; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
; RUN: llc -verify-machineinstrs -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 %s -o - | FileCheck -check-prefix=GFX9 %s
; RUN: llc -verify-machineinstrs -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1030 %s -o - | FileCheck -check-prefix=GFX10 %s
; RUN: llc -verify-machineinstrs -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1100 %s -o - | FileCheck -check-prefix=GFX11 %s

define amdgpu_kernel void @test(ptr addrspace(1) %src, ptr addrspace(1) %dst) {
; GFX9-LABEL: test:
; GFX9:       ; %bb.0: ; %entry
; GFX9-NEXT:    s_load_dword s4, s[8:9], 0x1c
; GFX9-NEXT:    s_load_dword s5, s[8:9], 0x38
; GFX9-NEXT:    s_load_dwordx4 s[0:3], s[8:9], 0x0
; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
; GFX9-NEXT:    s_and_b32 s4, s4, 0xffff
; GFX9-NEXT:    s_mul_i32 s12, s12, s4
; GFX9-NEXT:    s_add_i32 s5, s5, s12
; GFX9-NEXT:    v_add_u32_e32 v0, s5, v0
; GFX9-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; GFX9-NEXT:    v_lshlrev_b64 v[4:5], 4, v[0:1]
; GFX9-NEXT:    v_mov_b32_e32 v1, s1
; GFX9-NEXT:    v_add_co_u32_e32 v0, vcc, s0, v4
; GFX9-NEXT:    v_addc_co_u32_e32 v1, vcc, v1, v5, vcc
; GFX9-NEXT:    global_load_dwordx4 v[0:3], v[0:1], off
; GFX9-NEXT:    v_mov_b32_e32 v6, s3
; GFX9-NEXT:    v_add_co_u32_e32 v4, vcc, s2, v4
; GFX9-NEXT:    v_addc_co_u32_e32 v5, vcc, v6, v5, vcc
; GFX9-NEXT:    s_waitcnt vmcnt(0)
; GFX9-NEXT:    v_not_b32_e32 v3, v3
; GFX9-NEXT:    v_not_b32_e32 v2, v2
; GFX9-NEXT:    v_not_b32_e32 v1, v1
; GFX9-NEXT:    v_not_b32_e32 v0, v0
; GFX9-NEXT:    global_store_dwordx4 v[4:5], v[0:3], off
; GFX9-NEXT:    s_endpgm
;
; GFX10-LABEL: test:
; GFX10:       ; %bb.0: ; %entry
; GFX10-NEXT:    s_clause 0x2
; GFX10-NEXT:    s_load_dword s4, s[8:9], 0x1c
; GFX10-NEXT:    s_load_dword s5, s[8:9], 0x38
; GFX10-NEXT:    s_load_dwordx4 s[0:3], s[8:9], 0x0
; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
; GFX10-NEXT:    s_and_b32 s4, s4, 0xffff
; GFX10-NEXT:    s_mul_i32 s12, s12, s4
; GFX10-NEXT:    v_add3_u32 v0, s5, s12, v0
; GFX10-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; GFX10-NEXT:    v_lshlrev_b64 v[4:5], 4, v[0:1]
; GFX10-NEXT:    v_add_co_u32 v0, vcc_lo, s0, v4
; GFX10-NEXT:    v_add_co_ci_u32_e32 v1, vcc_lo, s1, v5, vcc_lo
; GFX10-NEXT:    v_add_co_u32 v4, vcc_lo, s2, v4
; GFX10-NEXT:    v_add_co_ci_u32_e32 v5, vcc_lo, s3, v5, vcc_lo
; GFX10-NEXT:    global_load_dwordx4 v[0:3], v[0:1], off
; GFX10-NEXT:    s_waitcnt vmcnt(0)
; GFX10-NEXT:    v_not_b32_e32 v3, v3
; GFX10-NEXT:    v_not_b32_e32 v2, v2
; GFX10-NEXT:    v_not_b32_e32 v1, v1
