mesa.git
15 years agomesa: minor code simplification in _mesa_GetVertexAttribfvARB()
Brian Paul [Thu, 21 May 2009 21:57:29 +0000 (15:57 -0600)]
mesa: minor code simplification in _mesa_GetVertexAttribfvARB()

15 years agomesa: remove const qualifier
Brian Paul [Thu, 21 May 2009 21:55:50 +0000 (15:55 -0600)]
mesa: remove const qualifier

15 years agomesa: VertexAttribPointer comments
Brian Paul [Thu, 21 May 2009 21:55:33 +0000 (15:55 -0600)]
mesa: VertexAttribPointer comments

15 years agomesa: call _mesa_update_array_object_max_element() before printing array info
Brian Paul [Thu, 21 May 2009 21:36:25 +0000 (15:36 -0600)]
mesa: call _mesa_update_array_object_max_element() before printing array info

15 years agomesa: added gl_array_object::Weight array field
Brian Paul [Thu, 21 May 2009 16:15:18 +0000 (10:15 -0600)]
mesa: added gl_array_object::Weight array field

We don't really implement vertex weights but in the VBO code this
fixes and odd case for the legacy_array[] setup.  Before, the
vbo->draw_prims() call was always indicating that the vertex weight
array was present/enabled when it really wasn't.

15 years agomesa: check FEATURE_point_size_array for PointSize array
Brian Paul [Thu, 21 May 2009 16:11:13 +0000 (10:11 -0600)]
mesa: check FEATURE_point_size_array for PointSize array

15 years agomesa: fix some potential state-restore issues in pop_texture_group()
Brian Paul [Thu, 21 May 2009 16:05:04 +0000 (10:05 -0600)]
mesa: fix some potential state-restore issues in pop_texture_group()

Call the _mesa_set_enable() functions instead of driver functions, etc.

Also, add missing code for 1D/2D texture arrays.

15 years agomesa: new _mesa_update_array_object_max_element() function
Brian Paul [Thu, 21 May 2009 15:56:41 +0000 (09:56 -0600)]
mesa: new _mesa_update_array_object_max_element() function

This will replace the code in state.c

15 years agovbo: return VP_NONE from get_program_mode() if running fixed-func vertex program
Brian Paul [Thu, 21 May 2009 15:52:33 +0000 (09:52 -0600)]
vbo: return VP_NONE from get_program_mode() if running fixed-func vertex program

If we're running a vertex program to emulated fixed-function, we still need
to treat vertex arrays/attributes as if we're in fixed-function mode.

This should probably be back-ported to Mesa 7.5 after a bit more testing.

15 years agovbo: move vp_mode enum to vbo_exec.h, use enum instead of GLuint
Brian Paul [Thu, 21 May 2009 15:50:55 +0000 (09:50 -0600)]
vbo: move vp_mode enum to vbo_exec.h, use enum instead of GLuint

15 years agoi965: fix whitespace in brw_tex_layout.c
Eric Anholt [Thu, 21 May 2009 04:15:46 +0000 (21:15 -0700)]
i965: fix whitespace in brw_tex_layout.c

The broken indentation was driving me crazy, so fix other stuff while
I'm here.

15 years agoi956: Make state dependency of SF on drawbuffer bounds match Mesa's.
Eric Anholt [Wed, 20 May 2009 22:59:07 +0000 (15:59 -0700)]
i956: Make state dependency of SF on drawbuffer bounds match Mesa's.

Noticed while debugging a weird 1D FBO testcase that left its existing
viewport and projection matrix in place when switching drawbuffers.  Didn't
fix the testcase, though.

15 years agointel: Fall back on any rendering to texture with no miptree.
Eric Anholt [Wed, 20 May 2009 21:16:34 +0000 (14:16 -0700)]
intel: Fall back on any rendering to texture with no miptree.

Fixes segfault on an fbo.c negative test for FBO with texture width/height
of 0.  Previously we just tested for border != 0 to work around this
segfault.

15 years agointel: Mark the FBO as incomplete if there's no intel_renderbuffer for it.
Eric Anholt [Wed, 20 May 2009 21:05:03 +0000 (14:05 -0700)]
intel: Mark the FBO as incomplete if there's no intel_renderbuffer for it.

