mesa.git
14 years agoradeon/r200/r600: reset bos when validating buffers during blit
Alex Deucher [Mon, 8 Mar 2010 15:04:02 +0000 (10:04 -0500)]
radeon/r200/r600: reset bos when validating buffers during blit

14 years agor600: recalculate point size, if point min/max size changes
Alex Deucher [Mon, 8 Mar 2010 15:00:59 +0000 (10:00 -0500)]
r600: recalculate point size, if point min/max size changes

14 years agor600: no need to flush on context init
Alex Deucher [Mon, 8 Mar 2010 14:57:26 +0000 (09:57 -0500)]
r600: no need to flush on context init

14 years agor300: don't enable EXT_packed_depth_stencil
Maciej Cencora [Sat, 13 Feb 2010 23:02:05 +0000 (00:02 +0100)]
r300: don't enable EXT_packed_depth_stencil

R300 hw doesn't support sampling from Z24_S8 or S8_Z24 formats.

14 years agoradeon: no need to emit full state twice after flush
Maciej Cencora [Sat, 16 Jan 2010 14:11:37 +0000 (15:11 +0100)]
radeon: no need to emit full state twice after flush

14 years agor300: allow src and dst BOs to be placed in GTT during blit
Maciej Cencora [Thu, 4 Mar 2010 17:35:55 +0000 (18:35 +0100)]
r300: allow src and dst BOs to be placed in GTT during blit

Fixes some relocation failures

14 years agor300: reset bos when validating buffers during blit
Maciej Cencora [Sat, 6 Feb 2010 17:25:36 +0000 (18:25 +0100)]
r300: reset bos when validating buffers during blit

14 years agoradeon: fallback to software in glCopyTexImage if blit isn't available
Maciej Cencora [Sat, 6 Feb 2010 18:37:23 +0000 (19:37 +0100)]
radeon: fallback to software in glCopyTexImage if blit isn't available

14 years agor300: VAP flush is needed only when vertex program or constants are changed
Maciej Cencora [Sun, 7 Mar 2010 11:02:28 +0000 (12:02 +0100)]
r300: VAP flush is needed only when vertex program or constants are changed

14 years agor300: recalculate point size, if point min/max size changes
Maciej Cencora [Sat, 6 Mar 2010 12:39:13 +0000 (13:39 +0100)]
r300: recalculate point size, if point min/max size changes

Fixes two wine d3d9 unit tests

14 years agor300: no need to flush on context init
Maciej Cencora [Sat, 6 Mar 2010 12:18:11 +0000 (13:18 +0100)]
r300: no need to flush on context init

14 years agor300: remove unnecessary code
Maciej Cencora [Sat, 6 Mar 2010 12:09:47 +0000 (13:09 +0100)]
r300: remove unnecessary code

_tnl_UpdateFixedFunctionProgram is already called in r300_draw.c

14 years agomesa: add render-to-texture case for MESA_FORMAT_S8_Z24
Brian Paul [Mon, 8 Mar 2010 16:38:07 +0000 (09:38 -0700)]
mesa: add render-to-texture case for MESA_FORMAT_S8_Z24

14 years agomesa: s/GL_DEPTH_STENCIL/GL_DEPTH_COMPONENT/ for MESA_FORMAT_Z16 renderbuffer
Brian Paul [Mon, 8 Mar 2010 16:34:46 +0000 (09:34 -0700)]
mesa: s/GL_DEPTH_STENCIL/GL_DEPTH_COMPONENT/ for MESA_FORMAT_Z16 renderbuffer

MESA_FORMAT_Z16 has no stencil bits.

14 years agost/vega: Fix OpenVG demo segfaults.
Chia-I Wu [Wed, 3 Mar 2010 15:51:13 +0000 (23:51 +0800)]
st/vega: Fix OpenVG demo segfaults.

When the paint is color, paint_bind_samplers binds a dummy sampler
without a texture.  It causes demos requiring a sampler (those use a
mask or an image) to crash.

14 years agogallivm: Initialize variables for default cases.
Vinson Lee [Sun, 7 Mar 2010 04:30:28 +0000 (20:30 -0800)]
gallivm: Initialize variables for default cases.

Fixes use of uninitialized variables in non-debug builds.

