Handle inline stacks in gcov-encoded sample profiles.

This patch adds support for reading sample profiles with inline stacks.
Inline stacks in a profile are generated when the sampled binary has
samples in inlined functions.

For instance, if main() calls foo() and foo() calls bar(), and bar() is
inlined into foo() and foo() inlined into main(), the profile may look
something like:

main total:364084 head:0
  [ ... ]
  2.3: _Z3fool total:243786
    1: 60149
    1.2: 38568
    1.4: 46511
    1.7: _Z3bari total:98558
      1.1: 52672
      1.2: 45886

At line 2, discriminator 3, main() calls foo(). In turn, foo() calls
bar() at line 1, discriminator 7.

In the textual format, this stacking of inline calls is represented
with indentation.

With this change, LLVM can now read sample profile files generated by
the create_gcov tool from https://github.com/google/autofdo.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@249644 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/include/llvm/ProfileData/SampleProf.h b/include/llvm/ProfileData/SampleProf.h
index ef5afc7eb3b0b39dfc03f5f08a93356a3c95b148..aa715df162fabbb37c55af4770fecffd82e21191 100644 (file)
--- a/include/llvm/ProfileData/SampleProf.h
+++ b/include/llvm/ProfileData/SampleProf.h
@@ -208,7 +208,7 @@ typedef DenseMap<CallsiteLocation, FunctionSamples> CallsiteSampleMap;
 class FunctionSamples {
 public:
   FunctionSamples() : TotalSamples(0), TotalHeadSamples(0) {}
-  void print(raw_ostream &OS = dbgs());
+  void print(raw_ostream &OS = dbgs(), unsigned Indent = 0) const;
   void addTotalSamples(unsigned Num) { TotalSamples += Num; }
   void addHeadSamples(unsigned Num) { TotalHeadSamples += Num; }
   void addBodySamples(int LineOffset, unsigned Discriminator, unsigned Num) {
@@ -302,6 +302,22 @@ private:
   /// are an offset from the start of the function.
   BodySampleMap BodySamples;
 
+  /// Map call sites to collected samples for the called function.
+  ///
+  /// Each entry in this map corresponds to all the samples
+  /// collected for the inlined function call at the given
+  /// location. For example, given:
+  ///
+  ///     void foo() {
+  ///  1    bar();
+  ///  ...
+  ///  8    baz();
+  ///     }
+  ///
+  /// If the bar() and baz() calls were inlined inside foo(), this
+  /// map will contain two entries.  One for all the samples collected
+  /// in the call to bar() at line offset 1, the other for all the samples
+  /// collected in the call to baz() at line offset 8.
   CallsiteSampleMap CallsiteSamples;
 };
 

diff --git a/include/llvm/ProfileData/SampleProfReader.h b/include/llvm/ProfileData/SampleProfReader.h
index 87c871f855b353a2efed9e38cda0baa63fac7b01..9f4d53902e1c4e1687c1f418f944321423448352 100644 (file)
--- a/include/llvm/ProfileData/SampleProfReader.h
+++ b/include/llvm/ProfileData/SampleProfReader.h
@@ -189,7 +189,7 @@ struct SourceInfo {
   uint32_t Discriminator;
 };
 
-typedef std::vector<SourceInfo> SourceStack;
+typedef SmallVector<FunctionSamples *, 10> InlineCallStack;
 
 // Supported histogram types in GCC.  Currently, we only need support for
 // call target histograms.
@@ -220,9 +220,8 @@ public:
 
 protected:
   std::error_code readNameTable();
-  std::error_code addSourceCount(StringRef Name, const SourceStack &Src,
-                                 uint64_t Count);
-  std::error_code readOneFunctionProfile(const SourceStack &Stack, bool Update);
+  std::error_code readOneFunctionProfile(const InlineCallStack &InlineStack,
+                                         bool Update, uint32_t Offset);
   std::error_code readFunctionProfiles();
   std::error_code skipNextWord();
   template <typename T> ErrorOr<T> readNumber();

diff --git a/include/llvm/ProfileData/SampleProfWriter.h b/include/llvm/ProfileData/SampleProfWriter.h
index 302a82d3286122d044cb3eb10ef908e828bde8bc..d3fa7429a6d4afb0326ec38b2db4800de02c077c 100644 (file)
--- a/include/llvm/ProfileData/SampleProfWriter.h
+++ b/include/llvm/ProfileData/SampleProfWriter.h
@@ -84,12 +84,18 @@ protected:
 class SampleProfileWriterText : public SampleProfileWriter {
 public:
   SampleProfileWriterText(StringRef F, std::error_code &EC)
-      : SampleProfileWriter(F, EC, sys::fs::F_Text) {}
+      : SampleProfileWriter(F, EC, sys::fs::F_Text), Indent(0) {}
 
   bool write(StringRef FName, const FunctionSamples &S) override;
   bool write(const Module &M, StringMap<FunctionSamples> &P) {
     return SampleProfileWriter::write(M, P);
   }
+
+private:
+  /// Indent level to use when writing.
+  ///
+  /// This is used when printing inlined callees.
+  unsigned Indent;
 };
 
