MachODump.cpp revision 36b56886974eae4f9c5ebc96befd3e7bfe5de338 
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/STLExtras.h"
16#include "llvm/ADT/StringExtras.h"
17#include "llvm/ADT/Triple.h"
18#include "llvm/DebugInfo/DIContext.h"
19#include "llvm/MC/MCAsmInfo.h"
20#include "llvm/MC/MCDisassembler.h"
21#include "llvm/MC/MCInst.h"
22#include "llvm/MC/MCInstPrinter.h"
23#include "llvm/MC/MCInstrAnalysis.h"
24#include "llvm/MC/MCInstrDesc.h"
25#include "llvm/MC/MCInstrInfo.h"
26#include "llvm/MC/MCRegisterInfo.h"
27#include "llvm/MC/MCSubtargetInfo.h"
28#include "llvm/Object/MachO.h"
29#include "llvm/Support/Casting.h"
30#include "llvm/Support/CommandLine.h"
31#include "llvm/Support/Debug.h"
32#include "llvm/Support/Format.h"
33#include "llvm/Support/GraphWriter.h"
34#include "llvm/Support/MachO.h"
35#include "llvm/Support/MemoryBuffer.h"
36#include "llvm/Support/TargetRegistry.h"
37#include "llvm/Support/TargetSelect.h"
38#include "llvm/Support/raw_ostream.h"
39#include "llvm/Support/system_error.h"
40#include <algorithm>
41#include <cstring>
42using namespace llvm;
43using namespace object;
44
45static cl::opt<bool>
46  UseDbg("g", cl::desc("Print line information from debug info if available"));
47
48static cl::opt<std::string>
49  DSYMFile("dsym", cl::desc("Use .dSYM file for debug info"));
50
51static const Target *GetTarget(const MachOObjectFile *MachOObj) {
52  // Figure out the target triple.
53  if (TripleName.empty()) {
54    llvm::Triple TT("unknown-unknown-unknown");
55    TT.setArch(Triple::ArchType(MachOObj->getArch()));
56    TripleName = TT.str();
57  }
58
59  // Get the target specific parser.
60  std::string Error;
61  const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error);
62  if (TheTarget)
63    return TheTarget;
64
65  errs() << "llvm-objdump: error: unable to get target for '" << TripleName
66         << "', see --version and --triple.\n";
67  return 0;
68}
69
70struct SymbolSorter {
71  bool operator()(const SymbolRef &A, const SymbolRef &B) {
72    SymbolRef::Type AType, BType;
73    A.getType(AType);
74    B.getType(BType);
75
76    uint64_t AAddr, BAddr;
77    if (AType != SymbolRef::ST_Function)
78      AAddr = 0;
79    else
80      A.getAddress(AAddr);
81    if (BType != SymbolRef::ST_Function)
82      BAddr = 0;
83    else
84      B.getAddress(BAddr);
85    return AAddr < BAddr;
86  }
87};
88
89// Types for the storted data in code table that is built before disassembly
90// and the predicate function to sort them.
91typedef std::pair<uint64_t, DiceRef> DiceTableEntry;
92typedef std::vector<DiceTableEntry> DiceTable;
93typedef DiceTable::iterator dice_table_iterator;
94
95static bool
96compareDiceTableEntries(const DiceTableEntry i,
97                        const DiceTableEntry j) {
98  return i.first == j.first;
99}
100
101static void DumpDataInCode(const char *bytes, uint64_t Size,
102                           unsigned short Kind) {
103  uint64_t Value;
104
105  switch (Kind) {
106  case MachO::DICE_KIND_DATA:
107    switch (Size) {
108    case 4:
109      Value = bytes[3] << 24 |
110              bytes[2] << 16 |
111              bytes[1] << 8 |
112              bytes[0];
113      outs() << "\t.long " << Value;
114      break;
115    case 2:
116      Value = bytes[1] << 8 |
117              bytes[0];
118      outs() << "\t.short " << Value;
119      break;
120    case 1:
121      Value = bytes[0];
122      outs() << "\t.byte " << Value;
123      break;
124    }
125    outs() << "\t@ KIND_DATA\n";
126    break;
127  case MachO::DICE_KIND_JUMP_TABLE8:
128    Value = bytes[0];
129    outs() << "\t.byte " << Value << "\t@ KIND_JUMP_TABLE8";
130    break;
131  case MachO::DICE_KIND_JUMP_TABLE16:
132    Value = bytes[1] << 8 |
133            bytes[0];
134    outs() << "\t.short " << Value << "\t@ KIND_JUMP_TABLE16";
135    break;
136  case MachO::DICE_KIND_JUMP_TABLE32:
137    Value = bytes[3] << 24 |
138            bytes[2] << 16 |
139            bytes[1] << 8 |
140            bytes[0];
