cppcheck/addons/y2038/y2038.py

283 lines
8.1 KiB
Python
Raw Normal View History

2017-06-04 22:51:48 +02:00
#!/usr/bin/env python
#
# cppcheck addon for Y2038 safeness detection
#
# Detects:
#
# 1. _TIME_BITS being defined to something else than 64 bits
# 2. _USE_TIME_BITS64 being defined when _TIME_BITS is not
# 3. Any Y2038-unsafe symbol when _USE_TIME_BITS64 is not defined.
#
# Example usage:
# $ cppcheck --dump path-to-src/
# $ y2038.py path-to-src/
#
# y2038.py will walk the source tree for .dump files.
import cppcheckdata
import sys
import os
import re
2017-06-04 22:51:48 +02:00
# --------------
# Error handling
2017-06-04 22:51:48 +02:00
# --------------
diagnostics = {}
def reportDiagnostic(template, configuration, file, line, severity, message):
# collect diagnostics by configuration
2017-06-04 22:51:48 +02:00
if configuration not in diagnostics:
diagnostics[configuration] = []
# add error to this configuration
diagnostics[configuration].append(
cppcheckdata.reportError(template, [[file, line]], severity, message))
def printDiagnostics():
for cfg in diagnostics:
sys.stderr.write('# Configuration "' + cfg + '":\n')
for diag in diagnostics[cfg]:
sys.stderr.write(diag + '\n')
def reportDirDiag(template, cfg, filename, linenr, directive, severity, msg):
reportDiagnostic(template, cfg.name,
directive.file, directive.linenr,
severity, msg)
if (filename != directive.file) or (linenr != directive.linenr):
reportDiagnostic(template, cfg.name,
filename, linenr, 'information',
directive.file + ' was included from here')
def reportTokDiag(template, cfg, token, severity, msg):
reportDiagnostic(template, cfg.name,
token.file, token.linenr,
severity, msg)
2017-06-04 22:51:48 +02:00
# --------------------------------------------
# #define/#undef detection regular expressions
2017-06-04 22:51:48 +02:00
# --------------------------------------------
# test for '#define _TIME_BITS 64'
re_define_time_bits_64 = re.compile(r'^\s*#\s*define\s+_TIME_BITS\s+64\s*$')
# test for '#define _TIME_BITS ...' (combine w/ above to test for 'not 64')
re_define_time_bits = re.compile(r'^\s*#\s*define\s+_TIME_BITS\s+.*$')
# test for '#undef _TIME_BITS' (if it ever happens)
re_undef_time_bits = re.compile(r'^\s*#\s*undef\s+_TIME_BITS\s*$')
# test for '#define _USE_TIME_BITS64'
re_define_use_time_bits64 = re.compile(r'^\s*#\s*define\s+_USE_TIME_BITS64\s*$')
# test for '#undef _USE_TIME_BITS64' (if it ever happens)
re_undef_use_time_bits64 = re.compile(r'^\s*#\s*undef\s+_USE_TIME_BITS64\s*$')
