[CSSPGO][llvm-profgen] Fix getCanonicalFnName usage in llvm-profgen

[wlei]

Summary

Previously we didn't support to keep the unique linkage name(-funique-internal-linkage-name) in llvm-profgen. As discussed in https://reviews.llvm.org/D96932, we choose to do canonicalization for it.

Now since "selected" is set as the default parameter of getCanonicalFnName in D96932, we don't need to add any attribute here for the previous usage and only fix the missing usage in the pseudo probe decoding.

[wenlei]

Looks like this is the only place where canonicalization is missing. Would be good to have a test case for proper profile name canonicalization (removing other suffix, but not unique), otherwise looks good.

[wlei]

In D98226#2612949, @wenlei wrote:

Looks like this is the only place where canonicalization is missing. Would be good to have a test case for proper profile name canonicalization (removing other suffix, but not unique), otherwise looks good.

Yeah, two places for the dwarf-based name canonicalization: one in the symbolizer and another is for the call target name. for probe, only need to call getCanonicalFnName after func name decoding.

Good point to add test case, will add it later.

[hoy]

llvm/tools/llvm-profgen/PseudoProbe.cpp

165

Give it "selected" to avoid dropping .__unique suffix?

[wlei]

llvm/tools/llvm-profgen/PseudoProbe.cpp

165

The "selected" is already the default parameter, you can find it in https://reviews.llvm.org/D96932.

[hoy]

llvm/tools/llvm-profgen/PseudoProbe.cpp

165

Sounds good, thanks.

[wlei]

Add test cases

[wenlei]

lgtm, thanks.

[wmi]

llvm/test/tools/llvm-profgen/fname-canonicalization.test

2

Since the default strategy of getCanonicalFnName is "selected" which is trying to keep ".__uniq." suffix if it exists, so I guess the test will behave the same no matter whether getCanonicalFnName is functioning or not, is that correct? If that is the case, maybe add another type of "suffix" like '.llvm.' in the test?

[wlei]

llvm/test/tools/llvm-profgen/fname-canonicalization.test

2

Yes. Good point to add a test for other suffix.

Added the '.llvm.1000' suffix to the symbolizer test(provide a switch --show-canonical-fname for this). With the --show-canonical-fname, it will get rid of the '.llvm.1000' suffix.

[wlei]

Addressing Wei's feedback

[wmi]

LGTM. Thanks.

llvm/test/tools/llvm-profgen/Inputs/unique-linkage-name-dwarf.perfscript

+PERF_RECORD_MMAP2 3021905/3021905: [0x400000(0x1000) @ 0 00:1d 306132807 1941166]: r-xp unique-linkage-name-dwarf.perfbin
+
+            4006de
+  5541f689495641d7
+ 0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x4006fb/0x4006d0/M/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0  0x400728/0x4006de/P/-/-/0

llvm/test/tools/llvm-profgen/Inputs/unique-linkage-name-probe.perfscript

+PERF_RECORD_MMAP2 3933222/3933222: [0x201000(0x1000) @ 0 00:1d 306175638 1941562]: r-xp unique-linkage-name-probe.perfbin
+
+            2017ce
+  5541f689495641d7
+ 0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0  0x201818/0x2017ce/P/-/-/0

