done`; \
+	  redo_results=`for i in $$redo_bases; do echo $$i.trs; done`; \
+	  if $(am__make_dryrun); then :; else \
+	    rm -f $$redo_logs && rm -f $$redo_results || exit 1; \
+	  fi; \
+	fi; \
+	if test -n "$$am__remaking_logs"; then \
+	  echo "fatal: making $(TEST_SUITE_LOG): possible infinite" \
+	       "recursion detected" >&2; \
+	elif test -n "$$redo_logs"; then \
+	  am__remaking_logs=yes $(MAKE) $(AM_MAKEFLAGS) $$redo_logs; \
+	fi; \
+	if $(am__make_dryrun); then :; else \
+	  st=0;  \
+	  errmsg="fatal: making $(TEST_SUITE_LOG): failed to create"; \
+	  for i in $$redo_bases; do \
+	    test -f $$i.trs && test -r $$i.trs \
+	      || { echo "$$errmsg $$i.trs" >&2; st=1; }; \
+	    test -f $$i.log && test -r $$i.log \
+	      || { echo "$$errmsg $$i.log" >&2; st=1; }; \
+	  done; \
+	  test $$st -eq 0 || exit 1; \
+	fi
+	@$(am__sh_e_setup); $(am__tty_colors); $(am__set_TESTS_bases); \
+	ws='[ 	]'; \
+	results=`for b in $$bases; do echo $$b.trs; done`; \
+	test -n "$$results" || results=/dev/null; \
+	all=`  grep "^$$ws*:test-result:"           $$results | wc -l`; \
+	pass=` grep "^$$ws*:test-result:$$ws*PASS"  $$results | wc -l`; \
+	fail=` grep "^$$ws*:test-result:$$ws*FAIL"  $$results | wc -l`; \
+	skip=` grep "^$$ws*:test-result:$$ws*SKIP"  $$results | wc -l`; \
+	xfail=`grep "^$$ws*:test-result:$$ws*XFAIL" $$results | wc -l`; \
+	xpass=`grep "^$$ws*:test-result:$$ws*XPASS" $$results | wc -l`; \
+	error=`grep "^$$ws*:test-result:$$ws*ERROR" $$results | wc -l`; \
+	if test `expr $$fail + $$xpass + $$error` -eq 0; then \
+	  success=true; \
+	else \
+	  success=false; \
+	fi; \
+	br='==================='; br=$$br$$br$$br$$br; \
+	result_count () \
+	{ \
+	    if test x"$$1" = x"--maybe-color"; then \
+	      maybe_colorize=yes; \
+	    elif test x"$$1" = x"--no-color"; then \
+	      maybe_colorize=no; \
+	    else \
+	      echo "$@: invalid 'result_count' usage" >&2; exit 4; \
+	    fi; \
+	    shift; \
+	    desc=$$1 count=$$2; \
+	    if test $$maybe_colorize = yes && test $$count -gt 0; then \
+	      color_start=$$3 color_end=$$std; \
+	    else \
+	      color_start= color_end=; \
+	    fi; \
+	    echo "$${color_start}# $$desc $$count$${color_end}"; \
+	}; \
+	create_testsuite_report () \
+	{ \
+	  result_count $$1 "TOTAL:" $$all   "$$brg"; \
+	  result_count $$1 "PASS: " $$pass  "$$grn"; \
+	  result_count $$1 "SKIP: " $$skip  "$$blu"; \
+	  result_count $$1 "XFAIL:" $$xfail "$$lgn"; \
+	  result_count $$1 "FAIL: " $$fail  "$$red"; \
+	  result_count $$1 "XPASS:" $$xpass "$$red"; \
+	  result_count $$1 "ERROR:" $$error "$$mgn"; \
+	}; \
+	{								\
+	  echo "$(PACKAGE_STRING): $(subdir)/$(TEST_SUITE_LOG)" |	\
+	    $(am__rst_title);						\
+	  create_testsuite_report --no-color;				\
+	  echo;								\
+	  echo ".. contents:: :depth: 2";				\
+	  echo;								\
+	  for b in $$bases; do echo $$b; done				\
+	    | $(am__create_global_log);					\
+	} >$(TEST_SUITE_LOG).tmp || exit 1;				\
+	mv $(TEST_SUITE_LOG).tmp $(TEST_SUITE_LOG);			\
+	if $$success; then						\
+	  col="$$grn";							\
+	 else								\
+	  col="$$red";							\
+	  test x"$$VERBOSE" = x || cat $(TEST_SUITE_LOG);		\
+	fi;								\
+	echo "$${col}$$br$${std}"; 					\
+	echo "$${col}Testsuite summary for $(PACKAGE_STRING)$${std}";	\
+	echo "$${col}$$br$${std}"; 					\
+	create_testsuite_report --maybe-color;				\
+	echo "$$col$$br$$std";						\
+	if $$success; then :; else					\
+	  echo "$${col}See $(subdir)/$(TEST_SUITE_LOG)$${std}";		\
+	  if test -n "$(PACKAGE_BUGREPORT)"; then			\
+	    echo "$${col}Please report to $(PACKAGE_BUGREPORT)$${std}";	\
+	  fi;								\
+	  echo "$$col$$br$$std";					\
+	fi;								\
+	$$success || exit 1
+
+check-TESTS: $(check_PROGRAMS)
+	@list='$(RECHECK_LOGS)';           test -z "$$list" || rm -f $$list
+	@list='$(RECHECK_LOGS:.log=.trs)'; test -z "$$list" || rm -f $$list
+	@test -z "$(TEST_SUITE_LOG)" || rm -f $(TEST_SUITE_LOG)
+	@set +e; $(am__set_TESTS_bases); \
+	log_list=`for i in $$bases; do echo $$i.log; done`; \
+	trs_list=`for i in $$bases; do echo $$i.trs; done`; \
+	log_list=`echo $$log_list`; trs_list=`echo $$trs_list`; \
+	$(MAKE) $(AM_MAKEFLAGS) $(TEST_SUITE_LOG) TEST_LOGS="$$log_list"; \
+	exit $$?;
+recheck: all $(check_PROGRAMS)
+	@test -z "$(TEST_SUITE_LOG)" || rm -f $(TEST_SUITE_LOG)
+	@set +e; $(am__set_TESTS_bases); \
+	bases=`for i in $$bases; do echo $$i; done \
+	         | $(am__list_recheck_tests)` || exit 1; \
+	log_list=`for i in $$bases; do echo $$i.log; done`; \
+	log_list=`echo $$log_list`; \
+	$(MAKE) $(AM_MAKEFLAGS) $(TEST_SUITE_LOG) \
+	        am__force_recheck=am--force-recheck \
+	        TEST_LOGS="$$log_list"; \
+	exit $$?
+set_test.log: set_test$(EXEEXT)
+	@p='set_test$(EXEEXT)'; \
+	b='set_test'; \
+	$(am__check_pre) $(LOG_DRIVER) --test-name "$$f" \
+	--log-file $$b.log --trs-file $$b.trs \
+	$(am__common_driver_flags) $(AM_LOG_DRIVER_FLAGS) $(LOG_DRIVER_FLAGS) -- $(LOG_COMPILE) \
+	"$$tst" $(AM_TESTS_FD_REDIRECT)
+.test.log:
+	@p='$<'; \
+	$(am__set_b); \
+	$(am__check_pre) $(TEST_LOG_DRIVER) --test-name "$$f" \
+	--log-file $$b.log --trs-file $$b.trs \
+	$(am__common_driver_flags) $(AM_TEST_LOG_DRIVER_FLAGS) $(TEST_LOG_DRIVER_FLAGS) -- $(TEST_LOG_COMPILE) \
+	"$$tst" $(AM_TESTS_FD_REDIRECT)
+@am__EXEEXT_TRUE@.test$(EXEEXT).log:
+@am__EXEEXT_TRUE@	@p='$<'; \
+@am__EXEEXT_TRUE@	$(am__set_b); \
+@am__EXEEXT_TRUE@	$(am__check_pre) $(TEST_LOG_DRIVER) --test-name "$$f" \
+@am__EXEEXT_TRUE@	--log-file $$b.log --trs-file $$b.trs \
+@am__EXEEXT_TRUE@	$(am__common_driver_flags) $(AM_TEST_LOG_DRIVER_FLAGS) $(TEST_LOG_DRIVER_FLAGS) -- $(TEST_LOG_COMPILE) \
+@am__EXEEXT_TRUE@	"$$tst" $(AM_TESTS_FD_REDIRECT)
+
+distdir: $(BUILT_SOURCES)
+	$(MAKE) $(AM_MAKEFLAGS) distdir-am
+
+distdir-am: $(DISTFILES)
+	@srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
+	topsrcdirstrip=`echo "$(top_srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
+	list='$(DISTFILES)'; \
+	  dist_files=`for file in $$list; do echo $$file; done | \
+	  sed -e "s|^$$srcdirstrip/||;t" \
+	      -e "s|^$$topsrcdirstrip/|$(top_builddir)/|;t"`; \
+	case $$dist_files in \
+	  */*) $(MKDIR_P) `echo "$$dist_files" | \
+			   sed '/\//!d;s|^|$(distdir)/|;s,/[^/]*$$,,' | \
+			   sort -u` ;; \
+	esac; \
+	for file in $$dist_files; do \
+	  if test -f $$file || test -d $$file; then d=.; else d=$(srcdir); fi; \
+	  if test -d $$d/$$file; then \
+	    dir=`echo "/$$file" | sed -e 's,/[^/]*$$,,'`; \
+	    if test -d "$(distdir)/$$file"; then \
+	      find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
+	    fi; \
+	    if test -d $(srcdir)/$$file && test $$d != $(srcdir); then \
+	      cp -fpR $(srcdir)/$$file "$(distdir)$$dir" || exit 1; \
+	      find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
+	    fi; \
+	    cp -fpR $$d/$$file "$(distdir)$$dir" || exit 1; \
+	  else \
+	    test -f "$(distdir)/$$file" \
+	    || cp -p $$d/$$file "$(distdir)/$$file" \
+	    || exit 1; \
+	  fi; \
+	done
+check-am: all-am
+	$(MAKE) $(AM_MAKEFLAGS) $(check_PROGRAMS)
+	$(MAKE) $(AM_MAKEFLAGS) check-TESTS
+check: check-am
+all-am: Makefile
+installdirs:
+install: install-am
+install-exec: install-exec-am
+install-data: install-data-am
+uninstall: uninstall-am
+
+install-am: all-am
+	@$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am
+
+installcheck: installcheck-am
+install-strip:
+	if test -z '$(STRIP)'; then \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	      install; \
+	else \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	    "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
+	fi
+mostlyclean-generic:
+	-test -z "$(TEST_LOGS)" || rm -f $(TEST_LOGS)
+	-test -z "$(TEST_LOGS:.log=.trs)" || rm -f $(TEST_LOGS:.log=.trs)
+	-test -z "$(TEST_SUITE_LOG)" || rm -f $(TEST_SUITE_LOG)
+
+clean-generic:
+
+distclean-generic:
+	-test -z "$(CONFIG_CLEAN_FILES)" || rm -f $(CONFIG_CLEAN_FILES)
+	-test . = "$(srcdir)" || test -z "$(CONFIG_CLEAN_VPATH_FILES)" || rm -f $(CONFIG_CLEAN_VPATH_FILES)
+
+maintainer-clean-generic:
+	@echo "This command is intended for maintainers to use"
+	@echo "it deletes files that may require special tools to rebuild."
+clean: clean-am
+
+clean-am: clean-checkPROGRAMS clean-generic clean-libtool \
+	mostlyclean-am
+
+distclean: distclean-am
+		-rm -f ./$(DEPDIR)/set_test.Po
+	-rm -f Makefile
+distclean-am: clean-am distclean-compile distclean-generic \
+	distclean-tags
+
+dvi: dvi-am
+
+dvi-am:
+
+html: html-am
+
+html-am:
+
+info: info-am
+
+info-am:
+
+install-data-am:
+
+install-dvi: install-dvi-am
+
+install-dvi-am:
+
+install-exec-am:
+
+install-html: install-html-am
+
+install-html-am:
+
+install-info: install-info-am
+
+install-info-am:
+
+install-man:
+
+install-pdf: install-pdf-am
+
+install-pdf-am:
+
+install-ps: install-ps-am
+
+install-ps-am:
+
+installcheck-am:
+
+maintainer-clean: maintainer-clean-am
+		-rm -f ./$(DEPDIR)/set_test.Po
+	-rm -f Makefile
+maintainer-clean-am: distclean-am maintainer-clean-generic
+
+mostlyclean: mostlyclean-am
+
+mostlyclean-am: mostlyclean-compile mostlyclean-generic \
+	mostlyclean-libtool
+
+pdf: pdf-am
+
+pdf-am:
+
+ps: ps-am
+
+ps-am:
+
+uninstall-am:
+
+.MAKE: check-am install-am install-strip
+
+.PHONY: CTAGS GTAGS TAGS all all-am am--depfiles check check-TESTS \
+	check-am clean clean-checkPROGRAMS clean-generic clean-libtool \
+	cscopelist-am ctags ctags-am distclean distclean-compile \
+	distclean-generic distclean-libtool distclean-tags distdir dvi \
+	dvi-am html html-am info info-am install install-am \
+	install-data install-data-am install-dvi install-dvi-am \
+	install-exec install-exec-am install-html install-html-am \
+	install-info install-info-am install-man install-pdf \
+	install-pdf-am install-ps install-ps-am install-strip \
+	installcheck installcheck-am installdirs maintainer-clean \
+	maintainer-clean-generic mostlyclean mostlyclean-compile \
+	mostlyclean-generic mostlyclean-libtool pdf pdf-am ps ps-am \
+	recheck tags tags-am uninstall uninstall-am
+
+.PRECIOUS: Makefile
+
+
+# Tell versions [3.59,3.63) of GNU make to not export all variables.
+# Otherwise a system limit (for SysV at least) may be exceeded.
+.NOEXPORT:
diff --git a/lib/mesa/src/util/tests/set/meson.build b/lib/mesa/src/util/tests/set/meson.build
new file mode 100644
index 000000000..add3fc560
--- /dev/null
+++ b/lib/mesa/src/util/tests/set/meson.build
@@ -0,0 +1,30 @@
+# Copyright © 2018 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+test(
+  'set',
+  executable(
+    'set_test',
+    'set_test.cpp',
+    dependencies : [dep_thread, dep_dl, idep_gtest],
+    include_directories : inc_common,
+    link_with : [libmesa_util],
+  )
+)
diff --git a/lib/mesa/src/util/tests/set/set_test.cpp b/lib/mesa/src/util/tests/set/set_test.cpp
new file mode 100644
index 000000000..a1eef0b3d
--- /dev/null
+++ b/lib/mesa/src/util/tests/set/set_test.cpp
@@ -0,0 +1,115 @@
+/*
+ * Copyright © 2018 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <gtest/gtest.h>
+#include "util/hash_table.h"
+#include "util/set.h"
+
+TEST(set, basic)
+{
+   struct set *s = _mesa_set_create(NULL, _mesa_hash_pointer,
+                                    _mesa_key_pointer_equal);
+   struct set_entry *entry;
+
+   const void *a = (const void *)10;
+   const void *b = (const void *)20;
+
+   _mesa_set_add(s, a);
+   _mesa_set_add(s, b);
+   EXPECT_EQ(s->entries, 2);
+
+   _mesa_set_add(s, a);
+   EXPECT_EQ(s->entries, 2);
+
+   entry = _mesa_set_search(s, a);
+   EXPECT_TRUE(entry);
+   EXPECT_EQ(entry->key, a);
+
+   _mesa_set_remove(s, entry);
+   EXPECT_EQ(s->entries, 1);
+
+   entry = _mesa_set_search(s, a);
+   EXPECT_FALSE(entry);
+
+   _mesa_set_destroy(s, NULL);
+}
+
+TEST(set, clone)
+{
+   struct set *s = _mesa_set_create(NULL, _mesa_hash_pointer,
+                                    _mesa_key_pointer_equal);
+   struct set_entry *entry;
+
+   const void *a = (const void *)10;
+   const void *b = (const void *)20;
+   const void *c = (const void *)30;
+
+   _mesa_set_add(s, a);
+   _mesa_set_add(s, b);
+   _mesa_set_add(s, c);
+
+   entry = _mesa_set_search(s, c);
+   EXPECT_TRUE(entry);
+   EXPECT_EQ(entry->key, c);
+
+   _mesa_set_remove(s, entry);
+   EXPECT_EQ(s->entries, 2);
+
+   struct set *clone = _mesa_set_clone(s, NULL);
+   EXPECT_EQ(clone->entries, 2);
+
+   EXPECT_TRUE(_mesa_set_search(clone, a));
+   EXPECT_TRUE(_mesa_set_search(clone, b));
+   EXPECT_FALSE(_mesa_set_search(clone, c));
+
+   _mesa_set_destroy(s, NULL);
+   _mesa_set_destroy(clone, NULL);
+}
+
+TEST(set, remove_key)
+{
+   struct set *s = _mesa_set_create(NULL, _mesa_hash_pointer,
+                                    _mesa_key_pointer_equal);
+
+   const void *a = (const void *)10;
+   const void *b = (const void *)20;
+   const void *c = (const void *)30;
+
+   _mesa_set_add(s, a);
+   _mesa_set_add(s, b);
+   EXPECT_EQ(s->entries, 2);
+
+   /* Remove existing key. */
+   _mesa_set_remove_key(s, a);
+   EXPECT_EQ(s->entries, 1);
+   EXPECT_FALSE(_mesa_set_search(s, a));
+   EXPECT_TRUE(_mesa_set_search(s, b));
+
+   /* Remove non-existing key. */
+   _mesa_set_remove_key(s, c);
+   EXPECT_EQ(s->entries, 1);
+   EXPECT_FALSE(_mesa_set_search(s, a));
+   EXPECT_TRUE(_mesa_set_search(s, b));
+
+   _mesa_set_destroy(s, NULL);
+}
diff --git a/lib/mesa/src/util/tests/string_buffer/Makefile.am b/lib/mesa/src/util/tests/string_buffer/Makefile.am
index bd04d8634..8bc8a9c9b 100644
--- a/lib/mesa/src/util/tests/string_buffer/Makefile.am
+++ b/lib/mesa/src/util/tests/string_buffer/Makefile.am
@@ -38,3 +38,5 @@ string_buffer_test_LDADD = \
 	$(top_builddir)/src/util/libmesautil.la \
 	$(PTHREAD_LIBS) \
 	$(DLOPEN_LIBS)
+
+EXTRA_DIST = meson.build
diff --git a/lib/mesa/src/util/tests/string_buffer/meson.build b/lib/mesa/src/util/tests/string_buffer/meson.build
new file mode 100644
index 000000000..9f42e3550
--- /dev/null
+++ b/lib/mesa/src/util/tests/string_buffer/meson.build
@@ -0,0 +1,30 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+test(
+  'string_buffer',
+  executable(
+    'string_buffer_test',
+    'string_buffer_test.cpp',
+    dependencies : [dep_thread, dep_dl, idep_gtest],
+    include_directories : inc_common,
+    link_with : [libmesa_util],
+  )
+)
diff --git a/lib/mesa/src/util/tests/string_buffer/string_buffer_test.cpp b/lib/mesa/src/util/tests/string_buffer/string_buffer_test.cpp
index 545f607fa..afb6dfb2a 100644
--- a/lib/mesa/src/util/tests/string_buffer/string_buffer_test.cpp
+++ b/lib/mesa/src/util/tests/string_buffer/string_buffer_test.cpp
@@ -95,15 +95,15 @@ TEST_F(string_buffer, string_buffer_tests)
    EXPECT_TRUE(strlen(buf->buf) == 0);
 
    /* Test a string with some formatting */
-   sprintf(str4, "Testing formatting %d, %f", 100, 1.0);
+   snprintf(str4, sizeof(str4), "Testing formatting %d, %f", 100, 1.0);
    EXPECT_TRUE(_mesa_string_buffer_printf(buf, "Testing formatting %d, %f", 100, 1.0));
    EXPECT_TRUE(strcmp(buf->buf, str4) == 0);
 
    /* Compile a string with some other formatting */
-   sprintf(str5, "Testing formatting %d, %x", 100, 0xDEADBEAF);
+   snprintf(str5, sizeof(str5), "Testing formatting %d, %x", 100, 0xDEADBEAF);
 
    /* Concatenate str5 to str4 */
-   strcat(str4, str5);
+   strncat(str4, str5, sizeof(str5));
 
