Skip to content

[nfc][AMDGPU] Autogenerated assertions applied to fold-imm-copy.mir #135071

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Apr 9, 2025
Merged

[nfc][AMDGPU] Autogenerated assertions applied to fold-imm-copy.mir #135071

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
57 changes: 37 additions & 20 deletions llvm/test/CodeGen/AMDGPU/fold-imm-copy.mir
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,14 +1,23 @@
# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 5
# RUN: llc -mtriple=amdgcn -run-pass si-fold-operands -verify-machineinstrs %s -o - | FileCheck -check-prefix=GCN %s

# GCN-LABEL: name: fold-imm-copy
# GCN: V_AND_B32_e32 65535

---
name: fold-imm-copy
tracksRegLiveness: true
body: |
bb.0:
liveins: $vgpr0, $sgpr0_sgpr1
; GCN-LABEL: name: fold-imm-copy
; GCN: liveins: $vgpr0, $sgpr0_sgpr1
; GCN-NEXT: {{ $}}
; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; GCN-NEXT: [[COPY1:%[0-9]+]]:sgpr_64 = COPY $sgpr0_sgpr1
; GCN-NEXT: [[S_LOAD_DWORDX4_IMM:%[0-9]+]]:sgpr_128 = S_LOAD_DWORDX4_IMM [[COPY1]], 9, 0
; GCN-NEXT: [[V_LSHLREV_B32_e64_:%[0-9]+]]:vgpr_32 = V_LSHLREV_B32_e64 2, [[COPY]], implicit $exec
; GCN-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE killed [[V_LSHLREV_B32_e64_]], %subreg.sub0, killed [[V_MOV_B32_e32_]], %subreg.sub1
; GCN-NEXT: [[BUFFER_LOAD_DWORD_ADDR64_:%[0-9]+]]:vgpr_32 = BUFFER_LOAD_DWORD_ADDR64 [[REG_SEQUENCE]], [[S_LOAD_DWORDX4_IMM]], 0, 4, 0, 0, implicit $exec
; GCN-NEXT: [[V_AND_B32_e32_:%[0-9]+]]:vgpr_32 = V_AND_B32_e32 65535, [[BUFFER_LOAD_DWORD_ADDR64_]], implicit $exec
%0:vgpr_32 = COPY $vgpr0
%1:sgpr_64 = COPY $sgpr0_sgpr1
%2:sgpr_128 = S_LOAD_DWORDX4_IMM %1, 9, 0
Expand All @@ -23,14 +32,19 @@ body: |
...

---
# GCN-LABEL: name: no_extra_fold_on_same_opnd
# The first XOR needs commuting to fold that immediate operand.
# GCN: V_XOR_B32_e32 0, %1
# GCN: V_XOR_B32_e32 %2, %4.sub0
name: no_extra_fold_on_same_opnd
tracksRegLiveness: true
body: |
bb.0:
; GCN-LABEL: name: no_extra_fold_on_same_opnd
; GCN: [[DEF:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
; GCN-NEXT: [[DEF1:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
; GCN-NEXT: [[DEF2:%[0-9]+]]:vgpr_32 = IMPLICIT_DEF
; GCN-NEXT: [[V_MOV_B32_e32_:%[0-9]+]]:vgpr_32 = V_MOV_B32_e32 0, implicit $exec
; GCN-NEXT: [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE killed [[DEF]], %subreg.sub0, killed [[V_MOV_B32_e32_]], %subreg.sub1
; GCN-NEXT: [[V_XOR_B32_e32_:%[0-9]+]]:vgpr_32 = V_XOR_B32_e32 0, [[DEF1]], implicit $exec
; GCN-NEXT: [[V_XOR_B32_e32_1:%[0-9]+]]:vgpr_32 = V_XOR_B32_e32 [[DEF2]], [[REG_SEQUENCE]].sub0, implicit $exec
%0:vgpr_32 = IMPLICIT_DEF
%1:vgpr_32 = IMPLICIT_DEF
%2:vgpr_32 = IMPLICIT_DEF
Expand All @@ -45,42 +59,44 @@ body: |
# Make sure the subreg index is not reinterpreted when folding
# immediates
#
# GCN-LABEL: name: clear_subreg_imm_fold{{$}}
# GCN: %1:sgpr_32 = S_MOV_B32 4294967288
# GCN: %2:sgpr_32 = S_MOV_B32 4294967295
name: clear_subreg_imm_fold
tracksRegLiveness: true
body: |
bb.0:
; GCN-LABEL: name: clear_subreg_imm_fold
; GCN: [[S_MOV_B32_:%[0-9]+]]:sgpr_32 = S_MOV_B32 4294967288
; GCN-NEXT: [[S_MOV_B32_1:%[0-9]+]]:sgpr_32 = S_MOV_B32 4294967295
; GCN-NEXT: S_ENDPGM 0, implicit [[S_MOV_B32_]], implicit [[S_MOV_B32_1]]
%0:sreg_64 = S_MOV_B64 -8
%1:sgpr_32 = COPY %0.sub0
%2:sgpr_32 = COPY %0.sub1
S_ENDPGM 0, implicit %1, implicit %2

