Paul Berry [Mon, 1 Aug 2011 20:06:06 +0000 (13:06 -0700)]
Revert "glsl: Skip processing the first function's body in do_dead_functions()."
opt_dead_functions contained a shortcut to skip processing the first
function's body, based on the assumption that IR functions are
topologically sorted, with callees always coming before their callers
(therefore the first function cannot contain any calls).
This assumption turns out not to be true in general. For example, the
following code snippet gets translated to IR that violates this
assumption:
void f();
void g();
void f() { g(); }
void g() { ... }
In practice, the shortcut didn't cause bugs because of a coincidence
of the circumstances in which opt_dead_functions is called:
(a) we do inlining right before dead function elimination, and
inlining (when successful) eliminates all calls.
(b) for user-defined functions, inlining is always successful, because
previous optimization passes (during compilation) have reduced
them to a form that is eligible for inlining.
(c) the function that appears first in the IR can't possibly call a
built-in function, because built-in functions are always emitted
before the function that calls them.
It seems unnecessarily fragile to have opt_dead_functions depend on
these coincidences. And the next patch in this series will break (c).
So I'm reverting the shortcut. The consequence will be a slight
increase in link time for complex shaders.
This reverts commit
c75427f4c8767e131e5fb3de44fbc9d904cb992d.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Bryan Cain [Sun, 7 Aug 2011 19:15:35 +0000 (14:15 -0500)]
st/mesa: inline st_prepare_fragment_program in st_translate_fragment_program
This reverts an unnecessary part of commit
4683529048ee and fixes misrendering
and an assertion failure in Cogs.
Fixes freedesktop.org bug 39888.
Reviewed-by: Brian Paul <brianp@vmware.com>
Brian Paul [Mon, 8 Aug 2011 15:01:13 +0000 (09:01 -0600)]
glext: upgrade to version 72
Brian Paul [Mon, 8 Aug 2011 15:00:57 +0000 (09:00 -0600)]
glsl: silence warning about trailing comma in enum list
Brian Paul [Mon, 8 Aug 2011 15:00:06 +0000 (09:00 -0600)]
gallium: silence warnings about trailing commas in enum lists
Brian Paul [Fri, 5 Aug 2011 21:01:41 +0000 (15:01 -0600)]
mesa: whitespace changes
Christoph Bumiller [Sun, 7 Aug 2011 13:34:07 +0000 (15:34 +0200)]
d3d1x: adapt to resource_resolve interface change
Christoph Bumiller [Mon, 25 Jul 2011 16:13:26 +0000 (18:13 +0200)]
nv50,nvc0: never convert in resource copy when format sizes match
If there are any cases left where the st thinks that RGBA -> BGRA
will swap components, it will get what it deserves.
Now the GPU's 2D engine goes unused. What a shame.
Christoph Bumiller [Fri, 5 Aug 2011 18:10:04 +0000 (20:10 +0200)]
st/mesa: don't resolve stencil twice
Marek Olšák [Sat, 6 Aug 2011 03:15:30 +0000 (05:15 +0200)]
winsys/radeon: disable use of the buffer busy-for-write flag
Kenneth Graunke [Sun, 31 Jul 2011 04:26:26 +0000 (21:26 -0700)]
mesa: In validate_program(), initialize errMsg for safety.
validate_program relies on validate_shader_program to fill in errMsg;
empirically, there exist cases where that doesn't happen.
While tracking those down may be worthwhile, initializing the string so
we don't try to ralloc_strdup random garbage also seems wise.
Fixes issues caught by valgrind while running some test case.
NOTE: This is a candidate for stable release branches.
Reviewed-by: Chad Versace <chad@chad-versace.us>
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Eric Anholt [Fri, 5 Aug 2011 19:47:25 +0000 (12:47 -0700)]
intel: Fix warnings from gl_constant_parameter changes.
Bryan Cain [Fri, 5 Aug 2011 19:37:33 +0000 (14:37 -0500)]
glsl_to_tgsi: replace open-coded swizzle_for_size()
This is a port of commit
4c7e215c7bb to glsl_to_tgsi.
Bryan Cain [Fri, 5 Aug 2011 19:09:37 +0000 (14:09 -0500)]
glsl_to_tgsi: try to avoid emitting a MOV_SAT to saturate an expression tree
This is a port of commit 62722d9 to glsl_to_tgsi, with minor aesthetic
changes (moved the declaration and assignment of new_inst inside the if block).
