cppcheck/addons/cert.py

476 lines
17 KiB
Python
Raw Normal View History

#!/usr/bin/env python3
2015-08-18 16:14:53 +02:00
#
# Cert: Some extra CERT checkers
#
# Cppcheck itself handles many CERT rules. Cppcheck warns when there is undefined behaviour.
2018-03-16 08:12:39 +01:00
# CERT Homepage: https://www.cert.org/secure-coding/
2015-08-18 16:14:53 +02:00
#
# Example usage of this addon (scan a sourcefile main.cpp)
# cppcheck --dump main.cpp
# python cert.py main.cpp.dump
import argparse
import cppcheckdata
2015-08-18 16:14:53 +02:00
import sys
import re
import subprocess
2015-08-21 10:55:19 +02:00
2017-10-15 16:18:29 +02:00
VERIFY = ('-verify' in sys.argv)
VERIFY_EXPECTED = []
VERIFY_ACTUAL = []
2017-08-15 20:34:54 +02:00
def reportError(token, severity, msg, id):
2017-10-15 16:18:29 +02:00
if VERIFY:
2019-04-09 20:36:24 +02:00
VERIFY_ACTUAL.append(str(token.linenr) + ':cert-' + id)
2017-10-15 16:18:29 +02:00
else:
2019-04-14 08:54:53 +02:00
cppcheckdata.reportError(token, severity, msg, 'cert', id)
def simpleMatch(token, pattern):
2020-11-11 14:24:55 +01:00
return cppcheckdata.simpleMatch(token, pattern)
2015-08-21 10:55:19 +02:00
def isUnpackedStruct(token):
if token.valueType is None:
return False
if token.valueType.typeScope is None:
return False
if token.valueType.typeScope.type != "Struct":
return False
startToken = token.valueType.typeScope.bodyStart
linenr = int(startToken.linenr)
for line in open(startToken.file):
linenr -= 1
if linenr == 0:
return True
if linenr < 3 and re.match(r'#pragma\s+pack\s*\(', line):
return False
return True
def isLocalUnpackedStruct(arg):
if arg and arg.str == '&' and not arg.astOperand2:
arg = arg.astOperand1
return arg and arg.variable and (arg.variable.isLocal or arg.variable.isArgument) and isUnpackedStruct(arg)
2015-08-21 10:55:19 +02:00
2015-08-18 16:14:53 +02:00
def isBitwiseOp(token):
2017-06-04 22:51:48 +02:00
return token and (token.str in {'&', '|', '^'})
2015-08-18 16:14:53 +02:00
2015-08-21 10:55:19 +02:00
2015-08-18 16:14:53 +02:00
def isComparisonOp(token):
2017-06-04 22:51:48 +02:00
return token and (token.str in {'==', '!=', '>', '>=', '<', '<='})
2015-08-18 16:14:53 +02:00
def isCast(expr):
if not expr or expr.str != '(' or not expr.astOperand1 or expr.astOperand2:
return False
if simpleMatch(expr, '( )'):
return False
return True
def isStandardFunction(token):
if token.function:
return False
prev = token.previous
if prev:
if prev.str == '.':
return False
if prev.str == '::':
prevprev = prev.previous
if prevprev and not prevprev.str == 'std':
return False
return True
2019-06-23 13:59:24 +02:00
# Is this a function call
def isFunctionCall(token, function_names, number_of_arguments=None):
if not token.isName:
return False
if token.str not in function_names:
return False
if (token.next is None) or token.next.str != '(' or token.next != token.astParent:
return False
if number_of_arguments is None:
return True
return len(cppcheckdata.getArguments(token)) == number_of_arguments
# EXP05-C
# do not attempt to cast away const
def exp05(data):
# TODO Reuse code in misra rule 11.8
for token in data.tokenlist:
if isCast(token):
# C-style cast
if not token.valueType:
continue
if not token.astOperand1.valueType:
continue
if token.valueType.pointer == 0:
continue
if token.astOperand1.valueType.pointer == 0:
continue
const1 = token.valueType.constness
const2 = token.astOperand1.valueType.constness
if (const1 % 2) < (const2 % 2):
2019-04-09 20:36:24 +02:00
reportError(token, 'style', "Attempt to cast away const", 'EXP05-C')
elif token.str == '(' and token.astOperand1 and token.astOperand2 and token.astOperand1.function:
function = token.astOperand1.function
arguments = cppcheckdata.getArguments(token.previous)
if not arguments:
continue
for argnr, argvar in function.argument.items():
