Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 1 | //===--- GlobalModuleIndex.cpp - Global Module Index ------------*- C++ -*-===// |
| 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 GlobalModuleIndex class. |
| 11 | // |
| 12 | //===----------------------------------------------------------------------===// |
| 13 | |
| 14 | #include "ASTReaderInternals.h" |
| 15 | #include "clang/Basic/FileManager.h" |
| 16 | #include "clang/Basic/OnDiskHashTable.h" |
| 17 | #include "clang/Serialization/ASTBitCodes.h" |
| 18 | #include "clang/Serialization/GlobalModuleIndex.h" |
| 19 | #include "llvm/ADT/DenseMap.h" |
| 20 | #include "llvm/ADT/MapVector.h" |
| 21 | #include "llvm/ADT/SmallString.h" |
| 22 | #include "llvm/ADT/StringExtras.h" |
| 23 | #include "llvm/Bitcode/BitstreamReader.h" |
| 24 | #include "llvm/Bitcode/BitstreamWriter.h" |
Douglas Gregor | 5100135 | 2013-01-23 22:45:24 +0000 | [diff] [blame] | 25 | #include "llvm/Support/FileSystem.h" |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 26 | #include "llvm/Support/LockFileManager.h" |
| 27 | #include "llvm/Support/MemoryBuffer.h" |
| 28 | #include "llvm/Support/PathV2.h" |
NAKAMURA Takumi | 1d32133 | 2013-01-25 01:47:07 +0000 | [diff] [blame] | 29 | #include <cstdio> |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 30 | using namespace clang; |
| 31 | using namespace serialization; |
| 32 | |
| 33 | //----------------------------------------------------------------------------// |
| 34 | // Shared constants |
| 35 | //----------------------------------------------------------------------------// |
| 36 | namespace { |
| 37 | enum { |
| 38 | /// \brief The block containing the index. |
| 39 | GLOBAL_INDEX_BLOCK_ID = llvm::bitc::FIRST_APPLICATION_BLOCKID |
| 40 | }; |
| 41 | |
| 42 | /// \brief Describes the record types in the index. |
| 43 | enum IndexRecordTypes { |
| 44 | /// \brief Contains version information and potentially other metadata, |
| 45 | /// used to determine if we can read this global index file. |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 46 | INDEX_METADATA, |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 47 | /// \brief Describes a module, including its file name and dependencies. |
| 48 | MODULE, |
| 49 | /// \brief The index for identifiers. |
| 50 | IDENTIFIER_INDEX |
| 51 | }; |
| 52 | } |
| 53 | |
| 54 | /// \brief The name of the global index file. |
| 55 | static const char * const IndexFileName = "modules.idx"; |
| 56 | |
| 57 | /// \brief The global index file version. |
| 58 | static const unsigned CurrentVersion = 1; |
| 59 | |
| 60 | //----------------------------------------------------------------------------// |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 61 | // Global module index reader. |
| 62 | //----------------------------------------------------------------------------// |
| 63 | |
| 64 | namespace { |
| 65 | |
| 66 | /// \brief Trait used to read the identifier index from the on-disk hash |
| 67 | /// table. |
| 68 | class IdentifierIndexReaderTrait { |
| 69 | public: |
| 70 | typedef StringRef external_key_type; |
| 71 | typedef StringRef internal_key_type; |
| 72 | typedef SmallVector<unsigned, 2> data_type; |
| 73 | |
| 74 | static bool EqualKey(const internal_key_type& a, const internal_key_type& b) { |
| 75 | return a == b; |
| 76 | } |
| 77 | |
| 78 | static unsigned ComputeHash(const internal_key_type& a) { |
| 79 | return llvm::HashString(a); |
| 80 | } |
| 81 | |
| 82 | static std::pair<unsigned, unsigned> |
| 83 | ReadKeyDataLength(const unsigned char*& d) { |
| 84 | using namespace clang::io; |
| 85 | unsigned KeyLen = ReadUnalignedLE16(d); |
| 86 | unsigned DataLen = ReadUnalignedLE16(d); |
| 87 | return std::make_pair(KeyLen, DataLen); |
| 88 | } |
| 89 | |
| 90 | static const internal_key_type& |
| 91 | GetInternalKey(const external_key_type& x) { return x; } |
| 92 | |
| 93 | static const external_key_type& |
| 94 | GetExternalKey(const internal_key_type& x) { return x; } |
| 95 | |
| 96 | static internal_key_type ReadKey(const unsigned char* d, unsigned n) { |
| 97 | return StringRef((const char *)d, n); |
| 98 | } |
| 99 | |
| 100 | static data_type ReadData(const internal_key_type& k, |
| 101 | const unsigned char* d, |
| 102 | unsigned DataLen) { |
| 103 | using namespace clang::io; |
| 104 | |
