elftools/elf/elffile.py

   1 #-------------------------------------------------------------------------------
   2 # elftools: elf/elffile.py
   3 #
   4 # ELFFile - main class for accessing ELF files
   5 #
   6 # Eli Bendersky (eliben@gmail.com)
   7 # This code is in the public domain
   8 #-------------------------------------------------------------------------------
   9 import io
  10 import struct
  11 import zlib
  12
  13 try:
  14     import resource
  15     PAGESIZE = resource.getpagesize()
  16 except ImportError:
  17     try:
  18         # Windows system
  19         import mmap
  20         PAGESIZE = mmap.PAGESIZE
  21     except ImportError:
  22         # Jython
  23         PAGESIZE = 4096
  24
  25 from ..common.py3compat import BytesIO
  26 from ..common.exceptions import ELFError
  27 from ..common.utils import struct_parse, elf_assert
  28 from .structs import ELFStructs
  29 from .sections import (
  30         Section, StringTableSection, SymbolTableSection,
  31         SymbolTableIndexSection, SUNWSyminfoTableSection, NullSection,
  32         NoteSection, StabSection, ARMAttributesSection)
  33 from .dynamic import DynamicSection, DynamicSegment
  34 from .relocation import RelocationSection, RelocationHandler
  35 from .gnuversions import (
  36         GNUVerNeedSection, GNUVerDefSection,
  37         GNUVerSymSection)
  38 from .segments import Segment, InterpSegment, NoteSegment
  39 from ..dwarf.dwarfinfo import DWARFInfo, DebugSectionDescriptor, DwarfConfig
  40 from ..ehabi.ehabiinfo import EHABIInfo
  41 from .hash import ELFHashSection, GNUHashSection
  42 from .constants import SHN_INDICES
  43
  44 class ELFFile(object):
  45     """ Creation: the constructor accepts a stream (file-like object) with the
  46         contents of an ELF file.
  47
  48         Accessible attributes:
  49
  50             stream:
  51                 The stream holding the data of the file - must be a binary
  52                 stream (bytes, not string).
  53
  54             elfclass:
  55                 32 or 64 - specifies the word size of the target machine
  56
  57             little_endian:
  58                 boolean - specifies the target machine's endianness
  59
  60             elftype:
  61                 string or int, either known value of E_TYPE enum defining ELF
  62                 type (e.g. executable, dynamic library or core dump) or integral
  63                 unparsed value
  64
  65             header:
  66                 the complete ELF file header
  67
  68             e_ident_raw:
  69                 the raw e_ident field of the header
  70     """
  71     def __init__(self, stream):
  72         self.stream = stream
  73         self._identify_file()
  74         self.structs = ELFStructs(
  75             little_endian=self.little_endian,
  76             elfclass=self.elfclass)
  77
  78         self.structs.create_basic_structs()
  79         self.header = self._parse_elf_header()
  80         self.structs.create_advanced_structs(
  81                 self['e_type'],
  82                 self['e_machine'],
  83                 self['e_ident']['EI_OSABI'])
  84         self.stream.seek(0)
  85         self.e_ident_raw = self.stream.read(16)
  86
  87         self._section_header_stringtable = \
  88             self._get_section_header_stringtable()
  89         self._section_name_map = None
  90
  91     def num_sections(self):
  92         """ Number of sections in the file
  93         """
  94         if self['e_shoff'] == 0:
  95             return 0
  96         # From the ELF ABI documentation at
  97         # https://refspecs.linuxfoundation.org/elf/gabi4+/ch4.sheader.html:
  98         # "e_shnum normally tells how many entries the section header table
  99         # contains. [...] If the number of sections is greater than or equal to
 100         # SHN_LORESERVE (0xff00), e_shnum has the value SHN_UNDEF (0) and the
 101         # actual number of section header table entries is contained in the
 102         # sh_size field of the section header at index 0 (otherwise, the sh_size
 103         # member of the initial entry contains 0)."
 104         if self['e_shnum'] == 0:
 105             return self._get_section_header(0)['sh_size']
 106         return self['e_shnum']
 107
 108     def get_section(self, n):
 109         """ Get the section at index #n from the file (Section object or a
 110             subclass)
 111         """
 112         section_header = self._get_section_header(n)
 113         return self._make_section(section_header)
 114
 115     def get_section_by_name(self, name):
 116         """ Get a section from the file, by name. Return None if no such
 117             section exists.
