tools/macho-dump/macho-dump.cpp

   1 //===-- macho-dump.cpp - Mach Object Dumping Tool -------------------------===//
   2 //
   3 //                     The LLVM Compiler Infrastructure
   4 //
   5 // This file is distributed under the University of Illinois Open Source
   6 // License. See LICENSE.TXT for details.
   7 //
   8 //===----------------------------------------------------------------------===//
   9 //
  10 // This is a testing tool for use with the MC/Mach-O LLVM components.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #include "llvm/Object/MachOObject.h"
  15 #include "llvm/ADT/StringExtras.h"
  16 #include "llvm/ADT/Twine.h"
  17 #include "llvm/Support/CommandLine.h"
  18 #include "llvm/Support/Format.h"
  19 #include "llvm/Support/ManagedStatic.h"
  20 #include "llvm/Support/MemoryBuffer.h"
  21 #include "llvm/Support/raw_ostream.h"
  22 #include "llvm/Support/system_error.h"
  23 using namespace llvm;
  24 using namespace llvm::object;
  25
  26 static cl::opt<std::string>
  27 InputFile(cl::Positional, cl::desc("<input file>"), cl::init("-"));
  28
  29 static cl::opt<bool>
  30 ShowSectionData("dump-section-data", cl::desc("Dump the contents of sections"),
  31                 cl::init(false));
  32
  33 ///
  34
  35 static const char *ProgramName;
  36
  37 static void Message(const char *Type, const Twine &Msg) {
  38   errs() << ProgramName << ": " << Type << ": " << Msg << "\n";
  39 }
  40
  41 static int Error(const Twine &Msg) {
  42   Message("error", Msg);
  43   return 1;
  44 }
  45
  46 static void Warning(const Twine &Msg) {
  47   Message("warning", Msg);
  48 }
  49
  50 ///
  51
  52 static int DumpHeader(MachOObject &Obj) {
  53   // Read the header.
  54   const macho::Header &Hdr = Obj.getHeader();
  55   outs() << "('cputype', " << Hdr.CPUType << ")\n";
  56   outs() << "('cpusubtype', " << Hdr.CPUSubtype << ")\n";
  57   outs() << "('filetype', " << Hdr.FileType << ")\n";
  58   outs() << "('num_load_commands', " << Hdr.NumLoadCommands << ")\n";
  59   outs() << "('load_commands_size', " << Hdr.SizeOfLoadCommands << ")\n";
  60   outs() << "('flag', " << Hdr.Flags << ")\n";
  61
  62   // Print extended header if 64-bit.
  63   if (Obj.is64Bit()) {
  64     const macho::Header64Ext &Hdr64 = Obj.getHeader64Ext();
  65     outs() << "('reserved', " << Hdr64.Reserved << ")\n";
  66   }
  67
  68   return 0;
  69 }
  70
  71 static void DumpSegmentCommandData(StringRef Name,
  72                                    uint64_t VMAddr, uint64_t VMSize,
  73                                    uint64_t FileOffset, uint64_t FileSize,
  74                                    uint32_t MaxProt, uint32_t InitProt,
  75                                    uint32_t NumSections, uint32_t Flags) {
  76   outs() << "  ('segment_name', '";
  77   outs().write_escaped(Name, /*UseHexEscapes=*/true) << "')\n";
  78   outs() << "  ('vm_addr', " << VMAddr << ")\n";
  79   outs() << "  ('vm_size', " << VMSize << ")\n";
  80   outs() << "  ('file_offset', " << FileOffset << ")\n";
  81   outs() << "  ('file_size', " << FileSize << ")\n";
  82   outs() << "  ('maxprot', " << MaxProt << ")\n";
  83   outs() << "  ('initprot', " << InitProt << ")\n";
  84   outs() << "  ('num_sections', " << NumSections << ")\n";
  85   outs() << "  ('flags', " << Flags << ")\n";
