1 //===- MachOObjectFile.cpp - Mach-O object file binding ---------*- 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 defines the MachOObjectFile class, which binds the MachOObject
11 // class to the generic ObjectFile wrapper.
12 //
13 //===----------------------------------------------------------------------===//
14 
15 #include "llvm/Object/MachO.h"
16 #include "llvm/ADT/STLExtras.h"
17 #include "llvm/ADT/StringSwitch.h"
18 #include "llvm/ADT/Triple.h"
19 #include "llvm/Support/DataExtractor.h"
20 #include "llvm/Support/Format.h"
21 #include "llvm/Support/Host.h"
22 #include "llvm/Support/MemoryBuffer.h"
23 #include "llvm/Support/raw_ostream.h"
24 #include <cctype>
25 #include <cstring>
26 #include <limits>
27 
28 using namespace llvm;
29 using namespace object;
30 
31 namespace {
32   struct section_base {
33     char sectname[16];
34     char segname[16];
35   };
36 }
37 
38 template<typename T>
39 static T getStruct(const MachOObjectFile *O, const char *P) {
40   T Cmd;
41   memcpy(&Cmd, P, sizeof(T));
42   if (O->isLittleEndian() != sys::IsLittleEndianHost)
43     MachO::swapStruct(Cmd);
44   return Cmd;
45 }
46 
47 static uint32_t
48 getSegmentLoadCommandNumSections(const MachOObjectFile *O,
49                                  const MachOObjectFile::LoadCommandInfo &L) {
50   if (O->is64Bit()) {
51     MachO::segment_command_64 S = O->getSegment64LoadCommand(L);
52     return S.nsects;
53   }
54   MachO::segment_command S = O->getSegmentLoadCommand(L);
55   return S.nsects;
56 }
57 
58 static const char *
59 getSectionPtr(const MachOObjectFile *O, MachOObjectFile::LoadCommandInfo L,
60               unsigned Sec) {
61   uintptr_t CommandAddr = reinterpret_cast<uintptr_t>(L.Ptr);
62 
63   bool Is64 = O->is64Bit();
64   unsigned SegmentLoadSize = Is64 ? sizeof(MachO::segment_command_64) :
65                                     sizeof(MachO::segment_command);
66   unsigned SectionSize = Is64 ? sizeof(MachO::section_64) :
67                                 sizeof(MachO::section);
68 
69   uintptr_t SectionAddr = CommandAddr + SegmentLoadSize + Sec * SectionSize;
70   return reinterpret_cast<const char*>(SectionAddr);
71 }
72 
73 static const char *getPtr(const MachOObjectFile *O, size_t Offset) {
74   return O->getData().substr(Offset, 1).data();
75 }
76 
77 static MachO::nlist_base
78 getSymbolTableEntryBase(const MachOObjectFile *O, DataRefImpl DRI) {
79   const char *P = reinterpret_cast<const char *>(DRI.p);
80   return getStruct<MachO::nlist_base>(O, P);
81 }
82 
83 static StringRef parseSegmentOrSectionName(const char *P) {
84   if (P[15] == 0)
85     // Null terminated.
86     return P;
87   // Not null terminated, so this is a 16 char string.
88   return StringRef(P, 16);
89 }
90 
91 // Helper to advance a section or symbol iterator multiple increments at a time.
92 template<class T>
93 static void advance(T &it, size_t Val) {
94   while (Val--)
95     ++it;
96 }
97 
98 static unsigned getCPUType(const MachOObjectFile *O) {
99   return O->getHeader().cputype;
100 }
101 
102 static void printRelocationTargetName(const MachOObjectFile *O,
103                                       const MachO::any_relocation_info &RE,
104                                       raw_string_ostream &fmt) {
105   bool IsScattered = O->isRelocationScattered(RE);
106 
107   // Target of a scattered relocation is an address.  In the interest of
108   // generating pretty output, scan through the symbol table looking for a
109   // symbol that aligns with that address.  If we find one, print it.
110   // Otherwise, we just print the hex address of the target.
111   if (IsScattered) {
112     uint32_t Val = O->getPlainRelocationSymbolNum(RE);
113 
114     for (const SymbolRef &Symbol : O->symbols()) {
115       std::error_code ec;
116       uint64_t Addr;
117       StringRef Name;
118 
119       if ((ec = Symbol.getAddress(Addr)))
120         report_fatal_error(ec.message());
121       if (Addr != Val)
122         continue;
123       if ((ec = Symbol.getName(Name)))
124         report_fatal_error(ec.message());
125       fmt << Name;
126       return;
127     }
128 
129     // If we couldn't find a symbol that this relocation refers to, try
130     // to find a section beginning instead.
131     for (const SectionRef &Section : O->sections()) {
132       std::error_code ec;
133       uint64_t Addr;
134       StringRef Name;
135 
136       if ((ec = Section.getAddress(Addr)))
137         report_fatal_error(ec.message());
138       if (Addr != Val)
139         continue;
140       if ((ec = Section.getName(Name)))
141         report_fatal_error(ec.message());
142       fmt << Name;
143       return;
144     }
145 
146     fmt << format("0x%x", Val);
147     return;
148   }
149 
150   StringRef S;
151   bool isExtern = O->getPlainRelocationExternal(RE);
152   uint64_t Val = O->getPlainRelocationSymbolNum(RE);
153 
154   if (isExtern) {
155     symbol_iterator SI = O->symbol_begin();
156     advance(SI, Val);
157     SI->getName(S);
158   } else {
159     section_iterator SI = O->section_begin();
160     // Adjust for the fact that sections are 1-indexed.
161     advance(SI, Val - 1);
162     SI->getName(S);
163   }
164 
165   fmt << S;
166 }
167 
168 static uint32_t
169 getPlainRelocationAddress(const MachO::any_relocation_info &RE) {
170   return RE.r_word0;
171 }
172 
173 static unsigned
174 getScatteredRelocationAddress(const MachO::any_relocation_info &RE) {
175   return RE.r_word0 & 0xffffff;
176 }
177 
178 static bool getPlainRelocationPCRel(const MachOObjectFile *O,
179                                     const MachO::any_relocation_info &RE) {
180   if (O->isLittleEndian())
181     return (RE.r_word1 >> 24) & 1;
182   return (RE.r_word1 >> 7) & 1;
183 }
184 
185 static bool
186 getScatteredRelocationPCRel(const MachOObjectFile *O,
187                             const MachO::any_relocation_info &RE) {
188   return (RE.r_word0 >> 30) & 1;
189 }
190 
191 static unsigned getPlainRelocationLength(const MachOObjectFile *O,
192                                          const MachO::any_relocation_info &RE) {
193   if (O->isLittleEndian())
194     return (RE.r_word1 >> 25) & 3;
195   return (RE.r_word1 >> 5) & 3;
196 }
197 
198 static unsigned
199 getScatteredRelocationLength(const MachO::any_relocation_info &RE) {
200   return (RE.r_word0 >> 28) & 3;
201 }
202 
203 static unsigned getPlainRelocationType(const MachOObjectFile *O,
204                                        const MachO::any_relocation_info &RE) {
205   if (O->isLittleEndian())
206     return RE.r_word1 >> 28;
207   return RE.r_word1 & 0xf;
208 }
209 
210 static unsigned
211 getScatteredRelocationType(const MachO::any_relocation_info &RE) {
212   return (RE.r_word0 >> 24) & 0xf;
213 }
214 
215 static uint32_t getSectionFlags(const MachOObjectFile *O,
216                                 DataRefImpl Sec) {
217   if (O->is64Bit()) {
218     MachO::section_64 Sect = O->getSection64(Sec);
219     return Sect.flags;
220   }
221   MachO::section Sect = O->getSection(Sec);
222   return Sect.flags;
223 }
224 
225 MachOObjectFile::MachOObjectFile(MemoryBufferRef Object, bool IsLittleEndian,
226                                  bool Is64bits, std::error_code &EC)
227     : ObjectFile(getMachOType(IsLittleEndian, Is64bits), Object),
228       SymtabLoadCmd(nullptr), DysymtabLoadCmd(nullptr),
229       DataInCodeLoadCmd(nullptr) {
230   uint32_t LoadCommandCount = this->getHeader().ncmds;
231   MachO::LoadCommandType SegmentLoadType = is64Bit() ?
232     MachO::LC_SEGMENT_64 : MachO::LC_SEGMENT;
233 
234   MachOObjectFile::LoadCommandInfo Load = getFirstLoadCommandInfo();
235   for (unsigned I = 0; ; ++I) {
236     if (Load.C.cmd == MachO::LC_SYMTAB) {
237       assert(!SymtabLoadCmd && "Multiple symbol tables");
238       SymtabLoadCmd = Load.Ptr;
239     } else if (Load.C.cmd == MachO::LC_DYSYMTAB) {
240       assert(!DysymtabLoadCmd && "Multiple dynamic symbol tables");
241       DysymtabLoadCmd = Load.Ptr;
242     } else if (Load.C.cmd == MachO::LC_DATA_IN_CODE) {
243       assert(!DataInCodeLoadCmd && "Multiple data in code tables");
244       DataInCodeLoadCmd = Load.Ptr;
245     } else if (Load.C.cmd == SegmentLoadType) {
246       uint32_t NumSections = getSegmentLoadCommandNumSections(this, Load);
247       for (unsigned J = 0; J < NumSections; ++J) {
248         const char *Sec = getSectionPtr(this, Load, J);
249         Sections.push_back(Sec);
250       }
251     } else if (Load.C.cmd == MachO::LC_LOAD_DYLIB ||
252                Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB ||
253                Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB ||
254                Load.C.cmd == MachO::LC_REEXPORT_DYLIB ||
255                Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) {
256       Libraries.push_back(Load.Ptr);
257     }
258 
259     if (I == LoadCommandCount - 1)
260       break;
261     else
262       Load = getNextLoadCommandInfo(Load);
263   }
264 }
265 
266 void MachOObjectFile::moveSymbolNext(DataRefImpl &Symb) const {
267   unsigned SymbolTableEntrySize = is64Bit() ?
