José Fonseca [Wed, 10 Feb 2010 19:59:53 +0000 (19:59 +0000)]
llvmpipe: Map rendertargets when bound as textures.
Fixes gearbox segfault (fdo 25960). Corruption remains.
Ian Romanick [Wed, 10 Feb 2010 19:43:02 +0000 (11:43 -0800)]
dri: Fix copy-and-paste brain damage in previous commit
A number of places in
3cce4a1e10361458630511543b7a8a6438544775 use
TRUE instead of GL_TRUE. This causes build failures in all of the
drivers that I don't typically build. Win.
Reported by sungami on IRC.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Ian Romanick [Mon, 8 Feb 2010 21:11:58 +0000 (13:11 -0800)]
intel: Don't expose GLX_SWAP_EXCHANGE_OML
We can't always guarantee that the swap will happen by exchange, so we
can't expose this mode. GLX_SWAP_UNDEFINED_OML already covers the
case where the swap *might be* by exchange.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Ian Romanick [Mon, 8 Feb 2010 19:21:29 +0000 (11:21 -0800)]
intel: Expose a minimal number of configs with accumulation buffer
Expose one config per color depth that includes accumulation buffer.
We could probably expose only one config with accumulation buffer, but
that would require figuring out the actual color depth. This is
easier and only exposes 2 useless configs.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Ian Romanick [Mon, 8 Feb 2010 19:09:15 +0000 (11:09 -0800)]
dri: Allow selective generation of accum. buffer configs
Modify the interface to driCreateConfigs allowing drivers to not
expose configs with an accumuation buffer. All of the drivers calling
function have been updated to pass true for the accumulation
selector. This maintains the current behavior.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Reviewed-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Ian Romanick [Mon, 8 Feb 2010 18:34:52 +0000 (10:34 -0800)]
intel: Stop exposing useless 24 depth/0 stencil configs
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Ian Romanick [Mon, 8 Feb 2010 18:33:35 +0000 (10:33 -0800)]
intel: Remove redundant init of depth_bits / stencil_bits in intelInitScreen2
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Ian Romanick [Mon, 8 Feb 2010 18:32:29 +0000 (10:32 -0800)]
intel: Update comment in intelInitScreen2 to noting DRI2 protocol issues
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
unknown [Wed, 10 Feb 2010 17:55:01 +0000 (10:55 -0700)]
Merge branch 'master' of git+ssh://git.freedesktop.org/git/mesa/mesa
Brian Paul [Wed, 10 Feb 2010 17:49:48 +0000 (10:49 -0700)]
glsl/pp: fix extension enable/disable options
Now the #extension name: disable/enable flags do the right thing.
Fixes glean/glsl1 "Preprocessor test (extension test 3)"
unknown [Wed, 10 Feb 2010 17:49:11 +0000 (10:49 -0700)]
Add new symbol to exports list
Michal Krol [Wed, 10 Feb 2010 17:04:07 +0000 (18:04 +0100)]
Simplify GLSL extension mechanism.
Since extension name and extension name string are the same, collapse
them into one name.
Michal Krol [Wed, 10 Feb 2010 16:55:29 +0000 (17:55 +0100)]
glsl/pp: Fix ifdef directive for extension names.
Brian Paul [Wed, 10 Feb 2010 15:27:03 +0000 (08:27 -0700)]
glsl: GLSL extensions have the GL_ prefix
Both the #extension directive name and the preprocessor symbol start
with the GL_ prefix. For example:
...
New glean/glsl1 tests have been added to test the #extension feature.
Zack Rusin [Wed, 10 Feb 2010 02:12:57 +0000 (21:12 -0500)]
st/xorg: fix a silly offset bug in xv
we were multiplying the difference in the wrong direction.
Kristian Høgsberg [Wed, 10 Feb 2010 12:56:13 +0000 (07:56 -0500)]
r600: Fix typo in __DRI_TEXTURE_FORMAT_RGBA rename
Oops, I n00bed it.
José Fonseca [Wed, 10 Feb 2010 11:25:26 +0000 (11:25 +0000)]
llvmpipe: Fix null pointer dereference when no depth buffer is bound.
