John Criswell | 71478b7 | 2003-09-19 20:24:23 +0000 | [diff] [blame^] | 1 | //===- gccld.cpp - LLVM 'ld' compatible linker ----------------------------===// |
| 2 | // |
| 3 | // This utility is intended to be compatible with GCC, and follows standard |
| 4 | // system 'ld' conventions. As such, the default output file is ./a.out. |
| 5 | // Additionally, this program outputs a shell script that is used to invoke LLI |
| 6 | // to execute the program. In this manner, the generated executable (a.out for |
| 7 | // example), is directly executable, whereas the bytecode file actually lives in |
| 8 | // the a.out.bc file generated by this program. Also, Force is on by default. |
| 9 | // |
| 10 | // Note that if someone (or a script) deletes the executable program generated, |
| 11 | // the .bc file will be left around. Considering that this is a temporary hack, |
| 12 | // I'm not too worried about this. |
| 13 | // |
| 14 | //===----------------------------------------------------------------------===// |
| 15 | |
| 16 | #include "llvm/Transforms/Utils/Linker.h" |
| 17 | #include "llvm/Module.h" |
| 18 | #include "llvm/PassManager.h" |
| 19 | #include "llvm/Bytecode/Reader.h" |
| 20 | #include "llvm/Bytecode/WriteBytecodePass.h" |
| 21 | #include "llvm/Target/TargetData.h" |
| 22 | #include "llvm/Transforms/IPO.h" |
| 23 | #include "llvm/Transforms/Scalar.h" |
| 24 | #include "Support/FileUtilities.h" |
| 25 | #include "Support/SystemUtils.h" |
| 26 | #include "Support/CommandLine.h" |
| 27 | #include "Support/Signals.h" |
| 28 | #include "Config/stdlib.h" |
| 29 | #include "gccld.h" |
| 30 | |
| 31 | #include <fstream> |
| 32 | #include <memory> |
| 33 | #include <set> |
| 34 | #include <algorithm> |
| 35 | |
| 36 | // |
| 37 | // Function: FileExists () |
| 38 | // |
| 39 | // Description: |
| 40 | // Determine if the specified filename exists and is readable. |
| 41 | // |
| 42 | // Inputs: |
| 43 | // FN - The name of the file. |
| 44 | // |
| 45 | // Outputs: |
| 46 | // None. |
| 47 | // |
| 48 | // Return Value: |
| 49 | // TRUE - The file exists and is readable. |
| 50 | // FALSE - The file does not exist or is unreadable. |
| 51 | // |
| 52 | static inline bool |
| 53 | FileExists(const std::string &FN) |
| 54 | { |
| 55 | return access(FN.c_str(), R_OK | F_OK) != -1; |
| 56 | } |
| 57 | |
| 58 | // |
| 59 | // Function: IsArchive () |
| 60 | // |
| 61 | // Description: |
| 62 | // Determine if the specified file is an ar archive. It determines this by |
| 63 | // checking the magic string at the beginning of the file. |
| 64 | // |
| 65 | // Inputs: |
| 66 | // filename - A C++ string containing the name of the file. |
| 67 | // |
| 68 | // Outputs: |
| 69 | // None. |
| 70 | // |
| 71 | // Return value: |
| 72 | // TRUE - The file is an archive. |
| 73 | // FALSE - The file is not an archive. |
| 74 | // |
| 75 | static inline bool |
| 76 | IsArchive (const std::string &filename) |
| 77 | { |
| 78 | std::string ArchiveMagic("!<arch>\012"); |
| 79 | char buf[1 + ArchiveMagic.size()]; |
| 80 | |
| 81 | std::ifstream f(filename.c_str()); |
| 82 | f.read(buf, ArchiveMagic.size()); |
| 83 | buf[ArchiveMagic.size()] = '\0'; |
| 84 | return ArchiveMagic == buf; |
| 85 | } |
| 86 | |
| 87 | // |
| 88 | // Function: FindLib () |
| 89 | // |
| 90 | // Description: |
| 91 | // This function locates a particular library. It will prepend and append |
| 92 | // various directories, prefixes, and suffixes until it can find the library. |