268     sizeof(MachO::nlist_64) :
269     sizeof(MachO::nlist);
270   Symb.p += SymbolTableEntrySize;
271 }
272 
273 std::error_code MachOObjectFile::getSymbolName(DataRefImpl Symb,
274                                                StringRef &Res) const {
275   StringRef StringTable = getStringTableData();
276   MachO::nlist_base Entry = getSymbolTableEntryBase(this, Symb);
277   const char *Start = &StringTable.data()[Entry.n_strx];
278   Res = StringRef(Start);
279   return object_error::success;
280 }
281 
282 // getIndirectName() returns the name of the alias'ed symbol who's string table
283 // index is in the n_value field.
284 std::error_code MachOObjectFile::getIndirectName(DataRefImpl Symb,
285                                                  StringRef &Res) const {
286   StringRef StringTable = getStringTableData();
287   uint64_t NValue;
288   if (is64Bit()) {
289     MachO::nlist_64 Entry = getSymbol64TableEntry(Symb);
290     NValue = Entry.n_value;
291     if ((Entry.n_type & MachO::N_TYPE) != MachO::N_INDR)
292       return object_error::parse_failed;
293   } else {
294     MachO::nlist Entry = getSymbolTableEntry(Symb);
295     NValue = Entry.n_value;
296     if ((Entry.n_type & MachO::N_TYPE) != MachO::N_INDR)
297       return object_error::parse_failed;
298   }
299   if (NValue >= StringTable.size())
300     return object_error::parse_failed;
301   const char *Start = &StringTable.data()[NValue];
302   Res = StringRef(Start);
303   return object_error::success;
304 }
305 
306 std::error_code MachOObjectFile::getSymbolAddress(DataRefImpl Symb,
307                                                   uint64_t &Res) const {
308   if (is64Bit()) {
309     MachO::nlist_64 Entry = getSymbol64TableEntry(Symb);
310     if ((Entry.n_type & MachO::N_TYPE) == MachO::N_UNDF &&
311         Entry.n_value == 0)
312       Res = UnknownAddressOrSize;
313     else
314       Res = Entry.n_value;
315   } else {
316     MachO::nlist Entry = getSymbolTableEntry(Symb);
317     if ((Entry.n_type & MachO::N_TYPE) == MachO::N_UNDF &&
318         Entry.n_value == 0)
319       Res = UnknownAddressOrSize;
320     else
321       Res = Entry.n_value;
322   }
323   return object_error::success;
324 }
325 
326 std::error_code MachOObjectFile::getSymbolAlignment(DataRefImpl DRI,
327                                                     uint32_t &Result) const {
328   uint32_t flags = getSymbolFlags(DRI);
329   if (flags & SymbolRef::SF_Common) {
330     MachO::nlist_base Entry = getSymbolTableEntryBase(this, DRI);
331     Result = 1 << MachO::GET_COMM_ALIGN(Entry.n_desc);
332   } else {
333     Result = 0;
334   }
335   return object_error::success;
336 }
337 
338 std::error_code MachOObjectFile::getSymbolSize(DataRefImpl DRI,
339                                                uint64_t &Result) const {
340   uint64_t BeginOffset;
341   uint64_t EndOffset = 0;
342   uint8_t SectionIndex;
343 
344   MachO::nlist_base Entry = getSymbolTableEntryBase(this, DRI);
345   uint64_t Value;
346   getSymbolAddress(DRI, Value);
347   if (Value == UnknownAddressOrSize) {
348     Result = UnknownAddressOrSize;
349     return object_error::success;
350   }
351 
352   BeginOffset = Value;
353 
354   SectionIndex = Entry.n_sect;
355   if (!SectionIndex) {
356     uint32_t flags = getSymbolFlags(DRI);
357     if (flags & SymbolRef::SF_Common)
358       Result = Value;
359     else
360       Result = UnknownAddressOrSize;
361     return object_error::success;
362   }
363   // Unfortunately symbols are unsorted so we need to touch all
364   // symbols from load command
365   for (const SymbolRef &Symbol : symbols()) {
366     DataRefImpl DRI = Symbol.getRawDataRefImpl();
367     Entry = getSymbolTableEntryBase(this, DRI);
368     getSymbolAddress(DRI, Value);
369     if (Value == UnknownAddressOrSize)
370       continue;
371     if (Entry.n_sect == SectionIndex && Value > BeginOffset)
372       if (!EndOffset || Value < EndOffset)
373         EndOffset = Value;
374   }
375   if (!EndOffset) {
376     uint64_t Size;
377     DataRefImpl Sec;
378     Sec.d.a = SectionIndex-1;
379     getSectionSize(Sec, Size);
380     getSectionAddress(Sec, EndOffset);
381     EndOffset += Size;
382   }
383   Result = EndOffset - BeginOffset;
384   return object_error::success;
385 }
386 
387 std::error_code MachOObjectFile::getSymbolType(DataRefImpl Symb,
388                                                SymbolRef::Type &Res) const {
389   MachO::nlist_base Entry = getSymbolTableEntryBase(this, Symb);
390   uint8_t n_type = Entry.n_type;
391 
392   Res = SymbolRef::ST_Other;
393 
394   // If this is a STAB debugging symbol, we can do nothing more.
395   if (n_type & MachO::N_STAB) {
396     Res = SymbolRef::ST_Debug;
397     return object_error::success;
398   }
399 
400   switch (n_type & MachO::N_TYPE) {
401     case MachO::N_UNDF :
402       Res = SymbolRef::ST_Unknown;
403       break;
404     case MachO::N_SECT :
405       Res = SymbolRef::ST_Function;
406       break;
407   }
408   return object_error::success;
409 }
410 
411 uint32_t MachOObjectFile::getSymbolFlags(DataRefImpl DRI) const {
412   MachO::nlist_base Entry = getSymbolTableEntryBase(this, DRI);
413 
414   uint8_t MachOType = Entry.n_type;
415   uint16_t MachOFlags = Entry.n_desc;
416 
417   uint32_t Result = SymbolRef::SF_None;
418 
419   if ((MachOType & MachO::N_TYPE) == MachO::N_UNDF)
420     Result |= SymbolRef::SF_Undefined;
421 
422   if ((MachOType & MachO::N_TYPE) == MachO::N_INDR)
423     Result |= SymbolRef::SF_Indirect;
424 
425   if (MachOType & MachO::N_STAB)
426     Result |= SymbolRef::SF_FormatSpecific;
427 
428   if (MachOType & MachO::N_EXT) {
429     Result |= SymbolRef::SF_Global;
430     if ((MachOType & MachO::N_TYPE) == MachO::N_UNDF) {
431       uint64_t Value;
432       getSymbolAddress(DRI, Value);
433       if (Value && Value != UnknownAddressOrSize)
434         Result |= SymbolRef::SF_Common;
435     }
436   }
437 
438   if (MachOFlags & (MachO::N_WEAK_REF | MachO::N_WEAK_DEF))
439     Result |= SymbolRef::SF_Weak;
440 
441   if (MachOFlags & (MachO::N_ARM_THUMB_DEF))
442     Result |= SymbolRef::SF_Thumb;
443 
444   if ((MachOType & MachO::N_TYPE) == MachO::N_ABS)
445     Result |= SymbolRef::SF_Absolute;
446 
447   return Result;
448 }
449 
450 std::error_code MachOObjectFile::getSymbolSection(DataRefImpl Symb,
451                                                   section_iterator &Res) const {
452   MachO::nlist_base Entry = getSymbolTableEntryBase(this, Symb);
453   uint8_t index = Entry.n_sect;
454 
455   if (index == 0) {
456     Res = section_end();
457   } else {
458     DataRefImpl DRI;
459     DRI.d.a = index - 1;
460     Res = section_iterator(SectionRef(DRI, this));
461   }
462 
463   return object_error::success;
464 }
465 
466 void MachOObjectFile::moveSectionNext(DataRefImpl &Sec) const {
467   Sec.d.a++;
468 }
469 
470 std::error_code MachOObjectFile::getSectionName(DataRefImpl Sec,
471                                                 StringRef &Result) const {
472   ArrayRef<char> Raw = getSectionRawName(Sec);
473   Result = parseSegmentOrSectionName(Raw.data());
474   return object_error::success;
475 }
476 
