1 //===- MachO.h - MachO object file implementation ---------------*- 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 declares the MachOObjectFile class, which implement the ObjectFile 11 // interface for MachO files. 12 // 13 //===----------------------------------------------------------------------===// 14 15 #ifndef LLVM_OBJECT_MACHO_H 16 #define LLVM_OBJECT_MACHO_H 17 18 #include "llvm/ADT/ArrayRef.h" 19 #include "llvm/ADT/SmallString.h" 20 #include "llvm/ADT/SmallVector.h" 21 #include "llvm/ADT/StringExtras.h" 22 #include "llvm/ADT/StringRef.h" 23 #include "llvm/ADT/Triple.h" 24 #include "llvm/ADT/iterator_range.h" 25 #include "llvm/BinaryFormat/MachO.h" 26 #include "llvm/MC/SubtargetFeature.h" 27 #include "llvm/Object/Binary.h" 28 #include "llvm/Object/ObjectFile.h" 29 #include "llvm/Object/SymbolicFile.h" 30 #include "llvm/Support/Error.h" 31 #include "llvm/Support/Format.h" 32 #include "llvm/Support/MemoryBuffer.h" 33 #include "llvm/Support/raw_ostream.h" 34 #include <cstdint> 35 #include <memory> 36 #include <string> 37 #include <system_error> 38 39 namespace llvm { 40 namespace object { 41 42 /// DiceRef - This is a value type class that represents a single 43 /// data in code entry in the table in a Mach-O object file. 44 class DiceRef { 45 DataRefImpl DicePimpl; 46 const ObjectFile *OwningObject = nullptr; 47 48 public: 49 DiceRef() = default; 50 DiceRef(DataRefImpl DiceP, const ObjectFile *Owner); 51 52 bool operator==(const DiceRef &Other) const; 53 bool operator<(const DiceRef &Other) const; 54 55 void moveNext(); 56 57 std::error_code getOffset(uint32_t &Result) const; 58 std::error_code getLength(uint16_t &Result) const; 59 std::error_code getKind(uint16_t &Result) const; 60 61 DataRefImpl getRawDataRefImpl() const; 62 const ObjectFile *getObjectFile() const; 63 }; 64 using dice_iterator = content_iterator<DiceRef>; 65 66 /// ExportEntry encapsulates the current-state-of-the-walk used when doing a 67 /// non-recursive walk of the trie data structure. This allows you to iterate 68 /// across all exported symbols using: 69 /// Error Err; 70 /// for (const llvm::object::ExportEntry &AnExport : Obj->exports(&Err)) { 71 /// } 72 /// if (Err) { report error ... 73 class ExportEntry { 74 public: 75 ExportEntry(Error *Err, const MachOObjectFile *O, ArrayRef<uint8_t> Trie); 76 77 StringRef name() const; 78 uint64_t flags() const; 79 uint64_t address() const; 80 uint64_t other() const; 81 StringRef otherName() const; 82 uint32_t nodeOffset() const; 83 84 bool operator==(const ExportEntry &) const; 85 86 void moveNext(); 87 88 private: 89 friend class MachOObjectFile; 90 91 void moveToFirst(); 92 void moveToEnd(); 93 uint64_t readULEB128(const uint8_t *&p, const char **error); 94 void pushDownUntilBottom(); 95 void pushNode(uint64_t Offset); 96 97 // Represents a node in the mach-o exports trie. 98 struct NodeState { 99 NodeState(const uint8_t *Ptr); 100 101 const uint8_t *Start; 102 const uint8_t *Current; 103 uint64_t Flags = 0; 104 uint64_t Address = 0; 105 uint64_t Other = 0; 106 const char *ImportName = nullptr; 107 unsigned ChildCount = 0; 108 unsigned NextChildIndex = 0; 109 unsigned ParentStringLength = 0; 110 bool IsExportNode = false; 111 }; 112 using NodeList = SmallVector<NodeState, 16>; 113 using node_iterator = NodeList::const_iterator; 114 115 Error *E; 116 const MachOObjectFile *O; 117 ArrayRef<uint8_t> Trie; 118 