[llvm-profdata] Support -detailed-summary for Sample Profile

[wenlei]

Summary

Add -detailed-summary support for sample profile dump to match that of instrumentation profile.

[wmi]

llvm/tools/llvm-profdata/llvm-profdata.cpp

839

Can we make it a member function of ProfileSummary? We may use the dump function in other places.

[wenlei]

Update

[wenlei]

llvm/tools/llvm-profdata/llvm-profdata.cpp

839

Sure, moved both to be member functions.

[wmi]

llvm/lib/IR/ProfileSummary.cpp

209–215

Why we need printSummary and printDetailedSummary instead of one print function? printSummary seems only print very limited information in ProfileSummary.

[wenlei]

llvm/lib/IR/ProfileSummary.cpp

209–215

That was mostly for sharing printDetailedSummary between InstrProf and SampleProf for llvm-profdata. The printSummary part is always printed for InstrProf, so separated out in order to share printDetailedSummary.

[wmi]

llvm/lib/IR/ProfileSummary.cpp

209–215

I mean can we have one print function, printSummary for example, and dump everything including the detail inside of it? Is it a problem for InstrProf test?

[wenlei]

llvm/lib/IR/ProfileSummary.cpp

209–215

Yeah, it causes problem for InstrProf tests. Stuff in printSummary is always printed for InstrProfile even without detailed-summary flag, and it's interleaved with other outputs. Tests actually leverage the interleaved structure for checking, so if we combine printSummary with printDetailedSummary, many tests need to be updated, and detail-summary need to be added to many tests too. We could do that, I was trying to avoid that churn though...

[wmi]

llvm/lib/IR/ProfileSummary.cpp

209–215

Ok, then can we let printDetailedSummary just print stuff in DetailedSummary and let printSummary print more information like NumCounts, TotalCounts, ....? It makes what those functions should print a little clearer.

InstrProf can print NumCounts and TotalCounts directly since it has already done that in this way.

[wenlei]

restructure prints

[wenlei]

llvm/lib/IR/ProfileSummary.cpp

209–215

Good point, changed.

[wmi]

Thanks, LGTM.

llvm/include/llvm/IR/ProfileSummary.h

 #include <algorithm>
 #include <cstdint>
 #include <vector>
 
 namespace llvm {
 
 class LLVMContext;
 class Metadata;
+class raw_ostream;
 
 // The profile summary is one or more (Cutoff, MinCount, NumCounts) triplets.
 // The semantics of counts depend on the type of profile. For instrumentation
 // profile, counts are block counts and for sample profile, counts are
 // per-line samples. Given a target counts percentile, we compute the minimum
 // number of counts needed to reach this target and the minimum among these
 // counts.
 struct ProfileSummaryEntry {
public:
   uint32_t getNumFunctions() { return NumFunctions; }
   uint64_t getMaxFunctionCount() { return MaxFunctionCount; }
   uint32_t getNumCounts() { return NumCounts; }
   uint64_t getTotalCount() { return TotalCount; }
   uint64_t getMaxCount() { return MaxCount; }
   uint64_t getMaxInternalCount() { return MaxInternalCount; }
   void setPartialProfile(bool PP) { Partial = PP; }
   bool isPartialProfile() { return Partial; }
+  void printSummary(raw_ostream &OS);
+  void printDetailedSummary(raw_ostream &OS);
 };
 
 } // end namespace llvm
 
 #endif // LLVM_IR_PROFILESUMMARY_H

llvm/lib/IR/ProfileSummary.cpp

ProfileSummary *ProfileSummary::getFromMD(Metadata *MD) {
 
   SummaryEntryVector Summary;
   if (!getSummaryFromMD(dyn_cast<MDTuple>(Tuple->getOperand(i++)), Summary))
     return nullptr;
   return new ProfileSummary(SummaryKind, std::move(Summary), TotalCount,
                             MaxCount, MaxInternalCount, MaxFunctionCount,
                             NumCounts, NumFunctions, IsPartialProfile);
 }
+
+void ProfileSummary::printSummary(raw_ostream &OS) {
+  OS << "Total functions: " << NumFunctions << "\n";
+  OS << "Maximum function count: " << MaxFunctionCount << "\n";
+  OS << "Maximum block count: " << MaxCount << "\n";
+  OS << "Total number of blocks: " << NumCounts << "\n";
+  OS << "Total count: " << TotalCount << "\n";
+}
+
+void ProfileSummary::printDetailedSummary(raw_ostream &OS) {
+  OS << "Detailed summary:\n";
+  for (auto Entry : DetailedSummary) {
+    OS << Entry.NumCounts << " blocks with count >= " << Entry.MinCount
+       << " account for "
+       << format("%0.6g", (float)Entry.Cutoff / Scale * 100)
+       << " percentage of the total counts.\n";
+  }
+}

llvm/test/tools/llvm-profdata/general.proftext

 # RUN: llvm-profdata show %t.profdata.dense | FileCheck %s -check-prefix=SUMMARY
 # SUMMARY-NOT: Counters:
 # SUMMARY-NOT: Functions shown:
 # SUMMARY: Total functions: 4
 # SUMMARY: Maximum function count: 2305843009213693952
 # SUMMARY: Maximum internal block count: 1152921504606846976
 
