mirrors/apparmor
1
0
Fork 0
mirror of https://gitlab.com/apparmor/apparmor.git synced 2025-03-04 16:35:02 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions
apparmor/utils/test/test-libapparmor-test_multi.py
Christian Boltz 962d4afb3d Fix aa-logprof crash on ptrace garbage log events 
(garbage) ptrace events like
    ... apparmor="DENIED" operation="ptrace" profile="/bin/netstat" pid=1962 comm="netstat" target=""
cause an empty name2 field, which leads to a crash in the tools.

This patch lets logparser.py ignore such garbage log events, which also
avoids the crash.

As usual, add some testcases.

test-libapparmor-test_multi.py needs some special handling to ignore the
empty name2 field in one of the testcases.


References: https://bugs.launchpad.net/apparmor/+bug/1689667


Acked-by: Seth Arnold <seth.arnold@canonical.com> for trunk and 2.11.

Older releases can't handle ptrace log events and therefore can't crash ;-)
2017-05-19 22:47:36 +02:00

275 lines
11 KiB
Python
Raw Blame History

#! /usr/bin/python3
# ------------------------------------------------------------------
#
# Copyright (C) 2015 Christian Boltz <apparmor@cboltz.de>
#
# This program is free software; you can redistribute it and/or
# modify it under the terms of version 2 of the GNU General Public
# License published by the Free Software Foundation.
#
# ------------------------------------------------------------------
import unittest
from common_test import AATest, setup_all_loops, setup_aa, read_file
import os
from apparmor.common import open_file_read
import apparmor.aa
from apparmor.logparser import ReadLog
class TestLibapparmorTestMulti(AATest):
'''Parse all libraries/libapparmor/testsuite/test_multi tests and compare the result with the *.out files'''
tests = 'invalid' # filled by parse_test_profiles()
def _run_test(self, params, expected):
# tests[][expected] is a dummy, replace it with the real values
expected = self._parse_libapparmor_test_multi(params)
with open_file_read('%s.in' % params) as f_in:
loglines = f_in.readlines()
loglines2 = []
for line in loglines:
if line.strip():
loglines2 += [line]
self.assertEqual(len(loglines2), 1, '%s.in should only contain one line!' % params)
parser = ReadLog('', '', '', '', '')
parsed_event = parser.parse_event(loglines2[0])
if parsed_event and expected:
parsed_items = dict(parsed_event.items())
# check if the line passes the regex in logparser.py
if not parser.RE_LOG_ALL.search(loglines2[0]):
raise Exception("Log event doesn't match RE_LOG_ALL")
for label in expected:
if label in [
'file', # filename of the *.in file
'event_type', # mapped to aamode
'audit_id', 'audit_sub_id', # not set nor relevant
'comm', # not set, and not too useful
# XXX most of the keywords listed below mean "TODO"
'fsuid', 'ouid', # file events
'flags', 'fs_type', # mount
'namespace', # file_lock only?? (at least the tests don't contain this in other event types with namespace)
'net_local_addr', 'net_foreign_addr', 'net_local_port', 'net_foreign_port', # detailed network events
'peer', 'signal', # signal
'src_name', # pivotroot
'dbus_bus', 'dbus_interface', 'dbus_member', 'dbus_path', # dbus
'peer_pid', 'peer_profile', # dbus
]:
pass
elif parsed_items['operation'] == 'exec' and label in ['sock_type', 'family', 'protocol']:
pass # XXX 'exec' + network? really?
elif parsed_items['operation'] == 'ptrace' and label == 'name2' and params.endswith('/ptrace_garbage_lp1689667_1'):
pass # libapparmor would better qualify this case as invalid event
elif not parsed_items.get(label, None):
raise Exception('parsed_items[%s] not set' % label)
elif not expected.get(label, None):
raise Exception('expected[%s] not set' % label)
else:
self.assertEqual(str(parsed_items[label]), expected[label], '%s differs' % label)
elif expected:
self.assertIsNone(parsed_event) # that's why we end up here
self.assertEqual(dict(), expected, 'parsed_event is none') # effectively print the content of expected
elif parsed_event:
self.assertIsNone(expected) # that's why we end up here
self.assertEqual(parsed_event, dict(), 'expected is none') # effectively print the content of parsed_event
else:
self.assertIsNone(expected) # that's why we end up here
self.assertIsNone(parsed_event) # that's why we end up here
self.assertEqual(parsed_event, expected) # both are None
# list of labels that use a different name in logparser.py than in the test_multi *.out files
# (additionally, .lower() is applied to all labels)
label_map = {
'Mask': 'request_mask',
'Command': 'comm',
'Token': 'magic_token',
'ErrorCode': 'error_code',
'Network family': 'family',
'Socket type': 'sock_type',
'Local addr': 'net_local_addr',
'Foreign addr': 'net_foreign_addr',
'Local port': 'net_local_port',
'Foreign port': 'net_foreign_port',
'Audit subid': 'audit_sub_id',
'Attribute': 'attr',
'Epoch': 'time',
}
def _parse_libapparmor_test_multi(self, file_with_path):
'''parse the libapparmor test_multi *.in tests and their expected result in *.out'''
with open_file_read('%s.out' % file_with_path) as f_in:
expected = f_in.readlines()
if expected[0].rstrip('\n') != 'START':
raise Exception("%s.out doesn't have 'START' in its first line! (%s)" % ( file_with_path, expected[0]))
