# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 3
# RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1031 -run-pass=machine-sink -o - %s | FileCheck %s
# A VGPR loop variable was incorrectly sunk into a flow block, past
# the si_end_cf reconvergence point.
---
name: machinesink_loop_vgpr_out_of_divergent_loop
tracksRegLiveness: true
machineFunctionInfo:
scratchRSrcReg: '$sgpr0_sgpr1_sgpr2_sgpr3'
frameOffsetReg: '$sgpr33'
stackPtrOffsetReg: '$sgpr32'
body: |
; CHECK-LABEL: name: machinesink_loop_vgpr_out_of_divergent_loop
; CHECK: bb.0:
; CHECK-NEXT: successors: %bb.1(0x80000000)
; CHECK-NEXT: liveins: $vgpr0, $vgpr1, $sgpr8
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr0
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr8
; CHECK-NEXT: [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.1:
; CHECK-NEXT: successors: %bb.6(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[SI_IF:%[0-9]+]]:sreg_32 = SI_IF [[COPY1]], %bb.6, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
; CHECK-NEXT: S_BRANCH %bb.2
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.3(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[SI_IF1:%[0-9]+]]:sreg_32 = SI_IF [[COPY1]], %bb.4, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
; CHECK-NEXT: S_BRANCH %bb.3
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_NOP 0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.5(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; CHECK-NEXT: [[SI_IF_BREAK:%[0-9]+]]:sreg_32 = SI_IF_BREAK killed [[SI_IF1]], [[SI_IF]], implicit-def dead $scc
; CHECK-NEXT: SI_LOOP [[SI_IF_BREAK]], %bb.2, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
; CHECK-NEXT: S_BRANCH %bb.5
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.5:
; CHECK-NEXT: successors: %bb.2(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[PHI:%[0-9]+]]:vgpr_32 = PHI [[COPY]], %bb.4
; CHECK-NEXT: SI_END_CF [[SI_IF_BREAK]], implicit-def dead $exec, implicit-def dead $scc, implicit $exec
; CHECK-NEXT: [[V_ADD_U32_e64_:%[0-9]+]]:vgpr_32 = V_ADD_U32_e64 [[COPY]], [[COPY1]], 0, implicit $exec
; CHECK-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */, implicit [[V_ADD_U32_e64_]]
; CHECK-NEXT: S_BRANCH %bb.2
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.6:
; CHECK-NEXT: successors: %bb.1(0x40000000), %bb.7(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: SI_LOOP [[SI_IF]], %bb.1, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
; CHECK-NEXT: S_BRANCH %bb.7
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.7:
; CHECK-NEXT: successors: %bb.7(0x40000000), %bb.8(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_CBRANCH_VCCNZ %bb.7, implicit undef $vcc
; CHECK-NEXT: S_BRANCH %bb.8
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.8:
; CHECK-NEXT: SI_RETURN
bb.0:
liveins: $vgpr0, $vgpr1, $sgpr8
%0:vgpr_32 = COPY $vgpr0
%1:sreg_32 = COPY $sgpr8
%2:vgpr_32 = COPY $vgpr1
bb.1:
%3:sreg_32 = SI_IF %1, %bb.6, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
S_BRANCH %bb.2
bb.2:
%4:sreg_32 = SI_IF %1, %bb.4, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
S_BRANCH %bb.3
bb.3:
S_NOP 0
bb.4:
INLINEASM &"", 1 /* sideeffect attdialect */
%5:vgpr_32 = V_ADD_U32_e64 %0, %1, 0, implicit $exec
%6:sreg_32 = SI_IF_BREAK killed %4, %3, implicit-def dead $scc
SI_LOOP %6, %bb.2, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
S_BRANCH %bb.5
bb.5:
%7:vgpr_32 = PHI %0, %bb.4
SI_END_CF %6, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
INLINEASM &"", 1, implicit %5
S_BRANCH %bb.2
bb.6:
SI_LOOP %3, %bb.1, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
S_BRANCH %bb.7
bb.7:
S_CBRANCH_VCCNZ %bb.7, implicit undef $vcc
S_BRANCH %bb.8
bb.8:
SI_RETURN
...
