1 //===- MachOObjectFile.cpp - Mach-O object file binding ---------*- C++ -*-===//
3 // The LLVM Compiler Infrastructure
5 // This file is distributed under the University of Illinois Open Source
6 // License. See LICENSE.TXT for details.
8 //===----------------------------------------------------------------------===//
10 // This file defines the MachOObjectFile class, which binds the MachOObject
11 // class to the generic ObjectFile wrapper.
13 //===----------------------------------------------------------------------===//
15 #include "llvm/Object/MachO.h"
16 #include "llvm/ADT/Triple.h"
17 #include "llvm/Support/DataExtractor.h"
18 #include "llvm/Support/Format.h"
19 #include "llvm/Support/Host.h"
20 #include "llvm/Support/MemoryBuffer.h"
21 #include "llvm/Support/raw_ostream.h"
27 using namespace object;
45 static void SwapValue(T &Value) {
46 Value = sys::SwapByteOrder(Value);
50 static void SwapStruct(T &Value);
53 void SwapStruct(MachO::any_relocation_info &H) {
59 void SwapStruct(MachO::load_command &L) {
65 void SwapStruct(nlist_base &S) {
71 void SwapStruct(MachO::section &S) {
79 SwapValue(S.reserved1);
80 SwapValue(S.reserved2);
84 void SwapStruct(MachO::section_64 &S) {
92 SwapValue(S.reserved1);
93 SwapValue(S.reserved2);
94 SwapValue(S.reserved3);
98 void SwapStruct(MachO::nlist &S) {
101 SwapValue(S.n_value);
105 void SwapStruct(MachO::nlist_64 &S) {
108 SwapValue(S.n_value);
112 void SwapStruct(MachO::mach_header &H) {
114 SwapValue(H.cputype);
115 SwapValue(H.cpusubtype);
116 SwapValue(H.filetype);
118 SwapValue(H.sizeofcmds);
123 void SwapStruct(MachO::mach_header_64 &H) {
125 SwapValue(H.cputype);
126 SwapValue(H.cpusubtype);
127 SwapValue(H.filetype);
129 SwapValue(H.sizeofcmds);
131 SwapValue(H.reserved);
135 void SwapStruct(MachO::symtab_command &C) {
137 SwapValue(C.cmdsize);
141 SwapValue(C.strsize);
145 void SwapStruct(MachO::dysymtab_command &C) {
147 SwapValue(C.cmdsize);
148 SwapValue(C.ilocalsym);
149 SwapValue(C.nlocalsym);
150 SwapValue(C.iextdefsym);
151 SwapValue(C.nextdefsym);
152 SwapValue(C.iundefsym);
153 SwapValue(C.nundefsym);
156 SwapValue(C.modtaboff);
157 SwapValue(C.nmodtab);
158 SwapValue(C.extrefsymoff);
159 SwapValue(C.nextrefsyms);
160 SwapValue(C.indirectsymoff);
161 SwapValue(C.nindirectsyms);
162 SwapValue(C.extreloff);
163 SwapValue(C.nextrel);
164 SwapValue(C.locreloff);
165 SwapValue(C.nlocrel);
169 void SwapStruct(MachO::linkedit_data_command &C) {
171 SwapValue(C.cmdsize);
172 SwapValue(C.dataoff);
173 SwapValue(C.datasize);
177 void SwapStruct(MachO::segment_command &C) {
179 SwapValue(C.cmdsize);
182 SwapValue(C.fileoff);
183 SwapValue(C.filesize);
184 SwapValue(C.maxprot);
185 SwapValue(C.initprot);
191 void SwapStruct(MachO::segment_command_64 &C) {
193 SwapValue(C.cmdsize);
196 SwapValue(C.fileoff);
197 SwapValue(C.filesize);
198 SwapValue(C.maxprot);
199 SwapValue(C.initprot);
205 void SwapStruct(uint32_t &C) {
210 void SwapStruct(MachO::linker_options_command &C) {
212 SwapValue(C.cmdsize);
217 void SwapStruct(MachO::data_in_code_entry &C) {
224 T getStruct(const MachOObjectFile *O, const char *P) {
226 memcpy(&Cmd, P, sizeof(T));
227 if (O->isLittleEndian() != sys::IsLittleEndianHost)
233 getSegmentLoadCommandNumSections(const MachOObjectFile *O,
234 const MachOObjectFile::LoadCommandInfo &L) {
236 MachO::segment_command_64 S = O->getSegment64LoadCommand(L);
239 MachO::segment_command S = O->getSegmentLoadCommand(L);
244 getSectionPtr(const MachOObjectFile *O, MachOObjectFile::LoadCommandInfo L,
246 uintptr_t CommandAddr = reinterpret_cast<uintptr_t>(L.Ptr);
248 bool Is64 = O->is64Bit();
249 unsigned SegmentLoadSize = Is64 ? sizeof(MachO::segment_command_64) :
250 sizeof(MachO::segment_command);
251 unsigned SectionSize = Is64 ? sizeof(MachO::section_64) :
252 sizeof(MachO::section);
254 uintptr_t SectionAddr = CommandAddr + SegmentLoadSize + Sec * SectionSize;
255 return reinterpret_cast<const char*>(SectionAddr);
258 static const char *getPtr(const MachOObjectFile *O, size_t Offset) {
259 return O->getData().substr(Offset, 1).data();
263 getSymbolTableEntryBase(const MachOObjectFile *O, DataRefImpl DRI) {
264 const char *P = reinterpret_cast<const char *>(DRI.p);
265 return getStruct<nlist_base>(O, P);
268 static StringRef parseSegmentOrSectionName(const char *P) {
272 // Not null terminated, so this is a 16 char string.
273 return StringRef(P, 16);
276 // Helper to advance a section or symbol iterator multiple increments at a time.
