Michal Krol [Tue, 1 Dec 2009 07:37:15 +0000 (08:37 +0100)]
draw: Update for renamed sampler/texture state setters.
Michal Krol [Tue, 1 Dec 2009 07:35:43 +0000 (08:35 +0100)]
cso: Add support for separate vertex sampler state.
Michal Krol [Tue, 1 Dec 2009 07:33:43 +0000 (08:33 +0100)]
gallium: Introduce separate vertex texture/sampler state.
Previously, gallium shared sampler and texture state between vertex
and fragment shader stages. This change generalises this concept by
providing separate entrypoints for vertex and fragment sampler state
setting.
A new capability bit is added to query the driver for the number
of samplers that can be utilised by a vertex and fragment shader
at the same time.
Michal Krol [Fri, 27 Nov 2009 11:33:17 +0000 (12:33 +0100)]
tgsi/sanity: Up MAX_REGISTERS to 1024.
Brian Paul [Mon, 30 Nov 2009 15:56:47 +0000 (08:56 -0700)]
st/mesa: handle front/back-face +1/-1 vs. 1/0 conversion
Fixes progs/glsl/twoside.c demo.
Richard Li [Sun, 29 Nov 2009 17:28:32 +0000 (12:28 -0500)]
r600 : clena up a bit for last commit.
Richard Li [Sun, 29 Nov 2009 17:12:19 +0000 (12:12 -0500)]
r600 : add read port allocation for uniform;
mapping ps input based on vs output;
fix bugs including constants updating for vs.
Christoph Bumiller [Sat, 28 Nov 2009 12:57:38 +0000 (13:57 +0100)]
nv50: update linkage on rasterizer change
We need to update VP_RESULT_MAP and/or COORD_REPLACE_MAP
when light_twoside and/or point_sprite are changed.
Christoph Bumiller [Sun, 29 Nov 2009 12:33:16 +0000 (13:33 +0100)]
nv50: match VP outputs to FP inputs ourselves
For each FP input, don't assume that the VP output will be
at the same position, but scan the semantics instead, then
put the correct output reg indices into VP_RESULT_MAP.
Position is still assumed to be the first output/input.
See
07fafc7c9346aa260829603bf3188596481e9e62, which renders
previous assumptions incorrect.
Christoph Bumiller [Thu, 26 Nov 2009 16:03:00 +0000 (17:03 +0100)]
nv50: don't permanently negate src in emit_ddy
Corbin Simpson [Sat, 28 Nov 2009 18:45:17 +0000 (10:45 -0800)]
util: Fix bad code.
Uf. How embarrassing.
Corbin Simpson [Sat, 28 Nov 2009 18:13:51 +0000 (10:13 -0800)]
util: Improve bitcount.
Sorry for not pushing this before, it got lost in stashes.
Christoph Bumiller [Fri, 27 Nov 2009 20:29:38 +0000 (21:29 +0100)]
nv50: do conversion of last insn to 64 bit format first
Simplifies things since the second to last one will then
be converted in the subsequent pass that ensures alignment
automatically.
Christoph Bumiller [Thu, 26 Nov 2009 15:59:39 +0000 (16:59 +0100)]
nv50: bswap32 the polygon stipple pattern
The hardware wants the pattern the same way it is
passed to glPolygonStipple.
Christoph Bumiller [Thu, 26 Nov 2009 15:58:59 +0000 (16:58 +0100)]
gallium/util: added util_bswap32()
Roland Scheidegger [Fri, 27 Nov 2009 16:41:42 +0000 (17:41 +0100)]
svga: fix for not using texture width/height/depth arrays
Roland Scheidegger [Fri, 27 Nov 2009 16:40:24 +0000 (17:40 +0100)]
r300g,llvmpipe: fix some more merge problems
Roland Scheidegger [Fri, 27 Nov 2009 16:15:47 +0000 (17:15 +0100)]
Merge branch 'width0'
Conflicts:
src/gallium/drivers/r300/r300_texture.c
src/gallium/state_trackers/xorg/xorg_exa.c
src/mesa/state_tracker/st_cb_texture.c
Roland Scheidegger [Thu, 26 Nov 2009 21:49:58 +0000 (22:49 +0100)]
gallium: fix more statetrackers/drivers for not using texture width/height/depth arrays
Michal Krol [Thu, 26 Nov 2009 19:38:43 +0000 (20:38 +0100)]
tgsi/exec: Fix orientation of DDY.
