summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2011-01-31glsl: Change texel offsets to a single vector rvalue.Kenneth Graunke
Having these as actual integer values makes it difficult to implement the texture*Offset built-in functions, since the offset is actually a function parameter (which doesn't have a constant value). The original rationale was that some hardware needs these offset baked into the instruction opcode. However, at least i965 should be able to support non-constant offsets. Others should be able to rely on inlining and constant propagation.
2011-01-31glsl: Re-synchronize ir_variable_mode and the printer's string array.Kenneth Graunke
Since the introduction of ir_var_system_value, system variables would be printed as "temporary" and temporaries would result in out-of-bounds array access, showing up as garbage in printed IR.
2011-01-31scons: Gracefully handle pkg-config errors with libdrm_radeon.Vinson Lee
Print warnings and continue build.
2011-01-31Remove the talloc sources from the Mesa repository.Kenneth Graunke
2011-01-31Remove talloc from the SCons build system.Kenneth Graunke
2011-01-31Remove talloc from the make and automake build systems.Kenneth Graunke
2011-01-31ralloc: a new MIT-licensed recursive memory allocator.Kenneth Graunke
2011-01-31Convert everything from the talloc API to the ralloc API.Kenneth Graunke
2011-01-31ralloc: Add a fake implementation of ralloc based on talloc.Kenneth Graunke
2011-01-31glx: Properly check for a valid fd in dri2CreateScreen().Henri Verbeet
Signed-off-by: Brian Paul <brianp@vmware.com>
2011-01-31svga: Fix translation of TGSI SSG opcode.Michel Dänzer
SVGA3D only supports SGN for vertex shaders, and this requires two additional temporary registers for intermediate results. For fragment shaders, lower to two CMPs and one ADD.
2011-01-31svga: TEXLDL opcode dst/src register information is correct.Michel Dänzer
2011-01-31svga: Print the number and mnemonic of the opcode we're missing information for.Michel Dänzer
Makes it easier to figure out which opcode it's about.
2011-01-31glx: Fix leaks in DRISW screen creation error paths.Henri Verbeet
Signed-off-by: Brian Paul <brianp@vmware.com>
2011-01-31glx: Fix leaks in DRI screen creation error paths.Henri Verbeet
Signed-off-by: Brian Paul <brianp@vmware.com>
2011-01-31glx: Fix leaks in DRI2 screen creation error paths.Henri Verbeet
Signed-off-by: Brian Paul <brianp@vmware.com>
2011-01-31glx: fix length of GLXGetFBConfigsSGIXJulien Cristau
The extra length is the size of the request *minus* the size of the VendorPrivate header, not the addition. NOTE: This is a candidate for the 7.9 and 7.10 branches Signed-off-by: Julien Cristau <jcristau@debian.org> Signed-off-by: Brian Paul <brianp@vmware.com>
2011-01-31glx: fix GLXChangeDrawableAttributesSGIX requestJulien Cristau
xGLXChangeDrawableAttributesSGIXReq follows the GLXVendorPrivate header with a drawable, number of attributes, and list of (type, value) attribute pairs. Don't forget to put the number of attributes in there. I don't think this can ever have worked. NOTE: This is a candidate for the 7.9 and 7.10 branches Signed-off-by: Julien Cristau <jcristau@debian.org> Signed-off-by: Brian Paul <brianp@vmware.com>
2011-01-31r600g: fix eg OQ properly.Dave Airlie
the context init is separate for these gpus.
2011-01-31r600g: fix OQ on evergreenAlex Deucher
6xx/7xx have a max of 4 DBs, evergreen have a max of 8. Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
2011-01-31r600g: fix occlusion query results.Dave Airlie
Like on some r5xx, there are multiple DB backends on the r600, we need to add up the query results from each of these to get the final correct value. So far I'm not 100% sure how to calculate the num_db, value setting it to 4 should be harmless enough until we do. This fixes occulsion_query piglit test on my rv740. Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-30r600g: remove some non-existent evergreen reg fieldsAlex Deucher
Signed-off-by: Alex Deucher <alexdeucher@gmail.com>
2011-01-31r600g: fix regression in cubemap tests since ↵Dave Airlie
eea1d8199b376f37027c14669e0bdf991a22872d Although CUBE is a reduction inst, it writes to more than just PV.X so we need to keep the dst channel. Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-31r600g: handle the write all cbufs property.Dave Airlie
This only works on r600/r700 so far, evergreen doesn't appear to have the multiwrite enable bit in the color control, so we may have to actually do a shader rewrite on EG hardware. remove some duplicate code reg defines also. Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-01-30util: Call tables should be const.Henri Verbeet
2011-01-30r600g: Update the flushed depth texture after drawing to the corresponding ↵Henri Verbeet
texture. I know Jerome will probably rewrite the way depth textures work sometime soon. For the time being this should at least make common depth texture usage for shadowing work properly though.
