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("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::stack<BasicBlock*> remaining_blocks;
 210         /* TODO:
 211          * Do all blocks get added properly? We should take care to remove
 212          * the other ones at the end of the function!
 213          */
 214         std::map<uint64_t, BasicBlock*> new_blocks;
 215         SectionRef text_section = sections[".text"];
 216         StringRef bytes;
 217         text_section.getContents(bytes);
 218         StringRefMemoryObject ref(bytes);
 219
 220         LOG4CXX_DEBUG(logger, "Handling function " << function->getName());
 221
 222         BasicBlock * block = manager->newBasicBlock(function->getStartAddress());
 223         remaining_blocks.push(block);
 224         new_blocks.insert(std::make_pair(block->getStartAddress(), block));
 225         function->addBasicBlock(block);
 226
 227         while (remaining_blocks.size()) {
 228                 BasicBlock * current_block = remaining_blocks.top();
 229                 remaining_blocks.pop();
 230
 231                 LOG4CXX_DEBUG(logger, "Handling Block starting at " << std::hex
 232                               << current_block->getStartAddress());
 233
 234                 uint64_t inst_size;
 235                 uint64_t base_address;
 236                 text_section.getAddress(base_address);
 237                 uint64_t current_address = current_block->getStartAddress() - base_address;
 238                 while(true) {
 239                         MCInst inst;
 240                         std::string buf;
 241                         llvm::raw_string_ostream s(buf);
 242
 243                         if(llvm::MCDisassembler::Success ==
 244                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 245                                 uint64_t jmptarget;
 246
 247                                 if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 248                                         jmptarget += base_address;
 249                                         if (!MIA->isIndirectBranch(inst)) {
 250                                                 if (MIA->isCall(inst)) {
 251                                                         if (NULL == manager->getFunction(jmptarget))
 252                                                                 disassembleFunctionAt(jmptarget);
 253                                                 } else {
 254                                                         current_block->setNextBlock(0, jmptarget);
 255                                                         if (new_blocks.find(jmptarget) == new_blocks.end()) {
 256                                                                 BasicBlock * block = manager->newBasicBlock(jmptarget);
 257                                                                 assert(block);
 258                                                                 new_blocks.insert(std::make_pair(block->getStartAddress(), block));
 259                                                                 function->addBasicBlock(block);
 260                                                                 remaining_blocks.push(block);
 261                                                         } else {
 262                                                                 LOG4CXX_DEBUG(logger, "Reusing Block starting at " << std::hex
 263                                                                               << current_block->getStartAddress());
 264                                                                 function->addBasicBlock(new_blocks.find(jmptarget)->second);
 265                                                         }
 266                                                         if (MIA->isConditionalBranch(inst)) {
 267                                                                 jmptarget = base_address + current_address + inst_size;
 268                                                                 current_block->setNextBlock(1, jmptarget);
 269                                                                 if (new_blocks.find(jmptarget) == new_blocks.end()) {
 270                                                                         BasicBlock * block = manager->newBasicBlock(jmptarget);
 271                                                                         assert(block);
 272                                                                         new_blocks.insert(std::make_pair(block->getStartAddress(), block));
 273                                                                         function->addBasicBlock(block);
 274                                                                         remaining_blocks.push(block);
 275                                                                 } else {
 276                                                                         LOG4CXX_DEBUG(logger, "Reusing Block starting at " << std::hex
 277                                                                                       << current_block->getStartAddress());
 278                                                                         function->addBasicBlock(new_blocks.find(jmptarget)->second);
 279                                                                 }
 280                                                         }
 281                                                 }
 282                                         }
 283                                 }
 284                         } else {
 285                                 inst_size = 0;
 286                         }
 287
 288
 289                         if (inst_size == 0 || MIA->isTerminator(inst) || MIA->isBranch(inst)) {
 290                                 current_block->setEndAddress(current_address + base_address + inst_size);
