Update test files.

As amdgpu-no-flat-scratch-init is set by opt, in most llc tests the
attribute is not set. This commit corrects this.

Author: jwanggit86

llvm/test/CodeGen/AMDGPU/GlobalISel/atomicrmw_udec_wrap.ll

@@ -121,15 +121,12 @@ define amdgpu_kernel void @addrspacecast(ptr addrspace(5) %ptr.private, ptr addr
   ret void
 }
 
-define amdgpu_kernel void @llvm_amdgcn_is_shared(ptr %ptr) {
+define amdgpu_kernel void @llvm_amdgcn_is_shared(ptr %ptr) #0 {
 ; GFX8V4-LABEL: llvm_amdgcn_is_shared:
 ; GFX8V4:       ; %bb.0:
 ; GFX8V4-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
 ; GFX8V4-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8V4-NEXT:    s_load_dword s0, s[6:7], 0x40
-; GFX8V4-NEXT:    s_add_i32 s12, s12, s17
-; GFX8V4-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
-; GFX8V4-NEXT:    s_mov_b32 flat_scratch_lo, s13
 ; GFX8V4-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8V4-NEXT:    s_cmp_eq_u32 s1, s0
 ; GFX8V4-NEXT:    s_cselect_b32 s0, 1, 0
@@ -143,9 +140,6 @@ define amdgpu_kernel void @llvm_amdgcn_is_shared(ptr %ptr) {
 ; GFX8V5-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
 ; GFX8V5-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8V5-NEXT:    s_load_dword s0, s[8:9], 0xcc
-; GFX8V5-NEXT:    s_add_i32 s12, s12, s17
-; GFX8V5-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
-; GFX8V5-NEXT:    s_mov_b32 flat_scratch_lo, s13
 ; GFX8V5-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8V5-NEXT:    s_cmp_eq_u32 s1, s0
 ; GFX8V5-NEXT:    s_cselect_b32 s0, 1, 0
@@ -183,15 +177,12 @@ define amdgpu_kernel void @llvm_amdgcn_is_shared(ptr %ptr) {
   ret void
 }
 
-define amdgpu_kernel void @llvm_amdgcn_is_private(ptr %ptr) {
+define amdgpu_kernel void @llvm_amdgcn_is_private(ptr %ptr) #0 {
 ; GFX8V4-LABEL: llvm_amdgcn_is_private:
 ; GFX8V4:       ; %bb.0:
 ; GFX8V4-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
 ; GFX8V4-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8V4-NEXT:    s_load_dword s0, s[6:7], 0x44
-; GFX8V4-NEXT:    s_add_i32 s12, s12, s17
-; GFX8V4-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
-; GFX8V4-NEXT:    s_mov_b32 flat_scratch_lo, s13
 ; GFX8V4-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8V4-NEXT:    s_cmp_eq_u32 s1, s0
 ; GFX8V4-NEXT:    s_cselect_b32 s0, 1, 0
@@ -205,9 +196,6 @@ define amdgpu_kernel void @llvm_amdgcn_is_private(ptr %ptr) {
 ; GFX8V5-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
 ; GFX8V5-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8V5-NEXT:    s_load_dword s0, s[8:9], 0xc8
-; GFX8V5-NEXT:    s_add_i32 s12, s12, s17
-; GFX8V5-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
-; GFX8V5-NEXT:    s_mov_b32 flat_scratch_lo, s13
 ; GFX8V5-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX8V5-NEXT:    s_cmp_eq_u32 s1, s0
 ; GFX8V5-NEXT:    s_cselect_b32 s0, 1, 0
@@ -245,7 +233,7 @@ define amdgpu_kernel void @llvm_amdgcn_is_private(ptr %ptr) {
   ret void
 }
 
-define amdgpu_kernel void @llvm_trap() {
+define amdgpu_kernel void @llvm_trap() #0 {
 ; GFX8V4-LABEL: llvm_trap:
 ; GFX8V4:       ; %bb.0:
 ; GFX8V4-NEXT:    s_mov_b64 s[0:1], s[6:7]
@@ -268,7 +256,7 @@ define amdgpu_kernel void @llvm_trap() {
   unreachable
 }
 