14 years agoprogs/demos: Add arbocclude2 to Makefile and .gitignore.
Vinson Lee [Sat, 6 Mar 2010 07:18:01 +0000 (23:18 -0800)]
progs/demos: Add arbocclude2 to Makefile and .gitignore.

14 years agoglx: Move initialization of ext_list_first_time to where it's storage is allocated
Brian Paul [Sat, 6 Mar 2010 00:32:36 +0000 (17:32 -0700)]
glx: Move initialization of ext_list_first_time to where it's storage is allocated

Move the initialization of ext_list_first_time from all of the DRI loader's
CreateScreen routines, to where the storage for the screen config is
allocated.

It needs to get set in the screen-config even if DRI is forced off
using LIBGL_ALWAYS_INDIRECT, so that psc->direct_support is initialized
correctly, otherwise __glXExtensionBitIsEnabled() always returns FALSE

Specifically, this causes a problem with an X server which advertises
GLX<=1.2, and the GLX_SGIX_fbconfig extension.
glXGetFBConfigFromVisualSGIX() uses __glXExtensionBitIsEnabled() to
check if the GLX_SGIX_fbconfig extension is available, but that function
won't return correct information because that data has never been
initialized, because ext_list_first_time was never set...

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Brian Paul <brianp@vmware.com>
(cherry picked from commit 96ab4d2b84178209ee59017458d9964b32b7e183)

14 years agodocs: remove merge conflict garbage
Brian Paul [Sat, 6 Mar 2010 00:32:07 +0000 (17:32 -0700)]
docs: remove merge conflict garbage

14 years agometa: Use the DrawBuffer's stencil size
Ian Romanick [Thu, 4 Mar 2010 01:20:27 +0000 (17:20 -0800)]
meta: Use the DrawBuffer's stencil size

Previously the code was erroneously using the stencil size of the
context instead of the stencil size of the DrawBuffer.  With FBOs
these may be different.  As a result, clearing the stencil buffer of
an FBO bound to a context that doesn't have stencil would fail.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
14 years agoswrast: Remove redundant test of the visual's accumulation buffer size
Ian Romanick [Thu, 4 Mar 2010 01:17:56 +0000 (17:17 -0800)]
swrast: Remove redundant test of the visual's accumulation buffer size

If the visual doesn't have an accumulation buffer, the renderbuffer
passed into _swrast_clear_accum_buffer will be NULL anyway.  There is
no reason the check the visual.  Moreover, the test erroneously checks
the context's visual instead of the visual of the current DrawBuffer.
With FBOs these may be different.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
14 years agoswrast: Adjust colors based on ReadBuffer visual, not context visual
Ian Romanick [Thu, 4 Mar 2010 01:16:11 +0000 (17:16 -0800)]
swrast: Adjust colors based on ReadBuffer visual, not context visual

In the presence of FBOs, the visual of the context may not match the,
possibly fake, visual of the current ReadBuffer.  Note that the caller
of adjust_colors correctly uses the visual of the ReadBuffer.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
14 years agoswrast: Remove stray comment that mentions ctx->Visual
Ian Romanick [Thu, 4 Mar 2010 01:14:55 +0000 (17:14 -0800)]
swrast: Remove stray comment that mentions ctx->Visual

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
14 years agogallivm: checkpoint: code gen for mipmap selection
Brian Paul [Fri, 5 Mar 2010 23:52:11 +0000 (16:52 -0700)]
gallivm: checkpoint: code gen for mipmap selection

14 years agollvmpipe: added code to set texture depth, max mipmap levels info
Brian Paul [Fri, 5 Mar 2010 23:51:35 +0000 (16:51 -0700)]
llvmpipe: added code to set texture depth, max mipmap levels info

14 years agogallivm: added methods for getting texture depth, num mipmap levels
Brian Paul [Fri, 5 Mar 2010 23:50:48 +0000 (16:50 -0700)]
gallivm: added methods for getting texture depth, num mipmap levels

14 years agogallivm: checkpoint: texture LOD computation code gen
Brian Paul [Fri, 5 Mar 2010 23:07:54 +0000 (16:07 -0700)]
gallivm: checkpoint: texture LOD computation code gen

14 years agogallivm: added field for sampler lod
Brian Paul [Fri, 5 Mar 2010 23:07:26 +0000 (16:07 -0700)]
gallivm: added field for sampler lod

14 years agoi965: Only use the effective subset of the WM sampler key.
Eric Anholt [Fri, 5 Mar 2010 22:13:24 +0000 (14:13 -0800)]
i965: Only use the effective subset of the WM sampler key.

