+#include "disassembler/Instruction.hxx"
#include "disassembler/llvm/LLVMDisassembler.hxx"
#include "core/InformationManager.hxx"
#include "core/Function.hxx"
using namespace llvm::object;
using std::error_code;
+namespace {
+ class COFFT {
+
+ };
+
+ class MACHOT {
+
+ };
+}
+
/*
*
*/
if (ELF64BEObjectFile * object = dyn_cast<ELF64BEObjectFile>(op)) {
return new LLVMDisassembler<ELFType<support::big, 2, true>>(filename, manager, object);
}
+ if (COFFObjectFile * object = dyn_cast<COFFObjectFile>(op)) {
+ return new LLVMDisassembler<COFFT>(filename, manager, object);
+ }
+ if (MachOObjectFile * object = dyn_cast<MachOObjectFile>(op)) {
+ return new LLVMDisassembler<MACHOT>(filename, manager, object);
+ }
return NULL;
}
template <typename ELFT>
LLVMDisassembler<ELFT>::LLVMDisassembler(const std::string& filename,
InformationManager* manager,
- ELFObjectFile<ELFT>* file)
+ ObjectFile* file)
: Disassembler()
- , logger(log4cxx::Logger::getLogger("LLVMDisassembler"))
+ , logger(log4cxx::Logger::getLogger("disassembler.LLVMDisassembler"))
, triple("unknown-unknown-unknown")
, manager(manager)
{
- LOG4CXX_DEBUG(logger, "Handling file" << filename);
+ LOG4CXX_DEBUG(logger, "Handling file " << filename);
if (!file) {
auto result = createBinary(filename);
binary.reset(result.get());
- o = dyn_cast<ELFObjectFile<ELFT>>(binary.get());
+ o = dyn_cast<ObjectFile>(binary.get());
} else {
o = file;
binary.reset(file);
RelInfo.reset(
target->createMCRelocationInfo(tripleName, Ctx));
if (RelInfo) {
- Symzer.reset(
- MCObjectSymbolizer::createObjectSymbolizer(Ctx, std::move(RelInfo), o));
- if (Symzer)
- DisAsm->setSymbolizer(std::move(Symzer));
+ // Symzer.reset(
+ // MCObjectSymbolizer::createObjectSymbolizer(Ctx, std::move(RelInfo), o));
+ // if (Symzer)
+ // DisAsm->setSymbolizer(std::move(Symzer));
}
RelInfo.release();
Symzer.release();
std::unique_ptr<MCObjectDisassembler> OD(
new MCObjectDisassembler(*o, *DisAsm, *MIA));
Mod.reset(OD->buildModule(false));
+
+ readSections();
}
template <typename ELFT>
void LLVMDisassembler<ELFT>::start() {
readSymbols();
- readSections();
disassemble();
readDynamicSymbols();
}
template <typename ELFT>
Function* LLVMDisassembler<ELFT>::disassembleFunctionAt(uint64_t address, const std::string& name) {
Function * function;
- SectionRef text_section = sections[".text"];
+ SectionRef text_section = getTextSection();
uint64_t base_address, size;
text_section.getAddress(base_address);
text_section.getSize(size);
function->setName(name);
}
disassembleFunction(function);
- manager->finishFunction(function);
}
return function;
template <typename ELFT>
void LLVMDisassembler<ELFT>::disassembleFunction(Function* function) {
+ std::vector<uint64_t> called_functions;
std::stack<BasicBlock*> remaining_blocks;
/* TODO:
* Do all blocks get added properly? We should take care to remove
* the other ones at the end of the function!
