mesa.git
14 years agoglsl2: Decompose matrix comparison into vector operations
Ian Romanick [Thu, 19 Aug 2010 00:53:47 +0000 (17:53 -0700)]
glsl2: Decompose matrix comparison into vector operations

14 years agotrace: Don't try to dump the rgba array if null
Jakob Bornecrantz [Fri, 27 Aug 2010 23:30:00 +0000 (00:30 +0100)]
trace: Don't try to dump the rgba array if null

14 years agoglsl: Rename a couple of common variable names in mat_op_to_vec.
Eric Anholt [Fri, 27 Aug 2010 22:34:42 +0000 (15:34 -0700)]
glsl: Rename a couple of common variable names in mat_op_to_vec.

It was easy while typing implementations to accidentally overwrite the
original expression or assignment variables.

14 years agoglsl: Make mat_op_to_vec allocate out of the IR's parent.
Eric Anholt [Fri, 27 Aug 2010 22:32:59 +0000 (15:32 -0700)]
glsl: Make mat_op_to_vec allocate out of the IR's parent.

This will reduce memory consumption of compiled shaders by not
dragging optimized-out children around.

14 years agoi965: Set the pop count on BRK/CONT inside of an if statement in the FS.
Eric Anholt [Fri, 27 Aug 2010 23:04:16 +0000 (16:04 -0700)]
i965: Set the pop count on BRK/CONT inside of an if statement in the FS.

This is the same as 8de8c97275e9555183a7e8f2238143657bbe60b2 for the
VS, and fixes glsl-fs-if-nested-loop and the mandelbrot demo.

Bug #29498

14 years agoglsl: Regenerate for double destroy fix.
Eric Anholt [Fri, 27 Aug 2010 22:24:23 +0000 (15:24 -0700)]
glsl: Regenerate for double destroy fix.

14 years agoglsl: Protect against double compiler-destroy.
Eric Anholt [Fri, 27 Aug 2010 22:23:29 +0000 (15:23 -0700)]
glsl: Protect against double compiler-destroy.

DRI was doing teardown when we close the last screen, then an atexit()
was added to call it as well.

14 years agodri: Get prototype for _mesa_destroy_compiler().
Eric Anholt [Fri, 27 Aug 2010 21:43:39 +0000 (14:43 -0700)]
dri: Get prototype for _mesa_destroy_compiler().

Bug #29665.

14 years agoi965: Don't strip negate/abs flags when assigning uniform locations.
Eric Anholt [Fri, 27 Aug 2010 21:15:42 +0000 (14:15 -0700)]
i965: Don't strip negate/abs flags when assigning uniform locations.

Fixes glsl-algebraic-sub-zero-4.

14 years agoi965: Add missing handling for BRW_OPCODE_SEL.
Eric Anholt [Fri, 27 Aug 2010 21:09:05 +0000 (14:09 -0700)]
i965: Add missing handling for BRW_OPCODE_SEL.

Fixes 4 piglit tests about min, max, and clamp.

14 years agoi965: Mask out higher bits of the result of BRW_CMP producing a boolean.
Eric Anholt [Fri, 27 Aug 2010 20:49:09 +0000 (13:49 -0700)]
i965: Mask out higher bits of the result of BRW_CMP producing a boolean.

When it says it sets the LSB, that's not just a hint as to where the
result goes.  Only the LSB is modified.  Fixes 20 piglit cases.

14 years agoi965: Fix the types of immediate integer values.
Eric Anholt [Fri, 27 Aug 2010 20:47:37 +0000 (13:47 -0700)]
i965: Fix the types of immediate integer values.

When we're trying to do integer ops, handing a float in doesn't help.

14 years agoi965: Add translation for RNDD and RNDZ.
Eric Anholt [Fri, 27 Aug 2010 20:29:45 +0000 (13:29 -0700)]
i965: Add translation for RNDD and RNDZ.

