X-Git-Url: https://git.siccegge.de//index.cgi?p=frida%2Ffrida.git;a=blobdiff_plain;f=src%2Fdisassembler%2Fllvm%2FLLVMDisassembler.cxx;h=a194d2144d459d407efa769596a2975e72ed0a44;hp=e1570758da27831e02d1ca63fe85fa326297bda1;hb=f47dd6b106b9ba1d6363277b47cdec2ac0448f6c;hpb=30ce0f02e529603965d15a3afa71c4612dd40c4a

diff --git a/src/disassembler/llvm/LLVMDisassembler.cxx b/src/disassembler/llvm/LLVMDisassembler.cxx
index e157075..a194d21 100644
--- a/src/disassembler/llvm/LLVMDisassembler.cxx
+++ b/src/disassembler/llvm/LLVMDisassembler.cxx
@@ -13,7 +13,6 @@ using namespace llvm::object;
  * ist sondern z.B. einfach nur Instruktionen oder ein Bootsektor oder
  * foo
  */
-
 LLVMDisassembler::LLVMDisassembler(const std::string& filename)
     : Disassembler(filename)
     , logger(log4cxx::Logger::getLogger("LLVMDisassembler"))
@@ -111,30 +110,45 @@ LLVMDisassembler::LLVMDisassembler(const std::string& filename)
     disassemble();
 }
 
+LLVMDisassembler::~LLVMDisassembler() {
+    std::for_each(functions.begin(), functions.end(),
+                  [](std::pair<uint64_t,LLVMFunction*> it) {
+                      delete it.second;
+                  });
+    std::for_each(blocks.begin(), blocks.end(),
+                  [](std::pair<uint64_t, LLVMBasicBlock*> it) {
+                      delete it.second;
+                  });
+}
 
+/*
+ * TODO: If we jump into some Basic Block we need to split it there into two
+ */
 void LLVMDisassembler::disassemble() {
     std::stack<LLVMFunction*> remaining_functions;
     std::stack<LLVMBasicBlock*> remaining_blocks;
     SectionRef text_section = sections[".text"];
 
-    std::for_each(symbols.begin(), symbols.end(),
-                  [&](std::pair<const std::string, SymbolRef> x) {
-                      uint64_t result;
-                      bool contains;
-                      SymbolRef::Type symbol_type;
+    for (auto x = symbols.begin(); x != symbols.end(); ++x) {
+        uint64_t result;
+        bool contains;
+        SymbolRef::Type symbol_type;
 
-                      if (text_section.containsSymbol(x.second, contains) || !contains)
-                          return;
 
-                      if (x.second.getType(symbol_type)
-                          || SymbolRef::ST_Function != symbol_type)
-                          return;
+        if (text_section.containsSymbol(x->second, contains) || !contains)
+            continue;
 
-                      if (!x.second.getAddress(result)) {
-                          remaining_functions.push(new LLVMFunction(x.first, result));
-                          LOG4CXX_DEBUG(logger, "Disasembling " << x.first);
-                      }
-                  });
+        if (x->second.getType(symbol_type)
+            || SymbolRef::ST_Function != symbol_type)
+            continue;
+
+        if (!x->second.getAddress(result)) {
+            LLVMFunction * fun = new LLVMFunction(x->first, result);
+            remaining_functions.push(fun);
+            functions.insert(std::make_pair(result, fun));
+            LOG4CXX_DEBUG(logger, "Disasembling " << x->first);
+        }
+    }
 
     StringRef bytes;
     text_section.getContents(bytes);
@@ -149,7 +163,9 @@ void LLVMDisassembler::disassemble() {
         // if ("_start" != current_function->getName())
         //  continue;
 
