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