SmallString<256> CumulativeString; 119 NodeList Stack; 120 bool Done = false; 121 122 iterator_range<node_iterator> nodes() const { 123 return make_range(Stack.begin(), Stack.end()); 124 } 125 }; 126 using export_iterator = content_iterator<ExportEntry>; 127 128 // Segment info so SegIndex/SegOffset pairs in a Mach-O Bind or Rebase entry 129 // can be checked and translated. Only the SegIndex/SegOffset pairs from 130 // checked entries are to be used with the segmentName(), sectionName() and 131 // address() methods below. 132 class BindRebaseSegInfo { 133 public: 134 BindRebaseSegInfo(const MachOObjectFile *Obj); 135 136 // Used to check a Mach-O Bind or Rebase entry for errors when iterating. 137 const char *checkSegAndOffset(int32_t SegIndex, uint64_t SegOffset, 138 bool endInvalid); 139 const char *checkCountAndSkip(uint32_t Count, uint32_t Skip, 140 uint8_t PointerSize, int32_t SegIndex, 141 uint64_t SegOffset); 142 // Used with valid SegIndex/SegOffset values from checked entries. 143 StringRef segmentName(int32_t SegIndex); 144 StringRef sectionName(int32_t SegIndex, uint64_t SegOffset); 145 uint64_t address(uint32_t SegIndex, uint64_t SegOffset); 146 147 private: 148 struct SectionInfo { 149 uint64_t Address; 150 uint64_t Size; 151 StringRef SectionName; 152 StringRef SegmentName; 153 uint64_t OffsetInSegment; 154 uint64_t SegmentStartAddress; 155 int32_t SegmentIndex; 156 }; 157 const SectionInfo &findSection(int32_t SegIndex, uint64_t SegOffset); 158 159 SmallVector<SectionInfo, 32> Sections; 160 int32_t MaxSegIndex; 161 }; 162 163 /// MachORebaseEntry encapsulates the current state in the decompression of 164 /// rebasing opcodes. This allows you to iterate through the compressed table of 165 /// rebasing using: 166 /// Error Err; 167 /// for (const llvm::object::MachORebaseEntry &Entry : Obj->rebaseTable(&Err)) { 168 /// } 169 /// if (Err) { report error ... 170 class MachORebaseEntry { 171 public: 172 MachORebaseEntry(Error *Err, const MachOObjectFile *O, 173 ArrayRef<uint8_t> opcodes, bool is64Bit); 174 175 int32_t segmentIndex() const; 176 uint64_t segmentOffset() const; 177 StringRef typeName() const; 178 StringRef segmentName() const; 179 StringRef sectionName() const; 180 uint64_t address() const; 181 182 bool operator==(const MachORebaseEntry &) const; 183 184 void moveNext(); 185 186 private: 187 friend class MachOObjectFile; 188 189 void moveToFirst(); 190 void moveToEnd(); 191 uint64_t readULEB128(const char **error); 192 193 Error *E; 194 const MachOObjectFile *O; 195 ArrayRef<uint8_t> Opcodes; 196 const uint8_t *Ptr; 197 uint64_t SegmentOffset = 0; 198 int32_t SegmentIndex = -1; 199 uint64_t RemainingLoopCount = 0; 200 uint64_t AdvanceAmount = 0; 201 uint8_t RebaseType = 0; 202 uint8_t PointerSize; 203 bool Done = false; 204 }; 205 using rebase_iterator = content_iterator<MachORebaseEntry>; 206 207 /// MachOBindEntry encapsulates the current state in the decompression of 208 /// binding opcodes. This allows you to iterate through the compressed table of 209 /// bindings using: 210 /// Error Err; 211 /// for (const llvm::object::MachOBindEntry &Entry : Obj->bindTable(&Err)) { 212 /// } 213 /// if (Err) { report error ... 