Eric Anholt [Mon, 2 May 2011 23:27:46 +0000 (16:27 -0700)]
ir_to_mesa: Replace open-coded swizzle_for_size()
Christopher James Halse Rogers [Thu, 4 Aug 2011 02:06:13 +0000 (12:06 +1000)]
glx/dri2: Paper over errors in DRI2Connect when indirect
DRI2 will throw BadRequest for this when the client is not local, but
DRI2 is an implementation detail and not something callers should have
to know about. Silently swallow errors in this case, and just propagate
the failure through DRI2Connect's return code.
Note: This is a candidate for the stable release branches.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=28125
Signed-off-by: Christopher James Halse Rogers <christopher.halse.rogers@canonical.com>
Eric Anholt [Fri, 22 Jul 2011 20:54:15 +0000 (13:54 -0700)]
ir_to_mesa: Try to avoid emitting a MOV_SAT to saturate an expression tree.
Fixes a regression in codegen quality for ff_fragment_shader
conversion to GLSL -- glean texCombine produces 7.5% fewer Mesa IR
instructions.
Eric Anholt [Fri, 22 Jul 2011 22:13:08 +0000 (15:13 -0700)]
prog_optimize: Add support for saturates to _mesa_merge_mov_into_inst.
This fixes the remaining regression from ff_fragment_shader in Mesa IR
instruction count, to now being a 1.9% win overall.
Eric Anholt [Tue, 26 Jul 2011 01:15:25 +0000 (18:15 -0700)]
i965: Add dumping for gen6 WM constants too.
This looks just like the VS dump for now.
Eric Anholt [Thu, 28 Jul 2011 16:52:03 +0000 (09:52 -0700)]
mesa: Remove dead "MemPool" field of gl_shader_state.
Eric Anholt [Tue, 26 Jul 2011 01:13:04 +0000 (18:13 -0700)]
i965/fs: Don't upload unused uniform components.
This saves both register space and upload bandwidth for unused values.
Note that previously we were relying on the visitor not initially
generating references to different sets of uniforms between the 8-wide
and 16-wide code generation, and now we're relying on them dead-code
eliminating the same stuff, too.
Eric Anholt [Thu, 28 Jul 2011 16:57:19 +0000 (09:57 -0700)]
i965/fs: Don't allocate the old backend's compile structs for our compile.
This saves some 35MB when the program only uses GLSL shaders.
Marek Olšák [Wed, 3 Aug 2011 19:01:31 +0000 (21:01 +0200)]
winsys/radeon: do the CS cleanup in the CS ioctl thread
Marek Olšák [Wed, 3 Aug 2011 18:57:48 +0000 (20:57 +0200)]
winsys/radeon: fix space checking
We should remove the relocations which caused a validation failure
from the list, so that the kernel receives only the validated ones.
NOTE: This is a candidate for the 7.11 branch.
Marek Olšák [Fri, 5 Aug 2011 05:07:46 +0000 (07:07 +0200)]
st/dri: remove an unused-but-set variable
Marek Olšák [Fri, 5 Aug 2011 05:02:25 +0000 (07:02 +0200)]
st/dri: remove a dummy function dri2_create_context
It does nothing besides calling dri_create_context with the same parameters.
Marek Olšák [Fri, 5 Aug 2011 04:57:07 +0000 (06:57 +0200)]
st/mesa: remove unused-but-set variables in st_glsl_to_tgsi.cpp
Marek Olšák [Fri, 5 Aug 2011 04:04:05 +0000 (06:04 +0200)]
r300g: handle new CAPs
Marek Olšák [Fri, 5 Aug 2011 04:03:18 +0000 (06:03 +0200)]
r300g: adapt to the resource_resolve interface change
Marek Olšák [Thu, 4 Aug 2011 05:38:13 +0000 (07:38 +0200)]
vbo: do not call _mesa_max_buffer_index in debug builds
That code drops performance in Unigine Heaven and Tropics
by a factor of 10. That's too crazy even for a debug build.
NOTE: This is a candidate for the 7.11 branch.
Reviewed-by: Brian Paul <brianp@vmware.com>
Chia-I Wu [Wed, 3 Aug 2011 15:39:07 +0000 (00:39 +0900)]
glsl: empty declarations should be valid
Unlike C++, empty declarations such as
float;
should be valid. The spec is not explicit about this actually.
Some apps that generate their shader sources may rely on this. This was
noted when porting one of them to Linux from Windows.
Reviewed-by: Chad Versace <chad@chad-versace.us>
Note: this is a candidate for the 7.11 branch.
