OSDN Git Service
Corbin Simpson [Wed, 18 Feb 2009 05:30:55 +0000 (21:30 -0800)]
r300-gallium: Consolidate state updates.
Corbin Simpson [Wed, 18 Feb 2009 03:23:09 +0000 (19:23 -0800)]
r300-gallium: Hook up new swtcl vbuf stage.
Hold on to your hats.
Corbin Simpson [Wed, 18 Feb 2009 03:04:39 +0000 (19:04 -0800)]
r300-gallium: Turn swtcl_emit into a vbuf_render stage.
Movin' out of the Stone Ages.
Corbin Simpson [Wed, 18 Feb 2009 02:45:19 +0000 (18:45 -0800)]
r300-gallium: Update r300_reg from classic Mesa.
Mostly needed a few defines for index buffers, but there's other goodies too.
Jakob Bornecrantz [Fri, 13 Feb 2009 16:53:49 +0000 (17:53 +0100)]
demos: Add polygon mode point to isosurf
Jakob Bornecrantz [Tue, 17 Feb 2009 15:04:02 +0000 (15:04 +0000)]
nv20: Use the new draw vbuf interface
Jakob Bornecrantz [Tue, 17 Feb 2009 14:47:42 +0000 (14:47 +0000)]
nv10: Use the new draw vbuf internface
Jakob Bornecrantz [Tue, 17 Feb 2009 14:37:20 +0000 (14:37 +0000)]
nv04: Use the new draw vbuf interface
Jakob Bornecrantz [Fri, 13 Feb 2009 10:37:25 +0000 (11:37 +0100)]
i915: Use the new draw vbuf interface
Keith Whitwell [Tue, 17 Feb 2009 12:39:05 +0000 (12:39 +0000)]
draw: second argument to unmap is max, not count
Keith Whitwell [Tue, 17 Feb 2009 12:20:05 +0000 (12:20 +0000)]
softpipe: update to new draw interfaces
Keith Whitwell [Tue, 17 Feb 2009 12:07:09 +0000 (12:07 +0000)]
draw: add map/unmap directives for swtnl driver interface
Previously draw module asked for a pointer into (mapped) vertex data,
which it would incrementally fill and emit draw commands against. This
was hard for the drivers to deal with, especially in the case where a
draw command would force a flush and thus an unmap of the vertex data.
With this change, the draw module explicitly maps & then unmaps vertex
data prior to emitting draw commands.
Brian Paul [Mon, 16 Feb 2009 18:50:05 +0000 (11:50 -0700)]
mesa: remove old comments
Note: the default value for EmitCondCodes is FALSE. This means the GLSL
compiler will emit code like this:
SEQ TEMP[0].x, A, B;
IF TEMP[0].x;
...
ENDIF
But if EmitCondCodes is TRUE, condition codes will be used instead:
SEQ.C TEMP[0].x, A, B;
IF (NE.xxxx);
...
ENDIF
Brian Paul [Mon, 16 Feb 2009 18:45:57 +0000 (11:45 -0700)]
i965: tell GLSL compiler to emit code using condition codes
The default for EmitCondCodes got flipped when gallium-0.2 was merged.
This fixes GLSL if/else/endif regressions.
Drivers that use GLSL should always explicitly set the flag to be safe.
Patrice Mandin [Mon, 16 Feb 2009 17:43:56 +0000 (18:43 +0100)]
nouveau: oops :-)
Dan Nicholson [Mon, 16 Feb 2009 17:28:35 +0000 (09:28 -0800)]
autoconf: Always enable glew
Maybe this should turn into an option (--disable-glew), but for now just
enable glew all the time.
Brian Paul [Mon, 16 Feb 2009 15:25:57 +0000 (08:25 -0700)]
glsl: silence some uninit var warnings
Brian Paul [Mon, 16 Feb 2009 15:25:33 +0000 (08:25 -0700)]
cell: use some SPU intrinsics to get slightly better code in eval_inputs()
Suggested by Jonathan Adamczewski. There may be more places to do this...
Corbin Simpson [Mon, 16 Feb 2009 11:56:03 +0000 (03:56 -0800)]
r300-gallium: Add verbosity level to debugging.
Makes it a bit more manageable to read through the console logs.
Corbin Simpson [Mon, 16 Feb 2009 11:55:16 +0000 (03:55 -0800)]
r300-gallium: Fix BEGIN_CS and END_CS counting and mismatch.
Corbin Simpson [Mon, 16 Feb 2009 11:33:56 +0000 (03:33 -0800)]
r300-gallium: Set up draw rasterizer.
