build: Replace OSMESA_VERSION with generic VERSION_NUMBER

Can be used by other modules.

build: Order AC_CONFIG_FILES list

Makefiles before .pc files before directories. Alphabetize files of the
same type.

build: Only build libmesa.la when needed

Namely, for Xlib-GLX, OSMesa, or test programs.

build: Remove duplicate DRI automake conditionals

build: Remove GLU_DIRS

build: Only generate dispatch assembly code that will be built

i965: don't clear resolve map when doing fast depth clears.

Previously, when performing a fast depth clear, we would also clear
the miptree's resolve map.  This destroyed important information,
since the resolve map contains information about needed resolves for
all levels and layers of the miptree, whereas a depth clear only
applies to a single level/layer combination at a time.  As a result,
resolves would sometimes fail to occur, leading to incorrect
rendering.

Fixes rendering artifacts with shadow maps in Unigine Heaven and
Unigine Sanctuary.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=50270

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>

i965/HiZ: remove assertion from intel_resolve_map_set().

There are three possible resolve map states for each (level, layer) of
a depth miptree: "needs HiZ resolve", "needs depth resolve", and
"needs neither".  When HiZ was first implemented on i965, any attempt
to directly transition between "needs HiZ resolve" and "needs depth
resolve" without passing through the "needs neither" state would have
been a bug indicating that a necessary resolve hadn't been performed.
Accordingly, intel_resolve_map_set() contained an assertion to verify
that no such direct transition happened.

However, now that we support fast depth clears, there is a valid
transition from the "needs HiZ resolve" to the "needs depth resolve"
state.  When doing a fast depth clear, the old state of the buffer is
irrelevant, since we are completely replacing it with the clear value,
so it is not necessary to do any resolves before clearing--we can
transition, if necessary, directly from the "needs HiZ resolve" state
to the "needs depth resolve" state.

To avoid spurious assertions in this valid case, this patch just
removes the assertion.

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>

radeonsi: remove old tilling handling

Just use the functionality provided by the surface manager instead.

This fixes just another bunch of piglit tests.

Signed-off-by: Christian König <deathsimple@vodafone.de>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

mesa/es: Validate glCreateShader targets in Mesa code rather than the ES wrapper

v2: Add proper core-profile filtering.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

mesa/es: Validate glGetProgramiv pnames in Mesa code rather than the ES wrapper

v2: Add proper core-profile and GLES3 filtering.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

mesa: Filter glGetProgramiv pnames based on available extensions

Previously you could always glGetProgramiv one of the transform feedback
or geometry shader enums even if the extension wasn't supported.

In addtion, this reverts part of bda6ad27.  I think the hunks involving
GL_PROGRAM_BINARY_LENGTH_OES were spurious.  Mesa has no support for any
other part of GL_OES_get_program_binary.

v2: Remove redundant return in get_programiv based on review feedback
from Matt Turner.

v3: Correctly handle UBO related enums.

v4: Emit the bad enum in the _mesa_error call based on review feedback
from Brian Paul.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

swrast: implement cubical depth texture sampling

Fixes a few more failures in the piglit copyteximage test.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

clover: Accept CL_MEM_READ_WRITE flag

Fix API functions for memory objects to accept CL_MEM_READ_WRITE flag.

Signed-off-by: Blaž Tomažič <blaz.tomazic@gmail.com>
[ Francisco Jerez: Drop incorrect change in clCreateSubBuffer. ]

radeon/llvm: Cleanup R600Instructions.td

main: fix ES compile breakage

mesa/swrast: fix GL_TEXTURE_2D_ARRAY texture fetches for dxt formats

As with the previous commit.

This fixes the last crash in the piglit copyteximage test but there's
still some failures.

mesa/swrast: fix GL_TEXTURE_2D_ARRAY texture fetches for latc/rgtc formats

Fix-up the texel fetch functions so that they handle 3D coords (as used for
array textures) and remove the "f_2d" part from their names.