| 105 | data_type Result; |
| 106 | while (DataLen > 0) { |
| 107 | unsigned ID = ReadUnalignedLE32(d); |
| 108 | Result.push_back(ID); |
| 109 | DataLen -= 4; |
| 110 | } |
| 111 | |
| 112 | return Result; |
| 113 | } |
| 114 | }; |
| 115 | |
| 116 | typedef OnDiskChainedHashTable<IdentifierIndexReaderTrait> IdentifierIndexTable; |
| 117 | |
| 118 | /// \brief Module information as it was loaded from the index file. |
| 119 | struct LoadedModuleInfo { |
| 120 | const FileEntry *File; |
| 121 | SmallVector<unsigned, 2> Dependencies; |
| 122 | SmallVector<unsigned, 2> ImportedBy; |
| 123 | }; |
| 124 | |
| 125 | } |
| 126 | |
| 127 | GlobalModuleIndex::GlobalModuleIndex(FileManager &FileMgr, |
| 128 | llvm::MemoryBuffer *Buffer, |
| 129 | llvm::BitstreamCursor Cursor) |
Douglas Gregor | 188bdcd | 2013-01-25 23:32:03 +0000 | [diff] [blame^] | 130 | : Buffer(Buffer), IdentifierIndex(), |
| 131 | NumIdentifierLookups(), NumIdentifierLookupHits() |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 132 | { |
| 133 | typedef llvm::DenseMap<unsigned, LoadedModuleInfo> LoadedModulesMap; |
| 134 | LoadedModulesMap LoadedModules; |
| 135 | |
| 136 | // Read the global index. |
| 137 | unsigned LargestID = 0; |
| 138 | bool InGlobalIndexBlock = false; |
| 139 | bool Done = false; |
| 140 | bool AnyOutOfDate = false; |
| 141 | while (!Done) { |
| 142 | llvm::BitstreamEntry Entry = Cursor.advance(); |
| 143 | |
| 144 | switch (Entry.Kind) { |
| 145 | case llvm::BitstreamEntry::Error: |
| 146 | return; |
| 147 | |
| 148 | case llvm::BitstreamEntry::EndBlock: |
| 149 | if (InGlobalIndexBlock) { |
| 150 | InGlobalIndexBlock = false; |
| 151 | Done = true; |
| 152 | continue; |
| 153 | } |
| 154 | return; |
| 155 | |
| 156 | |
| 157 | case llvm::BitstreamEntry::Record: |
| 158 | // Entries in the global index block are handled below. |
| 159 | if (InGlobalIndexBlock) |
| 160 | break; |
| 161 | |
| 162 | return; |
| 163 | |
| 164 | case llvm::BitstreamEntry::SubBlock: |
| 165 | if (!InGlobalIndexBlock && Entry.ID == GLOBAL_INDEX_BLOCK_ID) { |
| 166 | if (Cursor.EnterSubBlock(GLOBAL_INDEX_BLOCK_ID)) |
| 167 | return; |
| 168 | |
| 169 | InGlobalIndexBlock = true; |
| 170 | } else if (Cursor.SkipBlock()) { |
| 171 | return; |
| 172 | } |
| 173 | continue; |
| 174 | } |
| 175 | |
| 176 | SmallVector<uint64_t, 64> Record; |
| 177 | StringRef Blob; |
| 178 | switch ((IndexRecordTypes)Cursor.readRecord(Entry.ID, Record, &Blob)) { |
| 179 | case INDEX_METADATA: |
| 180 | // Make sure that the version matches. |
| 181 | if (Record.size() < 1 || Record[0] != CurrentVersion) |
| 182 | return; |
| 183 | break; |
| 184 | |
| 185 | case MODULE: { |
| 186 | unsigned Idx = 0; |
| 187 | unsigned ID = Record[Idx++]; |
| 188 | if (ID > LargestID) |
| 189 | LargestID = ID; |
| 190 | |
| 191 | off_t Size = Record[Idx++]; |
| 192 | time_t ModTime = Record[Idx++]; |
| 193 | |
| 194 | // File name. |
| 195 | unsigned NameLen = Record[Idx++]; |
| 196 | llvm::SmallString<64> FileName(Record.begin() + Idx, |
| 197 | Record.begin() + Idx + NameLen); |
| 198 | Idx += NameLen; |
| 199 | |
| 200 | // Dependencies |
| 201 | unsigned NumDeps = Record[Idx++]; |
| 202 | llvm::SmallVector<unsigned, 2> |
| 203 | Dependencies(Record.begin() + Idx, Record.begin() + Idx + NumDeps); |
| 204 | |
| 205 | // Find the file. If we can't find it, ignore it. |
| 206 | const FileEntry *File = FileMgr.getFile(FileName); |
| 207 | if (!File) { |
| 208 | AnyOutOfDate = true; |
| 209 | break; |
| 210 | } |
| 211 | |
| 212 | // If the module file is newer than the index, ignore it. |
| 213 | if (File->getSize() != Size || File->getModificationTime() != ModTime) { |
| 214 | AnyOutOfDate = true; |
| 215 | break; |
| 216 | } |
| 217 | |
| 218 | // Record this module. The dependencies will be resolved later. |
| 219 | LoadedModuleInfo &Info = LoadedModules[ID]; |
| 220 | Info.File = File; |
| 221 | Info.Dependencies.swap(Dependencies); |
| 222 | break; |
| 223 | } |
| 224 | |
| 225 | case IDENTIFIER_INDEX: |
| 226 | // Wire up the identifier index. |
| 227 | if (Record[0]) { |
| 228 | IdentifierIndex = IdentifierIndexTable::Create( |
| 229 | (const unsigned char *)Blob.data() + Record[0], |
| 230 | (const unsigned char *)Blob.data(), |
| 231 | IdentifierIndexReaderTrait()); |
| 232 | } |
| 233 | break; |
| 234 | } |
| 235 | } |