Alan Hourihane [Mon, 16 Feb 2009 11:54:58 +0000 (11:54 +0000)]
demos: fix glxpixmap and call glXWaitGL before calling XCopyArea.
Alan Hourihane [Mon, 16 Feb 2009 11:44:40 +0000 (11:44 +0000)]
dri2: support glXWaitX & glXWaitGL by using fake front buffer.
Corbin Simpson [Mon, 16 Feb 2009 11:13:16 +0000 (03:13 -0800)]
r300-gallium: Ooops, forgot to apply this stash.
"git stash": The cause of, and solution to, all my problems.
Corbin Simpson [Mon, 16 Feb 2009 11:03:28 +0000 (03:03 -0800)]
r300-gallium: Include-guard r300_emit.h
Corbin Simpson [Mon, 16 Feb 2009 10:53:34 +0000 (02:53 -0800)]
r300-gallium: Add draw_arrays and friends.
This is the last bit of Gallium-side plumbing for drawing things.
From this point on, the only missing parts should be in r3xx-specific
code areas...
Corbin Simpson [Mon, 16 Feb 2009 10:52:47 +0000 (02:52 -0800)]
r300-gallium: Add SC_CLIP_RULE to clear.
This is a register that is in r300_demo but not r300_surface, so adding it in
to see if it helps.
Brian Paul [Sun, 15 Feb 2009 19:08:47 +0000 (12:08 -0700)]
cell: minor Makefile clean-up
Brian Paul [Sun, 15 Feb 2009 19:08:02 +0000 (12:08 -0700)]
cell: new/tighter code for computing fragment program inputs
Brian Paul [Sun, 15 Feb 2009 16:25:22 +0000 (09:25 -0700)]
cell: combine eval_z(), eval_w() functions
Corbin Simpson [Sat, 14 Feb 2009 23:24:44 +0000 (15:24 -0800)]
r300-gallium: Fix scissors.
Don't use SCISSORS_OFFSET since we're DRI2,
and don't forget to set scissors in clear.
Corbin Simpson [Sat, 14 Feb 2009 12:41:29 +0000 (04:41 -0800)]
r300-gallium: Grab bag of goodies.
Some fixes from glisse, moar swtcl emit setup, cleanup a bunch of regs,
properly do clear flush, and BEGIN_CS count fixes.
Corbin Simpson [Sat, 14 Feb 2009 12:11:27 +0000 (04:11 -0800)]
r300-gallium: Clean up some clear registers.
Corbin Simpson [Sat, 14 Feb 2009 11:23:50 +0000 (03:23 -0800)]
r300-gallium: Use fui instead of a roll-my-own.
Man, util/u_math just gets better by the day.
Nicolai Haehnle [Sat, 14 Feb 2009 16:06:06 +0000 (17:06 +0100)]
r300: Redirect constant TEX coordinates
R3xx/R5xx fragment program texture constants must come from a hardware
register instead of the constant file, so we redirect if necessary during
the native rewrite phase.
The symptoms of this bug started appearing when the Mesa fixed function
texenvprogram code started using STATE_CURRENT_ATTRIB constants for
texture coordinates when the corresponding attributes were constant across
all vertices.
Signed-off-by: Nicolai Haehnle <nhaehnle@gmail.com>
Corbin Simpson [Sat, 14 Feb 2009 10:40:33 +0000 (02:40 -0800)]
radeon-gallium: Clean out extensions explicitly enabled by Mesa.
This cleans up some of the cruft from the old DRI setup, and
it turns out that only the GLSL extensions are still off if we
let st_extensions.c handle the setup instead.
Corbin Simpson [Sat, 14 Feb 2009 10:24:30 +0000 (02:24 -0800)]
r300-gallium: Update r300_screen comments and add a few formats.
Corbin Simpson [Sat, 14 Feb 2009 10:07:29 +0000 (02:07 -0800)]
r300-gallium: Enable all four colorbuffer render targets.
As far as I can tell all the state emission necessary has been set up.
Well, except for the fragment shader, but c'mon, gimme a break. :3
Corbin Simpson [Sat, 14 Feb 2009 10:06:17 +0000 (02:06 -0800)]
r300-gallium: Add Z/stencil buffer format emit.
Also set BEGIN_CS correctly.
Corbin Simpson [Sat, 14 Feb 2009 09:55:38 +0000 (01:55 -0800)]
r300-gallium: Emit Z/stencil buffer offset.
Corbin Simpson [Sat, 14 Feb 2009 07:20:43 +0000 (23:20 -0800)]
r300-gallium: Add derived state for vertex formats.
