pixman 0.12.0. Tested on a full ports build by naddy@.

author: Matthieu Herrb <matthieu@cvs.openbsd.org> 2008-09-23 19:11:41 +0000
committer: Matthieu Herrb <matthieu@cvs.openbsd.org> 2008-09-23 19:11:41 +0000
commit: 61c42f8ddcc41b05b5d28906533e3ca0b9aa1965 (patch)
tree: 2009f3d26880bf14c1b4e7953ee9b0cac8236df7 /lib/pixman
parent: f0d06ca61c104601dbdd10a4e58ae54f8d3f5332 (diff)
43 files changed, 12055 insertions, 3076 deletions
diff --git a/lib/pixman/Makefile.am b/lib/pixman/Makefile.am
index 542603eec..2cf692d5f 100644
--- a/lib/pixman/Makefile.am
+++ b/lib/pixman/Makefile.am
@@ -10,6 +10,7 @@ snapshot:
 	test -d "$(srcdir)/.git" && distdir=$$distdir-`cd "$(srcdir)" && git-rev-parse HEAD | cut -c 1-6`; \
 	$(MAKE) $(AM_MAKEFLAGS) distdir="$$distdir" dist
 
+GPGKEY=6FF7C1A8
 USERNAME=$$USER
 RELEASE_OR_SNAPSHOT = $$(if test "x$(CAIRO_VERSION_MINOR)" = "x$$(echo "$(CAIRO_VERSION_MINOR)/2*2" | bc)" ; then echo release; else echo snapshot; fi)
 RELEASE_CAIRO_HOST =	$(USERNAME)@cairographics.org
@@ -66,22 +67,24 @@ ensure-prev:
 	@if [[ "$(PREV)" == "" ]]; then							\
 		echo ""							          &&	\
 		echo "You must set the PREV variable on the make command line to" &&	\
-		echo "the git tag of the last release."				  &&	\
+		echo "the last version."				  &&	\
 		echo ""								  &&	\
 		echo "For example:"						  &&	\
-		echo "      make PREV=pixman-0.7.3"				  &&	\
+		echo "      make PREV=0.7.3"				  &&	\
 		echo ""								  &&	\
 		false;									\
 	fi
 
 release-check: ensure-prev release-verify-newer release-remove-old distcheck
 
+release-tag:
+	git-tag -u $(GPGKEY) -m "$(PACKAGE) $(VERSION) release" $(PACKAGE)-$(VERSION)
+
 release-upload: release-check $(tar_gz) $(tar_bz2) $(sha1_tgz) $(gpg_file)
 	mkdir -p releases
 	scp $(tar_gz) $(sha1_tgz) $(gpg_file) $(RELEASE_CAIRO_HOST):$(RELEASE_CAIRO_DIR)
 	scp $(tar_gz) $(tar_bz2) $(RELEASE_XORG_HOST):$(RELEASE_XORG_DIR)
 	ssh $(RELEASE_CAIRO_HOST) "rm -f $(RELEASE_CAIRO_DIR)/LATEST-$(PACKAGE)-[0-9]* && ln -s $(tar_gz) $(RELEASE_CAIRO_DIR)/LATEST-$(PACKAGE)-$(VERSION)"
-	git-tag -s -m "$(PACKAGE) $(VERSION) release" $(PACKAGE)-$(VERSION)
 
 release-publish-message: $(sha1_tgz) $(md5_tgz) $(sha1_tbz2) $(md5_tbz2) ensure-prev
 	@echo "Please follow the instructions in RELEASING to push stuff out and"
@@ -111,17 +114,17 @@ release-publish-message: $(sha1_tgz) $(md5_tgz) $(sha1_tbz2) $(md5_tbz2) ensure-
 	@echo ""
 	@echo "GPG signature:"
 	@echo "	$(RELEASE_CAIRO_URL)/$(gpg_file)"
-	@echo "	(signed by `getent passwd "$$USER" | cut -d: -f 5 | cut -d, -f 1`)"
+	@echo "	(signed by `git config --get user.name` <`git config --get user.email`>)"
 	@echo ""
 	@echo "Git:"
 	@echo "	git://git.freedesktop.org/git/pixman"
 	@echo "	tag: $(PACKAGE)-$(VERSION)"
 	@echo ""
 	@echo "Log:"
-	@git-log --no-merges "$(PACKAGE)-$(PREV)".."$(PACKAGE)-$(VERSION)" | git shortlog | awk '{ printf "\t"; print ; }' | cut -b0-80
+	@git-log --no-merges "$(PACKAGE)-$(PREV)".."$(PACKAGE)-$(VERSION)" | git shortlog | awk '{ printf "\t"; print ; }' | cut -b1-80
 	@echo "============================== CUT HERE =============================="
 	@echo ""
 
-release-publish: release-upload release-publish-message
+release-publish: release-upload release-tag release-publish-message
 
-.PHONY: release-upload release-publish release-publish-message
+.PHONY: release-upload release-publish release-publish-message release-tag
diff --git a/lib/pixman/Makefile.bsd-wrapper b/lib/pixman/Makefile.bsd-wrapper
index a8d4aaa8a..1350a7be3 100644
--- a/lib/pixman/Makefile.bsd-wrapper
+++ b/lib/pixman/Makefile.bsd-wrapper
@@ -1,5 +1,9 @@
-# $OpenBSD: Makefile.bsd-wrapper,v 1.5 2008/04/08 19:00:25 matthieu Exp $
+# $OpenBSD: Makefile.bsd-wrapper,v 1.6 2008/09/23 19:11:40 matthieu Exp $
 
-SHARED_LIBS=	pixman-1 11.0
+SHARED_LIBS=	pixman-1 12.0
+
+.if ${MACHINE} == amd64
+CONFIGURE_ARGS += --disable-sse2
+.endif
 
 .include <bsd.xorg.mk>
diff --git a/lib/pixman/Makefile.in b/lib/pixman/Makefile.in
index ddcae2103..7016a4800 100644
--- a/lib/pixman/Makefile.in
+++ b/lib/pixman/Makefile.in
@@ -38,9 +38,10 @@ build_triplet = @build@
 host_triplet = @host@
 DIST_COMMON = README $(am__configure_deps) $(srcdir)/Makefile.am \
 	$(srcdir)/Makefile.in $(srcdir)/config.h.in \
-	$(srcdir)/pixman-1.pc.in $(top_srcdir)/configure AUTHORS \
-	COPYING ChangeLog INSTALL NEWS TODO compile config.guess \
-	config.sub depcomp install-sh ltmain.sh missing
+	$(srcdir)/pixman-1-uninstalled.pc.in $(srcdir)/pixman-1.pc.in \
+	$(top_srcdir)/configure AUTHORS COPYING ChangeLog INSTALL NEWS \
+	TODO compile config.guess config.sub depcomp install-sh \
+	ltmain.sh missing
 subdir = .
 ACLOCAL_M4 = $(top_srcdir)/aclocal.m4
 am__aclocal_m4_deps = $(top_srcdir)/configure.ac
@@ -50,7 +51,7 @@ am__CONFIG_DISTCLEAN_FILES = config.status config.cache config.log \
  configure.lineno configure.status.lineno
 mkinstalldirs = $(SHELL) $(install_sh) -d
 CONFIG_HEADER = config.h
-CONFIG_CLEAN_FILES = pixman-1.pc
+CONFIG_CLEAN_FILES = pixman-1.pc pixman-1-uninstalled.pc
 SOURCES =
 DIST_SOURCES =
 RECURSIVE_TARGETS = all-recursive check-recursive dvi-recursive \
@@ -105,6 +106,7 @@ DEFS = @DEFS@
 DEPDIR = @DEPDIR@
 DEP_CFLAGS = @DEP_CFLAGS@
 DEP_LIBS = @DEP_LIBS@
+DSYMUTIL = @DSYMUTIL@
 ECHO = @ECHO@
 ECHO_C = @ECHO_C@
 ECHO_N = @ECHO_N@
@@ -130,6 +132,7 @@ LTLIBOBJS = @LTLIBOBJS@
 LT_VERSION_INFO = @LT_VERSION_INFO@
 MAKEINFO = @MAKEINFO@
 MMX_CFLAGS = @MMX_CFLAGS@
+NMEDIT = @NMEDIT@
 OBJEXT = @OBJEXT@
 PACKAGE = @PACKAGE@
 PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
@@ -138,26 +141,31 @@ PACKAGE_STRING = @PACKAGE_STRING@
 PACKAGE_TARNAME = @PACKAGE_TARNAME@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
+PERL = @PERL@
 PIXMAN_VERSION_MAJOR = @PIXMAN_VERSION_MAJOR@
 PIXMAN_VERSION_MICRO = @PIXMAN_VERSION_MICRO@
 PIXMAN_VERSION_MINOR = @PIXMAN_VERSION_MINOR@
 PKG_CONFIG = @PKG_CONFIG@
 RANLIB = @RANLIB@
+SED = @SED@
 SET_MAKE = @SET_MAKE@
 SHELL = @SHELL@
-SSE_CFLAGS = @SSE_CFLAGS@
+SSE2_CFLAGS = @SSE2_CFLAGS@
 STRIP = @STRIP@
 USE_MMX_FALSE = @USE_MMX_FALSE@
 USE_MMX_TRUE = @USE_MMX_TRUE@
 USE_SSE2_FALSE = @USE_SSE2_FALSE@
 USE_SSE2_TRUE = @USE_SSE2_TRUE@
-USE_SSE_FALSE = @USE_SSE_FALSE@
-USE_SSE_TRUE = @USE_SSE_TRUE@
+USE_VMX_FALSE = @USE_VMX_FALSE@
+USE_VMX_TRUE = @USE_VMX_TRUE@
 VERSION = @VERSION@
+VMX_CFLAGS = @VMX_CFLAGS@
 ac_ct_AR = @ac_ct_AR@
 ac_ct_CC = @ac_ct_CC@
 ac_ct_CXX = @ac_ct_CXX@
+ac_ct_DSYMUTIL = @ac_ct_DSYMUTIL@
 ac_ct_F77 = @ac_ct_F77@
+ac_ct_NMEDIT = @ac_ct_NMEDIT@
 ac_ct_RANLIB = @ac_ct_RANLIB@
 ac_ct_STRIP = @ac_ct_STRIP@
 ac_pt_PKG_CONFIG = @ac_pt_PKG_CONFIG@
@@ -201,6 +209,7 @@ target_alias = @target_alias@
 SUBDIRS = pixman test
 pkgconfigdir = $(libdir)/pkgconfig
 pkgconfig_DATA = pixman-1.pc
+GPGKEY = 6FF7C1A8
 USERNAME = $$USER
 RELEASE_OR_SNAPSHOT = $$(if test "x$(CAIRO_VERSION_MINOR)" = "x$$(echo "$(CAIRO_VERSION_MINOR)/2*2" | bc)" ; then echo release; else echo snapshot; fi)
 RELEASE_CAIRO_HOST = $(USERNAME)@cairographics.org
@@ -273,6 +282,8 @@ distclean-hdr:
 	-rm -f config.h stamp-h1
 pixman-1.pc: $(top_builddir)/config.status $(srcdir)/pixman-1.pc.in
 	cd $(top_builddir) && $(SHELL) ./config.status $@
+pixman-1-uninstalled.pc: $(top_builddir)/config.status $(srcdir)/pixman-1-uninstalled.pc.in
+	cd $(top_builddir) && $(SHELL) ./config.status $@
 
 mostlyclean-libtool:
 	-rm -f *.lo
@@ -719,22 +730,24 @@ ensure-prev:
 	@if [[ "$(PREV)" == "" ]]; then							\
 		echo ""							          &&	\
 		echo "You must set the PREV variable on the make command line to" &&	\
-		echo "the git tag of the last release."				  &&	\
+		echo "the last version."				  &&	\
 		echo ""								  &&	\
 		echo "For example:"						  &&	\
-		echo "      make PREV=pixman-0.7.3"				  &&	\
+		echo "      make PREV=0.7.3"				  &&	\
 		echo ""								  &&	\
 		false;									\
 	fi
 
 release-check: ensure-prev release-verify-newer release-remove-old distcheck
 
+release-tag:
+	git-tag -u $(GPGKEY) -m "$(PACKAGE) $(VERSION) release" $(PACKAGE)-$(VERSION)
+
 release-upload: release-check $(tar_gz) $(tar_bz2) $(sha1_tgz) $(gpg_file)
 	mkdir -p releases
 	scp $(tar_gz) $(sha1_tgz) $(gpg_file) $(RELEASE_CAIRO_HOST):$(RELEASE_CAIRO_DIR)
 	scp $(tar_gz) $(tar_bz2) $(RELEASE_XORG_HOST):$(RELEASE_XORG_DIR)
 	ssh $(RELEASE_CAIRO_HOST) "rm -f $(RELEASE_CAIRO_DIR)/LATEST-$(PACKAGE)-[0-9]* && ln -s $(tar_gz) $(RELEASE_CAIRO_DIR)/LATEST-$(PACKAGE)-$(VERSION)"
-	git-tag -s -m "$(PACKAGE) $(VERSION) release" $(PACKAGE)-$(VERSION)
 
 release-publish-message: $(sha1_tgz) $(md5_tgz) $(sha1_tbz2) $(md5_tbz2) ensure-prev
 	@echo "Please follow the instructions in RELEASING to push stuff out and"
@@ -764,20 +777,20 @@ release-publish-message: $(sha1_tgz) $(md5_tgz) $(sha1_tbz2) $(md5_tbz2) ensure-
 	@echo ""
 	@echo "GPG signature:"
 	@echo "	$(RELEASE_CAIRO_URL)/$(gpg_file)"
-	@echo "	(signed by `getent passwd "$$USER" | cut -d: -f 5 | cut -d, -f 1`)"
+	@echo "	(signed by `git config --get user.name` <`git config --get user.email`>)"
 	@echo ""
 	@echo "Git:"
 	@echo "	git://git.freedesktop.org/git/pixman"
 	@echo "	tag: $(PACKAGE)-$(VERSION)"
 	@echo ""
 	@echo "Log:"
-	@git-log --no-merges "$(PACKAGE)-$(PREV)".."$(PACKAGE)-$(VERSION)" | git shortlog | awk '{ printf "\t"; print ; }' | cut -b0-80
+	@git-log --no-merges "$(PACKAGE)-$(PREV)".."$(PACKAGE)-$(VERSION)" | git shortlog | awk '{ printf "\t"; print ; }' | cut -b1-80
 	@echo "============================== CUT HERE =============================="
 	@echo ""
 
-release-publish: release-upload release-publish-message
+release-publish: release-upload release-tag release-publish-message
 
-.PHONY: release-upload release-publish release-publish-message
+.PHONY: release-upload release-publish release-publish-message release-tag
 # 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/pixman/TODO b/lib/pixman/TODO
index 7ff58def8..47b9bc9d0 100644
--- a/lib/pixman/TODO
+++ b/lib/pixman/TODO
@@ -1,3 +1,11 @@
+  - SSE 2 issues:
+
+      - Use MM_HINT_NTA instead of MM_HINT_T0
+
+      - Use of fbCompositeOver_x888x8x8888sse2()
+
+      - Use pixmanFillsse2 and pixmanBltsse2
+
   - Update the RLEASING file
 
   - Things to keep in mind if breaking ABI:
@@ -19,7 +27,7 @@
         (They only exist because they are theoretically available to
         drivers).
 
-      - Regions should become 32 bit
+      - 16 bit regions should be deleted
 
       - There should only be one trap rasterization API.
 
@@ -27,9 +35,20 @@
         to indicate the actual depth. That way PIXMAN_x4c4 and PIXMAN_c8
 	won't collide.
 
+  - Maybe bite the bullet and make configure.ac generate a pixman-types.h
+    file that can be included from pixman.h to avoid the #ifdef magic
+    in pixman.h
+
   - Make pixman_region_point_in() survive a NULL box, then fix up
     pixman-compose.c
 
+      - Possibly look into inlining the fetch functions
+
+  - There is a bug with source clipping demonstrated by clip-test in the
+    test directory. If we interprete source clipping as given in
+    destination coordinates, which is probably the only sane choice,
+    then the result should have two red bars down the sides.
+    
   - Test suite
 
   - Add a general way of dealing with architecture specific
@@ -61,11 +80,6 @@
         this, I suggest just using 32-bit datatypes by setting
         IC_SHIFT to 5 for all machines.
 
-  - Consider whether calling regions region16 is really such a great
-    idea. Vlad wants 32 bit regions for Cairo. This will break X server
-    ABI, but should otherwise be mostly harmless, though a
-    pixman_region_get_boxes16() may be useful.
-
   - Consider optimizing the 8/16 bit solid fills in pixman-util.c by
     storing more than one value at a time.
 
@@ -79,6 +93,37 @@
     (0, 0). Cairo would have to make sure that the delta *within* a
     batch of trapezoids does not exceed 16 bit.
 
+  - Consider adding actual backends. Brain dump:
+
+    A backend is something that knows how to
+
+      - Create images
+      - Composite three images
+      - Rasterize trapezoids
+      - Do solid fills and blits
+
+    These operations are provided by a vtable that the backend will
+    create when it is initialized. Initial backends:
+
+      - VMX
+      - SSE2
+      - MMX
+      - Plain Old C
+
+    When the SIMD backends are initialized, they will be passed a
+    pointer to the Plain Old C backend that they can use for fallback
+    purposes.
+
+    Images would gain a vtable as well that would contain things like
+
+      - Read scanline
+      - Write scanline
+
+    (Or even read_patch/write_patch as suggested by Keith a while
+    back).
+
+    This could simplify the compositing code considerably.
+
   - Review the pixman_format_code_t enum to make sure it will support
     future formats. Some formats we will probably need:
 
@@ -130,8 +175,31 @@
 
     What about color spaces such a linear vs. srGB etc.?
 
+
 done:
 
+- Be consistent about calling sse2 sse2
+
+- Rename "SSE" to "MMX_EXTENSIONS". (Deleted mmx extensions).
+
+- Commented-out uses of fbCompositeCopyAreasse2()
+
+- Consider whether calling regions region16 is really such a great
+  idea. Vlad wants 32 bit regions for Cairo. This will break X server
+  ABI, but should otherwise be mostly harmless, though a
+  pixman_region_get_boxes16() may be useful.
+
+- Altivec signal issue (Company has fix, there is also a patch by
+  dwmw2 in rawhide).
+
+- Behdad's MMX issue - see list
+
+- SSE2 issues:
+    - Crashes in Mozilla because of unaligned stack. Possible fixes
+        - Make use of gcc 4.2 feature to align the stack
+        - Write some sort of trampoline that aligns the stack
+          before calling SSE functions.
+
 - Get rid of the switch-of-doom; replace it with a big table
   describing the various fast paths.
 
diff --git a/lib/pixman/aclocal.m4 b/lib/pixman/aclocal.m4
index bfbf5e1d9..b7278b304 100644
--- a/lib/pixman/aclocal.m4
+++ b/lib/pixman/aclocal.m4
@@ -13,7 +13,7 @@
 
 # libtool.m4 - Configure libtool for the host system. -*-Autoconf-*-
 
-# serial 48 AC_PROG_LIBTOOL
+# serial 52 AC_PROG_LIBTOOL
 
 
 # AC_PROVIDE_IFELSE(MACRO-NAME, IF-PROVIDED, IF-NOT-PROVIDED)
@@ -101,7 +101,6 @@ AC_REQUIRE([AC_DEPLIBS_CHECK_METHOD])dnl
 AC_REQUIRE([AC_OBJEXT])dnl
 AC_REQUIRE([AC_EXEEXT])dnl
 dnl
-
 AC_LIBTOOL_SYS_MAX_CMD_LEN
 AC_LIBTOOL_SYS_GLOBAL_SYMBOL_PIPE
 AC_LIBTOOL_OBJDIR
@@ -176,7 +175,7 @@ test -z "$STRIP" && STRIP=:
 test -z "$ac_objext" && ac_objext=o
 
 # Determine commands to create old-style static archives.
-old_archive_cmds='$AR $AR_FLAGS $oldlib$oldobjs$old_deplibs'
+old_archive_cmds='$AR $AR_FLAGS $oldlib$oldobjs'
 old_postinstall_cmds='chmod 644 $oldlib'
 old_postuninstall_cmds=
 
@@ -203,6 +202,8 @@ file_magic*)
   ;;
 esac
 
+_LT_REQUIRED_DARWIN_CHECKS
+
 AC_PROVIDE_IFELSE([AC_LIBTOOL_DLOPEN], enable_dlopen=yes, enable_dlopen=no)
 AC_PROVIDE_IFELSE([AC_LIBTOOL_WIN32_DLL],
 enable_win32_dll=yes, enable_win32_dll=no)
@@ -263,8 +264,9 @@ cc_basename=`$echo "X$cc_temp" | $Xsed -e 's%.*/%%' -e "s%^$host_alias-%%"`
 # Check for compiler boilerplate output or warnings with
 # the simple compiler test code.
 AC_DEFUN([_LT_COMPILER_BOILERPLATE],
-[ac_outfile=conftest.$ac_objext
-printf "$lt_simple_compile_test_code" >conftest.$ac_ext
+[AC_REQUIRE([LT_AC_PROG_SED])dnl
+ac_outfile=conftest.$ac_objext
+echo "$lt_simple_compile_test_code" >conftest.$ac_ext
 eval "$ac_compile" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_compiler_boilerplate=`cat conftest.err`
 $rm conftest*
@@ -276,13 +278,85 @@ $rm conftest*
 # Check for linker boilerplate output or warnings with
 # the simple link test code.
 AC_DEFUN([_LT_LINKER_BOILERPLATE],
-[ac_outfile=conftest.$ac_objext
-printf "$lt_simple_link_test_code" >conftest.$ac_ext
+[AC_REQUIRE([LT_AC_PROG_SED])dnl
+ac_outfile=conftest.$ac_objext
+echo "$lt_simple_link_test_code" >conftest.$ac_ext
 eval "$ac_link" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_linker_boilerplate=`cat conftest.err`
-$rm conftest*
+$rm -r conftest*
 ])# _LT_LINKER_BOILERPLATE
 
+# _LT_REQUIRED_DARWIN_CHECKS
+# --------------------------
+# Check for some things on darwin
+AC_DEFUN([_LT_REQUIRED_DARWIN_CHECKS],[
+  case $host_os in
+    rhapsody* | darwin*)
+    AC_CHECK_TOOL([DSYMUTIL], [dsymutil], [:])
+    AC_CHECK_TOOL([NMEDIT], [nmedit], [:])
+
+    AC_CACHE_CHECK([for -single_module linker flag],[lt_cv_apple_cc_single_mod],
+      [lt_cv_apple_cc_single_mod=no
+      if test -z "${LT_MULTI_MODULE}"; then
+   # By default we will add the -single_module flag. You can override
+   # by either setting the environment variable LT_MULTI_MODULE
+   # non-empty at configure time, or by adding -multi_module to the
+   # link flags.
+   echo "int foo(void){return 1;}" > conftest.c
+   $LTCC $LTCFLAGS $LDFLAGS -o libconftest.dylib \
+     -dynamiclib ${wl}-single_module conftest.c
+   if test -f libconftest.dylib; then
+     lt_cv_apple_cc_single_mod=yes
+     rm -rf libconftest.dylib*
+   fi
+   rm conftest.c
+      fi])
+    AC_CACHE_CHECK([for -exported_symbols_list linker flag],
+      [lt_cv_ld_exported_symbols_list],
+      [lt_cv_ld_exported_symbols_list=no
+      save_LDFLAGS=$LDFLAGS
+      echo "_main" > conftest.sym
+      LDFLAGS="$LDFLAGS -Wl,-exported_symbols_list,conftest.sym"
+      AC_LINK_IFELSE([AC_LANG_PROGRAM([],[])],
+   [lt_cv_ld_exported_symbols_list=yes],
+   [lt_cv_ld_exported_symbols_list=no])
+   LDFLAGS="$save_LDFLAGS"
+    ])
+    case $host_os in
+    rhapsody* | darwin1.[[0123]])
+      _lt_dar_allow_undefined='${wl}-undefined ${wl}suppress' ;;
+    darwin1.*)
+     _lt_dar_allow_undefined='${wl}-flat_namespace ${wl}-undefined ${wl}suppress' ;;
+    darwin*)
+      # if running on 10.5 or later, the deployment target defaults
+      # to the OS version, if on x86, and 10.4, the deployment
+      # target defaults to 10.4. Don't you love it?
+      case ${MACOSX_DEPLOYMENT_TARGET-10.0},$host in
+   10.0,*86*-darwin8*|10.0,*-darwin[[91]]*)
+     _lt_dar_allow_undefined='${wl}-undefined ${wl}dynamic_lookup' ;;
+   10.[[012]]*)
+     _lt_dar_allow_undefined='${wl}-flat_namespace ${wl}-undefined ${wl}suppress' ;;
+   10.*)
+     _lt_dar_allow_undefined='${wl}-undefined ${wl}dynamic_lookup' ;;
+      esac
+    ;;
+  esac
+    if test "$lt_cv_apple_cc_single_mod" = "yes"; then
+      _lt_dar_single_mod='$single_module'
+    fi
+    if test "$lt_cv_ld_exported_symbols_list" = "yes"; then
+      _lt_dar_export_syms=' ${wl}-exported_symbols_list,$output_objdir/${libname}-symbols.expsym'
+    else
+      _lt_dar_export_syms="~$NMEDIT -s \$output_objdir/\${libname}-symbols.expsym \${lib}"
+    fi
+    if test "$DSYMUTIL" != ":"; then
+      _lt_dsymutil="~$DSYMUTIL \$lib || :"
+    else
+      _lt_dsymutil=
+    fi
+    ;;
+  esac
+])
 
 # _LT_AC_SYS_LIBPATH_AIX
 # ----------------------
@@ -293,12 +367,20 @@ $rm conftest*
 # If we don't find anything, use the default library path according
 # to the aix ld manual.
 AC_DEFUN([_LT_AC_SYS_LIBPATH_AIX],
-[AC_LINK_IFELSE(AC_LANG_PROGRAM,[
-aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`
+[AC_REQUIRE([LT_AC_PROG_SED])dnl
+AC_LINK_IFELSE(AC_LANG_PROGRAM,[
+lt_aix_libpath_sed='
+    /Import File Strings/,/^$/ {
+	/^0/ {
+	    s/^0  *\(.*\)$/\1/
+	    p
+	}
+    }'
+aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
 # Check for a 64-bit object if we didn't find anything.
-if test -z "$aix_libpath"; then aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`; fi],[])
+if test -z "$aix_libpath"; then
+  aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
+fi],[])
 if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 ])# _LT_AC_SYS_LIBPATH_AIX
 
@@ -529,13 +611,17 @@ ia64-*-hpux*)
   rm -rf conftest*
   ;;
 
-x86_64-*linux*|ppc*-*linux*|powerpc*-*linux*|s390*-*linux*|sparc*-*linux*)
+x86_64-*kfreebsd*-gnu|x86_64-*linux*|ppc*-*linux*|powerpc*-*linux*| \
+s390*-*linux*|sparc*-*linux*)
   # Find out which ABI we are using.
   echo 'int i;' > conftest.$ac_ext
   if AC_TRY_EVAL(ac_compile); then
     case `/usr/bin/file conftest.o` in
     *32-bit*)
       case $host in
+        x86_64-*kfreebsd*-gnu)
+          LD="${LD-ld} -m elf_i386_fbsd"
+          ;;
         x86_64-*linux*)
           LD="${LD-ld} -m elf_i386"
           ;;
@@ -552,6 +638,9 @@ x86_64-*linux*|ppc*-*linux*|powerpc*-*linux*|s390*-*linux*|sparc*-*linux*)
       ;;
     *64-bit*)
       case $host in
+        x86_64-*kfreebsd*-gnu)
+          LD="${LD-ld} -m elf_x86_64_fbsd"
+          ;;
         x86_64-*linux*)
           LD="${LD-ld} -m elf_x86_64"
           ;;
@@ -592,7 +681,11 @@ sparc*-*solaris*)
     *64-bit*)
       case $lt_cv_prog_gnu_ld in
       yes*) LD="${LD-ld} -m elf64_sparc" ;;
-      *)    LD="${LD-ld} -64" ;;
+      *)
+        if ${LD-ld} -64 -r -o conftest2.o conftest.o >/dev/null 2>&1; then
+	  LD="${LD-ld} -64"
+	fi
+	;;
       esac
       ;;
     esac
@@ -623,7 +716,7 @@ AC_DEFUN([AC_LIBTOOL_COMPILER_OPTION],
 AC_CACHE_CHECK([$1], [$2],
   [$2=no
   ifelse([$4], , [ac_outfile=conftest.$ac_objext], [ac_outfile=$4])
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
    lt_compiler_flag="$3"
    # Insert the option either (1) after the last *FLAGS variable, or
    # (2) before a word containing "conftest.", or (3) at the end.
@@ -664,11 +757,12 @@ fi
 # ------------------------------------------------------------
 # Check whether the given compiler option works
 AC_DEFUN([AC_LIBTOOL_LINKER_OPTION],
-[AC_CACHE_CHECK([$1], [$2],
+[AC_REQUIRE([LT_AC_PROG_SED])dnl
+AC_CACHE_CHECK([$1], [$2],
   [$2=no
    save_LDFLAGS="$LDFLAGS"
    LDFLAGS="$LDFLAGS $3"
-   printf "$lt_simple_link_test_code" > conftest.$ac_ext
+   echo "$lt_simple_link_test_code" > conftest.$ac_ext
    if (eval $ac_link 2>conftest.err) && test -s conftest$ac_exeext; then
      # The linker can only warn and ignore the option if not recognized
      # So say no if there are warnings
@@ -684,7 +778,7 @@ AC_DEFUN([AC_LIBTOOL_LINKER_OPTION],
        $2=yes
      fi
    fi
-   $rm conftest*
+   $rm -r conftest*
    LDFLAGS="$save_LDFLAGS"
 ])
 
@@ -782,24 +876,27 @@ AC_CACHE_VAL([lt_cv_sys_max_cmd_len], [dnl
     fi
     ;;
   *)
-    # If test is not a shell built-in, we'll probably end up computing a
-    # maximum length that is only half of the actual maximum length, but
-    # we can't tell.
-    SHELL=${SHELL-${CONFIG_SHELL-/bin/sh}}
-    while (test "X"`$SHELL [$]0 --fallback-echo "X$teststring" 2>/dev/null` \
+    lt_cv_sys_max_cmd_len=`(getconf ARG_MAX) 2> /dev/null`
+    if test -n "$lt_cv_sys_max_cmd_len"; then
+      lt_cv_sys_max_cmd_len=`expr $lt_cv_sys_max_cmd_len \/ 4`
+      lt_cv_sys_max_cmd_len=`expr $lt_cv_sys_max_cmd_len \* 3`
+    else
+      SHELL=${SHELL-${CONFIG_SHELL-/bin/sh}}
+      while (test "X"`$SHELL [$]0 --fallback-echo "X$teststring" 2>/dev/null` \
 	       = "XX$teststring") >/dev/null 2>&1 &&
-	    new_result=`expr "X$teststring" : ".*" 2>&1` &&
-	    lt_cv_sys_max_cmd_len=$new_result &&
-	    test $i != 17 # 1/2 MB should be enough
-    do
-      i=`expr $i + 1`
-      teststring=$teststring$teststring
-    done
-    teststring=
-    # Add a significant safety factor because C++ compilers can tack on massive
-    # amounts of additional arguments before passing them to the linker.
-    # It appears as though 1/2 is a usable value.
-    lt_cv_sys_max_cmd_len=`expr $lt_cv_sys_max_cmd_len \/ 2`
+	      new_result=`expr "X$teststring" : ".*" 2>&1` &&
+	      lt_cv_sys_max_cmd_len=$new_result &&
+	      test $i != 17 # 1/2 MB should be enough
+      do
+        i=`expr $i + 1`
+        teststring=$teststring$teststring
+      done
+      teststring=
+      # Add a significant safety factor because C++ compilers can tack on massive
+      # amounts of additional arguments before passing them to the linker.
+      # It appears as though 1/2 is a usable value.
+      lt_cv_sys_max_cmd_len=`expr $lt_cv_sys_max_cmd_len \/ 2`
+    fi
     ;;
   esac
 ])
@@ -952,7 +1049,7 @@ else
     AC_CHECK_FUNC([shl_load],
 	  [lt_cv_dlopen="shl_load"],
       [AC_CHECK_LIB([dld], [shl_load],
-	    [lt_cv_dlopen="shl_load" lt_cv_dlopen_libs="-dld"],
+	    [lt_cv_dlopen="shl_load" lt_cv_dlopen_libs="-ldld"],
 	[AC_CHECK_FUNC([dlopen],
 	      [lt_cv_dlopen="dlopen"],
 	  [AC_CHECK_LIB([dl], [dlopen],
@@ -960,7 +1057,7 @@ else
 	    [AC_CHECK_LIB([svld], [dlopen],
 		  [lt_cv_dlopen="dlopen" lt_cv_dlopen_libs="-lsvld"],
 	      [AC_CHECK_LIB([dld], [dld_link],
-		    [lt_cv_dlopen="dld_link" lt_cv_dlopen_libs="-dld"])
+		    [lt_cv_dlopen="dld_link" lt_cv_dlopen_libs="-ldld"])
 	      ])
 	    ])
 	  ])
@@ -1026,7 +1123,8 @@ fi
 # ---------------------------------
 # Check to see if options -c and -o are simultaneously supported by compiler
 AC_DEFUN([AC_LIBTOOL_PROG_CC_C_O],
-[AC_REQUIRE([_LT_AC_SYS_COMPILER])dnl
+[AC_REQUIRE([LT_AC_PROG_SED])dnl
+AC_REQUIRE([_LT_AC_SYS_COMPILER])dnl
 AC_CACHE_CHECK([if $compiler supports -c -o file.$ac_objext],
   [_LT_AC_TAGVAR(lt_cv_prog_compiler_c_o, $1)],
   [_LT_AC_TAGVAR(lt_cv_prog_compiler_c_o, $1)=no
@@ -1034,7 +1132,7 @@ AC_CACHE_CHECK([if $compiler supports -c -o file.$ac_objext],
    mkdir conftest
    cd conftest
    mkdir out
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
 
    lt_compiler_flag="-o out/conftest2.$ac_objext"
    # Insert the option either (1) after the last *FLAGS variable, or
@@ -1174,6 +1272,7 @@ else
    darwin*)
        if test -n "$STRIP" ; then
          striplib="$STRIP -x"
+         old_striplib="$STRIP -S"
          AC_MSG_RESULT([yes])
        else
   AC_MSG_RESULT([no])
@@ -1191,7 +1290,8 @@ fi
 # -----------------------------
 # PORTME Fill in your ld.so characteristics
 AC_DEFUN([AC_LIBTOOL_SYS_DYNAMIC_LINKER],
-[AC_MSG_CHECKING([dynamic linker characteristics])
+[AC_REQUIRE([LT_AC_PROG_SED])dnl
+AC_MSG_CHECKING([dynamic linker characteristics])
 library_names_spec=
 libname_spec='lib$name'
 soname_spec=
@@ -1205,20 +1305,58 @@ shlibpath_overrides_runpath=unknown
 version_type=none
 dynamic_linker="$host_os ld.so"
 sys_lib_dlsearch_path_spec="/lib /usr/lib"
+m4_if($1,[],[
 if test "$GCC" = yes; then
-  sys_lib_search_path_spec=`$CC -print-search-dirs | grep "^libraries:" | $SED -e "s/^libraries://" -e "s,=/,/,g"`
-  if echo "$sys_lib_search_path_spec" | grep ';' >/dev/null ; then
+  case $host_os in
+    darwin*) lt_awk_arg="/^libraries:/,/LR/" ;;
+    *) lt_awk_arg="/^libraries:/" ;;
+  esac
+  lt_search_path_spec=`$CC -print-search-dirs | awk $lt_awk_arg | $SED -e "s/^libraries://" -e "s,=/,/,g"`
+  if echo "$lt_search_path_spec" | grep ';' >/dev/null ; then
     # if the path contains ";" then we assume it to be the separator
     # otherwise default to the standard path separator (i.e. ":") - it is
     # assumed that no part of a normal pathname contains ";" but that should
     # okay in the real world where ";" in dirpaths is itself problematic.
-    sys_lib_search_path_spec=`echo "$sys_lib_search_path_spec" | $SED -e 's/;/ /g'`
+    lt_search_path_spec=`echo "$lt_search_path_spec" | $SED -e 's/;/ /g'`
   else
-    sys_lib_search_path_spec=`echo "$sys_lib_search_path_spec" | $SED  -e "s/$PATH_SEPARATOR/ /g"`
+    lt_search_path_spec=`echo "$lt_search_path_spec" | $SED  -e "s/$PATH_SEPARATOR/ /g"`
   fi
+  # Ok, now we have the path, separated by spaces, we can step through it
+  # and add multilib dir if necessary.
+  lt_tmp_lt_search_path_spec=
+  lt_multi_os_dir=`$CC $CPPFLAGS $CFLAGS $LDFLAGS -print-multi-os-directory 2>/dev/null`
+  for lt_sys_path in $lt_search_path_spec; do
+    if test -d "$lt_sys_path/$lt_multi_os_dir"; then
+      lt_tmp_lt_search_path_spec="$lt_tmp_lt_search_path_spec $lt_sys_path/$lt_multi_os_dir"
+    else
+      test -d "$lt_sys_path" && \
+	lt_tmp_lt_search_path_spec="$lt_tmp_lt_search_path_spec $lt_sys_path"
+    fi
+  done
+  lt_search_path_spec=`echo $lt_tmp_lt_search_path_spec | awk '
+BEGIN {RS=" "; FS="/|\n";} {
+  lt_foo="";
+  lt_count=0;
+  for (lt_i = NF; lt_i > 0; lt_i--) {
+    if ($lt_i != "" && $lt_i != ".") {
+      if ($lt_i == "..") {
+        lt_count++;
+      } else {
+        if (lt_count == 0) {
+          lt_foo="/" $lt_i lt_foo;
+        } else {
+          lt_count--;
+        }
+      }
+    }
+  }
+  if (lt_foo != "") { lt_freq[[lt_foo]]++; }
+  if (lt_freq[[lt_foo]] == 1) { print lt_foo; }
+}'`
+  sys_lib_search_path_spec=`echo $lt_search_path_spec`
 else
   sys_lib_search_path_spec="/lib /usr/lib /usr/local/lib"
-fi
+fi])
 need_lib_prefix=unknown
 hardcode_into_libs=no
 
@@ -1236,7 +1374,7 @@ aix3*)
   soname_spec='${libname}${release}${shared_ext}$major'
   ;;
 
-aix4* | aix5*)
+aix[[4-9]]*)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -1375,12 +1513,8 @@ darwin* | rhapsody*)
   shlibpath_overrides_runpath=yes
   shlibpath_var=DYLD_LIBRARY_PATH
   shrext_cmds='`test .$module = .yes && echo .so || echo .dylib`'
-  # Apple's gcc prints 'gcc -print-search-dirs' doesn't operate the same.
-  if test "$GCC" = yes; then
-    sys_lib_search_path_spec=`$CC -print-search-dirs | tr "\n" "$PATH_SEPARATOR" | sed -e 's/libraries:/@libraries:/' | tr "@" "\n" | grep "^libraries:" | sed -e "s/^libraries://" -e "s,=/,/,g" -e "s,$PATH_SEPARATOR, ,g" -e "s,.*,& /lib /usr/lib /usr/local/lib,g"`
-  else
-    sys_lib_search_path_spec='/lib /usr/lib /usr/local/lib'
-  fi
+  m4_if([$1], [],[
+  sys_lib_search_path_spec="$sys_lib_search_path_spec /usr/local/lib"]) 
   sys_lib_dlsearch_path_spec='/usr/local/lib /lib /usr/lib'
   ;;
 
@@ -1397,18 +1531,6 @@ freebsd1*)
   dynamic_linker=no
   ;;
 
-kfreebsd*-gnu)
-  version_type=linux
-  need_lib_prefix=no
-  need_version=no
-  library_names_spec='${libname}${release}${shared_ext}$versuffix ${libname}${release}${shared_ext}$major ${libname}${shared_ext}'
-  soname_spec='${libname}${release}${shared_ext}$major'
-  shlibpath_var=LD_LIBRARY_PATH
-  shlibpath_overrides_runpath=no
-  hardcode_into_libs=yes
-  dynamic_linker='GNU ld.so'
-  ;;
-
 freebsd* | dragonfly*)
   # DragonFly does not have aout.  When/if they implement a new
   # versioning mechanism, adjust this.
@@ -1446,7 +1568,7 @@ freebsd* | dragonfly*)
     shlibpath_overrides_runpath=no
     hardcode_into_libs=yes
     ;;
-  freebsd*) # from 4.6 on
+  *) # from 4.6 on, and DragonFly
     shlibpath_overrides_runpath=yes
     hardcode_into_libs=yes
     ;;
@@ -1509,7 +1631,7 @@ hpux9* | hpux10* | hpux11*)
   postinstall_cmds='chmod 555 $lib'
   ;;
 
-interix3*)
+interix[[3-9]]*)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -1564,7 +1686,7 @@ linux*oldld* | linux*aout* | linux*coff*)
   ;;
 
 # This must be Linux ELF.
-linux*)
+linux* | k*bsd*-gnu)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -1580,7 +1702,7 @@ linux*)
 
   # Append ld.so.conf contents to the search path
   if test -f /etc/ld.so.conf; then
-    lt_ld_extra=`awk '/^include / { system(sprintf("cd /etc; cat %s", \[$]2)); skip = 1; } { if (!skip) print \[$]0; skip = 0; }' < /etc/ld.so.conf | $SED -e 's/#.*//;s/[:,	]/ /g;s/=[^=]*$//;s/=[^= ]* / /g;/^$/d' | tr '\n' ' '`
+    lt_ld_extra=`awk '/^include / { system(sprintf("cd /etc; cat %s 2>/dev/null", \[$]2)); skip = 1; } { if (!skip) print \[$]0; skip = 0; }' < /etc/ld.so.conf | $SED -e 's/#.*//;/^[ 	]*hwcap[ 	]/d;s/[:,	]/ /g;s/=[^=]*$//;s/=[^= ]* / /g;/^$/d' | tr '\n' ' '`
     sys_lib_dlsearch_path_spec="/lib /usr/lib $lt_ld_extra"
   fi
 
@@ -1593,18 +1715,6 @@ linux*)
   dynamic_linker='GNU/Linux ld.so'
   ;;
 
-knetbsd*-gnu)
-  version_type=linux
-  need_lib_prefix=no
-  need_version=no
-  library_names_spec='${libname}${release}${shared_ext}$versuffix ${libname}${release}${shared_ext}$major ${libname}${shared_ext}'
-  soname_spec='${libname}${release}${shared_ext}$major'
-  shlibpath_var=LD_LIBRARY_PATH
-  shlibpath_overrides_runpath=no
-  hardcode_into_libs=yes
-  dynamic_linker='GNU ld.so'
-  ;;
-
 netbsd*)
   version_type=sunos
   need_lib_prefix=no
@@ -1686,6 +1796,10 @@ osf3* | osf4* | osf5*)
   sys_lib_dlsearch_path_spec="$sys_lib_search_path_spec"
   ;;
 
+rdos*)
+  dynamic_linker=no
+  ;;
+
 solaris*)
   version_type=linux
   need_lib_prefix=no
@@ -1781,6 +1895,13 @@ esac
 AC_MSG_RESULT([$dynamic_linker])
 test "$dynamic_linker" = no && can_build_shared=no
 
+AC_CACHE_VAL([lt_cv_sys_lib_search_path_spec],
+[lt_cv_sys_lib_search_path_spec="$sys_lib_search_path_spec"])
+sys_lib_search_path_spec="$lt_cv_sys_lib_search_path_spec"
+AC_CACHE_VAL([lt_cv_sys_lib_dlsearch_path_spec],
+[lt_cv_sys_lib_dlsearch_path_spec="$sys_lib_dlsearch_path_spec"])
+sys_lib_dlsearch_path_spec="$lt_cv_sys_lib_dlsearch_path_spec"
+
 variables_saved_for_relink="PATH $shlibpath_var $runpath_var"
 if test "$GCC" = yes; then
   variables_saved_for_relink="$variables_saved_for_relink GCC_EXEC_PREFIX COMPILER_PATH LIBRARY_PATH"
@@ -1791,7 +1912,8 @@ fi
 # _LT_AC_TAGCONFIG
 # ----------------
 AC_DEFUN([_LT_AC_TAGCONFIG],
-[AC_ARG_WITH([tags],
+[AC_REQUIRE([LT_AC_PROG_SED])dnl
+AC_ARG_WITH([tags],
     [AC_HELP_STRING([--with-tags@<:@=TAGS@:>@],
         [include additional configurations @<:@automatic@:>@])],
     [tagnames="$withval"])
@@ -2052,7 +2174,7 @@ m4_ifndef([AC_PROG_EGREP], [AC_DEFUN([AC_PROG_EGREP],
 
 # AC_PATH_TOOL_PREFIX
 # -------------------
-# find a file program which can recognise shared library
+# find a file program which can recognize shared library
 AC_DEFUN([AC_PATH_TOOL_PREFIX],
 [AC_REQUIRE([AC_PROG_EGREP])dnl
 AC_MSG_CHECKING([for $1])
@@ -2115,7 +2237,7 @@ fi
 
 # AC_PATH_MAGIC
 # -------------
-# find a file program which can recognise a shared library
+# find a file program which can recognize a shared library
 AC_DEFUN([AC_PATH_MAGIC],
 [AC_PATH_TOOL_PREFIX(${ac_tool_prefix}file, /usr/bin$PATH_SEPARATOR$PATH)
 if test -z "$lt_cv_path_MAGIC_CMD"; then
@@ -2262,7 +2384,7 @@ esac
 # how to check for library dependencies
 #  -- PORTME fill in with the dynamic library characteristics
 AC_DEFUN([AC_DEPLIBS_CHECK_METHOD],
-[AC_CACHE_CHECK([how to recognise dependent libraries],
+[AC_CACHE_CHECK([how to recognize dependent libraries],
 lt_cv_deplibs_check_method,
 [lt_cv_file_magic_cmd='$MAGIC_CMD'
 lt_cv_file_magic_test_file=
@@ -2279,7 +2401,7 @@ lt_cv_deplibs_check_method='unknown'
 # whether `pass_all' will *always* work, you probably want this one.
 
 case $host_os in
-aix4* | aix5*)
+aix[[4-9]]*)
   lt_cv_deplibs_check_method=pass_all
   ;;
 
@@ -2301,16 +2423,22 @@ cygwin*)
 
 mingw* | pw32*)
   # Base MSYS/MinGW do not provide the 'file' command needed by
-  # func_win32_libid shell function, so use a weaker test based on 'objdump'.
-  lt_cv_deplibs_check_method='file_magic file format pei*-i386(.*architecture: i386)?'
-  lt_cv_file_magic_cmd='$OBJDUMP -f'
+  # func_win32_libid shell function, so use a weaker test based on 'objdump',
+  # unless we find 'file', for example because we are cross-compiling.
+  if ( file / ) >/dev/null 2>&1; then
+    lt_cv_deplibs_check_method='file_magic ^x86 archive import|^x86 DLL'
+    lt_cv_file_magic_cmd='func_win32_libid'
+  else
+    lt_cv_deplibs_check_method='file_magic file format pei*-i386(.*architecture: i386)?'
+    lt_cv_file_magic_cmd='$OBJDUMP -f'
+  fi
   ;;
 
 darwin* | rhapsody*)
   lt_cv_deplibs_check_method=pass_all
   ;;
 
-freebsd* | kfreebsd*-gnu | dragonfly*)
+freebsd* | dragonfly*)
   if echo __ELF__ | $CC -E - | grep __ELF__ > /dev/null; then
     case $host_cpu in
     i*86 )
@@ -2348,7 +2476,7 @@ hpux10.20* | hpux11*)
   esac
   ;;
 
-interix3*)
+interix[[3-9]]*)
   # PIC code is broken on Interix 3.x, that's why |\.a not |_pic\.a here
   lt_cv_deplibs_check_method='match_pattern /lib[[^/]]+(\.so|\.a)$'
   ;;
@@ -2364,7 +2492,7 @@ irix5* | irix6* | nonstopux*)
   ;;
 
 # This must be Linux ELF.
-linux*)
+linux* | k*bsd*-gnu)
   lt_cv_deplibs_check_method=pass_all
   ;;
 
@@ -2398,6 +2526,10 @@ osf3* | osf4* | osf5*)
   lt_cv_deplibs_check_method=pass_all
   ;;
 
+rdos*)
+  lt_cv_deplibs_check_method=pass_all
+  ;;
+
 solaris*)
   lt_cv_deplibs_check_method=pass_all
   ;;
@@ -2450,7 +2582,7 @@ AC_DEFUN([AC_PROG_NM],
   lt_cv_path_NM="$NM"
 else
   lt_nm_to_check="${ac_tool_prefix}nm"
-  if test -n "$ac_tool_prefix" && test "$build" = "$host"; then 
+  if test -n "$ac_tool_prefix" && test "$build" = "$host"; then
     lt_nm_to_check="$lt_nm_to_check nm"
   fi
   for lt_tmp_nm in $lt_nm_to_check; do
@@ -2666,10 +2798,10 @@ objext=o
 _LT_AC_TAGVAR(objext, $1)=$objext
 
 # Code to be used in simple compile tests
-lt_simple_compile_test_code="int some_variable = 0;\n"
+lt_simple_compile_test_code="int some_variable = 0;"
 
 # Code to be used in simple link tests
-lt_simple_link_test_code='int main(){return(0);}\n'
+lt_simple_link_test_code='int main(){return(0);}'
 
 _LT_AC_SYS_COMPILER
 
@@ -2705,7 +2837,7 @@ aix3*)
   fi
   ;;
 
-aix4* | aix5*)
+aix[[4-9]]*)
   if test "$host_cpu" != ia64 && test "$aix_use_runtimelinking" = no ; then
     test "$enable_shared" = yes && enable_static=no
   fi
@@ -2762,6 +2894,7 @@ _LT_AC_TAGVAR(postdep_objects, $1)=
 _LT_AC_TAGVAR(predeps, $1)=
 _LT_AC_TAGVAR(postdeps, $1)=
 _LT_AC_TAGVAR(compiler_lib_search_path, $1)=
+_LT_AC_TAGVAR(compiler_lib_search_dirs, $1)=
 
 # Source file extension for C++ test sources.
 ac_ext=cpp
@@ -2771,10 +2904,10 @@ objext=o
 _LT_AC_TAGVAR(objext, $1)=$objext
 
 # Code to be used in simple compile tests
-lt_simple_compile_test_code="int some_variable = 0;\n"
+lt_simple_compile_test_code="int some_variable = 0;"
 
 # Code to be used in simple link tests
-lt_simple_link_test_code='int main(int, char *[[]]) { return(0); }\n'
+lt_simple_link_test_code='int main(int, char *[[]]) { return(0); }'
 
 # ltmain only uses $CC for tagged configurations so make sure $CC is set.
 _LT_AC_SYS_COMPILER
@@ -2871,7 +3004,7 @@ case $host_os in
     # FIXME: insert proper C++ library support
     _LT_AC_TAGVAR(ld_shlibs, $1)=no
     ;;
-  aix4* | aix5*)
+  aix[[4-9]]*)
     if test "$host_cpu" = ia64; then
       # On IA64, the linker does run time linking by default, so we don't
       # have to do anything special.
@@ -2884,7 +3017,7 @@ case $host_os in
       # Test if we are trying to use run time linking or normal
       # AIX style linking. If -brtl is somewhere in LDFLAGS, we
       # need to do runtime linking.
-      case $host_os in aix4.[[23]]|aix4.[[23]].*|aix5*)
+      case $host_os in aix4.[[23]]|aix4.[[23]].*|aix[[5-9]]*)
 	for ld_flag in $LDFLAGS; do
 	  case $ld_flag in
 	  *-brtl*)
@@ -2920,7 +3053,7 @@ case $host_os in
 	   strings "$collect2name" | grep resolve_lib_name >/dev/null
 	then
 	  # We have reworked collect2
-	  _LT_AC_TAGVAR(hardcode_direct, $1)=yes
+	  :
 	else
 	  # We have old collect2
 	  _LT_AC_TAGVAR(hardcode_direct, $1)=unsupported
@@ -3030,59 +3163,31 @@ case $host_os in
     fi
   ;;
       darwin* | rhapsody*)
-        case $host_os in
-        rhapsody* | darwin1.[[012]])
-         _LT_AC_TAGVAR(allow_undefined_flag, $1)='${wl}-undefined ${wl}suppress'
-         ;;
-       *) # Darwin 1.3 on
-         if test -z ${MACOSX_DEPLOYMENT_TARGET} ; then
-           _LT_AC_TAGVAR(allow_undefined_flag, $1)='${wl}-flat_namespace ${wl}-undefined ${wl}suppress'
-         else
-           case ${MACOSX_DEPLOYMENT_TARGET} in
-             10.[[012]])
-               _LT_AC_TAGVAR(allow_undefined_flag, $1)='${wl}-flat_namespace ${wl}-undefined ${wl}suppress'
-               ;;
-             10.*)
-               _LT_AC_TAGVAR(allow_undefined_flag, $1)='${wl}-undefined ${wl}dynamic_lookup'
-               ;;
-           esac
-         fi
-         ;;
-        esac
       _LT_AC_TAGVAR(archive_cmds_need_lc, $1)=no
       _LT_AC_TAGVAR(hardcode_direct, $1)=no
       _LT_AC_TAGVAR(hardcode_automatic, $1)=yes
       _LT_AC_TAGVAR(hardcode_shlibpath_var, $1)=unsupported
       _LT_AC_TAGVAR(whole_archive_flag_spec, $1)=''
       _LT_AC_TAGVAR(link_all_deplibs, $1)=yes
-
-    if test "$GXX" = yes ; then
-      lt_int_apple_cc_single_mod=no
+      _LT_AC_TAGVAR(allow_undefined_flag, $1)="$_lt_dar_allow_undefined"
+      if test "$GXX" = yes ; then
       output_verbose_link_cmd='echo'
-      if $CC -dumpspecs 2>&1 | $EGREP 'single_module' >/dev/null ; then
-       lt_int_apple_cc_single_mod=yes
+      _LT_AC_TAGVAR(archive_cmds, $1)="\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring $_lt_dar_single_mod${_lt_dsymutil}"
+      _LT_AC_TAGVAR(module_cmds, $1)="\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dsymutil}"
+      _LT_AC_TAGVAR(archive_expsym_cmds, $1)="sed 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring ${_lt_dar_single_mod}${_lt_dar_export_syms}${_lt_dsymutil}"
+      _LT_AC_TAGVAR(module_expsym_cmds, $1)="sed -e 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dar_export_syms}${_lt_dsymutil}"
+      if test "$lt_cv_apple_cc_single_mod" != "yes"; then
+        _LT_AC_TAGVAR(archive_cmds, $1)="\$CC -r -keep_private_externs -nostdlib -o \${lib}-master.o \$libobjs~\$CC -dynamiclib \$allow_undefined_flag -o \$lib \${lib}-master.o \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring${_lt_dsymutil}"
+        _LT_AC_TAGVAR(archive_expsym_cmds, $1)="sed 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC -r -keep_private_externs -nostdlib -o \${lib}-master.o \$libobjs~\$CC -dynamiclib \$allow_undefined_flag -o \$lib \${lib}-master.o \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring${_lt_dar_export_syms}${_lt_dsymutil}"
       fi
-      if test "X$lt_int_apple_cc_single_mod" = Xyes ; then
-       _LT_AC_TAGVAR(archive_cmds, $1)='$CC -dynamiclib -single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring'
-      else
-          _LT_AC_TAGVAR(archive_cmds, $1)='$CC -r -keep_private_externs -nostdlib -o ${lib}-master.o $libobjs~$CC -dynamiclib $allow_undefined_flag -o $lib ${lib}-master.o $deplibs $compiler_flags -install_name $rpath/$soname $verstring'
-        fi
-        _LT_AC_TAGVAR(module_cmds, $1)='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
-        # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-          if test "X$lt_int_apple_cc_single_mod" = Xyes ; then
-            _LT_AC_TAGVAR(archive_expsym_cmds, $1)='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -dynamiclib -single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
-          else
-            _LT_AC_TAGVAR(archive_expsym_cmds, $1)='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -r -keep_private_externs -nostdlib -o ${lib}-master.o $libobjs~$CC -dynamiclib $allow_undefined_flag -o $lib ${lib}-master.o $deplibs $compiler_flags -install_name $rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
-          fi
-            _LT_AC_TAGVAR(module_expsym_cmds, $1)='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
       else
       case $cc_basename in
         xlc*)
          output_verbose_link_cmd='echo'
-          _LT_AC_TAGVAR(archive_cmds, $1)='$CC -qmkshrobj ${wl}-single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $verstring'
+          _LT_AC_TAGVAR(archive_cmds, $1)='$CC -qmkshrobj ${wl}-single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $xlcverstring'
           _LT_AC_TAGVAR(module_cmds, $1)='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
           # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-          _LT_AC_TAGVAR(archive_expsym_cmds, $1)='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj ${wl}-single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
+          _LT_AC_TAGVAR(archive_expsym_cmds, $1)='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj ${wl}-single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $xlcverstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           _LT_AC_TAGVAR(module_expsym_cmds, $1)='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           ;;
        *)
@@ -3116,7 +3221,7 @@ case $host_os in
   freebsd-elf*)
     _LT_AC_TAGVAR(archive_cmds_need_lc, $1)=no
     ;;
-  freebsd* | kfreebsd*-gnu | dragonfly*)
+  freebsd* | dragonfly*)
     # FreeBSD 3 and later use GNU C++ and GNU ld with standard ELF
     # conventions
     _LT_AC_TAGVAR(ld_shlibs, $1)=yes
@@ -3165,9 +3270,7 @@ case $host_os in
       _LT_AC_TAGVAR(hardcode_libdir_separator, $1)=:
 
       case $host_cpu in
-      hppa*64*|ia64*)
-	_LT_AC_TAGVAR(hardcode_libdir_flag_spec_ld, $1)='+b $libdir'
-        ;;
+      hppa*64*|ia64*) ;;
       *)
 	_LT_AC_TAGVAR(export_dynamic_flag_spec, $1)='${wl}-E'
         ;;
@@ -3235,7 +3338,7 @@ case $host_os in
 	;;
     esac
     ;;
-  interix3*)
+  interix[[3-9]]*)
     _LT_AC_TAGVAR(hardcode_direct, $1)=no
     _LT_AC_TAGVAR(hardcode_shlibpath_var, $1)=no
     _LT_AC_TAGVAR(hardcode_libdir_flag_spec, $1)='${wl}-rpath,$libdir'
@@ -3275,7 +3378,7 @@ case $host_os in
     _LT_AC_TAGVAR(hardcode_libdir_flag_spec, $1)='${wl}-rpath ${wl}$libdir'
     _LT_AC_TAGVAR(hardcode_libdir_separator, $1)=:
     ;;
-  linux*)
+  linux* | k*bsd*-gnu)
     case $cc_basename in
       KCC*)
 	# Kuck and Associates, Inc. (KAI) C++ Compiler
@@ -3327,7 +3430,7 @@ case $host_os in
 	_LT_AC_TAGVAR(export_dynamic_flag_spec, $1)='${wl}--export-dynamic'
 	_LT_AC_TAGVAR(whole_archive_flag_spec, $1)='${wl}--whole-archive$convenience ${wl}--no-whole-archive'
 	;;
-      pgCC*)
+      pgCC* | pgcpp*)
         # Portland Group C++ compiler
 	_LT_AC_TAGVAR(archive_cmds, $1)='$CC -shared $pic_flag $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags ${wl}-soname ${wl}$soname -o $lib'
   	_LT_AC_TAGVAR(archive_expsym_cmds, $1)='$CC -shared $pic_flag $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags ${wl}-soname ${wl}$soname ${wl}-retain-symbols-file ${wl}$export_symbols -o $lib'
@@ -3355,6 +3458,29 @@ case $host_os in
 	# dependencies.
 	output_verbose_link_cmd='templist=`$CC -shared $CFLAGS -v conftest.$objext 2>&1 | grep "ld"`; templist=`echo $templist | $SED "s/\(^.*ld.*\)\( .*ld .*$\)/\1/"`; list=""; for z in $templist; do case $z in conftest.$objext) list="$list $z";; *.$objext);; *) list="$list $z";;esac; done; echo $list'
 	;;
+      *)
+	case `$CC -V 2>&1 | sed 5q` in
+	*Sun\ C*)
+	  # Sun C++ 5.9
+	  _LT_AC_TAGVAR(no_undefined_flag, $1)=' -zdefs'
+	  _LT_AC_TAGVAR(archive_cmds, $1)='$CC -G${allow_undefined_flag} -h$soname -o $lib $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags'
+	  _LT_AC_TAGVAR(archive_expsym_cmds, $1)='$CC -G${allow_undefined_flag} -h$soname -o $lib $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags ${wl}-retain-symbols-file ${wl}$export_symbols'
+	  _LT_AC_TAGVAR(hardcode_libdir_flag_spec, $1)='-R$libdir'
+	  _LT_AC_TAGVAR(whole_archive_flag_spec, $1)='${wl}--whole-archive`new_convenience=; for conv in $convenience\"\"; do test -z \"$conv\" || new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}--no-whole-archive'
+
+	  # Not sure whether something based on
+	  # $CC $CFLAGS -v conftest.$objext -o libconftest$shared_ext 2>&1
+	  # would be better.
+	  output_verbose_link_cmd='echo'
+
+	  # Archives containing C++ object files must be created using
+	  # "CC -xar", where "CC" is the Sun C++ compiler.  This is
+	  # necessary to make sure instantiated templates are included
+	  # in the archive.
+	  _LT_AC_TAGVAR(old_archive_cmds, $1)='$CC -xar -o $oldlib $oldobjs'
+	  ;;
+	esac
+	;;
     esac
     ;;
   lynxos*)
@@ -3399,9 +3525,9 @@ case $host_os in
       _LT_AC_TAGVAR(archive_cmds, $1)='$CC -shared $pic_flag $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags -o $lib'
       _LT_AC_TAGVAR(hardcode_libdir_flag_spec, $1)='${wl}-rpath,$libdir'
       if test -z "`echo __ELF__ | $CC -E - | grep __ELF__`" || test "$host_os-$host_cpu" = "openbsd2.8-powerpc"; then
-        _LT_AC_TAGVAR(archive_expsym_cmds, $1)='$CC -shared $pic_flag $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags ${wl}-retain-symbols-file,$export_symbols -o $lib'
-        _LT_AC_TAGVAR(export_dynamic_flag_spec, $1)='${wl}-E'
-        _LT_AC_TAGVAR(whole_archive_flag_spec, $1)="$wlarc"'--whole-archive$convenience '"$wlarc"'--no-whole-archive'
+	_LT_AC_TAGVAR(archive_expsym_cmds, $1)='$CC -shared $pic_flag $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags ${wl}-retain-symbols-file,$export_symbols -o $lib'
+	_LT_AC_TAGVAR(export_dynamic_flag_spec, $1)='${wl}-E'
+	_LT_AC_TAGVAR(whole_archive_flag_spec, $1)="$wlarc"'--whole-archive$convenience '"$wlarc"'--no-whole-archive'
       fi
       output_verbose_link_cmd='echo'
     else
@@ -3568,15 +3694,10 @@ case $host_os in
 	case $host_os in
 	  solaris2.[[0-5]] | solaris2.[[0-5]].*) ;;
 	  *)
-	    # The C++ compiler is used as linker so we must use $wl
-	    # flag to pass the commands to the underlying system
-	    # linker. We must also pass each convience library through
-	    # to the system linker between allextract/defaultextract.
-	    # The C++ compiler will combine linker options so we
-	    # cannot just pass the convience library names through
-	    # without $wl.
+	    # The compiler driver will combine and reorder linker options,
+	    # but understands `-z linker_flag'.
 	    # Supported since Solaris 2.6 (maybe 2.5.1?)
-	    _LT_AC_TAGVAR(whole_archive_flag_spec, $1)='${wl}-z ${wl}allextract`for conv in $convenience\"\"; do test -n \"$conv\" && new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}-z ${wl}defaultextract'
+	    _LT_AC_TAGVAR(whole_archive_flag_spec, $1)='-z allextract$convenience -z defaultextract'
 	    ;;
 	esac
 	_LT_AC_TAGVAR(link_all_deplibs, $1)=yes
@@ -3623,6 +3744,12 @@ case $host_os in
 	  fi
 
 	  _LT_AC_TAGVAR(hardcode_libdir_flag_spec, $1)='${wl}-R $wl$libdir'
+	  case $host_os in
+	  solaris2.[[0-5]] | solaris2.[[0-5]].*) ;;
+	  *)
+	    _LT_AC_TAGVAR(whole_archive_flag_spec, $1)='${wl}-z ${wl}allextract$convenience ${wl}-z ${wl}defaultextract'
+	    ;;
+	  esac
 	fi
 	;;
     esac
@@ -3738,7 +3865,8 @@ lt_cv_prog_gnu_ld=$lt_save_with_gnu_ld
 # compiler output when linking a shared library.
 # Parse the compiler output and extract the necessary
 # objects, libraries and library flags.
-AC_DEFUN([AC_LIBTOOL_POSTDEP_PREDEP],[
+AC_DEFUN([AC_LIBTOOL_POSTDEP_PREDEP],
+[AC_REQUIRE([LT_AC_PROG_SED])dnl
 dnl we can't use the lt_simple_compile_test_code here,
 dnl because it contains code intended for an executable,
 dnl not a library.  It's possible we should let each
@@ -3863,10 +3991,15 @@ fi
 
 $rm -f confest.$objext
 
+_LT_AC_TAGVAR(compiler_lib_search_dirs, $1)=
+if test -n "$_LT_AC_TAGVAR(compiler_lib_search_path, $1)"; then
+  _LT_AC_TAGVAR(compiler_lib_search_dirs, $1)=`echo " ${_LT_AC_TAGVAR(compiler_lib_search_path, $1)}" | ${SED} -e 's! -L! !g' -e 's!^ !!'`
+fi
+
 # PORTME: override above test on systems where it is broken
 ifelse([$1],[CXX],
 [case $host_os in
-interix3*)
+interix[[3-9]]*)
   # Interix 3.5 installs completely hosed .la files for C++, so rather than
   # hack all around it, let's just trust "g++" to DTRT.
   _LT_AC_TAGVAR(predep_objects,$1)=
@@ -3874,19 +4007,51 @@ interix3*)
   _LT_AC_TAGVAR(postdeps,$1)=
   ;;
 
+linux*)
+  case `$CC -V 2>&1 | sed 5q` in
+  *Sun\ C*)
+    # Sun C++ 5.9
+    #
+    # The more standards-conforming stlport4 library is
+    # incompatible with the Cstd library. Avoid specifying
+    # it if it's in CXXFLAGS. Ignore libCrun as
+    # -library=stlport4 depends on it.
+    case " $CXX $CXXFLAGS " in
+    *" -library=stlport4 "*)
+      solaris_use_stlport4=yes
+      ;;
+    esac
+    if test "$solaris_use_stlport4" != yes; then
+      _LT_AC_TAGVAR(postdeps,$1)='-library=Cstd -library=Crun'
+    fi
+    ;;
+  esac
+  ;;
+
 solaris*)
   case $cc_basename in
   CC*)
+    # The more standards-conforming stlport4 library is
+    # incompatible with the Cstd library. Avoid specifying
+    # it if it's in CXXFLAGS. Ignore libCrun as
+    # -library=stlport4 depends on it.
+    case " $CXX $CXXFLAGS " in
+    *" -library=stlport4 "*)
+      solaris_use_stlport4=yes
+      ;;
+    esac
+
     # Adding this requires a known-good setup of shared libraries for
     # Sun compiler versions before 5.6, else PIC objects from an old
     # archive will be linked into the output, leading to subtle bugs.
-    _LT_AC_TAGVAR(postdeps,$1)='-lCstd -lCrun'
+    if test "$solaris_use_stlport4" != yes; then
+      _LT_AC_TAGVAR(postdeps,$1)='-library=Cstd -library=Crun'
+    fi
     ;;
   esac
   ;;
 esac
 ])
-
 case " $_LT_AC_TAGVAR(postdeps, $1) " in
 *" -lc "*) _LT_AC_TAGVAR(archive_cmds_need_lc, $1)=no ;;
 esac
@@ -3929,10 +4094,17 @@ objext=o
 _LT_AC_TAGVAR(objext, $1)=$objext
 
 # Code to be used in simple compile tests
-lt_simple_compile_test_code="      subroutine t\n      return\n      end\n"
+lt_simple_compile_test_code="\
+      subroutine t
+      return
+      end
+"
 
 # Code to be used in simple link tests
-lt_simple_link_test_code="      program t\n      end\n"
+lt_simple_link_test_code="\
+      program t
+      end
+"
 
 # ltmain only uses $CC for tagged configurations so make sure $CC is set.
 _LT_AC_SYS_COMPILER
@@ -3964,7 +4136,7 @@ aix3*)
     postinstall_cmds='$RANLIB $lib'
   fi
   ;;
-aix4* | aix5*)
+aix[[4-9]]*)
   if test "$host_cpu" != ia64 && test "$aix_use_runtimelinking" = no ; then
     test "$enable_shared" = yes && enable_static=no
   fi
@@ -4011,10 +4183,10 @@ objext=o
 _LT_AC_TAGVAR(objext, $1)=$objext
 
 # Code to be used in simple compile tests
-lt_simple_compile_test_code="class foo {}\n"
+lt_simple_compile_test_code="class foo {}"
 
 # Code to be used in simple link tests
-lt_simple_link_test_code='public class conftest { public static void main(String[[]] argv) {}; }\n'
+lt_simple_link_test_code='public class conftest { public static void main(String[[]] argv) {}; }'
 
 # ltmain only uses $CC for tagged configurations so make sure $CC is set.
 _LT_AC_SYS_COMPILER
@@ -4067,7 +4239,7 @@ objext=o
 _LT_AC_TAGVAR(objext, $1)=$objext
 
 # Code to be used in simple compile tests
-lt_simple_compile_test_code='sample MENU { MENUITEM "&Soup", 100, CHECKED }\n'
+lt_simple_compile_test_code='sample MENU { MENUITEM "&Soup", 100, CHECKED }'
 
 # Code to be used in simple link tests
 lt_simple_link_test_code="$lt_simple_compile_test_code"
@@ -4141,6 +4313,7 @@ if test -f "$ltmain"; then
     _LT_AC_TAGVAR(predeps, $1) \
     _LT_AC_TAGVAR(postdeps, $1) \
     _LT_AC_TAGVAR(compiler_lib_search_path, $1) \
+    _LT_AC_TAGVAR(compiler_lib_search_dirs, $1) \
     _LT_AC_TAGVAR(archive_cmds, $1) \
     _LT_AC_TAGVAR(archive_expsym_cmds, $1) \
     _LT_AC_TAGVAR(postinstall_cmds, $1) \
@@ -4156,6 +4329,7 @@ if test -f "$ltmain"; then
     _LT_AC_TAGVAR(module_cmds, $1) \
     _LT_AC_TAGVAR(module_expsym_cmds, $1) \
     _LT_AC_TAGVAR(lt_cv_prog_compiler_c_o, $1) \
+    _LT_AC_TAGVAR(fix_srcfile_path, $1) \
     _LT_AC_TAGVAR(exclude_expsyms, $1) \
     _LT_AC_TAGVAR(include_expsyms, $1); do
 
@@ -4202,7 +4376,7 @@ ifelse([$1], [],
 # Generated automatically by $PROGRAM (GNU $PACKAGE $VERSION$TIMESTAMP)
 # NOTE: Changes made to this file will be lost: look at ltmain.sh.
 #
-# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001
+# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
 # Free Software Foundation, Inc.
 #
 # This file is part of GNU Libtool:
@@ -4439,6 +4613,10 @@ predeps=$lt_[]_LT_AC_TAGVAR(predeps, $1)
 # shared library.
 postdeps=$lt_[]_LT_AC_TAGVAR(postdeps, $1)
 
+# The directories searched by this compiler when creating a shared
+# library
+compiler_lib_search_dirs=$lt_[]_LT_AC_TAGVAR(compiler_lib_search_dirs, $1)
+
 # The library search path used internally by the compiler when linking
 # a shared library.
 compiler_lib_search_path=$lt_[]_LT_AC_TAGVAR(compiler_lib_search_path, $1)
@@ -4527,7 +4705,7 @@ sys_lib_search_path_spec=$lt_sys_lib_search_path_spec
 sys_lib_dlsearch_path_spec=$lt_sys_lib_dlsearch_path_spec
 
 # Fix the shell variable \$srcfile for the compiler.
-fix_srcfile_path="$_LT_AC_TAGVAR(fix_srcfile_path, $1)"
+fix_srcfile_path=$lt_fix_srcfile_path
 
 # Set to yes if exported symbols are required.
 always_export_symbols=$_LT_AC_TAGVAR(always_export_symbols, $1)
@@ -4610,6 +4788,7 @@ fi
 # ---------------------------------
 AC_DEFUN([AC_LIBTOOL_SYS_GLOBAL_SYMBOL_PIPE],
 [AC_REQUIRE([AC_CANONICAL_HOST])
+AC_REQUIRE([LT_AC_PROG_SED])
 AC_REQUIRE([AC_PROG_NM])
 AC_REQUIRE([AC_OBJEXT])
 # Check for command to grab the raw symbol name followed by C symbol from nm.
@@ -4646,7 +4825,7 @@ hpux*) # Its linker distinguishes data from code symbols
   lt_cv_sys_global_symbol_to_cdecl="sed -n -e 's/^T .* \(.*\)$/extern int \1();/p' -e 's/^$symcode* .* \(.*\)$/extern char \1;/p'"
   lt_cv_sys_global_symbol_to_c_name_address="sed -n -e 's/^: \([[^ ]]*\) $/  {\\\"\1\\\", (lt_ptr) 0},/p' -e 's/^$symcode* \([[^ ]]*\) \([[^ ]]*\)$/  {\"\2\", (lt_ptr) \&\2},/p'"
   ;;
-linux*)
+linux* | k*bsd*-gnu)
   if test "$host_cpu" = ia64; then
     symcode='[[ABCDGIRSTW]]'
     lt_cv_sys_global_symbol_to_cdecl="sed -n -e 's/^T .* \(.*\)$/extern int \1();/p' -e 's/^$symcode* .* \(.*\)$/extern char \1;/p'"
@@ -4787,7 +4966,7 @@ EOF
     echo "$progname: failed program was:" >&AS_MESSAGE_LOG_FD
     cat conftest.$ac_ext >&5
   fi
-  rm -f conftest* conftst*
+  rm -rf conftest* conftst*
 
   # Do not use the global_symbol_pipe unless it works.
   if test "$pipe_works" = yes; then
@@ -4836,13 +5015,16 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
       # like `-m68040'.
       _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-m68020 -resident32 -malways-restore-a4'
       ;;
-    beos* | cygwin* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
+    beos* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
       # PIC is the default for these OSes.
       ;;
-    mingw* | os2* | pw32*)
+    mingw* | cygwin* | os2* | pw32*)
       # This hack is so that the source file can tell whether it is being
       # built for inclusion in a dll (and should export symbols for example).
-      _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-DDLL_EXPORT'
+      # Although the cygwin gcc ignores -fPIC, still need this for old-style
+      # (--disable-auto-import) libraries
+      m4_if([$1], [GCJ], [],
+	[_LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-DDLL_EXPORT'])
       ;;
     darwin* | rhapsody*)
       # PIC is the default on this platform
@@ -4853,7 +5035,7 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
       # DJGPP does not support shared libraries at all
       _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)=
       ;;
-    interix3*)
+    interix[[3-9]]*)
       # Interix 3.x gcc -fpic/-fPIC options generate broken code.
       # Instead, we relocate shared libraries at runtime.
       ;;
@@ -4879,7 +5061,7 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
     esac
   else
     case $host_os in
-      aix4* | aix5*)
+      aix[[4-9]]*)
 	# All AIX code is PIC.
 	if test "$host_cpu" = ia64; then
 	  # AIX 5 now supports IA64 processor
@@ -4919,7 +5101,7 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
 	    ;;
 	esac
 	;;
-      freebsd* | kfreebsd*-gnu | dragonfly*)
+      freebsd* | dragonfly*)
 	# FreeBSD uses GNU C++
 	;;
       hpux9* | hpux10* | hpux11*)
@@ -4962,7 +5144,7 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
 	    ;;
 	esac
 	;;
-      linux*)
+      linux* | k*bsd*-gnu)
 	case $cc_basename in
 	  KCC*)
 	    # KAI C++ Compiler
@@ -4975,7 +5157,7 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
 	    _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-KPIC'
 	    _LT_AC_TAGVAR(lt_prog_compiler_static, $1)='-static'
 	    ;;
-	  pgCC*)
+	  pgCC* | pgcpp*)
 	    # Portland Group C++ compiler.
 	    _LT_AC_TAGVAR(lt_prog_compiler_wl, $1)='-Wl,'
 	    _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-fpic'
@@ -4989,6 +5171,14 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
 	    _LT_AC_TAGVAR(lt_prog_compiler_static, $1)='-non_shared'
 	    ;;
 	  *)
+	    case `$CC -V 2>&1 | sed 5q` in
+	    *Sun\ C*)
+	      # Sun C++ 5.9
+	      _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-KPIC'
+	      _LT_AC_TAGVAR(lt_prog_compiler_static, $1)='-Bstatic'
+	      _LT_AC_TAGVAR(lt_prog_compiler_wl, $1)='-Qoption ld '
+	      ;;
+	    esac
 	    ;;
 	esac
 	;;
@@ -5109,14 +5299,17 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
       _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-m68020 -resident32 -malways-restore-a4'
       ;;
 
-    beos* | cygwin* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
+    beos* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
       # PIC is the default for these OSes.
       ;;
 
-    mingw* | pw32* | os2*)
+    mingw* | cygwin* | pw32* | os2*)
       # This hack is so that the source file can tell whether it is being
       # built for inclusion in a dll (and should export symbols for example).
-      _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-DDLL_EXPORT'
+      # Although the cygwin gcc ignores -fPIC, still need this for old-style
+      # (--disable-auto-import) libraries
+      m4_if([$1], [GCJ], [],
+	[_LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-DDLL_EXPORT'])
       ;;
 
     darwin* | rhapsody*)
@@ -5125,7 +5318,7 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
       _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-fno-common'
       ;;
 
-    interix3*)
+    interix[[3-9]]*)
       # Interix 3.x gcc -fpic/-fPIC options generate broken code.
       # Instead, we relocate shared libraries at runtime.
       ;;
@@ -5183,10 +5376,11 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
        esac
        ;;
 
-    mingw* | pw32* | os2*)
+    mingw* | cygwin* | pw32* | os2*)
       # This hack is so that the source file can tell whether it is being
       # built for inclusion in a dll (and should export symbols for example).
-      _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-DDLL_EXPORT'
+      m4_if([$1], [GCJ], [],
+	[_LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-DDLL_EXPORT'])
       ;;
 
     hpux9* | hpux10* | hpux11*)
@@ -5216,7 +5410,7 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
       _LT_AC_TAGVAR(lt_prog_compiler_static, $1)='-Bstatic'
       ;;
 
-    linux*)
+    linux* | k*bsd*-gnu)
       case $cc_basename in
       icc* | ecc*)
 	_LT_AC_TAGVAR(lt_prog_compiler_wl, $1)='-Wl,'
@@ -5235,6 +5429,22 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
         # All Alpha code is PIC.
         _LT_AC_TAGVAR(lt_prog_compiler_static, $1)='-non_shared'
         ;;
+      *)
+        case `$CC -V 2>&1 | sed 5q` in
+	*Sun\ C*)
+	  # Sun C 5.9
+	  _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-KPIC'
+	  _LT_AC_TAGVAR(lt_prog_compiler_static, $1)='-Bstatic'
+	  _LT_AC_TAGVAR(lt_prog_compiler_wl, $1)='-Wl,'
+	  ;;
+	*Sun\ F*)
+	  # Sun Fortran 8.3 passes all unrecognized flags to the linker
+	  _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-KPIC'
+	  _LT_AC_TAGVAR(lt_prog_compiler_static, $1)='-Bstatic'
+	  _LT_AC_TAGVAR(lt_prog_compiler_wl, $1)=''
+	  ;;
+	esac
+	;;
       esac
       ;;
 
@@ -5244,6 +5454,10 @@ AC_MSG_CHECKING([for $compiler option to produce PIC])
       _LT_AC_TAGVAR(lt_prog_compiler_static, $1)='-non_shared'
       ;;
 
+    rdos*)
+      _LT_AC_TAGVAR(lt_prog_compiler_static, $1)='-non_shared'
+      ;;
+
     solaris*)
       _LT_AC_TAGVAR(lt_prog_compiler_pic, $1)='-KPIC'
       _LT_AC_TAGVAR(lt_prog_compiler_static, $1)='-Bstatic'
@@ -5303,7 +5517,7 @@ AC_MSG_RESULT([$_LT_AC_TAGVAR(lt_prog_compiler_pic, $1)])
 #
 if test -n "$_LT_AC_TAGVAR(lt_prog_compiler_pic, $1)"; then
   AC_LIBTOOL_COMPILER_OPTION([if $compiler PIC flag $_LT_AC_TAGVAR(lt_prog_compiler_pic, $1) works],
-    _LT_AC_TAGVAR(lt_prog_compiler_pic_works, $1),
+    _LT_AC_TAGVAR(lt_cv_prog_compiler_pic_works, $1),
     [$_LT_AC_TAGVAR(lt_prog_compiler_pic, $1)ifelse([$1],[],[ -DPIC],[ifelse([$1],[CXX],[ -DPIC],[])])], [],
     [case $_LT_AC_TAGVAR(lt_prog_compiler_pic, $1) in
      "" | " "*) ;;
@@ -5327,7 +5541,7 @@ esac
 #
 wl=$_LT_AC_TAGVAR(lt_prog_compiler_wl, $1) eval lt_tmp_static_flag=\"$_LT_AC_TAGVAR(lt_prog_compiler_static, $1)\"
 AC_LIBTOOL_LINKER_OPTION([if $compiler static flag $lt_tmp_static_flag works],
-  _LT_AC_TAGVAR(lt_prog_compiler_static_works, $1),
+  _LT_AC_TAGVAR(lt_cv_prog_compiler_static_works, $1),
   $lt_tmp_static_flag,
   [],
   [_LT_AC_TAGVAR(lt_prog_compiler_static, $1)=])
@@ -5338,11 +5552,12 @@ AC_LIBTOOL_LINKER_OPTION([if $compiler static flag $lt_tmp_static_flag works],
 # ------------------------------------
 # See if the linker supports building shared libraries.
 AC_DEFUN([AC_LIBTOOL_PROG_LD_SHLIBS],
-[AC_MSG_CHECKING([whether the $compiler linker ($LD) supports shared libraries])
+[AC_REQUIRE([LT_AC_PROG_SED])dnl
+AC_MSG_CHECKING([whether the $compiler linker ($LD) supports shared libraries])
 ifelse([$1],[CXX],[
   _LT_AC_TAGVAR(export_symbols_cmds, $1)='$NM $libobjs $convenience | $global_symbol_pipe | $SED '\''s/.* //'\'' | sort | uniq > $export_symbols'
   case $host_os in
-  aix4* | aix5*)
+  aix[[4-9]]*)
     # If we're using GNU nm, then we don't want the "-C" option.
     # -C means demangle to AIX nm, but means don't demangle with GNU nm
     if $NM -V 2>&1 | grep 'GNU' > /dev/null; then
@@ -5355,12 +5570,13 @@ ifelse([$1],[CXX],[
     _LT_AC_TAGVAR(export_symbols_cmds, $1)="$ltdll_cmds"
   ;;
   cygwin* | mingw*)
-    _LT_AC_TAGVAR(export_symbols_cmds, $1)='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[[BCDGRS]] /s/.* \([[^ ]]*\)/\1 DATA/;/^.* __nm__/s/^.* __nm__\([[^ ]]*\) [[^ ]]*/\1 DATA/;/^I /d;/^[[AITW]] /s/.* //'\'' | sort | uniq > $export_symbols'
+    _LT_AC_TAGVAR(export_symbols_cmds, $1)='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[[BCDGRS]][[ ]]/s/.*[[ ]]\([[^ ]]*\)/\1 DATA/;/^.*[[ ]]__nm__/s/^.*[[ ]]__nm__\([[^ ]]*\)[[ ]][[^ ]]*/\1 DATA/;/^I[[ ]]/d;/^[[AITW]][[ ]]/s/.*[[ ]]//'\'' | sort | uniq > $export_symbols'
   ;;
   *)
     _LT_AC_TAGVAR(export_symbols_cmds, $1)='$NM $libobjs $convenience | $global_symbol_pipe | $SED '\''s/.* //'\'' | sort | uniq > $export_symbols'
   ;;
   esac
+  _LT_AC_TAGVAR(exclude_expsyms, $1)=['_GLOBAL_OFFSET_TABLE_|_GLOBAL__F[ID]_.*']
 ],[
   runpath_var=
   _LT_AC_TAGVAR(allow_undefined_flag, $1)=
@@ -5391,12 +5607,14 @@ ifelse([$1],[CXX],[
   # it will be wrapped by ` (' and `)$', so one must not match beginning or
   # end of line.  Example: `a|bc|.*d.*' will exclude the symbols `a' and `bc',
   # as well as any symbol that contains `d'.
-  _LT_AC_TAGVAR(exclude_expsyms, $1)="_GLOBAL_OFFSET_TABLE_"
+  _LT_AC_TAGVAR(exclude_expsyms, $1)=['_GLOBAL_OFFSET_TABLE_|_GLOBAL__F[ID]_.*']
   # Although _GLOBAL_OFFSET_TABLE_ is a valid symbol C name, most a.out
   # platforms (ab)use it in PIC code, but their linkers get confused if
   # the symbol is explicitly referenced.  Since portable code cannot
   # rely on this symbol name, it's probably fine to never include it in
   # preloaded symbol tables.
+  # Exclude shared library initialization/finalization symbols.
+dnl Note also adjust exclude_expsyms for C++ above.
   extract_expsyms_cmds=
   # Just being paranoid about ensuring that cc_basename is set.
   _LT_CC_BASENAME([$compiler])
@@ -5446,7 +5664,7 @@ ifelse([$1],[CXX],[
 
     # See if GNU ld supports shared libraries.
     case $host_os in
-    aix3* | aix4* | aix5*)
+    aix[[3-9]]*)
       # On AIX/PPC, the GNU linker is very broken
       if test "$host_cpu" != ia64; then
 	_LT_AC_TAGVAR(ld_shlibs, $1)=no
@@ -5494,7 +5712,7 @@ EOF
       _LT_AC_TAGVAR(allow_undefined_flag, $1)=unsupported
       _LT_AC_TAGVAR(always_export_symbols, $1)=no
       _LT_AC_TAGVAR(enable_shared_with_static_runtimes, $1)=yes
-      _LT_AC_TAGVAR(export_symbols_cmds, $1)='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[[BCDGRS]] /s/.* \([[^ ]]*\)/\1 DATA/'\'' | $SED -e '\''/^[[AITW]] /s/.* //'\'' | sort | uniq > $export_symbols'
+      _LT_AC_TAGVAR(export_symbols_cmds, $1)='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[[BCDGRS]][[ ]]/s/.*[[ ]]\([[^ ]]*\)/\1 DATA/'\'' -e '\''/^[[AITW]][[ ]]/s/.*[[ ]]//'\'' | sort | uniq > $export_symbols'
 
       if $LD --help 2>&1 | grep 'auto-import' > /dev/null; then
         _LT_AC_TAGVAR(archive_cmds, $1)='$CC -shared $libobjs $deplibs $compiler_flags -o $output_objdir/$soname ${wl}--enable-auto-image-base -Xlinker --out-implib -Xlinker $lib'
@@ -5512,7 +5730,7 @@ EOF
       fi
       ;;
 
-    interix3*)
+    interix[[3-9]]*)
       _LT_AC_TAGVAR(hardcode_direct, $1)=no
       _LT_AC_TAGVAR(hardcode_shlibpath_var, $1)=no
       _LT_AC_TAGVAR(hardcode_libdir_flag_spec, $1)='${wl}-rpath,$libdir'
@@ -5527,7 +5745,7 @@ EOF
       _LT_AC_TAGVAR(archive_expsym_cmds, $1)='sed "s,^,_," $export_symbols >$output_objdir/$soname.expsym~$CC -shared $pic_flag $libobjs $deplibs $compiler_flags ${wl}-h,$soname ${wl}--retain-symbols-file,$output_objdir/$soname.expsym ${wl}--image-base,`expr ${RANDOM-$$} % 4096 / 2 \* 262144 + 1342177280` -o $lib'
       ;;
 
-    linux*)
+    gnu* | linux* | k*bsd*-gnu)
       if $LD --help 2>&1 | grep ': supported targets:.* elf' > /dev/null; then
 	tmp_addflag=
 	case $cc_basename,$host_cpu in
@@ -5545,13 +5763,22 @@ EOF
 	ifc* | ifort*)			# Intel Fortran compiler
 	  tmp_addflag=' -nofor_main' ;;
 	esac
-	_LT_AC_TAGVAR(archive_cmds, $1)='$CC -shared'"$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname -o $lib'
+	case `$CC -V 2>&1 | sed 5q` in
+	*Sun\ C*)			# Sun C 5.9
+	  _LT_AC_TAGVAR(whole_archive_flag_spec, $1)='${wl}--whole-archive`new_convenience=; for conv in $convenience\"\"; do test -z \"$conv\" || new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}--no-whole-archive'
+	  tmp_sharedflag='-G' ;;
+	*Sun\ F*)			# Sun Fortran 8.3
+	  tmp_sharedflag='-G' ;;
+	*)
+	  tmp_sharedflag='-shared' ;;
+	esac
+	_LT_AC_TAGVAR(archive_cmds, $1)='$CC '"$tmp_sharedflag""$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname -o $lib'
 
 	if test $supports_anon_versioning = yes; then
 	  _LT_AC_TAGVAR(archive_expsym_cmds, $1)='$echo "{ global:" > $output_objdir/$libname.ver~
   cat $export_symbols | sed -e "s/\(.*\)/\1;/" >> $output_objdir/$libname.ver~
   $echo "local: *; };" >> $output_objdir/$libname.ver~
-	  $CC -shared'"$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname ${wl}-version-script ${wl}$output_objdir/$libname.ver -o $lib'
+	  $CC '"$tmp_sharedflag""$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname ${wl}-version-script ${wl}$output_objdir/$libname.ver -o $lib'
 	fi
       else
 	_LT_AC_TAGVAR(ld_shlibs, $1)=no
@@ -5591,7 +5818,7 @@ EOF
 
     sysv5* | sco3.2v5* | sco5v6* | unixware* | OpenUNIX*)
       case `$LD -v 2>&1` in
-        *\ [[01]].* | *\ 2.[[0-9]].* | *\ 2.1[[0-5]].*) 
+        *\ [[01]].* | *\ 2.[[0-9]].* | *\ 2.1[[0-5]].*)
 	_LT_AC_TAGVAR(ld_shlibs, $1)=no
 	cat <<_LT_EOF 1>&2
 
@@ -5656,7 +5883,7 @@ _LT_EOF
       fi
       ;;
 
-    aix4* | aix5*)
+    aix[[4-9]]*)
       if test "$host_cpu" = ia64; then
 	# On IA64, the linker does run time linking by default, so we don't
 	# have to do anything special.
@@ -5676,7 +5903,7 @@ _LT_EOF
 	# Test if we are trying to use run time linking or normal
 	# AIX style linking. If -brtl is somewhere in LDFLAGS, we
 	# need to do runtime linking.
-	case $host_os in aix4.[[23]]|aix4.[[23]].*|aix5*)
+	case $host_os in aix4.[[23]]|aix4.[[23]].*|aix[[5-9]]*)
 	  for ld_flag in $LDFLAGS; do
   	  if (test $ld_flag = "-brtl" || test $ld_flag = "-Wl,-brtl"); then
   	    aix_use_runtimelinking=yes
@@ -5710,7 +5937,7 @@ _LT_EOF
   	   strings "$collect2name" | grep resolve_lib_name >/dev/null
 	  then
   	  # We have reworked collect2
-  	  _LT_AC_TAGVAR(hardcode_direct, $1)=yes
+  	  :
 	  else
   	  # We have old collect2
   	  _LT_AC_TAGVAR(hardcode_direct, $1)=unsupported
@@ -5803,7 +6030,7 @@ _LT_EOF
       # The linker will automatically build a .lib file if we build a DLL.
       _LT_AC_TAGVAR(old_archive_From_new_cmds, $1)='true'
       # FIXME: Should let the user specify the lib program.
-      _LT_AC_TAGVAR(old_archive_cmds, $1)='lib /OUT:$oldlib$oldobjs$old_deplibs'
+      _LT_AC_TAGVAR(old_archive_cmds, $1)='lib -OUT:$oldlib$oldobjs$old_deplibs'
       _LT_AC_TAGVAR(fix_srcfile_path, $1)='`cygpath -w "$srcfile"`'
       _LT_AC_TAGVAR(enable_shared_with_static_runtimes, $1)=yes
       ;;
@@ -5836,19 +6063,18 @@ _LT_EOF
       _LT_AC_TAGVAR(link_all_deplibs, $1)=yes
     if test "$GCC" = yes ; then
     	output_verbose_link_cmd='echo'
-        _LT_AC_TAGVAR(archive_cmds, $1)='$CC -dynamiclib $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring'
-      _LT_AC_TAGVAR(module_cmds, $1)='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
-      # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-      _LT_AC_TAGVAR(archive_expsym_cmds, $1)='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -dynamiclib $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
-      _LT_AC_TAGVAR(module_expsym_cmds, $1)='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
+        _LT_AC_TAGVAR(archive_cmds, $1)="\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring $_lt_dar_single_mod${_lt_dsymutil}"
+        _LT_AC_TAGVAR(module_cmds, $1)="\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dsymutil}"
+        _LT_AC_TAGVAR(archive_expsym_cmds, $1)="sed 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring ${_lt_dar_single_mod}${_lt_dar_export_syms}${_lt_dsymutil}"
+        _LT_AC_TAGVAR(module_expsym_cmds, $1)="sed -e 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dar_export_syms}${_lt_dsymutil}"
     else
       case $cc_basename in
         xlc*)
          output_verbose_link_cmd='echo'
-         _LT_AC_TAGVAR(archive_cmds, $1)='$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $verstring'
+         _LT_AC_TAGVAR(archive_cmds, $1)='$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $xlcverstring'
          _LT_AC_TAGVAR(module_cmds, $1)='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
           # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-         _LT_AC_TAGVAR(archive_expsym_cmds, $1)='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
+         _LT_AC_TAGVAR(archive_expsym_cmds, $1)='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $xlcverstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           _LT_AC_TAGVAR(module_expsym_cmds, $1)='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           ;;
        *)
@@ -5888,7 +6114,7 @@ _LT_EOF
       ;;
 
     # FreeBSD 3 and greater uses gcc -shared to do shared libraries.
-    freebsd* | kfreebsd*-gnu | dragonfly*)
+    freebsd* | dragonfly*)
       _LT_AC_TAGVAR(archive_cmds, $1)='$CC -shared -o $lib $libobjs $deplibs $compiler_flags'
       _LT_AC_TAGVAR(hardcode_libdir_flag_spec, $1)='-R$libdir'
       _LT_AC_TAGVAR(hardcode_direct, $1)=yes
@@ -6011,15 +6237,15 @@ _LT_EOF
 
     openbsd*)
       if test -f /usr/libexec/ld.so; then
-        _LT_AC_TAGVAR(hardcode_direct, $1)=yes
-        _LT_AC_TAGVAR(hardcode_shlibpath_var, $1)=no
-        if test -z "`echo __ELF__ | $CC -E - | grep __ELF__`" || test "$host_os-$host_cpu" = "openbsd2.8-powerpc"; then
+	_LT_AC_TAGVAR(hardcode_direct, $1)=yes
+	_LT_AC_TAGVAR(hardcode_shlibpath_var, $1)=no
+	if test -z "`echo __ELF__ | $CC -E - | grep __ELF__`" || test "$host_os-$host_cpu" = "openbsd2.8-powerpc"; then
 	  _LT_AC_TAGVAR(archive_cmds, $1)='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags'
 	  _LT_AC_TAGVAR(archive_expsym_cmds, $1)='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-retain-symbols-file,$export_symbols'
 	  _LT_AC_TAGVAR(hardcode_libdir_flag_spec, $1)='${wl}-rpath,$libdir'
 	  _LT_AC_TAGVAR(export_dynamic_flag_spec, $1)='${wl}-E'
-        else
-         case $host_os in
+	else
+	  case $host_os in
 	   openbsd[[01]].* | openbsd2.[[0-7]] | openbsd2.[[0-7]].*)
 	     _LT_AC_TAGVAR(archive_cmds, $1)='$LD -Bshareable -o $lib $libobjs $deplibs $linker_flags'
 	     _LT_AC_TAGVAR(hardcode_libdir_flag_spec, $1)='-R$libdir'
@@ -6028,10 +6254,10 @@ _LT_EOF
 	     _LT_AC_TAGVAR(archive_cmds, $1)='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags'
 	     _LT_AC_TAGVAR(hardcode_libdir_flag_spec, $1)='${wl}-rpath,$libdir'
 	     ;;
-         esac
+	  esac
         fi
       else
-        _LT_AC_TAGVAR(ld_shlibs, $1)=no
+	_LT_AC_TAGVAR(ld_shlibs, $1)=no
       fi
       ;;
 
@@ -6090,17 +6316,16 @@ _LT_EOF
       case $host_os in
       solaris2.[[0-5]] | solaris2.[[0-5]].*) ;;
       *)
- 	# The compiler driver will combine linker options so we
- 	# cannot just pass the convience library names through
- 	# without $wl, iff we do not link with $LD.
- 	# Luckily, gcc supports the same syntax we need for Sun Studio.
+	# The compiler driver will combine and reorder linker options,
+	# but understands `-z linker_flag'.  GCC discards it without `$wl',
+	# but is careful enough not to reorder.
  	# Supported since Solaris 2.6 (maybe 2.5.1?)
- 	case $wlarc in
- 	'')
- 	  _LT_AC_TAGVAR(whole_archive_flag_spec, $1)='-z allextract$convenience -z defaultextract' ;;
- 	*)
- 	  _LT_AC_TAGVAR(whole_archive_flag_spec, $1)='${wl}-z ${wl}allextract`for conv in $convenience\"\"; do test -n \"$conv\" && new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}-z ${wl}defaultextract' ;;
- 	esac ;;
+	if test "$GCC" = yes; then
+	  _LT_AC_TAGVAR(whole_archive_flag_spec, $1)='${wl}-z ${wl}allextract$convenience ${wl}-z ${wl}defaultextract'
+	else
+	  _LT_AC_TAGVAR(whole_archive_flag_spec, $1)='-z allextract$convenience -z defaultextract'
+	fi
+	;;
       esac
       _LT_AC_TAGVAR(link_all_deplibs, $1)=yes
       ;;
@@ -6157,7 +6382,7 @@ _LT_EOF
       fi
       ;;
 
-    sysv4*uw2* | sysv5OpenUNIX* | sysv5UnixWare7.[[01]].[[10]]* | unixware7*)
+    sysv4*uw2* | sysv5OpenUNIX* | sysv5UnixWare7.[[01]].[[10]]* | unixware7* | sco3.2v5.0.[[024]]*)
       _LT_AC_TAGVAR(no_undefined_flag, $1)='${wl}-z,text'
       _LT_AC_TAGVAR(archive_cmds_need_lc, $1)=no
       _LT_AC_TAGVAR(hardcode_shlibpath_var, $1)=no
@@ -6232,7 +6457,7 @@ x|xyes)
       # to ld, don't add -lc before -lgcc.
       AC_MSG_CHECKING([whether -lc should be explicitly linked in])
       $rm conftest*
-      printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+      echo "$lt_simple_compile_test_code" > conftest.$ac_ext
 
       if AC_TRY_EVAL(ac_compile) 2>conftest.err; then
         soname=conftest
@@ -6335,6 +6560,30 @@ AC_DEFUN([LT_AC_PROG_RC],
 [AC_CHECK_TOOL(RC, windres, no)
 ])
 
+
+# Cheap backport of AS_EXECUTABLE_P and required macros
+# from Autoconf 2.59; we should not use $as_executable_p directly.
+
+# _AS_TEST_PREPARE
+# ----------------
+m4_ifndef([_AS_TEST_PREPARE],
+[m4_defun([_AS_TEST_PREPARE],
+[if test -x / >/dev/null 2>&1; then
+  as_executable_p='test -x'
+else
+  as_executable_p='test -f'
+fi
+])])# _AS_TEST_PREPARE
+
+# AS_EXECUTABLE_P
+# ---------------
+# Check whether a file is executable.
+m4_ifndef([AS_EXECUTABLE_P],
+[m4_defun([AS_EXECUTABLE_P],
+[AS_REQUIRE([_AS_TEST_PREPARE])dnl
+$as_executable_p $1[]dnl
+])])# AS_EXECUTABLE_P
+
 # NOTE: This macro has been submitted for inclusion into   #
 #  GNU Autoconf as AC_PROG_SED.  When it is available in   #
 #  a released version of Autoconf we should remove this    #
@@ -6355,12 +6604,13 @@ do
   test -z "$as_dir" && as_dir=.
   for lt_ac_prog in sed gsed; do
     for ac_exec_ext in '' $ac_executable_extensions; do
-      if $as_executable_p "$as_dir/$lt_ac_prog$ac_exec_ext"; then
+      if AS_EXECUTABLE_P(["$as_dir/$lt_ac_prog$ac_exec_ext"]); then
         lt_ac_sed_list="$lt_ac_sed_list $as_dir/$lt_ac_prog$ac_exec_ext"
       fi
     done
   done
 done
+IFS=$as_save_IFS
 lt_ac_max=0
 lt_ac_count=0
 # Add /usr/xpg4/bin/sed as it is typically found on Solaris
@@ -6393,6 +6643,7 @@ for lt_ac_sed in $lt_ac_sed_list /usr/xpg4/bin/sed; do
 done
 ])
 SED=$lt_cv_path_SED
+AC_SUBST([SED])
 AC_MSG_RESULT([$SED])
 ])
 
diff --git a/lib/pixman/config.h.in b/lib/pixman/config.h.in
index cdcf61ba3..64de572be 100644
--- a/lib/pixman/config.h.in
+++ b/lib/pixman/config.h.in
@@ -57,15 +57,21 @@
 /* use MMX compiler intrinsics */
 #undef USE_MMX
 
-/* use SSE compiler intrinsics */
-#undef USE_SSE
-
-/* use SSE compiler intrinsics */
+/* use SSE2 compiler intrinsics */
 #undef USE_SSE2
 
+/* use VMX compiler intrinsics */
+#undef USE_VMX
+
 /* Version number of package */
 #undef VERSION
 
 /* Define to 1 if your processor stores words with the most significant byte
    first (like Motorola and SPARC, unlike Intel and VAX). */
 #undef WORDS_BIGENDIAN
+
+/* Define to `__inline__' or `__inline' if that's what the C compiler
+   calls it, or to nothing if 'inline' is not supported under any name.  */
+#ifndef __cplusplus
+#undef inline
+#endif
diff --git a/lib/pixman/configure b/lib/pixman/configure
index 1535e84a0..462a1934c 100644
--- a/lib/pixman/configure
+++ b/lib/pixman/configure
@@ -1,6 +1,6 @@
 #! /bin/sh
 # Guess values for system-dependent variables and create Makefiles.
-# Generated by GNU Autoconf 2.59 for pixman 0.10.0.
+# Generated by GNU Autoconf 2.59 for pixman 0.12.0.
 #
 # Report bugs to <"sandmann@daimi.au.dk">.
 #
@@ -423,8 +423,8 @@ SHELL=${CONFIG_SHELL-/bin/sh}
 # Identity of this package.
 PACKAGE_NAME='pixman'
 PACKAGE_TARNAME='pixman'
-PACKAGE_VERSION='0.10.0'
-PACKAGE_STRING='pixman 0.10.0'
+PACKAGE_VERSION='0.12.0'
+PACKAGE_STRING='pixman 0.12.0'
 PACKAGE_BUGREPORT='"sandmann@daimi.au.dk"'
 
 # Factoring default headers for most tests.
@@ -464,7 +464,7 @@ ac_includes_default="\
 # include <unistd.h>
 #endif"
 
-ac_subst_vars='SHELL PATH_SEPARATOR PACKAGE_NAME PACKAGE_TARNAME PACKAGE_VERSION PACKAGE_STRING PACKAGE_BUGREPORT exec_prefix prefix program_transform_name bindir sbindir libexecdir datadir sysconfdir sharedstatedir localstatedir libdir includedir oldincludedir infodir mandir build_alias host_alias target_alias DEFS ECHO_C ECHO_N ECHO_T LIBS INSTALL_PROGRAM INSTALL_SCRIPT INSTALL_DATA CYGPATH_W PACKAGE VERSION ACLOCAL AUTOCONF AUTOMAKE AUTOHEADER MAKEINFO install_sh STRIP ac_ct_STRIP INSTALL_STRIP_PROGRAM mkdir_p AWK SET_MAKE am__leading_dot AMTAR am__tar am__untar CC CFLAGS LDFLAGS CPPFLAGS ac_ct_CC EXEEXT OBJEXT DEPDIR am__include am__quote AMDEP_TRUE AMDEP_FALSE AMDEPBACKSLASH CCDEPMODE am__fastdepCC_TRUE am__fastdepCC_FALSE build build_cpu build_vendor build_os host host_cpu host_vendor host_os EGREP LN_S ECHO AR ac_ct_AR RANLIB ac_ct_RANLIB CPP CXX CXXFLAGS ac_ct_CXX CXXDEPMODE am__fastdepCXX_TRUE am__fastdepCXX_FALSE CXXCPP F77 FFLAGS ac_ct_F77 LIBTOOL PIXMAN_VERSION_MAJOR PIXMAN_VERSION_MINOR PIXMAN_VERSION_MICRO LT_VERSION_INFO USE_MMX_TRUE USE_MMX_FALSE USE_SSE_TRUE USE_SSE_FALSE USE_SSE2_TRUE USE_SSE2_FALSE MMX_CFLAGS SSE_CFLAGS PKG_CONFIG ac_pt_PKG_CONFIG GTK_CFLAGS GTK_LIBS HAVE_GTK_TRUE HAVE_GTK_FALSE DEP_CFLAGS DEP_LIBS LIBOBJS LTLIBOBJS'
+ac_subst_vars='SHELL PATH_SEPARATOR PACKAGE_NAME PACKAGE_TARNAME PACKAGE_VERSION PACKAGE_STRING PACKAGE_BUGREPORT exec_prefix prefix program_transform_name bindir sbindir libexecdir datadir sysconfdir sharedstatedir localstatedir libdir includedir oldincludedir infodir mandir build_alias host_alias target_alias DEFS ECHO_C ECHO_N ECHO_T LIBS INSTALL_PROGRAM INSTALL_SCRIPT INSTALL_DATA CYGPATH_W PACKAGE VERSION ACLOCAL AUTOCONF AUTOMAKE AUTOHEADER MAKEINFO install_sh STRIP ac_ct_STRIP INSTALL_STRIP_PROGRAM mkdir_p AWK SET_MAKE am__leading_dot AMTAR am__tar am__untar CC CFLAGS LDFLAGS CPPFLAGS ac_ct_CC EXEEXT OBJEXT DEPDIR am__include am__quote AMDEP_TRUE AMDEP_FALSE AMDEPBACKSLASH CCDEPMODE am__fastdepCC_TRUE am__fastdepCC_FALSE build build_cpu build_vendor build_os host host_cpu host_vendor host_os SED EGREP LN_S ECHO AR ac_ct_AR RANLIB ac_ct_RANLIB DSYMUTIL ac_ct_DSYMUTIL NMEDIT ac_ct_NMEDIT CPP CXX CXXFLAGS ac_ct_CXX CXXDEPMODE am__fastdepCXX_TRUE am__fastdepCXX_FALSE CXXCPP F77 FFLAGS ac_ct_F77 LIBTOOL PIXMAN_VERSION_MAJOR PIXMAN_VERSION_MINOR PIXMAN_VERSION_MICRO LT_VERSION_INFO PERL USE_MMX_TRUE USE_MMX_FALSE USE_SSE2_TRUE USE_SSE2_FALSE MMX_CFLAGS SSE2_CFLAGS VMX_CFLAGS USE_VMX_TRUE USE_VMX_FALSE PKG_CONFIG ac_pt_PKG_CONFIG GTK_CFLAGS GTK_LIBS HAVE_GTK_TRUE HAVE_GTK_FALSE DEP_CFLAGS DEP_LIBS LIBOBJS LTLIBOBJS'
 ac_subst_files=''
 
 # Initialize some variables set by options.
@@ -965,7 +965,7 @@ if test "$ac_init_help" = "long"; then
   # Omit some internal or obsolete options to make the list less imposing.
   # This message is too long to be a string in the A/UX 3.1 sh.
   cat <<_ACEOF
-\`configure' configures pixman 0.10.0 to adapt to many kinds of systems.
+\`configure' configures pixman 0.12.0 to adapt to many kinds of systems.
 
 Usage: $0 [OPTION]... [VAR=VALUE]...
 
