ArchiveReader.cpp revision 8dde18fc505c438041987a540977a240f258332e
1//===-- ArchiveReader.cpp - Read LLVM archive files -------------*- C++ -*-===// 2// 3// The LLVM Compiler Infrastructure 4// 5// This file was developed by Reid Spencer and is distributed under the 6// University of Illinois Open Source License. See LICENSE.TXT for details. 7// 8//===----------------------------------------------------------------------===// 9// 10// Builds up standard unix archive files (.a) containing LLVM bytecode. 11// 12//===----------------------------------------------------------------------===// 13 14#include "ArchiveInternals.h" 15#include "llvm/Bytecode/Reader.h" 16 17using namespace llvm; 18 19/// Read a variable-bit-rate encoded unsigned integer 20inline unsigned readInteger(const char*&At, const char*End) { 21 unsigned Shift = 0; 22 unsigned Result = 0; 23 24 do { 25 if (At == End) 26 throw std::string("Ran out of data reading vbr_uint!"); 27 Result |= (unsigned)((*At++) & 0x7F) << Shift; 28 Shift += 7; 29 } while (At[-1] & 0x80); 30 return Result; 31} 32 33// Completely parse the Archive's symbol table and populate symTab member var. 34void 35Archive::parseSymbolTable(const void* data, unsigned size) { 36 const char* At = (const char*) data; 37 const char* End = At + size; 38 while (At < End) { 39 unsigned offset = readInteger(At, End); 40 unsigned length = readInteger(At, End); 41 if (At + length > End) 42 throw std::string("malformed symbol table"); 43 // we don't care if it can't be inserted (duplicate entry) 44 symTab.insert(std::make_pair(std::string(At, length), offset)); 45 At += length; 46 } 47 symTabSize = size; 48} 49 50// This member parses an ArchiveMemberHeader that is presumed to be pointed to 51// by At. The At pointer is updated to the byte just after the header, which 52// can be variable in size. 53ArchiveMember* 54Archive::parseMemberHeader(const char*& At, const char* End) { 55 assert(At + sizeof(ArchiveMemberHeader) < End && "Not enough data"); 56 57 // Cast archive member header 58 ArchiveMemberHeader* Hdr = (ArchiveMemberHeader*)At; 59 At += sizeof(ArchiveMemberHeader); 60 61 // Instantiate the ArchiveMember to be filled 62 ArchiveMember* member = new ArchiveMember(this); 63 64 // Extract the size and determine if the file is 65 // compressed or not (negative length). 66 int flags = 0; 67 int MemberSize = atoi(Hdr->size); 68 if (MemberSize < 0) { 69 flags |= ArchiveMember::CompressedFlag; 70 MemberSize = -MemberSize; 71 } 72 73 // Check the size of the member for sanity 74 if (At + MemberSize > End) 75 throw std::string("invalid member length in archive file"); 76 77 // Check the member signature 78 if (!Hdr->checkSignature()) 79 throw std::string("invalid file member signature"); 80 81 // Convert and check the member name 82 // The empty name ( '/' and 15 blanks) is for a foreign (non-LLVM) symbol 83 // table. The special name "//" and 14 blanks is for a string table, used 84 // for long file names. This library doesn't generate either of those but 85 // it will accept them. If the name starts with #1/ and the remainder is 86 // digits, then those digits specify the length of the name that is 87 // stored immediately following the header. The special name 88 // __LLVM_SYM_TAB__ identifies the symbol table for LLVM bytecode. 89 // Anything else is a regular, short filename that is terminated with 90 // a '/' and blanks. 91 92 std::string pathname; 93 unsigned index; 94 switch (Hdr->name[0]) { 95 case '#': 96 if (Hdr->name[1] == '1' && Hdr->name[2] == '/') { 97 if (isdigit(Hdr->name[3])) { 98 unsigned len = atoi(&Hdr->name[3]); 99 pathname.assign(At, len); 100 At += len; 101 MemberSize -= len; 102 flags |= ArchiveMember::HasLongFilenameFlag; 103 } else 104 throw std::string("invalid long filename"); 105 } else if (Hdr->name[1] == '_' && 106 (0 == memcmp(Hdr->name, ARFILE_LLVM_SYMTAB_NAME, 16))) { 107 // The member is using a long file name (>15 chars) format. 