src/disassembler/llvm/LLVMDisassembler.cxx

   1 #include "disassembler/llvm/LLVMDisassembler.hxx"
   2 #include "core/InformationManager.hxx"
   3 #include "core/Function.hxx"
   4 #include "core/BasicBlock.hxx"
   5
   6 #include <stack>
   7 #include <algorithm>
   8 #include <cassert>
   9
  10 using namespace llvm;
  11 using namespace llvm::object;
  12 using std::error_code;
  13
  14 namespace {
  15         class COFFT {
  16
  17         };
  18 }
  19
  20 /*
  21  *
  22  */
  23 Disassembler * createLLVMDisassembler(const std::string& filename, InformationManager* manager) {
  24         if (filename == "")
  25                 return NULL;
  26
  27         std::unique_ptr<Binary> o;
  28         o.reset(createBinary(filename).get());
  29         Binary * op = o.release();
  30
  31         // ELFType<endian, maxalign, 64bit>
  32         if (ELF32LEObjectFile * object = dyn_cast<ELF32LEObjectFile>(op)) {
  33                 return new LLVMDisassembler<ELFType<support::little, 2, false>>(filename, manager, object);
  34         }
  35         if (ELF64LEObjectFile * object = dyn_cast<ELF64LEObjectFile>(op)) {
  36                 return new LLVMDisassembler<ELFType<support::little, 2, true>>(filename, manager, object);
  37         }
  38         if (ELF32BEObjectFile * object = dyn_cast<ELF32BEObjectFile>(op)) {
  39                 return new LLVMDisassembler<ELFType<support::big, 2, false>>(filename, manager, object);
  40         }
  41         if (ELF64BEObjectFile * object = dyn_cast<ELF64BEObjectFile>(op)) {
  42                 return new LLVMDisassembler<ELFType<support::big, 2, true>>(filename, manager, object);
  43         }
  44         if (COFFObjectFile * object = dyn_cast<COFFObjectFile>(op)) {
  45                 return new LLVMDisassembler<COFFT>(filename, manager, object);
  46         }
  47
  48         return NULL;
  49 }
  50
  51 /*
  52  * TODO: fallback code falls die Datei kein ELF/PE/COFF/MacO/.. binary
  53  * ist sondern z.B. einfach nur Instruktionen oder ein Bootsektor oder
  54  * foo
  55  */
  56 template <typename ELFT>
  57 LLVMDisassembler<ELFT>::LLVMDisassembler(const std::string& filename,
  58                                          InformationManager* manager,
  59                                          ObjectFile* file)
  60         : Disassembler()
  61         , logger(log4cxx::Logger::getLogger("disassembler.LLVMDisassembler"))
  62         , triple("unknown-unknown-unknown")
  63         , manager(manager)
  64 {
  65         LOG4CXX_DEBUG(logger, "Handling file " << filename);
  66
  67         if (!file) {
  68                 auto result = createBinary(filename);
  69
  70                 error_code ec;
  71                 if ((ec = result.getError())) {
  72                         LOG4CXX_ERROR(logger, "Failed to load Binary" << ec.message());
  73                         binary = NULL;
  74                         return;
  75                 }
  76
  77                 binary.reset(result.get());
  78
  79                 o = dyn_cast<ObjectFile>(binary.get());
  80         } else {
  81                 o = file;
  82                 binary.reset(file);
  83         }
  84
  85         triple.setArch(Triple::ArchType(o->getArch()));
  86         std::string tripleName(triple.getTriple());
  87
  88         LOG4CXX_INFO(logger, "Architecture " << tripleName);
  89
  90
  91         std::string es;
  92         target = TargetRegistry::lookupTarget("", triple, es);
  93         if (!target) {
  94                 LOG4CXX_ERROR(logger, es);
  95                 return;
  96         }
  97
  98         LOG4CXX_INFO(logger, "Target " << target->getName());
  99
 100         MRI.reset(target->createMCRegInfo(tripleName));
 101         if (!MRI) {
 102                 LOG4CXX_ERROR(logger, "no register info for target " << tripleName);
 103                 return;
 104         }
 105
 106         // Set up disassembler.
