MachO.h revision d111c7844ec26448764ced627e153f406d730c5f
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  Expected<ArrayRef<uint8_t>>
301  getSectionContents(DataRefImpl Sec) const override;
302  uint64_t getSectionAlignment(DataRefImpl Sec) const override;
303  Expected<SectionRef> getSection(unsigned SectionIndex) const;
304  Expected<SectionRef> getSection(StringRef SectionName) const;
305  bool isSectionCompressed(DataRefImpl Sec) const override;
306  bool isSectionText(DataRefImpl Sec) const override;
307  bool isSectionData(DataRefImpl Sec) const override;
308  bool isSectionBSS(DataRefImpl Sec) const override;
309  bool isSectionVirtual(DataRefImpl Sec) const override;
310  bool isSectionBitcode(DataRefImpl Sec) const override;
311
312  /// When dsymutil generates the companion file, it strips all unnecessary
313  /// sections (e.g. everything in the _TEXT segment) by omitting their body
314  /// and setting the offset in their corresponding load command to zero.
315  ///
316  /// While the load command itself is valid, reading the section corresponds
317  /// to reading the number of bytes specified in the load command, starting
318  /// from offset 0 (i.e. the Mach-O header at the beginning of the file).
319  bool isSectionStripped(DataRefImpl Sec) const override;
320
321  relocation_iterator section_rel_begin(DataRefImpl Sec) const override;
322  relocation_iterator section_rel_end(DataRefImpl Sec) const override;
323
324  relocation_iterator extrel_begin() const;
325  relocation_iterator extrel_end() const;
326  iterator_range<relocation_iterator> external_relocations() const {
327    return make_range(extrel_begin(), extrel_end());
328  }
329
330  relocation_iterator locrel_begin() const;
331  relocation_iterator locrel_end() const;
332
333  void moveRelocationNext(DataRefImpl &Rel) const override;
334  uint64_t getRelocationOffset(DataRefImpl Rel) const override;
335  symbol_iterator getRelocationSymbol(DataRefImpl Rel) const override;
336  section_iterator getRelocationSection(DataRefImpl Rel) const;
337  uint64_t getRelocationType(DataRefImpl Rel) const override;
338  void getRelocationTypeName(DataRefImpl Rel,
339                             SmallVectorImpl<char> &Result) const override;
340  uint8_t getRelocationLength(DataRefImpl Rel) const;
341
342  // MachO specific.
343  std::error_code getLibraryShortNameByIndex(unsigned Index, StringRef &) const;
344  uint32_t getLibraryCount() const;
345
346  section_iterator getRelocationRelocatedSection(relocation_iterator Rel) const;
347
348  // TODO: Would be useful to have an iterator based version
349  // of the load command interface too.
350
351  basic_symbol_iterator symbol_begin() const override;
352  basic_symbol_iterator symbol_end() const override;
353
354  // MachO specific.
355  symbol_iterator getSymbolByIndex(unsigned Index) const;
356  uint64_t getSymbolIndex(DataRefImpl Symb) const;
357
358  section_iterator section_begin() const override;
359  section_iterator section_end() const override;
360
361  uint8_t getBytesInAddress() const override;
362
363  StringRef getFileFormatName() const override;
364  Triple::ArchType getArch() const override;
365  SubtargetFeatures getFeatures() const override { return SubtargetFeatures(); }
366  Triple getArchTriple(const char **McpuDefault = nullptr) const;
367
368  relocation_iterator section_rel_begin(unsigned Index) const;
369  relocation_iterator section_rel_end(unsigned Index) const;
370
371  dice_iterator begin_dices() const;
372  dice_iterator end_dices() const;
373
374  load_command_iterator begin_load_commands() const;
375  load_command_iterator end_load_commands() const;
376  iterator_range<load_command_iterator> load_commands() const;
377
378  /// For use iterating over all exported symbols.
379  iterator_range<export_iterator> exports(Error &Err) const;
380
381  /// For use examining a trie not in a MachOObjectFile.
382  static iterator_range<export_iterator> exports(Error &Err,
383                                                 ArrayRef<uint8_t> Trie,
384                                                 const MachOObjectFile *O =
385                                                                      nullptr);
386
387  /// For use iterating over all rebase table entries.
