xref: /llvm-project/llvm/test/Transforms/SampleProfile/csspgo-inline.ll (revision 339b8a0019658746217870215abca77291ad75b3) 
1; Test for CSSPGO's new early inliner using priority queue
2
3; Note that we need new pass manager to enable top-down processing for sample profile loader
4; Test we inlined the following in top-down order with old inliner
5;   main:3 @ _Z5funcAi
6;   main:3 @ _Z5funcAi:1 @ _Z8funcLeafi
7;   _Z5funcBi:1 @ _Z8funcLeafi
8; RUN: opt < %s -passes=sample-profile -sample-profile-file=%S/Inputs/profile-context-tracker.prof -sample-profile-inline-size -sample-profile-prioritized-inline=0 -profile-sample-accurate -S -pass-remarks=inline -o /dev/null 2>&1 | FileCheck %s --check-prefix=INLINE-BASE
9
10; RUN: llvm-profdata merge --sample --extbinary --use-md5 %S/Inputs/profile-context-tracker.prof -o %t.md5
11; RUN: opt < %s -passes=sample-profile -sample-profile-file=%t.md5 -sample-profile-inline-size -sample-profile-prioritized-inline=0 -profile-sample-accurate -S -pass-remarks=inline -o /dev/null 2>&1 | FileCheck %s --check-prefix=INLINE-BASE
12
13; RUN: llvm-profdata merge --sample --text --convert-sample-profile-layout=nest  %S/Inputs/profile-context-tracker.prof -o %t.prof
14; RUN: opt < %s -passes=sample-profile -sample-profile-file=%t.prof -sample-profile-inline-size -sample-profile-prioritized-inline=0 -profile-sample-accurate -S -pass-remarks=inline -o /dev/null 2>&1 | FileCheck %s --check-prefix=INLINE-BASE
15
16; With new FDO early inliner, callee entry count is used to drive inlining instead of callee total samples, so we get less inlining for given profile
17; RUN: opt < %s -passes=sample-profile -sample-profile-file=%S/Inputs/profile-context-tracker.prof -sample-profile-inline-size -profile-sample-accurate -S -pass-remarks=inline -o /dev/null 2>&1 | FileCheck %s --check-prefix=INLINE-NEW
18; RUN: opt < %s -passes=sample-profile -sample-profile-file=%t.prof -sample-profile-prioritized-inline -sample-profile-inline-size -profile-sample-accurate -S -pass-remarks=inline -o /dev/null 2>&1 | FileCheck %s --check-prefix=INLINE-NEW
19;
20; With new FDO early inliner, callee entry count is used to drive inlining instead of callee total samples, tuning hot cutoff can get us the same inlining
21; RUN: opt < %s -passes=sample-profile -sample-profile-file=%S/Inputs/profile-context-tracker.prof -sample-profile-inline-size -profile-summary-cutoff-hot=999990 -profile-sample-accurate -S -pass-remarks=inline -o /dev/null 2>&1 | FileCheck %s --check-prefix=INLINE-BASE
22;
23; With new FDO early inliner, callee entry count is used to drive inlining instead of callee total samples, tuning cold sample profile inline threshold can get us the same inlining
24; RUN: opt < %s -passes=sample-profile -sample-profile-file=%S/Inputs/profile-context-tracker.prof -sample-profile-inline-size -sample-profile-cold-inline-threshold=200 -profile-sample-accurate -S -pass-remarks=inline -o /dev/null 2>&1 | FileCheck %s --check-prefix=INLINE-BASE
25;
26; With new FDO early inliner and tuned cutoff, we can control inlining through size growth tuning knob.
