mesa.git
14 years agoast_to_hir: Reject function names that start with "gl_".
Kenneth Graunke [Fri, 20 Aug 2010 09:14:35 +0000 (02:14 -0700)]
ast_to_hir: Reject function names that start with "gl_".

Fixes piglit test redeclaration-03.vert.

14 years agoast_to_hir: Fix crash when a function shadows a variable.
Kenneth Graunke [Fri, 20 Aug 2010 09:04:52 +0000 (02:04 -0700)]
ast_to_hir: Fix crash when a function shadows a variable.

The code would attempt to add a new signature to the ir_function, which
didn't exist.  Simply bailing out/returning early seems reasonable.

Fixes piglit test redeclaration-02.vert, and fixes a crash in
redeclaration-03.vert (the test still fails).

14 years agogalahad, i915g: Copy over constant buffer index check.
Corbin Simpson [Fri, 20 Aug 2010 07:18:30 +0000 (00:18 -0700)]
galahad, i915g: Copy over constant buffer index check.

14 years agogalahad, i915g: Move over a few state asserts.
Corbin Simpson [Fri, 20 Aug 2010 07:00:40 +0000 (00:00 -0700)]
galahad, i915g: Move over a few state asserts.

14 years agogalahad: Make it obvious on stderr that Galahad's active.
Corbin Simpson [Fri, 20 Aug 2010 06:34:39 +0000 (23:34 -0700)]
galahad: Make it obvious on stderr that Galahad's active.

14 years agogallium/docs: Add Galahad
Jakob Bornecrantz [Fri, 20 Aug 2010 01:28:36 +0000 (02:28 +0100)]
gallium/docs: Add Galahad

14 years agoconfigs: Add -lstdc++ to default.
Vinson Lee [Fri, 20 Aug 2010 01:09:24 +0000 (18:09 -0700)]
configs: Add -lstdc++ to default.

This fixes the following error when trying to run glxinfo or glxgears
with swrast.

undefined symbol: __cxa_pure_virtual

14 years agomesa: Silence uninitialized variable warnings in dummy_enum_func.
Vinson Lee [Fri, 20 Aug 2010 00:07:28 +0000 (17:07 -0700)]
mesa: Silence uninitialized variable warnings in dummy_enum_func.

14 years agotranslate_sse: enable on Win64
Luca Barbieri [Thu, 19 Aug 2010 22:04:30 +0000 (00:04 +0200)]
translate_sse: enable on Win64

According to Vinson, enabling it causes no regressions

14 years agou_blitter: fix a memory leak
Marek Olšák [Thu, 19 Aug 2010 21:32:04 +0000 (23:32 +0200)]
u_blitter: fix a memory leak

14 years agor300g: do not use fastfill with 16-bit zbuffers
Marek Olšák [Thu, 19 Aug 2010 20:18:40 +0000 (22:18 +0200)]
r300g: do not use fastfill with 16-bit zbuffers

To my knowledge, there is no way to flush zmask and thus write the clear
value.

This fixes zbuffer reads, among other things.

14 years agor600g: update comments about ALU src operands
Alex Deucher [Thu, 19 Aug 2010 16:06:20 +0000 (12:06 -0400)]
r600g: update comments about ALU src operands

14 years agotranslate_sse: try to fix Win64
Luca Barbieri [Thu, 19 Aug 2010 08:07:58 +0000 (10:07 +0200)]
translate_sse: try to fix Win64

Not sure whether it works now (it is still disabled).

14 years agotranslate_sse: fix emit_load_sse2
Luca Barbieri [Thu, 19 Aug 2010 08:03:03 +0000 (10:03 +0200)]
translate_sse: fix emit_load_sse2

14 years agoi915: Remove unnecessary header from i830_context.c.
Vinson Lee [Thu, 19 Aug 2010 07:29:59 +0000 (00:29 -0700)]
i915: Remove unnecessary header from i830_context.c.