if argnr < 1 or argnr > len(arguments):
continue
if not argvar.isPointer:
continue
2018-04-12 20:23:50 +02:00
if (argvar.constness % 2) == 1: # data is const
continue
argtok = arguments[argnr - 1]
if not argtok.valueType:
continue
if argtok.valueType.pointer == 0:
continue
const2 = arguments[argnr - 1].valueType.constness
2018-04-12 20:23:50 +02:00
if (const2 % 2) == 1:
2019-04-09 20:36:24 +02:00
reportError(token, 'style', "Attempt to cast away const", 'EXP05-C')
# EXP42-C
# do not compare padding data
def exp42(data):
for token in data.tokenlist:
if token.str != '(' or not token.astOperand1 or token.astOperand1.str != 'memcmp':
continue
arg1 = None
arg2 = None
if token.astOperand2 and token.astOperand2.str == ',':
if token.astOperand2.astOperand1 and token.astOperand2.astOperand1.str == ',':
arg1 = token.astOperand2.astOperand1.astOperand1
arg2 = token.astOperand2.astOperand1.astOperand2
if isLocalUnpackedStruct(arg1) or isLocalUnpackedStruct(arg2):
reportError(
2017-08-15 20:34:54 +02:00
token, 'style', "Comparison of struct padding data " +
2019-04-09 20:36:24 +02:00
"(fix either by packing the struct using '#pragma pack' or by rewriting the comparison)", 'EXP42-C')
2017-06-04 22:51:48 +02:00
# EXP15-C
# Do not place a semicolon on the same line as an if, for or while statement
def exp15(data):
for scope in data.scopes:
if scope.type in ('If', 'For', 'While'):
token = scope.bodyStart.next
if token.str==';' and token.linenr==scope.bodyStart.linenr:
reportError(token, 'style', 'Do not place a semicolon on the same line as an IF, FOR or WHILE', 'EXP15-C')
2015-08-18 16:14:53 +02:00
# EXP46-C
# Do not use a bitwise operator with a Boolean-like operand
# int x = (a == b) & c;
def exp46(data):
for token in data.tokenlist:
if isBitwiseOp(token) and (isComparisonOp(token.astOperand1) or isComparisonOp(token.astOperand2)):
reportError(
2019-04-09 20:36:24 +02:00
token, 'style', 'Bitwise operator is used with a Boolean-like operand', 'EXP46-c')
2015-08-18 16:14:53 +02:00
# INT31-C
# Ensure that integer conversions do not result in lost or misinterpreted data
def int31(data, platform):
if not platform:
return
for token in data.tokenlist:
2022-01-20 21:09:39 +01:00
to_value_type = None
from_values = None
action = ''
if isCast(token):
to_value_type = token.valueType
from_values = token.astOperand1.values
action = 'casting'
elif token.str == '=' and token.astOperand1 and token.astOperand2:
to_value_type = token.astOperand1.valueType
from_values = token.astOperand2.values
action = 'assign'
else:
continue
2022-01-20 21:09:39 +01:00
if to_value_type is None or not from_values:
continue
bits = None
if token.valueType.pointer > 0:
bits = platform.pointer_bit
elif to_value_type.type == 'char':
bits = platform.char_bit
2022-01-20 21:09:39 +01:00
elif to_value_type.type == 'short':
bits = platform.short_bit
2022-01-20 21:09:39 +01:00
elif to_value_type.type == 'int':
bits = platform.int_bit
2022-01-20 21:09:39 +01:00
elif to_value_type.type == 'long':
bits = platform.long_bit
2022-01-20 21:09:39 +01:00
elif to_value_type.type == 'long long':
2017-10-16 13:35:07 +02:00
bits = platform.long_long_bit
else:
continue
2022-01-20 21:09:39 +01:00
if to_value_type.sign == 'unsigned':
2017-10-16 13:35:07 +02:00
found = False
2022-01-20 21:09:39 +01:00
for value in from_values:
if value.intvalue and value.intvalue < 0:
2017-10-16 13:35:07 +02:00
found = True
reportError(
token,
'style',
2022-01-20 21:09:39 +01:00
'Ensure that integer conversions do not result in lost or misinterpreted data (' + action + ' ' + str(value.intvalue) + ' to unsigned ' + token.valueType.type + ')',
2019-04-09 20:36:24 +02:00
'INT31-c')
2017-10-16 13:35:07 +02:00
break
if found:
continue
if bits >= 64:
continue
2017-10-16 13:35:07 +02:00
minval = 0
maxval = 1
if token.valueType.sign == 'signed':
