Skip to content

Commit f5a6fbd

Browse files
jdenny-ornljdenny-ornl
committed
Regenerate OpenMP tests from current clang
See llvm/test/Analysis/KernelInfo/openmp/README.md.
1 parent 0c30e7c commit f5a6fbd

File tree

2 files changed

+627
-639
lines changed

2 files changed

+627
-639
lines changed

‎llvm/test/Analysis/KernelInfo/openmp/amdgpu.ll‎

Lines changed: 59 additions & 66 deletions
Original file line numberDiff line numberDiff line change
@@ -16,13 +16,12 @@
1616
; CHECK-NEXT: remark: test.c:17:5: in artificial function '[[OFF_FUNC]]_debug__', direct call to defined function, callee is 'g'
1717
; CHECK-NEXT: remark: test.c:18:3: in artificial function '[[OFF_FUNC]]_debug__', direct call, callee is '@__kmpc_target_deinit'
1818
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', ExternalNotKernel = 0
19-
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', omp_target_thread_limit = 256
2019
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', amdgpu-max-num-workgroups[0] = 0
2120
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', amdgpu-max-num-workgroups[1] = 0
2221
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', amdgpu-max-num-workgroups[2] = 0
2322
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', amdgpu-flat-work-group-size[0] = 1
24-
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', amdgpu-flat-work-group-size[1] = 256
25-
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', amdgpu-waves-per-eu[0] = 1
23+
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', amdgpu-flat-work-group-size[1] = 1024
24+
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', amdgpu-waves-per-eu[0] = 4
2625
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', amdgpu-waves-per-eu[1] = 10
2726
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', Allocas = 3
2827
; CHECK-NEXT: remark: test.c:13:0: in artificial function '[[OFF_FUNC]]_debug__', AllocasStaticSizeSum = 20
@@ -39,12 +38,13 @@
3938
; CHECK-NEXT: remark: test.c:12:1: in artificial function '[[OFF_FUNC]]', 'load' instruction ('%[[#]]') accesses memory in flat address space
4039
; CHECK-NEXT: remark: test.c:12:1: in artificial function '[[OFF_FUNC]]', direct call to defined function, callee is artificial '[[OFF_FUNC]]_debug__'
4140
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', ExternalNotKernel = 0
41+
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', omp_target_thread_limit = 256
4242
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', amdgpu-max-num-workgroups[0] = 0
4343
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', amdgpu-max-num-workgroups[1] = 0
4444
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', amdgpu-max-num-workgroups[2] = 0
4545
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', amdgpu-flat-work-group-size[0] = 1
46-
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', amdgpu-flat-work-group-size[1] = 1024
47-
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', amdgpu-waves-per-eu[0] = 4
46+
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', amdgpu-flat-work-group-size[1] = 256
47+
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', amdgpu-waves-per-eu[0] = 1
4848
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', amdgpu-waves-per-eu[1] = 10
4949
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', Allocas = 1
5050
; CHECK-NEXT: remark: test.c:12:0: in artificial function '[[OFF_FUNC]]', AllocasStaticSizeSum = 8
@@ -95,82 +95,75 @@ target triple = "amdgcn-amd-amdhsa"
9595
@__omp_rtl_assume_threads_oversubscription = weak_odr hidden addrspace(1) constant i32 0
9696
@__omp_rtl_assume_no_thread_state = weak_odr hidden addrspace(1) constant i32 0
9797
@__omp_rtl_assume_no_nested_parallelism = weak_odr hidden addrspace(1) constant i32 0
98-
@0 = private unnamed_addr constant [57 x i8] c";test.c;__omp_offloading_fd02_71f35_h_l12_debug__;13;3;;\00", align 1
98+
@0 = private unnamed_addr constant [57 x i8] c";test.c;__omp_offloading_fd02_6f0c0_h_l12_debug__;13;3;;\00", align 1
9999
@1 = private unnamed_addr addrspace(1) constant %struct.ident_t { i32 0, i32 2, i32 0, i32 56, ptr @0 }, align 8
100-
@__omp_offloading_fd02_71f35_h_l12_dynamic_environment = weak_odr protected addrspace(1) global %struct.DynamicEnvironmentTy zeroinitializer
101-
@__omp_offloading_fd02_71f35_h_l12_kernel_environment = weak_odr protected addrspace(1) constant %struct.KernelEnvironmentTy { %struct.ConfigurationEnvironmentTy { i8 1, i8 1, i8 1, i32 1, i32 256, i32 -1, i32 -1, i32 0, i32 0 }, ptr addrspacecast (ptr addrspace(1) @1 to ptr), ptr addrspacecast (ptr addrspace(1) @__omp_offloading_fd02_71f35_h_l12_dynamic_environment to ptr) }
100+
@__omp_offloading_fd02_6f0c0_h_l12_dynamic_environment = weak_odr protected addrspace(1) global %struct.DynamicEnvironmentTy zeroinitializer
101+
@__omp_offloading_fd02_6f0c0_h_l12_kernel_environment = weak_odr protected addrspace(1) constant %struct.KernelEnvironmentTy { %struct.ConfigurationEnvironmentTy { i8 1, i8 1, i8 1, i32 1, i32 256, i32 -1, i32 -1, i32 0, i32 0 }, ptr addrspacecast (ptr addrspace(1) @1 to ptr), ptr addrspacecast (ptr addrspace(1) @__omp_offloading_fd02_6f0c0_h_l12_dynamic_environment to ptr) }
102102
@__oclc_ABI_version = weak_odr hidden local_unnamed_addr addrspace(4) constant i32 500
103103

