Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 1 | //===- MachO.h - MachO object file implementation ---------------*- C++ -*-===// |
| 2 | // |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame] | 3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
| 4 | // See https://llvm.org/LICENSE.txt for license information. |
| 5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 6 | // |
| 7 | //===----------------------------------------------------------------------===// |
| 8 | // |
| 9 | // This file declares the MachOObjectFile class, which implement the ObjectFile |
| 10 | // interface for MachO files. |
| 11 | // |
| 12 | //===----------------------------------------------------------------------===// |
| 13 | |
| 14 | #ifndef LLVM_OBJECT_MACHO_H |
| 15 | #define LLVM_OBJECT_MACHO_H |
| 16 | |
| 17 | #include "llvm/ADT/ArrayRef.h" |
| 18 | #include "llvm/ADT/SmallString.h" |
| 19 | #include "llvm/ADT/SmallVector.h" |
| 20 | #include "llvm/ADT/StringExtras.h" |
| 21 | #include "llvm/ADT/StringRef.h" |
| 22 | #include "llvm/ADT/Triple.h" |
| 23 | #include "llvm/ADT/iterator_range.h" |
| 24 | #include "llvm/BinaryFormat/MachO.h" |
| 25 | #include "llvm/MC/SubtargetFeature.h" |
| 26 | #include "llvm/Object/Binary.h" |
| 27 | #include "llvm/Object/ObjectFile.h" |
| 28 | #include "llvm/Object/SymbolicFile.h" |
| 29 | #include "llvm/Support/Error.h" |
| 30 | #include "llvm/Support/Format.h" |
| 31 | #include "llvm/Support/MemoryBuffer.h" |
| 32 | #include "llvm/Support/raw_ostream.h" |
| 33 | #include <cstdint> |
| 34 | #include <memory> |
| 35 | #include <string> |
| 36 | #include <system_error> |
| 37 | |
| 38 | namespace llvm { |
| 39 | namespace object { |
| 40 | |
| 41 | /// DiceRef - This is a value type class that represents a single |
| 42 | /// data in code entry in the table in a Mach-O object file. |
| 43 | class DiceRef { |
| 44 | DataRefImpl DicePimpl; |
| 45 | const ObjectFile *OwningObject = nullptr; |
| 46 | |
| 47 | public: |
| 48 | DiceRef() = default; |
| 49 | DiceRef(DataRefImpl DiceP, const ObjectFile *Owner); |
| 50 | |
| 51 | bool operator==(const DiceRef &Other) const; |
| 52 | bool operator<(const DiceRef &Other) const; |
| 53 | |
| 54 | void moveNext(); |
| 55 | |
| 56 | std::error_code getOffset(uint32_t &Result) const; |
| 57 | std::error_code getLength(uint16_t &Result) const; |
| 58 | std::error_code getKind(uint16_t &Result) const; |
| 59 | |
| 60 | DataRefImpl getRawDataRefImpl() const; |
| 61 | const ObjectFile *getObjectFile() const; |
| 62 | }; |
| 63 | using dice_iterator = content_iterator<DiceRef>; |
| 64 | |
| 65 | /// ExportEntry encapsulates the current-state-of-the-walk used when doing a |
| 66 | /// non-recursive walk of the trie data structure. This allows you to iterate |
| 67 | /// across all exported symbols using: |
| 68 | /// Error Err; |
| 69 | /// for (const llvm::object::ExportEntry &AnExport : Obj->exports(&Err)) { |
| 70 | /// } |
| 71 | /// if (Err) { report error ... |
| 72 | class ExportEntry { |
| 73 | public: |
| 74 | ExportEntry(Error *Err, const MachOObjectFile *O, ArrayRef<uint8_t> Trie); |
| 75 | |
| 76 | StringRef name() const; |
| 77 | uint64_t flags() const; |
| 78 | uint64_t address() const; |
| 79 | uint64_t other() const; |
| 80 | StringRef otherName() const; |
| 81 | uint32_t nodeOffset() const; |
| 82 | |
| 83 | bool operator==(const ExportEntry &) const; |
| 84 | |
| 85 | void moveNext(); |
| 86 | |
| 87 | private: |
| 88 | friend class MachOObjectFile; |
| 89 | |
| 90 | void moveToFirst(); |
| 91 | void moveToEnd(); |
| 92 | uint64_t readULEB128(const uint8_t *&p, const char **error); |
| 93 | void pushDownUntilBottom(); |
| 94 | void pushNode(uint64_t Offset); |
| 95 | |
| 96 | // Represents a node in the mach-o exports trie. |
| 97 | struct NodeState { |
| 98 | NodeState(const uint8_t *Ptr); |
| 99 | |
| 100 | const uint8_t *Start; |
| 101 | const uint8_t *Current; |
| 102 | uint64_t Flags = 0; |
| 103 | uint64_t Address = 0; |
| 104 | uint64_t Other = 0; |
| 105 | const char *ImportName = nullptr; |
| 106 | unsigned ChildCount = 0; |
| 107 | unsigned NextChildIndex = 0; |