477 std::error_code MachOObjectFile::getSectionAddress(DataRefImpl Sec,
478                                                    uint64_t &Res) const {
479   if (is64Bit()) {
480     MachO::section_64 Sect = getSection64(Sec);
481     Res = Sect.addr;
482   } else {
483     MachO::section Sect = getSection(Sec);
484     Res = Sect.addr;
485   }
486   return object_error::success;
487 }
488 
489 std::error_code MachOObjectFile::getSectionSize(DataRefImpl Sec,
490                                                 uint64_t &Res) const {
491   if (is64Bit()) {
492     MachO::section_64 Sect = getSection64(Sec);
493     Res = Sect.size;
494   } else {
495     MachO::section Sect = getSection(Sec);
496     Res = Sect.size;
497   }
498 
499   return object_error::success;
500 }
501 
502 std::error_code MachOObjectFile::getSectionContents(DataRefImpl Sec,
503                                                     StringRef &Res) const {
504   uint32_t Offset;
505   uint64_t Size;
506 
507   if (is64Bit()) {
508     MachO::section_64 Sect = getSection64(Sec);
509     Offset = Sect.offset;
510     Size = Sect.size;
511   } else {
512     MachO::section Sect = getSection(Sec);
513     Offset = Sect.offset;
514     Size = Sect.size;
515   }
516 
517   Res = this->getData().substr(Offset, Size);
518   return object_error::success;
519 }
520 
521 std::error_code MachOObjectFile::getSectionAlignment(DataRefImpl Sec,
522                                                      uint64_t &Res) const {
523   uint32_t Align;
524   if (is64Bit()) {
525     MachO::section_64 Sect = getSection64(Sec);
526     Align = Sect.align;
527   } else {
528     MachO::section Sect = getSection(Sec);
529     Align = Sect.align;
530   }
531 
532   Res = uint64_t(1) << Align;
533   return object_error::success;
534 }
535 
536 std::error_code MachOObjectFile::isSectionText(DataRefImpl Sec,
537                                                bool &Res) const {
538   uint32_t Flags = getSectionFlags(this, Sec);
539   Res = Flags & MachO::S_ATTR_PURE_INSTRUCTIONS;
540   return object_error::success;
541 }
542 
543 std::error_code MachOObjectFile::isSectionData(DataRefImpl Sec,
544                                                bool &Result) const {
545   uint32_t Flags = getSectionFlags(this, Sec);
546   unsigned SectionType = Flags & MachO::SECTION_TYPE;
547   Result = !(Flags & MachO::S_ATTR_PURE_INSTRUCTIONS) &&
548            !(SectionType == MachO::S_ZEROFILL ||
549              SectionType == MachO::S_GB_ZEROFILL);
550   return object_error::success;
551 }
552 
553 std::error_code MachOObjectFile::isSectionBSS(DataRefImpl Sec,
554                                               bool &Result) const {
555   uint32_t Flags = getSectionFlags(this, Sec);
556   unsigned SectionType = Flags & MachO::SECTION_TYPE;
557   Result = !(Flags & MachO::S_ATTR_PURE_INSTRUCTIONS) &&
558            (SectionType == MachO::S_ZEROFILL ||
559             SectionType == MachO::S_GB_ZEROFILL);
560   return object_error::success;
561 }
562 
563 std::error_code
564 MachOObjectFile::isSectionRequiredForExecution(DataRefImpl Sec,
565                                                bool &Result) const {
566   // FIXME: Unimplemented.
567   Result = true;
568   return object_error::success;
569 }
570 
571 std::error_code MachOObjectFile::isSectionVirtual(DataRefImpl Sec,
572                                                   bool &Result) const {
573   // FIXME: Unimplemented.
574   Result = false;
575   return object_error::success;
576 }
577 
578 std::error_code MachOObjectFile::isSectionZeroInit(DataRefImpl Sec,
579                                                    bool &Res) const {
580   uint32_t Flags = getSectionFlags(this, Sec);
581   unsigned SectionType = Flags & MachO::SECTION_TYPE;
582   Res = SectionType == MachO::S_ZEROFILL ||
583     SectionType == MachO::S_GB_ZEROFILL;
584   return object_error::success;
585 }
586 
587 std::error_code MachOObjectFile::isSectionReadOnlyData(DataRefImpl Sec,
588                                                        bool &Result) const {
589   // Consider using the code from isSectionText to look for __const sections.
590   // Alternately, emit S_ATTR_PURE_INSTRUCTIONS and/or S_ATTR_SOME_INSTRUCTIONS
591   // to use section attributes to distinguish code from data.
592 
593   // FIXME: Unimplemented.
594   Result = false;
595   return object_error::success;
596 }
597 
598 std::error_code MachOObjectFile::sectionContainsSymbol(DataRefImpl Sec,
599                                                        DataRefImpl Symb,
600                                                        bool &Result) const {
601   SymbolRef::Type ST;
602   this->getSymbolType(Symb, ST);
603   if (ST == SymbolRef::ST_Unknown) {
604     Result = false;
605     return object_error::success;
606   }
607 
608   uint64_t SectBegin, SectEnd;
609   getSectionAddress(Sec, SectBegin);
610   getSectionSize(Sec, SectEnd);
611   SectEnd += SectBegin;
612 
613   uint64_t SymAddr;
614   getSymbolAddress(Symb, SymAddr);
615   Result = (SymAddr >= SectBegin) && (SymAddr < SectEnd);
616 
617   return object_error::success;
618 }
619 
620 relocation_iterator MachOObjectFile::section_rel_begin(DataRefImpl Sec) const {
621   DataRefImpl Ret;
622   Ret.d.a = Sec.d.a;
623   Ret.d.b = 0;
624   return relocation_iterator(RelocationRef(Ret, this));
625 }
626 
627 relocation_iterator
628 MachOObjectFile::section_rel_end(DataRefImpl Sec) const {
629   uint32_t Num;
630   if (is64Bit()) {
631     MachO::section_64 Sect = getSection64(Sec);
632     Num = Sect.nreloc;
633   } else {
634     MachO::section Sect = getSection(Sec);
635     Num = Sect.nreloc;
636   }
637 
638   DataRefImpl Ret;
639   Ret.d.a = Sec.d.a;
640   Ret.d.b = Num;
641   return relocation_iterator(RelocationRef(Ret, this));
642 }
643 
644 void MachOObjectFile::moveRelocationNext(DataRefImpl &Rel) const {
645   ++Rel.d.b;
646 }
647 
648 std::error_code MachOObjectFile::getRelocationAddress(DataRefImpl Rel,
649                                                       uint64_t &Res) const {
650   uint64_t Offset;
651   getRelocationOffset(Rel, Offset);
652 
653   DataRefImpl Sec;
654   Sec.d.a = Rel.d.a;
655   uint64_t SecAddress;
656   getSectionAddress(Sec, SecAddress);
657   Res = SecAddress + Offset;
658   return object_error::success;
659 }
660 
661 std::error_code MachOObjectFile::getRelocationOffset(DataRefImpl Rel,
662                                                      uint64_t &Res) const {
663   assert(getHeader().filetype == MachO::MH_OBJECT &&
664          "Only implemented for MH_OBJECT");
665   MachO::any_relocation_info RE = getRelocation(Rel);
666   Res = getAnyRelocationAddress(RE);
667   return object_error::success;
668 }
669 
670 symbol_iterator
671 MachOObjectFile::getRelocationSymbol(DataRefImpl Rel) const {
672   MachO::any_relocation_info RE = getRelocation(Rel);
673   if (isRelocationScattered(RE))
674     return symbol_end();
675 
676   uint32_t SymbolIdx = getPlainRelocationSymbolNum(RE);
677   bool isExtern = getPlainRelocationExternal(RE);
678   if (!isExtern)
679     return symbol_end();
680 
681   MachO::symtab_command S = getSymtabLoadCommand();
682   unsigned SymbolTableEntrySize = is64Bit() ?
