tools/find_runtime_symbols/prepare_symbol_info.py

   1 #!/usr/bin/env python
   2 # Copyright (c) 2012 The Chromium Authors. All rights reserved.
   3 # Use of this source code is governed by a BSD-style license that can be
   4 # found in the LICENSE file.
   5
   6 import hashlib
   7 import json
   8 import logging
   9 import optparse
  10 import os
  11 import re
  12 import shutil
  13 import subprocess
  14 import sys
  15 import tempfile
  16
  17
  18 BASE_PATH = os.path.dirname(os.path.abspath(__file__))
  19 REDUCE_DEBUGLINE_PATH = os.path.join(BASE_PATH, 'reduce_debugline.py')
  20 _TOOLS_LINUX_PATH = os.path.join(BASE_PATH, os.pardir, 'linux')
  21 sys.path.insert(0, _TOOLS_LINUX_PATH)
  22
  23
  24 from procfs import ProcMaps  # pylint: disable=F0401
  25
  26
  27 LOGGER = logging.getLogger('prepare_symbol_info')
  28
  29
  30 def _dump_command_result(command, output_dir_path, basename, suffix):
  31   handle_out, filename_out = tempfile.mkstemp(
  32       suffix=suffix, prefix=basename + '.', dir=output_dir_path)
  33   handle_err, filename_err = tempfile.mkstemp(
  34       suffix=suffix + '.err', prefix=basename + '.', dir=output_dir_path)
  35   error = False
  36   try:
  37     subprocess.check_call(
  38         command, stdout=handle_out, stderr=handle_err, shell=True)
  39   except (OSError, subprocess.CalledProcessError):
  40     error = True
  41   finally:
  42     os.close(handle_err)
  43     os.close(handle_out)
  44
  45   if os.path.exists(filename_err):
  46     if LOGGER.getEffectiveLevel() <= logging.DEBUG:
  47       with open(filename_err, 'r') as f:
  48         for line in f:
  49           LOGGER.debug(line.rstrip())
  50     os.remove(filename_err)
  51
  52   if os.path.exists(filename_out) and (
  53       os.path.getsize(filename_out) == 0 or error):
  54     os.remove(filename_out)
  55     return None
  56
  57   if not os.path.exists(filename_out):
  58     return None
  59
  60   return filename_out
  61
  62
  63 def prepare_symbol_info(maps_path,
  64                         output_dir_path=None,
  65                         alternative_dirs=None,
  66                         use_tempdir=False,
  67                         use_source_file_name=False):
  68   """Prepares (collects) symbol information files for find_runtime_symbols.
  69
  70   1) If |output_dir_path| is specified, it tries collecting symbol information
  71   files in the given directory |output_dir_path|.
  72   1-a) If |output_dir_path| doesn't exist, create the directory and use it.
  73   1-b) If |output_dir_path| is an empty directory, use it.
  74   1-c) If |output_dir_path| is a directory which has 'files.json', assumes that
  75        files are already collected and just ignores it.
  76   1-d) Otherwise, depends on |use_tempdir|.
  77
  78   2) If |output_dir_path| is not specified, it tries to create a new directory
  79   depending on 'maps_path'.
  80
  81   If it cannot create a new directory, creates a temporary directory depending
  82   on |use_tempdir|.  If |use_tempdir| is False, returns None.
  83
  84   Args:
  85       maps_path: A path to a file which contains '/proc/<pid>/maps'.
  86       alternative_dirs: A mapping from a directory '/path/on/target' where the
  87           target process runs to a directory '/path/on/host' where the script
  88           reads the binary.  Considered to be used for Android binaries.
  89       output_dir_path: A path to a directory where files are prepared.
  90       use_tempdir: If True, it creates a temporary directory when it cannot
  91           create a new directory.
  92       use_source_file_name: If True, it adds reduced result of 'readelf -wL'
  93           to find source file names.
  94
  95   Returns:
  96       A pair of a path to the prepared directory and a boolean representing
  97       if it created a temporary directory or not.
  98   """
  99   alternative_dirs = alternative_dirs or {}
 100   if not output_dir_path:
 101     matched = re.match('^(.*)\.maps$', os.path.basename(maps_path))
 102     if matched:
 103       output_dir_path = matched.group(1) + '.pre'
 104   if not output_dir_path:
 105     matched = re.match('^/proc/(.*)/maps$', os.path.realpath(maps_path))
 106     if matched:
 107       output_dir_path = matched.group(1) + '.pre'
 108   if not output_dir_path:
 109     output_dir_path = os.path.basename(maps_path) + '.pre'
 110   # TODO(dmikurube): Find another candidate for output_dir_path.
 111
 112   used_tempdir = False
 113   LOGGER.info('Data for profiling will be collected in "%s".' % output_dir_path)
 114   if os.path.exists(output_dir_path):
 115     if os.path.isdir(output_dir_path) and not os.listdir(output_dir_path):
 116       LOGGER.warn('Using an empty existing directory "%s".' % output_dir_path)
 117     else:
 118       LOGGER.warn('A file or a directory exists at "%s".' % output_dir_path)
 119       if os.path.exists(os.path.join(output_dir_path, 'files.json')):
 120         LOGGER.warn('Using the existing directory "%s".' % output_dir_path)
 121         return output_dir_path, used_tempdir
 122       else:
 123         if use_tempdir:
 124           output_dir_path = tempfile.mkdtemp()
 125           used_tempdir = True
 126           LOGGER.warn('Using a temporary directory "%s".' % output_dir_path)
 127         else:
 128           LOGGER.warn('The directory "%s" is not available.' % output_dir_path)
 129           return None, used_tempdir
 130   else:
 131     LOGGER.info('Creating a new directory "%s".' % output_dir_path)
 132     try:
 133       os.mkdir(output_dir_path)
 134     except OSError:
 135       LOGGER.warn('A directory "%s" cannot be created.' % output_dir_path)
 136       if use_tempdir:
 137         output_dir_path = tempfile.mkdtemp()
 138         used_tempdir = True
 139         LOGGER.warn('Using a temporary directory "%s".' % output_dir_path)
 140       else:
 141         LOGGER.warn('The directory "%s" is not available.' % output_dir_path)
 142         return None, used_tempdir
 143
 144   shutil.copyfile(maps_path, os.path.join(output_dir_path, 'maps'))
 145
 146   with open(maps_path, mode='r') as f:
 147     maps = ProcMaps.load_file(f)
