src/disassembler/llvm/LLVMDisassembler.cxx

   1 #include "disassembler/Instruction.hxx"
   2 #include "disassembler/llvm/LLVMDisassembler.hxx"
   3 #include "core/InformationManager.hxx"
   4 #include "core/Function.hxx"
   5 #include "core/BasicBlock.hxx"
   6 #include "core/Exception.hxx"
   7 #include <boost/algorithm/string.hpp>
   8
   9 #include <stack>
  10 #include <algorithm>
  11 #include <cassert>
  12
  13 using namespace llvm;
  14 using namespace llvm::object;
  15 using std::error_code;
  16
  17 namespace {
  18         class COFFT {
  19
  20         };
  21
  22         class MACHOT {
  23
  24         };
  25 }
  26
  27 /*
  28  *
  29  */
  30 Disassembler * createLLVMDisassembler(const std::string& filename, InformationManager* manager) {
  31         log4cxx::LoggerPtr logger(log4cxx::Logger::getLogger("disassembler.LLVMDisassembler"));
  32         if (filename == "")
  33                 return NULL;
  34
  35         auto retval = createBinary(filename);
  36         if (error_code ec = retval.getError()) {
  37                 LOG4CXX_ERROR(logger, ec.message());
  38                 return NULL;
  39         }
  40 #if defined(LLVM_35)
  41         Binary * op = retval.get();
  42 #elif defined(LLVM_36)
  43         OwningBinary<Binary> ob;
  44         ob  = std::move(retval.get());
  45         Binary* op = ob.getBinary();
  46         auto foo = ob.takeBinary();
  47         foo.first.release();
  48         foo.second.release();
  49 #endif
  50
  51         try {
  52                 // ELFType<endian, maxalign, 64bit>
  53                 if (ELF32LEObjectFile * object = dyn_cast<ELF32LEObjectFile>(op)) {
  54                         return new LLVMDisassembler<ELFType<support::little, 2, false>>(filename, manager, object);
  55                 }
  56                 if (ELF64LEObjectFile * object = dyn_cast<ELF64LEObjectFile>(op)) {
  57                         return new LLVMDisassembler<ELFType<support::little, 2, true>>(filename, manager, object);
  58                 }
  59                 if (ELF32BEObjectFile * object = dyn_cast<ELF32BEObjectFile>(op)) {
  60                         return new LLVMDisassembler<ELFType<support::big, 2, false>>(filename, manager, object);
  61                 }
  62                 if (ELF64BEObjectFile * object = dyn_cast<ELF64BEObjectFile>(op)) {
  63                         return new LLVMDisassembler<ELFType<support::big, 2, true>>(filename, manager, object);
  64                 }
  65                 if (COFFObjectFile * object = dyn_cast<COFFObjectFile>(op)) {
  66                         return new LLVMDisassembler<COFFT>(filename, manager, object);
  67                 }
  68                 if (MachOObjectFile * object = dyn_cast<MachOObjectFile>(op)) {
  69                         return new LLVMDisassembler<MACHOT>(filename, manager, object);
  70                 }
  71         } catch (BinaryNotSupported& e) {
  72                 return NULL;
  73         }
  74         return NULL;
  75 }
  76
  77 /*
  78  * TODO: fallback code falls die Datei kein ELF/PE/COFF/MacO/.. binary
  79  * ist sondern z.B. einfach nur Instruktionen oder ein Bootsektor oder
  80  * foo
  81  */
  82 template <typename ELFT>
  83 LLVMDisassembler<ELFT>::LLVMDisassembler(const std::string& filename,
  84                                          InformationManager* manager,
  85                                          ObjectFile* file)
  86         : Disassembler()
  87         , logger(log4cxx::Logger::getLogger("disassembler.LLVMDisassembler"))
  88         , triple("unknown-unknown-unknown")
  89         , manager(manager)
  90 {
  91         LOG4CXX_DEBUG(logger, "Handling file " << filename);
  92
  93         if (!file) {
  94                 auto result = createBinary(filename);
  95
  96                 error_code ec;
  97                 if ((ec = result.getError())) {
  98                         LOG4CXX_ERROR(logger, "Failed to load Binary" << ec.message());
  99                         binary = NULL;
 100                         return;
 101                 }
 102
 103 #if defined(LLVM_35)
 104                 binary.reset(result.get());
 105 #elif defined(LLVM_36)
 106                 OwningBinary<Binary> ob;
 107                 ob  = std::move(result.get());
 108                 Binary* op = ob.getBinary();
 109
 110                 binary.reset(op);
 111 #endif
 112
 113                 o = dyn_cast<ObjectFile>(binary.get());