Fixes:
glsl-fs-any.
glsl1-integer division with uniform var

14 years agoi965: Add support for ir_binop_mod using do_mod_to_fract.
Eric Anholt [Fri, 27 Aug 2010 20:24:41 +0000 (13:24 -0700)]
i965: Add support for ir_binop_mod using do_mod_to_fract.

Fixes glsl-fs-mod.

14 years agor600g: fix vbo size
Jerome Glisse [Fri, 27 Aug 2010 20:52:29 +0000 (16:52 -0400)]
r600g: fix vbo size

Silence the kernel, vbo size is size - 1.

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agomesa: move null ptr check (fdo bug 29843)
Brian Paul [Fri, 27 Aug 2010 20:35:52 +0000 (14:35 -0600)]
mesa: move null ptr check (fdo bug 29843)

14 years agomesa: Use the format info to identify packed depth/stencil formats.
Nick Bowler [Fri, 27 Aug 2010 00:52:30 +0000 (17:52 -0700)]
mesa: Use the format info to identify packed depth/stencil formats.

Due to a misunderstanding of the Z24_X8 and X8_Z24 formats, the earlier
patch created depth/stencil wrappers for them.  This broke swrast.

Use the format info instead, which only identifies Z24_S8 and S8_Z24 as
packed depth/stencil.  It also has the advantage of being nicer code.

Signed-off-by: Nick Bowler <nbowler@draconx.ca>
Signed-off-by: Brian Paul <brianp@vmware.com>
14 years agomesa: use atexit() handler to release GLSL compiler memory
Brian Paul [Fri, 27 Aug 2010 17:54:32 +0000 (11:54 -0600)]
mesa: use atexit() handler to release GLSL compiler memory

This releases a bunch of memory that was showing up as leaks with
valgrind.

If atexit() isn't widely supported we may need to add some #ifdef
tests around the call.

14 years agoglsl2: restructure header file for C++ and C inclusion
Brian Paul [Fri, 27 Aug 2010 17:53:19 +0000 (11:53 -0600)]
glsl2: restructure header file for C++ and C inclusion

As it was, the header could not be cleanly #included by a C source.

14 years agoglsl2: remove 'extern' keyword in .c file
Brian Paul [Fri, 27 Aug 2010 17:15:03 +0000 (11:15 -0600)]
glsl2: remove 'extern' keyword in .c file

14 years agoi965: Fix swapped instructions in ir_unop_abs and ir_unop_neg.
Eric Anholt [Fri, 27 Aug 2010 20:19:28 +0000 (13:19 -0700)]
i965: Fix swapped instructions in ir_unop_abs and ir_unop_neg.

Fixes glsl-fs-neg and 5 other tests.

14 years agoi965: Add generate() handling for AND, OR, XOR.
Eric Anholt [Fri, 27 Aug 2010 20:14:15 +0000 (13:14 -0700)]
i965: Add generate() handling for AND, OR, XOR.

10 more piglit tests pass.

14 years agoi965: Add support for if instructions in the new FS backend.
Eric Anholt [Fri, 27 Aug 2010 19:54:12 +0000 (12:54 -0700)]
i965: Add support for if instructions in the new FS backend.

20 more piglit tests pass.

14 years agoi965: When encountering an unknown opcode in new FS backend, print its name.
Eric Anholt [Fri, 27 Aug 2010 19:53:48 +0000 (12:53 -0700)]
i965: When encountering an unknown opcode in new FS backend, print its name.

14 years agoi965: Fix the maximum grf counting in the new FS backend.
Eric Anholt [Fri, 27 Aug 2010 19:19:30 +0000 (12:19 -0700)]
i965: Fix the maximum grf counting in the new FS backend.

glsl-algebraic-rcp-rsq managed to use 33 registers, and we claimed to
only use 32, so the write to g32 would go stomping over the precious
g0 of some other thread.

14 years agoi965: Validate the IR tree after doing our custom optimization passes.
Eric Anholt [Fri, 27 Aug 2010 19:02:53 +0000 (12:02 -0700)]
i965: Validate the IR tree after doing our custom optimization passes.