Michal Krol [Thu, 26 Nov 2009 19:30:04 +0000 (20:30 +0100)]
tgsi/exec: Force return from a subroutine at ENDSUB.
Keith Whitwell [Thu, 26 Nov 2009 18:24:57 +0000 (19:24 +0100)]
draw: Fix max_index check.
We want to fallback to draw splitting when vertex element indices
might be too high for atomic draw path (currently limited to 4095).
Michal Krol [Thu, 26 Nov 2009 18:21:55 +0000 (19:21 +0100)]
tgsi/ureg: Add forgotten goto in ureg_DECL_constant().
José Fonseca [Thu, 26 Nov 2009 11:16:19 +0000 (11:16 +0000)]
llvmpipe: Fake missing SSSE3 when simulation less capabable machines.
SSE3 != SSSE3 and so far we only use the later.
José Fonseca [Thu, 26 Nov 2009 11:15:08 +0000 (11:15 +0000)]
llvmpipe: Update/correct CPU requirements.
There are no hard requirements at the moment.
We don't really emit any sse3 yet. Just some ssse3.
Thanks to Roland for spotting these incorrections.
Roland Scheidegger [Thu, 26 Nov 2009 00:15:25 +0000 (01:15 +0100)]
Merge commit 'origin/st-shader-varients'
Conflicts:
src/mesa/state_tracker/st_atom_shader.c
src/mesa/state_tracker/st_program.c
José Fonseca [Wed, 25 Nov 2009 18:06:12 +0000 (18:06 +0000)]
scons: Make it work with MinGW build of LLVM 2.6.
LLVM 2.5 is no longer supported on windows.
Dave Airlie [Wed, 25 Nov 2009 05:45:31 +0000 (15:45 +1000)]
r600: add ARB_texture_non_power_of_two support.
This makes the miptree rounds up to the near POT for each level for
all radeons, however since mipmaps aren't support with NPOT on previous
radeons this calculation shouldn't cause any problems. If it does
we can just make it r600 only.
I tested a few mipmap demos on r500 and they all seem to work.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Richard Li [Tue, 24 Nov 2009 21:00:25 +0000 (16:00 -0500)]
r600 : reset stack flag with one channel only.
Richard Li [Tue, 24 Nov 2009 17:16:39 +0000 (12:16 -0500)]
r600 : fix stack depth setting bug.
José Fonseca [Tue, 24 Nov 2009 16:01:01 +0000 (16:01 +0000)]
llvmpipe: Update instructions.
José Fonseca [Sun, 15 Nov 2009 20:14:03 +0000 (12:14 -0800)]
svga: Handle comment tokens when dumping.
José Fonseca [Tue, 24 Nov 2009 14:43:30 +0000 (14:43 +0000)]
svga: Use consistent names for public symbol names of shader dumping facilities.
José Fonseca [Tue, 24 Nov 2009 14:37:45 +0000 (14:37 +0000)]
svga: Use consistent file names for dumping facilities.
José Fonseca [Tue, 24 Nov 2009 14:24:38 +0000 (14:24 +0000)]
util: Describe a few more formats.
José Fonseca [Mon, 23 Nov 2009 11:21:11 +0000 (11:21 +0000)]
llvmpipe: Use assert instead of abort. Only verify functions on debug builds.
José Fonseca [Sun, 15 Nov 2009 14:46:48 +0000 (06:46 -0800)]
llvmpipe: Fix memory leak.
José Fonseca [Wed, 11 Nov 2009 01:52:53 +0000 (17:52 -0800)]
llvmpipe: Use the generic conversion routine for depths.
This allows for z32f depth format to work correctly.
