X-Git-Url: http://demsky.eecs.uci.edu/git/?a=blobdiff_plain;f=tools%2Fllvm-objdump%2FMachODump.cpp;h=a2f3bc8e8a7c2c2f28389e6669eddf462855fa82;hb=005b231d452c9d7f3db00ce746f9e40a032630fd;hp=042ac2dd8a5df29d786fdb64677bcdc97fe88613;hpb=0602444f70ea5d578817f68d7003d34c35522cff;p=oota-llvm.git diff --git a/tools/llvm-objdump/MachODump.cpp b/tools/llvm-objdump/MachODump.cpp index 042ac2dd8a5..a2f3bc8e8a7 100644 --- a/tools/llvm-objdump/MachODump.cpp +++ b/tools/llvm-objdump/MachODump.cpp @@ -13,11 +13,13 @@ #include "llvm-objdump.h" #include "llvm-c/Disassembler.h" +#include "llvm/ADT/Optional.h" #include "llvm/ADT/STLExtras.h" #include "llvm/ADT/StringExtras.h" #include "llvm/ADT/Triple.h" #include "llvm/Config/config.h" -#include "llvm/DebugInfo/DWARF/DIContext.h" +#include "llvm/DebugInfo/DIContext.h" +#include "llvm/DebugInfo/DWARF/DWARFContext.h" #include "llvm/MC/MCAsmInfo.h" #include "llvm/MC/MCContext.h" #include "llvm/MC/MCDisassembler.h" @@ -63,9 +65,8 @@ static cl::opt DSYMFile("dsym", static cl::opt FullLeadingAddr("full-leading-addr", cl::desc("Print full leading address")); -static cl::opt - PrintImmHex("print-imm-hex", - cl::desc("Use hex format for immediate values")); +static cl::opt NoLeadingAddr("no-leading-addr", + cl::desc("Print no leading address")); cl::opt llvm::UniversalHeaders("universal-headers", cl::desc("Print Mach-O universal headers " @@ -76,6 +77,12 @@ cl::opt cl::desc("Print archive headers for Mach-O archives " "(requires -macho)")); +cl::opt + ArchiveMemberOffsets("archive-member-offsets", + cl::desc("Print the offset to each archive member for " + "Mach-O archives (requires -macho and " + "-archive-headers)")); + cl::opt llvm::IndirectSymbols("indirect-symbols", cl::desc("Print indirect symbol table for Mach-O " @@ -91,14 +98,43 @@ cl::opt cl::desc("Print the linker optimization hints for " "Mach-O objects (requires -macho)")); -cl::list - llvm::DumpSections("section", - cl::desc("Prints the specified segment,section for " - "Mach-O objects (requires -macho)")); +cl::opt + llvm::InfoPlist("info-plist", + cl::desc("Print the info plist section as strings for " + "Mach-O objects (requires -macho)")); + +cl::opt + llvm::DylibsUsed("dylibs-used", + cl::desc("Print the shared libraries used for linked " + "Mach-O files (requires -macho)")); + +cl::opt + llvm::DylibId("dylib-id", + cl::desc("Print the shared library's id for the dylib Mach-O " + "file (requires -macho)")); + +cl::opt + llvm::NonVerbose("non-verbose", + cl::desc("Print the info for Mach-O objects in " + "non-verbose or numeric form (requires -macho)")); + +cl::opt + llvm::ObjcMetaData("objc-meta-data", + cl::desc("Print the Objective-C runtime meta data for " + "Mach-O files (requires -macho)")); + +cl::opt llvm::DisSymName( + "dis-symname", + cl::desc("disassemble just this symbol's instructions (requires -macho")); + +static cl::opt NoSymbolicOperands( + "no-symbolic-operands", + cl::desc("do not symbolic operands when disassembling (requires -macho)")); static cl::list ArchFlags("arch", cl::desc("architecture(s) from a Mach-O file to dump"), cl::ZeroOrMore); + bool ArchAll = false; static std::string ThumbTripleName; @@ -136,19 +172,8 @@ static const Target *GetTarget(const MachOObjectFile *MachOObj, struct SymbolSorter { bool operator()(const SymbolRef &A, const SymbolRef &B) { - SymbolRef::Type AType, BType; - A.getType(AType); - B.getType(BType); - - uint64_t AAddr, BAddr; - if (AType != SymbolRef::ST_Function) - AAddr = 0; - else - A.getAddress(AAddr); - if (BType != SymbolRef::ST_Function) - BAddr = 0; - else - B.getAddress(BAddr); + uint64_t AAddr = (A.getType() != SymbolRef::ST_Function) ? 0 : A.getValue(); + uint64_t BAddr = (B.getType() != SymbolRef::ST_Function) ? 0 : B.getValue(); return AAddr < BAddr; } }; @@ -173,7 +198,7 @@ static bool compareDiceTableEntries(const DiceTableEntry &i, return j.first >= i.first && j.first < i.first + Length; } -static uint64_t DumpDataInCode(const char *bytes, uint64_t Length, +static uint64_t DumpDataInCode(const uint8_t *bytes, uint64_t Length, unsigned short Kind) { uint32_t Value, Size = 1; @@ -182,19 +207,19 @@ static uint64_t DumpDataInCode(const char *bytes, uint64_t Length, case MachO::DICE_KIND_DATA: if (Length >= 4) { if (!NoShowRawInsn) - DumpBytes(StringRef(bytes, 4)); + dumpBytes(makeArrayRef(bytes, 4), outs()); Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0]; outs() << "\t.long " << Value; Size = 4; } else if (Length >= 2) { if (!NoShowRawInsn) - DumpBytes(StringRef(bytes, 2)); + dumpBytes(makeArrayRef(bytes, 2), outs()); Value = bytes[1] << 8 | bytes[0]; outs() << "\t.short " << Value; Size = 2; } else { if (!NoShowRawInsn) - DumpBytes(StringRef(bytes, 2)); + dumpBytes(makeArrayRef(bytes, 2), outs()); Value = bytes[0]; outs() << "\t.byte " << Value; Size = 1; @@ -206,14 +231,14 @@ static uint64_t DumpDataInCode(const char *bytes, uint64_t Length, break; case MachO::DICE_KIND_JUMP_TABLE8: if (!NoShowRawInsn) - DumpBytes(StringRef(bytes, 1)); + dumpBytes(makeArrayRef(bytes, 1), outs()); Value = bytes[0]; outs() << "\t.byte " << format("%3u", Value) << "\t@ KIND_JUMP_TABLE8\n"; Size = 1; break; case MachO::DICE_KIND_JUMP_TABLE16: if (!NoShowRawInsn) - DumpBytes(StringRef(bytes, 2)); + dumpBytes(makeArrayRef(bytes, 2), outs()); Value = bytes[1] << 8 | bytes[0]; outs() << "\t.short " << format("%5u", Value & 0xffff) << "\t@ KIND_JUMP_TABLE16\n"; @@ -222,7 +247,7 @@ static uint64_t DumpDataInCode(const char *bytes, uint64_t Length, case MachO::DICE_KIND_JUMP_TABLE32: case MachO::DICE_KIND_ABS_JUMP_TABLE32: if (!NoShowRawInsn) - DumpBytes(StringRef(bytes, 4)); + dumpBytes(makeArrayRef(bytes, 4), outs()); Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0]; outs() << "\t.long " << Value; if (Kind == MachO::DICE_KIND_JUMP_TABLE32) @@ -235,16 +260,16 @@ static uint64_t DumpDataInCode(const char *bytes, uint64_t Length, return Size; } -static void getSectionsAndSymbols(const MachO::mach_header Header, - MachOObjectFile *MachOObj, +static void getSectionsAndSymbols(MachOObjectFile *MachOObj, std::vector &Sections, std::vector &Symbols, SmallVectorImpl &FoundFns, uint64_t &BaseSegmentAddress) { for (const SymbolRef &Symbol : MachOObj->symbols()) { - StringRef SymName; - Symbol.getName(SymName); - if (!SymName.startswith("ltmp")) + ErrorOr SymName = Symbol.getName(); + if (std::error_code EC = SymName.getError()) + report_fatal_error(EC.message()); + if (!SymName->startswith("ltmp")) Symbols.push_back(Symbol); } @@ -254,10 +279,8 @@ static void getSectionsAndSymbols(const MachO::mach_header Header, Sections.push_back(Section); } - MachOObjectFile::LoadCommandInfo Command = - MachOObj->getFirstLoadCommandInfo(); bool BaseSegmentAddressSet = false; - for (unsigned i = 0;; ++i) { + for (const auto &Command : MachOObj->load_commands()) { if (Command.C.cmd == MachO::LC_FUNCTION_STARTS) { // We found a function starts segment, parse the addresses for later // consumption. @@ -273,11 +296,6 @@ static void getSectionsAndSymbols(const MachO::mach_header Header, BaseSegmentAddress = SLC.vmaddr; } } - - if (i == Header.ncmds - 1) - break; - else - Command = MachOObj->getNextLoadCommandInfo(Command); } } @@ -323,24 +341,25 @@ static void PrintIndirectSymbolTable(MachOObjectFile *O, bool verbose, continue; } outs() << format("%5u ", indirect_symbol); - MachO::symtab_command Symtab = O->getSymtabLoadCommand(); - if (indirect_symbol < Symtab.nsyms) { - symbol_iterator Sym = O->getSymbolByIndex(indirect_symbol); - SymbolRef Symbol = *Sym; - StringRef SymName; - Symbol.getName(SymName); - outs() << SymName; - } else { - outs() << "?"; + if (verbose) { + MachO::symtab_command Symtab = O->getSymtabLoadCommand(); + if (indirect_symbol < Symtab.nsyms) { + symbol_iterator Sym = O->getSymbolByIndex(indirect_symbol); + SymbolRef Symbol = *Sym; + ErrorOr SymName = Symbol.getName(); + if (std::error_code EC = SymName.getError()) + report_fatal_error(EC.message()); + outs() << *SymName; + } else { + outs() << "?"; + } } outs() << "\n"; } } static void PrintIndirectSymbols(MachOObjectFile *O, bool verbose) { - uint32_t LoadCommandCount = O->getHeader().ncmds; - MachOObjectFile::LoadCommandInfo Load = O->getFirstLoadCommandInfo(); - for (unsigned I = 0;; ++I) { + for (const auto &Load : O->load_commands()) { if (Load.C.cmd == MachO::LC_SEGMENT_64) { MachO::segment_command_64 Seg = O->getSegment64LoadCommand(Load); for (unsigned J = 0; J < Seg.nsects; ++J) { @@ -398,10 +417,6 @@ static void PrintIndirectSymbols(MachOObjectFile *O, bool verbose) { } } } - if (I == LoadCommandCount - 1) - break; - else - Load = O->getNextLoadCommandInfo(Load); } } @@ -504,21 +519,70 @@ static void PrintLinkOptHints(MachOObjectFile *O) { } } +static void PrintDylibs(MachOObjectFile *O, bool JustId) { + unsigned Index = 0; + for (const auto &Load : O->load_commands()) { + if ((JustId && Load.C.cmd == MachO::LC_ID_DYLIB) || + (!JustId && (Load.C.cmd == MachO::LC_ID_DYLIB || + Load.C.cmd == MachO::LC_LOAD_DYLIB || + Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB || + Load.C.cmd == MachO::LC_REEXPORT_DYLIB || + Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB || + Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB))) { + MachO::dylib_command dl = O->getDylibIDLoadCommand(Load); + if (dl.dylib.name < dl.cmdsize) { + const char *p = (const char *)(Load.Ptr) + dl.dylib.name; + if (JustId) + outs() << p << "\n"; + else { + outs() << "\t" << p; + outs() << " (compatibility version " + << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "." + << ((dl.dylib.compatibility_version >> 8) & 0xff) << "." + << (dl.dylib.compatibility_version & 0xff) << ","; + outs() << " current version " + << ((dl.dylib.current_version >> 16) & 0xffff) << "." + << ((dl.dylib.current_version >> 8) & 0xff) << "." + << (dl.dylib.current_version & 0xff) << ")\n"; + } + } else { + outs() << "\tBad offset (" << dl.dylib.name << ") for name of "; + if (Load.C.cmd == MachO::LC_ID_DYLIB) + outs() << "LC_ID_DYLIB "; + else if (Load.C.cmd == MachO::LC_LOAD_DYLIB) + outs() << "LC_LOAD_DYLIB "; + else if (Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB) + outs() << "LC_LOAD_WEAK_DYLIB "; + else if (Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB) + outs() << "LC_LAZY_LOAD_DYLIB "; + else if (Load.C.cmd == MachO::LC_REEXPORT_DYLIB) + outs() << "LC_REEXPORT_DYLIB "; + else if (Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) + outs() << "LC_LOAD_UPWARD_DYLIB "; + else + outs() << "LC_??? "; + outs() << "command " << Index++ << "\n"; + } + } + } +} + typedef DenseMap SymbolAddressMap; static void CreateSymbolAddressMap(MachOObjectFile *O, SymbolAddressMap *AddrMap) { // Create a map of symbol addresses to symbol names. for (const SymbolRef &Symbol : O->symbols()) { - SymbolRef::Type ST; - Symbol.getType(ST); + SymbolRef::Type ST = Symbol.getType(); if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data || ST == SymbolRef::ST_Other) { - uint64_t Address; - Symbol.getAddress(Address); - StringRef SymName; - Symbol.getName(SymName); - (*AddrMap)[Address] = SymName; + uint64_t Address = Symbol.getValue(); + ErrorOr SymNameOrErr = Symbol.getName(); + if (std::error_code EC = SymNameOrErr.getError()) + report_fatal_error(EC.message()); + StringRef SymName = *SymNameOrErr; + if (!SymName.startswith(".objc")) + (*AddrMap)[Address] = SymName; } } } @@ -537,15 +601,320 @@ static const char *GuessSymbolName(uint64_t value, SymbolAddressMap *AddrMap) { return SymbolName; } +static void DumpCstringChar(const char c) { + char p[2]; + p[0] = c; + p[1] = '\0'; + outs().write_escaped(p); +} + +static void DumpCstringSection(MachOObjectFile *O, const char *sect, + uint32_t sect_size, uint64_t sect_addr, + bool print_addresses) { + for (uint32_t i = 0; i < sect_size; i++) { + if (print_addresses) { + if (O->is64Bit()) + outs() << format("%016" PRIx64, sect_addr + i) << " "; + else + outs() << format("%08" PRIx64, sect_addr + i) << " "; + } + for (; i < sect_size && sect[i] != '\0'; i++) + DumpCstringChar(sect[i]); + if (i < sect_size && sect[i] == '\0') + outs() << "\n"; + } +} + +static void DumpLiteral4(uint32_t l, float f) { + outs() << format("0x%08" PRIx32, l); + if ((l & 0x7f800000) != 0x7f800000) + outs() << format(" (%.16e)\n", f); + else { + if (l == 0x7f800000) + outs() << " (+Infinity)\n"; + else if (l == 0xff800000) + outs() << " (-Infinity)\n"; + else if ((l & 0x00400000) == 0x00400000) + outs() << " (non-signaling Not-a-Number)\n"; + else + outs() << " (signaling Not-a-Number)\n"; + } +} + +static void DumpLiteral4Section(MachOObjectFile *O, const char *sect, + uint32_t sect_size, uint64_t sect_addr, + bool print_addresses) { + for (uint32_t i = 0; i < sect_size; i += sizeof(float)) { + if (print_addresses) { + if (O->is64Bit()) + outs() << format("%016" PRIx64, sect_addr + i) << " "; + else + outs() << format("%08" PRIx64, sect_addr + i) << " "; + } + float f; + memcpy(&f, sect + i, sizeof(float)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(f); + uint32_t l; + memcpy(&l, sect + i, sizeof(uint32_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(l); + DumpLiteral4(l, f); + } +} + +static void DumpLiteral8(MachOObjectFile *O, uint32_t l0, uint32_t l1, + double d) { + outs() << format("0x%08" PRIx32, l0) << " " << format("0x%08" PRIx32, l1); + uint32_t Hi, Lo; + Hi = (O->isLittleEndian()) ? l1 : l0; + Lo = (O->isLittleEndian()) ? l0 : l1; + + // Hi is the high word, so this is equivalent to if(isfinite(d)) + if ((Hi & 0x7ff00000) != 0x7ff00000) + outs() << format(" (%.16e)\n", d); + else { + if (Hi == 0x7ff00000 && Lo == 0) + outs() << " (+Infinity)\n"; + else if (Hi == 0xfff00000 && Lo == 0) + outs() << " (-Infinity)\n"; + else if ((Hi & 0x00080000) == 0x00080000) + outs() << " (non-signaling Not-a-Number)\n"; + else + outs() << " (signaling Not-a-Number)\n"; + } +} + +static void DumpLiteral8Section(MachOObjectFile *O, const char *sect, + uint32_t sect_size, uint64_t sect_addr, + bool print_addresses) { + for (uint32_t i = 0; i < sect_size; i += sizeof(double)) { + if (print_addresses) { + if (O->is64Bit()) + outs() << format("%016" PRIx64, sect_addr + i) << " "; + else + outs() << format("%08" PRIx64, sect_addr + i) << " "; + } + double d; + memcpy(&d, sect + i, sizeof(double)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(d); + uint32_t l0, l1; + memcpy(&l0, sect + i, sizeof(uint32_t)); + memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) { + sys::swapByteOrder(l0); + sys::swapByteOrder(l1); + } + DumpLiteral8(O, l0, l1, d); + } +} + +static void DumpLiteral16(uint32_t l0, uint32_t l1, uint32_t l2, uint32_t l3) { + outs() << format("0x%08" PRIx32, l0) << " "; + outs() << format("0x%08" PRIx32, l1) << " "; + outs() << format("0x%08" PRIx32, l2) << " "; + outs() << format("0x%08" PRIx32, l3) << "\n"; +} + +static void DumpLiteral16Section(MachOObjectFile *O, const char *sect, + uint32_t sect_size, uint64_t sect_addr, + bool print_addresses) { + for (uint32_t i = 0; i < sect_size; i += 16) { + if (print_addresses) { + if (O->is64Bit()) + outs() << format("%016" PRIx64, sect_addr + i) << " "; + else + outs() << format("%08" PRIx64, sect_addr + i) << " "; + } + uint32_t l0, l1, l2, l3; + memcpy(&l0, sect + i, sizeof(uint32_t)); + memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t)); + memcpy(&l2, sect + i + 2 * sizeof(uint32_t), sizeof(uint32_t)); + memcpy(&l3, sect + i + 3 * sizeof(uint32_t), sizeof(uint32_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) { + sys::swapByteOrder(l0); + sys::swapByteOrder(l1); + sys::swapByteOrder(l2); + sys::swapByteOrder(l3); + } + DumpLiteral16(l0, l1, l2, l3); + } +} + +static void DumpLiteralPointerSection(MachOObjectFile *O, + const SectionRef &Section, + const char *sect, uint32_t sect_size, + uint64_t sect_addr, + bool print_addresses) { + // Collect the literal sections in this Mach-O file. + std::vector LiteralSections; + for (const SectionRef &Section : O->sections()) { + DataRefImpl Ref = Section.getRawDataRefImpl(); + uint32_t section_type; + if (O->is64Bit()) { + const MachO::section_64 Sec = O->getSection64(Ref); + section_type = Sec.flags & MachO::SECTION_TYPE; + } else { + const MachO::section Sec = O->getSection(Ref); + section_type = Sec.flags & MachO::SECTION_TYPE; + } + if (section_type == MachO::S_CSTRING_LITERALS || + section_type == MachO::S_4BYTE_LITERALS || + section_type == MachO::S_8BYTE_LITERALS || + section_type == MachO::S_16BYTE_LITERALS) + LiteralSections.push_back(Section); + } + + // Set the size of the literal pointer. + uint32_t lp_size = O->is64Bit() ? 8 : 4; + + // Collect the external relocation symbols for the literal pointers. + std::vector> Relocs; + for (const RelocationRef &Reloc : Section.relocations()) { + DataRefImpl Rel; + MachO::any_relocation_info RE; + bool isExtern = false; + Rel = Reloc.getRawDataRefImpl(); + RE = O->getRelocation(Rel); + isExtern = O->getPlainRelocationExternal(RE); + if (isExtern) { + uint64_t RelocOffset = Reloc.getOffset(); + symbol_iterator RelocSym = Reloc.getSymbol(); + Relocs.push_back(std::make_pair(RelocOffset, *RelocSym)); + } + } + array_pod_sort(Relocs.begin(), Relocs.end()); + + // Dump each literal pointer. + for (uint32_t i = 0; i < sect_size; i += lp_size) { + if (print_addresses) { + if (O->is64Bit()) + outs() << format("%016" PRIx64, sect_addr + i) << " "; + else + outs() << format("%08" PRIx64, sect_addr + i) << " "; + } + uint64_t lp; + if (O->is64Bit()) { + memcpy(&lp, sect + i, sizeof(uint64_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(lp); + } else { + uint32_t li; + memcpy(&li, sect + i, sizeof(uint32_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(li); + lp = li; + } + + // First look for an external relocation entry for this literal pointer. + auto Reloc = std::find_if( + Relocs.begin(), Relocs.end(), + [&](const std::pair &P) { return P.first == i; }); + if (Reloc != Relocs.end()) { + symbol_iterator RelocSym = Reloc->second; + ErrorOr SymName = RelocSym->getName(); + if (std::error_code EC = SymName.getError()) + report_fatal_error(EC.message()); + outs() << "external relocation entry for symbol:" << *SymName << "\n"; + continue; + } + + // For local references see what the section the literal pointer points to. + auto Sect = std::find_if(LiteralSections.begin(), LiteralSections.end(), + [&](const SectionRef &R) { + return lp >= R.getAddress() && + lp < R.getAddress() + R.getSize(); + }); + if (Sect == LiteralSections.end()) { + outs() << format("0x%" PRIx64, lp) << " (not in a literal section)\n"; + continue; + } + + uint64_t SectAddress = Sect->getAddress(); + uint64_t SectSize = Sect->getSize(); + + StringRef SectName; + Sect->getName(SectName); + DataRefImpl Ref = Sect->getRawDataRefImpl(); + StringRef SegmentName = O->getSectionFinalSegmentName(Ref); + outs() << SegmentName << ":" << SectName << ":"; + + uint32_t section_type; + if (O->is64Bit()) { + const MachO::section_64 Sec = O->getSection64(Ref); + section_type = Sec.flags & MachO::SECTION_TYPE; + } else { + const MachO::section Sec = O->getSection(Ref); + section_type = Sec.flags & MachO::SECTION_TYPE; + } + + StringRef BytesStr; + Sect->getContents(BytesStr); + const char *Contents = reinterpret_cast(BytesStr.data()); + + switch (section_type) { + case MachO::S_CSTRING_LITERALS: + for (uint64_t i = lp - SectAddress; i < SectSize && Contents[i] != '\0'; + i++) { + DumpCstringChar(Contents[i]); + } + outs() << "\n"; + break; + case MachO::S_4BYTE_LITERALS: + float f; + memcpy(&f, Contents + (lp - SectAddress), sizeof(float)); + uint32_t l; + memcpy(&l, Contents + (lp - SectAddress), sizeof(uint32_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) { + sys::swapByteOrder(f); + sys::swapByteOrder(l); + } + DumpLiteral4(l, f); + break; + case MachO::S_8BYTE_LITERALS: { + double d; + memcpy(&d, Contents + (lp - SectAddress), sizeof(double)); + uint32_t l0, l1; + memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t)); + memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t), + sizeof(uint32_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) { + sys::swapByteOrder(f); + sys::swapByteOrder(l0); + sys::swapByteOrder(l1); + } + DumpLiteral8(O, l0, l1, d); + break; + } + case MachO::S_16BYTE_LITERALS: { + uint32_t l0, l1, l2, l3; + memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t)); + memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t), + sizeof(uint32_t)); + memcpy(&l2, Contents + (lp - SectAddress) + 2 * sizeof(uint32_t), + sizeof(uint32_t)); + memcpy(&l3, Contents + (lp - SectAddress) + 3 * sizeof(uint32_t), + sizeof(uint32_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) { + sys::swapByteOrder(l0); + sys::swapByteOrder(l1); + sys::swapByteOrder(l2); + sys::swapByteOrder(l3); + } + DumpLiteral16(l0, l1, l2, l3); + break; + } + } + } +} + static void DumpInitTermPointerSection(MachOObjectFile *O, const char *sect, uint32_t sect_size, uint64_t sect_addr, SymbolAddressMap *AddrMap, bool verbose) { uint32_t stride; - if (O->is64Bit()) - stride = sizeof(uint64_t); - else - stride = sizeof(uint32_t); + stride = (O->is64Bit()) ? sizeof(uint64_t) : sizeof(uint32_t); for (uint32_t i = 0; i < sect_size; i += stride) { const char *SymbolName = nullptr; if (O->is64Bit()) { @@ -582,7 +951,7 @@ static void DumpRawSectionContents(MachOObjectFile *O, const char *sect, if (O->is64Bit()) outs() << format("%016" PRIx64, addr) << "\t"; else - outs() << format("%08" PRIx64, sect) << "\t"; + outs() << format("%08" PRIx64, addr) << "\t"; for (j = 0; j < 16 && i + j < size; j++) { uint8_t byte_word = *(sect + i + j); outs() << format("%02" PRIx32, (uint32_t)byte_word) << " "; @@ -616,13 +985,19 @@ static void DumpRawSectionContents(MachOObjectFile *O, const char *sect, } } -static void DumpSectionContents(MachOObjectFile *O, bool verbose) { +static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF, + StringRef DisSegName, StringRef DisSectName); +static void DumpProtocolSection(MachOObjectFile *O, const char *sect, + uint32_t size, uint32_t addr); + +static void DumpSectionContents(StringRef Filename, MachOObjectFile *O, + bool verbose) { SymbolAddressMap AddrMap; if (verbose) CreateSymbolAddressMap(O, &AddrMap); - for (unsigned i = 0; i < DumpSections.size(); ++i) { - StringRef DumpSection = DumpSections[i]; + for (unsigned i = 0; i < FilterSections.size(); ++i) { + StringRef DumpSection = FilterSections[i]; std::pair DumpSegSectName; DumpSegSectName = DumpSection.split(','); StringRef DumpSegName, DumpSectName; @@ -640,17 +1015,17 @@ static void DumpSectionContents(MachOObjectFile *O, bool verbose) { StringRef SegName = O->getSectionFinalSegmentName(Ref); if ((DumpSegName.empty() || SegName == DumpSegName) && (SectName == DumpSectName)) { - outs() << "Contents of (" << SegName << "," << SectName - << ") section\n"; - uint32_t section_type; + + uint32_t section_flags; if (O->is64Bit()) { const MachO::section_64 Sec = O->getSection64(Ref); - section_type = Sec.flags & MachO::SECTION_TYPE; + section_flags = Sec.flags; } else { const MachO::section Sec = O->getSection(Ref); - section_type = Sec.flags & MachO::SECTION_TYPE; + section_flags = Sec.flags; } + uint32_t section_type = section_flags & MachO::SECTION_TYPE; StringRef BytesStr; Section.getContents(BytesStr); @@ -658,7 +1033,23 @@ static void DumpSectionContents(MachOObjectFile *O, bool verbose) { uint32_t sect_size = BytesStr.size(); uint64_t sect_addr = Section.getAddress(); + outs() << "Contents of (" << SegName << "," << SectName + << ") section\n"; + if (verbose) { + if ((section_flags & MachO::S_ATTR_PURE_INSTRUCTIONS) || + (section_flags & MachO::S_ATTR_SOME_INSTRUCTIONS)) { + DisassembleMachO(Filename, O, SegName, SectName); + continue; + } + if (SegName == "__TEXT" && SectName == "__info_plist") { + outs() << sect; + continue; + } + if (SegName == "__OBJC" && SectName == "__protocol") { + DumpProtocolSection(O, sect, sect_size, sect_addr); + continue; + } switch (section_type) { case MachO::S_REGULAR: DumpRawSectionContents(O, sect, sect_size, sect_addr); @@ -666,6 +1057,22 @@ static void DumpSectionContents(MachOObjectFile *O, bool verbose) { case MachO::S_ZEROFILL: outs() << "zerofill section and has no contents in the file\n"; break; + case MachO::S_CSTRING_LITERALS: + DumpCstringSection(O, sect, sect_size, sect_addr, !NoLeadingAddr); + break; + case MachO::S_4BYTE_LITERALS: + DumpLiteral4Section(O, sect, sect_size, sect_addr, !NoLeadingAddr); + break; + case MachO::S_8BYTE_LITERALS: + DumpLiteral8Section(O, sect, sect_size, sect_addr, !NoLeadingAddr); + break; + case MachO::S_16BYTE_LITERALS: + DumpLiteral16Section(O, sect, sect_size, sect_addr, !NoLeadingAddr); + break; + case MachO::S_LITERAL_POINTERS: + DumpLiteralPointerSection(O, Section, sect, sect_size, sect_addr, + !NoLeadingAddr); + break; case MachO::S_MOD_INIT_FUNC_POINTERS: case MachO::S_MOD_TERM_FUNC_POINTERS: DumpInitTermPointerSection(O, sect, sect_size, sect_addr, &AddrMap, @@ -688,6 +1095,24 @@ static void DumpSectionContents(MachOObjectFile *O, bool verbose) { } } +static void DumpInfoPlistSectionContents(StringRef Filename, + MachOObjectFile *O) { + for (const SectionRef &Section : O->sections()) { + StringRef SectName; + Section.getName(SectName); + DataRefImpl Ref = Section.getRawDataRefImpl(); + StringRef SegName = O->getSectionFinalSegmentName(Ref); + if (SegName == "__TEXT" && SectName == "__info_plist") { + outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; + StringRef BytesStr; + Section.getContents(BytesStr); + const char *sect = reinterpret_cast(BytesStr.data()); + outs() << sect; + return; + } + } +} + // checkMachOAndArchFlags() checks to see if the ObjectFile is a Mach-O file // and if it is and there is a list of architecture flags is specified then // check to make sure this Mach-O file is one of those architectures or all @@ -722,7 +1147,7 @@ static bool checkMachOAndArchFlags(ObjectFile *O, StringRef Filename) { return true; } -static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF); +static void printObjcMetaData(MachOObjectFile *O, bool verbose); // ProcessMachO() is passed a single opened Mach-O file, which may be an // archive member and or in a slice of a universal file. It prints the @@ -736,7 +1161,7 @@ static void ProcessMachO(StringRef Filename, MachOObjectFile *MachOOF, // UniversalHeaders or ArchiveHeaders. if (Disassemble || PrivateHeaders || ExportsTrie || Rebase || Bind || LazyBind || WeakBind || IndirectSymbols || DataInCode || LinkOptHints || - DumpSections.size() != 0) { + DylibsUsed || DylibId || ObjcMetaData || (FilterSections.size() != 0)) { outs() << Filename; if (!ArchiveMemberName.empty()) outs() << '(' << ArchiveMemberName << ')'; @@ -746,11 +1171,11 @@ static void ProcessMachO(StringRef Filename, MachOObjectFile *MachOOF, } if (Disassemble) - DisassembleMachO(Filename, MachOOF); + DisassembleMachO(Filename, MachOOF, "__TEXT", "__text"); if (IndirectSymbols) - PrintIndirectSymbols(MachOOF, true); + PrintIndirectSymbols(MachOOF, !NonVerbose); if (DataInCode) - PrintDataInCodeTable(MachOOF, true); + PrintDataInCodeTable(MachOOF, !NonVerbose); if (LinkOptHints) PrintLinkOptHints(MachOOF); if (Relocations) @@ -759,14 +1184,22 @@ static void ProcessMachO(StringRef Filename, MachOObjectFile *MachOOF, PrintSectionHeaders(MachOOF); if (SectionContents) PrintSectionContents(MachOOF); - if (DumpSections.size() != 0) - DumpSectionContents(MachOOF, true); + if (FilterSections.size() != 0) + DumpSectionContents(Filename, MachOOF, !NonVerbose); + if (InfoPlist) + DumpInfoPlistSectionContents(Filename, MachOOF); + if (DylibsUsed) + PrintDylibs(MachOOF, false); + if (DylibId) + PrintDylibs(MachOOF, true); if (SymbolTable) PrintSymbolTable(MachOOF); if (UnwindInfo) printMachOUnwindInfo(MachOOF); if (PrivateHeaders) printMachOFileHeader(MachOOF); + if (ObjcMetaData) + printObjcMetaData(MachOOF, !NonVerbose); if (ExportsTrie) printExportsTrie(MachOOF); if (Rebase) @@ -952,7 +1385,7 @@ static void printMachOUniversalHeaders(const object::MachOUniversalBinary *UB, } } -static void printArchiveChild(Archive::Child &C, bool verbose, +static void printArchiveChild(const Archive::Child &C, bool verbose, bool print_offset) { if (print_offset) outs() << C.getChildOffset() << "\t"; @@ -961,42 +1394,15 @@ static void printArchiveChild(Archive::Child &C, bool verbose, // FIXME: this first dash, "-", is for (Mode & S_IFMT) == S_IFREG. // But there is nothing in sys::fs::perms for S_IFMT or S_IFREG. outs() << "-"; - if (Mode & sys::fs::owner_read) - outs() << "r"; - else - outs() << "-"; - if (Mode & sys::fs::owner_write) - outs() << "w"; - else - outs() << "-"; - if (Mode & sys::fs::owner_exe) - outs() << "x"; - else - outs() << "-"; - if (Mode & sys::fs::group_read) - outs() << "r"; - else - outs() << "-"; - if (Mode & sys::fs::group_write) - outs() << "w"; - else - outs() << "-"; - if (Mode & sys::fs::group_exe) - outs() << "x"; - else - outs() << "-"; - if (Mode & sys::fs::others_read) - outs() << "r"; - else - outs() << "-"; - if (Mode & sys::fs::others_write) - outs() << "w"; - else - outs() << "-"; - if (Mode & sys::fs::others_exe) - outs() << "x"; - else - outs() << "-"; + outs() << ((Mode & sys::fs::owner_read) ? "r" : "-"); + outs() << ((Mode & sys::fs::owner_write) ? "w" : "-"); + outs() << ((Mode & sys::fs::owner_exe) ? "x" : "-"); + outs() << ((Mode & sys::fs::group_read) ? "r" : "-"); + outs() << ((Mode & sys::fs::group_write) ? "w" : "-"); + outs() << ((Mode & sys::fs::group_exe) ? "x" : "-"); + outs() << ((Mode & sys::fs::others_read) ? "r" : "-"); + outs() << ((Mode & sys::fs::others_write) ? "w" : "-"); + outs() << ((Mode & sys::fs::others_exe) ? "x" : "-"); } else { outs() << format("0%o ", Mode); } @@ -1005,8 +1411,10 @@ static void printArchiveChild(Archive::Child &C, bool verbose, outs() << format("%3d/", UID); unsigned GID = C.getGID(); outs() << format("%-3d ", GID); - uint64_t Size = C.getRawSize(); - outs() << format("%5" PRId64, Size) << " "; + ErrorOr Size = C.getRawSize(); + if (std::error_code EC = Size.getError()) + report_fatal_error(EC.message()); + outs() << format("%5" PRId64, Size.get()) << " "; StringRef RawLastModified = C.getRawLastModified(); if (verbose) { @@ -1040,14 +1448,11 @@ static void printArchiveChild(Archive::Child &C, bool verbose, } static void printArchiveHeaders(Archive *A, bool verbose, bool print_offset) { - if (A->hasSymbolTable()) { - Archive::child_iterator S = A->getSymbolTableChild(); - Archive::Child C = *S; - printArchiveChild(C, verbose, print_offset); - } - for (Archive::child_iterator I = A->child_begin(), E = A->child_end(); I != E; - ++I) { - Archive::Child C = *I; + for (Archive::child_iterator I = A->child_begin(false), E = A->child_end(); + I != E; ++I) { + if (std::error_code EC = I->getError()) + report_fatal_error(EC.message()); + const Archive::Child &C = **I; printArchiveChild(C, verbose, print_offset); } } @@ -1081,10 +1486,13 @@ void llvm::ParseInputMachO(StringRef Filename) { if (Archive *A = dyn_cast(&Bin)) { outs() << "Archive : " << Filename << "\n"; if (ArchiveHeaders) - printArchiveHeaders(A, true, false); + printArchiveHeaders(A, !NonVerbose, ArchiveMemberOffsets); for (Archive::child_iterator I = A->child_begin(), E = A->child_end(); I != E; ++I) { - ErrorOr> ChildOrErr = I->getAsBinary(); + if (std::error_code EC = I->getError()) + report_error(Filename, EC); + auto &C = I->get(); + ErrorOr> ChildOrErr = C.getAsBinary(); if (ChildOrErr.getError()) continue; if (MachOObjectFile *O = dyn_cast(&*ChildOrErr.get())) { @@ -1097,7 +1505,7 @@ void llvm::ParseInputMachO(StringRef Filename) { } if (UniversalHeaders) { if (MachOUniversalBinary *UB = dyn_cast(&Bin)) - printMachOUniversalHeaders(UB, true); + printMachOUniversalHeaders(UB, !NonVerbose); } if (MachOUniversalBinary *UB = dyn_cast(&Bin)) { // If we have a list of architecture flags specified dump only those. @@ -1128,11 +1536,14 @@ void llvm::ParseInputMachO(StringRef Filename) { outs() << " (architecture " << ArchitectureName << ")"; outs() << "\n"; if (ArchiveHeaders) - printArchiveHeaders(A.get(), true, false); + printArchiveHeaders(A.get(), !NonVerbose, ArchiveMemberOffsets); for (Archive::child_iterator AI = A->child_begin(), AE = A->child_end(); AI != AE; ++AI) { - ErrorOr> ChildOrErr = AI->getAsBinary(); + if (std::error_code EC = AI->getError()) + report_error(Filename, EC); + auto &C = AI->get(); + ErrorOr> ChildOrErr = C.getAsBinary(); if (ChildOrErr.getError()) continue; if (MachOObjectFile *O = @@ -1170,11 +1581,14 @@ void llvm::ParseInputMachO(StringRef Filename) { std::unique_ptr &A = *AOrErr; outs() << "Archive : " << Filename << "\n"; if (ArchiveHeaders) - printArchiveHeaders(A.get(), true, false); + printArchiveHeaders(A.get(), !NonVerbose, ArchiveMemberOffsets); for (Archive::child_iterator AI = A->child_begin(), AE = A->child_end(); AI != AE; ++AI) { - ErrorOr> ChildOrErr = AI->getAsBinary(); + if (std::error_code EC = AI->getError()) + report_error(Filename, EC); + auto &C = AI->get(); + ErrorOr> ChildOrErr = C.getAsBinary(); if (ChildOrErr.getError()) continue; if (MachOObjectFile *O = @@ -1207,10 +1621,13 @@ void llvm::ParseInputMachO(StringRef Filename) { outs() << " (architecture " << ArchitectureName << ")"; outs() << "\n"; if (ArchiveHeaders) - printArchiveHeaders(A.get(), true, false); + printArchiveHeaders(A.get(), !NonVerbose, ArchiveMemberOffsets); for (Archive::child_iterator AI = A->child_begin(), AE = A->child_end(); AI != AE; ++AI) { - ErrorOr> ChildOrErr = AI->getAsBinary(); + if (std::error_code EC = AI->getError()) + report_error(Filename, EC); + auto &C = AI->get(); + ErrorOr> ChildOrErr = C.getAsBinary(); if (ChildOrErr.getError()) continue; if (MachOObjectFile *O = @@ -1233,8 +1650,7 @@ void llvm::ParseInputMachO(StringRef Filename) { errs() << "llvm-objdump: '" << Filename << "': " << "Object is not a Mach-O file type.\n"; } else - errs() << "llvm-objdump: '" << Filename << "': " - << "Unrecognized file type.\n"; + report_error(Filename, object_error::invalid_file_type); } typedef std::pair BindInfoEntry; @@ -1255,6 +1671,7 @@ struct DisassembleInfo { uint64_t adrp_addr; uint32_t adrp_inst; BindTable *bindtable; + uint32_t depth; }; // SymbolizerGetOpInfo() is the operand information call back function. @@ -1272,8 +1689,8 @@ struct DisassembleInfo { // names and addends of the symbolic expression to add for the operand. The // value of TagType is currently 1 (for the LLVMOpInfo1 struct). If symbolic // information is returned then this function returns 1 else it returns 0. -int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, - uint64_t Size, int TagType, void *TagBuf) { +static int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, + uint64_t Size, int TagType, void *TagBuf) { struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo; struct LLVMOpInfo1 *op_info = (struct LLVMOpInfo1 *)TagBuf; uint64_t value = op_info->Value; @@ -1285,15 +1702,22 @@ int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, // If the TagType is not the value 1 which it code knows about or if no // verbose symbolic information is wanted then just return 0, indicating no // information is being returned. - if (TagType != 1 || info->verbose == false) + if (TagType != 1 || !info->verbose) return 0; unsigned int Arch = info->O->getArch(); if (Arch == Triple::x86) { if (Size != 1 && Size != 2 && Size != 4 && Size != 0) return 0; - // First search the section's relocation entries (if any) for an entry - // for this section offset. + if (info->O->getHeader().filetype != MachO::MH_OBJECT) { + // TODO: + // Search the external relocation entries of a fully linked image + // (if any) for an entry that matches this segment offset. + // uint32_t seg_offset = (Pc + Offset); + return 0; + } + // In MH_OBJECT filetypes search the section's relocation entries (if any) + // for an entry for this section offset. uint32_t sect_addr = info->S.getAddress(); uint32_t sect_offset = (Pc + Offset) - sect_addr; bool reloc_found = false; @@ -1304,8 +1728,7 @@ int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, bool r_scattered = false; uint32_t r_value, pair_r_value, r_type; for (const RelocationRef &Reloc : info->S.relocations()) { - uint64_t RelocOffset; - Reloc.getOffset(RelocOffset); + uint64_t RelocOffset = Reloc.getOffset(); if (RelocOffset == sect_offset) { Rel = Reloc.getRawDataRefImpl(); RE = info->O->getRelocation(Rel); @@ -1336,9 +1759,10 @@ int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, } } if (reloc_found && isExtern) { - StringRef SymName; - Symbol.getName(SymName); - const char *name = SymName.data(); + ErrorOr SymName = Symbol.getName(); + if (std::error_code EC = SymName.getError()) + report_fatal_error(EC.message()); + const char *name = SymName->data(); op_info->AddSymbol.Present = 1; op_info->AddSymbol.Name = name; // For i386 extern relocation entries the value in the instruction is @@ -1363,16 +1787,20 @@ int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, op_info->Value = offset; return 1; } - // TODO: - // Second search the external relocation entries of a fully linked image - // (if any) for an entry that matches this segment offset. - // uint32_t seg_offset = (Pc + Offset); return 0; - } else if (Arch == Triple::x86_64) { + } + if (Arch == Triple::x86_64) { if (Size != 1 && Size != 2 && Size != 4 && Size != 0) return 0; - // First search the section's relocation entries (if any) for an entry - // for this section offset. + if (info->O->getHeader().filetype != MachO::MH_OBJECT) { + // TODO: + // Search the external relocation entries of a fully linked image + // (if any) for an entry that matches this segment offset. + // uint64_t seg_offset = (Pc + Offset); + return 0; + } + // In MH_OBJECT filetypes search the section's relocation entries (if any) + // for an entry for this section offset. uint64_t sect_addr = info->S.getAddress(); uint64_t sect_offset = (Pc + Offset) - sect_addr; bool reloc_found = false; @@ -1381,8 +1809,7 @@ int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, bool isExtern = false; SymbolRef Symbol; for (const RelocationRef &Reloc : info->S.relocations()) { - uint64_t RelocOffset; - Reloc.getOffset(RelocOffset); + uint64_t RelocOffset = Reloc.getOffset(); if (RelocOffset == sect_offset) { Rel = Reloc.getRawDataRefImpl(); RE = info->O->getRelocation(Rel); @@ -1402,9 +1829,10 @@ int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, // is the offset from the external symbol. if (info->O->getAnyRelocationPCRel(RE)) op_info->Value -= Pc + Offset + Size; - StringRef SymName; - Symbol.getName(SymName); - const char *name = SymName.data(); + ErrorOr SymName = Symbol.getName(); + if (std::error_code EC = SymName.getError()) + report_fatal_error(EC.message()); + const char *name = SymName->data(); unsigned Type = info->O->getAnyRelocationType(RE); if (Type == MachO::X86_64_RELOC_SUBTRACTOR) { DataRefImpl RelNext = Rel; @@ -1418,9 +1846,10 @@ int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, op_info->SubtractSymbol.Name = name; symbol_iterator RelocSymNext = info->O->getSymbolByIndex(SymbolNum); Symbol = *RelocSymNext; - StringRef SymNameNext; - Symbol.getName(SymNameNext); - name = SymNameNext.data(); + ErrorOr SymNameNext = Symbol.getName(); + if (std::error_code EC = SymNameNext.getError()) + report_fatal_error(EC.message()); + name = SymNameNext->data(); } } // TODO: add the VariantKinds to op_info->VariantKind for relocation types @@ -1429,94 +1858,85 @@ int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, op_info->AddSymbol.Name = name; return 1; } - // TODO: - // Second search the external relocation entries of a fully linked image - // (if any) for an entry that matches this segment offset. - // uint64_t seg_offset = (Pc + Offset); return 0; - } else if (Arch == Triple::arm) { + } + if (Arch == Triple::arm) { if (Offset != 0 || (Size != 4 && Size != 2)) return 0; - // First search the section's relocation entries (if any) for an entry - // for this section offset. + if (info->O->getHeader().filetype != MachO::MH_OBJECT) { + // TODO: + // Search the external relocation entries of a fully linked image + // (if any) for an entry that matches this segment offset. + // uint32_t seg_offset = (Pc + Offset); + return 0; + } + // In MH_OBJECT filetypes search the section's relocation entries (if any) + // for an entry for this section offset. uint32_t sect_addr = info->S.getAddress(); uint32_t sect_offset = (Pc + Offset) - sect_addr; - bool reloc_found = false; DataRefImpl Rel; MachO::any_relocation_info RE; bool isExtern = false; SymbolRef Symbol; bool r_scattered = false; uint32_t r_value, pair_r_value, r_type, r_length, other_half; - for (const RelocationRef &Reloc : info->S.relocations()) { - uint64_t RelocOffset; - Reloc.getOffset(RelocOffset); - if (RelocOffset == sect_offset) { - Rel = Reloc.getRawDataRefImpl(); - RE = info->O->getRelocation(Rel); - r_length = info->O->getAnyRelocationLength(RE); - r_scattered = info->O->isRelocationScattered(RE); - if (r_scattered) { - r_value = info->O->getScatteredRelocationValue(RE); - r_type = info->O->getScatteredRelocationType(RE); - } else { - r_type = info->O->getAnyRelocationType(RE); - isExtern = info->O->getPlainRelocationExternal(RE); - if (isExtern) { - symbol_iterator RelocSym = Reloc.getSymbol(); - Symbol = *RelocSym; - } - } - if (r_type == MachO::ARM_RELOC_HALF || - r_type == MachO::ARM_RELOC_SECTDIFF || - r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF || - r_type == MachO::ARM_RELOC_HALF_SECTDIFF) { - DataRefImpl RelNext = Rel; - info->O->moveRelocationNext(RelNext); - MachO::any_relocation_info RENext; - RENext = info->O->getRelocation(RelNext); - other_half = info->O->getAnyRelocationAddress(RENext) & 0xffff; - if (info->O->isRelocationScattered(RENext)) - pair_r_value = info->O->getScatteredRelocationValue(RENext); - } - reloc_found = true; - break; + auto Reloc = + std::find_if(info->S.relocations().begin(), info->S.relocations().end(), + [&](const RelocationRef &Reloc) { + uint64_t RelocOffset = Reloc.getOffset(); + return RelocOffset == sect_offset; + }); + + if (Reloc == info->S.relocations().end()) + return 0; + + Rel = Reloc->getRawDataRefImpl(); + RE = info->O->getRelocation(Rel); + r_length = info->O->getAnyRelocationLength(RE); + r_scattered = info->O->isRelocationScattered(RE); + if (r_scattered) { + r_value = info->O->getScatteredRelocationValue(RE); + r_type = info->O->getScatteredRelocationType(RE); + } else { + r_type = info->O->getAnyRelocationType(RE); + isExtern = info->O->getPlainRelocationExternal(RE); + if (isExtern) { + symbol_iterator RelocSym = Reloc->getSymbol(); + Symbol = *RelocSym; } } - if (reloc_found && isExtern) { - StringRef SymName; - Symbol.getName(SymName); - const char *name = SymName.data(); + if (r_type == MachO::ARM_RELOC_HALF || + r_type == MachO::ARM_RELOC_SECTDIFF || + r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF || + r_type == MachO::ARM_RELOC_HALF_SECTDIFF) { + DataRefImpl RelNext = Rel; + info->O->moveRelocationNext(RelNext); + MachO::any_relocation_info RENext; + RENext = info->O->getRelocation(RelNext); + other_half = info->O->getAnyRelocationAddress(RENext) & 0xffff; + if (info->O->isRelocationScattered(RENext)) + pair_r_value = info->O->getScatteredRelocationValue(RENext); + } + + if (isExtern) { + ErrorOr SymName = Symbol.getName(); + if (std::error_code EC = SymName.getError()) + report_fatal_error(EC.message()); + const char *name = SymName->data(); op_info->AddSymbol.Present = 1; op_info->AddSymbol.Name = name; - if (value != 0) { - switch (r_type) { - case MachO::ARM_RELOC_HALF: - if ((r_length & 0x1) == 1) { - op_info->Value = value << 16 | other_half; - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16; - } else { - op_info->Value = other_half << 16 | value; - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16; - } - break; - default: - break; - } - } else { - switch (r_type) { - case MachO::ARM_RELOC_HALF: - if ((r_length & 0x1) == 1) { - op_info->Value = value << 16 | other_half; - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16; - } else { - op_info->Value = other_half << 16 | value; - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16; - } - break; - default: - break; + switch (r_type) { + case MachO::ARM_RELOC_HALF: + if ((r_length & 0x1) == 1) { + op_info->Value = value << 16 | other_half; + op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16; + } else { + op_info->Value = other_half << 16 | value; + op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16; } + break; + default: + break; } return 1; } @@ -1524,27 +1944,25 @@ int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, // return 0 so the code in tryAddingSymbolicOperand() can use the // SymbolLookUp call back with the branch target address to look up the // symbol and possiblity add an annotation for a symbol stub. - if (reloc_found && isExtern == 0 && (r_type == MachO::ARM_RELOC_BR24 || - r_type == MachO::ARM_THUMB_RELOC_BR22)) + if (isExtern == 0 && (r_type == MachO::ARM_RELOC_BR24 || + r_type == MachO::ARM_THUMB_RELOC_BR22)) return 0; uint32_t offset = 0; - if (reloc_found) { - if (r_type == MachO::ARM_RELOC_HALF || - r_type == MachO::ARM_RELOC_HALF_SECTDIFF) { - if ((r_length & 0x1) == 1) - value = value << 16 | other_half; - else - value = other_half << 16 | value; - } - if (r_scattered && (r_type != MachO::ARM_RELOC_HALF && - r_type != MachO::ARM_RELOC_HALF_SECTDIFF)) { - offset = value - r_value; - value = r_value; - } + if (r_type == MachO::ARM_RELOC_HALF || + r_type == MachO::ARM_RELOC_HALF_SECTDIFF) { + if ((r_length & 0x1) == 1) + value = value << 16 | other_half; + else + value = other_half << 16 | value; + } + if (r_scattered && (r_type != MachO::ARM_RELOC_HALF && + r_type != MachO::ARM_RELOC_HALF_SECTDIFF)) { + offset = value - r_value; + value = r_value; } - if (reloc_found && r_type == MachO::ARM_RELOC_HALF_SECTDIFF) { + if (r_type == MachO::ARM_RELOC_HALF_SECTDIFF) { if ((r_length & 0x1) == 1) op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16; else @@ -1566,18 +1984,13 @@ int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, return 1; } - if (reloc_found == false) - return 0; - op_info->AddSymbol.Present = 1; op_info->Value = offset; - if (reloc_found) { - if (r_type == MachO::ARM_RELOC_HALF) { - if ((r_length & 0x1) == 1) - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16; - else - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16; - } + if (r_type == MachO::ARM_RELOC_HALF) { + if ((r_length & 0x1) == 1) + op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16; + else + op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16; } const char *add = GuessSymbolName(value, info->AddrMap); if (add != nullptr) { @@ -1586,98 +1999,94 @@ int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, } op_info->AddSymbol.Value = value; return 1; - } else if (Arch == Triple::aarch64) { + } + if (Arch == Triple::aarch64) { if (Offset != 0 || Size != 4) return 0; - // First search the section's relocation entries (if any) for an entry - // for this section offset. + if (info->O->getHeader().filetype != MachO::MH_OBJECT) { + // TODO: + // Search the external relocation entries of a fully linked image + // (if any) for an entry that matches this segment offset. + // uint64_t seg_offset = (Pc + Offset); + return 0; + } + // In MH_OBJECT filetypes search the section's relocation entries (if any) + // for an entry for this section offset. uint64_t sect_addr = info->S.getAddress(); uint64_t sect_offset = (Pc + Offset) - sect_addr; - bool reloc_found = false; - DataRefImpl Rel; - MachO::any_relocation_info RE; - bool isExtern = false; - SymbolRef Symbol; - uint32_t r_type = 0; - for (const RelocationRef &Reloc : info->S.relocations()) { - uint64_t