    /* Now use the formatted append function again */
    EXPECT_TRUE(_mesa_string_buffer_printf(buf, "Testing formatting %d, %x", 100, 0xDEADBEAF));
diff --git a/lib/mesa/src/util/tests/vma/Makefile.am b/lib/mesa/src/util/tests/vma/Makefile.am
new file mode 100644
index 000000000..b9ca8f597
--- /dev/null
+++ b/lib/mesa/src/util/tests/vma/Makefile.am
@@ -0,0 +1,39 @@
+# Copyright © 2018 Intel Corporation
+#
+#  Permission is hereby granted, free of charge, to any person obtaining a
+#  copy of this software and associated documentation files (the "Software"),
+#  to deal in the Software without restriction, including without limitation
+#  the rights to use, copy, modify, merge, publish, distribute, sublicense,
+#  and/or sell copies of the Software, and to permit persons to whom the
+#  Software is furnished to do so, subject to the following conditions:
+#
+#  The above copyright notice and this permission notice (including the next
+#  paragraph) shall be included in all copies or substantial portions of the
+#  Software.
+#
+#  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+#  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+#  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+#  THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+#  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+#  FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+#  IN THE SOFTWARE.
+
+AM_CPPFLAGS = \
+	-I$(top_srcdir)/include \
+	-I$(top_srcdir)/src/util \
+	$(DEFINES)
+
+TESTS = vma_random_test
+
+check_PROGRAMS = $(TESTS)
+
+vma_random_test_SOURCES = \
+	vma_random_test.cpp
+
+vma_random_test_LDADD = \
+	$(top_builddir)/src/util/libmesautil.la
+
+vma_random_test_CXXFLAGS = $(CXX11_CXXFLAGS)
+
+EXTRA_DIST = meson.build
diff --git a/lib/mesa/src/util/tests/vma/Makefile.in b/lib/mesa/src/util/tests/vma/Makefile.in
new file mode 100644
index 000000000..92afaa356
--- /dev/null
+++ b/lib/mesa/src/util/tests/vma/Makefile.in
@@ -0,0 +1,1233 @@
+# Makefile.in generated by automake 1.16.1 from Makefile.am.
+# @configure_input@
+
+# Copyright (C) 1994-2018 Free Software Foundation, Inc.
+
+# This Makefile.in is free software; the Free Software Foundation
+# gives unlimited permission to copy and/or distribute it,
+# with or without modifications, as long as this notice is preserved.
+
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY, to the extent permitted by law; without
+# even the implied warranty of MERCHANTABILITY or FITNESS FOR A
+# PARTICULAR PURPOSE.
+
+@SET_MAKE@
+
+# Copyright © 2018 Intel Corporation
+#
+#  Permission is hereby granted, free of charge, to any person obtaining a
+#  copy of this software and associated documentation files (the "Software"),
+#  to deal in the Software without restriction, including without limitation
+#  the rights to use, copy, modify, merge, publish, distribute, sublicense,
+#  and/or sell copies of the Software, and to permit persons to whom the
+#  Software is furnished to do so, subject to the following conditions:
+#
+#  The above copyright notice and this permission notice (including the next
+#  paragraph) shall be included in all copies or substantial portions of the
+#  Software.
+#
+#  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+#  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+#  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+#  THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+#  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+#  FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+#  IN THE SOFTWARE.
+VPATH = @srcdir@
+am__is_gnu_make = { \
+  if test -z '$(MAKELEVEL)'; then \
+    false; \
+  elif test -n '$(MAKE_HOST)'; then \
+    true; \
+  elif test -n '$(MAKE_VERSION)' && test -n '$(CURDIR)'; then \
+    true; \
+  else \
+    false; \
+  fi; \
+}
+am__make_running_with_option = \
+  case $${target_option-} in \
+      ?) ;; \
+      *) echo "am__make_running_with_option: internal error: invalid" \
+              "target option '$${target_option-}' specified" >&2; \
+         exit 1;; \
+  esac; \
+  has_opt=no; \
+  sane_makeflags=$$MAKEFLAGS; \
+  if $(am__is_gnu_make); then \
+    sane_makeflags=$$MFLAGS; \
+  else \
+    case $$MAKEFLAGS in \
+      *\\[\ \	]*) \
+        bs=\\; \
+        sane_makeflags=`printf '%s\n' "$$MAKEFLAGS" \
+          | sed "s/$$bs$$bs[$$bs $$bs	]*//g"`;; \
+    esac; \
+  fi; \
+  skip_next=no; \
+  strip_trailopt () \
+  { \
+    flg=`printf '%s\n' "$$flg" | sed "s/$$1.*$$//"`; \
+  }; \
+  for flg in $$sane_makeflags; do \
+    test $$skip_next = yes && { skip_next=no; continue; }; \
+    case $$flg in \
+      *=*|--*) continue;; \
+        -*I) strip_trailopt 'I'; skip_next=yes;; \
+      -*I?*) strip_trailopt 'I';; \
+        -*O) strip_trailopt 'O'; skip_next=yes;; \
+      -*O?*) strip_trailopt 'O';; \
+        -*l) strip_trailopt 'l'; skip_next=yes;; \
+      -*l?*) strip_trailopt 'l';; \
+      -[dEDm]) skip_next=yes;; \
+      -[JT]) skip_next=yes;; \
+    esac; \
+    case $$flg in \
+      *$$target_option*) has_opt=yes; break;; \
+    esac; \
+  done; \
+  test $$has_opt = yes
+am__make_dryrun = (target_option=n; $(am__make_running_with_option))
+am__make_keepgoing = (target_option=k; $(am__make_running_with_option))
+pkgdatadir = $(datadir)/@PACKAGE@
+pkgincludedir = $(includedir)/@PACKAGE@
+pkglibdir = $(libdir)/@PACKAGE@
+pkglibexecdir = $(libexecdir)/@PACKAGE@
+am__cd = CDPATH="$${ZSH_VERSION+.}$(PATH_SEPARATOR)" && cd
+install_sh_DATA = $(install_sh) -c -m 644
+install_sh_PROGRAM = $(install_sh) -c
+install_sh_SCRIPT = $(install_sh) -c
+INSTALL_HEADER = $(INSTALL_DATA)
+transform = $(program_transform_name)
+NORMAL_INSTALL = :
+PRE_INSTALL = :
+POST_INSTALL = :
+NORMAL_UNINSTALL = :
+PRE_UNINSTALL = :
+POST_UNINSTALL = :
+build_triplet = @build@
+host_triplet = @host@
+target_triplet = @target@
+TESTS = vma_random_test$(EXEEXT)
+check_PROGRAMS = $(am__EXEEXT_1)
+subdir = src/util/tests/vma
+ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
+am__aclocal_m4_deps = $(top_srcdir)/m4/ax_check_compile_flag.m4 \
+	$(top_srcdir)/m4/ax_check_gnu_make.m4 \
+	$(top_srcdir)/m4/ax_check_python_mako_module.m4 \
+	$(top_srcdir)/m4/ax_gcc_builtin.m4 \
+	$(top_srcdir)/m4/ax_gcc_func_attribute.m4 \
+	$(top_srcdir)/m4/ax_prog_bison.m4 \
+	$(top_srcdir)/m4/ax_prog_flex.m4 \
+	$(top_srcdir)/m4/ax_pthread.m4 $(top_srcdir)/m4/libtool.m4 \
+	$(top_srcdir)/m4/ltoptions.m4 $(top_srcdir)/m4/ltsugar.m4 \
+	$(top_srcdir)/m4/ltversion.m4 $(top_srcdir)/m4/lt~obsolete.m4 \
+	$(top_srcdir)/VERSION $(top_srcdir)/configure.ac
+am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
+	$(ACLOCAL_M4)
+DIST_COMMON = $(srcdir)/Makefile.am $(am__DIST_COMMON)
+mkinstalldirs = $(install_sh) -d
+CONFIG_CLEAN_FILES =
+CONFIG_CLEAN_VPATH_FILES =
+am__EXEEXT_1 = vma_random_test$(EXEEXT)
+am_vma_random_test_OBJECTS =  \
+	vma_random_test-vma_random_test.$(OBJEXT)
+vma_random_test_OBJECTS = $(am_vma_random_test_OBJECTS)
+vma_random_test_DEPENDENCIES =  \
+	$(top_builddir)/src/util/libmesautil.la
+AM_V_lt = $(am__v_lt_@AM_V@)
+am__v_lt_ = $(am__v_lt_@AM_DEFAULT_V@)
+am__v_lt_0 = --silent
+am__v_lt_1 = 
+vma_random_test_LINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX \
+	$(AM_LIBTOOLFLAGS) $(LIBTOOLFLAGS) --mode=link $(CXXLD) \
+	$(vma_random_test_CXXFLAGS) $(CXXFLAGS) $(AM_LDFLAGS) \
+	$(LDFLAGS) -o $@
+AM_V_P = $(am__v_P_@AM_V@)
+am__v_P_ = $(am__v_P_@AM_DEFAULT_V@)
+am__v_P_0 = false
+am__v_P_1 = :
+AM_V_GEN = $(am__v_GEN_@AM_V@)
+am__v_GEN_ = $(am__v_GEN_@AM_DEFAULT_V@)
+am__v_GEN_0 = @echo "  GEN     " $@;
+am__v_GEN_1 = 
+AM_V_at = $(am__v_at_@AM_V@)
+am__v_at_ = $(am__v_at_@AM_DEFAULT_V@)
+am__v_at_0 = @
+am__v_at_1 = 
+DEFAULT_INCLUDES = -I.@am__isrc@
+depcomp = $(SHELL) $(top_srcdir)/bin/depcomp
+am__maybe_remake_depfiles = depfiles
+am__depfiles_remade = ./$(DEPDIR)/vma_random_test-vma_random_test.Po
+am__mv = mv -f
+CXXCOMPILE = $(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) \
+	$(AM_CPPFLAGS) $(CPPFLAGS) $(AM_CXXFLAGS) $(CXXFLAGS)
+LTCXXCOMPILE = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
+	$(LIBTOOLFLAGS) --mode=compile $(CXX) $(DEFS) \
+	$(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) \
+	$(AM_CXXFLAGS) $(CXXFLAGS)
+AM_V_CXX = $(am__v_CXX_@AM_V@)
+am__v_CXX_ = $(am__v_CXX_@AM_DEFAULT_V@)
+am__v_CXX_0 = @echo "  CXX     " $@;
+am__v_CXX_1 = 
+CXXLD = $(CXX)
+CXXLINK = $(LIBTOOL) $(AM_V_lt) --tag=CXX $(AM_LIBTOOLFLAGS) \
+	$(LIBTOOLFLAGS) --mode=link $(CXXLD) $(AM_CXXFLAGS) \
+	$(CXXFLAGS) $(AM_LDFLAGS) $(LDFLAGS) -o $@
+AM_V_CXXLD = $(am__v_CXXLD_@AM_V@)
+am__v_CXXLD_ = $(am__v_CXXLD_@AM_DEFAULT_V@)
+am__v_CXXLD_0 = @echo "  CXXLD   " $@;
+am__v_CXXLD_1 = 
+SOURCES = $(vma_random_test_SOURCES)
+DIST_SOURCES = $(vma_random_test_SOURCES)
+am__can_run_installinfo = \
+  case $$AM_UPDATE_INFO_DIR in \
+    n|no|NO) false;; \
+    *) (install-info --version) >/dev/null 2>&1;; \
+  esac
+am__tagged_files = $(HEADERS) $(SOURCES) $(TAGS_FILES) $(LISP)
+# Read a list of newline-separated strings from the standard input,
+# and print each of them once, without duplicates.  Input order is
+# *not* preserved.
+am__uniquify_input = $(AWK) '\
+  BEGIN { nonempty = 0; } \
+  { items[$$0] = 1; nonempty = 1; } \
+  END { if (nonempty) { for (i in items) print i; }; } \
+'
+# Make sure the list of sources is unique.  This is necessary because,
+# e.g., the same source file might be shared among _SOURCES variables
+# for different programs/libraries.
+am__define_uniq_tagged_files = \
+  list='$(am__tagged_files)'; \
+  unique=`for i in $$list; do \
+    if test -f "$$i"; then echo $$i; else echo $(srcdir)/$$i; fi; \
+  done | $(am__uniquify_input)`
+ETAGS = etags
+CTAGS = ctags
+am__tty_colors_dummy = \
+  mgn= red= grn= lgn= blu= brg= std=; \
+  am__color_tests=no
+am__tty_colors = { \
+  $(am__tty_colors_dummy); \
+  if test "X$(AM_COLOR_TESTS)" = Xno; then \
+    am__color_tests=no; \
+  elif test "X$(AM_COLOR_TESTS)" = Xalways; then \
+    am__color_tests=yes; \
+  elif test "X$$TERM" != Xdumb && { test -t 1; } 2>/dev/null; then \
+    am__color_tests=yes; \
+  fi; \
+  if test $$am__color_tests = yes; then \
+    red=''; \
+    grn=''; \
+    lgn=''; \
+    blu=''; \
+    mgn=''; \
+    brg=''; \
+    std=''; \
+  fi; \
+}
+am__vpath_adj_setup = srcdirstrip=`echo "$(srcdir)" | sed 's|.|.|g'`;
+am__vpath_adj = case $$p in \
+    $(srcdir)/*) f=`echo "$$p" | sed "s|^$$srcdirstrip/||"`;; \
+    *) f=$$p;; \
+  esac;
+am__strip_dir = f=`echo $$p | sed -e 's|^.*/||'`;
+am__install_max = 40
+am__nobase_strip_setup = \
+  srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*|]/\\\\&/g'`
+am__nobase_strip = \
+  for p in $$list; do echo "$$p"; done | sed -e "s|$$srcdirstrip/||"
+am__nobase_list = $(am__nobase_strip_setup); \
+  for p in $$list; do echo "$$p $$p"; done | \
+  sed "s| $$srcdirstrip/| |;"' / .*\//!s/ .*/ ./; s,\( .*\)/[^/]*$$,\1,' | \
+  $(AWK) 'BEGIN { files["."] = "" } { files[$$2] = files[$$2] " " $$1; \
+    if (++n[$$2] == $(am__install_max)) \
+      { print $$2, files[$$2]; n[$$2] = 0; files[$$2] = "" } } \
+    END { for (dir in files) print dir, files[dir] }'
+am__base_list = \
+  sed '$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;$$!N;s/\n/ /g' | \
+  sed '$$!N;$$!N;$$!N;$$!N;s/\n/ /g'
+am__uninstall_files_from_dir = { \
+  test -z "$$files" \
+    || { test ! -d "$$dir" && test ! -f "$$dir" && test ! -r "$$dir"; } \
+    || { echo " ( cd '$$dir' && rm -f" $$files ")"; \
+         $(am__cd) "$$dir" && rm -f $$files; }; \
+  }
+am__recheck_rx = ^[ 	]*:recheck:[ 	]*
+am__global_test_result_rx = ^[ 	]*:global-test-result:[ 	]*
+am__copy_in_global_log_rx = ^[ 	]*:copy-in-global-log:[ 	]*
+# A command that, given a newline-separated list of test names on the
+# standard input, print the name of the tests that are to be re-run
+# upon "make recheck".
+am__list_recheck_tests = $(AWK) '{ \
+  recheck = 1; \
+  while ((rc = (getline line < ($$0 ".trs"))) != 0) \
+    { \
+      if (rc < 0) \
+        { \
+          if ((getline line2 < ($$0 ".log")) < 0) \
+	    recheck = 0; \
+          break; \
+        } \
+      else if (line ~ /$(am__recheck_rx)[nN][Oo]/) \
+        { \
+          recheck = 0; \
+          break; \
+        } \
+      else if (line ~ /$(am__recheck_rx)[yY][eE][sS]/) \
+        { \
+          break; \
+        } \
+    }; \
+  if (recheck) \
+    print $$0; \
+  close ($$0 ".trs"); \
+  close ($$0 ".log"); \
+}'
+# A command that, given a newline-separated list of test names on the
+# standard input, create the global log from their .trs and .log files.
+am__create_global_log = $(AWK) ' \
+function fatal(msg) \
+{ \
+  print "fatal: making $@: " msg | "cat >&2"; \
+  exit 1; \
+} \
+function rst_section(header) \
+{ \
+  print header; \
+  len = length(header); \
+  for (i = 1; i <= len; i = i + 1) \
+    printf "="; \
+  printf "\n\n"; \
+} \
+{ \
+  copy_in_global_log = 1; \
+  global_test_result = "RUN"; \
+  while ((rc = (getline line < ($$0 ".trs"))) != 0) \
+    { \
+      if (rc < 0) \
+         fatal("failed to read from " $$0 ".trs"); \
+      if (line ~ /$(am__global_test_result_rx)/) \
+        { \
+          sub("$(am__global_test_result_rx)", "", line); \
+          sub("[ 	]*$$", "", line); \
+          global_test_result = line; \
+        } \
+      else if (line ~ /$(am__copy_in_global_log_rx)[nN][oO]/) \
+        copy_in_global_log = 0; \
+    }; \
+  if (copy_in_global_log) \
+    { \
+      rst_section(global_test_result ": " $$0); \
+      while ((rc = (getline line < ($$0 ".log"))) != 0) \
+      { \
+        if (rc < 0) \
+          fatal("failed to read from " $$0 ".log"); \
+        print line; \
+      }; \
+      printf "\n"; \
+    }; \
+  close ($$0 ".trs"); \
+  close ($$0 ".log"); \
+}'
+# Restructured Text title.
+am__rst_title = { sed 's/.*/   &   /;h;s/./=/g;p;x;s/ *$$//;p;g' && echo; }
+# Solaris 10 'make', and several other traditional 'make' implementations,
+# pass "-e" to $(SHELL), and POSIX 2008 even requires this.  Work around it
+# by disabling -e (using the XSI extension "set +e") if it's set.
+am__sh_e_setup = case $$- in *e*) set +e;; esac
+# Default flags passed to test drivers.
+am__common_driver_flags = \
+  --color-tests "$$am__color_tests" \
+  --enable-hard-errors "$$am__enable_hard_errors" \
+  --expect-failure "$$am__expect_failure"
+# To be inserted before the command running the test.  Creates the
+# directory for the log if needed.  Stores in $dir the directory
+# containing $f, in $tst the test, in $log the log.  Executes the
+# developer- defined test setup AM_TESTS_ENVIRONMENT (if any), and
+# passes TESTS_ENVIRONMENT.  Set up options for the wrapper that
+# will run the test scripts (or their associated LOG_COMPILER, if
+# thy have one).
+am__check_pre = \
+$(am__sh_e_setup);					\
+$(am__vpath_adj_setup) $(am__vpath_adj)			\
+$(am__tty_colors);					\
+srcdir=$(srcdir); export srcdir;			\
+case "$@" in						\
+  */*) am__odir=`echo "./$@" | sed 's|/[^/]*$$||'`;;	\
+    *) am__odir=.;; 					\
+esac;							\
+test "x$$am__odir" = x"." || test -d "$$am__odir" 	\
+  || $(MKDIR_P) "$$am__odir" || exit $$?;		\
+if test -f "./$$f"; then dir=./;			\
+elif test -f "$$f"; then dir=;				\
+else dir="$(srcdir)/"; fi;				\
+tst=$$dir$$f; log='$@'; 				\
+if test -n '$(DISABLE_HARD_ERRORS)'; then		\
+  am__enable_hard_errors=no; 				\
+else							\
+  am__enable_hard_errors=yes; 				\
+fi; 							\
+case " $(XFAIL_TESTS) " in				\
+  *[\ \	]$$f[\ \	]* | *[\ \	]$$dir$$f[\ \	]*) \
+    am__expect_failure=yes;;				\
+  *)							\
+    am__expect_failure=no;;				\
+esac; 							\
+$(AM_TESTS_ENVIRONMENT) $(TESTS_ENVIRONMENT)
+# A shell command to get the names of the tests scripts with any registered
+# extension removed (i.e., equivalently, the names of the test logs, with
+# the '.log' extension removed).  The result is saved in the shell variable
+# '$bases'.  This honors runtime overriding of TESTS and TEST_LOGS.  Sadly,
+# we cannot use something simpler, involving e.g., "$(TEST_LOGS:.log=)",
+# since that might cause problem with VPATH rewrites for suffix-less tests.
+# See also 'test-harness-vpath-rewrite.sh' and 'test-trs-basic.sh'.
+am__set_TESTS_bases = \
+  bases='$(TEST_LOGS)'; \
+  bases=`for i in $$bases; do echo $$i; done | sed 's/\.log$$//'`; \
+  bases=`echo $$bases`
+RECHECK_LOGS = $(TEST_LOGS)
+AM_RECURSIVE_TARGETS = check recheck
+TEST_SUITE_LOG = test-suite.log
+TEST_EXTENSIONS = @EXEEXT@ .test
+LOG_DRIVER = $(SHELL) $(top_srcdir)/bin/test-driver
+LOG_COMPILE = $(LOG_COMPILER) $(AM_LOG_FLAGS) $(LOG_FLAGS)
+am__set_b = \
+  case '$@' in \
+    */*) \
+      case '$*' in \
+        */*) b='$*';; \
+          *) b=`echo '$@' | sed 's/\.log$$//'`; \
+       esac;; \
+    *) \
+      b='$*';; \
+  esac
+am__test_logs1 = $(TESTS:=.log)
+am__test_logs2 = $(am__test_logs1:@EXEEXT@.log=.log)
+TEST_LOGS = $(am__test_logs2:.test.log=.log)
+TEST_LOG_DRIVER = $(SHELL) $(top_srcdir)/bin/test-driver
+TEST_LOG_COMPILE = $(TEST_LOG_COMPILER) $(AM_TEST_LOG_FLAGS) \
+	$(TEST_LOG_FLAGS)
+am__DIST_COMMON = $(srcdir)/Makefile.in $(top_srcdir)/bin/depcomp \
+	$(top_srcdir)/bin/test-driver
+DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
+ACLOCAL = @ACLOCAL@
+AMDGPU_CFLAGS = @AMDGPU_CFLAGS@
+AMDGPU_LIBS = @AMDGPU_LIBS@
+AMTAR = @AMTAR@
+AM_DEFAULT_VERBOSITY = @AM_DEFAULT_VERBOSITY@
+ANDROID_CFLAGS = @ANDROID_CFLAGS@
+ANDROID_LIBS = @ANDROID_LIBS@
+AR = @AR@
+AUTOCONF = @AUTOCONF@
+AUTOHEADER = @AUTOHEADER@
+AUTOMAKE = @AUTOMAKE@
+AWK = @AWK@
+BACKTRACE_CFLAGS = @BACKTRACE_CFLAGS@
+BACKTRACE_LIBS = @BACKTRACE_LIBS@
+BSYMBOLIC = @BSYMBOLIC@
+CC = @CC@
+CCAS = @CCAS@
+CCASDEPMODE = @CCASDEPMODE@
+CCASFLAGS = @CCASFLAGS@
+CCDEPMODE = @CCDEPMODE@
+CFLAGS = @CFLAGS@
+CLANG_RESOURCE_DIR = @CLANG_RESOURCE_DIR@
+CLOCK_LIB = @CLOCK_LIB@
+CLOVER_STD_OVERRIDE = @CLOVER_STD_OVERRIDE@
+CPP = @CPP@
+CPPFLAGS = @CPPFLAGS@
+CXX = @CXX@
+CXX11_CXXFLAGS = @CXX11_CXXFLAGS@
+CXXCPP = @CXXCPP@
+CXXDEPMODE = @CXXDEPMODE@
+CXXFLAGS = @CXXFLAGS@
+CYGPATH_W = @CYGPATH_W@
+D3D_DRIVER_INSTALL_DIR = @D3D_DRIVER_INSTALL_DIR@
+DEFINES = @DEFINES@
+DEFS = @DEFS@
+DEPDIR = @DEPDIR@
+DLLTOOL = @DLLTOOL@
+DLOPEN_LIBS = @DLOPEN_LIBS@
+DRI2PROTO_CFLAGS = @DRI2PROTO_CFLAGS@
+DRI2PROTO_LIBS = @DRI2PROTO_LIBS@
+DRIGL_CFLAGS = @DRIGL_CFLAGS@
+DRIGL_LIBS = @DRIGL_LIBS@
+DRI_DRIVER_INSTALL_DIR = @DRI_DRIVER_INSTALL_DIR@
+DRI_DRIVER_SEARCH_DIR = @DRI_DRIVER_SEARCH_DIR@
+DRI_LIB_DEPS = @DRI_LIB_DEPS@
+DRI_PC_REQ_PRIV = @DRI_PC_REQ_PRIV@
+DSYMUTIL = @DSYMUTIL@
+DUMPBIN = @DUMPBIN@
+ECHO_C = @ECHO_C@
+ECHO_N = @ECHO_N@
+ECHO_T = @ECHO_T@
+EGL_CFLAGS = @EGL_CFLAGS@
+EGL_LIB_DEPS = @EGL_LIB_DEPS@
+EGL_NATIVE_PLATFORM = @EGL_NATIVE_PLATFORM@
+EGREP = @EGREP@
+ETNAVIV_CFLAGS = @ETNAVIV_CFLAGS@
+ETNAVIV_LIBS = @ETNAVIV_LIBS@
+EXEEXT = @EXEEXT@
+EXPAT_CFLAGS = @EXPAT_CFLAGS@
+EXPAT_LIBS = @EXPAT_LIBS@
+FGREP = @FGREP@
+GALLIUM_PIPE_LOADER_DEFINES = @GALLIUM_PIPE_LOADER_DEFINES@
+GBM_PC_LIB_PRIV = @GBM_PC_LIB_PRIV@
+GBM_PC_REQ_PRIV = @GBM_PC_REQ_PRIV@
+GC_SECTIONS = @GC_SECTIONS@
+GLESv1_CM_LIB_DEPS = @GLESv1_CM_LIB_DEPS@
+GLESv1_CM_PC_LIB_PRIV = @GLESv1_CM_PC_LIB_PRIV@
+GLESv2_LIB_DEPS = @GLESv2_LIB_DEPS@
+GLESv2_PC_LIB_PRIV = @GLESv2_PC_LIB_PRIV@
+GLPROTO_CFLAGS = @GLPROTO_CFLAGS@
+GLPROTO_LIBS = @GLPROTO_LIBS@
+GLVND_CFLAGS = @GLVND_CFLAGS@
+GLVND_LIBS = @GLVND_LIBS@
+GLX_TLS = @GLX_TLS@
+GL_LIB = @GL_LIB@
+GL_LIB_DEPS = @GL_LIB_DEPS@
+GL_PC_CFLAGS = @GL_PC_CFLAGS@
+GL_PC_LIB_PRIV = @GL_PC_LIB_PRIV@
+GL_PC_REQ_PRIV = @GL_PC_REQ_PRIV@
+GL_PKGCONF_LIB = @GL_PKGCONF_LIB@
+GREP = @GREP@
+I915_CFLAGS = @I915_CFLAGS@
+I915_LIBS = @I915_LIBS@
+INDENT = @INDENT@
+INDENT_FLAGS = @INDENT_FLAGS@
+INSTALL = @INSTALL@
+INSTALL_DATA = @INSTALL_DATA@
+INSTALL_PROGRAM = @INSTALL_PROGRAM@
+INSTALL_SCRIPT = @INSTALL_SCRIPT@
+INSTALL_STRIP_PROGRAM = @INSTALL_STRIP_PROGRAM@
+LD = @LD@
+LDFLAGS = @LDFLAGS@
+LD_BUILD_ID = @LD_BUILD_ID@
+LD_NO_UNDEFINED = @LD_NO_UNDEFINED@
+LEX = @LEX@
+LEXLIB = @LEXLIB@
+LEX_OUTPUT_ROOT = @LEX_OUTPUT_ROOT@
+LIBATOMIC_LIBS = @LIBATOMIC_LIBS@
+LIBCLC_INCLUDEDIR = @LIBCLC_INCLUDEDIR@
+LIBCLC_LIBEXECDIR = @LIBCLC_LIBEXECDIR@
+LIBDRM_CFLAGS = @LIBDRM_CFLAGS@
+LIBDRM_LIBS = @LIBDRM_LIBS@
+LIBELF_CFLAGS = @LIBELF_CFLAGS@
+LIBELF_LIBS = @LIBELF_LIBS@
+LIBGLVND_DATADIR = @LIBGLVND_DATADIR@
+LIBOBJS = @LIBOBJS@
+LIBS = @LIBS@
+LIBSENSORS_LIBS = @LIBSENSORS_LIBS@
+LIBTOOL = @LIBTOOL@
+LIBUNWIND_CFLAGS = @LIBUNWIND_CFLAGS@
+LIBUNWIND_LIBS = @LIBUNWIND_LIBS@
+LIB_DIR = @LIB_DIR@
+LIB_EXT = @LIB_EXT@
+LIPO = @LIPO@
+LLVM_CFLAGS = @LLVM_CFLAGS@
+LLVM_CONFIG = @LLVM_CONFIG@
+LLVM_CXXFLAGS = @LLVM_CXXFLAGS@
+LLVM_INCLUDEDIR = @LLVM_INCLUDEDIR@
+LLVM_LDFLAGS = @LLVM_LDFLAGS@
+LLVM_LIBS = @LLVM_LIBS@
+LN_S = @LN_S@
+LTLIBOBJS = @LTLIBOBJS@
+LT_SYS_LIBRARY_PATH = @LT_SYS_LIBRARY_PATH@
+MAKEINFO = @MAKEINFO@
+MANIFEST_TOOL = @MANIFEST_TOOL@
+MKDIR_P = @MKDIR_P@
+MSVC2013_COMPAT_CFLAGS = @MSVC2013_COMPAT_CFLAGS@
+MSVC2013_COMPAT_CXXFLAGS = @MSVC2013_COMPAT_CXXFLAGS@
+NINE_MAJOR = @NINE_MAJOR@
+NINE_MINOR = @NINE_MINOR@
+NINE_PATCH = @NINE_PATCH@
+NINE_VERSION = @NINE_VERSION@
+NM = @NM@
+NMEDIT = @NMEDIT@
+NOUVEAU_CFLAGS = @NOUVEAU_CFLAGS@
+NOUVEAU_LIBS = @NOUVEAU_LIBS@
+NVVIEUX_CFLAGS = @NVVIEUX_CFLAGS@
+NVVIEUX_LIBS = @NVVIEUX_LIBS@
+OBJDUMP = @OBJDUMP@
+OBJEXT = @OBJEXT@
+OMX_BELLAGIO_CFLAGS = @OMX_BELLAGIO_CFLAGS@
+OMX_BELLAGIO_LIBS = @OMX_BELLAGIO_LIBS@
+OMX_BELLAGIO_LIB_INSTALL_DIR = @OMX_BELLAGIO_LIB_INSTALL_DIR@
+OMX_TIZONIA_CFLAGS = @OMX_TIZONIA_CFLAGS@
+OMX_TIZONIA_LIBS = @OMX_TIZONIA_LIBS@
+OMX_TIZONIA_LIB_INSTALL_DIR = @OMX_TIZONIA_LIB_INSTALL_DIR@
+OPENCL_LIBNAME = @OPENCL_LIBNAME@
+OPENCL_VERSION = @OPENCL_VERSION@
+OSMESA_LIB = @OSMESA_LIB@
+OSMESA_LIB_DEPS = @OSMESA_LIB_DEPS@
+OSMESA_PC_LIB_PRIV = @OSMESA_PC_LIB_PRIV@
+OSMESA_PC_REQ = @OSMESA_PC_REQ@
+OSMESA_VERSION = @OSMESA_VERSION@
+OTOOL = @OTOOL@
+OTOOL64 = @OTOOL64@
+PACKAGE = @PACKAGE@
+PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
+PACKAGE_NAME = @PACKAGE_NAME@
+PACKAGE_STRING = @PACKAGE_STRING@
+PACKAGE_TARNAME = @PACKAGE_TARNAME@
+PACKAGE_URL = @PACKAGE_URL@
+PACKAGE_VERSION = @PACKAGE_VERSION@
+PATH_SEPARATOR = @PATH_SEPARATOR@
+PKG_CONFIG = @PKG_CONFIG@
+PKG_CONFIG_LIBDIR = @PKG_CONFIG_LIBDIR@
+PKG_CONFIG_PATH = @PKG_CONFIG_PATH@
+POSIX_SHELL = @POSIX_SHELL@
+PTHREADSTUBS_CFLAGS = @PTHREADSTUBS_CFLAGS@
+PTHREADSTUBS_LIBS = @PTHREADSTUBS_LIBS@
+PTHREAD_CC = @PTHREAD_CC@
+PTHREAD_CFLAGS = @PTHREAD_CFLAGS@
+PTHREAD_LIBS = @PTHREAD_LIBS@
+PWR8_CFLAGS = @PWR8_CFLAGS@
+PYTHON = @PYTHON@
+PYTHON_EXEC_PREFIX = @PYTHON_EXEC_PREFIX@
+PYTHON_PLATFORM = @PYTHON_PLATFORM@
+PYTHON_PREFIX = @PYTHON_PREFIX@
+PYTHON_VERSION = @PYTHON_VERSION@
+RADEON_CFLAGS = @RADEON_CFLAGS@
+RADEON_LIBS = @RADEON_LIBS@
+RANLIB = @RANLIB@
+RM = @RM@
+SCANNER_ARG = @SCANNER_ARG@
+SED = @SED@
+SELINUX_CFLAGS = @SELINUX_CFLAGS@
+SELINUX_LIBS = @SELINUX_LIBS@
+SET_MAKE = @SET_MAKE@
+SHELL = @SHELL@
+SIMPENROSE_CFLAGS = @SIMPENROSE_CFLAGS@
+SIMPENROSE_LIBS = @SIMPENROSE_LIBS@
+SSE41_CFLAGS = @SSE41_CFLAGS@
+STRIP = @STRIP@
+SWR_AVX2_CXXFLAGS = @SWR_AVX2_CXXFLAGS@
+SWR_AVX_CXXFLAGS = @SWR_AVX_CXXFLAGS@
+SWR_KNL_CXXFLAGS = @SWR_KNL_CXXFLAGS@
+SWR_SKX_CXXFLAGS = @SWR_SKX_CXXFLAGS@
+V3D_SIMULATOR_CFLAGS = @V3D_SIMULATOR_CFLAGS@
+V3D_SIMULATOR_LIBS = @V3D_SIMULATOR_LIBS@
+VALGRIND_CFLAGS = @VALGRIND_CFLAGS@
+VALGRIND_LIBS = @VALGRIND_LIBS@
+VA_CFLAGS = @VA_CFLAGS@
+VA_LIBS = @VA_LIBS@
+VA_LIB_INSTALL_DIR = @VA_LIB_INSTALL_DIR@
+VA_MAJOR = @VA_MAJOR@
+VA_MINOR = @VA_MINOR@
+VC4_CFLAGS = @VC4_CFLAGS@
+VC4_LIBS = @VC4_LIBS@
+VDPAU_CFLAGS = @VDPAU_CFLAGS@
+VDPAU_LIBS = @VDPAU_LIBS@
+VDPAU_LIB_INSTALL_DIR = @VDPAU_LIB_INSTALL_DIR@
+VDPAU_MAJOR = @VDPAU_MAJOR@
+VDPAU_MINOR = @VDPAU_MINOR@
+VERSION = @VERSION@
+VISIBILITY_CFLAGS = @VISIBILITY_CFLAGS@
+VISIBILITY_CXXFLAGS = @VISIBILITY_CXXFLAGS@
+VL_CFLAGS = @VL_CFLAGS@
+VL_LIBS = @VL_LIBS@
+VULKAN_ICD_INSTALL_DIR = @VULKAN_ICD_INSTALL_DIR@
+WAYLAND_CLIENT_CFLAGS = @WAYLAND_CLIENT_CFLAGS@
+WAYLAND_CLIENT_LIBS = @WAYLAND_CLIENT_LIBS@
+WAYLAND_EGL_CFLAGS = @WAYLAND_EGL_CFLAGS@
+WAYLAND_EGL_LIBS = @WAYLAND_EGL_LIBS@
+WAYLAND_PROTOCOLS_CFLAGS = @WAYLAND_PROTOCOLS_CFLAGS@
+WAYLAND_PROTOCOLS_DATADIR = @WAYLAND_PROTOCOLS_DATADIR@
+WAYLAND_PROTOCOLS_LIBS = @WAYLAND_PROTOCOLS_LIBS@
+WAYLAND_SCANNER = @WAYLAND_SCANNER@
+WAYLAND_SCANNER_CFLAGS = @WAYLAND_SCANNER_CFLAGS@
+WAYLAND_SCANNER_LIBS = @WAYLAND_SCANNER_LIBS@
+WAYLAND_SERVER_CFLAGS = @WAYLAND_SERVER_CFLAGS@
+WAYLAND_SERVER_LIBS = @WAYLAND_SERVER_LIBS@
+WNO_OVERRIDE_INIT = @WNO_OVERRIDE_INIT@
+X11_INCLUDES = @X11_INCLUDES@
+XA_MAJOR = @XA_MAJOR@
+XA_MINOR = @XA_MINOR@
+XA_PATCH = @XA_PATCH@
+XA_VERSION = @XA_VERSION@
+XCB_DRI2_CFLAGS = @XCB_DRI2_CFLAGS@
+XCB_DRI2_LIBS = @XCB_DRI2_LIBS@
+XCB_DRI3_CFLAGS = @XCB_DRI3_CFLAGS@
+XCB_DRI3_LIBS = @XCB_DRI3_LIBS@
+XCB_DRI3_MODIFIERS_CFLAGS = @XCB_DRI3_MODIFIERS_CFLAGS@
+XCB_DRI3_MODIFIERS_LIBS = @XCB_DRI3_MODIFIERS_LIBS@
+XCB_RANDR_CFLAGS = @XCB_RANDR_CFLAGS@
+XCB_RANDR_LIBS = @XCB_RANDR_LIBS@
+XLIBGL_CFLAGS = @XLIBGL_CFLAGS@
+XLIBGL_LIBS = @XLIBGL_LIBS@
+XLIB_RANDR_CFLAGS = @XLIB_RANDR_CFLAGS@
+XLIB_RANDR_LIBS = @XLIB_RANDR_LIBS@
+XVMC_CFLAGS = @XVMC_CFLAGS@
+XVMC_LIBS = @XVMC_LIBS@
+XVMC_LIB_INSTALL_DIR = @XVMC_LIB_INSTALL_DIR@
+XVMC_MAJOR = @XVMC_MAJOR@
+XVMC_MINOR = @XVMC_MINOR@
+YACC = @YACC@
+YFLAGS = @YFLAGS@
+ZLIB_CFLAGS = @ZLIB_CFLAGS@
+ZLIB_LIBS = @ZLIB_LIBS@
+abs_builddir = @abs_builddir@
+abs_srcdir = @abs_srcdir@
+abs_top_builddir = @abs_top_builddir@
+abs_top_srcdir = @abs_top_srcdir@
+ac_ct_AR = @ac_ct_AR@
+ac_ct_CC = @ac_ct_CC@
+ac_ct_CXX = @ac_ct_CXX@
+ac_ct_DUMPBIN = @ac_ct_DUMPBIN@
+acv_mako_found = @acv_mako_found@
+am__include = @am__include@
+am__leading_dot = @am__leading_dot@
+am__quote = @am__quote@
+am__tar = @am__tar@
+am__untar = @am__untar@
+ax_pthread_config = @ax_pthread_config@
+bindir = @bindir@
+build = @build@
+build_alias = @build_alias@
+build_cpu = @build_cpu@
+build_os = @build_os@
+build_vendor = @build_vendor@
+builddir = @builddir@
+datadir = @datadir@
+datarootdir = @datarootdir@
+docdir = @docdir@
+dvidir = @dvidir@
+exec_prefix = @exec_prefix@
+host = @host@
+host_alias = @host_alias@
+host_cpu = @host_cpu@
+host_os = @host_os@
+host_vendor = @host_vendor@
+htmldir = @htmldir@
+ifGNUmake = @ifGNUmake@
+includedir = @includedir@
+infodir = @infodir@
+install_sh = @install_sh@
+libdir = @libdir@
+libexecdir = @libexecdir@
+localedir = @localedir@
+localstatedir = @localstatedir@
+mandir = @mandir@
+mkdir_p = @mkdir_p@
+oldincludedir = @oldincludedir@
+pdfdir = @pdfdir@
+pkgpyexecdir = @pkgpyexecdir@
+pkgpythondir = @pkgpythondir@
+prefix = @prefix@
+program_transform_name = @program_transform_name@
+psdir = @psdir@
+pyexecdir = @pyexecdir@
+pythondir = @pythondir@
+sbindir = @sbindir@
+sharedstatedir = @sharedstatedir@
+srcdir = @srcdir@
+sysconfdir = @sysconfdir@
+target = @target@
+target_alias = @target_alias@
+target_cpu = @target_cpu@
+target_os = @target_os@
+target_vendor = @target_vendor@
+top_build_prefix = @top_build_prefix@
+top_builddir = @top_builddir@
+top_srcdir = @top_srcdir@
+AM_CPPFLAGS = \
+	-I$(top_srcdir)/include \
+	-I$(top_srcdir)/src/util \
+	$(DEFINES)
+
+vma_random_test_SOURCES = \
+	vma_random_test.cpp
+
+vma_random_test_LDADD = \
+	$(top_builddir)/src/util/libmesautil.la
+
+vma_random_test_CXXFLAGS = $(CXX11_CXXFLAGS)
+EXTRA_DIST = meson.build
+all: all-am
+
+.SUFFIXES:
+.SUFFIXES: .cpp .lo .log .o .obj .test .test$(EXEEXT) .trs
+$(srcdir)/Makefile.in:  $(srcdir)/Makefile.am  $(am__configure_deps)
+	@for dep in $?; do \
+	  case '$(am__configure_deps)' in \
+	    *$$dep*) \
+	      ( cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh ) \
+	        && { if test -f $@; then exit 0; else break; fi; }; \
+	      exit 1;; \
+	  esac; \
+	done; \
+	echo ' cd $(top_srcdir) && $(AUTOMAKE) --foreign src/util/tests/vma/Makefile'; \
+	$(am__cd) $(top_srcdir) && \
+	  $(AUTOMAKE) --foreign src/util/tests/vma/Makefile
+Makefile: $(srcdir)/Makefile.in $(top_builddir)/config.status
+	@case '$?' in \
+	  *config.status*) \
+	    cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh;; \
+	  *) \
+	    echo ' cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__maybe_remake_depfiles)'; \
+	    cd $(top_builddir) && $(SHELL) ./config.status $(subdir)/$@ $(am__maybe_remake_depfiles);; \
+	esac;
+
+$(top_builddir)/config.status: $(top_srcdir)/configure $(CONFIG_STATUS_DEPENDENCIES)
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+
+$(top_srcdir)/configure:  $(am__configure_deps)
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+$(ACLOCAL_M4):  $(am__aclocal_m4_deps)
+	cd $(top_builddir) && $(MAKE) $(AM_MAKEFLAGS) am--refresh
+$(am__aclocal_m4_deps):
+
+clean-checkPROGRAMS:
+	@list='$(check_PROGRAMS)'; test -n "$$list" || exit 0; \
+	echo " rm -f" $$list; \
+	rm -f $$list || exit $$?; \
+	test -n "$(EXEEXT)" || exit 0; \
+	list=`for p in $$list; do echo "$$p"; done | sed 's/$(EXEEXT)$$//'`; \
+	echo " rm -f" $$list; \
+	rm -f $$list
+
+vma_random_test$(EXEEXT): $(vma_random_test_OBJECTS) $(vma_random_test_DEPENDENCIES) $(EXTRA_vma_random_test_DEPENDENCIES) 
+	@rm -f vma_random_test$(EXEEXT)
+	$(AM_V_CXXLD)$(vma_random_test_LINK) $(vma_random_test_OBJECTS) $(vma_random_test_LDADD) $(LIBS)
+
+mostlyclean-compile:
+	-rm -f *.$(OBJEXT)
+
+distclean-compile:
+	-rm -f *.tab.c
+
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/vma_random_test-vma_random_test.Po@am__quote@ # am--include-marker
+
+$(am__depfiles_remade):
+	@$(MKDIR_P) $(@D)
+	@echo '# dummy' >$@-t && $(am__mv) $@-t $@
+
+am--depfiles: $(am__depfiles_remade)
+
+.cpp.o:
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)depbase=`echo $@ | sed 's|[^/]*$$|$(DEPDIR)/&|;s|\.o$$||'`;\
+@am__fastdepCXX_TRUE@	$(CXXCOMPILE) -MT $@ -MD -MP -MF $$depbase.Tpo -c -o $@ $< &&\
+@am__fastdepCXX_TRUE@	$(am__mv) $$depbase.Tpo $$depbase.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ $<
+
+.cpp.obj:
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)depbase=`echo $@ | sed 's|[^/]*$$|$(DEPDIR)/&|;s|\.obj$$||'`;\
+@am__fastdepCXX_TRUE@	$(CXXCOMPILE) -MT $@ -MD -MP -MF $$depbase.Tpo -c -o $@ `$(CYGPATH_W) '$<'` &&\
+@am__fastdepCXX_TRUE@	$(am__mv) $$depbase.Tpo $$depbase.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=no @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXXCOMPILE) -c -o $@ `$(CYGPATH_W) '$<'`
+
+.cpp.lo:
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)depbase=`echo $@ | sed 's|[^/]*$$|$(DEPDIR)/&|;s|\.lo$$||'`;\
+@am__fastdepCXX_TRUE@	$(LTCXXCOMPILE) -MT $@ -MD -MP -MF $$depbase.Tpo -c -o $@ $< &&\
+@am__fastdepCXX_TRUE@	$(am__mv) $$depbase.Tpo $$depbase.Plo
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='$<' object='$@' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(LTCXXCOMPILE) -c -o $@ $<
+
+vma_random_test-vma_random_test.o: vma_random_test.cpp
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(vma_random_test_CXXFLAGS) $(CXXFLAGS) -MT vma_random_test-vma_random_test.o -MD -MP -MF $(DEPDIR)/vma_random_test-vma_random_test.Tpo -c -o vma_random_test-vma_random_test.o `test -f 'vma_random_test.cpp' || echo '$(srcdir)/'`vma_random_test.cpp
+@am__fastdepCXX_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/vma_random_test-vma_random_test.Tpo $(DEPDIR)/vma_random_test-vma_random_test.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='vma_random_test.cpp' object='vma_random_test-vma_random_test.o' libtool=no @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(vma_random_test_CXXFLAGS) $(CXXFLAGS) -c -o vma_random_test-vma_random_test.o `test -f 'vma_random_test.cpp' || echo '$(srcdir)/'`vma_random_test.cpp
+
+vma_random_test-vma_random_test.obj: vma_random_test.cpp
+@am__fastdepCXX_TRUE@	$(AM_V_CXX)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(vma_random_test_CXXFLAGS) $(CXXFLAGS) -MT vma_random_test-vma_random_test.obj -MD -MP -MF $(DEPDIR)/vma_random_test-vma_random_test.Tpo -c -o vma_random_test-vma_random_test.obj `if test -f 'vma_random_test.cpp'; then $(CYGPATH_W) 'vma_random_test.cpp'; else $(CYGPATH_W) '$(srcdir)/vma_random_test.cpp'; fi`
+@am__fastdepCXX_TRUE@	$(AM_V_at)$(am__mv) $(DEPDIR)/vma_random_test-vma_random_test.Tpo $(DEPDIR)/vma_random_test-vma_random_test.Po
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	$(AM_V_CXX)source='vma_random_test.cpp' object='vma_random_test-vma_random_test.obj' libtool=no @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCXX_FALSE@	DEPDIR=$(DEPDIR) $(CXXDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCXX_FALSE@	$(AM_V_CXX@am__nodep@)$(CXX) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(vma_random_test_CXXFLAGS) $(CXXFLAGS) -c -o vma_random_test-vma_random_test.obj `if test -f 'vma_random_test.cpp'; then $(CYGPATH_W) 'vma_random_test.cpp'; else $(CYGPATH_W) '$(srcdir)/vma_random_test.cpp'; fi`
+
+mostlyclean-libtool:
+	-rm -f *.lo
+
+clean-libtool:
+	-rm -rf .libs _libs
+
+ID: $(am__tagged_files)
+	$(am__define_uniq_tagged_files); mkid -fID $$unique
+tags: tags-am
+TAGS: tags
+
+tags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+	set x; \
+	here=`pwd`; \
+	$(am__define_uniq_tagged_files); \
+	shift; \
+	if test -z "$(ETAGS_ARGS)$$*$$unique"; then :; else \
+	  test -n "$$unique" || unique=$$empty_fix; \
+	  if test $$# -gt 0; then \
+	    $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
+	      "$$@" $$unique; \
+	  else \
+	    $(ETAGS) $(ETAGSFLAGS) $(AM_ETAGSFLAGS) $(ETAGS_ARGS) \
+	      $$unique; \
+	  fi; \
+	fi
+ctags: ctags-am
+
+CTAGS: ctags
+ctags-am: $(TAGS_DEPENDENCIES) $(am__tagged_files)
+	$(am__define_uniq_tagged_files); \
+	test -z "$(CTAGS_ARGS)$$unique" \
+	  || $(CTAGS) $(CTAGSFLAGS) $(AM_CTAGSFLAGS) $(CTAGS_ARGS) \
+	     $$unique
+
+GTAGS:
+	here=`$(am__cd) $(top_builddir) && pwd` \
+	  && $(am__cd) $(top_srcdir) \
+	  && gtags -i $(GTAGS_ARGS) "$$here"
+cscopelist: cscopelist-am
+
+cscopelist-am: $(am__tagged_files)
+	list='$(am__tagged_files)'; \
+	case "$(srcdir)" in \
+	  [\\/]* | ?:[\\/]*) sdir="$(srcdir)" ;; \
+	  *) sdir=$(subdir)/$(srcdir) ;; \
+	esac; \
+	for i in $$list; do \
+	  if test -f "$$i"; then \
+	    echo "$(subdir)/$$i"; \
+	  else \
+	    echo "$$sdir/$$i"; \
+	  fi; \
+	done >> $(top_builddir)/cscope.files
+
+distclean-tags:
+	-rm -f TAGS ID GTAGS GRTAGS GSYMS GPATH tags
+
+# Recover from deleted '.trs' file; this should ensure that
+# "rm -f foo.log; make foo.trs" re-run 'foo.test', and re-create
+# both 'foo.log' and 'foo.trs'.  Break the recipe in two subshells
+# to avoid problems with "make -n".
+.log.trs:
+	rm -f $< $@
+	$(MAKE) $(AM_MAKEFLAGS) $<
+
+# Leading 'am--fnord' is there to ensure the list of targets does not
+# expand to empty, as could happen e.g. with make check TESTS=''.
+am--fnord $(TEST_LOGS) $(TEST_LOGS:.log=.trs): $(am__force_recheck)
+am--force-recheck:
+	@:
+
+$(TEST_SUITE_LOG): $(TEST_LOGS)
+	@$(am__set_TESTS_bases); \
+	am__f_ok () { test -f "$$1" && test -r "$$1"; }; \
+	redo_bases=`for i in $$bases; do \
+	              am__f_ok $$i.trs && am__f_ok $$i.log || echo $$i; \
+	            done`; \
+	if test -n "$$redo_bases"; then \
+	  redo_logs=`for i in $$redo_bases; do echo $$i.log; done`; \
+	  redo_results=`for i in $$redo_bases; do echo $$i.trs; done`; \
+	  if $(am__make_dryrun); then :; else \
+	    rm -f $$redo_logs && rm -f $$redo_results || exit 1; \
+	  fi; \
+	fi; \
+	if test -n "$$am__remaking_logs"; then \
+	  echo "fatal: making $(TEST_SUITE_LOG): possible infinite" \
+	       "recursion detected" >&2; \
+	elif test -n "$$redo_logs"; then \
+	  am__remaking_logs=yes $(MAKE) $(AM_MAKEFLAGS) $$redo_logs; \
+	fi; \
+	if $(am__make_dryrun); then :; else \
+	  st=0;  \
+	  errmsg="fatal: making $(TEST_SUITE_LOG): failed to create"; \
+	  for i in $$redo_bases; do \
+	    test -f $$i.trs && test -r $$i.trs \
+	      || { echo "$$errmsg $$i.trs" >&2; st=1; }; \
+	    test -f $$i.log && test -r $$i.log \
+	      || { echo "$$errmsg $$i.log" >&2; st=1; }; \
+	  done; \
+	  test $$st -eq 0 || exit 1; \
+	fi
+	@$(am__sh_e_setup); $(am__tty_colors); $(am__set_TESTS_bases); \
+	ws='[ 	]'; \
+	results=`for b in $$bases; do echo $$b.trs; done`; \
+	test -n "$$results" || results=/dev/null; \
+	all=`  grep "^$$ws*:test-result:"           $$results | wc -l`; \
+	pass=` grep "^$$ws*:test-result:$$ws*PASS"  $$results | wc -l`; \
+	fail=` grep "^$$ws*:test-result:$$ws*FAIL"  $$results | wc -l`; \
+	skip=` grep "^$$ws*:test-result:$$ws*SKIP"  $$results | wc -l`; \
+	xfail=`grep "^$$ws*:test-result:$$ws*XFAIL" $$results | wc -l`; \
+	xpass=`grep "^$$ws*:test-result:$$ws*XPASS" $$results | wc -l`; \
+	error=`grep "^$$ws*:test-result:$$ws*ERROR" $$results | wc -l`; \
+	if test `expr $$fail + $$xpass + $$error` -eq 0; then \
+	  success=true; \
+	else \
+	  success=false; \
+	fi; \
+	br='==================='; br=$$br$$br$$br$$br; \
+	result_count () \
+	{ \
+	    if test x"$$1" = x"--maybe-color"; then \
+	      maybe_colorize=yes; \
+	    elif test x"$$1" = x"--no-color"; then \
+	      maybe_colorize=no; \
+	    else \
+	      echo "$@: invalid 'result_count' usage" >&2; exit 4; \
+	    fi; \
+	    shift; \
+	    desc=$$1 count=$$2; \
+	    if test $$maybe_colorize = yes && test $$count -gt 0; then \
+	      color_start=$$3 color_end=$$std; \
+	    else \
+	      color_start= color_end=; \
+	    fi; \
+	    echo "$${color_start}# $$desc $$count$${color_end}"; \
+	}; \
+	create_testsuite_report () \
+	{ \
+	  result_count $$1 "TOTAL:" $$all   "$$brg"; \
+	  result_count $$1 "PASS: " $$pass  "$$grn"; \
+	  result_count $$1 "SKIP: " $$skip  "$$blu"; \
+	  result_count $$1 "XFAIL:" $$xfail "$$lgn"; \
+	  result_count $$1 "FAIL: " $$fail  "$$red"; \
+	  result_count $$1 "XPASS:" $$xpass "$$red"; \
+	  result_count $$1 "ERROR:" $$error "$$mgn"; \
+	}; \
+	{								\
+	  echo "$(PACKAGE_STRING): $(subdir)/$(TEST_SUITE_LOG)" |	\
+	    $(am__rst_title);						\
+	  create_testsuite_report --no-color;				\
+	  echo;								\
+	  echo ".. contents:: :depth: 2";				\
+	  echo;								\
+	  for b in $$bases; do echo $$b; done				\
+	    | $(am__create_global_log);					\
+	} >$(TEST_SUITE_LOG).tmp || exit 1;				\
+	mv $(TEST_SUITE_LOG).tmp $(TEST_SUITE_LOG);			\
+	if $$success; then						\
+	  col="$$grn";							\
+	 else								\
+	  col="$$red";							\
+	  test x"$$VERBOSE" = x || cat $(TEST_SUITE_LOG);		\
+	fi;								\
+	echo "$${col}$$br$${std}"; 					\
+	echo "$${col}Testsuite summary for $(PACKAGE_STRING)$${std}";	\
+	echo "$${col}$$br$${std}"; 					\
+	create_testsuite_report --maybe-color;				\
+	echo "$$col$$br$$std";						\
+	if $$success; then :; else					\
+	  echo "$${col}See $(subdir)/$(TEST_SUITE_LOG)$${std}";		\
+	  if test -n "$(PACKAGE_BUGREPORT)"; then			\
+	    echo "$${col}Please report to $(PACKAGE_BUGREPORT)$${std}";	\
+	  fi;								\
+	  echo "$$col$$br$$std";					\
+	fi;								\
+	$$success || exit 1
+
+check-TESTS: $(check_PROGRAMS)
+	@list='$(RECHECK_LOGS)';           test -z "$$list" || rm -f $$list
+	@list='$(RECHECK_LOGS:.log=.trs)'; test -z "$$list" || rm -f $$list
+	@test -z "$(TEST_SUITE_LOG)" || rm -f $(TEST_SUITE_LOG)
+	@set +e; $(am__set_TESTS_bases); \
+	log_list=`for i in $$bases; do echo $$i.log; done`; \
+	trs_list=`for i in $$bases; do echo $$i.trs; done`; \
+	log_list=`echo $$log_list`; trs_list=`echo $$trs_list`; \
+	$(MAKE) $(AM_MAKEFLAGS) $(TEST_SUITE_LOG) TEST_LOGS="$$log_list"; \
+	exit $$?;
+recheck: all $(check_PROGRAMS)
+	@test -z "$(TEST_SUITE_LOG)" || rm -f $(TEST_SUITE_LOG)
+	@set +e; $(am__set_TESTS_bases); \
+	bases=`for i in $$bases; do echo $$i; done \
+	         | $(am__list_recheck_tests)` || exit 1; \
+	log_list=`for i in $$bases; do echo $$i.log; done`; \
+	log_list=`echo $$log_list`; \
+	$(MAKE) $(AM_MAKEFLAGS) $(TEST_SUITE_LOG) \
+	        am__force_recheck=am--force-recheck \
+	        TEST_LOGS="$$log_list"; \
+	exit $$?
+vma_random_test.log: vma_random_test$(EXEEXT)
+	@p='vma_random_test$(EXEEXT)'; \
+	b='vma_random_test'; \
+	$(am__check_pre) $(LOG_DRIVER) --test-name "$$f" \
+	--log-file $$b.log --trs-file $$b.trs \
+	$(am__common_driver_flags) $(AM_LOG_DRIVER_FLAGS) $(LOG_DRIVER_FLAGS) -- $(LOG_COMPILE) \
+	"$$tst" $(AM_TESTS_FD_REDIRECT)
+.test.log:
+	@p='$<'; \
+	$(am__set_b); \
+	$(am__check_pre) $(TEST_LOG_DRIVER) --test-name "$$f" \
+	--log-file $$b.log --trs-file $$b.trs \
+	$(am__common_driver_flags) $(AM_TEST_LOG_DRIVER_FLAGS) $(TEST_LOG_DRIVER_FLAGS) -- $(TEST_LOG_COMPILE) \
+	"$$tst" $(AM_TESTS_FD_REDIRECT)
+@am__EXEEXT_TRUE@.test$(EXEEXT).log:
+@am__EXEEXT_TRUE@	@p='$<'; \
+@am__EXEEXT_TRUE@	$(am__set_b); \
+@am__EXEEXT_TRUE@	$(am__check_pre) $(TEST_LOG_DRIVER) --test-name "$$f" \
+@am__EXEEXT_TRUE@	--log-file $$b.log --trs-file $$b.trs \
+@am__EXEEXT_TRUE@	$(am__common_driver_flags) $(AM_TEST_LOG_DRIVER_FLAGS) $(TEST_LOG_DRIVER_FLAGS) -- $(TEST_LOG_COMPILE) \
+@am__EXEEXT_TRUE@	"$$tst" $(AM_TESTS_FD_REDIRECT)
+
+distdir: $(BUILT_SOURCES)