| 93 | // |
| 94 | // Inputs: |
| 95 | // Filename - Name of the file to find. |
| 96 | // Paths - List of directories to search. |
| 97 | // |
| 98 | // Outputs: |
| 99 | // None. |
| 100 | // |
| 101 | // Return value: |
| 102 | // The name of the file is returned. |
| 103 | // If the file is not found, an empty string is returned. |
| 104 | // |
| 105 | static std::string |
| 106 | FindLib (const std::string & Filename, const std::vector<std::string> & Paths) |
| 107 | { |
| 108 | // |
| 109 | // Determine if the pathname can be found as it stands. |
| 110 | // |
| 111 | if (FileExists (Filename)) |
| 112 | { |
| 113 | return Filename; |
| 114 | } |
| 115 | |
| 116 | // |
| 117 | // If that doesn't work, convert the name into a library name. |
| 118 | // |
| 119 | std::string LibName = "lib" + Filename; |
| 120 | |
| 121 | // |
| 122 | // Iterate over the directories in Paths to see if we can find the library |
| 123 | // there. |
| 124 | // |
| 125 | for (unsigned Index = 0; Index != Paths.size(); ++Index) |
| 126 | { |
| 127 | std::string Directory = Paths[Index] + "/"; |
| 128 | |
| 129 | if (FileExists (Directory + LibName + ".bc")) |
| 130 | { |
| 131 | return Directory + LibName + ".bc"; |
| 132 | } |
| 133 | |
| 134 | if (FileExists (Directory + LibName + ".so")) |
| 135 | { |
| 136 | return Directory + LibName + ".so"; |
| 137 | } |
| 138 | |
| 139 | if (FileExists (Directory + LibName + ".a")) |
| 140 | { |
| 141 | return Directory + LibName + ".a"; |
| 142 | } |
| 143 | } |
| 144 | |
| 145 | // |
| 146 | // One last hope: Check LLVM_LIB_SEARCH_PATH. |
| 147 | // |
| 148 | char *SearchPath = getenv("LLVM_LIB_SEARCH_PATH"); |
| 149 | if (SearchPath == NULL) |
| 150 | { |
| 151 | return std::string(); |
| 152 | } |
| 153 | |
| 154 | LibName = std::string(SearchPath) + "/" + LibName; |
| 155 | if (FileExists (LibName)) |
| 156 | { |
| 157 | return LibName; |
| 158 | } |
| 159 | |
| 160 | return std::string(); |
| 161 | } |
| 162 | |
| 163 | // |
| 164 | // Function: GetAllDefinedSymbols () |
| 165 | // |
| 166 | // Description: |
| 167 | // Find all of the defined symbols in the specified module. |
| 168 | // |
| 169 | // Inputs: |
| 170 | // M - The module in which to find defined symbols. |
| 171 | // |
| 172 | // Outputs: |
| 173 | // DefinedSymbols - A set of C++ strings that will contain the name of all |
| 174 | // defined symbols. |
| 175 | // |
| 176 | // Return value: |
| 177 | // None. |
| 178 | // |
| 179 | void |
| 180 | GetAllDefinedSymbols (Module *M, std::set<std::string> &DefinedSymbols) |
| 181 | { |
| 182 | for (Module::iterator I = M->begin(), E = M->end(); I != E; ++I) |
| 183 | if (I->hasName() && !I->isExternal() && !I->hasInternalLinkage()) |
| 184 | DefinedSymbols.insert(I->getName()); |
| 185 | for (Module::giterator I = M->gbegin(), E = M->gend(); I != E; ++I) |
| 186 | if (I->hasName() && !I->isExternal() && !I->hasInternalLinkage()) |
| 187 | DefinedSymbols.insert(I->getName()); |
| 188 | } |
| 189 | |
| 190 | // |
| 191 | // Function: GetAllUndefinedSymbols () |
| 192 | // |
| 193 | // Description: |
| 194 | // This calculates the set of undefined symbols that still exist in an LLVM |
| 195 | // module. This is a bit tricky because there may be two symbols with the |
| 196 | // same name but different LLVM types that will be resolved to each other but |
| 197 | // aren't currently (thus we need to treat it as resolved). |