This wouldn't catch the last failure fixed in them, because we don't
validate assignments well (due to the fact that we've got a pretty
glaring inconsistency in how we handle assignment writemasking), but
it could catch other failure we may produce.

14 years agoi965: Add a bit of support for matrices to the new FS.
Eric Anholt [Fri, 27 Aug 2010 17:44:04 +0000 (10:44 -0700)]
i965: Add a bit of support for matrices to the new FS.

14 years agoi965: Fix destination writemasking in the new FS.
Eric Anholt [Fri, 27 Aug 2010 18:05:47 +0000 (11:05 -0700)]
i965: Fix destination writemasking in the new FS.

14 years agoi965: Fix swizzling in vector splitting for the new FS backend.
Eric Anholt [Fri, 27 Aug 2010 19:00:29 +0000 (12:00 -0700)]
i965: Fix swizzling in vector splitting for the new FS backend.

We weren't smearing a component of a split RHS out to reach an unsplit
LHS's writemask, so gl_FragColor (always unsplit) would often get
uninitialized values.

Fixes: glsl-algebraic-add-add-1 (and probably many others).
14 years agograw: fix the build (missing header)
Zack Rusin [Fri, 27 Aug 2010 17:09:38 +0000 (13:09 -0400)]
graw: fix the build (missing header)

14 years agomesa: free the fallback texture object in free_shared_state()
Brian Paul [Fri, 27 Aug 2010 16:51:47 +0000 (10:51 -0600)]
mesa: free the fallback texture object in free_shared_state()

14 years agomesa: fix double-underscore naming
Brian Paul [Fri, 27 Aug 2010 16:21:59 +0000 (10:21 -0600)]
mesa: fix double-underscore naming

14 years agollvmpipe: use util_iround in place of round
Keith Whitwell [Fri, 27 Aug 2010 12:40:23 +0000 (13:40 +0100)]
llvmpipe: use util_iround in place of round

Fix mingw build.

14 years agoutil: fix typo in MAX4
Keith Whitwell [Fri, 27 Aug 2010 12:29:00 +0000 (13:29 +0100)]
util: fix typo in MAX4

Thanks to Michal for spotting it.

14 years agollvmpipe: eliminate tri->dx, tri->dy values
Keith Whitwell [Fri, 27 Aug 2010 10:03:58 +0000 (11:03 +0100)]
llvmpipe: eliminate tri->dx, tri->dy values

Use an internal struct for line setup information.

14 years agollvmpipe: point sprites rasterization
Hui Qi Tay [Tue, 10 Aug 2010 10:41:32 +0000 (11:41 +0100)]
llvmpipe: point sprites rasterization

Point sprites now done in the rasterizer setup code instead of
going through the draw module.

14 years agollvmpipe: native point rasterization with better pixel rasterization
Hui Qi Tay [Fri, 27 Aug 2010 09:46:19 +0000 (10:46 +0100)]
llvmpipe: native point rasterization with better pixel rasterization

A few subpixel_snap and fixed width changes.

Conflicts:

src/gallium/drivers/llvmpipe/lp_setup_point.c

14 years agollvmpipe: native point rasterization
Hui Qi Tay [Fri, 27 Aug 2010 09:37:09 +0000 (10:37 +0100)]
llvmpipe: native point rasterization

Conflicts:

src/gallium/drivers/llvmpipe/lp_setup_context.h
src/gallium/drivers/llvmpipe/lp_setup_line.c
src/gallium/drivers/llvmpipe/lp_setup_tri.c

14 years agollvmpipe: update line rasterization code to current master
Keith Whitwell [Thu, 26 Aug 2010 19:09:22 +0000 (20:09 +0100)]
llvmpipe: update line rasterization code to current master

14 years agoutil: add MIN4, MAX4
Keith Whitwell [Thu, 26 Aug 2010 19:03:03 +0000 (20:03 +0100)]
util: add MIN4, MAX4

14 years agollvmpipe: native line rasterization with correct pixel rasterization
Hui Qi Tay [Wed, 4 Aug 2010 16:13:39 +0000 (17:13 +0100)]
llvmpipe: native line rasterization with correct pixel rasterization

Line rasterization that follows diamond exit rule.
Can still optimize logic for start/endpoints.