RelocOffset; - Reloc.getOffset(RelocOffset); - if (RelocOffset == sect_offset) { - Rel = Reloc.getRawDataRefImpl(); - RE = info->O->getRelocation(Rel); - r_type = info->O->getAnyRelocationType(RE); - if (r_type == MachO::ARM64_RELOC_ADDEND) { - DataRefImpl RelNext = Rel; - info->O->moveRelocationNext(RelNext); - MachO::any_relocation_info RENext = info->O->getRelocation(RelNext); - if (value == 0) { - value = info->O->getPlainRelocationSymbolNum(RENext); - op_info->Value = value; - } - } - // NOTE: Scattered relocations don't exist on arm64. - isExtern = info->O->getPlainRelocationExternal(RE); - if (isExtern) { - symbol_iterator RelocSym = Reloc.getSymbol(); - Symbol = *RelocSym; - } - reloc_found = true; - break; + auto Reloc = + std::find_if(info->S.relocations().begin(), info->S.relocations().end(), + [&](const RelocationRef &Reloc) { + uint64_t RelocOffset = Reloc.getOffset(); + return RelocOffset == sect_offset; + }); + + if (Reloc == info->S.relocations().end()) + return 0; + + DataRefImpl Rel = Reloc->getRawDataRefImpl(); + MachO::any_relocation_info RE = info->O->getRelocation(Rel); + uint32_t r_type = info->O->getAnyRelocationType(RE); + if (r_type == MachO::ARM64_RELOC_ADDEND) { + DataRefImpl RelNext = Rel; + info->O->moveRelocationNext(RelNext); + MachO::any_relocation_info RENext = info->O->getRelocation(RelNext); + if (value == 0) { + value = info->O->getPlainRelocationSymbolNum(RENext); + op_info->Value = value; } } - if (reloc_found && isExtern) { - StringRef SymName; - Symbol.getName(SymName); - const char *name = SymName.data(); - op_info->AddSymbol.Present = 1; - op_info->AddSymbol.Name = name; + // NOTE: Scattered relocations don't exist on arm64. + if (!info->O->getPlainRelocationExternal(RE)) + return 0; + ErrorOr SymName = Reloc->getSymbol()->getName(); + if (std::error_code EC = SymName.getError()) + report_fatal_error(EC.message()); + const char *name = SymName->data(); + op_info->AddSymbol.Present = 1; + op_info->AddSymbol.Name = name; - switch (r_type) { - case MachO::ARM64_RELOC_PAGE21: - /* @page */ - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGE; - break; - case MachO::ARM64_RELOC_PAGEOFF12: - /* @pageoff */ - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGEOFF; - break; - case MachO::ARM64_RELOC_GOT_LOAD_PAGE21: - /* @gotpage */ - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGE; - break; - case MachO::ARM64_RELOC_GOT_LOAD_PAGEOFF12: - /* @gotpageoff */ - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGEOFF; - break; - case MachO::ARM64_RELOC_TLVP_LOAD_PAGE21: - /* @tvlppage is not implemented in llvm-mc */ - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVP; - break; - case MachO::ARM64_RELOC_TLVP_LOAD_PAGEOFF12: - /* @tvlppageoff is not implemented in llvm-mc */ - op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVOFF; - break; - default: - case MachO::ARM64_RELOC_BRANCH26: - op_info->VariantKind = LLVMDisassembler_VariantKind_None; - break; - } - return 1; + switch (r_type) { + case MachO::ARM64_RELOC_PAGE21: + /* @page */ + op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGE; + break; + case MachO::ARM64_RELOC_PAGEOFF12: + /* @pageoff */ + op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGEOFF; + break; + case MachO::ARM64_RELOC_GOT_LOAD_PAGE21: + /* @gotpage */ + op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGE; + break; + case MachO::ARM64_RELOC_GOT_LOAD_PAGEOFF12: + /* @gotpageoff */ + op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGEOFF; + break; + case MachO::ARM64_RELOC_TLVP_LOAD_PAGE21: + /* @tvlppage is not implemented in llvm-mc */ + op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVP; + break; + case MachO::ARM64_RELOC_TLVP_LOAD_PAGEOFF12: + /* @tvlppageoff is not implemented in llvm-mc */ + op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVOFF; + break; + default: + case MachO::ARM64_RELOC_BRANCH26: + op_info->VariantKind = LLVMDisassembler_VariantKind_None; + break; } - return 0; - } else { - return 0; + return 1; } + return 0; } // GuessCstringPointer is passed the address of what might be a pointer to a // literal string in a cstring section. If that address is in a cstring section // it returns a pointer to that string. Else it returns nullptr. -const char *GuessCstringPointer(uint64_t ReferenceValue, - struct DisassembleInfo *info) { - uint32_t LoadCommandCount = info->O->getHeader().ncmds; - MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo(); - for (unsigned I = 0;; ++I) { +static const char *GuessCstringPointer(uint64_t ReferenceValue, + struct DisassembleInfo *info) { + for (const auto &Load : info->O->load_commands()) { if (Load.C.cmd == MachO::LC_SEGMENT_64) { MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load); for (unsigned J = 0; J < Seg.nsects; ++J) { @@ -1721,10 +2130,6 @@ const char *GuessCstringPointer(uint64_t ReferenceValue, } } } - if (I == LoadCommandCount - 1) - break; - else - Load = info->O->getNextLoadCommandInfo(Load); } return nullptr; } @@ -1735,11 +2140,9 @@ const char *GuessCstringPointer(uint64_t ReferenceValue, // symbol name being referenced by the stub or pointer. static const char *GuessIndirectSymbol(uint64_t ReferenceValue, struct DisassembleInfo *info) { - uint32_t LoadCommandCount = info->O->getHeader().ncmds; - MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo(); MachO::dysymtab_command Dysymtab = info->O->getDysymtabLoadCommand(); MachO::symtab_command Symtab = info->O->getSymtabLoadCommand(); - for (unsigned I = 0;; ++I) { + for (const auto &Load : info->O->load_commands()) { if (Load.C.cmd == MachO::LC_SEGMENT_64) { MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load); for (unsigned J = 0; J < Seg.nsects; ++J) { @@ -1766,9 +2169,10 @@ static const char *GuessIndirectSymbol(uint64_t ReferenceValue, if (indirect_symbol < Symtab.nsyms) { symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol); SymbolRef Symbol = *Sym; - StringRef SymName; - Symbol.getName(SymName); - const char *name = SymName.data(); + ErrorOr SymName = Symbol.getName(); + if (std::error_code EC = SymName.getError()) + report_fatal_error(EC.message()); + const char *name = SymName->data(); return name; } } @@ -1800,19 +2204,16 @@ static const char *GuessIndirectSymbol(uint64_t ReferenceValue, if (indirect_symbol < Symtab.nsyms) { symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol); SymbolRef Symbol = *Sym; - StringRef SymName; - Symbol.getName(SymName); - const char *name = SymName.data(); + ErrorOr SymName = Symbol.getName(); + if (std::error_code EC = SymName.getError()) + report_fatal_error(EC.message()); + const char *name = SymName->data(); return name; } } } } } - if (I == LoadCommandCount - 1) - break; - else - Load = info->O->getNextLoadCommandInfo(Load); } return nullptr; } @@ -1899,9 +2300,7 @@ static uint64_t GuessPointerPointer(uint64_t ReferenceValue, selref = false; msgref = false; cfstring = false; - uint32_t LoadCommandCount = info->O->getHeader().ncmds; - MachOObjectFile::LoadCommandInfo Load = info->O->getFirstLoadCommandInfo(); - for (unsigned I = 0;; ++I) { + for (const auto &Load : info->O->load_commands()) { if (Load.C.cmd == MachO::LC_SEGMENT_64) { MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load); for (unsigned J = 0; J < Seg.nsects; ++J) { @@ -1946,10 +2345,6 @@ static uint64_t GuessPointerPointer(uint64_t ReferenceValue, } } // TODO: Look for LC_SEGMENT for 32-bit Mach-O files. - if (I == LoadCommandCount - 1) - break; - else - Load = info->O->getNextLoadCommandInfo(Load); } return 0; } @@ -1959,14 +2354,26 @@ static uint64_t GuessPointerPointer(uint64_t ReferenceValue, // offset into the section, number of bytes left in the section past the offset // and which section is was being referenced. If the Address is not in a // section nullptr is returned. -const char *get_pointer_64(uint64_t Address, uint32_t &offset, uint32_t &left, - SectionRef &S, DisassembleInfo *info) { +static const char *get_pointer_64(uint64_t Address, uint32_t &offset, + uint32_t &left, SectionRef &S, + DisassembleInfo *info, + bool objc_only = false) { offset = 0; left = 0; S = SectionRef(); for (unsigned SectIdx = 0; SectIdx != info->Sections->size(); SectIdx++) { uint64_t SectAddress = ((*(info->Sections))[SectIdx]).getAddress(); uint64_t SectSize = ((*(info->Sections))[SectIdx]).getSize(); + if (SectSize == 0) + continue; + if (objc_only) { + StringRef SectName; + ((*(info->Sections))[SectIdx]).getName(SectName); + DataRefImpl Ref = ((*(info->Sections))[SectIdx]).getRawDataRefImpl(); + StringRef SegName = info->O->getSectionFinalSegmentName(Ref); + if (SegName != "__OBJC" && SectName != "__cstring") + continue; + } if (Address >= SectAddress && Address < SectAddress + SectSize) { S = (*(info->Sections))[SectIdx]; offset = Address - SectAddress; @@ -1979,13 +2386,23 @@ const char *get_pointer_64(uint64_t Address, uint32_t &offset, uint32_t &left, return nullptr; } +static const char *get_pointer_32(uint32_t Address, uint32_t &offset, + uint32_t &left, SectionRef &S, + DisassembleInfo *info, + bool objc_only = false) { + return get_pointer_64(Address, offset, left, S, info, objc_only); +} + // get_symbol_64() returns the name of a symbol (or nullptr) and the address of // the symbol indirectly through n_value. Based on the relocation information // for the specified section offset in the specified section reference. -const char *get_symbol_64(uint32_t sect_offset, SectionRef S, - DisassembleInfo *info, uint64_t &n_value) { +// If no relocation information is found and a non-zero ReferenceValue for the +// symbol is passed, look up that address in the info's AddrMap. +static const char *get_symbol_64(uint32_t sect_offset, SectionRef S, + DisassembleInfo *info, uint64_t &n_value, + uint64_t ReferenceValue = 0) { n_value = 0; - if (info->verbose == false) + if (!info->verbose) return nullptr; // See if there is an external relocation entry at the sect_offset. @@ -1995,8 +2412,7 @@ const char *get_symbol_64(uint32_t sect_offset, SectionRef S, bool isExtern = false; SymbolRef Symbol; for (const RelocationRef &Reloc : S.relocations()) { - uint64_t RelocOffset; - Reloc.getOffset(RelocOffset); + uint64_t RelocOffset = Reloc.getOffset(); if (RelocOffset == sect_offset) { Rel = Reloc.getRawDataRefImpl(); RE = info->O->getRelocation(Rel); @@ -2016,11 +2432,13 @@ const char *get_symbol_64(uint32_t sect_offset, SectionRef S, // and return its name. const char *SymbolName = nullptr; if (reloc_found && isExtern) { - Symbol.getAddress(n_value); - StringRef name; - Symbol.getName(name); - if (!name.empty()) { - SymbolName = name.data(); + n_value = Symbol.getValue(); + ErrorOr NameOrError = Symbol.getName(); + if (std::error_code EC = NameOrError.getError()) + report_fatal_error(EC.message()); + StringRef Name = *NameOrError; + if (!Name.empty()) { + SymbolName = Name.data(); return SymbolName; } } @@ -2034,17 +2452,20 @@ const char *get_symbol_64(uint32_t sect_offset, SectionRef S, // // NOTE: need add passing the database_offset to this routine. - // TODO: We did not find an external relocation entry so look up the - // ReferenceValue as an address of a symbol and if found return that symbol's - // name. - // - // NOTE: need add passing the ReferenceValue to this routine. Then that code - // would simply be this: - // SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap); + // We did not find an external relocation entry so look up the ReferenceValue + // as an address of a symbol and if found return that symbol's name. + SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap); return SymbolName; } +static const char *get_symbol_32(uint32_t sect_offset, SectionRef S, + DisassembleInfo *info, + uint32_t ReferenceValue) { + uint64_t n_value64; + return get_symbol_64(sect_offset, S, info, n_value64, ReferenceValue); +} + // These are structs in the Objective-C meta data and read to produce the // comments for disassembly. While these are part of the ABI they are no // public defintions. So the are here not in include/llvm/Support/MachO.h . @@ -2066,6 +2487,14 @@ struct class64_t { uint64_t data; // class_ro64_t * (64-bit pointer) }; +struct class32_t { + uint32_t isa; /* class32_t * (32-bit pointer) */ + uint32_t superclass; /* class32_t * (32-bit pointer) */ + uint32_t cache; /* Cache (32-bit pointer) */ + uint32_t vtable; /* IMP * (32-bit pointer) */ + uint32_t data; /* class_ro32_t * (32-bit pointer) */ +}; + struct class_ro64_t { uint32_t flags; uint32_t instanceStart; @@ -2080,6 +2509,276 @@ struct class_ro64_t { uint64_t baseProperties; // const struct objc_property_list (64-bit pointer) }; +struct class_ro32_t { + uint32_t flags; + uint32_t instanceStart; + uint32_t instanceSize; + uint32_t ivarLayout; /* const uint8_t * (32-bit pointer) */ + uint32_t name; /* const char * (32-bit pointer) */ + uint32_t baseMethods; /* const method_list_t * (32-bit pointer) */ + uint32_t baseProtocols; /* const protocol_list_t * (32-bit pointer) */ + uint32_t ivars; /* const ivar_list_t * (32-bit pointer) */ + uint32_t weakIvarLayout; /* const uint8_t * (32-bit pointer) */ + uint32_t baseProperties; /* const struct objc_property_list * + (32-bit pointer) */ +}; + +/* Values for class_ro{64,32}_t->flags */ +#define RO_META (1 << 0) +#define RO_ROOT (1 << 1) +#define RO_HAS_CXX_STRUCTORS (1 << 2) + +struct method_list64_t { + uint32_t entsize; + uint32_t count; + /* struct method64_t first; These structures follow inline */ +}; + +struct method_list32_t { + uint32_t entsize; + uint32_t count; + /* struct method32_t first; These structures follow inline */ +}; + +struct method64_t { + uint64_t name; /* SEL (64-bit pointer) */ + uint64_t types; /* const char * (64-bit pointer) */ + uint64_t imp; /* IMP (64-bit pointer) */ +}; + +struct method32_t { + uint32_t name; /* SEL (32-bit pointer) */ + uint32_t types; /* const char * (32-bit pointer) */ + uint32_t imp; /* IMP (32-bit pointer) */ +}; + +struct protocol_list64_t { + uint64_t count; /* uintptr_t (a 64-bit value) */ + /* struct protocol64_t * list[0]; These pointers follow inline */ +}; + +struct protocol_list32_t { + uint32_t count; /* uintptr_t (a 32-bit value) */ + /* struct protocol32_t * list[0]; These pointers follow inline */ +}; + +struct protocol64_t { + uint64_t isa; /* id * (64-bit pointer) */ + uint64_t name; /* const char * (64-bit pointer) */ + uint64_t protocols; /* struct protocol_list64_t * + (64-bit pointer) */ + uint64_t instanceMethods; /* method_list_t * (64-bit pointer) */ + uint64_t classMethods; /* method_list_t * (64-bit pointer) */ + uint64_t optionalInstanceMethods; /* method_list_t * (64-bit pointer) */ + uint64_t optionalClassMethods; /* method_list_t * (64-bit pointer) */ + uint64_t instanceProperties; /* struct objc_property_list * + (64-bit pointer) */ +}; + +struct protocol32_t { + uint32_t isa; /* id * (32-bit pointer) */ + uint32_t name; /* const char * (32-bit pointer) */ + uint32_t protocols; /* struct protocol_list_t * + (32-bit pointer) */ + uint32_t instanceMethods; /* method_list_t * (32-bit pointer) */ + uint32_t classMethods; /* method_list_t * (32-bit pointer) */ + uint32_t optionalInstanceMethods; /* method_list_t * (32-bit pointer) */ + uint32_t optionalClassMethods; /* method_list_t * (32-bit pointer) */ + uint32_t instanceProperties; /* struct objc_property_list * + (32-bit pointer) */ +}; + +struct ivar_list64_t { + uint32_t entsize; + uint32_t count; + /* struct ivar64_t first; These structures follow inline */ +}; + +struct ivar_list32_t { + uint32_t entsize; + uint32_t count; + /* struct ivar32_t first; These structures follow inline */ +}; + +struct ivar64_t { + uint64_t offset; /* uintptr_t * (64-bit pointer) */ + uint64_t name; /* const char * (64-bit pointer) */ + uint64_t type; /* const char * (64-bit pointer) */ + uint32_t alignment; + uint32_t size; +}; + +struct ivar32_t { + uint32_t offset; /* uintptr_t * (32-bit pointer) */ + uint32_t name; /* const char * (32-bit pointer) */ + uint32_t type; /* const char * (32-bit pointer) */ + uint32_t alignment; + uint32_t size; +}; + +struct objc_property_list64 { + uint32_t entsize; + uint32_t count; + /* struct objc_property64 first; These structures follow inline */ +}; + +struct objc_property_list32 { + uint32_t entsize; + uint32_t count; + /* struct objc_property32 first; These structures follow inline */ +}; + +struct objc_property64 { + uint64_t name; /* const char * (64-bit pointer) */ + uint64_t attributes; /* const char * (64-bit pointer) */ +}; + +struct objc_property32 { + uint32_t name; /* const char * (32-bit pointer) */ + uint32_t attributes; /* const char * (32-bit pointer) */ +}; + +struct category64_t { + uint64_t name; /* const char * (64-bit pointer) */ + uint64_t cls; /* struct class_t * (64-bit pointer) */ + uint64_t instanceMethods; /* struct method_list_t * (64-bit pointer) */ + uint64_t classMethods; /* struct method_list_t * (64-bit pointer) */ + uint64_t protocols; /* struct protocol_list_t * (64-bit pointer) */ + uint64_t instanceProperties; /* struct objc_property_list * + (64-bit pointer) */ +}; + +struct category32_t { + uint32_t name; /* const char * (32-bit pointer) */ + uint32_t cls; /* struct class_t * (32-bit pointer) */ + uint32_t instanceMethods; /* struct method_list_t * (32-bit pointer) */ + uint32_t classMethods; /* struct method_list_t * (32-bit pointer) */ + uint32_t protocols; /* struct protocol_list_t * (32-bit pointer) */ + uint32_t instanceProperties; /* struct objc_property_list * + (32-bit pointer) */ +}; + +struct objc_image_info64 { + uint32_t version; + uint32_t flags; +}; +struct objc_image_info32 { + uint32_t version; + uint32_t flags; +}; +struct imageInfo_t { + uint32_t version; + uint32_t flags; +}; +/* masks for objc_image_info.flags */ +#define OBJC_IMAGE_IS_REPLACEMENT (1 << 0) +#define OBJC_IMAGE_SUPPORTS_GC (1 << 1) + +struct message_ref64 { + uint64_t imp; /* IMP (64-bit pointer) */ + uint64_t sel; /* SEL (64-bit pointer) */ +}; + +struct message_ref32 { + uint32_t imp; /* IMP (32-bit pointer) */ + uint32_t sel; /* SEL (32-bit pointer) */ +}; + +// Objective-C 1 (32-bit only) meta data structs. + +struct objc_module_t { + uint32_t version; + uint32_t size; + uint32_t name; /* char * (32-bit pointer) */ + uint32_t symtab; /* struct objc_symtab * (32-bit pointer) */ +}; + +struct objc_symtab_t { + uint32_t sel_ref_cnt; + uint32_t refs; /* SEL * (32-bit pointer) */ + uint16_t cls_def_cnt; + uint16_t cat_def_cnt; + // uint32_t defs[1]; /* void * (32-bit pointer) variable size */ +}; + +struct objc_class_t { + uint32_t isa; /* struct objc_class * (32-bit pointer) */ + uint32_t super_class; /* struct objc_class * (32-bit pointer) */ + uint32_t name; /* const char * (32-bit pointer) */ + int32_t version; + int32_t info; + int32_t instance_size; + uint32_t ivars; /* struct objc_ivar_list * (32-bit pointer) */ + uint32_t methodLists; /* struct objc_method_list ** (32-bit pointer) */ + uint32_t cache; /* struct objc_cache * (32-bit pointer) */ + uint32_t protocols; /* struct objc_protocol_list * (32-bit pointer) */ +}; + +#define CLS_GETINFO(cls, infomask) ((cls)->info & (infomask)) +// class is not a metaclass +#define CLS_CLASS 0x1 +// class is a metaclass +#define CLS_META 0x2 + +struct objc_category_t { + uint32_t category_name; /* char * (32-bit pointer) */ + uint32_t class_name; /* char * (32-bit pointer) */ + uint32_t instance_methods; /* struct objc_method_list * (32-bit pointer) */ + uint32_t class_methods; /* struct objc_method_list * (32-bit pointer) */ + uint32_t protocols; /* struct objc_protocol_list * (32-bit ptr) */ +}; + +struct objc_ivar_t { + uint32_t ivar_name; /* char * (32-bit pointer) */ + uint32_t ivar_type; /* char * (32-bit pointer) */ + int32_t ivar_offset; +}; + +struct objc_ivar_list_t { + int32_t ivar_count; + // struct objc_ivar_t ivar_list[1]; /* variable length structure */ +}; + +struct objc_method_list_t { + uint32_t obsolete; /* struct objc_method_list * (32-bit pointer) */ + int32_t method_count; + // struct objc_method_t method_list[1]; /* variable length structure */ +}; + +struct objc_method_t { + uint32_t method_name; /* SEL, aka struct objc_selector * (32-bit pointer) */ + uint32_t method_types; /* char * (32-bit pointer) */ + uint32_t method_imp; /* IMP, aka function pointer, (*IMP)(id, SEL, ...) + (32-bit pointer) */ +}; + +struct objc_protocol_list_t { + uint32_t next; /* struct objc_protocol_list * (32-bit pointer) */ + int32_t count; + // uint32_t list[1]; /* Protocol *, aka struct objc_protocol_t * + // (32-bit pointer) */ +}; + +struct objc_protocol_t { + uint32_t isa; /* struct objc_class * (32-bit pointer) */ + uint32_t protocol_name; /* char * (32-bit pointer) */ + uint32_t protocol_list; /* struct objc_protocol_list * (32-bit pointer) */ + uint32_t instance_methods; /* struct objc_method_description_list * + (32-bit pointer) */ + uint32_t class_methods; /* struct objc_method_description_list * + (32-bit pointer) */ +}; + +struct objc_method_description_list_t { + int32_t count; + // struct objc_method_description_t list[1]; +}; + +struct objc_method_description_t { + uint32_t name; /* SEL, aka struct objc_selector * (32-bit pointer) */ + uint32_t types; /* char * (32-bit pointer) */ +}; + inline void swapStruct(struct cfstring64_t &cfs) { sys::swapByteOrder(cfs.isa); sys::swapByteOrder(cfs.flags); @@ -2095,6 +2794,14 @@ inline void swapStruct(struct class64_t &c) { sys::swapByteOrder(c.data); } +inline void swapStruct(struct class32_t &c) { + sys::swapByteOrder(c.isa); + sys::swapByteOrder(c.superclass); + sys::swapByteOrder(c.cache); + sys::swapByteOrder(c.vtable); + sys::swapByteOrder(c.data); +} + inline void swapStruct(struct class_ro64_t &cro) { sys::swapByteOrder(cro.flags); sys::swapByteOrder(cro.instanceStart); @@ -2109,6 +2816,238 @@ inline void swapStruct(struct class_ro64_t &cro) { sys::swapByteOrder(cro.baseProperties); } +inline void swapStruct(struct class_ro32_t &cro) { + sys::swapByteOrder(cro.flags); + sys::swapByteOrder(cro.instanceStart); + sys::swapByteOrder(cro.instanceSize); + sys::swapByteOrder(cro.ivarLayout); + sys::swapByteOrder(cro.name); + sys::swapByteOrder(cro.baseMethods); + sys::swapByteOrder(cro.baseProtocols); + sys::swapByteOrder(cro.ivars); + sys::swapByteOrder(cro.weakIvarLayout); + sys::swapByteOrder(cro.baseProperties); +} + +inline void swapStruct(struct method_list64_t &ml) { + sys::swapByteOrder(ml.entsize); + sys::swapByteOrder(ml.count); +} + +inline void swapStruct(struct method_list32_t &ml) { + sys::swapByteOrder(ml.entsize); + sys::swapByteOrder(ml.count); +} + +inline void swapStruct(struct method64_t &m) { + sys::swapByteOrder(m.name); + sys::swapByteOrder(m.types); + sys::swapByteOrder(m.imp); +} + +inline void swapStruct(struct method32_t &m) { + sys::swapByteOrder(m.name); + sys::swapByteOrder(m.types); + sys::swapByteOrder(m.imp); +} + +inline void swapStruct(struct protocol_list64_t &pl) { + sys::swapByteOrder(pl.count); +} + +inline void swapStruct(struct protocol_list32_t &pl) { + sys::swapByteOrder(pl.count); +} + +inline void swapStruct(struct protocol64_t &p) { + sys::swapByteOrder(p.isa); + sys::swapByteOrder(p.name); + sys::swapByteOrder(p.protocols); + sys::swapByteOrder(p.instanceMethods); + sys::swapByteOrder(p.classMethods); + sys::swapByteOrder(p.optionalInstanceMethods); + sys::swapByteOrder(p.optionalClassMethods); + sys::swapByteOrder(p.instanceProperties); +} + +inline void swapStruct(struct protocol32_t &p) { + sys::swapByteOrder(p.isa); + sys::swapByteOrder(p.name); + sys::swapByteOrder(p.protocols); + sys::swapByteOrder(p.instanceMethods); + sys::swapByteOrder(p.classMethods); + sys::swapByteOrder(p.optionalInstanceMethods); + sys::swapByteOrder(p.optionalClassMethods); + sys::swapByteOrder(p.instanceProperties); +} + +inline void swapStruct(struct ivar_list64_t &il) { + sys::swapByteOrder(il.entsize); + sys::swapByteOrder(il.count); +} + +inline