 114         } else {
 115                 o = file;
 116                 binary.reset(file);
 117         }
 118
 119         triple.setArch(Triple::ArchType(o->getArch()));
 120         std::string tripleName(triple.getTriple());
 121
 122         LOG4CXX_INFO(logger, "Architecture " << tripleName);
 123
 124
 125         std::string es;
 126         target = TargetRegistry::lookupTarget("", triple, es);
 127         if (!target) {
 128                 LOG4CXX_ERROR(logger, es);
 129                 BinaryNotSupported e;
 130                 throw e;
 131         }
 132
 133         LOG4CXX_INFO(logger, "Target " << target->getName());
 134
 135         MRI.reset(target->createMCRegInfo(tripleName));
 136         if (!MRI) {
 137                 LOG4CXX_ERROR(logger, "no register info for target " << tripleName);
 138                 BinaryNotSupported e;
 139                 throw e;
 140         }
 141
 142         // Set up disassembler.
 143         AsmInfo.reset(target->createMCAsmInfo(*MRI, tripleName));
 144         if (!AsmInfo) {
 145                 LOG4CXX_ERROR(logger, "no assembly info for target " << tripleName);
 146                 BinaryNotSupported e;
 147                 throw e;
 148         }
 149
 150         STI.reset(target->createMCSubtargetInfo(tripleName, "", ""));
 151         if (!STI) {
 152                 LOG4CXX_ERROR(logger, "no subtarget info for target " << tripleName);
 153                 BinaryNotSupported e;
 154                 throw e;
 155         }
 156
 157         MII.reset(target->createMCInstrInfo());
 158         if (!MII) {
 159                 LOG4CXX_ERROR(logger, "no instruction info for target " << tripleName);
 160                                 BinaryNotSupported e;
 161                 throw e;
 162         }
 163
 164         MOFI.reset(new MCObjectFileInfo);
 165         MCContext Ctx(AsmInfo.get(), MRI.get(), MOFI.get());
 166
 167         DisAsm.reset(target->createMCDisassembler(*STI, Ctx));
 168         if (!DisAsm) {
 169                 LOG4CXX_ERROR(logger, "no disassembler for target " << tripleName);
 170                 BinaryNotSupported e;
 171                 throw e;
 172         }
 173         RelInfo.reset(
 174                 target->createMCRelocationInfo(tripleName, Ctx));
 175         if (RelInfo) {
 176                 // Symzer.reset(
 177                 //      MCObjectSymbolizer::createObjectSymbolizer(Ctx, std::move(RelInfo), o));
 178                 // if (Symzer)
 179                 //      DisAsm->setSymbolizer(std::move(Symzer));
 180         }
 181         RelInfo.release();
 182         Symzer.release();
 183
 184         MIA.reset(target->createMCInstrAnalysis(MII.get()));
 185         if (!MIA) {
 186                 LOG4CXX_ERROR(logger, "no instruction analysis for target " << tripleName);
 187                 BinaryNotSupported e;
 188                 throw e;
 189         }
 190
 191         int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
 192         IP.reset(target->createMCInstPrinter(AsmPrinterVariant, *AsmInfo, *MII, *MRI, *STI));
 193         if (!IP) {
 194                 LOG4CXX_ERROR(logger, "no instruction printer for target " << tripleName);
 195                 BinaryNotSupported e;
 196                 throw e;
 197         }
 198
 199         IP->setPrintImmHex(llvm::HexStyle::C);
 200         IP->setPrintImmHex(true);
 201
 202 //      std::unique_ptr<MCObjectDisassembler> OD(
 203 //              new MCObjectDisassembler(*o, *DisAsm, *MIA));
 204         //Mod.reset(OD->buildModule(false));
 205
 206         readSections();
 207 }
 208
 209 template <typename ELFT>
 210 void LLVMDisassembler<ELFT>::start() {
 211         readSymbols();
 212         disassemble();
 213         readDynamicSymbols();
 214 }
 215
 216 template <typename ELFT>
 217 LLVMDisassembler<ELFT>::~LLVMDisassembler() {}
 218
 219 template <typename ELFT>
 220 Function* LLVMDisassembler<ELFT>::disassembleFunctionAt(uint64_t address, const std::string& name) {
 221         Function * function;
 222         SectionRef text_section = getTextSection();
 223         uint64_t base_address, size;
 224 #if defined(LLVM_35)
 225         text_section.getAddress(base_address);
 226         text_section.getSize(size);
 227 #elif defined(LLVM_36)
 228         base_address = text_section.getAddress();
 229         size = text_section.getSize();
 230 #endif
 231         if (address < base_address ||
