1 //===- MachO.h - MachO object file implementation ---------------*- C++ -*-===//
2 //
3 //                     The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 // This file declares the MachOObjectFile class, which implement the ObjectFile
11 // interface for MachO files.
12 //
13 //===----------------------------------------------------------------------===//
14 
15 #ifndef LLVM_OBJECT_MACHO_H
16 #define LLVM_OBJECT_MACHO_H
17 
18 #include "llvm/ADT/ArrayRef.h"
19 #include "llvm/ADT/SmallString.h"
20 #include "llvm/ADT/SmallVector.h"
21 #include "llvm/ADT/StringExtras.h"
22 #include "llvm/ADT/StringRef.h"
23 #include "llvm/ADT/Triple.h"
24 #include "llvm/ADT/iterator_range.h"
25 #include "llvm/BinaryFormat/MachO.h"
26 #include "llvm/MC/SubtargetFeature.h"
27 #include "llvm/Object/Binary.h"
28 #include "llvm/Object/ObjectFile.h"
29 #include "llvm/Object/SymbolicFile.h"
30 #include "llvm/Support/Error.h"
31 #include "llvm/Support/Format.h"
32 #include "llvm/Support/MemoryBuffer.h"
33 #include "llvm/Support/raw_ostream.h"
34 #include <cstdint>
35 #include <memory>
36 #include <string>
37 #include <system_error>
38 
39 namespace llvm {
40 namespace object {
41 
42 /// DiceRef - This is a value type class that represents a single
43 /// data in code entry in the table in a Mach-O object file.
44 class DiceRef {
45   DataRefImpl DicePimpl;
46   const ObjectFile *OwningObject = nullptr;
47 
48 public:
49   DiceRef() = default;
50   DiceRef(DataRefImpl DiceP, const ObjectFile *Owner);
51 
52   bool operator==(const DiceRef &Other) const;
53   bool operator<(const DiceRef &Other) const;
54 
55   void moveNext();
56 
57   std::error_code getOffset(uint32_t &Result) const;
58   std::error_code getLength(uint16_t &Result) const;
59   std::error_code getKind(uint16_t &Result) const;
60 
61   DataRefImpl getRawDataRefImpl() const;
62   const ObjectFile *getObjectFile() const;
63 };
64 using dice_iterator = content_iterator<DiceRef>;
65 
66 /// ExportEntry encapsulates the current-state-of-the-walk used when doing a
67 /// non-recursive walk of the trie data structure.  This allows you to iterate
68 /// across all exported symbols using:
69 ///      Error Err;
70 ///      for (const llvm::object::ExportEntry &AnExport : Obj->exports(&Err)) {
71 ///      }
72 ///      if (Err) { report error ...
73 class ExportEntry {
74 public:
75   ExportEntry(Error *Err, const MachOObjectFile *O, ArrayRef<uint8_t> Trie);
76 
77   StringRef name() const;
78   uint64_t flags() const;
79   uint64_t address() const;
80   uint64_t other() const;
81   StringRef otherName() const;
82   uint32_t nodeOffset() const;
83 
84   bool operator==(const ExportEntry &) const;
85 
86   void moveNext();
87 
88 private:
89   friend class MachOObjectFile;
90 
91   void moveToFirst();
92   void moveToEnd();
93   uint64_t readULEB128(const uint8_t *&p, const char **error);
94   void pushDownUntilBottom();
95   void pushNode(uint64_t Offset);
96 
97   // Represents a node in the mach-o exports trie.
98   struct NodeState {
99     NodeState(const uint8_t *Ptr);
100 
101     const uint8_t *Start;
102     const uint8_t *Current;
103     uint64_t Flags = 0;
104     uint64_t Address = 0;
105     uint64_t Other = 0;
106     const char *ImportName = nullptr;
107     unsigned ChildCount = 0;
108     unsigned NextChildIndex = 0;
109     unsigned ParentStringLength = 0;
110     bool IsExportNode = false;
111   };
112   using NodeList = SmallVector<NodeState, 16>;
113   using node_iterator = NodeList::const_iterator;
114 
115   Error *E;
116   const MachOObjectFile *O;
117   ArrayRef<uint8_t> Trie;
118   SmallString<256> CumulativeString;
119   NodeList Stack;
120   bool Done = false;
121 
nodes()122   iterator_range<node_iterator> nodes() const {
123     return make_range(Stack.begin(), Stack.end());
124   }
125 };
126 using export_iterator = content_iterator<ExportEntry>;
127 
128 // Segment info so SegIndex/SegOffset pairs in a Mach-O Bind or Rebase entry
129 // can be checked and translated.  Only the SegIndex/SegOffset pairs from
130 // checked entries are to be used with the segmentName(), sectionName() and
131 // address() methods below.