minval = -(1 << (bits - 1))
maxval = ((1 << (bits - 1)) - 1)
else:
minval = 0
maxval = ((1 << bits) - 1)
2022-01-20 21:09:39 +01:00
for value in from_values:
2018-10-29 01:39:52 +01:00
if value.intvalue and (value.intvalue < minval or value.intvalue > maxval):
2017-10-16 13:35:07 +02:00
destType = ''
if token.valueType.sign:
destType = token.valueType.sign + ' ' + token.valueType.type
else:
destType = token.valueType.type
reportError(
token,
'style',
2022-01-20 21:09:39 +01:00
'Ensure that integer conversions do not result in lost or misinterpreted data (' + action + ' ' + str(value.intvalue) + ' to ' + destType + ')',
2019-04-09 20:36:24 +02:00
'INT31-c')
break
2019-11-21 06:46:46 +01:00
# ENV33-C
# Do not call system()
def env33(data):
for token in data.tokenlist:
if isFunctionCall(token, ('system',), 1):
# Invalid syntax
if not token.next.astOperand2:
continue
# ENV33-C-EX1: It is permissible to call system() with a null
# pointer argument to determine the presence of a command processor
# for the system.
argValue = token.next.astOperand2.getValue(0)
if argValue and argValue.intvalue == 0 and argValue.isKnown():
continue
2019-11-21 06:46:46 +01:00
reportError(token, 'style', 'Do not call system()', 'ENV33-C')
# MSC24-C
# Do not use deprecated or obsolescent functions
def msc24(data):
for token in data.tokenlist:
if isFunctionCall(token, ('asctime',), 1):
reportError(token,'style','Do not use asctime() better use asctime_s()', 'MSC24-C')
elif isFunctionCall(token, ('atof',), 1):
reportError(token,'style','Do not use atof() better use strtod()', 'MSC24-C')
elif isFunctionCall(token, ('atoi',), 1):
reportError(token,'style','Do not use atoi() better use strtol()', 'MSC24-C')
elif isFunctionCall(token, ('atol',), 1):
reportError(token,'style','Do not use atol() better use strtol()', 'MSC24-C')
elif isFunctionCall(token, ('atoll',), 1):
reportError(token,'style','Do not use atoll() better use strtoll()', 'MSC24-C')
elif isFunctionCall(token, ('ctime',), 1):
reportError(token,'style','Do not use ctime() better use ctime_s()', 'MSC24-C')
elif isFunctionCall(token, ('fopen',), 2):
reportError(token,'style','Do not use fopen() better use fopen_s()', 'MSC24-C')
elif isFunctionCall(token, ('freopen',), 3):
reportError(token,'style','Do not use freopen() better use freopen_s()', 'MSC24-C')
elif isFunctionCall(token, ('rewind',), 1):
reportError(token,'style','Do not use rewind() better use fseek()', 'MSC24-C')
elif isFunctionCall(token, ('setbuf',), 2):
reportError(token,'style','Do not use setbuf() better use setvbuf()', 'MSC24-C')
# MSC30-C
# Do not use the rand() function for generating pseudorandom numbers
def msc30(data):
for token in data.tokenlist:
if simpleMatch(token, "rand ( )") and isStandardFunction(token):
2019-04-09 20:36:24 +02:00
reportError(token, 'style', 'Do not use the rand() function for generating pseudorandom numbers', 'MSC30-c')
# STR03-C
# Do not inadvertently truncate a string
def str03(data):
for token in data.tokenlist:
if not isFunctionCall(token, 'strncpy'):
continue
arguments = cppcheckdata.getArguments(token)
if len(arguments)!=3:
continue
if arguments[2].str=='(' and arguments[2].astOperand1.str=='sizeof':
reportError(token, 'style', 'Do not inadvertently truncate a string', 'STR03-C')
# STR05-C
# Use pointers to const when referring to string literals
def str05(data):
for token in data.tokenlist:
if token.isString:
parent = token.astParent
if parent is None:
continue
parentOp1 = parent.astOperand1
if parent.isAssignmentOp and parentOp1.valueType:
if (parentOp1.valueType.type in ('char', 'wchar_t')) and parentOp1.valueType.pointer and not parentOp1.valueType.constness:
reportError(parentOp1, 'style', 'Use pointers to const when referring to string literals', 'STR05-C')
2019-06-23 13:59:24 +02:00
