maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/Transforms/SampleProfile/offset.ll

83 lines
3.6 KiB
LLVM
Raw Normal View History

Tolerate negative offset when matching sample profile. In some cases (as illustrated in the unittest), lineno can be less than the heade_lineno because the function body are included from some other files. In this case, offset will be negative. This patch makes clang still able to match the profile to IR in this situation. http://reviews.llvm.org/D13914 llvm-svn: 250873
2015-10-21 09:22:27 +08:00
; RUN: opt < %s -sample-profile -sample-profile-file=%S/Inputs/offset.prof | opt -analyze -branch-prob | FileCheck %s
[PM] Port the Sample FDO to new PM (part-2) llvm-svn: 271072
2016-05-28 07:20:16 +08:00
; RUN: opt < %s -passes=sample-profile -sample-profile-file=%S/Inputs/offset.prof | opt -analyze -branch-prob | FileCheck %s
Tolerate negative offset when matching sample profile. In some cases (as illustrated in the unittest), lineno can be less than the heade_lineno because the function body are included from some other files. In this case, offset will be negative. This patch makes clang still able to match the profile to IR in this situation. http://reviews.llvm.org/D13914 llvm-svn: 250873
2015-10-21 09:22:27 +08:00
; Original C++ code for this test case:
;
; a.cc:
; #1
; #2
; #3
; #4
; #5 int foo(int a) {
; #6 #include "a.b"
; #7}
;
; a.b:
; #1 if (a > 0) {
; #2 return 10;
; #3 } else {
; #4 return 20;
; #5 }
; Function Attrs: nounwind uwtable
DI: Reverse direction of subprogram -> function edge. Previously, subprograms contained a metadata reference to the function they described. Because most clients need to get or set a subprogram for a given function rather than the other way around, this created unneeded inefficiency. For example, many passes needed to call the function llvm::makeSubprogramMap() to build a mapping from functions to subprograms, and the IR linker needed to fix up function references in a way that caused quadratic complexity in the IR linking phase of LTO. This change reverses the direction of the edge by storing the subprogram as function-level metadata and removing DISubprogram's function field. Since this is an IR change, a bitcode upgrade has been provided. Fixes PR23367. An upgrade script for textual IR for out-of-tree clients is attached to the PR. Differential Revision: http://reviews.llvm.org/D14265 llvm-svn: 252219
2015-11-06 06:03:56 +08:00
define i32 @_Z3fooi(i32 %a) #0 !dbg !4 {
Tolerate negative offset when matching sample profile. In some cases (as illustrated in the unittest), lineno can be less than the heade_lineno because the function body are included from some other files. In this case, offset will be negative. This patch makes clang still able to match the profile to IR in this situation. http://reviews.llvm.org/D13914 llvm-svn: 250873
2015-10-21 09:22:27 +08:00
entry:
%retval = alloca i32, align 4
%a.addr = alloca i32, align 4
store i32 %a, i32* %a.addr, align 4
call void @llvm.dbg.declare(metadata i32* %a.addr, metadata !11, metadata !12), !dbg !13
%0 = load i32, i32* %a.addr, align 4, !dbg !14
%cmp = icmp sgt i32 %0, 0, !dbg !18
br i1 %cmp, label %if.then, label %if.else, !dbg !19
Fine tuning of sample profile propagation algorithm. Summary: The refined propagation algorithm is more accurate and robust. Reviewers: davidxl, dnovillo Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D23224 llvm-svn: 278522
2016-08-13 00:22:12 +08:00
; CHECK: edge entry -> if.then probability is 0x0167ba82 / 0x80000000 = 1.10%
; CHECK: edge entry -> if.else probability is 0x7e98457e / 0x80000000 = 98.90% [HOT edge]
Tolerate negative offset when matching sample profile. In some cases (as illustrated in the unittest), lineno can be less than the heade_lineno because the function body are included from some other files. In this case, offset will be negative. This patch makes clang still able to match the profile to IR in this situation. http://reviews.llvm.org/D13914 llvm-svn: 250873
2015-10-21 09:22:27 +08:00
if.then: ; preds = %entry
store i32 10, i32* %retval, align 4, !dbg !20
br label %return, !dbg !20
if.else: ; preds = %entry
store i32 20, i32* %retval, align 4, !dbg !22
br label %return, !dbg !22
return: ; preds = %if.else, %if.then
%1 = load i32, i32* %retval, align 4, !dbg !24
ret i32 %1, !dbg !24
}
; Function Attrs: nounwind readnone
declare void @llvm.dbg.declare(metadata, metadata, metadata) #1
attributes #0 = { nounwind uwtable "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+fxsr,+mmx,+sse,+sse2" "unsafe-fp-math"="false" "use-soft-float"="false" }
attributes #1 = { nounwind readnone }
!llvm.dbg.cu = !{!0}
!llvm.module.flags = !{!8, !9}
!llvm.ident = !{!10}
