Brian Paul [Thu, 18 Mar 2010 21:04:31 +0000 (15:04 -0600)]
progs/tests: also test stencil INCR_WRAP mode if supported
Brian Paul [Thu, 18 Mar 2010 21:02:13 +0000 (15:02 -0600)]
gallivm: fix broken INCR/DECR stencil modes
We were mistakenly using the wrong data type for stencil values before.
Brian Paul [Thu, 18 Mar 2010 20:53:43 +0000 (14:53 -0600)]
progs/tests: comments and clean-ups
Brian Paul [Thu, 18 Mar 2010 19:02:53 +0000 (13:02 -0600)]
gallivm/llvmpipe: added lp_rast_shader_inputs::facing and pass through
The triangle rasterizer sets this field to indicate front/back-facing.
It gets passed into the generated fragment code as another parameter.
Used now for stencil front/back selection but will also be used for
fragment shaders in general (see TGSI_SEMANTIC_FACE).
With this commit two-sided stenciling mostly works but there's
still a bug or two...
Brian Paul [Thu, 18 Mar 2010 18:04:50 +0000 (12:04 -0600)]
llvmpipe: defines for RAST_WHOLE, RAST_EDGE_TEST
Alan Hourihane [Thu, 18 Mar 2010 18:01:05 +0000 (18:01 +0000)]
Add format B5G5R5X1
Brian Paul [Thu, 18 Mar 2010 17:31:38 +0000 (11:31 -0600)]
gallivm/llvmpipe: simplify front/back stencil ref value handling
Instead of passing an array, just pass two scalar values.
Brian Paul [Thu, 18 Mar 2010 17:18:01 +0000 (11:18 -0600)]
gallivm: checkpoint WIP two-sided stencil test
Brian Paul [Thu, 18 Mar 2010 15:30:13 +0000 (09:30 -0600)]
llvmpipe: set opaque = FALSE if stencil enabled
Fixes occasional bad tiles seen in some demos like progs/demos/reflect.c
Francisco Jerez [Thu, 18 Mar 2010 12:46:20 +0000 (13:46 +0100)]
dri/nouveau: Flush after texture validation.
Swizzling needs the destination surface in VRAM, but the subsequent
rendering operations making use of it are likely to not care. Fire the
ring after validation to leave the memory manager more room for
maneuvering.
Francisco Jerez [Thu, 18 Mar 2010 13:19:10 +0000 (14:19 +0100)]
dri/nouveau: Avoid pushbuf flushes in the middle of LMA setup.
Francisco Jerez [Thu, 18 Mar 2010 13:18:55 +0000 (14:18 +0100)]
dri/nouveau: Implement texcoord generation.
Francisco Jerez [Thu, 18 Mar 2010 13:13:36 +0000 (14:13 +0100)]
dri/nouveau: Implement texture matrices.
Francisco Jerez [Thu, 11 Mar 2010 20:48:51 +0000 (21:48 +0100)]
dri/nouveau: Some minor vertex submission fixes.
Xavier Chantry [Sat, 13 Mar 2010 18:28:07 +0000 (19:28 +0100)]
dri/nouveau: only reallocate texture when needed
nouveau reallocated the mipmap tree on every MIN_FILTER call to account
for mipmap change. We only need to do this if the texture does not fit
in the existing mipmap tree. This gives a big performance boost for a
game like bzflag which changes MIN_FILTER all the time for its font
rendering.
Signed-off-by: Xavier Chantry <chantry.xavier@gmail.com>
Signed-off-by: Francisco Jerez <currojerez@riseup.net>
José Fonseca [Thu, 18 Mar 2010 10:24:10 +0000 (10:24 +0000)]
llvmpipe: Fix crashes when there is no depth buffer bound.
Kristian Høgsberg [Thu, 18 Mar 2010 02:44:58 +0000 (22:44 -0400)]
intel: Call _mesa_make_current() after getting initial buffers
The default viewport is the window rectangle, which is set up by
_mesa_make_current(). To be able to do that we need to get the
window dimension (and buffers) first, so we have to call
intel_prepare_render() before we can call into _mesa_make_current().
