Vinson Lee [Wed, 27 Jan 2010 22:04:53 +0000 (14:04 -0800)]
r300g: Silence GCC "missing braces around initializer" warning.
Alex Deucher [Wed, 27 Jan 2010 17:18:02 +0000 (12:18 -0500)]
r600: fix warning
Alex Deucher [Wed, 27 Jan 2010 17:16:50 +0000 (12:16 -0500)]
r600: rv670 support 8 tex instructions just like other r6xx
also clarify some other const values.
Brian Paul [Wed, 27 Jan 2010 16:21:57 +0000 (09:21 -0700)]
cell: remove commas from structs wrapped in PIPE_ALIGN_TYPE macro
This avoids the need to make PIPE_ALIGN_TYPE a variadic macro.
Alan Hourihane [Wed, 27 Jan 2010 12:46:46 +0000 (12:46 +0000)]
Duplicate CCVERSION check code to compensate for different env CC.
Andre Maasikas [Wed, 27 Jan 2010 09:22:56 +0000 (11:22 +0200)]
r600: increase max texture units to 16
Andre Maasikas [Wed, 27 Jan 2010 08:20:15 +0000 (10:20 +0200)]
r600: fix XPD with writemask
same variable used for 2 different temp registers
fixes e.g. glsl/bump
Vinson Lee [Wed, 27 Jan 2010 07:28:23 +0000 (23:28 -0800)]
softpipe: Remove unnecessary header.
Vinson Lee [Wed, 27 Jan 2010 07:26:41 +0000 (23:26 -0800)]
i915g: Remove unnecessary headers.
Vinson Lee [Wed, 27 Jan 2010 06:56:58 +0000 (22:56 -0800)]
scons: Use '-Werror=' option on GCC 4.2.x and greater.
The existing code only checked for GCC 4.2.x and 4.3.x.
Corbin Simpson [Wed, 27 Jan 2010 06:08:11 +0000 (22:08 -0800)]
r300g: Turn the RS block into an atom.
At least one extraneous dirty was eliminated, as well as the chance for
avoiding dirty on shader change.
Corbin Simpson [Wed, 27 Jan 2010 05:49:02 +0000 (21:49 -0800)]
r300g: Make vertex_format into an atom.
Some delicious hax here.
Corbin Simpson [Wed, 27 Jan 2010 05:41:44 +0000 (21:41 -0800)]
r300g: Kill dead code for hashing custom state.
It just wasn't meant to be.
Corbin Simpson [Wed, 27 Jan 2010 04:11:39 +0000 (20:11 -0800)]
r300g: Add A8_UNORM texture format.
Keeps teeworld happy.
Dave Airlie [Wed, 27 Jan 2010 01:46:26 +0000 (11:46 +1000)]
radeon/r200/r300: don't clean non-emitted state.
So if we don't actually emit an atom to the hw because we don't
need it in the current state (e.g. lighting related atoms when
lighting is off) then don't mark it as clean, because when
lighting gets switched on we won't emit it at all.
This fixes funky gears colors.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Wed, 27 Jan 2010 01:02:32 +0000 (11:02 +1000)]
radeon: remove unused file
Eric Anholt [Wed, 27 Jan 2010 02:04:03 +0000 (18:04 -0800)]
intel: Remove dead code from having to clip copyteximage source rect.
mesa core does it now. If only it did so for other entrypoints.
Eric Anholt [Wed, 27 Jan 2010 02:01:37 +0000 (18:01 -0800)]
intel: Use a handy helper in glReadPixels source clipping.
Eric Anholt [Wed, 27 Jan 2010 01:47:03 +0000 (17:47 -0800)]
intel: Clean up stale comments about cliprects.
Eric Anholt [Wed, 27 Jan 2010 01:40:44 +0000 (17:40 -0800)]
intel: Remove the remaining cliprects code from DRI1.
Eric Anholt [Wed, 27 Jan 2010 01:36:35 +0000 (17:36 -0800)]
intel: Remove DRI1 junk from spans code.