108 // This format is standard for 4.4BSD and Mac OSX operating 109 // systems. LLVM uses it similarly. In this format, the 110 // remainder of the name field (after #1/) specifies the 111 // length of the file name which occupy the first bytes of 112 // the member's data. The pathname already has the #1/ stripped. 113 pathname.assign(ARFILE_LLVM_SYMTAB_NAME); 114 flags |= ArchiveMember::LLVMSymbolTableFlag; 115 } 116 break; 117 case '/': 118 if (Hdr->name[1]== '/') { 119 if (0 == memcmp(Hdr->name, ARFILE_STRTAB_NAME, 16)) { 120 pathname.assign(ARFILE_STRTAB_NAME); 121 flags |= ArchiveMember::StringTableFlag; 122 } else { 123 throw std::string("invalid string table name"); 124 } 125 } else if (Hdr->name[1] == ' ') { 126 if (0 == memcmp(Hdr->name, ARFILE_SVR4_SYMTAB_NAME, 16)) { 127 pathname.assign(ARFILE_SVR4_SYMTAB_NAME); 128 flags |= ArchiveMember::SVR4SymbolTableFlag; 129 } else { 130 throw std::string("invalid SVR4 symbol table name"); 131 } 132 } else if (isdigit(Hdr->name[1])) { 133 unsigned index = atoi(&Hdr->name[1]); 134 if (index < strtab.length()) { 135 const char* namep = strtab.c_str() + index; 136 const char* endp = strtab.c_str() + strtab.length(); 137 const char* p = namep; 138 const char* last_p = p; 139 while (p < endp) { 140 if (*p == '\n' && *last_p == '/') { 141 pathname.assign(namep, last_p - namep); 142 flags |= ArchiveMember::HasLongFilenameFlag; 143 break; 144 } 145 last_p = p; 146 p++; 147 } 148 if (p >= endp) 149 throw std::string("missing name termiantor in string table"); 150 } else { 151 throw std::string("name index beyond string table"); 152 } 153 } 154 break; 155 case '_': 156 if (Hdr->name[1] == '_' && 157 (0 == memcmp(Hdr->name, ARFILE_BSD4_SYMTAB_NAME, 16))) { 158 pathname.assign(ARFILE_BSD4_SYMTAB_NAME); 159 flags |= ArchiveMember::BSD4SymbolTableFlag; 160 break; 161 } 162 /* FALL THROUGH */ 163 164 default: 165 char* slash = (char*) memchr(Hdr->name, '/', 16); 166 if (slash == 0) 167 slash = Hdr->name + 16; 168 pathname.assign(Hdr->name, slash - Hdr->name); 169 break; 170 } 171 172 // Determine if this is a bytecode file 173 switch (sys::IdentifyFileType(At, 4)) { 174 case sys::BytecodeFileType: 175 flags |= ArchiveMember::BytecodeFlag; 176 break; 177 case sys::CompressedBytecodeFileType: 178 flags |= ArchiveMember::CompressedBytecodeFlag; 179 flags &= ~ArchiveMember::CompressedFlag; 180 break; 181 default: 182 flags &= ~(ArchiveMember::BytecodeFlag| 183 ArchiveMember::CompressedBytecodeFlag); 184 break; 185 } 186 187 // Fill in fields of the ArchiveMember 188 member->next = 0; 189 member->prev = 0; 190 member->parent = this; 191 member->path.setFile(pathname); 192 member->info.fileSize = MemberSize; 193 member->info.modTime.fromEpochTime(atoi(Hdr->date)); 194 sscanf(Hdr->mode, "%o", &(member->info.mode)); 195 member->info.user = atoi(Hdr->uid); 196 member->info.group = atoi(Hdr->gid); 197 member->flags = flags; 198 member->data = At; 199 200 return member; 201} 202 203void 204Archive::checkSignature() { 205 // Check the magic string at file's header 206 if (mapfile->size() < 8 || memcmp(base, ARFILE_MAGIC, 8)) 207 throw std::string("invalid signature for an archive file"); 208} 209 210// This function loads the entire archive and fully populates its ilist with 211// the members of the archive file. This is typically used in preparation for 212// editing the contents of the archive. 213void 214Archive::loadArchive() { 215 216 // Set up parsing 217 members.clear(); 218 symTab.clear(); 219 const char *At = base; 220 const char *End = base + mapfile->size(); 221 222 checkSignature(); 223 At += 8; // Skip the magic string. 224 225 bool seenSymbolTable = false; 226 bool foundFirstFile = false; 227 while (At < End) { 228 // parse the member header 229 const char* Save = At; 230 ArchiveMember* mbr = parseMemberHeader(At, End); 231 232 // check if this is the foreign symbol table 233 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) { 234 // We just save this but don't do anything special 235 // with it. It doesn't count as the "first file". 236 if (foreignST) { 237 // What? Multiple foreign symbol tables? Just chuck it 238 // and retain the last one found. 