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 # If elftools is not installed, maybe we're running from the root or scripts
  15 # dir of the source distribution
  16 try:
  17     import elftools
  18 except ImportError:
  19     sys.path.extend(['.', '..'])
  20
  21 from elftools import __version__
  22 from elftools.common.exceptions import ELFError
  23 from elftools.common.py3compat import (
  24         ifilter, byte2int, bytes2str, itervalues, str2bytes)
  25 from elftools.elf.elffile import ELFFile
  26 from elftools.elf.dynamic import DynamicSection, DynamicSegment
  27 from elftools.elf.enums import ENUM_D_TAG
  28 from elftools.elf.segments import InterpSegment
  29 from elftools.elf.sections import SymbolTableSection
  30 from elftools.elf.relocation import RelocationSection
  31 from elftools.elf.descriptions import (
  32     describe_ei_class, describe_ei_data, describe_ei_version,
  33     describe_ei_osabi, describe_e_type, describe_e_machine,
  34     describe_e_version_numeric, describe_p_type, describe_p_flags,
  35     describe_sh_type, describe_sh_flags,
  36     describe_symbol_type, describe_symbol_bind, describe_symbol_visibility,
  37     describe_symbol_shndx, describe_reloc_type, describe_dyn_tag,
  38     )
  39 from elftools.dwarf.dwarfinfo import DWARFInfo
  40 from elftools.dwarf.descriptions import (
  41     describe_reg_name, describe_attr_value, set_global_machine_arch,
  42     describe_CFI_instructions, describe_CFI_register_rule,
  43     describe_CFI_CFA_rule,
  44     )
  45 from elftools.dwarf.constants import (
  46     DW_LNS_copy, DW_LNS_set_file, DW_LNE_define_file)
  47 from elftools.dwarf.callframe import CIE, FDE
  48
  49
  50 class ReadElf(object):
  51     """ display_* methods are used to emit output into the output stream
  52     """
  53     def __init__(self, file, output):
  54         """ file:
  55                 stream object with the ELF file to read
  56
  57             output:
  58                 output stream to write to
  59         """
  60         self.elffile = ELFFile(file)
  61         self.output = output
  62
  63         # Lazily initialized if a debug dump is requested
  64         self._dwarfinfo = None
  65
  66     def display_file_header(self):
  67         """ Display the ELF file header
  68         """
  69         self._emitline('ELF Header:')
  70         self._emit('  Magic:   ')
  71         self._emitline(' '.join('%2.2x' % byte2int(b)
  72                                     for b in self.elffile.e_ident_raw))
  73         header = self.elffile.header
  74         e_ident = header['e_ident']
  75         self._emitline('  Class:                             %s' %
  76                 describe_ei_class(e_ident['EI_CLASS']))
  77         self._emitline('  Data:                              %s' %
  78                 describe_ei_data(e_ident['EI_DATA']))
  79         self._emitline('  Version:                           %s' %
  80                 describe_ei_version(e_ident['EI_VERSION']))
  81         self._emitline('  OS/ABI:                            %s' %
  82                 describe_ei_osabi(e_ident['EI_OSABI']))
  83         self._emitline('  ABI Version:                       %d' %
  84                 e_ident['EI_ABIVERSION'])
  85         self._emitline('  Type:                              %s' %
  86                 describe_e_type(header['e_type']))
  87         self._emitline('  Machine:                           %s' %
  88                 describe_e_machine(header['e_machine']))
  89         self._emitline('  Version:                           %s' %
  90                 describe_e_version_numeric(header['e_version']))
  91         self._emitline('  Entry point address:               %s' %
  92                 self._format_hex(header['e_entry']))
  93         self._emit('  Start of program headers:          %s' %
  94                 header['e_phoff'])
  95         self._emitline(' (bytes into file)')
  96         self._emit('  Start of section headers:          %s' %
  97                 header['e_shoff'])
  98         self._emitline(' (bytes into file)')
  99         self._emitline('  Flags:                             %s' %
 100                 self._format_hex(header['e_flags']))
 101         self._emitline('  Size of this header:               %s (bytes)' %
 102                 header['e_ehsize'])
 103         self._emitline('  Size of program headers:           %s (bytes)' %
 104                 header['e_phentsize'])
 105         self._emitline('  Number of program headers:         %s' %
 106                 header['e_phnum'])
 107         self._emitline('  Size of section headers:           %s (bytes)' %
 108                 header['e_shentsize'])
 109         self._emitline('  Number of section headers:         %s' %
 110                 header['e_shnum'])
 111         self._emitline('  Section header string table index: %s' %
 112                 header['e_shstrndx'])
 113
 114     def display_program_headers(self, show_heading=True):
 115         """ Display the ELF program headers.
