/*==============================================================================
* FUNCTION: FrontEnd::processProc
* OVERVIEW: Process a procedure, given a native (source machine) address.
* PARAMETERS: address - the address at which the procedure starts
* pProc - the procedure object
* frag - if true, this is just a fragment of a procedure
* spec - if true, this is a speculative decode
* os - the output stream for .rtl output
* NOTE: This is a sort of generic front end. For many processors, this will be overridden
* in the FrontEnd derived class, sometimes calling this function to do most of the work
* RETURNS: true for a good decode (no illegal instructions)
*============================================================================*/
bool FrontEnd::processProc(ADDRESS uAddr, UserProc* pProc, std::ofstream &os, bool frag /* = false */,
bool spec /* = false */) {
PBB pBB; // Pointer to the current basic block
std::cout<<"Entering Processing Proc\n";
// just in case you missed it
first_line = true;
if (AssProgram)
std::cout <<"Name Of Program : " << AssProgram->name << std::endl;
Boomerang::get()->alert_new(pProc);
// We have a set of CallStatement pointers. These may be disregarded if this is a speculative decode
// that fails (i.e. an illegal instruction is found). If not, this set will be used to add to the set of calls
// to be analysed in the cfg, and also to call newProc()
std::list<CallStatement*> callList;
// Indicates whether or not the next instruction to be decoded is the lexical successor of the current one.
// Will be true for all NCTs and for CTIs with a fall through branch.
bool sequentialDecode = true;
Cfg* pCfg = pProc->getCFG();
// If this is a speculative decode, the second time we decode the same address, we get no cfg. Else an error.
if (spec && (pCfg == 0))
return false;
assert(pCfg);
// Initialise the queue of control flow targets that have yet to be decoded.
targetQueue.initial(uAddr);
// Clear the pointer used by the caller prologue code to access the last call rtl of this procedure
//decoder.resetLastCall();
// ADDRESS initAddr = uAddr;
int nTotalBytes = 0;
ADDRESS startAddr = uAddr;
ADDRESS lastAddr = uAddr;
ADDRESS address = uAddr;
std::cout << "Start at address = " << uAddr << std::endl;
//------IMPORTANT------------------------------------------------------------------------
list<AssemblyLabel*>::iterator lbi;
list<AssemblyLine*>* temp_lines = new list<AssemblyLine*>();
if (AssProgram){
for(lbi = AssProgram->labelList->begin(); lbi != AssProgram->labelList->end(); ++lbi ){
if((*lbi)->address == uAddr){
temp_lines = (*lbi)->lineList;
std::cout << "***DECODE LABEL: " << (*lbi)->name << std::endl;
std::cout << "***AT ADDRESS: " << (*lbi)->address << std::endl;
std::cout << "***NUMBER OF INSTRUCTION: " << (*lbi)->lineList->size() << std::endl;
break;
}
}
}
list<AssemblyLine*>::iterator li;
if (temp_lines->size()>0)
li = temp_lines->begin();
//---------------------------------------------------------------------------------------
while ((uAddr = targetQueue.nextAddress(pCfg)) != NO_ADDRESS) {
// The list of RTLs for the current basic block
std::list<RTL*>* BB_rtls = new std::list<RTL*>();
// Keep decoding sequentially until a CTI without a fall through branch is decoded
//ADDRESS start = uAddr;
DecodeResult inst;
while (sequentialDecode) {
// Decode and classify the current source instruction
if (Boomerang::get()->traceDecoder)
LOG << "*" << uAddr << "\t";
// Decode the inst at uAddr.
if(ASS_FILE){
if(li != temp_lines->end()){
inst = decodeAssemblyInstruction(uAddr,"assemblySets.at(line)", (*li));
}
}
else
inst = decodeInstruction(uAddr);
// If invalid and we are speculating, just exit
if (spec && !inst.valid)
return false;
// Need to construct a new list of RTLs if a basic block has just been finished but decoding is
// continuing from its lexical successor
if (BB_rtls == NULL)
BB_rtls = new std::list<RTL*>();
RTL* pRtl = inst.rtl;
if (inst.valid == false) {
// Alert the watchers to the problem
Boomerang::get()->alert_baddecode(uAddr);
// An invalid instruction. Most likely because a call did not return (e.g. call _exit()), etc.
