tools/llvm-objdump/MachODump.cpp

   1 //===-- MachODump.cpp - Object file dumping utility for llvm --------------===//
   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 file implements the MachO-specific dumper for llvm-objdump.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #include "llvm-objdump.h"
  15 #include "MCFunction.h"
  16 #include "llvm/Support/MachO.h"
  17 #include "llvm/Object/MachOObject.h"
  18 #include "llvm/ADT/OwningPtr.h"
  19 #include "llvm/ADT/Triple.h"
  20 #include "llvm/ADT/STLExtras.h"
  21 #include "llvm/MC/MCAsmInfo.h"
  22 #include "llvm/MC/MCDisassembler.h"
  23 #include "llvm/MC/MCInst.h"
  24 #include "llvm/MC/MCInstPrinter.h"
  25 #include "llvm/MC/MCInstrAnalysis.h"
  26 #include "llvm/MC/MCInstrDesc.h"
  27 #include "llvm/MC/MCInstrInfo.h"
  28 #include "llvm/MC/MCSubtargetInfo.h"
  29 #include "llvm/Support/CommandLine.h"
  30 #include "llvm/Support/Debug.h"
  31 #include "llvm/Support/Format.h"
  32 #include "llvm/Support/GraphWriter.h"
  33 #include "llvm/Support/MemoryBuffer.h"
  34 #include "llvm/Support/TargetRegistry.h"
  35 #include "llvm/Support/TargetSelect.h"
  36 #include "llvm/Support/raw_ostream.h"
  37 #include "llvm/Support/system_error.h"
  38 #include <algorithm>
  39 #include <cstring>
  40 using namespace llvm;
  41 using namespace object;
  42
  43 static cl::opt<bool>
  44   CFG("cfg", cl::desc("Create a CFG for every symbol in the object file and"
  45                       "write it to a graphviz file (MachO-only)"));
  46
  47 static const Target *GetTarget(const MachOObject *MachOObj) {
  48   // Figure out the target triple.
  49   llvm::Triple TT("unknown-unknown-unknown");
  50   switch (MachOObj->getHeader().CPUType) {
  51   case llvm::MachO::CPUTypeI386:
  52     TT.setArch(Triple::ArchType(Triple::x86));
  53     break;
  54   case llvm::MachO::CPUTypeX86_64:
  55     TT.setArch(Triple::ArchType(Triple::x86_64));
  56     break;
  57   case llvm::MachO::CPUTypeARM:
  58     TT.setArch(Triple::ArchType(Triple::arm));
  59     break;
  60   case llvm::MachO::CPUTypePowerPC:
  61     TT.setArch(Triple::ArchType(Triple::ppc));
  62     break;
  63   case llvm::MachO::CPUTypePowerPC64:
  64     TT.setArch(Triple::ArchType(Triple::ppc64));
  65     break;
  66   }
  67
  68   TripleName = TT.str();
  69
  70   // Get the target specific parser.
  71   std::string Error;
  72   const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
  73   if (TheTarget)
  74     return TheTarget;
  75
  76   errs() << "llvm-objdump: error: unable to get target for '" << TripleName
  77          << "', see --version and --triple.\n";
  78   return 0;
  79 }
  80
  81 struct Section {
  82   char Name[16];
  83   uint64_t Address;
  84   uint64_t Size;
  85   uint32_t Offset;
  86   uint32_t NumRelocs;
  87   uint64_t RelocTableOffset;
  88 };
  89
  90 struct Symbol {
  91   uint64_t Value;
  92   uint32_t StringIndex;
  93   uint8_t SectionIndex;
  94   bool operator<(const Symbol &RHS) const { return Value < RHS.Value; }
  95 };
  96
  97
  98 template <typename T>
  99 static Section copySection(const T &Sect) {
 100   Section S;
 101   memcpy(S.Name, Sect->Name, 16);
 102   S.Address = Sect->Address;
 103   S.Size = Sect->Size;
 104   S.Offset = Sect->Offset;
 105   S.NumRelocs = Sect->NumRelocationTableEntries;
 106   S.RelocTableOffset = Sect->RelocationTableOffset;
 107   return S;
 108 }
 109
 110 template <typename T>
 111 static Symbol copySymbol(const T &STE) {
 112   Symbol S;
 113   S.StringIndex = STE->StringIndex;
 114   S.SectionIndex = STE->SectionIndex;
 115   S.Value = STE->Value;
 116   return S;
 117 }
 118
 119 // Print addtitional information about an address, if available.
