xref: /llvm-project/llvm/test/Transforms/SampleProfile/profile-topdown-order.ll (revision eecb8c5f06149baf970fa0943e9fb9a6afe00207) 
1;; Test for different function processing orders affecting inlining in sample profile loader.
2
3;; There is an SCC _Z5funcAi -> _Z8funcLeafi -> _Z5funcAi in the program.
4;; With -use-profiled-call-graph=0, the top-down processing order of
5;; that SCC is (_Z8funcLeafi, _Z5funcAi), which is determinined based on
6;; the static call graph. With -use-profiled-call-graph=1, call edges
7;; from profile are considered, thus the order becomes (_Z5funcAi, _Z8funcLeafi).
8;; While _Z8funcLeafi is not supposed to be inlined, the outlined entry counts
9;; are affected.
10; RUN: opt < %s -passes=sample-profile -use-profiled-call-graph=0 -sample-profile-file=%S/Inputs/profile-topdown-order.prof -S | FileCheck %s -check-prefix=STATIC
11; RUN: opt < %s -passes=sample-profile -use-profiled-call-graph=1 -sample-profile-file=%S/Inputs/profile-topdown-order.prof -S | FileCheck %s -check-prefix=DYNAMIC
12
13
14; STATIC:  define dso_local i32 @_Z8funcLeafi{{.*}} !prof ![[#PROF:]]
15; STATIC:  ![[#PROF]] = !{!"function_entry_count", i64 21}
16; DYNAMIC: define dso_local i32 @_Z8funcLeafi{{.*}} !prof ![[#PROF:]]
17; DYNAMIC: ![[#PROF]] = !{!"function_entry_count", i64 27}
18
19@factor = dso_local global i32 3, align 4, !dbg !0
20@fp = dso_local global ptr null, align 8
21
22define dso_local i32 @_Z5funcAi(i32 %x) local_unnamed_addr #0 !dbg !40 {
23entry:
24  %add = add nsw i32 %x, 100000, !dbg !44
25  %0 = load ptr, ptr @fp, align 8
26  %call = call i32 %0(i32 8), !dbg !45
27  %call1 = tail call i32 @_Z8funcLeafi(i32 %add), !dbg !46
28  ret i32 %call, !dbg !46
29}
30
31define dso_local i32 @main() local_unnamed_addr #0 !dbg !18 {
32entry:
33  store ptr @_Z3fibi, ptr @fp, align 8, !dbg !25
34  br label %for.body, !dbg !25
35
36for.cond.cleanup:                                 ; preds = %for.body
37  ret i32 %add3, !dbg !27
38
39for.body:                                         ; preds = %for.body, %entry
40  %x.011 = phi i32 [ 300000, %entry ], [ %dec, %for.body ]
41  %r.010 = phi i32 [ 0, %entry ], [ %add3, %for.body ]
42  %call = tail call i32 @_Z5funcBi(i32 %x.011), !dbg !32
43  %add = add nuw nsw i32 %x.011, 1, !dbg !31
44  %call1 = tail call i32 @_Z5funcAi(i32 %add), !dbg !28
45  %add2 = add i32 %call, %r.010, !dbg !34
46  %add3 = add i32 %add2, %call1, !dbg !35
47  %dec = add nsw i32 %x.011, -1, !dbg !36
48  %cmp = icmp eq i32 %x.011, 0, !dbg !38
49  br i1 %cmp, label %for.cond.cleanup, label %for.body, !dbg !25
50}
51
52; INLINE: define dso_local i32 @_Z8funcLeafi
53; NOINLINE: define dso_local i32 @_Z8funcLeafi
54; ICALL-INLINE: define dso_local i32 @_Z8funcLeafi
55; ICALL-NOINLINE: define dso_local i32 @_Z8funcLeafi
56define dso_local i32 @_Z8funcLeafi(i32 %x) local_unnamed_addr #1 !dbg !54 {
57entry:
58  %cmp = icmp sgt i32 %x, 0, !dbg !57
59  br i1 %cmp, label %while.body, label %while.cond2.preheader, !dbg !59
60
61while.cond2.preheader:                            ; preds = %entry
62  %cmp313 = icmp slt i32 %x, 0, !dbg !60
63  br i1 %cmp313, label %while.body4, label %if.end, !dbg !63
64
65while.body:                                       ; preds = %while.body, %entry
66  %x.addr.016 = phi i32 [ %sub, %while.body ], [ %x, %entry ]
67  %tmp = load volatile i32, ptr @factor, align 4, !dbg !64
68  %call = tail call i32 @_Z5funcAi(i32 %tmp), !dbg !67
69  %sub = sub nsw i32 %x.addr.016, %call, !dbg !68
70  %cmp1 = icmp sgt i32 %sub, 0, !dbg !69
