src/disassembler/llvm/LLVMDisassembler.cxx

   1 #include "disassembler/llvm/LLVMDisassembler.hxx"
   2 #include "disassembler/llvm/LLVMBasicBlock.hxx"
   3 #include "disassembler/llvm/LLVMFunction.hxx"
   4
   5 #include <stack>
   6 #include <algorithm>
   7
   8 using namespace llvm;
   9 using namespace llvm::object;
  10 using std::error_code;
  11
  12 /*
  13  * TODO: fallback code falls die Datei kein ELF/PE/COFF/MacO/.. binary
  14  * ist sondern z.B. einfach nur Instruktionen oder ein Bootsektor oder
  15  * foo
  16  */
  17 LLVMDisassembler::LLVMDisassembler(const std::string& filename)
  18     : Disassembler(filename)
  19     , logger(log4cxx::Logger::getLogger("LLVMDisassembler"))
  20     , triple("unknown-unknown-unknown")
  21 {
  22     LOG4CXX_DEBUG(logger, "Handling file" << filename);
  23     auto result = createBinary(filename);
  24
  25     error_code ec;
  26     if ((ec = result.getError())) {
  27         LOG4CXX_ERROR(logger, "Failed to load Binary" << ec.message());
  28         binary = NULL;
  29         return;
  30     }
  31
  32     binary.reset(result.get());
  33
  34     o = dyn_cast<ObjectFile>(binary.get());
  35
  36     triple.setArch(Triple::ArchType(o->getArch()));
  37     std::string tripleName(triple.getTriple());
  38
  39     LOG4CXX_INFO(logger, "Architecture " << tripleName);
  40
  41
  42     std::string es;
  43     target = TargetRegistry::lookupTarget("", triple, es);
  44     if (!target) {
  45         LOG4CXX_ERROR(logger, es);
  46         return;
  47     }
  48
  49     LOG4CXX_INFO(logger, "Target " << target->getName());
  50
  51     MRI.reset(target->createMCRegInfo(tripleName));
  52     if (!MRI) {
  53         LOG4CXX_ERROR(logger, "no register info for target " << tripleName);
  54         return;
  55     }
  56
  57     // Set up disassembler.
  58     AsmInfo.reset(target->createMCAsmInfo(*MRI, tripleName));
  59     if (!AsmInfo) {
  60         LOG4CXX_ERROR(logger, "no assembly info for target " << tripleName);
  61         return;
  62     }
  63
  64     STI.reset(target->createMCSubtargetInfo(tripleName, "", ""));
  65     if (!STI) {
  66         LOG4CXX_ERROR(logger, "no subtarget info for target " << tripleName);
  67         return;
  68     }
  69
  70     MII.reset(target->createMCInstrInfo());
  71     if (!MII) {
  72         LOG4CXX_ERROR(logger, "no instruction info for target " << tripleName);
  73         return;
  74     }
  75
  76     MOFI.reset(new MCObjectFileInfo);
  77     MCContext Ctx(AsmInfo.get(), MRI.get(), MOFI.get());
  78
  79     DisAsm.reset(target->createMCDisassembler(*STI, Ctx));
  80     if (!DisAsm) {
  81         LOG4CXX_ERROR(logger, "no disassembler for target " << tripleName);
  82         return;
  83     }
  84     RelInfo.reset(
  85         target->createMCRelocationInfo(tripleName, Ctx));
  86     if (RelInfo) {
  87         Symzer.reset(
  88             MCObjectSymbolizer::createObjectSymbolizer(Ctx, std::move(RelInfo), o));
  89         if (Symzer)
  90             DisAsm->setSymbolizer(std::move(Symzer));
  91     }
  92     RelInfo.release();
  93     Symzer.release();
  94
  95     MIA.reset(target->createMCInstrAnalysis(MII.get()));
  96     if (!MIA) {
  97         LOG4CXX_ERROR(logger, "no instruction analysis for target " << tripleName);
  98         return;
  99     }
 100
 101     int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
 102     IP.reset(target->createMCInstPrinter(AsmPrinterVariant, *AsmInfo, *MII, *MRI, *STI));
 103     if (!IP) {
 104         LOG4CXX_ERROR(logger, "no instruction printer for target " << tripleName);
 105         return;
 106     }
 107
 108     IP->setPrintImmHex(llvm::HexStyle::C);
 109     IP->setPrintImmHex(true);
 110
 111     std::unique_ptr<MCObjectDisassembler> OD(
 112         new MCObjectDisassembler(*o, *DisAsm, *MIA));
 113     Mod.reset(OD->buildModule(false));
 114
 115     readSymbols();
 116     readSections();
 117     disassemble();
 118 }
 119
 120 LLVMDisassembler::~LLVMDisassembler() {
 121     std::for_each(functions.begin(), functions.end(),
 122                   [](std::pair<uint64_t,LLVMFunction*> it) {
 123                       delete it.second;
 124                   });
 125     std::for_each(blocks.begin(), blocks.end(),
 126                   [](std::pair<uint64_t, LLVMBasicBlock*> it) {
 127                       delete it.second;
 128                   });
 129 }
 130
 131 void LLVMDisassembler::disassemble() {
 132     std::stack<LLVMFunction*> remaining_functions;
 133     std::stack<LLVMBasicBlock*> remaining_blocks;
 134     SectionRef text_section = sections[".text"];
 135
 136         // Assume all function symbols actually start a real function
 137     for (auto x = symbols.begin(); x != symbols.end(); ++x) {
 138         uint64_t result;
 139         bool contains;
 140         SymbolRef::Type symbol_type;
 141
 142
 143         if (text_section.containsSymbol(x->second, contains) || !contains)
 144             continue;
 145
 146         if (x->second.getType(symbol_type)
 147             || SymbolRef::ST_Function != symbol_type)
 148             continue;
 149
 150         if (!x->second.getAddress(result)) {
 151             LLVMFunction * fun = new LLVMFunction(x->first, result);
 152             remaining_functions.push(fun);
 153             functions.insert(std::make_pair(result, fun));
 154             LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
 155         }
 156     }
 157
 158     if (binary->isELF()) {
 159         bool is64bit = (binary->getData()[4] == 0x02);
 160