This happens to rendering with textures with a border, which had resulted
in a segfault on dereferencing the irb.

15 years agointel: Don't segfault on glGenerateMipmaps of a cube map with one face defined.
Eric Anholt [Wed, 20 May 2009 21:00:32 +0000 (14:00 -0700)]
intel: Don't segfault on glGenerateMipmaps of a cube map with one face defined.

This presumably applies to SGIS_generate_mipmaps as well.

15 years agointel: Use _mesa_CheckFramebufferStatusEXT insteaad of glCheck...
Eric Anholt [Wed, 20 May 2009 20:09:04 +0000 (13:09 -0700)]
intel: Use _mesa_CheckFramebufferStatusEXT insteaad of glCheck...

Fixes a segfault in our oglconform fbo test.

15 years agomesa: added debug functions for dumping color/depth/stencil buffers
Brian Paul [Thu, 21 May 2009 15:12:35 +0000 (09:12 -0600)]
mesa: added debug functions for dumping color/depth/stencil buffers

15 years agoi965: rename var: s/tmp/vs_inputs/
Brian Paul [Fri, 15 May 2009 15:14:24 +0000 (09:14 -0600)]
i965: rename var: s/tmp/vs_inputs/

15 years agomesa: remove pointless null ptr check, improved some error messages
Brian Paul [Thu, 14 May 2009 22:51:10 +0000 (16:51 -0600)]
mesa: remove pointless null ptr check, improved some error messages

15 years agomesa: added _mesa_print_arrays() for debugging
Brian Paul [Thu, 14 May 2009 22:25:32 +0000 (16:25 -0600)]
mesa: added _mesa_print_arrays() for debugging

15 years agomesa: updated comment for _MaxElement field
Brian Paul [Thu, 14 May 2009 22:23:34 +0000 (16:23 -0600)]
mesa: updated comment for _MaxElement field

It's the largest array index, plus one.

15 years agomesa: move gl_array_attrib::_MaxElement to gl_array_object::_MaxElement
Brian Paul [Thu, 14 May 2009 19:24:24 +0000 (13:24 -0600)]
mesa: move gl_array_attrib::_MaxElement to gl_array_object::_MaxElement

This value is per array object.

15 years agovbo: s/8/MAX_TEXTURE_COORD_UNITS/
Brian Paul [Thu, 14 May 2009 14:22:29 +0000 (08:22 -0600)]
vbo: s/8/MAX_TEXTURE_COORD_UNITS/

15 years agocell: perform triangle cull a little earlier
Jonathan Adamczewski [Thu, 21 May 2009 14:19:00 +0000 (08:19 -0600)]
cell: perform triangle cull a little earlier

In spu_tri.c:setup_sort_vertices() triangles are culled after the
vertices are sorted.  This patch moves the check a little earlier
and performs the actual check a little faster through intrinsics and
a little trickery.

Reduced code size and less work is done before a triangle is deemed
OK to skip.

15 years agocell: unroll inner loop of spu_render.c:cmd_render()
Jonathan Adamczewski [Thu, 21 May 2009 14:18:03 +0000 (08:18 -0600)]
cell: unroll inner loop of spu_render.c:cmd_render()

It was taking approximately 50 cycles to extract the vertex indices,
calculate the vertex_header pointers and call tri_draw() for each
three vertices - .

Unrolled, it takes less than 100 cycles to extract, unpack,
calculate pointers and call tri_draw() eight times.  It does have a
nasty jump-tabled switch.  I'm sure that there's a better way...

Code size of spu_render.o gets larger due to the extra constants and
work in the inner loop, there are extra stack saves and loads
because there are more registers in use, and an assert.  spu_tri.o
gets a little smaller.

15 years agost: add support for GL_EXT_vertex_array_bgra
Brian Paul [Thu, 21 May 2009 14:15:22 +0000 (08:15 -0600)]
st: add support for GL_EXT_vertex_array_bgra

15 years agor300-gallium: r500-fs: POW.
Corbin Simpson [Thu, 21 May 2009 06:22:16 +0000 (23:22 -0700)]
r300-gallium: r500-fs: POW.

I feel so unclean.