388  iterator_range<rebase_iterator> rebaseTable(Error &Err);
389
390  /// For use examining rebase opcodes in a MachOObjectFile.
391  static iterator_range<rebase_iterator> rebaseTable(Error &Err,
392                                                     MachOObjectFile *O,
393                                                     ArrayRef<uint8_t> Opcodes,
394                                                     bool is64);
395
396  /// For use iterating over all bind table entries.
397  iterator_range<bind_iterator> bindTable(Error &Err);
398
399  /// For use iterating over all lazy bind table entries.
400  iterator_range<bind_iterator> lazyBindTable(Error &Err);
401
402  /// For use iterating over all weak bind table entries.
403  iterator_range<bind_iterator> weakBindTable(Error &Err);
404
405  /// For use examining bind opcodes in a MachOObjectFile.
406  static iterator_range<bind_iterator> bindTable(Error &Err,
407                                                 MachOObjectFile *O,
408                                                 ArrayRef<uint8_t> Opcodes,
409                                                 bool is64,
410                                                 MachOBindEntry::Kind);
411
412  // Given a SegIndex, SegOffset, and PointerSize, verify a valid section exists
413  // that fully contains a pointer at that location. Multiple fixups in a bind
414  // (such as with the BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB opcode) can
415  // be tested via the Count and Skip parameters.
416  //
417  // This is used by MachOBindEntry::moveNext() to validate a MachOBindEntry.
418  const char *BindEntryCheckSegAndOffsets(int32_t SegIndex, uint64_t SegOffset,
419                                         uint8_t PointerSize, uint32_t Count=1,
420                                          uint32_t Skip=0) const {
421    return BindRebaseSectionTable->checkSegAndOffsets(SegIndex, SegOffset,
422                                                     PointerSize, Count, Skip);
423  }
424
425  // Given a SegIndex, SegOffset, and PointerSize, verify a valid section exists
426  // that fully contains a pointer at that location. Multiple fixups in a rebase
427  // (such as with the REBASE_OPCODE_DO_*_TIMES* opcodes) can be tested via the
428  // Count and Skip parameters.
429  //
430  // This is used by MachORebaseEntry::moveNext() to validate a MachORebaseEntry
431  const char *RebaseEntryCheckSegAndOffsets(int32_t SegIndex,
432                                            uint64_t SegOffset,
433                                            uint8_t PointerSize,
434                                            uint32_t Count=1,
435                                            uint32_t Skip=0) const {
436    return BindRebaseSectionTable->checkSegAndOffsets(SegIndex, SegOffset,
437                                                      PointerSize, Count, Skip);
438  }
439
440  /// For use with the SegIndex of a checked Mach-O Bind or Rebase entry to
441  /// get the segment name.
442  StringRef BindRebaseSegmentName(int32_t SegIndex) const {
443    return BindRebaseSectionTable->segmentName(SegIndex);
444  }
445
446  /// For use with a SegIndex,SegOffset pair from a checked Mach-O Bind or
447  /// Rebase entry to get the section name.
448  StringRef BindRebaseSectionName(uint32_t SegIndex, uint64_t SegOffset) const {
449    return BindRebaseSectionTable->sectionName(SegIndex, SegOffset);
450  }
451
452  /// For use with a SegIndex,SegOffset pair from a checked Mach-O Bind or
453  /// Rebase entry to get the address.
454  uint64_t BindRebaseAddress(uint32_t SegIndex, uint64_t SegOffset) const {
455    return BindRebaseSectionTable->address(SegIndex, SegOffset);
456  }
457
458  // In a MachO file, sections have a segment name. This is used in the .o
459  // files. They have a single segment, but this field specifies which segment
460  // a section should be put in the final object.
461  StringRef getSectionFinalSegmentName(DataRefImpl Sec) const;
462
463  // Names are stored as 16 bytes. These returns the raw 16 bytes without
464  // interpreting them as a C string.
465  ArrayRef<char> getSectionRawName(DataRefImpl Sec) const;
466  ArrayRef<char> getSectionRawFinalSegmentName(DataRefImpl Sec) const;
467
468  // MachO specific Info about relocations.