This reduces the driver size by over 1%.
Eric Anholt [Wed, 27 Jan 2010 01:11:11 +0000 (17:11 -0800)]
intel: Remove DRI1 junk from blit glBitmap.
Eric Anholt [Wed, 27 Jan 2010 01:02:55 +0000 (17:02 -0800)]
intel: Remove DRI1 junk from CopyPixels.
Eric Anholt [Wed, 27 Jan 2010 00:40:05 +0000 (16:40 -0800)]
intel: Remove DRI1 junk from glClear blit implementation.
Eric Anholt [Wed, 27 Jan 2010 00:35:27 +0000 (16:35 -0800)]
i965: Remove DRI1 leftovers from stipple offset handling.
Vinson Lee [Wed, 27 Jan 2010 01:00:15 +0000 (17:00 -0800)]
i965g: Silence uninitialized variable warning.
Vinson Lee [Wed, 27 Jan 2010 00:45:32 +0000 (16:45 -0800)]
softpipe: Remove unused variables.
Vinson Lee [Wed, 27 Jan 2010 00:37:10 +0000 (16:37 -0800)]
st/mesa: Silence uninitialized variable warning.
Eric Anholt [Wed, 27 Jan 2010 00:22:10 +0000 (16:22 -0800)]
intel: Fix PBO blit ReadPixels from an FBO.
Bug #25921 -- clutter PBO usage gave unreliable results.
Igor Oliveira [Mon, 25 Jan 2010 23:23:04 +0000 (19:23 -0400)]
docs: add documentation to double opcodes
Signed-off-by: Brian Paul <brianp@vmware.com>
Eric Anholt [Tue, 26 Jan 2010 22:43:23 +0000 (14:43 -0800)]
i965: Add support for EXT_draw_buffers2.
Eric Anholt [Tue, 26 Jan 2010 21:08:42 +0000 (13:08 -0800)]
i965: Fix fp fragment.position handling and enable HW part of ARB_fcc.
As with swrast, this fixes the default pixel center behavior which was
broken, and implements the previous behavior for integer. Fixes
piglit fp-arb-fragment-coord-conventions-none. The extension won't be
exposed until we get the GLSL part implemented.
The DRI1 origin_x/y parts are dropped since they're no longer relevant.
Eric Anholt [Tue, 26 Jan 2010 20:43:43 +0000 (12:43 -0800)]
swrast: Implement ARB_fragment_coord_conventions but don't enable.
This brings swrast's support up to the state of gallium, and fixes the
default center behavior of fragment.position.xy in piglit
fp-arb-fragment-coord-conventions-none.
The extension is not enabled currently because the GLSL part of the
extension isn't supported, so piglit
glsl-arb-fragment-coord-conventions-define fails as would any serious
test of the GLSL part.
Alan Hourihane [Tue, 26 Jan 2010 20:58:11 +0000 (20:58 +0000)]
support an 'embedded' platform target which turns off most parts of the
build.
Alan Hourihane [Tue, 26 Jan 2010 20:57:03 +0000 (20:57 +0000)]
Check for __USE_MISC for defining uint & ushort
Brian Paul [Tue, 26 Jan 2010 20:23:25 +0000 (13:23 -0700)]
Merge branch 'mesa_7_7_branch'
Merging was easier than cherry picking in this instance.
Erik Wien [Tue, 26 Jan 2010 20:19:30 +0000 (13:19 -0700)]
mesa: Don't bind DRAW/READ_FRAMEBUFFER separately without FBO blit support
If GL_EXT_framebuffer_blit was not supported _mesa_DeleteFramebuffersEXT
would raise an error when deleting the currently bound framebuffer. This
because it tried to bind the default DRAW- and READ_FRAMEBUFFER separately.
This patch binds the default FRAMEBUFFER instead in that case.
Encountered in the fbo/fbo-copyteximage piglit test on R600.
Patch cleaned up a bit by Brian Paul.