14 years agost/egl: Fix win32 build.
Chia-I Wu [Thu, 19 Aug 2010 05:45:29 +0000 (13:45 +0800)]
st/egl: Fix win32 build.

Add new source files to SConscripts.

14 years agor600g: add sin/cos
Dave Airlie [Thu, 19 Aug 2010 04:43:44 +0000 (14:43 +1000)]
r600g: add sin/cos

This pretty much ports the code from r600c, however it doesn't
always seem to work quite perfectly, but I can't find anything in this
code that is wrong. I'm guessing either literal input or constants
aren't working always.

14 years agor600g: add a chiprev type for r600/r700/evergreen instead of using family
Dave Airlie [Thu, 19 Aug 2010 04:43:11 +0000 (14:43 +1000)]
r600g: add a chiprev type for r600/r700/evergreen instead of using family

14 years agor600g: add SSG, SEQ, SGT and SNE
Dave Airlie [Thu, 19 Aug 2010 01:41:18 +0000 (11:41 +1000)]
r600g: add SSG, SEQ, SGT and SNE

14 years agoi965: Set the if stack pop count when breaking out of a loop inside an if.
Eric Anholt [Thu, 19 Aug 2010 04:21:05 +0000 (21:21 -0700)]
i965: Set the if stack pop count when breaking out of a loop inside an if.

Otherwise, we might end up with the if stack pointing at the wrong
place.  Fixes GPU hang with glsl-vs-if-loop.

14 years agoi965: Don't set the swizzle on an immediate value in the VS.
Eric Anholt [Thu, 19 Aug 2010 04:18:23 +0000 (21:18 -0700)]
i965: Don't set the swizzle on an immediate value in the VS.

Fixes glsl-vs-if-nested (70.0 is not <= 70.000648 thanks to the
swizzle bits getting set).  Some safety checks are added to make sure
this doesn't happen again as we increase the usage of immediate values
in program generation.

14 years agomesa: Fix the whining for link failures to actually be under MESA_GLSL=dump.
Eric Anholt [Thu, 19 Aug 2010 01:01:55 +0000 (18:01 -0700)]
mesa: Fix the whining for link failures to actually be under MESA_GLSL=dump.

14 years agoglsl: Replace sscanf in s_expression reader with strspn and strcspn.
Kenneth Graunke [Thu, 19 Aug 2010 01:03:22 +0000 (18:03 -0700)]
glsl: Replace sscanf in s_expression reader with strspn and strcspn.

This seems to give roughly a 20% speedup.

14 years agomesa: Remove unnecessary heaaders from shaderapi.c.
Vinson Lee [Thu, 19 Aug 2010 01:49:32 +0000 (18:49 -0700)]
mesa: Remove unnecessary heaaders from shaderapi.c.

14 years agomesa: Don't try to free components of a NULL uniform list.
Eric Anholt [Thu, 19 Aug 2010 00:42:34 +0000 (17:42 -0700)]
mesa: Don't try to free components of a NULL uniform list.

This might happen if we manage to trigger the right linker errors.

14 years agoglsl: Also strdup the names of uniform list entries for >vec4 types.
Eric Anholt [Thu, 19 Aug 2010 00:39:57 +0000 (17:39 -0700)]
glsl: Also strdup the names of uniform list entries for >vec4 types.

Fixes double-free since the fix to free all of the uniform list.

14 years agoglcpp: Refresh autogenerated lexer.
Kenneth Graunke [Thu, 19 Aug 2010 00:38:24 +0000 (17:38 -0700)]
glcpp: Refresh autogenerated lexer.

14 years agoglcpp: Add basic #line support (adapted from the main compiler).
Kenneth Graunke [Thu, 19 Aug 2010 00:38:05 +0000 (17:38 -0700)]
glcpp: Add basic #line support (adapted from the main compiler).

14 years agoir_to_mesa: Fix leak by improper freeing of a uniform list.
Eric Anholt [Thu, 19 Aug 2010 00:12:18 +0000 (17:12 -0700)]
ir_to_mesa: Fix leak by improper freeing of a uniform list.

