xref: /llvm-project/llvm/test/Transforms/SampleProfile/csspgo-use-preinliner.ll (revision 339b8a0019658746217870215abca77291ad75b3) 
1; Test for CSSPGO's sample loader inlining to make sure it can use llvm-profgen preinliner's decision
2
3; RUN: opt < %s -passes=sample-profile -sample-profile-file=%S/Inputs/csspgo-use-preinliner.prof -pass-remarks=inline -sample-profile-prioritized-inline -profile-sample-accurate -sample-profile-use-preinliner=0 -S 2>&1 | FileCheck %s --check-prefix=DEFAULT
4; RUN: opt < %s -passes=sample-profile -sample-profile-file=%S/Inputs/csspgo-use-preinliner.prof -pass-remarks=inline -sample-profile-prioritized-inline -profile-sample-accurate -sample-profile-use-preinliner=1 -S 2>&1 | FileCheck %s --check-prefix=PREINLINE
5
6; RUN: llvm-profdata merge --sample --text --convert-sample-profile-layout=nest  -generate-merged-base-profiles=0 %S/Inputs/csspgo-use-preinliner.prof -o %t.prof
7; RUN: opt < %s -passes=sample-profile -sample-profile-file=%t.prof -pass-remarks=inline -sample-profile-prioritized-inline -profile-sample-accurate -sample-profile-use-preinliner=0 -S 2>&1 | FileCheck %s --check-prefix=DEFAULT
8; RUN: opt < %s -passes=sample-profile -sample-profile-file=%t.prof -pass-remarks=inline -sample-profile-prioritized-inline -profile-sample-accurate -sample-profile-use-preinliner=1 -S 2>&1 | FileCheck %s --check-prefix=PREINLINE
9
10; DEFAULT: '_Z5funcAi' inlined into 'main'
11; DEFAULT-NOT: inlined into
12
13; PREINLINE-NOT: inlined into
14; `[main:3 @ _Z5funcAi]` does not have preinline decision, so it's up for loader inliner to decided.
15; PREINLINE: '_Z5funcAi' inlined into 'main'
16; `[main:3 @ _Z5funcAi:1 @ _Z8funcLeafi]` is inlined according to preinline decision.
17; PREINLINE: '_Z8funcLeafi' inlined into 'main'
18; Even though `[main:3.1 @ _Z5funcBi]` context is marked should inline, `_Z5funcBi` is a noinline function, so we honor that and don't inline.
19; When _Z5funcBi is promoted to be top level context-less profile, `[_Z5funcBi:1 @ _Z8funcLeafi]` becomes synthetic context, so preinline decision is ignored and we don't inline `_Z8funcLeafi`.
20; PREINLINE-NOT: inlined into
21
22@factor = dso_local global i32 3, align 4, !dbg !0
23
24define dso_local i32 @main() local_unnamed_addr #0 !dbg !18 {
25entry:
26  br label %for.body, !dbg !25
27
28for.cond.cleanup:                                 ; preds = %for.body
29  ret i32 %add3, !dbg !27
30
31for.body:                                         ; preds = %for.body, %entry
32  %x.011 = phi i32 [ 300000, %entry ], [ %dec, %for.body ]
33  %r.010 = phi i32 [ 0, %entry ], [ %add3, %for.body ]
34  %call = tail call i32 @_Z5funcBi(i32 %x.011), !dbg !32
35  %add = add nuw nsw i32 %x.011, 1, !dbg !31
36  %call1 = tail call i32 @_Z5funcAi(i32 %add), !dbg !28
37  %add2 = add i32 %call, %r.010, !dbg !34
38  %add3 = add i32 %add2, %call1, !dbg !35
39  %dec = add nsw i32 %x.011, -1, !dbg !36
40  %cmp = icmp eq i32 %x.011, 0, !dbg !38
41  br i1 %cmp, label %for.cond.cleanup, label %for.body, !dbg !25
42}
43
44define dso_local i32 @_Z5funcAi(i32 %x) local_unnamed_addr #1 !dbg !40 {
45entry:
46  %add = add nsw i32 %x, 100000, !dbg !44