void swapStruct(struct ivar_list32_t &il) { + sys::swapByteOrder(il.entsize); + sys::swapByteOrder(il.count); +} + +inline void swapStruct(struct ivar64_t &i) { + sys::swapByteOrder(i.offset); + sys::swapByteOrder(i.name); + sys::swapByteOrder(i.type); + sys::swapByteOrder(i.alignment); + sys::swapByteOrder(i.size); +} + +inline void swapStruct(struct ivar32_t &i) { + sys::swapByteOrder(i.offset); + sys::swapByteOrder(i.name); + sys::swapByteOrder(i.type); + sys::swapByteOrder(i.alignment); + sys::swapByteOrder(i.size); +} + +inline void swapStruct(struct objc_property_list64 &pl) { + sys::swapByteOrder(pl.entsize); + sys::swapByteOrder(pl.count); +} + +inline void swapStruct(struct objc_property_list32 &pl) { + sys::swapByteOrder(pl.entsize); + sys::swapByteOrder(pl.count); +} + +inline void swapStruct(struct objc_property64 &op) { + sys::swapByteOrder(op.name); + sys::swapByteOrder(op.attributes); +} + +inline void swapStruct(struct objc_property32 &op) { + sys::swapByteOrder(op.name); + sys::swapByteOrder(op.attributes); +} + +inline void swapStruct(struct category64_t &c) { + sys::swapByteOrder(c.name); + sys::swapByteOrder(c.cls); + sys::swapByteOrder(c.instanceMethods); + sys::swapByteOrder(c.classMethods); + sys::swapByteOrder(c.protocols); + sys::swapByteOrder(c.instanceProperties); +} + +inline void swapStruct(struct category32_t &c) { + sys::swapByteOrder(c.name); + sys::swapByteOrder(c.cls); + sys::swapByteOrder(c.instanceMethods); + sys::swapByteOrder(c.classMethods); + sys::swapByteOrder(c.protocols); + sys::swapByteOrder(c.instanceProperties); +} + +inline void swapStruct(struct objc_image_info64 &o) { + sys::swapByteOrder(o.version); + sys::swapByteOrder(o.flags); +} + +inline void swapStruct(struct objc_image_info32 &o) { + sys::swapByteOrder(o.version); + sys::swapByteOrder(o.flags); +} + +inline void swapStruct(struct imageInfo_t &o) { + sys::swapByteOrder(o.version); + sys::swapByteOrder(o.flags); +} + +inline void swapStruct(struct message_ref64 &mr) { + sys::swapByteOrder(mr.imp); + sys::swapByteOrder(mr.sel); +} + +inline void swapStruct(struct message_ref32 &mr) { + sys::swapByteOrder(mr.imp); + sys::swapByteOrder(mr.sel); +} + +inline void swapStruct(struct objc_module_t &module) { + sys::swapByteOrder(module.version); + sys::swapByteOrder(module.size); + sys::swapByteOrder(module.name); + sys::swapByteOrder(module.symtab); +} + +inline void swapStruct(struct objc_symtab_t &symtab) { + sys::swapByteOrder(symtab.sel_ref_cnt); + sys::swapByteOrder(symtab.refs); + sys::swapByteOrder(symtab.cls_def_cnt); + sys::swapByteOrder(symtab.cat_def_cnt); +} + +inline void swapStruct(struct objc_class_t &objc_class) { + sys::swapByteOrder(objc_class.isa); + sys::swapByteOrder(objc_class.super_class); + sys::swapByteOrder(objc_class.name); + sys::swapByteOrder(objc_class.version); + sys::swapByteOrder(objc_class.info); + sys::swapByteOrder(objc_class.instance_size); + sys::swapByteOrder(objc_class.ivars); + sys::swapByteOrder(objc_class.methodLists); + sys::swapByteOrder(objc_class.cache); + sys::swapByteOrder(objc_class.protocols); +} + +inline void swapStruct(struct objc_category_t &objc_category) { + sys::swapByteOrder(objc_category.category_name); + sys::swapByteOrder(objc_category.class_name); + sys::swapByteOrder(objc_category.instance_methods); + sys::swapByteOrder(objc_category.class_methods); + sys::swapByteOrder(objc_category.protocols); +} + +inline void swapStruct(struct objc_ivar_list_t &objc_ivar_list) { + sys::swapByteOrder(objc_ivar_list.ivar_count); +} + +inline void swapStruct(struct objc_ivar_t &objc_ivar) { + sys::swapByteOrder(objc_ivar.ivar_name); + sys::swapByteOrder(objc_ivar.ivar_type); + sys::swapByteOrder(objc_ivar.ivar_offset); +} + +inline void swapStruct(struct objc_method_list_t &method_list) { + sys::swapByteOrder(method_list.obsolete); + sys::swapByteOrder(method_list.method_count); +} + +inline void swapStruct(struct objc_method_t &method) { + sys::swapByteOrder(method.method_name); + sys::swapByteOrder(method.method_types); + sys::swapByteOrder(method.method_imp); +} + +inline void swapStruct(struct objc_protocol_list_t &protocol_list) { + sys::swapByteOrder(protocol_list.next); + sys::swapByteOrder(protocol_list.count); +} + +inline void swapStruct(struct objc_protocol_t &protocol) { + sys::swapByteOrder(protocol.isa); + sys::swapByteOrder(protocol.protocol_name); + sys::swapByteOrder(protocol.protocol_list); + sys::swapByteOrder(protocol.instance_methods); + sys::swapByteOrder(protocol.class_methods); +} + +inline void swapStruct(struct objc_method_description_list_t &mdl) { + sys::swapByteOrder(mdl.count); +} + +inline void swapStruct(struct objc_method_description_t &md) { + sys::swapByteOrder(md.name); + sys::swapByteOrder(md.types); +} + static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue, struct DisassembleInfo *info); @@ -2117,104 +3056,2495 @@ static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue, // address of the pointer, so when the pointer is zero as it can be in an .o // file, that is used to look for an external relocation entry with a symbol // name. -const char *get_objc2_64bit_class_name(uint64_t pointer_value, - uint64_t ReferenceValue, - struct DisassembleInfo *info) { +static const char *get_objc2_64bit_class_name(uint64_t pointer_value, + uint64_t ReferenceValue, + struct DisassembleInfo *info) { const char *r; uint32_t offset, left; SectionRef S; - // The pointer_value can be 0 in an object file and have a relocation - // entry for the class symbol at the ReferenceValue (the address of the - // pointer). - if (pointer_value == 0) { - r = get_pointer_64(ReferenceValue, offset, left, S, info); - if (r == nullptr || left < sizeof(uint64_t)) - return nullptr; - uint64_t n_value; - const char *symbol_name = get_symbol_64(offset, S, info, n_value); - if (symbol_name == nullptr) - return nullptr; - const char *class_name = strrchr(symbol_name, '$'); - if (class_name != nullptr && class_name[1] == '_' && class_name[2] != '\0') - return class_name + 2; - else - return nullptr; + // The pointer_value can be 0 in an object file and have a relocation + // entry for the class symbol at the ReferenceValue (the address of the + // pointer). + if (pointer_value == 0) { + r = get_pointer_64(ReferenceValue, offset, left, S, info); + if (r == nullptr || left < sizeof(uint64_t)) + return nullptr; + uint64_t n_value; + const char *symbol_name = get_symbol_64(offset, S, info, n_value); + if (symbol_name == nullptr) + return nullptr; + const char *class_name = strrchr(symbol_name, '$'); + if (class_name != nullptr && class_name[1] == '_' && class_name[2] != '\0') + return class_name + 2; + else + return nullptr; + } + + // The case were the pointer_value is non-zero and points to a class defined + // in this Mach-O file. + r = get_pointer_64(pointer_value, offset, left, S, info); + if (r == nullptr || left < sizeof(struct class64_t)) + return nullptr; + struct class64_t c; + memcpy(&c, r, sizeof(struct class64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(c); + if (c.data == 0) + return nullptr; + r = get_pointer_64(c.data, offset, left, S, info); + if (r == nullptr || left < sizeof(struct class_ro64_t)) + return nullptr; + struct class_ro64_t cro; + memcpy(&cro, r, sizeof(struct class_ro64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(cro); + if (cro.name == 0) + return nullptr; + const char *name = get_pointer_64(cro.name, offset, left, S, info); + return name; +} + +// get_objc2_64bit_cfstring_name is used for disassembly and is passed a +// pointer to a cfstring and returns its name or nullptr. +static const char *get_objc2_64bit_cfstring_name(uint64_t ReferenceValue, + struct DisassembleInfo *info) { + const char *r, *name; + uint32_t offset, left; + SectionRef S; + struct cfstring64_t cfs; + uint64_t cfs_characters; + + r = get_pointer_64(ReferenceValue, offset, left, S, info); + if (r == nullptr || left < sizeof(struct cfstring64_t)) + return nullptr; + memcpy(&cfs, r, sizeof(struct cfstring64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(cfs); + if (cfs.characters == 0) { + uint64_t n_value; + const char *symbol_name = get_symbol_64( + offset + offsetof(struct cfstring64_t, characters), S, info, n_value); + if (symbol_name == nullptr) + return nullptr; + cfs_characters = n_value; + } else + cfs_characters = cfs.characters; + name = get_pointer_64(cfs_characters, offset, left, S, info); + + return name; +} + +// get_objc2_64bit_selref() is used for disassembly and is passed a the address +// of a pointer to an Objective-C selector reference when the pointer value is +// zero as in a .o file and is likely to have a external relocation entry with +// who's symbol's n_value is the real pointer to the selector name. If that is +// the case the real pointer to the selector name is returned else 0 is +// returned +static uint64_t get_objc2_64bit_selref(uint64_t ReferenceValue, + struct DisassembleInfo *info) { + uint32_t offset, left; + SectionRef S; + + const char *r = get_pointer_64(ReferenceValue, offset, left, S, info); + if (r == nullptr || left < sizeof(uint64_t)) + return 0; + uint64_t n_value; + const char *symbol_name = get_symbol_64(offset, S, info, n_value); + if (symbol_name == nullptr) + return 0; + return n_value; +} + +static const SectionRef get_section(MachOObjectFile *O, const char *segname, + const char *sectname) { + for (const SectionRef &Section : O->sections()) { + StringRef SectName; + Section.getName(SectName); + DataRefImpl Ref = Section.getRawDataRefImpl(); + StringRef SegName = O->getSectionFinalSegmentName(Ref); + if (SegName == segname && SectName == sectname) + return Section; + } + return SectionRef(); +} + +static void +walk_pointer_list_64(const char *listname, const SectionRef S, + MachOObjectFile *O, struct DisassembleInfo *info, + void (*func)(uint64_t, struct DisassembleInfo *info)) { + if (S == SectionRef()) + return; + + StringRef SectName; + S.getName(SectName); + DataRefImpl Ref = S.getRawDataRefImpl(); + StringRef SegName = O->getSectionFinalSegmentName(Ref); + outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; + + StringRef BytesStr; + S.getContents(BytesStr); + const char *Contents = reinterpret_cast(BytesStr.data()); + + for (uint32_t i = 0; i < S.getSize(); i += sizeof(uint64_t)) { + uint32_t left = S.getSize() - i; + uint32_t size = left < sizeof(uint64_t) ? left : sizeof(uint64_t); + uint64_t p = 0; + memcpy(&p, Contents + i, size); + if (i + sizeof(uint64_t) > S.getSize()) + outs() << listname << " list pointer extends past end of (" << SegName + << "," << SectName << ") section\n"; + outs() << format("%016" PRIx64, S.getAddress() + i) << " "; + + if (O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(p); + + uint64_t n_value = 0; + const char *name = get_symbol_64(i, S, info, n_value, p); + if (name == nullptr) + name = get_dyld_bind_info_symbolname(S.getAddress() + i, info); + + if (n_value != 0) { + outs() << format("0x%" PRIx64, n_value); + if (p != 0) + outs() << " + " << format("0x%" PRIx64, p); + } else + outs() << format("0x%" PRIx64, p); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + p += n_value; + if (func) + func(p, info); + } +} + +static void +walk_pointer_list_32(const char *listname, const SectionRef S, + MachOObjectFile *O, struct DisassembleInfo *info, + void (*func)(uint32_t, struct DisassembleInfo *info)) { + if (S == SectionRef()) + return; + + StringRef SectName; + S.getName(SectName); + DataRefImpl Ref = S.getRawDataRefImpl(); + StringRef SegName = O->getSectionFinalSegmentName(Ref); + outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; + + StringRef BytesStr; + S.getContents(BytesStr); + const char *Contents = reinterpret_cast(BytesStr.data()); + + for (uint32_t i = 0; i < S.getSize(); i += sizeof(uint32_t)) { + uint32_t left = S.getSize() - i; + uint32_t size = left < sizeof(uint32_t) ? left : sizeof(uint32_t); + uint32_t p = 0; + memcpy(&p, Contents + i, size); + if (i + sizeof(uint32_t) > S.getSize()) + outs() << listname << " list pointer extends past end of (" << SegName + << "," << SectName << ") section\n"; + uint32_t Address = S.getAddress() + i; + outs() << format("%08" PRIx32, Address) << " "; + + if (O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(p); + outs() << format("0x%" PRIx32, p); + + const char *name = get_symbol_32(i, S, info, p); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + if (func) + func(p, info); + } +} + +static void print_layout_map(const char *layout_map, uint32_t left) { + if (layout_map == nullptr) + return; + outs() << " layout map: "; + do { + outs() << format("0x%02" PRIx32, (*layout_map) & 0xff) << " "; + left--; + layout_map++; + } while (*layout_map != '\0' && left != 0); + outs() << "\n"; +} + +static void print_layout_map64(uint64_t p, struct DisassembleInfo *info) { + uint32_t offset, left; + SectionRef S; + const char *layout_map; + + if (p == 0) + return; + layout_map = get_pointer_64(p, offset, left, S, info); + print_layout_map(layout_map, left); +} + +static void print_layout_map32(uint32_t p, struct DisassembleInfo *info) { + uint32_t offset, left; + SectionRef S; + const char *layout_map; + + if (p == 0) + return; + layout_map = get_pointer_32(p, offset, left, S, info); + print_layout_map(layout_map, left); +} + +static void print_method_list64_t(uint64_t p, struct DisassembleInfo *info, + const char *indent) { + struct method_list64_t ml; + struct method64_t m; + const char *r; + uint32_t offset, xoffset, left, i; + SectionRef S, xS; + const char *name, *sym_name; + uint64_t n_value; + + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&ml, '\0', sizeof(struct method_list64_t)); + if (left < sizeof(struct method_list64_t)) { + memcpy(&ml, r, left); + outs() << " (method_list_t entends past the end of the section)\n"; + } else + memcpy(&ml, r, sizeof(struct method_list64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(ml); + outs() << indent << "\t\t entsize " << ml.entsize << "\n"; + outs() << indent << "\t\t count " << ml.count << "\n"; + + p += sizeof(struct method_list64_t); + offset += sizeof(struct method_list64_t); + for (i = 0; i < ml.count; i++) { + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&m, '\0', sizeof(struct method64_t)); + if (left < sizeof(struct method64_t)) { + memcpy(&m, r, left); + outs() << indent << " (method_t extends past the end of the section)\n"; + } else + memcpy(&m, r, sizeof(struct method64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(m); + + outs() << indent << "\t\t name "; + sym_name = get_symbol_64(offset + offsetof(struct method64_t, name), S, + info, n_value, m.name); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (m.name != 0) + outs() << " + " << format("0x%" PRIx64, m.name); + } else + outs() << format("0x%" PRIx64, m.name); + name = get_pointer_64(m.name + n_value, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << indent << "\t\t types "; + sym_name = get_symbol_64(offset + offsetof(struct method64_t, types), S, + info, n_value, m.types); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (m.types != 0) + outs() << " + " << format("0x%" PRIx64, m.types); + } else + outs() << format("0x%" PRIx64, m.types); + name = get_pointer_64(m.types + n_value, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << indent << "\t\t imp "; + name = get_symbol_64(offset + offsetof(struct method64_t, imp), S, info, + n_value, m.imp); + if (info->verbose && name == nullptr) { + if (n_value != 0) { + outs() << format("0x%" PRIx64, n_value) << " "; + if (m.imp != 0) + outs() << "+ " << format("0x%" PRIx64, m.imp) << " "; + } else + outs() << format("0x%" PRIx64, m.imp) << " "; + } + if (name != nullptr) + outs() << name; + outs() << "\n"; + + p += sizeof(struct method64_t); + offset += sizeof(struct method64_t); + } +} + +static void print_method_list32_t(uint64_t p, struct DisassembleInfo *info, + const char *indent) { + struct method_list32_t ml; + struct method32_t m; + const char *r, *name; + uint32_t offset, xoffset, left, i; + SectionRef S, xS; + + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&ml, '\0', sizeof(struct method_list32_t)); + if (left < sizeof(struct method_list32_t)) { + memcpy(&ml, r, left); + outs() << " (method_list_t entends past the end of the section)\n"; + } else + memcpy(&ml, r, sizeof(struct method_list32_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(ml); + outs() << indent << "\t\t entsize " << ml.entsize << "\n"; + outs() << indent << "\t\t count " << ml.count << "\n"; + + p += sizeof(struct method_list32_t); + offset += sizeof(struct method_list32_t); + for (i = 0; i < ml.count; i++) { + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&m, '\0', sizeof(struct method32_t)); + if (left < sizeof(struct method32_t)) { + memcpy(&ml, r, left); + outs() << indent << " (method_t entends past the end of the section)\n"; + } else + memcpy(&m, r, sizeof(struct method32_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(m); + + outs() << indent << "\t\t name " << format("0x%" PRIx32, m.name); + name = get_pointer_32(m.name, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << indent << "\t\t types " << format("0x%" PRIx32, m.types); + name = get_pointer_32(m.types, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << indent << "\t\t imp " << format("0x%" PRIx32, m.imp); + name = get_symbol_32(offset + offsetof(struct method32_t, imp), S, info, + m.imp); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + p += sizeof(struct method32_t); + offset += sizeof(struct method32_t); + } +} + +static bool print_method_list(uint32_t p, struct DisassembleInfo *info) { + uint32_t offset, left, xleft; + SectionRef S; + struct objc_method_list_t method_list; + struct objc_method_t method; + const char *r, *methods, *name, *SymbolName; + int32_t i; + + r = get_pointer_32(p, offset, left, S, info, true); + if (r == nullptr) + return true; + + outs() << "\n"; + if (left > sizeof(struct objc_method_list_t)) { + memcpy(&method_list, r, sizeof(struct objc_method_list_t)); + } else { + outs() << "\t\t objc_method_list extends past end of the section\n"; + memset(&method_list, '\0', sizeof(struct objc_method_list_t)); + memcpy(&method_list, r, left); + } + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(method_list); + + outs() << "\t\t obsolete " + << format("0x%08" PRIx32, method_list.obsolete) << "\n"; + outs() << "\t\t method_count " << method_list.method_count << "\n"; + + methods = r + sizeof(struct objc_method_list_t); + for (i = 0; i < method_list.method_count; i++) { + if ((i + 1) * sizeof(struct objc_method_t) > left) { + outs() << "\t\t remaining method's extend past the of the section\n"; + break; + } + memcpy(&method, methods + i * sizeof(struct objc_method_t), + sizeof(struct objc_method_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(method); + + outs() << "\t\t method_name " + << format("0x%08" PRIx32, method.method_name); + if (info->verbose) { + name = get_pointer_32(method.method_name, offset, xleft, S, info, true); + if (name != nullptr) + outs() << format(" %.*s", xleft, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + + outs() << "\t\t method_types " + << format("0x%08" PRIx32, method.method_types); + if (info->verbose) { + name = get_pointer_32(method.method_types, offset, xleft, S, info, true); + if (name != nullptr) + outs() << format(" %.*s", xleft, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + + outs() << "\t\t method_imp " + << format("0x%08" PRIx32, method.method_imp) << " "; + if (info->verbose) { + SymbolName = GuessSymbolName(method.method_imp, info->AddrMap); + if (SymbolName != nullptr) + outs() << SymbolName; + } + outs() << "\n"; + } + return false; +} + +static void print_protocol_list64_t(uint64_t p, struct DisassembleInfo *info) { + struct protocol_list64_t pl; + uint64_t q, n_value; + struct protocol64_t pc; + const char *r; + uint32_t offset, xoffset, left, i; + SectionRef S, xS; + const char *name, *sym_name; + + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&pl, '\0', sizeof(struct protocol_list64_t)); + if (left < sizeof(struct protocol_list64_t)) { + memcpy(&pl, r, left); + outs() << " (protocol_list_t entends past the end of the section)\n"; + } else + memcpy(&pl, r, sizeof(struct protocol_list64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(pl); + outs() << " count " << pl.count << "\n"; + + p += sizeof(struct protocol_list64_t); + offset += sizeof(struct protocol_list64_t); + for (i = 0; i < pl.count; i++) { + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr) + return; + q = 0; + if (left < sizeof(uint64_t)) { + memcpy(&q, r, left); + outs() << " (protocol_t * entends past the end of the section)\n"; + } else + memcpy(&q, r, sizeof(uint64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(q); + + outs() << "\t\t list[" << i << "] "; + sym_name = get_symbol_64(offset, S, info, n_value, q); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (q != 0) + outs() << " + " << format("0x%" PRIx64, q); + } else + outs() << format("0x%" PRIx64, q); + outs() << " (struct protocol_t *)\n"; + + r = get_pointer_64(q + n_value, offset, left, S, info); + if (r == nullptr) + return; + memset(&pc, '\0', sizeof(struct protocol64_t)); + if (left < sizeof(struct protocol64_t)) { + memcpy(&pc, r, left); + outs() << " (protocol_t entends past the end of the section)\n"; + } else + memcpy(&pc, r, sizeof(struct protocol64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(pc); + + outs() << "\t\t\t isa " << format("0x%" PRIx64, pc.isa) << "\n"; + + outs() << "\t\t\t name "; + sym_name = get_symbol_64(offset + offsetof(struct protocol64_t, name), S, + info, n_value, pc.name); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (pc.name != 0) + outs() << " + " << format("0x%" PRIx64, pc.name); + } else + outs() << format("0x%" PRIx64, pc.name); + name = get_pointer_64(pc.name + n_value, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << "\t\t\tprotocols " << format("0x%" PRIx64, pc.protocols) << "\n"; + + outs() << "\t\t instanceMethods "; + sym_name = + get_symbol_64(offset + offsetof(struct protocol64_t, instanceMethods), + S, info, n_value, pc.instanceMethods); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (pc.instanceMethods != 0) + outs() << " + " << format("0x%" PRIx64, pc.instanceMethods); + } else + outs() << format("0x%" PRIx64, pc.instanceMethods); + outs() << " (struct method_list_t *)\n"; + if (pc.instanceMethods + n_value != 0) + print_method_list64_t(pc.instanceMethods + n_value, info, "\t"); + + outs() << "\t\t classMethods "; + sym_name = + get_symbol_64(offset + offsetof(struct protocol64_t, classMethods), S, + info, n_value, pc.classMethods); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (pc.classMethods != 0) + outs() << " + " << format("0x%" PRIx64, pc.classMethods); + } else + outs() << format("0x%" PRIx64, pc.classMethods); + outs() << " (struct method_list_t *)\n"; + if (pc.classMethods + n_value != 0) + print_method_list64_t(pc.classMethods + n_value, info, "\t"); + + outs() << "\t optionalInstanceMethods " + << format("0x%" PRIx64, pc.optionalInstanceMethods) << "\n"; + outs() << "\t optionalClassMethods " + << format("0x%" PRIx64, pc.optionalClassMethods) << "\n"; + outs() << "\t instanceProperties " + << format("0x%" PRIx64, pc.instanceProperties) << "\n"; + + p += sizeof(uint64_t); + offset += sizeof(uint64_t); + } +} + +static void print_protocol_list32_t(uint32_t p, struct DisassembleInfo *info) { + struct protocol_list32_t pl; + uint32_t q; + struct protocol32_t pc; + const char *r; + uint32_t offset, xoffset, left, i; + SectionRef S, xS; + const char *name; + + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&pl, '\0', sizeof(struct protocol_list32_t)); + if (left < sizeof(struct protocol_list32_t)) { + memcpy(&pl, r, left); + outs() << " (protocol_list_t entends past the end