 116             If show_heading is True, displays the heading for this information
 117             (Elf file type is...)
 118         """
 119         self._emitline()
 120         if self.elffile.num_segments() == 0:
 121             self._emitline('There are no program headers in this file.')
 122             return
 123
 124         elfheader = self.elffile.header
 125         if show_heading:
 126             self._emitline('Elf file type is %s' %
 127                 describe_e_type(elfheader['e_type']))
 128             self._emitline('Entry point is %s' %
 129                 self._format_hex(elfheader['e_entry']))
 130             # readelf weirness - why isn't e_phoff printed as hex? (for section
 131             # headers, it is...)
 132             self._emitline('There are %s program headers, starting at offset %s' % (
 133                 elfheader['e_phnum'], elfheader['e_phoff']))
 134             self._emitline()
 135
 136         self._emitline('Program Headers:')
 137
 138         # Now comes the table of program headers with their attributes. Note
 139         # that due to different formatting constraints of 32-bit and 64-bit
 140         # addresses, there are some conditions on elfclass here.
 141         #
 142         # First comes the table heading
 143         #
 144         if self.elffile.elfclass == 32:
 145             self._emitline('  Type           Offset   VirtAddr   PhysAddr   FileSiz MemSiz  Flg Align')
 146         else:
 147             self._emitline('  Type           Offset             VirtAddr           PhysAddr')
 148             self._emitline('                 FileSiz            MemSiz              Flags  Align')
 149
 150         # Now the entries
 151         #
 152         for segment in self.elffile.iter_segments():
 153             self._emit('  %-14s ' % describe_p_type(segment['p_type']))
 154
 155             if self.elffile.elfclass == 32:
 156                 self._emitline('%s %s %s %s %s %-3s %s' % (
 157                     self._format_hex(segment['p_offset'], fieldsize=6),
 158                     self._format_hex(segment['p_vaddr'], fullhex=True),
 159                     self._format_hex(segment['p_paddr'], fullhex=True),
 160                     self._format_hex(segment['p_filesz'], fieldsize=5),
 161                     self._format_hex(segment['p_memsz'], fieldsize=5),
 162                     describe_p_flags(segment['p_flags']),
 163                     self._format_hex(segment['p_align'])))
 164             else: # 64
 165                 self._emitline('%s %s %s' % (
 166                     self._format_hex(segment['p_offset'], fullhex=True),
 167                     self._format_hex(segment['p_vaddr'], fullhex=True),
 168                     self._format_hex(segment['p_paddr'], fullhex=True)))
 169                 self._emitline('                 %s %s  %-3s    %s' % (
 170                     self._format_hex(segment['p_filesz'], fullhex=True),
 171                     self._format_hex(segment['p_memsz'], fullhex=True),
 172                     describe_p_flags(segment['p_flags']),
 173                     # lead0x set to False for p_align, to mimic readelf.
 174                     # No idea why the difference from 32-bit mode :-|
 175                     self._format_hex(segment['p_align'], lead0x=False)))
 176
 177             if isinstance(segment, InterpSegment):
 178                 self._emitline('      [Requesting program interpreter: %s]' %
 179                     bytes2str(segment.get_interp_name()))
 180
 181         # Sections to segments mapping
 182         #
 183         if self.elffile.num_sections() == 0:
 184             # No sections? We're done
 185             return
 186
 187         self._emitline('\n Section to Segment mapping:')
 188         self._emitline('  Segment Sections...')