| 236 | |
| 237 | // If there are any modules that have gone out-of-date, prune out any modules |
| 238 | // that depend on them. |
| 239 | if (AnyOutOfDate) { |
| 240 | // First, build back links in the module dependency graph. |
| 241 | SmallVector<unsigned, 4> Stack; |
| 242 | for (LoadedModulesMap::iterator LM = LoadedModules.begin(), |
| 243 | LMEnd = LoadedModules.end(); |
| 244 | LM != LMEnd; ++LM) { |
| 245 | unsigned ID = LM->first; |
| 246 | |
| 247 | // If this module is out-of-date, push it onto the stack. |
| 248 | if (LM->second.File == 0) |
| 249 | Stack.push_back(ID); |
| 250 | |
| 251 | for (unsigned I = 0, N = LM->second.Dependencies.size(); I != N; ++I) { |
| 252 | unsigned DepID = LM->second.Dependencies[I]; |
| 253 | LoadedModulesMap::iterator Known = LoadedModules.find(DepID); |
| 254 | if (Known == LoadedModules.end() || !Known->second.File) { |
| 255 | // The dependency was out-of-date, so mark us as out of date. |
| 256 | // This is just an optimization. |
| 257 | if (LM->second.File) |
| 258 | Stack.push_back(ID); |
| 259 | |
| 260 | LM->second.File = 0; |
| 261 | continue; |
| 262 | } |
| 263 | |
| 264 | // Record this reverse dependency. |
| 265 | Known->second.ImportedBy.push_back(ID); |
| 266 | } |
| 267 | } |
| 268 | |
| 269 | // Second, walk the back links from out-of-date modules to those modules |
| 270 | // that depend on them, making those modules out-of-date as well. |
| 271 | while (!Stack.empty()) { |
| 272 | unsigned ID = Stack.back(); |
| 273 | Stack.pop_back(); |
| 274 | |
| 275 | LoadedModuleInfo &Info = LoadedModules[ID]; |
| 276 | for (unsigned I = 0, N = Info.ImportedBy.size(); I != N; ++I) { |
| 277 | unsigned FromID = Info.ImportedBy[I]; |
| 278 | if (LoadedModules[FromID].File) { |
| 279 | LoadedModules[FromID].File = 0; |
| 280 | Stack.push_back(FromID); |
| 281 | } |
| 282 | } |
| 283 | } |
| 284 | } |
| 285 | |
| 286 | // Allocate the vector containing information about all of the modules. |
| 287 | Modules.resize(LargestID + 1); |
| 288 | for (LoadedModulesMap::iterator LM = LoadedModules.begin(), |
| 289 | LMEnd = LoadedModules.end(); |
| 290 | LM != LMEnd; ++LM) { |
| 291 | if (!LM->second.File) |
| 292 | continue; |
| 293 | |
| 294 | Modules[LM->first].File = LM->second.File; |
| 295 | |
| 296 | // Resolve dependencies. Drop any we can't resolve due to out-of-date |
| 297 | // module files. |
| 298 | for (unsigned I = 0, N = LM->second.Dependencies.size(); I != N; ++I) { |
| 299 | unsigned DepID = LM->second.Dependencies[I]; |
| 300 | LoadedModulesMap::iterator Known = LoadedModules.find(DepID); |
| 301 | if (Known == LoadedModules.end() || !Known->second.File) |
| 302 | continue; |
| 303 | |
| 304 | Modules[LM->first].Dependencies.push_back(Known->second.File); |
| 305 | } |
| 306 | } |
| 307 | } |
| 308 | |
| 309 | GlobalModuleIndex::~GlobalModuleIndex() { } |
| 310 | |
| 311 | std::pair<GlobalModuleIndex *, GlobalModuleIndex::ErrorCode> |
| 312 | GlobalModuleIndex::readIndex(FileManager &FileMgr, StringRef Path) { |
| 313 | // Load the index file, if it's there. |
| 314 | llvm::SmallString<128> IndexPath; |
| 315 | IndexPath += Path; |
| 316 | llvm::sys::path::append(IndexPath, IndexFileName); |
| 317 | |
| 318 | llvm::OwningPtr<llvm::MemoryBuffer> Buffer( |
| 319 | FileMgr.getBufferForFile(IndexPath)); |
| 320 | if (!Buffer) |
| 321 | return std::make_pair((GlobalModuleIndex *)0, EC_NotFound); |
| 322 | |
| 323 | /// \brief The bitstream reader from which we'll read the AST file. |
| 324 | llvm::BitstreamReader Reader((const unsigned char *)Buffer->getBufferStart(), |
| 325 | (const unsigned char *)Buffer->getBufferEnd()); |
| 326 | |
| 327 | /// \brief The main bitstream cursor for the main block. |
| 328 | llvm::BitstreamCursor Cursor(Reader); |
| 329 | |
| 330 | // Sniff for the signature. |
| 331 | if (Cursor.Read(8) != 'B' || |
| 332 | Cursor.Read(8) != 'C' || |
| 333 | Cursor.Read(8) != 'G' || |
| 334 | Cursor.Read(8) != 'I') { |
| 335 | return std::make_pair((GlobalModuleIndex *)0, EC_IOError); |
| 336 | } |
| 337 | |
| 338 | return std::make_pair(new GlobalModuleIndex(FileMgr, Buffer.take(), Cursor), |
| 339 | EC_None); |
| 340 | } |
| 341 | |
| 342 | void GlobalModuleIndex::getKnownModules( |
| 343 | SmallVectorImpl<const FileEntry *> &ModuleFiles) { |
| 344 | ModuleFiles.clear(); |