llvm/test/tools/llvm-profgen/fname-canonicalization.test

+; Test unique linkage name decoding and canonicalization from dwarf debug info
+; RUN: llvm-profgen --perfscript=%S/Inputs/unique-linkage-name-dwarf.perfscript --binary=%S/Inputs/unique-linkage-name-dwarf.perfbin --output=%t --csprof-cold-thres=0
+; RUN: FileCheck %s --input-file %t --check-prefix=CHECK-DWARF-FNAME
+
+; CHECK-DWARF-FNAME:[main:1 @ foo]:309:0
+; CHECK-DWARF-FNAME: 2: 14
+; CHECK-DWARF-FNAME: 3: 29 bar.__uniq.26267048767521081047744692097241227776:14
+; CHECK-DWARF-FNAME:[main:1 @ foo:3 @ bar.__uniq.26267048767521081047744692097241227776]:84:0
+; CHECK-DWARF-FNAME: 1: 14
+
+
+; Test unique linkage name decoding and canonicalization from pseudo probe info
+; RUN: llvm-profgen --perfscript=%S/Inputs/unique-linkage-name-probe.perfscript --binary=%S/Inputs/unique-linkage-name-probe.perfbin --output=%t --csprof-cold-thres=0
+; RUN: FileCheck %s --input-file %t --check-prefix=CHECK-PROBE-FNAME
+
+; CHECK-PROBE-FNAME:[main:2 @ foo]:75:0
+; CHECK-PROBE-FNAME: 2: 15
+; CHECK-PROBE-FNAME: 3: 15
+; CHECK-PROBE-FNAME: 4: 15
+; CHECK-PROBE-FNAME: 6: 15
+; CHECK-PROBE-FNAME: 8: 15 _ZL3barii.__uniq.26267048767521081047744692097241227776:15
+; CHECK-PROBE-FNAME: !CFGChecksum: 563088904013236
+; CHECK-PROBE-FNAME:[main:2 @ foo:8 @ _ZL3barii.__uniq.26267048767521081047744692097241227776]:30:15
+; CHECK-PROBE-FNAME: 1: 15
+; CHECK-PROBE-FNAME: 2: 18446744073709551615
+; CHECK-PROBE-FNAME: 3: 18446744073709551615
+; CHECK-PROBE-FNAME: 4: 15
+; CHECK-PROBE-FNAME: !CFGChecksum: 72617220756
+
+
+; Original code:
+; Dwarf:  clang -O3 -funique-internal-linkage-names -g test.c -o a.out
+; Probe:  clang -O3 -funique-internal-linkage-names  -fexperimental-new-pass-manager -fuse-ld=lld -fpseudo-probe-for-profiling -fno-omit-frame-pointer -mno-omit-leaf-frame-pointer -Xclang -mdisable-tail-calls  -g test.c  -o a.out
+
+#include <stdio.h>
+
+static int bar(int x, int y) {
+  if (x % 3) {
+    return x - y;
+  }
+  return x + y;
+}
+
+void foo() {
+  int s, i = 0;
+  while (i++ < 4000 * 4000)
+    if (i % 91) s = bar(i, s); else s += 30;
+  printf("sum is %d\n", s);
+}
+
+int main() {
+  foo();
+  return 0;
+}

llvm/test/tools/llvm-profgen/symbolize.ll

 ; REQUIRES: x86-registered-target
 ; RUN: llc -filetype=obj %s -o %t
 ; RUN: llvm-profgen --binary=%t --perfscript=%s --output=%t1 --show-disassembly-only -x86-asm-syntax=intel --show-source-locations | FileCheck %s --match-full-lines
+; RUN: llvm-profgen --binary=%t --perfscript=%s --output=%t2 --show-disassembly-only -x86-asm-syntax=intel --show-source-locations --show-canonical-fname | FileCheck %s --match-full-lines  --check-prefix=CHECK-CANO
 
