1 //===- MachO.h - MachO object file implementation ---------------*- C++ -*-===//
2 //
3 // The LLVM Compiler Infrastructure
4 //
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
7 //
8 //===----------------------------------------------------------------------===//
9 //
10 // This file declares the MachOObjectFile class, which implement the ObjectFile
11 // interface for MachO files.
12 //
13 //===----------------------------------------------------------------------===//
14
15 #ifndef LLVM_OBJECT_MACHO_H
16 #define LLVM_OBJECT_MACHO_H
17
18 #include "llvm/ADT/ArrayRef.h"
19 #include "llvm/ADT/SmallVector.h"
20 #include "llvm/ADT/Triple.h"
21 #include "llvm/Object/ObjectFile.h"
22 #include "llvm/Support/MachO.h"
23
24 namespace llvm {
25 namespace object {
26
27 /// DiceRef - This is a value type class that represents a single
28 /// data in code entry in the table in a Mach-O object file.
29 class DiceRef {
30 DataRefImpl DicePimpl;
31 const ObjectFile *OwningObject;
32
33 public:
DiceRef()34 DiceRef() : OwningObject(nullptr) { }
35
36 DiceRef(DataRefImpl DiceP, const ObjectFile *Owner);
37
38 bool operator==(const DiceRef &Other) const;
39 bool operator<(const DiceRef &Other) const;
40
41 void moveNext();
42
43 std::error_code getOffset(uint32_t &Result) const;
44 std::error_code getLength(uint16_t &Result) const;
45 std::error_code getKind(uint16_t &Result) const;
46
47 DataRefImpl getRawDataRefImpl() const;
48 const ObjectFile *getObjectFile() const;
49 };
50 typedef content_iterator<DiceRef> dice_iterator;
51
52 /// ExportEntry encapsulates the current-state-of-the-walk used when doing a
53 /// non-recursive walk of the trie data structure. This allows you to iterate
54 /// across all exported symbols using:
55 /// for (const llvm::object::ExportEntry &AnExport : Obj->exports()) {
56 /// }
57 class ExportEntry {
58 public:
59 ExportEntry(ArrayRef<uint8_t> Trie);
60
61 StringRef name() const;
62 uint64_t flags() const;
63 uint64_t address() const;
64 uint64_t other() const;
65 StringRef otherName() const;
66 uint32_t nodeOffset() const;
67
68 bool operator==(const ExportEntry &) const;
69
70 void moveNext();
71
72 private:
73 friend class MachOObjectFile;
74 void moveToFirst();
75 void moveToEnd();
76 uint64_t readULEB128(const uint8_t *&p);
77 void pushDownUntilBottom();
78 void pushNode(uint64_t Offset);
79
80 // Represents a node in the mach-o exports trie.
81 struct NodeState {
82 NodeState(const uint8_t *Ptr);
83 const uint8_t *Start;
84 const uint8_t *Current;
85 uint64_t Flags;
86 uint64_t Address;
87 uint64_t Other;
88 const char *ImportName;
89 unsigned ChildCount;
90 unsigned NextChildIndex;
91 unsigned ParentStringLength;
92 bool IsExportNode;
93 };
94
95 ArrayRef<uint8_t> Trie;
96 SmallString<256> CumulativeString;
97 SmallVector<NodeState, 16> Stack;
98 bool Malformed;
99 bool Done;
100 };
101 typedef content_iterator<ExportEntry> export_iterator;
102
103 /// MachORebaseEntry encapsulates the current state in the decompression of
104 /// rebasing opcodes. This allows you to iterate through the compressed table of
105 /// rebasing using:
106 /// for (const llvm::object::MachORebaseEntry &Entry : Obj->rebaseTable()) {
107 /// }
108 class MachORebaseEntry {
109 public:
110 MachORebaseEntry(ArrayRef<uint8_t> opcodes, bool is64Bit);
111
112 uint32_t segmentIndex() const;
113 uint64_t segmentOffset() const;
114 StringRef typeName() const;
115
116 bool operator==(const MachORebaseEntry &) const;
117
118 void moveNext();
119
120 private:
121 friend class MachOObjectFile;
122 void moveToFirst();
123 void moveToEnd();
124 uint64_t readULEB128();
125
126 ArrayRef<uint8_t> Opcodes;
127 const uint8_t *Ptr;
128 uint64_t SegmentOffset;
129 uint32_t SegmentIndex;
130 uint64_t RemainingLoopCount;
131 uint64_t AdvanceAmount;
132 uint8_t RebaseType;
133 uint8_t PointerSize;
134 bool Malformed;
135 bool Done;
136 };
137 typedef content_iterator<MachORebaseEntry> rebase_iterator;
138