| 345 | for (unsigned I = 0, N = Modules.size(); I != N; ++I) { |
| 346 | if (Modules[I].File) |
| 347 | ModuleFiles.push_back(Modules[I].File); |
| 348 | } |
| 349 | } |
| 350 | |
| 351 | void GlobalModuleIndex::getModuleDependencies( |
| 352 | const clang::FileEntry *ModuleFile, |
| 353 | SmallVectorImpl<const clang::FileEntry *> &Dependencies) { |
| 354 | // If the file -> index mapping is empty, populate it now. |
| 355 | if (ModulesByFile.empty()) { |
| 356 | for (unsigned I = 0, N = Modules.size(); I != N; ++I) { |
| 357 | if (Modules[I].File) |
| 358 | ModulesByFile[Modules[I].File] = I; |
| 359 | } |
| 360 | } |
| 361 | |
| 362 | // Look for information about this module file. |
| 363 | llvm::DenseMap<const FileEntry *, unsigned>::iterator Known |
| 364 | = ModulesByFile.find(ModuleFile); |
| 365 | if (Known == ModulesByFile.end()) |
| 366 | return; |
| 367 | |
| 368 | // Record dependencies. |
| 369 | Dependencies = Modules[Known->second].Dependencies; |
| 370 | } |
| 371 | |
Douglas Gregor | 188bdcd | 2013-01-25 23:32:03 +0000 | [diff] [blame^] | 372 | bool GlobalModuleIndex::lookupIdentifier(StringRef Name, HitSet &Hits) { |
| 373 | Hits.clear(); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 374 | |
| 375 | // If there's no identifier index, there is nothing we can do. |
| 376 | if (!IdentifierIndex) |
| 377 | return false; |
| 378 | |
| 379 | // Look into the identifier index. |
| 380 | ++NumIdentifierLookups; |
| 381 | IdentifierIndexTable &Table |
| 382 | = *static_cast<IdentifierIndexTable *>(IdentifierIndex); |
| 383 | IdentifierIndexTable::iterator Known = Table.find(Name); |
| 384 | if (Known == Table.end()) { |
| 385 | return true; |
| 386 | } |
| 387 | |
| 388 | SmallVector<unsigned, 2> ModuleIDs = *Known; |
| 389 | for (unsigned I = 0, N = ModuleIDs.size(); I != N; ++I) { |
| 390 | unsigned ID = ModuleIDs[I]; |
| 391 | if (ID >= Modules.size() || !Modules[ID].File) |
| 392 | continue; |
| 393 | |
Douglas Gregor | 188bdcd | 2013-01-25 23:32:03 +0000 | [diff] [blame^] | 394 | Hits.insert(Modules[ID].File); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 395 | } |
| 396 | |
| 397 | ++NumIdentifierLookupHits; |
| 398 | return true; |
| 399 | } |
| 400 | |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 401 | void GlobalModuleIndex::printStats() { |
| 402 | std::fprintf(stderr, "*** Global Module Index Statistics:\n"); |
| 403 | if (NumIdentifierLookups) { |
| 404 | fprintf(stderr, " %u / %u identifier lookups succeeded (%f%%)\n", |
| 405 | NumIdentifierLookupHits, NumIdentifierLookups, |
| 406 | (double)NumIdentifierLookupHits*100.0/NumIdentifierLookups); |
| 407 | } |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 408 | std::fprintf(stderr, "\n"); |
| 409 | } |
| 410 | |
| 411 | //----------------------------------------------------------------------------// |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 412 | // Global module index writer. |
| 413 | //----------------------------------------------------------------------------// |
| 414 | |
| 415 | namespace { |
| 416 | /// \brief Provides information about a specific module file. |
| 417 | struct ModuleFileInfo { |
| 418 | /// \brief The numberic ID for this module file. |
| 419 | unsigned ID; |
| 420 | |
| 421 | /// \brief The set of modules on which this module depends. Each entry is |
| 422 | /// a module ID. |
| 423 | SmallVector<unsigned, 4> Dependencies; |
| 424 | }; |
| 425 | |
| 426 | /// \brief Builder that generates the global module index file. |
| 427 | class GlobalModuleIndexBuilder { |
| 428 | FileManager &FileMgr; |
| 429 | |
| 430 | /// \brief Mapping from files to module file information. |
| 431 | typedef llvm::MapVector<const FileEntry *, ModuleFileInfo> ModuleFilesMap; |
| 432 | |
| 433 | /// \brief Information about each of the known module files. |
| 434 | ModuleFilesMap ModuleFiles; |
| 435 | |
| 436 | /// \brief Mapping from identifiers to the list of module file IDs that |
| 437 | /// consider this identifier to be interesting. |
| 438 | typedef llvm::StringMap<SmallVector<unsigned, 2> > InterestingIdentifierMap; |
| 439 | |
| 440 | /// \brief A mapping from all interesting identifiers to the set of module |
| 441 | /// files in which those identifiers are considered interesting. |
| 442 | InterestingIdentifierMap InterestingIdentifiers; |
| 443 | |
