OSDN Git Service
Brian Paul [Tue, 11 Nov 2008 15:16:51 +0000 (08:16 -0700)]
gallium: massage sp_vbuf_draw() and sp_vbuf_draw_arrays() to look more alike
Also, update some comments.
Brian Paul [Sat, 22 Nov 2008 00:22:16 +0000 (17:22 -0700)]
mesa: issue error, don't crash, when calling a prototyped, but undefined function
Bug #18659.
Brian Paul [Fri, 21 Nov 2008 21:24:28 +0000 (14:24 -0700)]
mesa: better variable name: s/aux/store/
Brian Paul [Fri, 21 Nov 2008 17:03:19 +0000 (10:03 -0700)]
added progs/demos/fragcoord.c - tests gl_FragCoord attribute in fragment shader
Fragment's red/greenb/blue is a function gl_FragCoord.xyz
Eric Anholt [Fri, 21 Nov 2008 09:09:47 +0000 (17:09 +0800)]
i965: Add support for accelerated CopyTexSubImage.
There were hacks in EmitCopyBlit before to adjust offsets so that y=0 after
the offsets had been adjusted for a negative pitch. It appears that those
hacks were due to an unclear and surprising aspect of the hardware: inverting
the pitch results in the blit into the specified rectangle being inverted,
without the user needing to adjust y and base offset.
Tested with piglit copytexsubimage test on 915GM and GM965. Should fix
serious performance issues with ETQW and other applications.
Eric Anholt [Fri, 21 Nov 2008 04:05:21 +0000 (12:05 +0800)]
intel: Don't glBitmap fallback with scissoring enabled.
The blit bitmap code already handles scissoring. This is a 15-100% speedup on
blender benchmark.blend thanks to avoiding fallbacks. Bug #17951.
Eric Anholt [Fri, 14 Nov 2008 02:32:44 +0000 (18:32 -0800)]
i915: Don't overwrite i915's Viewport function from generic code.
Instead, have i965 and i915 both call the generic function from their Viewport.
Jakob Bornecrantz [Mon, 24 Nov 2008 19:05:47 +0000 (20:05 +0100)]
progs: Add ignores
Jakob Bornecrantz [Mon, 24 Nov 2008 19:01:48 +0000 (20:01 +0100)]
util: Add generic tile and detile functions
Brian [Mon, 24 Nov 2008 02:15:15 +0000 (19:15 -0700)]
softpipe: remove old/unneeded dependencies between TGSI exec and softpipe
Use tgsi_sampler struct as a base class. Softpipe subclasses it and adds
the fields it needs.
Brian Paul [Fri, 21 Nov 2008 20:57:57 +0000 (13:57 -0700)]
cell: only build Cell and softpipe drivers
Robert Ellison [Fri, 21 Nov 2008 18:42:14 +0000 (11:42 -0700)]
CELL: use variant-length fragment ops programs
This is a set of changes that optimizes the memory use of fragment
operation programs (by using and transmitting only as much memory as is
needed for the fragment ops programs, instead of maximal sizes), as well
as eliminate the dependency on hard-coded maximal program sizes. State
that is not dependent on fragment facing (i.e. that isn't using
two-sided stenciling) will only save and transmit a single
fragment operation program, instead of two identical programs.
- Added the ability to emit a LNOP (No Operation (Load)) instruction.
This is used to pad the generated fragment operations programs to
a multiple of 8 bytes, which is necessary for proper operation of
the dual instruction pipeline, and also required for proper SPU-side
decoding.
- Added the ability to allocate and manage a variant-length
struct cell_command_fragment_ops. This structure now puts the
generated function field at the end, where it can be as large
as necessary.
- On the PPU side, we now combine the generated front-facing and
back-facing code into a single variant-length buffer (and only use one
if the two sets of code are identical) for transmission to the SPU.
- On the SPU side, we pull the correct sizes out of the buffer,
allocate a new code buffer if the one we have isn't large enough,
and save the code to that buffer. The buffer is deallocated when
the SPU exits.
- Commented out the emit_fetch() static function, which was not being used.
José Fonseca [Thu, 20 Nov 2008 20:26:52 +0000 (05:26 +0900)]
glut: MinGW portability fixes.
Still, it doesn't run as well as the glut binaries...
José Fonseca [Thu, 20 Nov 2008 20:25:31 +0000 (05:25 +0900)]
gdi: Reimplement using the WGL statetracker.
José Fonseca [Thu, 20 Nov 2008 20:24:35 +0000 (05:24 +0900)]
mesa: Bring in new mesa sub-statetracker.
