xref: /llvm-project/llvm/test/Transforms/SampleProfile/csspgo-inline-debug.ll (revision 0271ae65a66367d802fa6866599d11e35f68450f) 
1; REQUIRES: asserts
2; Test that the new FDO inliner using prioty queue will not visit same call site again and again.
3; Use debug prints as repeated call site evaluation is not visible from final inline decision.
4
5; Note that we need new pass manager to enable top-down processing for sample profile loader
6; RUN: opt < %s -passes=sample-profile -sample-profile-file=%S/Inputs/profile-context-tracker.prof -sample-profile-inline-size -sample-profile-prioritized-inline=0 -debug-only=sample-context-tracker -o /dev/null 2>&1 | FileCheck %s --check-prefix=OLD-INLINE
7; RUN: opt < %s -passes=sample-profile -sample-profile-file=%S/Inputs/profile-context-tracker.prof -sample-profile-inline-size -sample-profile-prioritized-inline=1 -debug-only=sample-context-tracker -o /dev/null 2>&1 | FileCheck %s --check-prefix=NEW-INLINE
8
9; Old inliner will evaluate the same call site three times
10; OLD-INLINE:      Getting callee context for instr:   %call = tail call i32 @_Z5funcBi
11; OLD-INLINE-NEXT:   Callee context found: main:3.1 @ _Z5funcBi
12; OLD-INLINE:      Getting callee context for instr:   %call = tail call i32 @_Z5funcBi
13; OLD-INLINE-NEXT:   Callee context found: main:3.1 @ _Z5funcBi
14; OLD-INLINE:      Getting callee context for instr:   %call = tail call i32 @_Z5funcBi
15; OLD-INLINE-NEXT:   Callee context found: main:3.1 @ _Z5funcBi
16
17; New inliner only evaluate the same call site once
18; NEW-INLINE:      Getting callee context for instr:   %call = tail call i32 @_Z5funcBi
19; NEW-INLINE-NEXT:   Callee context found: main:3.1 @ _Z5funcBi
20; NEW-INLINE-NOT:  Getting callee context for instr:   %call = tail call i32 @_Z5funcBi
21; NEW-INLINE-NOT:    Callee context found: main:3.1 @ _Z5funcBi
22
23@factor = dso_local global i32 3, align 4, !dbg !0
24
25define dso_local i32 @main() local_unnamed_addr #0 !dbg !18 {
26entry:
27  br label %for.body, !dbg !25
28
29for.cond.cleanup:                                 ; preds = %for.body
30  ret i32 %add3, !dbg !27
31
32for.body:                                         ; preds = %for.body, %entry
33  %x.011 = phi i32 [ 300000, %entry ], [ %dec, %for.body ]
34  %r.010 = phi i32 [ 0, %entry ], [ %add3, %for.body ]
35  %call = tail call i32 @_Z5funcBi(i32 %x.011), !dbg !32
36  %add = add nuw nsw i32 %x.011, 1, !dbg !31
37  %call1 = tail call i32 @_Z5funcAi(i32 %add), !dbg !28
38  %add2 = add i32 %call, %r.010, !dbg !34
39  %add3 = add i32 %add2, %call1, !dbg !35
40  %dec = add nsw i32 %x.011, -1, !dbg !36
41  %cmp = icmp eq i32 %x.011, 0, !dbg !38
42  br i1 %cmp, label %for.cond.cleanup, label %for.body, !dbg !25
43}
44
45define dso_local i32 @_Z5funcAi(i32 %x) local_unnamed_addr #1 !dbg !40 {
46entry:
47  %add = add nsw i32 %x, 100000, !dbg !44
48  %call = tail call i32 @_Z8funcLeafi(i32 %add), !dbg !45
49  ret i32 %call, !dbg !46
50}
51
52define dso_local i32 @_Z8funcLeafi(i32 %x) local_unnamed_addr #1 !dbg !54 {
53entry:
54  %cmp = icmp sgt i32 %x, 0, !dbg !57
55  br i1 %cmp, label %while.body, label %while.cond2.preheader, !dbg !59
56
57while.cond2.preheader:                            ; preds = %entry
58  %cmp313 = icmp slt i32 %x, 0, !dbg !60
59  br i1 %cmp313, label %while.body4, label %if.end, !dbg !63
60
61while.body:                                       ; preds = %while.body, %entry
62  %x.addr.016 = phi i32 [ %sub, %while.body ], [ %x, %entry ]
63  %tmp = load volatile i32, ptr @factor, align 4, !dbg !64
64  %call = tail call i32 @_Z3fibi(i32 %tmp), !dbg !67
65  %sub = sub nsw i32 %x.addr.016, %call, !dbg !68
66  %cmp1 = icmp sgt i32 %sub, 0, !dbg !69
67  br i1 %cmp1, label %while.body, label %if.end, !dbg !71
68
69while.body4:                                      ; preds = %while.body4, %while.cond2.preheader
70  %x.addr.114 = phi i32 [ %add, %while.body4 ], [ %x, %while.cond2.preheader ]
71  %tmp1 = load volatile i32, ptr @factor, align 4, !dbg !72
72  %call5 = tail call i32 @_Z3fibi(i32 %tmp1), !dbg !74
73  %add = add nsw i32 %call5, %x.addr.114, !dbg !75
74  %cmp3 = icmp slt i32 %add, 0, !dbg !60
75  br i1 %cmp3, label %while.body4, label %if.end, !dbg !63
76
77if.end:                                           ; preds = %while.body4, %while.body, %while.cond2.preheader
78  %x.addr.2 = phi i32 [ 0, %while.cond2.preheader ], [ %sub, %while.body ], [ %add, %while.body4 ]
79  ret i32 %x.addr.2, !dbg !76
80}
81
82define dso_local i32 @_Z5funcBi(i32 %x) local_unnamed_addr #0 !dbg !47 {
83entry:
84  %sub = add nsw i32 %x, -100000, !dbg !51
