1#!/usr/bin/env python2
2
3"""Find Kconfig symbols that are referenced but not defined."""
4
5# (c) 2014-2015 Valentin Rothberg <[email protected]>
6# (c) 2014 Stefan Hengelein <[email protected]>
7#
8# Licensed under the terms of the GNU GPL License version 2
9
10
11import os
12import re
13import sys
14from subprocess import Popen, PIPE, STDOUT
15from optparse import OptionParser
16
17
18# regex expressions
19OPERATORS = r"&|\(|\)|\||\!"
20FEATURE = r"(?:\w*[A-Z0-9]\w*){2,}"
21DEF = r"^\s*(?:menu){,1}config\s+(" + FEATURE + r")\s*"
22EXPR = r"(?:" + OPERATORS + r"|\s|" + FEATURE + r")+"
23STMT = r"^\s*(?:if|select|depends\s+on)\s+" + EXPR
24SOURCE_FEATURE = r"(?:\W|\b)+[D]{,1}CONFIG_(" + FEATURE + r")"
25
26# regex objects
27REGEX_FILE_KCONFIG = re.compile(r".*Kconfig[\.\w+\-]*$")
28REGEX_FEATURE = re.compile(r"(" + FEATURE + r")")
29REGEX_SOURCE_FEATURE = re.compile(SOURCE_FEATURE)
30REGEX_KCONFIG_DEF = re.compile(DEF)
31REGEX_KCONFIG_EXPR = re.compile(EXPR)
32REGEX_KCONFIG_STMT = re.compile(STMT)
33REGEX_KCONFIG_HELP = re.compile(r"^\s+(help|---help---)\s*$")
34REGEX_FILTER_FEATURES = re.compile(r"[A-Za-z0-9]$")
35
36
37def parse_options():
38    """The user interface of this module."""
39    usage = "%prog [options]\n\n"                                              \
40            "Run this tool to detect Kconfig symbols that are referenced but " \
41            "not defined in\nKconfig.  The output of this tool has the "       \
42            "format \'Undefined symbol\\tFile list\'\n\n"                      \
43            "If no option is specified, %prog will default to check your\n"    \
44            "current tree.  Please note that specifying commits will "         \
45            "\'git reset --hard\'\nyour current tree!  You may save "          \
46            "uncommitted changes to avoid losing data."
47
48    parser = OptionParser(usage=usage)
49
50    parser.add_option('-c', '--commit', dest='commit', action='store',
51                      default="",
52                      help="Check if the specified commit (hash) introduces "
53                           "undefined Kconfig symbols.")
54
55    parser.add_option('-d', '--diff', dest='diff', action='store',
56                      default="",
57                      help="Diff undefined symbols between two commits.  The "
58                           "input format bases on Git log's "
59                           "\'commmit1..commit2\'.")
60
61    parser.add_option('-f', '--find', dest='find', action='store_true',
62                      default=False,
63                      help="Find and show commits that may cause symbols to be "
64                           "missing.  Required to run with --diff.")
65
66    parser.add_option('-i', '--ignore', dest='ignore', action='store',
67                      default="",
68                      help="Ignore files matching this pattern.  Note that "
69                           "the pattern needs to be a Python regex.  To "
70                           "ignore defconfigs, specify -i '.*defconfig'.")
71
72    parser.add_option('', '--force', dest='force', action='store_true',
73                      default=False,
74                      help="Reset current Git tree even when it's dirty.")
75
76    (opts, _) = parser.parse_args()
77
78    if opts.commit and opts.diff:
79        sys.exit("Please specify only one option at once.")
80
81    if opts.diff and not re.match(r"^[\w\-\.]+\.\.[\w\-\.]+$", opts.diff):
82        sys.exit("Please specify valid input in the following format: "
83                 "\'commmit1..commit2\'")
84
85    if opts.commit or opts.diff:
86        if not opts.force and tree_is_dirty():
87            sys.exit("The current Git tree is dirty (see 'git status').  "
88                     "Running this script may\ndelete important data since it "
89                     "calls 'git reset --hard' for some performance\nreasons. "
90                     " Please run this script in a clean Git tree or pass "
91                     "'--force' if you\nwant to ignore this warning and "
92                     "continue.")
93
94    if opts.commit:
95        opts.find = False
96
97    if opts.ignore:
98        try:
99            re.match(opts.ignore, "this/is/just/a/test.c")
100        except:
101            sys.exit("Please specify a valid Python regex.")
102
103    return opts
104
105
106def main():
107    """Main function of this module."""
108    opts = parse_options()
109
110    if opts.commit or opts.diff:
111        head = get_head()
112
113        # get commit range
114        commit_a = None
115        commit_b = None
116        if opts.commit:
117            commit_a = opts.commit + "~"
118            commit_b = opts.commit
119        elif opts.diff:
120            split = opts.diff.split("..")
121            commit_a = split[0]
122            commit_b = split[1]
123            undefined_a = {}
124            undefined_b = {}
125
126        # get undefined items before the commit
127        execute("git reset --hard %s" % commit_a)
128        undefined_a = check_symbols(opts.ignore)
129
130        # get undefined items for the commit
131        execute("git reset --hard %s" % commit_b)
132        undefined_b = check_symbols(opts.ignore)
133
134        # report cases that are present for the commit but not before
135        for feature in sorted(undefined_b):
136            # feature has not been undefined before
137            if not feature in undefined_a:
138                files = sorted(undefined_b.get(feature))
139                print "%s\t%s" % (feature, ", ".join(files))
140                if opts.find:
141                    commits = find_commits(feature, opts.diff)
142                    print commits
143            # check if there are new files that reference the undefined feature
144            else:
145                files = sorted(undefined_b.get(feature) -
146                               undefined_a.get(feature))
147                if files:
148                    print "%s\t%s" % (feature, ", ".join(files))
149                    if opts.find:
150                        commits = find_commits(feature, opts.diff)
151                        print commits
152
153        # reset to head
154        execute("git reset --hard %s" % head)
155
156    # default to check the entire tree
157    else:
158        undefined = check_symbols(opts.ignore)
159        for feature in sorted(undefined):
160            files = sorted(undefined.get(feature))
161            print "%s\t%s" % (feature, ", ".join(files))
162
163
164def execute(cmd):
165    """Execute %cmd and return stdout.  Exit in case of error."""