 189
 190         for nseg, segment in enumerate(self.elffile.iter_segments()):
 191             self._emit('   %2.2d     ' % nseg)
 192
 193             for section in self.elffile.iter_sections():
 194                 if (    not section.is_null() and
 195                         segment.section_in_segment(section)):
 196                     self._emit('%s ' % bytes2str(section.name))
 197
 198             self._emitline('')
 199
 200     def display_section_headers(self, show_heading=True):
 201         """ Display the ELF section headers
 202         """
 203         elfheader = self.elffile.header
 204         if show_heading:
 205             self._emitline('There are %s section headers, starting at offset %s' % (
 206                 elfheader['e_shnum'], self._format_hex(elfheader['e_shoff'])))
 207
 208         self._emitline('\nSection Header%s:' % (
 209             's' if elfheader['e_shnum'] > 1 else ''))
 210
 211         # Different formatting constraints of 32-bit and 64-bit addresses
 212         #
 213         if self.elffile.elfclass == 32:
 214             self._emitline('  [Nr] Name              Type            Addr     Off    Size   ES Flg Lk Inf Al')
 215         else:
 216             self._emitline('  [Nr] Name              Type             Address           Offset')
 217             self._emitline('       Size              EntSize          Flags  Link  Info  Align')
 218
 219         # Now the entries
 220         #
 221         for nsec, section in enumerate(self.elffile.iter_sections()):
 222             self._emit('  [%2u] %-17.17s %-15.15s ' % (
 223                 nsec, bytes2str(section.name), describe_sh_type(section['sh_type'])))
 224
 225             if self.elffile.elfclass == 32:
 226                 self._emitline('%s %s %s %s %3s %2s %3s %2s' % (
 227                     self._format_hex(section['sh_addr'], fieldsize=8, lead0x=False),
 228                     self._format_hex(section['sh_offset'], fieldsize=6, lead0x=False),
 229                     self._format_hex(section['sh_size'], fieldsize=6, lead0x=False),
 230                     self._format_hex(section['sh_entsize'], fieldsize=2, lead0x=False),
 231                     describe_sh_flags(section['sh_flags']),
 232                     section['sh_link'], section['sh_info'],
 233                     section['sh_addralign']))
 234             else: # 64
 235                 self._emitline(' %s  %s' % (
 236                     self._format_hex(section['sh_addr'], fullhex=True, lead0x=False),
 237                     self._format_hex(section['sh_offset'],
 238                         fieldsize=16 if section['sh_offset'] > 0xffffffff else 8,
 239                         lead0x=False)))
 240                 self._emitline('       %s  %s %3s      %2s   %3s     %s' % (
 241                     self._format_hex(section['sh_size'], fullhex=True, lead0x=False),
 242                     self._format_hex(section['sh_entsize'], fullhex=True, lead0x=False),
 243                     describe_sh_flags(section['sh_flags']),
 244                     section['sh_link'], section['sh_info'],
 245                     section['sh_addralign']))
 246
 247         self._emitline('Key to Flags:')
 248         self._emit('  W (write), A (alloc), X (execute), M (merge), S (strings)')
 249         if self.elffile['e_machine'] in ('EM_X86_64', 'EM_L10M'):
 250             self._emitline(', l (large)')
 251         else:
 252             self._emitline()
 253         self._emitline('  I (info), L (link order), G (group), T (TLS), E (exclude), x (unknown)')
 254         self._emitline('  O (extra OS processing required) o (OS specific), p (processor specific)')
 255
 256     def display_symbol_tables(self):
 257         """ Display the symbol tables contained in the file
 258         """
 259         for section in self.elffile.iter_sections():
 260             if not isinstance(section, SymbolTableSection):
 261                 continue
 262
 263             if section['sh_entsize'] == 0:
 264                 self._emitline("\nSymbol table '%s' has a sh_entsize of zero!" % (
 265                     bytes2str(section.name)))
 266                 continue
 267
 268             self._emitline("\nSymbol table '%s' contains %s entries:" % (
 269                 bytes2str(section.name), section.num_symbols()))
 270
 271             if self.elffile.elfclass == 32:
 272                 self._emitline('   Num:    Value  Size Type    Bind   Vis      Ndx Name')
 273             else: # 64
 274                 self._emitline('   Num:    Value          Size Type    Bind   Vis      Ndx Name')
 275
 276             for nsym, symbol in enumerate(section.iter_symbols()):
 277                 # symbol names are truncated to 25 chars, similarly to readelf
 278                 self._emitline('%6d: %s %5d %-7s %-6s %-7s %4s %.25s' % (
 279                     nsym,
 280                     self._format_hex(symbol['st_value'], fullhex=True, lead0x=False),
 281                     symbol['st_size'],
 282                     describe_symbol_type(symbol['st_info']['type']),
 283                     describe_symbol_bind(symbol['st_info']['bind']),
 284                     describe_symbol_visibility(symbol['st_other']['visibility']),
 285                     describe_symbol_shndx(symbol['st_shndx']),
 286                     bytes2str(symbol.name)))
 287
 288     def display_dynamic_tags(self):
 289         """ Display the dynamic tags contained in the file
 290         """
 291         for section in self.elffile.iter_sections():
 292             if not isinstance(section, DynamicSection):
 293                 continue
 294
 295             self._emitline("\nDynamic section at offset %s contains %s entries:" % (
 296                 self._format_hex(section['sh_offset']),
 297                 section.num_tags()))
 298             self._emitline("  Tag        Type                         Name/Value")
 299
 300             padding = 20 + (8 if self.elffile.elfclass == 32 else 0)