Next up: The evil RS block.
Brian Paul [Sat, 14 Feb 2009 00:17:52 +0000 (17:17 -0700)]
i965: rewrite the code for handling shader subroutine calls
Previously, the prog_instruction::Data field was used to map original Mesa
instructions to brw instructions in order to resolve subroutine calls. This
was a rather tangled mess. Plus it's an obstacle to implementing dynamic
allocation/growing of the instruction buffer (it's still a fixed size).
Mesa's GLSL compiler emits a label for each subroutine and CAL instruction.
Now we use those labels to patch the subroutine calls after code generation
has been done. We just keep a list of all CAL instructions that needs patching
and a list of all subroutine labels. It's a simple matter to resolve them.
This also consolidates some redundant post-emit code between brw_vs_emit.c and
brw_wm_glsl.c and removes some loops that cleared the prog_instruction::Data
fields at the end.
Plus, a bunch of new comments.
Brian Paul [Fri, 13 Feb 2009 23:56:25 +0000 (16:56 -0700)]
i965: add missing break for OPCODE_RET case
This doesn't effect correctness, but we were emitting an extraneous ADD.
Brian Paul [Thu, 12 Feb 2009 23:03:39 +0000 (16:03 -0700)]
i965: the return value of translate_insn() is never used. Make it void.
Brian Paul [Thu, 12 Feb 2009 22:40:04 +0000 (15:40 -0700)]
i965: minor clean-ups
Brian Paul [Thu, 12 Feb 2009 22:35:41 +0000 (15:35 -0700)]
i965: code clean-ups, comments, and minor refactoring
Brian Paul [Thu, 12 Feb 2009 22:35:02 +0000 (15:35 -0700)]
i965: updated comments
Brian Paul [Thu, 12 Feb 2009 22:04:21 +0000 (15:04 -0700)]
intel: turn on GL_ARB_shading_language_120
It's done in the Mesa GLSL compiler. The only part of it that might
matter in drivers is the centroid sampling option for MSAA.
Brian Paul [Thu, 12 Feb 2009 21:58:19 +0000 (14:58 -0700)]
i965: more reformatting/clean-up
Brian Paul [Thu, 12 Feb 2009 21:54:13 +0000 (14:54 -0700)]
i965: s/__inline/INLINE/
Brian Paul [Thu, 12 Feb 2009 21:53:30 +0000 (14:53 -0700)]
i965: formatting and indentation fixes
Brian Paul [Thu, 12 Feb 2009 21:47:56 +0000 (14:47 -0700)]
i965: fix inconsistant indentation in brw_wm.c
Robert Ellison [Fri, 13 Feb 2009 22:19:04 +0000 (15:19 -0700)]
i965: Eric Anholt's patch for bumping up texture sizes
I'm committing this because it fixes a conform failure; the failure occurs
on the TextureProxy test, where the test attempts to create proxy textures
at every level, but fails at the last level (border == 1, width == 1,
height == 1) because it's beyond MAX_TEXTURE_LEVELS.
Eric's original comment was:
idr said that in his review swrast was ready for it, and the 965 driver is
advertising it already though it has been resulting in many crashes due to
arrays using these defines not being big enough.
Corbin Simpson [Fri, 13 Feb 2009 16:14:42 +0000 (08:14 -0800)]
r300-gallium: Various thingies.
Add formats to framebuffer emit, fix up shader function names,
make sure fragment format is emitted for r500.
Dan Nicholson [Fri, 13 Feb 2009 15:31:24 +0000 (07:31 -0800)]
Add dummy install target for glew to fix 'make install'
Signed-off-by: Dan Nicholson <dbn.lists@gmail.com>
Dan Nicholson [Fri, 13 Feb 2009 14:34:09 +0000 (06:34 -0800)]
autoconf: Add GLEW needed by progs when building GLUT
Signed-off-by: Dan Nicholson <dbn.lists@gmail.com>
Brian Paul [Fri, 13 Feb 2009 15:11:01 +0000 (08:11 -0700)]
mesa: check if TNL state is null in _tnl_free_vertices() to avoid potential segfault
_tnl_free_vertices() is called from several places during context tear-down.
Depending on the order in which the swrast, swrast_setup and tnl context is
destroyed we could hit a null pointer here. This doesn't seem to be an
actual issue with any Mesa drivers, but let's be safe.
Brian Paul [Fri, 13 Feb 2009 14:44:02 +0000 (07:44 -0700)]
mesa: add additional texture size/limit assertions
Jakob Bornecrantz [Thu, 12 Feb 2009 02:59:07 +0000 (03:59 +0100)]
egl: Doesn't depend on xorg or pixman
Corbin Simpson [Fri, 13 Feb 2009 07:36:21 +0000 (23:36 -0800)]
r300-gallium: Take care of various bad dereferences in shader setup.
