mesa.git
14 years agoegl: Rework config lookup.
Chia-I Wu [Fri, 25 Sep 2009 15:43:49 +0000 (23:43 +0800)]
egl: Rework config lookup.

Make it similiar to how contexts and surfaces are looked up.  It should
be slightly faster, and work better with multiple displays.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
14 years agoegl: Rework configuration management.
Chia-I Wu [Fri, 25 Sep 2009 15:24:03 +0000 (23:24 +0800)]
egl: Rework configuration management.

This mainly implements the algorithms for configuration selection and
sorting, described in the spec.  User errors should also be correctly
detected and reported.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
14 years agoegl: Introduce config keys.
Chia-I Wu [Fri, 25 Sep 2009 14:54:34 +0000 (22:54 +0800)]
egl: Introduce config keys.

Config keys are almost config attributes.  A valid config attribute is a
valid config key, but a valid config key may not be a valid config
attribute.

This commit does not distinguish the differences.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
14 years agollvmpipe: Emit SSE intrinsics based on runtime cpu capability check.
José Fonseca [Tue, 29 Sep 2009 12:58:58 +0000 (13:58 +0100)]
llvmpipe: Emit SSE intrinsics based on runtime cpu capability check.

Note that llvmpipe still doesn't run on any processor yet: if you don't
have a recent processor with SSE4.1 you will still likely end up
hitting a code path for which a generic non-sse4 version is not
implemented yet.

14 years agoutil: Cleanup u_cpu_detect, build. Support X86_64 and detect SSE4.1 too.
José Fonseca [Tue, 29 Sep 2009 12:25:08 +0000 (13:25 +0100)]
util: Cleanup u_cpu_detect, build. Support X86_64 and detect SSE4.1 too.

I was waiting for the need to use this code to arise, and it finally came.

I've tested building this on Linux and Windows, both x86 and x64_64. But
it might break other platforms. Please bear with me and help me fix it.

Many thanks to Dennis Smit who submitted this, and Eric Anholt whose
work this was based on.

14 years agogallium: New PIPE_OS_UNIX to simplify code that is portable to all unices.
José Fonseca [Tue, 29 Sep 2009 12:17:36 +0000 (13:17 +0100)]
gallium: New PIPE_OS_UNIX to simplify code that is portable to all unices.

14 years agoi915: Fix MSVC build.
Vinson Lee [Tue, 29 Sep 2009 09:38:47 +0000 (10:38 +0100)]
i915: Fix MSVC build.

14 years agopython: Update for surface_buffer_create change.
José Fonseca [Mon, 28 Sep 2009 18:44:30 +0000 (19:44 +0100)]
python: Update for surface_buffer_create change.

14 years agog3dvl: assert.h -> util/u_debug.h
José Fonseca [Mon, 28 Sep 2009 18:02:34 +0000 (19:02 +0100)]
g3dvl: assert.h -> util/u_debug.h

14 years agog3dvl: Define PIPE_VIDEO_CODEC_UNKNOWN for failures.
José Fonseca [Mon, 28 Sep 2009 18:01:49 +0000 (19:01 +0100)]
g3dvl: Define PIPE_VIDEO_CODEC_UNKNOWN for failures.

gcc 4.4 seems particularly picky with int -> enum conversions.

14 years agor600: use CB_TARGET_MASK instead of CB_SHADER_MASK for setting color mask
Andre Maasikas [Mon, 28 Sep 2009 08:23:49 +0000 (11:23 +0300)]
r600: use CB_TARGET_MASK instead of CB_SHADER_MASK for setting color mask

makes blend functions work better

Signed-off-by: Dave Airlie <airlied@linux.ie>
14 years agor600: user correct alpha blend factor
Andre Maasikas [Mon, 28 Sep 2009 07:42:35 +0000 (10:42 +0300)]
r600: user correct alpha blend factor

Signed-off-by: Dave Airlie <airlied@linux.ie>
14 years agor600: clear position enable bit when when wpos is not used by FP
Andre Maasikas [Tue, 29 Sep 2009 06:46:29 +0000 (09:46 +0300)]
r600: clear position enable bit when when wpos is not used by FP

Makes doom3 alot nicer..