132 class BindRebaseSegInfo {
133 public:
134   BindRebaseSegInfo(const MachOObjectFile *Obj);
135 
136   // Used to check a Mach-O Bind or Rebase entry for errors when iterating.
137   const char *checkSegAndOffset(int32_t SegIndex, uint64_t SegOffset,
138                                 bool endInvalid);
139   const char *checkCountAndSkip(uint32_t Count, uint32_t Skip,
140                                 uint8_t PointerSize, int32_t SegIndex,
141                                 uint64_t SegOffset);
142   // Used with valid SegIndex/SegOffset values from checked entries.
143   StringRef segmentName(int32_t SegIndex);
144   StringRef sectionName(int32_t SegIndex, uint64_t SegOffset);
145   uint64_t address(uint32_t SegIndex, uint64_t SegOffset);
146 
147 private:
148   struct SectionInfo {
149     uint64_t Address;
150     uint64_t Size;
151     StringRef SectionName;
152     StringRef SegmentName;
153     uint64_t OffsetInSegment;
154     uint64_t SegmentStartAddress;
155     int32_t SegmentIndex;
156   };
157   const SectionInfo &findSection(int32_t SegIndex, uint64_t SegOffset);
158 
159   SmallVector<SectionInfo, 32> Sections;
160   int32_t MaxSegIndex;
161 };
162 
163 /// MachORebaseEntry encapsulates the current state in the decompression of
164 /// rebasing opcodes. This allows you to iterate through the compressed table of
165 /// rebasing using:
166 ///    Error Err;
167 ///    for (const llvm::object::MachORebaseEntry &Entry : Obj->rebaseTable(&Err)) {
168 ///    }
169 ///    if (Err) { report error ...
170 class MachORebaseEntry {
171 public:
172   MachORebaseEntry(Error *Err, const MachOObjectFile *O,
173                    ArrayRef<uint8_t> opcodes, bool is64Bit);
174 
175   int32_t segmentIndex() const;
176   uint64_t segmentOffset() const;
177   StringRef typeName() const;
178   StringRef segmentName() const;
179   StringRef sectionName() const;
180   uint64_t address() const;
181 
182   bool operator==(const MachORebaseEntry &) const;
183 
184   void moveNext();
185 
186 private:
187   friend class MachOObjectFile;
188 
189   void moveToFirst();
190   void moveToEnd();
191   uint64_t readULEB128(const char **error);
192 
193   Error *E;
194   const MachOObjectFile *O;
195   ArrayRef<uint8_t> Opcodes;
196   const uint8_t *Ptr;
197   uint64_t SegmentOffset = 0;
198   int32_t SegmentIndex = -1;
199   uint64_t RemainingLoopCount = 0;
200   uint64_t AdvanceAmount = 0;
201   uint8_t  RebaseType = 0;
202   uint8_t  PointerSize;
203   bool     Done = false;
204 };
205 using rebase_iterator = content_iterator<MachORebaseEntry>;
206 
207 /// MachOBindEntry encapsulates the current state in the decompression of
208 /// binding opcodes. This allows you to iterate through the compressed table of
209 /// bindings using:
210 ///    Error Err;
211 ///    for (const llvm::object::MachOBindEntry &Entry : Obj->bindTable(&Err)) {
212 ///    }
213 ///    if (Err) { report error ...
