Chia-I Wu [Sat, 23 Oct 2010 03:59:03 +0000 (11:59 +0800)]
egl: Add reference count for resources.
This is a really simple mechanism. There is no atomicity and the caller
is expected to hold the display lock.
Chia-I Wu [Sat, 23 Oct 2010 03:31:29 +0000 (11:31 +0800)]
st/egl: Fix native_mode refresh mode.
Define the unit to match _EGLMode's.
Chia-I Wu [Fri, 22 Oct 2010 18:52:14 +0000 (02:52 +0800)]
egl: Fix _eglModeLookup.
Internally a mode belongs to a screen. But functions like
eglGetModeAttribMESA treat a mode as a display resource: a mode can be
looked up without a screen. Considering how KMS works, it is better to
stick to the current implementation.
To properly support looking up a mode without a screen, this commit
assigns each mode (of all screens) a unique ID.
Chia-I Wu [Fri, 22 Oct 2010 18:11:21 +0000 (02:11 +0800)]
egl: Minor changes to the _EGLScreen interface.
Make _eglInitScreen take a display and rename _eglAddScreen to
_eglLinkScreen. Remove unused functions.
Chia-I Wu [Fri, 22 Oct 2010 16:47:22 +0000 (00:47 +0800)]
egl: Minor changes to the _EGLConfig interface.
Mainly to rename _eglAddConfig to _eglLinkConfig, along with a few clean
ups.
Chia-I Wu [Fri, 22 Oct 2010 16:37:19 +0000 (00:37 +0800)]
egl: Drop dpy argument from the link functions.
All display resources are already initialized with a display. Linking
simply links a resource to its display.
Eric Anholt [Fri, 22 Oct 2010 19:57:00 +0000 (12:57 -0700)]
i965: Add support for pull constants to the new FS backend.
Fixes glsl-fs-uniform-array-5, but not 6 which fails in ir_to_mesa.
Eric Anholt [Fri, 22 Oct 2010 20:17:38 +0000 (13:17 -0700)]
i965: Move the FS disasm/annotation printout to codegen time.
This makes it a lot easier to track down where we failed when some
code emit triggers an assert. Plus, less memory allocation for
codegen.
Dave Airlie [Fri, 22 Oct 2010 21:45:59 +0000 (07:45 +1000)]
r600g: not fatal if we can't get tiling info from kernel
Marek Olšák [Fri, 22 Oct 2010 18:33:25 +0000 (20:33 +0200)]
r300g: say no to PIPE_CAP_STREAM_OUTPUT and PIPE_CAP_PRIMITIVE_RESTART
Marek Olšák [Fri, 22 Oct 2010 17:45:05 +0000 (19:45 +0200)]
r300g: do not print get_param errors in non-debug build
Brian Paul [Fri, 22 Oct 2010 14:59:06 +0000 (08:59 -0600)]
mesa: move declaration before code
Brian Paul [Fri, 22 Oct 2010 14:58:35 +0000 (08:58 -0600)]
galahad: silence warnings
Francisco Jerez [Fri, 22 Oct 2010 11:43:57 +0000 (13:43 +0200)]
dri/nouveau: Force a "slow" Z clear if we're getting a new depth buffer.
Chia-I Wu [Fri, 22 Oct 2010 10:11:04 +0000 (18:11 +0800)]
egl: Move fallback routines to eglfallbacks.c.
We do not want them to be all over the places.
Chia-I Wu [Fri, 22 Oct 2010 09:09:40 +0000 (17:09 +0800)]
egl: Parse image attributes with _eglParseImageAttribList.
Avoid code duplications.
Chia-I Wu [Fri, 22 Oct 2010 08:36:47 +0000 (16:36 +0800)]
egl: Move attributes in _EGLImage to _EGLImageAttribs.
The opaque nature of EGLImage implies that extensions almost always
define their own attributes. Move attributes in _EGLImage to
_EGLImageAttribs and add a helper function to parse attribute lists.
Chia-I Wu [Fri, 22 Oct 2010 07:44:33 +0000 (15:44 +0800)]
egl_glx: Fix borken driver.
The driver was broken since
6eda3f311bc24999835003e404d5eda5599bc5de.
