Marek Olšák [Wed, 22 Dec 2010 00:12:20 +0000 (01:12 +0100)]
mesa: implement new texture format I16
Marek Olšák [Wed, 22 Dec 2010 00:11:13 +0000 (01:11 +0100)]
mesa: implement new texture format L16
Marek Olšák [Tue, 21 Dec 2010 23:32:33 +0000 (00:32 +0100)]
mesa: implement new texture format A16
Marek Olšák [Tue, 21 Dec 2010 22:46:32 +0000 (23:46 +0100)]
mesa: implement new texture format AL44
Radeon GPUs can do this. R600 can even do render-to-texture.
Packing and extracting aren't implemented, but we shouldn't hit them (I think).
Tested with swrast, softpipe, and r300g.
Marek Olšák [Tue, 21 Dec 2010 17:54:50 +0000 (18:54 +0100)]
mesa: implement new texture format ARGB2101010
Radeon GPUs do support GL_RGB10_A2.
Marek Olšák [Tue, 21 Dec 2010 04:50:58 +0000 (05:50 +0100)]
st/mesa: if Z32 is unsupported, prefer Z24 to Z16
Marek Olšák [Tue, 21 Dec 2010 04:45:46 +0000 (05:45 +0100)]
st/mesa: use RGBA16 for RGB12 and RGB16
To provide enough precision if a user wants it.
Marek Olšák [Tue, 21 Dec 2010 05:41:59 +0000 (06:41 +0100)]
st/mesa: use DXT SRGB formats for COMPRESSED_SRGB
And also check if the formats are supported to return something meaningful
if compression cannot be used.
Eric Anholt [Thu, 23 Dec 2010 06:48:24 +0000 (22:48 -0800)]
i965: Keep around a copy of the VS constant surface dumping code.
Just like everywhere else, I never trust my constant uploads to
correctly put constants in the right places, even though that's so
rarely where the issue is.
Eric Anholt [Thu, 23 Dec 2010 06:46:12 +0000 (22:46 -0800)]
i965: Correct the dp_read message descriptor setup on g4x.
It's mostly like gen4 message descriptor setup, except that the sizes
of type/control changed to be like gen5. Fixes 21 piglit cases on
gm45, including the regressions in bug #32311 from increased VS
constant buffer usage.
Zhenyu Wang [Thu, 23 Dec 2010 08:30:20 +0000 (16:30 +0800)]
i965: upload multisample state for fragment program change
This makes conformance tests stable on sandybridge D0 to track
multisample state before SF/WM state.
Zhenyu Wang [Thu, 23 Dec 2010 07:06:33 +0000 (15:06 +0800)]
i965: Use MI_FLUSH_DW for blt ring flush on sandybridge
Old MI_FLUSH command is deprecated on sandybridge blt.
Vinson Lee [Thu, 23 Dec 2010 09:03:32 +0000 (01:03 -0800)]
st/mesa: Remove unnecessary header.
Vinson Lee [Thu, 23 Dec 2010 03:44:00 +0000 (19:44 -0800)]
gallivm: Disable MMX-disabling code on llvm-2.9.
The disable-mmx option was removed in llvm-2.9svn by revisions 122188
and 122189.
Fixes FDO bug 32564.
Vinson Lee [Thu, 23 Dec 2010 00:48:19 +0000 (16:48 -0800)]
gallivm: Fix 'cast from pointer to integer of different size' warning.
Fixes this GCC warning.
lp_bld_const.h: In function 'lp_build_const_int_pointer':
lp_bld_const.h:137: warning: cast from pointer to integer of different size
Vinson Lee [Wed, 22 Dec 2010 08:57:52 +0000 (00:57 -0800)]
i915g: Remove unnecessary header.
Vinson Lee [Wed, 22 Dec 2010 08:55:41 +0000 (00:55 -0800)]
llvmpipe: Remove unnecessary headers.
Vinson Lee [Wed, 22 Dec 2010 08:52:05 +0000 (00:52 -0800)]
r300g: Remove unnecessary headers.
