llvm/test/CodeGen/AMDGPU/expand-waitcnt-profiling-no-outstanding.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc -mtriple=amdgpu9.00 -amdgpu-waitcnt-forcezero < %s | FileCheck --check-prefix=GFX9 %s
; RUN: llc -mtriple=amdgpu12.00 -amdgpu-waitcnt-forcezero < %s | FileCheck --check-prefix=GFX12 %s

; Regression test for an unsigned-underflow infinite loop in the Pre-GFX12
; waitcnt profiling expansion (WaitcntGeneratorPreGFX12::createNewWaitcnt).

define amdgpu_kernel void @no_outstanding(ptr addrspace(1) %p) #0 {
; GFX9-LABEL: no_outstanding:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_waitcnt expcnt(0)
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x24
; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v0, 0
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_waitcnt expcnt(0)
; GFX9-NEXT:    global_store_dword v0, v0, s[0:1]
; GFX9-NEXT:    s_endpgm
;
; GFX12-LABEL: no_outstanding:
; GFX12:       ; %bb.0:
; GFX12-NEXT:    s_wait_loadcnt 0x0
; GFX12-NEXT:    s_wait_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    s_load_b64 s[0:1], s[4:5], 0x24
; GFX12-NEXT:    s_wait_loadcnt 0x0
; GFX12-NEXT:    s_wait_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    v_mov_b32_e32 v0, 0
; GFX12-NEXT:    s_wait_loadcnt 0x0
; GFX12-NEXT:    s_wait_dscnt 0x0
; GFX12-NEXT:    s_wait_expcnt 0x0
; GFX12-NEXT:    s_wait_samplecnt 0x0
; GFX12-NEXT:    s_wait_bvhcnt 0x0
; GFX12-NEXT:    s_wait_kmcnt 0x0
; GFX12-NEXT:    global_store_b32 v0, v0, s[0:1]
; GFX12-NEXT:    s_endpgm
  store i32 0, ptr addrspace(1) %p
  ret void
}

attributes #0 = { "amdgpu-expand-waitcnt-profiling" }