From: Christoph Egger <siccegge@faui49man12.informatik.uni-erlangen.de>
Date: Thu, 8 Jan 2015 17:42:04 +0000 (+0100)
Subject: Split blocks before finishing function
X-Git-Tag: v0.1~139
X-Git-Url: https://git.siccegge.de//index.cgi?a=commitdiff_plain;h=9252262693432b33dbe4ffc60d79bbdc6fbb5f66;p=frida%2Ffrida.git

Split blocks before finishing function
---

diff --git a/src/disassembler/Function.hxx b/src/disassembler/Function.hxx
index ccca54c..57833ba 100644
--- a/src/disassembler/Function.hxx
+++ b/src/disassembler/Function.hxx
@@ -2,6 +2,7 @@
 #define INCLUDE__Function_hxx
 
 #include "disassembler/BasicBlock.hxx"
+#include <map>
 
 class Function {
 public:
@@ -17,9 +18,18 @@ public:
 	std::string getName() const {
 		return name;
 	}
+
+	void addBasicBlock(BasicBlock* block) {
+		_blocks.insert(std::make_pair(block->getStartAddress(), block));
+	}
+
+	std::map<uint64_t, BasicBlock*>& blocks() {
+		return _blocks;
+	}
 private:
 	std::string name;
 	uint64_t start_address;
+	std::map<uint64_t, BasicBlock*> _blocks;
 };
 
 #endif
diff --git a/src/disassembler/llvm/LLVMDisassembler.cxx b/src/disassembler/llvm/LLVMDisassembler.cxx
index 79c7a54..c73e703 100644
--- a/src/disassembler/llvm/LLVMDisassembler.cxx
+++ b/src/disassembler/llvm/LLVMDisassembler.cxx
@@ -138,7 +138,7 @@ Function* LLVMDisassembler::disassembleFunctionAt(uint64_t address, const std::s
 	text_section.getAddress(base_address);
 	text_section.getSize(size);
 
-	if (address < base_address || 
+	if (address < base_address ||
 	    address >= base_address + size) {
 		return NULL;
 	}
@@ -174,6 +174,7 @@ void LLVMDisassembler::disassembleFunction(LLVMFunction* function) {
 	LLVMBasicBlock * block = new LLVMBasicBlock(function->getStartAddress(), this);
 	remaining_blocks.push(block);
 	blocks.insert(std::make_pair(block->getStartAddress(), block));
+	function->addBasicBlock(block);
 
 	while (remaining_blocks.size()) {
 		LLVMBasicBlock * current_block = remaining_blocks.top();
@@ -206,6 +207,7 @@ void LLVMDisassembler::disassembleFunction(LLVMFunction* function) {
 							if (blocks.find(jmptarget) == blocks.end()) {
 								LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
 								blocks.insert(std::make_pair(block->getStartAddress(), block));
+								function->addBasicBlock(block);
 								remaining_blocks.push(block);
 							}
 							if (MIA->isConditionalBranch(inst)) {
@@ -214,6 +216,7 @@ void LLVMDisassembler::disassembleFunction(LLVMFunction* function) {
 								if (blocks.find(jmptarget) == blocks.end()) {
 									LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
 									blocks.insert(std::make_pair(block->getStartAddress(), block));
+									function->addBasicBlock(block);
 									remaining_blocks.push(block);
 								}
 							}
@@ -234,6 +237,7 @@ void LLVMDisassembler::disassembleFunction(LLVMFunction* function) {
 			current_address += inst_size;
 		}
 	}
+	splitBlocks(function);
 	LOG4CXX_DEBUG(logger, "Finished function " << function->getName());
 	manager->signal_new_function(function);
 }
@@ -293,19 +297,19 @@ void LLVMDisassembler::disassemble() {
 		LOG4CXX_INFO(logger, "No Symbols found, starting at the beginning of the text segment");
 		disassembleFunctionAt(text_entry);
 	}
-
-	splitBlocks();
 }
 
-void LLVMDisassembler::splitBlocks() {
+void LLVMDisassembler::splitBlocks(LLVMFunction* function) {
 	SectionRef text_section = sections[".text"];
 	StringRef bytes;
 	text_section.getContents(bytes);
 	StringRefMemoryObject ref(bytes);
 
 	// Split blocks where jumps are going inside the block
-	for (auto it = blocks.begin(); it != blocks.end(); ++it) {
-		LLVMBasicBlock * current_block = it->second;
+	for (auto it = function->blocks().begin();
+	     it != function->blocks().end();
+	     ++it) {
+		BasicBlock * current_block = it->second;
 		uint64_t inst_size;
 		uint64_t base_address;
 		text_section.getAddress(base_address);
@@ -317,8 +321,10 @@ void LLVMDisassembler::splitBlocks() {
 
 			if(llvm::MCDisassembler::Success ==
 			   DisAsm->getInstruction(inst, inst_size, ref, current_address, nulls(), nulls())) {
+				// See if some other block starts here
 				auto other = blocks.find(current_address + inst_size + base_address);
 
+				// Special case, other block starts here but we are at the end anyway
 				if (other != blocks.end()) {
 					uint64_t endaddress = current_address + inst_size + base_address;
 					if (endaddress != current_block->getEndAddress()) {
@@ -377,7 +383,8 @@ void LLVMDisassembler::forEachFunction(std::function<void (uint64_t, Function*)>
 }
 
 void LLVMDisassembler::printEachInstruction(uint64_t start, uint64_t end,
-                                            std::function<void (uint8_t*, size_t, const std::string&)> fun) {
+                                            std::function<void (uint8_t*, size_t,
+                                                                const std::string&)> fun) {
 	SectionRef text_section = sections[".text"];
 	uint64_t base_address;
 	text_section.getAddress(base_address);
diff --git a/src/disassembler/llvm/LLVMDisassembler.hxx b/src/disassembler/llvm/LLVMDisassembler.hxx
index 107767d..b0ab082 100644
--- a/src/disassembler/llvm/LLVMDisassembler.hxx
+++ b/src/disassembler/llvm/LLVMDisassembler.hxx
@@ -41,7 +41,7 @@ protected:
 private:
 	// http://llvm.org/docs/doxygen/html/MCObjectDisassembler_8cpp_source.html +197
 	void disassembleFunction(LLVMFunction* function);
-	void splitBlocks();
+	void splitBlocks(LLVMFunction* fun);
 	void disassemble();
 
 	void readSymbols();
diff --git a/src/disassembler/llvm/LLVMFunction.hxx b/src/disassembler/llvm/LLVMFunction.hxx
index 3bc813a..0bdb1dc 100644
--- a/src/disassembler/llvm/LLVMFunction.hxx
+++ b/src/disassembler/llvm/LLVMFunction.hxx
@@ -11,6 +11,3 @@ private:
 };
 
 #endif
-
-
-