scripts/glibcelf.py

   1 #!/usr/bin/python3
   2 # ELF support functionality for Python.
   3 # Copyright (C) 2022-2023 Free Software Foundation, Inc.
   4 # This file is part of the GNU C Library.
   5 #
   6 # The GNU C Library is free software; you can redistribute it and/or
   7 # modify it under the terms of the GNU Lesser General Public
   8 # License as published by the Free Software Foundation; either
   9 # version 2.1 of the License, or (at your option) any later version.
  10 #
  11 # The GNU C Library is distributed in the hope that it will be useful,
  12 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14 # Lesser General Public License for more details.
  15 #
  16 # You should have received a copy of the GNU Lesser General Public
  17 # License along with the GNU C Library; if not, see
  18 # <https://www.gnu.org/licenses/>.
  19
  20 """Basic ELF parser.
  21
  22 Use Image.readfile(path) to read an ELF file into memory and begin
  23 parsing it.
  24
  25 """
  26
  27 import collections
  28 import functools
  29 import os
  30 import struct
  31
  32 import glibcpp
  33
  34 class _MetaNamedValue(type):
  35     """Used to set up _NamedValue subclasses."""
  36
  37     @classmethod
  38     def __prepare__(metacls, cls, bases, **kwds):
  39         # Indicates an int-based class.  Needed for types like Shn.
  40         int_based = False
  41         for base in bases:
  42             if issubclass(base, int):
  43                 int_based = int
  44                 break
  45         return dict(by_value={},
  46                     by_name={},
  47                     prefix=None,
  48                     _int_based=int_based)
  49
  50     def __contains__(self, other):
  51         return other in self.by_value
  52
  53 class _NamedValue(metaclass=_MetaNamedValue):
  54     """Typed, named integer constants.
  55
  56     Constants have the following instance attributes:
  57
  58     name: The full name of the constant (e.g., "PT_NULL").
  59     short_name: The name with of the constant without the prefix ("NULL").
  60     value: The integer value of the constant.
  61
  62     The following class attributes are available:
  63
  64     by_value: A dict mapping integers to constants.
  65     by_name: A dict mapping strings to constants.
  66     prefix: A string that is removed from the start of short names, or None.
  67
  68     """
  69
  70     def __new__(cls, arg0, arg1=None):
  71         """Instance creation.
  72
  73         For the one-argument form, the argument must be a string, an
  74         int, or an instance of this class.  Strings are looked up via
  75         by_name.  Values are looked up via by_value; if value lookup
  76         fails, a new unnamed instance is returned.  Instances of this
  77         class a re returned as-is.
  78
  79         The two-argument form expects the name (a string) and the
  80         value (an integer).  A new instance is created in this case.
  81         The instance is not registered in the by_value/by_name
  82         dictionaries (but the caller can do that).
  83
  84         """
  85
  86         typ0 = type(arg0)
  87         if arg1 is None:
  88             if isinstance(typ0, cls):
  89                 # Re-use the existing object.
  90                 return arg0
  91             if typ0 is int:
  92                 by_value = cls.by_value
  93                 try:
  94                     return by_value[arg0]
  95                 except KeyError:
  96                     # Create a new object of the requested value.
  97                     if cls._int_based:
  98                         result = int.__new__(cls, arg0)
  99                     else:
 100                         result = object.__new__(cls)
 101                     result.value = arg0
 102                     result.name = None
 103                     return result
 104             if typ0 is str:
 105                 by_name = cls.by_name
 106                 try:
 107                     return by_name[arg0]
 108                 except KeyError:
 109                     raise ValueError('unknown {} constant: {!r}'.format(
 110                         cls.__name__, arg0))
 111         else:
 112             # Types for the two-argument form are rigid.
 113             if typ0 is not str and typ0 is not None:
 114                 raise ValueError('type {} of name {!r} should be str'.format(
 115                     typ0.__name__, arg0))
 116             if type(arg1) is not int:
 117                 raise ValueError('type {} of value {!r} should be int'.format(
 118                     type(arg1).__name__, arg1))
 119             # Create a new named constants.
 120             if cls._int_based:
 121                 result = int.__new__(cls, arg1)
 122             else:
 123                 result = object.__new__(cls)
 124             result.value = arg1
 125             result.name = arg0
 126             # Set up the short_name attribute.
 127             prefix = cls.prefix
 128             if prefix and arg0.startswith(prefix):
 129                 result.short_name = arg0[len(prefix):]
 130             else:
 131                 result.short_name = arg0
 132             return result
 133
 134     def __str__(self):
 135         name = self.name
 136         if name:
 137             return name
 138         else:
 139             return str(self.value)
 140
 141     def __repr__(self):
 142         name = self.name
 143         if name:
 144             return name
 145         else:
 146             return '{}({})'.format(self.__class__.__name__, self.value)
 147
 148     def __setattr__(self, name, value):
 149         # Prevent modification of the critical attributes once they
 150         # have been set.