 # RUN: llvm-profdata show --detailed-summary %t.profdata.dense | FileCheck %s -check-prefix=DETAILED-SUMMARY
-# DETAILED-SUMMARY: Detailed summary:
 # DETAILED-SUMMARY: Total number of blocks: 10
 # DETAILED-SUMMARY: Total count: 4539628424389557499
+# DETAILED-SUMMARY: Detailed summary:
 # DETAILED-SUMMARY: 3 blocks with count >= 576460752303423488 account for 80 percentage of the total counts.
 # DETAILED-SUMMARY: 4 blocks with count >= 288230376151711744 account for 90 percentage of the total counts.
 # DETAILED-SUMMARY: 4 blocks with count >= 288230376151711744 account for 95 percentage of the total counts.
 # DETAILED-SUMMARY: 6 blocks with count >= 72057594037927936 account for 99 percentage of the total counts.
 # DETAILED-SUMMARY: 6 blocks with count >= 72057594037927936 account for 99.9 percentage of the total counts.
 # DETAILED-SUMMARY: 6 blocks with count >= 72057594037927936 account for 99.99 percentage of the total counts.
 # DETAILED-SUMMARY: 6 blocks with count >= 72057594037927936 account for 99.999 percentage of the total counts.
 
 # RUN: llvm-profdata show --detailed-summary --detailed-summary-cutoffs=600000 %t.profdata | FileCheck %s -check-prefix=DETAILED-SUMMARY-2
 # DETAILED-SUMMARY-2: 2 blocks with count >= 1152921504606846976 account for 60 percentage of the total counts.
 #
 # RUN: llvm-profdata show --detailed-summary --detailed-summary-cutoffs=600000,900000,999999 %t.profdata | FileCheck %s -check-prefix=DETAILED-SUMMARY-3
 # DETAILED-SUMMARY-3: 2 blocks with count >= 1152921504606846976 account for 60 percentage of the total counts.
 # DETAILED-SUMMARY-3: 4 blocks with count >= 288230376151711744 account for 90 percentage of the total counts.
 # DETAILED-SUMMARY-3: 6 blocks with count >= 72057594037927936 account for 99.9999 percentage of the total counts.

llvm/test/tools/llvm-profdata/sample-summary.test

+; RUN: llvm-profdata show -sample -detailed-summary %S/Inputs/sample-profile.proftext | FileCheck %s
+
+; CHECK: Total functions: 3
+; CHECK-NEXT: Maximum function count: 1437
+; CHECK-NEXT: Maximum block count: 2080
+; CHECK-NEXT: Total number of blocks: 11
+; CHECK-NEXT: Total count: 12943
+; CHECK-NEXT: Detailed summary:
+; CHECK-NEXT: 1 blocks with count >= 2080 account for 1 percentage of the total counts.
+; CHECK-NEXT: 1 blocks with count >= 2080 account for 10 percentage of the total counts.
+; CHECK-NEXT: 2 blocks with count >= 2064 account for 20 percentage of the total counts.
+; CHECK-NEXT: 2 blocks with count >= 2064 account for 30 percentage of the total counts.
+; CHECK-NEXT: 3 blocks with count >= 2000 account for 40 percentage of the total counts.
+; CHECK-NEXT: 4 blocks with count >= 1437 account for 50 percentage of the total counts.
+; CHECK-NEXT: 6 blocks with count >= 1075 account for 60 percentage of the total counts.
+; CHECK-NEXT: 6 blocks with count >= 1075 account for 70 percentage of the total counts.
+; CHECK-NEXT: 7 blocks with count >= 1000 account for 80 percentage of the total counts.
+; CHECK-NEXT: 11 blocks with count >= 534 account for 90 percentage of the total counts.
+; CHECK-NEXT: 11 blocks with count >= 534 account for 95 percentage of the total counts.
+; CHECK-NEXT: 11 blocks with count >= 534 account for 99 percentage of the total counts.
+; CHECK-NEXT: 11 blocks with count >= 534 account for 99.9 percentage of the total counts.
+; CHECK-NEXT: 11 blocks with count >= 534 account for 99.99 percentage of the total counts.
+; CHECK-NEXT: 11 blocks with count >= 534 account for 99.999 percentage of the total counts.
+; CHECK-NEXT: 11 blocks with count >= 534 account for 99.9999 percentage of the total counts.