| 198 | // |
| 199 | // Inputs: |
| 200 | // M - The module in which to find undefined symbols. |
| 201 | // |
| 202 | // Outputs: |
| 203 | // UndefinedSymbols - A set of C++ strings containing the name of all |
| 204 | // undefined symbols. |
| 205 | // |
| 206 | // Return value: |
| 207 | // None. |
| 208 | // |
| 209 | void |
| 210 | GetAllUndefinedSymbols(Module *M, std::set<std::string> &UndefinedSymbols) |
| 211 | { |
| 212 | std::set<std::string> DefinedSymbols; |
| 213 | UndefinedSymbols.clear(); // Start out empty |
| 214 | |
| 215 | for (Module::iterator I = M->begin(), E = M->end(); I != E; ++I) |
| 216 | if (I->hasName()) { |
| 217 | if (I->isExternal()) |
| 218 | UndefinedSymbols.insert(I->getName()); |
| 219 | else if (!I->hasInternalLinkage()) |
| 220 | DefinedSymbols.insert(I->getName()); |
| 221 | } |
| 222 | for (Module::giterator I = M->gbegin(), E = M->gend(); I != E; ++I) |
| 223 | if (I->hasName()) { |
| 224 | if (I->isExternal()) |
| 225 | UndefinedSymbols.insert(I->getName()); |
| 226 | else if (!I->hasInternalLinkage()) |
| 227 | DefinedSymbols.insert(I->getName()); |
| 228 | } |
| 229 | |
| 230 | // Prune out any defined symbols from the undefined symbols set... |
| 231 | for (std::set<std::string>::iterator I = UndefinedSymbols.begin(); |
| 232 | I != UndefinedSymbols.end(); ) |
| 233 | if (DefinedSymbols.count(*I)) |
| 234 | UndefinedSymbols.erase(I++); // This symbol really is defined! |
| 235 | else |
| 236 | ++I; // Keep this symbol in the undefined symbols list |
| 237 | } |
| 238 | |
| 239 | |
| 240 | // |
| 241 | // Function: LoadObject () |
| 242 | // |
| 243 | // Description: |
| 244 | // Read the specified bytecode object file. |
| 245 | // |
| 246 | // Inputs: |
| 247 | // FN - The name of the file to load. |
| 248 | // |
| 249 | // Outputs: |
| 250 | // OutErrorMessage - The error message to give back to the caller. |
| 251 | // |
| 252 | // Return Value: |
| 253 | // A pointer to a module represening the bytecode file is returned. |
| 254 | // If an error occurs, the pointer is 0. |
| 255 | // |
| 256 | std::auto_ptr<Module> |
| 257 | LoadObject (const std::string & FN, std::string &OutErrorMessage) |
| 258 | { |
| 259 | std::string ErrorMessage; |
| 260 | Module *Result = ParseBytecodeFile(FN, &ErrorMessage); |
| 261 | if (Result) return std::auto_ptr<Module>(Result); |
| 262 | |
| 263 | OutErrorMessage = "Bytecode file '" + FN + "' corrupt!"; |
| 264 | if (ErrorMessage.size()) OutErrorMessage += ": " + ErrorMessage; |
| 265 | return std::auto_ptr<Module>(); |
| 266 | } |
| 267 | |
| 268 | // |
| 269 | // Function: LinkInArchive () |
| 270 | // |
| 271 | // Description: |
| 272 | // This function will open an archive library and link in all objects which |
| 273 | // provide symbols that are currently undefined. |
| 274 | // |
| 275 | // Inputs: |
| 276 | // M - The module in which to link the archives. |
| 277 | // Filename - The pathname of the archive. |
| 278 | // Verbose - Flags whether verbose messages should be printed. |
| 279 | // |
| 280 | // Outputs: |
| 281 | // ErrorMessage - A C++ string detailing what error occurred, if any. |
| 282 | // |
| 283 | // Return Value: |
| 284 | // TRUE - An error occurred. |
| 285 | // FALSE - No errors. |
| 286 | // |
| 287 | static bool |
| 288 | LinkInArchive (Module * M, |
| 289 | const std::string & Filename, |
| 290 | std::string & ErrorMessage, |
| 291 | bool Verbose) |
| 292 | { |
| 293 | // |
