#-------------------------------------------------------------------------------
import os, sys
from optparse import OptionParser
+from itertools import ifilter
import string
-
# If elftools is not installed, maybe we're running from the root or scripts
# dir of the source distribution
-#
try:
import elftools
except ImportError:
from elftools.common.exceptions import ELFError
from elftools.elf.elffile import ELFFile
from elftools.elf.segments import InterpSegment
-from elftools.elf.sections import SymbolTableSection, RelocationSection
+from elftools.elf.sections import SymbolTableSection
+from elftools.elf.relocation import RelocationSection
from elftools.elf.descriptions import (
describe_ei_class, describe_ei_data, describe_ei_version,
describe_ei_osabi, describe_e_type, describe_e_machine,
describe_symbol_type, describe_symbol_bind, describe_symbol_visibility,
describe_symbol_shndx, describe_reloc_type,
)
+from elftools.dwarf.dwarfinfo import DWARFInfo
+from elftools.dwarf.descriptions import (
+ describe_reg_name, describe_attr_value, set_global_machine_arch,
+ describe_CFI_instructions, describe_CFI_register_rule,
+ describe_CFI_CFA_rule,
+ )
+from elftools.dwarf.constants import (
+ DW_LNS_copy, DW_LNS_set_file, DW_LNE_define_file)
+from elftools.dwarf.callframe import CIE, FDE
class ReadElf(object):
"""
self.elffile = ELFFile(file)
self.output = output
+
+ # Lazily initialized if a debug dump is requested
+ self._dwarfinfo = None
def display_file_header(self):
""" Display the ELF file header
section['sh_addralign']))
self._emitline('Key to Flags:')
- self._emitline(' W (write), A (alloc), X (execute), M (merge), S (strings)')
- self._emitline(' I (info), L (link order), G (group), x (unknown)')
+ self._emit(' W (write), A (alloc), X (execute), M (merge), S (strings)')
+ if self.elffile['e_machine'] in ('EM_X86_64', 'EM_L10M'):
+ self._emitline(', l (large)')
+ else:
+ self._emitline()
+ self._emitline(' I (info), L (link order), G (group), T (TLS), E (exclude), x (unknown)')
self._emitline(' O (extra OS processing required) o (OS specific), p (processor specific)')
def display_symbol_tables(self):
self._format_hex(rel['r_info'],
fieldsize=hexwidth, lead0x=False),
describe_reloc_type(
- rel['r_info_type'], self.elffile['e_machine'])))
+ rel['r_info_type'], self.elffile)))
if rel['r_info_sym'] == 0:
self._emitline()
symbol_name = symsec.name
else:
symbol_name = symbol.name
- self._emit(' %s %s%s' % (
+ self._emit(' %s %s%22.22s' % (
self._format_hex(
symbol['st_value'],
fullhex=True, lead0x=False),
else:
self._emitline()
+ def display_debug_dump(self, dump_what):
+ """ Dump a DWARF section
+ """
+ self._init_dwarfinfo()
+ if self._dwarfinfo is None:
+ return
+
+ set_global_machine_arch(self.elffile.get_machine_arch())
+
+ if dump_what == 'info':
+ self._dump_debug_info()
+ elif dump_what == 'decodedline':
+ self._dump_debug_line_programs()
+ elif dump_what == 'frames':
+ self._dump_debug_frames()
+ elif dump_what == 'frames-interp':
+ self._dump_debug_frames_interp()
+ else:
+ self._emitline('debug dump not yet supported for "%s"' % dump_what)
+
def _format_hex(self, addr, fieldsize=None, fullhex=False, lead0x=True):
""" Format an address into a hexadecimal string.
self._emitline(' Note: This section has relocations against it, but these have NOT been applied to this dump.')
return
+ def _init_dwarfinfo(self):
+ """ Initialize the DWARF info contained in the file and assign it to
+ self._dwarfinfo.
+ Leave self._dwarfinfo at None if no DWARF info was found in the file
+ """
+ if self._dwarfinfo is not None:
+ return
+
+ if self.elffile.has_dwarf_info():
+ self._dwarfinfo = self.elffile.get_dwarf_info()
+ else:
+ self._dwarfinfo = None
+
+ def _dump_debug_info(self):
+ """ Dump the debugging info section.
