MachODump.cpp revision 0b8b771e9f2f251460a6f200c45efe9d55640d60 
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>
40using namespace llvm;
41using namespace object;
42
43static 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
47static 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
81struct 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
90struct 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
97static void DumpAddress(uint64_t Address, ArrayRef<Section> Sections,
98                        MachOObject *MachOObj, raw_ostream &OS) {
99  for (unsigned i = 0; i != Sections.size(); ++i) {
100    uint64_t addr = Address-Sections[i].Address;
101    if (Sections[i].Address <= Address &&
102        Sections[i].Address + Sections[i].Size > Address) {
103      StringRef bytes = MachOObj->getData(Sections[i].Offset,
104                                          Sections[i].Size);
105      if (!strcmp(Sections[i].Name, "__cstring"))
106        OS << '"' << bytes.substr(addr, bytes.find('\0', addr)) << '"';
107      if (!strcmp(Sections[i].Name, "__cfstring"))
108        OS << "@\"" << bytes.substr(addr, bytes.find('\0', addr)) << '"';
109    }
110  }
111}
112
113void llvm::DisassembleInputMachO(StringRef Filename) {
114  OwningPtr<MemoryBuffer> Buff;
115
116  if (error_code ec = MemoryBuffer::getFileOrSTDIN(Filename, Buff)) {
117    errs() << "llvm-objdump: " << Filename << ": " << ec.message() << "\n";
118    return;
119  }
120
121  OwningPtr<MachOObject> MachOObj(MachOObject::LoadFromBuffer(Buff.take()));
122
123  const Target *TheTarget = GetTarget(MachOObj.get());
124  if (!TheTarget) {
125    // GetTarget prints out stuff.
126    return;
127  }
128  const MCInstrInfo *InstrInfo = TheTarget->createMCInstrInfo();
129  OwningPtr<MCInstrAnalysis>
130    InstrAnalysis(TheTarget->createMCInstrAnalysis(InstrInfo));
131
132  // Set up disassembler.
133  OwningPtr<const MCAsmInfo> AsmInfo(TheTarget->createMCAsmInfo(TripleName));
134
135  if (!AsmInfo) {
136    errs() << "error: no assembly info for target " << TripleName << "\n";
137    return;
138  }
139
140  OwningPtr<const MCSubtargetInfo>
141    STI(TheTarget->createMCSubtargetInfo(TripleName, "", ""));
142
143  if (!STI) {
144    errs() << "error: no subtarget info for target " << TripleName << "\n";
145    return;
146  }
147
148  OwningPtr<const MCDisassembler> DisAsm(TheTarget->createMCDisassembler(*STI));
149  if (!DisAsm) {
150    errs() << "error: no disassembler for target " << TripleName << "\n";
151    return;
152  }
153
154  int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
155  OwningPtr<MCInstPrinter> IP(TheTarget->createMCInstPrinter(
156        AsmPrinterVariant, *AsmInfo, *STI));
157  if (!IP) {
158    errs() << "error: no instruction printer for target " << TripleName << '\n';
159    return;
160  }
161
162  outs() << '\n';
163  outs() << Filename << ":\n\n";
164
165  const macho::Header &Header = MachOObj->getHeader();
166
167  const MachOObject::LoadCommandInfo *SymtabLCI = 0;
168  for (unsigned i = 0; i != Header.NumLoadCommands; ++i) {
169    const MachOObject::LoadCommandInfo &LCI = MachOObj->getLoadCommandInfo(i);
170    switch (LCI.Command.Type) {
171    case macho::LCT_Symtab:
172      SymtabLCI = &LCI;
173      break;
174    }
175  }
176
177  // Read and register the symbol table data.