683     sizeof(MachO::nlist_64) :
684     sizeof(MachO::nlist);
685   uint64_t Offset = S.symoff + SymbolIdx * SymbolTableEntrySize;
686   DataRefImpl Sym;
687   Sym.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
688   return symbol_iterator(SymbolRef(Sym, this));
689 }
690 
691 std::error_code MachOObjectFile::getRelocationType(DataRefImpl Rel,
692                                                    uint64_t &Res) const {
693   MachO::any_relocation_info RE = getRelocation(Rel);
694   Res = getAnyRelocationType(RE);
695   return object_error::success;
696 }
697 
698 std::error_code
699 MachOObjectFile::getRelocationTypeName(DataRefImpl Rel,
700                                        SmallVectorImpl<char> &Result) const {
701   StringRef res;
702   uint64_t RType;
703   getRelocationType(Rel, RType);
704 
705   unsigned Arch = this->getArch();
706 
707   switch (Arch) {
708     case Triple::x86: {
709       static const char *const Table[] =  {
710         "GENERIC_RELOC_VANILLA",
711         "GENERIC_RELOC_PAIR",
712         "GENERIC_RELOC_SECTDIFF",
713         "GENERIC_RELOC_PB_LA_PTR",
714         "GENERIC_RELOC_LOCAL_SECTDIFF",
715         "GENERIC_RELOC_TLV" };
716 
717       if (RType > 5)
718         res = "Unknown";
719       else
720         res = Table[RType];
721       break;
722     }
723     case Triple::x86_64: {
724       static const char *const Table[] =  {
725         "X86_64_RELOC_UNSIGNED",
726         "X86_64_RELOC_SIGNED",
727         "X86_64_RELOC_BRANCH",
728         "X86_64_RELOC_GOT_LOAD",
729         "X86_64_RELOC_GOT",
730         "X86_64_RELOC_SUBTRACTOR",
731         "X86_64_RELOC_SIGNED_1",
732         "X86_64_RELOC_SIGNED_2",
733         "X86_64_RELOC_SIGNED_4",
734         "X86_64_RELOC_TLV" };
735 
736       if (RType > 9)
737         res = "Unknown";
738       else
739         res = Table[RType];
740       break;
741     }
742     case Triple::arm: {
743       static const char *const Table[] =  {
744         "ARM_RELOC_VANILLA",
745         "ARM_RELOC_PAIR",
746         "ARM_RELOC_SECTDIFF",
747         "ARM_RELOC_LOCAL_SECTDIFF",
748         "ARM_RELOC_PB_LA_PTR",
749         "ARM_RELOC_BR24",
750         "ARM_THUMB_RELOC_BR22",
751         "ARM_THUMB_32BIT_BRANCH",
752         "ARM_RELOC_HALF",
753         "ARM_RELOC_HALF_SECTDIFF" };
754 
755       if (RType > 9)
756         res = "Unknown";
757       else
758         res = Table[RType];
759       break;
760     }
761     case Triple::aarch64: {
762       static const char *const Table[] = {
763         "ARM64_RELOC_UNSIGNED",           "ARM64_RELOC_SUBTRACTOR",
764         "ARM64_RELOC_BRANCH26",           "ARM64_RELOC_PAGE21",
765         "ARM64_RELOC_PAGEOFF12",          "ARM64_RELOC_GOT_LOAD_PAGE21",
766         "ARM64_RELOC_GOT_LOAD_PAGEOFF12", "ARM64_RELOC_POINTER_TO_GOT",
767         "ARM64_RELOC_TLVP_LOAD_PAGE21",   "ARM64_RELOC_TLVP_LOAD_PAGEOFF12",
768         "ARM64_RELOC_ADDEND"
769       };
770 
771       if (RType >= array_lengthof(Table))
772         res = "Unknown";
773       else
774         res = Table[RType];
775       break;
776     }
777     case Triple::ppc: {
778       static const char *const Table[] =  {
779         "PPC_RELOC_VANILLA",
780         "PPC_RELOC_PAIR",
781         "PPC_RELOC_BR14",
782         "PPC_RELOC_BR24",
783         "PPC_RELOC_HI16",
784         "PPC_RELOC_LO16",
785         "PPC_RELOC_HA16",
786         "PPC_RELOC_LO14",
787         "PPC_RELOC_SECTDIFF",
788         "PPC_RELOC_PB_LA_PTR",
789         "PPC_RELOC_HI16_SECTDIFF",
790         "PPC_RELOC_LO16_SECTDIFF",
791         "PPC_RELOC_HA16_SECTDIFF",
792         "PPC_RELOC_JBSR",
793         "PPC_RELOC_LO14_SECTDIFF",
794         "PPC_RELOC_LOCAL_SECTDIFF" };
795 
796       if (RType > 15)
797         res = "Unknown";
798       else
799         res = Table[RType];
800       break;
801     }
802     case Triple::UnknownArch:
803       res = "Unknown";
804       break;
805   }
806   Result.append(res.begin(), res.end());
807   return object_error::success;
808 }
809 
810 std::error_code
811 MachOObjectFile::getRelocationValueString(DataRefImpl Rel,
812                                           SmallVectorImpl<char> &Result) const {
813   MachO::any_relocation_info RE = getRelocation(Rel);
814 
815   unsigned Arch = this->getArch();
816 
817   std::string fmtbuf;
818   raw_string_ostream fmt(fmtbuf);
819   unsigned Type = this->getAnyRelocationType(RE);
820   bool IsPCRel = this->getAnyRelocationPCRel(RE);
821 
822   // Determine any addends that should be displayed with the relocation.
823   // These require decoding the relocation type, which is triple-specific.
824 
825   // X86_64 has entirely custom relocation types.
826   if (Arch == Triple::x86_64) {
827     bool isPCRel = getAnyRelocationPCRel(RE);
828 
829     switch (Type) {
830       case MachO::X86_64_RELOC_GOT_LOAD:
831       case MachO::X86_64_RELOC_GOT: {
832         printRelocationTargetName(this, RE, fmt);
833         fmt << "@GOT";
834         if (isPCRel) fmt << "PCREL";
835         break;
836       }
837       case MachO::X86_64_RELOC_SUBTRACTOR: {
838         DataRefImpl RelNext = Rel;
839         moveRelocationNext(RelNext);
840         MachO::any_relocation_info RENext = getRelocation(RelNext);
841 
842         // X86_64_RELOC_SUBTRACTOR must be followed by a relocation of type
843         // X86_64_RELOC_UNSIGNED.
844         // NOTE: Scattered relocations don't exist on x86_64.
845         unsigned RType = getAnyRelocationType(RENext);
846         if (RType != MachO::X86_64_RELOC_UNSIGNED)
847           report_fatal_error("Expected X86_64_RELOC_UNSIGNED after "
848                              "X86_64_RELOC_SUBTRACTOR.");
849 
850         // The X86_64_RELOC_UNSIGNED contains the minuend symbol;
851         // X86_64_RELOC_SUBTRACTOR contains the subtrahend.
852         printRelocationTargetName(this, RENext, fmt);
853         fmt << "-";
854         printRelocationTargetName(this, RE, fmt);
855         break;
856       }
857       case MachO::X86_64_RELOC_TLV:
858         printRelocationTargetName(this, RE, fmt);
859         fmt << "@TLV";
860         if (isPCRel) fmt << "P";
861         break;
862       case MachO::X86_64_RELOC_SIGNED_1:
863         printRelocationTargetName(this, RE, fmt);
864         fmt << "-1";
865         break;
866       case MachO::X86_64_RELOC_SIGNED_2:
867         printRelocationTargetName(this, RE, fmt);
868         fmt << "-2";
869         break;
870       case MachO::X86_64_RELOC_SIGNED_4:
871         printRelocationTargetName(this, RE, fmt);
872         fmt << "-4";
873         break;
874       default:
875         printRelocationTargetName(this, RE, fmt);
876         break;
877     }
878   // X86 and ARM share some relocation types in common.
879   } else if (Arch == Triple::x86 || Arch == Triple::arm ||
880              Arch == Triple::ppc) {
881     // Generic relocation types...
882     switch (Type) {
883       case MachO::GENERIC_RELOC_PAIR: // prints no info
884         return object_error::success;
885       case MachO::GENERIC_RELOC_SECTDIFF: {
886         DataRefImpl RelNext = Rel;
887         moveRelocationNext(RelNext);
888         MachO::any_relocation_info RENext = getRelocation(RelNext);
889 
890         // X86 sect diff's must be followed by a relocation of type
891         // GENERIC_RELOC_PAIR.
892         unsigned RType = getAnyRelocationType(RENext);
893 
894         if (RType != MachO::GENERIC_RELOC_PAIR)
895           report_fatal_error("Expected GENERIC_RELOC_PAIR after "
896                              "GENERIC_RELOC_SECTDIFF.");
897 
898         printRelocationTargetName(this, RE, fmt);
899         fmt << "-";
900         printRelocationTargetName(this, RENext, fmt);
901         break;
902       }
903     }
904 
905     if (Arch == Triple::x86 || Arch == Triple::ppc) {
906       switch (Type) {
907         case MachO::GENERIC_RELOC_LOCAL_SECTDIFF: {
908           DataRefImpl RelNext = Rel;
909           moveRelocationNext(RelNext);
910           MachO::any_relocation_info RENext = getRelocation(RelNext);
911 
912           // X86 sect diff's must be followed by a relocation of type
913           // GENERIC_RELOC_PAIR.
914           unsigned RType = getAnyRelocationType(RENext);
915           if (RType != MachO::GENERIC_RELOC_PAIR)
916             report_fatal_error("Expected GENERIC_RELOC_PAIR after "
917                                "GENERIC_RELOC_LOCAL_SECTDIFF.");
918 
919           printRelocationTargetName(this, RE, fmt);
920           fmt << "-";
921           printRelocationTargetName(this, RENext, fmt);
922           break;
923         }
924         case MachO::GENERIC_RELOC_TLV: {
925           printRelocationTargetName(this, RE, fmt);
926           fmt << "@TLV";
927           if (IsPCRel) fmt << "P";
928           break;
929         }
930         default:
931           printRelocationTargetName(this, RE, fmt);
932       }
933     } else { // ARM-specific relocations
934       switch (Type) {
935         case MachO::ARM_RELOC_HALF:
936         case MachO::ARM_RELOC_HALF_SECTDIFF: {
937           // Half relocations steal a bit from the length field to encode
938           // whether this is an upper16 or a lower16 relocation.