Jesse Barnes [Tue, 26 Jan 2010 20:16:53 +0000 (22:16 +0200)]
intel: make sure we update the renderbuffers after a swap
Now that LOCK_HARDWARE is gone, we don't have a convenient place to
update the renderbuffers everywhere we need them. So grab new buffers
when we invalidate the old ones until we optimize things further.
Alan Hourihane [Tue, 26 Jan 2010 19:57:34 +0000 (19:57 +0000)]
add missing SCons.Util
Brian Paul [Tue, 26 Jan 2010 19:49:44 +0000 (12:49 -0700)]
vbo: if 'end' is out of bounds, clamp it
If we determine that the 'end' parameter to glDrawElements() is out of
bounds, clamp it to the max legal index value.
Brian Paul [Tue, 26 Jan 2010 19:47:51 +0000 (12:47 -0700)]
vbo: clamp DrawElements start/end to max possible values
Some apps are sloppy with their start/end values. Clamp them to
max possible values to prevent problems later.
Alan Hourihane [Tue, 26 Jan 2010 19:18:32 +0000 (19:18 +0000)]
fix typo
Alan Hourihane [Tue, 26 Jan 2010 19:14:16 +0000 (19:14 +0000)]
Allow the environment to override certain flags.
Alan Hourihane [Tue, 26 Jan 2010 19:13:27 +0000 (19:13 +0000)]
only use -Werror flags with gcc 4.2.x and 4.3.x.
could use a better test here though.
Aaron Plattner [Tue, 26 Jan 2010 18:14:11 +0000 (10:14 -0800)]
glxgears: Support GLX_EXT_swap_control for querying the current swap interval.
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
Brian Paul [Tue, 26 Jan 2010 17:53:34 +0000 (10:53 -0700)]
mesa: remove redundant _MaxElement computation
Eric added some new code to check if offset < obj_size before computing
_MaxElement but my original code was still present afterward and it
clobbered the _MaxElement value.
Not sure if this came from a bad merge or what.
Brian Paul [Tue, 26 Jan 2010 17:48:00 +0000 (10:48 -0700)]
progs/tests: use glDrawRangeElements() in bufferobj.c
Brian Paul [Tue, 26 Jan 2010 16:12:32 +0000 (09:12 -0700)]
docs: document Intel, R300 env vars
Chia-I Wu [Tue, 26 Jan 2010 10:41:15 +0000 (18:41 +0800)]
egl: Remove _eglOpenDriver and _eglCloseDriver.
_eglCloseDriver is no-op and _eglOpenDriver does nothing but call
_eglMatchDriver. Export _eglMatchDriver directly.
Chia-I Wu [Tue, 26 Jan 2010 10:34:29 +0000 (18:34 +0800)]
egl: Fix a segfault when a display is initialized again.
Reset dpy->MaxConfigs so that dpy->Configs is re-allocated.
Chia-I Wu [Tue, 26 Jan 2010 09:13:51 +0000 (17:13 +0800)]
egl: Remove _eglGetCurrentSurface and _eglGetCurrentDisplay.
They have little use in drivers since drivers need to work for multiple
current contexts.
Chia-I Wu [Tue, 26 Jan 2010 09:22:21 +0000 (17:22 +0800)]
egl: Fix leaks in eglReleaseThread.
There may be multiple bound contexts that should be unbound.
Chia-I Wu [Tue, 26 Jan 2010 08:53:40 +0000 (16:53 +0800)]
egl: Refactor _eglMakeCurrent.
Refactor _eglMakeCurrent into _eglCheckMakeCurrent,
_eglBindContextToSurface, and _eglBindContextToThread.
Chia-I Wu [Tue, 26 Jan 2010 08:52:14 +0000 (16:52 +0800)]
egl: Add _eglGetAPIContext.
It will return the currently bound context of the given API.
Chia-I Wu [Tue, 26 Jan 2010 07:16:49 +0000 (15:16 +0800)]
egl: Use a boolean to indicate whether a resource is linked.
An unlinked resource may still be a current resource such as current
surfaces. There might still be a need to know which display the
unlinked resource belongs to.
