OSDN Git Service
Brian Paul [Thu, 2 Sep 2010 20:11:53 +0000 (14:11 -0600)]
st/glx: added some comments
Luca Barbieri [Fri, 3 Sep 2010 20:06:41 +0000 (22:06 +0200)]
nvfx: implement LIT in fp
Ian Romanick [Thu, 2 Sep 2010 21:53:17 +0000 (14:53 -0700)]
glsl2: Use as_constant some places instead of constant_expression_value
The places where constant_expression_value are still used in loop
analysis are places where a new expression tree is created and
constant folding won't have happened. This is used, for example, when
we try to determine the maximal loop iteration count.
Based on review comments by Eric. "...rely on constant folding to
have done its job, instead of going all through the subtree again when
it wasn't a constant."
Ian Romanick [Fri, 27 Aug 2010 23:22:36 +0000 (16:22 -0700)]
glsl2: Allow copy / constant propagation into array indices
Ian Romanick [Fri, 27 Aug 2010 20:59:49 +0000 (13:59 -0700)]
glsl2: Add module to perform simple loop unrolling
Ian Romanick [Fri, 27 Aug 2010 22:41:20 +0000 (15:41 -0700)]
glsl2: Track the number of ir_loop_jump instructions that are in a loop
Ian Romanick [Fri, 27 Aug 2010 20:53:25 +0000 (13:53 -0700)]
ir_expression: Add static operator_string method
I've used this in quite a few debug commits that never reached an
up-stream tree.
Ian Romanick [Fri, 27 Aug 2010 20:53:56 +0000 (13:53 -0700)]
exec_node: Add insert_before that inserts an entire list
Ian Romanick [Fri, 27 Aug 2010 18:26:08 +0000 (11:26 -0700)]
glsl2: Eliminate zero-iteration loops
Ian Romanick [Thu, 26 Aug 2010 23:45:22 +0000 (16:45 -0700)]
glsl2: Perform initial bits of loop analysis during compilation
Ian Romanick [Thu, 26 Aug 2010 23:43:57 +0000 (16:43 -0700)]
glsl2: Add module to suss out loop control variables from loop analysis data
This is the next step on the road to loop unrolling
Ian Romanick [Thu, 26 Aug 2010 22:58:33 +0000 (15:58 -0700)]
glsl2: Add module to analyze variables used in loops
This is the first step eventually leading to loop unrolling.
Ian Romanick [Thu, 26 Aug 2010 22:49:33 +0000 (15:49 -0700)]
ir_to_mesa: Handle loops with loop controls set
The downside of our talloc usage is that we can't really make static
(i.e., not created with new) instances of our IR types. This leads to
a lot of unnecessary dynamic allocation in this patch.
Ian Romanick [Thu, 26 Aug 2010 22:22:06 +0000 (15:22 -0700)]
ir_validate: Validate loop control fields in ir_loop
Ian Romanick [Thu, 26 Aug 2010 22:11:26 +0000 (15:11 -0700)]
glsl2: Add cmp field to ir_loop
This reprents the type of comparison between the loop induction
variable and the loop termination value.
Ian Romanick [Thu, 5 Aug 2010 22:29:24 +0000 (15:29 -0700)]
glsl2: Set a flag when visiting the assignee of an assignment
Ian Romanick [Tue, 17 Aug 2010 01:02:11 +0000 (18:02 -0700)]
exec_list: Add pop_head
Ian Romanick [Thu, 12 Aug 2010 21:55:48 +0000 (14:55 -0700)]
ir_print_visitor: Print empty else blocks more compactly
Luca Barbieri [Fri, 3 Sep 2010 18:57:44 +0000 (20:57 +0200)]
nvfx: fix division by zero in vp-ignore-input
Luca Barbieri [Fri, 3 Sep 2010 18:54:48 +0000 (20:54 +0200)]
nvfx: report correct max lodbias
Fixes piglit lodbias
Luca Barbieri [Fri, 3 Sep 2010 18:36:29 +0000 (20:36 +0200)]
nvfx: remove message
Luca Barbieri [Fri, 3 Sep 2010 16:31:18 +0000 (18:31 +0200)]
nvfx: support indirect addressing in vps
Negative or huge offsets not yet supported.
Alex Deucher [Fri, 3 Sep 2010 16:13:47 +0000 (12:13 -0400)]
r600c: add proper returns for some evergreen functions
these weren't checked anyway.
Fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=29999
Luca Barbieri [Fri, 3 Sep 2010 13:44:27 +0000 (15:44 +0200)]
nvfx: fix support for more than 8 texture units (fixes etqw crash)
Dave Airlie [Fri, 3 Sep 2010 09:37:52 +0000 (19:37 +1000)]
r600g: fix segfault in state after rework
probably can improve this a bit.
Alex Deucher [Fri, 3 Sep 2010 05:13:41 +0000 (01:13 -0400)]
r600c: emit DB_HTILE_DATA_BASE on evergreen
Make the hw happy.
Dave Airlie [Fri, 3 Sep 2010 04:12:38 +0000 (14:12 +1000)]
r600g: refactor sample states into a reusable struct.
I will not cut-n-paste.
I will not cut-n-paste.
I will not cut-n-paste.
Dave Airlie [Fri, 3 Sep 2010 03:53:39 +0000 (13:53 +1000)]
r600g: reduce size of r600 context structure to !insane
Its now about 7.8k, and might actually fit in a cache.
Dave Airlie [Fri, 3 Sep 2010 03:07:40 +0000 (13:07 +1000)]
r600g: add texture border state.
Okay I finally wrapped my head around what r600_context_state is meant to be,
maybe I should just rename all the structs so that have distinct names.
I've no idea however why 16 is a good magic number for R600_MAX_RSTATE.
Dave Airlie [Fri, 3 Sep 2010 02:01:59 +0000 (12:01 +1000)]
r600g: deref old driver states for set entry points.
Dave Airlie [Fri, 3 Sep 2010 01:55:36 +0000 (11:55 +1000)]
r600g: drop r600_bind_state.
This was another ugly function that really wasn't needed.
The 3 calls to it from the gallium api were shorter than it,
and all the calls from the set_ functions were pointless.
Dave Airlie [Fri, 3 Sep 2010 01:35:08 +0000 (11:35 +1000)]
r600g: kill r600_context_state function
having some sort of locality of code really matters, just create
and setup state at time. Not sure if this is just further polishing of a bad thing,
but at least it makes it more readable.
Dave Airlie [Fri, 3 Sep 2010 00:55:02 +0000 (10:55 +1000)]
r600g: move lots of state inline helpers to separate header.
this gets them out of sight of the main codeflow.
Vinson Lee [Fri, 3 Sep 2010 00:00:53 +0000 (17:00 -0700)]
draw: Include missing headers in draw_vs_aos.h.
Include tgsi_exec.h for TGSI_EXEC_NUM_TEMPS.
Include draw_vs.h for draw_vs_varient.
Dave Airlie [Thu, 2 Sep 2010 23:39:04 +0000 (09:39 +1000)]
r600g: drop magic numbers in depth state.
this also fixes occulsion queries.
Vinson Lee [Thu, 2 Sep 2010 23:30:34 +0000 (16:30 -0700)]
util: Include missing header in u_linear.h.
Include p_compiler.h for size_t and boolean symbols.
Vinson Lee [Thu, 2 Sep 2010 23:12:58 +0000 (16:12 -0700)]
mesa: Fix printf-like warning.
Vinson Lee [Thu, 2 Sep 2010 23:03:32 +0000 (16:03 -0700)]
mesa: Fix printf-like warnings.
Jerome Glisse [Thu, 2 Sep 2010 20:53:33 +0000 (16:53 -0400)]
r600g: force unbind of previously bind sampler/sampler_view
Previously bind sampler/sampler_view can be converted and endup
overwritting the current state we want to schedule. Example :
bind texA texB to sampler_view[0] & sampler_view[1], render,
bind texB to sampler_view[0] render. Now state associated to
texB are set to configure sampler_view slot 0, but as we don't
unbind sampler_view[1] still point to texB state so we end up
with sampler_view[1] overwritting sampler_view[0], which gives
wrong rendering if next rendering bind texA to sampler_view[0],
it will endup as texB is bound to sampler_view[0]. If you are
not confuse at that point give me a call i will be buying you
beer.
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Brian Paul [Thu, 2 Sep 2010 18:58:58 +0000 (12:58 -0600)]
st/mesa: clamp gl_constants::MaxVarying against MAX_VARYING
Don't try to use more generic varying vars than core Mesa supports.
Fixes fd.o bug 29959.