214 class MachOBindEntry {
215 public:
216   enum class Kind { Regular, Lazy, Weak };
217 
218   MachOBindEntry(Error *Err, const MachOObjectFile *O,
219                  ArrayRef<uint8_t> Opcodes, bool is64Bit, MachOBindEntry::Kind);
220 
221   int32_t segmentIndex() const;
222   uint64_t segmentOffset() const;
223   StringRef typeName() const;
224   StringRef symbolName() const;
225   uint32_t flags() const;
226   int64_t addend() const;
227   int ordinal() const;
228 
229   StringRef segmentName() const;
230   StringRef sectionName() const;
231   uint64_t address() const;
232 
233   bool operator==(const MachOBindEntry &) const;
234 
235   void moveNext();
236 
237 private:
238   friend class MachOObjectFile;
239 
240   void moveToFirst();
241   void moveToEnd();
242   uint64_t readULEB128(const char **error);
243   int64_t readSLEB128(const char **error);
244 
245   Error *E;
246   const MachOObjectFile *O;
247   ArrayRef<uint8_t> Opcodes;
248   const uint8_t *Ptr;
249   uint64_t SegmentOffset = 0;
250   int32_t  SegmentIndex = -1;
251   StringRef SymbolName;
252   bool     LibraryOrdinalSet = false;
253   int      Ordinal = 0;
254   uint32_t Flags = 0;
255   int64_t  Addend = 0;
256   uint64_t RemainingLoopCount = 0;
257   uint64_t AdvanceAmount = 0;
258   uint8_t  BindType = 0;
259   uint8_t  PointerSize;
260   Kind     TableKind;
261   bool     Done = false;
262 };
263 using bind_iterator = content_iterator<MachOBindEntry>;
264 
265 class MachOObjectFile : public ObjectFile {
266 public:
267   struct LoadCommandInfo {
268     const char *Ptr;      // Where in memory the load command is.
269     MachO::load_command C; // The command itself.
270   };
271   using LoadCommandList = SmallVector<LoadCommandInfo, 4>;
272   using load_command_iterator = LoadCommandList::const_iterator;
273 
274   static Expected<std::unique_ptr<MachOObjectFile>>
275   create(MemoryBufferRef Object, bool IsLittleEndian, bool Is64Bits,
276          uint32_t UniversalCputype = 0, uint32_t UniversalIndex = 0);
277 
278   void moveSymbolNext(DataRefImpl &Symb) const override;
279 
280   uint64_t getNValue(DataRefImpl Sym) const;
281   Expected<StringRef> getSymbolName(DataRefImpl Symb) const override;
282 
283   // MachO specific.
284   Error checkSymbolTable() const;
285 
286   std::error_code getIndirectName(DataRefImpl Symb, StringRef &Res) const;
287   unsigned getSectionType(SectionRef Sec) const;
288 
289   Expected<uint64_t> getSymbolAddress(DataRefImpl Symb) const override;
290   uint32_t getSymbolAlignment(DataRefImpl Symb) const override;
291   uint64_t getCommonSymbolSizeImpl(DataRefImpl Symb) const override;
292   Expected<SymbolRef::Type> getSymbolType(DataRefImpl Symb) const override;
293   uint32_t getSymbolFlags(DataRefImpl Symb) const override;
294   Expected<section_iterator> getSymbolSection(DataRefImpl Symb) const override;
295   unsigned getSymbolSectionID(SymbolRef Symb) const;
296   unsigned getSectionID(SectionRef Sec) const;
297 
298   void moveSectionNext(DataRefImpl &Sec) const override;
299   std::error_code getSectionName(DataRefImpl Sec,
300                                  StringRef &Res) const override;
301   uint64_t getSectionAddress(DataRefImpl Sec) const override;
302   uint64_t getSectionIndex(DataRefImpl Sec) const override;
303   uint64_t getSectionSize(DataRefImpl Sec) const override;
304   std::error_code getSectionContents(DataRefImpl Sec,
305                                      StringRef &Res) const override;
306   uint64_t getSectionAlignment(DataRefImpl Sec) const override;
307   Expected<SectionRef> getSection(unsigned SectionIndex) const;
308   Expected<SectionRef> getSection(StringRef SectionName) const;
309   bool isSectionCompressed(DataRefImpl Sec) const override;
310   bool isSectionText(DataRefImpl Sec) const override;
311   bool isSectionData(DataRefImpl Sec) const override;
312   bool isSectionBSS(DataRefImpl Sec) const override;
313   bool isSectionVirtual(DataRefImpl Sec) const override;
314   bool isSectionBitcode(DataRefImpl Sec) const override;
315 
316   /// When dsymutil generates the companion file, it strips all unnecessary
317   /// sections (e.g. everything in the _TEXT segment) by omitting their body
318   /// and setting the offset in their corresponding load command to zero.