104104
; Function Attrs: convergent noinline norecurse nounwind optnone
105-
define internal void @__omp_offloading_fd02_71f35_h_l12_debug__(ptr noalias noundef %dyn_ptr) #0 !dbg !16 {
106-
entry:
107-
%dyn_ptr.addr = alloca ptr, align 8, addrspace(5)
108-
%i = alloca i32, align 4, addrspace(5)
109-
%a = alloca [2 x i32], align 4, addrspace(5)
110-
%dyn_ptr.addr.ascast = addrspacecast ptr addrspace(5) %dyn_ptr.addr to ptr
111-
%i.ascast = addrspacecast ptr addrspace(5) %i to ptr
112-
%a.ascast = addrspacecast ptr addrspace(5) %a to ptr
113-
store ptr %dyn_ptr, ptr %dyn_ptr.addr.ascast, align 8
114-
tail call void @llvm.dbg.declare(metadata ptr addrspace(5) %dyn_ptr.addr, metadata !24, metadata !DIExpression()), !dbg !25
115-
%0 = call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @__omp_offloading_fd02_71f35_h_l12_kernel_environment to ptr), ptr %dyn_ptr), !dbg !26
116-
%exec_user_code = icmp eq i32 %0, -1, !dbg !26
117-
br i1 %exec_user_code, label %user_code.entry, label %worker.exit, !dbg !26
118-
119-
user_code.entry: ; preds = %entry
120-
tail call void @llvm.dbg.declare(metadata ptr addrspace(5) %i, metadata !27, metadata !DIExpression()), !dbg !30
121-
tail call void @llvm.dbg.declare(metadata ptr addrspace(5) %a, metadata !31, metadata !DIExpression()), !dbg !35
122-
call void @f() #5, !dbg !36
123-
call void @g() #5, !dbg !37
105+
define internal void @__omp_offloading_fd02_6f0c0_h_l12_debug__(ptr noalias noundef %0) #0 !dbg !16 {
106+
%2 = alloca ptr, align 8, addrspace(5)
107+
%3 = alloca i32, align 4, addrspace(5)
108+
%4 = alloca [2 x i32], align 4, addrspace(5)
109+
%5 = addrspacecast ptr addrspace(5) %2 to ptr
110+
%6 = addrspacecast ptr addrspace(5) %3 to ptr
111+
%7 = addrspacecast ptr addrspace(5) %4 to ptr
112+
store ptr %0, ptr %5, align 8
113+
#dbg_declare(ptr addrspace(5) %2, !24, !DIExpression(), !25)
114+
%8 = call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @__omp_offloading_fd02_6f0c0_h_l12_kernel_environment to ptr), ptr %0), !dbg !26
115+
%9 = icmp eq i32 %8, -1, !dbg !26
116+
br i1 %9, label %10, label %11, !dbg !26
117+
118+
10: ; preds = %1
119+
#dbg_declare(ptr addrspace(5) %3, !27, !DIExpression(), !30)
120+
#dbg_declare(ptr addrspace(5) %4, !31, !DIExpression(), !35)
121+
call void @f() #4, !dbg !36
122+
call void @g() #4, !dbg !37
124123
call void @__kmpc_target_deinit(), !dbg !38
125124
ret void, !dbg !39
126125