14 years agoglsl2: Free the shader compiler at dri screen destruction.
Eric Anholt [Wed, 18 Aug 2010 23:56:39 +0000 (16:56 -0700)]
glsl2: Free the shader compiler at dri screen destruction.

Hooray, we can valgrind again without adding suppressions.  This also
adds an interface for use by an implementation of
glReleaseShaderCompiler().

14 years agoir_to_mesa: Don't leak the whole linked assembly program.
Eric Anholt [Wed, 18 Aug 2010 23:29:02 +0000 (16:29 -0700)]
ir_to_mesa: Don't leak the whole linked assembly program.

14 years agodri2: Clean up the common dri2 options at screen destroy.
Eric Anholt [Wed, 18 Aug 2010 23:17:10 +0000 (16:17 -0700)]
dri2: Clean up the common dri2 options at screen destroy.

14 years agoglsl: Fix leak-causing typo in destructor that made it another constructor.
Eric Anholt [Wed, 18 Aug 2010 22:59:31 +0000 (15:59 -0700)]
glsl: Fix leak-causing typo in destructor that made it another constructor.

14 years agoglsl: Fix uninitialized member in ir_hierarchical_vistor constructor.
Vinson Lee [Wed, 18 Aug 2010 23:10:15 +0000 (16:10 -0700)]
glsl: Fix uninitialized member in ir_hierarchical_vistor constructor.

Class member base_ir was not initialized by the default constructor.

14 years agomesa: fpclassify is available on OpenSolaris.
Vinson Lee [Wed, 18 Aug 2010 21:36:08 +0000 (14:36 -0700)]
mesa: fpclassify is available on OpenSolaris.

There is no explicit predefined macro to distinguish between OpenSolaris
and Solaris. This patch assumes that the difference is in the compilers.
OpenSolaris uses GCC and not the Sun Studio compiler. Assume that the
availability of fpclassify is due to GCC.

This patch was not tested on Solaris. It would break the build on
Solaris with GCC if GCC on Solaris does not have fpclassify.

14 years agoir_to_mesa: Fix leak of set_branchtargets temp data.
Eric Anholt [Wed, 18 Aug 2010 21:13:30 +0000 (14:13 -0700)]
ir_to_mesa: Fix leak of set_branchtargets temp data.

14 years agoglsl: Garbage collect old prototype for ir_to_mesa.
Eric Anholt [Wed, 18 Aug 2010 19:04:43 +0000 (12:04 -0700)]
glsl: Garbage collect old prototype for ir_to_mesa.

14 years agoir_to_mesa: Stop leaking the ir_instruction * annotation of our compile.
Eric Anholt [Wed, 18 Aug 2010 20:48:32 +0000 (13:48 -0700)]
ir_to_mesa: Stop leaking the ir_instruction * annotation of our compile.

14 years agomesa: Free linked shaders when deleting the shader program.
Eric Anholt [Wed, 18 Aug 2010 21:12:52 +0000 (14:12 -0700)]
mesa: Free linked shaders when deleting the shader program.

14 years agomesa: Free old linked shaders when relinking new shaders.
Eric Anholt [Wed, 18 Aug 2010 19:02:35 +0000 (12:02 -0700)]
mesa: Free old linked shaders when relinking new shaders.

14 years agoglsl2: Fix copy'n'paste hilarity leading to leaking in the refcount visitor.
Eric Anholt [Wed, 18 Aug 2010 18:42:42 +0000 (11:42 -0700)]
glsl2: Fix copy'n'paste hilarity leading to leaking in the refcount visitor.

14 years agoir_to_mesa: Free the ir_to_mesa temporary storage when we're done.
Eric Anholt [Wed, 18 Aug 2010 18:37:18 +0000 (11:37 -0700)]
ir_to_mesa: Free the ir_to_mesa temporary storage when we're done.

