scripts/readelf.py

   1 #!/usr/bin/env python
   2 #-------------------------------------------------------------------------------
   3 # scripts/readelf.py
   4 #
   5 # A clone of 'readelf' in Python, based on the pyelftools library
   6 #
   7 # Eli Bendersky (eliben@gmail.com)
   8 # This code is in the public domain
   9 #-------------------------------------------------------------------------------
  10 import os, sys
  11 from optparse import OptionParser
  12 import string
  13
  14
  15 # If elftools is not installed, maybe we're running from the root or scripts
  16 # dir of the source distribution
  17 #
  18 try:
  19     import elftools
  20 except ImportError:
  21     sys.path.extend(['.', '..'])
  22
  23 from elftools import __version__
  24 from elftools.common.exceptions import ELFError
  25 from elftools.elf.elffile import ELFFile
  26 from elftools.elf.segments import InterpSegment
  27 from elftools.elf.sections import SymbolTableSection
  28 from elftools.elf.descriptions import (
  29     describe_ei_class, describe_ei_data, describe_ei_version,
  30     describe_ei_osabi, describe_e_type, describe_e_machine,
  31     describe_e_version_numeric, describe_p_type, describe_p_flags,
  32     describe_sh_type, describe_sh_flags,
  33     describe_symbol_type, describe_symbol_bind, describe_symbol_visibility,
  34     describe_symbol_shndx,
  35     )
  36
  37
  38 class ReadElf(object):
  39     """ display_* methods are used to emit output into the output stream
  40     """
  41     def __init__(self, file, output):
  42         """ file:
  43                 stream object with the ELF file to read
  44
  45             output:
  46                 output stream to write to
  47         """
  48         self.elffile = ELFFile(file)
  49         self.output = output
  50
  51     def display_file_header(self):
  52         """ Display the ELF file header
  53         """
  54         self._emitline('ELF Header:')
  55         self._emit('  Magic:   ')
  56         self._emitline(' '.join('%2.2x' % ord(b)
  57                                     for b in self.elffile.e_ident_raw))
  58         header = self.elffile.header
  59         e_ident = header['e_ident']
  60         self._emitline('  Class:                             %s' %
  61                 describe_ei_class(e_ident['EI_CLASS']))
  62         self._emitline('  Data:                              %s' %
  63                 describe_ei_data(e_ident['EI_DATA']))
  64         self._emitline('  Version:                           %s' %
  65                 describe_ei_version(e_ident['EI_VERSION']))
  66         self._emitline('  OS/ABI:                            %s' %
  67                 describe_ei_osabi(e_ident['EI_OSABI']))
  68         self._emitline('  ABI Version:                       %d' %
  69                 e_ident['EI_ABIVERSION'])
  70         self._emitline('  Type:                              %s' %
  71                 describe_e_type(header['e_type']))
  72         self._emitline('  Machine:                           %s' %
  73                 describe_e_machine(header['e_machine']))
  74         self._emitline('  Version:                           %s' %
  75                 describe_e_version_numeric(header['e_version']))
  76         self._emitline('  Entry point address:               %s' %
  77                 self._format_hex(header['e_entry']))
  78         self._emit('  Start of program headers:          %s' %
  79                 header['e_phoff'])
  80         self._emitline(' (bytes into file)')
  81         self._emit('  Start of section headers:          %s' %
  82                 header['e_shoff'])
  83         self._emitline(' (bytes into file)')
  84         self._emitline('  Flags:                             %s' %
  85                 self._format_hex(header['e_flags']))
  86         self._emitline('  Size of this header:               %s (bytes)' %
  87                 header['e_ehsize'])
  88         self._emitline('  Size of program headers:           %s (bytes)' %
  89                 header['e_phentsize'])
  90         self._emitline('  Number of program headers:         %s' %
  91                 header['e_phnum'])
  92         self._emitline('  Size of section headers:           %s (bytes)' %
  93                 header['e_shentsize'])
  94         self._emitline('  Number of section headers:         %s' %
  95                 header['e_shnum'])
  96         self._emitline('  Section header string table index: %s' %
  97                 header['e_shstrndx'])
  98
  99     def display_program_headers(self, show_heading=True):
 100         """ Display the ELF program headers.