 232             address >= base_address + size) {
 233                 return NULL;
 234         }
 235
 236         if (NULL == (function = manager->getFunction(address))) {
 237
 238                 if (name == "") {
 239                         std::stringstream s;
 240                         s << "<Unnamed 0x" << std::hex << address << ">";
 241                         function = manager->newFunction(address);
 242                         function->setName(s.str());
 243                 } else {
 244                         function = manager->newFunction(address);
 245                         function->setName(name);
 246                 }
 247                 disassembleFunction(function);
 248         }
 249
 250         return function;
 251 }
 252
 253 template <typename ELFT>
 254 void LLVMDisassembler<ELFT>::disassembleFunction(Function* function) {
 255         std::vector<uint64_t> called_functions;
 256         std::stack<BasicBlock*> remaining_blocks;
 257         /* TODO:
 258          * Do all blocks get added properly? We should take care to remove
 259          * the other ones at the end of the function!
 260          */
 261         std::map<uint64_t, BasicBlock*> new_blocks;
 262         SectionRef text_section = getTextSection();
 263         StringRef bytes;
 264         text_section.getContents(bytes);
 265 #if defined(LLVM_35)
 266         StringRefMemoryObject ref(bytes);
 267 #elif defined(LLVM_36)
 268         ArrayRef<uint8_t> bytearray(reinterpret_cast<const uint8_t *>(bytes.data()),
 269                            bytes.size());
 270 #else
 271 #error LLVM != 3.5 | 3.6 not supported
 272 #endif
 273
 274         LOG4CXX_DEBUG(logger, "Handling function " << function->getName());
 275
 276         BasicBlock * block = manager->newBasicBlock(function->getStartAddress());
 277         remaining_blocks.push(block);
 278         new_blocks.insert(std::make_pair(block->getStartAddress(), block));
 279         function->addBasicBlock(block);
 280
 281         uint64_t base_address, size;
 282 #if defined(LLVM_35)
 283         text_section.getAddress(base_address);
 284         text_section.getSize(size);
 285 #elif defined(LLVM_36)
 286         base_address = text_section.getAddress();
 287         size = text_section.getSize();
 288 #endif
 289         LOG4CXX_DEBUG(logger, "Text section at " << std::hex << base_address << " with size " << size);
 290
 291         while (remaining_blocks.size()) {
 292                 BasicBlock * current_block = remaining_blocks.top();
 293                 remaining_blocks.pop();
 294
 295                 LOG4CXX_DEBUG(logger, "Handling Block starting at " << std::hex
 296                               << current_block->getStartAddress());
 297
 298                 uint64_t inst_size;
 299                 uint64_t current_address = current_block->getStartAddress() - base_address;
 300                 while(true) {
 301                         MCInst inst;
 302                         std::string buf;
 303                         llvm::raw_string_ostream s(buf);
 304
 305                         if(llvm::MCDisassembler::Success ==
 306 #if defined(LLVM_35)
 307                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 308 #elif defined(LLVM_36)
 309                                 DisAsm->getInstruction(inst, inst_size,
 310                                                        bytearray.slice(current_address),
 311                                                        base_address + current_address,
 312                                                        nulls(), nulls())) {
 313 #endif
 314                                 uint64_t jmptarget;
 315
 316                                 if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 317                                         jmptarget += base_address;
 318                                         if (!MIA->isIndirectBranch(inst)) {
 319                                                 if (MIA->isCall(inst)) {
 320                                                         if (NULL == manager->getFunction(jmptarget))
 321                                                                 called_functions.push_back(jmptarget);
 322                                                 } else {
 323                                                         current_block->setNextBlock(0, jmptarget);
 324                                                         if (new_blocks.find(jmptarget) == new_blocks.end()) {
