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" };
968 case Triple::UnknownArch:
972 Result.append(res.begin(), res.end());
973 return object_error::success;
977 MachOObjectFile::getRelocationValueString(DataRefImpl Rel,
978 SmallVectorImpl<char> &Result) const {
979 MachO::any_relocation_info RE = getRelocation(Rel);
981 unsigned Arch = this->getArch();
984 raw_string_ostream fmt(fmtbuf);
985 unsigned Type = this->getAnyRelocationType(RE);
986 bool IsPCRel = this->getAnyRelocationPCRel(RE);
988 // Determine any addends that should be displayed with the relocation.
989 // These require decoding the relocation type, which is triple-specific.
991 // X86_64 has entirely custom relocation types.
992 if (Arch == Triple::x86_64) {
993 bool isPCRel = getAnyRelocationPCRel(RE);
996 case MachO::X86_64_RELOC_GOT_LOAD:
997 case MachO::X86_64_RELOC_GOT: {
998 printRelocationTargetName(this, RE, fmt);
1000 if (isPCRel) fmt << "PCREL";
1003 case MachO::X86_64_RELOC_SUBTRACTOR: {
1004 DataRefImpl RelNext = Rel;
1006 MachO::any_relocation_info RENext = getRelocation(RelNext);
1008 // X86_64_RELOC_SUBTRACTOR must be followed by a relocation of type
1009 // X86_64_RELOC_UNSIGNED.
1010 // NOTE: Scattered relocations don't exist on x86_64.
1011 unsigned RType = getAnyRelocationType(RENext);
1012 if (RType != MachO::X86_64_RELOC_UNSIGNED)
1013 report_fatal_error("Expected X86_64_RELOC_UNSIGNED after "
1014 "X86_64_RELOC_SUBTRACTOR.");
1016 // The X86_64_RELOC_UNSIGNED contains the minuend symbol;
1017 // X86_64_RELOC_SUBTRACTOR contains the subtrahend.
1018 printRelocationTargetName(this, RENext, fmt);
1020 printRelocationTargetName(this, RE, fmt);
1023 case MachO::X86_64_RELOC_TLV:
1024 printRelocationTargetName(this, RE, fmt);
1026 if (isPCRel) fmt << "P";
1028 case MachO::X86_64_RELOC_SIGNED_1:
1029 printRelocationTargetName(this, RE, fmt);
1032 case MachO::X86_64_RELOC_SIGNED_2:
1033 printRelocationTargetName(this, RE, fmt);
1036 case MachO::X86_64_RELOC_SIGNED_4:
1037 printRelocationTargetName(this, RE, fmt);
1041 printRelocationTargetName(this, RE, fmt);
1044 // X86 and ARM share some relocation types in common.
1045 } else if (Arch == Triple::x86 || Arch == Triple::arm ||
1046 Arch == Triple::ppc) {
1047 // Generic relocation types...
1049 case MachO::GENERIC_RELOC_PAIR: // prints no info
1050 return object_error::success;
1051 case MachO::GENERIC_RELOC_SECTDIFF: {
1052 DataRefImpl RelNext = Rel;
1054 MachO::any_relocation_info RENext = getRelocation(RelNext);
1056 // X86 sect diff's must be followed by a relocation of type
1057 // GENERIC_RELOC_PAIR.
1058 unsigned RType = getAnyRelocationType(RENext);
1060 if (RType != MachO::GENERIC_RELOC_PAIR)
1061 report_fatal_error("Expected GENERIC_RELOC_PAIR after "
1062 "GENERIC_RELOC_SECTDIFF.");
1064 printRelocationTargetName(this, RE, fmt);
1066 printRelocationTargetName(this, RENext, fmt);
1071 if (Arch == Triple::x86 || Arch == Triple::ppc) {
1073 case MachO::GENERIC_RELOC_LOCAL_SECTDIFF: {
1074 DataRefImpl RelNext = Rel;
1076 MachO::any_relocation_info RENext = getRelocation(RelNext);
1078 // X86 sect diff's must be followed by a relocation of type
1079 // GENERIC_RELOC_PAIR.
1080 unsigned RType = getAnyRelocationType(RENext);
1081 if (RType != MachO::GENERIC_RELOC_PAIR)
1082 report_fatal_error("Expected GENERIC_RELOC_PAIR after "
1083 "GENERIC_RELOC_LOCAL_SECTDIFF.");
1085 printRelocationTargetName(this, RE, fmt);
1087 printRelocationTargetName(this, RENext, fmt);
1090 case MachO::GENERIC_RELOC_TLV: {
1091 printRelocationTargetName(this, RE, fmt);
1093 if (IsPCRel) fmt << "P";
1097 printRelocationTargetName(this, RE, fmt);
1099 } else { // ARM-specific relocations
1101 case MachO::ARM_RELOC_HALF:
1102 case MachO::ARM_RELOC_HALF_SECTDIFF: {
1103 // Half relocations steal a bit from the length field to encode
1104 // whether this is an upper16 or a lower16 relocation.