*/
std::map<uint64_t, BasicBlock*> new_blocks;
- SectionRef text_section = sections[".text"];
+ SectionRef text_section = getTextSection();
StringRef bytes;
text_section.getContents(bytes);
StringRefMemoryObject ref(bytes);
new_blocks.insert(std::make_pair(block->getStartAddress(), block));
function->addBasicBlock(block);
+ uint64_t base_address, size;
+ text_section.getAddress(base_address);
+ text_section.getSize(size);
+ LOG4CXX_DEBUG(logger, "Text section at " << std::hex << base_address << " with size " << size);
+
while (remaining_blocks.size()) {
BasicBlock * current_block = remaining_blocks.top();
remaining_blocks.pop();
<< current_block->getStartAddress());
uint64_t inst_size;
- uint64_t base_address;
- text_section.getAddress(base_address);
uint64_t current_address = current_block->getStartAddress() - base_address;
while(true) {
MCInst inst;
if (!MIA->isIndirectBranch(inst)) {
if (MIA->isCall(inst)) {
if (NULL == manager->getFunction(jmptarget))
- disassembleFunctionAt(jmptarget);
+ called_functions.push_back(jmptarget);
} else {
current_block->setNextBlock(0, jmptarget);
if (new_blocks.find(jmptarget) == new_blocks.end()) {
}
splitBlocks(function);
LOG4CXX_DEBUG(logger, "Finished function " << function->getName());
- manager->signal_new_function(function);
+ manager->finishFunction(function);
+ for (uint64_t address : called_functions)
+ disassembleFunctionAt(address);
}
template <typename ELFT>
void LLVMDisassembler<ELFT>::disassemble() {
- SectionRef text_section = sections[".text"];
+ SectionRef text_section = getTextSection();
std::vector<Function*> remaining_functions;
// Assume all function symbols actually start a real function
if (!x->second.getAddress(result)) {
Function * fun = manager->newFunction(result);
- fun->setName(x->first);
- remaining_functions.push_back(fun);
- LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
+ if (fun) {
+ fun->setName(x->first);
+ remaining_functions.push_back(fun);
+ LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
+ } else {
+ LOG4CXX_DEBUG(logger, "Function at " << std::hex << result
+ << " already disassembled as " << manager->getFunction(result)->getName());
+ }
}
}
}
if (binary->isELF()) {
- const ELFO * elffile = o->getELFFile();
- const typename ELFO::Elf_Ehdr * header = elffile->getHeader();
-
- _entryAddress = header->e_entry;
+ uint64_t _entryAddress = entryAddress();
LOG4CXX_DEBUG(logger, "Adding entryAddress at: " << std::hex << _entryAddress);
std::stringstream s;
s << "<_start 0x" << std::hex << _entryAddress << ">";
}
}
+template <>
+uint64_t LLVMDisassembler<COFFT>::entryAddress() {
+ const auto coffobject = dyn_cast<COFFObjectFile>(o);
+ const struct pe32_header* pe32_header;
+ const struct pe32plus_header* pe32plus_header;
+
+ coffobject->getPE32PlusHeader(pe32plus_header);
+
+ if (pe32plus_header) {
+ return pe32plus_header->AddressOfEntryPoint;
+ } else {
+ coffobject->getPE32Header(pe32_header);
+ return pe32_header->AddressOfEntryPoint;
+ }
+}
+
+template<>
+uint64_t LLVMDisassembler<MACHOT>::entryAddress() {
+ // TODO
+ return 0;
+}
+
+template <typename ELFT>
+uint64_t LLVMDisassembler<ELFT>::entryAddress() {
+ const auto elffile = dyn_cast<ELFObjectFile<ELFT>>(o)->getELFFile();
+ const auto * header = elffile->getHeader();
+
+ return header->e_entry;
+}
+
template <typename ELFT>
void LLVMDisassembler<ELFT>::splitBlocks(Function* function) {
- SectionRef text_section = sections[".text"];
+ SectionRef text_section = getTextSection();
StringRef bytes;
text_section.getContents(bytes);
StringRefMemoryObject ref(bytes);
+ LOG4CXX_DEBUG(logger, "Splitting Blocks in Function " << function->getName());
// Split blocks where jumps are going inside the block
for (auto it = function->blocks().begin();
