Age | Commit message (Collapse) | Author | |
---|---|---|---|
2006-06-08 | mark ARL operand as scalar (Roland Scheidegger) | Tilman Sauerbeck | |
2006-06-08 | struct op_names: get rid of the 'output operands' field as it's not used ↵ | Tilman Sauerbeck | |
(Roland Scheidegger) | |||
2006-06-08 | Improve slightly wrong CMP_SRCS test to avoid unencessary instructions. ↵ | Roland Scheidegger | |
Clean up the r200 vertex program code a bit. | |||
2006-06-06 | prevent possible program mix up with sw vps (Roland Scheidegger) | Aapo Tahkola | |
2006-06-06 | prevent run_arb_vertex_program from running tnl programs unless ↵ | Aapo Tahkola | |
ctx->_MaintainTnlProgram is set | |||
2006-06-06 | make sure tex format is supported. | Aapo Tahkola | |
2006-06-06 | various fixes (Roland Scheidegger) | Aapo Tahkola | |
2006-06-03 | Added simple lookup_bufferobj() function to wrap the _mesa_HashLookup() call | Brian Paul | |
and cast. Simplifies the code a little bit elsewhere. | |||
2006-06-03 | updates for using libdrm (Hanno Bock) | Brian Paul | |
2006-06-02 | minor clean-ups | Brian Paul | |
2006-06-02 | implement arb_vertex_program in hw for r200. Code contains still some hacks, ↵ | Roland Scheidegger | |
generic attribs cause a fallback, but otherwise it seems to work quite well. Passes all glean vertProg1 tests with the exception of the degnerated LIT case (which is a hw limitation), as well as runs the r200 render path of doom3/quake4 (1.1 patch needed for quake4). The code is heavily borrowed from the r300 driver as vertex programs encoding is almost identical. arb_vertex_program is not yet announced by default and still needs to be enabled via driconf. | |||
2006-06-02 | set per-context GL_DRAW_BUFFER state regardless of currently bound FBO | Brian Paul | |
2006-06-02 | fix depth-component test to allow rectangular textures | Brian Paul | |
2006-06-02 | remove code dealing with drmMinor version < 6 since all radeon drivers ↵ | Roland Scheidegger | |
request at least drmMinor 6 anyway. | |||
2006-06-01 | retry on EBUSY instead of EAGAIN in radeonWaitIrq (it appears this was the ↵ | Roland Scheidegger | |
intention as drm seems to never return EAGAIN) in all radeon drivers. | |||
2006-06-01 | Fix extended swizzling in vertex programs by introducing special swizzle ↵ | Roland Scheidegger | |
instruction, extend the 2 bit rsw field to 3 bit like used in other places. While here, also fix up rsw (negation), dph and try to fix up rsq with negative values (doesn't work, bug seems elsewhere) in the sse codegen code. | |||
2006-05-31 | SWZ is nearly identical to MOV. the difference is that SWZ supports ↵ | Tilman Sauerbeck | |
per-component negation, which MOV doesn't. as that's handled transparently, we can use the same code for them. | |||
2006-05-31 | pass src->NegateBase as-is in t_src() now, as the flags are equivalent to ↵ | Tilman Sauerbeck | |
r300's VSF_FLAGs now | |||
2006-05-31 | Added DirectFB GLUT implementation. | Claudio Ciccani | |
2006-05-31 | Fixed bug with multiple contexts. | Claudio Ciccani | |
Added support for ARGB4444 and ARGB2554. Added hardware accelerated Clear(). | |||
2006-05-30 | Define new NEGATE_* tokens for setting the src register's NegateBase field. | Brian Paul | |
Before, we were using GL_TRUE/GL_FALSE in a few places. Note: only the SWZ instruction can negate individual components. | |||
2006-05-30 | remove redundant 'const' | Brian Paul | |
2006-05-30 | free convolved image / fix mem leak (bug 7077) | Brian Paul | |
2006-05-30 | minor code cleanup, no actual changes | Tilman Sauerbeck | |
2006-05-30 | don't generate GL_INVALID_ENUM if GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB, ↵ | Tilman Sauerbeck | |
which is actually handled | |||
2006-05-30 | Remove not needed forward declaration. | Michal Krol | |
2006-05-29 | don't do the check for >= MAX_VERTEX_PROGRAM_ATTRIBS twice, it's already ↵ | Tilman Sauerbeck | |
done in parse_generic_attrib_num() | |||
2006-05-29 | generic attribute 0 wasn't handled correctly (Jesse Allen) | Brian Paul | |
2006-05-27 | fix some whitespace in previous commit | Roland Scheidegger | |
2006-05-27 | preparation for r200 hw vertex programs. Increase R200_CMD_BUF_SZ to 16k ↵ | Roland Scheidegger | |
instead of 8k (the reasons why it was exactly 8k are unclear). Add register defines, sanity code, fix potential issue with wrong statechange order when disabling fragment programs. | |||
2006-05-24 | The ctx->_TexEnvProgram that's allocated in update_state() doesn't seem to | Brian Paul | |
be needed (it's set in texenvprogram.c) and is a memory leak according to valgrind. Disable with #if 0 / #endif | |||
2006-05-24 | Roll _mesa_free_parameters() into calling function (the only caller). | Brian Paul | |
Replace assert(0) with _mesa_warning(). Use new _mesa_align_realloc() function. New comments, clean-ups. | |||
2006-05-24 | fix a small memory leak in _mesa_add_state_reference() | Brian Paul | |
2006-05-24 | added _mesa_align_realloc() | Brian Paul | |
2006-05-24 | use GLbitfield for StateFlags, improved comments | Brian Paul | |
2006-05-23 | Add const qualifiers in a number of places. | Brian Paul | |
2006-05-23 | init mask=0 to silence warning | Brian Paul | |
2006-05-23 | make mgaSetFence() and mgaWaitFence() static | Brian Paul | |
2006-05-23 | prototype i915_udpate_fog() to silence warning | Brian Paul | |
2006-05-23 | In cache_item(), there was sizeof(void) expression. | Brian Paul | |
Replace void *key parameter with const struct state_key * | |||
2006-05-22 | GL_INDEX_OFFSET didn't work with CI->RGB mappings. | Brian Paul | |
Added some const qualifiers. | |||
2006-05-22 | memory usage fixes for glean/conform, use a better hash function | Keith Whitwell | |
2006-05-22 | Fix typo | Keith Whitwell | |
2006-05-22 | Better caching for texenv programs. | Keith Whitwell | |
Initialize some values correctly. | |||
2006-05-22 | Speedup the venerable mm.[ch] allocator with doubly linked lists and a | Keith Whitwell | |
separate list of free segments. | |||
2006-05-20 | In gl_texture_image, replace ImageStride with an ImageOffsets array. | Brian Paul | |
Some hardware lays out 3D mipmaps in a manner that can't be expressed with a simple image stride. The ImageOffsets array is allocated and initialized to typical defaults in the _mesa_init_teximage_fields() function. If needed, a driver will then have to replace these offsets. TexStore and TexelFetch routines updated to use offsets array. | |||
2006-05-20 | remove some redundant functions, fix-up reference counting | Brian Paul | |
2006-05-20 | use new _mesa_remove_renderbuffer() | Brian Paul | |
2006-05-20 | use new _mesa_dereference_frame/renderbuffer() functions | Brian Paul | |
2006-05-20 | Added _mesa_remove_renderbuffer() as a counterpart to _mesa_add_renderbuffer(). | Brian Paul | |
Added _mesa_dereference_renderbuffer() to encapsulate renderbuffer reference count decrement, delete, locking. |