+ """
+ self._emitline('Contents of the .debug_info section:\n')
+
+ # Offset of the .debug_info section in the stream
+ section_offset = self._dwarfinfo.debug_info_sec.global_offset
+
+ for cu in self._dwarfinfo.iter_CUs():
+ self._emitline(' Compilation Unit @ offset %s:' %
+ self._format_hex(cu.cu_offset))
+ self._emitline(' Length: %s (%s)' % (
+ self._format_hex(cu['unit_length']),
+ '%s-bit' % cu.dwarf_format()))
+ self._emitline(' Version: %s' % cu['version']),
+ self._emitline(' Abbrev Offset: %s' % cu['debug_abbrev_offset']),
+ self._emitline(' Pointer Size: %s' % cu['address_size'])
+
+ # The nesting depth of each DIE within the tree of DIEs must be
+ # displayed. To implement this, a counter is incremented each time
+ # the current DIE has children, and decremented when a null die is
+ # encountered. Due to the way the DIE tree is serialized, this will
+ # correctly reflect the nesting depth
+ #
+ die_depth = 0
+ for die in cu.iter_DIEs():
+ if die.is_null():
+ die_depth -= 1
+ continue
+ self._emitline(' <%s><%x>: Abbrev Number: %s (%s)' % (
+ die_depth,
+ die.offset,
+ die.abbrev_code,
+ die.tag))
+
+ for attr in die.attributes.itervalues():
+ name = attr.name
+ # Unknown attribute values are passed-through as integers
+ if isinstance(name, int):
+ name = 'Unknown AT value: %x' % name
+ self._emitline(' <%2x> %-18s: %s' % (
+ attr.offset,
+ name,
+ describe_attr_value(
+ attr, die, section_offset)))
+
+ if die.has_children:
+ die_depth += 1
+
+ self._emitline()
+
+ def _dump_debug_line_programs(self):
+ """ Dump the (decoded) line programs from .debug_line
+ The programs are dumped in the order of the CUs they belong to.
+ """
+ self._emitline('Decoded dump of debug contents of section .debug_line:\n')
+
+ for cu in self._dwarfinfo.iter_CUs():
+ lineprogram = self._dwarfinfo.line_program_for_CU(cu)
+
+ cu_filename = ''
+ if len(lineprogram['include_directory']) > 0:
+ cu_filename = '%s/%s' % (
+ lineprogram['include_directory'][0],
+ lineprogram['file_entry'][0].name)
+ else:
+ cu_filename = lineprogram['file_entry'][0].name
+
+ self._emitline('CU: %s:' % cu_filename)
+ self._emitline('File name Line number Starting address')
+
+ # Print each state's file, line and address information. For some
+ # instructions other output is needed to be compatible with
+ # readelf.
+ for entry in lineprogram.get_entries():
+ state = entry.state
+ if state is None:
+ # Special handling for commands that don't set a new state
+ if entry.command == DW_LNS_set_file:
+ file_entry = lineprogram['file_entry'][entry.args[0] - 1]
+ if file_entry.dir_index == 0:
+ # current directory
+ self._emitline('\n./%s:[++]' % (
+ file_entry.name))
+ else:
+ self._emitline('\n%s/%s:' % (
+ lineprogram['include_directory'][file_entry.dir_index - 1],
+ file_entry.name))
+ elif entry.command == DW_LNE_define_file:
+ self._emitline('%s:' % (
+ lineprogram['include_directory'][entry.args[0].dir_index]))
+ elif not state.end_sequence:
+ # readelf doesn't print the state after end_sequence
+ # instructions. I think it's a bug but to be compatible
+ # I don't print them too.
+ self._emitline('%-35s %11d %18s' % (
+ lineprogram['file_entry'][state.file - 1].name,
+ state.line,
+ '0' if state.address == 0 else
+ self._format_hex(state.address)))
+ if entry.command == DW_LNS_copy:
+ # Another readelf oddity...