 118         """
 119         # The first time this method is called, construct a name to number
 120         # mapping
 121         #
 122         if self._section_name_map is None:
 123             self._make_section_name_map()
 124         secnum = self._section_name_map.get(name, None)
 125         return None if secnum is None else self.get_section(secnum)
 126
 127     def get_section_index(self, section_name):
 128         """ Gets the index of the section by name. Return None if no such
 129             section name exists.
 130         """
 131         # The first time this method is called, construct a name to number
 132         # mapping
 133         #
 134         if self._section_name_map is None:
 135             self._make_section_name_map()
 136         return self._section_name_map.get(section_name, None)
 137
 138     def iter_sections(self, type=None):
 139         """ Yield all the sections in the file. If the optional |type|
 140             parameter is passed, this method will only yield sections of the
 141             given type. The parameter value must be a string containing the
 142             name of the type as defined in the ELF specification, e.g.
 143             'SHT_SYMTAB'.
 144         """
 145         for i in range(self.num_sections()):
 146             section = self.get_section(i)
 147             if type is None or section['sh_type'] == type:
 148                 yield section
 149
 150     def num_segments(self):
 151         """ Number of segments in the file
 152         """
 153         # From: https://github.com/hjl-tools/x86-psABI/wiki/X86-psABI
 154         # Section: 4.1.2 Number of Program Headers
 155         # If the number of program headers is greater than or equal to
 156         # PN_XNUM (0xffff), this member has the value PN_XNUM
 157         # (0xffff). The actual number of program header table entries
 158         # is contained in the sh_info field of the section header at
 159         # index 0.
 160         if self['e_phnum'] < 0xffff:
 161             return self['e_phnum']
 162         else:
 163             return self.get_section(0)['sh_info']
 164
 165     def get_segment(self, n):
 166         """ Get the segment at index #n from the file (Segment object)
 167         """
 168         segment_header = self._get_segment_header(n)
 169         return self._make_segment(segment_header)
 170
 171     def iter_segments(self, type=None):
 172         """ Yield all the segments in the file. If the optional |type|
 173             parameter is passed, this method will only yield segments of the
 174             given type. The parameter value must be a string containing the
 175             name of the type as defined in the ELF specification, e.g.
 176             'PT_LOAD'.
 177         """
 178         for i in range(self.num_segments()):
 179             segment = self.get_segment(i)
 180             if type is None or segment['p_type'] == type:
 181                 yield segment
 182
 183     def address_offsets(self, start, size=1):
 184         """ Yield a file offset for each ELF segment containing a memory region.
 185
 186             A memory region is defined by the range [start...start+size). The
 187             offset of the region is yielded.
 188         """
 189         end = start + size
 190         # consider LOAD only to prevent same address being yielded twice
 191         for seg in self.iter_segments(type='PT_LOAD'):
 192             if (start >= seg['p_vaddr'] and
 193                 end <= seg['p_vaddr'] + seg['p_filesz']):
 194                 yield start - seg['p_vaddr'] + seg['p_offset']
 195
 196     def has_dwarf_info(self):
 197         """ Check whether this file appears to have debugging information.
 198             We assume that if it has the .debug_info or .zdebug_info section, it
 199             has all the other required sections as well.
 200         """
 201         return bool(self.get_section_by_name('.debug_info') or
 202             self.get_section_by_name('.zdebug_info') or
 203             self.get_section_by_name('.eh_frame'))
 204
 205     def get_dwarf_info(self, relocate_dwarf_sections=True):
 206         """ Return a DWARFInfo object representing the debugging information in
 207             this file.
 208
 209             If relocate_dwarf_sections is True, relocations for DWARF sections
 210             are looked up and applied.
 211         """
 212         # Expect that has_dwarf_info was called, so at least .debug_info is
 213         # present.
 214         # Sections that aren't found will be passed as None to DWARFInfo.