127-
worker.exit: ; preds = %entry
126+
11: ; preds = %1
128127
ret void, !dbg !26
129128
}
130129

130+
; Function Attrs: convergent mustprogress noinline norecurse nounwind optnone
131+
define weak_odr protected amdgpu_kernel void @__omp_offloading_fd02_6f0c0_h_l12(ptr noalias noundef %0) #1 !dbg !40 {
132+
%2 = alloca ptr, align 8, addrspace(5)
133+
%3 = addrspacecast ptr addrspace(5) %2 to ptr
134+
store ptr %0, ptr %3, align 8
135+
#dbg_declare(ptr addrspace(5) %2, !41, !DIExpression(), !42)
136+
%4 = load ptr, ptr %3, align 8, !dbg !43
137+
call void @__omp_offloading_fd02_6f0c0_h_l12_debug__(ptr %4) #5, !dbg !43
138+
ret void, !dbg !43
139+
}
140+
131141
declare i32 @__kmpc_target_init(ptr, ptr)
132142

133143
; Function Attrs: convergent
134-
declare void @f(...) #1
144+
declare void @f(...) #2
135145

136146
declare void @__kmpc_target_deinit()
137147

138-
; Function Attrs: convergent mustprogress noinline norecurse nounwind optnone
139-
define weak_odr protected amdgpu_kernel void @__omp_offloading_fd02_71f35_h_l12(ptr noalias noundef %dyn_ptr) #2 !dbg !40 {
140-
entry:
141-
%dyn_ptr.addr = alloca ptr, align 8, addrspace(5)
142-
%dyn_ptr.addr.ascast = addrspacecast ptr addrspace(5) %dyn_ptr.addr to ptr
143-
store ptr %dyn_ptr, ptr %dyn_ptr.addr.ascast, align 8
144-
tail call void @llvm.dbg.declare(metadata ptr addrspace(5) %dyn_ptr.addr, metadata !41, metadata !DIExpression()), !dbg !42
145-
%0 = load ptr, ptr %dyn_ptr.addr.ascast, align 8, !dbg !43
146-
call void @__omp_offloading_fd02_71f35_h_l12_debug__(ptr %0) #6, !dbg !43
147-
ret void, !dbg !43
148-
}
149-
150148
; Function Attrs: convergent noinline nounwind optnone
151149
define hidden void @g() #3 !dbg !44 {
152-
entry:
153-
%i = alloca i32, align 4, addrspace(5)
154-
%a = alloca [2 x i32], align 4, addrspace(5)
155-
%i.ascast = addrspacecast ptr addrspace(5) %i to ptr
156-
%a.ascast = addrspacecast ptr addrspace(5) %a to ptr
157-
tail call void @llvm.dbg.declare(metadata ptr addrspace(5) %i, metadata !47, metadata !DIExpression()), !dbg !48
158-
tail call void @llvm.dbg.declare(metadata ptr addrspace(5) %a, metadata !49, metadata !DIExpression()), !dbg !50
159-
call void @f() #5, !dbg !51
160-
call void @g() #5, !dbg !52
150+
%1 = alloca i32, align 4, addrspace(5)
151+
%2 = alloca [2 x i32], align 4, addrspace(5)
152+
%3 = addrspacecast ptr addrspace(5) %1 to ptr
153+
%4 = addrspacecast ptr addrspace(5) %2 to ptr
154+
#dbg_declare(ptr addrspace(5) %1, !47, !DIExpression(), !48)
155+
#dbg_declare(ptr addrspace(5) %2, !49, !DIExpression(), !50)
156+
call void @f() #4, !dbg !51
157+
call void @g() #4, !dbg !52
161158
ret void, !dbg !53
162159
}
163160