14 years agollvmpipe: native rasterization for lines
Hui Qi Tay [Mon, 19 Jul 2010 14:23:09 +0000 (15:23 +0100)]
llvmpipe: native rasterization for lines

Rasterize lines directly by treating them as 4-sided polygons.
Still need to check the exact pixel rasteration.

14 years agollvmpipe: add lp_setup_coef to makefile
Keith Whitwell [Fri, 27 Aug 2010 10:18:11 +0000 (11:18 +0100)]
llvmpipe: add lp_setup_coef to makefile

14 years agollvmpipe: intrinsics version of triangle coeficient calculation
Keith Whitwell [Sun, 22 Aug 2010 09:57:12 +0000 (10:57 +0100)]
llvmpipe: intrinsics version of triangle coeficient calculation

Looks nice, but makes almost no impact on performance - maybe
a percent or so in isosurf, nothing elsewhere.  May be of use
later on.

14 years agogallium: Remove unnecessary header from p_shader_tokens.h.
Vinson Lee [Fri, 27 Aug 2010 07:40:42 +0000 (00:40 -0700)]
gallium: Remove unnecessary header from p_shader_tokens.h.

Remove p_compiler.h.

14 years agotgsi: Include missing header in tgsi_parse.h.
Vinson Lee [Fri, 27 Aug 2010 07:34:32 +0000 (00:34 -0700)]
tgsi: Include missing header in tgsi_parse.h.

Include p_compiler.h for boolean and INLINE symbols.

14 years agotgsi: Include missing header in tgsi_info.h.
Vinson Lee [Fri, 27 Aug 2010 07:31:27 +0000 (00:31 -0700)]
tgsi: Include missing header in tgsi_info.h.

Include p_compiler.h for uint symbol.

14 years agotgsi: Include missing header in tgsi_dump.h.
Vinson Lee [Fri, 27 Aug 2010 07:26:59 +0000 (00:26 -0700)]
tgsi: Include missing header in tgsi_dump.h.

Include p_compiler.h for uint symbol.

14 years agogallium: Remove unnecessary header from p_format.h.
Vinson Lee [Fri, 27 Aug 2010 07:15:04 +0000 (00:15 -0700)]
gallium: Remove unnecessary header from p_format.h.

Remove p_compiler.h.

14 years agograw: Include missing header in graw.h.
Vinson Lee [Fri, 27 Aug 2010 07:07:38 +0000 (00:07 -0700)]
graw: Include missing header in graw.h.

Include p_compiler.h for PUBLIC symbol.

14 years agor300g: Include missing header in r300_texture.h.
Vinson Lee [Fri, 27 Aug 2010 07:04:50 +0000 (00:04 -0700)]
r300g: Include missing header in r300_texture.h.

Include p_compiler.h for uint32_t and boolean symbols.

14 years agoFix typo in function name "shading_laguage_version".
Kenneth Graunke [Fri, 27 Aug 2010 06:07:45 +0000 (23:07 -0700)]
Fix typo in function name "shading_laguage_version".

14 years agoRevert "r600g: simplify states"
Dave Airlie [Fri, 27 Aug 2010 05:45:58 +0000 (15:45 +1000)]
Revert "r600g: simplify states"

This reverts commit bd25e23bf3740f59ce8859848c715daeb9e9821f.

Apart from introducing a lot of hex magic numbers and being highly impenetable code,
it causes lots of lockups on an average piglit run that always runs without lockups.

Always run piglit before/after doing big things like this.

14 years agor600g: add initial if/else/endif support
Dave Airlie [Wed, 25 Aug 2010 05:57:41 +0000 (15:57 +1000)]
r600g: add initial if/else/endif support

this adds handling for some more CF instructions and conditions
also adds parameter for stack size emission

These seem to pass on VS with the stack size hack but not on FS,

TODO: fix FS + stack size calcs

14 years agor600g: optimise op2 and swapped op2 emission.
Dave Airlie [Wed, 25 Aug 2010 06:02:38 +0000 (16:02 +1000)]
r600g: optimise op2 and swapped op2 emission.

this makes op2 emission smaller, since it skips instructions
that don't write to the dst. not sure if this could have unwanted
side effects but try it and see.