The whole structure is 836 bytes, but if only the first one or two
samplers are used (as is common), the part that matters is only 56 or
108 bytes.  By using just that subset as the key (since the key size
is also part of the key), we improve firefox-talos-gfx performance by
1%.

14 years agointel: Remove support for the DRI1 TFP extension.
Eric Anholt [Fri, 5 Mar 2010 21:48:41 +0000 (13:48 -0800)]
intel: Remove support for the DRI1 TFP extension.

14 years agoi965: Remove the paths for making a const surface object for a NULL BO.
Eric Anholt [Fri, 5 Mar 2010 21:38:02 +0000 (13:38 -0800)]
i965: Remove the paths for making a const surface object for a NULL BO.

14 years agoi965: Remove the unused SDC BO pointer from the sampler state key.
Eric Anholt [Fri, 5 Mar 2010 21:24:24 +0000 (13:24 -0800)]
i965: Remove the unused SDC BO pointer from the sampler state key.

Since it's a BO pointer, it already lives outside the key in
brw->wm.sdc_bo[] which is used for sampler state lookup and setup.

14 years agoi965: Rename SNB brw_*_key structs to gen6_*_key structs.
Eric Anholt [Fri, 5 Mar 2010 21:19:09 +0000 (13:19 -0800)]
i965: Rename SNB brw_*_key structs to gen6_*_key structs.

14 years agoi965: Pack gen6 blend_state_key structure.
Eric Anholt [Fri, 5 Mar 2010 21:18:13 +0000 (13:18 -0800)]
i965: Pack gen6 blend_state_key structure.

14 years agointel: Only do batchbuffer debug if --enable-debug is used.
Eric Anholt [Fri, 5 Mar 2010 20:56:40 +0000 (12:56 -0800)]
intel: Only do batchbuffer debug if --enable-debug is used.

This saves 6.6KB on the 965 driver, and appears to speed
firefox-talos-gfx up by 1-2%.  Unlike many other asserts in the
driver, when we make a mistake that would trigger one of these it
generally shows up all the time for developers, so turning it off for
release seems fine.

14 years agointel: Replace batch macro contents with function calls.
Eric Anholt [Fri, 5 Mar 2010 20:47:21 +0000 (12:47 -0800)]
intel: Replace batch macro contents with function calls.

This manages to cut down another 3800 bytes.

14 years agointel: Move the assertions about reloc delta from the macros to the function.
Eric Anholt [Fri, 5 Mar 2010 20:40:45 +0000 (12:40 -0800)]
intel: Move the assertions about reloc delta from the macros to the function.

Cuts another 1800 bytes from the driver.

14 years agointel: Check that the batch is mapped per BEGIN, not each OUT.
Eric Anholt [Fri, 5 Mar 2010 20:27:46 +0000 (12:27 -0800)]
intel: Check that the batch is mapped per BEGIN, not each OUT.

Shaves 800 bytes off the driver.

14 years agoi965: Try to assign SNB SF inputs to outputs.
Eric Anholt [Fri, 5 Mar 2010 20:27:25 +0000 (12:27 -0800)]
i965: Try to assign SNB SF inputs to outputs.

14 years agoi965: Fix the SNB SF input size.
Eric Anholt [Fri, 5 Mar 2010 02:26:33 +0000 (18:26 -0800)]
i965: Fix the SNB SF input size.

14 years agoAlways return VOLATILE for ObjectPurgeable(VOLATILE)
Chris Wilson [Fri, 5 Mar 2010 23:10:45 +0000 (23:10 +0000)]
Always return VOLATILE for ObjectPurgeable(VOLATILE)

Fixes fdo bug 26128.

The spec mandates that VOLATILE is returned from
ObjectPurgeable(VOLATILE) irrespective of the actual status of the
object upon completion of marking it purgeable.