expected.pop(0)
exresult = dict()
for line in expected:
label, value = line.split(':', 1)
# test_multi doesn't always use the original labels :-/
if label in self.label_map.keys():
label = self.label_map[label]
label = label.replace(' ', '_').lower()
exresult[label] = value.strip()
if not exresult['event_type'].startswith('AA_RECORD_'):
raise Exception("event_type doesn't start with AA_RECORD_: %s in file %s" % (exresult['event_type'], file_with_path))
exresult['aamode'] = exresult['event_type'].replace('AA_RECORD_', '')
if exresult['aamode'] == 'ALLOWED':
exresult['aamode'] = 'PERMITTING'
if exresult['aamode'] == 'DENIED':
exresult['aamode'] = 'REJECTING'
if exresult['event_type'] == 'AA_RECORD_INVALID': # or exresult.get('error_code', 0) != 0: # XXX should events with errors be ignored?
exresult = None
return exresult
# tests that do not produce the expected profile (checked with assertNotEqual)
log_to_profile_known_failures = [
'testcase_dmesg_changeprofile_01', # change_profile not yet supported in logparser
'testcase_changeprofile_01', # change_profile not yet supported in logparser
'testcase_mount_01', # mount rules not yet supported in logparser
'testcase_pivotroot_01', # pivot_rot not yet supported in logparser
# exec events
'testcase01',
'testcase12',
'testcase13',
# null-* hats get ignored by handle_children() if it didn't see an exec event for that null-* hat
'syslog_datetime_01',
'syslog_datetime_02',
'syslog_datetime_03',
'syslog_datetime_04',
'syslog_datetime_05',
'syslog_datetime_06',
'syslog_datetime_07',
'syslog_datetime_08',
'syslog_datetime_09',
'syslog_datetime_10',
'syslog_datetime_11',
'syslog_datetime_12',
'syslog_datetime_13',
'syslog_datetime_14',
'syslog_datetime_15',
'syslog_datetime_16',
'syslog_datetime_17',
'syslog_datetime_18',
'testcase_network_send_receive',
]
# tests that cause crashes or need user interaction (will be skipped)
log_to_profile_skip = [
'testcase31', # XXX AppArmorBug: Log contains unknown mode mrwIxl
'testcase_dmesg_changehat_negative_error', # fails in write_header -> quote_if_needed because data is None
'testcase_syslog_changehat_negative_error', # fails in write_header -> quote_if_needed because data is None
'testcase_changehat_01', # interactive, asks to add a hat
]
class TestLogToProfile(AATest):
'''Check if the libraries/libapparmor/testsuite/test_multi tests result in the expected profile'''
tests = 'invalid' # filled by parse_test_profiles()
def _run_test(self, params, expected):
logfile = '%s.in' % params
profile_dummy_file = 'AATest_does_exist'
# we need to find out the profile name and aamode (complain vs. enforce mode) so that the test can access the correct place in storage
parser = ReadLog('', '', '', '', '')
parsed_event = parser.parse_event(read_file(logfile))
if not parsed_event: # AA_RECORD_INVALID
return
if params.split('/')[-1] in log_to_profile_skip:
return
aamode = parsed_event['aamode']
if aamode in['AUDIT', 'STATUS', 'HINT']: # ignore some event types # XXX maybe we shouldn't ignore AUDIT events?
return
if aamode not in ['PERMITTING', 'REJECTING']:
raise Exception('Unexpected aamode %s' % parsed_event['aamode'])
# cleanup apparmor.aa storage
apparmor.aa.log = dict()
apparmor.aa.aa = apparmor.aa.hasher()
apparmor.aa.prelog = apparmor.aa.hasher()
profile = parsed_event['profile']
hat = profile
if '//' in profile:
profile, hat = profile.split('//')
apparmor.aa.existing_profiles = {profile: profile_dummy_file}
log_reader = ReadLog(dict(), logfile, apparmor.aa.existing_profiles, '', [])
log = log_reader.read_log('')
for root in log:
apparmor.aa.handle_children('', '', root) # interactive for exec events!
log_dict = apparmor.aa.collapse_log()
apparmor.aa.filelist = apparmor.aa.hasher()
apparmor.aa.filelist[profile_dummy_file]['profiles'][profile] = True
new_profile = apparmor.aa.serialize_profile(log_dict[aamode][profile], profile, None)
expected_profile = read_file('%s.profile' % params)
if params.split('/')[-1] in log_to_profile_known_failures:
self.assertNotEqual(new_profile, expected_profile) # known failure
else:
self.assertEqual(new_profile, expected_profile)
def find_test_multi(log_dir):
'''find all log sniplets in the given log_dir'''
log_dir = os.path.abspath(log_dir)
tests = []
for root, dirs, files in os.walk(log_dir):
for file in files:
if file.endswith('.in'):
file_with_path = os.path.join(root, file[:-3]) # filename without '.in'
tests.append([file_with_path, True]) # True is a dummy testresult, parsing of the *.out files is done while running the tests
elif file.endswith('.out') or file.endswith('.err') or file.endswith('.profile'):
pass
else:
raise Exception('Found unknown file %s in libapparmor test_multi' % file)
return tests
print('Testing libapparmor test_multi tests...')
TestLibapparmorTestMulti.tests = find_test_multi('../../libraries/libapparmor/testsuite/test_multi/')
TestLogToProfile.tests = find_test_multi('../../libraries/libapparmor/testsuite/test_multi/')
setup_aa(apparmor.aa)
setup_all_loops(__name__)
if __name__ == '__main__':
unittest.main(verbosity=1) # reduced verbosity due to the big number of tests
Reference in a new issue View git blame Copy permalink