Fix formatting of some dynamic tag fields to match readelf
[pyelftools.git] / 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') or
314 tag.entry.d_tag.endswith('COUNT')):
315 parsed = '%i' % tag['d_val']
316 elif tag.entry.d_tag == 'DT_PLTREL':
317 s = describe_dyn_tag(tag.entry.d_val)
318 if s.startswith('DT_'):
319 s = s[3:]
320 parsed = '%s' % s
321 else:
322 parsed = '%#x' % tag['d_val']
323
324 self._emitline(" %s %-*s %s" % (
325 self._format_hex(ENUM_D_TAG.get(tag.entry.d_tag, tag.entry.d_tag),
326 fullhex=True, lead0x=True),
327 padding,
328 '(%s)' % (tag.entry.d_tag[3:],),
329 parsed))
330
331 def display_relocations(self):
332 """ Display the relocations contained in the file
333 """
334 has_relocation_sections = False
335 for section in self.elffile.iter_sections():
336 if not isinstance(section, RelocationSection):
337 continue
338
339 has_relocation_sections = True
340 self._emitline("\nRelocation section '%s' at offset %s contains %s entries:" % (
341 bytes2str(section.name),
342 self._format_hex(section['sh_offset']),
343 section.num_relocations()))
344 if section.is_RELA():
345 self._emitline(" Offset Info Type Sym. Value Sym. Name + Addend")
346 else:
347 self._emitline(" Offset Info Type Sym.Value Sym. Name")
348
349 # The symbol table section pointed to in sh_link
350 symtable = self.elffile.get_section(section['sh_link'])
351
352 for rel in section.iter_relocations():
353 hexwidth = 8 if self.elffile.elfclass == 32 else 12
354 self._emit('%s %s %-17.17s' % (
355 self._format_hex(rel['r_offset'],
356 fieldsize=hexwidth, lead0x=False),
357 self._format_hex(rel['r_info'],
358 fieldsize=hexwidth, lead0x=False),
359 describe_reloc_type(
360 rel['r_info_type'], self.elffile)))
361
362 if rel['r_info_sym'] == 0:
363 self._emitline()
364 continue
365
366 symbol = symtable.get_symbol(rel['r_info_sym'])
367 # Some symbols have zero 'st_name', so instead what's used is
368 # the name of the section they point at
369 if symbol['st_name'] == 0:
370 symsec = self.elffile.get_section(symbol['st_shndx'])
371 symbol_name = symsec.name
372 else:
373 symbol_name = symbol.name
374 self._emit(' %s %s%22.22s' % (
375 self._format_hex(
376 symbol['st_value'],
377 fullhex=True, lead0x=False),
378 ' ' if self.elffile.elfclass == 32 else '',
379 bytes2str(symbol_name)))
380 if section.is_RELA():
381 self._emit(' %s %x' % (
382 '+' if rel['r_addend'] >= 0 else '-',
383 abs(rel['r_addend'])))
384 self._emitline()
385
386 if not has_relocation_sections:
387 self._emitline('\nThere are no relocations in this file.')
388
389 def display_hex_dump(self, section_spec):
390 """ Display a hex dump of a section. section_spec is either a section
391 number or a name.
392 """
393 section = self._section_from_spec(section_spec)
394 if section is None:
395 self._emitline("Section '%s' does not exist in the file!" % (
396 section_spec))
397 return
398
399 self._emitline("\nHex dump of section '%s':" % bytes2str(section.name))
400 self._note_relocs_for_section(section)
401 addr = section['sh_addr']
402 data = section.data()
403 dataptr = 0
404
405 while dataptr < len(data):
406 bytesleft = len(data) - dataptr
407 # chunks of 16 bytes per line
408 linebytes = 16 if bytesleft > 16 else bytesleft
409
410 self._emit(' %s ' % self._format_hex(addr, fieldsize=8))
411 for i in range(16):
412 if i < linebytes:
413 self._emit('%2.2x' % byte2int(data[dataptr + i]))
414 else:
415 self._emit(' ')
416 if i % 4 == 3:
417 self._emit(' ')
418
419 for i in range(linebytes):
420 c = data[dataptr + i : dataptr + i + 1]
421 if byte2int(c[0]) >= 32 and byte2int(c[0]) < 0x7f:
422 self._emit(bytes2str(c))
423 else:
424 self._emit(bytes2str(b'.'))
