; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=bonaire -mattr=+flat-for-global -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,CI %s
; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=tonga -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,VI %s
; RUN: llc -global-isel -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX9 %s

; FIXME: Merge with other test. DS offset folding doesn't work due to
; register bank copies, and no return optimization is missing.


declare i32 @llvm.amdgcn.atomic.inc.i32.p1i32(i32 addrspace(1)* nocapture, i32, i32, i32, i1) #2
declare i32 @llvm.amdgcn.atomic.inc.i32.p3i32(i32 addrspace(3)* nocapture, i32, i32, i32, i1) #2
declare i32 @llvm.amdgcn.atomic.inc.i32.p0i32(i32* nocapture, i32, i32, i32, i1) #2

declare i64 @llvm.amdgcn.atomic.inc.i64.p1i64(i64 addrspace(1)* nocapture, i64, i32, i32, i1) #2
declare i64 @llvm.amdgcn.atomic.inc.i64.p3i64(i64 addrspace(3)* nocapture, i64, i32, i32, i1) #2
declare i64 @llvm.amdgcn.atomic.inc.i64.p0i64(i64* nocapture, i64, i32, i32, i1) #2

declare i32 @llvm.amdgcn.workitem.id.x() #1

define amdgpu_kernel void @lds_atomic_inc_ret_i32(i32 addrspace(1)* %out, i32 addrspace(3)* %ptr) #0 {
; CI-LABEL: lds_atomic_inc_ret_i32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    s_load_dword s2, s[4:5], 0x2
; CI-NEXT:    v_mov_b32_e32 v0, 42
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v1, s2
; CI-NEXT:    ds_inc_rtn_u32 v2, v1, v0
; CI-NEXT:    v_mov_b32_e32 v0, s0
; CI-NEXT:    v_mov_b32_e32 v1, s1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    flat_store_dword v[0:1], v2
; CI-NEXT:    s_endpgm
;
; VI-LABEL: lds_atomic_inc_ret_i32:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    s_load_dword s2, s[4:5], 0x8
; VI-NEXT:    v_mov_b32_e32 v0, 42
; VI-NEXT:    s_mov_b32 m0, -1
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v1, s2
; VI-NEXT:    ds_inc_rtn_u32 v2, v1, v0
; VI-NEXT:    v_mov_b32_e32 v0, s0
; VI-NEXT:    v_mov_b32_e32 v1, s1
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    flat_store_dword v[0:1], v2
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: lds_atomic_inc_ret_i32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    s_load_dword s2, s[4:5], 0x8
; GFX9-NEXT:    v_mov_b32_e32 v1, 42
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v0, s2
; GFX9-NEXT:    ds_inc_rtn_u32 v2, v0, v1
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_store_dword v[0:1], v2, off
; GFX9-NEXT:    s_endpgm
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p3i32(i32 addrspace(3)* %ptr, i32 42, i32 0, i32 0, i1 false), !noalias !0
  store i32 %result, i32 addrspace(1)* %out
  ret void
}

!0 = distinct !{!0, !"noalias-scope"}

define amdgpu_kernel void @lds_atomic_inc_ret_i32_offset(i32 addrspace(1)* %out, i32 addrspace(3)* %ptr) #0 {
; CI-LABEL: lds_atomic_inc_ret_i32_offset:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    s_load_dword s2, s[4:5], 0x2
; CI-NEXT:    v_mov_b32_e32 v0, 42
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_add_u32 s2, s2, 16
; CI-NEXT:    v_mov_b32_e32 v1, s2
; CI-NEXT:    ds_inc_rtn_u32 v2, v1, v0
; CI-NEXT:    v_mov_b32_e32 v0, s0
; CI-NEXT:    v_mov_b32_e32 v1, s1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    flat_store_dword v[0:1], v2
; CI-NEXT:    s_endpgm
;
; VI-LABEL: lds_atomic_inc_ret_i32_offset:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    s_load_dword s2, s[4:5], 0x8
; VI-NEXT:    v_mov_b32_e32 v0, 42
; VI-NEXT:    s_mov_b32 m0, -1
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    s_add_u32 s2, s2, 16
; VI-NEXT:    v_mov_b32_e32 v1, s2
; VI-NEXT:    ds_inc_rtn_u32 v2, v1, v0
; VI-NEXT:    v_mov_b32_e32 v0, s0
; VI-NEXT:    v_mov_b32_e32 v1, s1
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    flat_store_dword v[0:1], v2
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: lds_atomic_inc_ret_i32_offset:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    s_load_dword s2, s[4:5], 0x8
; GFX9-NEXT:    v_mov_b32_e32 v1, 42
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_add_u32 s2, s2, 16
; GFX9-NEXT:    v_mov_b32_e32 v0, s2
; GFX9-NEXT:    ds_inc_rtn_u32 v2, v0, v1
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_store_dword v[0:1], v2, off
; GFX9-NEXT:    s_endpgm
  %gep = getelementptr i32, i32 addrspace(3)* %ptr, i32 4
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p3i32(i32 addrspace(3)* %gep, i32 42, i32 0, i32 0, i1 false)
  store i32 %result, i32 addrspace(1)* %out
  ret void
}

define amdgpu_kernel void @lds_atomic_inc_noret_i32(i32 addrspace(3)* %ptr) nounwind {
; CI-LABEL: lds_atomic_inc_noret_i32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dword s0, s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v0, 42
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v1, s0
; CI-NEXT:    ds_inc_rtn_u32 v0, v1, v0
; CI-NEXT:    s_endpgm
;
; VI-LABEL: lds_atomic_inc_noret_i32:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dword s0, s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v0, 42
; VI-NEXT:    s_mov_b32 m0, -1
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v1, s0
; VI-NEXT:    ds_inc_rtn_u32 v0, v1, v0
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: lds_atomic_inc_noret_i32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v1, 42
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    ds_inc_rtn_u32 v0, v0, v1
; GFX9-NEXT:    s_endpgm
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p3i32(i32 addrspace(3)* %ptr, i32 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @lds_atomic_inc_noret_i32_offset(i32 addrspace(3)* %ptr) nounwind {
; CI-LABEL: lds_atomic_inc_noret_i32_offset:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dword s0, s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v0, 42
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_add_u32 s0, s0, 16
; CI-NEXT:    v_mov_b32_e32 v1, s0
; CI-NEXT:    ds_inc_rtn_u32 v0, v1, v0
; CI-NEXT:    s_endpgm
;
; VI-LABEL: lds_atomic_inc_noret_i32_offset:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dword s0, s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v0, 42
; VI-NEXT:    s_mov_b32 m0, -1
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    s_add_u32 s0, s0, 16
; VI-NEXT:    v_mov_b32_e32 v1, s0
; VI-NEXT:    ds_inc_rtn_u32 v0, v1, v0
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: lds_atomic_inc_noret_i32_offset:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v1, 42
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_add_u32 s0, s0, 16
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    ds_inc_rtn_u32 v0, v0, v1
; GFX9-NEXT:    s_endpgm
  %gep = getelementptr i32, i32 addrspace(3)* %ptr, i32 4
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p3i32(i32 addrspace(3)* %gep, i32 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @global_atomic_inc_ret_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %ptr) #0 {
; CI-LABEL: global_atomic_inc_ret_i32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v0, s2
; CI-NEXT:    v_mov_b32_e32 v1, s3
; CI-NEXT:    flat_atomic_inc v2, v[0:1], v2 glc
; CI-NEXT:    v_mov_b32_e32 v0, s0
; CI-NEXT:    v_mov_b32_e32 v1, s1
; CI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; CI-NEXT:    flat_store_dword v[0:1], v2
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_ret_i32:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v0, s2
; VI-NEXT:    v_mov_b32_e32 v1, s3
; VI-NEXT:    flat_atomic_inc v2, v[0:1], v2 glc
; VI-NEXT:    v_mov_b32_e32 v0, s0
; VI-NEXT:    v_mov_b32_e32 v1, s1
; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; VI-NEXT:    flat_store_dword v[0:1], v2
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_ret_i32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v0, s2
; GFX9-NEXT:    v_mov_b32_e32 v1, s3
; GFX9-NEXT:    global_atomic_inc v2, v[0:1], v2, off glc
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_store_dword v[0:1], v2, off
; GFX9-NEXT:    s_endpgm
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p1i32(i32 addrspace(1)* %ptr, i32 42, i32 0, i32 0, i1 false)
  store i32 %result, i32 addrspace(1)* %out
  ret void
}