 107         AsmInfo.reset(target->createMCAsmInfo(*MRI, tripleName));
 108         if (!AsmInfo) {
 109                 LOG4CXX_ERROR(logger, "no assembly info for target " << tripleName);
 110                 return;
 111         }
 112
 113         STI.reset(target->createMCSubtargetInfo(tripleName, "", ""));
 114         if (!STI) {
 115                 LOG4CXX_ERROR(logger, "no subtarget info for target " << tripleName);
 116                 return;
 117         }
 118
 119         MII.reset(target->createMCInstrInfo());
 120         if (!MII) {
 121                 LOG4CXX_ERROR(logger, "no instruction info for target " << tripleName);
 122                 return;
 123         }
 124
 125         MOFI.reset(new MCObjectFileInfo);
 126         MCContext Ctx(AsmInfo.get(), MRI.get(), MOFI.get());
 127
 128         DisAsm.reset(target->createMCDisassembler(*STI, Ctx));
 129         if (!DisAsm) {
 130                 LOG4CXX_ERROR(logger, "no disassembler for target " << tripleName);
 131                 return;
 132         }
 133         RelInfo.reset(
 134                 target->createMCRelocationInfo(tripleName, Ctx));
 135         if (RelInfo) {
 136                 // Symzer.reset(
 137                 //      MCObjectSymbolizer::createObjectSymbolizer(Ctx, std::move(RelInfo), o));
 138                 // if (Symzer)
 139                 //      DisAsm->setSymbolizer(std::move(Symzer));
 140         }
 141         RelInfo.release();
 142         Symzer.release();
 143
 144         MIA.reset(target->createMCInstrAnalysis(MII.get()));
 145         if (!MIA) {
 146                 LOG4CXX_ERROR(logger, "no instruction analysis for target " << tripleName);
 147                 return;
 148         }
 149
 150         int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
 151         IP.reset(target->createMCInstPrinter(AsmPrinterVariant, *AsmInfo, *MII, *MRI, *STI));
 152         if (!IP) {
 153                 LOG4CXX_ERROR(logger, "no instruction printer for target " << tripleName);
 154                 return;
 155         }
 156
 157         IP->setPrintImmHex(llvm::HexStyle::C);
 158         IP->setPrintImmHex(true);
 159
 160         std::unique_ptr<MCObjectDisassembler> OD(
 161                 new MCObjectDisassembler(*o, *DisAsm, *MIA));
 162         Mod.reset(OD->buildModule(false));
 163
 164         readSections();
 165 }
 166
 167 template <typename ELFT>
 168 void LLVMDisassembler<ELFT>::start() {
 169         readSymbols();
 170         disassemble();
 171         readDynamicSymbols();
 172 }
 173
 174 template <typename ELFT>
 175 LLVMDisassembler<ELFT>::~LLVMDisassembler() {}
 176
 177 template <typename ELFT>
 178 Function* LLVMDisassembler<ELFT>::disassembleFunctionAt(uint64_t address, const std::string& name) {
 179         Function * function;
 180         SectionRef text_section = sections[".text"];
 181         uint64_t base_address, size;
 182         text_section.getAddress(base_address);
 183         text_section.getSize(size);
 184
 185         if (address < base_address ||
 186             address >= base_address + size) {
 187                 return NULL;
 188         }
 189
 190         if (NULL == (function = manager->getFunction(address))) {
 191
 192                 if (name == "") {
 193                         std::stringstream s;
 194                         s << "<Unnamed 0x" << std::hex << address << ">";
 195                         function = manager->newFunction(address);
 196                         function->setName(s.str());
 197                 } else {
 198                         function = manager->newFunction(address);
 199                         function->setName(name);
 200                 }
 201                 disassembleFunction(function);
 202         }
 203
 204         return function;
 205 }
 206
 207 template <typename ELFT>
 208 void LLVMDisassembler<ELFT>::disassembleFunction(Function* function) {
 209         std::vector<uint64_t> called_functions;
 210         std::stack<BasicBlock*> remaining_blocks;
 211         /* TODO:
 212          * Do all blocks get added properly? We should take care to remove
 213          * the other ones at the end of the function!
