MachODump.cpp revision 54154f3bf1ae3d2dfd68cc9474cad061b3338a40 
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 "llvm/ADT/OwningPtr.h"
16#include "llvm/ADT/STLExtras.h"
17#include "llvm/ADT/StringExtras.h"
18#include "llvm/ADT/Triple.h"
19#include "llvm/DebugInfo/DIContext.h"
20#include "llvm/MC/MCAsmInfo.h"
21#include "llvm/MC/MCDisassembler.h"
22#include "llvm/MC/MCInst.h"
23#include "llvm/MC/MCInstPrinter.h"
24#include "llvm/MC/MCInstrAnalysis.h"
25#include "llvm/MC/MCInstrDesc.h"
26#include "llvm/MC/MCInstrInfo.h"
27#include "llvm/MC/MCRegisterInfo.h"
28#include "llvm/MC/MCSubtargetInfo.h"
29#include "llvm/Object/MachO.h"
30#include "llvm/Support/Casting.h"
31#include "llvm/Support/CommandLine.h"
32#include "llvm/Support/Debug.h"
33#include "llvm/Support/Format.h"
34#include "llvm/Support/GraphWriter.h"
35#include "llvm/Support/MachO.h"
36#include "llvm/Support/MemoryBuffer.h"
37#include "llvm/Support/TargetRegistry.h"
38#include "llvm/Support/TargetSelect.h"
39#include "llvm/Support/raw_ostream.h"
40#include "llvm/Support/system_error.h"
41#include <algorithm>
42#include <cstring>
43using namespace llvm;
44using namespace object;
45
46static cl::opt<bool>
47  UseDbg("g", cl::desc("Print line information from debug info if available"));
48
49static cl::opt<std::string>
50  DSYMFile("dsym", cl::desc("Use .dSYM file for debug info"));
51
52static const Target *GetTarget(const MachOObjectFile *MachOObj) {
53  // Figure out the target triple.
54  if (TripleName.empty()) {
55    llvm::Triple TT("unknown-unknown-unknown");
56    TT.setArch(Triple::ArchType(MachOObj->getArch()));
57    TripleName = TT.str();
58  }
59
60  // Get the target specific parser.
61  std::string Error;
62  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
63  if (TheTarget)
64    return TheTarget;
65
66  errs() << "llvm-objdump: error: unable to get target for '" << TripleName
67         << "', see --version and --triple.\n";
68  return 0;
69}
70
71struct SymbolSorter {
72  bool operator()(const SymbolRef &A, const SymbolRef &B) {
73    SymbolRef::Type AType, BType;
74    A.getType(AType);
75    B.getType(BType);
76
77    uint64_t AAddr, BAddr;
78    if (AType != SymbolRef::ST_Function)
79      AAddr = 0;
80    else
81      A.getAddress(AAddr);
82    if (BType != SymbolRef::ST_Function)
83      BAddr = 0;
84    else
85      B.getAddress(BAddr);
86    return AAddr < BAddr;
87  }
88};
89
90// Types for the storted data in code table that is built before disassembly
91// and the predicate function to sort them.
92typedef std::pair<uint64_t, DiceRef> DiceTableEntry;
93typedef std::vector<DiceTableEntry> DiceTable;
94typedef DiceTable::iterator dice_table_iterator;
95
96static bool
97compareDiceTableEntries(const DiceTableEntry i,
98                        const DiceTableEntry j) {
99  return i.first == j.first;
100}
101
102static void DumpDataInCode(const char *bytes, uint64_t Size,
103                           unsigned short Kind) {
104  uint64_t Value;
105
106  switch (Kind) {
107  case macho::Data:
108    switch (Size) {
109    case 4:
110      Value = bytes[3] << 24 |
111              bytes[2] << 16 |
112              bytes[1] << 8 |
113              bytes[0];
114      outs() << "\t.long " << Value;
115      break;
116    case 2:
117      Value = bytes[1] << 8 |
118              bytes[0];
119      outs() << "\t.short " << Value;
120      break;
121    case 1:
122      Value = bytes[0];
123      outs() << "\t.byte " << Value;
124      break;
125    }
126    outs() << "\t@ KIND_DATA\n";
127    break;
128  case macho::JumpTable8:
129    Value = bytes[0];
130    outs() << "\t.byte " << Value << "\t@ KIND_JUMP_TABLE8";
131    break;
132  case macho::JumpTable16:
133    Value = bytes[1] << 8 |
134            bytes[0];
135    outs() << "\t.short " << Value << "\t@ KIND_JUMP_TABLE16";
136    break;
137  case macho::JumpTable32:
138    Value = bytes[3] << 24 |
139            bytes[2] << 16 |
140            bytes[1] << 8 |
141            bytes[0];
142    outs() << "\t.long " << Value << "\t@ KIND_JUMP_TABLE32";
143    break;