define amdgpu_kernel void @global_atomic_inc_ret_i32_offset(i32 addrspace(1)* %out, i32 addrspace(1)* %ptr) #0 {
; CI-LABEL: global_atomic_inc_ret_i32_offset:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_add_u32 s2, s2, 16
; CI-NEXT:    s_addc_u32 s3, s3, 0
; CI-NEXT:    v_mov_b32_e32 v0, s2
; CI-NEXT:    v_mov_b32_e32 v1, s3
; CI-NEXT:    flat_atomic_inc v2, v[0:1], v2 glc
; CI-NEXT:    v_mov_b32_e32 v0, s0
; CI-NEXT:    v_mov_b32_e32 v1, s1
; CI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; CI-NEXT:    flat_store_dword v[0:1], v2
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_ret_i32_offset:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    s_add_u32 s2, s2, 16
; VI-NEXT:    s_addc_u32 s3, s3, 0
; VI-NEXT:    v_mov_b32_e32 v0, s2
; VI-NEXT:    v_mov_b32_e32 v1, s3
; VI-NEXT:    flat_atomic_inc v2, v[0:1], v2 glc
; VI-NEXT:    v_mov_b32_e32 v0, s0
; VI-NEXT:    v_mov_b32_e32 v1, s1
; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; VI-NEXT:    flat_store_dword v[0:1], v2
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_ret_i32_offset:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_add_u32 s2, s2, 16
; GFX9-NEXT:    s_addc_u32 s3, s3, 0
; GFX9-NEXT:    v_mov_b32_e32 v0, s2
; GFX9-NEXT:    v_mov_b32_e32 v1, s3
; GFX9-NEXT:    global_atomic_inc v2, v[0:1], v2, off glc
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_store_dword v[0:1], v2, off
; GFX9-NEXT:    s_endpgm
  %gep = getelementptr i32, i32 addrspace(1)* %ptr, i32 4
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p1i32(i32 addrspace(1)* %gep, i32 42, i32 0, i32 0, i1 false)
  store i32 %result, i32 addrspace(1)* %out
  ret void
}

define amdgpu_kernel void @global_atomic_inc_noret_i32(i32 addrspace(1)* %ptr) nounwind {
; CI-LABEL: global_atomic_inc_noret_i32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v0, s0
; CI-NEXT:    v_mov_b32_e32 v1, s1
; CI-NEXT:    flat_atomic_inc v0, v[0:1], v2 glc
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_noret_i32:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v0, s0
; VI-NEXT:    v_mov_b32_e32 v1, s1
; VI-NEXT:    flat_atomic_inc v0, v[0:1], v2 glc
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_noret_i32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    global_atomic_inc v0, v[0:1], v2, off glc
; GFX9-NEXT:    s_endpgm
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p1i32(i32 addrspace(1)* %ptr, i32 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @global_atomic_inc_noret_i32_offset(i32 addrspace(1)* %ptr) nounwind {
; CI-LABEL: global_atomic_inc_noret_i32_offset:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_add_u32 s0, s0, 16
; CI-NEXT:    s_addc_u32 s1, s1, 0
; CI-NEXT:    v_mov_b32_e32 v0, s0
; CI-NEXT:    v_mov_b32_e32 v1, s1
; CI-NEXT:    flat_atomic_inc v0, v[0:1], v2 glc
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_noret_i32_offset:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    s_add_u32 s0, s0, 16
; VI-NEXT:    s_addc_u32 s1, s1, 0
; VI-NEXT:    v_mov_b32_e32 v0, s0
; VI-NEXT:    v_mov_b32_e32 v1, s1
; VI-NEXT:    flat_atomic_inc v0, v[0:1], v2 glc
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_noret_i32_offset:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_add_u32 s0, s0, 16
; GFX9-NEXT:    s_addc_u32 s1, s1, 0
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    global_atomic_inc v0, v[0:1], v2, off glc
; GFX9-NEXT:    s_endpgm
  %gep = getelementptr i32, i32 addrspace(1)* %ptr, i32 4
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p1i32(i32 addrspace(1)* %gep, i32 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @global_atomic_inc_ret_i32_offset_addr64(i32 addrspace(1)* %out, i32 addrspace(1)* %ptr) #0 {
; CI-LABEL: global_atomic_inc_ret_i32_offset_addr64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; CI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; CI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 2
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v2, s2
; CI-NEXT:    v_mov_b32_e32 v3, s3
; CI-NEXT:    v_add_i32_e32 v4, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v5, vcc, v3, v1, vcc
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    v_add_i32_e32 v0, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; CI-NEXT:    v_add_i32_e32 v2, vcc, 20, v4
; CI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v5, vcc
; CI-NEXT:    v_mov_b32_e32 v4, 42
; CI-NEXT:    flat_atomic_inc v2, v[2:3], v4 glc
; CI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; CI-NEXT:    flat_store_dword v[0:1], v2
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_ret_i32_offset_addr64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; VI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; VI-NEXT:    v_lshlrev_b64 v[0:1], 2, v[0:1]
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v2, s2
; VI-NEXT:    v_mov_b32_e32 v3, s3
; VI-NEXT:    v_add_u32_e32 v4, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v5, vcc, v3, v1, vcc
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    v_add_u32_e32 v0, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; VI-NEXT:    v_add_u32_e32 v2, vcc, 20, v4
; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v5, vcc
; VI-NEXT:    v_mov_b32_e32 v4, 42
; VI-NEXT:    flat_atomic_inc v2, v[2:3], v4 glc
; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; VI-NEXT:    flat_store_dword v[0:1], v2
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_ret_i32_offset_addr64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GFX9-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; GFX9-NEXT:    v_lshlrev_b64 v[0:1], 2, v[0:1]
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v2, s2
; GFX9-NEXT:    v_mov_b32_e32 v3, s3
; GFX9-NEXT:    v_add_co_u32_e32 v4, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v5, vcc, v3, v1, vcc
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, v3, v1, vcc
; GFX9-NEXT:    v_add_co_u32_e32 v2, vcc, 20, v4
; GFX9-NEXT:    v_addc_co_u32_e32 v3, vcc, 0, v5, vcc
; GFX9-NEXT:    v_mov_b32_e32 v4, 42
; GFX9-NEXT:    global_atomic_inc v2, v[2:3], v4, off glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_store_dword v[0:1], v2, off
; GFX9-NEXT:    s_endpgm
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.tid = getelementptr i32, i32 addrspace(1)* %ptr, i32 %id
  %out.gep = getelementptr i32, i32 addrspace(1)* %out, i32 %id
  %gep = getelementptr i32, i32 addrspace(1)* %gep.tid, i32 5
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p1i32(i32 addrspace(1)* %gep, i32 42, i32 0, i32 0, i1 false)
  store i32 %result, i32 addrspace(1)* %out.gep
  ret void
}