166    pop = Popen(cmd, stdout=PIPE, stderr=STDOUT, shell=True)
167    (stdout, _) = pop.communicate()  # wait until finished
168    if pop.returncode != 0:
169        sys.exit(stdout)
170    return stdout
171
172
173def find_commits(symbol, diff):
174    """Find commits changing %symbol in the given range of %diff."""
175    commits = execute("git log --pretty=oneline --abbrev-commit -G %s %s"
176                      % (symbol, diff))
177    return commits
178
179
180def tree_is_dirty():
181    """Return true if the current working tree is dirty (i.e., if any file has
182    been added, deleted, modified, renamed or copied but not committed)."""
183    stdout = execute("git status --porcelain")
184    for line in stdout:
185        if re.findall(r"[URMADC]{1}", line[:2]):
186            return True
187    return False
188
189
190def get_head():
191    """Return commit hash of current HEAD."""
192    stdout = execute("git rev-parse HEAD")
193    return stdout.strip('\n')
194
195
196def check_symbols(ignore):
197    """Find undefined Kconfig symbols and return a dict with the symbol as key
198    and a list of referencing files as value.  Files matching %ignore are not
199    checked for undefined symbols."""
200    source_files = []
201    kconfig_files = []
202    defined_features = set()
203    referenced_features = dict()  # {feature: [files]}
204
205    # use 'git ls-files' to get the worklist
206    stdout = execute("git ls-files")
207    if len(stdout) > 0 and stdout[-1] == "\n":
208        stdout = stdout[:-1]
209
210    for gitfile in stdout.rsplit("\n"):
211        if ".git" in gitfile or "ChangeLog" in gitfile or      \
212                ".log" in gitfile or os.path.isdir(gitfile) or \
213                gitfile.startswith("tools/"):
214            continue
215        if REGEX_FILE_KCONFIG.match(gitfile):
216            kconfig_files.append(gitfile)
217        else:
218            # all non-Kconfig files are checked for consistency
219            source_files.append(gitfile)
220
221    for sfile in source_files:
222        if ignore and re.match(ignore, sfile):
223            # do not check files matching %ignore
224            continue
225        parse_source_file(sfile, referenced_features)
226
227    for kfile in kconfig_files:
228        if ignore and re.match(ignore, kfile):
229            # do not collect references for files matching %ignore
230            parse_kconfig_file(kfile, defined_features, dict())
231        else:
232            parse_kconfig_file(kfile, defined_features, referenced_features)
233
234    undefined = {}  # {feature: [files]}
235    for feature in sorted(referenced_features):
236        # filter some false positives
237        if feature == "FOO" or feature == "BAR" or \
238                feature == "FOO_BAR" or feature == "XXX":
239            continue
240        if feature not in defined_features:
241            if feature.endswith("_MODULE"):
242                # avoid false positives for kernel modules
243                if feature[:-len("_MODULE")] in defined_features:
244                    continue
245            undefined[feature] = referenced_features.get(feature)
246    return undefined
247
248
249def parse_source_file(sfile, referenced_features):
250    """Parse @sfile for referenced Kconfig features."""
251    lines = []
252    with open(sfile, "r") as stream:
253        lines = stream.readlines()
254
255    for line in lines:
256        if not "CONFIG_" in line:
257            continue
258        features = REGEX_SOURCE_FEATURE.findall(line)
259        for feature in features:
260            if not REGEX_FILTER_FEATURES.search(feature):
261                continue
262            sfiles = referenced_features.get(feature, set())
263            sfiles.add(sfile)
264            referenced_features[feature] = sfiles
265
266
267def get_features_in_line(line):
268    """Return mentioned Kconfig features in @line."""
269    return REGEX_FEATURE.findall(line)
270
271
272def parse_kconfig_file(kfile, defined_features, referenced_features):
273    """Parse @kfile and update feature definitions and references."""
274    lines = []
275    skip = False
276
277    with open(kfile, "r") as stream:
278        lines = stream.readlines()
279
280    for i in range(len(lines)):
281        line = lines[i]
282        line = line.strip('\n')
283        line = line.split("#")[0]  # ignore comments
284
285        if REGEX_KCONFIG_DEF.match(line):
286            feature_def = REGEX_KCONFIG_DEF.findall(line)
287            defined_features.add(feature_def[0])
288            skip = False
289        elif REGEX_KCONFIG_HELP.match(line):
290            skip = True
291        elif skip:
292            # ignore content of help messages
293            pass
294        elif REGEX_KCONFIG_STMT.match(line):
295            features = get_features_in_line(line)
296            # multi-line statements
297            while line.endswith("\\"):
298                i += 1
299                line = lines[i]
300                line = line.strip('\n')
301                features.extend(get_features_in_line(line))
302            for feature in set(features):
303                paths = referenced_features.get(feature, set())
304                paths.add(kfile)
305                referenced_features[feature] = paths
306
307
308if __name__ == "__main__":
309    main()
310