1105 bool isUpper = getAnyRelocationLength(RE) >> 1;
1108 fmt << ":upper16:(";
1110 fmt << ":lower16:(";
1111 printRelocationTargetName(this, RE, fmt);
1113 DataRefImpl RelNext = Rel;
1115 MachO::any_relocation_info RENext = getRelocation(RelNext);
1117 // ARM half relocs must be followed by a relocation of type
1119 unsigned RType = getAnyRelocationType(RENext);
1120 if (RType != MachO::ARM_RELOC_PAIR)
1121 report_fatal_error("Expected ARM_RELOC_PAIR after "
1124 // NOTE: The half of the target virtual address is stashed in the
1125 // address field of the secondary relocation, but we can't reverse
1126 // engineer the constant offset from it without decoding the movw/movt
1127 // instruction to find the other half in its immediate field.
1129 // ARM_RELOC_HALF_SECTDIFF encodes the second section in the
1130 // symbol/section pointer of the follow-on relocation.
1131 if (Type == MachO::ARM_RELOC_HALF_SECTDIFF) {
1133 printRelocationTargetName(this, RENext, fmt);
1140 printRelocationTargetName(this, RE, fmt);
1145 printRelocationTargetName(this, RE, fmt);
1148 Result.append(fmtbuf.begin(), fmtbuf.end());
1149 return object_error::success;
1153 MachOObjectFile::getRelocationHidden(DataRefImpl Rel, bool &Result) const {
1154 unsigned Arch = getArch();
1156 getRelocationType(Rel, Type);
1160 // On arches that use the generic relocations, GENERIC_RELOC_PAIR
1161 // is always hidden.
1162 if (Arch == Triple::x86 || Arch == Triple::arm || Arch == Triple::ppc) {
1163 if (Type == MachO::GENERIC_RELOC_PAIR) Result = true;
1164 } else if (Arch == Triple::x86_64) {
1165 // On x86_64, X86_64_RELOC_UNSIGNED is hidden only when it follows
1166 // an X86_64_RELOC_SUBTRACTOR.
1167 if (Type == MachO::X86_64_RELOC_UNSIGNED && Rel.d.a > 0) {
1168 DataRefImpl RelPrev = Rel;
1171 getRelocationType(RelPrev, PrevType);
1172 if (PrevType == MachO::X86_64_RELOC_SUBTRACTOR)
1177 return object_error::success;
1180 error_code MachOObjectFile::getLibraryNext(DataRefImpl LibData,
1181 LibraryRef &Res) const {
1182 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1185 error_code MachOObjectFile::getLibraryPath(DataRefImpl LibData,
1186 StringRef &Res) const {
1187 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1190 symbol_iterator MachOObjectFile::begin_symbols() const {
1193 return symbol_iterator(SymbolRef(DRI, this));
1195 MachO::symtab_command Symtab = getSymtabLoadCommand();
1196 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Symtab.symoff));
1197 return symbol_iterator(SymbolRef(DRI, this));
1200 symbol_iterator MachOObjectFile::end_symbols() const {
1203 return symbol_iterator(SymbolRef(DRI, this));
1205 MachO::symtab_command Symtab = getSymtabLoadCommand();
1206 unsigned SymbolTableEntrySize = is64Bit() ?