llvm/tools/llvm-profdata/llvm-profdata.cpp

 }
 
 static int showInstrProfile(const std::string &Filename, bool ShowCounts,
                             uint32_t TopN, bool ShowIndirectCallTargets,
                             bool ShowMemOPSizes, bool ShowDetailedSummary,
                             std::vector<uint32_t> DetailedSummaryCutoffs,
                             bool ShowAllFunctions, bool ShowCS,
                             uint64_t ValueCutoff, bool OnlyListBelow,
                             const std::string &ShowFunction, bool TextFormat,
                             raw_fd_ostream &OS) {
   auto ReaderOrErr = InstrProfReader::create(Filename);
   std::vector<uint32_t> Cutoffs = std::move(DetailedSummaryCutoffs);
   if (ShowDetailedSummary && Cutoffs.empty()) {
     Cutoffs = {800000, 900000, 950000, 990000, 999000, 999900, 999990};
   }
   InstrProfSummaryBuilder Builder(std::move(Cutoffs));
   if (Error E = ReaderOrErr.takeError())
static int showInstrProfile(const std::string &Filename, bool ShowCounts,
   }
 
   if (ShownFunctions && ShowMemOPSizes) {
     OS << "Statistics for memory intrinsic calls sizes profile:\n";
     showValueSitesStats(OS, IPVK_MemOPSize, VPStats[IPVK_MemOPSize]);
   }
 
   if (ShowDetailedSummary) {
-    OS << "Detailed summary:\n";
     OS << "Total number of blocks: " << PS->getNumCounts() << "\n";
     OS << "Total count: " << PS->getTotalCount() << "\n";
-    for (auto Entry : PS->getDetailedSummary()) {
-      OS << Entry.NumCounts << " blocks with count >= " << Entry.MinCount
-         << " account for "
-         << format("%0.6g", (float)Entry.Cutoff / ProfileSummary::Scale * 100)
-         << " percentage of the total counts.\n";
-    }
+    PS->printDetailedSummary(OS);
   }
   return 0;
 }
 
 static void showSectionInfo(sampleprof::SampleProfileReader *Reader,
                             raw_fd_ostream &OS) {
   if (!Reader->dumpSectionInfo(OS)) {
     WithColor::warning() << "-show-sec-info-only is only supported for "
                          << "sample profile in extbinary format and is "
                          << "ignored for other formats.\n";
     return;
   }
 }
 
 static int showSampleProfile(const std::string &Filename, bool ShowCounts,
-                             bool ShowAllFunctions,
+                             bool ShowAllFunctions, bool ShowDetailedSummary,
                              const std::string &ShowFunction,
                              bool ShowProfileSymbolList,
                              bool ShowSectionInfoOnly, raw_fd_ostream &OS) {
   using namespace sampleprof;
   LLVMContext Context;
   auto ReaderOrErr = SampleProfileReader::create(Filename, Context);
   if (std::error_code EC = ReaderOrErr.getError())
     exitWithErrorCode(EC, Filename);
else
     Reader->dumpFunctionProfile(ShowFunction, OS);
 
   if (ShowProfileSymbolList) {
     std::unique_ptr<sampleprof::ProfileSymbolList> ReaderList =
         Reader->getProfileSymbolList();
     ReaderList->dump(OS);
   }
 
+  if (ShowDetailedSummary) {
+    auto &PS = Reader->getSummary();
+    PS.printSummary(OS);
+    PS.printDetailedSummary(OS);
+  }
+
   return 0;
 }
 
 static int show_main(int argc, const char *argv[]) {
   cl::opt<std::string> Filename(cl::Positional, cl::Required,
                                 cl::desc("<profdata-file>"));
 
   cl::opt<bool> ShowCounts("counts", cl::init(false),
static int show_main(int argc, const char *argv[]) {
   if (ProfileKind == instr)
     return showInstrProfile(Filename, ShowCounts, TopNFunctions,
                             ShowIndirectCallTargets, ShowMemOPSizes,
                             ShowDetailedSummary, DetailedSummaryCutoffs,
                             ShowAllFunctions, ShowCS, ValueCutoff,
                             OnlyListBelow, ShowFunction, TextFormat, OS);
   else
     return showSampleProfile(Filename, ShowCounts, ShowAllFunctions,
-                             ShowFunction, ShowProfileSymbolList,
-                             ShowSectionInfoOnly, OS);
+                             ShowDetailedSummary, ShowFunction,
+                             ShowProfileSymbolList, ShowSectionInfoOnly, OS);
 }
 
 int main(int argc, const char *argv[]) {
   InitLLVM X(argc, argv);
 
   StringRef ProgName(sys::path::filename(argv[0]));
   if (argc > 1) {
     int (*func)(int, const char *[]) = nullptr;