 215
 216         section_names = ('.debug_info', '.debug_aranges', '.debug_abbrev',
 217                          '.debug_str', '.debug_line', '.debug_frame',
 218                          '.debug_loc', '.debug_ranges', '.debug_pubtypes',
 219                          '.debug_pubnames', '.debug_addr', '.debug_str_offsets')
 220
 221         compressed = bool(self.get_section_by_name('.zdebug_info'))
 222         if compressed:
 223             section_names = tuple(map(lambda x: '.z' + x[1:], section_names))
 224
 225         # As it is loaded in the process image, .eh_frame cannot be compressed
 226         section_names += ('.eh_frame', )
 227
 228         (debug_info_sec_name, debug_aranges_sec_name, debug_abbrev_sec_name,
 229          debug_str_sec_name, debug_line_sec_name, debug_frame_sec_name,
 230          debug_loc_sec_name, debug_ranges_sec_name, debug_pubtypes_name,
 231          debug_pubnames_name, debug_addr_name, debug_str_offsets_name,
 232          eh_frame_sec_name) = section_names
 233
 234         debug_sections = {}
 235         for secname in section_names:
 236             section = self.get_section_by_name(secname)
 237             if section is None:
 238                 debug_sections[secname] = None
 239             else:
 240                 dwarf_section = self._read_dwarf_section(
 241                     section,
 242                     relocate_dwarf_sections)
 243                 if compressed and secname.startswith('.z'):
 244                     dwarf_section = self._decompress_dwarf_section(dwarf_section)
 245                 debug_sections[secname] = dwarf_section
 246
 247         return DWARFInfo(
 248                 config=DwarfConfig(
 249                     little_endian=self.little_endian,
 250                     default_address_size=self.elfclass // 8,
 251                     machine_arch=self.get_machine_arch()),
 252                 debug_info_sec=debug_sections[debug_info_sec_name],
 253                 debug_aranges_sec=debug_sections[debug_aranges_sec_name],
 254                 debug_abbrev_sec=debug_sections[debug_abbrev_sec_name],
 255                 debug_frame_sec=debug_sections[debug_frame_sec_name],
 256                 eh_frame_sec=debug_sections[eh_frame_sec_name],
 257                 debug_str_sec=debug_sections[debug_str_sec_name],
 258                 debug_loc_sec=debug_sections[debug_loc_sec_name],
 259                 debug_ranges_sec=debug_sections[debug_ranges_sec_name],
 260                 debug_line_sec=debug_sections[debug_line_sec_name],
 261                 debug_pubtypes_sec=debug_sections[debug_pubtypes_name],
 262                 debug_pubnames_sec=debug_sections[debug_pubnames_name],
 263                 debug_addr_sec=debug_sections[debug_addr_name],
 264                 debug_str_offsets_sec=debug_sections[debug_str_offsets_name],
 265                 )
 266
 267     def has_ehabi_info(self):
 268         """ Check whether this file appears to have arm exception handler index table.
 269         """
 270         return any(self.iter_sections(type='SHT_ARM_EXIDX'))
 271
 272     def get_ehabi_infos(self):
 273         """ Generally, shared library and executable contain 1 .ARM.exidx section.
 274             Object file contains many .ARM.exidx sections.
 275             So we must traverse every section and filter sections whose type is SHT_ARM_EXIDX.
 276         """
 277         _ret = []
 278         if self['e_type'] == 'ET_REL':
 279             # TODO: support relocatable file
 280             assert False, "Current version of pyelftools doesn't support relocatable file."
 281         for section in self.iter_sections(type='SHT_ARM_EXIDX'):
 282             _ret.append(EHABIInfo(section, self.little_endian))
 283         return _ret if len(_ret) > 0 else None
 284
 285     def get_machine_arch(self):
 286         """ Return the machine architecture, as detected from the ELF header.
 287         """
 288         architectures = {
 289             'EM_M32'           : 'AT&T WE 32100',