278 static error_code advance(T &it, size_t Val) {
287 static void advanceTo(T &it, size_t Val) {
288 if (error_code ec = advance(it, Val))
289 report_fatal_error(ec.message());
292 static unsigned getCPUType(const MachOObjectFile *O) {
293 return O->getHeader().cputype;
296 static void printRelocationTargetName(const MachOObjectFile *O,
297 const MachO::any_relocation_info &RE,
298 raw_string_ostream &fmt) {
299 bool IsScattered = O->isRelocationScattered(RE);
301 // Target of a scattered relocation is an address. In the interest of
302 // generating pretty output, scan through the symbol table looking for a
303 // symbol that aligns with that address. If we find one, print it.
304 // Otherwise, we just print the hex address of the target.
306 uint32_t Val = O->getPlainRelocationSymbolNum(RE);
309 for (symbol_iterator SI = O->begin_symbols(), SE = O->end_symbols();
310 SI != SE; SI.increment(ec)) {
311 if (ec) report_fatal_error(ec.message());
316 if ((ec = SI->getAddress(Addr)))
317 report_fatal_error(ec.message());
318 if (Addr != Val) continue;
319 if ((ec = SI->getName(Name)))
320 report_fatal_error(ec.message());
325 // If we couldn't find a symbol that this relocation refers to, try
326 // to find a section beginning instead.
327 for (section_iterator SI = O->begin_sections(), SE = O->end_sections();
328 SI != SE; SI.increment(ec)) {
329 if (ec) report_fatal_error(ec.message());
334 if ((ec = SI->getAddress(Addr)))
335 report_fatal_error(ec.message());
336 if (Addr != Val) continue;
337 if ((ec = SI->getName(Name)))
338 report_fatal_error(ec.message());
343 fmt << format("0x%x", Val);
348 bool isExtern = O->getPlainRelocationExternal(RE);
349 uint64_t Val = O->getPlainRelocationSymbolNum(RE);
352 symbol_iterator SI = O->begin_symbols();
356 section_iterator SI = O->begin_sections();
357 // Adjust for the fact that sections are 1-indexed.
358 advanceTo(SI, Val - 1);
366 getPlainRelocationAddress(const MachO::any_relocation_info &RE) {
371 getScatteredRelocationAddress(const MachO::any_relocation_info &RE) {
372 return RE.r_word0 & 0xffffff;
375 static bool getPlainRelocationPCRel(const MachOObjectFile *O,
376 const MachO::any_relocation_info &RE) {
377 if (O->isLittleEndian())
378 return (RE.r_word1 >> 24) & 1;
379 return (RE.r_word1 >> 7) & 1;
383 getScatteredRelocationPCRel(const MachOObjectFile *O,
384 const MachO::any_relocation_info &RE) {
385 return (RE.r_word0 >> 30) & 1;
388 static unsigned getPlainRelocationLength(const MachOObjectFile *O,
389 const MachO::any_relocation_info &RE) {
390 if (O->isLittleEndian())
391 return (RE.r_word1 >> 25) & 3;
392 return (RE.r_word1 >> 5) & 3;
396 getScatteredRelocationLength(const MachO::any_relocation_info &RE) {
397 return (RE.r_word0 >> 28) & 3;
400 static unsigned getPlainRelocationType(const MachOObjectFile *O,
401 const MachO::any_relocation_info &RE) {
402 if (O->isLittleEndian())
403 return RE.r_word1 >> 28;
404 return RE.r_word1 & 0xf;
408 getScatteredRelocationType(const MachO::any_relocation_info &RE) {
409 return (RE.r_word0 >> 24) & 0xf;
412 static uint32_t getSectionFlags(const MachOObjectFile *O,
415 MachO::section_64 Sect = O->getSection64(Sec);
418 MachO::section Sect = O->getSection(Sec);
422 MachOObjectFile::MachOObjectFile(MemoryBuffer *Object,
423 bool IsLittleEndian, bool Is64bits,
425 : ObjectFile(getMachOType(IsLittleEndian, Is64bits), Object),
426 SymtabLoadCmd(NULL), DysymtabLoadCmd(NULL), DataInCodeLoadCmd(NULL) {
427 uint32_t LoadCommandCount = this->getHeader().ncmds;
428 MachO::LoadCommandType SegmentLoadType = is64Bit() ?
429 MachO::LC_SEGMENT_64 : MachO::LC_SEGMENT;
431 MachOObjectFile::LoadCommandInfo Load = getFirstLoadCommandInfo();
432 for (unsigned I = 0; ; ++I) {
433 if (Load.C.cmd == MachO::LC_SYMTAB) {
434 assert(!SymtabLoadCmd && "Multiple symbol tables");
435 SymtabLoadCmd = Load.Ptr;
436 } else if (Load.C.cmd == MachO::LC_DYSYMTAB) {
437 assert(!DysymtabLoadCmd && "Multiple dynamic symbol tables");
438 DysymtabLoadCmd = Load.Ptr;
439 } else if (Load.C.cmd == MachO::LC_DATA_IN_CODE) {
440 assert(!DataInCodeLoadCmd && "Multiple data in code tables");
441 DataInCodeLoadCmd = Load.Ptr;
442 } else if (Load.C.cmd == SegmentLoadType) {
443 uint32_t NumSections = getSegmentLoadCommandNumSections(this, Load);
444 for (unsigned J = 0; J < NumSections; ++J) {
445 const char *Sec = getSectionPtr(this, Load, J);
446 Sections.push_back(Sec);
450 if (I == LoadCommandCount - 1)
453 Load = getNextLoadCommandInfo(Load);
457 error_code MachOObjectFile::getSymbolNext(DataRefImpl Symb,
458 SymbolRef &Res) const {
459 unsigned SymbolTableEntrySize = is64Bit() ?