[PR27284] Reverse the ownership between DICompileUnit and DISubprogram. Currently each Function points to a DISubprogram and DISubprogram has a scope field. For member functions the scope is a DICompositeType. DIScopes point to the DICompileUnit to facilitate type uniquing. Distinct DISubprograms (with isDefinition: true) are not part of the type hierarchy and cannot be uniqued. This change removes the subprograms list from DICompileUnit and instead adds a pointer to the owning compile unit to distinct DISubprograms. This would make it easy for ThinLTO to strip unneeded DISubprograms and their transitively referenced debug info. Motivation ---------- Materializing DISubprograms is currently the most expensive operation when doing a ThinLTO build of clang. We want the DISubprogram to be stored in a separate Bitcode block (or the same block as the function body) so we can avoid having to expensively deserialize all DISubprograms together with the global metadata. If a function has been inlined into another subprogram we need to store a reference the block containing the inlined subprogram. Attached to https://llvm.org/bugs/show_bug.cgi?id=27284 is a python script that updates LLVM IR testcases to the new format. http://reviews.llvm.org/D19034 <rdar://problem/25256815> llvm-svn: 266446
2016-04-15 23:57:41 +08:00
!0 = distinct !DICompileUnit(language: DW_LANG_C_plus_plus, file: !1, producer: "clang version 3.8.0 (trunk 250750)", isOptimized: false, runtimeVersion: 0, emissionKind: NoDebug, enums: !2)
Tolerate negative offset when matching sample profile. In some cases (as illustrated in the unittest), lineno can be less than the heade_lineno because the function body are included from some other files. In this case, offset will be negative. This patch makes clang still able to match the profile to IR in this situation. http://reviews.llvm.org/D13914 llvm-svn: 250873
2015-10-21 09:22:27 +08:00
!1 = !DIFile(filename: "a.cc", directory: "/tmp")
!2 = !{}
[PR27284] Reverse the ownership between DICompileUnit and DISubprogram. Currently each Function points to a DISubprogram and DISubprogram has a scope field. For member functions the scope is a DICompositeType. DIScopes point to the DICompileUnit to facilitate type uniquing. Distinct DISubprograms (with isDefinition: true) are not part of the type hierarchy and cannot be uniqued. This change removes the subprograms list from DICompileUnit and instead adds a pointer to the owning compile unit to distinct DISubprograms. This would make it easy for ThinLTO to strip unneeded DISubprograms and their transitively referenced debug info. Motivation ---------- Materializing DISubprograms is currently the most expensive operation when doing a ThinLTO build of clang. We want the DISubprogram to be stored in a separate Bitcode block (or the same block as the function body) so we can avoid having to expensively deserialize all DISubprograms together with the global metadata. If a function has been inlined into another subprogram we need to store a reference the block containing the inlined subprogram. Attached to https://llvm.org/bugs/show_bug.cgi?id=27284 is a python script that updates LLVM IR testcases to the new format. http://reviews.llvm.org/D19034 <rdar://problem/25256815> llvm-svn: 266446
2016-04-15 23:57:41 +08:00
!4 = distinct !DISubprogram(name: "foo", linkageName: "_Z3fooi", scope: !1, file: !1, line: 5, type: !5, isLocal: false, isDefinition: true, scopeLine: 5, flags: DIFlagPrototyped, isOptimized: false, unit: !0, variables: !2)
Tolerate negative offset when matching sample profile. In some cases (as illustrated in the unittest), lineno can be less than the heade_lineno because the function body are included from some other files. In this case, offset will be negative. This patch makes clang still able to match the profile to IR in this situation. http://reviews.llvm.org/D13914 llvm-svn: 250873
2015-10-21 09:22:27 +08:00
!5 = !DISubroutineType(types: !6)
!6 = !{!7, !7}
!7 = !DIBasicType(name: "int", size: 32, align: 32, encoding: DW_ATE_signed)
!8 = !{i32 2, !"Dwarf Version", i32 4}
!9 = !{i32 2, !"Debug Info Version", i32 3}
!10 = !{!"clang version 3.8.0 (trunk 250750)"}
!11 = !DILocalVariable(name: "a", arg: 1, scope: !4, file: !1, line: 5, type: !7)
!12 = !DIExpression()
!13 = !DILocation(line: 5, column: 13, scope: !4)
!14 = !DILocation(line: 1, column: 5, scope: !15)
!15 = distinct !DILexicalBlock(scope: !17, file: !16, line: 1, column: 5)
!16 = !DIFile(filename: "./a.b", directory: "/tmp")
!17 = !DILexicalBlockFile(scope: !4, file: !16, discriminator: 0)
!18 = !DILocation(line: 1, column: 7, scope: !15)
!19 = !DILocation(line: 1, column: 5, scope: !17)
!20 = !DILocation(line: 2, column: 3, scope: !21)
!21 = distinct !DILexicalBlock(scope: !15, file: !16, line: 1, column: 12)
!22 = !DILocation(line: 4, column: 3, scope: !23)
!23 = distinct !DILexicalBlock(scope: !15, file: !16, line: 3, column: 8)
!24 = !DILocation(line: 7, column: 1, scope: !25)
!25 = !DILexicalBlockFile(scope: !4, file: !1, discriminator: 0)