 214          */
 215         std::map<uint64_t, BasicBlock*> new_blocks;
 216         SectionRef text_section = sections[".text"];
 217         StringRef bytes;
 218         text_section.getContents(bytes);
 219         StringRefMemoryObject ref(bytes);
 220
 221         LOG4CXX_DEBUG(logger, "Handling function " << function->getName());
 222
 223         BasicBlock * block = manager->newBasicBlock(function->getStartAddress());
 224         remaining_blocks.push(block);
 225         new_blocks.insert(std::make_pair(block->getStartAddress(), block));
 226         function->addBasicBlock(block);
 227
 228         while (remaining_blocks.size()) {
 229                 BasicBlock * current_block = remaining_blocks.top();
 230                 remaining_blocks.pop();
 231
 232                 LOG4CXX_DEBUG(logger, "Handling Block starting at " << std::hex
 233                               << current_block->getStartAddress());
 234
 235                 uint64_t inst_size;
 236                 uint64_t base_address;
 237                 text_section.getAddress(base_address);
 238                 uint64_t current_address = current_block->getStartAddress() - base_address;
 239                 while(true) {
 240                         MCInst inst;
 241                         std::string buf;
 242                         llvm::raw_string_ostream s(buf);
 243
 244                         if(llvm::MCDisassembler::Success ==
 245                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 246                                 uint64_t jmptarget;
 247
 248                                 if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 249                                         jmptarget += base_address;
 250                                         if (!MIA->isIndirectBranch(inst)) {
 251                                                 if (MIA->isCall(inst)) {
 252                                                         if (NULL == manager->getFunction(jmptarget))
 253                                                                 called_functions.push_back(jmptarget);
 254                                                 } else {
 255                                                         current_block->setNextBlock(0, jmptarget);
 256                                                         if (new_blocks.find(jmptarget) == new_blocks.end()) {
 257                                                                 BasicBlock * block = manager->newBasicBlock(jmptarget);
 258                                                                 assert(block);
 259                                                                 new_blocks.insert(std::make_pair(block->getStartAddress(), block));
 260                                                                 function->addBasicBlock(block);
 261                                                                 remaining_blocks.push(block);
 262                                                         } else {
 263                                                                 LOG4CXX_DEBUG(logger, "Reusing Block starting at " << std::hex
 264                                                                               << current_block->getStartAddress());
 265                                                                 function->addBasicBlock(new_blocks.find(jmptarget)->second);
 266                                                         }
 267                                                         if (MIA->isConditionalBranch(inst)) {
 268                                                                 jmptarget = base_address + current_address + inst_size;
 269                                                                 current_block->setNextBlock(1, jmptarget);
 270                                                                 if (new_blocks.find(jmptarget) == new_blocks.end()) {
 271                                                                         BasicBlock * block = manager->newBasicBlock(jmptarget);
 272                                                                         assert(block);
 273                                                                         new_blocks.insert(std::make_pair(block->getStartAddress(), block));
 274                                                                         function->addBasicBlock(block);
 275                                                                         remaining_blocks.push(block);
 276                                                                 } else {
 277                                                                         LOG4CXX_DEBUG(logger, "Reusing Block starting at " << std::hex
 278                                                                                       << current_block->getStartAddress());
 279                                                                         function->addBasicBlock(new_blocks.find(jmptarget)->second);
 280                                                                 }
 281                                                         }
 282                                                 }
 283                                         }
 284                                 }
 285                         } else {
 286                                 inst_size = 0;
 287                         }
 288
 289
 290                         if (inst_size == 0 || MIA->isTerminator(inst) || MIA->isBranch(inst)) {
 291                                 current_block->setEndAddress(current_address + base_address + inst_size);
 292                                 LOG4CXX_DEBUG(logger, "Finished Block at " << std::hex <<
 293                                               current_block->getEndAddress());
 294                                 break;