460 sizeof(MachO::nlist_64) :
461 sizeof(MachO::nlist);
462 Symb.p += SymbolTableEntrySize;
463 Res = SymbolRef(Symb, this);
464 return object_error::success;
467 error_code MachOObjectFile::getSymbolName(DataRefImpl Symb,
468 StringRef &Res) const {
469 StringRef StringTable = getStringTableData();
470 nlist_base Entry = getSymbolTableEntryBase(this, Symb);
471 const char *Start = &StringTable.data()[Entry.n_strx];
472 Res = StringRef(Start);
473 return object_error::success;
476 error_code MachOObjectFile::getSymbolAddress(DataRefImpl Symb,
477 uint64_t &Res) const {
479 MachO::nlist_64 Entry = getSymbol64TableEntry(Symb);
482 MachO::nlist Entry = getSymbolTableEntry(Symb);
485 return object_error::success;
489 MachOObjectFile::getSymbolFileOffset(DataRefImpl Symb,
490 uint64_t &Res) const {
491 nlist_base Entry = getSymbolTableEntryBase(this, Symb);
492 getSymbolAddress(Symb, Res);
496 SecRel.d.a = Entry.n_sect-1;
498 MachO::section_64 Sec = getSection64(SecRel);
499 Delta = Sec.offset - Sec.addr;
501 MachO::section Sec = getSection(SecRel);
502 Delta = Sec.offset - Sec.addr;
508 return object_error::success;
511 error_code MachOObjectFile::getSymbolAlignment(DataRefImpl DRI,
512 uint32_t &Result) const {
514 this->getSymbolFlags(DRI, flags);
515 if (flags & SymbolRef::SF_Common) {
516 nlist_base Entry = getSymbolTableEntryBase(this, DRI);
517 Result = 1 << MachO::GET_COMM_ALIGN(Entry.n_desc);
521 return object_error::success;
524 error_code MachOObjectFile::getSymbolSize(DataRefImpl DRI,
525 uint64_t &Result) const {
526 uint64_t BeginOffset;
527 uint64_t EndOffset = 0;
528 uint8_t SectionIndex;
530 nlist_base Entry = getSymbolTableEntryBase(this, DRI);
532 getSymbolAddress(DRI, Value);
536 SectionIndex = Entry.n_sect;
538 uint32_t flags = SymbolRef::SF_None;
539 this->getSymbolFlags(DRI, flags);
540 if (flags & SymbolRef::SF_Common)
543 Result = UnknownAddressOrSize;
544 return object_error::success;
546 // Unfortunately symbols are unsorted so we need to touch all
547 // symbols from load command
549 for (symbol_iterator I = begin_symbols(), E = end_symbols(); I != E;
551 DataRefImpl DRI = I->getRawDataRefImpl();
552 Entry = getSymbolTableEntryBase(this, DRI);
553 getSymbolAddress(DRI, Value);
554 if (Entry.n_sect == SectionIndex && Value > BeginOffset)
555 if (!EndOffset || Value < EndOffset)
561 Sec.d.a = SectionIndex-1;
562 getSectionSize(Sec, Size);
563 getSectionAddress(Sec, EndOffset);
566 Result = EndOffset - BeginOffset;
567 return object_error::success;
570 error_code MachOObjectFile::getSymbolType(DataRefImpl Symb,
571 SymbolRef::Type &Res) const {
572 nlist_base Entry = getSymbolTableEntryBase(this, Symb);
573 uint8_t n_type = Entry.n_type;
575 Res = SymbolRef::ST_Other;
577 // If this is a STAB debugging symbol, we can do nothing more.
578 if (n_type & MachO::N_STAB) {
579 Res = SymbolRef::ST_Debug;
580 return object_error::success;
583 switch (n_type & MachO::N_TYPE) {
585 Res = SymbolRef::ST_Unknown;
588 Res = SymbolRef::ST_Function;
591 return object_error::success;
594 error_code MachOObjectFile::getSymbolFlags(DataRefImpl DRI,
595 uint32_t &Result) const {
596 nlist_base Entry = getSymbolTableEntryBase(this, DRI);
598 uint8_t MachOType = Entry.n_type;
599 uint16_t MachOFlags = Entry.n_desc;
601 // TODO: Correctly set SF_ThreadLocal
602 Result = SymbolRef::SF_None;
604 if ((MachOType & MachO::N_TYPE) == MachO::N_UNDF)
605 Result |= SymbolRef::SF_Undefined;
607 if (MachOType & MachO::N_STAB)
608 Result |= SymbolRef::SF_FormatSpecific;
610 if (MachOType & MachO::N_EXT) {
611 Result |= SymbolRef::SF_Global;
612 if ((MachOType & MachO::N_TYPE) == MachO::N_UNDF) {
614 getSymbolAddress(DRI, Value);
616 Result |= SymbolRef::SF_Common;
620 if (MachOFlags & (MachO::N_WEAK_REF | MachO::N_WEAK_DEF))
621 Result |= SymbolRef::SF_Weak;
623 if ((MachOType & MachO::N_TYPE) == MachO::N_ABS)
624 Result |= SymbolRef::SF_Absolute;
626 return object_error::success;
630 MachOObjectFile::getSymbolSection(DataRefImpl Symb,
631 section_iterator &Res) const {
632 nlist_base Entry = getSymbolTableEntryBase(this, Symb);
633 uint8_t index = Entry.n_sect;
636 Res = end_sections();
640 Res = section_iterator(SectionRef(DRI, this));
643 return object_error::success;
646 error_code MachOObjectFile::getSymbolValue(DataRefImpl Symb,
647 uint64_t &Val) const {
648 report_fatal_error("getSymbolValue unimplemented in MachOObjectFile");
651 error_code MachOObjectFile::getSectionNext(DataRefImpl Sec,
652 SectionRef &Res) const {
654 Res = SectionRef(Sec, this);
655 return object_error::success;
659 MachOObjectFile::getSectionName(DataRefImpl Sec, StringRef &Result) const {
660 ArrayRef<char> Raw = getSectionRawName(Sec);
661 Result = parseSegmentOrSectionName(Raw.data());
662 return object_error::success;
666 MachOObjectFile::getSectionAddress(DataRefImpl Sec, uint64_t &Res) const {
668 MachO::section_64 Sect = getSection64(Sec);
671 MachO::section Sect = getSection(Sec);
674 return object_error::success;
678 MachOObjectFile::getSectionSize(DataRefImpl Sec, uint64_t &Res) const {
680 MachO::section_64 Sect = getSection64(Sec);
683 MachO::section Sect = getSection(Sec);
687 return object_error::success;
691 MachOObjectFile::getSectionContents(DataRefImpl Sec, StringRef &Res) const {
696 MachO::section_64 Sect = getSection64(Sec);
697 Offset = Sect.offset;
700 MachO::section Sect = getSection(Sec);
701 Offset = Sect.offset;
705 Res = this->getData().substr(Offset, Size);
706 return object_error::success;
710 MachOObjectFile::getSectionAlignment(DataRefImpl Sec, uint64_t &Res) const {
713 MachO::section_64 Sect = getSection64(Sec);
716 MachO::section Sect = getSection(Sec);
720 Res = uint64_t(1) << Align;
721 return object_error::success;
725 MachOObjectFile::isSectionText(DataRefImpl Sec, bool &Res) const {
726 uint32_t Flags = getSectionFlags(this, Sec);
727 Res = Flags & MachO::S_ATTR_PURE_INSTRUCTIONS;
728 return object_error::success;
731 error_code MachOObjectFile::isSectionData(DataRefImpl DRI, bool &Result) const {
732 // FIXME: Unimplemented.