+	$(MAKE) $(AM_MAKEFLAGS) distdir-am
+
+distdir-am: $(DISTFILES)
+	@srcdirstrip=`echo "$(srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
+	topsrcdirstrip=`echo "$(top_srcdir)" | sed 's/[].[^$$\\*]/\\\\&/g'`; \
+	list='$(DISTFILES)'; \
+	  dist_files=`for file in $$list; do echo $$file; done | \
+	  sed -e "s|^$$srcdirstrip/||;t" \
+	      -e "s|^$$topsrcdirstrip/|$(top_builddir)/|;t"`; \
+	case $$dist_files in \
+	  */*) $(MKDIR_P) `echo "$$dist_files" | \
+			   sed '/\//!d;s|^|$(distdir)/|;s,/[^/]*$$,,' | \
+			   sort -u` ;; \
+	esac; \
+	for file in $$dist_files; do \
+	  if test -f $$file || test -d $$file; then d=.; else d=$(srcdir); fi; \
+	  if test -d $$d/$$file; then \
+	    dir=`echo "/$$file" | sed -e 's,/[^/]*$$,,'`; \
+	    if test -d "$(distdir)/$$file"; then \
+	      find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
+	    fi; \
+	    if test -d $(srcdir)/$$file && test $$d != $(srcdir); then \
+	      cp -fpR $(srcdir)/$$file "$(distdir)$$dir" || exit 1; \
+	      find "$(distdir)/$$file" -type d ! -perm -700 -exec chmod u+rwx {} \;; \
+	    fi; \
+	    cp -fpR $$d/$$file "$(distdir)$$dir" || exit 1; \
+	  else \
+	    test -f "$(distdir)/$$file" \
+	    || cp -p $$d/$$file "$(distdir)/$$file" \
+	    || exit 1; \
+	  fi; \
+	done
+check-am: all-am
+	$(MAKE) $(AM_MAKEFLAGS) $(check_PROGRAMS)
+	$(MAKE) $(AM_MAKEFLAGS) check-TESTS
+check: check-am
+all-am: Makefile
+installdirs:
+install: install-am
+install-exec: install-exec-am
+install-data: install-data-am
+uninstall: uninstall-am
+
+install-am: all-am
+	@$(MAKE) $(AM_MAKEFLAGS) install-exec-am install-data-am
+
+installcheck: installcheck-am
+install-strip:
+	if test -z '$(STRIP)'; then \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	      install; \
+	else \
+	  $(MAKE) $(AM_MAKEFLAGS) INSTALL_PROGRAM="$(INSTALL_STRIP_PROGRAM)" \
+	    install_sh_PROGRAM="$(INSTALL_STRIP_PROGRAM)" INSTALL_STRIP_FLAG=-s \
+	    "INSTALL_PROGRAM_ENV=STRIPPROG='$(STRIP)'" install; \
+	fi
+mostlyclean-generic:
+	-test -z "$(TEST_LOGS)" || rm -f $(TEST_LOGS)
+	-test -z "$(TEST_LOGS:.log=.trs)" || rm -f $(TEST_LOGS:.log=.trs)
+	-test -z "$(TEST_SUITE_LOG)" || rm -f $(TEST_SUITE_LOG)
+
+clean-generic:
+
+distclean-generic:
+	-test -z "$(CONFIG_CLEAN_FILES)" || rm -f $(CONFIG_CLEAN_FILES)
+	-test . = "$(srcdir)" || test -z "$(CONFIG_CLEAN_VPATH_FILES)" || rm -f $(CONFIG_CLEAN_VPATH_FILES)
+
+maintainer-clean-generic:
+	@echo "This command is intended for maintainers to use"
+	@echo "it deletes files that may require special tools to rebuild."
+clean: clean-am
+
+clean-am: clean-checkPROGRAMS clean-generic clean-libtool \
+	mostlyclean-am
+
+distclean: distclean-am
+		-rm -f ./$(DEPDIR)/vma_random_test-vma_random_test.Po
+	-rm -f Makefile
+distclean-am: clean-am distclean-compile distclean-generic \
+	distclean-tags
+
+dvi: dvi-am
+
+dvi-am:
+
+html: html-am
+
+html-am:
+
+info: info-am
+
+info-am:
+
+install-data-am:
+
+install-dvi: install-dvi-am
+
+install-dvi-am:
+
+install-exec-am:
+
+install-html: install-html-am
+
+install-html-am:
+
+install-info: install-info-am
+
+install-info-am:
+
+install-man:
+
+install-pdf: install-pdf-am
+
+install-pdf-am:
+
+install-ps: install-ps-am
+
+install-ps-am:
+
+installcheck-am:
+
+maintainer-clean: maintainer-clean-am
+		-rm -f ./$(DEPDIR)/vma_random_test-vma_random_test.Po
+	-rm -f Makefile
+maintainer-clean-am: distclean-am maintainer-clean-generic
+
+mostlyclean: mostlyclean-am
+
+mostlyclean-am: mostlyclean-compile mostlyclean-generic \
+	mostlyclean-libtool
+
+pdf: pdf-am
+
+pdf-am:
+
+ps: ps-am
+
+ps-am:
+
+uninstall-am:
+
+.MAKE: check-am install-am install-strip
+
+.PHONY: CTAGS GTAGS TAGS all all-am am--depfiles check check-TESTS \
+	check-am clean clean-checkPROGRAMS clean-generic clean-libtool \
+	cscopelist-am ctags ctags-am distclean distclean-compile \
+	distclean-generic distclean-libtool distclean-tags distdir dvi \
+	dvi-am html html-am info info-am install install-am \
+	install-data install-data-am install-dvi install-dvi-am \
+	install-exec install-exec-am install-html install-html-am \
+	install-info install-info-am install-man install-pdf \
+	install-pdf-am install-ps install-ps-am install-strip \
+	installcheck installcheck-am installdirs maintainer-clean \
+	maintainer-clean-generic mostlyclean mostlyclean-compile \
+	mostlyclean-generic mostlyclean-libtool pdf pdf-am ps ps-am \
+	recheck tags tags-am uninstall uninstall-am
+
+.PRECIOUS: Makefile
+
+
+# Tell versions [3.59,3.63) of GNU make to not export all variables.
+# Otherwise a system limit (for SysV at least) may be exceeded.
+.NOEXPORT:
diff --git a/lib/mesa/src/util/tests/vma/meson.build b/lib/mesa/src/util/tests/vma/meson.build
new file mode 100644
index 000000000..53562db31
--- /dev/null
+++ b/lib/mesa/src/util/tests/vma/meson.build
@@ -0,0 +1,29 @@
+# Copyright © 2018 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+test(
+  'vma_random',
+  executable(
+    'vma_random_test',
+    'vma_random_test.cpp',
+    include_directories : [inc_include, inc_util],
+    link_with : [libmesa_util],
+  )
+)
diff --git a/lib/mesa/src/util/tests/vma/vma_random_test.cpp b/lib/mesa/src/util/tests/vma/vma_random_test.cpp
new file mode 100644
index 000000000..1f194fcdf
--- /dev/null
+++ b/lib/mesa/src/util/tests/vma/vma_random_test.cpp
@@ -0,0 +1,244 @@
+/*
+ * Copyright © 2018 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+/* it is a test after all */
+#undef NDEBUG
+
+#include <algorithm>
+#include <cassert>
+#include <climits>
+#include <cstdint>
+#include <cstdio>
+#include <cstdlib>
+#include <random>
+#include <set>
+#include <vector>
+
+#include <err.h>
+
+#include "vma.h"
+
+namespace {
+
+static const uint64_t MEM_PAGE_SIZE = 4096;
+
+struct allocation {
+   uint64_t start_page;
+   uint64_t num_pages;
+};
+
+struct allocation_less {
+   bool operator()(const allocation& lhs, const allocation& rhs) const
+   {
+      assert(lhs.start_page + lhs.num_pages > lhs.start_page);
+      return lhs.start_page + lhs.num_pages <= rhs.start_page;
+   }
+};
+
+constexpr uint64_t allocation_end_page(const allocation& a) {
+   return a.start_page + a.num_pages;
+}
+
+struct random_test {
+   static const uint64_t MEM_START_PAGE = 1;
+   static const uint64_t MEM_SIZE = 0xfffffffffffff000;
+   static const uint64_t MEM_PAGES = MEM_SIZE / MEM_PAGE_SIZE;
+
+   random_test(uint_fast32_t seed)
+      : heap_holes{allocation{MEM_START_PAGE, MEM_PAGES}}, rand{seed}
+   {
+      util_vma_heap_init(&heap, MEM_START_PAGE * MEM_PAGE_SIZE, MEM_SIZE);
+   }
+
+   void test(unsigned long count)
+   {
+      std::uniform_int_distribution<> one_to_thousand(1, 1000);
+      while (count-- > 0) {
+         int action = one_to_thousand(rand);
+         if (action == 1)          fill();
+         else if (action == 2)     empty();
+         else if (action < 374)    dealloc();
+         else                      alloc();
+      }
+   }
+
+   bool alloc(uint64_t size_order=52, uint64_t align_order=52)
+   {
+      std::geometric_distribution<> dist;
+
+      if (align_order > 51)
+         align_order = std::min(dist(rand), 51);
+      uint64_t align_pages = 1ULL << align_order;
+      uint64_t align = align_pages * MEM_PAGE_SIZE;
+
+      if (size_order > 51)
+         size_order = std::min(dist(rand), 51);
+      uint64_t size_pages = 1ULL << size_order;
+      uint64_t size = size_pages * MEM_PAGE_SIZE;
+
+      uint64_t addr = util_vma_heap_alloc(&heap, size, align);
+
+      if (addr == 0) {
+         /* assert no gaps are present in the tracker that could satisfy this
+          * allocation.
+          */
+         for (const auto& hole : heap_holes) {
+            uint64_t hole_alignment_pages =
+               (align_pages - (hole.start_page % align_pages)) % align_pages;
+            assert(hole.num_pages < size_pages + hole_alignment_pages);
+         }
+         return false;
+      } else {
+         assert(addr % align == 0);
+         uint64_t addr_page = addr / MEM_PAGE_SIZE;
+         allocation a{addr_page, size_pages};
+         auto i = heap_holes.find(a);
+         assert(i != end(heap_holes));
+         allocation hole = *i;
+
+         assert(hole.start_page <= addr_page);
+         assert(hole.num_pages >= size_pages + addr_page - hole.start_page);
+
+         heap_holes.erase(i);
+         if (hole.start_page < a.start_page) {
+            heap_holes.emplace(allocation{hole.start_page,
+                     a.start_page - hole.start_page});
+         }
+         if (allocation_end_page(hole) > allocation_end_page(a)) {
+            heap_holes.emplace(allocation{allocation_end_page(a),
+                     allocation_end_page(hole) - allocation_end_page(a)});
+         }
+
+         allocations.push_back(a);
+         return true;
+      }
+   }
+
+   void dealloc()
+   {
+      if (allocations.size() == 0)
+         return;
+
+      std::uniform_int_distribution<> dist(0, allocations.size() - 1);
+      int to_dealloc = dist(rand);
+
+      std::swap(allocations.at(to_dealloc), allocations.back());
+      allocation a = allocations.back();
+      allocations.pop_back();
+
+      util_vma_heap_free(&heap, a.start_page * MEM_PAGE_SIZE,
+                         a.num_pages * MEM_PAGE_SIZE);
+
+      assert(heap_holes.find(a) == end(heap_holes));
+      auto next = heap_holes.upper_bound(a);
+      if (next != end(heap_holes)) {
+         if (next->start_page == allocation_end_page(a)) {
+            allocation x {a.start_page, a.num_pages + next->num_pages};
+            next = heap_holes.erase(next);
+            next = heap_holes.insert(next, x);
+
+            if (next != begin(heap_holes)) {
+               auto prev = next;
+               prev--;
+               if (allocation_end_page(*prev) == next->start_page) {
+                  allocation x {prev->start_page,
+                        prev->num_pages + next->num_pages};
+
+                  heap_holes.erase(prev);
+                  next = heap_holes.erase(next);
+                  heap_holes.insert(next, x);
+               }
+            }
+
+            return;
+         }
+      }
+
+      if (next != begin(heap_holes)) {
+         auto prev = next;
+         prev--;
+         if (allocation_end_page(*prev) == a.start_page) {
+            allocation x {prev->start_page, prev->num_pages + a.num_pages};
+            next = heap_holes.erase(prev);
+            heap_holes.insert(next, x);
+
+            return;
+         }
+      }
+
+      heap_holes.emplace(a);
+   }
+
+   void fill()
+   {
+      for (int sz = 51; sz >= 0; sz--) {
+         while (alloc(sz, 0))
+            ;
+      }
+      assert(heap_holes.empty());
+   }
+
+   void empty()
+   {
+      while (allocations.size() != 0)
+         dealloc();
+      assert(heap_holes.size() == 1);
+      auto& hole = *begin(heap_holes);
+      assert(hole.start_page == MEM_START_PAGE && hole.num_pages == MEM_PAGES);
+   }
+
+   struct util_vma_heap heap;
+   std::set<allocation, allocation_less> heap_holes;
+   std::default_random_engine rand;
+   std::vector<allocation> allocations;
+};
+
+}
+
+int main(int argc, char **argv)
+{
+   unsigned long seed, count;
+   if (argc == 3) {
+      char *arg_end = NULL;
+      seed = strtoul(argv[1], &arg_end, 0);
+      if (!arg_end || *arg_end || seed == ULONG_MAX)
+         errx(1, "invalid seed \"%s\"", argv[1]);
+
+      arg_end = NULL;
+      count = strtoul(argv[2], &arg_end, 0);
+      if (!arg_end || *arg_end || count == ULONG_MAX)
+         errx(1, "invalid count \"%s\"", argv[2]);
+   } else if (argc == 1) {
+      /* importantly chosen prime numbers. */
+      seed = 8675309;
+      count = 2459;
+   } else {
+      errx(1, "USAGE: %s seed iter_count\n", argv[0]);
+   }
+
+   random_test r{(uint_fast32_t)seed};
+   r.test(count);
+
+   printf("ok\n");
+   return 0;
+}
diff --git a/lib/mesa/src/util/u_cpu_detect.c b/lib/mesa/src/util/u_cpu_detect.c
new file mode 100644
index 000000000..4dbb4d8fb
--- /dev/null
+++ b/lib/mesa/src/util/u_cpu_detect.c
@@ -0,0 +1,593 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Dennis Smit
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.  IN NO EVENT SHALL
+ * AUTHORS, COPYRIGHT HOLDERS, AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+/**
+ * @file
+ * CPU feature detection.
+ *
+ * @author Dennis Smit
+ * @author Based on the work of Eric Anholt <anholt@FreeBSD.org>
+ */
+
+#include "pipe/p_config.h"
+
+#include "util/u_debug.h"
+#include "u_cpu_detect.h"
+#include "c11/threads.h"
+
+#if defined(PIPE_ARCH_PPC)
+#if defined(PIPE_OS_APPLE)
+#include <sys/sysctl.h>
+#else
+#include <signal.h>
+#include <setjmp.h>
+#endif
+#endif
+
+#if defined(PIPE_OS_NETBSD) || defined(PIPE_OS_OPENBSD)
+#include <sys/param.h>
+#include <sys/sysctl.h>
+#include <machine/cpu.h>
+#endif
+
+#if defined(PIPE_OS_FREEBSD) || defined(PIPE_OS_DRAGONFLY)
+#include <sys/types.h>
+#include <sys/sysctl.h>
+#endif
+
+#if defined(PIPE_OS_LINUX)
+#include <signal.h>
+#include <fcntl.h>
+#include <elf.h>
+#endif
+
+#ifdef PIPE_OS_UNIX
+#include <unistd.h>
+#endif
+
+#if defined(HAS_ANDROID_CPUFEATURES)
+#include <cpu-features.h>
+#endif
+
+#if defined(PIPE_OS_WINDOWS)
+#include <windows.h>
+#if defined(PIPE_CC_MSVC)
+#include <intrin.h>
+#endif
+#endif
+
+
+#ifdef DEBUG
+DEBUG_GET_ONCE_BOOL_OPTION(dump_cpu, "GALLIUM_DUMP_CPU", FALSE)
+#endif
+
+
+struct util_cpu_caps util_cpu_caps;
+
+#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
+static int has_cpuid(void);
+#endif
+
+
+#if defined(PIPE_ARCH_PPC) && !defined(PIPE_OS_APPLE)
+static jmp_buf  __lv_powerpc_jmpbuf;
+static volatile sig_atomic_t __lv_powerpc_canjump = 0;
+
+static void
+sigill_handler(int sig)
+{
+   if (!__lv_powerpc_canjump) {
+      signal (sig, SIG_DFL);
+      raise (sig);
+   }
+
+   __lv_powerpc_canjump = 0;
+   longjmp(__lv_powerpc_jmpbuf, 1);
+}
+#endif
+
+#if defined(PIPE_ARCH_PPC)
+static void
+check_os_altivec_support(void)
+{
+#if defined(PIPE_OS_APPLE)
+   int sels[2] = {CTL_HW, HW_VECTORUNIT};
+   int has_vu = 0;
+   int len = sizeof (has_vu);
+   int err;
+
+   err = sysctl(sels, 2, &has_vu, &len, NULL, 0);
+
+   if (err == 0) {
+      if (has_vu != 0) {
+         util_cpu_caps.has_altivec = 1;
+      }
+   }
+#else /* !PIPE_OS_APPLE */
+   /* not on Apple/Darwin, do it the brute-force way */
+   /* this is borrowed from the libmpeg2 library */
+   signal(SIGILL, sigill_handler);
+   if (setjmp(__lv_powerpc_jmpbuf)) {
+      signal(SIGILL, SIG_DFL);
+   } else {
+      boolean enable_altivec = TRUE;    /* Default: enable  if available, and if not overridden */
+      boolean enable_vsx = TRUE;
+#ifdef DEBUG
+      /* Disabling Altivec code generation is not the same as disabling VSX code generation,
+       * which can be done simply by passing -mattr=-vsx to the LLVM compiler; cf.
+       * lp_build_create_jit_compiler_for_module().
+       * If you want to disable Altivec code generation, the best place to do it is here.
+       */
+      char *env_control = getenv("GALLIVM_ALTIVEC");    /* 1=enable (default); 0=disable */
+      if (env_control && env_control[0] == '0') {
+         enable_altivec = FALSE;
+      }
+#endif
+      /* VSX instructions can be explicitly enabled/disabled via GALLIVM_VSX=1 or 0 */
+      char *env_vsx = getenv("GALLIVM_VSX");
+      if (env_vsx && env_vsx[0] == '0') {
+         enable_vsx = FALSE;
+      }
+      if (enable_altivec) {
+         __lv_powerpc_canjump = 1;
+
+         __asm __volatile
+            ("mtspr 256, %0\n\t"
+             "vand %%v0, %%v0, %%v0"
+             :
+             : "r" (-1));
+
+         util_cpu_caps.has_altivec = 1;
+
+         if (enable_vsx) {
+            __asm __volatile("xxland %vs0, %vs0, %vs0");
+            util_cpu_caps.has_vsx = 1;
+         }
+         signal(SIGILL, SIG_DFL);
+      } else {
+         util_cpu_caps.has_altivec = 0;
+      }
+   }
+#endif /* !PIPE_OS_APPLE */
+}
+#endif /* PIPE_ARCH_PPC */
+
+
+#if defined(PIPE_ARCH_X86) || defined (PIPE_ARCH_X86_64)
+static int has_cpuid(void)
+{
+#if defined(PIPE_ARCH_X86)
+#if defined(PIPE_OS_GCC)
+   int a, c;
+
+   __asm __volatile
+      ("pushf\n"
+       "popl %0\n"
+       "movl %0, %1\n"
+       "xorl $0x200000, %0\n"
+       "push %0\n"
+       "popf\n"
+       "pushf\n"
+       "popl %0\n"
+       : "=a" (a), "=c" (c)
+       :
+       : "cc");
+
+   return a != c;
+#else
+   /* FIXME */
+   return 1;
+#endif
+#elif defined(PIPE_ARCH_X86_64)
+   return 1;
+#else
+   return 0;
+#endif
+}
+
+
+/**
+ * @sa cpuid.h included in gcc-4.3 onwards.
+ * @sa http://msdn.microsoft.com/en-us/library/hskdteyh.aspx
+ */
+static inline void
+cpuid(uint32_t ax, uint32_t *p)
+{
+#if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86)
+   __asm __volatile (
+     "xchgl %%ebx, %1\n\t"
+     "cpuid\n\t"
+     "xchgl %%ebx, %1"
+     : "=a" (p[0]),
+       "=S" (p[1]),
+       "=c" (p[2]),
+       "=d" (p[3])
+     : "0" (ax)
+   );
+#elif defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86_64)
+   __asm __volatile (
+     "cpuid\n\t"
+     : "=a" (p[0]),
+       "=b" (p[1]),
+       "=c" (p[2]),
+       "=d" (p[3])
+     : "0" (ax)
+   );
+#elif defined(PIPE_CC_MSVC)
+   __cpuid(p, ax);
+#else
+   p[0] = 0;
+   p[1] = 0;
+   p[2] = 0;
+   p[3] = 0;
+#endif
+}
+
+/**
+ * @sa cpuid.h included in gcc-4.4 onwards.
+ * @sa http://msdn.microsoft.com/en-us/library/hskdteyh%28v=vs.90%29.aspx
+ */
+static inline void
+cpuid_count(uint32_t ax, uint32_t cx, uint32_t *p)
+{
+#if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86)
+   __asm __volatile (
+     "xchgl %%ebx, %1\n\t"
+     "cpuid\n\t"
+     "xchgl %%ebx, %1"
+     : "=a" (p[0]),
+       "=S" (p[1]),
+       "=c" (p[2]),
+       "=d" (p[3])
+     : "0" (ax), "2" (cx)
+   );
+#elif defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86_64)
+   __asm __volatile (
+     "cpuid\n\t"
+     : "=a" (p[0]),
+       "=b" (p[1]),
+       "=c" (p[2]),
+       "=d" (p[3])
+     : "0" (ax), "2" (cx)
+   );
+#elif defined(PIPE_CC_MSVC)
+   __cpuidex(p, ax, cx);
+#else
+   p[0] = 0;
+   p[1] = 0;
+   p[2] = 0;
+   p[3] = 0;
+#endif
+}
+
+
+static inline uint64_t xgetbv(void)
+{
+#if defined(PIPE_CC_GCC)
+   uint32_t eax, edx;
+
+   __asm __volatile (
+     ".byte 0x0f, 0x01, 0xd0" // xgetbv isn't supported on gcc < 4.4
+     : "=a"(eax),
+       "=d"(edx)
+     : "c"(0)
+   );
+
+   return ((uint64_t)edx << 32) | eax;
+#elif defined(PIPE_CC_MSVC) && defined(_MSC_FULL_VER) && defined(_XCR_XFEATURE_ENABLED_MASK)
+   return _xgetbv(_XCR_XFEATURE_ENABLED_MASK);
+#else
+   return 0;
+#endif
+}
+
+
+#if defined(PIPE_ARCH_X86)
+PIPE_ALIGN_STACK static inline boolean sse2_has_daz(void)
+{
+   struct {
+      uint32_t pad1[7];
+      uint32_t mxcsr_mask;
+      uint32_t pad2[128-8];
+   } PIPE_ALIGN_VAR(16) fxarea;
+
+   fxarea.mxcsr_mask = 0;
+#if defined(PIPE_CC_GCC)
+   __asm __volatile ("fxsave %0" : "+m" (fxarea));
+#elif defined(PIPE_CC_MSVC) || defined(PIPE_CC_ICL)
+   _fxsave(&fxarea);
+#else
+   fxarea.mxcsr_mask = 0;
+#endif
+   return !!(fxarea.mxcsr_mask & (1 << 6));
+}
+#endif
+
+#endif /* X86 or X86_64 */
+
+#if defined(PIPE_ARCH_ARM)
+static void
+check_os_arm_support(void)
+{
+   /*
+    * On Android, the cpufeatures library is preferred way of checking
+    * CPU capabilities. However, it is not available for standalone Mesa
+    * builds, i.e. when Android build system (Android.mk-based) is not
+    * used. Because of this we cannot use PIPE_OS_ANDROID here, but rather
+    * have a separate macro that only gets enabled from respective Android.mk.
+    */
+#if defined(HAS_ANDROID_CPUFEATURES)
+   AndroidCpuFamily cpu_family = android_getCpuFamily();
+   uint64_t cpu_features = android_getCpuFeatures();
+
+   if (cpu_family == ANDROID_CPU_FAMILY_ARM) {
+      if (cpu_features & ANDROID_CPU_ARM_FEATURE_NEON)
+         util_cpu_caps.has_neon = 1;
+   }
+#elif defined(PIPE_OS_LINUX)
+    Elf32_auxv_t aux;
+    int fd;
+
+    fd = open("/proc/self/auxv", O_RDONLY | O_CLOEXEC);
+    if (fd >= 0) {
+       while (read(fd, &aux, sizeof(Elf32_auxv_t)) == sizeof(Elf32_auxv_t)) {
+          if (aux.a_type == AT_HWCAP) {
+             uint32_t hwcap = aux.a_un.a_val;
+
+             util_cpu_caps.has_neon = (hwcap >> 12) & 1;
+             break;
+          }
+       }
+       close (fd);
+    }
+#endif /* PIPE_OS_LINUX */
+}
+#endif /* PIPE_ARCH_ARM */
+
+static void
+get_cpu_topology(void)
+{
+   uint32_t regs[4];
+
+   /* Default. This is correct if L3 is not present or there is only one. */
+   util_cpu_caps.cores_per_L3 = util_cpu_caps.nr_cpus;
+
+#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
+   /* AMD Zen */
+   if (util_cpu_caps.x86_cpu_type == 0x17) {
+      /* Query the L3 cache topology information. */
+      cpuid_count(0x8000001D, 3, regs);
+      unsigned cache_level = (regs[0] >> 5) & 0x7;
+      unsigned cores_per_cache = ((regs[0] >> 14) & 0xfff) + 1;
+
+      if (cache_level == 3)
+         util_cpu_caps.cores_per_L3 = cores_per_cache;
+   }
+#endif
+}
+
+static void
+util_cpu_detect_once(void)
+{
+   memset(&util_cpu_caps, 0, sizeof util_cpu_caps);
+
+   /* Count the number of CPUs in system */
+#if defined(PIPE_OS_WINDOWS)
+   {
+      SYSTEM_INFO system_info;
+      GetSystemInfo(&system_info);
+      util_cpu_caps.nr_cpus = system_info.dwNumberOfProcessors;
+   }
+#elif defined(PIPE_OS_UNIX) && defined(_SC_NPROCESSORS_ONLN)
+   util_cpu_caps.nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
+   if (util_cpu_caps.nr_cpus == ~0)
+      util_cpu_caps.nr_cpus = 1;
+#elif defined(PIPE_OS_BSD)
+   {
+      int mib[2], ncpu;
+      int len;
+
+      mib[0] = CTL_HW;
+      mib[1] = HW_NCPU;
+
+      len = sizeof (ncpu);
+      sysctl(mib, 2, &ncpu, &len, NULL, 0);
+      util_cpu_caps.nr_cpus = ncpu;
+   }
+#else
+   util_cpu_caps.nr_cpus = 1;
+#endif
+
+   /* Make the fallback cacheline size nonzero so that it can be
+    * safely passed to align().
+    */
+   util_cpu_caps.cacheline = sizeof(void *);
+
+#if defined(PIPE_ARCH_X86) || defined(PIPE_ARCH_X86_64)
+   if (has_cpuid()) {
+      uint32_t regs[4];
+      uint32_t regs2[4];
+
+      util_cpu_caps.cacheline = 32;
+
+      /* Get max cpuid level */
+      cpuid(0x00000000, regs);
+
+      if (regs[0] >= 0x00000001) {
+         unsigned int cacheline;
+
+         cpuid (0x00000001, regs2);
+
+         util_cpu_caps.x86_cpu_type = (regs2[0] >> 8) & 0xf;
+         /* Add "extended family". */
+         if (util_cpu_caps.x86_cpu_type == 0xf)
+             util_cpu_caps.x86_cpu_type += ((regs2[0] >> 20) & 0xff);
+
+         /* general feature flags */
+         util_cpu_caps.has_tsc    = (regs2[3] >>  4) & 1; /* 0x0000010 */
+         util_cpu_caps.has_mmx    = (regs2[3] >> 23) & 1; /* 0x0800000 */
+         util_cpu_caps.has_sse    = (regs2[3] >> 25) & 1; /* 0x2000000 */
+         util_cpu_caps.has_sse2   = (regs2[3] >> 26) & 1; /* 0x4000000 */
+         util_cpu_caps.has_sse3   = (regs2[2] >>  0) & 1; /* 0x0000001 */
+         util_cpu_caps.has_ssse3  = (regs2[2] >>  9) & 1; /* 0x0000020 */
+         util_cpu_caps.has_sse4_1 = (regs2[2] >> 19) & 1;
+         util_cpu_caps.has_sse4_2 = (regs2[2] >> 20) & 1;
+         util_cpu_caps.has_popcnt = (regs2[2] >> 23) & 1;
+         util_cpu_caps.has_avx    = ((regs2[2] >> 28) & 1) && // AVX
+                                    ((regs2[2] >> 27) & 1) && // OSXSAVE
+                                    ((xgetbv() & 6) == 6);    // XMM & YMM
+         util_cpu_caps.has_f16c   = ((regs2[2] >> 29) & 1) && util_cpu_caps.has_avx;
+         util_cpu_caps.has_fma    = ((regs2[2] >> 12) & 1) && util_cpu_caps.has_avx;
+         util_cpu_caps.has_mmx2   = util_cpu_caps.has_sse; /* SSE cpus supports mmxext too */
+#if defined(PIPE_ARCH_X86_64)
+         util_cpu_caps.has_daz = 1;
+#else
+         util_cpu_caps.has_daz = util_cpu_caps.has_sse3 ||
+            (util_cpu_caps.has_sse2 && sse2_has_daz());
+#endif
+
+         cacheline = ((regs2[1] >> 8) & 0xFF) * 8;
+         if (cacheline > 0)
+            util_cpu_caps.cacheline = cacheline;
+      }
+      if (util_cpu_caps.has_avx && regs[0] >= 0x00000007) {
+         uint32_t regs7[4];
+         cpuid_count(0x00000007, 0x00000000, regs7);
+         util_cpu_caps.has_avx2 = (regs7[1] >> 5) & 1;
+      }
+
+      // check for avx512
+      if (((regs2[2] >> 27) & 1) && // OSXSAVE
+          (xgetbv() & (0x7 << 5)) && // OPMASK: upper-256 enabled by OS
+          ((xgetbv() & 6) == 6)) { // XMM/YMM enabled by OS
+         uint32_t regs3[4];
+         cpuid_count(0x00000007, 0x00000000, regs3);
+         util_cpu_caps.has_avx512f    = (regs3[1] >> 16) & 1;
+         util_cpu_caps.has_avx512dq   = (regs3[1] >> 17) & 1;
+         util_cpu_caps.has_avx512ifma = (regs3[1] >> 21) & 1;
+         util_cpu_caps.has_avx512pf   = (regs3[1] >> 26) & 1;
+         util_cpu_caps.has_avx512er   = (regs3[1] >> 27) & 1;
+         util_cpu_caps.has_avx512cd   = (regs3[1] >> 28) & 1;
+         util_cpu_caps.has_avx512bw   = (regs3[1] >> 30) & 1;
+         util_cpu_caps.has_avx512vl   = (regs3[1] >> 31) & 1;
+         util_cpu_caps.has_avx512vbmi = (regs3[2] >>  1) & 1;
+      }
+
+      if (regs[1] == 0x756e6547 && regs[2] == 0x6c65746e && regs[3] == 0x49656e69) {
+         /* GenuineIntel */
+         util_cpu_caps.has_intel = 1;
+      }
+
+      cpuid(0x80000000, regs);
+
+      if (regs[0] >= 0x80000001) {
+
+         cpuid(0x80000001, regs2);
+
+         util_cpu_caps.has_mmx  |= (regs2[3] >> 23) & 1;
+         util_cpu_caps.has_mmx2 |= (regs2[3] >> 22) & 1;
+         util_cpu_caps.has_3dnow = (regs2[3] >> 31) & 1;
+         util_cpu_caps.has_3dnow_ext = (regs2[3] >> 30) & 1;
+
+         util_cpu_caps.has_xop = util_cpu_caps.has_avx &&
+                                 ((regs2[2] >> 11) & 1);
+      }
+
+      if (regs[0] >= 0x80000006) {
+         /* should we really do this if the clflush size above worked? */
+         unsigned int cacheline;
+         cpuid(0x80000006, regs2);
+         cacheline = regs2[2] & 0xFF;
+         if (cacheline > 0)
+            util_cpu_caps.cacheline = cacheline;
+      }
+
+      if (!util_cpu_caps.has_sse) {
+         util_cpu_caps.has_sse2 = 0;
+         util_cpu_caps.has_sse3 = 0;
+         util_cpu_caps.has_ssse3 = 0;
+         util_cpu_caps.has_sse4_1 = 0;
+      }
+   }
+#endif /* PIPE_ARCH_X86 || PIPE_ARCH_X86_64 */
+
+#if defined(PIPE_ARCH_ARM)
+   check_os_arm_support();
+#endif
+
+#if defined(PIPE_ARCH_PPC)
+   check_os_altivec_support();
+#endif /* PIPE_ARCH_PPC */
+
+   get_cpu_topology();
+
+#ifdef DEBUG
+   if (debug_get_option_dump_cpu()) {
+      debug_printf("util_cpu_caps.nr_cpus = %u\n", util_cpu_caps.nr_cpus);
+
+      debug_printf("util_cpu_caps.x86_cpu_type = %u\n", util_cpu_caps.x86_cpu_type);
+      debug_printf("util_cpu_caps.cacheline = %u\n", util_cpu_caps.cacheline);
+
+      debug_printf("util_cpu_caps.has_tsc = %u\n", util_cpu_caps.has_tsc);
+      debug_printf("util_cpu_caps.has_mmx = %u\n", util_cpu_caps.has_mmx);
+      debug_printf("util_cpu_caps.has_mmx2 = %u\n", util_cpu_caps.has_mmx2);
+      debug_printf("util_cpu_caps.has_sse = %u\n", util_cpu_caps.has_sse);
+      debug_printf("util_cpu_caps.has_sse2 = %u\n", util_cpu_caps.has_sse2);
+      debug_printf("util_cpu_caps.has_sse3 = %u\n", util_cpu_caps.has_sse3);
+      debug_printf("util_cpu_caps.has_ssse3 = %u\n", util_cpu_caps.has_ssse3);
+      debug_printf("util_cpu_caps.has_sse4_1 = %u\n", util_cpu_caps.has_sse4_1);
+      debug_printf("util_cpu_caps.has_sse4_2 = %u\n", util_cpu_caps.has_sse4_2);
+      debug_printf("util_cpu_caps.has_avx = %u\n", util_cpu_caps.has_avx);
+      debug_printf("util_cpu_caps.has_avx2 = %u\n", util_cpu_caps.has_avx2);
+      debug_printf("util_cpu_caps.has_f16c = %u\n", util_cpu_caps.has_f16c);
+      debug_printf("util_cpu_caps.has_popcnt = %u\n", util_cpu_caps.has_popcnt);
+      debug_printf("util_cpu_caps.has_3dnow = %u\n", util_cpu_caps.has_3dnow);
+      debug_printf("util_cpu_caps.has_3dnow_ext = %u\n", util_cpu_caps.has_3dnow_ext);
+      debug_printf("util_cpu_caps.has_xop = %u\n", util_cpu_caps.has_xop);
+      debug_printf("util_cpu_caps.has_altivec = %u\n", util_cpu_caps.has_altivec);
+      debug_printf("util_cpu_caps.has_vsx = %u\n", util_cpu_caps.has_vsx);
+      debug_printf("util_cpu_caps.has_neon = %u\n", util_cpu_caps.has_neon);
+      debug_printf("util_cpu_caps.has_daz = %u\n", util_cpu_caps.has_daz);
+      debug_printf("util_cpu_caps.has_avx512f = %u\n", util_cpu_caps.has_avx512f);
+      debug_printf("util_cpu_caps.has_avx512dq = %u\n", util_cpu_caps.has_avx512dq);
+      debug_printf("util_cpu_caps.has_avx512ifma = %u\n", util_cpu_caps.has_avx512ifma);
+      debug_printf("util_cpu_caps.has_avx512pf = %u\n", util_cpu_caps.has_avx512pf);
+      debug_printf("util_cpu_caps.has_avx512er = %u\n", util_cpu_caps.has_avx512er);
+      debug_printf("util_cpu_caps.has_avx512cd = %u\n", util_cpu_caps.has_avx512cd);
+      debug_printf("util_cpu_caps.has_avx512bw = %u\n", util_cpu_caps.has_avx512bw);
+      debug_printf("util_cpu_caps.has_avx512vl = %u\n", util_cpu_caps.has_avx512vl);
+      debug_printf("util_cpu_caps.has_avx512vbmi = %u\n", util_cpu_caps.has_avx512vbmi);
+   }
+#endif
+}
+
+static once_flag cpu_once_flag = ONCE_FLAG_INIT;
+
+void
+util_cpu_detect(void)
+{
+   call_once(&cpu_once_flag, util_cpu_detect_once);
+}
diff --git a/lib/mesa/src/util/u_cpu_detect.h b/lib/mesa/src/util/u_cpu_detect.h
new file mode 100644
index 000000000..efc910d14
--- /dev/null
+++ b/lib/mesa/src/util/u_cpu_detect.h
@@ -0,0 +1,101 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 Dennis Smit
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * on the rights to use, copy, modify, merge, publish, distribute, sub
+ * license, and/or sell copies of the Software, and to permit persons to whom
+ * the Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.  IN NO EVENT SHALL
+ * AUTHORS, COPYRIGHT HOLDERS, AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ ***************************************************************************/
+
+/**
+ * @file
+ * CPU feature detection.
+ *
+ * @author Dennis Smit
+ * @author Based on the work of Eric Anholt <anholt@FreeBSD.org>
+ */
+
+#ifndef _UTIL_CPU_DETECT_H
+#define _UTIL_CPU_DETECT_H
+
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_config.h"
+
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+
+struct util_cpu_caps {
+   int nr_cpus;
+
+   /* Feature flags */
+   int x86_cpu_type;
+   unsigned cacheline;
+   unsigned cores_per_L3;
+
+   unsigned has_intel:1;
+   unsigned has_tsc:1;
+   unsigned has_mmx:1;
+   unsigned has_mmx2:1;
+   unsigned has_sse:1;
+   unsigned has_sse2:1;
+   unsigned has_sse3:1;
+   unsigned has_ssse3:1;
+   unsigned has_sse4_1:1;
+   unsigned has_sse4_2:1;
+   unsigned has_popcnt:1;
+   unsigned has_avx:1;
+   unsigned has_avx2:1;
+   unsigned has_f16c:1;
+   unsigned has_fma:1;
+   unsigned has_3dnow:1;
+   unsigned has_3dnow_ext:1;
+   unsigned has_xop:1;
+   unsigned has_altivec:1;
+   unsigned has_vsx:1;
+   unsigned has_daz:1;
+   unsigned has_neon:1;
+
+   unsigned has_avx512f:1;
+   unsigned has_avx512dq:1;
+   unsigned has_avx512ifma:1;
+   unsigned has_avx512pf:1;
+   unsigned has_avx512er:1;
+   unsigned has_avx512cd:1;
+   unsigned has_avx512bw:1;
+   unsigned has_avx512vl:1;
+   unsigned has_avx512vbmi:1;
+};
+
+extern struct util_cpu_caps
+util_cpu_caps;
+
+void util_cpu_detect(void);
+
+
+#ifdef	__cplusplus
+}
+#endif
+
+
+#endif /* _UTIL_CPU_DETECT_H */
diff --git a/lib/mesa/src/util/u_debug.c b/lib/mesa/src/util/u_debug.c
new file mode 100644
index 000000000..f6ed0138c
--- /dev/null
+++ b/lib/mesa/src/util/u_debug.c
@@ -0,0 +1,441 @@
+/**************************************************************************
+ *
+ * Copyright 2008 VMware, Inc.
+ * Copyright (c) 2008 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+
+#include "pipe/p_config.h"
+
+#include "pipe/p_compiler.h"
+#include "util/u_debug.h"
+#include "pipe/p_format.h"
+#include "pipe/p_state.h"
+#include "util/u_string.h"
+#include "util/u_math.h"
+#include <inttypes.h>
+
+#include <stdio.h>
+#include <limits.h> /* CHAR_BIT */
+#include <ctype.h> /* isalnum */
+
+#ifdef _WIN32
+#include <windows.h>
+#include <stdlib.h>
+#endif
+
+
+void
+_debug_vprintf(const char *format, va_list ap)
+{
+   static char buf[4096] = {'\0'};
+#if defined(PIPE_OS_WINDOWS) || defined(PIPE_SUBSYSTEM_EMBEDDED)
+   /* We buffer until we find a newline. */
+   size_t len = strlen(buf);
+   int ret = util_vsnprintf(buf + len, sizeof(buf) - len, format, ap);
+   if (ret > (int)(sizeof(buf) - len - 1) || util_strchr(buf + len, '\n')) {
+      os_log_message(buf);
+      buf[0] = '\0';
+   }
+#else
+   util_vsnprintf(buf, sizeof(buf), format, ap);
+   os_log_message(buf);
+#endif
+}
+
+
+void
+_pipe_debug_message(struct pipe_debug_callback *cb,
+                    unsigned *id,
+                    enum pipe_debug_type type,
+                    const char *fmt, ...)
+{
+   va_list args;
+   va_start(args, fmt);
+   if (cb && cb->debug_message)
+      cb->debug_message(cb->data, id, type, fmt, args);
+   va_end(args);
+}
+
+
+void
+debug_disable_error_message_boxes(void)
+{
+#ifdef _WIN32
+   /* When Windows' error message boxes are disabled for this process (as is
+    * typically the case when running tests in an automated fashion) we disable
+    * CRT message boxes too.
+    */
+   UINT uMode = SetErrorMode(0);
+   SetErrorMode(uMode);
+   if (uMode & SEM_FAILCRITICALERRORS) {
+      /* Disable assertion failure message box.
+       * http://msdn.microsoft.com/en-us/library/sas1dkb2.aspx
+       */
+      _set_error_mode(_OUT_TO_STDERR);
+#ifdef _MSC_VER
+      /* Disable abort message box.
+       * http://msdn.microsoft.com/en-us/library/e631wekh.aspx
+       */
+      _set_abort_behavior(0, _WRITE_ABORT_MSG | _CALL_REPORTFAULT);
+#endif
+   }
+#endif /* _WIN32 */
+}
+
+
+#ifdef DEBUG
+void
+debug_print_blob(const char *name, const void *blob, unsigned size)
+{
+   const unsigned *ublob = (const unsigned *)blob;
+   unsigned i;
+
+   debug_printf("%s (%d dwords%s)\n", name, size/4,
+                size%4 ? "... plus a few bytes" : "");
+
+   for (i = 0; i < size/4; i++) {
+      debug_printf("%d:\t%08x\n", i, ublob[i]);
+   }
+}
+#endif
+
+
+static boolean
+debug_get_option_should_print(void)
+{
+   static boolean first = TRUE;
+   static boolean value = FALSE;
+
+   if (!first)
+      return value;
+
+   /* Oh hey this will call into this function,
+    * but its cool since we set first to false
+    */
+   first = FALSE;
+   value = debug_get_bool_option("GALLIUM_PRINT_OPTIONS", FALSE);
+   /* XXX should we print this option? Currently it wont */
+   return value;
+}
+
+
+const char *
+debug_get_option(const char *name, const char *dfault)
+{
+   const char *result;
+
+   result = os_get_option(name);
+   if (!result)
+      result = dfault;
+
+   if (debug_get_option_should_print())
+      debug_printf("%s: %s = %s\n", __FUNCTION__, name,
+                   result ? result : "(null)");
+
+   return result;
+}
+
+
+boolean
+debug_get_bool_option(const char *name, boolean dfault)
+{
+   const char *str = os_get_option(name);
+   boolean result;
+
+   if (str == NULL)
+      result = dfault;
+   else if (!util_strcmp(str, "n"))
+      result = FALSE;
+   else if (!util_strcmp(str, "no"))
+      result = FALSE;
+   else if (!util_strcmp(str, "0"))
+      result = FALSE;
+   else if (!util_strcmp(str, "f"))
+      result = FALSE;
+   else if (!util_strcmp(str, "F"))
+      result = FALSE;
+   else if (!util_strcmp(str, "false"))
+      result = FALSE;
+   else if (!util_strcmp(str, "FALSE"))
+      result = FALSE;
+   else
+      result = TRUE;
+
+   if (debug_get_option_should_print())
+      debug_printf("%s: %s = %s\n", __FUNCTION__, name,
+                   result ? "TRUE" : "FALSE");
+
+   return result;
+}
+
+
+long
+debug_get_num_option(const char *name, long dfault)
+{
+   long result;
+   const char *str;
+
+   str = os_get_option(name);
+   if (!str) {
+      result = dfault;
+   } else {
+      char *endptr;
+
+      result = strtol(str, &endptr, 0);
+      if (str == endptr) {
+         /* Restore the default value when no digits were found. */
+         result = dfault;
+      }
+   }
+
+   if (debug_get_option_should_print())
+      debug_printf("%s: %s = %li\n", __FUNCTION__, name, result);
+
+   return result;
+}
+
+
+static boolean
+str_has_option(const char *str, const char *name)
+{
+   /* Empty string. */
+   if (!*str) {
+      return FALSE;
+   }
+
+   /* OPTION=all */
+   if (!util_strcmp(str, "all")) {
+      return TRUE;
+   }
+
+   /* Find 'name' in 'str' surrounded by non-alphanumeric characters. */
+   {
+      const char *start = str;
+      unsigned name_len = strlen(name);
+
+      /* 'start' is the beginning of the currently-parsed word,
+       * we increment 'str' each iteration.
+       * if we find either the end of string or a non-alphanumeric character,
+       * we compare 'start' up to 'str-1' with 'name'. */
+
+      while (1) {
+         if (!*str || !(isalnum(*str) || *str == '_')) {
+            if (str-start == name_len &&
+                !memcmp(start, name, name_len)) {
+               return TRUE;
+            }
+
+            if (!*str) {
+               return FALSE;
+            }
+
+            start = str+1;
+         }
+
+         str++;
+      }
+   }
+
+   return FALSE;
+}
+
+
+uint64_t
+debug_get_flags_option(const char *name,
+                       const struct debug_named_value *flags,
+                       uint64_t dfault)
+{
+   uint64_t result;
+   const char *str;
+   const struct debug_named_value *orig = flags;
+   unsigned namealign = 0;
+
+   str = os_get_option(name);
+   if (!str)
+      result = dfault;
+   else if (!util_strcmp(str, "help")) {
+      result = dfault;
+      _debug_printf("%s: help for %s:\n", __FUNCTION__, name);
+      for (; flags->name; ++flags)
+         namealign = MAX2(namealign, strlen(flags->name));
+      for (flags = orig; flags->name; ++flags)
+         _debug_printf("| %*s [0x%0*"PRIx64"]%s%s\n", namealign, flags->name,
+                      (int)sizeof(uint64_t)*CHAR_BIT/4, flags->value,
+                      flags->desc ? " " : "", flags->desc ? flags->desc : "");
+   }
+   else {
+      result = 0;
+      while (flags->name) {
+	 if (str_has_option(str, flags->name))
+	    result |= flags->value;
+	 ++flags;
+      }
+   }
+
+   if (debug_get_option_should_print()) {
+      if (str) {
+         debug_printf("%s: %s = 0x%"PRIx64" (%s)\n",
+                      __FUNCTION__, name, result, str);
+      } else {
+         debug_printf("%s: %s = 0x%"PRIx64"\n", __FUNCTION__, name, result);
+      }
+   }
+
+   return result;
+}
+
+
+void
+_debug_assert_fail(const char *expr, const char *file, unsigned line,
+                   const char *function)
+{
+   _debug_printf("%s:%u:%s: Assertion `%s' failed.\n",
+                 file, line, function, expr);
+   os_abort();
+}
+
+
+const char *
+debug_dump_enum(const struct debug_named_value *names,
+                unsigned long value)
+{
+   static char rest[64];
+
+   while (names->name) {
+      if (names->value == value)
+	 return names->name;
+      ++names;
+   }
+
+   util_snprintf(rest, sizeof(rest), "0x%08lx", value);
+   return rest;
+}
+
+
+const char *
+debug_dump_enum_noprefix(const struct debug_named_value *names,
+                         const char *prefix,
+                         unsigned long value)
+{
+   static char rest[64];
+
+   while (names->name) {
+      if (names->value == value) {
+         const char *name = names->name;
+         while (*name == *prefix) {
+            name++;
+            prefix++;
+         }
+         return name;
+      }
+      ++names;
+   }
+
+   util_snprintf(rest, sizeof(rest), "0x%08lx", value);
+   return rest;
+}
+
+
+const char *
+debug_dump_flags(const struct debug_named_value *names, unsigned long value)
+{
+   static char output[4096];
+   static char rest[256];
+   int first = 1;
+
+   output[0] = '\0';
+
+   while (names->name) {
+      if ((names->value & value) == names->value) {
+	 if (!first)
+	    util_strncat(output, "|", sizeof(output) - strlen(output) - 1);
+	 else
+	    first = 0;
+	 util_strncat(output, names->name, sizeof(output) - strlen(output) - 1);
+	 output[sizeof(output) - 1] = '\0';
+	 value &= ~names->value;
+      }
+      ++names;
+   }
+
+   if (value) {
+      if (!first)
+	 util_strncat(output, "|", sizeof(output) - strlen(output) - 1);
+      else
+	 first = 0;
+
+      util_snprintf(rest, sizeof(rest), "0x%08lx", value);
+      util_strncat(output, rest, sizeof(output) - strlen(output) - 1);
+      output[sizeof(output) - 1] = '\0';
+   }
+
+   if (first)
+      return "0";
+