| 444 | /// \brief Write the block-info block for the global module index file. |
| 445 | void emitBlockInfoBlock(llvm::BitstreamWriter &Stream); |
| 446 | |
| 447 | /// \brief Retrieve the module file information for the given file. |
| 448 | ModuleFileInfo &getModuleFileInfo(const FileEntry *File) { |
| 449 | llvm::MapVector<const FileEntry *, ModuleFileInfo>::iterator Known |
| 450 | = ModuleFiles.find(File); |
| 451 | if (Known != ModuleFiles.end()) |
| 452 | return Known->second; |
| 453 | |
| 454 | unsigned NewID = ModuleFiles.size(); |
| 455 | ModuleFileInfo &Info = ModuleFiles[File]; |
| 456 | Info.ID = NewID; |
| 457 | return Info; |
| 458 | } |
| 459 | |
| 460 | public: |
| 461 | explicit GlobalModuleIndexBuilder(FileManager &FileMgr) : FileMgr(FileMgr){} |
| 462 | |
| 463 | /// \brief Load the contents of the given module file into the builder. |
| 464 | /// |
| 465 | /// \returns true if an error occurred, false otherwise. |
| 466 | bool loadModuleFile(const FileEntry *File); |
| 467 | |
| 468 | /// \brief Write the index to the given bitstream. |
| 469 | void writeIndex(llvm::BitstreamWriter &Stream); |
| 470 | }; |
| 471 | } |
| 472 | |
| 473 | static void emitBlockID(unsigned ID, const char *Name, |
| 474 | llvm::BitstreamWriter &Stream, |
| 475 | SmallVectorImpl<uint64_t> &Record) { |
| 476 | Record.clear(); |
| 477 | Record.push_back(ID); |
| 478 | Stream.EmitRecord(llvm::bitc::BLOCKINFO_CODE_SETBID, Record); |
| 479 | |
| 480 | // Emit the block name if present. |
| 481 | if (Name == 0 || Name[0] == 0) return; |
| 482 | Record.clear(); |
| 483 | while (*Name) |
| 484 | Record.push_back(*Name++); |
| 485 | Stream.EmitRecord(llvm::bitc::BLOCKINFO_CODE_BLOCKNAME, Record); |
| 486 | } |
| 487 | |
| 488 | static void emitRecordID(unsigned ID, const char *Name, |
| 489 | llvm::BitstreamWriter &Stream, |
| 490 | SmallVectorImpl<uint64_t> &Record) { |
| 491 | Record.clear(); |
| 492 | Record.push_back(ID); |
| 493 | while (*Name) |
| 494 | Record.push_back(*Name++); |
| 495 | Stream.EmitRecord(llvm::bitc::BLOCKINFO_CODE_SETRECORDNAME, Record); |
| 496 | } |
| 497 | |
| 498 | void |
| 499 | GlobalModuleIndexBuilder::emitBlockInfoBlock(llvm::BitstreamWriter &Stream) { |
| 500 | SmallVector<uint64_t, 64> Record; |
| 501 | Stream.EnterSubblock(llvm::bitc::BLOCKINFO_BLOCK_ID, 3); |
| 502 | |
| 503 | #define BLOCK(X) emitBlockID(X ## _ID, #X, Stream, Record) |
| 504 | #define RECORD(X) emitRecordID(X, #X, Stream, Record) |
| 505 | BLOCK(GLOBAL_INDEX_BLOCK); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 506 | RECORD(INDEX_METADATA); |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 507 | RECORD(MODULE); |
| 508 | RECORD(IDENTIFIER_INDEX); |
| 509 | #undef RECORD |
| 510 | #undef BLOCK |
| 511 | |
| 512 | Stream.ExitBlock(); |
| 513 | } |
| 514 | |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 515 | namespace { |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 516 | class InterestingASTIdentifierLookupTrait |
| 517 | : public serialization::reader::ASTIdentifierLookupTraitBase { |
| 518 | |
| 519 | public: |
| 520 | /// \brief The identifier and whether it is "interesting". |
| 521 | typedef std::pair<StringRef, bool> data_type; |
| 522 | |
| 523 | data_type ReadData(const internal_key_type& k, |
| 524 | const unsigned char* d, |
| 525 | unsigned DataLen) { |
| 526 | // The first bit indicates whether this identifier is interesting. |
| 527 | // That's all we care about. |
| 528 | using namespace clang::io; |
| 529 | unsigned RawID = ReadUnalignedLE32(d); |
| 530 | bool IsInteresting = RawID & 0x01; |
| 531 | return std::make_pair(k, IsInteresting); |
| 532 | } |
| 533 | }; |
| 534 | } |
| 535 | |
| 536 | bool GlobalModuleIndexBuilder::loadModuleFile(const FileEntry *File) { |
| 537 | // Open the module file. |
| 538 | OwningPtr<llvm::MemoryBuffer> Buffer; |
| 539 | Buffer.reset(FileMgr.getBufferForFile(File)); |
| 540 | if (!Buffer) { |
| 541 | return true; |
| 542 | } |
| 543 | |
| 544 | // Initialize the input stream |
| 545 | llvm::BitstreamReader InStreamFile; |
| 546 | llvm::BitstreamCursor InStream; |
| 547 | InStreamFile.init((const unsigned char *)Buffer->getBufferStart(), |
| 548 | (const unsigned char *)Buffer->getBufferEnd()); |
| 549 | InStream.init(InStreamFile); |
| 550 | |
| 551 | // Sniff for the signature. |