@@ -1031,7 +1031,7 @@ fi
 
 if test -n "$ac_init_help"; then
   case $ac_init_help in
-     short | recursive ) echo "Configuration of pixman 0.10.0:";;
+     short | recursive ) echo "Configuration of pixman 0.12.0:";;
    esac
   cat <<\_ACEOF
 
@@ -1047,6 +1047,10 @@ Optional Features:
   --enable-fast-install[=PKGS]
                           optimize for fast installation [default=yes]
   --disable-libtool-lock  avoid locking (might break parallel builds)
+  --disable-mmx           disable MMX fast paths
+  --disable-sse2          disable SSE2 fast paths
+  --disable-vmx           disable VMX fast paths
+  --enable-gtk            enable tests using GTK+ [default=auto]
 
 Optional Packages:
   --with-PACKAGE[=ARG]    use PACKAGE [ARG=yes]
@@ -1173,7 +1177,7 @@ fi
 test -n "$ac_init_help" && exit 0
 if $ac_init_version; then
   cat <<\_ACEOF
-pixman configure 0.10.0
+pixman configure 0.12.0
 generated by GNU Autoconf 2.59
 
 Copyright (C) 2003 Free Software Foundation, Inc.
@@ -1187,7 +1191,7 @@ cat >&5 <<_ACEOF
 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.
 
-It was created by pixman $as_me 0.10.0, which was
+It was created by pixman $as_me 0.12.0, which was
 generated by GNU Autoconf 2.59.  Invocation command line was
 
   $ $0 $@
@@ -1785,7 +1789,7 @@ fi
 
 # Define the identity of the package.
  PACKAGE='pixman'
- VERSION='0.10.0'
+ VERSION='0.12.0'
 
 
 cat >>confdefs.h <<_ACEOF
@@ -3169,6 +3173,7 @@ do
     done
   done
 done
+IFS=$as_save_IFS
 lt_ac_max=0
 lt_ac_count=0
 # Add /usr/xpg4/bin/sed as it is typically found on Solaris
@@ -3203,6 +3208,7 @@ done
 fi
 
 SED=$lt_cv_path_SED
+
 echo "$as_me:$LINENO: result: $SED" >&5
 echo "${ECHO_T}$SED" >&6
 
@@ -3417,8 +3423,8 @@ else
 echo "${ECHO_T}no, using $LN_S" >&6
 fi
 
-echo "$as_me:$LINENO: checking how to recognise dependent libraries" >&5
-echo $ECHO_N "checking how to recognise dependent libraries... $ECHO_C" >&6
+echo "$as_me:$LINENO: checking how to recognize dependent libraries" >&5
+echo $ECHO_N "checking how to recognize dependent libraries... $ECHO_C" >&6
 if test "${lt_cv_deplibs_check_method+set}" = set; then
   echo $ECHO_N "(cached) $ECHO_C" >&6
 else
@@ -3437,7 +3443,7 @@ lt_cv_deplibs_check_method='unknown'
 # whether `pass_all' will *always* work, you probably want this one.
 
 case $host_os in
-aix4* | aix5*)
+aix[4-9]*)
   lt_cv_deplibs_check_method=pass_all
   ;;
 
@@ -3459,16 +3465,22 @@ cygwin*)
 
 mingw* | pw32*)
   # Base MSYS/MinGW do not provide the 'file' command needed by
-  # func_win32_libid shell function, so use a weaker test based on 'objdump'.
-  lt_cv_deplibs_check_method='file_magic file format pei*-i386(.*architecture: i386)?'
-  lt_cv_file_magic_cmd='$OBJDUMP -f'
+  # func_win32_libid shell function, so use a weaker test based on 'objdump',
+  # unless we find 'file', for example because we are cross-compiling.
+  if ( file / ) >/dev/null 2>&1; then
+    lt_cv_deplibs_check_method='file_magic ^x86 archive import|^x86 DLL'
+    lt_cv_file_magic_cmd='func_win32_libid'
+  else
+    lt_cv_deplibs_check_method='file_magic file format pei*-i386(.*architecture: i386)?'
+    lt_cv_file_magic_cmd='$OBJDUMP -f'
+  fi
   ;;
 
 darwin* | rhapsody*)
   lt_cv_deplibs_check_method=pass_all
   ;;
 
-freebsd* | kfreebsd*-gnu | dragonfly*)
+freebsd* | dragonfly*)
   if echo __ELF__ | $CC -E - | grep __ELF__ > /dev/null; then
     case $host_cpu in
     i*86 )
@@ -3506,7 +3518,7 @@ hpux10.20* | hpux11*)
   esac
   ;;
 
-interix3*)
+interix[3-9]*)
   # PIC code is broken on Interix 3.x, that's why |\.a not |_pic\.a here
   lt_cv_deplibs_check_method='match_pattern /lib[^/]+(\.so|\.a)$'
   ;;
@@ -3522,7 +3534,7 @@ irix5* | irix6* | nonstopux*)
   ;;
 
 # This must be Linux ELF.
-linux*)
+linux* | k*bsd*-gnu)
   lt_cv_deplibs_check_method=pass_all
   ;;
 
@@ -3556,6 +3568,10 @@ osf3* | osf4* | osf5*)
   lt_cv_deplibs_check_method=pass_all
   ;;
 
+rdos*)
+  lt_cv_deplibs_check_method=pass_all
+  ;;
+
 solaris*)
   lt_cv_deplibs_check_method=pass_all
   ;;
@@ -3643,7 +3659,7 @@ ia64-*-hpux*)
   ;;
 *-*-irix6*)
   # Find out which ABI we are using.
-  echo '#line 3646 "configure"' > conftest.$ac_ext
+  echo '#line 3662 "configure"' > conftest.$ac_ext
   if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5
   (eval $ac_compile) 2>&5
   ac_status=$?
@@ -3678,7 +3694,8 @@ ia64-*-hpux*)
   rm -rf conftest*
   ;;
 
-x86_64-*linux*|ppc*-*linux*|powerpc*-*linux*|s390*-*linux*|sparc*-*linux*)
+x86_64-*kfreebsd*-gnu|x86_64-*linux*|ppc*-*linux*|powerpc*-*linux*| \
+s390*-*linux*|sparc*-*linux*)
   # Find out which ABI we are using.
   echo 'int i;' > conftest.$ac_ext
   if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5
@@ -3689,6 +3706,9 @@ x86_64-*linux*|ppc*-*linux*|powerpc*-*linux*|s390*-*linux*|sparc*-*linux*)
     case `/usr/bin/file conftest.o` in
     *32-bit*)
       case $host in
+        x86_64-*kfreebsd*-gnu)
+          LD="${LD-ld} -m elf_i386_fbsd"
+          ;;
         x86_64-*linux*)
           LD="${LD-ld} -m elf_i386"
           ;;
@@ -3705,6 +3725,9 @@ x86_64-*linux*|ppc*-*linux*|powerpc*-*linux*|s390*-*linux*|sparc*-*linux*)
       ;;
     *64-bit*)
       case $host in
+        x86_64-*kfreebsd*-gnu)
+          LD="${LD-ld} -m elf_x86_64_fbsd"
+          ;;
         x86_64-*linux*)
           LD="${LD-ld} -m elf_x86_64"
           ;;
@@ -3811,7 +3834,11 @@ sparc*-*solaris*)
     *64-bit*)
       case $lt_cv_prog_gnu_ld in
       yes*) LD="${LD-ld} -m elf64_sparc" ;;
-      *)    LD="${LD-ld} -64" ;;
+      *)
+        if ${LD-ld} -64 -r -o conftest2.o conftest.o >/dev/null 2>&1; then
+	  LD="${LD-ld} -64"
+	fi
+	;;
       esac
       ;;
     esac
@@ -5242,7 +5269,7 @@ fi
 
 
 # Provide some information about the compiler.
-echo "$as_me:5245:" \
+echo "$as_me:5272:" \
      "checking for Fortran 77 compiler version" >&5
 ac_compiler=`set X $ac_compile; echo $2`
 { (eval echo "$as_me:$LINENO: \"$ac_compiler --version </dev/null >&5\"") >&5
@@ -5389,7 +5416,6 @@ ac_compiler_gnu=$ac_cv_c_compiler_gnu
 
 
 # Autoconf 2.13's AC_OBJEXT and AC_EXEEXT macros only works for C compilers!
-
 # find the maximum length of command line arguments
 echo "$as_me:$LINENO: checking the maximum length of command line arguments" >&5
 echo $ECHO_N "checking the maximum length of command line arguments... $ECHO_C" >&6
@@ -5476,24 +5502,27 @@ else
     fi
     ;;
   *)
-    # If test is not a shell built-in, we'll probably end up computing a
-    # maximum length that is only half of the actual maximum length, but
-    # we can't tell.
-    SHELL=${SHELL-${CONFIG_SHELL-/bin/sh}}
-    while (test "X"`$SHELL $0 --fallback-echo "X$teststring" 2>/dev/null` \
+    lt_cv_sys_max_cmd_len=`(getconf ARG_MAX) 2> /dev/null`
+    if test -n "$lt_cv_sys_max_cmd_len"; then
+      lt_cv_sys_max_cmd_len=`expr $lt_cv_sys_max_cmd_len \/ 4`
+      lt_cv_sys_max_cmd_len=`expr $lt_cv_sys_max_cmd_len \* 3`
+    else
+      SHELL=${SHELL-${CONFIG_SHELL-/bin/sh}}
+      while (test "X"`$SHELL $0 --fallback-echo "X$teststring" 2>/dev/null` \
 	       = "XX$teststring") >/dev/null 2>&1 &&
-	    new_result=`expr "X$teststring" : ".*" 2>&1` &&
-	    lt_cv_sys_max_cmd_len=$new_result &&
-	    test $i != 17 # 1/2 MB should be enough
-    do
-      i=`expr $i + 1`
-      teststring=$teststring$teststring
-    done
-    teststring=
-    # Add a significant safety factor because C++ compilers can tack on massive
-    # amounts of additional arguments before passing them to the linker.
-    # It appears as though 1/2 is a usable value.
-    lt_cv_sys_max_cmd_len=`expr $lt_cv_sys_max_cmd_len \/ 2`
+	      new_result=`expr "X$teststring" : ".*" 2>&1` &&
+	      lt_cv_sys_max_cmd_len=$new_result &&
+	      test $i != 17 # 1/2 MB should be enough
+      do
+        i=`expr $i + 1`
+        teststring=$teststring$teststring
+      done
+      teststring=
+      # Add a significant safety factor because C++ compilers can tack on massive
+      # amounts of additional arguments before passing them to the linker.
+      # It appears as though 1/2 is a usable value.
+      lt_cv_sys_max_cmd_len=`expr $lt_cv_sys_max_cmd_len \/ 2`
+    fi
     ;;
   esac
 
@@ -5510,6 +5539,7 @@ fi
 
 
 
+
 # Check for command to grab the raw symbol name followed by C symbol from nm.
 echo "$as_me:$LINENO: checking command to parse $NM output from $compiler object" >&5
 echo $ECHO_N "checking command to parse $NM output from $compiler object... $ECHO_C" >&6
@@ -5547,7 +5577,7 @@ hpux*) # Its linker distinguishes data from code symbols
   lt_cv_sys_global_symbol_to_cdecl="sed -n -e 's/^T .* \(.*\)$/extern int \1();/p' -e 's/^$symcode* .* \(.*\)$/extern char \1;/p'"
   lt_cv_sys_global_symbol_to_c_name_address="sed -n -e 's/^: \([^ ]*\) $/  {\\\"\1\\\", (lt_ptr) 0},/p' -e 's/^$symcode* \([^ ]*\) \([^ ]*\)$/  {\"\2\", (lt_ptr) \&\2},/p'"
   ;;
-linux*)
+linux* | k*bsd*-gnu)
   if test "$host_cpu" = ia64; then
     symcode='[ABCDGIRSTW]'
     lt_cv_sys_global_symbol_to_cdecl="sed -n -e 's/^T .* \(.*\)$/extern int \1();/p' -e 's/^$symcode* .* \(.*\)$/extern char \1;/p'"
@@ -5700,7 +5730,7 @@ EOF
     echo "$progname: failed program was:" >&5
     cat conftest.$ac_ext >&5
   fi
-  rm -f conftest* conftst*
+  rm -rf conftest* conftst*
 
   # Do not use the global_symbol_pipe unless it works.
   if test "$pipe_works" = yes; then
@@ -6050,7 +6080,7 @@ test -z "$STRIP" && STRIP=:
 test -z "$ac_objext" && ac_objext=o
 
 # Determine commands to create old-style static archives.
-old_archive_cmds='$AR $AR_FLAGS $oldlib$oldobjs$old_deplibs'
+old_archive_cmds='$AR $AR_FLAGS $oldlib$oldobjs'
 old_postinstall_cmds='chmod 644 $oldlib'
 old_postuninstall_cmds=
 
@@ -6212,6 +6242,289 @@ fi
   ;;
 esac
 
+
+  case $host_os in
+    rhapsody* | darwin*)
+    if test -n "$ac_tool_prefix"; then
+  # Extract the first word of "${ac_tool_prefix}dsymutil", so it can be a program name with args.
+set dummy ${ac_tool_prefix}dsymutil; ac_word=$2
+echo "$as_me:$LINENO: checking for $ac_word" >&5
+echo $ECHO_N "checking for $ac_word... $ECHO_C" >&6
+if test "${ac_cv_prog_DSYMUTIL+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  if test -n "$DSYMUTIL"; then
+  ac_cv_prog_DSYMUTIL="$DSYMUTIL" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if $as_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_DSYMUTIL="${ac_tool_prefix}dsymutil"
+    echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+done
+
+fi
+fi
+DSYMUTIL=$ac_cv_prog_DSYMUTIL
+if test -n "$DSYMUTIL"; then
+  echo "$as_me:$LINENO: result: $DSYMUTIL" >&5
+echo "${ECHO_T}$DSYMUTIL" >&6
+else
+  echo "$as_me:$LINENO: result: no" >&5
+echo "${ECHO_T}no" >&6
+fi
+
+fi
+if test -z "$ac_cv_prog_DSYMUTIL"; then
+  ac_ct_DSYMUTIL=$DSYMUTIL
+  # Extract the first word of "dsymutil", so it can be a program name with args.
+set dummy dsymutil; ac_word=$2
+echo "$as_me:$LINENO: checking for $ac_word" >&5
+echo $ECHO_N "checking for $ac_word... $ECHO_C" >&6
+if test "${ac_cv_prog_ac_ct_DSYMUTIL+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  if test -n "$ac_ct_DSYMUTIL"; then
+  ac_cv_prog_ac_ct_DSYMUTIL="$ac_ct_DSYMUTIL" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if $as_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_ac_ct_DSYMUTIL="dsymutil"
+    echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+done
+
+  test -z "$ac_cv_prog_ac_ct_DSYMUTIL" && ac_cv_prog_ac_ct_DSYMUTIL=":"
+fi
+fi
+ac_ct_DSYMUTIL=$ac_cv_prog_ac_ct_DSYMUTIL
+if test -n "$ac_ct_DSYMUTIL"; then
+  echo "$as_me:$LINENO: result: $ac_ct_DSYMUTIL" >&5
+echo "${ECHO_T}$ac_ct_DSYMUTIL" >&6
+else
+  echo "$as_me:$LINENO: result: no" >&5
+echo "${ECHO_T}no" >&6
+fi
+
+  DSYMUTIL=$ac_ct_DSYMUTIL
+else
+  DSYMUTIL="$ac_cv_prog_DSYMUTIL"
+fi
+
+    if test -n "$ac_tool_prefix"; then
+  # Extract the first word of "${ac_tool_prefix}nmedit", so it can be a program name with args.
+set dummy ${ac_tool_prefix}nmedit; ac_word=$2
+echo "$as_me:$LINENO: checking for $ac_word" >&5
+echo $ECHO_N "checking for $ac_word... $ECHO_C" >&6
+if test "${ac_cv_prog_NMEDIT+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  if test -n "$NMEDIT"; then
+  ac_cv_prog_NMEDIT="$NMEDIT" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if $as_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_NMEDIT="${ac_tool_prefix}nmedit"
+    echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+done
+
+fi
+fi
+NMEDIT=$ac_cv_prog_NMEDIT
+if test -n "$NMEDIT"; then
+  echo "$as_me:$LINENO: result: $NMEDIT" >&5
+echo "${ECHO_T}$NMEDIT" >&6
+else
+  echo "$as_me:$LINENO: result: no" >&5
+echo "${ECHO_T}no" >&6
+fi
+
+fi
+if test -z "$ac_cv_prog_NMEDIT"; then
+  ac_ct_NMEDIT=$NMEDIT
+  # Extract the first word of "nmedit", so it can be a program name with args.
+set dummy nmedit; ac_word=$2
+echo "$as_me:$LINENO: checking for $ac_word" >&5
+echo $ECHO_N "checking for $ac_word... $ECHO_C" >&6
+if test "${ac_cv_prog_ac_ct_NMEDIT+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  if test -n "$ac_ct_NMEDIT"; then
+  ac_cv_prog_ac_ct_NMEDIT="$ac_ct_NMEDIT" # Let the user override the test.
+else
+as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if $as_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_prog_ac_ct_NMEDIT="nmedit"
+    echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+done
+
+  test -z "$ac_cv_prog_ac_ct_NMEDIT" && ac_cv_prog_ac_ct_NMEDIT=":"
+fi
+fi
+ac_ct_NMEDIT=$ac_cv_prog_ac_ct_NMEDIT
+if test -n "$ac_ct_NMEDIT"; then
+  echo "$as_me:$LINENO: result: $ac_ct_NMEDIT" >&5
+echo "${ECHO_T}$ac_ct_NMEDIT" >&6
+else
+  echo "$as_me:$LINENO: result: no" >&5
+echo "${ECHO_T}no" >&6
+fi
+
+  NMEDIT=$ac_ct_NMEDIT
+else
+  NMEDIT="$ac_cv_prog_NMEDIT"
+fi
+
+
+    echo "$as_me:$LINENO: checking for -single_module linker flag" >&5
+echo $ECHO_N "checking for -single_module linker flag... $ECHO_C" >&6
+if test "${lt_cv_apple_cc_single_mod+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  lt_cv_apple_cc_single_mod=no
+      if test -z "${LT_MULTI_MODULE}"; then
+   # By default we will add the -single_module flag. You can override
+   # by either setting the environment variable LT_MULTI_MODULE
+   # non-empty at configure time, or by adding -multi_module to the
+   # link flags.
+   echo "int foo(void){return 1;}" > conftest.c
+   $LTCC $LTCFLAGS $LDFLAGS -o libconftest.dylib \
+     -dynamiclib ${wl}-single_module conftest.c
+   if test -f libconftest.dylib; then
+     lt_cv_apple_cc_single_mod=yes
+     rm -rf libconftest.dylib*
+   fi
+   rm conftest.c
+      fi
+fi
+echo "$as_me:$LINENO: result: $lt_cv_apple_cc_single_mod" >&5
+echo "${ECHO_T}$lt_cv_apple_cc_single_mod" >&6
+    echo "$as_me:$LINENO: checking for -exported_symbols_list linker flag" >&5
+echo $ECHO_N "checking for -exported_symbols_list linker flag... $ECHO_C" >&6
+if test "${lt_cv_ld_exported_symbols_list+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  lt_cv_ld_exported_symbols_list=no
+      save_LDFLAGS=$LDFLAGS
+      echo "_main" > conftest.sym
+      LDFLAGS="$LDFLAGS -Wl,-exported_symbols_list,conftest.sym"
+      cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+
+int
+main ()
+{
+
+  ;
+  return 0;
+}
+_ACEOF
+rm -f conftest.$ac_objext conftest$ac_exeext
+if { (eval echo "$as_me:$LINENO: \"$ac_link\"") >&5
+  (eval $ac_link) 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } &&
+	 { ac_try='test -z "$ac_c_werror_flag"
+			 || test ! -s conftest.err'
+  { (eval echo "$as_me:$LINENO: \"$ac_try\"") >&5
+  (eval $ac_try) 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; } &&
+	 { ac_try='test -s conftest$ac_exeext'
+  { (eval echo "$as_me:$LINENO: \"$ac_try\"") >&5
+  (eval $ac_try) 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+  lt_cv_ld_exported_symbols_list=yes
+else
+  echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+lt_cv_ld_exported_symbols_list=no
+fi
+rm -f conftest.err conftest.$ac_objext \
+      conftest$ac_exeext conftest.$ac_ext
+   LDFLAGS="$save_LDFLAGS"
+
+fi
+echo "$as_me:$LINENO: result: $lt_cv_ld_exported_symbols_list" >&5
+echo "${ECHO_T}$lt_cv_ld_exported_symbols_list" >&6
+    case $host_os in
+    rhapsody* | darwin1.[0123])
+      _lt_dar_allow_undefined='${wl}-undefined ${wl}suppress' ;;
+    darwin1.*)
+     _lt_dar_allow_undefined='${wl}-flat_namespace ${wl}-undefined ${wl}suppress' ;;
+    darwin*)
+      # if running on 10.5 or later, the deployment target defaults
+      # to the OS version, if on x86, and 10.4, the deployment
+      # target defaults to 10.4. Don't you love it?
+      case ${MACOSX_DEPLOYMENT_TARGET-10.0},$host in
+   10.0,*86*-darwin8*|10.0,*-darwin[91]*)
+     _lt_dar_allow_undefined='${wl}-undefined ${wl}dynamic_lookup' ;;
+   10.[012]*)
+     _lt_dar_allow_undefined='${wl}-flat_namespace ${wl}-undefined ${wl}suppress' ;;
+   10.*)
+     _lt_dar_allow_undefined='${wl}-undefined ${wl}dynamic_lookup' ;;
+      esac
+    ;;
+  esac
+    if test "$lt_cv_apple_cc_single_mod" = "yes"; then
+      _lt_dar_single_mod='$single_module'
+    fi
+    if test "$lt_cv_ld_exported_symbols_list" = "yes"; then
+      _lt_dar_export_syms=' ${wl}-exported_symbols_list,$output_objdir/${libname}-symbols.expsym'
+    else
+      _lt_dar_export_syms="~$NMEDIT -s \$output_objdir/\${libname}-symbols.expsym \${lib}"
+    fi
+    if test "$DSYMUTIL" != ":"; then
+      _lt_dsymutil="~$DSYMUTIL \$lib || :"
+    else
+      _lt_dsymutil=
+    fi
+    ;;
+  esac
+
+
 enable_dlopen=no
 enable_win32_dll=no
 
@@ -6250,10 +6563,10 @@ objext=o
 objext=$objext
 
 # Code to be used in simple compile tests
-lt_simple_compile_test_code="int some_variable = 0;\n"
+lt_simple_compile_test_code="int some_variable = 0;"
 
 # Code to be used in simple link tests
-lt_simple_link_test_code='int main(){return(0);}\n'
+lt_simple_link_test_code='int main(){return(0);}'
 
 
 # If no C compiler was specified, use CC.
@@ -6268,16 +6581,16 @@ compiler=$CC
 
 # save warnings/boilerplate of simple test code
 ac_outfile=conftest.$ac_objext
-printf "$lt_simple_compile_test_code" >conftest.$ac_ext
+echo "$lt_simple_compile_test_code" >conftest.$ac_ext
 eval "$ac_compile" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_compiler_boilerplate=`cat conftest.err`
 $rm conftest*
 
 ac_outfile=conftest.$ac_objext
-printf "$lt_simple_link_test_code" >conftest.$ac_ext
+echo "$lt_simple_link_test_code" >conftest.$ac_ext
 eval "$ac_link" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_linker_boilerplate=`cat conftest.err`
-$rm conftest*
+$rm -r conftest*
 
 
 
@@ -6294,7 +6607,7 @@ if test "${lt_cv_prog_compiler_rtti_exceptions+set}" = set; then
 else
   lt_cv_prog_compiler_rtti_exceptions=no
   ac_outfile=conftest.$ac_objext
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
    lt_compiler_flag="-fno-rtti -fno-exceptions"
    # Insert the option either (1) after the last *FLAGS variable, or
    # (2) before a word containing "conftest.", or (3) at the end.
@@ -6305,11 +6618,11 @@ else
    -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \
    -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \
    -e 's:$: $lt_compiler_flag:'`
-   (eval echo "\"\$as_me:6308: $lt_compile\"" >&5)
+   (eval echo "\"\$as_me:6621: $lt_compile\"" >&5)
    (eval "$lt_compile" 2>conftest.err)
    ac_status=$?
    cat conftest.err >&5
-   echo "$as_me:6312: \$? = $ac_status" >&5
+   echo "$as_me:6625: \$? = $ac_status" >&5
    if (exit $ac_status) && test -s "$ac_outfile"; then
      # The compiler can only warn and ignore the option if not recognized
      # So say no if there are warnings other than the usual output.
@@ -6360,13 +6673,15 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_pic='-m68020 -resident32 -malways-restore-a4'
       ;;
 
-    beos* | cygwin* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
+    beos* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
       # PIC is the default for these OSes.
       ;;
 
-    mingw* | pw32* | os2*)
+    mingw* | cygwin* | pw32* | os2*)
       # This hack is so that the source file can tell whether it is being
       # built for inclusion in a dll (and should export symbols for example).
+      # Although the cygwin gcc ignores -fPIC, still need this for old-style
+      # (--disable-auto-import) libraries
       lt_prog_compiler_pic='-DDLL_EXPORT'
       ;;
 
@@ -6376,7 +6691,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_pic='-fno-common'
       ;;
 
-    interix3*)
+    interix[3-9]*)
       # Interix 3.x gcc -fpic/-fPIC options generate broken code.
       # Instead, we relocate shared libraries at runtime.
       ;;
@@ -6434,7 +6749,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
        esac
        ;;
 
-    mingw* | pw32* | os2*)
+    mingw* | cygwin* | pw32* | os2*)
       # This hack is so that the source file can tell whether it is being
       # built for inclusion in a dll (and should export symbols for example).
       lt_prog_compiler_pic='-DDLL_EXPORT'
@@ -6467,7 +6782,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_static='-Bstatic'
       ;;
 
-    linux*)
+    linux* | k*bsd*-gnu)
       case $cc_basename in
       icc* | ecc*)
 	lt_prog_compiler_wl='-Wl,'
@@ -6486,6 +6801,22 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
         # All Alpha code is PIC.
         lt_prog_compiler_static='-non_shared'
         ;;
+      *)
+        case `$CC -V 2>&1 | sed 5q` in
+	*Sun\ C*)
+	  # Sun C 5.9
+	  lt_prog_compiler_pic='-KPIC'
+	  lt_prog_compiler_static='-Bstatic'
+	  lt_prog_compiler_wl='-Wl,'
+	  ;;
+	*Sun\ F*)
+	  # Sun Fortran 8.3 passes all unrecognized flags to the linker
+	  lt_prog_compiler_pic='-KPIC'
+	  lt_prog_compiler_static='-Bstatic'
+	  lt_prog_compiler_wl=''
+	  ;;
+	esac
+	;;
       esac
       ;;
 
@@ -6495,6 +6826,10 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_static='-non_shared'
       ;;
 
+    rdos*)
+      lt_prog_compiler_static='-non_shared'
+      ;;
+
     solaris*)
       lt_prog_compiler_pic='-KPIC'
       lt_prog_compiler_static='-Bstatic'
@@ -6557,12 +6892,12 @@ if test -n "$lt_prog_compiler_pic"; then
 
 echo "$as_me:$LINENO: checking if $compiler PIC flag $lt_prog_compiler_pic works" >&5
 echo $ECHO_N "checking if $compiler PIC flag $lt_prog_compiler_pic works... $ECHO_C" >&6
-if test "${lt_prog_compiler_pic_works+set}" = set; then
+if test "${lt_cv_prog_compiler_pic_works+set}" = set; then
   echo $ECHO_N "(cached) $ECHO_C" >&6
 else
-  lt_prog_compiler_pic_works=no
+  lt_cv_prog_compiler_pic_works=no
   ac_outfile=conftest.$ac_objext
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
    lt_compiler_flag="$lt_prog_compiler_pic -DPIC"
    # Insert the option either (1) after the last *FLAGS variable, or
    # (2) before a word containing "conftest.", or (3) at the end.
@@ -6573,27 +6908,27 @@ else
    -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \
    -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \
    -e 's:$: $lt_compiler_flag:'`
-   (eval echo "\"\$as_me:6576: $lt_compile\"" >&5)
+   (eval echo "\"\$as_me:6911: $lt_compile\"" >&5)
    (eval "$lt_compile" 2>conftest.err)
    ac_status=$?
    cat conftest.err >&5
-   echo "$as_me:6580: \$? = $ac_status" >&5
+   echo "$as_me:6915: \$? = $ac_status" >&5
    if (exit $ac_status) && test -s "$ac_outfile"; then
      # The compiler can only warn and ignore the option if not recognized
      # So say no if there are warnings other than the usual output.
      $echo "X$_lt_compiler_boilerplate" | $Xsed -e '/^$/d' >conftest.exp
      $SED '/^$/d; /^ *+/d' conftest.err >conftest.er2
      if test ! -s conftest.er2 || diff conftest.exp conftest.er2 >/dev/null; then
-       lt_prog_compiler_pic_works=yes
+       lt_cv_prog_compiler_pic_works=yes
      fi
    fi
    $rm conftest*
 
 fi
-echo "$as_me:$LINENO: result: $lt_prog_compiler_pic_works" >&5
-echo "${ECHO_T}$lt_prog_compiler_pic_works" >&6
+echo "$as_me:$LINENO: result: $lt_cv_prog_compiler_pic_works" >&5
+echo "${ECHO_T}$lt_cv_prog_compiler_pic_works" >&6
 
-if test x"$lt_prog_compiler_pic_works" = xyes; then
+if test x"$lt_cv_prog_compiler_pic_works" = xyes; then
     case $lt_prog_compiler_pic in
      "" | " "*) ;;
      *) lt_prog_compiler_pic=" $lt_prog_compiler_pic" ;;
@@ -6620,13 +6955,13 @@ esac
 wl=$lt_prog_compiler_wl eval lt_tmp_static_flag=\"$lt_prog_compiler_static\"
 echo "$as_me:$LINENO: checking if $compiler static flag $lt_tmp_static_flag works" >&5
 echo $ECHO_N "checking if $compiler static flag $lt_tmp_static_flag works... $ECHO_C" >&6
-if test "${lt_prog_compiler_static_works+set}" = set; then
+if test "${lt_cv_prog_compiler_static_works+set}" = set; then
   echo $ECHO_N "(cached) $ECHO_C" >&6
 else
-  lt_prog_compiler_static_works=no
+  lt_cv_prog_compiler_static_works=no
    save_LDFLAGS="$LDFLAGS"
    LDFLAGS="$LDFLAGS $lt_tmp_static_flag"
-   printf "$lt_simple_link_test_code" > conftest.$ac_ext
+   echo "$lt_simple_link_test_code" > conftest.$ac_ext
    if (eval $ac_link 2>conftest.err) && test -s conftest$ac_exeext; then
      # The linker can only warn and ignore the option if not recognized
      # So say no if there are warnings
@@ -6636,20 +6971,20 @@ else
        $echo "X$_lt_linker_boilerplate" | $Xsed -e '/^$/d' > conftest.exp
        $SED '/^$/d; /^ *+/d' conftest.err >conftest.er2
        if diff conftest.exp conftest.er2 >/dev/null; then
-         lt_prog_compiler_static_works=yes
+         lt_cv_prog_compiler_static_works=yes
        fi
      else
-       lt_prog_compiler_static_works=yes
+       lt_cv_prog_compiler_static_works=yes
      fi
    fi
-   $rm conftest*
+   $rm -r conftest*
    LDFLAGS="$save_LDFLAGS"
 
 fi
-echo "$as_me:$LINENO: result: $lt_prog_compiler_static_works" >&5
-echo "${ECHO_T}$lt_prog_compiler_static_works" >&6
+echo "$as_me:$LINENO: result: $lt_cv_prog_compiler_static_works" >&5
+echo "${ECHO_T}$lt_cv_prog_compiler_static_works" >&6
 
-if test x"$lt_prog_compiler_static_works" = xyes; then
+if test x"$lt_cv_prog_compiler_static_works" = xyes; then
     :
 else
     lt_prog_compiler_static=
@@ -6666,7 +7001,7 @@ else
    mkdir conftest
    cd conftest
    mkdir out
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
 
    lt_compiler_flag="-o out/conftest2.$ac_objext"
    # Insert the option either (1) after the last *FLAGS variable, or
@@ -6677,11 +7012,11 @@ else
    -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \
    -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \
    -e 's:$: $lt_compiler_flag:'`
-   (eval echo "\"\$as_me:6680: $lt_compile\"" >&5)
+   (eval echo "\"\$as_me:7015: $lt_compile\"" >&5)
    (eval "$lt_compile" 2>out/conftest.err)
    ac_status=$?
    cat out/conftest.err >&5
-   echo "$as_me:6684: \$? = $ac_status" >&5
+   echo "$as_me:7019: \$? = $ac_status" >&5
    if (exit $ac_status) && test -s out/conftest2.$ac_objext
    then
      # The compiler can only warn and ignore the option if not recognized
@@ -6761,12 +7096,13 @@ echo $ECHO_N "checking whether the $compiler linker ($LD) supports shared librar
   # it will be wrapped by ` (' and `)$', so one must not match beginning or
   # end of line.  Example: `a|bc|.*d.*' will exclude the symbols `a' and `bc',
   # as well as any symbol that contains `d'.
-  exclude_expsyms="_GLOBAL_OFFSET_TABLE_"
+  exclude_expsyms='_GLOBAL_OFFSET_TABLE_|_GLOBAL__F[ID]_.*'
   # Although _GLOBAL_OFFSET_TABLE_ is a valid symbol C name, most a.out
   # platforms (ab)use it in PIC code, but their linkers get confused if
   # the symbol is explicitly referenced.  Since portable code cannot
   # rely on this symbol name, it's probably fine to never include it in
   # preloaded symbol tables.
+  # Exclude shared library initialization/finalization symbols.
   extract_expsyms_cmds=
   # Just being paranoid about ensuring that cc_basename is set.
   for cc_temp in $compiler""; do
@@ -6825,7 +7161,7 @@ cc_basename=`$echo "X$cc_temp" | $Xsed -e 's%.*/%%' -e "s%^$host_alias-%%"`
 
     # See if GNU ld supports shared libraries.
     case $host_os in
-    aix3* | aix4* | aix5*)
+    aix[3-9]*)
       # On AIX/PPC, the GNU linker is very broken
       if test "$host_cpu" != ia64; then
 	ld_shlibs=no
@@ -6873,7 +7209,7 @@ EOF
       allow_undefined_flag=unsupported
       always_export_symbols=no
       enable_shared_with_static_runtimes=yes
-      export_symbols_cmds='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[BCDGRS] /s/.* \([^ ]*\)/\1 DATA/'\'' | $SED -e '\''/^[AITW] /s/.* //'\'' | sort | uniq > $export_symbols'
+      export_symbols_cmds='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[BCDGRS][ ]/s/.*[ ]\([^ ]*\)/\1 DATA/'\'' -e '\''/^[AITW][ ]/s/.*[ ]//'\'' | sort | uniq > $export_symbols'
 
       if $LD --help 2>&1 | grep 'auto-import' > /dev/null; then
         archive_cmds='$CC -shared $libobjs $deplibs $compiler_flags -o $output_objdir/$soname ${wl}--enable-auto-image-base -Xlinker --out-implib -Xlinker $lib'
@@ -6891,7 +7227,7 @@ EOF
       fi
       ;;
 
-    interix3*)
+    interix[3-9]*)
       hardcode_direct=no
       hardcode_shlibpath_var=no
       hardcode_libdir_flag_spec='${wl}-rpath,$libdir'
@@ -6906,7 +7242,7 @@ EOF
       archive_expsym_cmds='sed "s,^,_," $export_symbols >$output_objdir/$soname.expsym~$CC -shared $pic_flag $libobjs $deplibs $compiler_flags ${wl}-h,$soname ${wl}--retain-symbols-file,$output_objdir/$soname.expsym ${wl}--image-base,`expr ${RANDOM-$$} % 4096 / 2 \* 262144 + 1342177280` -o $lib'
       ;;
 
-    linux*)
+    gnu* | linux* | k*bsd*-gnu)
       if $LD --help 2>&1 | grep ': supported targets:.* elf' > /dev/null; then
 	tmp_addflag=
 	case $cc_basename,$host_cpu in
@@ -6924,13 +7260,22 @@ EOF
 	ifc* | ifort*)			# Intel Fortran compiler
 	  tmp_addflag=' -nofor_main' ;;
 	esac
-	archive_cmds='$CC -shared'"$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname -o $lib'
+	case `$CC -V 2>&1 | sed 5q` in
+	*Sun\ C*)			# Sun C 5.9
+	  whole_archive_flag_spec='${wl}--whole-archive`new_convenience=; for conv in $convenience\"\"; do test -z \"$conv\" || new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}--no-whole-archive'
+	  tmp_sharedflag='-G' ;;
+	*Sun\ F*)			# Sun Fortran 8.3
+	  tmp_sharedflag='-G' ;;
+	*)
+	  tmp_sharedflag='-shared' ;;
+	esac
+	archive_cmds='$CC '"$tmp_sharedflag""$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname -o $lib'
 
 	if test $supports_anon_versioning = yes; then
 	  archive_expsym_cmds='$echo "{ global:" > $output_objdir/$libname.ver~
   cat $export_symbols | sed -e "s/\(.*\)/\1;/" >> $output_objdir/$libname.ver~
   $echo "local: *; };" >> $output_objdir/$libname.ver~
-	  $CC -shared'"$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname ${wl}-version-script ${wl}$output_objdir/$libname.ver -o $lib'
+	  $CC '"$tmp_sharedflag""$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname ${wl}-version-script ${wl}$output_objdir/$libname.ver -o $lib'
 	fi
       else
 	ld_shlibs=no
@@ -7035,7 +7380,7 @@ _LT_EOF
       fi
       ;;
 
-    aix4* | aix5*)
+    aix[4-9]*)
       if test "$host_cpu" = ia64; then
 	# On IA64, the linker does run time linking by default, so we don't
 	# have to do anything special.
@@ -7055,7 +7400,7 @@ _LT_EOF
 	# Test if we are trying to use run time linking or normal
 	# AIX style linking. If -brtl is somewhere in LDFLAGS, we
 	# need to do runtime linking.
-	case $host_os in aix4.[23]|aix4.[23].*|aix5*)
+	case $host_os in aix4.[23]|aix4.[23].*|aix[5-9]*)
 	  for ld_flag in $LDFLAGS; do
   	  if (test $ld_flag = "-brtl" || test $ld_flag = "-Wl,-brtl"); then
   	    aix_use_runtimelinking=yes
@@ -7089,7 +7434,7 @@ _LT_EOF
   	   strings "$collect2name" | grep resolve_lib_name >/dev/null
 	  then
   	  # We have reworked collect2
-  	  hardcode_direct=yes
+  	  :
 	  else
   	  # We have old collect2
   	  hardcode_direct=unsupported
@@ -7167,11 +7512,18 @@ if { (eval echo "$as_me:$LINENO: \"$ac_link\"") >&5
   echo "$as_me:$LINENO: \$? = $ac_status" >&5
   (exit $ac_status); }; }; then
 
-aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`
+lt_aix_libpath_sed='
+    /Import File Strings/,/^$/ {
+	/^0/ {
+	    s/^0  *\(.*\)$/\1/
+	    p
+	}
+    }'
+aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
 # Check for a 64-bit object if we didn't find anything.
-if test -z "$aix_libpath"; then aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`; fi
+if test -z "$aix_libpath"; then
+  aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
+fi
 else
   echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
@@ -7228,11 +7580,18 @@ if { (eval echo "$as_me:$LINENO: \"$ac_link\"") >&5
   echo "$as_me:$LINENO: \$? = $ac_status" >&5
   (exit $ac_status); }; }; then
 
-aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`
+lt_aix_libpath_sed='
+    /Import File Strings/,/^$/ {
+	/^0/ {
+	    s/^0  *\(.*\)$/\1/
+	    p
+	}
+    }'
+aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
 # Check for a 64-bit object if we didn't find anything.
-if test -z "$aix_libpath"; then aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`; fi
+if test -z "$aix_libpath"; then
+  aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
+fi
 else
   echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
@@ -7284,7 +7643,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       # The linker will automatically build a .lib file if we build a DLL.
       old_archive_From_new_cmds='true'
       # FIXME: Should let the user specify the lib program.
-      old_archive_cmds='lib /OUT:$oldlib$oldobjs$old_deplibs'
+      old_archive_cmds='lib -OUT:$oldlib$oldobjs$old_deplibs'
       fix_srcfile_path='`cygpath -w "$srcfile"`'
       enable_shared_with_static_runtimes=yes
       ;;
@@ -7317,19 +7676,18 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       link_all_deplibs=yes
     if test "$GCC" = yes ; then
     	output_verbose_link_cmd='echo'
-        archive_cmds='$CC -dynamiclib $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring'
-      module_cmds='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
-      # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-      archive_expsym_cmds='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -dynamiclib $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
-      module_expsym_cmds='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
+        archive_cmds="\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring $_lt_dar_single_mod${_lt_dsymutil}"
+        module_cmds="\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dsymutil}"
+        archive_expsym_cmds="sed 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring ${_lt_dar_single_mod}${_lt_dar_export_syms}${_lt_dsymutil}"
+        module_expsym_cmds="sed -e 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dar_export_syms}${_lt_dsymutil}"
     else
       case $cc_basename in
         xlc*)
          output_verbose_link_cmd='echo'
-         archive_cmds='$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $verstring'
+         archive_cmds='$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $xlcverstring'
          module_cmds='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
           # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-         archive_expsym_cmds='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
+         archive_expsym_cmds='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $xlcverstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           module_expsym_cmds='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           ;;
        *)
@@ -7369,7 +7727,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       ;;
 
     # FreeBSD 3 and greater uses gcc -shared to do shared libraries.
-    freebsd* | kfreebsd*-gnu | dragonfly*)
+    freebsd* | dragonfly*)
       archive_cmds='$CC -shared -o $lib $libobjs $deplibs $compiler_flags'
       hardcode_libdir_flag_spec='-R$libdir'
       hardcode_direct=yes
@@ -7492,15 +7850,15 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 
     openbsd*)
       if test -f /usr/libexec/ld.so; then
-        hardcode_direct=yes
-        hardcode_shlibpath_var=no
-        if test -z "`echo __ELF__ | $CC -E - | grep __ELF__`" || test "$host_os-$host_cpu" = "openbsd2.8-powerpc"; then
+	hardcode_direct=yes
+	hardcode_shlibpath_var=no
+	if test -z "`echo __ELF__ | $CC -E - | grep __ELF__`" || test "$host_os-$host_cpu" = "openbsd2.8-powerpc"; then
 	  archive_cmds='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags'
 	  archive_expsym_cmds='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-retain-symbols-file,$export_symbols'
 	  hardcode_libdir_flag_spec='${wl}-rpath,$libdir'
 	  export_dynamic_flag_spec='${wl}-E'
-        else
-         case $host_os in
+	else
+	  case $host_os in
 	   openbsd[01].* | openbsd2.[0-7] | openbsd2.[0-7].*)
 	     archive_cmds='$LD -Bshareable -o $lib $libobjs $deplibs $linker_flags'
 	     hardcode_libdir_flag_spec='-R$libdir'
@@ -7509,10 +7867,10 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 	     archive_cmds='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags'
 	     hardcode_libdir_flag_spec='${wl}-rpath,$libdir'
 	     ;;
-         esac
+	  esac
         fi
       else
-        ld_shlibs=no
+	ld_shlibs=no
       fi
       ;;
 
@@ -7571,17 +7929,16 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       case $host_os in
       solaris2.[0-5] | solaris2.[0-5].*) ;;
       *)
- 	# The compiler driver will combine linker options so we
- 	# cannot just pass the convience library names through
- 	# without $wl, iff we do not link with $LD.
- 	# Luckily, gcc supports the same syntax we need for Sun Studio.
+	# The compiler driver will combine and reorder linker options,
+	# but understands `-z linker_flag'.  GCC discards it without `$wl',
+	# but is careful enough not to reorder.
  	# Supported since Solaris 2.6 (maybe 2.5.1?)
- 	case $wlarc in
- 	'')
- 	  whole_archive_flag_spec='-z allextract$convenience -z defaultextract' ;;
- 	*)
- 	  whole_archive_flag_spec='${wl}-z ${wl}allextract`for conv in $convenience\"\"; do test -n \"$conv\" && new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}-z ${wl}defaultextract' ;;
- 	esac ;;
+	if test "$GCC" = yes; then
+	  whole_archive_flag_spec='${wl}-z ${wl}allextract$convenience ${wl}-z ${wl}defaultextract'
+	else
+	  whole_archive_flag_spec='-z allextract$convenience -z defaultextract'
+	fi
+	;;
       esac
       link_all_deplibs=yes
       ;;
@@ -7638,7 +7995,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       fi
       ;;
 
-    sysv4*uw2* | sysv5OpenUNIX* | sysv5UnixWare7.[01].[10]* | unixware7*)
+    sysv4*uw2* | sysv5OpenUNIX* | sysv5UnixWare7.[01].[10]* | unixware7* | sco3.2v5.0.[024]*)
       no_undefined_flag='${wl}-z,text'
       archive_cmds_need_lc=no
       hardcode_shlibpath_var=no
@@ -7715,7 +8072,7 @@ x|xyes)
       echo "$as_me:$LINENO: checking whether -lc should be explicitly linked in" >&5
 echo $ECHO_N "checking whether -lc should be explicitly linked in... $ECHO_C" >&6
       $rm conftest*
-      printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+      echo "$lt_simple_compile_test_code" > conftest.$ac_ext
 
       if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5
   (eval $ac_compile) 2>&5
@@ -7773,17 +8130,55 @@ shlibpath_overrides_runpath=unknown
 version_type=none
 dynamic_linker="$host_os ld.so"
 sys_lib_dlsearch_path_spec="/lib /usr/lib"
+
 if test "$GCC" = yes; then
-  sys_lib_search_path_spec=`$CC -print-search-dirs | grep "^libraries:" | $SED -e "s/^libraries://" -e "s,=/,/,g"`
-  if echo "$sys_lib_search_path_spec" | grep ';' >/dev/null ; then
+  case $host_os in
+    darwin*) lt_awk_arg="/^libraries:/,/LR/" ;;
+    *) lt_awk_arg="/^libraries:/" ;;
+  esac
+  lt_search_path_spec=`$CC -print-search-dirs | awk $lt_awk_arg | $SED -e "s/^libraries://" -e "s,=/,/,g"`
+  if echo "$lt_search_path_spec" | grep ';' >/dev/null ; then
     # if the path contains ";" then we assume it to be the separator
     # otherwise default to the standard path separator (i.e. ":") - it is
     # assumed that no part of a normal pathname contains ";" but that should
     # okay in the real world where ";" in dirpaths is itself problematic.
-    sys_lib_search_path_spec=`echo "$sys_lib_search_path_spec" | $SED -e 's/;/ /g'`
+    lt_search_path_spec=`echo "$lt_search_path_spec" | $SED -e 's/;/ /g'`
   else
-    sys_lib_search_path_spec=`echo "$sys_lib_search_path_spec" | $SED  -e "s/$PATH_SEPARATOR/ /g"`
+    lt_search_path_spec=`echo "$lt_search_path_spec" | $SED  -e "s/$PATH_SEPARATOR/ /g"`
   fi
+  # Ok, now we have the path, separated by spaces, we can step through it
+  # and add multilib dir if necessary.
+  lt_tmp_lt_search_path_spec=
+  lt_multi_os_dir=`$CC $CPPFLAGS $CFLAGS $LDFLAGS -print-multi-os-directory 2>/dev/null`
+  for lt_sys_path in $lt_search_path_spec; do
+    if test -d "$lt_sys_path/$lt_multi_os_dir"; then
+      lt_tmp_lt_search_path_spec="$lt_tmp_lt_search_path_spec $lt_sys_path/$lt_multi_os_dir"
+    else
+      test -d "$lt_sys_path" && \
+	lt_tmp_lt_search_path_spec="$lt_tmp_lt_search_path_spec $lt_sys_path"
+    fi
+  done
+  lt_search_path_spec=`echo $lt_tmp_lt_search_path_spec | awk '
+BEGIN {RS=" "; FS="/|\n";} {
+  lt_foo="";
+  lt_count=0;
+  for (lt_i = NF; lt_i > 0; lt_i--) {
+    if ($lt_i != "" && $lt_i != ".") {
+      if ($lt_i == "..") {
+        lt_count++;
+      } else {
+        if (lt_count == 0) {
+          lt_foo="/" $lt_i lt_foo;
+        } else {
+          lt_count--;
+        }
+      }
+    }
+  }
+  if (lt_foo != "") { lt_freq[lt_foo]++; }
+  if (lt_freq[lt_foo] == 1) { print lt_foo; }
+}'`
+  sys_lib_search_path_spec=`echo $lt_search_path_spec`
 else
   sys_lib_search_path_spec="/lib /usr/lib /usr/local/lib"
 fi
@@ -7804,7 +8199,7 @@ aix3*)
   soname_spec='${libname}${release}${shared_ext}$major'
   ;;
 
-aix4* | aix5*)
+aix[4-9]*)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -7943,12 +8338,8 @@ darwin* | rhapsody*)
   shlibpath_overrides_runpath=yes
   shlibpath_var=DYLD_LIBRARY_PATH
   shrext_cmds='`test .$module = .yes && echo .so || echo .dylib`'
-  # Apple's gcc prints 'gcc -print-search-dirs' doesn't operate the same.
-  if test "$GCC" = yes; then
-    sys_lib_search_path_spec=`$CC -print-search-dirs | tr "\n" "$PATH_SEPARATOR" | sed -e 's/libraries:/@libraries:/' | tr "@" "\n" | grep "^libraries:" | sed -e "s/^libraries://" -e "s,=/,/,g" -e "s,$PATH_SEPARATOR, ,g" -e "s,.*,& /lib /usr/lib /usr/local/lib,g"`
-  else
-    sys_lib_search_path_spec='/lib /usr/lib /usr/local/lib'
-  fi
+
+  sys_lib_search_path_spec="$sys_lib_search_path_spec /usr/local/lib"
   sys_lib_dlsearch_path_spec='/usr/local/lib /lib /usr/lib'
   ;;
 
@@ -7965,18 +8356,6 @@ freebsd1*)
   dynamic_linker=no
   ;;
 
-kfreebsd*-gnu)
-  version_type=linux
-  need_lib_prefix=no
-  need_version=no
-  library_names_spec='${libname}${release}${shared_ext}$versuffix ${libname}${release}${shared_ext}$major ${libname}${shared_ext}'
-  soname_spec='${libname}${release}${shared_ext}$major'
-  shlibpath_var=LD_LIBRARY_PATH
-  shlibpath_overrides_runpath=no
-  hardcode_into_libs=yes
-  dynamic_linker='GNU ld.so'
-  ;;
-
 freebsd* | dragonfly*)
   # DragonFly does not have aout.  When/if they implement a new
   # versioning mechanism, adjust this.
@@ -8014,7 +8393,7 @@ freebsd* | dragonfly*)
     shlibpath_overrides_runpath=no
     hardcode_into_libs=yes
     ;;
-  freebsd*) # from 4.6 on
+  *) # from 4.6 on, and DragonFly
     shlibpath_overrides_runpath=yes
     hardcode_into_libs=yes
     ;;
@@ -8077,7 +8456,7 @@ hpux9* | hpux10* | hpux11*)
   postinstall_cmds='chmod 555 $lib'
   ;;
 
-interix3*)
+interix[3-9]*)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -8132,7 +8511,7 @@ linux*oldld* | linux*aout* | linux*coff*)
   ;;
 
 # This must be Linux ELF.
-linux*)
+linux* | k*bsd*-gnu)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -8148,7 +8527,7 @@ linux*)
 
   # Append ld.so.conf contents to the search path
   if test -f /etc/ld.so.conf; then
-    lt_ld_extra=`awk '/^include / { system(sprintf("cd /etc; cat %s", \$2)); skip = 1; } { if (!skip) print \$0; skip = 0; }' < /etc/ld.so.conf | $SED -e 's/#.*//;s/[:,	]/ /g;s/=[^=]*$//;s/=[^= ]* / /g;/^$/d' | tr '\n' ' '`
+    lt_ld_extra=`awk '/^include / { system(sprintf("cd /etc; cat %s 2>/dev/null", \$2)); skip = 1; } { if (!skip) print \$0; skip = 0; }' < /etc/ld.so.conf | $SED -e 's/#.*//;/^[ 	]*hwcap[ 	]/d;s/[:,	]/ /g;s/=[^=]*$//;s/=[^= ]* / /g;/^$/d' | tr '\n' ' '`
     sys_lib_dlsearch_path_spec="/lib /usr/lib $lt_ld_extra"
   fi
 
@@ -8161,18 +8540,6 @@ linux*)
   dynamic_linker='GNU/Linux ld.so'
   ;;
 
-knetbsd*-gnu)
-  version_type=linux
-  need_lib_prefix=no
-  need_version=no
-  library_names_spec='${libname}${release}${shared_ext}$versuffix ${libname}${release}${shared_ext}$major ${libname}${shared_ext}'
-  soname_spec='${libname}${release}${shared_ext}$major'
-  shlibpath_var=LD_LIBRARY_PATH
-  shlibpath_overrides_runpath=no
-  hardcode_into_libs=yes
-  dynamic_linker='GNU ld.so'
-  ;;
-
 netbsd*)
   version_type=sunos
   need_lib_prefix=no
@@ -8254,6 +8621,10 @@ osf3* | osf4* | osf5*)
   sys_lib_dlsearch_path_spec="$sys_lib_search_path_spec"
   ;;
 
+rdos*)
+  dynamic_linker=no
+  ;;
+
 solaris*)
   version_type=linux
   need_lib_prefix=no
@@ -8350,6 +8721,21 @@ echo "$as_me:$LINENO: result: $dynamic_linker" >&5
 echo "${ECHO_T}$dynamic_linker" >&6
 test "$dynamic_linker" = no && can_build_shared=no
 
+if test "${lt_cv_sys_lib_search_path_spec+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  lt_cv_sys_lib_search_path_spec="$sys_lib_search_path_spec"
+fi
+
+sys_lib_search_path_spec="$lt_cv_sys_lib_search_path_spec"
+if test "${lt_cv_sys_lib_dlsearch_path_spec+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  lt_cv_sys_lib_dlsearch_path_spec="$sys_lib_dlsearch_path_spec"
+fi
+
+sys_lib_dlsearch_path_spec="$lt_cv_sys_lib_dlsearch_path_spec"
+
 variables_saved_for_relink="PATH $shlibpath_var $runpath_var"
 if test "$GCC" = yes; then
   variables_saved_for_relink="$variables_saved_for_relink GCC_EXEC_PREFIX COMPILER_PATH LIBRARY_PATH"
@@ -8407,6 +8793,7 @@ else
    darwin*)
        if test -n "$STRIP" ; then
          striplib="$STRIP -x"
+         old_striplib="$STRIP -S"
          echo "$as_me:$LINENO: result: yes" >&5
 echo "${ECHO_T}yes" >&6
        else
@@ -8683,7 +9070,7 @@ fi
 echo "$as_me:$LINENO: result: $ac_cv_lib_dld_shl_load" >&5
 echo "${ECHO_T}$ac_cv_lib_dld_shl_load" >&6
 if test $ac_cv_lib_dld_shl_load = yes; then
-  lt_cv_dlopen="shl_load" lt_cv_dlopen_libs="-dld"
+  lt_cv_dlopen="shl_load" lt_cv_dlopen_libs="-ldld"
 else
   echo "$as_me:$LINENO: checking for dlopen" >&5
 echo $ECHO_N "checking for dlopen... $ECHO_C" >&6
@@ -8977,7 +9364,7 @@ fi
 echo "$as_me:$LINENO: result: $ac_cv_lib_dld_dld_link" >&5
 echo "${ECHO_T}$ac_cv_lib_dld_dld_link" >&6
 if test $ac_cv_lib_dld_dld_link = yes; then
-  lt_cv_dlopen="dld_link" lt_cv_dlopen_libs="-dld"
+  lt_cv_dlopen="dld_link" lt_cv_dlopen_libs="-ldld"
 fi
 
 
@@ -9026,7 +9413,7 @@ else
   lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2
   lt_status=$lt_dlunknown
   cat > conftest.$ac_ext <<EOF
-#line 9029 "configure"
+#line 9416 "configure"
 #include "confdefs.h"
 
 #if HAVE_DLFCN_H
@@ -9126,7 +9513,7 @@ else
   lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2
   lt_status=$lt_dlunknown
   cat > conftest.$ac_ext <<EOF
-#line 9129 "configure"
+#line 9516 "configure"
 #include "confdefs.h"
 
 #if HAVE_DLFCN_H
@@ -9253,7 +9640,7 @@ aix3*)
   fi
   ;;
 
-aix4* | aix5*)
+aix[4-9]*)
   if test "$host_cpu" != ia64 && test "$aix_use_runtimelinking" = no ; then
     test "$enable_shared" = yes && enable_static=no
   fi
@@ -9309,6 +9696,7 @@ if test -f "$ltmain"; then
     predeps \
     postdeps \
     compiler_lib_search_path \
+    compiler_lib_search_dirs \
     archive_cmds \
     archive_expsym_cmds \
     postinstall_cmds \
@@ -9324,6 +9712,7 @@ if test -f "$ltmain"; then
     module_cmds \
     module_expsym_cmds \
     lt_cv_prog_compiler_c_o \
+    fix_srcfile_path \
     exclude_expsyms \
     include_expsyms; do
 
@@ -9368,7 +9757,7 @@ echo "$as_me: creating $ofile" >&6;}
 # Generated automatically by $PROGRAM (GNU $PACKAGE $VERSION$TIMESTAMP)
 # NOTE: Changes made to this file will be lost: look at ltmain.sh.
 #
-# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001
+# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008
 # Free Software Foundation, Inc.
 #
 # This file is part of GNU Libtool:
@@ -9604,6 +9993,10 @@ predeps=$lt_predeps
 # shared library.
 postdeps=$lt_postdeps
 
+# The directories searched by this compiler when creating a shared
+# library
+compiler_lib_search_dirs=$lt_compiler_lib_search_dirs
+
 # The library search path used internally by the compiler when linking
 # a shared library.
 compiler_lib_search_path=$lt_compiler_lib_search_path
@@ -9692,7 +10085,7 @@ sys_lib_search_path_spec=$lt_sys_lib_search_path_spec
 sys_lib_dlsearch_path_spec=$lt_sys_lib_dlsearch_path_spec
 
 # Fix the shell variable \$srcfile for the compiler.
-fix_srcfile_path="$fix_srcfile_path"
+fix_srcfile_path=$lt_fix_srcfile_path
 
 # Set to yes if exported symbols are required.
 always_export_symbols=$always_export_symbols
@@ -9852,6 +10245,7 @@ postdep_objects_CXX=
 predeps_CXX=
 postdeps_CXX=
 compiler_lib_search_path_CXX=
+compiler_lib_search_dirs_CXX=
 
 # Source file extension for C++ test sources.
 ac_ext=cpp
@@ -9861,10 +10255,10 @@ objext=o
 objext_CXX=$objext
 
 # Code to be used in simple compile tests
-lt_simple_compile_test_code="int some_variable = 0;\n"
+lt_simple_compile_test_code="int some_variable = 0;"
 
 # Code to be used in simple link tests
-lt_simple_link_test_code='int main(int, char *[]) { return(0); }\n'
+lt_simple_link_test_code='int main(int, char *[]) { return(0); }'
 
 # ltmain only uses $CC for tagged configurations so make sure $CC is set.
 
@@ -9880,16 +10274,16 @@ compiler=$CC
 
 # save warnings/boilerplate of simple test code
 ac_outfile=conftest.$ac_objext
-printf "$lt_simple_compile_test_code" >conftest.$ac_ext
+echo "$lt_simple_compile_test_code" >conftest.$ac_ext
 eval "$ac_compile" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_compiler_boilerplate=`cat conftest.err`
 $rm conftest*
 
 ac_outfile=conftest.$ac_objext
-printf "$lt_simple_link_test_code" >conftest.$ac_ext
+echo "$lt_simple_link_test_code" >conftest.$ac_ext
 eval "$ac_link" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_linker_boilerplate=`cat conftest.err`
-$rm conftest*
+$rm -r conftest*
 
 
 # Allow CC to be a program name with arguments.
@@ -10096,7 +10490,7 @@ case $host_os in
     # FIXME: insert proper C++ library support
     ld_shlibs_CXX=no
     ;;
-  aix4* | aix5*)
+  aix[4-9]*)
     if test "$host_cpu" = ia64; then
       # On IA64, the linker does run time linking by default, so we don't
       # have to do anything special.
@@ -10109,7 +10503,7 @@ case $host_os in
       # Test if we are trying to use run time linking or normal
       # AIX style linking. If -brtl is somewhere in LDFLAGS, we
       # need to do runtime linking.
-      case $host_os in aix4.[23]|aix4.[23].*|aix5*)
+      case $host_os in aix4.[23]|aix4.[23].*|aix[5-9]*)
 	for ld_flag in $LDFLAGS; do
 	  case $ld_flag in
 	  *-brtl*)
@@ -10145,7 +10539,7 @@ case $host_os in
 	   strings "$collect2name" | grep resolve_lib_name >/dev/null
 	then
 	  # We have reworked collect2
-	  hardcode_direct_CXX=yes
+	  :
 	else
 	  # We have old collect2
 	  hardcode_direct_CXX=unsupported
@@ -10223,11 +10617,18 @@ if { (eval echo "$as_me:$LINENO: \"$ac_link\"") >&5
   echo "$as_me:$LINENO: \$? = $ac_status" >&5
   (exit $ac_status); }; }; then
 
-aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`
+lt_aix_libpath_sed='
+    /Import File Strings/,/^$/ {
+	/^0/ {
+	    s/^0  *\(.*\)$/\1/
+	    p
+	}
+    }'
+aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
 # Check for a 64-bit object if we didn't find anything.
-if test -z "$aix_libpath"; then aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`; fi
+if test -z "$aix_libpath"; then
+  aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
+fi
 else
   echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
@@ -10285,11 +10686,18 @@ if { (eval echo "$as_me:$LINENO: \"$ac_link\"") >&5
   echo "$as_me:$LINENO: \$? = $ac_status" >&5
   (exit $ac_status); }; }; then
 
-aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`
+lt_aix_libpath_sed='
+    /Import File Strings/,/^$/ {
+	/^0/ {
+	    s/^0  *\(.*\)$/\1/
+	    p
+	}
+    }'
+aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
 # Check for a 64-bit object if we didn't find anything.
-if test -z "$aix_libpath"; then aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`; fi
+if test -z "$aix_libpath"; then
+  aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
+fi
 else
   echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
@@ -10357,59 +10765,31 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
     fi
   ;;
       darwin* | rhapsody*)
-        case $host_os in
-        rhapsody* | darwin1.[012])
-         allow_undefined_flag_CXX='${wl}-undefined ${wl}suppress'
-         ;;
-       *) # Darwin 1.3 on
-         if test -z ${MACOSX_DEPLOYMENT_TARGET} ; then
-           allow_undefined_flag_CXX='${wl}-flat_namespace ${wl}-undefined ${wl}suppress'
-         else
-           case ${MACOSX_DEPLOYMENT_TARGET} in
-             10.[012])
-               allow_undefined_flag_CXX='${wl}-flat_namespace ${wl}-undefined ${wl}suppress'
-               ;;
-             10.*)
-               allow_undefined_flag_CXX='${wl}-undefined ${wl}dynamic_lookup'
-               ;;
-           esac
-         fi
-         ;;
-        esac
       archive_cmds_need_lc_CXX=no
       hardcode_direct_CXX=no
       hardcode_automatic_CXX=yes
       hardcode_shlibpath_var_CXX=unsupported
       whole_archive_flag_spec_CXX=''
       link_all_deplibs_CXX=yes
-
-    if test "$GXX" = yes ; then
-      lt_int_apple_cc_single_mod=no
+      allow_undefined_flag_CXX="$_lt_dar_allow_undefined"
+      if test "$GXX" = yes ; then
       output_verbose_link_cmd='echo'
-      if $CC -dumpspecs 2>&1 | $EGREP 'single_module' >/dev/null ; then
-       lt_int_apple_cc_single_mod=yes
+      archive_cmds_CXX="\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring $_lt_dar_single_mod${_lt_dsymutil}"
+      module_cmds_CXX="\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dsymutil}"
+      archive_expsym_cmds_CXX="sed 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring ${_lt_dar_single_mod}${_lt_dar_export_syms}${_lt_dsymutil}"
+      module_expsym_cmds_CXX="sed -e 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dar_export_syms}${_lt_dsymutil}"
+      if test "$lt_cv_apple_cc_single_mod" != "yes"; then
+        archive_cmds_CXX="\$CC -r -keep_private_externs -nostdlib -o \${lib}-master.o \$libobjs~\$CC -dynamiclib \$allow_undefined_flag -o \$lib \${lib}-master.o \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring${_lt_dsymutil}"
+        archive_expsym_cmds_CXX="sed 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC -r -keep_private_externs -nostdlib -o \${lib}-master.o \$libobjs~\$CC -dynamiclib \$allow_undefined_flag -o \$lib \${lib}-master.o \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring${_lt_dar_export_syms}${_lt_dsymutil}"
       fi
-      if test "X$lt_int_apple_cc_single_mod" = Xyes ; then
-       archive_cmds_CXX='$CC -dynamiclib -single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring'
-      else
-          archive_cmds_CXX='$CC -r -keep_private_externs -nostdlib -o ${lib}-master.o $libobjs~$CC -dynamiclib $allow_undefined_flag -o $lib ${lib}-master.o $deplibs $compiler_flags -install_name $rpath/$soname $verstring'
-        fi
-        module_cmds_CXX='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
-        # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-          if test "X$lt_int_apple_cc_single_mod" = Xyes ; then
-            archive_expsym_cmds_CXX='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -dynamiclib -single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
-          else
-            archive_expsym_cmds_CXX='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -r -keep_private_externs -nostdlib -o ${lib}-master.o $libobjs~$CC -dynamiclib $allow_undefined_flag -o $lib ${lib}-master.o $deplibs $compiler_flags -install_name $rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
-          fi
-            module_expsym_cmds_CXX='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
       else
       case $cc_basename in
         xlc*)
          output_verbose_link_cmd='echo'
-          archive_cmds_CXX='$CC -qmkshrobj ${wl}-single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $verstring'
+          archive_cmds_CXX='$CC -qmkshrobj ${wl}-single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $xlcverstring'
           module_cmds_CXX='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
           # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-          archive_expsym_cmds_CXX='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj ${wl}-single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
+          archive_expsym_cmds_CXX='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj ${wl}-single_module $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $xlcverstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           module_expsym_cmds_CXX='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           ;;
        *)
@@ -10443,7 +10823,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
   freebsd-elf*)
     archive_cmds_need_lc_CXX=no
     ;;
-  freebsd* | kfreebsd*-gnu | dragonfly*)
+  freebsd* | dragonfly*)
     # FreeBSD 3 and later use GNU C++ and GNU ld with standard ELF
     # conventions
     ld_shlibs_CXX=yes
@@ -10492,9 +10872,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       hardcode_libdir_separator_CXX=:
 
       case $host_cpu in
-      hppa*64*|ia64*)
-	hardcode_libdir_flag_spec_ld_CXX='+b $libdir'
-        ;;
+      hppa*64*|ia64*) ;;
       *)
 	export_dynamic_flag_spec_CXX='${wl}-E'
         ;;
@@ -10562,7 +10940,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 	;;
     esac
     ;;
-  interix3*)
+  interix[3-9]*)
     hardcode_direct_CXX=no
     hardcode_shlibpath_var_CXX=no
     hardcode_libdir_flag_spec_CXX='${wl}-rpath,$libdir'
@@ -10602,7 +10980,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
     hardcode_libdir_flag_spec_CXX='${wl}-rpath ${wl}$libdir'
     hardcode_libdir_separator_CXX=:
     ;;
-  linux*)
+  linux* | k*bsd*-gnu)
     case $cc_basename in
       KCC*)
 	# Kuck and Associates, Inc. (KAI) C++ Compiler
@@ -10654,7 +11032,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 	export_dynamic_flag_spec_CXX='${wl}--export-dynamic'
 	whole_archive_flag_spec_CXX='${wl}--whole-archive$convenience ${wl}--no-whole-archive'
 	;;
-      pgCC*)
+      pgCC* | pgcpp*)
         # Portland Group C++ compiler
 	archive_cmds_CXX='$CC -shared $pic_flag $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags ${wl}-soname ${wl}$soname -o $lib'
   	archive_expsym_cmds_CXX='$CC -shared $pic_flag $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags ${wl}-soname ${wl}$soname ${wl}-retain-symbols-file ${wl}$export_symbols -o $lib'
@@ -10682,6 +11060,29 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 	# dependencies.
 	output_verbose_link_cmd='templist=`$CC -shared $CFLAGS -v conftest.$objext 2>&1 | grep "ld"`; templist=`echo $templist | $SED "s/\(^.*ld.*\)\( .*ld .*$\)/\1/"`; list=""; for z in $templist; do case $z in conftest.$objext) list="$list $z";; *.$objext);; *) list="$list $z";;esac; done; echo $list'
 	;;
+      *)
+	case `$CC -V 2>&1 | sed 5q` in
+	*Sun\ C*)
+	  # Sun C++ 5.9
+	  no_undefined_flag_CXX=' -zdefs'
+	  archive_cmds_CXX='$CC -G${allow_undefined_flag} -h$soname -o $lib $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags'
+	  archive_expsym_cmds_CXX='$CC -G${allow_undefined_flag} -h$soname -o $lib $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags ${wl}-retain-symbols-file ${wl}$export_symbols'
+	  hardcode_libdir_flag_spec_CXX='-R$libdir'
+	  whole_archive_flag_spec_CXX='${wl}--whole-archive`new_convenience=; for conv in $convenience\"\"; do test -z \"$conv\" || new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}--no-whole-archive'
+
+	  # Not sure whether something based on
+	  # $CC $CFLAGS -v conftest.$objext -o libconftest$shared_ext 2>&1
+	  # would be better.
+	  output_verbose_link_cmd='echo'
+
+	  # Archives containing C++ object files must be created using
+	  # "CC -xar", where "CC" is the Sun C++ compiler.  This is
+	  # necessary to make sure instantiated templates are included
+	  # in the archive.
+	  old_archive_cmds_CXX='$CC -xar -o $oldlib $oldobjs'
+	  ;;
+	esac
+	;;
     esac
     ;;
   lynxos*)
@@ -10726,9 +11127,9 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       archive_cmds_CXX='$CC -shared $pic_flag $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags -o $lib'
       hardcode_libdir_flag_spec_CXX='${wl}-rpath,$libdir'
       if test -z "`echo __ELF__ | $CC -E - | grep __ELF__`" || test "$host_os-$host_cpu" = "openbsd2.8-powerpc"; then
-        archive_expsym_cmds_CXX='$CC -shared $pic_flag $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags ${wl}-retain-symbols-file,$export_symbols -o $lib'
-        export_dynamic_flag_spec_CXX='${wl}-E'
-        whole_archive_flag_spec_CXX="$wlarc"'--whole-archive$convenience '"$wlarc"'--no-whole-archive'
+	archive_expsym_cmds_CXX='$CC -shared $pic_flag $predep_objects $libobjs $deplibs $postdep_objects $compiler_flags ${wl}-retain-symbols-file,$export_symbols -o $lib'
+	export_dynamic_flag_spec_CXX='${wl}-E'
+	whole_archive_flag_spec_CXX="$wlarc"'--whole-archive$convenience '"$wlarc"'--no-whole-archive'
       fi
       output_verbose_link_cmd='echo'
     else
@@ -10895,15 +11296,10 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 	case $host_os in
 	  solaris2.[0-5] | solaris2.[0-5].*) ;;
 	  *)
-	    # The C++ compiler is used as linker so we must use $wl
-	    # flag to pass the commands to the underlying system
-	    # linker. We must also pass each convience library through
-	    # to the system linker between allextract/defaultextract.
-	    # The C++ compiler will combine linker options so we
-	    # cannot just pass the convience library names through
-	    # without $wl.
+	    # The compiler driver will combine and reorder linker options,
+	    # but understands `-z linker_flag'.
 	    # Supported since Solaris 2.6 (maybe 2.5.1?)
-	    whole_archive_flag_spec_CXX='${wl}-z ${wl}allextract`for conv in $convenience\"\"; do test -n \"$conv\" && new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}-z ${wl}defaultextract'
+	    whole_archive_flag_spec_CXX='-z allextract$convenience -z defaultextract'
 	    ;;
 	esac
 	link_all_deplibs_CXX=yes
@@ -10950,6 +11346,12 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 	  fi
 
 	  hardcode_libdir_flag_spec_CXX='${wl}-R $wl$libdir'
+	  case $host_os in
+	  solaris2.[0-5] | solaris2.[0-5].*) ;;
+	  *)
+	    whole_archive_flag_spec_CXX='${wl}-z ${wl}allextract$convenience ${wl}-z ${wl}defaultextract'
+	    ;;
+	  esac
 	fi
 	;;
     esac
@@ -11037,7 +11439,6 @@ test "$ld_shlibs_CXX" = no && can_build_shared=no
 GCC_CXX="$GXX"
 LD_CXX="$LD"
 
-
 cat > conftest.$ac_ext <<EOF
 class Foo
 {
@@ -11139,9 +11540,14 @@ fi
 
 $rm -f confest.$objext
 
+compiler_lib_search_dirs_CXX=
+if test -n "$compiler_lib_search_path_CXX"; then
+  compiler_lib_search_dirs_CXX=`echo " ${compiler_lib_search_path_CXX}" | ${SED} -e 's! -L! !g' -e 's!^ !!'`
+fi
+
 # PORTME: override above test on systems where it is broken
 case $host_os in
-interix3*)
+interix[3-9]*)
   # Interix 3.5 installs completely hosed .la files for C++, so rather than
   # hack all around it, let's just trust "g++" to DTRT.
   predep_objects_CXX=
@@ -11149,19 +11555,51 @@ interix3*)
   postdeps_CXX=
   ;;
 
+linux*)
+  case `$CC -V 2>&1 | sed 5q` in
+  *Sun\ C*)
+    # Sun C++ 5.9
+    #
+    # The more standards-conforming stlport4 library is
+    # incompatible with the Cstd library. Avoid specifying
+    # it if it's in CXXFLAGS. Ignore libCrun as
+    # -library=stlport4 depends on it.
+    case " $CXX $CXXFLAGS " in
+    *" -library=stlport4 "*)
+      solaris_use_stlport4=yes
+      ;;
+    esac
+    if test "$solaris_use_stlport4" != yes; then
+      postdeps_CXX='-library=Cstd -library=Crun'
+    fi
+    ;;
+  esac
+  ;;
+
 solaris*)
   case $cc_basename in
   CC*)
+    # The more standards-conforming stlport4 library is
+    # incompatible with the Cstd library. Avoid specifying
+    # it if it's in CXXFLAGS. Ignore libCrun as
+    # -library=stlport4 depends on it.
+    case " $CXX $CXXFLAGS " in
+    *" -library=stlport4 "*)
+      solaris_use_stlport4=yes
+      ;;
+    esac
+
     # Adding this requires a known-good setup of shared libraries for
     # Sun compiler versions before 5.6, else PIC objects from an old
     # archive will be linked into the output, leading to subtle bugs.
-    postdeps_CXX='-lCstd -lCrun'
+    if test "$solaris_use_stlport4" != yes; then
+      postdeps_CXX='-library=Cstd -library=Crun'
+    fi
     ;;
   esac
   ;;
 esac
 
-
 case " $postdeps_CXX " in
 *" -lc "*) archive_cmds_need_lc_CXX=no ;;
 esac
@@ -11192,12 +11630,14 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       # like `-m68040'.
       lt_prog_compiler_pic_CXX='-m68020 -resident32 -malways-restore-a4'
       ;;
-    beos* | cygwin* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
+    beos* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
       # PIC is the default for these OSes.
       ;;
-    mingw* | os2* | pw32*)
+    mingw* | cygwin* | os2* | pw32*)
       # This hack is so that the source file can tell whether it is being
       # built for inclusion in a dll (and should export symbols for example).
+      # Although the cygwin gcc ignores -fPIC, still need this for old-style
+      # (--disable-auto-import) libraries
       lt_prog_compiler_pic_CXX='-DDLL_EXPORT'
       ;;
     darwin* | rhapsody*)
@@ -11209,7 +11649,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       # DJGPP does not support shared libraries at all
       lt_prog_compiler_pic_CXX=
       ;;
-    interix3*)
+    interix[3-9]*)
       # Interix 3.x gcc -fpic/-fPIC options generate broken code.
       # Instead, we relocate shared libraries at runtime.
       ;;
@@ -11235,7 +11675,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
     esac
   else
     case $host_os in
-      aix4* | aix5*)
+      aix[4-9]*)
 	# All AIX code is PIC.
 	if test "$host_cpu" = ia64; then
 	  # AIX 5 now supports IA64 processor
@@ -11275,7 +11715,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
 	    ;;
 	esac
 	;;
-      freebsd* | kfreebsd*-gnu | dragonfly*)
+      freebsd* | dragonfly*)
 	# FreeBSD uses GNU C++
 	;;
       hpux9* | hpux10* | hpux11*)
@@ -11318,7 +11758,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
 	    ;;
 	esac
 	;;
-      linux*)
+      linux* | k*bsd*-gnu)
 	case $cc_basename in
 	  KCC*)
 	    # KAI C++ Compiler
@@ -11331,7 +11771,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
 	    lt_prog_compiler_pic_CXX='-KPIC'
 	    lt_prog_compiler_static_CXX='-static'
 	    ;;
-	  pgCC*)
+	  pgCC* | pgcpp*)
 	    # Portland Group C++ compiler.
 	    lt_prog_compiler_wl_CXX='-Wl,'
 	    lt_prog_compiler_pic_CXX='-fpic'
@@ -11345,6 +11785,14 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
 	    lt_prog_compiler_static_CXX='-non_shared'
 	    ;;
 	  *)
+	    case `$CC -V 2>&1 | sed 5q` in
+	    *Sun\ C*)
+	      # Sun C++ 5.9
+	      lt_prog_compiler_pic_CXX='-KPIC'
+	      lt_prog_compiler_static_CXX='-Bstatic'
+	      lt_prog_compiler_wl_CXX='-Qoption ld '
+	      ;;
+	    esac
 	    ;;
 	esac
 	;;
@@ -11454,12 +11902,12 @@ if test -n "$lt_prog_compiler_pic_CXX"; then
 
 echo "$as_me:$LINENO: checking if $compiler PIC flag $lt_prog_compiler_pic_CXX works" >&5
 echo $ECHO_N "checking if $compiler PIC flag $lt_prog_compiler_pic_CXX works... $ECHO_C" >&6
-if test "${lt_prog_compiler_pic_works_CXX+set}" = set; then
+if test "${lt_cv_prog_compiler_pic_works_CXX+set}" = set; then
   echo $ECHO_N "(cached) $ECHO_C" >&6
 else
-  lt_prog_compiler_pic_works_CXX=no
+  lt_cv_prog_compiler_pic_works_CXX=no
   ac_outfile=conftest.$ac_objext
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
    lt_compiler_flag="$lt_prog_compiler_pic_CXX -DPIC"
    # Insert the option either (1) after the last *FLAGS variable, or
    # (2) before a word containing "conftest.", or (3) at the end.
@@ -11470,27 +11918,27 @@ else
    -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \
    -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \
    -e 's:$: $lt_compiler_flag:'`
-   (eval echo "\"\$as_me:11473: $lt_compile\"" >&5)
+   (eval echo "\"\$as_me:11921: $lt_compile\"" >&5)
    (eval "$lt_compile" 2>conftest.err)
    ac_status=$?
    cat conftest.err >&5
-   echo "$as_me:11477: \$? = $ac_status" >&5
+   echo "$as_me:11925: \$? = $ac_status" >&5
    if (exit $ac_status) && test -s "$ac_outfile"; then
      # The compiler can only warn and ignore the option if not recognized
      # So say no if there are warnings other than the usual output.
      $echo "X$_lt_compiler_boilerplate" | $Xsed -e '/^$/d' >conftest.exp
      $SED '/^$/d; /^ *+/d' conftest.err >conftest.er2
      if test ! -s conftest.er2 || diff conftest.exp conftest.er2 >/dev/null; then
-       lt_prog_compiler_pic_works_CXX=yes
+       lt_cv_prog_compiler_pic_works_CXX=yes
      fi
    fi
    $rm conftest*
 
 fi
-echo "$as_me:$LINENO: result: $lt_prog_compiler_pic_works_CXX" >&5
-echo "${ECHO_T}$lt_prog_compiler_pic_works_CXX" >&6
+echo "$as_me:$LINENO: result: $lt_cv_prog_compiler_pic_works_CXX" >&5
+echo "${ECHO_T}$lt_cv_prog_compiler_pic_works_CXX" >&6
 
-if test x"$lt_prog_compiler_pic_works_CXX" = xyes; then
+if test x"$lt_cv_prog_compiler_pic_works_CXX" = xyes; then
     case $lt_prog_compiler_pic_CXX in
      "" | " "*) ;;
      *) lt_prog_compiler_pic_CXX=" $lt_prog_compiler_pic_CXX" ;;
@@ -11517,13 +11965,13 @@ esac
 wl=$lt_prog_compiler_wl_CXX eval lt_tmp_static_flag=\"$lt_prog_compiler_static_CXX\"
 echo "$as_me:$LINENO: checking if $compiler static flag $lt_tmp_static_flag works" >&5
 echo $ECHO_N "checking if $compiler static flag $lt_tmp_static_flag works... $ECHO_C" >&6
-if test "${lt_prog_compiler_static_works_CXX+set}" = set; then
+if test "${lt_cv_prog_compiler_static_works_CXX+set}" = set; then
   echo $ECHO_N "(cached) $ECHO_C" >&6
 else
-  lt_prog_compiler_static_works_CXX=no
+  lt_cv_prog_compiler_static_works_CXX=no
    save_LDFLAGS="$LDFLAGS"
    LDFLAGS="$LDFLAGS $lt_tmp_static_flag"
-   printf "$lt_simple_link_test_code" > conftest.$ac_ext
+   echo "$lt_simple_link_test_code" > conftest.$ac_ext
    if (eval $ac_link 2>conftest.err) && test -s conftest$ac_exeext; then
      # The linker can only warn and ignore the option if not recognized
      # So say no if there are warnings
@@ -11533,20 +11981,20 @@ else
        $echo "X$_lt_linker_boilerplate" | $Xsed -e '/^$/d' > conftest.exp
        $SED '/^$/d; /^ *+/d' conftest.err >conftest.er2
        if diff conftest.exp conftest.er2 >/dev/null; then
-         lt_prog_compiler_static_works_CXX=yes
+         lt_cv_prog_compiler_static_works_CXX=yes
        fi
      else
-       lt_prog_compiler_static_works_CXX=yes
+       lt_cv_prog_compiler_static_works_CXX=yes
      fi
    fi
-   $rm conftest*
+   $rm -r conftest*
    LDFLAGS="$save_LDFLAGS"
 
 fi
-echo "$as_me:$LINENO: result: $lt_prog_compiler_static_works_CXX" >&5
-echo "${ECHO_T}$lt_prog_compiler_static_works_CXX" >&6
+echo "$as_me:$LINENO: result: $lt_cv_prog_compiler_static_works_CXX" >&5
+echo "${ECHO_T}$lt_cv_prog_compiler_static_works_CXX" >&6
 
-if test x"$lt_prog_compiler_static_works_CXX" = xyes; then
+if test x"$lt_cv_prog_compiler_static_works_CXX" = xyes; then
     :
 else
     lt_prog_compiler_static_CXX=
@@ -11563,7 +12011,7 @@ else
    mkdir conftest
    cd conftest
    mkdir out
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
 
    lt_compiler_flag="-o out/conftest2.$ac_objext"
    # Insert the option either (1) after the last *FLAGS variable, or
@@ -11574,11 +12022,11 @@ else
    -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \
    -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \
    -e 's:$: $lt_compiler_flag:'`
-   (eval echo "\"\$as_me:11577: $lt_compile\"" >&5)
+   (eval echo "\"\$as_me:12025: $lt_compile\"" >&5)
    (eval "$lt_compile" 2>out/conftest.err)
    ac_status=$?
    cat out/conftest.err >&5
-   echo "$as_me:11581: \$? = $ac_status" >&5
+   echo "$as_me:12029: \$? = $ac_status" >&5
    if (exit $ac_status) && test -s out/conftest2.$ac_objext
    then
      # The compiler can only warn and ignore the option if not recognized
@@ -11631,7 +12079,7 @@ echo $ECHO_N "checking whether the $compiler linker ($LD) supports shared librar
 
   export_symbols_cmds_CXX='$NM $libobjs $convenience | $global_symbol_pipe | $SED '\''s/.* //'\'' | sort | uniq > $export_symbols'
   case $host_os in
-  aix4* | aix5*)
+  aix[4-9]*)
     # If we're using GNU nm, then we don't want the "-C" option.
     # -C means demangle to AIX nm, but means don't demangle with GNU nm
     if $NM -V 2>&1 | grep 'GNU' > /dev/null; then
@@ -11644,12 +12092,13 @@ echo $ECHO_N "checking whether the $compiler linker ($LD) supports shared librar
     export_symbols_cmds_CXX="$ltdll_cmds"
   ;;
   cygwin* | mingw*)
-    export_symbols_cmds_CXX='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[BCDGRS] /s/.* \([^ ]*\)/\1 DATA/;/^.* __nm__/s/^.* __nm__\([^ ]*\) [^ ]*/\1 DATA/;/^I /d;/^[AITW] /s/.* //'\'' | sort | uniq > $export_symbols'
+    export_symbols_cmds_CXX='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[BCDGRS][ ]/s/.*[ ]\([^ ]*\)/\1 DATA/;/^.*[ ]__nm__/s/^.*[ ]__nm__\([^ ]*\)[ ][^ ]*/\1 DATA/;/^I[ ]/d;/^[AITW][ ]/s/.*[ ]//'\'' | sort | uniq > $export_symbols'
   ;;
   *)
     export_symbols_cmds_CXX='$NM $libobjs $convenience | $global_symbol_pipe | $SED '\''s/.* //'\'' | sort | uniq > $export_symbols'
   ;;
   esac
+  exclude_expsyms_CXX='_GLOBAL_OFFSET_TABLE_|_GLOBAL__F[ID]_.*'
 
 echo "$as_me:$LINENO: result: $ld_shlibs_CXX" >&5
 echo "${ECHO_T}$ld_shlibs_CXX" >&6
@@ -11675,7 +12124,7 @@ x|xyes)
       echo "$as_me:$LINENO: checking whether -lc should be explicitly linked in" >&5
 echo $ECHO_N "checking whether -lc should be explicitly linked in... $ECHO_C" >&6
       $rm conftest*
-      printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+      echo "$lt_simple_compile_test_code" > conftest.$ac_ext
 
       if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5
   (eval $ac_compile) 2>&5
@@ -11733,20 +12182,7 @@ shlibpath_overrides_runpath=unknown
 version_type=none
 dynamic_linker="$host_os ld.so"
 sys_lib_dlsearch_path_spec="/lib /usr/lib"
-if test "$GCC" = yes; then
-  sys_lib_search_path_spec=`$CC -print-search-dirs | grep "^libraries:" | $SED -e "s/^libraries://" -e "s,=/,/,g"`
-  if echo "$sys_lib_search_path_spec" | grep ';' >/dev/null ; then
-    # if the path contains ";" then we assume it to be the separator
-    # otherwise default to the standard path separator (i.e. ":") - it is
-    # assumed that no part of a normal pathname contains ";" but that should
-    # okay in the real world where ";" in dirpaths is itself problematic.
-    sys_lib_search_path_spec=`echo "$sys_lib_search_path_spec" | $SED -e 's/;/ /g'`
-  else
-    sys_lib_search_path_spec=`echo "$sys_lib_search_path_spec" | $SED  -e "s/$PATH_SEPARATOR/ /g"`
-  fi
-else
-  sys_lib_search_path_spec="/lib /usr/lib /usr/local/lib"
-fi
+
 need_lib_prefix=unknown
 hardcode_into_libs=no
 
@@ -11764,7 +12200,7 @@ aix3*)
   soname_spec='${libname}${release}${shared_ext}$major'
   ;;
 
-aix4* | aix5*)
+aix[4-9]*)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -11903,12 +12339,7 @@ darwin* | rhapsody*)
   shlibpath_overrides_runpath=yes
   shlibpath_var=DYLD_LIBRARY_PATH
   shrext_cmds='`test .$module = .yes && echo .so || echo .dylib`'
-  # Apple's gcc prints 'gcc -print-search-dirs' doesn't operate the same.
-  if test "$GCC" = yes; then
-    sys_lib_search_path_spec=`$CC -print-search-dirs | tr "\n" "$PATH_SEPARATOR" | sed -e 's/libraries:/@libraries:/' | tr "@" "\n" | grep "^libraries:" | sed -e "s/^libraries://" -e "s,=/,/,g" -e "s,$PATH_SEPARATOR, ,g" -e "s,.*,& /lib /usr/lib /usr/local/lib,g"`
-  else
-    sys_lib_search_path_spec='/lib /usr/lib /usr/local/lib'
-  fi
+
   sys_lib_dlsearch_path_spec='/usr/local/lib /lib /usr/lib'
   ;;
 
@@ -11925,18 +12356,6 @@ freebsd1*)
   dynamic_linker=no
   ;;
 
-kfreebsd*-gnu)
-  version_type=linux
-  need_lib_prefix=no
-  need_version=no
-  library_names_spec='${libname}${release}${shared_ext}$versuffix ${libname}${release}${shared_ext}$major ${libname}${shared_ext}'
-  soname_spec='${libname}${release}${shared_ext}$major'
-  shlibpath_var=LD_LIBRARY_PATH
-  shlibpath_overrides_runpath=no
-  hardcode_into_libs=yes
-  dynamic_linker='GNU ld.so'
-  ;;
-
 freebsd* | dragonfly*)
   # DragonFly does not have aout.  When/if they implement a new
   # versioning mechanism, adjust this.
@@ -11974,7 +12393,7 @@ freebsd* | dragonfly*)
     shlibpath_overrides_runpath=no
     hardcode_into_libs=yes
     ;;
-  freebsd*) # from 4.6 on
+  *) # from 4.6 on, and DragonFly
     shlibpath_overrides_runpath=yes
     hardcode_into_libs=yes
     ;;
@@ -12037,7 +12456,7 @@ hpux9* | hpux10* | hpux11*)
   postinstall_cmds='chmod 555 $lib'
   ;;
 
-interix3*)
+interix[3-9]*)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -12092,7 +12511,7 @@ linux*oldld* | linux*aout* | linux*coff*)
   ;;
 
 # This must be Linux ELF.
-linux*)
+linux* | k*bsd*-gnu)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -12108,7 +12527,7 @@ linux*)
 
   # Append ld.so.conf contents to the search path
   if test -f /etc/ld.so.conf; then
-    lt_ld_extra=`awk '/^include / { system(sprintf("cd /etc; cat %s", \$2)); skip = 1; } { if (!skip) print \$0; skip = 0; }' < /etc/ld.so.conf | $SED -e 's/#.*//;s/[:,	]/ /g;s/=[^=]*$//;s/=[^= ]* / /g;/^$/d' | tr '\n' ' '`
+    lt_ld_extra=`awk '/^include / { system(sprintf("cd /etc; cat %s 2>/dev/null", \$2)); skip = 1; } { if (!skip) print \$0; skip = 0; }' < /etc/ld.so.conf | $SED -e 's/#.*//;/^[ 	]*hwcap[ 	]/d;s/[:,	]/ /g;s/=[^=]*$//;s/=[^= ]* / /g;/^$/d' | tr '\n' ' '`
     sys_lib_dlsearch_path_spec="/lib /usr/lib $lt_ld_extra"
   fi
 
@@ -12121,18 +12540,6 @@ linux*)
   dynamic_linker='GNU/Linux ld.so'
   ;;
 
-knetbsd*-gnu)
-  version_type=linux
-  need_lib_prefix=no
-  need_version=no
-  library_names_spec='${libname}${release}${shared_ext}$versuffix ${libname}${release}${shared_ext}$major ${libname}${shared_ext}'
-  soname_spec='${libname}${release}${shared_ext}$major'
-  shlibpath_var=LD_LIBRARY_PATH
-  shlibpath_overrides_runpath=no
-  hardcode_into_libs=yes
-  dynamic_linker='GNU ld.so'
-  ;;
-
 netbsd*)
   version_type=sunos
   need_lib_prefix=no
@@ -12214,6 +12621,10 @@ osf3* | osf4* | osf5*)
   sys_lib_dlsearch_path_spec="$sys_lib_search_path_spec"
   ;;
 
+rdos*)
+  dynamic_linker=no
+  ;;
+
 solaris*)
   version_type=linux
   need_lib_prefix=no
@@ -12310,6 +12721,21 @@ echo "$as_me:$LINENO: result: $dynamic_linker" >&5
 echo "${ECHO_T}$dynamic_linker" >&6
 test "$dynamic_linker" = no && can_build_shared=no
 
+if test "${lt_cv_sys_lib_search_path_spec+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  lt_cv_sys_lib_search_path_spec="$sys_lib_search_path_spec"
+fi
+
+sys_lib_search_path_spec="$lt_cv_sys_lib_search_path_spec"
+if test "${lt_cv_sys_lib_dlsearch_path_spec+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  lt_cv_sys_lib_dlsearch_path_spec="$sys_lib_dlsearch_path_spec"
+fi
+
+sys_lib_dlsearch_path_spec="$lt_cv_sys_lib_dlsearch_path_spec"
+
 variables_saved_for_relink="PATH $shlibpath_var $runpath_var"
 if test "$GCC" = yes; then
   variables_saved_for_relink="$variables_saved_for_relink GCC_EXEC_PREFIX COMPILER_PATH LIBRARY_PATH"
@@ -12393,6 +12819,7 @@ if test -f "$ltmain"; then
     predeps_CXX \
     postdeps_CXX \
     compiler_lib_search_path_CXX \
+    compiler_lib_search_dirs_CXX \
     archive_cmds_CXX \
     archive_expsym_cmds_CXX \
     postinstall_cmds_CXX \
@@ -12408,6 +12835,7 @@ if test -f "$ltmain"; then
     module_cmds_CXX \
     module_expsym_cmds_CXX \
     lt_cv_prog_compiler_c_o_CXX \
+    fix_srcfile_path_CXX \
     exclude_expsyms_CXX \
     include_expsyms_CXX; do
 
@@ -12640,6 +13068,10 @@ predeps=$lt_predeps_CXX
 # shared library.
 postdeps=$lt_postdeps_CXX
 
+# The directories searched by this compiler when creating a shared
+# library
+compiler_lib_search_dirs=$lt_compiler_lib_search_dirs_CXX
+
 # The library search path used internally by the compiler when linking
 # a shared library.
 compiler_lib_search_path=$lt_compiler_lib_search_path_CXX
@@ -12728,7 +13160,7 @@ sys_lib_search_path_spec=$lt_sys_lib_search_path_spec
 sys_lib_dlsearch_path_spec=$lt_sys_lib_dlsearch_path_spec
 
 # Fix the shell variable \$srcfile for the compiler.
-fix_srcfile_path="$fix_srcfile_path_CXX"
+fix_srcfile_path=$lt_fix_srcfile_path
 
 # Set to yes if exported symbols are required.
 always_export_symbols=$always_export_symbols_CXX
@@ -12819,10 +13251,17 @@ objext=o
 objext_F77=$objext
 
 # Code to be used in simple compile tests
-lt_simple_compile_test_code="      subroutine t\n      return\n      end\n"
+lt_simple_compile_test_code="\
+      subroutine t
+      return
+      end
+"
 
 # Code to be used in simple link tests
-lt_simple_link_test_code="      program t\n      end\n"
+lt_simple_link_test_code="\
+      program t
+      end
+"
 
 # ltmain only uses $CC for tagged configurations so make sure $CC is set.
 
@@ -12838,16 +13277,16 @@ compiler=$CC
 
 # save warnings/boilerplate of simple test code
 ac_outfile=conftest.$ac_objext
-printf "$lt_simple_compile_test_code" >conftest.$ac_ext
+echo "$lt_simple_compile_test_code" >conftest.$ac_ext
 eval "$ac_compile" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_compiler_boilerplate=`cat conftest.err`
 $rm conftest*
 
 ac_outfile=conftest.$ac_objext
-printf "$lt_simple_link_test_code" >conftest.$ac_ext
+echo "$lt_simple_link_test_code" >conftest.$ac_ext
 eval "$ac_link" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_linker_boilerplate=`cat conftest.err`
-$rm conftest*
+$rm -r conftest*
 
 
 # Allow CC to be a program name with arguments.
@@ -12885,7 +13324,7 @@ aix3*)
     postinstall_cmds='$RANLIB $lib'
   fi
   ;;
-aix4* | aix5*)
+aix[4-9]*)
   if test "$host_cpu" != ia64 && test "$aix_use_runtimelinking" = no ; then
     test "$enable_shared" = yes && enable_static=no
   fi
@@ -12931,13 +13370,15 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_pic_F77='-m68020 -resident32 -malways-restore-a4'
       ;;
 
-    beos* | cygwin* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
+    beos* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
       # PIC is the default for these OSes.
       ;;
 
-    mingw* | pw32* | os2*)
+    mingw* | cygwin* | pw32* | os2*)
       # This hack is so that the source file can tell whether it is being
       # built for inclusion in a dll (and should export symbols for example).
+      # Although the cygwin gcc ignores -fPIC, still need this for old-style
+      # (--disable-auto-import) libraries
       lt_prog_compiler_pic_F77='-DDLL_EXPORT'
       ;;
 
@@ -12947,7 +13388,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_pic_F77='-fno-common'
       ;;
 
-    interix3*)
+    interix[3-9]*)
       # Interix 3.x gcc -fpic/-fPIC options generate broken code.
       # Instead, we relocate shared libraries at runtime.
       ;;
@@ -13005,7 +13446,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
        esac
        ;;
 
-    mingw* | pw32* | os2*)
+    mingw* | cygwin* | pw32* | os2*)
       # This hack is so that the source file can tell whether it is being
       # built for inclusion in a dll (and should export symbols for example).
       lt_prog_compiler_pic_F77='-DDLL_EXPORT'
@@ -13038,7 +13479,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_static_F77='-Bstatic'
       ;;
 
-    linux*)
+    linux* | k*bsd*-gnu)
       case $cc_basename in
       icc* | ecc*)
 	lt_prog_compiler_wl_F77='-Wl,'
@@ -13057,6 +13498,22 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
         # All Alpha code is PIC.
         lt_prog_compiler_static_F77='-non_shared'
         ;;
+      *)
+        case `$CC -V 2>&1 | sed 5q` in
+	*Sun\ C*)
+	  # Sun C 5.9
+	  lt_prog_compiler_pic_F77='-KPIC'
+	  lt_prog_compiler_static_F77='-Bstatic'
+	  lt_prog_compiler_wl_F77='-Wl,'
+	  ;;
+	*Sun\ F*)
+	  # Sun Fortran 8.3 passes all unrecognized flags to the linker
+	  lt_prog_compiler_pic_F77='-KPIC'
+	  lt_prog_compiler_static_F77='-Bstatic'
+	  lt_prog_compiler_wl_F77=''
+	  ;;
+	esac
+	;;
       esac
       ;;
 
@@ -13066,6 +13523,10 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_static_F77='-non_shared'
       ;;
 
+    rdos*)
+      lt_prog_compiler_static_F77='-non_shared'
+      ;;
+
     solaris*)
       lt_prog_compiler_pic_F77='-KPIC'
       lt_prog_compiler_static_F77='-Bstatic'
@@ -13128,12 +13589,12 @@ if test -n "$lt_prog_compiler_pic_F77"; then
 
 echo "$as_me:$LINENO: checking if $compiler PIC flag $lt_prog_compiler_pic_F77 works" >&5
 echo $ECHO_N "checking if $compiler PIC flag $lt_prog_compiler_pic_F77 works... $ECHO_C" >&6
-if test "${lt_prog_compiler_pic_works_F77+set}" = set; then
+if test "${lt_cv_prog_compiler_pic_works_F77+set}" = set; then
   echo $ECHO_N "(cached) $ECHO_C" >&6
 else
-  lt_prog_compiler_pic_works_F77=no
+  lt_cv_prog_compiler_pic_works_F77=no
   ac_outfile=conftest.$ac_objext
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
    lt_compiler_flag="$lt_prog_compiler_pic_F77"
    # Insert the option either (1) after the last *FLAGS variable, or
    # (2) before a word containing "conftest.", or (3) at the end.
@@ -13144,27 +13605,27 @@ else
    -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \
    -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \
    -e 's:$: $lt_compiler_flag:'`
-   (eval echo "\"\$as_me:13147: $lt_compile\"" >&5)
+   (eval echo "\"\$as_me:13608: $lt_compile\"" >&5)
    (eval "$lt_compile" 2>conftest.err)
    ac_status=$?
    cat conftest.err >&5
-   echo "$as_me:13151: \$? = $ac_status" >&5
+   echo "$as_me:13612: \$? = $ac_status" >&5
    if (exit $ac_status) && test -s "$ac_outfile"; then
      # The compiler can only warn and ignore the option if not recognized
      # So say no if there are warnings other than the usual output.
      $echo "X$_lt_compiler_boilerplate" | $Xsed -e '/^$/d' >conftest.exp
      $SED '/^$/d; /^ *+/d' conftest.err >conftest.er2
      if test ! -s conftest.er2 || diff conftest.exp conftest.er2 >/dev/null; then
-       lt_prog_compiler_pic_works_F77=yes
+       lt_cv_prog_compiler_pic_works_F77=yes
      fi
    fi
    $rm conftest*
 
 fi
-echo "$as_me:$LINENO: result: $lt_prog_compiler_pic_works_F77" >&5
-echo "${ECHO_T}$lt_prog_compiler_pic_works_F77" >&6
+echo "$as_me:$LINENO: result: $lt_cv_prog_compiler_pic_works_F77" >&5
+echo "${ECHO_T}$lt_cv_prog_compiler_pic_works_F77" >&6
 
-if test x"$lt_prog_compiler_pic_works_F77" = xyes; then
+if test x"$lt_cv_prog_compiler_pic_works_F77" = xyes; then
     case $lt_prog_compiler_pic_F77 in
      "" | " "*) ;;
      *) lt_prog_compiler_pic_F77=" $lt_prog_compiler_pic_F77" ;;
@@ -13191,13 +13652,13 @@ esac
 wl=$lt_prog_compiler_wl_F77 eval lt_tmp_static_flag=\"$lt_prog_compiler_static_F77\"
 echo "$as_me:$LINENO: checking if $compiler static flag $lt_tmp_static_flag works" >&5
 echo $ECHO_N "checking if $compiler static flag $lt_tmp_static_flag works... $ECHO_C" >&6
-if test "${lt_prog_compiler_static_works_F77+set}" = set; then
+if test "${lt_cv_prog_compiler_static_works_F77+set}" = set; then
   echo $ECHO_N "(cached) $ECHO_C" >&6
 else
-  lt_prog_compiler_static_works_F77=no
+  lt_cv_prog_compiler_static_works_F77=no
    save_LDFLAGS="$LDFLAGS"
    LDFLAGS="$LDFLAGS $lt_tmp_static_flag"
-   printf "$lt_simple_link_test_code" > conftest.$ac_ext
+   echo "$lt_simple_link_test_code" > conftest.$ac_ext
    if (eval $ac_link 2>conftest.err) && test -s conftest$ac_exeext; then
      # The linker can only warn and ignore the option if not recognized
      # So say no if there are warnings
@@ -13207,20 +13668,20 @@ else
        $echo "X$_lt_linker_boilerplate" | $Xsed -e '/^$/d' > conftest.exp
        $SED '/^$/d; /^ *+/d' conftest.err >conftest.er2
        if diff conftest.exp conftest.er2 >/dev/null; then
-         lt_prog_compiler_static_works_F77=yes
+         lt_cv_prog_compiler_static_works_F77=yes
        fi
      else
-       lt_prog_compiler_static_works_F77=yes
+       lt_cv_prog_compiler_static_works_F77=yes
      fi
    fi
-   $rm conftest*
+   $rm -r conftest*
    LDFLAGS="$save_LDFLAGS"
 
 fi
-echo "$as_me:$LINENO: result: $lt_prog_compiler_static_works_F77" >&5
-echo "${ECHO_T}$lt_prog_compiler_static_works_F77" >&6
+echo "$as_me:$LINENO: result: $lt_cv_prog_compiler_static_works_F77" >&5
+echo "${ECHO_T}$lt_cv_prog_compiler_static_works_F77" >&6
 
-if test x"$lt_prog_compiler_static_works_F77" = xyes; then
+if test x"$lt_cv_prog_compiler_static_works_F77" = xyes; then
     :
 else
     lt_prog_compiler_static_F77=
@@ -13237,7 +13698,7 @@ else
    mkdir conftest
    cd conftest
    mkdir out
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
 
    lt_compiler_flag="-o out/conftest2.$ac_objext"
    # Insert the option either (1) after the last *FLAGS variable, or
@@ -13248,11 +13709,11 @@ else
    -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \
    -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \
    -e 's:$: $lt_compiler_flag:'`
-   (eval echo "\"\$as_me:13251: $lt_compile\"" >&5)
+   (eval echo "\"\$as_me:13712: $lt_compile\"" >&5)
    (eval "$lt_compile" 2>out/conftest.err)
    ac_status=$?
    cat out/conftest.err >&5
-   echo "$as_me:13255: \$? = $ac_status" >&5
+   echo "$as_me:13716: \$? = $ac_status" >&5
    if (exit $ac_status) && test -s out/conftest2.$ac_objext
    then
      # The compiler can only warn and ignore the option if not recognized
@@ -13332,12 +13793,13 @@ echo $ECHO_N "checking whether the $compiler linker ($LD) supports shared librar
   # it will be wrapped by ` (' and `)$', so one must not match beginning or
   # end of line.  Example: `a|bc|.*d.*' will exclude the symbols `a' and `bc',
   # as well as any symbol that contains `d'.
-  exclude_expsyms_F77="_GLOBAL_OFFSET_TABLE_"
+  exclude_expsyms_F77='_GLOBAL_OFFSET_TABLE_|_GLOBAL__F[ID]_.*'
   # Although _GLOBAL_OFFSET_TABLE_ is a valid symbol C name, most a.out
   # platforms (ab)use it in PIC code, but their linkers get confused if
   # the symbol is explicitly referenced.  Since portable code cannot
   # rely on this symbol name, it's probably fine to never include it in
   # preloaded symbol tables.
+  # Exclude shared library initialization/finalization symbols.
   extract_expsyms_cmds=
   # Just being paranoid about ensuring that cc_basename is set.
   for cc_temp in $compiler""; do
@@ -13396,7 +13858,7 @@ cc_basename=`$echo "X$cc_temp" | $Xsed -e 's%.*/%%' -e "s%^$host_alias-%%"`
 
     # See if GNU ld supports shared libraries.
     case $host_os in
-    aix3* | aix4* | aix5*)
+    aix[3-9]*)
       # On AIX/PPC, the GNU linker is very broken
       if test "$host_cpu" != ia64; then
 	ld_shlibs_F77=no
@@ -13444,7 +13906,7 @@ EOF
       allow_undefined_flag_F77=unsupported
       always_export_symbols_F77=no
       enable_shared_with_static_runtimes_F77=yes
-      export_symbols_cmds_F77='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[BCDGRS] /s/.* \([^ ]*\)/\1 DATA/'\'' | $SED -e '\''/^[AITW] /s/.* //'\'' | sort | uniq > $export_symbols'
+      export_symbols_cmds_F77='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[BCDGRS][ ]/s/.*[ ]\([^ ]*\)/\1 DATA/'\'' -e '\''/^[AITW][ ]/s/.*[ ]//'\'' | sort | uniq > $export_symbols'
 
       if $LD --help 2>&1 | grep 'auto-import' > /dev/null; then
         archive_cmds_F77='$CC -shared $libobjs $deplibs $compiler_flags -o $output_objdir/$soname ${wl}--enable-auto-image-base -Xlinker --out-implib -Xlinker $lib'
@@ -13462,7 +13924,7 @@ EOF
       fi
       ;;
 
-    interix3*)
+    interix[3-9]*)
       hardcode_direct_F77=no
       hardcode_shlibpath_var_F77=no
       hardcode_libdir_flag_spec_F77='${wl}-rpath,$libdir'
@@ -13477,7 +13939,7 @@ EOF
       archive_expsym_cmds_F77='sed "s,^,_," $export_symbols >$output_objdir/$soname.expsym~$CC -shared $pic_flag $libobjs $deplibs $compiler_flags ${wl}-h,$soname ${wl}--retain-symbols-file,$output_objdir/$soname.expsym ${wl}--image-base,`expr ${RANDOM-$$} % 4096 / 2 \* 262144 + 1342177280` -o $lib'
       ;;
 
-    linux*)
+    gnu* | linux* | k*bsd*-gnu)
       if $LD --help 2>&1 | grep ': supported targets:.* elf' > /dev/null; then
 	tmp_addflag=
 	case $cc_basename,$host_cpu in
@@ -13495,13 +13957,22 @@ EOF
 	ifc* | ifort*)			# Intel Fortran compiler
 	  tmp_addflag=' -nofor_main' ;;
 	esac
-	archive_cmds_F77='$CC -shared'"$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname -o $lib'
+	case `$CC -V 2>&1 | sed 5q` in
+	*Sun\ C*)			# Sun C 5.9
+	  whole_archive_flag_spec_F77='${wl}--whole-archive`new_convenience=; for conv in $convenience\"\"; do test -z \"$conv\" || new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}--no-whole-archive'
+	  tmp_sharedflag='-G' ;;
+	*Sun\ F*)			# Sun Fortran 8.3
+	  tmp_sharedflag='-G' ;;
+	*)
+	  tmp_sharedflag='-shared' ;;
+	esac
+	archive_cmds_F77='$CC '"$tmp_sharedflag""$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname -o $lib'
 
 	if test $supports_anon_versioning = yes; then
 	  archive_expsym_cmds_F77='$echo "{ global:" > $output_objdir/$libname.ver~
   cat $export_symbols | sed -e "s/\(.*\)/\1;/" >> $output_objdir/$libname.ver~
   $echo "local: *; };" >> $output_objdir/$libname.ver~
-	  $CC -shared'"$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname ${wl}-version-script ${wl}$output_objdir/$libname.ver -o $lib'
+	  $CC '"$tmp_sharedflag""$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname ${wl}-version-script ${wl}$output_objdir/$libname.ver -o $lib'
 	fi
       else
 	ld_shlibs_F77=no
@@ -13606,7 +14077,7 @@ _LT_EOF
       fi
       ;;
 