14 years agost/xorg: Add debug for fallbacks
Jakob Bornecrantz [Mon, 28 Sep 2009 18:19:26 +0000 (11:19 -0700)]
st/xorg: Add debug for fallbacks

14 years agost/xorg: Make debug printing optional
Jakob Bornecrantz [Sun, 27 Sep 2009 21:03:24 +0000 (14:03 -0700)]
st/xorg: Make debug printing optional

14 years agometa: Fix invalid PBO access from DrawPixels when trying to just alloc.
Eric Anholt [Mon, 28 Sep 2009 21:03:40 +0000 (14:03 -0700)]
meta: Fix invalid PBO access from DrawPixels when trying to just alloc.

This whole reuse of buffers (TexSubImage instead of TexImage, SubData
instead of Data) is bad for hardware drivers, but it's even worse when
we accidentally try to access the 2x2 PBO to fill the new 16x16 texture
we're creating, producing GL errors.

Fixes piglit pbo-drawpixels.  Bug #14163.

14 years agointel: Drop my generatemipmap code in favor of the new shared code.
Eric Anholt [Mon, 28 Sep 2009 18:42:31 +0000 (11:42 -0700)]
intel: Drop my generatemipmap code in favor of the new shared code.

14 years agointel: Remove some dead metaops code.
Eric Anholt [Tue, 22 Sep 2009 00:13:31 +0000 (17:13 -0700)]
intel: Remove some dead metaops code.

14 years agoMerge branch 'mesa_7_6_branch'
Ian Romanick [Mon, 28 Sep 2009 19:03:23 +0000 (12:03 -0700)]
Merge branch 'mesa_7_6_branch'

Conflicts:
Makefile
src/mesa/main/version.h

14 years agoMerge branch 'mesa_7_5_branch' into mesa_7_6_branch
Ian Romanick [Mon, 28 Sep 2009 19:02:39 +0000 (12:02 -0700)]
Merge branch 'mesa_7_5_branch' into mesa_7_6_branch

Conflicts:
src/mesa/main/version.h

14 years agodocs: 7.5.2 md5 sums
Ian Romanick [Mon, 28 Sep 2009 18:41:23 +0000 (11:41 -0700)]
docs: 7.5.2 md5 sums

14 years agoPrep for 7.5.2 release
Ian Romanick [Mon, 28 Sep 2009 18:33:52 +0000 (11:33 -0700)]
Prep for 7.5.2 release

14 years agodocs: 7.6 md5 sums
Ian Romanick [Mon, 28 Sep 2009 18:29:25 +0000 (11:29 -0700)]
docs: 7.6 md5 sums

14 years agoPrep for 7.6 release
Ian Romanick [Mon, 28 Sep 2009 18:22:54 +0000 (11:22 -0700)]
Prep for 7.6 release

14 years agost/mesa: check gl_texture_object::GenerateMipmap field when allocating texmem
Brian Paul [Mon, 28 Sep 2009 16:03:58 +0000 (10:03 -0600)]
st/mesa: check gl_texture_object::GenerateMipmap field when allocating texmem

In guess_and_alloc_texture() use the gl_texture_object::GenerateMipmap
field as another hint as to whether to allocate space for a whole mipmap.