 ; CHECK: Disassembly of section .text [0x0, 0x4a]:
-; CHECK: <funcA>:
-; CHECK:        0:  mov eax, edi                         funcA:0
-; CHECK:        2:  mov ecx, dword ptr [rip]             funcLeaf:2 @ funcA:1
-; CHECK:        8:  lea edx, [rcx + 3]                   fib:2 @ funcLeaf:2 @ funcA:1
-; CHECK:        b:  cmp ecx, 3                           fib:2 @ funcLeaf:2 @ funcA:1
-; CHECK:        e:  cmovl edx, ecx                       fib:2 @ funcLeaf:2 @ funcA:1
-; CHECK:       11:  sub eax, edx                         funcLeaf:2 @ funcA:1
-; CHECK:       13:  ret                                  funcA:2
+; CHECK: <funcA.llvm.1000>:
+; CHECK:        0:  mov eax, edi                         funcA.llvm.1000:0
+; CHECK:        2:  mov ecx, dword ptr [rip]             funcLeaf:2 @ funcA.llvm.1000:1
+; CHECK:        8:  lea edx, [rcx + 3]                   fib:2 @ funcLeaf:2 @ funcA.llvm.1000:1
+; CHECK:        b:  cmp ecx, 3                           fib:2 @ funcLeaf:2 @ funcA.llvm.1000:1
+; CHECK:        e:  cmovl edx, ecx                       fib:2 @ funcLeaf:2 @ funcA.llvm.1000:1
+; CHECK:       11:  sub eax, edx                         funcLeaf:2 @ funcA.llvm.1000:1
+; CHECK:       13:  ret                                  funcA.llvm.1000:2
 ; CHECK:       14:  nop word ptr cs:[rax + rax]
 ; CHECK:       1e:  nop
+; CHECK-CANO: <funcA>:
+; CHECK-CANO:        0: mov eax, edi                         funcA:0
+; CHECK-CANO:        2: mov ecx, dword ptr [rip]             funcLeaf:2 @ funcA:1
+; CHECK-CANO:        8: lea edx, [rcx + 3]                   fib:2 @ funcLeaf:2 @ funcA:1
+; CHECK-CANO:        b: cmp ecx, 3                           fib:2 @ funcLeaf:2 @ funcA:1
+; CHECK-CANO:        e: cmovl edx, ecx                       fib:2 @ funcLeaf:2 @ funcA:1
+; CHECK-CANO:       11: sub eax, edx                         funcLeaf:2 @ funcA:1
+; CHECK-CANO:       13: ret                                  funcA:2
+; CHECK-CANO:       14: nop word ptr cs:[rax + rax]
+; CHECK-CANO:       1e: nop
 ; CHECK: <funcLeaf>:
 ; CHECK:      20: mov eax, edi                           funcLeaf:1
 ; CHECK:      22: mov ecx, dword ptr [rip]               funcLeaf:2
 ; CHECK:      28: lea edx, [rcx + 3]                     fib:2 @ funcLeaf:2
 ; CHECK:      2b: cmp ecx, 3                             fib:2 @ funcLeaf:2
 ; CHECK:      2e: cmovl edx, ecx                         fib:2 @ funcLeaf:2
 ; CHECK:      31: sub eax, edx                           funcLeaf:2
 ; CHECK:      33: ret                                    funcLeaf:3
 ; CHECK:      34: nop word ptr cs:[rax + rax]
 ; CHECK:      3e: nop
 ; CHECK: <fib>:
 ; CHECK:      40: lea eax, [rdi + 3]                     fib:2
 ; CHECK:      43: cmp edi, 3                             fib:2
 ; CHECK:      46: cmovl eax, edi                         fib:2
 ; CHECK:      49: ret                                    fib:8
 
 target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
 
 @factor = dso_local global i32 3
 
-define dso_local i32 @funcA(i32 %x) !dbg !12 {
+define dso_local i32 @funcA.llvm.1000(i32 %x) !dbg !12 {
 entry:
   call void @llvm.dbg.value(metadata i32 %x, metadata !16, metadata !DIExpression()), !dbg !18
   call void @llvm.dbg.value(metadata i32 %x, metadata !19, metadata !DIExpression()), !dbg !22
   %0 = load volatile i32, i32* @factor, align 4, !dbg !24, !tbaa !25
   call void @llvm.dbg.value(metadata i32 %0, metadata !29, metadata !DIExpression()), !dbg !32
   %cmp.i.i = icmp slt i32 %0, 3, !dbg !34
   %add.i.i = add nsw i32 %0, 3, !dbg !36
   %retval.0.i.i = select i1 %cmp.i.i, i32 %0, i32 %add.i.i, !dbg !36

 !2 = distinct !DICompileUnit(language: DW_LANG_C99, file: !3, emissionKind: FullDebug)
 !3 = !DIFile(filename: "test.c", directory: "test")
 !4 = !{}
 !6 = !DIDerivedType(tag: DW_TAG_volatile_type, baseType: !7)
 !7 = !DIBasicType(name: "int", size: 32, encoding: DW_ATE_signed)
 !8 = !{i32 7, !"Dwarf Version", i32 4}
 !9 = !{i32 2, !"Debug Info Version", i32 3}
 !10 = !{i32 1, !"wchar_size", i32 4}
-!12 = distinct !DISubprogram(name: "funcA", scope: !3, file: !3, line: 6, type: !13, scopeLine: 6, flags: DIFlagPrototyped | DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !2, retainedNodes: !15)
+!12 = distinct !DISubprogram(name: "funcA.llvm.1000", scope: !3, file: !3, line: 6, type: !13, scopeLine: 6, flags: DIFlagPrototyped | DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !2, retainedNodes: !15)
 !13 = !DISubroutineType(types: !14)
 !14 = !{!7, !7}
 !15 = !{!16, !17}
 !16 = !DILocalVariable(name: "x", arg: 1, scope: !12, file: !3, line: 6, type: !7)
 !17 = !DILocalVariable(name: "r", scope: !12, file: !3, line: 7, type: !7)
 !18 = !DILocation(line: 0, scope: !12)
 !19 = !DILocalVariable(name: "x", arg: 1, scope: !20, file: !3, line: 22, type: !7)
 !20 = distinct !DISubprogram(name: "funcLeaf", scope: !3, file: !3, line: 22, type: !13, scopeLine: 23, flags: DIFlagPrototyped | DIFlagAllCallsDescribed, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !2, retainedNodes: !21)

