X-Git-Url: https://git.siccegge.de//index.cgi?p=frida%2Ffrida.git;a=blobdiff_plain;f=src%2Fdisassembler%2Fllvm%2FLLVMDisassembler.cxx;h=45cd04e25566d97f830df05be7a00f42ffbe29a8;hp=96418d86bbdded11fd4c530aef48f571c4300cd6;hb=9d5b0bc7a1a5670e30637c1e71c432a4fd9255ca;hpb=3e7900dab2b4618d0c4fbe48d312cb74666af842

diff --git a/src/disassembler/llvm/LLVMDisassembler.cxx b/src/disassembler/llvm/LLVMDisassembler.cxx
index 96418d8..45cd04e 100644
--- a/src/disassembler/llvm/LLVMDisassembler.cxx
+++ b/src/disassembler/llvm/LLVMDisassembler.cxx
@@ -1,7 +1,9 @@
+#include "disassembler/Instruction.hxx"
 #include "disassembler/llvm/LLVMDisassembler.hxx"
 #include "core/InformationManager.hxx"
 #include "core/Function.hxx"
 #include "core/BasicBlock.hxx"
+#include <boost/algorithm/string.hpp>
 
 #include <stack>
 #include <algorithm>
@@ -15,18 +17,32 @@ namespace {
 	class COFFT {
 
 	};
+
+	class MACHOT {
+
+	};
 }
 
 /*
  *
  */
 Disassembler * createLLVMDisassembler(const std::string& filename, InformationManager* manager) {
+	log4cxx::LoggerPtr logger(log4cxx::Logger::getLogger("disassembler.LLVMDisassembler"));
 	if (filename == "")
 		return NULL;
 
-	std::unique_ptr<Binary> o;
-	o.reset(createBinary(filename).get());
-	Binary * op = o.release();
+	auto retval = createBinary(filename);
+	if (error_code ec = retval.getError()) {
+		LOG4CXX_ERROR(logger, ec.message());
+		return NULL;
+	}
+
+	Binary * op = retval.get();
+
+	if (!op) {
+		LOG4CXX_ERROR(logger, "Could not open " << filename);
+		return NULL;
+	}
 
 	// ELFType<endian, maxalign, 64bit>
 	if (ELF32LEObjectFile * object = dyn_cast<ELF32LEObjectFile>(op)) {
@@ -44,6 +60,9 @@ Disassembler * createLLVMDisassembler(const std::string& filename, InformationMa
 	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;
 }
@@ -159,7 +178,7 @@ LLVMDisassembler<ELFT>::LLVMDisassembler(const std::string& filename,
 
 	std::unique_ptr<MCObjectDisassembler> OD(
 		new MCObjectDisassembler(*o, *DisAsm, *MIA));
-	Mod.reset(OD->buildModule(false));
+	//Mod.reset(OD->buildModule(false));
 
 	readSections();
 }
@@ -177,7 +196,7 @@ LLVMDisassembler<ELFT>::~LLVMDisassembler() {}
 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);
@@ -213,7 +232,7 @@ void LLVMDisassembler<ELFT>::disassembleFunction(Function* function) {
 	 * 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);
@@ -308,7 +327,7 @@ void LLVMDisassembler<ELFT>::disassembleFunction(Function* function) {
 
 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
@@ -327,9 +346,14 @@ void LLVMDisassembler<ELFT>::disassemble() {
 
 		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());
+			}
 		}
 	}
 
@@ -371,6 +395,12 @@ uint64_t LLVMDisassembler<COFFT>::entryAddress() {
 	}
 }
 
+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();
@@ -381,7 +411,7 @@ uint64_t LLVMDisassembler<ELFT>::entryAddress() {
 
 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);
@@ -440,6 +470,11 @@ void LLVMDisassembler<COFFT>::readDynamicSymbols() {
 	//TODO
 }
 
+template<>
+void LLVMDisassembler<MACHOT>::readDynamicSymbols() {
+	//TODO
+}
+
 template <typename ELFT>
 void LLVMDisassembler<ELFT>::readDynamicSymbols() {
 	const auto elffile = dyn_cast<ELFObjectFile<ELFT>>(o)->getELFFile();
@@ -452,7 +487,11 @@ void LLVMDisassembler<ELFT>::readDynamicSymbols() {
 			// 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);
 		}
 	}
@@ -497,12 +536,66 @@ void LLVMDisassembler<ELFT>::readSections() {
 // 	//               });
 // }
 
+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, boost::algorithm::trim_copy(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;
@@ -548,3 +641,13 @@ void LLVMDisassembler<ELFT>::printEachInstruction(uint64_t start, uint64_t end,
 		current_address += inst_size;
 	}
 }
+
+template <typename ELFT>
+SectionRef LLVMDisassembler<ELFT>::getTextSection() {
+	return sections[".text"];
+}
+
+template <>
+SectionRef LLVMDisassembler<MACHOT>::getTextSection() {
+	return sections["__text"];
+}