15 years agor300-gallium: r500-fs: LRP.
Corbin Simpson [Thu, 21 May 2009 04:52:11 +0000 (21:52 -0700)]
r300-gallium: r500-fs: LRP.

Goddammit. This cannot be the "easy way." :C

15 years agor300-gallium: r500-fs: Combine function.
Corbin Simpson [Wed, 20 May 2009 23:53:45 +0000 (16:53 -0700)]
r300-gallium: r500-fs: Combine function.

15 years agor300-gallium: Prevent assert when fogcoords are present.
Corbin Simpson [Wed, 20 May 2009 22:02:45 +0000 (15:02 -0700)]
r300-gallium: Prevent assert when fogcoords are present.

Seems like this file is the source of all bad logic. (Pun intended.)

15 years agor300-gallium: Another constantbuf shader recompile test.
Corbin Simpson [Wed, 20 May 2009 21:55:03 +0000 (14:55 -0700)]
r300-gallium: Another constantbuf shader recompile test.

Less briefly... Shaders need to be recompiled if their constantbuf
offsets have changed. However, since we only change them from shaders if
immediates need to be emitted, we shouldn't bother if the shader doesn't
use immediates.

15 years agor300-gallium: Raise constantbuf limits.
Corbin Simpson [Wed, 20 May 2009 21:38:22 +0000 (14:38 -0700)]
r300-gallium: Raise constantbuf limits.

Still not correct, but really I don't care.

15 years agor300-gallium: fs: Remove cruft from way back when.
Corbin Simpson [Wed, 20 May 2009 21:17:27 +0000 (14:17 -0700)]
r300-gallium: fs: Remove cruft from way back when.

15 years agoradeon-gallium: Add surface_buffer_create callback.
Corbin Simpson [Wed, 20 May 2009 20:21:17 +0000 (13:21 -0700)]
radeon-gallium: Add surface_buffer_create callback.

15 years agor300-gallium: Make surface_copy actually load the texture in shader.
Corbin Simpson [Wed, 20 May 2009 19:22:24 +0000 (12:22 -0700)]
r300-gallium: Make surface_copy actually load the texture in shader.

15 years agor300-gallium: Add missing R481 PCI ID.
Corbin Simpson [Wed, 20 May 2009 19:08:00 +0000 (12:08 -0700)]
r300-gallium: Add missing R481 PCI ID.

Per 74cb2aba on xf86-video-ati.

15 years agor300-gallium: Make surface_copy work, and refactor buffer validation.
Corbin Simpson [Wed, 20 May 2009 18:46:26 +0000 (11:46 -0700)]
r300-gallium: Make surface_copy work, and refactor buffer validation.

15 years agoradeon-gallium: Don't permit reading and writing a BO in one CS.
Corbin Simpson [Wed, 20 May 2009 14:18:08 +0000 (07:18 -0700)]
radeon-gallium: Don't permit reading and writing a BO in one CS.

This fixes some silent problems in current libdrm_radeon.

surface_copy still locks up hard.

15 years agost: reformatting, comments, var renaming
Brian Paul [Wed, 13 May 2009 01:51:44 +0000 (19:51 -0600)]
st: reformatting, comments, var renaming

15 years agoMerge branch 'mesa_7_5_branch'
Brian Paul [Tue, 19 May 2009 16:07:45 +0000 (10:07 -0600)]
Merge branch 'mesa_7_5_branch'

15 years agomesa: allow depth/stencil textures to be attached to GL_STENCIL_ATTACHMENT
Mathias Fröhlich [Tue, 19 May 2009 15:59:01 +0000 (09:59 -0600)]
mesa: allow depth/stencil textures to be attached to GL_STENCIL_ATTACHMENT

See sourceforge bug #2793846.

15 years agomesa: assign trb->Base.StencilBits in update_wrapper().
Brian Paul [Tue, 19 May 2009 15:57:01 +0000 (09:57 -0600)]
mesa: assign trb->Base.StencilBits in update_wrapper().

When we render to a depth/stencil texture there are stencil bits.