Vinson Lee [Fri, 5 Aug 2011 01:04:44 +0000 (18:04 -0700)]
mesa: Remove MSVC stdint typedefs from compiler.h.
MSVC can now include the stdint.h at include/c99/stdint.h.
Brian Paul [Thu, 4 Aug 2011 22:01:27 +0000 (16:01 -0600)]
mesa: pass correct constant type to _mesa_fetch_state()
Fixes assorted warnings about float vs. gl_constant_value pointers.
Brian Paul [Thu, 4 Aug 2011 22:00:06 +0000 (16:00 -0600)]
mesa: use gl_constant_value type in ARB program parser
Brian Paul [Thu, 4 Aug 2011 21:55:50 +0000 (15:55 -0600)]
mesa: use gl_constant_value type in _mesa_[Get]ProgramNamedParameter4fNV()
Brian Paul [Thu, 4 Aug 2011 21:55:13 +0000 (15:55 -0600)]
mesa: add st_glsl_to_tgsi.cpp to Sconscript
Brian Paul [Thu, 4 Aug 2011 21:32:09 +0000 (15:32 -0600)]
mesa: fix out of bounds array access in rtgc debug code
Fixes https://bugs.freedesktop.org/show_bug.cgi?id=39841
This would only be hit if someone set RGTC_DEBUG=1.
Brian Paul [Thu, 4 Aug 2011 19:07:50 +0000 (13:07 -0600)]
mesa: don't use K&R style function pointer calls
Bryan Cain [Thu, 4 Aug 2011 20:43:34 +0000 (15:43 -0500)]
Merge branch 'glsl-to-tgsi'
Conflicts:
src/mesa/state_tracker/st_atom_pixeltransfer.c
src/mesa/state_tracker/st_program.c
Bryan Cain [Thu, 4 Aug 2011 15:15:54 +0000 (10:15 -0500)]
st/mesa: replace duplicated create_color_map_texture() function with shared function
Bryan Cain [Tue, 2 Aug 2011 16:36:44 +0000 (11:36 -0500)]
glsl_to_tgsi: remove debugging printf
Brian Paul [Thu, 4 Aug 2011 14:22:31 +0000 (08:22 -0600)]
st/mesa: silence int/float and double/float conversion warnings
Brian Paul [Thu, 4 Aug 2011 14:22:31 +0000 (08:22 -0600)]
st/mesa: move declaration before code
Brian Paul [Thu, 4 Aug 2011 14:22:31 +0000 (08:22 -0600)]
mesa: minor comment updates in enable.c
Brian Paul [Thu, 4 Aug 2011 14:22:31 +0000 (08:22 -0600)]
mesa: whitespace, formatting fixes in GetTexParameter() code
Brian Paul [Thu, 4 Aug 2011 14:22:31 +0000 (08:22 -0600)]
mesa: add null ptr checks in GetTexParameterI[u]iv() functions
Brian Paul [Thu, 4 Aug 2011 14:22:31 +0000 (08:22 -0600)]
mesa: condense GL_TEXTURE_RESIDENT query code
Brian Paul [Thu, 4 Aug 2011 14:22:30 +0000 (08:22 -0600)]
mesa: make error handling in glGetTexParameter() a bit more concise
Christoph Bumiller [Thu, 28 Jul 2011 13:54:53 +0000 (15:54 +0200)]
nv50: implement resource_resolve with custom blit
Christoph Bumiller [Wed, 3 Aug 2011 14:01:41 +0000 (16:01 +0200)]
st/mesa: implement multisample resolve via BlitFramebuffer
Christoph Bumiller [Wed, 3 Aug 2011 13:43:16 +0000 (15:43 +0200)]
gallium: extend resource_resolve to accommodate BlitFramebuffer
Resolve via glBlitFramebuffer allows resolving a sub-region of a
renderbuffer to a different location in any mipmap level of some
other texture, and, with a new extension, even scaling. Therefore,
location and size parameters are needed.
The mask parameter was added because resolving only depth or only
stencil of a combined buffer is possible as well.
Full information about the blit operation allows the drivers to
take the most efficient path they possibly can.
Christoph Bumiller [Thu, 28 Jul 2011 13:26:01 +0000 (15:26 +0200)]
st/mesa: determine Const.MaxSamples in init_extensions
v2: Check for non-pow2 sample counts as well.
