fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-05-05 09:38:07 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

llvmpipe: fix arguments order given to vec_andc

Browse source 
This patch fixes a classic "confuse the enemy" bug.

_mm_andnot_si128 (SSE) and vec_andc (VMX) do the same operation, but the
arguments are opposite.

_mm_andnot_si128 performs "r = (~a) & b" while
vec_andc performs "r = a & (~b)"

To make sure this error won't return in another place, I added a wrapper
function, vec_andnot_si128, in u_pwr8.h, which makes the swap inside.

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
This commit is contained in:
Oded Gabbay 2016-01-17 14:25:32 +02:00
parent 02ac91d717
commit 529aa8249a
2 changed files with 7 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

6
src/gallium/auxiliary/util/u_pwr8.h
View file

@ -153,6 +153,12 @@ vec_mullo_epi32 (__m128i a, __m128i b)
return v;
}
static inline __m128i
vec_andnot_si128 (__m128i a, __m128i b)
{
return vec_andc (b, a);
}
static inline void
transpose4_epi32(const __m128i * restrict a,
const __m128i * restrict b,

2
src/gallium/drivers/llvmpipe/lp_setup_tri.c
View file

@ -556,7 +556,7 @@ do_triangle_ccw(struct lp_setup_context *setup,
/* Calculate trivial reject values:
*/
eo = vec_sub_epi32(vec_andc(dcdy_neg_mask, dcdy),
eo = vec_sub_epi32(vec_andnot_si128(dcdy_neg_mask, dcdy),
vec_and(dcdx_neg_mask, dcdx));
/* ei = _mm_sub_epi32(_mm_sub_epi32(dcdy, dcdx), eo); */