Mesa (master): freedreno/a3xx/compiler: fix CMP
Rob Clark
robclark at kemper.freedesktop.org
Sat Aug 24 18:02:48 UTC 2013
Module: Mesa
Branch: master
Commit: 8b250bb8aa31b781681491b1d2a8baa15a024c1c
URL: http://cgit.freedesktop.org/mesa/mesa/commit/?id=8b250bb8aa31b781681491b1d2a8baa15a024c1c
Author: Rob Clark <robclark at freedesktop.org>
Date: Tue Aug 20 13:46:30 2013 -0400
freedreno/a3xx/compiler: fix CMP
The 1st src to add.s needs (r) flag (repeat), otherwise it will end up:
add.s dst.xyzw, tmp.xxxx -1
instead of:
add.s dst.xyzw, tmp.xyzw, -1
Also, if we are using a temporary dst to avoid clobbering one of the src
registers, we actually need to use that as the dst for the sel
instruction.
Signed-off-by: Rob Clark <robclark at freedesktop.org>
---
src/gallium/drivers/freedreno/a3xx/fd3_compiler.c | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/gallium/drivers/freedreno/a3xx/fd3_compiler.c b/src/gallium/drivers/freedreno/a3xx/fd3_compiler.c
index eabe21c..07bede4 100644
--- a/src/gallium/drivers/freedreno/a3xx/fd3_compiler.c
+++ b/src/gallium/drivers/freedreno/a3xx/fd3_compiler.c
@@ -790,13 +790,13 @@ trans_cmp(const struct instr_translater *t,
instr = ir3_instr_create(ctx->ir, 2, OPC_ADD_S);
instr->repeat = 3;
add_dst_reg(ctx, instr, &tmp_dst, 0);
- add_src_reg(ctx, instr, &tmp_src, 0);
+ add_src_reg(ctx, instr, &tmp_src, 0)->flags |= IR3_REG_R;
ir3_reg_create(instr, 0, IR3_REG_IMMED)->iim_val = -1;
/* sel.{f32,f16} dst, src2, tmp, src1 */
instr = ir3_instr_create(ctx->ir, 3, ctx->so->half_precision ?
OPC_SEL_F16 : OPC_SEL_F32);
- vectorize(ctx, instr, &inst->Dst[0].Register, 3,
+ vectorize(ctx, instr, dst, 3,
&inst->Src[2].Register, 0,
&tmp_src, 0,
&inst->Src[1].Register, 0);
More information about the mesa-commit
mailing list