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/STLExtras.h"
17 #include "llvm/ADT/Triple.h"
18 #include "llvm/Support/DataExtractor.h"
19 #include "llvm/Support/Format.h"
20 #include "llvm/Support/Host.h"
21 #include "llvm/Support/MemoryBuffer.h"
22 #include "llvm/Support/raw_ostream.h"
28 using namespace object;
46 static void SwapValue(T &Value) {
47 Value = sys::SwapByteOrder(Value);
51 static void SwapStruct(T &Value);
54 void SwapStruct(MachO::any_relocation_info &H) {
60 void SwapStruct(MachO::load_command &L) {
66 void SwapStruct(nlist_base &S) {
72 void SwapStruct(MachO::section &S) {
80 SwapValue(S.reserved1);
81 SwapValue(S.reserved2);
85 void SwapStruct(MachO::section_64 &S) {
93 SwapValue(S.reserved1);
94 SwapValue(S.reserved2);
95 SwapValue(S.reserved3);
99 void SwapStruct(MachO::nlist &S) {
102 SwapValue(S.n_value);
106 void SwapStruct(MachO::nlist_64 &S) {
109 SwapValue(S.n_value);
113 void SwapStruct(MachO::mach_header &H) {
115 SwapValue(H.cputype);
116 SwapValue(H.cpusubtype);
117 SwapValue(H.filetype);
119 SwapValue(H.sizeofcmds);
124 void SwapStruct(MachO::mach_header_64 &H) {
126 SwapValue(H.cputype);
127 SwapValue(H.cpusubtype);
128 SwapValue(H.filetype);
130 SwapValue(H.sizeofcmds);
132 SwapValue(H.reserved);
136 void SwapStruct(MachO::symtab_command &C) {
138 SwapValue(C.cmdsize);
142 SwapValue(C.strsize);
146 void SwapStruct(MachO::dysymtab_command &C) {
148 SwapValue(C.cmdsize);
149 SwapValue(C.ilocalsym);
150 SwapValue(C.nlocalsym);
151 SwapValue(C.iextdefsym);
152 SwapValue(C.nextdefsym);
153 SwapValue(C.iundefsym);
154 SwapValue(C.nundefsym);
157 SwapValue(C.modtaboff);
158 SwapValue(C.nmodtab);
159 SwapValue(C.extrefsymoff);
160 SwapValue(C.nextrefsyms);
161 SwapValue(C.indirectsymoff);
162 SwapValue(C.nindirectsyms);
163 SwapValue(C.extreloff);
164 SwapValue(C.nextrel);
165 SwapValue(C.locreloff);
166 SwapValue(C.nlocrel);
170 void SwapStruct(MachO::linkedit_data_command &C) {
172 SwapValue(C.cmdsize);
173 SwapValue(C.dataoff);
174 SwapValue(C.datasize);
178 void SwapStruct(MachO::segment_command &C) {
180 SwapValue(C.cmdsize);
183 SwapValue(C.fileoff);
184 SwapValue(C.filesize);
185 SwapValue(C.maxprot);
186 SwapValue(C.initprot);
192 void SwapStruct(MachO::segment_command_64 &C) {
194 SwapValue(C.cmdsize);
197 SwapValue(C.fileoff);
198 SwapValue(C.filesize);
199 SwapValue(C.maxprot);
200 SwapValue(C.initprot);
206 void SwapStruct(uint32_t &C) {
211 void SwapStruct(MachO::linker_options_command &C) {
213 SwapValue(C.cmdsize);
218 void SwapStruct(MachO::version_min_command&C) {
220 SwapValue(C.cmdsize);
221 SwapValue(C.version);
222 SwapValue(C.reserved);
226 void SwapStruct(MachO::data_in_code_entry &C) {
233 T getStruct(const MachOObjectFile *O, const char *P) {
235 memcpy(&Cmd, P, sizeof(T));
236 if (O->isLittleEndian() != sys::IsLittleEndianHost)
242 getSegmentLoadCommandNumSections(const MachOObjectFile *O,
243 const MachOObjectFile::LoadCommandInfo &L) {
245 MachO::segment_command_64 S = O->getSegment64LoadCommand(L);
248 MachO::segment_command S = O->getSegmentLoadCommand(L);
253 getSectionPtr(const MachOObjectFile *O, MachOObjectFile::LoadCommandInfo L,
255 uintptr_t CommandAddr = reinterpret_cast<uintptr_t>(L.Ptr);
257 bool Is64 = O->is64Bit();
258 unsigned SegmentLoadSize = Is64 ? sizeof(MachO::segment_command_64) :
259 sizeof(MachO::segment_command);
260 unsigned SectionSize = Is64 ? sizeof(MachO::section_64) :
261 sizeof(MachO::section);
263 uintptr_t SectionAddr = CommandAddr + SegmentLoadSize + Sec * SectionSize;
264 return reinterpret_cast<const char*>(SectionAddr);
267 static const char *getPtr(const MachOObjectFile *O, size_t Offset) {
268 return O->getData().substr(Offset, 1).data();
272 getSymbolTableEntryBase(const MachOObjectFile *O, DataRefImpl DRI) {
273 const char *P = reinterpret_cast<const char *>(DRI.p);
274 return getStruct<nlist_base>(O, P);
277 static StringRef parseSegmentOrSectionName(const char *P) {
281 // Not null terminated, so this is a 16 char string.
282 return StringRef(P, 16);
285 // Helper to advance a section or symbol iterator multiple increments at a time.
287 static void advance(T &it, size_t Val) {
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);
308 for (const SymbolRef &Symbol : O->symbols()) {
313 if ((ec = Symbol.getAddress(Addr)))
314 report_fatal_error(ec.message());
317 if ((ec = Symbol.getName(Name)))
318 report_fatal_error(ec.message());
323 // If we couldn't find a symbol that this relocation refers to, try
324 // to find a section beginning instead.