319   ///
320   /// While the load command itself is valid, reading the section corresponds
321   /// to reading the number of bytes specified in the load command, starting
322   /// from offset 0 (i.e. the Mach-O header at the beginning of the file).
323   bool isSectionStripped(DataRefImpl Sec) const override;
324 
325   relocation_iterator section_rel_begin(DataRefImpl Sec) const override;
326   relocation_iterator section_rel_end(DataRefImpl Sec) const override;
327 
328   relocation_iterator extrel_begin() const;
329   relocation_iterator extrel_end() const;
external_relocations()330   iterator_range<relocation_iterator> external_relocations() const {
331     return make_range(extrel_begin(), extrel_end());
332   }
333 
334   relocation_iterator locrel_begin() const;
335   relocation_iterator locrel_end() const;
336 
337   void moveRelocationNext(DataRefImpl &Rel) const override;
338   uint64_t getRelocationOffset(DataRefImpl Rel) const override;
339   symbol_iterator getRelocationSymbol(DataRefImpl Rel) const override;
340   section_iterator getRelocationSection(DataRefImpl Rel) const;
341   uint64_t getRelocationType(DataRefImpl Rel) const override;
342   void getRelocationTypeName(DataRefImpl Rel,
343                              SmallVectorImpl<char> &Result) const override;
344   uint8_t getRelocationLength(DataRefImpl Rel) const;
345 
346   // MachO specific.
347   std::error_code getLibraryShortNameByIndex(unsigned Index, StringRef &) const;
348   uint32_t getLibraryCount() const;
349 
350   section_iterator getRelocationRelocatedSection(relocation_iterator Rel) const;
351 
352   // TODO: Would be useful to have an iterator based version
353   // of the load command interface too.
354 
355   basic_symbol_iterator symbol_begin() const override;
356   basic_symbol_iterator symbol_end() const override;
357 
358   // MachO specific.
359   symbol_iterator getSymbolByIndex(unsigned Index) const;
360   uint64_t getSymbolIndex(DataRefImpl Symb) const;
361 
362   section_iterator section_begin() const override;
363   section_iterator section_end() const override;
364 
365   uint8_t getBytesInAddress() const override;
366 
367   StringRef getFileFormatName() const override;
368   Triple::ArchType getArch() const override;
getFeatures()369   SubtargetFeatures getFeatures() const override { return SubtargetFeatures(); }
370   Triple getArchTriple(const char **McpuDefault = nullptr) const;
371 
372   relocation_iterator section_rel_begin(unsigned Index) const;
373   relocation_iterator section_rel_end(unsigned Index) const;
374 
375   dice_iterator begin_dices() const;
376   dice_iterator end_dices() const;
377 
378   load_command_iterator begin_load_commands() const;
379   load_command_iterator end_load_commands() const;
380   iterator_range<load_command_iterator> load_commands() const;
381 
382   /// For use iterating over all exported symbols.
383   iterator_range<export_iterator> exports(Error &Err) const;
384 
385   /// For use examining a trie not in a MachOObjectFile.
386   static iterator_range<export_iterator> exports(Error &Err,
387                                                  ArrayRef<uint8_t> Trie,
388                                                  const MachOObjectFile *O =
389                                                                       nullptr);
390 
391   /// For use iterating over all rebase table entries.
392   iterator_range<rebase_iterator> rebaseTable(Error &Err);
393 
394   /// For use examining rebase opcodes in a MachOObjectFile.
395   static iterator_range<rebase_iterator> rebaseTable(Error &Err,
396                                                      MachOObjectFile *O,
397                                                      ArrayRef<uint8_t> Opcodes,
398                                                      bool is64);
399 
400   /// For use iterating over all bind table entries.
401   iterator_range<bind_iterator> bindTable(Error &Err);
402 
403   /// For use iterating over all lazy bind table entries.
404   iterator_range<bind_iterator> lazyBindTable(Error &Err);
405 
406   /// For use iterating over all weak bind table entries.
407   iterator_range<bind_iterator> weakBindTable(Error &Err);
408 
409   /// For use examining bind opcodes in a MachOObjectFile.
410   static iterator_range<bind_iterator> bindTable(Error &Err,
411                                                  MachOObjectFile *O,
412                                                  ArrayRef<uint8_t> Opcodes,
413                                                  bool is64,
414                                                  MachOBindEntry::Kind);
415 
416   /// For use with a SegIndex,SegOffset pair in MachOBindEntry::moveNext() to
417   /// validate a MachOBindEntry.