 120 static void DumpAddress(uint64_t Address, ArrayRef<Section> Sections,
 121                         MachOObject *MachOObj, raw_ostream &OS) {
 122   for (unsigned i = 0; i != Sections.size(); ++i) {
 123     uint64_t addr = Address-Sections[i].Address;
 124     if (Sections[i].Address <= Address &&
 125         Sections[i].Address + Sections[i].Size > Address) {
 126       StringRef bytes = MachOObj->getData(Sections[i].Offset,
 127                                           Sections[i].Size);
 128       // Print constant strings.
 129       if (!strcmp(Sections[i].Name, "__cstring"))
 130         OS << '"' << bytes.substr(addr, bytes.find('\0', addr)) << '"';
 131       // Print constant CFStrings.
 132       if (!strcmp(Sections[i].Name, "__cfstring"))
 133         OS << "@\"" << bytes.substr(addr, bytes.find('\0', addr)) << '"';
 134     }
 135   }
 136 }
 137
 138 typedef std::map<uint64_t, MCFunction*> FunctionMapTy;
 139 typedef SmallVector<MCFunction, 16> FunctionListTy;
 140 static void createMCFunctionAndSaveCalls(StringRef Name,
 141                                          const MCDisassembler *DisAsm,
 142                                          MemoryObject &Object, uint64_t Start,
 143                                          uint64_t End,
 144                                          MCInstrAnalysis *InstrAnalysis,
 145                                          uint64_t Address,
 146                                          raw_ostream &DebugOut,
 147                                          FunctionMapTy &FunctionMap,
 148                                          FunctionListTy &Functions) {
 149   SmallVector<uint64_t, 16> Calls;
 150   MCFunction f =
 151     MCFunction::createFunctionFromMC(Name, DisAsm, Object, Start, End,
 152                                      InstrAnalysis, DebugOut, Calls);
 153   Functions.push_back(f);
 154   FunctionMap[Address] = &Functions.back();
 155
 156   // Add the gathered callees to the map.
 157   for (unsigned i = 0, e = Calls.size(); i != e; ++i)
 158     FunctionMap.insert(std::make_pair(Calls[i], (MCFunction*)0));
 159 }
 160
 161 // Write a graphviz file for the CFG inside an MCFunction.
 162 static void emitDOTFile(const char *FileName, const MCFunction &f,
 163                         MCInstPrinter *IP) {
 164   // Start a new dot file.
 165   std::string Error;
 166   raw_fd_ostream Out(FileName, Error);
 167   if (!Error.empty()) {
 168     errs() << "llvm-objdump: warning: " << Error << '\n';
 169     return;
 170   }
 171
 172   Out << "digraph " << f.getName() << " {\n";
 173   Out << "graph [ rankdir = \"LR\" ];\n";
 174   for (MCFunction::iterator i = f.begin(), e = f.end(); i != e; ++i) {
 175     bool hasPreds = false;
 176     // Only print blocks that have predecessors.
 177     // FIXME: Slow.
 178     for (MCFunction::iterator pi = f.begin(), pe = f.end(); pi != pe;
 179         ++pi)
 180       if (pi->second.contains(i->first)) {
 181         hasPreds = true;
 182         break;
 183       }
 184
 185     if (!hasPreds && i != f.begin())
 186       continue;
 187
 188     Out << '"' << i->first << "\" [ label=\"<a>";
 189     // Print instructions.
 190     for (unsigned ii = 0, ie = i->second.getInsts().size(); ii != ie;
 191         ++ii) {
 192       // Escape special chars and print the instruction in mnemonic form.