| 294 | // Find all of the symbols currently undefined in the bytecode program. |
| 295 | // If all the symbols are defined, the program is complete, and there is |
| 296 | // no reason to link in any archive files. |
| 297 | // |
| 298 | std::set<std::string> UndefinedSymbols; |
| 299 | GetAllUndefinedSymbols (M, UndefinedSymbols); |
| 300 | if (UndefinedSymbols.empty()) |
| 301 | { |
| 302 | if (Verbose) std::cerr << " No symbols undefined, don't link library!\n"; |
| 303 | return false; // No need to link anything in! |
| 304 | } |
| 305 | |
| 306 | // |
| 307 | // Load in the archive objects. |
| 308 | // |
| 309 | if (Verbose) std::cerr << " Loading '" << Filename << "'\n"; |
| 310 | std::vector<Module*> Objects; |
| 311 | if (ReadArchiveFile (Filename, Objects, &ErrorMessage)) |
| 312 | { |
| 313 | return true; |
| 314 | } |
| 315 | |
| 316 | // |
| 317 | // Figure out which symbols are defined by all of the modules in the archive. |
| 318 | // |
| 319 | std::vector<std::set<std::string> > DefinedSymbols; |
| 320 | DefinedSymbols.resize (Objects.size()); |
| 321 | for (unsigned i = 0; i != Objects.size(); ++i) |
| 322 | { |
| 323 | GetAllDefinedSymbols(Objects[i], DefinedSymbols[i]); |
| 324 | } |
| 325 | |
| 326 | // While we are linking in object files, loop. |
| 327 | bool Linked = true; |
| 328 | while (Linked) |
| 329 | { |
| 330 | Linked = false; |
| 331 | |
| 332 | for (unsigned i = 0; i != Objects.size(); ++i) { |
| 333 | // Consider whether we need to link in this module... we only need to |
| 334 | // link it in if it defines some symbol which is so far undefined. |
| 335 | // |
| 336 | const std::set<std::string> &DefSymbols = DefinedSymbols[i]; |
| 337 | |
| 338 | bool ObjectRequired = false; |
| 339 | for (std::set<std::string>::iterator I = UndefinedSymbols.begin(), |
| 340 | E = UndefinedSymbols.end(); I != E; ++I) |
| 341 | if (DefSymbols.count(*I)) { |
| 342 | if (Verbose) |
| 343 | std::cerr << " Found object providing symbol '" << *I << "'...\n"; |
| 344 | ObjectRequired = true; |
| 345 | break; |
| 346 | } |
| 347 | |
| 348 | // We DO need to link this object into the program... |
| 349 | if (ObjectRequired) { |
| 350 | if (LinkModules(M, Objects[i], &ErrorMessage)) |
| 351 | return true; // Couldn't link in the right object file... |
| 352 | |
| 353 | // Since we have linked in this object, delete it from the list of |
| 354 | // objects to consider in this archive file. |
| 355 | std::swap(Objects[i], Objects.back()); |
| 356 | std::swap(DefinedSymbols[i], DefinedSymbols.back()); |
| 357 | Objects.pop_back(); |
| 358 | DefinedSymbols.pop_back(); |
| 359 | --i; // Do not skip an entry |
| 360 | |
| 361 | // The undefined symbols set should have shrunk. |
| 362 | GetAllUndefinedSymbols(M, UndefinedSymbols); |
| 363 | Linked = true; // We have linked something in! |
| 364 | } |
| 365 | } |
| 366 | } |
| 367 | |
| 368 | return false; |
| 369 | } |
| 370 | |
| 371 | // |
| 372 | // Function: LinkInFile () |
| 373 | // |
| 374 | // Description: |
| 375 | // This function will open an archive library and link in all objects which |
| 376 | // provide symbols that are currently undefined. |
| 377 | // |
| 378 | // Inputs: |
| 379 | // HeadModule - The module in which to link the archives. |
| 380 | // Filename - The pathname of the archive. |
| 381 | // Verbose - Flags whether verbose messages should be printed. |
| 382 | // |
| 383 | // Outputs: |
| 384 | // ErrorMessage - A C++ string detailing what error occurred, if any. |