164-
; Function Attrs: nocallback nofree nosync nounwind speculatable willreturn memory(none)
165-
declare void @llvm.dbg.declare(metadata, metadata, metadata) #4
166-
167-
attributes #0 = { convergent noinline norecurse nounwind optnone "amdgpu-flat-work-group-size"="1,256" "frame-pointer"="all" "no-trapping-math"="true" "omp_target_thread_limit"="256" "stack-protector-buffer-size"="8" "target-cpu"="gfx906" "target-features"="+16-bit-insts,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot7-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64" }
168-
attributes #1 = { convergent "frame-pointer"="all" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx906" "target-features"="+16-bit-insts,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot7-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64" }
169-
attributes #2 = { convergent mustprogress noinline norecurse nounwind optnone "frame-pointer"="all" "kernel" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx906" "target-features"="+16-bit-insts,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot7-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64" "uniform-work-group-size"="true" }
161+
attributes #0 = { convergent noinline norecurse nounwind optnone "frame-pointer"="all" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx906" "target-features"="+16-bit-insts,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot7-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64" }
162+
attributes #1 = { convergent mustprogress noinline norecurse nounwind optnone "amdgpu-flat-work-group-size"="1,256" "frame-pointer"="all" "kernel" "no-trapping-math"="true" "omp_target_thread_limit"="256" "stack-protector-buffer-size"="8" "target-cpu"="gfx906" "target-features"="+16-bit-insts,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot7-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64" "uniform-work-group-size"="true" }
163+
attributes #2 = { convergent "frame-pointer"="all" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx906" "target-features"="+16-bit-insts,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot7-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64" }
170164
attributes #3 = { convergent noinline nounwind optnone "frame-pointer"="all" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="gfx906" "target-features"="+16-bit-insts,+ci-insts,+dl-insts,+dot1-insts,+dot10-insts,+dot2-insts,+dot7-insts,+dpp,+gfx8-insts,+gfx9-insts,+s-memrealtime,+s-memtime-inst,+wavefrontsize64" }
171-
attributes #4 = { nocallback nofree nosync nounwind speculatable willreturn memory(none) }
172-
attributes #5 = { convergent }
173-
attributes #6 = { nounwind }
165+
attributes #4 = { convergent }
166+
attributes #5 = { nounwind }
174167

175168
!llvm.dbg.cu = !{!0}
176169
!omp_offload.info = !{!2}
@@ -179,10 +172,10 @@ attributes #6 = { nounwind }
179172
!llvm.ident = !{!13, !14, !14, !14, !14, !14, !14, !14, !14, !14, !14, !14, !14, !14, !14, !14, !14}
180173
!opencl.ocl.version = !{!15, !15, !15, !15, !15, !15, !15, !15, !15, !15, !15, !15, !15, !15, !15, !15}
181174