 151         if name in ('name', 'value', 'short_name') and hasattr(self, name):
 152             raise AttributeError('can\'t set attribute {}'.format(name))
 153         object.__setattr__(self, name, value)
 154
 155 @functools.total_ordering
 156 class _TypedConstant(_NamedValue):
 157     """Base class for integer-valued optionally named constants.
 158
 159     This type is not an integer type.
 160
 161     """
 162
 163     def __eq__(self, other):
 164         return isinstance(other, self.__class__) and self.value == other.value
 165
 166     def __lt__(self, other):
 167         return isinstance(other, self.__class__) and self.value <= other.value
 168
 169     def __hash__(self):
 170         return hash(self.value)
 171
 172 class _IntConstant(_NamedValue, int):
 173     """Base class for integer-like optionally named constants.
 174
 175     Instances compare equal to the integer of the same value, and can
 176     be used in integer arithmetic.
 177
 178     """
 179
 180     pass
 181
 182 class _FlagConstant(_TypedConstant, int):
 183     pass
 184
 185 def _parse_elf_h():
 186     """Read ../elf/elf.h and return a dict with the constants in it."""
 187
 188     path = os.path.join(os.path.dirname(os.path.realpath(__file__)),
 189                         '..', 'elf', 'elf.h')
 190     class TokenizerReporter:
 191         """Report tokenizer errors to standard output."""
 192
 193         def __init__(self):
 194             self.errors = 0
 195
 196         def error(self, token, message):
 197             self.errors += 1
 198             print('{}:{}:{}: error: {}'.format(
 199                 path, token.line, token.column, message))
 200
 201     reporter = TokenizerReporter()
 202     with open(path) as inp:
 203         tokens = glibcpp.tokenize_c(inp.read(), reporter)
 204     if reporter.errors:
 205         raise IOError('parse error in elf.h')
 206
 207     class MacroReporter:
 208         """Report macro errors to standard output."""
 209
 210         def __init__(self):
 211             self.errors = 0
 212
 213         def error(self, line, message):
 214             self.errors += 1
 215             print('{}:{}: error: {}'.format(path, line, message))
 216
 217         def note(self, line, message):
 218             print('{}:{}: note: {}'.format(path, line, message))
 219
 220     reporter = MacroReporter()
 221     result = glibcpp.macro_eval(glibcpp.macro_definitions(tokens), reporter)
 222     if reporter.errors:
 223         raise IOError('parse error in elf.h')
 224
 225     return result
 226 _elf_h = _parse_elf_h()
 227 del _parse_elf_h
 228 _elf_h_processed = set()
 229
 230 def _register_elf_h(cls, prefix=None, skip=(), ranges=False, parent=None):
 231     prefix = prefix or cls.prefix
 232     if not prefix:
 233         raise ValueError('missing prefix for {}'.format(cls.__name__))
 234     by_value = cls.by_value
 235     by_name = cls.by_name
 236     processed = _elf_h_processed
 237
 238     skip = set(skip)
 239     skip.add(prefix + 'NUM')
 240     if ranges:
 241         skip.add(prefix + 'LOOS')
 242         skip.add(prefix + 'HIOS')
 243         skip.add(prefix + 'LOPROC')
 244         skip.add(prefix + 'HIPROC')
 245         cls.os_range = (_elf_h[prefix + 'LOOS'], _elf_h[prefix + 'HIOS'])
 246         cls.proc_range = (_elf_h[prefix + 'LOPROC'], _elf_h[prefix + 'HIPROC'])
 247
 248     # Inherit the prefix from the parent if not set.
 249     if parent and cls.prefix is None and parent.prefix is not None:
 250         cls.prefix = parent.prefix
 251
 252     processed_len_start = len(processed)
 253     for name, value in _elf_h.items():
 254         if name in skip or name in processed:
 255             continue
 256         if name.startswith(prefix):
 257             processed.add(name)
 258             if value in by_value:
 259                 raise ValueError('duplicate value {}: {}, {}'.format(
 260                     value, name, by_value[value]))
 261             obj = cls(name, value)
 262             by_value[value] = obj
 263             by_name[name] = obj
 264             setattr(cls, name, obj)
 265             if parent:
 266                 # Make the symbolic name available through the parent as well.
 267                 parent.by_name[name] = obj
 268                 setattr(parent, name, obj)
 269
 270     if len(processed) == processed_len_start:
 271         raise ValueError('nothing matched prefix {!r}'.format(prefix))
 272
 273 class ElfClass(_TypedConstant):
 274     """ELF word size.  Type of EI_CLASS values."""
 275 _register_elf_h(ElfClass, prefix='ELFCLASS')
 276
 277 class ElfData(_TypedConstant):
 278     """ELF endianess.  Type of EI_DATA values."""
 279 _register_elf_h(ElfData, prefix='ELFDATA')
 280
 281 class Machine(_TypedConstant):
 282     """ELF machine type.  Type of values in Ehdr.e_machine field."""
