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         RelrRelocationSection)
  36 from .gnuversions import (
  37         GNUVerNeedSection, GNUVerDefSection,
  38         GNUVerSymSection)
  39 from .segments import Segment, InterpSegment, NoteSegment
  40 from ..dwarf.dwarfinfo import DWARFInfo, DebugSectionDescriptor, DwarfConfig
  41 from ..ehabi.ehabiinfo import EHABIInfo
  42 from .hash import ELFHashSection, GNUHashSection
  43 from .constants import SHN_INDICES
  44
  45 class ELFFile(object):
  46     """ Creation: the constructor accepts a stream (file-like object) with the
  47         contents of an ELF file.
  48
  49         Accessible attributes:
  50
  51             stream:
  52                 The stream holding the data of the file - must be a binary
  53                 stream (bytes, not string).
  54
  55             elfclass:
  56                 32 or 64 - specifies the word size of the target machine
  57
  58             little_endian:
  59                 boolean - specifies the target machine's endianness
  60
  61             elftype:
  62                 string or int, either known value of E_TYPE enum defining ELF
  63                 type (e.g. executable, dynamic library or core dump) or integral
  64                 unparsed value
  65
  66             header:
  67                 the complete ELF file header
  68
  69             e_ident_raw:
  70                 the raw e_ident field of the header
  71     """
  72     def __init__(self, stream):
  73         self.stream = stream
  74         self._identify_file()
  75         self.structs = ELFStructs(
  76             little_endian=self.little_endian,
  77             elfclass=self.elfclass)
  78
  79         self.structs.create_basic_structs()
  80         self.header = self._parse_elf_header()
  81         self.structs.create_advanced_structs(
  82                 self['e_type'],
  83                 self['e_machine'],
  84                 self['e_ident']['EI_OSABI'])
  85         self.stream.seek(0)
  86         self.e_ident_raw = self.stream.read(16)
  87
  88         self._section_header_stringtable = \
  89             self._get_section_header_stringtable()
  90         self._section_name_map = None
  91
  92     def num_sections(self):
  93         """ Number of sections in the file
  94         """
  95         if self['e_shoff'] == 0:
  96             return 0
  97         # From the ELF ABI documentation at
  98         # https://refspecs.linuxfoundation.org/elf/gabi4+/ch4.sheader.html:
  99         # "e_shnum normally tells how many entries the section header table
 100         # contains. [...] If the number of sections is greater than or equal to
 101         # SHN_LORESERVE (0xff00), e_shnum has the value SHN_UNDEF (0) and the
 102         # actual number of section header table entries is contained in the
 103         # sh_size field of the section header at index 0 (otherwise, the sh_size
 104         # member of the initial entry contains 0)."
 105         if self['e_shnum'] == 0:
 106             return self._get_section_header(0)['sh_size']
 107         return self['e_shnum']
 108
 109     def get_section(self, n):
 110         """ Get the section at index #n from the file (Section object or a
 111             subclass)
 112         """
 113         section_header = self._get_section_header(n)
 114         return self._make_section(section_header)
 115
 116     def get_section_by_name(self, name):
 117         """ Get a section from the file, by name. Return None if no such
 118             section exists.
 119         """
 120         # The first time this method is called, construct a name to number
 121         # mapping
 122         #
 123         if self._section_name_map is None:
 124             self._make_section_name_map()
 125         secnum = self._section_name_map.get(name, None)
 126         return None if secnum is None else self.get_section(secnum)
 127
 128     def get_section_index(self, section_name):
 129         """ Gets the index of the section by name. Return None if no such
 130             section name exists.
 131         """
 132         # The first time this method is called, construct a name to number
 133         # mapping
 134         #
 135         if self._section_name_map is None:
 136             self._make_section_name_map()
 137         return self._section_name_map.get(section_name, None)
 138
 139     def iter_sections(self, type=None):
 140         """ Yield all the sections in the file. If the optional |type|
 141             parameter is passed, this method will only yield sections of the
 142             given type. The parameter value must be a string containing the
 143             name of the type as defined in the ELF specification, e.g.
 144             'SHT_SYMTAB'.
 145         """
 146         for i in range(self.num_sections()):
 147             section = self.get_section(i)
 148             if type is None or section['sh_type'] == type:
 149                 yield section
 150
 151     def num_segments(self):
 152         """ Number of segments in the file
 153         """
 154         # From: https://github.com/hjl-tools/x86-psABI/wiki/X86-psABI
 155         # Section: 4.1.2 Number of Program Headers
 156         # If the number of program headers is greater than or equal to
 157         # PN_XNUM (0xffff), this member has the value PN_XNUM
 158         # (0xffff). The actual number of program header table entries
 159         # is contained in the sh_info field of the section header at
 160         # index 0.