...

# GCN-LABEL: name: no_fold_imm_into_m0{{$}}
# GCN: %0:sreg_32 = S_MOV_B32 -8
# GCN-NEXT: $m0 = COPY %0

---
name: no_fold_imm_into_m0
tracksRegLiveness: true
body: |
bb.0:
; GCN-LABEL: name: no_fold_imm_into_m0
; GCN: [[S_MOV_B32_:%[0-9]+]]:sreg_32 = S_MOV_B32 -8
; GCN-NEXT: $m0 = COPY [[S_MOV_B32_]]
; GCN-NEXT: S_ENDPGM 0, implicit $m0
%0:sreg_32 = S_MOV_B32 -8
$m0 = COPY %0
S_ENDPGM 0, implicit $m0

...

# GCN-LABEL: name: fold_sgpr_imm_to_vgpr_copy{{$}}
# GCN: $vgpr0 = V_MOV_B32_e32 -8, implicit $exec
---
name: fold_sgpr_imm_to_vgpr_copy
tracksRegLiveness: true
body: |
bb.0:
; GCN-LABEL: name: fold_sgpr_imm_to_vgpr_copy
; GCN: $vgpr0 = V_MOV_B32_e32 -8, implicit $exec
; GCN-NEXT: S_ENDPGM 0, implicit $vgpr0
%0:sreg_32 = S_MOV_B32 -8
$vgpr0 = COPY %0
S_ENDPGM 0, implicit $vgpr0
Expand All @@ -90,16 +106,17 @@ body: |
# The users of $vgpr1 should not be visited for further immediate
# folding.

# GCN-LABEL: name: no_fold_physreg_users_vgpr{{$}}
# GCN: $vgpr1 = V_MOV_B32_e32 0, implicit $exec
# GCN-NEXT: S_NOP 0, implicit-def $vgpr1
# GCN-NEXT: %2:vgpr_32 = COPY $vgpr1
# GCN-NEXT: $vgpr2 = COPY %2
---
name: no_fold_physreg_users_vgpr
tracksRegLiveness: true
body: |
bb.0:
; GCN-LABEL: name: no_fold_physreg_users_vgpr
; GCN: $vgpr1 = V_MOV_B32_e32 0, implicit $exec
; GCN-NEXT: S_NOP 0, implicit-def $vgpr1
; GCN-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; GCN-NEXT: $vgpr2 = COPY [[COPY]]
; GCN-NEXT: S_ENDPGM 0
%0:sreg_32 = S_MOV_B32 0
%1:vgpr_32 = COPY %0
$vgpr1 = COPY %0
Expand Down