src/Binary.cxx

   1 #include "Binary.hxx"
   2
   3 #include <iostream>
   4 #include <string>
   5 #include <algorithm>
   6
   7 #include "llvm/Support/raw_ostream.h"
   8
   9 using namespace llvm;
  10 using namespace llvm::object;
  11
  12 namespace {
  13     bool error(error_code ec) {
  14         if (!ec) return false;
  15
  16         outs() << "error reading file: " << ec.message() << ".\n";
  17         outs().flush();
  18         return true;
  19     }
  20
  21     bool RelocAddressLess(RelocationRef a, RelocationRef b) {
  22         uint64_t a_addr, b_addr;
  23         if (error(a.getOffset(a_addr))) return false;
  24         if (error(b.getOffset(b_addr))) return false;
  25         return a_addr < b_addr;
  26     }
  27
  28     void DumpBytes(StringRef bytes) {
  29         static const char hex_rep[] = "0123456789abcdef";
  30         // FIXME: The real way to do this is to figure out the longest instruction
  31         //        and align to that size before printing. I'll fix this when I get
  32         //        around to outputting relocations.
  33         // 15 is the longest x86 instruction
  34         // 3 is for the hex rep of a byte + a space.
  35         // 1 is for the null terminator.
  36         enum { OutputSize = (15 * 3) + 1 };
  37         char output[OutputSize];
  38
  39         assert(bytes.size() <= 15
  40                && "DumpBytes only supports instructions of up to 15 bytes");
  41         memset(output, ' ', sizeof(output));
  42         unsigned index = 0;
  43         for (StringRef::iterator i = bytes.begin(),
  44                  e = bytes.end(); i != e; ++i) {
  45             output[index] = hex_rep[(*i & 0xF0) >> 4];
  46             output[index + 1] = hex_rep[*i & 0xF];
  47             index += 3;
  48         }
  49
  50         output[sizeof(output) - 1] = 0;
  51         outs() << output;
  52     }
  53
  54     std::map<std::string, SectionRef> readSections(const ObjectFile& o) {
  55         error_code ec;
  56         std::map<std::string, SectionRef> result;
  57         section_iterator i(o.begin_sections()), e(o.end_sections());
  58         for (; i != e; i.increment(ec)) {
  59             StringRef name;
  60             if (error(i->getName(name))) break;
  61
  62             result.insert(make_pair(name.str(), *i));
  63         }
  64         return result;
  65     }
  66
  67     std::map<std::string, SymbolRef> readSymbols(const ObjectFile& o) {
  68         error_code ec;
  69         std::map<std::string, SymbolRef> result;
  70         symbol_iterator si(o.begin_symbols()), se(o.end_symbols());
  71         for (; si != se; si.increment(ec)) {
  72             StringRef name;
  73             if (error(si->getName(name))) break;
  74
  75             result.insert(make_pair(name.str(), *si));
  76         }
  77         return result;
  78     }
  79 }
  80
  81 namespace qtlldb {
  82
  83     Binary::Binary(const std::string& filename)
  84         : triple("unkown-unknown-unknown")
  85     {
  86         std::string error;
  87
  88         createBinary(filename, binary);
  89         if (Archive *a = dyn_cast<Archive>(binary.get())) {
  90             std::cerr << "Got an archive!" << std::endl;
  91             return;
  92         }
  93
  94         o = dyn_cast<ObjectFile>(binary.get());
  95
  96         triple.setArch(Triple::ArchType(o->getArch()));
  97         std::string tripleName(triple.getTriple());
  98
  99         outs() << tripleName << "\n";
 100
 101         target = TargetRegistry::lookupTarget("", triple, error);
 102         if (!target) {
 103             std::cerr << error;
 104             return;
 105         }
 106
 107         outs() << target->getName() << "\n";
 108
 109         MRI.reset(target->createMCRegInfo(tripleName));
 110         if (!MRI) {
 111             std::cerr << "error: no register info for target " << tripleName << "\n";
 112             return;
 113         }
 114
 115         // Set up disassembler.