 301             for tag in section.iter_tags():
 302                 if tag.entry.d_tag == 'DT_NEEDED':
 303                     parsed = 'Shared library: [%s]' % tag.needed
 304                 elif tag.entry.d_tag == 'DT_RPATH':
 305                     parsed = 'Library rpath: [%s]' % tag.rpath
 306                 elif tag.entry.d_tag == 'DT_RUNPATH':
 307                     parsed = 'Library runpath: [%s]' % tag.runpath
 308                 elif tag.entry.d_tag == 'DT_SONAME':
 309                     parsed = 'Library soname: [%s]' % tag.soname
 310                 elif (tag.entry.d_tag.endswith('SZ') or
 311                       tag.entry.d_tag.endswith('ENT')):
 312                     parsed = '%i (bytes)' % tag['d_val']
 313                 elif tag.entry.d_tag.endswith('NUM'):
 314                     parsed = '%i' % tag['d_val']
 315                 elif tag.entry.d_tag == 'DT_PLTREL':
 316                     s = describe_dyn_tag(tag.entry.d_val)
 317                     if s.startswith('DT_'):
 318                         s = s[3:]
 319                     parsed = '%s' % s
 320                 else:
 321                     parsed = '%#x' % tag['d_val']
 322
 323                 self._emitline(" %s %-*s %s" % (
 324                     self._format_hex(ENUM_D_TAG.get(tag.entry.d_tag, tag.entry.d_tag),
 325                         fullhex=True, lead0x=True),
 326                     padding,
 327                     '(%s)' % (tag.entry.d_tag[3:],),
 328                     parsed))
 329
 330     def display_relocations(self):
 331         """ Display the relocations contained in the file
 332         """
 333         has_relocation_sections = False
 334         for section in self.elffile.iter_sections():
 335             if not isinstance(section, RelocationSection):
 336                 continue
 337
 338             has_relocation_sections = True
 339             self._emitline("\nRelocation section '%s' at offset %s contains %s entries:" % (
 340                 bytes2str(section.name),
 341                 self._format_hex(section['sh_offset']),
 342                 section.num_relocations()))
 343             if section.is_RELA():
 344                 self._emitline("  Offset          Info           Type           Sym. Value    Sym. Name + Addend")
 345             else:
 346                 self._emitline(" Offset     Info    Type            Sym.Value  Sym. Name")
 347
 348             # The symbol table section pointed to in sh_link
 349             symtable = self.elffile.get_section(section['sh_link'])
 350
 351             for rel in section.iter_relocations():
 352                 hexwidth = 8 if self.elffile.elfclass == 32 else 12
 353                 self._emit('%s  %s %-17.17s' % (
 354                     self._format_hex(rel['r_offset'],
 355                         fieldsize=hexwidth, lead0x=False),
 356                     self._format_hex(rel['r_info'],
 357                         fieldsize=hexwidth, lead0x=False),
 358                     describe_reloc_type(
 359                         rel['r_info_type'], self.elffile)))
 360
 361                 if rel['r_info_sym'] == 0:
 362                     self._emitline()
 363                     continue
 364
 365                 symbol = symtable.get_symbol(rel['r_info_sym'])
 366                 # Some symbols have zero 'st_name', so instead what's used is
 367                 # the name of the section they point at
 368                 if symbol['st_name'] == 0:
 369                     symsec = self.elffile.get_section(symbol['st_shndx'])
 370                     symbol_name = symsec.name
 371                 else:
 372                     symbol_name = symbol.name
 373                 self._emit(' %s %s%22.22s' % (
 374                     self._format_hex(
 375                         symbol['st_value'],
 376                         fullhex=True, lead0x=False),
 377                     '  ' if self.elffile.elfclass == 32 else '',
 378                     bytes2str(symbol_name)))
 379                 if section.is_RELA():
 380                     self._emit(' %s %x' % (
 381                         '+' if rel['r_addend'] >= 0 else '-',
 382                         abs(rel['r_addend'])))
 383                 self._emitline()
 384
 385         if not has_relocation_sections:
 386             self._emitline('\nThere are no relocations in this file.')
 387
 388     def display_hex_dump(self, section_spec):
 389         """ Display a hex dump of a section. section_spec is either a section
 390             number or a name.
 391         """
 392         section = self._section_from_spec(section_spec)
 393         if section is None:
 394             self._emitline("Section '%s' does not exist in the file!" % (
 395                 section_spec))
 396             return
 397
 398         self._emitline("\nHex dump of section '%s':" % bytes2str(section.name))
 399         self._note_relocs_for_section(section)
 400         addr = section['sh_addr']
 401         data = section.data()
 402         dataptr = 0
 403
 404         while dataptr < len(data):
 405             bytesleft = len(data) - dataptr
 406             # chunks of 16 bytes per line
 407             linebytes = 16 if bytesleft > 16 else bytesleft
 408
 409             self._emit('  %s ' % self._format_hex(addr, fieldsize=8))
 410             for i in range(16):
 411                 if i < linebytes:
 412                     self._emit('%2.2x' % byte2int(data[dataptr + i]))
 413                 else:
 414                     self._emit('  ')
 415                 if i % 4 == 3:
 416                     self._emit(' ')
 417
 418             for i in range(linebytes):
 419                 c = data[dataptr + i : dataptr + i + 1]
 420                 if byte2int(c[0]) >= 32 and byte2int(c[0]) < 0x7f:
 421                     self._emit(bytes2str(c))
 422                 else:
 423                     self._emit(bytes2str(b'.'))
