src/disassembler/llvm/LLVMDisassembler.cxx

   1 #include "disassembler/llvm/LLVMDisassembler.hxx"
   2 #include "disassembler/llvm/LLVMBasicBlock.hxx"
   3 #include "disassembler/llvm/LLVMFunction.hxx"
   4
   5 #include <stack>
   6 #include <algorithm>
   7
   8 using namespace llvm;
   9 using namespace llvm::object;
  10
  11 /*
  12  * TODO: fallback code falls die Datei kein ELF/PE/COFF/MacO/.. binary
  13  * ist sondern z.B. einfach nur Instruktionen oder ein Bootsektor oder
  14  * foo
  15  */
  16 LLVMDisassembler::LLVMDisassembler(const std::string& filename)
  17     : Disassembler(filename)
  18     , logger(log4cxx::Logger::getLogger("LLVMDisassembler"))
  19     , triple("unknown-unknown-unknown")
  20 {
  21     LOG4CXX_DEBUG(logger, "Handling file" << filename);
  22     auto result = createBinary(filename);
  23
  24     error_code ec;
  25     if ((ec = result.getError())) {
  26         LOG4CXX_ERROR(logger, "Failed to load Binary" << ec.message());
  27         binary = NULL;
  28         return;
  29     }
  30
  31     binary.reset(result.get());
  32
  33     o = dyn_cast<ObjectFile>(binary.get());
  34
  35     triple.setArch(Triple::ArchType(o->getArch()));
  36     std::string tripleName(triple.getTriple());
  37
  38     LOG4CXX_INFO(logger, "Architecture " << tripleName);
  39
  40
  41     std::string es;
  42     target = TargetRegistry::lookupTarget("", triple, es);
  43     if (!target) {
  44         LOG4CXX_ERROR(logger, es);
  45         return;
  46     }
  47
  48     LOG4CXX_INFO(logger, "Target " << target->getName());
  49
  50     MRI.reset(target->createMCRegInfo(tripleName));
  51     if (!MRI) {
  52         LOG4CXX_ERROR(logger, "no register info for target " << tripleName);
  53         return;
  54     }
  55
  56     // Set up disassembler.
  57     AsmInfo.reset(target->createMCAsmInfo(*MRI, tripleName));
  58     if (!AsmInfo) {
  59         LOG4CXX_ERROR(logger, "no assembly info for target " << tripleName);
  60         return;
  61     }
  62
  63     STI.reset(target->createMCSubtargetInfo(tripleName, "", ""));
  64     if (!STI) {
  65         LOG4CXX_ERROR(logger, "no subtarget info for target " << tripleName);
  66         return;
  67     }
  68
  69     MII.reset(target->createMCInstrInfo());
  70     if (!MII) {
  71         LOG4CXX_ERROR(logger, "no instruction info for target " << tripleName);
  72         return;
  73     }
  74
  75     DisAsm.reset(target->createMCDisassembler(*STI));
  76     if (!DisAsm) {
  77         LOG4CXX_ERROR(logger, "no disassembler for target " << tripleName);
  78         return;
  79     }
  80
  81     MOFI.reset(new MCObjectFileInfo);
  82     Ctx.reset(new MCContext(AsmInfo.get(), MRI.get(), MOFI.get()));
  83     RelInfo.reset(
  84         target->createMCRelocationInfo(tripleName, *Ctx.get()));
  85     if (RelInfo) {
  86         Symzer.reset(
  87             MCObjectSymbolizer::createObjectSymbolizer(*Ctx.get(), RelInfo, o));
  88         if (Symzer)
  89             DisAsm->setSymbolizer(Symzer);
  90     }
  91
  92     MIA.reset(target->createMCInstrAnalysis(MII.get()));
  93
  94     int AsmPrinterVariant = AsmInfo->getAssemblerDialect();
  95     IP.reset(target->createMCInstPrinter(AsmPrinterVariant, *AsmInfo, *MII, *MRI, *STI));
  96     if (!IP) {
  97         LOG4CXX_ERROR(logger, "no instruction printer for target " << tripleName);
  98         return;
  99     }
 100
 101     IP->setPrintImmHex(llvm::HexStyle::C);
 102     IP->setPrintImmHex(true);
 103
 104     OwningPtr<MCObjectDisassembler> OD(
 105         new MCObjectDisassembler(*o, *DisAsm, *MIA));
 106     Mod.reset(OD->buildModule(false));
 107
 108     readSymbols();
 109     readSections();
 110     disassemble();
 111 }
 112
 113 LLVMDisassembler::~LLVMDisassembler() {
 114     std::for_each(functions.begin(), functions.end(),
 115                   [](std::pair<uint64_t,LLVMFunction*> it) {
 116                       delete it.second;
 117                   });
 118     std::for_each(blocks.begin(), blocks.end(),
 119                   [](std::pair<uint64_t, LLVMBasicBlock*> it) {
 120                       delete it.second;
 121                   });
 122 }
 123
 124 void LLVMDisassembler::disassemble() {
 125     std::stack<LLVMFunction*> remaining_functions;
 126     std::stack<LLVMBasicBlock*> remaining_blocks;
 127     SectionRef text_section = sections[".text"];
 128
 129         // Assume all function symbols actually start a real function
 130     for (auto x = symbols.begin(); x != symbols.end(); ++x) {
 131         uint64_t result;
 132         bool contains;