 325                                                                 BasicBlock * block = manager->newBasicBlock(jmptarget);
 326                                                                 assert(block);
 327                                                                 new_blocks.insert(std::make_pair(block->getStartAddress(), block));
 328                                                                 function->addBasicBlock(block);
 329                                                                 remaining_blocks.push(block);
 330                                                         } else {
 331                                                                 LOG4CXX_DEBUG(logger, "Reusing Block starting at " << std::hex
 332                                                                               << current_block->getStartAddress());
 333                                                                 function->addBasicBlock(new_blocks.find(jmptarget)->second);
 334                                                         }
 335                                                         if (MIA->isConditionalBranch(inst)) {
 336                                                                 jmptarget = base_address + current_address + inst_size;
 337                                                                 current_block->setNextBlock(1, jmptarget);
 338                                                                 if (new_blocks.find(jmptarget) == new_blocks.end()) {
 339                                                                         BasicBlock * block = manager->newBasicBlock(jmptarget);
 340                                                                         assert(block);
 341                                                                         new_blocks.insert(std::make_pair(block->getStartAddress(), block));
 342                                                                         function->addBasicBlock(block);
 343                                                                         remaining_blocks.push(block);
 344                                                                 } else {
 345                                                                         LOG4CXX_DEBUG(logger, "Reusing Block starting at " << std::hex
 346                                                                                       << current_block->getStartAddress());
 347                                                                         function->addBasicBlock(new_blocks.find(jmptarget)->second);
 348                                                                 }
 349                                                         }
 350                                                 }
 351                                         }
 352                                 }
 353                         } else {
 354                                 inst_size = 0;
 355                         }
 356
 357
 358                         if (inst_size == 0 || MIA->isTerminator(inst) || MIA->isBranch(inst)) {
 359                                 current_block->setEndAddress(current_address + base_address + inst_size);
 360                                 LOG4CXX_DEBUG(logger, "Finished Block at " << std::hex <<
 361                                               current_block->getEndAddress());
 362                                 break;
 363                         }
 364                         current_address += inst_size;
 365                 }
 366         }
 367         splitBlocks(function);
 368         LOG4CXX_DEBUG(logger, "Finished function " << function->getName());
 369         manager->finishFunction(function);
 370         for (uint64_t address : called_functions)
 371                 disassembleFunctionAt(address);
 372 }
 373
 374 template <typename ELFT>
 375 void LLVMDisassembler<ELFT>::disassemble() {
 376         SectionRef text_section = getTextSection();
 377         std::vector<Function*> remaining_functions;
 378
 379         // Assume all function symbols actually start a real function
 380         for (auto x = symbols.begin(); x != symbols.end(); ++x) {
 381                 uint64_t result;
 382                 bool contains;
 383                 SymbolRef::Type symbol_type;
 384
 385 #if defined(LLVM_35)
 386                 if (text_section.containsSymbol(x->second, contains) || !contains)
 387 #elif defined(LLVM_36)
