summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* nvfx: fix fragment program constant updatesLuca Barbieri2010-04-181-1/+4
|
* r300g: do not tell st that a buffer is referenced by hwMarek Olšák2010-04-187-15/+33
| | | | It saves a few libdrm calls and unnecessary flushes.
* r300g: save vertex buffers around u_blitter callsLuca Barbieri2010-04-181-0/+1
| | | | | | Not strictly needed by the Mesa state tracker, but needed in general. Only compile tested.
* u_blitter: add support for saving vertex buffersLuca Barbieri2010-04-182-0/+24
| | | | | | | | | | | | Currently r300g does not save vertex buffer on blitter calls. It gets away with it because the current Mesa state tracker usually resets vertex buffers on every draw calls. However, this is wrong. nvfx won't be lucky because it needs to use the blitter inside draw calls.
* p_compiler: add likely/unlikely macrosLuca Barbieri2010-04-181-1/+39
| | | | | | Defined like the Linux kernel. A explanatory comment is included.
* r300g: fixup 0 stride buffer properly.Dave Airlie2010-04-181-8/+7
| | | | | | | Just set the max index to 1, this lets doom3 run and seems correct, though it would be better to just emit a constant like SVGA does. Signed-off-by: Dave Airlie <[email protected]>
* llvmpipe: Emit only the vertex attributes necessary for the FS, and ensure ↵José Fonseca2010-04-181-68/+67
| | | | | | | the first one is always position. With this we correctly handle vertex shaders whose output position is not in index zero.
* softpipe: Preserve/check the signal of the number of scanlines in an edge.José Fonseca2010-04-181-1/+2
| | | | It can become negative in some weird triangles.
* draw: Cosmetic cleanups and comments.José Fonseca2010-04-184-4/+10
|
* retrace: Fix vertex/index/constant buffer dumping.José Fonseca2010-04-181-3/+4
|
* trace: Handle transfers with zero strides correctly.José Fonseca2010-04-181-2/+10
|
* st/python: Set the bind flag for the default texture.José Fonseca2010-04-181-0/+1
|
* retrace: Remove deprecated code.José Fonseca2010-04-181-10/+0
|
* mesa: Restore comment too.José Fonseca2010-04-182-2/+2
|
* glsl: Fix handling of OPCODE_PRINT for no registers case.Vinson Lee2010-04-182-2/+2
| | | | | | | | | A register file value is unsigned so could never be -1. A value of 0 also aliased to PROGRAM_TEMPORARY. If an OPCODE_PRINT has no registers to print, set the register file value to PROGRAM_UNDEFINED and check for that value when executing this instruction.
* i965: Assert that an OPCODE_IF was seen before an OPCODE_ELSE.Vinson Lee2010-04-171-0/+1
|
* st/dri: Check for NULL before dereference and assignment.Vinson Lee2010-04-171-2/+10
|
* nvfx: Initialize variables.Vinson Lee2010-04-172-4/+4
|
* r300g: Cast rbuf->user_buffer to 'uint8_t *' before arithmetic.Vinson Lee2010-04-171-1/+1
| | | | | Fixes this SCons build error. "pointer of type 'void *' uses in arithmetic"
* nvfx: Remove unused variables.Vinson Lee2010-04-172-3/+0
|
* nv50: Initialize variables.Vinson Lee2010-04-172-6/+6
| | | | Silences GCC uninitialized variable warnings.
* nv50: Remove unused variable.Vinson Lee2010-04-171-2/+0
|
* i915g: Remove broken power of two functionJakob Bornecrantz2010-04-181-16/+8
|
* i915g: Add some helper functionsJakob Bornecrantz2010-04-181-7/+19
| | | | Mostly to make the lines not so long.
* i915g: Fix cubemap layoutsJakob Bornecrantz2010-04-181-101/+121
|
* i915g: Clean and add some debug printingJakob Bornecrantz2010-04-181-8/+18
| | | | Disabled by default
* gallium/tests: Make trivial apps depend on libsJakob Bornecrantz2010-04-181-1/+1
|
* r300g: various other cleanupsMarek Olšák2010-04-1812-86/+71
| | | | | | It seems to me that the resource management code is the most blurry. Also some of my notes about fastfill are wrong, removing them too.
* r300g: inline screen_buffer functionsMarek Olšák2010-04-181-84/+36
|
* r300g: remove unused declarationsMarek Olšák2010-04-181-4/+0
|
* r300g: add debugging options "notiling" and "noimmd" (for testing)Marek Olšák2010-04-184-9/+18
| | | | | notiling = Disable texture tiling noimmd = Disable immediate mode (this optimization was really worth it!)
* r300g: remove r300_transfer::ctxMarek Olšák2010-04-181-5/+1
|
* r300g: remove use of c++ keywordsMarek Olšák2010-04-184-15/+12
|
* r300g: always create microtiled zbuffer regardless of texture dimensionsMarek Olšák2010-04-181-2/+3
|
* i915g: Remove unnecessary headers.Vinson Lee2010-04-171-3/+0
|
* nvfx: Remove unnecessary headers.Vinson Lee2010-04-174-6/+0
|
* nv50: Remove redundant assignment.Vinson Lee2010-04-171-1/+1
|
* i965g: Remove redundant assert.Vinson Lee2010-04-171-1/+0
|
* os: Implement pipe_barrier for POSIX platforms without pthread_barrier_t.Vinson Lee2010-04-171-5/+26
| | | | This patch was tested on Mac OS X.
* i915g: Add a software only debuging winsysJakob Bornecrantz2010-04-178-0/+480
|
* gallium/tests: Fix trivial examplesJakob Bornecrantz2010-04-173-25/+29
|
* gallium/util: IgnoresJakob Bornecrantz2010-04-171-0/+1
|
* progs/tests: IgnoresJakob Bornecrantz2010-04-171-3/+14
|
* r300/compiler: Remove unnecessary header.Vinson Lee2010-04-171-2/+0
|
* draw llvm: Remove unnecessary header.Vinson Lee2010-04-161-1/+0
|
* llvmpipe: Remove unnecessary header.Vinson Lee2010-04-161-1/+0
|
* r300/compiler: Add radeon_program_tex.c to SCons build.Vinson Lee2010-04-161-0/+1
| | | | This was missed in commit f8a14186809356871ae74159c774e9e3959a22e5.
* r300/compiler: add emulation of all mirrored-clamp wrap modes for NPOT texturesMarek Olšák2010-04-173-3/+24
|
* r300/compiler: replace mirrored-repeat emulation with a faster versionMarek Olšák2010-04-171-62/+65
| | | | | Also, the Negate bitfield was 1 instead of RC_MASK_XYZ in the previous version, causing incorrect rendering.
* r300/compiler: add handy definitions for XYZ0 and smeared half swizzlingMarek Olšák2010-04-172-3/+5
|