 290             'EM_SPARC'         : 'SPARC',
 291             'EM_386'           : 'x86',
 292             'EM_68K'           : 'Motorola 68000',
 293             'EM_88K'           : 'Motorola 88000',
 294             'EM_IAMCU'         : 'Intel MCU',
 295             'EM_860'           : 'Intel 80860',
 296             'EM_MIPS'          : 'MIPS',
 297             'EM_S370'          : 'IBM System/370',
 298             'EM_MIPS_RS3_LE'   : 'MIPS RS3000 Little-endian',
 299             'EM_PARISC'        : 'Hewlett-Packard PA-RISC',
 300             'EM_VPP500'        : 'Fujitsu VPP500',
 301             'EM_SPARC32PLUS'   : 'Enhanced SPARC',
 302             'EM_960'           : 'Intel 80960',
 303             'EM_PPC'           : 'PowerPC',
 304             'EM_PPC64'         : '64-bit PowerPC',
 305             'EM_S390'          : 'IBM System/390',
 306             'EM_SPU'           : 'IBM SPU/SPC',
 307             'EM_V800'          : 'NEC V800',
 308             'EM_FR20'          : 'Fujitsu FR20',
 309             'EM_RH32'          : 'TRW RH-32',
 310             'EM_RCE'           : 'Motorola RCE',
 311             'EM_ARM'           : 'ARM',
 312             'EM_ALPHA'         : 'Digital Alpha',
 313             'EM_SH'            : 'Hitachi SH',
 314             'EM_SPARCV9'       : 'SPARC Version 9',
 315             'EM_TRICORE'       : 'Siemens TriCore embedded processor',
 316             'EM_ARC'           : 'Argonaut RISC Core, Argonaut Technologies Inc.',
 317             'EM_H8_300'        : 'Hitachi H8/300',
 318             'EM_H8_300H'       : 'Hitachi H8/300H',
 319             'EM_H8S'           : 'Hitachi H8S',
 320             'EM_H8_500'        : 'Hitachi H8/500',
 321             'EM_IA_64'         : 'Intel IA-64',
 322             'EM_MIPS_X'        : 'MIPS-X',
 323             'EM_COLDFIRE'      : 'Motorola ColdFire',
 324             'EM_68HC12'        : 'Motorola M68HC12',
 325             'EM_MMA'           : 'Fujitsu MMA',
 326             'EM_PCP'           : 'Siemens PCP',
 327             'EM_NCPU'          : 'Sony nCPU',
 328             'EM_NDR1'          : 'Denso NDR1',
 329             'EM_STARCORE'      : 'Motorola Star*Core',
 330             'EM_ME16'          : 'Toyota ME16',
 331             'EM_ST100'         : 'STMicroelectronics ST100',
 332             'EM_TINYJ'         : 'Advanced Logic TinyJ',
 333             'EM_X86_64'        : 'x64',
 334             'EM_PDSP'          : 'Sony DSP',
 335             'EM_PDP10'         : 'Digital Equipment PDP-10',
 336             'EM_PDP11'         : 'Digital Equipment PDP-11',
 337             'EM_FX66'          : 'Siemens FX66',
 338             'EM_ST9PLUS'       : 'STMicroelectronics ST9+ 8/16 bit',
 339             'EM_ST7'           : 'STMicroelectronics ST7 8-bit',
 340             'EM_68HC16'        : 'Motorola MC68HC16',
 341             'EM_68HC11'        : 'Motorola MC68HC11',
 342             'EM_68HC08'        : 'Motorola MC68HC08',
 343             'EM_68HC05'        : 'Motorola MC68HC05',
 344             'EM_SVX'           : 'Silicon Graphics SVx',
 345             'EM_ST19'          : 'STMicroelectronics ST19 8-bit',
 346             'EM_VAX'           : 'Digital VAX',
 347             'EM_CRIS'          : 'Axis Communications 32-bit',
 348             'EM_JAVELIN'       : 'Infineon Technologies 32-bit',
 349             'EM_FIREPATH'      : 'Element 14 64-bit DSP',
 350             'EM_ZSP'           : 'LSI Logic 16-bit DSP',
 351             'EM_MMIX'          : 'Donald Knuth\'s educational 64-bit',
 352             'EM_HUANY'         : 'Harvard University machine-independent object files',
 353             'EM_PRISM'         : 'SiTera Prism',
 354             'EM_AVR'           : 'Atmel AVR 8-bit',
 355             'EM_FR30'          : 'Fujitsu FR30',
 356             'EM_D10V'          : 'Mitsubishi D10V',
 357             'EM_D30V'          : 'Mitsubishi D30V',
 358             'EM_V850'          : 'NEC v850',
 359             'EM_M32R'          : 'Mitsubishi M32R',
