Brian Paul [Wed, 14 Oct 2009 21:11:12 +0000 (15:11 -0600)]
mesa: remove unused ctx->Driver.PrioritizeTextures() hook
Brian Paul [Wed, 14 Oct 2009 21:00:50 +0000 (15:00 -0600)]
mesa: remove unused ctx->Driver.TextureMatrix() hook
Jakob Bornecrantz [Tue, 13 Oct 2009 09:56:35 +0000 (11:56 +0200)]
st/xorg: Fix dpms include problems
Jakob Bornecrantz [Thu, 15 Oct 2009 00:24:53 +0000 (01:24 +0100)]
autoconf: Fix case for not having package
Zack Rusin [Sun, 11 Oct 2009 05:07:26 +0000 (01:07 -0400)]
st/xorg: get transparency on fills working (fixes Qt/KDE apps)
Jakob Bornecrantz [Tue, 13 Oct 2009 14:32:04 +0000 (15:32 +0100)]
i915g: Fix warnings
Alex Deucher [Wed, 14 Oct 2009 18:18:36 +0000 (14:18 -0400)]
r600: enable EXT_vertex_array_bgra extensions
José Fonseca [Wed, 14 Oct 2009 16:27:06 +0000 (17:27 +0100)]
util: Fix cpu detection on Windows. Cleanup.
José Fonseca [Wed, 14 Oct 2009 16:11:30 +0000 (17:11 +0100)]
llvmpipe: Use ALIGN_STACK.
José Fonseca [Wed, 14 Oct 2009 16:11:08 +0000 (17:11 +0100)]
gallium: New ALIGN_STACK macro to tell gcc to align stack pointer.
José Fonseca [Wed, 14 Oct 2009 15:16:40 +0000 (16:16 +0100)]
scons: Disable SSE intrinsics on MinGW.
-mstackrealign causes stack corruption on MinGW. And without it the ability
to use SSE instrinsics goes down the drain. Even if we use
__attribute__((force_align_arg_pointer)) for the functions we explicitly
use SSE instrinsics, the SSE code automatically generated by gcc will
cause assertion failures. What a nightmare.
Thankfully LLVM gets this right, so all runtime generated SSE code just
works. rtasm code doesn't assume 16byte alignment. Therefore the bulk of
our performance sensitive code is not affected by this.
Still, intrinsics can be convenient, and it would be nice
to get this working again some day, sp will try to get a reduced test
case.
Brian Paul [Wed, 14 Oct 2009 15:40:05 +0000 (09:40 -0600)]
Merge branch 'mesa_7_6_branch'
Brian Paul [Wed, 14 Oct 2009 15:37:40 +0000 (09:37 -0600)]
progs/Makefile: remove extra quoting
This reverts part of commit
115edf24a9128b79dfa5f30482c990e2cb898357 and
fixes the error: "/bin/sh: line 0: test: too many arguments".
Brian Paul [Wed, 14 Oct 2009 15:22:30 +0000 (09:22 -0600)]
Merge branch 'mesa_7_6_branch'
Brian Paul [Wed, 14 Oct 2009 13:52:26 +0000 (07:52 -0600)]
mesa: remove left-over debug printf
Vinson Lee [Wed, 14 Oct 2009 13:43:18 +0000 (07:43 -0600)]
prog/tests: Fix MSVC build.
Corbin Simpson [Wed, 14 Oct 2009 10:09:41 +0000 (03:09 -0700)]
r300g: Move ztop to derived state.
Need to get it into its own atom instead of piggybacking on DSA.
Corbin Simpson [Wed, 14 Oct 2009 08:58:18 +0000 (01:58 -0700)]
r300g: Compiler warning cleanup.
Dave Airlie [Wed, 14 Oct 2009 08:24:34 +0000 (18:24 +1000)]
r300g: remove buffer add that should be unnecessary.
This should be handled in the emit fine
Dave Airlie [Wed, 14 Oct 2009 08:24:05 +0000 (18:24 +1000)]
r300g: add QUERY to KITCHEN_SINK
I missed this, thanks to Corbin for pointing it out.
Dave Airlie [Wed, 14 Oct 2009 08:05:14 +0000 (18:05 +1000)]
r300g: add one more ZTOP disable bit.