+   return output;
+}
+
+
+
+#ifdef DEBUG
+int fl_indent = 0;
+const char* fl_function[1024];
+
+int
+debug_funclog_enter(const char* f, UNUSED const int line,
+                    UNUSED const char* file)
+{
+   int i;
+
+   for (i = 0; i < fl_indent; i++)
+      debug_printf("  ");
+   debug_printf("%s\n", f);
+
+   assert(fl_indent < 1023);
+   fl_function[fl_indent++] = f;
+
+   return 0;
+}
+
+void
+debug_funclog_exit(const char* f, UNUSED const int line,
+                   UNUSED const char* file)
+{
+   --fl_indent;
+   assert(fl_indent >= 0);
+   assert(fl_function[fl_indent] == f);
+}
+
+void
+debug_funclog_enter_exit(const char* f, UNUSED const int line,
+                         UNUSED const char* file)
+{
+   int i;
+   for (i = 0; i < fl_indent; i++)
+      debug_printf("  ");
+   debug_printf("%s\n", f);
+}
+#endif
diff --git a/lib/mesa/src/util/u_debug.h b/lib/mesa/src/util/u_debug.h
new file mode 100644
index 000000000..b3505caeb
--- /dev/null
+++ b/lib/mesa/src/util/u_debug.h
@@ -0,0 +1,460 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 VMware, Inc.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+/**
+ * @file
+ * Cross-platform debugging helpers.
+ * 
+ * For now it just has assert and printf replacements, but it might be extended 
+ * with stack trace reports and more advanced logging in the near future. 
+ * 
+ * @author Jose Fonseca <jfonseca@vmware.com>
+ */
+
+#ifndef U_DEBUG_H_
+#define U_DEBUG_H_
+
+
+#include "util/os_misc.h"
+
+#if defined(PIPE_OS_HAIKU)
+/* Haiku provides debug_printf in libroot with OS.h */
+#include <OS.h>
+#endif
+
+#include "pipe/p_defines.h"
+
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+
+#if defined(__GNUC__)
+#define _util_printf_format(fmt, list) __attribute__ ((format (printf, fmt, list)))
+#else
+#define _util_printf_format(fmt, list)
+#endif
+
+void _debug_vprintf(const char *format, va_list ap);
+   
+
+static inline void
+_debug_printf(const char *format, ...)
+{
+   va_list ap;
+   va_start(ap, format);
+   _debug_vprintf(format, ap);
+   va_end(ap);
+}
+
+
+/**
+ * Print debug messages.
+ *
+ * The actual channel used to output debug message is platform specific. To
+ * avoid misformating or truncation, follow these rules of thumb:
+ * - output whole lines
+ * - avoid outputing large strings (512 bytes is the current maximum length
+ * that is guaranteed to be printed in all platforms)
+ */
+#if !defined(PIPE_OS_HAIKU)
+static inline void
+debug_printf(const char *format, ...) _util_printf_format(1,2);
+
+static inline void
+debug_printf(const char *format, ...)
+{
+#ifdef DEBUG
+   va_list ap;
+   va_start(ap, format);
+   _debug_vprintf(format, ap);
+   va_end(ap);
+#else
+   (void) format; /* silence warning */
+#endif
+}
+#endif
+
+
+/*
+ * ... isn't portable so we need to pass arguments in parentheses.
+ *
+ * usage:
+ *    debug_printf_once(("answer: %i\n", 42));
+ */
+#define debug_printf_once(args) \
+   do { \
+      static boolean once = TRUE; \
+      if (once) { \
+         once = FALSE; \
+         debug_printf args; \
+      } \
+   } while (0)
+
+
+#ifdef DEBUG
+#define debug_vprintf(_format, _ap) _debug_vprintf(_format, _ap)
+#else
+#define debug_vprintf(_format, _ap) ((void)0)
+#endif
+
+
+#ifdef DEBUG
+/**
+ * Dump a blob in hex to the same place that debug_printf sends its
+ * messages.
+ */
+void debug_print_blob( const char *name, const void *blob, unsigned size );
+#else
+#define debug_print_blob(_name, _blob, _size) ((void)0)
+#endif
+
+
+/**
+ * Disable interactive error message boxes.
+ *
+ * Should be called as soon as possible for effectiveness.
+ */
+void
+debug_disable_error_message_boxes(void);
+
+
+/**
+ * Hard-coded breakpoint.
+ */
+#ifdef DEBUG
+#define debug_break() os_break()
+#else /* !DEBUG */
+#define debug_break() ((void)0)
+#endif /* !DEBUG */
+
+
+long
+debug_get_num_option(const char *name, long dfault);
+
+#ifdef _MSC_VER
+__declspec(noreturn)
+#endif
+void _debug_assert_fail(const char *expr, 
+                        const char *file, 
+                        unsigned line, 
+                        const char *function)
+#if defined(__GNUC__) && !defined(DEBUG)
+   __attribute__((noreturn))
+#endif
+;
+
+
+/** 
+ * Assert macro
+ * 
+ * Do not expect that the assert call terminates -- errors must be handled 
+ * regardless of assert behavior.
+ *
+ * For non debug builds the assert macro will expand to a no-op, so do not
+ * call functions with side effects in the assert expression.
+ */
+#ifndef NDEBUG
+#define debug_assert(expr) ((expr) ? (void)0 : _debug_assert_fail(#expr, __FILE__, __LINE__, __FUNCTION__))
+#else
+#define debug_assert(expr) (void)(0 && (expr))
+#endif
+
+
+/** Override standard assert macro */
+#ifdef assert
+#undef assert
+#endif
+#define assert(expr) debug_assert(expr)
+
+
+/**
+ * Output the current function name.
+ */
+#ifdef DEBUG
+#define debug_checkpoint() \
+   _debug_printf("%s\n", __FUNCTION__)
+#else
+#define debug_checkpoint() \
+   ((void)0) 
+#endif
+
+
+/**
+ * Output the full source code position.
+ */
+#ifdef DEBUG
+#define debug_checkpoint_full() \
+   _debug_printf("%s:%u:%s\n", __FILE__, __LINE__, __FUNCTION__)
+#else
+#define debug_checkpoint_full() \
+   ((void)0) 
+#endif
+
+
+/**
+ * Output a warning message. Muted on release version.
+ */
+#ifdef DEBUG
+#define debug_warning(__msg) \
+   _debug_printf("%s:%u:%s: warning: %s\n", __FILE__, __LINE__, __FUNCTION__, __msg)
+#else
+#define debug_warning(__msg) \
+   ((void)0) 
+#endif
+
+
+/**
+ * Emit a warning message, but only once.
+ */
+#ifdef DEBUG
+#define debug_warn_once(__msg) \
+   do { \
+      static bool warned = FALSE; \
+      if (!warned) { \
+         _debug_printf("%s:%u:%s: one time warning: %s\n", \
+                       __FILE__, __LINE__, __FUNCTION__, __msg); \
+         warned = TRUE; \
+      } \
+   } while (0)
+#else
+#define debug_warn_once(__msg) \
+   ((void)0) 
+#endif
+
+
+/**
+ * Output an error message. Not muted on release version.
+ */
+#ifdef DEBUG
+#define debug_error(__msg) \
+   _debug_printf("%s:%u:%s: error: %s\n", __FILE__, __LINE__, __FUNCTION__, __msg) 
+#else
+#define debug_error(__msg) \
+   _debug_printf("error: %s\n", __msg)
+#endif
+
+/**
+ * Output a debug log message to the debug info callback.
+ */
+#define pipe_debug_message(cb, type, fmt, ...) do { \
+   static unsigned id = 0; \
+   if ((cb) && (cb)->debug_message) { \
+      _pipe_debug_message(cb, &id, \
+                          PIPE_DEBUG_TYPE_ ## type, \
+                          fmt, ##__VA_ARGS__); \
+   } \
+} while (0)
+
+struct pipe_debug_callback;
+
+void
+_pipe_debug_message(
+   struct pipe_debug_callback *cb,
+   unsigned *id,
+   enum pipe_debug_type type,
+   const char *fmt, ...) _util_printf_format(4, 5);
+
+
+/**
+ * Used by debug_dump_enum and debug_dump_flags to describe symbols.
+ */
+struct debug_named_value
+{
+   const char *name;
+   uint64_t value;
+   const char *desc;
+};
+
+
+/**
+ * Some C pre-processor magic to simplify creating named values.
+ * 
+ * Example:
+ * @code
+ * static const debug_named_value my_names[] = {
+ *    DEBUG_NAMED_VALUE(MY_ENUM_VALUE_X),
+ *    DEBUG_NAMED_VALUE(MY_ENUM_VALUE_Y),
+ *    DEBUG_NAMED_VALUE(MY_ENUM_VALUE_Z),
+ *    DEBUG_NAMED_VALUE_END
+ * };
+ * 
+ *    ...
+ *    debug_printf("%s = %s\n", 
+ *                 name,
+ *                 debug_dump_enum(my_names, my_value));
+ *    ...
+ * @endcode
+ */
+#define DEBUG_NAMED_VALUE(__symbol) {#__symbol, (unsigned long)__symbol, NULL}
+#define DEBUG_NAMED_VALUE_WITH_DESCRIPTION(__symbol, __desc) {#__symbol, (unsigned long)__symbol, __desc}
+#define DEBUG_NAMED_VALUE_END {NULL, 0, NULL}
+
+
+/**
+ * Convert a enum value to a string.
+ */
+const char *
+debug_dump_enum(const struct debug_named_value *names, 
+                unsigned long value);
+
+const char *
+debug_dump_enum_noprefix(const struct debug_named_value *names, 
+                         const char *prefix,
+                         unsigned long value);
+
+
+/**
+ * Convert binary flags value to a string.
+ */
+const char *
+debug_dump_flags(const struct debug_named_value *names, 
+                 unsigned long value);
+
+
+/**
+ * Function enter exit loggers
+ */
+#ifdef DEBUG
+int debug_funclog_enter(const char* f, const int line, const char* file);
+void debug_funclog_exit(const char* f, const int line, const char* file);
+void debug_funclog_enter_exit(const char* f, const int line, const char* file);
+
+#define DEBUG_FUNCLOG_ENTER() \
+   int __debug_decleration_work_around = \
+      debug_funclog_enter(__FUNCTION__, __LINE__, __FILE__)
+#define DEBUG_FUNCLOG_EXIT() \
+   do { \
+      (void)__debug_decleration_work_around; \
+      debug_funclog_exit(__FUNCTION__, __LINE__, __FILE__); \
+      return; \
+   } while(0)
+#define DEBUG_FUNCLOG_EXIT_RET(ret) \
+   do { \
+      (void)__debug_decleration_work_around; \
+      debug_funclog_exit(__FUNCTION__, __LINE__, __FILE__); \
+      return ret; \
+   } while(0)
+#define DEBUG_FUNCLOG_ENTER_EXIT() \
+   debug_funclog_enter_exit(__FUNCTION__, __LINE__, __FILE__)
+
+#else
+#define DEBUG_FUNCLOG_ENTER() \
+   int __debug_decleration_work_around
+#define DEBUG_FUNCLOG_EXIT() \
+   do { (void)__debug_decleration_work_around; return; } while(0)
+#define DEBUG_FUNCLOG_EXIT_RET(ret) \
+   do { (void)__debug_decleration_work_around; return ret; } while(0)
+#define DEBUG_FUNCLOG_ENTER_EXIT()
+#endif
+
+
+/**
+ * Get option.
+ * 
+ * It is an alias for getenv on Linux. 
+ * 
+ * On Windows it reads C:\gallium.cfg, which is a text file with CR+LF line 
+ * endings with one option per line as
+ *  
+ *   NAME=value
+ * 
+ * This file must be terminated with an extra empty line.
+ */
+const char *
+debug_get_option(const char *name, const char *dfault);
+
+boolean
+debug_get_bool_option(const char *name, boolean dfault);
+
+long
+debug_get_num_option(const char *name, long dfault);
+
+uint64_t
+debug_get_flags_option(const char *name, 
+                       const struct debug_named_value *flags,
+                       uint64_t dfault);
+
+#define DEBUG_GET_ONCE_OPTION(suffix, name, dfault) \
+static const char * \
+debug_get_option_ ## suffix (void) \
+{ \
+   static boolean first = TRUE; \
+   static const char * value; \
+   if (first) { \
+      first = FALSE; \
+      value = debug_get_option(name, dfault); \
+   } \
+   return value; \
+}
+
+#define DEBUG_GET_ONCE_BOOL_OPTION(sufix, name, dfault) \
+static boolean \
+debug_get_option_ ## sufix (void) \
+{ \
+   static boolean first = TRUE; \
+   static boolean value; \
+   if (first) { \
+      first = FALSE; \
+      value = debug_get_bool_option(name, dfault); \
+   } \
+   return value; \
+}
+
+#define DEBUG_GET_ONCE_NUM_OPTION(sufix, name, dfault) \
+static long \
+debug_get_option_ ## sufix (void) \
+{ \
+   static boolean first = TRUE; \
+   static long value; \
+   if (first) { \
+      first = FALSE; \
+      value = debug_get_num_option(name, dfault); \
+   } \
+   return value; \
+}
+
+#define DEBUG_GET_ONCE_FLAGS_OPTION(sufix, name, flags, dfault) \
+static unsigned long \
+debug_get_option_ ## sufix (void) \
+{ \
+   static boolean first = TRUE; \
+   static unsigned long value; \
+   if (first) { \
+      first = FALSE; \
+      value = debug_get_flags_option(name, flags, dfault); \
+   } \
+   return value; \
+}
+
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif /* U_DEBUG_H_ */
diff --git a/lib/mesa/src/util/u_math.c b/lib/mesa/src/util/u_math.c
new file mode 100644
index 000000000..c58af911b
--- /dev/null
+++ b/lib/mesa/src/util/u_math.c
@@ -0,0 +1,137 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 VMware, Inc.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+
+#include "pipe/p_config.h"
+#include "util/u_math.h"
+#include "util/u_cpu_detect.h"
+
+#if defined(PIPE_ARCH_SSE)
+#include <xmmintrin.h>
+/* This is defined in pmmintrin.h, but it can only be included when -msse3 is
+ * used, so just define it here to avoid further. */
+#define _MM_DENORMALS_ZERO_MASK	0x0040
+#endif
+
+
+/** 2^x, for x in [-1.0, 1.0) */
+float pow2_table[POW2_TABLE_SIZE];
+
+
+static void
+init_pow2_table(void)
+{
+   int i;
+   for (i = 0; i < POW2_TABLE_SIZE; i++)
+      pow2_table[i] = exp2f((i - POW2_TABLE_OFFSET) / POW2_TABLE_SCALE);
+}
+
+
+/** log2(x), for x in [1.0, 2.0) */
+float log2_table[LOG2_TABLE_SIZE];
+
+
+static void 
+init_log2_table(void)
+{
+   unsigned i;
+   for (i = 0; i < LOG2_TABLE_SIZE; i++)
+      log2_table[i] = (float) log2(1.0 + i * (1.0 / LOG2_TABLE_SCALE));
+}
+
+
+/**
+ * One time init for math utilities.
+ */
+void
+util_init_math(void)
+{
+   static boolean initialized = FALSE;
+   if (!initialized) {
+      init_pow2_table();
+      init_log2_table();
+      initialized = TRUE;
+   }
+}
+
+/**
+ * Fetches the contents of the fpstate (mxcsr on x86) register.
+ *
+ * On platforms without support for it just returns 0.
+ */
+unsigned
+util_fpstate_get(void)
+{
+   unsigned mxcsr = 0;
+
+#if defined(PIPE_ARCH_SSE)
+   if (util_cpu_caps.has_sse) {
+      mxcsr = _mm_getcsr();
+   }
+#endif
+
+   return mxcsr;
+}
+
+/**
+ * Make sure that the fp treats the denormalized floating
+ * point numbers as zero.
+ *
+ * This is the behavior required by D3D10. OpenGL doesn't care.
+ */
+unsigned
+util_fpstate_set_denorms_to_zero(unsigned current_mxcsr)
+{
+#if defined(PIPE_ARCH_SSE)
+   if (util_cpu_caps.has_sse) {
+      /* Enable flush to zero mode */
+      current_mxcsr |= _MM_FLUSH_ZERO_MASK;
+      if (util_cpu_caps.has_daz) {
+         /* Enable denormals are zero mode */
+         current_mxcsr |= _MM_DENORMALS_ZERO_MASK;
+      }
+      util_fpstate_set(current_mxcsr);
+   }
+#endif
+   return current_mxcsr;
+}
+
+/**
+ * Set the state of the fpstate (mxcsr on x86) register.
+ *
+ * On platforms without support for it's a noop.
+ */
+void
+util_fpstate_set(unsigned mxcsr)
+{
+#if defined(PIPE_ARCH_SSE)
+   if (util_cpu_caps.has_sse) {
+      _mm_setcsr(mxcsr);
+   }
+#endif
+}
diff --git a/lib/mesa/src/util/u_math.h b/lib/mesa/src/util/u_math.h
new file mode 100644
index 000000000..e7dbbe5ca
--- /dev/null
+++ b/lib/mesa/src/util/u_math.h
@@ -0,0 +1,752 @@
+/**************************************************************************
+ * 
+ * Copyright 2008 VMware, Inc.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+/**
+ * Math utilities and approximations for common math functions.
+ * Reduced precision is usually acceptable in shaders...
+ *
+ * "fast" is used in the names of functions which are low-precision,
+ * or at least lower-precision than the normal C lib functions.
+ */
+
+
+#ifndef U_MATH_H
+#define U_MATH_H
+
+
+#include "pipe/p_compiler.h"
+
+#include "c99_math.h"
+#include <assert.h>
+#include <float.h>
+#include <stdarg.h>
+
+#include "bitscan.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+
+#ifndef M_SQRT2
+#define M_SQRT2 1.41421356237309504880
+#endif
+
+#define POW2_TABLE_SIZE_LOG2 9
+#define POW2_TABLE_SIZE (1 << POW2_TABLE_SIZE_LOG2)
+#define POW2_TABLE_OFFSET (POW2_TABLE_SIZE/2)
+#define POW2_TABLE_SCALE ((float)(POW2_TABLE_SIZE/2))
+extern float pow2_table[POW2_TABLE_SIZE];
+
+
+/**
+ * Initialize math module.  This should be called before using any
+ * other functions in this module.
+ */
+extern void
+util_init_math(void);
+
+
+union fi {
+   float f;
+   int32_t i;
+   uint32_t ui;
+};
+
+
+union di {
+   double d;
+   int64_t i;
+   uint64_t ui;
+};
+
+
+/**
+ * Extract the IEEE float32 exponent.
+ */
+static inline signed
+util_get_float32_exponent(float x)
+{
+   union fi f;
+
+   f.f = x;
+
+   return ((f.ui >> 23) & 0xff) - 127;
+}
+
+
+/**
+ * Fast version of 2^x
+ * Identity: exp2(a + b) = exp2(a) * exp2(b)
+ * Let ipart = int(x)
+ * Let fpart = x - ipart;
+ * So, exp2(x) = exp2(ipart) * exp2(fpart)
+ * Compute exp2(ipart) with i << ipart
+ * Compute exp2(fpart) with lookup table.
+ */
+static inline float
+util_fast_exp2(float x)
+{
+   int32_t ipart;
+   float fpart, mpart;
+   union fi epart;
+
+   if(x > 129.00000f)
+      return 3.402823466e+38f;
+
+   if (x < -126.99999f)
+      return 0.0f;
+
+   ipart = (int32_t) x;
+   fpart = x - (float) ipart;
+
+   /* same as
+    *   epart.f = (float) (1 << ipart)
+    * but faster and without integer overflow for ipart > 31
+    */
+   epart.i = (ipart + 127 ) << 23;
+
+   mpart = pow2_table[POW2_TABLE_OFFSET + (int)(fpart * POW2_TABLE_SCALE)];
+
+   return epart.f * mpart;
+}
+
+
+/**
+ * Fast approximation to exp(x).
+ */
+static inline float
+util_fast_exp(float x)
+{
+   const float k = 1.44269f; /* = log2(e) */
+   return util_fast_exp2(k * x);
+}
+
+
+#define LOG2_TABLE_SIZE_LOG2 16
+#define LOG2_TABLE_SCALE (1 << LOG2_TABLE_SIZE_LOG2)
+#define LOG2_TABLE_SIZE (LOG2_TABLE_SCALE + 1)
+extern float log2_table[LOG2_TABLE_SIZE];
+
+
+/**
+ * Fast approximation to log2(x).
+ */
+static inline float
+util_fast_log2(float x)
+{
+   union fi num;
+   float epart, mpart;
+   num.f = x;
+   epart = (float)(((num.i & 0x7f800000) >> 23) - 127);
+   /* mpart = log2_table[mantissa*LOG2_TABLE_SCALE + 0.5] */
+   mpart = log2_table[((num.i & 0x007fffff) + (1 << (22 - LOG2_TABLE_SIZE_LOG2))) >> (23 - LOG2_TABLE_SIZE_LOG2)];
+   return epart + mpart;
+}
+
+
+/**
+ * Fast approximation to x^y.
+ */
+static inline float
+util_fast_pow(float x, float y)
+{
+   return util_fast_exp2(util_fast_log2(x) * y);
+}
+
+
+/**
+ * Floor(x), returned as int.
+ */
+static inline int
+util_ifloor(float f)
+{
+   int ai, bi;
+   double af, bf;
+   union fi u;
+   af = (3 << 22) + 0.5 + (double) f;
+   bf = (3 << 22) + 0.5 - (double) f;
+   u.f = (float) af;  ai = u.i;
+   u.f = (float) bf;  bi = u.i;
+   return (ai - bi) >> 1;
+}
+
+
+/**
+ * Round float to nearest int.
+ */
+static inline int
+util_iround(float f)
+{
+#if defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86) 
+   int r;
+   __asm__ ("fistpl %0" : "=m" (r) : "t" (f) : "st");
+   return r;
+#elif defined(PIPE_CC_MSVC) && defined(PIPE_ARCH_X86)
+   int r;
+   _asm {
+      fld f
+      fistp r
+   }
+   return r;
+#else
+   if (f >= 0.0f)
+      return (int) (f + 0.5f);
+   else
+      return (int) (f - 0.5f);
+#endif
+}
+
+
+/**
+ * Approximate floating point comparison
+ */
+static inline boolean
+util_is_approx(float a, float b, float tol)
+{
+   return fabsf(b - a) <= tol;
+}
+
+
+/**
+ * util_is_X_inf_or_nan = test if x is NaN or +/- Inf
+ * util_is_X_nan        = test if x is NaN
+ * util_X_inf_sign      = return +1 for +Inf, -1 for -Inf, or 0 for not Inf
+ *
+ * NaN can be checked with x != x, however this fails with the fast math flag
+ **/
+
+
+/**
+ * Single-float
+ */
+static inline boolean
+util_is_inf_or_nan(float x)
+{
+   union fi tmp;
+   tmp.f = x;
+   return (tmp.ui & 0x7f800000) == 0x7f800000;
+}
+
+
+static inline boolean
+util_is_nan(float x)
+{
+   union fi tmp;
+   tmp.f = x;
+   return (tmp.ui & 0x7fffffff) > 0x7f800000;
+}
+
+
+static inline int
+util_inf_sign(float x)
+{
+   union fi tmp;
+   tmp.f = x;
+   if ((tmp.ui & 0x7fffffff) != 0x7f800000) {
+      return 0;
+   }
+
+   return (x < 0) ? -1 : 1;
+}
+
+
+/**
+ * Double-float
+ */
+static inline boolean
+util_is_double_inf_or_nan(double x)
+{
+   union di tmp;
+   tmp.d = x;
+   return (tmp.ui & 0x7ff0000000000000ULL) == 0x7ff0000000000000ULL;
+}
+
+
+static inline boolean
+util_is_double_nan(double x)
+{
+   union di tmp;
+   tmp.d = x;
+   return (tmp.ui & 0x7fffffffffffffffULL) > 0x7ff0000000000000ULL;
+}
+
+
+static inline int
+util_double_inf_sign(double x)
+{
+   union di tmp;
+   tmp.d = x;
+   if ((tmp.ui & 0x7fffffffffffffffULL) != 0x7ff0000000000000ULL) {
+      return 0;
+   }
+
+   return (x < 0) ? -1 : 1;
+}
+
+
+/**
+ * Half-float
+ */
+static inline boolean
+util_is_half_inf_or_nan(int16_t x)
+{
+   return (x & 0x7c00) == 0x7c00;
+}
+
+
+static inline boolean
+util_is_half_nan(int16_t x)
+{
+   return (x & 0x7fff) > 0x7c00;
+}
+
+
+static inline int
+util_half_inf_sign(int16_t x)
+{
+   if ((x & 0x7fff) != 0x7c00) {
+      return 0;
+   }
+
+   return (x < 0) ? -1 : 1;
+}
+
+
+/**
+ * Return float bits.
+ */
+static inline unsigned
+fui( float f )
+{
+   union fi fi;
+   fi.f = f;
+   return fi.ui;
+}
+
+static inline float
+uif(uint32_t ui)
+{
+   union fi fi;
+   fi.ui = ui;
+   return fi.f;
+}
+
+
+/**
+ * Convert ubyte to float in [0, 1].
+ */
+static inline float
+ubyte_to_float(ubyte ub)
+{
+   return (float) ub * (1.0f / 255.0f);
+}
+
+
+/**
+ * Convert float in [0,1] to ubyte in [0,255] with clamping.
+ */
+static inline ubyte
+float_to_ubyte(float f)
+{
+   /* return 0 for NaN too */
+   if (!(f > 0.0f)) {
+      return (ubyte) 0;
+   }
+   else if (f >= 1.0f) {
+      return (ubyte) 255;
+   }
+   else {
+      union fi tmp;
+      tmp.f = f;
+      tmp.f = tmp.f * (255.0f/256.0f) + 32768.0f;
+      return (ubyte) tmp.i;
+   }
+}
+
+static inline float
+byte_to_float_tex(int8_t b)
+{
+   return (b == -128) ? -1.0F : b * 1.0F / 127.0F;
+}
+
+static inline int8_t
+float_to_byte_tex(float f)
+{
+   return (int8_t) (127.0F * f);
+}
+
+/**
+ * Calc log base 2
+ */
+static inline unsigned
+util_logbase2(unsigned n)
+{
+#if defined(HAVE___BUILTIN_CLZ)
+   return ((sizeof(unsigned) * 8 - 1) - __builtin_clz(n | 1));
+#else
+   unsigned pos = 0;
+   if (n >= 1<<16) { n >>= 16; pos += 16; }
+   if (n >= 1<< 8) { n >>=  8; pos +=  8; }
+   if (n >= 1<< 4) { n >>=  4; pos +=  4; }
+   if (n >= 1<< 2) { n >>=  2; pos +=  2; }
+   if (n >= 1<< 1) {           pos +=  1; }
+   return pos;
+#endif
+}
+
+static inline uint64_t
+util_logbase2_64(uint64_t n)
+{
+#if defined(HAVE___BUILTIN_CLZLL)
+   return ((sizeof(uint64_t) * 8 - 1) - __builtin_clzll(n | 1));
+#else
+   uint64_t pos = 0ull;
+   if (n >= 1ull<<32) { n >>= 32; pos += 32; }
+   if (n >= 1ull<<16) { n >>= 16; pos += 16; }
+   if (n >= 1ull<< 8) { n >>=  8; pos +=  8; }
+   if (n >= 1ull<< 4) { n >>=  4; pos +=  4; }
+   if (n >= 1ull<< 2) { n >>=  2; pos +=  2; }
+   if (n >= 1ull<< 1) {           pos +=  1; }
+   return pos;
+#endif
+}
+
+/**
+ * Returns the ceiling of log n base 2, and 0 when n == 0. Equivalently,
+ * returns the smallest x such that n <= 2**x.
+ */
+static inline unsigned
+util_logbase2_ceil(unsigned n)
+{
+   if (n <= 1)
+      return 0;
+
+   return 1 + util_logbase2(n - 1);
+}
+
+static inline uint64_t
+util_logbase2_ceil64(uint64_t n)
+{
+   if (n <= 1)
+      return 0;
+
+   return 1ull + util_logbase2_64(n - 1);
+}
+
+/**
+ * Returns the smallest power of two >= x
+ */
+static inline unsigned
+util_next_power_of_two(unsigned x)
+{
+#if defined(HAVE___BUILTIN_CLZ)
+   if (x <= 1)
+       return 1;
+
+   return (1 << ((sizeof(unsigned) * 8) - __builtin_clz(x - 1)));
+#else
+   unsigned val = x;
+
+   if (x <= 1)
+      return 1;
+
+   if (util_is_power_of_two_or_zero(x))
+      return x;
+
+   val--;
+   val = (val >> 1) | val;
+   val = (val >> 2) | val;
+   val = (val >> 4) | val;
+   val = (val >> 8) | val;
+   val = (val >> 16) | val;
+   val++;
+   return val;
+#endif
+}
+
+static inline uint64_t
+util_next_power_of_two64(uint64_t x)
+{
+#if defined(HAVE___BUILTIN_CLZLL)
+   if (x <= 1)
+       return 1;
+
+   return (1ull << ((sizeof(uint64_t) * 8) - __builtin_clzll(x - 1)));
+#else
+   uint64_t val = x;
+
+   if (x <= 1)
+      return 1;
+
+   if (util_is_power_of_two_or_zero64(x))
+      return x;
+
+   val--;
+   val = (val >> 1)  | val;
+   val = (val >> 2)  | val;
+   val = (val >> 4)  | val;
+   val = (val >> 8)  | val;
+   val = (val >> 16) | val;
+   val = (val >> 32) | val;
+   val++;
+   return val;
+#endif
+}
+
+
+/**
+ * Return number of bits set in n.
+ */
+static inline unsigned
+util_bitcount(unsigned n)
+{
+#if defined(HAVE___BUILTIN_POPCOUNT)
+   return __builtin_popcount(n);
+#else
+   /* K&R classic bitcount.
+    *
+    * For each iteration, clear the LSB from the bitfield.
+    * Requires only one iteration per set bit, instead of
+    * one iteration per bit less than highest set bit.
+    */
+   unsigned bits;
+   for (bits = 0; n; bits++) {
+      n &= n - 1;
+   }
+   return bits;
+#endif
+}
+
+
+static inline unsigned
+util_bitcount64(uint64_t n)
+{
+#ifdef HAVE___BUILTIN_POPCOUNTLL
+   return __builtin_popcountll(n);
+#else
+   return util_bitcount(n) + util_bitcount(n >> 32);
+#endif
+}
+
+
+/**
+ * Reverse bits in n
+ * Algorithm taken from:
+ * http://stackoverflow.com/questions/9144800/c-reverse-bits-in-unsigned-integer
+ */
+static inline unsigned
+util_bitreverse(unsigned n)
+{
+    n = ((n >> 1) & 0x55555555u) | ((n & 0x55555555u) << 1);
+    n = ((n >> 2) & 0x33333333u) | ((n & 0x33333333u) << 2);
+    n = ((n >> 4) & 0x0f0f0f0fu) | ((n & 0x0f0f0f0fu) << 4);
+    n = ((n >> 8) & 0x00ff00ffu) | ((n & 0x00ff00ffu) << 8);
+    n = ((n >> 16) & 0xffffu) | ((n & 0xffffu) << 16);
+    return n;
+}
+
+/**
+ * Convert from little endian to CPU byte order.
+ */
+
+#ifdef PIPE_ARCH_BIG_ENDIAN
+#define util_le64_to_cpu(x) util_bswap64(x)
+#define util_le32_to_cpu(x) util_bswap32(x)
+#define util_le16_to_cpu(x) util_bswap16(x)
+#else
+#define util_le64_to_cpu(x) (x)
+#define util_le32_to_cpu(x) (x)
+#define util_le16_to_cpu(x) (x)
+#endif
+
+#define util_cpu_to_le64(x) util_le64_to_cpu(x)
+#define util_cpu_to_le32(x) util_le32_to_cpu(x)
+#define util_cpu_to_le16(x) util_le16_to_cpu(x)
+
+/**
+ * Reverse byte order of a 32 bit word.
+ */
+static inline uint32_t
+util_bswap32(uint32_t n)
+{
+#if defined(HAVE___BUILTIN_BSWAP32)
+   return __builtin_bswap32(n);
+#else
+   return (n >> 24) |
+          ((n >> 8) & 0x0000ff00) |
+          ((n << 8) & 0x00ff0000) |
+          (n << 24);
+#endif
+}
+
+/**
+ * Reverse byte order of a 64bit word.
+ */
+static inline uint64_t
+util_bswap64(uint64_t n)
+{
+#if defined(HAVE___BUILTIN_BSWAP64)
+   return __builtin_bswap64(n);
+#else
+   return ((uint64_t)util_bswap32((uint32_t)n) << 32) |
+          util_bswap32((n >> 32));
+#endif
+}
+
+
+/**
+ * Reverse byte order of a 16 bit word.
+ */
+static inline uint16_t
+util_bswap16(uint16_t n)
+{
+   return (n >> 8) |
+          (n << 8);
+}
+
+static inline void*
+util_memcpy_cpu_to_le32(void * restrict dest, const void * restrict src, size_t n)
+{
+#ifdef PIPE_ARCH_BIG_ENDIAN
+   size_t i, e;
+   assert(n % 4 == 0);
+
+   for (i = 0, e = n / 4; i < e; i++) {
+      uint32_t * restrict d = (uint32_t* restrict)dest;
+      const uint32_t * restrict s = (const uint32_t* restrict)src;
+      d[i] = util_bswap32(s[i]);
+   }
+   return dest;
+#else
+   return memcpy(dest, src, n);
+#endif
+}
+
+/**
+ * Clamp X to [MIN, MAX].
+ * This is a macro to allow float, int, uint, etc. types.
+ * We arbitrarily turn NaN into MIN.
+ */
+#define CLAMP( X, MIN, MAX )  ( (X)>(MIN) ? ((X)>(MAX) ? (MAX) : (X)) : (MIN) )
+
+#define MIN2( A, B )   ( (A)<(B) ? (A) : (B) )
+#define MAX2( A, B )   ( (A)>(B) ? (A) : (B) )
+
+#define MIN3( A, B, C ) ((A) < (B) ? MIN2(A, C) : MIN2(B, C))
+#define MAX3( A, B, C ) ((A) > (B) ? MAX2(A, C) : MAX2(B, C))
+
+#define MIN4( A, B, C, D ) ((A) < (B) ? MIN3(A, C, D) : MIN3(B, C, D))
+#define MAX4( A, B, C, D ) ((A) > (B) ? MAX3(A, C, D) : MAX3(B, C, D))
+
+
+/**
+ * Align a value, only works pot alignemnts.
+ */
+static inline int
+align(int value, int alignment)
+{
+   return (value + alignment - 1) & ~(alignment - 1);
+}
+
+static inline uint64_t
+align64(uint64_t value, unsigned alignment)
+{
+   return (value + alignment - 1) & ~((uint64_t)alignment - 1);
+}
+
+/**
+ * Works like align but on npot alignments.
+ */
+static inline size_t
+util_align_npot(size_t value, size_t alignment)
+{
+   if (value % alignment)
+      return value + (alignment - (value % alignment));
+   return value;
+}
+
+static inline unsigned
+u_minify(unsigned value, unsigned levels)
+{
+    return MAX2(1, value >> levels);
+}
+
+#ifndef COPY_4V
+#define COPY_4V( DST, SRC )         \
+do {                                \
+   (DST)[0] = (SRC)[0];             \
+   (DST)[1] = (SRC)[1];             \
+   (DST)[2] = (SRC)[2];             \
+   (DST)[3] = (SRC)[3];             \
+} while (0)
+#endif
+
+
+#ifndef COPY_4FV
+#define COPY_4FV( DST, SRC )  COPY_4V(DST, SRC)
+#endif
+
+
+#ifndef ASSIGN_4V
+#define ASSIGN_4V( DST, V0, V1, V2, V3 ) \
+do {                                     \
+   (DST)[0] = (V0);                      \
+   (DST)[1] = (V1);                      \
+   (DST)[2] = (V2);                      \
+   (DST)[3] = (V3);                      \
+} while (0)
+#endif
+
+
+static inline uint32_t
+util_unsigned_fixed(float value, unsigned frac_bits)
+{
+   return value < 0 ? 0 : (uint32_t)(value * (1<<frac_bits));
+}
+
+static inline int32_t
+util_signed_fixed(float value, unsigned frac_bits)
+{
+   return (int32_t)(value * (1<<frac_bits));
+}
+
+unsigned
+util_fpstate_get(void);
+unsigned
+util_fpstate_set_denorms_to_zero(unsigned current_fpstate);
+void
+util_fpstate_set(unsigned fpstate);
+
+
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* U_MATH_H */
diff --git a/lib/mesa/src/util/u_process.c b/lib/mesa/src/util/u_process.c
new file mode 100644
index 000000000..5e5927678
--- /dev/null
+++ b/lib/mesa/src/util/u_process.c
@@ -0,0 +1,125 @@
+/*
+ * Copyright © 2003 Felix Kuehling
+ * Copyright © 2018 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
+ * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NON-INFRINGEMENT. IN NO EVENT SHALL THE COPYRIGHT HOLDERS, AUTHORS
+ * AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ */
+
+#include "u_process.h"
+#include <string.h>
+#include <errno.h>
+#include <stdlib.h>
+
+#undef GET_PROGRAM_NAME
+
+#if (defined(__GNU_LIBRARY__) || defined(__GLIBC__)) && !defined(__UCLIBC__)
+#    if !defined(__GLIBC__) || (__GLIBC__ < 2)
+/* These aren't declared in any libc5 header */
+extern char *program_invocation_name, *program_invocation_short_name;
+#    endif
+static const char *
+__getProgramName()
+{
+   char * arg = strrchr(program_invocation_name, '/');
+   if (arg)
+      return arg+1;
+
+   /* If there was no '/' at all we likely have a windows like path from
+    * a wine application.
+    */
+   arg = strrchr(program_invocation_name, '\\');
+   if (arg)
+      return arg+1;
+
+   return program_invocation_name;
+}
+#    define GET_PROGRAM_NAME() __getProgramName()
+#elif defined(__CYGWIN__)
+#    define GET_PROGRAM_NAME() program_invocation_short_name
+#elif defined(__FreeBSD__) && (__FreeBSD__ >= 2)
+#    include <osreldate.h>
+#    if (__FreeBSD_version >= 440000)
+#        define GET_PROGRAM_NAME() getprogname()
+#    endif
+#elif defined(__NetBSD__) && defined(__NetBSD_Version__) && (__NetBSD_Version__ >= 106000100)
+#    define GET_PROGRAM_NAME() getprogname()
+#elif defined(__DragonFly__)
+#    define GET_PROGRAM_NAME() getprogname()
+#elif defined(__APPLE__)
+#    define GET_PROGRAM_NAME() getprogname()
+#elif defined(ANDROID)
+#    define GET_PROGRAM_NAME() getprogname()
+#elif defined(__sun)
+/* Solaris has getexecname() which returns the full path - return just
+   the basename to match BSD getprogname() */
+#    include <libgen.h>
+
+static const char *
+__getProgramName()
+{
+    static const char *progname;
+
+    if (progname == NULL) {
+        const char *e = getexecname();
+        if (e != NULL) {
+            /* Have to make a copy since getexecname can return a readonly
+               string, but basename expects to be able to modify its arg. */
+            char *n = strdup(e);
+            if (n != NULL) {
+                progname = basename(n);
+            }
+        }
+    }
+    return progname;
+}
+
+#    define GET_PROGRAM_NAME() __getProgramName()
+#endif
+
+#if !defined(GET_PROGRAM_NAME)
+#    if defined(__OpenBSD__) || defined(NetBSD) || defined(__UCLIBC__) || defined(ANDROID)
+/* This is a hack. It's said to work on OpenBSD, NetBSD and GNU.
+ * Rogelio M.Serrano Jr. reported it's also working with UCLIBC. It's
+ * used as a last resort, if there is no documented facility available. */
+static const char *
+__getProgramName()
+{
+    extern const char *__progname;
+    char * arg = strrchr(__progname, '/');
+    if (arg)
+        return arg+1;
+    else
+        return __progname;
+}
+#        define GET_PROGRAM_NAME() __getProgramName()
+#    else
+#        define GET_PROGRAM_NAME() ""
+#        pragma message ( "Warning: Per application configuration won't work with your OS version." )
+#    endif
+#endif
+
+const char *
+util_get_process_name(void)
+{
+   return GET_PROGRAM_NAME();
+}
diff --git a/lib/mesa/src/util/u_process.h b/lib/mesa/src/util/u_process.h
new file mode 100644
index 000000000..77f7cb1cb
--- /dev/null
+++ b/lib/mesa/src/util/u_process.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright © 2003 Felix Kuehling
+ * Copyright © 2018 Advanced Micro Devices, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
+ * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NON-INFRINGEMENT. IN NO EVENT SHALL THE COPYRIGHT HOLDERS, AUTHORS
+ * AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ */
+
+#ifndef PROCESS_H
+#define PROCESS_H
+
+const char *
+util_get_process_name(void);
+
+#endif
diff --git a/lib/mesa/src/util/vma.c b/lib/mesa/src/util/vma.c
new file mode 100644
index 000000000..c8f55031c
--- /dev/null
+++ b/lib/mesa/src/util/vma.c
@@ -0,0 +1,234 @@
+/*
+ * Copyright © 2018 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#include <stdlib.h>
+
+#include "util/u_math.h"
+#include "util/vma.h"
+
+struct util_vma_hole {
+   struct list_head link;
+   uint64_t offset;
+   uint64_t size;
+};
+
+#define util_vma_foreach_hole(_hole, _heap) \
+   list_for_each_entry(struct util_vma_hole, _hole, &(_heap)->holes, link)
+
+#define util_vma_foreach_hole_safe(_hole, _heap) \
+   list_for_each_entry_safe(struct util_vma_hole, _hole, &(_heap)->holes, link)
+
+void
+util_vma_heap_init(struct util_vma_heap *heap,
+                   uint64_t start, uint64_t size)
+{
+   list_inithead(&heap->holes);
+   util_vma_heap_free(heap, start, size);
+}
+
+void
+util_vma_heap_finish(struct util_vma_heap *heap)
+{
+   util_vma_foreach_hole_safe(hole, heap)
+      free(hole);
+}
+
+#ifndef NDEBUG
+static void
+util_vma_heap_validate(struct util_vma_heap *heap)
+{
+   uint64_t prev_offset = 0;
+   util_vma_foreach_hole(hole, heap) {
+      assert(hole->offset > 0);
+      assert(hole->size > 0);
+
+      if (&hole->link == heap->holes.next) {
+         /* This must be the top-most hole.  Assert that, if it overflows, it
+          * overflows to 0, i.e. 2^64.
+          */
+         assert(hole->size + hole->offset == 0 ||
+                hole->size + hole->offset > hole->offset);
+      } else {
+         /* This is not the top-most hole so it must not overflow and, in
+          * fact, must be strictly lower than the top-most hole.  If
+          * hole->size + hole->offset == prev_offset, then we failed to join
+          * holes during a util_vma_heap_free.
+          */
+         assert(hole->size + hole->offset > hole->offset &&
+                hole->size + hole->offset < prev_offset);
+      }
+      prev_offset = hole->offset;
+   }
+}
+#else
+#define util_vma_heap_validate(heap)
+#endif
+
+uint64_t
+util_vma_heap_alloc(struct util_vma_heap *heap,
+                    uint64_t size, uint64_t alignment)
+{
+   /* The caller is expected to reject zero-size allocations */
+   assert(size > 0);
+   assert(alignment > 0);
+
+   util_vma_heap_validate(heap);
+
+   util_vma_foreach_hole_safe(hole, heap) {
+      if (size > hole->size)
+         continue;
+
+      /* Compute the offset as the highest address where a chunk of the given
+       * size can be without going over the top of the hole.
+       *
+       * This calculation is known to not overflow because we know that
+       * hole->size + hole->offset can only overflow to 0 and size > 0.
+       */
+      uint64_t offset = (hole->size - size) + hole->offset;
+
+      /* Align the offset.  We align down and not up because we are allocating
+       * from the top of the hole and not the bottom.
+       */
+      offset = (offset / alignment) * alignment;
+
+      if (offset < hole->offset)
+         continue;
+
+      if (offset == hole->offset && size == hole->size) {
+         /* Just get rid of the hole. */
+         list_del(&hole->link);
+         free(hole);
+         util_vma_heap_validate(heap);
+         return offset;
+      }
+
+      assert(offset - hole->offset <= hole->size - size);
+      uint64_t waste = (hole->size - size) - (offset - hole->offset);
+      if (waste == 0) {
+         /* We allocated at the top.  Shrink the hole down. */
+         hole->size -= size;
+         util_vma_heap_validate(heap);
+         return offset;
+      }
+
+      if (offset == hole->offset) {
+         /* We allocated at the bottom. Shrink the hole up. */
+         hole->offset += size;
+         hole->size -= size;
+         util_vma_heap_validate(heap);
+         return offset;
+      }
+
+      /* We allocated in the middle.  We need to split the old hole into two
+       * holes, one high and one low.
+       */
+      struct util_vma_hole *high_hole = calloc(1, sizeof(*hole));
+      high_hole->offset = offset + size;
+      high_hole->size = waste;
+
+      /* Adjust the hole to be the amount of space left at he bottom of the
+       * original hole.
+       */
+      hole->size = offset - hole->offset;
+
+      /* Place the new hole before the old hole so that the list is in order
+       * from high to low.
+       */
+      list_addtail(&high_hole->link, &hole->link);
+
+      util_vma_heap_validate(heap);
+
+      return offset;
+   }
+
+   /* Failed to allocate */
+   return 0;
+}
+
+void
+util_vma_heap_free(struct util_vma_heap *heap,
+                   uint64_t offset, uint64_t size)
+{
+   /* An offset of 0 is reserved for allocation failure.  It is not a valid
+    * address and cannot be freed.
+    */
+   assert(offset > 0);
+
+   /* Freeing something with a size of 0 is also not valid. */
+   assert(size > 0);
+
+   /* It's possible for offset + size to wrap around if we touch the top of
+    * the 64-bit address space, but we cannot go any higher than 2^64.
+    */
+   assert(offset + size == 0 || offset + size > offset);
+
+   util_vma_heap_validate(heap);
+
+   /* Find immediately higher and lower holes if they exist. */
+   struct util_vma_hole *high_hole = NULL, *low_hole = NULL;
+   util_vma_foreach_hole(hole, heap) {
+      if (hole->offset <= offset) {
+         low_hole = hole;
+         break;
+      }
+      high_hole = hole;
+   }
+
+   if (high_hole)
+      assert(offset + size <= high_hole->offset);
+   bool high_adjacent = high_hole && offset + size == high_hole->offset;
+
+   if (low_hole) {
+      assert(low_hole->offset + low_hole->size > low_hole->offset);
+      assert(low_hole->offset + low_hole->size <= offset);
+   }
+   bool low_adjacent = low_hole && low_hole->offset + low_hole->size == offset;
+
+   if (low_adjacent && high_adjacent) {
+      /* Merge the two holes */
+      low_hole->size += size + high_hole->size;
+      list_del(&high_hole->link);
+      free(high_hole);
+   } else if (low_adjacent) {
+      /* Merge into the low hole */
+      low_hole->size += size;
+   } else if (high_adjacent) {
+      /* Merge into the high hole */
+      high_hole->offset = offset;
+      high_hole->size += size;
+   } else {
+      /* Neither hole is adjacent; make a new one */
+      struct util_vma_hole *hole = calloc(1, sizeof(*hole));
+
+      hole->offset = offset;
+      hole->size = size;
+
+      /* Add it after the high hole so we maintain high-to-low ordering */
+      if (high_hole)
+         list_add(&hole->link, &high_hole->link);
+      else
+         list_add(&hole->link, &heap->holes);
+   }
+
+   util_vma_heap_validate(heap);
+}
diff --git a/lib/mesa/src/util/vma.h b/lib/mesa/src/util/vma.h
new file mode 100644
index 000000000..ed69914e4
--- /dev/null
+++ b/lib/mesa/src/util/vma.h
@@ -0,0 +1,53 @@
+/*
+ * Copyright © 2018 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+#ifndef _UTIL_VMA_H
+#define _UTIL_VMA_H
+
+#include <stdint.h>
+
+#include "list.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct util_vma_heap {
+   struct list_head holes;
+};
+
+void util_vma_heap_init(struct util_vma_heap *heap,
+                        uint64_t start, uint64_t size);
+void util_vma_heap_finish(struct util_vma_heap *heap);
+
+uint64_t util_vma_heap_alloc(struct util_vma_heap *heap,
+                             uint64_t size, uint64_t alignment);
+
+void util_vma_heap_free(struct util_vma_heap *heap,
+                        uint64_t offset, uint64_t size);
+
+#ifdef __cplusplus
+} /* extern C */
+#endif
+
+#endif /* _UTIL_DEBUG_H */
diff --git a/lib/mesa/src/util/xmlconfig.h b/lib/mesa/src/util/xmlconfig.h
index 77aa14c20..cc8c6ef09 100644
--- a/lib/mesa/src/util/xmlconfig.h
+++ b/lib/mesa/src/util/xmlconfig.h
@@ -32,6 +32,7 @@
 