llvm/tools/llvm-profgen/ProfiledBinary.cpp

 cl::opt<bool> ShowDisassemblyOnly("show-disassembly-only", cl::ReallyHidden,
                                   cl::init(false), cl::ZeroOrMore,
                                   cl::desc("Print disassembled code."));
 
 cl::opt<bool> ShowSourceLocations("show-source-locations", cl::ReallyHidden,
                                   cl::init(false), cl::ZeroOrMore,
                                   cl::desc("Print source locations."));
 
+cl::opt<bool> ShowCanonicalFnName("show-canonical-fname", cl::ReallyHidden,
+                                  cl::init(false), cl::ZeroOrMore,
+                                  cl::desc("Print canonical function name."));
+
 cl::opt<bool> ShowPseudoProbe(
     "show-pseudo-probe", cl::ReallyHidden, cl::init(false), cl::ZeroOrMore,
     cl::desc("Print pseudo probe section and disassembled info."));
 
 namespace llvm {
 namespace sampleprof {
 
 static const Target *getTarget(const ObjectFile *Obj) {
bool ProfiledBinary::dissassembleSymbol(std::size_t SI, ArrayRef<uint8_t> Bytes,
   uint64_t SectSize = Section.getSize();
   uint64_t StartOffset = Symbols[SI].Addr - PreferredBaseAddress;
   uint64_t EndOffset = (SI + 1 < SE)
                            ? Symbols[SI + 1].Addr - PreferredBaseAddress
                            : SectionOffset + SectSize;
   if (StartOffset >= EndOffset)
     return true;
 
-  std::string &&SymbolName = Symbols[SI].Name.str();
+  StringRef SymbolName =
+      ShowCanonicalFnName
+          ? FunctionSamples::getCanonicalFnName(Symbols[SI].Name)
+          : Symbols[SI].Name;
   if (ShowDisassemblyOnly)
     outs() << '<' << SymbolName << ">:\n";
 
   auto WarnInvalidInsts = [](uint64_t Start, uint64_t End) {
     WithColor::warning() << "Invalid instructions at "
                          << format("%8" PRIx64, Start) << " - "
                          << format("%8" PRIx64, End) << "\n";
   };
if (ShowDisassemblyOnly) {
         IPrinter->printInst(&Inst, Offset + Size, "", *STI.get(), outs());
       else
         outs() << "\t<unknown>";
       if (ShowSourceLocations) {
         unsigned Cur = outs().tell() - Start;
         if (Cur < 40)
           outs().indent(40 - Cur);
         InstructionPointer IP(this, Offset);
-        outs() << getReversedLocWithContext(symbolize(IP));
+        outs() << getReversedLocWithContext(symbolize(IP, ShowCanonicalFnName));
       }
       outs() << "\n";
     }
 
     if (Disassembled) {
       const MCInstrDesc &MCDesc = MII->get(Inst.getOpcode());
       // Populate a vector of the symbolized callsite at this location
       // We don't need symbolized info for probe-based profile, just use an

llvm/tools/llvm-profgen/PseudoProbe.cpp

 #endif
   Data = Start;
   End = Data + Size;
 
   while (Data < End) {
     uint64_t GUID = readUnencodedNumber<uint64_t>();
     uint64_t Hash = readUnencodedNumber<uint64_t>();
     uint32_t NameSize = readUnsignedNumber<uint32_t>();
-    StringRef Name = readString(NameSize);
+    StringRef Name = FunctionSamples::getCanonicalFnName(readString(NameSize));
 
     // Initialize PseudoProbeFuncDesc and populate it into GUID2FuncDescMap
     GUID2FuncDescMap.emplace(GUID, PseudoProbeFuncDesc(GUID, Hash, Name));
   }
   assert(Data == End && "Have unprocessed data in pseudo_probe_desc section");
 }
 
 void PseudoProbeDecoder::buildAddress2ProbeMap(const uint8_t *Start,