// Best thing is to emit a INVALID BB, and continue with valid instructions
if (VERBOSE) {
LOG << "Warning: invalid instruction at " << uAddr << ": ";
// Emit the next 4 bytes for debugging
for (int ii=0; ii < 4; ii++)
LOG << (unsigned)(pBF->readNative1(uAddr + ii) & 0xFF) << " ";
LOG << "\n";
}
// Emit the RTL anyway, so we have the address and maybe some other clues
BB_rtls->push_back(new RTL(uAddr));
pBB = pCfg->newBB(BB_rtls, INVALID, 0);
sequentialDecode = false; BB_rtls = NULL; continue;
}
//pProc->unionDefine = new list<UnionDefine*>();
pProc->bitVar = AssProgram->bitVar;
pProc->replacement = AssProgram->replacement;
// alert the watchers that we have decoded an instruction
Boomerang::get()->alert_decode(uAddr, inst.numBytes);
nTotalBytes += inst.numBytes;
// Check if this is an already decoded jump instruction (from a previous pass with propagation etc)
// If so, we throw away the just decoded RTL (but we still may have needed to calculate the number
// of bytes.. ick.)
std::map<ADDRESS, RTL*>::iterator ff = previouslyDecoded.find(uAddr);
if (ff != previouslyDecoded.end())
pRtl = ff->second;
if (pRtl == NULL) {
// This can happen if an instruction is "cancelled", e.g. call to __main in a hppa program
// Just ignore the whole instruction
if (inst.numBytes > 0)
uAddr += inst.numBytes;
continue;
}
// Display RTL representation if asked
std::cout<<"RTL: "<<std::endl;
std::ostringstream st;
pRtl->print(st);
std::cout << st.str().c_str()<<std::endl;
ADDRESS uDest;
// For each Statement in the RTL
//std::list<Statement*>& sl = pRtl->getList();
std::list<Statement*> sl = pRtl->getList();
// Make a copy (!) of the list. This is needed temporarily to work around the following problem.
// We are currently iterating an RTL, which could be a return instruction. The RTL is passed to
// createReturnBlock; if this is not the first return statement, it will get cleared, and this will
// cause problems with the current iteration. The effects seem to be worse for MSVC/Windows.
// This problem will likely be easier to cope with when the RTLs are removed, and there are special
// Statements to mark the start of instructions (and their native address).
// FIXME: However, this workaround breaks logic below where a GOTO is changed to a CALL followed by a return
// if it points to the start of a known procedure
std::list<Statement*>::iterator ss;
#if 1
for (ss = sl.begin(); ss != sl.end(); ss++) { // }
#else
// The counter is introduced because ss != sl.end() does not work as it should
// FIXME: why? Does this really fix the problem?
int counter = sl.size();
for (ss = sl.begin(); counter > 0; ss++, counter--) {
#endif
Statement* s = *ss;
s->setProc(pProc); // let's do this really early!
if (refHints.find(pRtl->getAddress()) != refHints.end()) {
const char *nam = refHints[pRtl->getAddress()].c_str();
ADDRESS gu = prog->getGlobalAddr((char*)nam);
if (gu != NO_ADDRESS) {
s->searchAndReplace(new Const((int)gu), new Unary(opAddrOf, Location::global(nam, pProc)));
}
}
s->simplify();
GotoStatement* stmt_jump = static_cast<GotoStatement*>(s);
// Check for a call to an already existing procedure (including self recursive jumps), or to the PLT
// (note that a LibProc entry for the PLT function may not yet exist)
ADDRESS dest;
Proc* proc;
if (s->getKind() == STMT_GOTO) {
dest = stmt_jump->getFixedDest();
if (dest != NO_ADDRESS) {
proc = prog->findProc(dest);
if (proc == NULL) {
if(!ASS_FILE){
if (pBF->IsDynamicLinkedProc(dest))
proc = prog->setNewProc(dest);
}
}
if (proc != NULL && proc != (Proc*)-1) {
s = new CallStatement();