 193       std::string Str;
 194       raw_string_ostream OS(Str);
 195       IP->printInst(&i->second.getInsts()[ii].Inst, OS, "");
 196       Out << DOT::EscapeString(OS.str()) << '|';
 197     }
 198     Out << "<o>\" shape=\"record\" ];\n";
 199
 200     // Add edges.
 201     for (MCBasicBlock::succ_iterator si = i->second.succ_begin(),
 202         se = i->second.succ_end(); si != se; ++si)
 203       Out << i->first << ":o -> " << *si <<":a\n";
 204   }
 205   Out << "}\n";
 206 }
 207
 208 void llvm::DisassembleInputMachO(StringRef Filename) {
 209   OwningPtr<MemoryBuffer> Buff;
 210
 211   if (error_code ec = MemoryBuffer::getFileOrSTDIN(Filename, Buff)) {
 212     errs() << "llvm-objdump: " << Filename << ": " << ec.message() << "\n";
 213     return;
 214   }
 215
 216   OwningPtr<MachOObject> MachOObj(MachOObject::LoadFromBuffer(Buff.take()));
 217
 218   const Target *TheTarget = GetTarget(MachOObj.get());
 219   if (!TheTarget) {
 220     // GetTarget prints out stuff.
 221     return;
 222   }
 223   const MCInstrInfo *InstrInfo = TheTarget->createMCInstrInfo();
 224   OwningPtr<MCInstrAnalysis>
 225     InstrAnalysis(TheTarget->createMCInstrAnalysis(InstrInfo));
 226
 227   // Set up disassembler.
 228   OwningPtr<const MCAsmInfo> AsmInfo(TheTarget->createMCAsmInfo(TripleName));
 229   OwningPtr<const MCSubtargetInfo>
 230     STI(TheTarget->createMCSubtargetInfo(TripleName, "", ""));
 231   OwningPtr<const MCDisassembler> DisAsm(TheTarget->createMCDisassembler(*STI));
 232   int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
 233   OwningPtr<MCInstPrinter> IP(TheTarget->createMCInstPrinter(
 234                               AsmPrinterVariant, *AsmInfo, *STI));
 235
 236   if (!InstrAnalysis || !AsmInfo || !STI || !DisAsm || !IP) {
 237     errs() << "error: couldn't initialize disassmbler for target "
 238            << TripleName << '\n';
 239     return;
 240   }
 241
 242   outs() << '\n' << Filename << ":\n\n";
 243
 244   const macho::Header &Header = MachOObj->getHeader();
 245
 246   const MachOObject::LoadCommandInfo *SymtabLCI = 0;
 247   // First, find the symbol table segment.
 248   for (unsigned i = 0; i != Header.NumLoadCommands; ++i) {
 249     const MachOObject::LoadCommandInfo &LCI = MachOObj->getLoadCommandInfo(i);
 250     if (LCI.Command.Type == macho::LCT_Symtab) {
 251       SymtabLCI = &LCI;
 252       break;
 253     }
 254   }
 255
 256   // Read and register the symbol table data.
 257   InMemoryStruct<macho::SymtabLoadCommand> SymtabLC;
 258   MachOObj->ReadSymtabLoadCommand(*SymtabLCI, SymtabLC);
 259   MachOObj->RegisterStringTable(*SymtabLC);
 260
 261   std::vector<Section> Sections;
 262   std::vector<Symbol> Symbols;
 263   std::vector<Symbol> UnsortedSymbols; // FIXME: duplication
 264   SmallVector<uint64_t, 8> FoundFns;
 265
 266   // Make a list of all symbols in the object file.
 267   for (unsigned i = 0; i != Header.NumLoadCommands; ++i) {
 268     const MachOObject::LoadCommandInfo &LCI = MachOObj->getLoadCommandInfo(i);
 269     if (LCI.Command.Type == macho::LCT_Segment) {
 270       InMemoryStruct<macho::SegmentLoadCommand> SegmentLC;
 271       MachOObj->ReadSegmentLoadCommand(LCI, SegmentLC);
 272
 273       // Store the sections in this segment.