47  %call = tail call i32 @_Z8funcLeafi(i32 %add), !dbg !45
48  ret i32 %call, !dbg !46
49}
50
51define dso_local i32 @_Z8funcLeafi(i32 %x) local_unnamed_addr #1 !dbg !54 {
52entry:
53  %cmp = icmp sgt i32 %x, 0, !dbg !57
54  br i1 %cmp, label %while.body, label %while.cond2.preheader, !dbg !59
55
56while.cond2.preheader:                            ; preds = %entry
57  %cmp313 = icmp slt i32 %x, 0, !dbg !60
58  br i1 %cmp313, label %while.body4, label %if.end, !dbg !63
59
60while.body:                                       ; preds = %while.body, %entry
61  %x.addr.016 = phi i32 [ %sub, %while.body ], [ %x, %entry ]
62  %tmp = load volatile i32, ptr @factor, align 4, !dbg !64
63  %call = tail call i32 @_Z3fibi(i32 %tmp), !dbg !67
64  %sub = sub nsw i32 %x.addr.016, %call, !dbg !68
65  %cmp1 = icmp sgt i32 %sub, 0, !dbg !69
66  br i1 %cmp1, label %while.body, label %if.end, !dbg !71
67
68while.body4:                                      ; preds = %while.body4, %while.cond2.preheader
69  %x.addr.114 = phi i32 [ %add, %while.body4 ], [ %x, %while.cond2.preheader ]
70  %tmp1 = load volatile i32, ptr @factor, align 4, !dbg !72
71  %call5 = tail call i32 @_Z3fibi(i32 %tmp1), !dbg !74
72  %add = add nsw i32 %call5, %x.addr.114, !dbg !75
73  %cmp3 = icmp slt i32 %add, 0, !dbg !60
74  br i1 %cmp3, label %while.body4, label %if.end, !dbg !63
75
76if.end:                                           ; preds = %while.body4, %while.body, %while.cond2.preheader
77  %x.addr.2 = phi i32 [ 0, %while.cond2.preheader ], [ %sub, %while.body ], [ %add, %while.body4 ]
78  ret i32 %x.addr.2, !dbg !76
79}
80
81define dso_local i32 @_Z5funcBi(i32 %x) local_unnamed_addr #0 !dbg !47 {
82entry:
83  %sub = add nsw i32 %x, -100000, !dbg !51
84  %call = tail call i32 @_Z8funcLeafi(i32 %sub), !dbg !52
85  ret i32 %call, !dbg !53
86}
87
88declare i32 @_Z3fibi(i32)
89
90attributes #0 = { nofree noinline norecurse 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" }
91attributes #1 = { nofree norecurse 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" }
92
93!llvm.dbg.cu = !{!2}
94!llvm.module.flags = !{!14, !15, !16}
95!llvm.ident = !{!17}
96
97!0 = !DIGlobalVariableExpression(var: !1, expr: !DIExpression())
98!1 = distinct !DIGlobalVariable(name: "factor", scope: !2, file: !3, line: 21, type: !13, isLocal: false, isDefinition: true)
99!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)
100!3 = !DIFile(filename: "merged.cpp", directory: "/local/autofdo")
101!4 = !{}
102!5 = !{!6, !10, !11}
103!6 = !DISubprogram(name: "funcA", linkageName: "_Z5funcAi", scope: !3, file: !3, line: 6, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
104!7 = !DISubroutineType(types: !8)
105!8 = !{!9, !9}
106!9 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed)
107!10 = !DISubprogram(name: "funcB", linkageName: "_Z5funcBi", scope: !3, file: !3, line: 7, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
108!11 = !DISubprogram(name: "funcLeaf", linkageName: "_Z8funcLeafi", scope: !3, file: !3, line: 22, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
109!12 = !{!0}
110!13 = !DIDerivedType(tag: DW_TAG_volatile_type, baseType: !9)
111!14 = !{i32 7, !"Dwarf Version", i32 4}