214 class MachOBindEntry { 215 public: 216 enum class Kind { Regular, Lazy, Weak }; 217 218 MachOBindEntry(Error *Err, const MachOObjectFile *O, 219 ArrayRef<uint8_t> Opcodes, bool is64Bit, MachOBindEntry::Kind); 220 221 int32_t segmentIndex() const; 222 uint64_t segmentOffset() const; 223 StringRef typeName() const; 224 StringRef symbolName() const; 225 uint32_t flags() const; 226 int64_t addend() const; 227 int ordinal() const; 228 229 StringRef segmentName() const; 230 StringRef sectionName() const; 231 uint64_t address() const; 232 233 bool operator==(const MachOBindEntry &) const; 234 235 void moveNext(); 236 237 private: 238 friend class MachOObjectFile; 239 240 void moveToFirst(); 241 void moveToEnd(); 242 uint64_t readULEB128(const char **error); 243 int64_t readSLEB128(const char **error); 244 245 Error *E; 246 const MachOObjectFile *O; 247 ArrayRef<uint8_t> Opcodes; 248 const uint8_t *Ptr; 249 uint64_t SegmentOffset = 0; 250 int32_t SegmentIndex = -1; 251 StringRef SymbolName; 252 bool LibraryOrdinalSet = false; 253 int Ordinal = 0; 254 uint32_t Flags = 0; 255 int64_t Addend = 0; 256 uint64_t RemainingLoopCount = 0; 257 uint64_t AdvanceAmount = 0; 258 uint8_t BindType = 0; 259 uint8_t PointerSize; 260 Kind TableKind; 261 bool Done = false; 262 }; 263 using bind_iterator = content_iterator<MachOBindEntry>; 264 265 class MachOObjectFile : public ObjectFile { 266 public: 267 struct LoadCommandInfo { 268 const char *Ptr; // Where in memory the load command is. 269 MachO::load_command C; // The command itself. 270 }; 271 using LoadCommandList = SmallVector<LoadCommandInfo, 4>; 272 using load_command_iterator = LoadCommandList::const_iterator; 273 274 static Expected<std::unique_ptr<MachOObjectFile>> 275 create(MemoryBufferRef Object, bool IsLittleEndian, bool Is64Bits, 276 uint32_t UniversalCputype = 0, uint32_t UniversalIndex = 0); 277 278 void moveSymbolNext(DataRefImpl &Symb) const override; 279 280 uint64_t getNValue(DataRefImpl Sym) const; 281 Expected<StringRef> getSymbolName(DataRefImpl Symb) const override; 282 283 // MachO specific. 284 Error checkSymbolTable() const; 285 286 std::error_code getIndirectName(DataRefImpl Symb, StringRef &Res) const; 287 unsigned getSectionType(SectionRef Sec) const; 288 289 Expected<uint64_t> getSymbolAddress(DataRefImpl Symb) const override; 290 uint32_t getSymbolAlignment(DataRefImpl Symb) const override; 291 uint64_t getCommonSymbolSizeImpl(DataRefImpl Symb) const override; 292 Expected<SymbolRef::Type> getSymbolType(DataRefImpl Symb) const override; 293 uint32_t getSymbolFlags(DataRefImpl Symb) const override; 294 Expected<section_iterator> getSymbolSection(DataRefImpl Symb) const override; 295 unsigned getSymbolSectionID(SymbolRef Symb) const; 296 unsigned getSectionID(SectionRef Sec) const; 297 298 void moveSectionNext(DataRefImpl &Sec) const override; 299 std::error_code getSectionName(DataRefImpl Sec, 300 StringRef &Res) const override; 301 uint64_t getSectionAddress(DataRefImpl Sec) const override; 302 uint64_t getSectionIndex(DataRefImpl Sec) const override; 303 uint64_t getSectionSize(DataRefImpl Sec) const override; 304 std::error_code getSectionContents(DataRefImpl Sec, 305 StringRef &Res) const override; 306 uint64_t getSectionAlignment(DataRefImpl Sec) const override; 307 bool isSectionCompressed(DataRefImpl Sec) const override; 308 bool isSectionText(DataRefImpl Sec) const override; 309 bool isSectionData(DataRefImpl Sec) const override; 310 bool isSectionBSS(DataRefImpl Sec) const override; 311 bool isSectionVirtual(DataRefImpl Sec) const override; 312 bool isSectionBitcode(DataRefImpl Sec) const override; 313 314 /// When dsymutil generates the companion file, it strips all unnecessary 315 /// sections (e.g. everything in the _TEXT segment) by omitting their body 316 /// and setting the offset in their corresponding load command to zero. 317 /// 318 /// While the load command itself is valid, reading the section corresponds 319 /// to reading the number of bytes specified in the load command, starting 320 /// from offset 0 (i.e. the Mach-O header at the beginning of the file). 