tools/check_static.py

   1 #!/usr/bin/env python3
   2 # Wireshark - Network traffic analyzer
   3 # By Gerald Combs <gerald@wireshark.org>
   4 # Copyright 1998 Gerald Combs
   5 #
   6 # SPDX-License-Identifier: GPL-2.0-or-later
   7
   8 import os
   9 import re
  10 import subprocess
  11 import argparse
  12 import signal
  13
  14 # Look for dissector symbols that could/should be static.
  15 # This will not run on Windows, unless/until we check the platform
  16 # and use (I think) dumpbin.exe
  17 #
  18 # N.B. Will report false positives if symbols are extern'd rather than
  19 # declared in a header file.
  20
  21 # Try to exit soon after Ctrl-C is pressed.
  22 should_exit = False
  23
  24 def signal_handler(sig, frame):
  25     global should_exit
  26     should_exit = True
  27     print('You pressed Ctrl+C - exiting')
  28
  29 signal.signal(signal.SIGINT, signal_handler)
  30
  31 # Allow this as a default build folder name...
  32 build_folder = os.getcwd() + '-build'
  33
  34
  35 # Record which symbols are referred to (by a set of files).
  36 class CalledSymbols:
  37     def __init__(self):
  38         self.referred = set()
  39
  40     def addCalls(self, file):
  41         if should_exit:
  42             exit(1)
  43
  44         # Make sure that file is built.
  45         last_dir = os.path.split(os.path.dirname(file))[-1]
  46         if file.find('ui/cli') != -1:
  47             # A tshark target-only file
  48             object_file = os.path.join(build_folder, 'CMakeFiles', ('tshark' + '.dir'), file + '.o')
  49         elif file.find('ui/qt') != -1:
  50             object_file = os.path.join(build_folder, os.path.dirname(file), 'CMakeFiles', ('qtui' + '.dir'), os.path.basename(file) + '.o')
  51         else:
  52             if file.endswith('dissectors.c'):
  53                 object_file = os.path.join(build_folder, os.path.dirname(file), 'CMakeFiles', 'dissector-registration' + '.dir', os.path.basename(file) + '.o')
  54             else:
  55                 object_file = os.path.join(build_folder, os.path.dirname(file), 'CMakeFiles', last_dir + '.dir', os.path.basename(file) + '.o')
  56         if not os.path.exists(object_file):
  57             # Not built for whatever reason..
  58             #print('Warning -', object_file, 'does not exist')
  59             return
  60
  61         # Run command to check symbols.
  62         command = ['nm', object_file]
  63         for f in subprocess.check_output(command).splitlines():
  64             line = str(f)[2:-1]
  65             # Lines might, or might not, have an address before letter and symbol.
  66             p1 = re.compile(r'[0-9a-f]* ([a-zA-Z]) (.*)')
  67             p2 = re.compile(r'[ ]* ([a-zA-Z]) (.*)')
  68
  69             m = p1.match(line)
  70             if not m:
  71                 m = p2.match(line)
  72             if m:
  73                 letter = m.group(1)
  74                 function_name = m.group(2)
  75
  76                 # Only interested in undefined/external references to symbols.
  77                 if letter == 'U':
  78                     self.referred.add(function_name)
  79
  80
  81
  82 # Record which symbols are defined in a single dissector file.
  83 class DefinedSymbols:
  84     def __init__(self, file):
  85         self.filename = file
  86         self.global_symbols = {}       # map from defined symbol -> whole output-line
  87         self.header_file_contents = None
  88         self.from_generated_file = isGeneratedFile(file)
  89
  90         # Make sure that file is built.
  91         if self.filename.startswith('epan'):
  92             object_file = os.path.join(build_folder, 'epan', 'dissectors', 'CMakeFiles', 'dissectors.dir', os.path.basename(file) + '.o')
  93         elif self.filename.startswith('plugins'):
  94             plugin_base_dir = os.path.dirname(file)
  95             plugin_base_name = os.path.basename(plugin_base_dir)
  96             object_file = os.path.join(build_folder, plugin_base_dir, 'CMakeFiles', plugin_base_name + '.dir', os.path.basename(file) + '.o')
  97         else:
  98             #print("Warning - can't determine object file for ", self.filename)
  99             return
 100         if not os.path.exists(object_file):
 101             #print('Warning -', object_file, 'does not exist')
 102             return
 103
 104         # Get header file contents if available
 105         header_file= file.replace('.c', '.h')
 106         try:
 107             f = open(header_file, 'r')
 108             self.header_file_contents = f.read()
 109         except IOError:
 110             pass
 111
 112         # Run command to see which symbols are defined
 113         command = ['nm', object_file]
 114         for f in subprocess.check_output(command).splitlines():
 115             # Line consists of whitespace, [address], letter, symbolName
 116             line = str(f)[2:-1]
 117             p = re.compile(r'[0-9a-f]* ([a-zA-Z]) (.*)')
 118             m = p.match(line)
 119             if m:
 120                 letter = m.group(1)
 121                 function_name = m.group(2)
 122                 # Globally-defined symbols. Would be 't' or 'd' if already static..