 388                 if (!text_section.containsSymbol(x->second))
 389 #endif
 390                         continue;
 391
 392                 if (x->second.getType(symbol_type)
 393                     || SymbolRef::ST_Function != symbol_type)
 394                         continue;
 395
 396                 if (!x->second.getAddress(result)) {
 397                         Function * fun = manager->newFunction(result);
 398                         if (fun) {
 399                                 fun->setName(x->first);
 400                                 remaining_functions.push_back(fun);
 401                                 LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
 402                         } else {
 403                                 LOG4CXX_DEBUG(logger, "Function at " << std::hex << result
 404                                               << " already disassembled as " << manager->getFunction(result)->getName());
 405                         }
 406                 }
 407         }
 408
 409         for (Function* function : remaining_functions) {
 410                 disassembleFunction(function);
 411                 manager->finishFunction(function);
 412         }
 413
 414         if (binary->isELF()) {
 415                 uint64_t _entryAddress = entryAddress();
 416                 LOG4CXX_DEBUG(logger, "Adding entryAddress at: " << std::hex << _entryAddress);
 417                 std::stringstream s;
 418                 s << "<_start 0x" << std::hex << _entryAddress << ">";
 419
 420                 disassembleFunctionAt(_entryAddress, s.str());
 421         }
 422
 423         if (!manager->hasFunctions()) {
 424                 uint64_t text_entry;
 425 #if defined(LLVM_35)
 426                 text_section.getAddress(text_entry);
 427 #elif defined(LLVM_36)
 428                 text_entry = text_section.getAddress();
 429 #endif
 430                 LOG4CXX_INFO(logger, "No Symbols found, starting at the beginning of the text segment");
 431                 disassembleFunctionAt(text_entry);
 432         }
 433 }
 434
 435 template <>
 436 uint64_t LLVMDisassembler<COFFT>::entryAddress() {
 437         const auto coffobject = dyn_cast<COFFObjectFile>(o);
 438         const struct pe32_header* pe32_header;
 439         const struct pe32plus_header* pe32plus_header;
 440
 441         coffobject->getPE32PlusHeader(pe32plus_header);
 442
 443         if (pe32plus_header) {
 444                 return pe32plus_header->AddressOfEntryPoint;
 445         } else {
 446                 coffobject->getPE32Header(pe32_header);
 447                 return pe32_header->AddressOfEntryPoint;
 448         }
 449 }
 450
 451 template<>
 452 uint64_t LLVMDisassembler<MACHOT>::entryAddress() {
 453         // TODO
 454         return 0;
 455 }
 456
 457 template <typename ELFT>
 458 uint64_t LLVMDisassembler<ELFT>::entryAddress() {
 459         const auto elffile = dyn_cast<ELFObjectFile<ELFT>>(o)->getELFFile();
 460         const auto * header = elffile->getHeader();
 461
 462         return header->e_entry;
 463 }
 464
 465 template <typename ELFT>
 466 void LLVMDisassembler<ELFT>::splitBlocks(Function* function) {
 467         SectionRef text_section = getTextSection();
 468         StringRef bytes;
 469         text_section.getContents(bytes);
 470 #if defined(LLVM_35)
 471         StringRefMemoryObject ref(bytes);
 472 #elif defined(LLVM_36)
 473         ArrayRef<uint8_t> bytearray(reinterpret_cast<const uint8_t *>(bytes.data()),
 474                            bytes.size());
 475 #endif
 476
 477
 478         LOG4CXX_DEBUG(logger, "Splitting Blocks in Function " << function->getName());
 479         // Split blocks where jumps are going inside the block
 480         for (auto it = function->blocks().begin();
 481              it != function->blocks().end();
 482              ++it) {
 483                 BasicBlock * current_block = it->second;
 484                 if (current_block->getEndAddress() == 0) {
 485                         LOG4CXX_ERROR(logger, "UNFINISHED BLOCK " << std::hex << current_block->getStartAddress());
 486                         break;
 487                 }
 488                 uint64_t inst_size;
 489                 uint64_t base_address;
 490 #if defined(LLVM_35)
 491                 text_section.getAddress(base_address);
 492 #elif defined(LLVM_36)
 493                 base_address = text_section.getAddress();
 494 #endif