939           bool isUpper = getAnyRelocationLength(RE) >> 1;
940 
941           if (isUpper)
942             fmt << ":upper16:(";
943           else
944             fmt << ":lower16:(";
945           printRelocationTargetName(this, RE, fmt);
946 
947           DataRefImpl RelNext = Rel;
948           moveRelocationNext(RelNext);
949           MachO::any_relocation_info RENext = getRelocation(RelNext);
950 
951           // ARM half relocs must be followed by a relocation of type
952           // ARM_RELOC_PAIR.
953           unsigned RType = getAnyRelocationType(RENext);
954           if (RType != MachO::ARM_RELOC_PAIR)
955             report_fatal_error("Expected ARM_RELOC_PAIR after "
956                                "ARM_RELOC_HALF");
957 
958           // NOTE: The half of the target virtual address is stashed in the
959           // address field of the secondary relocation, but we can't reverse
960           // engineer the constant offset from it without decoding the movw/movt
961           // instruction to find the other half in its immediate field.
962 
963           // ARM_RELOC_HALF_SECTDIFF encodes the second section in the
964           // symbol/section pointer of the follow-on relocation.
965           if (Type == MachO::ARM_RELOC_HALF_SECTDIFF) {
966             fmt << "-";
967             printRelocationTargetName(this, RENext, fmt);
968           }
969 
970           fmt << ")";
971           break;
972         }
973         default: {
974           printRelocationTargetName(this, RE, fmt);
975         }
976       }
977     }
978   } else
979     printRelocationTargetName(this, RE, fmt);
980 
981   fmt.flush();
982   Result.append(fmtbuf.begin(), fmtbuf.end());
983   return object_error::success;
984 }
985 
986 std::error_code MachOObjectFile::getRelocationHidden(DataRefImpl Rel,
987                                                      bool &Result) const {
988   unsigned Arch = getArch();
989   uint64_t Type;
990   getRelocationType(Rel, Type);
991 
992   Result = false;
993 
994   // On arches that use the generic relocations, GENERIC_RELOC_PAIR
995   // is always hidden.
996   if (Arch == Triple::x86 || Arch == Triple::arm || Arch == Triple::ppc) {
997     if (Type == MachO::GENERIC_RELOC_PAIR) Result = true;
998   } else if (Arch == Triple::x86_64) {
999     // On x86_64, X86_64_RELOC_UNSIGNED is hidden only when it follows
1000     // an X86_64_RELOC_SUBTRACTOR.
1001     if (Type == MachO::X86_64_RELOC_UNSIGNED && Rel.d.a > 0) {
1002       DataRefImpl RelPrev = Rel;
1003       RelPrev.d.a--;
1004       uint64_t PrevType;
1005       getRelocationType(RelPrev, PrevType);
1006       if (PrevType == MachO::X86_64_RELOC_SUBTRACTOR)
1007         Result = true;
1008     }
1009   }
1010 
1011   return object_error::success;
1012 }
1013 
1014 //
1015 // guessLibraryShortName() is passed a name of a dynamic library and returns a
1016 // guess on what the short name is.  Then name is returned as a substring of the
1017 // StringRef Name passed in.  The name of the dynamic library is recognized as
1018 // a framework if it has one of the two following forms:
1019 //      Foo.framework/Versions/A/Foo
1020 //      Foo.framework/Foo
1021 // Where A and Foo can be any string.  And may contain a trailing suffix
1022 // starting with an underbar.  If the Name is recognized as a framework then
1023 // isFramework is set to true else it is set to false.  If the Name has a
1024 // suffix then Suffix is set to the substring in Name that contains the suffix
1025 // else it is set to a NULL StringRef.
1026 //
1027 // The Name of the dynamic library is recognized as a library name if it has
1028 // one of the two following forms:
1029 //      libFoo.A.dylib
1030 //      libFoo.dylib
1031 // The library may have a suffix trailing the name Foo of the form:
1032 //      libFoo_profile.A.dylib
1033 //      libFoo_profile.dylib
1034 //
1035 // The Name of the dynamic library is also recognized as a library name if it
1036 // has the following form:
1037 //      Foo.qtx
1038 //
1039 // If the Name of the dynamic library is none of the forms above then a NULL
1040 // StringRef is returned.
1041 //
1042 StringRef MachOObjectFile::guessLibraryShortName(StringRef Name,
1043                                                  bool &isFramework,
1044                                                  StringRef &Suffix) {
1045   StringRef Foo, F, DotFramework, V, Dylib, Lib, Dot, Qtx;
1046   size_t a, b, c, d, Idx;
1047 
1048   isFramework = false;
1049   Suffix = StringRef();
1050 
1051   // Pull off the last component and make Foo point to it
1052   a = Name.rfind('/');
1053   if (a == Name.npos || a == 0)
1054     goto guess_library;
1055   Foo = Name.slice(a+1, Name.npos);
1056 
1057   // Look for a suffix starting with a '_'
1058   Idx = Foo.rfind('_');
1059   if (Idx != Foo.npos && Foo.size() >= 2) {
1060     Suffix = Foo.slice(Idx, Foo.npos);
1061     Foo = Foo.slice(0, Idx);
1062   }
1063 
1064   // First look for the form Foo.framework/Foo
1065   b = Name.rfind('/', a);
1066   if (b == Name.npos)
1067     Idx = 0;
1068   else
1069     Idx = b+1;
1070   F = Name.slice(Idx, Idx + Foo.size());
1071   DotFramework = Name.slice(Idx + Foo.size(),
1072                             Idx + Foo.size() + sizeof(".framework/")-1);
1073   if (F == Foo && DotFramework == ".framework/") {
1074     isFramework = true;
1075     return Foo;
1076   }
1077 
1078   // Next look for the form Foo.framework/Versions/A/Foo
1079   if (b == Name.npos)
1080     goto guess_library;
1081   c =  Name.rfind('/', b);
1082   if (c == Name.npos || c == 0)
1083     goto guess_library;
1084   V = Name.slice(c+1, Name.npos);
1085   if (!V.startswith("Versions/"))
1086     goto guess_library;
1087   d =  Name.rfind('/', c);
1088   if (d == Name.npos)
1089     Idx = 0;
1090   else
1091     Idx = d+1;
1092   F = Name.slice(Idx, Idx + Foo.size());
1093   DotFramework = Name.slice(Idx + Foo.size(),
1094                             Idx + Foo.size() + sizeof(".framework/")-1);
1095   if (F == Foo && DotFramework == ".framework/") {
1096     isFramework = true;
1097     return Foo;
1098   }
1099 
1100 guess_library:
1101   // pull off the suffix after the "." and make a point to it
1102   a = Name.rfind('.');
1103   if (a == Name.npos || a == 0)
1104     return StringRef();
1105   Dylib = Name.slice(a, Name.npos);
1106   if (Dylib != ".dylib")
1107     goto guess_qtx;
1108 
1109   // First pull off the version letter for the form Foo.A.dylib if any.
1110   if (a >= 3) {
1111     Dot = Name.slice(a-2, a-1);
1112     if (Dot == ".")
1113       a = a - 2;
1114   }
1115 
1116   b = Name.rfind('/', a);
1117   if (b == Name.npos)
1118     b = 0;
1119   else
1120     b = b+1;
1121   // ignore any suffix after an underbar like Foo_profile.A.dylib
1122   Idx = Name.find('_', b);
1123   if (Idx != Name.npos && Idx != b) {
1124     Lib = Name.slice(b, Idx);
1125     Suffix = Name.slice(Idx, a);
1126   }
1127   else
1128     Lib = Name.slice(b, a);
1129   // There are incorrect library names of the form:
1130   // libATS.A_profile.dylib so check for these.
1131   if (Lib.size() >= 3) {
1132     Dot = Lib.slice(Lib.size()-2, Lib.size()-1);
1133     if (Dot == ".")
1134       Lib = Lib.slice(0, Lib.size()-2);
1135   }
1136   return Lib;
1137 
1138 guess_qtx:
1139   Qtx = Name.slice(a, Name.npos);
1140   if (Qtx != ".qtx")
1141     return StringRef();
1142   b = Name.rfind('/', a);
1143   if (b == Name.npos)
1144     Lib = Name.slice(0, a);
1145   else
1146     Lib = Name.slice(b+1, a);
1147   // There are library names of the form: QT.A.qtx so check for these.
1148   if (Lib.size() >= 3) {
1149     Dot = Lib.slice(Lib.size()-2, Lib.size()-1);
1150     if (Dot == ".")
1151       Lib = Lib.slice(0, Lib.size()-2);
1152   }
1153   return Lib;
1154 }
1155 
1156 // getLibraryShortNameByIndex() is used to get the short name of the library
1157 // for an undefined symbol in a linked Mach-O binary that was linked with the
1158 // normal two-level namespace default (that is MH_TWOLEVEL in the header).
1159 // It is passed the index (0 - based) of the library as translated from
1160 // GET_LIBRARY_ORDINAL (1 - based).
1161 std::error_code MachOObjectFile::getLibraryShortNameByIndex(unsigned Index,
1162                                                             StringRef &Res) {
1163   if (Index >= Libraries.size())
1164     return object_error::parse_failed;
1165 
1166   MachO::dylib_command D =
1167     getStruct<MachO::dylib_command>(this, Libraries[Index]);
1168   if (D.dylib.name >= D.cmdsize)
1169     return object_error::parse_failed;
1170 
1171   // If the cache of LibrariesShortNames is not built up do that first for
1172   // all the Libraries.