-    aix4* | aix5*)
+    aix[4-9]*)
       if test "$host_cpu" = ia64; then
 	# On IA64, the linker does run time linking by default, so we don't
 	# have to do anything special.
@@ -13626,7 +14097,7 @@ _LT_EOF
 	# Test if we are trying to use run time linking or normal
 	# AIX style linking. If -brtl is somewhere in LDFLAGS, we
 	# need to do runtime linking.
-	case $host_os in aix4.[23]|aix4.[23].*|aix5*)
+	case $host_os in aix4.[23]|aix4.[23].*|aix[5-9]*)
 	  for ld_flag in $LDFLAGS; do
   	  if (test $ld_flag = "-brtl" || test $ld_flag = "-Wl,-brtl"); then
   	    aix_use_runtimelinking=yes
@@ -13660,7 +14131,7 @@ _LT_EOF
   	   strings "$collect2name" | grep resolve_lib_name >/dev/null
 	  then
   	  # We have reworked collect2
-  	  hardcode_direct_F77=yes
+  	  :
 	  else
   	  # We have old collect2
   	  hardcode_direct_F77=unsupported
@@ -13728,11 +14199,18 @@ if { (eval echo "$as_me:$LINENO: \"$ac_link\"") >&5
   echo "$as_me:$LINENO: \$? = $ac_status" >&5
   (exit $ac_status); }; }; then
 
-aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`
+lt_aix_libpath_sed='
+    /Import File Strings/,/^$/ {
+	/^0/ {
+	    s/^0  *\(.*\)$/\1/
+	    p
+	}
+    }'
+aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
 # Check for a 64-bit object if we didn't find anything.
-if test -z "$aix_libpath"; then aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`; fi
+if test -z "$aix_libpath"; then
+  aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
+fi
 else
   echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
@@ -13779,11 +14257,18 @@ if { (eval echo "$as_me:$LINENO: \"$ac_link\"") >&5
   echo "$as_me:$LINENO: \$? = $ac_status" >&5
   (exit $ac_status); }; }; then
 
-aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`
+lt_aix_libpath_sed='
+    /Import File Strings/,/^$/ {
+	/^0/ {
+	    s/^0  *\(.*\)$/\1/
+	    p
+	}
+    }'
+aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
 # Check for a 64-bit object if we didn't find anything.
-if test -z "$aix_libpath"; then aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`; fi
+if test -z "$aix_libpath"; then
+  aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
+fi
 else
   echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
@@ -13835,7 +14320,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       # The linker will automatically build a .lib file if we build a DLL.
       old_archive_From_new_cmds_F77='true'
       # FIXME: Should let the user specify the lib program.
-      old_archive_cmds_F77='lib /OUT:$oldlib$oldobjs$old_deplibs'
+      old_archive_cmds_F77='lib -OUT:$oldlib$oldobjs$old_deplibs'
       fix_srcfile_path_F77='`cygpath -w "$srcfile"`'
       enable_shared_with_static_runtimes_F77=yes
       ;;
@@ -13868,19 +14353,18 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       link_all_deplibs_F77=yes
     if test "$GCC" = yes ; then
     	output_verbose_link_cmd='echo'
-        archive_cmds_F77='$CC -dynamiclib $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring'
-      module_cmds_F77='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
-      # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-      archive_expsym_cmds_F77='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -dynamiclib $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
-      module_expsym_cmds_F77='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
+        archive_cmds_F77="\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring $_lt_dar_single_mod${_lt_dsymutil}"
+        module_cmds_F77="\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dsymutil}"
+        archive_expsym_cmds_F77="sed 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring ${_lt_dar_single_mod}${_lt_dar_export_syms}${_lt_dsymutil}"
+        module_expsym_cmds_F77="sed -e 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dar_export_syms}${_lt_dsymutil}"
     else
       case $cc_basename in
         xlc*)
          output_verbose_link_cmd='echo'
-         archive_cmds_F77='$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $verstring'
+         archive_cmds_F77='$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $xlcverstring'
          module_cmds_F77='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
           # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-         archive_expsym_cmds_F77='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
+         archive_expsym_cmds_F77='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $xlcverstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           module_expsym_cmds_F77='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           ;;
        *)
@@ -13920,7 +14404,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       ;;
 
     # FreeBSD 3 and greater uses gcc -shared to do shared libraries.
-    freebsd* | kfreebsd*-gnu | dragonfly*)
+    freebsd* | dragonfly*)
       archive_cmds_F77='$CC -shared -o $lib $libobjs $deplibs $compiler_flags'
       hardcode_libdir_flag_spec_F77='-R$libdir'
       hardcode_direct_F77=yes
@@ -14043,15 +14527,15 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 
     openbsd*)
       if test -f /usr/libexec/ld.so; then
-        hardcode_direct_F77=yes
-        hardcode_shlibpath_var_F77=no
-        if test -z "`echo __ELF__ | $CC -E - | grep __ELF__`" || test "$host_os-$host_cpu" = "openbsd2.8-powerpc"; then
+	hardcode_direct_F77=yes
+	hardcode_shlibpath_var_F77=no
+	if test -z "`echo __ELF__ | $CC -E - | grep __ELF__`" || test "$host_os-$host_cpu" = "openbsd2.8-powerpc"; then
 	  archive_cmds_F77='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags'
 	  archive_expsym_cmds_F77='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-retain-symbols-file,$export_symbols'
 	  hardcode_libdir_flag_spec_F77='${wl}-rpath,$libdir'
 	  export_dynamic_flag_spec_F77='${wl}-E'
-        else
-         case $host_os in
+	else
+	  case $host_os in
 	   openbsd[01].* | openbsd2.[0-7] | openbsd2.[0-7].*)
 	     archive_cmds_F77='$LD -Bshareable -o $lib $libobjs $deplibs $linker_flags'
 	     hardcode_libdir_flag_spec_F77='-R$libdir'
@@ -14060,10 +14544,10 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 	     archive_cmds_F77='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags'
 	     hardcode_libdir_flag_spec_F77='${wl}-rpath,$libdir'
 	     ;;
-         esac
+	  esac
         fi
       else
-        ld_shlibs_F77=no
+	ld_shlibs_F77=no
       fi
       ;;
 
@@ -14122,17 +14606,16 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       case $host_os in
       solaris2.[0-5] | solaris2.[0-5].*) ;;
       *)
- 	# The compiler driver will combine linker options so we
- 	# cannot just pass the convience library names through
- 	# without $wl, iff we do not link with $LD.
- 	# Luckily, gcc supports the same syntax we need for Sun Studio.
+	# The compiler driver will combine and reorder linker options,
+	# but understands `-z linker_flag'.  GCC discards it without `$wl',
+	# but is careful enough not to reorder.
  	# Supported since Solaris 2.6 (maybe 2.5.1?)
- 	case $wlarc in
- 	'')
- 	  whole_archive_flag_spec_F77='-z allextract$convenience -z defaultextract' ;;
- 	*)
- 	  whole_archive_flag_spec_F77='${wl}-z ${wl}allextract`for conv in $convenience\"\"; do test -n \"$conv\" && new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}-z ${wl}defaultextract' ;;
- 	esac ;;
+	if test "$GCC" = yes; then
+	  whole_archive_flag_spec_F77='${wl}-z ${wl}allextract$convenience ${wl}-z ${wl}defaultextract'
+	else
+	  whole_archive_flag_spec_F77='-z allextract$convenience -z defaultextract'
+	fi
+	;;
       esac
       link_all_deplibs_F77=yes
       ;;
@@ -14189,7 +14672,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       fi
       ;;
 
-    sysv4*uw2* | sysv5OpenUNIX* | sysv5UnixWare7.[01].[10]* | unixware7*)
+    sysv4*uw2* | sysv5OpenUNIX* | sysv5UnixWare7.[01].[10]* | unixware7* | sco3.2v5.0.[024]*)
       no_undefined_flag_F77='${wl}-z,text'
       archive_cmds_need_lc_F77=no
       hardcode_shlibpath_var_F77=no
@@ -14266,7 +14749,7 @@ x|xyes)
       echo "$as_me:$LINENO: checking whether -lc should be explicitly linked in" >&5
 echo $ECHO_N "checking whether -lc should be explicitly linked in... $ECHO_C" >&6
       $rm conftest*
-      printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+      echo "$lt_simple_compile_test_code" > conftest.$ac_ext
 
       if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5
   (eval $ac_compile) 2>&5
@@ -14324,20 +14807,7 @@ shlibpath_overrides_runpath=unknown
 version_type=none
 dynamic_linker="$host_os ld.so"
 sys_lib_dlsearch_path_spec="/lib /usr/lib"
-if test "$GCC" = yes; then
-  sys_lib_search_path_spec=`$CC -print-search-dirs | grep "^libraries:" | $SED -e "s/^libraries://" -e "s,=/,/,g"`
-  if echo "$sys_lib_search_path_spec" | grep ';' >/dev/null ; then
-    # if the path contains ";" then we assume it to be the separator
-    # otherwise default to the standard path separator (i.e. ":") - it is
-    # assumed that no part of a normal pathname contains ";" but that should
-    # okay in the real world where ";" in dirpaths is itself problematic.
-    sys_lib_search_path_spec=`echo "$sys_lib_search_path_spec" | $SED -e 's/;/ /g'`
-  else
-    sys_lib_search_path_spec=`echo "$sys_lib_search_path_spec" | $SED  -e "s/$PATH_SEPARATOR/ /g"`
-  fi
-else
-  sys_lib_search_path_spec="/lib /usr/lib /usr/local/lib"
-fi
+
 need_lib_prefix=unknown
 hardcode_into_libs=no
 
@@ -14355,7 +14825,7 @@ aix3*)
   soname_spec='${libname}${release}${shared_ext}$major'
   ;;
 
-aix4* | aix5*)
+aix[4-9]*)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -14494,12 +14964,7 @@ darwin* | rhapsody*)
   shlibpath_overrides_runpath=yes
   shlibpath_var=DYLD_LIBRARY_PATH
   shrext_cmds='`test .$module = .yes && echo .so || echo .dylib`'
-  # Apple's gcc prints 'gcc -print-search-dirs' doesn't operate the same.
-  if test "$GCC" = yes; then
-    sys_lib_search_path_spec=`$CC -print-search-dirs | tr "\n" "$PATH_SEPARATOR" | sed -e 's/libraries:/@libraries:/' | tr "@" "\n" | grep "^libraries:" | sed -e "s/^libraries://" -e "s,=/,/,g" -e "s,$PATH_SEPARATOR, ,g" -e "s,.*,& /lib /usr/lib /usr/local/lib,g"`
-  else
-    sys_lib_search_path_spec='/lib /usr/lib /usr/local/lib'
-  fi
+
   sys_lib_dlsearch_path_spec='/usr/local/lib /lib /usr/lib'
   ;;
 
@@ -14516,18 +14981,6 @@ freebsd1*)
   dynamic_linker=no
   ;;
 
-kfreebsd*-gnu)
-  version_type=linux
-  need_lib_prefix=no
-  need_version=no
-  library_names_spec='${libname}${release}${shared_ext}$versuffix ${libname}${release}${shared_ext}$major ${libname}${shared_ext}'
-  soname_spec='${libname}${release}${shared_ext}$major'
-  shlibpath_var=LD_LIBRARY_PATH
-  shlibpath_overrides_runpath=no
-  hardcode_into_libs=yes
-  dynamic_linker='GNU ld.so'
-  ;;
-
 freebsd* | dragonfly*)
   # DragonFly does not have aout.  When/if they implement a new
   # versioning mechanism, adjust this.
@@ -14565,7 +15018,7 @@ freebsd* | dragonfly*)
     shlibpath_overrides_runpath=no
     hardcode_into_libs=yes
     ;;
-  freebsd*) # from 4.6 on
+  *) # from 4.6 on, and DragonFly
     shlibpath_overrides_runpath=yes
     hardcode_into_libs=yes
     ;;
@@ -14628,7 +15081,7 @@ hpux9* | hpux10* | hpux11*)
   postinstall_cmds='chmod 555 $lib'
   ;;
 
-interix3*)
+interix[3-9]*)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -14683,7 +15136,7 @@ linux*oldld* | linux*aout* | linux*coff*)
   ;;
 
 # This must be Linux ELF.
-linux*)
+linux* | k*bsd*-gnu)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -14699,7 +15152,7 @@ linux*)
 
   # Append ld.so.conf contents to the search path
   if test -f /etc/ld.so.conf; then
-    lt_ld_extra=`awk '/^include / { system(sprintf("cd /etc; cat %s", \$2)); skip = 1; } { if (!skip) print \$0; skip = 0; }' < /etc/ld.so.conf | $SED -e 's/#.*//;s/[:,	]/ /g;s/=[^=]*$//;s/=[^= ]* / /g;/^$/d' | tr '\n' ' '`
+    lt_ld_extra=`awk '/^include / { system(sprintf("cd /etc; cat %s 2>/dev/null", \$2)); skip = 1; } { if (!skip) print \$0; skip = 0; }' < /etc/ld.so.conf | $SED -e 's/#.*//;/^[ 	]*hwcap[ 	]/d;s/[:,	]/ /g;s/=[^=]*$//;s/=[^= ]* / /g;/^$/d' | tr '\n' ' '`
     sys_lib_dlsearch_path_spec="/lib /usr/lib $lt_ld_extra"
   fi
 
@@ -14712,18 +15165,6 @@ linux*)
   dynamic_linker='GNU/Linux ld.so'
   ;;
 
-knetbsd*-gnu)
-  version_type=linux
-  need_lib_prefix=no
-  need_version=no
-  library_names_spec='${libname}${release}${shared_ext}$versuffix ${libname}${release}${shared_ext}$major ${libname}${shared_ext}'
-  soname_spec='${libname}${release}${shared_ext}$major'
-  shlibpath_var=LD_LIBRARY_PATH
-  shlibpath_overrides_runpath=no
-  hardcode_into_libs=yes
-  dynamic_linker='GNU ld.so'
-  ;;
-
 netbsd*)
   version_type=sunos
   need_lib_prefix=no
@@ -14805,6 +15246,10 @@ osf3* | osf4* | osf5*)
   sys_lib_dlsearch_path_spec="$sys_lib_search_path_spec"
   ;;
 
+rdos*)
+  dynamic_linker=no
+  ;;
+
 solaris*)
   version_type=linux
   need_lib_prefix=no
@@ -14901,6 +15346,21 @@ echo "$as_me:$LINENO: result: $dynamic_linker" >&5
 echo "${ECHO_T}$dynamic_linker" >&6
 test "$dynamic_linker" = no && can_build_shared=no
 
+if test "${lt_cv_sys_lib_search_path_spec+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  lt_cv_sys_lib_search_path_spec="$sys_lib_search_path_spec"
+fi
+
+sys_lib_search_path_spec="$lt_cv_sys_lib_search_path_spec"
+if test "${lt_cv_sys_lib_dlsearch_path_spec+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  lt_cv_sys_lib_dlsearch_path_spec="$sys_lib_dlsearch_path_spec"
+fi
+
+sys_lib_dlsearch_path_spec="$lt_cv_sys_lib_dlsearch_path_spec"
+
 variables_saved_for_relink="PATH $shlibpath_var $runpath_var"
 if test "$GCC" = yes; then
   variables_saved_for_relink="$variables_saved_for_relink GCC_EXEC_PREFIX COMPILER_PATH LIBRARY_PATH"
@@ -14984,6 +15444,7 @@ if test -f "$ltmain"; then
     predeps_F77 \
     postdeps_F77 \
     compiler_lib_search_path_F77 \
+    compiler_lib_search_dirs_F77 \
     archive_cmds_F77 \
     archive_expsym_cmds_F77 \
     postinstall_cmds_F77 \
@@ -14999,6 +15460,7 @@ if test -f "$ltmain"; then
     module_cmds_F77 \
     module_expsym_cmds_F77 \
     lt_cv_prog_compiler_c_o_F77 \
+    fix_srcfile_path_F77 \
     exclude_expsyms_F77 \
     include_expsyms_F77; do
 
@@ -15231,6 +15693,10 @@ predeps=$lt_predeps_F77
 # shared library.
 postdeps=$lt_postdeps_F77
 
+# The directories searched by this compiler when creating a shared
+# library
+compiler_lib_search_dirs=$lt_compiler_lib_search_dirs_F77
+
 # The library search path used internally by the compiler when linking
 # a shared library.
 compiler_lib_search_path=$lt_compiler_lib_search_path_F77
@@ -15319,7 +15785,7 @@ sys_lib_search_path_spec=$lt_sys_lib_search_path_spec
 sys_lib_dlsearch_path_spec=$lt_sys_lib_dlsearch_path_spec
 
 # Fix the shell variable \$srcfile for the compiler.
-fix_srcfile_path="$fix_srcfile_path_F77"
+fix_srcfile_path=$lt_fix_srcfile_path
 
 # Set to yes if exported symbols are required.
 always_export_symbols=$always_export_symbols_F77
@@ -15378,10 +15844,10 @@ objext=o
 objext_GCJ=$objext
 
 # Code to be used in simple compile tests
-lt_simple_compile_test_code="class foo {}\n"
+lt_simple_compile_test_code="class foo {}"
 
 # Code to be used in simple link tests
-lt_simple_link_test_code='public class conftest { public static void main(String[] argv) {}; }\n'
+lt_simple_link_test_code='public class conftest { public static void main(String[] argv) {}; }'
 
 # ltmain only uses $CC for tagged configurations so make sure $CC is set.
 
@@ -15397,16 +15863,16 @@ compiler=$CC
 
 # save warnings/boilerplate of simple test code
 ac_outfile=conftest.$ac_objext
-printf "$lt_simple_compile_test_code" >conftest.$ac_ext
+echo "$lt_simple_compile_test_code" >conftest.$ac_ext
 eval "$ac_compile" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_compiler_boilerplate=`cat conftest.err`
 $rm conftest*
 
 ac_outfile=conftest.$ac_objext
-printf "$lt_simple_link_test_code" >conftest.$ac_ext
+echo "$lt_simple_link_test_code" >conftest.$ac_ext
 eval "$ac_link" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_linker_boilerplate=`cat conftest.err`
-$rm conftest*
+$rm -r conftest*
 
 
 # Allow CC to be a program name with arguments.
@@ -15444,7 +15910,7 @@ if test "${lt_cv_prog_compiler_rtti_exceptions+set}" = set; then
 else
   lt_cv_prog_compiler_rtti_exceptions=no
   ac_outfile=conftest.$ac_objext
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
    lt_compiler_flag="-fno-rtti -fno-exceptions"
    # Insert the option either (1) after the last *FLAGS variable, or
    # (2) before a word containing "conftest.", or (3) at the end.
@@ -15455,11 +15921,11 @@ else
    -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \
    -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \
    -e 's:$: $lt_compiler_flag:'`
-   (eval echo "\"\$as_me:15458: $lt_compile\"" >&5)
+   (eval echo "\"\$as_me:15924: $lt_compile\"" >&5)
    (eval "$lt_compile" 2>conftest.err)
    ac_status=$?
    cat conftest.err >&5
-   echo "$as_me:15462: \$? = $ac_status" >&5
+   echo "$as_me:15928: \$? = $ac_status" >&5
    if (exit $ac_status) && test -s "$ac_outfile"; then
      # The compiler can only warn and ignore the option if not recognized
      # So say no if there are warnings other than the usual output.
@@ -15510,14 +15976,16 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_pic_GCJ='-m68020 -resident32 -malways-restore-a4'
       ;;
 
-    beos* | cygwin* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
+    beos* | irix5* | irix6* | nonstopux* | osf3* | osf4* | osf5*)
       # PIC is the default for these OSes.
       ;;
 
-    mingw* | pw32* | os2*)
+    mingw* | cygwin* | pw32* | os2*)
       # This hack is so that the source file can tell whether it is being
       # built for inclusion in a dll (and should export symbols for example).
-      lt_prog_compiler_pic_GCJ='-DDLL_EXPORT'
+      # Although the cygwin gcc ignores -fPIC, still need this for old-style
+      # (--disable-auto-import) libraries
+
       ;;
 
     darwin* | rhapsody*)
@@ -15526,7 +15994,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_pic_GCJ='-fno-common'
       ;;
 
-    interix3*)
+    interix[3-9]*)
       # Interix 3.x gcc -fpic/-fPIC options generate broken code.
       # Instead, we relocate shared libraries at runtime.
       ;;
@@ -15584,10 +16052,10 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
        esac
        ;;
 
-    mingw* | pw32* | os2*)
+    mingw* | cygwin* | pw32* | os2*)
       # This hack is so that the source file can tell whether it is being
       # built for inclusion in a dll (and should export symbols for example).
-      lt_prog_compiler_pic_GCJ='-DDLL_EXPORT'
+
       ;;
 
     hpux9* | hpux10* | hpux11*)
@@ -15617,7 +16085,7 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_static_GCJ='-Bstatic'
       ;;
 
-    linux*)
+    linux* | k*bsd*-gnu)
       case $cc_basename in
       icc* | ecc*)
 	lt_prog_compiler_wl_GCJ='-Wl,'
@@ -15636,6 +16104,22 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
         # All Alpha code is PIC.
         lt_prog_compiler_static_GCJ='-non_shared'
         ;;
+      *)
+        case `$CC -V 2>&1 | sed 5q` in
+	*Sun\ C*)
+	  # Sun C 5.9
+	  lt_prog_compiler_pic_GCJ='-KPIC'
+	  lt_prog_compiler_static_GCJ='-Bstatic'
+	  lt_prog_compiler_wl_GCJ='-Wl,'
+	  ;;
+	*Sun\ F*)
+	  # Sun Fortran 8.3 passes all unrecognized flags to the linker
+	  lt_prog_compiler_pic_GCJ='-KPIC'
+	  lt_prog_compiler_static_GCJ='-Bstatic'
+	  lt_prog_compiler_wl_GCJ=''
+	  ;;
+	esac
+	;;
       esac
       ;;
 
@@ -15645,6 +16129,10 @@ echo $ECHO_N "checking for $compiler option to produce PIC... $ECHO_C" >&6
       lt_prog_compiler_static_GCJ='-non_shared'
       ;;
 
+    rdos*)
+      lt_prog_compiler_static_GCJ='-non_shared'
+      ;;
+
     solaris*)
       lt_prog_compiler_pic_GCJ='-KPIC'
       lt_prog_compiler_static_GCJ='-Bstatic'
@@ -15707,12 +16195,12 @@ if test -n "$lt_prog_compiler_pic_GCJ"; then
 
 echo "$as_me:$LINENO: checking if $compiler PIC flag $lt_prog_compiler_pic_GCJ works" >&5
 echo $ECHO_N "checking if $compiler PIC flag $lt_prog_compiler_pic_GCJ works... $ECHO_C" >&6
-if test "${lt_prog_compiler_pic_works_GCJ+set}" = set; then
+if test "${lt_cv_prog_compiler_pic_works_GCJ+set}" = set; then
   echo $ECHO_N "(cached) $ECHO_C" >&6
 else
-  lt_prog_compiler_pic_works_GCJ=no
+  lt_cv_prog_compiler_pic_works_GCJ=no
   ac_outfile=conftest.$ac_objext
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
    lt_compiler_flag="$lt_prog_compiler_pic_GCJ"
    # Insert the option either (1) after the last *FLAGS variable, or
    # (2) before a word containing "conftest.", or (3) at the end.
@@ -15723,27 +16211,27 @@ else
    -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \
    -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \
    -e 's:$: $lt_compiler_flag:'`
-   (eval echo "\"\$as_me:15726: $lt_compile\"" >&5)
+   (eval echo "\"\$as_me:16214: $lt_compile\"" >&5)
    (eval "$lt_compile" 2>conftest.err)
    ac_status=$?
    cat conftest.err >&5
-   echo "$as_me:15730: \$? = $ac_status" >&5
+   echo "$as_me:16218: \$? = $ac_status" >&5
    if (exit $ac_status) && test -s "$ac_outfile"; then
      # The compiler can only warn and ignore the option if not recognized
      # So say no if there are warnings other than the usual output.
      $echo "X$_lt_compiler_boilerplate" | $Xsed -e '/^$/d' >conftest.exp
      $SED '/^$/d; /^ *+/d' conftest.err >conftest.er2
      if test ! -s conftest.er2 || diff conftest.exp conftest.er2 >/dev/null; then
-       lt_prog_compiler_pic_works_GCJ=yes
+       lt_cv_prog_compiler_pic_works_GCJ=yes
      fi
    fi
    $rm conftest*
 
 fi
-echo "$as_me:$LINENO: result: $lt_prog_compiler_pic_works_GCJ" >&5
-echo "${ECHO_T}$lt_prog_compiler_pic_works_GCJ" >&6
+echo "$as_me:$LINENO: result: $lt_cv_prog_compiler_pic_works_GCJ" >&5
+echo "${ECHO_T}$lt_cv_prog_compiler_pic_works_GCJ" >&6
 
-if test x"$lt_prog_compiler_pic_works_GCJ" = xyes; then
+if test x"$lt_cv_prog_compiler_pic_works_GCJ" = xyes; then
     case $lt_prog_compiler_pic_GCJ in
      "" | " "*) ;;
      *) lt_prog_compiler_pic_GCJ=" $lt_prog_compiler_pic_GCJ" ;;
@@ -15770,13 +16258,13 @@ esac
 wl=$lt_prog_compiler_wl_GCJ eval lt_tmp_static_flag=\"$lt_prog_compiler_static_GCJ\"
 echo "$as_me:$LINENO: checking if $compiler static flag $lt_tmp_static_flag works" >&5
 echo $ECHO_N "checking if $compiler static flag $lt_tmp_static_flag works... $ECHO_C" >&6
-if test "${lt_prog_compiler_static_works_GCJ+set}" = set; then
+if test "${lt_cv_prog_compiler_static_works_GCJ+set}" = set; then
   echo $ECHO_N "(cached) $ECHO_C" >&6
 else
-  lt_prog_compiler_static_works_GCJ=no
+  lt_cv_prog_compiler_static_works_GCJ=no
    save_LDFLAGS="$LDFLAGS"
    LDFLAGS="$LDFLAGS $lt_tmp_static_flag"
-   printf "$lt_simple_link_test_code" > conftest.$ac_ext
+   echo "$lt_simple_link_test_code" > conftest.$ac_ext
    if (eval $ac_link 2>conftest.err) && test -s conftest$ac_exeext; then
      # The linker can only warn and ignore the option if not recognized
      # So say no if there are warnings
@@ -15786,20 +16274,20 @@ else
        $echo "X$_lt_linker_boilerplate" | $Xsed -e '/^$/d' > conftest.exp
        $SED '/^$/d; /^ *+/d' conftest.err >conftest.er2
        if diff conftest.exp conftest.er2 >/dev/null; then
-         lt_prog_compiler_static_works_GCJ=yes
+         lt_cv_prog_compiler_static_works_GCJ=yes
        fi
      else
-       lt_prog_compiler_static_works_GCJ=yes
+       lt_cv_prog_compiler_static_works_GCJ=yes
      fi
    fi
-   $rm conftest*
+   $rm -r conftest*
    LDFLAGS="$save_LDFLAGS"
 
 fi
-echo "$as_me:$LINENO: result: $lt_prog_compiler_static_works_GCJ" >&5
-echo "${ECHO_T}$lt_prog_compiler_static_works_GCJ" >&6
+echo "$as_me:$LINENO: result: $lt_cv_prog_compiler_static_works_GCJ" >&5
+echo "${ECHO_T}$lt_cv_prog_compiler_static_works_GCJ" >&6
 
-if test x"$lt_prog_compiler_static_works_GCJ" = xyes; then
+if test x"$lt_cv_prog_compiler_static_works_GCJ" = xyes; then
     :
 else
     lt_prog_compiler_static_GCJ=
@@ -15816,7 +16304,7 @@ else
    mkdir conftest
    cd conftest
    mkdir out
-   printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+   echo "$lt_simple_compile_test_code" > conftest.$ac_ext
 
    lt_compiler_flag="-o out/conftest2.$ac_objext"
    # Insert the option either (1) after the last *FLAGS variable, or
@@ -15827,11 +16315,11 @@ else
    -e 's:.*FLAGS}\{0,1\} :&$lt_compiler_flag :; t' \
    -e 's: [^ ]*conftest\.: $lt_compiler_flag&:; t' \
    -e 's:$: $lt_compiler_flag:'`
-   (eval echo "\"\$as_me:15830: $lt_compile\"" >&5)
+   (eval echo "\"\$as_me:16318: $lt_compile\"" >&5)
    (eval "$lt_compile" 2>out/conftest.err)
    ac_status=$?
    cat out/conftest.err >&5
-   echo "$as_me:15834: \$? = $ac_status" >&5
+   echo "$as_me:16322: \$? = $ac_status" >&5
    if (exit $ac_status) && test -s out/conftest2.$ac_objext
    then
      # The compiler can only warn and ignore the option if not recognized
@@ -15911,12 +16399,13 @@ echo $ECHO_N "checking whether the $compiler linker ($LD) supports shared librar
   # it will be wrapped by ` (' and `)$', so one must not match beginning or
   # end of line.  Example: `a|bc|.*d.*' will exclude the symbols `a' and `bc',
   # as well as any symbol that contains `d'.
-  exclude_expsyms_GCJ="_GLOBAL_OFFSET_TABLE_"
+  exclude_expsyms_GCJ='_GLOBAL_OFFSET_TABLE_|_GLOBAL__F[ID]_.*'
   # Although _GLOBAL_OFFSET_TABLE_ is a valid symbol C name, most a.out
   # platforms (ab)use it in PIC code, but their linkers get confused if
   # the symbol is explicitly referenced.  Since portable code cannot
   # rely on this symbol name, it's probably fine to never include it in
   # preloaded symbol tables.
+  # Exclude shared library initialization/finalization symbols.
   extract_expsyms_cmds=
   # Just being paranoid about ensuring that cc_basename is set.
   for cc_temp in $compiler""; do
@@ -15975,7 +16464,7 @@ cc_basename=`$echo "X$cc_temp" | $Xsed -e 's%.*/%%' -e "s%^$host_alias-%%"`
 
     # See if GNU ld supports shared libraries.
     case $host_os in
-    aix3* | aix4* | aix5*)
+    aix[3-9]*)
       # On AIX/PPC, the GNU linker is very broken
       if test "$host_cpu" != ia64; then
 	ld_shlibs_GCJ=no
@@ -16023,7 +16512,7 @@ EOF
       allow_undefined_flag_GCJ=unsupported
       always_export_symbols_GCJ=no
       enable_shared_with_static_runtimes_GCJ=yes
-      export_symbols_cmds_GCJ='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[BCDGRS] /s/.* \([^ ]*\)/\1 DATA/'\'' | $SED -e '\''/^[AITW] /s/.* //'\'' | sort | uniq > $export_symbols'
+      export_symbols_cmds_GCJ='$NM $libobjs $convenience | $global_symbol_pipe | $SED -e '\''/^[BCDGRS][ ]/s/.*[ ]\([^ ]*\)/\1 DATA/'\'' -e '\''/^[AITW][ ]/s/.*[ ]//'\'' | sort | uniq > $export_symbols'
 
       if $LD --help 2>&1 | grep 'auto-import' > /dev/null; then
         archive_cmds_GCJ='$CC -shared $libobjs $deplibs $compiler_flags -o $output_objdir/$soname ${wl}--enable-auto-image-base -Xlinker --out-implib -Xlinker $lib'
@@ -16041,7 +16530,7 @@ EOF
       fi
       ;;
 
-    interix3*)
+    interix[3-9]*)
       hardcode_direct_GCJ=no
       hardcode_shlibpath_var_GCJ=no
       hardcode_libdir_flag_spec_GCJ='${wl}-rpath,$libdir'
@@ -16056,7 +16545,7 @@ EOF
       archive_expsym_cmds_GCJ='sed "s,^,_," $export_symbols >$output_objdir/$soname.expsym~$CC -shared $pic_flag $libobjs $deplibs $compiler_flags ${wl}-h,$soname ${wl}--retain-symbols-file,$output_objdir/$soname.expsym ${wl}--image-base,`expr ${RANDOM-$$} % 4096 / 2 \* 262144 + 1342177280` -o $lib'
       ;;
 
-    linux*)
+    gnu* | linux* | k*bsd*-gnu)
       if $LD --help 2>&1 | grep ': supported targets:.* elf' > /dev/null; then
 	tmp_addflag=
 	case $cc_basename,$host_cpu in
@@ -16074,13 +16563,22 @@ EOF
 	ifc* | ifort*)			# Intel Fortran compiler
 	  tmp_addflag=' -nofor_main' ;;
 	esac
-	archive_cmds_GCJ='$CC -shared'"$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname -o $lib'
+	case `$CC -V 2>&1 | sed 5q` in
+	*Sun\ C*)			# Sun C 5.9
+	  whole_archive_flag_spec_GCJ='${wl}--whole-archive`new_convenience=; for conv in $convenience\"\"; do test -z \"$conv\" || new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}--no-whole-archive'
+	  tmp_sharedflag='-G' ;;
+	*Sun\ F*)			# Sun Fortran 8.3
+	  tmp_sharedflag='-G' ;;
+	*)
+	  tmp_sharedflag='-shared' ;;
+	esac
+	archive_cmds_GCJ='$CC '"$tmp_sharedflag""$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname -o $lib'
 
 	if test $supports_anon_versioning = yes; then
 	  archive_expsym_cmds_GCJ='$echo "{ global:" > $output_objdir/$libname.ver~
   cat $export_symbols | sed -e "s/\(.*\)/\1;/" >> $output_objdir/$libname.ver~
   $echo "local: *; };" >> $output_objdir/$libname.ver~
-	  $CC -shared'"$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname ${wl}-version-script ${wl}$output_objdir/$libname.ver -o $lib'
+	  $CC '"$tmp_sharedflag""$tmp_addflag"' $libobjs $deplibs $compiler_flags ${wl}-soname $wl$soname ${wl}-version-script ${wl}$output_objdir/$libname.ver -o $lib'
 	fi
       else
 	ld_shlibs_GCJ=no
@@ -16185,7 +16683,7 @@ _LT_EOF
       fi
       ;;
 
-    aix4* | aix5*)
+    aix[4-9]*)
       if test "$host_cpu" = ia64; then
 	# On IA64, the linker does run time linking by default, so we don't
 	# have to do anything special.
@@ -16205,7 +16703,7 @@ _LT_EOF
 	# Test if we are trying to use run time linking or normal
 	# AIX style linking. If -brtl is somewhere in LDFLAGS, we
 	# need to do runtime linking.
-	case $host_os in aix4.[23]|aix4.[23].*|aix5*)
+	case $host_os in aix4.[23]|aix4.[23].*|aix[5-9]*)
 	  for ld_flag in $LDFLAGS; do
   	  if (test $ld_flag = "-brtl" || test $ld_flag = "-Wl,-brtl"); then
   	    aix_use_runtimelinking=yes
@@ -16239,7 +16737,7 @@ _LT_EOF
   	   strings "$collect2name" | grep resolve_lib_name >/dev/null
 	  then
   	  # We have reworked collect2
-  	  hardcode_direct_GCJ=yes
+  	  :
 	  else
   	  # We have old collect2
   	  hardcode_direct_GCJ=unsupported
@@ -16317,11 +16815,18 @@ if { (eval echo "$as_me:$LINENO: \"$ac_link\"") >&5
   echo "$as_me:$LINENO: \$? = $ac_status" >&5
   (exit $ac_status); }; }; then
 
-aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`
+lt_aix_libpath_sed='
+    /Import File Strings/,/^$/ {
+	/^0/ {
+	    s/^0  *\(.*\)$/\1/
+	    p
+	}
+    }'
+aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
 # Check for a 64-bit object if we didn't find anything.
-if test -z "$aix_libpath"; then aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`; fi
+if test -z "$aix_libpath"; then
+  aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
+fi
 else
   echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
@@ -16378,11 +16883,18 @@ if { (eval echo "$as_me:$LINENO: \"$ac_link\"") >&5
   echo "$as_me:$LINENO: \$? = $ac_status" >&5
   (exit $ac_status); }; }; then
 
-aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`
+lt_aix_libpath_sed='
+    /Import File Strings/,/^$/ {
+	/^0/ {
+	    s/^0  *\(.*\)$/\1/
+	    p
+	}
+    }'
+aix_libpath=`dump -H conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
 # Check for a 64-bit object if we didn't find anything.
-if test -z "$aix_libpath"; then aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e '/Import File Strings/,/^$/ { /^0/ { s/^0  *\(.*\)$/\1/; p; }
-}'`; fi
+if test -z "$aix_libpath"; then
+  aix_libpath=`dump -HX64 conftest$ac_exeext 2>/dev/null | $SED -n -e "$lt_aix_libpath_sed"`
+fi
 else
   echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
@@ -16434,7 +16946,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       # The linker will automatically build a .lib file if we build a DLL.
       old_archive_From_new_cmds_GCJ='true'
       # FIXME: Should let the user specify the lib program.
-      old_archive_cmds_GCJ='lib /OUT:$oldlib$oldobjs$old_deplibs'
+      old_archive_cmds_GCJ='lib -OUT:$oldlib$oldobjs$old_deplibs'
       fix_srcfile_path_GCJ='`cygpath -w "$srcfile"`'
       enable_shared_with_static_runtimes_GCJ=yes
       ;;
@@ -16467,19 +16979,18 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       link_all_deplibs_GCJ=yes
     if test "$GCC" = yes ; then
     	output_verbose_link_cmd='echo'
-        archive_cmds_GCJ='$CC -dynamiclib $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring'
-      module_cmds_GCJ='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
-      # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-      archive_expsym_cmds_GCJ='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -dynamiclib $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags -install_name $rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
-      module_expsym_cmds_GCJ='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
+        archive_cmds_GCJ="\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring $_lt_dar_single_mod${_lt_dsymutil}"
+        module_cmds_GCJ="\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dsymutil}"
+        archive_expsym_cmds_GCJ="sed 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC -dynamiclib \$allow_undefined_flag -o \$lib \$libobjs \$deplibs \$compiler_flags -install_name \$rpath/\$soname \$verstring ${_lt_dar_single_mod}${_lt_dar_export_syms}${_lt_dsymutil}"
+        module_expsym_cmds_GCJ="sed -e 's,^,_,' < \$export_symbols > \$output_objdir/\${libname}-symbols.expsym~\$CC \$allow_undefined_flag -o \$lib -bundle \$libobjs \$deplibs \$compiler_flags${_lt_dar_export_syms}${_lt_dsymutil}"
     else
       case $cc_basename in
         xlc*)
          output_verbose_link_cmd='echo'
-         archive_cmds_GCJ='$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $verstring'
+         archive_cmds_GCJ='$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}`echo $rpath/$soname` $xlcverstring'
          module_cmds_GCJ='$CC $allow_undefined_flag -o $lib -bundle $libobjs $deplibs$compiler_flags'
           # Don't fix this by using the ld -exported_symbols_list flag, it doesn't exist in older darwin lds
-         archive_expsym_cmds_GCJ='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $verstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
+         archive_expsym_cmds_GCJ='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC -qmkshrobj $allow_undefined_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-install_name ${wl}$rpath/$soname $xlcverstring~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           module_expsym_cmds_GCJ='sed -e "s,#.*,," -e "s,^[    ]*,," -e "s,^\(..*\),_&," < $export_symbols > $output_objdir/${libname}-symbols.expsym~$CC $allow_undefined_flag  -o $lib -bundle $libobjs $deplibs$compiler_flags~nmedit -s $output_objdir/${libname}-symbols.expsym ${lib}'
           ;;
        *)
@@ -16519,7 +17030,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       ;;
 
     # FreeBSD 3 and greater uses gcc -shared to do shared libraries.
-    freebsd* | kfreebsd*-gnu | dragonfly*)
+    freebsd* | dragonfly*)
       archive_cmds_GCJ='$CC -shared -o $lib $libobjs $deplibs $compiler_flags'
       hardcode_libdir_flag_spec_GCJ='-R$libdir'
       hardcode_direct_GCJ=yes
@@ -16642,15 +17153,15 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 
     openbsd*)
       if test -f /usr/libexec/ld.so; then
-        hardcode_direct_GCJ=yes
-        hardcode_shlibpath_var_GCJ=no
-        if test -z "`echo __ELF__ | $CC -E - | grep __ELF__`" || test "$host_os-$host_cpu" = "openbsd2.8-powerpc"; then
+	hardcode_direct_GCJ=yes
+	hardcode_shlibpath_var_GCJ=no
+	if test -z "`echo __ELF__ | $CC -E - | grep __ELF__`" || test "$host_os-$host_cpu" = "openbsd2.8-powerpc"; then
 	  archive_cmds_GCJ='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags'
 	  archive_expsym_cmds_GCJ='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags ${wl}-retain-symbols-file,$export_symbols'
 	  hardcode_libdir_flag_spec_GCJ='${wl}-rpath,$libdir'
 	  export_dynamic_flag_spec_GCJ='${wl}-E'
-        else
-         case $host_os in
+	else
+	  case $host_os in
 	   openbsd[01].* | openbsd2.[0-7] | openbsd2.[0-7].*)
 	     archive_cmds_GCJ='$LD -Bshareable -o $lib $libobjs $deplibs $linker_flags'
 	     hardcode_libdir_flag_spec_GCJ='-R$libdir'
@@ -16659,10 +17170,10 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
 	     archive_cmds_GCJ='$CC -shared $pic_flag -o $lib $libobjs $deplibs $compiler_flags'
 	     hardcode_libdir_flag_spec_GCJ='${wl}-rpath,$libdir'
 	     ;;
-         esac
+	  esac
         fi
       else
-        ld_shlibs_GCJ=no
+	ld_shlibs_GCJ=no
       fi
       ;;
 
@@ -16721,17 +17232,16 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       case $host_os in
       solaris2.[0-5] | solaris2.[0-5].*) ;;
       *)
- 	# The compiler driver will combine linker options so we
- 	# cannot just pass the convience library names through
- 	# without $wl, iff we do not link with $LD.
- 	# Luckily, gcc supports the same syntax we need for Sun Studio.
+	# The compiler driver will combine and reorder linker options,
+	# but understands `-z linker_flag'.  GCC discards it without `$wl',
+	# but is careful enough not to reorder.
  	# Supported since Solaris 2.6 (maybe 2.5.1?)
- 	case $wlarc in
- 	'')
- 	  whole_archive_flag_spec_GCJ='-z allextract$convenience -z defaultextract' ;;
- 	*)
- 	  whole_archive_flag_spec_GCJ='${wl}-z ${wl}allextract`for conv in $convenience\"\"; do test -n \"$conv\" && new_convenience=\"$new_convenience,$conv\"; done; $echo \"$new_convenience\"` ${wl}-z ${wl}defaultextract' ;;
- 	esac ;;
+	if test "$GCC" = yes; then
+	  whole_archive_flag_spec_GCJ='${wl}-z ${wl}allextract$convenience ${wl}-z ${wl}defaultextract'
+	else
+	  whole_archive_flag_spec_GCJ='-z allextract$convenience -z defaultextract'
+	fi
+	;;
       esac
       link_all_deplibs_GCJ=yes
       ;;
@@ -16788,7 +17298,7 @@ if test -z "$aix_libpath"; then aix_libpath="/usr/lib:/lib"; fi
       fi
       ;;
 
-    sysv4*uw2* | sysv5OpenUNIX* | sysv5UnixWare7.[01].[10]* | unixware7*)
+    sysv4*uw2* | sysv5OpenUNIX* | sysv5UnixWare7.[01].[10]* | unixware7* | sco3.2v5.0.[024]*)
       no_undefined_flag_GCJ='${wl}-z,text'
       archive_cmds_need_lc_GCJ=no
       hardcode_shlibpath_var_GCJ=no
@@ -16865,7 +17375,7 @@ x|xyes)
       echo "$as_me:$LINENO: checking whether -lc should be explicitly linked in" >&5
 echo $ECHO_N "checking whether -lc should be explicitly linked in... $ECHO_C" >&6
       $rm conftest*
-      printf "$lt_simple_compile_test_code" > conftest.$ac_ext
+      echo "$lt_simple_compile_test_code" > conftest.$ac_ext
 
       if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5
   (eval $ac_compile) 2>&5
@@ -16923,20 +17433,7 @@ shlibpath_overrides_runpath=unknown
 version_type=none
 dynamic_linker="$host_os ld.so"
 sys_lib_dlsearch_path_spec="/lib /usr/lib"
-if test "$GCC" = yes; then
-  sys_lib_search_path_spec=`$CC -print-search-dirs | grep "^libraries:" | $SED -e "s/^libraries://" -e "s,=/,/,g"`
-  if echo "$sys_lib_search_path_spec" | grep ';' >/dev/null ; then
-    # if the path contains ";" then we assume it to be the separator
-    # otherwise default to the standard path separator (i.e. ":") - it is
-    # assumed that no part of a normal pathname contains ";" but that should
-    # okay in the real world where ";" in dirpaths is itself problematic.
-    sys_lib_search_path_spec=`echo "$sys_lib_search_path_spec" | $SED -e 's/;/ /g'`
-  else
-    sys_lib_search_path_spec=`echo "$sys_lib_search_path_spec" | $SED  -e "s/$PATH_SEPARATOR/ /g"`
-  fi
-else
-  sys_lib_search_path_spec="/lib /usr/lib /usr/local/lib"
-fi
+
 need_lib_prefix=unknown
 hardcode_into_libs=no
 
@@ -16954,7 +17451,7 @@ aix3*)
   soname_spec='${libname}${release}${shared_ext}$major'
   ;;
 
-aix4* | aix5*)
+aix[4-9]*)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -17093,12 +17590,7 @@ darwin* | rhapsody*)
   shlibpath_overrides_runpath=yes
   shlibpath_var=DYLD_LIBRARY_PATH
   shrext_cmds='`test .$module = .yes && echo .so || echo .dylib`'
-  # Apple's gcc prints 'gcc -print-search-dirs' doesn't operate the same.
-  if test "$GCC" = yes; then
-    sys_lib_search_path_spec=`$CC -print-search-dirs | tr "\n" "$PATH_SEPARATOR" | sed -e 's/libraries:/@libraries:/' | tr "@" "\n" | grep "^libraries:" | sed -e "s/^libraries://" -e "s,=/,/,g" -e "s,$PATH_SEPARATOR, ,g" -e "s,.*,& /lib /usr/lib /usr/local/lib,g"`
-  else
-    sys_lib_search_path_spec='/lib /usr/lib /usr/local/lib'
-  fi
+
   sys_lib_dlsearch_path_spec='/usr/local/lib /lib /usr/lib'
   ;;
 
@@ -17115,18 +17607,6 @@ freebsd1*)
   dynamic_linker=no
   ;;
 
-kfreebsd*-gnu)
-  version_type=linux
-  need_lib_prefix=no
-  need_version=no
-  library_names_spec='${libname}${release}${shared_ext}$versuffix ${libname}${release}${shared_ext}$major ${libname}${shared_ext}'
-  soname_spec='${libname}${release}${shared_ext}$major'
-  shlibpath_var=LD_LIBRARY_PATH
-  shlibpath_overrides_runpath=no
-  hardcode_into_libs=yes
-  dynamic_linker='GNU ld.so'
-  ;;
-
 freebsd* | dragonfly*)
   # DragonFly does not have aout.  When/if they implement a new
   # versioning mechanism, adjust this.
@@ -17164,7 +17644,7 @@ freebsd* | dragonfly*)
     shlibpath_overrides_runpath=no
     hardcode_into_libs=yes
     ;;
-  freebsd*) # from 4.6 on
+  *) # from 4.6 on, and DragonFly
     shlibpath_overrides_runpath=yes
     hardcode_into_libs=yes
     ;;
@@ -17227,7 +17707,7 @@ hpux9* | hpux10* | hpux11*)
   postinstall_cmds='chmod 555 $lib'
   ;;
 
-interix3*)
+interix[3-9]*)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -17282,7 +17762,7 @@ linux*oldld* | linux*aout* | linux*coff*)
   ;;
 
 # This must be Linux ELF.
-linux*)
+linux* | k*bsd*-gnu)
   version_type=linux
   need_lib_prefix=no
   need_version=no
@@ -17298,7 +17778,7 @@ linux*)
 
   # Append ld.so.conf contents to the search path
   if test -f /etc/ld.so.conf; then
-    lt_ld_extra=`awk '/^include / { system(sprintf("cd /etc; cat %s", \$2)); skip = 1; } { if (!skip) print \$0; skip = 0; }' < /etc/ld.so.conf | $SED -e 's/#.*//;s/[:,	]/ /g;s/=[^=]*$//;s/=[^= ]* / /g;/^$/d' | tr '\n' ' '`
+    lt_ld_extra=`awk '/^include / { system(sprintf("cd /etc; cat %s 2>/dev/null", \$2)); skip = 1; } { if (!skip) print \$0; skip = 0; }' < /etc/ld.so.conf | $SED -e 's/#.*//;/^[ 	]*hwcap[ 	]/d;s/[:,	]/ /g;s/=[^=]*$//;s/=[^= ]* / /g;/^$/d' | tr '\n' ' '`
     sys_lib_dlsearch_path_spec="/lib /usr/lib $lt_ld_extra"
   fi
 
@@ -17311,18 +17791,6 @@ linux*)
   dynamic_linker='GNU/Linux ld.so'
   ;;
 
-knetbsd*-gnu)
-  version_type=linux
-  need_lib_prefix=no
-  need_version=no
-  library_names_spec='${libname}${release}${shared_ext}$versuffix ${libname}${release}${shared_ext}$major ${libname}${shared_ext}'
-  soname_spec='${libname}${release}${shared_ext}$major'
-  shlibpath_var=LD_LIBRARY_PATH
-  shlibpath_overrides_runpath=no
-  hardcode_into_libs=yes
-  dynamic_linker='GNU ld.so'
-  ;;
-
 netbsd*)
   version_type=sunos
   need_lib_prefix=no
@@ -17404,6 +17872,10 @@ osf3* | osf4* | osf5*)
   sys_lib_dlsearch_path_spec="$sys_lib_search_path_spec"
   ;;
 
+rdos*)
+  dynamic_linker=no
+  ;;
+
 solaris*)
   version_type=linux
   need_lib_prefix=no
@@ -17500,6 +17972,21 @@ echo "$as_me:$LINENO: result: $dynamic_linker" >&5
 echo "${ECHO_T}$dynamic_linker" >&6
 test "$dynamic_linker" = no && can_build_shared=no
 
+if test "${lt_cv_sys_lib_search_path_spec+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  lt_cv_sys_lib_search_path_spec="$sys_lib_search_path_spec"
+fi
+
+sys_lib_search_path_spec="$lt_cv_sys_lib_search_path_spec"
+if test "${lt_cv_sys_lib_dlsearch_path_spec+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  lt_cv_sys_lib_dlsearch_path_spec="$sys_lib_dlsearch_path_spec"
+fi
+
+sys_lib_dlsearch_path_spec="$lt_cv_sys_lib_dlsearch_path_spec"
+
 variables_saved_for_relink="PATH $shlibpath_var $runpath_var"
 if test "$GCC" = yes; then
   variables_saved_for_relink="$variables_saved_for_relink GCC_EXEC_PREFIX COMPILER_PATH LIBRARY_PATH"
@@ -17583,6 +18070,7 @@ if test -f "$ltmain"; then
     predeps_GCJ \
     postdeps_GCJ \
     compiler_lib_search_path_GCJ \
+    compiler_lib_search_dirs_GCJ \
     archive_cmds_GCJ \
     archive_expsym_cmds_GCJ \
     postinstall_cmds_GCJ \
@@ -17598,6 +18086,7 @@ if test -f "$ltmain"; then
     module_cmds_GCJ \
     module_expsym_cmds_GCJ \
     lt_cv_prog_compiler_c_o_GCJ \
+    fix_srcfile_path_GCJ \
     exclude_expsyms_GCJ \
     include_expsyms_GCJ; do
 
@@ -17830,6 +18319,10 @@ predeps=$lt_predeps_GCJ
 # shared library.
 postdeps=$lt_postdeps_GCJ
 
+# The directories searched by this compiler when creating a shared
+# library
+compiler_lib_search_dirs=$lt_compiler_lib_search_dirs_GCJ
+
 # The library search path used internally by the compiler when linking
 # a shared library.
 compiler_lib_search_path=$lt_compiler_lib_search_path_GCJ
@@ -17918,7 +18411,7 @@ sys_lib_search_path_spec=$lt_sys_lib_search_path_spec
 sys_lib_dlsearch_path_spec=$lt_sys_lib_dlsearch_path_spec
 
 # Fix the shell variable \$srcfile for the compiler.
-fix_srcfile_path="$fix_srcfile_path_GCJ"
+fix_srcfile_path=$lt_fix_srcfile_path
 
 # Set to yes if exported symbols are required.
 always_export_symbols=$always_export_symbols_GCJ
@@ -17976,7 +18469,7 @@ objext=o
 objext_RC=$objext
 
 # Code to be used in simple compile tests
-lt_simple_compile_test_code='sample MENU { MENUITEM "&Soup", 100, CHECKED }\n'
+lt_simple_compile_test_code='sample MENU { MENUITEM "&Soup", 100, CHECKED }'
 
 # Code to be used in simple link tests
 lt_simple_link_test_code="$lt_simple_compile_test_code"
@@ -17995,16 +18488,16 @@ compiler=$CC
 
 # save warnings/boilerplate of simple test code
 ac_outfile=conftest.$ac_objext
-printf "$lt_simple_compile_test_code" >conftest.$ac_ext
+echo "$lt_simple_compile_test_code" >conftest.$ac_ext
 eval "$ac_compile" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_compiler_boilerplate=`cat conftest.err`
 $rm conftest*
 
 ac_outfile=conftest.$ac_objext
-printf "$lt_simple_link_test_code" >conftest.$ac_ext
+echo "$lt_simple_link_test_code" >conftest.$ac_ext
 eval "$ac_link" 2>&1 >/dev/null | $SED '/^$/d; /^ *+/d' >conftest.err
 _lt_linker_boilerplate=`cat conftest.err`
-$rm conftest*
+$rm -r conftest*
 
 
 # Allow CC to be a program name with arguments.
@@ -18064,6 +18557,7 @@ if test -f "$ltmain"; then
     predeps_RC \
     postdeps_RC \
     compiler_lib_search_path_RC \
+    compiler_lib_search_dirs_RC \
     archive_cmds_RC \
     archive_expsym_cmds_RC \
     postinstall_cmds_RC \
@@ -18079,6 +18573,7 @@ if test -f "$ltmain"; then
     module_cmds_RC \
     module_expsym_cmds_RC \
     lt_cv_prog_compiler_c_o_RC \
+    fix_srcfile_path_RC \
     exclude_expsyms_RC \
     include_expsyms_RC; do
 
@@ -18311,6 +18806,10 @@ predeps=$lt_predeps_RC
 # shared library.
 postdeps=$lt_postdeps_RC
 
+# The directories searched by this compiler when creating a shared
+# library
+compiler_lib_search_dirs=$lt_compiler_lib_search_dirs_RC
+
 # The library search path used internally by the compiler when linking
 # a shared library.
 compiler_lib_search_path=$lt_compiler_lib_search_path_RC
@@ -18399,7 +18898,7 @@ sys_lib_search_path_spec=$lt_sys_lib_search_path_spec
 sys_lib_dlsearch_path_spec=$lt_sys_lib_dlsearch_path_spec
 
 # Fix the shell variable \$srcfile for the compiler.
-fix_srcfile_path="$fix_srcfile_path_RC"
+fix_srcfile_path=$lt_fix_srcfile_path
 
 # Set to yes if exported symbols are required.
 always_export_symbols=$always_export_symbols_RC
@@ -18830,6 +19329,76 @@ presetting ac_cv_c_bigendian=no (or yes) will help" >&2;}
    { (exit 1); exit 1; }; } ;;
 esac
 
+echo "$as_me:$LINENO: checking for inline" >&5
+echo $ECHO_N "checking for inline... $ECHO_C" >&6
+if test "${ac_cv_c_inline+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  ac_cv_c_inline=no
+for ac_kw in inline __inline__ __inline; do
+  cat >conftest.$ac_ext <<_ACEOF
+/* confdefs.h.  */
+_ACEOF
+cat confdefs.h >>conftest.$ac_ext
+cat >>conftest.$ac_ext <<_ACEOF
+/* end confdefs.h.  */
+#ifndef __cplusplus
+typedef int foo_t;
+static $ac_kw foo_t static_foo () {return 0; }
+$ac_kw foo_t foo () {return 0; }
+#endif
+
+_ACEOF
+rm -f conftest.$ac_objext
+if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5
+  (eval $ac_compile) 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } &&
+	 { ac_try='test -z "$ac_c_werror_flag"
+			 || test ! -s conftest.err'
+  { (eval echo "$as_me:$LINENO: \"$ac_try\"") >&5
+  (eval $ac_try) 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; } &&
+	 { ac_try='test -s conftest.$ac_objext'
+  { (eval echo "$as_me:$LINENO: \"$ac_try\"") >&5
+  (eval $ac_try) 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+  ac_cv_c_inline=$ac_kw; break
+else
+  echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+fi
+rm -f conftest.err conftest.$ac_objext conftest.$ac_ext
+done
+
+fi
+echo "$as_me:$LINENO: result: $ac_cv_c_inline" >&5
+echo "${ECHO_T}$ac_cv_c_inline" >&6
+
+
+case $ac_cv_c_inline in
+  inline | yes) ;;
+  *)
+    case $ac_cv_c_inline in
+      no) ac_val=;;
+      *) ac_val=$ac_cv_c_inline;;
+    esac
+    cat >>confdefs.h <<_ACEOF
+#ifndef __cplusplus
+#define inline $ac_val
+#endif
+_ACEOF
+    ;;
+esac
 
 #
 # We ignore pixman_major in the version here because the major version should
@@ -18841,11 +19410,11 @@ esac
 
 
 
-LT_VERSION_INFO="10:0:10"
+LT_VERSION_INFO="12:0:12"
 
 PIXMAN_VERSION_MAJOR=0
 
-PIXMAN_VERSION_MINOR=10
+PIXMAN_VERSION_MINOR=12
 
 PIXMAN_VERSION_MICRO=0
 
@@ -18860,6 +19429,53 @@ if test "x$GCC" = "xyes"; then
   *[\ \	]-Wall[\ \	]*) ;;
   *) CFLAGS="$CFLAGS -Wall" ;;
   esac fi
+# Extract the first word of "perl", so it can be a program name with args.
+set dummy perl; ac_word=$2
+echo "$as_me:$LINENO: checking for $ac_word" >&5
+echo $ECHO_N "checking for $ac_word... $ECHO_C" >&6
+if test "${ac_cv_path_PERL+set}" = set; then
+  echo $ECHO_N "(cached) $ECHO_C" >&6
+else
+  case $PERL in
+  [\\/]* | ?:[\\/]*)
+  ac_cv_path_PERL="$PERL" # Let the user override the test with a path.
+  ;;
+  *)
+  as_save_IFS=$IFS; IFS=$PATH_SEPARATOR
+for as_dir in $PATH
+do
+  IFS=$as_save_IFS
+  test -z "$as_dir" && as_dir=.
+  for ac_exec_ext in '' $ac_executable_extensions; do
+  if $as_executable_p "$as_dir/$ac_word$ac_exec_ext"; then
+    ac_cv_path_PERL="$as_dir/$ac_word$ac_exec_ext"
+    echo "$as_me:$LINENO: found $as_dir/$ac_word$ac_exec_ext" >&5
+    break 2
+  fi
+done
+done
+
+  test -z "$ac_cv_path_PERL" && ac_cv_path_PERL="no"
+  ;;
+esac
+fi
+PERL=$ac_cv_path_PERL
+
+if test -n "$PERL"; then
+  echo "$as_me:$LINENO: result: $PERL" >&5
+echo "${ECHO_T}$PERL" >&6
+else
+  echo "$as_me:$LINENO: result: no" >&5
+echo "${ECHO_T}no" >&6
+fi
+
+if test "x$PERL" = xno; then
+    { { echo "$as_me:$LINENO: error: Perl is required to build pixman." >&5
+echo "$as_me: error: Perl is required to build pixman." >&2;}
+   { (exit 1); exit 1; }; }
+fi
+
+
 
 have_gcc4=no
 echo "$as_me:$LINENO: checking for -fvisibility" >&5
@@ -18868,7 +19484,7 @@ cat >conftest.$ac_ext <<_ACEOF
 
 #if defined(__GNUC__) && (__GNUC__ >= 4)
 #else
-#error Need GCC 4.0 for visibility
+error Need GCC 4.0 for visibility
 #endif
 int main () { return 0; }
 
@@ -18909,8 +19525,56 @@ fi
 echo "$as_me:$LINENO: result: $have_gcc4" >&5
 echo "${ECHO_T}$have_gcc4" >&6
 
+have_sunstudio8=no
+echo "$as_me:$LINENO: checking for -xldscope (Sun compilers)" >&5
+echo $ECHO_N "checking for -xldscope (Sun compilers)... $ECHO_C" >&6
+cat >conftest.$ac_ext <<_ACEOF
 
-MMX_CFLAGS="-mmmx -Winline --param inline-unit-growth=10000 --param large-function-growth=10000"
+#if defined(__SUNPRO_C) && (__SUNPRO_C >= 0x550)
+#else
+error Need Sun Studio 8 for visibility
+#endif
+int main () { return 0; }
+
+_ACEOF
+rm -f conftest.$ac_objext
+if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5
+  (eval $ac_compile) 2>conftest.er1
+  ac_status=$?
+  grep -v '^ *+' conftest.er1 >conftest.err
+  rm -f conftest.er1
+  cat conftest.err >&5
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); } &&
+	 { ac_try='test -z "$ac_c_werror_flag"
+			 || test ! -s conftest.err'
+  { (eval echo "$as_me:$LINENO: \"$ac_try\"") >&5
+  (eval $ac_try) 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; } &&
+	 { ac_try='test -s conftest.$ac_objext'
+  { (eval echo "$as_me:$LINENO: \"$ac_try\"") >&5
+  (eval $ac_try) 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; }; then
+  have_sunstudio8=yes
+else
+  echo "$as_me: failed program was:" >&5
+sed 's/^/| /' conftest.$ac_ext >&5
+
+fi
+rm -f conftest.err conftest.$ac_objext conftest.$ac_ext
+
+if test "x$have_sunstudio8" = "xyes"; then
+   CFLAGS="$CFLAGS -xldscope=hidden"
+fi
+echo "$as_me:$LINENO: result: $have_sunstudio8" >&5
+echo "${ECHO_T}$have_sunstudio8" >&6
+
+
+MMX_CFLAGS="-mmmx -Winline"
 
 have_mmx_intrinsics=no
 echo "$as_me:$LINENO: checking whether to use MMX intrinsics" >&5
@@ -18919,8 +19583,8 @@ xserver_save_CFLAGS=$CFLAGS
 CFLAGS="$CFLAGS $MMX_CFLAGS"
 cat >conftest.$ac_ext <<_ACEOF
 
-#if defined(__GNUC__) && (__GNUC__ < 3 || (__GNUC__ == 3 && __GNUC_MINOR__ < 4))
-#error "Need GCC >= 3.4 for MMX intrinsics"
+#if defined(__GNUC__) && (__GNUC__ < 3 || (__GNUC__ == 3 && __GNUC_MINOR__ < 3))
+error "Need GCC >= 3.3 for MMX intrinsics"
 #endif
 #include <mmintrin.h>
 int main () {
@@ -18958,8 +19622,18 @@ sed 's/^/| /' conftest.$ac_ext >&5
 fi
 rm -f conftest.err conftest.$ac_objext conftest.$ac_ext
 CFLAGS=$xserver_save_CFLAGS
-echo "$as_me:$LINENO: result: $have_mmx_intrinsics" >&5
-echo "${ECHO_T}$have_mmx_intrinsics" >&6
+
+# Check whether --enable-mmx or --disable-mmx was given.
+if test "${enable_mmx+set}" = set; then
+  enableval="$enable_mmx"
+  enable_mmx=$enableval
+else
+  enable_mmx=auto
+fi;
+
+if test $enable_mmx = no ; then
+   have_mmx_intrinsics=disabled
+fi
 
 if test $have_mmx_intrinsics = yes ; then
 
@@ -18971,6 +19645,14 @@ else
    MMX_CFLAGS=
 fi
 
+echo "$as_me:$LINENO: result: $have_mmx_intrinsics" >&5
+echo "${ECHO_T}$have_mmx_intrinsics" >&6
+if test $enable_mmx = yes && test $have_mmx_intrinsics = no ; then
+   { { echo "$as_me:$LINENO: error: MMX intrinsics not detected" >&5
+echo "$as_me: error: MMX intrinsics not detected" >&2;}
+   { (exit 1); exit 1; }; }
+fi
+
 
 
 if test $have_mmx_intrinsics = yes; then
@@ -18983,24 +19665,28 @@ fi
 
 
 
+SSE2_CFLAGS="-mmmx -msse2 -Winline"
 
-have_sse_intrinsics=no
-echo "$as_me:$LINENO: checking whether to use SSE intrinsics" >&5
-echo $ECHO_N "checking whether to use SSE intrinsics... $ECHO_C" >&6
+have_sse2_intrinsics=no
+echo "$as_me:$LINENO: checking whether to use SSE2 intrinsics" >&5
+echo $ECHO_N "checking whether to use SSE2 intrinsics... $ECHO_C" >&6
 xserver_save_CFLAGS=$CFLAGS
-CFLAGS="$CFLAGS -msse $MMX_CFLAGS"
+CFLAGS="$CFLAGS -msse2 $SSE2_CFLAGS"
 
 cat >conftest.$ac_ext <<_ACEOF
 
-#if !defined(__amd64__) && !defined(__x86_64__)
-#error "Need x86-64 for SSE"
+#if defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 2))
+#   if !defined(__amd64__) && !defined(__x86_64__)
+#      error "Need GCC >= 4.2 for SSE2 intrinsics on x86"
+#   endif
 #endif
 #include <mmintrin.h>
 #include <xmmintrin.h>
+#include <emmintrin.h>
 int main () {
-    __m64 v = _mm_cvtsi32_si64 (1);
-    v = _mm_shuffle_pi16 (v, _MM_SHUFFLE(3, 3, 3, 3));
-    return _mm_cvtsi64_si32 (v);
+    __m128i a, b, c;
+	c = _mm_xor_si128 (a, b);
+    return 0;
 }
 _ACEOF
 rm -f conftest.$ac_objext
@@ -19025,7 +19711,7 @@ if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5
   ac_status=$?
   echo "$as_me:$LINENO: \$? = $ac_status" >&5
   (exit $ac_status); }; }; then
-  have_sse_intrinsics=yes
+  have_sse2_intrinsics=yes
 else
   echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
@@ -19033,46 +19719,69 @@ sed 's/^/| /' conftest.$ac_ext >&5
 fi
 rm -f conftest.err conftest.$ac_objext conftest.$ac_ext
 CFLAGS=$xserver_save_CFLAGS
-echo "$as_me:$LINENO: result: $have_sse_intrinsics" >&5
-echo "${ECHO_T}$have_sse_intrinsics" >&6
 
-if test $have_sse_intrinsics = yes ; then
+# Check whether --enable-sse2 or --disable-sse2 was given.
+if test "${enable_sse2+set}" = set; then
+  enableval="$enable_sse2"
+  enable_sse2=$enableval
+else
+  enable_sse2=auto
+fi;
+
+if test $enable_sse2 = no ; then
+   have_sse2_intrinsics=disabled
+fi
+
+if test $have_sse2_intrinsics = yes ; then
 
 cat >>confdefs.h <<\_ACEOF
-#define USE_SSE 1
+#define USE_SSE2 1
 _ACEOF
 
-   MMX_CFLAGS="-msse $MMX_CFLAGS"
+fi
+
+echo "$as_me:$LINENO: result: $have_sse2_intrinsics" >&5
+echo "${ECHO_T}$have_sse2_intrinsics" >&6
+if test $enable_sse2 = yes && test $have_sse2_intrinsics = no ; then
+   { { echo "$as_me:$LINENO: error: SSE2 intrinsics not detected" >&5
+echo "$as_me: error: SSE2 intrinsics not detected" >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
 
 
-if test $have_sse_intrinsics = yes; then
-  USE_SSE_TRUE=
-  USE_SSE_FALSE='#'
+if test $have_sse2_intrinsics = yes; then
+  USE_SSE2_TRUE=
+  USE_SSE2_FALSE='#'
 else
-  USE_SSE_TRUE='#'
-  USE_SSE_FALSE=
+  USE_SSE2_TRUE='#'
+  USE_SSE2_FALSE=
 fi
 
 
 
 
-SSE_CFLAGS="-mmmx -msse2 -Winline --param inline-unit-growth=10000 --param large-function-growth=10000"
 
-have_sse2_intrinsics=no
-echo "$as_me:$LINENO: checking whether to use SSE2 intrinsics" >&5
-echo $ECHO_N "checking whether to use SSE2 intrinsics... $ECHO_C" >&6
-xserver_save_CFLAGS=$CFLAGS
-CFLAGS="$CFLAGS -msse2 $SSE_CFLAGS"
+if test -n "`$CC -v 2>&1 | grep version | grep Apple`"; then
+    VMX_CFLAGS="-faltivec"
+else
+    VMX_CFLAGS="-maltivec -mabi=altivec"
+fi
 
+have_vmx_intrinsics=no
+echo "$as_me:$LINENO: checking whether to use VMX/Altivec intrinsics" >&5
+echo $ECHO_N "checking whether to use VMX/Altivec intrinsics... $ECHO_C" >&6
+xserver_save_CFLAGS=$CFLAGS
+CFLAGS="$CFLAGS $VMX_CFLAGS"
 cat >conftest.$ac_ext <<_ACEOF
 
-#include <mmintrin.h>
-#include <xmmintrin.h>
+#if defined(__GNUC__) && (__GNUC__ < 3 || (__GNUC__ == 3 && __GNUC_MINOR__ < 4))
+error "Need GCC >= 3.4 for sane altivec support"
+#endif
+#include <altivec.h>
 int main () {
-    __m128i a, b, c;
-	c = _mm_xor_si128 (a, b);
+    vector unsigned int v = vec_splat_u32 (1);
+    v = vec_sub (v, v);
     return 0;
 }
 _ACEOF
@@ -19098,7 +19807,7 @@ if { (eval echo "$as_me:$LINENO: \"$ac_compile\"") >&5
   ac_status=$?
   echo "$as_me:$LINENO: \$? = $ac_status" >&5
   (exit $ac_status); }; }; then
-  have_sse2_intrinsics=yes
+  have_vmx_intrinsics=yes
 else
   echo "$as_me: failed program was:" >&5
 sed 's/^/| /' conftest.$ac_ext >&5
@@ -19106,30 +19815,57 @@ sed 's/^/| /' conftest.$ac_ext >&5
 fi
 rm -f conftest.err conftest.$ac_objext conftest.$ac_ext
 CFLAGS=$xserver_save_CFLAGS
-echo "$as_me:$LINENO: result: $have_sse2_intrinsics" >&5
-echo "${ECHO_T}$have_sse2_intrinsics" >&6
 
-if test $have_sse2_intrinsics = yes ; then
+# Check whether --enable-vmx or --disable-vmx was given.
+if test "${enable_vmx+set}" = set; then
+  enableval="$enable_vmx"
+  enable_vmx=$enableval
+else
+  enable_vmx=auto
+fi;
+
+if test $enable_vmx = no ; then
+   have_vmx_intrinsics=disabled
+fi
+
+if test $have_vmx_intrinsics = yes ; then
 
 cat >>confdefs.h <<\_ACEOF
-#define USE_SSE2 1
+#define USE_VMX 1
 _ACEOF
 
+else
+   VMX_CFLAGS=
+fi
+
+echo "$as_me:$LINENO: result: $have_vmx_intrinsics" >&5
+echo "${ECHO_T}$have_vmx_intrinsics" >&6
+if test $enable_vmx = yes && test $have_vmx_intrinsics = no ; then
+   { { echo "$as_me:$LINENO: error: VMX intrinsics not detected" >&5
+echo "$as_me: error: VMX intrinsics not detected" >&2;}
+   { (exit 1); exit 1; }; }
 fi
 
 
 
-if test $have_sse2_intrinsics = yes; then
-  USE_SSE2_TRUE=
-  USE_SSE2_FALSE='#'
-else
-  USE_SSE2_TRUE='#'
-  USE_SSE2_FALSE=
-fi
 
 
+if test $have_vmx_intrinsics = yes; then
+  USE_VMX_TRUE=
+  USE_VMX_FALSE='#'
+else
+  USE_VMX_TRUE='#'
+  USE_VMX_FALSE=
+fi
 
 
+# Check whether --enable-gtk or --disable-gtk was given.
+if test "${enable_gtk+set}" = set; then
+  enableval="$enable_gtk"
+  enable_gtk=$enableval
+else
+  enable_gtk=auto
+fi;
 
 
 
@@ -19236,6 +19972,19 @@ echo "${ECHO_T}no" >&6
 	fi
 
 fi
+if test $enable_gtk = auto ; then
+   if test -n "$PKG_CONFIG" && \
+    { (echo "$as_me:$LINENO: \$PKG_CONFIG --exists --print-errors \"gtk+-2.0\"") >&5
+  ($PKG_CONFIG --exists --print-errors "gtk+-2.0") 2>&5
+  ac_status=$?
+  echo "$as_me:$LINENO: \$? = $ac_status" >&5
+  (exit $ac_status); }; then
+  enable_gtk=yes
+else
+  enable_gtk=no
+fi
+fi
+if test $enable_gtk = yes ; then
 
 pkg_failed=no
 echo "$as_me:$LINENO: checking for GTK" >&5
@@ -19295,21 +20044,63 @@ fi
 	# Put the nasty error message in config.log where it belongs
 	echo "$GTK_PKG_ERRORS" >&5
 
-	echo "$as_me:$LINENO: result: no" >&5
-echo "${ECHO_T}no" >&6
-                HAVE_GTK=no
+	{ { echo "$as_me:$LINENO: error: Package requirements (gtk+-2.0) were not met:
+
+$GTK_PKG_ERRORS
+
+Consider adjusting the PKG_CONFIG_PATH environment variable if you
+installed software in a non-standard prefix.
+
+Alternatively, you may set the environment variables GTK_CFLAGS
+and GTK_LIBS to avoid the need to call pkg-config.
+See the pkg-config man page for more details.
+" >&5
+echo "$as_me: error: Package requirements (gtk+-2.0) were not met:
+
+$GTK_PKG_ERRORS
+
+Consider adjusting the PKG_CONFIG_PATH environment variable if you
+installed software in a non-standard prefix.
+
+Alternatively, you may set the environment variables GTK_CFLAGS
+and GTK_LIBS to avoid the need to call pkg-config.
+See the pkg-config man page for more details.
+" >&2;}
+   { (exit 1); exit 1; }; }
 elif test $pkg_failed = untried; then
-	HAVE_GTK=no
+	{ { echo "$as_me:$LINENO: error: The pkg-config script could not be found or is too old.  Make sure it
+is in your PATH or set the PKG_CONFIG environment variable to the full
+path to pkg-config.
+
+Alternatively, you may set the environment variables GTK_CFLAGS
+and GTK_LIBS to avoid the need to call pkg-config.
+See the pkg-config man page for more details.
+
+To get pkg-config, see <http://www.freedesktop.org/software/pkgconfig>.
+See \`config.log' for more details." >&5
+echo "$as_me: error: The pkg-config script could not be found or is too old.  Make sure it
+is in your PATH or set the PKG_CONFIG environment variable to the full
+path to pkg-config.
+
+Alternatively, you may set the environment variables GTK_CFLAGS
+and GTK_LIBS to avoid the need to call pkg-config.
+See the pkg-config man page for more details.
+
+To get pkg-config, see <http://www.freedesktop.org/software/pkgconfig>.
+See \`config.log' for more details." >&2;}
+   { (exit 1); exit 1; }; }
 else
 	GTK_CFLAGS=$pkg_cv_GTK_CFLAGS
 	GTK_LIBS=$pkg_cv_GTK_LIBS
         echo "$as_me:$LINENO: result: yes" >&5
 echo "${ECHO_T}yes" >&6
-	HAVE_GTK=yes
+	:
 fi
+fi
+
 
 
-if test "x$HAVE_GTK" = xyes; then
+if test "x$enable_gtk" = xyes; then
   HAVE_GTK_TRUE=
   HAVE_GTK_FALSE='#'
 else
@@ -19323,7 +20114,7 @@ fi
 
 
 
-                                                  ac_config_files="$ac_config_files pixman-1.pc Makefile pixman/Makefile pixman/pixman-version.h test/Makefile"
+                                                            ac_config_files="$ac_config_files pixman-1.pc pixman-1-uninstalled.pc Makefile pixman/Makefile pixman/pixman-version.h test/Makefile"
 cat >confcache <<\_ACEOF
 # This file is a shell script that caches the results of configure
 # tests run on this system so they can be shared between configure
@@ -19443,17 +20234,17 @@ echo "$as_me: error: conditional \"USE_MMX\" was never defined.
 Usually this means the macro was only invoked conditionally." >&2;}
    { (exit 1); exit 1; }; }
 fi
-if test -z "${USE_SSE_TRUE}" && test -z "${USE_SSE_FALSE}"; then
-  { { echo "$as_me:$LINENO: error: conditional \"USE_SSE\" was never defined.
+if test -z "${USE_SSE2_TRUE}" && test -z "${USE_SSE2_FALSE}"; then
+  { { echo "$as_me:$LINENO: error: conditional \"USE_SSE2\" was never defined.
 Usually this means the macro was only invoked conditionally." >&5
-echo "$as_me: error: conditional \"USE_SSE\" was never defined.
+echo "$as_me: error: conditional \"USE_SSE2\" was never defined.
 Usually this means the macro was only invoked conditionally." >&2;}
    { (exit 1); exit 1; }; }
 fi
-if test -z "${USE_SSE2_TRUE}" && test -z "${USE_SSE2_FALSE}"; then
-  { { echo "$as_me:$LINENO: error: conditional \"USE_SSE2\" was never defined.
+if test -z "${USE_VMX_TRUE}" && test -z "${USE_VMX_FALSE}"; then
+  { { echo "$as_me:$LINENO: error: conditional \"USE_VMX\" was never defined.
 Usually this means the macro was only invoked conditionally." >&5
-echo "$as_me: error: conditional \"USE_SSE2\" was never defined.
+echo "$as_me: error: conditional \"USE_VMX\" was never defined.
 Usually this means the macro was only invoked conditionally." >&2;}
    { (exit 1); exit 1; }; }
 fi
@@ -19735,7 +20526,7 @@ _ASBOX
 } >&5
 cat >&5 <<_CSEOF
 
-This file was extended by pixman $as_me 0.10.0, which was
+This file was extended by pixman $as_me 0.12.0, which was
 generated by GNU Autoconf 2.59.  Invocation command line was
 
   CONFIG_FILES    = $CONFIG_FILES
@@ -19798,7 +20589,7 @@ _ACEOF
 
 cat >>$CONFIG_STATUS <<_ACEOF
 ac_cs_version="\\
-pixman config.status 0.10.0
+pixman config.status 0.12.0
 configured by $0, generated by GNU Autoconf 2.59,
   with options \\"`echo "$ac_configure_args" | sed 's/[\\""\`\$]/\\\\&/g'`\\"
 
@@ -19909,6 +20700,7 @@ do
   case "$ac_config_target" in
   # Handling of arguments.
   "pixman-1.pc" ) CONFIG_FILES="$CONFIG_FILES pixman-1.pc" ;;
+  "pixman-1-uninstalled.pc" ) CONFIG_FILES="$CONFIG_FILES pixman-1-uninstalled.pc" ;;
   "Makefile" ) CONFIG_FILES="$CONFIG_FILES Makefile" ;;
   "pixman/Makefile" ) CONFIG_FILES="$CONFIG_FILES pixman/Makefile" ;;
   "pixman/pixman-version.h" ) CONFIG_FILES="$CONFIG_FILES pixman/pixman-version.h" ;;
@@ -20046,6 +20838,7 @@ s,@host@,$host,;t t
 s,@host_cpu@,$host_cpu,;t t
 s,@host_vendor@,$host_vendor,;t t
 s,@host_os@,$host_os,;t t
+s,@SED@,$SED,;t t
 s,@EGREP@,$EGREP,;t t
 s,@LN_S@,$LN_S,;t t
 s,@ECHO@,$ECHO,;t t
@@ -20053,6 +20846,10 @@ s,@AR@,$AR,;t t
 s,@ac_ct_AR@,$ac_ct_AR,;t t
 s,@RANLIB@,$RANLIB,;t t
 s,@ac_ct_RANLIB@,$ac_ct_RANLIB,;t t
+s,@DSYMUTIL@,$DSYMUTIL,;t t
+s,@ac_ct_DSYMUTIL@,$ac_ct_DSYMUTIL,;t t
+s,@NMEDIT@,$NMEDIT,;t t
+s,@ac_ct_NMEDIT@,$ac_ct_NMEDIT,;t t
 s,@CPP@,$CPP,;t t
 s,@CXX@,$CXX,;t t
 s,@CXXFLAGS@,$CXXFLAGS,;t t
@@ -20069,14 +20866,16 @@ s,@PIXMAN_VERSION_MAJOR@,$PIXMAN_VERSION_MAJOR,;t t
 s,@PIXMAN_VERSION_MINOR@,$PIXMAN_VERSION_MINOR,;t t
 s,@PIXMAN_VERSION_MICRO@,$PIXMAN_VERSION_MICRO,;t t
 s,@LT_VERSION_INFO@,$LT_VERSION_INFO,;t t
+s,@PERL@,$PERL,;t t
 s,@USE_MMX_TRUE@,$USE_MMX_TRUE,;t t
 s,@USE_MMX_FALSE@,$USE_MMX_FALSE,;t t
-s,@USE_SSE_TRUE@,$USE_SSE_TRUE,;t t
-s,@USE_SSE_FALSE@,$USE_SSE_FALSE,;t t
 s,@USE_SSE2_TRUE@,$USE_SSE2_TRUE,;t t
 s,@USE_SSE2_FALSE@,$USE_SSE2_FALSE,;t t
 s,@MMX_CFLAGS@,$MMX_CFLAGS,;t t
-s,@SSE_CFLAGS@,$SSE_CFLAGS,;t t
+s,@SSE2_CFLAGS@,$SSE2_CFLAGS,;t t
+s,@VMX_CFLAGS@,$VMX_CFLAGS,;t t
+s,@USE_VMX_TRUE@,$USE_VMX_TRUE,;t t
+s,@USE_VMX_FALSE@,$USE_VMX_FALSE,;t t
 s,@PKG_CONFIG@,$PKG_CONFIG,;t t
 s,@ac_pt_PKG_CONFIG@,$ac_pt_PKG_CONFIG,;t t
 s,@GTK_CFLAGS@,$GTK_CFLAGS,;t t
diff --git a/lib/pixman/configure.ac b/lib/pixman/configure.ac
index 18c744cea..eb8879ca0 100644
--- a/lib/pixman/configure.ac
+++ b/lib/pixman/configure.ac
@@ -24,22 +24,18 @@ AC_PREREQ([2.57])
 
 #   Pixman versioning scheme
 #
-#   - The git version must at all times have an odd MICRO version
-#     number.
+#   - The version in git has an odd MICRO version number
 #
-#   - If you add API, increment the MICRO version to the next largest
-#     odd number.
+#   - Released versions both development and stable have an even MICRO 
+#     version number
 #
-#   - If you release a version that contains new API, then increment
-#     MINOR and set MICRO to 0. 
+#   - Released development versions have an odd MINOR number
 #
-#   - If you release a new version that does not contain new API, then
-#     increment MICRO to the next even number.
+#   - Released stable versions have an event MINOR number
 #
-#   - After doing a release, increment MICRO again to make the version 
-#     number in git odd.
+#   - Versions that break ABI must have a new MAJOR number
 #
-#   - If you break the ABI, then
+#   - If you break the ABI, then at least this must be done:
 #
 #        - increment MAJOR
 #
@@ -54,9 +50,10 @@ AC_PREREQ([2.57])
 #      This ensures that binary incompatible versions can be installed
 #      in parallel.  See http://www106.pair.com/rhp/parallel.html for
 #      more information
+#
 
 m4_define([pixman_major], 0)
-m4_define([pixman_minor], 10)
+m4_define([pixman_minor], 12)
 m4_define([pixman_micro], 0)
 
 m4_define([pixman_version],[pixman_major.pixman_minor.pixman_micro])
@@ -70,7 +67,7 @@ AC_PROG_CC
 AC_PROG_LIBTOOL
 AC_CHECK_FUNCS([getisax])
 AC_C_BIGENDIAN
-
+AC_C_INLINE
 # 
 # We ignore pixman_major in the version here because the major version should
 # always be encoded in the actual library name. Ie., the soname is:
@@ -102,6 +99,12 @@ if test "x$GCC" = "xyes"; then
   *) CFLAGS="$CFLAGS -Wall" ;;
   esac fi changequote([,])dnl
 
+AC_PATH_PROG(PERL, perl, no)
+if test "x$PERL" = xno; then
+    AC_MSG_ERROR([Perl is required to build pixman.])
+fi
+AC_SUBST(PERL)
+
 dnl =========================================================================
 dnl -fvisibility stuff
 
@@ -110,7 +113,7 @@ AC_MSG_CHECKING(for -fvisibility)
 AC_COMPILE_IFELSE([
 #if defined(__GNUC__) && (__GNUC__ >= 4)
 #else
-#error Need GCC 4.0 for visibility
+error Need GCC 4.0 for visibility
 #endif
 int main () { return 0; } 
 ], have_gcc4=yes)
@@ -120,10 +123,25 @@ if test "x$have_gcc4" = "xyes"; then
 fi
 AC_MSG_RESULT($have_gcc4)
 
+have_sunstudio8=no
+AC_MSG_CHECKING([for -xldscope (Sun compilers)])
+AC_COMPILE_IFELSE([
+#if defined(__SUNPRO_C) && (__SUNPRO_C >= 0x550)
+#else
+error Need Sun Studio 8 for visibility
+#endif
+int main () { return 0; } 
+], have_sunstudio8=yes)
+
+if test "x$have_sunstudio8" = "xyes"; then
+   CFLAGS="$CFLAGS -xldscope=hidden"
+fi
+AC_MSG_RESULT($have_sunstudio8)
+
 dnl ===========================================================================
 dnl Check for MMX
 
-MMX_CFLAGS="-mmmx -Winline --param inline-unit-growth=10000 --param large-function-growth=10000"
+MMX_CFLAGS="-mmmx -Winline"
 
 have_mmx_intrinsics=no
 AC_MSG_CHECKING(whether to use MMX intrinsics)
@@ -131,7 +149,7 @@ xserver_save_CFLAGS=$CFLAGS
 CFLAGS="$CFLAGS $MMX_CFLAGS"
 AC_COMPILE_IFELSE([
 #if defined(__GNUC__) && (__GNUC__ < 3 || (__GNUC__ == 3 && __GNUC_MINOR__ < 4))
-#error "Need GCC >= 3.4 for MMX intrinsics"
+error "Need GCC >= 3.4 for MMX intrinsics"
 #endif
 #include <mmintrin.h>
 int main () {
@@ -139,7 +157,15 @@ int main () {
     return _mm_cvtsi64_si32 (v);
 }], have_mmx_intrinsics=yes)
 CFLAGS=$xserver_save_CFLAGS
-AC_MSG_RESULT($have_mmx_intrinsics)
+
+AC_ARG_ENABLE(mmx,
+   [AC_HELP_STRING([--disable-mmx],
+                   [disable MMX fast paths])],
+   [enable_mmx=$enableval], [enable_mmx=auto])
+
+if test $enable_mmx = no ; then
+   have_mmx_intrinsics=disabled
+fi
 
 if test $have_mmx_intrinsics = yes ; then
    AC_DEFINE(USE_MMX, 1, [use MMX compiler intrinsics])
@@ -147,77 +173,124 @@ else
    MMX_CFLAGS=
 fi
 
-AM_CONDITIONAL(USE_MMX, test $have_mmx_intrinsics = yes)
-
-dnl =======================================================
-
-dnl GCC 4.2 when compiling with -msse will generate SSE instructions
-dnl on its own.  This means anything compiled with -mss can only be
-dnl run after a runtime check for SSE.  Unfortunately, since we still
-dnl need to support MMX-but-not-SSE (such as the OLPC), this means we
-dnl can only use SSE when compiling for x86-64 (where SSE is always
-dnl supported).
-
-have_sse_intrinsics=no
-AC_MSG_CHECKING(whether to use SSE intrinsics)
-xserver_save_CFLAGS=$CFLAGS
-CFLAGS="$CFLAGS -msse $MMX_CFLAGS"
-
-AC_COMPILE_IFELSE([
-#if !defined(__amd64__) && !defined(__x86_64__)
-#error "Need x86-64 for SSE"
-#endif
-#include <mmintrin.h>
-#include <xmmintrin.h>
-int main () {
-    __m64 v = _mm_cvtsi32_si64 (1);
-    v = _mm_shuffle_pi16 (v, _MM_SHUFFLE(3, 3, 3, 3));
-    return _mm_cvtsi64_si32 (v);
-}], have_sse_intrinsics=yes)
-CFLAGS=$xserver_save_CFLAGS
-AC_MSG_RESULT($have_sse_intrinsics)
-
-if test $have_sse_intrinsics = yes ; then
-   AC_DEFINE(USE_SSE, 1, [use SSE compiler intrinsics])
-   MMX_CFLAGS="-msse $MMX_CFLAGS"
+AC_MSG_RESULT($have_mmx_intrinsics)
+if test $enable_mmx = yes && test $have_mmx_intrinsics = no ; then
+   AC_MSG_ERROR([MMX intrinsics not detected])
 fi
 
-AM_CONDITIONAL(USE_SSE, test $have_sse_intrinsics = yes)
-
+AM_CONDITIONAL(USE_MMX, test $have_mmx_intrinsics = yes)
 
 dnl ===========================================================================
 dnl Check for SSE2
 
-SSE_CFLAGS="-mmmx -msse2 -Winline --param inline-unit-growth=10000 --param large-function-growth=10000"
+SSE2_CFLAGS="-mmmx -msse2 -Winline"
 
 have_sse2_intrinsics=no
 AC_MSG_CHECKING(whether to use SSE2 intrinsics)
 xserver_save_CFLAGS=$CFLAGS
-CFLAGS="$CFLAGS -msse2 $SSE_CFLAGS"
+CFLAGS="$CFLAGS -msse2 $SSE2_CFLAGS"
 
 AC_COMPILE_IFELSE([
+#if defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 2))
+#   if !defined(__amd64__) && !defined(__x86_64__)
+#      error "Need GCC >= 4.2 for SSE2 intrinsics on x86"
+#   endif
+#endif
 #include <mmintrin.h>
 #include <xmmintrin.h>
+#include <emmintrin.h>
 int main () {
     __m128i a, b, c;
 	c = _mm_xor_si128 (a, b);
     return 0;
 }], have_sse2_intrinsics=yes)
 CFLAGS=$xserver_save_CFLAGS
-AC_MSG_RESULT($have_sse2_intrinsics)
+
+AC_ARG_ENABLE(sse2,
+   [AC_HELP_STRING([--disable-sse2],
+                   [disable SSE2 fast paths])],
+   [enable_sse2=$enableval], [enable_sse2=auto])
+
+if test $enable_sse2 = no ; then
+   have_sse2_intrinsics=disabled
+fi
 
 if test $have_sse2_intrinsics = yes ; then
-   AC_DEFINE(USE_SSE2, 1, [use SSE compiler intrinsics])
+   AC_DEFINE(USE_SSE2, 1, [use SSE2 compiler intrinsics])
+fi
+
+AC_MSG_RESULT($have_sse2_intrinsics)
+if test $enable_sse2 = yes && test $have_sse2_intrinsics = no ; then
+   AC_MSG_ERROR([SSE2 intrinsics not detected])
 fi
 
 AM_CONDITIONAL(USE_SSE2, test $have_sse2_intrinsics = yes)
 
 dnl ========================================================
 AC_SUBST(MMX_CFLAGS)
-AC_SUBST(SSE_CFLAGS)
+AC_SUBST(SSE2_CFLAGS)
+
+dnl Check for VMX/Altivec
+if test -n "`$CC -v 2>&1 | grep version | grep Apple`"; then
+    VMX_CFLAGS="-faltivec"
+else
+    VMX_CFLAGS="-maltivec -mabi=altivec"
+fi
+
+have_vmx_intrinsics=no
+AC_MSG_CHECKING(whether to use VMX/Altivec intrinsics)
+xserver_save_CFLAGS=$CFLAGS
+CFLAGS="$CFLAGS $VMX_CFLAGS"
+AC_COMPILE_IFELSE([
+#if defined(__GNUC__) && (__GNUC__ < 3 || (__GNUC__ == 3 && __GNUC_MINOR__ < 4))
+error "Need GCC >= 3.4 for sane altivec support"
+#endif
+#include <altivec.h>
+int main () {
+    vector unsigned int v = vec_splat_u32 (1);
+    v = vec_sub (v, v);
+    return 0;
+}], have_vmx_intrinsics=yes)
+CFLAGS=$xserver_save_CFLAGS
+
+AC_ARG_ENABLE(vmx,
+   [AC_HELP_STRING([--disable-vmx],
+                   [disable VMX fast paths])],
+   [enable_vmx=$enableval], [enable_vmx=auto])
+
+if test $enable_vmx = no ; then
+   have_vmx_intrinsics=disabled
+fi
+
+if test $have_vmx_intrinsics = yes ; then
+   AC_DEFINE(USE_VMX, 1, [use VMX compiler intrinsics])
+else
+   VMX_CFLAGS=
+fi
+
+AC_MSG_RESULT($have_vmx_intrinsics)
+if test $enable_vmx = yes && test $have_vmx_intrinsics = no ; then
+   AC_MSG_ERROR([VMX intrinsics not detected])
+fi
+
+AC_SUBST(VMX_CFLAGS)
+
+AM_CONDITIONAL(USE_VMX, test $have_vmx_intrinsics = yes)
+
+AC_ARG_ENABLE(gtk,
+   [AC_HELP_STRING([--enable-gtk],
+                   [enable tests using GTK+ [default=auto]])],
+   [enable_gtk=$enableval], [enable_gtk=auto])
+
+PKG_PROG_PKG_CONFIG
+if test $enable_gtk = auto ; then
+   PKG_CHECK_EXISTS([gtk+-2.0], [enable_gtk=yes], [enable_gtk=no])
+fi
+if test $enable_gtk = yes ; then
+   PKG_CHECK_MODULES(GTK, [gtk+-2.0])
+fi
 
-PKG_CHECK_MODULES(GTK, [gtk+-2.0], [HAVE_GTK=yes], [HAVE_GTK=no])
-AM_CONDITIONAL(HAVE_GTK, [test "x$HAVE_GTK" = xyes])
+AM_CONDITIONAL(HAVE_GTK, [test "x$enable_gtk" = xyes])
 
 AC_SUBST(GTK_CFLAGS)
 AC_SUBST(GTK_LIBS)
@@ -225,6 +298,7 @@ AC_SUBST(DEP_CFLAGS)
 AC_SUBST(DEP_LIBS)
 		  
 AC_OUTPUT([pixman-1.pc
+           pixman-1-uninstalled.pc
            Makefile
 	   pixman/Makefile
 	   pixman/pixman-version.h
diff --git a/lib/pixman/ltmain.sh b/lib/pixman/ltmain.sh
index 85faf0734..248cd4047 100644
--- a/lib/pixman/ltmain.sh
+++ b/lib/pixman/ltmain.sh
@@ -1,8 +1,8 @@
 # ltmain.sh - Provide generalized library-building support services.
 # NOTE: Changing this file will not affect anything until you rerun configure.
 #
-# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2003, 2004, 2005
-# Free Software Foundation, Inc.
+# Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2003, 2004, 2005, 2006,
+# 2007, 2008  Free Software Foundation, Inc.
 # Originally by Gordon Matzigkeit <gord@gnu.ai.mit.edu>, 1996
 #
 # This program is free software; you can redistribute it and/or modify
@@ -43,14 +43,22 @@ EXIT_FAILURE=1
 
 PROGRAM=ltmain.sh
 PACKAGE=libtool
-VERSION=1.5.22
-TIMESTAMP=" (1.1220.2.365 2005/12/18 22:14:06)"
-
-# See if we are running on zsh, and set the options which allow our
-# commands through without removal of \ escapes.
-if test -n "${ZSH_VERSION+set}" ; then
+VERSION=1.5.26
+TIMESTAMP=" (1.1220.2.493 2008/02/01 16:58:18)"
+
+# Be Bourne compatible (taken from Autoconf:_AS_BOURNE_COMPATIBLE).
+if test -n "${ZSH_VERSION+set}" && (emulate sh) >/dev/null 2>&1; then
+  emulate sh
+  NULLCMD=:
+  # Zsh 3.x and 4.x performs word splitting on ${1+"$@"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '${1+"$@"}'='"$@"'
   setopt NO_GLOB_SUBST
+else
+  case `(set -o) 2>/dev/null` in *posix*) set -o posix;; esac
 fi
+BIN_SH=xpg4; export BIN_SH # for Tru64
+DUALCASE=1; export DUALCASE # for MKS sh
 
 # Check that we have a working $echo.
 if test "X$1" = X--no-reexec; then
@@ -105,11 +113,19 @@ esac
 # These must not be set unconditionally because not all systems understand
 # e.g. LANG=C (notably SCO).
 # We save the old values to restore during execute mode.
-if test "${LC_ALL+set}" = set; then
-  save_LC_ALL="$LC_ALL"; LC_ALL=C; export LC_ALL
-fi
-if test "${LANG+set}" = set; then
-  save_LANG="$LANG"; LANG=C; export LANG
+lt_env=
+for lt_var in LANG LANGUAGE LC_ALL LC_CTYPE LC_COLLATE LC_MESSAGES
+do
+  eval "if test \"\${$lt_var+set}\" = set; then
+	  save_$lt_var=\$$lt_var
+	  lt_env=\"$lt_var=\$$lt_var \$lt_env\"
+	  $lt_var=C
+	  export $lt_var
+	fi"
+done
+
+if test -n "$lt_env"; then
+  lt_env="env $lt_env"
 fi
 
 # Make sure IFS has a sensible default
@@ -136,6 +152,8 @@ duplicate_deps=no
 preserve_args=
 lo2o="s/\\.lo\$/.${objext}/"
 o2lo="s/\\.${objext}\$/.lo/"
+extracted_archives=
+extracted_serial=0
 
 #####################################
 # Shell function definitions:
@@ -196,7 +214,13 @@ func_win32_libid ()
     if eval $OBJDUMP -f $1 | $SED -e '10q' 2>/dev/null | \
       $EGREP -e 'file format pe-i386(.*architecture: i386)?' >/dev/null ; then
       win32_nmres=`eval $NM -f posix -A $1 | \
-	$SED -n -e '1,100{/ I /{s,.*,import,;p;q;};}'`
+	$SED -n -e '1,100{
+		/ I /{
+			s,.*,import,
+			p
+			q
+			}
+		}'`
       case $win32_nmres in
       import*)  win32_libid_type="x86 archive import";;
       *)        win32_libid_type="x86 archive static";;
@@ -327,7 +351,17 @@ func_extract_archives ()
 	*) my_xabs=`pwd`"/$my_xlib" ;;
       esac
       my_xlib=`$echo "X$my_xlib" | $Xsed -e 's%^.*/%%'`
-      my_xdir="$my_gentop/$my_xlib"
+      my_xlib_u=$my_xlib
+      while :; do
+        case " $extracted_archives " in
+	*" $my_xlib_u "*)
+	  extracted_serial=`expr $extracted_serial + 1`
+	  my_xlib_u=lt$extracted_serial-$my_xlib ;;
+	*) break ;;
+	esac
+      done
+      extracted_archives="$extracted_archives $my_xlib_u"
+      my_xdir="$my_gentop/$my_xlib_u"
 
       $show "${rm}r $my_xdir"
       $run ${rm}r "$my_xdir"
@@ -454,11 +488,12 @@ do
     ;;
 
   --version)
-    $echo "$PROGRAM (GNU $PACKAGE) $VERSION$TIMESTAMP"
-    $echo
-    $echo "Copyright (C) 2005  Free Software Foundation, Inc."
-    $echo "This is free software; see the source for copying conditions.  There is NO"
-    $echo "warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE."
+    echo "\
+$PROGRAM (GNU $PACKAGE) $VERSION$TIMESTAMP
+
+Copyright (C) 2008  Free Software Foundation, Inc.
+This is free software; see the source for copying conditions.  There is NO
+warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE."
     exit $?
     ;;
 
@@ -755,9 +790,11 @@ if test -z "$show_help"; then
     *.class) xform=class ;;
     *.cpp) xform=cpp ;;
     *.cxx) xform=cxx ;;
-    *.f90) xform=f90 ;;
+    *.[fF][09]?) xform=[fF][09]. ;;
     *.for) xform=for ;;
     *.java) xform=java ;;
+    *.obj) xform=obj ;;
+    *.sx) xform=sx ;;
     esac
 
     libobj=`$echo "X$libobj" | $Xsed -e "s/\.$xform$/.lo/"`
@@ -926,7 +963,7 @@ EOF
       $run $rm "$lobj" "$output_obj"
 
       $show "$command"
-      if $run eval "$command"; then :
+      if $run eval $lt_env "$command"; then :
       else
 	test -n "$output_obj" && $run $rm $removelist
 	exit $EXIT_FAILURE
@@ -998,7 +1035,7 @@ EOF
       command="$command$suppress_output"
       $run $rm "$obj" "$output_obj"
       $show "$command"
-      if $run eval "$command"; then :
+      if $run eval $lt_env "$command"; then :
       else
 	$run $rm $removelist
 	exit $EXIT_FAILURE
@@ -1131,6 +1168,7 @@ EOF
     thread_safe=no
     vinfo=
     vinfo_number=no
+    single_module="${wl}-single_module"
 
     func_infer_tag $base_compile
 
@@ -1138,8 +1176,9 @@ EOF
     for arg
     do
       case $arg in
-      -all-static | -static)
-	if test "X$arg" = "X-all-static"; then
+      -all-static | -static | -static-libtool-libs)
+	case $arg in
+	-all-static)
 	  if test "$build_libtool_libs" = yes && test -z "$link_static_flag"; then
 	    $echo "$modename: warning: complete static linking is impossible in this configuration" 1>&2
 	  fi
@@ -1147,12 +1186,20 @@ EOF
 	    dlopen_self=$dlopen_self_static
 	  fi
 	  prefer_static_libs=yes
-	else
+	  ;;
+	-static)
 	  if test -z "$pic_flag" && test -n "$link_static_flag"; then
 	    dlopen_self=$dlopen_self_static
 	  fi
 	  prefer_static_libs=built
-	fi
+	  ;;
+	-static-libtool-libs)
+	  if test -z "$pic_flag" && test -n "$link_static_flag"; then
+	    dlopen_self=$dlopen_self_static
+	  fi
+	  prefer_static_libs=yes
+	  ;;
+	esac
 	build_libtool_libs=no
 	build_old_libs=yes
 	break
@@ -1600,7 +1647,7 @@ EOF
 	continue
 	;;
 
-     -mt|-mthreads|-kthread|-Kthread|-pthread|-pthreads|--thread-safe)
+     -mt|-mthreads|-kthread|-Kthread|-pthread|-pthreads|--thread-safe|-threads)
 	compiler_flags="$compiler_flags $arg"
 	compile_command="$compile_command $arg"
 	finalize_command="$finalize_command $arg"
@@ -1608,6 +1655,11 @@ EOF
 	continue
 	;;
 
+      -multi_module)
+	single_module="${wl}-multi_module"
+	continue
+	;;
+
       -module)
 	module=yes
 	continue
@@ -1621,10 +1673,11 @@ EOF
       # -m* pass through architecture-specific compiler args for GCC
       # -m*, -t[45]*, -txscale* pass through architecture-specific
       # compiler args for GCC
-      # -pg pass through profiling flag for GCC
+      # -p, -pg, --coverage, -fprofile-* pass through profiling flag for GCC
+      # -F/path gives path to uninstalled frameworks, gcc on darwin
       # @file GCC response files
-      -64|-mips[0-9]|-r[0-9][0-9]*|-xarch=*|-xtarget=*|+DA*|+DD*|-q*|-m*|-pg| \
-      -t[45]*|-txscale*|@*)
+      -64|-mips[0-9]|-r[0-9][0-9]*|-xarch=*|-xtarget=*|+DA*|+DD*|-q*|-m*| \
+      -t[45]*|-txscale*|-p|-pg|--coverage|-fprofile-*|-F*|@*)
 
 	# Unknown arguments in both finalize_command and compile_command need
 	# to be aesthetically quoted because they are evaled later.
@@ -1652,9 +1705,9 @@ EOF
 
       -no-install)
 	case $host in
-	*-*-cygwin* | *-*-mingw* | *-*-pw32* | *-*-os2*)
+	*-*-cygwin* | *-*-mingw* | *-*-pw32* | *-*-os2* | *-*-darwin*)
 	  # The PATH hackery in wrapper scripts is required on Windows
-	  # in order for the loader to find any dlls it needs.
+	  # and Darwin in order for the loader to find any dlls it needs.
 	  $echo "$modename: warning: \`-no-install' is ignored for $host" 1>&2
 	  $echo "$modename: warning: assuming \`-no-fast-install' instead" 1>&2
 	  fast_install=no
@@ -1713,7 +1766,7 @@ EOF
 	continue
 	;;
 
-      -static)
+      -static | -static-libtool-libs)
 	# The effects of -static are defined in a previous loop.
 	# We used to do the same as -all-static on platforms that
 	# didn't have a PIC flag, but the assumption that the effects
@@ -2106,7 +2159,7 @@ EOF
 	lib=
 	found=no
 	case $deplib in
-	-mt|-mthreads|-kthread|-Kthread|-pthread|-pthreads|--thread-safe)
+	-mt|-mthreads|-kthread|-Kthread|-pthread|-pthreads|--thread-safe|-threads)
 	  if test "$linkmode,$pass" = "prog,link"; then
 	    compile_deplibs="$deplib $compile_deplibs"
 	    finalize_deplibs="$deplib $finalize_deplibs"
@@ -2122,7 +2175,12 @@ EOF
 	    continue
 	  fi
 	  name=`$echo "X$deplib" | $Xsed -e 's/^-l//'`
-	  for searchdir in $newlib_search_path $lib_search_path $sys_lib_search_path $shlib_search_path; do
+	  if test "$linkmode" = lib; then
+	    searchdirs="$newlib_search_path $lib_search_path $compiler_lib_search_dirs $sys_lib_search_path $shlib_search_path"
+	  else
+	    searchdirs="$newlib_search_path $lib_search_path $sys_lib_search_path $shlib_search_path"
+	  fi
+	  for searchdir in $searchdirs; do
 	    for search_ext in .la $std_shrext .so .a; do
 	      # Search the libtool library
 	      lib="$searchdir/lib${name}${search_ext}"
@@ -2503,7 +2561,9 @@ EOF
 
 	if test "$linkmode,$pass" = "prog,link"; then
 	  if test -n "$library_names" &&
-	     { test "$prefer_static_libs" = no || test -z "$old_library"; }; then
+	     { { test "$prefer_static_libs" = no ||
+		 test "$prefer_static_libs,$installed" = "built,yes"; } ||
+	       test -z "$old_library"; }; then
 	    # We need to hardcode the library path
 	    if test -n "$shlibpath_var" && test -z "$avoidtemprpath" ; then
 	      # Make sure the rpath contains only unique directories.
@@ -2916,12 +2976,18 @@ EOF
 		  # we do not want to link against static libs,
 		  # but need to link against shared
 		  eval deplibrary_names=`${SED} -n -e 's/^library_names=\(.*\)$/\1/p' $deplib`
+		  eval deplibdir=`${SED} -n -e 's/^libdir=\(.*\)$/\1/p' $deplib`
 		  if test -n "$deplibrary_names" ; then
 		    for tmp in $deplibrary_names ; do
 		      depdepl=$tmp
 		    done
-		    if test -f "$path/$depdepl" ; then
+		    if test -f "$deplibdir/$depdepl" ; then
+		      depdepl="$deplibdir/$depdepl"
+	      	    elif test -f "$path/$depdepl" ; then
 		      depdepl="$path/$depdepl"
+		    else
+		      # Can't find it, oh well...
+		      depdepl=
 		    fi
 		    # do not add paths which are already there
 		    case " $newlib_search_path " in
@@ -3069,9 +3135,10 @@ EOF
 
     case $linkmode in
     oldlib)
-      if test -n "$deplibs"; then
-	$echo "$modename: warning: \`-l' and \`-L' are ignored for archives" 1>&2
-      fi
+      case " $deplibs" in
+      *\ -l* | *\ -L*)
+	$echo "$modename: warning: \`-l' and \`-L' are ignored for archives" 1>&2 ;;
+      esac
 
       if test -n "$dlfiles$dlprefiles" || test "$dlself" != no; then
 	$echo "$modename: warning: \`-dlopen' is ignored for archives" 1>&2
@@ -3199,7 +3266,7 @@ EOF
 	  # which has an extra 1 added just for fun
 	  #
 	  case $version_type in
-	  darwin|linux|osf|windows)
+	  darwin|linux|osf|windows|none)
 	    current=`expr $number_major + $number_minor`
 	    age="$number_minor"
 	    revision="$number_revision"
@@ -3210,9 +3277,10 @@ EOF
 	    age="0"
 	    ;;
 	  irix|nonstopux)
-	    current=`expr $number_major + $number_minor - 1`
+	    current=`expr $number_major + $number_minor`
 	    age="$number_minor"
 	    revision="$number_minor"
+	    lt_irix_increment=no
 	    ;;
 	  esac
 	  ;;
@@ -3271,7 +3339,8 @@ EOF
 	  versuffix="$major.$age.$revision"
 	  # Darwin ld doesn't like 0 for these options...
 	  minor_current=`expr $current + 1`
-	  verstring="${wl}-compatibility_version ${wl}$minor_current ${wl}-current_version ${wl}$minor_current.$revision"
+	  xlcverstring="${wl}-compatibility_version ${wl}$minor_current ${wl}-current_version ${wl}$minor_current.$revision"
+	  verstring="-compatibility_version $minor_current -current_version $minor_current.$revision"
 	  ;;
 
 	freebsd-aout)
@@ -3285,8 +3354,11 @@ EOF
 	  ;;
 
 	irix | nonstopux)
-	  major=`expr $current - $age + 1`
-
+	  if test "X$lt_irix_increment" = "Xno"; then
+	    major=`expr $current - $age`
+	  else
+	    major=`expr $current - $age + 1`
+	  fi
 	  case $version_type in
 	    nonstopux) verstring_prefix=nonstopux ;;
 	    *)         verstring_prefix=sgi ;;
@@ -3437,11 +3509,11 @@ EOF
       fi
 
       # Eliminate all temporary directories.
-      for path in $notinst_path; do
-	lib_search_path=`$echo "$lib_search_path " | ${SED} -e "s% $path % %g"`
-	deplibs=`$echo "$deplibs " | ${SED} -e "s% -L$path % %g"`
-	dependency_libs=`$echo "$dependency_libs " | ${SED} -e "s% -L$path % %g"`
-      done
+      #for path in $notinst_path; do
+      #	lib_search_path=`$echo "$lib_search_path " | ${SED} -e "s% $path % %g"`
+      #	deplibs=`$echo "$deplibs " | ${SED} -e "s% -L$path % %g"`
+      #	dependency_libs=`$echo "$dependency_libs " | ${SED} -e "s% -L$path % %g"`
+      #done
 
       if test -n "$xrpath"; then
 	# If the user specified any rpath flags, then add them.
@@ -3542,13 +3614,12 @@ EOF
 	  int main() { return 0; }
 EOF
 	  $rm conftest
-	  $LTCC $LTCFLAGS -o conftest conftest.c $deplibs
-	  if test "$?" -eq 0 ; then
+	  if $LTCC $LTCFLAGS -o conftest conftest.c $deplibs; then
 	    ldd_output=`ldd conftest`
 	    for i in $deplibs; do
 	      name=`expr $i : '-l\(.*\)'`
 	      # If $name is empty we are operating on a -L argument.
-              if test "$name" != "" && test "$name" -ne "0"; then
+              if test "$name" != "" && test "$name" != "0"; then
 		if test "X$allow_libtool_libs_with_static_runtimes" = "Xyes" ; then
 		  case " $predeps $postdeps " in
 		  *" $i "*)
@@ -3587,9 +3658,7 @@ EOF
 	      # If $name is empty we are operating on a -L argument.
               if test "$name" != "" && test "$name" != "0"; then
 		$rm conftest
-		$LTCC $LTCFLAGS -o conftest conftest.c $i
-		# Did it work?
-		if test "$?" -eq 0 ; then
+		if $LTCC $LTCFLAGS -o conftest conftest.c $i; then
 		  ldd_output=`ldd conftest`
 		  if test "X$allow_libtool_libs_with_static_runtimes" = "Xyes" ; then
 		    case " $predeps $postdeps " in
@@ -3621,7 +3690,7 @@ EOF
 		  droppeddeps=yes
 		  $echo
 		  $echo "*** Warning!  Library $i is needed by this library but I was not able to"
-		  $echo "***  make it link in!  You will probably need to install it or some"
+		  $echo "*** make it link in!  You will probably need to install it or some"
 		  $echo "*** library that it depends on before this library will be fully"
 		  $echo "*** functional.  Installing it before continuing would be even better."
 		fi
@@ -3907,7 +3976,10 @@ EOF
 	     test -n "$hardcode_libdirs"; then
 	    libdir="$hardcode_libdirs"
 	    if test -n "$hardcode_libdir_flag_spec_ld"; then
-	      eval dep_rpath=\"$hardcode_libdir_flag_spec_ld\"
+	      case $archive_cmds in
+	      *\$LD*) eval dep_rpath=\"$hardcode_libdir_flag_spec_ld\" ;;
+	      *)      eval dep_rpath=\"$hardcode_libdir_flag_spec\" ;;
+	      esac
 	    else
 	      eval dep_rpath=\"$hardcode_libdir_flag_spec\"
 	    fi
@@ -4217,9 +4289,10 @@ EOF
       ;;
 
     obj)
-      if test -n "$deplibs"; then
-	$echo "$modename: warning: \`-l' and \`-L' are ignored for objects" 1>&2
-      fi
+      case " $deplibs" in
+      *\ -l* | *\ -L*)
+	$echo "$modename: warning: \`-l' and \`-L' are ignored for objects" 1>&2 ;;
+      esac
 
       if test -n "$dlfiles$dlprefiles" || test "$dlself" != no; then
 	$echo "$modename: warning: \`-dlopen' is ignored for objects" 1>&2
@@ -4266,12 +4339,14 @@ EOF
       reload_conv_objs=
       gentop=
       # reload_cmds runs $LD directly, so let us get rid of
-      # -Wl from whole_archive_flag_spec
+      # -Wl from whole_archive_flag_spec and hope we can get by with
+      # turning comma into space..
       wl=
 
       if test -n "$convenience"; then
 	if test -n "$whole_archive_flag_spec"; then
-	  eval reload_conv_objs=\"\$reload_objs $whole_archive_flag_spec\"
+	  eval tmp_whole_archive_flags=\"$whole_archive_flag_spec\"
+	  reload_conv_objs=$reload_objs\ `$echo "X$tmp_whole_archive_flags" | $Xsed -e 's|,| |g'`
 	else
 	  gentop="$output_objdir/${obj}x"
 	  generated="$generated $gentop"
@@ -4719,16 +4794,16 @@ static const void *lt_preloaded_setup() {
           case $host in
           *cygwin* | *mingw* )
             if test -f "$output_objdir/${outputname}.def" ; then
-              compile_command=`$echo "X$compile_command" | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}.def $output_objdir/${outputname}S.${objext}%"`
-              finalize_command=`$echo "X$finalize_command" | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}.def $output_objdir/${outputname}S.${objext}%"`
+              compile_command=`$echo "X$compile_command" | $SP2NL | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}.def $output_objdir/${outputname}S.${objext}%" | $NL2SP`
+              finalize_command=`$echo "X$finalize_command" | $SP2NL | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}.def $output_objdir/${outputname}S.${objext}%" | $NL2SP`
             else
-              compile_command=`$echo "X$compile_command" | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}S.${objext}%"`
-              finalize_command=`$echo "X$finalize_command" | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}S.${objext}%"`
+              compile_command=`$echo "X$compile_command" | $SP2NL | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}S.${objext}%" | $NL2SP`
+              finalize_command=`$echo "X$finalize_command" | $SP2NL | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}S.${objext}%" | $NL2SP`
              fi
             ;;
           * )
-            compile_command=`$echo "X$compile_command" | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}S.${objext}%"`
-            finalize_command=`$echo "X$finalize_command" | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}S.${objext}%"`
+            compile_command=`$echo "X$compile_command" | $SP2NL | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}S.${objext}%" | $NL2SP`
+            finalize_command=`$echo "X$finalize_command" | $SP2NL | $Xsed -e "s%@SYMFILE@%$output_objdir/${outputname}S.${objext}%" | $NL2SP`
             ;;
           esac
 	  ;;
@@ -4743,13 +4818,13 @@ static const void *lt_preloaded_setup() {
 	# really was required.
 