Ian Romanick [Thu, 2 Sep 2010 17:11:54 +0000 (10:11 -0700)]
glsl2: Update TODO file
Jerome Glisse [Thu, 2 Sep 2010 15:32:32 +0000 (11:32 -0400)]
r600g: fix memory/bo leak
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Brian Paul [Thu, 2 Sep 2010 13:58:28 +0000 (07:58 -0600)]
mesa: fix code generation for ir_unop_sqrt
The CMP instruction needed to be flipped to properly handle
operand==0.
Fixes fd.o bug 29923.
Brian Paul [Thu, 2 Sep 2010 13:57:16 +0000 (07:57 -0600)]
mesa: fix some printf warnings with casts
Dave Airlie [Thu, 2 Sep 2010 11:25:42 +0000 (21:25 +1000)]
r600g: fix thinko in shadow code.
spotted by taiu on irc
Dave Airlie [Thu, 2 Sep 2010 06:39:32 +0000 (16:39 +1000)]
r600g: fix logicop, the 3d ROP is the 2D rop shifted twice.
Dave Airlie [Thu, 2 Sep 2010 05:51:23 +0000 (15:51 +1000)]
r600g: fix depth texture tests
Kenneth Graunke [Thu, 2 Sep 2010 03:03:17 +0000 (20:03 -0700)]
glsl: Apply implicit conversions to structure constructor parameters.
The code for handling implicit conversions should probably get
refactored, but for now, this is easy.
Fixes piglit test constructor-26.vert.
Kenneth Graunke [Thu, 2 Sep 2010 02:54:27 +0000 (19:54 -0700)]
glsl: Convert constant record constructor parameters to ir_constants.
I'm not sure if this is strictly necessary, but it seems wise.
Kenneth Graunke [Thu, 2 Sep 2010 02:44:25 +0000 (19:44 -0700)]
glsl: Reject structure constructors that have too many arguments.
Fixes piglit test constructor-27.vert.
Ian Romanick [Wed, 1 Sep 2010 21:16:53 +0000 (14:16 -0700)]
glsl2: Remove unnecessary glsl_symbol_table::get_function parameter return_constructors
Now that constructors are not generated as functions or stored in the
symbol table, there is no need to flag whether or not constructors
should be returned.
Ian Romanick [Wed, 1 Sep 2010 21:12:24 +0000 (14:12 -0700)]
glsl2: Remove unused method glsl_type::generate_constructor
Ian Romanick [Wed, 1 Sep 2010 21:10:39 +0000 (14:10 -0700)]
glsl2: Remove unused 'constructor' parameter from glsl_symbol_table::add_type
Ian Romanick [Wed, 1 Sep 2010 21:08:08 +0000 (14:08 -0700)]
glsl2: Don't generate constructor functions for structures
Ian Romanick [Wed, 1 Sep 2010 20:46:04 +0000 (13:46 -0700)]
glsl2: Emit structure constructors inline
Fixes piglit test cases glsl-[fv]s-all-0[12].
Dave Airlie [Thu, 2 Sep 2010 03:32:25 +0000 (13:32 +1000)]
r600g: add missing vertex fetch formats to the translation table.
fixes at least 2 more piglits.
Jerome Glisse [Thu, 2 Sep 2010 01:57:43 +0000 (21:57 -0400)]
r600g: fix binding of same texture to several target slot
One can bind same texture or sampler to different slot,
each slot needs it own state. The solution implemented
here is not exactly beautifull or optimal need to think
to somethings better.
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Kenneth Graunke [Wed, 1 Sep 2010 23:10:01 +0000 (16:10 -0700)]
glsl: Fix write mask in matrix-from-matrix constructors.
If the matrix being constructed was larger than the source matrix, it
would overwrite the lower-right part of the matrix with the wrong
values, rather than leaving it as the identity matrix.
For example, constructing a mat4 from a mat2 should only use a writemask
of "xy" when copying from the source, but was using "xyzw".
Fixes the code generated by piglit test constructor-23.vert.
Kenneth Graunke [Wed, 1 Sep 2010 22:37:13 +0000 (15:37 -0700)]
glsl: Add proper handling for constant matrix-from-matrix constructors.
Fixes piglit test case constructor-21.vert and changes
constructor-22.vert to give the correct output.
Kenneth Graunke [Wed, 1 Sep 2010 22:31:06 +0000 (15:31 -0700)]
glsl: Move generate_constructor_(matrix|vector) to ir_constant ctor.