14 years agomesa: Hang the compiled shader off the shader->ir, not the shader.
Eric Anholt [Wed, 18 Aug 2010 18:36:09 +0000 (11:36 -0700)]
mesa: Hang the compiled shader off the shader->ir, not the shader.

Otherwise, with repeated program recompile, we never free the results
of the previous compile.

14 years agomesa: Make MESA_GLSL=dump include when compile/link fails, and the info log.
Eric Anholt [Wed, 18 Aug 2010 18:35:15 +0000 (11:35 -0700)]
mesa: Make MESA_GLSL=dump include when compile/link fails, and the info log.

14 years agoglsl: Fix scoping bug in if statements.
Kenneth Graunke [Wed, 18 Aug 2010 20:54:50 +0000 (13:54 -0700)]
glsl: Fix scoping bug in if statements.

Fixes glslparsertest/glsl2/scoping-01.frag (successfully compiled but
should've failed) and scoping-02.frag (assertion triggered).

14 years agoglsl: Refresh autogenerated file builtin_function.cpp.
Kenneth Graunke [Wed, 18 Aug 2010 20:20:58 +0000 (13:20 -0700)]
glsl: Refresh autogenerated file builtin_function.cpp.

14 years agogenerate_builtins.py: Clean up generated output a bit.
Kenneth Graunke [Wed, 18 Aug 2010 20:17:27 +0000 (13:17 -0700)]
generate_builtins.py: Clean up generated output a bit.

This should make it easier to diff the output, clean up some of the
insane whitespace, and make the strings a bit smaller.

We'll probably need to split up the prototype strings eventually, but
for now, this gets it under the 65K mark.

14 years agoglsl/builtins: Add forgotten hyperbolic trig builtins in 1.30 profiles.
Kenneth Graunke [Wed, 18 Aug 2010 20:16:50 +0000 (13:16 -0700)]
glsl/builtins: Add forgotten hyperbolic trig builtins in 1.30 profiles.

14 years agoir_constant_expression: Implement equal/notEqual for booleans.
Kenneth Graunke [Wed, 18 Aug 2010 19:06:25 +0000 (12:06 -0700)]
ir_constant_expression: Implement equal/notEqual for booleans.

Calls to equal(bvec, bvec) or notEqual(bvec, bvec) previously caused an
assertion.  Fixes piglit tests glsl-const-builtin-equal-bool and
glsl-const-builtin-notEqual-bool.

14 years agor600: implement SSG instruction
Andre Maasikas [Wed, 18 Aug 2010 11:14:38 +0000 (14:14 +0300)]
r600: implement SSG instruction

14 years agogallivm: Use proper index to lookup predicate register array.
michal [Wed, 18 Aug 2010 11:16:42 +0000 (13:16 +0200)]
gallivm: Use proper index to lookup predicate register array.

Doesn't fix anything, as those indices were both always 0.

14 years agor600: implement DP2 opcode
Andre Maasikas [Wed, 18 Aug 2010 08:57:28 +0000 (11:57 +0300)]
r600: implement DP2 opcode

14 years agoglsl2: Regenerate builtin_function.cpp.
Kenneth Graunke [Wed, 18 Aug 2010 06:47:51 +0000 (23:47 -0700)]
glsl2: Regenerate builtin_function.cpp.

14 years agoglsl2: Fix cut and paste error in EXT_texture_array builtins.
Kenneth Graunke [Wed, 18 Aug 2010 06:47:11 +0000 (23:47 -0700)]
glsl2: Fix cut and paste error in EXT_texture_array builtins.

Fixes fd.o bug #29629.

14 years agoglcpp: Refresh generated files.
Carl Worth [Wed, 18 Aug 2010 06:22:42 +0000 (23:22 -0700)]
glcpp: Refresh generated files.

After a recent change to glcpp-parse.y (adding "redefined macro" error).

14 years agoglcpp: Add support for "redefined macro" error.
Carl Worth [Wed, 18 Aug 2010 06:20:58 +0000 (23:20 -0700)]
glcpp: Add support for "redefined macro" error.