 424
 425             self._emitline()
 426             addr += linebytes
 427             dataptr += linebytes
 428
 429         self._emitline()
 430
 431     def display_string_dump(self, section_spec):
 432         """ Display a strings dump of a section. section_spec is either a
 433             section number or a name.
 434         """
 435         section = self._section_from_spec(section_spec)
 436         if section is None:
 437             self._emitline("Section '%s' does not exist in the file!" % (
 438                 section_spec))
 439             return
 440
 441         self._emitline("\nString dump of section '%s':" % bytes2str(section.name))
 442
 443         found = False
 444         data = section.data()
 445         dataptr = 0
 446
 447         while dataptr < len(data):
 448             while ( dataptr < len(data) and
 449                     not (32 <= byte2int(data[dataptr]) <= 127)):
 450                 dataptr += 1
 451
 452             if dataptr >= len(data):
 453                 break
 454
 455             endptr = dataptr
 456             while endptr < len(data) and byte2int(data[endptr]) != 0:
 457                 endptr += 1
 458
 459             found = True
 460             self._emitline('  [%6x]  %s' % (
 461                 dataptr, bytes2str(data[dataptr:endptr])))
 462
 463             dataptr = endptr
 464
 465         if not found:
 466             self._emitline('  No strings found in this section.')
 467         else:
 468             self._emitline()
 469
 470     def display_debug_dump(self, dump_what):
 471         """ Dump a DWARF section
 472         """
 473         self._init_dwarfinfo()
 474         if self._dwarfinfo is None:
 475             return
 476
 477         set_global_machine_arch(self.elffile.get_machine_arch())
 478
 479         if dump_what == 'info':
 480             self._dump_debug_info()
 481         elif dump_what == 'decodedline':
 482             self._dump_debug_line_programs()
 483         elif dump_what == 'frames':
 484             self._dump_debug_frames()
 485         elif dump_what == 'frames-interp':
 486             self._dump_debug_frames_interp()
 487         else:
 488             self._emitline('debug dump not yet supported for "%s"' % dump_what)
 489
 490     def _format_hex(self, addr, fieldsize=None, fullhex=False, lead0x=True):
 491         """ Format an address into a hexadecimal string.
 492
 493             fieldsize:
 494                 Size of the hexadecimal field (with leading zeros to fit the
 495                 address into. For example with fieldsize=8, the format will
 496                 be %08x
 497                 If None, the minimal required field size will be used.
 498
 499             fullhex:
 500                 If True, override fieldsize to set it to the maximal size
 501                 needed for the elfclass
 502
 503             lead0x:
 504                 If True, leading 0x is added
 505         """
 506         s = '0x' if lead0x else ''
 507         if fullhex:
 508             fieldsize = 8 if self.elffile.elfclass == 32 else 16
 509         if fieldsize is None:
 510             field = '%x'
 511         else:
 512             field = '%' + '0%sx' % fieldsize
 513         return s + field % addr
 514
 515     def _section_from_spec(self, spec):
 516         """ Retrieve a section given a "spec" (either number or name).
 517             Return None if no such section exists in the file.
 518         """
 519         try:
 520             num = int(spec)
 521             if num < self.elffile.num_sections():
 522                 return self.elffile.get_section(num)
 523             else:
 524                 return None
 525         except ValueError:
 526             # Not a number. Must be a name then
 527             return self.elffile.get_section_by_name(str2bytes(spec))
 528
 529     def _note_relocs_for_section(self, section):
 530         """ If there are relocation sections pointing to the givne section,
 531             emit a note about it.
 532         """
 533         for relsec in self.elffile.iter_sections():
 534             if isinstance(relsec, RelocationSection):
 535                 info_idx = relsec['sh_info']
 536                 if self.elffile.get_section(info_idx) == section:
 537                     self._emitline('  Note: This section has relocations against it, but these have NOT been applied to this dump.')
 538                     return
 539
 540     def _init_dwarfinfo(self):
 541         """ Initialize the DWARF info contained in the file and assign it to
 542             self._dwarfinfo.
