Marek Olšák [Fri, 8 Apr 2011 21:16:17 +0000 (23:16 +0200)]
r300g: fix another crash with hyperz
https://bugs.freedesktop.org/show_bug.cgi?id=36086
Jon TURNEY [Mon, 14 Mar 2011 22:08:23 +0000 (22:08 +0000)]
Disable direct rendering on Cygwin
Add Cygwin platform-specific settings and drivers to build for dri driver:
- by default, disable direct rendering.
- if direct rendering is enabled, the swrast dridriver is the only one it's
sensible to try to build (this doesn't work at the moment as additional patches
are required to build a libGL which can load just swrast without the DRM headers,
even though there's no actual functional dependency)
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Julien Cristau <jcristau@debian.org>
nobled [Mon, 14 Mar 2011 22:08:22 +0000 (22:08 +0000)]
Disable direct rendering on GNU/Hurd
The Hurd kernel doesn't have DRM yet.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Julien Cristau <jcristau@debian.org>
Reviewed-By: Jakob Bornecrantz <wallbraker@gmail.com>
Samuel Thibault [Mon, 14 Mar 2011 22:08:21 +0000 (22:08 +0000)]
Only require libdrm if direct rendering is actually enabled.
Fix build when configured --with-driver=dri --disable-driglx-direct on targets
without drm e.g. GNU/Hurd and Cygwin
Based on the Debian patch file '05_hurd-ftbfs.diff' by Samuel Thibault.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Julien Cristau <jcristau@debian.org>
Reviewed-By: Jakob Bornecrantz <wallbraker@gmail.com>
Marek Olšák [Fri, 8 Apr 2011 18:42:03 +0000 (20:42 +0200)]
r300g: fix crash when adding a new relocation
No idea why it's so hard to reproduce.
Broken with:
c35572352e3e92683988ee8d151b47f4190d62f9
Thanks to Toni Spets for assistance.
Marek Olšák [Fri, 8 Apr 2011 18:05:33 +0000 (20:05 +0200)]
r300g: emit US_FORMAT on R520 only
Eric Anholt [Sun, 3 Apr 2011 04:11:32 +0000 (18:11 -1000)]
i965/fs: Remove broken optimization for live intervals in loops.
The theory here was to detect a temporary variable used within a loop,
and avoid considering it live across the entire loop. However, it was
overeager and failed when the first definition of the variable
appeared within the loop but was only conditionally defined.
Fixes glsl-fs-loop-redundant-condition.
Marek Olšák [Fri, 8 Apr 2011 17:31:26 +0000 (19:31 +0200)]
u_vbuf_mgr: fix crash with unsupported vertex format in hw VBOs
Marek Olšák [Fri, 8 Apr 2011 11:27:14 +0000 (13:27 +0200)]
r300g: fix crash with hyperz
r300_update_hyperz_state is no longer required to be called here.
Henri Verbeet [Mon, 17 Jan 2011 21:03:30 +0000 (22:03 +0100)]
r300g: Properly take the sampler view's first_level into account when sampling.
Signed-off-by: Brian Paul <brianp@vmware.com>
Signed-off-by: Marek Olšák <maraeo@gmail.com>
Fabian Bieler [Thu, 31 Mar 2011 10:32:52 +0000 (12:32 +0200)]
llvmpipe: Take the sampler view's first_level into account when sampling.
Henri Verbeet [Mon, 17 Jan 2011 21:03:30 +0000 (22:03 +0100)]
softpipe: Take the sampler view's first_level into account when sampling.
Signed-off-by: Brian Paul <brianp@vmware.com>
Henri Verbeet [Mon, 17 Jan 2011 21:03:30 +0000 (22:03 +0100)]
st/mesa: Clamp min_lod to the max miplevel as well.
Otherwise min_lod can potentially be larger than the clamped max_lod. The
code that follows will swap min_lod and max_lod in that case, resulting in a
max_lod larger than MAX_LEVEL.
Signed-off-by: Brian Paul <brianp@vmware.com>
Henri Verbeet [Mon, 17 Jan 2011 21:03:30 +0000 (22:03 +0100)]
st/mesa: Set samplers views' first_level.
Base level and min LOD aren't equivalent. In particular, min LOD has no
effect on image array selection for magnification and non-mipmapped
minification.