BindEntryCheckSegAndOffset(int32_t SegIndex,uint64_t SegOffset,bool endInvalid)418   const char *BindEntryCheckSegAndOffset(int32_t SegIndex, uint64_t SegOffset,
419                                          bool endInvalid) const {
420     return BindRebaseSectionTable->checkSegAndOffset(SegIndex, SegOffset,
421                                                      endInvalid);
422   }
423   /// For use in MachOBindEntry::moveNext() to validate a MachOBindEntry for
424   /// the BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB opcode.
BindEntryCheckCountAndSkip(uint32_t Count,uint32_t Skip,uint8_t PointerSize,int32_t SegIndex,uint64_t SegOffset)425   const char *BindEntryCheckCountAndSkip(uint32_t Count, uint32_t Skip,
426                                          uint8_t PointerSize, int32_t SegIndex,
427                                          uint64_t SegOffset) const {
428     return BindRebaseSectionTable->checkCountAndSkip(Count, Skip, PointerSize,
429                                                      SegIndex, SegOffset);
430   }
431 
432   /// For use with a SegIndex,SegOffset pair in MachORebaseEntry::moveNext() to
433   /// validate a MachORebaseEntry.
RebaseEntryCheckSegAndOffset(int32_t SegIndex,uint64_t SegOffset,bool endInvalid)434   const char *RebaseEntryCheckSegAndOffset(int32_t SegIndex, uint64_t SegOffset,
435                                            bool endInvalid) const {
436     return BindRebaseSectionTable->checkSegAndOffset(SegIndex, SegOffset,
437                                                      endInvalid);
438   }
439   /// For use in MachORebaseEntry::moveNext() to validate a MachORebaseEntry for
440   /// the REBASE_OPCODE_DO_*_TIMES* opcodes.
RebaseEntryCheckCountAndSkip(uint32_t Count,uint32_t Skip,uint8_t PointerSize,int32_t SegIndex,uint64_t SegOffset)441   const char *RebaseEntryCheckCountAndSkip(uint32_t Count, uint32_t Skip,
442                                          uint8_t PointerSize, int32_t SegIndex,
443                                          uint64_t SegOffset) const {
444     return BindRebaseSectionTable->checkCountAndSkip(Count, Skip, PointerSize,
445                                                      SegIndex, SegOffset);
446   }
447 
448   /// For use with the SegIndex of a checked Mach-O Bind or Rebase entry to
449   /// get the segment name.
BindRebaseSegmentName(int32_t SegIndex)450   StringRef BindRebaseSegmentName(int32_t SegIndex) const {
451     return BindRebaseSectionTable->segmentName(SegIndex);
452   }
453 
454   /// For use with a SegIndex,SegOffset pair from a checked Mach-O Bind or
455   /// Rebase entry to get the section name.
BindRebaseSectionName(uint32_t SegIndex,uint64_t SegOffset)456   StringRef BindRebaseSectionName(uint32_t SegIndex, uint64_t SegOffset) const {
457     return BindRebaseSectionTable->sectionName(SegIndex, SegOffset);
458   }
459 
460   /// For use with a SegIndex,SegOffset pair from a checked Mach-O Bind or
461   /// Rebase entry to get the address.
BindRebaseAddress(uint32_t SegIndex,uint64_t SegOffset)462   uint64_t BindRebaseAddress(uint32_t SegIndex, uint64_t SegOffset) const {
463     return BindRebaseSectionTable->address(SegIndex, SegOffset);
464   }
465 
466   // In a MachO file, sections have a segment name. This is used in the .o
467   // files. They have a single segment, but this field specifies which segment
468   // a section should be put in the final object.
469   StringRef getSectionFinalSegmentName(DataRefImpl Sec) const;
470 
471   // Names are stored as 16 bytes. These returns the raw 16 bytes without
472   // interpreting them as a C string.
473   ArrayRef<char> getSectionRawName(DataRefImpl Sec) const;
474   ArrayRef<char> getSectionRawFinalSegmentName(DataRefImpl Sec) const;
475 
476   // MachO specific Info about relocations.