734 return object_error::success;
737 error_code MachOObjectFile::isSectionBSS(DataRefImpl DRI, bool &Result) const {
738 // FIXME: Unimplemented.
740 return object_error::success;
744 MachOObjectFile::isSectionRequiredForExecution(DataRefImpl Sec,
745 bool &Result) const {
746 // FIXME: Unimplemented.
748 return object_error::success;
751 error_code MachOObjectFile::isSectionVirtual(DataRefImpl Sec,
752 bool &Result) const {
753 // FIXME: Unimplemented.
755 return object_error::success;
759 MachOObjectFile::isSectionZeroInit(DataRefImpl Sec, bool &Res) const {
760 uint32_t Flags = getSectionFlags(this, Sec);
761 unsigned SectionType = Flags & MachO::SECTION_TYPE;
762 Res = SectionType == MachO::S_ZEROFILL ||
763 SectionType == MachO::S_GB_ZEROFILL;
764 return object_error::success;
767 error_code MachOObjectFile::isSectionReadOnlyData(DataRefImpl Sec,
768 bool &Result) const {
769 // Consider using the code from isSectionText to look for __const sections.
770 // Alternately, emit S_ATTR_PURE_INSTRUCTIONS and/or S_ATTR_SOME_INSTRUCTIONS
771 // to use section attributes to distinguish code from data.
773 // FIXME: Unimplemented.
775 return object_error::success;
779 MachOObjectFile::sectionContainsSymbol(DataRefImpl Sec, DataRefImpl Symb,
780 bool &Result) const {
782 this->getSymbolType(Symb, ST);
783 if (ST == SymbolRef::ST_Unknown) {
785 return object_error::success;
788 uint64_t SectBegin, SectEnd;
789 getSectionAddress(Sec, SectBegin);
790 getSectionSize(Sec, SectEnd);
791 SectEnd += SectBegin;
794 getSymbolAddress(Symb, SymAddr);
795 Result = (SymAddr >= SectBegin) && (SymAddr < SectEnd);
797 return object_error::success;
800 relocation_iterator MachOObjectFile::section_rel_begin(DataRefImpl Sec) const {
803 MachO::section_64 Sect = getSection64(Sec);
804 Offset = Sect.reloff;
806 MachO::section Sect = getSection(Sec);
807 Offset = Sect.reloff;
811 Ret.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
812 return relocation_iterator(RelocationRef(Ret, this));
816 MachOObjectFile::section_rel_end(DataRefImpl Sec) const {
820 MachO::section_64 Sect = getSection64(Sec);
821 Offset = Sect.reloff;
824 MachO::section Sect = getSection(Sec);
825 Offset = Sect.reloff;
829 const MachO::any_relocation_info *P =
830 reinterpret_cast<const MachO::any_relocation_info *>(getPtr(this, Offset));
833 Ret.p = reinterpret_cast<uintptr_t>(P + Num);
834 return relocation_iterator(RelocationRef(Ret, this));
837 error_code MachOObjectFile::getRelocationNext(DataRefImpl Rel,
838 RelocationRef &Res) const {
839 const MachO::any_relocation_info *P =
840 reinterpret_cast<const MachO::any_relocation_info *>(Rel.p);
841 Rel.p = reinterpret_cast<uintptr_t>(P + 1);
842 Res = RelocationRef(Rel, this);
843 return object_error::success;
847 MachOObjectFile::getRelocationAddress(DataRefImpl Rel, uint64_t &Res) const {
848 report_fatal_error("getRelocationAddress not implemented in MachOObjectFile");
851 error_code MachOObjectFile::getRelocationOffset(DataRefImpl Rel,
852 uint64_t &Res) const {
853 MachO::any_relocation_info RE = getRelocation(Rel);
854 Res = getAnyRelocationAddress(RE);
855 return object_error::success;
859 MachOObjectFile::getRelocationSymbol(DataRefImpl Rel) const {
860 MachO::any_relocation_info RE = getRelocation(Rel);
861 uint32_t SymbolIdx = getPlainRelocationSymbolNum(RE);
862 bool isExtern = getPlainRelocationExternal(RE);
864 return end_symbols();
866 MachO::symtab_command S = getSymtabLoadCommand();
867 unsigned SymbolTableEntrySize = is64Bit() ?