  86 }
  87
  88 static int DumpSectionData(MachOObject &Obj, unsigned Index, StringRef Name,
  89                            StringRef SegmentName, uint64_t Address,
  90                            uint64_t Size, uint32_t Offset,
  91                            uint32_t Align, uint32_t RelocationTableOffset,
  92                            uint32_t NumRelocationTableEntries,
  93                            uint32_t Flags, uint32_t Reserved1,
  94                            uint32_t Reserved2, uint64_t Reserved3 = ~0ULL) {
  95   outs() << "    # Section " << Index << "\n";
  96   outs() << "   (('section_name', '";
  97   outs().write_escaped(Name, /*UseHexEscapes=*/true) << "')\n";
  98   outs() << "    ('segment_name', '";
  99   outs().write_escaped(SegmentName, /*UseHexEscapes=*/true) << "')\n";
 100   outs() << "    ('address', " << Address << ")\n";
 101   outs() << "    ('size', " << Size << ")\n";
 102   outs() << "    ('offset', " << Offset << ")\n";
 103   outs() << "    ('alignment', " << Align << ")\n";
 104   outs() << "    ('reloc_offset', " << RelocationTableOffset << ")\n";
 105   outs() << "    ('num_reloc', " << NumRelocationTableEntries << ")\n";
 106   outs() << "    ('flags', " << format("0x%x", Flags) << ")\n";
 107   outs() << "    ('reserved1', " << Reserved1 << ")\n";
 108   outs() << "    ('reserved2', " << Reserved2 << ")\n";
 109   if (Reserved3 != ~0ULL)
 110     outs() << "    ('reserved3', " << Reserved3 << ")\n";
 111   outs() << "   ),\n";
 112
 113   // Dump the relocation entries.
 114   int Res = 0;
 115   outs() << "  ('_relocations', [\n";
 116   for (unsigned i = 0; i != NumRelocationTableEntries; ++i) {
 117     InMemoryStruct<macho::RelocationEntry> RE;
 118     Obj.ReadRelocationEntry(RelocationTableOffset, i, RE);
 119     if (!RE) {
 120       Res = Error("unable to read relocation table entry '" + Twine(i) + "'");
 121       break;
 122     }
 123
 124     outs() << "    # Relocation " << i << "\n";
 125     outs() << "    (('word-0', " << format("0x%x", RE->Word0) << "),\n";
 126     outs() << "     ('word-1', " << format("0x%x", RE->Word1) << ")),\n";
 127   }
 128   outs() << "  ])\n";
 129
 130   // Dump the section data, if requested.
 131   if (ShowSectionData) {
 132     outs() << "  ('_section_data', '";
 133     StringRef Data = Obj.getData(Offset, Size);
 134     for (unsigned i = 0; i != Data.size(); ++i) {
 135       if (i && (i % 4) == 0)
 136         outs() << ' ';
 137       outs() << hexdigit((Data[i] >> 4) & 0xF, /*LowerCase=*/true);
 138       outs() << hexdigit((Data[i] >> 0) & 0xF, /*LowerCase=*/true);
 139     }
 140     outs() << "')\n";
 141   }
 142
 143   return Res;
 144 }
 145
 146 static int DumpSegmentCommand(MachOObject &Obj,
 147                                const MachOObject::LoadCommandInfo &LCI) {
 148   InMemoryStruct<macho::SegmentLoadCommand> SLC;
 149   Obj.ReadSegmentLoadCommand(LCI, SLC);
 150   if (!SLC)
 151     return Error("unable to read segment load command");
 152
 153   DumpSegmentCommandData(StringRef(SLC->Name, 16), SLC->VMAddress,
 154                          SLC->VMSize, SLC->FileOffset, SLC->FileSize,
 155                          SLC->MaxVMProtection, SLC->InitialVMProtection,
 156                          SLC->NumSections, SLC->Flags);
 157
 158   // Dump the sections.