José Fonseca [Wed, 10 Feb 2010 11:01:26 +0000 (11:01 +0000)]
llvmpipe: Leave depth buffer in swizzled format.
This fixes several assertion failures due to only Z32 being supported.
Vinson Lee [Wed, 10 Feb 2010 09:24:41 +0000 (01:24 -0800)]
glu/sgi: Initialize member of class Uarray.
Vinson Lee [Wed, 10 Feb 2010 08:35:48 +0000 (00:35 -0800)]
llvmpipe: Remove unnecessary headers.
Zack Rusin [Wed, 10 Feb 2010 05:07:51 +0000 (00:07 -0500)]
scons: export drawllvm after actually checking that llvm is present
Kristian Høgsberg [Wed, 10 Feb 2010 02:01:04 +0000 (21:01 -0500)]
egl_dri2: Only allocate config if we're going to keep it
Small optimization.
Kristian Høgsberg [Wed, 10 Feb 2010 01:49:40 +0000 (20:49 -0500)]
egl_dri2: Initialize config attributes with visual id and class from X
Kristian Høgsberg [Wed, 10 Feb 2010 00:59:33 +0000 (19:59 -0500)]
eglinfo: Also print visual type, assuming X visual types
This may not make a lot of sense for non-X EGL, but for EGL under X
it's very useful.
Kristian Høgsberg [Tue, 9 Feb 2010 21:06:23 +0000 (16:06 -0500)]
egl_dri2: Allow pbuffer and pixmap surfaces for all configs
Pauli Nieminen [Wed, 10 Feb 2010 01:20:02 +0000 (03:20 +0200)]
radeon: Fix printf formatings to match te values.
Pauli Nieminen [Wed, 10 Feb 2010 01:00:12 +0000 (03:00 +0200)]
radeon: Define EXT_framebuffer_object constants to match hw.
This hides the assertion failure in glean/fbo test. Underlying
problem when same texture is set twice to different attachments
will cause assertion.
Brian Paul [Wed, 10 Feb 2010 01:05:42 +0000 (18:05 -0700)]
mesa: move all limit/constant assertions into check_context_limits()
Francisco Jerez [Tue, 9 Feb 2010 23:36:22 +0000 (00:36 +0100)]
Drop incorrectly revived files during the 'gallium-nopointsizeminmax' merge.
Corbin Simpson [Tue, 9 Feb 2010 21:58:53 +0000 (13:58 -0800)]
r300g: One less assert.
This entire codepath needs to be cleaned to be more pliant towards
shader failure.
Corbin Simpson [Tue, 9 Feb 2010 21:53:30 +0000 (13:53 -0800)]
r300g: Don't write non-existent bit on non-r500.
Might help with compiz being funky with MRTs.
Corbin Simpson [Tue, 9 Feb 2010 21:50:54 +0000 (13:50 -0800)]
r300g: Correct colorbuffer measurements.
Also clarify changes from pointminmax.
Corbin Simpson [Tue, 9 Feb 2010 21:50:38 +0000 (13:50 -0800)]
r300g: Be ever-so-slightly more useful on bad shaders.
Kristian Høgsberg [Tue, 9 Feb 2010 20:54:59 +0000 (15:54 -0500)]
configure.ac: Enable egl demos when --enable-egl is passed
Kristian Høgsberg [Tue, 9 Feb 2010 20:52:30 +0000 (15:52 -0500)]
xeglgears: Add test case for eglBindTexImage()
Kristian Høgsberg [Tue, 9 Feb 2010 20:49:28 +0000 (15:49 -0500)]
egl_dri2: Implement eglBindTexImage()
Kristian Høgsberg [Tue, 9 Feb 2010 20:26:14 +0000 (15:26 -0500)]
dri_interface: Introduce DRI tokens for the texBuffer texture formats
This used to take GLX tokens, but the DRI interface can't depend on GLX
defines. We fix this by introducing DRI tokens that have the same value
as the GLX texture format tokens.
Kristian Høgsberg [Tue, 9 Feb 2010 15:57:43 +0000 (10:57 -0500)]
egl_dri2: Use a helper function for looking up and binding extensions
Brian Paul [Tue, 9 Feb 2010 20:10:05 +0000 (13:10 -0700)]
gallium/auxiliary: add -D__STDC_CONSTANT_MACROS
Not sure why this is needed now, after the gallivm re-org.