Conform to the spec, even though it seems wrong.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
14 years agollvmpipe: add pipe_thread_wait() calls
Brian Paul [Fri, 5 Mar 2010 21:24:02 +0000 (14:24 -0700)]
llvmpipe: add pipe_thread_wait() calls

Wait for threads to exit before cleaning up per-thread data.
Fixes hang on context destruction with glean makeCurrent test.
See fd.o bug 26536.

14 years agogallium/docs: document additional wrap mode restrictions
Brian Paul [Fri, 5 Mar 2010 20:30:24 +0000 (13:30 -0700)]
gallium/docs: document additional wrap mode restrictions

14 years agogallivm: minor arithmetic improvements
Brian Paul [Fri, 5 Mar 2010 20:27:51 +0000 (13:27 -0700)]
gallivm: minor arithmetic improvements

14 years agogallivm: implement non-normalized texture wrap modes
Brian Paul [Fri, 5 Mar 2010 20:21:56 +0000 (13:21 -0700)]
gallivm: implement non-normalized texture wrap modes

Note that only the PIPE_TEX_WRAP_CLAMP,CLAMP_TO_EDGE,CLAMP_TO_BORDER
modes work with non-normalized texcoords.

14 years agosoftpipe: implement PIPE_TEX_WRAP_CLAMP_TO_EDGE for unorm texture sampling
Brian Paul [Fri, 5 Mar 2010 19:38:05 +0000 (12:38 -0700)]
softpipe: implement PIPE_TEX_WRAP_CLAMP_TO_EDGE for unorm texture sampling

CLAMP_TO_BORDER and CLAMP_TO_EDGE were doing the same thing.

14 years agomesa: minor reformatting, new comments
Brian Paul [Fri, 5 Mar 2010 19:32:32 +0000 (12:32 -0700)]
mesa: minor reformatting, new comments

14 years agoprogs/xdemos: fix up GLX extension checking
Jesse Barnes [Fri, 5 Mar 2010 19:21:12 +0000 (11:21 -0800)]
progs/xdemos: fix up GLX extension checking

A few of my programs were using the client string in addition to the
main extensions string to see if a feature was present.  Correct this to
only check for the currently available and working extensions.

14 years agoDRI2: expose OML_sync_control if possible
Jesse Barnes [Fri, 5 Mar 2010 19:17:19 +0000 (11:17 -0800)]
DRI2: expose OML_sync_control if possible

If the server supports the OML related protocol, enable support for the
extension.

14 years agoGLX: remove debug prints from glXWaitForMscOML dispatch
Jesse Barnes [Fri, 5 Mar 2010 18:31:39 +0000 (10:31 -0800)]
GLX: remove debug prints from glXWaitForMscOML dispatch

Leftover from earlier commit.

14 years agoxdemos/omlsync: add waitformsc and swap interval code
Jesse Barnes [Fri, 5 Mar 2010 18:28:41 +0000 (10:28 -0800)]
xdemos/omlsync: add waitformsc and swap interval code

14 years agoprogs: Add arbocclude2 demo.
Michal Krol [Fri, 5 Mar 2010 17:42:42 +0000 (18:42 +0100)]
progs: Add arbocclude2 demo.

14 years agogallium/docs: updated/improve sampler state documentation
Brian Paul [Fri, 5 Mar 2010 16:53:37 +0000 (09:53 -0700)]
gallium/docs: updated/improve sampler state documentation

14 years agodocs: document GL_APPLE_object_purgeable
Brian Paul [Fri, 5 Mar 2010 16:24:15 +0000 (09:24 -0700)]
docs: document GL_APPLE_object_purgeable

14 years agomesa: whitespace fixes, 80-column wrapping, etc.
Brian Paul [Fri, 5 Mar 2010 16:23:43 +0000 (09:23 -0700)]
mesa: whitespace fixes, 80-column wrapping, etc.

14 years agoMerge branch 'object-purgeable'
Chris Wilson [Fri, 5 Mar 2010 15:03:51 +0000 (15:03 +0000)]
Merge branch 'object-purgeable'

Acked-by: Brian Paul <brianp@vmware.com>
14 years agoAPPLE_object_purgeable: intel
Chris Wilson [Wed, 13 Jan 2010 18:35:25 +0000 (18:35 +0000)]
APPLE_object_purgeable: intel

Implement support for purgeable objects by using the GEM madvise ioctl.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
14 years agoAPPLE_object_purgeable: core
Chris Wilson [Fri, 13 Nov 2009 12:19:35 +0000 (12:19 +0000)]
APPLE_object_purgeable: core

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
14 years agomesa/es: Update GLAPI and APISpec for GL_OES_EGL_image.
Chia-I Wu [Thu, 25 Feb 2010 08:37:04 +0000 (16:37 +0800)]
mesa/es: Update GLAPI and APISpec for GL_OES_EGL_image.

