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 # For running from development directory.
  15 sys.path.insert(0, '.')
  16
  17
  18 from elftools import __version__
  19 from elftools.common.exceptions import ELFError
  20 from elftools.common.py3compat import (
  21         ifilter, byte2int, bytes2str, itervalues, str2bytes)
  22 from elftools.elf.elffile import ELFFile
  23 from elftools.elf.dynamic import DynamicSection, DynamicSegment
  24 from elftools.elf.enums import ENUM_D_TAG
  25 from elftools.elf.segments import InterpSegment
  26 from elftools.elf.sections import SymbolTableSection
  27 from elftools.elf.relocation import RelocationSection
  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, describe_reloc_type, describe_dyn_tag,
  35     )
  36 from elftools.dwarf.dwarfinfo import DWARFInfo
  37 from elftools.dwarf.descriptions import (
  38     describe_reg_name, describe_attr_value, set_global_machine_arch,
  39     describe_CFI_instructions, describe_CFI_register_rule,
  40     describe_CFI_CFA_rule,
  41     )
  42 from elftools.dwarf.constants import (
  43     DW_LNS_copy, DW_LNS_set_file, DW_LNE_define_file)
  44 from elftools.dwarf.callframe import CIE, FDE
  45
  46
  47 class ReadElf(object):
  48     """ display_* methods are used to emit output into the output stream
  49     """
  50     def __init__(self, file, output):
  51         """ file:
  52                 stream object with the ELF file to read
  53
  54             output:
  55                 output stream to write to
  56         """
  57         self.elffile = ELFFile(file)
  58         self.output = output
  59
  60         # Lazily initialized if a debug dump is requested
  61         self._dwarfinfo = None
  62
  63     def display_file_header(self):
  64         """ Display the ELF file header
  65         """
  66         self._emitline('ELF Header:')
  67         self._emit('  Magic:   ')
  68         self._emitline(' '.join('%2.2x' % byte2int(b)
  69                                     for b in self.elffile.e_ident_raw))
  70         header = self.elffile.header
  71         e_ident = header['e_ident']
  72         self._emitline('  Class:                             %s' %
  73                 describe_ei_class(e_ident['EI_CLASS']))
  74         self._emitline('  Data:                              %s' %
  75                 describe_ei_data(e_ident['EI_DATA']))
  76         self._emitline('  Version:                           %s' %
  77                 describe_ei_version(e_ident['EI_VERSION']))
  78         self._emitline('  OS/ABI:                            %s' %
  79                 describe_ei_osabi(e_ident['EI_OSABI']))
  80         self._emitline('  ABI Version:                       %d' %
  81                 e_ident['EI_ABIVERSION'])
  82         self._emitline('  Type:                              %s' %
  83                 describe_e_type(header['e_type']))
  84         self._emitline('  Machine:                           %s' %
  85                 describe_e_machine(header['e_machine']))
  86         self._emitline('  Version:                           %s' %
  87                 describe_e_version_numeric(header['e_version']))
  88         self._emitline('  Entry point address:               %s' %
  89                 self._format_hex(header['e_entry']))
  90         self._emit('  Start of program headers:          %s' %
  91                 header['e_phoff'])
  92         self._emitline(' (bytes into file)')
  93         self._emit('  Start of section headers:          %s' %
  94                 header['e_shoff'])
  95         self._emitline(' (bytes into file)')
  96         self._emitline('  Flags:                             %s' %
  97                 self._format_hex(header['e_flags']))
  98         self._emitline('  Size of this header:               %s (bytes)' %
  99                 header['e_ehsize'])
 100         self._emitline('  Size of program headers:           %s (bytes)' %
 101                 header['e_phentsize'])
 102         self._emitline('  Number of program headers:         %s' %
 103                 header['e_phnum'])
 104         self._emitline('  Size of section headers:           %s (bytes)' %
 105                 header['e_shentsize'])
 106         self._emitline('  Number of section headers:         %s' %
 107                 header['e_shnum'])
 108         self._emitline('  Section header string table index: %s' %
 109                 header['e_shstrndx'])
 110
 111     def display_program_headers(self, show_heading=True):
 112         """ Display the ELF program headers.
 113             If show_heading is True, displays the heading for this information
 114             (Elf file type is...)
 115         """
 116         self._emitline()
 117         if self.elffile.num_segments() == 0:
 118             self._emitline('There are no program headers in this file.')
 119             return
 120
 121         elfheader = self.elffile.header
 122         if show_heading:
 123             self._emitline('Elf file type is %s' %
 124                 describe_e_type(elfheader['e_type']))
 125             self._emitline('Entry point is %s' %
 126                 self._format_hex(elfheader['e_entry']))
 127             # readelf weirness - why isn't e_phoff printed as hex? (for section
 128             # headers, it is...)