| 552 | if (InStream.Read(8) != 'C' || |
| 553 | InStream.Read(8) != 'P' || |
| 554 | InStream.Read(8) != 'C' || |
| 555 | InStream.Read(8) != 'H') { |
| 556 | return true; |
| 557 | } |
| 558 | |
| 559 | // Record this module file and assign it a unique ID (if it doesn't have |
| 560 | // one already). |
| 561 | unsigned ID = getModuleFileInfo(File).ID; |
| 562 | |
| 563 | // Search for the blocks and records we care about. |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 564 | enum { Other, ControlBlock, ASTBlock } State = Other; |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 565 | bool Done = false; |
| 566 | while (!Done) { |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 567 | llvm::BitstreamEntry Entry = InStream.advance(); |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 568 | switch (Entry.Kind) { |
| 569 | case llvm::BitstreamEntry::Error: |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 570 | Done = true; |
| 571 | continue; |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 572 | |
| 573 | case llvm::BitstreamEntry::Record: |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 574 | // In the 'other' state, just skip the record. We don't care. |
| 575 | if (State == Other) { |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 576 | InStream.skipRecord(Entry.ID); |
| 577 | continue; |
| 578 | } |
| 579 | |
| 580 | // Handle potentially-interesting records below. |
| 581 | break; |
| 582 | |
| 583 | case llvm::BitstreamEntry::SubBlock: |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 584 | if (Entry.ID == CONTROL_BLOCK_ID) { |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 585 | if (InStream.EnterSubBlock(CONTROL_BLOCK_ID)) |
| 586 | return true; |
| 587 | |
| 588 | // Found the control block. |
| 589 | State = ControlBlock; |
| 590 | continue; |
| 591 | } |
| 592 | |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 593 | if (Entry.ID == AST_BLOCK_ID) { |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 594 | if (InStream.EnterSubBlock(AST_BLOCK_ID)) |
| 595 | return true; |
| 596 | |
| 597 | // Found the AST block. |
| 598 | State = ASTBlock; |
| 599 | continue; |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 600 | } |
| 601 | |
| 602 | if (InStream.SkipBlock()) |
| 603 | return true; |
| 604 | |
| 605 | continue; |
| 606 | |
| 607 | case llvm::BitstreamEntry::EndBlock: |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 608 | State = Other; |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 609 | continue; |
| 610 | } |
| 611 | |
| 612 | // Read the given record. |
| 613 | SmallVector<uint64_t, 64> Record; |
| 614 | StringRef Blob; |
| 615 | unsigned Code = InStream.readRecord(Entry.ID, Record, &Blob); |
| 616 | |
| 617 | // Handle module dependencies. |
| 618 | if (State == ControlBlock && Code == IMPORTS) { |
| 619 | // Load each of the imported PCH files. |
| 620 | unsigned Idx = 0, N = Record.size(); |
| 621 | while (Idx < N) { |
| 622 | // Read information about the AST file. |
| 623 | |
| 624 | // Skip the imported kind |
| 625 | ++Idx; |
| 626 | |
| 627 | // Skip the import location |
| 628 | ++Idx; |
| 629 | |
| 630 | // Retrieve the imported file name. |
| 631 | unsigned Length = Record[Idx++]; |
| 632 | SmallString<128> ImportedFile(Record.begin() + Idx, |
| 633 | Record.begin() + Idx + Length); |
| 634 | Idx += Length; |
| 635 | |
| 636 | // Find the imported module file. |
| 637 | const FileEntry *DependsOnFile = FileMgr.getFile(ImportedFile); |
| 638 | if (!DependsOnFile) |
| 639 | return true; |
| 640 | |
| 641 | // Record the dependency. |
| 642 | unsigned DependsOnID = getModuleFileInfo(DependsOnFile).ID; |
| 643 | getModuleFileInfo(File).Dependencies.push_back(DependsOnID); |
| 644 | } |
| 645 | |
| 646 | continue; |
| 647 | } |
| 648 | |
| 649 | // Handle the identifier table |
| 650 | if (State == ASTBlock && Code == IDENTIFIER_TABLE && Record[0] > 0) { |
| 651 | typedef OnDiskChainedHashTable<InterestingASTIdentifierLookupTrait> |
| 652 | InterestingIdentifierTable; |
| 653 | llvm::OwningPtr<InterestingIdentifierTable> |
| 654 | Table(InterestingIdentifierTable::Create( |
| 655 | (const unsigned char *)Blob.data() + Record[0], |
| 656 | (const unsigned char *)Blob.data())); |
| 657 | for (InterestingIdentifierTable::data_iterator D = Table->data_begin(), |