321 bool isSectionStripped(DataRefImpl Sec) const override; 322 323 relocation_iterator section_rel_begin(DataRefImpl Sec) const override; 324 relocation_iterator section_rel_end(DataRefImpl Sec) const override; 325 326 relocation_iterator extrel_begin() const; 327 relocation_iterator extrel_end() const; 328 iterator_range<relocation_iterator> external_relocations() const { 329 return make_range(extrel_begin(), extrel_end()); 330 } 331 332 void moveRelocationNext(DataRefImpl &Rel) const override; 333 uint64_t getRelocationOffset(DataRefImpl Rel) const override; 334 symbol_iterator getRelocationSymbol(DataRefImpl Rel) const override; 335 section_iterator getRelocationSection(DataRefImpl Rel) const; 336 uint64_t getRelocationType(DataRefImpl Rel) const override; 337 void getRelocationTypeName(DataRefImpl Rel, 338 SmallVectorImpl<char> &Result) const override; 339 uint8_t getRelocationLength(DataRefImpl Rel) const; 340 341 // MachO specific. 342 std::error_code getLibraryShortNameByIndex(unsigned Index, StringRef &) const; 343 uint32_t getLibraryCount() const; 344 345 section_iterator getRelocationRelocatedSection(relocation_iterator Rel) const; 346 347 // TODO: Would be useful to have an iterator based version 348 // of the load command interface too. 349 350 basic_symbol_iterator symbol_begin() const override; 351 basic_symbol_iterator symbol_end() const override; 352 353 // MachO specific. 354 basic_symbol_iterator getSymbolByIndex(unsigned Index) const; 355 uint64_t getSymbolIndex(DataRefImpl Symb) const; 356 357 section_iterator section_begin() const override; 358 section_iterator section_end() const override; 359 360 uint8_t getBytesInAddress() const override; 361 362 StringRef getFileFormatName() const override; 363 unsigned getArch() const override; 364 SubtargetFeatures getFeatures() const override { return SubtargetFeatures(); } 365 Triple getArchTriple(const char **McpuDefault = nullptr) const; 366 367 relocation_iterator section_rel_begin(unsigned Index) const; 368 relocation_iterator section_rel_end(unsigned Index) const; 369 370 dice_iterator begin_dices() const; 371 dice_iterator end_dices() const; 372 373 load_command_iterator begin_load_commands() const; 374 load_command_iterator end_load_commands() const; 375 iterator_range<load_command_iterator> load_commands() const; 376 377 /// For use iterating over all exported symbols. 378 iterator_range<export_iterator> exports(Error &Err) const; 379 380 /// For use examining a trie not in a MachOObjectFile. 381 static iterator_range<export_iterator> exports(Error &Err, 382 ArrayRef<uint8_t> Trie, 383 const MachOObjectFile *O = 384 nullptr); 385 386 /// For use iterating over all rebase table entries. 387 iterator_range<rebase_iterator> rebaseTable(Error &Err); 388 389 /// For use examining rebase opcodes in a MachOObjectFile. 390 static iterator_range<rebase_iterator> rebaseTable(Error &Err, 391 MachOObjectFile *O, 392 ArrayRef<uint8_t> Opcodes, 393 bool is64); 394 395 /// For use iterating over all bind table entries. 