Fixes #26676 and #26678.
Brian Paul [Wed, 17 Mar 2010 22:26:41 +0000 (16:26 -0600)]
progs/trivial: added comments
Brian Paul [Wed, 17 Mar 2010 22:24:12 +0000 (16:24 -0600)]
gallivm/llvmpipe: basic stencil testing works
Most stencil demos look OK (modulo some unrelated rendering glitches).
Only single-sided stencil test works at this point.
There are probably some bugs to be found...
Brian Paul [Wed, 17 Mar 2010 21:09:35 +0000 (15:09 -0600)]
llvmpipe: remove incorrect depth test check
Brian Paul [Wed, 17 Mar 2010 21:07:45 +0000 (15:07 -0600)]
gallivm: added lp_build_andc()
Brian Paul [Wed, 17 Mar 2010 14:34:23 +0000 (08:34 -0600)]
llvmpipe: silence some pointer/casting warnings
Brian Paul [Wed, 17 Mar 2010 00:26:51 +0000 (18:26 -0600)]
gallivm/llvmpipe: more asst changes for stencil testing
Brian Paul [Tue, 16 Mar 2010 20:32:18 +0000 (14:32 -0600)]
gallivm/llmvpipe: pass stencil refs state into z/stencil build code
Brian Paul [Tue, 16 Mar 2010 20:11:43 +0000 (14:11 -0600)]
llvmpipe: use new LP_JIT_CTX_ enums for jit context field positions
Use the new enum values rather than integers in a few places.
Brian Paul [Tue, 16 Mar 2010 20:00:40 +0000 (14:00 -0600)]
llvmpipe: added stencil ref values to jit context state
Brian Paul [Tue, 16 Mar 2010 19:50:19 +0000 (13:50 -0600)]
llvmpipe: break lines
Eric Anholt [Wed, 17 Mar 2010 16:09:54 +0000 (09:09 -0700)]
intel: Replace mt->pitch with mt->region->pitch.
The pitch is not really an inherent part of the miptree, since it's
not part of any of the layout calculations, and it's dictated by the
libdrm-allocated region pitch now.
Eric Anholt [Wed, 17 Mar 2010 16:27:48 +0000 (09:27 -0700)]
intel: Return false like other blit failure paths if out of aperture.
The primary consumer of this (miptree relayout) already has this code
for handling failure, and the other paths want to know if failure
actually occurs and do something appropriate, which may not include
memcpy.
Eric Anholt [Wed, 17 Mar 2010 17:10:37 +0000 (10:10 -0700)]
intel: Respect src pitch in _mesa_copy_rect().
If a non-zero src_y was used, this would break piglit
depth-level-clamp.
Eric Anholt [Tue, 16 Mar 2010 23:20:03 +0000 (16:20 -0700)]
intel: Rely on allocated region pitch for the miptree pitch.
Bug #26966: 945 miptree pitch disagreement with libdrm.
Eric Anholt [Wed, 17 Mar 2010 16:26:37 +0000 (09:26 -0700)]
intel: Assert that the linear blits succeed.
We don't have any fallback code here, and we want to avoid this path
if failure would happen, so just assert.
Eric Anholt [Tue, 16 Mar 2010 23:11:05 +0000 (16:11 -0700)]
intel: Remove level_offset now that it's unused.
This is the last pitch-dependent part of miptree setup.
Eric Anholt [Tue, 16 Mar 2010 23:05:53 +0000 (16:05 -0700)]
intel: Remove extra tiling setting after allocating a tiled region.
Eric Anholt [Tue, 16 Mar 2010 21:02:22 +0000 (14:02 -0700)]
intel: Rename the z24_x8 depth spans to z24_s8 since they do stencil too.
Brian Paul [Wed, 17 Mar 2010 16:53:55 +0000 (10:53 -0600)]
llvmpipe: remove -m32 flag from linux-llvm config
Brian Paul [Wed, 17 Mar 2010 16:31:57 +0000 (10:31 -0600)]
st/mesa: plug in default for pipe_context::surface_copy() if needed
This lets us avoid conditionals and duplicated code in several places.
