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 | |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 60 | ModuleFileNameResolver::~ModuleFileNameResolver() { } |
| 61 | |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 62 | //----------------------------------------------------------------------------// |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 63 | // Global module index reader. |
| 64 | //----------------------------------------------------------------------------// |
| 65 | |
| 66 | namespace { |
| 67 | |
| 68 | /// \brief Trait used to read the identifier index from the on-disk hash |
| 69 | /// table. |
| 70 | class IdentifierIndexReaderTrait { |
| 71 | public: |
| 72 | typedef StringRef external_key_type; |
| 73 | typedef StringRef internal_key_type; |
| 74 | typedef SmallVector<unsigned, 2> data_type; |
| 75 | |
| 76 | static bool EqualKey(const internal_key_type& a, const internal_key_type& b) { |
| 77 | return a == b; |
| 78 | } |
| 79 | |
| 80 | static unsigned ComputeHash(const internal_key_type& a) { |
| 81 | return llvm::HashString(a); |
| 82 | } |
| 83 | |
| 84 | static std::pair<unsigned, unsigned> |
| 85 | ReadKeyDataLength(const unsigned char*& d) { |
| 86 | using namespace clang::io; |
| 87 | unsigned KeyLen = ReadUnalignedLE16(d); |
| 88 | unsigned DataLen = ReadUnalignedLE16(d); |
| 89 | return std::make_pair(KeyLen, DataLen); |
| 90 | } |
| 91 | |
| 92 | static const internal_key_type& |
| 93 | GetInternalKey(const external_key_type& x) { return x; } |
| 94 | |
| 95 | static const external_key_type& |
| 96 | GetExternalKey(const internal_key_type& x) { return x; } |
| 97 | |
| 98 | static internal_key_type ReadKey(const unsigned char* d, unsigned n) { |
| 99 | return StringRef((const char *)d, n); |
| 100 | } |
| 101 | |
| 102 | static data_type ReadData(const internal_key_type& k, |
| 103 | const unsigned char* d, |
| 104 | unsigned DataLen) { |
| 105 | using namespace clang::io; |
| 106 | |
| 107 | data_type Result; |
| 108 | while (DataLen > 0) { |
| 109 | unsigned ID = ReadUnalignedLE32(d); |
| 110 | Result.push_back(ID); |
| 111 | DataLen -= 4; |
| 112 | } |
| 113 | |
| 114 | return Result; |
| 115 | } |
| 116 | }; |
| 117 | |
| 118 | typedef OnDiskChainedHashTable<IdentifierIndexReaderTrait> IdentifierIndexTable; |
| 119 | |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 120 | } |
| 121 | |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 122 | GlobalModuleIndex::GlobalModuleIndex(llvm::MemoryBuffer *Buffer, |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 123 | llvm::BitstreamCursor Cursor) |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 124 | : Buffer(Buffer), Resolver(), IdentifierIndex(), |
Douglas Gregor | 188bdcd | 2013-01-25 23:32:03 +0000 | [diff] [blame] | 125 | NumIdentifierLookups(), NumIdentifierLookupHits() |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 126 | { |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 127 | // Read the global index. |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 128 | bool InGlobalIndexBlock = false; |
| 129 | bool Done = false; |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 130 | while (!Done) { |
| 131 | llvm::BitstreamEntry Entry = Cursor.advance(); |
| 132 | |
| 133 | switch (Entry.Kind) { |
| 134 | case llvm::BitstreamEntry::Error: |
| 135 | return; |
| 136 | |
| 137 | case llvm::BitstreamEntry::EndBlock: |
| 138 | if (InGlobalIndexBlock) { |
| 139 | InGlobalIndexBlock = false; |
| 140 | Done = true; |
| 141 | continue; |
| 142 | } |
| 143 | return; |
| 144 | |
| 145 | |
| 146 | case llvm::BitstreamEntry::Record: |
| 147 | // Entries in the global index block are handled below. |
| 148 | if (InGlobalIndexBlock) |
| 149 | break; |
| 150 | |
| 151 | return; |
| 152 | |
| 153 | case llvm::BitstreamEntry::SubBlock: |
| 154 | if (!InGlobalIndexBlock && Entry.ID == GLOBAL_INDEX_BLOCK_ID) { |
| 155 | if (Cursor.EnterSubBlock(GLOBAL_INDEX_BLOCK_ID)) |
| 156 | return; |
| 157 | |
| 158 | InGlobalIndexBlock = true; |
| 159 | } else if (Cursor.SkipBlock()) { |
| 160 | return; |
| 161 | } |
| 162 | continue; |
| 163 | } |
| 164 | |
| 165 | SmallVector<uint64_t, 64> Record; |
| 166 | StringRef Blob; |