# The same testcase, except the relevant instruction is scalar and
# could be legally sunk.
---
name: machinesink_loop_sgpr_out_of_divergent_loop
tracksRegLiveness: true
machineFunctionInfo:
scratchRSrcReg: '$sgpr0_sgpr1_sgpr2_sgpr3'
frameOffsetReg: '$sgpr33'
stackPtrOffsetReg: '$sgpr32'
body: |
; CHECK-LABEL: name: machinesink_loop_sgpr_out_of_divergent_loop
; CHECK: bb.0:
; CHECK-NEXT: successors: %bb.1(0x80000000)
; CHECK-NEXT: liveins: $sgpr8, $sgpr9, $sgpr10
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[COPY:%[0-9]+]]:sreg_32 = COPY $sgpr8
; CHECK-NEXT: [[COPY1:%[0-9]+]]:sreg_32 = COPY $sgpr9
; CHECK-NEXT: [[COPY2:%[0-9]+]]:sreg_32 = COPY $sgpr10
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.1:
; CHECK-NEXT: successors: %bb.6(0x40000000), %bb.2(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[SI_IF:%[0-9]+]]:sreg_32 = SI_IF [[COPY1]], %bb.6, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
; CHECK-NEXT: S_BRANCH %bb.2
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.2:
; CHECK-NEXT: successors: %bb.4(0x40000000), %bb.3(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[SI_IF1:%[0-9]+]]:sreg_32 = SI_IF [[COPY1]], %bb.4, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
; CHECK-NEXT: S_BRANCH %bb.3
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.3:
; CHECK-NEXT: successors: %bb.4(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_NOP 0
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.4:
; CHECK-NEXT: successors: %bb.2(0x40000000), %bb.5(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */
; CHECK-NEXT: [[SI_IF_BREAK:%[0-9]+]]:sreg_32 = SI_IF_BREAK killed [[SI_IF1]], [[SI_IF]], implicit-def dead $scc
; CHECK-NEXT: SI_LOOP [[SI_IF_BREAK]], %bb.2, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
; CHECK-NEXT: S_BRANCH %bb.5
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.5:
; CHECK-NEXT: successors: %bb.2(0x80000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: [[PHI:%[0-9]+]]:vgpr_32 = PHI [[COPY]], %bb.4
; CHECK-NEXT: SI_END_CF [[SI_IF_BREAK]], implicit-def dead $exec, implicit-def dead $scc, implicit $exec
; CHECK-NEXT: [[S_ADD_I32_:%[0-9]+]]:sreg_32 = S_ADD_I32 [[COPY]], [[COPY1]], implicit-def dead $scc
; CHECK-NEXT: INLINEASM &"", 1 /* sideeffect attdialect */, implicit [[S_ADD_I32_]]
; CHECK-NEXT: S_BRANCH %bb.2
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.6:
; CHECK-NEXT: successors: %bb.1(0x40000000), %bb.7(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: SI_LOOP [[SI_IF]], %bb.1, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
; CHECK-NEXT: S_BRANCH %bb.7
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.7:
; CHECK-NEXT: successors: %bb.7(0x40000000), %bb.8(0x40000000)
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: S_CBRANCH_VCCNZ %bb.7, implicit undef $vcc
; CHECK-NEXT: S_BRANCH %bb.8
; CHECK-NEXT: {{ $}}
; CHECK-NEXT: bb.8:
; CHECK-NEXT: SI_RETURN
bb.0:
liveins: $sgpr8, $sgpr9, $sgpr10
%0:sreg_32 = COPY $sgpr8
%1:sreg_32 = COPY $sgpr9
%2:sreg_32 = COPY $sgpr10
bb.1:
%3:sreg_32 = SI_IF %1, %bb.6, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
S_BRANCH %bb.2
bb.2:
%4:sreg_32 = SI_IF %1, %bb.4, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
S_BRANCH %bb.3
bb.3:
S_NOP 0
bb.4:
INLINEASM &"", 1 /* sideeffect attdialect */
%5:sreg_32 = S_ADD_I32 %0, %1, implicit-def dead $scc
%6:sreg_32 = SI_IF_BREAK killed %4, %3, implicit-def dead $scc
SI_LOOP %6, %bb.2, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
S_BRANCH %bb.5
bb.5:
%7:vgpr_32 = PHI %0, %bb.4
SI_END_CF %6, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
INLINEASM &"", 1, implicit %5
S_BRANCH %bb.2
bb.6:
SI_LOOP %3, %bb.1, implicit-def dead $exec, implicit-def dead $scc, implicit $exec
S_BRANCH %bb.7
bb.7:
S_CBRANCH_VCCNZ %bb.7, implicit undef $vcc
S_BRANCH %bb.8
bb.8:
SI_RETURN
...
Codebase Browser
llvm