144  default:
145    outs() << "\t@ data in code kind = " << Kind << "\n";
146    break;
147  }
148}
149
150static void
151getSectionsAndSymbols(const macho::Header Header,
152                      MachOObjectFile *MachOObj,
153                      std::vector<SectionRef> &Sections,
154                      std::vector<SymbolRef> &Symbols,
155                      SmallVectorImpl<uint64_t> &FoundFns,
156                      uint64_t &BaseSegmentAddress) {
157  error_code ec;
158  for (symbol_iterator SI = MachOObj->begin_symbols(),
159       SE = MachOObj->end_symbols(); SI != SE; SI.increment(ec))
160    Symbols.push_back(*SI);
161
162  for (section_iterator SI = MachOObj->begin_sections(),
163       SE = MachOObj->end_sections(); SI != SE; SI.increment(ec)) {
164    SectionRef SR = *SI;
165    StringRef SectName;
166    SR.getName(SectName);
167    Sections.push_back(*SI);
168  }
169
170  MachOObjectFile::LoadCommandInfo Command =
171    MachOObj->getFirstLoadCommandInfo();
172  bool BaseSegmentAddressSet = false;
173  for (unsigned i = 0; ; ++i) {
174    if (Command.C.Type == macho::LCT_FunctionStarts) {
175      // We found a function starts segment, parse the addresses for later
176      // consumption.
177      macho::LinkeditDataLoadCommand LLC =
178        MachOObj->getLinkeditDataLoadCommand(Command);
179
180      MachOObj->ReadULEB128s(LLC.DataOffset, FoundFns);
181    }
182    else if (Command.C.Type == macho::LCT_Segment) {
183      macho::SegmentLoadCommand SLC =
184        MachOObj->getSegmentLoadCommand(Command);
185      StringRef SegName = SLC.Name;
186      if(!BaseSegmentAddressSet && SegName != "__PAGEZERO") {
187        BaseSegmentAddressSet = true;
188        BaseSegmentAddress = SLC.VMAddress;
189      }
190    }
191
192    if (i == Header.NumLoadCommands - 1)
193      break;
194    else
195      Command = MachOObj->getNextLoadCommandInfo(Command);
196  }
197}
198
199static void DisassembleInputMachO2(StringRef Filename,
200                                   MachOObjectFile *MachOOF);
201
202void llvm::DisassembleInputMachO(StringRef Filename) {
203  OwningPtr<MemoryBuffer> Buff;
204
205  if (error_code ec = MemoryBuffer::getFileOrSTDIN(Filename, Buff)) {
206    errs() << "llvm-objdump: " << Filename << ": " << ec.message() << "\n";
207    return;
208  }
209
210  OwningPtr<MachOObjectFile> MachOOF(static_cast<MachOObjectFile*>(
211        ObjectFile::createMachOObjectFile(Buff.take())));
212
213  DisassembleInputMachO2(Filename, MachOOF.get());
214}
215
216static void DisassembleInputMachO2(StringRef Filename,
217                                   MachOObjectFile *MachOOF) {
218  const Target *TheTarget = GetTarget(MachOOF);
219  if (!TheTarget) {
220    // GetTarget prints out stuff.
221    return;
222  }
223  OwningPtr<const MCInstrInfo> InstrInfo(TheTarget->createMCInstrInfo());
224  OwningPtr<MCInstrAnalysis>
225    InstrAnalysis(TheTarget->createMCInstrAnalysis(InstrInfo.get()));
226
227  // Set up disassembler.
228  OwningPtr<const MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TripleName));
229  OwningPtr<const MCAsmInfo> AsmInfo(
230      TheTarget->createMCAsmInfo(*MRI, TripleName));
231  OwningPtr<const MCSubtargetInfo>
232    STI(TheTarget->createMCSubtargetInfo(TripleName, "", ""));
233  OwningPtr<const MCDisassembler> DisAsm(TheTarget->createMCDisassembler(*STI));
234  int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
235  OwningPtr<MCInstPrinter>
236    IP(TheTarget->createMCInstPrinter(AsmPrinterVariant, *AsmInfo, *InstrInfo,
237                                      *MRI, *STI));
238
239  if (!InstrAnalysis || !AsmInfo || !STI || !DisAsm || !IP) {
240    errs() << "error: couldn't initialize disassembler for target "
241           << TripleName << '\n';
242    return;
243  }
244
245  outs() << '\n' << Filename << ":\n\n";
246
247  macho::Header Header = MachOOF->getHeader();
248
249  // FIXME: FoundFns isn't used anymore. Using symbols/LC_FUNCTION_STARTS to
250  // determine function locations will eventually go in MCObjectDisassembler.
