| Age | Commit message (Collapse) | Author |
|---|
|
Builds on commit ddb0e18f6c5582d4d2cc59ffd16ad9c4639ed059 and fixes
regressions in glean clipFlat test.
We assume that Gallium drivers observe flatshade_first for all triangles
and that all the assorted per-triangle calls in the 'draw' module also
follow flatshade_first. Everything else builds on those rules.
Gallium does not use follow flatshade_first for GL quads, quad strips
and polygons; the "last" vertex is always the provoking vertex for those
prims. So now there are separate QUAD_FIRST_PV and QUAD_LAST_PV macros
in the draw primitive decomposition code instead of one QUAD macro.
|
|
Looks like coverity found a false positive for an out of bounds array write.
|
|
|
|
|
|
|
|
|
|
|
|
When the cube faces were stored in a compressed format, the img_stride
values were wrong and didn't match the per-face size computed in the
tex_image_face_size() function. This caused bad rendering or segfaults.
|
|
|
|
Start propogating NULL pointers from allocation functions and checks for
NULL in the callers...
|
|
|
|
|
|
|
|
This is a better place than in u_rect.c
|
|
|
|
|
|
|
|
|
|
This gives a ~30% shader optimization time improvement on blender.
Tested by comparing the dumped LLVM modules.
Current ordering:
time ~/llvm-git/obj/Release-Asserts/bin/opt l.bc -constprop -instcombine
-mem2reg -gvn -simplifycfg
real 0m1.126s
user 0m1.108s
sys 0m0.012s
With this patch:
time ~/llvm-git/obj/Release-Asserts/bin/opt l.bc -mem2reg -constprop -instcombine -gvn -simplifycfg
real 0m0.885s
user 0m0.880s
sys 0m0.000s
The overall improvement in blender is ~15%.
Blender without the patch takes 1m13s:
edwin 5934 87.6 11.5 729440 458296 pts/5 SLl+ 17:35 1:13 blender
Blender with the patch takes 1m3s:
edwin 5726 94.2 11.2 716424 446168 pts/5 SLl+ 17:32 1:03 blender
It is still slow with the patch, but better (most of the optimization time is
taken up by GVN, see LLVM PR7023).
Signed-off-by: Török Edwin <edwintorok@gmail.com>
Signed-off-by: José Fonseca <jfonseca@vmware.com>
|
|
|
|
It was only used for D3D's REP/END/BGNFOR/ENDFOR. D3D's aL register is
just like another address register now.
|
|
Current code only invalidated if the texture was different, however we
store swizzled values in the cache, so we need to invalidate in that case
also.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
|
|
|
|
|
|
Signed-off-by: Xavier Chantry <chantry.xavier@gmail.com>
Signed-off-by: José Fonseca <jfonseca@vmware.com>
|
|
All affected operations have already been moved to context.
More cleanup work can be done, in particular with the buffer transfers.
|
|
|
|
|
|
Don't use the dst alpha channel from the tile cache when it does not
exist in the true format.
|
|
Don't use the dst alpha channel from the swizzled tile when it does not
exist in the true format.
|
|
Fixes svga_screen_init_surface_functions implicit declaration warning.
|
|
|
|
No accelerated blitting for these, it's too messy.
|
|
|
|
This further reduces the number of DRM calls and flushes.
|
|
This reverts commit 3e0f1719d411b34f77fd2a053ecf83b43bb836ca.
|
|
These two should be tied together because what's set in VAP or stuffed in GA
should be rasterized in RS. Not doing so causes a hardlock.
The reason for the merge is that if stuffed texture coordinates (e.g. point
sprite texgen) happen to occupy the texcoord slot dedicated to fog or wpos,
the two must be relocated to other free slots, which needs remapping the vertex
shader outputs.
The rasterizer code is now literally a sequence read-rasterize-write.
|
|
|
|
Resetting tiling flags might happen after a texture is destroyed...
Looking at the kernel sources, we don't actually need to reset the tiling
flags.
|
|
|
|
|
|
Removed the param heaps, haven't been using them for a long
time now.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Ouch.
|