Vinson Lee [Wed, 22 Dec 2010 08:42:23 +0000 (00:42 -0800)]
svga: Remove unnecessary header.
Vinson Lee [Wed, 22 Dec 2010 08:38:42 +0000 (00:38 -0800)]
st/vega: Remove unnecessary headers.
Henri Verbeet [Wed, 22 Dec 2010 08:18:05 +0000 (09:18 +0100)]
r600g: Remove the unused "pframebuffer" field from r600_pipe_context.
Henri Verbeet [Wed, 22 Dec 2010 08:18:05 +0000 (09:18 +0100)]
r600g: r600_new() and r600_delete() are unused.
Zhenyu Wang [Wed, 22 Dec 2010 15:52:12 +0000 (10:52 -0500)]
i965: explicit tell header present for fb write on sandybridge
Determine header present for fb write by msg length is not right
for SIMD16 dispatch, and if there're more output attributes, header
present is not easy to tell from msg length. This explicitly adds
new param for fb write to say header present or not.
Fixes many cases' hang and failure in GL conformance test.
Chia-I Wu [Wed, 22 Dec 2010 07:13:52 +0000 (15:13 +0800)]
st/egl: Assorted fixes for dri2_display_get_configs.
Set window_bit only when the visual id is greater than zero. Correct
visual types. Skip slow configs as they are not relevant. Finally, do
not return duplicated configs.
Alex Deucher [Wed, 22 Dec 2010 06:30:41 +0000 (01:30 -0500)]
r600g: remove useless switch statements
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
Chia-I Wu [Wed, 22 Dec 2010 06:21:21 +0000 (14:21 +0800)]
st/egl: Fix eglCopyBuffers.
Flush before presenting.
Chia-I Wu [Wed, 22 Dec 2010 05:59:58 +0000 (13:59 +0800)]
st/egl: Plug pbuffer leaks.
Unreference validated resources or remove unnecessary validations.
Chia-I Wu [Wed, 22 Dec 2010 05:47:50 +0000 (13:47 +0800)]
st/egl: Allow single-buffered pixmaps.
All single-buffered configs were ignored before to make sure
EGL_RENDER_BUFFER is settable for window surfaces. It is better to
allow single-buffered configs and set EGL_WINDOW_BIT only for
double-buffered ones. This way there can be single-buffered pixmaps.
Dave Airlie [Wed, 22 Dec 2010 05:58:29 +0000 (15:58 +1000)]
r600g: drop unused code in evergreen.
this code was pretty much duplicated, thanks to Henri Verbeet on irc for
pointing it out.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Chia-I Wu [Wed, 22 Dec 2010 04:27:35 +0000 (12:27 +0800)]
st/egl: Remove native_config::samples.
Multisample buffers are never requested.
Chia-I Wu [Wed, 22 Dec 2010 04:22:42 +0000 (12:22 +0800)]
st/egl: Remove native_config::slow_config.
In direct rendering scenario, it is not needed until an EGLDisplay can
support both HW and SW pipe screens.
Chia-I Wu [Wed, 22 Dec 2010 04:19:12 +0000 (12:19 +0800)]
st/egl: Remove unnecessary egl_g3d_find_pixmap_config.
It was used to find a compatible config for a given pixmap. Now that a
config is optional for pixmap surface creation, the function is not
needed.
Chia-I Wu [Wed, 22 Dec 2010 04:14:44 +0000 (12:14 +0800)]
st/egl: Make config optional for create_pixmap_surface.
eglCopyBuffers or EGL_KHR_image_pixmap require creating a pixmap surface
without a config. Make it just work without relying on
is_pixmap_supported.
Dave Airlie [Wed, 22 Dec 2010 04:54:17 +0000 (14:54 +1000)]
r600g: fix evergreen segfaults.
evergreen was crashing running even gears here.
