creduce-clang-crash.py source code [clang_source_code/utils/creduce-clang-crash.py]

1	#!/usr/bin/env python
2	"""Calls C-Reduce to create a minimal reproducer for clang crashes.
3	"""
4
5	from argparse import ArgumentParser
6	import os
7	import re
8	import stat
9	import sys
10	import subprocess
11	import pipes
12	import shlex
13	import tempfile
14	import shutil
15	from distutils.spawn import find_executable
16
17	verbose = False
18	llvm_bin = None
19	creduce_cmd = None
20	not_cmd = None
21
22	def check_file(fname):
23	if not os.path.isfile(fname):
24	sys.exit("ERROR: %s does not exist" % (fname))
25	return fname
26
27	def check_cmd(cmd_name, cmd_dir, cmd_path=None):
28	"""
29	Returns absolute path to cmd_path if it is given,
30	or absolute path to cmd_dir/cmd_name.
31	"""
32	if cmd_path:
33	cmd = find_executable(cmd_path)
34	if cmd:
35	return cmd
36	sys.exit("ERROR: executable %s not found" % (cmd_path))
37
38	cmd = find_executable(cmd_name, path=cmd_dir)
39	if cmd:
40	return cmd
41	sys.exit("ERROR: %s not found in %s" % (cmd_name, cmd_dir))
42
43	def quote_cmd(cmd):
44	return ' '.join(arg if arg.startswith('$') else pipes.quote(arg)
45	for arg in cmd)
46
47	def get_crash_cmd(crash_script):
48	with open(crash_script) as f:
49	# Assume clang call is on the last line of the script
50	line = f.readlines()[-1]
51	cmd = shlex.split(line)
52
53	# Overwrite the script's clang with the user's clang path
54	new_clang = check_cmd('clang', llvm_bin)
55	cmd[0] = pipes.quote(new_clang)
56	return cmd
57
58	def has_expected_output(crash_cmd, expected_output):
59	p = subprocess.Popen(crash_cmd,
60	stdout=subprocess.PIPE,
61	stderr=subprocess.STDOUT)
62	crash_output, _ = p.communicate()
63	return all(msg in crash_output for msg in expected_output)
64
65	def get_expected_output(crash_cmd):
66	p = subprocess.Popen(crash_cmd,
67	stdout=subprocess.PIPE,
68	stderr=subprocess.STDOUT)
69	crash_output, _ = p.communicate()
70
71	# If there is an assertion failure, use that;
72	# otherwise use the last five stack trace functions
73	assertion_re = r'Assertion `([^\']+)\' failed'
74	assertion_match = re.search(assertion_re, crash_output)
75	if assertion_match:
76	return [assertion_match.group(1)]
77	else:
78	stacktrace_re = r'#[0-9]+\s+0[xX][0-9a-fA-F]+\s*([^(]+)\('
79	matches = re.findall(stacktrace_re, crash_output)
80	return matches[-5:]
81
82	def write_interestingness_test(testfile, crash_cmd, expected_output,
83	file_to_reduce):
84	filename = os.path.basename(file_to_reduce)
85	if filename not in crash_cmd:
86	sys.exit("ERROR: expected %s to be in the crash command" % filename)
87
88	# Replace all instances of file_to_reduce with a command line variable
89	output = ['#!/bin/bash',
90	'if [ -z "$1" ] ; then',
91	' f=%s' % (pipes.quote(filename)),
92	'else',
93	' f="$1"',
94	'fi']
95	cmd = ['$f' if s == filename else s for s in crash_cmd]
96
97	output.append('%s --crash %s >& t.log \|\| exit 1' % (pipes.quote(not_cmd),
98	quote_cmd(cmd)))
99
100	for msg in expected_output:
101	output.append('grep %s t.log \|\| exit 1' % pipes.quote(msg))
102
103	with open(testfile, 'w') as f:
104	f.write('\n'.join(output))
105	os.chmod(testfile, os.stat(testfile).st_mode \| stat.S_IEXEC)
106
107	def check_interestingness(testfile, file_to_reduce):
108	testfile = os.path.abspath(testfile)
109
110	# Check that the test considers the original file interesting
111	with open(os.devnull, 'w') as devnull:
112	returncode = subprocess.call(testfile, stdout=devnull)
113	if returncode:
114	sys.exit("The interestingness test does not pass for the original file.")
115
116	# Check that an empty file is not interesting
117	_, empty_file = tempfile.mkstemp()
118	with open(os.devnull, 'w') as devnull:
119	returncode = subprocess.call([testfile, empty_file], stdout=devnull)
120	os.remove(empty_file)
121	if not returncode:
122	sys.exit("The interestingness test passes for an empty file.")