Still missing the frag uses kill support, hopefully nha can point that out.
Dave Airlie [Wed, 14 Oct 2009 07:44:19 +0000 (17:44 +1000)]
r300g: port over last parts of oq support.
Add support for begin/end in each CS so we don't get any other
processes rendering in between.
TODO:
blame other parts of driver for this not working like Z.
Dave Airlie [Wed, 14 Oct 2009 07:14:43 +0000 (17:14 +1000)]
r300g: convert query to a state for emitting.
This means we don't emit in the begin query but when we have
to flush. Similiar to classic.
TODO:
make query object actually work.
Dave Airlie [Wed, 14 Oct 2009 06:53:12 +0000 (16:53 +1000)]
r300g: fixup arb occulsion query support.
1: add rv530 support
- num z pipes cap
- add proper start/finish query options for rv530
2: convert to use linked list properly.
3: add flushing required check.
4: initial Z top disabling support.
TODO:
make it actually work on my rv530.
Dave Airlie [Wed, 14 Oct 2009 05:13:25 +0000 (15:13 +1000)]
r300g: attempt to make bo space check sane.
This attempts to make r300g do proper bo space checking as opposed
to whatever it was doing now.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Wed, 14 Oct 2009 01:42:05 +0000 (11:42 +1000)]
r300g: fix case where texture unit 0 is disabled but unit 1 is enabled.
to reproduce, start texrect, disable 0 texture in menu.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Brian Paul [Tue, 13 Oct 2009 22:33:17 +0000 (16:33 -0600)]
mesa: minor tweak to printf string
Brian Paul [Tue, 13 Oct 2009 22:32:15 +0000 (16:32 -0600)]
mesa: rework _mesa_read_shader() debug hook
Look for shaders named "newshader_<CHECKSUM>" to replace the incoming
shader text. For debug purposes.
Brian Paul [Tue, 13 Oct 2009 22:18:06 +0000 (16:18 -0600)]
mesa: don't print pointer in _mesa_fprint_parameter_list()
Tom Fogal [Tue, 13 Oct 2009 17:55:34 +0000 (10:55 -0700)]
Fix build when PROGRAM_DIRS is empty.
SUBDIRS just takes PROGRAM_DIRS value. If PROGRAM_DIRS gets set
to the empty string (as can happen when building only OSMesa), a
'for' loop will lack anything to iterate over, causing a parse
error.
This fixes the issue by making sure SUBDIRS is the null string
when PROGRAM_DIRS is, and wrapping the for loops in if's, causing
them only to execute if there are directories to iterate over.
Tom Fogal [Tue, 13 Oct 2009 17:55:33 +0000 (10:55 -0700)]
ac: Fix AIX shared library builds.
AIX uses ".a" for both static and shared library extensions.
Ian Romanick [Fri, 9 Oct 2009 00:28:02 +0000 (17:28 -0700)]
Store clipping distance for user clip planes as part of vertex processing
Once the clipping distance is calculated and stored per vertex, the
distances can be re-used when clipping is actually performed. This
doesn't have any immediate benefit, but it paves the way for
implementing gl_ClipDistance in vertex shaders and result.clip[] in
vertex programs.
This has not produces any oglconform regressions on my G31 system
which uses software TNL.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Patrice Mandin [Tue, 13 Oct 2009 20:53:32 +0000 (22:53 +0200)]
nouveau: nv30: use texture width,height for render target dimensions
Chia-I Wu [Tue, 29 Sep 2009 08:11:06 +0000 (16:11 +0800)]
egl: Improve logging facility.
Add _eglSetLogger and _eglSetLogLevel to allow drivers to change the
message logger or report level.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Mon, 17 Aug 2009 07:53:54 +0000 (15:53 +0800)]
egl: Allow binding to any client API.
As a result, EGL_NONE is no longer a valid client API. And it is
possible that no config supports the current bound API.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Mon, 28 Sep 2009 09:39:07 +0000 (17:39 +0800)]
egl: Preload a driver if eglGetProcAddress is called early.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Sat, 15 Aug 2009 14:44:46 +0000 (22:44 +0800)]
egl: Remove core functions from eglGetProcAddress.
eglGetProcAddress may not be used to query core (non-extension)
functions.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Chia-I Wu [Mon, 28 Sep 2009 09:25:48 +0000 (17:25 +0800)]
egl: Add support for driver built-in.
