bin/symbols-check.py

   1 #!/usr/bin/env python
   2
   3 import argparse
   4 import os
   5 import platform
   6 import subprocess
   7
   8 # This list contains symbols that _might_ be exported for some platforms
   9 PLATFORM_SYMBOLS = [
  10     '__bss_end__',
  11     '__bss_start__',
  12     '__bss_start',
  13     '__end__',
  14     '_bss_end__',
  15     '_edata',
  16     '_end',
  17     '_fini',
  18     '_init',
  19 ]
  20
  21
  22 def get_symbols_nm(nm, lib):
  23     '''
  24     List all the (non platform-specific) symbols exported by the library
  25     using `nm`
  26     '''
  27     symbols = []
  28     platform_name = platform.system()
  29     output = subprocess.check_output([nm, '-gP', lib],
  30                                      stderr=open(os.devnull, 'w')).decode("ascii")
  31     for line in output.splitlines():
  32         fields = line.split()
  33         if len(fields) == 2 or fields[1] == 'U':
  34             continue
  35         symbol_name = fields[0]
  36         if platform_name == 'Linux':
  37             if symbol_name in PLATFORM_SYMBOLS:
  38                 continue
  39         elif platform_name == 'Darwin':
  40             assert symbol_name[0] == '_'
  41             symbol_name = symbol_name[1:]
  42         symbols.append(symbol_name)
  43     return symbols
  44
  45
  46 def get_symbols_dumpbin(dumpbin, lib):
  47     '''
  48     List all the (non platform-specific) symbols exported by the library
  49     using `dumpbin`
  50     '''
  51     symbols = []
  52     output = subprocess.check_output([dumpbin, '/exports', lib],
  53                                      stderr=open(os.devnull, 'w')).decode("ascii")
  54     for line in output.splitlines():
  55         fields = line.split()
  56         # The lines with the symbols are made of at least 4 columns; see details below
  57         if len(fields) < 4:
  58             continue
  59         try:
  60             # Making sure the first 3 columns are a dec counter, a hex counter
  61             # and a hex address
  62             _ = int(fields[0], 10)
  63             _ = int(fields[1], 16)
  64             _ = int(fields[2], 16)
  65         except ValueError:
  66             continue
  67         symbol_name = fields[3]
  68         # De-mangle symbols
  69         if symbol_name[0] == '_':
  70             symbol_name = symbol_name[1:].split('@')[0]
  71         symbols.append(symbol_name)
  72     return symbols
  73
  74
  75 def main():
  76     parser = argparse.ArgumentParser()
  77     parser.add_argument('--symbols-file',
  78                         action='store',
  79                         required=True,
  80                         help='path to file containing symbols')
  81     parser.add_argument('--lib',
  82                         action='store',
  83                         required=True,
  84                         help='path to library')
  85     parser.add_argument('--nm',
  86                         action='store',
  87                         help='path to binary (or name in $PATH)')
  88     parser.add_argument('--dumpbin',
  89                         action='store',
  90                         help='path to binary (or name in $PATH)')
  91     parser.add_argument('--ignore-symbol',
  92                         action='append',
  93                         help='do not process this symbol')
  94     args = parser.parse_args()
  95
  96     try:
  97         if platform.system() == 'Windows':
  98             if not args.dumpbin:
  99                 parser.error('--dumpbin is mandatory')
 100             lib_symbols = get_symbols_dumpbin(args.dumpbin, args.lib)
 101         else:
 102             if not args.nm:
 103                 parser.error('--nm is mandatory')
 104             lib_symbols = get_symbols_nm(args.nm, args.lib)
 105     except:
 106         # We can't run this test, but we haven't technically failed it either
 107         # Return the GNU "skip" error code
 108         exit(77)
 109     mandatory_symbols = []
 110     optional_symbols = []
 111     with open(args.symbols_file) as symbols_file:
 112         qualifier_optional = '(optional)'
 113         for line in symbols_file.readlines():
 114
 115             # Strip comments
 116             line = line.split('#')[0]
 117             line = line.strip()
 118             if not line:
 119                 continue
 120
 121             # Line format:
 122             # [qualifier] symbol
 123             qualifier = None
 124             symbol = None
 125
 126             fields = line.split()
 127             if len(fields) == 1:
 128                 symbol = fields[0]
 129             elif len(fields) == 2:
 130                 qualifier = fields[0]
 131                 symbol = fields[1]
 132             else:
 133                 print(args.symbols_file + ': invalid format: ' + line)
 134                 exit(1)
 135
 136             # The only supported qualifier is 'optional', which means the
 137             # symbol doesn't have to be exported by the library
 138             if qualifier and not qualifier == qualifier_optional:
 139                 print(args.symbols_file + ': invalid qualifier: ' + qualifier)
 140                 exit(1)
 141
 142             if qualifier == qualifier_optional:
 143                 optional_symbols.append(symbol)
 144             else:
 145                 mandatory_symbols.append(symbol)
 146
 147     unknown_symbols = []
 148     for symbol in lib_symbols:
 149         if symbol in mandatory_symbols:
 150             continue
 151         if symbol in optional_symbols:
 152             continue
 153         if args.ignore_symbol and symbol in args.ignore_symbol:
 154             continue
 155         if symbol[:2] == '_Z':
 156             # As ajax found out, the compiler intentionally exports symbols
 157             # that we explicitely asked it not to export, and we can't do
 158             # anything about it:
 159             # https://gcc.gnu.org/bugzilla/show_bug.cgi?id=36022#c4
 160             continue
 161         unknown_symbols.append(symbol)
 162
 163     missing_symbols = [
 164         sym for sym in mandatory_symbols if sym not in lib_symbols
 165     ]
 166
 167     for symbol in unknown_symbols:
 168         print(args.lib + ': unknown symbol exported: ' + symbol)
 169
 170     for symbol in missing_symbols:
 171         print(args.lib + ': missing symbol: ' + symbol)
 172
 173     if unknown_symbols or missing_symbols:
 174         exit(1)
 175     exit(0)
 176
 177
 178 if __name__ == '__main__':
 179     main()