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/SmallVector.h" 20 #include "llvm/ADT/Triple.h" 21 #include "llvm/Object/ObjectFile.h" 22 #include "llvm/Support/MachO.h" 23 24 namespace llvm { 25 namespace object { 26 27 /// DiceRef - This is a value type class that represents a single 28 /// data in code entry in the table in a Mach-O object file. 29 class DiceRef { 30 DataRefImpl DicePimpl; 31 const ObjectFile *OwningObject; 32 33 public: 34 DiceRef() : OwningObject(nullptr) { } 35 36 DiceRef(DataRefImpl DiceP, const ObjectFile *Owner); 37 38 bool operator==(const DiceRef &Other) const; 39 bool operator<(const DiceRef &Other) const; 40 41 void moveNext(); 42 43 std::error_code getOffset(uint32_t &Result) const; 44 std::error_code getLength(uint16_t &Result) const; 45 std::error_code getKind(uint16_t &Result) const; 46 47 DataRefImpl getRawDataRefImpl() const; 48 const ObjectFile *getObjectFile() const; 49 }; 50 typedef content_iterator<DiceRef> dice_iterator; 51 52 /// ExportEntry encapsulates the current-state-of-the-walk used when doing a 53 /// non-recursive walk of the trie data structure. This allows you to iterate 54 /// across all exported symbols using: 55 /// for (const llvm::object::ExportEntry &AnExport : Obj->exports()) { 56 /// } 57 class ExportEntry { 58 public: 59 ExportEntry(ArrayRef<uint8_t> Trie); 60 61 StringRef name() const; 62 uint64_t flags() const; 63 uint64_t address() const; 64 uint64_t other() const; 65 StringRef otherName() const; 66 uint32_t nodeOffset() const; 67 68 bool operator==(const ExportEntry &) const; 69 70 void moveNext(); 71 72 private: 73 friend class MachOObjectFile; 74 void moveToFirst(); 75 void moveToEnd(); 76 uint64_t readULEB128(const uint8_t *&p); 77 void pushDownUntilBottom(); 78 void pushNode(uint64_t Offset); 79 80 // Represents a node in the mach-o exports trie. 81 struct NodeState { 82 NodeState(const uint8_t *Ptr); 83 const uint8_t *Start; 84 const uint8_t *Current; 85 uint64_t Flags; 86 uint64_t Address; 87 uint64_t Other; 88 const char *ImportName; 89 unsigned ChildCount; 90 unsigned NextChildIndex; 91 unsigned ParentStringLength; 92 bool IsExportNode; 93 }; 94 95 ArrayRef<uint8_t> Trie; 96 SmallString<256> CumulativeString; 97 SmallVector<NodeState, 16> Stack; 98 bool Malformed; 99 bool Done; 100 }; 101 typedef content_iterator<ExportEntry> export_iterator; 102 103 /// MachORebaseEntry encapsulates the current state in the decompression of 104 /// rebasing opcodes. This allows you to iterate through the compressed table of 105 /// rebasing using: 106 /// for (const llvm::object::MachORebaseEntry &Entry : Obj->rebaseTable()) { 107 /// } 108 class MachORebaseEntry { 109 public: 110 MachORebaseEntry(ArrayRef<uint8_t> opcodes, bool is64Bit); 111 112 uint32_t segmentIndex() const; 113 uint64_t segmentOffset() const; 114 StringRef typeName() const; 115 116 bool operator==(const MachORebaseEntry &) const; 117 118 void moveNext(); 119 120 private: 121 friend class MachOObjectFile; 122 void moveToFirst(); 123 void moveToEnd(); 124 uint64_t readULEB128(); 125 126 ArrayRef<uint8_t> Opcodes; 127 const uint8_t *Ptr; 128 uint64_t SegmentOffset; 129 uint32_t SegmentIndex; 130 uint64_t RemainingLoopCount; 131 uint64_t AdvanceAmount; 132 uint8_t RebaseType; 133 uint8_t PointerSize; 134 bool Malformed; 135 bool Done; 136 }; 137 typedef content_iterator<MachORebaseEntry> rebase_iterator; 138 139 /// MachOBindEntry encapsulates the current state in the decompression of 140 /// binding opcodes. This allows you to iterate through the compressed table of 141 /// bindings using: 142 /// for (const llvm::object::MachOBindEntry &Entry : Obj->bindTable()) { 143 /// } 144 class MachOBindEntry { 145 public: 146 enum class Kind { Regular, Lazy, Weak }; 147 148 MachOBindEntry(ArrayRef<uint8_t> Opcodes, bool is64Bit, MachOBindEntry::Kind); 149 150 uint32_t segmentIndex() const; 151 uint64_t segmentOffset() const; 152 StringRef typeName() const; 153 StringRef symbolName() const; 154 uint32_t flags() const; 155 int64_t addend() const; 156 int ordinal() const; 157 158 bool operator==(const MachOBindEntry &) const; 159 160 void moveNext(); 161 162 private: 163 friend class MachOObjectFile; 164 void moveToFirst(); 165 void moveToEnd(); 166 uint64_t readULEB128(); 167 int64_t readSLEB128(); 168 169 ArrayRef<uint8_t> Opcodes; 170 const uint8_t *Ptr; 171 uint64_t SegmentOffset; 172 uint32_t SegmentIndex; 173 StringRef SymbolName; 174 int Ordinal; 175 uint32_t Flags; 176 int64_t Addend; 177 uint64_t RemainingLoopCount; 178 uint64_t AdvanceAmount; 179 uint8_t BindType; 180 uint8_t PointerSize; 181 Kind TableKind; 182 bool Malformed; 183 bool Done; 184 }; 185 typedef content_iterator<MachOBindEntry> bind_iterator; 186 187 class MachOObjectFile : public ObjectFile { 188 public: 189 struct LoadCommandInfo { 190 const char *Ptr; // Where in memory the load command is. 191 MachO::load_command C; // The command itself. 192 }; 193 194 MachOObjectFile(MemoryBufferRef Object, bool IsLittleEndian, bool Is64Bits, 195 std::error_code &EC); 196 197 void moveSymbolNext(DataRefImpl &Symb) const override; 198 std::error_code getSymbolName(DataRefImpl Symb, 199 StringRef &Res) const override; 200 201 // MachO specific. 202 std::error_code getIndirectName(DataRefImpl Symb, StringRef &Res) const; 203 unsigned getSectionType(SectionRef Sec) const; 204 205 std::error_code getSymbolAddress(DataRefImpl Symb, 206 uint64_t &Res) const override; 207 std::error_code getSymbolAlignment(DataRefImpl Symb, 208 uint32_t &Res) const override; 209 std::error_code getSymbolSize(DataRefImpl Symb, uint64_t &Res) const override; 210 std::error_code getSymbolType(DataRefImpl Symb, 211 SymbolRef::Type &Res) const override; 212 uint32_t getSymbolFlags(DataRefImpl Symb) const override; 213 std::error_code getSymbolSection(DataRefImpl Symb, 214 section_iterator &Res) const override; 215 216 void moveSectionNext(DataRefImpl &Sec) const override; 217 std::error_code getSectionName(DataRefImpl Sec, 218 StringRef &Res) const override; 219 uint64_t getSectionAddress(DataRefImpl Sec) const override; 220 uint64_t getSectionSize(DataRefImpl Sec) const override; 221 std::error_code getSectionContents(DataRefImpl Sec, 222 StringRef &Res) const override; 223 uint64_t getSectionAlignment(DataRefImpl Sec) const override; 224 bool isSectionText(DataRefImpl Sec) const override; 225 bool isSectionData(DataRefImpl Sec) const override; 226 bool isSectionBSS(DataRefImpl Sec) const override; 227 bool isSectionVirtual(DataRefImpl Sec) const override; 228 bool sectionContainsSymbol(DataRefImpl Sec, DataRefImpl Symb) const override; 229 relocation_iterator section_rel_begin(DataRefImpl Sec) const override; 230 relocation_iterator section_rel_end(DataRefImpl Sec) const override; 231 232 void moveRelocationNext(DataRefImpl &Rel) const override; 233 std::error_code getRelocationAddress(DataRefImpl Rel, 234 uint64_t &Res) const override; 235 std::error_code getRelocationOffset(DataRefImpl Rel, 236 uint64_t &Res) const override; 237 symbol_iterator getRelocationSymbol(DataRefImpl Rel) const override; 238 std::error_code getRelocationType(DataRefImpl Rel, 239 uint64_t &Res) const override; 240 std::error_code 241 getRelocationTypeName(DataRefImpl Rel, 242 SmallVectorImpl<char> &Result) const override; 243 std::error_code 244 getRelocationValueString(DataRefImpl Rel, 245 SmallVectorImpl<char> &Result) const override; 246 std::error_code getRelocationHidden(DataRefImpl Rel, 247 bool &Result) const override; 248 249 // MachO specific. 