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  *
  14  */
  15 Disassembler * createLLVMDisassembler(const std::string& filename, InformationManager* manager) {
  16         if (filename == "")
  17                 return NULL;
  18
  19         std::unique_ptr<Binary> o;
  20         o.reset(createBinary(filename).get());
  21         Binary * op = o.release();
  22
  23         // ELFType<endian, maxalign, 64bit>
  24         if (ELF32LEObjectFile * object = dyn_cast<ELF32LEObjectFile>(op)) {
  25                 return new LLVMDisassembler<ELFType<support::little, 2, false>>(filename, manager, object);
  26         }
  27         if (ELF64LEObjectFile * object = dyn_cast<ELF64LEObjectFile>(op)) {
  28                 return new LLVMDisassembler<ELFType<support::little, 2, true>>(filename, manager, object);
  29         }
  30         if (ELF32BEObjectFile * object = dyn_cast<ELF32BEObjectFile>(op)) {
  31                 return new LLVMDisassembler<ELFType<support::big, 2, false>>(filename, manager, object);
  32         }
  33         if (ELF64BEObjectFile * object = dyn_cast<ELF64BEObjectFile>(op)) {
  34                 return new LLVMDisassembler<ELFType<support::big, 2, true>>(filename, manager, object);
  35         }
  36
  37         return NULL;
  38 }
  39
  40 /*
  41  * TODO: fallback code falls die Datei kein ELF/PE/COFF/MacO/.. binary
  42  * ist sondern z.B. einfach nur Instruktionen oder ein Bootsektor oder
  43  * foo
  44  */
  45 template <typename ELFT>
  46 LLVMDisassembler<ELFT>::LLVMDisassembler(const std::string& filename,
  47                                          InformationManager* manager,
  48                                          ELFObjectFile<ELFT>* file)
  49         : Disassembler(filename, manager)
  50         , logger(log4cxx::Logger::getLogger("LLVMDisassembler"))
  51         , triple("unknown-unknown-unknown")
  52         , manager(manager)
  53 {
  54         LOG4CXX_DEBUG(logger, "Handling file" << filename);
  55
  56         if (!file) {
  57                 auto result = createBinary(filename);
  58
  59                 error_code ec;
  60                 if ((ec = result.getError())) {
  61                         LOG4CXX_ERROR(logger, "Failed to load Binary" << ec.message());
  62                         binary = NULL;
  63                         return;
  64                 }
  65
  66                 binary.reset(result.get());
  67
  68                 o = dyn_cast<ELFObjectFile<ELFT>>(binary.get());
  69         } else {
  70                 o = file;
  71                 binary.reset(file);
  72         }
  73
  74         triple.setArch(Triple::ArchType(o->getArch()));
  75         std::string tripleName(triple.getTriple());
  76
  77         LOG4CXX_INFO(logger, "Architecture " << tripleName);
  78
  79
  80         std::string es;
  81         target = TargetRegistry::lookupTarget("", triple, es);
  82         if (!target) {
  83                 LOG4CXX_ERROR(logger, es);
  84                 return;
  85         }
  86
  87         LOG4CXX_INFO(logger, "Target " << target->getName());
  88
  89         MRI.reset(target->createMCRegInfo(tripleName));
  90         if (!MRI) {
  91                 LOG4CXX_ERROR(logger, "no register info for target " << tripleName);
  92                 return;
  93         }
  94
  95         // Set up disassembler.