Helps fix swrast crashes in piglit's copyteximage test.  More to come.

mesa: code movement in teximage.c

To get rid of a forward declaration.

mesa: consolidate glTexImage and glCompressedTexImage code

There was a lot of similar or duplicated code before.
To minimize this patch's size, use a forward declaration for
compressed_texture_error_check().  Move the function in the next patch.

mesa: make glTexImage, glCompressedTexImage proxy code more alike

Next up, we can combine the teximage() and compressed_teximage() functions.

mesa: rename texpal.[ch] to texcompress_cpal.[ch]

To be consistent with other files related to texture compression.

mesa: s/GLuint/gl_format/ in _mesa_compressed_format_to_glenum()

No real change here, just use the right type.

mesa: new _mesa_num_tex_faces() helper

Not a real big help now, but will be useful for the
GL_ARB_texture_cube_map_array extension in the future.

mesa: make _mesa_get_proxy_tex_image() static

It's not used by any other file.

mesa: don't clear proxy image fields when regular GL error is generated

If a proxy texture call generates a regular GL error, we should not
clear the proxy image's width/height/depth/format fields.  Use a new
PROXY_ERROR token to distinguish proxy errors from regular GL errors.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>

mesa: fix glTexImage proxy texture error generation

When calling glTexImage() with a proxy target most error conditions should
generate a GL error.  We were erroneously doing the proxy-error behaviour
(where we zeroed-out the image's width/height/depth/format fields) in too
many places.

There's another issue with proxy textures, but that'll be fixed in the
next patch.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>

draw: Fix regression in draw_set_sampler(_views).

draw->samplers(_views) now has PIPE_SHADER_TYPES elements, instead of
PIPE_MAX_SAMPLERS as before.

Also, shader_stage must be less than PIPE_SHADER_TYPES to prevent buffer
overflow.

Trivial.

build: don't leave git_sha1.h.tmp after build/install

Fixes "`main/git_sha1.h.tmp': Permission denied" build error.
See https://bugs.freedesktop.org/show_bug.cgi?id=52064

Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>

radeon/llvm: Set End of Program bit on RAT instructions

This code was accidently dropped during the MCCodeEmitter conversion.

radeon/llvm: Use correct instruction for moving immediates

This should fix an assertion failure that was happening in some compute
shaders.

radeon/llvm: Fix some coding style issues

radeon/llvm: Pull changes from external version of the backend

radeon/llvm: Simplify the convert to ISA pass

radeon/llvm: Make sure to use the Text section in the AsmPrinter

build: Fix installation of GLES2 headers

Reported-by: U. Artie Eoff <ullysses.a.eoff@intel.com>
Tested-by: U. Artie Eoff <ullysses.a.eoff@intel.com>

build: Fix GLES linkage with libglapi

Reported-by: Ian Romanick <idr@freedesktop.org>

i965/msaa: Add sample-alpha-to-coverage support for multiple render targets

Render Target Write message should include source zero alpha value when
sample-alpha-to-coverage is enabled for an FBO with  multiple render targets.
Source zero alpha value is used as fragment coverage for all the render
targets.

This patch makes piglit tests draw-buffers-alpha-to-coverage and
alpha-to-coverage-no-draw-buffer-zero to pass on Sandybridge. No
regressions are observed with piglit all.tests.

V2: Revert all the changes made in emit_color_write() function to
include src0 alpha for targets > 0. Now handling this case in a if
block.

V3: Correctly calculate the instruction length for buffer zero.
Properly handle the case of dual_src_blend when alpha-to-coverage
is enabled.

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Paul Berry <stereotype441@gmail.com>

glsl/linker: Avoid buffer over-run in parcel_out_uniform_storage::visit_field

When too may uniforms are used, the error will be caught in
check_resources (src/glsl/linker.cpp).

NOTE: This is a candidate for the 8.0 branch.

Signed-off-by: Stéphane Marchesin <marcheu@chromium.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Tested-by: Benoit Jacob <bjacob@mozilla.com>

mesa/es: Validate glCompressedTexSubImage internalFormat in Mesa code rather than the ES wrapper

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Validate glCompressedTexImage internalFormat in Mesa code rather than the ES wrapper

v2: Add proper core-profile filtering.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Validate glCopyTexImage internalFormat in Mesa code rather than the ES wrapper

v2: Add GLES3 filtering.  I'm not 100% sure this is correct.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Validate glTexSubImage format and type in Mesa code rather than the ES wrapper

v2: Add proper GLES3 filtering.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Validate glTexImage format, type, and internalFormat in Mesa code rather than the ES wrapper

v2: Add proper GLES3 filtering.

v3: Collapse ALPHA, LUMINANCE, and LUMINANCE_ALPHA cases per review
comment from Ken.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Validate glTexImage border in Mesa code rather than the ES wrapper

Also validate glCopyTexImage border.  This fixes a bug in the APIspec.
Previously glTexImage3DOES could be passed a non-zero border without error.