250 std::error_code getLibraryShortNameByIndex(unsigned Index, StringRef &) const; 251 252 // TODO: Would be useful to have an iterator based version 253 // of the load command interface too. 254 255 basic_symbol_iterator symbol_begin_impl() const override; 256 basic_symbol_iterator symbol_end_impl() const override; 257 258 // MachO specific. 259 basic_symbol_iterator getSymbolByIndex(unsigned Index) const; 260 261 section_iterator section_begin() const override; 262 section_iterator section_end() const override; 263 264 uint8_t getBytesInAddress() const override; 265 266 StringRef getFileFormatName() const override; 267 unsigned getArch() const override; 268 Triple getArch(const char **McpuDefault, Triple *ThumbTriple) const; 269 270 relocation_iterator section_rel_begin(unsigned Index) const; 271 relocation_iterator section_rel_end(unsigned Index) const; 272 273 dice_iterator begin_dices() const; 274 dice_iterator end_dices() const; 275 276 /// For use iterating over all exported symbols. 277 iterator_range<export_iterator> exports() const; 278 279 /// For use examining a trie not in a MachOObjectFile. 280 static iterator_range<export_iterator> exports(ArrayRef<uint8_t> Trie); 281 282 /// For use iterating over all rebase table entries. 283 iterator_range<rebase_iterator> rebaseTable() const; 284 285 /// For use examining rebase opcodes not in a MachOObjectFile. 286 static iterator_range<rebase_iterator> rebaseTable(ArrayRef<uint8_t> Opcodes, 287 bool is64); 288 289 /// For use iterating over all bind table entries. 290 iterator_range<bind_iterator> bindTable() const; 291 292 /// For use iterating over all lazy bind table entries. 293 iterator_range<bind_iterator> lazyBindTable() const; 294 295 /// For use iterating over all lazy bind table entries. 296 iterator_range<bind_iterator> weakBindTable() const; 297 298 /// For use examining bind opcodes not in a MachOObjectFile. 299 static iterator_range<bind_iterator> bindTable(ArrayRef<uint8_t> Opcodes, 300 bool is64, 301 MachOBindEntry::Kind); 302 303 304 // In a MachO file, sections have a segment name. This is used in the .o 305 // files. They have a single segment, but this field specifies which segment 306 // a section should be put in in the final object. 307 StringRef getSectionFinalSegmentName(DataRefImpl Sec) const; 308 309 // Names are stored as 16 bytes. These returns the raw 16 bytes without 310 // interpreting them as a C string. 311 ArrayRef<char> getSectionRawName(DataRefImpl Sec) const; 312 ArrayRef<char> getSectionRawFinalSegmentName(DataRefImpl Sec) const; 313 314 // MachO specific Info about relocations. 315 bool isRelocationScattered(const MachO::any_relocation_info &RE) const; 316 unsigned getPlainRelocationSymbolNum( 317 const MachO::any_relocation_info &RE) const; 318 bool getPlainRelocationExternal(const MachO::any_relocation_info &RE) const; 319 bool getScatteredRelocationScattered( 320 const MachO::any_relocation_info &RE) const; 321 uint32_t getScatteredRelocationValue( 322 const MachO::any_relocation_info &RE) const; 323 uint32_t getScatteredRelocationType( 324 const MachO::any_relocation_info &RE) const; 325 unsigned getAnyRelocationAddress(const MachO::any_relocation_info &RE) const; 326 unsigned getAnyRelocationPCRel(const MachO::any_relocation_info &RE) const; 327 unsigned getAnyRelocationLength(const MachO::any_relocation_info &RE) const; 328 unsigned getAnyRelocationType(const MachO::any_relocation_info &RE) const; 329 SectionRef getRelocationSection(const MachO::any_relocation_info &RE) const; 330 331 // Walk load commands. 