| 167 | switch ((IndexRecordTypes)Cursor.readRecord(Entry.ID, Record, &Blob)) { |
| 168 | case INDEX_METADATA: |
| 169 | // Make sure that the version matches. |
| 170 | if (Record.size() < 1 || Record[0] != CurrentVersion) |
| 171 | return; |
| 172 | break; |
| 173 | |
| 174 | case MODULE: { |
| 175 | unsigned Idx = 0; |
| 176 | unsigned ID = Record[Idx++]; |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 177 | |
| 178 | // Make room for this module's information. |
| 179 | if (ID == Modules.size()) |
| 180 | Modules.push_back(ModuleInfo()); |
| 181 | else |
| 182 | Modules.resize(ID + 1); |
| 183 | |
| 184 | // Size/modification time for this module file at the time the |
| 185 | // global index was built. |
| 186 | Modules[ID].Size = Record[Idx++]; |
| 187 | Modules[ID].ModTime = Record[Idx++]; |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 188 | |
| 189 | // File name. |
| 190 | unsigned NameLen = Record[Idx++]; |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 191 | Modules[ID].FileName.assign(Record.begin() + Idx, |
| 192 | Record.begin() + Idx + NameLen); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 193 | Idx += NameLen; |
| 194 | |
| 195 | // Dependencies |
| 196 | unsigned NumDeps = Record[Idx++]; |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 197 | Modules[ID].Dependencies.insert(Modules[ID].Dependencies.end(), |
| 198 | Record.begin() + Idx, |
| 199 | Record.begin() + Idx + NumDeps); |
| 200 | Idx += NumDeps; |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 201 | |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 202 | // Make sure we're at the end of the record. |
| 203 | assert(Idx == Record.size() && "More module info?"); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 204 | break; |
| 205 | } |
| 206 | |
| 207 | case IDENTIFIER_INDEX: |
| 208 | // Wire up the identifier index. |
| 209 | if (Record[0]) { |
| 210 | IdentifierIndex = IdentifierIndexTable::Create( |
| 211 | (const unsigned char *)Blob.data() + Record[0], |
| 212 | (const unsigned char *)Blob.data(), |
| 213 | IdentifierIndexReaderTrait()); |
| 214 | } |
| 215 | break; |
| 216 | } |
| 217 | } |
| 218 | |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 219 | // Compute imported-by relation. |
| 220 | for (unsigned ID = 0, IDEnd = Modules.size(); ID != IDEnd; ++ID) { |
| 221 | for (unsigned D = 0, DEnd = Modules[ID].Dependencies.size(); |
| 222 | D != DEnd; ++D) { |
| 223 | Modules[Modules[ID].Dependencies[D]].ImportedBy.push_back(ID); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 224 | } |
| 225 | } |
| 226 | } |
| 227 | |
| 228 | GlobalModuleIndex::~GlobalModuleIndex() { } |
| 229 | |
| 230 | std::pair<GlobalModuleIndex *, GlobalModuleIndex::ErrorCode> |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 231 | GlobalModuleIndex::readIndex(StringRef Path) { |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 232 | // Load the index file, if it's there. |
| 233 | llvm::SmallString<128> IndexPath; |
| 234 | IndexPath += Path; |
| 235 | llvm::sys::path::append(IndexPath, IndexFileName); |
| 236 | |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 237 | llvm::OwningPtr<llvm::MemoryBuffer> Buffer; |
| 238 | if (llvm::MemoryBuffer::getFile(IndexPath, Buffer) != llvm::errc::success) |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 239 | return std::make_pair((GlobalModuleIndex *)0, EC_NotFound); |
| 240 | |
| 241 | /// \brief The bitstream reader from which we'll read the AST file. |
| 242 | llvm::BitstreamReader Reader((const unsigned char *)Buffer->getBufferStart(), |
| 243 | (const unsigned char *)Buffer->getBufferEnd()); |
| 244 | |
| 245 | /// \brief The main bitstream cursor for the main block. |
| 246 | llvm::BitstreamCursor Cursor(Reader); |
| 247 | |
| 248 | // Sniff for the signature. |
| 249 | if (Cursor.Read(8) != 'B' || |
| 250 | Cursor.Read(8) != 'C' || |
| 251 | Cursor.Read(8) != 'G' || |
| 252 | Cursor.Read(8) != 'I') { |
| 253 | return std::make_pair((GlobalModuleIndex *)0, EC_IOError); |
| 254 | } |
| 255 | |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 256 | return std::make_pair(new GlobalModuleIndex(Buffer.take(), Cursor), EC_None); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 257 | } |