1207 sizeof(MachO::nlist_64) :
1208 sizeof(MachO::nlist);
1209 unsigned Offset = Symtab.symoff +
1210 Symtab.nsyms * SymbolTableEntrySize;
1211 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
1212 return symbol_iterator(SymbolRef(DRI, this));
1215 symbol_iterator MachOObjectFile::begin_dynamic_symbols() const {
1217 report_fatal_error("Dynamic symbols unimplemented in MachOObjectFile");
1220 symbol_iterator MachOObjectFile::end_dynamic_symbols() const {
1222 report_fatal_error("Dynamic symbols unimplemented in MachOObjectFile");
1225 section_iterator MachOObjectFile::begin_sections() const {
1227 return section_iterator(SectionRef(DRI, this));
1230 section_iterator MachOObjectFile::end_sections() const {
1232 DRI.d.a = Sections.size();
1233 return section_iterator(SectionRef(DRI, this));
1236 library_iterator MachOObjectFile::begin_libraries_needed() const {
1238 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1241 library_iterator MachOObjectFile::end_libraries_needed() const {
1243 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1246 uint8_t MachOObjectFile::getBytesInAddress() const {
1247 return is64Bit() ? 8 : 4;
1250 StringRef MachOObjectFile::getFileFormatName() const {
1251 unsigned CPUType = getCPUType(this);
1254 case llvm::MachO::CPU_TYPE_I386:
1255 return "Mach-O 32-bit i386";
1256 case llvm::MachO::CPU_TYPE_ARM:
1257 return "Mach-O arm";
1258 case llvm::MachO::CPU_TYPE_POWERPC:
1259 return "Mach-O 32-bit ppc";
1261 assert((CPUType & llvm::MachO::CPU_ARCH_ABI64) == 0 &&
1262 "64-bit object file when we're not 64-bit?");
1263 return "Mach-O 32-bit unknown";
1267 // Make sure the cpu type has the correct mask.
1268 assert((CPUType & llvm::MachO::CPU_ARCH_ABI64)
1269 == llvm::MachO::CPU_ARCH_ABI64 &&
1270 "32-bit object file when we're 64-bit?");
1273 case llvm::MachO::CPU_TYPE_X86_64:
1274 return "Mach-O 64-bit x86-64";
1275 case llvm::MachO::CPU_TYPE_POWERPC64:
1276 return "Mach-O 64-bit ppc64";
1278 return "Mach-O 64-bit unknown";
1282 Triple::ArchType MachOObjectFile::getArch(uint32_t CPUType) {
1284 case llvm::MachO::CPU_TYPE_I386:
1286 case llvm::MachO::CPU_TYPE_X86_64:
1287 return Triple::x86_64;
1288 case llvm::MachO::CPU_TYPE_ARM:
1290 case llvm::MachO::CPU_TYPE_POWERPC:
1292 case llvm::MachO::CPU_TYPE_POWERPC64:
1293 return Triple::ppc64;
1295 return Triple::UnknownArch;
1299 unsigned MachOObjectFile::getArch() const {
1300 return getArch(getCPUType(this));
1303 StringRef MachOObjectFile::getLoadName() const {
1305 report_fatal_error("get_load_name() unimplemented in MachOObjectFile");
1308 relocation_iterator MachOObjectFile::section_rel_begin(unsigned Index) const {
1311 return section_rel_begin(DRI);
1314 relocation_iterator MachOObjectFile::section_rel_end(unsigned Index) const {
1317 return section_rel_end(DRI);
1320 dice_iterator MachOObjectFile::begin_dices() const {
1322 if (!DataInCodeLoadCmd)
1323 return dice_iterator(DiceRef(DRI, this));
1325 MachO::linkedit_data_command DicLC = getDataInCodeLoadCommand();
1326 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, DicLC.dataoff));
1327 return dice_iterator(DiceRef(DRI, this));
1330 dice_iterator MachOObjectFile::end_dices() const {
1332 if (!DataInCodeLoadCmd)
1333 return dice_iterator(DiceRef(DRI, this));
1335 MachO::linkedit_data_command DicLC = getDataInCodeLoadCommand();
1336 unsigned Offset = DicLC.dataoff + DicLC.datasize;
1337 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
1338 return dice_iterator(DiceRef(DRI, this));
1342 MachOObjectFile::getSectionFinalSegmentName(DataRefImpl Sec) const {
1343 ArrayRef<char> Raw = getSectionRawFinalSegmentName(Sec);
1344 return parseSegmentOrSectionName(Raw.data());
1348 MachOObjectFile::getSectionRawName(DataRefImpl Sec) const {
1349 const section_base *Base =
1350 reinterpret_cast<const section_base *>(Sections[Sec.d.a]);
1351 return ArrayRef<char>(Base->sectname);