141    outs() << "\t.long " << Value << "\t@ KIND_JUMP_TABLE32";
142    break;
143  default:
144    outs() << "\t@ data in code kind = " << Kind << "\n";
145    break;
146  }
147}
148
149static void getSectionsAndSymbols(const MachO::mach_header Header,
150                                  MachOObjectFile *MachOObj,
151                                  std::vector<SectionRef> &Sections,
152                                  std::vector<SymbolRef> &Symbols,
153                                  SmallVectorImpl<uint64_t> &FoundFns,
154                                  uint64_t &BaseSegmentAddress) {
155  for (const SymbolRef &Symbol : MachOObj->symbols())
156    Symbols.push_back(Symbol);
157
158  for (const SectionRef &Section : MachOObj->sections()) {
159    StringRef SectName;
160    Section.getName(SectName);
161    Sections.push_back(Section);
162  }
163
164  MachOObjectFile::LoadCommandInfo Command =
165      MachOObj->getFirstLoadCommandInfo();
166  bool BaseSegmentAddressSet = false;
167  for (unsigned i = 0; ; ++i) {
168    if (Command.C.cmd == MachO::LC_FUNCTION_STARTS) {
169      // We found a function starts segment, parse the addresses for later
170      // consumption.
171      MachO::linkedit_data_command LLC =
172        MachOObj->getLinkeditDataLoadCommand(Command);
173
174      MachOObj->ReadULEB128s(LLC.dataoff, FoundFns);
175    }
176    else if (Command.C.cmd == MachO::LC_SEGMENT) {
177      MachO::segment_command SLC =
178        MachOObj->getSegmentLoadCommand(Command);
179      StringRef SegName = SLC.segname;
180      if(!BaseSegmentAddressSet && SegName != "__PAGEZERO") {
181        BaseSegmentAddressSet = true;
182        BaseSegmentAddress = SLC.vmaddr;
183      }
184    }
185
186    if (i == Header.ncmds - 1)
187      break;
188    else
189      Command = MachOObj->getNextLoadCommandInfo(Command);
190  }
191}
192
193static void DisassembleInputMachO2(StringRef Filename,
194                                   MachOObjectFile *MachOOF);
195
196void llvm::DisassembleInputMachO(StringRef Filename) {
197  std::unique_ptr<MemoryBuffer> Buff;
198
199  if (error_code ec = MemoryBuffer::getFileOrSTDIN(Filename, Buff)) {
200    errs() << "llvm-objdump: " << Filename << ": " << ec.message() << "\n";
201    return;
202  }
203
204  std::unique_ptr<MachOObjectFile> MachOOF(static_cast<MachOObjectFile *>(
205      ObjectFile::createMachOObjectFile(Buff.release()).get()));
206
207  DisassembleInputMachO2(Filename, MachOOF.get());
208}
209
210static void DisassembleInputMachO2(StringRef Filename,
211                                   MachOObjectFile *MachOOF) {
212  const Target *TheTarget = GetTarget(MachOOF);
213  if (!TheTarget) {
214    // GetTarget prints out stuff.
215    return;
216  }
217  std::unique_ptr<const MCInstrInfo> InstrInfo(TheTarget->createMCInstrInfo());
218  std::unique_ptr<MCInstrAnalysis> InstrAnalysis(
219      TheTarget->createMCInstrAnalysis(InstrInfo.get()));
220
221  // Set up disassembler.
222  std::unique_ptr<const MCRegisterInfo> MRI(
223      TheTarget->createMCRegInfo(TripleName));
224  std::unique_ptr<const MCAsmInfo> AsmInfo(
225      TheTarget->createMCAsmInfo(*MRI, TripleName));
226  std::unique_ptr<const MCSubtargetInfo> STI(
227      TheTarget->createMCSubtargetInfo(TripleName, "", ""));
228  std::unique_ptr<const MCDisassembler> DisAsm(
229      TheTarget->createMCDisassembler(*STI));
230  int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
231  std::unique_ptr<MCInstPrinter> IP(TheTarget->createMCInstPrinter(
232      AsmPrinterVariant, *AsmInfo, *InstrInfo, *MRI, *STI));
233
234  if (!InstrAnalysis || !AsmInfo || !STI || !DisAsm || !IP) {
235    errs() << "error: couldn't initialize disassembler for target "
236           << TripleName << '\n';
237    return;
238  }
239
240  outs() << '\n' << Filename << ":\n\n";
241
242  MachO::mach_header Header = MachOOF->getHeader();
243
244  // FIXME: FoundFns isn't used anymore. Using symbols/LC_FUNCTION_STARTS to
245  // determine function locations will eventually go in MCObjectDisassembler.