27; RUN: opt < %s -passes=sample-profile -sample-profile-file=%S/Inputs/profile-context-tracker.prof -sample-profile-inline-size -profile-summary-cutoff-hot=999990 -sample-profile-inline-limit-min=0 -sample-profile-inline-growth-limit=1 -profile-sample-accurate -S -pass-remarks=inline -o /dev/null 2>&1 | FileCheck %s --allow-empty --check-prefix=INLINE-NEW-LIMIT1
28; RUN: opt < %s -passes=sample-profile -sample-profile-file=%S/Inputs/profile-context-tracker.prof -sample-profile-inline-size -profile-summary-cutoff-hot=999990 -sample-profile-inline-limit-min=10 -sample-profile-inline-growth-limit=1 -profile-sample-accurate -S -pass-remarks=inline -o /dev/null 2>&1 | FileCheck %s --check-prefix=INLINE-NEW-LIMIT2
29
30
31; INLINE-BASE: remark: merged.cpp:14:10: '_Z5funcAi' inlined into 'main' to match profiling context with (cost={{[0-9]+}}, threshold={{[0-9]+}}) at callsite main:3:10
32; INLINE-BASE: remark: merged.cpp:27:11: '_Z8funcLeafi' inlined into 'main' to match profiling context with (cost={{[0-9]+}}, threshold={{[0-9]+}}) at callsite _Z5funcAi:1:11 @ main:3:10
33; INLINE-BASE: remark: merged.cpp:33:11: '_Z8funcLeafi' inlined into '_Z5funcBi' to match profiling context with (cost={{[0-9]+}}, threshold={{[0-9]+}}) at callsite _Z5funcBi:1:11
34
35; INLINE-NEW: remark: merged.cpp:14:10: '_Z5funcAi' inlined into 'main' to match profiling context with (cost={{[0-9]+}}, threshold={{[0-9]+}}) at callsite main:3:10
36; INLINE-NEW-NOT: remark
37
38; INLINE-NEW-LIMIT1-NOT: remark
39
40; INLINE-NEW-LIMIT2: remark: merged.cpp:33:11: '_Z8funcLeafi' inlined into '_Z5funcBi' to match profiling context with (cost={{[0-9]+}}, threshold={{[0-9]+}}) at callsite _Z5funcBi:1:11
41; INLINE-NEW-LIMIT2: remark: merged.cpp:27:11: '_Z8funcLeafi' inlined into '_Z5funcAi' to match profiling context with (cost={{[0-9]+}}, threshold={{[0-9]+}}) at callsite _Z5funcAi:1:11;
42; INLINE-NEW-LIMIT2-NOT: remark
43
44@factor = dso_local global i32 3, align 4, !dbg !0
45
46define dso_local i32 @main() local_unnamed_addr #0 !dbg !18 {
47entry:
48  br label %for.body, !dbg !25
49
50for.cond.cleanup:                                 ; preds = %for.body
51  ret i32 %add3, !dbg !27
52
53for.body:                                         ; preds = %for.body, %entry
54  %x.011 = phi i32 [ 300000, %entry ], [ %dec, %for.body ]
55  %r.010 = phi i32 [ 0, %entry ], [ %add3, %for.body ]
56  %call = tail call i32 @_Z5funcBi(i32 %x.011), !dbg !32
57  %add = add nuw nsw i32 %x.011, 1, !dbg !31
58  %call1 = tail call i32 @_Z5funcAi(i32 %add), !dbg !28
59  %add2 = add i32 %call, %r.010, !dbg !34
60  %add3 = add i32 %add2, %call1, !dbg !35
61  %dec = add nsw i32 %x.011, -1, !dbg !36
62  %cmp = icmp eq i32 %x.011, 0, !dbg !38
63  br i1 %cmp, label %for.cond.cleanup, label %for.body, !dbg !25
64}
65
66define dso_local i32 @_Z5funcAi(i32 %x) local_unnamed_addr #1 !dbg !40 {
67entry:
68  %add = add nsw i32 %x, 100000, !dbg !44
69  %call = tail call i32 @_Z8funcLeafi(i32 %add), !dbg !45
70  ret i32 %call, !dbg !46
71}
72
73define dso_local i32 @_Z8funcLeafi(i32 %x) local_unnamed_addr #1 !dbg !54 {