 161         if self['e_phnum'] < 0xffff:
 162             return self['e_phnum']
 163         else:
 164             return self.get_section(0)['sh_info']
 165
 166     def get_segment(self, n):
 167         """ Get the segment at index #n from the file (Segment object)
 168         """
 169         segment_header = self._get_segment_header(n)
 170         return self._make_segment(segment_header)
 171
 172     def iter_segments(self, type=None):
 173         """ Yield all the segments in the file. If the optional |type|
 174             parameter is passed, this method will only yield segments of the
 175             given type. The parameter value must be a string containing the
 176             name of the type as defined in the ELF specification, e.g.
 177             'PT_LOAD'.
 178         """
 179         for i in range(self.num_segments()):
 180             segment = self.get_segment(i)
 181             if type is None or segment['p_type'] == type:
 182                 yield segment
 183
 184     def address_offsets(self, start, size=1):
 185         """ Yield a file offset for each ELF segment containing a memory region.
 186
 187             A memory region is defined by the range [start...start+size). The
 188             offset of the region is yielded.
 189         """
 190         end = start + size
 191         # consider LOAD only to prevent same address being yielded twice
 192         for seg in self.iter_segments(type='PT_LOAD'):
 193             if (start >= seg['p_vaddr'] and
 194                 end <= seg['p_vaddr'] + seg['p_filesz']):
 195                 yield start - seg['p_vaddr'] + seg['p_offset']
 196
 197     def has_dwarf_info(self):
 198         """ Check whether this file appears to have debugging information.
 199             We assume that if it has the .debug_info or .zdebug_info section, it
 200             has all the other required sections as well.
 201         """
 202         return bool(self.get_section_by_name('.debug_info') or
 203             self.get_section_by_name('.zdebug_info') or
 204             self.get_section_by_name('.eh_frame'))
 205
 206     def get_dwarf_info(self, relocate_dwarf_sections=True):
 207         """ Return a DWARFInfo object representing the debugging information in
 208             this file.
 209
 210             If relocate_dwarf_sections is True, relocations for DWARF sections
 211             are looked up and applied.
 212         """
 213         # Expect that has_dwarf_info was called, so at least .debug_info is
 214         # present.
 215         # Sections that aren't found will be passed as None to DWARFInfo.
 216
 217         section_names = ('.debug_info', '.debug_aranges', '.debug_abbrev',
 218                          '.debug_str', '.debug_line', '.debug_frame',
 219                          '.debug_loc', '.debug_ranges', '.debug_pubtypes',
 220                          '.debug_pubnames', '.debug_addr',
 221                          '.debug_str_offsets', '.debug_line_str')
 222
 223
 224         compressed = bool(self.get_section_by_name('.zdebug_info'))
 225         if compressed:
 226             section_names = tuple(map(lambda x: '.z' + x[1:], section_names))
 227
 228         # As it is loaded in the process image, .eh_frame cannot be compressed
 229         section_names += ('.eh_frame', )
 230
 231         (debug_info_sec_name, debug_aranges_sec_name, debug_abbrev_sec_name,
 232          debug_str_sec_name, debug_line_sec_name, debug_frame_sec_name,
 233          debug_loc_sec_name, debug_ranges_sec_name, debug_pubtypes_name,
 234          debug_pubnames_name, debug_addr_name, debug_str_offsets_name,
 235          debug_line_str_name, eh_frame_sec_name) = section_names
 236
 237         debug_sections = {}
 238         for secname in section_names:
 239             section = self.get_section_by_name(secname)
 240             if section is None:
 241                 debug_sections[secname] = None
 242             else:
 243                 dwarf_section = self._read_dwarf_section(
 244                     section,
 245                     relocate_dwarf_sections)
 246                 if compressed and secname.startswith('.z'):
 247                     dwarf_section = self._decompress_dwarf_section(dwarf_section)
 248                 debug_sections[secname] = dwarf_section
 249
 250         return DWARFInfo(
 251                 config=DwarfConfig(
 252                     little_endian=self.little_endian,
 253                     default_address_size=self.elfclass // 8,
 254                     machine_arch=self.get_machine_arch()),
 255                 debug_info_sec=debug_sections[debug_info_sec_name],
 256                 debug_aranges_sec=debug_sections[debug_aranges_sec_name],
 257                 debug_abbrev_sec=debug_sections[debug_abbrev_sec_name],
 258                 debug_frame_sec=debug_sections[debug_frame_sec_name],
 259                 eh_frame_sec=debug_sections[eh_frame_sec_name],
 260                 debug_str_sec=debug_sections[debug_str_sec_name],
 261                 debug_loc_sec=debug_sections[debug_loc_sec_name],
 262                 debug_ranges_sec=debug_sections[debug_ranges_sec_name],
 263                 debug_line_sec=debug_sections[debug_line_sec_name],
 264                 debug_pubtypes_sec=debug_sections[debug_pubtypes_name],
 265                 debug_pubnames_sec=debug_sections[debug_pubnames_name],
 266                 debug_addr_sec=debug_sections[debug_addr_name],
 267                 debug_str_offsets_sec=debug_sections[debug_str_offsets_name],
 268                 debug_line_str_sec=debug_sections[debug_line_str_name]
 269                 )
 270
 271     def has_ehabi_info(self):
 272         """ Check whether this file appears to have arm exception handler index table.