1355 MachOObjectFile::getSectionRawFinalSegmentName(DataRefImpl Sec) const {
1356 const section_base *Base =
1357 reinterpret_cast<const section_base *>(Sections[Sec.d.a]);
1358 return ArrayRef<char>(Base->segname);
1362 MachOObjectFile::isRelocationScattered(const MachO::any_relocation_info &RE)
1364 if (getCPUType(this) == MachO::CPU_TYPE_X86_64)
1366 return getPlainRelocationAddress(RE) & MachO::R_SCATTERED;
1369 unsigned MachOObjectFile::getPlainRelocationSymbolNum(
1370 const MachO::any_relocation_info &RE) const {
1371 if (isLittleEndian())
1372 return RE.r_word1 & 0xffffff;
1373 return RE.r_word1 >> 8;
1376 bool MachOObjectFile::getPlainRelocationExternal(
1377 const MachO::any_relocation_info &RE) const {
1378 if (isLittleEndian())
1379 return (RE.r_word1 >> 27) & 1;
1380 return (RE.r_word1 >> 4) & 1;
1383 bool MachOObjectFile::getScatteredRelocationScattered(
1384 const MachO::any_relocation_info &RE) const {
1385 return RE.r_word0 >> 31;
1388 uint32_t MachOObjectFile::getScatteredRelocationValue(
1389 const MachO::any_relocation_info &RE) const {
1393 unsigned MachOObjectFile::getAnyRelocationAddress(
1394 const MachO::any_relocation_info &RE) const {
1395 if (isRelocationScattered(RE))
1396 return getScatteredRelocationAddress(RE);
1397 return getPlainRelocationAddress(RE);
1400 unsigned MachOObjectFile::getAnyRelocationPCRel(
1401 const MachO::any_relocation_info &RE) const {
1402 if (isRelocationScattered(RE))
1403 return getScatteredRelocationPCRel(this, RE);
1404 return getPlainRelocationPCRel(this, RE);
1407 unsigned MachOObjectFile::getAnyRelocationLength(
1408 const MachO::any_relocation_info &RE) const {
1409 if (isRelocationScattered(RE))
1410 return getScatteredRelocationLength(RE);
1411 return getPlainRelocationLength(this, RE);
1415 MachOObjectFile::getAnyRelocationType(
1416 const MachO::any_relocation_info &RE) const {
1417 if (isRelocationScattered(RE))
1418 return getScatteredRelocationType(RE);
1419 return getPlainRelocationType(this, RE);
1423 MachOObjectFile::getRelocationSection(
1424 const MachO::any_relocation_info &RE) const {
1425 if (isRelocationScattered(RE) || getPlainRelocationExternal(RE))
1426 return *end_sections();
1427 unsigned SecNum = getPlainRelocationSymbolNum(RE) - 1;
1430 return SectionRef(DRI, this);
1433 MachOObjectFile::LoadCommandInfo
1434 MachOObjectFile::getFirstLoadCommandInfo() const {
1435 MachOObjectFile::LoadCommandInfo Load;
1437 unsigned HeaderSize = is64Bit() ? sizeof(MachO::mach_header_64) :
1438 sizeof(MachO::mach_header);
1439 Load.Ptr = getPtr(this, HeaderSize);
1440 Load.C = getStruct<MachO::load_command>(this, Load.Ptr);
1444 MachOObjectFile::LoadCommandInfo
1445 MachOObjectFile::getNextLoadCommandInfo(const LoadCommandInfo &L) const {
1446 MachOObjectFile::LoadCommandInfo Next;
1447 Next.Ptr = L.Ptr + L.C.cmdsize;
1448 Next.C = getStruct<MachO::load_command>(this, Next.Ptr);
1452 MachO::section MachOObjectFile::getSection(DataRefImpl DRI) const {
1453 return getStruct<MachO::section>(this, Sections[DRI.d.a]);
1456 MachO::section_64 MachOObjectFile::getSection64(DataRefImpl DRI) const {
1457 return getStruct<MachO::section_64>(this, Sections[DRI.d.a]);
1460 MachO::section MachOObjectFile::getSection(const LoadCommandInfo &L,
1461 unsigned Index) const {
1462 const char *Sec = getSectionPtr(this, L, Index);
1463 return getStruct<MachO::section>(this, Sec);
1466 MachO::section_64 MachOObjectFile::getSection64(const LoadCommandInfo &L,
1467 unsigned Index) const {
1468 const char *Sec = getSectionPtr(this, L, Index);
1469 return getStruct<MachO::section_64>(this, Sec);
1473 MachOObjectFile::getSymbolTableEntry(DataRefImpl DRI) const {
1474 const char *P = reinterpret_cast<const char *>(DRI.p);
1475 return getStruct<MachO::nlist>(this, P);
1479 MachOObjectFile::getSymbol64TableEntry(DataRefImpl DRI) const {
1480 const char *P = reinterpret_cast<const char *>(DRI.p);