CallStatement *call = static_cast<CallStatement*>(s);
call->setDest(dest);
call->setDestProc(proc);
call->setReturnAfterCall(true);
// also need to change it in the actual RTL
std::list<Statement*>::iterator ss1 = ss;
ss1++;
assert(ss1 == sl.end());
pRtl->replaceLastStmt(s);
*ss = s;
}
}
}
switch (s->getKind())
{
case STMT_GOTO: {
uDest = stmt_jump->getFixedDest();
// Handle one way jumps and computed jumps separately
if (uDest != NO_ADDRESS) {
BB_rtls->push_back(pRtl);
sequentialDecode = false;
pBB = pCfg->newBB(BB_rtls,ONEWAY,1);
BB_rtls = NULL; // Clear when make new BB
// Exit the switch now if the basic block already existed
if (pBB == 0) {
break;
}
// Add the out edge if it is to a destination within the
// procedure
if (uDest < pBF->getLimitTextHigh()) {
targetQueue.visit(pCfg, uDest, pBB);
pCfg->addOutEdge(pBB, uDest, true);
}
else {
std::cout<<"Entering Processing Proc5\n";
if (!ASS_FILE)
LOG << "Error: Instruction at " << uAddr << " branches beyond end of section, to "
<< uDest << "\n";
else{
targetQueue.visit(pCfg, uDest, pBB);
pCfg->addOutEdge(pBB, uDest, true);
}
}
}
break;
}
case STMT_CASE: {
Exp* pDest = stmt_jump->getDest();
if (pDest == NULL) { // Happens if already analysed (now redecoding)
// SWITCH_INFO* psi = ((CaseStatement*)stmt_jump)->getSwitchInfo();
BB_rtls->push_back(pRtl);
pBB = pCfg->newBB(BB_rtls, NWAY, 0); // processSwitch will update num outedges
pBB->processSwitch(pProc); // decode arms, set out edges, etc
sequentialDecode = false; // Don't decode after the jump
BB_rtls = NULL; // New RTLList for next BB
break; // Just leave it alone
}
// Check for indirect calls to library functions, especially in Win32 programs
if (pDest && pDest->getOper() == opMemOf &&
pDest->getSubExp1()->getOper() == opIntConst &&
pBF->IsDynamicLinkedProcPointer(((Const*)pDest->getSubExp1())->getAddr())) {
if (VERBOSE)
LOG << "jump to a library function: " << stmt_jump << ", replacing with a call/ret.\n";
// jump to a library function
// replace with a call ret
// TODO:
std::string func = pBF->GetDynamicProcName(
((Const*)stmt_jump->getDest()->getSubExp1())->getAddr());
//------------------------------------
CallStatement *call = new CallStatement;
call->setDest(stmt_jump->getDest()->clone());
LibProc *lp = pProc->getProg()->getLibraryProc(func.c_str());
if (lp == NULL)
LOG << "getLibraryProc returned NULL, aborting\n";
assert(lp);
call->setDestProc(lp);
std::list<Statement*>* stmt_list = new std::list<Statement*>;
stmt_list->push_back(call);
BB_rtls->push_back(new RTL(pRtl->getAddress(), stmt_list));
pBB = pCfg->newBB(BB_rtls, CALL, 1);
appendSyntheticReturn(pBB, pProc, pRtl);
sequentialDecode = false;
BB_rtls = NULL;
if (pRtl->getAddress() == pProc->getNativeAddress()) {
// it's a thunk
// Proc *lp = prog->findProc(func.c_str());
func = std::string("__imp_") + func;
pProc->setName(func.c_str());
//lp->setName(func.c_str());
Boomerang::get()->alert_update_signature(pProc);
}
callList.push_back(call);
ss = sl.end(); ss--; // get out of the loop
break;
}
BB_rtls->push_back(pRtl);
// We create the BB as a COMPJUMP type, then change to an NWAY if it turns out to be a switch stmt
pBB = pCfg->newBB(BB_rtls, COMPJUMP, 0);
LOG << "COMPUTED JUMP at " << uAddr << ", pDest = " << pDest << "\n";
if (Boomerang::get()->noDecompile) {
// try some hacks
if (pDest->isMemOf() && pDest->getSubExp1()->getOper() == opPlus &&
pDest->getSubExp1()->getSubExp2()->isIntConst()) {
// assume subExp2 is a jump table
ADDRESS jmptbl = ((Const*)pDest->getSubExp1()->getSubExp2())->getInt();
unsigned int i;
for (i = 0; ; i++) {