396 iterator_range<bind_iterator> bindTable(Error &Err); 397 398 /// For use iterating over all lazy bind table entries. 399 iterator_range<bind_iterator> lazyBindTable(Error &Err); 400 401 /// For use iterating over all weak bind table entries. 402 iterator_range<bind_iterator> weakBindTable(Error &Err); 403 404 /// For use examining bind opcodes in a MachOObjectFile. 405 static iterator_range<bind_iterator> bindTable(Error &Err, 406 MachOObjectFile *O, 407 ArrayRef<uint8_t> Opcodes, 408 bool is64, 409 MachOBindEntry::Kind); 410 411 /// For use with a SegIndex,SegOffset pair in MachOBindEntry::moveNext() to 412 /// validate a MachOBindEntry. 413 const char *BindEntryCheckSegAndOffset(int32_t SegIndex, uint64_t SegOffset, 414 bool endInvalid) const { 415 return BindRebaseSectionTable->checkSegAndOffset(SegIndex, SegOffset, 416 endInvalid); 417 } 418 /// For use in MachOBindEntry::moveNext() to validate a MachOBindEntry for 419 /// the BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB opcode. 420 const char *BindEntryCheckCountAndSkip(uint32_t Count, uint32_t Skip, 421 uint8_t PointerSize, int32_t SegIndex, 422 uint64_t SegOffset) const { 423 return BindRebaseSectionTable->checkCountAndSkip(Count, Skip, PointerSize, 424 SegIndex, SegOffset); 425 } 426 427 /// For use with a SegIndex,SegOffset pair in MachORebaseEntry::moveNext() to 428 /// validate a MachORebaseEntry. 429 const char *RebaseEntryCheckSegAndOffset(int32_t SegIndex, uint64_t SegOffset, 430 bool endInvalid) const { 431 return BindRebaseSectionTable->checkSegAndOffset(SegIndex, SegOffset, 432 endInvalid); 433 } 434 /// For use in MachORebaseEntry::moveNext() to validate a MachORebaseEntry for 435 /// the REBASE_OPCODE_DO_*_TIMES* opcodes. 436 const char *RebaseEntryCheckCountAndSkip(uint32_t Count, uint32_t Skip, 437 uint8_t PointerSize, int32_t SegIndex, 438 uint64_t SegOffset) const { 439 return BindRebaseSectionTable->checkCountAndSkip(Count, Skip, PointerSize, 440 SegIndex, SegOffset); 441 } 442 443 /// For use with the SegIndex of a checked Mach-O Bind or Rebase entry to 444 /// get the segment name. 445 StringRef BindRebaseSegmentName(int32_t SegIndex) const { 446 return BindRebaseSectionTable->segmentName(SegIndex); 447 } 448 449 /// For use with a SegIndex,SegOffset pair from a checked Mach-O Bind or 450 /// Rebase entry to get the section name. 451 StringRef BindRebaseSectionName(uint32_t SegIndex, uint64_t SegOffset) const { 452 return BindRebaseSectionTable->sectionName(SegIndex, SegOffset); 453 } 454 455 /// For use with a SegIndex,SegOffset pair from a checked Mach-O Bind or 456 /// Rebase entry to get the address. 457 uint64_t BindRebaseAddress(uint32_t SegIndex, uint64_t SegOffset) const { 458 return BindRebaseSectionTable->address(SegIndex, SegOffset); 459 } 460 461 // In a MachO file, sections have a segment name. This is used in the .o 462 // files. They have a single segment, but this field specifies which segment 463 // a section should be put in in the final object. 464 StringRef getSectionFinalSegmentName(DataRefImpl Sec) const; 465 466 // Names are stored as 16 bytes. These returns the raw 16 bytes without 467 // interpreting them as a C string. 468 ArrayRef<char> getSectionRawName(DataRefImpl Sec) const; 469 ArrayRef<char> getSectionRawFinalSegmentName(DataRefImpl Sec) const; 470 471 // MachO specific Info about relocations. 