139 /// MachOBindEntry encapsulates the current state in the decompression of
140 /// binding opcodes. This allows you to iterate through the compressed table of
141 /// bindings using:
142 /// for (const llvm::object::MachOBindEntry &Entry : Obj->bindTable()) {
143 /// }
144 class MachOBindEntry {
145 public:
146 enum class Kind { Regular, Lazy, Weak };
147
148 MachOBindEntry(ArrayRef<uint8_t> Opcodes, bool is64Bit, MachOBindEntry::Kind);
149
150 uint32_t segmentIndex() const;
151 uint64_t segmentOffset() const;
152 StringRef typeName() const;
153 StringRef symbolName() const;
154 uint32_t flags() const;
155 int64_t addend() const;
156 int ordinal() const;
157
158 bool operator==(const MachOBindEntry &) const;
159
160 void moveNext();
161
162 private:
163 friend class MachOObjectFile;
164 void moveToFirst();
165 void moveToEnd();
166 uint64_t readULEB128();
167 int64_t readSLEB128();
168
169 ArrayRef<uint8_t> Opcodes;
170 const uint8_t *Ptr;
171 uint64_t SegmentOffset;
172 uint32_t SegmentIndex;
173 StringRef SymbolName;
174 int Ordinal;
175 uint32_t Flags;
176 int64_t Addend;
177 uint64_t RemainingLoopCount;
178 uint64_t AdvanceAmount;
179 uint8_t BindType;
180 uint8_t PointerSize;
181 Kind TableKind;
182 bool Malformed;
183 bool Done;
184 };
185 typedef content_iterator<MachOBindEntry> bind_iterator;
186
187 class MachOObjectFile : public ObjectFile {
188 public:
189 struct LoadCommandInfo {
190 const char *Ptr; // Where in memory the load command is.
191 MachO::load_command C; // The command itself.
192 };
193 typedef SmallVector<LoadCommandInfo, 4> LoadCommandList;
194 typedef LoadCommandList::const_iterator load_command_iterator;
195
196 MachOObjectFile(MemoryBufferRef Object, bool IsLittleEndian, bool Is64Bits,
197 std::error_code &EC);
198
199 void moveSymbolNext(DataRefImpl &Symb) const override;
200
201 uint64_t getNValue(DataRefImpl Sym) const;
202 ErrorOr<StringRef> getSymbolName(DataRefImpl Symb) const override;
203
204 // MachO specific.
205 std::error_code getIndirectName(DataRefImpl Symb, StringRef &Res) const;
206 unsigned getSectionType(SectionRef Sec) const;
207
208 ErrorOr<uint64_t> getSymbolAddress(DataRefImpl Symb) const override;
209 uint32_t getSymbolAlignment(DataRefImpl Symb) const override;
210 uint64_t getCommonSymbolSizeImpl(DataRefImpl Symb) const override;
211 SymbolRef::Type getSymbolType(DataRefImpl Symb) const override;
212 uint32_t getSymbolFlags(DataRefImpl Symb) const override;
213 ErrorOr<section_iterator> getSymbolSection(DataRefImpl Symb) const override;
214 unsigned getSymbolSectionID(SymbolRef Symb) const;
215 unsigned getSectionID(SectionRef Sec) const;
216
217 void moveSectionNext(DataRefImpl &Sec) const override;
218 std::error_code getSectionName(DataRefImpl Sec,
219 StringRef &Res) const override;
220 uint64_t getSectionAddress(DataRefImpl Sec) const override;
221 uint64_t getSectionSize(DataRefImpl Sec) const override;
222 std::error_code getSectionContents(DataRefImpl Sec,
223 StringRef &Res) const override;
224 uint64_t getSectionAlignment(DataRefImpl Sec) const override;
225 bool isSectionText(DataRefImpl Sec) const override;
226 bool isSectionData(DataRefImpl Sec) const override;
227 bool isSectionBSS(DataRefImpl Sec) const override;
228 bool isSectionVirtual(DataRefImpl Sec) const override;
229 relocation_iterator section_rel_begin(DataRefImpl Sec) const override;
230 relocation_iterator section_rel_end(DataRefImpl Sec) const override;
231
232 void moveRelocationNext(DataRefImpl &Rel) const override;
233 uint64_t getRelocationOffset(DataRefImpl Rel) const override;
234 symbol_iterator getRelocationSymbol(DataRefImpl Rel) const override;
235 section_iterator getRelocationSection(DataRefImpl Rel) const;
236 uint64_t getRelocationType(DataRefImpl Rel) const override;
237 void getRelocationTypeName(DataRefImpl Rel,
238 SmallVectorImpl<char> &Result) const override;
239 uint8_t getRelocationLength(DataRefImpl Rel) const;
240
241 // MachO specific.