123
124	def clang_preprocess(file_to_reduce, crash_cmd, expected_output):
125	_, tmpfile = tempfile.mkstemp()
126	shutil.copy(file_to_reduce, tmpfile)
127
128	cmd = crash_cmd + ['-E', '-P']
129	p = subprocess.Popen(cmd,
130	stdout=subprocess.PIPE,
131	stderr=subprocess.STDOUT)
132	preprocessed, _ = p.communicate()
133
134	with open(file_to_reduce, 'w') as f:
135	f.write(preprocessed)
136
137	if has_expected_output(crash_cmd, expected_output):
138	if verbose:
139	print("Successfuly preprocessed with %s" % (quote_cmd(cmd)))
140	os.remove(tmpfile)
141	else:
142	if verbose:
143	print("Failed to preprocess with %s" % (quote_cmd(cmd)))
144	shutil.move(tmpfile, file_to_reduce)
145
146
147	def filter_args(args, opts_startswith=[]):
148	result = [arg for arg in args if all(not arg.startswith(a) for a in
149	opts_startswith)]
150	return result
151
152	def try_remove_args(cmd, expected_output, msg=None, extra_arg=None, **kwargs):
153	new_cmd = filter_args(cmd, **kwargs)
154	if extra_arg and extra_arg not in new_cmd:
155	new_cmd.append(extra_arg)
156	if new_cmd != cmd and has_expected_output(new_cmd, expected_output):
157	if msg and verbose:
158	print(msg)
159	return new_cmd
160	return cmd
161
162	def simplify_crash_cmd(crash_cmd, expected_output):
163	new_cmd = try_remove_args(crash_cmd, expected_output,
164	msg="Removed debug info options",
165	opts_startswith=["-gcodeview",
166	"-dwarf-column-info",
167	"-debug-info-kind=",
168	"-debugger-tuning=",
169	"-gdwarf"])
170	new_cmd = try_remove_args(new_cmd, expected_output,
171	msg="Replaced -W options with -w",
172	extra_arg='-w',
173	opts_startswith=["-W"])
174	new_cmd = try_remove_args(new_cmd, expected_output,
175	msg="Replaced optimization level with -O0",
176	extra_arg="-O0",
177	opts_startswith=["-O"])
178	return new_cmd
179
180	def main():
181	global verbose
182	global llvm_bin
183	global creduce_cmd
184	global not_cmd
185
186	parser = ArgumentParser(description=__doc__)
187	parser.add_argument('crash_script', type=str, nargs=1,
188	help="Name of the script that generates the crash.")
189	parser.add_argument('file_to_reduce', type=str, nargs=1,
190	help="Name of the file to be reduced.")
191	parser.add_argument('--llvm-bin', dest='llvm_bin', type=str,
192	required=True, help="Path to the LLVM bin directory.")
193	parser.add_argument('--llvm-not', dest='llvm_not', type=str,
194	help="The path to the `not` executable. "
195	"By default uses the llvm-bin directory.")
196	parser.add_argument('--creduce', dest='creduce', type=str,
197	help="The path to the `creduce` executable. "
198	"Required if `creduce` is not in PATH environment.")
199	parser.add_argument('-v', '--verbose', action='store_true')
200	args = parser.parse_args()
201
202	verbose = args.verbose
203	llvm_bin = os.path.abspath(args.llvm_bin)
204	creduce_cmd = check_cmd('creduce', None, args.creduce)
205	not_cmd = check_cmd('not', llvm_bin, args.llvm_not)
206	crash_script = check_file(args.crash_script[0])
207	file_to_reduce = check_file(args.file_to_reduce[0])
208
209	print("\nParsing the crash script and getting expected output...")
210	crash_cmd = get_crash_cmd(crash_script)
211
212	expected_output = get_expected_output(crash_cmd)
213	if len(expected_output) < 1:
214	sys.exit("ERROR: no crash was found")
215
216	print("\nSimplifying the crash command...")
217	crash_cmd = simplify_crash_cmd(crash_cmd, expected_output)
218
219	print("\nWriting interestingness test to file...")
220	testfile = os.path.splitext(file_to_reduce)[0] + '.test.sh'
221	write_interestingness_test(testfile, crash_cmd, expected_output,
222	file_to_reduce)
223	check_interestingness(testfile, file_to_reduce)
224
225	print("\nPreprocessing the file to reduce...")
226	clang_preprocess(file_to_reduce, crash_cmd, expected_output)
227
228	print("\nRunning C-Reduce...")
229	try:
230	p = subprocess.Popen([creduce_cmd, testfile, file_to_reduce])
231	p.communicate()
232	except KeyboardInterrupt:
233	# Hack to kill C-Reduce because it jumps into its own pgid
234	print('\n\nctrl-c detected, killed creduce')
235	p.kill()
236
237	# FIXME: reduce the clang crash command
238
239	if __name__ == '__main__':
240	main()
241

Clang Project