 273         """
 274         return any(self.iter_sections(type='SHT_ARM_EXIDX'))
 275
 276     def get_ehabi_infos(self):
 277         """ Generally, shared library and executable contain 1 .ARM.exidx section.
 278             Object file contains many .ARM.exidx sections.
 279             So we must traverse every section and filter sections whose type is SHT_ARM_EXIDX.
 280         """
 281         _ret = []
 282         if self['e_type'] == 'ET_REL':
 283             # TODO: support relocatable file
 284             assert False, "Current version of pyelftools doesn't support relocatable file."
 285         for section in self.iter_sections(type='SHT_ARM_EXIDX'):
 286             _ret.append(EHABIInfo(section, self.little_endian))
 287         return _ret if len(_ret) > 0 else None
 288
 289     def get_machine_arch(self):
 290         """ Return the machine architecture, as detected from the ELF header.
 291         """
 292         architectures = {
 293             'EM_M32'           : 'AT&T WE 32100',
 294             'EM_SPARC'         : 'SPARC',
 295             'EM_386'           : 'x86',
 296             'EM_68K'           : 'Motorola 68000',
 297             'EM_88K'           : 'Motorola 88000',
 298             'EM_IAMCU'         : 'Intel MCU',
 299             'EM_860'           : 'Intel 80860',
 300             'EM_MIPS'          : 'MIPS',
 301             'EM_S370'          : 'IBM System/370',
 302             'EM_MIPS_RS3_LE'   : 'MIPS RS3000 Little-endian',
 303             'EM_PARISC'        : 'Hewlett-Packard PA-RISC',
 304             'EM_VPP500'        : 'Fujitsu VPP500',
 305             'EM_SPARC32PLUS'   : 'Enhanced SPARC',
 306             'EM_960'           : 'Intel 80960',
 307             'EM_PPC'           : 'PowerPC',
 308             'EM_PPC64'         : '64-bit PowerPC',
 309             'EM_S390'          : 'IBM System/390',
 310             'EM_SPU'           : 'IBM SPU/SPC',
 311             'EM_V800'          : 'NEC V800',
 312             'EM_FR20'          : 'Fujitsu FR20',
 313             'EM_RH32'          : 'TRW RH-32',
 314             'EM_RCE'           : 'Motorola RCE',
 315             'EM_ARM'           : 'ARM',
 316             'EM_ALPHA'         : 'Digital Alpha',
 317             'EM_SH'            : 'Hitachi SH',
 318             'EM_SPARCV9'       : 'SPARC Version 9',
 319             'EM_TRICORE'       : 'Siemens TriCore embedded processor',
 320             'EM_ARC'           : 'Argonaut RISC Core, Argonaut Technologies Inc.',
 321             'EM_H8_300'        : 'Hitachi H8/300',
 322             'EM_H8_300H'       : 'Hitachi H8/300H',
 323             'EM_H8S'           : 'Hitachi H8S',
 324             'EM_H8_500'        : 'Hitachi H8/500',
 325             'EM_IA_64'         : 'Intel IA-64',
 326             'EM_MIPS_X'        : 'MIPS-X',
 327             'EM_COLDFIRE'      : 'Motorola ColdFire',
 328             'EM_68HC12'        : 'Motorola M68HC12',
 329             'EM_MMA'           : 'Fujitsu MMA',
 330             'EM_PCP'           : 'Siemens PCP',
 331             'EM_NCPU'          : 'Sony nCPU',
 332             'EM_NDR1'          : 'Denso NDR1',
 333             'EM_STARCORE'      : 'Motorola Star*Core',
 334             'EM_ME16'          : 'Toyota ME16',
 335             'EM_ST100'         : 'STMicroelectronics ST100',
 336             'EM_TINYJ'         : 'Advanced Logic TinyJ',
 337             'EM_X86_64'        : 'x64',