14 years agoMerge branch 'mesa_7_6_branch'
Brian Paul [Mon, 28 Sep 2009 16:02:20 +0000 (10:02 -0600)]
Merge branch 'mesa_7_6_branch'

14 years agoMerge branch 'mesa_7_5_branch' into mesa_7_6_branch
Brian Paul [Mon, 28 Sep 2009 15:59:59 +0000 (09:59 -0600)]
Merge branch 'mesa_7_5_branch' into mesa_7_6_branch

14 years agoFix build on non GLIBC platforms (FreeBSD at least)
Robert Noland [Tue, 22 Sep 2009 18:49:57 +0000 (11:49 -0700)]
Fix build on non GLIBC platforms (FreeBSD at least)

Build was broken by commit 9666529b5a5be1fcde82caadc2fe2efa5ea81e49

I'm not certain that this is entirely the correct fix since the demo
from bug #23774 seemed to work before the commit that broke the build.

Signed-off-by: Robert Noland <rnoland@2hip.net>
Signed-off-by: Brian Paul <brianp@vmware.com>
14 years agodocs: list additional 7.5.2 bug fixes
Brian Paul [Mon, 28 Sep 2009 15:43:42 +0000 (09:43 -0600)]
docs: list additional 7.5.2 bug fixes

14 years agodocs: document gallium mipmap generation fix
Brian Paul [Mon, 28 Sep 2009 15:38:50 +0000 (09:38 -0600)]
docs: document gallium mipmap generation fix

14 years agost/mesa: fix st_generate_mipmap() issues
Brian Paul [Fri, 25 Sep 2009 23:24:27 +0000 (17:24 -0600)]
st/mesa: fix st_generate_mipmap() issues

The main issue is we didn't always have a gallium texture object with
enough space to store the to-be-generated mipmap levels.  When that's
the case, allocate a new gallium texture and use st_texure_finalize()
to copy images from the old texture to the new one.

We also had the baseLevel parameter to st_render_mipmap() wrong.

14 years agost/mesa: fix/simplify st_texture_object::lastLevel calculation
Brian Paul [Mon, 28 Sep 2009 15:28:50 +0000 (09:28 -0600)]
st/mesa: fix/simplify st_texture_object::lastLevel calculation

Don't compute the st_texture_object::lastLevel field based on the texture
filters.  Use the _MaxLevel value that core Mesa computes for us.
When called from the GenerateMipmap path, we'll use the lastLevel field
as-is.

14 years agogallium/util: add sanity check assertions
Brian Paul [Fri, 25 Sep 2009 23:19:25 +0000 (17:19 -0600)]
gallium/util: add sanity check assertions

14 years agomesa: use _mesa_get_current_tex_unit() helper
Brian Paul [Mon, 28 Sep 2009 15:22:33 +0000 (09:22 -0600)]
mesa: use _mesa_get_current_tex_unit() helper

14 years agoprogs/perf: Include local headers before installed headers.
Vinson Lee [Mon, 28 Sep 2009 15:03:26 +0000 (09:03 -0600)]
progs/perf: Include local headers before installed headers.

Fixes compilation errors on platforms with older installed GL headers.

14 years agoprogs/tests: print usage info upon SPACE/u
Brian Paul [Fri, 25 Sep 2009 22:38:14 +0000 (16:38 -0600)]
progs/tests: print usage info upon SPACE/u

14 years agoprogs/tests: re-indent mipmap_limits.c
Brian Paul [Fri, 25 Sep 2009 22:37:15 +0000 (16:37 -0600)]
progs/tests: re-indent mipmap_limits.c

14 years agollvmpipe: Grab a ref when the fb is set.
Younes Manton [Sun, 27 Sep 2009 14:56:42 +0000 (10:56 -0400)]
llvmpipe: Grab a ref when the fb is set.

Based on softpipe commit a77226071f6814a53358a5d6caff685889d0e4ec.

14 years agosoftpipe: Fix MSVC build.
José Fonseca [Mon, 28 Sep 2009 12:03:03 +0000 (13:03 +0100)]
softpipe: Fix MSVC build.

14 years agog3dvl: Fix MSVC build.
José Fonseca [Mon, 28 Sep 2009 12:02:42 +0000 (13:02 +0100)]
g3dvl: Fix MSVC build.

pipe/p_compiler for integer types. No declarations out of scope.