 360             'EM_MN10300'       : 'Matsushita MN10300',
 361             'EM_MN10200'       : 'Matsushita MN10200',
 362             'EM_PJ'            : 'picoJava',
 363             'EM_OPENRISC'      : 'OpenRISC 32-bit',
 364             'EM_ARC_COMPACT'   : 'ARC International ARCompact',
 365             'EM_XTENSA'        : 'Tensilica Xtensa',
 366             'EM_VIDEOCORE'     : 'Alphamosaic VideoCore',
 367             'EM_TMM_GPP'       : 'Thompson Multimedia',
 368             'EM_NS32K'         : 'National Semiconductor 32000 series',
 369             'EM_TPC'           : 'Tenor Network TPC',
 370             'EM_SNP1K'         : 'Trebia SNP 1000',
 371             'EM_ST200'         : 'STMicroelectronics ST200',
 372             'EM_IP2K'          : 'Ubicom IP2xxx',
 373             'EM_MAX'           : 'MAX',
 374             'EM_CR'            : 'National Semiconductor CompactRISC',
 375             'EM_F2MC16'        : 'Fujitsu F2MC16',
 376             'EM_MSP430'        : 'Texas Instruments msp430',
 377             'EM_BLACKFIN'      : 'Analog Devices Blackfin',
 378             'EM_SE_C33'        : 'Seiko Epson S1C33',
 379             'EM_SEP'           : 'Sharp',
 380             'EM_ARCA'          : 'Arca RISC',
 381             'EM_UNICORE'       : 'PKU-Unity MPRC',
 382             'EM_EXCESS'        : 'eXcess',
 383             'EM_DXP'           : 'Icera Semiconductor Deep Execution Processor',
 384             'EM_ALTERA_NIOS2'  : 'Altera Nios II',
 385             'EM_CRX'           : 'National Semiconductor CompactRISC CRX',
 386             'EM_XGATE'         : 'Motorola XGATE',
 387             'EM_C166'          : 'Infineon C16x/XC16x',
 388             'EM_M16C'          : 'Renesas M16C',
 389             'EM_DSPIC30F'      : 'Microchip Technology dsPIC30F',
 390             'EM_CE'            : 'Freescale Communication Engine RISC core',
 391             'EM_M32C'          : 'Renesas M32C',
 392             'EM_TSK3000'       : 'Altium TSK3000',
 393             'EM_RS08'          : 'Freescale RS08',
 394             'EM_SHARC'         : 'Analog Devices SHARC',
 395             'EM_ECOG2'         : 'Cyan Technology eCOG2',
 396             'EM_SCORE7'        : 'Sunplus S+core7 RISC',
 397             'EM_DSP24'         : 'New Japan Radio (NJR) 24-bit DSP',
 398             'EM_VIDEOCORE3'    : 'Broadcom VideoCore III',
 399             'EM_LATTICEMICO32' : 'Lattice FPGA RISC',
 400             'EM_SE_C17'        : 'Seiko Epson C17',
 401             'EM_TI_C6000'      : 'TI TMS320C6000',
 402             'EM_TI_C2000'      : 'TI TMS320C2000',
 403             'EM_TI_C5500'      : 'TI TMS320C55x',
 404             'EM_TI_ARP32'      : 'TI Application Specific RISC, 32bit',
 405             'EM_TI_PRU'        : 'TI Programmable Realtime Unit',
 406             'EM_MMDSP_PLUS'    : 'STMicroelectronics 64bit VLIW',
 407             'EM_CYPRESS_M8C'   : 'Cypress M8C',
 408             'EM_R32C'          : 'Renesas R32C',
 409             'EM_TRIMEDIA'      : 'NXP Semiconductors TriMedia',
 410             'EM_QDSP6'         : 'QUALCOMM DSP6',
 411             'EM_8051'          : 'Intel 8051',
 412             'EM_STXP7X'        : 'STMicroelectronics STxP7x',
 413             'EM_NDS32'         : 'Andes Technology RISC',
 414             'EM_ECOG1'         : 'Cyan Technology eCOG1X',
 415             'EM_ECOG1X'        : 'Cyan Technology eCOG1X',
 416             'EM_MAXQ30'        : 'Dallas Semiconductor MAXQ30',
 417             'EM_XIMO16'        : 'New Japan Radio (NJR) 16-bit',
 418             'EM_MANIK'         : 'M2000 Reconfigurable RISC',
 419             'EM_CRAYNV2'       : 'Cray Inc. NV2',
 420             'EM_RX'            : 'Renesas RX',
 421             'EM_METAG'         : 'Imagination Technologies META',
 422             'EM_MCST_ELBRUS'   : 'MCST Elbrus',
 423             'EM_ECOG16'        : 'Cyan Technology eCOG16',
 424             'EM_CR16'          : 'National Semiconductor CompactRISC CR16 16-bit',
 425             'EM_ETPU'          : 'Freescale',