 	# Nullify the symbol file.
-	compile_command=`$echo "X$compile_command" | $Xsed -e "s% @SYMFILE@%%"`
-	finalize_command=`$echo "X$finalize_command" | $Xsed -e "s% @SYMFILE@%%"`
+	compile_command=`$echo "X$compile_command" | $SP2NL | $Xsed -e "s% @SYMFILE@%%" | $NL2SP`
+	finalize_command=`$echo "X$finalize_command" | $SP2NL | $Xsed -e "s% @SYMFILE@%%" | $NL2SP`
       fi
 
       if test "$need_relink" = no || test "$build_libtool_libs" != yes; then
 	# Replace the output file specification.
-	compile_command=`$echo "X$compile_command" | $Xsed -e 's%@OUTPUT@%'"$output"'%g'`
+	compile_command=`$echo "X$compile_command" | $SP2NL | $Xsed -e 's%@OUTPUT@%'"$output"'%g' | $NL2SP`
 	link_command="$compile_command$compile_rpath"
 
 	# We have no uninstalled library dependencies, so finalize right now.
@@ -4836,7 +4911,7 @@ static const void *lt_preloaded_setup() {
 	if test "$fast_install" != no; then
 	  link_command="$finalize_var$compile_command$finalize_rpath"
 	  if test "$fast_install" = yes; then
-	    relink_command=`$echo "X$compile_var$compile_command$compile_rpath" | $Xsed -e 's%@OUTPUT@%\$progdir/\$file%g'`
+	    relink_command=`$echo "X$compile_var$compile_command$compile_rpath" | $SP2NL | $Xsed -e 's%@OUTPUT@%\$progdir/\$file%g' | $NL2SP`
 	  else
 	    # fast_install is set to needless
 	    relink_command=
@@ -4873,7 +4948,7 @@ static const void *lt_preloaded_setup() {
 	  fi
 	done
 	relink_command="(cd `pwd`; $relink_command)"
-	relink_command=`$echo "X$relink_command" | $Xsed -e "$sed_quote_subst"`
+	relink_command=`$echo "X$relink_command" | $SP2NL | $Xsed -e "$sed_quote_subst" | $NL2SP`
       fi
 
       # Quote $echo for shipping.
@@ -5280,6 +5355,20 @@ EOF
 Xsed='${SED} -e 1s/^X//'
 sed_quote_subst='$sed_quote_subst'
 
+# Be Bourne compatible (taken from Autoconf:_AS_BOURNE_COMPATIBLE).
+if test -n \"\${ZSH_VERSION+set}\" && (emulate sh) >/dev/null 2>&1; then
+  emulate sh
+  NULLCMD=:
+  # Zsh 3.x and 4.x performs word splitting on \${1+\"\$@\"}, which
+  # is contrary to our usage.  Disable this feature.
+  alias -g '\${1+\"\$@\"}'='\"\$@\"'
+  setopt NO_GLOB_SUBST
+else
+  case \`(set -o) 2>/dev/null\` in *posix*) set -o posix;; esac
+fi
+BIN_SH=xpg4; export BIN_SH # for Tru64
+DUALCASE=1; export DUALCASE # for MKS sh
+
 # The HP-UX ksh and POSIX shell print the target directory to stdout
 # if CDPATH is set.
 (unset CDPATH) >/dev/null 2>&1 && unset CDPATH
@@ -5422,7 +5511,7 @@ else
 	  ;;
 	esac
 	$echo >> $output "\
-      \$echo \"\$0: cannot exec \$program \${1+\"\$@\"}\"
+      \$echo \"\$0: cannot exec \$program \$*\"
       exit $EXIT_FAILURE
     fi
   else
@@ -5608,7 +5697,7 @@ fi\
       done
       # Quote the link command for shipping.
       relink_command="(cd `pwd`; $SHELL $progpath $preserve_args --mode=relink $libtool_args @inst_prefix_dir@)"
-      relink_command=`$echo "X$relink_command" | $Xsed -e "$sed_quote_subst"`
+      relink_command=`$echo "X$relink_command" | $SP2NL | $Xsed -e "$sed_quote_subst" | $NL2SP`
       if test "$hardcode_automatic" = yes ; then
 	relink_command=
       fi
@@ -5957,9 +6046,9 @@ relink_command=\"$relink_command\""
 
 	  if test -n "$inst_prefix_dir"; then
 	    # Stick the inst_prefix_dir data into the link command.
-	    relink_command=`$echo "$relink_command" | $SED "s%@inst_prefix_dir@%-inst-prefix-dir $inst_prefix_dir%"`
+	    relink_command=`$echo "$relink_command" | $SP2NL | $SED "s%@inst_prefix_dir@%-inst-prefix-dir $inst_prefix_dir%" | $NL2SP`
 	  else
-	    relink_command=`$echo "$relink_command" | $SED "s%@inst_prefix_dir@%%"`
+	    relink_command=`$echo "$relink_command" | $SP2NL | $SED "s%@inst_prefix_dir@%%" | $NL2SP`
 	  fi
 
 	  $echo "$modename: warning: relinking \`$file'" 1>&2
@@ -6168,7 +6257,7 @@ relink_command=\"$relink_command\""
 	      file=`$echo "X$file$stripped_ext" | $Xsed -e 's%^.*/%%'`
 	      outputname="$tmpdir/$file"
 	      # Replace the output file specification.
-	      relink_command=`$echo "X$relink_command" | $Xsed -e 's%@OUTPUT@%'"$outputname"'%g'`
+	      relink_command=`$echo "X$relink_command" | $SP2NL | $Xsed -e 's%@OUTPUT@%'"$outputname"'%g' | $NL2SP`
 
 	      $show "$relink_command"
 	      if $run eval "$relink_command"; then :
@@ -6345,8 +6434,10 @@ relink_command=\"$relink_command\""
 	if test -f "$dir/$objdir/$dlname"; then
 	  dir="$dir/$objdir"
 	else
-	  $echo "$modename: cannot find \`$dlname' in \`$dir' or \`$dir/$objdir'" 1>&2
-	  exit $EXIT_FAILURE
+	  if test ! -f "$dir/$dlname"; then
+	    $echo "$modename: cannot find \`$dlname' in \`$dir' or \`$dir/$objdir'" 1>&2
+	    exit $EXIT_FAILURE
+	  fi
 	fi
 	;;
 
@@ -6410,12 +6501,12 @@ relink_command=\"$relink_command\""
       fi
 
       # Restore saved environment variables
-      if test "${save_LC_ALL+set}" = set; then
-	LC_ALL="$save_LC_ALL"; export LC_ALL
-      fi
-      if test "${save_LANG+set}" = set; then
-	LANG="$save_LANG"; export LANG
-      fi
+      for lt_var in LANG LANGUAGE LC_ALL LC_CTYPE LC_COLLATE LC_MESSAGES
+      do
+	eval "if test \"\${save_$lt_var+set}\" = set; then
+		$lt_var=\$save_$lt_var; export $lt_var
+	      fi"
+      done
 
       # Now prepare to actually exec the command.
       exec_cmd="\$cmd$args"
@@ -6772,9 +6863,9 @@ The following components of LINK-COMMAND are treated specially:
   -dlpreopen FILE   link in FILE and add its symbols to lt_preloaded_symbols
   -export-dynamic   allow symbols from OUTPUT-FILE to be resolved with dlsym(3)
   -export-symbols SYMFILE
-		    try to export only the symbols listed in SYMFILE
+                    try to export only the symbols listed in SYMFILE
   -export-symbols-regex REGEX
-		    try to export only the symbols matching REGEX
+                    try to export only the symbols matching REGEX
   -LLIBDIR          search LIBDIR for required installed libraries
   -lNAME            OUTPUT-FILE requires the installed library libNAME
   -module           build a library that can dlopened
@@ -6788,9 +6879,11 @@ The following components of LINK-COMMAND are treated specially:
   -release RELEASE  specify package release information
   -rpath LIBDIR     the created library will eventually be installed in LIBDIR
   -R[ ]LIBDIR       add LIBDIR to the runtime path of programs and libraries
-  -static           do not do any dynamic linking of libtool libraries
+  -static           do not do any dynamic linking of uninstalled libtool libraries
+  -static-libtool-libs
+                    do not do any dynamic linking of libtool libraries
   -version-info CURRENT[:REVISION[:AGE]]
-		    specify library version info [each variable defaults to 0]
+                    specify library version info [each variable defaults to 0]
 
 All other options (arguments beginning with \`-') are ignored.
 
diff --git a/lib/pixman/pixman-1-uninstalled.pc.in b/lib/pixman/pixman-1-uninstalled.pc.in
new file mode 100644
index 000000000..9a2afa1a5
--- /dev/null
+++ b/lib/pixman/pixman-1-uninstalled.pc.in
@@ -0,0 +1,11 @@
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+includedir=@includedir@
+
+Name: Pixman
+Description: The pixman library (version 1)
+Version: @PACKAGE_VERSION@
+Cflags: -I${pc_top_builddir}/${pcfiledir}/pixman
+Libs: ${pc_top_builddir}/${pcfiledir}/libpixman-1-so.a
+
diff --git a/lib/pixman/pixman/Makefile.am b/lib/pixman/pixman/Makefile.am
index 190df4c57..399fb7d5d 100644
--- a/lib/pixman/pixman/Makefile.am
+++ b/lib/pixman/pixman/Makefile.am
@@ -1,14 +1,19 @@
 lib_LTLIBRARIES = libpixman-1.la
-libpixman_1_la_LDFLAGS = -version-info $(LT_VERSION_INFO)
+libpixman_1_la_LDFLAGS = -version-info $(LT_VERSION_INFO) -no-undefined
 libpixman_1_la_LIBADD = @DEP_LIBS@ -lm
+libpixman_1_la_CFLAGS = -DPIXMAN_DISABLE_DEPRECATED
 libpixman_1_la_SOURCES =		\
 	pixman.h			\
 	pixman-access.c			\
 	pixman-access-accessors.c	\
-	pixman-region.c			\
+	pixman-region16.c		\
+	pixman-region32.c		\
 	pixman-private.h		\
 	pixman-image.c			\
-	pixman-combine.c		\
+	pixman-combine32.c		\
+	pixman-combine32.h		\
+	pixman-combine64.c		\
+	pixman-combine64.h		\
 	pixman-compose.c		\
 	pixman-compose-accessors.c	\
 	pixman-pict.c			\
@@ -27,7 +32,18 @@ libpixmanincludedir = $(includedir)/pixman-1/
 libpixmaninclude_HEADERS = pixman.h pixman-version.h
 noinst_LTLIBRARIES = 
 
-EXTRA_DIST = Makefile.win32
+pixman-combine32.c : combine.inc pixman-combine32.h combine.pl
+	$(PERL) $(srcdir)/combine.pl 8 < $(srcdir)/combine.inc > $@ || ($(RM) $@; exit 1)
+pixman-combine32.h : combine.h.inc combine.pl
+	$(PERL) $(srcdir)/combine.pl 8 < $(srcdir)/combine.h.inc > $@ || ($(RM) $@; exit 1)
+
+pixman-combine64.c : combine.inc pixman-combine64.h combine.pl
+	$(PERL) $(srcdir)/combine.pl 16 < $(srcdir)/combine.inc > $@ || ($(RM) $@; exit 1)
+pixman-combine64.h : combine.h.inc combine.pl
+	$(PERL) $(srcdir)/combine.pl 16 < $(srcdir)/combine.h.inc > $@ || ($(RM) $@; exit 1)
+
+EXTRA_DIST = Makefile.win32 combine.inc combine.pl pixman-region.c combine.h.inc
+CLEANFILES = pixman-combine32.c pixman-combine64.c pixman-combine32.h pixman-combine64.h
 
 # mmx code
 if USE_MMX
@@ -40,15 +56,26 @@ libpixman_mmx_la_LIBADD = $(DEP_LIBS)
 libpixman_1_la_LIBADD += libpixman-mmx.la
 endif
 
+# vmx code
+if USE_VMX
+noinst_LTLIBRARIES += libpixman-vmx.la
+libpixman_vmx_la_SOURCES = \
+	pixman-vmx.c \
+	pixman-vmx.h \
+	pixman-combine32.h
+libpixman_vmx_la_CFLAGS = $(DEP_CFLAGS) $(VMX_CFLAGS)
+libpixman_vmx_la_LIBADD = $(DEP_LIBS)
+libpixman_1_la_LIBADD += libpixman-vmx.la
+endif
 
 # sse2 code
 if USE_SSE2
-noinst_LTLIBRARIES += libpixman-sse.la
-libpixman_sse_la_SOURCES = \
-	pixman-sse.c \
-	pixman-sse.h
-libpixman_sse_la_CFLAGS = $(DEP_CFLAGS) $(SSE_CFLAGS)
-libpixman_sse_la_LIBADD = $(DEP_LIBS)
-libpixman_1_la_LIBADD += libpixman-sse.la
+noinst_LTLIBRARIES += libpixman-sse2.la
+libpixman_sse2_la_SOURCES = \
+	pixman-sse2.c \
+	pixman-sse2.h
+libpixman_sse2_la_CFLAGS = $(DEP_CFLAGS) $(SSE2_CFLAGS)
+libpixman_sse2_la_LIBADD = $(DEP_LIBS)
+libpixman_1_la_LIBADD += libpixman-sse2.la
 endif
 
diff --git a/lib/pixman/pixman/Makefile.in b/lib/pixman/pixman/Makefile.in
index d8e122e6e..02206f868 100644
--- a/lib/pixman/pixman/Makefile.in
+++ b/lib/pixman/pixman/Makefile.in
@@ -42,9 +42,13 @@ host_triplet = @host@
 @USE_MMX_TRUE@am__append_1 = libpixman-mmx.la
 @USE_MMX_TRUE@am__append_2 = libpixman-mmx.la
 
+# vmx code
+@USE_VMX_TRUE@am__append_3 = libpixman-vmx.la
+@USE_VMX_TRUE@am__append_4 = libpixman-vmx.la
+
 # sse2 code
-@USE_SSE2_TRUE@am__append_3 = libpixman-sse.la
-@USE_SSE2_TRUE@am__append_4 = libpixman-sse.la
+@USE_SSE2_TRUE@am__append_5 = libpixman-sse2.la
+@USE_SSE2_TRUE@am__append_6 = libpixman-sse2.la
 subdir = pixman
 DIST_COMMON = $(libpixmaninclude_HEADERS) $(srcdir)/Makefile.am \
 	$(srcdir)/Makefile.in $(srcdir)/pixman-version.h.in
@@ -66,30 +70,48 @@ am__installdirs = "$(DESTDIR)$(libdir)" \
 libLTLIBRARIES_INSTALL = $(INSTALL)
 LTLIBRARIES = $(lib_LTLIBRARIES) $(noinst_LTLIBRARIES)
 @USE_MMX_TRUE@am__DEPENDENCIES_1 = libpixman-mmx.la
-@USE_SSE2_TRUE@am__DEPENDENCIES_2 = libpixman-sse.la
+@USE_VMX_TRUE@am__DEPENDENCIES_2 = libpixman-vmx.la
+@USE_SSE2_TRUE@am__DEPENDENCIES_3 = libpixman-sse2.la
 libpixman_1_la_DEPENDENCIES = $(am__DEPENDENCIES_1) \
-	$(am__DEPENDENCIES_2)
-am_libpixman_1_la_OBJECTS = pixman-access.lo \
-	pixman-access-accessors.lo pixman-region.lo pixman-image.lo \
-	pixman-combine.lo pixman-compose.lo \
-	pixman-compose-accessors.lo pixman-pict.lo pixman-source.lo \
-	pixman-transformed.lo pixman-transformed-accessors.lo \
-	pixman-utils.lo pixman-edge.lo pixman-edge-accessors.lo \
-	pixman-trap.lo pixman-compute-region.lo pixman-timer.lo
+	$(am__DEPENDENCIES_2) $(am__DEPENDENCIES_3)
+am_libpixman_1_la_OBJECTS = libpixman_1_la-pixman-access.lo \
+	libpixman_1_la-pixman-access-accessors.lo \
+	libpixman_1_la-pixman-region16.lo \
+	libpixman_1_la-pixman-region32.lo \
+	libpixman_1_la-pixman-image.lo \
+	libpixman_1_la-pixman-combine32.lo \
+	libpixman_1_la-pixman-combine64.lo \
+	libpixman_1_la-pixman-compose.lo \
+	libpixman_1_la-pixman-compose-accessors.lo \
+	libpixman_1_la-pixman-pict.lo libpixman_1_la-pixman-source.lo \
+	libpixman_1_la-pixman-transformed.lo \
+	libpixman_1_la-pixman-transformed-accessors.lo \
+	libpixman_1_la-pixman-utils.lo libpixman_1_la-pixman-edge.lo \
+	libpixman_1_la-pixman-edge-accessors.lo \
+	libpixman_1_la-pixman-trap.lo \
+	libpixman_1_la-pixman-compute-region.lo \
+	libpixman_1_la-pixman-timer.lo
 libpixman_1_la_OBJECTS = $(am_libpixman_1_la_OBJECTS)
-am__DEPENDENCIES_3 =
-@USE_MMX_TRUE@libpixman_mmx_la_DEPENDENCIES = $(am__DEPENDENCIES_3)
+am__DEPENDENCIES_4 =
+@USE_MMX_TRUE@libpixman_mmx_la_DEPENDENCIES = $(am__DEPENDENCIES_4)
 am__libpixman_mmx_la_SOURCES_DIST = pixman-mmx.c pixman-mmx.h
 @USE_MMX_TRUE@am_libpixman_mmx_la_OBJECTS =  \
 @USE_MMX_TRUE@	libpixman_mmx_la-pixman-mmx.lo
 libpixman_mmx_la_OBJECTS = $(am_libpixman_mmx_la_OBJECTS)
 @USE_MMX_TRUE@am_libpixman_mmx_la_rpath =
-@USE_SSE2_TRUE@libpixman_sse_la_DEPENDENCIES = $(am__DEPENDENCIES_3)
-am__libpixman_sse_la_SOURCES_DIST = pixman-sse.c pixman-sse.h
-@USE_SSE2_TRUE@am_libpixman_sse_la_OBJECTS =  \
-@USE_SSE2_TRUE@	libpixman_sse_la-pixman-sse.lo
-libpixman_sse_la_OBJECTS = $(am_libpixman_sse_la_OBJECTS)
-@USE_SSE2_TRUE@am_libpixman_sse_la_rpath =
+@USE_SSE2_TRUE@libpixman_sse2_la_DEPENDENCIES = $(am__DEPENDENCIES_4)
+am__libpixman_sse2_la_SOURCES_DIST = pixman-sse2.c pixman-sse2.h
+@USE_SSE2_TRUE@am_libpixman_sse2_la_OBJECTS =  \
+@USE_SSE2_TRUE@	libpixman_sse2_la-pixman-sse2.lo
+libpixman_sse2_la_OBJECTS = $(am_libpixman_sse2_la_OBJECTS)
+@USE_SSE2_TRUE@am_libpixman_sse2_la_rpath =
+@USE_VMX_TRUE@libpixman_vmx_la_DEPENDENCIES = $(am__DEPENDENCIES_4)
+am__libpixman_vmx_la_SOURCES_DIST = pixman-vmx.c pixman-vmx.h \
+	pixman-combine32.h
+@USE_VMX_TRUE@am_libpixman_vmx_la_OBJECTS =  \
+@USE_VMX_TRUE@	libpixman_vmx_la-pixman-vmx.lo
+libpixman_vmx_la_OBJECTS = $(am_libpixman_vmx_la_OBJECTS)
+@USE_VMX_TRUE@am_libpixman_vmx_la_rpath =
 DEFAULT_INCLUDES = -I. -I$(srcdir) -I$(top_builddir)
 depcomp = $(SHELL) $(top_srcdir)/depcomp
 am__depfiles_maybe = depfiles
@@ -102,10 +124,11 @@ CCLD = $(CC)
 LINK = $(LIBTOOL) --tag=CC --mode=link $(CCLD) $(AM_CFLAGS) $(CFLAGS) \
 	$(AM_LDFLAGS) $(LDFLAGS) -o $@
 SOURCES = $(libpixman_1_la_SOURCES) $(libpixman_mmx_la_SOURCES) \
-	$(libpixman_sse_la_SOURCES)
+	$(libpixman_sse2_la_SOURCES) $(libpixman_vmx_la_SOURCES)
 DIST_SOURCES = $(libpixman_1_la_SOURCES) \
 	$(am__libpixman_mmx_la_SOURCES_DIST) \
-	$(am__libpixman_sse_la_SOURCES_DIST)
+	$(am__libpixman_sse2_la_SOURCES_DIST) \
+	$(am__libpixman_vmx_la_SOURCES_DIST)
 libpixmanincludeHEADERS_INSTALL = $(INSTALL_HEADER)
 HEADERS = $(libpixmaninclude_HEADERS)
 ETAGS = etags
@@ -134,6 +157,7 @@ DEFS = @DEFS@
 DEPDIR = @DEPDIR@
 DEP_CFLAGS = @DEP_CFLAGS@
 DEP_LIBS = @DEP_LIBS@
+DSYMUTIL = @DSYMUTIL@
 ECHO = @ECHO@
 ECHO_C = @ECHO_C@
 ECHO_N = @ECHO_N@
@@ -159,6 +183,7 @@ LTLIBOBJS = @LTLIBOBJS@
 LT_VERSION_INFO = @LT_VERSION_INFO@
 MAKEINFO = @MAKEINFO@
 MMX_CFLAGS = @MMX_CFLAGS@
+NMEDIT = @NMEDIT@
 OBJEXT = @OBJEXT@
 PACKAGE = @PACKAGE@
 PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
@@ -167,26 +192,31 @@ PACKAGE_STRING = @PACKAGE_STRING@
 PACKAGE_TARNAME = @PACKAGE_TARNAME@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
+PERL = @PERL@
 PIXMAN_VERSION_MAJOR = @PIXMAN_VERSION_MAJOR@
 PIXMAN_VERSION_MICRO = @PIXMAN_VERSION_MICRO@
 PIXMAN_VERSION_MINOR = @PIXMAN_VERSION_MINOR@
 PKG_CONFIG = @PKG_CONFIG@
 RANLIB = @RANLIB@
+SED = @SED@
 SET_MAKE = @SET_MAKE@
 SHELL = @SHELL@
-SSE_CFLAGS = @SSE_CFLAGS@
+SSE2_CFLAGS = @SSE2_CFLAGS@
 STRIP = @STRIP@
 USE_MMX_FALSE = @USE_MMX_FALSE@
 USE_MMX_TRUE = @USE_MMX_TRUE@
 USE_SSE2_FALSE = @USE_SSE2_FALSE@
 USE_SSE2_TRUE = @USE_SSE2_TRUE@
-USE_SSE_FALSE = @USE_SSE_FALSE@
-USE_SSE_TRUE = @USE_SSE_TRUE@
+USE_VMX_FALSE = @USE_VMX_FALSE@
+USE_VMX_TRUE = @USE_VMX_TRUE@
 VERSION = @VERSION@
+VMX_CFLAGS = @VMX_CFLAGS@
 ac_ct_AR = @ac_ct_AR@
 ac_ct_CC = @ac_ct_CC@
 ac_ct_CXX = @ac_ct_CXX@
+ac_ct_DSYMUTIL = @ac_ct_DSYMUTIL@
 ac_ct_F77 = @ac_ct_F77@
+ac_ct_NMEDIT = @ac_ct_NMEDIT@
 ac_ct_RANLIB = @ac_ct_RANLIB@
 ac_ct_STRIP = @ac_ct_STRIP@
 ac_pt_PKG_CONFIG = @ac_pt_PKG_CONFIG@
@@ -228,16 +258,22 @@ sharedstatedir = @sharedstatedir@
 sysconfdir = @sysconfdir@
 target_alias = @target_alias@
 lib_LTLIBRARIES = libpixman-1.la
-libpixman_1_la_LDFLAGS = -version-info $(LT_VERSION_INFO)
-libpixman_1_la_LIBADD = @DEP_LIBS@ -lm $(am__append_2) $(am__append_4)
+libpixman_1_la_LDFLAGS = -version-info $(LT_VERSION_INFO) -no-undefined
+libpixman_1_la_LIBADD = @DEP_LIBS@ -lm $(am__append_2) $(am__append_4) \
+	$(am__append_6)
+libpixman_1_la_CFLAGS = -DPIXMAN_DISABLE_DEPRECATED
 libpixman_1_la_SOURCES = \
 	pixman.h			\
 	pixman-access.c			\
 	pixman-access-accessors.c	\
-	pixman-region.c			\
+	pixman-region16.c		\
+	pixman-region32.c		\
 	pixman-private.h		\
 	pixman-image.c			\
-	pixman-combine.c		\
+	pixman-combine32.c		\
+	pixman-combine32.h		\
+	pixman-combine64.c		\
+	pixman-combine64.h		\
 	pixman-compose.c		\
 	pixman-compose-accessors.c	\
 	pixman-pict.c			\
@@ -254,20 +290,28 @@ libpixman_1_la_SOURCES = \
 
 libpixmanincludedir = $(includedir)/pixman-1/
 libpixmaninclude_HEADERS = pixman.h pixman-version.h
-noinst_LTLIBRARIES = $(am__append_1) $(am__append_3)
-EXTRA_DIST = Makefile.win32
+noinst_LTLIBRARIES = $(am__append_1) $(am__append_3) $(am__append_5)
+EXTRA_DIST = Makefile.win32 combine.inc combine.pl pixman-region.c combine.h.inc
+CLEANFILES = pixman-combine32.c pixman-combine64.c pixman-combine32.h pixman-combine64.h
 @USE_MMX_TRUE@libpixman_mmx_la_SOURCES = \
 @USE_MMX_TRUE@	pixman-mmx.c \
 @USE_MMX_TRUE@	pixman-mmx.h
 
 @USE_MMX_TRUE@libpixman_mmx_la_CFLAGS = $(DEP_CFLAGS) $(MMX_CFLAGS)
 @USE_MMX_TRUE@libpixman_mmx_la_LIBADD = $(DEP_LIBS)
-@USE_SSE2_TRUE@libpixman_sse_la_SOURCES = \
-@USE_SSE2_TRUE@	pixman-sse.c \
-@USE_SSE2_TRUE@	pixman-sse.h
-
-@USE_SSE2_TRUE@libpixman_sse_la_CFLAGS = $(DEP_CFLAGS) $(SSE_CFLAGS)
-@USE_SSE2_TRUE@libpixman_sse_la_LIBADD = $(DEP_LIBS)
+@USE_VMX_TRUE@libpixman_vmx_la_SOURCES = \
+@USE_VMX_TRUE@	pixman-vmx.c \
+@USE_VMX_TRUE@	pixman-vmx.h \
+@USE_VMX_TRUE@	pixman-combine32.h
+
+@USE_VMX_TRUE@libpixman_vmx_la_CFLAGS = $(DEP_CFLAGS) $(VMX_CFLAGS)
+@USE_VMX_TRUE@libpixman_vmx_la_LIBADD = $(DEP_LIBS)
+@USE_SSE2_TRUE@libpixman_sse2_la_SOURCES = \
+@USE_SSE2_TRUE@	pixman-sse2.c \
+@USE_SSE2_TRUE@	pixman-sse2.h
+
+@USE_SSE2_TRUE@libpixman_sse2_la_CFLAGS = $(DEP_CFLAGS) $(SSE2_CFLAGS)
+@USE_SSE2_TRUE@libpixman_sse2_la_LIBADD = $(DEP_LIBS)
 all: all-am
 
 .SUFFIXES:
@@ -343,8 +387,10 @@ libpixman-1.la: $(libpixman_1_la_OBJECTS) $(libpixman_1_la_DEPENDENCIES)
 	$(LINK) -rpath $(libdir) $(libpixman_1_la_LDFLAGS) $(libpixman_1_la_OBJECTS) $(libpixman_1_la_LIBADD) $(LIBS)
 libpixman-mmx.la: $(libpixman_mmx_la_OBJECTS) $(libpixman_mmx_la_DEPENDENCIES) 
 	$(LINK) $(am_libpixman_mmx_la_rpath) $(libpixman_mmx_la_LDFLAGS) $(libpixman_mmx_la_OBJECTS) $(libpixman_mmx_la_LIBADD) $(LIBS)
-libpixman-sse.la: $(libpixman_sse_la_OBJECTS) $(libpixman_sse_la_DEPENDENCIES) 
-	$(LINK) $(am_libpixman_sse_la_rpath) $(libpixman_sse_la_LDFLAGS) $(libpixman_sse_la_OBJECTS) $(libpixman_sse_la_LIBADD) $(LIBS)
+libpixman-sse2.la: $(libpixman_sse2_la_OBJECTS) $(libpixman_sse2_la_DEPENDENCIES) 
+	$(LINK) $(am_libpixman_sse2_la_rpath) $(libpixman_sse2_la_LDFLAGS) $(libpixman_sse2_la_OBJECTS) $(libpixman_sse2_la_LIBADD) $(LIBS)
+libpixman-vmx.la: $(libpixman_vmx_la_OBJECTS) $(libpixman_vmx_la_DEPENDENCIES) 
+	$(LINK) $(am_libpixman_vmx_la_rpath) $(libpixman_vmx_la_LDFLAGS) $(libpixman_vmx_la_OBJECTS) $(libpixman_vmx_la_LIBADD) $(LIBS)
 
 mostlyclean-compile:
 	-rm -f *.$(OBJEXT)
@@ -352,25 +398,28 @@ mostlyclean-compile:
 distclean-compile:
 	-rm -f *.tab.c
 
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-access-accessors.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-access.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-combine32.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-combine64.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-compose-accessors.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-compose.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-compute-region.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-edge-accessors.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-edge.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-image.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-pict.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-region16.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-region32.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-source.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-timer.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-transformed-accessors.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-transformed.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-trap.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_1_la-pixman-utils.Plo@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_mmx_la-pixman-mmx.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_sse_la-pixman-sse.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-access-accessors.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-access.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-combine.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-compose-accessors.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-compose.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-compute-region.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-edge-accessors.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-edge.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-image.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-pict.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-region.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-source.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-timer.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-transformed-accessors.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-transformed.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-trap.Plo@am__quote@
-@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/pixman-utils.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_sse2_la-pixman-sse2.Plo@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/libpixman_vmx_la-pixman-vmx.Plo@am__quote@
 
 .c.o:
 @am__fastdepCC_TRUE@	if $(COMPILE) -MT $@ -MD -MP -MF "$(DEPDIR)/$*.Tpo" -c -o $@ $<; \
@@ -393,6 +442,139 @@ distclean-compile:
 @AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
 @am__fastdepCC_FALSE@	$(LTCOMPILE) -c -o $@ $<
 
+libpixman_1_la-pixman-access.lo: pixman-access.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-access.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-access.Tpo" -c -o libpixman_1_la-pixman-access.lo `test -f 'pixman-access.c' || echo '$(srcdir)/'`pixman-access.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-access.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-access.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-access.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-access.c' object='libpixman_1_la-pixman-access.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-access.lo `test -f 'pixman-access.c' || echo '$(srcdir)/'`pixman-access.c
+
+libpixman_1_la-pixman-access-accessors.lo: pixman-access-accessors.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-access-accessors.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-access-accessors.Tpo" -c -o libpixman_1_la-pixman-access-accessors.lo `test -f 'pixman-access-accessors.c' || echo '$(srcdir)/'`pixman-access-accessors.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-access-accessors.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-access-accessors.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-access-accessors.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-access-accessors.c' object='libpixman_1_la-pixman-access-accessors.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-access-accessors.lo `test -f 'pixman-access-accessors.c' || echo '$(srcdir)/'`pixman-access-accessors.c
+
+libpixman_1_la-pixman-region16.lo: pixman-region16.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-region16.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-region16.Tpo" -c -o libpixman_1_la-pixman-region16.lo `test -f 'pixman-region16.c' || echo '$(srcdir)/'`pixman-region16.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-region16.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-region16.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-region16.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-region16.c' object='libpixman_1_la-pixman-region16.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-region16.lo `test -f 'pixman-region16.c' || echo '$(srcdir)/'`pixman-region16.c
+
+libpixman_1_la-pixman-region32.lo: pixman-region32.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-region32.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-region32.Tpo" -c -o libpixman_1_la-pixman-region32.lo `test -f 'pixman-region32.c' || echo '$(srcdir)/'`pixman-region32.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-region32.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-region32.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-region32.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-region32.c' object='libpixman_1_la-pixman-region32.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-region32.lo `test -f 'pixman-region32.c' || echo '$(srcdir)/'`pixman-region32.c
+
+libpixman_1_la-pixman-image.lo: pixman-image.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-image.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-image.Tpo" -c -o libpixman_1_la-pixman-image.lo `test -f 'pixman-image.c' || echo '$(srcdir)/'`pixman-image.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-image.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-image.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-image.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-image.c' object='libpixman_1_la-pixman-image.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-image.lo `test -f 'pixman-image.c' || echo '$(srcdir)/'`pixman-image.c
+
+libpixman_1_la-pixman-combine32.lo: pixman-combine32.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-combine32.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-combine32.Tpo" -c -o libpixman_1_la-pixman-combine32.lo `test -f 'pixman-combine32.c' || echo '$(srcdir)/'`pixman-combine32.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-combine32.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-combine32.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-combine32.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-combine32.c' object='libpixman_1_la-pixman-combine32.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-combine32.lo `test -f 'pixman-combine32.c' || echo '$(srcdir)/'`pixman-combine32.c
+
+libpixman_1_la-pixman-combine64.lo: pixman-combine64.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-combine64.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-combine64.Tpo" -c -o libpixman_1_la-pixman-combine64.lo `test -f 'pixman-combine64.c' || echo '$(srcdir)/'`pixman-combine64.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-combine64.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-combine64.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-combine64.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-combine64.c' object='libpixman_1_la-pixman-combine64.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-combine64.lo `test -f 'pixman-combine64.c' || echo '$(srcdir)/'`pixman-combine64.c
+
+libpixman_1_la-pixman-compose.lo: pixman-compose.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-compose.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-compose.Tpo" -c -o libpixman_1_la-pixman-compose.lo `test -f 'pixman-compose.c' || echo '$(srcdir)/'`pixman-compose.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-compose.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-compose.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-compose.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-compose.c' object='libpixman_1_la-pixman-compose.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-compose.lo `test -f 'pixman-compose.c' || echo '$(srcdir)/'`pixman-compose.c
+
+libpixman_1_la-pixman-compose-accessors.lo: pixman-compose-accessors.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-compose-accessors.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-compose-accessors.Tpo" -c -o libpixman_1_la-pixman-compose-accessors.lo `test -f 'pixman-compose-accessors.c' || echo '$(srcdir)/'`pixman-compose-accessors.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-compose-accessors.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-compose-accessors.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-compose-accessors.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-compose-accessors.c' object='libpixman_1_la-pixman-compose-accessors.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-compose-accessors.lo `test -f 'pixman-compose-accessors.c' || echo '$(srcdir)/'`pixman-compose-accessors.c
+
+libpixman_1_la-pixman-pict.lo: pixman-pict.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-pict.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-pict.Tpo" -c -o libpixman_1_la-pixman-pict.lo `test -f 'pixman-pict.c' || echo '$(srcdir)/'`pixman-pict.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-pict.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-pict.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-pict.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-pict.c' object='libpixman_1_la-pixman-pict.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-pict.lo `test -f 'pixman-pict.c' || echo '$(srcdir)/'`pixman-pict.c
+
+libpixman_1_la-pixman-source.lo: pixman-source.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-source.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-source.Tpo" -c -o libpixman_1_la-pixman-source.lo `test -f 'pixman-source.c' || echo '$(srcdir)/'`pixman-source.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-source.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-source.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-source.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-source.c' object='libpixman_1_la-pixman-source.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-source.lo `test -f 'pixman-source.c' || echo '$(srcdir)/'`pixman-source.c
+
+libpixman_1_la-pixman-transformed.lo: pixman-transformed.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-transformed.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-transformed.Tpo" -c -o libpixman_1_la-pixman-transformed.lo `test -f 'pixman-transformed.c' || echo '$(srcdir)/'`pixman-transformed.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-transformed.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-transformed.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-transformed.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-transformed.c' object='libpixman_1_la-pixman-transformed.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-transformed.lo `test -f 'pixman-transformed.c' || echo '$(srcdir)/'`pixman-transformed.c
+
+libpixman_1_la-pixman-transformed-accessors.lo: pixman-transformed-accessors.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-transformed-accessors.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-transformed-accessors.Tpo" -c -o libpixman_1_la-pixman-transformed-accessors.lo `test -f 'pixman-transformed-accessors.c' || echo '$(srcdir)/'`pixman-transformed-accessors.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-transformed-accessors.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-transformed-accessors.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-transformed-accessors.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-transformed-accessors.c' object='libpixman_1_la-pixman-transformed-accessors.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-transformed-accessors.lo `test -f 'pixman-transformed-accessors.c' || echo '$(srcdir)/'`pixman-transformed-accessors.c
+
+libpixman_1_la-pixman-utils.lo: pixman-utils.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-utils.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-utils.Tpo" -c -o libpixman_1_la-pixman-utils.lo `test -f 'pixman-utils.c' || echo '$(srcdir)/'`pixman-utils.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-utils.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-utils.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-utils.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-utils.c' object='libpixman_1_la-pixman-utils.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-utils.lo `test -f 'pixman-utils.c' || echo '$(srcdir)/'`pixman-utils.c
+
+libpixman_1_la-pixman-edge.lo: pixman-edge.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-edge.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-edge.Tpo" -c -o libpixman_1_la-pixman-edge.lo `test -f 'pixman-edge.c' || echo '$(srcdir)/'`pixman-edge.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-edge.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-edge.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-edge.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-edge.c' object='libpixman_1_la-pixman-edge.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-edge.lo `test -f 'pixman-edge.c' || echo '$(srcdir)/'`pixman-edge.c
+
+libpixman_1_la-pixman-edge-accessors.lo: pixman-edge-accessors.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-edge-accessors.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-edge-accessors.Tpo" -c -o libpixman_1_la-pixman-edge-accessors.lo `test -f 'pixman-edge-accessors.c' || echo '$(srcdir)/'`pixman-edge-accessors.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-edge-accessors.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-edge-accessors.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-edge-accessors.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-edge-accessors.c' object='libpixman_1_la-pixman-edge-accessors.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-edge-accessors.lo `test -f 'pixman-edge-accessors.c' || echo '$(srcdir)/'`pixman-edge-accessors.c
+
+libpixman_1_la-pixman-trap.lo: pixman-trap.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-trap.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-trap.Tpo" -c -o libpixman_1_la-pixman-trap.lo `test -f 'pixman-trap.c' || echo '$(srcdir)/'`pixman-trap.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-trap.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-trap.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-trap.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-trap.c' object='libpixman_1_la-pixman-trap.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-trap.lo `test -f 'pixman-trap.c' || echo '$(srcdir)/'`pixman-trap.c
+
+libpixman_1_la-pixman-compute-region.lo: pixman-compute-region.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-compute-region.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-compute-region.Tpo" -c -o libpixman_1_la-pixman-compute-region.lo `test -f 'pixman-compute-region.c' || echo '$(srcdir)/'`pixman-compute-region.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-compute-region.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-compute-region.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-compute-region.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-compute-region.c' object='libpixman_1_la-pixman-compute-region.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-compute-region.lo `test -f 'pixman-compute-region.c' || echo '$(srcdir)/'`pixman-compute-region.c
+
+libpixman_1_la-pixman-timer.lo: pixman-timer.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -MT libpixman_1_la-pixman-timer.lo -MD -MP -MF "$(DEPDIR)/libpixman_1_la-pixman-timer.Tpo" -c -o libpixman_1_la-pixman-timer.lo `test -f 'pixman-timer.c' || echo '$(srcdir)/'`pixman-timer.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_1_la-pixman-timer.Tpo" "$(DEPDIR)/libpixman_1_la-pixman-timer.Plo"; else rm -f "$(DEPDIR)/libpixman_1_la-pixman-timer.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-timer.c' object='libpixman_1_la-pixman-timer.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_1_la_CFLAGS) $(CFLAGS) -c -o libpixman_1_la-pixman-timer.lo `test -f 'pixman-timer.c' || echo '$(srcdir)/'`pixman-timer.c
+
 libpixman_mmx_la-pixman-mmx.lo: pixman-mmx.c
 @am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_mmx_la_CFLAGS) $(CFLAGS) -MT libpixman_mmx_la-pixman-mmx.lo -MD -MP -MF "$(DEPDIR)/libpixman_mmx_la-pixman-mmx.Tpo" -c -o libpixman_mmx_la-pixman-mmx.lo `test -f 'pixman-mmx.c' || echo '$(srcdir)/'`pixman-mmx.c; \
 @am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_mmx_la-pixman-mmx.Tpo" "$(DEPDIR)/libpixman_mmx_la-pixman-mmx.Plo"; else rm -f "$(DEPDIR)/libpixman_mmx_la-pixman-mmx.Tpo"; exit 1; fi
@@ -400,12 +582,19 @@ libpixman_mmx_la-pixman-mmx.lo: pixman-mmx.c
 @AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
 @am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_mmx_la_CFLAGS) $(CFLAGS) -c -o libpixman_mmx_la-pixman-mmx.lo `test -f 'pixman-mmx.c' || echo '$(srcdir)/'`pixman-mmx.c
 
-libpixman_sse_la-pixman-sse.lo: pixman-sse.c
-@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_sse_la_CFLAGS) $(CFLAGS) -MT libpixman_sse_la-pixman-sse.lo -MD -MP -MF "$(DEPDIR)/libpixman_sse_la-pixman-sse.Tpo" -c -o libpixman_sse_la-pixman-sse.lo `test -f 'pixman-sse.c' || echo '$(srcdir)/'`pixman-sse.c; \
-@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_sse_la-pixman-sse.Tpo" "$(DEPDIR)/libpixman_sse_la-pixman-sse.Plo"; else rm -f "$(DEPDIR)/libpixman_sse_la-pixman-sse.Tpo"; exit 1; fi
-@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-sse.c' object='libpixman_sse_la-pixman-sse.lo' libtool=yes @AMDEPBACKSLASH@
+libpixman_sse2_la-pixman-sse2.lo: pixman-sse2.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_sse2_la_CFLAGS) $(CFLAGS) -MT libpixman_sse2_la-pixman-sse2.lo -MD -MP -MF "$(DEPDIR)/libpixman_sse2_la-pixman-sse2.Tpo" -c -o libpixman_sse2_la-pixman-sse2.lo `test -f 'pixman-sse2.c' || echo '$(srcdir)/'`pixman-sse2.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_sse2_la-pixman-sse2.Tpo" "$(DEPDIR)/libpixman_sse2_la-pixman-sse2.Plo"; else rm -f "$(DEPDIR)/libpixman_sse2_la-pixman-sse2.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-sse2.c' object='libpixman_sse2_la-pixman-sse2.lo' libtool=yes @AMDEPBACKSLASH@
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_sse2_la_CFLAGS) $(CFLAGS) -c -o libpixman_sse2_la-pixman-sse2.lo `test -f 'pixman-sse2.c' || echo '$(srcdir)/'`pixman-sse2.c
+
+libpixman_vmx_la-pixman-vmx.lo: pixman-vmx.c
+@am__fastdepCC_TRUE@	if $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_vmx_la_CFLAGS) $(CFLAGS) -MT libpixman_vmx_la-pixman-vmx.lo -MD -MP -MF "$(DEPDIR)/libpixman_vmx_la-pixman-vmx.Tpo" -c -o libpixman_vmx_la-pixman-vmx.lo `test -f 'pixman-vmx.c' || echo '$(srcdir)/'`pixman-vmx.c; \
+@am__fastdepCC_TRUE@	then mv -f "$(DEPDIR)/libpixman_vmx_la-pixman-vmx.Tpo" "$(DEPDIR)/libpixman_vmx_la-pixman-vmx.Plo"; else rm -f "$(DEPDIR)/libpixman_vmx_la-pixman-vmx.Tpo"; exit 1; fi
+@AMDEP_TRUE@@am__fastdepCC_FALSE@	source='pixman-vmx.c' object='libpixman_vmx_la-pixman-vmx.lo' libtool=yes @AMDEPBACKSLASH@
 @AMDEP_TRUE@@am__fastdepCC_FALSE@	DEPDIR=$(DEPDIR) $(CCDEPMODE) $(depcomp) @AMDEPBACKSLASH@
-@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_sse_la_CFLAGS) $(CFLAGS) -c -o libpixman_sse_la-pixman-sse.lo `test -f 'pixman-sse.c' || echo '$(srcdir)/'`pixman-sse.c
+@am__fastdepCC_FALSE@	$(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS) $(libpixman_vmx_la_CFLAGS) $(CFLAGS) -c -o libpixman_vmx_la-pixman-vmx.lo `test -f 'pixman-vmx.c' || echo '$(srcdir)/'`pixman-vmx.c
 
 mostlyclean-libtool:
 	-rm -f *.lo
@@ -533,6 +722,7 @@ install-strip:
 mostlyclean-generic:
 
 clean-generic:
+	-test -z "$(CLEANFILES)" || rm -f $(CLEANFILES)
 
 distclean-generic:
 	-test -z "$(CONFIG_CLEAN_FILES)" || rm -f $(CONFIG_CLEAN_FILES)
@@ -606,6 +796,16 @@ uninstall-am: uninstall-info-am uninstall-libLTLIBRARIES \
 	tags uninstall uninstall-am uninstall-info-am \
 	uninstall-libLTLIBRARIES uninstall-libpixmanincludeHEADERS
 
+
+pixman-combine32.c : combine.inc pixman-combine32.h combine.pl
+	$(PERL) $(srcdir)/combine.pl 8 < $(srcdir)/combine.inc > $@ || ($(RM) $@; exit 1)
+pixman-combine32.h : combine.h.inc combine.pl
+	$(PERL) $(srcdir)/combine.pl 8 < $(srcdir)/combine.h.inc > $@ || ($(RM) $@; exit 1)
+
+pixman-combine64.c : combine.inc pixman-combine64.h combine.pl
+	$(PERL) $(srcdir)/combine.pl 16 < $(srcdir)/combine.inc > $@ || ($(RM) $@; exit 1)
+pixman-combine64.h : combine.h.inc combine.pl
+	$(PERL) $(srcdir)/combine.pl 16 < $(srcdir)/combine.h.inc > $@ || ($(RM) $@; exit 1)
 # 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/pixman/pixman/Makefile.win32 b/lib/pixman/pixman/Makefile.win32
index bee828902..ad30f8c46 100644
--- a/lib/pixman/pixman/Makefile.win32
+++ b/lib/pixman/pixman/Makefile.win32
@@ -3,32 +3,42 @@ LIBRARY     = pixman-1
 CC   = cl
 LINK = link
 
-ifeq ($(CFG),)
-CFG=release
+CFG_VAR = $(CFG)
+ifeq ($(CFG_VAR),)
+CFG_VAR=release
 endif
 
-ifeq ($(MMX),)
-MMX=1
+MMX_VAR = $(MMX)
+ifeq ($(MMX_VAR),)
+MMX_VAR=on
+endif
+
+SSE2_VAR = $(SSE2)
+ifeq ($(SSE2_VAR),)
+SSE2_VAR=on
 endif
 
 CFLAGS     = -MD -nologo -D_CRT_SECURE_NO_DEPRECATE -D_CRT_NONSTDC_NO_DEPRECATE -I../pixman/src -I. -DPACKAGE=$(LIBRARY) -DPACKAGE_VERSION="" -DPACKAGE_BUGREPORT=""
 MMX_CFLAGS = -DUSE_MMX -w14710 -w14714
+SSE2_CFLAGS = -DUSE_SSE2
 
 # optimization flags
-ifeq ($(CFG),debug)
+ifeq ($(CFG_VAR),debug)
 CFLAGS += -Od -Zi
 else
 CFLAGS += -O2
 endif
 
 SOURCES = \
-	pixman-region.c				\
 	pixman-image.c					\
 	pixman-access.c					\
 	pixman-access-accessors.c			\
-	pixman-combine.c				\
+	pixman-region16.c		\
+	pixman-region32.c		\
 	pixman-compose.c				\
 	pixman-compose-accessors.c	\
+	pixman-combine32.c				\
+	pixman-combine64.c				\
 	pixman-pict.c					\
 	pixman-source.c					\
 	pixman-transformed.c				\
@@ -42,24 +52,31 @@ SOURCES = \
 	$(NULL)
 
 # MMX compilation flags
-ifeq ($(MMX),1)
+ifeq ($(MMX_VAR),on)
 CFLAGS += $(MMX_CFLAGS)
 SOURCES += pixman-mmx.c
 endif
 
-OBJECTS     = $(patsubst %.c, $(CFG)/%.obj, $(SOURCES))
+# SSE2 compilation flags
+ifeq ($(SSE2_VAR),on)
+CFLAGS += $(SSE2_CFLAGS)
+SOURCES += pixman-sse2.c
+endif
+
+OBJECTS     = $(patsubst %.c, $(CFG_VAR)/%.obj, $(SOURCES))
 
 # targets
-all: inform informMMX $(CFG)/$(LIBRARY).lib
+all: inform informMMX informSSE2 $(CFG_VAR)/$(LIBRARY).lib
 	@exit 0
 clean: inform clean_r 
 	@exit 0
-pixman: inform informMMX $(CFG)/$(LIBRARY).lib 
+pixman: inform informMMX informSSE2 $(CFG_VAR)/$(LIBRARY).lib 
 	@exit 0
 
 inform:
 ifneq ($(CFG),release)
 ifneq ($(CFG),debug)
+ifneq ($(CFG),)
 	@echo "Invalid specified configuration option : "$(CFG)"."
 	@echo
 	@echo -n "Possible choices for configuration are "
@@ -67,26 +84,56 @@ ifneq ($(CFG),debug)
 	@echo ""
 	@exit 1
 endif
+	@echo "Using default RELEASE configuration... (use CFG=release or CFG=debug)"
+endif
 endif
 
 informMMX:
-ifneq ($(MMX),0)
-ifneq ($(MMX),1)
-	@echo "Invalid specified MMX option : "$(MMX)"."
+ifneq ($(MMX),off)
+ifneq ($(MMX),on)
+ifneq ($(MMX),)
+	@echo "Invalid specified MMX option : "$(MMX_VAR)"."
 	@echo
-	@echo -n "Possible choices for MMX are 0 or 1"
+	@echo -n "Possible choices for MMX are 'on' or 'off'"
 	@echo ""
 	@exit 1
 endif
+	@echo "Setting MMX flag to default value 'on'... (use MMX=on or MMX=off)"
+endif
+endif
+
+informSSE2:
+ifneq ($(SSE2),off)
+ifneq ($(SSE2),on)
+ifneq ($(SSE2),)
+	@echo "Invalid specified SSE option : "$(SSE2)"."
+	@echo
+	@echo -n "Possible choices for SSE2 are 'on' or 'off'"
+	@echo ""
+	@exit 1
+endif
+	@echo "Setting SSE2 flag to default value 'on'... (use SSE2=on or SSE2=off)"
+endif
 endif
 
 # pixman compilation and linking
-$(CFG)/%.obj: %.c
-	@mkdir -p $(CFG)
+$(CFG_VAR)/%.obj: %.c
+	@mkdir -p $(CFG_VAR)
 	@$(CC) -c $(CFLAGS) -Fo"$@" $<
 
-$(CFG)/$(LIBRARY).lib: $(OBJECTS)
+$(CFG_VAR)/$(LIBRARY).lib: $(OBJECTS)
 	lib -NOLOGO -OUT:$@ $(OBJECTS) || exit 0
 
+pixman-combine32.c: combine.inc pixman-combine32.h combine.pl
+	perl ./combine.pl 8 < $< > $@ || ($(RM) $@; exit 1)
+pixman-combine32.h: combine.h.inc combine.pl
+	perl ./combine.pl 8 < $< > $@ || ($(RM) $@; exit 1)
+
+pixman-combine64.c: combine.inc pixman-combine64.h combine.pl
+	perl ./combine.pl 16 < $< > $@ || ($(RM) $@; exit 1)
+pixman-combine64.h: combine.h.inc combine.pl
+	perl ./combine.pl 16 < $< > $@ || ($(RM) $@; exit 1)
+
 clean_r:
-	@rm -f $(CFG)/*.obj $(CFG)/*.lib $(CFG)/*.pdb $(CFG)/*.ilk || exit 0
+	@rm -f $(CFG_VAR)/*.obj $(CFG_VAR)/*.lib $(CFG_VAR)/*.pdb $(CFG)/*.ilk || exit 0
+	@rm -f $(CFG)/*.obj $(CFG)/*.lib $(CFG)/*.pdb $(CFG)/*.ilk pixman-combine32.c pixman-combine64.c || exit 0
diff --git a/lib/pixman/pixman/combine.h.inc b/lib/pixman/pixman/combine.h.inc
new file mode 100644
index 000000000..8c70cb78d
--- /dev/null
+++ b/lib/pixman/pixman/combine.h.inc
@@ -0,0 +1,227 @@
+
+#define COMPONENT_SIZE
+#define MASK
+#define ONE_HALF
+
+#define G_SHIFT
+#define B_SHIFT
+#define A_SHIFT
+#define G_MASK
+#define B_MASK
+#define A_MASK
+
+#define RB_MASK
+#define AG_MASK
+#define RB_ONE_HALF
+#define RB_MASK_PLUS_ONE
+
+#define Alpha(x) ((x) >> A_SHIFT)
+
+/*
+ * Helper macros.
+ */
+
+#define IntMult(a,b,t) ( (t) = (a) * (b) + ONE_HALF, ( ( ( (t)>>G_SHIFT ) + (t) )>>G_SHIFT ) )
+#define IntDiv(a,b)    (((comp2_t) (a) * MASK) / (b))
+
+#define GetComp(v,i)   ((comp2_t) (comp1_t) ((v) >> i))
+
+#define Add(x,y,i,t)   ((t) = GetComp(x,i) + GetComp(y,i),              \
+                        (comp4_t) ((comp1_t) ((t) | (0 - ((t) >> G_SHIFT)))) << (i))
+
+#define FbGen(x,y,i,ax,ay,t,u,v) ((t) = (IntMult(GetComp(y,i),ay,(u)) + \
+					 IntMult(GetComp(x,i),ax,(v))), \
+				  (comp4_t) ((comp1_t) ((t) |		\
+							 (0 - ((t) >> G_SHIFT)))) << (i))
+
+/*
+  The methods below use some tricks to be able to do two color
+  components at the same time.
+*/
+
+/*
+  x_c = (x_c * a) / 255
+*/
+#define FbByteMul(x, a) do {                                            \
+        comp4_t t = ((x & RB_MASK) * a) + RB_ONE_HALF;                  \
+        t = (t + ((t >> COMPONENT_SIZE) & RB_MASK)) >> COMPONENT_SIZE;  \
+        t &= RB_MASK;                                                   \
+                                                                        \
+        x = (((x >> COMPONENT_SIZE) & RB_MASK) * a) + RB_ONE_HALF;      \
+        x = (x + ((x >> COMPONENT_SIZE) & RB_MASK));                    \
+        x &= RB_MASK << COMPONENT_SIZE;                                 \
+        x += t;                                                         \
+    } while (0)
+
+/*
+  x_c = (x_c * a) / 255 + y
+*/
+#define FbByteMulAdd(x, a, y) do {                                      \
+        /* multiply and divide: trunc((i + 128)*257/65536) */           \
+        comp4_t t = ((x & RB_MASK) * a) + RB_ONE_HALF;                  \
+        t = (t + ((t >> COMPONENT_SIZE) & RB_MASK)) >> COMPONENT_SIZE;  \
+        t &= RB_MASK;                                                   \
+                                                                        \
+        /* add */                                                       \
+        t += y & RB_MASK;                                               \
+                                                                        \
+        /* saturate */                                                  \
+        t |= RB_MASK_PLUS_ONE - ((t >> COMPONENT_SIZE) & RB_MASK);      \
+        t &= RB_MASK;                                                   \
+                                                                        \
+        /* multiply and divide */                                       \
+        x = (((x >> COMPONENT_SIZE) & RB_MASK) * a) + RB_ONE_HALF;      \
+        x = (x + ((x >> COMPONENT_SIZE) & RB_MASK)) >> COMPONENT_SIZE;  \
+        x &= RB_MASK;                                                   \
+                                                                        \
+        /* add */                                                       \
+        x += (y >> COMPONENT_SIZE) & RB_MASK;                           \
+                                                                        \
+        /* saturate */                                                  \
+        x |= RB_MASK_PLUS_ONE - ((x >> COMPONENT_SIZE) & RB_MASK);      \
+        x &= RB_MASK;                                                   \
+                                                                        \
+        /* recombine */                                                 \
+        x <<= COMPONENT_SIZE;                                           \
+        x += t;                                                         \
+    } while (0)
+
+/*
+  x_c = (x_c * a + y_c * b) / 255
+*/
+#define FbByteAddMul(x, a, y, b) do {                                   \
+        comp4_t t;                                                      \
+        comp4_t r = (x >> A_SHIFT) * a + (y >> A_SHIFT) * b + ONE_HALF; \
+        r += (r >> G_SHIFT);                                            \
+        r >>= G_SHIFT;                                                  \
+                                                                        \
+        t = (x & G_MASK) * a + (y & G_MASK) * b;                        \
+        t += (t >> G_SHIFT) + (ONE_HALF << G_SHIFT);                    \
+        t >>= B_SHIFT;                                                  \
+                                                                        \
+        t |= r << B_SHIFT;                                              \
+        t |= RB_MASK_PLUS_ONE - ((t >> G_SHIFT) & RB_MASK);             \
+        t &= RB_MASK;                                                   \
+        t <<= G_SHIFT;                                                  \
+                                                                        \
+        r = ((x >> B_SHIFT) & MASK) * a +                               \
+            ((y >> B_SHIFT) & MASK) * b + ONE_HALF;                     \
+        r += (r >> G_SHIFT);                                            \
+        r >>= G_SHIFT;                                                  \
+                                                                        \
+        x = (x & MASK) * a + (y & MASK) * b + ONE_HALF;                 \
+        x += (x >> G_SHIFT);                                            \
+        x >>= G_SHIFT;                                                  \
+        x |= r << B_SHIFT;                                              \
+        x |= RB_MASK_PLUS_ONE - ((x >> G_SHIFT) & RB_MASK);             \
+        x &= RB_MASK;                                                   \
+        x |= t;                                                         \
+    } while (0)
+
+/*
+  x_c = (x_c * a + y_c *b) / 256
+*/
+#define FbByteAddMul_256(x, a, y, b) do {                               \
+        comp4_t t = (x & RB_MASK) * a + (y & RB_MASK) * b;              \
+        t >>= G_SHIFT;                                                  \
+        t &= RB_MASK;                                                   \
+                                                                        \
+        x = ((x >> G_SHIFT) & RB_MASK) * a +                            \
+            ((y >> G_SHIFT) & RB_MASK) * b;                             \
+        x &= AG_MASK;                                                   \
+        x += t;                                                         \
+    } while (0)
+
+/*
+  x_c = (x_c * a_c) / 255
+*/
+#define FbByteMulC(x, a) do {                                           \
+        comp4_t t;                                                      \
+        comp4_t r = (x & MASK) * (a & MASK);                            \
+        r |= (x & B_MASK) * ((a >> B_SHIFT) & MASK);                    \
+        r += RB_ONE_HALF;                                               \
+        r = (r + ((r >> G_SHIFT) & RB_MASK)) >> G_SHIFT;                \
+        r &= RB_MASK;                                                   \
+                                                                        \
+        x >>= G_SHIFT;                                                  \
+        t = (x & MASK) * ((a >> G_SHIFT) & MASK);                       \
+        t |= (x & B_MASK) * (a >> A_SHIFT);                             \
+        t += RB_ONE_HALF;                                               \
+        t = t + ((t >> G_SHIFT) & RB_MASK);                             \
+        x = r | (t & AG_MASK);                                          \
+    } while (0)
+
+/*
+  x_c = (x_c * a) / 255 + y
+*/
+#define FbByteMulAddC(x, a, y) do {                                     \
+        comp4_t t;                                                      \
+        comp4_t r = (x & MASK) * (a & MASK);                            \
+        r |= (x & B_MASK) * ((a >> B_SHIFT) & MASK);                    \
+        r += RB_ONE_HALF;                                               \
+        r = (r + ((r >> G_SHIFT) & RB_MASK)) >> G_SHIFT;                \
+        r &= RB_MASK;                                                   \
+        r += y & RB_MASK;                                               \
+        r |= RB_MASK_PLUS_ONE - ((r >> G_SHIFT) & RB_MASK);             \
+        r &= RB_MASK;                                                   \
+                                                                        \
+        x >>= G_SHIFT;                                                  \
+        t = (x & MASK) * ((a >> G_SHIFT) & MASK);                       \
+        t |= (x & B_MASK) * (a >> A_SHIFT);                             \
+        t += RB_ONE_HALF;                                               \
+        t = (t + ((t >> G_SHIFT) & RB_MASK)) >> G_SHIFT;                \
+        t &= RB_MASK;                                                   \
+        t += (y >> G_SHIFT) & RB_MASK;                                  \
+        t |= RB_MASK_PLUS_ONE - ((t >> G_SHIFT) & RB_MASK);             \
+        t &= RB_MASK;                                                   \
+        x = r | (t << G_SHIFT);                                         \
+    } while (0)
+
+/*
+  x_c = (x_c * a_c + y_c * b) / 255
+*/
+#define FbByteAddMulC(x, a, y, b) do {                                  \
+        comp4_t t;                                                      \
+        comp4_t r = (x >> A_SHIFT) * (a >> A_SHIFT) +                   \
+                     (y >> A_SHIFT) * b;                                \
+        r += (r >> G_SHIFT) + ONE_HALF;                                 \
+        r >>= G_SHIFT;                                                  \
+                                                                        \
+        t = (x & G_MASK) * ((a >> G_SHIFT) & MASK) + (y & G_MASK) * b;  \
+        t += (t >> G_SHIFT) + (ONE_HALF << G_SHIFT);                    \
+        t >>= B_SHIFT;                                                  \
+                                                                        \
+        t |= r << B_SHIFT;                                              \
+        t |= RB_MASK_PLUS_ONE - ((t >> G_SHIFT) & RB_MASK);             \
+        t &= RB_MASK;                                                   \
+        t <<= G_SHIFT;                                                  \
+                                                                        \
+        r = ((x >> B_SHIFT) & MASK) * ((a >> B_SHIFT) & MASK) +         \
+            ((y >> B_SHIFT) & MASK) * b + ONE_HALF;                     \
+        r += (r >> G_SHIFT);                                            \
+        r >>= G_SHIFT;                                                  \
+                                                                        \
+        x = (x & MASK) * (a & MASK) + (y & MASK) * b + ONE_HALF;        \
+        x += (x >> G_SHIFT);                                            \
+        x >>= G_SHIFT;                                                  \
+        x |= r << B_SHIFT;                                              \
+        x |= RB_MASK_PLUS_ONE - ((x >> G_SHIFT) & RB_MASK);             \
+        x &= RB_MASK;                                                   \
+        x |= t;                                                         \
+    } while (0)
+
+/*
+  x_c = min(x_c + y_c, 255)
+*/
+#define FbByteAdd(x, y) do {                                            \
+        comp4_t t;                                                      \
+        comp4_t r = (x & RB_MASK) + (y & RB_MASK);                      \
+        r |= RB_MASK_PLUS_ONE - ((r >> G_SHIFT) & RB_MASK);             \
+        r &= RB_MASK;                                                   \
+                                                                        \
+        t = ((x >> G_SHIFT) & RB_MASK) + ((y >> G_SHIFT) & RB_MASK);    \
+        t |= RB_MASK_PLUS_ONE - ((t >> G_SHIFT) & RB_MASK);             \
+        r |= (t & RB_MASK) << G_SHIFT;                                  \
+        x = r;                                                          \
+    } while (0)
+
diff --git a/lib/pixman/pixman/pixman-combine.c b/lib/pixman/pixman/combine.inc
index 0d9c066c7..9f88dee15 100644
--- a/lib/pixman/pixman/pixman-combine.c
+++ b/lib/pixman/pixman/combine.inc
@@ -5,6 +5,9 @@
 #include <string.h>
 
 #include "pixman-private.h"
+
+#include "pixman-combine.h"
+
 /*
  * There are two ways of handling alpha -- either as a single unified value or
  * a separate value for each component, hence each macro must have two
@@ -17,13 +20,13 @@
 /*
  * Combine src and mask
  */
-FASTCALL void
-pixman_fbCombineMaskU (uint32_t *src, const uint32_t *mask, int width)
+FASTCALL static void
+pixman_fbCombineMaskU (comp4_t *src, const comp4_t *mask, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t a = *(mask + i) >> 24;
-        uint32_t s = *(src + i);
+        comp4_t a = *(mask + i) >> A_SHIFT;
+        comp4_t s = *(src + i);
         FbByteMul(s, a);
         *(src + i) = s;
     }
@@ -34,26 +37,26 @@ pixman_fbCombineMaskU (uint32_t *src, const uint32_t *mask, int width)
  */
 
 FASTCALL static void
-fbCombineClear (uint32_t *dest, const uint32_t *src, int width)
+fbCombineClear (comp4_t *dest, const comp4_t *src, int width)
 {
-    memset(dest, 0, width*sizeof(uint32_t));
+    memset(dest, 0, width*sizeof(comp4_t));
 }
 
 FASTCALL static void
-fbCombineSrcU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineSrcU (comp4_t *dest, const comp4_t *src, int width)
 {
-    memcpy(dest, src, width*sizeof(uint32_t));
+    memcpy(dest, src, width*sizeof(comp4_t));
 }
 
 /* if the Src is opaque, call fbCombineSrcU */
 FASTCALL static void
-fbCombineOverU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineOverU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t s = *(src + i);
-        uint32_t d = *(dest + i);
-        uint32_t ia = Alpha(~s);
+        comp4_t s = *(src + i);
+        comp4_t d = *(dest + i);
+        comp4_t ia = Alpha(~s);
 
         FbByteMulAdd(d, ia, s);
 	*(dest + i) = d;
@@ -62,13 +65,13 @@ fbCombineOverU (uint32_t *dest, const uint32_t *src, int width)
 
 /* if the Dst is opaque, this is a noop */
 FASTCALL static void
-fbCombineOverReverseU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineOverReverseU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t s = *(src + i);
-        uint32_t d = *(dest + i);
-        uint32_t ia = Alpha(~*(dest + i));
+        comp4_t s = *(src + i);
+        comp4_t d = *(dest + i);
+        comp4_t ia = Alpha(~*(dest + i));
         FbByteMulAdd(s, ia, d);
 	*(dest + i) = s;
     }
@@ -76,12 +79,12 @@ fbCombineOverReverseU (uint32_t *dest, const uint32_t *src, int width)
 
 /* if the Dst is opaque, call fbCombineSrcU */
 FASTCALL static void
-fbCombineInU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineInU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t s = *(src + i);
-        uint32_t a = Alpha(*(dest + i));
+        comp4_t s = *(src + i);
+        comp4_t a = Alpha(*(dest + i));
         FbByteMul(s, a);
 	*(dest + i) = s;
     }
@@ -89,12 +92,12 @@ fbCombineInU (uint32_t *dest, const uint32_t *src, int width)
 
 /* if the Src is opaque, this is a noop */
 FASTCALL static void
-fbCombineInReverseU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineInReverseU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t d = *(dest + i);
-        uint32_t a = Alpha(*(src + i));
+        comp4_t d = *(dest + i);
+        comp4_t a = Alpha(*(src + i));
         FbByteMul(d, a);
 	*(dest + i) = d;
     }
@@ -102,12 +105,12 @@ fbCombineInReverseU (uint32_t *dest, const uint32_t *src, int width)
 
 /* if the Dst is opaque, call fbCombineClear */
 FASTCALL static void
-fbCombineOutU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineOutU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t s = *(src + i);
-        uint32_t a = Alpha(~*(dest + i));
+        comp4_t s = *(src + i);
+        comp4_t a = Alpha(~*(dest + i));
         FbByteMul(s, a);
 	*(dest + i) = s;
     }
@@ -115,12 +118,12 @@ fbCombineOutU (uint32_t *dest, const uint32_t *src, int width)
 
 /* if the Src is opaque, call fbCombineClear */
 FASTCALL static void
-fbCombineOutReverseU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineOutReverseU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t d = *(dest + i);
-        uint32_t a = Alpha(~*(src + i));
+        comp4_t d = *(dest + i);
+        comp4_t a = Alpha(~*(src + i));
         FbByteMul(d, a);
 	*(dest + i) = d;
     }
@@ -130,14 +133,14 @@ fbCombineOutReverseU (uint32_t *dest, const uint32_t *src, int width)
 /* if the Dst is opaque, call fbCombineOverU */
 /* if both the Src and Dst are opaque, call fbCombineSrcU */
 FASTCALL static void
-fbCombineAtopU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineAtopU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t s = *(src + i);
-        uint32_t d = *(dest + i);
-        uint32_t dest_a = Alpha(d);
-        uint32_t src_ia = Alpha(~s);
+        comp4_t s = *(src + i);
+        comp4_t d = *(dest + i);
+        comp4_t dest_a = Alpha(d);
+        comp4_t src_ia = Alpha(~s);
 
         FbByteAddMul(s, dest_a, d, src_ia);
 	*(dest + i) = s;
@@ -148,14 +151,14 @@ fbCombineAtopU (uint32_t *dest, const uint32_t *src, int width)
 /* if the Dst is opaque, call fbCombineInReverseU */
 /* if both the Src and Dst are opaque, call fbCombineDstU */
 FASTCALL static void
-fbCombineAtopReverseU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineAtopReverseU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t s = *(src + i);
-        uint32_t d = *(dest + i);
-        uint32_t src_a = Alpha(s);
-        uint32_t dest_ia = Alpha(~d);
+        comp4_t s = *(src + i);
+        comp4_t d = *(dest + i);
+        comp4_t src_a = Alpha(s);
+        comp4_t dest_ia = Alpha(~d);
 
         FbByteAddMul(s, dest_ia, d, src_a);
 	*(dest + i) = s;
@@ -166,14 +169,14 @@ fbCombineAtopReverseU (uint32_t *dest, const uint32_t *src, int width)
 /* if the Dst is opaque, call fbCombineOverReverseU */
 /* if both the Src and Dst are opaque, call fbCombineClear */
 FASTCALL static void
-fbCombineXorU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineXorU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t s = *(src + i);
-        uint32_t d = *(dest + i);
-        uint32_t src_ia = Alpha(~s);
-        uint32_t dest_ia = Alpha(~d);
+        comp4_t s = *(src + i);
+        comp4_t d = *(dest + i);
+        comp4_t src_ia = Alpha(~s);
+        comp4_t dest_ia = Alpha(~d);
 
         FbByteAddMul(s, dest_ia, d, src_ia);
 	*(dest + i) = s;
@@ -181,12 +184,12 @@ fbCombineXorU (uint32_t *dest, const uint32_t *src, int width)
 }
 
 FASTCALL static void
-fbCombineAddU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineAddU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t s = *(src + i);
-        uint32_t d = *(dest + i);
+        comp4_t s = *(src + i);
+        comp4_t d = *(dest + i);
         FbByteAdd(d, s);
 	*(dest + i) = d;
     }
@@ -196,19 +199,19 @@ fbCombineAddU (uint32_t *dest, const uint32_t *src, int width)
 /* if the Dst is opaque, call fbCombineAddU */
 /* if both the Src and Dst are opaque, call fbCombineAddU */
 FASTCALL static void
-fbCombineSaturateU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineSaturateU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t  s = *(src + i);
-        uint32_t d = *(dest + i);
-        uint16_t  sa, da;
+        comp4_t s = *(src + i);
+        comp4_t d = *(dest + i);
+        comp2_t sa, da;
 
-        sa = s >> 24;
-        da = ~d >> 24;
+        sa = s >> A_SHIFT;
+        da = ~d >> A_SHIFT;
         if (sa > da)
         {
-            sa = FbIntDiv(da, sa);
+            sa = IntDiv(da, sa);
             FbByteMul(s, sa);
         };
         FbByteAdd(d, s);
@@ -257,20 +260,20 @@ fbCombineSaturateU (uint32_t *dest, const uint32_t *src, int width)
 #define CombineXor	(CombineAOut|CombineBOut)
 
 /* portion covered by a but not b */
-FASTCALL static uint8_t
-fbCombineDisjointOutPart (uint8_t a, uint8_t b)
+FASTCALL static comp1_t
+fbCombineDisjointOutPart (comp1_t a, comp1_t b)
 {
     /* min (1, (1-b) / a) */
 
     b = ~b;		    /* 1 - b */
     if (b >= a)		    /* 1 - b >= a -> (1-b)/a >= 1 */
-	return 0xff;	    /* 1 */
-    return FbIntDiv(b,a);   /* (1-b) / a */
+	return MASK;	    /* 1 */
+    return IntDiv(b,a);     /* (1-b) / a */
 }
 
 /* portion covered by both a and b */
-FASTCALL static uint8_t
-fbCombineDisjointInPart (uint8_t a, uint8_t b)
+FASTCALL static comp1_t
+fbCombineDisjointInPart (comp1_t a, comp1_t b)
 {
     /* max (1-(1-b)/a,0) */
     /*  = - min ((1-b)/a - 1, 0) */
@@ -279,12 +282,12 @@ fbCombineDisjointInPart (uint8_t a, uint8_t b)
     b = ~b;		    /* 1 - b */
     if (b >= a)		    /* 1 - b >= a -> (1-b)/a >= 1 */
 	return 0;	    /* 1 - 1 */
-    return ~FbIntDiv(b,a);  /* 1 - (1-b) / a */
+    return ~IntDiv(b,a);    /* 1 - (1-b) / a */
 }
 
 /* portion covered by a but not b */
-FASTCALL static uint8_t
-fbCombineConjointOutPart (uint8_t a, uint8_t b)
+FASTCALL static comp1_t
+fbCombineConjointOutPart (comp1_t a, comp1_t b)
 {
     /* max (1-b/a,0) */
     /* = 1-min(b/a,1) */
@@ -293,31 +296,31 @@ fbCombineConjointOutPart (uint8_t a, uint8_t b)
 
     if (b >= a)		    /* b >= a -> b/a >= 1 */
 	return 0x00;	    /* 0 */
-    return ~FbIntDiv(b,a);   /* 1 - b/a */
+    return ~IntDiv(b,a);    /* 1 - b/a */
 }
 
 /* portion covered by both a and b */
-FASTCALL static uint8_t
-fbCombineConjointInPart (uint8_t a, uint8_t b)
+FASTCALL static comp1_t
+fbCombineConjointInPart (comp1_t a, comp1_t b)
 {
     /* min (1,b/a) */
 
     if (b >= a)		    /* b >= a -> b/a >= 1 */
-	return 0xff;	    /* 1 */
-    return FbIntDiv(b,a);   /* b/a */
+	return MASK;	    /* 1 */
+    return IntDiv(b,a);     /* b/a */
 }
 
 FASTCALL static void
-fbCombineDisjointGeneralU (uint32_t *dest, const uint32_t *src, int width, uint8_t combine)
+fbCombineDisjointGeneralU (comp4_t *dest, const comp4_t *src, int width, comp1_t combine)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t s = *(src + i);
-        uint32_t d = *(dest + i);
-        uint32_t m,n,o,p;
-        uint16_t Fa, Fb, t, u, v;
-        uint8_t sa = s >> 24;
-        uint8_t da = d >> 24;
+        comp4_t s = *(src + i);
+        comp4_t d = *(dest + i);
+        comp4_t m,n,o,p;
+        comp2_t Fa, Fb, t, u, v;
+        comp1_t sa = s >> A_SHIFT;
+        comp1_t da = d >> A_SHIFT;
 
         switch (combine & CombineA) {
         default:
@@ -330,7 +333,7 @@ fbCombineDisjointGeneralU (uint32_t *dest, const uint32_t *src, int width, uint8
             Fa = fbCombineDisjointInPart (sa, da);
             break;
         case CombineA:
-            Fa = 0xff;
+            Fa = MASK;
             break;
         }
 
@@ -345,32 +348,32 @@ fbCombineDisjointGeneralU (uint32_t *dest, const uint32_t *src, int width, uint8
             Fb = fbCombineDisjointInPart (da, sa);
             break;
         case CombineB:
-            Fb = 0xff;
+            Fb = MASK;
             break;
         }
         m = FbGen (s,d,0,Fa,Fb,t, u, v);
-        n = FbGen (s,d,8,Fa,Fb,t, u, v);
-        o = FbGen (s,d,16,Fa,Fb,t, u, v);
-        p = FbGen (s,d,24,Fa,Fb,t, u, v);
+        n = FbGen (s,d,G_SHIFT,Fa,Fb,t, u, v);
+        o = FbGen (s,d,B_SHIFT,Fa,Fb,t, u, v);
+        p = FbGen (s,d,A_SHIFT,Fa,Fb,t, u, v);
         s = m|n|o|p;
 	*(dest + i) = s;
     }
 }
 
 FASTCALL static void
-fbCombineDisjointOverU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineDisjointOverU (comp4_t *dest, const comp4_t *src, int width)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t  s = *(src + i);
-        uint16_t  a = s >> 24;
+        comp4_t s = *(src + i);
+        comp2_t a = s >> A_SHIFT;
 
         if (a != 0x00)
         {
-            if (a != 0xff)
+            if (a != MASK)
             {
-                uint32_t d = *(dest + i);
-                a = fbCombineDisjointOutPart (d >> 24, a);
+                comp4_t d = *(dest + i);
+                a = fbCombineDisjointOutPart (d >> A_SHIFT, a);
                 FbByteMulAdd(d, a, s);
                 s = d;
             }
@@ -380,58 +383,58 @@ fbCombineDisjointOverU (uint32_t *dest, const uint32_t *src, int width)
 }
 
 FASTCALL static void
-fbCombineDisjointInU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineDisjointInU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineDisjointGeneralU (dest, src, width, CombineAIn);
 }
 
 FASTCALL static void
-fbCombineDisjointInReverseU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineDisjointInReverseU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineDisjointGeneralU (dest, src, width, CombineBIn);
 }
 
 FASTCALL static void
-fbCombineDisjointOutU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineDisjointOutU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineDisjointGeneralU (dest, src, width, CombineAOut);
 }
 
 FASTCALL static void
-fbCombineDisjointOutReverseU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineDisjointOutReverseU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineDisjointGeneralU (dest, src, width, CombineBOut);
 }
 
 FASTCALL static void
-fbCombineDisjointAtopU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineDisjointAtopU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineDisjointGeneralU (dest, src, width, CombineAAtop);
 }
 
 FASTCALL static void
-fbCombineDisjointAtopReverseU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineDisjointAtopReverseU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineDisjointGeneralU (dest, src, width, CombineBAtop);
 }
 
 FASTCALL static void
-fbCombineDisjointXorU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineDisjointXorU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineDisjointGeneralU (dest, src, width, CombineXor);
 }
 
 FASTCALL static void
-fbCombineConjointGeneralU (uint32_t *dest, const uint32_t *src, int width, uint8_t combine)
+fbCombineConjointGeneralU (comp4_t *dest, const comp4_t *src, int width, comp1_t combine)
 {
     int i;
     for (i = 0; i < width; ++i) {
-        uint32_t  s = *(src + i);
-        uint32_t d = *(dest + i);
-        uint32_t  m,n,o,p;
-        uint16_t  Fa, Fb, t, u, v;
-        uint8_t sa = s >> 24;
-        uint8_t da = d >> 24;
+        comp4_t s = *(src + i);
+        comp4_t d = *(dest + i);
+        comp4_t m,n,o,p;
+        comp2_t Fa, Fb, t, u, v;
+        comp1_t sa = s >> A_SHIFT;
+        comp1_t da = d >> A_SHIFT;
 
         switch (combine & CombineA) {
         default:
@@ -444,7 +447,7 @@ fbCombineConjointGeneralU (uint32_t *dest, const uint32_t *src, int width, uint8
             Fa = fbCombineConjointInPart (sa, da);
             break;
         case CombineA:
-            Fa = 0xff;
+            Fa = MASK;
             break;
         }
 
@@ -459,71 +462,71 @@ fbCombineConjointGeneralU (uint32_t *dest, const uint32_t *src, int width, uint8
             Fb = fbCombineConjointInPart (da, sa);
             break;
         case CombineB:
-            Fb = 0xff;
+            Fb = MASK;
             break;
         }
         m = FbGen (s,d,0,Fa,Fb,t, u, v);
-        n = FbGen (s,d,8,Fa,Fb,t, u, v);
-        o = FbGen (s,d,16,Fa,Fb,t, u, v);
-        p = FbGen (s,d,24,Fa,Fb,t, u, v);
+        n = FbGen (s,d,G_SHIFT,Fa,Fb,t, u, v);
+        o = FbGen (s,d,B_SHIFT,Fa,Fb,t, u, v);
+        p = FbGen (s,d,A_SHIFT,Fa,Fb,t, u, v);
         s = m|n|o|p;
 	*(dest + i) = s;
     }
 }
 
 FASTCALL static void
-fbCombineConjointOverU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineConjointOverU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineConjointGeneralU (dest, src, width, CombineAOver);
 }
 
 
 FASTCALL static void
-fbCombineConjointOverReverseU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineConjointOverReverseU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineConjointGeneralU (dest, src, width, CombineBOver);
 }
 
 
 FASTCALL static void
-fbCombineConjointInU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineConjointInU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineConjointGeneralU (dest, src, width, CombineAIn);
 }
 
 
 FASTCALL static void
-fbCombineConjointInReverseU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineConjointInReverseU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineConjointGeneralU (dest, src, width, CombineBIn);
 }
 
 FASTCALL static void
-fbCombineConjointOutU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineConjointOutU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineConjointGeneralU (dest, src, width, CombineAOut);
 }
 
 FASTCALL static void
-fbCombineConjointOutReverseU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineConjointOutReverseU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineConjointGeneralU (dest, src, width, CombineBOut);
 }
 
 FASTCALL static void
-fbCombineConjointAtopU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineConjointAtopU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineConjointGeneralU (dest, src, width, CombineAAtop);
 }
 
 FASTCALL static void
-fbCombineConjointAtopReverseU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineConjointAtopReverseU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineConjointGeneralU (dest, src, width, CombineBAtop);
 }
 
 FASTCALL static void
-fbCombineConjointXorU (uint32_t *dest, const uint32_t *src, int width)
+fbCombineConjointXorU (comp4_t *dest, const comp4_t *src, int width)
 {
     fbCombineConjointGeneralU (dest, src, width, CombineXor);
 }
@@ -533,12 +536,12 @@ fbCombineConjointXorU (uint32_t *dest, const uint32_t *src, int width)
 /********************************************************************************/
 
 FASTCALL static void
-fbCombineMaskC (uint32_t *src, uint32_t *mask)
+fbCombineMaskC (comp4_t *src, comp4_t *mask)
 {
-    uint32_t a = *mask;
+    comp4_t a = *mask;
 
-    uint32_t	x;
-    uint16_t	xa;
+    comp4_t	x;
+    comp2_t	xa;
 
     if (!a)
     {
@@ -547,16 +550,16 @@ fbCombineMaskC (uint32_t *src, uint32_t *mask)
     }
 
     x = *(src);
-    if (a == 0xffffffff)
+    if (a == ~0)
     {
-	x = x >> 24;
-	x |= x << 8;
-	x |= x << 16;
+	x = x >> A_SHIFT;
+	x |= x << G_SHIFT;
+	x |= x << B_SHIFT;
 	*(mask) = x;
 	return;
     }
 
-    xa = x >> 24;
+    xa = x >> A_SHIFT;
     FbByteMulC(x, a);
     *(src) = x;
     FbByteMul(a, xa);
@@ -564,10 +567,10 @@ fbCombineMaskC (uint32_t *src, uint32_t *mask)
 }
 
 FASTCALL static void
-fbCombineMaskValueC (uint32_t *src, const uint32_t *mask)
+fbCombineMaskValueC (comp4_t *src, const comp4_t *mask)
 {
-    uint32_t a = *mask;
-    uint32_t	x;
+    comp4_t a = *mask;
+    comp4_t	x;
 
     if (!a)
     {
@@ -575,7 +578,7 @@ fbCombineMaskValueC (uint32_t *src, const uint32_t *mask)
 	return;
     }
 
-    if (a == 0xffffffff)
+    if (a == ~0)
 	return;
 
     x = *(src);
@@ -584,22 +587,22 @@ fbCombineMaskValueC (uint32_t *src, const uint32_t *mask)
 }
 
 FASTCALL static void
-fbCombineMaskAlphaC (const uint32_t *src, uint32_t *mask)
+fbCombineMaskAlphaC (const comp4_t *src, comp4_t *mask)
 {
-    uint32_t a = *(mask);
-    uint32_t	x;
+    comp4_t a = *(mask);
+    comp4_t	x;
 
     if (!a)
 	return;
 
-    x = *(src) >> 24;
-    if (x == 0xff)
+    x = *(src) >> A_SHIFT;
+    if (x == MASK)
 	return;
-    if (a == 0xffffffff)
+    if (a == ~0)
     {
-	x = x >> 24;
-	x |= x << 8;
-	x |= x << 16;
+	x = x >> A_SHIFT;
+	x |= x << G_SHIFT;
+	x |= x << B_SHIFT;
 	*(mask) = x;
 	return;
     }
@@ -611,19 +614,19 @@ fbCombineMaskAlphaC (const uint32_t *src, uint32_t *mask)
 
 
 FASTCALL static void
-fbCombineClearC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineClearC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
-    memset(dest, 0, width*sizeof(uint32_t));
+    memset(dest, 0, width*sizeof(comp4_t));
 }
 
 FASTCALL static void
-fbCombineSrcC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineSrcC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-	uint32_t s = *(src + i);
-	uint32_t m = *(mask + i);
+	comp4_t s = *(src + i);
+	comp4_t m = *(mask + i);
 
 	fbCombineMaskValueC (&s, &m);
 
@@ -632,23 +635,23 @@ fbCombineSrcC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 }
 
 FASTCALL static void
-fbCombineOverC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineOverC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-	uint32_t s = *(src + i);
-	uint32_t m = *(mask + i);
-	uint32_t a;
+	comp4_t s = *(src + i);
+	comp4_t m = *(mask + i);
+	comp4_t a;
 
 	fbCombineMaskC (&s, &m);
 
 	a = ~m;
-        if (a != 0xffffffff)
+        if (a != ~0)
         {
             if (a)
             {
-                uint32_t d = *(dest + i);
+                comp4_t d = *(dest + i);
                 FbByteMulAddC(d, a, s);
                 s = d;
             }
@@ -658,22 +661,22 @@ fbCombineOverC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 }
 
 FASTCALL static void
-fbCombineOverReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineOverReverseC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-        uint32_t d = *(dest + i);
-        uint32_t a = ~d >> 24;
+        comp4_t d = *(dest + i);
+        comp4_t a = ~d >> A_SHIFT;
 
         if (a)
         {
-            uint32_t s = *(src + i);
-	    uint32_t m = *(mask + i);
+            comp4_t s = *(src + i);
+	    comp4_t m = *(mask + i);
 
 	    fbCombineMaskValueC (&s, &m);
 
-            if (a != 0xff)
+            if (a != MASK)
             {
                 FbByteMulAdd(s, a, d);
             }
@@ -683,21 +686,21 @@ fbCombineOverReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 }
 
 FASTCALL static void
-fbCombineInC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineInC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-        uint32_t d = *(dest + i);
-        uint16_t a = d >> 24;
-        uint32_t s = 0;
+        comp4_t d = *(dest + i);
+        comp2_t a = d >> A_SHIFT;
+        comp4_t s = 0;
         if (a)
         {
-	    uint32_t m = *(mask + i);
+	    comp4_t m = *(mask + i);
 
 	    s = *(src + i);
 	    fbCombineMaskValueC (&s, &m);
-            if (a != 0xff)
+            if (a != MASK)
             {
                 FbByteMul(s, a);
             }
@@ -707,21 +710,21 @@ fbCombineInC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 }
 
 FASTCALL static void
-fbCombineInReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineInReverseC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-        uint32_t s = *(src + i);
-        uint32_t m = *(mask + i);
-        uint32_t a;
+        comp4_t s = *(src + i);
+        comp4_t m = *(mask + i);
+        comp4_t a;
 
 	fbCombineMaskAlphaC (&s, &m);
 
 	a = m;
-        if (a != 0xffffffff)
+        if (a != ~0)
         {
-            uint32_t d = 0;
+            comp4_t d = 0;
             if (a)
             {
                 d = *(dest + i);
@@ -733,22 +736,22 @@ fbCombineInReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 }
 
 FASTCALL static void
-fbCombineOutC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineOutC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-        uint32_t d = *(dest + i);
-        uint16_t a = ~d >> 24;
-        uint32_t s = 0;
+        comp4_t d = *(dest + i);
+        comp2_t a = ~d >> A_SHIFT;
+        comp4_t s = 0;
         if (a)
         {
-	    uint32_t m = *(mask + i);
+	    comp4_t m = *(mask + i);
 
 	    s = *(src + i);
 	    fbCombineMaskValueC (&s, &m);
 
-            if (a != 0xff)
+            if (a != MASK)
             {
                 FbByteMul(s, a);
             }
@@ -758,21 +761,21 @@ fbCombineOutC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 }
 
 FASTCALL static void
-fbCombineOutReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineOutReverseC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-	uint32_t s = *(src + i);
-	uint32_t m = *(mask + i);
-	uint32_t a;
+	comp4_t s = *(src + i);
+	comp4_t m = *(mask + i);
+	comp4_t a;
 
 	fbCombineMaskAlphaC (&s, &m);
 
         a = ~m;
-        if (a != 0xffffffff)
+        if (a != ~0)
         {
-            uint32_t d = 0;
+            comp4_t d = 0;
             if (a)
             {
                 d = *(dest + i);
@@ -784,16 +787,16 @@ fbCombineOutReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 }
 
 FASTCALL static void
-fbCombineAtopC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineAtopC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-        uint32_t d = *(dest + i);
-        uint32_t s = *(src + i);
-        uint32_t m = *(mask + i);
-        uint32_t ad;
-        uint16_t as = d >> 24;
+        comp4_t d = *(dest + i);
+        comp4_t s = *(src + i);
+        comp4_t m = *(mask + i);
+        comp4_t ad;
+        comp2_t as = d >> A_SHIFT;
 
 	fbCombineMaskC (&s, &m);
 
@@ -805,17 +808,17 @@ fbCombineAtopC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 }
 
 FASTCALL static void
-fbCombineAtopReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineAtopReverseC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
 
-        uint32_t d = *(dest + i);
-        uint32_t s = *(src + i);
-        uint32_t m = *(mask + i);
-        uint32_t ad;
-        uint16_t as = ~d >> 24;
+        comp4_t d = *(dest + i);
+        comp4_t s = *(src + i);
+        comp4_t m = *(mask + i);
+        comp4_t ad;
+        comp2_t as = ~d >> A_SHIFT;
 
 	fbCombineMaskC (&s, &m);
 
@@ -827,16 +830,16 @@ fbCombineAtopReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 }
 
 FASTCALL static void
-fbCombineXorC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineXorC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-        uint32_t d = *(dest + i);
-        uint32_t s = *(src + i);
-        uint32_t m = *(mask + i);
-        uint32_t ad;
-        uint16_t as = ~d >> 24;
+        comp4_t d = *(dest + i);
+        comp4_t s = *(src + i);
+        comp4_t m = *(mask + i);
+        comp4_t ad;
+        comp2_t as = ~d >> A_SHIFT;
 
 	fbCombineMaskC (&s, &m);
 
@@ -848,14 +851,14 @@ fbCombineXorC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 }
 
 FASTCALL static void
-fbCombineAddC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineAddC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-        uint32_t s = *(src + i);
-        uint32_t m = *(mask + i);
-        uint32_t d = *(dest + i);
+        comp4_t s = *(src + i);
+        comp4_t m = *(mask + i);
+        comp4_t d = *(dest + i);
 
 	fbCombineMaskValueC (&s, &m);
 
@@ -865,15 +868,15 @@ fbCombineAddC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 }
 
 FASTCALL static void
-fbCombineSaturateC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineSaturateC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-        uint32_t  s, d;
-        uint16_t  sa, sr, sg, sb, da;
-        uint16_t  t, u, v;
-        uint32_t  m,n,o,p;
+        comp4_t s, d;
+        comp2_t sa, sr, sg, sb, da;
+        comp2_t t, u, v;
+        comp4_t m,n,o,p;
 
         d = *(dest + i);
         s = *(src + i);
@@ -881,53 +884,53 @@ fbCombineSaturateC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
 
 	fbCombineMaskC (&s, &m);
 
-        sa = (m >> 24);
-        sr = (m >> 16) & 0xff;
-        sg = (m >>  8) & 0xff;
-        sb = (m      ) & 0xff;
-        da = ~d >> 24;
+        sa = (m >> A_SHIFT);
+        sr = (m >> B_SHIFT) & MASK;
+        sg = (m >> G_SHIFT) & MASK;
+        sb =  m             & MASK;
+        da = ~d >> A_SHIFT;
 
         if (sb <= da)
-            m = FbAdd(s,d,0,t);
+            m = Add(s,d,0,t);
         else
-            m = FbGen (s, d, 0, (da << 8) / sb, 0xff, t, u, v);
+            m = FbGen (s, d, 0, (da << G_SHIFT) / sb, MASK, t, u, v);
 
         if (sg <= da)
-            n = FbAdd(s,d,8,t);
+            n = Add(s,d,G_SHIFT,t);
         else
-            n = FbGen (s, d, 8, (da << 8) / sg, 0xff, t, u, v);
+            n = FbGen (s, d, G_SHIFT, (da << G_SHIFT) / sg, MASK, t, u, v);
 
         if (sr <= da)
-            o = FbAdd(s,d,16,t);
+            o = Add(s,d,B_SHIFT,t);
         else
-            o = FbGen (s, d, 16, (da << 8) / sr, 0xff, t, u, v);
+            o = FbGen (s, d, B_SHIFT, (da << G_SHIFT) / sr, MASK, t, u, v);
 
         if (sa <= da)
-            p = FbAdd(s,d,24,t);
+            p = Add(s,d,A_SHIFT,t);
         else
-            p = FbGen (s, d, 24, (da << 8) / sa, 0xff, t, u, v);
+            p = FbGen (s, d, A_SHIFT, (da << G_SHIFT) / sa, MASK, t, u, v);
 
 	*(dest + i) = m|n|o|p;
     }
 }
 
 FASTCALL static void
-fbCombineDisjointGeneralC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width, uint8_t combine)
+fbCombineDisjointGeneralC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width, comp1_t combine)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-        uint32_t  s, d;
-        uint32_t  m,n,o,p;
-        uint32_t  Fa, Fb;
-        uint16_t  t, u, v;
-        uint32_t  sa;
-        uint8_t   da;
+        comp4_t s, d;
+        comp4_t m,n,o,p;
+        comp4_t Fa, Fb;
+        comp2_t t, u, v;
+        comp4_t sa;
+        comp1_t da;
 
         s = *(src + i);
         m = *(mask + i);
         d = *(dest + i);
-        da = d >> 24;
+        da = d >> A_SHIFT;
 
 	fbCombineMaskC (&s, &m);
 
@@ -938,21 +941,21 @@ fbCombineDisjointGeneralC (uint32_t *dest, uint32_t *src, uint32_t *mask, int wi
             Fa = 0;
             break;
         case CombineAOut:
-            m = fbCombineDisjointOutPart ((uint8_t) (sa >> 0), da);
-            n = fbCombineDisjointOutPart ((uint8_t) (sa >> 8), da) << 8;
-            o = fbCombineDisjointOutPart ((uint8_t) (sa >> 16), da) << 16;
-            p = fbCombineDisjointOutPart ((uint8_t) (sa >> 24), da) << 24;
+            m = (comp4_t)fbCombineDisjointOutPart ((comp1_t) (sa >> 0), da);
+            n = (comp4_t)fbCombineDisjointOutPart ((comp1_t) (sa >> G_SHIFT), da) << G_SHIFT;
+            o = (comp4_t)fbCombineDisjointOutPart ((comp1_t) (sa >> B_SHIFT), da) << B_SHIFT;
+            p = (comp4_t)fbCombineDisjointOutPart ((comp1_t) (sa >> A_SHIFT), da) << A_SHIFT;
             Fa = m|n|o|p;
             break;
         case CombineAIn:
-            m = fbCombineDisjointInPart ((uint8_t) (sa >> 0), da);
-            n = fbCombineDisjointInPart ((uint8_t) (sa >> 8), da) << 8;
-            o = fbCombineDisjointInPart ((uint8_t) (sa >> 16), da) << 16;
-            p = fbCombineDisjointInPart ((uint8_t) (sa >> 24), da) << 24;
+            m = (comp4_t)fbCombineDisjointInPart ((comp1_t) (sa >> 0), da);
+            n = (comp4_t)fbCombineDisjointInPart ((comp1_t) (sa >> G_SHIFT), da) << G_SHIFT;
+            o = (comp4_t)fbCombineDisjointInPart ((comp1_t) (sa >> B_SHIFT), da) << B_SHIFT;
+            p = (comp4_t)fbCombineDisjointInPart ((comp1_t) (sa >> A_SHIFT), da) << A_SHIFT;
             Fa = m|n|o|p;
             break;
         case CombineA:
-            Fa = 0xffffffff;
+            Fa = ~0;
             break;
         }
 
@@ -961,97 +964,97 @@ fbCombineDisjointGeneralC (uint32_t *dest, uint32_t *src, uint32_t *mask, int wi
             Fb = 0;
             break;
         case CombineBOut:
-            m = fbCombineDisjointOutPart (da, (uint8_t) (sa >> 0));
-            n = fbCombineDisjointOutPart (da, (uint8_t) (sa >> 8)) << 8;
-            o = fbCombineDisjointOutPart (da, (uint8_t) (sa >> 16)) << 16;
-            p = fbCombineDisjointOutPart (da, (uint8_t) (sa >> 24)) << 24;
+            m = (comp4_t)fbCombineDisjointOutPart (da, (comp1_t) (sa >> 0));
+            n = (comp4_t)fbCombineDisjointOutPart (da, (comp1_t) (sa >> G_SHIFT)) << G_SHIFT;
+            o = (comp4_t)fbCombineDisjointOutPart (da, (comp1_t) (sa >> B_SHIFT)) << B_SHIFT;
+            p = (comp4_t)fbCombineDisjointOutPart (da, (comp1_t) (sa >> A_SHIFT)) << A_SHIFT;
             Fb = m|n|o|p;
             break;
         case CombineBIn:
-            m = fbCombineDisjointInPart (da, (uint8_t) (sa >> 0));
-            n = fbCombineDisjointInPart (da, (uint8_t) (sa >> 8)) << 8;
-            o = fbCombineDisjointInPart (da, (uint8_t) (sa >> 16)) << 16;
-            p = fbCombineDisjointInPart (da, (uint8_t) (sa >> 24)) << 24;
+            m = (comp4_t)fbCombineDisjointInPart (da, (comp1_t) (sa >> 0));
+            n = (comp4_t)fbCombineDisjointInPart (da, (comp1_t) (sa >> G_SHIFT)) << G_SHIFT;
+            o = (comp4_t)fbCombineDisjointInPart (da, (comp1_t) (sa >> B_SHIFT)) << B_SHIFT;
+            p = (comp4_t)fbCombineDisjointInPart (da, (comp1_t) (sa >> A_SHIFT)) << A_SHIFT;
             Fb = m|n|o|p;
             break;
         case CombineB:
-            Fb = 0xffffffff;
+            Fb = ~0;
             break;
         }
-        m = FbGen (s,d,0,FbGet8(Fa,0),FbGet8(Fb,0),t, u, v);
-        n = FbGen (s,d,8,FbGet8(Fa,8),FbGet8(Fb,8),t, u, v);
-        o = FbGen (s,d,16,FbGet8(Fa,16),FbGet8(Fb,16),t, u, v);
-        p = FbGen (s,d,24,FbGet8(Fa,24),FbGet8(Fb,24),t, u, v);
+        m = FbGen (s,d,0,GetComp(Fa,0),GetComp(Fb,0),t, u, v);
+        n = FbGen (s,d,G_SHIFT,GetComp(Fa,G_SHIFT),GetComp(Fb,G_SHIFT),t, u, v);
+        o = FbGen (s,d,B_SHIFT,GetComp(Fa,B_SHIFT),GetComp(Fb,B_SHIFT),t, u, v);
+        p = FbGen (s,d,A_SHIFT,GetComp(Fa,A_SHIFT),GetComp(Fb,A_SHIFT),t, u, v);
         s = m|n|o|p;
 	*(dest + i) = s;
     }
 }
 
 FASTCALL static void
-fbCombineDisjointOverC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineDisjointOverC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineDisjointGeneralC (dest, src, mask, width, CombineAOver);
 }
 
 FASTCALL static void
-fbCombineDisjointInC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineDisjointInC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineDisjointGeneralC (dest, src, mask, width, CombineAIn);
 }
 
 FASTCALL static void
-fbCombineDisjointInReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineDisjointInReverseC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineDisjointGeneralC (dest, src, mask, width, CombineBIn);
 }
 
 FASTCALL static void
-fbCombineDisjointOutC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineDisjointOutC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineDisjointGeneralC (dest, src, mask, width, CombineAOut);
 }
 
 FASTCALL static void
-fbCombineDisjointOutReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineDisjointOutReverseC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineDisjointGeneralC (dest, src, mask, width, CombineBOut);
 }
 