 291                                 LOG4CXX_DEBUG(logger, "Finished Block at " << std::hex <<
 292                                               current_block->getEndAddress());
 293                                 break;
 294                         }
 295                         current_address += inst_size;
 296                 }
 297         }
 298         splitBlocks(function);
 299         LOG4CXX_DEBUG(logger, "Finished function " << function->getName());
 300         manager->finishFunction(function);
 301 }
 302
 303 template <typename ELFT>
 304 void LLVMDisassembler<ELFT>::disassemble() {
 305         SectionRef text_section = sections[".text"];
 306         std::vector<Function*> remaining_functions;
 307
 308         // Assume all function symbols actually start a real function
 309         for (auto x = symbols.begin(); x != symbols.end(); ++x) {
 310                 uint64_t result;
 311                 bool contains;
 312                 SymbolRef::Type symbol_type;
 313
 314
 315                 if (text_section.containsSymbol(x->second, contains) || !contains)
 316                         continue;
 317
 318                 if (x->second.getType(symbol_type)
 319                     || SymbolRef::ST_Function != symbol_type)
 320                         continue;
 321
 322                 if (!x->second.getAddress(result)) {
 323                         Function * fun = manager->newFunction(result);
 324                         fun->setName(x->first);
 325                         remaining_functions.push_back(fun);
 326                         LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
 327                 }
 328         }
 329
 330         for (Function* function : remaining_functions) {
 331                 disassembleFunction(function);
 332                 manager->finishFunction(function);
 333         }
 334
 335         if (binary->isELF()) {
 336                 uint64_t _entryAddress = entryAddress();
 337                 LOG4CXX_DEBUG(logger, "Adding entryAddress at: " << std::hex << _entryAddress);
 338                 std::stringstream s;
 339                 s << "<_start 0x" << std::hex << _entryAddress << ">";
 340
 341                 disassembleFunctionAt(_entryAddress, s.str());
 342         }
 343
 344         if (!manager->hasFunctions()) {
 345                 uint64_t text_entry;
 346                 text_section.getAddress(text_entry);
 347                 LOG4CXX_INFO(logger, "No Symbols found, starting at the beginning of the text segment");
 348                 disassembleFunctionAt(text_entry);
 349         }
 350 }
 351
 352 template <>
 353 uint64_t LLVMDisassembler<COFFT>::entryAddress() {
 354         const auto coffobject = dyn_cast<COFFObjectFile>(o);
 355         const struct pe32_header* pe32_header;
 356         const struct pe32plus_header* pe32plus_header;
 357
 358         coffobject->getPE32PlusHeader(pe32plus_header);
 359
 360         if (pe32plus_header) {
 361                 return pe32plus_header->AddressOfEntryPoint;
 362         } else {
 363                 coffobject->getPE32Header(pe32_header);
 364                 return pe32_header->AddressOfEntryPoint;
 365         }
 366 }
 367
 368 template <typename ELFT>
 369 uint64_t LLVMDisassembler<ELFT>::entryAddress() {
 370         const auto elffile = dyn_cast<ELFObjectFile<ELFT>>(o)->getELFFile();
 371         const auto * header = elffile->getHeader();
 372
 373         return header->e_entry;
 374 }
 375
 376 template <typename ELFT>
 377 void LLVMDisassembler<ELFT>::splitBlocks(Function* function) {
 378         SectionRef text_section = sections[".text"];
 379         StringRef bytes;
 380         text_section.getContents(bytes);
 381         StringRefMemoryObject ref(bytes);
 382
 383         // Split blocks where jumps are going inside the block
 384         for (auto it = function->blocks().begin();
 385              it != function->blocks().end();
 386              ++it) {
 387                 BasicBlock * current_block = it->second;
 388                 uint64_t inst_size;
 389                 uint64_t base_address;
 390                 text_section.getAddress(base_address);
 391                 uint64_t current_address = current_block->getStartAddress() - base_address;
 392                 while(current_block->getEndAddress() - base_address > current_address) {
 393                         MCInst inst;
 394                         std::string buf;
 395                         llvm::raw_string_ostream s(buf);
 396
 397                         if(llvm::MCDisassembler::Success ==
 398                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 399                                 // See if some other block starts here
 400                                 BasicBlock* other = manager->getBasicBlock(current_address
 401                                                                            + inst_size
 402                                                                            + base_address);
 403
 404                                 // Special case, other block starts here but we are at the end anyway
 405                                 if (other != NULL) {
 406                                         uint64_t endaddress = current_address + inst_size + base_address;
 407                                         if (endaddress != current_block->getEndAddress()) {
 408                                                 LOG4CXX_DEBUG(logger, "Shortening block starting at "
 409                                                               << std::hex
 410                                                               << current_block->getStartAddress()
 411                                                               << " now ending at "