 #include "util/mesa-sha1.h"
 #include "util/ralloc.h"
+#include <string.h>
 
 #define STRING_CONF_MAXLEN 25
 
@@ -101,10 +102,11 @@ void driParseOptionInfo (driOptionCache *info,
 			 const char *configOptions);
 /** \brief Initialize option cache from info and parse configuration files
  *
- * To be called in <driver>CreateContext. screenNum and driverName select
- * device sections. */
+ * To be called in <driver>CreateContext. screenNum, driverName and
+ * kernelDriverName select device sections. */
 void driParseConfigFiles (driOptionCache *cache, const driOptionCache *info,
-			  int screenNum, const char *driverName);
+			  int screenNum, const char *driverName,
+			  const char *kernelDriverName);
 /** \brief Destroy option info
  *
  * To be called in <driver>DestroyScreen */
diff --git a/lib/mesa/src/util/xmlpool/LINGUAS b/lib/mesa/src/util/xmlpool/LINGUAS
new file mode 100644
index 000000000..362017651
--- /dev/null
+++ b/lib/mesa/src/util/xmlpool/LINGUAS
@@ -0,0 +1 @@
+ca es de nl sv fr
diff --git a/lib/mesa/src/util/xmlpool/POTFILES b/lib/mesa/src/util/xmlpool/POTFILES
new file mode 100644
index 000000000..d68d7009b
--- /dev/null
+++ b/lib/mesa/src/util/xmlpool/POTFILES
@@ -0,0 +1 @@
+src/util/xmlpool/t_options.h
diff --git a/lib/mesa/src/util/xmlpool/SConscript b/lib/mesa/src/util/xmlpool/SConscript
index fa42554d3..0a00cefa6 100644
--- a/lib/mesa/src/util/xmlpool/SConscript
+++ b/lib/mesa/src/util/xmlpool/SConscript
@@ -8,7 +8,7 @@ xmlpool_options, = env.CodeGenerate(
     target = 'options.h',
     script = 'gen_xmlpool.py',
     source = ['t_options.h'],
-    command = python_cmd + ' $SCRIPT $SOURCE ' + LOCALEDIR + ' > $TARGET'
+    command = python_cmd + ' $SCRIPT --template $SOURCE --output $TARGET --localedir ' + LOCALEDIR
 )
 
 Export('xmlpool_options')
diff --git a/lib/mesa/src/util/xmlpool/ca.gmo b/lib/mesa/src/util/xmlpool/ca.gmo
new file mode 100644
index 000000000..2786d3f4d
--- /dev/null
+++ b/lib/mesa/src/util/xmlpool/ca.gmo
diff --git a/lib/mesa/src/util/xmlpool/ca.po b/lib/mesa/src/util/xmlpool/ca.po
index 03bf29613..91621f283 100644
--- a/lib/mesa/src/util/xmlpool/ca.po
+++ b/lib/mesa/src/util/xmlpool/ca.po
@@ -39,14 +39,6 @@ msgstr ""
 msgid "Debugging"
 msgstr "Depuració"
 
-#: t_options.h:60
-msgid "Disable 3D acceleration"
-msgstr "Deshabilita l'acceleració 3D"
-
-#: t_options.h:65
-msgid "Show performance boxes"
-msgstr "Mostra les caixes de rendiment"
-
 #: t_options.h:70
 msgid "Enable flushing batchbuffer after each draw call"
 msgstr "Habilita el buidatge del batchbuffer després de cada trucada de dibuix"
@@ -75,10 +67,6 @@ msgstr ""
 "Deshabilita les continuacions de línia basades en barra invertida en la font "
 "GLSL"
 
-#: t_options.h:100
-msgid "Disable GL_ARB_shader_bit_encoding"
-msgstr "Deshabilita el GL_ARB_shader_bit_encoding"
-
 #: t_options.h:105
 msgid ""
 "Force a default GLSL version for shaders that lack an explicit #version line"
@@ -94,87 +82,6 @@ msgstr "Permet les directives #extension GLSL en el mitjà dels shaders"
 msgid "Image Quality"
 msgstr "Qualitat d'imatge"
 
-#: t_options.h:133
-msgid "Texture color depth"
-msgstr "Profunditat de color de textura"
-
-#: t_options.h:134
-msgid "Prefer frame buffer color depth"
-msgstr "Prefereix profunditat de color del framebuffer"
-
-#: t_options.h:135
-msgid "Prefer 32 bits per texel"
-msgstr "Prefereix 32 bits per texel"
-
-#: t_options.h:136
-msgid "Prefer 16 bits per texel"
-msgstr "Prefereix 16 bits per texel"
-
-#: t_options.h:137
-msgid "Force 16 bits per texel"
-msgstr "Força 16 bits per texel"
-
-#: t_options.h:143
-msgid "Initial maximum value for anisotropic texture filtering"
-msgstr "Valor màxim inicial per a la filtració de textura anisòtropa"
-
-#: t_options.h:148
-msgid "Forbid negative texture LOD bias"
-msgstr ""
-"Prohibeix una parcialitat negativa del Nivell de Detalle (LOD) de les "
-"textures"
-
-#: t_options.h:153
-msgid ""
-"Enable S3TC texture compression even if software support is not available"
-msgstr ""
-"Habilita la compressió de textures S3TC encara que el suport de programari "
-"no estigui disponible"
-
-#: t_options.h:160
-msgid "Initial color reduction method"
-msgstr "Mètode inicial de reducció de color"
-
-#: t_options.h:161
-msgid "Round colors"
-msgstr "Colors arrodonits"
-
-#: t_options.h:162
-msgid "Dither colors"
-msgstr "Colors tramats"
-
-#: t_options.h:170
-msgid "Color rounding method"
-msgstr "Mètode d'arrodoniment de color"
-
-#: t_options.h:171
-msgid "Round color components downward"
-msgstr "Arrodoneix els components de color a baix"
-
-#: t_options.h:172
-msgid "Round to nearest color"
-msgstr "Arrodoneix al color més proper"
-
-#: t_options.h:181
-msgid "Color dithering method"
-msgstr "Mètode de tramat de color"
-
-#: t_options.h:182
-msgid "Horizontal error diffusion"
-msgstr "Difusió d'error horitzontal"
-
-#: t_options.h:183
-msgid "Horizontal error diffusion, reset error at line start"
-msgstr "Difusió d'error horitzontal, reinicia l'error a l'inici de la línia"
-
-#: t_options.h:184
-msgid "Ordered 2D color dithering"
-msgstr "Tramat de color 2D ordenat"
-
-#: t_options.h:190
-msgid "Floating point depth buffer"
-msgstr "Buffer de profunditat de punt flotant"
-
 #: t_options.h:195
 msgid "A post-processing filter to cel-shade the output"
 msgstr "Un filtre de postprocessament per a aplicar cel shading a la sortida"
@@ -212,46 +119,6 @@ msgstr ""
 msgid "Performance"
 msgstr "Rendiment"
 
-#: t_options.h:238
-msgid "TCL mode (Transformation, Clipping, Lighting)"
-msgstr "Mode TCL (Transformació, Retall, Il·luminació)"
-
-#: t_options.h:239
-msgid "Use software TCL pipeline"
-msgstr "Utilitza la canonada TCL de programari"
-
-#: t_options.h:240
-msgid "Use hardware TCL as first TCL pipeline stage"
-msgstr "Utilitza el TCL de maquinari com a la primera fase de la canonada TCL"
-
-#: t_options.h:241
-msgid "Bypass the TCL pipeline"
-msgstr "Passa per alt la canonada TCL"
-
-#: t_options.h:242
-msgid ""
-"Bypass the TCL pipeline with state-based machine code generated on-the-fly"
-msgstr ""
-"Passa per alt la canonada TCL amb codi de màquina basat en estats, generat "
-"sobre la marxa"
-
-#: t_options.h:251
-msgid "Method to limit rendering latency"
-msgstr "Mètode per a limitar la latència de renderització"
-
-#: t_options.h:252
-msgid "Busy waiting for the graphics hardware"
-msgstr "Espera activa pel maquinari de gràfics"
-
-#: t_options.h:253
-msgid "Sleep for brief intervals while waiting for the graphics hardware"
-msgstr "Dorm per intervals breus mentre s'espera al maquinari de gràfics"
-
-#: t_options.h:254
-msgid "Let the graphics hardware emit a software interrupt and sleep"
-msgstr ""
-"Deixa que el maquinari de gràfics emeti una interrupció de programari i dormi"
-
 #: t_options.h:264
 msgid "Synchronization with vertical refresh (swap intervals)"
 msgstr "Sincronització amb refresc vertical (intervals d'intercanvi)"
@@ -277,44 +144,6 @@ msgstr ""
 "Sempre sincronitza amb el refresc vertical, l'aplicació tria l'interval "
 "mínim d'intercanvi"
 