This allows an EGL driver to be compiled together with libEGL.so. It
eliminates the need to specify a driver, or support module loading on
new platforms.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Zack Rusin [Sat, 10 Oct 2009 09:24:32 +0000 (05:24 -0400)]
st/xorg: solid fills with masks are supported
gradients are supported, but not enabled by default due to little
testing they got
Brian Paul [Tue, 13 Oct 2009 16:37:39 +0000 (10:37 -0600)]
intel: fix broken sw generate mipmap path
Need to restore code that fixed up the intel_texture_image state.
Zack Rusin [Sat, 10 Oct 2009 08:45:41 +0000 (04:45 -0400)]
st/xorg: fix and enable by default xrender acceleration
src in mask was broken
Brian Paul [Tue, 13 Oct 2009 15:04:54 +0000 (09:04 -0600)]
mesa: whitespace fixes
Brian Paul [Tue, 13 Oct 2009 15:04:14 +0000 (09:04 -0600)]
mesa: do RTT check in _mesa_meta_check_generate_mipmap_fallback()
We need to check that we can actually render to the texture's format
before doing mipmap generation.
This may fix bug 24219.
Corbin Simpson [Tue, 13 Oct 2009 04:26:46 +0000 (21:26 -0700)]
r300g: More debug flags.
Corbin Simpson [Tue, 13 Oct 2009 03:55:57 +0000 (20:55 -0700)]
r300g: Surface debug.
It gets really annoying watching r300g tell me how it's filling surfaces.
Or falling back during filling surfaces.
Corbin Simpson [Tue, 13 Oct 2009 03:47:00 +0000 (20:47 -0700)]
r300g: Fallback on surfaces we can't render to or from.
Still not sure why st keeps handing down things we can't render to.
Corbin Simpson [Tue, 6 Oct 2009 23:00:27 +0000 (16:00 -0700)]
r300g: Prevent multiple-use textures from getting incorrectly approved.
Corbin Simpson [Tue, 6 Oct 2009 22:30:39 +0000 (15:30 -0700)]
r300g: Clean up texture formats.
Brian Paul [Tue, 13 Oct 2009 00:11:31 +0000 (18:11 -0600)]
mesa: save/set/restore texture base/wrap state in blitframebuffer_texture()
Brian Paul [Tue, 13 Oct 2009 00:09:32 +0000 (18:09 -0600)]
mesa: minor clean up in check_begin_texture_render()
Brian Paul [Tue, 13 Oct 2009 00:08:42 +0000 (18:08 -0600)]
intel: whitespace/formatting clean-up
Brian Paul [Mon, 12 Oct 2009 23:41:47 +0000 (17:41 -0600)]
intel: replace extern decl with #include
Brian Paul [Mon, 12 Oct 2009 23:41:14 +0000 (17:41 -0600)]
intel: added INLINE keyword to minify()
This is mainly just to silence some warnings.
Brian Paul [Mon, 12 Oct 2009 23:40:33 +0000 (17:40 -0600)]
intel: code clean-ups
Brian Paul [Mon, 12 Oct 2009 23:37:06 +0000 (17:37 -0600)]
intel: pass zslice to intel_miptree_image_offset()
This lets us get rid of intel_miptree_depth_offsets() and simplify all
of the calling code.
Brian Paul [Mon, 12 Oct 2009 23:32:16 +0000 (17:32 -0600)]
i965: remove unused var
Vinson Lee [Tue, 13 Oct 2009 00:05:05 +0000 (18:05 -0600)]
prog/tests: Fix MSVC build.
Brian Paul [Fri, 9 Oct 2009 14:29:28 +0000 (08:29 -0600)]
mesa: print program Id when printing
Patrice Mandin [Mon, 12 Oct 2009 19:03:26 +0000 (21:03 +0200)]
Add support for more 8 and 16 bits formats
Andre Maasikas [Mon, 12 Oct 2009 11:57:45 +0000 (14:57 +0300)]
r600: implement ProgramStringNotify
need this to properly test with piglit/glean vert/fragprog tests
copied mostly from r300, many thanks to osiris, nha, airlied, others...