Chia-I Wu [Tue, 26 Jan 2010 07:10:18 +0000 (15:10 +0800)]
egl: Mark _eglCheckResource as PUBLIC.
To support extensions such as GL_OES_EGL_image, the drivers need a way
to check if a given EGLImageKHR is valid.
Vinson Lee [Tue, 26 Jan 2010 08:45:54 +0000 (00:45 -0800)]
glsl/pp: Silence GCC "no previous prototype" warning.
Eric Anholt [Mon, 25 Jan 2010 22:53:50 +0000 (14:53 -0800)]
intel: Don't do client-side frame throttling with DRI2 SwapBuffers.
The server side does the throttling on our behalf now by putting the
client to sleep, so we don't need our previous hacks for limiting the
number of outstanding frames. Same effect as
7d4e674b212c9dc6408c13913a399bd4a2b9a1e3.
Eric Anholt [Mon, 25 Jan 2010 22:48:57 +0000 (14:48 -0800)]
Revert "intel: Use the new DRI2 flush invalidate entrypoint to signal frame done."
This reverts commit
7d4e674b212c9dc6408c13913a399bd4a2b9a1e3. It broke
throttling in the non-new-DRI2 case.
Eric Anholt [Tue, 19 Jan 2010 23:26:56 +0000 (15:26 -0800)]
i965: Remove unnecessary malloc/free in VS binding table setup.
Eric Anholt [Tue, 26 Jan 2010 06:27:46 +0000 (22:27 -0800)]
i965: Fix build after merge of mesa stable branch.
Chia-I Wu [Tue, 26 Jan 2010 03:35:30 +0000 (11:35 +0800)]
docs: Mention EGL and OpenGL ES in 7.8 release notes.
Chia-I Wu [Thu, 21 Jan 2010 09:58:28 +0000 (17:58 +0800)]
glx: Build GLX normally.
GLX was built specially and before Mesa core because libglapi.a could
not be built with IN_DRI_DRIVER defined. This is no longer the case
since
6e99e6ddbf488f6955e34ef0bc438fdcb4d90f74.
It works fine in my (limited) testing with both direct and indirect
rendering. I also compare the outputs after preprocessing (gcc -E) with
or without this commit, and they are identical.
Chia-I Wu [Tue, 26 Jan 2010 02:54:45 +0000 (10:54 +0800)]
docs: Update EGL documentation.
Add --enable-gallium-swrast to the example. Document missing
--with-egl-driver-dir and update TODOs.
Chia-I Wu [Tue, 26 Jan 2010 02:46:32 +0000 (10:46 +0800)]
egl/xdri: Update with glxclient.h change.
__glXFindDisplay is no longer static since
7a9329ba99fe1242c07fbf4fd04e7a4dbeba2e72.
Robert Bragg [Wed, 20 Jan 2010 03:01:14 +0000 (03:01 +0000)]
mesa: misc GLX_INTEL_swap_event fixes
Fixup a few issues found through testing:
- update GLX names to match glproto
- register DRI2WireToEvent so it actually gets called
Signed-off-by: Robert Bragg <robert@linux.intel.com>
Brian Paul [Mon, 25 Jan 2010 21:46:17 +0000 (14:46 -0700)]
Merge branch 'mesa_7_7_branch'
Conflicts:
src/mesa/drivers/dri/intel/intel_screen.c
src/mesa/drivers/dri/intel/intel_swapbuffers.c
src/mesa/drivers/dri/r300/r300_emit.c
src/mesa/drivers/dri/r300/r300_ioctl.c
src/mesa/drivers/dri/r300/r300_tex.c
src/mesa/drivers/dri/r300/r300_texstate.c
Alex Deucher [Mon, 25 Jan 2010 21:19:19 +0000 (16:19 -0500)]
r300: rv350+ support FLT16_2/4 vertex formats
Corbin Simpson [Mon, 25 Jan 2010 19:27:50 +0000 (11:27 -0800)]
r300g: Finish and enable immediate mode.
Based on Marek's immd-mode patch.