Some code cleanup is still in order.
José Fonseca [Thu, 20 Nov 2008 18:40:48 +0000 (03:40 +0900)]
scons: Build progs.
Just demos and trivial dirs for starters.
Robert Ellison [Thu, 20 Nov 2008 17:00:51 +0000 (10:00 -0700)]
CELL: improve twiddling/untwiddling error text
As suggested by Brian Paul: in the case of a twiddling error, instead
of reporting the bad format number (which is all but unusable), report
the more useful enum name.
Alan Hourihane [Thu, 20 Nov 2008 13:44:13 +0000 (13:44 +0000)]
Merge commit 'origin/master' into gallium-0.2
Xiang, Haihao [Thu, 20 Nov 2008 08:54:16 +0000 (16:54 +0800)]
mesa: fix shadow sampling unit issue.
texture comparison logic is bypassed if the currently bound texture is not
a depth/depth_stencil texture.
airlied [Thu, 20 Nov 2008 11:27:45 +0000 (21:27 +1000)]
intel: fix i830 comment + backwards VB offsets.
According to Keith the docs have these offsets the other way around
airlied [Thu, 20 Nov 2008 11:14:45 +0000 (21:14 +1000)]
intel: fix i8xx vbo enable bit
airlied [Thu, 20 Nov 2008 08:18:20 +0000 (18:18 +1000)]
intel: add lots of i830 engine to intel_decode debug
Brian Paul [Wed, 19 Nov 2008 23:04:37 +0000 (16:04 -0700)]
mesa: minor comment reformattting
Brian Paul [Wed, 19 Nov 2008 22:08:46 +0000 (15:08 -0700)]
mesa: glsl compiler debug code
RETURN0 macro reports file/line before returning zero.
Brian Paul [Wed, 19 Nov 2008 21:12:25 +0000 (14:12 -0700)]
mesa: rework GLSL array code generation
We now express arrays in terms of indirect addressing. For example:
dst = a[i];
becomes:
MOV dst, TEMP[1 + TEMP[2].y];
At instruction-emit time indirect addressing is converted into ARL/
ADDR-relative form:
ARL ADDR.x, TEMP[2].y;
MOV dst, TEMP[1 + ADDR.x];
This fixes a number of array-related issues. Arrays of arrays and complex
array/struct nesting works now.
There may be some regressions, but more work is coming.
Michal Krol [Wed, 19 Nov 2008 19:31:38 +0000 (20:31 +0100)]
scons: Fix toolchain selection.
Brian Paul [Fri, 14 Nov 2008 20:19:42 +0000 (13:19 -0700)]
mesa: don't realloc instruction buffer so often
Brian Paul [Fri, 14 Nov 2008 20:19:31 +0000 (13:19 -0700)]
mesa: updated comment
Keith Whitwell [Wed, 19 Nov 2008 16:04:18 +0000 (16:04 +0000)]
Merge commit 'origin/gallium-0.1' into gallium-0.2
José Fonseca [Wed, 19 Nov 2008 16:02:03 +0000 (01:02 +0900)]
python/retrace: Dump constants.
José Fonseca [Wed, 19 Nov 2008 16:01:48 +0000 (01:01 +0900)]
python: Allow to read from buffers.
José Fonseca [Wed, 19 Nov 2008 11:06:52 +0000 (20:06 +0900)]
python/retrace: Fix formatting of shaders.
José Fonseca [Wed, 19 Nov 2008 11:06:04 +0000 (20:06 +0900)]
python/retrace: Ignore irrelevant calls.
Alan Hourihane [Wed, 19 Nov 2008 09:35:37 +0000 (09:35 +0000)]
add SRGB formats
José Fonseca [Wed, 19 Nov 2008 08:17:06 +0000 (17:17 +0900)]
python/retrace: Highlight the trace dump to help to visualize.
José Fonseca [Wed, 19 Nov 2008 07:23:01 +0000 (16:23 +0900)]
python/retrace: Use the usual BSD-style license.
Xiang, Haihao [Wed, 19 Nov 2008 03:22:35 +0000 (11:22 +0800)]
mesa: clamp luminance if needed.
This fixes glReadPixels(GL_LUMINANCE, GL_FLOAT)/glGetTexImage(GL_LUMINANCE, GL_FLOAT) issue
on fixed-point color buffers.
Michal Krol [Fri, 14 Nov 2008 12:31:06 +0000 (13:31 +0100)]
tgsi: Return 0.0 for negative constant register indices.
Michal Krol [Fri, 14 Nov 2008 12:26:01 +0000 (13:26 +0100)]
tgsi: Keep address register as a floating point.