define amdgpu_kernel void @global_atomic_inc_noret_i32_offset_addr64(i32 addrspace(1)* %ptr) #0 {
; CI-LABEL: global_atomic_inc_noret_i32_offset_addr64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; CI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 2
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    v_add_i32_e32 v0, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; CI-NEXT:    v_add_i32_e32 v0, vcc, 20, v0
; CI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    flat_atomic_inc v0, v[0:1], v2 glc
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_noret_i32_offset_addr64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; VI-NEXT:    v_lshlrev_b64 v[0:1], 2, v[0:1]
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    v_add_u32_e32 v0, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; VI-NEXT:    v_add_u32_e32 v0, vcc, 20, v0
; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    flat_atomic_inc v0, v[0:1], v2 glc
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_noret_i32_offset_addr64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; GFX9-NEXT:    v_lshlrev_b64 v[0:1], 2, v[0:1]
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, v3, v1, vcc
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, 20, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    global_atomic_inc v0, v[0:1], v2, off glc
; GFX9-NEXT:    s_endpgm
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.tid = getelementptr i32, i32 addrspace(1)* %ptr, i32 %id
  %gep = getelementptr i32, i32 addrspace(1)* %gep.tid, i32 5
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p1i32(i32 addrspace(1)* %gep, i32 42, i32 0, i32 0, i1 false)
  ret void
}

@lds0 = internal addrspace(3) global [512 x i32] undef, align 4

define amdgpu_kernel void @atomic_inc_shl_base_lds_0_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %add_use) #0 {
; CI-LABEL: atomic_inc_shl_base_lds_0_i32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; CI-NEXT:    v_add_i32_e32 v2, vcc, 2, v0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; CI-NEXT:    v_add_i32_e32 v0, vcc, 0, v0
; CI-NEXT:    v_mov_b32_e32 v1, 9
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_inc_rtn_u32 v3, v0, v1 offset:8
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v0, s2
; CI-NEXT:    v_mov_b32_e32 v1, s3
; CI-NEXT:    flat_store_dword v[0:1], v2
; CI-NEXT:    v_mov_b32_e32 v0, s0
; CI-NEXT:    v_mov_b32_e32 v1, s1
; CI-NEXT:    flat_store_dword v[0:1], v3
; CI-NEXT:    s_endpgm
;
; VI-LABEL: atomic_inc_shl_base_lds_0_i32:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; VI-NEXT:    v_add_u32_e32 v2, vcc, 2, v0
; VI-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; VI-NEXT:    v_add_u32_e32 v0, vcc, 0, v0
; VI-NEXT:    v_mov_b32_e32 v1, 9
; VI-NEXT:    s_mov_b32 m0, -1
; VI-NEXT:    ds_inc_rtn_u32 v3, v0, v1 offset:8
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v0, s2
; VI-NEXT:    v_mov_b32_e32 v1, s3
; VI-NEXT:    flat_store_dword v[0:1], v2
; VI-NEXT:    v_mov_b32_e32 v0, s0
; VI-NEXT:    v_mov_b32_e32 v1, s1
; VI-NEXT:    flat_store_dword v[0:1], v3
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: atomic_inc_shl_base_lds_0_i32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GFX9-NEXT:    v_add_u32_e32 v2, 2, v0
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 2, v0
; GFX9-NEXT:    v_add_u32_e32 v0, 0, v0
; GFX9-NEXT:    v_mov_b32_e32 v1, 9
; GFX9-NEXT:    ds_inc_rtn_u32 v3, v0, v1 offset:8
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v0, s2
; GFX9-NEXT:    v_mov_b32_e32 v1, s3
; GFX9-NEXT:    global_store_dword v[0:1], v2, off
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    global_store_dword v[0:1], v3, off
; GFX9-NEXT:    s_endpgm
  %tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %idx.0 = add nsw i32 %tid.x, 2
  %arrayidx0 = getelementptr inbounds [512 x i32], [512 x i32] addrspace(3)* @lds0, i32 0, i32 %idx.0
  %val0 = call i32 @llvm.amdgcn.atomic.inc.i32.p3i32(i32 addrspace(3)* %arrayidx0, i32 9, i32 0, i32 0, i1 false)
  store i32 %idx.0, i32 addrspace(1)* %add_use
  store i32 %val0, i32 addrspace(1)* %out
  ret void
}

define amdgpu_kernel void @lds_atomic_inc_ret_i64(i64 addrspace(1)* %out, i64 addrspace(3)* %ptr) #0 {
; CI-LABEL: lds_atomic_inc_ret_i64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    s_load_dword s2, s[4:5], 0x2
; CI-NEXT:    v_mov_b32_e32 v0, 42
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v2, s2
; CI-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; CI-NEXT:    s_endpgm
;
; VI-LABEL: lds_atomic_inc_ret_i64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    s_load_dword s2, s[4:5], 0x8
; VI-NEXT:    v_mov_b32_e32 v0, 42
; VI-NEXT:    v_mov_b32_e32 v1, 0
; VI-NEXT:    s_mov_b32 m0, -1
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v2, s2
; VI-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: lds_atomic_inc_ret_i64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    s_load_dword s2, s[4:5], 0x8
; GFX9-NEXT:    v_mov_b32_e32 v0, 42
; GFX9-NEXT:    v_mov_b32_e32 v1, 0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v2, s2
; GFX9-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_store_dwordx2 v[2:3], v[0:1], off
; GFX9-NEXT:    s_endpgm
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p3i64(i64 addrspace(3)* %ptr, i64 42, i32 0, i32 0, i1 false)
  store i64 %result, i64 addrspace(1)* %out
  ret void
}