ADDRESS uDest = pBF->readNative4(jmptbl + i * 4);
if (pBF->getLimitTextLow() <= uDest && uDest < pBF->getLimitTextHigh()) {
LOG << " guessed uDest " << uDest << "\n";
targetQueue.visit(pCfg, uDest, pBB);
pCfg->addOutEdge(pBB, uDest, true);
} else
break;
}
pBB->updateType(NWAY, i);
}
}
sequentialDecode = false;
BB_rtls = NULL; // New RTLList for next BB
break;
}
case STMT_BRANCH: {
uDest = stmt_jump->getFixedDest();
BB_rtls->push_back(pRtl);
pBB = pCfg->newBB(BB_rtls, TWOWAY, 2);
// Stop decoding sequentially if the basic block already existed otherwise complete the basic block
if (pBB == 0)
sequentialDecode = false;
else {
// Add the out edge if it is to a destination within the procedure
if (!ASS_FILE){
if (uDest < pBF->getLimitTextHigh()) {
targetQueue.visit(pCfg, uDest, pBB);
pCfg->addOutEdge(pBB, uDest, true);
}
else
LOG << "Error: Instruction at " << uAddr << " branches beyond end of section, to "
<< uDest << "\n";
}
else {
targetQueue.visit(pCfg, uDest, pBB);
pCfg->addOutEdge(pBB, uDest, true);
}
// Add the fall-through outedge
pCfg->addOutEdge(pBB, uAddr + inst.numBytes);
}
// Create the list of RTLs for the next basic block and continue with the next instruction.
BB_rtls = NULL;
break;
}
case STMT_CALL: {
CallStatement* call = static_cast<CallStatement*>(s);
// Check for a dynamic linked library function
// TODO: solution dont use pBF
if (!ASS_FILE){
if (call->getDest()->getOper() == opMemOf &&
call->getDest()->getSubExp1()->getOper() == opIntConst &&
pBF->IsDynamicLinkedProcPointer(((Const*)call->getDest()->getSubExp1())->getAddr())) {
// Dynamic linked proc pointers are treated as static.
const char *nam = pBF->GetDynamicProcName( ((Const*)call->getDest()->getSubExp1())->getAddr());
Proc *p = pProc->getProg()->getLibraryProc(nam);
call->setDestProc(p);
call->setIsComputed(false);
}
}
else {
if (call->getDest()->getOper() == opMemOf &&
call->getDest()->getSubExp1()->getOper() == opIntConst &&
funcsType.find(((Const*)call->getDest()->getSubExp1())->getAddr())->second) {
// Dynamic linked proc pointers are treated as static.
const char *nam = namesList.find(((Const*)call->getDest()->getSubExp1())->getAddr())->second;
Proc *p = pProc->getProg()->getLibraryProc(nam);
call->setDestProc(p);
call->setIsComputed(false);
}
}
// Is the called function a thunk calling a library function?
// A "thunk" is a function which only consists of: "GOTO library_function"
// Should i modify
if (!ASS_FILE){
if( call && call->getFixedDest() != NO_ADDRESS ) {
// Get the address of the called function.
ADDRESS callAddr=call->getFixedDest();
// It should not be in the PLT either, but getLimitTextHigh() takes this into account
if (callAddr < pBF->getLimitTextHigh()) {
// Decode it.
DecodeResult decoded=decodeInstruction(callAddr);
if (decoded.valid) { // is the instruction decoded succesfully?
// Yes, it is. Create a Statement from it.
RTL *rtl = decoded.rtl;
Statement* first_statement = *rtl->getList().begin();
if (first_statement) {
first_statement->setProc(pProc);
first_statement->simplify();
GotoStatement* stmt_jump = static_cast<GotoStatement*>(first_statement);
// In fact it's a computed (looked up) jump, so the jump seems to be a case
// statement.
//TODO : We dont handle this case
if ( first_statement->getKind() == STMT_CASE &&
stmt_jump->getDest()->getOper() == opMemOf &&
stmt_jump->getDest()->getSubExp1()->getOper() == opIntConst &&
pBF->IsDynamicLinkedProcPointer(((Const*)stmt_jump->getDest()->getSubExp1())->
getAddr())) // Is it an "DynamicLinkedProcPointer"?
{
// Yes, it's a library function. Look up it's name.
