elftools/elf/structs.py

   1 #-------------------------------------------------------------------------------
   2 # elftools: elf/structs.py
   3 #
   4 # Encapsulation of Construct structs for parsing an ELF file, adjusted for
   5 # correct endianness and word-size.
   6 #
   7 # Eli Bendersky (eliben@gmail.com)
   8 # This code is in the public domain
   9 #-------------------------------------------------------------------------------
  10 from elftools.construct.macros import AlignedStruct, IfThenElse, UNInt8
  11 from ..construct import (
  12     UBInt8, UBInt16, UBInt32, UBInt64,
  13     ULInt8, ULInt16, ULInt32, ULInt64,
  14     SBInt32, SLInt32, SBInt64, SLInt64,
  15     Struct, Array, Enum, Padding, BitStruct, BitField, Value, String, CString,
  16     Switch, Field
  17     )
  18 from ..common.construct_utils import ULEB128
  19 from ..common.utils import roundup
  20 from .enums import *
  21
  22
  23 class ELFStructs(object):
  24     """ Accessible attributes:
  25
  26             Elf_{byte|half|word|word64|addr|offset|sword|xword|xsword}:
  27                 Data chunks, as specified by the ELF standard, adjusted for
  28                 correct endianness and word-size.
  29
  30             Elf_Ehdr:
  31                 ELF file header
  32
  33             Elf_Phdr:
  34                 Program header
  35
  36             Elf_Shdr:
  37                 Section header
  38
  39             Elf_Sym:
  40                 Symbol table entry
  41
  42             Elf_Rel, Elf_Rela:
  43                 Entries in relocation sections
  44     """
  45     def __init__(self, little_endian=True, elfclass=32):
  46         assert elfclass == 32 or elfclass == 64
  47         self.little_endian = little_endian
  48         self.elfclass = elfclass
  49         self.e_type = None
  50         self.e_machine = None
  51         self.e_ident_osabi = None
  52
  53     def __getstate__(self):
  54         return self.little_endian, self.elfclass, self.e_type, self.e_machine, self.e_ident_osabi
  55
  56     def __setstate__(self, state):
  57         self.little_endian, self.elfclass, e_type, e_machine, e_osabi = state
  58         self.create_basic_structs()
  59         self.create_advanced_structs(e_type, e_machine, e_osabi)
  60
  61     def create_basic_structs(self):
  62         """ Create word-size related structs and ehdr struct needed for
  63             initial determining of ELF type.
  64         """
  65         if self.little_endian:
  66             self.Elf_byte = ULInt8
  67             self.Elf_half = ULInt16
  68             self.Elf_word = ULInt32
  69             self.Elf_word64 = ULInt64
  70             self.Elf_addr = ULInt32 if self.elfclass == 32 else ULInt64
  71             self.Elf_offset = self.Elf_addr
  72             self.Elf_sword = SLInt32
  73             self.Elf_xword = ULInt32 if self.elfclass == 32 else ULInt64
  74             self.Elf_sxword = SLInt32 if self.elfclass == 32 else SLInt64
  75         else:
  76             self.Elf_byte = UBInt8
  77             self.Elf_half = UBInt16
  78             self.Elf_word = UBInt32
  79             self.Elf_word64 = UBInt64
  80             self.Elf_addr = UBInt32 if self.elfclass == 32 else UBInt64
  81             self.Elf_offset = self.Elf_addr
  82             self.Elf_sword = SBInt32
  83             self.Elf_xword = UBInt32 if self.elfclass == 32 else UBInt64
  84             self.Elf_sxword = SBInt32 if self.elfclass == 32 else SBInt64
  85         self._create_ehdr()
  86         self._create_leb128()
  87         self._create_ntbs()
  88
  89     def create_advanced_structs(self, e_type=None, e_machine=None, e_ident_osabi=None):
  90         """ Create all ELF structs except the ehdr. They may possibly depend
  91             on provided e_type and/or e_machine parsed from ehdr.