define amdgpu_kernel void @lds_atomic_inc_ret_i64_offset(i64 addrspace(1)* %out, i64 addrspace(3)* %ptr) #0 {
; CI-LABEL: lds_atomic_inc_ret_i64_offset:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    s_load_dword s2, s[4:5], 0x2
; CI-NEXT:    v_mov_b32_e32 v0, 42
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_add_u32 s2, s2, 32
; CI-NEXT:    v_mov_b32_e32 v2, s2
; CI-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; CI-NEXT:    s_endpgm
;
; VI-LABEL: lds_atomic_inc_ret_i64_offset:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    s_load_dword s2, s[4:5], 0x8
; VI-NEXT:    v_mov_b32_e32 v0, 42
; VI-NEXT:    v_mov_b32_e32 v1, 0
; VI-NEXT:    s_mov_b32 m0, -1
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    s_add_u32 s2, s2, 32
; VI-NEXT:    v_mov_b32_e32 v2, s2
; VI-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: lds_atomic_inc_ret_i64_offset:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    s_load_dword s2, s[4:5], 0x8
; GFX9-NEXT:    v_mov_b32_e32 v0, 42
; GFX9-NEXT:    v_mov_b32_e32 v1, 0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_add_u32 s2, s2, 32
; GFX9-NEXT:    v_mov_b32_e32 v2, s2
; GFX9-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_store_dwordx2 v[2:3], v[0:1], off
; GFX9-NEXT:    s_endpgm
  %gep = getelementptr i64, i64 addrspace(3)* %ptr, i32 4
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p3i64(i64 addrspace(3)* %gep, i64 42, i32 0, i32 0, i1 false)
  store i64 %result, i64 addrspace(1)* %out
  ret void
}

define amdgpu_kernel void @lds_atomic_inc_noret_i64(i64 addrspace(3)* %ptr) nounwind {
; CI-LABEL: lds_atomic_inc_noret_i64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dword s0, s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v0, 42
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; CI-NEXT:    s_endpgm
;
; VI-LABEL: lds_atomic_inc_noret_i64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dword s0, s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v0, 42
; VI-NEXT:    v_mov_b32_e32 v1, 0
; VI-NEXT:    s_mov_b32 m0, -1
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: lds_atomic_inc_noret_i64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v0, 42
; GFX9-NEXT:    v_mov_b32_e32 v1, 0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; GFX9-NEXT:    s_endpgm
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p3i64(i64 addrspace(3)* %ptr, i64 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @lds_atomic_inc_noret_i64_offset(i64 addrspace(3)* %ptr) nounwind {
; CI-LABEL: lds_atomic_inc_noret_i64_offset:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dword s0, s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v0, 42
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_add_u32 s0, s0, 32
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; CI-NEXT:    s_endpgm
;
; VI-LABEL: lds_atomic_inc_noret_i64_offset:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dword s0, s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v0, 42
; VI-NEXT:    v_mov_b32_e32 v1, 0
; VI-NEXT:    s_mov_b32 m0, -1
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    s_add_u32 s0, s0, 32
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: lds_atomic_inc_noret_i64_offset:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dword s0, s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v0, 42
; GFX9-NEXT:    v_mov_b32_e32 v1, 0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_add_u32 s0, s0, 32
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1]
; GFX9-NEXT:    s_endpgm
  %gep = getelementptr i64, i64 addrspace(3)* %ptr, i32 4
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p3i64(i64 addrspace(3)* %gep, i64 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @global_atomic_inc_ret_i64(i64 addrspace(1)* %out, i64 addrspace(1)* %ptr) #0 {
; CI-LABEL: global_atomic_inc_ret_i64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    v_mov_b32_e32 v3, 0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v0, s2
; CI-NEXT:    v_mov_b32_e32 v1, s3
; CI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; CI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_ret_i64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    v_mov_b32_e32 v3, 0
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v0, s2
; VI-NEXT:    v_mov_b32_e32 v1, s3
; VI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; VI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_ret_i64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    v_mov_b32_e32 v3, 0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v0, s2
; GFX9-NEXT:    v_mov_b32_e32 v1, s3
; GFX9-NEXT:    global_atomic_inc_x2 v[0:1], v[0:1], v[2:3], off glc
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_store_dwordx2 v[2:3], v[0:1], off
; GFX9-NEXT:    s_endpgm
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p1i64(i64 addrspace(1)* %ptr, i64 42, i32 0, i32 0, i1 false)
  store i64 %result, i64 addrspace(1)* %out
  ret void
}

define amdgpu_kernel void @global_atomic_inc_ret_i64_offset(i64 addrspace(1)* %out, i64 addrspace(1)* %ptr) #0 {
; CI-LABEL: global_atomic_inc_ret_i64_offset:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    v_mov_b32_e32 v3, 0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_add_u32 s2, s2, 32
; CI-NEXT:    s_addc_u32 s3, s3, 0
; CI-NEXT:    v_mov_b32_e32 v0, s2
; CI-NEXT:    v_mov_b32_e32 v1, s3
; CI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; CI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_ret_i64_offset:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    v_mov_b32_e32 v3, 0
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    s_add_u32 s2, s2, 32
; VI-NEXT:    s_addc_u32 s3, s3, 0
; VI-NEXT:    v_mov_b32_e32 v0, s2
; VI-NEXT:    v_mov_b32_e32 v1, s3
; VI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; VI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_ret_i64_offset:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    v_mov_b32_e32 v3, 0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_add_u32 s2, s2, 32
; GFX9-NEXT:    s_addc_u32 s3, s3, 0
; GFX9-NEXT:    v_mov_b32_e32 v0, s2
; GFX9-NEXT:    v_mov_b32_e32 v1, s3
; GFX9-NEXT:    global_atomic_inc_x2 v[0:1], v[0:1], v[2:3], off glc
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_store_dwordx2 v[2:3], v[0:1], off
; GFX9-NEXT:    s_endpgm
  %gep = getelementptr i64, i64 addrspace(1)* %ptr, i32 4
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p1i64(i64 addrspace(1)* %gep, i64 42, i32 0, i32 0, i1 false)
  store i64 %result, i64 addrspace(1)* %out
  ret void
}