 129             self._emitline('There are %s program headers, starting at offset %s' % (
 130                 elfheader['e_phnum'], elfheader['e_phoff']))
 131             self._emitline()
 132
 133         self._emitline('Program Headers:')
 134
 135         # Now comes the table of program headers with their attributes. Note
 136         # that due to different formatting constraints of 32-bit and 64-bit
 137         # addresses, there are some conditions on elfclass here.
 138         #
 139         # First comes the table heading
 140         #
 141         if self.elffile.elfclass == 32:
 142             self._emitline('  Type           Offset   VirtAddr   PhysAddr   FileSiz MemSiz  Flg Align')
 143         else:
 144             self._emitline('  Type           Offset             VirtAddr           PhysAddr')
 145             self._emitline('                 FileSiz            MemSiz              Flags  Align')
 146
 147         # Now the entries
 148         #
 149         for segment in self.elffile.iter_segments():
 150             self._emit('  %-14s ' % describe_p_type(segment['p_type']))
 151
 152             if self.elffile.elfclass == 32:
 153                 self._emitline('%s %s %s %s %s %-3s %s' % (
 154                     self._format_hex(segment['p_offset'], fieldsize=6),
 155                     self._format_hex(segment['p_vaddr'], fullhex=True),
 156                     self._format_hex(segment['p_paddr'], fullhex=True),
 157                     self._format_hex(segment['p_filesz'], fieldsize=5),
 158                     self._format_hex(segment['p_memsz'], fieldsize=5),
 159                     describe_p_flags(segment['p_flags']),
 160                     self._format_hex(segment['p_align'])))
 161             else: # 64
 162                 self._emitline('%s %s %s' % (
 163                     self._format_hex(segment['p_offset'], fullhex=True),
 164                     self._format_hex(segment['p_vaddr'], fullhex=True),
 165                     self._format_hex(segment['p_paddr'], fullhex=True)))
 166                 self._emitline('                 %s %s  %-3s    %s' % (
 167                     self._format_hex(segment['p_filesz'], fullhex=True),
 168                     self._format_hex(segment['p_memsz'], fullhex=True),
 169                     describe_p_flags(segment['p_flags']),
 170                     # lead0x set to False for p_align, to mimic readelf.
 171                     # No idea why the difference from 32-bit mode :-|
 172                     self._format_hex(segment['p_align'], lead0x=False)))
 173
 174             if isinstance(segment, InterpSegment):
 175                 self._emitline('      [Requesting program interpreter: %s]' %
 176                     bytes2str(segment.get_interp_name()))
 177
 178         # Sections to segments mapping
 179         #
 180         if self.elffile.num_sections() == 0:
 181             # No sections? We're done
 182             return
 183
 184         self._emitline('\n Section to Segment mapping:')
 185         self._emitline('  Segment Sections...')
 186
 187         for nseg, segment in enumerate(self.elffile.iter_segments()):
 188             self._emit('   %2.2d     ' % nseg)
 189
 190             for section in self.elffile.iter_sections():
 191                 if (    not section.is_null() and
 192                         segment.section_in_segment(section)):
 193                     self._emit('%s ' % bytes2str(section.name))
 194
 195             self._emitline('')
 196
 197     def display_section_headers(self, show_heading=True):
 198         """ Display the ELF section headers
 199         """
 200         elfheader = self.elffile.header
 201         if show_heading:
 202             self._emitline('There are %s section headers, starting at offset %s' % (
 203                 elfheader['e_shnum'], self._format_hex(elfheader['e_shoff'])))
 204
 205         self._emitline('\nSection Header%s:' % (
 206             's' if elfheader['e_shnum'] > 1 else ''))
 207
 208         # Different formatting constraints of 32-bit and 64-bit addresses
 209         #
 210         if self.elffile.elfclass == 32:
 211             self._emitline('  [Nr] Name              Type            Addr     Off    Size   ES Flg Lk Inf Al')
 212         else:
 213             self._emitline('  [Nr] Name              Type             Address           Offset')
 214             self._emitline('       Size              EntSize          Flags  Link  Info  Align')
 215
 216         # Now the entries
 217         #
 218         for nsec, section in enumerate(self.elffile.iter_sections()):
 219             self._emit('  [%2u] %-17.17s %-15.15s ' % (
 220                 nsec, bytes2str(section.name), describe_sh_type(section['sh_type'])))
 221
 222             if self.elffile.elfclass == 32:
 223                 self._emitline('%s %s %s %s %3s %2s %3s %2s' % (
 224                     self._format_hex(section['sh_addr'], fieldsize=8, lead0x=False),
 225                     self._format_hex(section['sh_offset'], fieldsize=6, lead0x=False),