 283     prefix = 'EM_'
 284 _register_elf_h(Machine, skip=('EM_ARC_A5',))
 285
 286 class Et(_TypedConstant):
 287     """ELF file type.  Type of ET_* values and the Ehdr.e_type field."""
 288     prefix = 'ET_'
 289 _register_elf_h(Et, ranges=True)
 290
 291 class Shn(_IntConstant):
 292     """ELF reserved section indices."""
 293     prefix = 'SHN_'
 294 class ShnMIPS(Shn):
 295     """Supplemental SHN_* constants for EM_MIPS."""
 296 class ShnPARISC(Shn):
 297     """Supplemental SHN_* constants for EM_PARISC."""
 298 _register_elf_h(ShnMIPS, prefix='SHN_MIPS_', parent=Shn)
 299 _register_elf_h(ShnPARISC, prefix='SHN_PARISC_', parent=Shn)
 300 _register_elf_h(Shn, skip='SHN_LORESERVE SHN_HIRESERVE'.split(), ranges=True)
 301
 302 class Sht(_TypedConstant):
 303     """ELF section types.  Type of SHT_* values."""
 304     prefix = 'SHT_'
 305 class ShtALPHA(Sht):
 306     """Supplemental SHT_* constants for EM_ALPHA."""
 307 class ShtARC(Sht):
 308     """Supplemental SHT_* constants for EM_ARC."""
 309 class ShtARM(Sht):
 310     """Supplemental SHT_* constants for EM_ARM."""
 311 class ShtCSKY(Sht):
 312     """Supplemental SHT_* constants for EM_CSKY."""
 313 class ShtIA_64(Sht):
 314     """Supplemental SHT_* constants for EM_IA_64."""
 315 class ShtMIPS(Sht):
 316     """Supplemental SHT_* constants for EM_MIPS."""
 317 class ShtPARISC(Sht):
 318     """Supplemental SHT_* constants for EM_PARISC."""
 319 class ShtRISCV(Sht):
 320     """Supplemental SHT_* constants for EM_RISCV."""
 321 _register_elf_h(ShtALPHA, prefix='SHT_ALPHA_', parent=Sht)
 322 _register_elf_h(ShtARC, prefix='SHT_ARC_', parent=Sht)
 323 _register_elf_h(ShtARM, prefix='SHT_ARM_', parent=Sht)
 324 _register_elf_h(ShtCSKY, prefix='SHT_CSKY_', parent=Sht)
 325 _register_elf_h(ShtIA_64, prefix='SHT_IA_64_', parent=Sht)
 326 _register_elf_h(ShtMIPS, prefix='SHT_MIPS_', parent=Sht)
 327 _register_elf_h(ShtPARISC, prefix='SHT_PARISC_', parent=Sht)
 328 _register_elf_h(ShtRISCV, prefix='SHT_RISCV_', parent=Sht)
 329 _register_elf_h(Sht, ranges=True,
 330                 skip='SHT_LOSUNW SHT_HISUNW SHT_LOUSER SHT_HIUSER'.split())
 331
 332 class Pf(_FlagConstant):
 333     """Program header flags.  Type of Phdr.p_flags values."""
 334     prefix = 'PF_'
 335 class PfARM(Pf):
 336     """Supplemental PF_* flags for EM_ARM."""
 337 class PfHP(Pf):
 338     """Supplemental PF_* flags for HP-UX."""
 339 class PfIA_64(Pf):
 340     """Supplemental PF_* flags for EM_IA_64."""
 341 class PfMIPS(Pf):
 342     """Supplemental PF_* flags for EM_MIPS."""
 343 class PfPARISC(Pf):
 344     """Supplemental PF_* flags for EM_PARISC."""
 345 _register_elf_h(PfARM, prefix='PF_ARM_', parent=Pf)
 346 _register_elf_h(PfHP, prefix='PF_HP_', parent=Pf)
 347 _register_elf_h(PfIA_64, prefix='PF_IA_64_', parent=Pf)
 348 _register_elf_h(PfMIPS, prefix='PF_MIPS_', parent=Pf)
 349 _register_elf_h(PfPARISC, prefix='PF_PARISC_', parent=Pf)
 350 _register_elf_h(Pf, skip='PF_MASKOS PF_MASKPROC'.split())
 351
 352 class Shf(_FlagConstant):
 353     """Section flags.  Type of Shdr.sh_type values."""
 354     prefix = 'SHF_'
 355 class ShfALPHA(Shf):
 356     """Supplemental SHF_* constants for EM_ALPHA."""
 357 class ShfARM(Shf):
 358     """Supplemental SHF_* constants for EM_ARM."""
 359 class ShfIA_64(Shf):
 360     """Supplemental SHF_* constants for EM_IA_64."""
 361 class ShfMIPS(Shf):
 362     """Supplemental SHF_* constants for EM_MIPS."""
 363 class ShfPARISC(Shf):
 364     """Supplemental SHF_* constants for EM_PARISC."""