182-
!0 = distinct !DICompileUnit(language: DW_LANG_C11, file: !1, producer: "clang version 19.0.0git (/tmp/llvm/clang 5a5e94265d423fa9eb39dc1b855511195f8dc0fe)", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, splitDebugInlining: false, nameTableKind: None)
183-
!1 = !DIFile(filename: "test.c", directory: "/tmp", checksumkind: CSK_MD5, checksum: "eff61a7cf33c8dd1bd6933250fc90157")
184-
!2 = !{i32 0, i32 64770, i32 466741, !"h", i32 12, i32 0, i32 0}
185-
!3 = !{ptr @__omp_offloading_fd02_71f35_h_l12, !"kernel", i32 1}
175+
!0 = distinct !DICompileUnit(language: DW_LANG_C11, file: !1, producer: "clang version 20.0.0git (/tmp/llvm/clang 0c30e7ceeb36294f4523da2590101314ca1c662d)", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, splitDebugInlining: false, nameTableKind: None)
176+
!1 = !DIFile(filename: "test.c", directory: "/tmp", checksumkind: CSK_MD5, checksum: "854099697e49b3ca7d3b3c08503e6fef")
177+
!2 = !{i32 0, i32 64770, i32 454848, !"h", i32 12, i32 0, i32 0}
178+
!3 = !{ptr @__omp_offloading_fd02_6f0c0_h_l12, !"kernel", i32 1}
186179
!4 = !{i32 1, !"amdhsa_code_object_version", i32 500}
187180
!5 = !{i32 7, !"Dwarf Version", i32 5}
188181
!6 = !{i32 2, !"Debug Info Version", i32 3}
@@ -192,10 +185,10 @@ attributes #6 = { nounwind }
192185
!10 = !{i32 8, !"PIC Level", i32 2}
193186
!11 = !{i32 7, !"frame-pointer", i32 2}
194187
!12 = !{i32 4, !"amdgpu_hostcall", i32 1}
195-
!13 = !{!"clang version 19.0.0git (/tmp/llvm/clang 5a5e94265d423fa9eb39dc1b855511195f8dc0fe)"}
188+
!13 = !{!"clang version 20.0.0git (/tmp/llvm/clang 0c30e7ceeb36294f4523da2590101314ca1c662d)"}
196189
!14 = !{!"AMD clang version 17.0.0 (https://github.com/RadeonOpenCompute/llvm-project roc-6.0.2 24012 af27734ed982b52a9f1be0f035ac91726fc697e4)"}
197190
!15 = !{i32 2, i32 0}
198-
!16 = distinct !DISubprogram(name: "__omp_offloading_fd02_71f35_h_l12_debug__", scope: !17, file: !17, line: 13, type: !18, scopeLine: 13, flags: DIFlagArtificial | DIFlagPrototyped, spFlags: DISPFlagLocalToUnit | DISPFlagDefinition, unit: !0, retainedNodes: !23)
191+
!16 = distinct !DISubprogram(name: "__omp_offloading_fd02_6f0c0_h_l12_debug__", scope: !17, file: !17, line: 13, type: !18, scopeLine: 13, flags: DIFlagArtificial | DIFlagPrototyped, spFlags: DISPFlagLocalToUnit | DISPFlagDefinition, unit: !0, retainedNodes: !23)
199192
!17 = !DIFile(filename: "test.c", directory: "/tmp")
200193
!18 = !DISubroutineType(types: !19)
201194
!19 = !{null, !20}
@@ -219,7 +212,7 @@ attributes #6 = { nounwind }
219212
!37 = !DILocation(line: 17, column: 5, scope: !28)
220213
!38 = !DILocation(line: 18, column: 3, scope: !28)
221214
!39 = !DILocation(line: 18, column: 3, scope: !16)
222-
!40 = distinct !DISubprogram(name: "__omp_offloading_fd02_71f35_h_l12", scope: !17, file: !17, line: 12, type: !18, scopeLine: 12, flags: DIFlagArtificial | DIFlagPrototyped, spFlags: DISPFlagLocalToUnit | DISPFlagDefinition, unit: !0, retainedNodes: !23)
215+
!40 = distinct !DISubprogram(name: "__omp_offloading_fd02_6f0c0_h_l12", scope: !17, file: !17, line: 12, type: !18, scopeLine: 12, flags: DIFlagArtificial | DIFlagPrototyped, spFlags: DISPFlagLocalToUnit | DISPFlagDefinition, unit: !0, retainedNodes: !23)
223216
!41 = !DILocalVariable(name: "dyn_ptr", arg: 1, scope: !40, type: !20, flags: DIFlagArtificial)
224217
!42 = !DILocation(line: 0, scope: !40)
225218
!43 = !DILocation(line: 12, column: 1, scope: !40)

0 commit comments

Comments
 (0)