 159   int Res = 0;
 160   outs() << "  ('sections', [\n";
 161   for (unsigned i = 0; i != SLC->NumSections; ++i) {
 162     InMemoryStruct<macho::Section> Sect;
 163     Obj.ReadSection(LCI, i, Sect);
 164     if (!SLC) {
 165       Res = Error("unable to read section '" + Twine(i) + "'");
 166       break;
 167     }
 168
 169     if ((Res = DumpSectionData(Obj, i, StringRef(Sect->Name, 16),
 170                                StringRef(Sect->SegmentName, 16), Sect->Address,
 171                                Sect->Size, Sect->Offset, Sect->Align,
 172                                Sect->RelocationTableOffset,
 173                                Sect->NumRelocationTableEntries, Sect->Flags,
 174                                Sect->Reserved1, Sect->Reserved2)))
 175       break;
 176   }
 177   outs() << "  ])\n";
 178
 179   return Res;
 180 }
 181
 182 static int DumpSegment64Command(MachOObject &Obj,
 183                                const MachOObject::LoadCommandInfo &LCI) {
 184   InMemoryStruct<macho::Segment64LoadCommand> SLC;
 185   Obj.ReadSegment64LoadCommand(LCI, SLC);
 186   if (!SLC)
 187     return Error("unable to read segment load command");
 188
 189   DumpSegmentCommandData(StringRef(SLC->Name, 16), SLC->VMAddress,
 190                          SLC->VMSize, SLC->FileOffset, SLC->FileSize,
 191                          SLC->MaxVMProtection, SLC->InitialVMProtection,
 192                          SLC->NumSections, SLC->Flags);
 193
 194   // Dump the sections.
 195   int Res = 0;
 196   outs() << "  ('sections', [\n";
 197   for (unsigned i = 0; i != SLC->NumSections; ++i) {
 198     InMemoryStruct<macho::Section64> Sect;
 199     Obj.ReadSection64(LCI, i, Sect);
 200     if (!SLC) {
 201       Res = Error("unable to read section '" + Twine(i) + "'");
 202       break;
 203     }
 204
 205     if ((Res = DumpSectionData(Obj, i, StringRef(Sect->Name, 16),
 206                                StringRef(Sect->SegmentName, 16), Sect->Address,
 207                                Sect->Size, Sect->Offset, Sect->Align,
 208                                Sect->RelocationTableOffset,
 209                                Sect->NumRelocationTableEntries, Sect->Flags,
 210                                Sect->Reserved1, Sect->Reserved2,
 211                                Sect->Reserved3)))
 212       break;
 213   }
 214   outs() << "  ])\n";
 215
 216   return 0;
 217 }
 218
 219 static void DumpSymbolTableEntryData(MachOObject &Obj,
 220                                      unsigned Index, uint32_t StringIndex,
 221                                      uint8_t Type, uint8_t SectionIndex,
 222                                      uint16_t Flags, uint64_t Value) {
 223   outs() << "    # Symbol " << Index << "\n";
 224   outs() << "   (('n_strx', " << StringIndex << ")\n";
 225   outs() << "    ('n_type', " << format("0x%x", Type) << ")\n";
 226   outs() << "    ('n_sect', " << uint32_t(SectionIndex) << ")\n";
 227   outs() << "    ('n_desc', " << Flags << ")\n";
 228   outs() << "    ('n_value', " << Value << ")\n";
 229   outs() << "    ('_string', '" << Obj.getStringAtIndex(StringIndex) << "')\n";
 230   outs() << "   ),\n";
 231 }
 232
 233 static int DumpSymtabCommand(MachOObject &Obj,
 234                              const MachOObject::LoadCommandInfo &LCI) {
 235   InMemoryStruct<macho::SymtabLoadCommand> SLC;
 236   Obj.ReadSymtabLoadCommand(LCI, SLC);
 237   if (!SLC)
 238     return Error("unable to read segment load command");
 239
 240   outs() << "  ('symoff', " << SLC->SymbolTableOffset << ")\n";
 241   outs() << "  ('nsyms', " << SLC->NumSymbolTableEntries << ")\n";
 242   outs() << "  ('stroff', " << SLC->StringTableOffset << ")\n";
 243   outs() << "  ('strsize', " << SLC->StringTableSize << ")\n";
 244
 245   // Cache the string table data.