Carefully avoiding printing any error when the new definition matches
the existing definition.

This fixes the recently-added 088-redefine-macro-legitimate.c and
089-redefine-macro-error.c tests as well as glsparsertest/preprocess1
in piglit.

14 years agoglcpp: Add two new tests for testing redefined macros.
Carl Worth [Wed, 18 Aug 2010 06:19:01 +0000 (23:19 -0700)]
glcpp: Add two new tests for testing redefined macros.

The specification says that redefining a macro is an error, unless the
new definitions is identical to the old one, (identical replacement
lists but ignoring differing amounts of whitespace).

14 years agoglcpp: Allow standalone glcpp to accept a filename as first argument.
Carl Worth [Wed, 18 Aug 2010 06:13:56 +0000 (23:13 -0700)]
glcpp: Allow standalone glcpp to accept a filename as first argument.

This is useful for debugging the preprocessor.

14 years agor600g: add FRC, FLR, DDX and DDY
Dave Airlie [Wed, 18 Aug 2010 06:10:16 +0000 (16:10 +1000)]
r600g: add FRC, FLR, DDX and DDY

the first two are straight op2's and the DDX/DDY are taken from r600c.

14 years agor600g: add SGE and SLE opcodes
Dave Airlie [Wed, 18 Aug 2010 05:53:29 +0000 (15:53 +1000)]
r600g: add SGE and SLE opcodes

fixes fp-set-01 and glsl-fs-step

14 years agor600g: add TXB support
Dave Airlie [Wed, 18 Aug 2010 05:26:28 +0000 (15:26 +1000)]
r600g: add TXB support

fixes biased texturing tests

14 years agor600g: fix TXP vs TEX in shader.
Dave Airlie [Wed, 18 Aug 2010 05:05:34 +0000 (15:05 +1000)]
r600g: fix TXP vs TEX in shader.

Don't do perspective for TEX, and also copy input to a temporary for TEX

also add tex opcode names

14 years agoglcpp: Fix 064-version.c expected result to track recent change.
Carl Worth [Wed, 18 Aug 2010 05:23:43 +0000 (22:23 -0700)]
glcpp: Fix 064-version.c expected result to track recent change.

In commit 6be3a8b70af4ba4fa4d037d54ecf6d5f055edbc9, the #version directive
was fixed to stop generating a spurious newline. Here we simply update
the expected result for the single test which includes a #version directive.

14 years agoglcpp: Regenerated glcpp-lex.c from previous commit.
Carl Worth [Wed, 18 Aug 2010 05:22:13 +0000 (22:22 -0700)]
glcpp: Regenerated glcpp-lex.c from previous commit.

The previous commit changed glcpp-lex.l so we commit the resulting
generated file here.

14 years agoglcpp: Don't include the newline when discarding single-line comments
Carl Worth [Wed, 18 Aug 2010 05:17:09 +0000 (22:17 -0700)]
glcpp: Don't include the newline when discarding single-line comments

Matching the newline here meant having to do some redundant work here,
(incrementing line number, resetting column number, and returning a
NEWLINE token), that could otherwise simply be left to the existing rule
which matches a newline.

Worse, when the comment rule matches the newline as well, the parser
can lookahead and see a token for something that should actually be skipped.

For example, in a case like this:

#if 0 // comment here
fail
#else
win
#endif

Both fail and win appear in the output, (not that the condition is being
evaluated incorrectly---merely that one token after the comment's newline
was being lexed/parse regardless of the condition).

This commit fixes the above test case, (which is also remarkably similar
to 087-if-comments which now passes).

14 years agor600g: add two simple tgsi opcodes.
Dave Airlie [Wed, 18 Aug 2010 03:48:04 +0000 (13:48 +1000)]
r600g: add two simple tgsi opcodes.

makes glsl-fs-log2 and glsl1-integer division with uniform var pass

14 years agor600g: fix point size
Dave Airlie [Wed, 18 Aug 2010 03:29:41 +0000 (13:29 +1000)]
r600g: fix point size

fixes piglit pointAtten and point-sprite tests

14 years agoir_to_mesa: Allow ir_return in main().
Eric Anholt [Wed, 18 Aug 2010 02:41:27 +0000 (19:41 -0700)]
ir_to_mesa: Allow ir_return in main().