 101             If show_heading is True, displays the heading for this information
 102             (Elf file type is...)
 103         """
 104         self._emitline()
 105         if self.elffile.num_segments() == 0:
 106             self._emitline('There are no program headers in this file.')
 107             return
 108
 109         elfheader = self.elffile.header
 110         if show_heading:
 111             self._emitline('Elf file type is %s' %
 112                 describe_e_type(elfheader['e_type']))
 113             self._emitline('Entry point is %s' %
 114                 self._format_hex(elfheader['e_entry']))
 115             # readelf weirness - why isn't e_phoff printed as hex? (for section
 116             # headers, it is...)
 117             self._emitline('There are %s program headers, starting at offset %s' % (
 118                 elfheader['e_phnum'], elfheader['e_phoff']))
 119             self._emitline()
 120
 121         self._emitline('Program Headers:')
 122
 123         # Now comes the table of program headers with their attributes. Note
 124         # that due to different formatting constraints of 32-bit and 64-bit
 125         # addresses, there are some conditions on elfclass here.
 126         #
 127         # First comes the table heading
 128         #
 129         if self.elffile.elfclass == 32:
 130             self._emitline('  Type           Offset   VirtAddr   PhysAddr   FileSiz MemSiz  Flg Align')
 131         else:
 132             self._emitline('  Type           Offset             VirtAddr           PhysAddr')
 133             self._emitline('                 FileSiz            MemSiz              Flags  Align')
 134
 135         # Now the entries
 136         #
 137         for segment in self.elffile.iter_segments():
 138             self._emit('  %-14s ' % describe_p_type(segment['p_type']))
 139
 140             if self.elffile.elfclass == 32:
 141                 self._emitline('%s %s %s %s %s %-3s %s' % (
 142                     self._format_hex(segment['p_offset'], fieldsize=6),
 143                     self._format_hex(segment['p_vaddr'], fullhex=True),
 144                     self._format_hex(segment['p_paddr'], fullhex=True),
 145                     self._format_hex(segment['p_filesz'], fieldsize=5),
 146                     self._format_hex(segment['p_memsz'], fieldsize=5),
 147                     describe_p_flags(segment['p_flags']),
 148                     self._format_hex(segment['p_align'])))
 149             else: # 64
 150                 self._emitline('%s %s %s' % (
 151                     self._format_hex(segment['p_offset'], fullhex=True),
 152                     self._format_hex(segment['p_vaddr'], fullhex=True),
 153                     self._format_hex(segment['p_paddr'], fullhex=True)))
 154                 self._emitline('                 %s %s  %-3s    %s' % (
 155                     self._format_hex(segment['p_filesz'], fullhex=True),
 156                     self._format_hex(segment['p_memsz'], fullhex=True),
 157                     describe_p_flags(segment['p_flags']),
 158                     # lead0x set to False for p_align, to mimic readelf.
 159                     # No idea why the difference from 32-bit mode :-|
 160                     self._format_hex(segment['p_align'], lead0x=False)))
 161
 162             if isinstance(segment, InterpSegment):
 163                 self._emitline('      [Requesting program interpreter: %s]' %
 164                     segment.get_interp_name())
 165
 166         # Sections to segments mapping
 167         #
 168         if self.elffile.num_sections() == 0:
 169             # No sections? We're done
 170             return
 171
 172         self._emitline('\n Section to Segment mapping:')
 173         self._emitline('  Segment Sections...')