 543             Leave self._dwarfinfo at None if no DWARF info was found in the file
 544         """
 545         if self._dwarfinfo is not None:
 546             return
 547
 548         if self.elffile.has_dwarf_info():
 549             self._dwarfinfo = self.elffile.get_dwarf_info()
 550         else:
 551             self._dwarfinfo = None
 552
 553     def _dump_debug_info(self):
 554         """ Dump the debugging info section.
 555         """
 556         self._emitline('Contents of the .debug_info section:\n')
 557
 558         # Offset of the .debug_info section in the stream
 559         section_offset = self._dwarfinfo.debug_info_sec.global_offset
 560
 561         for cu in self._dwarfinfo.iter_CUs():
 562             self._emitline('  Compilation Unit @ offset %s:' %
 563                 self._format_hex(cu.cu_offset))
 564             self._emitline('   Length:        %s (%s)' % (
 565                 self._format_hex(cu['unit_length']),
 566                 '%s-bit' % cu.dwarf_format()))
 567             self._emitline('   Version:       %s' % cu['version']),
 568             self._emitline('   Abbrev Offset: %s' % (
 569                 self._format_hex(cu['debug_abbrev_offset']))),
 570             self._emitline('   Pointer Size:  %s' % cu['address_size'])
 571
 572             # The nesting depth of each DIE within the tree of DIEs must be
 573             # displayed. To implement this, a counter is incremented each time
 574             # the current DIE has children, and decremented when a null die is
 575             # encountered. Due to the way the DIE tree is serialized, this will
 576             # correctly reflect the nesting depth
 577             #
 578             die_depth = 0
 579             for die in cu.iter_DIEs():
 580                 self._emitline(' <%s><%x>: Abbrev Number: %s%s' % (
 581                     die_depth,
 582                     die.offset,
 583                     die.abbrev_code,
 584                     (' (%s)' % die.tag) if not die.is_null() else ''))
 585                 if die.is_null():
 586                     die_depth -= 1
 587                     continue
 588
 589                 for attr in itervalues(die.attributes):
 590                     name = attr.name
 591                     # Unknown attribute values are passed-through as integers
 592                     if isinstance(name, int):
 593                         name = 'Unknown AT value: %x' % name
 594                     self._emitline('    <%2x>   %-18s: %s' % (
 595                         attr.offset,
 596                         name,
 597                         describe_attr_value(
 598                             attr, die, section_offset)))
 599
 600                 if die.has_children:
 601                     die_depth += 1
 602
 603         self._emitline()
 604
 605     def _dump_debug_line_programs(self):
 606         """ Dump the (decoded) line programs from .debug_line
 607             The programs are dumped in the order of the CUs they belong to.
 608         """
 609         self._emitline('Decoded dump of debug contents of section .debug_line:\n')
 610
 611         for cu in self._dwarfinfo.iter_CUs():
 612             lineprogram = self._dwarfinfo.line_program_for_CU(cu)
 613
 614             cu_filename = bytes2str(lineprogram['file_entry'][0].name)
 615             if len(lineprogram['include_directory']) > 0:
 616                 dir_index = lineprogram['file_entry'][0].dir_index
 617                 if dir_index > 0:
 618                     dir = lineprogram['include_directory'][dir_index - 1]
 619                 else:
 620                     dir = '.'
 621                 cu_filename = '%s/%s' % (bytes2str(dir), cu_filename)
 622
 623             self._emitline('CU: %s:' % cu_filename)
 624             self._emitline('File name                            Line number    Starting address')
 625
 626             # Print each state's file, line and address information. For some
 627             # instructions other output is needed to be compatible with
 628             # readelf.
 629             for entry in lineprogram.get_entries():
 630                 state = entry.state
 631                 if state is None:
 632                     # Special handling for commands that don't set a new state
 633                     if entry.command == DW_LNS_set_file:
 634                         file_entry = lineprogram['file_entry'][entry.args[0] - 1]
 635                         if file_entry.dir_index == 0:
 636                             # current directory
 637                             self._emitline('\n./%s:[++]' % (
 638                                 bytes2str(file_entry.name)))
 639                         else:
 640                             self._emitline('\n%s/%s:' % (
 641                                 bytes2str(lineprogram['include_directory'][file_entry.dir_index - 1]),
 642                                 bytes2str(file_entry.name)))
 643                     elif entry.command == DW_LNE_define_file:
 644                         self._emitline('%s:' % (
 645                             bytes2str(lineprogram['include_directory'][entry.args[0].dir_index])))
 646                 elif not state.end_sequence:
 647                     # readelf doesn't print the state after end_sequence
 648                     # instructions. I think it's a bug but to be compatible
 649                     # I don't print them too.