of the section)\n"; + } else + memcpy(&pl, r, sizeof(struct protocol_list32_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(pl); + outs() << " count " << pl.count << "\n"; + + p += sizeof(struct protocol_list32_t); + offset += sizeof(struct protocol_list32_t); + for (i = 0; i < pl.count; i++) { + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + q = 0; + if (left < sizeof(uint32_t)) { + memcpy(&q, r, left); + outs() << " (protocol_t * entends past the end of the section)\n"; + } else + memcpy(&q, r, sizeof(uint32_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(q); + outs() << "\t\t list[" << i << "] " << format("0x%" PRIx32, q) + << " (struct protocol_t *)\n"; + r = get_pointer_32(q, offset, left, S, info); + if (r == nullptr) + return; + memset(&pc, '\0', sizeof(struct protocol32_t)); + if (left < sizeof(struct protocol32_t)) { + memcpy(&pc, r, left); + outs() << " (protocol_t entends past the end of the section)\n"; + } else + memcpy(&pc, r, sizeof(struct protocol32_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(pc); + outs() << "\t\t\t isa " << format("0x%" PRIx32, pc.isa) << "\n"; + outs() << "\t\t\t name " << format("0x%" PRIx32, pc.name); + name = get_pointer_32(pc.name, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + outs() << "\t\t\tprotocols " << format("0x%" PRIx32, pc.protocols) << "\n"; + outs() << "\t\t instanceMethods " + << format("0x%" PRIx32, pc.instanceMethods) + << " (struct method_list_t *)\n"; + if (pc.instanceMethods != 0) + print_method_list32_t(pc.instanceMethods, info, "\t"); + outs() << "\t\t classMethods " << format("0x%" PRIx32, pc.classMethods) + << " (struct method_list_t *)\n"; + if (pc.classMethods != 0) + print_method_list32_t(pc.classMethods, info, "\t"); + outs() << "\t optionalInstanceMethods " + << format("0x%" PRIx32, pc.optionalInstanceMethods) << "\n"; + outs() << "\t optionalClassMethods " + << format("0x%" PRIx32, pc.optionalClassMethods) << "\n"; + outs() << "\t instanceProperties " + << format("0x%" PRIx32, pc.instanceProperties) << "\n"; + p += sizeof(uint32_t); + offset += sizeof(uint32_t); + } +} + +static void print_indent(uint32_t indent) { + for (uint32_t i = 0; i < indent;) { + if (indent - i >= 8) { + outs() << "\t"; + i += 8; + } else { + for (uint32_t j = i; j < indent; j++) + outs() << " "; + return; + } + } +} + +static bool print_method_description_list(uint32_t p, uint32_t indent, + struct DisassembleInfo *info) { + uint32_t offset, left, xleft; + SectionRef S; + struct objc_method_description_list_t mdl; + struct objc_method_description_t md; + const char *r, *list, *name; + int32_t i; + + r = get_pointer_32(p, offset, left, S, info, true); + if (r == nullptr) + return true; + + outs() << "\n"; + if (left > sizeof(struct objc_method_description_list_t)) { + memcpy(&mdl, r, sizeof(struct objc_method_description_list_t)); + } else { + print_indent(indent); + outs() << " objc_method_description_list extends past end of the section\n"; + memset(&mdl, '\0', sizeof(struct objc_method_description_list_t)); + memcpy(&mdl, r, left); + } + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(mdl); + + print_indent(indent); + outs() << " count " << mdl.count << "\n"; + + list = r + sizeof(struct objc_method_description_list_t); + for (i = 0; i < mdl.count; i++) { + if ((i + 1) * sizeof(struct objc_method_description_t) > left) { + print_indent(indent); + outs() << " remaining list entries extend past the of the section\n"; + break; + } + print_indent(indent); + outs() << " list[" << i << "]\n"; + memcpy(&md, list + i * sizeof(struct objc_method_description_t), + sizeof(struct objc_method_description_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(md); + + print_indent(indent); + outs() << " name " << format("0x%08" PRIx32, md.name); + if (info->verbose) { + name = get_pointer_32(md.name, offset, xleft, S, info, true); + if (name != nullptr) + outs() << format(" %.*s", xleft, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + + print_indent(indent); + outs() << " types " << format("0x%08" PRIx32, md.types); + if (info->verbose) { + name = get_pointer_32(md.types, offset, xleft, S, info, true); + if (name != nullptr) + outs() << format(" %.*s", xleft, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + } + return false; +} + +static bool print_protocol_list(uint32_t p, uint32_t indent, + struct DisassembleInfo *info); + +static bool print_protocol(uint32_t p, uint32_t indent, + struct DisassembleInfo *info) { + uint32_t offset, left; + SectionRef S; + struct objc_protocol_t protocol; + const char *r, *name; + + r = get_pointer_32(p, offset, left, S, info, true); + if (r == nullptr) + return true; + + outs() << "\n"; + if (left >= sizeof(struct objc_protocol_t)) { + memcpy(&protocol, r, sizeof(struct objc_protocol_t)); + } else { + print_indent(indent); + outs() << " Protocol extends past end of the section\n"; + memset(&protocol, '\0', sizeof(struct objc_protocol_t)); + memcpy(&protocol, r, left); + } + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(protocol); + + print_indent(indent); + outs() << " isa " << format("0x%08" PRIx32, protocol.isa) + << "\n"; + + print_indent(indent); + outs() << " protocol_name " + << format("0x%08" PRIx32, protocol.protocol_name); + if (info->verbose) { + name = get_pointer_32(protocol.protocol_name, offset, left, S, info, true); + if (name != nullptr) + outs() << format(" %.*s", left, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + + print_indent(indent); + outs() << " protocol_list " + << format("0x%08" PRIx32, protocol.protocol_list); + if (print_protocol_list(protocol.protocol_list, indent + 4, info)) + outs() << " (not in an __OBJC section)\n"; + + print_indent(indent); + outs() << " instance_methods " + << format("0x%08" PRIx32, protocol.instance_methods); + if (print_method_description_list(protocol.instance_methods, indent, info)) + outs() << " (not in an __OBJC section)\n"; + + print_indent(indent); + outs() << " class_methods " + << format("0x%08" PRIx32, protocol.class_methods); + if (print_method_description_list(protocol.class_methods, indent, info)) + outs() << " (not in an __OBJC section)\n"; + + return false; +} + +static bool print_protocol_list(uint32_t p, uint32_t indent, + struct DisassembleInfo *info) { + uint32_t offset, left, l; + SectionRef S; + struct objc_protocol_list_t protocol_list; + const char *r, *list; + int32_t i; + + r = get_pointer_32(p, offset, left, S, info, true); + if (r == nullptr) + return true; + + outs() << "\n"; + if (left > sizeof(struct objc_protocol_list_t)) { + memcpy(&protocol_list, r, sizeof(struct objc_protocol_list_t)); + } else { + outs() << "\t\t objc_protocol_list_t extends past end of the section\n"; + memset(&protocol_list, '\0', sizeof(struct objc_protocol_list_t)); + memcpy(&protocol_list, r, left); + } + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(protocol_list); + + print_indent(indent); + outs() << " next " << format("0x%08" PRIx32, protocol_list.next) + << "\n"; + print_indent(indent); + outs() << " count " << protocol_list.count << "\n"; + + list = r + sizeof(struct objc_protocol_list_t); + for (i = 0; i < protocol_list.count; i++) { + if ((i + 1) * sizeof(uint32_t) > left) { + outs() << "\t\t remaining list entries extend past the of the section\n"; + break; + } + memcpy(&l, list + i * sizeof(uint32_t), sizeof(uint32_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(l); + + print_indent(indent); + outs() << " list[" << i << "] " << format("0x%08" PRIx32, l); + if (print_protocol(l, indent, info)) + outs() << "(not in an __OBJC section)\n"; + } + return false; +} + +static void print_ivar_list64_t(uint64_t p, struct DisassembleInfo *info) { + struct ivar_list64_t il; + struct ivar64_t i; + const char *r; + uint32_t offset, xoffset, left, j; + SectionRef S, xS; + const char *name, *sym_name, *ivar_offset_p; + uint64_t ivar_offset, n_value; + + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&il, '\0', sizeof(struct ivar_list64_t)); + if (left < sizeof(struct ivar_list64_t)) { + memcpy(&il, r, left); + outs() << " (ivar_list_t entends past the end of the section)\n"; + } else + memcpy(&il, r, sizeof(struct ivar_list64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(il); + outs() << " entsize " << il.entsize << "\n"; + outs() << " count " << il.count << "\n"; + + p += sizeof(struct ivar_list64_t); + offset += sizeof(struct ivar_list64_t); + for (j = 0; j < il.count; j++) { + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&i, '\0', sizeof(struct ivar64_t)); + if (left < sizeof(struct ivar64_t)) { + memcpy(&i, r, left); + outs() << " (ivar_t entends past the end of the section)\n"; + } else + memcpy(&i, r, sizeof(struct ivar64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(i); + + outs() << "\t\t\t offset "; + sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, offset), S, + info, n_value, i.offset); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (i.offset != 0) + outs() << " + " << format("0x%" PRIx64, i.offset); + } else + outs() << format("0x%" PRIx64, i.offset); + ivar_offset_p = get_pointer_64(i.offset + n_value, xoffset, left, xS, info); + if (ivar_offset_p != nullptr && left >= sizeof(*ivar_offset_p)) { + memcpy(&ivar_offset, ivar_offset_p, sizeof(ivar_offset)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(ivar_offset); + outs() << " " << ivar_offset << "\n"; + } else + outs() << "\n"; + + outs() << "\t\t\t name "; + sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, name), S, info, + n_value, i.name); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (i.name != 0) + outs() << " + " << format("0x%" PRIx64, i.name); + } else + outs() << format("0x%" PRIx64, i.name); + name = get_pointer_64(i.name + n_value, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << "\t\t\t type "; + sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, type), S, info, + n_value, i.name); + name = get_pointer_64(i.type + n_value, xoffset, left, xS, info); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (i.type != 0) + outs() << " + " << format("0x%" PRIx64, i.type); + } else + outs() << format("0x%" PRIx64, i.type); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << "\t\t\talignment " << i.alignment << "\n"; + outs() << "\t\t\t size " << i.size << "\n"; + + p += sizeof(struct ivar64_t); + offset += sizeof(struct ivar64_t); + } +} + +static void print_ivar_list32_t(uint32_t p, struct DisassembleInfo *info) { + struct ivar_list32_t il; + struct ivar32_t i; + const char *r; + uint32_t offset, xoffset, left, j; + SectionRef S, xS; + const char *name, *ivar_offset_p; + uint32_t ivar_offset; + + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&il, '\0', sizeof(struct ivar_list32_t)); + if (left < sizeof(struct ivar_list32_t)) { + memcpy(&il, r, left); + outs() << " (ivar_list_t entends past the end of the section)\n"; + } else + memcpy(&il, r, sizeof(struct ivar_list32_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(il); + outs() << " entsize " << il.entsize << "\n"; + outs() << " count " << il.count << "\n"; + + p += sizeof(struct ivar_list32_t); + offset += sizeof(struct ivar_list32_t); + for (j = 0; j < il.count; j++) { + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&i, '\0', sizeof(struct ivar32_t)); + if (left < sizeof(struct ivar32_t)) { + memcpy(&i, r, left); + outs() << " (ivar_t entends past the end of the section)\n"; + } else + memcpy(&i, r, sizeof(struct ivar32_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(i); + + outs() << "\t\t\t offset " << format("0x%" PRIx32, i.offset); + ivar_offset_p = get_pointer_32(i.offset, xoffset, left, xS, info); + if (ivar_offset_p != nullptr && left >= sizeof(*ivar_offset_p)) { + memcpy(&ivar_offset, ivar_offset_p, sizeof(ivar_offset)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(ivar_offset); + outs() << " " << ivar_offset << "\n"; + } else + outs() << "\n"; + + outs() << "\t\t\t name " << format("0x%" PRIx32, i.name); + name = get_pointer_32(i.name, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << "\t\t\t type " << format("0x%" PRIx32, i.type); + name = get_pointer_32(i.type, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << "\t\t\talignment " << i.alignment << "\n"; + outs() << "\t\t\t size " << i.size << "\n"; + + p += sizeof(struct ivar32_t); + offset += sizeof(struct ivar32_t); + } +} + +static void print_objc_property_list64(uint64_t p, + struct DisassembleInfo *info) { + struct objc_property_list64 opl; + struct objc_property64 op; + const char *r; + uint32_t offset, xoffset, left, j; + SectionRef S, xS; + const char *name, *sym_name; + uint64_t n_value; + + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&opl, '\0', sizeof(struct objc_property_list64)); + if (left < sizeof(struct objc_property_list64)) { + memcpy(&opl, r, left); + outs() << " (objc_property_list entends past the end of the section)\n"; + } else + memcpy(&opl, r, sizeof(struct objc_property_list64)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(opl); + outs() << " entsize " << opl.entsize << "\n"; + outs() << " count " << opl.count << "\n"; + + p += sizeof(struct objc_property_list64); + offset += sizeof(struct objc_property_list64); + for (j = 0; j < opl.count; j++) { + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&op, '\0', sizeof(struct objc_property64)); + if (left < sizeof(struct objc_property64)) { + memcpy(&op, r, left); + outs() << " (objc_property entends past the end of the section)\n"; + } else + memcpy(&op, r, sizeof(struct objc_property64)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(op); + + outs() << "\t\t\t name "; + sym_name = get_symbol_64(offset + offsetof(struct objc_property64, name), S, + info, n_value, op.name); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (op.name != 0) + outs() << " + " << format("0x%" PRIx64, op.name); + } else + outs() << format("0x%" PRIx64, op.name); + name = get_pointer_64(op.name + n_value, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << "\t\t\tattributes "; + sym_name = + get_symbol_64(offset + offsetof(struct objc_property64, attributes), S, + info, n_value, op.attributes); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (op.attributes != 0) + outs() << " + " << format("0x%" PRIx64, op.attributes); + } else + outs() << format("0x%" PRIx64, op.attributes); + name = get_pointer_64(op.attributes + n_value, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + p += sizeof(struct objc_property64); + offset += sizeof(struct objc_property64); + } +} + +static void print_objc_property_list32(uint32_t p, + struct DisassembleInfo *info) { + struct objc_property_list32 opl; + struct objc_property32 op; + const char *r; + uint32_t offset, xoffset, left, j; + SectionRef S, xS; + const char *name; + + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&opl, '\0', sizeof(struct objc_property_list32)); + if (left < sizeof(struct objc_property_list32)) { + memcpy(&opl, r, left); + outs() << " (objc_property_list entends past the end of the section)\n"; + } else + memcpy(&opl, r, sizeof(struct objc_property_list32)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(opl); + outs() << " entsize " << opl.entsize << "\n"; + outs() << " count " << opl.count << "\n"; + + p += sizeof(struct objc_property_list32); + offset += sizeof(struct objc_property_list32); + for (j = 0; j < opl.count; j++) { + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&op, '\0', sizeof(struct objc_property32)); + if (left < sizeof(struct objc_property32)) { + memcpy(&op, r, left); + outs() << " (objc_property entends past the end of the section)\n"; + } else + memcpy(&op, r, sizeof(struct objc_property32)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(op); + + outs() << "\t\t\t name " << format("0x%" PRIx32, op.name); + name = get_pointer_32(op.name, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << "\t\t\tattributes " << format("0x%" PRIx32, op.attributes); + name = get_pointer_32(op.attributes, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + p += sizeof(struct objc_property32); + offset += sizeof(struct objc_property32); + } +} + +static bool print_class_ro64_t(uint64_t p, struct DisassembleInfo *info, + bool &is_meta_class) { + struct class_ro64_t cro; + const char *r; + uint32_t offset, xoffset, left; + SectionRef S, xS; + const char *name, *sym_name; + uint64_t n_value; + + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr || left < sizeof(struct class_ro64_t)) + return false; + memset(&cro, '\0', sizeof(struct class_ro64_t)); + if (left < sizeof(struct class_ro64_t)) { + memcpy(&cro, r, left); + outs() << " (class_ro_t entends past the end of the section)\n"; + } else + memcpy(&cro, r, sizeof(struct class_ro64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(cro); + outs() << " flags " << format("0x%" PRIx32, cro.flags); + if (cro.flags & RO_META) + outs() << " RO_META"; + if (cro.flags & RO_ROOT) + outs() << " RO_ROOT"; + if (cro.flags & RO_HAS_CXX_STRUCTORS) + outs() << " RO_HAS_CXX_STRUCTORS"; + outs() << "\n"; + outs() << " instanceStart " << cro.instanceStart << "\n"; + outs() << " instanceSize " << cro.instanceSize << "\n"; + outs() << " reserved " << format("0x%" PRIx32, cro.reserved) + << "\n"; + outs() << " ivarLayout " << format("0x%" PRIx64, cro.ivarLayout) + << "\n"; + print_layout_map64(cro.ivarLayout, info); + + outs() << " name "; + sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, name), S, + info, n_value, cro.name); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (cro.name != 0) + outs() << " + " << format("0x%" PRIx64, cro.name); + } else + outs() << format("0x%" PRIx64, cro.name); + name = get_pointer_64(cro.name + n_value, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << " baseMethods "; + sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, baseMethods), + S, info, n_value, cro.baseMethods); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (cro.baseMethods != 0) + outs() << " + " << format("0x%" PRIx64, cro.baseMethods); + } else + outs() << format("0x%" PRIx64, cro.baseMethods); + outs() << " (struct method_list_t *)\n"; + if (cro.baseMethods + n_value != 0) + print_method_list64_t(cro.baseMethods + n_value, info, ""); + + outs() << " baseProtocols "; + sym_name = + get_symbol_64(offset + offsetof(struct class_ro64_t, baseProtocols), S, + info, n_value, cro.baseProtocols); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (cro.baseProtocols != 0) + outs() << " + " << format("0x%" PRIx64, cro.baseProtocols); + } else + outs() << format("0x%" PRIx64, cro.baseProtocols); + outs() << "\n"; + if (cro.baseProtocols + n_value != 0) + print_protocol_list64_t(cro.baseProtocols + n_value, info); + + outs() << " ivars "; + sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, ivars), S, + info, n_value, cro.ivars); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (cro.ivars != 0) + outs() << " + " << format("0x%" PRIx64, cro.ivars); + } else + outs() << format("0x%" PRIx64, cro.ivars); + outs() << "\n"; + if (cro.ivars + n_value != 0) + print_ivar_list64_t(cro.ivars + n_value, info); + + outs() << " weakIvarLayout "; + sym_name = + get_symbol_64(offset + offsetof(struct class_ro64_t, weakIvarLayout), S, + info, n_value, cro.weakIvarLayout); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (cro.weakIvarLayout != 0) + outs() << " + " << format("0x%" PRIx64, cro.weakIvarLayout); + } else + outs() << format("0x%" PRIx64, cro.weakIvarLayout); + outs() << "\n"; + print_layout_map64(cro.weakIvarLayout + n_value, info); + + outs() << " baseProperties "; + sym_name = + get_symbol_64(offset + offsetof(struct class_ro64_t, baseProperties), S, + info, n_value, cro.baseProperties); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (cro.baseProperties != 0) + outs() << " + " << format("0x%" PRIx64, cro.baseProperties); + } else + outs() << format("0x%" PRIx64, cro.baseProperties); + outs() << "\n"; + if (cro.baseProperties + n_value != 0) + print_objc_property_list64(cro.baseProperties + n_value, info); + + is_meta_class = (cro.flags & RO_META) != 0; + return true; +} + +static bool print_class_ro32_t(uint32_t p, struct DisassembleInfo *info, + bool &is_meta_class) { + struct class_ro32_t cro; + const char *r; + uint32_t offset, xoffset, left; + SectionRef S, xS; + const char *name; + + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return false; + memset(&cro, '\0', sizeof(struct class_ro32_t)); + if (left < sizeof(struct class_ro32_t)) { + memcpy(&cro, r, left); + outs() << " (class_ro_t entends past the end of the section)\n"; + } else + memcpy(&cro, r, sizeof(struct class_ro32_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(cro); + outs() << " flags " << format("0x%" PRIx32, cro.flags); + if (cro.flags & RO_META) + outs() << " RO_META"; + if (cro.flags & RO_ROOT) + outs() << " RO_ROOT"; + if (cro.flags & RO_HAS_CXX_STRUCTORS) + outs() << " RO_HAS_CXX_STRUCTORS"; + outs() << "\n"; + outs() << " instanceStart " << cro.instanceStart << "\n"; + outs() << " instanceSize " << cro.instanceSize << "\n"; + outs() << " ivarLayout " << format("0x%" PRIx32, cro.ivarLayout) + << "\n"; + print_layout_map32(cro.ivarLayout, info); + + outs() << " name " << format("0x%" PRIx32, cro.name); + name = get_pointer_32(cro.name, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << " baseMethods " + << format("0x%" PRIx32, cro.baseMethods) + << " (struct method_list_t *)\n"; + if (cro.baseMethods != 0) + print_method_list32_t(cro.baseMethods, info, ""); + + outs() << " baseProtocols " + << format("0x%" PRIx32, cro.baseProtocols) << "\n"; + if (cro.baseProtocols != 0) + print_protocol_list32_t(cro.baseProtocols, info); + outs() << " ivars " << format("0x%" PRIx32, cro.ivars) + << "\n"; + if (cro.ivars != 0) + print_ivar_list32_t(cro.ivars, info); + outs() << " weakIvarLayout " + << format("0x%" PRIx32, cro.weakIvarLayout) << "\n"; + print_layout_map32(cro.weakIvarLayout, info); + outs() << " baseProperties " + << format("0x%" PRIx32, cro.baseProperties) << "\n"; + if (cro.baseProperties != 0) + print_objc_property_list32(cro.baseProperties, info); + is_meta_class = (cro.flags & RO_META) != 0; + return true; +} + +static void print_class64_t(uint64_t p, struct DisassembleInfo *info) { + struct class64_t c; + const char *r; + uint32_t offset, left; + SectionRef S; + const char *name; + uint64_t isa_n_value, n_value; + + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr || left < sizeof(struct class64_t)) + return; + memset(&c, '\0', sizeof(struct class64_t)); + if (left < sizeof(struct class64_t)) { + memcpy(&c, r, left); + outs() << " (class_t entends past the end of the section)\n"; + } else + memcpy(&c, r, sizeof(struct class64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(c); + + outs() << " isa " << format("0x%" PRIx64, c.isa); + name = get_symbol_64(offset + offsetof(struct class64_t, isa), S, info, + isa_n_value, c.isa); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + outs() << " superclass " << format("0x%" PRIx64, c.superclass); + name = get_symbol_64(offset + offsetof(struct class64_t, superclass), S, info, + n_value, c.superclass); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + outs() << " cache " << format("0x%" PRIx64, c.cache); + name = get_symbol_64(offset + offsetof(struct class64_t, cache), S, info, + n_value, c.cache); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + outs() << " vtable " << format("0x%" PRIx64, c.vtable); + name = get_symbol_64(offset + offsetof(struct class64_t, vtable), S, info, + n_value, c.vtable); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + name = get_symbol_64(offset + offsetof(struct class64_t, data), S, info, + n_value, c.data); + outs() << " data "; + if (n_value != 0) { + if (info->verbose && name != nullptr) + outs() << name; + else + outs() << format("0x%" PRIx64, n_value); + if (c.data != 0) + outs() << " + " << format("0x%" PRIx64, c.data); + } else + outs() << format("0x%" PRIx64, c.data); + outs() << " (struct class_ro_t *)"; + + // This is a Swift class if some of the low bits of the pointer are set. + if ((c.data + n_value) & 0x7) + outs() << " Swift class"; + outs() << "\n"; + bool is_meta_class; + if (!print_class_ro64_t((c.data + n_value) & ~0x7, info, is_meta_class)) + return; + + if (!is_meta_class && + c.isa + isa_n_value != p && + c.isa + isa_n_value != 0 && + info->depth < 100) { + info->depth++; + outs() << "Meta Class\n"; + print_class64_t(c.isa + isa_n_value, info); + } +} + +static void print_class32_t(uint32_t p, struct DisassembleInfo *info) { + struct class32_t c; + const char *r; + uint32_t offset, left; + SectionRef S; + const char *name; + + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&c, '\0', sizeof(struct class32_t)); + if (left < sizeof(struct class32_t)) { + memcpy(&c, r, left); + outs() << " (class_t entends past the end of the section)\n"; + } else + memcpy(&c, r, sizeof(struct class32_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(c); + + outs() << " isa " << format("0x%" PRIx32, c.isa); + name = + get_symbol_32(offset + offsetof(struct class32_t, isa), S, info, c.isa); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + outs() << " superclass " << format("0x%" PRIx32, c.superclass); + name = get_symbol_32(offset + offsetof(struct class32_t, superclass), S, info, + c.superclass); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + outs() << " cache " << format("0x%" PRIx32, c.cache); + name = get_symbol_32(offset + offsetof(struct class32_t, cache), S, info, + c.cache); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + outs() << " vtable " << format("0x%" PRIx32, c.vtable); + name = get_symbol_32(offset + offsetof(struct class32_t, vtable), S, info, + c.vtable); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + name = + get_symbol_32(offset + offsetof(struct class32_t, data), S, info, c.data); + outs() << " data " << format("0x%" PRIx32, c.data) + << " (struct class_ro_t *)"; + + // This is a Swift class if some of the low bits of the pointer are set. + if (c.data & 0x3) + outs() << " Swift class"; + outs() << "\n"; + bool is_meta_class; + if (!print_class_ro32_t(c.data & ~0x3, info, is_meta_class)) + return; + + if (!is_meta_class) { + outs() << "Meta Class\n"; + print_class32_t(c.isa, info); + } +} + +static void print_objc_class_t(struct objc_class_t *objc_class, + struct DisassembleInfo *info) { + uint32_t offset, left, xleft; + const char *name, *p, *ivar_list; + SectionRef S; + int32_t i; + struct objc_ivar_list_t objc_ivar_list; + struct objc_ivar_t ivar; + + outs() << "\t\t isa " << format("0x%08" PRIx32, objc_class->isa); + if (info->verbose && CLS_GETINFO(objc_class, CLS_META)) { + name = get_pointer_32(objc_class->isa, offset, left, S, info, true); + if (name != nullptr) + outs() << format(" %.