define amdgpu_kernel void @global_atomic_inc_noret_i64(i64 addrspace(1)* %ptr) nounwind {
; CI-LABEL: global_atomic_inc_noret_i64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    v_mov_b32_e32 v3, 0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v0, s0
; CI-NEXT:    v_mov_b32_e32 v1, s1
; CI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_noret_i64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    v_mov_b32_e32 v3, 0
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v0, s0
; VI-NEXT:    v_mov_b32_e32 v1, s1
; VI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_noret_i64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    v_mov_b32_e32 v3, 0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    global_atomic_inc_x2 v[0:1], v[0:1], v[2:3], off glc
; GFX9-NEXT:    s_endpgm
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p1i64(i64 addrspace(1)* %ptr, i64 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @global_atomic_inc_noret_i64_offset(i64 addrspace(1)* %ptr) nounwind {
; CI-LABEL: global_atomic_inc_noret_i64_offset:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    v_mov_b32_e32 v3, 0
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    s_add_u32 s0, s0, 32
; CI-NEXT:    s_addc_u32 s1, s1, 0
; CI-NEXT:    v_mov_b32_e32 v0, s0
; CI-NEXT:    v_mov_b32_e32 v1, s1
; CI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_noret_i64_offset:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    v_mov_b32_e32 v3, 0
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    s_add_u32 s0, s0, 32
; VI-NEXT:    s_addc_u32 s1, s1, 0
; VI-NEXT:    v_mov_b32_e32 v0, s0
; VI-NEXT:    v_mov_b32_e32 v1, s1
; VI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_noret_i64_offset:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    v_mov_b32_e32 v3, 0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_add_u32 s0, s0, 32
; GFX9-NEXT:    s_addc_u32 s1, s1, 0
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    global_atomic_inc_x2 v[0:1], v[0:1], v[2:3], off glc
; GFX9-NEXT:    s_endpgm
  %gep = getelementptr i64, i64 addrspace(1)* %ptr, i32 4
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p1i64(i64 addrspace(1)* %gep, i64 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @global_atomic_inc_ret_i64_offset_addr64(i64 addrspace(1)* %out, i64 addrspace(1)* %ptr) #0 {
; CI-LABEL: global_atomic_inc_ret_i64_offset_addr64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; CI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; CI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 3
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v2, s2
; CI-NEXT:    v_mov_b32_e32 v3, s3
; CI-NEXT:    v_add_i32_e32 v4, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v5, vcc, v3, v1, vcc
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    v_add_i32_e32 v0, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; CI-NEXT:    v_add_i32_e32 v2, vcc, 40, v4
; CI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v5, vcc
; CI-NEXT:    v_mov_b32_e32 v4, 42
; CI-NEXT:    v_mov_b32_e32 v5, 0
; CI-NEXT:    flat_atomic_inc_x2 v[2:3], v[2:3], v[4:5] glc
; CI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; CI-NEXT:    flat_store_dwordx2 v[0:1], v[2:3]
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_ret_i64_offset_addr64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; VI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; VI-NEXT:    v_lshlrev_b64 v[0:1], 3, v[0:1]
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v2, s2
; VI-NEXT:    v_mov_b32_e32 v3, s3
; VI-NEXT:    v_add_u32_e32 v4, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v5, vcc, v3, v1, vcc
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    v_add_u32_e32 v0, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; VI-NEXT:    v_add_u32_e32 v2, vcc, 40, v4
; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v5, vcc
; VI-NEXT:    v_mov_b32_e32 v4, 42
; VI-NEXT:    v_mov_b32_e32 v5, 0
; VI-NEXT:    flat_atomic_inc_x2 v[2:3], v[2:3], v[4:5] glc
; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; VI-NEXT:    flat_store_dwordx2 v[0:1], v[2:3]
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_ret_i64_offset_addr64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GFX9-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; GFX9-NEXT:    v_lshlrev_b64 v[0:1], 3, v[0:1]
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v2, s2
; GFX9-NEXT:    v_mov_b32_e32 v3, s3
; GFX9-NEXT:    v_add_co_u32_e32 v4, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v5, vcc, v3, v1, vcc
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, v3, v1, vcc
; GFX9-NEXT:    v_add_co_u32_e32 v2, vcc, 40, v4
; GFX9-NEXT:    v_addc_co_u32_e32 v3, vcc, 0, v5, vcc
; GFX9-NEXT:    v_mov_b32_e32 v4, 42
; GFX9-NEXT:    v_mov_b32_e32 v5, 0
; GFX9-NEXT:    global_atomic_inc_x2 v[2:3], v[2:3], v[4:5], off glc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    global_store_dwordx2 v[0:1], v[2:3], off
; GFX9-NEXT:    s_endpgm
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.tid = getelementptr i64, i64 addrspace(1)* %ptr, i32 %id
  %out.gep = getelementptr i64, i64 addrspace(1)* %out, i32 %id
  %gep = getelementptr i64, i64 addrspace(1)* %gep.tid, i32 5
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p1i64(i64 addrspace(1)* %gep, i64 42, i32 0, i32 0, i1 false)
  store i64 %result, i64 addrspace(1)* %out.gep
  ret void
}