1173   if (LibrariesShortNames.size() == 0) {
1174     for (unsigned i = 0; i < Libraries.size(); i++) {
1175       MachO::dylib_command D =
1176         getStruct<MachO::dylib_command>(this, Libraries[i]);
1177       if (D.dylib.name >= D.cmdsize) {
1178         LibrariesShortNames.push_back(StringRef());
1179         continue;
1180       }
1181       const char *P = (const char *)(Libraries[i]) + D.dylib.name;
1182       StringRef Name = StringRef(P);
1183       StringRef Suffix;
1184       bool isFramework;
1185       StringRef shortName = guessLibraryShortName(Name, isFramework, Suffix);
1186       if (shortName == StringRef())
1187         LibrariesShortNames.push_back(Name);
1188       else
1189         LibrariesShortNames.push_back(shortName);
1190     }
1191   }
1192 
1193   Res = LibrariesShortNames[Index];
1194   return object_error::success;
1195 }
1196 
1197 basic_symbol_iterator MachOObjectFile::symbol_begin_impl() const {
1198   return getSymbolByIndex(0);
1199 }
1200 
1201 basic_symbol_iterator MachOObjectFile::symbol_end_impl() const {
1202   DataRefImpl DRI;
1203   if (!SymtabLoadCmd)
1204     return basic_symbol_iterator(SymbolRef(DRI, this));
1205 
1206   MachO::symtab_command Symtab = getSymtabLoadCommand();
1207   unsigned SymbolTableEntrySize = is64Bit() ?
1208     sizeof(MachO::nlist_64) :
1209     sizeof(MachO::nlist);
1210   unsigned Offset = Symtab.symoff +
1211     Symtab.nsyms * SymbolTableEntrySize;
1212   DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
1213   return basic_symbol_iterator(SymbolRef(DRI, this));
1214 }
1215 
1216 basic_symbol_iterator MachOObjectFile::getSymbolByIndex(unsigned Index) const {
1217   DataRefImpl DRI;
1218   if (!SymtabLoadCmd)
1219     return basic_symbol_iterator(SymbolRef(DRI, this));
1220 
1221   MachO::symtab_command Symtab = getSymtabLoadCommand();
1222   assert(Index < Symtab.nsyms && "Requested symbol index is out of range.");
1223   unsigned SymbolTableEntrySize =
1224     is64Bit() ? sizeof(MachO::nlist_64) : sizeof(MachO::nlist);
1225   DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Symtab.symoff));
1226   DRI.p += Index * SymbolTableEntrySize;
1227   return basic_symbol_iterator(SymbolRef(DRI, this));
1228 }
1229 
1230 section_iterator MachOObjectFile::section_begin() const {
1231   DataRefImpl DRI;
1232   return section_iterator(SectionRef(DRI, this));
1233 }
1234 
1235 section_iterator MachOObjectFile::section_end() const {
1236   DataRefImpl DRI;
1237   DRI.d.a = Sections.size();
1238   return section_iterator(SectionRef(DRI, this));
1239 }
1240 
1241 uint8_t MachOObjectFile::getBytesInAddress() const {
1242   return is64Bit() ? 8 : 4;
1243 }
1244 
1245 StringRef MachOObjectFile::getFileFormatName() const {
1246   unsigned CPUType = getCPUType(this);
1247   if (!is64Bit()) {
1248     switch (CPUType) {
1249     case llvm::MachO::CPU_TYPE_I386:
1250       return "Mach-O 32-bit i386";
1251     case llvm::MachO::CPU_TYPE_ARM:
1252       return "Mach-O arm";
1253     case llvm::MachO::CPU_TYPE_POWERPC:
1254       return "Mach-O 32-bit ppc";
1255     default:
1256       assert((CPUType & llvm::MachO::CPU_ARCH_ABI64) == 0 &&
1257              "64-bit object file when we're not 64-bit?");
1258       return "Mach-O 32-bit unknown";
1259     }
1260   }
1261 
1262   // Make sure the cpu type has the correct mask.
1263   assert((CPUType & llvm::MachO::CPU_ARCH_ABI64)
1264          == llvm::MachO::CPU_ARCH_ABI64 &&
1265          "32-bit object file when we're 64-bit?");
1266 
1267   switch (CPUType) {
1268   case llvm::MachO::CPU_TYPE_X86_64:
1269     return "Mach-O 64-bit x86-64";
1270   case llvm::MachO::CPU_TYPE_ARM64:
1271     return "Mach-O arm64";
1272   case llvm::MachO::CPU_TYPE_POWERPC64:
1273     return "Mach-O 64-bit ppc64";
1274   default:
1275     return "Mach-O 64-bit unknown";
1276   }
1277 }
1278 
1279 Triple::ArchType MachOObjectFile::getArch(uint32_t CPUType) {
1280   switch (CPUType) {
1281   case llvm::MachO::CPU_TYPE_I386:
1282     return Triple::x86;
1283   case llvm::MachO::CPU_TYPE_X86_64:
1284     return Triple::x86_64;
1285   case llvm::MachO::CPU_TYPE_ARM:
1286     return Triple::arm;
1287   case llvm::MachO::CPU_TYPE_ARM64:
1288     return Triple::aarch64;
1289   case llvm::MachO::CPU_TYPE_POWERPC:
1290     return Triple::ppc;
1291   case llvm::MachO::CPU_TYPE_POWERPC64:
1292     return Triple::ppc64;
1293   default:
1294     return Triple::UnknownArch;
1295   }
1296 }
1297 
1298 Triple MachOObjectFile::getArch(uint32_t CPUType, uint32_t CPUSubType,
1299                                 const char **McpuDefault) {
1300   if (McpuDefault)
1301     *McpuDefault = nullptr;
1302 
1303   switch (CPUType) {
1304   case MachO::CPU_TYPE_I386:
1305     switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) {
1306     case MachO::CPU_SUBTYPE_I386_ALL:
1307       return Triple("i386-apple-darwin");
1308     default:
1309       return Triple();
1310     }
1311   case MachO::CPU_TYPE_X86_64:
1312     switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) {
1313     case MachO::CPU_SUBTYPE_X86_64_ALL:
1314       return Triple("x86_64-apple-darwin");
1315     case MachO::CPU_SUBTYPE_X86_64_H:
1316       return Triple("x86_64h-apple-darwin");
1317     default:
1318       return Triple();
1319     }
1320   case MachO::CPU_TYPE_ARM:
1321     switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) {
1322     case MachO::CPU_SUBTYPE_ARM_V4T:
1323       return Triple("armv4t-apple-darwin");
1324     case MachO::CPU_SUBTYPE_ARM_V5TEJ:
1325       return Triple("armv5e-apple-darwin");
1326     case MachO::CPU_SUBTYPE_ARM_XSCALE:
1327       return Triple("xscale-apple-darwin");
1328     case MachO::CPU_SUBTYPE_ARM_V6:
1329       return Triple("armv6-apple-darwin");
1330     case MachO::CPU_SUBTYPE_ARM_V6M:
1331       if (McpuDefault)
1332         *McpuDefault = "cortex-m0";
1333       return Triple("armv6m-apple-darwin");
1334     case MachO::CPU_SUBTYPE_ARM_V7:
1335       return Triple("armv7-apple-darwin");
1336     case MachO::CPU_SUBTYPE_ARM_V7EM:
1337       if (McpuDefault)
1338         *McpuDefault = "cortex-m4";
1339       return Triple("armv7em-apple-darwin");
1340     case MachO::CPU_SUBTYPE_ARM_V7K:
1341       return Triple("armv7k-apple-darwin");
1342     case MachO::CPU_SUBTYPE_ARM_V7M:
1343       if (McpuDefault)
1344         *McpuDefault = "cortex-m3";
1345       return Triple("armv7m-apple-darwin");
1346     case MachO::CPU_SUBTYPE_ARM_V7S:
1347       return Triple("armv7s-apple-darwin");
1348     default:
1349       return Triple();
1350     }
1351   case MachO::CPU_TYPE_ARM64:
1352     switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) {
1353     case MachO::CPU_SUBTYPE_ARM64_ALL:
1354       return Triple("arm64-apple-darwin");
1355     default:
1356       return Triple();
1357     }
1358   case MachO::CPU_TYPE_POWERPC:
1359     switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) {
1360     case MachO::CPU_SUBTYPE_POWERPC_ALL:
1361       return Triple("ppc-apple-darwin");
1362     default:
1363       return Triple();
1364     }
1365   case MachO::CPU_TYPE_POWERPC64:
1366     switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) {
1367     case MachO::CPU_SUBTYPE_POWERPC_ALL:
1368       return Triple("ppc64-apple-darwin");
1369     default:
1370       return Triple();
1371     }
1372   default:
1373     return Triple();
1374   }
1375 }
1376 
1377 Triple MachOObjectFile::getThumbArch(uint32_t CPUType, uint32_t CPUSubType,
1378                                      const char **McpuDefault) {
1379   if (McpuDefault)
1380     *McpuDefault = nullptr;
1381 
1382   switch (CPUType) {
1383   case MachO::CPU_TYPE_ARM:
1384     switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) {
1385     case MachO::CPU_SUBTYPE_ARM_V4T:
1386       return Triple("thumbv4t-apple-darwin");
1387     case MachO::CPU_SUBTYPE_ARM_V5TEJ:
1388       return Triple("thumbv5e-apple-darwin");
1389     case MachO::CPU_SUBTYPE_ARM_XSCALE:
1390       return Triple("xscale-apple-darwin");
1391     case MachO::CPU_SUBTYPE_ARM_V6:
1392       return Triple("thumbv6-apple-darwin");
1393     case MachO::CPU_SUBTYPE_ARM_V6M:
1394       if (McpuDefault)
1395         *McpuDefault = "cortex-m0";
1396       return Triple("thumbv6m-apple-darwin");
1397     case MachO::CPU_SUBTYPE_ARM_V7:
1398       return Triple("thumbv7-apple-darwin");
1399     case MachO::CPU_SUBTYPE_ARM_V7EM:
1400       if (McpuDefault)
1401         *McpuDefault = "cortex-m4";
1402       return Triple("thumbv7em-apple-darwin");
1403     case MachO::CPU_SUBTYPE_ARM_V7K:
1404       return Triple("thumbv7k-apple-darwin");
1405     case MachO::CPU_SUBTYPE_ARM_V7M:
1406       if (McpuDefault)
1407         *McpuDefault = "cortex-m3";
1408       return Triple("thumbv7m-apple-darwin");
1409     case MachO::CPU_SUBTYPE_ARM_V7S:
1410       return Triple("thumbv7s-apple-darwin");
1411     default:
1412       return Triple();
1413     }
1414   default:
1415     return Triple();
1416   }
1417 }
1418 
1419 Triple MachOObjectFile::getArch(uint32_t CPUType, uint32_t CPUSubType,
1420                                 const char **McpuDefault,
1421 				Triple *ThumbTriple) {
1422   Triple T = MachOObjectFile::getArch(CPUType, CPUSubType, McpuDefault);
1423   *ThumbTriple = MachOObjectFile::getThumbArch(CPUType, CPUSubType,
1424                                                McpuDefault);
1425   return T;
1426 }
1427 
1428 Triple MachOObjectFile::getHostArch() {
1429   return Triple(sys::getDefaultTargetTriple());
1430 }
1431 
1432 bool MachOObjectFile::isValidArch(StringRef ArchFlag) {
1433   return StringSwitch<bool>(ArchFlag)
1434       .Case("i386", true)
1435       .Case("x86_64", true)
1436       .Case("x86_64h", true)
1437       .Case("armv4t", true)
1438       .Case("arm", true)
1439       .Case("armv5e", true)
1440       .Case("armv6", true)
1441       .Case("armv6m", true)
1442       .Case("armv7em", true)
1443       .Case("armv7k", true)
1444       .Case("armv7m", true)
1445       .Case("armv7s", true)
1446       .Case("arm64", true)
1447       .Case("ppc", true)
1448       .Case("ppc64", true)
1449       .Default(false);
1450 }
1451 
1452 unsigned MachOObjectFile::getArch() const {
1453   return getArch(getCPUType(this));
1454 }
1455 
1456 Triple MachOObjectFile::getArch(const char **McpuDefault,
1457                                 Triple *ThumbTriple) const {
1458   Triple T;
1459   if (is64Bit()) {
1460     MachO::mach_header_64 H_64;
1461     H_64 = getHeader64();
1462     T = MachOObjectFile::getArch(H_64.cputype, H_64.cpusubtype, McpuDefault);
1463     *ThumbTriple = MachOObjectFile::getThumbArch(H_64.cputype, H_64.cpusubtype,
1464                                                  McpuDefault);
1465   } else {
1466     MachO::mach_header H;
1467     H = getHeader();
1468     T = MachOObjectFile::getArch(H.cputype, H.cpusubtype, McpuDefault);
1469     *ThumbTriple = MachOObjectFile::getThumbArch(H.cputype, H.cpusubtype,
1470                                                  McpuDefault);
1471   }
1472   return T;
1473 }
1474 
1475 relocation_iterator MachOObjectFile::section_rel_begin(unsigned Index) const {
1476   DataRefImpl DRI;
1477   DRI.d.a = Index;
1478   return section_rel_begin(DRI);
1479 }
1480 
1481 relocation_iterator MachOObjectFile::section_rel_end(unsigned Index) const {
1482   DataRefImpl DRI;
1483   DRI.d.a = Index;
1484   return section_rel_end(DRI);
1485 }
1486 
1487 dice_iterator MachOObjectFile::begin_dices() const {
1488   DataRefImpl DRI;
1489   if (!DataInCodeLoadCmd)
1490     return dice_iterator(DiceRef(DRI, this));
1491 
1492   MachO::linkedit_data_command DicLC = getDataInCodeLoadCommand();
1493   DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, DicLC.dataoff));
1494   return dice_iterator(DiceRef(DRI, this));
1495 }
1496 
1497 dice_iterator MachOObjectFile::end_dices() const {
1498   DataRefImpl DRI;
1499   if (!DataInCodeLoadCmd)
1500     return dice_iterator(DiceRef(DRI, this));
1501 
1502   MachO::linkedit_data_command DicLC = getDataInCodeLoadCommand();
1503   unsigned Offset = DicLC.dataoff + DicLC.datasize;
1504   DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
1505   return dice_iterator(DiceRef(DRI, this));
1506 }
1507 
1508 StringRef
1509 MachOObjectFile::getSectionFinalSegmentName(DataRefImpl Sec) const {
1510   ArrayRef<char> Raw = getSectionRawFinalSegmentName(Sec);
1511   return parseSegmentOrSectionName(Raw.data());
1512 }
1513 
1514 ArrayRef<char>
1515 MachOObjectFile::getSectionRawName(DataRefImpl Sec) const {
1516   const section_base *Base =
1517     reinterpret_cast<const section_base *>(Sections[Sec.d.a]);
1518   return ArrayRef<char>(Base->sectname);
1519 }
1520 
1521 ArrayRef<char>
1522 MachOObjectFile::getSectionRawFinalSegmentName(DataRefImpl Sec) const {
1523   const section_base *Base =
1524     reinterpret_cast<const section_base *>(Sections[Sec.d.a]);
1525   return ArrayRef<char>(Base->segname);
1526 }
1527 
1528 bool
1529 MachOObjectFile::isRelocationScattered(const MachO::any_relocation_info &RE)
1530   const {
1531   if (getCPUType(this) == MachO::CPU_TYPE_X86_64)
1532     return false;
1533   return getPlainRelocationAddress(RE) & MachO::R_SCATTERED;
1534 }
1535 
1536 unsigned MachOObjectFile::getPlainRelocationSymbolNum(
1537     const MachO::any_relocation_info &RE) const {
1538   if (isLittleEndian())
1539     return RE.r_word1 & 0xffffff;
1540   return RE.r_word1 >> 8;
1541 }
1542 
1543 bool MachOObjectFile::getPlainRelocationExternal(
1544     const MachO::any_relocation_info &RE) const {
1545   if (isLittleEndian())
1546     return (RE.r_word1 >> 27) & 1;
1547   return (RE.r_word1 >> 4) & 1;
1548 }
1549 
1550 bool MachOObjectFile::getScatteredRelocationScattered(
1551     const MachO::any_relocation_info &RE) const {
1552   return RE.r_word0 >> 31;
1553 }
1554 
1555 uint32_t MachOObjectFile::getScatteredRelocationValue(
1556     const MachO::any_relocation_info &RE) const {
1557   return RE.r_word1;
1558 }
1559 
1560 unsigned MachOObjectFile::getAnyRelocationAddress(
1561     const MachO::any_relocation_info &RE) const {
1562   if (isRelocationScattered(RE))
1563     return getScatteredRelocationAddress(RE);
1564   return getPlainRelocationAddress(RE);
1565 }
1566 
1567 unsigned MachOObjectFile::getAnyRelocationPCRel(
1568     const MachO::any_relocation_info &RE) const {
1569   if (isRelocationScattered(RE))
1570     return getScatteredRelocationPCRel(this, RE);
1571   return getPlainRelocationPCRel(this, RE);
1572 }
1573 
1574 unsigned MachOObjectFile::getAnyRelocationLength(
1575     const MachO::any_relocation_info &RE) const {
1576   if (isRelocationScattered(RE))
1577     return getScatteredRelocationLength(RE);
1578   return getPlainRelocationLength(this, RE);
1579 }
1580 
1581 unsigned
1582 MachOObjectFile::getAnyRelocationType(
1583                                    const MachO::any_relocation_info &RE) const {
1584   if (isRelocationScattered(RE))
1585     return getScatteredRelocationType(RE);
1586   return getPlainRelocationType(this, RE);
1587 }
1588 
1589 SectionRef
1590 MachOObjectFile::getRelocationSection(
1591                                    const MachO::any_relocation_info &RE) const {
1592   if (isRelocationScattered(RE) || getPlainRelocationExternal(RE))
1593     return *section_end();
