lib/ProfileData/SampleProfWriter.cpp

   1 //===- SampleProfWriter.cpp - Write LLVM sample profile data --------------===//
   2 //
   3 //                      The LLVM Compiler Infrastructure
   4 //
   5 // This file is distributed under the University of Illinois Open Source
   6 // License. See LICENSE.TXT for details.
   7 //
   8 //===----------------------------------------------------------------------===//
   9 //
  10 // This file implements the class that writes LLVM sample profiles. It
  11 // supports two file formats: text and binary. The textual representation
  12 // is useful for debugging and testing purposes. The binary representation
  13 // is more compact, resulting in smaller file sizes. However, they can
  14 // both be used interchangeably.
  15 //
  16 // See lib/ProfileData/SampleProfReader.cpp for documentation on each of the
  17 // supported formats.
  18 //
  19 //===----------------------------------------------------------------------===//
  20
  21 #include "llvm/ProfileData/SampleProfWriter.h"
  22 #include "llvm/ADT/StringRef.h"
  23 #include "llvm/ProfileData/ProfileCommon.h"
  24 #include "llvm/ProfileData/SampleProf.h"
  25 #include "llvm/Support/Endian.h"
  26 #include "llvm/Support/EndianStream.h"
  27 #include "llvm/Support/ErrorOr.h"
  28 #include "llvm/Support/FileSystem.h"
  29 #include "llvm/Support/LEB128.h"
  30 #include "llvm/Support/MD5.h"
  31 #include "llvm/Support/raw_ostream.h"
  32 #include <algorithm>
  33 #include <cstdint>
  34 #include <memory>
  35 #include <set>
  36 #include <system_error>
  37 #include <utility>
  38 #include <vector>
  39
  40 using namespace llvm;
  41 using namespace sampleprof;
  42
  43 std::error_code
  44 SampleProfileWriter::write(const StringMap<FunctionSamples> &ProfileMap) {
  45   if (std::error_code EC = writeHeader(ProfileMap))
  46     return EC;
  47
  48   // Sort the ProfileMap by total samples.
  49   typedef std::pair<StringRef, const FunctionSamples *> NameFunctionSamples;
  50   std::vector<NameFunctionSamples> V;
  51   for (const auto &I : ProfileMap)
  52     V.push_back(std::make_pair(I.getKey(), &I.second));
  53
  54   std::stable_sort(
  55       V.begin(), V.end(),
  56       [](const NameFunctionSamples &A, const NameFunctionSamples &B) {
  57         if (A.second->getTotalSamples() == B.second->getTotalSamples())
  58           return A.first > B.first;
  59         return A.second->getTotalSamples() > B.second->getTotalSamples();
  60       });
  61
  62   for (const auto &I : V) {
  63     if (std::error_code EC = write(*I.second))
  64       return EC;
  65   }
  66   return sampleprof_error::success;
  67 }
  68
  69 std::error_code SampleProfileWriterCompactBinary::write(
  70     const StringMap<FunctionSamples> &ProfileMap) {
  71   if (std::error_code EC = SampleProfileWriter::write(ProfileMap))
  72     return EC;
  73   if (std::error_code EC = writeFuncOffsetTable())
  74     return EC;
  75   return sampleprof_error::success;
  76 }
  77
  78 /// Write samples to a text file.
  79 ///
  80 /// Note: it may be tempting to implement this in terms of
  81 /// FunctionSamples::print().  Please don't.  The dump functionality is intended
  82 /// for debugging and has no specified form.
  83 ///
  84 /// The format used here is more structured and deliberate because
  85 /// it needs to be parsed by the SampleProfileReaderText class.