I didn't expect that this would really work, but it turns out there
are shaders in the wild that do it.

Fixes: (with swrast)
glsl-fs-main-return
glsl-vs-main-return

14 years agoi965: Throw a link error when we see a "return" in main().
Eric Anholt [Wed, 18 Aug 2010 02:51:00 +0000 (19:51 -0700)]
i965: Throw a link error when we see a "return" in main().

We'll need to use the HALT instruction to do this right, like returns
from other functions.

14 years agor600g: fixup pitch alignment like r600c.
Dave Airlie [Wed, 18 Aug 2010 01:49:51 +0000 (11:49 +1000)]
r600g: fixup pitch alignment like r600c.

This still needs work, passes tex3d, fbo-scissor-bitmap, scissor-bitmap

14 years agotranslate_sse: don't overwrite source buffer pointer
Luca Barbieri [Wed, 18 Aug 2010 02:14:43 +0000 (04:14 +0200)]
translate_sse: don't overwrite source buffer pointer

We were putting the source pointer in a register used as a temporary,
breaking all paths that don't read the data in a single instruction.

14 years agoglcpp: Add testcase for #if handling bug that breaks Savage2.
Eric Anholt [Wed, 18 Aug 2010 00:50:05 +0000 (17:50 -0700)]
glcpp: Add testcase for #if handling bug that breaks Savage2.

14 years agor600g: fix height calcs for miptree
Dave Airlie [Wed, 18 Aug 2010 00:50:19 +0000 (10:50 +1000)]
r600g: fix height calcs for miptree

h needs to be rounded up, this probably needs revisiting when we get
to tiling etc.

fixes fbo-generatemipmap-npot

14 years agor600g: emit texture level offset in CB/DB setup.
Dave Airlie [Tue, 17 Aug 2010 23:56:21 +0000 (09:56 +1000)]
r600g: emit texture level offset in CB/DB setup.

8 more piglit tests pass,
fbo-clearmipmap, fbo-copyteximage, fbo-generatemipmap,
fbo-generatemipmap-nonsquare, fbo-generatemipmap-scissor,
fbo-generatemipmap-viewport, gen-teximage, gen-texsubimage

14 years agoir_to_mesa: Fix implementation of ir_binop_equal, ir_binop_notequal.
Eric Anholt [Wed, 18 Aug 2010 00:24:42 +0000 (17:24 -0700)]
ir_to_mesa: Fix implementation of ir_binop_equal, ir_binop_notequal.

These binops are the vector-to-bool comparisons, not vec-to-bvec.  We
likely want both operations avilable as expression, since 915 and 965
FS naturally does the vector version, while 965 VS can also naturally
do the scalar version.  However, we can save that until later.

Fixes:
glsl-fs-vec4-operator-equal.shader_test
glsl-fs-vec4-operator-notequal.shader_test
glsl-vs-vec4-operator-equal.shader_test
glsl-vs-vec4-operator-notequal.shader_test

14 years agoi915: Add support for SSG opcode.
Eric Anholt [Tue, 17 Aug 2010 23:12:15 +0000 (16:12 -0700)]
i915: Add support for SSG opcode.

Fixes glsl-fs-sign and many of the tests of trig builtins.

14 years agoi915: Add support for reading output regs in the FS.
Eric Anholt [Tue, 17 Aug 2010 22:51:34 +0000 (15:51 -0700)]
i915: Add support for reading output regs in the FS.

Fixes glsl-unused-varying and many others, since we produce an output
reg read any time gl_FragColor is written inside an if statement.

14 years agoi915: Add support for OPCODE_DP2.
Eric Anholt [Tue, 17 Aug 2010 22:42:40 +0000 (15:42 -0700)]
i915: Add support for OPCODE_DP2.