71  br i1 %cmp1, label %while.body, label %if.end, !dbg !71
72
73while.body4:                                      ; preds = %while.body4, %while.cond2.preheader
74  %x.addr.114 = phi i32 [ %add, %while.body4 ], [ %x, %while.cond2.preheader ]
75  %tmp1 = load volatile i32, ptr @factor, align 4, !dbg !72
76  %call5 = tail call i32 @_Z5funcBi(i32 %tmp1), !dbg !74
77  %add = add nsw i32 %call5, %x.addr.114, !dbg !75
78  %cmp3 = icmp slt i32 %add, 0, !dbg !60
79  br i1 %cmp3, label %while.body4, label %if.end, !dbg !63
80
81if.end:                                           ; preds = %while.body4, %while.body, %while.cond2.preheader
82  %x.addr.2 = phi i32 [ 0, %while.cond2.preheader ], [ %sub, %while.body ], [ %add, %while.body4 ]
83  ret i32 %x.addr.2, !dbg !76
84}
85
86define dso_local i32 @_Z5funcBi(i32 %x) local_unnamed_addr #0 !dbg !47 {
87entry:
88  %sub = add nsw i32 %x, -100000, !dbg !51
89  %call = tail call i32 @_Z8funcLeafi(i32 %sub), !dbg !52
90  ret i32 %call, !dbg !53
91}
92
93define dso_local i32 @_Z3fibi(i32 %x) local_unnamed_addr #1 !dbg !77 {
94entry:
95  %sub = add nsw i32 %x, -100000, !dbg !78
96  %call = tail call i32 @_Z3foo(i32 %sub), !dbg !78
97  ret i32 %sub, !dbg !78
98}
99
100declare i32 @_Z3foo(i32)
101
102attributes #0 = { nofree noinline nounwind uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "frame-pointer"="none" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" "use-sample-profile" }
103attributes #1 = { nofree nounwind uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "frame-pointer"="none" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" "use-sample-profile" }
104
105!llvm.dbg.cu = !{!2}
106!llvm.module.flags = !{!14, !15, !16}
107!llvm.ident = !{!17}
108
109!0 = !DIGlobalVariableExpression(var: !1, expr: !DIExpression())
110!1 = distinct !DIGlobalVariable(name: "factor", scope: !2, file: !3, line: 21, type: !13, isLocal: false, isDefinition: true)
111!2 = distinct !DICompileUnit(language: DW_LANG_C_plus_plus_14, file: !3, producer: "clang version 11.0.0", isOptimized: true, runtimeVersion: 0, emissionKind: FullDebug, enums: !4, retainedTypes: !5, globals: !12, splitDebugInlining: false, debugInfoForProfiling: true, nameTableKind: None)
112!3 = !DIFile(filename: "merged.cpp", directory: "/local/autofdo")
113!4 = !{}
114!5 = !{!6, !10, !11}
115!6 = !DISubprogram(name: "funcA", linkageName: "_Z5funcAi", scope: !3, file: !3, line: 6, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
116!7 = !DISubroutineType(types: !8)
117!8 = !{!9, !9}
118!9 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed)
119!10 = !DISubprogram(name: "funcB", linkageName: "_Z5funcBi", scope: !3, file: !3, line: 7, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
120!11 = !DISubprogram(name: "funcLeaf", linkageName: "_Z8funcLeafi", scope: !3, file: !3, line: 22, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
121!12 = !{!0}
122!13 = !DIDerivedType(tag: DW_TAG_volatile_type, baseType: !9)
123!14 = !{i32 7, !"Dwarf Version", i32 4}
124!15 = !{i32 2, !"Debug Info Version", i32 3}
125!16 = !{i32 1, !"wchar_size", i32 4}
126!17 = !{!"clang version 11.0.0"}
127!18 = distinct !DISubprogram(name: "main", scope: !3, file: !3, line: 11, type: !19, scopeLine: 11, flags: DIFlagPrototyped | DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !2, retainedNodes: !21)
128!19 = !DISubroutineType(types: !20)
129!20 = !{!9}
130!21 = !{!22, !23}
131!22 = !DILocalVariable(name: "r", scope: !18, file: !3, line: 12, type: !9)