14 years agor600g: add exp support in theory.
Dave Airlie [Tue, 24 Aug 2010 01:47:56 +0000 (11:47 +1000)]
r600g: add exp support in theory.

though it isn't passing the test, and this instruction is pure bonghits.

14 years agor600g: add DPH support.
Dave Airlie [Tue, 24 Aug 2010 00:29:33 +0000 (10:29 +1000)]
r600g: add DPH support.

14 years agoi965: Add preliminary support for uniforms to the new FS backend.
Eric Anholt [Thu, 26 Aug 2010 23:39:41 +0000 (16:39 -0700)]
i965: Add preliminary support for uniforms to the new FS backend.

+269 piglits

14 years agoi965: Abort on gl_FragDepth in the new FS backend for now.
Eric Anholt [Fri, 27 Aug 2010 00:04:30 +0000 (17:04 -0700)]
i965: Abort on gl_FragDepth in the new FS backend for now.

It hangs the GPU due to FB_WRITE handling being incomplete.  There are
bigger issues to handle first.

14 years agoi965: Fix up and actually enable the NewShader and NewShaderProgram hooks.
Eric Anholt [Thu, 26 Aug 2010 23:59:55 +0000 (16:59 -0700)]
i965: Fix up and actually enable the NewShader and NewShaderProgram hooks.

14 years agoi965: Hack in avoidance of c++ reserved keyword in libdrm.
Eric Anholt [Thu, 26 Aug 2010 22:43:00 +0000 (15:43 -0700)]
i965: Hack in avoidance of c++ reserved keyword in libdrm.

I'm also fixing this upstream in libdrm, but this avoids new libdrm
dependency for the moment.

14 years agoi965: Add GLSL IR-level source annotation and comments to new FS debug.
Eric Anholt [Thu, 26 Aug 2010 21:42:06 +0000 (14:42 -0700)]
i965: Add GLSL IR-level source annotation and comments to new FS debug.

This should make debugging way easier, as now we have context for
reading large programs.

14 years agoi965: Use the implied move in brw_math() in the new FS.
Eric Anholt [Thu, 26 Aug 2010 21:09:54 +0000 (14:09 -0700)]
i965: Use the implied move in brw_math() in the new FS.

14 years agoi965: Add support for in varyings to the new FS codegen.
Eric Anholt [Tue, 17 Aug 2010 04:53:02 +0000 (21:53 -0700)]
i965: Add support for in varyings to the new FS codegen.

At least some tests, like glsl-vs-sign, now work.

14 years agoi965: Start building the codegen visitor.
Eric Anholt [Mon, 16 Aug 2010 01:58:58 +0000 (18:58 -0700)]
i965: Start building the codegen visitor.

This can successfully emit a real program that generates magenta now.

14 years agoi965: Start building direct GLSL2 IR to 965 assembly codegen.
Eric Anholt [Thu, 26 Aug 2010 19:12:00 +0000 (12:12 -0700)]
i965: Start building direct GLSL2 IR to 965 assembly codegen.

Our channel-expressions and vector-splitting changes now happen into a
private copy of the IR that we maintain for ourselves.  Uniform
assignment still happens by the core, so we continue using Mesa IR
generation not just for swrast fallbacks but also for uniform values
(since there's no storage for their contents other than
shader_program->FragmentProgram->Parameters->ParameterValues).  And
most importantly, at the moment no actual codegen is hooked up other
than emitting our favorite color to the framebuffer.

14 years agoi965: Add new pass to split vectors into scalar variables
Eric Anholt [Thu, 26 Aug 2010 19:02:26 +0000 (12:02 -0700)]
i965: Add new pass to split vectors into scalar variables

Combined with the previous pass, this lets other optimization passes
do their work thanks to ir_tree_grafting.  Still have regression in
instruction count with INTEL_NEW_FS, but register count is even
better.