| 258 | |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 259 | void |
| 260 | GlobalModuleIndex::getKnownModules(SmallVectorImpl<ModuleFile *> &ModuleFiles) { |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 261 | ModuleFiles.clear(); |
| 262 | for (unsigned I = 0, N = Modules.size(); I != N; ++I) { |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 263 | if (ModuleFile *File = resolveModuleFile(I)) |
| 264 | ModuleFiles.push_back(File); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 265 | } |
| 266 | } |
| 267 | |
| 268 | void GlobalModuleIndex::getModuleDependencies( |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 269 | ModuleFile *File, |
| 270 | SmallVectorImpl<ModuleFile *> &Dependencies) { |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 271 | // If the file -> index mapping is empty, populate it now. |
| 272 | if (ModulesByFile.empty()) { |
| 273 | for (unsigned I = 0, N = Modules.size(); I != N; ++I) { |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 274 | resolveModuleFile(I); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 275 | } |
| 276 | } |
| 277 | |
| 278 | // Look for information about this module file. |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 279 | llvm::DenseMap<ModuleFile *, unsigned>::iterator Known |
| 280 | = ModulesByFile.find(File); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 281 | if (Known == ModulesByFile.end()) |
| 282 | return; |
| 283 | |
| 284 | // Record dependencies. |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 285 | Dependencies.clear(); |
| 286 | ArrayRef<unsigned> StoredDependencies = Modules[Known->second].Dependencies; |
| 287 | for (unsigned I = 0, N = StoredDependencies.size(); I != N; ++I) { |
| 288 | if (ModuleFile *MF = resolveModuleFile(I)) |
| 289 | Dependencies.push_back(MF); |
| 290 | } |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 291 | } |
| 292 | |
Douglas Gregor | 188bdcd | 2013-01-25 23:32:03 +0000 | [diff] [blame] | 293 | bool GlobalModuleIndex::lookupIdentifier(StringRef Name, HitSet &Hits) { |
| 294 | Hits.clear(); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 295 | |
| 296 | // If there's no identifier index, there is nothing we can do. |
| 297 | if (!IdentifierIndex) |
| 298 | return false; |
| 299 | |
| 300 | // Look into the identifier index. |
| 301 | ++NumIdentifierLookups; |
| 302 | IdentifierIndexTable &Table |
| 303 | = *static_cast<IdentifierIndexTable *>(IdentifierIndex); |
| 304 | IdentifierIndexTable::iterator Known = Table.find(Name); |
| 305 | if (Known == Table.end()) { |
| 306 | return true; |
| 307 | } |
| 308 | |
| 309 | SmallVector<unsigned, 2> ModuleIDs = *Known; |
| 310 | for (unsigned I = 0, N = ModuleIDs.size(); I != N; ++I) { |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 311 | if (ModuleFile *File = resolveModuleFile(ModuleIDs[I])) |
| 312 | Hits.insert(File); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 313 | } |
| 314 | |
| 315 | ++NumIdentifierLookupHits; |
| 316 | return true; |
| 317 | } |
| 318 | |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 319 | ModuleFile *GlobalModuleIndex::resolveModuleFile(unsigned ID) { |
| 320 | assert(ID < Modules.size() && "Out-of-bounds module index"); |
| 321 | |
| 322 | // If we already have a module file, return it. |
| 323 | if (Modules[ID].File) |
| 324 | return Modules[ID].File; |
| 325 | |
| 326 | // If we don't have a file name, or if there is no resolver, we can't |
| 327 | // resolve this. |
| 328 | if (Modules[ID].FileName.empty() || !Resolver) |
| 329 | return 0; |
| 330 | |
| 331 | // Try to resolve this module file. |
| 332 | ModuleFile *File; |
| 333 | if (Resolver->resolveModuleFileName(Modules[ID].FileName, Modules[ID].Size, |
| 334 | Modules[ID].ModTime, File)) { |
| 335 | // Clear out the module files for anything that depends on this module. |
| 336 | llvm::SmallVector<unsigned, 8> Stack; |
| 337 | |
| 338 | Stack.push_back(ID); |
| 339 | while (!Stack.empty()) { |
| 340 | unsigned Victim = Stack.back(); |
| 341 | Stack.pop_back(); |
| 342 | |
| 343 | // Mark this file as ignored. |
| 344 | Modules[Victim].File = 0; |