This generates the entrypoints and dispatches for GL_OES_EGL_image.
There is no real support yet.

14 years agost/egl: Add support for EGL_KHR_image_base and EGL_KHR_image_pixmap.
Chia-I Wu [Fri, 5 Mar 2010 03:18:16 +0000 (11:18 +0800)]
st/egl: Add support for EGL_KHR_image_base and EGL_KHR_image_pixmap.

These extensions are not quite useful until the client APIs gain support
for the respective EGLImage extensions.

14 years agost/egl: Add get_param to native display interface.
Chia-I Wu [Fri, 5 Mar 2010 03:00:15 +0000 (11:00 +0800)]
st/egl: Add get_param to native display interface.

get_param can be used to query the parameters of a native display.
There is only NATIVE_PARAM_USE_NATIVE_BUFFER right now.  It queries
whether the window/pixmap surfaces use the native buffers instead of
private buffers.

14 years agost/egl: Move modeset interface to native_modeset.h.
Chia-I Wu [Fri, 5 Mar 2010 02:40:15 +0000 (10:40 +0800)]
st/egl: Move modeset interface to native_modeset.h.

native.h is getting more and more complex.  Move the independent modeset
interface to native_modeset.h to simplify native.h a little.

14 years agost/egl: Add event support to the native display interface.
Chia-I Wu [Sun, 21 Feb 2010 02:58:22 +0000 (10:58 +0800)]
st/egl: Add event support to the native display interface.

There is only invalid_surface event right now.  When EGL receives the
event, it sets the force_validate flag of the context binding to the
surface.  This helps skip an unnecessary check.

14 years agoRevert "mesa: Remove pointless comparison of unsigned integer with a negative constant."
Vinson Lee [Fri, 5 Mar 2010 01:40:36 +0000 (17:40 -0800)]
Revert "mesa: Remove pointless comparison of unsigned integer with a negative constant."

This reverts commit a05fdbcb719ac64e6be842372813f0f4ca2f4f93.

Removing the comparison is wrong. The comparison with -1 should be changed
to another value (probably PROGRAM_UNDEFINED) along with another
change in the shader assembler.

Conflicts:

src/mesa/shader/prog_execute.c

14 years agointel: Remove non-kernel-exec-fencing support.
Eric Anholt [Thu, 4 Mar 2010 23:47:19 +0000 (15:47 -0800)]
intel: Remove non-kernel-exec-fencing support.

Shaves 60k off the driver from removing the broken spans code.  This
means we now require 2.6.29, which seems fair given that it's a year
old and we've removed support for non-KMS already in the last release
of 2D.

14 years agointel: Remove the unused s8 spans code. Not hit during no_rast piglit.
Eric Anholt [Thu, 4 Mar 2010 23:33:21 +0000 (15:33 -0800)]
intel: Remove the unused s8 spans code.  Not hit during no_rast piglit.

Shaves 5.5k off of the driver.

14 years agoFix comparison of unsigned value against < 0.
Michel Dänzer [Thu, 4 Mar 2010 23:15:40 +0000 (00:15 +0100)]
Fix comparison of unsigned value against < 0.

14 years agosoftpipe: minor tweaks to texture filtering code
Brian Paul [Thu, 4 Mar 2010 22:53:41 +0000 (15:53 -0700)]
softpipe: minor tweaks to texture filtering code

14 years agogallivm: implement texture border color, plus tweaks to some wrap modes
Brian Paul [Thu, 4 Mar 2010 22:50:19 +0000 (15:50 -0700)]
gallivm: implement texture border color, plus tweaks to some wrap modes

The progs/test/texwrap demo looks pretty good, but there are still some
tiny differences from softpipe.  There may be a sub-pixel texcoord
interpolation error somewhere.