477   bool isRelocationScattered(const MachO::any_relocation_info &RE) const;
478   unsigned getPlainRelocationSymbolNum(
479                                     const MachO::any_relocation_info &RE) const;
480   bool getPlainRelocationExternal(const MachO::any_relocation_info &RE) const;
481   bool getScatteredRelocationScattered(
482                                     const MachO::any_relocation_info &RE) const;
483   uint32_t getScatteredRelocationValue(
484                                     const MachO::any_relocation_info &RE) const;
485   uint32_t getScatteredRelocationType(
486                                     const MachO::any_relocation_info &RE) const;
487   unsigned getAnyRelocationAddress(const MachO::any_relocation_info &RE) const;
488   unsigned getAnyRelocationPCRel(const MachO::any_relocation_info &RE) const;
489   unsigned getAnyRelocationLength(const MachO::any_relocation_info &RE) const;
490   unsigned getAnyRelocationType(const MachO::any_relocation_info &RE) const;
491   SectionRef getAnyRelocationSection(const MachO::any_relocation_info &RE) const;
492 
493   // MachO specific structures.
494   MachO::section getSection(DataRefImpl DRI) const;
495   MachO::section_64 getSection64(DataRefImpl DRI) const;
496   MachO::section getSection(const LoadCommandInfo &L, unsigned Index) const;
497   MachO::section_64 getSection64(const LoadCommandInfo &L,unsigned Index) const;
498   MachO::nlist getSymbolTableEntry(DataRefImpl DRI) const;
499   MachO::nlist_64 getSymbol64TableEntry(DataRefImpl DRI) const;
500 
501   MachO::linkedit_data_command
502   getLinkeditDataLoadCommand(const LoadCommandInfo &L) const;
503   MachO::segment_command
504   getSegmentLoadCommand(const LoadCommandInfo &L) const;
505   MachO::segment_command_64
506   getSegment64LoadCommand(const LoadCommandInfo &L) const;
507   MachO::linker_option_command
508   getLinkerOptionLoadCommand(const LoadCommandInfo &L) const;
509   MachO::version_min_command
510   getVersionMinLoadCommand(const LoadCommandInfo &L) const;
511   MachO::note_command
512   getNoteLoadCommand(const LoadCommandInfo &L) const;
513   MachO::build_version_command
514   getBuildVersionLoadCommand(const LoadCommandInfo &L) const;
515   MachO::build_tool_version
516   getBuildToolVersion(unsigned index) const;
517   MachO::dylib_command
518   getDylibIDLoadCommand(const LoadCommandInfo &L) const;
519   MachO::dyld_info_command
520   getDyldInfoLoadCommand(const LoadCommandInfo &L) const;
521   MachO::dylinker_command
522   getDylinkerCommand(const LoadCommandInfo &L) const;
523   MachO::uuid_command
524   getUuidCommand(const LoadCommandInfo &L) const;
525   MachO::rpath_command
526   getRpathCommand(const LoadCommandInfo &L) const;
527   MachO::source_version_command
528   getSourceVersionCommand(const LoadCommandInfo &L) const;
529   MachO::entry_point_command
530   getEntryPointCommand(const LoadCommandInfo &L) const;
531   MachO::encryption_info_command
532   getEncryptionInfoCommand(const LoadCommandInfo &L) const;
533   MachO::encryption_info_command_64
534   getEncryptionInfoCommand64(const LoadCommandInfo &L) const;
535   MachO::sub_framework_command
536   getSubFrameworkCommand(const LoadCommandInfo &L) const;
537   MachO::sub_umbrella_command
538   getSubUmbrellaCommand(const LoadCommandInfo &L) const;
539   MachO::sub_library_command
540   getSubLibraryCommand(const LoadCommandInfo &L) const;
541   MachO::sub_client_command
542   getSubClientCommand(const LoadCommandInfo &L) const;
543   MachO::routines_command
544   getRoutinesCommand(const LoadCommandInfo &L) const;
545   MachO::routines_command_64
546   getRoutinesCommand64(const LoadCommandInfo &L) const;
547   MachO::thread_command
548   getThreadCommand(const LoadCommandInfo &L) const;
549 
550   MachO::any_relocation_info getRelocation(DataRefImpl Rel) const;
551   MachO::data_in_code_entry getDice(DataRefImpl Rel) const;