-#: t_options.h:276
-msgid "Use HyperZ to boost performance"
-msgstr "Utilitza el HyperZ per a augmentar el rendiment"
-
-#: t_options.h:281
-msgid "Number of texture units used"
-msgstr "Nombre d'unitats de textura utilitzades"
-
-#: t_options.h:286
-msgid "Texture filtering quality vs. speed, AKA “brilinear” texture filtering"
-msgstr ""
-"Qualitat vs. velocitat de filtració de textura, àlies filtració \"brilinear"
-"\" de textura"
-
-#: t_options.h:294
-msgid "Used types of texture memory"
-msgstr "Tipus utilitzats de memòria de textura"
-
-#: t_options.h:295
-msgid "All available memory"
-msgstr "Tota la memòria disponible"
-
-#: t_options.h:296
-msgid "Only card memory (if available)"
-msgstr "Només memòria de targeta (si està disponible)"
-
-#: t_options.h:297
-msgid "Only GART (AGP/PCIE) memory (if available)"
-msgstr "Només memòria GART (AGP/PCIE) (si està disponible)"
-
-#: t_options.h:309
-msgid "Features that are not hardware-accelerated"
-msgstr "Característiques no accelerades per maquinari"
-
-#: t_options.h:313
-msgid "Enable extension GL_ARB_vertex_program"
-msgstr "Habilita l'extensió GL_ARB_vertex_program"
-
 #: t_options.h:323
 msgid "Miscellaneous"
 msgstr "Miscel·lània"
diff --git a/lib/mesa/src/util/xmlpool/de.gmo b/lib/mesa/src/util/xmlpool/de.gmo
new file mode 100644
index 000000000..313e4b779
--- /dev/null
+++ b/lib/mesa/src/util/xmlpool/de.gmo
diff --git a/lib/mesa/src/util/xmlpool/de.po b/lib/mesa/src/util/xmlpool/de.po
index 7b20d00a6..8c1d3df3f 100644
--- a/lib/mesa/src/util/xmlpool/de.po
+++ b/lib/mesa/src/util/xmlpool/de.po
@@ -21,14 +21,6 @@ msgstr ""
 msgid "Debugging"
 msgstr "Fehlersuche"
 
-#: t_options.h:60
-msgid "Disable 3D acceleration"
-msgstr "3D-Beschleunigung abschalten"
-
-#: t_options.h:65
-msgid "Show performance boxes"
-msgstr "Zeige Performanceboxen"
-
 #: t_options.h:70
 msgid "Enable flushing batchbuffer after each draw call"
 msgstr "Aktiviere sofortige Leerung des Stapelpuffers nach jedem Zeichenaufruf"
@@ -54,10 +46,6 @@ msgstr ""
 msgid "Disable backslash-based line continuations in GLSL source"
 msgstr ""
 
-#: t_options.h:100
-msgid "Disable GL_ARB_shader_bit_encoding"
-msgstr ""
-
 #: t_options.h:105
 msgid ""
 "Force a default GLSL version for shaders that lack an explicit #version line"
@@ -71,85 +59,6 @@ msgstr ""
 msgid "Image Quality"
 msgstr "Bildqualität"
 
-#: t_options.h:133
-msgid "Texture color depth"
-msgstr "Texturfarbtiefe"
-
-#: t_options.h:134
-msgid "Prefer frame buffer color depth"
-msgstr "Bevorzuge Farbtiefe des Framebuffers"
-
-#: t_options.h:135
-msgid "Prefer 32 bits per texel"
-msgstr "Bevorzuge 32 bits pro Texel"
-
-#: t_options.h:136
-msgid "Prefer 16 bits per texel"
-msgstr "Bevorzuge 16 bits pro Texel"
-
-#: t_options.h:137
-msgid "Force 16 bits per texel"
-msgstr "Erzwinge 16 bits pro Texel"
-
-#: t_options.h:143
-msgid "Initial maximum value for anisotropic texture filtering"
-msgstr "Initialer Maximalwert für anisotropische Texturfilterung"
-
-#: t_options.h:148
-msgid "Forbid negative texture LOD bias"
-msgstr "Verbiete negative Textur-Detailgradverschiebung"
-
-#: t_options.h:153
-msgid ""
-"Enable S3TC texture compression even if software support is not available"
-msgstr ""
-"Aktiviere S3TC Texturkomprimierung auch wenn die nötige "
-"Softwareunterstützung fehlt"
-
-#: t_options.h:160
-msgid "Initial color reduction method"
-msgstr "Initiale Farbreduktionsmethode"
-
-#: t_options.h:161
-msgid "Round colors"
-msgstr "Farben runden"
-
-#: t_options.h:162
-msgid "Dither colors"
-msgstr "Farben rastern"
-
-#: t_options.h:170
-msgid "Color rounding method"
-msgstr "Farbrundungsmethode"
-
-#: t_options.h:171
-msgid "Round color components downward"
-msgstr "Farbkomponenten abrunden"
-
-#: t_options.h:172
-msgid "Round to nearest color"
-msgstr "Zur ähnlichsten Farbe runden"
-
-#: t_options.h:181
-msgid "Color dithering method"
-msgstr "Farbrasterungsmethode"
-
-#: t_options.h:182
-msgid "Horizontal error diffusion"
-msgstr "Horizontale Fehlerstreuung"
-
-#: t_options.h:183
-msgid "Horizontal error diffusion, reset error at line start"
-msgstr "Horizontale Fehlerstreuung, Fehler am Zeilenanfang zurücksetzen"
-
-#: t_options.h:184
-msgid "Ordered 2D color dithering"
-msgstr "Geordnete 2D Farbrasterung"
-
-#: t_options.h:190
-msgid "Floating point depth buffer"
-msgstr "Fließkomma z-Puffer"
-
 #: t_options.h:195
 msgid "A post-processing filter to cel-shade the output"
 msgstr "Nachbearbeitungsfilter für Cell Shading"
@@ -186,46 +95,6 @@ msgstr ""
 msgid "Performance"
 msgstr "Leistung"
 
-#: t_options.h:238
-msgid "TCL mode (Transformation, Clipping, Lighting)"
-msgstr "TCL-Modus (Transformation, Clipping, Licht)"
-
-#: t_options.h:239
-msgid "Use software TCL pipeline"
-msgstr "Benutze die Software-TCL-Pipeline"
-
-#: t_options.h:240
-msgid "Use hardware TCL as first TCL pipeline stage"
-msgstr "Benutze Hardware TCL als erste Stufe der TCL-Pipeline"
-
-#: t_options.h:241
-msgid "Bypass the TCL pipeline"
-msgstr "Umgehe die TCL-Pipeline"
-
-#: t_options.h:242
-msgid ""
-"Bypass the TCL pipeline with state-based machine code generated on-the-fly"
-msgstr ""
-"Umgehe die TCL-Pipeline mit zur Laufzeit erzeugtem, zustandsbasiertem "
-"Maschinencode"
-
-#: t_options.h:251
-msgid "Method to limit rendering latency"
-msgstr "Methode zur Begrenzung der Bildverzögerung"
-
-#: t_options.h:252
-msgid "Busy waiting for the graphics hardware"
-msgstr "Aktives Warten auf die Grafikhardware"
-
-#: t_options.h:253
-msgid "Sleep for brief intervals while waiting for the graphics hardware"
-msgstr "Kurze Schlafintervalle beim Warten auf die Grafikhardware"
-
-#: t_options.h:254
-msgid "Let the graphics hardware emit a software interrupt and sleep"
-msgstr ""
-"Die Grafikhardware eine Softwareunterbrechnung erzeugen lassen und schlafen"
-
 #: t_options.h:264
 msgid "Synchronization with vertical refresh (swap intervals)"
 msgstr "Synchronisation mit der vertikalen Bildwiederholung"
@@ -252,44 +121,6 @@ msgstr ""
 "Immer mit der Bildwiederholung synchronisieren, Anwendung wählt das minimale "
 "Bildintervall"
 
-#: t_options.h:276
-msgid "Use HyperZ to boost performance"
-msgstr "HyperZ zur Leistungssteigerung verwenden"
-
-#: t_options.h:281
-msgid "Number of texture units used"
-msgstr "Anzahl der benutzten Textureinheiten"
-
-#: t_options.h:286
-msgid "Texture filtering quality vs. speed, AKA “brilinear” texture filtering"
-msgstr ""
-"Texturfilterqualität versus -geschwindigkeit, auch bekannt als „brilineare“ "
-"Texturfilterung"
-
-#: t_options.h:294
-msgid "Used types of texture memory"
-msgstr "Benutzte Arten von Texturspeicher"
-
-#: t_options.h:295
-msgid "All available memory"
-msgstr "Aller verfügbarer Speicher"
-
-#: t_options.h:296
-msgid "Only card memory (if available)"
-msgstr "Nur Grafikspeicher (falls verfügbar)"
-
-#: t_options.h:297
-msgid "Only GART (AGP/PCIE) memory (if available)"
-msgstr "Nur GART-Speicher (AGP/PCIE) (falls verfügbar)"
-
-#: t_options.h:309
-msgid "Features that are not hardware-accelerated"
-msgstr "Funktionalität, die nicht hardwarebeschleunigt ist"
-
-#: t_options.h:313
-msgid "Enable extension GL_ARB_vertex_program"
-msgstr "Erweiterung GL_ARB_vertex_program aktivieren"
-
 #: t_options.h:323
 msgid "Miscellaneous"
 msgstr ""
diff --git a/lib/mesa/src/util/xmlpool/es.gmo b/lib/mesa/src/util/xmlpool/es.gmo
new file mode 100644
index 000000000..f4732492f
--- /dev/null
+++ b/lib/mesa/src/util/xmlpool/es.gmo
diff --git a/lib/mesa/src/util/xmlpool/es.po b/lib/mesa/src/util/xmlpool/es.po
index f9d950ac1..e5f44c8ef 100644
--- a/lib/mesa/src/util/xmlpool/es.po
+++ b/lib/mesa/src/util/xmlpool/es.po
@@ -24,14 +24,6 @@ msgstr ""
 msgid "Debugging"
 msgstr "Depuración"
 
-#: t_options.h:60
-msgid "Disable 3D acceleration"
-msgstr "Deshabilitar aceleración 3D"
-
-#: t_options.h:65
-msgid "Show performance boxes"
-msgstr "Mostrar cajas de rendimiento"
-
 #: t_options.h:70
 msgid "Enable flushing batchbuffer after each draw call"
 msgstr "Habilitar vaciado del batchbuffer después de cada llamada de dibujo"
@@ -59,10 +51,6 @@ msgstr ""
 "Deshabilitar continuaciones de línea basadas en barra inversa en el código "
 "GLSL"
 
-#: t_options.h:100
-msgid "Disable GL_ARB_shader_bit_encoding"
-msgstr "Deshabilitar GL_ARB_shader_bit_encoding"
-
 #: t_options.h:105
 msgid ""
 "Force a default GLSL version for shaders that lack an explicit #version line"
@@ -78,85 +66,6 @@ msgstr "Permite directivas #extension GLSL en medio de los shaders"
 msgid "Image Quality"
 msgstr "Calidad de imagen"
 
-#: t_options.h:133
-msgid "Texture color depth"
-msgstr "Profundidad de color de textura"
-
-#: t_options.h:134
-msgid "Prefer frame buffer color depth"
-msgstr "Preferir profundidad de color del framebuffer"
-
-#: t_options.h:135
-msgid "Prefer 32 bits per texel"
-msgstr "Preferir 32 bits por texel"
-
-#: t_options.h:136
-msgid "Prefer 16 bits per texel"
-msgstr "Preferir 16 bits por texel"
-
-#: t_options.h:137
-msgid "Force 16 bits per texel"
-msgstr "Forzar a 16 bits por texel"
-
-#: t_options.h:143
-msgid "Initial maximum value for anisotropic texture filtering"
-msgstr "Valor máximo inicial para filtrado anisotrópico de textura"
-
-#: t_options.h:148
-msgid "Forbid negative texture LOD bias"
-msgstr "Prohibir valores negativos de Nivel De Detalle (LOD) de texturas"
-
-#: t_options.h:153
-msgid ""
-"Enable S3TC texture compression even if software support is not available"
-msgstr ""
-"Habilitar la compresión de texturas S3TC incluso si el soporte por software "
-"no está disponible"
-
-#: t_options.h:160
-msgid "Initial color reduction method"
-msgstr "Método inicial de reducción de color"
-
-#: t_options.h:161
-msgid "Round colors"
-msgstr "Colores redondeados"
-
-#: t_options.h:162
-msgid "Dither colors"
-msgstr "Colores suavizados"
-
-#: t_options.h:170
-msgid "Color rounding method"
-msgstr "Método de redondeo de colores"
-
-#: t_options.h:171
-msgid "Round color components downward"
-msgstr "Redondear hacia abajo los componentes de color"
-
-#: t_options.h:172
-msgid "Round to nearest color"
-msgstr "Redondear al color más cercano"
-
-#: t_options.h:181
-msgid "Color dithering method"
-msgstr "Método de suavizado de color"
-
-#: t_options.h:182
-msgid "Horizontal error diffusion"
-msgstr "Difusión de error horizontal"
-
-#: t_options.h:183
-msgid "Horizontal error diffusion, reset error at line start"
-msgstr "Difusión de error horizontal, reiniciar error al comienzo de línea"
-
-#: t_options.h:184
-msgid "Ordered 2D color dithering"
-msgstr "Suavizado de color 2D ordenado"
-
-#: t_options.h:190
-msgid "Floating point depth buffer"
-msgstr "Búfer de profundidad en coma flotante"
-
 #: t_options.h:195
 msgid "A post-processing filter to cel-shade the output"
 msgstr "Un filtro de postprocesamiento para aplicar cel shading a la salida"
@@ -193,46 +102,6 @@ msgstr ""
 msgid "Performance"
 msgstr "Rendimiento"
 
-#: t_options.h:238
-msgid "TCL mode (Transformation, Clipping, Lighting)"
-msgstr "Modo TCL (Transformación, Recorte, Iluminación)"
-
-#: t_options.h:239
-msgid "Use software TCL pipeline"
-msgstr "Usar tubería TCL por software"
-
-#: t_options.h:240
-msgid "Use hardware TCL as first TCL pipeline stage"
-msgstr "Usar TCL por hardware en la primera fase de la tubería TCL"
-
-#: t_options.h:241
-msgid "Bypass the TCL pipeline"
-msgstr "Pasar por alto la tubería TCL"
-
-#: t_options.h:242
-msgid ""
-"Bypass the TCL pipeline with state-based machine code generated on-the-fly"
-msgstr ""
-"Pasar por alto la tubería TCL con código máquina basado en estados, generado "
-"al vuelo"
-
-#: t_options.h:251
-msgid "Method to limit rendering latency"
-msgstr "Método para limitar la latencia de renderización"
-
-#: t_options.h:252
-msgid "Busy waiting for the graphics hardware"
-msgstr "Esperar activamente al hardware gráfico"
-
-#: t_options.h:253
-msgid "Sleep for brief intervals while waiting for the graphics hardware"
-msgstr "Dormir en intervalos cortos mientras se espera al hardware gráfico"
-
-#: t_options.h:254
-msgid "Let the graphics hardware emit a software interrupt and sleep"
-msgstr ""
-"Permitir que el hardware gráfico emita una interrupción de software y duerma"
-
 #: t_options.h:264
 msgid "Synchronization with vertical refresh (swap intervals)"
 msgstr "Sincronización con el refresco vertical (intervalos de intercambio)"
@@ -261,44 +130,6 @@ msgstr ""
 "Sincronizar siempre con el refresco vertical, la aplicación elige el "
 "intervalo de intercambio mínimo"
 
-#: t_options.h:276
-msgid "Use HyperZ to boost performance"
-msgstr "Usar HyperZ para potenciar rendimiento"
-
-#: t_options.h:281
-msgid "Number of texture units used"
-msgstr "Número de unidades de textura usadas"
-
-#: t_options.h:286
-msgid "Texture filtering quality vs. speed, AKA “brilinear” texture filtering"
-msgstr ""
-"Calidad de filtrado de textura vs. velocidad, alias filtrado \"brilinear\" "
-"de textura"
-
-#: t_options.h:294
-msgid "Used types of texture memory"
-msgstr "Tipos de memoria de textura usados"
-
-#: t_options.h:295
-msgid "All available memory"
-msgstr "Toda la memoria disponible"
-
-#: t_options.h:296
-msgid "Only card memory (if available)"
-msgstr "Solo memoria de tarjeta (si está disponible)"
-
-#: t_options.h:297
-msgid "Only GART (AGP/PCIE) memory (if available)"
-msgstr "Solo memoria GART (AGP/PCIE) (si está disponible)"
-
-#: t_options.h:309
-msgid "Features that are not hardware-accelerated"
-msgstr "Características no aceleradas por hardware"
-
-#: t_options.h:313
-msgid "Enable extension GL_ARB_vertex_program"
-msgstr "Habilitar la extensión GL_ARB_vertex_program"
-
 #: t_options.h:323
 msgid "Miscellaneous"
 msgstr "Misceláneos"
diff --git a/lib/mesa/src/util/xmlpool/fr.gmo b/lib/mesa/src/util/xmlpool/fr.gmo
new file mode 100644
index 000000000..380d43cd9
--- /dev/null
+++ b/lib/mesa/src/util/xmlpool/fr.gmo
diff --git a/lib/mesa/src/util/xmlpool/fr.po b/lib/mesa/src/util/xmlpool/fr.po
index fa069652c..aa8706cbb 100644
--- a/lib/mesa/src/util/xmlpool/fr.po
+++ b/lib/mesa/src/util/xmlpool/fr.po
@@ -21,14 +21,6 @@ msgstr ""
 msgid "Debugging"
 msgstr "Debogage"
 
-#: t_options.h:60
-msgid "Disable 3D acceleration"
-msgstr "Désactiver l'accélération 3D"
-
-#: t_options.h:65
-msgid "Show performance boxes"
-msgstr "Afficher les boîtes de performance"
-
 #: t_options.h:70
 msgid "Enable flushing batchbuffer after each draw call"
 msgstr ""
@@ -53,10 +45,6 @@ msgstr ""
 msgid "Disable backslash-based line continuations in GLSL source"
 msgstr ""
 
-#: t_options.h:100
-msgid "Disable GL_ARB_shader_bit_encoding"
-msgstr ""
-
 #: t_options.h:105
 msgid ""
 "Force a default GLSL version for shaders that lack an explicit #version line"
@@ -70,84 +58,6 @@ msgstr ""
 msgid "Image Quality"
 msgstr "Qualité d'image"
 
-#: t_options.h:133
-msgid "Texture color depth"
-msgstr "Profondeur de texture"
-
-#: t_options.h:134
-msgid "Prefer frame buffer color depth"
-msgstr "Profondeur de couleur"
-
-#: t_options.h:135
-msgid "Prefer 32 bits per texel"
-msgstr "Préférer 32 bits par texel"
-
-#: t_options.h:136
-msgid "Prefer 16 bits per texel"
-msgstr "Prérérer 16 bits par texel"
-
-#: t_options.h:137
-msgid "Force 16 bits per texel"
-msgstr "Forcer 16 bits par texel"
-
-#: t_options.h:143
-msgid "Initial maximum value for anisotropic texture filtering"
-msgstr "Valeur maximale initiale pour le filtrage anisotropique de texture"
-
-#: t_options.h:148
-msgid "Forbid negative texture LOD bias"
-msgstr "Interdire le LOD bias negatif"
-
-#: t_options.h:153
-msgid ""
-"Enable S3TC texture compression even if software support is not available"
-msgstr ""
-"Activer la compression de texture S3TC même si le support logiciel est absent"
-
-#: t_options.h:160
-msgid "Initial color reduction method"
-msgstr "Technique de réduction de couleurs"
-
-#: t_options.h:161
-msgid "Round colors"
-msgstr "Arrondir les valeurs de couleur"
-
-#: t_options.h:162
-msgid "Dither colors"
-msgstr "Tramer les couleurs"
-
-#: t_options.h:170
-msgid "Color rounding method"
-msgstr "Méthode d'arrondi des couleurs"
-
-#: t_options.h:171
-msgid "Round color components downward"
-msgstr "Arrondi à l'inférieur"
-
-#: t_options.h:172
-msgid "Round to nearest color"
-msgstr "Arrondi au plus proche"
-
-#: t_options.h:181
-msgid "Color dithering method"
-msgstr "Méthode de tramage"
-
-#: t_options.h:182
-msgid "Horizontal error diffusion"
-msgstr "Diffusion d'erreur horizontale"
-
-#: t_options.h:183
-msgid "Horizontal error diffusion, reset error at line start"
-msgstr "Diffusion d'erreur horizontale, réinitialisé pour chaque ligne"
-
-#: t_options.h:184
-msgid "Ordered 2D color dithering"
-msgstr "Tramage ordonné des couleurs"
-
-#: t_options.h:190
-msgid "Floating point depth buffer"
-msgstr "Z-buffer en virgule flottante"
-
 #: t_options.h:195
 msgid "A post-processing filter to cel-shade the output"
 msgstr ""
@@ -180,45 +90,6 @@ msgstr ""
 msgid "Performance"
 msgstr "Performance"
 
-#: t_options.h:238
-msgid "TCL mode (Transformation, Clipping, Lighting)"
-msgstr "Mode de TCL (Transformation, Clipping, Eclairage)"
-
-#: t_options.h:239
-msgid "Use software TCL pipeline"
-msgstr "Utiliser un pipeline TCL logiciel"
-
-#: t_options.h:240
-msgid "Use hardware TCL as first TCL pipeline stage"
-msgstr "Utiliser le TCL matériel pour le premier niveau de pipeline"
-
-#: t_options.h:241
-msgid "Bypass the TCL pipeline"
-msgstr "Court-circuiter le pipeline TCL"
-
-#: t_options.h:242
-msgid ""
-"Bypass the TCL pipeline with state-based machine code generated on-the-fly"
-msgstr ""
-"Court-circuiter le pipeline TCL par une machine à états qui génère le codede "
-"TCL à la volée"
-
-#: t_options.h:251
-msgid "Method to limit rendering latency"
-msgstr "Méthode d'attente de la carte graphique"
-
-#: t_options.h:252
-msgid "Busy waiting for the graphics hardware"
-msgstr "Attente active de la carte graphique"
-
-#: t_options.h:253
-msgid "Sleep for brief intervals while waiting for the graphics hardware"
-msgstr "Attente utilisant usleep()"
-
-#: t_options.h:254
-msgid "Let the graphics hardware emit a software interrupt and sleep"
-msgstr "Utiliser les interruptions"
-
 #: t_options.h:264
 msgid "Synchronization with vertical refresh (swap intervals)"
 msgstr "Synchronisation de l'affichage avec le balayage vertical"
@@ -249,43 +120,6 @@ msgstr ""
 "Toujours synchroniser avec le balayage vertical, l'application choisit "
 "l'intervalle minimal"
 
-#: t_options.h:276
-msgid "Use HyperZ to boost performance"
-msgstr "Utiliser le HyperZ pour améliorer les performances"
-
-#: t_options.h:281
-msgid "Number of texture units used"
-msgstr "Nombre d'unités de texture"
-
-#: t_options.h:286
-msgid "Texture filtering quality vs. speed, AKA “brilinear” texture filtering"
-msgstr ""
-"Qualité/performance du filtrage trilinéaire de texture (filtrage brilinéaire)"
-
-#: t_options.h:294
-msgid "Used types of texture memory"
-msgstr "Types de mémoire de texture"
-
-#: t_options.h:295
-msgid "All available memory"
-msgstr "Utiliser toute la mémoire disponible"
-
-#: t_options.h:296
-msgid "Only card memory (if available)"
-msgstr "Utiliser uniquement la mémoire graphique (si disponible)"
-
-#: t_options.h:297
-msgid "Only GART (AGP/PCIE) memory (if available)"
-msgstr "Utiliser uniquement la mémoire GART (AGP/PCIE) (si disponible)"
-
-#: t_options.h:309
-msgid "Features that are not hardware-accelerated"
-msgstr "Fonctionnalités ne bénéficiant pas d'une accélération matérielle"
-
-#: t_options.h:313
-msgid "Enable extension GL_ARB_vertex_program"
-msgstr "Activer l'extension GL_ARB_vertex_program"
-
 #: t_options.h:323
 msgid "Miscellaneous"
 msgstr ""
diff --git a/lib/mesa/src/util/xmlpool/gen_xmlpool.py b/lib/mesa/src/util/xmlpool/gen_xmlpool.py
index eb68a6517..3a3523126 100644
--- a/lib/mesa/src/util/xmlpool/gen_xmlpool.py
+++ b/lib/mesa/src/util/xmlpool/gen_xmlpool.py
@@ -1,4 +1,4 @@
-
+# encoding=utf-8
 #
 # Usage:
 #     gen_xmlpool.py /path/to/t_option.h localedir lang lang lang ...
@@ -7,69 +7,68 @@
 # `{localedir}/{language}/LC_MESSAGES/options.mo`.
 #
 
-import sys
+from __future__ import print_function, unicode_literals
+import argparse
 import gettext
+import io
+import os
 import re
+import sys
 
-# Path to t_options.h
-template_header_path = sys.argv[1]
-
-localedir = sys.argv[2]
-
-# List of supported languages
-languages = sys.argv[3:]
+if sys.version_info < (3, 0):
+    gettext_method = 'ugettext'
+else:
+    gettext_method = 'gettext'
 
 # Escape special characters in C strings
-def escapeCString (s):
+def escapeCString(s):
     escapeSeqs = {'\a' : '\\a', '\b' : '\\b', '\f' : '\\f', '\n' : '\\n',
                   '\r' : '\\r', '\t' : '\\t', '\v' : '\\v', '\\' : '\\\\'}
     # " -> '' is a hack. Quotes (") aren't possible in XML attributes.
     # Better use Unicode characters for typographic quotes in option
     # descriptions and translations.
+    last_quote = '”'
     i = 0
     r = ''
-    while i < len(s):
-        # Special case: escape double quote with \u201c or \u201d, depending
+    for c in s:
+        # Special case: escape double quote with “ or ”, depending
         # on whether it's an open or close quote. This is needed because plain
         # double quotes are not possible in XML attributes.
-        if s[i] == '"':
-            if i == len(s)-1 or s[i+1].isspace():
-                # close quote
-                q = u'\u201c'
+        if c == '"':
+            if last_quote == '”':
+                q = '“'
             else:
-                # open quote
-                q = u'\u201d'
+                q = '”'
+            last_quote = q
             r = r + q
-        elif escapeSeqs.has_key(s[i]):
-            r = r + escapeSeqs[s[i]]
+        elif c in escapeSeqs:
+            r = r + escapeSeqs[c]
         else:
-            r = r + s[i]
-        i = i + 1
+            r = r + c
     return r
 
 # Expand escape sequences in C strings (needed for gettext lookup)
-def expandCString (s):
+def expandCString(s):
     escapeSeqs = {'a' : '\a', 'b' : '\b', 'f' : '\f', 'n' : '\n',
                   'r' : '\r', 't' : '\t', 'v' : '\v',
                   '"' : '"', '\\' : '\\'}
-    i = 0
     escape = False
     hexa = False
     octa = False
     num = 0
     digits = 0
-    r = ''
-    while i < len(s):
+    r = u''
+    for c in s:
         if not escape:
-            if s[i] == '\\':
+            if c == '\\':
                 escape = True
             else:
-                r = r + s[i]
+                r = r + c
         elif hexa:
-            if (s[i] >= '0' and s[i] <= '9') or \
-               (s[i] >= 'a' and s[i] <= 'f') or \
-               (s[i] >= 'A' and s[i] <= 'F'):
-                num = num * 16 + int(s[i],16)
+            if (c >= '0' and c <= '9') or \
+               (c >= 'a' and c <= 'f') or \
+               (c >= 'A' and c <= 'F'):
+                num = num * 16 + int(c, 16)
                 digits = digits + 1
             else:
                 digits = 2
@@ -78,8 +77,8 @@ def expandCString (s):
                 escape = False
                 r = r + chr(num)
         elif octa:
-            if s[i] >= '0' and s[i] <= '7':
-                num = num * 8 + int(s[i],8)
+            if c >= '0' and c <= '7':
+                num = num * 8 + int(c, 8)
                 digits = digits + 1
             else:
                 digits = 3
@@ -88,24 +87,23 @@ def expandCString (s):
                 escape = False
                 r = r + chr(num)
         else:
-            if escapeSeqs.has_key(s[i]):
-                r = r + escapeSeqs[s[i]]
+            if c in escapeSeqs:
+                r = r + escapeSeqs[c]
                 escape = False
-            elif s[i] >= '0' and s[i] <= '7':
+            elif c >= '0' and c <= '7':
                 octa = True
-                num = int(s[i],8)
+                num = int(c, 8)
                 if num <= 3:
                     digits = 1
                 else:
                     digits = 2
-            elif s[i] == 'x' or s[i] == 'X':
+            elif c == 'x' or c == 'X':
                 hexa = True
                 num = 0
                 digits = 0
             else:
-                r = r + s[i]
+                r = r + c
                 escape = False
-        i = i + 1
     return r
 
 # Expand matches. The first match is always a DESC or DESC_BEGIN match.
@@ -113,7 +111,7 @@ def expandCString (s):
 #
 # DESC, DESC_BEGIN format: \1 \2=<lang> \3 \4=gettext(" \5=<text> \6=") \7
 # ENUM format:             \1 \2=gettext(" \3=<text> \4=") \5
-def expandMatches (matches, translations, end=None):
+def expandMatches(matches, translations, outfile, end=None):
     assert len(matches) > 0
     nTranslations = len(translations)
     i = 0
@@ -124,80 +122,95 @@ def expandMatches (matches, translations, end=None):
         # are extended with a backslash.
         suffix = ''
         if len(matches) == 1 and i < len(translations) and \
-               not matches[0].expand (r'\7').endswith('\\'):
+               not matches[0].expand(r'\7').endswith('\\'):
             suffix = ' \\'
-        # Expand the description line. Need to use ugettext in order to allow
-        # non-ascii unicode chars in the original English descriptions.
-        text = escapeCString (trans.ugettext (unicode (expandCString (
-            matches[0].expand (r'\5')), "utf-8"))).encode("utf-8")
-        print matches[0].expand (r'\1' + lang + r'\3"' + text + r'"\7') + suffix
+        text = escapeCString(getattr(trans, gettext_method)(expandCString(
+            matches[0].expand (r'\5'))))
+        text = (matches[0].expand(r'\1' + lang + r'\3"' + text + r'"\7') + suffix)
+
+        outfile.write(text + '\n')
+
         # Expand any subsequent enum lines
         for match in matches[1:]:
-            text = escapeCString (trans.ugettext (unicode (expandCString (
-                match.expand (r'\3')), "utf-8"))).encode("utf-8")
-            print match.expand (r'\1"' + text + r'"\5')
+            text = escapeCString(getattr(trans, gettext_method)(expandCString(
+                match.expand(r'\3'))))
+            text = match.expand(r'\1"' + text + r'"\5')
+
+            outfile.write(text + '\n')
 
         # Expand description end
         if end:
-            print end,
-
-# Compile a list of translation classes to all supported languages.
-# The first translation is always a NullTranslations.
-translations = [("en", gettext.NullTranslations())]
-for lang in languages:
-    try:
-        trans = gettext.translation ("options", localedir, [lang])
-    except IOError:
-        sys.stderr.write ("Warning: language '%s' not found.\n" % lang)
-        continue
-    translations.append ((lang, trans))
+            outfile.write(end)
 
 # Regular expressions:
-reLibintl_h  = re.compile (r'#\s*include\s*<libintl.h>')
-reDESC       = re.compile (r'(\s*DRI_CONF_DESC\s*\(\s*)([a-z]+)(\s*,\s*)(gettext\s*\(\s*")(.*)("\s*\))(\s*\)[ \t]*\\?)$')
-reDESC_BEGIN = re.compile (r'(\s*DRI_CONF_DESC_BEGIN\s*\(\s*)([a-z]+)(\s*,\s*)(gettext\s*\(\s*")(.*)("\s*\))(\s*\)[ \t]*\\?)$')
-reENUM       = re.compile (r'(\s*DRI_CONF_ENUM\s*\([^,]+,\s*)(gettext\s*\(\s*")(.*)("\s*\))(\s*\)[ \t]*\\?)$')
-reDESC_END   = re.compile (r'\s*DRI_CONF_DESC_END')
-
-# Print a header
-print \
-"/***********************************************************************\n" \
-" ***        THIS FILE IS GENERATED AUTOMATICALLY. DON'T EDIT!        ***\n" \
-" ***********************************************************************/"
-
-# Process the options template and generate options.h with all
-# translations.
-template = file (template_header_path, "r")
-descMatches = []
-for line in template:
-    if len(descMatches) > 0:
-        matchENUM     = reENUM    .match (line)
-        matchDESC_END = reDESC_END.match (line)
-        if matchENUM:
-            descMatches.append (matchENUM)
-        elif matchDESC_END:
-            expandMatches (descMatches, translations, line)
+reLibintl_h = re.compile(r'#\s*include\s*<libintl.h>')
+reDESC = re.compile(r'(\s*DRI_CONF_DESC\s*\(\s*)([a-z]+)(\s*,\s*)(gettext\s*\(\s*")(.*)("\s*\))(\s*\)[ \t]*\\?)$')
+reDESC_BEGIN = re.compile(r'(\s*DRI_CONF_DESC_BEGIN\s*\(\s*)([a-z]+)(\s*,\s*)(gettext\s*\(\s*")(.*)("\s*\))(\s*\)[ \t]*\\?)$')
+reENUM = re.compile(r'(\s*DRI_CONF_ENUM\s*\([^,]+,\s*)(gettext\s*\(\s*")(.*)("\s*\))(\s*\)[ \t]*\\?)$')
+reDESC_END = re.compile(r'\s*DRI_CONF_DESC_END')
+
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--template', required=True)
+    parser.add_argument('--output', required=True)
+    parser.add_argument('--localedir', required=True)
+    parser.add_argument('--languages', nargs='*', default=[])
+    args = parser.parse_args()
+
+    # Compile a list of translation classes to all supported languages.
+    # The first translation is always a NullTranslations.
+    translations = [("en", gettext.NullTranslations())]
+    for lang in args.languages:
+        try:
+            filename = os.path.join(args.localedir, '{}.gmo'.format(lang))
+            with io.open(filename, 'rb') as f:
+                trans = gettext.GNUTranslations(f)
+        except (IOError, OSError):
+            print("Warning: language '%s' not found." % lang, file=sys.stderr)
+            continue
+        translations.append((lang, trans))
+
+    with io.open(args.output, mode='wt', encoding='utf-8') as output:
+        output.write("/* This is file is generated automatically. Don't edit!  */\n")
+
+        # Process the options template and generate options.h with all
+        # translations.
+        with io.open(args.template, mode="rt", encoding='utf-8') as template:
             descMatches = []
-        else:
-            sys.stderr.write (
-                "Warning: unexpected line inside description dropped:\n%s\n" \
-                % line)
-        continue
-    if reLibintl_h.search (line):
-        # Ignore (comment out) #include <libintl.h>
-        print "/* %s * commented out by gen_xmlpool.py */" % line
-        continue
-    matchDESC       = reDESC      .match (line)
-    matchDESC_BEGIN = reDESC_BEGIN.match (line)
-    if matchDESC:
-        assert len(descMatches) == 0
-        expandMatches ([matchDESC], translations)
-    elif matchDESC_BEGIN:
-        assert len(descMatches) == 0
-        descMatches = [matchDESC_BEGIN]
-    else:
-        print line,
-
-if len(descMatches) > 0:
-    sys.stderr.write ("Warning: unterminated description at end of file.\n")
-    expandMatches (descMatches, translations)
+            for line in template:
+                if descMatches:
+                    matchENUM = reENUM.match(line)
+                    matchDESC_END = reDESC_END.match(line)
+                    if matchENUM:
+                        descMatches.append(matchENUM)
+                    elif matchDESC_END:
+                        expandMatches(descMatches, translations, output, line)
+                        descMatches = []
+                    else:
+                        print("Warning: unexpected line inside description dropped:\n",
+                              line, file=sys.stderr)
+                    continue
+                if reLibintl_h.search(line):
+                    # Ignore (comment out) #include <libintl.h>
+                    output.write("/* %s * commented out by gen_xmlpool.py */\n" % line)
+                    continue
+                matchDESC = reDESC.match(line)
+                matchDESC_BEGIN = reDESC_BEGIN.match(line)
+                if matchDESC:
+                    assert not descMatches
+                    expandMatches([matchDESC], translations, output)
+                elif matchDESC_BEGIN:
+                    assert not descMatches
+                    descMatches = [matchDESC_BEGIN]
+                else:
+
+                    output.write(line)
+
+        if descMatches:
+            print("Warning: unterminated description at end of file.", file=sys.stderr)
+            expandMatches(descMatches, translations, output)
+
+
+if __name__ == '__main__':
+    main()
diff --git a/lib/mesa/src/util/xmlpool/meson.build b/lib/mesa/src/util/xmlpool/meson.build
new file mode 100644
index 000000000..81184fa6b
--- /dev/null
+++ b/lib/mesa/src/util/xmlpool/meson.build
@@ -0,0 +1,40 @@
+# Copyright © 2017 Intel Corporation
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in
+# all copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+_langs = ['ca', 'es', 'de', 'nl', 'sv', 'fr']
+
+_langs_po_files = []
+foreach lang : _langs
+  _langs_po_files += files(lang + '.po')
+endforeach
+
+xmlpool_options_h = custom_target(
+  'xmlpool_options.h',
+  input : ['gen_xmlpool.py', 't_options.h'],
+  output : 'options.h',
+  command : [
+    prog_python, '@INPUT0@', '--template', '@INPUT1@', '--output', '@OUTPUT@',
+    '--localedir', meson.current_build_dir(), '--languages',  _langs,
+  ],
+  depend_files : _langs_po_files,
+)
+
+i18n = import('i18n')
+i18n.gettext('xmlpool', install : false)
diff --git a/lib/mesa/src/util/xmlpool/nl.gmo b/lib/mesa/src/util/xmlpool/nl.gmo
new file mode 100644
index 000000000..a67fb5a8a
--- /dev/null
+++ b/lib/mesa/src/util/xmlpool/nl.gmo
diff --git a/lib/mesa/src/util/xmlpool/nl.po b/lib/mesa/src/util/xmlpool/nl.po
index 86cb6e96d..50a53cc9c 100644
--- a/lib/mesa/src/util/xmlpool/nl.po
+++ b/lib/mesa/src/util/xmlpool/nl.po
@@ -21,14 +21,6 @@ msgstr ""
 msgid "Debugging"
 msgstr "Debuggen"
 