Kenneth Graunke [Wed, 1 Sep 2010 22:04:57 +0000 (15:04 -0700)]
ast_function: Fix check for "too few components".
This was triggering even for matrix-from-matrix constructors. It is
perfectly legal to construct a mat3 from a mat2 - the rest will be
filled in by the identity matrix.
Changes piglit test constructor-23.vert from FAIL to PASS, but the
generated code is incorrect.
Kenneth Graunke [Wed, 1 Sep 2010 21:49:53 +0000 (14:49 -0700)]
ast_function: Remove bogus cases from generate_constructor_matrix.
There are no integer matrix types, so switching on them is silly.
Dave Airlie [Thu, 2 Sep 2010 01:16:31 +0000 (11:16 +1000)]
r600g: fix incorrect state naming in pipe_sampler vs pipe_sampler_view
fixes problems in valgrind with uninitialised values.
Eric Anholt [Wed, 1 Sep 2010 23:06:32 +0000 (16:06 -0700)]
ir_to_mesa: Load all the STATE_VAR elements of a builtin uniform to a temp.
Like the constant handling and the handling of other uniforms, we add
the whole thing to the Parameters, avoiding messy, incomplete logic
for adding just the elements of a builting uniform that get used.
This means that a driver that relies only on ParameterValues[] for its
parameters will have an increased parameter load, but drivers
generally don't do that (since they have other params they need to
handle, too).
Fixes glsl-fs-statevar-call (testcase for Ember). Bug #29687.
Regresses glsl-vs-array-04 on 965. Thanks to a slight change in
register allocation, this test of undefined behavior now wraps around
the register space and unexpectedly reads the constant value it's
trying to compare to. The test should probably not look at the
resulting color, since behavior is undefined.
Eric Anholt [Wed, 1 Sep 2010 21:56:43 +0000 (14:56 -0700)]
ir_to_mesa: Add a little helper for emitting link failure messages.
Eric Anholt [Wed, 1 Sep 2010 21:46:22 +0000 (14:46 -0700)]
mesa: Fix many printf-like warnings.
Most of these are just typecasting to long to match the arg type. I
don't really care too much about getting a GLsizei or whatever
appropriate type in. However, there were a number of real bugs, like
missing arguments or passing floats to integer format specifiers. My
favorite: printflike("%s, argument") is missing an argument.
Eric Anholt [Wed, 1 Sep 2010 21:39:50 +0000 (14:39 -0700)]
mesa: Add __printf__ attribute to printf-like functions to get warnings.
Dave Airlie [Wed, 1 Sep 2010 23:45:37 +0000 (09:45 +1000)]
mesa/st: remove check for buffer/elements = 0
shown by the glsl-vs-point-size failing on r600g.
the test passes on softpipe and I get a full piglit test run completing on r600g.
Jerome Glisse [Wed, 1 Sep 2010 22:09:37 +0000 (18:09 -0400)]
r600g: silence compiler warning
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Jerome Glisse [Wed, 1 Sep 2010 22:04:38 +0000 (18:04 -0400)]
r600g: refix db/cb state
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Brian Paul [Wed, 1 Sep 2010 19:21:51 +0000 (13:21 -0600)]
mesa: fix out of bounds memory read in mipmap gen code
Out of bounds reads could happen for reducing WxH to WxH/2
or WxH to W/2xH.
Fixes fd.o bug 29918.
Brian Paul [Wed, 1 Sep 2010 18:46:57 +0000 (12:46 -0600)]
mesa: more prog_execute.c debug code
Kenneth Graunke [Wed, 1 Sep 2010 18:47:52 +0000 (11:47 -0700)]
glsl: Add forgotten implementations of equal/notEqual on bvecs.
Alex Deucher [Wed, 1 Sep 2010 17:57:52 +0000 (13:57 -0400)]
r600g: fix up default state differences between r6xx and r7xx
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Brian Paul [Wed, 1 Sep 2010 18:39:04 +0000 (12:39 -0600)]
st/glx: re-order destruction of buffers, visuals
Free the buffers before the visuals. Fixes valgrind warning
reported in fd.o bug 29919.
Eric Anholt [Wed, 1 Sep 2010 17:41:36 +0000 (10:41 -0700)]
i965: DP2 produces a scalar result like DP3, DP4, etc.
Fixes glsl-fs-dot-vec2-2.