 161         uint64_t entry(0);
 162         for (int i(0); i < (is64bit? 8 : 4); ++i) {
 163             if (binary->isLittleEndian()) {
 164                 entry |= (unsigned int)((unsigned char)binary->getData()[0x18 + i]) << 8*i;
 165             } else {
 166                 entry = entry << 8;
 167                 entry |= (unsigned char)binary->getData()[0x18 + i];
 168             }
 169         }
 170         if (functions.find(entry) == functions.end()) {
 171             LOG4CXX_DEBUG(logger, "Adding entry at: " << std::hex << entry);
 172             std::stringstream s;
 173             s << "<_start 0x" << std::hex << entry << ">";
 174             LLVMFunction * fun = new LLVMFunction(s.str(), entry);
 175             functions.insert(std::make_pair(entry, fun));
 176             remaining_functions.push(fun);
 177         }
 178     }
 179
 180     if (functions.empty()) {
 181         uint64_t text_entry;
 182         text_section.getAddress(text_entry);
 183         LOG4CXX_INFO(logger, "No Symbols found, starting at the beginning of the text segment");
 184
 185         std::stringstream s;
 186         s << "<Unnamed 0x" << std::hex << text_entry << ">";
 187         LLVMFunction * fun = new LLVMFunction(s.str(), text_entry);
 188         functions.insert(std::make_pair(text_entry, fun));
 189         remaining_functions.push(fun);
 190     }
 191
 192     StringRef bytes;
 193     text_section.getContents(bytes);
 194     StringRefMemoryObject ref(bytes);
 195
 196     while (remaining_functions.size()) {
 197         LLVMFunction * current_function = remaining_functions.top();
 198         remaining_functions.pop();
 199
 200         LOG4CXX_DEBUG(logger, "Handling function " << current_function->getName());
 201
 202         LLVMBasicBlock * block = new LLVMBasicBlock(current_function->getStartAddress(), this);
 203         remaining_blocks.push(block);
 204         blocks.insert(std::make_pair(block->getStartAddress(), block));
 205
 206         while (remaining_blocks.size()) {
 207             LLVMBasicBlock * current_block = remaining_blocks.top();
 208             remaining_blocks.pop();
 209
 210             LOG4CXX_DEBUG(logger, "Handling Block starting at " << std::hex << current_block->getStartAddress());
 211
 212             uint64_t inst_size;
 213             uint64_t base_address;
 214             text_section.getAddress(base_address);
 215             uint64_t current_address = current_block->getStartAddress() - base_address;
 216             while(true) {
 217                 MCInst inst;
 218                 std::string buf;
 219                 llvm::raw_string_ostream s(buf);
 220
 221                 if(llvm::MCDisassembler::Success ==
 222                    DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 223
 224                     uint64_t jmptarget;
 225                     if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 226                         jmptarget += base_address;
 227                         if (!MIA->isIndirectBranch(inst)) {
 228                             if (MIA->isCall(inst)) {
 229                                 if (functions.find(jmptarget) == functions.end()) {
 230                                     std::stringstream s;
 231                                     s << "<Unnamed 0x" << std::hex << jmptarget << ">";
 232                                     LLVMFunction * fun = new LLVMFunction(s.str(), jmptarget);
 233                                     functions.insert(std::make_pair(jmptarget, fun));
 234                                     remaining_functions.push(fun);
 235                                 }
 236                             } else {
 237                                                                 current_block->setNextBlock(0, jmptarget);
 238                                 if (blocks.find(jmptarget) == blocks.end()) {
 239                                     LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
 240                                     blocks.insert(std::make_pair(block->getStartAddress(), block));
 241                                     remaining_blocks.push(block);
 242                                 }
 243                                 if (MIA->isConditionalBranch(inst)) {
 244                                     jmptarget = base_address + current_address + inst_size;
 245                                                                         current_block->setNextBlock(1, jmptarget);
 246                                     if (blocks.find(jmptarget) == blocks.end()) {
 247                                         LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
 248                                         blocks.insert(std::make_pair(block->getStartAddress(), block));
 249                                         remaining_blocks.push(block);
 250                                     }
 251                                 }
 252                             }
 253                         }
 254                     }
 255                 } else {
 256                     inst_size = 0;
 257                 }
 258
 259
 260                 if (inst_size == 0 || MIA->isTerminator(inst) || MIA->isBranch(inst)) {
 261                     current_block->setEndAddress(current_address + base_address + inst_size);
 262                     LOG4CXX_DEBUG(logger, "Finished Block at " << std::hex <<
 263                                   current_block->getEndAddress());
 264                     break;
 265                 }
 266                 current_address += inst_size;
 267             }