14 years agor300g: fix r300g cause GPU hang issue.
Cooper Yuan [Mon, 28 Sep 2009 09:55:38 +0000 (17:55 +0800)]
r300g: fix r300g cause GPU hang issue.

why there are two input position semantic tags is that ureg doesn't
set vs input semantic due to commit: 6d8dbd3d1ec888

so use vs input index instead of semantic name.

14 years agoxvmc: Fail on unsupported formats, operations.
Younes Manton [Mon, 28 Sep 2009 04:17:33 +0000 (00:17 -0400)]
xvmc: Fail on unsupported formats, operations.

14 years agog3dvl: Missing semicolon.
Younes Manton [Mon, 28 Sep 2009 03:54:36 +0000 (23:54 -0400)]
g3dvl: Missing semicolon.

14 years agog3dvl: Formatting and cleanups.
Younes Manton [Mon, 28 Sep 2009 03:14:52 +0000 (23:14 -0400)]
g3dvl: Formatting and cleanups.

14 years agog3dvl: Move XvMC under the Xorg state tracker.
Younes Manton [Mon, 28 Sep 2009 01:54:20 +0000 (21:54 -0400)]
g3dvl: Move XvMC under the Xorg state tracker.

14 years agog3dvl: Implement XvMC using pipe_video_context.
Younes Manton [Mon, 28 Sep 2009 00:18:02 +0000 (20:18 -0400)]
g3dvl: Implement XvMC using pipe_video_context.

14 years agog3dvl: pipe_video_context interface, softpipe impl, auxiliary libs
Younes Manton [Sun, 27 Sep 2009 23:49:06 +0000 (19:49 -0400)]
g3dvl: pipe_video_context interface, softpipe impl, auxiliary libs

14 years agoutil: Add util_next_power_of_two() for rounding a uint up to a POT.
Younes Manton [Sun, 27 Sep 2009 18:40:14 +0000 (14:40 -0400)]
util: Add util_next_power_of_two() for rounding a uint up to a POT.

14 years agor300g: fix erroneous condition
Maciej Cencora [Sun, 27 Sep 2009 20:30:45 +0000 (22:30 +0200)]
r300g: fix erroneous condition

14 years agor300g: don't force vertex position for HW TCL path
Maciej Cencora [Sun, 27 Sep 2009 20:28:46 +0000 (22:28 +0200)]
r300g: don't force vertex position for HW TCL path

It could be generated by vertex shader.

14 years agor300g: add some assertions
Maciej Cencora [Sun, 27 Sep 2009 20:20:41 +0000 (22:20 +0200)]
r300g: add some assertions

Not sure why we are getting a shader with two inputs with position semantic, but we don't know how to handle it correctly so it's better to stop the app than lock the machine.

14 years agor300g: plug memory leak
Maciej Cencora [Sun, 27 Sep 2009 20:18:49 +0000 (22:18 +0200)]
r300g: plug memory leak

14 years agor300g: add some debugging info
Maciej Cencora [Sun, 27 Sep 2009 20:15:15 +0000 (22:15 +0200)]
r300g: add some debugging info

14 years agor300g: fix r300 rs path
Dave Airlie [Sun, 27 Sep 2009 20:42:25 +0000 (06:42 +1000)]
r300g: fix r300 rs path

14 years agor300g: rewrite RS state setup.
Dave Airlie [Sun, 27 Sep 2009 10:34:13 +0000 (20:34 +1000)]
r300g: rewrite RS state setup.

Not 100% sure this is correct, but its more correct than what was here previous
however it may require changes in the input routing for the frag shader.