 123                 if letter in 'TD':
 124                     self.addDefinedSymbol(function_name, line)
 125
 126     def addDefinedSymbol(self, symbol, line):
 127         self.global_symbols[symbol] = line
 128
 129     # Check if a given symbol is mentioned in headers
 130     def mentionedInHeaders(self, symbol):
 131         if self.header_file_contents:
 132              if self.header_file_contents.find(symbol) != -1:
 133                 return True
 134         # Also check some of the 'common' header files that don't match the dissector file name.
 135         # TODO: could cache the contents of these files?
 136         common_mismatched_headers = [ os.path.join('epan', 'dissectors', 'packet-ncp-int.h'),
 137                                       os.path.join('epan', 'dissectors', 'packet-mq.h'),
 138                                       os.path.join('epan', 'dissectors', 'packet-ip.h'),
 139                                       os.path.join('epan', 'dissectors', 'packet-gsm_a_common.h'),
 140                                       os.path.join('epan', 'dissectors', 'packet-epl.h'),
 141                                       os.path.join('epan', 'dissectors', 'packet-bluetooth.h'),
 142                                       os.path.join('epan', 'dissectors', 'packet-dcerpc.h'),
 143                                       os.path.join('epan', 'ip_opts.h'),
 144                                       os.path.join('epan', 'eap.h')]
 145         for hf in common_mismatched_headers:
 146             try:
 147                 f = open(hf)
 148                 contents = f.read()
 149                 if contents.find(symbol) != -1:
 150                     return True
 151             except EnvironmentError:
 152                 pass
 153
 154         return False
 155
 156     def checkIfSymbolsAreCalled(self, called_symbols):
 157         global issues_found
 158         for f in self.global_symbols:
 159             if f not in called_symbols:
 160                 mentioned_in_header = self.mentionedInHeaders(f)
 161                 fun = self.global_symbols[f]
 162                 print(self.filename, '' if not self.from_generated_file else '(GENERATED)',
 163                       '(' + fun + ')',
 164                       'is not referred to so could be static?', '(declared in header)' if mentioned_in_header else '')
 165                 issues_found += 1
 166
 167
 168
 169 # Helper functions.
 170
 171 def isDissectorFile(filename):
 172     # Ignoring usb.c & errno.c
 173     p = re.compile(r'(packet|file)-.*\.c')
 174     return p.match(filename)
 175
 176 # Test for whether the given dissector file was automatically generated.
 177 def isGeneratedFile(filename):
 178     # Check file exists - e.g. may have been deleted in a recent commit.
 179     if not os.path.exists(filename):
 180         return False
 181
 182     if not filename.endswith('.c'):
 183         return False
 184
 185     # Open file
 186     f_read = open(os.path.join(filename), 'r')
 187     lines_tested = 0
 188     for line in f_read:
 189         # The comment to say that its generated is near the top, so give up once
 190         # get a few lines down.
 191         if lines_tested > 10:
 192             f_read.close()
 193             return False
 194         if (line.find('Generated automatically') != -1 or
 195             line.find('Autogenerated from') != -1 or
 196             line.find('is autogenerated') != -1 or
 197             line.find('automatically generated by Pidl') != -1 or
 198             line.find('Created by: The Qt Meta Object Compiler') != -1 or
 199             line.find('This file was generated') != -1 or
 200             line.find('This filter was automatically generated') != -1):
 201
 202             f_read.close()
 203             return True
 204         lines_tested = lines_tested + 1
 205
 206     # OK, looks like a hand-written file!
 207     f_read.close()
 208     return False
 209
 210
 211 def findDissectorFilesInFolder(folder, include_generated):
 212     # Look at files in sorted order, to give some idea of how far through is.
 213     tmp_files = []
 214
 215     for f in sorted(os.listdir(folder)):
 216         if should_exit:
 217             return
 218         if isDissectorFile(f):
 219             if include_generated or not isGeneratedFile(os.path.join('epan', 'dissectors', f)):
 220                 filename = os.path.join(folder, f)
 221                 tmp_files.append(filename)
 222     return tmp_files
 223
 224 def findFilesInFolder(folder):
 225     # Look at files in sorted order, to give some idea of how far through is.