325 for (const SectionRef &Section : O->sections()) {
330 if ((ec = Section.getAddress(Addr)))
331 report_fatal_error(ec.message());
334 if ((ec = Section.getName(Name)))
335 report_fatal_error(ec.message());
340 fmt << format("0x%x", Val);
345 bool isExtern = O->getPlainRelocationExternal(RE);
346 uint64_t Val = O->getPlainRelocationSymbolNum(RE);
349 symbol_iterator SI = O->symbol_begin();
353 section_iterator SI = O->section_begin();
354 // Adjust for the fact that sections are 1-indexed.
355 advance(SI, Val - 1);
363 getPlainRelocationAddress(const MachO::any_relocation_info &RE) {
368 getScatteredRelocationAddress(const MachO::any_relocation_info &RE) {
369 return RE.r_word0 & 0xffffff;
372 static bool getPlainRelocationPCRel(const MachOObjectFile *O,
373 const MachO::any_relocation_info &RE) {
374 if (O->isLittleEndian())
375 return (RE.r_word1 >> 24) & 1;
376 return (RE.r_word1 >> 7) & 1;
380 getScatteredRelocationPCRel(const MachOObjectFile *O,
381 const MachO::any_relocation_info &RE) {
382 return (RE.r_word0 >> 30) & 1;
385 static unsigned getPlainRelocationLength(const MachOObjectFile *O,
386 const MachO::any_relocation_info &RE) {
387 if (O->isLittleEndian())
388 return (RE.r_word1 >> 25) & 3;
389 return (RE.r_word1 >> 5) & 3;
393 getScatteredRelocationLength(const MachO::any_relocation_info &RE) {
394 return (RE.r_word0 >> 28) & 3;
397 static unsigned getPlainRelocationType(const MachOObjectFile *O,
398 const MachO::any_relocation_info &RE) {
399 if (O->isLittleEndian())
400 return RE.r_word1 >> 28;
401 return RE.r_word1 & 0xf;
405 getScatteredRelocationType(const MachO::any_relocation_info &RE) {
406 return (RE.r_word0 >> 24) & 0xf;
409 static uint32_t getSectionFlags(const MachOObjectFile *O,
412 MachO::section_64 Sect = O->getSection64(Sec);
415 MachO::section Sect = O->getSection(Sec);
419 MachOObjectFile::MachOObjectFile(MemoryBuffer *Object, bool IsLittleEndian,
420 bool Is64bits, error_code &EC,
422 : ObjectFile(getMachOType(IsLittleEndian, Is64bits), Object, BufferOwned),
423 SymtabLoadCmd(NULL), DysymtabLoadCmd(NULL), DataInCodeLoadCmd(NULL) {
424 uint32_t LoadCommandCount = this->getHeader().ncmds;
425 MachO::LoadCommandType SegmentLoadType = is64Bit() ?
426 MachO::LC_SEGMENT_64 : MachO::LC_SEGMENT;
428 MachOObjectFile::LoadCommandInfo Load = getFirstLoadCommandInfo();
429 for (unsigned I = 0; ; ++I) {
430 if (Load.C.cmd == MachO::LC_SYMTAB) {
431 assert(!SymtabLoadCmd && "Multiple symbol tables");
432 SymtabLoadCmd = Load.Ptr;
433 } else if (Load.C.cmd == MachO::LC_DYSYMTAB) {
434 assert(!DysymtabLoadCmd && "Multiple dynamic symbol tables");
435 DysymtabLoadCmd = Load.Ptr;
436 } else if (Load.C.cmd == MachO::LC_DATA_IN_CODE) {
437 assert(!DataInCodeLoadCmd && "Multiple data in code tables");
438 DataInCodeLoadCmd = Load.Ptr;
439 } else if (Load.C.cmd == SegmentLoadType) {
440 uint32_t NumSections = getSegmentLoadCommandNumSections(this, Load);
441 for (unsigned J = 0; J < NumSections; ++J) {
442 const char *Sec = getSectionPtr(this, Load, J);
443 Sections.push_back(Sec);
447 if (I == LoadCommandCount - 1)
450 Load = getNextLoadCommandInfo(Load);
454 void MachOObjectFile::moveSymbolNext(DataRefImpl &Symb) const {
455 unsigned SymbolTableEntrySize = is64Bit() ?
456 sizeof(MachO::nlist_64) :
457 sizeof(MachO::nlist);
458 Symb.p += SymbolTableEntrySize;
461 error_code MachOObjectFile::getSymbolName(DataRefImpl Symb,
462 StringRef &Res) const {
463 StringRef StringTable = getStringTableData();
464 nlist_base Entry = getSymbolTableEntryBase(this, Symb);
465 const char *Start = &StringTable.data()[Entry.n_strx];
466 Res = StringRef(Start);
467 return object_error::success;
470 error_code MachOObjectFile::getSymbolAddress(DataRefImpl Symb,
471 uint64_t &Res) const {
473 MachO::nlist_64 Entry = getSymbol64TableEntry(Symb);
476 MachO::nlist Entry = getSymbolTableEntry(Symb);
479 return object_error::success;
483 MachOObjectFile::getSymbolFileOffset(DataRefImpl Symb,
484 uint64_t &Res) const {
485 nlist_base Entry = getSymbolTableEntryBase(this, Symb);
486 getSymbolAddress(Symb, Res);
490 SecRel.d.a = Entry.n_sect-1;
492 MachO::section_64 Sec = getSection64(SecRel);
493 Delta = Sec.offset - Sec.addr;
495 MachO::section Sec = getSection(SecRel);
496 Delta = Sec.offset - Sec.addr;
502 return object_error::success;
505 error_code MachOObjectFile::getSymbolAlignment(DataRefImpl DRI,
506 uint32_t &Result) const {
507 uint32_t flags = getSymbolFlags(DRI);
508 if (flags & SymbolRef::SF_Common) {
509 nlist_base Entry = getSymbolTableEntryBase(this, DRI);
510 Result = 1 << MachO::GET_COMM_ALIGN(Entry.n_desc);
514 return object_error::success;
517 error_code MachOObjectFile::getSymbolSize(DataRefImpl DRI,
518 uint64_t &Result) const {
519 uint64_t BeginOffset;
520 uint64_t EndOffset = 0;
521 uint8_t SectionIndex;
523 nlist_base Entry = getSymbolTableEntryBase(this, DRI);
525 getSymbolAddress(DRI, Value);
529 SectionIndex = Entry.n_sect;
531 uint32_t flags = getSymbolFlags(DRI);
532 if (flags & SymbolRef::SF_Common)
535 Result = UnknownAddressOrSize;
536 return object_error::success;
538 // Unfortunately symbols are unsorted so we need to touch all
539 // symbols from load command
540 for (const SymbolRef &Symbol : symbols()) {
541 DataRefImpl DRI = Symbol.getRawDataRefImpl();
542 Entry = getSymbolTableEntryBase(this, DRI);
543 getSymbolAddress(DRI, Value);
544 if (Entry.n_sect == SectionIndex && Value > BeginOffset)
545 if (!EndOffset || Value < EndOffset)
551 Sec.d.a = SectionIndex-1;
552 getSectionSize(Sec, Size);
553 getSectionAddress(Sec, EndOffset);
556 Result = EndOffset - BeginOffset;
557 return object_error::success;
560 error_code MachOObjectFile::getSymbolType(DataRefImpl Symb,
561 SymbolRef::Type &Res) const {
562 nlist_base Entry = getSymbolTableEntryBase(this, Symb);
563 uint8_t n_type = Entry.n_type;
565 Res = SymbolRef::ST_Other;
567 // If this is a STAB debugging symbol, we can do nothing more.