Corbin Simpson [Mon, 25 Jan 2010 19:04:55 +0000 (11:04 -0800)]
radeong: Make sure that we're not emitting relocations for local buffers.
Split from Marek's immd-mode patch.
Corbin Simpson [Mon, 25 Jan 2010 19:04:15 +0000 (11:04 -0800)]
radeong: Make is_r3xx inline to avoid warnings.
Split from Marek's immd-mode patch.
Corbin Simpson [Mon, 25 Jan 2010 19:01:16 +0000 (11:01 -0800)]
radeong: Clean up domain usages.
Split from Marek's immd-mode patch.
Corbin Simpson [Mon, 25 Jan 2010 18:52:19 +0000 (10:52 -0800)]
Revert "r300g,radeong: finish and enable the immediate mode"
This reverts commit
112239e9a66a155d36fe2ad0ab130e6f26eff298.
Jakob Bornecrantz [Mon, 25 Jan 2010 19:07:43 +0000 (20:07 +0100)]
st/xorg: Fix crash on resize with libkms
Igor Oliveira [Mon, 25 Jan 2010 16:53:53 +0000 (09:53 -0700)]
vega: fix incorrect samplers, textures indexes in blend_bind_samplers()
Fixes fd.o bug 25863.
Signed-off-by: Brian Paul <brianp@vmware.com>
Luca Barbieri [Wed, 20 Jan 2010 01:39:27 +0000 (17:39 -0800)]
Fix compressed texture loads for non-minimal pitches again
My commit
eea6a7639f767b1d30b6ef1f91a9c49e3f3b78f0 does a memcpy of height
lines, but that's wrong because the texture has a block layout and we
must thus use the number of vertical blocks instead of the height.
Signed-off-by: Brian Paul <brianp@vmware.com>
Chia-I Wu [Mon, 25 Jan 2010 14:29:51 +0000 (22:29 +0800)]
winsys/drm: Correctly install EGL drivers.
Remove extraneous quotes that prevent some drivers to be installed when
there are more than one display in EGL_DISPLAYS.
José Fonseca [Mon, 25 Jan 2010 13:17:47 +0000 (13:17 +0000)]
pipebuffer: Reapply the mesa_7_7_branch's changes.
After the last mesa_7_7_branch merge the old master code remained,
instead of the newer mesa_7_7_branch's code. This commit makes both
branches match, modulo interface changes.
Note: future mesa_7_7_branch -> master merges will most likely hit
conflicts in these files. Unless otherwise stated, the mesa_7_7_branch
is the reference.
José Fonseca [Mon, 25 Jan 2010 11:54:54 +0000 (11:54 +0000)]
llvmpipe: Remove lp_tex_sample_c.c again.
Popped back to life in some merge. Not necessary even as a reference,
since it is identical to softpipe's sp_tex_sample.c with renamed symbols.
Jakob Bornecrantz [Mon, 25 Jan 2010 11:27:47 +0000 (12:27 +0100)]
vmware/xorg: Do buffer round-robin logic differently
Jakob Bornecrantz [Fri, 22 Jan 2010 17:28:40 +0000 (18:28 +0100)]
st/xorg: Improve options and print them to log
Set 2D acceleration to off by default
Get fallback debugging from the Xorg config
Also print if 3D acceleration is enabled
Jakob Bornecrantz [Thu, 21 Jan 2010 20:32:28 +0000 (21:32 +0100)]
vmware/xorg: Export pci_probe function for ugly chain-loading
Jakob Bornecrantz [Thu, 21 Jan 2010 20:23:22 +0000 (21:23 +0100)]
st/xorg: Export helper function for detecting modesetting
Jakob Bornecrantz [Thu, 21 Jan 2010 20:28:14 +0000 (21:28 +0100)]
st/xorg: Fix warning
Corbin Simpson [Mon, 25 Jan 2010 09:12:30 +0000 (01:12 -0800)]
gallium-docs: Add note on scissor.
Marek Olšák [Sun, 17 Jan 2010 23:15:52 +0000 (00:15 +0100)]
r300g,radeong: finish and enable the immediate mode
Nearly 100% performance increase in glxgears.