 495                 uint64_t current_address = current_block->getStartAddress() - base_address;
 496                 while(current_block->getEndAddress() - base_address > current_address) {
 497                         MCInst inst;
 498                         std::string buf;
 499                         llvm::raw_string_ostream s(buf);
 500
 501                         if(llvm::MCDisassembler::Success ==
 502 #if defined(LLVM_35)
 503                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 504 #elif defined(LLVM_36)
 505                                 DisAsm->getInstruction(inst, inst_size,
 506                                                        bytearray.slice(current_address),
 507                                                        base_address + current_address,
 508                                                        nulls(), nulls())) {
 509 #endif
 510
 511                                 // See if some other block starts here
 512                                 BasicBlock* other = manager->getBasicBlock(current_address
 513                                                                            + inst_size
 514                                                                            + base_address);
 515
 516                                 // Special case, other block starts here but we are at the end anyway
 517                                 if (other != NULL) {
 518                                         uint64_t endaddress = current_address + inst_size + base_address;
 519                                         if (endaddress != current_block->getEndAddress()) {
 520                                                 LOG4CXX_DEBUG(logger, "Shortening block starting at "
 521                                                               << std::hex
 522                                                               << current_block->getStartAddress()
 523                                                               << " now ending at "
 524                                                               << other->getStartAddress());
 525                                                 function->addBasicBlock(other);
 526                                                 current_block->setEndAddress(endaddress);
 527                                                 current_block->setNextBlock(0, other->getStartAddress());
 528                                                 current_block->setNextBlock(1, 0);
 529                                         }
 530                                 }
 531                         } else {
 532                                 inst_size = 1;
 533                         }
 534                         current_address += inst_size;
 535                 }
 536         }
 537 }
 538
 539 template<>
 540 void LLVMDisassembler<COFFT>::readDynamicSymbols() {
 541         //TODO
 542 }
 543
 544 template<>
 545 void LLVMDisassembler<MACHOT>::readDynamicSymbols() {
 546         //TODO
 547 }
 548
 549 template <typename ELFT>
 550 void LLVMDisassembler<ELFT>::readDynamicSymbols() {
 551         const auto elffile = dyn_cast<ELFObjectFile<ELFT>>(o)->getELFFile();
 552         for (auto it = elffile->begin_dynamic_symbols(),
 553                      end = elffile->end_dynamic_symbols();
 554              it != end;
 555              ++it) {
 556                 if (it->getType() == 2) { // Function
 557                         bool is_default;
 558                         // TODO: Error handling
 559                         std::string symbolname = *(elffile->getSymbolName(it));
 560                         std::string symbolversion = *(elffile->getSymbolVersion(nullptr, &*it, is_default));
 561                         // TODO: actually get the symbol address from relocations
 562                         Function* f = manager->newDynamicFunction(0);
 563                         f->setName(symbolname + (is_default? "@@" : "@") + symbolversion);
 564                         manager->finishFunction(f);
 565
 566                         LOG4CXX_DEBUG(logger, "Adding dynamic Symbol " << symbolname << (is_default? "@@" : "@") << symbolversion);
 567                 }
 568         }
 569 }
 570
 571 template <typename ELFT>
 572 void LLVMDisassembler<ELFT>::readSymbols() {
 573         error_code ec;
 574         symbol_iterator si(o->symbol_begin()), se(o->symbol_end());
 575         for (; si != se; ++si) {
 576                 StringRef name;