568 if (n_type & MachO::N_STAB) {
569 Res = SymbolRef::ST_Debug;
570 return object_error::success;
573 switch (n_type & MachO::N_TYPE) {
575 Res = SymbolRef::ST_Unknown;
578 Res = SymbolRef::ST_Function;
581 return object_error::success;
584 uint32_t MachOObjectFile::getSymbolFlags(DataRefImpl DRI) const {
585 nlist_base Entry = getSymbolTableEntryBase(this, DRI);
587 uint8_t MachOType = Entry.n_type;
588 uint16_t MachOFlags = Entry.n_desc;
590 uint32_t Result = SymbolRef::SF_None;
592 if ((MachOType & MachO::N_TYPE) == MachO::N_UNDF)
593 Result |= SymbolRef::SF_Undefined;
595 if (MachOType & MachO::N_STAB)
596 Result |= SymbolRef::SF_FormatSpecific;
598 if (MachOType & MachO::N_EXT) {
599 Result |= SymbolRef::SF_Global;
600 if ((MachOType & MachO::N_TYPE) == MachO::N_UNDF) {
602 getSymbolAddress(DRI, Value);
604 Result |= SymbolRef::SF_Common;
608 if (MachOFlags & (MachO::N_WEAK_REF | MachO::N_WEAK_DEF))
609 Result |= SymbolRef::SF_Weak;
611 if ((MachOType & MachO::N_TYPE) == MachO::N_ABS)
612 Result |= SymbolRef::SF_Absolute;
618 MachOObjectFile::getSymbolSection(DataRefImpl Symb,
619 section_iterator &Res) const {
620 nlist_base Entry = getSymbolTableEntryBase(this, Symb);
621 uint8_t index = Entry.n_sect;
628 Res = section_iterator(SectionRef(DRI, this));
631 return object_error::success;
634 void MachOObjectFile::moveSectionNext(DataRefImpl &Sec) const {
639 MachOObjectFile::getSectionName(DataRefImpl Sec, StringRef &Result) const {
640 ArrayRef<char> Raw = getSectionRawName(Sec);
641 Result = parseSegmentOrSectionName(Raw.data());
642 return object_error::success;
646 MachOObjectFile::getSectionAddress(DataRefImpl Sec, uint64_t &Res) const {
648 MachO::section_64 Sect = getSection64(Sec);
651 MachO::section Sect = getSection(Sec);
654 return object_error::success;
658 MachOObjectFile::getSectionSize(DataRefImpl Sec, uint64_t &Res) const {
660 MachO::section_64 Sect = getSection64(Sec);
663 MachO::section Sect = getSection(Sec);
667 return object_error::success;
671 MachOObjectFile::getSectionContents(DataRefImpl Sec, StringRef &Res) const {
676 MachO::section_64 Sect = getSection64(Sec);
677 Offset = Sect.offset;
680 MachO::section Sect = getSection(Sec);
681 Offset = Sect.offset;
685 Res = this->getData().substr(Offset, Size);
686 return object_error::success;
690 MachOObjectFile::getSectionAlignment(DataRefImpl Sec, uint64_t &Res) const {
693 MachO::section_64 Sect = getSection64(Sec);
696 MachO::section Sect = getSection(Sec);
700 Res = uint64_t(1) << Align;
701 return object_error::success;
705 MachOObjectFile::isSectionText(DataRefImpl Sec, bool &Res) const {
706 uint32_t Flags = getSectionFlags(this, Sec);
707 Res = Flags & MachO::S_ATTR_PURE_INSTRUCTIONS;
708 return object_error::success;
711 error_code MachOObjectFile::isSectionData(DataRefImpl DRI, bool &Result) const {
712 // FIXME: Unimplemented.
714 return object_error::success;
717 error_code MachOObjectFile::isSectionBSS(DataRefImpl DRI, bool &Result) const {
718 // FIXME: Unimplemented.
720 return object_error::success;
724 MachOObjectFile::isSectionRequiredForExecution(DataRefImpl Sec,
725 bool &Result) const {
726 // FIXME: Unimplemented.
728 return object_error::success;
731 error_code MachOObjectFile::isSectionVirtual(DataRefImpl Sec,
732 bool &Result) const {
733 // FIXME: Unimplemented.