332 LoadCommandInfo getFirstLoadCommandInfo() const; 333 LoadCommandInfo getNextLoadCommandInfo(const LoadCommandInfo &L) const; 334 335 // MachO specific structures. 336 MachO::section getSection(DataRefImpl DRI) const; 337 MachO::section_64 getSection64(DataRefImpl DRI) const; 338 MachO::section getSection(const LoadCommandInfo &L, unsigned Index) const; 339 MachO::section_64 getSection64(const LoadCommandInfo &L,unsigned Index) const; 340 MachO::nlist getSymbolTableEntry(DataRefImpl DRI) const; 341 MachO::nlist_64 getSymbol64TableEntry(DataRefImpl DRI) const; 342 343 MachO::linkedit_data_command 344 getLinkeditDataLoadCommand(const LoadCommandInfo &L) const; 345 MachO::segment_command 346 getSegmentLoadCommand(const LoadCommandInfo &L) const; 347 MachO::segment_command_64 348 getSegment64LoadCommand(const LoadCommandInfo &L) const; 349 MachO::linker_option_command 350 getLinkerOptionLoadCommand(const LoadCommandInfo &L) const; 351 MachO::version_min_command 352 getVersionMinLoadCommand(const LoadCommandInfo &L) const; 353 MachO::dylib_command 354 getDylibIDLoadCommand(const LoadCommandInfo &L) const; 355 MachO::dyld_info_command 356 getDyldInfoLoadCommand(const LoadCommandInfo &L) const; 357 MachO::dylinker_command 358 getDylinkerCommand(const LoadCommandInfo &L) const; 359 MachO::uuid_command 360 getUuidCommand(const LoadCommandInfo &L) const; 361 MachO::rpath_command 362 getRpathCommand(const LoadCommandInfo &L) const; 363 MachO::source_version_command 364 getSourceVersionCommand(const LoadCommandInfo &L) const; 365 MachO::entry_point_command 366 getEntryPointCommand(const LoadCommandInfo &L) const; 367 MachO::encryption_info_command 368 getEncryptionInfoCommand(const LoadCommandInfo &L) const; 369 MachO::encryption_info_command_64 370 getEncryptionInfoCommand64(const LoadCommandInfo &L) const; 371 MachO::sub_framework_command 372 getSubFrameworkCommand(const LoadCommandInfo &L) const; 373 MachO::sub_umbrella_command 374 getSubUmbrellaCommand(const LoadCommandInfo &L) const; 375 MachO::sub_library_command 376 getSubLibraryCommand(const LoadCommandInfo &L) const; 377 MachO::sub_client_command 378 getSubClientCommand(const LoadCommandInfo &L) const; 379 MachO::routines_command 380 getRoutinesCommand(const LoadCommandInfo &L) const; 381 MachO::routines_command_64 382 getRoutinesCommand64(const LoadCommandInfo &L) const; 383 MachO::thread_command 384 getThreadCommand(const LoadCommandInfo &L) const; 385 386 MachO::any_relocation_info getRelocation(DataRefImpl Rel) const; 387 MachO::data_in_code_entry getDice(DataRefImpl Rel) const; 388 MachO::mach_header getHeader() const; 389 MachO::mach_header_64 getHeader64() const; 390 uint32_t 391 getIndirectSymbolTableEntry(const MachO::dysymtab_command &DLC, 392 unsigned Index) const; 393 MachO::data_in_code_entry getDataInCodeTableEntry(uint32_t DataOffset, 394 unsigned Index) const; 395 MachO::symtab_command getSymtabLoadCommand() const; 396 MachO::dysymtab_command getDysymtabLoadCommand() const; 397 MachO::linkedit_data_command getDataInCodeLoadCommand() const; 398 MachO::linkedit_data_command getLinkOptHintsLoadCommand() const; 399 ArrayRef<uint8_t> getDyldInfoRebaseOpcodes() const; 400 ArrayRef<uint8_t> getDyldInfoBindOpcodes() const; 401 