472 bool isRelocationScattered(const MachO::any_relocation_info &RE) const; 473 unsigned getPlainRelocationSymbolNum( 474 const MachO::any_relocation_info &RE) const; 475 bool getPlainRelocationExternal(const MachO::any_relocation_info &RE) const; 476 bool getScatteredRelocationScattered( 477 const MachO::any_relocation_info &RE) const; 478 uint32_t getScatteredRelocationValue( 479 const MachO::any_relocation_info &RE) const; 480 uint32_t getScatteredRelocationType( 481 const MachO::any_relocation_info &RE) const; 482 unsigned getAnyRelocationAddress(const MachO::any_relocation_info &RE) const; 483 unsigned getAnyRelocationPCRel(const MachO::any_relocation_info &RE) const; 484 unsigned getAnyRelocationLength(const MachO::any_relocation_info &RE) const; 485 unsigned getAnyRelocationType(const MachO::any_relocation_info &RE) const; 486 SectionRef getAnyRelocationSection(const MachO::any_relocation_info &RE) const; 487 488 // MachO specific structures. 489 MachO::section getSection(DataRefImpl DRI) const; 490 MachO::section_64 getSection64(DataRefImpl DRI) const; 491 MachO::section getSection(const LoadCommandInfo &L, unsigned Index) const; 492 MachO::section_64 getSection64(const LoadCommandInfo &L,unsigned Index) const; 493 MachO::nlist getSymbolTableEntry(DataRefImpl DRI) const; 494 MachO::nlist_64 getSymbol64TableEntry(DataRefImpl DRI) const; 495 496 MachO::linkedit_data_command 497 getLinkeditDataLoadCommand(const LoadCommandInfo &L) const; 498 MachO::segment_command 499 getSegmentLoadCommand(const LoadCommandInfo &L) const; 500 MachO::segment_command_64 501 getSegment64LoadCommand(const LoadCommandInfo &L) const; 502 MachO::linker_option_command 503 getLinkerOptionLoadCommand(const LoadCommandInfo &L) const; 504 MachO::version_min_command 505 getVersionMinLoadCommand(const LoadCommandInfo &L) const; 506 MachO::note_command 507 getNoteLoadCommand(const LoadCommandInfo &L) const; 508 MachO::build_version_command 509 getBuildVersionLoadCommand(const LoadCommandInfo &L) const; 510 MachO::build_tool_version 511 getBuildToolVersion(unsigned index) const; 512 MachO::dylib_command 513 getDylibIDLoadCommand(const LoadCommandInfo &L) const; 514 MachO::dyld_info_command 515 getDyldInfoLoadCommand(const LoadCommandInfo &L) const; 516 MachO::dylinker_command 517 getDylinkerCommand(const LoadCommandInfo &L) const; 518 MachO::uuid_command 519 getUuidCommand(const LoadCommandInfo &L) const; 520 MachO::rpath_command 521 getRpathCommand(const LoadCommandInfo &L) const; 522 MachO::source_version_command 523 getSourceVersionCommand(const LoadCommandInfo &L) const; 524 MachO::entry_point_command 525 getEntryPointCommand(const LoadCommandInfo &L) const; 526 MachO::encryption_info_command 527 getEncryptionInfoCommand(const LoadCommandInfo &L) const; 528 MachO::encryption_info_command_64 529 getEncryptionInfoCommand64(const LoadCommandInfo &L) const; 530 MachO::sub_framework_command 531 getSubFrameworkCommand(const LoadCommandInfo &L) const; 532 MachO::sub_umbrella_command 533 getSubUmbrellaCommand(const LoadCommandInfo &L) const; 534 MachO::sub_library_command 535 getSubLibraryCommand(const LoadCommandInfo &L) const; 536 MachO::sub_client_command 537 getSubClientCommand(const LoadCommandInfo &L) const; 538 MachO::routines_command 539 getRoutinesCommand(const LoadCommandInfo &L) const; 540 MachO::routines_command_64 541 