239 delete foreignST; 240 } 241 foreignST = mbr; 242 At += mbr->getSize(); 243 if ((intptr_t(At) & 1) == 1) 244 At++; 245 } else if (mbr->isStringTable()) { 246 // Simply suck the entire string table into a string 247 // variable. This will be used to get the names of the 248 // members that use the "/ddd" format for their names 249 // (SVR4 style long names). 250 strtab.assign(At, mbr->getSize()); 251 At += mbr->getSize(); 252 if ((intptr_t(At) & 1) == 1) 253 At++; 254 delete mbr; 255 } else if (mbr->isLLVMSymbolTable()) { 256 // This is the LLVM symbol table for the archive. If we've seen it 257 // already, its an error. Otherwise, parse the symbol table and move on. 258 if (seenSymbolTable) 259 throw std::string("invalid archive: multiple symbol tables"); 260 parseSymbolTable(mbr->getData(), mbr->getSize()); 261 seenSymbolTable = true; 262 At += mbr->getSize(); 263 if ((intptr_t(At) & 1) == 1) 264 At++; 265 delete mbr; // We don't need this member in the list of members. 266 } else { 267 // This is just a regular file. If its the first one, save its offset. 268 // Otherwise just push it on the list and move on to the next file. 269 if (!foundFirstFile) { 270 firstFileOffset = Save - base; 271 foundFirstFile = true; 272 } 273 members.push_back(mbr); 274 At += mbr->getSize(); 275 if ((intptr_t(At) & 1) == 1) 276 At++; 277 } 278 } 279} 280 281// Open and completely load the archive file. 282Archive* 283Archive::OpenAndLoad(const sys::Path& file) { 284 285 Archive* result = new Archive(file, true); 286 287 result->loadArchive(); 288 289 return result; 290} 291 292// Get all the bytecode modules from the archive 293bool 294Archive::getAllModules(std::vector<Module*>& Modules, std::string* ErrMessage) { 295 296 for (iterator I=begin(), E=end(); I != E; ++I) { 297 if (I->isBytecode() || I->isCompressedBytecode()) { 298 std::string FullMemberName = archPath.get() + 299 "(" + I->getPath().get() + ")"; 300 Module* M = ParseBytecodeBuffer((const unsigned char*)I->getData(), 301 I->getSize(), FullMemberName, ErrMessage); 302 if (!M) 303 return true; 304 305 Modules.push_back(M); 306 } 307 } 308 return false; 309} 310 311// Load just the symbol table from the archive file 312void 313Archive::loadSymbolTable() { 314 315 // Set up parsing 316 members.clear(); 317 symTab.clear(); 318 const char *At = base; 319 const char *End = base + mapfile->size(); 320 321 // Make sure we're dealing with an archive 322 checkSignature(); 323 324 At += 8; // Skip signature 325 326 // Parse the first file member header 327 const char* FirstFile = At; 328 ArchiveMember* mbr = parseMemberHeader(At, End); 329 330 if (mbr->isSVR4SymbolTable() || mbr->isBSD4SymbolTable()) { 331 // Skip the foreign symbol table, we don't do anything with it 332 At += mbr->getSize(); 333 if ((intptr_t(At) & 1) == 1) 334 At++; 335 delete mbr; 336 337 // Read the next one 338 FirstFile = At; 339 mbr = parseMemberHeader(At, End); 340 } 341 342 if (mbr->isStringTable()) { 343 // Process the string table entry 344 strtab.assign((const char*)mbr->getData(), mbr->getSize()); 345 At += mbr->getSize(); 346 if ((intptr_t(At) & 1) == 1) 347 At++; 348 delete mbr; 349 // Get the next one 350 FirstFile = At; 351 mbr = parseMemberHeader(At, End); 352 } 353 354 // See if its the symbol table 355 if (mbr->isLLVMSymbolTable()) { 356 parseSymbolTable(mbr->getData(), mbr->getSize()); 357 At += mbr->getSize(); 358 if ((intptr_t(At) & 1) == 1) 359 At++; 360 FirstFile = At; 361 } else { 362 // There's no symbol table in the file. We have to rebuild it from scratch 363 // because the intent of this method is to get the symbol table loaded so 364 // it can be searched efficiently. 365 // Add the member to the members list 366 members.push_back(mbr); 367 } 368 369 firstFileOffset = FirstFile - base; 370} 371 372// Open the archive and load just the symbol tables 373Archive* 374Archive::OpenAndLoadSymbols(const sys::Path& file) { 375 Archive* result = new Archive(file, true); 376 377 result->loadSymbolTable(); 378 379 return result; 380} 381 382// Look up one symbol in the symbol table and return a ModuleProvider for the 383// module that defines that symbol. 