 246   Obj.RegisterStringTable(*SLC);
 247
 248   // Dump the string data.
 249   outs() << "  ('_string_data', '";
 250   outs().write_escaped(Obj.getStringTableData(),
 251                        /*UseHexEscapes=*/true) << "')\n";
 252
 253   // Dump the symbol table.
 254   int Res = 0;
 255   outs() << "  ('_symbols', [\n";
 256   for (unsigned i = 0; i != SLC->NumSymbolTableEntries; ++i) {
 257     if (Obj.is64Bit()) {
 258       InMemoryStruct<macho::Symbol64TableEntry> STE;
 259       Obj.ReadSymbol64TableEntry(SLC->SymbolTableOffset, i, STE);
 260       if (!STE) {
 261         Res = Error("unable to read symbol: '" + Twine(i) + "'");
 262         break;
 263       }
 264
 265       DumpSymbolTableEntryData(Obj, i, STE->StringIndex, STE->Type,
 266                                STE->SectionIndex, STE->Flags, STE->Value);
 267     } else {
 268       InMemoryStruct<macho::SymbolTableEntry> STE;
 269       Obj.ReadSymbolTableEntry(SLC->SymbolTableOffset, i, STE);
 270       if (!SLC) {
 271         Res = Error("unable to read symbol: '" + Twine(i) + "'");
 272         break;
 273       }
 274
 275       DumpSymbolTableEntryData(Obj, i, STE->StringIndex, STE->Type,
 276                                STE->SectionIndex, STE->Flags, STE->Value);
 277     }
 278   }
 279   outs() << "  ])\n";
 280
 281   return Res;
 282 }
 283
 284 static int DumpDysymtabCommand(MachOObject &Obj,
 285                              const MachOObject::LoadCommandInfo &LCI) {
 286   InMemoryStruct<macho::DysymtabLoadCommand> DLC;
 287   Obj.ReadDysymtabLoadCommand(LCI, DLC);
 288   if (!DLC)
 289     return Error("unable to read segment load command");
 290
 291   outs() << "  ('ilocalsym', " << DLC->LocalSymbolsIndex << ")\n";
 292   outs() << "  ('nlocalsym', " << DLC->NumLocalSymbols << ")\n";
 293   outs() << "  ('iextdefsym', " << DLC->ExternalSymbolsIndex << ")\n";
 294   outs() << "  ('nextdefsym', " << DLC->NumExternalSymbols << ")\n";
 295   outs() << "  ('iundefsym', " << DLC->UndefinedSymbolsIndex << ")\n";
 296   outs() << "  ('nundefsym', " << DLC->NumUndefinedSymbols << ")\n";
 297   outs() << "  ('tocoff', " << DLC->TOCOffset << ")\n";
 298   outs() << "  ('ntoc', " << DLC->NumTOCEntries << ")\n";
 299   outs() << "  ('modtaboff', " << DLC->ModuleTableOffset << ")\n";
 300   outs() << "  ('nmodtab', " << DLC->NumModuleTableEntries << ")\n";
 301   outs() << "  ('extrefsymoff', " << DLC->ReferenceSymbolTableOffset << ")\n";
 302   outs() << "  ('nextrefsyms', "
 303          << DLC->NumReferencedSymbolTableEntries << ")\n";
 304   outs() << "  ('indirectsymoff', " << DLC->IndirectSymbolTableOffset << ")\n";
 305   outs() << "  ('nindirectsyms', "
 306          << DLC->NumIndirectSymbolTableEntries << ")\n";
 307   outs() << "  ('extreloff', " << DLC->ExternalRelocationTableOffset << ")\n";
 308   outs() << "  ('nextrel', " << DLC->NumExternalRelocationTableEntries << ")\n";
 309   outs() << "  ('locreloff', " << DLC->LocalRelocationTableOffset << ")\n";
 310   outs() << "  ('nlocrel', " << DLC->NumLocalRelocationTableEntries << ")\n";
 311
 312   // Dump the indirect symbol table.