José Fonseca [Tue, 18 Nov 2008 10:13:32 +0000 (19:13 +0900)]
scons: Support MinGW32 cross compiler.
To build an alternative opengl32.dll with Gallium's software-rasterizer from a
debian-based distribution run:
sudo apt-get install mingw32
scons platform=windows toolchain=crossmingw machine=x86 winsys=gdi dri=no
José Fonseca [Mon, 17 Nov 2008 07:40:21 +0000 (16:40 +0900)]
gallium: Use costum log2 for all windows builds.
José Fonseca [Thu, 13 Nov 2008 11:34:10 +0000 (20:34 +0900)]
gallium: State when there are no memory leaks detected.
José Fonseca [Tue, 11 Nov 2008 14:27:27 +0000 (23:27 +0900)]
gallium: Make handle_table reentrant.
Ensure that the object has consistent state also when calling the destroy
callback. Namely, ensure the object passed to the callback is removed from
the table prior to calling the destroy callback to avoid a infinite loop or
double free.
José Fonseca [Tue, 4 Nov 2008 01:53:02 +0000 (10:53 +0900)]
gallium: Yet another WinCE portability fix.
José Fonseca [Thu, 13 Nov 2008 11:34:10 +0000 (20:34 +0900)]
gallium: State when there are no memory leaks detected.
José Fonseca [Tue, 11 Nov 2008 14:27:27 +0000 (23:27 +0900)]
gallium: Make handle_table reentrant.
Ensure that the object has consistent state also when calling the destroy
callback. Namely, ensure the object passed to the callback is removed from
the table prior to calling the destroy callback to avoid a infinite loop or
double free.
Stephane Marchesin [Sun, 16 Nov 2008 19:32:05 +0000 (20:32 +0100)]
gallivm: fix some small stuff.
Jakob Bornecrantz [Sun, 16 Nov 2008 11:33:58 +0000 (12:33 +0100)]
i915: Silence warnings
Keith Whitwell [Sat, 15 Nov 2008 16:53:24 +0000 (16:53 +0000)]
Merge commit 'origin/master' into gallium-0.2
Conflicts:
src/mesa/shader/prog_print.c
Keith Whitwell [Sat, 15 Nov 2008 16:23:31 +0000 (16:23 +0000)]
Merge commit 'origin/gallium-0.1' into gallium-0.2
Conflicts:
src/mesa/drivers/dri/common/dri_util.c
Jakob Bornecrantz [Sat, 15 Nov 2008 11:10:32 +0000 (12:10 +0100)]
i915: Silence warning
Brian Paul [Fri, 14 Nov 2008 19:55:05 +0000 (12:55 -0700)]
gallium: increase table size for fast log/pow functions
The various conformance tests pass now.
Brian Paul [Thu, 13 Nov 2008 15:58:47 +0000 (08:58 -0700)]
util: Use OpenGL rasterization rules in blits and mipmap generation.
Keith Whitwell [Fri, 14 Nov 2008 17:59:29 +0000 (17:59 +0000)]
translate: pull in prefetch and other optimizations from draw_vs_aos.c
Keith Whitwell [Wed, 5 Nov 2008 11:31:57 +0000 (11:31 +0000)]
trivial: more tests
Alan Hourihane [Fri, 14 Nov 2008 14:30:49 +0000 (14:30 +0000)]
mesa: fix build
Brian Paul [Fri, 14 Nov 2008 01:18:07 +0000 (18:18 -0700)]
mesa: no longer need Writemask field in GLSL IR nodes
The Swizzle and Size fields carry all the info we need now.
Brian Paul [Fri, 14 Nov 2008 00:02:11 +0000 (17:02 -0700)]
mesa: revamp GLSL instruction emit code
This is a step toward better array handling code. In particular, when more
than one operand of an instruction uses indirect addressing, we'll need some
temporary instructions and registers. By converting IR storage to instruction
operands all in one place (emit_instruction()) we can be smarter about this.
Also, somewhat better handling of dst register swizzle/writemask handling.
This results in tighter writemasks on some instructions which is good for
SOA execution.
And, cleaner instruction commenting with inst_comment().
Next: remove some more dead code and additional clean-ups...
Brian Paul [Thu, 13 Nov 2008 23:37:52 +0000 (16:37 -0700)]
mesa: make writemask_string() non-static
Brian Paul [Thu, 13 Nov 2008 23:31:27 +0000 (16:31 -0700)]
mesa: remove some do-nothing GLSL code
Brian Paul [Thu, 13 Nov 2008 15:47:36 +0000 (08:47 -0700)]
mesa: fix accidental regression in GLSL built-in texture matrix lookup
Was broken by commit
9aca9a4b72b2a7b378e50bd88f9c3324d07375ec.