 FASTCALL static void
-fbCombineDisjointAtopC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineDisjointAtopC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineDisjointGeneralC (dest, src, mask, width, CombineAAtop);
 }
 
 FASTCALL static void
-fbCombineDisjointAtopReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineDisjointAtopReverseC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineDisjointGeneralC (dest, src, mask, width, CombineBAtop);
 }
 
 FASTCALL static void
-fbCombineDisjointXorC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineDisjointXorC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineDisjointGeneralC (dest, src, mask, width, CombineXor);
 }
 
 FASTCALL static void
-fbCombineConjointGeneralC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width, uint8_t combine)
+fbCombineConjointGeneralC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width, comp1_t combine)
 {
     int i;
 
     for (i = 0; i < width; ++i) {
-        uint32_t  s, d;
-        uint32_t  m,n,o,p;
-        uint32_t  Fa, Fb;
-        uint16_t  t, u, v;
-        uint32_t  sa;
-        uint8_t   da;
+        comp4_t s, d;
+        comp4_t m,n,o,p;
+        comp4_t Fa, Fb;
+        comp2_t t, u, v;
+        comp4_t sa;
+        comp1_t da;
 
         s = *(src + i);
         m = *(mask + i);
         d = *(dest + i);
-        da = d >> 24;
+        da = d >> A_SHIFT;
 
 	fbCombineMaskC (&s, &m);
 
@@ -1062,21 +1065,21 @@ fbCombineConjointGeneralC (uint32_t *dest, uint32_t *src, uint32_t *mask, int wi
             Fa = 0;
             break;
         case CombineAOut:
-            m = fbCombineConjointOutPart ((uint8_t) (sa >> 0), da);
-            n = fbCombineConjointOutPart ((uint8_t) (sa >> 8), da) << 8;
-            o = fbCombineConjointOutPart ((uint8_t) (sa >> 16), da) << 16;
-            p = fbCombineConjointOutPart ((uint8_t) (sa >> 24), da) << 24;
+            m = (comp4_t)fbCombineConjointOutPart ((comp1_t) (sa >> 0), da);
+            n = (comp4_t)fbCombineConjointOutPart ((comp1_t) (sa >> G_SHIFT), da) << G_SHIFT;
+            o = (comp4_t)fbCombineConjointOutPart ((comp1_t) (sa >> B_SHIFT), da) << B_SHIFT;
+            p = (comp4_t)fbCombineConjointOutPart ((comp1_t) (sa >> A_SHIFT), da) << A_SHIFT;
             Fa = m|n|o|p;
             break;
         case CombineAIn:
-            m = fbCombineConjointInPart ((uint8_t) (sa >> 0), da);
-            n = fbCombineConjointInPart ((uint8_t) (sa >> 8), da) << 8;
-            o = fbCombineConjointInPart ((uint8_t) (sa >> 16), da) << 16;
-            p = fbCombineConjointInPart ((uint8_t) (sa >> 24), da) << 24;
+            m = (comp4_t)fbCombineConjointInPart ((comp1_t) (sa >> 0), da);
+            n = (comp4_t)fbCombineConjointInPart ((comp1_t) (sa >> G_SHIFT), da) << G_SHIFT;
+            o = (comp4_t)fbCombineConjointInPart ((comp1_t) (sa >> B_SHIFT), da) << B_SHIFT;
+            p = (comp4_t)fbCombineConjointInPart ((comp1_t) (sa >> A_SHIFT), da) << A_SHIFT;
             Fa = m|n|o|p;
             break;
         case CombineA:
-            Fa = 0xffffffff;
+            Fa = ~0;
             break;
         }
 
@@ -1085,87 +1088,87 @@ fbCombineConjointGeneralC (uint32_t *dest, uint32_t *src, uint32_t *mask, int wi
             Fb = 0;
             break;
         case CombineBOut:
-            m = fbCombineConjointOutPart (da, (uint8_t) (sa >> 0));
-            n = fbCombineConjointOutPart (da, (uint8_t) (sa >> 8)) << 8;
-            o = fbCombineConjointOutPart (da, (uint8_t) (sa >> 16)) << 16;
-            p = fbCombineConjointOutPart (da, (uint8_t) (sa >> 24)) << 24;
+            m = (comp4_t)fbCombineConjointOutPart (da, (comp1_t) (sa >> 0));
+            n = (comp4_t)fbCombineConjointOutPart (da, (comp1_t) (sa >> G_SHIFT)) << G_SHIFT;
+            o = (comp4_t)fbCombineConjointOutPart (da, (comp1_t) (sa >> B_SHIFT)) << B_SHIFT;
+            p = (comp4_t)fbCombineConjointOutPart (da, (comp1_t) (sa >> A_SHIFT)) << A_SHIFT;
             Fb = m|n|o|p;
             break;
         case CombineBIn:
-            m = fbCombineConjointInPart (da, (uint8_t) (sa >> 0));
-            n = fbCombineConjointInPart (da, (uint8_t) (sa >> 8)) << 8;
-            o = fbCombineConjointInPart (da, (uint8_t) (sa >> 16)) << 16;
-            p = fbCombineConjointInPart (da, (uint8_t) (sa >> 24)) << 24;
+            m = (comp4_t)fbCombineConjointInPart (da, (comp1_t) (sa >> 0));
+            n = (comp4_t)fbCombineConjointInPart (da, (comp1_t) (sa >> G_SHIFT)) << G_SHIFT;
+            o = (comp4_t)fbCombineConjointInPart (da, (comp1_t) (sa >> B_SHIFT)) << B_SHIFT;
+            p = (comp4_t)fbCombineConjointInPart (da, (comp1_t) (sa >> A_SHIFT)) << A_SHIFT;
             Fb = m|n|o|p;
             break;
         case CombineB:
-            Fb = 0xffffffff;
+            Fb = ~0;
             break;
         }
-        m = FbGen (s,d,0,FbGet8(Fa,0),FbGet8(Fb,0),t, u, v);
-        n = FbGen (s,d,8,FbGet8(Fa,8),FbGet8(Fb,8),t, u, v);
-        o = FbGen (s,d,16,FbGet8(Fa,16),FbGet8(Fb,16),t, u, v);
-        p = FbGen (s,d,24,FbGet8(Fa,24),FbGet8(Fb,24),t, u, v);
+        m = FbGen (s,d,0,GetComp(Fa,0),GetComp(Fb,0),t, u, v);
+        n = FbGen (s,d,G_SHIFT,GetComp(Fa,G_SHIFT),GetComp(Fb,G_SHIFT),t, u, v);
+        o = FbGen (s,d,B_SHIFT,GetComp(Fa,B_SHIFT),GetComp(Fb,B_SHIFT),t, u, v);
+        p = FbGen (s,d,A_SHIFT,GetComp(Fa,A_SHIFT),GetComp(Fb,A_SHIFT),t, u, v);
         s = m|n|o|p;
 	*(dest + i) = s;
     }
 }
 
 FASTCALL static void
-fbCombineConjointOverC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineConjointOverC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineConjointGeneralC (dest, src, mask, width, CombineAOver);
 }
 
 FASTCALL static void
-fbCombineConjointOverReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineConjointOverReverseC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineConjointGeneralC (dest, src, mask, width, CombineBOver);
 }
 
 FASTCALL static void
-fbCombineConjointInC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineConjointInC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineConjointGeneralC (dest, src, mask, width, CombineAIn);
 }
 
 FASTCALL static void
-fbCombineConjointInReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineConjointInReverseC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineConjointGeneralC (dest, src, mask, width, CombineBIn);
 }
 
 FASTCALL static void
-fbCombineConjointOutC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineConjointOutC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineConjointGeneralC (dest, src, mask, width, CombineAOut);
 }
 
 FASTCALL static void
-fbCombineConjointOutReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineConjointOutReverseC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineConjointGeneralC (dest, src, mask, width, CombineBOut);
 }
 
 FASTCALL static void
-fbCombineConjointAtopC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineConjointAtopC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineConjointGeneralC (dest, src, mask, width, CombineAAtop);
 }
 
 FASTCALL static void
-fbCombineConjointAtopReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineConjointAtopReverseC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineConjointGeneralC (dest, src, mask, width, CombineBAtop);
 }
 
 FASTCALL static void
-fbCombineConjointXorC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+fbCombineConjointXorC (comp4_t *dest, comp4_t *src, comp4_t *mask, int width)
 {
     fbCombineConjointGeneralC (dest, src, mask, width, CombineXor);
 }
 
-CombineFuncU pixman_fbCombineFuncU[] = {
+static CombineFuncU pixman_fbCombineFuncU[] = {
     fbCombineClear,
     fbCombineSrcU,
     NULL, /* CombineDst */
@@ -1212,7 +1215,7 @@ CombineFuncU pixman_fbCombineFuncU[] = {
     fbCombineConjointXorU,
 };
 
-CombineFuncC pixman_fbCombineFuncC[] = {
+static CombineFuncC pixman_fbCombineFuncC[] = {
     fbCombineClearC,
     fbCombineSrcC,
     NULL, /* Dest */
@@ -1258,3 +1261,9 @@ CombineFuncC pixman_fbCombineFuncC[] = {
     fbCombineConjointAtopReverseC,
     fbCombineConjointXorC,
 };
+
+FbComposeFunctions pixman_composeFunctions = {
+    pixman_fbCombineFuncU,
+    pixman_fbCombineFuncC,
+    pixman_fbCombineMaskU
+};
diff --git a/lib/pixman/pixman/combine.pl b/lib/pixman/pixman/combine.pl
new file mode 100644
index 000000000..f8df60374
--- /dev/null
+++ b/lib/pixman/pixman/combine.pl
@@ -0,0 +1,81 @@
+$usage = "Usage: combine.pl { 8 | 16 } < combine.inc";
+
+$#ARGV == 0 or die $usage;
+
+# Get the component size.
+$size = int($ARGV[0]);
+$size == 8 or $size == 16 or die $usage;
+
+$pixel_size = $size * 4;
+$half_pixel_size = $size * 2;
+
+sub mask {
+    my $str = shift;
+    my $suffix;
+    $suffix = "ULL" if $size > 8;
+
+    return "0x" . $str . $suffix;
+}
+
+# Generate mask strings.
+$nibbles = $size / 4;
+$mask = "f" x $nibbles;
+$zero_mask = "0" x $nibbles;
+$one_half = "8" . "0" x ($nibbles - 1);
+
+print "/* WARNING: This file is generated by combine.pl from combine.inc.\n";
+print "   Please edit one of those files rather than this one. */\n";
+print "\n";
+
+print "#line 1 \"combine.inc\"\n";
+
+$mask_ = mask($mask);
+$one_half_ = mask($one_half);
+$g_mask = mask($mask . $zero_mask);
+$b_mask = mask($mask . $zero_mask x 2);
+$a_mask = mask($mask . $zero_mask x 3);
+$rb_mask = mask($mask . $zero_mask . $mask);
+$ag_mask = mask($mask . $zero_mask . $mask . $zero_mask);
+$rb_one_half = mask($one_half . $zero_mask . $one_half);
+$rb_mask_plus_one = mask("1" . $zero_mask x 2 . "1" .  $zero_mask);
+
+while (<STDIN>) {
+    # Mask and 1/2 value for a single component.
+    s/#define COMPONENT_SIZE\b/$& $size/;
+    s/#define MASK\b/$& $mask_/;
+    s/#define ONE_HALF\b/$& $one_half_/;
+
+    # Shifts and masks for green, blue, and alpha.
+    s/#define G_SHIFT\b/$& $size/;
+    s/#define B_SHIFT\b/$& $size * 2/;
+    s/#define A_SHIFT\b/$& $size * 3/;
+    s/#define G_MASK\b/$& $g_mask/;
+    s/#define B_MASK\b/$& $b_mask/;
+    s/#define A_MASK\b/$& $a_mask/;
+
+    # Special values for dealing with red + blue at the same time.
+    s/#define RB_MASK\b/$& $rb_mask/;
+    s/#define AG_MASK\b/$& $ag_mask/;
+    s/#define RB_ONE_HALF\b/$& $rb_one_half/;
+    s/#define RB_MASK_PLUS_ONE\b/$& $rb_mask_plus_one/;
+
+    # Add 32/64 suffix to combining function types.
+    s/\bCombineFuncC\b/CombineFuncC$pixel_size/;
+    s/\bCombineFuncU\b/CombineFuncU$pixel_size/;
+    s/\bCombineMaskU\b/CombineMaskU$pixel_size/;
+    s/\bFbComposeFunctions\b/FbComposeFunctions$pixel_size/;
+
+    # Convert comp*_t values into the appropriate real types.
+    s/comp1_t/uint${size}_t/g;
+    s/comp2_t/uint${half_pixel_size}_t/g;
+    s/comp4_t/uint${pixel_size}_t/g;
+
+    # Change the function table name for the 64-bit version.
+    s/pixman_composeFunctions/pixman_composeFunctions64/ if $size == 16;
+
+    # Change the header for the 64-bit version
+    s/pixman-combine.h/pixman-combine64.h/ if $size == 16;
+    s/pixman-combine.h/pixman-combine32.h/ if $size == 8;
+
+    print;
+}
diff --git a/lib/pixman/pixman/pixman-access.c b/lib/pixman/pixman/pixman-access.c
index 94af2065c..d01f8bd59 100644
--- a/lib/pixman/pixman/pixman-access.c
+++ b/lib/pixman/pixman/pixman-access.c
@@ -2,6 +2,7 @@
  *
  * Copyright © 2000 Keith Packard, member of The XFree86 Project, Inc.
  *             2005 Lars Knoll & Zack Rusin, Trolltech
+ *             2008 Aaron Plattner, NVIDIA Corporation
  *
  * Permission to use, copy, modify, distribute, and sell this software and its
  * documentation for any purpose is hereby granted without fee, provided that
@@ -27,19 +28,15 @@
 #include <config.h>
 #endif
 
+#include <stdlib.h>
 #include <string.h>
+#include <assert.h>
 
 #include "pixman-private.h"
 
-#ifdef PIXMAN_FB_ACCESSORS
-#define FETCH_PROC_FOR_PICTURE pixman_fetchProcForPicture_accessors
-#define FETCH_PIXEL_PROC_FOR_PICTURE pixman_fetchPixelProcForPicture_accessors
-#define STORE_PROC_FOR_PICTURE pixman_storeProcForPicture_accessors
-#else
-#define FETCH_PROC_FOR_PICTURE pixman_fetchProcForPicture
-#define FETCH_PIXEL_PROC_FOR_PICTURE pixman_fetchPixelProcForPicture
-#define STORE_PROC_FOR_PICTURE pixman_storeProcForPicture
-#endif
+#define Red(x) (((x) >> 16) & 0xff)
+#define Green(x) (((x) >> 8) & 0xff)
+#define Blue(x) ((x) & 0xff)
 
 /*
  * YV12 setup and access macros
@@ -123,6 +120,52 @@ fbFetch_x8b8g8r8 (bits_image_t *pict, int x, int y, int width, uint32_t *buffer)
 }
 
 static FASTCALL void
+fbFetch_a2b10g10r10 (bits_image_t *pict, int x, int y, int width, uint64_t *buffer)
+{
+    const uint32_t *bits = pict->bits + y*pict->rowstride;
+    const uint32_t *pixel = bits + x;
+    const uint32_t *end = pixel + width;
+    while (pixel < end) {
+        uint32_t p = READ(pict, pixel++);
+        uint64_t a = p >> 30;
+        uint64_t b = (p >> 20) & 0x3ff;
+        uint64_t g = (p >> 10) & 0x3ff;
+        uint64_t r = p & 0x3ff;
+
+        r = r << 6 | r >> 4;
+        g = g << 6 | g >> 4;
+        b = b << 6 | b >> 4;
+
+        a <<= 62;
+        a |= a >> 2;
+        a |= a >> 4;
+        a |= a >> 8;
+
+        *buffer++ = a << 48 | r << 32 | g << 16 | b;
+    }
+}
+
+static FASTCALL void
+fbFetch_x2b10g10r10 (bits_image_t *pict, int x, int y, int width, uint64_t *buffer)
+{
+    const uint32_t *bits = pict->bits + y*pict->rowstride;
+    const uint32_t *pixel = (uint32_t *)bits + x;
+    const uint32_t *end = pixel + width;
+    while (pixel < end) {
+        uint32_t p = READ(pict, pixel++);
+        uint64_t b = (p >> 20) & 0x3ff;
+        uint64_t g = (p >> 10) & 0x3ff;
+        uint64_t r = p & 0x3ff;
+
+        r = r << 6 | r >> 4;
+        g = g << 6 | g >> 4;
+        b = b << 6 | b >> 4;
+
+        *buffer++ = 0xffffULL << 48 | r << 32 | g << 16 | b;
+    }
+}
+
+static FASTCALL void
 fbFetch_r8g8b8 (bits_image_t *pict, int x, int y, int width, uint32_t *buffer)
 {
     const uint32_t *bits = pict->bits + y*pict->rowstride;
@@ -643,13 +686,16 @@ fbFetch_yv12 (bits_image_t *pict, int x, int line, int width, uint32_t *buffer)
     }
 }
 
-fetchProc FETCH_PROC_FOR_PICTURE (bits_image_t * pict)
+fetchProc32 ACCESS(pixman_fetchProcForPicture32) (bits_image_t * pict)
 {
     switch(pict->format) {
     case PIXMAN_a8r8g8b8: return fbFetch_a8r8g8b8;
     case PIXMAN_x8r8g8b8: return fbFetch_x8r8g8b8;
     case PIXMAN_a8b8g8r8: return fbFetch_a8b8g8r8;
     case PIXMAN_x8b8g8r8: return fbFetch_x8b8g8r8;
+    /* These two require wide compositing */
+    case PIXMAN_a2b10g10r10: return NULL;
+    case PIXMAN_x2b10g10r10: return NULL;
 
         /* 24bpp formats */
     case PIXMAN_r8g8b8: return fbFetch_r8g8b8;
@@ -699,8 +745,66 @@ fetchProc FETCH_PROC_FOR_PICTURE (bits_image_t * pict)
     return NULL;
 }
 
+static FASTCALL void
+fbFetch64_generic (bits_image_t *pict, int x, int y, int width, uint64_t *buffer)
+{
+    fetchProc32 fetch32 = ACCESS(pixman_fetchProcForPicture32) (pict);
+
+    // Fetch the pixels into the first half of buffer and then expand them in
+    // place.
+    fetch32(pict, x, y, width, (uint32_t*)buffer);
+    pixman_expand(buffer, (uint32_t*)buffer, pict->format, width);
+}
+
+fetchProc64 ACCESS(pixman_fetchProcForPicture64) (bits_image_t * pict)
+{
+    switch(pict->format) {
+    case PIXMAN_a2b10g10r10: return fbFetch_a2b10g10r10;
+    case PIXMAN_x2b10g10r10: return fbFetch_x2b10g10r10;
+    default: return fbFetch64_generic;
+    }
+}
+
 /**************************** Pixel wise fetching *****************************/
 
+static FASTCALL uint64_t
+fbFetchPixel_a2b10g10r10 (bits_image_t *pict, int offset, int line)
+{
+    uint32_t *bits = pict->bits + line*pict->rowstride;
+    uint32_t p = READ(pict, bits + offset);
+    uint64_t a = p >> 30;
+    uint64_t b = (p >> 20) & 0x3ff;
+    uint64_t g = (p >> 10) & 0x3ff;
+    uint64_t r = p & 0x3ff;
+
+    r = r << 6 | r >> 4;
+    g = g << 6 | g >> 4;
+    b = b << 6 | b >> 4;
+
+    a <<= 62;
+    a |= a >> 2;
+    a |= a >> 4;
+    a |= a >> 8;
+
+    return a << 48 | r << 32 | g << 16 | b;
+}
+
+static FASTCALL uint64_t
+fbFetchPixel_x2b10g10r10 (bits_image_t *pict, int offset, int line)
+{
+    uint32_t *bits = pict->bits + line*pict->rowstride;
+    uint32_t p = READ(pict, bits + offset);
+    uint64_t b = (p >> 20) & 0x3ff;
+    uint64_t g = (p >> 10) & 0x3ff;
+    uint64_t r = p & 0x3ff;
+
+    r = r << 6 | r >> 4;
+    g = g << 6 | g >> 4;
+    b = b << 6 | b >> 4;
+
+    return 0xffffULL << 48 | r << 32 | g << 16 | b;
+}
+
 static FASTCALL uint32_t
 fbFetchPixel_a8r8g8b8 (bits_image_t *pict, int offset, int line)
 {
@@ -1154,13 +1258,34 @@ fbFetchPixel_yv12 (bits_image_t *pict, int offset, int line)
 	(b >= 0 ? b < 0x1000000 ? (b >> 16) & 0x0000ff : 0x0000ff : 0);
 }
 
-fetchPixelProc FETCH_PIXEL_PROC_FOR_PICTURE (bits_image_t * pict)
+/*
+ * XXX: The transformed fetch path only works at 32-bpp so far.  When all paths
+ * have wide versions, this can be removed.
+ *
+ * WARNING: This function loses precision!
+ */
+static FASTCALL uint32_t
+fbFetchPixel32_generic_lossy (bits_image_t *pict, int offset, int line)
+{
+    fetchPixelProc64 fetchPixel64 = ACCESS(pixman_fetchPixelProcForPicture64) (pict);
+    const uint64_t argb16Pixel = fetchPixel64(pict, offset, line);
+    uint32_t argb8Pixel;
+
+    pixman_contract(&argb8Pixel, &argb16Pixel, 1);
+
+    return argb8Pixel;
+}
+
+fetchPixelProc32 ACCESS(pixman_fetchPixelProcForPicture32) (bits_image_t * pict)
 {
     switch(pict->format) {
     case PIXMAN_a8r8g8b8: return fbFetchPixel_a8r8g8b8;
     case PIXMAN_x8r8g8b8: return fbFetchPixel_x8r8g8b8;
     case PIXMAN_a8b8g8r8: return fbFetchPixel_a8b8g8r8;
     case PIXMAN_x8b8g8r8: return fbFetchPixel_x8b8g8r8;
+    /* These two require wide compositing */
+    case PIXMAN_a2b10g10r10: return fbFetchPixel32_generic_lossy;
+    case PIXMAN_x2b10g10r10: return fbFetchPixel32_generic_lossy;
 
         /* 24bpp formats */
     case PIXMAN_r8g8b8: return fbFetchPixel_r8g8b8;
@@ -1210,12 +1335,62 @@ fetchPixelProc FETCH_PIXEL_PROC_FOR_PICTURE (bits_image_t * pict)
     return NULL;
 }
 
+static FASTCALL uint64_t
+fbFetchPixel64_generic (bits_image_t *pict, int offset, int line)
+{
+    fetchPixelProc32 fetchPixel32 = ACCESS(pixman_fetchPixelProcForPicture32) (pict);
+    uint32_t argb8Pixel = fetchPixel32(pict, offset, line);
+    uint64_t argb16Pixel;
+
+    pixman_expand(&argb16Pixel, &argb8Pixel, pict->format, 1);
+
+    return argb16Pixel;
+}
+
+fetchPixelProc64 ACCESS(pixman_fetchPixelProcForPicture64) (bits_image_t * pict)
+{
+    switch(pict->format) {
+    case PIXMAN_a2b10g10r10: return fbFetchPixel_a2b10g10r10;
+    case PIXMAN_x2b10g10r10: return fbFetchPixel_x2b10g10r10;
+    default: return fbFetchPixel64_generic;
+    }
+}
+
 /*********************************** Store ************************************/
 
 #define Splita(v)	uint32_t	a = ((v) >> 24), r = ((v) >> 16) & 0xff, g = ((v) >> 8) & 0xff, b = (v) & 0xff
 #define Split(v)	uint32_t	r = ((v) >> 16) & 0xff, g = ((v) >> 8) & 0xff, b = (v) & 0xff
 
 static FASTCALL void
+fbStore_a2b10g10r10 (pixman_image_t *image,
+		     uint32_t *bits, const uint64_t *values, int x, int width, const pixman_indexed_t * indexed)
+{
+    int i;
+    uint32_t *pixel = bits + x;
+    for (i = 0; i < width; ++i) {
+        WRITE(image, pixel++,
+            ((values[i] >> 32) & 0xc0000000) | // A
+            ((values[i] >> 38) & 0x3ff) |      // R
+            ((values[i] >> 12) & 0xffc00) |    // G
+            ((values[i] << 14) & 0x3ff00000)); // B
+    }
+}
+
+static FASTCALL void
+fbStore_x2b10g10r10 (pixman_image_t *image,
+		     uint32_t *bits, const uint64_t *values, int x, int width, const pixman_indexed_t * indexed)
+{
+    int i;
+    uint32_t *pixel = bits + x;
+    for (i = 0; i < width; ++i) {
+        WRITE(image, pixel++,
+            ((values[i] >> 38) & 0x3ff) |      // R
+            ((values[i] >> 12) & 0xffc00) |    // G
+            ((values[i] << 14) & 0x3ff00000)); // B
+    }
+}
+
+static FASTCALL void
 fbStore_a8r8g8b8 (pixman_image_t *image,
 		  uint32_t *bits, const uint32_t *values, int x, int width, const pixman_indexed_t * indexed)
 {
@@ -1465,7 +1640,7 @@ fbStore_b2g3r3 (pixman_image_t *image,
 	Split(values[i]);
 	WRITE(image, pixel++,
 	      ((b     ) & 0xc0) |
-	      ((g >> 2) & 0x1c) |
+	      ((g >> 2) & 0x38) |
 	      ((r >> 5)       ));
     }
 }
@@ -1634,7 +1809,7 @@ fbStore_g1 (pixman_image_t *image,
 }
 
 
-storeProc STORE_PROC_FOR_PICTURE (bits_image_t * pict)
+storeProc32 ACCESS(pixman_storeProcForPicture32) (bits_image_t * pict)
 {
     switch(pict->format) {
     case PIXMAN_a8r8g8b8: return fbStore_a8r8g8b8;
@@ -1684,3 +1859,128 @@ storeProc STORE_PROC_FOR_PICTURE (bits_image_t * pict)
         return NULL;
     }
 }
+
+/*
+ * Contracts a 64bpp image to 32bpp and then stores it using a regular 32-bit
+ * store proc.
+ */
+static FASTCALL void
+fbStore64_generic (pixman_image_t *image,
+		   uint32_t *bits, const uint64_t *values, int x, int width, const pixman_indexed_t * indexed)
+{
+    bits_image_t *pict = (bits_image_t*)image;
+    storeProc32 store32 = ACCESS(pixman_storeProcForPicture32) (pict);
+    uint32_t *argb8Pixels;
+
+    assert(image->common.type == BITS);
+    assert(store32);
+
+    argb8Pixels = pixman_malloc_ab (width, sizeof(uint32_t));
+    if (!argb8Pixels) return;
+
+    // Contract the scanline.  We could do this in place if values weren't
+    // const.
+    pixman_contract(argb8Pixels, values, width);
+    store32(image, bits, argb8Pixels, x, width, indexed);
+
+    free(argb8Pixels);
+}
+
+storeProc64 ACCESS(pixman_storeProcForPicture64) (bits_image_t * pict)
+{
+    switch(pict->format) {
+    case PIXMAN_a2b10g10r10: return fbStore_a2b10g10r10;
+    case PIXMAN_x2b10g10r10: return fbStore_x2b10g10r10;
+    default: return fbStore64_generic;
+    }
+}
+
+#ifndef PIXMAN_FB_ACCESSORS
+/*
+ * Helper routine to expand a color component from 0 < n <= 8 bits to 16 bits by
+ * replication.
+ */
+static inline uint64_t expand16(const uint8_t val, int nbits)
+{
+    // Start out with the high bit of val in the high bit of result.
+    uint16_t result = (uint16_t)val << (16 - nbits);
+
+    if (nbits == 0)
+        return 0;
+
+    // Copy the bits in result, doubling the number of bits each time, until we
+    // fill all 16 bits.
+    while (nbits < 16) {
+        result |= result >> nbits;
+        nbits *= 2;
+    }
+
+    return result;
+}
+
+/*
+ * This function expands images from ARGB8 format to ARGB16.  To preserve
+ * precision, it needs to know the original source format.  For example, if the
+ * source was PIXMAN_x1r5g5b5 and the red component contained bits 12345, then
+ * the expanded value is 12345123.  To correctly expand this to 16 bits, it
+ * should be 1234512345123451 and not 1234512312345123.
+ */
+void pixman_expand(uint64_t *dst, const uint32_t *src,
+                   pixman_format_code_t format, int width)
+{
+    /*
+     * Determine the sizes of each component and the masks and shifts required
+     * to extract them from the source pixel.
+     */
+    const int a_size = PIXMAN_FORMAT_A(format),
+              r_size = PIXMAN_FORMAT_R(format),
+              g_size = PIXMAN_FORMAT_G(format),
+              b_size = PIXMAN_FORMAT_B(format);
+    const int a_shift = 32 - a_size,
+              r_shift = 24 - r_size,
+              g_shift = 16 - g_size,
+              b_shift =  8 - b_size;
+    const uint8_t a_mask = ~(~0 << a_size),
+                  r_mask = ~(~0 << r_size),
+                  g_mask = ~(~0 << g_size),
+                  b_mask = ~(~0 << b_size);
+    int i;
+
+    /* Start at the end so that we can do the expansion in place when src == dst */
+    for (i = width - 1; i >= 0; i--)
+    {
+        const uint32_t pixel = src[i];
+        // Extract the components.
+        const uint8_t a = (pixel >> a_shift) & a_mask,
+                      r = (pixel >> r_shift) & r_mask,
+                      g = (pixel >> g_shift) & g_mask,
+                      b = (pixel >> b_shift) & b_mask;
+        const uint64_t a16 = a_size ? expand16(a, a_size) : 0xffff,
+                       r16 = expand16(r, r_size),
+                       g16 = expand16(g, g_size),
+                       b16 = expand16(b, b_size);
+
+        dst[i] = a16 << 48 | r16 << 32 | g16 << 16 | b16;
+    }
+}
+
+/*
+ * Contracting is easier than expanding.  We just need to truncate the
+ * components.
+ */
+void pixman_contract(uint32_t *dst, const uint64_t *src, int width)
+{
+    int i;
+
+    /* Start at the beginning so that we can do the contraction in place when
+     * src == dst */
+    for (i = 0; i < width; i++)
+    {
+        const uint8_t a = src[i] >> 56,
+                      r = src[i] >> 40,
+                      g = src[i] >> 24,
+                      b = src[i] >> 8;
+        dst[i] = a << 24 | r << 16 | g << 8 | b;
+    }
+}
+#endif // PIXMAN_FB_ACCESSORS
diff --git a/lib/pixman/pixman/pixman-compose.c b/lib/pixman/pixman/pixman-compose.c
index 6f524d6f8..c583ea60e 100644
--- a/lib/pixman/pixman/pixman-compose.c
+++ b/lib/pixman/pixman/pixman-compose.c
@@ -2,6 +2,7 @@
  *
  * Copyright © 2000 Keith Packard, member of The XFree86 Project, Inc.
  *             2005 Lars Knoll & Zack Rusin, Trolltech
+ *             2008 Aaron Plattner, NVIDIA Corporation
  *
  * Permission to use, copy, modify, distribute, and sell this software and its
  * documentation for any purpose is hereby granted without fee, provided that
@@ -37,29 +38,8 @@
 
 #ifdef PIXMAN_FB_ACCESSORS
 #define PIXMAN_COMPOSITE_RECT_GENERAL pixman_composite_rect_general_accessors
-#define PIXMAN_COMPOSE_FUNCTIONS pixman_composeFunctions_accessors
-
-#define FETCH_PROC_FOR_PICTURE pixman_fetchProcForPicture_accessors
-#define FETCH_PIXEL_PROC_FOR_PICTURE pixman_fetchPixelProcForPicture_accessors
-#define STORE_PROC_FOR_PICTURE pixman_storeProcForPicture_accessors
-
-#define FB_FETCH_TRANSFORMED fbFetchTransformed_accessors
-#define FB_FETCH_EXTERNAL_ALPHA fbFetchExternalAlpha_accessors
-#define FB_STORE_EXTERNAL_ALPHA fbStoreExternalAlpha_accessors
-
 #else
-
 #define PIXMAN_COMPOSITE_RECT_GENERAL pixman_composite_rect_general_no_accessors
-#define PIXMAN_COMPOSE_FUNCTIONS pixman_composeFunctions
-
-#define FETCH_PROC_FOR_PICTURE pixman_fetchProcForPicture
-#define FETCH_PIXEL_PROC_FOR_PICTURE pixman_fetchPixelProcForPicture
-#define STORE_PROC_FOR_PICTURE pixman_storeProcForPicture
-
-#define FB_FETCH_TRANSFORMED fbFetchTransformed
-#define FB_FETCH_EXTERNAL_ALPHA fbFetchExternalAlpha
-#define FB_STORE_EXTERNAL_ALPHA fbStoreExternalAlpha
-
 #endif
 
 static unsigned int
@@ -126,7 +106,20 @@ static void fbFetchSolid(bits_image_t * pict, int x, int y, int width, uint32_t
 {
     uint32_t color;
     uint32_t *end;
-    fetchPixelProc fetch = FETCH_PIXEL_PROC_FOR_PICTURE(pict);
+    fetchPixelProc32 fetch = ACCESS(pixman_fetchPixelProcForPicture32)(pict);
+
+    color = fetch(pict, 0, 0);
+
+    end = buffer + width;
+    while (buffer < end)
+	*(buffer++) = color;
+}
+
+static void fbFetchSolid64(bits_image_t * pict, int x, int y, int width, uint64_t *buffer, void *unused, uint32_t unused2)
+{
+    uint64_t color;
+    uint64_t *end;
+    fetchPixelProc64 fetch = ACCESS(pixman_fetchPixelProcForPicture64)(pict);
 
     color = fetch(pict, 0, 0);
 
@@ -137,26 +130,38 @@ static void fbFetchSolid(bits_image_t * pict, int x, int y, int width, uint32_t
 
 static void fbFetch(bits_image_t * pict, int x, int y, int width, uint32_t *buffer, uint32_t *mask, uint32_t maskBits)
 {
-    fetchProc fetch = FETCH_PROC_FOR_PICTURE(pict);
+    fetchProc32 fetch = ACCESS(pixman_fetchProcForPicture32)(pict);
 
     fetch(pict, x, y, width, buffer);
 }
 
-#ifdef PIXMAN_FB_ACCESSORS	/* The accessor version can't be parameterized from outside */
-static const
-#endif
-FbComposeFunctions PIXMAN_COMPOSE_FUNCTIONS = {
-    pixman_fbCombineFuncU,
-    pixman_fbCombineFuncC,
-    pixman_fbCombineMaskU
-};
+static void fbFetch64(bits_image_t * pict, int x, int y, int width, uint64_t *buffer, void *unused, uint32_t unused2)
+{
+    fetchProc64 fetch = ACCESS(pixman_fetchProcForPicture64)(pict);
+
+    fetch(pict, x, y, width, buffer);
+}
 
 static void
 fbStore(bits_image_t * pict, int x, int y, int width, uint32_t *buffer)
 {
     uint32_t *bits;
     int32_t stride;
-    storeProc store = STORE_PROC_FOR_PICTURE(pict);
+    storeProc32 store = ACCESS(pixman_storeProcForPicture32)(pict);
+    const pixman_indexed_t * indexed = pict->indexed;
+
+    bits = pict->bits;
+    stride = pict->rowstride;
+    bits += y*stride;
+    store((pixman_image_t *)pict, bits, buffer, x, width, indexed);
+}
+
+static void
+fbStore64(bits_image_t * pict, int x, int y, int width, uint64_t *buffer)
+{
+    uint32_t *bits;
+    int32_t stride;
+    storeProc64 store = ACCESS(pixman_storeProcForPicture64)(pict);
     const pixman_indexed_t * indexed = pict->indexed;
 
     bits = pict->bits;
@@ -169,15 +174,70 @@ typedef void (*scanStoreProc)(pixman_image_t *, int, int, int, uint32_t *);
 typedef void (*scanFetchProc)(pixman_image_t *, int, int, int, uint32_t *,
 			      uint32_t *, uint32_t);
 
+static inline scanFetchProc get_fetch_source_pict(const int wide)
+{
+    if (wide)
+	return (scanFetchProc)pixmanFetchSourcePict64;
+    else
+	return (scanFetchProc)pixmanFetchSourcePict;
+}
+
+static inline scanFetchProc get_fetch_solid(const int wide)
+{
+    if (wide)
+	return (scanFetchProc)fbFetchSolid64;
+    else
+	return (scanFetchProc)fbFetchSolid;
+}
+
+static inline scanFetchProc get_fetch(const int wide)
+{
+    if (wide)
+	return (scanFetchProc)fbFetch64;
+    else
+	return (scanFetchProc)fbFetch;
+}
+
+static inline scanFetchProc get_fetch_external_alpha(const int wide)
+{
+    if (wide)
+	return (scanFetchProc)ACCESS(fbFetchExternalAlpha64);
+    else
+	return (scanFetchProc)ACCESS(fbFetchExternalAlpha);
+}
+
+static inline scanFetchProc get_fetch_transformed(const int wide)
+{
+    if (wide)
+	return (scanFetchProc)ACCESS(fbFetchTransformed64);
+    else
+	return (scanFetchProc)ACCESS(fbFetchTransformed);
+}
+
+static inline scanStoreProc get_store(const int wide)
+{
+    if (wide)
+	return (scanStoreProc)fbStore64;
+    else
+	return (scanStoreProc)fbStore;
+}
+
+static inline scanStoreProc get_store_external_alpha(const int wide)
+{
+    if (wide)
+	return (scanStoreProc)ACCESS(fbStoreExternalAlpha64);
+    else
+	return (scanStoreProc)ACCESS(fbStoreExternalAlpha);
+}
+
 #ifndef PIXMAN_FB_ACCESSORS
 static
 #endif
 void
 PIXMAN_COMPOSITE_RECT_GENERAL (const FbComposeData *data,
-			       uint32_t *scanline_buffer)
+			       void *src_buffer, void *mask_buffer, 
+			       void *dest_buffer, const int wide)
 {
-    uint32_t *src_buffer = scanline_buffer;
-    uint32_t *dest_buffer = src_buffer + data->width;
     int i;
     scanStoreProc store;
     scanFetchProc fetchSrc = NULL, fetchMask = NULL, fetchDest = NULL;
@@ -191,7 +251,7 @@ PIXMAN_COMPOSITE_RECT_GENERAL (const FbComposeData *data,
         fetchSrc = NULL;
     else if (IS_SOURCE_IMAGE (data->src))
     {
-	fetchSrc = (scanFetchProc)pixmanFetchSourcePict;
+	fetchSrc = get_fetch_source_pict(wide);
 	srcClass = SourcePictureClassify ((source_image_t *)data->src,
 					  data->xSrc, data->ySrc,
 					  data->width, data->height);
@@ -202,23 +262,23 @@ PIXMAN_COMPOSITE_RECT_GENERAL (const FbComposeData *data,
 
 	if (bits->common.alpha_map)
 	{
-	    fetchSrc = (scanFetchProc)FB_FETCH_EXTERNAL_ALPHA;
+	    fetchSrc = get_fetch_external_alpha(wide);
 	}
 	else if ((bits->common.repeat == PIXMAN_REPEAT_NORMAL || bits->common.repeat == PIXMAN_REPEAT_PAD) &&
 		 bits->width == 1 &&
 		 bits->height == 1)
 	{
-	    fetchSrc = (scanFetchProc)fbFetchSolid;
+	    fetchSrc = get_fetch_solid(wide);
 	    srcClass = SOURCE_IMAGE_CLASS_HORIZONTAL;
 	}
 	else if (!bits->common.transform && bits->common.filter != PIXMAN_FILTER_CONVOLUTION
                 && bits->common.repeat != PIXMAN_REPEAT_PAD)
 	{
-	    fetchSrc = (scanFetchProc)fbFetch;
+	    fetchSrc = get_fetch(wide);
 	}
 	else
 	{
-	    fetchSrc = (scanFetchProc)FB_FETCH_TRANSFORMED;
+	    fetchSrc = get_fetch_transformed(wide);
 	}
     }
 
@@ -241,34 +301,34 @@ PIXMAN_COMPOSITE_RECT_GENERAL (const FbComposeData *data,
 
 	    if (bits->common.alpha_map)
 	    {
-		fetchMask = (scanFetchProc)FB_FETCH_EXTERNAL_ALPHA;
+		fetchMask = get_fetch_external_alpha(wide);
 	    }
 	    else if ((bits->common.repeat == PIXMAN_REPEAT_NORMAL || bits->common.repeat == PIXMAN_REPEAT_PAD) &&
 		     bits->width == 1 && bits->height == 1)
 	    {
-		fetchMask = (scanFetchProc)fbFetchSolid;
+		fetchMask = get_fetch_solid(wide);
 		maskClass = SOURCE_IMAGE_CLASS_HORIZONTAL;
 	    }
 	    else if (!bits->common.transform && bits->common.filter != PIXMAN_FILTER_CONVOLUTION
                     && bits->common.repeat != PIXMAN_REPEAT_PAD)
-		fetchMask = (scanFetchProc)fbFetch;
+		fetchMask = get_fetch(wide);
 	    else
-		fetchMask = (scanFetchProc)FB_FETCH_TRANSFORMED;
+		fetchMask = get_fetch_transformed(wide);
 	}
     }
 
     if (data->dest->common.alpha_map)
     {
-	fetchDest = (scanFetchProc)FB_FETCH_EXTERNAL_ALPHA;
-	store = (scanStoreProc)FB_STORE_EXTERNAL_ALPHA;
+	fetchDest = get_fetch_external_alpha(wide);
+	store = get_store_external_alpha(wide);
 
 	if (data->op == PIXMAN_OP_CLEAR || data->op == PIXMAN_OP_SRC)
 	    fetchDest = NULL;
     }
     else
     {
-	fetchDest = (scanFetchProc)fbFetch;
-	store = (scanStoreProc)fbStore;
+	fetchDest = get_fetch(wide);
+	store = get_store(wide);
 
 	switch (data->op)
 	{
@@ -282,7 +342,11 @@ PIXMAN_COMPOSITE_RECT_GENERAL (const FbComposeData *data,
 	    switch (data->dest->bits.format) {
 	    case PIXMAN_a8r8g8b8:
 	    case PIXMAN_x8r8g8b8:
-		store = NULL;
+		// Skip the store step and composite directly into the
+		// destination if the output format of the compose func matches
+		// the destination format.
+		if (!wide)
+		    store = NULL;
 		break;
 	    default:
 		break;
@@ -312,8 +376,9 @@ PIXMAN_COMPOSITE_RECT_GENERAL (const FbComposeData *data,
 	data->mask->common.component_alpha &&
 	PIXMAN_FORMAT_RGB (data->mask->bits.format))
     {
-	uint32_t *mask_buffer = dest_buffer + data->width;
-	CombineFuncC compose = PIXMAN_COMPOSE_FUNCTIONS.combineC[data->op];
+	CombineFuncC32 compose =
+	    wide ? (CombineFuncC32)pixman_composeFunctions64.combineC[data->op] :
+		   pixman_composeFunctions.combineC[data->op];
 	if (!compose)
 	    return;
 
@@ -376,14 +441,14 @@ PIXMAN_COMPOSITE_RECT_GENERAL (const FbComposeData *data,
     }
     else
     {
-	uint32_t *src_mask_buffer = 0, *mask_buffer = 0;
-	CombineFuncU compose = PIXMAN_COMPOSE_FUNCTIONS.combineU[data->op];
+	void *src_mask_buffer = 0;
+	const int useMask = (fetchMask != NULL);
+	CombineFuncU32 compose =
+	    wide ? (CombineFuncU32)pixman_composeFunctions64.combineU[data->op] :
+		   pixman_composeFunctions.combineU[data->op];
 	if (!compose)
 	    return;
 
-	if (fetchMask)
-	    mask_buffer = dest_buffer + data->width;
-
 	for (i = 0; i < data->height; ++i) {
 	    /* fill first half of scanline with source */
 	    if (fetchSrc)
@@ -404,9 +469,13 @@ PIXMAN_COMPOSITE_RECT_GENERAL (const FbComposeData *data,
 		    fetchSrc (data->src, data->xSrc, data->ySrc + i,
 			      data->width, src_buffer, 0, 0);
 
-		    if (mask_buffer)
+		    if (useMask)
 		    {
-			PIXMAN_COMPOSE_FUNCTIONS.combineU[PIXMAN_OP_IN] (mask_buffer, src_buffer, data->width);
+			if (wide)
+			    pixman_composeFunctions64.combineU[PIXMAN_OP_IN] (mask_buffer, src_buffer, data->width);
+			else
+			    pixman_composeFunctions.combineU[PIXMAN_OP_IN] (mask_buffer, src_buffer, data->width);
+
 			src_mask_buffer = mask_buffer;
 		    }
 		    else
@@ -417,13 +486,19 @@ PIXMAN_COMPOSITE_RECT_GENERAL (const FbComposeData *data,
 		else
 		{
 		    fetchSrc (data->src, data->xSrc, data->ySrc + i,
-			      data->width, src_buffer, mask_buffer,
-			      0xff000000);
-
-		    if (mask_buffer)
-			PIXMAN_COMPOSE_FUNCTIONS.combineMaskU (src_buffer,
-							       mask_buffer,
-							       data->width);
+			      data->width, src_buffer,
+			      useMask ? mask_buffer : NULL, 0xff000000);
+
+		    if (useMask) {
+			if (wide)
+			    pixman_composeFunctions64.combineMaskU (src_buffer,
+								    mask_buffer,
+								    data->width);
+			else
+			    pixman_composeFunctions.combineMaskU (src_buffer,
+								  mask_buffer,
+								  data->width);
+		    }
 
 		    src_mask_buffer = src_buffer;
 		}
@@ -433,7 +508,10 @@ PIXMAN_COMPOSITE_RECT_GENERAL (const FbComposeData *data,
 		fetchMask (data->mask, data->xMask, data->yMask + i,
 			   data->width, mask_buffer, 0, 0);
 
-		PIXMAN_COMPOSE_FUNCTIONS.combineU[PIXMAN_OP_IN] (mask_buffer, src_buffer, data->width);
+		if (wide)
+		    pixman_composeFunctions64.combineU[PIXMAN_OP_IN] (mask_buffer, src_buffer, data->width);
+		else
+		    pixman_composeFunctions.combineU[PIXMAN_OP_IN] (mask_buffer, src_buffer, data->width);
 
 		src_mask_buffer = mask_buffer;
 	    }
@@ -465,10 +543,35 @@ PIXMAN_COMPOSITE_RECT_GENERAL (const FbComposeData *data,
 
 #ifndef PIXMAN_FB_ACCESSORS
 
+#define SCANLINE_BUFFER_LENGTH 2048
+
 void
-pixman_composite_rect_general (const FbComposeData *data,
-			       uint32_t *scanline_buffer)
+pixman_composite_rect_general (const FbComposeData *data)
 {
+    uint32_t _scanline_buffer[SCANLINE_BUFFER_LENGTH * 3];
+    const pixman_format_code_t srcFormat = data->src->type == BITS ? data->src->bits.format : 0;
+    const pixman_format_code_t maskFormat = data->mask && data->mask->type == BITS ? data->mask->bits.format : 0;
+    const pixman_format_code_t destFormat = data->dest->type == BITS ? data->dest->bits.format : 0;
+    const int srcWide = PIXMAN_FORMAT_16BPC(srcFormat);
+    const int maskWide = data->mask && PIXMAN_FORMAT_16BPC(maskFormat);
+    const int destWide = PIXMAN_FORMAT_16BPC(destFormat);
+    const int wide = srcWide || maskWide || destWide;
+    const int Bpp = wide ? 8 : 4;
+    uint8_t *scanline_buffer = (uint8_t*)_scanline_buffer;
+    uint8_t *src_buffer, *mask_buffer, *dest_buffer;
+
+    if (data->width * Bpp > SCANLINE_BUFFER_LENGTH * sizeof(uint32_t))
+    {
+	scanline_buffer = pixman_malloc_abc (data->width, 3, Bpp);
+
+	if (!scanline_buffer)
+	    return;
+    }
+
+    src_buffer = scanline_buffer;
+    mask_buffer = src_buffer + data->width * Bpp;
+    dest_buffer = mask_buffer + data->width * Bpp;
+
     if (data->src->common.read_func			||
 	data->src->common.write_func			||
 	(data->mask && data->mask->common.read_func)	||
@@ -476,12 +579,18 @@ pixman_composite_rect_general (const FbComposeData *data,
 	data->dest->common.read_func			||
 	data->dest->common.write_func)
     {
-	pixman_composite_rect_general_accessors (data, scanline_buffer);
+	pixman_composite_rect_general_accessors (data, src_buffer, mask_buffer,
+						 dest_buffer, wide);
     }
     else
     {
-	pixman_composite_rect_general_no_accessors (data, scanline_buffer);
+	pixman_composite_rect_general_no_accessors (data, src_buffer,
+						    mask_buffer, dest_buffer,
+						    wide);
     }
+
+    if ((void*)scanline_buffer != (void*)_scanline_buffer)
+	free (scanline_buffer);
 }
 
 #endif
diff --git a/lib/pixman/pixman/pixman-compute-region.c b/lib/pixman/pixman/pixman-compute-region.c
index f9b9de718..a93cee0ba 100644
--- a/lib/pixman/pixman/pixman-compute-region.c
+++ b/lib/pixman/pixman/pixman-compute-region.c
@@ -21,7 +21,10 @@
  * PERFORMANCE OF THIS SOFTWARE.
  */
 
+#ifdef HAVE_CONFIG_H
 #include <config.h>
+#endif
+
 #include <stdlib.h>
 #include <stdio.h>
 #include "pixman-private.h"
@@ -29,16 +32,16 @@
 #define BOUND(v)	(int16_t) ((v) < INT16_MIN ? INT16_MIN : (v) > INT16_MAX ? INT16_MAX : (v))
 
 static inline pixman_bool_t
-miClipPictureReg (pixman_region16_t *	pRegion,
-		  pixman_region16_t *	pClip,
+miClipPictureReg (pixman_region32_t *	pRegion,
+		  pixman_region32_t *	pClip,
 		  int		dx,
 		  int		dy)
 {
-    if (pixman_region_n_rects(pRegion) == 1 &&
-	pixman_region_n_rects(pClip) == 1)
+    if (pixman_region32_n_rects(pRegion) == 1 &&
+	pixman_region32_n_rects(pClip) == 1)
     {
-	pixman_box16_t *  pRbox = pixman_region_rectangles(pRegion, NULL);
-	pixman_box16_t *  pCbox = pixman_region_rectangles(pClip, NULL);
+	pixman_box32_t *  pRbox = pixman_region32_rectangles(pRegion, NULL);
+	pixman_box32_t *  pCbox = pixman_region32_rectangles(pClip, NULL);
 	int	v;
 	
 	if (pRbox->x1 < (v = pCbox->x1 + dx))
@@ -52,26 +55,26 @@ miClipPictureReg (pixman_region16_t *	pRegion,
 	if (pRbox->x1 >= pRbox->x2 ||
 	    pRbox->y1 >= pRbox->y2)
 	{
-	    pixman_region_init (pRegion);
+	    pixman_region32_init (pRegion);
 	}
     }
-    else if (!pixman_region_not_empty (pClip))
+    else if (!pixman_region32_not_empty (pClip))
 	return FALSE;
     else
     {
 	if (dx || dy)
-	    pixman_region_translate (pRegion, -dx, -dy);
-	if (!pixman_region_intersect (pRegion, pRegion, pClip))
+	    pixman_region32_translate (pRegion, -dx, -dy);
+	if (!pixman_region32_intersect (pRegion, pRegion, pClip))
 	    return FALSE;
 	if (dx || dy)
-	    pixman_region_translate(pRegion, dx, dy);
+	    pixman_region32_translate(pRegion, dx, dy);
     }
-    return pixman_region_not_empty(pRegion);
+    return pixman_region32_not_empty(pRegion);
 }
 
 
 static inline pixman_bool_t
-miClipPictureSrc (pixman_region16_t *	pRegion,
+miClipPictureSrc (pixman_region32_t *	pRegion,
 		  pixman_image_t *	pPicture,
 		  int		dx,
 		  int		dy)
@@ -95,13 +98,13 @@ miClipPictureSrc (pixman_region16_t *	pRegion,
 	 */
 	if (pPicture->common.has_client_clip)
 	{
-	    pixman_region_translate ( pRegion, dx, dy);
+	    pixman_region32_translate ( pRegion, dx, dy);
 	    
-	    if (!pixman_region_intersect (pRegion, pRegion, 
-					  (pixman_region16_t *) pPicture->common.src_clip))
+	    if (!pixman_region32_intersect (pRegion, pRegion, 
+					    pPicture->common.src_clip))
 		return FALSE;
 	    
-	    pixman_region_translate ( pRegion, -dx, -dy);
+	    pixman_region32_translate ( pRegion, -dx, -dy);
 	}
 	    
 	return TRUE;
@@ -121,18 +124,18 @@ miClipPictureSrc (pixman_region16_t *	pRegion,
  */
 
 pixman_bool_t
-pixman_compute_composite_region (pixman_region16_t *	pRegion,
-				 pixman_image_t *	pSrc,
-				 pixman_image_t *	pMask,
-				 pixman_image_t *	pDst,
-				 int16_t		xSrc,
-				 int16_t		ySrc,
-				 int16_t		xMask,
-				 int16_t		yMask,
-				 int16_t		xDst,
-				 int16_t		yDst,
-				 uint16_t	width,
-				 uint16_t	height)
+pixman_compute_composite_region32 (pixman_region32_t *	pRegion,
+				   pixman_image_t *	pSrc,
+				   pixman_image_t *	pMask,
+				   pixman_image_t *	pDst,
+				   int16_t		xSrc,
+				   int16_t		ySrc,
+				   int16_t		xMask,
+				   int16_t		yMask,
+				   int16_t		xDst,
+				   int16_t		yDst,
+				   uint16_t		width,
+				   uint16_t		height)
 {
     int		v;
     
@@ -147,13 +150,13 @@ pixman_compute_composite_region (pixman_region16_t *	pRegion,
     if (pRegion->extents.x1 >= pRegion->extents.x2 ||
 	pRegion->extents.y1 >= pRegion->extents.y2)
     {
-	pixman_region_init (pRegion);
+	pixman_region32_init (pRegion);
 	return FALSE;
     }
     /* clip against dst */
     if (!miClipPictureReg (pRegion, &pDst->common.clip_region, 0, 0))
     {
-	pixman_region_fini (pRegion);
+	pixman_region32_fini (pRegion);
 	return FALSE;
     }
     if (pDst->common.alpha_map)
@@ -162,14 +165,14 @@ pixman_compute_composite_region (pixman_region16_t *	pRegion,
 			       -pDst->common.alpha_origin.x,
 			       -pDst->common.alpha_origin.y))
 	{
-	    pixman_region_fini (pRegion);
+	    pixman_region32_fini (pRegion);
 	    return FALSE;
 	}
     }
     /* clip against src */
     if (!miClipPictureSrc (pRegion, pSrc, xDst - xSrc, yDst - ySrc))
     {
-	pixman_region_fini (pRegion);
+	pixman_region32_fini (pRegion);
 	return FALSE;
     }
     if (pSrc->common.alpha_map)
@@ -178,7 +181,7 @@ pixman_compute_composite_region (pixman_region16_t *	pRegion,
 			       xDst - (xSrc + pSrc->common.alpha_origin.x),
 			       yDst - (ySrc + pSrc->common.alpha_origin.y)))
 	{
-	    pixman_region_fini (pRegion);
+	    pixman_region32_fini (pRegion);
 	    return FALSE;
 	}
     }
@@ -187,7 +190,7 @@ pixman_compute_composite_region (pixman_region16_t *	pRegion,
     {
 	if (!miClipPictureSrc (pRegion, pMask, xDst - xMask, yDst - yMask))
 	{
-	    pixman_region_fini (pRegion);
+	    pixman_region32_fini (pRegion);
 	    return FALSE;
 	}	
 	if (pMask->common.alpha_map)
@@ -196,7 +199,7 @@ pixman_compute_composite_region (pixman_region16_t *	pRegion,
 				   xDst - (xMask + pMask->common.alpha_origin.x),
 				   yDst - (yMask + pMask->common.alpha_origin.y)))
 	    {
-		pixman_region_fini (pRegion);
+		pixman_region32_fini (pRegion);
 		return FALSE;
 	    }
 	}
@@ -204,3 +207,36 @@ pixman_compute_composite_region (pixman_region16_t *	pRegion,
     
     return TRUE;
 }
+
+PIXMAN_EXPORT pixman_bool_t
+pixman_compute_composite_region (pixman_region16_t *	pRegion,
+				 pixman_image_t *	pSrc,
+				 pixman_image_t *	pMask,
+				 pixman_image_t *	pDst,
+				 int16_t		xSrc,
+				 int16_t		ySrc,
+				 int16_t		xMask,
+				 int16_t		yMask,
+				 int16_t		xDst,
+				 int16_t		yDst,
+				 uint16_t	width,
+				 uint16_t	height)
+{
+    pixman_region32_t r32;
+    pixman_bool_t retval;
+
+    pixman_region32_init (&r32);
+    
+    retval = pixman_compute_composite_region32 (&r32, pSrc, pMask, pDst,
+						xSrc, ySrc, xMask, yMask, xDst, yDst,
+						width, height);
+
+    if (retval)
+    {
+	if (!pixman_region16_copy_from_region32 (pRegion, &r32))
+	    retval = FALSE;
+    }
+    
+    pixman_region32_fini (&r32);
+    return retval;
+}
diff --git a/lib/pixman/pixman/pixman-edge.c b/lib/pixman/pixman/pixman-edge.c
index 717284f7d..b9246af5f 100644
--- a/lib/pixman/pixman/pixman-edge.c
+++ b/lib/pixman/pixman/pixman-edge.c
@@ -315,7 +315,7 @@ PIXMAN_RASTERIZE_EDGES (pixman_image_t *image,
 
 #ifndef PIXMAN_FB_ACCESSORS
 
-void
+PIXMAN_EXPORT void
 pixman_rasterize_edges (pixman_image_t *image,
 			pixman_edge_t	*l,
 			pixman_edge_t	*r,
diff --git a/lib/pixman/pixman/pixman-image.c b/lib/pixman/pixman/pixman-image.c
index 9b6846e16..e80c47935 100644
--- a/lib/pixman/pixman/pixman-image.c
+++ b/lib/pixman/pixman/pixman-image.c
@@ -30,6 +30,8 @@
 
 #include "pixman-private.h"
 
+#define Alpha(x) ((x) >> 24)
+
 static void
 init_source_image (source_image_t *image)
 {
@@ -79,8 +81,8 @@ allocate_image (void)
     {
 	image_common_t *common = &image->common;
 
-	pixman_region_init (&common->full_region);
-	pixman_region_init (&common->clip_region);
+	pixman_region32_init (&common->full_region);
+	pixman_region32_init (&common->clip_region);
 	common->src_clip = &common->full_region;
 	common->has_client_clip = FALSE;
 	common->transform = NULL;
@@ -99,7 +101,7 @@ allocate_image (void)
 }
 
 /* Ref Counting */
-pixman_image_t *
+PIXMAN_EXPORT pixman_image_t *
 pixman_image_ref (pixman_image_t *image)
 {
     image->common.ref_count++;
@@ -108,7 +110,7 @@ pixman_image_ref (pixman_image_t *image)
 }
 
 /* returns TRUE when the image is freed */
-pixman_bool_t
+PIXMAN_EXPORT pixman_bool_t
 pixman_image_unref (pixman_image_t *image)
 {
     image_common_t *common = (image_common_t *)image;
@@ -117,8 +119,8 @@ pixman_image_unref (pixman_image_t *image)
 
     if (common->ref_count == 0)
     {
-	pixman_region_fini (&common->clip_region);
-	pixman_region_fini (&common->full_region);
+	pixman_region32_fini (&common->clip_region);
+	pixman_region32_fini (&common->full_region);
 
 	if (common->transform)
 	    free (common->transform);
@@ -156,7 +158,7 @@ pixman_image_unref (pixman_image_t *image)
 }
 
 /* Constructors */
-pixman_image_t *
+PIXMAN_EXPORT pixman_image_t *
 pixman_image_create_solid_fill (pixman_color_t *color)
 {
     pixman_image_t *img = allocate_image();
@@ -171,7 +173,7 @@ pixman_image_create_solid_fill (pixman_color_t *color)
     return img;
 }
 
-pixman_image_t *
+PIXMAN_EXPORT pixman_image_t *
 pixman_image_create_linear_gradient (pixman_point_fixed_t         *p1,
 				     pixman_point_fixed_t         *p2,
 				     const pixman_gradient_stop_t *stops,
@@ -204,7 +206,7 @@ pixman_image_create_linear_gradient (pixman_point_fixed_t         *p1,
 }
 
 
-pixman_image_t *
+PIXMAN_EXPORT pixman_image_t *
 pixman_image_create_radial_gradient (pixman_point_fixed_t         *inner,
 				     pixman_point_fixed_t         *outer,
 				     pixman_fixed_t                inner_radius,
@@ -248,7 +250,7 @@ pixman_image_create_radial_gradient (pixman_point_fixed_t         *inner,
     return image;
 }
 
-pixman_image_t *
+PIXMAN_EXPORT pixman_image_t *
 pixman_image_create_conical_gradient (pixman_point_fixed_t *center,
 				      pixman_fixed_t angle,
 				      const pixman_gradient_stop_t *stops,
@@ -321,20 +323,20 @@ create_bits (pixman_format_code_t format,
 static void
 reset_clip_region (pixman_image_t *image)
 {
-    pixman_region_fini (&image->common.clip_region);
+    pixman_region32_fini (&image->common.clip_region);
 
     if (image->type == BITS)
     {
-	pixman_region_init_rect (&image->common.clip_region, 0, 0,
-				 image->bits.width, image->bits.height);
+	pixman_region32_init_rect (&image->common.clip_region, 0, 0,
+				   image->bits.width, image->bits.height);
     }
     else
     {
-	pixman_region_init (&image->common.clip_region);
+	pixman_region32_init (&image->common.clip_region);
     }
 }
 
-pixman_image_t *
+PIXMAN_EXPORT pixman_image_t *
 pixman_image_create_bits (pixman_format_code_t  format,
 			  int                   width,
 			  int                   height,
@@ -376,15 +378,34 @@ pixman_image_create_bits (pixman_format_code_t  format,
 								  */
     image->bits.indexed = NULL;
 
-    pixman_region_fini (&image->common.full_region);
-    pixman_region_init_rect (&image->common.full_region, 0, 0,
-			     image->bits.width, image->bits.height);
+    pixman_region32_fini (&image->common.full_region);
+    pixman_region32_init_rect (&image->common.full_region, 0, 0,
+			       image->bits.width, image->bits.height);
 
     reset_clip_region (image);
     return image;
 }
 
-pixman_bool_t
+PIXMAN_EXPORT pixman_bool_t
+pixman_image_set_clip_region32 (pixman_image_t *image,
+				pixman_region32_t *region)
+{
+    image_common_t *common = (image_common_t *)image;
+
+    if (region)
+    {
+	return pixman_region32_copy (&common->clip_region, region);
+    }
+    else
+    {
+	reset_clip_region (image);
+
+	return TRUE;
+    }
+}
+
+
+PIXMAN_EXPORT pixman_bool_t
 pixman_image_set_clip_region (pixman_image_t    *image,
 			      pixman_region16_t *region)
 {
@@ -392,7 +413,7 @@ pixman_image_set_clip_region (pixman_image_t    *image,
 
     if (region)
     {
-	return pixman_region_copy (&common->clip_region, region);
+	return pixman_region32_copy_from_region16 (&common->clip_region, region);
     }
     else
     {
@@ -404,14 +425,14 @@ pixman_image_set_clip_region (pixman_image_t    *image,
 
 /* Sets whether the clip region includes a clip region set by the client
  */
-void
+PIXMAN_EXPORT void
 pixman_image_set_has_client_clip (pixman_image_t *image,
 				  pixman_bool_t	  client_clip)
 {
     image->common.has_client_clip = client_clip;
 }
 
-pixman_bool_t
+PIXMAN_EXPORT pixman_bool_t
 pixman_image_set_transform (pixman_image_t           *image,
 			    const pixman_transform_t *transform)
 {
@@ -445,14 +466,14 @@ pixman_image_set_transform (pixman_image_t           *image,
     return TRUE;
 }
 
-void
+PIXMAN_EXPORT void
 pixman_image_set_repeat (pixman_image_t  *image,
 			 pixman_repeat_t  repeat)
 {
     image->common.repeat = repeat;
 }
 
-pixman_bool_t
+PIXMAN_EXPORT pixman_bool_t
 pixman_image_set_filter (pixman_image_t       *image,
 			 pixman_filter_t       filter,
 			 const pixman_fixed_t *params,
@@ -485,7 +506,7 @@ pixman_image_set_filter (pixman_image_t       *image,
     return TRUE;
 }
 
-void
+PIXMAN_EXPORT void
 pixman_image_set_source_clipping (pixman_image_t  *image,
 				  pixman_bool_t    source_clipping)
 {
@@ -501,7 +522,7 @@ pixman_image_set_source_clipping (pixman_image_t  *image,
  * copy the content of indexed. Doing this copying is simply
  * way, way too expensive.
  */
-void
+PIXMAN_EXPORT void
 pixman_image_set_indexed (pixman_image_t	 *image,
 			  const pixman_indexed_t *indexed)
 {
@@ -510,7 +531,7 @@ pixman_image_set_indexed (pixman_image_t	 *image,
     bits->indexed = indexed;
 }
 
-void
+PIXMAN_EXPORT void
 pixman_image_set_alpha_map (pixman_image_t *image,
 			    pixman_image_t *alpha_map,
 			    int16_t         x,
@@ -535,7 +556,7 @@ pixman_image_set_alpha_map (pixman_image_t *image,
     common->alpha_origin.y = y;
 }
 
-void
+PIXMAN_EXPORT void
 pixman_image_set_component_alpha   (pixman_image_t       *image,
 				    pixman_bool_t         component_alpha)
 {
@@ -543,7 +564,7 @@ pixman_image_set_component_alpha   (pixman_image_t       *image,
 }
 
 
-void
+PIXMAN_EXPORT void
 pixman_image_set_accessors (pixman_image_t             *image,
 			    pixman_read_memory_func_t	read_func,
 			    pixman_write_memory_func_t	write_func)
@@ -554,7 +575,7 @@ pixman_image_set_accessors (pixman_image_t             *image,
     image->common.write_func = write_func;
 }
 
-uint32_t *
+PIXMAN_EXPORT uint32_t *
 pixman_image_get_data (pixman_image_t *image)
 {
     if (image->type == BITS)
@@ -563,7 +584,7 @@ pixman_image_get_data (pixman_image_t *image)
     return NULL;
 }
 
-int
+PIXMAN_EXPORT int
 pixman_image_get_width (pixman_image_t *image)
 {
     if (image->type == BITS)
@@ -572,7 +593,7 @@ pixman_image_get_width (pixman_image_t *image)
     return 0;
 }
 
-int
+PIXMAN_EXPORT int
 pixman_image_get_height (pixman_image_t *image)
 {
     if (image->type == BITS)
@@ -581,7 +602,7 @@ pixman_image_get_height (pixman_image_t *image)
     return 0;
 }
 
-int
+PIXMAN_EXPORT int
 pixman_image_get_stride (pixman_image_t *image)
 {
     if (image->type == BITS)
@@ -590,7 +611,7 @@ pixman_image_get_stride (pixman_image_t *image)
     return 0;
 }
 
-int
+PIXMAN_EXPORT int
 pixman_image_get_depth (pixman_image_t *image)
 {
     if (image->type == BITS)
@@ -599,7 +620,7 @@ pixman_image_get_depth (pixman_image_t *image)
     return 0;
 }
 
-pixman_bool_t
+static pixman_bool_t
 color_to_pixel (pixman_color_t *color,
 		uint32_t       *pixel,
 		pixman_format_code_t format)
@@ -640,7 +661,7 @@ color_to_pixel (pixman_color_t *color,
     return TRUE;
 }
 
-pixman_bool_t
+PIXMAN_EXPORT pixman_bool_t
 pixman_image_fill_rectangles (pixman_op_t		    op,
 			      pixman_image_t		   *dest,
 			      pixman_color_t		   *color,
@@ -677,23 +698,23 @@ pixman_image_fill_rectangles (pixman_op_t		    op,
 	{
 	    for (i = 0; i < n_rects; ++i)
 	    {
-		pixman_region16_t fill_region;
+		pixman_region32_t fill_region;
 		int n_boxes, j;
-		pixman_box16_t *boxes;
+		pixman_box32_t *boxes;
 
-		pixman_region_init_rect (&fill_region, rects[i].x, rects[i].y, rects[i].width, rects[i].height);
-		pixman_region_intersect (&fill_region, &fill_region, &dest->common.clip_region);
+		pixman_region32_init_rect (&fill_region, rects[i].x, rects[i].y, rects[i].width, rects[i].height);
+		pixman_region32_intersect (&fill_region, &fill_region, &dest->common.clip_region);
 
-		boxes = pixman_region_rectangles (&fill_region, &n_boxes);
+		boxes = pixman_region32_rectangles (&fill_region, &n_boxes);
 		for (j = 0; j < n_boxes; ++j)
 		{
-		    const pixman_box16_t *box = &(boxes[j]);
+		    const pixman_box32_t *box = &(boxes[j]);
 		    pixman_fill (dest->bits.bits, dest->bits.rowstride, PIXMAN_FORMAT_BPP (dest->bits.format),
 				 box->x1, box->y1, box->x2 - box->x1, box->y2 - box->y1,
 				 pixel);
 		}
 
-		pixman_region_fini (&fill_region);
+		pixman_region32_fini (&fill_region);
 	    }
 	    return TRUE;
 	}
@@ -717,3 +738,91 @@ pixman_image_fill_rectangles (pixman_op_t		    op,
 
     return TRUE;
 }
+
+pixman_bool_t
+pixman_image_can_get_solid (pixman_image_t *image)
+{
+    if (image->type == SOLID)
+	return TRUE;
+
+    if (image->type != BITS	||
+	image->bits.width != 1	||
+	image->bits.height != 1)
+    {
+	return FALSE;
+    }
+
+    if (image->common.repeat != PIXMAN_REPEAT_NORMAL)
+	return FALSE;
+
+    switch (image->bits.format)
+    {
+    case PIXMAN_a8r8g8b8:
+    case PIXMAN_x8r8g8b8:
+    case PIXMAN_a8b8g8r8:
+    case PIXMAN_x8b8g8r8:
+    case PIXMAN_r8g8b8:
+    case PIXMAN_b8g8r8:
+    case PIXMAN_r5g6b5:
+    case PIXMAN_b5g6r5:
+	return TRUE;
+    default:
+	return FALSE;
+    }
+}
+
+pixman_bool_t
+pixman_image_is_opaque(pixman_image_t *image)
+{
+    int i = 0;
+    int gradientNumberOfColors = 0;
+
+    if(image->common.alpha_map)
+        return FALSE;
+
+    switch(image->type)
+    {
+    case BITS:
+        if(PIXMAN_FORMAT_A(image->bits.format))
+            return FALSE;
+        break;
+
+    case LINEAR:
+    case CONICAL:
+    case RADIAL:
+        gradientNumberOfColors = image->gradient.n_stops;
+        i=0;
+        while(i<gradientNumberOfColors)
+        {
+            if(image->gradient.stops[i].color.alpha != 0xffff)
+                return FALSE;
+            i++;
+        }
+        break;
+
+    case SOLID:
+         if(Alpha(image->solid.color) != 0xff)
+            return FALSE;
+        break;
+    }
+
+    /* Convolution filters can introduce translucency if the sum of the weights
+       is lower than 1. */
+    if (image->common.filter == PIXMAN_FILTER_CONVOLUTION)
+         return FALSE;
+
+    if (image->common.repeat == PIXMAN_REPEAT_NONE)
+    {
+        if (image->common.filter != PIXMAN_FILTER_NEAREST)
+            return FALSE;
+
+        if (image->common.transform)
+            return FALSE;
+
+	/* Gradients do not necessarily cover the entire compositing area */
+	if (image->type == LINEAR || image->type == CONICAL || image->type == RADIAL)
+	    return FALSE;
+    }
+
+     return TRUE;
+}
diff --git a/lib/pixman/pixman/pixman-mmx.c b/lib/pixman/pixman/pixman-mmx.c
index 8afbcb8eb..a856fe4ef 100644
--- a/lib/pixman/pixman/pixman-mmx.c
+++ b/lib/pixman/pixman/pixman-mmx.c
@@ -36,9 +36,6 @@
 #ifdef USE_MMX
 
 #include <mmintrin.h>
-#ifdef USE_SSE
-#include <xmmintrin.h> /* for _mm_shuffle_pi16 and _MM_SHUFFLE */
-#endif
 
 #include "pixman-mmx.h"
 
@@ -142,28 +139,33 @@ static const MMXData c =
 #endif
 };
 
-#ifdef _MSC_VER
-#undef inline
-#define inline __forceinline
-#endif
-
 #ifdef __GNUC__
-#define MC(x) ((__m64) c.mmx_##x)
+#    ifdef __ICC
+#        define MC(x)  M64(c.mmx_##x)
+#    else
+#        define MC(x) ((__m64)c.mmx_##x)
+#    endif
+#    define inline __inline__ __attribute__ ((__always_inline__))
 #endif
+
 #ifdef _MSC_VER
-#define MC(x) c.mmx_##x
+#    define MC(x) c.mmx_##x
+#    undef inline
+#    define inline __forceinline
 #endif
 
 static inline __m64
 M64 (ullong x)
 {
-#ifdef __GNUC__
+#ifdef __ICC
+    return _mm_cvtsi64_m64 (x);
+#elif defined (__GNUC__)
     return (__m64)x;
 #endif
 
 #ifdef _MSC_VER
     __m64 res;
-    
+
     res.m64_u64 = x;
     return res;
 #endif
@@ -172,7 +174,9 @@ M64 (ullong x)
 static inline ullong
 ULLONG (__m64 x)
 {
-#ifdef __GNUC__
+#ifdef __ICC
+    return _mm_cvtm64_si64 (x);
+#elif defined (__GNUC__)
     return (ullong)x;
 #endif
 
@@ -220,28 +224,6 @@ pix_add (__m64 a, __m64 b)
     return  _mm_adds_pu8 (a, b);
 }
 
-#ifdef USE_SSE
-
-static inline __m64
-expand_alpha (__m64 pixel)
-{
-    return _mm_shuffle_pi16 (pixel, _MM_SHUFFLE(3, 3, 3, 3));
-}
-
-static inline __m64
-expand_alpha_rev (__m64 pixel)
-{
-    return _mm_shuffle_pi16 (pixel, _MM_SHUFFLE(0, 0, 0, 0));
-}
-
-static inline __m64
-invert_colors (__m64 pixel)
-{
-    return _mm_shuffle_pi16 (pixel, _MM_SHUFFLE(3, 0, 1, 2));
-}
-
-#else
-
 static inline __m64
 expand_alpha (__m64 pixel)
 {
@@ -293,8 +275,6 @@ invert_colors (__m64 pixel)
     return x;
 }
 
-#endif
-
 static inline __m64
 over (__m64 src, __m64 srca, __m64 dest)
 {
@@ -400,6 +380,12 @@ expand8888 (__m64 in, int pos)
 }
 
 static inline __m64
+expandx888 (__m64 in, int pos)
+{
+    return _mm_or_si64 (expand8888 (in, pos), MC(full_alpha));
+}
+
+static inline __m64
 pack565 (__m64 pixel, __m64 target, int pos)
 {
     __m64 p = pixel;
@@ -1379,36 +1365,36 @@ fbCompositeSrc_x888xnx8888mmx (pixman_op_t op,
 	    __m64 vs7 = *(__m64 *)(src + 14);
 
 	    vd0 = pack8888 (
-		in_over (expand8888 (vs0, 0), srca, vmask, expand8888 (vd0, 0)),
-		in_over (expand8888 (vs0, 1), srca, vmask, expand8888 (vd0, 1)));
+		in_over (expandx888 (vs0, 0), srca, vmask, expand8888 (vd0, 0)),
+		in_over (expandx888 (vs0, 1), srca, vmask, expand8888 (vd0, 1)));
 
 	    vd1 = pack8888 (
-		in_over (expand8888 (vs1, 0), srca, vmask, expand8888 (vd1, 0)),
-		in_over (expand8888 (vs1, 1), srca, vmask, expand8888 (vd1, 1)));
+		in_over (expandx888 (vs1, 0), srca, vmask, expand8888 (vd1, 0)),
+		in_over (expandx888 (vs1, 1), srca, vmask, expand8888 (vd1, 1)));
 
 	    vd2 = pack8888 (
-		in_over (expand8888 (vs2, 0), srca, vmask, expand8888 (vd2, 0)),
-		in_over (expand8888 (vs2, 1), srca, vmask, expand8888 (vd2, 1)));
+		in_over (expandx888 (vs2, 0), srca, vmask, expand8888 (vd2, 0)),
+		in_over (expandx888 (vs2, 1), srca, vmask, expand8888 (vd2, 1)));
 
 	    vd3 = pack8888 (
-		in_over (expand8888 (vs3, 0), srca, vmask, expand8888 (vd3, 0)),
-		in_over (expand8888 (vs3, 1), srca, vmask, expand8888 (vd3, 1)));
+		in_over (expandx888 (vs3, 0), srca, vmask, expand8888 (vd3, 0)),
+		in_over (expandx888 (vs3, 1), srca, vmask, expand8888 (vd3, 1)));
 
 	    vd4 = pack8888 (
-		in_over (expand8888 (vs4, 0), srca, vmask, expand8888 (vd4, 0)),
-		in_over (expand8888 (vs4, 1), srca, vmask, expand8888 (vd4, 1)));
+		in_over (expandx888 (vs4, 0), srca, vmask, expand8888 (vd4, 0)),
+		in_over (expandx888 (vs4, 1), srca, vmask, expand8888 (vd4, 1)));
 
 	    vd5 = pack8888 (
-		in_over (expand8888 (vs5, 0), srca, vmask, expand8888 (vd5, 0)),
-		in_over (expand8888 (vs5, 1), srca, vmask, expand8888 (vd5, 1)));
+		in_over (expandx888 (vs5, 0), srca, vmask, expand8888 (vd5, 0)),
+		in_over (expandx888 (vs5, 1), srca, vmask, expand8888 (vd5, 1)));
 
             vd6 = pack8888 (
-		in_over (expand8888 (vs6, 0), srca, vmask, expand8888 (vd6, 0)),
-		in_over (expand8888 (vs6, 1), srca, vmask, expand8888 (vd6, 1)));
+		in_over (expandx888 (vs6, 0), srca, vmask, expand8888 (vd6, 0)),
+		in_over (expandx888 (vs6, 1), srca, vmask, expand8888 (vd6, 1)));
 
 	    vd7 = pack8888 (
-		in_over (expand8888 (vs7, 0), srca, vmask, expand8888 (vd7, 0)),
-		in_over (expand8888 (vs7, 1), srca, vmask, expand8888 (vd7, 1)));
+		in_over (expandx888 (vs7, 0), srca, vmask, expand8888 (vd7, 0)),
+		in_over (expandx888 (vs7, 1), srca, vmask, expand8888 (vd7, 1)));
 
 	    *(__m64 *)(dst + 0) = vd0;
 	    *(__m64 *)(dst + 2) = vd1;
@@ -1726,13 +1712,27 @@ pixman_fill_mmx (uint32_t *bits,
     __m64	v1, v2, v3, v4, v5, v6, v7;
 #endif
 
-    if (bpp == 16 && (xor >> 16 != (xor & 0xffff)))
+    if (bpp != 16 && bpp != 32 && bpp != 8)
 	return FALSE;
 
-    if (bpp != 16 && bpp != 32)
+    if (bpp == 16 && (xor >> 16 != (xor & 0xffff)))
 	return FALSE;
 
-    if (bpp == 16)
+    if (bpp == 8 &&
+	((xor >> 16 != (xor & 0xffff)) ||
+	 (xor >> 24 != (xor & 0x00ff) >> 16)))
+    {
+	return FALSE;
+    }
+    
+    if (bpp == 8)
+    {
+	stride = stride * (int) sizeof (uint32_t) / 1;
+	byte_line = (uint8_t *)(((uint8_t *)bits) + stride * y + x);
+	byte_width = width;
+	stride *= 1;
+    }
+    else if (bpp == 16)
     {
 	stride = stride * (int) sizeof (uint32_t) / 2;
 	byte_line = (uint8_t *)(((uint16_t *)bits) + stride * y + x);
@@ -1771,6 +1771,13 @@ pixman_fill_mmx (uint32_t *bits,
 	byte_line += stride;
 	w = byte_width;
 
+	while (w >= 1 && ((unsigned long)d & 1))
+	{
+	    *(uint8_t *)d = (xor & 0xff);
+	    w--;
+	    d++;
+	}
+	
 	while (w >= 2 && ((unsigned long)d & 3))
 	{
 	    *(uint16_t *)d = xor;
@@ -1824,12 +1831,19 @@ pixman_fill_mmx (uint32_t *bits,
 	    w -= 4;
 	    d += 4;
 	}
-	if (w >= 2)
+	while (w >= 2)
 	{
 	    *(uint16_t *)d = xor;
 	    w -= 2;
 	    d += 2;
 	}
+	while (w >= 1)
+	{
+	    *(uint8_t *)d = (xor & 0xff);
+	    w--;
+	    d++;
+	}
+	
     }
 
     _mm_empty();
diff --git a/lib/pixman/pixman/pixman-pict.c b/lib/pixman/pixman/pixman-pict.c
index 8976ff209..193ea287c 100644
--- a/lib/pixman/pixman/pixman-pict.c
+++ b/lib/pixman/pixman/pixman-pict.c
@@ -30,10 +30,15 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
-
 #include "pixman-private.h"
 #include "pixman-mmx.h"
-#include "pixman-sse.h"
+#include "pixman-vmx.h"
+#include "pixman-sse2.h"
+#include "pixman-combine32.h"
+
+#ifdef __GNUC__
+#   define inline __inline__ __attribute__ ((__always_inline__))
+#endif
 
 #define FbFullMask(n)   ((n) == 32 ? (uint32_t)-1 : ((((uint32_t) 1) << n) - 1))
 
@@ -47,21 +52,17 @@ typedef void (* CompositeFunc) (pixman_op_t,
 				int16_t, int16_t, int16_t, int16_t, int16_t, int16_t,
 				uint16_t, uint16_t);
 
-uint32_t
-fbOver (uint32_t x, uint32_t y)
+static inline uint32_t
+fbOver (uint32_t src, uint32_t dest)
 {
-    uint16_t  a = ~x >> 24;
-    uint16_t  t;
-    uint32_t  m,n,o,p;
+    // dest = (dest * (255 - alpha)) / 255 + src
+    uint32_t a = ~src >> 24; // 255 - alpha == 255 + (~alpha + 1) == ~alpha
+    FbByteMulAdd(dest, a, src);
 
-    m = FbOverU(x,y,0,a,t);
-    n = FbOverU(x,y,8,a,t);
-    o = FbOverU(x,y,16,a,t);
-    p = FbOverU(x,y,24,a,t);
-    return m|n|o|p;
+    return dest;
 }
 
-uint32_t
+static uint32_t
 fbOver24 (uint32_t x, uint32_t y)
 {
     uint16_t  a = ~x >> 24;
@@ -74,7 +75,7 @@ fbOver24 (uint32_t x, uint32_t y)
     return m|n|o;
 }
 
-uint32_t
+static uint32_t
 fbIn (uint32_t x, uint8_t y)
 {
     uint16_t  a = y;
@@ -1107,7 +1108,7 @@ pixman_image_composite_rect  (pixman_op_t                   op,
 			      int16_t                       dest_y,
 			      uint16_t                      width,
 			      uint16_t                      height);
-void
+static void
 fbCompositeSolidFill (pixman_op_t op,
 		      pixman_image_t * pSrc,
 		      pixman_image_t * pMask,
@@ -1187,22 +1188,22 @@ pixman_walk_composite_region (pixman_op_t op,
 			      CompositeFunc compositeRect)
 {
     int		    n;
-    const pixman_box16_t *pbox;
+    const pixman_box32_t *pbox;
     int		    w, h, w_this, h_this;
     int		    x_msk, y_msk, x_src, y_src, x_dst, y_dst;
-    pixman_region16_t reg;
-    pixman_region16_t *region;
+    pixman_region32_t reg;
+    pixman_region32_t *region;
 
-    pixman_region_init (&reg);
-    if (!pixman_compute_composite_region (&reg, pSrc, pMask, pDst,
-					  xSrc, ySrc, xMask, yMask, xDst, yDst, width, height))
+    pixman_region32_init (&reg);
+    if (!pixman_compute_composite_region32 (&reg, pSrc, pMask, pDst,
+					    xSrc, ySrc, xMask, yMask, xDst, yDst, width, height))
     {
 	return;
     }
 
     region = &reg;
 
-    pbox = pixman_region_rectangles (region, &n);
+    pbox = pixman_region32_rectangles (region, &n);
     while (n--)
     {
 	h = pbox->y2 - pbox->y1;
@@ -1258,43 +1259,9 @@ pixman_walk_composite_region (pixman_op_t op,
 	}
 	pbox++;
     }
-    pixman_region_fini (&reg);
+    pixman_region32_fini (&reg);
 }
 
-static pixman_bool_t
-can_get_solid (pixman_image_t *image)
-{
-    if (image->type == SOLID)
-	return TRUE;
-
-    if (image->type != BITS	||
-	image->bits.width != 1	||
-	image->bits.height != 1)
-    {
-	return FALSE;
-    }
-
-    if (image->common.repeat != PIXMAN_REPEAT_NORMAL)
-	return FALSE;
-
-    switch (image->bits.format)
-    {
-    case PIXMAN_a8r8g8b8:
-    case PIXMAN_x8r8g8b8:
-    case PIXMAN_a8b8g8r8:
-    case PIXMAN_x8b8g8r8:
-    case PIXMAN_r8g8b8:
-    case PIXMAN_b8g8r8:
-    case PIXMAN_r5g6b5:
-    case PIXMAN_b5g6r5:
-	return TRUE;
-    default:
-	return FALSE;
-    }
-}
-
-#define SCANLINE_BUFFER_LENGTH 2048
-
 static void
 pixman_image_composite_rect  (pixman_op_t                   op,
 			      pixman_image_t               *src,
@@ -1310,20 +1277,10 @@ pixman_image_composite_rect  (pixman_op_t                   op,
 			      uint16_t                      height)
 {
     FbComposeData compose_data;
-    uint32_t _scanline_buffer[SCANLINE_BUFFER_LENGTH * 3];
-    uint32_t *scanline_buffer = _scanline_buffer;
 
     return_if_fail (src != NULL);
     return_if_fail (dest != NULL);
 
-    if (width > SCANLINE_BUFFER_LENGTH)
-    {
-	scanline_buffer = (uint32_t *)pixman_malloc_abc (width, 3, sizeof (uint32_t));
-
-	if (!scanline_buffer)
-	    return;
-    }
-
     compose_data.op = op;
     compose_data.src = src;
     compose_data.mask = mask;
@@ -1337,10 +1294,7 @@ pixman_image_composite_rect  (pixman_op_t                   op,
     compose_data.width = width;
     compose_data.height = height;
 
-    pixman_composite_rect_general (&compose_data, scanline_buffer);
-
-    if (scanline_buffer != _scanline_buffer)
-	free (scanline_buffer);
+    pixman_composite_rect_general (&compose_data);
 }
 
 /* These "formats" both have depth 0, so they
@@ -1428,6 +1382,8 @@ static const FastPathInfo mmx_fast_paths[] =
 
     { PIXMAN_OP_SRC, PIXMAN_a8r8g8b8,  PIXMAN_null,	PIXMAN_a8r8g8b8, fbCompositeCopyAreammx, 0 },
     { PIXMAN_OP_SRC, PIXMAN_a8b8g8r8,  PIXMAN_null,	PIXMAN_a8b8g8r8, fbCompositeCopyAreammx, 0 },
+    { PIXMAN_OP_SRC, PIXMAN_a8r8g8b8,  PIXMAN_null,	PIXMAN_x8r8g8b8, fbCompositeCopyAreammx, 0 },
+    { PIXMAN_OP_SRC, PIXMAN_a8b8g8r8,  PIXMAN_null,	PIXMAN_x8b8g8r8, fbCompositeCopyAreammx, 0 },
     { PIXMAN_OP_SRC, PIXMAN_x8r8g8b8,  PIXMAN_null,	PIXMAN_x8r8g8b8, fbCompositeCopyAreammx, 0 },
     { PIXMAN_OP_SRC, PIXMAN_x8b8g8r8,  PIXMAN_null,	PIXMAN_x8b8g8r8, fbCompositeCopyAreammx, 0 },
     { PIXMAN_OP_SRC, PIXMAN_r5g6b5,    PIXMAN_null,     PIXMAN_r5g6b5,   fbCompositeCopyAreammx, 0 },
@@ -1439,12 +1395,91 @@ static const FastPathInfo mmx_fast_paths[] =
 #endif
 
 #ifdef USE_SSE2
-static const FastPathInfo sse_fast_paths[] =
+static const FastPathInfo sse2_fast_paths[] =
+{
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8,       PIXMAN_r5g6b5,   fbCompositeSolidMask_nx8x0565sse2,     0 },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8,       PIXMAN_b5g6r5,   fbCompositeSolidMask_nx8x0565sse2,     0 },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_null,     PIXMAN_a8r8g8b8, fbCompositeSolid_nx8888sse2,           0 },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_null,     PIXMAN_x8r8g8b8, fbCompositeSolid_nx8888sse2,           0 },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_null,     PIXMAN_r5g6b5,   fbCompositeSolid_nx0565sse2,           0 },
+    { PIXMAN_OP_OVER, PIXMAN_a8r8g8b8, PIXMAN_null,     PIXMAN_a8r8g8b8, fbCompositeSrc_8888x8888sse2,          0 },
+    { PIXMAN_OP_OVER, PIXMAN_a8r8g8b8, PIXMAN_null,     PIXMAN_x8r8g8b8, fbCompositeSrc_8888x8888sse2,          0 },
+    { PIXMAN_OP_OVER, PIXMAN_a8b8g8r8, PIXMAN_null,     PIXMAN_a8b8g8r8, fbCompositeSrc_8888x8888sse2,          0 },
+    { PIXMAN_OP_OVER, PIXMAN_a8b8g8r8, PIXMAN_null,     PIXMAN_x8b8g8r8, fbCompositeSrc_8888x8888sse2,          0 },
+    { PIXMAN_OP_OVER, PIXMAN_a8r8g8b8, PIXMAN_null,     PIXMAN_r5g6b5,   fbCompositeSrc_8888x0565sse2,          0 },
+    { PIXMAN_OP_OVER, PIXMAN_a8b8g8r8, PIXMAN_null,     PIXMAN_b5g6r5,   fbCompositeSrc_8888x0565sse2,          0 },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8,       PIXMAN_a8r8g8b8, fbCompositeSolidMask_nx8x8888sse2,     0 },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8,       PIXMAN_x8r8g8b8, fbCompositeSolidMask_nx8x8888sse2,     0 },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8,       PIXMAN_a8b8g8r8, fbCompositeSolidMask_nx8x8888sse2,     0 },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8,       PIXMAN_x8b8g8r8, fbCompositeSolidMask_nx8x8888sse2,     0 },
+#if 0
+    /* FIXME: This code are buggy in MMX version, now the bug was translated to SSE2 version */
+    { PIXMAN_OP_OVER, PIXMAN_x8r8g8b8, PIXMAN_a8,       PIXMAN_x8r8g8b8, fbCompositeOver_x888x8x8888sse2,       0 },
+    { PIXMAN_OP_OVER, PIXMAN_x8r8g8b8, PIXMAN_a8,       PIXMAN_a8r8g8b8, fbCompositeOver_x888x8x8888sse2,       0 },
+    { PIXMAN_OP_OVER, PIXMAN_x8b8g8r8, PIXMAN_a8,       PIXMAN_x8b8g8r8, fbCompositeOver_x888x8x8888sse2,       0 },
+    { PIXMAN_OP_OVER, PIXMAN_x8b8g8r8, PIXMAN_a8,       PIXMAN_a8r8g8b8, fbCompositeOver_x888x8x8888sse2,       0 },
+#endif
+    { PIXMAN_OP_OVER, PIXMAN_x8r8g8b8, PIXMAN_a8,       PIXMAN_a8r8g8b8, fbCompositeSrc_x888xnx8888sse2,        NEED_SOLID_MASK },
+    { PIXMAN_OP_OVER, PIXMAN_x8r8g8b8, PIXMAN_a8,       PIXMAN_x8r8g8b8, fbCompositeSrc_x888xnx8888sse2,        NEED_SOLID_MASK },
+    { PIXMAN_OP_OVER, PIXMAN_x8b8g8r8, PIXMAN_a8,       PIXMAN_a8b8g8r8, fbCompositeSrc_x888xnx8888sse2,        NEED_SOLID_MASK },
+    { PIXMAN_OP_OVER, PIXMAN_x8b8g8r8, PIXMAN_a8,       PIXMAN_x8b8g8r8, fbCompositeSrc_x888xnx8888sse2,        NEED_SOLID_MASK },
+    { PIXMAN_OP_OVER, PIXMAN_a8r8g8b8, PIXMAN_a8,       PIXMAN_a8r8g8b8, fbCompositeSrc_8888x8x8888sse2,        NEED_SOLID_MASK },
+    { PIXMAN_OP_OVER, PIXMAN_a8r8g8b8, PIXMAN_a8,       PIXMAN_x8r8g8b8, fbCompositeSrc_8888x8x8888sse2,        NEED_SOLID_MASK },
+    { PIXMAN_OP_OVER, PIXMAN_a8b8g8r8, PIXMAN_a8,       PIXMAN_a8b8g8r8, fbCompositeSrc_8888x8x8888sse2,        NEED_SOLID_MASK },
+    { PIXMAN_OP_OVER, PIXMAN_a8b8g8r8, PIXMAN_a8,       PIXMAN_x8b8g8r8, fbCompositeSrc_8888x8x8888sse2,        NEED_SOLID_MASK },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8r8g8b8, PIXMAN_a8r8g8b8, fbCompositeSolidMask_nx8888x8888Csse2, NEED_COMPONENT_ALPHA },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8r8g8b8, PIXMAN_x8r8g8b8, fbCompositeSolidMask_nx8888x8888Csse2, NEED_COMPONENT_ALPHA },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8b8g8r8, PIXMAN_a8b8g8r8, fbCompositeSolidMask_nx8888x8888Csse2, NEED_COMPONENT_ALPHA },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8b8g8r8, PIXMAN_x8b8g8r8, fbCompositeSolidMask_nx8888x8888Csse2, NEED_COMPONENT_ALPHA },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8r8g8b8, PIXMAN_r5g6b5,   fbCompositeSolidMask_nx8888x0565Csse2, NEED_COMPONENT_ALPHA },
+    { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8b8g8r8, PIXMAN_b5g6r5,   fbCompositeSolidMask_nx8888x0565Csse2, NEED_COMPONENT_ALPHA },
+    { PIXMAN_OP_OVER, PIXMAN_x8b8g8r8, PIXMAN_a8r8g8b8, PIXMAN_a8r8g8b8, fbCompositeSrc_8888RevNPx8888sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8b8g8r8, PIXMAN_a8b8g8r8, PIXMAN_a8r8g8b8, fbCompositeSrc_8888RevNPx8888sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8b8g8r8, PIXMAN_a8r8g8b8, PIXMAN_x8r8g8b8, fbCompositeSrc_8888RevNPx8888sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8b8g8r8, PIXMAN_a8b8g8r8, PIXMAN_x8r8g8b8, fbCompositeSrc_8888RevNPx8888sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8r8g8b8, PIXMAN_a8r8g8b8, PIXMAN_a8b8g8r8, fbCompositeSrc_8888RevNPx8888sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8r8g8b8, PIXMAN_a8b8g8r8, PIXMAN_a8b8g8r8, fbCompositeSrc_8888RevNPx8888sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8r8g8b8, PIXMAN_a8r8g8b8, PIXMAN_x8b8g8r8, fbCompositeSrc_8888RevNPx8888sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8r8g8b8, PIXMAN_a8b8g8r8, PIXMAN_x8b8g8r8, fbCompositeSrc_8888RevNPx8888sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8b8g8r8, PIXMAN_a8r8g8b8, PIXMAN_r5g6b5,   fbCompositeSrc_8888RevNPx0565sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8b8g8r8, PIXMAN_a8b8g8r8, PIXMAN_r5g6b5,   fbCompositeSrc_8888RevNPx0565sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8r8g8b8, PIXMAN_a8r8g8b8, PIXMAN_b5g6r5,   fbCompositeSrc_8888RevNPx0565sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8r8g8b8, PIXMAN_a8b8g8r8, PIXMAN_b5g6r5,   fbCompositeSrc_8888RevNPx0565sse2,     NEED_PIXBUF },
+    { PIXMAN_OP_OVER, PIXMAN_x8r8g8b8, PIXMAN_null,     PIXMAN_x8r8g8b8, fbCompositeCopyAreasse2,               0 },
+    { PIXMAN_OP_OVER, PIXMAN_x8b8g8r8, PIXMAN_null,     PIXMAN_x8b8g8r8, fbCompositeCopyAreasse2,               0 },
+
+    { PIXMAN_OP_ADD,  PIXMAN_a8,       PIXMAN_null,     PIXMAN_a8,       fbCompositeSrcAdd_8000x8000sse2,       0 },
+    { PIXMAN_OP_ADD,  PIXMAN_a8r8g8b8, PIXMAN_null,     PIXMAN_a8r8g8b8, fbCompositeSrcAdd_8888x8888sse2,       0 },
+    { PIXMAN_OP_ADD,  PIXMAN_a8b8g8r8, PIXMAN_null,     PIXMAN_a8b8g8r8, fbCompositeSrcAdd_8888x8888sse2,       0 },
+    { PIXMAN_OP_ADD,  PIXMAN_solid,    PIXMAN_a8,       PIXMAN_a8,       fbCompositeSrcAdd_8888x8x8sse2,        0 },
+
+    { PIXMAN_OP_SRC, PIXMAN_solid,     PIXMAN_a8,       PIXMAN_a8r8g8b8, fbCompositeSolidMaskSrc_nx8x8888sse2,  0 },
+    { PIXMAN_OP_SRC, PIXMAN_solid,     PIXMAN_a8,       PIXMAN_x8r8g8b8, fbCompositeSolidMaskSrc_nx8x8888sse2,  0 },
+    { PIXMAN_OP_SRC, PIXMAN_solid,     PIXMAN_a8,       PIXMAN_a8b8g8r8, fbCompositeSolidMaskSrc_nx8x8888sse2,  0 },
+    { PIXMAN_OP_SRC, PIXMAN_solid,     PIXMAN_a8,       PIXMAN_x8b8g8r8, fbCompositeSolidMaskSrc_nx8x8888sse2,  0 },
+
+    { PIXMAN_OP_SRC, PIXMAN_a8r8g8b8,  PIXMAN_null,     PIXMAN_a8r8g8b8, fbCompositeCopyAreasse2,               0 },
+    { PIXMAN_OP_SRC, PIXMAN_a8b8g8r8,  PIXMAN_null,     PIXMAN_a8b8g8r8, fbCompositeCopyAreasse2,               0 },
+    { PIXMAN_OP_SRC, PIXMAN_x8r8g8b8,  PIXMAN_null,     PIXMAN_x8r8g8b8, fbCompositeCopyAreasse2,               0 },
+    { PIXMAN_OP_SRC, PIXMAN_x8b8g8r8,  PIXMAN_null,     PIXMAN_x8b8g8r8, fbCompositeCopyAreasse2,               0 },
+    { PIXMAN_OP_SRC, PIXMAN_r5g6b5,    PIXMAN_null,     PIXMAN_r5g6b5,   fbCompositeCopyAreasse2,               0 },
+    { PIXMAN_OP_SRC, PIXMAN_b5g6r5,    PIXMAN_null,     PIXMAN_b5g6r5,   fbCompositeCopyAreasse2,               0 },
+
+    { PIXMAN_OP_IN,  PIXMAN_a8,        PIXMAN_null,     PIXMAN_a8,       fbCompositeIn_8x8sse2,                 0 },
+    { PIXMAN_OP_IN,  PIXMAN_solid,     PIXMAN_a8,       PIXMAN_a8,       fbCompositeIn_nx8x8sse2,               0 },
+
+    { PIXMAN_OP_NONE },
+};
+#endif
+
+#ifdef USE_VMX
+static const FastPathInfo vmx_fast_paths[] =
 {
     { PIXMAN_OP_NONE },
 };
 #endif
 
+
 static const FastPathInfo c_fast_paths[] =
 {
     { PIXMAN_OP_OVER, PIXMAN_solid,    PIXMAN_a8,       PIXMAN_r5g6b5,   fbCompositeSolidMask_nx8x0565, 0 },
@@ -1552,7 +1587,7 @@ get_fast_path (const FastPathInfo *fast_paths,
 	if (info->op != op)
 	    continue;
 
-	if ((info->src_format == PIXMAN_solid && can_get_solid (pSrc))		||
+	if ((info->src_format == PIXMAN_solid && pixman_image_can_get_solid (pSrc))		||
 	    (pSrc->type == BITS && info->src_format == pSrc->bits.format))
 	{
 	    valid_src = TRUE;
@@ -1594,7 +1629,101 @@ get_fast_path (const FastPathInfo *fast_paths,
     return NULL;
 }
 
-void
+/*
+ * Operator optimizations based on source or destination opacity
+ */
+typedef struct
+{
+    pixman_op_t			op;
+    pixman_op_t			opSrcDstOpaque;
+    pixman_op_t			opSrcOpaque;
+    pixman_op_t			opDstOpaque;
+} OptimizedOperatorInfo;
+
+static const OptimizedOperatorInfo optimized_operators[] =
+{
+    /* Input Operator           SRC&DST Opaque          SRC Opaque              DST Opaque      */
+    { PIXMAN_OP_OVER,           PIXMAN_OP_SRC,          PIXMAN_OP_SRC,          PIXMAN_OP_OVER },
+    { PIXMAN_OP_OVER_REVERSE,   PIXMAN_OP_DST,          PIXMAN_OP_OVER_REVERSE, PIXMAN_OP_DST },
+    { PIXMAN_OP_IN,             PIXMAN_OP_SRC,          PIXMAN_OP_IN,           PIXMAN_OP_SRC },
+    { PIXMAN_OP_IN_REVERSE,     PIXMAN_OP_DST,          PIXMAN_OP_DST,          PIXMAN_OP_IN_REVERSE },
+    { PIXMAN_OP_OUT,            PIXMAN_OP_CLEAR,        PIXMAN_OP_OUT,          PIXMAN_OP_CLEAR },
+    { PIXMAN_OP_OUT_REVERSE,    PIXMAN_OP_CLEAR,        PIXMAN_OP_CLEAR,        PIXMAN_OP_OUT_REVERSE },
+    { PIXMAN_OP_ATOP,           PIXMAN_OP_SRC,          PIXMAN_OP_IN,           PIXMAN_OP_OVER },
+    { PIXMAN_OP_ATOP_REVERSE,   PIXMAN_OP_DST,          PIXMAN_OP_OVER_REVERSE, PIXMAN_OP_IN_REVERSE },
+    { PIXMAN_OP_XOR,            PIXMAN_OP_CLEAR,        PIXMAN_OP_OUT,          PIXMAN_OP_OUT_REVERSE },
+    { PIXMAN_OP_SATURATE,       PIXMAN_OP_DST,          PIXMAN_OP_OVER_REVERSE, PIXMAN_OP_DST },
+    { PIXMAN_OP_NONE }
+};
+
+/*
+ * Check if the current operator could be optimized
+ */
+static const OptimizedOperatorInfo*
+pixman_operator_can_be_optimized(pixman_op_t op)
+{
+    const OptimizedOperatorInfo *info;
+
+    for (info = optimized_operators; info->op != PIXMAN_OP_NONE; info++)
+    {
+        if(info->op == op)
+            return info;
+    }
+    return NULL;
+}
+
+/*
+ * Optimize the current operator based on opacity of source or destination
+ * The output operator should be mathematically equivalent to the source.
+ */
+static pixman_op_t
+pixman_optimize_operator(pixman_op_t op, pixman_image_t *pSrc, pixman_image_t *pMask, pixman_image_t *pDst )
+{
+    pixman_bool_t is_source_opaque;
+    pixman_bool_t is_dest_opaque;
+    const OptimizedOperatorInfo *info = pixman_operator_can_be_optimized(op);
+
+    if(!info || pMask)
+        return op;
+
+    is_source_opaque = pixman_image_is_opaque(pSrc);
+    is_dest_opaque = pixman_image_is_opaque(pDst);
+
+    if(is_source_opaque == FALSE && is_dest_opaque == FALSE)
+        return op;
+
+    if(is_source_opaque && is_dest_opaque)
+        return info->opSrcDstOpaque;
+    else if(is_source_opaque)
+        return info->opSrcOpaque;
+    else if(is_dest_opaque)
+        return info->opDstOpaque;
+
+    return op;
+
+}
+
+#if defined(USE_SSE2) && defined(__GNUC__) && !defined(__x86_64__) && !defined(__amd64__)
+
+/*
+ * Work around GCC bug causing crashes in Mozilla with SSE2
+ * 
+ * When using SSE2 intrinsics, gcc assumes that the stack is 16 byte
+ * aligned. Unfortunately some code, such as Mozilla and Mono contain
+ * code that aligns the stack to 4 bytes.
+ *
+ * The __force_align_arg_pointer__ makes gcc generate a prologue that
+ * realigns the stack pointer to 16 bytes.
+ *
+ * On x86-64 this is not necessary because the standard ABI already
+ * calls for a 16 byte aligned stack.
+ *
+ * See https://bugs.freedesktop.org/show_bug.cgi?id=15693
+ */
+
+__attribute__((__force_align_arg_pointer__))
+#endif
+PIXMAN_EXPORT void
 pixman_image_composite (pixman_op_t      op,
 			pixman_image_t * pSrc,
 			pixman_image_t * pMask,
@@ -1617,14 +1746,18 @@ pixman_image_composite (pixman_op_t      op,
     pixman_bool_t dstAlphaMap = pDst->common.alpha_map != NULL;
     CompositeFunc func = NULL;
 
-#ifdef USE_SSE2
-    fbComposeSetupSSE();
-#endif
-    
 #ifdef USE_MMX
     fbComposeSetupMMX();
 #endif
 
+#ifdef USE_VMX
+    fbComposeSetupVMX();
+#endif
+
+#ifdef USE_SSE2
+    fbComposeSetupSSE2();
+#endif
+
     if (srcRepeat && srcTransform &&
 	pSrc->bits.width == 1 &&
 	pSrc->bits.height == 1)
@@ -1650,7 +1783,15 @@ pixman_image_composite (pixman_op_t      op,
 	}
     }
 
-    if ((pSrc->type == BITS || can_get_solid (pSrc)) && (!pMask || pMask->type == BITS)
+    /*
+    * Check if we can replace our operator by a simpler one if the src or dest are opaque
+    * The output operator should be mathematically equivalent to the source.
+    */
+    op = pixman_optimize_operator(op, pSrc, pMask, pDst);
+    if(op == PIXMAN_OP_DST)
+        return;
+
+    if ((pSrc->type == BITS || pixman_image_can_get_solid (pSrc)) && (!pMask || pMask->type == BITS)
         && !srcTransform && !maskTransform
         && !maskAlphaMap && !srcAlphaMap && !dstAlphaMap
         && (pSrc->common.filter != PIXMAN_FILTER_CONVOLUTION)
@@ -1674,17 +1815,21 @@ pixman_image_composite (pixman_op_t      op,
 	info = NULL;
 	
 #ifdef USE_SSE2
-	if (pixman_have_sse ())
-	    info = get_fast_path (sse_fast_paths, op, pSrc, pMask, pDst, pixbuf);
-	if (!info)
+	if (pixman_have_sse2 ())
+	    info = get_fast_path (sse2_fast_paths, op, pSrc, pMask, pDst, pixbuf);
 #endif
 
 #ifdef USE_MMX
-
-	if (pixman_have_mmx())
+	if (!info && pixman_have_mmx())
 	    info = get_fast_path (mmx_fast_paths, op, pSrc, pMask, pDst, pixbuf);
-	if (!info)
 #endif
+
+#ifdef USE_VMX
+
+	if (!info && pixman_have_vmx())
+	    info = get_fast_path (vmx_fast_paths, op, pSrc, pMask, pDst, pixbuf);
+#endif
+        if (!info)
 	    info = get_fast_path (c_fast_paths, op, pSrc, pMask, pDst, pixbuf);
 
 	if (info)
@@ -1751,6 +1896,59 @@ pixman_image_composite (pixman_op_t      op,
 }
 
 
+#ifdef USE_VMX
+/* The CPU detection code needs to be in a file not compiled with
+ * "-maltivec -mabi=altivec", as gcc would try to save vector register
+ * across function calls causing SIGILL on cpus without Altivec/vmx.
+ */
+static pixman_bool_t initialized = FALSE;
+static volatile pixman_bool_t have_vmx = TRUE;
+
+#ifdef __APPLE__
+#include <sys/sysctl.h>
+
+pixman_bool_t pixman_have_vmx (void) {
+    if(!initialized) {
+        size_t length = sizeof(have_vmx);
+        int error =
+            sysctlbyname("hw.optional.altivec", &have_vmx, &length, NULL, 0);
+        if(error) have_vmx = FALSE;
+        initialized = TRUE;
+    }
+    return have_vmx;
+}
+
+#else
+#include <signal.h>
+#include <setjmp.h>
+
+static jmp_buf jump_env;
+
+static void vmx_test(int sig, siginfo_t *si, void *unused) {
+    longjmp (jump_env, 1);
+}
+
+pixman_bool_t pixman_have_vmx (void) {
+    struct sigaction sa, osa;
+    int jmp_result;
+    if (!initialized) {
+        sa.sa_flags = SA_SIGINFO;
+        sigemptyset(&sa.sa_mask);
+        sa.sa_sigaction = vmx_test;
+        sigaction(SIGILL, &sa, &osa);
+	jmp_result = setjmp (jump_env);
+	if (jmp_result == 0) {
+	    asm volatile ( "vor 0, 0, 0" );
+	}
+        sigaction(SIGILL, &osa, NULL);
+	have_vmx = (jmp_result == 0);
+        initialized = TRUE;
+    }
+    return have_vmx;
+}
+#endif /* __APPLE__ */
+#endif /* USE_VMX */
+
 #ifdef USE_MMX
 /* The CPU detection code needs to be in a file not compiled with
  * "-mmmx -msse", as gcc would generate CMOV instructions otherwise
@@ -1951,19 +2149,19 @@ pixman_have_mmx (void)
 