178  InMemoryStruct<macho::SymtabLoadCommand> SymtabLC;
179  MachOObj->ReadSymtabLoadCommand(*SymtabLCI, SymtabLC);
180  MachOObj->RegisterStringTable(*SymtabLC);
181
182  std::vector<Section> Sections;
183  std::vector<Symbol> Symbols;
184  std::vector<Symbol> UnsortedSymbols; // FIXME: duplication
185  SmallVector<uint64_t, 8> FoundFns;
186
187  for (unsigned i = 0; i != Header.NumLoadCommands; ++i) {
188    const MachOObject::LoadCommandInfo &LCI = MachOObj->getLoadCommandInfo(i);
189    if (LCI.Command.Type == macho::LCT_Segment) {
190      InMemoryStruct<macho::SegmentLoadCommand> SegmentLC;
191      MachOObj->ReadSegmentLoadCommand(LCI, SegmentLC);
192
193      for (unsigned SectNum = 0; SectNum != SegmentLC->NumSections; ++SectNum) {
194        InMemoryStruct<macho::Section> Sect;
195        MachOObj->ReadSection(LCI, SectNum, Sect);
196
197        Section S;
198        memcpy(S.Name, Sect->Name, 16);
199        S.Address = Sect->Address;
200        S.Size = Sect->Size;
201        S.Offset = Sect->Offset;
202        S.NumRelocs = Sect->NumRelocationTableEntries;
203        S.RelocTableOffset = Sect->RelocationTableOffset;
204        Sections.push_back(S);
205
206        for (unsigned i = 0; i != SymtabLC->NumSymbolTableEntries; ++i) {
207          InMemoryStruct<macho::SymbolTableEntry> STE;
208          MachOObj->ReadSymbolTableEntry(SymtabLC->SymbolTableOffset, i, STE);
209
210          Symbol S;
211          S.StringIndex = STE->StringIndex;
212          S.SectionIndex = STE->SectionIndex;
213          S.Value = STE->Value;
214          Symbols.push_back(S);
215          UnsortedSymbols.push_back(Symbols.back());
216        }
217      }
218    } else if (LCI.Command.Type == macho::LCT_Segment64) {
219      InMemoryStruct<macho::Segment64LoadCommand> Segment64LC;
220      MachOObj->ReadSegment64LoadCommand(LCI, Segment64LC);
221
222      for (unsigned SectNum = 0; SectNum != Segment64LC->NumSections; ++SectNum) {
223        InMemoryStruct<macho::Section64> Sect64;
224        MachOObj->ReadSection64(LCI, SectNum, Sect64);
225
226        Section S;
227        memcpy(S.Name, Sect64->Name, 16);
228        S.Address = Sect64->Address;
229        S.Size = Sect64->Size;
230        S.Offset = Sect64->Offset;
231        S.NumRelocs = Sect64->NumRelocationTableEntries;
232        S.RelocTableOffset = Sect64->RelocationTableOffset;
233        Sections.push_back(S);
234
235        for (unsigned i = 0; i != SymtabLC->NumSymbolTableEntries; ++i) {
236          InMemoryStruct<macho::Symbol64TableEntry> STE;
237          MachOObj->ReadSymbol64TableEntry(SymtabLC->SymbolTableOffset, i, STE);
238
239          Symbol S;
240          S.StringIndex = STE->StringIndex;
241          S.SectionIndex = STE->SectionIndex;
242          S.Value = STE->Value;
243          Symbols.push_back(S);
244          UnsortedSymbols.push_back(Symbols.back());
245        }
246      }
247    } else if (LCI.Command.Type == macho::LCT_FunctionStarts) {
248      InMemoryStruct<macho::LinkeditDataLoadCommand> LLC;
249      MachOObj->ReadLinkeditDataLoadCommand(LCI, LLC);
250
251      MachOObj->ReadULEB128s(LLC->DataOffset, FoundFns);
252    }
253  }
254
255  std::map<uint64_t, MCFunction*> FunctionMap;
256
257  // Sort the symbols by address, just in case they didn't come in that way.
258  array_pod_sort(Symbols.begin(), Symbols.end());
259
260#ifndef NDEBUG
261  raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls();
262#else
263  raw_ostream &DebugOut = nulls();
264#endif
265
266  SmallVector<MCFunction, 16> Functions;
267
268  for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) {
269    if (strcmp(Sections[SectIdx].Name, "__text"))
270      continue;
271
272    uint64_t VMAddr = Sections[SectIdx].Address - Sections[SectIdx].Offset;
273    for (unsigned i = 0, e = FoundFns.size(); i != e; ++i)
274      FunctionMap.insert(std::pair<uint64_t,MCFunction*>(FoundFns[i]+VMAddr,0));
275
276    StringRef Bytes = MachOObj->getData(Sections[SectIdx].Offset,
277                                        Sections[SectIdx].Size);
278    StringRefMemoryObject memoryObject(Bytes);
279    bool symbolTableWorked = false;
280
281    std::vector<std::pair<uint64_t, uint32_t> > Relocs;
282    for (unsigned j = 0; j != Sections[SectIdx].NumRelocs; ++j) {
283      InMemoryStruct<macho::RelocationEntry> RE;
284      MachOObj->ReadRelocationEntry(Sections[SectIdx].RelocTableOffset, j, RE);
285      Relocs.push_back(std::make_pair(RE->Word0, RE->Word1 & 0xffffff));
286    }
287    array_pod_sort(Relocs.begin(), Relocs.end());
288
289    for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) {
290      if ((unsigned)Symbols[SymIdx].SectionIndex - 1 != SectIdx)
291        continue;
292
293      uint64_t Start = Symbols[SymIdx].Value - Sections[SectIdx].Address;
