src - OpenBSD base system

Age	Commit message (Collapse)	Author
2013-02-27	pte bits have changed on gen6+, for now always set to uncached	Jonathan Gray

2013-02-27	sync execbuffer domain tracking and flushing with linux 3.4.33	Jonathan Gray

2013-02-26	enable more of i915_gem_object_pin_to_display_plane	Jonathan Gray

2013-02-26	unstub i915_gem_object_set_cache_level	Jonathan Gray

2013-02-26	set a default cache level	Jonathan Gray

2013-02-26	unstub i915_gem_object_flush_cpu_write_domain	Jonathan Gray

2013-02-26	add a i915_gem_clflush_object implementation, not yet used	Jonathan Gray

2013-02-26	add some additional flushes found in linux 3.4.33	Jonathan Gray

2013-02-25	Bring domain handling code close to Linux 3.4.33.	Mark Kettenis

2013-02-25	claim to support BSD/BLT rings	Jonathan Gray

2013-02-25	init some vblank vars	Jonathan Gray

2013-02-25	add some callbacks and a var for upcoming vblank changes	Jonathan Gray

2013-02-25	enable ivybridge parity task	Jonathan Gray

2013-02-25	use the ring flush callbacks in the accel fb functions	Jonathan Gray

2013-02-25	i915_get_fence_size is no longer needed	Jonathan Gray

2013-02-25	sync the chip reset code with linux, we can now reset more than just 965	Jonathan Gray

2013-02-24	Bring inactive list management closer to what Linux does.	Mark Kettenis
	Now we no longer call i915_gem_object_move_to_inactive() on objects that are not active.
2013-02-24	Oops, really remove all places where I915_GPU_WRITE was used.	Mark Kettenis

2013-02-24	Change I915_GPU_WRITE into a bitfield and adjust related code to be similar	Mark Kettenis
	to what Linux does.
2013-02-24	Bring i915_gem_object_move_to_active() a bit more in line with Linux.	Mark Kettenis

2013-02-24	Bring i915_gem_fault() a bit more in line with the Linux code.	Mark Kettenis

2013-02-24	Implement i915_gem_object_is_inactive().	Mark Kettenis

2013-02-24	Remove #ifdef'ed out code. The replacement in i915_gem.c is here to stay.	Mark Kettenis

2013-02-24	Get rid of I915_FENCED_EXEC. It's effectively unused.	Mark Kettenis

2013-02-24	Move i915_gem_process_flushing() to i915_gem.c and rename it to	Mark Kettenis
	i915_gem_process_flushing_list(). Bring it closer to the Linux version.
2013-02-24	Remove mm.gpu_write_list; it isn't used anymore.	Mark Kettenis

2013-02-24	some more M_NOWAIT -> M_WAITOK	Jonathan Gray

2013-02-24	some M_NOWAIT modesetting mallocs should be M_WAITOK	Jonathan Gray

2013-02-23	Hook up panel backlight control to wscons infrastructure.	Mark Kettenis

2013-02-23	Turns out mm.suspended gets reset before the hardware is properly initialized	Mark Kettenis
	in the resume path. As a consequence we may do some debug printfs before the rings are properly set up again. This makes us hang on resume on chips that have the seperate blitter ring.
2013-02-23	Restore framebuffer console mode in inteldrm_show_screen(). Gives us some	Mark Kettenis
	rudimentary vt switch capability if multiple screens are enabled.
2013-02-23	Use the blitter ring for submitting blitter commands on chips that have it.	Mark Kettenis
	Make framebuffer acceleration work on sandybridge. Hopefully this works on newer chips as well.
2013-02-23	Run common vga attachment code after drm attached instead of before.	Mark Kettenis
	Let wsdisplay ignore the console indicator if a wsdisplay console has already been attached.
2013-02-23	Don't use the accelerated framebuffer operations in the suspend resume path.	Mark Kettenis

2013-02-23	Framebuffer acceleration routines. Disabled for now.	Mark Kettenis

2013-02-23	setup hotplug workqs	Jonathan Gray
	with help from kettenis
2013-02-23	Don't clear wsdisplay_console_initted since we just want to override the	Mark Kettenis
	wsdisplay console attachment.
2013-02-23	Don't set cn_tab if we're just overriding the console attachment.	Mark Kettenis

2013-02-23	Eliminate request_lock. Linux doesn't have it and the kernel lock should be	Mark Kettenis
	good enough to protect us.
2013-02-23	unstub intel_render_ring_init_dri, not currently called	Jonathan Gray
	seems it is for the old DRI1 interface we no longer support with inteldrm?
2013-02-23	add i915_kernel_lost_context for ums (replaces inteldrm_update_ring).	Jonathan Gray

2013-02-23	don't call i915_gem_object_get_fence in i915_gem_process_flushing	Jonathan Gray
	seems to have never been there in the original linux code and is now causing problems. Diff from kettenis.
2013-02-22	The libdrm_intel code is totally retarded and doesn't actually look at the	Mark Kettenis
	value returned by I915_PARAM_HAS_RELAXED_FENCING. If the ioctl succeeds, it assumes relaxed fencing is supported. So as long as we don't actually support it, we need to fail with EINVAL.
2013-02-22	Remove some bits of old interrupt code that are no longer useful.	Mark Kettenis

2013-02-22	Unstub cleanup_pipe_control().	Mark Kettenis

2013-02-22	Don't forget to increase the fence pin count.	Mark Kettenis

2013-02-22	add some missing chipset flush calls	Jonathan Gray

2013-02-22	remove list lock removed by linux back in 2010	Jonathan Gray
	commit de227ef0907258359d53e3e1530c1f3678eb2bb9 Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Sat Jul 3 07:58:38 2010 +0100 drm/i915: Kill the active list spinlock This spinlock only served debugging purposes in a time when we could not be sure of the mutex ever being released upon a GPU hang. As we now should be able rely on hangcheck to do the job for us (and that error reporting should not itself require the struct mutex) we can kill the incomplete attempt at protection. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
2013-02-22	init rps.hw_lock	Jonathan Gray

2013-02-22	init rps.lock	Jonathan Gray