-define amdgpu_kernel void @llvm_debugtrap() {
+define amdgpu_kernel void @llvm_debugtrap() #0 {
 ; GFX8V4-LABEL: llvm_debugtrap:
 ; GFX8V4:       ; %bb.0:
 ; GFX8V4-NEXT:    s_trap 3
@@ -288,13 +276,10 @@ define amdgpu_kernel void @llvm_debugtrap() {
   unreachable
 }
 
-define amdgpu_kernel void @llvm_amdgcn_queue_ptr(ptr addrspace(1) %ptr)  {
+define amdgpu_kernel void @llvm_amdgcn_queue_ptr(ptr addrspace(1) %ptr)  #0 {
 ; GFX8V4-LABEL: llvm_amdgcn_queue_ptr:
 ; GFX8V4:       ; %bb.0:
-; GFX8V4-NEXT:    s_add_i32 s12, s12, s17
 ; GFX8V4-NEXT:    v_mov_b32_e32 v0, s6
-; GFX8V4-NEXT:    s_mov_b32 flat_scratch_lo, s13
-; GFX8V4-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
 ; GFX8V4-NEXT:    v_mov_b32_e32 v1, s7
 ; GFX8V4-NEXT:    s_add_u32 s0, s8, 8
 ; GFX8V4-NEXT:    flat_load_ubyte v0, v[0:1] glc
@@ -320,10 +305,7 @@ define amdgpu_kernel void @llvm_amdgcn_queue_ptr(ptr addrspace(1) %ptr)  {
 ;
 ; GFX8V5-LABEL: llvm_amdgcn_queue_ptr:
 ; GFX8V5:       ; %bb.0:
-; GFX8V5-NEXT:    s_add_i32 s12, s12, s17
 ; GFX8V5-NEXT:    v_mov_b32_e32 v0, s6
-; GFX8V5-NEXT:    s_mov_b32 flat_scratch_lo, s13
-; GFX8V5-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
 ; GFX8V5-NEXT:    v_mov_b32_e32 v1, s7
 ; GFX8V5-NEXT:    s_add_u32 s0, s8, 8
 ; GFX8V5-NEXT:    flat_load_ubyte v0, v[0:1] glc
@@ -402,3 +384,5 @@ declare void @llvm.debugtrap()
 
 !llvm.module.flags = !{!0}
 !0 = !{i32 1, !"amdhsa_code_object_version", i32 CODE_OBJECT_VERSION}
+
+attributes #0 = { "amdgpu-no-flat-scratch-init" }

llvm/test/CodeGen/AMDGPU/GlobalISel/atomicrmw_uinc_wrap.ll

@@ -9,7 +9,7 @@ define amdgpu_kernel void @v_insert_v64i32_varidx(ptr addrspace(1) %out.ptr, ptr
 ; GCN:       ; %bb.0:
 ; GCN-NEXT:    s_load_dwordx4 s[20:23], s[8:9], 0x0
 ; GCN-NEXT:    s_load_dwordx2 s[24:25], s[8:9], 0x10
-; GCN-NEXT:    s_add_u32 s0, s0, s17
+; GCN-NEXT:    s_add_u32 s0, s0, s15
 ; GCN-NEXT:    s_addc_u32 s1, s1, 0
 ; GCN-NEXT:    v_mov_b32_e32 v64, 0
 ; GCN-NEXT:    s_waitcnt lgkmcnt(0)
@@ -257,4 +257,4 @@ define amdgpu_kernel void @v_insert_v64i32_varidx(ptr addrspace(1) %out.ptr, ptr
   ret void
 }
 
-attributes #0 = { "amdgpu-flat-work-group-size"="1,256" "amdgpu-waves-per-eu"="1,10" }
+attributes #0 = { "amdgpu-flat-work-group-size"="1,256" "amdgpu-waves-per-eu"="1,10" "amdgpu-no-flat-scratch-init" }

