Files
clang-p2996/llvm/test/CodeGen/AMDGPU/GlobalISel/inst-select-amdgpu-wave-address.mir
Matt Arsenault 7f26a1027f AMDGPU/GlobalISel: Introduce pseudo to copy sp in call sequences
Arbitrary stack pointers are accessed using MUBUF instructions with
the voffset field, which is interpreted as the swizzled address. We
want to fold fold into the MUBUF form to use the SP in the SGPR
offset, and previously we were special casing the interpretation of
the pointer value if the access memory operand said it was relative to
the stack pointer.

690f5b7a01 removed this check, and moved
the DAG path to special casing copies from SGPRs. This is not an
entirely sound approach, since it's still changing the interpretation
of pointer values based the context.

Introduce a new pseudo which corresponds to the wave-to-vector address
transform. This way the memory instruction has consistent semantics
where the incoming pointer is always interpreted as a vector address,
and we're not obligated to optimize into the MUBUF offset-only
addressing mode. The DAG should probably have an equivalent pseudo.

This should fix some correctness issues, and folding this into
addressing modes will be a future optimization patch.
2022-01-19 10:13:31 -05:00

42 lines
1.6 KiB
YAML

# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
# RUN: llc -march=amdgcn -mcpu=gfx1031 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck -check-prefix=WAVE32 %s
# RUN: llc -march=amdgcn -mcpu=gfx1031 -mattr=+wavefrontsize64 -run-pass=instruction-select -verify-machineinstrs %s -o - | FileCheck -check-prefix=WAVE64 %s
---
name: wave_address_s
legalized: true
regBankSelected: true
tracksRegLiveness: true
machineFunctionInfo:
stackPtrOffsetReg: $sgpr32
body: |
bb.0:
; WAVE32-LABEL: name: wave_address_s
; WAVE32: [[S_LSHR_B32_:%[0-9]+]]:sreg_32 = S_LSHR_B32 $sgpr32, 5, implicit-def $scc
; WAVE32-NEXT: S_ENDPGM 0, implicit [[S_LSHR_B32_]]
; WAVE64-LABEL: name: wave_address_s
; WAVE64: [[S_LSHR_B32_:%[0-9]+]]:sreg_32 = S_LSHR_B32 $sgpr32, 6, implicit-def $scc
; WAVE64-NEXT: S_ENDPGM 0, implicit [[S_LSHR_B32_]]
%0:sgpr(p5) = G_AMDGPU_WAVE_ADDRESS $sgpr32
S_ENDPGM 0, implicit %0
...
---
name: wave_address_v
legalized: true
regBankSelected: true
tracksRegLiveness: true
machineFunctionInfo:
stackPtrOffsetReg: $sgpr32
body: |
bb.0:
; WAVE32-LABEL: name: wave_address_v
; WAVE32: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 5, $sgpr32, implicit $exec
; WAVE32-NEXT: S_ENDPGM 0, implicit [[V_LSHRREV_B32_e64_]]
; WAVE64-LABEL: name: wave_address_v
; WAVE64: [[V_LSHRREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHRREV_B32_e64 6, $sgpr32, implicit $exec
; WAVE64-NEXT: S_ENDPGM 0, implicit [[V_LSHRREV_B32_e64_]]
%0:vgpr(p5) = G_AMDGPU_WAVE_ADDRESS $sgpr32
S_ENDPGM 0, implicit %0
...