| 345 | Modules[Victim].FileName.clear(); |
| 346 | |
| 347 | // Push any not-yet-ignored imported modules onto the stack. |
| 348 | for (unsigned I = 0, N = Modules[Victim].ImportedBy.size(); I != N; ++I) { |
| 349 | unsigned NextVictim = Modules[Victim].ImportedBy[I]; |
| 350 | if (!Modules[NextVictim].FileName.empty()) |
| 351 | Stack.push_back(NextVictim); |
| 352 | } |
| 353 | } |
| 354 | |
| 355 | return 0; |
| 356 | } |
| 357 | |
| 358 | // If we have a module file, record it. |
| 359 | if (File) { |
| 360 | Modules[ID].File = File; |
| 361 | ModulesByFile[File] = ID; |
| 362 | } |
| 363 | |
| 364 | return File; |
| 365 | } |
| 366 | |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 367 | void GlobalModuleIndex::printStats() { |
| 368 | std::fprintf(stderr, "*** Global Module Index Statistics:\n"); |
| 369 | if (NumIdentifierLookups) { |
| 370 | fprintf(stderr, " %u / %u identifier lookups succeeded (%f%%)\n", |
| 371 | NumIdentifierLookupHits, NumIdentifierLookups, |
| 372 | (double)NumIdentifierLookupHits*100.0/NumIdentifierLookups); |
| 373 | } |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 374 | std::fprintf(stderr, "\n"); |
| 375 | } |
| 376 | |
| 377 | //----------------------------------------------------------------------------// |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 378 | // Global module index writer. |
| 379 | //----------------------------------------------------------------------------// |
| 380 | |
| 381 | namespace { |
| 382 | /// \brief Provides information about a specific module file. |
| 383 | struct ModuleFileInfo { |
| 384 | /// \brief The numberic ID for this module file. |
| 385 | unsigned ID; |
| 386 | |
| 387 | /// \brief The set of modules on which this module depends. Each entry is |
| 388 | /// a module ID. |
| 389 | SmallVector<unsigned, 4> Dependencies; |
| 390 | }; |
| 391 | |
| 392 | /// \brief Builder that generates the global module index file. |
| 393 | class GlobalModuleIndexBuilder { |
| 394 | FileManager &FileMgr; |
| 395 | |
| 396 | /// \brief Mapping from files to module file information. |
| 397 | typedef llvm::MapVector<const FileEntry *, ModuleFileInfo> ModuleFilesMap; |
| 398 | |
| 399 | /// \brief Information about each of the known module files. |
| 400 | ModuleFilesMap ModuleFiles; |
| 401 | |
| 402 | /// \brief Mapping from identifiers to the list of module file IDs that |
| 403 | /// consider this identifier to be interesting. |
| 404 | typedef llvm::StringMap<SmallVector<unsigned, 2> > InterestingIdentifierMap; |
| 405 | |
| 406 | /// \brief A mapping from all interesting identifiers to the set of module |
| 407 | /// files in which those identifiers are considered interesting. |
| 408 | InterestingIdentifierMap InterestingIdentifiers; |
| 409 | |
| 410 | /// \brief Write the block-info block for the global module index file. |
| 411 | void emitBlockInfoBlock(llvm::BitstreamWriter &Stream); |
| 412 | |
| 413 | /// \brief Retrieve the module file information for the given file. |
| 414 | ModuleFileInfo &getModuleFileInfo(const FileEntry *File) { |
| 415 | llvm::MapVector<const FileEntry *, ModuleFileInfo>::iterator Known |
| 416 | = ModuleFiles.find(File); |
| 417 | if (Known != ModuleFiles.end()) |
| 418 | return Known->second; |
| 419 | |
| 420 | unsigned NewID = ModuleFiles.size(); |
| 421 | ModuleFileInfo &Info = ModuleFiles[File]; |
| 422 | Info.ID = NewID; |
| 423 | return Info; |
| 424 | } |
| 425 | |
| 426 | public: |
| 427 | explicit GlobalModuleIndexBuilder(FileManager &FileMgr) : FileMgr(FileMgr){} |
| 428 | |
| 429 | /// \brief Load the contents of the given module file into the builder. |
| 430 | /// |
| 431 | /// \returns true if an error occurred, false otherwise. |
| 432 | bool loadModuleFile(const FileEntry *File); |
| 433 | |
| 434 | /// \brief Write the index to the given bitstream. |
| 435 | void writeIndex(llvm::BitstreamWriter &Stream); |
| 436 | }; |
| 437 | } |
| 438 | |
| 439 | static void emitBlockID(unsigned ID, const char *Name, |
| 440 | llvm::BitstreamWriter &Stream, |
| 441 | SmallVectorImpl<uint64_t> &Record) { |
| 442 | Record.clear(); |
| 443 | Record.push_back(ID); |
| 444 | Stream.EmitRecord(llvm::bitc::BLOCKINFO_CODE_SETBID, Record); |