Benjamin Franzke [Wed, 29 Jun 2011 06:49:39 +0000 (08:49 +0200)]
egl/gbm: Fix EGL_DEFAULT_DISPLAY
Benjamin Franzke [Thu, 4 Aug 2011 11:37:42 +0000 (13:37 +0200)]
gbm: link gbm_gallium_drm.so against math library
This avoids the following runtime error with EGL on platforms that
require linking with libm for nontrivial math functions:
failed to load module: /xorg/lib64/gbm/gbm_gallium_drm.so: undefined
symbol: powf
(Based on Kristóf RALOVICHs patch and Ian's suggestions in
http://lists.freedesktop.org/archives/mesa-dev/2011-August/010036.html)
RALOVICH, Kristóf [Sun, 31 Jul 2011 21:49:43 +0000 (23:49 +0200)]
gbm/dri: avoid crash in dri_screen_create
Marek Olšák [Thu, 4 Aug 2011 01:23:12 +0000 (03:23 +0200)]
r600g: remove more of unused code
This is a follow-up to
f6df430a85141f6a384c18079fb5b2ad848dac0d.
Vadim Girlin [Wed, 3 Aug 2011 11:35:02 +0000 (15:35 +0400)]
r600g: take into account force_add_cf in pops
When we have two ENDIFs in a row, we shouldn't modify the pop_count
for the same alu clause twice.
Fixes https://bugs.freedesktop.org/show_bug.cgi?id=38163
Note: this is a candidate for the 7.11 branch.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Vadim Girlin [Tue, 2 Aug 2011 21:04:19 +0000 (01:04 +0400)]
r600g: use backend mask for occlusion queries
Use backend_map kernel query if supported, otherwise analyze ZPASS_DONE
results to get the mask.
Fixes lockups with predicated rendering due to incorrect query buffer
initialization on some cards.
Note: this is a candidate for the 7.11 branch.
Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Marek Olšák [Tue, 2 Aug 2011 22:52:55 +0000 (00:52 +0200)]
winsys/radeon: remove dummy function pb_buffer
Marek Olšák [Tue, 2 Aug 2011 23:13:06 +0000 (01:13 +0200)]
util: fix a typo in util_format_swizzle_4f
Reported by Gustaw Smolarczyk.
Eric Anholt [Tue, 2 Aug 2011 20:49:05 +0000 (13:49 -0700)]
radeon: Remove some remaining set-but-unused variables.
These looked more like copy-and-paste to me than the others (which
looked more like possibly someone forgot to write some code in a
refactor), so I didn't verify where they came from.
Eric Anholt [Tue, 2 Aug 2011 20:47:18 +0000 (13:47 -0700)]
radeon: Remove set-but-unused variables in radeon_lock.c
These have been unused since this function's introduction in the FBO
support development around 2009.
Eric Anholt [Tue, 2 Aug 2011 20:41:59 +0000 (13:41 -0700)]
radeon: Remove set-but-unused variables in radeonSetTexBuffer2() variants.
These have been unused since 2009.
Eric Anholt [Tue, 2 Aug 2011 20:39:43 +0000 (13:39 -0700)]
radeon: Remove set-but-unused log2depth variable.
r100 doesn't support 3D GL_EXT_texture3D.
Eric Anholt [Tue, 2 Aug 2011 20:36:57 +0000 (13:36 -0700)]
radeon: Remove set-but-unused color_mask variable.
This has been around since the initial import in 2003 and never used.
Eric Anholt [Mon, 1 Aug 2011 23:06:59 +0000 (16:06 -0700)]
intel: Fix unused variable warning.
Marek Olšák [Sat, 23 Jul 2011 02:29:59 +0000 (04:29 +0200)]
r600g: remove dummy function r600_bo_offset
Always returned 0.
Marek Olšák [Fri, 22 Jul 2011 16:45:30 +0000 (18:45 +0200)]
r600g: remove unused code
Marek Olšák [Mon, 1 Aug 2011 23:04:58 +0000 (01:04 +0200)]
gallium/util: add functions for manipulating swizzles
Some of those have been in drivers already.