All configs fail to pass _eglValidateConfig.
Chia-I Wu [Fri, 22 Oct 2010 06:55:31 +0000 (14:55 +0800)]
egl_glx: Drop the use of [SG]ET_CONFIG_ATTRIB.
_EGLConfig can be directly dereferenced now. Since egl_glx is the last
user of the macros, drop the macros too.
Chia-I Wu [Fri, 22 Oct 2010 07:03:11 +0000 (15:03 +0800)]
egl_dri2: Drop the use of _egl[SG]etConfigKey.
_EGLConfig can be directly dereferenced now.
Brian Paul [Fri, 22 Oct 2010 01:55:01 +0000 (19:55 -0600)]
winsys/xlib: rename xm->xlib
Move away from the old Mesa-oriented names.
Brian Paul [Fri, 22 Oct 2010 01:49:32 +0000 (19:49 -0600)]
winsys/xlib: fix up allocation/dealloction of XImage
Fixes a crash upon exit when using remote display.
Brian Paul [Fri, 22 Oct 2010 01:37:11 +0000 (19:37 -0600)]
winsys/xlib: use Bool type for shm field
Brian Paul [Fri, 22 Oct 2010 01:17:31 +0000 (19:17 -0600)]
winsys/xlib: formatting fixes
Brian Paul [Fri, 22 Oct 2010 01:05:40 +0000 (19:05 -0600)]
Merge branch 'primitive-restart-cleanup'
Conflicts:
docs/relnotes-7.10.html
This branch is a re-do of the primitive-restart branch with all
the intermediate/temporary stuff cleaned out.
Brian Paul [Fri, 22 Oct 2010 01:03:39 +0000 (19:03 -0600)]
docs: added GL_NV_primitive_restart extension
Brian Paul [Fri, 22 Oct 2010 01:03:38 +0000 (19:03 -0600)]
llvmpipe: enable primitive restart
Brian Paul [Fri, 22 Oct 2010 01:03:38 +0000 (19:03 -0600)]
softpipe: enable primitive restart
Brian Paul [Fri, 22 Oct 2010 01:03:38 +0000 (19:03 -0600)]
draw: implement primitive splitting for primitive restart
Brian Paul [Fri, 22 Oct 2010 01:03:38 +0000 (19:03 -0600)]
st/mesa: support for primitive restart
Brian Paul [Fri, 22 Oct 2010 01:03:38 +0000 (19:03 -0600)]
gallium: new CAP, state for primitive restart
Brian Paul [Fri, 22 Oct 2010 01:03:38 +0000 (19:03 -0600)]
vbo: support for primitive restart
We handle splitting of glDrawArrays() calls into two primitives here
so that drivers don't have to worry about it.
Brian Paul [Fri, 22 Oct 2010 01:03:38 +0000 (19:03 -0600)]
mesa: plug in primitive restart function
Brian Paul [Fri, 22 Oct 2010 01:03:38 +0000 (19:03 -0600)]
mesa: regenerated files with primitive restart
Brian Paul [Fri, 22 Oct 2010 01:03:38 +0000 (19:03 -0600)]
mesa: API spec for primitive restart
Brian Paul [Fri, 22 Oct 2010 01:03:38 +0000 (19:03 -0600)]
mesa: set/get primitive restart state
Brian Paul [Fri, 22 Oct 2010 01:03:38 +0000 (19:03 -0600)]
mesa: driver hook for primitive restart
Eric Anholt [Thu, 21 Oct 2010 23:54:03 +0000 (16:54 -0700)]
i965: Be more aggressive in tracking live/dead intervals within loops.
Fixes glsl-fs-convolution-2, which was blowing up due to the array
access insanity getting at the uniform values within the loop. Each
temporary was considered live across the whole loop.