-#: t_options.h:60
-msgid "Disable 3D acceleration"
-msgstr "3D versnelling uitschakelen"
-
-#: t_options.h:65
-msgid "Show performance boxes"
-msgstr "Laat prestatie boxjes zien"
-
 #: t_options.h:70
 msgid "Enable flushing batchbuffer after each draw call"
 msgstr ""
@@ -53,10 +45,6 @@ msgstr ""
 msgid "Disable backslash-based line continuations in GLSL source"
 msgstr ""
 
-#: t_options.h:100
-msgid "Disable GL_ARB_shader_bit_encoding"
-msgstr ""
-
 #: t_options.h:105
 msgid ""
 "Force a default GLSL version for shaders that lack an explicit #version line"
@@ -70,69 +58,6 @@ msgstr ""
 msgid "Image Quality"
 msgstr "Beeldkwaliteit"
 
-#: t_options.h:133
-msgid "Texture color depth"
-msgstr "Textuurkleurendiepte"
-
-#: t_options.h:134
-msgid "Prefer frame buffer color depth"
-msgstr "Prefereer kaderbufferkleurdiepte"
-
-#: t_options.h:135
-msgid "Prefer 32 bits per texel"
-msgstr "Prefereer 32 bits per texel"
-
-#: t_options.h:136
-msgid "Prefer 16 bits per texel"
-msgstr "Prefereer 16 bits per texel"
-
-#: t_options.h:137
-msgid "Force 16 bits per texel"
-msgstr "Dwing 16 bits per texel af"
-
-#: t_options.h:143
-msgid "Initial maximum value for anisotropic texture filtering"
-msgstr "Initïele maximum waarde voor anisotrophische textuur filtering"
-
-#: t_options.h:148
-msgid "Forbid negative texture LOD bias"
-msgstr "Verbied negatief niveau detailonderscheid (LOD) van texturen"
-
-#: t_options.h:153
-msgid ""
-"Enable S3TC texture compression even if software support is not available"
-msgstr ""
-"Schakel S3TC textuurcompressie in, zelfs als softwareondersteuning niet "
-"aanwezig is"
-
-#: t_options.h:160
-msgid "Initial color reduction method"
-msgstr "Initïele kleurreductie methode"
-
-#: t_options.h:161
-msgid "Round colors"
-msgstr "Rond kleuren af"
-
-#: t_options.h:162
-msgid "Dither colors"
-msgstr "Rasteriseer kleuren"
-
-#: t_options.h:170
-msgid "Color rounding method"
-msgstr "Kleurafrondingmethode"
-
-#: t_options.h:171
-msgid "Round color components downward"
-msgstr "Rond kleurencomponenten af naar beneden"
-
-#: t_options.h:172
-msgid "Round to nearest color"
-msgstr "Rond af naar dichtsbijzijnde kleur"
-
-#: t_options.h:181
-msgid "Color dithering method"
-msgstr "Kleurrasteriseringsmethode"
-
 #: t_options.h:182
 msgid "Horizontal error diffusion"
 msgstr "Horizontale foutdiffusie"
@@ -145,10 +70,6 @@ msgstr "Horizontale foutdiffusie, zet fout bij lijnbegin terug"
 msgid "Ordered 2D color dithering"
 msgstr "Geordende 2D kleurrasterisering"
 
-#: t_options.h:190
-msgid "Floating point depth buffer"
-msgstr "Dieptebuffer als commagetal"
-
 #: t_options.h:195
 msgid "A post-processing filter to cel-shade the output"
 msgstr ""
@@ -181,48 +102,6 @@ msgstr ""
 msgid "Performance"
 msgstr "Prestatie"
 
-#: t_options.h:238
-msgid "TCL mode (Transformation, Clipping, Lighting)"
-msgstr "TCL-modus (Transformatie, Clipping, Licht)"
-
-#: t_options.h:239
-msgid "Use software TCL pipeline"
-msgstr "Gebruik software TCL pijpleiding"
-
-#: t_options.h:240
-msgid "Use hardware TCL as first TCL pipeline stage"
-msgstr "Gebruik hardware TCL as eerste TCL pijpleiding trap"
-
-#: t_options.h:241
-msgid "Bypass the TCL pipeline"
-msgstr "Omzeil de TCL pijpleiding"
-
-#: t_options.h:242
-msgid ""
-"Bypass the TCL pipeline with state-based machine code generated on-the-fly"
-msgstr ""
-"Omzeil de TCL pijpleiding met staatgebaseerde machinecode die tijdens "
-"executie gegenereerd wordt"
-
-#: t_options.h:251
-msgid "Method to limit rendering latency"
-msgstr "Methode om beeldopbouwvertraging te onderdrukken"
-
-#: t_options.h:252
-msgid "Busy waiting for the graphics hardware"
-msgstr "Actief wachten voor de grafische hardware"
-
-#: t_options.h:253
-msgid "Sleep for brief intervals while waiting for the graphics hardware"
-msgstr ""
-"Slaap voor korte intervallen tijdens het wachten op de grafische hardware"
-
-#: t_options.h:254
-msgid "Let the graphics hardware emit a software interrupt and sleep"
-msgstr ""
-"Laat de grafische hardware een software onderbreking uitzenden en in slaap "
-"vallen"
-
 #: t_options.h:264
 msgid "Synchronization with vertical refresh (swap intervals)"
 msgstr "Synchronisatie met verticale verversing (interval omwisselen)"
@@ -249,44 +128,6 @@ msgstr ""
 "Synchroniseer altijd met verticale verversing, de applicatie kiest het "
 "minimum omwisselingsinterval"
 
-#: t_options.h:276
-msgid "Use HyperZ to boost performance"
-msgstr "Gebruik HyperZ om de prestaties te verbeteren"
-
-#: t_options.h:281
-msgid "Number of texture units used"
-msgstr "Aantal textuureenheden in gebruik"
-
-#: t_options.h:286
-msgid "Texture filtering quality vs. speed, AKA “brilinear” texture filtering"
-msgstr ""
-"Textuurfilterkwaliteit versus -snelheid, ookwel bekend als “brilineaire” "
-"textuurfiltering"
-
-#: t_options.h:294
-msgid "Used types of texture memory"
-msgstr "Gebruikte soorten textuurgeheugen"
-
-#: t_options.h:295
-msgid "All available memory"
-msgstr "Al het beschikbaar geheugen"
-
-#: t_options.h:296
-msgid "Only card memory (if available)"
-msgstr "Alleen geheugen op de kaart (als het aanwezig is)"
-
-#: t_options.h:297
-msgid "Only GART (AGP/PCIE) memory (if available)"
-msgstr "Alleen GART (AGP/PCIE) geheugen (als het aanwezig is)"
-
-#: t_options.h:309
-msgid "Features that are not hardware-accelerated"
-msgstr "Eigenschappen die niet hardwareversneld zijn"
-
-#: t_options.h:313
-msgid "Enable extension GL_ARB_vertex_program"
-msgstr "Zet uitbreiding GL_ARB_vertex_program aan"
-
 #: t_options.h:323
 msgid "Miscellaneous"
 msgstr ""
diff --git a/lib/mesa/src/util/xmlpool/options.h b/lib/mesa/src/util/xmlpool/options.h
index ebf969d46..0c3fa9e4a 100644
--- a/lib/mesa/src/util/xmlpool/options.h
+++ b/lib/mesa/src/util/xmlpool/options.h
@@ -1,6 +1,4 @@
-/***********************************************************************
- ***        THIS FILE IS GENERATED AUTOMATICALLY. DON'T EDIT!        ***
- ***********************************************************************/
+/* This is file is generated automatically. Don't edit!  */
 /*
  * XML DRI client-side driver configuration
  * Copyright (C) 2003 Felix Kuehling
@@ -57,35 +55,13 @@
  */
 #define DRI_CONF_SECTION_DEBUG \
 DRI_CONF_SECTION_BEGIN \
-	DRI_CONF_DESC(en,"Debugging") \
-	DRI_CONF_DESC(ca,"Depuració") \
-	DRI_CONF_DESC(de,"Fehlersuche") \
-	DRI_CONF_DESC(es,"Depuración") \
-	DRI_CONF_DESC(nl,"Debuggen") \
-	DRI_CONF_DESC(fr,"Debogage") \
-	DRI_CONF_DESC(sv,"Felsökning")
-
-#define DRI_CONF_NO_RAST(def) \
-DRI_CONF_OPT_BEGIN_B(no_rast, def) \
-        DRI_CONF_DESC(en,"Disable 3D acceleration") \
-        DRI_CONF_DESC(ca,"Deshabilita l'acceleració 3D") \
-        DRI_CONF_DESC(de,"3D-Beschleunigung abschalten") \
-        DRI_CONF_DESC(es,"Deshabilitar aceleración 3D") \
-        DRI_CONF_DESC(nl,"3D versnelling uitschakelen") \
-        DRI_CONF_DESC(fr,"Désactiver l'accélération 3D") \
-        DRI_CONF_DESC(sv,"Inaktivera 3D-accelerering") \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_PERFORMANCE_BOXES(def) \
-DRI_CONF_OPT_BEGIN_B(performance_boxes, def) \
-        DRI_CONF_DESC(en,"Show performance boxes") \
-        DRI_CONF_DESC(ca,"Mostra les caixes de rendiment") \
-        DRI_CONF_DESC(de,"Zeige Performanceboxen") \
-        DRI_CONF_DESC(es,"Mostrar cajas de rendimiento") \
-        DRI_CONF_DESC(nl,"Laat prestatie boxjes zien") \
-        DRI_CONF_DESC(fr,"Afficher les boîtes de performance") \
-        DRI_CONF_DESC(sv,"Visa prestandarutor") \
-DRI_CONF_OPT_END
+        DRI_CONF_DESC(en,"Debugging") \
+        DRI_CONF_DESC(ca,"Depuració") \
+        DRI_CONF_DESC(de,"Fehlersuche") \
+        DRI_CONF_DESC(es,"Depuración") \
+        DRI_CONF_DESC(nl,"Debuggen") \
+        DRI_CONF_DESC(fr,"Debogage") \
+        DRI_CONF_DESC(sv,"Felsökning")
 
 #define DRI_CONF_ALWAYS_FLUSH_BATCH(def) \
 DRI_CONF_OPT_BEGIN_B(always_flush_batch, def) \
@@ -111,13 +87,13 @@ DRI_CONF_OPT_END
 
 #define DRI_CONF_DISABLE_THROTTLING(def) \
 DRI_CONF_OPT_BEGIN_B(disable_throttling, def) \
-	DRI_CONF_DESC(en,"Disable throttling on first batch after flush") \
-	DRI_CONF_DESC(ca,"Deshabilita la regulació en el primer lot després de buidar") \
-	DRI_CONF_DESC(de,"Disable throttling on first batch after flush") \
-	DRI_CONF_DESC(es,"Deshabilitar regulación del primer lote después de vaciar") \
-	DRI_CONF_DESC(nl,"Disable throttling on first batch after flush") \
-	DRI_CONF_DESC(fr,"Disable throttling on first batch after flush") \
-	DRI_CONF_DESC(sv,"Disable throttling on first batch after flush") \
+        DRI_CONF_DESC(en,"Disable throttling on first batch after flush") \
+        DRI_CONF_DESC(ca,"Deshabilita la regulació en el primer lot després de buidar") \
+        DRI_CONF_DESC(de,"Disable throttling on first batch after flush") \
+        DRI_CONF_DESC(es,"Deshabilitar regulación del primer lote después de vaciar") \
+        DRI_CONF_DESC(nl,"Disable throttling on first batch after flush") \
+        DRI_CONF_DESC(fr,"Disable throttling on first batch after flush") \
+        DRI_CONF_DESC(sv,"Disable throttling on first batch after flush") \
 DRI_CONF_OPT_END
 
 #define DRI_CONF_FORCE_GLSL_EXTENSIONS_WARN(def) \
@@ -164,17 +140,6 @@ DRI_CONF_OPT_BEGIN_B(disable_glsl_line_continuations, def) \
         DRI_CONF_DESC(sv,"Disable backslash-based line continuations in GLSL source") \
 DRI_CONF_OPT_END
 
-#define DRI_CONF_DISABLE_SHADER_BIT_ENCODING(def) \
-DRI_CONF_OPT_BEGIN_B(disable_shader_bit_encoding, def) \
-        DRI_CONF_DESC(en,"Disable GL_ARB_shader_bit_encoding") \
-        DRI_CONF_DESC(ca,"Deshabilita el GL_ARB_shader_bit_encoding") \
-        DRI_CONF_DESC(de,"Disable GL_ARB_shader_bit_encoding") \
-        DRI_CONF_DESC(es,"Deshabilitar GL_ARB_shader_bit_encoding") \
-        DRI_CONF_DESC(nl,"Disable GL_ARB_shader_bit_encoding") \
-        DRI_CONF_DESC(fr,"Disable GL_ARB_shader_bit_encoding") \
-        DRI_CONF_DESC(sv,"Disable GL_ARB_shader_bit_encoding") \
-DRI_CONF_OPT_END
-
 #define DRI_CONF_FORCE_GLSL_VERSION(def) \
 DRI_CONF_OPT_BEGIN_V(force_glsl_version, int, def, "0:999") \
         DRI_CONF_DESC(en,"Force a default GLSL version for shaders that lack an explicit #version line") \
@@ -197,6 +162,28 @@ DRI_CONF_OPT_BEGIN_B(allow_glsl_extension_directive_midshader, def) \
         DRI_CONF_DESC(sv,"Allow GLSL #extension directives in the middle of shaders") \
 DRI_CONF_OPT_END
 
+#define DRI_CONF_ALLOW_GLSL_BUILTIN_CONST_EXPRESSION(def) \
+DRI_CONF_OPT_BEGIN_B(allow_glsl_builtin_const_expression, def) \
+        DRI_CONF_DESC(en,"Allow builtins as part of constant expressions") \
+        DRI_CONF_DESC(ca,"Allow builtins as part of constant expressions") \
+        DRI_CONF_DESC(de,"Allow builtins as part of constant expressions") \
+        DRI_CONF_DESC(es,"Allow builtins as part of constant expressions") \
+        DRI_CONF_DESC(nl,"Allow builtins as part of constant expressions") \
+        DRI_CONF_DESC(fr,"Allow builtins as part of constant expressions") \
+        DRI_CONF_DESC(sv,"Allow builtins as part of constant expressions") \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_ALLOW_GLSL_RELAXED_ES(def) \
+DRI_CONF_OPT_BEGIN_B(allow_glsl_relaxed_es, def) \
+        DRI_CONF_DESC(en,"Allow some relaxation of GLSL ES shader restrictions") \
+        DRI_CONF_DESC(ca,"Allow some relaxation of GLSL ES shader restrictions") \
+        DRI_CONF_DESC(de,"Allow some relaxation of GLSL ES shader restrictions") \
+        DRI_CONF_DESC(es,"Allow some relaxation of GLSL ES shader restrictions") \
+        DRI_CONF_DESC(nl,"Allow some relaxation of GLSL ES shader restrictions") \
+        DRI_CONF_DESC(fr,"Allow some relaxation of GLSL ES shader restrictions") \
+        DRI_CONF_DESC(sv,"Allow some relaxation of GLSL ES shader restrictions") \
+DRI_CONF_OPT_END
+
 #define DRI_CONF_ALLOW_GLSL_BUILTIN_VARIABLE_REDECLARATION(def) \
 DRI_CONF_OPT_BEGIN_B(allow_glsl_builtin_variable_redeclaration, def) \
         DRI_CONF_DESC(en,"Allow GLSL built-in variables to be redeclared verbatim") \
@@ -252,95 +239,40 @@ DRI_CONF_OPT_BEGIN_B(allow_glsl_cross_stage_interpolation_mismatch, def) \
         DRI_CONF_DESC(sv,"Allow interpolation qualifier mismatch across shader stages") \
 DRI_CONF_OPT_END
 
+#define DRI_CONF_ALLOW_GLSL_LAYOUT_QUALIFIER_ON_FUNCTION_PARAMETERS(def) \
+DRI_CONF_OPT_BEGIN_B(allow_glsl_layout_qualifier_on_function_parameters, def) \
+        DRI_CONF_DESC(en,"Allow layout qualifiers on function parameters.") \
+        DRI_CONF_DESC(ca,"Allow layout qualifiers on function parameters.") \
+        DRI_CONF_DESC(de,"Allow layout qualifiers on function parameters.") \
+        DRI_CONF_DESC(es,"Allow layout qualifiers on function parameters.") \
+        DRI_CONF_DESC(nl,"Allow layout qualifiers on function parameters.") \
+        DRI_CONF_DESC(fr,"Allow layout qualifiers on function parameters.") \
+        DRI_CONF_DESC(sv,"Allow layout qualifiers on function parameters.") \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_FORCE_COMPAT_PROFILE(def) \
+DRI_CONF_OPT_BEGIN_B(force_compat_profile, def) \
+        DRI_CONF_DESC(en,"Force an OpenGL compatibility context") \
+        DRI_CONF_DESC(ca,"Force an OpenGL compatibility context") \
+        DRI_CONF_DESC(de,"Force an OpenGL compatibility context") \
+        DRI_CONF_DESC(es,"Force an OpenGL compatibility context") \
+        DRI_CONF_DESC(nl,"Force an OpenGL compatibility context") \
+        DRI_CONF_DESC(fr,"Force an OpenGL compatibility context") \
+        DRI_CONF_DESC(sv,"Force an OpenGL compatibility context") \
+DRI_CONF_OPT_END
+
 /**
  * \brief Image quality-related options
  */
 #define DRI_CONF_SECTION_QUALITY \
 DRI_CONF_SECTION_BEGIN \
-	DRI_CONF_DESC(en,"Image Quality") \
-	DRI_CONF_DESC(ca,"Qualitat d'imatge") \
-	DRI_CONF_DESC(de,"Bildqualität") \
-	DRI_CONF_DESC(es,"Calidad de imagen") \
-	DRI_CONF_DESC(nl,"Beeldkwaliteit") \
-	DRI_CONF_DESC(fr,"Qualité d'image") \
-	DRI_CONF_DESC(sv,"Bildkvalitet")
-
-#define DRI_CONF_EXCESS_MIPMAP(def) \
-DRI_CONF_OPT_BEGIN_B(excess_mipmap, def) \
-	DRI_CONF_DESC(en,"Enable extra mipmap level") \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_TEXTURE_DEPTH_FB       0
-#define DRI_CONF_TEXTURE_DEPTH_32       1
-#define DRI_CONF_TEXTURE_DEPTH_16       2
-#define DRI_CONF_TEXTURE_DEPTH_FORCE_16 3
-#define DRI_CONF_TEXTURE_DEPTH(def) \
-DRI_CONF_OPT_BEGIN_V(texture_depth,enum,def,"0:3") \
-	DRI_CONF_DESC_BEGIN(en,"Texture color depth") \
-                DRI_CONF_ENUM(0,"Prefer frame buffer color depth") \
-                DRI_CONF_ENUM(1,"Prefer 32 bits per texel") \
-                DRI_CONF_ENUM(2,"Prefer 16 bits per texel") \
-                DRI_CONF_ENUM(3,"Force 16 bits per texel") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(ca,"Profunditat de color de textura") \
-                DRI_CONF_ENUM(0,"Prefereix profunditat de color del framebuffer") \
-                DRI_CONF_ENUM(1,"Prefereix 32 bits per texel") \
-                DRI_CONF_ENUM(2,"Prefereix 16 bits per texel") \
-                DRI_CONF_ENUM(3,"Força 16 bits per texel") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(de,"Texturfarbtiefe") \
-                DRI_CONF_ENUM(0,"Bevorzuge Farbtiefe des Framebuffers") \
-                DRI_CONF_ENUM(1,"Bevorzuge 32 bits pro Texel") \
-                DRI_CONF_ENUM(2,"Bevorzuge 16 bits pro Texel") \
-                DRI_CONF_ENUM(3,"Erzwinge 16 bits pro Texel") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(es,"Profundidad de color de textura") \
-                DRI_CONF_ENUM(0,"Preferir profundidad de color del framebuffer") \
-                DRI_CONF_ENUM(1,"Preferir 32 bits por texel") \
-                DRI_CONF_ENUM(2,"Preferir 16 bits por texel") \
-                DRI_CONF_ENUM(3,"Forzar a 16 bits por texel") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(nl,"Textuurkleurendiepte") \
-                DRI_CONF_ENUM(0,"Prefereer kaderbufferkleurdiepte") \
-                DRI_CONF_ENUM(1,"Prefereer 32 bits per texel") \
-                DRI_CONF_ENUM(2,"Prefereer 16 bits per texel") \
-                DRI_CONF_ENUM(3,"Dwing 16 bits per texel af") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(fr,"Profondeur de texture") \
-                DRI_CONF_ENUM(0,"Profondeur de couleur") \
-                DRI_CONF_ENUM(1,"Préférer 32 bits par texel") \
-                DRI_CONF_ENUM(2,"Prérérer 16 bits par texel") \
-                DRI_CONF_ENUM(3,"Forcer 16 bits par texel") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(sv,"Färgdjup för texturer") \
-                DRI_CONF_ENUM(0,"Föredra färgdjupet för framebuffer") \
-                DRI_CONF_ENUM(1,"Föredra 32 bitar per texel") \
-                DRI_CONF_ENUM(2,"Föredra 16 bitar per texel") \
-                DRI_CONF_ENUM(3,"Tvinga 16 bitar per texel") \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_DEF_MAX_ANISOTROPY(def,range) \
-DRI_CONF_OPT_BEGIN_V(def_max_anisotropy,float,def,range) \
-        DRI_CONF_DESC(en,"Initial maximum value for anisotropic texture filtering") \
-        DRI_CONF_DESC(ca,"Valor màxim inicial per a la filtració de textura anisòtropa") \
-        DRI_CONF_DESC(de,"Initialer Maximalwert für anisotropische Texturfilterung") \
-        DRI_CONF_DESC(es,"Valor máximo inicial para filtrado anisotrópico de textura") \
-        DRI_CONF_DESC(nl,"Initïele maximum waarde voor anisotrophische textuur filtering") \
-        DRI_CONF_DESC(fr,"Valeur maximale initiale pour le filtrage anisotropique de texture") \
-        DRI_CONF_DESC(sv,"Initialt maximalt värde för anisotropisk texturfiltrering") \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_NO_NEG_LOD_BIAS(def) \
-DRI_CONF_OPT_BEGIN_B(no_neg_lod_bias, def) \
-        DRI_CONF_DESC(en,"Forbid negative texture LOD bias") \
-        DRI_CONF_DESC(ca,"Prohibeix una parcialitat negativa del Nivell de Detalle (LOD) de les textures") \
-        DRI_CONF_DESC(de,"Verbiete negative Textur-Detailgradverschiebung") \
-        DRI_CONF_DESC(es,"Prohibir valores negativos de Nivel De Detalle (LOD) de texturas") \
-        DRI_CONF_DESC(nl,"Verbied negatief niveau detailonderscheid (LOD) van texturen") \
-        DRI_CONF_DESC(fr,"Interdire le LOD bias negatif") \
-        DRI_CONF_DESC(sv,"Förbjud negativ LOD-kompensation för texturer") \
-DRI_CONF_OPT_END
+        DRI_CONF_DESC(en,"Image Quality") \
+        DRI_CONF_DESC(ca,"Qualitat d'imatge") \
+        DRI_CONF_DESC(de,"Bildqualität") \
+        DRI_CONF_DESC(es,"Calidad de imagen") \
+        DRI_CONF_DESC(nl,"Beeldkwaliteit") \
+        DRI_CONF_DESC(fr,"Qualité d'image") \
+        DRI_CONF_DESC(sv,"Bildkvalitet")
 
 #define DRI_CONF_PRECISE_TRIG(def) \
 DRI_CONF_OPT_BEGIN_B(precise_trig, def) \
@@ -353,127 +285,6 @@ DRI_CONF_OPT_BEGIN_B(precise_trig, def) \
         DRI_CONF_DESC(sv,"Prefer accuracy over performance in trig functions") \
 DRI_CONF_OPT_END
 
-#define DRI_CONF_COLOR_REDUCTION_ROUND 0
-#define DRI_CONF_COLOR_REDUCTION_DITHER 1
-#define DRI_CONF_COLOR_REDUCTION(def) \
-DRI_CONF_OPT_BEGIN_V(color_reduction,enum,def,"0:1") \
-        DRI_CONF_DESC_BEGIN(en,"Initial color reduction method") \
-                DRI_CONF_ENUM(0,"Round colors") \
-                DRI_CONF_ENUM(1,"Dither colors") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(ca,"Mètode inicial de reducció de color") \
-                DRI_CONF_ENUM(0,"Colors arrodonits") \
-                DRI_CONF_ENUM(1,"Colors tramats") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(de,"Initiale Farbreduktionsmethode") \
-                DRI_CONF_ENUM(0,"Farben runden") \
-                DRI_CONF_ENUM(1,"Farben rastern") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(es,"Método inicial de reducción de color") \
-                DRI_CONF_ENUM(0,"Colores redondeados") \
-                DRI_CONF_ENUM(1,"Colores suavizados") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(nl,"Initïele kleurreductie methode") \
-                DRI_CONF_ENUM(0,"Rond kleuren af") \
-                DRI_CONF_ENUM(1,"Rasteriseer kleuren") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(fr,"Technique de réduction de couleurs") \
-                DRI_CONF_ENUM(0,"Arrondir les valeurs de couleur") \
-                DRI_CONF_ENUM(1,"Tramer les couleurs") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(sv,"Initial färgminskningsmetod") \
-                DRI_CONF_ENUM(0,"Avrunda färger") \
-                DRI_CONF_ENUM(1,"Utjämna färger") \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_ROUND_TRUNC 0
-#define DRI_CONF_ROUND_ROUND 1
-#define DRI_CONF_ROUND_MODE(def) \
-DRI_CONF_OPT_BEGIN_V(round_mode,enum,def,"0:1") \
-	DRI_CONF_DESC_BEGIN(en,"Color rounding method") \
-                DRI_CONF_ENUM(0,"Round color components downward") \
-                DRI_CONF_ENUM(1,"Round to nearest color") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(ca,"Mètode d'arrodoniment de color") \
-                DRI_CONF_ENUM(0,"Arrodoneix els components de color a baix") \
-                DRI_CONF_ENUM(1,"Arrodoneix al color més proper") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(de,"Farbrundungsmethode") \
-                DRI_CONF_ENUM(0,"Farbkomponenten abrunden") \
-                DRI_CONF_ENUM(1,"Zur ähnlichsten Farbe runden") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(es,"Método de redondeo de colores") \
-                DRI_CONF_ENUM(0,"Redondear hacia abajo los componentes de color") \
-                DRI_CONF_ENUM(1,"Redondear al color más cercano") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(nl,"Kleurafrondingmethode") \
-                DRI_CONF_ENUM(0,"Rond kleurencomponenten af naar beneden") \
-                DRI_CONF_ENUM(1,"Rond af naar dichtsbijzijnde kleur") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(fr,"Méthode d'arrondi des couleurs") \
-                DRI_CONF_ENUM(0,"Arrondi à l'inférieur") \
-                DRI_CONF_ENUM(1,"Arrondi au plus proche") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(sv,"Färgavrundningsmetod") \
-                DRI_CONF_ENUM(0,"Avrunda färdkomponenter nedåt") \
-                DRI_CONF_ENUM(1,"Avrunda till närmsta färg") \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_DITHER_XERRORDIFF 0
-#define DRI_CONF_DITHER_XERRORDIFFRESET 1
-#define DRI_CONF_DITHER_ORDERED 2
-#define DRI_CONF_DITHER_MODE(def) \
-DRI_CONF_OPT_BEGIN_V(dither_mode,enum,def,"0:2") \
-	DRI_CONF_DESC_BEGIN(en,"Color dithering method") \
-                DRI_CONF_ENUM(0,"Horizontal error diffusion") \
-                DRI_CONF_ENUM(1,"Horizontal error diffusion, reset error at line start") \
-                DRI_CONF_ENUM(2,"Ordered 2D color dithering") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(ca,"Mètode de tramat de color") \
-                DRI_CONF_ENUM(0,"Difusió d'error horitzontal") \
-                DRI_CONF_ENUM(1,"Difusió d'error horitzontal, reinicia l'error a l'inici de la línia") \
-                DRI_CONF_ENUM(2,"Tramat de color 2D ordenat") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(de,"Farbrasterungsmethode") \
-                DRI_CONF_ENUM(0,"Horizontale Fehlerstreuung") \
-                DRI_CONF_ENUM(1,"Horizontale Fehlerstreuung, Fehler am Zeilenanfang zurücksetzen") \
-                DRI_CONF_ENUM(2,"Geordnete 2D Farbrasterung") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(es,"Método de suavizado de color") \
-                DRI_CONF_ENUM(0,"Difusión de error horizontal") \
-                DRI_CONF_ENUM(1,"Difusión de error horizontal, reiniciar error al comienzo de línea") \
-                DRI_CONF_ENUM(2,"Suavizado de color 2D ordenado") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(nl,"Kleurrasteriseringsmethode") \
-                DRI_CONF_ENUM(0,"Horizontale foutdiffusie") \
-                DRI_CONF_ENUM(1,"Horizontale foutdiffusie, zet fout bij lijnbegin terug") \
-                DRI_CONF_ENUM(2,"Geordende 2D kleurrasterisering") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(fr,"Méthode de tramage") \
-                DRI_CONF_ENUM(0,"Diffusion d'erreur horizontale") \
-                DRI_CONF_ENUM(1,"Diffusion d'erreur horizontale, réinitialisé pour chaque ligne") \
-                DRI_CONF_ENUM(2,"Tramage ordonné des couleurs") \
-        DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(sv,"Färgutjämningsmetod") \
-                DRI_CONF_ENUM(0,"Horisontell felspridning") \
-                DRI_CONF_ENUM(1,"Horisontell felspridning, återställ fel vid radbörjan") \
-                DRI_CONF_ENUM(2,"Ordnad 2D-färgutjämning") \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_FLOAT_DEPTH(def) \
-DRI_CONF_OPT_BEGIN_B(float_depth, def) \
-        DRI_CONF_DESC(en,"Floating point depth buffer") \
-        DRI_CONF_DESC(ca,"Buffer de profunditat de punt flotant") \
-        DRI_CONF_DESC(de,"Fließkomma z-Puffer") \
-        DRI_CONF_DESC(es,"Búfer de profundidad en coma flotante") \
-        DRI_CONF_DESC(nl,"Dieptebuffer als commagetal") \
-        DRI_CONF_DESC(fr,"Z-buffer en virgule flottante") \
-        DRI_CONF_DESC(sv,"Buffert för flytande punktdjup") \
-DRI_CONF_OPT_END
-
 #define DRI_CONF_PP_CELSHADE(def) \
 DRI_CONF_OPT_BEGIN_V(pp_celshade,enum,def,"0:1") \
         DRI_CONF_DESC(en,"A post-processing filter to cel-shade the output") \
@@ -555,98 +366,6 @@ DRI_CONF_SECTION_BEGIN \
         DRI_CONF_DESC(fr,"Performance") \
         DRI_CONF_DESC(sv,"Prestanda")
 
-#define DRI_CONF_TCL_SW 0
-#define DRI_CONF_TCL_PIPELINED 1
-#define DRI_CONF_TCL_VTXFMT 2
-#define DRI_CONF_TCL_CODEGEN 3
-#define DRI_CONF_TCL_MODE(def) \
-DRI_CONF_OPT_BEGIN_V(tcl_mode,enum,def,"0:3") \
-        DRI_CONF_DESC_BEGIN(en,"TCL mode (Transformation, Clipping, Lighting)") \
-                DRI_CONF_ENUM(0,"Use software TCL pipeline") \
-                DRI_CONF_ENUM(1,"Use hardware TCL as first TCL pipeline stage") \
-                DRI_CONF_ENUM(2,"Bypass the TCL pipeline") \
-                DRI_CONF_ENUM(3,"Bypass the TCL pipeline with state-based machine code generated on-the-fly") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(ca,"Mode TCL (Transformació, Retall, Il·luminació)") \
-                DRI_CONF_ENUM(0,"Utilitza la canonada TCL de programari") \
-                DRI_CONF_ENUM(1,"Utilitza el TCL de maquinari com a la primera fase de la canonada TCL") \
-                DRI_CONF_ENUM(2,"Passa per alt la canonada TCL") \
-                DRI_CONF_ENUM(3,"Passa per alt la canonada TCL amb codi de màquina basat en estats, generat sobre la marxa") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(de,"TCL-Modus (Transformation, Clipping, Licht)") \
-                DRI_CONF_ENUM(0,"Benutze die Software-TCL-Pipeline") \
-                DRI_CONF_ENUM(1,"Benutze Hardware TCL als erste Stufe der TCL-Pipeline") \
-                DRI_CONF_ENUM(2,"Umgehe die TCL-Pipeline") \
-                DRI_CONF_ENUM(3,"Umgehe die TCL-Pipeline mit zur Laufzeit erzeugtem, zustandsbasiertem Maschinencode") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(es,"Modo TCL (Transformación, Recorte, Iluminación)") \
-                DRI_CONF_ENUM(0,"Usar tubería TCL por software") \
-                DRI_CONF_ENUM(1,"Usar TCL por hardware en la primera fase de la tubería TCL") \
-                DRI_CONF_ENUM(2,"Pasar por alto la tubería TCL") \
-                DRI_CONF_ENUM(3,"Pasar por alto la tubería TCL con código máquina basado en estados, generado al vuelo") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(nl,"TCL-modus (Transformatie, Clipping, Licht)") \
-                DRI_CONF_ENUM(0,"Gebruik software TCL pijpleiding") \
-                DRI_CONF_ENUM(1,"Gebruik hardware TCL as eerste TCL pijpleiding trap") \
-                DRI_CONF_ENUM(2,"Omzeil de TCL pijpleiding") \
-                DRI_CONF_ENUM(3,"Omzeil de TCL pijpleiding met staatgebaseerde machinecode die tijdens executie gegenereerd wordt") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(fr,"Mode de TCL (Transformation, Clipping, Eclairage)") \
-                DRI_CONF_ENUM(0,"Utiliser un pipeline TCL logiciel") \
-                DRI_CONF_ENUM(1,"Utiliser le TCL matériel pour le premier niveau de pipeline") \
-                DRI_CONF_ENUM(2,"Court-circuiter le pipeline TCL") \
-                DRI_CONF_ENUM(3,"Court-circuiter le pipeline TCL par une machine à états qui génère le codede TCL à la volée") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(sv,"TCL-läge (Transformation, Clipping, Lighting)") \
-                DRI_CONF_ENUM(0,"Använd programvaru-TCL-rörledning") \
-                DRI_CONF_ENUM(1,"Använd maskinvaru-TCL som första TCL-rörledningssteg") \
-                DRI_CONF_ENUM(2,"Kringgå TCL-rörledningen") \
-                DRI_CONF_ENUM(3,"Kringgå TCL-rörledningen med tillståndsbaserad maskinkod som direktgenereras") \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_FTHROTTLE_BUSY 0
-#define DRI_CONF_FTHROTTLE_USLEEPS 1
-#define DRI_CONF_FTHROTTLE_IRQS 2
-#define DRI_CONF_FTHROTTLE_MODE(def) \
-DRI_CONF_OPT_BEGIN_V(fthrottle_mode,enum,def,"0:2") \
-        DRI_CONF_DESC_BEGIN(en,"Method to limit rendering latency") \
-                DRI_CONF_ENUM(0,"Busy waiting for the graphics hardware") \
-                DRI_CONF_ENUM(1,"Sleep for brief intervals while waiting for the graphics hardware") \
-                DRI_CONF_ENUM(2,"Let the graphics hardware emit a software interrupt and sleep") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(ca,"Mètode per a limitar la latència de renderització") \
-                DRI_CONF_ENUM(0,"Espera activa pel maquinari de gràfics") \
-                DRI_CONF_ENUM(1,"Dorm per intervals breus mentre s'espera al maquinari de gràfics") \
-                DRI_CONF_ENUM(2,"Deixa que el maquinari de gràfics emeti una interrupció de programari i dormi") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(de,"Methode zur Begrenzung der Bildverzögerung") \
-                DRI_CONF_ENUM(0,"Aktives Warten auf die Grafikhardware") \
-                DRI_CONF_ENUM(1,"Kurze Schlafintervalle beim Warten auf die Grafikhardware") \
-                DRI_CONF_ENUM(2,"Die Grafikhardware eine Softwareunterbrechnung erzeugen lassen und schlafen") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(es,"Método para limitar la latencia de renderización") \
-                DRI_CONF_ENUM(0,"Esperar activamente al hardware gráfico") \
-                DRI_CONF_ENUM(1,"Dormir en intervalos cortos mientras se espera al hardware gráfico") \
-                DRI_CONF_ENUM(2,"Permitir que el hardware gráfico emita una interrupción de software y duerma") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(nl,"Methode om beeldopbouwvertraging te onderdrukken") \
-                DRI_CONF_ENUM(0,"Actief wachten voor de grafische hardware") \
-                DRI_CONF_ENUM(1,"Slaap voor korte intervallen tijdens het wachten op de grafische hardware") \
-                DRI_CONF_ENUM(2,"Laat de grafische hardware een software onderbreking uitzenden en in slaap vallen") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(fr,"Méthode d'attente de la carte graphique") \
-                DRI_CONF_ENUM(0,"Attente active de la carte graphique") \
-                DRI_CONF_ENUM(1,"Attente utilisant usleep()") \
-                DRI_CONF_ENUM(2,"Utiliser les interruptions") \
-        DRI_CONF_DESC_END \
-        DRI_CONF_DESC_BEGIN(sv,"Metod för att begränsa renderingslatens") \
-                DRI_CONF_ENUM(0,"Upptagen med att vänta på grafikhårdvaran") \
-                DRI_CONF_ENUM(1,"Sov i korta intervall under väntan på grafikhårdvaran") \
-                DRI_CONF_ENUM(2,"Låt grafikhårdvaran sända ut ett programvaruavbrott och sov") \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
 #define DRI_CONF_VBLANK_NEVER 0
 #define DRI_CONF_VBLANK_DEF_INTERVAL_0 1
 #define DRI_CONF_VBLANK_DEF_INTERVAL_1 2
@@ -697,83 +416,6 @@ DRI_CONF_OPT_BEGIN_V(vblank_mode,enum,def,"0:3") \
         DRI_CONF_DESC_END \
 DRI_CONF_OPT_END
 