 /// \brief Sample-based profile writer (binary format).

diff --git a/lib/ProfileData/SampleProfReader.cpp b/lib/ProfileData/SampleProfReader.cpp
index 94f87996bea57fe58e930ff5b27d4e0a2f105bb2..6d7d182e46c8325a3a5ac1ef6149b6c6dffbd9b4 100644 (file)
--- a/lib/ProfileData/SampleProfReader.cpp
+++ b/lib/ProfileData/SampleProfReader.cpp
@@ -132,13 +132,14 @@ using namespace llvm;
 /// \brief Print the samples collected for a function on stream \p OS.
 ///
 /// \param OS Stream to emit the output to.
-void FunctionSamples::print(raw_ostream &OS) {
+void FunctionSamples::print(raw_ostream &OS, unsigned Indent) const {
   OS << TotalSamples << ", " << TotalHeadSamples << ", " << BodySamples.size()
      << " sampled lines\n";
   for (const auto &SI : BodySamples) {
     LineLocation Loc = SI.first;
     const SampleRecord &Sample = SI.second;
-    OS << "\tline offset: " << Loc.LineOffset
+    OS.indent(Indent);
+    OS << "line offset: " << Loc.LineOffset
        << ", discriminator: " << Loc.Discriminator
        << ", number of samples: " << Sample.getSamples();
     if (Sample.hasCalls()) {
@@ -148,7 +149,15 @@ void FunctionSamples::print(raw_ostream &OS) {
     }
     OS << "\n";
   }
-  OS << "\n";
+  for (const auto &CS : CallsiteSamples) {
+    CallsiteLocation Loc = CS.first;
+    const FunctionSamples &CalleeSamples = CS.second;
+    OS.indent(Indent);
+    OS << "line offset: " << Loc.LineOffset
+       << ", discriminator: " << Loc.Discriminator
+       << ", inlined callee: " << Loc.CalleeName << ": ";
+    CalleeSamples.print(OS, Indent + 2);
+  }
 }
 
 /// \brief Dump the function profile for \p FName.
@@ -266,7 +275,7 @@ static bool ParseLine(const StringRef &Input, bool &IsCallsite, unsigned &Depth,
 std::error_code SampleProfileReaderText::read() {
   line_iterator LineIt(*Buffer, /*SkipBlanks=*/true, '#');
 
-  SmallVector<FunctionSamples *, 10> InlineStack;
+  InlineCallStack InlineStack;
 
   for (; !LineIt.is_at_eof(); ++LineIt) {
     if ((*LineIt)[(*LineIt).find_first_not_of(' ')] == '#')
@@ -559,31 +568,18 @@ std::error_code SampleProfileReaderGCC::readFunctionProfiles() {
   if (!GcovBuffer.readInt(NumFunctions))
     return sampleprof_error::truncated;
 
-  SourceStack Stack;
+  InlineCallStack Stack;
   for (uint32_t I = 0; I < NumFunctions; ++I)
-    if (std::error_code EC = readOneFunctionProfile(Stack, true))
+    if (std::error_code EC = readOneFunctionProfile(Stack, true, 0))
       return EC;
 
   return sampleprof_error::success;
 }
 
-std::error_code SampleProfileReaderGCC::addSourceCount(StringRef Name,
-                                                       const SourceStack &Src,
-                                                       uint64_t Count) {
-  if (Src.size() == 0 || Src[0].Malformed())
-    return sampleprof_error::malformed;
-  FunctionSamples &FProfile = Profiles[Name];
-  FProfile.addTotalSamples(Count);
-  // FIXME(dnovillo) - Properly update inline stack for FnName.
-  FProfile.addBodySamples(Src[0].Line, Src[0].Discriminator, Count);
-  return sampleprof_error::success;
-}
-
-std::error_code
-SampleProfileReaderGCC::readOneFunctionProfile(const SourceStack &Stack,
-                                               bool Update) {
+std::error_code SampleProfileReaderGCC::readOneFunctionProfile(
+    const InlineCallStack &InlineStack, bool Update, uint32_t Offset) {
   uint64_t HeadCount = 0;
-  if (Stack.size() == 0)
+  if (InlineStack.size() == 0)
     if (!GcovBuffer.readInt64(HeadCount))
       return sampleprof_error::truncated;
 