Fixes glsl-fs-dot-vec2.

14 years agoi915: Enable ARB_fragment_shader by default.
Eric Anholt [Tue, 17 Aug 2010 22:07:22 +0000 (15:07 -0700)]
i915: Enable ARB_fragment_shader by default.

Now that we have glsl2 with if flattening in place, most shaders will
just work.  Remaining failing shaders will mostly be due to loop
unrolling (in progress), some possible if flattening failures in
inlining functions (planning on fixing), and the register/instruction
count limits.

While the GLSL and GLSL-ES specs say that shaders shouldn't fail to
compile/link due to register/instruction limits, in practice we're not
the first vendor to expose GLSL on hardware with these limitations.
The benefit to application developers of providing a better language
for GPU programming is greater than the pain of having to handle
instruction limits (which they had to for ARB_fp on this hardware
anyway)

14 years agoglsl2: Fix transpose of rows and cols
Ian Romanick [Tue, 17 Aug 2010 22:57:48 +0000 (15:57 -0700)]
glsl2: Fix transpose of rows and cols

This error led to an assertion failure for some constructors of
non-square matrices.  It only occured in matrices where the number of
columns was greater than the number of rows.  It didn't even always
occur on those.

Fixes piglit glslparsertest case constructor-16.vert.

14 years agor600c: Handle reads from PROGRAM_OUTPUT
Henri Verbeet [Tue, 17 Aug 2010 22:25:32 +0000 (18:25 -0400)]
r600c: Handle reads from PROGRAM_OUTPUT

with glsl2, reads from outputs are legal

14 years agolinker: Demote user-defined varyings in the VS-only case
Ian Romanick [Tue, 17 Aug 2010 21:55:50 +0000 (14:55 -0700)]
linker: Demote user-defined varyings in the VS-only case

Fixes piglit test case glsl-vs-ff-frag and bugzilla #29623.

14 years agoprog_optimize: Only merge writes to temporary registers
Benjamin Segovia [Tue, 17 Aug 2010 03:40:09 +0000 (20:40 -0700)]
prog_optimize: Only merge writes to temporary registers

In one optimization pass, register files may have been messed therefore
merging instructions which use the same index in two different register
files.

14 years agor600g: fix fake pixel output
Jerome Glisse [Tue, 17 Aug 2010 21:23:20 +0000 (17:23 -0400)]
r600g: fix fake pixel output

Signed-off-by: Jerome Glisse <jglisse@redhat.com>
14 years agoi965: Add support for DP2 in the VS.
Eric Anholt [Tue, 17 Aug 2010 20:59:08 +0000 (13:59 -0700)]
i965: Add support for DP2 in the VS.

Fixes glsl-vs-dot-vec2.

14 years agoglsl: When doing algebraic simplification, make sure the type still matches.
Eric Anholt [Tue, 17 Aug 2010 20:24:50 +0000 (13:24 -0700)]
glsl: When doing algebraic simplification, make sure the type still matches.

When simplifying (vec4(1.0) / (float(x))) to rcp(float(x)), we forgot
to produce a vec4, angering ir_validate when starting alien-arena.

Fixes:
glsl-algebraic-add-zero-2
glsl-algebraic-div-one-2
glsl-algebraic-mul-one-2
glsl-algebraic-sub-zero-3
glsl-algebraic-rcp-sqrt-2

14 years agoglsl: Make ir_algebraic new expressions allocate out of the parent.
Eric Anholt [Tue, 17 Aug 2010 20:27:44 +0000 (13:27 -0700)]
glsl: Make ir_algebraic new expressions allocate out of the parent.

This could reduce the amount of memory used by a shader tree after
optimization, and increases consistency with other passes.