getRoutinesCommand64(const LoadCommandInfo &L) const; 542 MachO::thread_command 543 getThreadCommand(const LoadCommandInfo &L) const; 544 545 MachO::any_relocation_info getRelocation(DataRefImpl Rel) const; 546 MachO::data_in_code_entry getDice(DataRefImpl Rel) const; 547 const MachO::mach_header &getHeader() const; 548 const MachO::mach_header_64 &getHeader64() const; 549 uint32_t 550 getIndirectSymbolTableEntry(const MachO::dysymtab_command &DLC, 551 unsigned Index) const; 552 MachO::data_in_code_entry getDataInCodeTableEntry(uint32_t DataOffset, 553 unsigned Index) const; 554 MachO::symtab_command getSymtabLoadCommand() const; 555 MachO::dysymtab_command getDysymtabLoadCommand() const; 556 MachO::linkedit_data_command getDataInCodeLoadCommand() const; 557 MachO::linkedit_data_command getLinkOptHintsLoadCommand() const; 558 ArrayRef<uint8_t> getDyldInfoRebaseOpcodes() const; 559 ArrayRef<uint8_t> getDyldInfoBindOpcodes() const; 560 ArrayRef<uint8_t> getDyldInfoWeakBindOpcodes() const; 561 ArrayRef<uint8_t> getDyldInfoLazyBindOpcodes() const; 562 ArrayRef<uint8_t> getDyldInfoExportsTrie() const; 563 ArrayRef<uint8_t> getUuid() const; 564 565 StringRef getStringTableData() const; 566 bool is64Bit() const; 567 void ReadULEB128s(uint64_t Index, SmallVectorImpl<uint64_t> &Out) const; 568 569 static StringRef guessLibraryShortName(StringRef Name, bool &isFramework, 570 StringRef &Suffix); 571 572 static Triple::ArchType getArch(uint32_t CPUType); 573 static Triple getArchTriple(uint32_t CPUType, uint32_t CPUSubType, 574 const char **McpuDefault = nullptr, 575 const char **ArchFlag = nullptr); 576 static bool isValidArch(StringRef ArchFlag); 577 static Triple getHostArch(); 578 579 bool isRelocatableObject() const override; 580 581 StringRef mapDebugSectionName(StringRef Name) const override; 582 583 bool hasPageZeroSegment() const { return HasPageZeroSegment; } 584 585 static bool classof(const Binary *v) { 586 return v->isMachO(); 587 } 588 589 static uint32_t 590 getVersionMinMajor(MachO::version_min_command &C, bool SDK) { 591 uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version; 592 return (VersionOrSDK >> 16) & 0xffff; 593 } 594 595 static uint32_t 596 getVersionMinMinor(MachO::version_min_command &C, bool SDK) { 597 uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version; 598 return (VersionOrSDK >> 8) & 0xff; 599 } 600 601 static uint32_t 602 getVersionMinUpdate(MachO::version_min_command &C, bool SDK) { 603 uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version; 604 return VersionOrSDK & 0xff; 605 } 606 607 static std::string getBuildPlatform(uint32_t platform) { 608 switch (platform) { 609 case MachO::PLATFORM_MACOS: return "macos"; 610 case MachO::PLATFORM_IOS: return "ios"; 611 case MachO::PLATFORM_TVOS: return "tvos"; 612 case MachO::PLATFORM_WATCHOS: return "watchos"; 613 case MachO::PLATFORM_BRIDGEOS: return "bridgeos"; 614 default: 615 std::string ret; 616 raw_string_ostream ss(ret); 617 ss << format_hex(platform, 8, true); 618 return ss.str(); 619 } 620 } 621 622 static std::string getBuildTool(uint32_t tools) { 623 switch (tools) { 624 case MachO::TOOL_CLANG: return "clang"; 625 case MachO::TOOL_SWIFT: return "swift"; 626 case MachO::TOOL_LD: return "ld"; 627 default: 628 std::string ret; 629 raw_string_ostream ss(ret); 630 ss << format_hex(tools, 8, true); 631 return ss.str(); 632 } 633 } 634 635 static std::string