Files
TheRock/compiler/amd-llvm/llvm/test/CodeGen/AMDGPU/eliminate-frame-index-scalar-scc-clobber.mir
T
derek.holloway 7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

139 lines
8.2 KiB
YAML

# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx942 -verify-machineinstrs -run-pass=prologepilog %s -o - | FileCheck -check-prefix=FLATSCRW64 %s
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1100 -verify-machineinstrs -run-pass=prologepilog %s -o - | FileCheck -check-prefix=FLATSCRW32 %s
# RUN: llc -mtriple=amdgcn-amd-amdpal -mcpu=gfx900 -verify-machineinstrs -run-pass=prologepilog %s -o - | FileCheck -check-prefix=MUBUF %s
---
# S_CSELECT_B32 with a frame index true-operand, preceded immediately by
# S_AND_B64 which produces SCC. The frame index has a non-zero offset so
# the ADDC path applies. Verify ADDC+BITCMP1+BITSET0 is generated.
name: s_cselect_b32_fi_after_scc_producer
tracksRegLiveness: true
stack:
- { id: 0, size: 64, alignment: 4 }
- { id: 1, size: 4, alignment: 4 }
machineFunctionInfo:
scratchRSrcReg: '$sgpr0_sgpr1_sgpr2_sgpr3'
frameOffsetReg: '$sgpr33'
stackPtrOffsetReg: '$sgpr32'
body: |
bb.0:
liveins: $sgpr4, $sgpr5, $sgpr6
; S_AND_B64 produces SCC. The regscavenger backward scan from S_CSELECT_B32
; will call stepBackward on S_AND_B64, removing SCC from liveness. Without
; the fix, RS->isRegUsed(SCC) returns false -> NeedSaveSCC=false -> plain
; S_ADD_I32 is inserted before S_CSELECT_B32 -> SCC clobbered -> wrong result.
; FLATSCRW64-LABEL: name: s_cselect_b32_fi_after_scc_producer
; FLATSCRW64: liveins: $sgpr4, $sgpr5, $sgpr6
; FLATSCRW64-NEXT: {{ $}}
; FLATSCRW64-NEXT: frame-setup CFI_INSTRUCTION escape 0x0f, 0x09, 0x90, 0x40, 0x94, 0x04, 0x36, 0x24, 0x36, 0xe9, 0x02
; FLATSCRW64-NEXT: frame-setup CFI_INSTRUCTION llvm_register_pair $pc_reg, $sgpr30, 32, $sgpr31, 32
; FLATSCRW64-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr4
; FLATSCRW64-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr5
; FLATSCRW64-NEXT: renamable $sgpr4_sgpr5 = S_AND_B64 killed renamable $sgpr4_sgpr5, $exec, implicit-def $scc
; FLATSCRW64-NEXT: $sgpr5 = S_ADDC_U32 $sgpr32, 64, implicit-def $scc, implicit $scc
; FLATSCRW64-NEXT: S_BITCMP1_B32 $sgpr5, 0, implicit-def $scc
; FLATSCRW64-NEXT: $sgpr5 = S_BITSET0_B32 0, $sgpr5
; FLATSCRW64-NEXT: renamable $sgpr4 = S_CSELECT_B32 killed $sgpr5, $sgpr6, implicit $scc
; FLATSCRW64-NEXT: SI_RETURN implicit $sgpr4
;
; FLATSCRW32-LABEL: name: s_cselect_b32_fi_after_scc_producer
; FLATSCRW32: liveins: $sgpr4, $sgpr5, $sgpr6
; FLATSCRW32-NEXT: {{ $}}
; FLATSCRW32-NEXT: frame-setup CFI_INSTRUCTION escape 0x0f, 0x09, 0x90, 0x40, 0x94, 0x04, 0x35, 0x24, 0x36, 0xe9, 0x02