Signed-off-by: Brian Paul <brianp@vmware.com>
Marek Olšák [Fri, 8 Apr 2011 02:37:19 +0000 (04:37 +0200)]
r300g: do not advertise R16F and RG16F on DRM < 2.8.0
Marek Olšák [Fri, 8 Apr 2011 01:11:22 +0000 (03:11 +0200)]
r300g: fix texturing with sizes > 2048 on r500
Marek Olšák [Thu, 7 Apr 2011 21:36:18 +0000 (23:36 +0200)]
r300g: add 'radeon: ' prefix to error messages in winsys
Marek Olšák [Tue, 5 Apr 2011 23:23:51 +0000 (01:23 +0200)]
mesa: fix dstRowDiff computation in RGTC texstore functions
Copied from libtxc_dxtn, this fixes NPOT RGTC1 textures with r300g.
I also did the same for RGTC2.
Kristian Høgsberg [Thu, 7 Apr 2011 22:27:06 +0000 (18:27 -0400)]
egl_dri2.c: Open drm device with O_CLOEXEC
Not a good fd to leak.
Henri Verbeet [Thu, 7 Apr 2011 20:21:20 +0000 (22:21 +0200)]
r600g: Unreference framebuffer state on context destruction.
Signed-off-by: Henri Verbeet <hverbeet@gmail.com>
Henri Verbeet [Thu, 7 Apr 2011 20:21:20 +0000 (22:21 +0200)]
r600g: Use symbolic names for NUMBER_TYPE.
Signed-off-by: Henri Verbeet <hverbeet@gmail.com>
Brian Paul [Thu, 7 Apr 2011 19:56:45 +0000 (13:56 -0600)]
docs: replace llvmpipe/README with docs/llvmpipe.html
Brian Paul [Thu, 7 Apr 2011 19:43:00 +0000 (13:43 -0600)]
llvmpipe: document issue with LLVM 2.8 and earlier with AVX
Brian Paul [Thu, 7 Apr 2011 19:41:52 +0000 (13:41 -0600)]
util: add cpu detection for sse 4.2 and avx
Brian Paul [Thu, 7 Apr 2011 19:41:52 +0000 (13:41 -0600)]
util: simplify bit shifting in util_cpu_detect()
Brian Paul [Thu, 7 Apr 2011 17:25:46 +0000 (11:25 -0600)]
svga: another tweak to adjust_x term (use -0.5)
José Fonseca [Thu, 7 Apr 2011 12:35:25 +0000 (13:35 +0100)]
mesa: Update _ElementSize in a few more places.
José Fonseca [Thu, 7 Apr 2011 11:19:14 +0000 (12:19 +0100)]
mesa/st: Fix user buffer size computation when stride is zero.
Although for GL a zero stride means tightly packed elements, Mesa
internally uses zero strides for constant arrays.
Therefore user buffers need to be defined from
buffer_offset + src_offset + min_index*stride
to
buffer_offset + src_offset + max_index*stride + elem_size
Simplifying the later with (max_index + 1)*stride will give zero
sized buffers.
This change also aggregates the st_context's info about user buffers
into a single array.
Brian Paul [Thu, 7 Apr 2011 11:07:32 +0000 (12:07 +0100)]
mesa: Update _ElementSize.
José Fonseca [Wed, 6 Apr 2011 14:10:19 +0000 (15:10 +0100)]
mesa/st: Prevent 'end' < 'start' in vbo_exec_DrawRangeElementsBaseVertex()
We adjust 'end' to fit into _MaxElement, but that may result into a 'start'
value bigger than 'end' being passed downstream, causing havoc.
This could be seen with arb_robustness_draw-vbo-bounds, due to an
application bug.
José Fonseca [Wed, 6 Apr 2011 13:38:26 +0000 (14:38 +0100)]
mesa/st: Handle feedback draw when VBO offsets are not sorted in ascending order.
Simply port the same logic from setup_interleaved_attribs().
Avoids overflow in mustpass.c and feedback.c conform tests.
Zou Nan hai [Thu, 7 Apr 2011 13:07:57 +0000 (21:07 +0800)]
i965: clear global offset to zero in m0.2 for VS DP read.
Signed-off-by: Zou Nan hai <nanhai.zou@intel.com>
Vinson Lee [Wed, 6 Apr 2011 23:13:15 +0000 (16:13 -0700)]
mesa: Fix allocation size of extension_indices array.
