llvm/test/CodeGen/AMDGPU/dag-preserve-disjoint-flag.ll

; NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 5
; RUN: llc -mtriple=amdgcn-amd-amdpal -mcpu=gfx900 -stop-after=finalize-isel -simplify-mir -o - %s | FileCheck %s

; Make sure disjoint flag is preserved on or instructions through selection

define amdgpu_ps i32 @s_or_i32_disjoint(i32 inreg %a, i32 inreg %b) {
  ; CHECK-LABEL: name: s_or_i32_disjoint
  ; CHECK: bb.0 (%ir-block.0):
  ; CHECK-NEXT:   liveins: $sgpr0, $sgpr1
  ; CHECK-NEXT: {{  $}}
  ; CHECK-NEXT:   [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr1
  ; CHECK-NEXT:   [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr0
  ; CHECK-NEXT:   %3:sreg_32 = disjoint S_OR_B32 [[COPY1]], [[COPY]], implicit-def dead $scc
  ; CHECK-NEXT:   $sgpr0 = COPY %3
  ; CHECK-NEXT:   SI_RETURN_TO_EPILOG $sgpr0
  %result = or disjoint i32 %a, %b
  ret i32 %result
}

define amdgpu_ps <2 x i32> @s_or_v2i32_disjoint(<2 x i32> inreg %a, <2 x i32> inreg %b) {
  ; CHECK-LABEL: name: s_or_v2i32_disjoint
  ; CHECK: bb.0 (%ir-block.0):
  ; CHECK-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3
  ; CHECK-NEXT: {{  $}}
  ; CHECK-NEXT:   [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr3
  ; CHECK-NEXT:   [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr2
  ; CHECK-NEXT:   [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr1
  ; CHECK-NEXT:   [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr0
  ; CHECK-NEXT:   %5:sreg_32 = disjoint S_OR_B32 [[COPY3]], [[COPY1]], implicit-def dead $scc
  ; CHECK-NEXT:   %6:sreg_32 = disjoint S_OR_B32 [[COPY2]], [[COPY]], implicit-def dead $scc
  ; CHECK-NEXT:   $sgpr0 = COPY %5
  ; CHECK-NEXT:   $sgpr1 = COPY %6
  ; CHECK-NEXT:   SI_RETURN_TO_EPILOG $sgpr0, $sgpr1
  %result = or disjoint <2 x i32> %a, %b
  ret <2 x i32> %result
}

define i32 @v_or_i32_disjoint(i32 %a, i32 %b) {
  ; CHECK-LABEL: name: v_or_i32_disjoint
  ; CHECK: bb.0 (%ir-block.0):
  ; CHECK-NEXT:   liveins: $vgpr0, $vgpr1
  ; CHECK-NEXT: {{  $}}
  ; CHECK-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; CHECK-NEXT:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; CHECK-NEXT:   %10:vgpr_32 = disjoint V_OR_B32_e64 [[COPY1]], [[COPY]], implicit $exec
  ; CHECK-NEXT:   $vgpr0 = COPY %10
  ; CHECK-NEXT:   SI_RETURN implicit $vgpr0
  %result = or disjoint i32 %a, %b
  ret i32 %result
}

define <2 x i32> @v_or_v2i32_disjoint(<2 x i32> %a, <2 x i32> %b) {
  ; CHECK-LABEL: name: v_or_v2i32_disjoint
  ; CHECK: bb.0 (%ir-block.0):
  ; CHECK-NEXT:   liveins: $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; CHECK-NEXT: {{  $}}
  ; CHECK-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr3
  ; CHECK-NEXT:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; CHECK-NEXT:   [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; CHECK-NEXT:   [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; CHECK-NEXT:   %12:vgpr_32 = disjoint V_OR_B32_e64 [[COPY3]], [[COPY1]], implicit $exec
  ; CHECK-NEXT:   %13:vgpr_32 = disjoint V_OR_B32_e64 [[COPY2]], [[COPY]], implicit $exec
  ; CHECK-NEXT:   $vgpr0 = COPY %12
  ; CHECK-NEXT:   $vgpr1 = COPY %13
  ; CHECK-NEXT:   SI_RETURN implicit $vgpr0, implicit $vgpr1
  %result = or disjoint <2 x i32> %a, %b
  ret <2 x i32> %result
}