  92         """
  93         self.e_type = e_type
  94         self.e_machine = e_machine
  95         self.e_ident_osabi = e_ident_osabi
  96
  97         self._create_phdr()
  98         self._create_shdr()
  99         self._create_chdr()
 100         self._create_sym()
 101         self._create_rel()
 102         self._create_dyn()
 103         self._create_sunw_syminfo()
 104         self._create_gnu_verneed()
 105         self._create_gnu_verdef()
 106         self._create_gnu_versym()
 107         self._create_gnu_abi()
 108         self._create_gnu_property()
 109         self._create_note(e_type)
 110         self._create_stabs()
 111         self._create_arm_attributes()
 112         self._create_elf_hash()
 113         self._create_gnu_hash()
 114
 115     #-------------------------------- PRIVATE --------------------------------#
 116
 117     def _create_ehdr(self):
 118         self.Elf_Ehdr = Struct('Elf_Ehdr',
 119             Struct('e_ident',
 120                 Array(4, self.Elf_byte('EI_MAG')),
 121                 Enum(self.Elf_byte('EI_CLASS'), **ENUM_EI_CLASS),
 122                 Enum(self.Elf_byte('EI_DATA'), **ENUM_EI_DATA),
 123                 Enum(self.Elf_byte('EI_VERSION'), **ENUM_E_VERSION),
 124                 Enum(self.Elf_byte('EI_OSABI'), **ENUM_EI_OSABI),
 125                 self.Elf_byte('EI_ABIVERSION'),
 126                 Padding(7)
 127             ),
 128             Enum(self.Elf_half('e_type'), **ENUM_E_TYPE),
 129             Enum(self.Elf_half('e_machine'), **ENUM_E_MACHINE),
 130             Enum(self.Elf_word('e_version'), **ENUM_E_VERSION),
 131             self.Elf_addr('e_entry'),
 132             self.Elf_offset('e_phoff'),
 133             self.Elf_offset('e_shoff'),
 134             self.Elf_word('e_flags'),
 135             self.Elf_half('e_ehsize'),
 136             self.Elf_half('e_phentsize'),
 137             self.Elf_half('e_phnum'),
 138             self.Elf_half('e_shentsize'),
 139             self.Elf_half('e_shnum'),
 140             self.Elf_half('e_shstrndx'),
 141         )
 142
 143     def _create_leb128(self):
 144         self.Elf_uleb128 = ULEB128
 145
 146     def _create_ntbs(self):
 147         self.Elf_ntbs = CString
 148
 149     def _create_phdr(self):
 150         p_type_dict = ENUM_P_TYPE_BASE
 151         if self.e_machine == 'EM_ARM':
 152             p_type_dict = ENUM_P_TYPE_ARM
 153         elif self.e_machine == 'EM_AARCH64':
 154             p_type_dict = ENUM_P_TYPE_AARCH64
 155         elif self.e_machine == 'EM_MIPS':
 156             p_type_dict = ENUM_P_TYPE_MIPS
 157
 158         if self.elfclass == 32:
 159             self.Elf_Phdr = Struct('Elf_Phdr',
 160                 Enum(self.Elf_word('p_type'), **p_type_dict),
 161                 self.Elf_offset('p_offset'),
 162                 self.Elf_addr('p_vaddr'),
 163                 self.Elf_addr('p_paddr'),
 164                 self.Elf_word('p_filesz'),
 165                 self.Elf_word('p_memsz'),
 166                 self.Elf_word('p_flags'),
 167                 self.Elf_word('p_align'),
 168             )
 169         else: # 64
 170             self.Elf_Phdr = Struct('Elf_Phdr',
 171                 Enum(self.Elf_word('p_type'), **p_type_dict),
 172                 self.Elf_word('p_flags'),
 173                 self.Elf_offset('p_offset'),
 174                 self.Elf_addr('p_vaddr'),
 175                 self.Elf_addr('p_paddr'),
 176                 self.Elf_xword('p_filesz'),
 177                 self.Elf_xword('p_memsz'),
 178                 self.Elf_xword('p_align'),
 179             )
 180
 181     def _create_shdr(self):
 182         """Section header parsing.