; GFX10-NEXT:    v_not_b32_e32 v0, v0
; GFX10-NEXT:    global_store_dwordx4 v[4:5], v[0:3], off
; GFX10-NEXT:    s_endpgm
;
; GFX11-LABEL: test:
; GFX11:       ; %bb.0: ; %entry
; GFX11-NEXT:    s_clause 0x2
; GFX11-NEXT:    s_load_b32 s6, s[4:5], 0x1c
; GFX11-NEXT:    s_load_b32 s7, s[4:5], 0x38
; GFX11-NEXT:    s_load_b128 s[0:3], s[4:5], 0x0
; GFX11-NEXT:    v_and_b32_e32 v0, 0x3ff, v0
; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
; GFX11-NEXT:    s_and_b32 s4, s6, 0xffff
; GFX11-NEXT:    s_delay_alu instid0(SALU_CYCLE_1)
; GFX11-NEXT:    s_mul_i32 s13, s13, s4
; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instid1(SALU_CYCLE_1)
; GFX11-NEXT:    v_add3_u32 v0, s7, s13, v0
; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_1)
; GFX11-NEXT:    v_ashrrev_i32_e32 v1, 31, v0
; GFX11-NEXT:    v_lshlrev_b64 v[4:5], 4, v[0:1]
; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_2)
; GFX11-NEXT:    v_add_co_u32 v0, vcc_lo, s0, v4
; GFX11-NEXT:    v_add_co_ci_u32_e32 v1, vcc_lo, s1, v5, vcc_lo
; GFX11-NEXT:    v_add_co_u32 v4, vcc_lo, s2, v4
; GFX11-NEXT:    v_add_co_ci_u32_e32 v5, vcc_lo, s3, v5, vcc_lo
; GFX11-NEXT:    global_load_b128 v[0:3], v[0:1], off
; GFX11-NEXT:    s_waitcnt vmcnt(0)
; GFX11-NEXT:    v_not_b32_e32 v3, v3
; GFX11-NEXT:    v_not_b32_e32 v2, v2
; GFX11-NEXT:    v_not_b32_e32 v1, v1
; GFX11-NEXT:    v_not_b32_e32 v0, v0
; GFX11-NEXT:    global_store_b128 v[4:5], v[0:3], off
; GFX11-NEXT:    s_endpgm
entry:
  %implicitarg.ptr = tail call ptr addrspace(4) @llvm.amdgcn.implicitarg.ptr()
  %arg.1.ptr = getelementptr inbounds i8, ptr addrspace(4) %implicitarg.ptr, i64 40
  %arg.1 = load i64, ptr addrspace(4) %arg.1.ptr, align 8
  %workgroup.id.x = tail call i32 @llvm.amdgcn.workgroup.id.x()
  %arg.2.ptr = getelementptr inbounds i8, ptr addrspace(4) %implicitarg.ptr, i64 12
  %arg.2 = load i16, ptr addrspace(4) %arg.2.ptr, align 4
  %arg.2.ext = zext i16 %arg.2 to i32
  %mul = mul i32 %workgroup.id.x, %arg.2.ext
  %workitem.id.x = tail call i32 @llvm.amdgcn.workitem.id.x()
  %add = add i32 %mul, %workitem.id.x
  %add.ext = zext i32 %add to i64
  %add.1 = add i64 %arg.1, %add.ext
  %sext = shl i64 %add.1, 32
  %idxprom = ashr exact i64 %sext, 32
  %arrayidx = getelementptr inbounds <16 x i8>, ptr addrspace(1) %src, i64 %idxprom
  %arrayval = load <16 x i8>, ptr addrspace(1) %arrayidx, align 16
  %not = xor <16 x i8> %arrayval, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
  %arrayidx2 = getelementptr inbounds <16 x i8>, ptr addrspace(1) %dst, i64 %idxprom
  store <16 x i8> %not, ptr addrspace(1) %arrayidx2, align 16
  ret void
}