246  // FIXME: Using the -cfg command line option, this code used to be able to
247  // annotate relocations with the referenced symbol's name, and if this was
248  // inside a __[cf]string section, the data it points to. This is now replaced
249  // by the upcoming MCSymbolizer, which needs the appropriate setup done above.
250  std::vector<SectionRef> Sections;
251  std::vector<SymbolRef> Symbols;
252  SmallVector<uint64_t, 8> FoundFns;
253  uint64_t BaseSegmentAddress;
254
255  getSectionsAndSymbols(Header, MachOOF, Sections, Symbols, FoundFns,
256                        BaseSegmentAddress);
257
258  // Sort the symbols by address, just in case they didn't come in that way.
259  std::sort(Symbols.begin(), Symbols.end(), SymbolSorter());
260
261  // Build a data in code table that is sorted on by the address of each entry.
262  uint64_t BaseAddress = 0;
263  if (Header.filetype == MachO::MH_OBJECT)
264    Sections[0].getAddress(BaseAddress);
265  else
266    BaseAddress = BaseSegmentAddress;
267  DiceTable Dices;
268  for (dice_iterator DI = MachOOF->begin_dices(), DE = MachOOF->end_dices();
269       DI != DE; ++DI) {
270    uint32_t Offset;
271    DI->getOffset(Offset);
272    Dices.push_back(std::make_pair(BaseAddress + Offset, *DI));
273  }
274  array_pod_sort(Dices.begin(), Dices.end());
275
276#ifndef NDEBUG
277  raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls();
278#else
279  raw_ostream &DebugOut = nulls();
280#endif
281
282  std::unique_ptr<DIContext> diContext;
283  ObjectFile *DbgObj = MachOOF;
284  // Try to find debug info and set up the DIContext for it.
285  if (UseDbg) {
286    // A separate DSym file path was specified, parse it as a macho file,
287    // get the sections and supply it to the section name parsing machinery.
288    if (!DSYMFile.empty()) {
289      std::unique_ptr<MemoryBuffer> Buf;
290      if (error_code ec = MemoryBuffer::getFileOrSTDIN(DSYMFile, Buf)) {
291        errs() << "llvm-objdump: " << Filename << ": " << ec.message() << '\n';
292        return;
293      }
294      DbgObj = ObjectFile::createMachOObjectFile(Buf.release()).get();
295    }
296
297    // Setup the DIContext
298    diContext.reset(DIContext::getDWARFContext(DbgObj));
299  }
300
301  for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) {
302
303    bool SectIsText = false;
304    Sections[SectIdx].isText(SectIsText);
305    if (SectIsText == false)
306      continue;
307
308    StringRef SectName;
309    if (Sections[SectIdx].getName(SectName) ||
310        SectName != "__text")
311      continue; // Skip non-text sections
312
313    DataRefImpl DR = Sections[SectIdx].getRawDataRefImpl();
314
315    StringRef SegmentName = MachOOF->getSectionFinalSegmentName(DR);
316    if (SegmentName != "__TEXT")
317      continue;
318
319    StringRef Bytes;
320    Sections[SectIdx].getContents(Bytes);
321    StringRefMemoryObject memoryObject(Bytes);
322    bool symbolTableWorked = false;
323
324    // Parse relocations.
325    std::vector<std::pair<uint64_t, SymbolRef>> Relocs;
326    for (const RelocationRef &Reloc : Sections[SectIdx].relocations()) {
327      uint64_t RelocOffset, SectionAddress;
328      Reloc.getOffset(RelocOffset);
329      Sections[SectIdx].getAddress(SectionAddress);
330      RelocOffset -= SectionAddress;
331
332      symbol_iterator RelocSym = Reloc.getSymbol();
333
334      Relocs.push_back(std::make_pair(RelocOffset, *RelocSym));
335    }
336    array_pod_sort(Relocs.begin(), Relocs.end());
337
338    // Disassemble symbol by symbol.
339    for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) {
340      StringRef SymName;
341      Symbols[SymIdx].getName(SymName);
342
343      SymbolRef::Type ST;
344      Symbols[SymIdx].getType(ST);
345      if (ST != SymbolRef::ST_Function)
346        continue;
347
348      // Make sure the symbol is defined in this section.