| 108 | unsigned ParentStringLength = 0; |
| 109 | bool IsExportNode = false; |
| 110 | }; |
| 111 | using NodeList = SmallVector<NodeState, 16>; |
| 112 | using node_iterator = NodeList::const_iterator; |
| 113 | |
| 114 | Error *E; |
| 115 | const MachOObjectFile *O; |
| 116 | ArrayRef<uint8_t> Trie; |
| 117 | SmallString<256> CumulativeString; |
| 118 | NodeList Stack; |
| 119 | bool Done = false; |
| 120 | |
| 121 | iterator_range<node_iterator> nodes() const { |
| 122 | return make_range(Stack.begin(), Stack.end()); |
| 123 | } |
| 124 | }; |
| 125 | using export_iterator = content_iterator<ExportEntry>; |
| 126 | |
| 127 | // Segment info so SegIndex/SegOffset pairs in a Mach-O Bind or Rebase entry |
| 128 | // can be checked and translated. Only the SegIndex/SegOffset pairs from |
| 129 | // checked entries are to be used with the segmentName(), sectionName() and |
| 130 | // address() methods below. |
| 131 | class BindRebaseSegInfo { |
| 132 | public: |
| 133 | BindRebaseSegInfo(const MachOObjectFile *Obj); |
| 134 | |
| 135 | // Used to check a Mach-O Bind or Rebase entry for errors when iterating. |
| 136 | const char *checkSegAndOffset(int32_t SegIndex, uint64_t SegOffset, |
| 137 | bool endInvalid); |
| 138 | const char *checkCountAndSkip(uint32_t Count, uint32_t Skip, |
| 139 | uint8_t PointerSize, int32_t SegIndex, |
| 140 | uint64_t SegOffset); |
| 141 | // Used with valid SegIndex/SegOffset values from checked entries. |
| 142 | StringRef segmentName(int32_t SegIndex); |
| 143 | StringRef sectionName(int32_t SegIndex, uint64_t SegOffset); |
| 144 | uint64_t address(uint32_t SegIndex, uint64_t SegOffset); |
| 145 | |
| 146 | private: |
| 147 | struct SectionInfo { |
| 148 | uint64_t Address; |
| 149 | uint64_t Size; |
| 150 | StringRef SectionName; |
| 151 | StringRef SegmentName; |
| 152 | uint64_t OffsetInSegment; |
| 153 | uint64_t SegmentStartAddress; |
| 154 | int32_t SegmentIndex; |
| 155 | }; |
| 156 | const SectionInfo &findSection(int32_t SegIndex, uint64_t SegOffset); |
| 157 | |
| 158 | SmallVector<SectionInfo, 32> Sections; |
| 159 | int32_t MaxSegIndex; |
| 160 | }; |
| 161 | |
| 162 | /// MachORebaseEntry encapsulates the current state in the decompression of |
| 163 | /// rebasing opcodes. This allows you to iterate through the compressed table of |
| 164 | /// rebasing using: |
| 165 | /// Error Err; |
| 166 | /// for (const llvm::object::MachORebaseEntry &Entry : Obj->rebaseTable(&Err)) { |
| 167 | /// } |
| 168 | /// if (Err) { report error ... |
| 169 | class MachORebaseEntry { |
| 170 | public: |
| 171 | MachORebaseEntry(Error *Err, const MachOObjectFile *O, |
| 172 | ArrayRef<uint8_t> opcodes, bool is64Bit); |
| 173 | |
| 174 | int32_t segmentIndex() const; |
| 175 | uint64_t segmentOffset() const; |
| 176 | StringRef typeName() const; |
| 177 | StringRef segmentName() const; |
| 178 | StringRef sectionName() const; |
| 179 | uint64_t address() const; |
| 180 | |
| 181 | bool operator==(const MachORebaseEntry &) const; |
| 182 | |
| 183 | void moveNext(); |
| 184 | |
| 185 | private: |
| 186 | friend class MachOObjectFile; |
| 187 | |
| 188 | void moveToFirst(); |
| 189 | void moveToEnd(); |
| 190 | uint64_t readULEB128(const char **error); |
| 191 | |
| 192 | Error *E; |
| 193 | const MachOObjectFile *O; |
| 194 | ArrayRef<uint8_t> Opcodes; |
| 195 | const uint8_t *Ptr; |
| 196 | uint64_t SegmentOffset = 0; |
| 197 | int32_t SegmentIndex = -1; |
| 198 | uint64_t RemainingLoopCount = 0; |
| 199 | uint64_t AdvanceAmount = 0; |
| 200 | uint8_t RebaseType = 0; |
| 201 | uint8_t PointerSize; |
| 202 | bool Done = false; |
| 203 | }; |
| 204 | using rebase_iterator = content_iterator<MachORebaseEntry>; |
| 205 | |
| 206 | /// MachOBindEntry encapsulates the current state in the decompression of |
| 207 | /// binding opcodes. This allows you to iterate through the compressed table of |
| 208 | /// bindings using: |
| 209 | /// Error Err; |
| 210 | /// for (const llvm::object::MachOBindEntry &Entry : Obj->bindTable(&Err)) { |
| 211 | /// } |
| 212 | /// if (Err) { report error ... |
| 213 | class MachOBindEntry { |