 116         AsmInfo.reset(target->createMCAsmInfo(*MRI, tripleName));
 117         if (!AsmInfo) {
 118             std::cerr << "error: no assembly info for target " << tripleName << "\n";
 119             return;
 120         }
 121
 122         STI.reset(target->createMCSubtargetInfo(tripleName, "", ""));
 123         if (!STI) {
 124             errs() << "error: no subtarget info for target " << tripleName << "\n";
 125             return;
 126         }
 127
 128         MII.reset(target->createMCInstrInfo());
 129         if (!MII) {
 130             std::cerr << "error: no instruction info for target " << tripleName << "\n";
 131             return;
 132         }
 133
 134         DisAsm.reset(target->createMCDisassembler(*STI));
 135         if (!DisAsm) {
 136             std::cerr << "error: no disassembler for target " << tripleName << "\n";
 137             return;
 138         }
 139
 140         MOFI.reset(new MCObjectFileInfo);
 141         Ctx.reset(new MCContext(AsmInfo.get(), MRI.get(), MOFI.get()));
 142         RelInfo.reset(
 143             target->createMCRelocationInfo(tripleName, *Ctx.get()));
 144         if (RelInfo) {
 145             Symzer.reset(
 146                 MCObjectSymbolizer::createObjectSymbolizer(*Ctx.get(), RelInfo, o));
 147             if (Symzer)
 148                 DisAsm->setSymbolizer(Symzer);
 149         }
 150
 151         MIA.reset(target->createMCInstrAnalysis(MII.get()));
 152
 153         int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
 154         IP.reset(target->createMCInstPrinter(AsmPrinterVariant, *AsmInfo, *MII, *MRI, *STI));
 155         if (!IP) {
 156             std::cerr << "error: no instruction printer for target " << tripleName
 157                       << '\n';
 158             return;
 159         }
 160
 161         OwningPtr<MCObjectDisassembler> OD(
 162             new MCObjectDisassembler(*o, *DisAsm, *MIA));
 163         Mod.reset(OD->buildModule(/* withCFG */ false));
 164
 165         symbols = readSymbols(*o);
 166         sections = readSections(*o);
 167
 168         // for_each(sections.begin(), sections.end(), [](const std::pair<std::string, SectionRef>& i){
 169         //         std::cout << "Section: " << i.first << std::endl;
 170         //     });
 171     }
 172
 173
 174
 175     std::vector<std::string> Binary::getSymbols() {
 176         error_code ec;
 177         SectionRef r = sections[".text"];
 178         std::vector<std::string> result;
 179         for_each(symbols.begin(), symbols.end(), [&](const std::pair<std::string, SymbolRef>& i) {
 180                      bool contains;
 181                      if (!error(r.containsSymbol(i.second, contains)) && contains)
 182                          result.push_back(i.first);
 183                  });
 184         return result;
 185     }
 186
 187     void Binary::for_each_instruction(const std::string& function,
 188                                       std::function<void (long, std::string, std::string)> callback) {
 189         StringRef bytes;
 190         uint64_t base_address, address, ssize, size(0), index, end;
 191         StringRefMemoryObject memoryObject("");
 192
 193         if (symbols.end() != symbols.find(function)) {
 194             SymbolRef ref;
 195             section_iterator sec(o->begin_sections());
 196
 197             ref = symbols.at(function);
 198             if (error(ref.getSection(sec))) return;
 199             if (error(ref.getAddress(address))) return;
 200             if (address == UnknownAddressOrSize) return;
 201             if (error(ref.getSize(ssize))) return;
 202             if (error(sec->getAddress(base_address))) return;
 203             if (error(sec->getContents(bytes))) return;
 204             memoryObject = bytes;
 205
 206         }
 207         else if (sections.end() != sections.find(function)) {
 208             SectionRef sref = sections.at(function);
 209             if (error(sref.getAddress(address))) return;
 210             if (address == UnknownAddressOrSize) return;
 211             if (error(sref.getSize(ssize))) return;
 212             if (error(sref.getContents(bytes))) return;
 213             base_address = address;
 214             memoryObject = bytes;
 215         }
 216
 217
 218         // outs() << "Start for_each_instruction " << function << "\n";
 219
 220
 221         for (end = address + ssize - base_address, index = address - base_address; index < end; index += size) {
 222             MCInst Inst;
 223
 224             if (DisAsm->getInstruction(Inst, size, memoryObject, index,
 225                                        nulls(), nulls())) {
 226                 std::string buf;
 227                 llvm::raw_string_ostream s(buf);
 228                 IP->printInst(&Inst, s, "");
 229
 230                 if (index + size < bytes.str().length())
 231                     callback(base_address + index, bytes.str().substr(index, size), s.str());
 232
 233             } else {
 234                 errs() << "warning: invalid instruction encoding\n";
 235                 if (size == 0)
 236                     size = 1; // skip illegible bytes
 237             }
 238         }
 239         // outs() << "End for_each_instruction\n";
 240
 241     }
 242
 243     void Binary::disassemble() {
 244         for (MCModule::const_atom_iterator AI = Mod->atom_begin(),
 245                  AE = Mod->atom_end();
 246              AI != AE; ++AI) {
 247
 248             if ((*AI)->getKind() != llvm::MCAtom::TextAtom)
 249                 continue;
 250
 251             outs() << "\n\nAtom " << (*AI)->getName() << ": \n";
 252             if (const MCTextAtom *TA = dyn_cast<MCTextAtom>(*AI)) {
 253                 for (MCTextAtom::const_iterator II = TA->begin(), IE = TA->end();
 254                      II != IE;
 255                      ++II) {
 256 //                II->Inst.dump();
 257                     IP->printInst(&II->Inst, outs(), "");
 258                     outs() << "\n";
 259                 }
 260             }
 261         }
 262
 263         outs() << "binary " << triple.getArchName() << "\n";
 264     }
 265
 266     void Binary::disassemble_functions() {
 267         error_code ec;
 268         for (section_iterator i = o->begin_sections(),
 269                  e = o->end_sections();
 270              i != e; i.increment(ec)) {
 271             if (error(ec)) break;
 272             bool text;
 273             if (error(i->isText(text))) break;
 274             if (!text) continue;
 275
 276             uint64_t SectionAddr;
 277             if (error(i->getAddress(SectionAddr))) break;
 278
 279             // Make a list of all the symbols in this section.