425
426 self._emitline()
427 addr += linebytes
428 dataptr += linebytes
429
430 self._emitline()
431
432 def display_string_dump(self, section_spec):
433 """ Display a strings dump of a section. section_spec is either a
434 section number or a name.
435 """
436 section = self._section_from_spec(section_spec)
437 if section is None:
438 self._emitline("Section '%s' does not exist in the file!" % (
439 section_spec))
440 return
441
442 self._emitline("\nString dump of section '%s':" % bytes2str(section.name))
443
444 found = False
445 data = section.data()
446 dataptr = 0
447
448 while dataptr < len(data):
449 while ( dataptr < len(data) and
450 not (32 <= byte2int(data[dataptr]) <= 127)):
451 dataptr += 1
452
453 if dataptr >= len(data):
454 break
455
456 endptr = dataptr
457 while endptr < len(data) and byte2int(data[endptr]) != 0:
458 endptr += 1
459
460 found = True
461 self._emitline(' [%6x] %s' % (
462 dataptr, bytes2str(data[dataptr:endptr])))
463
464 dataptr = endptr
465
466 if not found:
467 self._emitline(' No strings found in this section.')
468 else:
469 self._emitline()
470
471 def display_debug_dump(self, dump_what):
472 """ Dump a DWARF section
473 """
474 self._init_dwarfinfo()
475 if self._dwarfinfo is None:
476 return
477
478 set_global_machine_arch(self.elffile.get_machine_arch())
479
480 if dump_what == 'info':
481 self._dump_debug_info()
482 elif dump_what == 'decodedline':
483 self._dump_debug_line_programs()
484 elif dump_what == 'frames':
485 self._dump_debug_frames()
486 elif dump_what == 'frames-interp':
487 self._dump_debug_frames_interp()
488 else:
489 self._emitline('debug dump not yet supported for "%s"' % dump_what)
490
491 def _format_hex(self, addr, fieldsize=None, fullhex=False, lead0x=True):
492 """ Format an address into a hexadecimal string.
493
494 fieldsize:
495 Size of the hexadecimal field (with leading zeros to fit the
496 address into. For example with fieldsize=8, the format will
497 be %08x
498 If None, the minimal required field size will be used.
499
500 fullhex:
501 If True, override fieldsize to set it to the maximal size
502 needed for the elfclass
503
504 lead0x:
505 If True, leading 0x is added
506 """
507 s = '0x' if lead0x else ''
508 if fullhex:
509 fieldsize = 8 if self.elffile.elfclass == 32 else 16
510 if fieldsize is None:
511 field = '%x'
512 else:
513 field = '%' + '0%sx' % fieldsize
514 return s + field % addr
515
516 def _section_from_spec(self, spec):
517 """ Retrieve a section given a "spec" (either number or name).
518 Return None if no such section exists in the file.
519 """
520 try:
521 num = int(spec)
522 if num < self.elffile.num_sections():
523 return self.elffile.get_section(num)
524 else:
525 return None
526 except ValueError:
527 # Not a number. Must be a name then
528 return self.elffile.get_section_by_name(str2bytes(spec))
529
530 def _note_relocs_for_section(self, section):
531 """ If there are relocation sections pointing to the givne section,
532 emit a note about it.
533 """
534 for relsec in self.elffile.iter_sections():
535 if isinstance(relsec, RelocationSection):
536 info_idx = relsec['sh_info']
537 if self.elffile.get_section(info_idx) == section:
538 self._emitline(' Note: This section has relocations against it, but these have NOT been applied to this dump.')
539 return
540
541 def _init_dwarfinfo(self):
542 """ Initialize the DWARF info contained in the file and assign it to
543 self._dwarfinfo.
544 Leave self._dwarfinfo at None if no DWARF info was found in the file
545 """
546 if self._dwarfinfo is not None:
547 return
548
549 if self.elffile.has_dwarf_info():
550 self._dwarfinfo = self.elffile.get_dwarf_info()
551 else:
552 self._dwarfinfo = None
553
554 def _dump_debug_info(self):
555 """ Dump the debugging info section.