Reviewed-by: Brian Paul <brianp@vmware.com>
Brian Paul [Wed, 6 Apr 2011 20:22:39 +0000 (14:22 -0600)]
mesa: fix error message text
Brian Paul [Wed, 6 Apr 2011 20:16:57 +0000 (14:16 -0600)]
mesa: consolidate code in _mesa_update_array_max_element()
Ian Romanick [Wed, 6 Apr 2011 20:46:58 +0000 (13:46 -0700)]
mesa: Add link to 7.10.2 release notes
Ian Romanick [Wed, 6 Apr 2011 20:41:43 +0000 (13:41 -0700)]
docs: update news.html with 7.10.2 release
(cherry picked from commit
b0866f6cfdfcaf23686b1ae2a99627d468196db2)
Ian Romanick [Wed, 6 Apr 2011 20:46:20 +0000 (13:46 -0700)]
Import 7.10.2 release notes from 7.10 branch
Julian Adams [Wed, 6 Apr 2011 19:04:08 +0000 (21:04 +0200)]
r600g: Fix non-independent blend state.
This fixes piglit fbo/fbo-drawbuffers-blend-add.
Signed-off-by: Henri Verbeet <hverbeet@gmail.com>
Brian Paul [Wed, 6 Apr 2011 19:00:14 +0000 (13:00 -0600)]
mesa: init/release the default texture buffer object
Brian Paul [Wed, 6 Apr 2011 18:45:21 +0000 (12:45 -0600)]
mesa: fix popping of texture state for GL_TEXTURE_BUFFER
Fixes http://bugs.freedesktop.org/show_bug.cgi?id=36032
Brian Paul [Wed, 6 Apr 2011 18:33:28 +0000 (12:33 -0600)]
mesa: TEXTURE_BUFFER fix-up
Fixes http://bugs.freedesktop.org/show_bug.cgi?id=36033
Tom Stellard [Wed, 6 Apr 2011 06:49:23 +0000 (23:49 -0700)]
r300/compiler: Don't try to convert RGB to Alpha in full instructions
Note: This is a candidate for the 7.10 branch.
José Fonseca [Wed, 6 Apr 2011 07:25:22 +0000 (08:25 +0100)]
mesa: Recognize MESA_VERBOSE=all as shortcut for all debug options.
José Fonseca [Tue, 5 Apr 2011 20:35:11 +0000 (21:35 +0100)]
draw: Fix thinko in debug warnings.
José Fonseca [Tue, 5 Apr 2011 19:13:44 +0000 (20:13 +0100)]
trace: Correct/cleanup.
Observe the actual type hierarchy and polymorphism of sampler views and
surface state.
s/texture/resource/
etc.
José Fonseca [Mon, 4 Apr 2011 10:48:07 +0000 (11:48 +0100)]
Revert "trace: Dump blend state as a binary blob."
This reverts commit
dcdf94c87c087186f963b6e201b2989c61a76ef0.
No retrace support anymore, so human-legible representation of blend
state is much more useful than a blob.
José Fonseca [Mon, 4 Apr 2011 10:41:30 +0000 (11:41 +0100)]
st/python: Remove bindings, and all its dependencies.
Because:
- bindings are not fully automatic, and they are broken most of the time
- unit tests/samples can be written in C on top of graw
- tracing/retracing is more useful at API levels with stable ABIs such as
GL, producing traces that cover more layers of the driver stack and and
can be used for regression testing
Brian Paul [Tue, 5 Apr 2011 23:52:47 +0000 (17:52 -0600)]
swrast: simplify assertion to silence warning
Brian Paul [Wed, 6 Apr 2011 01:02:07 +0000 (19:02 -0600)]
mesa: core support for GL_ARB_texture_buffer_object
No GLSL or driver support yet.
Brian Paul [Wed, 6 Apr 2011 00:50:39 +0000 (18:50 -0600)]
mesa: regenerated API entrypoints for GL_ARB_texture_buffer_object
Brian Paul [Wed, 6 Apr 2011 00:48:26 +0000 (18:48 -0600)]
mesa: API spec for GL_ARB_texture_buffer_object
Vinson Lee [Tue, 5 Apr 2011 22:32:39 +0000 (15:32 -0700)]
scons: Generate empty git_sha1.h for all platforms.
Fixes MinGW SCons build.
