mesa.git
15 years agollvmpipe: avoid flushing depth buffer cache on swapbuffers
Keith Whitwell [Thu, 23 Jul 2009 10:14:39 +0000 (11:14 +0100)]
llvmpipe: avoid flushing depth buffer cache on swapbuffers

There's no need to push out depth buffer contents on swapbuffers.

Note that this change doesn't throw away depth buffer changes, it simply
holds them in the cache over calls to swapbuffers.  The hope is
that swapbuffers will be followed by a clear() which means in that case
we won't have to write the changes out.

15 years agollvmpipe: also shortcircuit non-texture tile lookups
Keith Whitwell [Wed, 22 Jul 2009 14:36:25 +0000 (15:36 +0100)]
llvmpipe: also shortcircuit non-texture tile lookups

15 years agollvmpipe: shortcircuit repeated lookups of the same tile
Keith Whitwell [Wed, 22 Jul 2009 14:08:42 +0000 (15:08 +0100)]
llvmpipe: shortcircuit repeated lookups of the same tile

The lp_tile_cache is often called repeatedly to look up the same
tile.  Add a cache (to the cache) of the single tile most recently
retreived and make a quick inline check to see if this matches the
subsequent request.

Add a tile_address bitfield struct to make this check easier.

15 years agollvmpipe: remove unused vars in lp_setup.c
Keith Whitwell [Fri, 17 Jul 2009 11:12:04 +0000 (12:12 +0100)]
llvmpipe: remove unused vars in lp_setup.c

15 years agollvmpipe: use bitwise logic to setup quad masks in lp_setup
Keith Whitwell [Fri, 17 Jul 2009 11:03:51 +0000 (12:03 +0100)]
llvmpipe: use bitwise logic to setup quad masks in lp_setup

15 years agollvmpipe: simplify flush_spans
Keith Whitwell [Fri, 17 Jul 2009 09:47:32 +0000 (10:47 +0100)]
llvmpipe: simplify flush_spans

No loss of performance, but simpler code.

15 years agollvmpipe: make some small steps to flush texture cache less frequently
Keith Whitwell [Fri, 17 Jul 2009 09:44:22 +0000 (10:44 +0100)]
llvmpipe: make some small steps to flush texture cache less frequently

No performance gain yet, but the code is a bit cleaner.

15 years agollvmpipe: remove backwards dependency from tilecache to llvmpipe
Keith Whitwell [Thu, 16 Jul 2009 13:14:32 +0000 (14:14 +0100)]
llvmpipe: remove backwards dependency from tilecache to llvmpipe

The tile cache is a utility, it shouldn't know anything about the
entity which is making use of it (ie llvmpipe).

Remove llvmpipe parameter to all the tilecache function calls, and
also remove the need to keep a llvmpipe pointer in the sampler structs.

15 years agollvmpipe: Rename preprocessor symbols too.
José Fonseca [Mon, 27 Jul 2009 10:36:24 +0000 (11:36 +0100)]
llvmpipe: Rename preprocessor symbols too.

15 years agollvmpipe: Pixel packing/unpacking and loop code generators.
José Fonseca [Mon, 27 Jul 2009 00:23:15 +0000 (01:23 +0100)]
llvmpipe: Pixel packing/unpacking and loop code generators.

Just a small proof of concept plus a standalone test app. Not integrated
with the rest of the driver yet.

15 years agoutil: Pixel format database.
José Fonseca [Sun, 26 Jul 2009 22:50:00 +0000 (23:50 +0100)]
util: Pixel format database.

There are some inconsistencies in pipe_format, but above all, there
simply aren't enough bits in an enum to conveniently store all
information about a pixel format we need to be able to dynamically
generate pixel packing/unpacking code.

15 years agollvmpipe: Fork softpipe for experimentation with llvm.
José Fonseca [Sun, 26 Jul 2009 22:44:38 +0000 (23:44 +0100)]
llvmpipe: Fork softpipe for experimentation with llvm.