Ian Romanick [Wed, 1 Sep 2010 17:13:21 +0000 (10:13 -0700)]
glsl2: Perform algebraic simplifications on logical binary operators
Reduces glsl-vs-all-01 from 42 Mesa IR instructions (including the
END) to 17.
Ian Romanick [Wed, 1 Sep 2010 17:12:55 +0000 (10:12 -0700)]
glsl2: Allow ir_constant::zero to create boolean constants
Jerome Glisse [Mon, 30 Aug 2010 21:56:59 +0000 (17:56 -0400)]
r600g: avoid dynamic allocation of states
Make state statically allocated, this kills a bunch of code
and avoid intensive use of malloc/free. There is still a lot
of useless duplicate function wrapping that can be kill. This
doesn't improve yet performance, needs to avoid memcpy states
in radeon_ctx_set_draw and to avoid rebuilding vs_resources,
dsa, scissor, cb_cntl, ... states at each draw command.
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Jerome Glisse [Wed, 1 Sep 2010 17:04:42 +0000 (13:04 -0400)]
Revert "Revert "r600g: precompute some of the hw state""
This reverts commit
1fa7245c348cb7aced81f1672140f64cb6450e2f.
Conflicts:
src/gallium/drivers/r600/r600_state.c
Patrice Mandin [Wed, 1 Sep 2010 16:12:11 +0000 (18:12 +0200)]
nouveau/nvfx: Remove enforcement of bit depth being same as front buffer
Signed-off-by: Patrice Mandin <patmandin@gmail.com>
Vladimir Vukicevic [Wed, 1 Sep 2010 14:54:21 +0000 (08:54 -0600)]
mesa: initialize dummy framebuffer and renderbuffer mutexes
See fd.o bug 29909.
Signed-off-by: Brian Paul <brianp@vmware.com>
Jon TURNEY [Fri, 27 Aug 2010 21:52:41 +0000 (22:52 +0100)]
Cygwin: Adjust mklib so -linker and -cplusplus options are processed more like they are for linux
It looks like we were ignoring -linker when -noprefix wasn't present, and
when -noprefix was present, -linker was mandatory and -cplusplus ignored.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Brian Paul <brianp@vmware.com>
Jon TURNEY [Sat, 24 Jul 2010 11:05:34 +0000 (12:05 +0100)]
Cygwin: Teach mklib/minstall to properly install libraries on cygwin
Teach mklib/minstall more about cygwin so libraries are properly installed
Have mklib install the .dll into the lib/ staging directory as well
Have minstall install the .dll into PREFIX/bin at the same time as
installing the .dll.a link library into PREFIX/lib
mklib uses a '-' rather than a '.' as the separator before the version
number in library names on cygwin. Change the install globs so they match
library names like that.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Brian Paul <brianp@vmware.com>
Jon TURNEY [Sat, 24 Jul 2010 11:06:23 +0000 (12:06 +0100)]
Cygwin: Change mklib not to report the full archname on cygwin
Change mklib not to report the full archname when building a library for cygwin
(which is something like 'CYGWIN_NT-5.1' or 'CYGWIN_NT-6.1-WOW64' and kind of
confusing), but just 'CYGWIN'.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Brian Paul <brianp@vmware.com>
Jon TURNEY [Fri, 19 Feb 2010 22:38:57 +0000 (22:38 +0000)]
Cygwin: Have mklib exit with error code if link fails
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Brian Paul <brianp@vmware.com>
Ian Romanick [Wed, 1 Sep 2010 13:34:58 +0000 (06:34 -0700)]
glsl2: Disallow function declarations within function definitions in GLSL 1.20
The GLSL 1.20 spec specifically disallows this, but it was allowed in
GLSL 1.10.
Fixes piglit test cases local-function-0[13].frag and bugzilla #29921.
Andre Maasikas [Wed, 1 Sep 2010 06:19:51 +0000 (09:19 +0300)]
r600: cube mipmap levels are aligned to 8 faces only starting from r7xx
Dave Airlie [Wed, 1 Sep 2010 04:28:55 +0000 (14:28 +1000)]
r600g: correct cb/zb offset emits.
This fixes fbo-3d and fbo-cubemap
Dave Airlie [Wed, 1 Sep 2010 04:56:04 +0000 (14:56 +1000)]
Revert "r600g: precompute some of the hw state"
This reverts commit
de0b76cab22caa9fc7260f80acb8f151ccced6c5, its pre-computes the texture state wrong,
you can't just use an array of levels, since you can have FBOs to depth texture slices inside a level as well
it would get really messy quickly. Probably need to split commits like this up into pieces for each piece
of state, so we can revert bits easier in case of regressions.