 133         SymbolRef::Type symbol_type;
 134
 135
 136         if (text_section.containsSymbol(x->second, contains) || !contains)
 137             continue;
 138
 139         if (x->second.getType(symbol_type)
 140             || SymbolRef::ST_Function != symbol_type)
 141             continue;
 142
 143         if (!x->second.getAddress(result)) {
 144             LLVMFunction * fun = new LLVMFunction(x->first, result);
 145             remaining_functions.push(fun);
 146             functions.insert(std::make_pair(result, fun));
 147             LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
 148         }
 149     }
 150
 151     StringRef bytes;
 152     text_section.getContents(bytes);
 153     StringRefMemoryObject ref(bytes);
 154
 155     while (remaining_functions.size()) {
 156         LLVMFunction * current_function = remaining_functions.top();
 157         remaining_functions.pop();
 158
 159         LOG4CXX_DEBUG(logger, "Handling function " << current_function->getName());
 160
 161         LLVMBasicBlock * block = new LLVMBasicBlock(current_function->getStartAddress(), this);
 162         remaining_blocks.push(block);
 163         blocks.insert(std::make_pair(block->getStartAddress(), block));
 164
 165         while (remaining_blocks.size()) {
 166             LLVMBasicBlock * current_block = remaining_blocks.top();
 167             remaining_blocks.pop();
 168
 169             LOG4CXX_DEBUG(logger, "Handling Block starting at " << std::hex << current_block->getStartAddress());
 170
 171             uint64_t inst_size;
 172             uint64_t base_address;
 173             text_section.getAddress(base_address);
 174             uint64_t current_address = current_block->getStartAddress() - base_address;
 175             while(true) {
 176                 MCInst inst;
 177                 std::string buf;
 178                 llvm::raw_string_ostream s(buf);
 179
 180                 if(llvm::MCDisassembler::Success ==
 181                    DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 182
 183                     uint64_t jmptarget;
 184                     if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 185                         jmptarget += base_address;
 186                         if (!MIA->isIndirectBranch(inst)) {
 187                             if (MIA->isCall(inst)) {
 188                                 if (functions.find(jmptarget) == functions.end()) {
 189                                     std::stringstream s;
 190                                     s << "<Unnamed 0x" << std::hex << jmptarget << ">";
 191                                     LLVMFunction * fun = new LLVMFunction(s.str(), jmptarget);
 192                                     functions.insert(std::make_pair(jmptarget, fun));
 193                                     remaining_functions.push(fun);
 194                                 }
 195                             } else {
 196                                 if (blocks.find(jmptarget) == blocks.end()) {
 197                                     LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
 198                                     blocks.insert(std::make_pair(block->getStartAddress(), block));
 199                                     current_block->setNextBlock(0, block->getStartAddress());
 200                                     remaining_blocks.push(block);
 201                                 }
 202                                 if (MIA->isConditionalBranch(inst)) {
 203                                     jmptarget = base_address + current_address + inst_size;
 204                                     if (blocks.find(jmptarget) == blocks.end()) {
 205                                         LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
 206                                         blocks.insert(std::make_pair(block->getStartAddress(), block));
 207                                         current_block->setNextBlock(1, block->getStartAddress());
 208                                         remaining_blocks.push(block);
 209                                     }
 210                                 }
 211                             }
 212                         }
 213                     }
 214                 } else {
 215                     inst_size = 0;
 216                 }
 217
 218
 219                 if (inst_size == 0 || MIA->isTerminator(inst) || MIA->isBranch(inst)) {
 220                     current_block->setEndAddress(current_address + base_address + inst_size);
 221                     LOG4CXX_DEBUG(logger, "Finished Block at " << std::hex <<