 295                         }
 296                         current_address += inst_size;
 297                 }
 298         }
 299         splitBlocks(function);
 300         LOG4CXX_DEBUG(logger, "Finished function " << function->getName());
 301         manager->finishFunction(function);
 302         for (uint64_t address : called_functions)
 303                 disassembleFunctionAt(address);
 304 }
 305
 306 template <typename ELFT>
 307 void LLVMDisassembler<ELFT>::disassemble() {
 308         SectionRef text_section = sections[".text"];
 309         std::vector<Function*> remaining_functions;
 310
 311         // Assume all function symbols actually start a real function
 312         for (auto x = symbols.begin(); x != symbols.end(); ++x) {
 313                 uint64_t result;
 314                 bool contains;
 315                 SymbolRef::Type symbol_type;
 316
 317
 318                 if (text_section.containsSymbol(x->second, contains) || !contains)
 319                         continue;
 320
 321                 if (x->second.getType(symbol_type)
 322                     || SymbolRef::ST_Function != symbol_type)
 323                         continue;
 324
 325                 if (!x->second.getAddress(result)) {
 326                         Function * fun = manager->newFunction(result);
 327                         fun->setName(x->first);
 328                         remaining_functions.push_back(fun);
 329                         LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
 330                 }
 331         }
 332
 333         for (Function* function : remaining_functions) {
 334                 disassembleFunction(function);
 335                 manager->finishFunction(function);
 336         }
 337
 338         if (binary->isELF()) {
 339                 uint64_t _entryAddress = entryAddress();
 340                 LOG4CXX_DEBUG(logger, "Adding entryAddress at: " << std::hex << _entryAddress);
 341                 std::stringstream s;
 342                 s << "<_start 0x" << std::hex << _entryAddress << ">";
 343
 344                 disassembleFunctionAt(_entryAddress, s.str());
 345         }
 346
 347         if (!manager->hasFunctions()) {
 348                 uint64_t text_entry;
 349                 text_section.getAddress(text_entry);
 350                 LOG4CXX_INFO(logger, "No Symbols found, starting at the beginning of the text segment");
 351                 disassembleFunctionAt(text_entry);
 352         }
 353 }
 354
 355 template <>
 356 uint64_t LLVMDisassembler<COFFT>::entryAddress() {
 357         const auto coffobject = dyn_cast<COFFObjectFile>(o);
 358         const struct pe32_header* pe32_header;
 359         const struct pe32plus_header* pe32plus_header;
 360
 361         coffobject->getPE32PlusHeader(pe32plus_header);
 362
 363         if (pe32plus_header) {
 364                 return pe32plus_header->AddressOfEntryPoint;
 365         } else {
 366                 coffobject->getPE32Header(pe32_header);
 367                 return pe32_header->AddressOfEntryPoint;
 368         }
 369 }
 370
 371 template <typename ELFT>
 372 uint64_t LLVMDisassembler<ELFT>::entryAddress() {
 373         const auto elffile = dyn_cast<ELFObjectFile<ELFT>>(o)->getELFFile();
 374         const auto * header = elffile->getHeader();
 375
 376         return header->e_entry;
 377 }
 378
 379 template <typename ELFT>
 380 void LLVMDisassembler<ELFT>::splitBlocks(Function* function) {
 381         SectionRef text_section = sections[".text"];
 382         StringRef bytes;
 383         text_section.getContents(bytes);
 384         StringRefMemoryObject ref(bytes);
 385
 386         // Split blocks where jumps are going inside the block
 387         for (auto it = function->blocks().begin();
 388              it != function->blocks().end();
 389              ++it) {
 390                 BasicBlock * current_block = it->second;
 391                 uint64_t inst_size;
 392                 uint64_t base_address;
 393                 text_section.getAddress(base_address);
 394                 uint64_t current_address = current_block->getStartAddress() - base_address;
 395                 while(current_block->getEndAddress() - base_address > current_address) {
 396                         MCInst inst;
 397                         std::string buf;
 398                         llvm::raw_string_ostream s(buf);
 399
 400                         if(llvm::MCDisassembler::Success ==
 401                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 402                                 // See if some other block starts here
 403                                 BasicBlock* other = manager->getBasicBlock(current_address
 404                                                                            + inst_size
 405                                                                            + base_address);
 406
 407                                 // Special case, other block starts here but we are at the end anyway
 408                                 if (other != NULL) {
 409                                         uint64_t endaddress = current_address + inst_size + base_address;
 410                                         if (endaddress != current_block->getEndAddress()) {
 411                                                 LOG4CXX_DEBUG(logger, "Shortening block starting at "
 412                                                               << std::hex
 413                                                               << current_block->getStartAddress()
 414                                                               << " now ending at "
 415                                                               << other->getStartAddress());