Brian Paul [Thu, 21 Oct 2010 23:04:38 +0000 (17:04 -0600)]
docs: add GL_ARB_texture_rg to release notes
Brian Paul [Thu, 21 Oct 2010 23:04:25 +0000 (17:04 -0600)]
docs: update texture red/green support in GL3.txt
Brian Paul [Thu, 21 Oct 2010 22:39:07 +0000 (16:39 -0600)]
st/mesa: added cases for GL_COMPRESSED_RED/RG in st_choose_format()
Brian Paul [Thu, 21 Oct 2010 22:38:18 +0000 (16:38 -0600)]
mesa: add missing cases for packing red/green images
Brian Paul [Thu, 21 Oct 2010 22:36:23 +0000 (16:36 -0600)]
mesa: add GL_RG case to _mesa_source_buffer_exists()
Fixes failure with glReadPixels(format=GL_RG)
Brian Paul [Wed, 20 Oct 2010 23:18:40 +0000 (17:18 -0600)]
draw: fix typo in comment
Eric Anholt [Thu, 21 Oct 2010 22:07:45 +0000 (15:07 -0700)]
i965: Correct scratch space allocation.
One, it was allocating increments of 1kb, but per thread scratch space
is a power of two. Two, the new FS wasn't getting total_scratch set
at all, so everyone thought they had 1kb and writes beyond 1kb would
go stomping on a neighbor thread.
With this plus the previous register spilling for the new FS,
glsl-fs-convolution-1 passes.
Eric Anholt [Thu, 21 Oct 2010 21:40:49 +0000 (14:40 -0700)]
i965: Don't emit register spill offsets directly into g0.
g0 is used by others, and is expected to be left exactly as it was
dispatched to us. So manually move g0 into our message reg when
spilling/unspilling and update the offset in the MRF. Fixes failures
in texture sampling after having spilled a register.
Eric Anholt [Tue, 19 Oct 2010 16:25:51 +0000 (09:25 -0700)]
i965: Add support for register spilling.
It can be tested with if (0) replaced with if (1) to force spilling for all
virtual GRFs. Some simple tests work, but large texturing tests fail.
Eric Anholt [Wed, 20 Oct 2010 22:21:53 +0000 (15:21 -0700)]
i965: Fix gl_FrontFacing emit on pre-gen6.
It's amazing this code worked. Basically, we would get lucky in
register allocation and the tests using frontfacing would happen to
allocate gl_FrontFacing storage and the instructions generating
gl_FrontFacing but pointing at another register to the same hardware
register. Noticed during register spilling debug, when suddenly they
didn't get allocatd the same storage.
Eric Anholt [Wed, 20 Oct 2010 17:26:29 +0000 (10:26 -0700)]
i965: Split register allocation out of the ever-growing brw_fs.cpp.
Kenneth Graunke [Thu, 21 Oct 2010 18:43:39 +0000 (11:43 -0700)]
Refresh autogenerated file builtin_function.cpp.
Since this is just generated by python, it's questionable whether this
should continue to live in the repository - Mesa already has other
things generated from python as part of the build process.
Kenneth Graunke [Thu, 21 Oct 2010 18:42:45 +0000 (11:42 -0700)]
generate_builtins.py: Output large strings as arrays of characters.
This works around MSVC's 65535 byte limit, unfortunately at the expense
of any semblance of readability and much larger file size. Hopefully I
can implement a better solution later, but for now this fixes the build.
Vinson Lee [Thu, 21 Oct 2010 18:27:35 +0000 (11:27 -0700)]
gallivm: Silence uninitialized variable warning.
Fixes this GCC warning.
gallivm/lp_bld_tgsi_aos.c: In function 'lp_build_tgsi_aos':
gallivm/lp_bld_tgsi_aos.c:516: warning: 'dst0' may be used uninitialized in this function
gallivm/lp_bld_tgsi_aos.c:516: note: 'dst0' was declared here
Vinson Lee [Thu, 21 Oct 2010 18:21:03 +0000 (11:21 -0700)]
gallivm: Silence uninitialized variable warnings.
Fixes these GCC warnings.
gallivm/lp_bld_sample_aos.c: In function 'lp_build_sample_image_nearest':
gallivm/lp_bld_sample_aos.c:271: warning: 't_ipart' may be used uninitialized in this function
gallivm/lp_bld_sample_aos.c:271: warning: 'r_ipart' may be used uninitialized in this function
Vinson Lee [Thu, 21 Oct 2010 18:10:15 +0000 (11:10 -0700)]
gallivm: Silence uninitialized variable warnings.