ArrayRef<uint8_t> getDyldInfoWeakBindOpcodes() const; 402 ArrayRef<uint8_t> getDyldInfoLazyBindOpcodes() const; 403 ArrayRef<uint8_t> getDyldInfoExportsTrie() const; 404 ArrayRef<uint8_t> getUuid() const; 405 406 StringRef getStringTableData() const; 407 bool is64Bit() const; 408 void ReadULEB128s(uint64_t Index, SmallVectorImpl<uint64_t> &Out) const; 409 410 static StringRef guessLibraryShortName(StringRef Name, bool &isFramework, 411 StringRef &Suffix); 412 413 static Triple::ArchType getArch(uint32_t CPUType); 414 static Triple getArch(uint32_t CPUType, uint32_t CPUSubType, 415 const char **McpuDefault = nullptr); 416 static Triple getThumbArch(uint32_t CPUType, uint32_t CPUSubType, 417 const char **McpuDefault = nullptr); 418 static Triple getArch(uint32_t CPUType, uint32_t CPUSubType, 419 const char **McpuDefault, Triple *ThumbTriple); 420 static bool isValidArch(StringRef ArchFlag); 421 static Triple getHostArch(); 422 423 bool isRelocatableObject() const override; 424 425 bool hasPageZeroSegment() const { return HasPageZeroSegment; } 426 427 static bool classof(const Binary *v) { 428 return v->isMachO(); 429 } 430 431 private: 432 typedef SmallVector<const char*, 1> SectionList; 433 SectionList Sections; 434 typedef SmallVector<const char*, 1> LibraryList; 435 LibraryList Libraries; 436 typedef SmallVector<StringRef, 1> LibraryShortName; 437 mutable LibraryShortName LibrariesShortNames; 438 const char *SymtabLoadCmd; 439 const char *DysymtabLoadCmd; 440 const char *DataInCodeLoadCmd; 441 const char *LinkOptHintsLoadCmd; 442 const char *DyldInfoLoadCmd; 443 const char *UuidLoadCmd; 444 bool HasPageZeroSegment; 445 }; 446 447 /// DiceRef 448 inline DiceRef::DiceRef(DataRefImpl DiceP, const ObjectFile *Owner) 449 : DicePimpl(DiceP) , OwningObject(Owner) {} 450 451 inline bool DiceRef::operator==(const DiceRef &Other) const { 452 return DicePimpl == Other.DicePimpl; 453 } 454 455 inline bool DiceRef::operator<(const DiceRef &Other) const { 456 return DicePimpl < Other.DicePimpl; 457 } 458 459 inline void DiceRef::moveNext() { 460 const MachO::data_in_code_entry *P = 461 reinterpret_cast<const MachO::data_in_code_entry *>(DicePimpl.p); 462 DicePimpl.p = reinterpret_cast<uintptr_t>(P + 1); 463 } 464 465 // Since a Mach-O data in code reference, a DiceRef, can only be created when 466 // the OwningObject ObjectFile is a MachOObjectFile a static_cast<> is used for 467 // the methods that get the values of the fields of the reference. 468 469 inline std::error_code DiceRef::getOffset(uint32_t &Result) const { 470 const MachOObjectFile *MachOOF = 471 static_cast<const MachOObjectFile *>(OwningObject); 472 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl); 473 Result = Dice.offset; 474 return object_error::success; 475 } 476 477 inline std::error_code DiceRef::getLength(uint16_t &Result) const { 478 const MachOObjectFile *MachOOF = 479 static_cast<const MachOObjectFile *>(OwningObject); 480 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl); 481 Result = Dice.length; 482 return object_error::success; 483 } 484 485 inline std::error_code DiceRef::getKind(uint16_t &Result) const { 486 const MachOObjectFile *MachOOF = 487 static_cast<const MachOObjectFile *>(OwningObject); 488 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl); 489 Result = Dice.kind; 490 return object_error::success; 491 } 492 493 inline DataRefImpl DiceRef::getRawDataRefImpl() const { 494 return DicePimpl; 495 } 496 497 inline const ObjectFile *DiceRef::getObjectFile() const { 498 return OwningObject; 499 } 500 501 } 502 } 503 504 #endif 505 506