ADDRESS a = ((Const*)stmt_jump->getDest()->getSubExp1())->getAddr();
// TODO : We dont handle this case
const char *nam = pBF->GetDynamicProcName(a);
// Assign the proc to the call
Proc *p = pProc->getProg()->getLibraryProc(nam);
if (call->getDestProc()) {
// prevent unnecessary __imp procs
prog->removeProc(call->getDestProc()->getName());
}
call->setDestProc(p);
call->setIsComputed(false);
call->setDest(Location::memOf(new Const(a)));
}
}
}
}
}
}
// Treat computed and static calls separately
if (call->isComputed()) {
BB_rtls->push_back(pRtl);
pBB = pCfg->newBB(BB_rtls, COMPCALL, 1);
// Stop decoding sequentially if the basic block already
// existed otherwise complete the basic block
if (pBB == 0)
sequentialDecode = false;
else
pCfg->addOutEdge(pBB, uAddr + inst.numBytes);
// Add this call to the list of calls to analyse. We won't
// be able to analyse it's callee(s), of course.
callList.push_back(call);
}
else { // Static call
// Find the address of the callee.
ADDRESS uNewAddr = call->getFixedDest();
// Calls with 0 offset (i.e. call the next instruction) are simply pushing the PC to the
// stack. Treat these as non-control flow instructions and continue.
if (uNewAddr == uAddr + inst.numBytes)
break;
// Call the virtual helper function. If implemented, will check for machine specific funcion
// calls
if (helperFunc(uNewAddr, uAddr, BB_rtls)) {
// We have already added to BB_rtls
pRtl = NULL; // Discard the call semantics
break;
}
BB_rtls->push_back(pRtl);
// Add this non computed call site to the set of call sites which need to be analysed later.
//pCfg->addCall(call);
callList.push_back(call);
// Record the called address as the start of a new procedure if it didn't already exist.
if (uNewAddr && uNewAddr != NO_ADDRESS && pProc->getProg()->findProc(uNewAddr) == NULL) {
callList.push_back(call);
//newProc(pProc->getProg(), uNewAddr);
if (Boomerang::get()->traceDecoder)
LOG << "p" << uNewAddr << "\t";
}
// Check if this is the _exit or exit function. May prevent us from attempting to decode
// invalid instructions, and getting invalid stack height errors
const char* name;
if (!ASS_FILE){
name = pBF->SymbolByAddress(uNewAddr);
if (name == NULL && call->getDest()->isMemOf() &&
call->getDest()->getSubExp1()->isIntConst()) {
ADDRESS a = ((Const*)call->getDest()->getSubExp1())->getInt();
if (pBF->IsDynamicLinkedProcPointer(a))
name = pBF->GetDynamicProcName(a);
}
}
else {
name = namesList.find(uNewAddr)->second;
}
if (name && noReturnCallDest(name)) {
// Make sure it has a return appended (so there is only one exit from the function)
//call->setReturnAfterCall(true); // I think only the Sparc frontend cares
// Create the new basic block
pBB = pCfg->newBB(BB_rtls, CALL, 1);
appendSyntheticReturn(pBB, pProc, pRtl);
// Stop decoding sequentially
sequentialDecode = false;
}
else {
// Create the new basic block
pBB = pCfg->newBB(BB_rtls, CALL, 1);
if (call->isReturnAfterCall()) {
// Constuct the RTLs for the new basic block
std::list<RTL*>* rtls = new std::list<RTL*>();
// The only RTL in the basic block is one with a ReturnStatement
std::list<Statement*>* sl = new std::list<Statement*>;
sl->push_back(new ReturnStatement());
rtls->push_back(new RTL(pRtl->getAddress()+1, sl));
BasicBlock* returnBB = pCfg->newBB(rtls, RET, 0);
// Add out edge from call to return
pCfg->addOutEdge(pBB, returnBB);
// Put a label on the return BB (since it's an orphan); a jump will be reqd
pCfg->setLabel(returnBB);
pBB->setJumpReqd();
// Mike: do we need to set return locations?
// This ends the function
sequentialDecode = false;
}
else {
// Add the fall through edge if the block didn't
// already exist
if (pBB != NULL)
pCfg->addOutEdge(pBB, uAddr+inst.numBytes);
}
}
}
extraProcessCall(call, BB_rtls);
// Create the list of RTLs for the next basic block and continue with the next instruction.
BB_rtls = NULL;
break;
}
case STMT_RET: {
// Stop decoding sequentially
sequentialDecode = false;
pBB = createReturnBlock(pProc, BB_rtls, pRtl);
// Create the list of RTLs for the next basic block and
// continue with the next instruction.