 365 _register_elf_h(ShfALPHA, prefix='SHF_ALPHA_', parent=Shf)
 366 _register_elf_h(ShfARM, prefix='SHF_ARM_', parent=Shf)
 367 _register_elf_h(ShfIA_64, prefix='SHF_IA_64_', parent=Shf)
 368 _register_elf_h(ShfMIPS, prefix='SHF_MIPS_', parent=Shf)
 369 _register_elf_h(ShfPARISC, prefix='SHF_PARISC_', parent=Shf)
 370 _register_elf_h(Shf, skip='SHF_MASKOS SHF_MASKPROC'.split())
 371
 372 class Stb(_TypedConstant):
 373     """ELF symbol binding type."""
 374     prefix = 'STB_'
 375 _register_elf_h(Stb, ranges=True)
 376
 377 class Stt(_TypedConstant):
 378     """ELF symbol type."""
 379     prefix = 'STT_'
 380 class SttARM(Sht):
 381     """Supplemental STT_* constants for EM_ARM."""
 382 class SttPARISC(Sht):
 383     """Supplemental STT_* constants for EM_PARISC."""
 384 class SttSPARC(Sht):
 385     """Supplemental STT_* constants for EM_SPARC."""
 386     STT_SPARC_REGISTER = 13
 387 class SttX86_64(Sht):
 388     """Supplemental STT_* constants for EM_X86_64."""
 389 _register_elf_h(SttARM, prefix='STT_ARM_', parent=Stt)
 390 _register_elf_h(SttPARISC, prefix='STT_PARISC_', parent=Stt)
 391 _register_elf_h(SttSPARC, prefix='STT_SPARC_', parent=Stt)
 392 _register_elf_h(Stt, ranges=True)
 393
 394
 395 class Pt(_TypedConstant):
 396     """ELF program header types.  Type of Phdr.p_type."""
 397     prefix = 'PT_'
 398 class PtAARCH64(Pt):
 399     """Supplemental PT_* constants for EM_AARCH64."""
 400 class PtARM(Pt):
 401     """Supplemental PT_* constants for EM_ARM."""
 402 class PtHP(Pt):
 403     """Supplemental PT_* constants for HP-U."""
 404 class PtIA_64(Pt):
 405     """Supplemental PT_* constants for EM_IA_64."""
 406 class PtMIPS(Pt):
 407     """Supplemental PT_* constants for EM_MIPS."""
 408 class PtPARISC(Pt):
 409     """Supplemental PT_* constants for EM_PARISC."""
 410 class PtRISCV(Pt):
 411     """Supplemental PT_* constants for EM_RISCV."""
 412 _register_elf_h(PtAARCH64, prefix='PT_AARCH64_', parent=Pt)
 413 _register_elf_h(PtARM, prefix='PT_ARM_', parent=Pt)
 414 _register_elf_h(PtHP, prefix='PT_HP_', parent=Pt)
 415 _register_elf_h(PtIA_64, prefix='PT_IA_64_', parent=Pt)
 416 _register_elf_h(PtMIPS, prefix='PT_MIPS_', parent=Pt)
 417 _register_elf_h(PtPARISC, prefix='PT_PARISC_', parent=Pt)
 418 _register_elf_h(PtRISCV, prefix='PT_RISCV_', parent=Pt)
 419 _register_elf_h(Pt, skip='PT_LOSUNW PT_HISUNW'.split(), ranges=True)
 420
 421 class Dt(_TypedConstant):
 422     """ELF dynamic segment tags.  Type of Dyn.d_val."""
 423     prefix = 'DT_'
 424 class DtAARCH64(Dt):
 425     """Supplemental DT_* constants for EM_AARCH64."""
 426 class DtALPHA(Dt):
 427     """Supplemental DT_* constants for EM_ALPHA."""
 428 class DtALTERA_NIOS2(Dt):
 429     """Supplemental DT_* constants for EM_ALTERA_NIOS2."""
 430 class DtIA_64(Dt):
 431     """Supplemental DT_* constants for EM_IA_64."""
 432 class DtMIPS(Dt):
 433     """Supplemental DT_* constants for EM_MIPS."""
 434 class DtPPC(Dt):
 435     """Supplemental DT_* constants for EM_PPC."""
 436 class DtPPC64(Dt):
 437     """Supplemental DT_* constants for EM_PPC64."""
 438 class DtRISCV(Dt):
 439     """Supplemental DT_* constants for EM_RISCV."""
 440 class DtSPARC(Dt):
 441     """Supplemental DT_* constants for EM_SPARC."""