José Fonseca [Wed, 11 Nov 2009 01:51:06 +0000 (17:51 -0800)]
llvmpipe: Be more conservative with the supported formats.
We'll likely support much more formats, but doing this allows to run
more testsuites without immediately hit assertion failures.
José Fonseca [Wed, 11 Nov 2009 00:56:43 +0000 (16:56 -0800)]
wgl: Fix copy'n'paste typo in comment.
José Fonseca [Wed, 11 Nov 2009 00:55:44 +0000 (16:55 -0800)]
llvmpipe: Fix typo in comparison operator.
Michal Krol [Tue, 24 Nov 2009 10:58:01 +0000 (11:58 +0100)]
tgsi: Document Declaration Semantic token and FACE semantic name.
Michal Krol [Tue, 24 Nov 2009 10:22:03 +0000 (11:22 +0100)]
slang: Fix allocation size.
We don't need 16K+ to store a single pointer.
Michal Krol [Tue, 24 Nov 2009 10:17:16 +0000 (11:17 +0100)]
mesa: Fix pointer arithmetic.
Michal Krol [Tue, 24 Nov 2009 08:03:41 +0000 (09:03 +0100)]
tgsi: Implement predicated instructions in exec.
Michal Krol [Tue, 24 Nov 2009 08:02:29 +0000 (09:02 +0100)]
tgsi: Add ureg_DECL_loop().
Michal Krol [Tue, 24 Nov 2009 08:01:48 +0000 (09:01 +0100)]
tgsi: Account for gallium shader token representation changes.
Michal Krol [Tue, 24 Nov 2009 07:55:40 +0000 (08:55 +0100)]
gallium: Refactor the instruction predicate TGSI token.
Rename it to tgsi_instruction_predicate -- it's no longer an extended
token. Its presence is indicated by a new flag in tgsi_instruction that
indicates whether an instruction is predicated.
Also, change predicate index representation to match the other
tokens that specify register indices.
Dave Airlie [Tue, 24 Nov 2009 01:56:45 +0000 (11:56 +1000)]
r300: fix swtcl bo leak problem.
We can get a lot of swtcl bo allocations - need to probably abstract
this a bit further.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Tue, 24 Nov 2009 00:49:34 +0000 (10:49 +1000)]
radeon/r200/r300/r600: make bo mapping be explicit
This moves the bo mapping outside the DMA layer and makes it explicit,
this should in theory make it simpler to split the clean up the dma/cmdbuf
linkage that I created before that is broken.
Tested on: r600, rv380 (tcl/no-tcl), rv200 (tcl/no-tcl)
Signed-off-by: Dave Airlie <airlied@redhat.com>
Brian Paul [Tue, 24 Nov 2009 01:09:46 +0000 (18:09 -0700)]
mesa: use gcc __builtin_popcount()
Brian Paul [Tue, 24 Nov 2009 01:06:19 +0000 (18:06 -0700)]
r300g: use util_bitcount()
Brian Paul [Tue, 24 Nov 2009 01:04:47 +0000 (18:04 -0700)]
egl: use util_bitcount()
Brian Paul [Tue, 24 Nov 2009 01:04:22 +0000 (18:04 -0700)]
gallium/util: added util_bitcount()
Maciej Cencora [Mon, 23 Nov 2009 21:14:48 +0000 (22:14 +0100)]
Merge commit 'origin/mesa_7_7_branch'
Maciej Cencora [Mon, 23 Nov 2009 20:59:08 +0000 (21:59 +0100)]
radeon: fix errors in miptree related function
- typo
- memory leak
- off by one (spotted by airlied)
Maciej Cencora [Mon, 23 Nov 2009 20:28:22 +0000 (21:28 +0100)]
Merge branch 'mesa_7_7_branch' of anongit.freedesktop.org/git/mesa/mesa into mesa_7_7_branch
Jakob Bornecrantz [Mon, 23 Nov 2009 18:59:02 +0000 (19:59 +0100)]
svga: Scrub Makefiles a bit
Remove x86 specific hacks. Not that they will ever be used on
none x86 arches, but they are built by default. And the way the
flags where added was a hack.