 226                     self._format_hex(section['sh_size'], fieldsize=6, lead0x=False),
 227                     self._format_hex(section['sh_entsize'], fieldsize=2, lead0x=False),
 228                     describe_sh_flags(section['sh_flags']),
 229                     section['sh_link'], section['sh_info'],
 230                     section['sh_addralign']))
 231             else: # 64
 232                 self._emitline(' %s  %s' % (
 233                     self._format_hex(section['sh_addr'], fullhex=True, lead0x=False),
 234                     self._format_hex(section['sh_offset'],
 235                         fieldsize=16 if section['sh_offset'] > 0xffffffff else 8,
 236                         lead0x=False)))
 237                 self._emitline('       %s  %s %3s      %2s   %3s     %s' % (
 238                     self._format_hex(section['sh_size'], fullhex=True, lead0x=False),
 239                     self._format_hex(section['sh_entsize'], fullhex=True, lead0x=False),
 240                     describe_sh_flags(section['sh_flags']),
 241                     section['sh_link'], section['sh_info'],
 242                     section['sh_addralign']))
 243
 244         self._emitline('Key to Flags:')
 245         self._emit('  W (write), A (alloc), X (execute), M (merge), S (strings)')
 246         if self.elffile['e_machine'] in ('EM_X86_64', 'EM_L10M'):
 247             self._emitline(', l (large)')
 248         else:
 249             self._emitline()
 250         self._emitline('  I (info), L (link order), G (group), T (TLS), E (exclude), x (unknown)')
 251         self._emitline('  O (extra OS processing required) o (OS specific), p (processor specific)')
 252
 253     def display_symbol_tables(self):
 254         """ Display the symbol tables contained in the file
 255         """
 256         for section in self.elffile.iter_sections():
 257             if not isinstance(section, SymbolTableSection):
 258                 continue
 259
 260             if section['sh_entsize'] == 0:
 261                 self._emitline("\nSymbol table '%s' has a sh_entsize of zero!" % (
 262                     bytes2str(section.name)))
 263                 continue
 264
 265             self._emitline("\nSymbol table '%s' contains %s entries:" % (
 266                 bytes2str(section.name), section.num_symbols()))
 267
 268             if self.elffile.elfclass == 32:
 269                 self._emitline('   Num:    Value  Size Type    Bind   Vis      Ndx Name')
 270             else: # 64
 271                 self._emitline('   Num:    Value          Size Type    Bind   Vis      Ndx Name')
 272
 273             for nsym, symbol in enumerate(section.iter_symbols()):
 274                 # symbol names are truncated to 25 chars, similarly to readelf
 275                 self._emitline('%6d: %s %5d %-7s %-6s %-7s %4s %.25s' % (
 276                     nsym,
 277                     self._format_hex(symbol['st_value'], fullhex=True, lead0x=False),
 278                     symbol['st_size'],
 279                     describe_symbol_type(symbol['st_info']['type']),
 280                     describe_symbol_bind(symbol['st_info']['bind']),
 281                     describe_symbol_visibility(symbol['st_other']['visibility']),
 282                     describe_symbol_shndx(symbol['st_shndx']),
 283                     bytes2str(symbol.name)))
 284
 285     def display_dynamic_tags(self):
 286         """ Display the dynamic tags contained in the file
 287         """
 288         for section in self.elffile.iter_sections():
 289             if not isinstance(section, DynamicSection):
 290                 continue
 291
 292             self._emitline("\nDynamic section at offset %s contains %s entries:" % (
 293                 self._format_hex(section['sh_offset']),
 294                 section.num_tags()))
 295             self._emitline("  Tag        Type                         Name/Value")
 296
 297             padding = 20 + (8 if self.elffile.elfclass == 32 else 0)
 298             for tag in section.iter_tags():
 299                 if tag.entry.d_tag == 'DT_NEEDED':
 300                     parsed = 'Shared library: [%s]' % bytes2str(tag.needed)
 301                 elif tag.entry.d_tag == 'DT_RPATH':
 302                     parsed = 'Library rpath: [%s]' % bytes2str(tag.rpath)
 303                 elif tag.entry.d_tag == 'DT_RUNPATH':
 304                     parsed = 'Library runpath: [%s]' % bytes2str(tag.runpath)
 305                 elif tag.entry.d_tag == 'DT_SONAME':
 306                     parsed = 'Library soname: [%s]' % bytes2str(tag.soname)
 307                 elif (tag.entry.d_tag.endswith('SZ') or
 308                       tag.entry.d_tag.endswith('ENT')):
 309                     parsed = '%i (bytes)' % tag['d_val']
 310                 elif (tag.entry.d_tag.endswith('NUM') or
 311                       tag.entry.d_tag.endswith('COUNT')):
 312                     parsed = '%i' % tag['d_val']
 313                 elif tag.entry.d_tag == 'DT_PLTREL':