469  bool isRelocationScattered(const MachO::any_relocation_info &RE) const;
470  unsigned getPlainRelocationSymbolNum(
471                                    const MachO::any_relocation_info &RE) const;
472  bool getPlainRelocationExternal(const MachO::any_relocation_info &RE) const;
473  bool getScatteredRelocationScattered(
474                                    const MachO::any_relocation_info &RE) const;
475  uint32_t getScatteredRelocationValue(
476                                    const MachO::any_relocation_info &RE) const;
477  uint32_t getScatteredRelocationType(
478                                    const MachO::any_relocation_info &RE) const;
479  unsigned getAnyRelocationAddress(const MachO::any_relocation_info &RE) const;
480  unsigned getAnyRelocationPCRel(const MachO::any_relocation_info &RE) const;
481  unsigned getAnyRelocationLength(const MachO::any_relocation_info &RE) const;
482  unsigned getAnyRelocationType(const MachO::any_relocation_info &RE) const;
483  SectionRef getAnyRelocationSection(const MachO::any_relocation_info &RE) const;
484
485  // MachO specific structures.
486  MachO::section getSection(DataRefImpl DRI) const;
487  MachO::section_64 getSection64(DataRefImpl DRI) const;
488  MachO::section getSection(const LoadCommandInfo &L, unsigned Index) const;
489  MachO::section_64 getSection64(const LoadCommandInfo &L,unsigned Index) const;
490  MachO::nlist getSymbolTableEntry(DataRefImpl DRI) const;
491  MachO::nlist_64 getSymbol64TableEntry(DataRefImpl DRI) const;
492
493  MachO::linkedit_data_command
494  getLinkeditDataLoadCommand(const LoadCommandInfo &L) const;
495  MachO::segment_command
496  getSegmentLoadCommand(const LoadCommandInfo &L) const;
497  MachO::segment_command_64
498  getSegment64LoadCommand(const LoadCommandInfo &L) const;
499  MachO::linker_option_command
500  getLinkerOptionLoadCommand(const LoadCommandInfo &L) const;
501  MachO::version_min_command
502  getVersionMinLoadCommand(const LoadCommandInfo &L) const;
503  MachO::note_command
504  getNoteLoadCommand(const LoadCommandInfo &L) const;
505  MachO::build_version_command
506  getBuildVersionLoadCommand(const LoadCommandInfo &L) const;
507  MachO::build_tool_version
508  getBuildToolVersion(unsigned index) const;
509  MachO::dylib_command
510  getDylibIDLoadCommand(const LoadCommandInfo &L) const;
511  MachO::dyld_info_command
512  getDyldInfoLoadCommand(const LoadCommandInfo &L) const;
513  MachO::dylinker_command
514  getDylinkerCommand(const LoadCommandInfo &L) const;
515  MachO::uuid_command
516  getUuidCommand(const LoadCommandInfo &L) const;
517  MachO::rpath_command
518  getRpathCommand(const LoadCommandInfo &L) const;
519  MachO::source_version_command
520  getSourceVersionCommand(const LoadCommandInfo &L) const;
521  MachO::entry_point_command
522  getEntryPointCommand(const LoadCommandInfo &L) const;
523  MachO::encryption_info_command
524  getEncryptionInfoCommand(const LoadCommandInfo &L) const;
525  MachO::encryption_info_command_64
526  getEncryptionInfoCommand64(const LoadCommandInfo &L) const;
527  MachO::sub_framework_command
528  getSubFrameworkCommand(const LoadCommandInfo &L) const;
529  MachO::sub_umbrella_command
530  getSubUmbrellaCommand(const LoadCommandInfo &L) const;
531  MachO::sub_library_command
532  getSubLibraryCommand(const LoadCommandInfo &L) const;
533  MachO::sub_client_command
534  getSubClientCommand(const LoadCommandInfo &L) const;
535  MachO::routines_command
536  getRoutinesCommand(const LoadCommandInfo &L) const;
537  MachO::routines_command_64
538  getRoutinesCommand64(const LoadCommandInfo &L) const;
539  MachO::thread_command
540  getThreadCommand(const LoadCommandInfo &L) const;