  96         AsmInfo.reset(target->createMCAsmInfo(*MRI, tripleName));
  97         if (!AsmInfo) {
  98                 LOG4CXX_ERROR(logger, "no assembly info for target " << tripleName);
  99                 return;
 100         }
 101
 102         STI.reset(target->createMCSubtargetInfo(tripleName, "", ""));
 103         if (!STI) {
 104                 LOG4CXX_ERROR(logger, "no subtarget info for target " << tripleName);
 105                 return;
 106         }
 107
 108         MII.reset(target->createMCInstrInfo());
 109         if (!MII) {
 110                 LOG4CXX_ERROR(logger, "no instruction info for target " << tripleName);
 111                 return;
 112         }
 113
 114         MOFI.reset(new MCObjectFileInfo);
 115         MCContext Ctx(AsmInfo.get(), MRI.get(), MOFI.get());
 116
 117         DisAsm.reset(target->createMCDisassembler(*STI, Ctx));
 118         if (!DisAsm) {
 119                 LOG4CXX_ERROR(logger, "no disassembler for target " << tripleName);
 120                 return;
 121         }
 122         RelInfo.reset(
 123                 target->createMCRelocationInfo(tripleName, Ctx));
 124         if (RelInfo) {
 125                 Symzer.reset(
 126                         MCObjectSymbolizer::createObjectSymbolizer(Ctx, std::move(RelInfo), o));
 127                 if (Symzer)
 128                         DisAsm->setSymbolizer(std::move(Symzer));
 129         }
 130         RelInfo.release();
 131         Symzer.release();
 132
 133         MIA.reset(target->createMCInstrAnalysis(MII.get()));
 134         if (!MIA) {
 135                 LOG4CXX_ERROR(logger, "no instruction analysis for target " << tripleName);
 136                 return;
 137         }
 138
 139         int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
 140         IP.reset(target->createMCInstPrinter(AsmPrinterVariant, *AsmInfo, *MII, *MRI, *STI));
 141         if (!IP) {
 142                 LOG4CXX_ERROR(logger, "no instruction printer for target " << tripleName);
 143                 return;
 144         }
 145
 146         IP->setPrintImmHex(llvm::HexStyle::C);
 147         IP->setPrintImmHex(true);
 148
 149         std::unique_ptr<MCObjectDisassembler> OD(
 150                 new MCObjectDisassembler(*o, *DisAsm, *MIA));
 151         Mod.reset(OD->buildModule(false));
 152 }
 153
 154 template <typename ELFT>
 155 void LLVMDisassembler<ELFT>::start() {
 156         readSymbols();
 157         readSections();
 158         disassemble();
 159 }
 160
 161 template <typename ELFT>
 162 LLVMDisassembler<ELFT>::~LLVMDisassembler() {
 163         std::for_each(functions.begin(), functions.end(),
 164                       [](std::pair<uint64_t,LLVMFunction*> it) {
 165                               delete it.second;
 166                       });
 167         std::for_each(blocks.begin(), blocks.end(),
 168                       [](std::pair<uint64_t, LLVMBasicBlock*> it) {
 169                               delete it.second;
 170                       });
 171 }
 172
 173 template <typename ELFT>
 174 Function* LLVMDisassembler<ELFT>::disassembleFunctionAt(uint64_t address, const std::string& name) {
 175         SectionRef text_section = sections[".text"];
 176         uint64_t base_address, size;
 177         text_section.getAddress(base_address);
 178         text_section.getSize(size);
 179
 180         if (address < base_address ||
 181             address >= base_address + size) {
 182                 return NULL;
 183         }
 184
 185         if (functions.find(address) != functions.end()) {
 186                 return functions[address];
 187         }
 188
 189         LLVMFunction * function;
 190         if (name == "") {
 191                 std::stringstream s;
 192                 s << "<Unnamed 0x" << std::hex << address << ">";
 193                 function = new LLVMFunction(s.str(), address);
 194         } else {
 195                 function = new LLVMFunction(name, address);
 196         }
 197         functions.insert(std::make_pair(address, function));
 198
 199         disassembleFunction(function);
 200
 201         return function;
 202 }
 203
 204 template <typename ELFT>
 205 void LLVMDisassembler<ELFT>::disassembleFunction(LLVMFunction* function) {
 206         std::stack<LLVMBasicBlock*> remaining_blocks;
 207         SectionRef text_section = sections[".text"];
 208         StringRef bytes;
 209         text_section.getContents(bytes);
 210         StringRefMemoryObject ref(bytes);
 211
 212         LOG4CXX_DEBUG(logger, "Handling function " << function->getName());
 213
 214         LLVMBasicBlock * block = new LLVMBasicBlock(function->getStartAddress(), this);
 215         remaining_blocks.push(block);
 216         blocks.insert(std::make_pair(block->getStartAddress(), block));
 217         function->addBasicBlock(block);
 218
 219         while (remaining_blocks.size()) {
 220                 LLVMBasicBlock * current_block = remaining_blocks.top();
 221                 remaining_blocks.pop();
 222
 223                 LOG4CXX_DEBUG(logger, "Handling Block starting at " << std::hex << current_block->getStartAddress());
 224
 225                 uint64_t inst_size;
 226                 uint64_t base_address;
 227                 text_section.getAddress(base_address);
 228                 uint64_t current_address = current_block->getStartAddress() - base_address;
 229                 while(true) {
 230                         MCInst inst;
 231                         std::string buf;
 232                         llvm::raw_string_ostream s(buf);
 233
 234                         if(llvm::MCDisassembler::Success ==
 235                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 236                                 uint64_t jmptarget;
 237
 238                                 if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 239                                         jmptarget += base_address;
 240                                         if (!MIA->isIndirectBranch(inst)) {
 241                                                 if (MIA->isCall(inst)) {
 242                                                         if (functions.find(jmptarget) == functions.end()) {
 243                                                                 disassembleFunctionAt(jmptarget);
 244                                                         }