 314                     s = describe_dyn_tag(tag.entry.d_val)
 315                     if s.startswith('DT_'):
 316                         s = s[3:]
 317                     parsed = '%s' % s
 318                 else:
 319                     parsed = '%#x' % tag['d_val']
 320
 321                 self._emitline(" %s %-*s %s" % (
 322                     self._format_hex(ENUM_D_TAG.get(tag.entry.d_tag, tag.entry.d_tag),
 323                         fullhex=True, lead0x=True),
 324                     padding,
 325                     '(%s)' % (tag.entry.d_tag[3:],),
 326                     parsed))
 327
 328     def display_relocations(self):
 329         """ Display the relocations contained in the file
 330         """
 331         has_relocation_sections = False
 332         for section in self.elffile.iter_sections():
 333             if not isinstance(section, RelocationSection):
 334                 continue
 335
 336             has_relocation_sections = True
 337             self._emitline("\nRelocation section '%s' at offset %s contains %s entries:" % (
 338                 bytes2str(section.name),
 339                 self._format_hex(section['sh_offset']),
 340                 section.num_relocations()))
 341             if section.is_RELA():
 342                 self._emitline("  Offset          Info           Type           Sym. Value    Sym. Name + Addend")
 343             else:
 344                 self._emitline(" Offset     Info    Type            Sym.Value  Sym. Name")
 345
 346             # The symbol table section pointed to in sh_link
 347             symtable = self.elffile.get_section(section['sh_link'])
 348
 349             for rel in section.iter_relocations():
 350                 hexwidth = 8 if self.elffile.elfclass == 32 else 12
 351                 self._emit('%s  %s %-17.17s' % (
 352                     self._format_hex(rel['r_offset'],
 353                         fieldsize=hexwidth, lead0x=False),
 354                     self._format_hex(rel['r_info'],
 355                         fieldsize=hexwidth, lead0x=False),
 356                     describe_reloc_type(
 357                         rel['r_info_type'], self.elffile)))
 358
 359                 if rel['r_info_sym'] == 0:
 360                     self._emitline()
 361                     continue
 362
 363                 symbol = symtable.get_symbol(rel['r_info_sym'])
 364                 # Some symbols have zero 'st_name', so instead what's used is
 365                 # the name of the section they point at
 366                 if symbol['st_name'] == 0:
 367                     symsec = self.elffile.get_section(symbol['st_shndx'])
 368                     symbol_name = symsec.name
 369                 else:
 370                     symbol_name = symbol.name
 371                 self._emit(' %s %s%22.22s' % (
 372                     self._format_hex(
 373                         symbol['st_value'],
 374                         fullhex=True, lead0x=False),
 375                     '  ' if self.elffile.elfclass == 32 else '',
 376                     bytes2str(symbol_name)))
 377                 if section.is_RELA():
 378                     self._emit(' %s %x' % (
 379                         '+' if rel['r_addend'] >= 0 else '-',
 380                         abs(rel['r_addend'])))
 381                 self._emitline()
 382
 383         if not has_relocation_sections:
 384             self._emitline('\nThere are no relocations in this file.')
 385
 386     def display_hex_dump(self, section_spec):
 387         """ Display a hex dump of a section. section_spec is either a section
 388             number or a name.
 389         """
 390         section = self._section_from_spec(section_spec)
 391         if section is None:
 392             self._emitline("Section '%s' does not exist in the file!" % (
 393                 section_spec))
 394             return
 395
 396         self._emitline("\nHex dump of section '%s':" % bytes2str(section.name))
 397         self._note_relocs_for_section(section)
 398         addr = section['sh_addr']
 399         data = section.data()
 400         dataptr = 0
 401
 402         while dataptr < len(data):
 403             bytesleft = len(data) - dataptr
 404             # chunks of 16 bytes per line
 405             linebytes = 16 if bytesleft > 16 else bytesleft
 406
 407             self._emit('  %s ' % self._format_hex(addr, fieldsize=8))
 408             for i in range(16):
 409                 if i < linebytes:
 410                     self._emit('%2.2x' % byte2int(data[dataptr + i]))
 411                 else:
 412                     self._emit('  ')
 413                 if i % 4 == 3:
 414                     self._emit(' ')
 415
 416             for i in range(linebytes):
 417                 c = data[dataptr + i : dataptr + i + 1]
 418                 if byte2int(c[0]) >= 32 and byte2int(c[0]) < 0x7f:
 419                     self._emit(bytes2str(c))
 420                 else:
 421                     self._emit(bytes2str(b'.'))