735 return object_error::success;
739 MachOObjectFile::isSectionZeroInit(DataRefImpl Sec, bool &Res) const {
740 uint32_t Flags = getSectionFlags(this, Sec);
741 unsigned SectionType = Flags & MachO::SECTION_TYPE;
742 Res = SectionType == MachO::S_ZEROFILL ||
743 SectionType == MachO::S_GB_ZEROFILL;
744 return object_error::success;
747 error_code MachOObjectFile::isSectionReadOnlyData(DataRefImpl Sec,
748 bool &Result) const {
749 // Consider using the code from isSectionText to look for __const sections.
750 // Alternately, emit S_ATTR_PURE_INSTRUCTIONS and/or S_ATTR_SOME_INSTRUCTIONS
751 // to use section attributes to distinguish code from data.
753 // FIXME: Unimplemented.
755 return object_error::success;
759 MachOObjectFile::sectionContainsSymbol(DataRefImpl Sec, DataRefImpl Symb,
760 bool &Result) const {
762 this->getSymbolType(Symb, ST);
763 if (ST == SymbolRef::ST_Unknown) {
765 return object_error::success;
768 uint64_t SectBegin, SectEnd;
769 getSectionAddress(Sec, SectBegin);
770 getSectionSize(Sec, SectEnd);
771 SectEnd += SectBegin;
774 getSymbolAddress(Symb, SymAddr);
775 Result = (SymAddr >= SectBegin) && (SymAddr < SectEnd);
777 return object_error::success;
780 relocation_iterator MachOObjectFile::section_rel_begin(DataRefImpl Sec) const {
783 MachO::section_64 Sect = getSection64(Sec);
784 Offset = Sect.reloff;
786 MachO::section Sect = getSection(Sec);
787 Offset = Sect.reloff;
791 Ret.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
792 return relocation_iterator(RelocationRef(Ret, this));
796 MachOObjectFile::section_rel_end(DataRefImpl Sec) const {
800 MachO::section_64 Sect = getSection64(Sec);
801 Offset = Sect.reloff;
804 MachO::section Sect = getSection(Sec);
805 Offset = Sect.reloff;
809 const MachO::any_relocation_info *P =
810 reinterpret_cast<const MachO::any_relocation_info *>(getPtr(this, Offset));
813 Ret.p = reinterpret_cast<uintptr_t>(P + Num);
814 return relocation_iterator(RelocationRef(Ret, this));
817 bool MachOObjectFile::section_rel_empty(DataRefImpl Sec) const {
819 MachO::section_64 Sect = getSection64(Sec);
820 return Sect.nreloc == 0;
822 MachO::section Sect = getSection(Sec);
823 return Sect.nreloc == 0;
827 void MachOObjectFile::moveRelocationNext(DataRefImpl &Rel) const {
828 const MachO::any_relocation_info *P =
829 reinterpret_cast<const MachO::any_relocation_info *>(Rel.p);
830 Rel.p = reinterpret_cast<uintptr_t>(P + 1);
834 MachOObjectFile::getRelocationAddress(DataRefImpl Rel, uint64_t &Res) const {
835 report_fatal_error("getRelocationAddress not implemented in MachOObjectFile");
838 error_code MachOObjectFile::getRelocationOffset(DataRefImpl Rel,
839 uint64_t &Res) const {
840 MachO::any_relocation_info RE = getRelocation(Rel);
841 Res = getAnyRelocationAddress(RE);
842 return object_error::success;
846 MachOObjectFile::getRelocationSymbol(DataRefImpl Rel) const {
847 MachO::any_relocation_info RE = getRelocation(Rel);
848 uint32_t SymbolIdx = getPlainRelocationSymbolNum(RE);
849 bool isExtern = getPlainRelocationExternal(RE);
853 MachO::symtab_command S = getSymtabLoadCommand();
854 unsigned SymbolTableEntrySize = is64Bit() ?
855 sizeof(MachO::nlist_64) :
856 sizeof(MachO::nlist);
857 uint64_t Offset = S.symoff + SymbolIdx * SymbolTableEntrySize;
859 Sym.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
860 return symbol_iterator(SymbolRef(Sym, this));
863 error_code MachOObjectFile::getRelocationType(DataRefImpl Rel,
864 uint64_t &Res) const {
865 MachO::any_relocation_info RE = getRelocation(Rel);
866 Res = getAnyRelocationType(RE);
867 return object_error::success;
871 MachOObjectFile::getRelocationTypeName(DataRefImpl Rel,
872 SmallVectorImpl<char> &Result) const {
875 getRelocationType(Rel, RType);
877 unsigned Arch = this->getArch();
881 static const char *const Table[] = {
882 "GENERIC_RELOC_VANILLA",
883 "GENERIC_RELOC_PAIR",
884 "GENERIC_RELOC_SECTDIFF",
885 "GENERIC_RELOC_PB_LA_PTR",
886 "GENERIC_RELOC_LOCAL_SECTDIFF",
887 "GENERIC_RELOC_TLV" };
895 case Triple::x86_64: {
896 static const char *const Table[] = {
897 "X86_64_RELOC_UNSIGNED",
898 "X86_64_RELOC_SIGNED",
899 "X86_64_RELOC_BRANCH",
900 "X86_64_RELOC_GOT_LOAD",
902 "X86_64_RELOC_SUBTRACTOR",
903 "X86_64_RELOC_SIGNED_1",
904 "X86_64_RELOC_SIGNED_2",
905 "X86_64_RELOC_SIGNED_4",
906 "X86_64_RELOC_TLV" };
915 static const char *const Table[] = {
918 "ARM_RELOC_SECTDIFF",
919 "ARM_RELOC_LOCAL_SECTDIFF",
920 "ARM_RELOC_PB_LA_PTR",
922 "ARM_THUMB_RELOC_BR22",
923 "ARM_THUMB_32BIT_BRANCH",
925 "ARM_RELOC_HALF_SECTDIFF" };
934 case Triple::aarch64: {
935 static const char *const Table[] = {
936 "ARM64_RELOC_UNSIGNED", "ARM64_RELOC_SUBTRACTOR",
937 "ARM64_RELOC_BRANCH26", "ARM64_RELOC_PAGE21",
938 "ARM64_RELOC_PAGEOFF12", "ARM64_RELOC_GOT_LOAD_PAGE21",
939 "ARM64_RELOC_GOT_LOAD_PAGEOFF12", "ARM64_RELOC_POINTER_TO_GOT",
940 "ARM64_RELOC_TLVP_LOAD_PAGE21", "ARM64_RELOC_TLVP_LOAD_PAGEOFF12",
944 if (RType >= array_lengthof(Table))
951 static const char *const Table[] = {
960 "PPC_RELOC_SECTDIFF",
961 "PPC_RELOC_PB_LA_PTR",
962 "PPC_RELOC_HI16_SECTDIFF",
963 "PPC_RELOC_LO16_SECTDIFF",
964 "PPC_RELOC_HA16_SECTDIFF",
966 "PPC_RELOC_LO14_SECTDIFF",
967 "PPC_RELOC_LOCAL_SECTDIFF" };
975 case Triple::UnknownArch:
979 Result.append(res.begin(), res.end());
980 return object_error::success;
984 MachOObjectFile::getRelocationValueString(DataRefImpl Rel,
985 SmallVectorImpl<char> &Result) const {
986 MachO::any_relocation_info RE = getRelocation(Rel);
988 unsigned Arch = this->getArch();
991 raw_string_ostream fmt(fmtbuf);
992 unsigned Type = this->getAnyRelocationType(RE);
993 bool IsPCRel = this->getAnyRelocationPCRel(RE);
995 // Determine any addends that should be displayed with the relocation.