2011-01-30st/vega: Disable blending when the paint is opaque.Chia-I Wu
When the paint is opaque (currently, solid color with alpha 1.0f), no blending is needed for VG_BLEND_SRC_OVER. This eliminates the serious performance hit introduced by 859106f196ade77f59f8787b071739901cd1a843 for a common scenario.
2011-01-30st/vega: Remove an invalid sanity check.Chia-I Wu
Before create_handle returns, obj->handle is 0. Calling handle_to_object will fail.
2011-01-30st/vega: s/vg[A-Z]/vega[A-Z]/.Chia-I Wu
2011-01-30scons: We have C++ in several libraries, so always link with the C++ compilerJosé Fonseca
Prevents missing symbols in libGL.so when LLVM is disabled.
2011-01-30r600g: Fix void pointer arithmetic.Vinson Lee
Fixes SCons build.
2011-01-30r600g: fixes a segfault in the piglit fbo-genmipmap-formats test.Dave Airlie
should be no need to unset this ptr here and if we don't end up using the blitter we've just broken the state.
2011-01-29r300/compiler: Standardize the number of bits used by swizzle fieldsTom Stellard
Swizzles are now defined everywhere as a field with 12 bits that contains 4 channels worth of meaningful information. Any channel that is unused is set to RC_SWIZZLE_UNUSED. This change is necessary because rgb instructions and alpha instructions were initializing channels that would never be used (channel 3 for rgb and channels 1-3 for alpha) with 0 (aka RC_SWIZZLE_X). This made it impossible to use generic helper functions for swizzles, because sometimes a channel value of 0 meant unused and other times it meant RC_SWIZZLE_X. All hacks that tried to guess how many channels were relevant have also been removed.
2011-01-30r300g: upload translated indices via the uploaderMarek Olšák
2011-01-30r300g: rework vertex format fallbackMarek Olšák
1) Only translate the [min_index, max_index] range. 2) Upload translated vertices via the uploader. 3) Rename valid_vertex_buffer[] to real_vertex_buffer[]
2011-01-30r600g: upload translated indices via the uploaderMarek Olšák
2011-01-30r600g: rework vertex format fallbackMarek Olšák
1) Only translate the [min_index, max_index] range. 2) Upload translated vertices via the uploader.
2011-01-30r600g: fix vertex format fallbackMarek Olšák
This fixes: - piglit/draw-vertices - piglit/draw-vertices-half-float
2011-01-30r600g: rework vertex buffer uploadsMarek Olšák
Only upload the [min_index, max_index] range instead of [0, userbuf_size]. This an important optimization. Framerate in Lightsmark: Before: 22 fps After: 75 fps The same optimization is already in r300g.
2011-01-30r600g: consolidate set_constant_buffer functionsMarek Olšák
2011-01-30r600g: consolidate vertex_buffer_update functionsMarek Olšák
2011-01-30r600g: consolidate draw_vbo functions (v2)Marek Olšák
Added a conditional to spi_update per Dave's comment.
2011-01-30r600g: make r600_drawl inherit pipe_draw_infoMarek Olšák
2011-01-30r600g: add back u_upload_mgr integrationMarek Olšák
I can't see a performance difference with this code, which means all the driver-specific code removed in this commit was unnecessary. Now we use u_upload_mgr in a slightly different way than we did before it got dropped. I am not restoring the original code "as is" due to latest u_upload_mgr changes that r300g performance benefits from. This also fixes: - piglit/fp-kil
2011-01-30nvc0: implement transform feedback stateChristoph Bumiller
2011-01-29nvc0: enable PIPE_CAP_ARRAY_TEXTURES and fix themChristoph Bumiller
2011-01-30egl_dri2: Export glapi symbols for DRI drivers.Chia-I Wu
When an app loads libEGL.so dynamically with RTLD_LOCAL, loading DRI drivers would fail because of missing glapi symbols. This commit makes egl_dri2 load libglapi.so with RTLD_GLOBAL to export glapi symbols for future symbol resolutions. The same trick can be found in GLX. However, egl_dri2 can only do so when --enable-shared-glapi is given. Because, otherwise, both libGL.so and libglapi.so define glapi symbols and egl_dri2 cannot tell which library to load.
2011-01-30egl: Make the transition to built-in drivers more smooth.Chia-I Wu
When the user sets EGL_DRIVER to egl_dri2 (or egl_glx), make sure the built-in driver is used. The user might leave the outdated egl_dri2.so (or egl_glx.so) on the filesystem and we do not want to load it.
2011-01-29mapi: Workaround a bug in makedepend.Chia-I Wu
makedepend would crash when a source includes a header indirectly, such as #define HEADER "some-header.h" #include HEADER Do not define HEADER (makedepend would detects this as an incomplete include) and add the dependency manually in the Makefile. This should hopefully fix bug #33374.
2011-01-29u_blitter: use user buffers instead of real buffersMarek Olšák
User buffers may be the fastest way to upload data.