 422
 423             self._emitline()
 424             addr += linebytes
 425             dataptr += linebytes
 426
 427         self._emitline()
 428
 429     def display_string_dump(self, section_spec):
 430         """ Display a strings dump of a section. section_spec is either a
 431             section number or a name.
 432         """
 433         section = self._section_from_spec(section_spec)
 434         if section is None:
 435             self._emitline("Section '%s' does not exist in the file!" % (
 436                 section_spec))
 437             return
 438
 439         self._emitline("\nString dump of section '%s':" % bytes2str(section.name))
 440
 441         found = False
 442         data = section.data()
 443         dataptr = 0
 444
 445         while dataptr < len(data):
 446             while ( dataptr < len(data) and
 447                     not (32 <= byte2int(data[dataptr]) <= 127)):
 448                 dataptr += 1
 449
 450             if dataptr >= len(data):
 451                 break
 452
 453             endptr = dataptr
 454             while endptr < len(data) and byte2int(data[endptr]) != 0:
 455                 endptr += 1
 456
 457             found = True
 458             self._emitline('  [%6x]  %s' % (
 459                 dataptr, bytes2str(data[dataptr:endptr])))
 460
 461             dataptr = endptr
 462
 463         if not found:
 464             self._emitline('  No strings found in this section.')
 465         else:
 466             self._emitline()
 467
 468     def display_debug_dump(self, dump_what):
 469         """ Dump a DWARF section
 470         """
 471         self._init_dwarfinfo()
 472         if self._dwarfinfo is None:
 473             return
 474
 475         set_global_machine_arch(self.elffile.get_machine_arch())
 476
 477         if dump_what == 'info':
 478             self._dump_debug_info()
 479         elif dump_what == 'decodedline':
 480             self._dump_debug_line_programs()
 481         elif dump_what == 'frames':
 482             self._dump_debug_frames()
 483         elif dump_what == 'frames-interp':
 484             self._dump_debug_frames_interp()
 485         else:
 486             self._emitline('debug dump not yet supported for "%s"' % dump_what)
 487
 488     def _format_hex(self, addr, fieldsize=None, fullhex=False, lead0x=True):
 489         """ Format an address into a hexadecimal string.
 490
 491             fieldsize:
 492                 Size of the hexadecimal field (with leading zeros to fit the
 493                 address into. For example with fieldsize=8, the format will
 494                 be %08x
 495                 If None, the minimal required field size will be used.
 496
 497             fullhex:
 498                 If True, override fieldsize to set it to the maximal size
 499                 needed for the elfclass
 500
 501             lead0x:
 502                 If True, leading 0x is added
 503         """
 504         s = '0x' if lead0x else ''
 505         if fullhex:
 506             fieldsize = 8 if self.elffile.elfclass == 32 else 16
 507         if fieldsize is None:
 508             field = '%x'
 509         else:
 510             field = '%' + '0%sx' % fieldsize
 511         return s + field % addr
 512
 513     def _section_from_spec(self, spec):
 514         """ Retrieve a section given a "spec" (either number or name).
 515             Return None if no such section exists in the file.
 516         """
 517         try:
 518             num = int(spec)
 519             if num < self.elffile.num_sections():
 520                 return self.elffile.get_section(num)
 521             else:
 522                 return None
 523         except ValueError:
 524             # Not a number. Must be a name then
 525             return self.elffile.get_section_by_name(str2bytes(spec))
 526
 527     def _note_relocs_for_section(self, section):
 528         """ If there are relocation sections pointing to the givne section,
 529             emit a note about it.
 530         """
 531         for relsec in self.elffile.iter_sections():
 532             if isinstance(relsec, RelocationSection):
 533                 info_idx = relsec['sh_info']
 534                 if self.elffile.get_section(info_idx) == section:
 535                     self._emitline('  Note: This section has relocations against it, but these have NOT been applied to this dump.')
 536                     return
 537
 538     def _init_dwarfinfo(self):
 539         """ Initialize the DWARF info contained in the file and assign it to
 540             self._dwarfinfo.
