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                 manager->finishFunction(function);
 203         }
 204
 205         return function;
 206 }
 207
 208 template <typename ELFT>
 209 void LLVMDisassembler<ELFT>::disassembleFunction(Function* function) {
 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                                                                 disassembleFunctionAt(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->signal_new_function(function);
 302 }
 303
 304 template <typename ELFT>
 305 void LLVMDisassembler<ELFT>::disassemble() {
 306         SectionRef text_section = sections[".text"];
 307         std::vector<Function*> remaining_functions;
 308
 309         // Assume all function symbols actually start a real function
 310         for (auto x = symbols.begin(); x != symbols.end(); ++x) {
 311                 uint64_t result;
 312                 bool contains;
 313                 SymbolRef::Type symbol_type;
 314
 315
 316                 if (text_section.containsSymbol(x->second, contains) || !contains)
 317                         continue;
 318
 319                 if (x->second.getType(symbol_type)
 320                     || SymbolRef::ST_Function != symbol_type)
 321                         continue;
 322
 323                 if (!x->second.getAddress(result)) {
 324                         Function * fun = manager->newFunction(result);
 325                         fun->setName(x->first);
 326                         remaining_functions.push_back(fun);
 327                         LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
 328                 }
 329         }
 330
 331         for (Function* function : remaining_functions) {
 332                 disassembleFunction(function);
 333                 manager->finishFunction(function);
 334         }
 335
 336         if (binary->isELF()) {
 337                 uint64_t _entryAddress = entryAddress();
 338                 LOG4CXX_DEBUG(logger, "Adding entryAddress at: " << std::hex << _entryAddress);
 339                 std::stringstream s;
 340                 s << "<_start 0x" << std::hex << _entryAddress << ">";
 341
 342                 disassembleFunctionAt(_entryAddress, s.str());
 343         }
 344
 345         if (!manager->hasFunctions()) {
 346                 uint64_t text_entry;
 347                 text_section.getAddress(text_entry);
 348                 LOG4CXX_INFO(logger, "No Symbols found, starting at the beginning of the text segment");
 349                 disassembleFunctionAt(text_entry);
 350         }
 351 }
 352
 353 template <>
 354 uint64_t LLVMDisassembler<COFFT>::entryAddress() {
 355         const auto coffobject = dyn_cast<COFFObjectFile>(o);
 356         const struct pe32_header* pe32_header;
 357         const struct pe32plus_header* pe32plus_header;
 358
 359         coffobject->getPE32PlusHeader(pe32plus_header);
 360
 361         if (pe32plus_header) {
 362                 return pe32plus_header->AddressOfEntryPoint;
 363         } else {
 364                 coffobject->getPE32Header(pe32_header);
 365                 return pe32_header->AddressOfEntryPoint;
 366         }
 367 }
 368
 369 template <typename ELFT>
 370 uint64_t LLVMDisassembler<ELFT>::entryAddress() {
 371         const auto elffile = dyn_cast<ELFObjectFile<ELFT>>(o)->getELFFile();
 372         const auto * header = elffile->getHeader();
 373
 374         return header->e_entry;
 375 }
 376
 377 template <typename ELFT>
 378 void LLVMDisassembler<ELFT>::splitBlocks(Function* function) {
 379         SectionRef text_section = sections[".text"];
 380         StringRef bytes;
 381         text_section.getContents(bytes);
 382         StringRefMemoryObject ref(bytes);
 383
 384         // Split blocks where jumps are going inside the block
 385         for (auto it = function->blocks().begin();
 386              it != function->blocks().end();
 387              ++it) {
 388                 BasicBlock * current_block = it->second;
 389                 uint64_t inst_size;
 390                 uint64_t base_address;
 391                 text_section.getAddress(base_address);
 392                 uint64_t current_address = current_block->getStartAddress() - base_address;
 393                 while(current_block->getEndAddress() - base_address > current_address) {
 394                         MCInst inst;
 395                         std::string buf;
 396                         llvm::raw_string_ostream s(buf);
 397
 398                         if(llvm::MCDisassembler::Success ==
 399                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 400                                 // See if some other block starts here
 401                                 BasicBlock* other = manager->getBasicBlock(current_address
 402                                                                            + inst_size
 403                                                                            + base_address);
 404
 405                                 // Special case, other block starts here but we are at the end anyway
 406                                 if (other != NULL) {
 407                                         uint64_t endaddress = current_address + inst_size + base_address;
 408                                         if (endaddress != current_block->getEndAddress()) {
 409                                                 LOG4CXX_DEBUG(logger, "Shortening block starting at "
 410                                                               << std::hex
 411                                                               << current_block->getStartAddress()
 412                                                               << " now ending at "
 413                                                               << other->getStartAddress());