it != function->blocks().end();
++it) {
BasicBlock * current_block = it->second;
+ if (current_block->getEndAddress() == 0) {
+ LOG4CXX_ERROR(logger, "UNFINISHED BLOCK " << std::hex << current_block->getStartAddress());
+ break;
+ }
uint64_t inst_size;
uint64_t base_address;
text_section.getAddress(base_address);
}
}
+template<>
+void LLVMDisassembler<COFFT>::readDynamicSymbols() {
+ //TODO
+}
+
+template<>
+void LLVMDisassembler<MACHOT>::readDynamicSymbols() {
+ //TODO
+}
+
template <typename ELFT>
void LLVMDisassembler<ELFT>::readDynamicSymbols() {
- const ELFO * elffile = o->getELFFile();
- for (typename ELFO::Elf_Sym_Iter
- it = elffile->begin_dynamic_symbols(),
+ const auto elffile = dyn_cast<ELFObjectFile<ELFT>>(o)->getELFFile();
+ for (auto it = elffile->begin_dynamic_symbols(),
end = elffile->end_dynamic_symbols();
it != end;
++it) {
// TODO: Error handling
std::string symbolname = *(elffile->getSymbolName(it));
std::string symbolversion = *(elffile->getSymbolVersion(nullptr, &*it, is_default));
- manager->signal_new_dyn_symbol(symbolname + (is_default? "@@" : "@") + symbolversion);
+ // TODO: actually get the symbol address from relocations
+ Function* f = manager->newDynamicFunction(0);
+ f->setName(symbolname + (is_default? "@@" : "@") + symbolversion);
+ manager->finishFunction(f);
+
LOG4CXX_DEBUG(logger, "Adding dynamic Symbol " << symbolname << (is_default? "@@" : "@") << symbolversion);
}
}
// // });
// }
+template <typename ELFT>
+std::vector<Instruction> LLVMDisassembler<ELFT>::getInstructions(const BasicBlock *block) {
+ std::vector<Instruction> result;
+ SectionRef text_section = getTextSection();
+ uint64_t base_address;
+ text_section.getAddress(base_address);
+ uint64_t current_address = block->getStartAddress() - base_address;
+ uint64_t end_position = block->getEndAddress() - base_address;
+
+ StringRef bytes;
+ text_section.getContents(bytes);
+ StringRefMemoryObject ref(bytes);
+
+ while (current_address < end_position) {
+ uint64_t inst_size;
+ MCInst inst;
+ std::string buf;
+ llvm::raw_string_ostream s(buf);
+
+ if(llvm::MCDisassembler::Success ==
+ DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
+
+ uint8_t bytes[inst_size+2];
+ ref.readBytes(current_address, inst_size, bytes);
+
+ uint64_t jmptarget;
+ std::string ref("");
+ IP->printInst(&inst, s, "");
+ if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
+ std::stringstream stream;
+ if (MIA->isCall(inst))
+ stream << "function:";
+ else
+ stream << "block:";
+
+ stream << std::hex << (base_address + jmptarget);
+ ref = stream.str();
+ }
+ result.push_back(Instruction(current_address + base_address, s.str(),
+ std::vector<uint8_t>(bytes, bytes+inst_size), ref));
+ } else {
+ LOG4CXX_WARN(logger, "Invalid byte at" << std::hex << current_address + base_address);
+ uint8_t bytes[1];
+ ref.readBytes(current_address, 1, bytes);
+ result.push_back(Instruction(current_address + base_address, "Invalid Instruction",
+ std::vector<uint8_t>(bytes, bytes+1), ""));
+ inst_size = 1;
+ }
+
+ current_address += inst_size;
+ }
+ return result;
+}
+
template <typename ELFT>
void LLVMDisassembler<ELFT>::printEachInstruction(uint64_t start, uint64_t end,
std::function<void (uint8_t*, size_t,
const std::string&,
const std::string&)> fun) {
- SectionRef text_section = sections[".text"];
+ SectionRef text_section = getTextSection();
uint64_t base_address;
text_section.getAddress(base_address);
uint64_t current_address = start - base_address;
current_address += inst_size;
}
}
+
+template <typename ELFT>
+SectionRef LLVMDisassembler<ELFT>::getTextSection() {
+ return sections[".text"];
+}
+
+template <>
+SectionRef LLVMDisassembler<MACHOT>::getTextSection() {
+ return sections["__text"];
+}
projects / frida / frida.git / blobdiff