 426             'EM_SLE9X'         : 'Infineon Technologies SLE9X',
 427             'EM_L10M'          : 'Intel L10M',
 428             'EM_K10M'          : 'Intel K10M',
 429             'EM_AARCH64'       : 'AArch64',
 430             'EM_AVR32'         : 'Atmel 32-bit',
 431             'EM_STM8'          : 'STMicroeletronics STM8 8-bit',
 432             'EM_TILE64'        : 'Tilera TILE64',
 433             'EM_TILEPRO'       : 'Tilera TILEPro',
 434             'EM_MICROBLAZE'    : 'Xilinx MicroBlaze 32-bit RISC',
 435             'EM_CUDA'          : 'NVIDIA CUDA',
 436             'EM_TILEGX'        : 'Tilera TILE-Gx',
 437             'EM_CLOUDSHIELD'   : 'CloudShield',
 438             'EM_COREA_1ST'     : 'KIPO-KAIST Core-A 1st generation',
 439             'EM_COREA_2ND'     : 'KIPO-KAIST Core-A 2nd generation',
 440             'EM_ARC_COMPACT2'  : 'Synopsys ARCompact V2',
 441             'EM_OPEN8'         : 'Open8 8-bit RISC',
 442             'EM_RL78'          : 'Renesas RL78',
 443             'EM_VIDEOCORE5'    : 'Broadcom VideoCore V',
 444             'EM_78KOR'         : 'Renesas 78KOR',
 445             'EM_56800EX'       : 'Freescale 56800EX',
 446             'EM_BA1'           : 'Beyond BA1',
 447             'EM_BA2'           : 'Beyond BA2',
 448             'EM_XCORE'         : 'XMOS xCORE',
 449             'EM_MCHP_PIC'      : 'Microchip 8-bit PIC',
 450             'EM_INTEL205'      : 'Reserved by Intel',
 451             'EM_INTEL206'      : 'Reserved by Intel',
 452             'EM_INTEL207'      : 'Reserved by Intel',
 453             'EM_INTEL208'      : 'Reserved by Intel',
 454             'EM_INTEL209'      : 'Reserved by Intel',
 455             'EM_KM32'          : 'KM211 KM32 32-bit',
 456             'EM_KMX32'         : 'KM211 KMX32 32-bit',
 457             'EM_KMX16'         : 'KM211 KMX16 16-bit',
 458             'EM_KMX8'          : 'KM211 KMX8 8-bit',
 459             'EM_KVARC'         : 'KM211 KVARC',
 460             'EM_CDP'           : 'Paneve CDP',
 461             'EM_COGE'          : 'Cognitive',
 462             'EM_COOL'          : 'Bluechip Systems CoolEngine',
 463             'EM_NORC'          : 'Nanoradio Optimized RISC',
 464             'EM_CSR_KALIMBA'   : 'CSR Kalimba',
 465             'EM_Z80'           : 'Zilog Z80',
 466             'EM_VISIUM'        : 'VISIUMcore',
 467             'EM_FT32'          : 'FTDI Chip FT32 32-bit RISC',
 468             'EM_MOXIE'         : 'Moxie',
 469             'EM_AMDGPU'        : 'AMD GPU',
 470             'EM_RISCV'         : 'RISC-V'
 471         }
 472
 473         return architectures.get(self['e_machine'], '<unknown>')
 474
 475     def get_shstrndx(self):
 476         """ Find the string table section index for the section header table
 477         """
 478         # From https://refspecs.linuxfoundation.org/elf/gabi4+/ch4.eheader.html:
 479         # If the section name string table section index is greater than or
 480         # equal to SHN_LORESERVE (0xff00), this member has the value SHN_XINDEX
 481         # (0xffff) and the actual index of the section name string table section
 482         # is contained in the sh_link field of the section header at index 0.
 483         if self['e_shstrndx'] != SHN_INDICES.SHN_XINDEX:
 484             return self['e_shstrndx']
 485         else:
 486             return self._get_section_header(0)['sh_link']
 487
 488     #-------------------------------- PRIVATE --------------------------------#
 489
 490     def __getitem__(self, name):
 491         """ Implement dict-like access to header entries
 492         """
 493         return self.header[name]
 494
 495     def _identify_file(self):
 496         """ Verify the ELF file and identify its class and endianness.
 497         """
 498         # Note: this code reads the stream directly, without using ELFStructs,
 499         # since we don't yet know its exact format. ELF was designed to be
 500         # read like this - its e_ident field is word-size and endian agnostic.