15 years agoPut 'extern' first on the line to silence GCC warnings.
Ian Romanick [Sat, 29 Aug 2009 03:12:36 +0000 (20:12 -0700)]
Put 'extern' first on the line to silence GCC warnings.

15 years agoi965: Use VBOs in the VBO module on 965, now that we have ARB_map_buffer_range.
Eric Anholt [Fri, 28 Aug 2009 22:01:56 +0000 (15:01 -0700)]
i965: Use VBOs in the VBO module on 965, now that we have ARB_map_buffer_range.

This looks like it's a small win on blender.

15 years agovbo: Fix array pointer calculation for MapBufferRange-mapped vertex data.
Eric Anholt [Fri, 28 Aug 2009 21:44:55 +0000 (14:44 -0700)]
vbo: Fix array pointer calculation for MapBufferRange-mapped vertex data.

We would end up with the offset from the start of the mapping rather than
the offset from the start of the buffer.

15 years agointel: Add support for GL_ARB_map_buffer_range.
Eric Anholt [Thu, 27 Aug 2009 01:08:52 +0000 (18:08 -0700)]
intel: Add support for GL_ARB_map_buffer_range.

Passes glean's bufferObject test, and should provide good performance in the
cases applications are expected to use.

15 years agoi965: Implement ARB_oq CheckQuery in the intended way.
Eric Anholt [Fri, 28 Aug 2009 01:34:11 +0000 (18:34 -0700)]
i965: Implement ARB_oq CheckQuery in the intended way.

Previously we blocked because I hadn't added the libdrm function.  Now it's
there, so update your libdrm.

15 years agoradeon: Fix swtcl emit pediction.
Pauli Nieminen [Fri, 28 Aug 2009 21:46:08 +0000 (00:46 +0300)]
radeon: Fix swtcl emit pediction.

Problem was to find the correct place to run prediction. Only place that is
called for every primitive is ALLOC_VERTS so we have to do prediction there
before allocation.

15 years agoi915g: Add cursor case for scanout layout
Jakob Bornecrantz [Fri, 28 Aug 2009 21:39:29 +0000 (22:39 +0100)]
i915g: Add cursor case for scanout layout

15 years agoi915g: Don't forget to init buffer ptrs
Jakob Bornecrantz [Fri, 28 Aug 2009 21:33:42 +0000 (22:33 +0100)]
i915g: Don't forget to init buffer ptrs

15 years agost/xorg: disable composite accelration for now
Zack Rusin [Fri, 28 Aug 2009 20:19:53 +0000 (16:19 -0400)]
st/xorg: disable composite accelration for now

15 years agost/xorg: bind samplers before composition
Zack Rusin [Fri, 28 Aug 2009 19:43:46 +0000 (15:43 -0400)]
st/xorg: bind samplers before composition

15 years agost/xorg: normalize coords for the src operands in the vertex shader
Zack Rusin [Fri, 28 Aug 2009 19:36:05 +0000 (15:36 -0400)]
st/xorg: normalize coords for the src operands in the vertex shader

15 years agoradeon: Fix scissors for dri1.
Pauli Nieminen [Fri, 28 Aug 2009 17:41:15 +0000 (20:41 +0300)]
radeon: Fix scissors for dri1.

DRI1 has to set scissors in framebuffer coordinates.

15 years agor300g: pretty up the new drm code a bit
Joakim Sindholt [Fri, 28 Aug 2009 18:05:52 +0000 (20:05 +0200)]
r300g: pretty up the new drm code a bit

As it turns out, I made some boners. This corrects them (I hope)

15 years agost/xorg: If migrating from a sw pixmap to a hw pixmap upload old contents
Jakob Bornecrantz [Fri, 28 Aug 2009 14:56:59 +0000 (16:56 +0200)]
st/xorg: If migrating from a sw pixmap to a hw pixmap upload old contents

15 years agost/xorg: Null devPrivate.ptr when unmapping
Jakob Bornecrantz [Fri, 28 Aug 2009 14:56:02 +0000 (16:56 +0200)]
st/xorg: Null devPrivate.ptr when unmapping

15 years agost/xorg: Fix scons build.
Michel Dänzer [Fri, 28 Aug 2009 15:42:07 +0000 (17:42 +0200)]
st/xorg: Fix scons build.