14 years agor300g: silence compiler warning
Dave Airlie [Sun, 27 Sep 2009 10:31:55 +0000 (20:31 +1000)]
r300g: silence compiler warning

14 years agoi915g: Submit direct vertex buffers
Jakob Bornecrantz [Sun, 27 Sep 2009 20:12:11 +0000 (13:12 -0700)]
i915g: Submit direct vertex buffers

14 years agoi915g: Fix warning
Jakob Bornecrantz [Sun, 27 Sep 2009 20:11:49 +0000 (13:11 -0700)]
i915g: Fix warning

14 years agoi915g: Tile shared buffers as well
Jakob Bornecrantz [Wed, 23 Sep 2009 18:57:18 +0000 (11:57 -0700)]
i915g: Tile shared buffers as well

14 years agoi915g: Make sure to map tiled buffers via the gtt
Jakob Bornecrantz [Wed, 23 Sep 2009 18:54:22 +0000 (11:54 -0700)]
i915g: Make sure to map tiled buffers via the gtt

14 years agoi915g: Use boolean
Jakob Bornecrantz [Wed, 23 Sep 2009 18:53:50 +0000 (11:53 -0700)]
i915g: Use boolean

14 years agoi915g: Enable reuse of buffers
Jakob Bornecrantz [Wed, 23 Sep 2009 00:00:46 +0000 (17:00 -0700)]
i915g: Enable reuse of buffers

14 years agost/egl: Remove buffer from screen
Jakob Bornecrantz [Wed, 23 Sep 2009 03:22:13 +0000 (20:22 -0700)]
st/egl: Remove buffer from screen

It is no longer needed since the new drm api code,
and it incorrectly checked if the buffer where there
for testing completeness when it should have checked
the texture instead.

14 years agor600 : Enable draw_prim.
Richard Li [Sun, 27 Sep 2009 18:47:12 +0000 (14:47 -0400)]
r600 : Enable draw_prim.

14 years agosoftpipe: Grab a ref when the fb is set.
Younes Manton [Sun, 27 Sep 2009 14:56:42 +0000 (10:56 -0400)]
softpipe: Grab a ref when the fb is set.

Nasty bug when the surface is freed and another is allocated right on
top of it. The next time we set the fb state SP thinks it's the same
surface and doesn't flush, and when the flush eventually happens the
surface belongs to a completely different texture.

14 years agoprogs/rbug: Link to math
Jakob Bornecrantz [Sat, 26 Sep 2009 19:35:19 +0000 (12:35 -0700)]
progs/rbug: Link to math

14 years agor300g: add tx depth support in register.
Dave Airlie [Sat, 26 Sep 2009 09:32:46 +0000 (19:32 +1000)]
r300g: add tx depth support in register.

also enable cube/3d bits in txformat reg

14 years agor300g: only pass complete texture state to hw setup function
Dave Airlie [Sat, 26 Sep 2009 08:38:39 +0000 (18:38 +1000)]
r300g: only pass complete texture state to hw setup function

No point passing things twice here, also allows more state
to be setup properly.

14 years agor300g: add missing break in OQ emit
Dave Airlie [Sat, 26 Sep 2009 08:38:07 +0000 (18:38 +1000)]
r300g: add missing break in OQ emit

14 years agollvmpipe: Allow building with LLVM 2.6 too.
José Fonseca [Sat, 26 Sep 2009 08:33:32 +0000 (09:33 +0100)]
llvmpipe: Allow building with LLVM 2.6 too.

14 years agor300g: report GL1.5, enable cap bits for OQ and shadow.
Dave Airlie [Sat, 26 Sep 2009 08:25:00 +0000 (18:25 +1000)]
r300g: report GL1.5, enable cap bits for OQ and shadow.

Its not like it works well on 1.3 so may as well reach for greater
heights.

Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agor300g: add z16 unorm texture format
Dave Airlie [Sat, 26 Sep 2009 08:24:34 +0000 (18:24 +1000)]
r300g: add z16 unorm texture format

14 years agor300g: fix texture pitch to correct value.
Dave Airlie [Sat, 26 Sep 2009 06:39:13 +0000 (16:39 +1000)]
r300g: fix texture pitch to correct value.

pitch is pixels - 1, not bytes.