Kenneth Graunke [Tue, 5 Apr 2011 20:22:30 +0000 (13:22 -0700)]
ir_to_mesa: Use gl_register_file enum type rather than 'int'.
src_reg already used this; make dst_reg use it too.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Kenneth Graunke [Tue, 5 Apr 2011 20:19:50 +0000 (13:19 -0700)]
ir_to_mesa: Unprefix ir_to_mesa_undef* and ir_to_mesa_address_reg.
Rename ir_to_mesa_undef to undef_src, for clarity.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Kenneth Graunke [Tue, 5 Apr 2011 19:22:42 +0000 (12:22 -0700)]
ir_to_mesa: Use emit overloads to avoid passing undef registers.
Makes the code just a little bit cleaner.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Kenneth Graunke [Tue, 5 Apr 2011 19:16:57 +0000 (12:16 -0700)]
ir_to_mesa: Rename ir_to_mesa_emit_*_opX methods to emit_*.
There's really no need for a prefix on member functions, and overloading
takes care of the _op1/_op2 distinction quite nicely. Eric already made
a similar change in the i965 FS backend.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Kenneth Graunke [Tue, 5 Apr 2011 19:01:59 +0000 (12:01 -0700)]
ir_to_mesa: Use constructors to convert between src_reg and dst_reg.
Rather than ir_to_mesa_dst_reg_from_src and ir_to_mesa_src_reg_from_dst.
The new constructors are marked 'explicit' so that the compiler can
catch cases where source and destination registers were accidentally
interchanged.
This also necessitated using constructors to initialize the undef and
address registers, as well as adding a default constructor.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Kenneth Graunke [Tue, 5 Apr 2011 18:43:21 +0000 (11:43 -0700)]
ir_to_mesa: Remove the "ir_to_mesa_" prefix on src_reg/dst_reg types.
Both classes are completely private to ir_to_mesa.cpp, so there won't be
any name conflicts with other parts of Mesa. The prefix simply makes it
harder to read.
Also, use a class rather than typedef structs.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Kenneth Graunke [Tue, 5 Apr 2011 18:30:36 +0000 (11:30 -0700)]
ir_to_mesa: Rename src_reg and dst_reg variables to src and dst.
This is in preparation from removing the "ir_to_mesa_" prefix on the
src_reg and dst_reg types, which would cause a naming conflict.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Brian Paul [Tue, 5 Apr 2011 20:07:41 +0000 (14:07 -0600)]
scons: generate empty git_sha1.h file for now
My feeble attempt to invoke the extract_git_sha1 script from
SConscript didn't work. Hopefully this will do for now.
Ian Romanick [Thu, 31 Mar 2011 18:42:01 +0000 (11:42 -0700)]
mesa: Include GIT SHA1 in GL version string
Reviewed-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Ian Romanick [Tue, 1 Mar 2011 22:10:49 +0000 (14:10 -0800)]
dri: Remove driver date from renderer string
Reviewed-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Ian Romanick [Mon, 4 Apr 2011 20:35:26 +0000 (13:35 -0700)]
ir_to_mesa: Handle shadow compare w/projection and LOD bias correctly
The code would previously handle the projection, then swizzle the
shadow comparitor into place. However, when the projection is done
"by hand," as in the TXB case, the unprojected shadow comparitor would
over-write the projected shadow comparitor.
Shadow comparison with projection and LOD is an extremely rare case in
real application code, so it shouldn't matter that we don't handle
that case with the greatest efficiency.
NOTE: This is a candidate for the stable branches.
Reviewed-by: Brian Paul <brianp@vmware.com>
References: https://bugs.freedesktop.org/show_bug.cgi?id=32395
Marek Olšák [Tue, 5 Apr 2011 15:57:02 +0000 (17:57 +0200)]
r300g: postpone fragment shader state validation until draw_vbo
Tom Stellard [Tue, 5 Apr 2011 07:04:45 +0000 (00:04 -0700)]
r300/compiler: Fix vertex shader MAD instructions with constant swizzles
NOTE: This is a candidate for the 7.9 and 7.10 branches.