 577                 if ((ec = si->getName(name))) {
 578                         LOG4CXX_ERROR(logger, ec.message());
 579                         break;
 580                 }
 581                 LOG4CXX_DEBUG(logger, "Added symbol " << name.str());
 582                 symbols.insert(make_pair(name.str(), *si));
 583         }
 584 }
 585
 586 template <typename ELFT>
 587 void LLVMDisassembler<ELFT>::readSections() {
 588         error_code ec;
 589         section_iterator i(o->section_begin()), e(o->section_end());
 590         for (; i != e; ++i) {
 591                 StringRef name;
 592                 if ((ec = i->getName(name))) {
 593                         LOG4CXX_ERROR(logger, ec.message());
 594                         break;
 595                 }
 596                 LOG4CXX_DEBUG(logger, "Added section " << name.str());
 597                 sections.insert(make_pair(name.str(), *i));
 598         }
 599
 600 }
 601
 602 // template <typename ELFT>
 603 // void LLVMDisassembler<ELFT>::forEachFunction(std::function<void (uint64_t, Function*)> callback) {
 604 //      // std::for_each(functions.begin(), functions.end(),
 605 //      //               [&](std::pair<uint64_t, Function*> x) {
 606 //      //                    callback(x.first, x.second);
 607 //      //               });
 608 // }
 609
 610 template <typename ELFT>
 611 std::vector<Instruction> LLVMDisassembler<ELFT>::getInstructions(const BasicBlock *block) {
 612         std::vector<Instruction> result;
 613         SectionRef text_section = getTextSection();
 614         uint64_t base_address;
 615 #if defined(LLVM_35)
 616         text_section.getAddress(base_address);
 617 #elif defined(LLVM_36)
 618         base_address = text_section.getAddress();
 619 #endif
 620
 621         uint64_t current_address = block->getStartAddress() - base_address;
 622         uint64_t end_position = block->getEndAddress() - base_address;
 623
 624         StringRef bytes;
 625         text_section.getContents(bytes);
 626 #if defined(LLVM_35)
 627         StringRefMemoryObject ref(bytes);
 628 #elif defined(LLVM_36)
 629         ArrayRef<uint8_t> bytearray(reinterpret_cast<const uint8_t *>(bytes.data()),
 630                            bytes.size());
 631 #endif
 632
 633
 634         while (current_address < end_position) {
 635                 uint64_t inst_size;
 636                 MCInst inst;
 637                 std::string buf;
 638                 llvm::raw_string_ostream s(buf);
 639
 640                 if(llvm::MCDisassembler::Success ==
 641 #if defined(LLVM_35)
 642                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 643 #elif defined(LLVM_36)
 644                                 DisAsm->getInstruction(inst, inst_size,
 645                                                        bytearray.slice(current_address),
 646                                                        base_address + current_address,
 647                                                        nulls(), nulls())) {
 648 #endif
 649
 650                         uint8_t bytes[inst_size+2];
 651 #if defined(LLVM_35)
 652                         ref.readBytes(current_address, inst_size, bytes);
 653 #elif defined(LLVM_36)
 654                         size_t bytesindex(0);
 655                         for (uint8_t byte : bytearray.slice(current_address, inst_size)) {
 656                                 bytes[bytesindex++] = byte;
 657                         }
 658 #endif
 659
 660                         uint64_t jmptarget;
 661                         std::string ref("");
 662                         IP->printInst(&inst, s, "");
 663                         if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 664                                 std::stringstream stream;
 665                                 if (MIA->isCall(inst))
 666                                         stream << "function:";
 667                                 else
 668                                         stream << "block:";
 669
 670                                 stream << std::hex << (base_address + jmptarget);
 671                                 ref = stream.str();
 672                         }
 673                         result.push_back(Instruction(current_address + base_address, boost::algorithm::trim_copy(s.str()),
 674                                                      std::vector<uint8_t>(bytes, bytes+inst_size), ref));
 675                 } else {