85  %call = tail call i32 @_Z8funcLeafi(i32 %sub), !dbg !52
86  ret i32 %call, !dbg !53
87}
88
89declare i32 @_Z3fibi(i32)
90
91attributes #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" }
92attributes #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" }
93
94!llvm.dbg.cu = !{!2}
95!llvm.module.flags = !{!14, !15, !16}
96!llvm.ident = !{!17}
97
98!0 = !DIGlobalVariableExpression(var: !1, expr: !DIExpression())
99!1 = distinct !DIGlobalVariable(name: "factor", scope: !2, file: !3, line: 21, type: !13, isLocal: false, isDefinition: true)
100!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)
101!3 = !DIFile(filename: "merged.cpp", directory: "/local/autofdo")
102!4 = !{}
103!5 = !{!6, !10, !11}
104!6 = !DISubprogram(name: "funcA", linkageName: "_Z5funcAi", scope: !3, file: !3, line: 6, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
105!7 = !DISubroutineType(types: !8)
106!8 = !{!9, !9}
107!9 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed)
108!10 = !DISubprogram(name: "funcB", linkageName: "_Z5funcBi", scope: !3, file: !3, line: 7, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
109!11 = !DISubprogram(name: "funcLeaf", linkageName: "_Z8funcLeafi", scope: !3, file: !3, line: 22, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
110!12 = !{!0}
111!13 = !DIDerivedType(tag: DW_TAG_volatile_type, baseType: !9)
112!14 = !{i32 7, !"Dwarf Version", i32 4}
113!15 = !{i32 2, !"Debug Info Version", i32 3}
114!16 = !{i32 1, !"wchar_size", i32 4}
115!17 = !{!"clang version 11.0.0"}
116!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)
117!19 = !DISubroutineType(types: !20)
118!20 = !{!9}
119!21 = !{!22, !23}
120!22 = !DILocalVariable(name: "r", scope: !18, file: !3, line: 12, type: !9)
121!23 = !DILocalVariable(name: "x", scope: !24, file: !3, line: 13, type: !9)
122!24 = distinct !DILexicalBlock(scope: !18, file: !3, line: 13, column: 3)
123!25 = !DILocation(line: 13, column: 3, scope: !26)
124!26 = !DILexicalBlockFile(scope: !24, file: !3, discriminator: 2)
125!27 = !DILocation(line: 17, column: 3, scope: !18)
126!28 = !DILocation(line: 14, column: 10, scope: !29)
127!29 = distinct !DILexicalBlock(scope: !30, file: !3, line: 13, column: 37)
128!30 = distinct !DILexicalBlock(scope: !24, file: !3, line: 13, column: 3)
129!31 = !DILocation(line: 14, column: 29, scope: !29)
130!32 = !DILocation(line: 14, column: 21, scope: !33)
131!33 = !DILexicalBlockFile(scope: !29, file: !3, discriminator: 2)
132!34 = !DILocation(line: 14, column: 19, scope: !29)
133!35 = !DILocation(line: 14, column: 7, scope: !29)
134!36 = !DILocation(line: 13, column: 33, scope: !37)
135!37 = !DILexicalBlockFile(scope: !30, file: !3, discriminator: 6)
136!38 = !DILocation(line: 13, column: 26, scope: !39)
137!39 = !DILexicalBlockFile(scope: !30, file: !3, discriminator: 2)
138!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)
139!44 = !DILocation(line: 27, column: 22, scope: !40)
140!45 = !DILocation(line: 27, column: 11, scope: !40)
141!46 = !DILocation(line: 29, column: 3, scope: !40)
142!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)
143!51 = !DILocation(line: 33, column: 22, scope: !47)
144!52 = !DILocation(line: 33, column: 11, scope: !47)
145!53 = !DILocation(line: 35, column: 3, scope: !47)
146!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)
147!57 = !DILocation(line: 49, column: 9, scope: !58)
148!58 = distinct !DILexicalBlock(scope: !54, file: !3, line: 49, column: 7)
149!59 = !DILocation(line: 49, column: 7, scope: !54)
150!60 = !DILocation(line: 58, column: 14, scope: !61)
151!61 = !DILexicalBlockFile(scope: !62, file: !3, discriminator: 2)
152!62 = distinct !DILexicalBlock(scope: !58, file: !3, line: 56, column: 8)
153!63 = !DILocation(line: 58, column: 5, scope: !61)
154!64 = !DILocation(line: 52, column: 16, scope: !65)
155!65 = distinct !DILexicalBlock(scope: !66, file: !3, line: 51, column: 19)
156!66 = distinct !DILexicalBlock(scope: !58, file: !3, line: 49, column: 14)
157!67 = !DILocation(line: 52, column: 12, scope: !65)
158!68 = !DILocation(line: 52, column: 9, scope: !65)
159!69 = !DILocation(line: 51, column: 14, scope: !70)
160!70 = !DILexicalBlockFile(scope: !66, file: !3, discriminator: 2)
161!71 = !DILocation(line: 51, column: 5, scope: !70)
162!72 = !DILocation(line: 59, column: 16, scope: !73)
163!73 = distinct !DILexicalBlock(scope: !62, file: !3, line: 58, column: 19)
164!74 = !DILocation(line: 59, column: 12, scope: !73)
165!75 = !DILocation(line: 59, column: 9, scope: !73)
166!76 = !DILocation(line: 63, column: 3, scope: !54)
167