996 // These require decoding the relocation type, which is triple-specific.
998 // X86_64 has entirely custom relocation types.
999 if (Arch == Triple::x86_64) {
1000 bool isPCRel = getAnyRelocationPCRel(RE);
1003 case MachO::X86_64_RELOC_GOT_LOAD:
1004 case MachO::X86_64_RELOC_GOT: {
1005 printRelocationTargetName(this, RE, fmt);
1007 if (isPCRel) fmt << "PCREL";
1010 case MachO::X86_64_RELOC_SUBTRACTOR: {
1011 DataRefImpl RelNext = Rel;
1013 MachO::any_relocation_info RENext = getRelocation(RelNext);
1015 // X86_64_RELOC_SUBTRACTOR must be followed by a relocation of type
1016 // X86_64_RELOC_UNSIGNED.
1017 // NOTE: Scattered relocations don't exist on x86_64.
1018 unsigned RType = getAnyRelocationType(RENext);
1019 if (RType != MachO::X86_64_RELOC_UNSIGNED)
1020 report_fatal_error("Expected X86_64_RELOC_UNSIGNED after "
1021 "X86_64_RELOC_SUBTRACTOR.");
1023 // The X86_64_RELOC_UNSIGNED contains the minuend symbol;
1024 // X86_64_RELOC_SUBTRACTOR contains the subtrahend.
1025 printRelocationTargetName(this, RENext, fmt);
1027 printRelocationTargetName(this, RE, fmt);
1030 case MachO::X86_64_RELOC_TLV:
1031 printRelocationTargetName(this, RE, fmt);
1033 if (isPCRel) fmt << "P";
1035 case MachO::X86_64_RELOC_SIGNED_1:
1036 printRelocationTargetName(this, RE, fmt);
1039 case MachO::X86_64_RELOC_SIGNED_2:
1040 printRelocationTargetName(this, RE, fmt);
1043 case MachO::X86_64_RELOC_SIGNED_4:
1044 printRelocationTargetName(this, RE, fmt);
1048 printRelocationTargetName(this, RE, fmt);
1051 // X86 and ARM share some relocation types in common.
1052 } else if (Arch == Triple::x86 || Arch == Triple::arm ||
1053 Arch == Triple::ppc) {
1054 // Generic relocation types...
1056 case MachO::GENERIC_RELOC_PAIR: // prints no info
1057 return object_error::success;
1058 case MachO::GENERIC_RELOC_SECTDIFF: {
1059 DataRefImpl RelNext = Rel;
1061 MachO::any_relocation_info RENext = getRelocation(RelNext);
1063 // X86 sect diff's must be followed by a relocation of type
1064 // GENERIC_RELOC_PAIR.
1065 unsigned RType = getAnyRelocationType(RENext);
1067 if (RType != MachO::GENERIC_RELOC_PAIR)
1068 report_fatal_error("Expected GENERIC_RELOC_PAIR after "
1069 "GENERIC_RELOC_SECTDIFF.");
1071 printRelocationTargetName(this, RE, fmt);
1073 printRelocationTargetName(this, RENext, fmt);
1078 if (Arch == Triple::x86 || Arch == Triple::ppc) {
1080 case MachO::GENERIC_RELOC_LOCAL_SECTDIFF: {
1081 DataRefImpl RelNext = Rel;
1083 MachO::any_relocation_info RENext = getRelocation(RelNext);
1085 // X86 sect diff's must be followed by a relocation of type
1086 // GENERIC_RELOC_PAIR.
1087 unsigned RType = getAnyRelocationType(RENext);
1088 if (RType != MachO::GENERIC_RELOC_PAIR)
1089 report_fatal_error("Expected GENERIC_RELOC_PAIR after "
1090 "GENERIC_RELOC_LOCAL_SECTDIFF.");
1092 printRelocationTargetName(this, RE, fmt);
1094 printRelocationTargetName(this, RENext, fmt);
1097 case MachO::GENERIC_RELOC_TLV: {
1098 printRelocationTargetName(this, RE, fmt);
1100 if (IsPCRel) fmt << "P";
1104 printRelocationTargetName(this, RE, fmt);
1106 } else { // ARM-specific relocations
1108 case MachO::ARM_RELOC_HALF:
1109 case MachO::ARM_RELOC_HALF_SECTDIFF: {
1110 // Half relocations steal a bit from the length field to encode
1111 // whether this is an upper16 or a lower16 relocation.