552   const MachO::mach_header &getHeader() const;
553   const MachO::mach_header_64 &getHeader64() const;
554   uint32_t
555   getIndirectSymbolTableEntry(const MachO::dysymtab_command &DLC,
556                               unsigned Index) const;
557   MachO::data_in_code_entry getDataInCodeTableEntry(uint32_t DataOffset,
558                                                     unsigned Index) const;
559   MachO::symtab_command getSymtabLoadCommand() const;
560   MachO::dysymtab_command getDysymtabLoadCommand() const;
561   MachO::linkedit_data_command getDataInCodeLoadCommand() const;
562   MachO::linkedit_data_command getLinkOptHintsLoadCommand() const;
563   ArrayRef<uint8_t> getDyldInfoRebaseOpcodes() const;
564   ArrayRef<uint8_t> getDyldInfoBindOpcodes() const;
565   ArrayRef<uint8_t> getDyldInfoWeakBindOpcodes() const;
566   ArrayRef<uint8_t> getDyldInfoLazyBindOpcodes() const;
567   ArrayRef<uint8_t> getDyldInfoExportsTrie() const;
568   ArrayRef<uint8_t> getUuid() const;
569 
570   StringRef getStringTableData() const;
571   bool is64Bit() const;
572   void ReadULEB128s(uint64_t Index, SmallVectorImpl<uint64_t> &Out) const;
573 
574   static StringRef guessLibraryShortName(StringRef Name, bool &isFramework,
575                                          StringRef &Suffix);
576 
577   static Triple::ArchType getArch(uint32_t CPUType);
578   static Triple getArchTriple(uint32_t CPUType, uint32_t CPUSubType,
579                               const char **McpuDefault = nullptr,
580                               const char **ArchFlag = nullptr);
581   static bool isValidArch(StringRef ArchFlag);
582   static Triple getHostArch();
583 
584   bool isRelocatableObject() const override;
585 
586   StringRef mapDebugSectionName(StringRef Name) const override;
587 
hasPageZeroSegment()588   bool hasPageZeroSegment() const { return HasPageZeroSegment; }
589 
classof(const Binary * v)590   static bool classof(const Binary *v) {
591     return v->isMachO();
592   }
593 
594   static uint32_t
getVersionMinMajor(MachO::version_min_command & C,bool SDK)595   getVersionMinMajor(MachO::version_min_command &C, bool SDK) {
596     uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
597     return (VersionOrSDK >> 16) & 0xffff;
598   }
599 
600   static uint32_t
getVersionMinMinor(MachO::version_min_command & C,bool SDK)601   getVersionMinMinor(MachO::version_min_command &C, bool SDK) {
602     uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
603     return (VersionOrSDK >> 8) & 0xff;
604   }
605 
606   static uint32_t
getVersionMinUpdate(MachO::version_min_command & C,bool SDK)607   getVersionMinUpdate(MachO::version_min_command &C, bool SDK) {
608     uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
609     return VersionOrSDK & 0xff;
610   }
611 
getBuildPlatform(uint32_t platform)612   static std::string getBuildPlatform(uint32_t platform) {
613     switch (platform) {
614     case MachO::PLATFORM_MACOS: return "macos";
615     case MachO::PLATFORM_IOS: return "ios";
616     case MachO::PLATFORM_TVOS: return "tvos";
617     case MachO::PLATFORM_WATCHOS: return "watchos";
618     case MachO::PLATFORM_BRIDGEOS: return "bridgeos";
619     case MachO::PLATFORM_IOSSIMULATOR: return "iossimulator";
620     case MachO::PLATFORM_TVOSSIMULATOR: return "tvossimulator";
621     case MachO::PLATFORM_WATCHOSSIMULATOR: return "watchossimulator";
622     default:
623       std::string ret;
624       raw_string_ostream ss(ret);
625       ss << format_hex(platform, 8, true);
626       return ss.str();
627     }
628   }
629 
getBuildTool(uint32_t tools)630   static std::string getBuildTool(uint32_t tools) {
631     switch (tools) {
632     case MachO::TOOL_CLANG: return "clang";
633     case MachO::TOOL_SWIFT: return "swift";
634     case MachO::TOOL_LD: return "ld";
635     default:
636       std::string ret;
637       raw_string_ostream ss(ret);