294      uint64_t End = (SymIdx+1 == Symbols.size() ||
295          Symbols[SymIdx].SectionIndex != Symbols[SymIdx+1].SectionIndex) ?
296          Sections[SectIdx].Size :
297          Symbols[SymIdx+1].Value - Sections[SectIdx].Address;
298      uint64_t Size;
299
300      if (Start >= End)
301        continue;
302
303      symbolTableWorked = true;
304
305      if (!CFG) {
306        outs() << MachOObj->getStringAtIndex(Symbols[SymIdx].StringIndex)
307          << ":\n";
308        for (uint64_t Index = Start; Index < End; Index += Size) {
309          MCInst Inst;
310
311          if (DisAsm->getInstruction(Inst, Size, memoryObject, Index,
312                                     DebugOut, nulls())) {
313            outs() << format("%8llx:\t", Sections[SectIdx].Address + Index);
314            DumpBytes(StringRef(Bytes.data() + Index, Size));
315            IP->printInst(&Inst, outs(), "");
316            outs() << "\n";
317          } else {
318            errs() << "llvm-objdump: warning: invalid instruction encoding\n";
319            if (Size == 0)
320              Size = 1; // skip illegible bytes
321          }
322        }
323      } else {
324        // Create CFG and use it for disassembly.
325        SmallVector<uint64_t, 16> Calls;
326        MCFunction f =
327          MCFunction::createFunctionFromMC(
328              MachOObj->getStringAtIndex(Symbols[SymIdx].StringIndex),
329              DisAsm.get(),
330              memoryObject, Start, End,
331              InstrAnalysis.get(), DebugOut,
332              Calls);
333
334        Functions.push_back(f);
335        FunctionMap[Start] = &Functions.back();
336
337        for (unsigned i = 0, e = Calls.size(); i != e; ++i)
338          FunctionMap.insert(std::pair<uint64_t, MCFunction*>(Calls[i], 0));
339      }
340    }
341
342    if (CFG) {
343      if (!symbolTableWorked) {
344        // Create CFG and use it for disassembly.
345        SmallVector<uint64_t, 16> Calls;
346        MCFunction f =
347          MCFunction::createFunctionFromMC("__TEXT", DisAsm.get(),
348              memoryObject, 0, Sections[SectIdx].Size,
349              InstrAnalysis.get(), DebugOut,
350              Calls);
351
352        Functions.push_back(f);
353        FunctionMap[Sections[SectIdx].Offset] = &Functions.back();
354
355        for (unsigned i = 0, e = Calls.size(); i != e; ++i)
356          FunctionMap.insert(std::pair<uint64_t, MCFunction*>(Calls[i], 0));
357      }
358      for (std::map<uint64_t, MCFunction*>::iterator mi = FunctionMap.begin(),
359           me = FunctionMap.end(); mi != me; ++mi)
360        if (mi->second == 0) {
361          SmallVector<uint64_t, 16> Calls;
362          MCFunction f =
363            MCFunction::createFunctionFromMC("unknown", DisAsm.get(),
364                                             memoryObject, mi->first,
365                                             Sections[SectIdx].Size,
366                                             InstrAnalysis.get(), DebugOut,
367                                             Calls);
368          Functions.push_back(f);
369          mi->second = &Functions.back();
370          for (unsigned i = 0, e = Calls.size(); i != e; ++i)
371            if (FunctionMap.insert(std::pair<uint64_t, MCFunction*>(Calls[i],0))
372                                                                        .second)
373              mi = FunctionMap.begin();
374        }
375
376      DenseSet<uint64_t> PrintedBlocks;
377      for (unsigned ffi = 0, ffe = Functions.size(); ffi != ffe; ++ffi) {
378        MCFunction &f = Functions[ffi];
379        for (MCFunction::iterator fi = f.begin(), fe = f.end(); fi != fe; ++fi){
380          if (!PrintedBlocks.insert(fi->first).second)
381            continue;
382          bool hasPreds = FunctionMap.find(fi->first) != FunctionMap.end();
383
384          // Only print blocks that have predecessors.