 183
 184         Depends on e_machine because of machine-specific values in sh_type.
 185         """
 186         sh_type_dict = ENUM_SH_TYPE_BASE
 187         if self.e_machine == 'EM_ARM':
 188             sh_type_dict = ENUM_SH_TYPE_ARM
 189         elif self.e_machine == 'EM_X86_64':
 190             sh_type_dict = ENUM_SH_TYPE_AMD64
 191         elif self.e_machine == 'EM_MIPS':
 192             sh_type_dict = ENUM_SH_TYPE_MIPS
 193
 194         self.Elf_Shdr = Struct('Elf_Shdr',
 195             self.Elf_word('sh_name'),
 196             Enum(self.Elf_word('sh_type'), **sh_type_dict),
 197             self.Elf_xword('sh_flags'),
 198             self.Elf_addr('sh_addr'),
 199             self.Elf_offset('sh_offset'),
 200             self.Elf_xword('sh_size'),
 201             self.Elf_word('sh_link'),
 202             self.Elf_word('sh_info'),
 203             self.Elf_xword('sh_addralign'),
 204             self.Elf_xword('sh_entsize'),
 205         )
 206
 207     def _create_chdr(self):
 208         # Structure of compressed sections header. It is documented in Oracle
 209         # "Linker and Libraries Guide", Part IV ELF Application Binary
 210         # Interface, Chapter 13 Object File Format, Section Compression:
 211         # https://docs.oracle.com/cd/E53394_01/html/E54813/section_compression.html
 212         fields = [
 213             Enum(self.Elf_word('ch_type'), **ENUM_ELFCOMPRESS_TYPE),
 214             self.Elf_xword('ch_size'),
 215             self.Elf_xword('ch_addralign'),
 216         ]
 217         if self.elfclass == 64:
 218             fields.insert(1, self.Elf_word('ch_reserved'))
 219         self.Elf_Chdr = Struct('Elf_Chdr', *fields)
 220
 221     def _create_rel(self):
 222         # r_info is also taken apart into r_info_sym and r_info_type. This is
 223         # done in Value to avoid endianity issues while parsing.
 224         if self.elfclass == 32:
 225             fields = [self.Elf_xword('r_info'),
 226                       Value('r_info_sym',
 227                             lambda ctx: (ctx['r_info'] >> 8) & 0xFFFFFF),
 228                       Value('r_info_type',
 229                             lambda ctx: ctx['r_info'] & 0xFF)]
 230         elif self.e_machine == 'EM_MIPS': # ELF64 MIPS
 231             fields = [
 232                 # The MIPS ELF64 specification
 233                 # (https://www.linux-mips.org/pub/linux/mips/doc/ABI/elf64-2.4.pdf)
 234                 # provides a non-standard relocation structure definition.
 235                 self.Elf_word('r_sym'),
 236                 self.Elf_byte('r_ssym'),
 237                 self.Elf_byte('r_type3'),
 238                 self.Elf_byte('r_type2'),
 239                 self.Elf_byte('r_type'),
 240
 241                 # Synthetize usual fields for compatibility with other
 242                 # architectures. This allows relocation consumers (including
 243                 # our readelf tests) to work without worrying about MIPS64
 244                 # oddities.
 245                 Value('r_info_sym', lambda ctx: ctx['r_sym']),
 246                 Value('r_info_ssym', lambda ctx: ctx['r_ssym']),
 247                 Value('r_info_type', lambda ctx: ctx['r_type']),
 248                 Value('r_info_type2', lambda ctx: ctx['r_type2']),
 249                 Value('r_info_type3', lambda ctx: ctx['r_type3']),
 250                 Value('r_info',
 251                       lambda ctx: (ctx['r_sym'] << 32)
 252                                   | (ctx['r_ssym'] << 24)
 253                                   | (ctx['r_type3'] << 16)
 254                                   | (ctx['r_type2'] << 8)
 255                                   | ctx['r_type']),
 256             ]
 257         else: # Other 64 ELFs
 258             fields = [self.Elf_xword('r_info'),
 259                       Value('r_info_sym',
 260                             lambda ctx: (ctx['r_info'] >> 32) & 0xFFFFFFFF),
 261                       Value('r_info_type',
 262                             lambda ctx: ctx['r_info'] & 0xFFFFFFFF)]
 263
 264         self.Elf_Rel = Struct('Elf_Rel',