  86 std::error_code SampleProfileWriterText::write(const FunctionSamples &S) {
  87   auto &OS = *OutputStream;
  88   OS << S.getName() << ":" << S.getTotalSamples();
  89   if (Indent == 0)
  90     OS << ":" << S.getHeadSamples();
  91   OS << "\n";
  92
  93   SampleSorter<LineLocation, SampleRecord> SortedSamples(S.getBodySamples());
  94   for (const auto &I : SortedSamples.get()) {
  95     LineLocation Loc = I->first;
  96     const SampleRecord &Sample = I->second;
  97     OS.indent(Indent + 1);
  98     if (Loc.Discriminator == 0)
  99       OS << Loc.LineOffset << ": ";
 100     else
 101       OS << Loc.LineOffset << "." << Loc.Discriminator << ": ";
 102
 103     OS << Sample.getSamples();
 104
 105     for (const auto &J : Sample.getCallTargets())
 106       OS << " " << J.first() << ":" << J.second;
 107     OS << "\n";
 108   }
 109
 110   SampleSorter<LineLocation, FunctionSamplesMap> SortedCallsiteSamples(
 111       S.getCallsiteSamples());
 112   Indent += 1;
 113   for (const auto &I : SortedCallsiteSamples.get())
 114     for (const auto &FS : I->second) {
 115       LineLocation Loc = I->first;
 116       const FunctionSamples &CalleeSamples = FS.second;
 117       OS.indent(Indent);
 118       if (Loc.Discriminator == 0)
 119         OS << Loc.LineOffset << ": ";
 120       else
 121         OS << Loc.LineOffset << "." << Loc.Discriminator << ": ";
 122       if (std::error_code EC = write(CalleeSamples))
 123         return EC;
 124     }
 125   Indent -= 1;
 126
 127   return sampleprof_error::success;
 128 }
 129
 130 std::error_code SampleProfileWriterBinary::writeNameIdx(StringRef FName) {
 131   const auto &ret = NameTable.find(FName);
 132   if (ret == NameTable.end())
 133     return sampleprof_error::truncated_name_table;
 134   encodeULEB128(ret->second, *OutputStream);
 135   return sampleprof_error::success;
 136 }
 137
 138 void SampleProfileWriterBinary::addName(StringRef FName) {
 139   NameTable.insert(std::make_pair(FName, 0));
 140 }
 141
 142 void SampleProfileWriterBinary::addNames(const FunctionSamples &S) {
 143   // Add all the names in indirect call targets.
 144   for (const auto &I : S.getBodySamples()) {
 145     const SampleRecord &Sample = I.second;
 146     for (const auto &J : Sample.getCallTargets())
 147       addName(J.first());
 148   }
 149
 150   // Recursively add all the names for inlined callsites.
 151   for (const auto &J : S.getCallsiteSamples())
 152     for (const auto &FS : J.second) {
 153       const FunctionSamples &CalleeSamples = FS.second;
 154       addName(CalleeSamples.getName());
 155       addNames(CalleeSamples);
 156     }
 157 }
 158
 159 void SampleProfileWriterBinary::stablizeNameTable(std::set<StringRef> &V) {
 160   // Sort the names to make NameTable deterministic.
 161   for (const auto &I : NameTable)
 162     V.insert(I.first);
 163   int i = 0;
 164   for (const StringRef &N : V)
 165     NameTable[N] = i++;
 166 }
 167
 168 std::error_code SampleProfileWriterRawBinary::writeNameTable() {
 169   auto &OS = *OutputStream;
 170   std::set<StringRef> V;
 171   stablizeNameTable(V);
 172
 173   // Write out the name table.
 174   encodeULEB128(NameTable.size(), OS);
 175   for (auto N : V) {
 176     OS << N;
 177     encodeULEB128(0, OS);
 178   }
 179   return sampleprof_error::success;
 180 }
 181
 182 std::error_code SampleProfileWriterCompactBinary::writeFuncOffsetTable() {
 183   auto &OS = *OutputStream;
 184
 185   // Fill the slot remembered by TableOffset with the offset of FuncOffsetTable.
 186   auto &OFS = static_cast<raw_fd_ostream &>(OS);
 187   uint64_t FuncOffsetTableStart = OS.tell();
 188   if (OFS.seek(TableOffset) == (uint64_t)-1)
 189     return sampleprof_error::ostream_seek_unsupported;
 190   support::endian::Writer Writer(*OutputStream, support::little);
 191   Writer.write(FuncOffsetTableStart);
 192   if (OFS.seek(FuncOffsetTableStart) == (uint64_t)-1)
 193     return sampleprof_error::ostream_seek_unsupported;
 194
 195   // Write out the table size.
 196   encodeULEB128(FuncOffsetTable.size(), OS);
 197
 198   // Write out FuncOffsetTable.