385          // FIXME: Slow.
386          for (MCFunction::iterator pi = f.begin(), pe = f.end(); pi != pe;
387              ++pi)
388            if (pi->second.contains(fi->first)) {
389              hasPreds = true;
390              break;
391            }
392
393          // Data block.
394          if (!hasPreds && fi != f.begin()) {
395            uint64_t End = llvm::next(fi) == fe ? Sections[SectIdx].Size :
396                                                  llvm::next(fi)->first;
397            outs() << "# " << End-fi->first << " bytes of data:\n";
398            for (unsigned pos = fi->first; pos != End; ++pos) {
399              outs() << format("%8x:\t", Sections[SectIdx].Address + pos);
400              DumpBytes(StringRef(Bytes.data() + pos, 1));
401              outs() << format("\t.byte 0x%02x\n", (uint8_t)Bytes[pos]);
402            }
403            continue;
404          }
405
406          if (fi->second.contains(fi->first))
407            outs() << "# Loop begin:\n";
408
409          for (unsigned ii = 0, ie = fi->second.getInsts().size(); ii != ie;
410               ++ii) {
411            const MCDecodedInst &Inst = fi->second.getInsts()[ii];
412            if (FunctionMap.find(Sections[SectIdx].Address + Inst.Address) !=
413                FunctionMap.end())
414              outs() << FunctionMap[Sections[SectIdx].Address + Inst.Address]->
415                                                             getName() << ":\n";
416            outs() << format("%8llx:\t", Sections[SectIdx].Address +
417                                         Inst.Address);
418            DumpBytes(StringRef(Bytes.data() + Inst.Address, Inst.Size));
419            // Simple loops.
420            if (fi->second.contains(fi->first))
421              outs() << '\t';
422            IP->printInst(&Inst.Inst, outs(), "");
423            for (unsigned j = 0; j != Relocs.size(); ++j)
424              if (Relocs[j].first >= Sections[SectIdx].Address + Inst.Address &&
425                  Relocs[j].first < Sections[SectIdx].Address + Inst.Address +
426                                    Inst.Size) {
427                outs() << "\t# "
428                   << MachOObj->getStringAtIndex(
429                                  UnsortedSymbols[Relocs[j].second].StringIndex)
430                   << ' ';
431                DumpAddress(UnsortedSymbols[Relocs[j].second].Value, Sections,
432                            MachOObj.get(), outs());
433              }
434            uint64_t targ = InstrAnalysis->evaluateBranch(Inst.Inst,
435                                                          Inst.Address,
436                                                          Inst.Size);
437            if (targ != -1ULL)
438              DumpAddress(targ, Sections, MachOObj.get(), outs());
439
440            outs() << '\n';
441          }
442        }
443
444        // Start a new dot file.
445        std::string Error;
446        raw_fd_ostream Out((f.getName().str() + ".dot").c_str(), Error);
447        if (!Error.empty()) {
448          errs() << "llvm-objdump: warning: " << Error << '\n';
449          continue;
450        }
451
452        Out << "digraph " << f.getName() << " {\n";
453        Out << "graph [ rankdir = \"LR\" ];\n";
454        for (MCFunction::iterator i = f.begin(), e = f.end(); i != e; ++i) {
455          bool hasPreds = false;
456          // Only print blocks that have predecessors.
457          // FIXME: Slow.
458          for (MCFunction::iterator pi = f.begin(), pe = f.end(); pi != pe;
459               ++pi)
460            if (pi->second.contains(i->first)) {
461              hasPreds = true;
462              break;
463            }
464
465          if (!hasPreds && i != f.begin())
466            continue;
467
468          Out << '"' << i->first << "\" [ label=\"<a>";
469          // Print instructions.
470          for (unsigned ii = 0, ie = i->second.getInsts().size(); ii != ie;
471               ++ii) {
472            // Escape special chars and print the instruction in mnemonic form.
473            std::string Str;
474            raw_string_ostream OS(Str);
475            IP->printInst(&i->second.getInsts()[ii].Inst, OS, "");
476            Out << DOT::EscapeString(OS.str()) << '|';
477          }
478          Out << "<o>\" shape=\"record\" ];\n";
479
480          // Add edges.
481          for (MCBasicBlock::succ_iterator si = i->second.succ_begin(),
482              se = i->second.succ_end(); si != se; ++si)
483            Out << i->first << ":o -> " << *si <<":a\n";
484        }
485        Out << "}\n";
486      }
487    }
488  }
489}
490