 #ifdef USE_SSE2
 pixman_bool_t
-pixman_have_sse (void)
+pixman_have_sse2 (void)
 {
     static pixman_bool_t initialized = FALSE;
-    static pixman_bool_t sse_present;
+    static pixman_bool_t sse2_present;
 
     if (!initialized)
     {
         unsigned int features = detectCPUFeatures();
-        sse_present = (features & (MMX|MMX_Extensions|SSE|SSE2)) == (MMX|MMX_Extensions|SSE|SSE2);
+        sse2_present = (features & (MMX|MMX_Extensions|SSE|SSE2)) == (MMX|MMX_Extensions|SSE|SSE2);
         initialized = TRUE;
     }
 
-    return sse_present;
+    return sse2_present;
 }
 #endif
 
diff --git a/lib/pixman/pixman/pixman-private.h b/lib/pixman/pixman/pixman-private.h
index 5767c1084..e26a0c3de 100644
--- a/lib/pixman/pixman/pixman-private.h
+++ b/lib/pixman/pixman/pixman-private.h
@@ -143,15 +143,25 @@ typedef struct point point_t;
  */
 
 #define FASTCALL
-typedef FASTCALL void (*CombineMaskU) (uint32_t *src, const uint32_t *mask, int width);
-typedef FASTCALL void (*CombineFuncU) (uint32_t *dest, const uint32_t *src, int width);
-typedef FASTCALL void (*CombineFuncC) (uint32_t *dest, uint32_t *src, uint32_t *mask, int width);
-typedef FASTCALL void (*fetchProc)(bits_image_t *pict, int x, int y, int width,
-                                   uint32_t *buffer);
-typedef FASTCALL uint32_t (*fetchPixelProc)(bits_image_t *pict, int offset, int line);
-typedef FASTCALL void (*storeProc)(pixman_image_t *, uint32_t *bits,
-                                   const uint32_t *values, int x, int width,
-                                   const pixman_indexed_t *);
+typedef FASTCALL void (*CombineMaskU32) (uint32_t *src, const uint32_t *mask, int width);
+typedef FASTCALL void (*CombineFuncU32) (uint32_t *dest, const uint32_t *src, int width);
+typedef FASTCALL void (*CombineFuncC32) (uint32_t *dest, uint32_t *src, uint32_t *mask, int width);
+typedef FASTCALL void (*fetchProc32)(bits_image_t *pict, int x, int y, int width,
+                                     uint32_t *buffer);
+typedef FASTCALL uint32_t (*fetchPixelProc32)(bits_image_t *pict, int offset, int line);
+typedef FASTCALL void (*storeProc32)(pixman_image_t *, uint32_t *bits,
+                                     const uint32_t *values, int x, int width,
+                                     const pixman_indexed_t *);
+
+typedef FASTCALL void (*CombineMaskU64) (uint64_t *src, const uint64_t *mask, int width);
+typedef FASTCALL void (*CombineFuncU64) (uint64_t *dest, const uint64_t *src, int width);
+typedef FASTCALL void (*CombineFuncC64) (uint64_t *dest, uint64_t *src, uint64_t *mask, int width);
+typedef FASTCALL void (*fetchProc64)(bits_image_t *pict, int x, int y, int width,
+                                     uint64_t *buffer);
+typedef FASTCALL uint64_t (*fetchPixelProc64)(bits_image_t *pict, int offset, int line);
+typedef FASTCALL void (*storeProc64)(pixman_image_t *, uint32_t *bits,
+                                     const uint64_t *values, int x, int width,
+                                     const pixman_indexed_t *);
 
 typedef struct _FbComposeData {
     uint8_t	 op;
@@ -168,28 +178,49 @@ typedef struct _FbComposeData {
     uint16_t	 height;
 } FbComposeData;
 
-typedef struct _FbComposeFunctions {
-    CombineFuncU *combineU;
-    CombineFuncC *combineC;
-    CombineMaskU combineMaskU;
-} FbComposeFunctions;
+typedef struct _FbComposeFunctions32 {
+    CombineFuncU32 *combineU;
+    CombineFuncC32 *combineC;
+    CombineMaskU32 combineMaskU;
+} FbComposeFunctions32;
 
-extern FbComposeFunctions pixman_composeFunctions;
+typedef struct _FbComposeFunctions64 {
+    CombineFuncU64 *combineU;
+    CombineFuncC64 *combineC;
+    CombineMaskU64 combineMaskU;
+} FbComposeFunctions64;
 
-void pixman_composite_rect_general_accessors (const FbComposeData *data,
-					      uint32_t *scanline_buffer);
-void pixman_composite_rect_general (const FbComposeData *data,
-				    uint32_t *scanline_buffer);
+extern FbComposeFunctions32 pixman_composeFunctions;
+extern FbComposeFunctions64 pixman_composeFunctions64;
 
-fetchProc pixman_fetchProcForPicture (bits_image_t *);
-fetchPixelProc pixman_fetchPixelProcForPicture (bits_image_t *);
-storeProc pixman_storeProcForPicture (bits_image_t *);
-fetchProc pixman_fetchProcForPicture_accessors (bits_image_t *);
-fetchPixelProc pixman_fetchPixelProcForPicture_accessors (bits_image_t *);
-storeProc pixman_storeProcForPicture_accessors (bits_image_t *);
+void pixman_composite_rect_general_accessors (const FbComposeData *data,
+                                              void *src_buffer,
+                                              void *mask_buffer,
+                                              void *dest_buffer,
+                                              const int wide);
+void pixman_composite_rect_general (const FbComposeData *data);
+
+fetchProc32 pixman_fetchProcForPicture32 (bits_image_t *);
+fetchPixelProc32 pixman_fetchPixelProcForPicture32 (bits_image_t *);
+storeProc32 pixman_storeProcForPicture32 (bits_image_t *);
+fetchProc32 pixman_fetchProcForPicture32_accessors (bits_image_t *);
+fetchPixelProc32 pixman_fetchPixelProcForPicture32_accessors (bits_image_t *);
+storeProc32 pixman_storeProcForPicture32_accessors (bits_image_t *);
+
+fetchProc64 pixman_fetchProcForPicture64 (bits_image_t *);
+fetchPixelProc64 pixman_fetchPixelProcForPicture64 (bits_image_t *);
+storeProc64 pixman_storeProcForPicture64 (bits_image_t *);
+fetchProc64 pixman_fetchProcForPicture64_accessors (bits_image_t *);
+fetchPixelProc64 pixman_fetchPixelProcForPicture64_accessors (bits_image_t *);
+storeProc64 pixman_storeProcForPicture64_accessors (bits_image_t *);
+
+void pixman_expand(uint64_t *dst, const uint32_t *src, pixman_format_code_t, int width);
+void pixman_contract(uint32_t *dst, const uint64_t *src, int width);
 
 void pixmanFetchSourcePict(source_image_t *, int x, int y, int width,
                            uint32_t *buffer, uint32_t *mask, uint32_t maskBits);
+void pixmanFetchSourcePict64(source_image_t *, int x, int y, int width,
+                             uint64_t *buffer, uint64_t *mask, uint32_t maskBits);
 
 void fbFetchTransformed(bits_image_t *, int x, int y, int width,
                         uint32_t *buffer, uint32_t *mask, uint32_t maskBits);
@@ -207,6 +238,22 @@ void fbFetchExternalAlpha_accessors(bits_image_t *, int x, int y, int width,
                                     uint32_t *buffer, uint32_t *mask,
                                     uint32_t maskBits);
 
+void fbFetchTransformed64(bits_image_t *, int x, int y, int width,
+                          uint64_t *buffer, uint64_t *mask, uint32_t maskBits);
+void fbStoreExternalAlpha64(bits_image_t *, int x, int y, int width,
+                            uint64_t *buffer);
+void fbFetchExternalAlpha64(bits_image_t *, int x, int y, int width,
+                            uint64_t *buffer, uint64_t *mask, uint32_t maskBits);
+
+void fbFetchTransformed64_accessors(bits_image_t *, int x, int y, int width,
+                                    uint64_t *buffer, uint64_t *mask,
+                                    uint32_t maskBits);
+void fbStoreExternalAlpha64_accessors(bits_image_t *, int x, int y, int width,
+                                      uint64_t *buffer);
+void fbFetchExternalAlpha64_accessors(bits_image_t *, int x, int y, int width,
+                                      uint64_t *buffer, uint64_t *mask,
+                                      uint32_t maskBits);
+
 /* end */
 
 typedef enum
@@ -236,9 +283,9 @@ struct image_common
 {
     image_type_t		type;
     int32_t			ref_count;
-    pixman_region16_t		full_region;
-    pixman_region16_t		clip_region;
-    pixman_region16_t	       *src_clip;
+    pixman_region32_t		full_region;
+    pixman_region32_t		clip_region;
+    pixman_region32_t	       *src_clip;
     pixman_bool_t               has_client_clip;
     pixman_transform_t	       *transform;
     pixman_repeat_t		repeat;
@@ -332,10 +379,6 @@ union pixman_image
 };
 
 
-extern CombineFuncU pixman_fbCombineFuncU[];
-extern CombineFuncC pixman_fbCombineFuncC[];
-FASTCALL void pixman_fbCombineMaskU (uint32_t *src, const uint32_t *mask, int width);
-
 #define LOG2_BITMAP_PAD 5
 #define FB_STIP_SHIFT	LOG2_BITMAP_PAD
 #define FB_STIP_UNIT	(1 << FB_STIP_SHIFT)
@@ -397,11 +440,6 @@ FASTCALL void pixman_fbCombineMaskU (uint32_t *src, const uint32_t *mask, int wi
      WRITE(img, (a)+2, (uint8_t) ((v) >> 16))))
 #endif
 
-#define Alpha(x) ((x) >> 24)
-#define Red(x) (((x) >> 16) & 0xff)
-#define Green(x) (((x) >> 8) & 0xff)
-#define Blue(x) ((x) & 0xff)
-
 #define CvtR8G8B8toY15(s)       (((((s) >> 16) & 0xff) * 153 + \
                                   (((s) >>  8) & 0xff) * 301 +		\
                                   (((s)      ) & 0xff) * 58) >> 2)
@@ -445,192 +483,11 @@ FASTCALL void pixman_fbCombineMaskU (uint32_t *src, const uint32_t *mask, int wi
 
 #define FbInC(x,i,a,t) ((uint32_t) FbIntMult(FbGet8(x,i),FbGet8(a,i),(t)) << (i))
 
-#define FbGen(x,y,i,ax,ay,t,u,v) ((t) = (FbIntMult(FbGet8(y,i),ay,(u)) + \
-					 FbIntMult(FbGet8(x,i),ax,(v))), \
-				  (uint32_t) ((uint8_t) ((t) |		\
-							 (0 - ((t) >> 8)))) << (i))
-
 #define FbAdd(x,y,i,t)	((t) = FbGet8(x,i) + FbGet8(y,i),		\
 			 (uint32_t) ((uint8_t) ((t) | (0 - ((t) >> 8)))) << (i))
 
-
-/*
-  The methods below use some tricks to be able to do two color
-  components at the same time.
-*/
-
-/*
-  x_c = (x_c * a) / 255
-*/
-#define FbByteMul(x, a) do {					    \
-        uint32_t t = ((x & 0xff00ff) * a) + 0x800080;               \
-        t = (t + ((t >> 8) & 0xff00ff)) >> 8;			    \
-        t &= 0xff00ff;						    \
-								    \
-        x = (((x >> 8) & 0xff00ff) * a) + 0x800080;		    \
-        x = (x + ((x >> 8) & 0xff00ff));			    \
-        x &= 0xff00ff00;					    \
-        x += t;							    \
-    } while (0)
-
-/*
-  x_c = (x_c * a) / 255 + y
-*/
-#define FbByteMulAdd(x, a, y) do {				    \
-        uint32_t t = ((x & 0xff00ff) * a) + 0x800080;               \
-        t = (t + ((t >> 8) & 0xff00ff)) >> 8;			    \
-        t &= 0xff00ff;						    \
-        t += y & 0xff00ff;					    \
-        t |= 0x1000100 - ((t >> 8) & 0xff00ff);			    \
-        t &= 0xff00ff;						    \
-								    \
-        x = (((x >> 8) & 0xff00ff) * a) + 0x800080;                 \
-        x = (x + ((x >> 8) & 0xff00ff)) >> 8;                       \
-        x &= 0xff00ff;                                              \
-        x += (y >> 8) & 0xff00ff;                                   \
-        x |= 0x1000100 - ((x >> 8) & 0xff00ff);                     \
-        x &= 0xff00ff;                                              \
-        x <<= 8;                                                    \
-        x += t;                                                     \
-    } while (0)
-
-/*
-  x_c = (x_c * a + y_c * b) / 255
-*/
-#define FbByteAddMul(x, a, y, b) do {                                   \
-        uint32_t t;							\
-        uint32_t r = (x >> 24) * a + (y >> 24) * b + 0x80;		\
-        r += (r >> 8);                                                  \
-        r >>= 8;                                                        \
-									\
-        t = (x & 0xff00) * a + (y & 0xff00) * b;                        \
-        t += (t >> 8) + 0x8000;                                         \
-        t >>= 16;                                                       \
-									\
-        t |= r << 16;                                                   \
-        t |= 0x1000100 - ((t >> 8) & 0xff00ff);                         \
-        t &= 0xff00ff;                                                  \
-        t <<= 8;                                                        \
-									\
-        r = ((x >> 16) & 0xff) * a + ((y >> 16) & 0xff) * b + 0x80;     \
-        r += (r >> 8);                                                  \
-        r >>= 8;                                                        \
-									\
-        x = (x & 0xff) * a + (y & 0xff) * b + 0x80;                     \
-        x += (x >> 8);                                                  \
-        x >>= 8;                                                        \
-        x |= r << 16;                                                   \
-        x |= 0x1000100 - ((x >> 8) & 0xff00ff);                         \
-        x &= 0xff00ff;                                                  \
-        x |= t;                                                         \
-    } while (0)
-
-/*
-  x_c = (x_c * a + y_c *b) / 256
-*/
-#define FbByteAddMul_256(x, a, y, b) do {                               \
-        uint32_t t = (x & 0xff00ff) * a + (y & 0xff00ff) * b;		\
-        t >>= 8;                                                        \
-        t &= 0xff00ff;                                                  \
-									\
-        x = ((x >> 8) & 0xff00ff) * a + ((y >> 8) & 0xff00ff) * b;      \
-        x &= 0xff00ff00;                                                \
-        x += t;                                                         \
-    } while (0)
-
-/*
-  x_c = (x_c * a_c) / 255
-*/
-#define FbByteMulC(x, a) do {				  \
-        uint32_t t;                                       \
-        uint32_t r = (x & 0xff) * (a & 0xff);             \
-        r |= (x & 0xff0000) * ((a >> 16) & 0xff);	  \
-	r += 0x800080;					  \
-        r = (r + ((r >> 8) & 0xff00ff)) >> 8;		  \
-        r &= 0xff00ff;					  \
-							  \
-        x >>= 8;					  \
-        t = (x & 0xff) * ((a >> 8) & 0xff);		  \
-        t |= (x & 0xff0000) * (a >> 24);		  \
-        t += 0x800080;					  \
-        t = t + ((t >> 8) & 0xff00ff);			  \
-        x = r | (t & 0xff00ff00);			  \
-							  \
-    } while (0)
-
-/*
-  x_c = (x_c * a) / 255 + y
-*/
-#define FbByteMulAddC(x, a, y) do {				      \
-        uint32_t t;                                                   \
-        uint32_t r = (x & 0xff) * (a & 0xff);                         \
-        r |= (x & 0xff0000) * ((a >> 16) & 0xff);		      \
-	r += 0x800080;						      \
-	r = (r + ((r >> 8) & 0xff00ff)) >> 8;			      \
-        r &= 0xff00ff;						      \
-        r += y & 0xff00ff;					      \
-        r |= 0x1000100 - ((r >> 8) & 0xff00ff);			      \
-        r &= 0xff00ff;						      \
-								      \
-        x >>= 8;                                                       \
-        t = (x & 0xff) * ((a >> 8) & 0xff);                            \
-        t |= (x & 0xff0000) * (a >> 24);                               \
-	t += 0x800080;                                                 \
-        t = (t + ((t >> 8) & 0xff00ff)) >> 8;			       \
-        t &= 0xff00ff;                                                 \
-        t += (y >> 8) & 0xff00ff;                                      \
-        t |= 0x1000100 - ((t >> 8) & 0xff00ff);                        \
-        t &= 0xff00ff;                                                 \
-        x = r | (t << 8);                                              \
-    } while (0)
-
-/*
-  x_c = (x_c * a_c + y_c * b) / 255
-*/
-#define FbByteAddMulC(x, a, y, b) do {                                  \
-        uint32_t t;							\
-        uint32_t r = (x >> 24) * (a >> 24) + (y >> 24) * b;		\
-        r += (r >> 8) + 0x80;                                           \
-        r >>= 8;                                                        \
-									\
-        t = (x & 0xff00) * ((a >> 8) & 0xff) + (y & 0xff00) * b;        \
-        t += (t >> 8) + 0x8000;                                         \
-        t >>= 16;                                                       \
-									\
-        t |= r << 16;                                                   \
-        t |= 0x1000100 - ((t >> 8) & 0xff00ff);                         \
-        t &= 0xff00ff;                                                  \
-        t <<= 8;                                                        \
-									\
-        r = ((x >> 16) & 0xff) * ((a >> 16) & 0xff) + ((y >> 16) & 0xff) * b + 0x80; \
-        r += (r >> 8);                                                  \
-        r >>= 8;                                                        \
-									\
-        x = (x & 0xff) * (a & 0xff) + (y & 0xff) * b + 0x80;            \
-        x += (x >> 8);                                                  \
-        x >>= 8;                                                        \
-        x |= r << 16;                                                   \
-        x |= 0x1000100 - ((x >> 8) & 0xff00ff);                         \
-        x &= 0xff00ff;                                                  \
-        x |= t;                                                         \
-    } while (0)
-
-/*
-  x_c = min(x_c + y_c, 255)
-*/
-#define FbByteAdd(x, y) do {                                            \
-        uint32_t t;							\
-        uint32_t r = (x & 0xff00ff) + (y & 0xff00ff);			\
-        r |= 0x1000100 - ((r >> 8) & 0xff00ff);                         \
-        r &= 0xff00ff;                                                  \
-									\
-        t = ((x >> 8) & 0xff00ff) + ((y >> 8) & 0xff00ff);              \
-        t |= 0x1000100 - ((t >> 8) & 0xff00ff);                         \
-        r |= (t & 0xff00ff) << 8;                                       \
-        x = r;                                                          \
-    } while (0)
-
 #define div_255(x) (((x) + 0x80 + (((x) + 0x80) >> 8)) >> 8)
+#define div_65535(x) (((x) + 0x8000 + (((x) + 0x8000) >> 16)) >> 16)
 
 #define MOD(a,b) ((a) < 0 ? ((b) - ((-(a) - 1) % (b))) - 1 : (a) % (b))
 
@@ -685,6 +542,8 @@ FASTCALL void pixman_fbCombineMaskU (uint32_t *src, const uint32_t *mask, int wi
 
 #ifdef PIXMAN_FB_ACCESSORS
 
+#define ACCESS(sym) sym##_accessors
+
 #define READ(img, ptr)							\
     ((img)->common.read_func ((ptr), sizeof(*(ptr))))
 #define WRITE(img, ptr,val)						\
@@ -710,6 +569,8 @@ FASTCALL void pixman_fbCombineMaskU (uint32_t *src, const uint32_t *mask, int wi
 
 #else
 
+#define ACCESS(sym) sym
+
 #define READ(img, ptr)		(*(ptr))
 #define WRITE(img, ptr, val)	(*(ptr) = (val))
 #define MEMCPY_WRAPPED(img, dst, src, size)					\
@@ -786,6 +647,10 @@ FASTCALL void pixman_fbCombineMaskU (uint32_t *src, const uint32_t *mask, int wi
     } while (0)
 
 
+#define PIXMAN_FORMAT_16BPC(f)	(PIXMAN_FORMAT_A(f) > 8 || \
+				 PIXMAN_FORMAT_R(f) > 8 || \
+				 PIXMAN_FORMAT_G(f) > 8 || \
+				 PIXMAN_FORMAT_B(f) > 8)
 /*
  * Edges
  */
@@ -841,6 +706,44 @@ pixman_rasterize_edges_accessors (pixman_image_t *image,
 				  pixman_fixed_t	t,
 				  pixman_fixed_t	b);
 
+pixman_bool_t
+pixman_image_is_opaque(pixman_image_t *image);
+
+pixman_bool_t
+pixman_image_can_get_solid (pixman_image_t *image);
+
+pixman_bool_t
+pixman_compute_composite_region32 (pixman_region32_t *	pRegion,
+				   pixman_image_t *	pSrc,
+				   pixman_image_t *	pMask,
+				   pixman_image_t *	pDst,
+				   int16_t		xSrc,
+				   int16_t		ySrc,
+				   int16_t		xMask,
+				   int16_t		yMask,
+				   int16_t		xDst,
+				   int16_t		yDst,
+				   uint16_t		width,
+				   uint16_t		height);
+
+/* GCC visibility */
+#if defined(__GNUC__) && __GNUC__ >= 4
+#define PIXMAN_EXPORT __attribute__ ((visibility("default")))
+/* Sun Studio 8 visibility */
+#elif defined(__SUNPRO_C) && (__SUNPRO_C >= 0x550)
+#define PIXMAN_EXPORT __global
+#else
+#define PIXMAN_EXPORT
+#endif
+
+/* Region Helpers */
+pixman_bool_t pixman_region32_copy_from_region16 (pixman_region32_t *dst,
+						  pixman_region16_t *src);
+pixman_bool_t pixman_region16_copy_from_region32 (pixman_region16_t *dst,
+						  pixman_region32_t *src);
+void pixman_region_internal_set_static_pointers (pixman_box16_t *empty_box,
+						 pixman_region16_data_t *empty_data,
+						 pixman_region16_data_t *broken_data);
 
 #ifdef PIXMAN_TIMING
 
diff --git a/lib/pixman/pixman/pixman-region.c b/lib/pixman/pixman/pixman-region.c
index 496ce7728..559454a7e 100644
--- a/lib/pixman/pixman/pixman-region.c
+++ b/lib/pixman/pixman/pixman-region.c
@@ -45,29 +45,19 @@ SOFTWARE.
 
 ******************************************************************/
 
-#ifdef HAVE_CONFIG_H
-#include <config.h>
-#endif
-
 #include <stdlib.h>
 #include <limits.h>
 #include <string.h>
 #include <stdio.h>
 
-#include "pixman-private.h"
-
-typedef struct pixman_region16_point {
-    int x, y;
-} pixman_region16_point_t;
-
 #define PIXREGION_NIL(reg) ((reg)->data && !(reg)->data->numRects)
 /* not a region */
 #define PIXREGION_NAR(reg)	((reg)->data == pixman_brokendata)
 #define PIXREGION_NUM_RECTS(reg) ((reg)->data ? (reg)->data->numRects : 1)
 #define PIXREGION_SIZE(reg) ((reg)->data ? (reg)->data->size : 0)
-#define PIXREGION_RECTS(reg) ((reg)->data ? (pixman_box16_t *)((reg)->data + 1) \
+#define PIXREGION_RECTS(reg) ((reg)->data ? (box_type_t *)((reg)->data + 1) \
 			               : &(reg)->extents)
-#define PIXREGION_BOXPTR(reg) ((pixman_box16_t *)((reg)->data + 1))
+#define PIXREGION_BOXPTR(reg) ((box_type_t *)((reg)->data + 1))
 #define PIXREGION_BOX(reg,i) (&PIXREGION_BOXPTR(reg)[i])
 #define PIXREGION_TOP(reg) PIXREGION_BOX(reg, (reg)->data->numRects)
 #define PIXREGION_END(reg) PIXREGION_BOX(reg, (reg)->data->numRects - 1)
@@ -82,20 +72,20 @@ typedef struct pixman_region16_point {
 #define assert(expr)
 #endif
 
-#define good(reg) assert(pixman_region_selfcheck(reg))
+#define good(reg) assert(PREFIX(_selfcheck) (reg))
 
 #undef MIN
 #define MIN(a,b) ((a) < (b) ? (a) : (b))
 #undef MAX
 #define MAX(a,b) ((a) > (b) ? (a) : (b))
 
-static const pixman_box16_t _pixman_region_emptyBox = {0, 0, 0, 0};
-static const pixman_region16_data_t _pixman_region_emptyData = {0, 0};
-static const pixman_region16_data_t _pixman_brokendata = {0, 0};
+static const box_type_t PREFIX(_emptyBox_) = {0, 0, 0, 0};
+static const region_data_type_t PREFIX(_emptyData_) = {0, 0};
+static const region_data_type_t PREFIX(_brokendata_) = {0, 0};
 
-static pixman_box16_t *pixman_region_emptyBox = (pixman_box16_t *)&_pixman_region_emptyBox;
-static pixman_region16_data_t *pixman_region_emptyData = (pixman_region16_data_t *)&_pixman_region_emptyData;
-static pixman_region16_data_t *pixman_brokendata = (pixman_region16_data_t *)&_pixman_brokendata;
+static box_type_t *pixman_region_emptyBox = (box_type_t *)&PREFIX(_emptyBox_);
+static region_data_type_t *pixman_region_emptyData = (region_data_type_t *)&PREFIX(_emptyData_);
+static region_data_type_t *pixman_brokendata = (region_data_type_t *)&PREFIX(_brokendata_);
 
 /* This function exists only to make it possible to preserve the X ABI - it should
  * go away at first opportunity.
@@ -106,9 +96,9 @@ static pixman_region16_data_t *pixman_brokendata = (pixman_region16_data_t *)&_p
  * work.
  */
 void
-pixman_region_set_static_pointers (pixman_box16_t *empty_box,
-				   pixman_region16_data_t *empty_data,
-				   pixman_region16_data_t *broken_data)
+PREFIX(_internal_set_static_pointers) (box_type_t *empty_box,
+				       region_data_type_t *empty_data,
+				       region_data_type_t *broken_data)
 {
     pixman_region_emptyBox = empty_box;
     pixman_region_emptyData = empty_data;
@@ -116,7 +106,7 @@ pixman_region_set_static_pointers (pixman_box16_t *empty_box,
 }
 
 static pixman_bool_t
-pixman_break (pixman_region16_t *pReg);
+pixman_break (region_type_t *pReg);
 
 /*
  * The functions in this file implement the Region abstraction used extensively
@@ -190,14 +180,14 @@ pixman_break (pixman_region16_t *pReg);
 static size_t
 PIXREGION_SZOF(size_t n)
 {
-    size_t size = n * sizeof(pixman_box16_t);
-    if (n > UINT32_MAX / sizeof(pixman_box16_t))
+    size_t size = n * sizeof(box_type_t);
+    if (n > UINT32_MAX / sizeof(box_type_t))
         return 0;
 
-    if (sizeof(pixman_region16_data_t) > UINT32_MAX - size)
+    if (sizeof(region_data_type_t) > UINT32_MAX - size)
         return 0;
 
-    return size + sizeof(pixman_region16_data_t);
+    return size + sizeof(region_data_type_t);
 }
 
 static void *
@@ -245,12 +235,12 @@ if (!(pReg)->data || (((pReg)->data->numRects + (n)) > (pReg)->data->size)) \
 #define DOWNSIZE(reg,numRects)						\
     if (((numRects) < ((reg)->data->size >> 1)) && ((reg)->data->size > 50)) \
     {									\
-	pixman_region16_data_t * NewData;				\
+	region_data_type_t * NewData;				\
 	size_t data_size = PIXREGION_SZOF(numRects);			\
 	if (!data_size)							\
 	    NewData = NULL;						\
 	else								\
-	    NewData = (pixman_region16_data_t *)realloc((reg)->data, data_size); \
+	    NewData = (region_data_type_t *)realloc((reg)->data, data_size); \
 	if (NewData)							\
 	{								\
 	    NewData->size = (numRects);					\
@@ -258,14 +248,14 @@ if (!(pReg)->data || (((pReg)->data->numRects + (n)) > (pReg)->data->size)) \
 	}								\
     }
 
-pixman_bool_t
-pixman_region_equal(reg1, reg2)
-    pixman_region16_t * reg1;
-    pixman_region16_t * reg2;
+PIXMAN_EXPORT pixman_bool_t
+PREFIX(_equal) (reg1, reg2)
+    region_type_t * reg1;
+    region_type_t * reg2;
 {
     int i;
-    pixman_box16_t *rects1;
-    pixman_box16_t *rects2;
+    box_type_t *rects1;
+    box_type_t *rects2;
 
     if (reg1->extents.x1 != reg2->extents.x1) return FALSE;
     if (reg1->extents.x2 != reg2->extents.x2) return FALSE;
@@ -285,12 +275,12 @@ pixman_region_equal(reg1, reg2)
 }
 
 int
-pixman_region16_print(rgn)
-    pixman_region16_t * rgn;
+PREFIX(_print) (rgn)
+    region_type_t * rgn;
 {
     int num, size;
     int i;
-    pixman_box16_t * rects;
+    box_type_t * rects;
 
     num = PIXREGION_NUM_RECTS(rgn);
     size = PIXREGION_SIZE(rgn);
@@ -306,16 +296,16 @@ pixman_region16_print(rgn)
 }
 
 
-void
-pixman_region_init (pixman_region16_t *region)
+PIXMAN_EXPORT void
+PREFIX(_init) (region_type_t *region)
 {
     region->extents = *pixman_region_emptyBox;
     region->data = pixman_region_emptyData;
 }
 
-void
-pixman_region_init_rect (pixman_region16_t *region,
-			 int x, int y, unsigned int width, unsigned int height)
+PIXMAN_EXPORT void
+PREFIX(_init_rect) (region_type_t *region,
+		    int x, int y, unsigned int width, unsigned int height)
 {
     region->extents.x1 = x;
     region->extents.y1 = y;
@@ -324,35 +314,29 @@ pixman_region_init_rect (pixman_region16_t *region,
     region->data = NULL;
 }
 
-void
-pixman_region_init_with_extents (pixman_region16_t *region, pixman_box16_t *extents)
+PIXMAN_EXPORT void
+PREFIX(_init_with_extents) (region_type_t *region, box_type_t *extents)
 {
     region->extents = *extents;
     region->data = NULL;
 }
 
-void
-pixman_region_fini (pixman_region16_t *region)
+PIXMAN_EXPORT void
+PREFIX(_fini) (region_type_t *region)
 {
     good (region);
     freeData (region);
 }
 
-int
-pixman_region_n_rects (pixman_region16_t *region)
+PIXMAN_EXPORT int
+PREFIX(_n_rects) (region_type_t *region)
 {
     return PIXREGION_NUM_RECTS (region);
 }
 
-pixman_box16_t *
-pixman_region_rects (pixman_region16_t *region)
-{
-    return PIXREGION_RECTS (region);
-}
-
-pixman_box16_t *
-pixman_region_rectangles (pixman_region16_t *region,
-			  int		    *n_rects)
+PIXMAN_EXPORT box_type_t *
+PREFIX(_rectangles) (region_type_t *region,
+				  int		    *n_rects)
 {
     if (n_rects)
 	*n_rects = PIXREGION_NUM_RECTS (region);
@@ -361,7 +345,7 @@ pixman_region_rectangles (pixman_region16_t *region,
 }
 
 static pixman_bool_t
-pixman_break (pixman_region16_t *region)
+pixman_break (region_type_t *region)
 {
     freeData (region);
     region->extents = *pixman_region_emptyBox;
@@ -370,9 +354,9 @@ pixman_break (pixman_region16_t *region)
 }
 
 static pixman_bool_t
-pixman_rect_alloc (pixman_region16_t * region, int n)
+pixman_rect_alloc (region_type_t * region, int n)
 {
-    pixman_region16_data_t *data;
+    region_data_type_t *data;
 
     if (!region->data)
     {
@@ -404,7 +388,7 @@ pixman_rect_alloc (pixman_region16_t * region, int n)
 	if (!data_size)
 	    data = NULL;
 	else
-	    data = (pixman_region16_data_t *)realloc(region->data, PIXREGION_SZOF(n));
+	    data = (region_data_type_t *)realloc(region->data, PIXREGION_SZOF(n));
 	if (!data)
 	    return pixman_break (region);
 	region->data = data;
@@ -413,8 +397,8 @@ pixman_rect_alloc (pixman_region16_t * region, int n)
     return TRUE;
 }
 
-pixman_bool_t
-pixman_region_copy (pixman_region16_t *dst, pixman_region16_t *src)
+PIXMAN_EXPORT pixman_bool_t
+PREFIX(_copy) (region_type_t *dst, region_type_t *src)
 {
     good(dst);
     good(src);
@@ -437,7 +421,7 @@ pixman_region_copy (pixman_region16_t *dst, pixman_region16_t *src)
     }
     dst->data->numRects = src->data->numRects;
     memmove((char *)PIXREGION_BOXPTR(dst),(char *)PIXREGION_BOXPTR(src),
-	  dst->data->numRects * sizeof(pixman_box16_t));
+	  dst->data->numRects * sizeof(box_type_t));
     return TRUE;
 }
 
@@ -465,12 +449,12 @@ pixman_region_copy (pixman_region16_t *dst, pixman_region16_t *src)
  */
 static inline int
 pixman_coalesce (
-    pixman_region16_t *	region,	    	/* Region to coalesce		     */
+    region_type_t *	region,	    	/* Region to coalesce		     */
     int	    	  	prevStart,  	/* Index of start of previous band   */
     int	    	  	curStart)   	/* Index of start of current band    */
 {
-    pixman_box16_t *	pPrevBox;   	/* Current box in previous band	     */
-    pixman_box16_t *	pCurBox;    	/* Current box in current band       */
+    box_type_t *	pPrevBox;   	/* Current box in previous band	     */
+    box_type_t *	pCurBox;    	/* Current box in current band       */
     int  	numRects;	/* Number rectangles in both bands   */
     int	y2;		/* Bottom of current band	     */
     /*
@@ -548,13 +532,13 @@ pixman_coalesce (
 
 static inline pixman_bool_t
 pixman_region_appendNonO (
-    pixman_region16_t *	region,
-    pixman_box16_t *	r,
-    pixman_box16_t *  	  	rEnd,
+    region_type_t *	region,
+    box_type_t *	r,
+    box_type_t *  	  	rEnd,
     int  	y1,
     int  	y2)
 {
-    pixman_box16_t *	pNextRect;
+    box_type_t *	pNextRect;
     int	newRects;
 
     newRects = rEnd - r;
@@ -590,7 +574,7 @@ pixman_region_appendNonO (
     if ((newRects = rEnd - r)) {					\
 	RECTALLOC(newReg, newRects);					\
 	memmove((char *)PIXREGION_TOP(newReg),(char *)r, 			\
-              newRects * sizeof(pixman_box16_t));				\
+              newRects * sizeof(box_type_t));				\
 	newReg->data->numRects += newRects;				\
     }									\
 }
@@ -625,20 +609,20 @@ pixman_region_appendNonO (
  */
 
 typedef pixman_bool_t (*OverlapProcPtr)(
-    pixman_region16_t	 *region,
-    pixman_box16_t *r1,
-    pixman_box16_t *r1End,
-    pixman_box16_t *r2,
-    pixman_box16_t *r2End,
-    short    	 y1,
-    short    	 y2,
+    region_type_t	 *region,
+    box_type_t *r1,
+    box_type_t *r1End,
+    box_type_t *r2,
+    box_type_t *r2End,
+    int    	 y1,
+    int    	 y2,
     int		 *pOverlap);
 
 static pixman_bool_t
 pixman_op(
-    pixman_region16_t *newReg,		    /* Place to store result	     */
-    pixman_region16_t *       reg1,		    /* First region in operation     */
-    pixman_region16_t *       reg2,		    /* 2d region in operation        */
+    region_type_t *newReg,		    /* Place to store result	     */
+    region_type_t *       reg1,		    /* First region in operation     */
+    region_type_t *       reg2,		    /* 2d region in operation        */
     OverlapProcPtr  overlapFunc,            /* Function to call for over-
 					     * lapping bands		     */
     int	    appendNon1,		    /* Append non-overlapping bands  */
@@ -647,21 +631,21 @@ pixman_op(
 					    /* in region 2 ? */
     int	    *pOverlap)
 {
-    pixman_box16_t * r1;			    /* Pointer into first region     */
-    pixman_box16_t * r2;			    /* Pointer into 2d region	     */
-    pixman_box16_t *	    r1End;		    /* End of 1st region	     */
-    pixman_box16_t *	    r2End;		    /* End of 2d region		     */
-    short	    ybot;		    /* Bottom of intersection	     */
-    short	    ytop;		    /* Top of intersection	     */
-    pixman_region16_data_t *	    oldData;		    /* Old data for newReg	     */
+    box_type_t * r1;			    /* Pointer into first region     */
+    box_type_t * r2;			    /* Pointer into 2d region	     */
+    box_type_t *	    r1End;		    /* End of 1st region	     */
+    box_type_t *	    r2End;		    /* End of 2d region		     */
+    int	    ybot;		    /* Bottom of intersection	     */
+    int	    ytop;		    /* Top of intersection	     */
+    region_data_type_t *	    oldData;		    /* Old data for newReg	     */
     int		    prevBand;		    /* Index of start of
 					     * previous band in newReg       */
     int		    curBand;		    /* Index of start of current
 					     * band in newReg		     */
-    pixman_box16_t * r1BandEnd;		    /* End of current band in r1     */
-    pixman_box16_t * r2BandEnd;		    /* End of current band in r2     */
-    short	    top;		    /* Top of non-overlapping band   */
-    short	    bot;		    /* Bottom of non-overlapping band*/
+    box_type_t * r1BandEnd;		    /* End of current band in r1     */
+    box_type_t * r2BandEnd;		    /* End of current band in r2     */
+    int	    top;		    /* Top of non-overlapping band   */
+    int	    bot;		    /* Bottom of non-overlapping band*/
     int    r1y1;		    /* Temps for r1->y1 and r2->y1   */
     int    r2y1;
     int		    newSize;
@@ -690,7 +674,7 @@ pixman_op(
     assert(r1 != r1End);
     assert(r2 != r2End);
 
-    oldData = (pixman_region16_data_t *)NULL;
+    oldData = (region_data_type_t *)NULL;
     if (((newReg == reg1) && (newSize > 1)) ||
 	((newReg == reg2) && (numRects > 1)))
     {
@@ -848,7 +832,7 @@ pixman_op(
     {
 	newReg->extents = *PIXREGION_BOXPTR(newReg);
 	freeData(newReg);
-	newReg->data = (pixman_region16_data_t *)NULL;
+	newReg->data = (region_data_type_t *)NULL;
     }
     else
     {
@@ -874,9 +858,9 @@ pixman_op(
  *-----------------------------------------------------------------------
  */
 static void
-pixman_set_extents (pixman_region16_t *region)
+pixman_set_extents (region_type_t *region)
 {
-    pixman_box16_t *box, *boxEnd;
+    box_type_t *box, *boxEnd;
 
     if (!region->data)
 	return;
@@ -932,18 +916,18 @@ pixman_set_extents (pixman_region16_t *region)
  */
 /*ARGSUSED*/
 static pixman_bool_t
-pixman_region_intersectO (pixman_region16_t *region,
-			  pixman_box16_t    *r1,
-			  pixman_box16_t    *r1End,
-			  pixman_box16_t    *r2,
-			  pixman_box16_t    *r2End,
-			  short    	     y1,
-			  short    	     y2,
+pixman_region_intersectO (region_type_t *region,
+			  box_type_t    *r1,
+			  box_type_t    *r1End,
+			  box_type_t    *r2,
+			  box_type_t    *r2End,
+			  int    	     y1,
+			  int    	     y2,
 			  int		    *pOverlap)
 {
     int  	x1;
     int  	x2;
-    pixman_box16_t *	pNextRect;
+    box_type_t *	pNextRect;
 
     pNextRect = PIXREGION_TOP(region);
 
@@ -977,10 +961,10 @@ pixman_region_intersectO (pixman_region16_t *region,
     return TRUE;
 }
 
-pixman_bool_t
-pixman_region_intersect (pixman_region16_t * 	newReg,
-			 pixman_region16_t * 	reg1,
-			 pixman_region16_t *	reg2)
+PIXMAN_EXPORT pixman_bool_t
+PREFIX(_intersect) (region_type_t * 	newReg,
+			 region_type_t * 	reg1,
+			 region_type_t *	reg2)
 {
     good(reg1);
     good(reg2);
@@ -1009,19 +993,19 @@ pixman_region_intersect (pixman_region16_t * 	newReg,
 	newReg->extents.x2 = MIN(reg1->extents.x2, reg2->extents.x2);
 	newReg->extents.y2 = MIN(reg1->extents.y2, reg2->extents.y2);
 	freeData(newReg);
-	newReg->data = (pixman_region16_data_t *)NULL;
+	newReg->data = (region_data_type_t *)NULL;
     }
     else if (!reg2->data && SUBSUMES(&reg2->extents, &reg1->extents))
     {
-	return pixman_region_copy(newReg, reg1);
+	return PREFIX(_copy) (newReg, reg1);
     }
     else if (!reg1->data && SUBSUMES(&reg1->extents, &reg2->extents))
     {
-	return pixman_region_copy(newReg, reg2);
+	return PREFIX(_copy) (newReg, reg2);
     }
     else if (reg1 == reg2)
     {
-	return pixman_region_copy(newReg, reg1);
+	return PREFIX(_copy) (newReg, reg1);
     }
     else
     {
@@ -1073,16 +1057,16 @@ pixman_region_intersect (pixman_region16_t * 	newReg,
  */
 static pixman_bool_t
 pixman_region_unionO (
-    pixman_region16_t	 *region,
-    pixman_box16_t *r1,
-    pixman_box16_t *r1End,
-    pixman_box16_t *r2,
-    pixman_box16_t *r2End,
-    short	  y1,
-    short	  y2,
+    region_type_t	 *region,
+    box_type_t *r1,
+    box_type_t *r1End,
+    box_type_t *r2,
+    box_type_t *r2End,
+    int	  y1,
+    int	  y2,
     int		  *pOverlap)
 {
-    pixman_box16_t *     pNextRect;
+    box_type_t *     pNextRect;
     int        x1;     /* left and right side of current union */
     int        x2;
 
@@ -1134,29 +1118,29 @@ pixman_region_unionO (
 /* Convenience function for performing union of region with a
  * single rectangle
  */
-pixman_bool_t
-pixman_region_union_rect (pixman_region16_t *dest,
-			  pixman_region16_t *source,
+PIXMAN_EXPORT pixman_bool_t
+PREFIX(_union_rect) (region_type_t *dest,
+			  region_type_t *source,
 			  int x, int y,
 			  unsigned int width, unsigned int height)
 {
-    pixman_region16_t region;
+    region_type_t region;
 
     if (!width || !height)
-	return pixman_region_copy (dest, source);
+	return PREFIX(_copy) (dest, source);
     region.data = NULL;
     region.extents.x1 = x;
     region.extents.y1 = y;
     region.extents.x2 = x + width;
     region.extents.y2 = y + height;
 
-    return pixman_region_union (dest, source, &region);
+    return PREFIX(_union) (dest, source, &region);
 }
 
-pixman_bool_t
-pixman_region_union (pixman_region16_t *newReg,
-		     pixman_region16_t *reg1,
-		     pixman_region16_t *reg2)
+PIXMAN_EXPORT pixman_bool_t
+PREFIX(_union) (region_type_t *newReg,
+		     region_type_t *reg1,
+		     region_type_t *reg2)
 {
     int overlap; /* result ignored */
 
@@ -1173,7 +1157,7 @@ pixman_region_union (pixman_region16_t *newReg,
      */
     if (reg1 == reg2)
     {
-	return pixman_region_copy(newReg, reg1);
+	return PREFIX(_copy) (newReg, reg1);
     }
 
     /*
@@ -1184,7 +1168,7 @@ pixman_region_union (pixman_region16_t *newReg,
 	if (PIXREGION_NAR(reg1))
 	    return pixman_break (newReg);
         if (newReg != reg2)
-	    return pixman_region_copy(newReg, reg2);
+	    return PREFIX(_copy) (newReg, reg2);
         return TRUE;
     }
 
@@ -1196,7 +1180,7 @@ pixman_region_union (pixman_region16_t *newReg,
 	if (PIXREGION_NAR(reg2))
 	    return pixman_break (newReg);
         if (newReg != reg1)
-	    return pixman_region_copy(newReg, reg1);
+	    return PREFIX(_copy) (newReg, reg1);
         return TRUE;
     }
 
@@ -1206,7 +1190,7 @@ pixman_region_union (pixman_region16_t *newReg,
     if (!reg1->data && SUBSUMES(&reg1->extents, &reg2->extents))
     {
         if (newReg != reg1)
-	    return pixman_region_copy(newReg, reg1);
+	    return PREFIX(_copy) (newReg, reg1);
         return TRUE;
     }
 
@@ -1216,7 +1200,7 @@ pixman_region_union (pixman_region16_t *newReg,
     if (!reg2->data && SUBSUMES(&reg2->extents, &reg1->extents))
     {
         if (newReg != reg2)
-	    return pixman_region_copy(newReg, reg2);
+	    return PREFIX(_copy) (newReg, reg2);
         return TRUE;
     }
 
@@ -1235,112 +1219,9 @@ pixman_region_union (pixman_region16_t *newReg,
  *	    Batch Rectangle Union
  *====================================================================*/
 
-/*-
- *-----------------------------------------------------------------------
- * pixman_region_append --
- *
- *      "Append" the rgn rectangles onto the end of dstrgn, maintaining
- *      knowledge of YX-banding when it's easy.  Otherwise, dstrgn just
- *      becomes a non-y-x-banded random collection of rectangles, and not
- *      yet a true region.  After a sequence of appends, the caller must
- *      call pixman_region_validate to ensure that a valid region is
- *      constructed.
- *
- * Results:
- *	TRUE if successful.
- *
- * Side Effects:
- *      dstrgn is modified if rgn has rectangles.
- *
- */
-pixman_bool_t
-pixman_region_append (pixman_region16_t * dstrgn,
-		      pixman_region16_t * rgn)
-{
-    int numRects, dnumRects, size;
-    pixman_box16_t *new, *old;
-    int prepend;
-
-    if (PIXREGION_NAR(rgn))
-	return pixman_break (dstrgn);
-
-    if (!rgn->data && (dstrgn->data == pixman_region_emptyData))
-    {
-	dstrgn->extents = rgn->extents;
-	dstrgn->data = (pixman_region16_data_t *)NULL;
-	return TRUE;
-    }
-
-    numRects = PIXREGION_NUM_RECTS(rgn);
-    if (!numRects)
-	return TRUE;
-    prepend = FALSE;
-    size = numRects;
-    dnumRects = PIXREGION_NUM_RECTS(dstrgn);
-    if (!dnumRects && (size < 200))
-	size = 200; /* XXX pick numbers out of a hat */
-    RECTALLOC(dstrgn, size);
-    old = PIXREGION_RECTS(rgn);
-    if (!dnumRects)
-	dstrgn->extents = rgn->extents;
-    else if (dstrgn->extents.x2 > dstrgn->extents.x1)
-    {
-	pixman_box16_t *first, *last;
-
-	first = old;
-	last = PIXREGION_BOXPTR(dstrgn) + (dnumRects - 1);
-	if ((first->y1 > last->y2) ||
-	    ((first->y1 == last->y1) && (first->y2 == last->y2) &&
-	     (first->x1 > last->x2)))
-	{
-	    if (rgn->extents.x1 < dstrgn->extents.x1)
-		dstrgn->extents.x1 = rgn->extents.x1;
-	    if (rgn->extents.x2 > dstrgn->extents.x2)
-		dstrgn->extents.x2 = rgn->extents.x2;
-	    dstrgn->extents.y2 = rgn->extents.y2;
-	}
-	else
-	{
-	    first = PIXREGION_BOXPTR(dstrgn);
-	    last = old + (numRects - 1);
-	    if ((first->y1 > last->y2) ||
-		((first->y1 == last->y1) && (first->y2 == last->y2) &&
-		 (first->x1 > last->x2)))
-	    {
-		prepend = TRUE;
-		if (rgn->extents.x1 < dstrgn->extents.x1)
-		    dstrgn->extents.x1 = rgn->extents.x1;
-		if (rgn->extents.x2 > dstrgn->extents.x2)
-		    dstrgn->extents.x2 = rgn->extents.x2;
-		dstrgn->extents.y1 = rgn->extents.y1;
-	    }
-	    else
-		dstrgn->extents.x2 = dstrgn->extents.x1;
-	}
-    }
-    if (prepend)
-    {
-	new = PIXREGION_BOX(dstrgn, numRects);
-	if (dnumRects == 1)
-	    *new = *PIXREGION_BOXPTR(dstrgn);
-	else
-	    memmove((char *)new,(char *)PIXREGION_BOXPTR(dstrgn),
-		  dnumRects * sizeof(pixman_box16_t));
-	new = PIXREGION_BOXPTR(dstrgn);
-    }
-    else
-	new = PIXREGION_BOXPTR(dstrgn) + dnumRects;
-    if (numRects == 1)
-	*new = *old;
-    else
-	memmove((char *)new, (char *)old, numRects * sizeof(pixman_box16_t));
-    dstrgn->data->numRects += numRects;
-    return TRUE;
-}
-
 #define ExchangeRects(a, b) \
 {			    \
-    pixman_box16_t     t;	    \
+    box_type_t     t;	    \
     t = rects[a];	    \
     rects[a] = rects[b];    \
     rects[b] = t;	    \
@@ -1348,13 +1229,13 @@ pixman_region_append (pixman_region16_t * dstrgn,
 
 static void
 QuickSortRects(
-    pixman_box16_t     rects[],
+    box_type_t     rects[],
     int        numRects)
 {
     int	y1;
     int	x1;
     int        i, j;
-    pixman_box16_t *r;
+    box_type_t *r;
 
     /* Always called with numRects > 1 */
 
@@ -1438,13 +1319,13 @@ QuickSortRects(
  *-----------------------------------------------------------------------
  */
 
-pixman_bool_t
-pixman_region_validate(pixman_region16_t * badreg,
-		       int *pOverlap)
+static pixman_bool_t
+validate (region_type_t * badreg,
+	  int *pOverlap)
 {
     /* Descriptor for regions under construction  in Step 2. */
     typedef struct {
-	pixman_region16_t   reg;
+	region_type_t   reg;
 	int	    prevBand;
 	int	    curBand;
     } RegionInfo;
@@ -1456,10 +1337,10 @@ pixman_region_validate(pixman_region16_t * badreg,
 	     int	i;	    /* Index into rects			    */
     int	j;	    /* Index into ri			    */
     RegionInfo *rit;       /* &ri[j]				    */
-    pixman_region16_t *  reg;        /* ri[j].reg			    */
-    pixman_box16_t *	box;	    /* Current box in rects		    */
-    pixman_box16_t *	riBox;      /* Last box in ri[j].reg		    */
-    pixman_region16_t *  hreg;       /* ri[j_half].reg			    */
+    region_type_t *  reg;        /* ri[j].reg			    */
+    box_type_t *	box;	    /* Current box in rects		    */
+    box_type_t *	riBox;      /* Last box in ri[j].reg		    */
+    region_type_t *  hreg;       /* ri[j_half].reg			    */
     pixman_bool_t ret = TRUE;
 
     *pOverlap = FALSE;
@@ -1481,7 +1362,7 @@ pixman_region_validate(pixman_region16_t * badreg,
 	if ((numRects) == 1)
 	{
 	    freeData(badreg);
-	    badreg->data = (pixman_region16_data_t *) NULL;
+	    badreg->data = (region_data_type_t *) NULL;
 	}
 	else
 	{
@@ -1580,7 +1461,7 @@ pixman_region_validate(pixman_region16_t * badreg,
 	rit->prevBand = 0;
 	rit->curBand = 0;
 	rit->reg.extents = *box;
-	rit->reg.data = (pixman_region16_data_t *)NULL;
+	rit->reg.data = (region_data_type_t *)NULL;
 	if (!pixman_rect_alloc(&rit->reg, (i+numRI) / numRI)) /* MUST force allocation */
 	    goto bail;
 NextRect: ;
@@ -1599,7 +1480,7 @@ NextRect: ;
 	if (reg->data->numRects == 1) /* keep unions happy below */
 	{
 	    freeData(reg);
-	    reg->data = (pixman_region16_data_t *)NULL;
+	    reg->data = (region_data_type_t *)NULL;
 	}
     }
 
@@ -1660,16 +1541,16 @@ bail:
 /*ARGSUSED*/
 static pixman_bool_t
 pixman_region_subtractO (
-    pixman_region16_t *	region,
-    pixman_box16_t *	r1,
-    pixman_box16_t *  	  	r1End,
-    pixman_box16_t *	r2,
-    pixman_box16_t *  	  	r2End,
-    short  	y1,
-             short  	y2,
+    region_type_t *	region,
+    box_type_t *	r1,
+    box_type_t *  	  	r1End,
+    box_type_t *	r2,
+    box_type_t *  	  	r2End,
+    int  	y1,
+    int  	y2,
     int		*pOverlap)
 {
-    pixman_box16_t *	pNextRect;
+    box_type_t *	pNextRect;
     int  	x1;
 
     x1 = r1->x1;
@@ -1781,10 +1662,10 @@ pixman_region_subtractO (
  *
  *-----------------------------------------------------------------------
  */
-pixman_bool_t
-pixman_region_subtract(pixman_region16_t *	regD,
-		       pixman_region16_t * 	regM,
-		       pixman_region16_t *	regS)
+PIXMAN_EXPORT pixman_bool_t
+PREFIX(_subtract) (region_type_t *	regD,
+		       region_type_t * 	regM,
+		       region_type_t *	regS)
 {
     int overlap; /* result ignored */
 
@@ -1797,7 +1678,7 @@ pixman_region_subtract(pixman_region16_t *	regD,
     {
 	if (PIXREGION_NAR (regS))
 	    return pixman_break (regD);
-	return pixman_region_copy(regD, regM);
+	return PREFIX(_copy) (regD, regM);
     }
     else if (regM == regS)
     {
@@ -1846,11 +1727,11 @@ pixman_region_subtract(pixman_region16_t *	regD,
  *-----------------------------------------------------------------------
  */
 pixman_bool_t
-pixman_region_inverse(pixman_region16_t * 	  newReg,       /* Destination region */
-		      pixman_region16_t * 	  reg1,         /* Region to invert */
-		      pixman_box16_t *     	  invRect) 	/* Bounding box for inversion */
+PIXMAN_EXPORT PREFIX(_inverse) (region_type_t * 	  newReg,       /* Destination region */
+		      region_type_t * 	  reg1,         /* Region to invert */
+		      box_type_t *     	  invRect) 	/* Bounding box for inversion */
 {
-    pixman_region16_t	  invReg;   	/* Quick and dirty region made from the
+    region_type_t	  invReg;   	/* Quick and dirty region made from the
 				 * bounding box */
     int	  overlap;	/* result ignored */
 
@@ -1863,7 +1744,7 @@ pixman_region_inverse(pixman_region16_t * 	  newReg,       /* Destination region
 	    return pixman_break (newReg);
 	newReg->extents = *invRect;
 	freeData(newReg);
-	newReg->data = (pixman_region16_data_t *)NULL;
+	newReg->data = (region_data_type_t *)NULL;
         return TRUE;
     }
 
@@ -1871,7 +1752,7 @@ pixman_region_inverse(pixman_region16_t * 	  newReg,       /* Destination region
        do yucky substraction for overlaps, and
        just throw away rectangles in region 2 that aren't in region 1 */
     invReg.extents = *invRect;
-    invReg.data = (pixman_region16_data_t *)NULL;
+    invReg.data = (region_data_type_t *)NULL;
     if (!pixman_op(newReg, &invReg, reg1, pixman_region_subtractO, TRUE, FALSE, &overlap))
 	return FALSE;
 
@@ -1905,13 +1786,13 @@ pixman_region_inverse(pixman_region16_t * 	  newReg,       /* Destination region
  */
 
 pixman_region_overlap_t
-pixman_region_contains_rectangle(pixman_region16_t *  region,
-				 pixman_box16_t *     prect)
+PIXMAN_EXPORT PREFIX(_contains_rectangle) (region_type_t *  region,
+				 box_type_t *     prect)
 {
     int	x;
     int	y;
-    pixman_box16_t *     pbox;
-    pixman_box16_t *     pboxEnd;
+    box_type_t *     pbox;
+    box_type_t *     pboxEnd;
     int			partIn, partOut;
     int			numRects;
 
@@ -2005,16 +1886,16 @@ pixman_region_contains_rectangle(pixman_region16_t *  region,
     }
 }
 
-/* pixman_region_translate (region, x, y)
+/* PREFIX(_translate) (region, x, y)
    translates in place
 */
 
-void
-pixman_region_translate (pixman_region16_t * region, int x, int y)
+PIXMAN_EXPORT void
+PREFIX(_translate) (region_type_t * region, int x, int y)
 {
     int x1, x2, y1, y2;
     int nbox;
-    pixman_box16_t * pbox;
+    box_type_t * pbox;
 
     good(region);
     region->extents.x1 = x1 = region->extents.x1 + x;
@@ -2053,7 +1934,7 @@ pixman_region_translate (pixman_region16_t * region, int x, int y)
 	region->extents.y2 = SHRT_MAX;
     if (region->data && (nbox = region->data->numRects))
     {
-	pixman_box16_t * pboxout;
+	box_type_t * pboxout;
 
 	for (pboxout = pbox = PIXREGION_BOXPTR(region); nbox--; pbox++)
 	{
@@ -2083,7 +1964,7 @@ pixman_region_translate (pixman_region16_t * region, int x, int y)
 	    {
 		region->extents = *PIXREGION_BOXPTR(region);
 		freeData(region);
-		region->data = (pixman_region16_data_t *)NULL;
+		region->data = (region_data_type_t *)NULL;
 	    }
 	    else
 		pixman_set_extents(region);
@@ -2091,53 +1972,24 @@ pixman_region_translate (pixman_region16_t * region, int x, int y)
     }
 }
 
-/* XXX: Do we need this?
-static pixman_bool_t
-pixman_region16_data_copy(pixman_region16_t * dst, pixman_region16_t * src)
-{
-    good(dst);
-    good(src);
-    if (dst->data)
-	return TRUE;
-    if (dst == src)
-	return TRUE;
-    if (!src->data || !src->data->size)
-    {
-	freeData(dst);
-	dst->data = (pixman_region16_data_t *)NULL;
-	return TRUE;
-    }
-    if (!dst->data || (dst->data->size < src->data->numRects))
-    {
-	freeData(dst);
-	dst->data = allocData(src->data->numRects);
-	if (!dst->data)
-	    return pixman_break (dst);
-    }
-    dst->data->size = src->data->size;
-    dst->data->numRects = src->data->numRects;
-    return TRUE;
-}
-*/
-
-void
-pixman_region_reset(pixman_region16_t *region, pixman_box16_t *box)
+PIXMAN_EXPORT void
+PREFIX(_reset) (region_type_t *region, box_type_t *box)
 {
     good(region);
     assert(box->x1<=box->x2);
     assert(box->y1<=box->y2);
     region->extents = *box;
     freeData(region);
-    region->data = (pixman_region16_data_t *)NULL;
+    region->data = (region_data_type_t *)NULL;
 }
 
 /* box is "return" value */
-int
-pixman_region_contains_point(pixman_region16_t * region,
+PIXMAN_EXPORT int
+PREFIX(_contains_point) (region_type_t * region,
 			     int x, int y,
-			     pixman_box16_t * box)
+			     box_type_t * box)
 {
-    pixman_box16_t *pbox, *pboxEnd;
+    box_type_t *pbox, *pboxEnd;
     int numRects;
 
     good(region);
@@ -2165,150 +2017,20 @@ pixman_region_contains_point(pixman_region16_t * region,
     return(FALSE);
 }
 
-int
-pixman_region_not_empty(pixman_region16_t * region)
+PIXMAN_EXPORT int
+PREFIX(_not_empty) (region_type_t * region)
 {
     good(region);
     return(!PIXREGION_NIL(region));
 }
 
-/* XXX: Do we need this?
-static int
-pixman_region16_broken(pixman_region16_t * region)
-{
-    good(region);
-    return (PIXREGION_NAR(region));
-}
-*/
-
-void
-pixman_region_empty(pixman_region16_t * region)
-{
-    good(region);
-    freeData(region);
-    region->extents.x2 = region->extents.x1;
-    region->extents.y2 = region->extents.y1;
-    region->data = pixman_region_emptyData;
-}
-
-pixman_box16_t *
-pixman_region_extents(pixman_region16_t * region)
+PIXMAN_EXPORT box_type_t *
+PREFIX(_extents) (region_type_t * region)
 {
     good(region);
     return(&region->extents);
 }
 
-#define ExchangeSpans(a, b)				    \
-{							    \
-    pixman_region16_point_t tpt;					    \
-    int    tw;						    \
-							    \
-    tpt = spans[a]; spans[a] = spans[b]; spans[b] = tpt;    \
-    tw = widths[a]; widths[a] = widths[b]; widths[b] = tw;  \
-}
-
-/* ||| I should apply the merge sort code to rectangle sorting above, and see
-   if mapping time can be improved.  But right now I've been at work 12 hours,
-   so forget it.
-*/
-
-static void QuickSortSpans(
-    pixman_region16_point_t spans[],
-    int	    widths[],
-    int	    numSpans)
-{
-    int	    y;
-    int	    i, j, m;
-    pixman_region16_point_t *r;
-
-    /* Always called with numSpans > 1 */
-    /* Sorts only by y, doesn't bother to sort by x */
-
-    do
-    {
-	if (numSpans < 9)
-	{
-	    /* Do insertion sort */
-	    int yprev;
-
-	    yprev = spans[0].y;
-	    i = 1;
-	    do
-	    { /* while i != numSpans */
-		y = spans[i].y;
-		if (yprev > y)
-		{
-		    /* spans[i] is out of order.  Move into proper location. */
-		    pixman_region16_point_t tpt;
-		    int	    tw, k;
-
-		    for (j = 0; y >= spans[j].y; j++) {}
-		    tpt = spans[i];
-		    tw  = widths[i];
-		    for (k = i; k != j; k--)
-		    {
-			spans[k] = spans[k-1];
-			widths[k] = widths[k-1];
-		    }
-		    spans[j] = tpt;
-		    widths[j] = tw;
-		    y = spans[i].y;
-		} /* if out of order */
-		yprev = y;
-		i++;
-	    } while (i != numSpans);
-	    return;
-	}
-
-	/* Choose partition element, stick in location 0 */
-	m = numSpans / 2;
-	if (spans[m].y > spans[0].y)		ExchangeSpans(m, 0);
-	if (spans[m].y > spans[numSpans-1].y)   ExchangeSpans(m, numSpans-1);
-	if (spans[m].y > spans[0].y)		ExchangeSpans(m, 0);
-	y = spans[0].y;
-
-        /* Partition array */
-        i = 0;
-        j = numSpans;
-        do
-	{
-	    r = &(spans[i]);
-	    do
-	    {
-		r++;
-		i++;
-            } while (i != numSpans && r->y < y);
-	    r = &(spans[j]);
-	    do
-	    {
-		r--;
-		j--;
-            } while (y < r->y);
-            if (i < j)
-		ExchangeSpans(i, j);
-        } while (i < j);
-
-        /* Move partition element back to middle */
-        ExchangeSpans(0, j);
-
-	/* Recurse */
-        if (numSpans-j-1 > 1)
-	    QuickSortSpans(&spans[j+1], &widths[j+1], numSpans-j-1);
-        numSpans = j;
-    } while (numSpans > 1);
-}
-
-#define NextBand()						    \
-{								    \
-    clipy1 = pboxBandStart->y1;					    \
-    clipy2 = pboxBandStart->y2;					    \
-    pboxBandEnd = pboxBandStart + 1;				    \
-    while (pboxBandEnd != pboxLast && pboxBandEnd->y1 == clipy1) {  \
-	pboxBandEnd++;						    \
-    }								    \
-    for (; ppt != pptLast && ppt->y < clipy1; ppt++, pwidth++) {} \
-}
-
 /*
     Clip a list of scanlines to a region.  The caller has allocated the
     space.  FSorted is non-zero if the scanline origins are in ascending
@@ -2316,158 +2038,9 @@ static void QuickSortSpans(
     returns the number of new, clipped scanlines.
 */
 
-#ifdef XXX_DO_WE_NEED_THIS
-static int
-pixman_region16_clip_spans(
-    pixman_region16_t 		*prgnDst,
-    pixman_region16_point_t 	*ppt,
-    int	    		*pwidth,
-    int			nspans,
-    pixman_region16_point_t 	*pptNew,
-    int			*pwidthNew,
-    int			fSorted)
-{
-    pixman_region16_point_t 	*pptLast;
-    int			*pwidthNewStart;	/* the vengeance of Xerox! */
-    int	y, x1, x2;
-    int	numRects;
-
-    good(prgnDst);
-    pptLast = ppt + nspans;
-    pwidthNewStart = pwidthNew;
-
-    if (!prgnDst->data)
-    {
-	/* Do special fast code with clip boundaries in registers(?) */
-	/* It doesn't pay much to make use of fSorted in this case,
-	   so we lump everything together. */
-
-	   int clipx1, clipx2, clipy1, clipy2;
-
-	clipx1 = prgnDst->extents.x1;
-	clipy1 = prgnDst->extents.y1;
-	clipx2 = prgnDst->extents.x2;
-	clipy2 = prgnDst->extents.y2;
-
-	for (; ppt != pptLast; ppt++, pwidth++)
-	{
-	    y = ppt->y;
-	    x1 = ppt->x;
-	    if (clipy1 <= y && y < clipy2)
-	    {
-		x2 = x1 + *pwidth;
-		if (x1 < clipx1)    x1 = clipx1;
-		if (x2 > clipx2)    x2 = clipx2;
-		if (x1 < x2)
-		{
-		    /* part of span in clip rectangle */
-		    pptNew->x = x1;
-		    pptNew->y = y;
-		    *pwidthNew = x2 - x1;
-		    pptNew++;
-		    pwidthNew++;
-		}
-	    }
-	} /* end for */
-
-    }
-    else if ((numRects = prgnDst->data->numRects))
-    {
-	/* Have to clip against many boxes */
-	pixman_box16_t *pboxBandStart, *pboxBandEnd;
-	pixman_box16_t *pbox;
-	pixman_box16_t *pboxLast;
-	int	clipy1, clipy2;
-
-	/* In this case, taking advantage of sorted spans gains more than
-	   the sorting costs. */
-	if ((! fSorted) && (nspans > 1))
-	    QuickSortSpans(ppt, pwidth, nspans);
-
-	pboxBandStart = PIXREGION_BOXPTR(prgnDst);
-	pboxLast = pboxBandStart + numRects;
-
-	NextBand();
-
-	for (; ppt != pptLast; )
-	{
-	    y = ppt->y;
-	    if (y < clipy2)
-	    {
-		/* span is in the current band */
-		pbox = pboxBandStart;
-		x1 = ppt->x;
-		x2 = x1 + *pwidth;
-		do
-		{ /* For each box in band */
-		    int    newx1, newx2;
-
-		    newx1 = x1;
-		    newx2 = x2;
-		    if (newx1 < pbox->x1)   newx1 = pbox->x1;
-		    if (newx2 > pbox->x2)   newx2 = pbox->x2;
-		    if (newx1 < newx2)
-		    {
-			/* Part of span in clip rectangle */
-			pptNew->x = newx1;
-			pptNew->y = y;
-			*pwidthNew = newx2 - newx1;
-			pptNew++;
-			pwidthNew++;
-		    }
-		    pbox++;
-		} while (pbox != pboxBandEnd);
-		ppt++;
-		pwidth++;
-	    }
-	    else
-	    {
-		/* Move to next band, adjust ppt as needed */
-		pboxBandStart = pboxBandEnd;
-		if (pboxBandStart == pboxLast)
-		    break; /* We're completely done */
-		NextBand();
-	    }
-	}
-    }
-    return (pwidthNew - pwidthNewStart);
-}
-
-/* find the band in a region with the most rectangles */
-static int
-pixman_region16_find_max_band(pixman_region16_t * prgn)
-{
-    int nbox;
-    pixman_box16_t * pbox;
-    int nThisBand;
-    int nMaxBand = 0;
-    short yThisBand;
-
-    good(prgn);
-    nbox = PIXREGION_NUM_RECTS(prgn);
-    pbox = PIXREGION_RECTS(prgn);
-
-    while(nbox > 0)
-    {
-	yThisBand = pbox->y1;
-	nThisBand = 0;
-	while((nbox > 0) && (pbox->y1 == yThisBand))
-	{
-	    nbox--;
-	    pbox++;
-	    nThisBand++;
-	}
-	if (nThisBand > nMaxBand)
-	    nMaxBand = nThisBand;
-    }
-    return (nMaxBand);
-}
-#endif /* XXX_DO_WE_NEED_THIS */
-
-
-pixman_bool_t
-pixman_region_selfcheck (reg)
-    pixman_region16_t * reg;
+PIXMAN_EXPORT pixman_bool_t
+PREFIX(_selfcheck) (reg)
+    region_type_t * reg;
 {
     int i, numRects;
 
@@ -2483,8 +2056,8 @@ pixman_region_selfcheck (reg)
 	return (!reg->data);
     else
     {
-	pixman_box16_t * pboxP, * pboxN;
-	pixman_box16_t box;
+	box_type_t * pboxP, * pboxN;
+	box_type_t box;
 
 	pboxP = PIXREGION_RECTS(reg);
 	box = *pboxP;
@@ -2511,16 +2084,16 @@ pixman_region_selfcheck (reg)
     }
 }
 
-pixman_bool_t
-pixman_region_init_rects (pixman_region16_t *region,
-			  pixman_box16_t *boxes, int count)
+PIXMAN_EXPORT pixman_bool_t
+PREFIX(_init_rects) (region_type_t *region,
+		     box_type_t *boxes, int count)
 {
     int overlap;
 
     /* if it's 1, then we just want to set the extents, so call
      * the existing method. */
     if (count == 1) {
-       pixman_region_init_rect(region,
+       PREFIX(_init_rect) (region,
                                boxes[0].x1,
                                boxes[0].y1,
                                boxes[0].x2 - boxes[0].x1,
@@ -2528,7 +2101,7 @@ pixman_region_init_rects (pixman_region16_t *region,
        return TRUE;
     }
 
-    pixman_region_init(region);
+    PREFIX(_init) (region);
 
     /* if it's 0, don't call pixman_rect_alloc -- 0 rectangles is
      * a special case, and causing pixman_rect_alloc would cause
@@ -2542,10 +2115,10 @@ pixman_region_init_rects (pixman_region16_t *region,
 	return FALSE;
 
     /* Copy in the rects */
-    memcpy (PIXREGION_RECTS(region), boxes, sizeof(pixman_box16_t) * count);
+    memcpy (PIXREGION_RECTS(region), boxes, sizeof(box_type_t) * count);
     region->data->numRects = count;
 
     /* Validate */
     region->extents.x1 = region->extents.x2 = 0;
-    return pixman_region_validate (region, &overlap);
+    return validate (region, &overlap);
 }
diff --git a/lib/pixman/pixman/pixman-region16.c b/lib/pixman/pixman/pixman-region16.c
new file mode 100644
index 000000000..acee0946b
--- /dev/null
+++ b/lib/pixman/pixman/pixman-region16.c
@@ -0,0 +1,83 @@
+/*
+ * Copyright © 2008 Red Hat, Inc.
+ *
+ * Permission to use, copy, modify, distribute, and sell this software
+ * and its documentation for any purpose is hereby granted without
+ * fee, provided that the above copyright notice appear in all copies
+ * and that both that copyright notice and this permission notice
+ * appear in supporting documentation, and that the name of
+ * Red Hat, Inc. not be used in advertising or publicity pertaining to
+ * distribution of the software without specific, written prior
+ * permission. Red Hat, Inc. makes no representations about the
+ * suitability of this software for any purpose.  It is provided "as
+ * is" without express or implied warranty.
+ *
+ * RED HAT, INC. DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS
+ * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
+ * FITNESS, IN NO EVENT SHALL RED HAT, INC. BE LIABLE FOR ANY SPECIAL,
+ * INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER
+ * RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION
+ * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR
+ * IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ *
+ * Author: Soren Sandmann <sandmann@redhat.com>
+ */
+#ifdef HAVE_CONFIG_H
+#include <config.h>
+#endif
+
+#undef PIXMAN_DISABLE_DEPRECATED
+
+#include "pixman-private.h"
+
+#include <stdlib.h>
+
+typedef pixman_box16_t		box_type_t;
+typedef pixman_region16_data_t	region_data_type_t;
+typedef pixman_region16_t	region_type_t;
+
+typedef struct {
+    int x, y;
+} point_type_t;
+
+#define PREFIX(x) pixman_region##x
+
+PIXMAN_EXPORT void
+pixman_region_set_static_pointers (pixman_box16_t *empty_box,
+				   pixman_region16_data_t *empty_data,
+				   pixman_region16_data_t *broken_data)
+{
+    pixman_region_internal_set_static_pointers (empty_box, empty_data, broken_data);
+}
+
+pixman_bool_t
+pixman_region16_copy_from_region32 (pixman_region16_t *dst,
+				    pixman_region32_t *src)
+{
+    int n_boxes, i;
+    pixman_box32_t *boxes32;
+    pixman_box16_t *boxes16;
+    pixman_bool_t retval;
+    
+    boxes32 = pixman_region32_rectangles (src, &n_boxes);
+
+    boxes16 = pixman_malloc_ab (n_boxes, sizeof (pixman_box16_t));
+
+    if (!boxes16)
+	return FALSE;
+    
+    for (i = 0; i < n_boxes; ++i)
+    {
+	boxes16[i].x1 = boxes32[i].x1;
+	boxes16[i].y1 = boxes32[i].y1;
+	boxes16[i].x2 = boxes32[i].x2;
+	boxes16[i].y2 = boxes32[i].y2;
+    }
+
+    pixman_region_fini (dst);
+    retval = pixman_region_init_rects (dst, boxes16, n_boxes);
+    free (boxes16);
+    return retval;
+}
+
+#include "pixman-region.c"
diff --git a/lib/pixman/pixman/pixman-region32.c b/lib/pixman/pixman/pixman-region32.c
new file mode 100644
index 000000000..8a30d1dc7
--- /dev/null
+++ b/lib/pixman/pixman/pixman-region32.c
@@ -0,0 +1,73 @@
+/*
+ * Copyright © 2008 Red Hat, Inc.
+ *
+ * Permission to use, copy, modify, distribute, and sell this software
+ * and its documentation for any purpose is hereby granted without
+ * fee, provided that the above copyright notice appear in all copies
+ * and that both that copyright notice and this permission notice
+ * appear in supporting documentation, and that the name of
+ * Red Hat, Inc. not be used in advertising or publicity pertaining to
+ * distribution of the software without specific, written prior
+ * permission. Red Hat, Inc. makes no representations about the
+ * suitability of this software for any purpose.  It is provided "as
+ * is" without express or implied warranty.
+ *
+ * RED HAT, INC. DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS
+ * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
+ * FITNESS, IN NO EVENT SHALL RED HAT, INC. BE LIABLE FOR ANY SPECIAL,
+ * INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER
+ * RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION
+ * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR
+ * IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ *
+ * Author: Soren Sandmann <sandmann@redhat.com>
+ */
+#ifdef HAVE_CONFIG_H
+#include <config.h>
+#endif
+
+#include "pixman-private.h"
+
+#include <stdlib.h>
+
+typedef pixman_box32_t		box_type_t;
+typedef pixman_region32_data_t	region_data_type_t;
+typedef pixman_region32_t	region_type_t;
+
+typedef struct {
+    int x, y;
+} point_type_t;
+
+#define PREFIX(x) pixman_region32##x
+
+pixman_bool_t
+pixman_region32_copy_from_region16 (pixman_region32_t *dst,
+				    pixman_region16_t *src)
+{
+    int n_boxes, i;
+    pixman_box16_t *boxes16;
+    pixman_box32_t *boxes32;
+    pixman_bool_t retval;
+    
+    boxes16 = pixman_region_rectangles (src, &n_boxes);
+
+    boxes32 = pixman_malloc_ab (n_boxes, sizeof (pixman_box32_t));
+
+    if (!boxes32)
+	return FALSE;
+    
+    for (i = 0; i < n_boxes; ++i)
+    {
+	boxes32[i].x1 = boxes16[i].x1;
+	boxes32[i].y1 = boxes16[i].y1;
+	boxes32[i].x2 = boxes16[i].x2;
+	boxes32[i].y2 = boxes16[i].y2;
+    }
+
+    pixman_region32_fini (dst);
+    retval = pixman_region32_init_rects (dst, boxes32, n_boxes);
+    free (boxes32);
+    return retval;
+}
+
+#include "pixman-region.c"
diff --git a/lib/pixman/pixman/pixman-source.c b/lib/pixman/pixman/pixman-source.c
index a5a423587..6a640fa67 100644
--- a/lib/pixman/pixman/pixman-source.c
+++ b/lib/pixman/pixman/pixman-source.c
@@ -27,6 +27,7 @@
 #include <config.h>
 #endif
 