| 658 | DEnd = Table->data_end(); |
| 659 | D != DEnd; ++D) { |
| 660 | std::pair<StringRef, bool> Ident = *D; |
| 661 | if (Ident.second) |
| 662 | InterestingIdentifiers[Ident.first].push_back(ID); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 663 | else |
| 664 | (void)InterestingIdentifiers[Ident.first]; |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 665 | } |
| 666 | } |
| 667 | |
| 668 | // FIXME: Handle the selector table. |
| 669 | |
| 670 | // We don't care about this record. |
| 671 | } |
| 672 | |
| 673 | return false; |
| 674 | } |
| 675 | |
| 676 | namespace { |
| 677 | |
| 678 | /// \brief Trait used to generate the identifier index as an on-disk hash |
| 679 | /// table. |
| 680 | class IdentifierIndexWriterTrait { |
| 681 | public: |
| 682 | typedef StringRef key_type; |
| 683 | typedef StringRef key_type_ref; |
| 684 | typedef SmallVector<unsigned, 2> data_type; |
| 685 | typedef const SmallVector<unsigned, 2> &data_type_ref; |
| 686 | |
| 687 | static unsigned ComputeHash(key_type_ref Key) { |
| 688 | return llvm::HashString(Key); |
| 689 | } |
| 690 | |
| 691 | std::pair<unsigned,unsigned> |
| 692 | EmitKeyDataLength(raw_ostream& Out, key_type_ref Key, data_type_ref Data) { |
| 693 | unsigned KeyLen = Key.size(); |
| 694 | unsigned DataLen = Data.size() * 4; |
| 695 | clang::io::Emit16(Out, KeyLen); |
| 696 | clang::io::Emit16(Out, DataLen); |
| 697 | return std::make_pair(KeyLen, DataLen); |
| 698 | } |
| 699 | |
| 700 | void EmitKey(raw_ostream& Out, key_type_ref Key, unsigned KeyLen) { |
| 701 | Out.write(Key.data(), KeyLen); |
| 702 | } |
| 703 | |
| 704 | void EmitData(raw_ostream& Out, key_type_ref Key, data_type_ref Data, |
| 705 | unsigned DataLen) { |
| 706 | for (unsigned I = 0, N = Data.size(); I != N; ++I) |
| 707 | clang::io::Emit32(Out, Data[I]); |
| 708 | } |
| 709 | }; |
| 710 | |
| 711 | } |
| 712 | |
| 713 | void GlobalModuleIndexBuilder::writeIndex(llvm::BitstreamWriter &Stream) { |
| 714 | using namespace llvm; |
| 715 | |
| 716 | // Emit the file header. |
| 717 | Stream.Emit((unsigned)'B', 8); |
| 718 | Stream.Emit((unsigned)'C', 8); |
| 719 | Stream.Emit((unsigned)'G', 8); |
| 720 | Stream.Emit((unsigned)'I', 8); |
| 721 | |
| 722 | // Write the block-info block, which describes the records in this bitcode |
| 723 | // file. |
| 724 | emitBlockInfoBlock(Stream); |
| 725 | |
| 726 | Stream.EnterSubblock(GLOBAL_INDEX_BLOCK_ID, 3); |
| 727 | |
| 728 | // Write the metadata. |
| 729 | SmallVector<uint64_t, 2> Record; |
| 730 | Record.push_back(CurrentVersion); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 731 | Stream.EmitRecord(INDEX_METADATA, Record); |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 732 | |
| 733 | // Write the set of known module files. |
| 734 | for (ModuleFilesMap::iterator M = ModuleFiles.begin(), |
| 735 | MEnd = ModuleFiles.end(); |
| 736 | M != MEnd; ++M) { |
| 737 | Record.clear(); |
| 738 | Record.push_back(M->second.ID); |
| 739 | Record.push_back(M->first->getSize()); |
| 740 | Record.push_back(M->first->getModificationTime()); |
| 741 | |
| 742 | // File name |
| 743 | StringRef Name(M->first->getName()); |
| 744 | Record.push_back(Name.size()); |
| 745 | Record.append(Name.begin(), Name.end()); |
| 746 | |
| 747 | // Dependencies |
| 748 | Record.push_back(M->second.Dependencies.size()); |
| 749 | Record.append(M->second.Dependencies.begin(), M->second.Dependencies.end()); |
| 750 | Stream.EmitRecord(MODULE, Record); |
| 751 | } |
| 752 | |
| 753 | // Write the identifier -> module file mapping. |
| 754 | { |
| 755 | OnDiskChainedHashTableGenerator<IdentifierIndexWriterTrait> Generator; |
| 756 | IdentifierIndexWriterTrait Trait; |
| 757 | |
| 758 | // Populate the hash table. |
| 759 | for (InterestingIdentifierMap::iterator I = InterestingIdentifiers.begin(), |
| 760 | IEnd = InterestingIdentifiers.end(); |
| 761 | I != IEnd; ++I) { |
| 762 | Generator.insert(I->first(), I->second, Trait); |
| 763 | } |
| 764 | |
| 765 | // Create the on-disk hash table in a buffer. |
| 766 | SmallString<4096> IdentifierTable; |
| 767 | uint32_t BucketOffset; |
| 768 | { |
| 769 | llvm::raw_svector_ostream Out(IdentifierTable); |
| 770 | // Make sure that no bucket is at offset 0 |
| 771 | clang::io::Emit32(Out, 0); |
| 772 | BucketOffset = Generator.Emit(Out, Trait); |
| 773 | } |
| 774 | |