242 std::error_code getLibraryShortNameByIndex(unsigned Index, StringRef &) const;
243
244 section_iterator getRelocationRelocatedSection(relocation_iterator Rel) const;
245
246 // TODO: Would be useful to have an iterator based version
247 // of the load command interface too.
248
249 basic_symbol_iterator symbol_begin_impl() const override;
250 basic_symbol_iterator symbol_end_impl() const override;
251
252 // MachO specific.
253 basic_symbol_iterator getSymbolByIndex(unsigned Index) const;
254
255 section_iterator section_begin() const override;
256 section_iterator section_end() const override;
257
258 uint8_t getBytesInAddress() const override;
259
260 StringRef getFileFormatName() const override;
261 unsigned getArch() const override;
262 Triple getArch(const char **McpuDefault, Triple *ThumbTriple) const;
263
264 relocation_iterator section_rel_begin(unsigned Index) const;
265 relocation_iterator section_rel_end(unsigned Index) const;
266
267 dice_iterator begin_dices() const;
268 dice_iterator end_dices() const;
269
270 load_command_iterator begin_load_commands() const;
271 load_command_iterator end_load_commands() const;
272 iterator_range<load_command_iterator> load_commands() const;
273
274 /// For use iterating over all exported symbols.
275 iterator_range<export_iterator> exports() const;
276
277 /// For use examining a trie not in a MachOObjectFile.
278 static iterator_range<export_iterator> exports(ArrayRef<uint8_t> Trie);
279
280 /// For use iterating over all rebase table entries.
281 iterator_range<rebase_iterator> rebaseTable() const;
282
283 /// For use examining rebase opcodes not in a MachOObjectFile.
284 static iterator_range<rebase_iterator> rebaseTable(ArrayRef<uint8_t> Opcodes,
285 bool is64);
286
287 /// For use iterating over all bind table entries.
288 iterator_range<bind_iterator> bindTable() const;
289
290 /// For use iterating over all lazy bind table entries.
291 iterator_range<bind_iterator> lazyBindTable() const;
292
293 /// For use iterating over all lazy bind table entries.
294 iterator_range<bind_iterator> weakBindTable() const;
295
296 /// For use examining bind opcodes not in a MachOObjectFile.
297 static iterator_range<bind_iterator> bindTable(ArrayRef<uint8_t> Opcodes,
298 bool is64,
299 MachOBindEntry::Kind);
300
301
302 // In a MachO file, sections have a segment name. This is used in the .o
303 // files. They have a single segment, but this field specifies which segment
304 // a section should be put in in the final object.
305 StringRef getSectionFinalSegmentName(DataRefImpl Sec) const;
306
307 // Names are stored as 16 bytes. These returns the raw 16 bytes without
308 // interpreting them as a C string.
309 ArrayRef<char> getSectionRawName(DataRefImpl Sec) const;
310 ArrayRef<char> getSectionRawFinalSegmentName(DataRefImpl Sec) const;
311
312 // MachO specific Info about relocations.
313 bool isRelocationScattered(const MachO::any_relocation_info &RE) const;
314 unsigned getPlainRelocationSymbolNum(
315 const MachO::any_relocation_info &RE) const;
316 bool getPlainRelocationExternal(const MachO::any_relocation_info &RE) const;
317 bool getScatteredRelocationScattered(
318 const MachO::any_relocation_info &RE) const;
319 uint32_t getScatteredRelocationValue(
320 const MachO::any_relocation_info &RE) const;
321 uint32_t getScatteredRelocationType(
322 const MachO::any_relocation_info &RE) const;
323 unsigned getAnyRelocationAddress(const MachO::any_relocation_info &RE) const;
324 unsigned getAnyRelocationPCRel(const MachO::any_relocation_info &RE) const;
325 unsigned getAnyRelocationLength(const MachO::any_relocation_info &RE) const;
326 unsigned getAnyRelocationType(const MachO::any_relocation_info &RE) const;
327 SectionRef getAnyRelocationSection(const MachO::any_relocation_info &RE) const;
328
329 // MachO specific structures.