Brian Paul [Wed, 17 Mar 2010 16:17:04 +0000 (10:17 -0600)]
progs/samples: improve copy.c demo
If the test image was larger than the window, nothing was drawn because
of invalid raster position. Use glWindowPos instead of glRasterPos.
Also, use integer src/dst coordinates to avoid grabbing black pixels
outside of the src image region.
Brian Paul [Wed, 17 Mar 2010 16:13:51 +0000 (10:13 -0600)]
Merge branch '7.8'
Conflicts:
src/mesa/state_tracker/st_cb_drawpixels.c
Brian Paul [Wed, 17 Mar 2010 16:11:09 +0000 (10:11 -0600)]
progs/samples: silence warnings
Brian Paul [Wed, 17 Mar 2010 16:06:27 +0000 (10:06 -0600)]
st/mesa: fix glCopyPixels bugs/crashes when src region need clipping
Use the _mesa_clip_readpixels() function to clip the src region against
the buffer's bounds. Neatly, the resulting pixel unpack object's
SkipPixels/SkipRows fields can be used to determine the position of the
region in the destination texture.
Fixes crash in progs/samples/copy.c and probably other cases.
Brian Paul [Wed, 17 Mar 2010 15:04:26 +0000 (09:04 -0600)]
mesa: rename params in prototype to match implementation
Brian Paul [Wed, 17 Mar 2010 14:44:09 +0000 (08:44 -0600)]
glapi: fix assorted warnings
And replace some instances of GLuint with unsigned int to avoid pulling in
GL/gl.h
Brian Paul [Wed, 17 Mar 2010 14:41:47 +0000 (08:41 -0600)]
swrast: remove unused compute_coveragei() function
Corbin Simpson [Wed, 17 Mar 2010 11:35:14 +0000 (04:35 -0700)]
st/mesa: Fix build breakage.
Nearly certain this is what was intended; it compiles, but I'm not sure
this path is ever hit in my tests.
Pauli Nieminen [Wed, 17 Mar 2010 09:26:48 +0000 (11:26 +0200)]
Merge branch '7.8' into master
Conflicts:
Makefile
src/mesa/main/version.h
Brian Paul [Tue, 16 Mar 2010 22:45:54 +0000 (16:45 -0600)]
cell: build identity driver too
Brian Paul [Tue, 16 Mar 2010 22:45:36 +0000 (16:45 -0600)]
cell: add missing semicolon
Christoph Bumiller [Tue, 16 Mar 2010 21:32:42 +0000 (22:32 +0100)]
nv50: support more formats in surface_copy,fill
Fixes corrupted fonts in bzFlag, where we've been silently
failing to copy I8 mipmaps to a new miptree.
Print an error message on unsupported format now, since we
can't return failure.
Eric Anholt [Tue, 16 Mar 2010 20:23:23 +0000 (13:23 -0700)]
i965: Fix readpixels from ReadBuffer != DrawBuffer.
Fixes piglit fbo-readdrawpix.
Eric Anholt [Tue, 16 Mar 2010 20:14:38 +0000 (13:14 -0700)]
i965: Fix inversion for glCopyPixels to/from FBOs.
fixes piglit fbo-copypix.
Eric Anholt [Tue, 16 Mar 2010 19:59:59 +0000 (12:59 -0700)]
intel: Remove more code for x8z24 visuals, since we only do s8z24.
Eric Anholt [Tue, 16 Mar 2010 18:23:08 +0000 (11:23 -0700)]
Revert "i965: Do VS SGT, SLT, and friends using CMP, SEL instead of CMP, MOV, MOV."
This reverts commit
8ef3b1834a896927bdd4f2aea552cdb732849da9. Fixes
piglit glsl-vs-if.
Eric Anholt [Tue, 16 Mar 2010 18:22:29 +0000 (11:22 -0700)]
Revert "i965: Do FS SLT, SGT, and friends using CMP, SEL instead of CMP, MOV, MOV."