1594   unsigned SecNum = getPlainRelocationSymbolNum(RE) - 1;
1595   DataRefImpl DRI;
1596   DRI.d.a = SecNum;
1597   return SectionRef(DRI, this);
1598 }
1599 
1600 MachOObjectFile::LoadCommandInfo
1601 MachOObjectFile::getFirstLoadCommandInfo() const {
1602   MachOObjectFile::LoadCommandInfo Load;
1603 
1604   unsigned HeaderSize = is64Bit() ? sizeof(MachO::mach_header_64) :
1605                                     sizeof(MachO::mach_header);
1606   Load.Ptr = getPtr(this, HeaderSize);
1607   Load.C = getStruct<MachO::load_command>(this, Load.Ptr);
1608   return Load;
1609 }
1610 
1611 MachOObjectFile::LoadCommandInfo
1612 MachOObjectFile::getNextLoadCommandInfo(const LoadCommandInfo &L) const {
1613   MachOObjectFile::LoadCommandInfo Next;
1614   Next.Ptr = L.Ptr + L.C.cmdsize;
1615   Next.C = getStruct<MachO::load_command>(this, Next.Ptr);
1616   return Next;
1617 }
1618 
1619 MachO::section MachOObjectFile::getSection(DataRefImpl DRI) const {
1620   return getStruct<MachO::section>(this, Sections[DRI.d.a]);
1621 }
1622 
1623 MachO::section_64 MachOObjectFile::getSection64(DataRefImpl DRI) const {
1624   return getStruct<MachO::section_64>(this, Sections[DRI.d.a]);
1625 }
1626 
1627 MachO::section MachOObjectFile::getSection(const LoadCommandInfo &L,
1628                                            unsigned Index) const {
1629   const char *Sec = getSectionPtr(this, L, Index);
1630   return getStruct<MachO::section>(this, Sec);
1631 }
1632 
1633 MachO::section_64 MachOObjectFile::getSection64(const LoadCommandInfo &L,
1634                                                 unsigned Index) const {
1635   const char *Sec = getSectionPtr(this, L, Index);
1636   return getStruct<MachO::section_64>(this, Sec);
1637 }
1638 
1639 MachO::nlist
1640 MachOObjectFile::getSymbolTableEntry(DataRefImpl DRI) const {
1641   const char *P = reinterpret_cast<const char *>(DRI.p);
1642   return getStruct<MachO::nlist>(this, P);
1643 }
1644 
1645 MachO::nlist_64
1646 MachOObjectFile::getSymbol64TableEntry(DataRefImpl DRI) const {
1647   const char *P = reinterpret_cast<const char *>(DRI.p);
1648   return getStruct<MachO::nlist_64>(this, P);
1649 }
1650 
1651 MachO::linkedit_data_command
1652 MachOObjectFile::getLinkeditDataLoadCommand(const LoadCommandInfo &L) const {
1653   return getStruct<MachO::linkedit_data_command>(this, L.Ptr);
1654 }
1655 
1656 MachO::segment_command
1657 MachOObjectFile::getSegmentLoadCommand(const LoadCommandInfo &L) const {
1658   return getStruct<MachO::segment_command>(this, L.Ptr);
1659 }
1660 
1661 MachO::segment_command_64
1662 MachOObjectFile::getSegment64LoadCommand(const LoadCommandInfo &L) const {
1663   return getStruct<MachO::segment_command_64>(this, L.Ptr);
1664 }
1665 
1666 MachO::linker_options_command
1667 MachOObjectFile::getLinkerOptionsLoadCommand(const LoadCommandInfo &L) const {
1668   return getStruct<MachO::linker_options_command>(this, L.Ptr);
1669 }
1670 
1671 MachO::version_min_command
1672 MachOObjectFile::getVersionMinLoadCommand(const LoadCommandInfo &L) const {
1673   return getStruct<MachO::version_min_command>(this, L.Ptr);
1674 }
1675 
1676 MachO::dylib_command
1677 MachOObjectFile::getDylibIDLoadCommand(const LoadCommandInfo &L) const {
1678   return getStruct<MachO::dylib_command>(this, L.Ptr);
1679 }
1680 
1681 
1682 MachO::any_relocation_info
1683 MachOObjectFile::getRelocation(DataRefImpl Rel) const {
1684   DataRefImpl Sec;
1685   Sec.d.a = Rel.d.a;
1686   uint32_t Offset;
1687   if (is64Bit()) {
1688     MachO::section_64 Sect = getSection64(Sec);
1689     Offset = Sect.reloff;
1690   } else {
1691     MachO::section Sect = getSection(Sec);
1692     Offset = Sect.reloff;
1693   }
1694 
1695   auto P = reinterpret_cast<const MachO::any_relocation_info *>(
1696       getPtr(this, Offset)) + Rel.d.b;
1697   return getStruct<MachO::any_relocation_info>(
1698       this, reinterpret_cast<const char *>(P));
1699 }
1700 
1701 MachO::data_in_code_entry
1702 MachOObjectFile::getDice(DataRefImpl Rel) const {
1703   const char *P = reinterpret_cast<const char *>(Rel.p);
1704   return getStruct<MachO::data_in_code_entry>(this, P);
1705 }
1706 
1707 MachO::mach_header MachOObjectFile::getHeader() const {
1708   return getStruct<MachO::mach_header>(this, getPtr(this, 0));
1709 }
1710 
1711 MachO::mach_header_64 MachOObjectFile::getHeader64() const {
1712   return getStruct<MachO::mach_header_64>(this, getPtr(this, 0));
1713 }
1714 
1715 uint32_t MachOObjectFile::getIndirectSymbolTableEntry(
1716                                              const MachO::dysymtab_command &DLC,
1717                                              unsigned Index) const {
1718   uint64_t Offset = DLC.indirectsymoff + Index * sizeof(uint32_t);
1719   return getStruct<uint32_t>(this, getPtr(this, Offset));
1720 }
1721 
1722 MachO::data_in_code_entry
1723 MachOObjectFile::getDataInCodeTableEntry(uint32_t DataOffset,
1724                                          unsigned Index) const {
1725   uint64_t Offset = DataOffset + Index * sizeof(MachO::data_in_code_entry);
1726   return getStruct<MachO::data_in_code_entry>(this, getPtr(this, Offset));
1727 }
1728 
1729 MachO::symtab_command MachOObjectFile::getSymtabLoadCommand() const {
1730   return getStruct<MachO::symtab_command>(this, SymtabLoadCmd);
1731 }
1732 
1733 MachO::dysymtab_command MachOObjectFile::getDysymtabLoadCommand() const {
1734   return getStruct<MachO::dysymtab_command>(this, DysymtabLoadCmd);
1735 }
1736 
1737 MachO::linkedit_data_command
1738 MachOObjectFile::getDataInCodeLoadCommand() const {
1739   if (DataInCodeLoadCmd)
1740     return getStruct<MachO::linkedit_data_command>(this, DataInCodeLoadCmd);
1741 
1742   // If there is no DataInCodeLoadCmd return a load command with zero'ed fields.
1743   MachO::linkedit_data_command Cmd;
1744   Cmd.cmd = MachO::LC_DATA_IN_CODE;
1745   Cmd.cmdsize = sizeof(MachO::linkedit_data_command);
1746   Cmd.dataoff = 0;
1747   Cmd.datasize = 0;
1748   return Cmd;
1749 }
1750 
1751 StringRef MachOObjectFile::getStringTableData() const {
1752   MachO::symtab_command S = getSymtabLoadCommand();
1753   return getData().substr(S.stroff, S.strsize);
1754 }
1755 
1756 bool MachOObjectFile::is64Bit() const {
1757   return getType() == getMachOType(false, true) ||
1758     getType() == getMachOType(true, true);
1759 }
1760 
1761 void MachOObjectFile::ReadULEB128s(uint64_t Index,
1762                                    SmallVectorImpl<uint64_t> &Out) const {
1763   DataExtractor extractor(ObjectFile::getData(), true, 0);
1764 
1765   uint32_t offset = Index;
1766   uint64_t data = 0;
1767   while (uint64_t delta = extractor.getULEB128(&offset)) {
1768     data += delta;
1769     Out.push_back(data);
1770   }
1771 }
1772 
1773 bool MachOObjectFile::isRelocatableObject() const {
1774   return getHeader().filetype == MachO::MH_OBJECT;
1775 }
1776 
1777 ErrorOr<std::unique_ptr<MachOObjectFile>>
1778 ObjectFile::createMachOObjectFile(MemoryBufferRef Buffer) {
1779   StringRef Magic = Buffer.getBuffer().slice(0, 4);
1780   std::error_code EC;
1781   std::unique_ptr<MachOObjectFile> Ret;
1782   if (Magic == "\xFE\xED\xFA\xCE")
1783     Ret.reset(new MachOObjectFile(Buffer, false, false, EC));
1784   else if (Magic == "\xCE\xFA\xED\xFE")
1785     Ret.reset(new MachOObjectFile(Buffer, true, false, EC));
1786   else if (Magic == "\xFE\xED\xFA\xCF")
1787     Ret.reset(new MachOObjectFile(Buffer, false, true, EC));
1788   else if (Magic == "\xCF\xFA\xED\xFE")
1789     Ret.reset(new MachOObjectFile(Buffer, true, true, EC));
1790   else
1791     return object_error::parse_failed;
1792 
1793   if (EC)
1794     return EC;
1795   return std::move(Ret);
1796 }
1797 
1798