 280             std::vector<std::pair<uint64_t, StringRef> > Symbols;
 281             for (symbol_iterator si = o->begin_symbols(),
 282                      se = o->end_symbols();
 283                  si != se; si.increment(ec)) {
 284                 bool contains;
 285                 if (!error(i->containsSymbol(*si, contains)) && contains) {
 286                     uint64_t Address;
 287                     if (error(si->getAddress(Address))) break;
 288                     if (Address == UnknownAddressOrSize) continue;
 289                     Address -= SectionAddr;
 290
 291                     StringRef Name;
 292                     if (error(si->getName(Name))) break;
 293
 294                     outs() << "\nXXX " << Name << "\n";
 295
 296                     Symbols.push_back(std::make_pair(Address, Name));
 297                 }
 298             }
 299
 300             // Sort the symbols by address, just in case they didn't come in that way.
 301             array_pod_sort(Symbols.begin(), Symbols.end());
 302
 303             // Make a list of all the relocations for this section.
 304             std::vector<RelocationRef> Rels;
 305             // if (InlineRelocs) {
 306             //     for (relocation_iterator ri = i->begin_relocations(),
 307             //              re = i->end_relocations();
 308             //          ri != re; ri.increment(ec)) {
 309             //         if (error(ec)) break;
 310             //         Rels.push_back(*ri);
 311             //     }
 312             // }
 313
 314             // Sort relocations by address.
 315             std::sort(Rels.begin(), Rels.end(), RelocAddressLess);
 316
 317             StringRef SegmentName = "";
 318             // if (const MachOObjectFile *MachO =
 319             //     dyn_cast<const MachOObjectFile>(o)) {
 320             //     DataRefImpl DR = i->getRawDataRefImpl();
 321             //     SegmentName = MachO->getSectionFinalSegmentName(DR);
 322             // }
 323             StringRef name;
 324             if (error(i->getName(name))) break;
 325             outs() << "Disassembly of section ";
 326             if (!SegmentName.empty())
 327                 outs() << SegmentName << ",";
 328             outs() << name << ':';
 329
 330             // If the section has no symbols just insert a dummy one and disassemble
 331             // the whole section.
 332             if (Symbols.empty())
 333                 Symbols.push_back(std::make_pair(0, name));
 334
 335
 336             StringRef Bytes;
 337             if (error(i->getContents(Bytes))) break;
 338             StringRefMemoryObject memoryObject(Bytes);
 339             uint64_t Size;
 340             uint64_t Index;
 341             uint64_t SectSize;
 342             if (error(i->getSize(SectSize))) break;
 343
 344             std::vector<RelocationRef>::const_iterator rel_cur = Rels.begin();
 345             std::vector<RelocationRef>::const_iterator rel_end = Rels.end();
 346             // Disassemble symbol by symbol.
 347             for (unsigned si = 0, se = Symbols.size(); si != se; ++si) {
 348                 uint64_t Start = Symbols[si].first;
 349                 uint64_t End;
 350                 // The end is either the size of the section or the beginning of the next
 351                 // symbol.
 352                 if (si == se - 1)
 353                     End = SectSize;
 354                 // Make sure this symbol takes up space.
 355                 else if (Symbols[si + 1].first != Start)
 356                     End = Symbols[si + 1].first - 1;
 357                 else
 358                     // This symbol has the same address as the next symbol. Skip it.