Ian Romanick [Thu, 10 Feb 2011 21:20:26 +0000 (13:20 -0800)]
i915: Only emit program errors when INTEL_DEBUG=wm or INTEL_DEBUG=fallbacks
This makes piglit a lot more happy. The errors are logged when
INTEL_DEBUG=fallbacks because the application is about to hit a big
software fallback. We frequently ask people to run applications that
are hitting software fallbacks with INTEL_DEBUG=fallbacks so the we
can help them debug the reason for the software fallback.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Ian Romanick [Mon, 25 Jul 2011 23:41:39 +0000 (16:41 -0700)]
i915: Fail without crashing if a Mesa IR program uses too many registers
This can only happen in GLSL shaders because assembly shaders that use
too many temps are rejected by core Mesa. It is easiest to make this
happen with shaders that contain flow-control that could not be lowered.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Ian Romanick [Mon, 25 Jul 2011 22:58:07 +0000 (15:58 -0700)]
ir_to_mesa: Emit warnings instead of errors for IR that can't be lowered
Rely on the driver to do the right thing. This probably means falling
back to software. Page 88 of the OpenGL 2.1 spec specifically says:
"A shader should not fail to compile, and a program object should
not fail to link due to lack of instruction space or lack of
temporary variables. Implementations should ensure that all valid
shaders and program objects may be successfully compiled, linked
and executed."
There is no provision for saying "No" to a valid shader that is
difficult for the hardware to handle, so stop doing that.
On i915 this causes a large number of piglit tests to change from FAIL
to WARN. The warning is because the driver still emits messages to
stderr like "i915_program_error: Unsupported opcode: BGNLOOP".
It also fixes ES2 conformance CorrectFull_frag and CorrectParse1_frag
on i915 (and probably other hardware that can't handle loops).
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Ian Romanick [Mon, 25 Jul 2011 22:55:59 +0000 (15:55 -0700)]
ir_to_mesa: Use Add linker_error instead of fail_link
The functions were almost identical.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Ian Romanick [Thu, 28 Jul 2011 22:10:17 +0000 (15:10 -0700)]
mesa: Ensure that gl_shader_program::InfoLog is never NULL
This prevents assertion failures in ralloc_strcat. The ralloc_free in
_mesa_free_shader_program_data can be omitted because freeing the
gl_shader_program in _mesa_delete_shader_program will take care of
this automatically.
A bunch of this code could use a refactor to use ralloc a bit more
effectively. A bunch of the things that are allocated with malloc and
owned by the gl_shader_program should be allocated with ralloc (using
the gl_shader_program as the context).
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Ian Romanick [Thu, 28 Jul 2011 21:09:06 +0000 (14:09 -0700)]
linker: Make linker_{error,warning} generally available
linker_warning is a new function. It's identical to linker_error
except that it doesn't set LinkStatus=false and it prepends "warning: "
on messages instead of "error: ".
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Ian Romanick [Thu, 28 Jul 2011 21:04:09 +0000 (14:04 -0700)]
linker: Make linker_error set LinkStatus to false
Remove the other places that set LinkStatus to false since they all
immediately follow a call to linker_error. The function linker_error
was previously known as linker_error_printf. The name was changed
because it may seem surprising that a printf function will set an
error flag.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Kenneth Graunke [Sat, 30 Jul 2011 23:44:49 +0000 (16:44 -0700)]
i965/gen5+: Fix incorrect miptree layout for non-power-of-two cubemaps.
For power-of-two sizes, h0 == mt->height0 since it's already a multiple
of two. However, for NPOT, they're different; h1 should be computed
based on the original size.
Fixes piglit test "cubemap npot" and oglconform test "textureNPOT".
NOTE: This is a candidate for stable release branches.
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Bryan Cain [Wed, 27 Jul 2011 21:39:40 +0000 (16:39 -0500)]
glsl_to_tgsi: copy reladdr in st_src_reg(st_dst_reg) constructor
This is a glsl_to_tgsi port of commit
f7cd9a858c04.
Bryan Cain [Wed, 27 Jul 2011 21:36:10 +0000 (16:36 -0500)]
glsl_to_tgsi: add each relative address to the previous
This is a glsl_to_tgsi port of commit
d6e1a8f71437.
Bryan Cain [Wed, 27 Jul 2011 20:45:16 +0000 (15:45 -0500)]
glsl_to_tgsi: lower all ir_quadop_vector expressions
Unlike Mesa IR, TGSI doesn't have a SWZ opcode.
Bryan Cain [Wed, 27 Jul 2011 20:20:19 +0000 (15:20 -0500)]
glsl_to_tgsi: rework immediate tracking to not use gl_program_parameter_list
Bryan Cain [Fri, 22 Jul 2011 18:24:42 +0000 (13:24 -0500)]
glsl_to_tgsi: update comments
Bryan Cain [Fri, 22 Jul 2011 18:23:26 +0000 (13:23 -0500)]
glsl_to_tgsi: make coding style more consistent
Bryan Cain [Thu, 21 Jul 2011 21:29:56 +0000 (16:29 -0500)]
glsl_to_tgsi: make assignment hack safer
Fixes an assertion failure in piglit test glsl-texcoord-array.