| 385 | // |
| 386 | // Return Value: |
| 387 | // TRUE - An error occurred. |
| 388 | // FALSE - No errors. |
| 389 | // |
| 390 | static bool |
| 391 | LinkInFile (Module * HeadModule, |
| 392 | const std::string & Filename, |
| 393 | std::string & ErrorMessage, |
| 394 | bool Verbose) |
| 395 | { |
| 396 | std::auto_ptr<Module> M(LoadObject(Filename, ErrorMessage)); |
| 397 | if (M.get() == 0) |
| 398 | { |
| 399 | return true; |
| 400 | } |
| 401 | |
| 402 | if (Verbose) std::cerr << "Linking in '" << Filename << "'\n"; |
| 403 | |
| 404 | return (LinkModules (HeadModule, M.get(), &ErrorMessage)); |
| 405 | } |
| 406 | |
| 407 | // |
| 408 | // Function: LinkFiles () |
| 409 | // |
| 410 | // Description: |
| 411 | // This function takes a module and a list of files and links them all |
| 412 | // together. It locates the file either in the current directory, as it's |
| 413 | // absolute or relative pathname, or as a file somewhere in |
| 414 | // LLVM_LIB_SEARCH_PATH. |
| 415 | // |
| 416 | // Inputs: |
| 417 | // progname - The name of the program (infamous argv[0]). |
| 418 | // HeadModule - The module under which all files will be linked. |
| 419 | // Files - A vector of C++ strings indicating the LLVM bytecode filenames |
| 420 | // to be linked. The names can refer to a mixture of pure LLVM |
| 421 | // bytecode files and archive (ar) formatted files. |
| 422 | // Verbose - Flags whether verbose output should be printed while linking. |
| 423 | // |
| 424 | // Outputs: |
| 425 | // HeadModule - The module will have the specified LLVM bytecode files linked |
| 426 | // in. |
| 427 | // |
| 428 | // Return value: |
| 429 | // FALSE - No errors. |
| 430 | // TRUE - Some error occurred. |
| 431 | // |
| 432 | bool |
| 433 | LinkFiles (const char * progname, |
| 434 | Module * HeadModule, |
| 435 | const std::vector<std::string> & Files, |
| 436 | bool Verbose) |
| 437 | { |
| 438 | // String in which to receive error messages. |
| 439 | std::string ErrorMessage; |
| 440 | |
| 441 | // Full pathname of the file |
| 442 | std::string Pathname; |
| 443 | |
| 444 | // Get the library search path from the environment |
| 445 | char *SearchPath = getenv("LLVM_LIB_SEARCH_PATH"); |
| 446 | |
| 447 | for (unsigned i = 1; i < Files.size(); ++i) |
| 448 | { |
| 449 | // |
| 450 | // Determine where this file lives. |
| 451 | // |
| 452 | if (FileExists (Files[i])) |
| 453 | { |
| 454 | Pathname = Files[i]; |
| 455 | } |
| 456 | else |
| 457 | { |
| 458 | if (SearchPath == NULL) |
| 459 | { |
| 460 | std::cerr << "Cannot find " << Files[i]; |
| 461 | return true; |
| 462 | } |
| 463 | |
| 464 | Pathname = std::string(SearchPath)+"/"+Files[i]; |
| 465 | if (!FileExists (Pathname)) |
| 466 | { |
| 467 | std::cerr << "Cannot find " << Files[i]; |
| 468 | return true; |
| 469 | } |
| 470 | } |
| 471 | |
| 472 | // |
| 473 | // A user may specify an ar archive without -l, perhaps because it |
| 474 | // is not installed as a library. Detect that and link the library. |
| 475 | // |
| 476 | if (IsArchive(Pathname)) |
| 477 | { |
| 478 | if (Verbose) |
| 479 | { |
| 480 | std::cerr << "Linking archive '" << Files[i] << "'\n"; |
| 481 | } |
| 482 | |
| 483 | if (LinkInArchive (HeadModule, Pathname, ErrorMessage, Verbose)) |
| 484 | { |
| 485 | PrintAndReturn(progname, ErrorMessage, |
| 486 | ": Error linking in '" + Files[i] + "'"); |
| 487 | return true; |
| 488 | } |