| 214 | public: |
| 215 | enum class Kind { Regular, Lazy, Weak }; |
| 216 | |
| 217 | MachOBindEntry(Error *Err, const MachOObjectFile *O, |
| 218 | ArrayRef<uint8_t> Opcodes, bool is64Bit, MachOBindEntry::Kind); |
| 219 | |
| 220 | int32_t segmentIndex() const; |
| 221 | uint64_t segmentOffset() const; |
| 222 | StringRef typeName() const; |
| 223 | StringRef symbolName() const; |
| 224 | uint32_t flags() const; |
| 225 | int64_t addend() const; |
| 226 | int ordinal() const; |
| 227 | |
| 228 | StringRef segmentName() const; |
| 229 | StringRef sectionName() const; |
| 230 | uint64_t address() const; |
| 231 | |
| 232 | bool operator==(const MachOBindEntry &) const; |
| 233 | |
| 234 | void moveNext(); |
| 235 | |
| 236 | private: |
| 237 | friend class MachOObjectFile; |
| 238 | |
| 239 | void moveToFirst(); |
| 240 | void moveToEnd(); |
| 241 | uint64_t readULEB128(const char **error); |
| 242 | int64_t readSLEB128(const char **error); |
| 243 | |
| 244 | Error *E; |
| 245 | const MachOObjectFile *O; |
| 246 | ArrayRef<uint8_t> Opcodes; |
| 247 | const uint8_t *Ptr; |
| 248 | uint64_t SegmentOffset = 0; |
| 249 | int32_t SegmentIndex = -1; |
| 250 | StringRef SymbolName; |
| 251 | bool LibraryOrdinalSet = false; |
| 252 | int Ordinal = 0; |
| 253 | uint32_t Flags = 0; |
| 254 | int64_t Addend = 0; |
| 255 | uint64_t RemainingLoopCount = 0; |
| 256 | uint64_t AdvanceAmount = 0; |
| 257 | uint8_t BindType = 0; |
| 258 | uint8_t PointerSize; |
| 259 | Kind TableKind; |
| 260 | bool Done = false; |
| 261 | }; |
| 262 | using bind_iterator = content_iterator<MachOBindEntry>; |
| 263 | |
| 264 | class MachOObjectFile : public ObjectFile { |
| 265 | public: |
| 266 | struct LoadCommandInfo { |
| 267 | const char *Ptr; // Where in memory the load command is. |
| 268 | MachO::load_command C; // The command itself. |
| 269 | }; |
| 270 | using LoadCommandList = SmallVector<LoadCommandInfo, 4>; |
| 271 | using load_command_iterator = LoadCommandList::const_iterator; |
| 272 | |
| 273 | static Expected<std::unique_ptr<MachOObjectFile>> |
| 274 | create(MemoryBufferRef Object, bool IsLittleEndian, bool Is64Bits, |
| 275 | uint32_t UniversalCputype = 0, uint32_t UniversalIndex = 0); |
| 276 | |
| 277 | void moveSymbolNext(DataRefImpl &Symb) const override; |
| 278 | |
| 279 | uint64_t getNValue(DataRefImpl Sym) const; |
| 280 | Expected<StringRef> getSymbolName(DataRefImpl Symb) const override; |
| 281 | |
| 282 | // MachO specific. |
| 283 | Error checkSymbolTable() const; |
| 284 | |
| 285 | std::error_code getIndirectName(DataRefImpl Symb, StringRef &Res) const; |
| 286 | unsigned getSectionType(SectionRef Sec) const; |
| 287 | |
| 288 | Expected<uint64_t> getSymbolAddress(DataRefImpl Symb) const override; |
| 289 | uint32_t getSymbolAlignment(DataRefImpl Symb) const override; |
| 290 | uint64_t getCommonSymbolSizeImpl(DataRefImpl Symb) const override; |
| 291 | Expected<SymbolRef::Type> getSymbolType(DataRefImpl Symb) const override; |
| 292 | uint32_t getSymbolFlags(DataRefImpl Symb) const override; |
| 293 | Expected<section_iterator> getSymbolSection(DataRefImpl Symb) const override; |
| 294 | unsigned getSymbolSectionID(SymbolRef Symb) const; |
| 295 | unsigned getSectionID(SectionRef Sec) const; |
| 296 | |
| 297 | void moveSectionNext(DataRefImpl &Sec) const override; |
| 298 | std::error_code getSectionName(DataRefImpl Sec, |
| 299 | StringRef &Res) const override; |
| 300 | uint64_t getSectionAddress(DataRefImpl Sec) const override; |
| 301 | uint64_t getSectionIndex(DataRefImpl Sec) const override; |
| 302 | uint64_t getSectionSize(DataRefImpl Sec) const override; |
| 303 | std::error_code getSectionContents(DataRefImpl Sec, |
| 304 | StringRef &Res) const override; |
| 305 | uint64_t getSectionAlignment(DataRefImpl Sec) const override; |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame] | 306 | Expected<SectionRef> getSection(unsigned SectionIndex) const; |
| 307 | Expected<SectionRef> getSection(StringRef SectionName) const; |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 308 | bool isSectionCompressed(DataRefImpl Sec) const override; |
