X-Git-Url: https://git.siccegge.de//index.cgi?p=frida%2Ffrida.git;a=blobdiff_plain;f=src%2Fdisassembler%2Fllvm%2FLLVMDisassembler.cxx;h=21766bb344cb708a473188f577659d03723bd1c5;hp=cdcd3cb2cd516883400d08a9b47bef5c94a0341d;hb=099f1e8222109bf7397eff6da0c511a07906c9cd;hpb=4b2336f2066f76d2be017ad9731865ac6743abcb

diff --git a/src/disassembler/llvm/LLVMDisassembler.cxx b/src/disassembler/llvm/LLVMDisassembler.cxx
index cdcd3cb..21766bb 100644
--- a/src/disassembler/llvm/LLVMDisassembler.cxx
+++ b/src/disassembler/llvm/LLVMDisassembler.cxx
@@ -15,6 +15,10 @@ namespace {
 	class COFFT {
 
 	};
+
+	class MACHOT {
+
+	};
 }
 
 /*
@@ -44,6 +48,9 @@ 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;
 }
@@ -58,7 +65,7 @@ LLVMDisassembler<ELFT>::LLVMDisassembler(const std::string& filename,
                                          InformationManager* manager,
                                          ObjectFile* file)
 	: Disassembler()
-	, logger(log4cxx::Logger::getLogger("LLVMDisassembler"))
+	, logger(log4cxx::Logger::getLogger("disassembler.LLVMDisassembler"))
 	, triple("unknown-unknown-unknown")
 	, manager(manager)
 {
@@ -133,10 +140,10 @@ LLVMDisassembler<ELFT>::LLVMDisassembler(const std::string& filename,
 	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();
@@ -177,7 +184,7 @@ 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);
@@ -199,7 +206,6 @@ Function* LLVMDisassembler<ELFT>::disassembleFunctionAt(uint64_t address, const
 			function->setName(name);
 		}
 		disassembleFunction(function);
-		manager->finishFunction(function);
 	}
 
 	return function;
@@ -207,13 +213,14 @@ Function* LLVMDisassembler<ELFT>::disassembleFunctionAt(uint64_t address, const
 
 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);
@@ -225,6 +232,11 @@ void LLVMDisassembler<ELFT>::disassembleFunction(Function* function) {
 	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();
@@ -233,8 +245,6 @@ void LLVMDisassembler<ELFT>::disassembleFunction(Function* function) {
 		              << 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;
@@ -250,7 +260,7 @@ void LLVMDisassembler<ELFT>::disassembleFunction(Function* function) {
 					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()) {
@@ -298,12 +308,14 @@ void LLVMDisassembler<ELFT>::disassembleFunction(Function* function) {
 	}
 	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
@@ -322,9 +334,14 @@ 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());
+			}
 		}
 	}
 
@@ -366,6 +383,12 @@ 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();
@@ -376,16 +399,21 @@ 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);
 
+	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);
@@ -430,6 +458,11 @@ 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();
@@ -492,7 +525,7 @@ 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;
@@ -538,3 +571,13 @@ 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"];
+}