556 """
557 self._emitline('Contents of the .debug_info section:\n')
558
559 # Offset of the .debug_info section in the stream
560 section_offset = self._dwarfinfo.debug_info_sec.global_offset
561
562 for cu in self._dwarfinfo.iter_CUs():
563 self._emitline(' Compilation Unit @ offset %s:' %
564 self._format_hex(cu.cu_offset))
565 self._emitline(' Length: %s (%s)' % (
566 self._format_hex(cu['unit_length']),
567 '%s-bit' % cu.dwarf_format()))
568 self._emitline(' Version: %s' % cu['version']),
569 self._emitline(' Abbrev Offset: %s' % (
570 self._format_hex(cu['debug_abbrev_offset']))),
571 self._emitline(' Pointer Size: %s' % cu['address_size'])
572
573 # The nesting depth of each DIE within the tree of DIEs must be
574 # displayed. To implement this, a counter is incremented each time
575 # the current DIE has children, and decremented when a null die is
576 # encountered. Due to the way the DIE tree is serialized, this will
577 # correctly reflect the nesting depth
578 #
579 die_depth = 0
580 for die in cu.iter_DIEs():
581 self._emitline(' <%s><%x>: Abbrev Number: %s%s' % (
582 die_depth,
583 die.offset,
584 die.abbrev_code,
585 (' (%s)' % die.tag) if not die.is_null() else ''))
586 if die.is_null():
587 die_depth -= 1
588 continue
589
590 for attr in itervalues(die.attributes):
591 name = attr.name
592 # Unknown attribute values are passed-through as integers
593 if isinstance(name, int):
594 name = 'Unknown AT value: %x' % name
595 self._emitline(' <%2x> %-18s: %s' % (
596 attr.offset,
597 name,
598 describe_attr_value(
599 attr, die, section_offset)))
600
601 if die.has_children:
602 die_depth += 1
603
604 self._emitline()
605
606 def _dump_debug_line_programs(self):
607 """ Dump the (decoded) line programs from .debug_line
608 The programs are dumped in the order of the CUs they belong to.
609 """
610 self._emitline('Decoded dump of debug contents of section .debug_line:\n')
611
612 for cu in self._dwarfinfo.iter_CUs():
613 lineprogram = self._dwarfinfo.line_program_for_CU(cu)
614
615 cu_filename = bytes2str(lineprogram['file_entry'][0].name)
616 if len(lineprogram['include_directory']) > 0:
617 dir_index = lineprogram['file_entry'][0].dir_index
618 if dir_index > 0:
619 dir = lineprogram['include_directory'][dir_index - 1]
620 else:
621 dir = '.'
622 cu_filename = '%s/%s' % (bytes2str(dir), cu_filename)
623
624 self._emitline('CU: %s:' % cu_filename)
625 self._emitline('File name Line number Starting address')
626
627 # Print each state's file, line and address information. For some
628 # instructions other output is needed to be compatible with
629 # readelf.
630 for entry in lineprogram.get_entries():
631 state = entry.state
632 if state is None:
633 # Special handling for commands that don't set a new state
634 if entry.command == DW_LNS_set_file:
635 file_entry = lineprogram['file_entry'][entry.args[0] - 1]
636 if file_entry.dir_index == 0:
637 # current directory
638 self._emitline('\n./%s:[++]' % (
639 bytes2str(file_entry.name)))
640 else:
641 self._emitline('\n%s/%s:' % (
642 bytes2str(lineprogram['include_directory'][file_entry.dir_index - 1]),
643 bytes2str(file_entry.name)))
644 elif entry.command == DW_LNE_define_file:
645 self._emitline('%s:' % (
646 bytes2str(lineprogram['include_directory'][entry.args[0].dir_index])))
647 elif not state.end_sequence:
648 # readelf doesn't print the state after end_sequence
649 # instructions. I think it's a bug but to be compatible
650 # I don't print them too.
651 self._emitline('%-35s %11d %18s' % (
652 bytes2str(lineprogram['file_entry'][state.file - 1].name),
653 state.line,
654 '0' if state.address == 0 else
655 self._format_hex(state.address)))
656 if entry.command == DW_LNS_copy:
657 # Another readelf oddity...