Fixes these GCC warnings.
gallivm/lp_bld_sample_aos.c: In function 'lp_build_sample_image_linear':
gallivm/lp_bld_sample_aos.c:439: warning: 'r_ipart' may be used uninitialized in this function
gallivm/lp_bld_sample_aos.c:438: warning: 't_ipart' may be used uninitialized in this function
gallivm/lp_bld_sample_aos.c:438: warning: 't_fpart' may be used uninitialized in this function
gallivm/lp_bld_sample_aos.c:439: warning: 'r_fpart' may be used uninitialized in this function
gallivm/lp_bld_sample_aos.c:438: warning: 't_fpart_hi' may be used uninitialized in this function
gallivm/lp_bld_sample_aos.c:438: warning: 't_fpart_lo' may be used uninitialized in this function
gallivm/lp_bld_sample_aos.c:439: warning: 'r_fpart_hi' may be used uninitialized in this function
gallivm/lp_bld_sample_aos.c:439: warning: 'r_fpart_lo' may be used uninitialized in this function
Chia-I Wu [Fri, 15 Oct 2010 02:09:37 +0000 (10:09 +0800)]
mesa: Remove unused vtxfmt_tmp.h.
It was used by the "neutral" tnl module that was dropped in
81ccb3e2ce708619f4c23537a237d61bdffdd35f.
Dave Airlie [Thu, 21 Oct 2010 09:11:23 +0000 (19:11 +1000)]
r600g: start splitting out common code from eg/r600.
no point duplicating code that doesn't touch hw, also make it easier
to spot mistakes
Dave Airlie [Thu, 21 Oct 2010 07:34:25 +0000 (17:34 +1000)]
r600g: initial translate state support
Vinson Lee [Thu, 21 Oct 2010 08:47:52 +0000 (01:47 -0700)]
draw: Remove unnecessary header.
Vinson Lee [Thu, 21 Oct 2010 08:44:48 +0000 (01:44 -0700)]
llvmpipe: Remove unnecessary header.
Kenneth Graunke [Tue, 19 Oct 2010 18:57:15 +0000 (11:57 -0700)]
glsl: Refresh autogenerated file builtin_function.cpp.
Kenneth Graunke [Tue, 19 Oct 2010 18:56:47 +0000 (11:56 -0700)]
glsl: Add support for GLSL 1.30's modf built-in.
Kenneth Graunke [Thu, 21 Oct 2010 04:53:36 +0000 (21:53 -0700)]
glcpp: Refresh autogenerated lexer file.
Kenneth Graunke [Thu, 21 Oct 2010 04:51:03 +0000 (21:51 -0700)]
glcpp: Return NEWLINE token for newlines inside multi-line comments.
This is necessary for the main compiler to get correct line numbers.
Dave Airlie [Thu, 21 Oct 2010 03:40:45 +0000 (13:40 +1000)]
r600g: add texture tiling enable under a debug option.
At the moment you need kernel patches to have texture tiling work
with the kernel CS checker, so once they are upstream and the drm version
is bumped we can make this enable flip the other way most likely.
Dave Airlie [Thu, 21 Oct 2010 03:37:54 +0000 (13:37 +1000)]
r600g: add texture tiling alignment support.
this sets things up to align stride/height with tile sizes,
it also adds support for the 2D/1D array mode cross over point.
Dave Airlie [Thu, 21 Oct 2010 03:36:01 +0000 (13:36 +1000)]
r600g: introduce a per-driver resource flag for transfers.
this is to be used to decide not to tile a surface being used for transfers.
Dave Airlie [Thu, 21 Oct 2010 03:31:27 +0000 (13:31 +1000)]
r600g: add r600 surface to store the aligned height.
we need to know the aligned height when binding the surface to cb/zb,
not the gallium surface height.
Dave Airlie [Thu, 21 Oct 2010 03:27:07 +0000 (13:27 +1000)]
r600g: start adding hooks for aligning width/height for tiles.
Dave Airlie [Thu, 21 Oct 2010 03:26:04 +0000 (13:26 +1000)]
r600g: move to per-miplevel array mode.
Since the hw transitions from 2D->1D sampling below the 2D macrotile
size we need to keep track of the array mode per level so we can
render to it using the CB.