llvm/test/CodeGen/AMDGPU/GlobalISel/extractelement.ll

@@ -11,16 +11,13 @@ define amdgpu_kernel void @use_lds_globals(ptr addrspace(1) %out, ptr addrspace(
 ; CHECK-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
 ; CHECK-NEXT:    v_mov_b32_e32 v0, 4
 ; CHECK-NEXT:    s_mov_b32 m0, -1
-; CHECK-NEXT:    s_add_i32 s12, s12, s17
 ; CHECK-NEXT:    ds_read_b32 v2, v0
-; CHECK-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
+; CHECK-NEXT:    v_mov_b32_e32 v3, 9
 ; CHECK-NEXT:    s_waitcnt lgkmcnt(0)
 ; CHECK-NEXT:    s_add_u32 s0, s0, 4
 ; CHECK-NEXT:    s_addc_u32 s1, s1, 0
 ; CHECK-NEXT:    v_mov_b32_e32 v0, s0
-; CHECK-NEXT:    s_mov_b32 flat_scratch_lo, s13
 ; CHECK-NEXT:    v_mov_b32_e32 v1, s1
-; CHECK-NEXT:    v_mov_b32_e32 v3, 9
 ; CHECK-NEXT:    flat_store_dword v[0:1], v2
 ; CHECK-NEXT:    v_mov_b32_e32 v0, 0x200
 ; CHECK-NEXT:    ds_write_b32 v0, v3
@@ -34,4 +31,4 @@ entry:
   ret void
 }
 
-attributes #0 = { nounwind }
+attributes #0 = { nounwind "amdgpu-no-flat-scratch-init" }

llvm/test/CodeGen/AMDGPU/GlobalISel/implicit-kernarg-backend-usage-global-isel.ll

@@ -1,14 +1,11 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -global-isel -mtriple=amdgcn--amdhsa -mcpu=hawaii -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
 
-define amdgpu_kernel void @test_wave64(i32 %arg0, [8 x i32], i64 %saved) {
+define amdgpu_kernel void @test_wave64(i32 %arg0, [8 x i32], i64 %saved) #0 {
 ; GCN-LABEL: test_wave64:
 ; GCN:       ; %bb.0: ; %entry
 ; GCN-NEXT:    s_load_dword s2, s[8:9], 0x0
 ; GCN-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0xa
-; GCN-NEXT:    s_add_i32 s12, s12, s17
-; GCN-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
-; GCN-NEXT:    s_mov_b32 flat_scratch_lo, s13
 ; GCN-NEXT:    s_waitcnt lgkmcnt(0)
 ; GCN-NEXT:    s_cmp_eq_u32 s2, 0
 ; GCN-NEXT:    s_cselect_b32 s2, 1, 0
@@ -28,3 +25,5 @@ entry:
 }
 
 declare i64 @llvm.amdgcn.if.break.i64(i1, i64)
+
+attributes #0 = { "amdgpu-no-flat-scratch-init" }

llvm/test/CodeGen/AMDGPU/GlobalISel/insertelement-stack-lower.ll

@@ -37,14 +37,11 @@ define double @v_trig_preop_f64_imm(double %a, i32 %b) {
   ret double %result
 }
 
-define amdgpu_kernel void @s_trig_preop_f64(double %a, i32 %b) {
+define amdgpu_kernel void @s_trig_preop_f64(double %a, i32 %b) #1 {
 ; CI-LABEL: s_trig_preop_f64:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dword s2, s[8:9], 0x2
 ; CI-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
-; CI-NEXT:    s_add_i32 s12, s12, s17
-; CI-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
-; CI-NEXT:    s_mov_b32 flat_scratch_lo, s13
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    v_mov_b32_e32 v0, s2
 ; CI-NEXT:    v_trig_preop_f64 v[0:1], s[0:1], v0
@@ -62,9 +59,6 @@ define amdgpu_kernel void @s_trig_preop_f64(double %a, i32 %b) {
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dword s2, s[8:9], 0x8
 ; VI-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
-; VI-NEXT:    s_add_i32 s12, s12, s17
-; VI-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
-; VI-NEXT:    s_mov_b32 flat_scratch_lo, s13
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
 ; VI-NEXT:    v_mov_b32_e32 v0, s2
 ; VI-NEXT:    v_trig_preop_f64 v[0:1], s[0:1], v0
@@ -82,8 +76,6 @@ define amdgpu_kernel void @s_trig_preop_f64(double %a, i32 %b) {
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_load_dword s2, s[8:9], 0x8
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
-; GFX9-NEXT:    s_add_u32 flat_scratch_lo, s12, s17
-; GFX9-NEXT:    s_addc_u32 flat_scratch_hi, s13, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_mov_b32_e32 v0, s2
 ; GFX9-NEXT:    v_trig_preop_f64 v[0:1], s[0:1], v0
@@ -93,10 +85,6 @@ define amdgpu_kernel void @s_trig_preop_f64(double %a, i32 %b) {
 ;
 ; GFX10-LABEL: s_trig_preop_f64:
 ; GFX10:       ; %bb.0:
-; GFX10-NEXT:    s_add_u32 s12, s12, s17
-; GFX10-NEXT:    s_addc_u32 s13, s13, 0
-; GFX10-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_LO), s12
-; GFX10-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_HI), s13
 ; GFX10-NEXT:    s_clause 0x1
 ; GFX10-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
 ; GFX10-NEXT:    s_load_dword s2, s[8:9], 0x8
@@ -121,13 +109,10 @@ define amdgpu_kernel void @s_trig_preop_f64(double %a, i32 %b) {
   ret void
 }
 