 199   for (auto entry : FuncOffsetTable) {
 200     writeNameIdx(entry.first);
 201     encodeULEB128(entry.second, OS);
 202   }
 203   return sampleprof_error::success;
 204 }
 205
 206 std::error_code SampleProfileWriterCompactBinary::writeNameTable() {
 207   auto &OS = *OutputStream;
 208   std::set<StringRef> V;
 209   stablizeNameTable(V);
 210
 211   // Write out the name table.
 212   encodeULEB128(NameTable.size(), OS);
 213   for (auto N : V) {
 214     encodeULEB128(MD5Hash(N), OS);
 215   }
 216   return sampleprof_error::success;
 217 }
 218
 219 std::error_code SampleProfileWriterRawBinary::writeMagicIdent() {
 220   auto &OS = *OutputStream;
 221   // Write file magic identifier.
 222   encodeULEB128(SPMagic(), OS);
 223   encodeULEB128(SPVersion(), OS);
 224   return sampleprof_error::success;
 225 }
 226
 227 std::error_code SampleProfileWriterCompactBinary::writeMagicIdent() {
 228   auto &OS = *OutputStream;
 229   // Write file magic identifier.
 230   encodeULEB128(SPMagic(SPF_Compact_Binary), OS);
 231   encodeULEB128(SPVersion(), OS);
 232   return sampleprof_error::success;
 233 }
 234
 235 std::error_code SampleProfileWriterBinary::writeHeader(
 236     const StringMap<FunctionSamples> &ProfileMap) {
 237   writeMagicIdent();
 238
 239   computeSummary(ProfileMap);
 240   if (auto EC = writeSummary())
 241     return EC;
 242
 243   // Generate the name table for all the functions referenced in the profile.
 244   for (const auto &I : ProfileMap) {
 245     addName(I.first());
 246     addNames(I.second);
 247   }
 248
 249   writeNameTable();
 250   return sampleprof_error::success;
 251 }
 252
 253 std::error_code SampleProfileWriterCompactBinary::writeHeader(
 254     const StringMap<FunctionSamples> &ProfileMap) {
 255   support::endian::Writer Writer(*OutputStream, support::little);
 256   if (auto EC = SampleProfileWriterBinary::writeHeader(ProfileMap))
 257     return EC;
 258
 259   // Reserve a slot for the offset of function offset table. The slot will
 260   // be populated with the offset of FuncOffsetTable later.
 261   TableOffset = OutputStream->tell();
 262   Writer.write(static_cast<uint64_t>(-2));
 263   return sampleprof_error::success;
 264 }
 265
 266 std::error_code SampleProfileWriterBinary::writeSummary() {
 267   auto &OS = *OutputStream;
 268   encodeULEB128(Summary->getTotalCount(), OS);
 269   encodeULEB128(Summary->getMaxCount(), OS);
 270   encodeULEB128(Summary->getMaxFunctionCount(), OS);
 271   encodeULEB128(Summary->getNumCounts(), OS);
 272   encodeULEB128(Summary->getNumFunctions(), OS);
 273   std::vector<ProfileSummaryEntry> &Entries = Summary->getDetailedSummary();
 274   encodeULEB128(Entries.size(), OS);
 275   for (auto Entry : Entries) {
 276     encodeULEB128(Entry.Cutoff, OS);
 277     encodeULEB128(Entry.MinCount, OS);
 278     encodeULEB128(Entry.NumCounts, OS);
 279   }
 280   return sampleprof_error::success;
 281 }
 282 std::error_code SampleProfileWriterBinary::writeBody(const FunctionSamples &S) {
 283   auto &OS = *OutputStream;
 284
 285   if (std::error_code EC = writeNameIdx(S.getName()))
 286     return EC;
 287
 288   encodeULEB128(S.getTotalSamples(), OS);
 289
 290   // Emit all the body samples.
 291   encodeULEB128(S.getBodySamples().size(), OS);
 292   for (const auto &I : S.getBodySamples()) {
 293     LineLocation Loc = I.first;
 294     const SampleRecord &Sample = I.second;
 295     encodeULEB128(Loc.LineOffset, OS);
 296     encodeULEB128(Loc.Discriminator, OS);
 297     encodeULEB128(Sample.getSamples(), OS);
 298     encodeULEB128(Sample.getCallTargets().size(), OS);
 299     for (const auto &J : Sample.getCallTargets()) {
 300       StringRef Callee = J.first();
 301       uint64_t CalleeSamples = J.second;
 302       if (std::error_code EC = writeNameIdx(Callee))
 303         return EC;
 304       encodeULEB128(CalleeSamples, OS);
 305     }
 306   }
 307
 308   // Recursively emit all the callsite samples.