*s", left, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + + outs() << "\t super_class " + << format("0x%08" PRIx32, objc_class->super_class); + if (info->verbose) { + name = get_pointer_32(objc_class->super_class, offset, left, S, info, true); + if (name != nullptr) + outs() << format(" %.*s", left, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + + outs() << "\t\t name " << format("0x%08" PRIx32, objc_class->name); + if (info->verbose) { + name = get_pointer_32(objc_class->name, offset, left, S, info, true); + if (name != nullptr) + outs() << format(" %.*s", left, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + + outs() << "\t\t version " << format("0x%08" PRIx32, objc_class->version) + << "\n"; + + outs() << "\t\t info " << format("0x%08" PRIx32, objc_class->info); + if (info->verbose) { + if (CLS_GETINFO(objc_class, CLS_CLASS)) + outs() << " CLS_CLASS"; + else if (CLS_GETINFO(objc_class, CLS_META)) + outs() << " CLS_META"; + } + outs() << "\n"; + + outs() << "\t instance_size " + << format("0x%08" PRIx32, objc_class->instance_size) << "\n"; + + p = get_pointer_32(objc_class->ivars, offset, left, S, info, true); + outs() << "\t\t ivars " << format("0x%08" PRIx32, objc_class->ivars); + if (p != nullptr) { + if (left > sizeof(struct objc_ivar_list_t)) { + outs() << "\n"; + memcpy(&objc_ivar_list, p, sizeof(struct objc_ivar_list_t)); + } else { + outs() << " (entends past the end of the section)\n"; + memset(&objc_ivar_list, '\0', sizeof(struct objc_ivar_list_t)); + memcpy(&objc_ivar_list, p, left); + } + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(objc_ivar_list); + outs() << "\t\t ivar_count " << objc_ivar_list.ivar_count << "\n"; + ivar_list = p + sizeof(struct objc_ivar_list_t); + for (i = 0; i < objc_ivar_list.ivar_count; i++) { + if ((i + 1) * sizeof(struct objc_ivar_t) > left) { + outs() << "\t\t remaining ivar's extend past the of the section\n"; + break; + } + memcpy(&ivar, ivar_list + i * sizeof(struct objc_ivar_t), + sizeof(struct objc_ivar_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(ivar); + + outs() << "\t\t\tivar_name " << format("0x%08" PRIx32, ivar.ivar_name); + if (info->verbose) { + name = get_pointer_32(ivar.ivar_name, offset, xleft, S, info, true); + if (name != nullptr) + outs() << format(" %.*s", xleft, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + + outs() << "\t\t\tivar_type " << format("0x%08" PRIx32, ivar.ivar_type); + if (info->verbose) { + name = get_pointer_32(ivar.ivar_type, offset, xleft, S, info, true); + if (name != nullptr) + outs() << format(" %.*s", xleft, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + + outs() << "\t\t ivar_offset " + << format("0x%08" PRIx32, ivar.ivar_offset) << "\n"; + } + } else { + outs() << " (not in an __OBJC section)\n"; + } + + outs() << "\t\t methods " << format("0x%08" PRIx32, objc_class->methodLists); + if (print_method_list(objc_class->methodLists, info)) + outs() << " (not in an __OBJC section)\n"; + + outs() << "\t\t cache " << format("0x%08" PRIx32, objc_class->cache) + << "\n"; + + outs() << "\t\tprotocols " << format("0x%08" PRIx32, objc_class->protocols); + if (print_protocol_list(objc_class->protocols, 16, info)) + outs() << " (not in an __OBJC section)\n"; +} + +static void print_objc_objc_category_t(struct objc_category_t *objc_category, + struct DisassembleInfo *info) { + uint32_t offset, left; + const char *name; + SectionRef S; + + outs() << "\t category name " + << format("0x%08" PRIx32, objc_category->category_name); + if (info->verbose) { + name = get_pointer_32(objc_category->category_name, offset, left, S, info, + true); + if (name != nullptr) + outs() << format(" %.*s", left, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + + outs() << "\t\t class name " + << format("0x%08" PRIx32, objc_category->class_name); + if (info->verbose) { + name = + get_pointer_32(objc_category->class_name, offset, left, S, info, true); + if (name != nullptr) + outs() << format(" %.*s", left, name); + else + outs() << " (not in an __OBJC section)"; + } + outs() << "\n"; + + outs() << "\t instance methods " + << format("0x%08" PRIx32, objc_category->instance_methods); + if (print_method_list(objc_category->instance_methods, info)) + outs() << " (not in an __OBJC section)\n"; + + outs() << "\t class methods " + << format("0x%08" PRIx32, objc_category->class_methods); + if (print_method_list(objc_category->class_methods, info)) + outs() << " (not in an __OBJC section)\n"; +} + +static void print_category64_t(uint64_t p, struct DisassembleInfo *info) { + struct category64_t c; + const char *r; + uint32_t offset, xoffset, left; + SectionRef S, xS; + const char *name, *sym_name; + uint64_t n_value; + + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&c, '\0', sizeof(struct category64_t)); + if (left < sizeof(struct category64_t)) { + memcpy(&c, r, left); + outs() << " (category_t entends past the end of the section)\n"; + } else + memcpy(&c, r, sizeof(struct category64_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(c); + + outs() << " name "; + sym_name = get_symbol_64(offset + offsetof(struct category64_t, name), S, + info, n_value, c.name); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (c.name != 0) + outs() << " + " << format("0x%" PRIx64, c.name); + } else + outs() << format("0x%" PRIx64, c.name); + name = get_pointer_64(c.name + n_value, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + outs() << " cls "; + sym_name = get_symbol_64(offset + offsetof(struct category64_t, cls), S, info, + n_value, c.cls); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (c.cls != 0) + outs() << " + " << format("0x%" PRIx64, c.cls); + } else + outs() << format("0x%" PRIx64, c.cls); + outs() << "\n"; + if (c.cls + n_value != 0) + print_class64_t(c.cls + n_value, info); + + outs() << " instanceMethods "; + sym_name = + get_symbol_64(offset + offsetof(struct category64_t, instanceMethods), S, + info, n_value, c.instanceMethods); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (c.instanceMethods != 0) + outs() << " + " << format("0x%" PRIx64, c.instanceMethods); + } else + outs() << format("0x%" PRIx64, c.instanceMethods); + outs() << "\n"; + if (c.instanceMethods + n_value != 0) + print_method_list64_t(c.instanceMethods + n_value, info, ""); + + outs() << " classMethods "; + sym_name = get_symbol_64(offset + offsetof(struct category64_t, classMethods), + S, info, n_value, c.classMethods); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (c.classMethods != 0) + outs() << " + " << format("0x%" PRIx64, c.classMethods); + } else + outs() << format("0x%" PRIx64, c.classMethods); + outs() << "\n"; + if (c.classMethods + n_value != 0) + print_method_list64_t(c.classMethods + n_value, info, ""); + + outs() << " protocols "; + sym_name = get_symbol_64(offset + offsetof(struct category64_t, protocols), S, + info, n_value, c.protocols); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (c.protocols != 0) + outs() << " + " << format("0x%" PRIx64, c.protocols); + } else + outs() << format("0x%" PRIx64, c.protocols); + outs() << "\n"; + if (c.protocols + n_value != 0) + print_protocol_list64_t(c.protocols + n_value, info); + + outs() << "instanceProperties "; + sym_name = + get_symbol_64(offset + offsetof(struct category64_t, instanceProperties), + S, info, n_value, c.instanceProperties); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (c.instanceProperties != 0) + outs() << " + " << format("0x%" PRIx64, c.instanceProperties); + } else + outs() << format("0x%" PRIx64, c.instanceProperties); + outs() << "\n"; + if (c.instanceProperties + n_value != 0) + print_objc_property_list64(c.instanceProperties + n_value, info); +} + +static void print_category32_t(uint32_t p, struct DisassembleInfo *info) { + struct category32_t c; + const char *r; + uint32_t offset, left; + SectionRef S, xS; + const char *name; + + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&c, '\0', sizeof(struct category32_t)); + if (left < sizeof(struct category32_t)) { + memcpy(&c, r, left); + outs() << " (category_t entends past the end of the section)\n"; + } else + memcpy(&c, r, sizeof(struct category32_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(c); + + outs() << " name " << format("0x%" PRIx32, c.name); + name = get_symbol_32(offset + offsetof(struct category32_t, name), S, info, + c.name); + if (name) + outs() << " " << name; + outs() << "\n"; + + outs() << " cls " << format("0x%" PRIx32, c.cls) << "\n"; + if (c.cls != 0) + print_class32_t(c.cls, info); + outs() << " instanceMethods " << format("0x%" PRIx32, c.instanceMethods) + << "\n"; + if (c.instanceMethods != 0) + print_method_list32_t(c.instanceMethods, info, ""); + outs() << " classMethods " << format("0x%" PRIx32, c.classMethods) + << "\n"; + if (c.classMethods != 0) + print_method_list32_t(c.classMethods, info, ""); + outs() << " protocols " << format("0x%" PRIx32, c.protocols) << "\n"; + if (c.protocols != 0) + print_protocol_list32_t(c.protocols, info); + outs() << "instanceProperties " << format("0x%" PRIx32, c.instanceProperties) + << "\n"; + if (c.instanceProperties != 0) + print_objc_property_list32(c.instanceProperties, info); +} + +static void print_message_refs64(SectionRef S, struct DisassembleInfo *info) { + uint32_t i, left, offset, xoffset; + uint64_t p, n_value; + struct message_ref64 mr; + const char *name, *sym_name; + const char *r; + SectionRef xS; + + if (S == SectionRef()) + return; + + StringRef SectName; + S.getName(SectName); + DataRefImpl Ref = S.getRawDataRefImpl(); + StringRef SegName = info->O->getSectionFinalSegmentName(Ref); + outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; + offset = 0; + for (i = 0; i < S.getSize(); i += sizeof(struct message_ref64)) { + p = S.getAddress() + i; + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&mr, '\0', sizeof(struct message_ref64)); + if (left < sizeof(struct message_ref64)) { + memcpy(&mr, r, left); + outs() << " (message_ref entends past the end of the section)\n"; + } else + memcpy(&mr, r, sizeof(struct message_ref64)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(mr); + + outs() << " imp "; + name = get_symbol_64(offset + offsetof(struct message_ref64, imp), S, info, + n_value, mr.imp); + if (n_value != 0) { + outs() << format("0x%" PRIx64, n_value) << " "; + if (mr.imp != 0) + outs() << "+ " << format("0x%" PRIx64, mr.imp) << " "; + } else + outs() << format("0x%" PRIx64, mr.imp) << " "; + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + outs() << " sel "; + sym_name = get_symbol_64(offset + offsetof(struct message_ref64, sel), S, + info, n_value, mr.sel); + if (n_value != 0) { + if (info->verbose && sym_name != nullptr) + outs() << sym_name; + else + outs() << format("0x%" PRIx64, n_value); + if (mr.sel != 0) + outs() << " + " << format("0x%" PRIx64, mr.sel); + } else + outs() << format("0x%" PRIx64, mr.sel); + name = get_pointer_64(mr.sel + n_value, xoffset, left, xS, info); + if (name != nullptr) + outs() << format(" %.*s", left, name); + outs() << "\n"; + + offset += sizeof(struct message_ref64); + } +} + +static void print_message_refs32(SectionRef S, struct DisassembleInfo *info) { + uint32_t i, left, offset, xoffset, p; + struct message_ref32 mr; + const char *name, *r; + SectionRef xS; + + if (S == SectionRef()) + return; + + StringRef SectName; + S.getName(SectName); + DataRefImpl Ref = S.getRawDataRefImpl(); + StringRef SegName = info->O->getSectionFinalSegmentName(Ref); + outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; + offset = 0; + for (i = 0; i < S.getSize(); i += sizeof(struct message_ref64)) { + p = S.getAddress() + i; + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&mr, '\0', sizeof(struct message_ref32)); + if (left < sizeof(struct message_ref32)) { + memcpy(&mr, r, left); + outs() << " (message_ref entends past the end of the section)\n"; + } else + memcpy(&mr, r, sizeof(struct message_ref32)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(mr); + + outs() << " imp " << format("0x%" PRIx32, mr.imp); + name = get_symbol_32(offset + offsetof(struct message_ref32, imp), S, info, + mr.imp); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + outs() << " sel " << format("0x%" PRIx32, mr.sel); + name = get_pointer_32(mr.sel, xoffset, left, xS, info); + if (name != nullptr) + outs() << " " << name; + outs() << "\n"; + + offset += sizeof(struct message_ref32); + } +} + +static void print_image_info64(SectionRef S, struct DisassembleInfo *info) { + uint32_t left, offset, swift_version; + uint64_t p; + struct objc_image_info64 o; + const char *r; + + if (S == SectionRef()) + return; + + StringRef SectName; + S.getName(SectName); + DataRefImpl Ref = S.getRawDataRefImpl(); + StringRef SegName = info->O->getSectionFinalSegmentName(Ref); + outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; + p = S.getAddress(); + r = get_pointer_64(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&o, '\0', sizeof(struct objc_image_info64)); + if (left < sizeof(struct objc_image_info64)) { + memcpy(&o, r, left); + outs() << " (objc_image_info entends past the end of the section)\n"; + } else + memcpy(&o, r, sizeof(struct objc_image_info64)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(o); + outs() << " version " << o.version << "\n"; + outs() << " flags " << format("0x%" PRIx32, o.flags); + if (o.flags & OBJC_IMAGE_IS_REPLACEMENT) + outs() << " OBJC_IMAGE_IS_REPLACEMENT"; + if (o.flags & OBJC_IMAGE_SUPPORTS_GC) + outs() << " OBJC_IMAGE_SUPPORTS_GC"; + swift_version = (o.flags >> 8) & 0xff; + if (swift_version != 0) { + if (swift_version == 1) + outs() << " Swift 1.0"; + else if (swift_version == 2) + outs() << " Swift 1.1"; + else + outs() << " unknown future Swift version (" << swift_version << ")"; + } + outs() << "\n"; +} + +static void print_image_info32(SectionRef S, struct DisassembleInfo *info) { + uint32_t left, offset, swift_version, p; + struct objc_image_info32 o; + const char *r; + + StringRef SectName; + S.getName(SectName); + DataRefImpl Ref = S.getRawDataRefImpl(); + StringRef SegName = info->O->getSectionFinalSegmentName(Ref); + outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; + p = S.getAddress(); + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&o, '\0', sizeof(struct objc_image_info32)); + if (left < sizeof(struct objc_image_info32)) { + memcpy(&o, r, left); + outs() << " (objc_image_info entends past the end of the section)\n"; + } else + memcpy(&o, r, sizeof(struct objc_image_info32)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(o); + outs() << " version " << o.version << "\n"; + outs() << " flags " << format("0x%" PRIx32, o.flags); + if (o.flags & OBJC_IMAGE_IS_REPLACEMENT) + outs() << " OBJC_IMAGE_IS_REPLACEMENT"; + if (o.flags & OBJC_IMAGE_SUPPORTS_GC) + outs() << " OBJC_IMAGE_SUPPORTS_GC"; + swift_version = (o.flags >> 8) & 0xff; + if (swift_version != 0) { + if (swift_version == 1) + outs() << " Swift 1.0"; + else if (swift_version == 2) + outs() << " Swift 1.1"; + else + outs() << " unknown future Swift version (" << swift_version << ")"; + } + outs() << "\n"; +} + +static void print_image_info(SectionRef S, struct DisassembleInfo *info) { + uint32_t left, offset, p; + struct imageInfo_t o; + const char *r; + + StringRef SectName; + S.getName(SectName); + DataRefImpl Ref = S.getRawDataRefImpl(); + StringRef SegName = info->O->getSectionFinalSegmentName(Ref); + outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; + p = S.getAddress(); + r = get_pointer_32(p, offset, left, S, info); + if (r == nullptr) + return; + memset(&o, '\0', sizeof(struct imageInfo_t)); + if (left < sizeof(struct imageInfo_t)) { + memcpy(&o, r, left); + outs() << " (imageInfo entends past the end of the section)\n"; + } else + memcpy(&o, r, sizeof(struct imageInfo_t)); + if (info->O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(o); + outs() << " version " << o.version << "\n"; + outs() << " flags " << format("0x%" PRIx32, o.flags); + if (o.flags & 0x1) + outs() << " F&C"; + if (o.flags & 0x2) + outs() << " GC"; + if (o.flags & 0x4) + outs() << " GC-only"; + else + outs() << " RR"; + outs() << "\n"; +} + +static void printObjc2_64bit_MetaData(MachOObjectFile *O, bool verbose) { + SymbolAddressMap AddrMap; + if (verbose) + CreateSymbolAddressMap(O, &AddrMap); + + std::vector Sections; + for (const SectionRef &Section : O->sections()) { + StringRef SectName; + Section.getName(SectName); + Sections.push_back(Section); + } + + struct DisassembleInfo info; + // Set up the block of info used by the Symbolizer call backs. + info.verbose = verbose; + info.O = O; + info.AddrMap = &AddrMap; + info.Sections = &Sections; + info.class_name = nullptr; + info.selector_name = nullptr; + info.method = nullptr; + info.demangled_name = nullptr; + info.bindtable = nullptr; + info.adrp_addr = 0; + info.adrp_inst = 0; + + info.depth = 0; + SectionRef CL = get_section(O, "__OBJC2", "__class_list"); + if (CL == SectionRef()) + CL = get_section(O, "__DATA", "__objc_classlist"); + info.S = CL; + walk_pointer_list_64("class", CL, O, &info, print_class64_t); + + SectionRef CR = get_section(O, "__OBJC2", "__class_refs"); + if (CR == SectionRef()) + CR = get_section(O, "__DATA", "__objc_classrefs"); + info.S = CR; + walk_pointer_list_64("class refs", CR, O, &info, nullptr); + + SectionRef SR = get_section(O, "__OBJC2", "__super_refs"); + if (SR == SectionRef()) + SR = get_section(O, "__DATA", "__objc_superrefs"); + info.S = SR; + walk_pointer_list_64("super refs", SR, O, &info, nullptr); + + SectionRef CA = get_section(O, "__OBJC2", "__category_list"); + if (CA == SectionRef()) + CA = get_section(O, "__DATA", "__objc_catlist"); + info.S = CA; + walk_pointer_list_64("category", CA, O, &info, print_category64_t); + + SectionRef PL = get_section(O, "__OBJC2", "__protocol_list"); + if (PL == SectionRef()) + PL = get_section(O, "__DATA", "__objc_protolist"); + info.S = PL; + walk_pointer_list_64("protocol", PL, O, &info, nullptr); + + SectionRef MR = get_section(O, "__OBJC2", "__message_refs"); + if (MR == SectionRef()) + MR = get_section(O, "__DATA", "__objc_msgrefs"); + info.S = MR; + print_message_refs64(MR, &info); + + SectionRef II = get_section(O, "__OBJC2", "__image_info"); + if (II == SectionRef()) + II = get_section(O, "__DATA", "__objc_imageinfo"); + info.S = II; + print_image_info64(II, &info); + + if (info.bindtable != nullptr) + delete info.bindtable; +} + +static void printObjc2_32bit_MetaData(MachOObjectFile *O, bool verbose) { + SymbolAddressMap AddrMap; + if (verbose) + CreateSymbolAddressMap(O, &AddrMap); + + std::vector Sections; + for (const SectionRef &Section : O->sections()) { + StringRef SectName; + Section.getName(SectName); + Sections.push_back(Section); + } + + struct DisassembleInfo info; + // Set up the block of info used by the Symbolizer call backs. + info.verbose = verbose; + info.O = O; + info.AddrMap = &AddrMap; + info.Sections = &Sections; + info.class_name = nullptr; + info.selector_name = nullptr; + info.method = nullptr; + info.demangled_name = nullptr; + info.bindtable = nullptr; + info.adrp_addr = 0; + info.adrp_inst = 0; + + const SectionRef CL = get_section(O, "__OBJC2", "__class_list"); + if (CL != SectionRef()) { + info.S = CL; + walk_pointer_list_32("class", CL, O, &info, print_class32_t); + } else { + const SectionRef CL = get_section(O, "__DATA", "__objc_classlist"); + info.S = CL; + walk_pointer_list_32("class", CL, O, &info, print_class32_t); + } + + const SectionRef CR = get_section(O, "__OBJC2", "__class_refs"); + if (CR != SectionRef()) { + info.S = CR; + walk_pointer_list_32("class refs", CR, O, &info, nullptr); + } else { + const SectionRef CR = get_section(O, "__DATA", "__objc_classrefs"); + info.S = CR; + walk_pointer_list_32("class refs", CR, O, &info, nullptr); + } + + const SectionRef SR = get_section(O, "__OBJC2", "__super_refs"); + if (SR != SectionRef()) { + info.S = SR; + walk_pointer_list_32("super refs", SR, O, &info, nullptr); + } else { + const SectionRef SR = get_section(O, "__DATA", "__objc_superrefs"); + info.S = SR; + walk_pointer_list_32("super refs", SR, O, &info, nullptr); + } + + const SectionRef CA = get_section(O, "__OBJC2", "__category_list"); + if (CA != SectionRef()) { + info.S = CA; + walk_pointer_list_32("category", CA, O, &info, print_category32_t); + } else { + const SectionRef CA = get_section(O, "__DATA", "__objc_catlist"); + info.S = CA; + walk_pointer_list_32("category", CA, O, &info, print_category32_t); + } + + const SectionRef PL = get_section(O, "__OBJC2", "__protocol_list"); + if (PL != SectionRef()) { + info.S = PL; + walk_pointer_list_32("protocol", PL, O, &info, nullptr); + } else { + const SectionRef PL = get_section(O, "__DATA", "__objc_protolist"); + info.S = PL; + walk_pointer_list_32("protocol", PL, O, &info, nullptr); + } + + const SectionRef MR = get_section(O, "__OBJC2", "__message_refs"); + if (MR != SectionRef()) { + info.S = MR; + print_message_refs32(MR, &info); + } else { + const SectionRef MR = get_section(O, "__DATA", "__objc_msgrefs"); + info.S = MR; + print_message_refs32(MR, &info); + } + + const SectionRef II = get_section(O, "__OBJC2", "__image_info"); + if (II != SectionRef()) { + info.S = II; + print_image_info32(II, &info); + } else { + const SectionRef II = get_section(O, "__DATA", "__objc_imageinfo"); + info.S = II; + print_image_info32(II, &info); + } +} + +static bool printObjc1_32bit_MetaData(MachOObjectFile *O, bool verbose) { + uint32_t i, j, p, offset, xoffset, left, defs_left, def; + const char *r, *name, *defs; + struct objc_module_t module; + SectionRef S, xS; + struct objc_symtab_t symtab; + struct objc_class_t objc_class; + struct objc_category_t objc_category; + + outs() << "Objective-C segment\n"; + S = get_section(O, "__OBJC", "__module_info"); + if (S == SectionRef()) + return false; + + SymbolAddressMap AddrMap; + if (verbose) + CreateSymbolAddressMap(O, &AddrMap); + + std::vector Sections; + for (const SectionRef &Section : O->sections()) { + StringRef SectName; + Section.getName(SectName); + Sections.push_back(Section); + } + + struct DisassembleInfo info; + // Set up the block of info used by the Symbolizer call backs. + info.verbose = verbose; + info.O = O; + info.AddrMap = &AddrMap; + info.Sections = &Sections; + info.class_name = nullptr; + info.selector_name = nullptr; + info.method = nullptr; + info.demangled_name = nullptr; + info.bindtable = nullptr; + info.adrp_addr = 0; + info.adrp_inst = 0; + + for (i = 0; i < S.getSize(); i += sizeof(struct objc_module_t)) { + p = S.getAddress() + i; + r = get_pointer_32(p, offset, left, S, &info, true); + if (r == nullptr) + return true; + memset(&module, '\0', sizeof(struct objc_module_t)); + if (left < sizeof(struct objc_module_t)) { + memcpy(&module, r, left); + outs() << " (module extends past end of __module_info section)\n"; + } else + memcpy(&module, r, sizeof(struct objc_module_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(module); + + outs() << "Module " << format("0x%" PRIx32, p) << "\n"; + outs() << " version " << module.version << "\n"; + outs() << " size " << module.size << "\n"; + outs() << " name "; + name = get_pointer_32(module.name, xoffset, left, xS, &info, true); + if (name != nullptr) + outs() << format("%.