Dave Airlie [Thu, 21 Oct 2010 03:20:14 +0000 (13:20 +1000)]
r600g: all non-0 mipmap levels need to be w/h aligned to POT.
this adds a new minify function to the driver to ensure this.
Vinson Lee [Wed, 20 Oct 2010 22:27:48 +0000 (15:27 -0700)]
swrast: Print out format on unexpected failure in _swrast_DrawPixels.
Kenneth Graunke [Wed, 20 Oct 2010 20:52:16 +0000 (13:52 -0700)]
mesa: Remove FEATURE_ARB_shading_language_120 macro.
Everything should be able to support 1.20 at this point.
Kenneth Graunke [Wed, 20 Oct 2010 21:59:40 +0000 (14:59 -0700)]
glsl: Remove useless ir_shader enumeration value.
Vinson Lee [Wed, 20 Oct 2010 21:10:26 +0000 (14:10 -0700)]
glsl: Add assert for unhandled ir_shader case.
Silences this GCC warning.
ast_to_hir.cpp: In function 'void apply_type_qualifier_to_variable(const
ast_type_qualifier*, ir_variable*, _mesa_glsl_parse_state*, YYLTYPE*)'
ast_to_hir.cpp:1768: warning: enumeration value 'ir_shader' not handled
in switch
Brian Paul [Wed, 20 Oct 2010 20:54:38 +0000 (14:54 -0600)]
draw: use float version of LLVM Mul/Add instructions
LLVM 2.8 is pickier about int vs float instructions and operands.
Brian Paul [Wed, 20 Oct 2010 20:47:32 +0000 (14:47 -0600)]
llvmpipe/draw: always enable LLVMAddInstructionCombiningPass()
We were working around an LLVM 2.5 bug but we're using LLVM 2.6 or later now.
This basically reverts commit
baddcbc5225e12052b3bc8c07a8b65243d76574d.
This fixes the piglit bug/tri-tex-crash.c failure.
Orion Poplawski [Wed, 20 Oct 2010 19:54:57 +0000 (15:54 -0400)]
osmesa: link against libtalloc
Otherwise consumers have to, and that's lame.
Signed-off-by: Adam Jackson <ajax@redhat.com>
Vinson Lee [Wed, 20 Oct 2010 19:44:08 +0000 (12:44 -0700)]
r600g: Ensure r600_src is initialized in tgsi_exp function.
Silences these GCC warnings.
r600_shader.c: In function 'tgsi_exp':
r600_shader.c:2339: warning: 'r600_src[0].rel' is used uninitialized in this function
r600_shader.c:2339: warning: 'r600_src[0].abs' is used uninitialized in this function
r600_shader.c:2339: warning: 'r600_src[0].neg' is used uninitialized in this function
r600_shader.c:2339: warning: 'r600_src[0].chan' is used uninitialized in this function
r600_shader.c:2339: warning: 'r600_src[0].sel' is used uninitialized in this function
Vinson Lee [Wed, 20 Oct 2010 06:48:59 +0000 (23:48 -0700)]
draw: Move loop variable declaration outside for loop.
Fixes MSVC build.
Keith Whitwell [Wed, 20 Oct 2010 05:11:49 +0000 (22:11 -0700)]
draw: make sure viewport gets updated in draw llvm shader
The viewport state was being baked in at compile time (oops...)
Keith Whitwell [Wed, 20 Oct 2010 04:41:28 +0000 (21:41 -0700)]
Merge branch 'llvm-cliptest-viewport'
Hui Qi Tay [Mon, 18 Oct 2010 17:48:18 +0000 (18:48 +0100)]
draw: corrections to allow for different cliptest cases
Eric Anholt [Tue, 19 Oct 2010 19:32:55 +0000 (12:32 -0700)]
i965: Use the new style of IF statement with embedded comparison on gen6.
"Everyone else" does it this way, so follow suit. It's fewer
instructions, anyway.
Eric Anholt [Tue, 19 Oct 2010 21:50:42 +0000 (14:50 -0700)]
i965: Set the source operand types for gen6 if/else/endif to integer.
I don't think this should matter, but I'm not sure, and it's
recommended by a kernel checker in fulsim.