| 309 | bool isSectionText(DataRefImpl Sec) const override; |
| 310 | bool isSectionData(DataRefImpl Sec) const override; |
| 311 | bool isSectionBSS(DataRefImpl Sec) const override; |
| 312 | bool isSectionVirtual(DataRefImpl Sec) const override; |
| 313 | bool isSectionBitcode(DataRefImpl Sec) const override; |
| 314 | |
| 315 | /// When dsymutil generates the companion file, it strips all unnecessary |
| 316 | /// sections (e.g. everything in the _TEXT segment) by omitting their body |
| 317 | /// and setting the offset in their corresponding load command to zero. |
| 318 | /// |
| 319 | /// While the load command itself is valid, reading the section corresponds |
| 320 | /// to reading the number of bytes specified in the load command, starting |
| 321 | /// from offset 0 (i.e. the Mach-O header at the beginning of the file). |
| 322 | bool isSectionStripped(DataRefImpl Sec) const override; |
| 323 | |
| 324 | relocation_iterator section_rel_begin(DataRefImpl Sec) const override; |
| 325 | relocation_iterator section_rel_end(DataRefImpl Sec) const override; |
| 326 | |
| 327 | relocation_iterator extrel_begin() const; |
| 328 | relocation_iterator extrel_end() const; |
| 329 | iterator_range<relocation_iterator> external_relocations() const { |
| 330 | return make_range(extrel_begin(), extrel_end()); |
| 331 | } |
| 332 | |
| 333 | relocation_iterator locrel_begin() const; |
| 334 | relocation_iterator locrel_end() const; |
Andrew Scull | cdfcccc | 2018-10-05 20:58:37 +0100 | [diff] [blame] | 335 | |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 336 | void moveRelocationNext(DataRefImpl &Rel) const override; |
| 337 | uint64_t getRelocationOffset(DataRefImpl Rel) const override; |
| 338 | symbol_iterator getRelocationSymbol(DataRefImpl Rel) const override; |
| 339 | section_iterator getRelocationSection(DataRefImpl Rel) const; |
| 340 | uint64_t getRelocationType(DataRefImpl Rel) const override; |
| 341 | void getRelocationTypeName(DataRefImpl Rel, |
| 342 | SmallVectorImpl<char> &Result) const override; |
| 343 | uint8_t getRelocationLength(DataRefImpl Rel) const; |
| 344 | |
| 345 | // MachO specific. |
| 346 | std::error_code getLibraryShortNameByIndex(unsigned Index, StringRef &) const; |
| 347 | uint32_t getLibraryCount() const; |
| 348 | |
| 349 | section_iterator getRelocationRelocatedSection(relocation_iterator Rel) const; |
| 350 | |
| 351 | // TODO: Would be useful to have an iterator based version |
| 352 | // of the load command interface too. |
| 353 | |
| 354 | basic_symbol_iterator symbol_begin() const override; |
| 355 | basic_symbol_iterator symbol_end() const override; |
| 356 | |
| 357 | // MachO specific. |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame] | 358 | symbol_iterator getSymbolByIndex(unsigned Index) const; |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 359 | uint64_t getSymbolIndex(DataRefImpl Symb) const; |
| 360 | |
| 361 | section_iterator section_begin() const override; |
| 362 | section_iterator section_end() const override; |
| 363 | |
| 364 | uint8_t getBytesInAddress() const override; |
| 365 | |
| 366 | StringRef getFileFormatName() const override; |
| 367 | Triple::ArchType getArch() const override; |
| 368 | SubtargetFeatures getFeatures() const override { return SubtargetFeatures(); } |
| 369 | Triple getArchTriple(const char **McpuDefault = nullptr) const; |
| 370 | |
| 371 | relocation_iterator section_rel_begin(unsigned Index) const; |
| 372 | relocation_iterator section_rel_end(unsigned Index) const; |
| 373 | |
| 374 | dice_iterator begin_dices() const; |
| 375 | dice_iterator end_dices() const; |
| 376 | |
| 377 | load_command_iterator begin_load_commands() const; |
| 378 | load_command_iterator end_load_commands() const; |
| 379 | iterator_range<load_command_iterator> load_commands() const; |
| 380 | |
| 381 | /// For use iterating over all exported symbols. |
| 382 | iterator_range<export_iterator> exports(Error &Err) const; |
| 383 | |
| 384 | /// For use examining a trie not in a MachOObjectFile. |
| 385 | static iterator_range<export_iterator> exports(Error &Err, |
