X-Git-Url: https://git.siccegge.de//index.cgi?p=frida%2Ffrida.git;a=blobdiff_plain;f=src%2Fdisassembler%2Fllvm%2FLLVMDisassembler.cxx;h=990ed4f53c19c6f24255d2fb735d1bedb4ba8722;hp=96d08830e4f26087df5cd3feb373f5f47a781bf5;hb=1ef802ee9a14e6274b3e60db873965b794f49abe;hpb=cab6494b97d1626c7b9285b69df324d9c9953614

diff --git a/src/disassembler/llvm/LLVMDisassembler.cxx b/src/disassembler/llvm/LLVMDisassembler.cxx
index 96d0883..990ed4f 100644
--- a/src/disassembler/llvm/LLVMDisassembler.cxx
+++ b/src/disassembler/llvm/LLVMDisassembler.cxx
@@ -1,3 +1,4 @@
+#include "disassembler/Instruction.hxx"
 #include "disassembler/llvm/LLVMDisassembler.hxx"
 #include "core/InformationManager.hxx"
 #include "core/Function.hxx"
@@ -15,6 +16,10 @@ namespace {
 	class COFFT {
 
 	};
+
+	class MACHOT {
+
+	};
 }
 
 /*
@@ -44,6 +49,9 @@ 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;
 }
@@ -177,7 +185,7 @@ 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 +221,7 @@ 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);
@@ -225,6 +233,11 @@ 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 +246,6 @@ 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;
@@ -305,7 +316,7 @@ 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
@@ -324,9 +335,14 @@ 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());
+			}
 		}
 	}
 
@@ -368,6 +384,12 @@ 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();
@@ -378,16 +400,21 @@ 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);
@@ -432,6 +459,11 @@ 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();
@@ -444,7 +476,11 @@ 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);
 		}
 	}
@@ -489,12 +525,66 @@ 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, 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;
@@ -540,3 +630,13 @@ 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"];
+}