define amdgpu_kernel void @global_atomic_inc_noret_i64_offset_addr64(i64 addrspace(1)* %ptr) #0 {
; CI-LABEL: global_atomic_inc_noret_i64_offset_addr64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; CI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 3
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    v_add_i32_e32 v0, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; CI-NEXT:    v_add_i32_e32 v0, vcc, 40, v0
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; CI-NEXT:    v_mov_b32_e32 v3, 0
; CI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; CI-NEXT:    s_endpgm
;
; VI-LABEL: global_atomic_inc_noret_i64_offset_addr64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; VI-NEXT:    v_lshlrev_b64 v[0:1], 3, v[0:1]
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    v_add_u32_e32 v0, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; VI-NEXT:    v_add_u32_e32 v0, vcc, 40, v0
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; VI-NEXT:    v_mov_b32_e32 v3, 0
; VI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: global_atomic_inc_noret_i64_offset_addr64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; GFX9-NEXT:    v_lshlrev_b64 v[0:1], 3, v[0:1]
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, v3, v1, vcc
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, 40, v0
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
; GFX9-NEXT:    v_mov_b32_e32 v3, 0
; GFX9-NEXT:    global_atomic_inc_x2 v[0:1], v[0:1], v[2:3], off glc
; GFX9-NEXT:    s_endpgm
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.tid = getelementptr i64, i64 addrspace(1)* %ptr, i32 %id
  %gep = getelementptr i64, i64 addrspace(1)* %gep.tid, i32 5
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p1i64(i64 addrspace(1)* %gep, i64 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_ret_i32(i32* %out, i32* %ptr) #0 {
; GCN-LABEL: flat_atomic_inc_ret_i32:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GCN-NEXT:    v_mov_b32_e32 v2, 42
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    v_mov_b32_e32 v0, s2
; GCN-NEXT:    v_mov_b32_e32 v1, s3
; GCN-NEXT:    flat_atomic_inc v2, v[0:1], v2 glc
; GCN-NEXT:    v_mov_b32_e32 v0, s0
; GCN-NEXT:    v_mov_b32_e32 v1, s1
; GCN-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT:    flat_store_dword v[0:1], v2
; GCN-NEXT:    s_endpgm
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p0i32(i32* %ptr, i32 42, i32 0, i32 0, i1 false)
  store i32 %result, i32* %out
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_ret_i32_offset(i32* %out, i32* %ptr) #0 {
; GCN-LABEL: flat_atomic_inc_ret_i32_offset:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GCN-NEXT:    v_mov_b32_e32 v2, 42
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    s_add_u32 s2, s2, 16
; GCN-NEXT:    s_addc_u32 s3, s3, 0
; GCN-NEXT:    v_mov_b32_e32 v0, s2
; GCN-NEXT:    v_mov_b32_e32 v1, s3
; GCN-NEXT:    flat_atomic_inc v2, v[0:1], v2 glc
; GCN-NEXT:    v_mov_b32_e32 v0, s0
; GCN-NEXT:    v_mov_b32_e32 v1, s1
; GCN-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT:    flat_store_dword v[0:1], v2
; GCN-NEXT:    s_endpgm
  %gep = getelementptr i32, i32* %ptr, i32 4
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p0i32(i32* %gep, i32 42, i32 0, i32 0, i1 false)
  store i32 %result, i32* %out
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_noret_i32(i32* %ptr) nounwind {
; GCN-LABEL: flat_atomic_inc_noret_i32:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GCN-NEXT:    v_mov_b32_e32 v2, 42
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    v_mov_b32_e32 v0, s0
; GCN-NEXT:    v_mov_b32_e32 v1, s1
; GCN-NEXT:    flat_atomic_inc v0, v[0:1], v2 glc
; GCN-NEXT:    s_endpgm
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p0i32(i32* %ptr, i32 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_noret_i32_offset(i32* %ptr) nounwind {
; GCN-LABEL: flat_atomic_inc_noret_i32_offset:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GCN-NEXT:    v_mov_b32_e32 v2, 42
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    s_add_u32 s0, s0, 16
; GCN-NEXT:    s_addc_u32 s1, s1, 0
; GCN-NEXT:    v_mov_b32_e32 v0, s0
; GCN-NEXT:    v_mov_b32_e32 v1, s1
; GCN-NEXT:    flat_atomic_inc v0, v[0:1], v2 glc
; GCN-NEXT:    s_endpgm
  %gep = getelementptr i32, i32* %ptr, i32 4
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p0i32(i32* %gep, i32 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_ret_i32_offset_addr64(i32* %out, i32* %ptr) #0 {
; CI-LABEL: flat_atomic_inc_ret_i32_offset_addr64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; CI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; CI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 2
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v2, s2
; CI-NEXT:    v_mov_b32_e32 v3, s3
; CI-NEXT:    v_add_i32_e32 v4, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v5, vcc, v3, v1, vcc
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    v_add_i32_e32 v0, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; CI-NEXT:    v_add_i32_e32 v2, vcc, 20, v4
; CI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v5, vcc
; CI-NEXT:    v_mov_b32_e32 v4, 42
; CI-NEXT:    flat_atomic_inc v2, v[2:3], v4 glc
; CI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; CI-NEXT:    flat_store_dword v[0:1], v2
; CI-NEXT:    s_endpgm
;
; VI-LABEL: flat_atomic_inc_ret_i32_offset_addr64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; VI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; VI-NEXT:    v_lshlrev_b64 v[0:1], 2, v[0:1]
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v2, s2
; VI-NEXT:    v_mov_b32_e32 v3, s3
; VI-NEXT:    v_add_u32_e32 v4, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v5, vcc, v3, v1, vcc
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    v_add_u32_e32 v0, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; VI-NEXT:    v_add_u32_e32 v2, vcc, 20, v4
; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v5, vcc
; VI-NEXT:    v_mov_b32_e32 v4, 42
; VI-NEXT:    flat_atomic_inc v2, v[2:3], v4 glc
; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; VI-NEXT:    flat_store_dword v[0:1], v2
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: flat_atomic_inc_ret_i32_offset_addr64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GFX9-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; GFX9-NEXT:    v_lshlrev_b64 v[0:1], 2, v[0:1]
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v2, s2
; GFX9-NEXT:    v_mov_b32_e32 v3, s3
; GFX9-NEXT:    v_add_co_u32_e32 v4, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v5, vcc, v3, v1, vcc
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, v3, v1, vcc
; GFX9-NEXT:    v_add_co_u32_e32 v2, vcc, 20, v4
; GFX9-NEXT:    v_addc_co_u32_e32 v3, vcc, 0, v5, vcc
; GFX9-NEXT:    v_mov_b32_e32 v4, 42
; GFX9-NEXT:    flat_atomic_inc v2, v[2:3], v4 glc
; GFX9-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; GFX9-NEXT:    flat_store_dword v[0:1], v2
; GFX9-NEXT:    s_endpgm
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.tid = getelementptr i32, i32* %ptr, i32 %id
  %out.gep = getelementptr i32, i32* %out, i32 %id
  %gep = getelementptr i32, i32* %gep.tid, i32 5
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p0i32(i32* %gep, i32 42, i32 0, i32 0, i1 false)
  store i32 %result, i32* %out.gep
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_noret_i32_offset_addr64(i32* %ptr) #0 {
; CI-LABEL: flat_atomic_inc_noret_i32_offset_addr64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; CI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 2
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    v_add_i32_e32 v0, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; CI-NEXT:    v_add_i32_e32 v0, vcc, 20, v0
; CI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    flat_atomic_inc v0, v[0:1], v2 glc
; CI-NEXT:    s_endpgm
;
; VI-LABEL: flat_atomic_inc_noret_i32_offset_addr64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; VI-NEXT:    v_lshlrev_b64 v[0:1], 2, v[0:1]
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    v_add_u32_e32 v0, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; VI-NEXT:    v_add_u32_e32 v0, vcc, 20, v0
; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    flat_atomic_inc v0, v[0:1], v2 glc
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: flat_atomic_inc_noret_i32_offset_addr64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; GFX9-NEXT:    v_lshlrev_b64 v[0:1], 2, v[0:1]
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, v3, v1, vcc
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, 20, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    flat_atomic_inc v0, v[0:1], v2 glc
; GFX9-NEXT:    s_endpgm
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.tid = getelementptr i32, i32* %ptr, i32 %id
  %gep = getelementptr i32, i32* %gep.tid, i32 5
  %result = call i32 @llvm.amdgcn.atomic.inc.i32.p0i32(i32* %gep, i32 42, i32 0, i32 0, i1 false)
  ret void
}