| 445 | |
| 446 | // Emit the block name if present. |
| 447 | if (Name == 0 || Name[0] == 0) return; |
| 448 | Record.clear(); |
| 449 | while (*Name) |
| 450 | Record.push_back(*Name++); |
| 451 | Stream.EmitRecord(llvm::bitc::BLOCKINFO_CODE_BLOCKNAME, Record); |
| 452 | } |
| 453 | |
| 454 | static void emitRecordID(unsigned ID, const char *Name, |
| 455 | llvm::BitstreamWriter &Stream, |
| 456 | SmallVectorImpl<uint64_t> &Record) { |
| 457 | Record.clear(); |
| 458 | Record.push_back(ID); |
| 459 | while (*Name) |
| 460 | Record.push_back(*Name++); |
| 461 | Stream.EmitRecord(llvm::bitc::BLOCKINFO_CODE_SETRECORDNAME, Record); |
| 462 | } |
| 463 | |
| 464 | void |
| 465 | GlobalModuleIndexBuilder::emitBlockInfoBlock(llvm::BitstreamWriter &Stream) { |
| 466 | SmallVector<uint64_t, 64> Record; |
| 467 | Stream.EnterSubblock(llvm::bitc::BLOCKINFO_BLOCK_ID, 3); |
| 468 | |
| 469 | #define BLOCK(X) emitBlockID(X ## _ID, #X, Stream, Record) |
| 470 | #define RECORD(X) emitRecordID(X, #X, Stream, Record) |
| 471 | BLOCK(GLOBAL_INDEX_BLOCK); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 472 | RECORD(INDEX_METADATA); |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 473 | RECORD(MODULE); |
| 474 | RECORD(IDENTIFIER_INDEX); |
| 475 | #undef RECORD |
| 476 | #undef BLOCK |
| 477 | |
| 478 | Stream.ExitBlock(); |
| 479 | } |
| 480 | |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 481 | namespace { |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 482 | class InterestingASTIdentifierLookupTrait |
| 483 | : public serialization::reader::ASTIdentifierLookupTraitBase { |
| 484 | |
| 485 | public: |
| 486 | /// \brief The identifier and whether it is "interesting". |
| 487 | typedef std::pair<StringRef, bool> data_type; |
| 488 | |
| 489 | data_type ReadData(const internal_key_type& k, |
| 490 | const unsigned char* d, |
| 491 | unsigned DataLen) { |
| 492 | // The first bit indicates whether this identifier is interesting. |
| 493 | // That's all we care about. |
| 494 | using namespace clang::io; |
| 495 | unsigned RawID = ReadUnalignedLE32(d); |
| 496 | bool IsInteresting = RawID & 0x01; |
| 497 | return std::make_pair(k, IsInteresting); |
| 498 | } |
| 499 | }; |
| 500 | } |
| 501 | |
| 502 | bool GlobalModuleIndexBuilder::loadModuleFile(const FileEntry *File) { |
| 503 | // Open the module file. |
| 504 | OwningPtr<llvm::MemoryBuffer> Buffer; |
Douglas Gregor | 8e31d06 | 2013-02-06 18:08:37 +0000 | [diff] [blame] | 505 | std::string ErrorStr; |
| 506 | Buffer.reset(FileMgr.getBufferForFile(File, &ErrorStr, /*isVolatile=*/true)); |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 507 | if (!Buffer) { |
| 508 | return true; |
| 509 | } |
| 510 | |
| 511 | // Initialize the input stream |
| 512 | llvm::BitstreamReader InStreamFile; |
| 513 | llvm::BitstreamCursor InStream; |
| 514 | InStreamFile.init((const unsigned char *)Buffer->getBufferStart(), |
| 515 | (const unsigned char *)Buffer->getBufferEnd()); |
| 516 | InStream.init(InStreamFile); |
| 517 | |
| 518 | // Sniff for the signature. |
| 519 | if (InStream.Read(8) != 'C' || |
| 520 | InStream.Read(8) != 'P' || |
| 521 | InStream.Read(8) != 'C' || |
| 522 | InStream.Read(8) != 'H') { |
| 523 | return true; |
| 524 | } |
| 525 | |
| 526 | // Record this module file and assign it a unique ID (if it doesn't have |
| 527 | // one already). |
| 528 | unsigned ID = getModuleFileInfo(File).ID; |
| 529 | |
| 530 | // Search for the blocks and records we care about. |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 531 | enum { Other, ControlBlock, ASTBlock } State = Other; |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 532 | bool Done = false; |
| 533 | while (!Done) { |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 534 | llvm::BitstreamEntry Entry = InStream.advance(); |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 535 | switch (Entry.Kind) { |
| 536 | case llvm::BitstreamEntry::Error: |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 537 | Done = true; |
| 538 | continue; |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 539 | |