 359                     continue;
 360
 361                 outs() << '\n' << Symbols[si].second << ":\n";
 362
 363 #ifndef NDEBUG
 364                 raw_ostream &DebugOut = DebugFlag ? dbgs() : nulls();
 365 #else
 366                 raw_ostream &DebugOut = nulls();
 367 #endif
 368
 369                 for (Index = Start; Index < End; Index += Size) {
 370                     MCInst Inst;
 371
 372                     if (DisAsm->getInstruction(Inst, Size, memoryObject, Index,
 373                                                DebugOut, nulls())) {
 374                         outs() << format("%8" PRIx64 ":", SectionAddr + Index);
 375                         outs() << "\t";
 376                         DumpBytes(StringRef(Bytes.data() + Index, Size));
 377
 378                         IP->printInst(&Inst, outs(), "");
 379                         outs() << "\n";
 380                     } else {
 381                         errs() << "warning: invalid instruction encoding\n";
 382                         if (Size == 0)
 383                             Size = 1; // skip illegible bytes
 384                     }
 385
 386                     // Print relocation for instruction.
 387                     while (rel_cur != rel_end) {
 388                         bool hidden = false;
 389                         uint64_t addr;
 390                         SmallString<16> name;
 391                         SmallString<32> val;
 392
 393                         // If this relocation is hidden, skip it.
 394                         if (error(rel_cur->getHidden(hidden))) goto skip_print_rel;
 395                         if (hidden) goto skip_print_rel;
 396
 397                         if (error(rel_cur->getOffset(addr))) goto skip_print_rel;
 398                         // Stop when rel_cur's address is past the current instruction.
 399                         if (addr >= Index + Size) break;
 400                         if (error(rel_cur->getTypeName(name))) goto skip_print_rel;
 401                         if (error(rel_cur->getValueString(val))) goto skip_print_rel;
 402
 403                         outs() << format("\t\t\t%8" PRIx64 ": ", SectionAddr + addr) << name
 404                                << "\t" << val << "\n";
 405
 406                       skip_print_rel:
 407                         ++rel_cur;
 408                     }
 409                 }
 410             }
 411         }
 412     }
 413
 414     void Binary::disassemble_cfg() {
 415         for (MCModule::const_func_iterator FI = Mod->func_begin(),
 416                  FE = Mod->func_end();
 417              FI != FE; ++FI) {
 418             static int filenum = 0;
 419             std::string FileName = std::string("dot/") + (Twine((*FI)->getName()) + "_" + utostr(filenum) + ".dot").str();
 420
 421             std::cerr << FileName << std::endl;
 422
 423             // Start a new dot file.
 424             std::string Error;
 425             raw_fd_ostream Out(FileName.c_str(), Error);
 426             if (!Error.empty()) {
 427                 errs() << "llvm-objdump: warning: " << Error << '\n';
 428                 return;
 429             }
 430
 431             Out << "digraph \"" << (*FI)->getName() << "\" {\n";
 432             Out << "graph [ rankdir = \"LR\" ];\n";
 433             for (MCFunction::const_iterator i = (*FI)->begin(), e = (*FI)->end(); i != e; ++i) {
 434                 // Only print blocks that have predecessors.
 435                 bool hasPreds = (*i)->pred_begin() != (*i)->pred_end();
 436
 437                 if (!hasPreds && i != (*FI)->begin())
 438                     continue;
 439
 440                 Out << '"' << (*i)->getInsts()->getBeginAddr() << "\" [ label=\"<a>";
 441                 // Print instructions.
 442                 for (unsigned ii = 0, ie = (*i)->getInsts()->size(); ii != ie;
 443                      ++ii) {
 444                     if (ii != 0) // Not the first line, start a new row.
 445                         Out << '|';
 446                     if (ii + 1 == ie) // Last line, add an end id.
 447                         Out << "<o>";
 448
 449                     // Escape special chars and print the instruction in mnemonic form.
 450                     std::string Str;
 451                     raw_string_ostream OS(Str);
 452                     IP->printInst(&(*i)->getInsts()->at(ii).Inst, OS, "");
 453                     Out << DOT::EscapeString(OS.str());
 454                 }
 455                 Out << "\" shape=\"record\" ];\n";
 456
 457                 // Add edges.
 458                 for (MCBasicBlock::succ_const_iterator si = (*i)->succ_begin(),
 459                          se = (*i)->succ_end(); si != se; ++si)
 460                     Out << (*i)->getInsts()->getBeginAddr() << ":o -> "
 461                         << (*si)->getInsts()->getBeginAddr() << ":a\n";
 462             }
 463             Out << "}\n";
 464
 465             ++filenum;
 466         }
 467     }
 468 }