1481 return getStruct<MachO::nlist_64>(this, P);
1484 MachO::linkedit_data_command
1485 MachOObjectFile::getLinkeditDataLoadCommand(const LoadCommandInfo &L) const {
1486 return getStruct<MachO::linkedit_data_command>(this, L.Ptr);
1489 MachO::segment_command
1490 MachOObjectFile::getSegmentLoadCommand(const LoadCommandInfo &L) const {
1491 return getStruct<MachO::segment_command>(this, L.Ptr);
1494 MachO::segment_command_64
1495 MachOObjectFile::getSegment64LoadCommand(const LoadCommandInfo &L) const {
1496 return getStruct<MachO::segment_command_64>(this, L.Ptr);
1499 MachO::linker_options_command
1500 MachOObjectFile::getLinkerOptionsLoadCommand(const LoadCommandInfo &L) const {
1501 return getStruct<MachO::linker_options_command>(this, L.Ptr);
1504 MachO::any_relocation_info
1505 MachOObjectFile::getRelocation(DataRefImpl Rel) const {
1506 const char *P = reinterpret_cast<const char *>(Rel.p);
1507 return getStruct<MachO::any_relocation_info>(this, P);
1510 MachO::data_in_code_entry
1511 MachOObjectFile::getDice(DataRefImpl Rel) const {
1512 const char *P = reinterpret_cast<const char *>(Rel.p);
1513 return getStruct<MachO::data_in_code_entry>(this, P);
1516 MachO::mach_header MachOObjectFile::getHeader() const {
1517 return getStruct<MachO::mach_header>(this, getPtr(this, 0));
1520 MachO::mach_header_64 MachOObjectFile::getHeader64() const {
1521 return getStruct<MachO::mach_header_64>(this, getPtr(this, 0));
1524 uint32_t MachOObjectFile::getIndirectSymbolTableEntry(
1525 const MachO::dysymtab_command &DLC,
1526 unsigned Index) const {
1527 uint64_t Offset = DLC.indirectsymoff + Index * sizeof(uint32_t);
1528 return getStruct<uint32_t>(this, getPtr(this, Offset));
1531 MachO::data_in_code_entry
1532 MachOObjectFile::getDataInCodeTableEntry(uint32_t DataOffset,
1533 unsigned Index) const {
1534 uint64_t Offset = DataOffset + Index * sizeof(MachO::data_in_code_entry);
1535 return getStruct<MachO::data_in_code_entry>(this, getPtr(this, Offset));
1538 MachO::symtab_command MachOObjectFile::getSymtabLoadCommand() const {
1539 return getStruct<MachO::symtab_command>(this, SymtabLoadCmd);
1542 MachO::dysymtab_command MachOObjectFile::getDysymtabLoadCommand() const {
1543 return getStruct<MachO::dysymtab_command>(this, DysymtabLoadCmd);
1546 MachO::linkedit_data_command
1547 MachOObjectFile::getDataInCodeLoadCommand() const {
1548 if (DataInCodeLoadCmd)
1549 return getStruct<MachO::linkedit_data_command>(this, DataInCodeLoadCmd);
1551 // If there is no DataInCodeLoadCmd return a load command with zero'ed fields.
1552 MachO::linkedit_data_command Cmd;
1553 Cmd.cmd = MachO::LC_DATA_IN_CODE;
1554 Cmd.cmdsize = sizeof(MachO::linkedit_data_command);
1560 StringRef MachOObjectFile::getStringTableData() const {
1561 MachO::symtab_command S = getSymtabLoadCommand();
1562 return getData().substr(S.stroff, S.strsize);
1565 bool MachOObjectFile::is64Bit() const {
1566 return getType() == getMachOType(false, true) ||
1567 getType() == getMachOType(true, true);
1570 void MachOObjectFile::ReadULEB128s(uint64_t Index,
1571 SmallVectorImpl<uint64_t> &Out) const {
1572 DataExtractor extractor(ObjectFile::getData(), true, 0);
1574 uint32_t offset = Index;
1576 while (uint64_t delta = extractor.getULEB128(&offset)) {
1578 Out.push_back(data);
1582 ObjectFile *ObjectFile::createMachOObjectFile(MemoryBuffer *Buffer) {
1583 StringRef Magic = Buffer->getBuffer().slice(0, 4);
1585 OwningPtr<ObjectFile> Ret;
1586 if (Magic == "\xFE\xED\xFA\xCE")
1587 Ret.reset(new MachOObjectFile(Buffer, false, false, ec));
1588 else if (Magic == "\xCE\xFA\xED\xFE")
1589 Ret.reset(new MachOObjectFile(Buffer, true, false, ec));
1590 else if (Magic == "\xFE\xED\xFA\xCF")
1591 Ret.reset(new MachOObjectFile(Buffer, false, true, ec));
1592 else if (Magic == "\xCF\xFA\xED\xFE")
1593 Ret.reset(new MachOObjectFile(Buffer, true, true, ec));
1604 } // end namespace object
1605 } // end namespace llvm