 541             Leave self._dwarfinfo at None if no DWARF info was found in the file
 542         """
 543         if self._dwarfinfo is not None:
 544             return
 545
 546         if self.elffile.has_dwarf_info():
 547             self._dwarfinfo = self.elffile.get_dwarf_info()
 548         else:
 549             self._dwarfinfo = None
 550
 551     def _dump_debug_info(self):
 552         """ Dump the debugging info section.
 553         """
 554         self._emitline('Contents of the .debug_info section:\n')
 555
 556         # Offset of the .debug_info section in the stream
 557         section_offset = self._dwarfinfo.debug_info_sec.global_offset
 558
 559         for cu in self._dwarfinfo.iter_CUs():
 560             self._emitline('  Compilation Unit @ offset %s:' %
 561                 self._format_hex(cu.cu_offset))
 562             self._emitline('   Length:        %s (%s)' % (
 563                 self._format_hex(cu['unit_length']),
 564                 '%s-bit' % cu.dwarf_format()))
 565             self._emitline('   Version:       %s' % cu['version']),
 566             self._emitline('   Abbrev Offset: %s' % (
 567                 self._format_hex(cu['debug_abbrev_offset']))),
 568             self._emitline('   Pointer Size:  %s' % cu['address_size'])
 569
 570             # The nesting depth of each DIE within the tree of DIEs must be
 571             # displayed. To implement this, a counter is incremented each time
 572             # the current DIE has children, and decremented when a null die is
 573             # encountered. Due to the way the DIE tree is serialized, this will
 574             # correctly reflect the nesting depth
 575             #
 576             die_depth = 0
 577             for die in cu.iter_DIEs():
 578                 self._emitline(' <%s><%x>: Abbrev Number: %s%s' % (
 579                     die_depth,
 580                     die.offset,
 581                     die.abbrev_code,
 582                     (' (%s)' % die.tag) if not die.is_null() else ''))
 583                 if die.is_null():
 584                     die_depth -= 1
 585                     continue
 586
 587                 for attr in itervalues(die.attributes):
 588                     name = attr.name
 589                     # Unknown attribute values are passed-through as integers
 590                     if isinstance(name, int):
 591                         name = 'Unknown AT value: %x' % name
 592                     self._emitline('    <%2x>   %-18s: %s' % (
 593                         attr.offset,
 594                         name,
 595                         describe_attr_value(
 596                             attr, die, section_offset)))
 597
 598                 if die.has_children:
 599                     die_depth += 1
 600
 601         self._emitline()
 602
 603     def _dump_debug_line_programs(self):
 604         """ Dump the (decoded) line programs from .debug_line
 605             The programs are dumped in the order of the CUs they belong to.
 606         """
 607         self._emitline('Decoded dump of debug contents of section .debug_line:\n')
 608
 609         for cu in self._dwarfinfo.iter_CUs():
 610             lineprogram = self._dwarfinfo.line_program_for_CU(cu)
 611
 612             cu_filename = bytes2str(lineprogram['file_entry'][0].name)
 613             if len(lineprogram['include_directory']) > 0:
 614                 dir_index = lineprogram['file_entry'][0].dir_index
 615                 if dir_index > 0:
 616                     dir = lineprogram['include_directory'][dir_index - 1]
 617                 else:
 618                     dir = b'.'
 619                 cu_filename = '%s/%s' % (bytes2str(dir), cu_filename)
 620
 621             self._emitline('CU: %s:' % cu_filename)
 622             self._emitline('File name                            Line number    Starting address')
 623
 624             # Print each state's file, line and address information. For some
 625             # instructions other output is needed to be compatible with
 626             # readelf.
 627             for entry in lineprogram.get_entries():
 628                 state = entry.state
 629                 if state is None:
 630                     # Special handling for commands that don't set a new state
 631                     if entry.command == DW_LNS_set_file:
 632                         file_entry = lineprogram['file_entry'][entry.args[0] - 1]
 633                         if file_entry.dir_index == 0:
 634                             # current directory
 635                             self._emitline('\n./%s:[++]' % (
 636                                 bytes2str(file_entry.name)))
 637                         else:
 638                             self._emitline('\n%s/%s:' % (
 639                                 bytes2str(lineprogram['include_directory'][file_entry.dir_index - 1]),
 640                                 bytes2str(file_entry.name)))
 641                     elif entry.command == DW_LNE_define_file:
 642                         self._emitline('%s:' % (
 643                             bytes2str(lineprogram['include_directory'][entry.args[0].dir_index])))
 644                 elif not state.end_sequence:
 645                     # readelf doesn't print the state after end_sequence
 646                     # instructions. I think it's a bug but to be compatible
 647                     # I don't print them too.