 313   int Res = 0;
 314   outs() << "  ('_indirect_symbols', [\n";
 315   for (unsigned i = 0; i != DLC->NumIndirectSymbolTableEntries; ++i) {
 316     InMemoryStruct<macho::IndirectSymbolTableEntry> ISTE;
 317     Obj.ReadIndirectSymbolTableEntry(*DLC, i, ISTE);
 318     if (!ISTE) {
 319       Res = Error("unable to read segment load command");
 320       break;
 321     }
 322
 323     outs() << "    # Indirect Symbol " << i << "\n";
 324     outs() << "    (('symbol_index', "
 325            << format("0x%x", ISTE->Index) << "),),\n";
 326   }
 327   outs() << "  ])\n";
 328
 329   return Res;
 330 }
 331
 332 static int DumpLoadCommand(MachOObject &Obj, unsigned Index) {
 333   const MachOObject::LoadCommandInfo &LCI = Obj.getLoadCommandInfo(Index);
 334   int Res = 0;
 335
 336   outs() << "  # Load Command " << Index << "\n"
 337          << " (('command', " << LCI.Command.Type << ")\n"
 338          << "  ('size', " << LCI.Command.Size << ")\n";
 339   switch (LCI.Command.Type) {
 340   case macho::LCT_Segment:
 341     Res = DumpSegmentCommand(Obj, LCI);
 342     break;
 343   case macho::LCT_Segment64:
 344     Res = DumpSegment64Command(Obj, LCI);
 345     break;
 346   case macho::LCT_Symtab:
 347     Res = DumpSymtabCommand(Obj, LCI);
 348     break;
 349   case macho::LCT_Dysymtab:
 350     Res = DumpDysymtabCommand(Obj, LCI);
 351     break;
 352   default:
 353     Warning("unknown load command: " + Twine(LCI.Command.Type));
 354     break;
 355   }
 356   outs() << " ),\n";
 357
 358   return Res;
 359 }
 360
 361 int main(int argc, char **argv) {
 362   ProgramName = argv[0];
 363   llvm_shutdown_obj Y;  // Call llvm_shutdown() on exit.
 364
 365   cl::ParseCommandLineOptions(argc, argv, "llvm Mach-O dumping tool\n");
 366
 367   // Load the input file.
 368   std::string ErrorStr;
 369   OwningPtr<MemoryBuffer> InputBuffer;
 370   if (error_code ec = MemoryBuffer::getFileOrSTDIN(InputFile, InputBuffer))
 371     return Error("unable to read input: '" + ec.message() + "'");
 372
 373   // Construct the Mach-O wrapper object.
 374   OwningPtr<MachOObject> InputObject(
 375     MachOObject::LoadFromBuffer(InputBuffer.take(), &ErrorStr));
 376   if (!InputObject)
 377     return Error("unable to load object: '" + ErrorStr + "'");
 378
 379   if (int Res = DumpHeader(*InputObject))
 380     return Res;
 381
 382   // Print the load commands.
 383   int Res = 0;
 384   outs() << "('load_commands', [\n";
 385   for (unsigned i = 0; i != InputObject->getHeader().NumLoadCommands; ++i)
 386     if ((Res = DumpLoadCommand(*InputObject, i)))
 387       break;
 388   outs() << "])\n";
 389
 390   return Res;
 391 }