15 years agor300g: support new drm_api
Joakim Sindholt [Fri, 28 Aug 2009 15:13:49 +0000 (17:13 +0200)]
r300g: support new drm_api

15 years agodrm_api: Operate on textures instead of buffers
Jakob Bornecrantz [Thu, 27 Aug 2009 17:57:29 +0000 (18:57 +0100)]
drm_api: Operate on textures instead of buffers

Most use cases just got the buffer from the texture
and then called into one of the get_handle functions.

Also with this patch it would be easier to move to a
generic function for getting handles from textures
and textures from handles, that is exposed via the screen.

15 years agost/xorg: Fix some warnings
Jakob Bornecrantz [Thu, 27 Aug 2009 16:56:30 +0000 (17:56 +0100)]
st/xorg: Fix some warnings

15 years agost/dri: Fix some warnings
Jakob Bornecrantz [Thu, 27 Aug 2009 16:38:33 +0000 (17:38 +0100)]
st/dri: Fix some warnings

15 years agost/xorg: Pass dri driver name as driverName on screen
Jakob Bornecrantz [Fri, 28 Aug 2009 10:32:04 +0000 (12:32 +0200)]
st/xorg: Pass dri driver name as driverName on screen

15 years agoRevert "st/xorg/dri: Pass texture formats via the DRI2 protocol flags."
Jakob Bornecrantz [Fri, 28 Aug 2009 09:43:19 +0000 (11:43 +0200)]
Revert "st/xorg/dri: Pass texture formats via the DRI2 protocol flags."

This reverts commit a33e9ea3b9b7ac98d6e9bc385bbbe7eac2eb73d0.

As this change breaks both radeon and nouveau

15 years agor100/r200: Bring back old PolygonStripple for DRI1.
Pauli Nieminen [Fri, 28 Aug 2009 02:42:41 +0000 (05:42 +0300)]
r100/r200: Bring back old PolygonStripple for DRI1.

DRI1 didn't have support for command buffer emit for stripple.

15 years agor100/r200: Share PolygonStripple code.
Pauli Nieminen [Fri, 28 Aug 2009 01:58:50 +0000 (04:58 +0300)]
r100/r200: Share PolygonStripple code.

15 years agor200: Fix piglit paths test.
Pauli Nieminen [Fri, 28 Aug 2009 01:12:17 +0000 (04:12 +0300)]
r200: Fix piglit paths test.

Polygon stipple has to go to command buffer because special ioctl is disabled.
Piglit doesn't like HyperZ warning so disable it for kms.

15 years agor600: fix scissor also.
Dave Airlie [Fri, 28 Aug 2009 01:05:29 +0000 (11:05 +1000)]
r600: fix scissor also.

I forgot about r600 being here.

15 years agoradeon: fix scissors harder.
Dave Airlie [Fri, 28 Aug 2009 01:01:38 +0000 (11:01 +1000)]
radeon: fix scissors harder.

this makes gnome-shell work on r300 for me

15 years agointel: Add support for ARB_copy_buffer.
Eric Anholt [Thu, 27 Aug 2009 23:53:50 +0000 (16:53 -0700)]
intel: Add support for ARB_copy_buffer.

Passes glean's bufferObject test for this extension.

15 years agor300: Fix emit prediction to account scissor emitting correctly.
Pauli Nieminen [Thu, 27 Aug 2009 23:04:00 +0000 (02:04 +0300)]
r300: Fix emit prediction to account scissor emitting correctly.