Fabian Bieler [Fri, 25 Mar 2011 10:06:01 +0000 (11:06 +0100)]
st/mesa: Apply LOD from texture object
Signed-off-by: Brian Paul <brianp@vmware.com>
Fabian Bieler [Fri, 25 Mar 2011 10:04:41 +0000 (11:04 +0100)]
st/mesa: Apply LOD bias from correct texture unit
Signed-off-by: Brian Paul <brianp@vmware.com>
Brian Paul [Tue, 5 Apr 2011 13:53:05 +0000 (07:53 -0600)]
mesa: fix alpha value for texstore_rgbx8888
Silences constant overflow compiler warning.
Fabian Bieler [Tue, 5 Apr 2011 13:51:01 +0000 (07:51 -0600)]
mesa: Guard against null pointer deref in fbo validation
This matches the behaviour below when numSamples is compared.
At least with the gallium state tracker this can actually occur if st_render_texture fails.
Signed-off-by: Brian Paul <brianp@vmware.com>
Brian Paul [Tue, 5 Apr 2011 13:51:01 +0000 (07:51 -0600)]
mesa: added _mesa_get_attachment_teximage() helpers
Brian Paul [Tue, 5 Apr 2011 13:51:01 +0000 (07:51 -0600)]
mesa: 80-column wrapping
Dave Airlie [Tue, 5 Apr 2011 06:54:08 +0000 (16:54 +1000)]
r300c: fix build since last r300g commit
Marek Olšák [Tue, 5 Apr 2011 04:21:26 +0000 (06:21 +0200)]
r300g: fix RG/LATC1_SNORM by doing UNORM->SNORM conversion in the shader
Marek Olšák [Tue, 5 Apr 2011 04:18:18 +0000 (06:18 +0200)]
r300/compiler: implement the CND opcode
No one uses it now, but I will need it for a lowering pass.
Marek Olšák [Tue, 5 Apr 2011 00:44:03 +0000 (02:44 +0200)]
r300/compiler: set the MSB of ADDR for inline constants
The docs say so.
Brian Paul [Mon, 4 Apr 2011 23:42:25 +0000 (17:42 -0600)]
st/mesa: minor assorted clean-ups and fixes
Brian Paul [Mon, 4 Apr 2011 23:36:40 +0000 (17:36 -0600)]
st/mesa: use 'array' local var to simplify the code a bit
Brian Paul [Mon, 4 Apr 2011 23:24:09 +0000 (17:24 -0600)]
st/mesa: fix zero-sized user vertex buffer bug
Commit
4c4ab5668cd6df573db7b065f0493fb80ac70ab8 didn't properly
handle the stride==0 case.
Fixes https://bugs.freedesktop.org/show_bug.cgi?id=35961
Ian Romanick [Sun, 3 Apr 2011 01:56:33 +0000 (18:56 -0700)]
mesa: Add fall-back formats for unsupported snorm formats
This is always the way with real hardware and desktop OpenGL. Some
hardware can't do some formats natively. The alpha-only, luminance,
and intensity formats are usually the most problematic. Some sized
formats can also be problematic. This patch provides fall-back
formats for those that are not natively supported.
At some point it would be interesting to try providing
device-independent conversions using EXT_texture_swizzle. The drivers
that support EXT_texture_swizzle could, for example, see
GL_LUMINANCE16_SNORM as MESA_FORMAT_SIGNED_R16 with a { r, r, r, 1 }
swizzle. Care would need to be taken to prevent issues with using
those textures for FBO rendering.
This is the rest of the fix for glean's pixelFormats test on i965.
Reviewed-by: Marek Olšák <maraeo@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Ian Romanick [Sun, 3 Apr 2011 01:49:13 +0000 (18:49 -0700)]
i965: Add the missing supportable EXT_texture_snorm formats
This class of hardware can natively sample all of the snorm surface
formats that DX10 requires, but it can't do some of the legacy GL
formats. In particular, all of the alpha, luminance, and intensity
formats are unsupported.
This partially fixes the breakage in glean's pixelFormats test since
GL_EXT_texture_snorm support was added to Mesa.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Bryan Cain [Mon, 4 Apr 2011 21:07:16 +0000 (23:07 +0200)]
nv50: prevent NV_OP_SELECT from having flags_def defined
Brian Paul [Mon, 4 Apr 2011 20:37:16 +0000 (14:37 -0600)]
mesa: work around failed assertion for GL_RGBA16_SNORM accum buffer
We use this format to represent the accum buffer. No snorm texture
sampling or rendering takes place.
Fixes failed assertion with swrast and any app using the accum buffer
(and glxinfo).