 648                     self._emitline('%-35s  %11d  %18s' % (
 649                         bytes2str(lineprogram['file_entry'][state.file - 1].name),
 650                         state.line,
 651                         '0' if state.address == 0 else
 652                                self._format_hex(state.address)))
 653                 if entry.command == DW_LNS_copy:
 654                     # Another readelf oddity...
 655                     self._emitline()
 656
 657     def _dump_debug_frames(self):
 658         """ Dump the raw frame information from .debug_frame
 659         """
 660         if not self._dwarfinfo.has_CFI():
 661             return
 662         self._emitline('Contents of the .debug_frame section:')
 663
 664         for entry in self._dwarfinfo.CFI_entries():
 665             if isinstance(entry, CIE):
 666                 self._emitline('\n%08x %08x %08x CIE' % (
 667                     entry.offset, entry['length'], entry['CIE_id']))
 668                 self._emitline('  Version:               %d' % entry['version'])
 669                 self._emitline('  Augmentation:          "%s"' % bytes2str(entry['augmentation']))
 670                 self._emitline('  Code alignment factor: %u' % entry['code_alignment_factor'])
 671                 self._emitline('  Data alignment factor: %d' % entry['data_alignment_factor'])
 672                 self._emitline('  Return address column: %d' % entry['return_address_register'])
 673                 self._emitline()
 674             else: # FDE
 675                 self._emitline('\n%08x %08x %08x FDE cie=%08x pc=%08x..%08x' % (
 676                     entry.offset,
 677                     entry['length'],
 678                     entry['CIE_pointer'],
 679                     entry.cie.offset,
 680                     entry['initial_location'],
 681                     entry['initial_location'] + entry['address_range']))
 682
 683             self._emit(describe_CFI_instructions(entry))
 684         self._emitline()
 685
 686     def _dump_debug_frames_interp(self):
 687         """ Dump the interpreted (decoded) frame information from .debug_frame
 688         """
 689         if not self._dwarfinfo.has_CFI():
 690             return
 691
 692         self._emitline('Contents of the .debug_frame section:')
 693
 694         for entry in self._dwarfinfo.CFI_entries():
 695             if isinstance(entry, CIE):
 696                 self._emitline('\n%08x %08x %08x CIE "%s" cf=%d df=%d ra=%d' % (
 697                     entry.offset,
 698                     entry['length'],
 699                     entry['CIE_id'],
 700                     bytes2str(entry['augmentation']),
 701                     entry['code_alignment_factor'],
 702                     entry['data_alignment_factor'],
 703                     entry['return_address_register']))
 704                 ra_regnum = entry['return_address_register']
 705             else: # FDE
 706                 self._emitline('\n%08x %08x %08x FDE cie=%08x pc=%08x..%08x' % (
 707                     entry.offset,
 708                     entry['length'],
 709                     entry['CIE_pointer'],
 710                     entry.cie.offset,
 711                     entry['initial_location'],
 712                     entry['initial_location'] + entry['address_range']))
 713                 ra_regnum = entry.cie['return_address_register']
 714
 715             # Print the heading row for the decoded table
 716             self._emit('   LOC')
 717             self._emit('  ' if entry.structs.address_size == 4 else '          ')
 718             self._emit(' CFA      ')
 719
 720             # Decode the table nad look at the registers it describes.
 721             # We build reg_order here to match readelf's order. In particular,
 722             # registers are sorted by their number, and the register matching
 723             # ra_regnum is always listed last with a special heading.
 724             decoded_table = entry.get_decoded()
 725             reg_order = sorted(ifilter(
 726                 lambda r: r != ra_regnum,
 727                 decoded_table.reg_order))
 728
 729             # Headings for the registers
 730             for regnum in reg_order:
 731                 self._emit('%-6s' % describe_reg_name(regnum))
 732             self._emitline('ra      ')
 733
 734             # Now include ra_regnum in reg_order to print its values similarly
 735             # to the other registers.