 226     tmp_files = []
 227
 228     for f in sorted(os.listdir(folder)):
 229         if should_exit:
 230             return
 231         if f.endswith('.c') or f.endswith('.cpp'):
 232             filename = os.path.join(folder, f)
 233             tmp_files.append(filename)
 234     return tmp_files
 235
 236
 237 def is_dissector_file(filename):
 238     p = re.compile(r'.*(packet|file)-.*\.c')
 239     return p.match(filename)
 240
 241
 242
 243
 244 #################################################################
 245 # Main logic.
 246
 247 # command-line args.  Controls which dissector files should be checked.
 248 # If no args given, will just scan epan/dissectors folder.
 249 parser = argparse.ArgumentParser(description='Check calls in dissectors')
 250 parser.add_argument('--build-folder', action='store', default='',
 251                     help='build folder', required=False)
 252 parser.add_argument('--file', action='append',
 253                     help='specify individual dissector file to test')
 254 parser.add_argument('--commits', action='store',
 255                     help='last N commits to check')
 256 parser.add_argument('--open', action='store_true',
 257                     help='check open files')
 258
 259 args = parser.parse_args()
 260
 261 issues_found = 0
 262
 263 # Get files from wherever command-line args indicate.
 264 files = []
 265
 266 if args.build_folder:
 267     build_folder = args.build_folder
 268
 269 if args.file:
 270     # Add specified file(s)
 271     for f in args.file:
 272         if not os.path.isfile(f) and not f.startswith('epan'):
 273             f = os.path.join('epan', 'dissectors', f)
 274         if not os.path.isfile(f):
 275             print('Chosen file', f, 'does not exist.')
 276             exit(1)
 277         else:
 278             files.append(f)
 279 elif args.commits:
 280     # Get files affected by specified number of commits.
 281     command = ['git', 'diff', '--name-only', 'HEAD~' + args.commits]
 282     files = [f.decode('utf-8')
 283              for f in subprocess.check_output(command).splitlines()]
 284     # Will examine dissector files only
 285     files = list(filter(lambda f : is_dissector_file(f), files))
 286 elif args.open:
 287     # Unstaged changes.
 288     command = ['git', 'diff', '--name-only']
 289     files = [f.decode('utf-8')
 290              for f in subprocess.check_output(command).splitlines()]
 291     # Only interested in dissector files.
 292     files = list(filter(lambda f : is_dissector_file(f), files))
 293     # Staged changes.
 294     command = ['git', 'diff', '--staged', '--name-only']
 295     files_staged = [f.decode('utf-8')
 296                     for f in subprocess.check_output(command).splitlines()]
 297     # Only interested in dissector files.
 298     files_staged = list(filter(lambda f : is_dissector_file(f), files_staged))
 299     for f in files:
 300         files.append(f)
 301     for f in files_staged:
 302         if f not in files:
 303             files.append(f)
 304 else:
 305     # Find all dissector files from folder.
 306     files = findDissectorFilesInFolder(os.path.join('epan', 'dissectors'),
 307                                        include_generated=True)
 308
 309
 310 # If scanning a subset of files, list them here.
 311 print('Examining:')
 312 if args.file or args.commits or args.open:
 313     if files:
 314         print(' '.join(files), '\n')
 315     else:
 316         print('No files to check.\n')
 317 else:
 318     print('All dissector modules\n')
 319
 320
 321 if not os.path.isdir(build_folder):
 322     print('Build directory not valid', build_folder, '- please set with --build-folder')
 323     exit(1)
 324
 325
 326 # Get the set of called functions and referred-to data.
 327 called = CalledSymbols()
 328 for d in findDissectorFilesInFolder(os.path.join('epan', 'dissectors'), include_generated=True):
 329     called.addCalls(d)
 330 called.addCalls(os.path.join('epan', 'dissectors', 'dissectors.c'))
 331 # Also check calls from GUI code
 332 for d in findFilesInFolder('ui'):
 333     called.addCalls(d)
 334 for d in findFilesInFolder(os.path.join('ui', 'qt')):
 335     called.addCalls(d)
 336 # These are from tshark..
 337 for d in findFilesInFolder(os.path.join('ui', 'cli')):
 338     called.addCalls(d)
 339
 340
 341 # Now check identified dissector files.
 342 for f in files:
 343     if should_exit:
 344         exit(1)
 345     # Are these symbols called - or could they be deleted or static????
 346     DefinedSymbols(f).checkIfSymbolsAreCalled(called.referred)
 347
 348 # Show summary.
 349 print(issues_found, 'issues found')