-        remaining_blocks.push(new LLVMBasicBlock(current_function->getStartAddress()));
+        LLVMBasicBlock * block = new LLVMBasicBlock(current_function->getStartAddress(), this);
+        remaining_blocks.push(block);
+        blocks.insert(std::make_pair(block->getStartAddress(), block));
 
         while (remaining_blocks.size()) {
             LLVMBasicBlock * current_block = remaining_blocks.top();
@@ -184,18 +200,31 @@ void LLVMDisassembler::disassemble() {
 
                     uint64_t jmptarget;
                     if (MIA->evaluateBranch(inst, current_address, inst_size, jmptarget)) {
-						jmptarget += base_address;
+                        jmptarget += base_address;
                         if (!MIA->isIndirectBranch(inst)) {
                             if (MIA->isCall(inst)) {
-								if (blocks.find(jmptarget) == blocks.end())
-									remaining_functions.push(new LLVMFunction("<Unnamed>", jmptarget));
+                                if (functions.find(jmptarget) == functions.end()) {
+                                    std::stringstream s;
+                                    s << "<Unnamed 0x" << std::hex << jmptarget << ">";
+                                    LLVMFunction * fun = new LLVMFunction(s.str(), jmptarget);
+                                    functions.insert(std::make_pair(jmptarget, fun));
+                                    remaining_functions.push(fun);
+                                }
                             } else {
-								if (blocks.find(jmptarget) == blocks.end())
-                                remaining_blocks.push(new LLVMBasicBlock(jmptarget));
+                                if (blocks.find(jmptarget) == blocks.end()) {
+                                    LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
+                                    blocks.insert(std::make_pair(block->getStartAddress(), block));
+                                    current_block->setNextBlock(0, block->getStartAddress());
+                                    remaining_blocks.push(block);
+                                }
                                 if (MIA->isConditionalBranch(inst)) {
-									jmptarget = base_address + current_address + inst_size;
-									if (blocks.find(jmptarget) == blocks.end())
-										remaining_blocks.push(new LLVMBasicBlock(jmptarget));
+                                    jmptarget = base_address + current_address + inst_size;
+                                    if (blocks.find(jmptarget) == blocks.end()) {
+                                        LLVMBasicBlock * block = new LLVMBasicBlock(jmptarget, this);
+                                        blocks.insert(std::make_pair(block->getStartAddress(), block));
+                                        current_block->setNextBlock(1, block->getStartAddress());
+                                        remaining_blocks.push(block);
+                                    }
                                 }
                             }
                         }
@@ -206,9 +235,9 @@ void LLVMDisassembler::disassemble() {
 
 
                 if (inst_size == 0 || MIA->isTerminator(inst) || MIA->isBranch(inst)) {
-                    current_block->setEndAddress(current_address + base_address);
-					blocks.insert(std::make_pair(current_block->getStartAddress(), current_block));
-                    LOG4CXX_DEBUG(logger, "Finished Block at " << current_block->getEndAddress());
+                    current_block->setEndAddress(current_address + base_address + inst_size);
+                    LOG4CXX_DEBUG(logger, "Finished Block at " << std::hex <<
+                                  current_block->getEndAddress());
                     break;
                 }
                 current_address += inst_size;
@@ -247,6 +276,43 @@ void LLVMDisassembler::readSections() {
 
 }
 
-BasicBlock * LLVMDisassembler::generateControlFlowGraph(uint64_t address) {
+void LLVMDisassembler::forEachFunction(std::function<void (uint64_t, Function*)> callback) {
+    std::for_each(functions.begin(), functions.end(),
+                  [&](std::pair<uint64_t, LLVMFunction*> x) {
+                      callback(x.first, x.second);
+                  });
+}
+
+void LLVMDisassembler::printEachInstruction(uint64_t start, uint64_t end,
+											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);
+    uint64_t current_address = start - base_address;
+
+    StringRef bytes;
+    text_section.getContents(bytes);
+    StringRefMemoryObject ref(bytes);
+
+    while (current_address < end - base_address) {
+        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);
+
+            IP->printInst(&inst, s, "");
+			fun(bytes, inst_size, s.str());
+        } else {
+			fun(NULL, 0, "Invalid Byte");
+			inst_size = 1;
+		}
+
+		current_address += inst_size;
+    }
 }