15 years agomesa: print more info when valid_texture_object() fails
Brian Paul [Tue, 19 May 2009 15:21:27 +0000 (09:21 -0600)]
mesa: print more info when valid_texture_object() fails

15 years agotrace: Improve shader wrapping
Jakob Bornecrantz [Sat, 16 May 2009 02:13:15 +0000 (03:13 +0100)]
trace: Improve shader wrapping

15 years agost/dri: Only create new textures if drawable has changed
Jakob Bornecrantz [Mon, 18 May 2009 01:50:15 +0000 (02:50 +0100)]
st/dri: Only create new textures if drawable has changed

15 years agor300-gallium: Fix (another) wrong value in MSPOS.
Corbin Simpson [Mon, 18 May 2009 16:50:30 +0000 (09:50 -0700)]
r300-gallium: Fix (another) wrong value in MSPOS.

Again, thanks to agd5f.

15 years agoradeon-gallium: Remove BO validation debug.
Corbin Simpson [Mon, 18 May 2009 16:47:37 +0000 (09:47 -0700)]
radeon-gallium: Remove BO validation debug.

It appears that that area of code "just works" much like classic Mesa's
version, so might as well not waste scrollback on it.

15 years agor300-gallium: Cleanup viewport state setup.
Corbin Simpson [Mon, 18 May 2009 16:41:21 +0000 (09:41 -0700)]
r300-gallium: Cleanup viewport state setup.

15 years agor300-gallium: Always do VTE, never software viewport.
Corbin Simpson [Mon, 18 May 2009 16:40:13 +0000 (09:40 -0700)]
r300-gallium: Always do VTE, never software viewport.

This makes glxgears draw properly with SW TCL.

15 years agoMerge branch 'mesa_7_5_branch'
Brian Paul [Mon, 18 May 2009 16:36:50 +0000 (10:36 -0600)]
Merge branch 'mesa_7_5_branch'

Conflicts:

Makefile
src/mesa/main/version.h

15 years agost/mesa: fix incorrect src/dst stride params to _mesa_generate_mipmap_level()
Brian Paul [Mon, 18 May 2009 16:28:04 +0000 (10:28 -0600)]
st/mesa: fix incorrect src/dst stride params to _mesa_generate_mipmap_level()

The stride needs to be in texels, not bytes.

15 years agomesa: comments for _mesa_generate_mipmap_level()
Brian Paul [Mon, 18 May 2009 16:27:31 +0000 (10:27 -0600)]
mesa: comments for _mesa_generate_mipmap_level()

15 years agosoftpipe: add texture target sanity check assertion
Brian Paul [Mon, 18 May 2009 16:13:44 +0000 (10:13 -0600)]
softpipe: add texture target sanity check assertion

15 years agost: fix incorrect target parameter to screen->is_format_supported()
Brian Paul [Mon, 18 May 2009 16:12:36 +0000 (10:12 -0600)]
st: fix incorrect target parameter to screen->is_format_supported()

We were passing a GL texture target instead of a pipe_texture_target enum.

15 years agomesa: added linux-osmesa-static config
Brian Paul [Mon, 18 May 2009 14:46:17 +0000 (08:46 -0600)]
mesa: added linux-osmesa-static config

Contributed by Nicolas Noble.  See SF bug #2792536

15 years agodocs: link to 7.5 relnotes
Brian Paul [Mon, 18 May 2009 13:17:46 +0000 (07:17 -0600)]
docs: link to 7.5 relnotes

15 years agodocs: emphasize the incomplete status of the nouveau/R300 gallium drivers
Brian Paul [Mon, 18 May 2009 13:17:36 +0000 (07:17 -0600)]
docs: emphasize the incomplete status of the nouveau/R300 gallium drivers

15 years agoInitialize psp->waitX/waitGL for swrast_dri.so.
Aidan Thornton [Mon, 18 May 2009 10:47:00 +0000 (12:47 +0200)]
Initialize psp->waitX/waitGL for swrast_dri.so.

Fixes http://bugs.freedesktop.org/show_bug.cgi?id=21053 .

15 years agor300-gallium: Enable GLSL for r500.
Corbin Simpson [Mon, 18 May 2009 04:41:25 +0000 (21:41 -0700)]
r300-gallium: Enable GLSL for r500.

Before you get all excited, this is *not* to be construed as actual support
for GLSL shaders. The GL version is still 1.3, and stuff still sucks. Just
flicking it on so that it can be tested and developed a bit easier.