868 sizeof(MachO::nlist_64) :
869 sizeof(MachO::nlist);
870 uint64_t Offset = S.symoff + SymbolIdx * SymbolTableEntrySize;
872 Sym.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
873 return symbol_iterator(SymbolRef(Sym, this));
876 error_code MachOObjectFile::getRelocationType(DataRefImpl Rel,
877 uint64_t &Res) const {
878 MachO::any_relocation_info RE = getRelocation(Rel);
879 Res = getAnyRelocationType(RE);
880 return object_error::success;
884 MachOObjectFile::getRelocationTypeName(DataRefImpl Rel,
885 SmallVectorImpl<char> &Result) const {
888 getRelocationType(Rel, RType);
890 unsigned Arch = this->getArch();
894 static const char *const Table[] = {
895 "GENERIC_RELOC_VANILLA",
896 "GENERIC_RELOC_PAIR",
897 "GENERIC_RELOC_SECTDIFF",
898 "GENERIC_RELOC_PB_LA_PTR",
899 "GENERIC_RELOC_LOCAL_SECTDIFF",
900 "GENERIC_RELOC_TLV" };
908 case Triple::x86_64: {
909 static const char *const Table[] = {
910 "X86_64_RELOC_UNSIGNED",
911 "X86_64_RELOC_SIGNED",
912 "X86_64_RELOC_BRANCH",
913 "X86_64_RELOC_GOT_LOAD",
915 "X86_64_RELOC_SUBTRACTOR",
916 "X86_64_RELOC_SIGNED_1",
917 "X86_64_RELOC_SIGNED_2",
918 "X86_64_RELOC_SIGNED_4",
919 "X86_64_RELOC_TLV" };
928 static const char *const Table[] = {
931 "ARM_RELOC_SECTDIFF",
932 "ARM_RELOC_LOCAL_SECTDIFF",
933 "ARM_RELOC_PB_LA_PTR",
935 "ARM_THUMB_RELOC_BR22",
936 "ARM_THUMB_32BIT_BRANCH",
938 "ARM_RELOC_HALF_SECTDIFF" };
947 static const char *const Table[] = {
956 "PPC_RELOC_SECTDIFF",
957 "PPC_RELOC_PB_LA_PTR",
958 "PPC_RELOC_HI16_SECTDIFF",
959 "PPC_RELOC_LO16_SECTDIFF",
960 "PPC_RELOC_HA16_SECTDIFF",
962 "PPC_RELOC_LO14_SECTDIFF",
963 "PPC_RELOC_LOCAL_SECTDIFF" };
971 case Triple::UnknownArch:
975 Result.append(res.begin(), res.end());
976 return object_error::success;
980 MachOObjectFile::getRelocationValueString(DataRefImpl Rel,
981 SmallVectorImpl<char> &Result) const {
982 MachO::any_relocation_info RE = getRelocation(Rel);
984 unsigned Arch = this->getArch();
987 raw_string_ostream fmt(fmtbuf);
988 unsigned Type = this->getAnyRelocationType(RE);
989 bool IsPCRel = this->getAnyRelocationPCRel(RE);
991 // Determine any addends that should be displayed with the relocation.
992 // These require decoding the relocation type, which is triple-specific.
994 // X86_64 has entirely custom relocation types.
995 if (Arch == Triple::x86_64) {
996 bool isPCRel = getAnyRelocationPCRel(RE);
999 case MachO::X86_64_RELOC_GOT_LOAD:
1000 case MachO::X86_64_RELOC_GOT: {
1001 printRelocationTargetName(this, RE, fmt);
1003 if (isPCRel) fmt << "PCREL";
1006 case MachO::X86_64_RELOC_SUBTRACTOR: {
1007 DataRefImpl RelNext = Rel;
1009 MachO::any_relocation_info RENext = getRelocation(RelNext);
1011 // X86_64_RELOC_SUBTRACTOR must be followed by a relocation of type
1012 // X86_64_RELOC_UNSIGNED.
1013 // NOTE: Scattered relocations don't exist on x86_64.
1014 unsigned RType = getAnyRelocationType(RENext);
1015 if (RType != MachO::X86_64_RELOC_UNSIGNED)
1016 report_fatal_error("Expected X86_64_RELOC_UNSIGNED after "
1017 "X86_64_RELOC_SUBTRACTOR.");
1019 // The X86_64_RELOC_UNSIGNED contains the minuend symbol;
1020 // X86_64_RELOC_SUBTRACTOR contains the subtrahend.
1021 printRelocationTargetName(this, RENext, fmt);
1023 printRelocationTargetName(this, RE, fmt);
1026 case MachO::X86_64_RELOC_TLV:
1027 printRelocationTargetName(this, RE, fmt);
1029 if (isPCRel) fmt << "P";
1031 case MachO::X86_64_RELOC_SIGNED_1:
1032 printRelocationTargetName(this, RE, fmt);
1035 case MachO::X86_64_RELOC_SIGNED_2:
1036 printRelocationTargetName(this, RE, fmt);
1039 case MachO::X86_64_RELOC_SIGNED_4:
1040 printRelocationTargetName(this, RE, fmt);
1044 printRelocationTargetName(this, RE, fmt);
1047 // X86 and ARM share some relocation types in common.
1048 } else if (Arch == Triple::x86 || Arch == Triple::arm ||
1049 Arch == Triple::ppc) {
1050 // Generic relocation types...
1052 case MachO::GENERIC_RELOC_PAIR: // prints no info
1053 return object_error::success;
1054 case MachO::GENERIC_RELOC_SECTDIFF: {
1055 DataRefImpl RelNext = Rel;
1057 MachO::any_relocation_info RENext = getRelocation(RelNext);
1059 // X86 sect diff's must be followed by a relocation of type
1060 // GENERIC_RELOC_PAIR.
1061 unsigned RType = getAnyRelocationType(RENext);
1063 if (RType != MachO::GENERIC_RELOC_PAIR)
1064 report_fatal_error("Expected GENERIC_RELOC_PAIR after "
1065 "GENERIC_RELOC_SECTDIFF.");
1067 printRelocationTargetName(this, RE, fmt);
1069 printRelocationTargetName(this, RENext, fmt);
1074 if (Arch == Triple::x86 || Arch == Triple::ppc) {
1076 case MachO::GENERIC_RELOC_LOCAL_SECTDIFF: {
1077 DataRefImpl RelNext = Rel;
1079 MachO::any_relocation_info RENext = getRelocation(RelNext);
1081 // X86 sect diff's must be followed by a relocation of type
1082 // GENERIC_RELOC_PAIR.
1083 unsigned RType = getAnyRelocationType(RENext);
1084 if (RType != MachO::GENERIC_RELOC_PAIR)
1085 report_fatal_error("Expected GENERIC_RELOC_PAIR after "
1086 "GENERIC_RELOC_LOCAL_SECTDIFF.");
1088 printRelocationTargetName(this, RE, fmt);
1090 printRelocationTargetName(this, RENext, fmt);
1093 case MachO::GENERIC_RELOC_TLV: {
1094 printRelocationTargetName(this, RE, fmt);
1096 if (IsPCRel) fmt << "P";
1100 printRelocationTargetName(this, RE, fmt);
1102 } else { // ARM-specific relocations
1104 case MachO::ARM_RELOC_HALF:
1105 case MachO::ARM_RELOC_HALF_SECTDIFF: {
1106 // Half relocations steal a bit from the length field to encode
1107 // whether this is an upper16 or a lower16 relocation.