541
542  MachO::any_relocation_info getRelocation(DataRefImpl Rel) const;
543  MachO::data_in_code_entry getDice(DataRefImpl Rel) const;
544  const MachO::mach_header &getHeader() const;
545  const MachO::mach_header_64 &getHeader64() const;
546  uint32_t
547  getIndirectSymbolTableEntry(const MachO::dysymtab_command &DLC,
548                              unsigned Index) const;
549  MachO::data_in_code_entry getDataInCodeTableEntry(uint32_t DataOffset,
550                                                    unsigned Index) const;
551  MachO::symtab_command getSymtabLoadCommand() const;
552  MachO::dysymtab_command getDysymtabLoadCommand() const;
553  MachO::linkedit_data_command getDataInCodeLoadCommand() const;
554  MachO::linkedit_data_command getLinkOptHintsLoadCommand() const;
555  ArrayRef<uint8_t> getDyldInfoRebaseOpcodes() const;
556  ArrayRef<uint8_t> getDyldInfoBindOpcodes() const;
557  ArrayRef<uint8_t> getDyldInfoWeakBindOpcodes() const;
558  ArrayRef<uint8_t> getDyldInfoLazyBindOpcodes() const;
559  ArrayRef<uint8_t> getDyldInfoExportsTrie() const;
560  ArrayRef<uint8_t> getUuid() const;
561
562  StringRef getStringTableData() const;
563  bool is64Bit() const;
564  void ReadULEB128s(uint64_t Index, SmallVectorImpl<uint64_t> &Out) const;
565
566  static StringRef guessLibraryShortName(StringRef Name, bool &isFramework,
567                                         StringRef &Suffix);
568
569  static Triple::ArchType getArch(uint32_t CPUType);
570  static Triple getArchTriple(uint32_t CPUType, uint32_t CPUSubType,
571                              const char **McpuDefault = nullptr,
572                              const char **ArchFlag = nullptr);
573  static bool isValidArch(StringRef ArchFlag);
574  static ArrayRef<StringRef> getValidArchs();
575  static Triple getHostArch();
576
577  bool isRelocatableObject() const override;
578
579  StringRef mapDebugSectionName(StringRef Name) const override;
580
581  bool hasPageZeroSegment() const { return HasPageZeroSegment; }
582
583  static bool classof(const Binary *v) {
584    return v->isMachO();
585  }
586
587  static uint32_t
588  getVersionMinMajor(MachO::version_min_command &C, bool SDK) {
589    uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
590    return (VersionOrSDK >> 16) & 0xffff;
591  }
592
593  static uint32_t
594  getVersionMinMinor(MachO::version_min_command &C, bool SDK) {
595    uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
596    return (VersionOrSDK >> 8) & 0xff;
597  }
598
599  static uint32_t
600  getVersionMinUpdate(MachO::version_min_command &C, bool SDK) {
601    uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
602    return VersionOrSDK & 0xff;
603  }
604
605  static std::string getBuildPlatform(uint32_t platform) {
606    switch (platform) {
607    case MachO::PLATFORM_MACOS: return "macos";
608    case MachO::PLATFORM_IOS: return "ios";
609    case MachO::PLATFORM_TVOS: return "tvos";
610    case MachO::PLATFORM_WATCHOS: return "watchos";
611    case MachO::PLATFORM_BRIDGEOS: return "bridgeos";
612    case MachO::PLATFORM_MACCATALYST: return "macCatalyst";
613    case MachO::PLATFORM_IOSSIMULATOR: return "iossimulator";
614    case MachO::PLATFORM_TVOSSIMULATOR: return "tvossimulator";
615    case MachO::PLATFORM_WATCHOSSIMULATOR: return "watchossimulator";
616    default:
617      std::string ret;
618      raw_string_ostream ss(ret);
619      ss << format_hex(platform, 8, true);
620      return ss.str();
621    }
622  }
623
624  static std::string getBuildTool(uint32_t tools) {
625    switch (tools) {
626    case MachO::TOOL_CLANG: return "clang";
627    case MachO::TOOL_SWIFT: return "swift";
628    case MachO::TOOL_LD: return "ld";