15 years agor300-gallium: r500-fs: DDX and DDY support.
Corbin Simpson [Mon, 18 May 2009 04:40:16 +0000 (21:40 -0700)]
r300-gallium: r500-fs: DDX and DDY support.

Oh, look, GLSL instructions. I wonder what I'll do next.

15 years agodri-gallium: Add GLSL support.
Corbin Simpson [Mon, 18 May 2009 04:39:11 +0000 (21:39 -0700)]
dri-gallium: Add GLSL support.

Oh, look, it's more features. :3

15 years agor300-gallium: Add half-right COS and SIN.
Corbin Simpson [Mon, 18 May 2009 04:10:07 +0000 (21:10 -0700)]
r300-gallium: Add half-right COS and SIN.

HW trig does a premultiply by 2pi, where Mesa does another premultiply by pi.
This is a problem.

15 years agor300-gallium: Size mismatch.
Corbin Simpson [Mon, 18 May 2009 03:49:39 +0000 (20:49 -0700)]
r300-gallium: Size mismatch.

15 years agor300-gallium, radeon-gallium: Nuke gb_pipes from orbit.
Corbin Simpson [Mon, 18 May 2009 00:03:15 +0000 (17:03 -0700)]
r300-gallium, radeon-gallium: Nuke gb_pipes from orbit.

See the previous commit for an explanation. This is just all the support code
for GB_TILE_CONFIG.

15 years agor300-gallium: Don't set GB_TILE_CONFIG (in userspace.)
Corbin Simpson [Sun, 17 May 2009 23:42:54 +0000 (16:42 -0700)]
r300-gallium: Don't set GB_TILE_CONFIG (in userspace.)

This accompanies kernel patches that make GB_TILE_CONFIG's various members
completely controlled in DRM.

GB_TILE_CONFIG has the following controls:
 - The number of GB (pixel) pipes enabled
 - The size and style of tiling
 - Subpixel precision (either 1/12 or 1/16)

Per airlied and glisse, userspace and kernel will now agree (always) on
a subpixel precision of 1/12, and tiling will always be kernel-controlled.

15 years agor300-gallium: Clean up more invariant state.
Corbin Simpson [Sun, 17 May 2009 20:23:39 +0000 (13:23 -0700)]
r300-gallium: Clean up more invariant state.

GA_ENHANCE is now the kernel's problem.

15 years agor300-gallium: Update XXX.
Corbin Simpson [Sun, 17 May 2009 19:51:18 +0000 (12:51 -0700)]
r300-gallium: Update XXX.

Lops work fine as long as HW TCL is off. (I think I know why.)

15 years agor300-gallium: Correct default MSPOS.
Corbin Simpson [Sun, 17 May 2009 18:58:53 +0000 (11:58 -0700)]
r300-gallium: Correct default MSPOS.

Per agd5f.

15 years agor300-gallium: vs: Fix vert shader init.
Corbin Simpson [Sun, 17 May 2009 17:33:56 +0000 (10:33 -0700)]
r300-gallium: vs: Fix vert shader init.

Makes the last three commits suck much less. :3

15 years agor300-gallium: r500-fs: Enable depth writes, kinda.
Corbin Simpson [Sun, 17 May 2009 17:30:59 +0000 (10:30 -0700)]
r300-gallium: r500-fs: Enable depth writes, kinda.

Should work, but doesn't. Hm.

15 years agor300-gallium: Comment out useless debugging code.
Corbin Simpson [Sun, 17 May 2009 16:44:14 +0000 (09:44 -0700)]
r300-gallium: Comment out useless debugging code.

Those parts are nearly solid compared to the shaders.

15 years agor300-gallium: vs: Dupe tokens, better debug, count spurious insts.
Corbin Simpson [Sun, 17 May 2009 16:39:06 +0000 (09:39 -0700)]
r300-gallium: vs: Dupe tokens, better debug, count spurious insts.

15 years agor300-gallium: Die on bad texture formats.
Corbin Simpson [Sat, 16 May 2009 18:17:05 +0000 (11:17 -0700)]
r300-gallium: Die on bad texture formats.

Odds are good that we'll die later anyway, so we might as well do it before
we start dancing on random memory.