1108 bool isUpper = getAnyRelocationLength(RE) >> 1;
1111 fmt << ":upper16:(";
1113 fmt << ":lower16:(";
1114 printRelocationTargetName(this, RE, fmt);
1116 DataRefImpl RelNext = Rel;
1118 MachO::any_relocation_info RENext = getRelocation(RelNext);
1120 // ARM half relocs must be followed by a relocation of type
1122 unsigned RType = getAnyRelocationType(RENext);
1123 if (RType != MachO::ARM_RELOC_PAIR)
1124 report_fatal_error("Expected ARM_RELOC_PAIR after "
1127 // NOTE: The half of the target virtual address is stashed in the
1128 // address field of the secondary relocation, but we can't reverse
1129 // engineer the constant offset from it without decoding the movw/movt
1130 // instruction to find the other half in its immediate field.
1132 // ARM_RELOC_HALF_SECTDIFF encodes the second section in the
1133 // symbol/section pointer of the follow-on relocation.
1134 if (Type == MachO::ARM_RELOC_HALF_SECTDIFF) {
1136 printRelocationTargetName(this, RENext, fmt);
1143 printRelocationTargetName(this, RE, fmt);
1148 printRelocationTargetName(this, RE, fmt);
1151 Result.append(fmtbuf.begin(), fmtbuf.end());
1152 return object_error::success;
1156 MachOObjectFile::getRelocationHidden(DataRefImpl Rel, bool &Result) const {
1157 unsigned Arch = getArch();
1159 getRelocationType(Rel, Type);
1163 // On arches that use the generic relocations, GENERIC_RELOC_PAIR
1164 // is always hidden.
1165 if (Arch == Triple::x86 || Arch == Triple::arm || Arch == Triple::ppc) {
1166 if (Type == MachO::GENERIC_RELOC_PAIR) Result = true;
1167 } else if (Arch == Triple::x86_64) {
1168 // On x86_64, X86_64_RELOC_UNSIGNED is hidden only when it follows
1169 // an X86_64_RELOC_SUBTRACTOR.
1170 if (Type == MachO::X86_64_RELOC_UNSIGNED && Rel.d.a > 0) {
1171 DataRefImpl RelPrev = Rel;
1174 getRelocationType(RelPrev, PrevType);
1175 if (PrevType == MachO::X86_64_RELOC_SUBTRACTOR)
1180 return object_error::success;
1183 error_code MachOObjectFile::getLibraryNext(DataRefImpl LibData,
1184 LibraryRef &Res) const {
1185 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1188 error_code MachOObjectFile::getLibraryPath(DataRefImpl LibData,
1189 StringRef &Res) const {
1190 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1193 symbol_iterator MachOObjectFile::begin_symbols() const {
1196 return symbol_iterator(SymbolRef(DRI, this));
1198 MachO::symtab_command Symtab = getSymtabLoadCommand();
1199 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Symtab.symoff));
1200 return symbol_iterator(SymbolRef(DRI, this));
1203 symbol_iterator MachOObjectFile::end_symbols() const {
1206 return symbol_iterator(SymbolRef(DRI, this));
1208 MachO::symtab_command Symtab = getSymtabLoadCommand();
1209 unsigned SymbolTableEntrySize = is64Bit() ?
1210 sizeof(MachO::nlist_64) :
1211 sizeof(MachO::nlist);
1212 unsigned Offset = Symtab.symoff +
1213 Symtab.nsyms * SymbolTableEntrySize;
1214 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
1215 return symbol_iterator(SymbolRef(DRI, this));
1218 symbol_iterator MachOObjectFile::begin_dynamic_symbols() const {
1220 report_fatal_error("Dynamic symbols unimplemented in MachOObjectFile");
1223 symbol_iterator MachOObjectFile::end_dynamic_symbols() const {
1225 report_fatal_error("Dynamic symbols unimplemented in MachOObjectFile");
1228 section_iterator MachOObjectFile::begin_sections() const {
1230 return section_iterator(SectionRef(DRI, this));
1233 section_iterator MachOObjectFile::end_sections() const {
1235 DRI.d.a = Sections.size();
1236 return section_iterator(SectionRef(DRI, this));
1239 library_iterator MachOObjectFile::begin_libraries_needed() const {
1241 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1244 library_iterator MachOObjectFile::end_libraries_needed() const {
1246 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1249 uint8_t MachOObjectFile::getBytesInAddress() const {
1250 return is64Bit() ? 8 : 4;
1253 StringRef MachOObjectFile::getFileFormatName() const {
1254 unsigned CPUType = getCPUType(this);
1257 case llvm::MachO::CPU_TYPE_I386:
1258 return "Mach-O 32-bit i386";
1259 case llvm::MachO::CPU_TYPE_ARM:
1260 return "Mach-O arm";
1261 case llvm::MachO::CPU_TYPE_POWERPC:
1262 return "Mach-O 32-bit ppc";
1264 assert((CPUType & llvm::MachO::CPU_ARCH_ABI64) == 0 &&
1265 "64-bit object file when we're not 64-bit?");
1266 return "Mach-O 32-bit unknown";
1270 // Make sure the cpu type has the correct mask.