1112 bool isUpper = getAnyRelocationLength(RE) >> 1;
1115 fmt << ":upper16:(";
1117 fmt << ":lower16:(";
1118 printRelocationTargetName(this, RE, fmt);
1120 DataRefImpl RelNext = Rel;
1122 MachO::any_relocation_info RENext = getRelocation(RelNext);
1124 // ARM half relocs must be followed by a relocation of type
1126 unsigned RType = getAnyRelocationType(RENext);
1127 if (RType != MachO::ARM_RELOC_PAIR)
1128 report_fatal_error("Expected ARM_RELOC_PAIR after "
1131 // NOTE: The half of the target virtual address is stashed in the
1132 // address field of the secondary relocation, but we can't reverse
1133 // engineer the constant offset from it without decoding the movw/movt
1134 // instruction to find the other half in its immediate field.
1136 // ARM_RELOC_HALF_SECTDIFF encodes the second section in the
1137 // symbol/section pointer of the follow-on relocation.
1138 if (Type == MachO::ARM_RELOC_HALF_SECTDIFF) {
1140 printRelocationTargetName(this, RENext, fmt);
1147 printRelocationTargetName(this, RE, fmt);
1152 printRelocationTargetName(this, RE, fmt);
1155 Result.append(fmtbuf.begin(), fmtbuf.end());
1156 return object_error::success;
1160 MachOObjectFile::getRelocationHidden(DataRefImpl Rel, bool &Result) const {
1161 unsigned Arch = getArch();
1163 getRelocationType(Rel, Type);
1167 // On arches that use the generic relocations, GENERIC_RELOC_PAIR
1168 // is always hidden.
1169 if (Arch == Triple::x86 || Arch == Triple::arm || Arch == Triple::ppc) {
1170 if (Type == MachO::GENERIC_RELOC_PAIR) Result = true;
1171 } else if (Arch == Triple::x86_64) {
1172 // On x86_64, X86_64_RELOC_UNSIGNED is hidden only when it follows
1173 // an X86_64_RELOC_SUBTRACTOR.
1174 if (Type == MachO::X86_64_RELOC_UNSIGNED && Rel.d.a > 0) {
1175 DataRefImpl RelPrev = Rel;
1178 getRelocationType(RelPrev, PrevType);
1179 if (PrevType == MachO::X86_64_RELOC_SUBTRACTOR)
1184 return object_error::success;
1187 error_code MachOObjectFile::getLibraryNext(DataRefImpl LibData,
1188 LibraryRef &Res) const {
1189 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1192 error_code MachOObjectFile::getLibraryPath(DataRefImpl LibData,
1193 StringRef &Res) const {
1194 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1197 basic_symbol_iterator MachOObjectFile::symbol_begin_impl() const {
1200 return basic_symbol_iterator(SymbolRef(DRI, this));
1202 MachO::symtab_command Symtab = getSymtabLoadCommand();
1203 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Symtab.symoff));
1204 return basic_symbol_iterator(SymbolRef(DRI, this));
1207 basic_symbol_iterator MachOObjectFile::symbol_end_impl() const {
1210 return basic_symbol_iterator(SymbolRef(DRI, this));
1212 MachO::symtab_command Symtab = getSymtabLoadCommand();
1213 unsigned SymbolTableEntrySize = is64Bit() ?
1214 sizeof(MachO::nlist_64) :
1215 sizeof(MachO::nlist);
1216 unsigned Offset = Symtab.symoff +
1217 Symtab.nsyms * SymbolTableEntrySize;
1218 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
1219 return basic_symbol_iterator(SymbolRef(DRI, this));
1222 section_iterator MachOObjectFile::section_begin() const {
1224 return section_iterator(SectionRef(DRI, this));
1227 section_iterator MachOObjectFile::section_end() const {
1229 DRI.d.a = Sections.size();
1230 return section_iterator(SectionRef(DRI, this));
1233 library_iterator MachOObjectFile::needed_library_begin() const {
1235 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1238 library_iterator MachOObjectFile::needed_library_end() const {
1240 report_fatal_error("Needed libraries unimplemented in MachOObjectFile");
1243 uint8_t MachOObjectFile::getBytesInAddress() const {
1244 return is64Bit() ? 8 : 4;
1247 StringRef MachOObjectFile::getFileFormatName() const {
1248 unsigned CPUType = getCPUType(this);
1251 case llvm::MachO::CPU_TYPE_I386:
1252 return "Mach-O 32-bit i386";
1253 case llvm::MachO::CPU_TYPE_ARM:
1254 return "Mach-O arm";
1255 case llvm::MachO::CPU_TYPE_POWERPC:
1256 return "Mach-O 32-bit ppc";
1258 assert((CPUType & llvm::MachO::CPU_ARCH_ABI64) == 0 &&
1259 "64-bit object file when we're not 64-bit?");
1260 return "Mach-O 32-bit unknown";
1264 // Make sure the cpu type has the correct mask.