There's some room for optimization.  Many of the wrap modes compute
intermediate values that are constant for the texture size (see the
min/max values).  These could be computed earlier and stored somewhere
for later use.

14 years agogallivm: added texture border_color field
Brian Paul [Thu, 4 Mar 2010 22:52:01 +0000 (15:52 -0700)]
gallivm: added texture border_color field

14 years agogallivm: added question/comment
Brian Paul [Thu, 4 Mar 2010 22:52:19 +0000 (15:52 -0700)]
gallivm: added question/comment

14 years agollvmpipe: new comment
Brian Paul [Thu, 4 Mar 2010 22:52:37 +0000 (15:52 -0700)]
llvmpipe: new comment

14 years agogallivm: comments
Brian Paul [Thu, 4 Mar 2010 22:49:39 +0000 (15:49 -0700)]
gallivm: comments

14 years agodri/nouveau: Remove some CI remnants.
Francisco Jerez [Thu, 4 Mar 2010 22:19:49 +0000 (23:19 +0100)]
dri/nouveau: Remove some CI remnants.

14 years agodri/nouveau: Trivially add GL_ARB_texture_mirrored_repeat.
Andrew Randrianasulu [Thu, 4 Mar 2010 22:15:47 +0000 (23:15 +0100)]
dri/nouveau: Trivially add GL_ARB_texture_mirrored_repeat.

Signed-off-by: Francisco Jerez <currojerez@riseup.net>
14 years agodri/nouveau: Enable GL_EXT_stencil_wrap.
Andrew Randrianasulu [Mon, 1 Mar 2010 23:45:50 +0000 (23:45 +0000)]
dri/nouveau: Enable GL_EXT_stencil_wrap.

Signed-off-by: Francisco Jerez <currojerez@riseup.net>
14 years agodri/nouveau: Add GL_EXT_stencil_wrap for nv04.
Andrew Randrianasulu [Mon, 1 Mar 2010 23:58:17 +0000 (23:58 +0000)]
dri/nouveau: Add GL_EXT_stencil_wrap for nv04.

Signed-off-by: Francisco Jerez <currojerez@riseup.net>
14 years agodri/nouveau: Fix rb->DataType for z24s8 renderbuffers.
Francisco Jerez [Thu, 4 Mar 2010 20:40:47 +0000 (21:40 +0100)]
dri/nouveau: Fix rb->DataType for z24s8 renderbuffers.

14 years agodri/nouveau: Try to keep client buffers smaller than the scratch VBO length.
Francisco Jerez [Thu, 4 Mar 2010 20:32:07 +0000 (21:32 +0100)]
dri/nouveau: Try to keep client buffers smaller than the scratch VBO length.

14 years agodri/nouveau: Pack client arrays as they're copied to the real BO.
Francisco Jerez [Thu, 4 Mar 2010 20:27:11 +0000 (21:27 +0100)]
dri/nouveau: Pack client arrays as they're copied to the real BO.

14 years agoxdemos: add OML sync test
Jesse Barnes [Thu, 4 Mar 2010 21:14:59 +0000 (13:14 -0800)]
xdemos: add OML sync test

Add program to test glXSwapBuffersMscOML behavior.  Based on glsync.c.

14 years agor300g: Fix indexbuf upper limits.
Corbin Simpson [Thu, 4 Mar 2010 20:46:20 +0000 (12:46 -0800)]
r300g: Fix indexbuf upper limits.

Wine tends to pessimistically use ~0 for its max index, but r300s
only can go up to 2^24-1, causing the kernel checker to freak out.

Civ4 is marginally improved now. Still crashes, but not as bad.

14 years agoAPPLE_object_purgeable: autogenerated files
Chris Wilson [Thu, 12 Nov 2009 10:15:49 +0000 (10:15 +0000)]
APPLE_object_purgeable: autogenerated files

14 years agoAPPLE_object_purgeable: xml
Chris Wilson [Thu, 12 Nov 2009 10:12:56 +0000 (10:12 +0000)]
APPLE_object_purgeable: xml

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
14 years agogallivm: rework and implement more texture wrap modes
Brian Paul [Thu, 4 Mar 2010 17:51:31 +0000 (10:51 -0700)]
gallivm: rework and implement more texture wrap modes

All the texture wrap modes are now implemented for linear and nearest
sampling.