 442 _dt_skip = '''
 443 DT_ENCODING DT_PROCNUM
 444 DT_ADDRRNGLO DT_ADDRRNGHI DT_ADDRNUM
 445 DT_VALRNGLO DT_VALRNGHI DT_VALNUM
 446 DT_VERSIONTAGNUM DT_EXTRANUM
 447 DT_AARCH64_NUM
 448 DT_ALPHA_NUM
 449 DT_IA_64_NUM
 450 DT_MIPS_NUM
 451 DT_PPC_NUM
 452 DT_PPC64_NUM
 453 DT_SPARC_NUM
 454 '''.strip().split()
 455 _register_elf_h(DtAARCH64, prefix='DT_AARCH64_', skip=_dt_skip, parent=Dt)
 456 _register_elf_h(DtALPHA, prefix='DT_ALPHA_', skip=_dt_skip, parent=Dt)
 457 _register_elf_h(DtALTERA_NIOS2, prefix='DT_NIOS2_', skip=_dt_skip, parent=Dt)
 458 _register_elf_h(DtIA_64, prefix='DT_IA_64_', skip=_dt_skip, parent=Dt)
 459 _register_elf_h(DtMIPS, prefix='DT_MIPS_', skip=_dt_skip, parent=Dt)
 460 _register_elf_h(DtPPC, prefix='DT_PPC_', skip=_dt_skip, parent=Dt)
 461 _register_elf_h(DtPPC64, prefix='DT_PPC64_', skip=_dt_skip, parent=Dt)
 462 _register_elf_h(DtRISCV, prefix='DT_RISCV_', skip=_dt_skip, parent=Dt)
 463 _register_elf_h(DtSPARC, prefix='DT_SPARC_', skip=_dt_skip, parent=Dt)
 464 _register_elf_h(Dt, skip=_dt_skip, ranges=True)
 465 del _dt_skip
 466
 467 # Constant extraction is complete.
 468 del _register_elf_h
 469 del _elf_h
 470
 471 class StInfo:
 472     """ELF symbol binding and type.  Type of the Sym.st_info field."""
 473     def __init__(self, arg0, arg1=None):
 474         if isinstance(arg0, int) and arg1 is None:
 475             self.bind = Stb(arg0 >> 4)
 476             self.type = Stt(arg0 & 15)
 477         else:
 478             self.bind = Stb(arg0)
 479             self.type = Stt(arg1)
 480
 481     def value(self):
 482         """Returns the raw value for the bind/type combination."""
 483         return (self.bind.value() << 4) | (self.type.value())
 484
 485 # Type in an ELF file.  Used for deserialization.
 486 _Layout = collections.namedtuple('_Layout', 'unpack size')
 487
 488 def _define_layouts(baseclass: type, layout32: str, layout64: str,
 489                     types=None, fields32=None):
 490     """Assign variants dict to baseclass.
 491
 492     The variants dict is indexed by (ElfClass, ElfData) pairs, and its
 493     values are _Layout instances.
 494
 495     """
 496     struct32 = struct.Struct(layout32)
 497     struct64 = struct.Struct(layout64)
 498
 499     # Check that the struct formats yield the right number of components.
 500     for s in (struct32, struct64):
 501         example = s.unpack(b' ' * s.size)
 502         if len(example) != len(baseclass._fields):
 503             raise ValueError('{!r} yields wrong field count: {} != {}'.format(
 504                 s.format, len(example),  len(baseclass._fields)))
 505
 506     # Check that field names in types are correct.
 507     if types is None:
 508         types = ()
 509     for n in types:
 510         if n not in baseclass._fields:
 511             raise ValueError('{} does not have field {!r}'.format(
 512                 baseclass.__name__, n))
 513
 514     if fields32 is not None \
 515        and set(fields32) != set(baseclass._fields):
 516         raise ValueError('{!r} is not a permutation of the fields {!r}'.format(
 517             fields32, baseclass._fields))
 518
 519     def unique_name(name, used_names = (set((baseclass.__name__,))
 520                                         | set(baseclass._fields)
 521                                         | {n.__name__
 522                                            for n in (types or {}).values()})):
 523         """Find a name that is not used for a class or field name."""
 524         candidate = name
 525         n = 0
 526         while candidate in used_names:
 527             n += 1
 528             candidate = '{}{}'.format(name, n)
 529         used_names.add(candidate)
 530         return candidate
 531
 532     blob_name = unique_name('blob')
 533     struct_unpack_name = unique_name('struct_unpack')
 534     comps_name = unique_name('comps')
 535
 536     layouts = {}
 537     for (bits, elfclass, layout, fields) in (
 538             (32, ElfClass.ELFCLASS32, layout32, fields32),
 539             (64, ElfClass.ELFCLASS64, layout64, None),
 540     ):
 541         for (elfdata, structprefix, funcsuffix) in (
 542                 (ElfData.ELFDATA2LSB, '<', 'LE'),
 543                 (ElfData.ELFDATA2MSB, '>', 'BE'),
 544         ):
 545             env = {
 546                 baseclass.__name__: baseclass,
 547                 struct_unpack_name: struct.unpack,
 548             }
 549
 550             # Add the type converters.