Bryan Cain [Thu, 21 Jul 2011 20:49:26 +0000 (15:49 -0500)]
glsl_to_tgsi: separate immediates from array constants during IR translation
Before, if any uniform or constant array was accessed with indirect
addressing, st_translate_program() would emit uniform constants in the place
of immediates. This behavior was unavoidable with ir_to_mesa/mesa_to_tgsi, but
glsl_to_tgsi can work around it since the GLSL IR backend and the TGSI
emission are both inside the state tracker.
Bryan Cain [Sun, 10 Jul 2011 22:36:04 +0000 (17:36 -0500)]
glsl_to_tgsi: fix mistakes in get_pixel_transfer_visitor()
I noticed these issues while working on get_bitmap_visitor().
Bryan Cain [Sun, 10 Jul 2011 22:17:38 +0000 (17:17 -0500)]
st/mesa, glsl_to_tgsi: support glBitmap with a GLSL fragment shader active
Bryan Cain [Sat, 9 Jul 2011 02:12:08 +0000 (21:12 -0500)]
st/mesa, glsl_to_tgsi: support glDrawPixels/glCopyPixels with a GLSL fragment shader active
Since this was previously implemented using Mesa IR and _mesa_combine_programs,
this commit adds a new code path that works with glsl_to_tgsi.
Bryan Cain [Mon, 4 Jul 2011 13:44:12 +0000 (08:44 -0500)]
glsl_to_tgsi: replace MAX_PROGRAM_TEMPS (256) with MAX_TEMPS (4096)
Bryan Cain [Thu, 30 Jun 2011 18:42:37 +0000 (13:42 -0500)]
r200, r600c, i965: fix build
Bryan Cain [Mon, 27 Jun 2011 22:40:10 +0000 (17:40 -0500)]
glsl_to_tgsi: always run copy_propagate() and eliminate_dead_code_advanced()
These two passes are written to handle indirect addressing properly.
Bryan Cain [Mon, 27 Jun 2011 22:25:50 +0000 (17:25 -0500)]
glsl_to_tgsi: fix mistake in new dead code elimination pass
The conditions of IF opcodes were not being counted as reads, which sometimes
led to the condition register being wrong or undefined.
Bryan Cain [Mon, 27 Jun 2011 22:11:07 +0000 (17:11 -0500)]
glsl_to_tgsi: improve assignment handling
This is a hack, but it's better than emitting an unnecessary MOV instruction
and hoping the optimization passes clean it up.
Bryan Cain [Sat, 25 Jun 2011 04:17:30 +0000 (23:17 -0500)]
glsl_to_tgsi: improve eliminate_dead_code_advanced()
Bryan Cain [Sat, 25 Jun 2011 03:32:26 +0000 (22:32 -0500)]
glsl_to_tgsi: use the correct writemask in try_emit_mad() and try_emit_sat()
Bryan Cain [Sat, 25 Jun 2011 01:37:53 +0000 (20:37 -0500)]
glsl_to_tgsi: use a more specific condition for gl_FragDepth hack in generating assignments
This reduces the number of instructions in the fragment shader of
glsl-fs-atan-2 from 174 to 146 with EmitNoIfs enabled.
Bryan Cain [Fri, 24 Jun 2011 23:45:04 +0000 (18:45 -0500)]
glsl_to_tgsi: add a better, more advanced dead code elimination pass
Bryan Cain [Fri, 24 Jun 2011 00:53:37 +0000 (19:53 -0500)]
glsl_to_tgsi: remove handling of XPD opcode in compile_tgsi_instruction()
The opcode is never emitted by the glsl_to_tgsi_visitor, so its special case in
compile_tgsi_instruction() was dead code.
Bryan Cain [Fri, 24 Jun 2011 00:35:36 +0000 (19:35 -0500)]
glsl_to_tgsi: use swizzle_for_size for src reg in conditional moves
This prevents the copy propagation pass from being confused by undefined
channels and thus missing optimization opportunities.
Emil Velikov [Tue, 21 Jun 2011 20:52:19 +0000 (21:52 +0100)]
glsl_to_tgsi: execute merge_registers() after eliminate_dead_code()
Fixes a regression unintentionally introduced by "glsl_to_tgsi: fix shaders with
indirect addressing of temps" that caused missing leaves in 3dmark01 test 4 (Nature)
and missing/displaced textures on human models in Counter-Strike: Source.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Bryan Cain <bryancain3@gmail.com>