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     for (auto x = symbols.begin(); x != symbols.end(); ++x) {
 130         uint64_t result;
 131         bool contains;
 132         SymbolRef::Type symbol_type;
 133
 134
 135         if (text_section.containsSymbol(x->second, contains) || !contains)
 136             continue;
 137
 138         if (x->second.getType(symbol_type)
 139             || SymbolRef::ST_Function != symbol_type)
 140             continue;
 141
 142         if (!x->second.getAddress(result)) {
 143             LLVMFunction * fun = new LLVMFunction(x->first, result);
 144             remaining_functions.push(fun);
 145             functions.insert(std::make_pair(result, fun));
 146             LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
 147         }
 148     }
 149
 150     StringRef bytes;
 151     text_section.getContents(bytes);
 152     StringRefMemoryObject ref(bytes);
 153
 154     while (remaining_functions.size()) {
 155         LLVMFunction * current_function = remaining_functions.top();
 156         remaining_functions.pop();
 157
 158         LOG4CXX_DEBUG(logger, "Handling function " << current_function->getName());
 159
 160         // if ("_start" != current_function->getName())
 161         //  continue;
 162
 163         LLVMBasicBlock * block = new LLVMBasicBlock(current_function->getStartAddress(), this);
 164         remaining_blocks.push(block);
 165         blocks.insert(std::make_pair(block->getStartAddress(), block));
 166
 167         while (remaining_blocks.size()) {
 168             LLVMBasicBlock * current_block = remaining_blocks.top();
 169             remaining_blocks.pop();
 170
 171             LOG4CXX_DEBUG(logger, "Handling Block starting at " << std::hex << current_block->getStartAddress());
 172
 173             uint64_t inst_size;
 174             uint64_t base_address;
 175             text_section.getAddress(base_address);
 176             uint64_t current_address = current_block->getStartAddress() - base_address;
 177             while(true) {
 178                 MCInst inst;
 179                 std::string buf;
 180                 llvm::raw_string_ostream s(buf);
 181
 182                 if(llvm::MCDisassembler::Success ==
 183                    DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 184
 185                     uint8_t bytes[inst_size+2];
 186                     ref.readBytes(current_address, inst_size, bytes);
 187                     s << '\t';
 188                     for(uint8_t* cur = bytes; cur < bytes + inst_size; ++cur) {
 189                         s.write_hex(*cur);
 190                         s << ' ';
 191                     }
 192                     s << '\t';
 193
 194                     IP->printInst(&inst, s, "");
 195
 196                     LOG4CXX_DEBUG(logger, std::hex << current_address + base_address << s.str());
 197
 198                     uint64_t jmptarget;
 199                     if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
 200                         jmptarget += base_address;
 201                         if (!MIA->isIndirectBranch(inst)) {
 202                             if (MIA->isCall(inst)) {
 203                                 if (functions.find(jmptarget) == functions.end()) {
 204                                     std::stringstream s;
 205                                     s << "<Unnamed 0x" << std::hex << jmptarget << ">";
 206                                     LLVMFunction * fun = new LLVMFunction(s.str(), jmptarget);
 207                                     functions.insert(std::make_pair(jmptarget, fun));
 208                                     remaining_functions.push(fun);
 209                                 }
 210                             } else {
 211                                 if (blocks.find(jmptarget) == blocks.end()) {
 212                                     LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
 213                                     blocks.insert(std::make_pair(block->getStartAddress(), block));
 214                                     current_block->setNextBlock(0, block->getStartAddress());
 215                                     remaining_blocks.push(block);
 216                                 }
 217                                 if (MIA->isConditionalBranch(inst)) {
 218                                     jmptarget = base_address + current_address + inst_size;
 219                                     if (blocks.find(jmptarget) == blocks.end()) {
 220                                         LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
 221                                         blocks.insert(std::make_pair(block->getStartAddress(), block));
 222                                         current_block->setNextBlock(1, block->getStartAddress());
 223                                         remaining_blocks.push(block);
 224                                     }
 225                                 }
 226                             }
 227                         }
 228                     }
 229                 } else {
 230                     inst_size = 0;
 231                 }
 232
 233