1271 assert((CPUType & llvm::MachO::CPU_ARCH_ABI64)
1272 == llvm::MachO::CPU_ARCH_ABI64 &&
1273 "32-bit object file when we're 64-bit?");
1276 case llvm::MachO::CPU_TYPE_X86_64:
1277 return "Mach-O 64-bit x86-64";
1278 case llvm::MachO::CPU_TYPE_POWERPC64:
1279 return "Mach-O 64-bit ppc64";
1281 return "Mach-O 64-bit unknown";
1285 Triple::ArchType MachOObjectFile::getArch(uint32_t CPUType) {
1287 case llvm::MachO::CPU_TYPE_I386:
1289 case llvm::MachO::CPU_TYPE_X86_64:
1290 return Triple::x86_64;
1291 case llvm::MachO::CPU_TYPE_ARM:
1293 case llvm::MachO::CPU_TYPE_POWERPC:
1295 case llvm::MachO::CPU_TYPE_POWERPC64:
1296 return Triple::ppc64;
1298 return Triple::UnknownArch;
1302 unsigned MachOObjectFile::getArch() const {
1303 return getArch(getCPUType(this));
1306 StringRef MachOObjectFile::getLoadName() const {
1308 report_fatal_error("get_load_name() unimplemented in MachOObjectFile");
1311 relocation_iterator MachOObjectFile::section_rel_begin(unsigned Index) const {
1314 return section_rel_begin(DRI);
1317 relocation_iterator MachOObjectFile::section_rel_end(unsigned Index) const {
1320 return section_rel_end(DRI);
1323 dice_iterator MachOObjectFile::begin_dices() const {
1325 if (!DataInCodeLoadCmd)
1326 return dice_iterator(DiceRef(DRI, this));
1328 MachO::linkedit_data_command DicLC = getDataInCodeLoadCommand();
1329 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, DicLC.dataoff));
1330 return dice_iterator(DiceRef(DRI, this));
1333 dice_iterator MachOObjectFile::end_dices() const {
1335 if (!DataInCodeLoadCmd)
1336 return dice_iterator(DiceRef(DRI, this));
1338 MachO::linkedit_data_command DicLC = getDataInCodeLoadCommand();
1339 unsigned Offset = DicLC.dataoff + DicLC.datasize;
1340 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
1341 return dice_iterator(DiceRef(DRI, this));
1345 MachOObjectFile::getSectionFinalSegmentName(DataRefImpl Sec) const {
1346 ArrayRef<char> Raw = getSectionRawFinalSegmentName(Sec);
1347 return parseSegmentOrSectionName(Raw.data());
1351 MachOObjectFile::getSectionRawName(DataRefImpl Sec) const {
1352 const section_base *Base =
1353 reinterpret_cast<const section_base *>(Sections[Sec.d.a]);
1354 return ArrayRef<char>(Base->sectname);
1358 MachOObjectFile::getSectionRawFinalSegmentName(DataRefImpl Sec) const {
1359 const section_base *Base =
1360 reinterpret_cast<const section_base *>(Sections[Sec.d.a]);
1361 return ArrayRef<char>(Base->segname);
1365 MachOObjectFile::isRelocationScattered(const MachO::any_relocation_info &RE)
1367 if (getCPUType(this) == MachO::CPU_TYPE_X86_64)
1369 return getPlainRelocationAddress(RE) & MachO::R_SCATTERED;
1372 unsigned MachOObjectFile::getPlainRelocationSymbolNum(
1373 const MachO::any_relocation_info &RE) const {
1374 if (isLittleEndian())
1375 return RE.r_word1 & 0xffffff;
1376 return RE.r_word1 >> 8;
1379 bool MachOObjectFile::getPlainRelocationExternal(
1380 const MachO::any_relocation_info &RE) const {
1381 if (isLittleEndian())
1382 return (RE.r_word1 >> 27) & 1;
1383 return (RE.r_word1 >> 4) & 1;
1386 bool MachOObjectFile::getScatteredRelocationScattered(
1387 const MachO::any_relocation_info &RE) const {
1388 return RE.r_word0 >> 31;
1391 uint32_t MachOObjectFile::getScatteredRelocationValue(
1392 const MachO::any_relocation_info &RE) const {
1396 unsigned MachOObjectFile::getAnyRelocationAddress(
1397 const MachO::any_relocation_info &RE) const {
1398 if (isRelocationScattered(RE))
1399 return getScatteredRelocationAddress(RE);
1400 return getPlainRelocationAddress(RE);
1403 unsigned MachOObjectFile::getAnyRelocationPCRel(
1404 const MachO::any_relocation_info &RE) const {
1405 if (isRelocationScattered(RE))
1406 return getScatteredRelocationPCRel(this, RE);
1407 return getPlainRelocationPCRel(this, RE);
1410 unsigned MachOObjectFile::getAnyRelocationLength(
1411 const MachO::any_relocation_info &RE) const {
1412 if (isRelocationScattered(RE))
1413 return getScatteredRelocationLength(RE);
1414 return getPlainRelocationLength(this, RE);
1418 MachOObjectFile::getAnyRelocationType(
1419 const MachO::any_relocation_info &RE) const {
1420 if (isRelocationScattered(RE))
1421 return getScatteredRelocationType(RE);
1422 return getPlainRelocationType(this, RE);
1426 MachOObjectFile::getRelocationSection(
1427 const MachO::any_relocation_info &RE) const {
1428 if (isRelocationScattered(RE) || getPlainRelocationExternal(RE))
1429 return *end_sections();
1430 unsigned SecNum = getPlainRelocationSymbolNum(RE) - 1;
1433 return SectionRef(DRI, this);
1436 MachOObjectFile::LoadCommandInfo
1437 MachOObjectFile::getFirstLoadCommandInfo() const {
1438 MachOObjectFile::LoadCommandInfo Load;
1440 unsigned HeaderSize = is64Bit() ? sizeof(MachO::mach_header_64) :
1441 sizeof(MachO::mach_header);
1442 Load.Ptr = getPtr(this, HeaderSize);
1443 Load.C = getStruct<MachO::load_command>(this, Load.Ptr);
1447 MachOObjectFile::LoadCommandInfo
1448 MachOObjectFile::getNextLoadCommandInfo(const LoadCommandInfo &L) const {
1449 MachOObjectFile::LoadCommandInfo Next;
1450 Next.Ptr = L.Ptr + L.C.cmdsize;
