# RUN: llc -mtriple=amdgcn -mcpu=fiji -verify-machineinstrs -run-pass si-insert-waits -o - %s | FileCheck %s

--- |
  define float @waitcnt-permute(i32 %x, i32 %y) {
  entry:
    %0 = call i32 @llvm.amdgcn.ds.bpermute(i32 %x, i32 %y)
    %1 = bitcast i32 %0 to float
    %2 = fadd float 1.000000e+00, %1
    ret float %2
  }

  declare i32 @llvm.amdgcn.ds.bpermute(i32, i32)

...
---
# CHECK-LABEL: name: waitcnt-permute{{$}}
# CHECK: DS_BPERMUTE_B32
# CHECK-NEXT: S_WAITCNT 127

name:            waitcnt-permute
liveins:
  - { reg: '%vgpr0' }
  - { reg: '%vgpr1' }
  - { reg: '%sgpr30_sgpr31' }
body:             |
  bb.0:
    liveins: %vgpr0, %vgpr1, %sgpr30_sgpr31

    %vgpr0 = DS_BPERMUTE_B32 killed %vgpr0, killed %vgpr1, 0, implicit %exec
    %vgpr0 = V_ADD_F32_e32 1065353216, killed %vgpr0, implicit %exec
    S_SETPC_B64_return killed %sgpr30_sgpr31, implicit killed %vgpr0

...