 412                                                               << other->getStartAddress());
 413                                                 function->addBasicBlock(other);
 414                                                 current_block->setEndAddress(endaddress);
 415                                                 current_block->setNextBlock(0, other->getStartAddress());
 416                                                 current_block->setNextBlock(1, 0);
 417                                         }
 418                                 }
 419                         } else {
 420                                 inst_size = 1;
 421                         }
 422                         current_address += inst_size;
 423                 }
 424         }
 425 }
 426
 427 template<>
 428 void LLVMDisassembler<COFFT>::readDynamicSymbols() {
 429         //TODO
 430 }
 431
 432 template <typename ELFT>
 433 void LLVMDisassembler<ELFT>::readDynamicSymbols() {
 434         const auto elffile = dyn_cast<ELFObjectFile<ELFT>>(o)->getELFFile();
 435         for (auto it = elffile->begin_dynamic_symbols(),
 436                      end = elffile->end_dynamic_symbols();
 437              it != end;
 438              ++it) {
 439                 if (it->getType() == 2) { // Function
 440                         bool is_default;
 441                         // TODO: Error handling
 442                         std::string symbolname = *(elffile->getSymbolName(it));
 443                         std::string symbolversion = *(elffile->getSymbolVersion(nullptr, &*it, is_default));
 444                         manager->signal_new_dyn_symbol(symbolname + (is_default? "@@" : "@") + symbolversion);
 445                         LOG4CXX_DEBUG(logger, "Adding dynamic Symbol " << symbolname << (is_default? "@@" : "@") << symbolversion);
 446                 }
 447         }
 448 }
 449
 450 template <typename ELFT>
 451 void LLVMDisassembler<ELFT>::readSymbols() {
 452         error_code ec;
 453         symbol_iterator si(o->symbol_begin()), se(o->symbol_end());
 454         for (; si != se; ++si) {
 455                 StringRef name;
 456                 if ((ec = si->getName(name))) {
 457                         LOG4CXX_ERROR(logger, ec.message());
 458                         break;
 459                 }
 460                 LOG4CXX_DEBUG(logger, "Added symbol " << name.str());
 461                 symbols.insert(make_pair(name.str(), *si));
 462         }
 463 }
 464
 465 template <typename ELFT>
 466 void LLVMDisassembler<ELFT>::readSections() {
 467         error_code ec;
 468         section_iterator i(o->section_begin()), e(o->section_end());
 469         for (; i != e; ++i) {
 470                 StringRef name;
 471                 if ((ec = i->getName(name))) {
 472                         LOG4CXX_ERROR(logger, ec.message());
 473                         break;
 474                 }
 475                 LOG4CXX_DEBUG(logger, "Added section " << name.str());
 476                 sections.insert(make_pair(name.str(), *i));
 477         }
 478
 479 }
 480
 481 // template <typename ELFT>
 482 // void LLVMDisassembler<ELFT>::forEachFunction(std::function<void (uint64_t, Function*)> callback) {
 483 //      // std::for_each(functions.begin(), functions.end(),
 484 //      //               [&](std::pair<uint64_t, Function*> x) {
 485 //      //                    callback(x.first, x.second);
 486 //      //               });
 487 // }
 488
 489 template <typename ELFT>
 490 void LLVMDisassembler<ELFT>::printEachInstruction(uint64_t start, uint64_t end,
 491                                                   std::function<void (uint8_t*, size_t,
 492                                                                           const std::string&,
 493                                                                           const std::string&)> fun) {
 494         SectionRef text_section = sections[".text"];
 495         uint64_t base_address;
 496         text_section.getAddress(base_address);
 497         uint64_t current_address = start - base_address;
 498
 499         StringRef bytes;
 500         text_section.getContents(bytes);
 501         StringRefMemoryObject ref(bytes);
 502
 503         while (current_address < end - base_address) {
 504                 uint64_t inst_size;
 505                 MCInst inst;
 506                 std::string buf;
 507                 llvm::raw_string_ostream s(buf);
 508
 509                 if(llvm::MCDisassembler::Success ==
 510                    DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 511
 512                         uint8_t bytes[inst_size+2];
 513                         ref.readBytes(current_address, inst_size, bytes);
 514
 515                         uint64_t jmptarget;
 516                         std::string ref("");
 517                         IP->printInst(&inst, s, "");
 518                         if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 519                                 std::stringstream stream;
 520                                 if (MIA->isCall(inst))
 521                                         stream << "function:";
 522                                 else
 523                                         stream << "block:";
 524
 525                                 stream << std::hex << (base_address + jmptarget);
 526                                 ref = stream.str();
 527                         }
 528
 529
 530                         fun(bytes, inst_size, s.str(), ref);
 531                 } else {
 532                         LOG4CXX_WARN(logger, "Invalid byte at" << std::hex << current_address + base_address);
 533                         fun(NULL, 0, "Invalid Byte", "");
 534                         inst_size = 1;
 535                 }
 536
 537                 current_address += inst_size;
 538         }
 539 }