 414                                                 function->addBasicBlock(other);
 415                                                 current_block->setEndAddress(endaddress);
 416                                                 current_block->setNextBlock(0, other->getStartAddress());
 417                                                 current_block->setNextBlock(1, 0);
 418                                         }
 419                                 }
 420                         } else {
 421                                 inst_size = 1;
 422                         }
 423                         current_address += inst_size;
 424                 }
 425         }
 426 }
 427
 428 template<>
 429 void LLVMDisassembler<COFFT>::readDynamicSymbols() {
 430         //TODO
 431 }
 432
 433 template <typename ELFT>
 434 void LLVMDisassembler<ELFT>::readDynamicSymbols() {
 435         const auto elffile = dyn_cast<ELFObjectFile<ELFT>>(o)->getELFFile();
 436         for (auto it = elffile->begin_dynamic_symbols(),
 437                      end = elffile->end_dynamic_symbols();
 438              it != end;
 439              ++it) {
 440                 if (it->getType() == 2) { // Function
 441                         bool is_default;
 442                         // TODO: Error handling
 443                         std::string symbolname = *(elffile->getSymbolName(it));
 444                         std::string symbolversion = *(elffile->getSymbolVersion(nullptr, &*it, is_default));
 445                         manager->signal_new_dyn_symbol(symbolname + (is_default? "@@" : "@") + symbolversion);
 446                         LOG4CXX_DEBUG(logger, "Adding dynamic Symbol " << symbolname << (is_default? "@@" : "@") << symbolversion);
 447                 }
 448         }
 449 }
 450
 451 template <typename ELFT>
 452 void LLVMDisassembler<ELFT>::readSymbols() {
 453         error_code ec;
 454         symbol_iterator si(o->symbol_begin()), se(o->symbol_end());
 455         for (; si != se; ++si) {
 456                 StringRef name;
 457                 if ((ec = si->getName(name))) {
 458                         LOG4CXX_ERROR(logger, ec.message());
 459                         break;
 460                 }
 461                 LOG4CXX_DEBUG(logger, "Added symbol " << name.str());
 462                 symbols.insert(make_pair(name.str(), *si));
 463         }
 464 }
 465
 466 template <typename ELFT>
 467 void LLVMDisassembler<ELFT>::readSections() {
 468         error_code ec;
 469         section_iterator i(o->section_begin()), e(o->section_end());
 470         for (; i != e; ++i) {
 471                 StringRef name;
 472                 if ((ec = i->getName(name))) {
 473                         LOG4CXX_ERROR(logger, ec.message());
 474                         break;
 475                 }
 476                 LOG4CXX_DEBUG(logger, "Added section " << name.str());
 477                 sections.insert(make_pair(name.str(), *i));
 478         }
 479
 480 }
 481
 482 // template <typename ELFT>
 483 // void LLVMDisassembler<ELFT>::forEachFunction(std::function<void (uint64_t, Function*)> callback) {
 484 //      // std::for_each(functions.begin(), functions.end(),
 485 //      //               [&](std::pair<uint64_t, Function*> x) {
 486 //      //                    callback(x.first, x.second);
 487 //      //               });
 488 // }
 489
 490 template <typename ELFT>
 491 void LLVMDisassembler<ELFT>::printEachInstruction(uint64_t start, uint64_t end,
 492                                                   std::function<void (uint8_t*, size_t,
 493                                                                           const std::string&,
 494                                                                           const std::string&)> fun) {
 495         SectionRef text_section = sections[".text"];
 496         uint64_t base_address;
 497         text_section.getAddress(base_address);
 498         uint64_t current_address = start - base_address;
 499
 500         StringRef bytes;
 501         text_section.getContents(bytes);
 502         StringRefMemoryObject ref(bytes);
 503
 504         while (current_address < end - base_address) {
 505                 uint64_t inst_size;
 506                 MCInst inst;
 507                 std::string buf;
 508                 llvm::raw_string_ostream s(buf);
 509
 510                 if(llvm::MCDisassembler::Success ==
 511                    DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 512
 513                         uint8_t bytes[inst_size+2];
 514                         ref.readBytes(current_address, inst_size, bytes);
 515
 516                         uint64_t jmptarget;
 517                         std::string ref("");
 518                         IP->printInst(&inst, s, "");
 519                         if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 520                                 std::stringstream stream;
 521                                 if (MIA->isCall(inst))
 522                                         stream << "function:";
 523                                 else
 524                                         stream << "block:";
 525
 526                                 stream << std::hex << (base_address + jmptarget);
 527                                 ref = stream.str();
 528                         }
 529
 530
 531                         fun(bytes, inst_size, s.str(), ref);
 532                 } else {
 533                         LOG4CXX_WARN(logger, "Invalid byte at" << std::hex << current_address + base_address);
 534                         fun(NULL, 0, "Invalid Byte", "");
 535                         inst_size = 1;
 536                 }
 537
 538                 current_address += inst_size;
 539         }
 540 }