def fetch_contents(fn): """Fetch the reference count for the contents of function `fn` in the database. Returns the tuple `(func, address, tags)` where the `address` and `tags` fields are both dictionaries containing the reference count for the addresses and tag names. The field `func` contains the address of the function. """ addr, tags = {}, {} for ea in func.iterate(fn): ui.navigation.auto(ea) # grab both comment types so that we can decode them. after decoding, # then iterate through all of their keys and tally up their counts. repeatable, nonrepeatable = (db.comment(ea, repeatable=item) for item in [True, False]) for items in map(internal.comment.decode, [repeatable, nonrepeatable]): #items.pop('name', None) for name in items: addr[ea] = addr.get(ea, 0) + 1 tags[name] = tags.get(name, 0) + 1 continue continue return func.address(fn), addr, tags
def search_func(addr, funcname, curr_addr_list=None): """ Given an address and function name, recursively look from the given address for calls to the given function name. Modifies the 'calls' function tag with the function name and True or False depending if the function contains the wanted function or not. """ if curr_addr_list == None: curr_addr_list = [addr] print("Addr: {}".format(addr)) curr_calls = fn.tag(addr).get('calls', {}) for ea in fn.iterate(addr): if funcname in db.disasm(ea): curr_calls[funcname] = True fn.tag(addr, 'calls', curr_calls) return True if not ins.isCall(ea): continue call = ins.op_value(ea, 0) # Don't cycle ourselves if call == addr: print("Ignoring recursive loop on function: {}".format(hex(addr))) continue # Try to know if the call operand is a valid address # Bail if not.. try: print(hex(call)) except: continue # Check if this function has been analyzed already # and return the result # Cannot return False, because that will exit the loop and not # continue to iterate over the rest of the instrutions in the function call_cache = fn.tag(call).get('calls', {}) print("Call cache: {}".format(call_cache)) if funcname in call_cache: if call_cache[funcname]: curr_calls[funcname] = call_cache[funcname] fn.tag(addr, 'calls', curr_calls) return True # Don't call functions we are currently looking into if call in curr_addr_list: continue # Only now ca curr_addr_list.append(call) search_func(call, funcname=funcname, curr_addr_list=curr_addr_list) curr_calls[funcname] = False fn.tag(addr, 'calls', curr_calls) return False
def search_dyn_calls(addr, curr_addr_list=None, parents=None): dyn_call = namedtuple('dyn_call', ['call', 'parents']) hex = '{:x}'.format # print(hex(addr), curr_addr_list) if curr_addr_list == None: curr_addr_list = [addr] if parents == None: parents = [] calls = [] #print("Addr: {}".format(addr)) for ea in fn.iterate(addr): if not ins.isCall(ea): continue call = ins.op_value(ea, 0) # Don't cycle ourselves if call == addr: # print("Ignoring recursive loop on function: {}".format(hex(addr))) continue # Don't call functions we are currently looking into if call in curr_addr_list: continue """ .text:3F6E66AD 0B4 call ds:memmove Python>x = ins.op_value(h(), 0) Python>print(x.base) None .text:3F6E6208 008 call dword ptr [eax] Python>x.base <instruction.register.eax(0,dt_dword) 'eax' 0:+32> """ # we want non int/long for calls if isinstance(call, (int, long)) and not ins.op_type(ea, 0) in ('phrase'): # Only now call the function once curr_addr_list.append(call) new_parents = copy.deepcopy(parents) new_parents.append(ea) calls.extend( search_dyn_call(call, curr_addr_list=curr_addr_list, parents=new_parents)) elif isinstance( call, ins.intelop.OffsetBaseIndexScale) and call.base == None: # Ignore 'call ds:memmove' or 'call ds:atoi' # print("OffsetBase", call.offset, hex(ea)[2:-1], db.disasm(ea), parents) pass else: calls.append(dyn_call(ea, parents)) return calls