Unbreaks glxinfo.
Corbin Simpson [Fri, 13 Feb 2009 04:47:15 +0000 (20:47 -0800)]
r300-gallium: Fix linker error a few linker warnings.
A few prototypes, a missing header, a misspelled macro.
Corbin Simpson [Fri, 13 Feb 2009 04:35:17 +0000 (20:35 -0800)]
r300-gallium: Fix build errors.
Joakim Sindholt [Fri, 13 Feb 2009 04:08:54 +0000 (05:08 +0100)]
r300-gallium: fix OUT_CS_ONE_REG and use where applicable
Signed-off-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Corbin Simpson [Fri, 13 Feb 2009 04:20:41 +0000 (20:20 -0800)]
r300-gallium: r300 passthrough shader, static shader objects, and clear code.
Corbin Simpson [Fri, 13 Feb 2009 04:01:09 +0000 (20:01 -0800)]
r300-gallium: Add r300 passthrough shader.
Joakim Sindholt [Fri, 13 Feb 2009 01:34:34 +0000 (02:34 +0100)]
r300: minor fixes and clear up some surface_fill
Signed-off-by: Corbin Simpson <MostAwesomeDude@gmail.com>
Corbin Simpson [Fri, 13 Feb 2009 00:53:06 +0000 (16:53 -0800)]
r300-gallium: Add r500 passthrough shader assembly.
This allows a simple passthrough fragment shader to be provided on r500.
Corbin Simpson [Fri, 13 Feb 2009 00:52:15 +0000 (16:52 -0800)]
r300-gallium: Fix typo in texture buffer size request.
Jakob Bornecrantz [Fri, 13 Feb 2009 00:17:11 +0000 (01:17 +0100)]
intel: Only build winsys with built state trackers
Jakob Bornecrantz [Thu, 12 Feb 2009 23:57:47 +0000 (00:57 +0100)]
gallium: Add a bunch of autoconf options
./configure --help for more info
Benjamin Close [Thu, 12 Feb 2009 23:37:26 +0000 (10:07 +1030)]
galium: Makefile fixes: -Werror=implicit-function-declaration -> -Werror-implicit-function-declaration
Found-By: Tinderbox
Jakob Bornecrantz [Mon, 2 Feb 2009 00:30:05 +0000 (01:30 +0100)]
intel: Make gem export the drm_api_hocks
Jakob Bornecrantz [Mon, 2 Feb 2009 00:33:38 +0000 (01:33 +0100)]
intel: Remove the old depricated dri1 winsys
Jakob Bornecrantz [Mon, 2 Feb 2009 00:32:11 +0000 (01:32 +0100)]
intel: Remove the ttm backend
RIP ttm, its been fun knowing you.
Robert Ellison [Thu, 12 Feb 2009 20:47:36 +0000 (13:47 -0700)]
glDrawBuffers(n==0) is valid
According to the GL spec, calling glDrawBuffers() with n == 0 is a
valid operation (and essentially prevents drawing to any buffers).
But _msa_DrawBuffersARB() was producing a GL_INVALID_VALUE error in
this case.
This fix adjusts the error check, and makes a small change to the
ctx->Driver.DrawBuffer() call below to ensure that, if n == 0,
Driver.DrawBuffer() is called with GL_NONE and that buffers[0] is
*not* referenced in this case (since we don't know whether it is valid).
Internal identifier: 365833
Brian Paul [Thu, 12 Feb 2009 17:14:36 +0000 (10:14 -0700)]
mesa: use new ST_CALLOC_STRUCT() macro in gallium state tracker
Brian Paul [Thu, 12 Feb 2009 17:08:25 +0000 (10:08 -0700)]
mesa: consistantly use mesa memory-functions in gallium state tracker
Use _mesa_malloc(), _mesa_free(), etc everywhere, not malloc(), free(), etc.
Still using CALLOC_STRUCT() at this point.
Brian Paul [Thu, 12 Feb 2009 16:21:50 +0000 (09:21 -0700)]
mesa: don't include m_xform.h where not needed
Brian Paul [Thu, 12 Feb 2009 16:17:18 +0000 (09:17 -0700)]
mesa: move _mesa_transform_vector() from m_xform.c to m_matrix.c
m_xform.c is omitted from gallium builds but _mesa_transform_vector() is
still needed.