getVersionString(uint32_t version) { 636 uint32_t major = (version >> 16) & 0xffff; 637 uint32_t minor = (version >> 8) & 0xff; 638 uint32_t update = version & 0xff; 639 640 SmallString<32> Version; 641 Version = utostr(major) + "." + utostr(minor); 642 if (update != 0) 643 Version += "." + utostr(update); 644 return Version.str(); 645 } 646 647 private: 648 MachOObjectFile(MemoryBufferRef Object, bool IsLittleEndian, bool Is64Bits, 649 Error &Err, uint32_t UniversalCputype = 0, 650 uint32_t UniversalIndex = 0); 651 652 uint64_t getSymbolValueImpl(DataRefImpl Symb) const override; 653 654 union { 655 MachO::mach_header_64 Header64; 656 MachO::mach_header Header; 657 }; 658 using SectionList = SmallVector<const char*, 1>; 659 SectionList Sections; 660 using LibraryList = SmallVector<const char*, 1>; 661 LibraryList Libraries; 662 LoadCommandList LoadCommands; 663 using LibraryShortName = SmallVector<StringRef, 1>; 664 using BuildToolList = SmallVector<const char*, 1>; 665 BuildToolList BuildTools; 666 mutable LibraryShortName LibrariesShortNames; 667 std::unique_ptr<BindRebaseSegInfo> BindRebaseSectionTable; 668 const char *SymtabLoadCmd = nullptr; 669 const char *DysymtabLoadCmd = nullptr; 670 const char *DataInCodeLoadCmd = nullptr; 671 const char *LinkOptHintsLoadCmd = nullptr; 672 const char *DyldInfoLoadCmd = nullptr; 673 const char *UuidLoadCmd = nullptr; 674 bool HasPageZeroSegment = false; 675 }; 676 677 /// DiceRef 678 inline DiceRef::DiceRef(DataRefImpl DiceP, const ObjectFile *Owner) 679 : DicePimpl(DiceP) , OwningObject(Owner) {} 680 681 inline bool DiceRef::operator==(const DiceRef &Other) const { 682 return DicePimpl == Other.DicePimpl; 683 } 684 685 inline bool DiceRef::operator<(const DiceRef &Other) const { 686 return DicePimpl < Other.DicePimpl; 687 } 688 689 inline void DiceRef::moveNext() { 690 const MachO::data_in_code_entry *P = 691 reinterpret_cast<const MachO::data_in_code_entry *>(DicePimpl.p); 692 DicePimpl.p = reinterpret_cast<uintptr_t>(P + 1); 693 } 694 695 // Since a Mach-O data in code reference, a DiceRef, can only be created when 696 // the OwningObject ObjectFile is a MachOObjectFile a static_cast<> is used for 697 // the methods that get the values of the fields of the reference. 698 699 inline std::error_code DiceRef::getOffset(uint32_t &Result) const { 700 const MachOObjectFile *MachOOF = 701 static_cast<const MachOObjectFile *>(OwningObject); 702 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl); 703 Result = Dice.offset; 704 return std::error_code(); 705 } 706 707 inline std::error_code DiceRef::getLength(uint16_t &Result) const { 708 const MachOObjectFile *MachOOF = 709 static_cast<const MachOObjectFile *>(OwningObject); 710 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl); 711 Result = Dice.length; 712 return std::error_code(); 713 } 714 715 inline std::error_code DiceRef::getKind(uint16_t &Result) const { 716 const MachOObjectFile *MachOOF = 717 static_cast<const MachOObjectFile *>(OwningObject); 718 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl); 719 Result = Dice.kind; 720 return std::error_code(); 721 } 722 723 inline DataRefImpl DiceRef::getRawDataRefImpl() const { 724 return DicePimpl; 725 } 726 727 inline const ObjectFile *DiceRef::getObjectFile() const { 728 return OwningObject; 729 } 730 731 } // end namespace object 732 } // end namespace llvm 733 734 #endif // LLVM_OBJECT_MACHO_H 735