def _find_leafs(self):
# Loop through every function
for func_ea in idautils.Functions():
# Count the number of xrefs to this function
func = idaapi.get_func(func_ea)
if func:
leaf_function = True
ea = ida_shims.start_ea(func)
end_ea = ida_shims.end_ea(func)
# Loop through all instructions in this function looking
# for call instructions; if found, then this is not a leaf.
while ea <= end_ea:
insn = ida_shims.decode_insn(ea)
if idaapi.is_call_insn(ea):
leaf_function = False
break
ea = ida_shims.next_head(ea)
if leaf_function:
self.functions.append(
Function(start=ida_shims.start_ea(func),
end=ida_shims.end_ea(func),
leaf=True,
loop=self.has_loop(func),
argc=self.argp.argc(func)))
# Sort leafs by xref count, largest first
self.functions.sort(key=lambda f: f.xrefs, reverse=True)
def _get_code_block(self, ea):
for block in self.blocks:
start_ea = ida_shims.start_ea(block)
end_ea = ida_shims.end_ea(block)
if start_ea <= ea and end_ea > ea:
return block
return None
def colorize_node(self, ea, color):
func = idaapi.get_func(ea)
if func:
for block in idaapi.FlowChart(func):
block_start_ea = ida_shims.start_ea(block)
block_end_ea = ida_shims.end_ea(block)
if block_start_ea <= ea and block_end_ea > ea:
ea = block_start_ea
while ea < block_end_ea:
idaapi.set_item_color(ea, color)
ea = ida_shims.next_head(ea)
break
def argc(self, function):
'''
Counts the number of arguments used by the specified function.
'''
argv = set()
notargv = set()
ea = ida_shims.start_ea(function)
end_ea = ida_shims.end_ea(function)
if self.arch.unknown:
return 0
while ea < end_ea:
insn = ida_shims.decode_insn(ea)
features = ida_shims.get_canon_feature(insn)
for n in range(0, len(self.USE_OPND)):
ops = ida_shims.get_operands(insn)
if ops[n].type in [
idaapi.o_reg, idaapi.o_displ, idaapi.o_phrase
]:
try:
regname = self.arch.registers[ops[n].reg]
index = self.arch.argv.index(regname)
except ValueError:
continue
if features & self.USE_OPND[n] and regname not in notargv:
argv.update(self.arch.argv[:index + 1])
for n in range(0, len(self.CHANGE_OPND)):
ops = ida_shims.get_operands(insn)
if ops[n].type in [
idaapi.o_reg, idaapi.o_displ, idaapi.o_phrase
]:
try:
regname = self.arch.registers[ops[n].reg]
index = self.arch.argv.index(regname)
except ValueError:
continue
if regname not in argv:
notargv.update(self.arch.argv[index:])
if argv.union(notargv) == set(self.arch.argv):
break
# TODO: Use idc.NextHead(ea) instead...
ea += self.arch.insn_size
return len(argv)
def has_loop(self, func):
'''
A naive method for checking to see if a function contains a loop.
Works pretty well for simple functions though.
'''
func_start_ea = ida_shims.start_ea(func)
blocks = [func_start_ea]
for block in idaapi.FlowChart(func):
end_ea = ida_shims.end_ea(block)
blocks.append(end_ea)
for block in blocks:
for xref in idautils.XrefsTo(block):
xref_func = idaapi.get_func(xref.frm)
xref_start_ea = ida_shims.start_ea(xref_func)
if xref_func and xref_start_ea == func_start_ea:
if xref.frm >= block:
return True
return False
def __init__(self, start_ea, end_ea, quiet=False):
end_func = idaapi.get_func(end_ea)
start_func = idaapi.get_func(start_ea)
if not start_func:
raise AlleyCatException("Address 0x%X is not part of a function!" %
start_ea)
if not end_func:
raise AlleyCatException("Address 0x%X is not part of a function!" %
end_ea)
start_func_ea = ida_shims.start_ea(start_func)
end_func_ea = ida_shims.end_ea(end_func)
if start_func_ea != end_func_ea:
raise AlleyCatException("The start and end addresses are not part "
"of the same function!")
self.func = start_func
self.blocks = [block for block in idaapi.FlowChart(self.func)]
end_block = self._get_code_block(start_ea)
start_block = self._get_code_block(end_ea)
if not end_block:
raise AlleyCatException("Failed to find the code block associated "
"with address 0x%X" % start_ea)
if not start_block:
raise AlleyCatException("Failed to find the code block associated "
"with address 0x%X" % end_ea)
start_block_ea = ida_shims.start_ea(start_block)
end_block_ea = ida_shims.start_ea(end_block)
super(AlleyCatCodePaths, self).__init__(
start_block_ea, end_block_ea, quiet)
def block(self, block):
'''
Returns a tuple:
([formal, block, signatures], [fuzzy, block, signatures],
set([unique, immediate, values]), [called, function, names])
'''
formal = []
fuzzy = []
functions = []
immediates = []
ea = ida_shims.start_ea(block)
while ea < ida_shims.end_ea(block):
insn = ida_shims.decode_insn(ea)
# Get a list of all data/code refs from the current instruction
drefs = [x for x in idautils.DataRefsFrom(ea)]
crefs = [x for x in idautils.CodeRefsFrom(ea, False)]
# Add all instruction mnemonics to the formal block hash
formal.append(ida_shims.print_insn_mnem(ea))
# If this is a call instruction, be sure to note the name of the
# function being called. This is used to apply call-based
# signatures to functions.
#
# For fuzzy signatures, we can't use the actual name or EA of the
# function, but rather just want to note that a function call was
# made.
#
# Formal signatures already have the call instruction mnemonic,
# which is more specific than just saying that a call was made.
if idaapi.is_call_insn(ea):
for cref in crefs:
func_name = ida_shims.get_name(cref)
if func_name:
functions.append(func_name)
fuzzy.append("funcref")
# If there are data references from the instruction, check to see
# if any of them are strings. These are looked up in the
# pre-generated strings dictionary.
#
# String values are easily identifiable, and are used as part of
# both the fuzzy and the formal signatures.
#
# It is more difficult to determine if non-string values are
# constants or not; for both fuzzy and formal signatures, just use
# "data" to indicate that some data was referenced.
elif drefs:
for dref in drefs:
if self.strings.has_key(dref):
formal.append(self.strings[dref].value)
fuzzy.append(self.strings[dref].value)
else:
formal.append("dataref")
fuzzy.append("dataref")
# If there are no data or code references from the instruction, use
# every operand as part of the formal signature.
#
# Fuzzy signatures are only concerned with interesting immediate
# values, that is, values that are greater than 65,535, are not
# memory addresses, and are not displayed as negative values.
elif not drefs and not crefs:
ops = ida_shims.get_operands(insn)
for n in range(0, len(ops)):
opnd_text = ida_shims.print_operand(ea, n)
formal.append(opnd_text)
if ops[n].type == idaapi.o_imm and \
not opnd_text.startswith('-'):
if ops[n].value >= 0xFFFF:
if ida_shims.get_full_flags(ops[n].value) == 0:
fuzzy.append(str(ops[n].value))
immediates.append(ops[n].value)
ea = ida_shims.next_head(ea)
return (self.sighash(''.join(formal)), self.sighash(''.join(fuzzy)),
immediates, functions)