Brian Paul [Thu, 12 Feb 2009 16:15:38 +0000 (09:15 -0700)]
mesa: remove unused functions in m_xform.[ch]
The functions are:
_mesa_project_points()
_mesa_transform_bounds3()
_mesa_transform_bounds2()
_mesa_transform_point_sz()
Brian Paul [Thu, 12 Feb 2009 15:58:12 +0000 (08:58 -0700)]
mesa: restore FLUSH_VERTICES() in _mesa_notifySwapBuffers()
Brian Paul [Thu, 12 Feb 2009 15:30:58 +0000 (08:30 -0700)]
mesa: remove empty, unneeded mathmod.h header
José Fonseca [Thu, 12 Feb 2009 15:49:55 +0000 (15:49 +0000)]
glut: Fix broken font symbols when gcc visibility attributes used.
Jakob Bornecrantz [Wed, 11 Feb 2009 15:53:53 +0000 (16:53 +0100)]
egl: Use only libdrm structs for kms
José Fonseca [Thu, 12 Feb 2009 13:54:20 +0000 (13:54 +0000)]
progs: Prevent clash with min macro.
José Fonseca [Thu, 12 Feb 2009 13:52:16 +0000 (13:52 +0000)]
glut: Don't try to guess system headers.
Just include them, especially such basic header as stdlib.h
Keith Whitwell [Thu, 12 Feb 2009 13:30:01 +0000 (13:30 +0000)]
tgsi: keep immediate file info uptodate
Make sure the stats for TGSI_FILE_IMMEDIATE are uptodate. Previously
we just had immediate_count, but file_*[TGSI_FILE_IMMEDIATE] were
bogus.
José Fonseca [Thu, 12 Feb 2009 12:59:58 +0000 (12:59 +0000)]
scons: Unbreak mingw builds.
José Fonseca [Thu, 12 Feb 2009 11:58:04 +0000 (11:58 +0000)]
scons: Build xlib state tracker regardless of DRI settings.
José Fonseca [Thu, 12 Feb 2009 11:57:45 +0000 (11:57 +0000)]
scons: Build DLLs/EXEs with more memory debugger friendlier settings.
Brian Paul [Thu, 12 Feb 2009 05:13:17 +0000 (22:13 -0700)]
softpipe: asst clean-ups, const correctness, comments
Brian Paul [Thu, 12 Feb 2009 05:03:34 +0000 (22:03 -0700)]
softpipe: remove unneeded #include
Brian Paul [Thu, 12 Feb 2009 05:01:43 +0000 (22:01 -0700)]
softpipe: asst comments, clean-ups
Brian Paul [Thu, 12 Feb 2009 04:51:00 +0000 (21:51 -0700)]
softpipe: remove some old polygon stipple stuff and do some clean-ups
Brian [Thu, 12 Feb 2009 04:44:38 +0000 (21:44 -0700)]
softpipe: updated comments
Brian [Thu, 12 Feb 2009 04:38:20 +0000 (21:38 -0700)]
softpipe: rename PRIM_x to QUAD_PRIM_x
Brian [Thu, 12 Feb 2009 04:33:59 +0000 (21:33 -0700)]
softpipe: rename sp_headers.h to sp_quad.h
This header describes the quad-related datatypes afterall.
Brian [Thu, 12 Feb 2009 04:30:21 +0000 (21:30 -0700)]
softpipe: rename single-include preprocessor symbol, add comments
Brian [Thu, 12 Feb 2009 04:25:10 +0000 (21:25 -0700)]
softpipe: rename sp_quad.[ch] -> sp_quad_pipe.[ch]
Be more consistant with 'draw' module.
Robert Ellison [Thu, 12 Feb 2009 01:01:34 +0000 (18:01 -0700)]
Fix an i965 assertion failure on glClear()
While running conform with render-to-texture:
conform -d 33 -v 2 -t -direct
the i965 driver failed this assertion:
intel_clear.c:77: intel_clear_tris: Assertion `(mask & ~((1 << BUFFER_BACK_LEFT) | (1 << BUFFER_FRONT_LEFT) | (1 << BUFFER_DEPTH) | (1 << BUFFER_STENCIL))) == 0' failed.
The problem is that intel_clear_tris() is called by intelClear() to
clear any and all of the available color buffers, but intel_clear_tris()
actually only handles the back left and front left color buffers; so
the assertion fails as soon as you try to clear a non-standard color
buffer.
The fix is to have intelClear() only call intel_clear_tris() with
buffers that intel_clear_tris() can support. intelClear() already backs
down to _swrast_Clear() for all buffers that aren't handled explicitly.