 148
 149   LOGGER.debug('Listing up symbols.')
 150   files = {}
 151   for entry in maps.iter(ProcMaps.executable):
 152     LOGGER.debug('  %016x-%016x +%06x %s' % (
 153         entry.begin, entry.end, entry.offset, entry.name))
 154     binary_path = entry.name
 155     for target_path, host_path in alternative_dirs.iteritems():
 156       if entry.name.startswith(target_path):
 157         binary_path = entry.name.replace(target_path, host_path, 1)
 158     if not (ProcMaps.EXECUTABLE_PATTERN.match(binary_path) or
 159             (os.path.isfile(binary_path) and os.access(binary_path, os.X_OK))):
 160       continue
 161     nm_filename = _dump_command_result(
 162         'nm -n --format bsd %s | c++filt' % binary_path,
 163         output_dir_path, os.path.basename(binary_path), '.nm')
 164     if not nm_filename:
 165       continue
 166     readelf_e_filename = _dump_command_result(
 167         'readelf -eW %s' % binary_path,
 168         output_dir_path, os.path.basename(binary_path), '.readelf-e')
 169     if not readelf_e_filename:
 170       continue
 171     readelf_debug_decodedline_file = None
 172     if use_source_file_name:
 173       readelf_debug_decodedline_file = _dump_command_result(
 174           'readelf -wL %s | %s' % (binary_path, REDUCE_DEBUGLINE_PATH),
 175           output_dir_path, os.path.basename(binary_path), '.readelf-wL')
 176
 177     files[entry.name] = {}
 178     files[entry.name]['nm'] = {
 179         'file': os.path.basename(nm_filename),
 180         'format': 'bsd',
 181         'mangled': False}
 182     files[entry.name]['readelf-e'] = {
 183         'file': os.path.basename(readelf_e_filename)}
 184     if readelf_debug_decodedline_file:
 185       files[entry.name]['readelf-debug-decodedline-file'] = {
 186           'file': os.path.basename(readelf_debug_decodedline_file)}
 187
 188     files[entry.name]['size'] = os.stat(binary_path).st_size
 189
 190     with open(binary_path, 'rb') as entry_f:
 191       md5 = hashlib.md5()
 192       sha1 = hashlib.sha1()
 193       chunk = entry_f.read(1024 * 1024)
 194       while chunk:
 195         md5.update(chunk)
 196         sha1.update(chunk)
 197         chunk = entry_f.read(1024 * 1024)
 198       files[entry.name]['sha1'] = sha1.hexdigest()
 199       files[entry.name]['md5'] = md5.hexdigest()
 200
 201   with open(os.path.join(output_dir_path, 'files.json'), 'w') as f:
 202     json.dump(files, f, indent=2, sort_keys=True)
 203
 204   LOGGER.info('Collected symbol information at "%s".' % output_dir_path)
 205   return output_dir_path, used_tempdir
 206
 207
 208 def main():
 209   if not sys.platform.startswith('linux'):
 210     sys.stderr.write('This script work only on Linux.')
 211     return 1
 212
 213   option_parser = optparse.OptionParser(
 214       '%s /path/to/maps [/path/to/output_data_dir/]' % sys.argv[0])
 215   option_parser.add_option('--alternative-dirs', dest='alternative_dirs',
 216                            metavar='/path/on/target@/path/on/host[:...]',
 217                            help='Read files in /path/on/host/ instead of '
 218                            'files in /path/on/target/.')
 219   option_parser.add_option('--verbose', dest='verbose', action='store_true',
 220                            help='Enable verbose mode.')
 221   options, args = option_parser.parse_args(sys.argv)
 222   alternative_dirs_dict = {}
 223   if options.alternative_dirs:
 224     for alternative_dir_pair in options.alternative_dirs.split(':'):
 225       target_path, host_path = alternative_dir_pair.split('@', 1)
 226       alternative_dirs_dict[target_path] = host_path
 227
 228   LOGGER.setLevel(logging.DEBUG)
 229   handler = logging.StreamHandler()
 230   if options.verbose:
 231     handler.setLevel(logging.DEBUG)
 232   else:
 233     handler.setLevel(logging.INFO)
 234   formatter = logging.Formatter('%(message)s')
 235   handler.setFormatter(formatter)
 236   LOGGER.addHandler(handler)
 237
 238   if len(args) < 2:
 239     option_parser.error('Argument error.')
 240     return 1
 241   elif len(args) == 2:
 242     result, _ = prepare_symbol_info(args[1],
 243                                     alternative_dirs=alternative_dirs_dict)
 244   else:
 245     result, _ = prepare_symbol_info(args[1], args[2],
 246                                     alternative_dirs=alternative_dirs_dict)
 247
 248   return not result
 249
 250
 251 if __name__ == '__main__':
 252   sys.exit(main())