This is a 7.10 candidate if its broken the same.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Marek Olšák [Wed, 22 Dec 2010 01:22:09 +0000 (02:22 +0100)]
r300g: fix precision issues with B10G10R10A2
Marek Olšák [Tue, 21 Dec 2010 18:26:27 +0000 (19:26 +0100)]
r300g: support B10G10R10A2 render targets only with DRM 2.8.0 or later versions
Eric Anholt [Tue, 21 Dec 2010 21:04:07 +0000 (13:04 -0800)]
i965: Avoid using float type for raw moves, to work around SNB issue.
The SNB alt-mode math does the denorm and inf reduction even for a
"raw MOV" like we do for g0 message header setup, where we are moving
values that aren't actually floats. Just use UD type, where raw MOVs
really are raw MOVs.
Fixes glxgears since
c52adfc2e1d130effea940e75690897eb5d3ceaa, but no
piglit tests had regressed(!)
Jerome Glisse [Tue, 21 Dec 2010 15:49:53 +0000 (10:49 -0500)]
r600g: avoid segfault
Candidates 7.10
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Chris Wilson [Tue, 21 Dec 2010 11:55:11 +0000 (11:55 +0000)]
intel: Check for unsupported texture when finishing using as a render target
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=32541
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Vinson Lee [Tue, 21 Dec 2010 09:25:04 +0000 (01:25 -0800)]
st/mesa: Clean up header file inclusion in st_format.h.
Vinson Lee [Tue, 21 Dec 2010 09:17:37 +0000 (01:17 -0800)]
st/mesa: Clean up header file inclusion in st_draw.h.
Ben Skeggs [Mon, 20 Dec 2010 03:39:36 +0000 (13:39 +1000)]
nouveau: fix includes for latest libdrm
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Tom Fogal [Tue, 21 Dec 2010 02:24:29 +0000 (19:24 -0700)]
Regenerate gl_mangle.h.
NOTE: This is a candidate for the 7.10 branch.
Jerome Glisse [Mon, 20 Dec 2010 20:30:42 +0000 (15:30 -0500)]
r600g: properly unset vertex buffer
Fix bug http://bugs.freedesktop.org/show_bug.cgi?id=32455
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Vinson Lee [Mon, 20 Dec 2010 19:40:54 +0000 (11:40 -0800)]
st/python: remove unused 'buf' parameter in pipe_buffer_unmap
This is a follow-up to commit
ec51092a72e2dff1e9b1362d813fe4691cda89b7.
Fixes SCons build.
Marek Olšák [Sun, 19 Dec 2010 03:17:43 +0000 (04:17 +0100)]
gallium: remove unused 'buf' parameter in pipe_buffer_unmap
Vinson Lee [Mon, 20 Dec 2010 09:24:26 +0000 (01:24 -0800)]
st/mesa: Remove comment cruft from st_context.h.
This was unintentionally added by commit
1525fb4afec0adc164948b4060ec1c0359441cd3.
Vinson Lee [Mon, 20 Dec 2010 09:15:04 +0000 (01:15 -0800)]
st/mesa: Clean up header file inclusion in st_cb_texture.h.
Vinson Lee [Mon, 20 Dec 2010 09:00:26 +0000 (01:00 -0800)]
st/mesa: Clean up header file inclusion in st_cb_readpixels.h.
Marek Olšák [Sun, 19 Dec 2010 01:54:43 +0000 (02:54 +0100)]
r300g: optimize the fallback for misaligned ushort indices
Vinson Lee [Sat, 18 Dec 2010 09:44:52 +0000 (01:44 -0800)]
st/mesa: Clean up header file inclusion in st_cb_program.h.
Vinson Lee [Sat, 18 Dec 2010 09:28:18 +0000 (01:28 -0800)]
st/mesa: Clean up header file inclusion in st_cb_accum.h.
Vinson Lee [Sat, 18 Dec 2010 09:16:53 +0000 (01:16 -0800)]
mesa: Clean up header file inclusion in prog_statevars.h.
Dave Airlie [Sat, 18 Dec 2010 07:48:21 +0000 (17:48 +1000)]
mesa: fix queryobj whitespace.
Had done this before pushing but forgot to amend, doh.
Dave Airlie [Sat, 11 Sep 2010 20:31:30 +0000 (06:31 +1000)]
mesa/swrast/st: add ARB_occlusion_query2 support.