Brian Paul [Tue, 9 Feb 2010 20:08:23 +0000 (13:08 -0700)]
llvmpipe: remove .s rule (already in Makefile.template)
Brian Paul [Tue, 9 Feb 2010 20:07:58 +0000 (13:07 -0700)]
llvmpipe: lp_tile_surface.c isn't committed yet
Pauli Nieminen [Tue, 9 Feb 2010 19:37:19 +0000 (21:37 +0200)]
radeon: Add some debug output for fbo support
Pauli Nieminen [Mon, 8 Feb 2010 09:49:10 +0000 (11:49 +0200)]
r200: Fix UMS notto emit stp.
Polgon stipples are handled by kernel in UMS. Mark the
state as never to be emited for UMS.
Roland Scheidegger [Tue, 9 Feb 2010 18:32:21 +0000 (19:32 +0100)]
Merge branch 'gallium-nopointsizeminmax'
Conflicts:
src/gallium/drivers/nv10/nv10_state.c
src/gallium/drivers/nv20/nv20_state.c
src/gallium/drivers/nv50/nv50_program.c
Brian Paul [Tue, 9 Feb 2010 18:21:41 +0000 (11:21 -0700)]
docs: fix typo: s/osdemo/osdemos/
(cherry picked from commit
0371956d66cdc8a1629753fca9cb3cdc49b0d145)
Chris Li [Sat, 6 Feb 2010 05:25:17 +0000 (21:25 -0800)]
gallium: Adding the %.s rule to generate asm code
Signed-off-by: José Fonseca <jfonseca@vmware.com>
Chris Li [Sat, 6 Feb 2010 05:22:39 +0000 (21:22 -0800)]
gallium: Using the proper pattern rules for Makefile template.
The .o.c rule is consider obsolete.
Signed-off-by: José Fonseca <jfonseca@vmware.com>
Chris Li [Sat, 6 Feb 2010 03:41:32 +0000 (19:41 -0800)]
llvmpipe: Make llvm unit test depend on libllvmpipe.a
Signed-off-by: José Fonseca <jfonseca@vmware.com>
Roland Scheidegger [Mon, 8 Feb 2010 17:48:08 +0000 (18:48 +0100)]
st/mesa: fix wrong initialization of MaxPointSize
Jakob Bornecrantz [Sat, 6 Feb 2010 00:30:29 +0000 (01:30 +0100)]
vmware/core: Support drm interface version 1.0.0
Jakob Bornecrantz [Sat, 6 Feb 2010 00:29:47 +0000 (01:29 +0100)]
vmware/core: Use correct flags in scanout code
Jakob Bornecrantz [Sat, 6 Feb 2010 00:21:36 +0000 (01:21 +0100)]
vmware/core: Handle new scanout field
Jakob Bornecrantz [Fri, 5 Feb 2010 23:56:24 +0000 (00:56 +0100)]
vmware/core: Update header to latest version
Jakob Bornecrantz [Fri, 29 Jan 2010 16:11:36 +0000 (17:11 +0100)]
vmware/core: Remove old ioctl structs not used
Leftover that was never used in this driver
Michal Krol [Tue, 9 Feb 2010 15:20:55 +0000 (16:20 +0100)]
llvmpipe: Remove reference to prefilter field.
Kristian Høgsberg [Tue, 9 Feb 2010 15:21:19 +0000 (10:21 -0500)]
glx: Fix TOP setting in src/glx Makefile
Argh, forgot to commit this fix before pushing.
Kristian Høgsberg [Tue, 9 Feb 2010 14:58:36 +0000 (09:58 -0500)]
Retire miniglx and move the actual glx code up to src/glx
Kristian Høgsberg [Tue, 9 Feb 2010 14:49:56 +0000 (09:49 -0500)]
egl_dri2: Fix initialization with EGL_DEFAULT_DISPLAY
Brian Paul [Fri, 5 Feb 2010 19:57:11 +0000 (12:57 -0700)]
fix bug in perspective corrected interpolants for lines
Brian Paul [Fri, 5 Feb 2010 17:54:40 +0000 (10:54 -0700)]
progs/tests: add line drawing option to cylwrap.c
Brian Paul [Fri, 5 Feb 2010 17:21:15 +0000 (10:21 -0700)]
progs/tests: add x/y rotation to cylwrap.c
Brian Paul [Fri, 5 Feb 2010 17:18:25 +0000 (10:18 -0700)]
st/mesa: remove special-case cyl-wrap code
Cylinder wrap mode works with perspective interpolation now.