74entry:
75  %cmp = icmp sgt i32 %x, 0, !dbg !57
76  br i1 %cmp, label %while.body, label %while.cond2.preheader, !dbg !59
77
78while.cond2.preheader:                            ; preds = %entry
79  %cmp313 = icmp slt i32 %x, 0, !dbg !60
80  br i1 %cmp313, label %while.body4, label %if.end, !dbg !63
81
82while.body:                                       ; preds = %while.body, %entry
83  %x.addr.016 = phi i32 [ %sub, %while.body ], [ %x, %entry ]
84  %tmp = load volatile i32, ptr @factor, align 4, !dbg !64
85  %call = tail call i32 @_Z3fibi(i32 %tmp), !dbg !67
86  %sub = sub nsw i32 %x.addr.016, %call, !dbg !68
87  %cmp1 = icmp sgt i32 %sub, 0, !dbg !69
88  br i1 %cmp1, label %while.body, label %if.end, !dbg !71
89
90while.body4:                                      ; preds = %while.body4, %while.cond2.preheader
91  %x.addr.114 = phi i32 [ %add, %while.body4 ], [ %x, %while.cond2.preheader ]
92  %tmp1 = load volatile i32, ptr @factor, align 4, !dbg !72
93  %call5 = tail call i32 @_Z3fibi(i32 %tmp1), !dbg !74
94  %add = add nsw i32 %call5, %x.addr.114, !dbg !75
95  %cmp3 = icmp slt i32 %add, 0, !dbg !60
96  br i1 %cmp3, label %while.body4, label %if.end, !dbg !63
97
98if.end:                                           ; preds = %while.body4, %while.body, %while.cond2.preheader
99  %x.addr.2 = phi i32 [ 0, %while.cond2.preheader ], [ %sub, %while.body ], [ %add, %while.body4 ]
100  ret i32 %x.addr.2, !dbg !76
101}
102
103define dso_local i32 @_Z5funcBi(i32 %x) local_unnamed_addr #0 !dbg !47 {
104entry:
105  %sub = add nsw i32 %x, -100000, !dbg !51
106  %call = tail call i32 @_Z8funcLeafi(i32 %sub), !dbg !52
107  ret i32 %call, !dbg !53
108}
109
110declare i32 @_Z3fibi(i32)
111
112attributes #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" }
113attributes #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" }
114
115!llvm.dbg.cu = !{!2}
116!llvm.module.flags = !{!14, !15, !16}
117!llvm.ident = !{!17}
118
119!0 = !DIGlobalVariableExpression(var: !1, expr: !DIExpression())
120!1 = distinct !DIGlobalVariable(name: "factor", scope: !2, file: !3, line: 21, type: !13, isLocal: false, isDefinition: true)
121!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)
122!3 = !DIFile(filename: "merged.cpp", directory: "/local/autofdo")
123!4 = !{}
124!5 = !{!6, !10, !11}
125!6 = !DISubprogram(name: "funcA", linkageName: "_Z5funcAi", scope: !3, file: !3, line: 6, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
126!7 = !DISubroutineType(types: !8)
127!8 = !{!9, !9}
128!9 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed)
129!10 = !DISubprogram(name: "funcB", linkageName: "_Z5funcBi", scope: !3, file: !3, line: 7, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
130!11 = !DISubprogram(name: "funcLeaf", linkageName: "_Z8funcLeafi", scope: !3, file: !3, line: 22, type: !7, flags: DIFlagPrototyped, spFlags: DISPFlagOptimized, retainedNodes: !4)
131!12 = !{!0}
132!13 = !DIDerivedType(tag: DW_TAG_volatile_type, baseType: !9)
133!14 = !{i32 7, !"Dwarf Version", i32 4}
134!15 = !{i32 2, !"Debug Info Version", i32 3}