This also break 5 piglit tests, and valgrind starts to warn about invalid read/writes after this.
Dave Airlie [Wed, 1 Sep 2010 03:54:38 +0000 (13:54 +1000)]
r600g: fix typo causing segfault.
fixes warning that
r600_blit.c: In function ‘r600_resource_copy_region’:
r600_blit.c:136: warning: passing argument 1 of ‘util_resource_copy_region’ from incompatible pointer type
and also 7 more piglit tests.
Dave Airlie [Wed, 1 Sep 2010 01:37:39 +0000 (11:37 +1000)]
r600g: fix glean texCube and shadows.
add cube and shadow support to the texture code.
Brian Paul [Wed, 1 Sep 2010 01:14:18 +0000 (19:14 -0600)]
gallivm: fix bug in nested conditionals
This, plus the previous commit fix fd.o bug 29806.
Brian Paul [Wed, 1 Sep 2010 01:13:09 +0000 (19:13 -0600)]
glsl2: fix bug in atan(y, x) function
When x==0, the result was wrong. Fixes piglit glsl-fs-atan-1.shader_test
Eric Anholt [Tue, 31 Aug 2010 20:02:59 +0000 (13:02 -0700)]
ir_to_mesa: When emitting a pixel kill, flag that we did so.
Both i965 and swrast rely on UsesKill to determine whether to do early
depth writes. Fixes glsl-fs-discard-02.
Bug #29835.
Keith Whitwell [Tue, 24 Aug 2010 22:05:57 +0000 (23:05 +0100)]
llvmpipe: slightly simplify build_mask
Keith Whitwell [Sun, 22 Aug 2010 21:56:54 +0000 (22:56 +0100)]
llvmpipe: combine linear mask calculation
Keith Whitwell [Sun, 15 Aug 2010 17:19:52 +0000 (18:19 +0100)]
llvmpipe: intrinsics versions of build_mask functions
Ian Romanick [Tue, 31 Aug 2010 21:44:13 +0000 (14:44 -0700)]
glsl2: Write vector constructor constants in a single assignment
Make two passes over the constructor parameters. Write all of the
constants in a single write, then write the non-constants one at a
time. This causes the fragment shader
varying float g;
void main()
{
gl_FragColor = vec4(0.0, g, 0.0, 1.0);
}
to generate
(function main
(signature void (parameters )
(
(declare (temporary ) vec4 vec_ctor@0x8580058)
(assign (constant bool (1)) (xzw) (var_ref vec_ctor@0x8580058) (constant vec4 (0.000000 0.000000 0.000000 1.000000)) )
(assign (constant bool (1)) (y) (var_ref vec_ctor@0x8580058) (swiz xxxx (var_ref g@0x8580218) ))
(assign (constant bool (1)) (xyzw) (var_ref gl_FragColor@0x84d32a0) (var_ref vec_ctor@0x8580058) )
))
)
instead of
(function main
(signature void (parameters )
(
(declare (temporary ) vec4 vec_ctor@0x8580058)
(assign (constant bool (1)) (x) (var_ref vec_ctor@0x8580058) (constant vec4 (0.000000 0.000000 0.000000 1.000000)) )
(assign (constant bool (1)) (y) (var_ref vec_ctor@0x8580058) (swiz xxxx (var_ref g@0x8580218) ))
(assign (constant bool (1)) (z) (var_ref vec_ctor@0x8580058) (constant vec4 (0.000000 0.000000 0.000000 1.000000)) )
(assign (constant bool (1)) (w) (var_ref vec_ctor@0x8580058) (constant vec4 (0.000000 0.000000 0.000000 1.000000)) )
(assign (constant bool (1)) (xyzw) (var_ref gl_FragColor@0x84d32a0) (var_ref vec_ctor@0x8580058) )
))
)
A similar optimization could be done for matrix constructors, but it
is a little more complicate there.
Eric Anholt [Tue, 31 Aug 2010 18:16:09 +0000 (11:16 -0700)]
ir_to_mesa: Sort the uniform list we're adding to Parameters[] order.
Fixes glsl-uniform-linking-1 and failure to link a shader in Unigine.
An alternative here would be to just ditch using _mesa_add_parameter
and build the initial params list on our own, but that would require
two walks of the list as well.
Bug #29822