define amdgpu_ps i64 @s_or_i64_disjoint(i64 inreg %a, i64 inreg %b) {
  ; CHECK-LABEL: name: s_or_i64_disjoint
  ; CHECK: bb.0 (%ir-block.0):
  ; CHECK-NEXT:   liveins: $sgpr0, $sgpr1, $sgpr2, $sgpr3
  ; CHECK-NEXT: {{  $}}
  ; CHECK-NEXT:   [[COPY:%[0-9]+]]:sgpr_32 = COPY $sgpr3
  ; CHECK-NEXT:   [[COPY1:%[0-9]+]]:sgpr_32 = COPY $sgpr2
  ; CHECK-NEXT:   [[COPY2:%[0-9]+]]:sgpr_32 = COPY $sgpr1
  ; CHECK-NEXT:   [[COPY3:%[0-9]+]]:sgpr_32 = COPY $sgpr0
  ; CHECK-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:sgpr_64 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY]], %subreg.sub1
  ; CHECK-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:sgpr_64 = REG_SEQUENCE [[COPY3]], %subreg.sub0, [[COPY2]], %subreg.sub1
  ; CHECK-NEXT:   %7:sreg_64 = disjoint S_OR_B64 killed [[REG_SEQUENCE1]], killed [[REG_SEQUENCE]], implicit-def dead $scc
  ; CHECK-NEXT:   [[COPY4:%[0-9]+]]:sreg_32 = COPY %7.sub1
  ; CHECK-NEXT:   [[COPY5:%[0-9]+]]:sreg_32 = COPY %7.sub0
  ; CHECK-NEXT:   $sgpr0 = COPY [[COPY5]]
  ; CHECK-NEXT:   $sgpr1 = COPY [[COPY4]]
  ; CHECK-NEXT:   SI_RETURN_TO_EPILOG $sgpr0, $sgpr1
  %result = or disjoint i64 %a, %b
  ret i64 %result
}

define i64 @v_or_i64_disjoint(i64 %a, i64 %b) {
  ; CHECK-LABEL: name: v_or_i64_disjoint
  ; CHECK: bb.0 (%ir-block.0):
  ; CHECK-NEXT:   liveins: $vgpr0, $vgpr1, $vgpr2, $vgpr3
  ; CHECK-NEXT: {{  $}}
  ; CHECK-NEXT:   [[COPY:%[0-9]+]]:vgpr_32 = COPY $vgpr3
  ; CHECK-NEXT:   [[COPY1:%[0-9]+]]:vgpr_32 = COPY $vgpr2
  ; CHECK-NEXT:   [[COPY2:%[0-9]+]]:vgpr_32 = COPY $vgpr1
  ; CHECK-NEXT:   [[COPY3:%[0-9]+]]:vgpr_32 = COPY $vgpr0
  ; CHECK-NEXT:   [[DEF:%[0-9]+]]:sgpr_32 = IMPLICIT_DEF
  ; CHECK-NEXT:   [[DEF1:%[0-9]+]]:sgpr_32 = IMPLICIT_DEF
  ; CHECK-NEXT:   [[REG_SEQUENCE:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY3]], %subreg.sub0, [[COPY2]], %subreg.sub1
  ; CHECK-NEXT:   [[DEF2:%[0-9]+]]:sgpr_32 = IMPLICIT_DEF
  ; CHECK-NEXT:   [[DEF3:%[0-9]+]]:sgpr_32 = IMPLICIT_DEF
  ; CHECK-NEXT:   [[REG_SEQUENCE1:%[0-9]+]]:vreg_64 = REG_SEQUENCE [[COPY1]], %subreg.sub0, [[COPY]], %subreg.sub1
  ; CHECK-NEXT:   [[COPY4:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE1]].sub1
  ; CHECK-NEXT:   [[COPY5:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub1
  ; CHECK-NEXT:   [[V_OR_B32_e64_:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 killed [[COPY5]], killed [[COPY4]], implicit $exec
  ; CHECK-NEXT:   [[COPY6:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE1]].sub0
  ; CHECK-NEXT:   [[COPY7:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE]].sub0
  ; CHECK-NEXT:   [[V_OR_B32_e64_1:%[0-9]+]]:vgpr_32 = V_OR_B32_e64 killed [[COPY7]], killed [[COPY6]], implicit $exec
  ; CHECK-NEXT:   [[REG_SEQUENCE2:%[0-9]+]]:vreg_64 = REG_SEQUENCE killed [[V_OR_B32_e64_1]], %subreg.sub0, killed [[V_OR_B32_e64_]], %subreg.sub1
  ; CHECK-NEXT:   [[COPY8:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE2]].sub1
  ; CHECK-NEXT:   [[COPY9:%[0-9]+]]:vgpr_32 = COPY [[REG_SEQUENCE2]].sub0
  ; CHECK-NEXT:   $vgpr0 = COPY [[COPY9]]
  ; CHECK-NEXT:   $vgpr1 = COPY [[COPY8]]
  ; CHECK-NEXT:   SI_RETURN implicit $vgpr0, implicit $vgpr1
  %result = or disjoint i64 %a, %b
  ret i64 %result
}