1265 assert((CPUType & llvm::MachO::CPU_ARCH_ABI64)
1266 == llvm::MachO::CPU_ARCH_ABI64 &&
1267 "32-bit object file when we're 64-bit?");
1270 case llvm::MachO::CPU_TYPE_X86_64:
1271 return "Mach-O 64-bit x86-64";
1272 case llvm::MachO::CPU_TYPE_ARM64:
1273 return "Mach-O arm64";
1274 case llvm::MachO::CPU_TYPE_POWERPC64:
1275 return "Mach-O 64-bit ppc64";
1277 return "Mach-O 64-bit unknown";
1281 Triple::ArchType MachOObjectFile::getArch(uint32_t CPUType) {
1283 case llvm::MachO::CPU_TYPE_I386:
1285 case llvm::MachO::CPU_TYPE_X86_64:
1286 return Triple::x86_64;
1287 case llvm::MachO::CPU_TYPE_ARM:
1289 case llvm::MachO::CPU_TYPE_ARM64:
1290 return Triple::arm64;
1291 case llvm::MachO::CPU_TYPE_POWERPC:
1293 case llvm::MachO::CPU_TYPE_POWERPC64:
1294 return Triple::ppc64;
1296 return Triple::UnknownArch;
1300 unsigned MachOObjectFile::getArch() const {
1301 return getArch(getCPUType(this));
1304 StringRef MachOObjectFile::getLoadName() const {
1306 report_fatal_error("get_load_name() unimplemented in MachOObjectFile");
1309 relocation_iterator MachOObjectFile::section_rel_begin(unsigned Index) const {
1312 return section_rel_begin(DRI);
1315 relocation_iterator MachOObjectFile::section_rel_end(unsigned Index) const {
1318 return section_rel_end(DRI);
1321 dice_iterator MachOObjectFile::begin_dices() const {
1323 if (!DataInCodeLoadCmd)
1324 return dice_iterator(DiceRef(DRI, this));
1326 MachO::linkedit_data_command DicLC = getDataInCodeLoadCommand();
1327 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, DicLC.dataoff));
1328 return dice_iterator(DiceRef(DRI, this));
1331 dice_iterator MachOObjectFile::end_dices() const {
1333 if (!DataInCodeLoadCmd)
1334 return dice_iterator(DiceRef(DRI, this));
1336 MachO::linkedit_data_command DicLC = getDataInCodeLoadCommand();
1337 unsigned Offset = DicLC.dataoff + DicLC.datasize;
1338 DRI.p = reinterpret_cast<uintptr_t>(getPtr(this, Offset));
1339 return dice_iterator(DiceRef(DRI, this));
1343 MachOObjectFile::getSectionFinalSegmentName(DataRefImpl Sec) const {
1344 ArrayRef<char> Raw = getSectionRawFinalSegmentName(Sec);
1345 return parseSegmentOrSectionName(Raw.data());
1349 MachOObjectFile::getSectionRawName(DataRefImpl Sec) const {
1350 const section_base *Base =
1351 reinterpret_cast<const section_base *>(Sections[Sec.d.a]);
1352 return ArrayRef<char>(Base->sectname);
1356 MachOObjectFile::getSectionRawFinalSegmentName(DataRefImpl Sec) const {
1357 const section_base *Base =
1358 reinterpret_cast<const section_base *>(Sections[Sec.d.a]);
1359 return ArrayRef<char>(Base->segname);
1363 MachOObjectFile::isRelocationScattered(const MachO::any_relocation_info &RE)
1365 if (getCPUType(this) == MachO::CPU_TYPE_X86_64)
1367 return getPlainRelocationAddress(RE) & MachO::R_SCATTERED;
1370 unsigned MachOObjectFile::getPlainRelocationSymbolNum(
1371 const MachO::any_relocation_info &RE) const {
1372 if (isLittleEndian())
1373 return RE.r_word1 & 0xffffff;
1374 return RE.r_word1 >> 8;
1377 bool MachOObjectFile::getPlainRelocationExternal(
1378 const MachO::any_relocation_info &RE) const {
1379 if (isLittleEndian())
1380 return (RE.r_word1 >> 27) & 1;
1381 return (RE.r_word1 >> 4) & 1;
1384 bool MachOObjectFile::getScatteredRelocationScattered(
1385 const MachO::any_relocation_info &RE) const {
1386 return RE.r_word0 >> 31;
1389 uint32_t MachOObjectFile::getScatteredRelocationValue(
1390 const MachO::any_relocation_info &RE) const {
1394 unsigned MachOObjectFile::getAnyRelocationAddress(
1395 const MachO::any_relocation_info &RE) const {
1396 if (isRelocationScattered(RE))
1397 return getScatteredRelocationAddress(RE);
1398 return getPlainRelocationAddress(RE);
1401 unsigned MachOObjectFile::getAnyRelocationPCRel(
1402 const MachO::any_relocation_info &RE) const {
1403 if (isRelocationScattered(RE))
1404 return getScatteredRelocationPCRel(this, RE);
1405 return getPlainRelocationPCRel(this, RE);
1408 unsigned MachOObjectFile::getAnyRelocationLength(
1409 const MachO::any_relocation_info &RE) const {
1410 if (isRelocationScattered(RE))
1411 return getScatteredRelocationLength(RE);
1412 return getPlainRelocationLength(this, RE);
1416 MachOObjectFile::getAnyRelocationType(
1417 const MachO::any_relocation_info &RE) const {
1418 if (isRelocationScattered(RE))
1419 return getScatteredRelocationType(RE);
1420 return getPlainRelocationType(this, RE);
1424 MachOObjectFile::getRelocationSection(
1425 const MachO::any_relocation_info &RE) const {
1426 if (isRelocationScattered(RE) || getPlainRelocationExternal(RE))
1427 return *section_end();
1428 unsigned SecNum = getPlainRelocationSymbolNum(RE) - 1;
1431 return SectionRef(DRI, this);
1434 MachOObjectFile::LoadCommandInfo
1435 MachOObjectFile::getFirstLoadCommandInfo() const {
1436 MachOObjectFile::LoadCommandInfo Load;
1438 unsigned HeaderSize = is64Bit() ? sizeof(MachO::mach_header_64) :
1439 sizeof(MachO::mach_header);
1440 Load.Ptr = getPtr(this, HeaderSize);
1441 Load.C = getStruct<MachO::load_command>(this, Load.Ptr);
1445 MachOObjectFile::LoadCommandInfo
1446 MachOObjectFile::getNextLoadCommandInfo(const LoadCommandInfo &L) const {
1447 MachOObjectFile::LoadCommandInfo Next;
1448 Next.Ptr = L.Ptr + L.C.cmdsize;
1449 Next.C = getStruct<MachO::load_command>(this, Next.Ptr);