658 self._emitline()
659
660 def _dump_debug_frames(self):
661 """ Dump the raw frame information from .debug_frame
662 """
663 if not self._dwarfinfo.has_CFI():
664 return
665 self._emitline('Contents of the .debug_frame section:')
666
667 for entry in self._dwarfinfo.CFI_entries():
668 if isinstance(entry, CIE):
669 self._emitline('\n%08x %08x %08x CIE' % (
670 entry.offset, entry['length'], entry['CIE_id']))
671 self._emitline(' Version: %d' % entry['version'])
672 self._emitline(' Augmentation: "%s"' % bytes2str(entry['augmentation']))
673 self._emitline(' Code alignment factor: %u' % entry['code_alignment_factor'])
674 self._emitline(' Data alignment factor: %d' % entry['data_alignment_factor'])
675 self._emitline(' Return address column: %d' % entry['return_address_register'])
676 self._emitline()
677 else: # FDE
678 self._emitline('\n%08x %08x %08x FDE cie=%08x pc=%08x..%08x' % (
679 entry.offset,
680 entry['length'],
681 entry['CIE_pointer'],
682 entry.cie.offset,
683 entry['initial_location'],
684 entry['initial_location'] + entry['address_range']))
685
686 self._emit(describe_CFI_instructions(entry))
687 self._emitline()
688
689 def _dump_debug_frames_interp(self):
690 """ Dump the interpreted (decoded) frame information from .debug_frame
691 """
692 if not self._dwarfinfo.has_CFI():
693 return
694
695 self._emitline('Contents of the .debug_frame section:')
696
697 for entry in self._dwarfinfo.CFI_entries():
698 if isinstance(entry, CIE):
699 self._emitline('\n%08x %08x %08x CIE "%s" cf=%d df=%d ra=%d' % (
700 entry.offset,
701 entry['length'],
702 entry['CIE_id'],
703 bytes2str(entry['augmentation']),
704 entry['code_alignment_factor'],
705 entry['data_alignment_factor'],
706 entry['return_address_register']))
707 ra_regnum = entry['return_address_register']
708 else: # FDE
709 self._emitline('\n%08x %08x %08x FDE cie=%08x pc=%08x..%08x' % (
710 entry.offset,
711 entry['length'],
712 entry['CIE_pointer'],
713 entry.cie.offset,
714 entry['initial_location'],
715 entry['initial_location'] + entry['address_range']))
716 ra_regnum = entry.cie['return_address_register']
717
718 # Print the heading row for the decoded table
719 self._emit(' LOC')
720 self._emit(' ' if entry.structs.address_size == 4 else ' ')
721 self._emit(' CFA ')
722
723 # Decode the table nad look at the registers it describes.
724 # We build reg_order here to match readelf's order. In particular,
725 # registers are sorted by their number, and the register matching
726 # ra_regnum is always listed last with a special heading.
727 decoded_table = entry.get_decoded()
728 reg_order = sorted(ifilter(
729 lambda r: r != ra_regnum,
730 decoded_table.reg_order))
731
732 # Headings for the registers
733 for regnum in reg_order:
734 self._emit('%-6s' % describe_reg_name(regnum))
735 self._emitline('ra ')
736
737 # Now include ra_regnum in reg_order to print its values similarly
738 # to the other registers.