This gets my vote for most pointless extension of all time, I'm guessing
some driver could possibly optimise for this instead of counting it might
just get a true/false, but I'm not really sure.
need this to eventually advertise 3.3 despite its total uselessness.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Chia-I Wu [Fri, 10 Dec 2010 19:27:54 +0000 (03:27 +0800)]
mapi: Clean up sources.mk.
Rename MAPI_GLAPI_SOURCES to MAPI_UTIL_SOURCES. Rename macro
MAPI_GLAPI_CURRENT to MAPI_MODE_UTIL. Update the comments to make it
clear that mapi may be used in two ways and how.
Chia-I Wu [Fri, 10 Dec 2010 09:14:04 +0000 (17:14 +0800)]
mapi: Clean up u_current interface.
Try not to use macros to make u_current.h appear to be glapi.h. Use
u_current.h to implement glapi.h instead whenever possible.
Chia-I Wu [Thu, 16 Dec 2010 19:11:53 +0000 (03:11 +0800)]
mapi: Add ABI-tag note.
TLS requires kernel >= 2.4.20. Per glapi.
Kenneth Graunke [Sat, 18 Dec 2010 03:28:47 +0000 (19:28 -0800)]
Refresh autogenerated file builtin_function.cpp.
NOTE: The 7.9 and 7.10 branches will need their builtins refreshed too.
Rather than cherry-picking this commit, run 'make builtins'.
Kenneth Graunke [Sat, 18 Dec 2010 03:25:37 +0000 (19:25 -0800)]
glsl/builtins: Compute the correct value for smoothstep(vec, vec, vec).
These mistakenly computed 't' instead of t * t * (3.0 - 2.0 * t).
Also, properly vectorize the smoothstep(float, float, vec) variants.
NOTE: This is a candidate for the 7.9 and 7.10 branches.
José Fonseca [Fri, 17 Dec 2010 20:08:35 +0000 (20:08 +0000)]
gallivm: Cleanup util_format_xxx_fetch_xxx call generation.
No need to register function prototypes in the module now that we call
the C function pointer directly -- less LLVM objects lying around.
Limited testing with lp_test_format.
Kenneth Graunke [Mon, 13 Dec 2010 19:19:38 +0000 (11:19 -0800)]
glsl: Expose a public glsl_type::void_type const pointer.
This is analogous to glsl_type::int_type and all the others.
Marek Olšák [Fri, 17 Dec 2010 07:59:32 +0000 (08:59 +0100)]
r300g: finally fix the texture corruption on r3xx-r4xx
Even though a bound texture stays bound when calling set_fragment_sampler_views,
it must be assigned a new cache region depending on the occupancy of other
texture units.
This fixes:
https://bugs.freedesktop.org/show_bug.cgi?id=28800
Thanks to Álmos <aaalmosss@gmail.com> for finding the bug in the code.
NOTE: This is a candidate for both the 7.9 and 7.10 branches.
Kenneth Graunke [Fri, 17 Dec 2010 01:40:00 +0000 (17:40 -0800)]
Remove OES_compressed_paletted_texture from the ES2 extension list.
We don't support it.
Brian Paul [Fri, 17 Dec 2010 01:17:20 +0000 (18:17 -0700)]
softpipe: remove sp_tex_tile_cache_border_color()
With swizzling done at the end of texture sampling, we can greatly
simplify swizzling of the border color.
Fixes http://bugs.freedesktop.org/show_bug.cgi?id=32460
Brian Paul [Fri, 17 Dec 2010 00:38:16 +0000 (17:38 -0700)]
softpipe: fix depth texture sampling regression
We need to keep using the pipe_get_tile_swizzle() even though there's
no swizzling because we need to explicitly pass in the surface format.