 501         self.stream.seek(0)
 502         magic = self.stream.read(4)
 503         elf_assert(magic == b'\x7fELF', 'Magic number does not match')
 504
 505         ei_class = self.stream.read(1)
 506         if ei_class == b'\x01':
 507             self.elfclass = 32
 508         elif ei_class == b'\x02':
 509             self.elfclass = 64
 510         else:
 511             raise ELFError('Invalid EI_CLASS %s' % repr(ei_class))
 512
 513         ei_data = self.stream.read(1)
 514         if ei_data == b'\x01':
 515             self.little_endian = True
 516         elif ei_data == b'\x02':
 517             self.little_endian = False
 518         else:
 519             raise ELFError('Invalid EI_DATA %s' % repr(ei_data))
 520
 521     def _section_offset(self, n):
 522         """ Compute the offset of section #n in the file
 523         """
 524         return self['e_shoff'] + n * self['e_shentsize']
 525
 526     def _segment_offset(self, n):
 527         """ Compute the offset of segment #n in the file
 528         """
 529         return self['e_phoff'] + n * self['e_phentsize']
 530
 531     def _make_segment(self, segment_header):
 532         """ Create a Segment object of the appropriate type
 533         """
 534         segtype = segment_header['p_type']
 535         if segtype == 'PT_INTERP':
 536             return InterpSegment(segment_header, self.stream)
 537         elif segtype == 'PT_DYNAMIC':
 538             return DynamicSegment(segment_header, self.stream, self)
 539         elif segtype == 'PT_NOTE':
 540             return NoteSegment(segment_header, self.stream, self)
 541         else:
 542             return Segment(segment_header, self.stream)
 543
 544     def _get_section_header(self, n):
 545         """ Find the header of section #n, parse it and return the struct
 546         """
 547         return struct_parse(
 548             self.structs.Elf_Shdr,
 549             self.stream,
 550             stream_pos=self._section_offset(n))
 551
 552     def _get_section_name(self, section_header):
 553         """ Given a section header, find this section's name in the file's
 554             string table
 555         """
 556         name_offset = section_header['sh_name']
 557         return self._section_header_stringtable.get_string(name_offset)
 558
 559     def _make_section(self, section_header):
 560         """ Create a section object of the appropriate type
 561         """
 562         name = self._get_section_name(section_header)
 563         sectype = section_header['sh_type']
 564
 565         if sectype == 'SHT_STRTAB':
 566             return StringTableSection(section_header, name, self)
 567         elif sectype == 'SHT_NULL':
 568             return NullSection(section_header, name, self)
 569         elif sectype in ('SHT_SYMTAB', 'SHT_DYNSYM', 'SHT_SUNW_LDYNSYM'):
 570             return self._make_symbol_table_section(section_header, name)
 571         elif sectype == 'SHT_SYMTAB_SHNDX':
 572             return self._make_symbol_table_index_section(section_header, name)
 573         elif sectype == 'SHT_SUNW_syminfo':
 574             return self._make_sunwsyminfo_table_section(section_header, name)
 575         elif sectype == 'SHT_GNU_verneed':
 576             return self._make_gnu_verneed_section(section_header, name)
 577         elif sectype == 'SHT_GNU_verdef':
 578             return self._make_gnu_verdef_section(section_header, name)
 579         elif sectype == 'SHT_GNU_versym':
 580             return self._make_gnu_versym_section(section_header, name)
 581         elif sectype in ('SHT_REL', 'SHT_RELA'):
 582             return RelocationSection(section_header, name, self)
 583         elif sectype == 'SHT_DYNAMIC':
 584             return DynamicSection(section_header, name, self)
 585         elif sectype == 'SHT_NOTE':
 586             return NoteSection(section_header, name, self)
 587         elif sectype == 'SHT_PROGBITS' and name == '.stab':
 588             return StabSection(section_header, name, self)
 589         elif sectype == 'SHT_ARM_ATTRIBUTES':
 590             return ARMAttributesSection(section_header, name, self)
 591         elif sectype == 'SHT_HASH':
 592             return self._make_elf_hash_section(section_header, name)
 593         elif sectype == 'SHT_GNU_HASH':
 594             return self._make_gnu_hash_section(section_header, name)
 595         else:
 596             return Section(section_header, name, self)
 597
 598     def _make_section_name_map(self):
 599         self._section_name_map = {}
 600         for i, sec in enumerate(self.iter_sections()):
 601             self._section_name_map[sec.name] = i
 602
 603     def _make_symbol_table_section(self, section_header, name):
 604         """ Create a SymbolTableSection
 605         """
 606         linked_strtab_index = section_header['sh_link']
 607         strtab_section = self.get_section(linked_strtab_index)
 608         return SymbolTableSection(
 609             section_header, name,
 610             elffile=self,
 611             stringtable=strtab_section)
 612
 613     def _make_symbol_table_index_section(self, section_header, name):
 614         """ Create a SymbolTableIndexSection object
 615         """
 616         linked_symtab_index = section_header['sh_link']
 617         return SymbolTableIndexSection(
 618             section_header, name, elffile=self,
 619             symboltable=linked_symtab_index)
 620
 621     def _make_sunwsyminfo_table_section(self, section_header, name):
 622         """ Create a SUNWSyminfoTableSection
 623         """
 624         linked_strtab_index = section_header['sh_link']