However, texture border color is not yet supported so some wrap modes
will produce int coords outside the texture bounds.  We'll get garbage
values insted of the texture border color for now.

14 years agogallivm: added lp_build_fract()
Brian Paul [Thu, 4 Mar 2010 17:50:26 +0000 (10:50 -0700)]
gallivm: added lp_build_fract()

14 years agogallivm: added lp_build_set_sign()
Brian Paul [Thu, 4 Mar 2010 16:45:34 +0000 (09:45 -0700)]
gallivm: added lp_build_set_sign()

14 years agogallivm: added lp_build_negate()
Brian Paul [Thu, 4 Mar 2010 15:36:26 +0000 (08:36 -0700)]
gallivm: added lp_build_negate()

14 years agogallivm: clarify unsigned vs. signed integer type construction
Brian Paul [Sat, 27 Feb 2010 00:44:10 +0000 (17:44 -0700)]
gallivm: clarify unsigned vs. signed integer type construction

The lp_int_type() function was creating an unsigned type.  So rename
that function to lp_uint_type() and create new lp_int_type() that
creates a signed type.

14 years agoi915: Enable texture tiling by default.
Eric Anholt [Wed, 3 Mar 2010 19:51:51 +0000 (11:51 -0800)]
i915: Enable texture tiling by default.

This now passes piglit testcases, and shows performance improvements
on openarena.  Blit-heavy apps may see degradation, but the
expectation is that the common case of 3D rendering will see
significant wins.

This regresses gen-teximage on my 855, and no piglit tests on my 945.

14 years agoi915: Fix up i830 for tiled drawing offsets.
Eric Anholt [Wed, 3 Mar 2010 19:49:39 +0000 (11:49 -0800)]
i915: Fix up i830 for tiled drawing offsets.

Corresponds to b87406e55f029d29594ae76a4b39a4fe1007fe4f.

14 years agofix getprocaddress test
George Sapountzis [Thu, 4 Mar 2010 14:33:39 +0000 (16:33 +0200)]
fix getprocaddress test

14 years agoDRI: Put back missing break-statement
Ian Romanick [Thu, 4 Mar 2010 09:56:20 +0000 (01:56 -0800)]
DRI: Put back missing break-statement

The break in the __DRI_ATTRIB_RENDER_TYPE case was accidentally
removed in commit 5cf2c5851bcd29c2d53bb04ab692b4b156f5a74d.  This puts
it back.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
14 years agoRevert "mesa: Fix unsigned comparison."
Vinson Lee [Thu, 4 Mar 2010 09:51:50 +0000 (01:51 -0800)]
Revert "mesa: Fix unsigned comparison."

This reverts commit f9504e75f02586a8561733e0e2711c65efa2979d.

This patch is incorrect.

14 years agoglsl: Remove unsigned greater than or equal zero comparison.
Vinson Lee [Thu, 4 Mar 2010 09:35:28 +0000 (01:35 -0800)]
glsl: Remove unsigned greater than or equal zero comparison.

14 years agomesa: Fix unsigned comparison.
Vinson Lee [Thu, 4 Mar 2010 09:24:44 +0000 (01:24 -0800)]
mesa: Fix unsigned comparison.

14 years agoglsl: Remove unsigned greater than or equal zero comparison.
Vinson Lee [Thu, 4 Mar 2010 09:10:44 +0000 (01:10 -0800)]
glsl: Remove unsigned greater than or equal zero comparison.

14 years agoglsl: Return NULL on _slan_gen_asm error path on non-debug builds.
Vinson Lee [Thu, 4 Mar 2010 08:58:36 +0000 (00:58 -0800)]
glsl: Return NULL on _slan_gen_asm error path on non-debug builds.

Exit the function early on the error path, instead of continuing, to
prevent a null pointer dereference later on.

14 years agoglsl: Add assert to check input to strcmp.
Vinson Lee [Thu, 4 Mar 2010 08:53:52 +0000 (00:53 -0800)]
glsl: Add assert to check input to strcmp.

14 years agomesa: Add asserts to check inputs to memcpy.
Vinson Lee [Thu, 4 Mar 2010 08:41:21 +0000 (00:41 -0800)]
mesa: Add asserts to check inputs to memcpy.