Fixes http://bugs.freedesktop.org/show_bug.cgi?id=32459
Brian Paul [Thu, 16 Dec 2010 21:29:15 +0000 (14:29 -0700)]
gallivm: fix copy&paste error from previous commit
Fixes piglit regression, http://bugs.freedesktop.org/show_bug.cgi?id=32452
NOTE: This is a candidate for the 7.10 branch
richard [Thu, 16 Dec 2010 20:52:55 +0000 (15:52 -0500)]
r600c : inline vertex format is not updated in an app, switch to use vfetch constants. For the 7.9 and 7.10 branches as well.
Eric Anholt [Wed, 15 Dec 2010 20:10:03 +0000 (12:10 -0800)]
intel: Support glCopyTexImage() from XRGB8888 to ARGB8888.
The only mismatch between the two is that we have to clear the
destination's alpha to 1.0. Fixes WOW performance on my Ironlake,
from a few frames a second to almost playable.
Eric Anholt [Wed, 15 Dec 2010 19:54:38 +0000 (11:54 -0800)]
intel: Try to sanely check that formats match for CopyTexImage.
Before, we were going off of a couple of known (hopeful) matches
between internalFormats and the cpp of the read buffer. Instead, we
can now just look at the gl_format of the two to see if they match.
We should avoid bad blits that might have been possible before, but
also allow different internalFormats to work without having to
enumerate each one.
Eric Anholt [Wed, 15 Dec 2010 19:46:37 +0000 (11:46 -0800)]
intel: Drop commented intel_flush from copy_teximage.
The blit that follows appears in the command stream so it's serialized
with previous rendering. Any queued vertices in the tnl layer were
already flushed up in mesa/main/.
Eric Anholt [Wed, 15 Dec 2010 19:37:45 +0000 (11:37 -0800)]
intel: Update renderbuffers before looking up CopyTexImage's read buffer.
Not fixing a particular bug, just noticed by code inspection.
Brian Paul [Thu, 16 Dec 2010 17:13:02 +0000 (10:13 -0700)]
gallivm: work around LLVM 2.6 bug when calling C functions
Create a constant int pointer to the C function, then cast it to the
function's type. This avoids using trampoline code which seem to be
inadvertantly freed by LLVM in some situations (which leads to segfaults).
The root issue and work-around were found by José.
NOTE: This is a candidate for the 7.10 branch
Brian Paul [Thu, 16 Dec 2010 15:50:42 +0000 (08:50 -0700)]
draw: s/varient/variant/
Brian Paul [Thu, 16 Dec 2010 15:50:30 +0000 (08:50 -0700)]
svga: s/varient/variant/
Brian Paul [Thu, 16 Dec 2010 15:47:37 +0000 (08:47 -0700)]
i965g: s/varient/variant/
Brian Paul [Thu, 16 Dec 2010 15:46:35 +0000 (08:46 -0700)]
i915g: s/varient/variant/
Brian Paul [Thu, 16 Dec 2010 15:44:12 +0000 (08:44 -0700)]
softpipe: s/varient/variant
Brian Paul [Thu, 16 Dec 2010 15:43:21 +0000 (08:43 -0700)]
st/mesa: s/varient/variant
Eric Anholt [Wed, 15 Dec 2010 23:15:02 +0000 (15:15 -0800)]
i965: Set the alternative floating point mode on gen6 VS and WM.
This matches how we did the math instructions pre-gen6, though it
applies to non-math as well.
Fixes vp1-LIT test 2 (degenerate case: 0 ^ 0 -> 1)
Shuang He [Wed, 15 Dec 2010 03:16:49 +0000 (11:16 +0800)]
i915: Fix INTEL_DEBUG=wm segmentation fault
The program should be disassembled after it's uploaded
Ben Skeggs [Thu, 16 Dec 2010 01:12:52 +0000 (11:12 +1000)]
nvfx: fix fragprog word swapping on big-endian machines
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Jerome Glisse [Wed, 15 Dec 2010 20:27:16 +0000 (15:27 -0500)]
gallium: properly check for src->dst blit compatibilities
Spotted by Christoph Bumiller & Jose Fonseca
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Fredrik Höglund [Wed, 15 Dec 2010 19:00:42 +0000 (20:00 +0100)]
r600g: fix pow(0, 0) evaluating to NaN
We have to use the non-IEEE compliant version of MUL here, since
log2(0) is -inf, and 0 * -inf is NaN in IEEE arithmetic.