629    default:
630      std::string ret;
631      raw_string_ostream ss(ret);
632      ss << format_hex(tools, 8, true);
633      return ss.str();
634    }
635  }
636
637  static std::string getVersionString(uint32_t version) {
638    uint32_t major = (version >> 16) & 0xffff;
639    uint32_t minor = (version >> 8) & 0xff;
640    uint32_t update = version & 0xff;
641
642    SmallString<32> Version;
643    Version = utostr(major) + "." + utostr(minor);
644    if (update != 0)
645      Version += "." + utostr(update);
646    return Version.str();
647  }
648
649private:
650  MachOObjectFile(MemoryBufferRef Object, bool IsLittleEndian, bool Is64Bits,
651                  Error &Err, uint32_t UniversalCputype = 0,
652                  uint32_t UniversalIndex = 0);
653
654  uint64_t getSymbolValueImpl(DataRefImpl Symb) const override;
655
656  union {
657    MachO::mach_header_64 Header64;
658    MachO::mach_header Header;
659  };
660  using SectionList = SmallVector<const char*, 1>;
661  SectionList Sections;
662  using LibraryList = SmallVector<const char*, 1>;
663  LibraryList Libraries;
664  LoadCommandList LoadCommands;
665  using LibraryShortName = SmallVector<StringRef, 1>;
666  using BuildToolList = SmallVector<const char*, 1>;
667  BuildToolList BuildTools;
668  mutable LibraryShortName LibrariesShortNames;
669  std::unique_ptr<BindRebaseSegInfo> BindRebaseSectionTable;
670  const char *SymtabLoadCmd = nullptr;
671  const char *DysymtabLoadCmd = nullptr;
672  const char *DataInCodeLoadCmd = nullptr;
673  const char *LinkOptHintsLoadCmd = nullptr;
674  const char *DyldInfoLoadCmd = nullptr;
675  const char *UuidLoadCmd = nullptr;
676  bool HasPageZeroSegment = false;
677};
678
679/// DiceRef
680inline DiceRef::DiceRef(DataRefImpl DiceP, const ObjectFile *Owner)
681  : DicePimpl(DiceP) , OwningObject(Owner) {}
682
683inline bool DiceRef::operator==(const DiceRef &Other) const {
684  return DicePimpl == Other.DicePimpl;
685}
686
687inline bool DiceRef::operator<(const DiceRef &Other) const {
688  return DicePimpl < Other.DicePimpl;
689}
690
691inline void DiceRef::moveNext() {
692  const MachO::data_in_code_entry *P =
693    reinterpret_cast<const MachO::data_in_code_entry *>(DicePimpl.p);
694  DicePimpl.p = reinterpret_cast<uintptr_t>(P + 1);
695}
696
697// Since a Mach-O data in code reference, a DiceRef, can only be created when
698// the OwningObject ObjectFile is a MachOObjectFile a static_cast<> is used for
699// the methods that get the values of the fields of the reference.
700
701inline std::error_code DiceRef::getOffset(uint32_t &Result) const {
702  const MachOObjectFile *MachOOF =
703    static_cast<const MachOObjectFile *>(OwningObject);
704  MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
705  Result = Dice.offset;
706  return std::error_code();
707}
708
709inline std::error_code DiceRef::getLength(uint16_t &Result) const {
710  const MachOObjectFile *MachOOF =
711    static_cast<const MachOObjectFile *>(OwningObject);
712  MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
713  Result = Dice.length;
714  return std::error_code();
715}
716
717inline std::error_code DiceRef::getKind(uint16_t &Result) const {
718  const MachOObjectFile *MachOOF =
719    static_cast<const MachOObjectFile *>(OwningObject);
720  MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
721  Result = Dice.kind;
722  return std::error_code();
723}
724
725inline DataRefImpl DiceRef::getRawDataRefImpl() const {
726  return DicePimpl;
727}
728
729inline const ObjectFile *DiceRef::getObjectFile() const {
730  return OwningObject;
731}
732
733} // end namespace object
734} // end namespace llvm
735
736#endif // LLVM_OBJECT_MACHO_H
737