*s", left, name); + else + outs() << format("0x%08" PRIx32, module.name) + << "(not in an __OBJC section)"; + outs() << "\n"; + + r = get_pointer_32(module.symtab, xoffset, left, xS, &info, true); + if (module.symtab == 0 || r == nullptr) { + outs() << " symtab " << format("0x%08" PRIx32, module.symtab) + << " (not in an __OBJC section)\n"; + continue; + } + outs() << " symtab " << format("0x%08" PRIx32, module.symtab) << "\n"; + memset(&symtab, '\0', sizeof(struct objc_symtab_t)); + defs_left = 0; + defs = nullptr; + if (left < sizeof(struct objc_symtab_t)) { + memcpy(&symtab, r, left); + outs() << "\tsymtab extends past end of an __OBJC section)\n"; + } else { + memcpy(&symtab, r, sizeof(struct objc_symtab_t)); + if (left > sizeof(struct objc_symtab_t)) { + defs_left = left - sizeof(struct objc_symtab_t); + defs = r + sizeof(struct objc_symtab_t); + } + } + if (O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(symtab); + + outs() << "\tsel_ref_cnt " << symtab.sel_ref_cnt << "\n"; + r = get_pointer_32(symtab.refs, xoffset, left, xS, &info, true); + outs() << "\trefs " << format("0x%08" PRIx32, symtab.refs); + if (r == nullptr) + outs() << " (not in an __OBJC section)"; + outs() << "\n"; + outs() << "\tcls_def_cnt " << symtab.cls_def_cnt << "\n"; + outs() << "\tcat_def_cnt " << symtab.cat_def_cnt << "\n"; + if (symtab.cls_def_cnt > 0) + outs() << "\tClass Definitions\n"; + for (j = 0; j < symtab.cls_def_cnt; j++) { + if ((j + 1) * sizeof(uint32_t) > defs_left) { + outs() << "\t(remaining class defs entries entends past the end of the " + << "section)\n"; + break; + } + memcpy(&def, defs + j * sizeof(uint32_t), sizeof(uint32_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(def); + + r = get_pointer_32(def, xoffset, left, xS, &info, true); + outs() << "\tdefs[" << j << "] " << format("0x%08" PRIx32, def); + if (r != nullptr) { + if (left > sizeof(struct objc_class_t)) { + outs() << "\n"; + memcpy(&objc_class, r, sizeof(struct objc_class_t)); + } else { + outs() << " (entends past the end of the section)\n"; + memset(&objc_class, '\0', sizeof(struct objc_class_t)); + memcpy(&objc_class, r, left); + } + if (O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(objc_class); + print_objc_class_t(&objc_class, &info); + } else { + outs() << "(not in an __OBJC section)\n"; + } + + if (CLS_GETINFO(&objc_class, CLS_CLASS)) { + outs() << "\tMeta Class"; + r = get_pointer_32(objc_class.isa, xoffset, left, xS, &info, true); + if (r != nullptr) { + if (left > sizeof(struct objc_class_t)) { + outs() << "\n"; + memcpy(&objc_class, r, sizeof(struct objc_class_t)); + } else { + outs() << " (entends past the end of the section)\n"; + memset(&objc_class, '\0', sizeof(struct objc_class_t)); + memcpy(&objc_class, r, left); + } + if (O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(objc_class); + print_objc_class_t(&objc_class, &info); + } else { + outs() << "(not in an __OBJC section)\n"; + } + } + } + if (symtab.cat_def_cnt > 0) + outs() << "\tCategory Definitions\n"; + for (j = 0; j < symtab.cat_def_cnt; j++) { + if ((j + symtab.cls_def_cnt + 1) * sizeof(uint32_t) > defs_left) { + outs() << "\t(remaining category defs entries entends past the end of " + << "the section)\n"; + break; + } + memcpy(&def, defs + (j + symtab.cls_def_cnt) * sizeof(uint32_t), + sizeof(uint32_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) + sys::swapByteOrder(def); + + r = get_pointer_32(def, xoffset, left, xS, &info, true); + outs() << "\tdefs[" << j + symtab.cls_def_cnt << "] " + << format("0x%08" PRIx32, def); + if (r != nullptr) { + if (left > sizeof(struct objc_category_t)) { + outs() << "\n"; + memcpy(&objc_category, r, sizeof(struct objc_category_t)); + } else { + outs() << " (entends past the end of the section)\n"; + memset(&objc_category, '\0', sizeof(struct objc_category_t)); + memcpy(&objc_category, r, left); + } + if (O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(objc_category); + print_objc_objc_category_t(&objc_category, &info); + } else { + outs() << "(not in an __OBJC section)\n"; + } + } } + const SectionRef II = get_section(O, "__OBJC", "__image_info"); + if (II != SectionRef()) + print_image_info(II, &info); - // The case were the pointer_value is non-zero and points to a class defined - // in this Mach-O file. - r = get_pointer_64(pointer_value, offset, left, S, info); - if (r == nullptr || left < sizeof(struct class64_t)) - return nullptr; - struct class64_t c; - memcpy(&c, r, sizeof(struct class64_t)); - if (info->O->isLittleEndian() != sys::IsLittleEndianHost) - swapStruct(c); - if (c.data == 0) - return nullptr; - r = get_pointer_64(c.data, offset, left, S, info); - if (r == nullptr || left < sizeof(struct class_ro64_t)) - return nullptr; - struct class_ro64_t cro; - memcpy(&cro, r, sizeof(struct class_ro64_t)); - if (info->O->isLittleEndian() != sys::IsLittleEndianHost) - swapStruct(cro); - if (cro.name == 0) - return nullptr; - const char *name = get_pointer_64(cro.name, offset, left, S, info); - return name; + return true; } -// get_objc2_64bit_cfstring_name is used for disassembly and is passed a -// pointer to a cfstring and returns its name or nullptr. -const char *get_objc2_64bit_cfstring_name(uint64_t ReferenceValue, - struct DisassembleInfo *info) { - const char *r, *name; - uint32_t offset, left; - SectionRef S; - struct cfstring64_t cfs; - uint64_t cfs_characters; +static void DumpProtocolSection(MachOObjectFile *O, const char *sect, + uint32_t size, uint32_t addr) { + SymbolAddressMap AddrMap; + CreateSymbolAddressMap(O, &AddrMap); - r = get_pointer_64(ReferenceValue, offset, left, S, info); - if (r == nullptr || left < sizeof(struct cfstring64_t)) - return nullptr; - memcpy(&cfs, r, sizeof(struct cfstring64_t)); - if (info->O->isLittleEndian() != sys::IsLittleEndianHost) - swapStruct(cfs); - if (cfs.characters == 0) { - uint64_t n_value; - const char *symbol_name = get_symbol_64( - offset + offsetof(struct cfstring64_t, characters), S, info, n_value); - if (symbol_name == nullptr) - return nullptr; - cfs_characters = n_value; - } else - cfs_characters = cfs.characters; - name = get_pointer_64(cfs_characters, offset, left, S, info); + std::vector Sections; + for (const SectionRef &Section : O->sections()) { + StringRef SectName; + Section.getName(SectName); + Sections.push_back(Section); + } - return name; + struct DisassembleInfo info; + // Set up the block of info used by the Symbolizer call backs. + info.verbose = true; + info.O = O; + info.AddrMap = &AddrMap; + info.Sections = &Sections; + info.class_name = nullptr; + info.selector_name = nullptr; + info.method = nullptr; + info.demangled_name = nullptr; + info.bindtable = nullptr; + info.adrp_addr = 0; + info.adrp_inst = 0; + + const char *p; + struct objc_protocol_t protocol; + uint32_t left, paddr; + for (p = sect; p < sect + size; p += sizeof(struct objc_protocol_t)) { + memset(&protocol, '\0', sizeof(struct objc_protocol_t)); + left = size - (p - sect); + if (left < sizeof(struct objc_protocol_t)) { + outs() << "Protocol extends past end of __protocol section\n"; + memcpy(&protocol, p, left); + } else + memcpy(&protocol, p, sizeof(struct objc_protocol_t)); + if (O->isLittleEndian() != sys::IsLittleEndianHost) + swapStruct(protocol); + paddr = addr + (p - sect); + outs() << "Protocol " << format("0x%" PRIx32, paddr); + if (print_protocol(paddr, 0, &info)) + outs() << "(not in an __OBJC section)\n"; + } } -// get_objc2_64bit_selref() is used for disassembly and is passed a the address -// of a pointer to an Objective-C selector reference when the pointer value is -// zero as in a .o file and is likely to have a external relocation entry with -// who's symbol's n_value is the real pointer to the selector name. If that is -// the case the real pointer to the selector name is returned else 0 is -// returned -uint64_t get_objc2_64bit_selref(uint64_t ReferenceValue, - struct DisassembleInfo *info) { - uint32_t offset, left; - SectionRef S; - - const char *r = get_pointer_64(ReferenceValue, offset, left, S, info); - if (r == nullptr || left < sizeof(uint64_t)) - return 0; - uint64_t n_value; - const char *symbol_name = get_symbol_64(offset, S, info, n_value); - if (symbol_name == nullptr) - return 0; - return n_value; +static void printObjcMetaData(MachOObjectFile *O, bool verbose) { + if (O->is64Bit()) + printObjc2_64bit_MetaData(O, verbose); + else { + MachO::mach_header H; + H = O->getHeader(); + if (H.cputype == MachO::CPU_TYPE_ARM) + printObjc2_32bit_MetaData(O, verbose); + else { + // This is the 32-bit non-arm cputype case. Which is normally + // the first Objective-C ABI. But it may be the case of a + // binary for the iOS simulator which is the second Objective-C + // ABI. In that case printObjc1_32bit_MetaData() will determine that + // and return false. + if (!printObjc1_32bit_MetaData(O, verbose)) + printObjc2_32bit_MetaData(O, verbose); + } + } } // GuessLiteralPointer returns a string which for the item in the Mach-O file @@ -2235,41 +5565,43 @@ uint64_t get_objc2_64bit_selref(uint64_t ReferenceValue, // // If there is no item in the Mach-O file for the address passed in as // ReferenceValue nullptr is returned and ReferenceType is unchanged. -const char *GuessLiteralPointer(uint64_t ReferenceValue, uint64_t ReferencePC, - uint64_t *ReferenceType, - struct DisassembleInfo *info) { +static const char *GuessLiteralPointer(uint64_t ReferenceValue, + uint64_t ReferencePC, + uint64_t *ReferenceType, + struct DisassembleInfo *info) { // First see if there is an external relocation entry at the ReferencePC. - uint64_t sect_addr = info->S.getAddress(); - uint64_t sect_offset = ReferencePC - sect_addr; - bool reloc_found = false; - DataRefImpl Rel; - MachO::any_relocation_info RE; - bool isExtern = false; - SymbolRef Symbol; - for (const RelocationRef &Reloc : info->S.relocations()) { - uint64_t RelocOffset; - Reloc.getOffset(RelocOffset); - if (RelocOffset == sect_offset) { - Rel = Reloc.getRawDataRefImpl(); - RE = info->O->getRelocation(Rel); - if (info->O->isRelocationScattered(RE)) - continue; - isExtern = info->O->getPlainRelocationExternal(RE); - if (isExtern) { - symbol_iterator RelocSym = Reloc.getSymbol(); - Symbol = *RelocSym; + if (info->O->getHeader().filetype == MachO::MH_OBJECT) { + uint64_t sect_addr = info->S.getAddress(); + uint64_t sect_offset = ReferencePC - sect_addr; + bool reloc_found = false; + DataRefImpl Rel; + MachO::any_relocation_info RE; + bool isExtern = false; + SymbolRef Symbol; + for (const RelocationRef &Reloc : info->S.relocations()) { + uint64_t RelocOffset = Reloc.getOffset(); + if (RelocOffset == sect_offset) { + Rel = Reloc.getRawDataRefImpl(); + RE = info->O->getRelocation(Rel); + if (info->O->isRelocationScattered(RE)) + continue; + isExtern = info->O->getPlainRelocationExternal(RE); + if (isExtern) { + symbol_iterator RelocSym = Reloc.getSymbol(); + Symbol = *RelocSym; + } + reloc_found = true; + break; } - reloc_found = true; - break; } - } - // If there is an external relocation entry for a symbol in a section - // then used that symbol's value for the value of the reference. - if (reloc_found && isExtern) { - if (info->O->getAnyRelocationPCRel(RE)) { - unsigned Type = info->O->getAnyRelocationType(RE); - if (Type == MachO::X86_64_RELOC_SIGNED) { - Symbol.getAddress(ReferenceValue); + // If there is an external relocation entry for a symbol in a section + // then used that symbol's value for the value of the reference. + if (reloc_found && isExtern) { + if (info->O->getAnyRelocationPCRel(RE)) { + unsigned Type = info->O->getAnyRelocationType(RE); + if (Type == MachO::X86_64_RELOC_SIGNED) { + ReferenceValue = Symbol.getValue(); + } } } } @@ -2279,7 +5611,7 @@ const char *GuessLiteralPointer(uint64_t ReferenceValue, uint64_t ReferencePC, bool classref, selref, msgref, cfstring; uint64_t pointer_value = GuessPointerPointer(ReferenceValue, info, classref, selref, msgref, cfstring); - if (classref == true && pointer_value == 0) { + if (classref && pointer_value == 0) { // Note the ReferenceValue is a pointer into the __objc_classrefs section. // And the pointer_value in that section is typically zero as it will be // set by dyld as part of the "bind information". @@ -2295,7 +5627,7 @@ const char *GuessLiteralPointer(uint64_t ReferenceValue, uint64_t ReferencePC, } } - if (classref == true) { + if (classref) { *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref; const char *name = get_objc2_64bit_class_name(pointer_value, ReferenceValue, info); @@ -2306,13 +5638,13 @@ const char *GuessLiteralPointer(uint64_t ReferenceValue, uint64_t ReferencePC, return name; } - if (cfstring == true) { + if (cfstring) { *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_CFString_Ref; const char *name = get_objc2_64bit_cfstring_name(ReferenceValue, info); return name; } - if (selref == true && pointer_value == 0) + if (selref && pointer_value == 0) pointer_value = get_objc2_64bit_selref(ReferenceValue, info); if (pointer_value != 0) @@ -2320,10 +5652,10 @@ const char *GuessLiteralPointer(uint64_t ReferenceValue, uint64_t ReferencePC, const char *name = GuessCstringPointer(ReferenceValue, info); if (name) { - if (pointer_value != 0 && selref == true) { + if (pointer_value != 0 && selref) { *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Selector_Ref; info->selector_name = name; - } else if (pointer_value != 0 && msgref == true) { + } else if (pointer_value != 0 && msgref) { info->class_name = nullptr; *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message_Ref; info->selector_name = name; @@ -2371,13 +5703,14 @@ const char *GuessLiteralPointer(uint64_t ReferenceValue, uint64_t ReferencePC, // SymbolValue is checked to be an address of literal pointer, symbol pointer, // or an Objective-C meta data reference. If so the output ReferenceType is // set to correspond to that as well as setting the ReferenceName. -const char *SymbolizerSymbolLookUp(void *DisInfo, uint64_t ReferenceValue, - uint64_t *ReferenceType, - uint64_t ReferencePC, - const char **ReferenceName) { +static const char *SymbolizerSymbolLookUp(void *DisInfo, + uint64_t ReferenceValue, + uint64_t *ReferenceType, + uint64_t ReferencePC, + const char **ReferenceName) { struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo; // If no verbose symbolic information is wanted then just return nullptr. - if (info->verbose == false) { + if (!info->verbose) { *ReferenceName = nullptr; *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None; return nullptr; @@ -2523,7 +5856,6 @@ static void emitComments(raw_svector_ostream &CommentStream, formatted_raw_ostream &FormattedOS, const MCAsmInfo &MAI) { // Flush the stream before taking its content. - CommentStream.flush(); StringRef Comments = CommentsToEmit.str(); // Get the default information for printing a comment. const char *CommentBegin = MAI.getCommentString(); @@ -2544,10 +5876,10 @@ static void emitComments(raw_svector_ostream &CommentStream, // Tell the comment stream that the vector changed underneath it. CommentsToEmit.clear(); - CommentStream.resync(); } -static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { +static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF, + StringRef DisSegName, StringRef DisSectName) { const char *McpuDefault = nullptr; const Target *ThumbTarget = nullptr; const Target *TheTarget = GetTarget(MachOOF, &McpuDefault, &ThumbTarget); @@ -2594,7 +5926,7 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { } int AsmPrinterVariant = AsmInfo->getAssemblerDialect(); std::unique_ptr IP(TheTarget->createMCInstPrinter( - AsmPrinterVariant, *AsmInfo, *InstrInfo, *MRI, *STI)); + Triple(TripleName), AsmPrinterVariant, *AsmInfo, *InstrInfo, *MRI)); // Set the display preference for hex vs. decimal immediates. IP->setPrintImmHex(PrintImmHex); // Comment stream and backing vector. @@ -2642,8 +5974,8 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { } int ThumbAsmPrinterVariant = ThumbAsmInfo->getAssemblerDialect(); ThumbIP.reset(ThumbTarget->createMCInstPrinter( - ThumbAsmPrinterVariant, *ThumbAsmInfo, *ThumbInstrInfo, *ThumbMRI, - *ThumbSTI)); + Triple(ThumbTripleName), ThumbAsmPrinterVariant, *ThumbAsmInfo, + *ThumbInstrInfo, *ThumbMRI)); // Set the display preference for hex vs. decimal immediates. ThumbIP->setPrintImmHex(PrintImmHex); } @@ -2665,7 +5997,7 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { SmallVector FoundFns; uint64_t BaseSegmentAddress; - getSectionsAndSymbols(Header, MachOOF, Sections, Symbols, FoundFns, + getSectionsAndSymbols(MachOOF, Sections, Symbols, FoundFns, BaseSegmentAddress); // Sort the symbols by address, just in case they didn't come in that way. @@ -2712,31 +6044,21 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { } // Setup the DIContext - diContext.reset(DIContext::getDWARFContext(*DbgObj)); + diContext.reset(new DWARFContextInMemory(*DbgObj)); } - // TODO: For now this only disassembles the (__TEXT,__text) section (see the - // checks in the code below at the top of this loop). It should allow a - // darwin otool(1) like -s option to disassemble any named segment & section - // that is marked as containing instructions with the attributes - // S_ATTR_PURE_INSTRUCTIONS or S_ATTR_SOME_INSTRUCTIONS in the flags field of - // the section structure. - outs() << "(__TEXT,__text) section\n"; + if (FilterSections.size() == 0) + outs() << "(" << DisSegName << "," << DisSectName << ") section\n"; for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) { - - bool SectIsText = Sections[SectIdx].isText(); - if (SectIsText == false) - continue; - StringRef SectName; - if (Sections[SectIdx].getName(SectName) || SectName != "__text") - continue; // Skip non-text sections + if (Sections[SectIdx].getName(SectName) || SectName != DisSectName) + continue; DataRefImpl DR = Sections[SectIdx].getRawDataRefImpl(); StringRef SegmentName = MachOOF->getSectionFinalSegmentName(DR); - if (SegmentName != "__TEXT") + if (SegmentName != DisSegName) continue; StringRef BytesStr; @@ -2747,37 +6069,30 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { bool symbolTableWorked = false; - // Parse relocations. - std::vector> Relocs; - for (const RelocationRef &Reloc : Sections[SectIdx].relocations()) { - uint64_t RelocOffset; - Reloc.getOffset(RelocOffset); - uint64_t SectionAddress = Sections[SectIdx].getAddress(); - RelocOffset -= SectionAddress; - - symbol_iterator RelocSym = Reloc.getSymbol(); - - Relocs.push_back(std::make_pair(RelocOffset, *RelocSym)); - } - array_pod_sort(Relocs.begin(), Relocs.end()); - // Create a map of symbol addresses to symbol names for use by // the SymbolizerSymbolLookUp() routine. SymbolAddressMap AddrMap; + bool DisSymNameFound = false; for (const SymbolRef &Symbol : MachOOF->symbols()) { - SymbolRef::Type ST; - Symbol.getType(ST); + SymbolRef::Type ST = Symbol.getType(); if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data || ST == SymbolRef::ST_Other) { - uint64_t Address; - Symbol.getAddress(Address); - StringRef SymName; - Symbol.getName(SymName); + uint64_t Address = Symbol.getValue(); + ErrorOr SymNameOrErr = Symbol.getName(); + if (std::error_code EC = SymNameOrErr.getError()) + report_fatal_error(EC.message()); + StringRef SymName = *SymNameOrErr; AddrMap[Address] = SymName; + if (!DisSymName.empty() && DisSymName == SymName) + DisSymNameFound = true; } } + if (!DisSymName.empty() && !DisSymNameFound) { + outs() << "Can't find -dis-symname: " << DisSymName << "\n"; + return; + } // Set up the block of info used by the Symbolizer call backs. - SymbolizerInfo.verbose = true; + SymbolizerInfo.verbose = !NoSymbolicOperands; SymbolizerInfo.O = MachOOF; SymbolizerInfo.S = Sections[SectIdx]; SymbolizerInfo.AddrMap = &AddrMap; @@ -2790,7 +6105,7 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { SymbolizerInfo.adrp_addr = 0; SymbolizerInfo.adrp_inst = 0; // Same for the ThumbSymbolizer - ThumbSymbolizerInfo.verbose = true; + ThumbSymbolizerInfo.verbose = !NoSymbolicOperands; ThumbSymbolizerInfo.O = MachOOF; ThumbSymbolizerInfo.S = Sections[SectIdx]; ThumbSymbolizerInfo.AddrMap = &AddrMap; @@ -2805,12 +6120,13 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { // Disassemble symbol by symbol. for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) { - StringRef SymName; - Symbols[SymIdx].getName(SymName); + ErrorOr SymNameOrErr = Symbols[SymIdx].getName(); + if (std::error_code EC = SymNameOrErr.getError()) + report_fatal_error(EC.message()); + StringRef SymName = *SymNameOrErr; - SymbolRef::Type ST; - Symbols[SymIdx].getType(ST); - if (ST != SymbolRef::ST_Function) + SymbolRef::Type ST = Symbols[SymIdx].getType(); + if (ST != SymbolRef::ST_Function && ST != SymbolRef::ST_Data) continue; // Make sure the symbol is defined in this section. @@ -2818,10 +6134,13 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { if (!containsSym) continue; + // If we are only disassembling one symbol see if this is that symbol. + if (!DisSymName.empty() && DisSymName != SymName) + continue; + // Start at the address of the symbol relative to the section's address. - uint64_t Start = 0; + uint64_t Start = Symbols[SymIdx].getValue(); uint64_t SectionAddress = Sections[SectIdx].getAddress(); - Symbols[SymIdx].getAddress(Start); Start -= SectionAddress; // Stop disassembling either at the beginning of the next symbol or at @@ -2830,12 +6149,11 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { uint64_t NextSym = 0; uint64_t NextSymIdx = SymIdx + 1; while (Symbols.size() > NextSymIdx) { - SymbolRef::Type NextSymType; - Symbols[NextSymIdx].getType(NextSymType); + SymbolRef::Type NextSymType = Symbols[NextSymIdx].getType(); if (NextSymType == SymbolRef::ST_Function) { containsNextSym = Sections[SectIdx].containsSymbol(Symbols[NextSymIdx]); - Symbols[NextSymIdx].getAddress(NextSym); + NextSym = Symbols[NextSymIdx].getValue(); NextSym -= SectionAddress; break; } @@ -2858,13 +6176,15 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { MCInst Inst; uint64_t PC = SectAddress + Index; - if (FullLeadingAddr) { - if (MachOOF->is64Bit()) - outs() << format("%016" PRIx64, PC); - else - outs() << format("%08" PRIx64, PC); - } else { - outs() << format("%8" PRIx64 ":", PC); + if (!NoLeadingAddr) { + if (FullLeadingAddr) { + if (MachOOF->is64Bit()) + outs() << format("%016" PRIx64, PC); + else + outs() << format("%08" PRIx64, PC); + } else { + outs() << format("%8" PRIx64 ":", PC); + } } if (!NoShowRawInsn) outs() << "\t"; @@ -2881,9 +6201,7 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { DTI->second.getLength(Length); uint16_t Kind; DTI->second.getKind(Kind); - Size = DumpDataInCode(reinterpret_cast(Bytes.data()) + - Index, - Length, Kind); + Size = DumpDataInCode(Bytes.data() + Index, Length, Kind); if ((Kind == MachO::DICE_KIND_JUMP_TABLE8) && (PC == (DTI->first + Length - 1)) && (Length & 1)) Size++; @@ -2902,16 +6220,14 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { DebugOut, Annotations); if (gotInst) { if (!NoShowRawInsn) { - DumpBytes(StringRef( - reinterpret_cast(Bytes.data()) + Index, Size)); + dumpBytes(makeArrayRef(Bytes.data() + Index, Size), outs()); } formatted_raw_ostream FormattedOS(outs()); - Annotations.flush(); StringRef AnnotationsStr = Annotations.str(); if (isThumb) - ThumbIP->printInst(&Inst, FormattedOS, AnnotationsStr); + ThumbIP->printInst(&Inst, FormattedOS, AnnotationsStr, *ThumbSTI); else - IP->printInst(&Inst, FormattedOS, AnnotationsStr); + IP->printInst(&Inst, FormattedOS, AnnotationsStr, *STI); emitComments(CommentStream, CommentsToEmit, FormattedOS, *AsmInfo); // Print debug info. @@ -2956,21 +6272,21 @@ static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF) { uint64_t PC = SectAddress + Index; if (DisAsm->getInstruction(Inst, InstSize, Bytes.slice(Index), PC, DebugOut, nulls())) { - if (FullLeadingAddr) { - if (MachOOF->is64Bit()) - outs() << format("%016" PRIx64, PC); - else - outs() << format("%08" PRIx64, PC); - } else { - outs() << format("%8" PRIx64 ":", PC); + if (!NoLeadingAddr) { + if (FullLeadingAddr) { + if (MachOOF->is64Bit()) + outs() << format("%016" PRIx64, PC); + else + outs() << format("%08" PRIx64, PC); + } else { + outs() << format("%8" PRIx64 ":", PC); + } } if (!NoShowRawInsn) { outs() << "\t"; - DumpBytes( - StringRef(reinterpret_cast(Bytes.data()) + Index, - InstSize)); + dumpBytes(makeArrayRef(Bytes.data() + Index, InstSize), outs()); } - IP->printInst(&Inst, outs(), ""); + IP->printInst(&Inst, outs(), "", *STI); outs() << "\n"; } else { unsigned int Arch = MachOOF->getArch(); @@ -3068,13 +6384,16 @@ static void findUnwindRelocNameAddend(const MachOObjectFile *Obj, const RelocationRef &Reloc, uint64_t Addr, StringRef &Name, uint64_t &Addend) { if (Reloc.getSymbol() != Obj->symbol_end()) { - Reloc.getSymbol()->getName(Name); + ErrorOr NameOrErr = Reloc.getSymbol()->getName(); + if (std::error_code EC = NameOrErr.getError()) + report_fatal_error(EC.message()); + Name = *NameOrErr; Addend = Addr; return; } auto RE = Obj->getRelocation(Reloc.getRawDataRefImpl()); - SectionRef RelocSection = Obj->getRelocationSection(RE); + SectionRef RelocSection = Obj->getAnyRelocationSection(RE); uint64_t SectionAddr = RelocSection.getAddress(); @@ -3090,11 +6409,13 @@ static void findUnwindRelocNameAddend(const MachOObjectFile *Obj, // Go back one so that SymbolAddress <= Addr. --Sym; - section_iterator SymSection = Obj->section_end(); - Sym->second.getSection(SymSection); + section_iterator SymSection = *Sym->second.getSection(); if (RelocSection == *SymSection) { // There's a valid symbol in the same section before this reference. - Sym->second.getName(Name); + ErrorOr NameOrErr = Sym->second.getName(); + if (std::error_code EC = NameOrErr.getError()) + report_fatal_error(EC.message()); + Name = *NameOrErr; Addend = Addr - Sym->first; return; } @@ -3111,7 +6432,7 @@ static void printUnwindRelocDest(const MachOObjectFile *Obj, StringRef Name; uint64_t Addend; - if (!Reloc.getObjectFile()) + if (!Reloc.getObject()) return; findUnwindRelocNameAddend(Obj, Symbols, Reloc, Addr, Name, Addend); @@ -3147,8 +6468,7 @@ printMachOCompactUnwindSection(const MachOObjectFile *Obj, // Next we need to look at the relocations to find out what objects are // actually being referred to. for (const RelocationRef &Reloc : CompactUnwind.relocations()) { - uint64_t RelocAddress; - Reloc.getOffset(RelocAddress); + uint64_t RelocAddress = Reloc.getOffset(); uint32_t EntryIdx = RelocAddress / EntrySize; uint32_t OffsetInEntry = RelocAddress - EntryIdx * EntrySize; @@ -3184,7 +6504,7 @@ printMachOCompactUnwindSection(const MachOObjectFile *Obj, << format("0x%08" PRIx32, Entry.CompactEncoding) << '\n'; // 4. The personality function, if present. - if (Entry.PersonalityReloc.getObjectFile()) { + if (Entry.PersonalityReloc.getObject()) { outs() << " personality function: " << format("0x%" PRIx64, Entry.PersonalityAddr) << ' '; printUnwindRelocDest(Obj, Symbols, Entry.PersonalityReloc, @@ -3193,7 +6513,7 @@ printMachOCompactUnwindSection(const MachOObjectFile *Obj, } // 5. This entry's language-specific data area. - if (Entry.LSDAReloc.getObjectFile()) { + if (Entry.LSDAReloc.getObject()) { outs() << " LSDA: " << format("0x%" PRIx64, Entry.LSDAAddr) << ' '; printUnwindRelocDest(Obj, Symbols, Entry.LSDAReloc, Entry.LSDAAddr); @@ -3427,18 +6747,272 @@ static void printMachOUnwindInfoSection(const MachOObjectFile *Obj, } } +static unsigned getSizeForEncoding(bool is64Bit, + unsigned symbolEncoding) { + unsigned format = symbolEncoding & 0x0f; + switch (format) { + default: llvm_unreachable("Unknown Encoding"); + case dwarf::DW_EH_PE_absptr: + case dwarf::DW_EH_PE_signed: + return is64Bit ? 8 : 4; + case dwarf::DW_EH_PE_udata2: + case dwarf::DW_EH_PE_sdata2: + return 2; + case dwarf::DW_EH_PE_udata4: + case dwarf::DW_EH_PE_sdata4: + return 4; + case dwarf::DW_EH_PE_udata8: + case dwarf::DW_EH_PE_sdata8: + return 8; + } +} + +static uint64_t readPointer(const char *&Pos, bool is64Bit, unsigned Encoding) { + switch (getSizeForEncoding(is64Bit, Encoding)) { + case 2: + return readNext(Pos); + break; + case 4: + return readNext(Pos); + break; + case 8: + return readNext(Pos); + break; + default: + llvm_unreachable("Illegal data size"); + } +} + +static void printMachOEHFrameSection(const MachOObjectFile *Obj, + std::map &Symbols, + const SectionRef &EHFrame) { + if (!Obj->isLittleEndian()) { + outs() << "warning: cannot handle big endian __eh_frame section\n"; + return; + } + + bool is64Bit = Obj->is64Bit(); + + outs() << "Contents of __eh_frame section:\n"; + + StringRef Contents; + EHFrame.getContents(Contents); + + /// A few fields of the CIE are used when decoding the FDE's. This struct + /// will cache those fields we need so that we don't have to decode it + /// repeatedly for each FDE that references it. + struct DecodedCIE { + Optional FDEPointerEncoding; + Optional LSDAPointerEncoding; + bool hasAugmentationLength; + }; + + // Map from the start offset of the CIE to the cached data for that CIE. + DenseMap CachedCIEs; + + for (const char *Pos = Contents.data(), *End = Contents.end(); Pos != End; ) { + + const char *EntryStartPos = Pos; + + uint64_t Length = readNext(Pos); + if (Length == 0xffffffff) + Length = readNext(Pos); + + // Save the Pos so that we can check the length we encoded against what we + // end up decoding. + const char *PosAfterLength = Pos; + const char *EntryEndPos = PosAfterLength + Length; + + assert(EntryEndPos <= End && + "__eh_frame entry length exceeds section size"); + + uint32_t ID = readNext(Pos); + if (ID == 0) { + // This is a CIE. + + uint32_t Version = readNext(Pos); + + // Parse a null terminated augmentation string + SmallString<8> AugmentationString; + for (uint8_t Char = readNext(Pos); Char; + Char = readNext(Pos)) + AugmentationString.push_back(Char); + + // Optionally parse the EH data if the augmentation string says it's there. + Optional EHData; + if (StringRef(AugmentationString).count("eh")) + EHData = is64Bit ? readNext(Pos) : readNext(Pos); + + unsigned ULEBByteCount; + uint64_t CodeAlignmentFactor = decodeULEB128((const uint8_t *)Pos, + &ULEBByteCount); + Pos += ULEBByteCount; + + int64_t DataAlignmentFactor = decodeSLEB128((const uint8_t *)Pos, + &ULEBByteCount); + Pos += ULEBByteCount; + + uint32_t ReturnAddressRegister = readNext(Pos); + + Optional AugmentationLength; + Optional LSDAPointerEncoding; + Optional PersonalityEncoding; + Optional Personality; + Optional FDEPointerEncoding; + if (!AugmentationString.empty() && AugmentationString.front() == 'z') { + AugmentationLength = decodeULEB128((const uint8_t *)Pos, + &ULEBByteCount); + Pos += ULEBByteCount; + + // Walk the augmentation string to get all the augmentation data. + for (unsigned i = 1, e = AugmentationString.size(); i != e; ++i) { + char Char = AugmentationString[i]; + switch (Char) { + case 'e': + assert((i + 1) != e && AugmentationString[i + 1] == 'h' && + "Expected 'eh' in augmentation string"); + break; + case 'L': + assert(!LSDAPointerEncoding && "Duplicate LSDA encoding"); + LSDAPointerEncoding = readNext(Pos); + break; + case 'P': { + assert(!Personality && "Duplicate personality"); + PersonalityEncoding = readNext(Pos); + Personality = readPointer(Pos, is64Bit, *PersonalityEncoding); + break; + } + case 'R': + assert(!FDEPointerEncoding && "Duplicate FDE encoding"); + FDEPointerEncoding = readNext(Pos); + break; + case 'z': + llvm_unreachable("'z' must be first in the augmentation string"); + } + } + } + + outs() << "CIE:\n"; + outs() << " Length: " << Length << "\n"; + outs() << " CIE ID: " << ID << "\n"; + outs() << " Version: " << Version << "\n"; + outs() << " Augmentation String: " << AugmentationString << "\n"; + if (EHData) + outs() << " EHData: " << *EHData << "\n"; + outs() << " Code Alignment Factor: " << CodeAlignmentFactor << "\n"; + outs() << " Data Alignment Factor: " << DataAlignmentFactor << "\n"; + outs() << " Return Address Register: " << ReturnAddressRegister << "\n"; + if (AugmentationLength) { + outs() << " Augmentation Data Length: " << *AugmentationLength << "\n"; + if (LSDAPointerEncoding) { + outs() << " FDE LSDA Pointer Encoding: " + << *LSDAPointerEncoding << "\n"; + } + if (Personality) { + outs() << " Personality Encoding: " << *PersonalityEncoding << "\n"; + outs() << " Personality: " << *Personality << "\n"; + } + if (FDEPointerEncoding) { + outs() << " FDE Address Pointer Encoding: " + << *FDEPointerEncoding << "\n"; + } + } + // FIXME: Handle instructions. + // For now just emit some bytes + outs() << " Instructions:\n "; + dumpBytes(makeArrayRef((const uint8_t*)Pos, (const uint8_t*)EntryEndPos), + outs()); + outs() << "\n"; + Pos = EntryEndPos; + + // Cache this entry. + uint64_t Offset = EntryStartPos - Contents.data(); + CachedCIEs[Offset] = { FDEPointerEncoding, LSDAPointerEncoding, + AugmentationLength.hasValue() }; + continue; + } + + // This is an FDE. + // The CIE pointer for an FDE is the same location as the ID which we + // already read. + uint32_t CIEPointer = ID; + + const char *CIEStart = PosAfterLength - CIEPointer; + assert(CIEStart >= Contents.data() && + "FDE points to CIE before the __eh_frame start"); + + uint64_t CIEOffset = CIEStart - Contents.data(); + auto CIEIt = CachedCIEs.find(CIEOffset); + if (CIEIt == CachedCIEs.end()) + llvm_unreachable("Couldn't find CIE at offset in to __eh_frame section"); + + const DecodedCIE &CIE = CIEIt->getSecond(); + assert(CIE.FDEPointerEncoding && + "FDE references CIE which did not set pointer encoding"); + + uint64_t PCPointerSize = getSizeForEncoding(is64Bit, + *CIE.FDEPointerEncoding); + + uint64_t PCBegin = readPointer(Pos, is64Bit, *CIE.FDEPointerEncoding); + uint64_t PCRange = readPointer(Pos, is64Bit, *CIE.FDEPointerEncoding); + + Optional AugmentationLength; + uint32_t LSDAPointerSize; + Optional LSDAPointer; + if (CIE.hasAugmentationLength) { + unsigned ULEBByteCount; + AugmentationLength = decodeULEB128((const uint8_t *)Pos, + &ULEBByteCount); + Pos += ULEBByteCount; + + // Decode the LSDA if the CIE augmentation string said we should. + if (CIE.LSDAPointerEncoding) { + LSDAPointerSize = getSizeForEncoding(is64Bit, *CIE.LSDAPointerEncoding); + LSDAPointer = readPointer(Pos, is64Bit, *CIE.LSDAPointerEncoding); + } + } + + outs() << "FDE:\n"; + outs() << " Length: " << Length << "\n"; + outs() << " CIE Offset: " << CIEOffset << "\n"; + + if (PCPointerSize == 8) { + outs() << format(" PC Begin: %016" PRIx64, PCBegin) << "\n"; + outs() << format(" PC Range: %016" PRIx64, PCRange) << "\n"; + } else { + outs() << format(" PC Begin: %08" PRIx64, PCBegin) << "\n"; + outs() << format(" PC Range: %08" PRIx64, PCRange) << "\n"; + } + if (AugmentationLength) { + outs() << " Augmentation Data Length: " << *AugmentationLength << "\n"; + if (LSDAPointer) { + if (LSDAPointerSize == 8) + outs() << format(" LSDA Pointer: %016\n" PRIx64, *LSDAPointer); + else + outs() << format(" LSDA Pointer: %08\n" PRIx64, *LSDAPointer); + } + } + + // FIXME: Handle instructions. + // For now just emit some bytes + outs() << " Instructions:\n "; + dumpBytes(makeArrayRef((const uint8_t*)Pos, (const uint8_t*)EntryEndPos), + outs()); + outs() << "\n"; + Pos = EntryEndPos; + } +} + void llvm::printMachOUnwindInfo(const MachOObjectFile *Obj) { std::map Symbols; for (const SymbolRef &SymRef : Obj->symbols()) { // Discard any undefined or absolute symbols. They're not going to take part // in the convenience lookup for unwind info and just take up resources. - section_iterator Section = Obj->section_end(); - SymRef.getSection(Section); + section_iterator Section = *SymRef.getSection(); if (Section == Obj->section_end()) continue; - uint64_t Addr; - SymRef.getAddress(Addr); + uint64_t Addr = SymRef.getValue(); Symbols.insert(std::make_pair(Addr, SymRef)); } @@ -3450,7 +7024,7 @@ void llvm::printMachOUnwindInfo(const MachOObjectFile *Obj) { else if (SectName == "__unwind_info") printMachOUnwindInfoSection(Obj, Symbols, Section); else if (SectName == "__eh_frame") - outs() << "llvm-objdump: warning: unhandled __eh_frame section\n"; + printMachOEHFrameSection(Obj, Symbols, Section); } } @@ -3776,36 +7350,20 @@ static void PrintSegmentCommand(uint32_t cmd, uint32_t cmdsize, MachO::VM_PROT_EXECUTE)) != 0) outs() << " maxprot ?" << format("0x%08" PRIx32, maxprot) << "\n"; else { - if (maxprot & MachO::VM_PROT_READ) - outs() << " maxprot r"; - else - outs() << " maxprot -"; - if (maxprot & MachO::VM_PROT_WRITE) - outs() << "w"; - else - outs() << "-"; - if (maxprot & MachO::VM_PROT_EXECUTE) - outs() << "x\n"; - else - outs() << "-\n"; + outs() << " maxprot "; + outs() << ((maxprot & MachO::VM_PROT_READ) ? "r" : "-"); + outs() << ((maxprot & MachO::VM_PROT_WRITE) ? "w" : "-"); + outs() << ((maxprot & MachO::VM_PROT_EXECUTE) ? "x\n" : "-\n"); } if ((initprot & ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE | MachO::VM_PROT_EXECUTE)) != 0) outs() << " initprot ?" << format("0x%08" PRIx32, initprot) << "\n"; else { - if (initprot & MachO::VM_PROT_READ) - outs() << " initprot r"; - else - outs() << " initprot -"; - if (initprot & MachO::VM_PROT_WRITE) - outs() << "w"; - else - outs() << "-"; - if (initprot & MachO::VM_PROT_EXECUTE) - outs() << "x\n"; - else - outs() << "-\n"; + outs() << " initprot "; + outs() << ((initprot & MachO::VM_PROT_READ) ? "r" : "-"); + outs() << ((initprot & MachO::VM_PROT_WRITE) ? "w" : "-"); + outs() << ((initprot & MachO::VM_PROT_EXECUTE) ? "x\n" : "-\n"); } } else { outs() << " maxprot " << format("0x%08" PRIx32, maxprot) << "\n"; @@ -4259,26 +7817,11 @@ static void PrintUuidLoadCommand(MachO::uuid_command uuid) { else outs() << "\n"; outs() << " uuid "; - outs() << format("%02" PRIX32, uuid.uuid[0]); - outs() << format("%02" PRIX32, uuid.uuid[1]); - outs() << format("%02" PRIX32, uuid.uuid[2]); - outs() << format("%02" PRIX32, uuid.uuid[3]); - outs() << "-"; - outs() << format("%02" PRIX32, uuid.uuid[4]); - outs() << format("%02" PRIX32, uuid.uuid[5]); - outs() << "-"; - outs() << format("%02" PRIX32, uuid.uuid[6]); - outs() << format("%02" PRIX32, uuid.uuid[7]); - outs() << "-"; - outs() << format("%02" PRIX32, uuid.uuid[8]); - outs() << format("%02" PRIX32, uuid.uuid[9]); - outs() << "-"; - outs() << format("%02" PRIX32, uuid.uuid[10]); - outs() << format("%02" PRIX32, uuid.uuid[11]); - outs() << format("%02" PRIX32, uuid.uuid[12]); - outs() << format("%02" PRIX32, uuid.uuid[13]); - outs() << format("%02" PRIX32, uuid.uuid[14]); - outs() << format("%02" PRIX32, uuid.uuid[15]); + for (int i = 0; i < 16; ++i) { + outs() << format("%02" PRIX32, uuid.uuid[i]); + if (i == 3 || i == 5 || i == 7 || i == 9) + outs() << "-"; + } outs() << "\n"; } @@ -4298,30 +7841,47 @@ static void PrintRpathLoadCommand(MachO::rpath_command rpath, const char *Ptr) { } static void PrintVersionMinLoadCommand(MachO::version_min_command vd) { - if (vd.cmd == MachO::LC_VERSION_MIN_MACOSX) - outs() << " cmd LC_VERSION_MIN_MACOSX\n"; - else if (vd.cmd == MachO::LC_VERSION_MIN_IPHONEOS) - outs() << " cmd LC_VERSION_MIN_IPHONEOS\n"; - else - outs() << " cmd " << vd.cmd << " (?)\n"; + StringRef LoadCmdName; + switch (vd.cmd) { + case MachO::LC_VERSION_MIN_MACOSX: + LoadCmdName = "LC_VERSION_MIN_MACOSX"; + break; + case MachO::LC_VERSION_MIN_IPHONEOS: + LoadCmdName = "LC_VERSION_MIN_IPHONEOS"; + break; + case MachO::LC_VERSION_MIN_TVOS: + LoadCmdName = "LC_VERSION_MIN_TVOS"; + break; + case MachO::LC_VERSION_MIN_WATCHOS: + LoadCmdName = "LC_VERSION_MIN_WATCHOS"; + break; + default: + llvm_unreachable("Unknown version min load command"); + } + + outs() << " cmd " << LoadCmdName << '\n'; outs() << " cmdsize " << vd.cmdsize; if (vd.cmdsize != sizeof(struct MachO::version_min_command)) outs() << " Incorrect size\n"; else outs() << "\n"; - outs() << " version " << ((vd.version >> 16) & 0xffff) << "." - << ((vd.version >> 8) & 0xff); - if ((vd.version & 0xff) != 0) - outs() << "." << (vd.version & 0xff); + outs() << " version " + << MachOObjectFile::getVersionMinMajor(vd, false) << "." + << MachOObjectFile::getVersionMinMinor(vd, false); + uint32_t Update = MachOObjectFile::getVersionMinUpdate(vd, false); + if (Update != 0) + outs() << "." << Update; outs() << "\n"; if (vd.sdk == 0) outs() << " sdk n/a"; else { - outs() << " sdk " << ((vd.sdk >> 16) & 0xffff) << "." - << ((vd.sdk >> 8) & 0xff); + outs() << " sdk " + << MachOObjectFile::getVersionMinMajor(vd, true) << "." + << MachOObjectFile::getVersionMinMinor(vd, true); } - if ((vd.sdk & 0xff) != 0) - outs() << "." << (vd.sdk & 0xff); + Update = MachOObjectFile::getVersionMinUpdate(vd, true); + if (Update != 0) + outs() << "." << Update; outs() << "\n"; } @@ -4957,15 +8517,12 @@ static void PrintLinkEditDataCommand(MachO::linkedit_data_command ld, outs() << "\n"; } -static void PrintLoadCommands(const MachOObjectFile *Obj, uint32_t ncmds, - uint32_t filetype, uint32_t cputype, - bool verbose) { - if (ncmds == 0) - return; +static void PrintLoadCommands(const MachOObjectFile *Obj, uint32_t filetype, + uint32_t cputype, bool verbose) { StringRef Buf = Obj->getData(); - MachOObjectFile::LoadCommandInfo Command = Obj->getFirstLoadCommandInfo(); - for (unsigned i = 0;; ++i) { - outs() << "Load command " << i << "\n"; + unsigned Index = 0; + for (const auto &Command : Obj->load_commands()) { + outs() << "Load command " << Index++ << "\n"; if (Command.C.cmd == MachO::LC_SEGMENT) { MachO::segment_command SLC = Obj->getSegmentLoadCommand(Command); const char *sg_segname = SLC.segname; @@ -5017,7 +8574,9 @@ static void PrintLoadCommands(const MachOObjectFile *Obj, uint32_t ncmds, MachO::rpath_command Rpath = Obj->getRpathCommand(Command); PrintRpathLoadCommand(Rpath, Command.Ptr); } else if (Command.C.cmd == MachO::LC_VERSION_MIN_MACOSX || - Command.C.cmd == MachO::LC_VERSION_MIN_IPHONEOS) { + Command.C.cmd == MachO::LC_VERSION_MIN_IPHONEOS || + Command.C.cmd == MachO::LC_VERSION_MIN_TVOS || + Command.C.cmd == MachO::LC_VERSION_MIN_WATCHOS) { MachO::version_min_command Vd = Obj->getVersionMinLoadCommand(Command); PrintVersionMinLoadCommand(Vd); } else if (Command.C.cmd == MachO::LC_SOURCE_VERSION) { @@ -5084,14 +8643,10 @@ static void PrintLoadCommands(const MachOObjectFile *Obj, uint32_t ncmds, // TODO: get and print the raw bytes of the load command. } // TODO: print all the other kinds of load commands. - if (i == ncmds - 1) - break; - else - Command = Obj->getNextLoadCommandInfo(Command); } } -static void getAndPrintMachHeader(const MachOObjectFile *Obj, uint32_t &ncmds, +static void getAndPrintMachHeader(const MachOObjectFile *Obj, uint32_t &filetype, uint32_t &cputype, bool verbose) { if (Obj->is64Bit()) { @@ -5099,7 +8654,6 @@ static void getAndPrintMachHeader(const MachOObjectFile *Obj, uint32_t &ncmds, H_64 = Obj->getHeader64(); PrintMachHeader(H_64.magic, H_64.cputype, H_64.cpusubtype, H_64.filetype, H_64.ncmds, H_64.sizeofcmds, H_64.flags, verbose); - ncmds = H_64.ncmds; filetype = H_64.filetype; cputype = H_64.cputype; } else { @@ -5107,7 +8661,6 @@ static void getAndPrintMachHeader(const MachOObjectFile *Obj, uint32_t &ncmds, H = Obj->getHeader(); PrintMachHeader(H.magic, H.cputype, H.cpusubtype, H.filetype, H.ncmds, H.sizeofcmds, H.flags, verbose); - ncmds = H.ncmds; filetype = H.filetype; cputype = H.cputype; } @@ -5115,11 +8668,10 @@ static void getAndPrintMachHeader(const MachOObjectFile *Obj, uint32_t &ncmds, void llvm::printMachOFileHeader(const object::ObjectFile *Obj) { const MachOObjectFile *file = dyn_cast(Obj); - uint32_t ncmds = 0; uint32_t filetype = 0; uint32_t cputype = 0; - getAndPrintMachHeader(file, ncmds, filetype, cputype, true); - PrintLoadCommands(file, ncmds, filetype, cputype, true); + getAndPrintMachHeader(file, filetype, cputype, !NonVerbose); + PrintLoadCommands(file, filetype, cputype, !NonVerbose); } //===----------------------------------------------------------------------===// @@ -5194,6 +8746,7 @@ public: StringRef segmentName(uint32_t SegIndex); StringRef sectionName(uint32_t SegIndex, uint64_t SegOffset); uint64_t address(uint32_t SegIndex, uint64_t SegOffset); + bool isValidSegIndexAndOffset(uint32_t SegIndex, uint64_t SegOffset); private: struct SectionInfo { @@ -5217,8 +8770,7 @@ SegInfo::SegInfo(const object::MachOObjectFile *Obj) { uint64_t CurSegAddress; for (const SectionRef &Section : Obj->sections()) { SectionInfo Info; - if (error(Section.getName(Info.SectionName))) - return; + error(Section.getName(Info.SectionName)); Info.Address = Section.getAddress(); Info.Size = Section.getSize(); Info.SegmentName = @@ -5243,6 +8795,20 @@ StringRef SegInfo::segmentName(uint32_t SegIndex) { llvm_unreachable("invalid segIndex"); } +bool SegInfo::isValidSegIndexAndOffset(uint32_t SegIndex, + uint64_t OffsetInSeg) { + for (const SectionInfo &SI : Sections) { + if (SI.SegmentIndex != SegIndex) + continue; + if (SI.OffsetInSegment > OffsetInSeg) + continue; + if (OffsetInSeg >= (SI.OffsetInSegment + SI.Size)) + continue; + return true; + } + return false; +} + const SegInfo::SectionInfo &SegInfo::findSection(uint32_t SegIndex, uint64_t OffsetInSeg) { for (const SectionInfo &SI : Sections) { @@ -5411,6 +8977,8 @@ static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue, for (const llvm::object::MachOBindEntry &Entry : info->O->bindTable()) { uint32_t SegIndex = Entry.segmentIndex(); uint64_t OffsetInSeg = Entry.segmentOffset(); + if (!sectionTable.isValidSegIndexAndOffset(SegIndex, OffsetInSeg)) + continue; uint64_t Address = sectionTable.address(SegIndex, OffsetInSeg); const char *SymbolName = nullptr; StringRef name = Entry.symbolName();