15 years agor300-gallium: Update floating-point params too.
Corbin Simpson [Sat, 16 May 2009 17:56:17 +0000 (10:56 -0700)]
r300-gallium: Update floating-point params too.

Even though we *can* render 10,000-pixel-wide lines, let's not advertise it.

15 years agor300-gallium: Update screen caps.
Corbin Simpson [Sat, 16 May 2009 17:46:55 +0000 (10:46 -0700)]
r300-gallium: Update screen caps.

Anisotropic filtering should work, and OQ is broken.

15 years agointel-gallium: Fix trace_drm integration.
Corbin Simpson [Sat, 16 May 2009 17:03:46 +0000 (10:03 -0700)]
intel-gallium: Fix trace_drm integration.

Compile-tested only, sorry.

15 years agoCreate common trace_drm code, add to radeon_winsys.
Corbin Simpson [Sat, 16 May 2009 16:58:54 +0000 (09:58 -0700)]
Create common trace_drm code, add to radeon_winsys.

15 years agor300-gallium: Various cleanups leftover from before.
Corbin Simpson [Sat, 16 May 2009 15:47:36 +0000 (08:47 -0700)]
r300-gallium: Various cleanups leftover from before.

BEGIN/END_CS pair, a few asserts, and a slightly more correct VTE setup.

15 years agogallium: remove occlusion_count flag from depth-stencil state
Keith Whitwell [Sat, 16 May 2009 16:25:26 +0000 (17:25 +0100)]
gallium: remove occlusion_count flag from depth-stencil state

Drivers can just keep track of whether they are within a query
by monitoring the begin/end query callbacks.  The flag adds no
information beyond that.

Only softpipe was examining this flag -- it has been fixed up
and retested with demos/arbocclude.

15 years agogallium-intel: Enable trace driver
Jakob Bornecrantz [Sat, 16 May 2009 00:21:02 +0000 (01:21 +0100)]
gallium-intel: Enable trace driver

15 years agotrace: Export enabled status
Jakob Bornecrantz [Sat, 16 May 2009 00:49:11 +0000 (01:49 +0100)]
trace: Export enabled status

15 years agotrace: Unwrap buffer in texture_blanket
Jakob Bornecrantz [Sat, 16 May 2009 00:19:44 +0000 (01:19 +0100)]
trace: Unwrap buffer in texture_blanket

15 years agotrace: If either shader is disabled don't draw
Jakob Bornecrantz [Fri, 15 May 2009 03:59:24 +0000 (05:59 +0200)]
trace: If either shader is disabled don't draw

15 years agotrace: Put shaders on a list in the context
Jakob Bornecrantz [Fri, 15 May 2009 03:30:43 +0000 (05:30 +0200)]
trace: Put shaders on a list in the context

15 years agotrace: Move state dump functions to tr_dump_state.[c|h]
Jakob Bornecrantz [Fri, 15 May 2009 02:17:00 +0000 (04:17 +0200)]
trace: Move state dump functions to tr_dump_state.[c|h]

15 years agointel: Don't complain on falling back from PBO fastpaths.
Eric Anholt [Sat, 16 May 2009 00:32:21 +0000 (17:32 -0700)]
intel: Don't complain on falling back from PBO fastpaths.

Instead, stash the debug info under the handy debug flag.

Bug #20053

15 years agomesa: Mark FBOs with compressed color attachments as FBO-incomplete.
Eric Anholt [Fri, 15 May 2009 23:24:59 +0000 (16:24 -0700)]
mesa: Mark FBOs with compressed color attachments as FBO-incomplete.

Both EXT_fbo and ARB_fbo agree on this.  Fixes a segfault in the metaops
mipmap generation in Intel for SGIS_generate_mipmap of S3TC textures in
Regnum Online.

Bug #21654.

15 years agogallium-intel: Build with scons
Jakob Bornecrantz [Fri, 15 May 2009 21:54:07 +0000 (22:54 +0100)]
gallium-intel: Build with scons

15 years agost/egl: Use proper Makefile
Jakob Bornecrantz [Fri, 15 May 2009 21:47:31 +0000 (22:47 +0100)]
st/egl: Use proper Makefile

15 years agoi915: Fix 945 cube map layout for the small mipmaps along the bottom.
Steinar H. Gunderson [Tue, 12 May 2009 18:32:03 +0000 (11:32 -0700)]
i915: Fix 945 cube map layout for the small mipmaps along the bottom.