251  // FIXME: Using the -cfg command line option, this code used to be able to
252  // annotate relocations with the referenced symbol's name, and if this was
253  // inside a __[cf]string section, the data it points to. This is now replaced
254  // by the upcoming MCSymbolizer, which needs the appropriate setup done above.
255  std::vector<SectionRef> Sections;
256  std::vector<SymbolRef> Symbols;
257  SmallVector<uint64_t, 8> FoundFns;
258  uint64_t BaseSegmentAddress;
259
260  getSectionsAndSymbols(Header, MachOOF, Sections, Symbols, FoundFns,
261                        BaseSegmentAddress);
262
263  // Make a copy of the unsorted symbol list. FIXME: duplication
264  std::vector<SymbolRef> UnsortedSymbols(Symbols);
265  // Sort the symbols by address, just in case they didn't come in that way.
266  std::sort(Symbols.begin(), Symbols.end(), SymbolSorter());
267
268  // Build a data in code table that is sorted on by the address of each entry.
269  uint64_t BaseAddress = 0;
270  if (Header.FileType == macho::HFT_Object)
271    Sections[0].getAddress(BaseAddress);
272  else
273    BaseAddress = BaseSegmentAddress;
274  DiceTable Dices;
275  error_code ec;
276  for (dice_iterator DI = MachOOF->begin_dices(), DE = MachOOF->end_dices();
277       DI != DE; DI.increment(ec)){
278    uint32_t Offset;
279    DI->getOffset(Offset);
280    Dices.push_back(std::make_pair(BaseAddress + Offset, *DI));
281  }
282  array_pod_sort(Dices.begin(), Dices.end());
283
284#ifndef NDEBUG
285  raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls();
286#else
287  raw_ostream &DebugOut = nulls();
288#endif
289
290  OwningPtr<DIContext> diContext;
291  ObjectFile *DbgObj = MachOOF;
292  // Try to find debug info and set up the DIContext for it.
293  if (UseDbg) {
294    // A separate DSym file path was specified, parse it as a macho file,
295    // get the sections and supply it to the section name parsing machinery.
296    if (!DSYMFile.empty()) {
297      OwningPtr<MemoryBuffer> Buf;
298      if (error_code ec = MemoryBuffer::getFileOrSTDIN(DSYMFile.c_str(), Buf)) {
299        errs() << "llvm-objdump: " << Filename << ": " << ec.message() << '\n';
300        return;
301      }
302      DbgObj = ObjectFile::createMachOObjectFile(Buf.take());
303    }
304
305    // Setup the DIContext
306    diContext.reset(DIContext::getDWARFContext(DbgObj));
307  }
308
309  for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) {
310
311    bool SectIsText = false;
312    Sections[SectIdx].isText(SectIsText);
313    if (SectIsText == false)
314      continue;
315
316    StringRef SectName;
317    if (Sections[SectIdx].getName(SectName) ||
318        SectName != "__text")
319      continue; // Skip non-text sections
320
321    DataRefImpl DR = Sections[SectIdx].getRawDataRefImpl();
322
323    StringRef SegmentName = MachOOF->getSectionFinalSegmentName(DR);
324    if (SegmentName != "__TEXT")
325      continue;
326
327    StringRef Bytes;
328    Sections[SectIdx].getContents(Bytes);
329    StringRefMemoryObject memoryObject(Bytes);
330    bool symbolTableWorked = false;
331
332    // Parse relocations.
333    std::vector<std::pair<uint64_t, SymbolRef> > Relocs;
334    error_code ec;
335    for (relocation_iterator RI = Sections[SectIdx].begin_relocations(),
336         RE = Sections[SectIdx].end_relocations(); RI != RE; RI.increment(ec)) {
337      uint64_t RelocOffset, SectionAddress;
338      RI->getOffset(RelocOffset);
339      Sections[SectIdx].getAddress(SectionAddress);
340      RelocOffset -= SectionAddress;
341
342      symbol_iterator RelocSym = RI->getSymbol();
343
344      Relocs.push_back(std::make_pair(RelocOffset, *RelocSym));
345    }
346    array_pod_sort(Relocs.begin(), Relocs.end());
347
348    // Disassemble symbol by symbol.