-#define DRI_CONF_HYPERZ_DISABLED 0
-#define DRI_CONF_HYPERZ_ENABLED 1
-#define DRI_CONF_HYPERZ(def) \
-DRI_CONF_OPT_BEGIN_B(hyperz, def) \
-        DRI_CONF_DESC(en,"Use HyperZ to boost performance") \
-        DRI_CONF_DESC(ca,"Utilitza el HyperZ per a augmentar el rendiment") \
-        DRI_CONF_DESC(de,"HyperZ zur Leistungssteigerung verwenden") \
-        DRI_CONF_DESC(es,"Usar HyperZ para potenciar rendimiento") \
-        DRI_CONF_DESC(nl,"Gebruik HyperZ om de prestaties te verbeteren") \
-        DRI_CONF_DESC(fr,"Utiliser le HyperZ pour améliorer les performances") \
-        DRI_CONF_DESC(sv,"Använd HyperZ för att maximera prestandan") \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_MAX_TEXTURE_UNITS(def,min,max) \
-DRI_CONF_OPT_BEGIN_V(texture_units,int,def, # min ":" # max ) \
-        DRI_CONF_DESC(en,"Number of texture units used") \
-        DRI_CONF_DESC(ca,"Nombre d'unitats de textura utilitzades") \
-        DRI_CONF_DESC(de,"Anzahl der benutzten Textureinheiten") \
-        DRI_CONF_DESC(es,"Número de unidades de textura usadas") \
-        DRI_CONF_DESC(nl,"Aantal textuureenheden in gebruik") \
-        DRI_CONF_DESC(fr,"Nombre d'unités de texture") \
-        DRI_CONF_DESC(sv,"Antal använda texturenheter") \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_TEXTURE_BLEND_QUALITY(def,range) \
-DRI_CONF_OPT_BEGIN_V(texture_blend_quality,float,def,range) \
-	DRI_CONF_DESC(en,"Texture filtering quality vs. speed, AKA “brilinear” texture filtering") \
-	DRI_CONF_DESC(ca,"Qualitat vs. velocitat de filtració de textura, àlies filtració ”brilinear“ de textura") \
-	DRI_CONF_DESC(de,"Texturfilterqualität versus -geschwindigkeit, auch bekannt als „brilineare“ Texturfilterung") \
-	DRI_CONF_DESC(es,"Calidad de filtrado de textura vs. velocidad, alias filtrado ”brilinear“ de textura") \
-	DRI_CONF_DESC(nl,"Textuurfilterkwaliteit versus -snelheid, ookwel bekend als “brilineaire” textuurfiltering") \
-	DRI_CONF_DESC(fr,"Qualité/performance du filtrage trilinéaire de texture (filtrage brilinéaire)") \
-	DRI_CONF_DESC(sv,"Texturfiltreringskvalitet mot hastighet, även kallad ”brilinear”-texturfiltrering") \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_TEXTURE_HEAPS_ALL 0
-#define DRI_CONF_TEXTURE_HEAPS_CARD 1
-#define DRI_CONF_TEXTURE_HEAPS_GART 2
-#define DRI_CONF_TEXTURE_HEAPS(def) \
-DRI_CONF_OPT_BEGIN_V(texture_heaps,enum,def,"0:2") \
-	DRI_CONF_DESC_BEGIN(en,"Used types of texture memory") \
-		DRI_CONF_ENUM(0,"All available memory") \
-		DRI_CONF_ENUM(1,"Only card memory (if available)") \
-		DRI_CONF_ENUM(2,"Only GART (AGP/PCIE) memory (if available)") \
-	DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(ca,"Tipus utilitzats de memòria de textura") \
-		DRI_CONF_ENUM(0,"Tota la memòria disponible") \
-		DRI_CONF_ENUM(1,"Només memòria de targeta (si està disponible)") \
-		DRI_CONF_ENUM(2,"Només memòria GART (AGP/PCIE) (si està disponible)") \
-	DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(de,"Benutzte Arten von Texturspeicher") \
-		DRI_CONF_ENUM(0,"Aller verfügbarer Speicher") \
-		DRI_CONF_ENUM(1,"Nur Grafikspeicher (falls verfügbar)") \
-		DRI_CONF_ENUM(2,"Nur GART-Speicher (AGP/PCIE) (falls verfügbar)") \
-	DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(es,"Tipos de memoria de textura usados") \
-		DRI_CONF_ENUM(0,"Toda la memoria disponible") \
-		DRI_CONF_ENUM(1,"Solo memoria de tarjeta (si está disponible)") \
-		DRI_CONF_ENUM(2,"Solo memoria GART (AGP/PCIE) (si está disponible)") \
-	DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(nl,"Gebruikte soorten textuurgeheugen") \
-		DRI_CONF_ENUM(0,"Al het beschikbaar geheugen") \
-		DRI_CONF_ENUM(1,"Alleen geheugen op de kaart (als het aanwezig is)") \
-		DRI_CONF_ENUM(2,"Alleen GART (AGP/PCIE) geheugen (als het aanwezig is)") \
-	DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(fr,"Types de mémoire de texture") \
-		DRI_CONF_ENUM(0,"Utiliser toute la mémoire disponible") \
-		DRI_CONF_ENUM(1,"Utiliser uniquement la mémoire graphique (si disponible)") \
-		DRI_CONF_ENUM(2,"Utiliser uniquement la mémoire GART (AGP/PCIE) (si disponible)") \
-	DRI_CONF_DESC_END \
-	DRI_CONF_DESC_BEGIN(sv,"Använda typer av texturminne") \
-		DRI_CONF_ENUM(0,"Allt tillgängligt minne") \
-		DRI_CONF_ENUM(1,"Endast kortminne (om tillgängligt)") \
-		DRI_CONF_ENUM(2,"Endast GART-minne (AGP/PCIE) (om tillgängligt)") \
-	DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
 #define DRI_CONF_MESA_GLTHREAD(def) \
 DRI_CONF_OPT_BEGIN_B(mesa_glthread, def) \
         DRI_CONF_DESC(en,"Enable offloading GL driver work to a separate thread") \
@@ -818,30 +460,15 @@ DRI_CONF_OPT_BEGIN_B(glx_disable_oml_sync_control, def) \
    DRI_CONF_DESC(sv, "Disable the GLX_OML_sync_control extension") \
 DRI_CONF_OPT_END
 
-
-/**
- * \brief Software-fallback options.  To allow using features (like
- * GL_ARB_vertex_program) on GPUs that don't otherwise support the feature.
- */
-#define DRI_CONF_SECTION_SOFTWARE \
-DRI_CONF_SECTION_BEGIN \
-        DRI_CONF_DESC(en,"Features that are not hardware-accelerated") \
-        DRI_CONF_DESC(ca,"Característiques no accelerades per maquinari") \
-        DRI_CONF_DESC(de,"Funktionalität, die nicht hardwarebeschleunigt ist") \
-        DRI_CONF_DESC(es,"Características no aceleradas por hardware") \
-        DRI_CONF_DESC(nl,"Eigenschappen die niet hardwareversneld zijn") \
-        DRI_CONF_DESC(fr,"Fonctionnalités ne bénéficiant pas d'une accélération matérielle") \
-        DRI_CONF_DESC(sv,"Funktioner som inte är hårdvaruaccelererade")
-
-#define DRI_CONF_ARB_VERTEX_PROGRAM(def) \
-DRI_CONF_OPT_BEGIN_B(arb_vertex_program, def) \
-        DRI_CONF_DESC(en,"Enable extension GL_ARB_vertex_program") \
-        DRI_CONF_DESC(ca,"Habilita l'extensió GL_ARB_vertex_program") \
-        DRI_CONF_DESC(de,"Erweiterung GL_ARB_vertex_program aktivieren") \
-        DRI_CONF_DESC(es,"Habilitar la extensión GL_ARB_vertex_program") \
-        DRI_CONF_DESC(nl,"Zet uitbreiding GL_ARB_vertex_program aan") \
-        DRI_CONF_DESC(fr,"Activer l'extension GL_ARB_vertex_program") \
-        DRI_CONF_DESC(sv,"Aktivera tillägget GL_ARB_vertex_program") \
+#define DRI_CONF_DISABLE_SGI_VIDEO_SYNC(def) \
+DRI_CONF_OPT_BEGIN_B(glx_disable_sgi_video_sync, def) \
+   DRI_CONF_DESC(en, "Disable the GLX_SGI_video_sync extension") \
+   DRI_CONF_DESC(ca, "Disable the GLX_SGI_video_sync extension") \
+   DRI_CONF_DESC(de, "Disable the GLX_SGI_video_sync extension") \
+   DRI_CONF_DESC(es, "Disable the GLX_SGI_video_sync extension") \
+   DRI_CONF_DESC(nl, "Disable the GLX_SGI_video_sync extension") \
+   DRI_CONF_DESC(fr, "Disable the GLX_SGI_video_sync extension") \
+   DRI_CONF_DESC(sv, "Disable the GLX_SGI_video_sync extension") \
 DRI_CONF_OPT_END
 
 
@@ -881,6 +508,17 @@ DRI_CONF_OPT_BEGIN_B(glsl_zero_init, def) \
         DRI_CONF_DESC(sv,"Force uninitialized variables to default to zero") \
 DRI_CONF_OPT_END
 
+#define DRI_CONF_ALLOW_RGB10_CONFIGS(def) \
+DRI_CONF_OPT_BEGIN_B(allow_rgb10_configs, def) \
+DRI_CONF_DESC(en,"Allow exposure of visuals and fbconfigs with rgb10a2 formats") \
+DRI_CONF_DESC(ca,"Allow exposure of visuals and fbconfigs with rgb10a2 formats") \
+DRI_CONF_DESC(de,"Allow exposure of visuals and fbconfigs with rgb10a2 formats") \
+DRI_CONF_DESC(es,"Allow exposure of visuals and fbconfigs with rgb10a2 formats") \
+DRI_CONF_DESC(nl,"Allow exposure of visuals and fbconfigs with rgb10a2 formats") \
+DRI_CONF_DESC(fr,"Allow exposure of visuals and fbconfigs with rgb10a2 formats") \
+DRI_CONF_DESC(sv,"Allow exposure of visuals and fbconfigs with rgb10a2 formats") \
+DRI_CONF_OPT_END
+
 /**
  * \brief Initialization configuration options
  */
@@ -905,6 +543,17 @@ DRI_CONF_OPT_BEGIN(device_id, string, def) \
         DRI_CONF_DESC(sv,"Define the graphic device to use if possible") \
 DRI_CONF_OPT_END
 
+#define DRI_CONF_DRI_DRIVER(def) \
+DRI_CONF_OPT_BEGIN(dri_driver, string, def) \
+        DRI_CONF_DESC(en,"Override the DRI driver to load") \
+        DRI_CONF_DESC(ca,"Override the DRI driver to load") \
+        DRI_CONF_DESC(de,"Override the DRI driver to load") \
+        DRI_CONF_DESC(es,"Override the DRI driver to load") \
+        DRI_CONF_DESC(nl,"Override the DRI driver to load") \
+        DRI_CONF_DESC(fr,"Override the DRI driver to load") \
+        DRI_CONF_DESC(sv,"Override the DRI driver to load") \
+DRI_CONF_OPT_END
+
 /**
  * \brief Gallium-Nine specific configuration options
  */
@@ -1026,3 +675,8 @@ DRI_CONF_OPT_END
 DRI_CONF_OPT_BEGIN_B(radeonsi_clear_db_cache_before_clear, def) \
         DRI_CONF_DESC(en,"Clear DB cache before fast depth clear") \
 DRI_CONF_OPT_END
+
+#define DRI_CONF_RADEONSI_ZERO_ALL_VRAM_ALLOCS(def) \
+DRI_CONF_OPT_BEGIN_B(radeonsi_zerovram, def) \
+        DRI_CONF_DESC(en,"Zero all vram allocations") \
+DRI_CONF_OPT_END
diff --git a/lib/mesa/src/util/xmlpool/sv.gmo b/lib/mesa/src/util/xmlpool/sv.gmo
new file mode 100644
index 000000000..8393f427e
--- /dev/null
+++ b/lib/mesa/src/util/xmlpool/sv.gmo
diff --git a/lib/mesa/src/util/xmlpool/sv.po b/lib/mesa/src/util/xmlpool/sv.po
index d8d7353f4..36e5a9b65 100644
--- a/lib/mesa/src/util/xmlpool/sv.po
+++ b/lib/mesa/src/util/xmlpool/sv.po
@@ -21,14 +21,6 @@ msgstr ""
 msgid "Debugging"
 msgstr "Felsökning"
 
-#: t_options.h:60
-msgid "Disable 3D acceleration"
-msgstr "Inaktivera 3D-accelerering"
-
-#: t_options.h:65
-msgid "Show performance boxes"
-msgstr "Visa prestandarutor"
-
 #: t_options.h:70
 msgid "Enable flushing batchbuffer after each draw call"
 msgstr ""
@@ -53,10 +45,6 @@ msgstr ""
 msgid "Disable backslash-based line continuations in GLSL source"
 msgstr ""
 
-#: t_options.h:100
-msgid "Disable GL_ARB_shader_bit_encoding"
-msgstr ""
-
 #: t_options.h:105
 msgid ""
 "Force a default GLSL version for shaders that lack an explicit #version line"
@@ -70,63 +58,6 @@ msgstr ""
 msgid "Image Quality"
 msgstr "Bildkvalitet"
 
-#: t_options.h:133
-msgid "Texture color depth"
-msgstr "Färgdjup för texturer"
-
-#: t_options.h:134
-msgid "Prefer frame buffer color depth"
-msgstr "Föredra färgdjupet för framebuffer"
-
-#: t_options.h:135
-msgid "Prefer 32 bits per texel"
-msgstr "Föredra 32 bitar per texel"
-
-#: t_options.h:136
-msgid "Prefer 16 bits per texel"
-msgstr "Föredra 16 bitar per texel"
-
-#: t_options.h:137
-msgid "Force 16 bits per texel"
-msgstr "Tvinga 16 bitar per texel"
-
-#: t_options.h:143
-msgid "Initial maximum value for anisotropic texture filtering"
-msgstr "Initialt maximalt värde för anisotropisk texturfiltrering"
-
-#: t_options.h:148
-msgid "Forbid negative texture LOD bias"
-msgstr "Förbjud negativ LOD-kompensation för texturer"
-
-#: t_options.h:153
-msgid ""
-"Enable S3TC texture compression even if software support is not available"
-msgstr "Aktivera S3TC-texturkomprimering även om programvarustöd saknas"
-
-#: t_options.h:160
-msgid "Initial color reduction method"
-msgstr "Initial färgminskningsmetod"
-
-#: t_options.h:161
-msgid "Round colors"
-msgstr "Avrunda färger"
-
-#: t_options.h:162
-msgid "Dither colors"
-msgstr "Utjämna färger"
-
-#: t_options.h:170
-msgid "Color rounding method"
-msgstr "Färgavrundningsmetod"
-
-#: t_options.h:171
-msgid "Round color components downward"
-msgstr "Avrunda färdkomponenter nedåt"
-
-#: t_options.h:172
-msgid "Round to nearest color"
-msgstr "Avrunda till närmsta färg"
-
 #: t_options.h:181
 msgid "Color dithering method"
 msgstr "Färgutjämningsmetod"
@@ -143,10 +74,6 @@ msgstr "Horisontell felspridning, återställ fel vid radbörjan"
 msgid "Ordered 2D color dithering"
 msgstr "Ordnad 2D-färgutjämning"
 
-#: t_options.h:190
-msgid "Floating point depth buffer"
-msgstr "Buffert för flytande punktdjup"
-
 #: t_options.h:195
 msgid "A post-processing filter to cel-shade the output"
 msgstr ""
@@ -179,44 +106,6 @@ msgstr ""
 msgid "Performance"
 msgstr "Prestanda"
 
-#: t_options.h:238
-msgid "TCL mode (Transformation, Clipping, Lighting)"
-msgstr "TCL-läge (Transformation, Clipping, Lighting)"
-
-#: t_options.h:239
-msgid "Use software TCL pipeline"
-msgstr "Använd programvaru-TCL-rörledning"
-
-#: t_options.h:240
-msgid "Use hardware TCL as first TCL pipeline stage"
-msgstr "Använd maskinvaru-TCL som första TCL-rörledningssteg"
-
-#: t_options.h:241
-msgid "Bypass the TCL pipeline"
-msgstr "Kringgå TCL-rörledningen"
-
-#: t_options.h:242
-msgid ""
-"Bypass the TCL pipeline with state-based machine code generated on-the-fly"
-msgstr ""
-"Kringgå TCL-rörledningen med tillståndsbaserad maskinkod som direktgenereras"
-
-#: t_options.h:251
-msgid "Method to limit rendering latency"
-msgstr "Metod för att begränsa renderingslatens"
-
-#: t_options.h:252
-msgid "Busy waiting for the graphics hardware"
-msgstr "Upptagen med att vänta på grafikhårdvaran"
-
-#: t_options.h:253
-msgid "Sleep for brief intervals while waiting for the graphics hardware"
-msgstr "Sov i korta intervall under väntan på grafikhårdvaran"
-
-#: t_options.h:254
-msgid "Let the graphics hardware emit a software interrupt and sleep"
-msgstr "Låt grafikhårdvaran sända ut ett programvaruavbrott och sov"
-
 #: t_options.h:264
 msgid "Synchronization with vertical refresh (swap intervals)"
 msgstr "Synkronisering med vertikal uppdatering (växlingsintervall)"
@@ -241,44 +130,6 @@ msgstr ""
 "Synkronisera alltid med vertikal uppdatering, programmet väljer den minsta "
 "växlingsintervallen"
 
-#: t_options.h:276
-msgid "Use HyperZ to boost performance"
-msgstr "Använd HyperZ för att maximera prestandan"
-
-#: t_options.h:281
-msgid "Number of texture units used"
-msgstr "Antal använda texturenheter"
-
-#: t_options.h:286
-msgid "Texture filtering quality vs. speed, AKA “brilinear” texture filtering"
-msgstr ""
-"Texturfiltreringskvalitet mot hastighet, även kallad \"brilinear\"-"
-"texturfiltrering"
-
-#: t_options.h:294
-msgid "Used types of texture memory"
-msgstr "Använda typer av texturminne"
-
-#: t_options.h:295
-msgid "All available memory"
-msgstr "Allt tillgängligt minne"
-
-#: t_options.h:296
-msgid "Only card memory (if available)"
-msgstr "Endast kortminne (om tillgängligt)"
-
-#: t_options.h:297
-msgid "Only GART (AGP/PCIE) memory (if available)"
-msgstr "Endast GART-minne (AGP/PCIE) (om tillgängligt)"
-
-#: t_options.h:309
-msgid "Features that are not hardware-accelerated"
-msgstr "Funktioner som inte är hårdvaruaccelererade"
-
-#: t_options.h:313
-msgid "Enable extension GL_ARB_vertex_program"
-msgstr "Aktivera tillägget GL_ARB_vertex_program"
-
 #: t_options.h:323
 msgid "Miscellaneous"
 msgstr ""
diff --git a/lib/mesa/src/util/xmlpool/t_options.h b/lib/mesa/src/util/xmlpool/t_options.h
index bd553085c..e0a30f5fd 100644
--- a/lib/mesa/src/util/xmlpool/t_options.h
+++ b/lib/mesa/src/util/xmlpool/t_options.h
@@ -53,17 +53,7 @@
  */
 #define DRI_CONF_SECTION_DEBUG \
 DRI_CONF_SECTION_BEGIN \
-	DRI_CONF_DESC(en,gettext("Debugging"))
-
-#define DRI_CONF_NO_RAST(def) \
-DRI_CONF_OPT_BEGIN_B(no_rast, def) \
-        DRI_CONF_DESC(en,gettext("Disable 3D acceleration")) \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_PERFORMANCE_BOXES(def) \
-DRI_CONF_OPT_BEGIN_B(performance_boxes, def) \
-        DRI_CONF_DESC(en,gettext("Show performance boxes")) \
-DRI_CONF_OPT_END
+        DRI_CONF_DESC(en,gettext("Debugging"))
 
 #define DRI_CONF_ALWAYS_FLUSH_BATCH(def) \
 DRI_CONF_OPT_BEGIN_B(always_flush_batch, def) \
@@ -77,7 +67,7 @@ DRI_CONF_OPT_END
 
 #define DRI_CONF_DISABLE_THROTTLING(def) \
 DRI_CONF_OPT_BEGIN_B(disable_throttling, def) \
-	DRI_CONF_DESC(en,gettext("Disable throttling on first batch after flush")) \
+        DRI_CONF_DESC(en,gettext("Disable throttling on first batch after flush")) \
 DRI_CONF_OPT_END
 
 #define DRI_CONF_FORCE_GLSL_EXTENSIONS_WARN(def) \
@@ -100,11 +90,6 @@ DRI_CONF_OPT_BEGIN_B(disable_glsl_line_continuations, def) \
         DRI_CONF_DESC(en,gettext("Disable backslash-based line continuations in GLSL source")) \
 DRI_CONF_OPT_END
 
-#define DRI_CONF_DISABLE_SHADER_BIT_ENCODING(def) \
-DRI_CONF_OPT_BEGIN_B(disable_shader_bit_encoding, def) \
-        DRI_CONF_DESC(en,gettext("Disable GL_ARB_shader_bit_encoding")) \
-DRI_CONF_OPT_END
-
 #define DRI_CONF_FORCE_GLSL_VERSION(def) \
 DRI_CONF_OPT_BEGIN_V(force_glsl_version, int, def, "0:999") \
         DRI_CONF_DESC(en,gettext("Force a default GLSL version for shaders that lack an explicit #version line")) \
@@ -115,6 +100,16 @@ DRI_CONF_OPT_BEGIN_B(allow_glsl_extension_directive_midshader, def) \
         DRI_CONF_DESC(en,gettext("Allow GLSL #extension directives in the middle of shaders")) \
 DRI_CONF_OPT_END
 
+#define DRI_CONF_ALLOW_GLSL_BUILTIN_CONST_EXPRESSION(def) \
+DRI_CONF_OPT_BEGIN_B(allow_glsl_builtin_const_expression, def) \
+        DRI_CONF_DESC(en,gettext("Allow builtins as part of constant expressions")) \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_ALLOW_GLSL_RELAXED_ES(def) \
+DRI_CONF_OPT_BEGIN_B(allow_glsl_relaxed_es, def) \
+        DRI_CONF_DESC(en,gettext("Allow some relaxation of GLSL ES shader restrictions")) \
+DRI_CONF_OPT_END
+
 #define DRI_CONF_ALLOW_GLSL_BUILTIN_VARIABLE_REDECLARATION(def) \
 DRI_CONF_OPT_BEGIN_B(allow_glsl_builtin_variable_redeclaration, def) \
         DRI_CONF_DESC(en,gettext("Allow GLSL built-in variables to be redeclared verbatim")) \
@@ -140,84 +135,28 @@ DRI_CONF_OPT_BEGIN_B(allow_glsl_cross_stage_interpolation_mismatch, def) \
         DRI_CONF_DESC(en,gettext("Allow interpolation qualifier mismatch across shader stages")) \
 DRI_CONF_OPT_END
 
+#define DRI_CONF_ALLOW_GLSL_LAYOUT_QUALIFIER_ON_FUNCTION_PARAMETERS(def) \
+DRI_CONF_OPT_BEGIN_B(allow_glsl_layout_qualifier_on_function_parameters, def) \
+        DRI_CONF_DESC(en,gettext("Allow layout qualifiers on function parameters.")) \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_FORCE_COMPAT_PROFILE(def) \
+DRI_CONF_OPT_BEGIN_B(force_compat_profile, def) \
+        DRI_CONF_DESC(en,gettext("Force an OpenGL compatibility context")) \
+DRI_CONF_OPT_END
+
 /**
  * \brief Image quality-related options
  */
 #define DRI_CONF_SECTION_QUALITY \
 DRI_CONF_SECTION_BEGIN \
-	DRI_CONF_DESC(en,gettext("Image Quality"))
-
-#define DRI_CONF_EXCESS_MIPMAP(def) \
-DRI_CONF_OPT_BEGIN_B(excess_mipmap, def) \
-	DRI_CONF_DESC(en,"Enable extra mipmap level") \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_TEXTURE_DEPTH_FB       0
-#define DRI_CONF_TEXTURE_DEPTH_32       1
-#define DRI_CONF_TEXTURE_DEPTH_16       2
-#define DRI_CONF_TEXTURE_DEPTH_FORCE_16 3
-#define DRI_CONF_TEXTURE_DEPTH(def) \
-DRI_CONF_OPT_BEGIN_V(texture_depth,enum,def,"0:3") \
-	DRI_CONF_DESC_BEGIN(en,gettext("Texture color depth")) \
-                DRI_CONF_ENUM(0,gettext("Prefer frame buffer color depth")) \
-                DRI_CONF_ENUM(1,gettext("Prefer 32 bits per texel")) \
-                DRI_CONF_ENUM(2,gettext("Prefer 16 bits per texel")) \
-                DRI_CONF_ENUM(3,gettext("Force 16 bits per texel")) \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_DEF_MAX_ANISOTROPY(def,range) \
-DRI_CONF_OPT_BEGIN_V(def_max_anisotropy,float,def,range) \
-        DRI_CONF_DESC(en,gettext("Initial maximum value for anisotropic texture filtering")) \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_NO_NEG_LOD_BIAS(def) \
-DRI_CONF_OPT_BEGIN_B(no_neg_lod_bias, def) \
-        DRI_CONF_DESC(en,gettext("Forbid negative texture LOD bias")) \
-DRI_CONF_OPT_END
+        DRI_CONF_DESC(en,gettext("Image Quality"))
 
 #define DRI_CONF_PRECISE_TRIG(def) \
 DRI_CONF_OPT_BEGIN_B(precise_trig, def) \
         DRI_CONF_DESC(en,gettext("Prefer accuracy over performance in trig functions")) \
 DRI_CONF_OPT_END
 
-#define DRI_CONF_COLOR_REDUCTION_ROUND 0
-#define DRI_CONF_COLOR_REDUCTION_DITHER 1
-#define DRI_CONF_COLOR_REDUCTION(def) \
-DRI_CONF_OPT_BEGIN_V(color_reduction,enum,def,"0:1") \
-        DRI_CONF_DESC_BEGIN(en,gettext("Initial color reduction method")) \
-                DRI_CONF_ENUM(0,gettext("Round colors")) \
-                DRI_CONF_ENUM(1,gettext("Dither colors")) \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_ROUND_TRUNC 0
-#define DRI_CONF_ROUND_ROUND 1
-#define DRI_CONF_ROUND_MODE(def) \
-DRI_CONF_OPT_BEGIN_V(round_mode,enum,def,"0:1") \
-	DRI_CONF_DESC_BEGIN(en,gettext("Color rounding method")) \
-                DRI_CONF_ENUM(0,gettext("Round color components downward")) \
-                DRI_CONF_ENUM(1,gettext("Round to nearest color")) \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_DITHER_XERRORDIFF 0
-#define DRI_CONF_DITHER_XERRORDIFFRESET 1
-#define DRI_CONF_DITHER_ORDERED 2
-#define DRI_CONF_DITHER_MODE(def) \
-DRI_CONF_OPT_BEGIN_V(dither_mode,enum,def,"0:2") \
-	DRI_CONF_DESC_BEGIN(en,gettext("Color dithering method")) \
-                DRI_CONF_ENUM(0,gettext("Horizontal error diffusion")) \
-                DRI_CONF_ENUM(1,gettext("Horizontal error diffusion, reset error at line start")) \
-                DRI_CONF_ENUM(2,gettext("Ordered 2D color dithering")) \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_FLOAT_DEPTH(def) \
-DRI_CONF_OPT_BEGIN_B(float_depth, def) \
-        DRI_CONF_DESC(en,gettext("Floating point depth buffer")) \
-DRI_CONF_OPT_END
-
 #define DRI_CONF_PP_CELSHADE(def) \
 DRI_CONF_OPT_BEGIN_V(pp_celshade,enum,def,"0:1") \
         DRI_CONF_DESC(en,gettext("A post-processing filter to cel-shade the output")) \
@@ -257,32 +196,6 @@ DRI_CONF_OPT_END
 DRI_CONF_SECTION_BEGIN \
         DRI_CONF_DESC(en,gettext("Performance"))
 
-#define DRI_CONF_TCL_SW 0
-#define DRI_CONF_TCL_PIPELINED 1
-#define DRI_CONF_TCL_VTXFMT 2
-#define DRI_CONF_TCL_CODEGEN 3
-#define DRI_CONF_TCL_MODE(def) \
-DRI_CONF_OPT_BEGIN_V(tcl_mode,enum,def,"0:3") \
-        DRI_CONF_DESC_BEGIN(en,gettext("TCL mode (Transformation, Clipping, Lighting)")) \
-                DRI_CONF_ENUM(0,gettext("Use software TCL pipeline")) \
-                DRI_CONF_ENUM(1,gettext("Use hardware TCL as first TCL pipeline stage")) \
-                DRI_CONF_ENUM(2,gettext("Bypass the TCL pipeline")) \
-                DRI_CONF_ENUM(3,gettext("Bypass the TCL pipeline with state-based machine code generated on-the-fly")) \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_FTHROTTLE_BUSY 0
-#define DRI_CONF_FTHROTTLE_USLEEPS 1
-#define DRI_CONF_FTHROTTLE_IRQS 2
-#define DRI_CONF_FTHROTTLE_MODE(def) \
-DRI_CONF_OPT_BEGIN_V(fthrottle_mode,enum,def,"0:2") \
-        DRI_CONF_DESC_BEGIN(en,gettext("Method to limit rendering latency")) \
-                DRI_CONF_ENUM(0,gettext("Busy waiting for the graphics hardware")) \
-                DRI_CONF_ENUM(1,gettext("Sleep for brief intervals while waiting for the graphics hardware")) \
-                DRI_CONF_ENUM(2,gettext("Let the graphics hardware emit a software interrupt and sleep")) \
-        DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
 #define DRI_CONF_VBLANK_NEVER 0
 #define DRI_CONF_VBLANK_DEF_INTERVAL_0 1
 #define DRI_CONF_VBLANK_DEF_INTERVAL_1 2
@@ -297,35 +210,6 @@ DRI_CONF_OPT_BEGIN_V(vblank_mode,enum,def,"0:3") \
         DRI_CONF_DESC_END \
 DRI_CONF_OPT_END
 
-#define DRI_CONF_HYPERZ_DISABLED 0
-#define DRI_CONF_HYPERZ_ENABLED 1
-#define DRI_CONF_HYPERZ(def) \
-DRI_CONF_OPT_BEGIN_B(hyperz, def) \
-        DRI_CONF_DESC(en,gettext("Use HyperZ to boost performance")) \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_MAX_TEXTURE_UNITS(def,min,max) \
-DRI_CONF_OPT_BEGIN_V(texture_units,int,def, # min ":" # max ) \
-        DRI_CONF_DESC(en,gettext("Number of texture units used")) \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_TEXTURE_BLEND_QUALITY(def,range) \
-DRI_CONF_OPT_BEGIN_V(texture_blend_quality,float,def,range) \
-	DRI_CONF_DESC(en,gettext("Texture filtering quality vs. speed, AKA “brilinear” texture filtering")) \
-DRI_CONF_OPT_END
-
-#define DRI_CONF_TEXTURE_HEAPS_ALL 0
-#define DRI_CONF_TEXTURE_HEAPS_CARD 1
-#define DRI_CONF_TEXTURE_HEAPS_GART 2
-#define DRI_CONF_TEXTURE_HEAPS(def) \
-DRI_CONF_OPT_BEGIN_V(texture_heaps,enum,def,"0:2") \
-	DRI_CONF_DESC_BEGIN(en,gettext("Used types of texture memory")) \
-		DRI_CONF_ENUM(0,gettext("All available memory")) \
-		DRI_CONF_ENUM(1,gettext("Only card memory (if available)")) \
-		DRI_CONF_ENUM(2,gettext("Only GART (AGP/PCIE) memory (if available)")) \
-	DRI_CONF_DESC_END \
-DRI_CONF_OPT_END
-
 #define DRI_CONF_MESA_GLTHREAD(def) \
 DRI_CONF_OPT_BEGIN_B(mesa_glthread, def) \
         DRI_CONF_DESC(en,gettext("Enable offloading GL driver work to a separate thread")) \
@@ -346,18 +230,9 @@ DRI_CONF_OPT_BEGIN_B(glx_disable_oml_sync_control, def) \
    DRI_CONF_DESC(en, gettext("Disable the GLX_OML_sync_control extension")) \
 DRI_CONF_OPT_END
 
-
-/**
- * \brief Software-fallback options.  To allow using features (like
- * GL_ARB_vertex_program) on GPUs that don't otherwise support the feature.
- */
-#define DRI_CONF_SECTION_SOFTWARE \
-DRI_CONF_SECTION_BEGIN \
-        DRI_CONF_DESC(en,gettext("Features that are not hardware-accelerated"))
-
-#define DRI_CONF_ARB_VERTEX_PROGRAM(def) \
-DRI_CONF_OPT_BEGIN_B(arb_vertex_program, def) \
-        DRI_CONF_DESC(en,gettext("Enable extension GL_ARB_vertex_program")) \
+#define DRI_CONF_DISABLE_SGI_VIDEO_SYNC(def) \
+DRI_CONF_OPT_BEGIN_B(glx_disable_sgi_video_sync, def) \
+   DRI_CONF_DESC(en, gettext("Disable the GLX_SGI_video_sync extension")) \
 DRI_CONF_OPT_END
 
 
@@ -379,6 +254,11 @@ DRI_CONF_OPT_BEGIN_B(glsl_zero_init, def) \
         DRI_CONF_DESC(en,gettext("Force uninitialized variables to default to zero")) \
 DRI_CONF_OPT_END
 
+#define DRI_CONF_ALLOW_RGB10_CONFIGS(def) \
+DRI_CONF_OPT_BEGIN_B(allow_rgb10_configs, def) \
+DRI_CONF_DESC(en,gettext("Allow exposure of visuals and fbconfigs with rgb10a2 formats")) \
+DRI_CONF_OPT_END
+
 /**
  * \brief Initialization configuration options
  */
@@ -391,6 +271,11 @@ DRI_CONF_OPT_BEGIN(device_id, string, def) \
         DRI_CONF_DESC(en,gettext("Define the graphic device to use if possible")) \
 DRI_CONF_OPT_END
 
+#define DRI_CONF_DRI_DRIVER(def) \
+DRI_CONF_OPT_BEGIN(dri_driver, string, def) \
+        DRI_CONF_DESC(en,gettext("Override the DRI driver to load")) \
+DRI_CONF_OPT_END
+
 /**
  * \brief Gallium-Nine specific configuration options
  */
@@ -452,3 +337,8 @@ DRI_CONF_OPT_END
 DRI_CONF_OPT_BEGIN_B(radeonsi_clear_db_cache_before_clear, def) \
         DRI_CONF_DESC(en,"Clear DB cache before fast depth clear") \
 DRI_CONF_OPT_END
+
+#define DRI_CONF_RADEONSI_ZERO_ALL_VRAM_ALLOCS(def) \
+DRI_CONF_OPT_BEGIN_B(radeonsi_zerovram, def) \
+        DRI_CONF_DESC(en,"Zero all vram allocations") \
+DRI_CONF_OPT_END