| 489 | } |
| 490 | else |
| 491 | { |
| 492 | if (Verbose) |
| 493 | { |
| 494 | std::cerr << "Linking file '" << Files[i] << "'\n"; |
| 495 | } |
| 496 | |
| 497 | if (LinkInFile (HeadModule, Pathname, ErrorMessage, Verbose)) |
| 498 | { |
| 499 | PrintAndReturn(progname, ErrorMessage, |
| 500 | ": error linking in '" + Files[i] + "'"); |
| 501 | return true; |
| 502 | } |
| 503 | } |
| 504 | } |
| 505 | |
| 506 | return false; |
| 507 | } |
| 508 | |
| 509 | // |
| 510 | // Function: LinkLibraries () |
| 511 | // |
| 512 | // Description: |
| 513 | // This function takes the specified library files and links them into the |
| 514 | // main bytecode object file. |
| 515 | // |
| 516 | // Inputs: |
| 517 | // progname - The name of the program (infamous argv[0]). |
| 518 | // HeadModule - The module into which all necessary libraries will be linked. |
| 519 | // Libraries - The list of libraries to link into the module. |
| 520 | // LibPaths - The list of library paths in which to find libraries. |
| 521 | // Verbose - Flags whether verbose messages should be printed. |
| 522 | // Native - Flags whether native code is being generated. |
| 523 | // |
| 524 | // Outputs: |
| 525 | // HeadModule - The module will have all necessary libraries linked in. |
| 526 | // |
| 527 | // Return value: |
| 528 | // FALSE - No error. |
| 529 | // TRUE - Error. |
| 530 | // |
| 531 | bool |
| 532 | LinkLibraries (const char * progname, |
| 533 | Module * HeadModule, |
| 534 | const std::vector<std::string> & Libraries, |
| 535 | const std::vector<std::string> & LibPaths, |
| 536 | bool Verbose, |
| 537 | bool Native) |
| 538 | { |
| 539 | // String in which to receive error messages. |
| 540 | std::string ErrorMessage; |
| 541 | |
| 542 | for (unsigned i = 1; i < Libraries.size(); ++i) |
| 543 | { |
| 544 | // |
| 545 | // Determine where this library lives. |
| 546 | // |
| 547 | std::string Pathname = FindLib (Libraries[i], LibPaths); |
| 548 | if (Pathname.empty()) |
| 549 | { |
| 550 | // |
| 551 | // If the pathname does not exist, then continue to the next one if |
| 552 | // we're doing a native link and give an error if we're doing a bytecode |
| 553 | // link. |
| 554 | // |
| 555 | if (Native) |
| 556 | { |
| 557 | continue; |
| 558 | } |
| 559 | else |
| 560 | { |
| 561 | PrintAndReturn (progname, "Cannot find " + Libraries[i]); |
| 562 | return true; |
| 563 | } |
| 564 | } |
| 565 | |
| 566 | // |
| 567 | // A user may specify an ar archive without -l, perhaps because it |
| 568 | // is not installed as a library. Detect that and link the library. |
| 569 | // |
| 570 | if (IsArchive(Pathname)) |
| 571 | { |
| 572 | if (Verbose) |
| 573 | { |
| 574 | std::cerr << "Linking archive '" << Libraries[i] << "'\n"; |
| 575 | } |
| 576 | |
| 577 | if (LinkInArchive (HeadModule, Pathname, ErrorMessage, Verbose)) |
| 578 | { |
| 579 | PrintAndReturn(progname, ErrorMessage, |
| 580 | ": Error linking in '" + Libraries[i] + "'"); |
| 581 | return true; |
| 582 | } |
| 583 | } |
| 584 | else |
| 585 | { |
| 586 | if (Verbose) |
| 587 | { |
| 588 | std::cerr << "Linking file '" << Libraries[i] << "'\n"; |
| 589 | } |
| 590 | |
| 591 | if (LinkInFile (HeadModule, Pathname, ErrorMessage, Verbose)) |
| 592 | { |
| 593 | PrintAndReturn(progname, ErrorMessage, |
| 594 | ": error linking in '" + Libraries[i] + "'"); |
| 595 | return true; |
| 596 | } |
| 597 | } |
| 598 | } |
| 599 | |
| 600 | return false; |
| 601 | } |