These align1 opcodes do partial writes of 64-bit data. The problem is that we
want to use them to write on the same register to implement packDouble2x32 and
from the point of view of DCE, since both opcodes write to the same register,
only the last one stands and decides to eliminate the first, which is
not correct, so prevent this from happening.
v2: Make a helper in vec4_instruction to know if the instruction is an
align1 partial write. This will come in handy when we implement a
simd splitting pass in a later patch.
---
src/mesa/drivers/dri/i965/brw_ir_vec4.h | 6 ++++++
src/mesa/drivers/dri/i965/brw_vec4_dead_code_eliminate.cpp | 3 ++-
2 files changed, 8 insertions(+), 1 deletion(-)
diff --git a/src/mesa/drivers/dri/i965/brw_ir_vec4.h
b/src/mesa/drivers/dri/i965/brw_ir_vec4.h
index a8e5f4a..7451f44 100644
--- a/src/mesa/drivers/dri/i965/brw_ir_vec4.h
+++ b/src/mesa/drivers/dri/i965/brw_ir_vec4.h
@@ -232,6 +232,12 @@ public:
bool can_change_types() const;
bool has_source_and_destination_hazard() const;
+ bool is_align1_partial_write()
+ {
+ return opcode == VEC4_OPCODE_SET_LOW_32BIT ||
+ opcode == VEC4_OPCODE_SET_HIGH_32BIT;
+ }
+
bool reads_flag()
{
return predicate || opcode == VS_OPCODE_UNPACK_FLAGS_SIMD4X2;
diff --git a/src/mesa/drivers/dri/i965/brw_vec4_dead_code_eliminate.cpp
b/src/mesa/drivers/dri/i965/brw_vec4_dead_code_eliminate.cpp
index 50706a9..950c6c8 100644
--- a/src/mesa/drivers/dri/i965/brw_vec4_dead_code_eliminate.cpp
+++ b/src/mesa/drivers/dri/i965/brw_vec4_dead_code_eliminate.cpp
@@ -109,7 +109,8 @@ vec4_visitor::dead_code_eliminate()
}
}
- if (inst->dst.file == VGRF && !inst->predicate) {
+ if (inst->dst.file == VGRF && !inst->predicate &&
+ !inst->is_align1_partial_write()) {
for (unsigned i = 0; i < regs_written(inst); i++) {
for (int c = 0; c < 4; c++) {
if (inst->dst.writemask & (1 << c)) {
--
2.7.4
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