739 reg_order.append(ra_regnum)
740 for line in decoded_table.table:
741 self._emit(self._format_hex(
742 line['pc'], fullhex=True, lead0x=False))
743 self._emit(' %-9s' % describe_CFI_CFA_rule(line['cfa']))
744
745 for regnum in reg_order:
746 if regnum in line:
747 s = describe_CFI_register_rule(line[regnum])
748 else:
749 s = 'u'
750 self._emit('%-6s' % s)
751 self._emitline()
752 self._emitline()
753
754 def _emit(self, s=''):
755 """ Emit an object to output
756 """
757 self.output.write(str(s))
758
759 def _emitline(self, s=''):
760 """ Emit an object to output, followed by a newline
761 """
762 self.output.write(str(s) + '\n')
763
764
765 SCRIPT_DESCRIPTION = 'Display information about the contents of ELF format files'
766 VERSION_STRING = '%%prog: based on pyelftools %s' % __version__
767
768
769 def main(stream=None):
770 # parse the command-line arguments and invoke ReadElf
771 optparser = OptionParser(
772 usage='usage: %prog [options] <elf-file>',
773 description=SCRIPT_DESCRIPTION,
774 add_help_option=False, # -h is a real option of readelf
775 prog='readelf.py',
776 version=VERSION_STRING)
777 optparser.add_option('-d', '--dynamic',
778 action='store_true', dest='show_dynamic_tags',
779 help='Display the dynamic section')
780 optparser.add_option('-H', '--help',
781 action='store_true', dest='help',
782 help='Display this information')
783 optparser.add_option('-h', '--file-header',
784 action='store_true', dest='show_file_header',
785 help='Display the ELF file header')
786 optparser.add_option('-l', '--program-headers', '--segments',
787 action='store_true', dest='show_program_header',
788 help='Display the program headers')
789 optparser.add_option('-S', '--section-headers', '--sections',
790 action='store_true', dest='show_section_header',
791 help="Display the sections' headers")
792 optparser.add_option('-e', '--headers',
793 action='store_true', dest='show_all_headers',
794 help='Equivalent to: -h -l -S')
795 optparser.add_option('-s', '--symbols', '--syms',
796 action='store_true', dest='show_symbols',
797 help='Display the symbol table')
798 optparser.add_option('-r', '--relocs',
799 action='store_true', dest='show_relocs',
800 help='Display the relocations (if present)')
801 optparser.add_option('-x', '--hex-dump',
802 action='store', dest='show_hex_dump', metavar='<number|name>',
803 help='Dump the contents of section <number|name> as bytes')
804 optparser.add_option('-p', '--string-dump',
805 action='store', dest='show_string_dump', metavar='<number|name>',
806 help='Dump the contents of section <number|name> as strings')
807 optparser.add_option('--debug-dump',
808 action='store', dest='debug_dump_what', metavar='<what>',
809 help=(
810 'Display the contents of DWARF debug sections. <what> can ' +
811 'one of {info,decodedline,frames,frames-interp}'))
812
813 options, args = optparser.parse_args()
814
815 if options.help or len(args) == 0:
816 optparser.print_help()
817 sys.exit(0)
818
819 if options.show_all_headers:
820 do_file_header = do_section_header = do_program_header = True
821 else:
822 do_file_header = options.show_file_header
823 do_section_header = options.show_section_header
824 do_program_header = options.show_program_header
825
826 with open(args[0], 'rb') as file:
827 try:
828 readelf = ReadElf(file, stream or sys.stdout)
829 if do_file_header:
830 readelf.display_file_header()
831 if do_section_header:
832 readelf.display_section_headers(
833 show_heading=not do_file_header)
834 if do_program_header:
835 readelf.display_program_headers(
836 show_heading=not do_file_header)
837 if options.show_dynamic_tags:
838 readelf.display_dynamic_tags()
839 if options.show_symbols:
840 readelf.display_symbol_tables()
841 if options.show_relocs:
842 readelf.display_relocations()
843 if options.show_hex_dump:
844 readelf.display_hex_dump(options.show_hex_dump)
845 if options.show_string_dump:
846 readelf.display_string_dump(options.show_string_dump)
847 if options.debug_dump_what:
848 readelf.display_debug_dump(options.debug_dump_what)
849 except ELFError as ex:
850 sys.stderr.write('ELF error: %s\n' % ex)
851 sys.exit(1)
852
853
854 def profile_main():
855 # Run 'main' redirecting its output to readelfout.txt
856 # Saves profiling information in readelf.profile
857 PROFFILE = 'readelf.profile'
858 import cProfile
859 cProfile.run('main(open("readelfout.txt", "w"))', PROFFILE)
860
861 # Dig in some profiling stats
862 import pstats
863 p = pstats.Stats(PROFFILE)
864 p.sort_stats('cumulative').print_stats(25)
865
866
867 #-------------------------------------------------------------------------------
868 if __name__ == '__main__':
869 main()
870 #profile_main()
871
872