 268         }
 269         LOG4CXX_DEBUG(logger, "Finished function " << current_function->getName());
 270     }
 271
 272         // Split blocks where jumps are going inside the block
 273         for (auto it = blocks.begin(); it != blocks.end(); ++it) {
 274                 LLVMBasicBlock * current_block = it->second;
 275                 uint64_t inst_size;
 276                 uint64_t base_address;
 277                 text_section.getAddress(base_address);
 278                 uint64_t current_address = current_block->getStartAddress() - base_address;
 279                 while(current_block->getEndAddress() - base_address > current_address) {
 280                         MCInst inst;
 281                         std::string buf;
 282                         llvm::raw_string_ostream s(buf);
 283
 284                         if(llvm::MCDisassembler::Success ==
 285                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 286                                 auto other = blocks.find(current_address + inst_size + base_address);
 287
 288                                 if (other != blocks.end()) {
 289                                         uint64_t endaddress = current_address + inst_size + base_address;
 290                                         if (endaddress != current_block->getEndAddress()) {
 291                                                 LOG4CXX_DEBUG(logger, "Shortening block starting at "
 292                                                                           << std::hex
 293                                                                           << current_block->getStartAddress()
 294                                                                           << " now ending at "
 295                                                                           << other->first);
 296                                                 current_block->setEndAddress(endaddress);
 297                                                 current_block->setNextBlock(0, other->first);
 298                                                 current_block->setNextBlock(1, 0);
 299                                         }
 300                                 }
 301                         } else {
 302                                 inst_size = 1;
 303                         }
 304                         current_address += inst_size;
 305                 }
 306         }
 307 }
 308
 309 void LLVMDisassembler::readSymbols() {
 310     error_code ec;
 311     symbol_iterator si(o->symbol_begin()), se(o->symbol_end());
 312     for (; si != se; ++si) {
 313         StringRef name;
 314         if ((ec = si->getName(name))) {
 315             LOG4CXX_ERROR(logger, ec.message());
 316             break;
 317         }
 318         LOG4CXX_DEBUG(logger, "Added symbol " << name.str());
 319         symbols.insert(make_pair(name.str(), *si));
 320     }
 321 }
 322
 323 void LLVMDisassembler::readSections() {
 324     error_code ec;
 325     section_iterator i(o->section_begin()), e(o->section_end());
 326     for (; i != e; ++i) {
 327         StringRef name;
 328         if ((ec = i->getName(name))) {
 329             LOG4CXX_ERROR(logger, ec.message());
 330             break;
 331         }
 332         LOG4CXX_DEBUG(logger, "Added section " << name.str());
 333         sections.insert(make_pair(name.str(), *i));
 334     }
 335
 336 }
 337
 338 void LLVMDisassembler::forEachFunction(std::function<void (uint64_t, Function*)> callback) {
 339     std::for_each(functions.begin(), functions.end(),
 340                   [&](std::pair<uint64_t, LLVMFunction*> x) {
 341                       callback(x.first, x.second);
 342                   });
 343 }
 344
 345 void LLVMDisassembler::printEachInstruction(uint64_t start, uint64_t end,
 346                                                                                         std::function<void (uint8_t*, size_t, const std::string&)> fun) {
 347     SectionRef text_section = sections[".text"];
 348     uint64_t base_address;
 349     text_section.getAddress(base_address);
 350     uint64_t current_address = start - base_address;
 351
 352     StringRef bytes;
 353     text_section.getContents(bytes);
 354     StringRefMemoryObject ref(bytes);
 355
 356     while (current_address < end - base_address) {
 357         uint64_t inst_size;
 358         MCInst inst;
 359                 std::string buf;
 360                 llvm::raw_string_ostream s(buf);
 361
 362         if(llvm::MCDisassembler::Success ==
 363            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 364
 365             uint8_t bytes[inst_size+2];
 366             ref.readBytes(current_address, inst_size, bytes);
 367
 368                         uint64_t jmptarget;
 369                         if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 370                                 std::stringstream stream;
 371                                 stream << std::hex << (base_address + jmptarget);
 372                                 IP->printInst(&inst, s, stream.str());
 373                         } else
 374                                 IP->printInst(&inst, s, "");
 375
 376                         fun(bytes, inst_size, s.str());
 377         } else {
 378                         LOG4CXX_WARN(logger, "Invalid byte at" << std::hex << current_address + base_address);
 379                         fun(NULL, 0, "Invalid Byte");
 380                         inst_size = 1;
 381                 }
 382
 383                 current_address += inst_size;
 384     }
 385 }