[Mesa-dev] [PATCH] intel/compiler: Set swizzle to BRW_SWIZZLE_XXXX for scalar region

[Sagar Ghuge]

When RepCtrl is set, the swizzle field is ignored by the hardware. In
order to ensure a 1-to-1 correspondence between the human-readable
disassembly and the binary instruction encoding always set the swizzle
to XXXX (all zeros) when it is unused due to RepCtrl

Signed-off-by: Sagar Ghuge <***@intel.com>
Reviewed-by: Matt Turner <***@gmail.com>
---
src/intel/compiler/brw_eu_emit.c | 19 ++++++++++++++++++-
1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/src/intel/compiler/brw_eu_emit.c b/src/intel/compiler/brw_eu_emit.c
index 4630b83b1a..483037345e 100644
--- a/src/intel/compiler/brw_eu_emit.c
+++ b/src/intel/compiler/brw_eu_emit.c
@@ -833,7 +833,15 @@ brw_inst *brw_##OP(struct brw_codegen *p, \
struct brw_reg src0, \
struct brw_reg src1, \
struct brw_reg src2) \
-{ \
+{ \
+ if (p->current->access_mode == BRW_ALIGN_16) { \
+ if (src0.vstride == BRW_VERTICAL_STRIDE_0) \
+ src0.swizzle = BRW_SWIZZLE_XXXX; \
+ if (src1.vstride == BRW_VERTICAL_STRIDE_0) \
+ src1.swizzle = BRW_SWIZZLE_XXXX; \
+ if (src2.vstride == BRW_VERTICAL_STRIDE_0) \
+ src2.swizzle = BRW_SWIZZLE_XXXX; \
+ } \
return brw_alu3(p, BRW_OPCODE_##OP, dest, src0, src1, src2); \
}

@@ -854,6 +862,15 @@ brw_inst *brw_##OP(struct brw_codegen *p, \
assert(src0.type == BRW_REGISTER_TYPE_DF); \
assert(src1.type == BRW_REGISTER_TYPE_DF); \
assert(src2.type == BRW_REGISTER_TYPE_DF); \
+ } \
+ \
+ if (p->current->access_mode == BRW_ALIGN_16) { \
+ if (src0.vstride == BRW_VERTICAL_STRIDE_0) \
+ src0.swizzle = BRW_SWIZZLE_XXXX; \
+ if (src1.vstride == BRW_VERTICAL_STRIDE_0) \
+ src1.swizzle = BRW_SWIZZLE_XXXX; \
+ if (src2.vstride == BRW_VERTICAL_STRIDE_0) \
+ src2.swizzle = BRW_SWIZZLE_XXXX; \
} \
return brw_alu3(p, BRW_OPCODE_##OP, dest, src0, src1, src2); \
}

[Matt Turner]

Post by Sagar Ghuge
When RepCtrl is set, the swizzle field is ignored by the hardware. In
order to ensure a 1-to-1 correspondence between the human-readable
disassembly and the binary instruction encoding always set the swizzle
to XXXX (all zeros) when it is unused due to RepCtrl

... and committed!