14 years agoir_constant: Don't assert on out-of-bounds array accesses
Ian Romanick [Tue, 17 Aug 2010 19:57:28 +0000 (12:57 -0700)]
ir_constant: Don't assert on out-of-bounds array accesses

Several optimization paths, including constant folding, can lead to
accessing an ir_constant array with an out of bounds index.  The GLSL
spec lets us produce "undefined" results, but it does not let us
crash.

Fixes piglit test case glsl-array-bounds-01 and glsl-array-bounds-03.

14 years agomesa: Dump shader source before validating the shader.
Eric Anholt [Tue, 17 Aug 2010 19:37:49 +0000 (12:37 -0700)]
mesa: Dump shader source before validating the shader.

This will make extracting source to produce minimal testcases for
shader compile issues easier.

14 years agor600c: fix dword miscount in blit emit code
Alex Deucher [Tue, 17 Aug 2010 14:42:06 +0000 (10:42 -0400)]
r600c: fix dword miscount in blit emit code

14 years agotargets/egl: Link with DRI_LIB_DEPS.
Chia-I Wu [Tue, 17 Aug 2010 11:24:18 +0000 (19:24 +0800)]
targets/egl: Link with DRI_LIB_DEPS.

Use DRI_LIB_DEPS when linking GL/GLES state trackers.  This fixes
missing talloc symbol errors, and is hopefully more future proof.

14 years agogallivm: Fix and re-enable MMX-disabling code
nobled [Mon, 16 Aug 2010 16:46:14 +0000 (16:46 +0000)]
gallivm: Fix and re-enable MMX-disabling code

Signed-off-by: José Fonseca <jfonseca@vmware.com>
14 years agomesa: fix es1/2 build hopefully
Dave Airlie [Tue, 17 Aug 2010 10:54:45 +0000 (20:54 +1000)]
mesa: fix es1/2 build hopefully

needed to add cpp rules and includes properly for es1/es2

14 years agor300g: fix context destroy under hyperz
Dave Airlie [Tue, 17 Aug 2010 09:01:18 +0000 (19:01 +1000)]
r300g: fix context destroy under hyperz

we were destroying the mm before unrefing all the objects, so segfault.

Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agoscons: Add -fno-common for 64-bit builds on Mac OS X.
Vinson Lee [Tue, 17 Aug 2010 07:20:58 +0000 (00:20 -0700)]
scons: Add -fno-common for 64-bit builds on Mac OS X.

This option is also needed for 64-bit builds if llvm is enabled.
Other the build fails during linking.

14 years agor600g: kill event type magic number in winsys
Dave Airlie [Tue, 17 Aug 2010 05:16:53 +0000 (15:16 +1000)]
r600g: kill event type magic number in winsys

these events have names, use them.

14 years agoconfigs: Add -ltalloc to linux-dri.
Vinson Lee [Tue, 17 Aug 2010 05:41:44 +0000 (22:41 -0700)]
configs: Add -ltalloc to linux-dri.

Fixes build after glsl2 branch merge, which added talloc dependency.

14 years agor600g: add user clip plane support.
Dave Airlie [Tue, 17 Aug 2010 03:40:15 +0000 (13:40 +1000)]
r600g: add user clip plane support.

Apart from the fact that the radeon.h/r600_states.h editing is a nightmare, this
wasn't so bad.

passes piglit user-clip test now also trivial tests.

Signed-off-by: Dave Airlie <airlied@redhat.com>
14 years agoi965: Use the implied move available in most brw_wm_emit brw_math() calls.
Eric Anholt [Tue, 17 Aug 2010 02:18:11 +0000 (19:18 -0700)]
i965: Use the implied move available in most brw_wm_emit brw_math() calls.

This saves an extra message reg move in the program, though I'm not
clear on whether it will have any performance impact other than cache
footprint.  It will also fix those math calls on Sandybridge, where
the brw_eu_emit.c brw_math() support relies on the implied move being
used.

14 years agoi965: Add disasm for Compr4 instruction compression.
Eric Anholt [Fri, 9 Jul 2010 00:05:42 +0000 (17:05 -0700)]
i965: Add disasm for Compr4 instruction compression.