 650                     self._emitline('%-35s  %11d  %18s' % (
 651                         bytes2str(lineprogram['file_entry'][state.file - 1].name),
 652                         state.line,
 653                         '0' if state.address == 0 else
 654                                self._format_hex(state.address)))
 655                 if entry.command == DW_LNS_copy:
 656                     # Another readelf oddity...
 657                     self._emitline()
 658
 659     def _dump_debug_frames(self):
 660         """ Dump the raw frame information from .debug_frame
 661         """
 662         if not self._dwarfinfo.has_CFI():
 663             return
 664         self._emitline('Contents of the .debug_frame section:')
 665
 666         for entry in self._dwarfinfo.CFI_entries():
 667             if isinstance(entry, CIE):
 668                 self._emitline('\n%08x %08x %08x CIE' % (
 669                     entry.offset, entry['length'], entry['CIE_id']))
 670                 self._emitline('  Version:               %d' % entry['version'])
 671                 self._emitline('  Augmentation:          "%s"' % bytes2str(entry['augmentation']))
 672                 self._emitline('  Code alignment factor: %u' % entry['code_alignment_factor'])
 673                 self._emitline('  Data alignment factor: %d' % entry['data_alignment_factor'])
 674                 self._emitline('  Return address column: %d' % entry['return_address_register'])
 675                 self._emitline()
 676             else: # FDE
 677                 self._emitline('\n%08x %08x %08x FDE cie=%08x pc=%08x..%08x' % (
 678                     entry.offset,
 679                     entry['length'],
 680                     entry['CIE_pointer'],
 681                     entry.cie.offset,
 682                     entry['initial_location'],
 683                     entry['initial_location'] + entry['address_range']))
 684
 685             self._emit(describe_CFI_instructions(entry))
 686         self._emitline()
 687
 688     def _dump_debug_frames_interp(self):
 689         """ Dump the interpreted (decoded) frame information from .debug_frame
 690         """
 691         if not self._dwarfinfo.has_CFI():
 692             return
 693
 694         self._emitline('Contents of the .debug_frame section:')
 695
 696         for entry in self._dwarfinfo.CFI_entries():
 697             if isinstance(entry, CIE):
 698                 self._emitline('\n%08x %08x %08x CIE "%s" cf=%d df=%d ra=%d' % (
 699                     entry.offset,
 700                     entry['length'],
 701                     entry['CIE_id'],
 702                     bytes2str(entry['augmentation']),
 703                     entry['code_alignment_factor'],
 704                     entry['data_alignment_factor'],
 705                     entry['return_address_register']))
 706                 ra_regnum = entry['return_address_register']
 707             else: # FDE
 708                 self._emitline('\n%08x %08x %08x FDE cie=%08x pc=%08x..%08x' % (
 709                     entry.offset,
 710                     entry['length'],
 711                     entry['CIE_pointer'],
 712                     entry.cie.offset,
 713                     entry['initial_location'],
 714                     entry['initial_location'] + entry['address_range']))
 715                 ra_regnum = entry.cie['return_address_register']
 716
 717             # Print the heading row for the decoded table
 718             self._emit('   LOC')
 719             self._emit('  ' if entry.structs.address_size == 4 else '          ')
 720             self._emit(' CFA      ')
 721
 722             # Decode the table nad look at the registers it describes.
 723             # We build reg_order here to match readelf's order. In particular,
 724             # registers are sorted by their number, and the register matching
 725             # ra_regnum is always listed last with a special heading.
 726             decoded_table = entry.get_decoded()
 727             reg_order = sorted(ifilter(
 728                 lambda r: r != ra_regnum,
 729                 decoded_table.reg_order))
 730
 731             # Headings for the registers
 732             for regnum in reg_order:
 733                 self._emit('%-6s' % describe_reg_name(regnum))
 734             self._emitline('ra      ')
 735
 736             # Now include ra_regnum in reg_order to print its values similarly
 737             # to the other registers.