Andre Maasikas [Mon, 12 Oct 2009 09:58:40 +0000 (12:58 +0300)]
r600: LIT dst.y gets value from src.x
seems I overlooked this when removing hardcoded swizzles for this
one previously
Andre Maasikas [Mon, 12 Oct 2009 09:20:26 +0000 (12:20 +0300)]
r600: DPH adds w comp of second operand, so set first one to 1 instead
Zack Rusin [Mon, 12 Oct 2009 01:52:10 +0000 (21:52 -0400)]
st/xorg: initialize pipe in the renderer
Zack Rusin [Fri, 9 Oct 2009 13:52:17 +0000 (09:52 -0400)]
st/xorg: lots of rendering and xv changes
extract all the rendering code to xorg_rendedrer, make both exa and xv
share that code. in the process cleanup the rendering code and implement
a lot more of the xv infrastructure.
Zack Rusin [Tue, 6 Oct 2009 16:38:47 +0000 (12:38 -0400)]
configs: fix some remains of the i915simple driver
Zack Rusin [Thu, 1 Oct 2009 01:22:48 +0000 (21:22 -0400)]
st/xorg: implement basics of xv
Eric Anholt [Sun, 11 Oct 2009 18:16:03 +0000 (11:16 -0700)]
i965: Fix the last valid address setting for the index buffer.
Again, last valid address, not first invalid address. Fixes regression
in
255e5be265133280293bbfd8b2f9b74b2dec50bb that the kernel now catches
and caused piglit draw_elements_base_vertex to fail.
Eric Anholt [Sun, 11 Oct 2009 18:04:09 +0000 (11:04 -0700)]
i965: Fix the bounds emitted in the vertex buffer packets.
It's the address of the last valid byte, not the address of the first
invalid byte.
This should also fix problems with rendering with the new sanity checks in
the kernel.
Nicolai Hähnle [Sun, 11 Oct 2009 10:40:07 +0000 (12:40 +0200)]
r300g: Fix fragment program constants upload on R300
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
Dave Airlie [Sun, 11 Oct 2009 09:12:24 +0000 (19:12 +1000)]
r300g: fix blending default state + alpha separate.
this makes the default state same as r300
Vinson Lee [Sat, 10 Oct 2009 16:45:16 +0000 (10:45 -0600)]
progs/perf: Fix MSVC build.
Brian Paul [Sat, 10 Oct 2009 15:25:34 +0000 (09:25 -0600)]
Merge branch 'mesa_7_6_branch'
Brian Paul [Sat, 10 Oct 2009 15:12:00 +0000 (09:12 -0600)]
softpipe: fix multi-drawbuffers regression
This is part of the fix for bug 24401.
Brian Paul [Sat, 10 Oct 2009 15:18:14 +0000 (09:18 -0600)]
softpipe: revert
564df9dc5f6335eb8dc68f3c69cf054d2142663c
This change silenced valgrind warnings but broke progs/tests/drawbuffers.
The problem is we don't know the surface's state when we start caching it
(it may or may not be initialized/cleared/etc). So "clearing" it here was
presumptuous. Leaving the code in place (but disabled) for reference and
when using valgrind.
Fixes bug 24401
Cooper Yuan [Sat, 10 Oct 2009 06:41:44 +0000 (14:41 +0800)]
r300g: add video surface create and destroy functions
Alex Deucher [Fri, 9 Oct 2009 19:44:32 +0000 (15:44 -0400)]
radeon: fix scissor regression
fixes fdo bug 24248
Brian Paul [Fri, 9 Oct 2009 19:23:07 +0000 (13:23 -0600)]
docs: fix aux buffer bug 24426
Brian Paul [Fri, 9 Oct 2009 19:20:28 +0000 (13:20 -0600)]
st/mesa: create aux buffers according to visual
Fixes bug 24426 for gallium.
Brian Paul [Fri, 9 Oct 2009 19:22:00 +0000 (13:22 -0600)]
mesa: fix incorrect assertion in _mesa_add_aux_renderbuffers()
Fixes bug 24426.