| 386 | ArrayRef<uint8_t> Trie, |
| 387 | const MachOObjectFile *O = |
| 388 | nullptr); |
| 389 | |
| 390 | /// For use iterating over all rebase table entries. |
| 391 | iterator_range<rebase_iterator> rebaseTable(Error &Err); |
| 392 | |
| 393 | /// For use examining rebase opcodes in a MachOObjectFile. |
| 394 | static iterator_range<rebase_iterator> rebaseTable(Error &Err, |
| 395 | MachOObjectFile *O, |
| 396 | ArrayRef<uint8_t> Opcodes, |
| 397 | bool is64); |
| 398 | |
| 399 | /// For use iterating over all bind table entries. |
| 400 | iterator_range<bind_iterator> bindTable(Error &Err); |
| 401 | |
| 402 | /// For use iterating over all lazy bind table entries. |
| 403 | iterator_range<bind_iterator> lazyBindTable(Error &Err); |
| 404 | |
| 405 | /// For use iterating over all weak bind table entries. |
| 406 | iterator_range<bind_iterator> weakBindTable(Error &Err); |
| 407 | |
| 408 | /// For use examining bind opcodes in a MachOObjectFile. |
| 409 | static iterator_range<bind_iterator> bindTable(Error &Err, |
| 410 | MachOObjectFile *O, |
| 411 | ArrayRef<uint8_t> Opcodes, |
| 412 | bool is64, |
| 413 | MachOBindEntry::Kind); |
| 414 | |
| 415 | /// For use with a SegIndex,SegOffset pair in MachOBindEntry::moveNext() to |
| 416 | /// validate a MachOBindEntry. |
| 417 | const char *BindEntryCheckSegAndOffset(int32_t SegIndex, uint64_t SegOffset, |
| 418 | bool endInvalid) const { |
| 419 | return BindRebaseSectionTable->checkSegAndOffset(SegIndex, SegOffset, |
| 420 | endInvalid); |
| 421 | } |
| 422 | /// For use in MachOBindEntry::moveNext() to validate a MachOBindEntry for |
| 423 | /// the BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB opcode. |
| 424 | const char *BindEntryCheckCountAndSkip(uint32_t Count, uint32_t Skip, |
| 425 | uint8_t PointerSize, int32_t SegIndex, |
| 426 | uint64_t SegOffset) const { |
| 427 | return BindRebaseSectionTable->checkCountAndSkip(Count, Skip, PointerSize, |
| 428 | SegIndex, SegOffset); |
| 429 | } |
| 430 | |
| 431 | /// For use with a SegIndex,SegOffset pair in MachORebaseEntry::moveNext() to |
| 432 | /// validate a MachORebaseEntry. |
| 433 | const char *RebaseEntryCheckSegAndOffset(int32_t SegIndex, uint64_t SegOffset, |
| 434 | bool endInvalid) const { |
| 435 | return BindRebaseSectionTable->checkSegAndOffset(SegIndex, SegOffset, |
| 436 | endInvalid); |
| 437 | } |
| 438 | /// For use in MachORebaseEntry::moveNext() to validate a MachORebaseEntry for |
| 439 | /// the REBASE_OPCODE_DO_*_TIMES* opcodes. |
| 440 | const char *RebaseEntryCheckCountAndSkip(uint32_t Count, uint32_t Skip, |
| 441 | uint8_t PointerSize, int32_t SegIndex, |
| 442 | uint64_t SegOffset) const { |
| 443 | return BindRebaseSectionTable->checkCountAndSkip(Count, Skip, PointerSize, |
| 444 | SegIndex, SegOffset); |
| 445 | } |
| 446 | |
| 447 | /// For use with the SegIndex of a checked Mach-O Bind or Rebase entry to |
| 448 | /// get the segment name. |
| 449 | StringRef BindRebaseSegmentName(int32_t SegIndex) const { |
| 450 | return BindRebaseSectionTable->segmentName(SegIndex); |
| 451 | } |
| 452 | |
| 453 | /// For use with a SegIndex,SegOffset pair from a checked Mach-O Bind or |
| 454 | /// Rebase entry to get the section name. |
| 455 | StringRef BindRebaseSectionName(uint32_t SegIndex, uint64_t SegOffset) const { |
| 456 | return BindRebaseSectionTable->sectionName(SegIndex, SegOffset); |
| 457 | } |
| 458 | |
| 459 | /// For use with a SegIndex,SegOffset pair from a checked Mach-O Bind or |
| 460 | /// Rebase entry to get the address. |
| 461 | uint64_t BindRebaseAddress(uint32_t SegIndex, uint64_t SegOffset) const { |
| 462 | return BindRebaseSectionTable->address(SegIndex, SegOffset); |
| 463 | } |
| 464 | |
| 465 | // In a MachO file, sections have a segment name. This is used in the .o |
| 466 | // files. They have a single segment, but this field specifies which segment |
| 467 | // a section should be put in the final object. |
| 468 | StringRef getSectionFinalSegmentName(DataRefImpl Sec) const; |
| 469 | |
| 470 | // Names are stored as 16 bytes. These returns the raw 16 bytes without |