 736             reg_order.append(ra_regnum)
 737             for line in decoded_table.table:
 738                 self._emit(self._format_hex(
 739                     line['pc'], fullhex=True, lead0x=False))
 740                 self._emit(' %-9s' % describe_CFI_CFA_rule(line['cfa']))
 741
 742                 for regnum in reg_order:
 743                     if regnum in line:
 744                         s = describe_CFI_register_rule(line[regnum])
 745                     else:
 746                         s = 'u'
 747                     self._emit('%-6s' % s)
 748                 self._emitline()
 749         self._emitline()
 750
 751     def _emit(self, s=''):
 752         """ Emit an object to output
 753         """
 754         self.output.write(str(s))
 755
 756     def _emitline(self, s=''):
 757         """ Emit an object to output, followed by a newline
 758         """
 759         self.output.write(str(s) + '\n')
 760
 761
 762 SCRIPT_DESCRIPTION = 'Display information about the contents of ELF format files'
 763 VERSION_STRING = '%%prog: based on pyelftools %s' % __version__
 764
 765
 766 def main(stream=None):
 767     # parse the command-line arguments and invoke ReadElf
 768     optparser = OptionParser(
 769             usage='usage: %prog [options] <elf-file>',
 770             description=SCRIPT_DESCRIPTION,
 771             add_help_option=False, # -h is a real option of readelf
 772             prog='readelf.py',
 773             version=VERSION_STRING)
 774     optparser.add_option('-d', '--dynamic',
 775             action='store_true', dest='show_dynamic_tags',
 776             help='Display the dynamic section')
 777     optparser.add_option('-H', '--help',
 778             action='store_true', dest='help',
 779             help='Display this information')
 780     optparser.add_option('-h', '--file-header',
 781             action='store_true', dest='show_file_header',
 782             help='Display the ELF file header')
 783     optparser.add_option('-l', '--program-headers', '--segments',
 784             action='store_true', dest='show_program_header',
 785             help='Display the program headers')
 786     optparser.add_option('-S', '--section-headers', '--sections',
 787             action='store_true', dest='show_section_header',
 788             help="Display the sections' headers")
 789     optparser.add_option('-e', '--headers',
 790             action='store_true', dest='show_all_headers',
 791             help='Equivalent to: -h -l -S')
 792     optparser.add_option('-s', '--symbols', '--syms',
 793             action='store_true', dest='show_symbols',
 794             help='Display the symbol table')
 795     optparser.add_option('-r', '--relocs',
 796             action='store_true', dest='show_relocs',
 797             help='Display the relocations (if present)')
 798     optparser.add_option('-x', '--hex-dump',
 799             action='store', dest='show_hex_dump', metavar='<number|name>',
 800             help='Dump the contents of section <number|name> as bytes')
 801     optparser.add_option('-p', '--string-dump',
 802             action='store', dest='show_string_dump', metavar='<number|name>',
 803             help='Dump the contents of section <number|name> as strings')
 804     optparser.add_option('--debug-dump',
 805             action='store', dest='debug_dump_what', metavar='<what>',
 806             help=(
 807                 'Display the contents of DWARF debug sections. <what> can ' +
 808                 'one of {info,decodedline,frames,frames-interp}'))
 809
 810     options, args = optparser.parse_args()
 811
 812     if options.help or len(args) == 0:
 813         optparser.print_help()
 814         sys.exit(0)
 815
 816     if options.show_all_headers:
 817         do_file_header = do_section_header = do_program_header = True
 818     else:
 819         do_file_header = options.show_file_header
 820         do_section_header = options.show_section_header
 821         do_program_header = options.show_program_header
 822
 823     with open(args[0], 'rb') as file:
 824         try:
 825             readelf = ReadElf(file, stream or sys.stdout)
 826             if do_file_header:
 827                 readelf.display_file_header()
 828             if do_section_header:
 829                 readelf.display_section_headers(
 830                         show_heading=not do_file_header)
 831             if do_program_header:
 832                 readelf.display_program_headers(
 833                         show_heading=not do_file_header)
 834             if options.show_dynamic_tags:
 835                 readelf.display_dynamic_tags()
 836             if options.show_symbols:
 837                 readelf.display_symbol_tables()
 838             if options.show_relocs:
 839                 readelf.display_relocations()
 840             if options.show_hex_dump:
 841                 readelf.display_hex_dump(options.show_hex_dump)
 842             if options.show_string_dump:
 843                 readelf.display_string_dump(options.show_string_dump)
 844             if options.debug_dump_what:
 845                 readelf.display_debug_dump(options.debug_dump_what)
 846         except ELFError as ex:
 847             sys.stderr.write('ELF error: %s\n' % ex)
 848             sys.exit(1)
 849
 850
 851 def profile_main():
 852     # Run 'main' redirecting its output to readelfout.txt
 853     # Saves profiling information in readelf.profile
 854     PROFFILE = 'readelf.profile'
 855     import cProfile
 856     cProfile.run('main(open("readelfout.txt", "w"))', PROFFILE)
 857
 858     # Dig in some profiling stats
 859     import pstats
 860     p = pstats.Stats(PROFFILE)
 861     p.sort_stats('cumulative').print_stats(25)
 862
 863
 864 #-------------------------------------------------------------------------------
 865 if __name__ == '__main__':
 866     main()
 867     #profile_main()
 868
 869