 676                         LOG4CXX_WARN(logger, "Invalid byte at" << std::hex << current_address + base_address);
 677                         uint8_t bytes[1];
 678 #if defined(LLVM_35)
 679                         ref.readBytes(current_address, 1, bytes);
 680 #elif defined(LLVM_36)
 681                         bytes[0] = bytearray[current_address];
 682 #endif
 683                         result.push_back(Instruction(current_address + base_address, "Invalid Instruction",
 684                                                      std::vector<uint8_t>(bytes, bytes+1), ""));
 685                         inst_size = 1;
 686                 }
 687
 688                 current_address += inst_size;
 689         }
 690         return result;
 691 }
 692
 693 template <typename ELFT>
 694 void LLVMDisassembler<ELFT>::printEachInstruction(uint64_t start, uint64_t end,
 695                                                   std::function<void (uint8_t*, size_t,
 696                                                                           const std::string&,
 697                                                                           const std::string&)> fun) {
 698         SectionRef text_section = getTextSection();
 699         uint64_t base_address;
 700 #if defined(LLVM_35)
 701         text_section.getAddress(base_address);
 702 #elif defined(LLVM_36)
 703         base_address = text_section.getAddress();
 704 #endif
 705
 706         uint64_t current_address = start - base_address;
 707
 708         StringRef bytes;
 709         text_section.getContents(bytes);
 710 #if defined(LLVM_35)
 711         StringRefMemoryObject ref(bytes);
 712 #elif defined(LLVM_36)
 713         ArrayRef<uint8_t> bytearray(reinterpret_cast<const uint8_t *>(bytes.data()),
 714                            bytes.size());
 715 #endif
 716
 717
 718         while (current_address < end - base_address) {
 719                 uint64_t inst_size;
 720                 MCInst inst;
 721                 std::string buf;
 722                 llvm::raw_string_ostream s(buf);
 723
 724                 if(llvm::MCDisassembler::Success ==
 725 #if defined(LLVM_35)
 726                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 727 #elif defined(LLVM_36)
 728                                 DisAsm->getInstruction(inst, inst_size,
 729                                                        bytearray.slice(current_address),
 730                                                        base_address + current_address,
 731                                                        nulls(), nulls())) {
 732 #endif
 733
 734                         uint8_t bytes[inst_size+2];
 735 #if defined(LLVM_35)
 736                         ref.readBytes(current_address, inst_size, bytes);
 737 #elif defined(LLVM_36)
 738                         size_t bytesindex(0);
 739                         for (uint8_t byte : bytearray.slice(current_address, inst_size)) {
 740                                 bytes[bytesindex++] = byte;
 741                         }
 742 #endif
 743
 744                         uint64_t jmptarget;
 745                         std::string ref("");
 746                         IP->printInst(&inst, s, "");
 747                         if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 748                                 std::stringstream stream;
 749                                 if (MIA->isCall(inst))
 750                                         stream << "function:";
 751                                 else
 752                                         stream << "block:";
 753
 754                                 stream << std::hex << (base_address + jmptarget);
 755                                 ref = stream.str();
 756                         }
 757
 758
 759                         fun(bytes, inst_size, s.str(), ref);
 760                 } else {
 761                         LOG4CXX_WARN(logger, "Invalid byte at" << std::hex << current_address + base_address);
 762                         fun(NULL, 0, "Invalid Byte", "");
 763                         inst_size = 1;
 764                 }
 765
 766                 current_address += inst_size;
 767         }
 768 }
 769
 770 template <typename ELFT>
 771 SectionRef LLVMDisassembler<ELFT>::getTextSection() {
 772         return sections[".text"];
 773 }
 774
 775 template <>
 776 SectionRef LLVMDisassembler<MACHOT>::getTextSection() {
 777         return sections["__text"];
 778 }