@@ -597,15 +593,31 @@ SampleProfileReaderGCC::readOneFunctionProfile(const SourceStack &Stack,
   if (!GcovBuffer.readInt(NumPosCounts))
     return sampleprof_error::truncated;
 
-  uint32_t NumCallSites;
-  if (!GcovBuffer.readInt(NumCallSites))
+  uint32_t NumCallsites;
+  if (!GcovBuffer.readInt(NumCallsites))
     return sampleprof_error::truncated;
 
-  if (Stack.size() == 0) {
-    FunctionSamples &FProfile = Profiles[Name];
-    FProfile.addHeadSamples(HeadCount);
-    if (FProfile.getTotalSamples() > 0)
+  FunctionSamples *FProfile = nullptr;
+  if (InlineStack.size() == 0) {
+    // If this is a top function that we have already processed, do not
+    // update its profile again.  This happens in the presence of
+    // function aliases.  Since these aliases share the same function
+    // body, there will be identical replicated profiles for the
+    // original function.  In this case, we simply not bother updating
+    // the profile of the original function.
+    FProfile = &Profiles[Name];
+    FProfile->addHeadSamples(HeadCount);
+    if (FProfile->getTotalSamples() > 0)
       Update = false;
+  } else {
+    // Otherwise, we are reading an inlined instance. The top of the
+    // inline stack contains the profile of the caller. Insert this
+    // callee in the caller's CallsiteMap.
+    FunctionSamples *CallerProfile = InlineStack.front();
+    uint32_t LineOffset = Offset >> 16;
+    uint32_t Discriminator = Offset & 0xffff;
+    FProfile = &CallerProfile->functionSamplesAt(
+        CallsiteLocation(LineOffset, Discriminator, Name));
   }
 
   for (uint32_t I = 0; I < NumPosCounts; ++I) {
@@ -621,13 +633,28 @@ SampleProfileReaderGCC::readOneFunctionProfile(const SourceStack &Stack,
     if (!GcovBuffer.readInt64(Count))
       return sampleprof_error::truncated;
 
-    SourceInfo Info(Name, "", "", 0, Offset >> 16, Offset & 0xffff);
-    SourceStack NewStack;
-    NewStack.push_back(Info);
-    NewStack.insert(NewStack.end(), Stack.begin(), Stack.end());
-    if (Update)
-      addSourceCount(NewStack[NewStack.size() - 1].FuncName, NewStack, Count);
+    // The line location is encoded in the offset as:
+    //   high 16 bits: line offset to the start of the function.
+    //   low 16 bits: discriminator.
+    uint32_t LineOffset = Offset >> 16;
+    uint32_t Discriminator = Offset & 0xffff;
+
+    InlineCallStack NewStack;
+    NewStack.push_back(FProfile);
+    NewStack.insert(NewStack.end(), InlineStack.begin(), InlineStack.end());
+    if (Update) {
+      // Walk up the inline stack, adding the samples on this line to
+      // the total sample count of the callers in the chain.
+      for (auto CallerProfile : NewStack)
+        CallerProfile->addTotalSamples(Count);
+
+      // Update the body samples for the current profile.
+      FProfile->addBodySamples(LineOffset, Discriminator, Count);
+    }
 
+    // Process the list of functions called at an indirect call site.
+    // These are all the targets that a function pointer (or virtual
+    // function) resolved at runtime.
     for (uint32_t J = 0; J < NumTargets; J++) {
       uint32_t HistVal;
       if (!GcovBuffer.readInt(HistVal))
@@ -647,24 +674,25 @@ SampleProfileReaderGCC::readOneFunctionProfile(const SourceStack &Stack,
 
       if (Update) {
         FunctionSamples &TargetProfile = Profiles[TargetName];
-        TargetProfile.addBodySamples(NewStack[0].Line,
-                                     NewStack[0].Discriminator, TargetCount);
+        TargetProfile.addCalledTargetSamples(LineOffset, Discriminator,
+                                             TargetName, TargetCount);
       }
     }
   }
 
-  for (uint32_t I = 0; I < NumCallSites; I++) {
+  // Process all the inlined callers into the current function. These
+  // are all the callsites that were inlined into this function.
+  for (uint32_t I = 0; I < NumCallsites; I++) {
     // The offset is encoded as:
     //   high 16 bits: line offset to the start of the function.
     //   low 16 bits: discriminator.
     uint32_t Offset;
     if (!GcovBuffer.readInt(Offset))
       return sampleprof_error::truncated;
-    SourceInfo Info(Name, "", "", 0, Offset >> 16, Offset & 0xffff);
-    SourceStack NewStack;
-    NewStack.push_back(Info);
-    NewStack.insert(NewStack.end(), Stack.begin(), Stack.end());
-    if (std::error_code EC = readOneFunctionProfile(NewStack, Update))
+    InlineCallStack NewStack;
+    NewStack.push_back(FProfile);
+    NewStack.insert(NewStack.end(), InlineStack.begin(), InlineStack.end());
+    if (std::error_code EC = readOneFunctionProfile(NewStack, Update, Offset))
       return EC;
   }
 