BB_rtls = NULL; // New RTLList for next BB
}
break;
case STMT_BOOLASSIGN:
// This is just an ordinary instruction; no control transfer
// Fall through
case STMT_JUNCTION:
// FIXME: Do we need to do anything here?
case STMT_ASSIGN:
case STMT_PHIASSIGN:
case STMT_IMPASSIGN:
case STMT_IMPREF:
// Do nothing
break;
} // switch (s->getKind())
}
if (BB_rtls && pRtl)
// If non null, we haven't put this RTL into a the current BB as yet
BB_rtls->push_back(pRtl);
if (inst.reDecode)
// Special case: redecode the last instruction, without advancing uAddr by numBytes
continue;
uAddr += inst.numBytes;
if (uAddr > lastAddr)
lastAddr = uAddr;
// If sequentially decoding, check if the next address happens to be the start of an existing BB. If so,
// finish off the current BB (if any RTLs) as a fallthrough, and no need to decode again (unless it's an
// incomplete BB, then we do decode it).
// In fact, mustn't decode twice, because it will muck up the coverage, but also will cause subtle problems
// like add a call to the list of calls to be processed, then delete the call RTL (e.g. Pentium 134.perl
// benchmark)
if (sequentialDecode && pCfg->existsBB(uAddr)) {
// Create the fallthrough BB, if there are any RTLs at all
if (BB_rtls) {
PBB pBB = pCfg->newBB(BB_rtls, FALL, 1);
// Add an out edge to this address
if (pBB) {
pCfg->addOutEdge(pBB, uAddr);
BB_rtls = NULL; // Need new list of RTLs
}
}
// Pick a new address to decode from, if the BB is complete
if (!pCfg->isIncomplete(uAddr))
sequentialDecode = false;
}
if(AssProgram)
++ li ;
} // while sequentialDecode
// Add this range to the coverage
// pProc->addRange(start, uAddr);
// Must set sequentialDecode back to true
sequentialDecode = true;
} // while nextAddress() != NO_ADDRESS
//ProgWatcher *w = prog->getWatcher();
//if (w)
// w->alert_done(pProc, initAddr, lastAddr, nTotalBytes);
// Add the callees to the set of CallStatements, and also to the Prog object
std::list<CallStatement*>::iterator it;
for (it = callList.begin(); it != callList.end(); it++) {
ADDRESS dest = (*it)->getFixedDest();
// Don't speculatively decode procs that are outside of the main text section, apart from dynamically
// linked ones (in the .plt)
// TODO: change pBF pointers
if (!ASS_FILE){
if (pBF->IsDynamicLinkedProc(dest) || !spec || (dest < pBF->getLimitTextHigh())) {
pCfg->addCall(*it);
// Don't visit the destination of a register call
Proc *np = (*it)->getDestProc();
if (np == NULL && dest != NO_ADDRESS) {
//np = newProc(pProc->getProg(), dest);
np = pProc->getProg()->setNewProc(dest);
}
if (np != NULL) {
np->setFirstCaller(pProc);
pProc->addCallee(np);
}
}
}
else{
pCfg->addCall(*it);
// Don't visit the destination of a register call
Proc *np = (*it)->getDestProc();
if (np == NULL && dest != NO_ADDRESS) {
//np = newProc(pProc->getProg(), dest);
np = pProc->getProg()->setNewProc(dest);
}
if (np != NULL) {
np->setFirstCaller(pProc);
pProc->addCallee(np);
}
}
}
Boomerang::get()->alert_decode(pProc, startAddr, lastAddr, nTotalBytes);
std::cout<< "finished processing proc " << pProc->getName() << " at address " << pProc->getNativeAddress() << "\n";
if (VERBOSE)
LOG << "finished processing proc " << pProc->getName() << " at address " << pProc->getNativeAddress() << "\n";
return true;
}
/*==============================================================================
* FUNCTION: FrontEnd::getInst
* OVERVIEW: Fetch the smallest (nop-sized) instruction, in an endianness independent manner
* NOTE: Frequently overridden
* PARAMETERS: addr - host address to getch from
* RETURNS: An integer with the instruction in it
*============================================================================*/
int FrontEnd::getInst(int addr)
{
return (int)(*(unsigned char*)addr);
}
/*==============================================================================
* FUNCTION: TargetQueue::visit
* OVERVIEW: Visit a destination as a label, i.e. check whether we need to queue it as a new BB to create later.