349    for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) {
350      StringRef SymName;
351      Symbols[SymIdx].getName(SymName);
352
353      SymbolRef::Type ST;
354      Symbols[SymIdx].getType(ST);
355      if (ST != SymbolRef::ST_Function)
356        continue;
357
358      // Make sure the symbol is defined in this section.
359      bool containsSym = false;
360      Sections[SectIdx].containsSymbol(Symbols[SymIdx], containsSym);
361      if (!containsSym)
362        continue;
363
364      // Start at the address of the symbol relative to the section's address.
365      uint64_t SectionAddress = 0;
366      uint64_t Start = 0;
367      Sections[SectIdx].getAddress(SectionAddress);
368      Symbols[SymIdx].getAddress(Start);
369      Start -= SectionAddress;
370
371      // Stop disassembling either at the beginning of the next symbol or at
372      // the end of the section.
373      bool containsNextSym = false;
374      uint64_t NextSym = 0;
375      uint64_t NextSymIdx = SymIdx+1;
376      while (Symbols.size() > NextSymIdx) {
377        SymbolRef::Type NextSymType;
378        Symbols[NextSymIdx].getType(NextSymType);
379        if (NextSymType == SymbolRef::ST_Function) {
380          Sections[SectIdx].containsSymbol(Symbols[NextSymIdx],
381                                           containsNextSym);
382          Symbols[NextSymIdx].getAddress(NextSym);
383          NextSym -= SectionAddress;
384          break;
385        }
386        ++NextSymIdx;
387      }
388
389      uint64_t SectSize;
390      Sections[SectIdx].getSize(SectSize);
391      uint64_t End = containsNextSym ?  NextSym : SectSize;
392      uint64_t Size;
393
394      symbolTableWorked = true;
395
396      outs() << SymName << ":\n";
397      DILineInfo lastLine;
398      for (uint64_t Index = Start; Index < End; Index += Size) {
399        MCInst Inst;
400
401        uint64_t SectAddress = 0;
402        Sections[SectIdx].getAddress(SectAddress);
403        outs() << format("%8" PRIx64 ":\t", SectAddress + Index);
404
405        // Check the data in code table here to see if this is data not an
406        // instruction to be disassembled.
407        DiceTable Dice;
408        Dice.push_back(std::make_pair(SectAddress + Index, DiceRef()));
409        dice_table_iterator DTI = std::search(Dices.begin(), Dices.end(),
410                                              Dice.begin(), Dice.end(),
411                                              compareDiceTableEntries);
412        if (DTI != Dices.end()){
413          uint16_t Length;
414          DTI->second.getLength(Length);
415          DumpBytes(StringRef(Bytes.data() + Index, Length));
416          uint16_t Kind;
417          DTI->second.getKind(Kind);
418          DumpDataInCode(Bytes.data() + Index, Length, Kind);
419          continue;
420        }
421
422        if (DisAsm->getInstruction(Inst, Size, memoryObject, Index,
423                                   DebugOut, nulls())) {
424          DumpBytes(StringRef(Bytes.data() + Index, Size));
425          IP->printInst(&Inst, outs(), "");
426
427          // Print debug info.
428          if (diContext) {
429            DILineInfo dli =
430              diContext->getLineInfoForAddress(SectAddress + Index);
431            // Print valid line info if it changed.
432            if (dli != lastLine && dli.getLine() != 0)
433              outs() << "\t## " << dli.getFileName() << ':'
434                << dli.getLine() << ':' << dli.getColumn();
435            lastLine = dli;
436          }
437          outs() << "\n";
438        } else {
439          errs() << "llvm-objdump: warning: invalid instruction encoding\n";
440          if (Size == 0)
441            Size = 1; // skip illegible bytes
442        }
443      }
444    }
445    if (!symbolTableWorked) {
446      // Reading the symbol table didn't work, disassemble the whole section.
447      uint64_t SectAddress;
448      Sections[SectIdx].getAddress(SectAddress);
449      uint64_t SectSize;
450      Sections[SectIdx].getSize(SectSize);
451      uint64_t InstSize;
452      for (uint64_t Index = 0; Index < SectSize; Index += InstSize) {
453        MCInst Inst;
454
455        if (DisAsm->getInstruction(Inst, InstSize, memoryObject, Index,
456                                   DebugOut, nulls())) {
457          outs() << format("%8" PRIx64 ":\t", SectAddress + Index);
458          DumpBytes(StringRef(Bytes.data() + Index, InstSize));
459          IP->printInst(&Inst, outs(), "");
460          outs() << "\n";
461        } else {
462          errs() << "llvm-objdump: warning: invalid instruction encoding\n";
463          if (InstSize == 0)
464            InstSize = 1; // skip illegible bytes
465        }
466      }
467    }
468  }
469}
470