Michal Krol [Fri, 5 Feb 2010 16:12:48 +0000 (17:12 +0100)]
softpipe: Properly implement cylindrical wrapping.
Michal Krol [Fri, 5 Feb 2010 14:16:07 +0000 (15:16 +0100)]
Revert "softpipe: Implement cylindrical wrapping for linear interpolator."
This reverts commit
444cd62ca33a45bd39e8408979a0a01c246c6381.
Brian Paul [Thu, 4 Feb 2010 23:43:08 +0000 (16:43 -0700)]
progs/tests: new test program for exercising cylindrical tex wrap mode
Brian Paul [Thu, 4 Feb 2010 23:36:55 +0000 (16:36 -0700)]
st/mesa: check for PROG_PARAM_BIT_CYL_WRAP flag
Brian Paul [Thu, 4 Feb 2010 23:40:47 +0000 (16:40 -0700)]
mesa: add back-door support for cylindrical texture wrap mode
If the texture priority field is 0.125 set the PROG_PARAM_BIT_CYL_WRAP
flag. The gallium state tracker will look for this flag.
This feature is only for testing purposes and may be remove at some
point. But if it's useful we could write a GL/Mesa extension to
expose it properly.
Michal Krol [Wed, 3 Feb 2010 14:49:10 +0000 (15:49 +0100)]
gallium: Remove prefilter member from pipe_sampler_state struct.
Michal Krol [Wed, 3 Feb 2010 14:45:32 +0000 (15:45 +0100)]
docs: Document TGSI Declaration token.
Michal Krol [Wed, 3 Feb 2010 14:30:17 +0000 (15:30 +0100)]
softpipe: Implement cylindrical wrapping for linear interpolator.
Michal Krol [Wed, 3 Feb 2010 14:29:21 +0000 (15:29 +0100)]
tgsi: Gather cylindrical wrap info in tgsi_shader_info struct.
Michal Krol [Wed, 3 Feb 2010 14:28:50 +0000 (15:28 +0100)]
tgsi: Add ureg_DECL_fs_input_cyl().
Allows one to declare fragment shader inputs with cylindrical wrap info.
Michal Krol [Wed, 3 Feb 2010 14:28:03 +0000 (15:28 +0100)]
gallium: Add cylindrical wrap info to TGSI declaration.
Michal Krol [Wed, 3 Feb 2010 10:19:04 +0000 (11:19 +0100)]
tgsi: Remove tgsi_dump_c.[ch].
Little utility after development stabilisation, use tgsi_dump instead.
Michal Krol [Tue, 9 Feb 2010 13:25:41 +0000 (14:25 +0100)]
mesa: Enable true refcounting for NullBufferObj.
This object can be shared with another context, so we cannot just
delete it when the owning context is being destroyed.
Ensuring that buffer objects are properly refcounted guarantees
NullBufferObj is destroyed when all references to it are removed.
Michal Krol [Mon, 8 Feb 2010 18:43:42 +0000 (19:43 +0100)]
mesa: Protect buffer objects reference counting with a mutex.
José Fonseca [Mon, 1 Feb 2010 21:33:04 +0000 (21:33 +0000)]
mesa: Ensure object refcount is null when destroying the buffer.
Lets see if this is not too pedantic. Obj pointers are never exposed to
GL apps so it should be possible to get this right.
Furthermore apps with GL widgets and test suits create and destroy many
contexts and objects, so bad reference counting is not really an option.
José Fonseca [Mon, 1 Feb 2010 21:30:50 +0000 (21:30 +0000)]
mesa: Fix null buffer object reference counting.
Always use _mesa_reference_buffer_object, and never call
ctx->Driver.DeleteBuffer() directly to prevent dangling pointers to the
null buffer object.
This fixes crash/assertions in sharedtex_mt and Autodesk Mudbox.