This reverts commit
46450c1f3f93bf4dc96696fc7e0f0eb808d9c08a. I was
wrong about null reg behavior -- it reads undefined, not 0. And
they're not kidding.
Eric Anholt [Sat, 13 Mar 2010 01:12:14 +0000 (17:12 -0800)]
meta: Fix up restoration of state if _mesa_map_pbo_source() fails.
Eric Anholt [Sat, 13 Mar 2010 01:09:51 +0000 (17:09 -0800)]
meta: Properly refcount our saved programs and texobjs.
Found while debugging bug #24119.
Brian Paul [Tue, 16 Mar 2010 19:31:42 +0000 (13:31 -0600)]
gallium: add target-helpers/wrap_screen.c to C_SOURCES
Was commented out before.
Eric Anholt [Tue, 9 Mar 2010 22:22:51 +0000 (14:22 -0800)]
i965: Fix ENDLOOP to only patch up this loop's BREAK and CONT.
Corresponds to
d225a25e21a24508aea3b877c78beb35502e942d and fixes
piglit glsl-fs-loop-nested. Bug #25173.
(cherry picked from commit
a81836ee2fe5092d695b717addf8cec91f569777)
Eric Anholt [Tue, 9 Mar 2010 19:56:14 +0000 (11:56 -0800)]
i965: Unalias all GLSL source regs from the destination regs used.
We were doing it ad-hoc before, as instructions with potential
aliasing problems were identified. But thanks to swizzling basically
anything can have aliasing, so just do it generally at source reg
setup time. This is somewhat inefficient, because sometimes an
operation doesn't need unaliasing protection if the swizzling is safe,
but the unaliasing before didn't cover those cases either.
Fixes piglit glsl-fs-loop.
(cherry picked from commit
6b194dab6b4d9f12cdd54c699b23c0d3420a49c2)
Eric Anholt [Tue, 9 Mar 2010 17:56:42 +0000 (09:56 -0800)]
i965: Fix nested loops in the VS.
We were patching up all the break and continues between the start of
our loop and the end of our loop, even if they were breaks/continues
for an inner loop. Avoiding patching already patched breaks/continues
fixes piglit glsl-vs-loop-nested.
(cherry picked from commit
f6f547d87ea68f44c50a0b0231b7360ca94b2975)
Eric Anholt [Tue, 9 Mar 2010 00:08:33 +0000 (16:08 -0800)]
i965: Fix up VP constbuf leak on program delete.
(cherry picked from commit
7f6d2754d586545ab6c970acffdd897294879039)
Eric Anholt [Mon, 8 Mar 2010 20:12:31 +0000 (12:12 -0800)]
i965: Fix up the handling of point sprite coordinate replacement.
The code was walking over the regs of pairs of attributes and checking
whether the attribute with a given reg index had point sprite enabled.
So the point sprite setup code was rarely even getting executed.
Instead, we need to determine which channels of a reg need point
sprite coordinate replacement. In addition, it was multiplying the
attribute by 1/w, when it's supposed to cover (0, 1) in each direction
regardless of w, and it wasn't filling in the Z and W components of
the texcoord as specified.
Fixes piglit point-sprite and the spriteblast demo. Bug #24431, #22245.
(cherry picked from commit
bc632d04370566c1156cbd0345fe303834f0b910)
Eric Anholt [Wed, 10 Mar 2010 19:00:40 +0000 (11:00 -0800)]
i965: Add support for the CMP opcode in the GLSL path.
This would be triggered by use of sqrt() along with control flow.
Fixes piglit-fs-sqrt-branch and a bug in Yo Frankie!.
(cherry picked from commit
48dca99feb394febc3af44e14f23fb12a9cc9204)
Eric Anholt [Wed, 10 Mar 2010 18:38:20 +0000 (10:38 -0800)]
i965: Fix the response len of masked sampler messages for 8-wide dispatch.