 265                               self.Elf_addr('r_offset'),
 266                               *fields)
 267
 268         fields_and_addend = fields + [self.Elf_sxword('r_addend')]
 269         self.Elf_Rela = Struct('Elf_Rela',
 270                                self.Elf_addr('r_offset'),
 271                                *fields_and_addend
 272         )
 273
 274     def _create_dyn(self):
 275         d_tag_dict = dict(ENUM_D_TAG_COMMON)
 276         if self.e_machine in ENUMMAP_EXTRA_D_TAG_MACHINE:
 277             d_tag_dict.update(ENUMMAP_EXTRA_D_TAG_MACHINE[self.e_machine])
 278         elif self.e_ident_osabi == 'ELFOSABI_SOLARIS':
 279             d_tag_dict.update(ENUM_D_TAG_SOLARIS)
 280
 281         self.Elf_Dyn = Struct('Elf_Dyn',
 282             Enum(self.Elf_sxword('d_tag'), **d_tag_dict),
 283             self.Elf_xword('d_val'),
 284             Value('d_ptr', lambda ctx: ctx['d_val']),
 285         )
 286
 287     def _create_sym(self):
 288         # st_info is hierarchical. To access the type, use
 289         # container['st_info']['type']
 290         st_info_struct = BitStruct('st_info',
 291             Enum(BitField('bind', 4), **ENUM_ST_INFO_BIND),
 292             Enum(BitField('type', 4), **ENUM_ST_INFO_TYPE))
 293         # st_other is hierarchical. To access the visibility,
 294         # use container['st_other']['visibility']
 295         st_other_struct = BitStruct('st_other',
 296             # https://openpowerfoundation.org/wp-content/uploads/2016/03/ABI64BitOpenPOWERv1.1_16July2015_pub4.pdf
 297             # See 3.4.1 Symbol Values.
 298             Enum(BitField('local', 3), **ENUM_ST_LOCAL),
 299             Padding(2),
 300             Enum(BitField('visibility', 3), **ENUM_ST_VISIBILITY))
 301         if self.elfclass == 32:
 302             self.Elf_Sym = Struct('Elf_Sym',
 303                 self.Elf_word('st_name'),
 304                 self.Elf_addr('st_value'),
 305                 self.Elf_word('st_size'),
 306                 st_info_struct,
 307                 st_other_struct,
 308                 Enum(self.Elf_half('st_shndx'), **ENUM_ST_SHNDX),
 309             )
 310         else:
 311             self.Elf_Sym = Struct('Elf_Sym',
 312                 self.Elf_word('st_name'),
 313                 st_info_struct,
 314                 st_other_struct,
 315                 Enum(self.Elf_half('st_shndx'), **ENUM_ST_SHNDX),
 316                 self.Elf_addr('st_value'),
 317                 self.Elf_xword('st_size'),
 318             )
 319
 320     def _create_sunw_syminfo(self):
 321         self.Elf_Sunw_Syminfo = Struct('Elf_Sunw_Syminfo',
 322             Enum(self.Elf_half('si_boundto'), **ENUM_SUNW_SYMINFO_BOUNDTO),
 323             self.Elf_half('si_flags'),
 324         )
 325
 326     def _create_gnu_verneed(self):
 327         # Structure of "version needed" entries is documented in
 328         # Oracle "Linker and Libraries Guide", Chapter 13 Object File Format
 329         self.Elf_Verneed = Struct('Elf_Verneed',
 330             self.Elf_half('vn_version'),
 331             self.Elf_half('vn_cnt'),
 332             self.Elf_word('vn_file'),
 333             self.Elf_word('vn_aux'),
 334             self.Elf_word('vn_next'),
 335         )
 336         self.Elf_Vernaux = Struct('Elf_Vernaux',
 337             self.Elf_word('vna_hash'),
 338             self.Elf_half('vna_flags'),
 339             self.Elf_half('vna_other'),
 340             self.Elf_word('vna_name'),
 341             self.Elf_word('vna_next'),
 342         )
 343
 344     def _create_gnu_verdef(self):
 345         # Structure of "version definition" entries are documented in
 346         # Oracle "Linker and Libraries Guide", Chapter 13 Object File Format
 347         self.Elf_Verdef = Struct('Elf_Verdef',
 348             self.Elf_half('vd_version'),
 349             self.Elf_half('vd_flags'),
 350             self.Elf_half('vd_ndx'),
 351             self.Elf_half('vd_cnt'),
 352             self.Elf_word('vd_hash'),
 353             self.Elf_word('vd_aux'),
 354             self.Elf_word('vd_next'),
 355         )
 356         self.Elf_Verdaux = Struct('Elf_Verdaux',