 245                                                 } else {
 246                                                         current_block->setNextBlock(0, jmptarget);
 247                                                         if (blocks.find(jmptarget) == blocks.end()) {
 248                                                                 LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
 249                                                                 blocks.insert(std::make_pair(block->getStartAddress(), block));
 250                                                                 function->addBasicBlock(block);
 251                                                                 remaining_blocks.push(block);
 252                                                         } else {
 253                                                                 LOG4CXX_DEBUG(logger, "Reusing Block starting at " << std::hex << current_block->getStartAddress());
 254                                                                 function->addBasicBlock(blocks.find(jmptarget)->second);
 255                                                         }
 256                                                         if (MIA->isConditionalBranch(inst)) {
 257                                                                 jmptarget = base_address + current_address + inst_size;
 258                                                                 current_block->setNextBlock(1, jmptarget);
 259                                                                 if (blocks.find(jmptarget) == blocks.end()) {
 260                                                                         LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
 261                                                                         blocks.insert(std::make_pair(block->getStartAddress(), block));
 262                                                                         function->addBasicBlock(block);
 263                                                                         remaining_blocks.push(block);
 264                                                                 } else {
 265                                                                         LOG4CXX_DEBUG(logger, "Reusing Block starting at " << std::hex << current_block->getStartAddress());
 266                                                                         function->addBasicBlock(blocks.find(jmptarget)->second);
 267                                                                 }
 268                                                         }
 269                                                 }
 270                                         }
 271                                 }
 272                         } else {
 273                                 inst_size = 0;
 274                         }
 275
 276
 277                         if (inst_size == 0 || MIA->isTerminator(inst) || MIA->isBranch(inst)) {
 278                                 current_block->setEndAddress(current_address + base_address + inst_size);
 279                                 LOG4CXX_DEBUG(logger, "Finished Block at " << std::hex <<
 280                                               current_block->getEndAddress());
 281                                 break;
 282                         }
 283                         current_address += inst_size;
 284                 }
 285         }
 286         splitBlocks(function);
 287         LOG4CXX_DEBUG(logger, "Finished function " << function->getName());
 288         manager->signal_new_function(function);
 289 }
 290
 291 template <typename ELFT>
 292 void LLVMDisassembler<ELFT>::disassemble() {
 293         SectionRef text_section = sections[".text"];
 294         std::vector<LLVMFunction*> remaining_functions;
 295
 296         // Assume all function symbols actually start a real function
 297         for (auto x = symbols.begin(); x != symbols.end(); ++x) {
 298                 uint64_t result;
 299                 bool contains;
 300                 SymbolRef::Type symbol_type;
 301
 302
 303                 if (text_section.containsSymbol(x->second, contains) || !contains)
 304                         continue;
 305
 306                 if (x->second.getType(symbol_type)
 307                     || SymbolRef::ST_Function != symbol_type)
 308                         continue;
 309
 310                 if (!x->second.getAddress(result)) {
 311                         LLVMFunction * fun = new LLVMFunction(x->first, result);
 312                         remaining_functions.push_back(fun);
 313                         functions.insert(std::make_pair(result, fun));
 314                         LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
 315                 }
 316         }
 317
 318         for (LLVMFunction* function : remaining_functions) {
 319                 disassembleFunction(function);
 320         }
 321
 322         if (binary->isELF()) {
 323                 typedef ELFFile<ELFT> ELFO;
 324                 const ELFO * elffile = o->getELFFile();
 325                 const typename ELFO::Elf_Ehdr * header = elffile->getHeader();
 326
 327                 _entryAddress = header->e_entry;
 328                 LOG4CXX_DEBUG(logger, "Adding entryAddress at: " << std::hex << _entryAddress);
 329                 std::stringstream s;
 330                 s << "<_start 0x" << std::hex << _entryAddress << ">";
 331
 332                 disassembleFunctionAt(_entryAddress, s.str());
 333         }
 334
 335         if (functions.empty()) {
 336                 uint64_t text_entry;
 337                 text_section.getAddress(text_entry);
 338                 LOG4CXX_INFO(logger, "No Symbols found, starting at the beginning of the text segment");
 339                 disassembleFunctionAt(text_entry);
 340         }
 341 }
 342
 343 template <typename ELFT>
 344 void LLVMDisassembler<ELFT>::splitBlocks(LLVMFunction* function) {
 345         SectionRef text_section = sections[".text"];
 346         StringRef bytes;
 347         text_section.getContents(bytes);
 348         StringRefMemoryObject ref(bytes);
 349
 350         // Split blocks where jumps are going inside the block
 351         for (auto it = function->blocks().begin();
 352              it != function->blocks().end();
 353              ++it) {