1453 MachO::section MachOObjectFile::getSection(DataRefImpl DRI) const {
1454 return getStruct<MachO::section>(this, Sections[DRI.d.a]);
1457 MachO::section_64 MachOObjectFile::getSection64(DataRefImpl DRI) const {
1458 return getStruct<MachO::section_64>(this, Sections[DRI.d.a]);
1461 MachO::section MachOObjectFile::getSection(const LoadCommandInfo &L,
1462 unsigned Index) const {
1463 const char *Sec = getSectionPtr(this, L, Index);
1464 return getStruct<MachO::section>(this, Sec);
1467 MachO::section_64 MachOObjectFile::getSection64(const LoadCommandInfo &L,
1468 unsigned Index) const {
1469 const char *Sec = getSectionPtr(this, L, Index);
1470 return getStruct<MachO::section_64>(this, Sec);
1474 MachOObjectFile::getSymbolTableEntry(DataRefImpl DRI) const {
1475 const char *P = reinterpret_cast<const char *>(DRI.p);
1476 return getStruct<MachO::nlist>(this, P);
1480 MachOObjectFile::getSymbol64TableEntry(DataRefImpl DRI) const {
1481 const char *P = reinterpret_cast<const char *>(DRI.p);
1482 return getStruct<MachO::nlist_64>(this, P);
1485 MachO::linkedit_data_command
1486 MachOObjectFile::getLinkeditDataLoadCommand(const LoadCommandInfo &L) const {
1487 return getStruct<MachO::linkedit_data_command>(this, L.Ptr);
1490 MachO::segment_command
1491 MachOObjectFile::getSegmentLoadCommand(const LoadCommandInfo &L) const {
1492 return getStruct<MachO::segment_command>(this, L.Ptr);
1495 MachO::segment_command_64
1496 MachOObjectFile::getSegment64LoadCommand(const LoadCommandInfo &L) const {
1497 return getStruct<MachO::segment_command_64>(this, L.Ptr);
1500 MachO::linker_options_command
1501 MachOObjectFile::getLinkerOptionsLoadCommand(const LoadCommandInfo &L) const {
1502 return getStruct<MachO::linker_options_command>(this, L.Ptr);
1505 MachO::version_min_command
1506 MachOObjectFile::getVersionMinLoadCommand(const LoadCommandInfo &L) const {
1507 return getStruct<MachO::version_min_command>(this, L.Ptr);
1510 MachO::any_relocation_info
1511 MachOObjectFile::getRelocation(DataRefImpl Rel) const {
1512 const char *P = reinterpret_cast<const char *>(Rel.p);
1513 return getStruct<MachO::any_relocation_info>(this, P);
1516 MachO::data_in_code_entry
1517 MachOObjectFile::getDice(DataRefImpl Rel) const {
1518 const char *P = reinterpret_cast<const char *>(Rel.p);
1519 return getStruct<MachO::data_in_code_entry>(this, P);
1522 MachO::mach_header MachOObjectFile::getHeader() const {
1523 return getStruct<MachO::mach_header>(this, getPtr(this, 0));
1526 MachO::mach_header_64 MachOObjectFile::getHeader64() const {
1527 return getStruct<MachO::mach_header_64>(this, getPtr(this, 0));
1530 uint32_t MachOObjectFile::getIndirectSymbolTableEntry(
1531 const MachO::dysymtab_command &DLC,
1532 unsigned Index) const {
1533 uint64_t Offset = DLC.indirectsymoff + Index * sizeof(uint32_t);
1534 return getStruct<uint32_t>(this, getPtr(this, Offset));
1537 MachO::data_in_code_entry
1538 MachOObjectFile::getDataInCodeTableEntry(uint32_t DataOffset,
1539 unsigned Index) const {
1540 uint64_t Offset = DataOffset + Index * sizeof(MachO::data_in_code_entry);
1541 return getStruct<MachO::data_in_code_entry>(this, getPtr(this, Offset));
1544 MachO::symtab_command MachOObjectFile::getSymtabLoadCommand() const {
1545 return getStruct<MachO::symtab_command>(this, SymtabLoadCmd);
1548 MachO::dysymtab_command MachOObjectFile::getDysymtabLoadCommand() const {
1549 return getStruct<MachO::dysymtab_command>(this, DysymtabLoadCmd);
1552 MachO::linkedit_data_command
1553 MachOObjectFile::getDataInCodeLoadCommand() const {
1554 if (DataInCodeLoadCmd)
1555 return getStruct<MachO::linkedit_data_command>(this, DataInCodeLoadCmd);
1557 // If there is no DataInCodeLoadCmd return a load command with zero'ed fields.
1558 MachO::linkedit_data_command Cmd;
1559 Cmd.cmd = MachO::LC_DATA_IN_CODE;
1560 Cmd.cmdsize = sizeof(MachO::linkedit_data_command);
1566 StringRef MachOObjectFile::getStringTableData() const {
1567 MachO::symtab_command S = getSymtabLoadCommand();
1568 return getData().substr(S.stroff, S.strsize);
1571 bool MachOObjectFile::is64Bit() const {
1572 return getType() == getMachOType(false, true) ||
1573 getType() == getMachOType(true, true);
1576 void MachOObjectFile::ReadULEB128s(uint64_t Index,
1577 SmallVectorImpl<uint64_t> &Out) const {
1578 DataExtractor extractor(ObjectFile::getData(), true, 0);
1580 uint32_t offset = Index;
1582 while (uint64_t delta = extractor.getULEB128(&offset)) {
1584 Out.push_back(data);
1588 ErrorOr<ObjectFile *> ObjectFile::createMachOObjectFile(MemoryBuffer *Buffer,
1590 StringRef Magic = Buffer->getBuffer().slice(0, 4);
1592 std::unique_ptr<MachOObjectFile> Ret;
1593 if (Magic == "\xFE\xED\xFA\xCE")
1594 Ret.reset(new MachOObjectFile(Buffer, false, false, EC, BufferOwned));
1595 else if (Magic == "\xCE\xFA\xED\xFE")
1596 Ret.reset(new MachOObjectFile(Buffer, true, false, EC, BufferOwned));
1597 else if (Magic == "\xFE\xED\xFA\xCF")
1598 Ret.reset(new MachOObjectFile(Buffer, false, true, EC, BufferOwned));
1599 else if (Magic == "\xCF\xFA\xED\xFE")
1600 Ret.reset(new MachOObjectFile(Buffer, true, true, EC, BufferOwned));
1603 return object_error::parse_failed;
1608 return Ret.release();
1611 } // end namespace object
1612 } // end namespace llvm