132!23 = !DILocalVariable(name: "x", scope: !24, file: !3, line: 13, type: !9)
133!24 = distinct !DILexicalBlock(scope: !18, file: !3, line: 13, column: 3)
134!25 = !DILocation(line: 13, column: 3, scope: !26)
135!26 = !DILexicalBlockFile(scope: !24, file: !3, discriminator: 2)
136!27 = !DILocation(line: 17, column: 3, scope: !18)
137!28 = !DILocation(line: 14, column: 10, scope: !29)
138!29 = distinct !DILexicalBlock(scope: !30, file: !3, line: 13, column: 37)
139!30 = distinct !DILexicalBlock(scope: !24, file: !3, line: 13, column: 3)
140!31 = !DILocation(line: 14, column: 29, scope: !29)
141!32 = !DILocation(line: 14, column: 21, scope: !33)
142!33 = !DILexicalBlockFile(scope: !29, file: !3, discriminator: 2)
143!34 = !DILocation(line: 14, column: 19, scope: !29)
144!35 = !DILocation(line: 14, column: 7, scope: !29)
145!36 = !DILocation(line: 13, column: 33, scope: !37)
146!37 = !DILexicalBlockFile(scope: !30, file: !3, discriminator: 6)
147!38 = !DILocation(line: 13, column: 26, scope: !39)
148!39 = !DILexicalBlockFile(scope: !30, file: !3, discriminator: 2)
149!40 = distinct !DISubprogram(name: "funcA", linkageName: "_Z5funcAi", scope: !3, file: !3, line: 26, type: !7, scopeLine: 26, flags: DIFlagPrototyped | DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !2)
150!44 = !DILocation(line: 26, column: 22, scope: !40)
151!45 = !DILocation(line: 28, column: 11, scope: !40)
152!46 = !DILocation(line: 27, column: 3, scope: !40)
153!47 = distinct !DISubprogram(name: "funcB", linkageName: "_Z5funcBi", scope: !3, file: !3, line: 32, type: !7, scopeLine: 32, flags: DIFlagPrototyped | DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !2)
154!51 = !DILocation(line: 33, column: 22, scope: !47)
155!52 = !DILocation(line: 33, column: 11, scope: !47)
156!53 = !DILocation(line: 35, column: 3, scope: !47)
157!54 = distinct !DISubprogram(name: "funcLeaf", linkageName: "_Z8funcLeafi", scope: !3, file: !3, line: 48, type: !7, scopeLine: 48, flags: DIFlagPrototyped | DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !2)
158!57 = !DILocation(line: 49, column: 9, scope: !58)
159!58 = distinct !DILexicalBlock(scope: !54, file: !3, line: 49, column: 7)
160!59 = !DILocation(line: 49, column: 7, scope: !54)
161!60 = !DILocation(line: 58, column: 14, scope: !61)
162!61 = !DILexicalBlockFile(scope: !62, file: !3, discriminator: 2)
163!62 = distinct !DILexicalBlock(scope: !58, file: !3, line: 56, column: 8)
164!63 = !DILocation(line: 58, column: 5, scope: !61)
165!64 = !DILocation(line: 52, column: 16, scope: !65)
166!65 = distinct !DILexicalBlock(scope: !66, file: !3, line: 51, column: 19)
167!66 = distinct !DILexicalBlock(scope: !58, file: !3, line: 49, column: 14)
168!67 = !DILocation(line: 52, column: 12, scope: !65)
169!68 = !DILocation(line: 52, column: 9, scope: !65)
170!69 = !DILocation(line: 51, column: 14, scope: !70)
171!70 = !DILexicalBlockFile(scope: !66, file: !3, discriminator: 2)
172!71 = !DILocation(line: 51, column: 5, scope: !70)
173!72 = !DILocation(line: 59, column: 16, scope: !73)
174!73 = distinct !DILexicalBlock(scope: !62, file: !3, line: 58, column: 19)
175!74 = !DILocation(line: 59, column: 12, scope: !73)
176!75 = !DILocation(line: 59, column: 9, scope: !73)
177!76 = !DILocation(line: 63, column: 3, scope: !54)
178!77 = distinct !DISubprogram(name: "funcB", linkageName: "_Z3fibi", scope: !3, file: !3, line: 32, type: !7, scopeLine: 32, flags: DIFlagPrototyped | DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !2)
179!78 = !DILocation(line: 33, column: 22, scope: !77)
180