The bad response length would hang the GPU with a masked sample in a
shader using control flow. For 8-wide, the response length is always
4, and masked slots are just not written to. brw_wm_glsl.c already
allocates registers in the right locations.
Fixes piglit glsl-fs-bug25902 (fd.o bug #25902).
(cherry picked from commit
f6d210c284751ac50a8d6358de7e75a1ff1e4ac7)
(cherry picked from commit
dc8c0359448cdae7b367552ba58783c04b199778)
Michal Krol [Tue, 16 Mar 2010 18:39:09 +0000 (19:39 +0100)]
st/mesa: Invalidate sampler view when texture object changes.
Ian Romanick [Tue, 16 Mar 2010 14:48:14 +0000 (07:48 -0700)]
Remove nonexistant files from tarballs rule of Makefile
The Makefile should eventually be fixed to stop building tarballs if
one of the individual calls to tar fails.
Ian Romanick [Tue, 16 Mar 2010 14:38:51 +0000 (07:38 -0700)]
mesa: set version string to 7.8-rc1
Matthew W. S. Bell [Mon, 15 Mar 2010 01:18:02 +0000 (01:18 +0000)]
Correct GL_EQUIV code in r67/7xx.
From
247e121106e8d3e389f2e5a6edf13ea70ac18df7 Mon Sep 17 00:00:00 2001
These seem to be documented in
<http://www.svgopen.org/2003/papers/RasterOperationsUsingFilterElements/index.html>.
Roland Scheidegger [Tue, 16 Mar 2010 14:49:23 +0000 (15:49 +0100)]
st/xorg: fix up xorg state tracker to use cso changes
use cso_set_fragment_sampler_views instead of cso_set_sampler_textures
Pauli Nieminen [Mon, 15 Mar 2010 08:30:18 +0000 (10:30 +0200)]
radeon: Fix buffer object unmap to be called only once for dma buffers.
If flush happens inside radeonRefillCurrentMaRegion the last dma buffer would
be unmapped twice. Unmapping buffer when moving buffer to wait list fixes the
mapping error.
Brian Paul [Tue, 16 Mar 2010 13:56:56 +0000 (07:56 -0600)]
regenerate gl_mangle.h to get new EGLImage functions
José Fonseca [Tue, 16 Mar 2010 13:48:09 +0000 (13:48 +0000)]
libgl-xlib: Use a simple GALLIUM_DRIVER env var to select the pipe driver.
GALLIUM_DRIVER is being used in many other places, and it easier to
memorizing and understand than all the GALLIUM_NO_XXX.
José Fonseca [Tue, 16 Mar 2010 13:46:10 +0000 (13:46 +0000)]
gallium: Silence warning.
Roland Scheidegger [Mon, 15 Mar 2010 20:55:08 +0000 (21:55 +0100)]
gallium: change remaining util functions to use cso sampler views
changes arguments of util_blit_pixels_tex and util_gen_mipmap to
struct pipe_sampler_view * instead of struct pipe_texture *.
Michal Krol [Tue, 16 Mar 2010 09:58:33 +0000 (10:58 +0100)]
gallium/docs: Create a separate section for Sampler Views.
Keith Whitwell [Tue, 16 Mar 2010 09:02:38 +0000 (09:02 +0000)]
Merge commit 'origin/master' into gallium-sampler-view
This branch already seems to have the nv50_tex.c fix.
Conflicts:
src/gallium/drivers/nv50/nv50_tex.c
Chia-I Wu [Tue, 16 Mar 2010 07:48:03 +0000 (15:48 +0800)]
i965g: Fix use of an uninitialized variable.
