fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-02-01 06:50:22 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

gallivm: work-around trilinear mipmap filtering regression with LLVM 2.8

Browse source 
The bug only happens on the AOS / fixed-pt path.
This commit is contained in:
Brian Paul 2010-10-13 12:35:38 -06:00
parent bee22ed6b9
commit e487b665aa
1 changed files with 20 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

20
src/gallium/auxiliary/gallivm/lp_bld_sample_aos.c
View file

@ -873,6 +873,26 @@ lp_build_sample_mipmap(struct lp_build_sample_context *bld,
lod_fpart = LLVMBuildTrunc(builder, lod_fpart, h16_bld.elem_type, "");
lod_fpart = lp_build_broadcast_scalar(&h16_bld, lod_fpart);
#if HAVE_LLVM == 0x208
/* This is a work-around for a bug in LLVM 2.8.
* Evidently, something goes wrong in the construction of the
* lod_fpart short[8] vector. Adding this no-effect shuffle seems
* to force the vector to be properly constructed.
* Tested with mesa-demos/src/tests/mipmap_limits.c (press t, f).
*/
{
LLVMValueRef shuffles[8], shuffle;
int i;
assert(h16_bld.type.length <= Elements(shuffles));
for (i = 0; i < h16_bld.type.length; i++)
shuffles[i] = lp_build_const_int32(2 * (i & 1));
shuffle = LLVMConstVector(shuffles, h16_bld.type.length);
lod_fpart = LLVMBuildShuffleVector(builder,
lod_fpart, lod_fpart,
shuffle, "");
}
#endif
colors0_lo = lp_build_lerp(&h16_bld, lod_fpart,
colors0_lo, colors1_lo);
colors0_hi = lp_build_lerp(&h16_bld, lod_fpart,