Brian Paul [Thu, 13 Nov 2008 15:46:45 +0000 (08:46 -0700)]
mesa: use the tighter definition of GLSL ftransform() from the gallium branches
Brian Paul [Wed, 12 Nov 2008 23:47:03 +0000 (16:47 -0700)]
mesa: remove unused/obsolete __NormalMatrixTranspose matrix
Brian Paul [Wed, 12 Nov 2008 23:44:47 +0000 (16:44 -0700)]
mesa: fix bug in GLSL built-in matrix state lookup
Brian Paul [Wed, 12 Nov 2008 18:12:10 +0000 (11:12 -0700)]
mesa: tweak program register printing for RelAddr case
Alan Hourihane [Thu, 13 Nov 2008 19:38:18 +0000 (19:38 +0000)]
build default set of progs + egl
Pekka Paalanen [Thu, 13 Nov 2008 18:44:55 +0000 (20:44 +0200)]
Fix configs/linux-dri to build winsys again.
SRC_DIRS was being reset, which left winsys unbuilt.
Signed-off-by: Pekka Paalanen <pq@iki.fi>
Robert Ellison [Thu, 13 Nov 2008 18:22:12 +0000 (11:22 -0700)]
CELL: fix stencil twiddling, stencil invert
Many stencil tests were failing because of a failure to read the
stencil buffer, due to "twiddling" (or "untwiddling") "an unsupported
texture format". This is fixed for the case of a stencil/Z S824Z format
(which twiddles just like the 32-bit color formats).
tests/stencilwrap.c was failing on the GL_INVERT test, because
the emitted code for "spe_xori" turned out not to be an actual
"xori" instruction, but rather a "stqd" instruction, because
of a typo in the rtasm code. This is now fixed, and
tests/stencil_wrap now works.
Stephane Marchesin [Thu, 13 Nov 2008 17:06:06 +0000 (18:06 +0100)]
Merge branch 'gallium-0.2' of git+ssh://marcheu@git.freedesktop.org/git/nouveau/mesa into gallium-0.2
Ben Skeggs [Thu, 13 Nov 2008 16:47:47 +0000 (16:47 +0000)]
gallium: add st_set_teximage_surface for EXT_texture_from_pixmap
support.
Alan Hourihane [Thu, 13 Nov 2008 13:16:03 +0000 (13:16 +0000)]
mesa: fix generation of fixed function state when no vp exists
Alan Hourihane [Thu, 13 Nov 2008 13:16:03 +0000 (13:16 +0000)]
mesa: fix generation of fixed function state when no vp exists
Michal Krol [Wed, 12 Nov 2008 22:23:49 +0000 (23:23 +0100)]
tgsi: More comments on source register indirect and 2D indexing.
Eric Anholt [Wed, 12 Nov 2008 21:37:16 +0000 (13:37 -0800)]
i965: Upload state on primitive switch, don't just prepare it.
This was a regression in
59b2c2adbbece27ccf54e58b598ea29cb3a5aa85 that broke
blender, among other apps.
Eric Anholt [Wed, 12 Nov 2008 21:34:20 +0000 (13:34 -0800)]
i965: Fix VB refcount leak on aperture overflow.
Pekka Paalanen [Wed, 12 Nov 2008 20:37:01 +0000 (22:37 +0200)]
nv04..nv30: fix pipe_surface init
Signed-off-by: Pekka Paalanen <pq@iki.fi>
Pekka Paalanen [Wed, 12 Nov 2008 21:13:22 +0000 (23:13 +0200)]
Merge commit 'origin/gallium-0.1' into gallium-0.2
* commit 'origin/gallium-0.1':
Mesa: fix number of buffers in st_draw_vbo().
Nouveau: fix nv20_vertex_layout()
Nouveau: update nv20 miptree according to nv40.
Nouveau: build, link and use nv20.
Nouveau: name replace for nv20.
Nouveau: Rename nv20/ files as nv20.
Nouveau: copy nv30_vertprog.c to nv20.
Nouveau: start nv20 by copying the nv10 tree.
Eric Anholt [Tue, 4 Nov 2008 20:34:29 +0000 (12:34 -0800)]
Add glsync demo program from jbarnes for testing vblank synchronization.
Eric Anholt [Mon, 3 Nov 2008 02:50:59 +0000 (18:50 -0800)]
i965: Fix up VS max_threads for G4X and removing a magic number.