Chia-I Wu [Tue, 16 Mar 2010 07:12:46 +0000 (15:12 +0800)]
Merge remote branch 'origin/gallium-st-api'
Ben Skeggs [Tue, 16 Mar 2010 03:18:39 +0000 (13:18 +1000)]
nv50: fix texturing from >=4GiB mark
Jakob Bornecrantz [Tue, 16 Mar 2010 01:17:33 +0000 (01:17 +0000)]
llvmpipe: Fix rebase typo
Jakob Bornecrantz [Thu, 11 Mar 2010 03:33:03 +0000 (03:33 +0000)]
winsys/sw: Add a software winsys layered on a pipe
Jakob Bornecrantz [Tue, 16 Mar 2010 00:12:37 +0000 (00:12 +0000)]
i965g: Fix after context transfers
Brian Paul [Tue, 16 Mar 2010 00:19:04 +0000 (18:19 -0600)]
gallivm: checkpoint: stencil test code
Brian Paul [Mon, 15 Mar 2010 20:00:23 +0000 (14:00 -0600)]
gallivm/llvmpipe: rename some constant building functions
Brian Paul [Mon, 15 Mar 2010 19:54:43 +0000 (13:54 -0600)]
gallivm: fix up some #includes
Brian Paul [Mon, 15 Mar 2010 19:50:40 +0000 (13:50 -0600)]
gallivm/llvmpipe: rename os_llvm.h to lp_bld.h
The llvm wrapper wasn't really an OS thing.
Use lp_bld.h for now but we eventually should rename/re-prefix all the
files/functions in the gallivm/ directory.
José Fonseca [Mon, 15 Mar 2010 21:03:59 +0000 (21:03 +0000)]
libgl-xlib: Obey GALLIUM_SOFTPIPE define.
Dave Airlie [Mon, 15 Mar 2010 20:49:38 +0000 (06:49 +1000)]
gallium: fix frontbuffer rendering with r300g
No idea if this is the correct fix, but it makes it work again at least.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Brian Paul [Mon, 15 Mar 2010 17:46:41 +0000 (11:46 -0600)]
llvmpipe: updated status in README file
Brian Paul [Mon, 15 Mar 2010 17:31:59 +0000 (11:31 -0600)]
progs/objviewer: remove debug code
Brian Paul [Mon, 15 Mar 2010 17:27:24 +0000 (11:27 -0600)]
gallivm: fix typo/bug in lp_build_sgn()
We were never returning -1 as a result. This fixes some inverted/flipped
faces with cube mapping.
Keith Whitwell [Mon, 15 Mar 2010 16:56:19 +0000 (17:56 +0100)]
cso: Do not hold references to bound textures.
Sampler views already hold references to those.
Brian Paul [Mon, 15 Mar 2010 16:51:56 +0000 (10:51 -0600)]
gallivm: add min_lod==max_lod special-case code in lp_build_lod_selector()
When min_lod==max_lod we don't need to go through all the work of
computing the lod from partial derivatives. This is hit by the mipmap
generation utility code.
Brian Paul [Mon, 15 Mar 2010 16:43:05 +0000 (10:43 -0600)]
gallivm: asst clean-ups, comments, etc.
Brian Paul [Mon, 15 Mar 2010 16:36:56 +0000 (10:36 -0600)]
gallivm: implement minification/magnification selection
Use the LOD value to determine whether to use the minification vs.
magnification filter.
All mipmap sampling modes work now.
Brian Paul [Mon, 15 Mar 2010 16:31:20 +0000 (10:31 -0600)]
gallivm: fix incorrect clamp in lp_build_linear_mip_levels()
Brian Paul [Mon, 15 Mar 2010 16:29:04 +0000 (10:29 -0600)]
gallivm: fix incorrect floor(), itrunc()
LLVMBuildFPTrunc() should be used for double->float conversion, not
float->int conversion.
There should be a better way to compute floor(), ceil(), etc that doesn't
involve float->int->float conversion.
Marcin Baczyński [Sat, 13 Mar 2010 13:26:45 +0000 (14:26 +0100)]
Replace _mesa_strtod with _mesa_strtof.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Michal Krol [Mon, 15 Mar 2010 14:04:12 +0000 (15:04 +0100)]
st/mesa: Fix a call to st_get_stobj_sampler_view().
Michal Krol [Mon, 15 Mar 2010 14:03:44 +0000 (15:03 +0100)]
st/mesa: Pass in correct sampler view object to blitter.