Brian Paul [Thu, 8 Oct 2009 15:17:13 +0000 (09:17 -0600)]
mesa: regenerated gl_mangle.h file
Jakob Bornecrantz [Thu, 8 Oct 2009 12:43:22 +0000 (14:43 +0200)]
st/xorg: Old X servers don't deal well with failing accesses
Jakob Bornecrantz [Thu, 8 Oct 2009 12:40:19 +0000 (14:40 +0200)]
st/xorg: More fallback debugging
Change the fallback debugging around a bit and do the old commit correctly
Jakob Bornecrantz [Thu, 8 Oct 2009 12:41:06 +0000 (14:41 +0200)]
Revert "st/xorg: Debug fallbacks for composite as well"
This commit included a change that should have been in
its own commit, and turns out that you can do what was
suposed to go in it in much better way as well.
This reverts commit
5080e8bea6ae5cdb116023a5e2d8dbbb762bd69d.
Jakob Bornecrantz [Thu, 8 Oct 2009 11:38:34 +0000 (13:38 +0200)]
st/xorg: Debug fallbacks for composite as well
Jakob Bornecrantz [Thu, 8 Oct 2009 11:13:36 +0000 (13:13 +0200)]
st/xorg: Use A8 textures for depth 8 pixmaps
There is no hardware out there that can render to I8 textures.
Jakob Bornecrantz [Thu, 8 Oct 2009 11:05:55 +0000 (13:05 +0200)]
st/xorg: Fix depth stencil buffers on old X servers
Sanity checking is for the weak.
Vinson Lee [Fri, 9 Oct 2009 13:24:04 +0000 (07:24 -0600)]
progs/tests: Add tests to Makefile.
Brian Paul [Thu, 8 Oct 2009 19:00:37 +0000 (13:00 -0600)]
softpipe: restore/fix print_vertex() debug helper
Jakob Bornecrantz [Fri, 9 Oct 2009 12:57:23 +0000 (13:57 +0100)]
progs/perf: Ignores!
José Fonseca [Fri, 9 Oct 2009 12:22:42 +0000 (13:22 +0100)]
util: Force ESI register for cpuid's ebx result.
Fixes a segfault and better code. Unfortunately using an arbitrary
register ("=r") causes the gcc to abort when the code is optimized saying
it can't satisfy the constraint. Setting seems to do the trick.
José Fonseca [Fri, 9 Oct 2009 10:29:33 +0000 (11:29 +0100)]
llvmpipe: Eliminate constant mapping/unmapping.
José Fonseca [Tue, 6 Oct 2009 15:59:07 +0000 (16:59 +0100)]
progs: Remove unused empty file.
José Fonseca [Mon, 5 Oct 2009 17:32:36 +0000 (18:32 +0100)]
gdi: Fix the build when llvmpipe is requested (the default) but llvm is not present.
Andre Maasikas [Fri, 9 Oct 2009 07:46:12 +0000 (10:46 +0300)]
r600: fixup KIL instruction a bit
- KILLGT takes 2 arguments
- arb KIL has no dst register
- add TODO about clause ending but currently piglit fp-kil passes and
does not hang the card
Brian Paul [Fri, 9 Oct 2009 02:49:32 +0000 (20:49 -0600)]
docs: bump libdrm version requirement
Probably need to bump other version requirements too...
Eric Anholt [Fri, 9 Oct 2009 00:18:12 +0000 (17:18 -0700)]
mesa: Bump the required libdrm version to reflect reality.
Not all drivers require the latest version, but distros are good at
tracking it since we've got a good record as far as not regressing, and
it sure beats having a weird compile failure in someone else's driver
later on.
Eric Anholt [Fri, 2 Oct 2009 01:16:52 +0000 (18:16 -0700)]
i965: Use bo_references for the state cache delete function.
This appears to shave about 3% off the CPU usage in cairo-gl for firefox.
Eric Anholt [Fri, 2 Oct 2009 00:53:12 +0000 (17:53 -0700)]
intel: Use new drm_intel_bo_references() to avoid flushing.
Brian Paul [Thu, 8 Oct 2009 18:50:42 +0000 (12:50 -0600)]
mesa: free display list state after freeing shared state
Fixes bug 24402.
Brian Paul [Thu, 8 Oct 2009 15:28:07 +0000 (09:28 -0600)]
Merge branch 'mesa_7_6_branch'
Conflicts:
src/mesa/drivers/common/meta.c