 625         strtab_section = self.get_section(linked_strtab_index)
 626         return SUNWSyminfoTableSection(
 627             section_header, name,
 628             elffile=self,
 629             symboltable=strtab_section)
 630
 631     def _make_gnu_verneed_section(self, section_header, name):
 632         """ Create a GNUVerNeedSection
 633         """
 634         linked_strtab_index = section_header['sh_link']
 635         strtab_section = self.get_section(linked_strtab_index)
 636         return GNUVerNeedSection(
 637             section_header, name,
 638             elffile=self,
 639             stringtable=strtab_section)
 640
 641     def _make_gnu_verdef_section(self, section_header, name):
 642         """ Create a GNUVerDefSection
 643         """
 644         linked_strtab_index = section_header['sh_link']
 645         strtab_section = self.get_section(linked_strtab_index)
 646         return GNUVerDefSection(
 647             section_header, name,
 648             elffile=self,
 649             stringtable=strtab_section)
 650
 651     def _make_gnu_versym_section(self, section_header, name):
 652         """ Create a GNUVerSymSection
 653         """
 654         linked_strtab_index = section_header['sh_link']
 655         strtab_section = self.get_section(linked_strtab_index)
 656         return GNUVerSymSection(
 657             section_header, name,
 658             elffile=self,
 659             symboltable=strtab_section)
 660
 661     def _make_elf_hash_section(self, section_header, name):
 662         linked_symtab_index = section_header['sh_link']
 663         symtab_section = self.get_section(linked_symtab_index)
 664         return ELFHashSection(
 665             section_header, name, self, symtab_section
 666         )
 667
 668     def _make_gnu_hash_section(self, section_header, name):
 669         linked_symtab_index = section_header['sh_link']
 670         symtab_section = self.get_section(linked_symtab_index)
 671         return GNUHashSection(
 672             section_header, name, self, symtab_section
 673         )
 674
 675     def _get_segment_header(self, n):
 676         """ Find the header of segment #n, parse it and return the struct
 677         """
 678         return struct_parse(
 679             self.structs.Elf_Phdr,
 680             self.stream,
 681             stream_pos=self._segment_offset(n))
 682
 683     def _get_section_header_stringtable(self):
 684         """ Get the string table section corresponding to the section header
 685             table.
 686         """
 687         stringtable_section_num = self.get_shstrndx()
 688         return StringTableSection(
 689                 header=self._get_section_header(stringtable_section_num),
 690                 name='',
 691                 elffile=self)
 692
 693     def _parse_elf_header(self):
 694         """ Parses the ELF file header and assigns the result to attributes
 695             of this object.
 696         """
 697         return struct_parse(self.structs.Elf_Ehdr, self.stream, stream_pos=0)
 698
 699     def _read_dwarf_section(self, section, relocate_dwarf_sections):
 700         """ Read the contents of a DWARF section from the stream and return a
 701             DebugSectionDescriptor. Apply relocations if asked to.
 702         """
 703         # The section data is read into a new stream, for processing
 704         section_stream = BytesIO()
 705         section_stream.write(section.data())
 706
 707         if relocate_dwarf_sections:
 708             reloc_handler = RelocationHandler(self)
 709             reloc_section = reloc_handler.find_relocations_for_section(section)
 710             if reloc_section is not None:
 711                 reloc_handler.apply_section_relocations(
 712                         section_stream, reloc_section)
 713
 714         return DebugSectionDescriptor(
 715                 stream=section_stream,
 716                 name=section.name,
 717                 global_offset=section['sh_offset'],
 718                 size=section.data_size,
 719                 address=section['sh_addr'])
 720
 721     @staticmethod
 722     def _decompress_dwarf_section(section):
 723         """ Returns the uncompressed contents of the provided DWARF section.
 724         """
 725         # TODO: support other compression formats from readelf.c
 726         assert section.size > 12, 'Unsupported compression format.'
 727
 728         section.stream.seek(0)
 729         # According to readelf.c the content should contain "ZLIB"
 730         # followed by the uncompressed section size - 8 bytes in
 731         # big-endian order
 732         compression_type = section.stream.read(4)
 733         assert compression_type == b'ZLIB', \
 734             'Invalid compression type: %r' % (compression_type)
 735
 736         uncompressed_size = struct.unpack('>Q', section.stream.read(8))[0]
 737
 738         decompressor = zlib.decompressobj()
 739         uncompressed_stream = BytesIO()
 740         while True:
 741             chunk = section.stream.read(PAGESIZE)
 742             if not chunk:
 743                 break
 744             uncompressed_stream.write(decompressor.decompress(chunk))
 745         uncompressed_stream.write(decompressor.flush())
 746
 747         uncompressed_stream.seek(0, io.SEEK_END)
 748         size = uncompressed_stream.tell()
 749         assert uncompressed_size == size, \
 750                 'Wrong uncompressed size: expected %r, but got %r' % (
 751                     uncompressed_size, size,
 752                 )
 753
 754         return section._replace(stream=uncompressed_stream, size=size)