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