| 540 | case llvm::BitstreamEntry::Record: |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 541 | // In the 'other' state, just skip the record. We don't care. |
| 542 | if (State == Other) { |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 543 | InStream.skipRecord(Entry.ID); |
| 544 | continue; |
| 545 | } |
| 546 | |
| 547 | // Handle potentially-interesting records below. |
| 548 | break; |
| 549 | |
| 550 | case llvm::BitstreamEntry::SubBlock: |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 551 | if (Entry.ID == CONTROL_BLOCK_ID) { |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 552 | if (InStream.EnterSubBlock(CONTROL_BLOCK_ID)) |
| 553 | return true; |
| 554 | |
| 555 | // Found the control block. |
| 556 | State = ControlBlock; |
| 557 | continue; |
| 558 | } |
| 559 | |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 560 | if (Entry.ID == AST_BLOCK_ID) { |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 561 | if (InStream.EnterSubBlock(AST_BLOCK_ID)) |
| 562 | return true; |
| 563 | |
| 564 | // Found the AST block. |
| 565 | State = ASTBlock; |
| 566 | continue; |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 567 | } |
| 568 | |
| 569 | if (InStream.SkipBlock()) |
| 570 | return true; |
| 571 | |
| 572 | continue; |
| 573 | |
| 574 | case llvm::BitstreamEntry::EndBlock: |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 575 | State = Other; |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 576 | continue; |
| 577 | } |
| 578 | |
| 579 | // Read the given record. |
| 580 | SmallVector<uint64_t, 64> Record; |
| 581 | StringRef Blob; |
| 582 | unsigned Code = InStream.readRecord(Entry.ID, Record, &Blob); |
| 583 | |
| 584 | // Handle module dependencies. |
| 585 | if (State == ControlBlock && Code == IMPORTS) { |
| 586 | // Load each of the imported PCH files. |
| 587 | unsigned Idx = 0, N = Record.size(); |
| 588 | while (Idx < N) { |
| 589 | // Read information about the AST file. |
| 590 | |
| 591 | // Skip the imported kind |
| 592 | ++Idx; |
| 593 | |
| 594 | // Skip the import location |
| 595 | ++Idx; |
| 596 | |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 597 | // Load stored size/modification time. |
| 598 | off_t StoredSize = (off_t)Record[Idx++]; |
| 599 | time_t StoredModTime = (time_t)Record[Idx++]; |
| 600 | |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 601 | // Retrieve the imported file name. |
| 602 | unsigned Length = Record[Idx++]; |
| 603 | SmallString<128> ImportedFile(Record.begin() + Idx, |
| 604 | Record.begin() + Idx + Length); |
| 605 | Idx += Length; |
| 606 | |
| 607 | // Find the imported module file. |
Douglas Gregor | ea14a87 | 2013-02-08 21:27:45 +0000 | [diff] [blame] | 608 | const FileEntry *DependsOnFile |
| 609 | = FileMgr.getFile(ImportedFile, /*openFile=*/false, |
| 610 | /*cacheFailure=*/false); |
Douglas Gregor | 677e15f | 2013-03-19 00:28:20 +0000 | [diff] [blame^] | 611 | if (!DependsOnFile || |
| 612 | (StoredSize != DependsOnFile->getSize()) || |
| 613 | (StoredModTime != DependsOnFile->getModificationTime())) |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 614 | return true; |
| 615 | |
| 616 | // Record the dependency. |
| 617 | unsigned DependsOnID = getModuleFileInfo(DependsOnFile).ID; |
| 618 | getModuleFileInfo(File).Dependencies.push_back(DependsOnID); |
| 619 | } |
| 620 | |
| 621 | continue; |
| 622 | } |
| 623 | |
| 624 | // Handle the identifier table |
| 625 | if (State == ASTBlock && Code == IDENTIFIER_TABLE && Record[0] > 0) { |
| 626 | typedef OnDiskChainedHashTable<InterestingASTIdentifierLookupTrait> |
| 627 | InterestingIdentifierTable; |
| 628 | llvm::OwningPtr<InterestingIdentifierTable> |
| 629 | Table(InterestingIdentifierTable::Create( |
| 630 | (const unsigned char *)Blob.data() + Record[0], |
| 631 | (const unsigned char *)Blob.data())); |
| 632 | for (InterestingIdentifierTable::data_iterator D = Table->data_begin(), |
| 633 | DEnd = Table->data_end(); |
| 634 | D != DEnd; ++D) { |
| 635 | std::pair<StringRef, bool> Ident = *D; |
| 636 | if (Ident.second) |
| 637 | InterestingIdentifiers[Ident.first].push_back(ID); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 638 | else |