14 years agoi965: Add a pass for the FS to reduce vector expressions down to scalar.
Eric Anholt [Fri, 13 Aug 2010 09:20:40 +0000 (02:20 -0700)]
i965: Add a pass for the FS to reduce vector expressions down to scalar.

This is a step towards implementing a GLSL IR backend for the 965
fragment shader.  Because it has downsides with the current codegen,
it is hidden under the environment variable INTEL_NEW_FS.

This results in an increase in instruction count at the moment (1444
-> 1752 for glsl-fs-raytrace, 345 -> 359 on my demo), because dot
products are turned into a series of multiplies and adds instead of a
custom expansion of MULs and MACs, and by not splitting the variable
types up we don't get tree grafting and thus there are extra moves of
temporary storage.  However, register count drops for the non-GLSL
path (64 -> 56 on my demo shader) because the register allocator sees
all the sub-operations.

14 years agoi965: Start building 965 FS backend.
Eric Anholt [Wed, 11 Aug 2010 03:39:06 +0000 (20:39 -0700)]
i965: Start building 965 FS backend.

14 years agollvmpipe: fix PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS query
Brian Paul [Thu, 26 Aug 2010 17:45:25 +0000 (11:45 -0600)]
llvmpipe: fix PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS query

Fixes crashes in glean glsl1 and demos/src/glsl/vert-tex.
See comments for details.

14 years agograw: Add copyright headers to the interfaces.
José Fonseca [Thu, 26 Aug 2010 05:52:12 +0000 (06:52 +0100)]
graw: Add copyright headers to the interfaces.

14 years agoir_to_mesa: Don't assume that an ir_dereference_array is of a variable.
Eric Anholt [Thu, 26 Aug 2010 18:23:09 +0000 (11:23 -0700)]
ir_to_mesa: Don't assume that an ir_dereference_array is of a variable.

Fixes:
glsl-array-bounds-02 (software)
glsl-array-bounds-04
glsl-array-bounds-06 (software)
glsl-array-bounds-08

14 years agoglsl: Add a quick hack to constant folding to reduce duplicated work.
Eric Anholt [Thu, 26 Aug 2010 17:42:47 +0000 (10:42 -0700)]
glsl: Add a quick hack to constant folding to reduce duplicated work.

Reduces runtime of glsl-max-varyings 92% on my system.

14 years agost/mesa: Remove unnecessary header.
Vinson Lee [Thu, 26 Aug 2010 17:38:28 +0000 (10:38 -0700)]
st/mesa: Remove unnecessary header.

14 years agoscons: Fix inverted logic.
José Fonseca [Thu, 26 Aug 2010 17:28:29 +0000 (18:28 +0100)]
scons: Fix inverted logic.

14 years agodocs: update news.html with 7.8.2 release info
Brian Paul [Thu, 26 Aug 2010 17:24:22 +0000 (11:24 -0600)]
docs: update news.html with 7.8.2 release info

14 years agodocs: remove links to tungstengraphics.com, and misc updates
Brian Paul [Thu, 26 Aug 2010 17:20:31 +0000 (11:20 -0600)]
docs: remove links to tungstengraphics.com, and misc updates

14 years agodocs: remove link to old memory.html page
Brian Paul [Thu, 26 Aug 2010 17:18:06 +0000 (11:18 -0600)]
docs: remove link to old memory.html page

14 years agoscons: Add glsl_symbol_table.cpp
José Fonseca [Thu, 26 Aug 2010 17:19:57 +0000 (18:19 +0100)]
scons: Add glsl_symbol_table.cpp

14 years agoi965: Add support for destination RelAddr writes in the VS.
Eric Anholt [Thu, 26 Aug 2010 15:47:24 +0000 (08:47 -0700)]
i965: Add support for destination RelAddr writes in the VS.