def content(cls, ea): '''Iterate through every tag belonging to the contents of the function at `ea`.''' F = func.by(ea) # iterate through every address in the function for ea in func.iterate(F): ui.navigation.set(ea) # yield the tags res = db.tag(ea) if res: yield ea, res return
def fetch_function(f): addr, tags = {}, {} for ea in fn.iterate(f): res = db.tag(ea) #res.pop('name', None) for k, v in res.iteritems(): addr[ea] = addr.get(ea, 0) + 1 tags[k] = tags.get(k, 0) + 1 continue ea = f return fn.top(ea), addr, tags
def fetch_contents(fn): """Fetch the reference count for the contents of function `fn` in the database. Returns the tuple `(func, address, tags)` where the `address` and `tags` fields are both dictionaries containing the reference count for the addresses and tag names. The field `func` contains the address of the function. """ addr, tags = {}, {} for ea in func.iterate(fn): ui.navigation.auto(ea) res = db.tag(ea) #res.pop('name', None) for k, v in six.iteritems(res): addr[ea] = addr.get(ea, 0) + 1 tags[k] = tags.get(k, 0) + 1 continue return func.address(fn), addr, tags
def _search_func(addr, func, curr_addr_list=None): if curr_addr_list == None: curr_addr_list = [addr] for ea in fn.iterate(addr): func(ea) if not ins.isCall(ea): continue call = ins.op_value(ea, 0) # Don't cycle ourselves if call == addr: # print("Ignoring recursive loop on function: {}".format(hex(addr))) continue # Try to know if the call operand is a valid address # Bail if not.. if not isinstance(call, (int, long)): continue """ try: x = hex(call) except: continue """ # Don't call functions we are currently looking into if call in curr_addr_list: continue # Only now ca curr_addr_list.append(call) _search_func(call, func, curr_addr_list=curr_addr_list)
def function(ea): '''Yield each tag defined within a function.''' for ea in fn.iterate(ea): t = db.tag(ea) if t: yield ea, t return
def search_malloc(addr, funcname='malloc', curr_addr_list=None): # print(hex(addr), curr_addr_list) if curr_addr_list == None: curr_addr_list = [addr] print("Addr: {}".format(addr)) curr_calls = fn.tag(addr).get('calls', {}) for ea in fn.iterate(addr): if not ins.isCall(ea): continue if ins.op_type(0) == 'reg': """ Handle this case - malloc(100) mov ebp, ds:malloc push 100 call ebp """ if funcname in db.disasm(ea): # Push before malloc "should" be within 20 instructions """ Handle this case - malloc(100) push 100 mov eax, 10 mov ebx, 20 call malloc """ search_addr = db.prev(ea) for _ in range(20): if ins.mnem(search_addr) == 'push': break search_addr = db.prev(search_addr) print("FOUND PUSH FOR MALLOC: {}".format(hex(search_addr))) malloc_value = ins.op_value(search_addr, 0) if isinstance(malloc_value, (int, long)): curr_calls[funcname] = malloc_value else: curr_calls[funcname] = 'variable' fn.tag(addr, 'calls', curr_calls) return True call = ins.op_value(ea, 0) # Don't cycle ourselves if call == addr: print("Ignoring recursive loop on function: {}".format(hex(addr))) continue # Try to know if the call operand is a valid address # Bail if not.. try: print(hex(call)) except: continue # Check if this function has been analyzed already # and return the result # Cannot return False, because that will exit the loop and not # continue to iterate over the rest of the instrutions in the function call_cache = fn.tag(call).get('calls', {}) print("Call cache: {}".format(call_cache)) if funcname in call_cache: if call_cache[funcname]: curr_calls[funcname] = call_cache[funcname] fn.tag(addr, 'calls', curr_calls) return True # Don't call functions we are currently looking into if call in curr_addr_list: continue # Only now ca curr_addr_list.append(call) search_malloc(call, funcname=funcname, curr_addr_list=curr_addr_list) curr_calls[funcname] = False fn.tag(addr, 'calls', curr_calls) return False