 354                 BasicBlock * current_block = it->second;
 355                 uint64_t inst_size;
 356                 uint64_t base_address;
 357                 text_section.getAddress(base_address);
 358                 uint64_t current_address = current_block->getStartAddress() - base_address;
 359                 while(current_block->getEndAddress() - base_address > current_address) {
 360                         MCInst inst;
 361                         std::string buf;
 362                         llvm::raw_string_ostream s(buf);
 363
 364                         if(llvm::MCDisassembler::Success ==
 365                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 366                                 // See if some other block starts here
 367                                 auto other = blocks.find(current_address + inst_size + base_address);
 368
 369                                 // Special case, other block starts here but we are at the end anyway
 370                                 if (other != blocks.end()) {
 371                                         uint64_t endaddress = current_address + inst_size + base_address;
 372                                         if (endaddress != current_block->getEndAddress()) {
 373                                                 LOG4CXX_DEBUG(logger, "Shortening block starting at "
 374                                                               << std::hex
 375                                                               << current_block->getStartAddress()
 376                                                               << " now ending at "
 377                                                               << other->first);
 378                                                 function->addBasicBlock(other->second);
 379                                                 current_block->setEndAddress(endaddress);
 380                                                 current_block->setNextBlock(0, other->first);
 381                                                 current_block->setNextBlock(1, 0);
 382                                         }
 383                                 }
 384                         } else {
 385                                 inst_size = 1;
 386                         }
 387                         current_address += inst_size;
 388                 }
 389         }
 390 }
 391
 392 template <typename ELFT>
 393 void LLVMDisassembler<ELFT>::readSymbols() {
 394         error_code ec;
 395         symbol_iterator si(o->symbol_begin()), se(o->symbol_end());
 396         for (; si != se; ++si) {
 397                 StringRef name;
 398                 if ((ec = si->getName(name))) {
 399                         LOG4CXX_ERROR(logger, ec.message());
 400                         break;
 401                 }
 402                 LOG4CXX_DEBUG(logger, "Added symbol " << name.str());
 403                 symbols.insert(make_pair(name.str(), *si));
 404         }
 405 }
 406
 407 template <typename ELFT>
 408 void LLVMDisassembler<ELFT>::readSections() {
 409         error_code ec;
 410         section_iterator i(o->section_begin()), e(o->section_end());
 411         for (; i != e; ++i) {
 412                 StringRef name;
 413                 if ((ec = i->getName(name))) {
 414                         LOG4CXX_ERROR(logger, ec.message());
 415                         break;
 416                 }
 417                 LOG4CXX_DEBUG(logger, "Added section " << name.str());
 418                 sections.insert(make_pair(name.str(), *i));
 419         }
 420
 421 }
 422
 423 template <typename ELFT>
 424 void LLVMDisassembler<ELFT>::forEachFunction(std::function<void (uint64_t, Function*)> callback) {
 425         std::for_each(functions.begin(), functions.end(),
 426                       [&](std::pair<uint64_t, LLVMFunction*> x) {
 427                               callback(x.first, x.second);
 428                       });
 429 }
 430
 431 template <typename ELFT>
 432 void LLVMDisassembler<ELFT>::printEachInstruction(uint64_t start, uint64_t end,
 433                                                   std::function<void (uint8_t*, size_t,
 434                                                                           const std::string&,
 435                                                                           const std::string&)> fun) {
 436         SectionRef text_section = sections[".text"];
 437         uint64_t base_address;
 438         text_section.getAddress(base_address);
 439         uint64_t current_address = start - base_address;
 440
 441         StringRef bytes;
 442         text_section.getContents(bytes);
 443         StringRefMemoryObject ref(bytes);
 444
 445         while (current_address < end - base_address) {
 446                 uint64_t inst_size;
 447                 MCInst inst;
 448                 std::string buf;
 449                 llvm::raw_string_ostream s(buf);
 450
 451                 if(llvm::MCDisassembler::Success ==
 452                    DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 453
 454                         uint8_t bytes[inst_size+2];
 455                         ref.readBytes(current_address, inst_size, bytes);
 456
 457                         uint64_t jmptarget;
 458                         std::string ref("");
 459                         IP->printInst(&inst, s, "");
 460                         if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 461                                 std::stringstream stream;
 462                                 if (MIA->isCall(inst))
 463                                         stream << "function:";
 464                                 else
 465                                         stream << "block:";
 466
 467                                 stream << std::hex << (base_address + jmptarget);
 468                                 ref = stream.str();
 469                         }
 470
 471
 472                         fun(bytes, inst_size, s.str(), ref);
 473                 } else {
 474                         LOG4CXX_WARN(logger, "Invalid byte at" << std::hex << current_address + base_address);
 475                         fun(NULL, 0, "Invalid Byte", "");
 476                         inst_size = 1;
 477                 }
 478
 479                 current_address += inst_size;
 480         }
 481 }