| 471 | // interpreting them as a C string. |
| 472 | ArrayRef<char> getSectionRawName(DataRefImpl Sec) const; |
| 473 | ArrayRef<char> getSectionRawFinalSegmentName(DataRefImpl Sec) const; |
| 474 | |
| 475 | // MachO specific Info about relocations. |
| 476 | bool isRelocationScattered(const MachO::any_relocation_info &RE) const; |
| 477 | unsigned getPlainRelocationSymbolNum( |
| 478 | const MachO::any_relocation_info &RE) const; |
| 479 | bool getPlainRelocationExternal(const MachO::any_relocation_info &RE) const; |
| 480 | bool getScatteredRelocationScattered( |
| 481 | const MachO::any_relocation_info &RE) const; |
| 482 | uint32_t getScatteredRelocationValue( |
| 483 | const MachO::any_relocation_info &RE) const; |
| 484 | uint32_t getScatteredRelocationType( |
| 485 | const MachO::any_relocation_info &RE) const; |
| 486 | unsigned getAnyRelocationAddress(const MachO::any_relocation_info &RE) const; |
| 487 | unsigned getAnyRelocationPCRel(const MachO::any_relocation_info &RE) const; |
| 488 | unsigned getAnyRelocationLength(const MachO::any_relocation_info &RE) const; |
| 489 | unsigned getAnyRelocationType(const MachO::any_relocation_info &RE) const; |
| 490 | SectionRef getAnyRelocationSection(const MachO::any_relocation_info &RE) const; |
| 491 | |
| 492 | // MachO specific structures. |
| 493 | MachO::section getSection(DataRefImpl DRI) const; |
| 494 | MachO::section_64 getSection64(DataRefImpl DRI) const; |
| 495 | MachO::section getSection(const LoadCommandInfo &L, unsigned Index) const; |
| 496 | MachO::section_64 getSection64(const LoadCommandInfo &L,unsigned Index) const; |
| 497 | MachO::nlist getSymbolTableEntry(DataRefImpl DRI) const; |
| 498 | MachO::nlist_64 getSymbol64TableEntry(DataRefImpl DRI) const; |
| 499 | |
| 500 | MachO::linkedit_data_command |
| 501 | getLinkeditDataLoadCommand(const LoadCommandInfo &L) const; |
| 502 | MachO::segment_command |
| 503 | getSegmentLoadCommand(const LoadCommandInfo &L) const; |
| 504 | MachO::segment_command_64 |
| 505 | getSegment64LoadCommand(const LoadCommandInfo &L) const; |
| 506 | MachO::linker_option_command |
| 507 | getLinkerOptionLoadCommand(const LoadCommandInfo &L) const; |
| 508 | MachO::version_min_command |
| 509 | getVersionMinLoadCommand(const LoadCommandInfo &L) const; |
| 510 | MachO::note_command |
| 511 | getNoteLoadCommand(const LoadCommandInfo &L) const; |
| 512 | MachO::build_version_command |
| 513 | getBuildVersionLoadCommand(const LoadCommandInfo &L) const; |
| 514 | MachO::build_tool_version |
| 515 | getBuildToolVersion(unsigned index) const; |
| 516 | MachO::dylib_command |
| 517 | getDylibIDLoadCommand(const LoadCommandInfo &L) const; |
| 518 | MachO::dyld_info_command |
| 519 | getDyldInfoLoadCommand(const LoadCommandInfo &L) const; |
| 520 | MachO::dylinker_command |
| 521 | getDylinkerCommand(const LoadCommandInfo &L) const; |
| 522 | MachO::uuid_command |
| 523 | getUuidCommand(const LoadCommandInfo &L) const; |
| 524 | MachO::rpath_command |
| 525 | getRpathCommand(const LoadCommandInfo &L) const; |
| 526 | MachO::source_version_command |
| 527 | getSourceVersionCommand(const LoadCommandInfo &L) const; |
| 528 | MachO::entry_point_command |
| 529 | getEntryPointCommand(const LoadCommandInfo &L) const; |
| 530 | MachO::encryption_info_command |
| 531 | getEncryptionInfoCommand(const LoadCommandInfo &L) const; |
| 532 | MachO::encryption_info_command_64 |
| 533 | getEncryptionInfoCommand64(const LoadCommandInfo &L) const; |
| 534 | MachO::sub_framework_command |
| 535 | getSubFrameworkCommand(const LoadCommandInfo &L) const; |
| 536 | MachO::sub_umbrella_command |
| 537 | getSubUmbrellaCommand(const LoadCommandInfo &L) const; |
| 538 | MachO::sub_library_command |
| 539 | getSubLibraryCommand(const LoadCommandInfo &L) const; |
| 540 | MachO::sub_client_command |
| 541 | getSubClientCommand(const LoadCommandInfo &L) const; |
| 542 | MachO::routines_command |
| 543 | getRoutinesCommand(const LoadCommandInfo &L) const; |
| 544 | MachO::routines_command_64 |
| 545 | getRoutinesCommand64(const LoadCommandInfo &L) const; |
| 546 | MachO::thread_command |