112!15 = !{i32 2, !"Debug Info Version", i32 3}
113!16 = !{i32 1, !"wchar_size", i32 4}
114!17 = !{!"clang version 11.0.0"}
115!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)
116!19 = !DISubroutineType(types: !20)
117!20 = !{!9}
118!21 = !{!22, !23}
119!22 = !DILocalVariable(name: "r", scope: !18, file: !3, line: 12, type: !9)
120!23 = !DILocalVariable(name: "x", scope: !24, file: !3, line: 13, type: !9)
121!24 = distinct !DILexicalBlock(scope: !18, file: !3, line: 13, column: 3)
122!25 = !DILocation(line: 13, column: 3, scope: !26)
123!26 = !DILexicalBlockFile(scope: !24, file: !3, discriminator: 2)
124!27 = !DILocation(line: 17, column: 3, scope: !18)
125!28 = !DILocation(line: 14, column: 10, scope: !29)
126!29 = distinct !DILexicalBlock(scope: !30, file: !3, line: 13, column: 37)
127!30 = distinct !DILexicalBlock(scope: !24, file: !3, line: 13, column: 3)
128!31 = !DILocation(line: 14, column: 29, scope: !29)
129!32 = !DILocation(line: 14, column: 21, scope: !33)
130!33 = !DILexicalBlockFile(scope: !29, file: !3, discriminator: 2)
131!34 = !DILocation(line: 14, column: 19, scope: !29)
132!35 = !DILocation(line: 14, column: 7, scope: !29)
133!36 = !DILocation(line: 13, column: 33, scope: !37)
134!37 = !DILexicalBlockFile(scope: !30, file: !3, discriminator: 6)
135!38 = !DILocation(line: 13, column: 26, scope: !39)
136!39 = !DILexicalBlockFile(scope: !30, file: !3, discriminator: 2)
137!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)
138!44 = !DILocation(line: 27, column: 22, scope: !40)
139!45 = !DILocation(line: 27, column: 11, scope: !40)
140!46 = !DILocation(line: 29, column: 3, scope: !40)
141!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)
142!51 = !DILocation(line: 33, column: 22, scope: !47)
143!52 = !DILocation(line: 33, column: 11, scope: !47)
144!53 = !DILocation(line: 35, column: 3, scope: !47)
145!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)
146!57 = !DILocation(line: 49, column: 9, scope: !58)
147!58 = distinct !DILexicalBlock(scope: !54, file: !3, line: 49, column: 7)
148!59 = !DILocation(line: 49, column: 7, scope: !54)
149!60 = !DILocation(line: 58, column: 14, scope: !61)
150!61 = !DILexicalBlockFile(scope: !62, file: !3, discriminator: 2)
151!62 = distinct !DILexicalBlock(scope: !58, file: !3, line: 56, column: 8)
152!63 = !DILocation(line: 58, column: 5, scope: !61)
153!64 = !DILocation(line: 52, column: 16, scope: !65)
154!65 = distinct !DILexicalBlock(scope: !66, file: !3, line: 51, column: 19)
155!66 = distinct !DILexicalBlock(scope: !58, file: !3, line: 49, column: 14)
156!67 = !DILocation(line: 52, column: 12, scope: !65)
157!68 = !DILocation(line: 52, column: 9, scope: !65)
158!69 = !DILocation(line: 51, column: 14, scope: !70)
159!70 = !DILexicalBlockFile(scope: !66, file: !3, discriminator: 2)
160!71 = !DILocation(line: 51, column: 5, scope: !70)
161!72 = !DILocation(line: 59, column: 16, scope: !73)
162!73 = distinct !DILexicalBlock(scope: !62, file: !3, line: 58, column: 19)
163!74 = !DILocation(line: 59, column: 12, scope: !73)
164!75 = !DILocation(line: 59, column: 9, scope: !73)
165!76 = !DILocation(line: 63, column: 3, scope: !54)
166