 174
 175         for nseg, segment in enumerate(self.elffile.iter_segments()):
 176             self._emit('   %2.2d     ' % nseg)
 177
 178             for section in self.elffile.iter_sections():
 179                 if (    not section.is_null() and
 180                         segment.section_in_segment(section)):
 181                     self._emit('%s ' % section.name)
 182
 183             self._emitline('')
 184
 185     def display_section_headers(self, show_heading=True):
 186         """ Display the ELF section headers
 187         """
 188         elfheader = self.elffile.header
 189         if show_heading:
 190             self._emitline('There are %s section headers, starting at offset %s' % (
 191                 elfheader['e_shnum'], self._format_hex(elfheader['e_shoff'])))
 192
 193         self._emitline('\nSection Header%s:' % (
 194             's' if elfheader['e_shnum'] > 1 else ''))
 195
 196         # Different formatting constraints of 32-bit and 64-bit addresses
 197         #
 198         if self.elffile.elfclass == 32:
 199             self._emitline('  [Nr] Name              Type            Addr     Off    Size   ES Flg Lk Inf Al')
 200         else:
 201             self._emitline('  [Nr] Name              Type             Address           Offset')
 202             self._emitline('       Size              EntSize          Flags  Link  Info  Align')
 203
 204         # Now the entries
 205         #
 206         for nsec, section in enumerate(self.elffile.iter_sections()):
 207             self._emit('  [%2u] %-17.17s %-15.15s ' % (
 208                 nsec, section.name, describe_sh_type(section['sh_type'])))
 209
 210             if self.elffile.elfclass == 32:
 211                 self._emitline('%s %s %s %s %3s %2s %3s %2s' % (
 212                     self._format_hex(section['sh_addr'], fieldsize=8, lead0x=False),
 213                     self._format_hex(section['sh_offset'], fieldsize=6, lead0x=False),
 214                     self._format_hex(section['sh_size'], fieldsize=6, lead0x=False),
 215                     self._format_hex(section['sh_entsize'], fieldsize=2, lead0x=False),
 216                     describe_sh_flags(section['sh_flags']),
 217                     section['sh_link'], section['sh_info'],
 218                     section['sh_addralign']))
 219             else: # 64
 220                 self._emitline(' %s  %s' % (
 221                     self._format_hex(section['sh_addr'], fullhex=True, lead0x=False),
 222                     self._format_hex(section['sh_offset'],
 223                         fieldsize=16 if section['sh_offset'] > 0xffffffff else 8,
 224                         lead0x=False)))
 225                 self._emitline('       %s  %s %3s      %2s   %3s     %s' % (
 226                     self._format_hex(section['sh_size'], fullhex=True, lead0x=False),
 227                     self._format_hex(section['sh_entsize'], fullhex=True, lead0x=False),
 228                     describe_sh_flags(section['sh_flags']),
 229                     section['sh_link'], section['sh_info'],
 230                     section['sh_addralign']))
 231
 232         self._emitline('Key to Flags:')
 233         self._emitline('  W (write), A (alloc), X (execute), M (merge), S (strings)')
 234         self._emitline('  I (info), L (link order), G (group), x (unknown)')
 235         self._emitline('  O (extra OS processing required) o (OS specific), p (processor specific)')
 236
 237     def display_symbol_tables(self):
 238         """ Display the symbol tables contained in the file
 239         """
 240         for section in self.elffile.iter_sections():
 241             if not isinstance(section, SymbolTableSection):
 242                 continue
 243
 244             if section['sh_entsize'] == 0:
 245                 self._emitline("\nSymbol table '%s' has a sh_entsize of zero!" % (
 246                     section.name))
 247                 continue
 248
 249             self._emitline("\nSymbol table '%s' contains %s entries:" % (
 250                 section.name, section.num_symbols()))
 251
 252             if self.elffile.elfclass == 32:
 253                 self._emitline('   Num:    Value  Size Type    Bind   Vis      Ndx Name')
 254             else: # 64
 255                 self._emitline('   Num:    Value          Size Type    Bind   Vis      Ndx Name')
 256
 257             for nsym, symbol in enumerate(section.iter_symbols()):
 258                 # symbol names are truncated to 25 chars, similarly to readelf
 259                 self._emitline('%6d: %s %5d %-7s %-6s %-7s %4s %.25s' % (
 260                     nsym,
 261                     self._format_hex(symbol['st_value'], fullhex=True, lead0x=False),
 262                     symbol['st_size'],
 263                     describe_symbol_type(symbol['st_info']['type']),
 264                     describe_symbol_bind(symbol['st_info']['bind']),
 265                     describe_symbol_visibility(symbol['st_other']['visibility']),
 266                     describe_symbol_shndx(symbol['st_shndx']),
 267                     symbol.name))
 268
 269     def display_hex_dump(self, section_spec):
 270         """ Display a hex dump of a section. section_spec is either a section
 271             number or a name.