+#include <stdlib.h>
 #include <math.h>
 
 #include "pixman-private.h"
@@ -679,3 +680,30 @@ void pixmanFetchSourcePict(source_image_t * pict, int x, int y, int width,
         }
     }
 }
+
+/*
+ * For now, just evaluate the source picture at 32bpp and expand.  We could
+ * produce smoother gradients by evaluating them at higher color depth, but
+ * that's a project for the future.
+ */
+void pixmanFetchSourcePict64(source_image_t * pict, int x, int y, int width,
+                             uint64_t *buffer, uint64_t *mask, uint32_t maskBits)
+{
+    uint32_t *mask8 = NULL;
+
+    // Contract the mask image, if one exists, so that the 32-bit fetch function
+    // can use it.
+    if (mask) {
+        mask8 = pixman_malloc_ab(width, sizeof(uint32_t));
+        pixman_contract(mask8, mask, width);
+    }
+
+    // Fetch the source image into the first half of buffer.
+    pixmanFetchSourcePict(pict, x, y, width, (uint32_t*)buffer, mask8,
+                          maskBits);
+
+    // Expand from 32bpp to 64bpp in place.
+    pixman_expand(buffer, (uint32_t*)buffer, PIXMAN_a8r8g8b8, width);
+
+    free(mask8);
+}
diff --git a/lib/pixman/pixman/pixman-sse.c b/lib/pixman/pixman/pixman-sse.c
deleted file mode 100644
index eb1d3d5b4..000000000
--- a/lib/pixman/pixman/pixman-sse.c
+++ /dev/null
@@ -1,51 +0,0 @@
-/*
- * Copyright © 2008 Rodrigo Kumpera
- *
- * Permission to use, copy, modify, distribute, and sell this software and its
- * documentation for any purpose is hereby granted without fee, provided that
- * the above copyright notice appear in all copies and that both that
- * copyright notice and this permission notice appear in supporting
- * documentation, and that the name of Red Hat not be used in advertising or
- * publicity pertaining to distribution of the software without specific,
- * written prior permission.  Red Hat makes no representations about the
- * suitability of this software for any purpose.  It is provided "as is"
- * without express or implied warranty.
- *
- * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS
- * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
- * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY
- * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
- * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN
- * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
- * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
- * SOFTWARE.
- *
- * Author:  Rodrigo Kumpera (kumpera@gmail.com)
- * 
- */
-#ifdef HAVE_CONFIG_H
-#include <config.h>
-#endif
-
-#include "pixman-sse.h"
-
-#ifdef USE_SSE2
-
-void
-fbComposeSetupSSE(void)
-{
-    static pixman_bool_t initialized = FALSE;
-
-    if (initialized)
-	return;
-    
-    /* check if we have SSE2 support and initialize accordingly */
-    if (pixman_have_sse())
-    {
-    }
-
-    initialized = TRUE;
-}
-
-
-#endif /* USE_SSE2 */
diff --git a/lib/pixman/pixman/pixman-sse.h b/lib/pixman/pixman/pixman-sse.h
deleted file mode 100644
index 65cf21d5b..000000000
--- a/lib/pixman/pixman/pixman-sse.h
+++ /dev/null
@@ -1,53 +0,0 @@
-/*
- * Copyright © 2008 Rodrigo Kumpera
- *
- * Permission to use, copy, modify, distribute, and sell this software and its
- * documentation for any purpose is hereby granted without fee, provided that
- * the above copyright notice appear in all copies and that both that
- * copyright notice and this permission notice appear in supporting
- * documentation, and that the name of Red Hat not be used in advertising or
- * publicity pertaining to distribution of the software without specific,
- * written prior permission.  Red Hat makes no representations about the
- * suitability of this software for any purpose.  It is provided "as is"
- * without express or implied warranty.
- *
- * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS
- * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
- * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY
- * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
- * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN
- * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
- * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
- * SOFTWARE.
- *
- * Author:  Rodrigo Kumpera (kumpera@gmail.com)
- * 
- */
-#ifndef _PIXMAN_SSE_H_
-#define _PIXMAN_SSE_H_
-
-#ifdef HAVE_DIX_CONFIG_H
-#include <dix-config.h>
-#endif
-
-#include "pixman-private.h"
-
-#ifdef USE_SSE2
-
-#if !defined(__amd64__) && !defined(__x86_64__)
-pixman_bool_t pixman_have_sse(void);
-#else
-#define pixman_have_sse() TRUE
-#endif
-
-#else
-#define pixman_have_sse() FALSE
-#endif
-
-#ifdef USE_SSE2
-
-void fbComposeSetupSSE(void);
-
-#endif /* USE_SSE2 */
-
-#endif /* _PIXMAN_SSE_H_ */
diff --git a/lib/pixman/pixman/pixman-sse2.c b/lib/pixman/pixman/pixman-sse2.c
new file mode 100644
index 000000000..0f36436b6
--- /dev/null
+++ b/lib/pixman/pixman/pixman-sse2.c
@@ -0,0 +1,4726 @@
+/*
+ * Copyright © 2008 Rodrigo Kumpera
+ * Copyright © 2008 André Tupinambá
+ *
+ * Permission to use, copy, modify, distribute, and sell this software and its
+ * documentation for any purpose is hereby granted without fee, provided that
+ * the above copyright notice appear in all copies and that both that
+ * copyright notice and this permission notice appear in supporting
+ * documentation, and that the name of Red Hat not be used in advertising or
+ * publicity pertaining to distribution of the software without specific,
+ * written prior permission.  Red Hat makes no representations about the
+ * suitability of this software for any purpose.  It is provided "as is"
+ * without express or implied warranty.
+ *
+ * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS
+ * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
+ * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY
+ * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN
+ * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
+ * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
+ * SOFTWARE.
+ *
+ * Author:  Rodrigo Kumpera (kumpera@gmail.com)
+ *          André Tupinambá (andrelrt@gmail.com)
+ * 
+ * Based on work by Owen Taylor and Søren Sandmann
+ */
+#ifdef HAVE_CONFIG_H
+#include <config.h>
+#endif
+
+#include <mmintrin.h>
+#include <xmmintrin.h> /* for _mm_shuffle_pi16 and _MM_SHUFFLE */
+#include <emmintrin.h> /* for SSE2 intrinsics */
+
+#include "pixman-sse2.h"
+
+#ifdef USE_SSE2
+
+#ifdef _MSC_VER
+#undef inline
+#define inline __forceinline
+#endif
+
+#ifdef __GNUC__
+#    define inline __inline__ __attribute__ ((__always_inline__))
+#endif
+
+/* -------------------------------------------------------------------------------------------------
+ * Locals
+ */
+
+static __m64 xMask0080;
+static __m64 xMask00ff;
+static __m64 xMask0101;
+static __m64 xMaskAlpha;
+
+static __m64 xMask565rgb;
+static __m64 xMask565Unpack;
+
+static __m128i Mask0080;
+static __m128i Mask00ff;
+static __m128i Mask0101;
+static __m128i Maskffff;
+static __m128i Maskff000000;
+static __m128i MaskAlpha;
+
+static __m128i Mask565r;
+static __m128i Mask565g1, Mask565g2;
+static __m128i Mask565b;
+static __m128i MaskRed;
+static __m128i MaskGreen;
+static __m128i MaskBlue;
+
+static __m128i Mask565FixRB;
+static __m128i Mask565FixG;
+
+/* -------------------------------------------------------------------------------------------------
+ * SSE2 Inlines
+ */
+static inline __m128i
+unpack_32_1x128 (uint32_t data)
+{
+    return _mm_unpacklo_epi8 (_mm_cvtsi32_si128 (data), _mm_setzero_si128());
+}
+
+static inline void
+unpack_128_2x128 (__m128i data, __m128i* dataLo, __m128i* dataHi)
+{
+    *dataLo = _mm_unpacklo_epi8 (data, _mm_setzero_si128 ());
+    *dataHi = _mm_unpackhi_epi8 (data, _mm_setzero_si128 ());
+}
+
+static inline __m128i
+unpack565to8888 (__m128i lo)
+{
+    __m128i r, g, b, rb, t;
+    
+    r = _mm_and_si128 (_mm_slli_epi32 (lo, 8), MaskRed);
+    g = _mm_and_si128 (_mm_slli_epi32 (lo, 5), MaskGreen);
+    b = _mm_and_si128 (_mm_slli_epi32 (lo, 3), MaskBlue);
+
+    rb = _mm_or_si128 (r, b);
+    t  = _mm_and_si128 (rb, Mask565FixRB);
+    t  = _mm_srli_epi32 (t, 5);
+    rb = _mm_or_si128 (rb, t);
+
+    t  = _mm_and_si128 (g, Mask565FixG);
+    t  = _mm_srli_epi32 (t, 6);
+    g  = _mm_or_si128 (g, t);
+    
+    return _mm_or_si128 (rb, g);
+}
+
+static inline void
+unpack565_128_4x128 (__m128i data, __m128i* data0, __m128i* data1, __m128i* data2, __m128i* data3)
+{
+    __m128i lo, hi;
+
+    lo = _mm_unpacklo_epi16 (data, _mm_setzero_si128 ());
+    hi = _mm_unpackhi_epi16 (data, _mm_setzero_si128 ());
+
+    lo = unpack565to8888 (lo);
+    hi = unpack565to8888 (hi);
+
+    unpack_128_2x128 (lo, data0, data1);
+    unpack_128_2x128 (hi, data2, data3);
+}
+
+static inline uint16_t
+pack565_32_16 (uint32_t pixel)
+{
+    return (uint16_t) (((pixel>>8) & 0xf800) | ((pixel>>5) & 0x07e0) | ((pixel>>3) & 0x001f));
+}
+
+static inline __m128i
+pack_2x128_128 (__m128i lo, __m128i hi)
+{
+    return _mm_packus_epi16 (lo, hi);
+}
+
+static inline __m128i
+pack565_2x128_128 (__m128i lo, __m128i hi)
+{
+    __m128i data;
+    __m128i r, g1, g2, b;
+
+    data = pack_2x128_128 ( lo, hi );
+
+    r  = _mm_and_si128 (data , Mask565r);
+    g1 = _mm_and_si128 (_mm_slli_epi32 (data , 3), Mask565g1);
+    g2 = _mm_and_si128 (_mm_srli_epi32 (data , 5), Mask565g2);
+    b  = _mm_and_si128 (_mm_srli_epi32 (data , 3), Mask565b);
+
+    return _mm_or_si128 (_mm_or_si128 (_mm_or_si128 (r, g1), g2), b);
+}
+
+static inline __m128i
+pack565_4x128_128 (__m128i* xmm0, __m128i* xmm1, __m128i* xmm2, __m128i* xmm3)
+{
+    return _mm_packus_epi16 (pack565_2x128_128 (*xmm0, *xmm1), pack565_2x128_128 (*xmm2, *xmm3));
+}
+
+static inline uint32_t
+packAlpha (__m128i x)
+{
+    return _mm_cvtsi128_si32 (_mm_packus_epi16 (_mm_packus_epi16 (_mm_srli_epi32 (x, 24),
+                                                                  _mm_setzero_si128 ()),
+                                                _mm_setzero_si128 ()));
+}
+
+static inline __m128i
+expandPixel_32_1x128 (uint32_t data)
+{
+    return _mm_shuffle_epi32 (unpack_32_1x128 (data), _MM_SHUFFLE(1, 0, 1, 0));
+}
+
+static inline __m128i
+expandAlpha_1x128 (__m128i data)
+{
+    return _mm_shufflehi_epi16 (_mm_shufflelo_epi16 (data, _MM_SHUFFLE(3, 3, 3, 3)), _MM_SHUFFLE(3, 3, 3, 3));
+}
+
+static inline void
+expandAlpha_2x128 (__m128i dataLo, __m128i dataHi, __m128i* alphaLo, __m128i* alphaHi)
+{
+    __m128i lo, hi;
+
+    lo = _mm_shufflelo_epi16 (dataLo, _MM_SHUFFLE(3, 3, 3, 3));
+    hi = _mm_shufflelo_epi16 (dataHi, _MM_SHUFFLE(3, 3, 3, 3));
+    *alphaLo = _mm_shufflehi_epi16 (lo, _MM_SHUFFLE(3, 3, 3, 3));
+    *alphaHi = _mm_shufflehi_epi16 (hi, _MM_SHUFFLE(3, 3, 3, 3));
+}
+
+static inline void
+expandAlphaRev_2x128 (__m128i dataLo, __m128i dataHi, __m128i* alphaLo, __m128i* alphaHi)
+{
+    __m128i lo, hi;
+
+    lo = _mm_shufflelo_epi16 (dataLo, _MM_SHUFFLE(0, 0, 0, 0));
+    hi = _mm_shufflelo_epi16 (dataHi, _MM_SHUFFLE(0, 0, 0, 0));
+    *alphaLo = _mm_shufflehi_epi16 (lo, _MM_SHUFFLE(0, 0, 0, 0));
+    *alphaHi = _mm_shufflehi_epi16 (hi, _MM_SHUFFLE(0, 0, 0, 0));
+}
+
+static inline void
+pixMultiply_2x128 (__m128i* dataLo, __m128i* dataHi, __m128i* alphaLo, __m128i* alphaHi, __m128i* retLo, __m128i* retHi)
+{
+    __m128i lo, hi;
+
+    lo = _mm_mullo_epi16 (*dataLo, *alphaLo);
+    hi = _mm_mullo_epi16 (*dataHi, *alphaHi);
+    lo = _mm_adds_epu16 (lo, Mask0080);
+    hi = _mm_adds_epu16 (hi, Mask0080);
+    *retLo = _mm_mulhi_epu16 (lo, Mask0101);
+    *retHi = _mm_mulhi_epu16 (hi, Mask0101);
+}
+
+static inline void
+pixAddMultiply_2x128 (__m128i* srcLo, __m128i* srcHi, __m128i* alphaDstLo, __m128i* alphaDstHi,
+                      __m128i* dstLo, __m128i* dstHi, __m128i* alphaSrcLo, __m128i* alphaSrcHi,
+                      __m128i* retLo, __m128i* retHi)
+{
+    __m128i lo, hi;
+    __m128i mulLo, mulHi;
+
+    lo = _mm_mullo_epi16 (*srcLo, *alphaDstLo);
+    hi = _mm_mullo_epi16 (*srcHi, *alphaDstHi);
+    mulLo = _mm_mullo_epi16 (*dstLo, *alphaSrcLo);
+    mulHi = _mm_mullo_epi16 (*dstHi, *alphaSrcHi);
+    lo = _mm_adds_epu16 (lo, Mask0080);
+    hi = _mm_adds_epu16 (hi, Mask0080);
+    lo = _mm_adds_epu16 (lo, mulLo);
+    hi = _mm_adds_epu16 (hi, mulHi);
+    *retLo = _mm_mulhi_epu16 (lo, Mask0101);
+    *retHi = _mm_mulhi_epu16 (hi, Mask0101);
+}
+
+static inline void
+negate_2x128 (__m128i dataLo, __m128i dataHi, __m128i* negLo, __m128i* negHi)
+{
+    *negLo = _mm_xor_si128 (dataLo, Mask00ff);
+    *negHi = _mm_xor_si128 (dataHi, Mask00ff);
+}
+
+static inline void
+invertColors_2x128 (__m128i dataLo, __m128i dataHi, __m128i* invLo, __m128i* invHi)
+{
+    __m128i lo, hi;
+
+    lo = _mm_shufflelo_epi16 (dataLo, _MM_SHUFFLE(3, 0, 1, 2));
+    hi = _mm_shufflelo_epi16 (dataHi, _MM_SHUFFLE(3, 0, 1, 2));
+    *invLo = _mm_shufflehi_epi16 (lo, _MM_SHUFFLE(3, 0, 1, 2));
+    *invHi = _mm_shufflehi_epi16 (hi, _MM_SHUFFLE(3, 0, 1, 2));
+}
+
+static inline void
+over_2x128 (__m128i* srcLo, __m128i* srcHi, __m128i* alphaLo, __m128i* alphaHi, __m128i* dstLo, __m128i* dstHi)
+{
+    __m128i t1, t2;
+
+    negate_2x128 (*alphaLo, *alphaHi, &t1, &t2);
+
+    pixMultiply_2x128 (dstLo, dstHi, &t1, &t2, dstLo, dstHi);
+
+    *dstLo = _mm_adds_epu8 (*srcLo, *dstLo);
+    *dstHi = _mm_adds_epu8 (*srcHi, *dstHi);
+}
+
+static inline void
+overRevNonPre_2x128 (__m128i srcLo, __m128i srcHi, __m128i* dstLo, __m128i* dstHi)
+{
+    __m128i lo, hi;
+    __m128i alphaLo, alphaHi;
+
+    expandAlpha_2x128 (srcLo, srcHi, &alphaLo, &alphaHi);
+
+    lo = _mm_or_si128 (alphaLo, MaskAlpha);
+    hi = _mm_or_si128 (alphaHi, MaskAlpha);
+
+    invertColors_2x128 (srcLo, srcHi, &srcLo, &srcHi);
+
+    pixMultiply_2x128 (&srcLo, &srcHi, &lo, &hi, &lo, &hi);
+
+    over_2x128 (&lo, &hi, &alphaLo, &alphaHi, dstLo, dstHi);
+}
+
+static inline void
+inOver_2x128 (__m128i* srcLo,  __m128i* srcHi,  __m128i*  alphaLo, __m128i*  alphaHi,
+              __m128i* maskLo, __m128i* maskHi, __m128i* dstLo,   __m128i* dstHi)
+{
+    __m128i sLo, sHi;
+    __m128i aLo, aHi;
+
+    pixMultiply_2x128 (  srcLo,   srcHi, maskLo, maskHi, &sLo, &sHi);
+    pixMultiply_2x128 (alphaLo, alphaHi, maskLo, maskHi, &aLo, &aHi);
+
+    over_2x128 (&sLo, &sHi, &aLo, &aHi, dstLo, dstHi);
+}
+
+static inline void
+cachePrefetch (__m128i* addr)
+{
+    _mm_prefetch (addr, _MM_HINT_T0);
+}
+
+static inline void
+cachePrefetchNext (__m128i* addr)
+{
+    _mm_prefetch (addr + 4, _MM_HINT_T0); // 64 bytes ahead
+}
+
+/* load 4 pixels from a 16-byte boundary aligned address */
+static inline __m128i
+load128Aligned (__m128i* src)
+{
+    return _mm_load_si128 (src);
+}
+
+/* load 4 pixels from a unaligned address */
+static inline __m128i
+load128Unaligned (__m128i* src)
+{
+    return _mm_loadu_si128 (src);
+}
+
+/* save 4 pixels using Write Combining memory on a 16-byte boundary aligned address */
+static inline void
+save128WriteCombining (__m128i* dst, __m128i data)
+{
+    _mm_stream_si128 (dst, data);
+}
+
+/* save 4 pixels on a 16-byte boundary aligned address */
+static inline void
+save128Aligned (__m128i* dst, __m128i data)
+{
+    _mm_store_si128 (dst, data);
+}
+
+/* save 4 pixels on a unaligned address */
+static inline void
+save128Unaligned (__m128i* dst, __m128i data)
+{
+    _mm_storeu_si128 (dst, data);
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * MMX inlines
+ */
+
+static inline __m64
+unpack_32_1x64 (uint32_t data)
+{
+    return _mm_unpacklo_pi8 (_mm_cvtsi32_si64 (data), _mm_setzero_si64());
+}
+
+static inline __m64
+expandAlpha_1x64 (__m64 data)
+{
+    return _mm_shuffle_pi16 (data, _MM_SHUFFLE(3, 3, 3, 3));
+}
+
+static inline __m64
+expandAlphaRev_1x64 (__m64 data)
+{
+    return _mm_shuffle_pi16 (data, _MM_SHUFFLE(0, 0, 0, 0));
+}
+
+static inline __m64
+expandPixel_8_1x64 (uint8_t data)
+{
+    return _mm_shuffle_pi16 (unpack_32_1x64 ((uint32_t)data), _MM_SHUFFLE(0, 0, 0, 0));
+}
+
+static inline __m64
+pixMultiply_1x64 (__m64 data, __m64 alpha)
+{
+    return _mm_mulhi_pu16 (_mm_adds_pu16 (_mm_mullo_pi16 (data, alpha),
+                                          xMask0080),
+                           xMask0101);
+}
+
+static inline __m64
+pixAddMultiply_1x64 (__m64* src, __m64* alphaDst, __m64* dst, __m64* alphaSrc)
+{
+    return _mm_mulhi_pu16 (_mm_adds_pu16 (_mm_adds_pu16 (_mm_mullo_pi16 (*src, *alphaDst),
+                                                         xMask0080),
+                                          _mm_mullo_pi16 (*dst, *alphaSrc)),
+                           xMask0101);
+}
+
+static inline __m64
+negate_1x64 (__m64 data)
+{
+    return _mm_xor_si64 (data, xMask00ff);
+}
+
+static inline __m64
+invertColors_1x64 (__m64 data)
+{
+    return _mm_shuffle_pi16 (data, _MM_SHUFFLE(3, 0, 1, 2));
+}
+
+static inline __m64
+over_1x64 (__m64 src, __m64 alpha, __m64 dst)
+{
+    return _mm_adds_pu8 (src, pixMultiply_1x64 (dst, negate_1x64 (alpha)));
+}
+
+static inline __m64
+inOver_1x64 (__m64* src, __m64* alpha, __m64* mask, __m64* dst)
+{
+    return over_1x64 (pixMultiply_1x64 (*src, *mask),
+                      pixMultiply_1x64 (*alpha, *mask),
+                      *dst);
+}
+
+static inline __m64
+overRevNonPre_1x64 (__m64 src, __m64 dst)
+{
+    __m64 alpha = expandAlpha_1x64 (src);
+
+    return over_1x64 (pixMultiply_1x64 (invertColors_1x64 (src),
+                                        _mm_or_si64 (alpha, xMaskAlpha)),
+                      alpha,
+                      dst);
+}
+
+static inline uint32_t
+pack_1x64_32( __m64 data )
+{
+    return _mm_cvtsi64_si32 (_mm_packs_pu16 (data, _mm_setzero_si64()));
+}
+
+/* Expand 16 bits positioned at @pos (0-3) of a mmx register into
+ *
+ *    00RR00GG00BB
+ *
+ * --- Expanding 565 in the low word ---
+ *
+ * m = (m << (32 - 3)) | (m << (16 - 5)) | m;
+ * m = m & (01f0003f001f);
+ * m = m * (008404100840);
+ * m = m >> 8;
+ *
+ * Note the trick here - the top word is shifted by another nibble to
+ * avoid it bumping into the middle word
+ */
+static inline __m64
+expand565_16_1x64 (uint16_t pixel)
+{
+    __m64 p;
+    __m64 t1, t2;
+
+    p = _mm_cvtsi32_si64 ((uint32_t) pixel);
+
+    t1 = _mm_slli_si64 (p, 36 - 11);
+    t2 = _mm_slli_si64 (p, 16 - 5);
+
+    p = _mm_or_si64 (t1, p);
+    p = _mm_or_si64 (t2, p);
+    p = _mm_and_si64 (p, xMask565rgb);
+    p = _mm_mullo_pi16 (p, xMask565Unpack);
+
+    return _mm_srli_pi16 (p, 8);
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * Compose Core transformations
+ */
+static inline uint32_t
+coreCombineOverUPixelsse2 (uint32_t src, uint32_t dst)
+{
+    uint8_t     a;
+    __m64       ms;
+
+    a = src >> 24;
+
+    if (a == 0xff)
+    {
+        return src;
+    }
+    else if (a)
+    {
+        ms = unpack_32_1x64 (src);
+        return pack_1x64_32 (over_1x64 (ms, expandAlpha_1x64 (ms), unpack_32_1x64 (dst)));
+    }
+
+    return dst;
+}
+
+static inline void
+coreCombineOverUsse2 (uint32_t* pd, const uint32_t* ps, int w)
+{
+    uint32_t pa;
+    uint32_t s, d;
+
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmAlphaLo, xmmAlphaHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    /* Align dst on a 16-byte boundary */
+    while (w &&
+           ((unsigned long)pd & 15))
+    {
+        d = *pd;
+        s = *ps++;
+
+        *pd++ = coreCombineOverUPixelsse2 (s, d);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+
+        /* I'm loading unaligned because I'm not sure about the address alignment. */
+        xmmSrcHi = load128Unaligned ((__m128i*) ps);
+
+        /* Check the alpha channel */
+        pa = packAlpha (xmmSrcHi);
+
+        if (pa == 0xffffffff)
+        {
+            save128Aligned ((__m128i*)pd, xmmSrcHi);
+        }
+        else if (pa)
+        {
+            xmmDstHi = load128Aligned ((__m128i*) pd);
+
+            unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+            unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+            expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi);
+
+            over_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi, &xmmDstLo, &xmmDstHi);
+
+            /* rebuid the 4 pixel data and save*/
+            save128Aligned ((__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+        }
+
+        w -= 4;
+        ps += 4;
+        pd += 4;
+    }
+
+    while (w)
+    {
+        d = *pd;
+        s = *ps++;
+
+        *pd++ = coreCombineOverUPixelsse2 (s, d);
+        w--;
+    }
+}
+
+static inline void
+coreCombineOverReverseUsse2 (uint32_t* pd, const uint32_t* ps, int w)
+{
+    uint32_t s, d;
+
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmAlphaLo, xmmAlphaHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    /* Align dst on a 16-byte boundary */
+    while (w &&
+           ((unsigned long)pd & 15))
+    {
+        d = *pd;
+        s = *ps++;
+
+        *pd++ = coreCombineOverUPixelsse2 (d, s);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+
+        /* I'm loading unaligned because I'm not sure about the address alignment. */
+        xmmSrcHi = load128Unaligned ((__m128i*) ps);
+        xmmDstHi = load128Aligned ((__m128i*) pd);
+
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmAlphaLo, &xmmAlphaHi);
+
+        over_2x128 (&xmmDstLo, &xmmDstHi, &xmmAlphaLo, &xmmAlphaHi, &xmmSrcLo, &xmmSrcHi);
+
+        /* rebuid the 4 pixel data and save*/
+        save128Aligned ((__m128i*)pd, pack_2x128_128 (xmmSrcLo, xmmSrcHi));
+
+        w -= 4;
+        ps += 4;
+        pd += 4;
+    }
+
+    while (w)
+    {
+        d = *pd;
+        s = *ps++;
+
+        *pd++ = coreCombineOverUPixelsse2 (d, s);
+        w--;
+    }
+}
+
+static inline uint32_t
+coreCombineInUPixelsse2 (uint32_t src, uint32_t dst)
+{
+    uint32_t maska = src >> 24;
+
+    if (maska == 0)
+    {
+        return 0;
+    }
+    else if (maska != 0xff)
+    {
+        return pack_1x64_32(pixMultiply_1x64 (unpack_32_1x64 (dst), expandAlpha_1x64 (unpack_32_1x64 (src))));
+    }
+
+    return dst;
+}
+
+static inline void
+coreCombineInUsse2 (uint32_t* pd, const uint32_t* ps, int w)
+{
+    uint32_t s, d;
+
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w && ((unsigned long) pd & 15))
+    {
+        s = *ps++;
+        d = *pd;
+
+        *pd++ = coreCombineInUPixelsse2 (d, s);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+
+        xmmDstHi = load128Aligned ((__m128i*) pd);
+        xmmSrcHi = load128Unaligned ((__m128i*) ps);
+
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        pixMultiply_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmDstLo, &xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+        save128Aligned ((__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        d = *pd;
+
+        *pd++ = coreCombineInUPixelsse2 (d, s);
+        w--;
+    }
+}
+
+static inline void
+coreCombineReverseInUsse2 (uint32_t* pd, const uint32_t* ps, int w)
+{
+    uint32_t s, d;
+
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w && ((unsigned long) pd & 15))
+    {
+        s = *ps++;
+        d = *pd;
+
+        *pd++ = coreCombineInUPixelsse2 (s, d);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+
+        xmmDstHi = load128Aligned ((__m128i*) pd);
+        xmmSrcHi = load128Unaligned ((__m128i*) ps);
+
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+        pixMultiply_2x128 (&xmmDstLo, &xmmDstHi, &xmmSrcLo, &xmmSrcHi, &xmmDstLo, &xmmDstHi);
+
+        save128Aligned ((__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        d = *pd;
+
+        *pd++ = coreCombineInUPixelsse2 (s, d);
+        w--;
+    }
+}
+
+static inline void
+coreCombineReverseOutUsse2 (uint32_t* pd, const uint32_t* ps, int w)
+{
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w && ((unsigned long) pd & 15))
+    {
+        uint32_t s = *ps++;
+        uint32_t d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (d), negate_1x64 (expandAlpha_1x64 (unpack_32_1x64 (s)))));
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w >= 4)
+    {
+        __m128i xmmSrcLo, xmmSrcHi;
+        __m128i xmmDstLo, xmmDstHi;
+
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+
+        xmmSrcHi = load128Unaligned ((__m128i*) ps);
+        xmmDstHi = load128Aligned ((__m128i*) pd);
+
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+        expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        negate_2x128      (xmmSrcLo, xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+
+        pixMultiply_2x128 (&xmmDstLo, &xmmDstHi, &xmmSrcLo, &xmmSrcHi, &xmmDstLo, &xmmDstHi);
+
+        save128Aligned ((__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        uint32_t s = *ps++;
+        uint32_t d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (d), negate_1x64 (expandAlpha_1x64 (unpack_32_1x64 (s)))));
+        w--;
+    }
+}
+
+static inline void
+coreCombineOutUsse2 (uint32_t* pd, const uint32_t* ps, int w)
+{
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w && ((unsigned long) pd & 15))
+    {
+        uint32_t s = *ps++;
+        uint32_t d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (s), negate_1x64 (expandAlpha_1x64 (unpack_32_1x64 (d)))));
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w >= 4)
+    {
+        __m128i xmmSrcLo, xmmSrcHi;
+        __m128i xmmDstLo, xmmDstHi;
+
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+
+        xmmSrcHi = load128Unaligned ((__m128i*) ps);
+        xmmDstHi = load128Aligned ((__m128i*) pd);
+
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmDstLo, &xmmDstHi);
+        negate_2x128      (xmmDstLo, xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+        pixMultiply_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmDstLo, &xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+        save128Aligned ((__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        uint32_t s = *ps++;
+        uint32_t d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (s), negate_1x64 (expandAlpha_1x64 (unpack_32_1x64 (d)))));
+        w--;
+    }
+}
+
+static inline uint32_t
+coreCombineAtopUPixelsse2 (uint32_t src, uint32_t dst)
+{
+    __m64 s = unpack_32_1x64 (src);
+    __m64 d = unpack_32_1x64 (dst);
+
+    __m64 sa = negate_1x64 (expandAlpha_1x64 (s));
+    __m64 da = expandAlpha_1x64 (d);
+
+    return pack_1x64_32 (pixAddMultiply_1x64 (&s, &da, &d, &sa));
+}
+
+static inline void
+coreCombineAtopUsse2 (uint32_t* pd, const uint32_t* ps, int w)
+{
+    uint32_t s, d;
+
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmAlphaSrcLo, xmmAlphaSrcHi;
+    __m128i xmmAlphaDstLo, xmmAlphaDstHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w && ((unsigned long) pd & 15))
+    {
+        s = *ps++;
+        d = *pd;
+
+        *pd++ = coreCombineAtopUPixelsse2 (s, d);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+
+        xmmSrcHi = load128Unaligned ((__m128i*) ps);
+        xmmDstHi = load128Aligned ((__m128i*) pd);
+
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+        expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi);
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmAlphaDstLo, &xmmAlphaDstHi);
+
+        negate_2x128 (xmmAlphaSrcLo, xmmAlphaSrcHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi);
+
+        pixAddMultiply_2x128 ( &xmmSrcLo, &xmmSrcHi, &xmmAlphaDstLo, &xmmAlphaDstHi,
+                               &xmmDstLo, &xmmDstHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi,
+                               &xmmDstLo, &xmmDstHi );
+
+        save128Aligned ((__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        d = *pd;
+
+        *pd++ = coreCombineAtopUPixelsse2 (s, d);
+        w--;
+    }
+}
+
+static inline uint32_t
+coreCombineReverseAtopUPixelsse2 (uint32_t src, uint32_t dst)
+{
+    __m64 s = unpack_32_1x64 (src);
+    __m64 d = unpack_32_1x64 (dst);
+
+    __m64 sa = expandAlpha_1x64 (s);
+    __m64 da = negate_1x64 (expandAlpha_1x64 (d));
+
+    return pack_1x64_32 (pixAddMultiply_1x64 (&s, &da, &d, &sa));
+}
+
+static inline void
+coreCombineReverseAtopUsse2 (uint32_t* pd, const uint32_t* ps, int w)
+{
+    uint32_t s, d;
+
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmAlphaSrcLo, xmmAlphaSrcHi;
+    __m128i xmmAlphaDstLo, xmmAlphaDstHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w && ((unsigned long) pd & 15))
+    {
+        s = *ps++;
+        d = *pd;
+
+        *pd++ = coreCombineReverseAtopUPixelsse2 (s, d);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+
+        xmmSrcHi = load128Unaligned ((__m128i*) ps);
+        xmmDstHi = load128Aligned ((__m128i*) pd);
+
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+        expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi);
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmAlphaDstLo, &xmmAlphaDstHi);
+
+        negate_2x128 (xmmAlphaDstLo, xmmAlphaDstHi, &xmmAlphaDstLo, &xmmAlphaDstHi);
+
+        pixAddMultiply_2x128 ( &xmmSrcLo, &xmmSrcHi, &xmmAlphaDstLo, &xmmAlphaDstHi,
+                               &xmmDstLo, &xmmDstHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi,
+                               &xmmDstLo, &xmmDstHi );
+
+        save128Aligned ((__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        d = *pd;
+
+        *pd++ = coreCombineReverseAtopUPixelsse2 (s, d);
+        w--;
+    }
+}
+
+static inline uint32_t
+coreCombineXorUPixelsse2 (uint32_t src, uint32_t dst)
+{
+    __m64 s = unpack_32_1x64 (src);
+    __m64 d = unpack_32_1x64 (dst);
+
+    __m64 negD = negate_1x64 (expandAlpha_1x64 (d));
+    __m64 negS = negate_1x64 (expandAlpha_1x64 (s));
+
+    return pack_1x64_32 (pixAddMultiply_1x64 (&s, &negD, &d, &negS));
+}
+
+static inline void
+coreCombineXorUsse2 (uint32_t* dst, const uint32_t* src, int width)
+{
+    int w = width;
+    uint32_t s, d;
+    uint32_t* pd = dst;
+    const uint32_t* ps = src;
+
+    __m128i xmmSrc, xmmSrcLo, xmmSrcHi;
+    __m128i xmmDst, xmmDstLo, xmmDstHi;
+    __m128i xmmAlphaSrcLo, xmmAlphaSrcHi;
+    __m128i xmmAlphaDstLo, xmmAlphaDstHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w && ((unsigned long) pd & 15))
+    {
+        s = *ps++;
+        d = *pd;
+
+        *pd++ = coreCombineXorUPixelsse2 (s, d);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+
+        xmmSrc = load128Unaligned ((__m128i*) ps);
+        xmmDst = load128Aligned ((__m128i*) pd);
+
+        unpack_128_2x128 (xmmSrc, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmDst, &xmmDstLo, &xmmDstHi);
+
+        expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi);
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmAlphaDstLo, &xmmAlphaDstHi);
+
+        negate_2x128 (xmmAlphaSrcLo, xmmAlphaSrcHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi);
+        negate_2x128 (xmmAlphaDstLo, xmmAlphaDstHi, &xmmAlphaDstLo, &xmmAlphaDstHi);
+
+        pixAddMultiply_2x128 ( &xmmSrcLo, &xmmSrcHi, &xmmAlphaDstLo, &xmmAlphaDstHi,
+                               &xmmDstLo, &xmmDstHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi,
+                               &xmmDstLo, &xmmDstHi );
+
+        save128Aligned ((__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        d = *pd;
+
+        *pd++ = coreCombineXorUPixelsse2 (s, d);
+        w--;
+    }
+}
+
+static inline void
+coreCombineAddUsse2 (uint32_t* dst, const uint32_t* src, int width)
+{
+    int w = width;
+    uint32_t s,d;
+    uint32_t* pd = dst;
+    const uint32_t* ps = src;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        d = *pd;
+        *pd++ = _mm_cvtsi64_si32 (_mm_adds_pu8 (_mm_cvtsi32_si64 (s), _mm_cvtsi32_si64 (d)));
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+
+        save128Aligned( (__m128i*)pd,
+                        _mm_adds_epu8( load128Unaligned((__m128i*)ps),
+                                       load128Aligned  ((__m128i*)pd)) );
+        pd += 4;
+        ps += 4;
+        w -= 4;
+    }
+
+    while (w--)
+    {
+        s = *ps++;
+        d = *pd;
+        *pd++ = _mm_cvtsi64_si32 (_mm_adds_pu8 (_mm_cvtsi32_si64 (s), _mm_cvtsi32_si64 (d)));
+    }
+}
+
+static inline uint32_t
+coreCombineSaturateUPixelsse2 (uint32_t src, uint32_t dst)
+{
+    __m64 ms = unpack_32_1x64 (src);
+    __m64 md = unpack_32_1x64 (dst);
+    uint32_t sa = src >> 24;
+    uint32_t da = ~dst >> 24;
+
+    if (sa > da)
+    {
+        ms = pixMultiply_1x64 (ms, expandAlpha_1x64 (unpack_32_1x64 (FbIntDiv(da, sa) << 24)));
+    }
+
+    return pack_1x64_32 (_mm_adds_pu16 (md, ms));
+}
+
+static inline void
+coreCombineSaturateUsse2 (uint32_t *pd, const uint32_t *ps, int w)
+{
+    uint32_t s,d;
+
+    uint32_t packCmp;
+    __m128i xmmSrc, xmmDst;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        d = *pd;
+        *pd++ = coreCombineSaturateUPixelsse2 (s, d);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+
+        xmmDst = load128Aligned  ((__m128i*)pd);
+        xmmSrc = load128Unaligned((__m128i*)ps);
+
+        packCmp = _mm_movemask_epi8 (_mm_cmpgt_epi32 (_mm_srli_epi32 (xmmSrc, 24),
+                                                      _mm_srli_epi32 (_mm_xor_si128 (xmmDst, Maskff000000), 24)));
+
+        /* if some alpha src is grater than respective ~alpha dst */
+        if (packCmp)
+        {
+            s = *ps++;
+            d = *pd;
+            *pd++ = coreCombineSaturateUPixelsse2 (s, d);
+
+            s = *ps++;
+            d = *pd;
+            *pd++ = coreCombineSaturateUPixelsse2 (s, d);
+
+            s = *ps++;
+            d = *pd;
+            *pd++ = coreCombineSaturateUPixelsse2 (s, d);
+
+            s = *ps++;
+            d = *pd;
+            *pd++ = coreCombineSaturateUPixelsse2 (s, d);
+        }
+        else
+        {
+            save128Aligned ((__m128i*)pd, _mm_adds_epu8 (xmmDst, xmmSrc));
+
+            pd += 4;
+            ps += 4;
+        }
+
+        w -= 4;
+    }
+
+    while (w--)
+    {
+        s = *ps++;
+        d = *pd;
+        *pd++ = coreCombineSaturateUPixelsse2 (s, d);
+    }
+}
+
+static inline void
+coreCombineSrcCsse2 (uint32_t* pd, const uint32_t* ps, const uint32_t *pm, int w)
+{
+    uint32_t s, m;
+
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmMaskLo, xmmMaskHi;
+    __m128i xmmDstLo, xmmDstHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        m = *pm++;
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (m)));
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+        cachePrefetchNext ((__m128i*)pm);
+
+        xmmSrcHi = load128Unaligned ((__m128i*)ps);
+        xmmMaskHi = load128Unaligned ((__m128i*)pm);
+
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        pixMultiply_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmMaskLo, &xmmMaskHi, &xmmDstLo, &xmmDstHi);
+
+        save128Aligned( (__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        pm += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        m = *pm++;
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (m)));
+        w--;
+    }
+}
+
+static inline uint32_t
+coreCombineOverCPixelsse2 (uint32_t src, uint32_t mask, uint32_t dst)
+{
+    __m64 s = unpack_32_1x64 (src);
+    __m64 expAlpha = expandAlpha_1x64 (s);
+    __m64 unpkMask = unpack_32_1x64 (mask);
+    __m64 unpkDst  = unpack_32_1x64 (dst);
+
+    return pack_1x64_32 (inOver_1x64 (&s, &expAlpha, &unpkMask, &unpkDst));
+}
+
+static inline void
+coreCombineOverCsse2 (uint32_t* pd, const uint32_t* ps, const uint32_t *pm, int w)
+{
+    uint32_t s, m, d;
+
+    __m128i xmmAlphaLo, xmmAlphaHi;
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmMaskLo, xmmMaskHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = coreCombineOverCPixelsse2 (s, m, d);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+        cachePrefetchNext ((__m128i*)pm);
+
+        xmmDstHi = load128Aligned ((__m128i*)pd);
+        xmmSrcHi = load128Unaligned ((__m128i*)ps);
+        xmmMaskHi = load128Unaligned ((__m128i*)pm);
+
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi);
+
+        inOver_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi, &xmmMaskLo, &xmmMaskHi, &xmmDstLo, &xmmDstHi);
+
+        save128Aligned( (__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        pm += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = coreCombineOverCPixelsse2 (s, m, d);
+        w--;
+    }
+}
+
+static inline uint32_t
+coreCombineOverReverseCPixelsse2 (uint32_t src, uint32_t mask, uint32_t dst)
+{
+    __m64 d = unpack_32_1x64 (dst);
+
+	return pack_1x64_32(over_1x64 (d, expandAlpha_1x64 (d), pixMultiply_1x64 (unpack_32_1x64 (src), unpack_32_1x64 (mask))));
+}
+
+static inline void
+coreCombineOverReverseCsse2 (uint32_t* pd, const uint32_t* ps, const uint32_t *pm, int w)
+{
+    uint32_t s, m, d;
+
+    __m128i xmmAlphaLo, xmmAlphaHi;
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmMaskLo, xmmMaskHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = coreCombineOverReverseCPixelsse2 (s, m, d);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+        cachePrefetchNext ((__m128i*)pm);
+
+        xmmDstHi = load128Aligned ((__m128i*)pd);
+        xmmSrcHi = load128Unaligned ((__m128i*)ps);
+        xmmMaskHi = load128Unaligned ((__m128i*)pm);
+
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmAlphaLo, &xmmAlphaHi);
+        pixMultiply_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmMaskLo, &xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        over_2x128 (&xmmDstLo, &xmmDstHi, &xmmAlphaLo, &xmmAlphaHi, &xmmMaskLo, &xmmMaskHi);
+
+        save128Aligned( (__m128i*)pd, pack_2x128_128 (xmmMaskLo, xmmMaskHi));
+
+        ps += 4;
+        pd += 4;
+        pm += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = coreCombineOverReverseCPixelsse2 (s, m, d);
+        w--;
+    }
+}
+
+static inline void
+coreCombineInCsse2 (uint32_t *pd, uint32_t *ps, uint32_t *pm, int w)
+{
+    uint32_t s, m, d;
+
+    __m128i xmmAlphaLo, xmmAlphaHi;
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmMaskLo, xmmMaskHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (pixMultiply_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (m)),
+                                                expandAlpha_1x64 (unpack_32_1x64 (d))));
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+        cachePrefetchNext ((__m128i*)pm);
+
+        xmmDstHi = load128Aligned ((__m128i*)pd);
+        xmmSrcHi = load128Unaligned ((__m128i*)ps);
+        xmmMaskHi = load128Unaligned ((__m128i*)pm);
+
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmAlphaLo, &xmmAlphaHi);
+        pixMultiply_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmMaskLo, &xmmMaskHi, &xmmDstLo, &xmmDstHi);
+
+        pixMultiply_2x128 (&xmmDstLo, &xmmDstHi, &xmmAlphaLo, &xmmAlphaHi, &xmmDstLo, &xmmDstHi);
+
+        save128Aligned( (__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        pm += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (pixMultiply_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (m)),
+                                                expandAlpha_1x64 (unpack_32_1x64 (d))));
+        w--;
+    }
+}
+
+static inline void
+coreCombineInReverseCsse2 (uint32_t *pd, uint32_t *ps, uint32_t *pm, int w)
+{
+    uint32_t s, m, d;
+
+    __m128i xmmAlphaLo, xmmAlphaHi;
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmMaskLo, xmmMaskHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (d),
+                                                pixMultiply_1x64 (unpack_32_1x64 (m),
+                                                                  expandAlpha_1x64 (unpack_32_1x64 (s)))));
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+        cachePrefetchNext ((__m128i*)pm);
+
+        xmmDstHi = load128Aligned ((__m128i*)pd);
+        xmmSrcHi = load128Unaligned ((__m128i*)ps);
+        xmmMaskHi = load128Unaligned ((__m128i*)pm);
+
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi);
+        pixMultiply_2x128 (&xmmMaskLo, &xmmMaskHi, &xmmAlphaLo, &xmmAlphaHi, &xmmAlphaLo, &xmmAlphaHi);
+
+        pixMultiply_2x128 (&xmmDstLo, &xmmDstHi, &xmmAlphaLo, &xmmAlphaHi, &xmmDstLo, &xmmDstHi);
+
+        save128Aligned( (__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        pm += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (d),
+                                                pixMultiply_1x64 (unpack_32_1x64 (m),
+                                                                  expandAlpha_1x64 (unpack_32_1x64 (s)))));
+        w--;
+    }
+}
+
+static inline void
+coreCombineOutCsse2 (uint32_t *pd, uint32_t *ps, uint32_t *pm, int w)
+{
+    uint32_t s, m, d;
+
+    __m128i xmmAlphaLo, xmmAlphaHi;
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmMaskLo, xmmMaskHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (pixMultiply_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (m)),
+                                                negate_1x64 (expandAlpha_1x64 (unpack_32_1x64 (d)))));
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+        cachePrefetchNext ((__m128i*)pm);
+
+        xmmDstHi = load128Aligned ((__m128i*)pd);
+        xmmSrcHi = load128Unaligned ((__m128i*)ps);
+        xmmMaskHi = load128Unaligned ((__m128i*)pm);
+
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmAlphaLo, &xmmAlphaHi);
+        negate_2x128 (xmmAlphaLo, xmmAlphaHi, &xmmAlphaLo, &xmmAlphaHi);
+
+        pixMultiply_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmMaskLo, &xmmMaskHi, &xmmDstLo, &xmmDstHi);
+        pixMultiply_2x128 (&xmmDstLo, &xmmDstHi, &xmmAlphaLo, &xmmAlphaHi, &xmmDstLo, &xmmDstHi);
+
+        save128Aligned( (__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        pm += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (pixMultiply_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (m)),
+                                                negate_1x64 (expandAlpha_1x64 (unpack_32_1x64 (d)))));
+        w--;
+    }
+}
+
+static inline void
+coreCombineOutReverseCsse2 (uint32_t *pd, uint32_t *ps, uint32_t *pm, int w)
+{
+    uint32_t s, m, d;
+
+    __m128i xmmAlphaLo, xmmAlphaHi;
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmMaskLo, xmmMaskHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (d),
+                                                negate_1x64 (pixMultiply_1x64 (unpack_32_1x64 (m),
+                                                                               expandAlpha_1x64 (unpack_32_1x64 (s))))));
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+        cachePrefetchNext ((__m128i*)pm);
+
+        xmmDstHi = load128Aligned ((__m128i*)pd);
+        xmmSrcHi = load128Unaligned ((__m128i*)ps);
+        xmmMaskHi = load128Unaligned ((__m128i*)pm);
+
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi);
+
+        pixMultiply_2x128 (&xmmMaskLo, &xmmMaskHi, &xmmAlphaLo, &xmmAlphaHi, &xmmMaskLo, &xmmMaskHi);
+
+        negate_2x128 (xmmMaskLo, xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        pixMultiply_2x128 (&xmmDstLo, &xmmDstHi, &xmmMaskLo, &xmmMaskHi, &xmmDstLo, &xmmDstHi);
+
+        save128Aligned( (__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        pm += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (d),
+                                                negate_1x64 (pixMultiply_1x64 (unpack_32_1x64 (m),
+                                                                               expandAlpha_1x64 (unpack_32_1x64 (s))))));
+        w--;
+    }
+}
+
+static inline uint32_t
+coreCombineAtopCPixelsse2 (uint32_t src, uint32_t mask, uint32_t dst)
+{
+    __m64 m = unpack_32_1x64 (mask);
+    __m64 s = unpack_32_1x64 (src);
+    __m64 d = unpack_32_1x64 (dst);
+    __m64 sa = expandAlpha_1x64 (s);
+    __m64 da = expandAlpha_1x64 (d);
+
+    s = pixMultiply_1x64 (s, m);
+    m = negate_1x64 (pixMultiply_1x64 (m, sa));
+
+    return pack_1x64_32 (pixAddMultiply_1x64 (&d, &m, &s, &da));
+}
+
+static inline void
+coreCombineAtopCsse2 (uint32_t *pd, uint32_t *ps, uint32_t *pm, int w)
+{
+    uint32_t s, m, d;
+
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmAlphaSrcLo, xmmAlphaSrcHi;
+    __m128i xmmAlphaDstLo, xmmAlphaDstHi;
+    __m128i xmmMaskLo, xmmMaskHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = coreCombineAtopCPixelsse2 (s, m, d);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+        cachePrefetchNext ((__m128i*)pm);
+
+        xmmDstHi = load128Aligned ((__m128i*)pd);
+        xmmSrcHi = load128Unaligned ((__m128i*)ps);
+        xmmMaskHi = load128Unaligned ((__m128i*)pm);
+
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi);
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmAlphaDstLo, &xmmAlphaDstHi);
+
+        pixMultiply_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmMaskLo, &xmmMaskHi, &xmmSrcLo, &xmmSrcHi);
+        pixMultiply_2x128 (&xmmMaskLo, &xmmMaskHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi, &xmmMaskLo, &xmmMaskHi);
+
+        negate_2x128 (xmmMaskLo, xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        pixAddMultiply_2x128 (&xmmDstLo, &xmmDstHi, &xmmMaskLo, &xmmMaskHi,
+                              &xmmSrcLo, &xmmSrcHi, &xmmAlphaDstLo, &xmmAlphaDstHi,
+                              &xmmDstLo, &xmmDstHi);
+
+        save128Aligned( (__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        pm += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = coreCombineAtopCPixelsse2 (s, m, d);
+        w--;
+    }
+}
+
+static inline uint32_t
+coreCombineReverseAtopCPixelsse2 (uint32_t src, uint32_t mask, uint32_t dst)
+{
+    __m64 m = unpack_32_1x64 (mask);
+    __m64 s = unpack_32_1x64 (src);
+    __m64 d = unpack_32_1x64 (dst);
+
+    __m64 da = negate_1x64 (expandAlpha_1x64 (d));
+    __m64 sa = expandAlpha_1x64 (s);
+
+    s = pixMultiply_1x64 (s, m);
+    m = pixMultiply_1x64 (m, sa);
+
+    return pack_1x64_32 (pixAddMultiply_1x64 (&d, &m, &s, &da));
+}
+
+static inline void
+coreCombineReverseAtopCsse2 (uint32_t *pd, uint32_t *ps, uint32_t *pm, int w)
+{
+    uint32_t s, m, d;
+
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmAlphaSrcLo, xmmAlphaSrcHi;
+    __m128i xmmAlphaDstLo, xmmAlphaDstHi;
+    __m128i xmmMaskLo, xmmMaskHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = coreCombineReverseAtopCPixelsse2 (s, m, d);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+        cachePrefetchNext ((__m128i*)pm);
+
+        xmmDstHi = load128Aligned ((__m128i*)pd);
+        xmmSrcHi = load128Unaligned ((__m128i*)ps);
+        xmmMaskHi = load128Unaligned ((__m128i*)pm);
+
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi);
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmAlphaDstLo, &xmmAlphaDstHi);
+
+        pixMultiply_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmMaskLo, &xmmMaskHi, &xmmSrcLo, &xmmSrcHi);
+        pixMultiply_2x128 (&xmmMaskLo, &xmmMaskHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi, &xmmMaskLo, &xmmMaskHi);
+
+        negate_2x128 (xmmAlphaDstLo, xmmAlphaDstHi, &xmmAlphaDstLo, &xmmAlphaDstHi);
+
+        pixAddMultiply_2x128 (&xmmDstLo, &xmmDstHi, &xmmMaskLo, &xmmMaskHi,
+                              &xmmSrcLo, &xmmSrcHi, &xmmAlphaDstLo, &xmmAlphaDstHi,
+                              &xmmDstLo, &xmmDstHi);
+
+        save128Aligned( (__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        pm += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = coreCombineReverseAtopCPixelsse2 (s, m, d);
+        w--;
+    }
+}
+
+static inline uint32_t
+coreCombineXorCPixelsse2 (uint32_t src, uint32_t mask, uint32_t dst)
+{
+    __m64 a = unpack_32_1x64 (mask);
+    __m64 s = unpack_32_1x64 (src);
+    __m64 d = unpack_32_1x64 (dst);
+
+    __m64 alphaDst = negate_1x64 (pixMultiply_1x64 (a, expandAlpha_1x64 (s)));
+    __m64 dest      = pixMultiply_1x64 (s, a);
+    __m64 alphaSrc = negate_1x64 (expandAlpha_1x64 (d));
+
+    return pack_1x64_32 (pixAddMultiply_1x64 (&d,
+                                              &alphaDst,
+                                              &dest,
+                                              &alphaSrc));
+}
+
+static inline void
+coreCombineXorCsse2 (uint32_t *pd, uint32_t *ps, uint32_t *pm, int w)
+{
+    uint32_t s, m, d;
+
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmAlphaSrcLo, xmmAlphaSrcHi;
+    __m128i xmmAlphaDstLo, xmmAlphaDstHi;
+    __m128i xmmMaskLo, xmmMaskHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = coreCombineXorCPixelsse2 (s, m, d);
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+        cachePrefetchNext ((__m128i*)pm);
+
+        xmmDstHi = load128Aligned ((__m128i*)pd);
+        xmmSrcHi = load128Unaligned ((__m128i*)ps);
+        xmmMaskHi = load128Unaligned ((__m128i*)pm);
+
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi);
+        expandAlpha_2x128 (xmmDstLo, xmmDstHi, &xmmAlphaDstLo, &xmmAlphaDstHi);
+
+        pixMultiply_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmMaskLo, &xmmMaskHi, &xmmSrcLo, &xmmSrcHi);
+        pixMultiply_2x128 (&xmmMaskLo, &xmmMaskHi, &xmmAlphaSrcLo, &xmmAlphaSrcHi, &xmmMaskLo, &xmmMaskHi);
+
+        negate_2x128 (xmmAlphaDstLo, xmmAlphaDstHi, &xmmAlphaDstLo, &xmmAlphaDstHi);
+        negate_2x128 (xmmMaskLo, xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+        pixAddMultiply_2x128 (&xmmDstLo, &xmmDstHi, &xmmMaskLo, &xmmMaskHi,
+                              &xmmSrcLo, &xmmSrcHi, &xmmAlphaDstLo, &xmmAlphaDstHi,
+                              &xmmDstLo, &xmmDstHi);
+
+        save128Aligned( (__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+        ps += 4;
+        pd += 4;
+        pm += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = coreCombineXorCPixelsse2 (s, m, d);
+        w--;
+    }
+}
+
+static inline void
+coreCombineAddCsse2 (uint32_t *pd, uint32_t *ps, uint32_t *pm, int w)
+{
+    uint32_t s, m, d;
+
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+    __m128i xmmMaskLo, xmmMaskHi;
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w && (unsigned long)pd & 15)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = pack_1x64_32 (_mm_adds_pu8 (pixMultiply_1x64 (unpack_32_1x64 (s),
+                                                              unpack_32_1x64 (m)),
+                                            unpack_32_1x64 (d)));
+        w--;
+    }
+
+    /* call prefetch hint to optimize cache load*/
+    cachePrefetch ((__m128i*)ps);
+    cachePrefetch ((__m128i*)pd);
+    cachePrefetch ((__m128i*)pm);
+
+    while (w >= 4)
+    {
+        /* fill cache line with next memory */
+        cachePrefetchNext ((__m128i*)ps);
+        cachePrefetchNext ((__m128i*)pd);
+        cachePrefetchNext ((__m128i*)pm);
+
+        xmmSrcHi = load128Unaligned ((__m128i*)ps);
+        xmmMaskHi = load128Unaligned ((__m128i*)pm);
+        xmmDstHi = load128Aligned ((__m128i*)pd);
+
+        unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+        unpack_128_2x128 (xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+        unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+        pixMultiply_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmMaskLo, &xmmMaskHi, &xmmSrcLo, &xmmSrcHi);
+
+        save128Aligned( (__m128i*)pd, pack_2x128_128 (_mm_adds_epu8 (xmmSrcLo, xmmDstLo),
+                                                      _mm_adds_epu8 (xmmSrcHi, xmmDstHi)));
+
+        ps += 4;
+        pd += 4;
+        pm += 4;
+        w -= 4;
+    }
+
+    while (w)
+    {
+        s = *ps++;
+        m = *pm++;
+        d = *pd;
+
+        *pd++ = pack_1x64_32 (_mm_adds_pu8 (pixMultiply_1x64 (unpack_32_1x64 (s),
+                                                              unpack_32_1x64 (m)),
+                                            unpack_32_1x64 (d)));
+        w--;
+    }
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbComposeSetupSSE2
+ */
+static inline __m64
+createMask_16_64 (uint16_t mask)
+{
+    return _mm_set1_pi16 (mask);
+}
+
+static inline __m128i
+createMask_16_128 (uint16_t mask)
+{
+    return _mm_set1_epi16 (mask);
+}
+
+static inline __m64
+createMask_2x32_64 (uint32_t mask0, uint32_t mask1)
+{
+    return _mm_set_pi32 (mask0, mask1);
+}
+
+static inline __m128i
+createMask_2x32_128 (uint32_t mask0, uint32_t mask1)
+{
+    return _mm_set_epi32 (mask0, mask1, mask0, mask1);
+}
+
+/* SSE2 code patch for fbcompose.c */
+
+static FASTCALL void
+sse2CombineMaskU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineReverseInUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineOverU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineOverUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineOverReverseU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineOverReverseUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineInU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineInUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineInReverseU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineReverseInUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineOutU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineOutUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineOutReverseU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineReverseOutUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineAtopU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineAtopUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineAtopReverseU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineReverseAtopUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineXorU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineXorUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineAddU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineAddUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineSaturateU (uint32_t *dst, const uint32_t *src, int width)
+{
+    coreCombineSaturateUsse2 (dst, src, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineSrcC (uint32_t *dst, uint32_t *src, uint32_t *mask, int width)
+{
+    coreCombineSrcCsse2 (dst, src, mask, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineOverC (uint32_t *dst, uint32_t *src, uint32_t *mask, int width)
+{
+    coreCombineOverCsse2 (dst, src, mask, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineOverReverseC (uint32_t *dst, uint32_t *src, uint32_t *mask, int width)
+{
+    coreCombineOverReverseCsse2 (dst, src, mask, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineInC (uint32_t *dst, uint32_t *src, uint32_t *mask, int width)
+{
+    coreCombineInCsse2 (dst, src, mask, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineInReverseC (uint32_t *dst, uint32_t *src, uint32_t *mask, int width)
+{
+    coreCombineInReverseCsse2 (dst, src, mask, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineOutC (uint32_t *dst, uint32_t *src, uint32_t *mask, int width)
+{
+    coreCombineOutCsse2 (dst, src, mask, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineOutReverseC (uint32_t *dst, uint32_t *src, uint32_t *mask, int width)
+{
+    coreCombineOutReverseCsse2 (dst, src, mask, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineAtopC (uint32_t *dst, uint32_t *src, uint32_t *mask, int width)
+{
+    coreCombineAtopCsse2 (dst, src, mask, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineAtopReverseC (uint32_t *dst, uint32_t *src, uint32_t *mask, int width)
+{
+    coreCombineReverseAtopCsse2 (dst, src, mask, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineXorC (uint32_t *dst, uint32_t *src, uint32_t *mask, int width)
+{
+    coreCombineXorCsse2 (dst, src, mask, width);
+    _mm_empty();
+}
+
+static FASTCALL void
+sse2CombineAddC (uint32_t *dst, uint32_t *src, uint32_t *mask, int width)
+{
+    coreCombineAddCsse2 (dst, src, mask, width);
+    _mm_empty();
+}
+
+void
+fbComposeSetupSSE2(void)
+{
+    static pixman_bool_t initialized = FALSE;
+
+    if (initialized)
+	return;
+    
+    /* check if we have SSE2 support and initialize accordingly */
+    if (pixman_have_sse2())
+    {
+        /* SSE2 constants */
+        Mask565r  = createMask_2x32_128 (0x00f80000, 0x00f80000);
+        Mask565g1 = createMask_2x32_128 (0x00070000, 0x00070000);
+        Mask565g2 = createMask_2x32_128 (0x000000e0, 0x000000e0);
+        Mask565b  = createMask_2x32_128 (0x0000001f, 0x0000001f);
+        MaskRed   = createMask_2x32_128 (0x00f80000, 0x00f80000);
+        MaskGreen = createMask_2x32_128 (0x0000fc00, 0x0000fc00);
+        MaskBlue  = createMask_2x32_128 (0x000000f8, 0x000000f8);
+	Mask565FixRB = createMask_2x32_128 (0x00e000e0, 0x00e000e0);
+	Mask565FixG = createMask_2x32_128  (0x0000c000, 0x0000c000);
+        Mask0080 = createMask_16_128 (0x0080);
+        Mask00ff = createMask_16_128 (0x00ff);
+        Mask0101 = createMask_16_128 (0x0101);
+        Maskffff = createMask_16_128 (0xffff);
+        Maskff000000 = createMask_2x32_128 (0xff000000, 0xff000000);
+        MaskAlpha = createMask_2x32_128 (0x00ff0000, 0x00000000);
+
+        /* MMX constants */
+        xMask565rgb = createMask_2x32_64 (0x000001f0, 0x003f001f);
+        xMask565Unpack = createMask_2x32_64 (0x00000084, 0x04100840);
+
+        xMask0080 = createMask_16_64 (0x0080);
+        xMask00ff = createMask_16_64 (0x00ff);
+        xMask0101 = createMask_16_64 (0x0101);
+        xMaskAlpha = createMask_2x32_64 (0x00ff0000, 0x00000000);
+
+        /* SSE code patch for fbcompose.c */
+        pixman_composeFunctions.combineU[PIXMAN_OP_OVER] = sse2CombineOverU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_OVER_REVERSE] = sse2CombineOverReverseU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_IN] = sse2CombineInU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_IN_REVERSE] = sse2CombineInReverseU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_OUT] = sse2CombineOutU;
+
+        pixman_composeFunctions.combineU[PIXMAN_OP_OUT_REVERSE] = sse2CombineOutReverseU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_ATOP] = sse2CombineAtopU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_ATOP_REVERSE] = sse2CombineAtopReverseU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_XOR] = sse2CombineXorU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_ADD] = sse2CombineAddU;
+
+        pixman_composeFunctions.combineU[PIXMAN_OP_SATURATE] = sse2CombineSaturateU;
+
+        pixman_composeFunctions.combineC[PIXMAN_OP_SRC] = sse2CombineSrcC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_OVER] = sse2CombineOverC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_OVER_REVERSE] = sse2CombineOverReverseC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_IN] = sse2CombineInC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_IN_REVERSE] = sse2CombineInReverseC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_OUT] = sse2CombineOutC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_OUT_REVERSE] = sse2CombineOutReverseC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_ATOP] = sse2CombineAtopC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_ATOP_REVERSE] = sse2CombineAtopReverseC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_XOR] = sse2CombineXorC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_ADD] = sse2CombineAddC;
+
+        pixman_composeFunctions.combineMaskU = sse2CombineMaskU;
+    }
+
+    initialized = TRUE;
+
+    _mm_empty();
+}
+
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSolid_nx8888
+ */
+
+void
+fbCompositeSolid_nx8888sse2 (pixman_op_t op,
+			    pixman_image_t * pSrc,
+			    pixman_image_t * pMask,
+			    pixman_image_t * pDst,
+			    int16_t	xSrc,
+			    int16_t	ySrc,
+			    int16_t	xMask,
+			    int16_t	yMask,
+			    int16_t	xDst,
+			    int16_t	yDst,
+			    uint16_t	width,
+			    uint16_t	height)
+{
+    uint32_t	src;
+    uint32_t	*dstLine, *dst, d;
+    uint16_t	w;
+    int	dstStride;
+    __m128i xmmSrc, xmmAlpha;
+    __m128i xmmDst, xmmDstLo, xmmDstHi;
+
+    fbComposeGetSolid(pSrc, src, pDst->bits.format);
+
+    if (src >> 24 == 0)
+	return;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
+
+    xmmSrc = expandPixel_32_1x128 (src);
+    xmmAlpha = expandAlpha_1x128 (xmmSrc);
+
+    while (height--)
+    {
+        dst = dstLine;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)dst);
+
+        dstLine += dstStride;
+        w = width;
+
+        while (w && (unsigned long)dst & 15)
+        {
+            d = *dst;
+            *dst++ = pack_1x64_32 (over_1x64 (_mm_movepi64_pi64 (xmmSrc),
+                                              _mm_movepi64_pi64 (xmmAlpha),
+                                              unpack_32_1x64 (d)));
+            w--;
+        }
+
+        cachePrefetch ((__m128i*)dst);
+
+        while (w >= 4)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)dst);
+
+            xmmDst = load128Aligned ((__m128i*)dst);
+
+            unpack_128_2x128 (xmmDst, &xmmDstLo, &xmmDstHi);
+
+            over_2x128 (&xmmSrc, &xmmSrc, &xmmAlpha, &xmmAlpha, &xmmDstLo, &xmmDstHi);
+
+            /* rebuid the 4 pixel data and save*/
+            save128Aligned ((__m128i*)dst, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+            w -= 4;
+            dst += 4;
+        }
+
+        while (w)
+        {
+            d = *dst;
+            *dst++ = pack_1x64_32 (over_1x64 (_mm_movepi64_pi64 (xmmSrc),
+                                              _mm_movepi64_pi64 (xmmAlpha),
+                                              unpack_32_1x64 (d)));
+            w--;
+        }
+
+    }
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSolid_nx0565
+ */
+void
+fbCompositeSolid_nx0565sse2 (pixman_op_t op,
+			    pixman_image_t * pSrc,
+			    pixman_image_t * pMask,
+			    pixman_image_t * pDst,
+			    int16_t	xSrc,
+			    int16_t	ySrc,
+			    int16_t	xMask,
+			    int16_t	yMask,
+			    int16_t	xDst,
+			    int16_t	yDst,
+			    uint16_t	width,
+			    uint16_t	height)
+{
+    uint32_t	src;
+    uint16_t	*dstLine, *dst, d;
+    uint16_t	w;
+    int	        dstStride;
+    __m128i xmmSrc, xmmAlpha;
+    __m128i xmmDst, xmmDst0, xmmDst1, xmmDst2, xmmDst3;
+
+    fbComposeGetSolid(pSrc, src, pDst->bits.format);
+
+    if (src >> 24 == 0)
+        return;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint16_t, dstStride, dstLine, 1);
+
+    xmmSrc = expandPixel_32_1x128 (src);
+    xmmAlpha = expandAlpha_1x128 (xmmSrc);
+
+    while (height--)
+    {
+        dst = dstLine;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)dst);
+
+        dstLine += dstStride;
+        w = width;
+
+        while (w && (unsigned long)dst & 15)
+        {
+            d = *dst;
+
+            *dst++ = pack565_32_16 (pack_1x64_32 (over_1x64 (_mm_movepi64_pi64 (xmmSrc),
+                                                             _mm_movepi64_pi64 (xmmAlpha),
+                                                             expand565_16_1x64 (d))));
+            w--;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)dst);
+
+        while (w >= 8)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)dst);
+
+	    xmmDst = load128Aligned ((__m128i*)dst);
+	    
+	    unpack565_128_4x128 (xmmDst, &xmmDst0, &xmmDst1, &xmmDst2, &xmmDst3);
+	    
+            over_2x128 (&xmmSrc, &xmmSrc, &xmmAlpha, &xmmAlpha, &xmmDst0, &xmmDst1);
+            over_2x128 (&xmmSrc, &xmmSrc, &xmmAlpha, &xmmAlpha, &xmmDst2, &xmmDst3);
+
+            xmmDst = pack565_4x128_128 (&xmmDst0, &xmmDst1, &xmmDst2, &xmmDst3);
+            save128Aligned ((__m128i*)dst, xmmDst);
+
+            dst += 8;
+            w -= 8;
+        }
+
+        while (w--)
+        {
+            d = *dst;
+            *dst++ = pack565_32_16 (pack_1x64_32 (over_1x64 (_mm_movepi64_pi64 (xmmSrc),
+                                                             _mm_movepi64_pi64 (xmmAlpha),
+                                                             expand565_16_1x64 (d))));
+        }
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSolidMask_nx8888x8888C
+ */
+
+void
+fbCompositeSolidMask_nx8888x8888Csse2 (pixman_op_t op,
+				      pixman_image_t * pSrc,
+				      pixman_image_t * pMask,
+				      pixman_image_t * pDst,
+				      int16_t	xSrc,
+				      int16_t	ySrc,
+				      int16_t	xMask,
+				      int16_t	yMask,
+				      int16_t	xDst,
+				      int16_t	yDst,
+				      uint16_t	width,
+				      uint16_t	height)
+{
+    uint32_t	src, srca;
+    uint32_t	*dstLine, d;
+    uint32_t	*maskLine, m;
+    uint32_t    packCmp;
+    int	dstStride, maskStride;
+
+    __m128i xmmSrc, xmmAlpha;
+    __m128i xmmDst, xmmDstLo, xmmDstHi;
+    __m128i xmmMask, xmmMaskLo, xmmMaskHi;
+
+    __m64 mmxSrc, mmxAlpha, mmxMask, mmxDst;
+
+    fbComposeGetSolid(pSrc, src, pDst->bits.format);
+
+    srca = src >> 24;
+    if (srca == 0)
+	return;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pMask, xMask, yMask, uint32_t, maskStride, maskLine, 1);
+
+    xmmSrc = _mm_unpacklo_epi8 (createMask_2x32_128 (src, src), _mm_setzero_si128 ());
+    xmmAlpha = expandAlpha_1x128 (xmmSrc);
+    mmxSrc   = _mm_movepi64_pi64 (xmmSrc);
+    mmxAlpha = _mm_movepi64_pi64 (xmmAlpha);
+
+    while (height--)
+    {
+        int w = width;
+        uint32_t *pm = (uint32_t *)maskLine;
+        uint32_t *pd = (uint32_t *)dstLine;
+
+        dstLine += dstStride;
+        maskLine += maskStride;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)pd);
+        cachePrefetch ((__m128i*)pm);
+
+        while (w && (unsigned long)pd & 15)
+        {
+            m = *pm++;
+
+            if (m)
+            {
+                d = *pd;
+                mmxMask = unpack_32_1x64 (m);
+                mmxDst = unpack_32_1x64 (d);
+
+                *pd = pack_1x64_32 (inOver_1x64 (&mmxSrc,
+                                                 &mmxAlpha,
+                                                 &mmxMask,
+                                                 &mmxDst));
+            }
+
+            pd++;
+            w--;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)pd);
+        cachePrefetch ((__m128i*)pm);
+
+        while (w >= 4)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)pd);
+            cachePrefetchNext ((__m128i*)pm);
+
+            xmmMask = load128Unaligned ((__m128i*)pm);
+
+            packCmp = _mm_movemask_epi8 (_mm_cmpeq_epi32 (xmmMask, _mm_setzero_si128()));
+
+            /* if all bits in mask are zero, packCmp are equal to 0xffff */
+            if (packCmp != 0xffff)
+            {
+                xmmDst = load128Aligned ((__m128i*)pd);
+
+                unpack_128_2x128 (xmmMask, &xmmMaskLo, &xmmMaskHi);
+                unpack_128_2x128 (xmmDst, &xmmDstLo, &xmmDstHi);
+
+                inOver_2x128 (&xmmSrc, &xmmSrc, &xmmAlpha, &xmmAlpha, &xmmMaskLo, &xmmMaskHi, &xmmDstLo, &xmmDstHi);
+
+                save128Aligned ((__m128i*)pd, pack_2x128_128 (xmmDstLo, xmmDstHi));
+            }
+
+            pd += 4;
+            pm += 4;
+            w -= 4;
+        }
+
+        while (w)
+        {
+            m = *pm++;
+
+            if (m)
+            {
+                d = *pd;
+                mmxMask = unpack_32_1x64 (m);
+                mmxDst = unpack_32_1x64 (d);
+
+                *pd = pack_1x64_32 (inOver_1x64 (&mmxSrc,
+                                                 &mmxAlpha,
+                                                 &mmxMask,
+                                                 &mmxDst));
+            }
+
+            pd++;
+            w--;
+        }
+    }
+
+    _mm_empty();
+}
+
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSrc_8888x8x8888
+ */
+
+void
+fbCompositeSrc_8888x8x8888sse2 (pixman_op_t op,
+			       pixman_image_t * pSrc,
+			       pixman_image_t * pMask,
+			       pixman_image_t * pDst,
+			       int16_t	xSrc,
+			       int16_t	ySrc,
+			       int16_t      xMask,
+			       int16_t      yMask,
+			       int16_t      xDst,
+			       int16_t      yDst,
+			       uint16_t     width,
+			       uint16_t     height)
+{
+    uint32_t	*dstLine, *dst;
+    uint32_t	*srcLine, *src;
+    uint32_t	mask;
+    uint16_t	w;
+    int	dstStride, srcStride;
+
+    __m128i xmmMask;
+    __m128i xmmSrc, xmmSrcLo, xmmSrcHi;
+    __m128i xmmDst, xmmDstLo, xmmDstHi;
+    __m128i xmmAlphaLo, xmmAlphaHi;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pSrc, xSrc, ySrc, uint32_t, srcStride, srcLine, 1);
+    fbComposeGetSolid (pMask, mask, pDst->bits.format);
+
+    xmmMask = createMask_16_128 (mask >> 24);
+
+    while (height--)
+    {
+        dst = dstLine;
+        dstLine += dstStride;
+        src = srcLine;
+        srcLine += srcStride;
+        w = width;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)dst);
+        cachePrefetch ((__m128i*)src);
+
+        while (w && (unsigned long)dst & 15)
+        {
+            uint32_t s = *src++;
+            uint32_t d = *dst;
+
+            __m64 ms = unpack_32_1x64 (s);
+            __m64 alpha    = expandAlpha_1x64 (ms);
+            __m64 dest     = _mm_movepi64_pi64 (xmmMask);
+            __m64 alphaDst = unpack_32_1x64 (d);
+
+            *dst++ = pack_1x64_32 (inOver_1x64 (&ms,
+                                                &alpha,
+                                                &dest,
+                                                &alphaDst));
+
+            w--;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)dst);
+        cachePrefetch ((__m128i*)src);
+
+        while (w >= 4)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)dst);
+            cachePrefetchNext ((__m128i*)src);
+
+            xmmSrc = load128Unaligned ((__m128i*)src);
+            xmmDst = load128Aligned ((__m128i*)dst);
+
+            unpack_128_2x128 (xmmSrc, &xmmSrcLo, &xmmSrcHi);
+            unpack_128_2x128 (xmmDst, &xmmDstLo, &xmmDstHi);
+            expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi);
+
+            inOver_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi, &xmmMask, &xmmMask, &xmmDstLo, &xmmDstHi);
+
+            save128Aligned( (__m128i*)dst, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+            dst += 4;
+            src += 4;
+            w -= 4;
+        }
+
+        while (w)
+        {
+            uint32_t s = *src++;
+            uint32_t d = *dst;
+
+            __m64 ms = unpack_32_1x64 (s);
+            __m64 alpha = expandAlpha_1x64 (ms);
+            __m64 mask  = _mm_movepi64_pi64 (xmmMask);
+            __m64 dest  = unpack_32_1x64 (d);
+
+            *dst++ = pack_1x64_32 (inOver_1x64 (&ms,
+                                                &alpha,
+                                                &mask,
+                                                &dest));
+
+            w--;
+        }
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSrc_x888xnx8888
+ */
+void
+fbCompositeSrc_x888xnx8888sse2 (pixman_op_t op,
+			       pixman_image_t * pSrc,
+			       pixman_image_t * pMask,
+			       pixman_image_t * pDst,
+			       int16_t	xSrc,
+			       int16_t	ySrc,
+			       int16_t      xMask,
+			       int16_t      yMask,
+			       int16_t      xDst,
+			       int16_t      yDst,
+			       uint16_t     width,
+			       uint16_t     height)
+{
+    uint32_t	*dstLine, *dst;
+    uint32_t	*srcLine, *src;
+    uint32_t	mask;
+    int	dstStride, srcStride;
+    uint16_t	w;
+
+    __m128i xmmMask, xmmAlpha;
+    __m128i xmmSrc, xmmSrcLo, xmmSrcHi;
+    __m128i xmmDst, xmmDstLo, xmmDstHi;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pSrc, xSrc, ySrc, uint32_t, srcStride, srcLine, 1);
+    fbComposeGetSolid (pMask, mask, pDst->bits.format);
+
+    xmmMask = createMask_16_128 (mask >> 24);
+    xmmAlpha = Mask00ff;
+
+    while (height--)
+    {
+        dst = dstLine;
+        dstLine += dstStride;
+        src = srcLine;
+        srcLine += srcStride;
+        w = width;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)dst);
+        cachePrefetch ((__m128i*)src);
+
+        while (w && (unsigned long)dst & 15)
+        {
+            uint32_t s = (*src++) | 0xff000000;
+            uint32_t d = *dst;
+
+            __m64 src   = unpack_32_1x64 (s);
+            __m64 alpha = _mm_movepi64_pi64 (xmmAlpha);
+            __m64 mask  = _mm_movepi64_pi64 (xmmMask);
+            __m64 dest  = unpack_32_1x64 (d);
+
+            *dst++ = pack_1x64_32 (inOver_1x64 (&src,
+                                                &alpha,
+                                                &mask,
+                                                &dest));
+
+            w--;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)dst);
+        cachePrefetch ((__m128i*)src);
+
+        while (w >= 4)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)dst);
+            cachePrefetchNext ((__m128i*)src);
+
+            xmmSrc = _mm_or_si128 (load128Unaligned ((__m128i*)src), Maskff000000);
+            xmmDst = load128Aligned ((__m128i*)dst);
+
+            unpack_128_2x128 (xmmSrc, &xmmSrcLo, &xmmSrcHi);
+            unpack_128_2x128 (xmmDst, &xmmDstLo, &xmmDstHi);
+
+            inOver_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmAlpha, &xmmAlpha, &xmmMask, &xmmMask, &xmmDstLo, &xmmDstHi);
+
+            save128Aligned( (__m128i*)dst, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+            dst += 4;
+            src += 4;
+            w -= 4;
+
+        }
+
+        while (w)
+        {
+            uint32_t s = (*src++) | 0xff000000;
+            uint32_t d = *dst;
+
+            __m64 src  = unpack_32_1x64 (s);
+            __m64 alpha = _mm_movepi64_pi64 (xmmAlpha);
+            __m64 mask  = _mm_movepi64_pi64 (xmmMask);
+            __m64 dest  = unpack_32_1x64 (d);
+
+            *dst++ = pack_1x64_32 (inOver_1x64 (&src,
+                                                &alpha,
+                                                &mask,
+                                                &dest));
+
+            w--;
+        }
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSrc_8888x8888
+ */
+void
+fbCompositeSrc_8888x8888sse2 (pixman_op_t op,
+			     pixman_image_t * pSrc,
+			     pixman_image_t * pMask,
+			     pixman_image_t * pDst,
+			     int16_t	xSrc,
+			     int16_t	ySrc,
+			     int16_t      xMask,
+			     int16_t      yMask,
+			     int16_t      xDst,
+			     int16_t      yDst,
+			     uint16_t     width,
+			     uint16_t     height)
+{
+    int	        dstStride, srcStride;
+    uint32_t	*dstLine, *dst;
+    uint32_t	*srcLine, *src;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pSrc, xSrc, ySrc, uint32_t, srcStride, srcLine, 1);
+
+    dst = dstLine;
+    src = srcLine;
+
+    while (height--)
+    {
+        coreCombineOverUsse2 (dst, src, width);
+
+        dst += dstStride;
+        src += srcStride;
+    }
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSrc_8888x0565
+ */
+static inline uint16_t
+fbCompositeSrc_8888x0565pixel (uint32_t src, uint16_t dst)
+{
+    __m64       ms;
+
+    ms = unpack_32_1x64 (src);
+    return pack565_32_16( pack_1x64_32 (over_1x64 (ms,
+                                                   expandAlpha_1x64 (ms),
+                                                   expand565_16_1x64 (dst))));
+}
+
+void
+fbCompositeSrc_8888x0565sse2 (pixman_op_t op,
+			     pixman_image_t * pSrc,
+			     pixman_image_t * pMask,
+			     pixman_image_t * pDst,
+			     int16_t      xSrc,
+			     int16_t      ySrc,
+			     int16_t      xMask,
+			     int16_t      yMask,
+			     int16_t      xDst,
+			     int16_t      yDst,
+			     uint16_t     width,
+			     uint16_t     height)
+{
+    uint16_t	*dstLine, *dst, d;
+    uint32_t	*srcLine, *src, s;
+    int	dstStride, srcStride;
+    uint16_t	w;
+
+    __m128i xmmAlphaLo, xmmAlphaHi;
+    __m128i xmmSrc, xmmSrcLo, xmmSrcHi;
+    __m128i xmmDst, xmmDst0, xmmDst1, xmmDst2, xmmDst3;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint16_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pSrc, xSrc, ySrc, uint32_t, srcStride, srcLine, 1);
+
+#if 0
+    /* FIXME
+     *
+     * I copy the code from MMX one and keep the fixme.
+     * If it's a problem there, probably is a problem here.
+     */
+    assert (pSrc->pDrawable == pMask->pDrawable);
+#endif
+
+    while (height--)
+    {
+        dst = dstLine;
+        src = srcLine;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)src);
+        cachePrefetch ((__m128i*)dst);
+
+        dstLine += dstStride;
+        srcLine += srcStride;
+        w = width;
+
+        /* Align dst on a 16-byte boundary */
+        while (w &&
+               ((unsigned long)dst & 15))
+        {
+            s = *src++;
+            d = *dst;
+
+            *dst++ = fbCompositeSrc_8888x0565pixel (s, d);
+            w--;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)src);
+        cachePrefetch ((__m128i*)dst);
+
+        /* It's a 8 pixel loop */
+        while (w >= 8)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)src);
+            cachePrefetchNext ((__m128i*)dst);
+
+            /* I'm loading unaligned because I'm not sure about the address alignment. */
+            xmmSrc = load128Unaligned ((__m128i*) src);
+            xmmDst = load128Aligned ((__m128i*) dst);
+
+            /* Unpacking */
+            unpack_128_2x128 (xmmSrc, &xmmSrcLo, &xmmSrcHi);
+            unpack565_128_4x128 (xmmDst, &xmmDst0, &xmmDst1, &xmmDst2, &xmmDst3);
+            expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi);
+
+            /* I'm loading next 4 pixels from memory before to optimze the memory read. */
+            xmmSrc = load128Unaligned ((__m128i*) (src+4));
+
+            over_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi, &xmmDst0, &xmmDst1);
+
+            /* Unpacking */
+            unpack_128_2x128 (xmmSrc, &xmmSrcLo, &xmmSrcHi);
+            expandAlpha_2x128 (xmmSrcLo, xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi);
+
+            over_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmAlphaLo, &xmmAlphaHi, &xmmDst2, &xmmDst3);
+
+            save128Aligned ((__m128i*)dst, pack565_4x128_128 (&xmmDst0, &xmmDst1, &xmmDst2, &xmmDst3));
+
+            w -= 8;
+            dst += 8;
+            src += 8;
+        }
+
+        while (w--)
+        {
+            s = *src++;
+            d = *dst;
+
+            *dst++ = fbCompositeSrc_8888x0565pixel (s, d);
+        }
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSolidMask_nx8x8888
+ */
+
+void
+fbCompositeSolidMask_nx8x8888sse2 (pixman_op_t op,
+				  pixman_image_t * pSrc,
+				  pixman_image_t * pMask,
+				  pixman_image_t * pDst,
+				  int16_t      xSrc,
+				  int16_t      ySrc,
+				  int16_t      xMask,
+				  int16_t      yMask,
+				  int16_t      xDst,
+				  int16_t      yDst,
+				  uint16_t     width,
+				  uint16_t     height)
+{
+    uint32_t	src, srca;
+    uint32_t	*dstLine, *dst;
+    uint8_t	*maskLine, *mask;
+    int	dstStride, maskStride;
+    uint16_t	w;
+    uint32_t m, d;
+
+    __m128i xmmSrc, xmmAlpha, xmmDef;
+    __m128i xmmDst, xmmDstLo, xmmDstHi;
+    __m128i xmmMask, xmmMaskLo, xmmMaskHi;
+
+    __m64 mmxSrc, mmxAlpha, mmxMask, mmxDest;
+
+    fbComposeGetSolid(pSrc, src, pDst->bits.format);
+
+    srca = src >> 24;
+    if (srca == 0)
+	return;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pMask, xMask, yMask, uint8_t, maskStride, maskLine, 1);
+
+    xmmDef = createMask_2x32_128 (src, src);
+    xmmSrc = expandPixel_32_1x128 (src);
+    xmmAlpha = expandAlpha_1x128 (xmmSrc);
+    mmxSrc   = _mm_movepi64_pi64 (xmmSrc);
+    mmxAlpha = _mm_movepi64_pi64 (xmmAlpha);
+
+    while (height--)
+    {
+        dst = dstLine;
+        dstLine += dstStride;
+        mask = maskLine;
+        maskLine += maskStride;
+        w = width;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w && (unsigned long)dst & 15)
+        {
+            uint8_t m = *mask++;
+
+            if (m)
+            {
+                d = *dst;
+                mmxMask = expandPixel_8_1x64 (m);
+                mmxDest = unpack_32_1x64 (d);
+
+                *dst = pack_1x64_32 (inOver_1x64 (&mmxSrc,
+                                                  &mmxAlpha,
+                                                  &mmxMask,
+                                                  &mmxDest));
+            }
+
+            w--;
+            dst++;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w >= 4)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)mask);
+            cachePrefetchNext ((__m128i*)dst);
+
+            m = *((uint32_t*)mask);
+
+            if (srca == 0xff && m == 0xffffffff)
+            {
+                save128Aligned ((__m128i*)dst, xmmDef);
+            }
+            else if (m)
+            {
+                xmmDst = load128Aligned ((__m128i*) dst);
+                xmmMask = unpack_32_1x128 (m);
+                xmmMask = _mm_unpacklo_epi8 (xmmMask, _mm_setzero_si128());
+
+                /* Unpacking */
+                unpack_128_2x128 (xmmDst, &xmmDstLo, &xmmDstHi);
+                unpack_128_2x128 (xmmMask, &xmmMaskLo, &xmmMaskHi);
+
+                expandAlphaRev_2x128 (xmmMaskLo, xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+                inOver_2x128 (&xmmSrc, &xmmSrc, &xmmAlpha, &xmmAlpha, &xmmMaskLo, &xmmMaskHi, &xmmDstLo, &xmmDstHi);
+
+                save128Aligned ((__m128i*)dst, pack_2x128_128 (xmmDstLo, xmmDstHi));
+            }
+
+            w -= 4;
+            dst += 4;
+            mask += 4;
+        }
+
+        while (w)
+        {
+            uint8_t m = *mask++;
+
+            if (m)
+            {
+                d = *dst;
+                mmxMask = expandPixel_8_1x64 (m);
+                mmxDest = unpack_32_1x64 (d);
+
+                *dst = pack_1x64_32 (inOver_1x64 (&mmxSrc,
+                                                  &mmxAlpha,
+                                                  &mmxMask,
+                                                  &mmxDest));
+            }
+
+            w--;
+            dst++;
+        }
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSolidMask_nx8x8888
+ */
+
+pixman_bool_t
+pixmanFillsse2 (uint32_t *bits,
+		 int stride,
+		 int bpp,
+		 int x,
+		 int y,
+		 int width,
+		 int height,
+		 uint32_t data)
+{
+    uint32_t	byte_width;
+    uint8_t	    *byte_line;
+
+    __m128i xmmDef;
+
+    if (bpp == 16 && (data >> 16 != (data & 0xffff)))
+	return FALSE;
+
+    if (bpp != 16 && bpp != 32)
+	return FALSE;
+
+    if (bpp == 16)
+    {
+        stride = stride * (int) sizeof (uint32_t) / 2;
+        byte_line = (uint8_t *)(((uint16_t *)bits) + stride * y + x);
+        byte_width = 2 * width;
+        stride *= 2;
+    }
+    else
+    {
+        stride = stride * (int) sizeof (uint32_t) / 4;
+        byte_line = (uint8_t *)(((uint32_t *)bits) + stride * y + x);
+        byte_width = 4 * width;
+        stride *= 4;
+    }
+
+    cachePrefetch ((__m128i*)byte_line);
+    xmmDef = createMask_2x32_128 (data, data);
+
+    while (height--)
+    {
+        int w;
+        uint8_t *d = byte_line;
+        byte_line += stride;
+        w = byte_width;
+
+
+        cachePrefetchNext ((__m128i*)d);
+
+        while (w >= 2 && ((unsigned long)d & 3))
+        {
+            *(uint16_t *)d = data;
+            w -= 2;
+            d += 2;
+        }
+
+        while (w >= 4 && ((unsigned long)d & 15))
+        {
+            *(uint32_t *)d = data;
+
+            w -= 4;
+            d += 4;
+        }
+
+        cachePrefetchNext ((__m128i*)d);
+
+        while (w >= 128)
+        {
+            cachePrefetch (((__m128i*)d) + 12);
+
+            save128Aligned ((__m128i*)(d),     xmmDef);
+            save128Aligned ((__m128i*)(d+16),  xmmDef);
+            save128Aligned ((__m128i*)(d+32),  xmmDef);
+            save128Aligned ((__m128i*)(d+48),  xmmDef);
+            save128Aligned ((__m128i*)(d+64),  xmmDef);
+            save128Aligned ((__m128i*)(d+80),  xmmDef);
+            save128Aligned ((__m128i*)(d+96),  xmmDef);
+            save128Aligned ((__m128i*)(d+112), xmmDef);
+
+            d += 128;
+            w -= 128;
+        }
+
+        if (w >= 64)
+        {
+            cachePrefetch (((__m128i*)d) + 8);
+
+            save128Aligned ((__m128i*)(d),     xmmDef);
+            save128Aligned ((__m128i*)(d+16),  xmmDef);
+            save128Aligned ((__m128i*)(d+32),  xmmDef);
+            save128Aligned ((__m128i*)(d+48),  xmmDef);
+
+            d += 64;
+            w -= 64;
+        }
+
+        cachePrefetchNext ((__m128i*)d);
+
+        if (w >= 32)
+        {
+            save128Aligned ((__m128i*)(d),     xmmDef);
+            save128Aligned ((__m128i*)(d+16),  xmmDef);
+
+            d += 32;
+            w -= 32;
+        }
+
+        if (w >= 16)
+        {
+            save128Aligned ((__m128i*)(d),     xmmDef);
+
+            d += 16;
+            w -= 16;
+        }
+
+        cachePrefetchNext ((__m128i*)d);
+
+        while (w >= 4)
+        {
+            *(uint32_t *)d = data;
+
+            w -= 4;
+            d += 4;
+        }
+
+        if (w >= 2)
+        {
+            *(uint16_t *)d = data;
+            w -= 2;
+            d += 2;
+        }
+    }
+
+    _mm_empty();
+    return TRUE;
+}
+
+void
+fbCompositeSolidMaskSrc_nx8x8888sse2 (pixman_op_t op,
+				     pixman_image_t * pSrc,
+				     pixman_image_t * pMask,
+				     pixman_image_t * pDst,
+				     int16_t      xSrc,
+				     int16_t      ySrc,
+				     int16_t      xMask,
+				     int16_t      yMask,
+				     int16_t      xDst,
+				     int16_t      yDst,
+				     uint16_t     width,
+				     uint16_t     height)
+{
+    uint32_t	src, srca;
+    uint32_t	*dstLine, *dst;
+    uint8_t	*maskLine, *mask;
+    int	dstStride, maskStride;
+    uint16_t	w;
+    uint32_t    m;
+
+    __m128i xmmSrc, xmmDef;
+    __m128i xmmMask, xmmMaskLo, xmmMaskHi;
+
+    fbComposeGetSolid(pSrc, src, pDst->bits.format);
+
+    srca = src >> 24;
+    if (srca == 0)
+    {
+        pixmanFillsse2 (pDst->bits.bits, pDst->bits.rowstride,
+                        PIXMAN_FORMAT_BPP (pDst->bits.format),
+                        xDst, yDst, width, height, 0);
+        return;
+    }
+
+    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pMask, xMask, yMask, uint8_t, maskStride, maskLine, 1);
+
+    xmmDef = createMask_2x32_128 (src, src);
+    xmmSrc = expandPixel_32_1x128 (src);
+
+    while (height--)
+    {
+        dst = dstLine;
+        dstLine += dstStride;
+        mask = maskLine;
+        maskLine += maskStride;
+        w = width;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w && (unsigned long)dst & 15)
+        {
+            uint8_t m = *mask++;
+
+            if (m)
+            {
+                *dst = pack_1x64_32 (pixMultiply_1x64 (_mm_movepi64_pi64 (xmmSrc), expandPixel_8_1x64 (m)));
+            }
+            else
+            {
+                *dst = 0;
+            }
+
+            w--;
+            dst++;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w >= 4)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)mask);
+            cachePrefetchNext ((__m128i*)dst);
+
+            m = *((uint32_t*)mask);
+
+            if (srca == 0xff && m == 0xffffffff)
+            {
+                save128Aligned ((__m128i*)dst, xmmDef);
+            }
+            else if (m)
+            {
+                xmmMask = unpack_32_1x128 (m);
+                xmmMask = _mm_unpacklo_epi8 (xmmMask, _mm_setzero_si128());
+
+                /* Unpacking */
+                unpack_128_2x128 (xmmMask, &xmmMaskLo, &xmmMaskHi);
+
+                expandAlphaRev_2x128 (xmmMaskLo, xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+                pixMultiply_2x128 (&xmmSrc, &xmmSrc, &xmmMaskLo, &xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+                save128Aligned ((__m128i*)dst, pack_2x128_128 (xmmMaskLo, xmmMaskHi));
+            }
+            else
+            {
+                save128Aligned ((__m128i*)dst, _mm_setzero_si128());
+            }
+
+            w -= 4;
+            dst += 4;
+            mask += 4;
+        }
+
+        while (w)
+        {
+            uint8_t m = *mask++;
+
+            if (m)
+            {
+                *dst = pack_1x64_32 (pixMultiply_1x64 (_mm_movepi64_pi64 (xmmSrc), expandPixel_8_1x64 (m)));
+            }
+            else
+            {
+                *dst = 0;
+            }
+
+            w--;
+            dst++;
+        }
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSolidMask_nx8x0565
+ */
+
+void
+fbCompositeSolidMask_nx8x0565sse2 (pixman_op_t op,
+				  pixman_image_t * pSrc,
+				  pixman_image_t * pMask,
+				  pixman_image_t * pDst,
+				  int16_t      xSrc,
+				  int16_t      ySrc,
+				  int16_t      xMask,
+				  int16_t      yMask,
+				  int16_t      xDst,
+				  int16_t      yDst,
+				  uint16_t     width,
+				  uint16_t     height)
+{
+    uint32_t	src, srca;
+    uint16_t	*dstLine, *dst, d;
+    uint8_t	*maskLine, *mask;
+    int	dstStride, maskStride;
+    uint16_t	w;
+    uint32_t m;
+    __m64 mmxSrc, mmxAlpha, mmxMask, mmxDest;
+
+    __m128i xmmSrc, xmmAlpha;
+    __m128i xmmMask, xmmMaskLo, xmmMaskHi;
+    __m128i xmmDst, xmmDst0, xmmDst1, xmmDst2, xmmDst3;
+
+    fbComposeGetSolid(pSrc, src, pDst->bits.format);
+
+    srca = src >> 24;
+    if (srca == 0)
+	return;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint16_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pMask, xMask, yMask, uint8_t, maskStride, maskLine, 1);
+
+    xmmSrc = expandPixel_32_1x128 (src);
+    xmmAlpha = expandAlpha_1x128 (xmmSrc);
+    mmxSrc = _mm_movepi64_pi64 (xmmSrc);
+    mmxAlpha = _mm_movepi64_pi64 (xmmAlpha);
+
+    while (height--)
+    {
+        dst = dstLine;
+        dstLine += dstStride;
+        mask = maskLine;
+        maskLine += maskStride;
+        w = width;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w && (unsigned long)dst & 15)
+        {
+            m = *mask++;
+
+            if (m)
+            {
+                d = *dst;
+                mmxMask = expandAlphaRev_1x64 (unpack_32_1x64 (m));
+                mmxDest = expand565_16_1x64 (d);
+
+                *dst = pack565_32_16 (pack_1x64_32 (inOver_1x64 (&mmxSrc,
+                                                                 &mmxAlpha,
+                                                                 &mmxMask,
+                                                                 &mmxDest)));
+            }
+
+            w--;
+            dst++;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w >= 8)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)mask);
+            cachePrefetchNext ((__m128i*)dst);
+
+            xmmDst = load128Aligned ((__m128i*) dst);
+            unpack565_128_4x128 (xmmDst, &xmmDst0, &xmmDst1, &xmmDst2, &xmmDst3);
+
+            m = *((uint32_t*)mask);
+            mask += 4;
+
+            if (m)
+            {
+                xmmMask = unpack_32_1x128 (m);
+                xmmMask = _mm_unpacklo_epi8 (xmmMask, _mm_setzero_si128());
+
+                /* Unpacking */
+                unpack_128_2x128 (xmmMask, &xmmMaskLo, &xmmMaskHi);
+
+                expandAlphaRev_2x128 (xmmMaskLo, xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+                inOver_2x128 (&xmmSrc, &xmmSrc, &xmmAlpha, &xmmAlpha, &xmmMaskLo, &xmmMaskHi, &xmmDst0, &xmmDst1);
+            }
+
+            m = *((uint32_t*)mask);
+            mask += 4;
+
+            if (m)
+            {
+                xmmMask = unpack_32_1x128 (m);
+                xmmMask = _mm_unpacklo_epi8 (xmmMask, _mm_setzero_si128());
+
+                /* Unpacking */
+                unpack_128_2x128 (xmmMask, &xmmMaskLo, &xmmMaskHi);
+
+                expandAlphaRev_2x128 (xmmMaskLo, xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+                inOver_2x128 (&xmmSrc, &xmmSrc, &xmmAlpha, &xmmAlpha, &xmmMaskLo, &xmmMaskHi, &xmmDst2, &xmmDst3);
+            }
+
+            save128Aligned ((__m128i*)dst, pack565_4x128_128 (&xmmDst0, &xmmDst1, &xmmDst2, &xmmDst3));
+
+            w -= 8;
+            dst += 8;
+        }
+
+        while (w)
+        {
+            m = *mask++;
+
+            if (m)
+            {
+                d = *dst;
+                mmxMask = expandAlphaRev_1x64 (unpack_32_1x64 (m));
+                mmxDest = expand565_16_1x64 (d);
+
+                *dst = pack565_32_16 (pack_1x64_32 (inOver_1x64 (&mmxSrc,
+                                                                 &mmxAlpha,
+                                                                 &mmxMask,
+                                                                 &mmxDest)));
+            }
+
+            w--;
+            dst++;
+        }
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSrc_8888RevNPx0565
+ */
+
+void
+fbCompositeSrc_8888RevNPx0565sse2 (pixman_op_t op,
+				  pixman_image_t * pSrc,
+				  pixman_image_t * pMask,
+				  pixman_image_t * pDst,
+				  int16_t      xSrc,
+				  int16_t      ySrc,
+				  int16_t      xMask,
+				  int16_t      yMask,
+				  int16_t      xDst,
+				  int16_t      yDst,
+				  uint16_t     width,
+				  uint16_t     height)
+{
+    uint16_t	*dstLine, *dst, d;
+    uint32_t	*srcLine, *src, s;
+    int	dstStride, srcStride;
+    uint16_t	w;
+    uint32_t    packCmp;
+
+    __m64 ms;
+    __m128i xmmSrc, xmmSrcLo, xmmSrcHi;
+    __m128i xmmDst, xmmDst0, xmmDst1, xmmDst2, xmmDst3;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint16_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pSrc, xSrc, ySrc, uint32_t, srcStride, srcLine, 1);
+
+#if 0
+    /* FIXME
+     *
+     * I copy the code from MMX one and keep the fixme.
+     * If it's a problem there, probably is a problem here.
+     */
+    assert (pSrc->pDrawable == pMask->pDrawable);
+#endif
+
+    while (height--)
+    {
+        dst = dstLine;
+        dstLine += dstStride;
+        src = srcLine;
+        srcLine += srcStride;
+        w = width;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)src);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w && (unsigned long)dst & 15)
+        {
+            s = *src++;
+            d = *dst;
+
+            ms = unpack_32_1x64 (s);
+
+            *dst++ = pack565_32_16 (pack_1x64_32 (overRevNonPre_1x64(ms, expand565_16_1x64 (d))));
+            w--;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)src);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w >= 8)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)src);
+            cachePrefetchNext ((__m128i*)dst);
+
+            /* First round */
+            xmmSrc = load128Unaligned((__m128i*)src);
+            xmmDst = load128Aligned  ((__m128i*)dst);
+
+            packCmp = packAlpha (xmmSrc);
+
+            unpack565_128_4x128 (xmmDst, &xmmDst0, &xmmDst1, &xmmDst2, &xmmDst3);
+            unpack_128_2x128 (xmmSrc, &xmmSrcLo, &xmmSrcHi);
+
+            /* preload next round*/
+            xmmSrc = load128Unaligned((__m128i*)(src+4));
+            /* preload next round*/
+
+            if (packCmp == 0xffffffff)
+            {
+                invertColors_2x128 (xmmSrcLo, xmmSrcHi, &xmmDst0, &xmmDst1);
+            }
+            else if (packCmp)
+            {
+                overRevNonPre_2x128 (xmmSrcLo, xmmSrcHi, &xmmDst0, &xmmDst1);
+            }
+
+            /* Second round */
+            packCmp = packAlpha (xmmSrc);
+
+            unpack_128_2x128 (xmmSrc, &xmmSrcLo, &xmmSrcHi);
+
+            if (packCmp == 0xffffffff)
+            {
+                invertColors_2x128 (xmmSrcLo, xmmSrcHi, &xmmDst2, &xmmDst3);
+            }
+            else if (packCmp)
+            {
+                overRevNonPre_2x128 (xmmSrcLo, xmmSrcHi, &xmmDst2, &xmmDst3);
+            }
+
+            save128Aligned ((__m128i*)dst, pack565_4x128_128 (&xmmDst0, &xmmDst1, &xmmDst2, &xmmDst3));
+
+            w -= 8;
+            src += 8;
+            dst += 8;
+        }
+
+        while (w)
+        {
+            s = *src++;
+            d = *dst;
+
+            ms = unpack_32_1x64 (s);
+
+            *dst++ = pack565_32_16 (pack_1x64_32 (overRevNonPre_1x64(ms, expand565_16_1x64 (d))));
+            w--;
+        }
+    }
+
+    _mm_empty();
+}
+
+/* "8888RevNP" is GdkPixbuf's format: ABGR, non premultiplied */
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSrc_8888RevNPx8888
+ */
+
+void
+fbCompositeSrc_8888RevNPx8888sse2 (pixman_op_t op,
+				  pixman_image_t * pSrc,
+				  pixman_image_t * pMask,
+				  pixman_image_t * pDst,
+				  int16_t      xSrc,
+				  int16_t      ySrc,
+				  int16_t      xMask,
+				  int16_t      yMask,
+				  int16_t      xDst,
+				  int16_t      yDst,
+				  uint16_t     width,
+				  uint16_t     height)
+{
+    uint32_t	*dstLine, *dst, d;
+    uint32_t	*srcLine, *src, s;
+    int	dstStride, srcStride;
+    uint16_t	w;
+    uint32_t    packCmp;
+
+    __m128i xmmSrcLo, xmmSrcHi;
+    __m128i xmmDstLo, xmmDstHi;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pSrc, xSrc, ySrc, uint32_t, srcStride, srcLine, 1);
+
+#if 0
+    /* FIXME
+     *
+     * I copy the code from MMX one and keep the fixme.
+     * If it's a problem there, probably is a problem here.
+     */
+    assert (pSrc->pDrawable == pMask->pDrawable);
+#endif
+
+    while (height--)
+    {
+        dst = dstLine;
+        dstLine += dstStride;
+        src = srcLine;
+        srcLine += srcStride;
+        w = width;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)src);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w && (unsigned long)dst & 15)
+        {
+            s = *src++;
+            d = *dst;
+
+            *dst++ = pack_1x64_32 (overRevNonPre_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (d)));
+
+            w--;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)src);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w >= 4)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)src);
+            cachePrefetchNext ((__m128i*)dst);
+
+            xmmSrcHi = load128Unaligned((__m128i*)src);
+
+            packCmp = packAlpha (xmmSrcHi);
+
+            unpack_128_2x128 (xmmSrcHi, &xmmSrcLo, &xmmSrcHi);
+
+            if (packCmp == 0xffffffff)
+            {
+                invertColors_2x128( xmmSrcLo, xmmSrcHi, &xmmDstLo, &xmmDstHi);
+
+                save128Aligned ((__m128i*)dst, pack_2x128_128 (xmmDstLo, xmmDstHi));
+            }
+            else if (packCmp)
+            {
+                xmmDstHi = load128Aligned  ((__m128i*)dst);
+
+                unpack_128_2x128 (xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+                overRevNonPre_2x128 (xmmSrcLo, xmmSrcHi, &xmmDstLo, &xmmDstHi);
+
+                save128Aligned ((__m128i*)dst, pack_2x128_128 (xmmDstLo, xmmDstHi));
+            }
+
+            w -= 4;
+            dst += 4;
+            src += 4;
+        }
+
+        while (w)
+        {
+            s = *src++;
+            d = *dst;
+
+            *dst++ = pack_1x64_32 (overRevNonPre_1x64 (unpack_32_1x64 (s), unpack_32_1x64 (d)));
+
+            w--;
+        }
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSolidMask_nx8888x0565C
+ */
+
+void
+fbCompositeSolidMask_nx8888x0565Csse2 (pixman_op_t op,
+				      pixman_image_t * pSrc,
+				      pixman_image_t * pMask,
+				      pixman_image_t * pDst,
+				      int16_t      xSrc,
+				      int16_t      ySrc,
+				      int16_t      xMask,
+				      int16_t      yMask,
+				      int16_t      xDst,
+				      int16_t      yDst,
+				      uint16_t     width,
+				      uint16_t     height)
+{
+    uint32_t	src, srca;
+    uint16_t	*dstLine, *dst, d;
+    uint32_t	*maskLine, *mask, m;
+    int	dstStride, maskStride;
+    int w;
+    uint32_t packCmp;
+
+    __m128i xmmSrc, xmmAlpha;
+    __m128i xmmMask, xmmMaskLo, xmmMaskHi;
+    __m128i xmmDst, xmmDst0, xmmDst1, xmmDst2, xmmDst3;
+
+    __m64 mmxSrc, mmxAlpha, mmxMask, mmxDest;
+
+    fbComposeGetSolid(pSrc, src, pDst->bits.format);
+
+    srca = src >> 24;
+    if (srca == 0)
+        return;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint16_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pMask, xMask, yMask, uint32_t, maskStride, maskLine, 1);
+
+    xmmSrc = expandPixel_32_1x128 (src);
+    xmmAlpha = expandAlpha_1x128 (xmmSrc);
+    mmxSrc = _mm_movepi64_pi64 (xmmSrc);
+    mmxAlpha = _mm_movepi64_pi64 (xmmAlpha);
+
+    while (height--)
+    {
+        w = width;
+        mask = maskLine;
+        dst = dstLine;
+        maskLine += maskStride;
+        dstLine += dstStride;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w && ((unsigned long)dst & 15))
+        {
+            m = *(uint32_t *) mask;
+
+            if (m)
+            {
+                d = *dst;
+                mmxMask = unpack_32_1x64 (m);
+                mmxDest = expand565_16_1x64 (d);
+
+                *dst = pack565_32_16 (pack_1x64_32 (inOver_1x64 (&mmxSrc,
+                                                                 &mmxAlpha,
+                                                                 &mmxMask,
+                                                                 &mmxDest)));
+            }
+
+            w--;
+            dst++;
+            mask++;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w >= 8)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)mask);
+            cachePrefetchNext ((__m128i*)dst);
+
+            /* First round */
+            xmmMask = load128Unaligned((__m128i*)mask);
+            xmmDst = load128Aligned((__m128i*)dst);
+
+            packCmp = _mm_movemask_epi8 (_mm_cmpeq_epi32 (xmmMask, _mm_setzero_si128()));
+
+            unpack565_128_4x128 (xmmDst, &xmmDst0, &xmmDst1, &xmmDst2, &xmmDst3);
+            unpack_128_2x128 (xmmMask, &xmmMaskLo, &xmmMaskHi);
+
+            /* preload next round*/
+            xmmMask = load128Unaligned((__m128i*)(mask+4));
+            /* preload next round*/
+
+            if (packCmp != 0xffff)
+            {
+                inOver_2x128(&xmmSrc, &xmmSrc, &xmmAlpha, &xmmAlpha, &xmmMaskLo, &xmmMaskHi, &xmmDst0, &xmmDst1);
+            }
+
+            /* Second round */
+            packCmp = _mm_movemask_epi8 (_mm_cmpeq_epi32 (xmmMask, _mm_setzero_si128()));
+
+            unpack_128_2x128 (xmmMask, &xmmMaskLo, &xmmMaskHi);
+
+            if (packCmp != 0xffff)
+            {
+                inOver_2x128(&xmmSrc, &xmmSrc, &xmmAlpha, &xmmAlpha, &xmmMaskLo, &xmmMaskHi, &xmmDst2, &xmmDst3);
+            }
+
+            save128Aligned ((__m128i*)dst, pack565_4x128_128 (&xmmDst0, &xmmDst1, &xmmDst2, &xmmDst3));
+
+            w -= 8;
+            dst += 8;
+            mask += 8;
+        }
+
+        while (w)
+        {
+            m = *(uint32_t *) mask;
+
+            if (m)
+            {
+                d = *dst;
+                mmxMask = unpack_32_1x64 (m);
+                mmxDest = expand565_16_1x64 (d);
+
+                *dst = pack565_32_16 (pack_1x64_32 (inOver_1x64 (&mmxSrc,
+                                                                 &mmxAlpha,
+                                                                 &mmxMask,
+                                                                 &mmxDest)));
+            }
+
+            w--;
+            dst++;
+            mask++;
+        }
+    }
+
+    _mm_empty ();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeIn_nx8x8
+ */
+
+void
+fbCompositeIn_nx8x8sse2 (pixman_op_t op,
+			pixman_image_t * pSrc,
+			pixman_image_t * pMask,
+			pixman_image_t * pDst,
+			int16_t      xSrc,
+			int16_t      ySrc,
+			int16_t      xMask,
+			int16_t      yMask,
+			int16_t      xDst,
+			int16_t      yDst,
+			uint16_t     width,
+			uint16_t     height)
+{
+    uint8_t	*dstLine, *dst;
+    uint8_t	*maskLine, *mask;
+    int	dstStride, maskStride;
+    uint16_t	w, d, m;
+    uint32_t	src;
+    uint8_t	sa;
+
+    __m128i xmmAlpha;
+    __m128i xmmMask, xmmMaskLo, xmmMaskHi;
+    __m128i xmmDst, xmmDstLo, xmmDstHi;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint8_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pMask, xMask, yMask, uint8_t, maskStride, maskLine, 1);
+
+    fbComposeGetSolid(pSrc, src, pDst->bits.format);
+
+    sa = src >> 24;
+    if (sa == 0)
+        return;
+
+    xmmAlpha = expandAlpha_1x128 (expandPixel_32_1x128 (src));
+
+    while (height--)
+    {
+        dst = dstLine;
+        dstLine += dstStride;
+        mask = maskLine;
+        maskLine += maskStride;
+        w = width;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w && ((unsigned long)dst & 15))
+        {
+            m = (uint32_t) *mask++;
+            d = (uint32_t) *dst;
+
+            *dst++ = (uint8_t) pack_1x64_32 (pixMultiply_1x64 (pixMultiply_1x64 (_mm_movepi64_pi64 (xmmAlpha), unpack_32_1x64 (m)),
+                                                               unpack_32_1x64 (d)));
+            w--;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w >= 16)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)mask);
+            cachePrefetchNext ((__m128i*)dst);
+
+            xmmMask = load128Unaligned((__m128i*)mask);
+            xmmDst = load128Aligned((__m128i*)dst);
+
+            unpack_128_2x128 (xmmMask, &xmmMaskLo, &xmmMaskHi);
+            unpack_128_2x128 (xmmDst, &xmmDstLo, &xmmDstHi);
+
+            pixMultiply_2x128 (&xmmAlpha, &xmmAlpha, &xmmMaskLo, &xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+            pixMultiply_2x128 (&xmmMaskLo, &xmmMaskHi, &xmmDstLo, &xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+            save128Aligned ((__m128i*)dst, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+            mask += 16;
+            dst += 16;
+            w -= 16;
+        }
+
+        while (w)
+        {
+            m = (uint32_t) *mask++;
+            d = (uint32_t) *dst;
+
+            *dst++ = (uint8_t) pack_1x64_32 (pixMultiply_1x64 (pixMultiply_1x64 (_mm_movepi64_pi64 (xmmAlpha), unpack_32_1x64 (m)),
+                                                               unpack_32_1x64 (d)));
+            w--;
+        }
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeIn_8x8
+ */
+
+void
+fbCompositeIn_8x8sse2 (pixman_op_t op,
+		      pixman_image_t * pSrc,
+		      pixman_image_t * pMask,
+		      pixman_image_t * pDst,
+		      int16_t      xSrc,
+		      int16_t      ySrc,
+		      int16_t      xMask,
+		      int16_t      yMask,
+		      int16_t      xDst,
+		      int16_t      yDst,
+		      uint16_t     width,
+		      uint16_t     height)
+{
+    uint8_t	*dstLine, *dst;
+    uint8_t	*srcLine, *src;
+    int	srcStride, dstStride;
+    uint16_t	w;
+    uint32_t    s, d;
+
+    __m128i xmmSrc, xmmSrcLo, xmmSrcHi;
+    __m128i xmmDst, xmmDstLo, xmmDstHi;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint8_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pSrc, xSrc, ySrc, uint8_t, srcStride, srcLine, 1);
+
+    while (height--)
+    {
+        dst = dstLine;
+        dstLine += dstStride;
+        src = srcLine;
+        srcLine += srcStride;
+        w = width;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)src);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w && ((unsigned long)dst & 15))
+        {
+            s = (uint32_t) *src++;
+            d = (uint32_t) *dst;
+
+            *dst++ = (uint8_t) pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (s),unpack_32_1x64 (d)));
+            w--;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)src);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w >= 16)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)src);
+            cachePrefetchNext ((__m128i*)dst);
+
+            xmmSrc = load128Unaligned((__m128i*)src);
+            xmmDst = load128Aligned((__m128i*)dst);
+
+            unpack_128_2x128 (xmmSrc, &xmmSrcLo, &xmmSrcHi);
+            unpack_128_2x128 (xmmDst, &xmmDstLo, &xmmDstHi);
+
+            pixMultiply_2x128 (&xmmSrcLo, &xmmSrcHi, &xmmDstLo, &xmmDstHi, &xmmDstLo, &xmmDstHi);
+
+            save128Aligned ((__m128i*)dst, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+            src += 16;
+            dst += 16;
+            w -= 16;
+        }
+
+        while (w)
+        {
+            s = (uint32_t) *src++;
+            d = (uint32_t) *dst;
+
+            *dst++ = (uint8_t) pack_1x64_32 (pixMultiply_1x64 (unpack_32_1x64 (s),unpack_32_1x64 (d)));
+            w--;
+        }
+    }
+
+    _mm_empty ();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSrcAdd_8888x8x8
+ */
+
+void
+fbCompositeSrcAdd_8888x8x8sse2 (pixman_op_t op,
+			       pixman_image_t * pSrc,
+			       pixman_image_t * pMask,
+			       pixman_image_t * pDst,
+			       int16_t      xSrc,
+			       int16_t      ySrc,
+			       int16_t      xMask,
+			       int16_t      yMask,
+			       int16_t      xDst,
+			       int16_t      yDst,
+			       uint16_t     width,
+			       uint16_t     height)
+{
+    uint8_t	*dstLine, *dst;
+    uint8_t	*maskLine, *mask;
+    int	dstStride, maskStride;
+    uint16_t	w;
+    uint32_t	src;
+    uint8_t	sa;
+    uint32_t m, d;
+
+    __m128i xmmAlpha;
+    __m128i xmmMask, xmmMaskLo, xmmMaskHi;
+    __m128i xmmDst, xmmDstLo, xmmDstHi;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint8_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pMask, xMask, yMask, uint8_t, maskStride, maskLine, 1);
+
+    fbComposeGetSolid(pSrc, src, pDst->bits.format);
+
+    sa = src >> 24;
+    if (sa == 0)
+        return;
+
+    xmmAlpha = expandAlpha_1x128 (expandPixel_32_1x128 (src));
+
+    while (height--)
+    {
+        dst = dstLine;
+        dstLine += dstStride;
+        mask = maskLine;
+        maskLine += maskStride;
+        w = width;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w && ((unsigned long)dst & 15))
+        {
+            m = (uint32_t) *mask++;
+            d = (uint32_t) *dst;
+
+            *dst++ = (uint8_t) pack_1x64_32 (_mm_adds_pu16 (pixMultiply_1x64 (_mm_movepi64_pi64 (xmmAlpha), unpack_32_1x64 (m)),
+                                                                              unpack_32_1x64 (d)));
+            w--;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)mask);
+        cachePrefetch ((__m128i*)dst);
+
+        while (w >= 16)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)mask);
+            cachePrefetchNext ((__m128i*)dst);
+
+            xmmMask = load128Unaligned((__m128i*)mask);
+            xmmDst = load128Aligned((__m128i*)dst);
+
+            unpack_128_2x128 (xmmMask, &xmmMaskLo, &xmmMaskHi);
+            unpack_128_2x128 (xmmDst, &xmmDstLo, &xmmDstHi);
+
+            pixMultiply_2x128 (&xmmAlpha, &xmmAlpha, &xmmMaskLo, &xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+            xmmDstLo = _mm_adds_epu16 (xmmMaskLo, xmmDstLo);
+            xmmDstHi = _mm_adds_epu16 (xmmMaskHi, xmmDstHi);
+
+            save128Aligned ((__m128i*)dst, pack_2x128_128 (xmmDstLo, xmmDstHi));
+
+            mask += 16;
+            dst += 16;
+            w -= 16;
+        }
+
+        while (w)
+        {
+            m = (uint32_t) *mask++;
+            d = (uint32_t) *dst;
+
+            *dst++ = (uint8_t) pack_1x64_32 (_mm_adds_pu16 (pixMultiply_1x64 (_mm_movepi64_pi64 (xmmAlpha), unpack_32_1x64 (m)),
+                                                                              unpack_32_1x64 (d)));
+            w--;
+        }
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSrcAdd_8000x8000
+ */
+
+void
+fbCompositeSrcAdd_8000x8000sse2 (pixman_op_t op,
+				pixman_image_t * pSrc,
+				pixman_image_t * pMask,
+				pixman_image_t * pDst,
+				int16_t      xSrc,
+				int16_t      ySrc,
+				int16_t      xMask,
+				int16_t      yMask,
+				int16_t      xDst,
+				int16_t      yDst,
+				uint16_t     width,
+				uint16_t     height)
+{
+    uint8_t	*dstLine, *dst;
+    uint8_t	*srcLine, *src;
+    int	dstStride, srcStride;
+    uint16_t	w;
+    uint16_t	t;
+
+    fbComposeGetStart (pSrc, xSrc, ySrc, uint8_t, srcStride, srcLine, 1);
+    fbComposeGetStart (pDst, xDst, yDst, uint8_t, dstStride, dstLine, 1);
+
+    while (height--)
+    {
+        dst = dstLine;
+        src = srcLine;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)src);
+        cachePrefetch ((__m128i*)dst);
+
+        dstLine += dstStride;
+        srcLine += srcStride;
+        w = width;
+
+        /* Small head */
+        while (w && (unsigned long)dst & 3)
+        {
+            t = (*dst) + (*src++);
+            *dst++ = t | (0 - (t >> 8));
+            w--;
+        }
+
+        coreCombineAddUsse2 ((uint32_t*)dst, (uint32_t*)src, w >> 2);
+
+        /* Small tail */
+        dst += w & 0xfffc;
+        src += w & 0xfffc;
+
+        w &= 3;
+
+        while (w)
+        {
+            t = (*dst) + (*src++);
+            *dst++ = t | (0 - (t >> 8));
+            w--;
+        }
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeSrcAdd_8888x8888
+ */
+void
+fbCompositeSrcAdd_8888x8888sse2 (pixman_op_t 	op,
+				pixman_image_t *	pSrc,
+				pixman_image_t *	pMask,
+				pixman_image_t *	 pDst,
+				int16_t		 xSrc,
+				int16_t      ySrc,
+				int16_t      xMask,
+				int16_t      yMask,
+				int16_t      xDst,
+				int16_t      yDst,
+				uint16_t     width,
+				uint16_t     height)
+{
+    uint32_t	*dstLine, *dst;
+    uint32_t	*srcLine, *src;
+    int	dstStride, srcStride;
+
+    fbComposeGetStart (pSrc, xSrc, ySrc, uint32_t, srcStride, srcLine, 1);
+    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
+
+    while (height--)
+    {
+        dst = dstLine;
+        dstLine += dstStride;
+        src = srcLine;
+        srcLine += srcStride;
+
+        coreCombineAddUsse2 (dst, src, width);
+    }
+
+    _mm_empty();
+}
+
+/* -------------------------------------------------------------------------------------------------
+ * fbCompositeCopyAreasse2
+ */
+
+pixman_bool_t
+pixmanBltsse2 (uint32_t *src_bits,
+		uint32_t *dst_bits,
+		int src_stride,
+		int dst_stride,
+		int src_bpp,
+		int dst_bpp,
+		int src_x, int src_y,
+		int dst_x, int dst_y,
+		int width, int height)
+{
+    uint8_t *	src_bytes;
+    uint8_t *	dst_bytes;
+    int		byte_width;
+
+    if (src_bpp != dst_bpp)
+        return FALSE;
+
+    if (src_bpp == 16)
+    {
+        src_stride = src_stride * (int) sizeof (uint32_t) / 2;
+        dst_stride = dst_stride * (int) sizeof (uint32_t) / 2;
+        src_bytes = (uint8_t *)(((uint16_t *)src_bits) + src_stride * (src_y) + (src_x));
+        dst_bytes = (uint8_t *)(((uint16_t *)dst_bits) + dst_stride * (dst_y) + (dst_x));
+        byte_width = 2 * width;
+        src_stride *= 2;
+        dst_stride *= 2;
+    }
+    else if (src_bpp == 32)
+    {
+        src_stride = src_stride * (int) sizeof (uint32_t) / 4;
+        dst_stride = dst_stride * (int) sizeof (uint32_t) / 4;
+        src_bytes = (uint8_t *)(((uint32_t *)src_bits) + src_stride * (src_y) + (src_x));
+        dst_bytes = (uint8_t *)(((uint32_t *)dst_bits) + dst_stride * (dst_y) + (dst_x));
+        byte_width = 4 * width;
+        src_stride *= 4;
+        dst_stride *= 4;
+    }
+    else
+    {
+        return FALSE;
+    }
+
+    cachePrefetch ((__m128i*)src_bytes);
+    cachePrefetch ((__m128i*)dst_bytes);
+
+    while (height--)
+    {
+        int w;
+        uint8_t *s = src_bytes;
+        uint8_t *d = dst_bytes;
+        src_bytes += src_stride;
+        dst_bytes += dst_stride;
+        w = byte_width;
+
+        cachePrefetchNext ((__m128i*)s);
+        cachePrefetchNext ((__m128i*)d);
+
+        while (w >= 2 && ((unsigned long)d & 3))
+        {
+            *(uint16_t *)d = *(uint16_t *)s;
+            w -= 2;
+            s += 2;
+            d += 2;
+        }
+
+        while (w >= 4 && ((unsigned long)d & 15))
+        {
+            *(uint32_t *)d = *(uint32_t *)s;
+
+            w -= 4;
+            s += 4;
+            d += 4;
+        }
+
+        cachePrefetchNext ((__m128i*)s);
+        cachePrefetchNext ((__m128i*)d);
+
+        while (w >= 64)
+        {
+            __m128i xmm0, xmm1, xmm2, xmm3;
+
+            /* 128 bytes ahead */
+            cachePrefetch (((__m128i*)s) + 8);
+            cachePrefetch (((__m128i*)d) + 8);
+
+            xmm0 = load128Unaligned ((__m128i*)(s));
+            xmm1 = load128Unaligned ((__m128i*)(s+16));
+            xmm2 = load128Unaligned ((__m128i*)(s+32));
+            xmm3 = load128Unaligned ((__m128i*)(s+48));
+
+            save128Aligned ((__m128i*)(d),    xmm0);
+            save128Aligned ((__m128i*)(d+16), xmm1);
+            save128Aligned ((__m128i*)(d+32), xmm2);
+            save128Aligned ((__m128i*)(d+48), xmm3);
+
+            s += 64;
+            d += 64;
+            w -= 64;
+        }
+
+        cachePrefetchNext ((__m128i*)s);
+        cachePrefetchNext ((__m128i*)d);
+
+        while (w >= 16)
+        {
+            save128Aligned ((__m128i*)d, load128Unaligned ((__m128i*)s) );
+
+            w -= 16;
+            d += 16;
+            s += 16;
+        }
+
+        cachePrefetchNext ((__m128i*)s);
+        cachePrefetchNext ((__m128i*)d);
+
+        while (w >= 4)
+        {
+            *(uint32_t *)d = *(uint32_t *)s;
+
+            w -= 4;
+            s += 4;
+            d += 4;
+        }
+
+        if (w >= 2)
+        {
+            *(uint16_t *)d = *(uint16_t *)s;
+            w -= 2;
+            s += 2;
+            d += 2;
+        }
+    }
+
+    _mm_empty();
+
+    return TRUE;
+}
+
+void
+fbCompositeCopyAreasse2 (pixman_op_t       op,
+			pixman_image_t *	pSrc,
+			pixman_image_t *	pMask,
+			pixman_image_t *	pDst,
+			int16_t		xSrc,
+			int16_t		ySrc,
+			int16_t		xMask,
+			int16_t		yMask,
+			int16_t		xDst,
+			int16_t		yDst,
+			uint16_t		width,
+			uint16_t		height)
+{
+    pixmanBltsse2 (pSrc->bits.bits,
+		    pDst->bits.bits,
+		    pSrc->bits.rowstride,
+		    pDst->bits.rowstride,
+		    PIXMAN_FORMAT_BPP (pSrc->bits.format),
+		    PIXMAN_FORMAT_BPP (pDst->bits.format),
+		    xSrc, ySrc, xDst, yDst, width, height);
+}
+
+#if 0
+/* This code are buggy in MMX version, now the bug was translated to SSE2 version */
+void
+fbCompositeOver_x888x8x8888sse2 (pixman_op_t      op,
+				pixman_image_t * pSrc,
+				pixman_image_t * pMask,
+				pixman_image_t * pDst,
+				int16_t      xSrc,
+				int16_t      ySrc,
+				int16_t      xMask,
+				int16_t      yMask,
+				int16_t      xDst,
+				int16_t      yDst,
+				uint16_t     width,
+				uint16_t     height)
+{
+    uint32_t	*src, *srcLine, s;
+    uint32_t    *dst, *dstLine, d;
+    uint8_t	    *mask, *maskLine;
+    uint32_t    m;
+    int		 srcStride, maskStride, dstStride;
+    uint16_t w;
+
+    __m128i xmmSrc, xmmSrcLo, xmmSrcHi;
+    __m128i xmmDst, xmmDstLo, xmmDstHi;
+    __m128i xmmMask, xmmMaskLo, xmmMaskHi;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
+    fbComposeGetStart (pMask, xMask, yMask, uint8_t, maskStride, maskLine, 1);
+    fbComposeGetStart (pSrc, xSrc, ySrc, uint32_t, srcStride, srcLine, 1);
+
+    while (height--)
+    {
+        src = srcLine;
+        srcLine += srcStride;
+        dst = dstLine;
+        dstLine += dstStride;
+        mask = maskLine;
+        maskLine += maskStride;
+
+        w = width;
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)src);
+        cachePrefetch ((__m128i*)dst);
+        cachePrefetch ((__m128i*)mask);
+
+        while (w && (unsigned long)dst & 15)
+        {
+            s = 0xff000000 | *src++;
+            m = (uint32_t) *mask++;
+            d = *dst;
+
+            __m64 ms = unpack_32_1x64 (s);
+
+            if (m != 0xff)
+            {
+                ms = inOver_1x64 (ms,
+                                  xMask00ff,
+                                  expandAlphaRev_1x64 (unpack_32_1x64 (m)),
+                                  unpack_32_1x64 (d));
+            }
+
+            *dst++ = pack_1x64_32 (ms);
+            w--;
+        }
+
+        /* call prefetch hint to optimize cache load*/
+        cachePrefetch ((__m128i*)src);
+        cachePrefetch ((__m128i*)dst);
+        cachePrefetch ((__m128i*)mask);
+
+        while (w >= 4)
+        {
+            /* fill cache line with next memory */
+            cachePrefetchNext ((__m128i*)src);
+            cachePrefetchNext ((__m128i*)dst);
+            cachePrefetchNext ((__m128i*)mask);
+
+            m = *(uint32_t*) mask;
+            xmmSrc = _mm_or_si128 (load128Unaligned ((__m128i*)src), Maskff000000);
+
+            if (m == 0xffffffff)
+            {
+                save128Aligned ((__m128i*)dst, xmmSrc);
+            }
+            else
+            {
+                xmmDst = load128Aligned ((__m128i*)dst);
+
+                xmmMask = _mm_unpacklo_epi16 (unpack_32_1x128 (m), _mm_setzero_si128());
+
+                unpack_128_2x128 (xmmSrc, &xmmSrcLo, &xmmSrcHi);
+                unpack_128_2x128 (xmmMask, &xmmMaskLo, &xmmMaskHi);
+                unpack_128_2x128 (xmmDst, &xmmDstLo, &xmmDstHi);
+
+                expandAlphaRev_2x128 (xmmMaskLo, xmmMaskHi, &xmmMaskLo, &xmmMaskHi);
+
+                inOver_2x128 (xmmSrcLo, xmmSrcHi, Mask00ff, Mask00ff, xmmMaskLo, xmmMaskHi, &xmmDstLo, &xmmDstHi);
+
+                save128Aligned( (__m128i*)dst, pack_2x128_128 (xmmDstLo, xmmDstHi));
+            }
+
+            src += 4;
+            dst += 4;
+            mask += 4;
+            w -= 4;
+        }
+
+        while (w)
+        {
+            m = (uint32_t) *mask++;
+
+            if (m)
+            {
+                s = 0xff000000 | *src;
+
+                if (m == 0xff)
+                {
+                    *dst = s;
+                }
+                else
+                {
+                    d = *dst;
+
+                    *dst = pack_1x64_32 (inOver_1x64 (unpack_32_1x64 (s),
+                                                      xMask00ff,
+                                                      expandAlphaRev_1x64 (unpack_32_1x64 (m)),
+                                                      unpack_32_1x64 (d)));
+                }
+
+            }
+
+            src++;
+            dst++;
+            w--;
+        }
+    }
+
+    _mm_empty();
+}
+#endif /* #if 0 */
+
+#endif /* USE_SSE2 */
diff --git a/lib/pixman/pixman/pixman-sse2.h b/lib/pixman/pixman/pixman-sse2.h
new file mode 100644
index 000000000..2bcbf971b
--- /dev/null
+++ b/lib/pixman/pixman/pixman-sse2.h
@@ -0,0 +1,358 @@
+/*
+ * Copyright © 2008 Rodrigo Kumpera
+ * Copyright © 2008 André Tupinambá
+ *
+ * Permission to use, copy, modify, distribute, and sell this software and its
+ * documentation for any purpose is hereby granted without fee, provided that
+ * the above copyright notice appear in all copies and that both that
+ * copyright notice and this permission notice appear in supporting
+ * documentation, and that the name of Red Hat not be used in advertising or
+ * publicity pertaining to distribution of the software without specific,
+ * written prior permission.  Red Hat makes no representations about the
+ * suitability of this software for any purpose.  It is provided "as is"
+ * without express or implied warranty.
+ *
+ * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS
+ * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
+ * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY
+ * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN
+ * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
+ * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
+ * SOFTWARE.
+ *
+ * Author:  Rodrigo Kumpera (kumpera@gmail.com)
+ *          André Tupinambá (andrelrt@gmail.com)
+ * 
+ * Based on work by Owen Taylor and Søren Sandmann
+ */
+#ifndef _PIXMAN_SSE_H_
+#define _PIXMAN_SSE_H_
+
+#ifdef HAVE_DIX_CONFIG_H
+#include <dix-config.h>
+#endif
+
+#include "pixman-private.h"
+
+#ifdef USE_SSE2
+
+#if !defined(__amd64__) && !defined(__x86_64__)
+pixman_bool_t pixman_have_sse2(void);
+#else
+#define pixman_have_sse2() TRUE
+#endif
+
+#else
+#define pixman_have_sse2() FALSE
+#endif
+
+#ifdef USE_SSE2
+
+void fbComposeSetupSSE2(void);
+
+pixman_bool_t
+pixmanFillsse2 (uint32_t *bits,
+		 int stride,
+		 int bpp,
+		 int x,
+		 int y,
+		 int width,
+		 int height,
+		 uint32_t data);
+
+pixman_bool_t
+pixmanBltsse2 (uint32_t *src_bits,
+		uint32_t *dst_bits,
+		int src_stride,
+		int dst_stride,
+		int src_bpp,
+		int dst_bpp,
+		int src_x, int src_y,
+		int dst_x, int dst_y,
+		int width, int height);
+
+void
+fbCompositeSolid_nx8888sse2 (pixman_op_t op,
+			    pixman_image_t * pSrc,
+			    pixman_image_t * pMask,
+			    pixman_image_t * pDst,
+			    int16_t	xSrc,
+			    int16_t	ySrc,
+			    int16_t	xMask,
+			    int16_t	yMask,
+			    int16_t	xDst,
+			    int16_t	yDst,
+			    uint16_t	width,
+			    uint16_t	height);
+
+void
+fbCompositeSolid_nx0565sse2 (pixman_op_t op,
+			    pixman_image_t * pSrc,
+			    pixman_image_t * pMask,
+			    pixman_image_t * pDst,
+			    int16_t	xSrc,
+			    int16_t	ySrc,
+			    int16_t	xMask,
+			    int16_t	yMask,
+			    int16_t	xDst,
+			    int16_t	yDst,
+			    uint16_t	width,
+			    uint16_t	height);
+
+void
+fbCompositeSolidMask_nx8888x8888Csse2 (pixman_op_t op,
+				      pixman_image_t * pSrc,
+				      pixman_image_t * pMask,
+				      pixman_image_t * pDst,
+				      int16_t	xSrc,
+				      int16_t	ySrc,
+				      int16_t	xMask,
+				      int16_t	yMask,
+				      int16_t	xDst,
+				      int16_t	yDst,
+				      uint16_t	width,
+				      uint16_t	height);
+
+void
+fbCompositeSrc_8888x8x8888sse2 (pixman_op_t op,
+			       pixman_image_t * pSrc,
+			       pixman_image_t * pMask,
+			       pixman_image_t * pDst,
+			       int16_t	xSrc,
+			       int16_t	ySrc,
+			       int16_t      xMask,
+			       int16_t      yMask,
+			       int16_t      xDst,
+			       int16_t      yDst,
+			       uint16_t     width,
+			       uint16_t     height);
+
+void
+fbCompositeSrc_x888xnx8888sse2 (pixman_op_t op,
+			       pixman_image_t * pSrc,
+			       pixman_image_t * pMask,
+			       pixman_image_t * pDst,
+			       int16_t	xSrc,
+			       int16_t	ySrc,
+			       int16_t      xMask,
+			       int16_t      yMask,
+			       int16_t      xDst,
+			       int16_t      yDst,
+			       uint16_t     width,
+			       uint16_t     height);
+
+void
+fbCompositeSrc_8888x8888sse2 (pixman_op_t op,
+                 pixman_image_t * pSrc,
+                 pixman_image_t * pMask,
+                 pixman_image_t * pDst,
+                 int16_t    xSrc,
+                 int16_t    ySrc,
+                 int16_t      xMask,
+                 int16_t      yMask,
+                 int16_t      xDst,
+                 int16_t      yDst,
+                 uint16_t     width,
+                 uint16_t     height);
+
+void
+fbCompositeSrc_8888x0565sse2 (pixman_op_t op,
+                 pixman_image_t * pSrc,
+                 pixman_image_t * pMask,
+                 pixman_image_t * pDst,
+                 int16_t      xSrc,
+                 int16_t      ySrc,
+                 int16_t      xMask,
+                 int16_t      yMask,
+                 int16_t      xDst,
+                 int16_t      yDst,
+                 uint16_t     width,
+                 uint16_t     height);
+
+void
+fbCompositeSolidMask_nx8x8888sse2 (pixman_op_t op,
+				  pixman_image_t * pSrc,
+				  pixman_image_t * pMask,
+				  pixman_image_t * pDst,
+				  int16_t      xSrc,
+				  int16_t      ySrc,
+				  int16_t      xMask,
+				  int16_t      yMask,
+				  int16_t      xDst,
+				  int16_t      yDst,
+				  uint16_t     width,
+				  uint16_t     height);
+
+void
+fbCompositeSolidMaskSrc_nx8x8888sse2 (pixman_op_t op,
+				     pixman_image_t * pSrc,
+				     pixman_image_t * pMask,
+				     pixman_image_t * pDst,
+				     int16_t      xSrc,
+				     int16_t      ySrc,
+				     int16_t      xMask,
+				     int16_t      yMask,
+				     int16_t      xDst,
+				     int16_t      yDst,
+				     uint16_t     width,
+				     uint16_t     height);
+
+void
+fbCompositeSolidMask_nx8x0565sse2 (pixman_op_t op,
+				  pixman_image_t * pSrc,
+				  pixman_image_t * pMask,
+				  pixman_image_t * pDst,
+				  int16_t      xSrc,
+				  int16_t      ySrc,
+				  int16_t      xMask,
+				  int16_t      yMask,
+				  int16_t      xDst,
+				  int16_t      yDst,
+				  uint16_t     width,
+				  uint16_t     height);
+
+void
+fbCompositeSrc_8888RevNPx0565sse2 (pixman_op_t op,
+				  pixman_image_t * pSrc,
+				  pixman_image_t * pMask,
+				  pixman_image_t * pDst,
+				  int16_t      xSrc,
+				  int16_t      ySrc,
+				  int16_t      xMask,
+				  int16_t      yMask,
+				  int16_t      xDst,
+				  int16_t      yDst,
+				  uint16_t     width,
+				  uint16_t     height);
+
+void
+fbCompositeSrc_8888RevNPx8888sse2 (pixman_op_t op,
+				  pixman_image_t * pSrc,
+				  pixman_image_t * pMask,
+				  pixman_image_t * pDst,
+				  int16_t      xSrc,
+				  int16_t      ySrc,
+				  int16_t      xMask,
+				  int16_t      yMask,
+				  int16_t      xDst,
+				  int16_t      yDst,
+				  uint16_t     width,
+				  uint16_t     height);
+
+void
+fbCompositeSolidMask_nx8888x0565Csse2 (pixman_op_t op,
+				      pixman_image_t * pSrc,
+				      pixman_image_t * pMask,
+				      pixman_image_t * pDst,
+				      int16_t      xSrc,
+				      int16_t      ySrc,
+				      int16_t      xMask,
+				      int16_t      yMask,
+				      int16_t      xDst,
+				      int16_t      yDst,
+				      uint16_t     width,
+				      uint16_t     height);
+
+void
+fbCompositeIn_nx8x8sse2 (pixman_op_t op,
+			pixman_image_t * pSrc,
+			pixman_image_t * pMask,
+			pixman_image_t * pDst,
+			int16_t      xSrc,
+			int16_t      ySrc,
+			int16_t      xMask,
+			int16_t      yMask,
+			int16_t      xDst,
+			int16_t      yDst,
+			uint16_t     width,
+			uint16_t     height);
+
+void
+fbCompositeIn_8x8sse2 (pixman_op_t op,
+		      pixman_image_t * pSrc,
+		      pixman_image_t * pMask,
+		      pixman_image_t * pDst,
+		      int16_t      xSrc,
+		      int16_t      ySrc,
+		      int16_t      xMask,
+		      int16_t      yMask,
+		      int16_t      xDst,
+		      int16_t      yDst,
+		      uint16_t     width,
+		      uint16_t     height);
+
+void
+fbCompositeSrcAdd_8888x8x8sse2 (pixman_op_t op,
+			       pixman_image_t * pSrc,
+			       pixman_image_t * pMask,
+			       pixman_image_t * pDst,
+			       int16_t      xSrc,
+			       int16_t      ySrc,
+			       int16_t      xMask,
+			       int16_t      yMask,
+			       int16_t      xDst,
+			       int16_t      yDst,
+			       uint16_t     width,
+			       uint16_t     height);
+
+
+void
+fbCompositeSrcAdd_8000x8000sse2 (pixman_op_t op,
+                pixman_image_t * pSrc,
+                pixman_image_t * pMask,
+                pixman_image_t * pDst,
+                int16_t      xSrc,
+                int16_t      ySrc,
+                int16_t      xMask,
+                int16_t      yMask,
+                int16_t      xDst,
+                int16_t      yDst,
+                uint16_t     width,
+                uint16_t     height);
+
+void
+fbCompositeSrcAdd_8888x8888sse2 (pixman_op_t    op,
+                pixman_image_t *    pSrc,
+                pixman_image_t *    pMask,
+                pixman_image_t *     pDst,
+                int16_t      xSrc,
+                int16_t      ySrc,
+                int16_t      xMask,
+                int16_t      yMask,
+                int16_t      xDst,
+                int16_t      yDst,
+                uint16_t     width,
+                uint16_t     height);
+
+void
+fbCompositeCopyAreasse2 (pixman_op_t       op,
+			pixman_image_t *	pSrc,
+			pixman_image_t *	pMask,
+			pixman_image_t *	pDst,
+			int16_t		xSrc,
+			int16_t		ySrc,
+			int16_t		xMask,
+			int16_t		yMask,
+			int16_t		xDst,
+			int16_t		yDst,
+			uint16_t		width,
+			uint16_t		height);
+
+void
+fbCompositeOver_x888x8x8888sse2 (pixman_op_t      op,
+				pixman_image_t * pSrc,
+				pixman_image_t * pMask,
+				pixman_image_t * pDst,
+				int16_t      xSrc,
+				int16_t      ySrc,
+				int16_t      xMask,
+				int16_t      yMask,
+				int16_t      xDst,
+				int16_t      yDst,
+				uint16_t     width,
+				uint16_t     height);
+
+#endif /* USE_SSE2 */
+
+#endif /* _PIXMAN_SSE_H_ */
diff --git a/lib/pixman/pixman/pixman-transformed.c b/lib/pixman/pixman/pixman-transformed.c
index 19085ca87..5ad92ae9f 100644
--- a/lib/pixman/pixman/pixman-transformed.c
+++ b/lib/pixman/pixman/pixman-transformed.c
@@ -2,6 +2,7 @@
  *
  * Copyright © 2000 Keith Packard, member of The XFree86 Project, Inc.
  *             2005 Lars Knoll & Zack Rusin, Trolltech
+ *             2008 Aaron Plattner, NVIDIA Corporation
  *
  * Permission to use, copy, modify, distribute, and sell this software and its
  * documentation for any purpose is hereby granted without fee, provided that
@@ -31,52 +32,41 @@
 