Bug #21691.

15 years agoi915: Use Stencil.Enabled instead of Stencil._Enabled in DrawBuffers.
Eric Anholt [Fri, 15 May 2009 19:32:51 +0000 (12:32 -0700)]
i915: Use Stencil.Enabled instead of Stencil._Enabled in DrawBuffers.

The _Enabled field isn't updated at the point that DrawBuffers is called,
and the Driver.Enable() function does the testing for stencil buffer
presence anyway.

bug #21608 for Radeon

15 years agoi915: Only use the new 945 cube layout for compressed textures.
Eric Anholt [Fri, 15 May 2009 18:41:42 +0000 (11:41 -0700)]
i915: Only use the new 945 cube layout for compressed textures.

The docs actually explain this, but not in a terribly clear manner.
This nearly fixes the piglit cubemap testcase, except that something's
going wrong with the nearest filtering at 2x2 sizes in the testcase.
Looks good by visual inspection, though.

Bug #21692

15 years agomesa: bump version to 7.5-rc2
Brian Paul [Fri, 15 May 2009 14:03:56 +0000 (08:03 -0600)]
mesa: bump version to 7.5-rc2

15 years agodocs: updates from the 7.4 branch
Brian Paul [Fri, 15 May 2009 14:02:40 +0000 (08:02 -0600)]
docs: updates from the 7.4 branch

15 years agoi965: Fix varying payload reg assignment for the non-GLSL-instructions path.
Eric Anholt [Thu, 14 May 2009 17:56:32 +0000 (10:56 -0700)]
i965: Fix varying payload reg assignment for the non-GLSL-instructions path.

I don't have a testcase for this, but it seems clearly wrong.

15 years agoi965: Fix register allocation of GLSL fp inputs.
Eric Anholt [Thu, 14 May 2009 16:49:45 +0000 (09:49 -0700)]
i965: Fix register allocation of GLSL fp inputs.

Before, if the VP output something that is in the attributes coming into
the WM but which isn't used by the WM, then WM would end up reading subsequent
varyings from the wrong places.  This was visible with a GLSL demo
using gl_PointSize in the VS and a varying in the WM, as point size is in
the VUE but not used by the WM.  There is now a regression test in piglit,
glsl-unused-varying.

15 years agointel: Use FRONT_AND_BACK for StencilOp as well.
Eric Anholt [Thu, 14 May 2009 02:08:17 +0000 (19:08 -0700)]
intel: Use FRONT_AND_BACK for StencilOp as well.

15 years agointel: Use GL_FRONT_AND_BACK for stencil clearing.
Eric Anholt [Thu, 14 May 2009 01:18:29 +0000 (18:18 -0700)]
intel: Use GL_FRONT_AND_BACK for stencil clearing.

This comes from a radeon-rewrite fallback fix, but may also fix stencil
clear failure when the polygon winding mode is flipped.

15 years agor300-gallium: Correct VTE setup for surface_fill, make surface_copy emit right.
Corbin Simpson [Thu, 14 May 2009 15:17:08 +0000 (08:17 -0700)]
r300-gallium: Correct VTE setup for surface_fill, make surface_copy emit right.

15 years agoi965: fix 1D texture borders with GL_CLAMP_TO_BORDER
Robert Ellison [Thu, 14 May 2009 02:40:23 +0000 (20:40 -0600)]
i965: fix 1D texture borders with GL_CLAMP_TO_BORDER

With 1D textures, GL_TEXTURE_WRAP_T should be ignored (only
GL_TEXTURE_WRAP_S should be respected).  But the i965 hardware
seems to follow the value of GL_TEXTURE_WRAP_T even when sampling
1D textures.

This fix forces GL_TEXTURE_WRAP_T to be GL_REPEAT whenever 1D
textures are used; this allows the texture to be sampled
correctly, avoiding "imaginary" border elements in the T direction.

This bug was demonstrated in the Piglit tex1d-2dborder test.
With this fix, that test passes.