| 547 | getThreadCommand(const LoadCommandInfo &L) const; |
| 548 | |
| 549 | MachO::any_relocation_info getRelocation(DataRefImpl Rel) const; |
| 550 | MachO::data_in_code_entry getDice(DataRefImpl Rel) const; |
| 551 | const MachO::mach_header &getHeader() const; |
| 552 | const MachO::mach_header_64 &getHeader64() const; |
| 553 | uint32_t |
| 554 | getIndirectSymbolTableEntry(const MachO::dysymtab_command &DLC, |
| 555 | unsigned Index) const; |
| 556 | MachO::data_in_code_entry getDataInCodeTableEntry(uint32_t DataOffset, |
| 557 | unsigned Index) const; |
| 558 | MachO::symtab_command getSymtabLoadCommand() const; |
| 559 | MachO::dysymtab_command getDysymtabLoadCommand() const; |
| 560 | MachO::linkedit_data_command getDataInCodeLoadCommand() const; |
| 561 | MachO::linkedit_data_command getLinkOptHintsLoadCommand() const; |
| 562 | ArrayRef<uint8_t> getDyldInfoRebaseOpcodes() const; |
| 563 | ArrayRef<uint8_t> getDyldInfoBindOpcodes() const; |
| 564 | ArrayRef<uint8_t> getDyldInfoWeakBindOpcodes() const; |
| 565 | ArrayRef<uint8_t> getDyldInfoLazyBindOpcodes() const; |
| 566 | ArrayRef<uint8_t> getDyldInfoExportsTrie() const; |
| 567 | ArrayRef<uint8_t> getUuid() const; |
| 568 | |
| 569 | StringRef getStringTableData() const; |
| 570 | bool is64Bit() const; |
| 571 | void ReadULEB128s(uint64_t Index, SmallVectorImpl<uint64_t> &Out) const; |
| 572 | |
| 573 | static StringRef guessLibraryShortName(StringRef Name, bool &isFramework, |
| 574 | StringRef &Suffix); |
| 575 | |
| 576 | static Triple::ArchType getArch(uint32_t CPUType); |
| 577 | static Triple getArchTriple(uint32_t CPUType, uint32_t CPUSubType, |
| 578 | const char **McpuDefault = nullptr, |
| 579 | const char **ArchFlag = nullptr); |
| 580 | static bool isValidArch(StringRef ArchFlag); |
| 581 | static Triple getHostArch(); |
| 582 | |
| 583 | bool isRelocatableObject() const override; |
| 584 | |
| 585 | StringRef mapDebugSectionName(StringRef Name) const override; |
| 586 | |
| 587 | bool hasPageZeroSegment() const { return HasPageZeroSegment; } |
| 588 | |
| 589 | static bool classof(const Binary *v) { |
| 590 | return v->isMachO(); |
| 591 | } |
| 592 | |
| 593 | static uint32_t |
| 594 | getVersionMinMajor(MachO::version_min_command &C, bool SDK) { |
| 595 | uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version; |
| 596 | return (VersionOrSDK >> 16) & 0xffff; |
| 597 | } |
| 598 | |
| 599 | static uint32_t |
| 600 | getVersionMinMinor(MachO::version_min_command &C, bool SDK) { |
| 601 | uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version; |
| 602 | return (VersionOrSDK >> 8) & 0xff; |
| 603 | } |
| 604 | |
| 605 | static uint32_t |
| 606 | getVersionMinUpdate(MachO::version_min_command &C, bool SDK) { |
| 607 | uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version; |
| 608 | return VersionOrSDK & 0xff; |
| 609 | } |
| 610 | |
| 611 | static std::string getBuildPlatform(uint32_t platform) { |
| 612 | switch (platform) { |
| 613 | case MachO::PLATFORM_MACOS: return "macos"; |
| 614 | case MachO::PLATFORM_IOS: return "ios"; |
| 615 | case MachO::PLATFORM_TVOS: return "tvos"; |
| 616 | case MachO::PLATFORM_WATCHOS: return "watchos"; |
| 617 | case MachO::PLATFORM_BRIDGEOS: return "bridgeos"; |
Andrew Walbran | 16937d0 | 2019-10-22 13:54:20 +0100 | [diff] [blame] | 618 | case MachO::PLATFORM_IOSSIMULATOR: return "iossimulator"; |
| 619 | case MachO::PLATFORM_TVOSSIMULATOR: return "tvossimulator"; |
| 620 | case MachO::PLATFORM_WATCHOSSIMULATOR: return "watchossimulator"; |
Andrew Scull | 5e1ddfa | 2018-08-14 10:06:54 +0100 | [diff] [blame] | 621 | default: |
| 622 | std::string ret; |
| 623 | raw_string_ostream ss(ret); |
| 624 | ss << format_hex(platform, 8, true); |
| 625 | return ss.str(); |
| 626 | } |
| 627 | } |
| 628 | |
| 629 | static std::string getBuildTool(uint32_t tools) { |
| 630 | switch (tools) { |
| 631 | case MachO::TOOL_CLANG: return "clang"; |
| 632 | case MachO::TOOL_SWIFT: return "swift"; |
| 633 | case MachO::TOOL_LD: return "ld"; |
| 634 | default: |
| 635 | std::string ret; |
| 636 | raw_string_ostream ss(ret); |