 272         """
 273         section = self._section_from_spec(section_spec)
 274         if section is None:
 275             self._emitline("Section '%s' does not exist in the file!" % (
 276                 section_spec))
 277             return
 278
 279         self._emitline("\nHex dump of section '%s':" % section.name)
 280
 281         addr = section['sh_addr']
 282         data = section.data()
 283         dataptr = 0
 284
 285         while dataptr < len(data):
 286             bytesleft = len(data) - dataptr
 287             # chunks of 16 bytes per line
 288             linebytes = 16 if bytesleft > 16 else bytesleft
 289
 290             self._emit('  %s ' % self._format_hex(addr, fieldsize=8))
 291             for i in range(16):
 292                 if i < linebytes:
 293                     self._emit('%2.2x' % ord(data[dataptr + i]))
 294                 else:
 295                     self._emit('  ')
 296                 if i % 4 == 3:
 297                     self._emit(' ')
 298
 299             for i in range(linebytes):
 300                 c = data[dataptr + i]
 301                 if c >= ' ' and ord(c) <= 0x7f:
 302                     self._emit(c)
 303                 else:
 304                     self._emit('.')
 305
 306             self._emitline()
 307             addr += linebytes
 308             dataptr += linebytes
 309
 310         self._emitline()
 311
 312     def display_string_dump(self, section_spec):
 313         """ Display a strings dump of a section. section_spec is either a
 314             section number or a name.
 315         """
 316         section = self._section_from_spec(section_spec)
 317         if section is None:
 318             self._emitline("Section '%s' does not exist in the file!" % (
 319                 section_spec))
 320             return
 321
 322         printables = set(string.printable)
 323         self._emitline("\nString dump of section '%s':" % section.name)
 324
 325         found = False
 326         data = section.data()
 327         dataptr = 0
 328
 329         while dataptr < len(data):
 330             while dataptr < len(data) and data[dataptr] not in printables:
 331                 dataptr += 1
 332
 333             if dataptr >= len(data):
 334                 break
 335
 336             endptr = dataptr
 337             while endptr < len(data) and data[endptr] != '\x00':
 338                 endptr += 1
 339
 340             found = True
 341             self._emitline('  [%6x]  %s' % (
 342                 dataptr, data[dataptr:endptr]))
 343
 344             dataptr = endptr
 345
 346         if not found:
 347             self._emitline('  No strings found in this section.')
 348         else:
 349             self._emitline()
 350
 351     def _format_hex(self, addr, fieldsize=None, fullhex=False, lead0x=True):
 352         """ Format an address into a hexadecimal string.
 353
 354             fieldsize:
 355                 Size of the hexadecimal field (with leading zeros to fit the
 356                 address into. For example with fieldsize=8, the format will
 357                 be %08x
 358                 If None, the minimal required field size will be used.
 359
 360             fullhex:
 361                 If True, override fieldsize to set it to the maximal size
 362                 needed for the elfclass
 363
 364             lead0x:
 365                 If True, leading 0x is added
 366         """
 367         s = '0x' if lead0x else ''
 368         if fullhex:
 369             fieldsize = 8 if self.elffile.elfclass == 32 else 16
 370         if fieldsize is None:
 371             field = '%x'
 372         else:
 373             field = '%' + '0%sx' % fieldsize
 374         return s + field % addr
 375
 376     def _section_from_spec(self, spec):
 377         """ Retrieve a section given a "spec" (either number or name).