 357             self.Elf_word('vda_name'),
 358             self.Elf_word('vda_next'),
 359         )
 360
 361     def _create_gnu_versym(self):
 362         # Structure of "version symbol" entries are documented in
 363         # Oracle "Linker and Libraries Guide", Chapter 13 Object File Format
 364         self.Elf_Versym = Struct('Elf_Versym',
 365             Enum(self.Elf_half('ndx'), **ENUM_VERSYM),
 366         )
 367
 368     def _create_gnu_abi(self):
 369         # Structure of GNU ABI notes is documented in
 370         # https://code.woboq.org/userspace/glibc/csu/abi-note.S.html
 371         self.Elf_abi = Struct('Elf_abi',
 372             Enum(self.Elf_word('abi_os'), **ENUM_NOTE_ABI_TAG_OS),
 373             self.Elf_word('abi_major'),
 374             self.Elf_word('abi_minor'),
 375             self.Elf_word('abi_tiny'),
 376         )
 377
 378     def _create_gnu_property(self):
 379         # Structure of GNU property notes is documented in
 380         # https://github.com/hjl-tools/linux-abi/wiki/linux-abi-draft.pdf
 381         def roundup_padding(ctx):
 382             if self.elfclass == 32:
 383                 return roundup(ctx.pr_datasz, 2) - ctx.pr_datasz
 384             return roundup(ctx.pr_datasz, 3) - ctx.pr_datasz
 385
 386         self.Elf_Prop = Struct('Elf_Prop',
 387             Enum(self.Elf_word('pr_type'), **ENUM_NOTE_GNU_PROPERTY_TYPE),
 388             self.Elf_word('pr_datasz'),
 389             Switch('pr_data',
 390                 lambda ctx: (ctx.pr_type, ctx.pr_datasz, self.elfclass),
 391                 {
 392                     ('GNU_PROPERTY_STACK_SIZE', 4, 32): self.Elf_word('pr_data'),
 393                     ('GNU_PROPERTY_STACK_SIZE', 8, 64): self.Elf_word64('pr_data')
 394                 },
 395                 default=Field('pr_data', lambda ctx: ctx.pr_datasz)
 396             ),
 397             Padding(roundup_padding)
 398         )
 399
 400     def _create_note(self, e_type=None):
 401         # Structure of "PT_NOTE" section
 402
 403         self.Elf_ugid = self.Elf_half if self.elfclass == 32 and self.e_machine in {
 404             'EM_MN10300',
 405             'EM_ARM',
 406             'EM_CRIS',
 407             'EM_CYGNUS_FRV',
 408             'EM_386',
 409             'EM_M32R',
 410             'EM_68K',
 411             'EM_S390',
 412             'EM_SH',
 413             'EM_SPARC',
 414         } else self.Elf_word
 415
 416         self.Elf_Nhdr = Struct('Elf_Nhdr',
 417             self.Elf_word('n_namesz'),
 418             self.Elf_word('n_descsz'),
 419             Enum(self.Elf_word('n_type'),
 420                  **(ENUM_NOTE_N_TYPE if e_type != "ET_CORE"
 421                     else ENUM_CORE_NOTE_N_TYPE)),
 422         )
 423
 424         # A process psinfo structure according to
 425         # http://elixir.free-electrons.com/linux/v2.6.35/source/include/linux/elfcore.h#L84
 426         if self.elfclass == 32:
 427             self.Elf_Prpsinfo = Struct('Elf_Prpsinfo',
 428                 self.Elf_byte('pr_state'),
 429                 String('pr_sname', 1),
 430                 self.Elf_byte('pr_zomb'),
 431                 self.Elf_byte('pr_nice'),
 432                 self.Elf_xword('pr_flag'),
 433                 self.Elf_ugid('pr_uid'),
 434                 self.Elf_ugid('pr_gid'),
 435                 self.Elf_word('pr_pid'),
 436                 self.Elf_word('pr_ppid'),
 437                 self.Elf_word('pr_pgrp'),
 438                 self.Elf_word('pr_sid'),
 439                 String('pr_fname', 16),
 440                 String('pr_psargs', 80),
 441             )
 442         else: # 64
 443             self.Elf_Prpsinfo = Struct('Elf_Prpsinfo',
 444                 self.Elf_byte('pr_state'),
 445                 String('pr_sname', 1),
 446                 self.Elf_byte('pr_zomb'),
 447                 self.Elf_byte('pr_nice'),
 448                 Padding(4),
 449                 self.Elf_xword('pr_flag'),
 450                 self.Elf_ugid('pr_uid'),
 451                 self.Elf_ugid('pr_gid'),
 452                 self.Elf_word('pr_pid'),