| 639 | (void)InterestingIdentifiers[Ident.first]; |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 640 | } |
| 641 | } |
| 642 | |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 643 | // We don't care about this record. |
| 644 | } |
| 645 | |
| 646 | return false; |
| 647 | } |
| 648 | |
| 649 | namespace { |
| 650 | |
| 651 | /// \brief Trait used to generate the identifier index as an on-disk hash |
| 652 | /// table. |
| 653 | class IdentifierIndexWriterTrait { |
| 654 | public: |
| 655 | typedef StringRef key_type; |
| 656 | typedef StringRef key_type_ref; |
| 657 | typedef SmallVector<unsigned, 2> data_type; |
| 658 | typedef const SmallVector<unsigned, 2> &data_type_ref; |
| 659 | |
| 660 | static unsigned ComputeHash(key_type_ref Key) { |
| 661 | return llvm::HashString(Key); |
| 662 | } |
| 663 | |
| 664 | std::pair<unsigned,unsigned> |
| 665 | EmitKeyDataLength(raw_ostream& Out, key_type_ref Key, data_type_ref Data) { |
| 666 | unsigned KeyLen = Key.size(); |
| 667 | unsigned DataLen = Data.size() * 4; |
| 668 | clang::io::Emit16(Out, KeyLen); |
| 669 | clang::io::Emit16(Out, DataLen); |
| 670 | return std::make_pair(KeyLen, DataLen); |
| 671 | } |
| 672 | |
| 673 | void EmitKey(raw_ostream& Out, key_type_ref Key, unsigned KeyLen) { |
| 674 | Out.write(Key.data(), KeyLen); |
| 675 | } |
| 676 | |
| 677 | void EmitData(raw_ostream& Out, key_type_ref Key, data_type_ref Data, |
| 678 | unsigned DataLen) { |
| 679 | for (unsigned I = 0, N = Data.size(); I != N; ++I) |
| 680 | clang::io::Emit32(Out, Data[I]); |
| 681 | } |
| 682 | }; |
| 683 | |
| 684 | } |
| 685 | |
| 686 | void GlobalModuleIndexBuilder::writeIndex(llvm::BitstreamWriter &Stream) { |
| 687 | using namespace llvm; |
| 688 | |
| 689 | // Emit the file header. |
| 690 | Stream.Emit((unsigned)'B', 8); |
| 691 | Stream.Emit((unsigned)'C', 8); |
| 692 | Stream.Emit((unsigned)'G', 8); |
| 693 | Stream.Emit((unsigned)'I', 8); |
| 694 | |
| 695 | // Write the block-info block, which describes the records in this bitcode |
| 696 | // file. |
| 697 | emitBlockInfoBlock(Stream); |
| 698 | |
| 699 | Stream.EnterSubblock(GLOBAL_INDEX_BLOCK_ID, 3); |
| 700 | |
| 701 | // Write the metadata. |
| 702 | SmallVector<uint64_t, 2> Record; |
| 703 | Record.push_back(CurrentVersion); |
Douglas Gregor | 1a49d97 | 2013-01-25 01:03:03 +0000 | [diff] [blame] | 704 | Stream.EmitRecord(INDEX_METADATA, Record); |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 705 | |
| 706 | // Write the set of known module files. |
| 707 | for (ModuleFilesMap::iterator M = ModuleFiles.begin(), |
| 708 | MEnd = ModuleFiles.end(); |
| 709 | M != MEnd; ++M) { |
| 710 | Record.clear(); |
| 711 | Record.push_back(M->second.ID); |
| 712 | Record.push_back(M->first->getSize()); |
| 713 | Record.push_back(M->first->getModificationTime()); |
| 714 | |
| 715 | // File name |
| 716 | StringRef Name(M->first->getName()); |
| 717 | Record.push_back(Name.size()); |
| 718 | Record.append(Name.begin(), Name.end()); |
| 719 | |
| 720 | // Dependencies |
| 721 | Record.push_back(M->second.Dependencies.size()); |
| 722 | Record.append(M->second.Dependencies.begin(), M->second.Dependencies.end()); |
| 723 | Stream.EmitRecord(MODULE, Record); |
| 724 | } |
| 725 | |
| 726 | // Write the identifier -> module file mapping. |
| 727 | { |
| 728 | OnDiskChainedHashTableGenerator<IdentifierIndexWriterTrait> Generator; |
| 729 | IdentifierIndexWriterTrait Trait; |
| 730 | |
| 731 | // Populate the hash table. |
| 732 | for (InterestingIdentifierMap::iterator I = InterestingIdentifiers.begin(), |
| 733 | IEnd = InterestingIdentifiers.end(); |
| 734 | I != IEnd; ++I) { |
| 735 | Generator.insert(I->first(), I->second, Trait); |
| 736 | } |
| 737 | |
| 738 | // Create the on-disk hash table in a buffer. |
| 739 | SmallString<4096> IdentifierTable; |
| 740 | uint32_t BucketOffset; |
| 741 | { |
| 742 | llvm::raw_svector_ostream Out(IdentifierTable); |
| 743 | // Make sure that no bucket is at offset 0 |
| 744 | clang::io::Emit32(Out, 0); |
| 745 | BucketOffset = Generator.Emit(Out, Trait); |
| 746 | } |
| 747 | |
| 748 | // Create a blob abbreviation |
| 749 | BitCodeAbbrev *Abbrev = new BitCodeAbbrev(); |
| 750 | Abbrev->Add(BitCodeAbbrevOp(IDENTIFIER_INDEX)); |