 234                 if (inst_size == 0 || MIA->isTerminator(inst) || MIA->isBranch(inst)) {
 235                     current_block->setEndAddress(current_address + base_address + inst_size);
 236                     LOG4CXX_DEBUG(logger, "Finished Block at " << std::hex <<
 237                                   current_block->getEndAddress());
 238                     break;
 239                 }
 240                 current_address += inst_size;
 241             }
 242         }
 243         LOG4CXX_DEBUG(logger, "Finished function " << current_function->getName());
 244     }
 245
 246         // Split blocks where jumps are going inside the block
 247         for (auto it = blocks.begin(); it != blocks.end(); ++it) {
 248                 LLVMBasicBlock * current_block = it->second;
 249                 uint64_t inst_size;
 250                 uint64_t base_address;
 251                 text_section.getAddress(base_address);
 252                 uint64_t current_address = current_block->getStartAddress() - base_address;
 253                 while(current_block->getEndAddress() - base_address != current_address) {
 254                         MCInst inst;
 255                         std::string buf;
 256                         llvm::raw_string_ostream s(buf);
 257
 258                         if(llvm::MCDisassembler::Success ==
 259                            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 260                                 auto other = blocks.find(current_address + inst_size + base_address);
 261
 262                                 if (other != blocks.end()) {
 263                                         LOG4CXX_DEBUG(logger, "Shortening block starting at "
 264                                                                   << std::hex
 265                                                                   << current_block->getStartAddress()
 266                                                                   << " now ending at "
 267                                                                   << other->first);
 268                                         current_block->setEndAddress(current_address + inst_size + base_address);
 269                                         current_block->setNextBlock(0, other->first);
 270                                         current_block->setNextBlock(1, 0);
 271                                 }
 272                         } else {
 273                                 inst_size = 1;
 274                         }
 275                         current_address += inst_size;
 276                 }
 277         }
 278 }
 279
 280 void LLVMDisassembler::readSymbols() {
 281     error_code ec;
 282     symbol_iterator si(o->symbol_begin()), se(o->symbol_end());
 283     for (; si != se; ++si) {
 284         StringRef name;
 285         if ((ec = si->getName(name))) {
 286             LOG4CXX_ERROR(logger, ec.message());
 287             break;
 288         }
 289         LOG4CXX_DEBUG(logger, "Added symbol " << name.str());
 290         symbols.insert(make_pair(name.str(), *si));
 291     }
 292 }
 293
 294 void LLVMDisassembler::readSections() {
 295     error_code ec;
 296     section_iterator i(o->section_begin()), e(o->section_end());
 297     for (; i != e; ++i) {
 298         StringRef name;
 299         if ((ec = i->getName(name))) {
 300             LOG4CXX_ERROR(logger, ec.message());
 301             break;
 302         }
 303         LOG4CXX_DEBUG(logger, "Added section " << name.str());
 304         sections.insert(make_pair(name.str(), *i));
 305     }
 306
 307 }
 308
 309 void LLVMDisassembler::forEachFunction(std::function<void (uint64_t, Function*)> callback) {
 310     std::for_each(functions.begin(), functions.end(),
 311                   [&](std::pair<uint64_t, LLVMFunction*> x) {
 312                       callback(x.first, x.second);
 313                   });
 314 }
 315
 316 void LLVMDisassembler::printEachInstruction(uint64_t start, uint64_t end,
 317                                                                                         std::function<void (uint8_t*, size_t, const std::string&)> fun) {
 318     SectionRef text_section = sections[".text"];
 319     uint64_t base_address;
 320     text_section.getAddress(base_address);
 321     uint64_t current_address = start - base_address;
 322
 323     StringRef bytes;
 324     text_section.getContents(bytes);
 325     StringRefMemoryObject ref(bytes);
 326
 327     while (current_address < end - base_address) {
 328         uint64_t inst_size;
 329         MCInst inst;
 330                 std::string buf;
 331                 llvm::raw_string_ostream s(buf);
 332
 333         if(llvm::MCDisassembler::Success ==
 334            DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
 335
 336             uint8_t bytes[inst_size+2];
 337             ref.readBytes(current_address, inst_size, bytes);
 338
 339             IP->printInst(&inst, s, "");
 340                         fun(bytes, inst_size, s.str());
 341         } else {
 342                         fun(NULL, 0, "Invalid Byte");
 343                         inst_size = 1;
 344                 }
 345
 346                 current_address += inst_size;
 347     }
 348 }