@lds1 = internal addrspace(3) global [512 x i64] undef, align 8

define amdgpu_kernel void @atomic_inc_shl_base_lds_0_i64(i64 addrspace(1)* %out, i32 addrspace(1)* %add_use) #0 {
; CI-LABEL: atomic_inc_shl_base_lds_0_i64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; CI-NEXT:    v_add_i32_e32 v4, vcc, 2, v0
; CI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; CI-NEXT:    v_add_i32_e32 v2, vcc, 0, v0
; CI-NEXT:    v_mov_b32_e32 v0, 9
; CI-NEXT:    v_mov_b32_e32 v1, 0
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1] offset:16
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v2, s2
; CI-NEXT:    v_mov_b32_e32 v3, s3
; CI-NEXT:    flat_store_dword v[2:3], v4
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; CI-NEXT:    s_endpgm
;
; VI-LABEL: atomic_inc_shl_base_lds_0_i64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; VI-NEXT:    v_add_u32_e32 v4, vcc, 2, v0
; VI-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; VI-NEXT:    v_add_u32_e32 v2, vcc, 0, v0
; VI-NEXT:    v_mov_b32_e32 v0, 9
; VI-NEXT:    v_mov_b32_e32 v1, 0
; VI-NEXT:    s_mov_b32 m0, -1
; VI-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1] offset:16
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v2, s2
; VI-NEXT:    v_mov_b32_e32 v3, s3
; VI-NEXT:    flat_store_dword v[2:3], v4
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: atomic_inc_shl_base_lds_0_i64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GFX9-NEXT:    v_add_u32_e32 v4, 2, v0
; GFX9-NEXT:    v_lshlrev_b32_e32 v0, 3, v0
; GFX9-NEXT:    v_add_u32_e32 v2, 0, v0
; GFX9-NEXT:    v_mov_b32_e32 v0, 9
; GFX9-NEXT:    v_mov_b32_e32 v1, 0
; GFX9-NEXT:    ds_inc_rtn_u64 v[0:1], v2, v[0:1] offset:16
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v2, s2
; GFX9-NEXT:    v_mov_b32_e32 v3, s3
; GFX9-NEXT:    global_store_dword v[2:3], v4, off
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    global_store_dwordx2 v[2:3], v[0:1], off
; GFX9-NEXT:    s_endpgm
  %tid.x = tail call i32 @llvm.amdgcn.workitem.id.x() #1
  %idx.0 = add nsw i32 %tid.x, 2
  %arrayidx0 = getelementptr inbounds [512 x i64], [512 x i64] addrspace(3)* @lds1, i32 0, i32 %idx.0
  %val0 = call i64 @llvm.amdgcn.atomic.inc.i64.p3i64(i64 addrspace(3)* %arrayidx0, i64 9, i32 0, i32 0, i1 false)
  store i32 %idx.0, i32 addrspace(1)* %add_use
  store i64 %val0, i64 addrspace(1)* %out
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_ret_i64(i64* %out, i64* %ptr) #0 {
; GCN-LABEL: flat_atomic_inc_ret_i64:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GCN-NEXT:    v_mov_b32_e32 v2, 42
; GCN-NEXT:    v_mov_b32_e32 v3, 0
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    v_mov_b32_e32 v0, s2
; GCN-NEXT:    v_mov_b32_e32 v1, s3
; GCN-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; GCN-NEXT:    v_mov_b32_e32 v3, s1
; GCN-NEXT:    v_mov_b32_e32 v2, s0
; GCN-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; GCN-NEXT:    s_endpgm
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p0i64(i64* %ptr, i64 42, i32 0, i32 0, i1 false)
  store i64 %result, i64* %out
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_ret_i64_offset(i64* %out, i64* %ptr) #0 {
; GCN-LABEL: flat_atomic_inc_ret_i64_offset:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GCN-NEXT:    v_mov_b32_e32 v2, 42
; GCN-NEXT:    v_mov_b32_e32 v3, 0
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    s_add_u32 s2, s2, 32
; GCN-NEXT:    s_addc_u32 s3, s3, 0
; GCN-NEXT:    v_mov_b32_e32 v0, s2
; GCN-NEXT:    v_mov_b32_e32 v1, s3
; GCN-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; GCN-NEXT:    v_mov_b32_e32 v3, s1
; GCN-NEXT:    v_mov_b32_e32 v2, s0
; GCN-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; GCN-NEXT:    flat_store_dwordx2 v[2:3], v[0:1]
; GCN-NEXT:    s_endpgm
  %gep = getelementptr i64, i64* %ptr, i32 4
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p0i64(i64* %gep, i64 42, i32 0, i32 0, i1 false)
  store i64 %result, i64* %out
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_noret_i64(i64* %ptr) nounwind {
; GCN-LABEL: flat_atomic_inc_noret_i64:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GCN-NEXT:    v_mov_b32_e32 v2, 42
; GCN-NEXT:    v_mov_b32_e32 v3, 0
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    v_mov_b32_e32 v0, s0
; GCN-NEXT:    v_mov_b32_e32 v1, s1
; GCN-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; GCN-NEXT:    s_endpgm
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p0i64(i64* %ptr, i64 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_noret_i64_offset(i64* %ptr) nounwind {
; GCN-LABEL: flat_atomic_inc_noret_i64_offset:
; GCN:       ; %bb.0:
; GCN-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GCN-NEXT:    v_mov_b32_e32 v2, 42
; GCN-NEXT:    v_mov_b32_e32 v3, 0
; GCN-NEXT:    s_waitcnt lgkmcnt(0)
; GCN-NEXT:    s_add_u32 s0, s0, 32
; GCN-NEXT:    s_addc_u32 s1, s1, 0
; GCN-NEXT:    v_mov_b32_e32 v0, s0
; GCN-NEXT:    v_mov_b32_e32 v1, s1
; GCN-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; GCN-NEXT:    s_endpgm
  %gep = getelementptr i64, i64* %ptr, i32 4
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p0i64(i64* %gep, i64 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_ret_i64_offset_addr64(i64* %out, i64* %ptr) #0 {
; CI-LABEL: flat_atomic_inc_ret_i64_offset_addr64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; CI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; CI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 3
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v2, s2
; CI-NEXT:    v_mov_b32_e32 v3, s3
; CI-NEXT:    v_add_i32_e32 v4, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v5, vcc, v3, v1, vcc
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    v_add_i32_e32 v0, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; CI-NEXT:    v_add_i32_e32 v2, vcc, 40, v4
; CI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v5, vcc
; CI-NEXT:    v_mov_b32_e32 v4, 42
; CI-NEXT:    v_mov_b32_e32 v5, 0
; CI-NEXT:    flat_atomic_inc_x2 v[2:3], v[2:3], v[4:5] glc
; CI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; CI-NEXT:    flat_store_dwordx2 v[0:1], v[2:3]
; CI-NEXT:    s_endpgm
;
; VI-LABEL: flat_atomic_inc_ret_i64_offset_addr64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; VI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; VI-NEXT:    v_lshlrev_b64 v[0:1], 3, v[0:1]
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v2, s2
; VI-NEXT:    v_mov_b32_e32 v3, s3
; VI-NEXT:    v_add_u32_e32 v4, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v5, vcc, v3, v1, vcc
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    v_add_u32_e32 v0, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; VI-NEXT:    v_add_u32_e32 v2, vcc, 40, v4
; VI-NEXT:    v_addc_u32_e32 v3, vcc, 0, v5, vcc
; VI-NEXT:    v_mov_b32_e32 v4, 42
; VI-NEXT:    v_mov_b32_e32 v5, 0
; VI-NEXT:    flat_atomic_inc_x2 v[2:3], v[2:3], v[4:5] glc
; VI-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; VI-NEXT:    flat_store_dwordx2 v[0:1], v[2:3]
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: flat_atomic_inc_ret_i64_offset_addr64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GFX9-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; GFX9-NEXT:    v_lshlrev_b64 v[0:1], 3, v[0:1]
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v2, s2
; GFX9-NEXT:    v_mov_b32_e32 v3, s3
; GFX9-NEXT:    v_add_co_u32_e32 v4, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v5, vcc, v3, v1, vcc
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, v3, v1, vcc
; GFX9-NEXT:    v_add_co_u32_e32 v2, vcc, 40, v4
; GFX9-NEXT:    v_addc_co_u32_e32 v3, vcc, 0, v5, vcc
; GFX9-NEXT:    v_mov_b32_e32 v4, 42
; GFX9-NEXT:    v_mov_b32_e32 v5, 0
; GFX9-NEXT:    flat_atomic_inc_x2 v[2:3], v[2:3], v[4:5] glc
; GFX9-NEXT:    s_waitcnt vmcnt(0) lgkmcnt(0)
; GFX9-NEXT:    flat_store_dwordx2 v[0:1], v[2:3]
; GFX9-NEXT:    s_endpgm
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.tid = getelementptr i64, i64* %ptr, i32 %id
  %out.gep = getelementptr i64, i64* %out, i32 %id
  %gep = getelementptr i64, i64* %gep.tid, i32 5
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p0i64(i64* %gep, i64 42, i32 0, i32 0, i1 false)
  store i64 %result, i64* %out.gep
  ret void
}