-define amdgpu_kernel void @s_trig_preop_f64_imm(double %a, i32 %b) {
+define amdgpu_kernel void @s_trig_preop_f64_imm(double %a, i32 %b) #1 {
 ; CI-LABEL: s_trig_preop_f64_imm:
 ; CI:       ; %bb.0:
 ; CI-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
-; CI-NEXT:    s_add_i32 s12, s12, s17
-; CI-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
-; CI-NEXT:    s_mov_b32 flat_scratch_lo, s13
 ; CI-NEXT:    s_waitcnt lgkmcnt(0)
 ; CI-NEXT:    v_trig_preop_f64 v[0:1], s[0:1], 7
 ; CI-NEXT:    s_add_u32 s0, s0, 4
@@ -143,9 +128,6 @@ define amdgpu_kernel void @s_trig_preop_f64_imm(double %a, i32 %b) {
 ; VI-LABEL: s_trig_preop_f64_imm:
 ; VI:       ; %bb.0:
 ; VI-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
-; VI-NEXT:    s_add_i32 s12, s12, s17
-; VI-NEXT:    s_lshr_b32 flat_scratch_hi, s12, 8
-; VI-NEXT:    s_mov_b32 flat_scratch_lo, s13
 ; VI-NEXT:    s_waitcnt lgkmcnt(0)
 ; VI-NEXT:    v_trig_preop_f64 v[0:1], s[0:1], 7
 ; VI-NEXT:    s_add_u32 s0, s0, 4
@@ -161,8 +143,6 @@ define amdgpu_kernel void @s_trig_preop_f64_imm(double %a, i32 %b) {
 ; GFX9-LABEL: s_trig_preop_f64_imm:
 ; GFX9:       ; %bb.0:
 ; GFX9-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
-; GFX9-NEXT:    s_add_u32 flat_scratch_lo, s12, s17
-; GFX9-NEXT:    s_addc_u32 flat_scratch_hi, s13, 0
 ; GFX9-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX9-NEXT:    v_trig_preop_f64 v[0:1], s[0:1], 7
 ; GFX9-NEXT:    flat_store_dwordx2 v[0:1], v[0:1]
@@ -171,10 +151,6 @@ define amdgpu_kernel void @s_trig_preop_f64_imm(double %a, i32 %b) {
 ;
 ; GFX10-LABEL: s_trig_preop_f64_imm:
 ; GFX10:       ; %bb.0:
-; GFX10-NEXT:    s_add_u32 s12, s12, s17
-; GFX10-NEXT:    s_addc_u32 s13, s13, 0
-; GFX10-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_LO), s12
-; GFX10-NEXT:    s_setreg_b32 hwreg(HW_REG_FLAT_SCR_HI), s13
 ; GFX10-NEXT:    s_load_dwordx2 s[0:1], s[8:9], 0x0
 ; GFX10-NEXT:    s_waitcnt lgkmcnt(0)
 ; GFX10-NEXT:    v_trig_preop_f64 v[0:1], s[0:1], 7
@@ -198,3 +174,4 @@ define amdgpu_kernel void @s_trig_preop_f64_imm(double %a, i32 %b) {
 declare double @llvm.amdgcn.trig.preop.f64(double, i32) #0
 
 attributes #0 = { nounwind readnone speculatable }
+attributes #1 = { "amdgpu-no-flat-scratch-init" }