14 years agoMerge branch 'asm-shader-rework-2'
Ian Romanick [Fri, 25 Sep 2009 23:57:38 +0000 (16:57 -0700)]
Merge branch 'asm-shader-rework-2'

Conflicts:
src/mesa/shader/program_parse.tab.c

14 years agointel: Handle GL_RGB8 for glCopyTex(Sub)Image.
Michel Dänzer [Fri, 25 Sep 2009 18:59:44 +0000 (20:59 +0200)]
intel: Handle GL_RGB8 for glCopyTex(Sub)Image.

Avoids an unnecessary fallback.

14 years agost/xorg: Re-enable accelerated fills and copies.
Michel Dänzer [Fri, 25 Sep 2009 18:49:53 +0000 (20:49 +0200)]
st/xorg: Re-enable accelerated fills and copies.

These seem to work well enough now with the new code. Composite acceleration
isn't quite there yet and thus remains disabled in xorg_composite_bind_state()
for now.

14 years agost/xorg: Flush render cache if but only if a source has pending write operations.
Michel Dänzer [Fri, 25 Sep 2009 18:49:53 +0000 (20:49 +0200)]
st/xorg: Flush render cache if but only if a source has pending write operations.

14 years agost/xorg: Reject Composite acceleration for some cases not working yet.
Michel Dänzer [Fri, 25 Sep 2009 18:49:53 +0000 (20:49 +0200)]
st/xorg: Reject Composite acceleration for some cases not working yet.

14 years agost/xorg: Make sure struct is fully initialized.
Michel Dänzer [Fri, 25 Sep 2009 18:49:00 +0000 (20:49 +0200)]
st/xorg: Make sure struct is fully initialized.

gcc complained about a missing initializer.

14 years agost/xorg: Use generic semantic for Composite mask coordinates.
Michel Dänzer [Fri, 25 Sep 2009 18:38:49 +0000 (20:38 +0200)]
st/xorg: Use generic semantic for Composite mask coordinates.

14 years agost/xorg: Flesh out EXA PrepareComposite hook a little.
Michel Dänzer [Fri, 25 Sep 2009 18:38:49 +0000 (20:38 +0200)]
st/xorg: Flesh out EXA PrepareComposite hook a little.

Check that the formats are supported, and don't crash with source-only
pictures.

14 years agost/xorg: Use I8 format instead of A8 for depth 8 pixmaps.
Michel Dänzer [Fri, 25 Sep 2009 18:38:49 +0000 (20:38 +0200)]
st/xorg: Use I8 format instead of A8 for depth 8 pixmaps.

Seems to work better for Composite acceleration.

14 years agost/xorg: Bind rasterizer state for copies.
Michel Dänzer [Fri, 25 Sep 2009 18:38:49 +0000 (20:38 +0200)]
st/xorg: Bind rasterizer state for copies.

14 years agost/xorg: Source-only pictures always have format PICT_a8r8g8b8.
Michel Dänzer [Fri, 25 Sep 2009 18:38:49 +0000 (20:38 +0200)]
st/xorg: Source-only pictures always have format PICT_a8r8g8b8.

See xserver/render/picture.c:createSourcePicture(). This both simplifies the
code and avoids a crash because pFormat is NULL.

14 years agost/xorg: Better checks for unsupported component alpha pictures.
Michel Dänzer [Fri, 25 Sep 2009 18:38:49 +0000 (20:38 +0200)]
st/xorg: Better checks for unsupported component alpha pictures.

14 years agosoftpipe: Grab fs output z from the correct file.
Michal Krol [Fri, 25 Sep 2009 08:57:33 +0000 (10:57 +0200)]
softpipe: Grab fs output z from the correct file.