330 MachO::section getSection(DataRefImpl DRI) const;
331 MachO::section_64 getSection64(DataRefImpl DRI) const;
332 MachO::section getSection(const LoadCommandInfo &L, unsigned Index) const;
333 MachO::section_64 getSection64(const LoadCommandInfo &L,unsigned Index) const;
334 MachO::nlist getSymbolTableEntry(DataRefImpl DRI) const;
335 MachO::nlist_64 getSymbol64TableEntry(DataRefImpl DRI) const;
336
337 MachO::linkedit_data_command
338 getLinkeditDataLoadCommand(const LoadCommandInfo &L) const;
339 MachO::segment_command
340 getSegmentLoadCommand(const LoadCommandInfo &L) const;
341 MachO::segment_command_64
342 getSegment64LoadCommand(const LoadCommandInfo &L) const;
343 MachO::linker_option_command
344 getLinkerOptionLoadCommand(const LoadCommandInfo &L) const;
345 MachO::version_min_command
346 getVersionMinLoadCommand(const LoadCommandInfo &L) const;
347 MachO::dylib_command
348 getDylibIDLoadCommand(const LoadCommandInfo &L) const;
349 MachO::dyld_info_command
350 getDyldInfoLoadCommand(const LoadCommandInfo &L) const;
351 MachO::dylinker_command
352 getDylinkerCommand(const LoadCommandInfo &L) const;
353 MachO::uuid_command
354 getUuidCommand(const LoadCommandInfo &L) const;
355 MachO::rpath_command
356 getRpathCommand(const LoadCommandInfo &L) const;
357 MachO::source_version_command
358 getSourceVersionCommand(const LoadCommandInfo &L) const;
359 MachO::entry_point_command
360 getEntryPointCommand(const LoadCommandInfo &L) const;
361 MachO::encryption_info_command
362 getEncryptionInfoCommand(const LoadCommandInfo &L) const;
363 MachO::encryption_info_command_64
364 getEncryptionInfoCommand64(const LoadCommandInfo &L) const;
365 MachO::sub_framework_command
366 getSubFrameworkCommand(const LoadCommandInfo &L) const;
367 MachO::sub_umbrella_command
368 getSubUmbrellaCommand(const LoadCommandInfo &L) const;
369 MachO::sub_library_command
370 getSubLibraryCommand(const LoadCommandInfo &L) const;
371 MachO::sub_client_command
372 getSubClientCommand(const LoadCommandInfo &L) const;
373 MachO::routines_command
374 getRoutinesCommand(const LoadCommandInfo &L) const;
375 MachO::routines_command_64
376 getRoutinesCommand64(const LoadCommandInfo &L) const;
377 MachO::thread_command
378 getThreadCommand(const LoadCommandInfo &L) const;
379
380 MachO::any_relocation_info getRelocation(DataRefImpl Rel) const;
381 MachO::data_in_code_entry getDice(DataRefImpl Rel) const;
382 const MachO::mach_header &getHeader() const;
383 const MachO::mach_header_64 &getHeader64() const;
384 uint32_t
385 getIndirectSymbolTableEntry(const MachO::dysymtab_command &DLC,
386 unsigned Index) const;
387 MachO::data_in_code_entry getDataInCodeTableEntry(uint32_t DataOffset,
388 unsigned Index) const;
389 MachO::symtab_command getSymtabLoadCommand() const;
390 MachO::dysymtab_command getDysymtabLoadCommand() const;
391 MachO::linkedit_data_command getDataInCodeLoadCommand() const;
392 MachO::linkedit_data_command getLinkOptHintsLoadCommand() const;
393 ArrayRef<uint8_t> getDyldInfoRebaseOpcodes() const;
394 ArrayRef<uint8_t> getDyldInfoBindOpcodes() const;
395 ArrayRef<uint8_t> getDyldInfoWeakBindOpcodes() const;
396 ArrayRef<uint8_t> getDyldInfoLazyBindOpcodes() const;
397 ArrayRef<uint8_t> getDyldInfoExportsTrie() const;
398 ArrayRef<uint8_t> getUuid() const;
399
400 StringRef getStringTableData() const;
401 bool is64Bit() const;
402 void ReadULEB128s(uint64_t Index, SmallVectorImpl<uint64_t> &Out) const;
403
404 static StringRef guessLibraryShortName(StringRef Name, bool &isFramework,
405 StringRef &Suffix);
406
407 static Triple::ArchType getArch(uint32_t CPUType);
408 static Triple getArch(uint32_t CPUType, uint32_t CPUSubType,
409 const char **McpuDefault = nullptr);
410 static Triple getThumbArch(uint32_t CPUType, uint32_t CPUSubType,
411 const char **McpuDefault = nullptr);
412 static Triple getArch(uint32_t CPUType, uint32_t CPUSubType,