Brian Paul [Mon, 4 Apr 2011 17:30:46 +0000 (11:30 -0600)]
docs: update prerequisites, remove old demo info
Brian Paul [Mon, 4 Apr 2011 17:16:51 +0000 (11:16 -0600)]
docs: update info about Mesa packaging/contents
Marek Olšák [Mon, 4 Apr 2011 16:55:08 +0000 (18:55 +0200)]
r300/compiler: apply the texture swizzle to shadow pass and fail values too
Piglit tests:
- glsl-fs-shadow2d-01
- glsl-fs-shadow2d-02
- glsl-fs-shadow2d-03
- fs-shadow2d-red-01
- fs-shadow2d-red-02
- fs-shadow2d-red-03
NOTE: This is a candidate for the stable branches.
Marek Olšák [Mon, 4 Apr 2011 16:35:16 +0000 (18:35 +0200)]
r300/compiler: propagate SaturateMode down to the result of shadow comparison
NOTE: This is a candidate for the stable branches.
Alex Deucher [Mon, 4 Apr 2011 15:15:18 +0000 (11:15 -0400)]
r600g: add some additional ontario pci ids
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Alex Deucher [Mon, 4 Apr 2011 15:13:44 +0000 (11:13 -0400)]
r600c: add new ontario pci ids
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Marek Olšák [Sun, 3 Apr 2011 19:25:40 +0000 (21:25 +0200)]
r300g: do not wait for a busy BO if neither GPU nor CPU is changing it
Improves frame rate in apps with at least one user vertex buffer and
a hw index buffer.
Marek Olšák [Sun, 3 Apr 2011 19:18:03 +0000 (21:18 +0200)]
r300g: remove unused RADEON_PB_USAGE_CACHE
Marek Olšák [Sun, 3 Apr 2011 17:26:54 +0000 (19:26 +0200)]
r300g: tell the GLSL compiler to lower the continue opcode
NOTE: This is a candidate for the stable branches.
Marek Olšák [Sun, 3 Apr 2011 17:17:19 +0000 (19:17 +0200)]
r300g: avoid mapping the same buffer twice
Shouldn't happen, but you never know.
Marek Olšák [Sun, 3 Apr 2011 16:14:49 +0000 (18:14 +0200)]
r300g: handle DISCARD_WHOLE_RESOURCE for buffers
Marek Olšák [Sun, 3 Apr 2011 15:32:47 +0000 (17:32 +0200)]
r300g: remove the redundant reference counter in radeon_bo
We already have pb_buffer::reference::count.
Tom Stellard [Sun, 3 Apr 2011 06:14:12 +0000 (23:14 -0700)]
Revert "r300/compiler: Remove obsolete compiler passes"
This reverts commit
9f013a8233197d4a0482661cb37cfeac1a61b804.
These passes are still need for non-GLSL paths like g3dvl and ARB
programs.
Kenneth Graunke [Sun, 3 Apr 2011 08:39:28 +0000 (01:39 -0700)]
i965/fs: Switch W and 1/W in Sandybridge interpolation setup.
Various documentation mentions that "W" is handed to the WM stage,
but further digging seems to indicate that they really mean 1/W.
The code here is still unclear, but changing this fixes piglit
test "fragcoord_w" on Sandybridge as well as a Khronos ES2 conformance
test. I also tested 3DMarkMobile ES2.0's taiji and hoverjet demos, as
well as Nexuiz, just to be safe.
NOTE: This is a candidate for the 7.10 branch.
Kenneth Graunke [Sun, 3 Apr 2011 07:57:30 +0000 (00:57 -0700)]
i965: Fix null register use in Sandybridge implied move resolution.
Fixes regressions caused by commit
9a21bc6401, namely GPU hangs when
running gnome-shell or compiz (Mesa bugs #35820 and #35853).
I incorrectly refactored the case that dealt with ARF_NULL; even in that
case, the source register needs to be changed to the MRF.
NOTE: This is a candidate for the 7.10 branch (if
9a21bc6401 is
cherry-picked, take this one too).
Eric Anholt [Tue, 29 Mar 2011 20:22:13 +0000 (13:22 -0700)]
i965: Fix the VS thread limits for GT1, and clarify the WM limits on both.
José Fonseca [Fri, 1 Apr 2011 19:13:40 +0000 (20:13 +0100)]
tests: Use elts in translate_test.