Eric Anholt [Tue, 19 Oct 2010 19:55:04 +0000 (12:55 -0700)]
i965: Add EU emit support for gen6's new IF instruction with comparison.
Ian Romanick [Wed, 20 Oct 2010 00:59:10 +0000 (17:59 -0700)]
linker: Improve handling of unread/unwritten shader inputs/outputs
Previously some shader input or outputs that hadn't received location
assignments could slip through. This could happen when a shader
contained user-defined varyings and was used with either
fixed-function or assembly shaders.
See the piglit tests glsl-[fv]s-user-varying-ff and
sso-user-varying-0[12].
NOTE: this is a candidate for the 7.9 branch.
Chad Versace [Fri, 15 Oct 2010 21:46:00 +0000 (14:46 -0700)]
glsl: Commit generated file glsl_lexer.cpp
Changes are due to commit "glsl: Fix lexer rule for ^=".
Chad Versace [Fri, 15 Oct 2010 21:44:28 +0000 (14:44 -0700)]
glsl: Fix lexer rule for ^=
The caret is a special character, and needs to be quoted or escaped.
Chad Versace [Fri, 15 Oct 2010 19:08:28 +0000 (12:08 -0700)]
glsl: Implement ast-to-hir for bit-logic ops
Implement by adding to ast_expression::hir() the following cases:
- ast_and_assign
- ast_or_assign
- ast_xor_assign
Chad Versace [Fri, 15 Oct 2010 18:28:05 +0000 (11:28 -0700)]
glsl: Define bit_logic_result_type() in ast_to_hir.cpp
This function type checks the operands of and returns the result type of
bit-logic operations. It replaces the type checking performed in the
following cases of ast_expression::hir() :
- ast_bit_and
- ast_bit_or
- ast_bit_xor
Chad Versace [Fri, 15 Oct 2010 17:05:50 +0000 (10:05 -0700)]
glsl: Implement ast-to-hir for bit-shift-assignment
Implement by adding to ast_expression::hir() these cases:
- ast_ls_assign
- ast_rs_assign
Chad Versace [Fri, 15 Oct 2010 16:49:46 +0000 (09:49 -0700)]
glsl: Define shift_result_type() in ast_to_hir.cpp
This function type checks the operands of and returns the result type of
bit-shift operations. It replaces the type checking performed in the following
cases of ast_expression::hir() :
- ast_lshift
- ast_rshift
Eric Anholt [Mon, 18 Oct 2010 22:43:20 +0000 (15:43 -0700)]
i965: Disable thread dispatch when the FS doesn't do any work.
This should reduce the cost of generating shadow maps, for example.
No performance difference measured in nexuiz, though it does trigger
this path.
Eric Anholt [Tue, 19 Oct 2010 16:50:44 +0000 (09:50 -0700)]
i965: Remove the gen6 emit_mi_flushes I sprinkled around the driver.
These were for debugging in bringup. Now that relatively complicated
apps are working, they haven't helped debug anything in quite a while.
Eric Anholt [Tue, 19 Oct 2010 16:44:20 +0000 (09:44 -0700)]
i965: Tell the shader compiler when we expect depth writes for gen6.
This fixes hangs in some Z-writes-in-shaders tests, though other
pieces don't come out correctly.
Bug #30392: hang in fbo-fblit-d24s8. (still fails with bad color drawn
to some targets)
Vinson Lee [Tue, 19 Oct 2010 17:14:11 +0000 (10:14 -0700)]
llvmpipe: Initialize variable.
Vinson Lee [Tue, 19 Oct 2010 17:02:28 +0000 (10:02 -0700)]
llvmpipe: Initialize state variable in debug_bin function.
Vinson Lee [Tue, 19 Oct 2010 16:49:15 +0000 (09:49 -0700)]
st/xorg: Fix memory leak on error path.
Brian Paul [Tue, 19 Oct 2010 15:14:17 +0000 (09:14 -0600)]
gallivm: fix incorrect type for zero vector in emit_kilp()
http://bugs.freedesktop.org/show_bug.cgi?id=30974
Brian Paul [Mon, 18 Oct 2010 15:07:54 +0000 (09:07 -0600)]
mesa: fix mesa version string construction
Now that MESA_MINOR=10, we no longer need the extra '0' in the
version string.