 #include "pixman-private.h"
 
-#ifdef PIXMAN_FB_ACCESSORS
-#define FETCH_PROC_FOR_PICTURE pixman_fetchProcForPicture_accessors
-#define FETCH_PIXEL_PROC_FOR_PICTURE pixman_fetchPixelProcForPicture_accessors
-#define STORE_PROC_FOR_PICTURE pixman_storeProcForPicture_accessors
+#define Alpha(x) ((x) >> 24)
+#define Red(x) (((x) >> 16) & 0xff)
+#define Green(x) (((x) >> 8) & 0xff)
+#define Blue(x) ((x) & 0xff)
 
-#define FB_FETCH_TRANSFORMED fbFetchTransformed_accessors
-#define FB_FETCH_EXTERNAL_ALPHA fbFetchExternalAlpha_accessors
-#define FB_STORE_EXTERNAL_ALPHA fbStoreExternalAlpha_accessors
-
-#else
-
-#define FETCH_PROC_FOR_PICTURE pixman_fetchProcForPicture
-#define FETCH_PIXEL_PROC_FOR_PICTURE pixman_fetchPixelProcForPicture
-#define STORE_PROC_FOR_PICTURE pixman_storeProcForPicture
-
-#define FB_FETCH_TRANSFORMED fbFetchTransformed
-#define FB_FETCH_EXTERNAL_ALPHA fbFetchExternalAlpha
-#define FB_STORE_EXTERNAL_ALPHA fbStoreExternalAlpha
-
-#endif
+#define Alpha64(x) ((x) >> 48)
+#define Red64(x) (((x) >> 32) & 0xffff)
+#define Green64(x) (((x) >> 16) & 0xffff)
+#define Blue64(x) ((x) & 0xffff)
 
 /*
  * Fetch from region strategies
  */
-typedef FASTCALL uint32_t (*fetchFromRegionProc)(bits_image_t *pict, int x, int y, uint32_t *buffer, fetchPixelProc fetch, pixman_box16_t *box);
+typedef FASTCALL uint32_t (*fetchFromRegionProc)(bits_image_t *pict, int x, int y, uint32_t *buffer, fetchPixelProc32 fetch, pixman_box32_t *box);
 
 static inline uint32_t
-fbFetchFromNoRegion(bits_image_t *pict, int x, int y, uint32_t *buffer, fetchPixelProc fetch, pixman_box16_t *box)
+fbFetchFromNoRegion(bits_image_t *pict, int x, int y, uint32_t *buffer, fetchPixelProc32 fetch, pixman_box32_t *box)
 {
     return fetch (pict, x, y);
 }
 
 static uint32_t
-fbFetchFromNRectangles(bits_image_t *pict, int x, int y, uint32_t *buffer, fetchPixelProc fetch, pixman_box16_t *box)
+fbFetchFromNRectangles(bits_image_t *pict, int x, int y, uint32_t *buffer, fetchPixelProc32 fetch, pixman_box32_t *box)
 {
-    pixman_box16_t box2;
-    if (pixman_region_contains_point (pict->common.src_clip, x, y, &box2))
+    pixman_box32_t box2;
+    if (pixman_region32_contains_point (pict->common.src_clip, x, y, &box2))
         return fbFetchFromNoRegion(pict, x, y, buffer, fetch, box);
     else
         return 0;
 }
 
 static uint32_t
-fbFetchFromOneRectangle(bits_image_t *pict, int x, int y, uint32_t *buffer, fetchPixelProc fetch, pixman_box16_t *box)
+fbFetchFromOneRectangle(bits_image_t *pict, int x, int y, uint32_t *buffer, fetchPixelProc32 fetch, pixman_box32_t *box)
 {
-    pixman_box16_t box2 = *box;
+    pixman_box32_t box2 = *box;
     return ((x < box2.x1) | (x >= box2.x2) | (y < box2.y1) | (y >= box2.y2)) ?
         0 : fbFetchFromNoRegion(pict, x, y, buffer, fetch, box);
 }
@@ -87,15 +77,15 @@ fbFetchFromOneRectangle(bits_image_t *pict, int x, int y, uint32_t *buffer, fetc
 static void
 fbFetchTransformed_Nearest_Normal(bits_image_t * pict, int width, uint32_t *buffer, uint32_t *mask, uint32_t maskBits, pixman_bool_t affine, pixman_vector_t v, pixman_vector_t unit)
 {
-    pixman_box16_t* box = NULL;
-    fetchPixelProc   fetch;
+    pixman_box32_t* box = NULL;
+    fetchPixelProc32   fetch;
     fetchFromRegionProc fetchFromRegion;
     int x, y, i;
 
     /* initialize the two function pointers */
-    fetch = FETCH_PIXEL_PROC_FOR_PICTURE(pict);
+    fetch = ACCESS(pixman_fetchPixelProcForPicture32)(pict);
 
-    if(pixman_region_n_rects (pict->common.src_clip) == 1)
+    if(pixman_region32_n_rects (pict->common.src_clip) == 1)
         fetchFromRegion = fbFetchFromNoRegion;
     else
         fetchFromRegion = fbFetchFromNRectangles;
@@ -133,15 +123,15 @@ fbFetchTransformed_Nearest_Normal(bits_image_t * pict, int width, uint32_t *buff
 static void
 fbFetchTransformed_Nearest_Pad(bits_image_t * pict, int width, uint32_t *buffer, uint32_t *mask, uint32_t maskBits, pixman_bool_t affine, pixman_vector_t v, pixman_vector_t unit)
 {
-    pixman_box16_t *box = NULL;
-    fetchPixelProc   fetch;
+    pixman_box32_t *box = NULL;
+    fetchPixelProc32   fetch;
     fetchFromRegionProc fetchFromRegion;
     int x, y, i;
 
     /* initialize the two function pointers */
-    fetch = FETCH_PIXEL_PROC_FOR_PICTURE(pict);
+    fetch = ACCESS(pixman_fetchPixelProcForPicture32)(pict);
 
-    if(pixman_region_n_rects (pict->common.src_clip) == 1)
+    if(pixman_region32_n_rects (pict->common.src_clip) == 1)
         fetchFromRegion = fbFetchFromNoRegion;
     else
         fetchFromRegion = fbFetchFromNRectangles;
@@ -180,15 +170,15 @@ fbFetchTransformed_Nearest_Pad(bits_image_t * pict, int width, uint32_t *buffer,
 static void
 fbFetchTransformed_Nearest_General(bits_image_t * pict, int width, uint32_t *buffer, uint32_t *mask, uint32_t maskBits, pixman_bool_t affine, pixman_vector_t v, pixman_vector_t unit)
 {
-    pixman_box16_t *box = NULL;
-    fetchPixelProc   fetch;
+    pixman_box32_t *box = NULL;
+    fetchPixelProc32   fetch;
     fetchFromRegionProc fetchFromRegion;
     int x, y, i;
 
     /* initialize the two function pointers */
-    fetch = FETCH_PIXEL_PROC_FOR_PICTURE(pict);
+    fetch = ACCESS(pixman_fetchPixelProcForPicture32)(pict);
 
-    if(pixman_region_n_rects (pict->common.src_clip) == 1)
+    if(pixman_region32_n_rects (pict->common.src_clip) == 1)
     {
         box = &(pict->common.src_clip->extents);
         fetchFromRegion = fbFetchFromOneRectangle;
@@ -223,15 +213,15 @@ fbFetchTransformed_Nearest_General(bits_image_t * pict, int width, uint32_t *buf
 static void
 fbFetchTransformed_Bilinear_Normal(bits_image_t * pict, int width, uint32_t *buffer, uint32_t *mask, uint32_t maskBits, pixman_bool_t affine, pixman_vector_t v, pixman_vector_t unit)
 {
-    pixman_box16_t *box = NULL;
-    fetchPixelProc   fetch;
+    pixman_box32_t *box = NULL;
+    fetchPixelProc32   fetch;
     fetchFromRegionProc fetchFromRegion;
     int i;
 
     /* initialize the two function pointers */
-    fetch = FETCH_PIXEL_PROC_FOR_PICTURE(pict);
+    fetch = ACCESS(pixman_fetchPixelProcForPicture32)(pict);
 
-    if(pixman_region_n_rects (pict->common.src_clip) == 1)
+    if(pixman_region32_n_rects (pict->common.src_clip) == 1)
         fetchFromRegion = fbFetchFromNoRegion;
     else
         fetchFromRegion = fbFetchFromNRectangles;
@@ -300,15 +290,15 @@ fbFetchTransformed_Bilinear_Normal(bits_image_t * pict, int width, uint32_t *buf
 static void
 fbFetchTransformed_Bilinear_Pad(bits_image_t * pict, int width, uint32_t *buffer, uint32_t *mask, uint32_t maskBits, pixman_bool_t affine, pixman_vector_t v, pixman_vector_t unit)
 {
-    pixman_box16_t *box = NULL;
-    fetchPixelProc   fetch;
+    pixman_box32_t *box = NULL;
+    fetchPixelProc32   fetch;
     fetchFromRegionProc fetchFromRegion;
     int i;
 
     /* initialize the two function pointers */
-    fetch = FETCH_PIXEL_PROC_FOR_PICTURE(pict);
+    fetch = ACCESS(pixman_fetchPixelProcForPicture32)(pict);
 
-    if(pixman_region_n_rects (pict->common.src_clip) == 1)
+    if(pixman_region32_n_rects (pict->common.src_clip) == 1)
         fetchFromRegion = fbFetchFromNoRegion;
     else
         fetchFromRegion = fbFetchFromNRectangles;
@@ -377,15 +367,15 @@ fbFetchTransformed_Bilinear_Pad(bits_image_t * pict, int width, uint32_t *buffer
 static void
 fbFetchTransformed_Bilinear_General(bits_image_t * pict, int width, uint32_t *buffer, uint32_t *mask, uint32_t maskBits, pixman_bool_t affine, pixman_vector_t v, pixman_vector_t unit)
 {
-    pixman_box16_t *box = NULL;
-    fetchPixelProc   fetch;
+    pixman_box32_t *box = NULL;
+    fetchPixelProc32   fetch;
     fetchFromRegionProc fetchFromRegion;
     int i;
 
     /* initialize the two function pointers */
-    fetch = FETCH_PIXEL_PROC_FOR_PICTURE(pict);
+    fetch = ACCESS(pixman_fetchPixelProcForPicture32)(pict);
 
-    if(pixman_region_n_rects (pict->common.src_clip) == 1)
+    if(pixman_region32_n_rects (pict->common.src_clip) == 1)
     {
         box = &(pict->common.src_clip->extents);
         fetchFromRegion = fbFetchFromOneRectangle;
@@ -456,8 +446,8 @@ fbFetchTransformed_Bilinear_General(bits_image_t * pict, int width, uint32_t *bu
 static void
 fbFetchTransformed_Convolution(bits_image_t * pict, int width, uint32_t *buffer, uint32_t *mask, uint32_t maskBits, pixman_bool_t affine, pixman_vector_t v, pixman_vector_t unit)
 {
-    pixman_box16_t dummy;
-    fetchPixelProc fetch;
+    pixman_box32_t dummy;
+    fetchPixelProc32 fetch;
     int i;
 
     pixman_fixed_t *params = pict->common.filter_params;
@@ -465,7 +455,7 @@ fbFetchTransformed_Convolution(bits_image_t * pict, int width, uint32_t *buffer,
     int32_t cheight = pixman_fixed_to_int(params[1]);
     int xoff = (params[0] - pixman_fixed_1) >> 1;
     int yoff = (params[1] - pixman_fixed_1) >> 1;
-    fetch = FETCH_PIXEL_PROC_FOR_PICTURE(pict);
+    fetch = ACCESS(pixman_fetchPixelProcForPicture32)(pict);
 
     params += 2;
     for (i = 0; i < width; ++i) {
@@ -518,7 +508,7 @@ fbFetchTransformed_Convolution(bits_image_t * pict, int width, uint32_t *buffer,
                                 default:
                                     tx = x;
                             }
-                            if (pixman_region_contains_point (pict->common.src_clip, tx, ty, &dummy)) {
+                            if (pixman_region32_contains_point (pict->common.src_clip, tx, ty, &dummy)) {
                                 uint32_t c = fetch(pict, tx, ty);
 
                                 srtot += Red(c) * *p;
@@ -567,7 +557,8 @@ adjust (pixman_vector_t *v, pixman_vector_t *u, pixman_fixed_t adjustment)
 }
 
 void
-FB_FETCH_TRANSFORMED(bits_image_t * pict, int x, int y, int width, uint32_t *buffer, uint32_t *mask, uint32_t maskBits)
+ACCESS(fbFetchTransformed)(bits_image_t * pict, int x, int y, int width,
+                           uint32_t *buffer, uint32_t *mask, uint32_t maskBits)
 {
     uint32_t     *bits;
     int32_t    stride;
@@ -602,7 +593,7 @@ FB_FETCH_TRANSFORMED(bits_image_t * pict, int x, int y, int width, uint32_t *buf
 
     /* This allows filtering code to pretend that pixels are located at integer coordinates */
     adjust (&v, &unit, -(pixman_fixed_1 / 2));
-    
+
     if (pict->common.filter == PIXMAN_FILTER_NEAREST || pict->common.filter == PIXMAN_FILTER_FAST)
     {
 	/* Round down to closest integer, ensuring that 0.5 rounds to 0, not 1 */
@@ -611,7 +602,6 @@ FB_FETCH_TRANSFORMED(bits_image_t * pict, int x, int y, int width, uint32_t *buf
         if (pict->common.repeat == PIXMAN_REPEAT_NORMAL)
         {
             fbFetchTransformed_Nearest_Normal(pict, width, buffer, mask, maskBits, affine, v, unit);
-
         }
         else if (pict->common.repeat == PIXMAN_REPEAT_PAD)
         {
@@ -647,27 +637,52 @@ FB_FETCH_TRANSFORMED(bits_image_t * pict, int x, int y, int width, uint32_t *buf
     }
 }
 
+void
+ACCESS(fbFetchTransformed64)(bits_image_t * pict, int x, int y, int width,
+                             uint64_t *buffer, uint64_t *mask, uint32_t maskBits)
+{
+    // TODO: Don't lose precision for wide pictures!
+    uint32_t *mask8 = NULL;
+
+    // Contract the mask image, if one exists, so that the 32-bit fetch function
+    // can use it.
+    if (mask) {
+        mask8 = pixman_malloc_ab(width, sizeof(uint32_t));
+        pixman_contract(mask8, mask, width);
+    }
+
+    // Fetch the image into the first half of buffer.
+    ACCESS(fbFetchTransformed)(pict, x, y, width, (uint32_t*)buffer, mask8,
+                               maskBits);
+
+    // Expand from 32bpp to 64bpp in place.
+    pixman_expand(buffer, (uint32_t*)buffer, PIXMAN_a8r8g8b8, width);
+
+    free(mask8);
+}
+
 #define SCANLINE_BUFFER_LENGTH 2048
 
 void
-FB_FETCH_EXTERNAL_ALPHA(bits_image_t * pict, int x, int y, int width,
-                        uint32_t *buffer, uint32_t *mask, uint32_t maskBits)
+ACCESS(fbFetchExternalAlpha)(bits_image_t * pict, int x, int y, int width,
+                             uint32_t *buffer, uint32_t *mask,
+                             uint32_t maskBits)
 {
     int i;
     uint32_t _alpha_buffer[SCANLINE_BUFFER_LENGTH];
     uint32_t *alpha_buffer = _alpha_buffer;
 
     if (!pict->common.alpha_map) {
-        FB_FETCH_TRANSFORMED (pict, x, y, width, buffer, mask, maskBits);
+        ACCESS(fbFetchTransformed) (pict, x, y, width, buffer, mask, maskBits);
 	return;
     }
     if (width > SCANLINE_BUFFER_LENGTH)
         alpha_buffer = (uint32_t *) pixman_malloc_ab (width, sizeof(uint32_t));
 
-    FB_FETCH_TRANSFORMED(pict, x, y, width, buffer, mask, maskBits);
-    FB_FETCH_TRANSFORMED((bits_image_t *)pict->common.alpha_map, x - pict->common.alpha_origin.x,
-			 y - pict->common.alpha_origin.y, width, alpha_buffer,
-			 mask, maskBits);
+    ACCESS(fbFetchTransformed)(pict, x, y, width, buffer, mask, maskBits);
+    ACCESS(fbFetchTransformed)((bits_image_t *)pict->common.alpha_map, x - pict->common.alpha_origin.x,
+                               y - pict->common.alpha_origin.y, width,
+                               alpha_buffer, mask, maskBits);
     for (i = 0; i < width; ++i) {
         if (!mask || mask[i] & maskBits)
 	{
@@ -684,14 +699,53 @@ FB_FETCH_EXTERNAL_ALPHA(bits_image_t * pict, int x, int y, int width,
 }
 
 void
-FB_STORE_EXTERNAL_ALPHA(bits_image_t * pict, int x, int y, int width,
-                        uint32_t *buffer)
+ACCESS(fbFetchExternalAlpha64)(bits_image_t * pict, int x, int y, int width,
+                               uint64_t *buffer, uint64_t *mask,
+                               uint32_t maskBits)
+{
+    int i;
+    uint64_t _alpha_buffer[SCANLINE_BUFFER_LENGTH];
+    uint64_t *alpha_buffer = _alpha_buffer;
+    uint64_t maskBits64;
+
+    if (!pict->common.alpha_map) {
+        ACCESS(fbFetchTransformed64) (pict, x, y, width, buffer, mask, maskBits);
+	return;
+    }
+    if (width > SCANLINE_BUFFER_LENGTH)
+        alpha_buffer = (uint64_t *) pixman_malloc_ab (width, sizeof(uint64_t));
+
+    ACCESS(fbFetchTransformed64)(pict, x, y, width, buffer, mask, maskBits);
+    ACCESS(fbFetchTransformed64)((bits_image_t *)pict->common.alpha_map, x - pict->common.alpha_origin.x,
+                                 y - pict->common.alpha_origin.y, width,
+                                 alpha_buffer, mask, maskBits);
+
+    pixman_expand(&maskBits64, &maskBits, PIXMAN_a8r8g8b8, 1);
+
+    for (i = 0; i < width; ++i) {
+        if (!mask || mask[i] & maskBits64)
+	{
+	    int64_t a = alpha_buffer[i]>>48;
+	    *(buffer + i) = (a << 48)
+		| (div_65535(Red64(*(buffer + i)) * a) << 32)
+		| (div_65535(Green64(*(buffer + i)) * a) << 16)
+		| (div_65535(Blue64(*(buffer + i)) * a));
+	}
+    }
+
+    if (alpha_buffer != _alpha_buffer)
+        free(alpha_buffer);
+}
+
+void
+ACCESS(fbStoreExternalAlpha)(bits_image_t * pict, int x, int y, int width,
+                             uint32_t *buffer)
 {
     uint32_t *bits, *alpha_bits;
     int32_t stride, astride;
     int ax, ay;
-    storeProc store;
-    storeProc astore;
+    storeProc32 store;
+    storeProc32 astore;
     const pixman_indexed_t * indexed = pict->indexed;
     const pixman_indexed_t * aindexed;
 
@@ -702,8 +756,8 @@ FB_STORE_EXTERNAL_ALPHA(bits_image_t * pict, int x, int y, int width,
 	return;
     }
 
-    store = STORE_PROC_FOR_PICTURE(pict);
-    astore = STORE_PROC_FOR_PICTURE(pict->common.alpha_map);
+    store = ACCESS(pixman_storeProcForPicture32)(pict);
+    astore = ACCESS(pixman_storeProcForPicture32)(pict->common.alpha_map);
     aindexed = pict->common.alpha_map->indexed;
 
     ax = x;
@@ -724,3 +778,36 @@ FB_STORE_EXTERNAL_ALPHA(bits_image_t * pict, int x, int y, int width,
 	   alpha_bits, buffer, ax - pict->common.alpha_origin.x, width, aindexed);
 }
 
+void
+ACCESS(fbStoreExternalAlpha64)(bits_image_t * pict, int x, int y, int width,
+                               uint64_t *buffer)
+{
+    uint32_t *bits, *alpha_bits;
+    int32_t stride, astride;
+    int ax, ay;
+    storeProc64 store;
+    storeProc64 astore;
+    const pixman_indexed_t * indexed = pict->indexed;
+    const pixman_indexed_t * aindexed;
+
+    store = ACCESS(pixman_storeProcForPicture64)(pict);
+    astore = ACCESS(pixman_storeProcForPicture64)(pict->common.alpha_map);
+    aindexed = pict->common.alpha_map->indexed;
+
+    ax = x;
+    ay = y;
+
+    bits = pict->bits;
+    stride = pict->rowstride;
+
+    alpha_bits = pict->common.alpha_map->bits;
+    astride = pict->common.alpha_map->rowstride;
+
+    bits       += y*stride;
+    alpha_bits += (ay - pict->common.alpha_origin.y)*astride;
+
+
+    store((pixman_image_t *)pict, bits, buffer, x, width, indexed);
+    astore((pixman_image_t *)pict->common.alpha_map,
+	   alpha_bits, buffer, ax - pict->common.alpha_origin.x, width, aindexed);
+}
diff --git a/lib/pixman/pixman/pixman-trap.c b/lib/pixman/pixman/pixman-trap.c
index 0dca6ccf0..28dacafcc 100644
--- a/lib/pixman/pixman/pixman-trap.c
+++ b/lib/pixman/pixman/pixman-trap.c
@@ -29,7 +29,7 @@
 
 typedef uint32_t FbBits;
 
-void
+PIXMAN_EXPORT void
 pixman_add_traps (pixman_image_t *	image,
 		  int16_t	x_off,
 		  int16_t	y_off,
@@ -115,12 +115,12 @@ dump_image (pixman_image_t *image,
     }
 }
 
-void
-pixman_add_trapezoids       (pixman_image_t      *image,
-			     int16_t              x_off,
-			     int                      y_off,
-			     int                      ntraps,
-			     const pixman_trapezoid_t *traps)
+PIXMAN_EXPORT void
+pixman_add_trapezoids (pixman_image_t           *image,
+		       int16_t                   x_off,
+		       int                       y_off,
+		       int                       ntraps,
+		       const pixman_trapezoid_t *traps)
 {
     int i;
 
@@ -143,7 +143,7 @@ pixman_add_trapezoids       (pixman_image_t      *image,
 #endif
 }
 
-void
+PIXMAN_EXPORT void
 pixman_rasterize_trapezoid (pixman_image_t *    image,
 			    const pixman_trapezoid_t *trap,
 			    int			x_off,
diff --git a/lib/pixman/pixman/pixman-utils.c b/lib/pixman/pixman/pixman-utils.c
index b100d096c..21d9a2fd2 100644
--- a/lib/pixman/pixman/pixman-utils.c
+++ b/lib/pixman/pixman/pixman-utils.c
@@ -30,7 +30,7 @@
 #include "pixman-private.h"
 #include "pixman-mmx.h"
 
-pixman_bool_t
+PIXMAN_EXPORT pixman_bool_t
 pixman_transform_point_3d (pixman_transform_t *transform,
 			   pixman_vector_t *vector)
 {
@@ -62,7 +62,7 @@ pixman_transform_point_3d (pixman_transform_t *transform,
     return TRUE;
 }
 
-pixman_bool_t
+PIXMAN_EXPORT pixman_bool_t
 pixman_blt (uint32_t *src_bits,
 	    uint32_t *dst_bits,
 	    int src_stride,
@@ -156,7 +156,7 @@ pixman_fill32 (uint32_t *bits,
     }
 }
 
-pixman_bool_t
+PIXMAN_EXPORT pixman_bool_t
 pixman_fill (uint32_t *bits,
 	     int stride,
 	     int bpp,
@@ -204,7 +204,7 @@ pixman_fill (uint32_t *bits,
  * grid row
  */
 
-pixman_fixed_t
+PIXMAN_EXPORT pixman_fixed_t
 pixman_sample_ceil_y (pixman_fixed_t y, int n)
 {
     pixman_fixed_t   f = pixman_fixed_frac(y);
@@ -225,7 +225,7 @@ pixman_sample_ceil_y (pixman_fixed_t y, int n)
  * Compute the largest value no greater than y which is on a
  * grid row
  */
-pixman_fixed_t
+PIXMAN_EXPORT pixman_fixed_t
 pixman_sample_floor_y (pixman_fixed_t y, int n)
 {
     pixman_fixed_t   f = pixman_fixed_frac(y);
@@ -243,7 +243,7 @@ pixman_sample_floor_y (pixman_fixed_t y, int n)
 /*
  * Step an edge by any amount (including negative values)
  */
-void
+PIXMAN_EXPORT void
 pixman_edge_step (pixman_edge_t *e, int n)
 {
     pixman_fixed_48_16_t	ne;
@@ -298,7 +298,7 @@ _pixman_edge_tMultiInit (pixman_edge_t *e, int n, pixman_fixed_t *stepx_p, pixma
  * Initialize one edge structure given the line endpoints and a
  * starting y value
  */
-void
+PIXMAN_EXPORT void
 pixman_edge_init (pixman_edge_t	*e,
 		  int		n,
 		  pixman_fixed_t		y_start,
@@ -342,7 +342,7 @@ pixman_edge_init (pixman_edge_t	*e,
  * Initialize one edge structure given a line, starting y value
  * and a pixel offset for the line
  */
-void
+PIXMAN_EXPORT void
 pixman_line_fixed_edge_init (pixman_edge_t *e,
 			     int	    n,
 			     pixman_fixed_t	    y,
@@ -428,7 +428,7 @@ pixman_malloc_abc (unsigned int a,
  *
  * Return value: the encoded version.
  **/
-int
+PIXMAN_EXPORT int
 pixman_version (void)
 {
     return PIXMAN_VERSION;
@@ -445,7 +445,7 @@ pixman_version (void)
  *
  * Return value: a string containing the version.
  **/
-const char*
+PIXMAN_EXPORT const char*
 pixman_version_string (void)
 {
     return PIXMAN_VERSION_STRING;
@@ -462,11 +462,13 @@ pixman_version_string (void)
  * Currently, all pixman_format_code_t values are supported
  * except for the YUV formats.
  **/
-pixman_bool_t
+PIXMAN_EXPORT pixman_bool_t
 pixman_format_supported_destination (pixman_format_code_t format)
 {
     switch (format) {
     /* 32 bpp formats */
+    case PIXMAN_a2b10g10r10:
+    case PIXMAN_x2b10g10r10:
     case PIXMAN_a8r8g8b8:
     case PIXMAN_x8r8g8b8:
     case PIXMAN_a8b8g8r8:
@@ -530,11 +532,13 @@ pixman_format_supported_destination (pixman_format_code_t format)
  *
  * Currently, all pixman_format_code_t values are supported.
  **/
-pixman_bool_t
+PIXMAN_EXPORT pixman_bool_t
 pixman_format_supported_source (pixman_format_code_t format)
 {
     switch (format) {
     /* 32 bpp formats */
+    case PIXMAN_a2b10g10r10:
+    case PIXMAN_x2b10g10r10:
     case PIXMAN_a8r8g8b8:
     case PIXMAN_x8r8g8b8:
     case PIXMAN_a8b8g8r8:
diff --git a/lib/pixman/pixman/pixman-version.h b/lib/pixman/pixman/pixman-version.h
deleted file mode 100644
index a57031484..000000000
--- a/lib/pixman/pixman/pixman-version.h
+++ /dev/null
@@ -1,50 +0,0 @@
-/*
- * Copyright © 2008 Red Hat, 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 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.
- *
- * Author: Carl D. Worth <cworth@cworth.org>
- */
-
-#ifndef PIXMAN_VERSION_H__
-#define PIXMAN_VERSION_H__
-
-#ifndef PIXMAN_H__
-#  error pixman-version.h should only be included by pixman.h
-#endif
-
-#define PIXMAN_VERSION_MAJOR 0
-#define PIXMAN_VERSION_MINOR 10
-#define PIXMAN_VERSION_MICRO 0
-
-#define PIXMAN_VERSION_STRING "0.10.0"
-
-#define PIXMAN_VERSION_ENCODE(major, minor, micro) (	\
-	  ((major) * 10000)				\
-	+ ((minor) *   100)				\
-	+ ((micro) *     1))
-
-#define PIXMAN_VERSION PIXMAN_VERSION_ENCODE(	\
-	PIXMAN_VERSION_MAJOR,			\
-	PIXMAN_VERSION_MINOR,			\
-	PIXMAN_VERSION_MICRO)
-
-#endif /* PIXMAN_VERSION_H__ */
diff --git a/lib/pixman/pixman/pixman-vmx.c b/lib/pixman/pixman/pixman-vmx.c
new file mode 100644
index 000000000..8c8a2a364
--- /dev/null
+++ b/lib/pixman/pixman/pixman-vmx.c
@@ -0,0 +1,1068 @@
+/*
+ * Copyright © 2007 Luca Barbato
+ *
+ * Permission to use, copy, modify, distribute, and sell this software and its
+ * documentation for any purpose is hereby granted without fee, provided that
+ * the above copyright notice appear in all copies and that both that
+ * copyright notice and this permission notice appear in supporting
+ * documentation, and that the name of Luca Barbato not be used in advertising or
+ * publicity pertaining to distribution of the software without specific,
+ * written prior permission.  Luca Barbato makes no representations about the
+ * suitability of this software for any purpose.  It is provided "as is"
+ * without express or implied warranty.
+ *
+ * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS
+ * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
+ * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY
+ * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN
+ * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
+ * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
+ * SOFTWARE.
+ *
+ * Author:  Luca Barbato (lu_zero@gentoo.org)
+ *
+ * Based on fbmmx.c by Owen Taylor, Søren Sandmann and Nicholas Miell
+ */
+
+#include <config.h>
+#include "pixman-vmx.h"
+#include "pixman-combine32.h"
+#include <altivec.h>
+
+#ifdef __GNUC__
+#   define inline __inline__ __attribute__ ((__always_inline__))
+#endif
+
+static inline vector unsigned int
+splat_alpha (vector unsigned int pix) {
+    return vec_perm (pix, pix,
+    (vector unsigned char)AVV(0x00,0x00,0x00,0x00, 0x04,0x04,0x04,0x04,
+                               0x08,0x08,0x08,0x08, 0x0C,0x0C,0x0C,0x0C));
+}
+
+static inline vector unsigned int
+pix_multiply (vector unsigned int p, vector unsigned int a)
+{
+    vector unsigned short hi, lo, mod;
+    /* unpack to short */
+    hi = (vector unsigned short)
+                    vec_mergeh ((vector unsigned char)AVV(0),
+                                (vector unsigned char)p);
+    mod = (vector unsigned short)
+                    vec_mergeh ((vector unsigned char)AVV(0),
+                                (vector unsigned char)a);
+
+    hi = vec_mladd (hi, mod, (vector unsigned short)
+                            AVV(0x0080,0x0080,0x0080,0x0080,
+                                 0x0080,0x0080,0x0080,0x0080));
+
+    hi = vec_adds (hi, vec_sr (hi, vec_splat_u16 (8)));
+
+    hi = vec_sr (hi, vec_splat_u16 (8));
+
+    /* unpack to short */
+    lo = (vector unsigned short)
+                    vec_mergel ((vector unsigned char)AVV(0),
+                                (vector unsigned char)p);
+    mod = (vector unsigned short)
+                    vec_mergel ((vector unsigned char)AVV(0),
+                                (vector unsigned char)a);
+
+    lo = vec_mladd (lo, mod, (vector unsigned short)
+                            AVV(0x0080,0x0080,0x0080,0x0080,
+                                 0x0080,0x0080,0x0080,0x0080));
+
+    lo = vec_adds (lo, vec_sr (lo, vec_splat_u16 (8)));
+
+    lo = vec_sr (lo, vec_splat_u16 (8));
+
+    return (vector unsigned int)vec_packsu (hi, lo);
+}
+
+static inline vector unsigned int
+pix_add (vector unsigned int a, vector unsigned int b)
+{
+    return (vector unsigned int)vec_adds ((vector unsigned char)a,
+                     (vector unsigned char)b);
+}
+
+static inline vector unsigned int
+pix_add_mul (vector unsigned int x, vector unsigned int a,
+             vector unsigned int y, vector unsigned int b)
+{
+    vector unsigned short hi, lo, mod, hiy, loy, mody;
+
+    hi = (vector unsigned short)
+                    vec_mergeh ((vector unsigned char)AVV(0),
+                                (vector unsigned char)x);
+    mod = (vector unsigned short)
+                    vec_mergeh ((vector unsigned char)AVV(0),
+                                (vector unsigned char)a);
+    hiy = (vector unsigned short)
+                    vec_mergeh ((vector unsigned char)AVV(0),
+                                (vector unsigned char)y);
+    mody = (vector unsigned short)
+                    vec_mergeh ((vector unsigned char)AVV(0),
+                                (vector unsigned char)b);
+
+    hi = vec_mladd (hi, mod, (vector unsigned short)
+                             AVV(0x0080,0x0080,0x0080,0x0080,
+                                  0x0080,0x0080,0x0080,0x0080));
+
+    hi = vec_mladd (hiy, mody, hi);
+
+    hi = vec_adds (hi, vec_sr (hi, vec_splat_u16 (8)));
+
+    hi = vec_sr (hi, vec_splat_u16 (8));
+
+    lo = (vector unsigned short)
+                    vec_mergel ((vector unsigned char)AVV(0),
+                                (vector unsigned char)x);
+    mod = (vector unsigned short)
+                    vec_mergel ((vector unsigned char)AVV(0),
+                                (vector unsigned char)a);
+
+    loy = (vector unsigned short)
+                    vec_mergel ((vector unsigned char)AVV(0),
+                                (vector unsigned char)y);
+    mody = (vector unsigned short)
+                    vec_mergel ((vector unsigned char)AVV(0),
+                                (vector unsigned char)b);
+
+    lo = vec_mladd (lo, mod, (vector unsigned short)
+                             AVV(0x0080,0x0080,0x0080,0x0080,
+                                  0x0080,0x0080,0x0080,0x0080));
+
+    lo = vec_mladd (loy, mody, lo);
+
+    lo = vec_adds (lo, vec_sr (lo, vec_splat_u16 (8)));
+
+    lo = vec_sr (lo, vec_splat_u16 (8));
+
+    return (vector unsigned int)vec_packsu (hi, lo);
+}
+
+static inline vector unsigned int
+negate (vector unsigned int src)
+{
+    return vec_nor (src, src);
+}
+/* dest*~srca + src */
+static inline vector unsigned int
+over (vector unsigned int src, vector unsigned int srca,
+      vector unsigned int dest)
+{
+    vector unsigned char tmp = (vector unsigned char)
+                                pix_multiply (dest, negate (srca));
+    tmp = vec_adds ((vector unsigned char)src, tmp);
+    return (vector unsigned int)tmp;
+}
+
+/* in == pix_multiply */
+#define in_over(src, srca, mask, dest) over (pix_multiply (src, mask),\
+                                             pix_multiply (srca, mask), dest)
+
+
+#define COMPUTE_SHIFT_MASK(source) \
+    source ## _mask = vec_lvsl (0, source);
+
+#define COMPUTE_SHIFT_MASKS(dest, source) \
+    dest ## _mask = vec_lvsl (0, dest); \
+    source ## _mask = vec_lvsl (0, source); \
+    store_mask = vec_lvsr (0, dest);
+
+#define COMPUTE_SHIFT_MASKC(dest, source, mask) \
+    mask ## _mask = vec_lvsl (0, mask); \
+    dest ## _mask = vec_lvsl (0, dest); \
+    source ## _mask = vec_lvsl (0, source); \
+    store_mask = vec_lvsr (0, dest);
+
+/* notice you have to declare temp vars...
+ * Note: tmp3 and tmp4 must remain untouched!
+ */
+
+#define LOAD_VECTORS(dest, source) \
+        tmp1 = (typeof(tmp1))vec_ld(0, source); \
+        tmp2 = (typeof(tmp2))vec_ld(15, source); \
+        tmp3 = (typeof(tmp3))vec_ld(0, dest); \
+        v ## source = (typeof(v ## source)) \
+                       vec_perm(tmp1, tmp2, source ## _mask); \
+        tmp4 = (typeof(tmp4))vec_ld(15, dest); \
+        v ## dest = (typeof(v ## dest)) \
+                     vec_perm(tmp3, tmp4, dest ## _mask);
+
+#define LOAD_VECTORSC(dest, source, mask) \
+        tmp1 = (typeof(tmp1))vec_ld(0, source); \
+        tmp2 = (typeof(tmp2))vec_ld(15, source); \
+        tmp3 = (typeof(tmp3))vec_ld(0, dest); \
+        v ## source = (typeof(v ## source)) \
+                       vec_perm(tmp1, tmp2, source ## _mask); \
+        tmp4 = (typeof(tmp4))vec_ld(15, dest); \
+        tmp1 = (typeof(tmp1))vec_ld(0, mask); \
+        v ## dest = (typeof(v ## dest)) \
+                     vec_perm(tmp3, tmp4, dest ## _mask); \
+        tmp2 = (typeof(tmp2))vec_ld(15, mask); \
+        v ## mask = (typeof(v ## mask)) \
+                     vec_perm(tmp1, tmp2, mask ## _mask);
+#define STORE_VECTOR(dest) \
+        edges = vec_perm (tmp4, tmp3, dest ## _mask); \
+        tmp3 = vec_perm ((vector unsigned char)v ## dest, edges, store_mask); \
+        tmp1 = vec_perm (edges, (vector unsigned char)v ## dest, store_mask); \
+        vec_st ((vector unsigned int) tmp3, 15, dest ); \
+        vec_st ((vector unsigned int) tmp1, 0, dest );
+
+static FASTCALL void
+vmxCombineMaskU (uint32_t *src, const uint32_t *msk, int width)
+{
+    int i;
+    vector unsigned int  vsrc, vmsk;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         src_mask, msk_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKS(src, msk)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORS(src, msk)
+
+        vsrc = pix_multiply (vsrc, splat_alpha (vmsk));
+
+        STORE_VECTOR(src)
+
+        msk+=4;
+        src+=4;
+    }
+
+    for (i = width%4; --i >= 0;) {
+        uint32_t a = msk[i] >> 24;
+        uint32_t s = src[i];
+        FbByteMul (s, a);
+        src[i] = s;
+    }
+}
+
+static FASTCALL void
+vmxCombineOverU (uint32_t *dest, const uint32_t *src, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKS(dest, src)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORS(dest, src)
+
+        vdest = over (vsrc, splat_alpha (vsrc), vdest);
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        uint32_t ia = Alpha (~s);
+
+        FbByteMulAdd (d, ia, s);
+        dest[i] = d;
+    }
+}
+
+
+static FASTCALL void
+vmxCombineOverReverseU (uint32_t *dest, const uint32_t *src, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKS(dest, src)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORS(dest, src)
+
+        vdest = over (vdest, splat_alpha (vdest) , vsrc);
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        uint32_t ia = Alpha (~dest[i]);
+
+        FbByteMulAdd (s, ia, d);
+        dest[i] = s;
+    }
+}
+
+static FASTCALL void
+vmxCombineInU (uint32_t *dest, const uint32_t *src, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKS(dest, src)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORS(dest, src)
+
+        vdest = pix_multiply (vsrc, splat_alpha (vdest));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+
+        uint32_t s = src[i];
+        uint32_t a = Alpha (dest[i]);
+        FbByteMul (s, a);
+        dest[i] = s;
+    }
+}
+
+static FASTCALL void
+vmxCombineInReverseU (uint32_t *dest, const uint32_t *src, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKS(dest, src)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORS(dest, src)
+
+        vdest = pix_multiply (vdest, splat_alpha (vsrc));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t d = dest[i];
+        uint32_t a = Alpha (src[i]);
+        FbByteMul (d, a);
+        dest[i] = d;
+    }
+}
+
+static FASTCALL void
+vmxCombineOutU (uint32_t *dest, const uint32_t *src, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKS(dest, src)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORS(dest, src)
+
+        vdest = pix_multiply (vsrc, splat_alpha (negate (vdest)));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t s = src[i];
+        uint32_t a = Alpha (~dest[i]);
+        FbByteMul (s, a);
+        dest[i] = s;
+    }
+}
+
+static FASTCALL void
+vmxCombineOutReverseU (uint32_t *dest, const uint32_t *src, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKS(dest, src)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORS(dest, src)
+
+        vdest = pix_multiply (vdest, splat_alpha (negate (vsrc)));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t d = dest[i];
+        uint32_t a = Alpha (~src[i]);
+        FbByteMul (d, a);
+        dest[i] = d;
+    }
+}
+
+static FASTCALL void
+vmxCombineAtopU (uint32_t *dest, const uint32_t *src, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKS(dest, src)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORS(dest, src)
+
+        vdest = pix_add_mul (vsrc, splat_alpha (vdest),
+                            vdest, splat_alpha (negate (vsrc)));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        uint32_t dest_a = Alpha (d);
+        uint32_t src_ia = Alpha (~s);
+
+        FbByteAddMul (s, dest_a, d, src_ia);
+        dest[i] = s;
+    }
+}
+
+static FASTCALL void
+vmxCombineAtopReverseU (uint32_t *dest, const uint32_t *src, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKS(dest, src)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORS(dest, src)
+
+        vdest = pix_add_mul (vdest, splat_alpha (vsrc),
+                            vsrc, splat_alpha (negate (vdest)));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        uint32_t src_a = Alpha (s);
+        uint32_t dest_ia = Alpha (~d);
+
+        FbByteAddMul (s, dest_ia, d, src_a);
+        dest[i] = s;
+    }
+}
+
+static FASTCALL void
+vmxCombineXorU (uint32_t *dest, const uint32_t *src, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKS(dest, src)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORS (dest, src)
+
+        vdest = pix_add_mul (vsrc, splat_alpha (negate (vdest)),
+                            vdest, splat_alpha (negate (vsrc)));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        uint32_t src_ia = Alpha (~s);
+        uint32_t dest_ia = Alpha (~d);
+
+        FbByteAddMul (s, dest_ia, d, src_ia);
+        dest[i] = s;
+    }
+}
+
+static FASTCALL void
+vmxCombineAddU (uint32_t *dest, const uint32_t *src, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKS(dest, src)
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORS(dest, src)
+
+        vdest = pix_add (vsrc, vdest);
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        FbByteAdd (d, s);
+        dest[i] = d;
+    }
+}
+
+static FASTCALL void
+vmxCombineSrcC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc, vmask;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, mask_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKC(dest, src, mask);
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORSC(dest, src, mask)
+
+        vdest = pix_multiply (vsrc, vmask);
+
+        STORE_VECTOR(dest)
+
+        mask+=4;
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t a = mask[i];
+        uint32_t s = src[i];
+        FbByteMulC (s, a);
+        dest[i] = s;
+    }
+}
+
+static FASTCALL void
+vmxCombineOverC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc, vmask;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, mask_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKC(dest, src, mask);
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORSC(dest, src, mask)
+
+        vdest = in_over (vsrc, splat_alpha (vsrc), vmask, vdest);
+
+        STORE_VECTOR(dest)
+
+        mask+=4;
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t a = mask[i];
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        FbByteMulC (s, a);
+        FbByteMulAddC (d, ~a, s);
+        dest[i] = d;
+    }
+}
+
+static FASTCALL void
+vmxCombineOverReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc, vmask;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, mask_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKC(dest, src, mask);
+    /* printf("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORSC (dest, src, mask)
+
+        vdest = over (vdest, splat_alpha (vdest), pix_multiply (vsrc, vmask));
+
+        STORE_VECTOR(dest)
+
+        mask+=4;
+        src+=4;
+        dest+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t a = mask[i];
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        uint32_t da = Alpha (d);
+        FbByteMulC (s, a);
+        FbByteMulAddC (s, ~da, d);
+        dest[i] = s;
+    }
+}
+
+static FASTCALL void
+vmxCombineInC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc, vmask;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, mask_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKC(dest, src, mask)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORSC(dest, src, mask)
+
+        vdest = pix_multiply (pix_multiply (vsrc, vmask), splat_alpha (vdest));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+        mask+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t a = mask[i];
+        uint32_t s = src[i];
+        uint32_t da = Alpha (dest[i]);
+        FbByteMul (s, a);
+        FbByteMul (s, da);
+        dest[i] = s;
+    }
+}
+
+static FASTCALL void
+vmxCombineInReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc, vmask;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, mask_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKC(dest, src, mask)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORSC(dest, src, mask)
+
+        vdest = pix_multiply (vdest, pix_multiply (vmask, splat_alpha (vsrc)));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+        mask+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t a = mask[i];
+        uint32_t d = dest[i];
+        uint32_t sa = Alpha (src[i]);
+        FbByteMul (a, sa);
+        FbByteMulC (d, a);
+        dest[i] = d;
+    }
+}
+
+static FASTCALL void
+vmxCombineOutC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc, vmask;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, mask_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKC(dest, src, mask)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORSC(dest, src, mask)
+
+        vdest = pix_multiply (pix_multiply (vsrc, vmask), splat_alpha (vdest));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+        mask+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t a = mask[i];
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        uint32_t da = Alpha (~d);
+        FbByteMulC (s, a);
+        FbByteMulC (s, da);
+        dest[i] = s;
+    }
+}
+
+static FASTCALL void
+vmxCombineOutReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc, vmask;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, mask_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKC(dest, src, mask)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORSC(dest, src, mask)
+
+        vdest = pix_multiply (vdest,
+                             negate (pix_multiply (vmask, splat_alpha (vsrc))));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+        mask+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t a = mask[i];
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        uint32_t sa = Alpha (s);
+        FbByteMulC (a, sa);
+        FbByteMulC (d, ~a);
+        dest[i] = d;
+    }
+}
+
+static FASTCALL void
+vmxCombineAtopC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc, vmask;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, mask_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKC(dest, src, mask)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORSC(dest, src, mask)
+
+        vdest = pix_add_mul (pix_multiply (vsrc, vmask), splat_alpha (vdest),
+                            vdest,
+                            negate (pix_multiply (vmask,
+                                                splat_alpha (vmask))));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+        mask+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t a = mask[i];
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        uint32_t sa = Alpha (s);
+        uint32_t da = Alpha (d);
+
+        FbByteMulC (s, a);
+        FbByteMul (a, sa);
+        FbByteAddMulC (d, ~a, s, da);
+        dest[i] = d;
+    }
+}
+
+static FASTCALL void
+vmxCombineAtopReverseC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc, vmask;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, mask_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKC(dest, src, mask)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORSC(dest, src, mask)
+
+        vdest = pix_add_mul (vdest,
+                            pix_multiply (vmask, splat_alpha (vsrc)),
+                            pix_multiply (vsrc, vmask),
+                            negate (splat_alpha (vdest)));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+        mask+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t a = mask[i];
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        uint32_t sa = Alpha (s);
+        uint32_t da = Alpha (d);
+
+        FbByteMulC (s, a);
+        FbByteMul (a, sa);
+        FbByteAddMulC (d, a, s, ~da);
+        dest[i] = d;
+    }
+}
+
+static FASTCALL void
+vmxCombineXorC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc, vmask;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, mask_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKC(dest, src, mask)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORSC(dest, src, mask)
+
+        vdest = pix_add_mul (vdest,
+                            negate (pix_multiply (vmask, splat_alpha (vsrc))),
+                            pix_multiply (vsrc, vmask),
+                            negate (splat_alpha (vdest)));
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+        mask+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t a = mask[i];
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+        uint32_t sa = Alpha (s);
+        uint32_t da = Alpha (d);
+
+        FbByteMulC (s, a);
+        FbByteMul (a, sa);
+        FbByteAddMulC (d, ~a, s, ~da);
+        dest[i] = d;
+    }
+}
+
+static FASTCALL void
+vmxCombineAddC (uint32_t *dest, uint32_t *src, uint32_t *mask, int width)
+{
+    int i;
+    vector unsigned int  vdest, vsrc, vmask;
+    vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
+                         dest_mask, mask_mask, src_mask, store_mask;
+
+    COMPUTE_SHIFT_MASKC(dest, src, mask)
+
+    /* printf ("%s\n",__PRETTY_FUNCTION__); */
+    for (i = width/4; i > 0; i--) {
+
+        LOAD_VECTORSC(dest, src, mask)
+
+        vdest = pix_add (pix_multiply (vsrc, vmask), vdest);
+
+        STORE_VECTOR(dest)
+
+        src+=4;
+        dest+=4;
+        mask+=4;
+    }
+
+    for (i = width%4; --i >=0;) {
+        uint32_t a = mask[i];
+        uint32_t s = src[i];
+        uint32_t d = dest[i];
+
+        FbByteMulC (s, a);
+        FbByteAdd (s, d);
+        dest[i] = s;
+    }
+}
+
+
+#if 0
+void
+fbCompositeSolid_nx8888vmx (pixman_operator_t	op,
+			    pixman_image_t * pSrc,
+			    pixman_image_t * pMask,
+			    pixman_image_t * pDst,
+			    int16_t	xSrc,
+			    int16_t	ySrc,
+			    int16_t	xMask,
+			    int16_t	yMask,
+			    int16_t	xDst,
+			    int16_t	yDst,
+			    uint16_t	width,
+			    uint16_t	height)
+{
+    uint32_t	src;
+    uint32_t	*dstLine, *dst;
+    int	dstStride;
+
+    fbComposeGetSolid (pSrc, pDst, src);
+
+    if (src >> 24 == 0)
+	return;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
+
+    while (height--)
+    {
+	dst = dstLine;
+	dstLine += dstStride;
+	/* XXX vmxCombineOverU (dst, src, width); */
+    }
+}
+
+void
+fbCompositeSolid_nx0565vmx (pixman_operator_t	op,
+			    pixman_image_t * pSrc,
+			    pixman_image_t * pMask,
+			    pixman_image_t * pDst,
+			    int16_t	xSrc,
+			    int16_t	ySrc,
+			    int16_t	xMask,
+			    int16_t	yMask,
+			    int16_t	xDst,
+			    int16_t	yDst,
+			    uint16_t	width,
+			    uint16_t	height)
+{
+    uint32_t	src;
+    uint16_t	*dstLine, *dst;
+    uint16_t	w;
+    int	dstStride;
+
+    fbComposeGetSolid (pSrc, pDst, src);
+
+    if (src >> 24 == 0)
+	return;
+
+    fbComposeGetStart (pDst, xDst, yDst, uint16_t, dstStride, dstLine, 1);
+
+    while (height--)
+    {
+	dst = dstLine;
+	dstLine += dstStride;
+       vmxCombineOverU565(dst, src, width);
+    }
+}
+
+#endif
+
+void fbComposeSetupVMX (void)
+{
+    /* check if we have VMX support and initialize accordingly */
+    if (pixman_have_vmx ()) {
+        pixman_composeFunctions.combineU[PIXMAN_OP_OVER] = vmxCombineOverU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_OVER_REVERSE] = vmxCombineOverReverseU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_IN] = vmxCombineInU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_IN_REVERSE] = vmxCombineInReverseU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_OUT] = vmxCombineOutU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_OUT_REVERSE] = vmxCombineOutReverseU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_ATOP] = vmxCombineAtopU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_ATOP_REVERSE] = vmxCombineAtopReverseU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_XOR] = vmxCombineXorU;
+        pixman_composeFunctions.combineU[PIXMAN_OP_ADD] = vmxCombineAddU;
+
+        pixman_composeFunctions.combineC[PIXMAN_OP_SRC] = vmxCombineSrcC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_OVER] = vmxCombineOverC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_OVER_REVERSE] = vmxCombineOverReverseC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_IN] = vmxCombineInC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_IN_REVERSE] = vmxCombineInReverseC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_OUT] = vmxCombineOutC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_OUT_REVERSE] = vmxCombineOutReverseC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_ATOP] = vmxCombineAtopC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_ATOP_REVERSE] = vmxCombineAtopReverseC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_XOR] = vmxCombineXorC;
+        pixman_composeFunctions.combineC[PIXMAN_OP_ADD] = vmxCombineAddC;
+
+        pixman_composeFunctions.combineMaskU = vmxCombineMaskU;
+    }
+}
diff --git a/lib/pixman/pixman/pixman-vmx.h b/lib/pixman/pixman/pixman-vmx.h
new file mode 100644
index 000000000..70cb53a7f
--- /dev/null
+++ b/lib/pixman/pixman/pixman-vmx.h
@@ -0,0 +1,308 @@
+/*
+ * Copyright Â© 2007 Luca Barbato
+ *
+ * Permission to use, copy, modify, distribute, and sell this software and its
+ * documentation for any purpose is hereby granted without fee, provided that
+ * the above copyright notice appear in all copies and that both that
+ * copyright notice and this permission notice appear in supporting
+ * documentation, and that the name of Luca Barbato not be used in advertising or
+ * publicity pertaining to distribution of the software without specific,
+ * written prior permission.  Luca Barbato makes no representations about the
+ * suitability of this software for any purpose.  It is provided "as is"
+ * without express or implied warranty.
+ *
+ * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS
+ * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND
+ * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY
+ * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN
+ * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
+ * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
+ * SOFTWARE.
+ *
+ * Author:  Luca Barbato (lu_zero@gentoo.org)
+ *
+ * Based on work by Owen Taylor, Søren Sandmann and Lars Knoll
+ */
+
+#include "pixman-private.h"
+
+#ifdef USE_VMX
+
+pixman_bool_t pixman_have_vmx(void);
+
+#else
+#define pixman_have_vmx() FALSE
+#endif
+
+#ifdef USE_VMX
+
+#define AVV(x...) {x}
+
+void fbComposeSetupVMX (void);
+
+#if 0
+void fbCompositeIn_nx8x8vmx (pixman_operator_t	op,
+			     pixman_image_t * pSrc,
+			     pixman_image_t * pMask,
+			     pixman_image_t * pDst,
+			     INT16      xSrc,
+			     INT16      ySrc,
+			     INT16      xMask,
+			     INT16      yMask,
+			     INT16      xDst,
+			     INT16      yDst,
+			     CARD16     width,
+			     CARD16     height);
+
+void fbCompositeSolidMask_nx8888x0565Cvmx (pixman_operator_t      op,
+					   pixman_image_t * pSrc,
+					   pixman_image_t * pMask,
+					   pixman_image_t * pDst,
+					   INT16      xSrc,
+					   INT16      ySrc,
+					   INT16      xMask,
+					   INT16      yMask,
+					   INT16      xDst,
+					   INT16      yDst,
+					   CARD16     width,
+					   CARD16     height);
+
+void fbCompositeSrcAdd_8888x8888vmx (pixman_operator_t	op,
+				     pixman_image_t *	pSrc,
+				     pixman_image_t *	pMask,
+				     pixman_image_t *	pDst,
+				     INT16	xSrc,
+				     INT16      ySrc,
+				     INT16      xMask,
+				     INT16      yMask,
+				     INT16      xDst,
+				     INT16      yDst,
+				     CARD16     width,
+				     CARD16     height);
+
+void fbCompositeSolidMask_nx8888x8888Cvmx (pixman_operator_t	op,
+					   pixman_image_t *	pSrc,
+					   pixman_image_t *	pMask,
+					   pixman_image_t *	pDst,
+					   INT16	xSrc,
+					   INT16	ySrc,
+					   INT16	xMask,
+					   INT16	yMask,
+					   INT16	xDst,
+					   INT16	yDst,
+					   CARD16	width,
+					   CARD16	height);
+
+void fbCompositeSolidMask_nx8x8888vmx (pixman_operator_t      op,
+				       pixman_image_t * pSrc,
+				       pixman_image_t * pMask,
+				       pixman_image_t * pDst,
+				       INT16      xSrc,
+				       INT16      ySrc,
+				       INT16      xMask,
+				       INT16      yMask,
+				       INT16      xDst,
+				       INT16      yDst,
+				       CARD16     width,
+				       CARD16     height);
+
+void fbCompositeSolidMaskSrc_nx8x8888vmx (pixman_operator_t      op,
+					  pixman_image_t * pSrc,
+					  pixman_image_t * pMask,
+					  pixman_image_t * pDst,
+					  INT16      xSrc,
+					  INT16      ySrc,
+					  INT16      xMask,
+					  INT16      yMask,
+					  INT16      xDst,
+					  INT16      yDst,
+					  CARD16     width,
+					  CARD16     height);
+
+void fbCompositeSrcAdd_8888x8x8vmx (pixman_operator_t   op,
+				    pixman_image_t * pSrc,
+				    pixman_image_t * pMask,
+				    pixman_image_t * pDst,
+				    INT16      xSrc,
+				    INT16      ySrc,
+				    INT16      xMask,
+				    INT16      yMask,
+				    INT16      xDst,
+				    INT16      yDst,
+				    CARD16     width,
+				    CARD16     height);
+
+void fbCompositeIn_8x8vmx (pixman_operator_t	op,
+			   pixman_image_t * pSrc,
+			   pixman_image_t * pMask,
+			   pixman_image_t * pDst,
+			   INT16      xSrc,
+			   INT16      ySrc,
+			   INT16      xMask,
+			   INT16      yMask,
+			   INT16      xDst,
+			   INT16      yDst,
+			   CARD16     width,
+			   CARD16     height);
+
+void fbCompositeSrcAdd_8000x8000vmx (pixman_operator_t	op,
+				     pixman_image_t * pSrc,
+				     pixman_image_t * pMask,
+				     pixman_image_t * pDst,
+				     INT16      xSrc,
+				     INT16      ySrc,
+				     INT16      xMask,
+				     INT16      yMask,
+				     INT16      xDst,
+				     INT16      yDst,
+				     CARD16     width,
+				     CARD16     height);
+
+void fbCompositeSrc_8888RevNPx8888vmx (pixman_operator_t      op,
+				       pixman_image_t * pSrc,
+				       pixman_image_t * pMask,
+				       pixman_image_t * pDst,
+				       INT16      xSrc,
+				       INT16      ySrc,
+				       INT16      xMask,
+				       INT16      yMask,
+				       INT16      xDst,
+				       INT16      yDst,
+				       CARD16     width,
+				       CARD16     height);
+
+void fbCompositeSrc_8888x0565vmx (pixman_operator_t      op,
+				  pixman_image_t * pSrc,
+				  pixman_image_t * pMask,
+				  pixman_image_t * pDst,
+				  INT16      xSrc,
+				  INT16      ySrc,
+				  INT16      xMask,
+				  INT16      yMask,
+				  INT16      xDst,
+				  INT16      yDst,
+				  CARD16     width,
+				  CARD16     height);
+
+void fbCompositeSrc_8888RevNPx0565vmx (pixman_operator_t      op,
+				       pixman_image_t * pSrc,
+				       pixman_image_t * pMask,
+				       pixman_image_t * pDst,
+				       INT16      xSrc,
+				       INT16      ySrc,
+				       INT16      xMask,
+				       INT16      yMask,
+				       INT16      xDst,
+				       INT16      yDst,
+				       CARD16     width,
+				       CARD16     height);
+
+void fbCompositeSolid_nx8888vmx (pixman_operator_t		op,
+				 pixman_image_t *	pSrc,
+				 pixman_image_t *	pMask,
+				 pixman_image_t *	pDst,
+				 INT16		xSrc,
+				 INT16		ySrc,
+				 INT16		xMask,
+				 INT16		yMask,
+				 INT16		xDst,
+				 INT16		yDst,
+				 CARD16		width,
+				 CARD16		height);
+
+void fbCompositeSolid_nx0565vmx (pixman_operator_t		op,
+				 pixman_image_t *	pSrc,
+				 pixman_image_t *	pMask,
+				 pixman_image_t *	pDst,
+				 INT16		xSrc,
+				 INT16		ySrc,
+				 INT16		xMask,
+				 INT16		yMask,
+				 INT16		xDst,
+				 INT16		yDst,
+				 CARD16		width,
+				 CARD16		height);
+
+void fbCompositeSolidMask_nx8x0565vmx (pixman_operator_t      op,
+				       pixman_image_t * pSrc,
+				       pixman_image_t * pMask,
+				       pixman_image_t * pDst,
+				       INT16      xSrc,
+				       INT16      ySrc,
+				       INT16      xMask,
+				       INT16      yMask,
+				       INT16      xDst,
+				       INT16      yDst,
+				       CARD16     width,
+				       CARD16     height);
+
+void fbCompositeSrc_x888x8x8888vmx (pixman_operator_t	op,
+				    pixman_image_t *  pSrc,
+				    pixman_image_t *  pMask,
+				    pixman_image_t *  pDst,
+				    INT16	xSrc,
+				    INT16	ySrc,
+				    INT16       xMask,
+				    INT16       yMask,
+				    INT16       xDst,
+				    INT16       yDst,
+				    CARD16      width,
+				    CARD16      height);
+
+void fbCompositeSrc_8888x8x8888vmx (pixman_operator_t	op,
+				    pixman_image_t *  pSrc,
+				    pixman_image_t *  pMask,
+				    pixman_image_t *  pDst,
+				    INT16	xSrc,
+				    INT16	ySrc,
+				    INT16       xMask,
+				    INT16       yMask,
+				    INT16       xDst,
+				    INT16       yDst,
+				    CARD16      width,
+				    CARD16      height);
+
+void fbCompositeSrc_8888x8888vmx (pixman_operator_t      op,
+				  pixman_image_t * pSrc,
+				  pixman_image_t * pMask,
+				  pixman_image_t * pDst,
+				  INT16      xSrc,
+				  INT16      ySrc,
+				  INT16      xMask,
+				  INT16      yMask,
+				  INT16      xDst,
+				  INT16      yDst,
+				  CARD16     width,
+				  CARD16     height);
+
+pixman_bool_t fbCopyAreavmx (FbPixels	*pSrc,
+		    FbPixels	*pDst,
+		    int		src_x,
+		    int		src_y,
+		    int		dst_x,
+		    int		dst_y,
+		    int		width,
+		    int		height);
+
+void fbCompositeCopyAreavmx (pixman_operator_t	op,
+			     pixman_image_t *	pSrc,
+			     pixman_image_t *	pMask,
+			     pixman_image_t *	pDst,
+			     INT16	xSrc,
+			     INT16      ySrc,
+			     INT16      xMask,
+			     INT16      yMask,
+			     INT16      xDst,
+			     INT16      yDst,
+			     CARD16     width,
+			     CARD16     height);
+
+pixman_bool_t fbSolidFillvmx (FbPixels	*pDraw,
+		     int		x,
+		     int		y,
+		     int		width,
+		     int		height,
+		     FbBits		xor);
+#endif
+#endif /* USE_VMX */
diff --git a/lib/pixman/pixman/pixman.h b/lib/pixman/pixman/pixman.h
index 6a27a3fbc..36d91a900 100644
--- a/lib/pixman/pixman/pixman.h
+++ b/lib/pixman/pixman/pixman.h
@@ -74,10 +74,7 @@ SOFTWARE.
 /*
  * Standard integers
  */
-#if defined (__SVR4) && defined (__sun)
-#  include <sys/int_types.h>
-#  include <stdint.h>
-#elif defined (__OpenBSD__)
+#if defined (_SVR4) || defined (SVR4) || defined (__OpenBSD__) || defined (_sgi) || defined (__sun) || defined (sun)
 #  include <inttypes.h>
 #elif defined (_MSC_VER)
 typedef __int8 int8_t;
@@ -88,17 +85,12 @@ typedef __int32 int32_t;
 typedef unsigned __int32 uint32_t;
 typedef __int64 int64_t;
 typedef unsigned __int64 uint64_t;
+#elif defined (_AIX)
+#  include <sys/inttypes.h>
 #else
 #  include <stdint.h>
 #endif
 
-/* GCC visibility */
-#if defined(__GNUC__) && __GNUC__ >= 4
-#define PIXMAN_EXPORT __attribute__ ((visibility("default")))
-#else
-#define PIXMAN_EXPORT
-#endif
-
 /*
  * Boolean
  */
@@ -167,10 +159,6 @@ struct pixman_transform
     pixman_fixed_t	matrix[3][3];
 };
 
-PIXMAN_EXPORT
-pixman_bool_t pixman_transform_point_3d (pixman_transform_t *transform,
-					 pixman_vector_t    *vector);
-
 /* Don't blame me, blame XRender */
 typedef enum
 {
@@ -274,112 +262,186 @@ typedef enum
     PIXMAN_REGION_PART
 } pixman_region_overlap_t;
 
-PIXMAN_EXPORT
-int			pixman_version (void);
-
-PIXMAN_EXPORT
-const char*		pixman_version_string (void);
-
 /* This function exists only to make it possible to preserve the X ABI - it should
  * go away at first opportunity.
  */
-PIXMAN_EXPORT
-void		        pixman_region_set_static_pointers (pixman_box16_t *empty_box,
+void                    pixman_region_set_static_pointers (pixman_box16_t         *empty_box,
 							   pixman_region16_data_t *empty_data,
 							   pixman_region16_data_t *broken_data);
 
+
 /* creation/destruction */
-PIXMAN_EXPORT
-void                    pixman_region_init              (pixman_region16_t *region);
-PIXMAN_EXPORT
-void                    pixman_region_init_rect         (pixman_region16_t *region,
-							 int                x,
-							 int                y,
-							 unsigned int       width,
-							 unsigned int       height);
-PIXMAN_EXPORT
-void                    pixman_region_init_with_extents (pixman_region16_t *region,
-							 pixman_box16_t    *extents);
-PIXMAN_EXPORT
-void                    pixman_region_fini              (pixman_region16_t *region);
+void                    pixman_region_init                (pixman_region16_t      *region);
+void                    pixman_region_init_rect           (pixman_region16_t      *region,
+							   int                     x,
+							   int                     y,
+							   unsigned int            width,
+							   unsigned int            height);
+pixman_bool_t           pixman_region_init_rects          (pixman_region16_t      *region,
+							   pixman_box16_t         *boxes,
+							   int                     count);
+void                    pixman_region_init_with_extents   (pixman_region16_t      *region,
+							   pixman_box16_t         *extents);
+void                    pixman_region_fini                (pixman_region16_t      *region);
+
 
 /* manipulation */
-PIXMAN_EXPORT
-void                    pixman_region_translate  (pixman_region16_t *region,
-						  int                x,
-						  int                y);
-PIXMAN_EXPORT
-pixman_bool_t           pixman_region_copy       (pixman_region16_t *dest,
-						  pixman_region16_t *source);
-PIXMAN_EXPORT
-pixman_bool_t           pixman_region_intersect  (pixman_region16_t *newReg,
-						  pixman_region16_t *reg1,
-						  pixman_region16_t *reg2);
-PIXMAN_EXPORT
-pixman_bool_t           pixman_region_union      (pixman_region16_t *newReg,
-						  pixman_region16_t *reg1,
-						  pixman_region16_t *reg2);
-PIXMAN_EXPORT
-pixman_bool_t           pixman_region_union_rect (pixman_region16_t *dest,
-						  pixman_region16_t *source,
-						  int                x,
-						  int                y,
-						  unsigned int       width,
-						  unsigned int       height);
-PIXMAN_EXPORT
-pixman_bool_t           pixman_region_subtract   (pixman_region16_t *regD,
-						  pixman_region16_t *regM,
-						  pixman_region16_t *regS);
-PIXMAN_EXPORT
-pixman_bool_t           pixman_region_inverse    (pixman_region16_t *newReg,
-						  pixman_region16_t *reg1,
-						  pixman_box16_t    *invRect);
-PIXMAN_EXPORT
-pixman_bool_t           pixman_region_contains_point (pixman_region16_t *region,
-						      int x, int y, pixman_box16_t *box);
-PIXMAN_EXPORT
-pixman_region_overlap_t pixman_region_contains_rectangle (pixman_region16_t *pixman_region16_t,
-							  pixman_box16_t *prect);
-PIXMAN_EXPORT
-pixman_bool_t           pixman_region_not_empty (pixman_region16_t *region);
-PIXMAN_EXPORT
-pixman_box16_t *        pixman_region_extents (pixman_region16_t *region);
-PIXMAN_EXPORT
-int                     pixman_region_n_rects (pixman_region16_t *region);
-PIXMAN_EXPORT
-pixman_box16_t *        pixman_region_rectangles (pixman_region16_t *region,
-						  int		    *n_rects);
-PIXMAN_EXPORT
-pixman_bool_t		pixman_region_equal (pixman_region16_t *region1,
-					     pixman_region16_t *region2);
-PIXMAN_EXPORT
-pixman_bool_t		pixman_region_selfcheck (pixman_region16_t *region);
-PIXMAN_EXPORT
-void			pixman_region_reset (pixman_region16_t *region, pixman_box16_t *box);
-PIXMAN_EXPORT
-pixman_bool_t		pixman_region_init_rects (pixman_region16_t *region,
-						  pixman_box16_t *boxes, int count);
-
-/* Copy / Fill */
-PIXMAN_EXPORT
-pixman_bool_t pixman_blt (uint32_t *src_bits,
-			  uint32_t *dst_bits,
-			  int src_stride,
-			  int dst_stride,
-			  int src_bpp,
-			  int dst_bpp,
-			  int src_x, int src_y,
-			  int dst_x, int dst_y,
-			  int width, int height);
-PIXMAN_EXPORT
-pixman_bool_t pixman_fill (uint32_t *bits,
-			   int stride,
-			   int bpp,
-			   int x,
-			   int y,
-			   int width,
-			   int height,
-			   uint32_t _xor);
+void                    pixman_region_translate           (pixman_region16_t      *region,
+							   int                     x,
+							   int                     y);
+pixman_bool_t           pixman_region_copy                (pixman_region16_t      *dest,
+							   pixman_region16_t      *source);
+pixman_bool_t           pixman_region_intersect           (pixman_region16_t      *newReg,
+							   pixman_region16_t      *reg1,
+							   pixman_region16_t      *reg2);
+pixman_bool_t           pixman_region_union               (pixman_region16_t      *newReg,
+							   pixman_region16_t      *reg1,
+							   pixman_region16_t      *reg2);
+pixman_bool_t           pixman_region_union_rect          (pixman_region16_t      *dest,
+							   pixman_region16_t      *source,
+							   int                     x,
+							   int                     y,
+							   unsigned int            width,
+							   unsigned int            height);
+pixman_bool_t           pixman_region_subtract            (pixman_region16_t      *regD,
+							   pixman_region16_t      *regM,
+							   pixman_region16_t      *regS);
+pixman_bool_t           pixman_region_inverse             (pixman_region16_t      *newReg,
+							   pixman_region16_t      *reg1,
+							   pixman_box16_t         *invRect);
+pixman_bool_t           pixman_region_contains_point      (pixman_region16_t      *region,
+							   int                     x,
+							   int                     y,
+							   pixman_box16_t         *box);
+pixman_region_overlap_t pixman_region_contains_rectangle  (pixman_region16_t      *pixman_region16_t,
+							   pixman_box16_t         *prect);
+pixman_bool_t           pixman_region_not_empty           (pixman_region16_t      *region);
+pixman_box16_t *        pixman_region_extents             (pixman_region16_t      *region);
+int                     pixman_region_n_rects             (pixman_region16_t      *region);
+pixman_box16_t *        pixman_region_rectangles          (pixman_region16_t      *region,
+							   int                    *n_rects);
+pixman_bool_t           pixman_region_equal               (pixman_region16_t      *region1,
+							   pixman_region16_t      *region2);
+pixman_bool_t           pixman_region_selfcheck           (pixman_region16_t      *region);
+void                    pixman_region_reset               (pixman_region16_t      *region,
+							   pixman_box16_t         *box);
+
+/*
+ * 32 bit regions
+ */
+typedef struct pixman_region32_data	pixman_region32_data_t;
+typedef struct pixman_box32		pixman_box32_t;
+typedef struct pixman_rectangle32	pixman_rectangle32_t;
+typedef struct pixman_region32		pixman_region32_t;
+
+struct pixman_region32_data {
+    long		size;
+    long		numRects;
+/*  pixman_box32_t	rects[size];   in memory but not explicitly declared */
+};
+
+struct pixman_rectangle32
+{
+    int32_t x, y;
+    uint32_t width, height;
+};
+
+struct pixman_box32
+{
+    int32_t x1, y1, x2, y2;
+};
+
+struct pixman_region32
+{
+    pixman_box32_t          extents;
+    pixman_region32_data_t  *data;
+};
+
+/* creation/destruction */
+void                    pixman_region32_init               (pixman_region32_t *region);
+void                    pixman_region32_init_rect          (pixman_region32_t *region,
+							    int                x,
+							    int                y,
+							    unsigned int       width,
+							    unsigned int       height);
+pixman_bool_t           pixman_region32_init_rects         (pixman_region32_t *region,
+							    pixman_box32_t    *boxes,
+							    int                count);
+void                    pixman_region32_init_with_extents  (pixman_region32_t *region,
+							    pixman_box32_t    *extents);
+void                    pixman_region32_fini               (pixman_region32_t *region);
+
+
+/* manipulation */
+void                    pixman_region32_translate          (pixman_region32_t *region,
+							    int                x,
+							    int                y);
+pixman_bool_t           pixman_region32_copy               (pixman_region32_t *dest,
+							    pixman_region32_t *source);
+pixman_bool_t           pixman_region32_intersect          (pixman_region32_t *newReg,
+							    pixman_region32_t *reg1,
+							    pixman_region32_t *reg2);
+pixman_bool_t           pixman_region32_union              (pixman_region32_t *newReg,
+							    pixman_region32_t *reg1,
+							    pixman_region32_t *reg2);
+pixman_bool_t           pixman_region32_union_rect         (pixman_region32_t *dest,
+							    pixman_region32_t *source,
+							    int                x,
+							    int                y,
+							    unsigned int       width,
+							    unsigned int       height);
+pixman_bool_t           pixman_region32_subtract           (pixman_region32_t *regD,
+							    pixman_region32_t *regM,
+							    pixman_region32_t *regS);
+pixman_bool_t           pixman_region32_inverse            (pixman_region32_t *newReg,
+							    pixman_region32_t *reg1,
+							    pixman_box32_t    *invRect);
+pixman_bool_t           pixman_region32_contains_point     (pixman_region32_t *region,
+							    int                x,
+							    int                y,
+							    pixman_box32_t    *box);
+pixman_region_overlap_t pixman_region32_contains_rectangle (pixman_region32_t *region,
+							    pixman_box32_t    *prect);
+pixman_bool_t           pixman_region32_not_empty          (pixman_region32_t *region);
+pixman_box32_t *        pixman_region32_extents            (pixman_region32_t *region);
+int                     pixman_region32_n_rects            (pixman_region32_t *region);
+pixman_box32_t *        pixman_region32_rectangles         (pixman_region32_t *region,
+							    int               *n_rects);
+pixman_bool_t           pixman_region32_equal              (pixman_region32_t *region1,
+							    pixman_region32_t *region2);
+pixman_bool_t           pixman_region32_selfcheck          (pixman_region32_t *region);
+void                    pixman_region32_reset              (pixman_region32_t *region,
+							    pixman_box32_t    *box);
+
+
+/* Copy / Fill / Misc */
+pixman_bool_t pixman_blt                (uint32_t           *src_bits,
+					 uint32_t           *dst_bits,
+					 int                 src_stride,
+					 int                 dst_stride,
+					 int                 src_bpp,
+					 int                 dst_bpp,
+					 int                 src_x,
+					 int                 src_y,
+					 int                 dst_x,
+					 int                 dst_y,
+					 int                 width,
+					 int                 height);
+pixman_bool_t pixman_fill               (uint32_t           *bits,
+					 int                 stride,
+					 int                 bpp,
+					 int                 x,
+					 int                 y,
+					 int                 width,
+					 int                 height,
+					 uint32_t            _xor);
+
+pixman_bool_t pixman_transform_point_3d (pixman_transform_t *transform,
+					 pixman_vector_t    *vector);
+
+int           pixman_version            (void);
+const char*   pixman_version_string     (void);
+
 /*
  * Images
  */
@@ -450,7 +512,9 @@ typedef enum {
     PIXMAN_x8r8g8b8 =	PIXMAN_FORMAT(32,PIXMAN_TYPE_ARGB,0,8,8,8),
     PIXMAN_a8b8g8r8 =	PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,8,8,8,8),
     PIXMAN_x8b8g8r8 =	PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,0,8,8,8),
-    
+    PIXMAN_x2b10g10r10 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,0,10,10,10),
+    PIXMAN_a2b10g10r10 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,2,10,10,10),
+
 /* 24bpp formats */
     PIXMAN_r8g8b8 =	PIXMAN_FORMAT(24,PIXMAN_TYPE_ARGB,0,8,8,8),
     PIXMAN_b8g8r8 =	PIXMAN_FORMAT(24,PIXMAN_TYPE_ABGR,0,8,8,8),
@@ -500,37 +564,29 @@ typedef enum {
 
 /* YUV formats */
     PIXMAN_yuy2 =	PIXMAN_FORMAT(16,PIXMAN_TYPE_YUY2,0,0,0,0),
-    PIXMAN_yv12 =	PIXMAN_FORMAT(12,PIXMAN_TYPE_YV12,0,0,0,0),
+    PIXMAN_yv12 =	PIXMAN_FORMAT(12,PIXMAN_TYPE_YV12,0,0,0,0)
 } pixman_format_code_t;
 
 /* Querying supported format values. */
-PIXMAN_EXPORT
-pixman_bool_t	pixman_format_supported_destination (pixman_format_code_t format);
-
-PIXMAN_EXPORT
-pixman_bool_t	pixman_format_supported_source (pixman_format_code_t format);
+pixman_bool_t pixman_format_supported_destination (pixman_format_code_t format);
+pixman_bool_t pixman_format_supported_source      (pixman_format_code_t format);
 
 /* Constructors */
-PIXMAN_EXPORT
 pixman_image_t *pixman_image_create_solid_fill       (pixman_color_t               *color);
-PIXMAN_EXPORT
 pixman_image_t *pixman_image_create_linear_gradient  (pixman_point_fixed_t         *p1,
 						      pixman_point_fixed_t         *p2,
 						      const pixman_gradient_stop_t *stops,
 						      int                           n_stops);
-PIXMAN_EXPORT
 pixman_image_t *pixman_image_create_radial_gradient  (pixman_point_fixed_t         *inner,
 						      pixman_point_fixed_t         *outer,
 						      pixman_fixed_t                inner_radius,
 						      pixman_fixed_t                outer_radius,
 						      const pixman_gradient_stop_t *stops,
 						      int                           n_stops);
-PIXMAN_EXPORT
 pixman_image_t *pixman_image_create_conical_gradient (pixman_point_fixed_t         *center,
 						      pixman_fixed_t                angle,
 						      const pixman_gradient_stop_t *stops,
 						      int                           n_stops);
-PIXMAN_EXPORT
 pixman_image_t *pixman_image_create_bits             (pixman_format_code_t          format,
 						      int                           width,
 						      int                           height,
@@ -538,96 +594,74 @@ pixman_image_t *pixman_image_create_bits             (pixman_format_code_t
 						      int                           rowstride_bytes);
 
 /* Destructor */
-PIXMAN_EXPORT
 pixman_image_t *pixman_image_ref                     (pixman_image_t               *image);
-PIXMAN_EXPORT
 pixman_bool_t   pixman_image_unref                   (pixman_image_t               *image);
 
 
 /* Set properties */
-PIXMAN_EXPORT
 pixman_bool_t   pixman_image_set_clip_region         (pixman_image_t               *image,
 						      pixman_region16_t            *region);
-PIXMAN_EXPORT
+pixman_bool_t   pixman_image_set_clip_region32       (pixman_image_t               *image,
+						      pixman_region32_t            *region);
 void		pixman_image_set_has_client_clip     (pixman_image_t               *image,
 						      pixman_bool_t		    clien_clip);
-PIXMAN_EXPORT
 pixman_bool_t   pixman_image_set_transform           (pixman_image_t               *image,
 						      const pixman_transform_t     *transform);
-PIXMAN_EXPORT
 void            pixman_image_set_repeat              (pixman_image_t               *image,
 						      pixman_repeat_t               repeat);
-PIXMAN_EXPORT
 pixman_bool_t   pixman_image_set_filter              (pixman_image_t               *image,
 						      pixman_filter_t               filter,
 						      const pixman_fixed_t         *filter_params,
 						      int                           n_filter_params);
-PIXMAN_EXPORT
-void            pixman_image_set_filter_params       (pixman_image_t               *image,
-						      pixman_fixed_t               *params,
-						      int                           n_params);
-PIXMAN_EXPORT
 void		pixman_image_set_source_clipping     (pixman_image_t		   *image,
 						      pixman_bool_t                 source_clipping);
-PIXMAN_EXPORT
 void            pixman_image_set_alpha_map           (pixman_image_t               *image,
 						      pixman_image_t               *alpha_map,
 						      int16_t                       x,
 						      int16_t                       y);
-PIXMAN_EXPORT
 void            pixman_image_set_component_alpha     (pixman_image_t               *image,
 						      pixman_bool_t                 component_alpha);
-PIXMAN_EXPORT
 void		pixman_image_set_accessors	     (pixman_image_t		   *image,
 						      pixman_read_memory_func_t	    read_func,
 						      pixman_write_memory_func_t    write_func);
-PIXMAN_EXPORT
 void		pixman_image_set_indexed	     (pixman_image_t		   *image,
 						      const pixman_indexed_t	   *indexed);
-PIXMAN_EXPORT
 uint32_t       *pixman_image_get_data                (pixman_image_t               *image);
-PIXMAN_EXPORT
 int		pixman_image_get_width               (pixman_image_t               *image);
-PIXMAN_EXPORT
 int             pixman_image_get_height              (pixman_image_t               *image);
-PIXMAN_EXPORT
 int		pixman_image_get_stride              (pixman_image_t               *image);
-PIXMAN_EXPORT
 int		pixman_image_get_depth               (pixman_image_t		   *image);
-PIXMAN_EXPORT
 pixman_bool_t	pixman_image_fill_rectangles	     (pixman_op_t		    op,
 						      pixman_image_t		   *image,
 						      pixman_color_t		   *color,
 						      int			    n_rects,
-						      const pixman_rectangle16_t	   *rects);
+						      const pixman_rectangle16_t   *rects);
 
 /* Composite */
-PIXMAN_EXPORT
-pixman_bool_t   pixman_compute_composite_region (pixman_region16_t *	pRegion,
-						 pixman_image_t *	pSrc,
-						 pixman_image_t *	pMask,
-						 pixman_image_t *	pDst,
-						 int16_t		xSrc,
-						 int16_t		ySrc,
-						 int16_t		xMask,
-						 int16_t		yMask,
-						 int16_t		xDst,
-						 int16_t		yDst,
-						 uint16_t		width,
-						 uint16_t		height);
-PIXMAN_EXPORT
-void		pixman_image_composite          (pixman_op_t		    op,
-						 pixman_image_t		   *src,
-						 pixman_image_t               *mask,
-						 pixman_image_t               *dest,
-						 int16_t                       src_x,
-						 int16_t                       src_y,
-						 int16_t                       mask_x,
-						 int16_t                       mask_y,
-						 int16_t                       dest_x,
-						 int16_t                       dest_y,
-						 uint16_t                      width,
-						 uint16_t                      height);
+pixman_bool_t pixman_compute_composite_region (pixman_region16_t *pRegion,
+					       pixman_image_t    *pSrc,
+					       pixman_image_t    *pMask,
+					       pixman_image_t    *pDst,
+					       int16_t            xSrc,
+					       int16_t            ySrc,
+					       int16_t            xMask,
+					       int16_t            yMask,
+					       int16_t            xDst,
+					       int16_t            yDst,
+					       uint16_t           width,
+					       uint16_t           height);
+void          pixman_image_composite          (pixman_op_t        op,
+					       pixman_image_t    *src,
+					       pixman_image_t    *mask,
+					       pixman_image_t    *dest,
+					       int16_t            src_x,
+					       int16_t            src_y,
+					       int16_t            mask_x,
+					       int16_t            mask_y,
+					       int16_t            dest_x,
+					       int16_t            dest_y,
+					       uint16_t           width,
+					       uint16_t           height);
 
 /*
  * Trapezoids
@@ -680,52 +714,44 @@ struct pixman_trap
     pixman_span_fix_t	top, bot;
 };
 
-PIXMAN_EXPORT
-pixman_fixed_t pixman_sample_ceil_y        (pixman_fixed_t       y,
-					    int                  bpp);
-PIXMAN_EXPORT
-pixman_fixed_t pixman_sample_floor_y       (pixman_fixed_t       y,
-					    int                  bpp);
-PIXMAN_EXPORT
-void           pixman_edge_step            (pixman_edge_t       *e,
-					    int                  n);
-PIXMAN_EXPORT
-void           pixman_edge_init            (pixman_edge_t       *e,
-					    int                  bpp,
-					    pixman_fixed_t       y_start,
-					    pixman_fixed_t       x_top,
-					    pixman_fixed_t       y_top,
-					    pixman_fixed_t       x_bot,
-					    pixman_fixed_t       y_bot);
-PIXMAN_EXPORT
-void           pixman_line_fixed_edge_init (pixman_edge_t       *e,
-					    int                  bpp,
-					    pixman_fixed_t       y,
+pixman_fixed_t pixman_sample_ceil_y        (pixman_fixed_t             y,
+					    int                        bpp);
+pixman_fixed_t pixman_sample_floor_y       (pixman_fixed_t             y,
+					    int                        bpp);
+void           pixman_edge_step            (pixman_edge_t             *e,
+					    int                        n);
+void           pixman_edge_init            (pixman_edge_t             *e,
+					    int                        bpp,
+					    pixman_fixed_t             y_start,
+					    pixman_fixed_t             x_top,
+					    pixman_fixed_t             y_top,
+					    pixman_fixed_t             x_bot,
+					    pixman_fixed_t             y_bot);
+void           pixman_line_fixed_edge_init (pixman_edge_t             *e,
+					    int                        bpp,
+					    pixman_fixed_t             y,
 					    const pixman_line_fixed_t *line,
-					    int                  x_off,
-					    int                  y_off);
-PIXMAN_EXPORT
-void           pixman_rasterize_edges      (pixman_image_t      *image,
-					    pixman_edge_t       *l,
-					    pixman_edge_t       *r,
-					    pixman_fixed_t       t,
-					    pixman_fixed_t       b);
-PIXMAN_EXPORT
-void           pixman_add_traps            (pixman_image_t      *image,
-					    int16_t              x_off,
-					    int16_t              y_off,
-					    int                  ntrap,
-					    pixman_trap_t       *traps);
-PIXMAN_EXPORT
-void	       pixman_add_trapezoids       (pixman_image_t      *image,
-					    int16_t              x_off,
-					    int                  y_off,
-					    int                  ntraps,
+					    int                        x_off,
+					    int                        y_off);
+void           pixman_rasterize_edges      (pixman_image_t            *image,
+					    pixman_edge_t             *l,
+					    pixman_edge_t             *r,
+					    pixman_fixed_t             t,
+					    pixman_fixed_t             b);
+void           pixman_add_traps            (pixman_image_t            *image,
+					    int16_t                    x_off,
+					    int16_t                    y_off,
+					    int                        ntrap,
+					    pixman_trap_t             *traps);
+void           pixman_add_trapezoids       (pixman_image_t            *image,
+					    int16_t                    x_off,
+					    int                        y_off,
+					    int                        ntraps,
 					    const pixman_trapezoid_t  *traps);
-PIXMAN_EXPORT
-void           pixman_rasterize_trapezoid  (pixman_image_t      *image,
+void           pixman_rasterize_trapezoid  (pixman_image_t            *image,
 					    const pixman_trapezoid_t  *trap,
-					    int                  x_off,
-					    int                  y_off);
+					    int                        x_off,
+					    int                        y_off);
+
 
 #endif /* PIXMAN_H__ */
diff --git a/lib/pixman/test/Makefile.am b/lib/pixman/test/Makefile.am
index 3db77b341..33e2200db 100644
--- a/lib/pixman/test/Makefile.am
+++ b/lib/pixman/test/Makefile.am
@@ -1,15 +1,19 @@
 if HAVE_GTK
-TESTPROGRAMS = \
+TESTPROGRAMS =			\
+	clip-test		\
 	composite-test		\
 	gradient-test		\
+	region-test		\
 	fetch-test
 
 noinst_PROGRAMS = $(TESTPROGRAMS)
 
-INCLUDES = -I$(top_srcdir)/pixman $(GTK_CFLAGS)
+INCLUDES = -I$(top_srcdir)/pixman -I$(top_builddir)/pixman $(GTK_CFLAGS)
 
 composite_test_LDADD =	$(top_builddir)/pixman/libpixman-1.la $(GTK_LIBS)
 gradient_test_LDADD = $(top_builddir)/pixman/libpixman-1.la $(GTK_LIBS)
 fetch_test_LDADD = $(top_builddir)/pixman/libpixman-1.la
+region_test_LDADD = $(top_builddir)/pixman/libpixman-1.la
+clip_test_LDADD = $(top_builddir)/pixman/libpixman-1.la $(GTK_LIBS)
 
 endif
diff --git a/lib/pixman/test/Makefile.in b/lib/pixman/test/Makefile.in
index c49b472b8..c58c3688a 100644
--- a/lib/pixman/test/Makefile.in
+++ b/lib/pixman/test/Makefile.in
@@ -46,12 +46,18 @@ am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
 mkinstalldirs = $(SHELL) $(install_sh) -d
 CONFIG_HEADER = $(top_builddir)/config.h
 CONFIG_CLEAN_FILES =
-@HAVE_GTK_TRUE@am__EXEEXT_1 = composite-test$(EXEEXT) \
-@HAVE_GTK_TRUE@	gradient-test$(EXEEXT) fetch-test$(EXEEXT)
+@HAVE_GTK_TRUE@am__EXEEXT_1 = clip-test$(EXEEXT) \
+@HAVE_GTK_TRUE@	composite-test$(EXEEXT) gradient-test$(EXEEXT) \
+@HAVE_GTK_TRUE@	region-test$(EXEEXT) fetch-test$(EXEEXT)
 PROGRAMS = $(noinst_PROGRAMS)
+clip_test_SOURCES = clip-test.c
+clip_test_OBJECTS = clip-test.$(OBJEXT)
+am__DEPENDENCIES_1 =
+@HAVE_GTK_TRUE@clip_test_DEPENDENCIES =  \
+@HAVE_GTK_TRUE@	$(top_builddir)/pixman/libpixman-1.la \
+@HAVE_GTK_TRUE@	$(am__DEPENDENCIES_1)
 composite_test_SOURCES = composite-test.c
 composite_test_OBJECTS = composite-test.$(OBJEXT)
-am__DEPENDENCIES_1 =
 @HAVE_GTK_TRUE@composite_test_DEPENDENCIES =  \
 @HAVE_GTK_TRUE@	$(top_builddir)/pixman/libpixman-1.la \
 @HAVE_GTK_TRUE@	$(am__DEPENDENCIES_1)
@@ -64,6 +70,10 @@ gradient_test_OBJECTS = gradient-test.$(OBJEXT)
 @HAVE_GTK_TRUE@gradient_test_DEPENDENCIES =  \
 @HAVE_GTK_TRUE@	$(top_builddir)/pixman/libpixman-1.la \
 @HAVE_GTK_TRUE@	$(am__DEPENDENCIES_1)
+region_test_SOURCES = region-test.c
+region_test_OBJECTS = region-test.$(OBJEXT)
+@HAVE_GTK_TRUE@region_test_DEPENDENCIES =  \
+@HAVE_GTK_TRUE@	$(top_builddir)/pixman/libpixman-1.la
 DEFAULT_INCLUDES = -I. -I$(srcdir) -I$(top_builddir)
 depcomp = $(SHELL) $(top_srcdir)/depcomp
 am__depfiles_maybe = depfiles
@@ -75,8 +85,10 @@ LTCOMPILE = $(LIBTOOL) --tag=CC --mode=compile $(CC) $(DEFS) \
 CCLD = $(CC)
 LINK = $(LIBTOOL) --tag=CC --mode=link $(CCLD) $(AM_CFLAGS) $(CFLAGS) \
 	$(AM_LDFLAGS) $(LDFLAGS) -o $@
-SOURCES = composite-test.c fetch-test.c gradient-test.c
-DIST_SOURCES = composite-test.c fetch-test.c gradient-test.c
+SOURCES = clip-test.c composite-test.c fetch-test.c gradient-test.c \
+	region-test.c
+DIST_SOURCES = clip-test.c composite-test.c fetch-test.c \
+	gradient-test.c region-test.c
 ETAGS = etags
 CTAGS = ctags
 DISTFILES = $(DIST_COMMON) $(DIST_SOURCES) $(TEXINFOS) $(EXTRA_DIST)
@@ -103,6 +115,7 @@ DEFS = @DEFS@
 DEPDIR = @DEPDIR@
 DEP_CFLAGS = @DEP_CFLAGS@
 DEP_LIBS = @DEP_LIBS@
+DSYMUTIL = @DSYMUTIL@
 ECHO = @ECHO@
 ECHO_C = @ECHO_C@
 ECHO_N = @ECHO_N@
@@ -128,6 +141,7 @@ LTLIBOBJS = @LTLIBOBJS@
 LT_VERSION_INFO = @LT_VERSION_INFO@
 MAKEINFO = @MAKEINFO@
 MMX_CFLAGS = @MMX_CFLAGS@
+NMEDIT = @NMEDIT@
 OBJEXT = @OBJEXT@
 PACKAGE = @PACKAGE@
 PACKAGE_BUGREPORT = @PACKAGE_BUGREPORT@
@@ -136,26 +150,31 @@ PACKAGE_STRING = @PACKAGE_STRING@
 PACKAGE_TARNAME = @PACKAGE_TARNAME@
 PACKAGE_VERSION = @PACKAGE_VERSION@
 PATH_SEPARATOR = @PATH_SEPARATOR@
+PERL = @PERL@
 PIXMAN_VERSION_MAJOR = @PIXMAN_VERSION_MAJOR@
 PIXMAN_VERSION_MICRO = @PIXMAN_VERSION_MICRO@
 PIXMAN_VERSION_MINOR = @PIXMAN_VERSION_MINOR@
 PKG_CONFIG = @PKG_CONFIG@
 RANLIB = @RANLIB@
+SED = @SED@
 SET_MAKE = @SET_MAKE@
 SHELL = @SHELL@
-SSE_CFLAGS = @SSE_CFLAGS@
+SSE2_CFLAGS = @SSE2_CFLAGS@
 STRIP = @STRIP@
 USE_MMX_FALSE = @USE_MMX_FALSE@
 USE_MMX_TRUE = @USE_MMX_TRUE@
 USE_SSE2_FALSE = @USE_SSE2_FALSE@
 USE_SSE2_TRUE = @USE_SSE2_TRUE@
-USE_SSE_FALSE = @USE_SSE_FALSE@
-USE_SSE_TRUE = @USE_SSE_TRUE@
+USE_VMX_FALSE = @USE_VMX_FALSE@
+USE_VMX_TRUE = @USE_VMX_TRUE@
 VERSION = @VERSION@
+VMX_CFLAGS = @VMX_CFLAGS@
 ac_ct_AR = @ac_ct_AR@
 ac_ct_CC = @ac_ct_CC@
 ac_ct_CXX = @ac_ct_CXX@
+ac_ct_DSYMUTIL = @ac_ct_DSYMUTIL@
 ac_ct_F77 = @ac_ct_F77@
+ac_ct_NMEDIT = @ac_ct_NMEDIT@
 ac_ct_RANLIB = @ac_ct_RANLIB@
 ac_ct_STRIP = @ac_ct_STRIP@
 ac_pt_PKG_CONFIG = @ac_pt_PKG_CONFIG@
@@ -197,14 +216,18 @@ sharedstatedir = @sharedstatedir@
 sysconfdir = @sysconfdir@
 target_alias = @target_alias@
 @HAVE_GTK_TRUE@TESTPROGRAMS = \
+@HAVE_GTK_TRUE@	clip-test		\
 @HAVE_GTK_TRUE@	composite-test		\
 @HAVE_GTK_TRUE@	gradient-test		\
+@HAVE_GTK_TRUE@	region-test		\
 @HAVE_GTK_TRUE@	fetch-test
 
-@HAVE_GTK_TRUE@INCLUDES = -I$(top_srcdir)/pixman $(GTK_CFLAGS)
+@HAVE_GTK_TRUE@INCLUDES = -I$(top_srcdir)/pixman -I$(top_builddir)/pixman $(GTK_CFLAGS)
 @HAVE_GTK_TRUE@composite_test_LDADD = $(top_builddir)/pixman/libpixman-1.la $(GTK_LIBS)
 @HAVE_GTK_TRUE@gradient_test_LDADD = $(top_builddir)/pixman/libpixman-1.la $(GTK_LIBS)
 @HAVE_GTK_TRUE@fetch_test_LDADD = $(top_builddir)/pixman/libpixman-1.la
+@HAVE_GTK_TRUE@region_test_LDADD = $(top_builddir)/pixman/libpixman-1.la
+@HAVE_GTK_TRUE@clip_test_LDADD = $(top_builddir)/pixman/libpixman-1.la $(GTK_LIBS)
 all: all-am
 
 .SUFFIXES:
@@ -245,6 +268,9 @@ clean-noinstPROGRAMS:
 	  echo " rm -f $$p $$f"; \
 	  rm -f $$p $$f ; \
 	done
+clip-test$(EXEEXT): $(clip_test_OBJECTS) $(clip_test_DEPENDENCIES) 
+	@rm -f clip-test$(EXEEXT)
+	$(LINK) $(clip_test_LDFLAGS) $(clip_test_OBJECTS) $(clip_test_LDADD) $(LIBS)
 composite-test$(EXEEXT): $(composite_test_OBJECTS) $(composite_test_DEPENDENCIES) 
 	@rm -f composite-test$(EXEEXT)
 	$(LINK) $(composite_test_LDFLAGS) $(composite_test_OBJECTS) $(composite_test_LDADD) $(LIBS)
@@ -254,6 +280,9 @@ fetch-test$(EXEEXT): $(fetch_test_OBJECTS) $(fetch_test_DEPENDENCIES)
 gradient-test$(EXEEXT): $(gradient_test_OBJECTS) $(gradient_test_DEPENDENCIES) 
 	@rm -f gradient-test$(EXEEXT)
 	$(LINK) $(gradient_test_LDFLAGS) $(gradient_test_OBJECTS) $(gradient_test_LDADD) $(LIBS)
+region-test$(EXEEXT): $(region_test_OBJECTS) $(region_test_DEPENDENCIES) 
+	@rm -f region-test$(EXEEXT)
+	$(LINK) $(region_test_LDFLAGS) $(region_test_OBJECTS) $(region_test_LDADD) $(LIBS)
 
 mostlyclean-compile:
 	-rm -f *.$(OBJEXT)
@@ -261,9 +290,11 @@ mostlyclean-compile:
 distclean-compile:
 	-rm -f *.tab.c
 
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/clip-test.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/composite-test.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/fetch-test.Po@am__quote@
 @AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/gradient-test.Po@am__quote@
+@AMDEP_TRUE@@am__include@ @am__quote@./$(DEPDIR)/region-test.Po@am__quote@
 
 .c.o:
 @am__fastdepCC_TRUE@	if $(COMPILE) -MT $@ -MD -MP -MF "$(DEPDIR)/$*.Tpo" -c -o $@ $<; \
diff --git a/lib/pixman/test/clip-test.c b/lib/pixman/test/clip-test.c
new file mode 100644
index 000000000..457e97a2b
--- /dev/null
+++ b/lib/pixman/test/clip-test.c
@@ -0,0 +1,159 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <gtk/gtk.h>
+#include "pixman.h"
+
+GdkPixbuf *
+pixbuf_from_argb32 (uint32_t *bits,
+		    int width,
+		    int height,
+		    int stride)
+{
+    GdkPixbuf *pixbuf = gdk_pixbuf_new (GDK_COLORSPACE_RGB, TRUE,
+					8, width, height);
+    int p_stride = gdk_pixbuf_get_rowstride (pixbuf);
+    guint32 *p_bits = (guint32 *)gdk_pixbuf_get_pixels (pixbuf);
+    int w, h;
+    
+    for (h = 0; h < height; ++h)
+    {
+	for (w = 0; w < width; ++w)
+	{
+	    uint32_t argb = bits[h * stride + w];
+	    guint32 abgr;
+	    
+	    abgr = (argb & 0xff000000) |
+		(argb & 0xff) << 16 |
+		(argb & 0x00ff00) |
+		(argb & 0xff0000) >> 16;
+	    
+	    p_bits[h * (p_stride / 4) + w] = abgr;
+	}
+    }
+    
+    return pixbuf;
+}
+
+static gboolean
+on_expose (GtkWidget *widget, GdkEventExpose *expose, gpointer data)
+{
+    GdkPixbuf *pixbuf = data;
+    
+    gdk_draw_pixbuf (widget->window, NULL,
+		     pixbuf, 0, 0, 0, 0,
+		     gdk_pixbuf_get_width (pixbuf),
+		     gdk_pixbuf_get_height (pixbuf),
+		     GDK_RGB_DITHER_NONE,
+		     0, 0);
+    
+    return TRUE;
+}
+
+static void
+show_window (pixman_image_t *img, int w, int h, int stride)
+{
+    GdkPixbuf *pixbuf;
+    
+    GtkWidget *window = gtk_window_new (GTK_WINDOW_TOPLEVEL);
+    
+    pixbuf = pixbuf_from_argb32 (pixman_image_get_data (img), w, h, stride);
+    
+    g_signal_connect (window, "expose_event", G_CALLBACK (on_expose), pixbuf);
+    g_signal_connect (window, "delete_event", G_CALLBACK (gtk_main_quit), NULL);
+    
+    gtk_widget_show (window);
+    
+    gtk_main ();
+}
+
+#define WIDTH 200
+#define HEIGHT 200
+    
+static pixman_image_t *
+create_solid_bits (uint32_t pixel)
+{
+    uint32_t *pixels = malloc (WIDTH * HEIGHT * 4);
+    int i;
+    
+    for (i = 0; i < WIDTH * HEIGHT; ++i)
+	pixels[i] = pixel;
+
+    return pixman_image_create_bits (PIXMAN_a8r8g8b8,
+				     WIDTH, HEIGHT, 
+				     pixels,
+				     WIDTH * 4);
+}
+
+int
+main (int argc, char **argv)
+{
+    pixman_image_t *gradient_img;
+    pixman_image_t *src_img, *dst_img;
+    pixman_gradient_stop_t stops[2] =
+	{
+	    { pixman_int_to_fixed (0), { 0xffff, 0x0000, 0x0000, 0xffff } },
+	    { pixman_int_to_fixed (1), { 0xffff, 0xffff, 0x0000, 0xffff } }
+	};
+    pixman_point_fixed_t p1 = { 0, 0 };
+    pixman_point_fixed_t p2 = { pixman_int_to_fixed (WIDTH),
+				pixman_int_to_fixed (HEIGHT) };
+    pixman_point_fixed_t c_inner;
+    pixman_point_fixed_t c_outer;
+    pixman_fixed_t r_inner;
+    pixman_fixed_t r_outer;
+    pixman_region32_t clip_region;
+    pixman_transform_t trans = {
+	{ { pixman_double_to_fixed (1.3), pixman_double_to_fixed (0), pixman_double_to_fixed (-0.5), },
+	  { pixman_double_to_fixed (0), pixman_double_to_fixed (1), pixman_double_to_fixed (-0.5), },
+	  { pixman_double_to_fixed (0), pixman_double_to_fixed (0), pixman_double_to_fixed (1.0) } 
+	}
+    };
+    
+    gtk_init (&argc, &argv);
+    
+    src_img = create_solid_bits (0xff0000ff);
+    
+    c_inner.x = pixman_double_to_fixed (100.0);
+    c_inner.y = pixman_double_to_fixed (100.0);
+    c_outer.x = pixman_double_to_fixed (100.0);
+    c_outer.y = pixman_double_to_fixed (100.0);
+    r_inner = 0;
+    r_outer = pixman_double_to_fixed (100.0);
+    
+    gradient_img = pixman_image_create_radial_gradient (&c_inner, &c_outer,
+							r_inner, r_outer,
+							stops, 2);
+
+#if 0
+    gradient_img = pixman_image_create_linear_gradient  (&p1, &p2,
+							 stops, 2);
+    
+#endif
+
+    pixman_image_composite (PIXMAN_OP_OVER, gradient_img, NULL, src_img,
+			    0, 0, 0, 0, 0, 0, WIDTH, HEIGHT);
+    
+    pixman_region32_init_rect (&clip_region, 50, 0, 100, 200);
+    pixman_image_set_clip_region32 (src_img, &clip_region);
+    pixman_image_set_source_clipping (src_img, TRUE);
+    pixman_image_set_transform (src_img, &trans);
+    pixman_image_set_repeat (src_img, PIXMAN_REPEAT_NORMAL);
+    
+    dst_img = create_solid_bits (0xffff0000);
+    pixman_image_composite (PIXMAN_OP_OVER, src_img, NULL, dst_img,
+			    0, 0, 0, 0, 0, 0, WIDTH, HEIGHT);
+    
+
+#if 0
+    printf ("0, 0: %x\n", src[0]);
+    printf ("10, 10: %x\n", src[10 * 10 + 10]);
+    printf ("w, h: %x\n", src[(HEIGHT - 1) * 100 + (WIDTH - 1)]);
+#endif
+    
+    show_window (dst_img, WIDTH, HEIGHT, WIDTH);
+    
+    pixman_image_unref (gradient_img);
+    pixman_image_unref (src_img);
+    
+    return 0;
+}
diff --git a/lib/pixman/test/region-test.c b/lib/pixman/test/region-test.c
new file mode 100644
index 000000000..e214e9b89
--- /dev/null
+++ b/lib/pixman/test/region-test.c
@@ -0,0 +1,23 @@
+#include <assert.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include "pixman.h"
+
+/* This used to go into an infinite loop before pixman-region.c
+ * was fixed to not use explict "short" variables
+ */
+int
+main ()
+{
+    pixman_region32_t r1;
+    pixman_region32_t r2;
+    pixman_region32_t r3;
+
+    pixman_region32_init_rect (&r1, 0, 0, 20, 64000);
+    pixman_region32_init_rect (&r2, 0, 0, 20, 64000);
+    pixman_region32_init_rect (&r3, 0, 0, 20, 64000);
+
+    pixman_region32_subtract (&r1, &r2, &r3);
+
+}
+
author	Matthieu Herrb <matthieu@cvs.openbsd.org>	2008-09-23 19:11:41 +0000
committer	Matthieu Herrb <matthieu@cvs.openbsd.org>	2008-09-23 19:11:41 +0000
commit	61c42f8ddcc41b05b5d28906533e3ca0b9aa1965 (patch)
tree	2009f3d26880bf14c1b4e7953ee9b0cac8236df7 /lib/pixman
parent	f0d06ca61c104601dbdd10a4e58ae54f8d3f5332 (diff)