 416                                                 function->addBasicBlock(other);
 417                                                 current_block->setEndAddress(endaddress);
 418                                                 current_block->setNextBlock(0, other->getStartAddress());
 419                                                 current_block->setNextBlock(1, 0);
 420                                         }
 421                                 }
 422                         } else {
 423                                 inst_size = 1;
 424                         }
 425                         current_address += inst_size;
 426                 }
 427         }
 428 }
 429
 430 template<>
 431 void LLVMDisassembler<COFFT>::readDynamicSymbols() {
 432         //TODO
 433 }
 434
 435 template <typename ELFT>
 436 void LLVMDisassembler<ELFT>::readDynamicSymbols() {
 437         const auto elffile = dyn_cast<ELFObjectFile<ELFT>>(o)->getELFFile();
 438         for (auto it = elffile->begin_dynamic_symbols(),
 439                      end = elffile->end_dynamic_symbols();
 440              it != end;
 441              ++it) {
 442                 if (it->getType() == 2) { // Function
 443                         bool is_default;
 444                         // TODO: Error handling
 445                         std::string symbolname = *(elffile->getSymbolName(it));
 446                         std::string symbolversion = *(elffile->getSymbolVersion(nullptr, &*it, is_default));
 447                         manager->signal_new_dyn_symbol(symbolname + (is_default? "@@" : "@") + symbolversion);
 448                         LOG4CXX_DEBUG(logger, "Adding dynamic Symbol " << symbolname << (is_default? "@@" : "@") << symbolversion);
 449                 }
 450         }
 451 }
 452
 453 template <typename ELFT>
 454 void LLVMDisassembler<ELFT>::readSymbols() {
 455         error_code ec;
 456         symbol_iterator si(o->symbol_begin()), se(o->symbol_end());
 457         for (; si != se; ++si) {
 458                 StringRef name;
 459                 if ((ec = si->getName(name))) {
 460                         LOG4CXX_ERROR(logger, ec.message());
 461                         break;
 462                 }
 463                 LOG4CXX_DEBUG(logger, "Added symbol " << name.str());
 464                 symbols.insert(make_pair(name.str(), *si));
 465         }
 466 }
 467
 468 template <typename ELFT>
 469 void LLVMDisassembler<ELFT>::readSections() {
 470         error_code ec;
 471         section_iterator i(o->section_begin()), e(o->section_end());
 472         for (; i != e; ++i) {
 473                 StringRef name;
 474                 if ((ec = i->getName(name))) {
 475                         LOG4CXX_ERROR(logger, ec.message());
 476                         break;
 477                 }
 478                 LOG4CXX_DEBUG(logger, "Added section " << name.str());
 479                 sections.insert(make_pair(name.str(), *i));
 480         }
 481
 482 }
 483
 484 // template <typename ELFT>
 485 // void LLVMDisassembler<ELFT>::forEachFunction(std::function<void (uint64_t, Function*)> callback) {
 486 //      // std::for_each(functions.begin(), functions.end(),
 487 //      //               [&](std::pair<uint64_t, Function*> x) {
 488 //      //                    callback(x.first, x.second);
 489 //      //               });
 490 // }
 491
 492 template <typename ELFT>
 493 void LLVMDisassembler<ELFT>::printEachInstruction(uint64_t start, uint64_t end,
 494                                                   std::function<void (uint8_t*, size_t,
 495                                                                           const std::string&,
 496                                                                           const std::string&)> fun) {
 497         SectionRef text_section = sections[".text"];
 498         uint64_t base_address;
 499         text_section.getAddress(base_address);
 500         uint64_t current_address = start - base_address;
 501
 502         StringRef bytes;
 503         text_section.getContents(bytes);
 504         StringRefMemoryObject ref(bytes);
 505
 506         while (current_address < end - base_address) {
 507                 uint64_t inst_size;
 508                 MCInst inst;
 509                 std::string buf;
 510                 llvm::raw_string_ostream s(buf);
 511
 512                 if(llvm::MCDisassembler::Success ==
 513                    DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 514
 515                         uint8_t bytes[inst_size+2];
 516                         ref.readBytes(current_address, inst_size, bytes);
 517
 518                         uint64_t jmptarget;
 519                         std::string ref("");
 520                         IP->printInst(&inst, s, "");
 521                         if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 522                                 std::stringstream stream;
 523                                 if (MIA->isCall(inst))
 524                                         stream << "function:";
 525                                 else
 526                                         stream << "block:";
 527
 528                                 stream << std::hex << (base_address + jmptarget);
 529                                 ref = stream.str();
 530                         }
 531
 532
 533                         fun(bytes, inst_size, s.str(), ref);
 534                 } else {
 535                         LOG4CXX_WARN(logger, "Invalid byte at" << std::hex << current_address + base_address);
 536                         fun(NULL, 0, "Invalid Byte", "");
 537                         inst_size = 1;
 538                 }
 539
 540                 current_address += inst_size;
 541         }
 542 }