| 751 | Abbrev->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Fixed, 32)); |
| 752 | Abbrev->Add(BitCodeAbbrevOp(BitCodeAbbrevOp::Blob)); |
| 753 | unsigned IDTableAbbrev = Stream.EmitAbbrev(Abbrev); |
| 754 | |
| 755 | // Write the identifier table |
| 756 | Record.clear(); |
| 757 | Record.push_back(IDENTIFIER_INDEX); |
| 758 | Record.push_back(BucketOffset); |
| 759 | Stream.EmitRecordWithBlob(IDTableAbbrev, Record, IdentifierTable.str()); |
| 760 | } |
| 761 | |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 762 | Stream.ExitBlock(); |
| 763 | } |
| 764 | |
| 765 | GlobalModuleIndex::ErrorCode |
| 766 | GlobalModuleIndex::writeIndex(FileManager &FileMgr, StringRef Path) { |
| 767 | llvm::SmallString<128> IndexPath; |
| 768 | IndexPath += Path; |
| 769 | llvm::sys::path::append(IndexPath, IndexFileName); |
| 770 | |
| 771 | // Coordinate building the global index file with other processes that might |
| 772 | // try to do the same. |
| 773 | llvm::LockFileManager Locked(IndexPath); |
| 774 | switch (Locked) { |
| 775 | case llvm::LockFileManager::LFS_Error: |
| 776 | return EC_IOError; |
| 777 | |
| 778 | case llvm::LockFileManager::LFS_Owned: |
| 779 | // We're responsible for building the index ourselves. Do so below. |
| 780 | break; |
| 781 | |
| 782 | case llvm::LockFileManager::LFS_Shared: |
| 783 | // Someone else is responsible for building the index. We don't care |
| 784 | // when they finish, so we're done. |
| 785 | return EC_Building; |
| 786 | } |
| 787 | |
| 788 | // The module index builder. |
| 789 | GlobalModuleIndexBuilder Builder(FileMgr); |
| 790 | |
| 791 | // Load each of the module files. |
| 792 | llvm::error_code EC; |
| 793 | for (llvm::sys::fs::directory_iterator D(Path, EC), DEnd; |
| 794 | D != DEnd && !EC; |
| 795 | D.increment(EC)) { |
| 796 | // If this isn't a module file, we don't care. |
| 797 | if (llvm::sys::path::extension(D->path()) != ".pcm") { |
| 798 | // ... unless it's a .pcm.lock file, which indicates that someone is |
| 799 | // in the process of rebuilding a module. They'll rebuild the index |
| 800 | // at the end of that translation unit, so we don't have to. |
| 801 | if (llvm::sys::path::extension(D->path()) == ".pcm.lock") |
| 802 | return EC_Building; |
| 803 | |
| 804 | continue; |
| 805 | } |
| 806 | |
| 807 | // If we can't find the module file, skip it. |
| 808 | const FileEntry *ModuleFile = FileMgr.getFile(D->path()); |
| 809 | if (!ModuleFile) |
| 810 | continue; |
| 811 | |
| 812 | // Load this module file. |
| 813 | if (Builder.loadModuleFile(ModuleFile)) |
| 814 | return EC_IOError; |
| 815 | } |
| 816 | |
| 817 | // The output buffer, into which the global index will be written. |
| 818 | SmallVector<char, 16> OutputBuffer; |
| 819 | { |
| 820 | llvm::BitstreamWriter OutputStream(OutputBuffer); |
| 821 | Builder.writeIndex(OutputStream); |
| 822 | } |
| 823 | |
| 824 | // Write the global index file to a temporary file. |
| 825 | llvm::SmallString<128> IndexTmpPath; |
| 826 | int TmpFD; |
| 827 | if (llvm::sys::fs::unique_file(IndexPath + "-%%%%%%%%", TmpFD, IndexTmpPath)) |
| 828 | return EC_IOError; |
| 829 | |
| 830 | // Open the temporary global index file for output. |
NAKAMURA Takumi | d2db16f | 2013-01-24 08:20:11 +0000 | [diff] [blame] | 831 | llvm::raw_fd_ostream Out(TmpFD, true); |
Douglas Gregor | a6b00fc | 2013-01-23 22:38:11 +0000 | [diff] [blame] | 832 | if (Out.has_error()) |
| 833 | return EC_IOError; |
| 834 | |
| 835 | // Write the index. |
| 836 | Out.write(OutputBuffer.data(), OutputBuffer.size()); |
| 837 | Out.close(); |
| 838 | if (Out.has_error()) |
| 839 | return EC_IOError; |
| 840 | |
| 841 | // Remove the old index file. It isn't relevant any more. |
| 842 | bool OldIndexExisted; |
| 843 | llvm::sys::fs::remove(IndexPath.str(), OldIndexExisted); |
| 844 | |
| 845 | // Rename the newly-written index file to the proper name. |
| 846 | if (llvm::sys::fs::rename(IndexTmpPath.str(), IndexPath.str())) { |
| 847 | // Rename failed; just remove the |
| 848 | llvm::sys::fs::remove(IndexTmpPath.str(), OldIndexExisted); |
| 849 | return EC_IOError; |
| 850 | } |
| 851 | |
| 852 | // We're done. |
| 853 | return EC_None; |
| 854 | } |