| 637 | ss << format_hex(tools, 8, true); |
| 638 | return ss.str(); |
| 639 | } |
| 640 | } |
| 641 | |
| 642 | static std::string getVersionString(uint32_t version) { |
| 643 | uint32_t major = (version >> 16) & 0xffff; |
| 644 | uint32_t minor = (version >> 8) & 0xff; |
| 645 | uint32_t update = version & 0xff; |
| 646 | |
| 647 | SmallString<32> Version; |
| 648 | Version = utostr(major) + "." + utostr(minor); |
| 649 | if (update != 0) |
| 650 | Version += "." + utostr(update); |
| 651 | return Version.str(); |
| 652 | } |
| 653 | |
| 654 | private: |
| 655 | MachOObjectFile(MemoryBufferRef Object, bool IsLittleEndian, bool Is64Bits, |
| 656 | Error &Err, uint32_t UniversalCputype = 0, |
| 657 | uint32_t UniversalIndex = 0); |
| 658 | |
| 659 | uint64_t getSymbolValueImpl(DataRefImpl Symb) const override; |
| 660 | |
| 661 | union { |
| 662 | MachO::mach_header_64 Header64; |
| 663 | MachO::mach_header Header; |
| 664 | }; |
| 665 | using SectionList = SmallVector<const char*, 1>; |
| 666 | SectionList Sections; |
| 667 | using LibraryList = SmallVector<const char*, 1>; |
| 668 | LibraryList Libraries; |
| 669 | LoadCommandList LoadCommands; |
| 670 | using LibraryShortName = SmallVector<StringRef, 1>; |
| 671 | using BuildToolList = SmallVector<const char*, 1>; |
| 672 | BuildToolList BuildTools; |
| 673 | mutable LibraryShortName LibrariesShortNames; |
| 674 | std::unique_ptr<BindRebaseSegInfo> BindRebaseSectionTable; |
| 675 | const char *SymtabLoadCmd = nullptr; |
| 676 | const char *DysymtabLoadCmd = nullptr; |
| 677 | const char *DataInCodeLoadCmd = nullptr; |
| 678 | const char *LinkOptHintsLoadCmd = nullptr; |
| 679 | const char *DyldInfoLoadCmd = nullptr; |
| 680 | const char *UuidLoadCmd = nullptr; |
| 681 | bool HasPageZeroSegment = false; |
| 682 | }; |
| 683 | |
| 684 | /// DiceRef |
| 685 | inline DiceRef::DiceRef(DataRefImpl DiceP, const ObjectFile *Owner) |
| 686 | : DicePimpl(DiceP) , OwningObject(Owner) {} |
| 687 | |
| 688 | inline bool DiceRef::operator==(const DiceRef &Other) const { |
| 689 | return DicePimpl == Other.DicePimpl; |
| 690 | } |
| 691 | |
| 692 | inline bool DiceRef::operator<(const DiceRef &Other) const { |
| 693 | return DicePimpl < Other.DicePimpl; |
| 694 | } |
| 695 | |
| 696 | inline void DiceRef::moveNext() { |
| 697 | const MachO::data_in_code_entry *P = |
| 698 | reinterpret_cast<const MachO::data_in_code_entry *>(DicePimpl.p); |
| 699 | DicePimpl.p = reinterpret_cast<uintptr_t>(P + 1); |
| 700 | } |
| 701 | |
| 702 | // Since a Mach-O data in code reference, a DiceRef, can only be created when |
| 703 | // the OwningObject ObjectFile is a MachOObjectFile a static_cast<> is used for |
| 704 | // the methods that get the values of the fields of the reference. |
| 705 | |
| 706 | inline std::error_code DiceRef::getOffset(uint32_t &Result) const { |
| 707 | const MachOObjectFile *MachOOF = |
| 708 | static_cast<const MachOObjectFile *>(OwningObject); |
| 709 | MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl); |
| 710 | Result = Dice.offset; |
| 711 | return std::error_code(); |
| 712 | } |
| 713 | |
| 714 | inline std::error_code DiceRef::getLength(uint16_t &Result) const { |
| 715 | const MachOObjectFile *MachOOF = |
| 716 | static_cast<const MachOObjectFile *>(OwningObject); |
| 717 | MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl); |
| 718 | Result = Dice.length; |
| 719 | return std::error_code(); |
| 720 | } |
| 721 | |
| 722 | inline std::error_code DiceRef::getKind(uint16_t &Result) const { |
| 723 | const MachOObjectFile *MachOOF = |
| 724 | static_cast<const MachOObjectFile *>(OwningObject); |
| 725 | MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl); |
| 726 | Result = Dice.kind; |
| 727 | return std::error_code(); |
| 728 | } |
| 729 | |
| 730 | inline DataRefImpl DiceRef::getRawDataRefImpl() const { |
| 731 | return DicePimpl; |
| 732 | } |
| 733 | |
| 734 | inline const ObjectFile *DiceRef::getObjectFile() const { |
| 735 | return OwningObject; |
| 736 | } |
| 737 | |
| 738 | } // end namespace object |
| 739 | } // end namespace llvm |
| 740 | |
| 741 | #endif // LLVM_OBJECT_MACHO_H |