638       ss << format_hex(tools, 8, true);
639       return ss.str();
640     }
641   }
642 
getVersionString(uint32_t version)643   static std::string getVersionString(uint32_t version) {
644     uint32_t major = (version >> 16) & 0xffff;
645     uint32_t minor = (version >> 8) & 0xff;
646     uint32_t update = version & 0xff;
647 
648     SmallString<32> Version;
649     Version = utostr(major) + "." + utostr(minor);
650     if (update != 0)
651       Version += "." + utostr(update);
652     return Version.str();
653   }
654 
655 private:
656   MachOObjectFile(MemoryBufferRef Object, bool IsLittleEndian, bool Is64Bits,
657                   Error &Err, uint32_t UniversalCputype = 0,
658                   uint32_t UniversalIndex = 0);
659 
660   uint64_t getSymbolValueImpl(DataRefImpl Symb) const override;
661 
662   union {
663     MachO::mach_header_64 Header64;
664     MachO::mach_header Header;
665   };
666   using SectionList = SmallVector<const char*, 1>;
667   SectionList Sections;
668   using LibraryList = SmallVector<const char*, 1>;
669   LibraryList Libraries;
670   LoadCommandList LoadCommands;
671   using LibraryShortName = SmallVector<StringRef, 1>;
672   using BuildToolList = SmallVector<const char*, 1>;
673   BuildToolList BuildTools;
674   mutable LibraryShortName LibrariesShortNames;
675   std::unique_ptr<BindRebaseSegInfo> BindRebaseSectionTable;
676   const char *SymtabLoadCmd = nullptr;
677   const char *DysymtabLoadCmd = nullptr;
678   const char *DataInCodeLoadCmd = nullptr;
679   const char *LinkOptHintsLoadCmd = nullptr;
680   const char *DyldInfoLoadCmd = nullptr;
681   const char *UuidLoadCmd = nullptr;
682   bool HasPageZeroSegment = false;
683 };
684 
685 /// DiceRef
DiceRef(DataRefImpl DiceP,const ObjectFile * Owner)686 inline DiceRef::DiceRef(DataRefImpl DiceP, const ObjectFile *Owner)
687   : DicePimpl(DiceP) , OwningObject(Owner) {}
688 
689 inline bool DiceRef::operator==(const DiceRef &Other) const {
690   return DicePimpl == Other.DicePimpl;
691 }
692 
693 inline bool DiceRef::operator<(const DiceRef &Other) const {
694   return DicePimpl < Other.DicePimpl;
695 }
696 
moveNext()697 inline void DiceRef::moveNext() {
698   const MachO::data_in_code_entry *P =
699     reinterpret_cast<const MachO::data_in_code_entry *>(DicePimpl.p);
700   DicePimpl.p = reinterpret_cast<uintptr_t>(P + 1);
701 }
702 
703 // Since a Mach-O data in code reference, a DiceRef, can only be created when
704 // the OwningObject ObjectFile is a MachOObjectFile a static_cast<> is used for
705 // the methods that get the values of the fields of the reference.
706 
getOffset(uint32_t & Result)707 inline std::error_code DiceRef::getOffset(uint32_t &Result) const {
708   const MachOObjectFile *MachOOF =
709     static_cast<const MachOObjectFile *>(OwningObject);
710   MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
711   Result = Dice.offset;
712   return std::error_code();
713 }
714 
getLength(uint16_t & Result)715 inline std::error_code DiceRef::getLength(uint16_t &Result) const {
716   const MachOObjectFile *MachOOF =
717     static_cast<const MachOObjectFile *>(OwningObject);
718   MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
719   Result = Dice.length;
720   return std::error_code();
721 }
722 
getKind(uint16_t & Result)723 inline std::error_code DiceRef::getKind(uint16_t &Result) const {
724   const MachOObjectFile *MachOOF =
725     static_cast<const MachOObjectFile *>(OwningObject);
726   MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
727   Result = Dice.kind;
728   return std::error_code();
729 }
730 
getRawDataRefImpl()731 inline DataRefImpl DiceRef::getRawDataRefImpl() const {
732   return DicePimpl;
733 }
734 
getObjectFile()735 inline const ObjectFile *DiceRef::getObjectFile() const {
736   return OwningObject;
737 }
738 
739 } // end namespace object
740 } // end namespace llvm
741 
742 #endif // LLVM_OBJECT_MACHO_H
743