* Note: at present, it is important to visit an address BEFORE an out edge is added to that address.
* This is because adding an out edge enters the address into the Cfg's BB map, and it looks like the
* BB has already been visited, and it gets overlooked. It would be better to have a scheme whereby
* the order of calling these functions (i.e. visit() and AddOutEdge()) did not matter.
* PARAMETERS: pCfg - the enclosing CFG
* uNewAddr - the address to be checked
* pNewBB - set to the lower part of the BB if the address
* already exists as a non explicit label (BB has to be split)
* RETURNS: <nothing>
*============================================================================*/
void TargetQueue::visit(Cfg* pCfg, ADDRESS uNewAddr, PBB& pNewBB) {
// Find out if we've already parsed the destination
bool bParsed = pCfg->label(uNewAddr, pNewBB);
// Add this address to the back of the local queue,
// if not already processed
if (!bParsed) {
targets.push(uNewAddr);
if (Boomerang::get()->traceDecoder)
LOG << ">" << uNewAddr << "\t";
}
}
ProcStatus ProcDecompiler::tryDecompileRecursive(UserProc *proc)
{
/* Cycle detection logic:
* *********************
* cycleGrp is an initially null pointer to a set of procedures, representing the procedures
* involved in the current recursion group, if any. These procedures have to be analysed
* together as a group, after individual pre-group analysis. child is a set of procedures,
* cleared at the top of decompile(), representing the cycles associated with the current
* procedure and all of its children. If this is empty, the current procedure is not involved in
* recursion, and can be decompiled up to and including removing unused statements. callStack is
* an initially empty list of procedures, representing the call stack from the current entry
* point to the current procedure, inclusive. If (after all children have been processed:
* important!) the first element in callStack and also cycleGrp is the current procedure, we
* have the maximal set of distinct cycles, so we can do the recursion group analysis and return
* an empty set. At the end of the recursion group analysis, the whole group is complete, ready
* for the global analyses.
*
* cycleSet decompile(ProcList callStack) // call stack initially empty
* child = new ProcSet
* push this proc to the call stack
* for each child c called by this proc
* if c has already been visited but not finished
* // have new cycle
* if c is in callStack
* // this is a completely new cycle
* insert every proc from c to the end of callStack into child
* else
* // this is a new branch of an existing cycle
* child = c->cycleGrp
* find first element f of callStack that is in cycleGrp
* insert every proc after f to the end of callStack into child
* for each element e of child
* insert e->cycleGrp into child
* e->cycleGrp = child
* else
* // no new cycle
* tmp = c->decompile(callStack)
* child = union(child, tmp)
* set return statement in call to that of c
*
* if (child empty)
* earlyDecompile()
* child = middleDecompile()
* removeUnusedStatments() // Not involved in recursion
* else
* // Is involved in recursion
* find first element f in callStack that is also in cycleGrp
* if (f == this) // The big test: have we got the complete strongly connected component?