 274       for (unsigned SectNum = 0; SectNum != SegmentLC->NumSections; ++SectNum) {
 275         InMemoryStruct<macho::Section> Sect;
 276         MachOObj->ReadSection(LCI, SectNum, Sect);
 277         Sections.push_back(copySection(Sect));
 278
 279         // Store the symbols in this section.
 280         for (unsigned i = 0; i != SymtabLC->NumSymbolTableEntries; ++i) {
 281           InMemoryStruct<macho::SymbolTableEntry> STE;
 282           MachOObj->ReadSymbolTableEntry(SymtabLC->SymbolTableOffset, i, STE);
 283           Symbols.push_back(copySymbol(STE));
 284           UnsortedSymbols.push_back(Symbols.back());
 285         }
 286       }
 287     } else if (LCI.Command.Type == macho::LCT_Segment64) {
 288       InMemoryStruct<macho::Segment64LoadCommand> Segment64LC;
 289       MachOObj->ReadSegment64LoadCommand(LCI, Segment64LC);
 290
 291       // Store the sections in this segment.
 292       for (unsigned SectNum = 0; SectNum != Segment64LC->NumSections;
 293            ++SectNum) {
 294         InMemoryStruct<macho::Section64> Sect64;
 295         MachOObj->ReadSection64(LCI, SectNum, Sect64);
 296         Sections.push_back(copySection(Sect64));
 297
 298         // Store the symbols in this section.
 299         for (unsigned i = 0; i != SymtabLC->NumSymbolTableEntries; ++i) {
 300           InMemoryStruct<macho::Symbol64TableEntry> STE;
 301           MachOObj->ReadSymbol64TableEntry(SymtabLC->SymbolTableOffset, i, STE);
 302           Symbols.push_back(copySymbol(STE));
 303           UnsortedSymbols.push_back(Symbols.back());
 304         }
 305       }
 306     } else if (LCI.Command.Type == macho::LCT_FunctionStarts) {
 307       // We found a function starts segment, parse the addresses for later
 308       // consumption.
 309       InMemoryStruct<macho::LinkeditDataLoadCommand> LLC;
 310       MachOObj->ReadLinkeditDataLoadCommand(LCI, LLC);
 311
 312       MachOObj->ReadULEB128s(LLC->DataOffset, FoundFns);
 313     }
 314   }
 315
 316
 317   // Sort the symbols by address, just in case they didn't come in that way.
 318   array_pod_sort(Symbols.begin(), Symbols.end());
 319
 320 #ifndef NDEBUG
 321   raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls();
 322 #else
 323   raw_ostream &DebugOut = nulls();
 324 #endif
 325
 326   FunctionMapTy FunctionMap;
 327   FunctionListTy Functions;
 328
 329   for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) {
 330     if (strcmp(Sections[SectIdx].Name, "__text"))
 331       continue; // Skip non-text sections
 332
 333     // Insert the functions from the function starts segment into our map.
 334     uint64_t VMAddr = Sections[SectIdx].Address - Sections[SectIdx].Offset;
 335     for (unsigned i = 0, e = FoundFns.size(); i != e; ++i)
 336       FunctionMap.insert(std::make_pair(FoundFns[i]+VMAddr, (MCFunction*)0));
 337
 338     StringRef Bytes = MachOObj->getData(Sections[SectIdx].Offset,
 339                                         Sections[SectIdx].Size);
 340     StringRefMemoryObject memoryObject(Bytes);
 341     bool symbolTableWorked = false;
 342
 343     // Parse relocations.
 344     std::vector<std::pair<uint64_t, uint32_t> > Relocs;
 345     for (unsigned j = 0; j != Sections[SectIdx].NumRelocs; ++j) {
 346       InMemoryStruct<macho::RelocationEntry> RE;
 347       MachOObj->ReadRelocationEntry(Sections[SectIdx].RelocTableOffset, j, RE);
 348       Relocs.push_back(std::make_pair(RE->Word0, RE->Word1 & 0xffffff));
 349     }
 350     array_pod_sort(Relocs.begin(), Relocs.end());
 351
 352     // Disassemble symbol by symbol.
