tools/cygprofile/cyglog_to_orderfile.py

   1 #!/usr/bin/python
   2 # Copyright 2015 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 """Symbolizes a log file produced by cyprofile instrumentation.
   7
   8 Given a log file and the binary being profiled, creates an orderfile.
   9 """
  10
  11 import logging
  12 import multiprocessing
  13 import optparse
  14 import os
  15 import tempfile
  16 import string
  17 import sys
  18
  19 import cygprofile_utils
  20 import symbol_extractor
  21
  22
  23 def _ParseLogLines(log_file_lines):
  24   """Parses a merged cyglog produced by mergetraces.py.
  25
  26   Args:
  27     log_file_lines: array of lines in log file produced by profiled run
  28     lib_name: library or executable containing symbols
  29
  30     Below is an example of a small log file:
  31     5086e000-52e92000 r-xp 00000000 b3:02 51276      libchromeview.so
  32     secs       usecs      pid:threadid    func
  33     START
  34     1314897086 795828     3587:1074648168 0x509e105c
  35     1314897086 795874     3587:1074648168 0x509e0eb4
  36     1314897086 796326     3587:1074648168 0x509e0e3c
  37     1314897086 796552     3587:1074648168 0x509e07bc
  38     END
  39
  40   Returns:
  41     An ordered list of callee offsets.
  42   """
  43   call_lines = []
  44   vm_start = 0
  45   line = log_file_lines[0]
  46   assert 'r-xp' in line
  47   end_index = line.find('-')
  48   vm_start = int(line[:end_index], 16)
  49   for line in log_file_lines[3:]:
  50     fields = line.split()
  51     if len(fields) == 4:
  52       call_lines.append(fields)
  53     else:
  54       assert fields[0] == 'END'
  55   # Convert strings to int in fields.
  56   call_info = []
  57   for call_line in call_lines:
  58     addr = int(call_line[3], 16)
  59     if vm_start < addr:
  60       addr -= vm_start
  61       call_info.append(addr)
  62   return call_info
  63
  64
  65 def _GroupLibrarySymbolInfosByOffset(lib_filename):
  66   """Returns a dict {offset: [SymbolInfo]} from a library."""
  67   symbol_infos = symbol_extractor.SymbolInfosFromBinary(lib_filename)
  68   return symbol_extractor.GroupSymbolInfosByOffset(symbol_infos)
  69
  70
  71 class SymbolNotFoundException(Exception):
  72   def __init__(self, value):
  73     super(SymbolNotFoundException, self).__init__(value)
  74     self.value = value
  75
  76   def __str__(self):
  77     return repr(self.value)
  78
  79
  80 def _FindSymbolInfosAtOffset(offset_to_symbol_infos, offset):
  81   """Finds all SymbolInfo at a given offset.
  82
  83   Args:
  84     offset_to_symbol_infos: {offset: [SymbolInfo]}
  85     offset: offset to look the symbols at
  86
  87   Returns:
  88     The list of SymbolInfo at the given offset
  89
  90   Raises:
  91     SymbolNotFoundException if the offset doesn't match any symbol.
  92   """
  93   if offset in offset_to_symbol_infos:
  94     return offset_to_symbol_infos[offset]
  95   elif offset % 2 and (offset - 1) in offset_to_symbol_infos:
  96     # On ARM, odd addresses are used to signal thumb instruction. They are
  97     # generated by setting the LSB to 1 (see
  98     # http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.dui0471e/Babfjhia.html).
  99     # TODO(lizeb): Make sure this hack doesn't propagate to other archs.
 100     return offset_to_symbol_infos[offset - 1]
 101   else:
 102     raise SymbolNotFoundException(offset)
 103
 104
 105 def _GetObjectFileNames(obj_dir):
 106   """Returns the list of object files in a directory."""
 107   obj_files = []
 108   for (dirpath, _, filenames) in os.walk(obj_dir):
 109     for file_name in filenames:
 110       if file_name.endswith('.o'):
 111         obj_files.append(os.path.join(dirpath, file_name))
 112   return obj_files
 113
 114
 115 def _AllSymbolInfos(object_filenames):
 116   """Returns a list of SymbolInfo from an iterable of filenames."""
 117   pool = multiprocessing.Pool()
 118   # Hopefully the object files are in the page cache at this step, so IO should
 119   # not be a problem (hence no concurrency limit on the pool).
 120   symbol_infos_nested = pool.map(
 121       symbol_extractor.SymbolInfosFromBinary, object_filenames)
 122   result = []
 123   for symbol_infos in symbol_infos_nested:
 124     result += symbol_infos
 125   return result
 126
 127
 128 def _GetSymbolToSectionMapFromObjectFiles(obj_dir):
 129   """ Creates a mapping from symbol to linker section name by scanning all
 130       the object files.