 551             if types:
 552                 for cls in types.values():
 553                     env[cls.__name__] = cls
 554
 555             funcname = ''.join(
 556                 ('unpack_', baseclass.__name__, str(bits), funcsuffix))
 557
 558             code = '''
 559 def {funcname}({blob_name}):
 560 '''.format(funcname=funcname, blob_name=blob_name)
 561
 562             indent = ' ' * 4
 563             unpack_call = '{}({!r}, {})'.format(
 564                 struct_unpack_name, structprefix + layout, blob_name)
 565             field_names = ', '.join(baseclass._fields)
 566             if types is None and fields is None:
 567                 code += '{}return {}({})\n'.format(
 568                     indent, baseclass.__name__, unpack_call)
 569             else:
 570                 # Destructuring tuple assignment.
 571                 if fields is None:
 572                     code += '{}{} = {}\n'.format(
 573                         indent, field_names, unpack_call)
 574                 else:
 575                     # Use custom field order.
 576                     code += '{}{} = {}\n'.format(
 577                         indent, ', '.join(fields), unpack_call)
 578
 579                 # Perform the type conversions.
 580                 for n in baseclass._fields:
 581                     if n in types:
 582                         code += '{}{} = {}({})\n'.format(
 583                             indent, n, types[n].__name__, n)
 584                 # Create the named tuple.
 585                 code += '{}return {}({})\n'.format(
 586                     indent, baseclass.__name__, field_names)
 587
 588             exec(code, env)
 589             layouts[(elfclass, elfdata)] = _Layout(
 590                 env[funcname], struct.calcsize(layout))
 591     baseclass.layouts = layouts
 592
 593
 594 # Corresponds to EI_* indices into Elf*_Ehdr.e_indent.
 595 class Ident(collections.namedtuple('Ident',
 596     'ei_mag ei_class ei_data ei_version ei_osabi ei_abiversion ei_pad')):
 597
 598     def __new__(cls, *args):
 599         """Construct an object from a blob or its constituent fields."""
 600         if len(args) == 1:
 601             return cls.unpack(args[0])
 602         return cls.__base__.__new__(cls, *args)
 603
 604     @staticmethod
 605     def unpack(blob: memoryview) -> 'Ident':
 606         """Parse raws data into a tuple."""
 607         ei_mag, ei_class, ei_data, ei_version, ei_osabi, ei_abiversion, \
 608             ei_pad = struct.unpack('4s5B7s', blob)
 609         return Ident(ei_mag, ElfClass(ei_class), ElfData(ei_data),
 610                      ei_version, ei_osabi, ei_abiversion, ei_pad)
 611     size = 16
 612
 613 # Corresponds to Elf32_Ehdr and Elf64_Ehdr.
 614 Ehdr = collections.namedtuple('Ehdr',
 615    'e_ident e_type e_machine e_version e_entry e_phoff e_shoff e_flags'
 616     + ' e_ehsize e_phentsize e_phnum e_shentsize e_shnum e_shstrndx')
 617 _define_layouts(Ehdr,
 618                 layout32='16s2H5I6H',
 619                 layout64='16s2HI3QI6H',
 620                 types=dict(e_ident=Ident,
 621                            e_machine=Machine,
 622                            e_type=Et,
 623                            e_shstrndx=Shn))
 624
 625 # Corresponds to Elf32_Phdr and Elf64_Pdhr.  Order follows the latter.
 626 Phdr = collections.namedtuple('Phdr',
 627     'p_type p_flags p_offset p_vaddr p_paddr p_filesz p_memsz p_align')
 628 _define_layouts(Phdr,
 629                 layout32='8I',
 630                 fields32=('p_type', 'p_offset', 'p_vaddr', 'p_paddr',
 631                           'p_filesz', 'p_memsz', 'p_flags', 'p_align'),
 632                 layout64='2I6Q',
 633             types=dict(p_type=Pt, p_flags=Pf))
 634
 635
 636 # Corresponds to Elf32_Shdr and Elf64_Shdr.
 637 class Shdr(collections.namedtuple('Shdr',
 638     'sh_name sh_type sh_flags sh_addr sh_offset sh_size sh_link sh_info'
 639     + ' sh_addralign sh_entsize')):
 640     def resolve(self, strtab: 'StringTable') -> 'Shdr':
 641         """Resolve sh_name using a string table."""
 642         return self.__class__(strtab.get(self[0]), *self[1:])
 643 _define_layouts(Shdr,
 644                 layout32='10I',
 645                 layout64='2I4Q2I2Q',
 646                 types=dict(sh_type=Sht,
 647                            sh_flags=Shf,
 648                            sh_link=Shn))
 649
 650 # Corresponds to Elf32_Dyn and Elf64_Dyn.  The nesting through the
 651 # d_un union is skipped, and d_ptr is missing (its representation in
 652 # Python would be identical to d_val).
 653 Dyn = collections.namedtuple('Dyn', 'd_tag d_val')
 654 _define_layouts(Dyn,
 655                 layout32='2i',
 656                 layout64='2q',
 657                 types=dict(d_tag=Dt))
 658
 659 # Corresponds to Elf32_Sym and Elf64_Sym.
 660 class Sym(collections.namedtuple('Sym',
 661     'st_name st_info st_other st_shndx st_value st_size')):
 662     def resolve(self, strtab: 'StringTable') -> 'Sym':
 663         """Resolve st_name using a string table."""