| 775 | // Create a blob abbreviation |
| 776 | BitCodeAbbrev *Abbrev = new BitCodeAbbrev(); |
| 777 | Abbrev->Add(BitCodeAbbrevOp(IDENTIFIER_INDEX)); |
| 778 | Abbrev->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 32)); |
| 779 | Abbrev->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Blob)); |
| 780 | unsigned IDTableAbbrev = Stream.EmitAbbrev(Abbrev); |
| 781 | |
| 782 | // Write the identifier table |
| 783 | Record.clear(); |
| 784 | Record.push_back(IDENTIFIER_INDEX); |
| 785 | Record.push_back(BucketOffset); |
| 786 | Stream.EmitRecordWithBlob(IDTableAbbrev, Record, IdentifierTable.str()); |
| 787 | } |
| 788 | |
| 789 | // FIXME: Selectors. |
| 790 | |
| 791 | Stream.ExitBlock(); |
| 792 | } |
| 793 | |
| 794 | GlobalModuleIndex::ErrorCode |
| 795 | GlobalModuleIndex::writeIndex(FileManager &FileMgr, StringRef Path) { |
| 796 | llvm::SmallString<128> IndexPath; |
| 797 | IndexPath += Path; |
| 798 | llvm::sys::path::append(IndexPath, IndexFileName); |
| 799 | |
| 800 | // Coordinate building the global index file with other processes that might |
| 801 | // try to do the same. |
| 802 | llvm::LockFileManager Locked(IndexPath); |
| 803 | switch (Locked) { |
| 804 | case llvm::LockFileManager::LFS_Error: |
| 805 | return EC_IOError; |
| 806 | |
| 807 | case llvm::LockFileManager::LFS_Owned: |
| 808 | // We're responsible for building the index ourselves. Do so below. |
| 809 | break; |
| 810 | |
| 811 | case llvm::LockFileManager::LFS_Shared: |
| 812 | // Someone else is responsible for building the index. We don't care |
| 813 | // when they finish, so we're done. |
| 814 | return EC_Building; |
| 815 | } |
| 816 | |
| 817 | // The module index builder. |
| 818 | GlobalModuleIndexBuilder Builder(FileMgr); |
| 819 | |
| 820 | // Load each of the module files. |
| 821 | llvm::error_code EC; |
| 822 | for (llvm::sys::fs::directory_iterator D(Path, EC), DEnd; |
| 823 | D != DEnd && !EC; |
| 824 | D.increment(EC)) { |
| 825 | // If this isn't a module file, we don't care. |
| 826 | if (llvm::sys::path::extension(D->path()) != ".pcm") { |
| 827 | // ... unless it's a .pcm.lock file, which indicates that someone is |
| 828 | // in the process of rebuilding a module. They'll rebuild the index |
| 829 | // at the end of that translation unit, so we don't have to. |
| 830 | if (llvm::sys::path::extension(D->path()) == ".pcm.lock") |
| 831 | return EC_Building; |
| 832 | |
| 833 | continue; |
| 834 | } |
| 835 | |
| 836 | // If we can't find the module file, skip it. |
| 837 | const FileEntry *ModuleFile = FileMgr.getFile(D->path()); |
| 838 | if (!ModuleFile) |
| 839 | continue; |
| 840 | |
| 841 | // Load this module file. |
| 842 | if (Builder.loadModuleFile(ModuleFile)) |
| 843 | return EC_IOError; |
| 844 | } |
| 845 | |
| 846 | // The output buffer, into which the global index will be written. |
| 847 | SmallVector<char, 16> OutputBuffer; |
| 848 | { |
| 849 | llvm::BitstreamWriter OutputStream(OutputBuffer); |
| 850 | Builder.writeIndex(OutputStream); |
| 851 | } |
| 852 | |
| 853 | // Write the global index file to a temporary file. |
| 854 | llvm::SmallString<128> IndexTmpPath; |
| 855 | int TmpFD; |
| 856 | if (llvm::sys::fs::unique_file(IndexPath + "-%%%%%%%%", TmpFD, IndexTmpPath)) |
| 857 | return EC_IOError; |
| 858 | |
| 859 | // Open the temporary global index file for output. |
NAKAMURA Takumi | d2db16f | 2013-01-24 08:20:11 +0000 | [diff] [blame] | 860 | llvm::raw_fd_ostream Out(TmpFD, true); |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 861 | if (Out.has_error()) |
| 862 | return EC_IOError; |
| 863 | |
| 864 | // Write the index. |
| 865 | Out.write(OutputBuffer.data(), OutputBuffer.size()); |
| 866 | Out.close(); |
| 867 | if (Out.has_error()) |
| 868 | return EC_IOError; |
| 869 | |
| 870 | // Remove the old index file. It isn't relevant any more. |
| 871 | bool OldIndexExisted; |
| 872 | llvm::sys::fs::remove(IndexPath.str(), OldIndexExisted); |
| 873 | |
| 874 | // Rename the newly-written index file to the proper name. |
| 875 | if (llvm::sys::fs::rename(IndexTmpPath.str(), IndexPath.str())) { |
| 876 | // Rename failed; just remove the |
| 877 | llvm::sys::fs::remove(IndexTmpPath.str(), OldIndexExisted); |
| 878 | return EC_IOError; |
| 879 | } |
| 880 | |
| 881 | // We're done. |
| 882 | return EC_None; |
| 883 | } |