 309   uint64_t NumCallsites = 0;
 310   for (const auto &J : S.getCallsiteSamples())
 311     NumCallsites += J.second.size();
 312   encodeULEB128(NumCallsites, OS);
 313   for (const auto &J : S.getCallsiteSamples())
 314     for (const auto &FS : J.second) {
 315       LineLocation Loc = J.first;
 316       const FunctionSamples &CalleeSamples = FS.second;
 317       encodeULEB128(Loc.LineOffset, OS);
 318       encodeULEB128(Loc.Discriminator, OS);
 319       if (std::error_code EC = writeBody(CalleeSamples))
 320         return EC;
 321     }
 322
 323   return sampleprof_error::success;
 324 }
 325
 326 /// Write samples of a top-level function to a binary file.
 327 ///
 328 /// \returns true if the samples were written successfully, false otherwise.
 329 std::error_code SampleProfileWriterBinary::write(const FunctionSamples &S) {
 330   encodeULEB128(S.getHeadSamples(), *OutputStream);
 331   return writeBody(S);
 332 }
 333
 334 std::error_code
 335 SampleProfileWriterCompactBinary::write(const FunctionSamples &S) {
 336   uint64_t Offset = OutputStream->tell();
 337   StringRef Name = S.getName();
 338   FuncOffsetTable[Name] = Offset;
 339   encodeULEB128(S.getHeadSamples(), *OutputStream);
 340   return writeBody(S);
 341 }
 342
 343 /// Create a sample profile file writer based on the specified format.
 344 ///
 345 /// \param Filename The file to create.
 346 ///
 347 /// \param Format Encoding format for the profile file.
 348 ///
 349 /// \returns an error code indicating the status of the created writer.
 350 ErrorOr<std::unique_ptr<SampleProfileWriter>>
 351 SampleProfileWriter::create(StringRef Filename, SampleProfileFormat Format) {
 352   std::error_code EC;
 353   std::unique_ptr<raw_ostream> OS;
 354   if (Format == SPF_Binary || Format == SPF_Compact_Binary)
 355     OS.reset(new raw_fd_ostream(Filename, EC, sys::fs::F_None));
 356   else
 357     OS.reset(new raw_fd_ostream(Filename, EC, sys::fs::F_Text));
 358   if (EC)
 359     return EC;
 360
 361   return create(OS, Format);
 362 }
 363
 364 /// Create a sample profile stream writer based on the specified format.
 365 ///
 366 /// \param OS The output stream to store the profile data to.
 367 ///
 368 /// \param Format Encoding format for the profile file.
 369 ///
 370 /// \returns an error code indicating the status of the created writer.
 371 ErrorOr<std::unique_ptr<SampleProfileWriter>>
 372 SampleProfileWriter::create(std::unique_ptr<raw_ostream> &OS,
 373                             SampleProfileFormat Format) {
 374   std::error_code EC;
 375   std::unique_ptr<SampleProfileWriter> Writer;
 376
 377   if (Format == SPF_Binary)
 378     Writer.reset(new SampleProfileWriterRawBinary(OS));
 379   else if (Format == SPF_Compact_Binary)
 380     Writer.reset(new SampleProfileWriterCompactBinary(OS));
 381   else if (Format == SPF_Text)
 382     Writer.reset(new SampleProfileWriterText(OS));
 383   else if (Format == SPF_GCC)
 384     EC = sampleprof_error::unsupported_writing_format;
 385   else
 386     EC = sampleprof_error::unrecognized_format;
 387
 388   if (EC)
 389     return EC;
 390
 391   return std::move(Writer);
 392 }
 393
 394 void SampleProfileWriter::computeSummary(
 395     const StringMap<FunctionSamples> &ProfileMap) {
 396   SampleProfileSummaryBuilder Builder(ProfileSummaryBuilder::DefaultCutoffs);
 397   for (const auto &I : ProfileMap) {
 398     const FunctionSamples &Profile = I.second;
 399     Builder.addRecord(Profile);
 400   }
 401   Summary = Builder.getSummary();
 402 }