 664         return self.__class__(strtab.get(self[0]), *self[1:])
 665 _define_layouts(Sym,
 666                 layout32='3I2BH',
 667                 layout64='I2BH2Q',
 668                 fields32=('st_name', 'st_value', 'st_size', 'st_info',
 669                           'st_other', 'st_shndx'),
 670                 types=dict(st_shndx=Shn,
 671                            st_info=StInfo))
 672
 673 # Corresponds to Elf32_Rel and Elf64_Rel.
 674 Rel = collections.namedtuple('Rel', 'r_offset r_info')
 675 _define_layouts(Rel,
 676                 layout32='2I',
 677                 layout64='2Q')
 678
 679 # Corresponds to Elf32_Rel and Elf64_Rel.
 680 Rela = collections.namedtuple('Rela', 'r_offset r_info r_addend')
 681 _define_layouts(Rela,
 682                 layout32='3I',
 683                 layout64='3Q')
 684
 685 class StringTable:
 686     """ELF string table."""
 687     def __init__(self, blob):
 688         """Create a new string table backed by the data in the blob.
 689
 690         blob: a memoryview-like object
 691
 692         """
 693         self.blob = blob
 694
 695     def get(self, index) -> bytes:
 696         """Returns the null-terminated byte string at the index."""
 697         blob = self.blob
 698         endindex = index
 699         while True:
 700             if blob[endindex] == 0:
 701                 return bytes(blob[index:endindex])
 702             endindex += 1
 703
 704 class Image:
 705     """ELF image parser."""
 706     def __init__(self, image):
 707         """Create an ELF image from binary image data.
 708
 709         image: a memoryview-like object that supports efficient range
 710         subscripting.
 711
 712         """
 713         self.image = image
 714         ident = self.read(Ident, 0)
 715         classdata = (ident.ei_class, ident.ei_data)
 716         # Set self.Ehdr etc. to the subtypes with the right parsers.
 717         for typ in (Ehdr, Phdr, Shdr, Dyn, Sym, Rel, Rela):
 718             setattr(self, typ.__name__, typ.layouts.get(classdata, None))
 719
 720         if self.Ehdr is not None:
 721             self.ehdr = self.read(self.Ehdr, 0)
 722             self._shdr_num = self._compute_shdr_num()
 723         else:
 724             self.ehdr = None
 725             self._shdr_num = 0
 726
 727         self._section = {}
 728         self._stringtab = {}
 729
 730         if self._shdr_num > 0:
 731             self._shdr_strtab = self._find_shdr_strtab()
 732         else:
 733             self._shdr_strtab = None
 734
 735     @staticmethod
 736     def readfile(path: str) -> 'Image':
 737         """Reads the ELF file at the specified path."""
 738         with open(path, 'rb') as inp:
 739             return Image(memoryview(inp.read()))
 740
 741     def _compute_shdr_num(self) -> int:
 742         """Computes the actual number of section headers."""
 743         shnum = self.ehdr.e_shnum
 744         if shnum == 0:
 745             if self.ehdr.e_shoff == 0 or self.ehdr.e_shentsize == 0:
 746                 # No section headers.
 747                 return 0
 748             # Otherwise the extension mechanism is used (which may be
 749             # needed because e_shnum is just 16 bits).
 750             return self.read(self.Shdr, self.ehdr.e_shoff).sh_size
 751         return shnum
 752
 753     def _find_shdr_strtab(self) -> StringTable:
 754         """Finds the section header string table (maybe via extensions)."""
 755         shstrndx = self.ehdr.e_shstrndx
 756         if shstrndx == Shn.SHN_XINDEX:
 757             shstrndx = self.read(self.Shdr, self.ehdr.e_shoff).sh_link
 758         return self._find_stringtab(shstrndx)
 759
 760     def read(self, typ: type, offset:int ):
 761         """Reads an object at a specific offset.
 762
 763         The type must have been enhanced using _define_variants.
 764
 765         """
 766         return typ.unpack(self.image[offset: offset + typ.size])
 767
 768     def phdrs(self) -> Phdr:
 769         """Generator iterating over the program headers."""
 770         if self.ehdr is None:
 771             return
 772         size = self.ehdr.e_phentsize
 773         if size != self.Phdr.size:
 774             raise ValueError('Unexpected Phdr size in ELF header: {} != {}'
 775                              .format(size, self.Phdr.size))
 776
 777         offset = self.ehdr.e_phoff
 778         for _ in range(self.ehdr.e_phnum):
 779             yield self.read(self.Phdr, offset)
 780             offset += size
 781
 782     def shdrs(self, resolve: bool=True) -> Shdr:
 783         """Generator iterating over the section headers.
 784
 785         If resolve, section names are automatically translated
 786         using the section header string table.