NOTE: This is a candidate for stable release branches.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa: Generate an error when glCopyTexImage border is invalid

NOTE: This is a candidate for stable release branches.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Add support for GL_APPLE_texture_max_level

This is desktop OpenGL functionality that has always existed.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Validate glGetTexParameter pnames in Mesa code rather than the ES wrapper

This also adds a missing extension (and API) check around
GL_TEXTURE_CROP_RECT_OES.

v2: Add proper core-profile and GLES3 filtering.  GL_TEXTURE_MAX_LEVEL
is (incorrectly) accepted in ES contexts.  A future patch will add
GL_APPLE_texture_max_level, and meta really needs this.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Validate glTexParameter pnames in Mesa code rather than the ES wrapper

This also adds a missing extension (and API) check around
GL_TEXTURE_CROP_RECT_OES.

v2: Add proper core-profile, GLES1, and GLES3 filtering.  GL_TEXTURE_MAX_LEVEL
is (incorrectly) accepted in ES contexts.  A future patch will add
GL_APPLE_texture_max_level, and meta really needs this.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Remove redundant glBindTexture target validation

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa: Filter glBindTexture targets based on supported features.

Fixed the piglit test arb_texture_buffer_object-negative-unsupported.

NOTE: This is a candidate for stable release branches.

v2: Add proper core-profile and GLES3 filtering.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Validate tex image targets in Mesa code rather than the ES wrapper

This should take care of all the TexImage, TexSubImage, CopyTexImage,
CompressedTexImage3DOES, and CopyTexSubImage type paths.

v2: Add proper core-profile and GLES3 filtering.

v3: Squash the CompressedTexImage3DOES patch per review comment from
Ken.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Validate EGLImageTargetTexture2DOES target in Mesa code rather than the ES wrapper

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Validate glTexParameter targets in Mesa code rather than the ES wrapper

Ditto for glGetTexParameter targets.

v2: Add proper core-profile and GLES3 filtering.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa/es: Validate GL_TEXTURE_WRAP param in Mesa code rather than the ES wrapper

v2: Add proper core-profile filtering.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

mesa: Refactor validate_texture_wrap_mode to use a switch-statement

This makes the next couple changes a little easier.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>

meta: Don't modify GL_GENERATE_MIPMAP state when it doesn't exist

This is a bit of a hack.  _mesa_meta_GenerateMipmap shouldn't even be
used in contexts where GL_GENERATE_MIPMAP doesn't exist (i.e., core
profile and ES2) because it uses fixed-function, and fixed-function
doesn't exist there either!

A GLSL-based _mesa_meta_GenerateMipmap should be available soon.  When
that is available, this patch will be irrelevant and should be reverted.

v2: Change (ctx->API != API_OPENGLES2 && ctx->API != API_OPENGL_CORE) to
(ctx->API == API_OPENGL || ctx->API == API_OPENGLES) based on review
comment from Brian Paul.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

build/glsl: fix android build v2

Commit 77a3efc6b907943903190b385fdf107c4acfcdca broke android build that
sets its own value for GLSL_SRCDIR before including Makefile.sources.
Patch moves overriding the value after include, this works as GLSL_SRCDIR
variable gets expanded only later.

Reviewed-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>

automake: convert es1api

automake: convert es2api

st/dri: pass config options to the state tracker

Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>

st/mesa: accept and handle configuration options from st/dri

Currently there is a single option - force_glsl_extensions_warn.

Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>

st/dri: add force_glsl_extensions_warn option to dri options

Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>

st/dri: use driver name for driconf section lookup

The name is taken from the driver_descriptor, so it will be the same as
expected by driconf utility.

Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>

swrast: add DRM_DRIVER_DESCRIPTOR to store driver name

egl_dri2: Fix segmentation fault

The segmentation fault occurs when DRI2 is not loaded up and
dri2_setup_screen() function deferences dri2_dpy->dri2 (since it's NULL
at this point).

This patch fixes the segmentation fault by checking if dri2 pointer is
not NULL before deferencing it.