define amdgpu_kernel void @flat_atomic_inc_noret_i64_offset_addr64(i64* %ptr) #0 {
; CI-LABEL: flat_atomic_inc_noret_i64_offset_addr64:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; CI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; CI-NEXT:    v_lshl_b64 v[0:1], v[0:1], 3
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    v_add_i32_e32 v0, vcc, v2, v0
; CI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; CI-NEXT:    v_add_i32_e32 v0, vcc, 40, v0
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; CI-NEXT:    v_mov_b32_e32 v3, 0
; CI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; CI-NEXT:    s_endpgm
;
; VI-LABEL: flat_atomic_inc_noret_i64_offset_addr64:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; VI-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; VI-NEXT:    v_lshlrev_b64 v[0:1], 3, v[0:1]
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    v_add_u32_e32 v0, vcc, v2, v0
; VI-NEXT:    v_addc_u32_e32 v1, vcc, v3, v1, vcc
; VI-NEXT:    v_add_u32_e32 v0, vcc, 40, v0
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
; VI-NEXT:    v_mov_b32_e32 v3, 0
; VI-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: flat_atomic_inc_noret_i64_offset_addr64:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[4:5], 0x0
; GFX9-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; GFX9-NEXT:    v_lshlrev_b64 v[0:1], 3, v[0:1]
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v3, s1
; GFX9-NEXT:    v_mov_b32_e32 v2, s0
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, v2, v0
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, v3, v1, vcc
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, 40, v0
; GFX9-NEXT:    v_mov_b32_e32 v2, 42
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, 0, v1, vcc
; GFX9-NEXT:    v_mov_b32_e32 v3, 0
; GFX9-NEXT:    flat_atomic_inc_x2 v[0:1], v[0:1], v[2:3] glc
; GFX9-NEXT:    s_endpgm
  %id = call i32 @llvm.amdgcn.workitem.id.x()
  %gep.tid = getelementptr i64, i64* %ptr, i32 %id
  %gep = getelementptr i64, i64* %gep.tid, i32 5
  %result = call i64 @llvm.amdgcn.atomic.inc.i64.p0i64(i64* %gep, i64 42, i32 0, i32 0, i1 false)
  ret void
}

define amdgpu_kernel void @nocse_lds_atomic_inc_ret_i32(i32 addrspace(1)* %out0, i32 addrspace(1)* %out1, i32 addrspace(3)* %ptr) #0 {
; CI-LABEL: nocse_lds_atomic_inc_ret_i32:
; CI:       ; %bb.0:
; CI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; CI-NEXT:    s_load_dword s4, s[4:5], 0x4
; CI-NEXT:    v_mov_b32_e32 v2, 42
; CI-NEXT:    s_mov_b32 m0, -1
; CI-NEXT:    s_waitcnt lgkmcnt(0)
; CI-NEXT:    v_mov_b32_e32 v0, s2
; CI-NEXT:    v_mov_b32_e32 v3, s4
; CI-NEXT:    ds_inc_rtn_u32 v4, v3, v2
; CI-NEXT:    ds_inc_rtn_u32 v5, v3, v2
; CI-NEXT:    v_mov_b32_e32 v3, s1
; CI-NEXT:    v_mov_b32_e32 v2, s0
; CI-NEXT:    v_mov_b32_e32 v1, s3
; CI-NEXT:    s_waitcnt lgkmcnt(1)
; CI-NEXT:    flat_store_dword v[2:3], v4
; CI-NEXT:    s_waitcnt lgkmcnt(1)
; CI-NEXT:    flat_store_dword v[0:1], v5
; CI-NEXT:    s_endpgm
;
; VI-LABEL: nocse_lds_atomic_inc_ret_i32:
; VI:       ; %bb.0:
; VI-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; VI-NEXT:    s_load_dword s4, s[4:5], 0x10
; VI-NEXT:    v_mov_b32_e32 v2, 42
; VI-NEXT:    s_mov_b32 m0, -1
; VI-NEXT:    s_waitcnt lgkmcnt(0)
; VI-NEXT:    v_mov_b32_e32 v0, s2
; VI-NEXT:    v_mov_b32_e32 v3, s4
; VI-NEXT:    ds_inc_rtn_u32 v4, v3, v2
; VI-NEXT:    ds_inc_rtn_u32 v5, v3, v2
; VI-NEXT:    v_mov_b32_e32 v3, s1
; VI-NEXT:    v_mov_b32_e32 v2, s0
; VI-NEXT:    v_mov_b32_e32 v1, s3
; VI-NEXT:    s_waitcnt lgkmcnt(1)
; VI-NEXT:    flat_store_dword v[2:3], v4
; VI-NEXT:    s_waitcnt lgkmcnt(1)
; VI-NEXT:    flat_store_dword v[0:1], v5
; VI-NEXT:    s_endpgm
;
; GFX9-LABEL: nocse_lds_atomic_inc_ret_i32:
; GFX9:       ; %bb.0:
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[4:5], 0x0
; GFX9-NEXT:    s_load_dword s4, s[4:5], 0x10
; GFX9-NEXT:    v_mov_b32_e32 v0, 42
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    v_mov_b32_e32 v2, s2
; GFX9-NEXT:    v_mov_b32_e32 v1, s4
; GFX9-NEXT:    ds_inc_rtn_u32 v4, v1, v0
; GFX9-NEXT:    ds_inc_rtn_u32 v5, v1, v0
; GFX9-NEXT:    v_mov_b32_e32 v0, s0
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    v_mov_b32_e32 v3, s3
; GFX9-NEXT:    s_waitcnt lgkmcnt(1)
; GFX9-NEXT:    global_store_dword v[0:1], v4, off
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    global_store_dword v[2:3], v5, off
; GFX9-NEXT:    s_endpgm
  %result0 = call i32 @llvm.amdgcn.atomic.inc.i32.p3i32(i32 addrspace(3)* %ptr, i32 42, i32 0, i32 0, i1 false)
  %result1 = call i32 @llvm.amdgcn.atomic.inc.i32.p3i32(i32 addrspace(3)* %ptr, i32 42, i32 0, i32 0, i1 false)

  store i32 %result0, i32 addrspace(1)* %out0
  store i32 %result1, i32 addrspace(1)* %out1
  ret void
}

attributes #0 = { nounwind }
attributes #1 = { nounwind readnone }
attributes #2 = { nounwind argmemonly }