* recursionGroupAnalysis() // Yes, we have
* child = new ProcSet // Don't add these processed cycles to the parent
* remove last element (= this) from callStack
* return child
*/
Project *project = proc->getProg()->getProject();
LOG_MSG("%1 procedure '%2'", (proc->getStatus() >= PROC_VISITED) ? "Re-visiting" : "Visiting",
proc->getName());
project->alertDiscovered(proc);
// Prevent infinite loops when there are cycles in the call graph (should never happen now)
if (proc->getStatus() >= PROC_FINAL) {
LOG_WARN("Proc %1 already has status PROC_FINAL", proc->getName());
return PROC_FINAL; // Already decompiled
}
if (proc->getStatus() < PROC_DECODED) {
// Can happen e.g. if a callee is visible only after analysing a switch statement
// Actually decoding for the first time, not REdecoding
if (!proc->getProg()->reDecode(proc)) {
return PROC_UNDECODED;
}
}
if (proc->getStatus() < PROC_VISITED) {
proc->setStatus(PROC_VISITED); // We have at least visited this proc "on the way down"
}
m_callStack.push_back(proc);
if (project->getSettings()->verboseOutput) {
printCallStack();
}
if (project->getSettings()->decodeChildren) {
// Recurse to callees first, to perform a depth first search
for (BasicBlock *bb : *proc->getCFG()) {
if (bb->getType() != BBType::Call) {
continue;
}
// The call Statement will be in the last RTL in this BB
CallStatement *call = static_cast<CallStatement *>(bb->getRTLs()->back()->getHlStmt());
if (!call->isCall()) {
LOG_WARN("BB at address %1 is a CALL but last stmt is not a call: %2",
bb->getLowAddr(), call);
continue;
}
assert(call->isCall());
UserProc *callee = dynamic_cast<UserProc *>(call->getDestProc());
if (callee == nullptr) { // not an user proc, or missing dest
continue;
}
if (callee->getStatus() == PROC_FINAL) {
// Already decompiled, but the return statement still needs to be set for this call
call->setCalleeReturn(callee->getRetStmt());
continue;
}
// check if the callee has already been visited but not done (apart from global
// analyses). This means that we have found a new cycle or a part of an existing cycle
if ((callee->getStatus() >= PROC_VISITED) && (callee->getStatus() <= PROC_EARLYDONE)) {
// if callee is in callStack
ProcList::iterator calleeIt = std::find(m_callStack.begin(), m_callStack.end(),
callee);
if (calleeIt != m_callStack.end()) {
// This is a completely new cycle
std::shared_ptr<ProcSet> newRecursionGroup(new ProcSet());
newRecursionGroup->insert(calleeIt, m_callStack.end());
createRecursionGoup(newRecursionGroup);
}
else if (callee->getRecursionGroup()) {
// This is a new branch of an existing cycle that was visited previously
std::shared_ptr<ProcSet> recursionGroup = callee->getRecursionGroup();
// Find first element func of callStack that is in callee->recursionGroup
ProcList::iterator _pi = std::find_if(
m_callStack.begin(), m_callStack.end(), [callee](UserProc *func) {
return callee->getRecursionGroup()->find(func) !=
callee->getRecursionGroup()->end();
});
// Insert every proc after func to the end of path into child
assert(_pi != m_callStack.end());
for (auto it = std::next(_pi); it != m_callStack.end(); ++it) {
addToRecursionGroup(*it, recursionGroup);
}
}
proc->setStatus(PROC_INCYCLE);
}
else {
// No new cycle
LOG_VERBOSE("Preparing to decompile callee '%1' of '%2'", callee->getName(),
proc->getName());
callee->promoteSignature();
tryDecompileRecursive(callee);
// Child has at least done middleDecompile(), possibly more
call->setCalleeReturn(callee->getRetStmt());
if (proc->getStatus() != PROC_INCYCLE &&
m_recursionGroups.find(proc) != m_recursionGroups.end()) {
proc->setStatus(PROC_INCYCLE);
proc->setRecursionGroup(m_recursionGroups.find(proc)->second);
}
}
}
}
// if no child involved in recursion
if (proc->getStatus() != PROC_INCYCLE) {
project->alertDecompiling(proc);
LOG_MSG("Decompiling procedure '%1'", proc->getName());
earlyDecompile(proc);
middleDecompile(proc);
if (project->getSettings()->verboseOutput) {
printCallStack();
}
}
if (proc->getStatus() != PROC_INCYCLE) {
lateDecompile(proc); // Do the whole works
proc->setStatus(PROC_FINAL);
project->alertEndDecompile(proc);
}
else if (m_recursionGroups.find(proc) != m_recursionGroups.end()) {
// This proc's callees, and hence this proc, is/are involved in recursion.
// Find first element f in path that is also in our recursion group
ProcList::iterator f = std::find_if(
m_callStack.begin(), m_callStack.end(), [proc](UserProc *func) {
return proc->getRecursionGroup()->find(func) != proc->getRecursionGroup()->end();
});
// The big test: have we found the whole strongly connected component (in the call graph)?
if (*f == proc) {
// Yes, process these procs as a group
recursionGroupAnalysis(proc->getRecursionGroup());
proc->setStatus(PROC_FINAL);
project->alertEndDecompile(proc);
}
}
// Remove last element (= this) from path
assert(!m_callStack.empty());
assert(m_callStack.back() == proc);
m_callStack.pop_back();
LOG_MSG("Finished decompile of '%1'", proc->getName());
if (project->getSettings()->verboseOutput) {
printCallStack();
}
return proc->getStatus();
}