 787
 788         """
 789         if self._shdr_num == 0:
 790             return
 791
 792         size = self.ehdr.e_shentsize
 793         if size != self.Shdr.size:
 794             raise ValueError('Unexpected Shdr size in ELF header: {} != {}'
 795                              .format(size, self.Shdr.size))
 796
 797         offset = self.ehdr.e_shoff
 798         for _ in range(self._shdr_num):
 799             shdr = self.read(self.Shdr, offset)
 800             if resolve:
 801                 shdr = shdr.resolve(self._shdr_strtab)
 802             yield shdr
 803             offset += size
 804
 805     def dynamic(self) -> Dyn:
 806         """Generator iterating over the dynamic segment."""
 807         for phdr in self.phdrs():
 808             if phdr.p_type == Pt.PT_DYNAMIC:
 809                 # Pick the first dynamic segment, like the loader.
 810                 if phdr.p_filesz == 0:
 811                     # Probably separated debuginfo.
 812                     return
 813                 offset = phdr.p_offset
 814                 end = offset + phdr.p_memsz
 815                 size = self.Dyn.size
 816                 while True:
 817                     next_offset = offset + size
 818                     if next_offset > end:
 819                         raise ValueError(
 820                             'Dynamic segment size {} is not a multiple of Dyn size {}'.format(
 821                                 phdr.p_memsz, size))
 822                     yield self.read(self.Dyn, offset)
 823                     if next_offset == end:
 824                         return
 825                     offset = next_offset
 826
 827     def syms(self, shdr: Shdr, resolve: bool=True) -> Sym:
 828         """A generator iterating over a symbol table.
 829
 830         If resolve, symbol names are automatically translated using
 831         the string table for the symbol table.
 832
 833         """
 834         assert shdr.sh_type == Sht.SHT_SYMTAB
 835         size = shdr.sh_entsize
 836         if size != self.Sym.size:
 837             raise ValueError('Invalid symbol table entry size {}'.format(size))
 838         offset = shdr.sh_offset
 839         end = shdr.sh_offset + shdr.sh_size
 840         if resolve:
 841             strtab = self._find_stringtab(shdr.sh_link)
 842         while offset < end:
 843             sym = self.read(self.Sym, offset)
 844             if resolve:
 845                 sym = sym.resolve(strtab)
 846             yield sym
 847             offset += size
 848         if offset != end:
 849             raise ValueError('Symbol table is not a multiple of entry size')
 850
 851     def lookup_string(self, strtab_index: int, strtab_offset: int) -> bytes:
 852         """Looks up a string in a string table identified by its link index."""
 853         try:
 854             strtab = self._stringtab[strtab_index]
 855         except KeyError:
 856             strtab = self._find_stringtab(strtab_index)
 857         return strtab.get(strtab_offset)
 858
 859     def find_section(self, shndx: Shn) -> Shdr:
 860         """Returns the section header for the indexed section.
 861
 862         The section name is not resolved.
 863         """
 864         try:
 865             return self._section[shndx]
 866         except KeyError:
 867             pass
 868         if shndx in Shn:
 869             raise ValueError('Reserved section index {}'.format(shndx))
 870         idx = shndx.value
 871         if idx < 0 or idx > self._shdr_num:
 872             raise ValueError('Section index {} out of range [0, {})'.format(
 873                 idx, self._shdr_num))
 874         shdr = self.read(
 875             self.Shdr, self.ehdr.e_shoff + idx * self.Shdr.size)
 876         self._section[shndx] = shdr
 877         return shdr
 878
 879     def _find_stringtab(self, sh_link: int) -> StringTable:
 880         if sh_link in self._stringtab:
 881             return self._stringtab
 882         if sh_link < 0 or sh_link >= self._shdr_num:
 883             raise ValueError('Section index {} out of range [0, {})'.format(
 884                 sh_link, self._shdr_num))
 885         shdr = self.read(
 886             self.Shdr, self.ehdr.e_shoff + sh_link * self.Shdr.size)
 887         if shdr.sh_type != Sht.SHT_STRTAB:
 888             raise ValueError(
 889                 'Section {} is not a string table: {}'.format(
 890                     sh_link, shdr.sh_type))
 891         strtab = StringTable(
 892             self.image[shdr.sh_offset:shdr.sh_offset + shdr.sh_size])
 893         # This could retrain essentially arbitrary amounts of data,
 894         # but caching string tables seems important for performance.
 895         self._stringtab[sh_link] = strtab
 896         return strtab
 897
 898 def elf_hash(s):
 899     """Computes the ELF hash of the string."""
 900     acc = 0
 901     for ch in s:
 902         if type(ch) is not int:
 903             ch = ord(ch)
 904         acc = ((acc << 4) + ch) & 0xffffffff
 905         top = acc & 0xf0000000
 906         acc = (acc ^ (top >> 24)) & ~top
 907     return acc
 908
 909 def gnu_hash(s):
 910     """Computes the GNU hash of the string."""
 911     h = 5381
 912     for ch in s:
 913         if type(ch) is not int:
 914             ch = ord(ch)
 915         h = (h * 33 + ch) & 0xffffffff
 916     return h
 917
 918 __all__ = [name for name in dir() if name[0].isupper()]