As far as I can read in the docs, VS threads can be 1:1 with the pairs of
VUE handles allocated for them. Also, G4X can run twice as many threads as
before (though we won't unless the we bump the preferred URB entries for VS).
Eric Anholt [Mon, 3 Nov 2008 02:49:33 +0000 (18:49 -0800)]
i965: Fix up SF max_threads.
We were dividing the number of URB entries by two to get number of threads,
which looks suspiciously like a copy'n'paste-o from brw_vs_state.c. Also, the
maximum number of threads is 24, not 12.
Eric Anholt [Mon, 3 Nov 2008 02:43:10 +0000 (18:43 -0800)]
i965: Fix up clip min_nr_entries, preferred_nr_entries, and max_threads.
The clip thread could potentially deadlock when processing tristrips since
being moved back to dual-thread mode, as the two threads could each have 4 VUEs
referenced and not be able to allocate another one since SF processing
wasn't able to continue (needing 5 entries before it freed 2).
In constrained URB mode, similar deadlock could even have occurred with
polygons (so we cut back max_threads if we can't handle it any primitive type).
Eric Anholt [Mon, 3 Nov 2008 02:42:17 +0000 (18:42 -0800)]
i965: Update WM maximum threads for G4X.
Eric Anholt [Mon, 3 Nov 2008 02:48:25 +0000 (18:48 -0800)]
i965: Add a big comment explaining my understanding of URB management.
It shouldn't offer anything new over what's in the docs (except for G4X notes),
but here it's all in one place.
Kristian Høgsberg [Wed, 12 Nov 2008 20:24:45 +0000 (15:24 -0500)]
glFlush before CopySubBuffer, fix coordinates and extension name typo.
Robert Ellison [Wed, 12 Nov 2008 19:23:52 +0000 (12:23 -0700)]
CELL: fix stencil test bugs
Fixed a boneheaded error in the generation of SPU code that calculates
the results of the stencil test. Basically, all the greater than/less than
calculations were exactly inverted: they were coded as though the
given comparison took the stencil value as a left-hand operand and the
reference value as a right-hand operand, but the actual semantics always
put the reference as the left-hand operand and the stencil as the right-hand
operand.
With this fix, tests/dinoshade runs, as do all the other Mesa tests
and samples that use stencil (and that don't use texture formats
unsupported by Cell).
Brian Paul [Wed, 12 Nov 2008 18:09:12 +0000 (11:09 -0700)]
gallium: add missing prototypes
Brian Paul [Wed, 12 Nov 2008 18:05:34 +0000 (11:05 -0700)]
cell: move semicolons to silence warnings w/ other compilers
Brian Paul [Wed, 12 Nov 2008 18:04:17 +0000 (11:04 -0700)]
cell: include cell_surface.h
Brian Paul [Wed, 12 Nov 2008 18:01:40 +0000 (11:01 -0700)]
cell: fix typo in EMIT_ macro
Brian Paul [Wed, 12 Nov 2008 18:01:05 +0000 (11:01 -0700)]
cell: added -Wmissing-prototypes
Brian Paul [Wed, 12 Nov 2008 18:00:41 +0000 (11:00 -0700)]
cell: include cell_pipe_state.h
Michal Krol [Wed, 12 Nov 2008 18:02:41 +0000 (19:02 +0100)]
util: Optimise log2().
Michal Krol [Wed, 12 Nov 2008 18:01:46 +0000 (19:01 +0100)]
util: Optimise log2().
Michal Krol [Wed, 12 Nov 2008 17:44:20 +0000 (18:44 +0100)]
rtasm: Use INLINE keyword. Compile for all platforms, not only GALLIUM_CELL.
Michal Krol [Wed, 12 Nov 2008 16:14:07 +0000 (17:14 +0100)]
util: Add log2() definition for MSC.
Michal Krol [Wed, 12 Nov 2008 17:19:20 +0000 (18:19 +0100)]
tgsi: Fix a bug with saving/restoring xmm registers upon func call.
Michal Krol [Wed, 12 Nov 2008 17:13:58 +0000 (18:13 +0100)]
rtasm: Compile only for GALLIUM_CELL.
Michal Krol [Wed, 12 Nov 2008 16:03:58 +0000 (17:03 +0100)]
draw: Add missing include.
Michal Krol [Wed, 12 Nov 2008 16:14:07 +0000 (17:14 +0100)]
util: Add log2() definition for MSC.
Michal Krol [Wed, 12 Nov 2008 16:03:58 +0000 (17:03 +0100)]
draw: Add missing include.