José Fonseca [Mon, 1 Feb 2010 21:22:10 +0000 (21:22 +0000)]
mesa: Always do proper ref counting of shared state.
José Fonseca [Thu, 28 Jan 2010 20:51:52 +0000 (20:51 +0000)]
svga: Actually call fence_reference in texture downloads.
Vinson Lee [Tue, 9 Feb 2010 09:35:14 +0000 (01:35 -0800)]
progs/xdemos: Silence uninitialized variable warning.
Vinson Lee [Tue, 9 Feb 2010 09:26:11 +0000 (01:26 -0800)]
r300g: Fix off-by-one errors in array bounds assertions.
Zack Rusin [Tue, 9 Feb 2010 05:00:35 +0000 (00:00 -0500)]
scons: don't assume llvm is always present
Zack Rusin [Tue, 9 Feb 2010 02:50:33 +0000 (21:50 -0500)]
llvmpipe: switch to using dynamic stack allocation instead of registers
with mutable vars we don't need to follow the phi nodes. meaning that
control flow becomes trivial as we don't have scan the rest of the tgsi
to figure out the variable usage anymore. futhermore the memory2register
pass promotes alloca/store/load to registers while inserting the right phi
nodes. so we get simplicity and performance.
Vinson Lee [Tue, 9 Feb 2010 02:22:29 +0000 (18:22 -0800)]
identity: Remove unnecessary header.
Vinson Lee [Tue, 9 Feb 2010 02:14:49 +0000 (18:14 -0800)]
softpipe: Remove unnecessary headers.
Brian Paul [Tue, 9 Feb 2010 00:51:48 +0000 (17:51 -0700)]
gallivm: added lp_bld_misc.cpp to sources list
Brian Paul [Tue, 9 Feb 2010 00:40:02 +0000 (17:40 -0700)]
llvmpipe: include gallivm/lp_bld_misc.h
Brian Paul [Tue, 9 Feb 2010 00:39:08 +0000 (17:39 -0700)]
gallivm: added gallivm/lp_bld_misc.cpp to Makefile
Brian Paul [Tue, 9 Feb 2010 00:38:35 +0000 (17:38 -0700)]
llvmpipe: remove files moved to gallivm/ from llvmpipe/Makefile
Brian Paul [Tue, 9 Feb 2010 00:34:59 +0000 (17:34 -0700)]
llvmpipe: s/textured/texture/
Brian Paul [Mon, 8 Feb 2010 20:09:18 +0000 (13:09 -0700)]
llvmpipe: added counters for color tile clear, load, store
Brian Paul [Mon, 8 Feb 2010 20:07:31 +0000 (13:07 -0700)]
llvmpipe: use new os time functions
Francisco Jerez [Mon, 8 Feb 2010 22:15:03 +0000 (23:15 +0100)]
glx: Fix SwapBuffers regression introduced by
01923fb72d.
After that commit, some dri2 protocol symbols were being checked from
places that weren't including dri2proto.h, effectively disabling some
valuable SwapBuffers codepaths.
Zack Rusin [Mon, 8 Feb 2010 23:05:22 +0000 (18:05 -0500)]
llvmpipe: export the tgsi translation code to a common layer
the llvmpipe tgsi translation is a lot more complete than what was in
gallivm so replacing the latter with the former. this is needed since
the draw llvm paths will use the same code. effectively the proven
llvmpipe code becomes gallivm.
Corbin Simpson [Mon, 8 Feb 2010 22:03:28 +0000 (14:03 -0800)]
radeong: Provide drm_api::destroy hook.
Pedantic Corbin is pedantic.
Corbin Simpson [Mon, 8 Feb 2010 21:37:55 +0000 (13:37 -0800)]
radeong: Always return something for create_screen.
Prevents a segfault on e.g. r600.
Keith Whitwell [Mon, 8 Feb 2010 19:45:02 +0000 (19:45 +0000)]
st/dri: fix compile breakage
Keith Whitwell [Mon, 8 Feb 2010 17:15:47 +0000 (17:15 +0000)]
egl: re-fix context_create typo
Keith Whitwell [Mon, 8 Feb 2010 17:13:15 +0000 (17:13 +0000)]
svga: remove dead debug sleep