Fixes: glsl-vs-varying-array
14 years agoi965: Fix the test for variable indexing of shader inputs.
Eric Anholt [Thu, 26 Aug 2010 16:10:19 +0000 (09:10 -0700)]
i965: Fix the test for variable indexing of shader inputs.

Shader inputs appear in source registers, not dst registers.  Catches
unsupported shaders in glsl-fs-varying-array and Humus
RaytracedShadows.

14 years agoglsl2: Remove a couple FINISHME comments that have already been resolved
Ian Romanick [Thu, 26 Aug 2010 16:24:58 +0000 (09:24 -0700)]
glsl2: Remove a couple FINISHME comments that have already been resolved

14 years agoglsl: Move built-ins to live beyond the global scope.
Kenneth Graunke [Mon, 23 Aug 2010 21:52:06 +0000 (14:52 -0700)]
glsl: Move built-ins to live beyond the global scope.

Per the GLSL 1.20 specification (presumably a clarification of 1.10).

Also, when creating user functions, make a new ir_function that shadows the
built-in ir_function, rather than adding new signatures.  User functions
are supposed to hide built-ins, not overload them.

Fixes piglit tests redeclaration-{04, 12, 14}.vert.

14 years agoglsl: Move is_built_in flag from ir_function_signature to ir_function.
Kenneth Graunke [Sat, 21 Aug 2010 03:04:39 +0000 (20:04 -0700)]
glsl: Move is_built_in flag from ir_function_signature to ir_function.

Also rename it to "is_builtin" for consistency.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
14 years agoglsl: Refactor variable declaration handling.
Kenneth Graunke [Tue, 24 Aug 2010 08:45:49 +0000 (01:45 -0700)]
glsl: Refactor variable declaration handling.

Moving the check for an earlier variable declaration helps cleanly
separate out the re-declaration vs. new declaration code a bit.  With
that in place, conflicts between variable names and structure types or
function names aren't caught by the earlier "redeclaration" error
message, so check the return type on glsl_symbol_table::add_variable
and issue an error there.  If one occurs, don't emit the initializer.

Fixes redeclaration-01.vert and redeclaration-09.vert.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
14 years agoglsl: Don't add overloads to existing structure constructors.
Kenneth Graunke [Thu, 26 Aug 2010 00:10:16 +0000 (17:10 -0700)]
glsl: Don't add overloads to existing structure constructors.

Instead, make a new ir_function and try to add it to the symbol table.

Fixes piglit test redeclaration-08.vert.

14 years agoglsl: Remove name_declared_this_scope check when adding functions.
Kenneth Graunke [Wed, 25 Aug 2010 23:37:46 +0000 (16:37 -0700)]
glsl: Remove name_declared_this_scope check when adding functions.

Instead, rely on the symbol table's rules.

Fixes redeclaration-02.vert.

14 years agoglsl: Use a single shared namespace in the symbol table.
Kenneth Graunke [Sun, 22 Aug 2010 03:23:18 +0000 (20:23 -0700)]
glsl: Use a single shared namespace in the symbol table.

As of 1.20, variable names, function names, and structure type names all
share a single namespace, and should conflict with one another in the
same scope, or hide each other in nested scopes.

However, in 1.10, variables and functions can share the same name in the
same scope.  Structure types, however, conflict with/hide both.

Fixes piglit tests redeclaration-06.vert, redeclaration-11.vert,
redeclaration-19.vert, and struct-05.vert.

14 years agost/mesa: add missing packed depth/stencil formats in st_format_datatype()
Brian Paul [Thu, 26 Aug 2010 15:20:18 +0000 (09:20 -0600)]
st/mesa: add missing packed depth/stencil formats in st_format_datatype()

Fixes llvmpipe regression from one of the prev commits.

14 years agomesa: fix mixed-up function call name
Brian Paul [Thu, 26 Aug 2010 14:59:43 +0000 (08:59 -0600)]
mesa: fix mixed-up function call name

14 years agomesa: Identify packed depth/stencil buffers using the Format field.
Nick Bowler [Thu, 26 Aug 2010 14:26:21 +0000 (07:26 -0700)]
mesa: Identify packed depth/stencil buffers using the Format field.