 222                                   current_block->getEndAddress());
 223                     break;
 224                 }
 225                 current_address += inst_size;
 226             }
 227         }
 228         LOG4CXX_DEBUG(logger, "Finished function " << current_function->getName());
 229     }
 230
 231         // Split blocks where jumps are going inside the block
 232         for (auto it = blocks.begin(); it != blocks.end(); ++it) {
 233                 LLVMBasicBlock * current_block = it->second;
 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(current_block->getEndAddress() - base_address != current_address) {
 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                                 auto other = blocks.find(current_address + inst_size + base_address);
 246
 247                                 if (other != blocks.end()) {
 248                                         LOG4CXX_DEBUG(logger, "Shortening block starting at "
 249                                                                   << std::hex
 250                                                                   << current_block->getStartAddress()
 251                                                                   << " now ending at "
 252                                                                   << other->first);
 253                                         current_block->setEndAddress(current_address + inst_size + base_address);
 254                                         current_block->setNextBlock(0, other->first);
 255                                         current_block->setNextBlock(1, 0);
 256                                 }
 257                         } else {
 258                                 inst_size = 1;
 259                         }
 260                         current_address += inst_size;
 261                 }
 262         }
 263 }
 264
 265 void LLVMDisassembler::readSymbols() {
 266     error_code ec;
 267     symbol_iterator si(o->symbol_begin()), se(o->symbol_end());
 268     for (; si != se; ++si) {
 269         StringRef name;
 270         if ((ec = si->getName(name))) {
 271             LOG4CXX_ERROR(logger, ec.message());
 272             break;
 273         }
 274         LOG4CXX_DEBUG(logger, "Added symbol " << name.str());
 275         symbols.insert(make_pair(name.str(), *si));
 276     }
 277 }
 278
 279 void LLVMDisassembler::readSections() {
 280     error_code ec;
 281     section_iterator i(o->section_begin()), e(o->section_end());
 282     for (; i != e; ++i) {
 283         StringRef name;
 284         if ((ec = i->getName(name))) {
 285             LOG4CXX_ERROR(logger, ec.message());
 286             break;
 287         }
 288         LOG4CXX_DEBUG(logger, "Added section " << name.str());
 289         sections.insert(make_pair(name.str(), *i));
 290     }
 291
 292 }
 293
 294 void LLVMDisassembler::forEachFunction(std::function<void (uint64_t, Function*)> callback) {
 295     std::for_each(functions.begin(), functions.end(),
 296                   [&](std::pair<uint64_t, LLVMFunction*> x) {
 297                       callback(x.first, x.second);
 298                   });
 299 }
 300
 301 void LLVMDisassembler::printEachInstruction(uint64_t start, uint64_t end,
 302                                                                                         std::function<void (uint8_t*, size_t, const std::string&)> fun) {
 303     SectionRef text_section = sections[".text"];
 304     uint64_t base_address;
 305     text_section.getAddress(base_address);
 306     uint64_t current_address = start - base_address;
 307
 308     StringRef bytes;
 309     text_section.getContents(bytes);
 310     StringRefMemoryObject ref(bytes);
 311
 312     while (current_address < end - base_address) {
 313         uint64_t inst_size;
 314         MCInst inst;
 315                 std::string buf;
 316                 llvm::raw_string_ostream s(buf);
 317
 318         if(llvm::MCDisassembler::Success ==
 319            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 320
 321             uint8_t bytes[inst_size+2];
 322             ref.readBytes(current_address, inst_size, bytes);
 323
 324             IP->printInst(&inst, s, "");
 325                         fun(bytes, inst_size, s.str());
 326         } else {
 327                         LOG4CXX_WARN(logger, "Invalid byte at" << std::hex << current_address + base_address);
 328                         fun(NULL, 0, "Invalid Byte");
 329                         inst_size = 1;
 330                 }
 331
 332                 current_address += inst_size;
 333     }
 334 }