; FLATSCRW32-NEXT: frame-setup CFI_INSTRUCTION llvm_register_pair $pc_reg, $sgpr30, 32, $sgpr31, 32
; FLATSCRW32-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr4
; FLATSCRW32-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr5
; FLATSCRW32-NEXT: renamable $sgpr4_sgpr5 = S_AND_B64 killed renamable $sgpr4_sgpr5, $exec, implicit-def $scc
; FLATSCRW32-NEXT: $sgpr5 = S_ADDC_U32 $sgpr32, 64, implicit-def $scc, implicit $scc
; FLATSCRW32-NEXT: S_BITCMP1_B32 $sgpr5, 0, implicit-def $scc
; FLATSCRW32-NEXT: $sgpr5 = S_BITSET0_B32 0, $sgpr5
; FLATSCRW32-NEXT: renamable $sgpr4 = S_CSELECT_B32 killed $sgpr5, $sgpr6, implicit $scc
; FLATSCRW32-NEXT: SI_RETURN implicit $sgpr4
;
; MUBUF-LABEL: name: s_cselect_b32_fi_after_scc_producer
; MUBUF: liveins: $sgpr4, $sgpr5, $sgpr6
; MUBUF-NEXT: {{ $}}
; MUBUF-NEXT: frame-setup CFI_INSTRUCTION llvm_def_aspace_cfa $sgpr32, 0, 6
; MUBUF-NEXT: frame-setup CFI_INSTRUCTION llvm_register_pair $pc_reg, $sgpr30, 32, $sgpr31, 32
; MUBUF-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr4
; MUBUF-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr5
; MUBUF-NEXT: renamable $sgpr4_sgpr5 = S_AND_B64 killed renamable $sgpr4_sgpr5, $exec, implicit-def $scc
; MUBUF-NEXT: $sgpr5 = S_LSHR_B32 $sgpr32, 6, implicit-def $scc
; MUBUF-NEXT: $sgpr5 = S_ADD_I32 killed $sgpr5, 64, implicit-def $scc
; MUBUF-NEXT: renamable $sgpr4 = S_CSELECT_B32 killed $sgpr5, $sgpr6, implicit $scc
; MUBUF-NEXT: SI_RETURN implicit $sgpr4
renamable $sgpr4_sgpr5 = S_AND_B64 killed renamable $sgpr4_sgpr5, $exec, implicit-def $scc
; %stack.1 is at offset 64 from frame base (non-zero), so eliminateFrameIndex
; cannot use the trivial FrameReg-only path. Verify it uses ADDC+BITCMP1+BITSET0.
renamable $sgpr4 = S_CSELECT_B32 %stack.1, $sgpr6, implicit $scc
SI_RETURN implicit $sgpr4
...
---
# S_ADDC_U32 with a frame index source-operand, preceded by S_AND_B64 which
# produces SCC. S_ADDC_U32 reads SCC, so frame index elimination must use the
# ADDC+BITCMP1+BITSET0 sequence to preserve the incoming SCC value.
name: s_addc_u32_fi_after_and_scc_producer
tracksRegLiveness: true
stack:
- { id: 0, size: 64, alignment: 4 }
- { id: 1, size: 4, alignment: 4 }
machineFunctionInfo:
scratchRSrcReg: '$sgpr0_sgpr1_sgpr2_sgpr3'
frameOffsetReg: '$sgpr33'
stackPtrOffsetReg: '$sgpr32'
body: |
bb.0:
liveins: $sgpr4, $sgpr5, $sgpr6
; S_AND_B64 produces SCC. S_ADDC_U32 has a frame index source operand and
; reads SCC (implicit $scc), so frame index elimination must preserve SCC
; using ADDC+BITCMP1+BITSET0 before expanding the frame index.