Michal Krol [Mon, 23 Nov 2009 12:22:04 +0000 (13:22 +0100)]
slang: Check return value from emit_instruction().
Michal Krol [Mon, 23 Nov 2009 10:32:58 +0000 (11:32 +0100)]
tgsi: Clamp the source argument in micro_exp2() to avoid Inf.
Michal Krol [Mon, 23 Nov 2009 10:29:29 +0000 (11:29 +0100)]
tgsi: Remove code that actually had no effect.
Michal Krol [Mon, 23 Nov 2009 09:51:07 +0000 (10:51 +0100)]
softpipe: Initialise TGSI machine's Face.
Michal Krol [Mon, 23 Nov 2009 09:49:41 +0000 (10:49 +0100)]
tgsi: Fix POSITION and FACE fragment shader inputs.
Michal Krol [Mon, 23 Nov 2009 09:40:05 +0000 (10:40 +0100)]
tgsi: Bring BGNFOR/ENDFOR implementation up to spec.
Michal Krol [Thu, 19 Nov 2009 12:05:58 +0000 (13:05 +0100)]
tgsi: Fake TXD implementation in exec.
Michal Krol [Thu, 19 Nov 2009 12:01:08 +0000 (13:01 +0100)]
tgsi: Add execution debugging facilities to exec.
Michal Krol [Thu, 19 Nov 2009 11:52:58 +0000 (12:52 +0100)]
slang: Be more robust with memory in concat_shaders().
Vinson Lee [Mon, 23 Nov 2009 06:52:59 +0000 (01:52 -0500)]
mesa: Initialize variable in _mesa_get_texel_fetch_func.
Vinson Lee [Mon, 23 Nov 2009 06:30:32 +0000 (01:30 -0500)]
glx: Prevent potential null pointer deference in driCreateContext.
Vinson Lee [Mon, 23 Nov 2009 06:09:06 +0000 (01:09 -0500)]
glu/sgi: Fix memory leak in bitmapBuild2DMipmaps.
Vinson Lee [Mon, 23 Nov 2009 05:57:37 +0000 (00:57 -0500)]
glu/sgi: Fix memory leak in gluScaleImage3D.
Richard Li [Mon, 23 Nov 2009 02:31:46 +0000 (21:31 -0500)]
r600 : add support for shader instruction trunc and discard.
Dave Airlie [Sun, 22 Nov 2009 20:44:29 +0000 (06:44 +1000)]
r600: hopefully fix segfault.
Dave Airlie [Sun, 22 Nov 2009 20:31:29 +0000 (06:31 +1000)]
r600: fix inline issues
Richard Li [Sun, 22 Nov 2009 17:58:28 +0000 (12:58 -0500)]
r600 : add stack depth calculation, enable CF pop.
Richard Li [Fri, 20 Nov 2009 21:36:55 +0000 (16:36 -0500)]
r600 : use cf for all pop now, left optimization for future.
Maciej Cencora [Sun, 22 Nov 2009 14:08:46 +0000 (15:08 +0100)]
r300: fix VP source conflict resolution on 64-bit machines
On 32bit machines we were lucky because the sizeof(reg) == sizeof(rc_src_register).
On 64bit machines pointers are 8 bytes long, so we were overwriting other data.
Maciej Cencora [Sun, 22 Nov 2009 13:13:18 +0000 (14:13 +0100)]
r300: fix SIN/COS/SCS instructions for R300 fp
Vinson Lee [Sun, 22 Nov 2009 06:57:35 +0000 (01:57 -0500)]
glu/sgi: Fix memory leak in gluScaleImage.
Vinson Lee [Sun, 22 Nov 2009 06:45:53 +0000 (01:45 -0500)]
glu: Fix memory leak in __gl_meshMakeEdge.
Vinson Lee [Sun, 22 Nov 2009 06:26:32 +0000 (01:26 -0500)]
svga: Prevent potential null pointer deference in vmw_surface.c.