413 const char **McpuDefault, Triple *ThumbTriple);
414 static bool isValidArch(StringRef ArchFlag);
415 static Triple getHostArch();
416
417 bool isRelocatableObject() const override;
418
hasPageZeroSegment()419 bool hasPageZeroSegment() const { return HasPageZeroSegment; }
420
classof(const Binary * v)421 static bool classof(const Binary *v) {
422 return v->isMachO();
423 }
424
425 static uint32_t
getVersionMinMajor(MachO::version_min_command & C,bool SDK)426 getVersionMinMajor(MachO::version_min_command &C, bool SDK) {
427 uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
428 return (VersionOrSDK >> 16) & 0xffff;
429 }
430
431 static uint32_t
getVersionMinMinor(MachO::version_min_command & C,bool SDK)432 getVersionMinMinor(MachO::version_min_command &C, bool SDK) {
433 uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
434 return (VersionOrSDK >> 8) & 0xff;
435 }
436
437 static uint32_t
getVersionMinUpdate(MachO::version_min_command & C,bool SDK)438 getVersionMinUpdate(MachO::version_min_command &C, bool SDK) {
439 uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
440 return VersionOrSDK & 0xff;
441 }
442
443 private:
444 uint64_t getSymbolValueImpl(DataRefImpl Symb) const override;
445
446 union {
447 MachO::mach_header_64 Header64;
448 MachO::mach_header Header;
449 };
450 typedef SmallVector<const char*, 1> SectionList;
451 SectionList Sections;
452 typedef SmallVector<const char*, 1> LibraryList;
453 LibraryList Libraries;
454 LoadCommandList LoadCommands;
455 typedef SmallVector<StringRef, 1> LibraryShortName;
456 mutable LibraryShortName LibrariesShortNames;
457 const char *SymtabLoadCmd;
458 const char *DysymtabLoadCmd;
459 const char *DataInCodeLoadCmd;
460 const char *LinkOptHintsLoadCmd;
461 const char *DyldInfoLoadCmd;
462 const char *UuidLoadCmd;
463 bool HasPageZeroSegment;
464 };
465
466 /// DiceRef
DiceRef(DataRefImpl DiceP,const ObjectFile * Owner)467 inline DiceRef::DiceRef(DataRefImpl DiceP, const ObjectFile *Owner)
468 : DicePimpl(DiceP) , OwningObject(Owner) {}
469
470 inline bool DiceRef::operator==(const DiceRef &Other) const {
471 return DicePimpl == Other.DicePimpl;
472 }
473
474 inline bool DiceRef::operator<(const DiceRef &Other) const {
475 return DicePimpl < Other.DicePimpl;
476 }
477
moveNext()478 inline void DiceRef::moveNext() {
479 const MachO::data_in_code_entry *P =
480 reinterpret_cast<const MachO::data_in_code_entry *>(DicePimpl.p);
481 DicePimpl.p = reinterpret_cast<uintptr_t>(P + 1);
482 }
483
484 // Since a Mach-O data in code reference, a DiceRef, can only be created when
485 // the OwningObject ObjectFile is a MachOObjectFile a static_cast<> is used for
486 // the methods that get the values of the fields of the reference.
487
getOffset(uint32_t & Result)488 inline std::error_code DiceRef::getOffset(uint32_t &Result) const {
489 const MachOObjectFile *MachOOF =
490 static_cast<const MachOObjectFile *>(OwningObject);
491 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
492 Result = Dice.offset;
493 return std::error_code();
494 }
495
getLength(uint16_t & Result)496 inline std::error_code DiceRef::getLength(uint16_t &Result) const {
497 const MachOObjectFile *MachOOF =
498 static_cast<const MachOObjectFile *>(OwningObject);
499 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
500 Result = Dice.length;
501 return std::error_code();
502 }
503
getKind(uint16_t & Result)504 inline std::error_code DiceRef::getKind(uint16_t &Result) const {
505 const MachOObjectFile *MachOOF =
506 static_cast<const MachOObjectFile *>(OwningObject);
507 MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
508 Result = Dice.kind;
509 return std::error_code();
510 }
511
getRawDataRefImpl()512 inline DataRefImpl DiceRef::getRawDataRefImpl() const {
513 return DicePimpl;
514 }
515
getObjectFile()516 inline const ObjectFile *DiceRef::getObjectFile() const {
517 return OwningObject;
518 }
519
520 }
521 }
522
523 #endif
524