 378             Return None if no such section exists in the file.
 379         """
 380         try:
 381             num = int(spec)
 382             if num < self.elffile.num_sections():
 383                 return self.elffile.get_section(num)
 384             else:
 385                 return None
 386         except ValueError:
 387             # Not a number. Must be a name then
 388             return self.elffile.get_section_by_name(spec)
 389
 390     def _emit(self, s=''):
 391         """ Emit an object to output
 392         """
 393         self.output.write(str(s))
 394
 395     def _emitline(self, s=''):
 396         """ Emit an object to output, followed by a newline
 397         """
 398         self.output.write(str(s) + '\n')
 399
 400
 401 SCRIPT_DESCRIPTION = 'Display information about the contents of ELF format files'
 402 VERSION_STRING = '%%prog: based on pyelftools %s' % __version__
 403
 404
 405 def main():
 406     # parse the command-line arguments and invoke ReadElf
 407     optparser = OptionParser(
 408             usage='usage: %prog [options] <elf-file>',
 409             description=SCRIPT_DESCRIPTION,
 410             add_help_option=False, # -h is a real option of readelf
 411             prog='readelf.py',
 412             version=VERSION_STRING)
 413     optparser.add_option('-H', '--help',
 414             action='store_true', dest='help',
 415             help='Display this information')
 416     optparser.add_option('-h', '--file-header',
 417             action='store_true', dest='show_file_header',
 418             help='Display the ELF file header')
 419     optparser.add_option('-l', '--program-headers', '--segments',
 420             action='store_true', dest='show_program_header',
 421             help='Display the program headers')
 422     optparser.add_option('-S', '--section-headers', '--sections',
 423             action='store_true', dest='show_section_header',
 424             help="Display the sections' headers")
 425     optparser.add_option('-e', '--headers',
 426             action='store_true', dest='show_all_headers',
 427             help='Equivalent to: -h -l -S')
 428     optparser.add_option('-s', '--symbols', '--syms',
 429             action='store_true', dest='show_symbols',
 430             help='Display the symbol table')
 431     optparser.add_option('-x', '--hex-dump',
 432             action='store', dest='show_hex_dump', metavar='<number|name>',
 433             help='Dump the contents of section <number|name> as bytes')
 434     optparser.add_option('-p', '--string-dump',
 435             action='store', dest='show_string_dump', metavar='<number|name>',
 436             help='Dump the contents of section <number|name> as strings')
 437
 438     options, args = optparser.parse_args()
 439
 440     if options.help or len(args) == 0:
 441         optparser.print_help()
 442         sys.exit(0)
 443
 444     if options.show_all_headers:
 445         do_file_header = do_section_header = do_program_header = True
 446     else:
 447         do_file_header = options.show_file_header
 448         do_section_header = options.show_section_header
 449         do_program_header = options.show_program_header
 450
 451     with open(args[0], 'rb') as file:
 452         try:
 453             readelf = ReadElf(file, sys.stdout)
 454             if do_file_header:
 455                 readelf.display_file_header()
 456             if do_section_header:
 457                 readelf.display_section_headers(
 458                         show_heading=not do_file_header)
 459             if do_program_header:
 460                 readelf.display_program_headers(
 461                         show_heading=not do_file_header)
 462             if options.show_symbols:
 463                 readelf.display_symbol_tables()
 464             if options.show_hex_dump:
 465                 readelf.display_hex_dump(options.show_hex_dump)
 466             if options.show_string_dump:
 467                 readelf.display_string_dump(options.show_string_dump)
 468         except ELFError as ex:
 469             sys.stderr.write('ELF error: %s\n' % ex)
 470             sys.exit(1)
 471
 472
 473 #-------------------------------------------------------------------------------
 474 if __name__ == '__main__':
 475     main()
 476