Intel sometimes uses packed depth/stencil buffers even when only a depth
buffer or only a stencil buffer was requested.  Common code currently
uses the _BaseFormat field to determine whether a depth/stencil wrapper
is necessary.  But unless the user explicitly requested a packed
depth/stencil buffer, the _BaseFormat field does not encode this
information, and the required wrappers are not created.

The problem was introduced by commit 45e76d2665b38b ("mesa: remove a
bunch of gl_renderbuffer fields"), which killed off the _ActualFormat
field upon which the decision to create a wrapper used to be made.  This
patch changes the logic to use the Format field instead, which is more
like the old code.

Fixes fdo bug 27590.

Signed-off-by: Nick Bowler <nbowler@draconx.ca>
Signed-off-by: Brian Paul <brianp@vmware.com>
14 years agointel: Merge identical cases in switch statement.
Nick Bowler [Wed, 25 Aug 2010 13:14:42 +0000 (06:14 -0700)]
intel: Merge identical cases in switch statement.

Signed-off-by: Nick Bowler <nbowler@draconx.ca>
Signed-off-by: Brian Paul <brianp@vmware.com>
14 years agor600: fix vertex buffer size calculation
Andre Maasikas [Thu, 26 Aug 2010 12:22:21 +0000 (15:22 +0300)]
r600: fix vertex buffer size calculation

when we dont know max_index we cannot calculate vb size from count
anymore - just use the bo size.
Also added an assert to remind that we dont handle GL_INT GL_DOUBLE
upload when we dont' know max_index - will fix later

14 years agograw: Undo late loading of graw drivers.
José Fonseca [Thu, 26 Aug 2010 10:37:42 +0000 (11:37 +0100)]
graw: Undo late loading of graw drivers.

Keith prefers a clean separation between graw applications and
implementations, where apps do not link libgallium.a but instead
get all functionality they need via graw interface.

Although this is not incompatible with late loading of graw drivers, it
it would make it very hard to maintain, as wrappers for every utility
symbol exposed in graw would have to be written or generated somehow.

14 years agost/mesa: Fix glEGLImageTargetTexture2DOES.
Chia-I Wu [Wed, 25 Aug 2010 18:15:22 +0000 (02:15 +0800)]
st/mesa: Fix glEGLImageTargetTexture2DOES.

stObj->pt should be set in st_bind_surface, just as in st_TexImage.  On
the other hand, st_TexImage should unreference stObj->pt.  It also needs
to initialize the texture image again as _mesa_clear_texture_object
clears the image.

14 years agost/egl: Add support for EGL_MESA_image_drm.
Chia-I Wu [Wed, 25 Aug 2010 17:49:18 +0000 (01:49 +0800)]
st/egl: Add support for EGL_MESA_image_drm.

14 years agost/egl: Add support for EGL_KHR_surfaceless_*.
Chia-I Wu [Thu, 26 Aug 2010 07:10:14 +0000 (01:10 -0600)]
st/egl: Add support for EGL_KHR_surfaceless_*.

14 years agost/mesa: Add support for surfaceless current contexts.
Chia-I Wu [Wed, 25 Aug 2010 16:51:28 +0000 (00:51 +0800)]
st/mesa: Add support for surfaceless current contexts.

A surfaceless current context is a context that is made current without
draw and read framebuffers.  Such contexts can only render to FBOs.

14 years agost/egl: Make KMS support optional in KMS backend.
Chia-I Wu [Wed, 25 Aug 2010 16:38:59 +0000 (00:38 +0800)]
st/egl: Make KMS support optional in KMS backend.

It should be called DRM backend now.

14 years agor300g: Include missing header in r300_texture_desc.h.
Vinson Lee [Thu, 26 Aug 2010 08:38:23 +0000 (01:38 -0700)]
r300g: Include missing header in r300_texture_desc.h.

Include p_format.h for enum pipe_format symbol.

Fixes r300g build.