15 years agoMerge branch 'mesa_7_5_branch'
Brian Paul [Thu, 27 Aug 2009 22:54:32 +0000 (16:54 -0600)]
Merge branch 'mesa_7_5_branch'

15 years agodocs: fix selection/feedback culling bug
Brian Paul [Thu, 27 Aug 2009 22:50:44 +0000 (16:50 -0600)]
docs: fix selection/feedback culling bug

15 years agoswrast: fix incorrect tri culling in selection/feedback mode.
Brian Paul [Thu, 27 Aug 2009 22:50:03 +0000 (16:50 -0600)]
swrast: fix incorrect tri culling in selection/feedback mode.

See bug 16866.

15 years agoi915: Fix undefined symbol as of eabe12df44a41e97fb5736959e8864ddbd01be14
Eric Anholt [Thu, 27 Aug 2009 22:33:33 +0000 (15:33 -0700)]
i915: Fix undefined symbol as of eabe12df44a41e97fb5736959e8864ddbd01be14

15 years agost/xorg: set vertex and fragment shader for the current op
Zack Rusin [Thu, 27 Aug 2009 22:17:54 +0000 (18:17 -0400)]
st/xorg: set vertex and fragment shader for the current op

15 years agost/xorg: bind rasterizer state
Zack Rusin [Thu, 27 Aug 2009 22:08:58 +0000 (18:08 -0400)]
st/xorg: bind rasterizer state

15 years agost/xorg: set blend state for exa
Zack Rusin [Thu, 27 Aug 2009 22:06:15 +0000 (18:06 -0400)]
st/xorg: set blend state for exa

15 years agost/xorg: bind framebuffer and viewport for exa
Zack Rusin [Thu, 27 Aug 2009 21:39:20 +0000 (17:39 -0400)]
st/xorg: bind framebuffer and viewport for exa

15 years agost/xorg: create and destroy shader cache
Zack Rusin [Wed, 26 Aug 2009 22:32:36 +0000 (18:32 -0400)]
st/xorg: create and destroy shader cache

15 years agost/xorg: Set shared usage of pixmaps for use via DRI2.
Michel Dänzer [Thu, 27 Aug 2009 18:29:32 +0000 (20:29 +0200)]
st/xorg: Set shared usage of pixmaps for use via DRI2.

Also allow pixmaps to have both the shared and displayed usage flags, as is
the case for the screen pixmap now.

15 years agost/dri: Make sure the front left renderbuffer is there for texture-from-pixmap.
Michel Dänzer [Thu, 27 Aug 2009 18:27:28 +0000 (20:27 +0200)]
st/dri: Make sure the front left renderbuffer is there for texture-from-pixmap.

15 years agost/xorg: Don't hardcode ExaPrepareCopy to fall back to software.
Michel Dänzer [Thu, 27 Aug 2009 18:27:28 +0000 (20:27 +0200)]
st/xorg: Don't hardcode ExaPrepareCopy to fall back to software.

15 years agost/xorg: Flush context if necessary in ExaPrepareAccess.
Michel Dänzer [Thu, 27 Aug 2009 18:27:28 +0000 (20:27 +0200)]
st/xorg: Flush context if necessary in ExaPrepareAccess.

15 years agost/dri: Only ask for fake front buffer for single buffer visuals by default.
Michel Dänzer [Thu, 27 Aug 2009 18:17:53 +0000 (20:17 +0200)]
st/dri: Only ask for fake front buffer for single buffer visuals by default.

15 years agost/xorg/dri: Pass texture formats via the DRI2 protocol flags.
Michel Dänzer [Thu, 27 Aug 2009 18:17:53 +0000 (20:17 +0200)]
st/xorg/dri: Pass texture formats via the DRI2 protocol flags.

No need to guess.

15 years agor300: Cleanup reported native program limits
Nicolai Hähnle [Thu, 27 Aug 2009 15:51:12 +0000 (17:51 +0200)]
r300: Cleanup reported native program limits

The fragment program native limits are actually higher on R4xx
as reported by Alex Deucher, but using those requires some
bit-twiddling changes to program handling.

Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
15 years agor300: Debug messages are written to stderr, so fflush that
Nicolai Hähnle [Thu, 27 Aug 2009 15:12:22 +0000 (17:12 +0200)]
r300: Debug messages are written to stderr, so fflush that

Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
15 years agoradeon/r200/r300: Fix swtcl prediction to work after primitie change.
Pauli Nieminen [Thu, 27 Aug 2009 11:21:13 +0000 (14:21 +0300)]
radeon/r200/r300: Fix swtcl prediction to work after primitie change.

Swtcl calls flush everytime primitive changes so prediction has to made again
after flushing.

15 years agomesa: direct program debug output to stderr instead of stdout
marvin24 [Thu, 27 Aug 2009 15:22:51 +0000 (09:22 -0600)]
mesa: direct program debug output to stderr instead of stdout

15 years agoMerge branch 'mesa_7_5_branch'
Brian Paul [Thu, 27 Aug 2009 15:14:09 +0000 (09:14 -0600)]
Merge branch 'mesa_7_5_branch'

15 years agogallium/util: added support for SRGB formats
Brian Paul [Thu, 27 Aug 2009 15:10:34 +0000 (09:10 -0600)]
gallium/util: added support for SRGB formats

Fixes glean/texture_srgb failure, bug #23449.

15 years agogallium/util: added cases for SRGB formats
Brian Paul [Thu, 27 Aug 2009 15:09:56 +0000 (09:09 -0600)]
gallium/util: added cases for SRGB formats

15 years agoprogs: EGL/X progs should link to libX11.
Chia-I Wu [Wed, 26 Aug 2009 04:44:02 +0000 (12:44 +0800)]
progs: EGL/X progs should link to libX11.

Since 5a459d58fca2b71cb77c39f98df8a81ce6298421, libEGL no longer links
to libX11.  Add the dependency to affected progs and cleanup
prog/egl/Makefile.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
15 years agomesa: move decls before code
Vinson Lee [Thu, 27 Aug 2009 13:40:37 +0000 (07:40 -0600)]
mesa: move decls before code

15 years agor300g: Correct scissor setting, subtract 1 from window's width and height
Cooper Yuan [Thu, 27 Aug 2009 09:46:59 +0000 (17:46 +0800)]
r300g: Correct scissor setting, subtract 1 from window's width and height

15 years agoradeon: fix scissor calcs.
Dave Airlie [Thu, 27 Aug 2009 05:36:35 +0000 (15:36 +1000)]
radeon: fix scissor calcs.

For non-FBOs we need to invert, for FBOs the scissors are non-inverted.

no matter what we need to clamp them to the buffer sizes.

15 years agoi965: Increase assmebly shader program parameter limits
Ian Romanick [Thu, 27 Aug 2009 05:04:23 +0000 (22:04 -0700)]
i965: Increase assmebly shader program parameter limits

Increase the number of native program parameters to the same values
exposed by GLSL.

15 years agoARB prog: Change handling of program parameter limits
Ian Romanick [Mon, 24 Aug 2009 19:54:48 +0000 (12:54 -0700)]
ARB prog: Change handling of program parameter limits

Several changes are made to program parameter limits.  Several of the
non-NATIVE limits are set higher.  All of the NATIVE limits are set to
zero in the core Mesa code.  Each driver must set the actual value in
its context creation routine.  If the NATIVE value remains zero, this
indicates that hardware shaders may not be supported.

Each of the preceeding changes matches the bahavior of Apple's shader
assembler, so it seems safe.

Finally, we limit the value of MaxEnvParams to be no greater than
MaxNativeAttribs.  At least one case has been found where an
application does the wrong thing if MaxNativeAttribs < MaxEnvParams.

See also bugzilla #23490.

15 years agoradeon: fix max indx/vertex emission due to state checker
Dave Airlie [Fri, 24 Jul 2009 01:46:20 +0000 (11:46 +1000)]
radeon: fix max indx/vertex emission due to state checker

15 years agomesa: gitignore
Younes Manton [Thu, 27 Aug 2009 00:13:11 +0000 (20:13 -0400)]
mesa: gitignore

15 years agonv04: Handle more Z formats.
Younes Manton [Thu, 27 Aug 2009 00:07:51 +0000 (20:07 -0400)]
nv04: Handle more Z formats.