 131   """
 132   object_files = _GetObjectFileNames(obj_dir)
 133   symbol_to_section_map = {}
 134   symbol_warnings = cygprofile_utils.WarningCollector(300)
 135   symbol_infos = _AllSymbolInfos(object_files)
 136   for symbol_info in symbol_infos:
 137     symbol = symbol_info.name
 138     if symbol.startswith('.LTHUNK'):
 139       continue
 140     section = symbol_info.section
 141     if ((symbol in symbol_to_section_map) and
 142         (symbol_to_section_map[symbol] != symbol_info.section)):
 143       symbol_warnings.Write('Symbol ' + symbol +
 144                             ' in conflicting sections ' + section +
 145                             ' and ' + symbol_to_section_map[symbol])
 146     elif not section.startswith('.text'):
 147       symbol_warnings.Write('Symbol ' + symbol +
 148                             ' in incorrect section ' + section)
 149     else:
 150       symbol_to_section_map[symbol] = section
 151   symbol_warnings.WriteEnd('bad sections')
 152   return symbol_to_section_map
 153
 154
 155 def _WarnAboutDuplicates(offsets):
 156   """Warns about duplicate offsets.
 157
 158   Args:
 159     offsets: list of offsets to check for duplicates
 160
 161   Returns:
 162     True if there are no duplicates, False otherwise.
 163   """
 164   seen_offsets = set()
 165   ok = True
 166   for offset in offsets:
 167     if offset not in seen_offsets:
 168       seen_offsets.add(offset)
 169     else:
 170       ok = False
 171       logging.warning('Duplicate offset: ' + hex(offset))
 172   return ok
 173
 174
 175 def _OutputOrderfile(offsets, offset_to_symbol_infos, symbol_to_section_map,
 176                      output_file):
 177   """Outputs the orderfile to output_file.
 178
 179   Args:
 180     offsets: Iterable of offsets to match to section names
 181     offset_to_symbol_infos: {offset: [SymbolInfo]}
 182     symbol_to_section_map: {name: section}
 183     output_file: file-like object to write the results to
 184   """
 185   success = True
 186   unknown_symbol_warnings = cygprofile_utils.WarningCollector(300)
 187   symbol_not_found_warnings = cygprofile_utils.WarningCollector(300)
 188   output_sections = set()
 189   for offset in offsets:
 190     try:
 191       symbol_infos = _FindSymbolInfosAtOffset(offset_to_symbol_infos, offset)
 192       for symbol_info in symbol_infos:
 193         if symbol_info.name in symbol_to_section_map:
 194           section = symbol_to_section_map[symbol_info.name]
 195           if not section in output_sections:
 196             output_file.write(section + '\n')
 197             output_sections.add(section)
 198         else:
 199           unknown_symbol_warnings.Write(
 200               'No known section for symbol ' + symbol_info.name)
 201     except SymbolNotFoundException:
 202       symbol_not_found_warnings.Write(
 203           'Did not find function in binary. offset: ' + hex(offset))
 204       success = False
 205   unknown_symbol_warnings.WriteEnd('no known section for symbol.')
 206   symbol_not_found_warnings.WriteEnd('symbol not found in the binary.')
 207   return success
 208
 209
 210 def main():
 211   parser = optparse.OptionParser(usage=
 212       'usage: %prog [options] <merged_cyglog> <library> <output_filename>')
 213   parser.add_option('--target-arch', action='store', dest='arch',
 214                     choices=['arm', 'arm64', 'x86', 'x86_64', 'x64', 'mips'],
 215                     help='The target architecture for libchrome.so')
 216   options, argv = parser.parse_args(sys.argv)
 217   if not options.arch:
 218     options.arch = cygprofile_utils.DetectArchitecture()
 219   if len(argv) != 4:
 220     parser.print_help()
 221     return 1
 222   (log_filename, lib_filename, output_filename) = argv[1:]
 223   symbol_extractor.SetArchitecture(options.arch)
 224
 225   obj_dir = os.path.abspath(os.path.join(
 226       os.path.dirname(lib_filename), '../obj'))
 227
 228   log_file_lines = map(string.rstrip, open(log_filename).readlines())
 229   offsets = _ParseLogLines(log_file_lines)
 230   _WarnAboutDuplicates(offsets)
 231
 232   offset_to_symbol_infos = _GroupLibrarySymbolInfosByOffset(lib_filename)
 233   symbol_to_section_map = _GetSymbolToSectionMapFromObjectFiles(obj_dir)
 234
 235   success = False
 236   temp_filename = None
 237   output_file = None
 238   try:
 239     (fd, temp_filename) = tempfile.mkstemp(dir=os.path.dirname(output_filename))
 240     output_file = os.fdopen(fd, 'w')
 241     ok = _OutputOrderfile(
 242         offsets, offset_to_symbol_infos, symbol_to_section_map, output_file)
 243     output_file.close()
 244     os.rename(temp_filename, output_filename)
 245     temp_filename = None
 246     success = ok
 247   finally:
 248     if output_file:
 249       output_file.close()
 250     if temp_filename:
 251       os.remove(temp_filename)
 252
 253   return 0 if success else 1
 254
 255
 256 if __name__ == '__main__':
 257   logging.basicConfig(level=logging.INFO)
 258   sys.exit(main())