1451 Next.C = getStruct<MachO::load_command>(this, Next.Ptr);
1455 MachO::section MachOObjectFile::getSection(DataRefImpl DRI) const {
1456 return getStruct<MachO::section>(this, Sections[DRI.d.a]);
1459 MachO::section_64 MachOObjectFile::getSection64(DataRefImpl DRI) const {
1460 return getStruct<MachO::section_64>(this, Sections[DRI.d.a]);
1463 MachO::section MachOObjectFile::getSection(const LoadCommandInfo &L,
1464 unsigned Index) const {
1465 const char *Sec = getSectionPtr(this, L, Index);
1466 return getStruct<MachO::section>(this, Sec);
1469 MachO::section_64 MachOObjectFile::getSection64(const LoadCommandInfo &L,
1470 unsigned Index) const {
1471 const char *Sec = getSectionPtr(this, L, Index);
1472 return getStruct<MachO::section_64>(this, Sec);
1476 MachOObjectFile::getSymbolTableEntry(DataRefImpl DRI) const {
1477 const char *P = reinterpret_cast<const char *>(DRI.p);
1478 return getStruct<MachO::nlist>(this, P);
1482 MachOObjectFile::getSymbol64TableEntry(DataRefImpl DRI) const {
1483 const char *P = reinterpret_cast<const char *>(DRI.p);
1484 return getStruct<MachO::nlist_64>(this, P);
1487 MachO::linkedit_data_command
1488 MachOObjectFile::getLinkeditDataLoadCommand(const LoadCommandInfo &L) const {
1489 return getStruct<MachO::linkedit_data_command>(this, L.Ptr);
1492 MachO::segment_command
1493 MachOObjectFile::getSegmentLoadCommand(const LoadCommandInfo &L) const {
1494 return getStruct<MachO::segment_command>(this, L.Ptr);
1497 MachO::segment_command_64
1498 MachOObjectFile::getSegment64LoadCommand(const LoadCommandInfo &L) const {
1499 return getStruct<MachO::segment_command_64>(this, L.Ptr);
1502 MachO::linker_options_command
1503 MachOObjectFile::getLinkerOptionsLoadCommand(const LoadCommandInfo &L) const {
1504 return getStruct<MachO::linker_options_command>(this, L.Ptr);
1507 MachO::any_relocation_info
1508 MachOObjectFile::getRelocation(DataRefImpl Rel) const {
1509 const char *P = reinterpret_cast<const char *>(Rel.p);
1510 return getStruct<MachO::any_relocation_info>(this, P);
1513 MachO::data_in_code_entry
1514 MachOObjectFile::getDice(DataRefImpl Rel) const {
1515 const char *P = reinterpret_cast<const char *>(Rel.p);
1516 return getStruct<MachO::data_in_code_entry>(this, P);
1519 MachO::mach_header MachOObjectFile::getHeader() const {
1520 return getStruct<MachO::mach_header>(this, getPtr(this, 0));
1523 MachO::mach_header_64 MachOObjectFile::getHeader64() const {
1524 return getStruct<MachO::mach_header_64>(this, getPtr(this, 0));
1527 uint32_t MachOObjectFile::getIndirectSymbolTableEntry(
1528 const MachO::dysymtab_command &DLC,
1529 unsigned Index) const {
1530 uint64_t Offset = DLC.indirectsymoff + Index * sizeof(uint32_t);
1531 return getStruct<uint32_t>(this, getPtr(this, Offset));
1534 MachO::data_in_code_entry
1535 MachOObjectFile::getDataInCodeTableEntry(uint32_t DataOffset,
1536 unsigned Index) const {
1537 uint64_t Offset = DataOffset + Index * sizeof(MachO::data_in_code_entry);
1538 return getStruct<MachO::data_in_code_entry>(this, getPtr(this, Offset));
1541 MachO::symtab_command MachOObjectFile::getSymtabLoadCommand() const {
1542 return getStruct<MachO::symtab_command>(this, SymtabLoadCmd);
1545 MachO::dysymtab_command MachOObjectFile::getDysymtabLoadCommand() const {
1546 return getStruct<MachO::dysymtab_command>(this, DysymtabLoadCmd);
1549 MachO::linkedit_data_command
1550 MachOObjectFile::getDataInCodeLoadCommand() const {
1551 if (DataInCodeLoadCmd)
1552 return getStruct<MachO::linkedit_data_command>(this, DataInCodeLoadCmd);
1554 // If there is no DataInCodeLoadCmd return a load command with zero'ed fields.
1555 MachO::linkedit_data_command Cmd;
1556 Cmd.cmd = MachO::LC_DATA_IN_CODE;
1557 Cmd.cmdsize = sizeof(MachO::linkedit_data_command);
1563 StringRef MachOObjectFile::getStringTableData() const {
1564 MachO::symtab_command S = getSymtabLoadCommand();
1565 return getData().substr(S.stroff, S.strsize);
1568 bool MachOObjectFile::is64Bit() const {
1569 return getType() == getMachOType(false, true) ||
1570 getType() == getMachOType(true, true);
1573 void MachOObjectFile::ReadULEB128s(uint64_t Index,
1574 SmallVectorImpl<uint64_t> &Out) const {
1575 DataExtractor extractor(ObjectFile::getData(), true, 0);
1577 uint32_t offset = Index;
1579 while (uint64_t delta = extractor.getULEB128(&offset)) {
1581 Out.push_back(data);
1585 ErrorOr<ObjectFile *> ObjectFile::createMachOObjectFile(MemoryBuffer *Buffer) {
1586 StringRef Magic = Buffer->getBuffer().slice(0, 4);
1588 OwningPtr<MachOObjectFile> Ret;
1589 if (Magic == "\xFE\xED\xFA\xCE")
1590 Ret.reset(new MachOObjectFile(Buffer, false, false, EC));
1591 else if (Magic == "\xCE\xFA\xED\xFE")
1592 Ret.reset(new MachOObjectFile(Buffer, true, false, EC));
1593 else if (Magic == "\xFE\xED\xFA\xCF")
1594 Ret.reset(new MachOObjectFile(Buffer, false, true, EC));
1595 else if (Magic == "\xCF\xFA\xED\xFE")
1596 Ret.reset(new MachOObjectFile(Buffer, true, true, EC));
1599 return object_error::parse_failed;
1607 } // end namespace object
1608 } // end namespace llvm