; FLATSCRW64-LABEL: name: s_addc_u32_fi_after_and_scc_producer
; FLATSCRW64: liveins: $sgpr4, $sgpr5, $sgpr6
; FLATSCRW64-NEXT: {{ $}}
; FLATSCRW64-NEXT: frame-setup CFI_INSTRUCTION escape 0x0f, 0x09, 0x90, 0x40, 0x94, 0x04, 0x36, 0x24, 0x36, 0xe9, 0x02
; FLATSCRW64-NEXT: frame-setup CFI_INSTRUCTION llvm_register_pair $pc_reg, $sgpr30, 32, $sgpr31, 32
; FLATSCRW64-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr4
; FLATSCRW64-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr5
; FLATSCRW64-NEXT: renamable $sgpr4_sgpr5 = S_AND_B64 killed renamable $sgpr4_sgpr5, $exec, implicit-def $scc
; FLATSCRW64-NEXT: $sgpr5 = S_ADDC_U32 $sgpr32, 64, implicit-def $scc, implicit $scc
; FLATSCRW64-NEXT: S_BITCMP1_B32 $sgpr5, 0, implicit-def $scc
; FLATSCRW64-NEXT: $sgpr5 = S_BITSET0_B32 0, $sgpr5
; FLATSCRW64-NEXT: renamable $sgpr4 = S_ADDC_U32 killed $sgpr5, $sgpr6, implicit-def $scc, implicit $scc
; FLATSCRW64-NEXT: SI_RETURN implicit $sgpr4
;
; FLATSCRW32-LABEL: name: s_addc_u32_fi_after_and_scc_producer
; FLATSCRW32: liveins: $sgpr4, $sgpr5, $sgpr6
; FLATSCRW32-NEXT: {{ $}}
; FLATSCRW32-NEXT: frame-setup CFI_INSTRUCTION escape 0x0f, 0x09, 0x90, 0x40, 0x94, 0x04, 0x35, 0x24, 0x36, 0xe9, 0x02
; FLATSCRW32-NEXT: frame-setup CFI_INSTRUCTION llvm_register_pair $pc_reg, $sgpr30, 32, $sgpr31, 32
; FLATSCRW32-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr4
; FLATSCRW32-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr5
; FLATSCRW32-NEXT: renamable $sgpr4_sgpr5 = S_AND_B64 killed renamable $sgpr4_sgpr5, $exec, implicit-def $scc
; FLATSCRW32-NEXT: $sgpr5 = S_ADDC_U32 $sgpr32, 64, implicit-def $scc, implicit $scc
; FLATSCRW32-NEXT: S_BITCMP1_B32 $sgpr5, 0, implicit-def $scc
; FLATSCRW32-NEXT: $sgpr5 = S_BITSET0_B32 0, $sgpr5
; FLATSCRW32-NEXT: renamable $sgpr4 = S_ADDC_U32 killed $sgpr5, $sgpr6, implicit-def $scc, implicit $scc
; FLATSCRW32-NEXT: SI_RETURN implicit $sgpr4
;
; MUBUF-LABEL: name: s_addc_u32_fi_after_and_scc_producer
; MUBUF: liveins: $sgpr4, $sgpr5, $sgpr6
; MUBUF-NEXT: {{ $}}
; MUBUF-NEXT: frame-setup CFI_INSTRUCTION llvm_def_aspace_cfa $sgpr32, 0, 6
; MUBUF-NEXT: frame-setup CFI_INSTRUCTION llvm_register_pair $pc_reg, $sgpr30, 32, $sgpr31, 32
; MUBUF-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr4
; MUBUF-NEXT: frame-setup CFI_INSTRUCTION undefined $sgpr5
; MUBUF-NEXT: renamable $sgpr4_sgpr5 = S_AND_B64 killed renamable $sgpr4_sgpr5, $exec, implicit-def $scc
; MUBUF-NEXT: $sgpr5 = S_LSHR_B32 $sgpr32, 6, implicit-def $scc
; MUBUF-NEXT: $sgpr5 = S_ADD_I32 killed $sgpr5, 64, implicit-def $scc
; MUBUF-NEXT: renamable $sgpr4 = S_ADDC_U32 killed $sgpr5, $sgpr6, implicit-def $scc, implicit $scc
; MUBUF-NEXT: SI_RETURN implicit $sgpr4
renamable $sgpr4_sgpr5 = S_AND_B64 killed renamable $sgpr4_sgpr5, $exec, implicit-def $scc
renamable $sgpr4 = S_ADDC_U32 %stack.1, $sgpr6, implicit-def $scc, implicit $scc
SI_RETURN implicit $sgpr4
...