384ModuleProvider* 385Archive::findModuleDefiningSymbol(const std::string& symbol) { 386 SymTabType::iterator SI = symTab.find(symbol); 387 if (SI == symTab.end()) 388 return 0; 389 390 // The symbol table was previously constructed assuming that the members were 391 // written without the symbol table header. Because VBR encoding is used, the 392 // values could not be adjusted to account for the offset of the symbol table 393 // because that could affect the size of the symbol table due to VBR encoding. 394 // We now have to account for this by adjusting the offset by the size of the 395 // symbol table and its header. 396 unsigned fileOffset = 397 SI->second + // offset in symbol-table-less file 398 firstFileOffset; // add offset to first "real" file in archive 399 400 // See if the module is already loaded 401 ModuleMap::iterator MI = modules.find(fileOffset); 402 if (MI != modules.end()) 403 return MI->second.first; 404 405 // Module hasn't been loaded yet, we need to load it 406 const char* modptr = base + fileOffset; 407 ArchiveMember* mbr = parseMemberHeader(modptr, base + mapfile->size()); 408 409 // Now, load the bytecode module to get the ModuleProvider 410 std::string FullMemberName = archPath.get() + "(" + 411 mbr->getPath().get() + ")"; 412 ModuleProvider* mp = getBytecodeBufferModuleProvider( 413 (const unsigned char*) mbr->getData(), mbr->getSize(), 414 FullMemberName, 0); 415 416 modules.insert(std::make_pair(fileOffset, std::make_pair(mp, mbr))); 417 418 return mp; 419} 420 421// Look up multiple symbols in the symbol table and return a set of 422// ModuleProviders that define those symbols. 423void 424Archive::findModulesDefiningSymbols(std::set<std::string>& symbols, 425 std::set<ModuleProvider*>& result) 426{ 427 assert(mapfile && base && "Can't findModulesDefiningSymbols on new archive"); 428 if (symTab.empty()) { 429 // We don't have a symbol table, so we must build it now but lets also 430 // make sure that we populate the modules table as we do this to ensure 431 // that we don't load them twice when findModuleDefiningSymbol is called 432 // below. 433 434 // Get a pointer to the first file 435 const char* At = ((const char*)base) + firstFileOffset; 436 const char* End = ((const char*)base) + mapfile->size(); 437 438 while ( At < End) { 439 // Compute the offset to be put in the symbol table 440 unsigned offset = At - base - firstFileOffset; 441 442 // Parse the file's header 443 ArchiveMember* mbr = parseMemberHeader(At, End); 444 445 // If it contains symbols 446 if (mbr->isBytecode() || mbr->isCompressedBytecode()) { 447 // Get the symbols 448 std::vector<std::string> symbols; 449 std::string FullMemberName = archPath.get() + "(" + 450 mbr->getPath().get() + ")"; 451 ModuleProvider* MP = GetBytecodeSymbols((const unsigned char*)At, 452 mbr->getSize(), FullMemberName, symbols); 453 454 if (MP) { 455 // Insert the module's symbols into the symbol table 456 for (std::vector<std::string>::iterator I = symbols.begin(), 457 E=symbols.end(); I != E; ++I ) { 458 symTab.insert(std::make_pair(*I, offset)); 459 } 460 // Insert the ModuleProvider and the ArchiveMember into the table of 461 // modules. 462 modules.insert(std::make_pair(offset, std::make_pair(MP, mbr))); 463 } else { 464 throw std::string("Can't parse bytecode member: ") + 465 mbr->getPath().get(); 466 } 467 } 468 469 // Go to the next file location 470 At += mbr->getSize(); 471 if ((intptr_t(At) & 1) == 1) 472 At++; 473 } 474 } 475 476 // At this point we have a valid symbol table (one way or another) so we 477 // just use it to quickly find the symbols requested. 478 479 for (std::set<std::string>::iterator I=symbols.begin(), 480 E=symbols.end(); I != E;) { 481 // See if this symbol exists 482 ModuleProvider* mp = findModuleDefiningSymbol(*I); 483 if (mp) { 484 // The symbol exists, insert the ModuleProvider into our result, 485 // duplicates wil be ignored 486 result.insert(mp); 487 488 // Remove the symbol now that its been resolved, being careful to 489 // post-increment the iterator. 490 symbols.erase(I++); 491 } else { 492 ++I; 493 } 494 } 495} 496