 738             reg_order.append(ra_regnum)
 739             for line in decoded_table.table:
 740                 self._emit(self._format_hex(
 741                     line['pc'], fullhex=True, lead0x=False))
 742                 self._emit(' %-9s' % describe_CFI_CFA_rule(line['cfa']))
 743
 744                 for regnum in reg_order:
 745                     if regnum in line:
 746                         s = describe_CFI_register_rule(line[regnum])
 747                     else:
 748                         s = 'u'
 749                     self._emit('%-6s' % s)
 750                 self._emitline()
 751         self._emitline()
 752
 753     def _emit(self, s=''):
 754         """ Emit an object to output
 755         """
 756         self.output.write(str(s))
 757
 758     def _emitline(self, s=''):
 759         """ Emit an object to output, followed by a newline
 760         """
 761         self.output.write(str(s) + '\n')
 762
 763
 764 SCRIPT_DESCRIPTION = 'Display information about the contents of ELF format files'
 765 VERSION_STRING = '%%prog: based on pyelftools %s' % __version__
 766
 767
 768 def main(stream=None):
 769     # parse the command-line arguments and invoke ReadElf
 770     optparser = OptionParser(
 771             usage='usage: %prog [options] <elf-file>',
 772             description=SCRIPT_DESCRIPTION,
 773             add_help_option=False, # -h is a real option of readelf
 774             prog='readelf.py',
 775             version=VERSION_STRING)
 776     optparser.add_option('-d', '--dynamic',
 777             action='store_true', dest='show_dynamic_tags',
 778             help='Display the dynamic section')
 779     optparser.add_option('-H', '--help',
 780             action='store_true', dest='help',
 781             help='Display this information')
 782     optparser.add_option('-h', '--file-header',
 783             action='store_true', dest='show_file_header',
 784             help='Display the ELF file header')
 785     optparser.add_option('-l', '--program-headers', '--segments',
 786             action='store_true', dest='show_program_header',
 787             help='Display the program headers')
 788     optparser.add_option('-S', '--section-headers', '--sections',
 789             action='store_true', dest='show_section_header',
 790             help="Display the sections' headers")
 791     optparser.add_option('-e', '--headers',
 792             action='store_true', dest='show_all_headers',
 793             help='Equivalent to: -h -l -S')
 794     optparser.add_option('-s', '--symbols', '--syms',
 795             action='store_true', dest='show_symbols',
 796             help='Display the symbol table')
 797     optparser.add_option('-r', '--relocs',
 798             action='store_true', dest='show_relocs',
 799             help='Display the relocations (if present)')
 800     optparser.add_option('-x', '--hex-dump',
 801             action='store', dest='show_hex_dump', metavar='<number|name>',
 802             help='Dump the contents of section <number|name> as bytes')
 803     optparser.add_option('-p', '--string-dump',
 804             action='store', dest='show_string_dump', metavar='<number|name>',
 805             help='Dump the contents of section <number|name> as strings')
 806     optparser.add_option('--debug-dump',
 807             action='store', dest='debug_dump_what', metavar='<what>',
 808             help=(
 809                 'Display the contents of DWARF debug sections. <what> can ' +
 810                 'one of {info,decodedline,frames,frames-interp}'))
 811
 812     options, args = optparser.parse_args()
 813
 814     if options.help or len(args) == 0:
 815         optparser.print_help()
 816         sys.exit(0)
 817
 818     if options.show_all_headers:
 819         do_file_header = do_section_header = do_program_header = True
 820     else:
 821         do_file_header = options.show_file_header
 822         do_section_header = options.show_section_header
 823         do_program_header = options.show_program_header
 824
 825     with open(args[0], 'rb') as file:
 826         try:
 827             readelf = ReadElf(file, stream or sys.stdout)
 828             if do_file_header:
 829                 readelf.display_file_header()
 830             if do_section_header:
 831                 readelf.display_section_headers(
 832                         show_heading=not do_file_header)
 833             if do_program_header:
 834                 readelf.display_program_headers(
 835                         show_heading=not do_file_header)
 836             if options.show_dynamic_tags:
 837                 readelf.display_dynamic_tags()
 838             if options.show_symbols:
 839                 readelf.display_symbol_tables()
 840             if options.show_relocs:
 841                 readelf.display_relocations()
 842             if options.show_hex_dump:
 843                 readelf.display_hex_dump(options.show_hex_dump)
 844             if options.show_string_dump:
 845                 readelf.display_string_dump(options.show_string_dump)
 846             if options.debug_dump_what:
 847                 readelf.display_debug_dump(options.debug_dump_what)
 848         except ELFError as ex:
 849             sys.stderr.write('ELF error: %s\n' % ex)
 850             sys.exit(1)
 851
 852
 853 def profile_main():
 854     # Run 'main' redirecting its output to readelfout.txt
 855     # Saves profiling information in readelf.profile
 856     PROFFILE = 'readelf.profile'
 857     import cProfile
 858     cProfile.run('main(open("readelfout.txt", "w"))', PROFFILE)
 859
 860     # Dig in some profiling stats
 861     import pstats
 862     p = pstats.Stats(PROFFILE)
 863     p.sort_stats('cumulative').print_stats(25)
 864
 865
 866 #-------------------------------------------------------------------------------
 867 if __name__ == '__main__':
 868     main()
 869     #profile_main()
 870
 871