135!16 = !{i32 1, !"wchar_size", i32 4}
136!17 = !{!"clang version 11.0.0"}
137!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)
138!19 = !DISubroutineType(types: !20)
139!20 = !{!9}
140!21 = !{!22, !23}
141!22 = !DILocalVariable(name: "r", scope: !18, file: !3, line: 12, type: !9)
142!23 = !DILocalVariable(name: "x", scope: !24, file: !3, line: 13, type: !9)
143!24 = distinct !DILexicalBlock(scope: !18, file: !3, line: 13, column: 3)
144!25 = !DILocation(line: 13, column: 3, scope: !26)
145!26 = !DILexicalBlockFile(scope: !24, file: !3, discriminator: 2)
146!27 = !DILocation(line: 17, column: 3, scope: !18)
147!28 = !DILocation(line: 14, column: 10, scope: !29)
148!29 = distinct !DILexicalBlock(scope: !30, file: !3, line: 13, column: 37)
149!30 = distinct !DILexicalBlock(scope: !24, file: !3, line: 13, column: 3)
150!31 = !DILocation(line: 14, column: 29, scope: !29)
151!32 = !DILocation(line: 14, column: 21, scope: !33)
152!33 = !DILexicalBlockFile(scope: !29, file: !3, discriminator: 2)
153!34 = !DILocation(line: 14, column: 19, scope: !29)
154!35 = !DILocation(line: 14, column: 7, scope: !29)
155!36 = !DILocation(line: 13, column: 33, scope: !37)
156!37 = !DILexicalBlockFile(scope: !30, file: !3, discriminator: 6)
157!38 = !DILocation(line: 13, column: 26, scope: !39)
158!39 = !DILexicalBlockFile(scope: !30, file: !3, discriminator: 2)
159!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)
160!44 = !DILocation(line: 27, column: 22, scope: !40)
161!45 = !DILocation(line: 27, column: 11, scope: !40)
162!46 = !DILocation(line: 29, column: 3, scope: !40)
163!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)
164!51 = !DILocation(line: 33, column: 22, scope: !47)
165!52 = !DILocation(line: 33, column: 11, scope: !47)
166!53 = !DILocation(line: 35, column: 3, scope: !47)
167!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)
168!57 = !DILocation(line: 49, column: 9, scope: !58)
169!58 = distinct !DILexicalBlock(scope: !54, file: !3, line: 49, column: 7)
170!59 = !DILocation(line: 49, column: 7, scope: !54)
171!60 = !DILocation(line: 58, column: 14, scope: !61)
172!61 = !DILexicalBlockFile(scope: !62, file: !3, discriminator: 2)
173!62 = distinct !DILexicalBlock(scope: !58, file: !3, line: 56, column: 8)
174!63 = !DILocation(line: 58, column: 5, scope: !61)
175!64 = !DILocation(line: 52, column: 16, scope: !65)
176!65 = distinct !DILexicalBlock(scope: !66, file: !3, line: 51, column: 19)
177!66 = distinct !DILexicalBlock(scope: !58, file: !3, line: 49, column: 14)
178!67 = !DILocation(line: 52, column: 12, scope: !65)
179!68 = !DILocation(line: 52, column: 9, scope: !65)
180!69 = !DILocation(line: 51, column: 14, scope: !70)
181!70 = !DILexicalBlockFile(scope: !66, file: !3, discriminator: 2)
182!71 = !DILocation(line: 51, column: 5, scope: !70)
183!72 = !DILocation(line: 59, column: 16, scope: !73)
184!73 = distinct !DILexicalBlock(scope: !62, file: !3, line: 58, column: 19)
185!74 = !DILocation(line: 59, column: 12, scope: !73)
186!75 = !DILocation(line: 59, column: 9, scope: !73)
187!76 = !DILocation(line: 63, column: 3, scope: !54)
188