 453                 self.Elf_word('pr_ppid'),
 454                 self.Elf_word('pr_pgrp'),
 455                 self.Elf_word('pr_sid'),
 456                 String('pr_fname', 16),
 457                 String('pr_psargs', 80),
 458             )
 459
 460         # A PT_NOTE of type NT_FILE matching the definition in
 461         # https://chromium.googlesource.com/
 462         # native_client/nacl-binutils/+/upstream/master/binutils/readelf.c
 463         # Line 15121
 464         self.Elf_Nt_File = Struct('Elf_Nt_File',
 465                                   self.Elf_xword("num_map_entries"),
 466                                   self.Elf_xword("page_size"),
 467                                   Array(lambda ctx: ctx.num_map_entries,
 468                                         Struct('Elf_Nt_File_Entry',
 469                                              self.Elf_addr('vm_start'),
 470                                              self.Elf_addr('vm_end'),
 471                                              self.Elf_offset('page_offset'))),
 472                                   Array(lambda ctx: ctx.num_map_entries,
 473                                         CString('filename')))
 474
 475     def _create_stabs(self):
 476         # Structure of one stabs entry, see binutils/bfd/stabs.c
 477         # Names taken from https://sourceware.org/gdb/current/onlinedocs/stabs.html#Overview
 478         self.Elf_Stabs = Struct('Elf_Stabs',
 479             self.Elf_word('n_strx'),
 480             self.Elf_byte('n_type'),
 481             self.Elf_byte('n_other'),
 482             self.Elf_half('n_desc'),
 483             self.Elf_word('n_value'),
 484         )
 485
 486     def _create_arm_attributes(self):
 487         # Structure of a build attributes subsection header. A subsection is
 488         # either public to all tools that process the ELF file or private to
 489         # the vendor's tools.
 490         self.Elf_Attr_Subsection_Header = Struct('Elf_Attr_Subsection',
 491                                                  self.Elf_word('length'),
 492                                                  self.Elf_ntbs('vendor_name',
 493                                                                encoding='utf-8')
 494         )
 495
 496         # Structure of a build attribute tag.
 497         self.Elf_Attribute_Tag = Struct('Elf_Attribute_Tag',
 498                                         Enum(self.Elf_uleb128('tag'),
 499                                              **ENUM_ATTR_TAG_ARM)
 500         )
 501
 502     def _create_elf_hash(self):
 503         # Structure of the old SYSV-style hash table header. It is documented
 504         # in the Oracle "Linker and Libraries Guide", Part IV ELF Application
 505         # Binary Interface, Chapter 14 Object File Format, Section Hash Table
 506         # Section:
 507         # https://docs.oracle.com/cd/E53394_01/html/E54813/chapter6-48031.html
 508
 509         self.Elf_Hash = Struct('Elf_Hash',
 510                                self.Elf_word('nbuckets'),
 511                                self.Elf_word('nchains'),
 512                                Array(lambda ctx: ctx['nbuckets'], self.Elf_word('buckets')),
 513                                Array(lambda ctx: ctx['nchains'], self.Elf_word('chains')))
 514
 515     def _create_gnu_hash(self):
 516         # Structure of the GNU-style hash table header. Documentation for this
 517         # table is mostly in the GLIBC source code, a good explanation of the
 518         # format can be found in this blog post:
 519         # https://flapenguin.me/2017/05/10/elf-lookup-dt-gnu-hash/
 520         self.Gnu_Hash = Struct('Gnu_Hash',
 521                                self.Elf_word('nbuckets'),
 522                                self.Elf_word('symoffset'),
 523                                self.Elf_word('bloom_size'),
 524                                self.Elf_word('bloom_shift'),
 525                                Array(lambda ctx: ctx['bloom_size'], self.Elf_xword('bloom')),
 526                                Array(lambda ctx: ctx['nbuckets'], self.Elf_word('buckets')))