diff --git a/lib/ProfileData/SampleProfWriter.cpp b/lib/ProfileData/SampleProfWriter.cpp
index c95267ad976b896d8b69d85e755fe7529c2077cb..e6a4d7d132103aba5c527203e6ea399728851411 100644 (file)
--- a/lib/ProfileData/SampleProfWriter.cpp
+++ b/lib/ProfileData/SampleProfWriter.cpp
@@ -31,15 +31,15 @@ using namespace llvm;
 
 /// \brief Write samples to a text file.
 bool SampleProfileWriterText::write(StringRef FName, const FunctionSamples &S) {
-  if (S.empty())
-    return true;
-
-  OS << FName << ":" << S.getTotalSamples() << ":" << S.getHeadSamples()
-     << "\n";
+  OS << FName << ":" << S.getTotalSamples();
+  if (Indent == 0)
+    OS << ":" << S.getHeadSamples();
+  OS << "\n";
 
   for (const auto &I : S.getBodySamples()) {
     LineLocation Loc = I.first;
     const SampleRecord &Sample = I.second;
+    OS.indent(Indent + 1);
     if (Loc.Discriminator == 0)
       OS << Loc.LineOffset << ": ";
     else
@@ -52,6 +52,19 @@ bool SampleProfileWriterText::write(StringRef FName, const FunctionSamples &S) {
     OS << "\n";
   }
 
+  Indent += 1;
+  for (const auto &I : S.getCallsiteSamples()) {
+    CallsiteLocation Loc = I.first;
+    const FunctionSamples &CalleeSamples = I.second;
+    OS.indent(Indent);
+    if (Loc.Discriminator == 0)
+      OS << Loc.LineOffset << ": ";
+    else
+      OS << Loc.LineOffset << "." << Loc.Discriminator << ": ";
+    write(Loc.CalleeName, CalleeSamples);
+  }
+  Indent -= 1;
+
   return true;
 }
 

diff --git a/test/tools/llvm-profdata/Inputs/gcc-sample-profile.gcov b/test/tools/llvm-profdata/Inputs/gcc-sample-profile.gcov
new file mode 100644 (file)
index 0000000..0099280
Binary files /dev/null and b/test/tools/llvm-profdata/Inputs/gcc-sample-profile.gcov differ

diff --git a/test/tools/llvm-profdata/gcc-gcov-sample-profile.test b/test/tools/llvm-profdata/gcc-gcov-sample-profile.test
new file mode 100644 (file)
index 0000000..5d9c30f
--- /dev/null
+++ b/test/tools/llvm-profdata/gcc-gcov-sample-profile.test
@@ -0,0 +1,24 @@
+Tests for sample profiles encoded in GCC's gcov format.
+
+1- Show all functions. This profile has a single main() function with several
+   inlined callees.
+RUN: llvm-profdata show --sample %p/Inputs/gcc-sample-profile.gcov | FileCheck %s --check-prefix=SHOW1
+SHOW1: Function: main: 364084, 0, 6 sampled lines
+SHOW1: line offset: 2, discriminator: 3, inlined callee: _Z3fool: 243786, 0, 3 sampled lines
+SHOW1:   line offset: 1, discriminator: 3, inlined callee: _Z3bari: 0, 0, 2 sampled lines
+SHOW1:   line offset: 1, discriminator: 8, inlined callee: _Z3bari: 0, 0, 2 sampled lines
+SHOW1:   line offset: 1, discriminator: 7, inlined callee: _Z3bari: 98558, 0, 2 sampled lines
+
+2- Convert the profile to text encoding and check that they are both
+   identical.
+RUN: llvm-profdata merge --sample %p/Inputs/gcc-sample-profile.gcov --text -o - | llvm-profdata show --sample - -o %t-text
+RUN: llvm-profdata show --sample %p/Inputs/gcc-sample-profile.gcov -o %t-gcov
+RUN: diff %t-text %t-gcov
+
+4- Merge the gcov and text encodings of the profile and check that the
+   counters have doubled.
+RUN: llvm-profdata merge --sample --text %p/Inputs/gcc-sample-profile.gcov -o %t-gcov
+RUN: llvm-profdata merge --sample --text %p/Inputs/gcc-sample-profile.gcov %t-gcov -o - | FileCheck %s --check-prefix=MERGE1
+MERGE1: main:728168:0
+MERGE1: 2.3: 120298
+MERGE1: 2.3: _Z3fool:487572