Based on the patch from Luo Jinghua.

15 years agor300/compiler: Fix vertex program MAD emit
Nicolai Hähnle [Wed, 26 Aug 2009 21:31:37 +0000 (23:31 +0200)]
r300/compiler: Fix vertex program MAD emit

Only use the macro variant of MAD when absolutely necessary.
Apparently it cannot deal with relative addressing.

Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
15 years agor300/compiler: Fix bug in rc_find_free_temporary
Nicolai Hähnle [Wed, 26 Aug 2009 20:53:24 +0000 (22:53 +0200)]
r300/compiler: Fix bug in rc_find_free_temporary

Find used temporaries even if they are only written to in dead code.
This fixes a bug in the NQSSADCE stage.

Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
15 years agoegl: Remove Xdpy from EGLDisplay.
Chia-I Wu [Wed, 26 Aug 2009 08:06:39 +0000 (16:06 +0800)]
egl: Remove Xdpy from EGLDisplay.

It is not used anymore.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
15 years agoegl_glx: Make fbconfigs and visuals per display.
Chia-I Wu [Wed, 26 Aug 2009 06:45:17 +0000 (14:45 +0800)]
egl_glx: Make fbconfigs and visuals per display.

This is to allow a driver to drive multiple displays.  Remove the use of
_EGL_PLATFORM_X and obsolete code along the way.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
15 years agoegl_softpipe: Make winsys and pipe screen per display.
Chia-I Wu [Wed, 26 Aug 2009 07:42:35 +0000 (15:42 +0800)]
egl_softpipe: Make winsys and pipe screen per display.

This is to allow a driver to drive multiple displays.  Remove the use of
_EGL_PLATFORM_X along the way.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
15 years agoEGL_i915: Make struct drm_device per display.
Chia-I Wu [Wed, 26 Aug 2009 08:39:29 +0000 (16:39 +0800)]
EGL_i915: Make struct drm_device per display.

This is to allow a driver to drive multiple displays.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
15 years agoMerge branch 'fix_r300_swtcl' into push_master
Pauli Nieminen [Wed, 26 Aug 2009 21:08:30 +0000 (00:08 +0300)]
Merge branch 'fix_r300_swtcl' into push_master

15 years agoradeon: Fix loop en condition so debug output doesn't flood meaningless values.
Pauli Nieminen [Wed, 26 Aug 2009 21:07:05 +0000 (00:07 +0300)]
radeon: Fix loop en condition so debug output doesn't flood meaningless values.

15 years agoradeon: Make RADEON_CMDBUF more fine grained.
Pauli Nieminen [Wed, 26 Aug 2009 21:05:57 +0000 (00:05 +0300)]
radeon: Make RADEON_CMDBUF more fine grained.

15 years agoradeon/r200/r300: Fix swtcl flushing not to invalidate dma region.
Pauli Nieminen [Wed, 26 Aug 2009 19:24:25 +0000 (22:24 +0300)]
radeon/r200/r300: Fix swtcl flushing not to invalidate dma region.

We were check command buffer sizes too alte so allocated dma regions
were freed before relocations so space checking failed.

15 years agoMerge branch 'mesa_7_5_branch'
Brian Paul [Wed, 26 Aug 2009 20:49:15 +0000 (14:49 -0600)]
Merge branch 'mesa_7_5_branch'

15 years agoi965: init the tex_units_used field
Brian Paul [Wed, 26 Aug 2009 20:47:50 +0000 (14:47 -0600)]
i965: init the tex_units_used field

15 years agoi965: fix incorrect tex unit in emit_tex() and emit_txb()
Brian Paul [Wed, 26 Aug 2009 20:46:58 +0000 (14:46 -0600)]
i965: fix incorrect tex unit in emit_tex() and emit_txb()

The instructions we're translating already went through the brw_wm_pass_fp()
function which does the sampler->texture unit mapping.  We were applying
the sample->unit mapping a second time in the GLSL texture emitters.
Often, this made no difference but other times it could lead to accessing
an invalid texture and could cause a GPU lockup.