 353     for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) {
 354       // Make sure the symbol is defined in this section.
 355       if ((unsigned)Symbols[SymIdx].SectionIndex - 1 != SectIdx)
 356         continue;
 357
 358       // Start at the address of the symbol relative to the section's address.
 359       uint64_t Start = Symbols[SymIdx].Value - Sections[SectIdx].Address;
 360       // Stop disassembling either at the beginning of the next symbol or at
 361       // the end of the section.
 362       uint64_t End = (SymIdx+1 == Symbols.size() ||
 363           Symbols[SymIdx].SectionIndex != Symbols[SymIdx+1].SectionIndex) ?
 364           Sections[SectIdx].Size :
 365           Symbols[SymIdx+1].Value - Sections[SectIdx].Address;
 366       uint64_t Size;
 367
 368       if (Start >= End)
 369         continue;
 370
 371       symbolTableWorked = true;
 372
 373       if (!CFG) {
 374         // Normal disassembly, print addresses, bytes and mnemonic form.
 375         outs() << MachOObj->getStringAtIndex(Symbols[SymIdx].StringIndex)
 376           << ":\n";
 377         for (uint64_t Index = Start; Index < End; Index += Size) {
 378           MCInst Inst;
 379
 380           if (DisAsm->getInstruction(Inst, Size, memoryObject, Index,
 381                                      DebugOut, nulls())) {
 382             outs() << format("%8llx:\t", Sections[SectIdx].Address + Index);
 383             DumpBytes(StringRef(Bytes.data() + Index, Size));
 384             IP->printInst(&Inst, outs(), "");
 385             outs() << "\n";
 386           } else {
 387             errs() << "llvm-objdump: warning: invalid instruction encoding\n";
 388             if (Size == 0)
 389               Size = 1; // skip illegible bytes
 390           }
 391         }
 392       } else {
 393         // Create CFG and use it for disassembly.
 394         createMCFunctionAndSaveCalls(
 395             MachOObj->getStringAtIndex(Symbols[SymIdx].StringIndex),
 396             DisAsm.get(), memoryObject, Start, End, InstrAnalysis.get(),
 397             Start, DebugOut, FunctionMap, Functions);
 398       }
 399     }
 400
 401     if (CFG) {
 402       if (!symbolTableWorked) {
 403         // Reading the symbol table didn't work, create a big __TEXT symbol.
 404         createMCFunctionAndSaveCalls("__TEXT", DisAsm.get(), memoryObject,
 405                                      0, Sections[SectIdx].Size,
 406                                      InstrAnalysis.get(),
 407                                      Sections[SectIdx].Offset, DebugOut,
 408                                      FunctionMap, Functions);
 409       }
 410       for (std::map<uint64_t, MCFunction*>::iterator mi = FunctionMap.begin(),
 411            me = FunctionMap.end(); mi != me; ++mi)
 412         if (mi->second == 0) {
 413           // Create functions for the remaining callees we have gathered,
 414           // but we didn't find a name for them.
 415           SmallVector<uint64_t, 16> Calls;
 416           MCFunction f =
 417             MCFunction::createFunctionFromMC("unknown", DisAsm.get(),
 418                                              memoryObject, mi->first,
 419                                              Sections[SectIdx].Size,
 420                                              InstrAnalysis.get(), DebugOut,
 421                                              Calls);
 422           Functions.push_back(f);
 423           mi->second = &Functions.back();
 424           for (unsigned i = 0, e = Calls.size(); i != e; ++i) {
 425             std::pair<uint64_t, MCFunction*> p(Calls[i], (MCFunction*)0);
 426             if (FunctionMap.insert(p).second)
 427               mi = FunctionMap.begin();
 428           }
 429         }
 430
 431       DenseSet<uint64_t> PrintedBlocks;
 432       for (unsigned ffi = 0, ffe = Functions.size(); ffi != ffe; ++ffi) {
 433         MCFunction &f = Functions[ffi];
 434         for (MCFunction::iterator fi = f.begin(), fe = f.end(); fi != fe; ++fi){
 435           if (!PrintedBlocks.insert(fi->first).second)
 436             continue; // We already printed this block.