# STR07-C
# Use the bounds-checking interfaces for string manipulation
def str07(data):
if data.standards.c=='c89' or data.standards.c=='c99':
return
2019-06-23 13:59:24 +02:00
for token in data.tokenlist:
if not isFunctionCall(token, ('strcpy', 'strcat')):
continue
args = cppcheckdata.getArguments(token)
if len(args)!=2:
continue
if args[1].isString:
continue
reportError(token, 'style', 'Use the bounds-checking interfaces %s_s()' % token.str, 'STR07-C')
# STR11-C
# Do not specify the bound of a character array initialized with a string literal
def str11(data):
for token in data.tokenlist:
if not token.isString:
continue
strlen = token.strlen
parent = token.astParent
if parent is None:
continue
parentOp1 = parent.astOperand1
if parentOp1 is None or parentOp1.str!='[':
continue
if not parent.isAssignmentOp:
continue
varToken = parentOp1.astOperand1
if varToken is None or not varToken.isName:
continue
if varToken.variable is None:
continue
if varToken != varToken.variable.nameToken:
continue
valueToken = parentOp1.astOperand2
if valueToken is None:
continue
if valueToken.isNumber and int(valueToken.str)==strlen:
reportError(valueToken, 'style', 'Do not specify the bound of a character array initialized with a string literal', 'STR11-C')
# API01-C
# Avoid laying out strings in memory directly before sensitive data
def api01(data):
for scope in data.scopes:
if scope.type!='Struct':
continue
token = scope.bodyStart
string_found = False
# loop through the complete struct
while token != scope.bodyEnd:
if token.isName and token.variable:
is_string = False
if token.variable.isArray:
type_token = token.variable.typeStartToken
while type_token and type_token.isName:
if type_token.str in ('char', 'wchar_t') and not type_token.isExpandedMacro:
is_string = True
type_token = type_token.next
if is_string:
string_found = True
elif string_found and not token.variable.isConst:
reportError(token, 'style', 'Avoid laying out strings in memory directly before sensitive data', 'API01-C')
# reset flags to report other positions in the same struct
string_found = False
token = token.next
2017-10-15 16:18:29 +02:00
def get_args_parser():
parser = cppcheckdata.ArgumentParser()
parser.add_argument("-verify", help=argparse.SUPPRESS, action="store_true")
return parser
2017-10-15 16:18:29 +02:00
if __name__ == '__main__':
parser = get_args_parser()
args = parser.parse_args()
path_premium_addon = cppcheckdata.get_path_premium_addon()
if args.verify:
VERIFY = True
if not args.dumpfile:
if not args.quiet:
print("no input files.")
sys.exit(0)
for dumpfile in args.dumpfile:
if not args.quiet:
print('Checking %s...' % dumpfile)
if path_premium_addon:
premium_command = [path_premium_addon, '--cert', dumpfile]
if args.cli:
premium_command.append('--cli')
for line in subprocess.check_output(premium_command).decode('ascii').split('\n'):
if line.find('cert-') > 0:
print(line.strip())
data = cppcheckdata.CppcheckData(dumpfile)
if VERIFY:
VERIFY_ACTUAL = []
VERIFY_EXPECTED = []
for tok in data.rawTokens:
if tok.str.startswith('//') and 'TODO' not in tok.str:
for word in tok.str[2:].split(' '):
if re.match(r'cert-[A-Z][A-Z][A-Z][0-9][0-9].*',word):
VERIFY_EXPECTED.append(str(tok.linenr) + ':' + word)
for cfg in data.iterconfigurations():
if not args.quiet:
print('Checking %s, config %s...' % (dumpfile, cfg.name))
exp05(cfg)
exp42(cfg)
exp46(cfg)
exp15(cfg)
int31(cfg, data.platform)
str03(cfg)
str05(cfg)
str07(cfg)
str11(cfg)
2019-11-21 06:46:46 +01:00
env33(cfg)
msc24(cfg)
msc30(cfg)
api01(cfg)
if VERIFY:
fail = False
for expected in VERIFY_EXPECTED:
if expected not in VERIFY_ACTUAL:
print('Expected but not seen: ' + expected)
fail = True
for actual in VERIFY_ACTUAL:
if actual not in VERIFY_EXPECTED:
print('Not expected: ' + actual)
fail = True
if fail:
sys.exit(1)
sys.exit(cppcheckdata.EXIT_CODE)