15 years agoi965: clean-up tex target switches
Brian Paul [Wed, 26 Aug 2009 20:43:45 +0000 (14:43 -0600)]
i965: clean-up tex target switches

15 years agoglsl: asst. clean-ups in set_program_uniform()
Brian Paul [Wed, 26 Aug 2009 20:35:45 +0000 (14:35 -0600)]
glsl: asst. clean-ups in set_program_uniform()

Remove redunant type check for samplers (assert instead).
Move some local vars.  Update comments.

15 years agoglsl: signal that the program needs to be re-translated when samplers change
Brian Paul [Wed, 26 Aug 2009 20:29:50 +0000 (14:29 -0600)]
glsl: signal that the program needs to be re-translated when samplers change

15 years agoprogs/glsl: asst. changes in shtest.c
Brian Paul [Wed, 26 Aug 2009 18:16:07 +0000 (12:16 -0600)]
progs/glsl: asst. changes in shtest.c

15 years agoi965: added texture unit sanity check
Brian Paul [Wed, 26 Aug 2009 18:07:52 +0000 (12:07 -0600)]
i965: added texture unit sanity check

Check that all the textures needed by the current fragment program
actually exist and are valid.

15 years agoi965: keep track of which texture units the fragment shader accesses
Brian Paul [Wed, 26 Aug 2009 18:07:03 +0000 (12:07 -0600)]
i965: keep track of which texture units the fragment shader accesses

We'll use this for debug/sanity checking.

15 years agodocs: document sampler array bug fix
Brian Paul [Wed, 26 Aug 2009 18:04:35 +0000 (12:04 -0600)]
docs: document sampler array bug fix

15 years agoi965: clean up texture target switches
Brian Paul [Wed, 26 Aug 2009 17:05:04 +0000 (11:05 -0600)]
i965: clean up texture target switches

15 years agomesa: var renaming, new assertion
Brian Paul [Wed, 26 Aug 2009 16:58:06 +0000 (10:58 -0600)]
mesa: var renaming, new assertion

15 years agomesa: additional instruction field size assertions
Brian Paul [Wed, 26 Aug 2009 16:57:18 +0000 (10:57 -0600)]
mesa: additional instruction field size assertions

15 years agoMerge branch 'mesa_7_5_branch'
Brian Paul [Wed, 26 Aug 2009 18:01:23 +0000 (12:01 -0600)]
Merge branch 'mesa_7_5_branch'

15 years agoprogs/glsl: handle window resizes in texaaline.c
Brian Paul [Wed, 26 Aug 2009 17:56:19 +0000 (11:56 -0600)]
progs/glsl: handle window resizes in texaaline.c

15 years agoprogs/glsl: add special Makefile rule for samplers_array
Brian Paul [Wed, 26 Aug 2009 17:55:15 +0000 (11:55 -0600)]
progs/glsl: add special Makefile rule for samplers_array

15 years agoprogs/glsl: change samplers.c to better test sampler/texture indexing
Brian Paul [Wed, 26 Aug 2009 17:53:25 +0000 (11:53 -0600)]
progs/glsl: change samplers.c to better test sampler/texture indexing

Now the left half is yellow and the right half is red, with the gradients
going in opposite directions.

15 years agoglsl: fix bug in sampler array indexing
Brian Paul [Wed, 26 Aug 2009 17:39:24 +0000 (11:39 -0600)]
glsl: fix bug in sampler array indexing

Need to add the 'offset' parameter when indexing the parameter array.
Before, if we were setting arrays of samplers, we were actually only
setting the 0th sampler's value.

Because of how progs/glsl/samplers.c is constructed, this wasn't showing
up as a failure in the samplers_array output.