14 years agosoftpipe: Do not advertise support for L16 and YCBCR formats.
Michal Krol [Fri, 25 Sep 2009 08:48:19 +0000 (10:48 +0200)]
softpipe: Do not advertise support for L16 and YCBCR formats.

14 years agoradeon: Fix newlines.
Michal Krol [Fri, 25 Sep 2009 16:51:55 +0000 (18:51 +0200)]
radeon: Fix newlines.

14 years agoMerge branch 'mesa_7_6_branch'
Brian Paul [Fri, 25 Sep 2009 15:00:19 +0000 (09:00 -0600)]
Merge branch 'mesa_7_6_branch'

Conflicts:

src/mesa/drivers/dri/intel/intel_clear.c

14 years agomesa: move declaration before code
Vinson Lee [Fri, 25 Sep 2009 14:51:57 +0000 (08:51 -0600)]
mesa: move declaration before code

14 years agonv50: fix TEX for WriteMask not equal 0xf
Christoph Bumiller [Fri, 25 Sep 2009 08:53:01 +0000 (10:53 +0200)]
nv50: fix TEX for WriteMask not equal 0xf

If you e.g. only need alpha, it ends up in the first reg,
not the last, as it would when reading rgb too.

14 years agonv50: actually enable view volume clipping
Christoph Bumiller [Thu, 24 Sep 2009 15:37:08 +0000 (17:37 +0200)]
nv50: actually enable view volume clipping

Until now, only primitives wholly outside the view volume
were not drawn.
This was only visibile when using a viewport smaller than
the window size, naturally.

14 years agonv50: RCP and RSQ cannot load from VP inputs
Christoph Bumiller [Thu, 24 Sep 2009 15:24:48 +0000 (17:24 +0200)]
nv50: RCP and RSQ cannot load from VP inputs

14 years agonv50: fix CEIL and TRUNC
Christoph Bumiller [Fri, 25 Sep 2009 08:33:02 +0000 (10:33 +0200)]
nv50: fix CEIL and TRUNC

Separated the integer rounding mode flag for cvt.

14 years agonv50: implement BGNLOOP, BRK, ENDLOOP
Christoph Bumiller [Fri, 25 Sep 2009 08:24:40 +0000 (10:24 +0200)]
nv50: implement BGNLOOP, BRK, ENDLOOP

There's a good chance a loop won't execute correctly
though since our TEMP allocation assumes programs to
be executed linearly. Will fix later.

14 years agonv50: implement IF, ELSE, ENDIF opcodes
Christoph Bumiller [Thu, 24 Sep 2009 14:39:56 +0000 (16:39 +0200)]
nv50: implement IF, ELSE, ENDIF opcodes

14 years agor300g: add texture format for xvmc
Cooper Yuan [Fri, 25 Sep 2009 07:15:20 +0000 (15:15 +0800)]
r300g: add texture format for xvmc

14 years agomesa/main: Make FEATURE_convolve follow feature conventions.
Chia-I Wu [Wed, 23 Sep 2009 08:56:20 +0000 (16:56 +0800)]
mesa/main: Make FEATURE_convolve follow feature conventions.

As shown in mfeatures.h, this allows users of convolve.h to work without
knowing if the feature is available.

14 years agomesa/main: Make FEATURE_colortable follow feature conventions.
Chia-I Wu [Mon, 7 Sep 2009 09:17:11 +0000 (17:17 +0800)]
mesa/main: Make FEATURE_colortable follow feature conventions.

As shown in mfeatures.h, this allows users of colortab.h to work
without knowing if the feature is available.

14 years agomesa/main: Make FEATURE_pixel_transfer follow feature conventions.
Chia-I Wu [Mon, 7 Sep 2009 08:59:27 +0000 (16:59 +0800)]
mesa/main: Make FEATURE_pixel_transfer follow feature conventions.

As shown in mfeatures.h, this allows users of pixel.h to work without
knowing if the feature is available.