349      bool containsSym = false;
350      Sections[SectIdx].containsSymbol(Symbols[SymIdx], containsSym);
351      if (!containsSym)
352        continue;
353
354      // Start at the address of the symbol relative to the section's address.
355      uint64_t SectionAddress = 0;
356      uint64_t Start = 0;
357      Sections[SectIdx].getAddress(SectionAddress);
358      Symbols[SymIdx].getAddress(Start);
359      Start -= SectionAddress;
360
361      // Stop disassembling either at the beginning of the next symbol or at
362      // the end of the section.
363      bool containsNextSym = false;
364      uint64_t NextSym = 0;
365      uint64_t NextSymIdx = SymIdx+1;
366      while (Symbols.size() > NextSymIdx) {
367        SymbolRef::Type NextSymType;
368        Symbols[NextSymIdx].getType(NextSymType);
369        if (NextSymType == SymbolRef::ST_Function) {
370          Sections[SectIdx].containsSymbol(Symbols[NextSymIdx],
371                                           containsNextSym);
372          Symbols[NextSymIdx].getAddress(NextSym);
373          NextSym -= SectionAddress;
374          break;
375        }
376        ++NextSymIdx;
377      }
378
379      uint64_t SectSize;
380      Sections[SectIdx].getSize(SectSize);
381      uint64_t End = containsNextSym ?  NextSym : SectSize;
382      uint64_t Size;
383
384      symbolTableWorked = true;
385
386      outs() << SymName << ":\n";
387      DILineInfo lastLine;
388      for (uint64_t Index = Start; Index < End; Index += Size) {
389        MCInst Inst;
390
391        uint64_t SectAddress = 0;
392        Sections[SectIdx].getAddress(SectAddress);
393        outs() << format("%8" PRIx64 ":\t", SectAddress + Index);
394
395        // Check the data in code table here to see if this is data not an
396        // instruction to be disassembled.
397        DiceTable Dice;
398        Dice.push_back(std::make_pair(SectAddress + Index, DiceRef()));
399        dice_table_iterator DTI = std::search(Dices.begin(), Dices.end(),
400                                              Dice.begin(), Dice.end(),
401                                              compareDiceTableEntries);
402        if (DTI != Dices.end()){
403          uint16_t Length;
404          DTI->second.getLength(Length);
405          DumpBytes(StringRef(Bytes.data() + Index, Length));
406          uint16_t Kind;
407          DTI->second.getKind(Kind);
408          DumpDataInCode(Bytes.data() + Index, Length, Kind);
409          continue;
410        }
411
412        if (DisAsm->getInstruction(Inst, Size, memoryObject, Index,
413                                   DebugOut, nulls())) {
414          DumpBytes(StringRef(Bytes.data() + Index, Size));
415          IP->printInst(&Inst, outs(), "");
416
417          // Print debug info.
418          if (diContext) {
419            DILineInfo dli =
420              diContext->getLineInfoForAddress(SectAddress + Index);
421            // Print valid line info if it changed.
422            if (dli != lastLine && dli.getLine() != 0)
423              outs() << "\t## " << dli.getFileName() << ':'
424                << dli.getLine() << ':' << dli.getColumn();
425            lastLine = dli;
426          }
427          outs() << "\n";
428        } else {
429          errs() << "llvm-objdump: warning: invalid instruction encoding\n";
430          if (Size == 0)
431            Size = 1; // skip illegible bytes
432        }
433      }
434    }
435    if (!symbolTableWorked) {
436      // Reading the symbol table didn't work, disassemble the whole section.
437      uint64_t SectAddress;
438      Sections[SectIdx].getAddress(SectAddress);
439      uint64_t SectSize;
440      Sections[SectIdx].getSize(SectSize);
441      uint64_t InstSize;
442      for (uint64_t Index = 0; Index < SectSize; Index += InstSize) {
443        MCInst Inst;
444
445        if (DisAsm->getInstruction(Inst, InstSize, memoryObject, Index,
446                                   DebugOut, nulls())) {
447          outs() << format("%8" PRIx64 ":\t", SectAddress + Index);
448          DumpBytes(StringRef(Bytes.data() + Index, InstSize));
449          IP->printInst(&Inst, outs(), "");
450          outs() << "\n";
451        } else {
452          errs() << "llvm-objdump: warning: invalid instruction encoding\n";
453          if (InstSize == 0)
454            InstSize = 1; // skip illegible bytes
455        }
456      }
457    }
458  }
459}
460