Vinson Lee [Sun, 22 Nov 2009 06:20:07 +0000 (01:20 -0500)]
svga: Fix memory leak in vmw_screen_ioctl.c
Vinson Lee [Sun, 22 Nov 2009 06:10:38 +0000 (01:10 -0500)]
st/xorg: Prevent potential null pointer deference in xorg_xv.c.
Vinson Lee [Sun, 22 Nov 2009 05:55:52 +0000 (00:55 -0500)]
st/xorg: Reorder cases in switch statement.
Silences missing break statement warning.
Corbin Simpson [Sun, 22 Nov 2009 03:13:26 +0000 (19:13 -0800)]
r300g: Build fix.
Oops.
Marek Olšák [Fri, 20 Nov 2009 03:52:49 +0000 (04:52 +0100)]
r300g: do not reallocate r300_vertex_info and r300_rs_block all the time
Marek Olšák [Sat, 21 Nov 2009 04:51:13 +0000 (05:51 +0100)]
r300g: clean up vs/fs tabs
Instead of vs_tab, we use vs_output_tab and it's local now. fs_tab hasn't
been used anywhere, so I removed it and r300_update_fs_tab too.
Maciej Cencora [Sat, 21 Nov 2009 14:56:23 +0000 (15:56 +0100)]
radeon: fix compressed mipmapped textures
Tested on r300 only, other cards may require adjusting texture_compressed_row_align.
Maciej Cencora [Sat, 21 Nov 2009 14:56:02 +0000 (15:56 +0100)]
radeon: fix glCompressedTexSubImage
Maciej Cencora [Sat, 21 Nov 2009 14:56:23 +0000 (15:56 +0100)]
radeon: fix compressed mipmapped textures
Tested on r300 only, other cards may require adjusting texture_compressed_row_align.
Maciej Cencora [Sat, 21 Nov 2009 14:56:02 +0000 (15:56 +0100)]
radeon: fix glCompressedTexSubImage
Roland Scheidegger [Sat, 21 Nov 2009 12:39:30 +0000 (04:39 -0800)]
intel: make CopyTex[Sub]Image fallback debug messages more consistent
Roland Scheidegger [Sat, 21 Nov 2009 12:22:50 +0000 (04:22 -0800)]
mesa: handle different RowStride in _mesa_get_compressed_teximage
drivers storing compressed textures with non-native stride but relying on
_mesa_get_compressed_teximage for GetCompressedTexImage otherwise won't
work correctly (for instance i965 compressed mipmaps).
Eric Anholt [Sat, 21 Nov 2009 12:33:21 +0000 (13:33 +0100)]
i965: Fix several memory leaks on exit.
Bug #25194.
Marek Olšák [Thu, 19 Nov 2009 21:40:11 +0000 (22:40 +0100)]
r300g: silence warnings
Marek Olšák [Thu, 19 Nov 2009 19:41:19 +0000 (20:41 +0100)]
r300g: fix typo in r300_reg.h to prevent the RS unit from doing random things
And reorder fragment shader inputs so that the colors are before texcoords,
as is allocated by the shader compiler. This commit makes VS->FS attribute
routing work on R500.
Marek Olšák [Sun, 15 Nov 2009 15:41:25 +0000 (16:41 +0100)]
r300g: fix updating a vertex format
We must update PSC when we change the vertex format, e.g. vertex colors
from RGBA to BGRA.
Marek Olšák [Sat, 14 Nov 2009 22:27:20 +0000 (23:27 +0100)]
r300g: add texture lod clamping
These now work:
piglit/lodclamp
piglit/levelclamp
Corbin Simpson [Sun, 15 Nov 2009 04:25:15 +0000 (05:25 +0100)]
r300g: Begin separating HW TCL and SW TCL state and setup.
This patch removes draw_context entirely from the HW TCL path and cleans up
a few other things along the way. Hopefully, nothing got broken.
Thanks to Marek Olšák for testing, review, and pointing out my bugs. :3
Vinson Lee [Sat, 21 Nov 2009 00:33:25 +0000 (16:33 -0800)]
st/egl: Fix memory leak in egl_tracker.c.