2017-06-04 22:51:48 +02:00
# --------------------------------
# List of Y2038-unsafe identifiers
2017-06-04 22:51:48 +02:00
# --------------------------------
# This is WIP. Eventually it should contain all identifiers (types
# and functions) which would be affected by the Y2038 bug.
2017-06-04 22:51:48 +02:00
id_Y2038 = {
# Y2038-unsafe types by definition
'time_t'
# Types using Y2038-unsafe types
'lastlog',
'msqid_ds',
'semid_ds',
'timeb',
'timespec',
'timeval',
'utimbuf',
'itimerspec',
'stat',
'clnt_ops',
'elf_prstatus',
'itimerval',
'ntptimeval',
'rusage',
'timex',
'utmp',
'utmpx',
# APIs using 2038-unsafe types
'ctime',
'ctime_r',
'difftime',
'gmtime',
'gmtime_r',
'localtime',
'localtime_r',
'mktime',
'stime',
'timegm',
'timelocal',
'time',
'msgctl',
'ftime',
'aio_suspend',
'clock_getres',
'clock_gettime',
'clock_nanosleep',
'clock_settime',
'futimens',
'mq_timedreceive',
'mq_timedsend',
'nanosleep',
'pselect',
'pthread_cond_timedwait',
'pthread_mutex_timedlock',
'pthread_rwlock_timedrdlock',
'pthread_rwlock_timedwrlock',
'sched_rr_get_interval',
'sem_timedwait',
'sigtimedwait',
'timespec_get',
'utimensat',
'adjtime',
'pmap_rmtcall',
'clntudp_bufcreate',
'clntudp_create',
'futimes',
'gettimeofday',
'lutimes',
'select',
'settimeofday',
'utimes',
'utime',
'timerfd_gettime',
'timerfd_settime',
'timer_gettime',
'timer_settime',
'fstatat',
'fstat',
'__fxstatat',
'__fxstat',
'lstat',
'__lxstat',
'stat',
'__xstat',
'struct itimerval',
'setitimer',
'getitimer',
'ntp_gettime',
'getrusage',
'wait3',
'wait4',
'adjtimex',
'ntp_adjtime',
'getutent_r',
'getutent',
'getutid_r',
'getutid',
'getutline_r',
'getutline',
'login',
'pututline',
'updwtmp',
'getutxent',
'getutxid',
'getutxline',
'pututxline'
2017-06-04 22:51:48 +02:00
}
2017-06-04 22:51:48 +02:00
# return all files ending in .dump among or under the given paths
def find_dump_files(paths):
dumpfiles = []
for path in paths:
if path.endswith('.dump'):
2017-06-04 22:51:48 +02:00
if path not in dumpfiles:
dumpfiles.append(path)
else:
for (top, subdirs, files) in os.walk(path):
for file in files:
if file.endswith('.dump'):
2016-07-23 13:44:05 +02:00
f = top + '/' + file
2017-06-04 22:51:48 +02:00
if f not in dumpfiles:
dumpfiles.append(f)
dumpfiles.sort()
return dumpfiles
2017-06-04 22:51:48 +02:00
# -----------------
# Let's get to work
2017-06-04 22:51:48 +02:00
# -----------------
# extend cppcheck parser with our own options
parser = cppcheckdata.ArgumentParser()
parser.add_argument('-q', '--quiet', action='store_true',
2016-07-23 13:44:05 +02:00
help='do not print "Checking ..." lines')
parser.add_argument('paths', nargs='+', metavar='path',
2016-07-23 13:44:05 +02:00
help='path to dump file or directory')
2017-06-04 22:51:48 +02:00
# parse command line
args = parser.parse_args()
# now operate on each file in turn
dumpfiles = find_dump_files(args.paths)
for dumpfile in dumpfiles:
if not args.quiet:
print('Checking ' + dumpfile + '...')
srcfile = dumpfile.rstrip('.dump')
# at the start of the check, we don't know if code is Y2038 safe
y2038safe = False
# load XML from .dump file
data = cppcheckdata.parsedump(dumpfile)
# go through each configuration
for cfg in data.configurations:
if not args.quiet:
print('Checking ' + dumpfile + ', config "' + cfg.name + '"...')
safe_ranges = []
safe = -1
time_bits_defined = False
2017-04-19 06:46:53 +02:00
srclinenr = '0'
for directive in cfg.directives:
# track source line number
if directive.file == srcfile:
srclinenr = directive.linenr
# check for correct _TIME_BITS if present
if re_define_time_bits_64.match(directive.str):
time_bits_defined = True
elif re_define_time_bits.match(directive.str):
reportDirDiag(args.template, cfg, srcfile, srclinenr,
directive, 'error',
'_TIME_BITS must be defined equal to 64')
time_bits_defined = False
elif re_undef_time_bits.match(directive.str):
time_bits_defined = False
# check for _USE_TIME_BITS64 (un)definition
if re_define_use_time_bits64.match(directive.str):
safe = int(srclinenr)
# warn about _TIME_BITS not being defined
2017-06-04 22:51:48 +02:00
if not time_bits_defined:
2016-07-23 13:44:05 +02:00
reportDirDiag(args.template,
2016-07-25 18:03:44 +02:00
cfg, srcfile, srclinenr, directive, 'warning',
'_USE_TIME_BITS64 is defined but _TIME_BITS was not')
elif re_undef_use_time_bits64.match(directive.str):
unsafe = int(srclinenr)
# do we have a safe..unsafe area?
if safe > 0 and unsafe > safe:
safe_ranges.append((safe, unsafe))
safe = -1
# check end of source beyond last directive
if len(cfg.tokenlist) > 0:
unsafe = int(cfg.tokenlist[-1].linenr)
if safe > 0 and unsafe > safe:
safe_ranges.append((safe, unsafe))
# go through all tokens
for token in cfg.tokenlist:
if token.str in id_Y2038:
if not any(lower <= int(token.linenr) <= upper
for (lower, upper) in safe_ranges):
reportTokDiag(args.template, cfg, token, 'warning',
token.str + ' is Y2038-unsafe')
token = token.next
printDiagnostics()