Marek Olšák [Sun, 17 Jan 2010 22:13:51 +0000 (23:13 +0100)]
r300g: fix zero-area scissor
Marek Olšák [Sat, 16 Jan 2010 23:57:04 +0000 (00:57 +0100)]
r300g,radeong: set tiling flags in libdrm
Marek Olšák [Sat, 23 Jan 2010 16:19:44 +0000 (17:19 +0100)]
radeong: do not emit a zero-sized command stream
Fixing a hardlock introduced in
fcbd285e421903ee0a65f19f5d633b25b5923c24 and
reproducible with piglit/bugs/fdo23489.
Chia-I Wu [Mon, 25 Jan 2010 03:55:48 +0000 (11:55 +0800)]
egl: Native types are renamed in EGL 1.3.
Rename Native*Type to EGLNative*Type.
Chia-I Wu [Mon, 25 Jan 2010 03:39:44 +0000 (11:39 +0800)]
egl: Make resource void pointer in _eglCheckResource.
This emphasizes the fact that the resource to be checked could really be
invalid and have an unknown type.
Chia-I Wu [Mon, 25 Jan 2010 03:34:09 +0000 (11:34 +0800)]
egl: Install EGL headers.
Install EGL (and KHR) headers along with the library.
Chia-I Wu [Sat, 15 Aug 2009 14:58:13 +0000 (22:58 +0800)]
egl: Add support for EGL_KHR_image.
Individual drivers still need to implement the API hooks.
Chia-I Wu [Sat, 15 Aug 2009 14:57:47 +0000 (22:57 +0800)]
egl: Update headers.
Update to the current versions found at
http://www.khronos.org/registry/egl/. There is one modification in
khrplatform.h for GCC visibility.
Igor Oliveira [Sun, 24 Jan 2010 16:26:31 +0000 (12:26 -0400)]
egl: check if driver_name is null
Brian Paul [Mon, 25 Jan 2010 01:18:17 +0000 (18:18 -0700)]
st/mesa: fix int->uint conversion for negative scissor bound values
Based on a patch by Xavier Chantry <chantry.xavier@gmail.com>:
If x+width or y+height is negative, then maxx or maxy will get a bogus value
when converting that to unsigned. Fix this by setting 0 as minimal value.
This was also triggered by teeworlds, but only with some combination of
resolution and map section. For example upper part of dm2 at 1280x1024.
Brian Paul [Mon, 25 Jan 2010 00:54:44 +0000 (17:54 -0700)]
mesa: move _mesa_debug() call earlier in _mesa_Scissor
Part of a patch from Xavier Chantry <chantry.xavier@gmail.com>
Xavier Chantry [Sat, 23 Jan 2010 16:27:21 +0000 (17:27 +0100)]
st/mesa: fix unsigned/signed breakage in scissor
commit
53174afeeb introduced a portability change that converted GLint x,y
to GLuint. That breaks when x and y are negative, which seems to be allowed,
and which at least one game uses : teeworlds.
Rather than simply reverting the change, it seems possible to convert the
16bit unsigned to GLint so that comparisons are made between signed integers
instead. This hopefully does not break anything while keeping MSVC happy.
Signed-off-by: Xavier Chantry <chantry.xavier@gmail.com>
Signed-off-by: Brian Paul <brianp@vmware.com>
Marcin Slusarz [Sun, 24 Jan 2010 18:27:59 +0000 (19:27 +0100)]
nv50: fix uninitialized variable in nv50_revdep_reorder
"unsafe" is never initialized, but used
(found by valgrind)
Marcin Slusarz [Wed, 20 Jan 2010 22:27:07 +0000 (23:27 +0100)]
nv50: fix crash in nv50_pre_pipebuffer_map (nv50_screen->cur_ctx)
nv50_pre_pipebuffer_map references screen->cur_ctx which points
to freed memory after the context is destroyed.
This crash is easily triggerable by progs/xdemos/glxcontexts.