candidates for 7.10 branch
Jerome Glisse [Wed, 15 Dec 2010 17:07:09 +0000 (12:07 -0500)]
r600g: need to reference upload buffer as the might still live accross flush
Can't get away from referencing upload buffer as after flush a vertex buffer
using the upload buffer might still be active. Likely need to simplify the
pipe_refence a bit so we don't waste so much cpu time in it.
candidates for 7.10 branch
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Brian Paul [Tue, 14 Dec 2010 21:15:20 +0000 (14:15 -0700)]
st/mesa: fix incorrect prev pointer in destroy_program_variants()
Brian Paul [Tue, 14 Dec 2010 20:01:00 +0000 (13:01 -0700)]
softpipe: do texture swizzle during texture sampling
Instead of when we read texture tiles. Now swizzling happens after
the shadow depth compare step. This fixes the piglit glsl-fs-shadow2d*
tests (except for proj+bias because of a GLSL bug).
Brian Paul [Tue, 14 Dec 2010 19:46:01 +0000 (12:46 -0700)]
mesa: more program debug code
Brian Paul [Tue, 14 Dec 2010 19:45:36 +0000 (12:45 -0700)]
tgsi: document texture opcodes
Brian Paul [Mon, 13 Dec 2010 15:41:08 +0000 (08:41 -0700)]
glsl: new glsl_strtod() wrapper to fix decimal point interpretation
We always want to use '.' as the decimal point.
See http://bugs.freedesktop.org/show_bug.cgi?id=24531
NOTE: this is a candidate for the 7.10 branch.
Brian Paul [Tue, 14 Dec 2010 17:38:15 +0000 (10:38 -0700)]
gallivm: do texture swizzle after shadow compare
We need to swizzle after the shadow comparison so that the GL_DEPTH_MODE
functionality is handled properly.
This fixes all the piglit glsl-fs-shadow2d*.shader_test cases, except
for glsl-fs-shadow2dproj-bias.shader_test which fails because of a
bug in the GLSL compiler (fd.o 32395).
Brian Paul [Tue, 14 Dec 2010 15:36:39 +0000 (08:36 -0700)]
st/mesa: rename the varient release functions
Jerome Glisse [Tue, 14 Dec 2010 18:50:46 +0000 (13:50 -0500)]
r600g: fix segfault when translating vertex buffer
Note the support for non float vertex draw likely regressed need to
find what we want to do there.
candidates for 7.10 branches
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
Vinson Lee [Tue, 14 Dec 2010 08:39:57 +0000 (00:39 -0800)]
mesa: Clean up header file inclusion in prog_optimize.h.
Vinson Lee [Tue, 14 Dec 2010 08:30:24 +0000 (00:30 -0800)]
mesa: Clean up header file inclusion in prog_cache.h.
Vinson Lee [Tue, 14 Dec 2010 08:22:27 +0000 (00:22 -0800)]
mesa: Clean up header file inclusion in nvvertparse.h.
Eric Anholt [Mon, 13 Dec 2010 19:32:19 +0000 (11:32 -0800)]
i965: Add support for using the BLT ring on gen6.
Eric Anholt [Mon, 13 Dec 2010 23:18:13 +0000 (15:18 -0800)]
i965: Improve the hacks for ARB_fp scalar^scalar POW on gen6.
This is still awful, but my ability to care about reworking the old
backend so we can just get a temporary value into a POW is awfully low
since the new backend does this all sensibly.
Fixes:
fp1-LIT test 1
fp1-LIT test 3 (case x < 0)
fp1-POW test (exponentiation)
fp-lit-mask
Brian Paul [Tue, 14 Dec 2010 00:34:07 +0000 (17:34 -0700)]
st/mesa: 80-columns wrapping, whitespace fixes in st_cb_bitmap.c
Brian Paul [Tue, 14 Dec 2010 00:29:56 +0000 (17:29 -0700)]
st/mesa: add geom program code in destroy_program_variants()