Signed-off-by: Paulo Alcantara <pcacjr@profusion.mobi>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Brian Paul <brianp@vmware.com>

radeon/llvm: Use the MCCodeEmitter for R600

radeon/llvm: Use the MCCodeEmitter for SI

radeon/llvm: Set 64BitPtr feature bit for SI

radeon/llvm: Lower RETFLAG DAG Node to S_ENDPGM on SI

radeon/llvm: Add AsmPrinter

radeon/llvm: Mark JUMP as a pseudo instruction

radeon/llvm: Remove the last uses of MachineOperand flags

radeon/llvm: Add flag operand to some instructions

This new operand replaces the MachineOperand flags in LLVM, which
will be deprecated soon.  Eventually all instructions should have a flag
operand, but for now this operand has only been added to instructions
that need it.

radeon/llvm: Encapsulate setting of MachineOperand flags

MachineOperand flags will be removed soon, so it is convienent to
have only one function that modifies them.

build: Link DRI drivers with dricore in case of no direct rendering

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

build: Only build libmesagallium.la if building Gallium

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

build: Clean glx Makefile.am

mapi/glapi is already built when make is run in src/glx.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

build: Put mapi/shared-glapi in CORE_DIRS

SRC_DIRS was overwritten (visible in the second hunk).

Also don't require mapi/shared-glapi to be built for GLES.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

build: Only allow shared-glapi with DRI

Tested-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

build: Set sensible DRI/X11/OSMesa defaults

Tested-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

build: Print whether shared-glapi is enabled

Tested-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

build/x11: Force usage of C++ linker

Tested-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

build/x11: Don't link against shared-glapi

Tested-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

build: Remove deprecated --with-driver= flag

Tested-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

radeonsi: rework vertex format handling

Preventing piglit's draw-vertices test from hanging the GPU.

Signed-off-by: Christian König <deathsimple@vodafone.de>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

radeonsi: fix SPI_PS_INPUT_ENA handling

We need to enable at least one interpolation mode,
otherwise the GPU will hang.

Signed-off-by: Christian König <deathsimple@vodafone.de>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>

r600g: fix lockups with dual_src_blend v2

Disable blending when dual_src_blend is enabled and number of color exports
in the current fragment shader is less than 2.

Fixes lockups with ext_framebuffer_multisample-
alpha-to-coverage-dual-src-blend piglit test.

Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>

st/dri: Add shared usage on buffers created

Tested-by: Scott Moreau <oreaus@gmail.com>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

gbm: Add shared usage on images created

Tested-by: Scott Moreau <oreaus@gmail.com>
Signed-off-by: Jakob Bornecrantz <jakob@vmware.com>

mesa: Fix generic compressed texture formats' handling in glTexImage/glCopyTexImage

The generic texture formats should be accepted by the <internalformat>
parameter of TexImage1D, TexImage2D, TexImage3D, CopyTexImage1D, and
CopyTexImage2D functions. When the application specifies a generic
format, the driver is free to pick an uncompressed format.

This patch reverts the changes due to following commit:
commit a36581ccc06693231011c3fe136207e73191b1ce
mesa: do more teximage error checking for generic compressed formats

This patch fixes compressed texture format failures in intel oglconform
pxconv-gettex test case:
https://bugs.freedesktop.org/show_bug.cgi?id=47220

Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Brian Paul <brianp@vmware.com>

radeon/llvm: ExpandSpecialInstrs - Add support for cube instructions

radeon/llvm: ExpandSpecialInstrs - Add support for vector instructions

radeon/llvm: Add R600ExpandSpecialInstrs pass

This pass expends reduction instructions into a MachineInstrBundle that
contains 4 instruction, one for each instruction slot.

radeon/llvm: Add helper function for getting sub reg indices

radeonsi: Handle NULL sampler views getting passed in by the state tracker.

Don't dereference NULL pointers, and if all views are NULL, don't generate an
invalid PM4 packet which locks up the GPU.

Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

APIspec: Remove cruft about AMD_compressed_???_texture

Mesa doesn't support these extensions, and it seems unlikely that it
ever will

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa/es: Remove redundant glFramebufferTexture3D textarget validation

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa/es: Remove redundant glGetShaderiv pname validation

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Brian Paul <brianp@vmware.com>

mesa/es: Remove redundant glCompressedTexImage border validation

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Brian Paul <brianp@vmware.com>