+ self._emitline()
+
+ def _dump_debug_frames(self):
+ """ Dump the raw frame information from .debug_frame
+ """
+ if not self._dwarfinfo.has_CFI():
+ return
+ self._emitline('Contents of the .debug_frame section:')
+
+ for entry in self._dwarfinfo.CFI_entries():
+ if isinstance(entry, CIE):
+ self._emitline('\n%08x %08x %08x CIE' % (
+ entry.offset, entry['length'], entry['CIE_id']))
+ self._emitline(' Version: %d' % entry['version'])
+ self._emitline(' Augmentation: "%s"' % entry['augmentation'])
+ self._emitline(' Code alignment factor: %u' % entry['code_alignment_factor'])
+ self._emitline(' Data alignment factor: %d' % entry['data_alignment_factor'])
+ self._emitline(' Return address column: %d' % entry['return_address_register'])
+ self._emitline()
+ else: # FDE
+ self._emitline('\n%08x %08x %08x FDE cie=%08x pc=%08x..%08x' % (
+ entry.offset,
+ entry['length'],
+ entry['CIE_pointer'],
+ entry.cie.offset,
+ entry['initial_location'],
+ entry['initial_location'] + entry['address_range']))
+
+ self._emit(describe_CFI_instructions(entry))
+ self._emitline()
+
+ def _dump_debug_frames_interp(self):
+ """ Dump the interpreted (decoded) frame information from .debug_frame
+ """
+ if not self._dwarfinfo.has_CFI():
+ return
+
+ self._emitline('Contents of the .debug_frame section:')
+
+ for entry in self._dwarfinfo.CFI_entries():
+ if isinstance(entry, CIE):
+ self._emitline('\n%08x %08x %08x CIE "%s" cf=%d df=%d ra=%d' % (
+ entry.offset,
+ entry['length'],
+ entry['CIE_id'],
+ entry['augmentation'],
+ entry['code_alignment_factor'],
+ entry['data_alignment_factor'],
+ entry['return_address_register']))
+ ra_regnum = entry['return_address_register']
+ else: # FDE
+ self._emitline('\n%08x %08x %08x FDE cie=%08x pc=%08x..%08x' % (
+ entry.offset,
+ entry['length'],
+ entry['CIE_pointer'],
+ entry.cie.offset,
+ entry['initial_location'],
+ entry['initial_location'] + entry['address_range']))
+ ra_regnum = entry.cie['return_address_register']
+
+ # Print the heading row for the decoded table
+ self._emit(' LOC')
+ self._emit(' ' if entry.structs.address_size == 4 else ' ')
+ self._emit(' CFA ')
+
+ # Decode the table nad look at the registers it describes.
+ # We build reg_order here to match readelf's order. In particular,
+ # registers are sorted by their number, and the register matching
+ # ra_regnum is always listed last with a special heading.
+ decoded_table = entry.get_decoded()
+ reg_order = sorted(ifilter(
+ lambda r: r != ra_regnum,
+ decoded_table.reg_order))
+
+ # Headings for the registers
+ for regnum in reg_order:
+ self._emit('%-6s' % describe_reg_name(regnum))
+ self._emitline('ra ')
+
+ # Now include ra_regnum in reg_order to print its values similarly
+ # to the other registers.
+ reg_order.append(ra_regnum)
+ for line in decoded_table.table:
+ self._emit(self._format_hex(
+ line['pc'], fullhex=True, lead0x=False))
+ self._emit(' %-9s' % describe_CFI_CFA_rule(line['cfa']))
+
+ for regnum in reg_order:
+ if regnum in line:
+ s = describe_CFI_register_rule(line[regnum])
+ else:
+ s = 'u'
+ self._emit('%-6s' % s)
+ self._emitline()
+ self._emitline()
+
def _emit(self, s=''):
""" Emit an object to output
"""
VERSION_STRING = '%%prog: based on pyelftools %s' % __version__
-def main():
+def main(stream=None):
# parse the command-line arguments and invoke ReadElf
optparser = OptionParser(
usage='usage: %prog [options] <elf-file>',
optparser.add_option('-p', '--string-dump',
action='store', dest='show_string_dump', metavar='<number|name>',
help='Dump the contents of section <number|name> as strings')
+ optparser.add_option('--debug-dump',
+ action='store', dest='debug_dump_what', metavar='<what>',
+ help=(
+ 'Display the contents of DWARF debug sections. <what> can ' +
+ 'one of {info,decodedline,frames,frames-interp}'))
options, args = optparser.parse_args()
with open(args[0], 'rb') as file:
try:
- readelf = ReadElf(file, sys.stdout)
+ readelf = ReadElf(file, stream or sys.stdout)
if do_file_header:
readelf.display_file_header()
if do_section_header:
readelf.display_hex_dump(options.show_hex_dump)
if options.show_string_dump:
readelf.display_string_dump(options.show_string_dump)
+ if options.debug_dump_what:
+ readelf.display_debug_dump(options.debug_dump_what)
except ELFError as ex:
sys.stderr.write('ELF error: %s\n' % ex)
sys.exit(1)
+def profile_main():
+ # Run 'main' redirecting its output to readelfout.txt
+ # Saves profiling information in readelf.profile
+ PROFFILE = 'readelf.profile'
+ import cProfile
+ cProfile.run('main(open("readelfout.txt", "w"))', PROFFILE)
+
+ # Dig in some profiling stats
+ import pstats
+ p = pstats.Stats(PROFFILE)
+ p.sort_stats('cumulative').print_stats(25)
+
+
#-------------------------------------------------------------------------------
if __name__ == '__main__':
main()
+ #profile_main()
+