 437
 438           // We assume a block has predecessors when it's the first block after
 439           // a symbol.
 440           bool hasPreds = FunctionMap.find(fi->first) != FunctionMap.end();
 441
 442           // See if this block has predecessors.
 443           // FIXME: Slow.
 444           for (MCFunction::iterator pi = f.begin(), pe = f.end(); pi != pe;
 445               ++pi)
 446             if (pi->second.contains(fi->first)) {
 447               hasPreds = true;
 448               break;
 449             }
 450
 451           // No predecessors, this is a data block. Print as .byte directives.
 452           if (!hasPreds) {
 453             uint64_t End = llvm::next(fi) == fe ? Sections[SectIdx].Size :
 454                                                   llvm::next(fi)->first;
 455             outs() << "# " << End-fi->first << " bytes of data:\n";
 456             for (unsigned pos = fi->first; pos != End; ++pos) {
 457               outs() << format("%8x:\t", Sections[SectIdx].Address + pos);
 458               DumpBytes(StringRef(Bytes.data() + pos, 1));
 459               outs() << format("\t.byte 0x%02x\n", (uint8_t)Bytes[pos]);
 460             }
 461             continue;
 462           }
 463
 464           if (fi->second.contains(fi->first)) // Print a header for simple loops
 465             outs() << "# Loop begin:\n";
 466
 467           // Walk over the instructions and print them.
 468           for (unsigned ii = 0, ie = fi->second.getInsts().size(); ii != ie;
 469                ++ii) {
 470             const MCDecodedInst &Inst = fi->second.getInsts()[ii];
 471
 472             // If there's a symbol at this address, print its name.
 473             if (FunctionMap.find(Sections[SectIdx].Address + Inst.Address) !=
 474                 FunctionMap.end())
 475               outs() << FunctionMap[Sections[SectIdx].Address + Inst.Address]->
 476                                                              getName() << ":\n";
 477
 478             outs() << format("%8llx:\t", Sections[SectIdx].Address +
 479                                          Inst.Address);
 480             DumpBytes(StringRef(Bytes.data() + Inst.Address, Inst.Size));
 481
 482             if (fi->second.contains(fi->first)) // Indent simple loops.
 483               outs() << '\t';
 484
 485             IP->printInst(&Inst.Inst, outs(), "");
 486
 487             // Look for relocations inside this instructions, if there is one
 488             // print its target and additional information if availbable.
 489             for (unsigned j = 0; j != Relocs.size(); ++j)
 490               if (Relocs[j].first >= Sections[SectIdx].Address + Inst.Address &&
 491                   Relocs[j].first < Sections[SectIdx].Address + Inst.Address +
 492                                     Inst.Size) {
 493                 outs() << "\t# "
 494                    << MachOObj->getStringAtIndex(
 495                                   UnsortedSymbols[Relocs[j].second].StringIndex)
 496                    << ' ';
 497                 DumpAddress(UnsortedSymbols[Relocs[j].second].Value, Sections,
 498                             MachOObj.get(), outs());
 499               }
 500
 501             // If this instructions contains an address, see if we can evaluate
 502             // it and print additional information.
 503             uint64_t targ = InstrAnalysis->evaluateBranch(Inst.Inst,
 504                                                           Inst.Address,
 505                                                           Inst.Size);
 506             if (targ != -1ULL)
 507               DumpAddress(targ, Sections, MachOObj.get(), outs());
 508
 509             outs() << '\n';
 510           }
 511         }
 512
 513         emitDOTFile((f.getName().str() + ".dot").c_str(), f, IP.get());
 514       }
 515     }
 516   }
 517 }