 338             'EM_PDSP'          : 'Sony DSP',
 339             'EM_PDP10'         : 'Digital Equipment PDP-10',
 340             'EM_PDP11'         : 'Digital Equipment PDP-11',
 341             'EM_FX66'          : 'Siemens FX66',
 342             'EM_ST9PLUS'       : 'STMicroelectronics ST9+ 8/16 bit',
 343             'EM_ST7'           : 'STMicroelectronics ST7 8-bit',
 344             'EM_68HC16'        : 'Motorola MC68HC16',
 345             'EM_68HC11'        : 'Motorola MC68HC11',
 346             'EM_68HC08'        : 'Motorola MC68HC08',
 347             'EM_68HC05'        : 'Motorola MC68HC05',
 348             'EM_SVX'           : 'Silicon Graphics SVx',
 349             'EM_ST19'          : 'STMicroelectronics ST19 8-bit',
 350             'EM_VAX'           : 'Digital VAX',
 351             'EM_CRIS'          : 'Axis Communications 32-bit',
 352             'EM_JAVELIN'       : 'Infineon Technologies 32-bit',
 353             'EM_FIREPATH'      : 'Element 14 64-bit DSP',
 354             'EM_ZSP'           : 'LSI Logic 16-bit DSP',
 355             'EM_MMIX'          : 'Donald Knuth\'s educational 64-bit',
 356             'EM_HUANY'         : 'Harvard University machine-independent object files',
 357             'EM_PRISM'         : 'SiTera Prism',
 358             'EM_AVR'           : 'Atmel AVR 8-bit',
 359             'EM_FR30'          : 'Fujitsu FR30',
 360             'EM_D10V'          : 'Mitsubishi D10V',
 361             'EM_D30V'          : 'Mitsubishi D30V',
 362             'EM_V850'          : 'NEC v850',
 363             'EM_M32R'          : 'Mitsubishi M32R',
 364             'EM_MN10300'       : 'Matsushita MN10300',
 365             'EM_MN10200'       : 'Matsushita MN10200',
 366             'EM_PJ'            : 'picoJava',
 367             'EM_OPENRISC'      : 'OpenRISC 32-bit',
 368             'EM_ARC_COMPACT'   : 'ARC International ARCompact',
 369             'EM_XTENSA'        : 'Tensilica Xtensa',
 370             'EM_VIDEOCORE'     : 'Alphamosaic VideoCore',
 371             'EM_TMM_GPP'       : 'Thompson Multimedia',
 372             'EM_NS32K'         : 'National Semiconductor 32000 series',
 373             'EM_TPC'           : 'Tenor Network TPC',
 374             'EM_SNP1K'         : 'Trebia SNP 1000',
 375             'EM_ST200'         : 'STMicroelectronics ST200',
 376             'EM_IP2K'          : 'Ubicom IP2xxx',
 377             'EM_MAX'           : 'MAX',
 378             'EM_CR'            : 'National Semiconductor CompactRISC',
 379             'EM_F2MC16'        : 'Fujitsu F2MC16',
 380             'EM_MSP430'        : 'Texas Instruments msp430',
 381             'EM_BLACKFIN'      : 'Analog Devices Blackfin',
 382             'EM_SE_C33'        : 'Seiko Epson S1C33',
 383             'EM_SEP'           : 'Sharp',
 384             'EM_ARCA'          : 'Arca RISC',
 385             'EM_UNICORE'       : 'PKU-Unity MPRC',
 386             'EM_EXCESS'        : 'eXcess',
 387             'EM_DXP'           : 'Icera Semiconductor Deep Execution Processor',
 388             'EM_ALTERA_NIOS2'  : 'Altera Nios II',
 389             'EM_CRX'           : 'National Semiconductor CompactRISC CRX',
 390             'EM_XGATE'         : 'Motorola XGATE',
 391             'EM_C166'          : 'Infineon C16x/XC16x',
 392             'EM_M16C'          : 'Renesas M16C',
 393             'EM_DSPIC30F'      : 'Microchip Technology dsPIC30F',
 394             'EM_CE'            : 'Freescale Communication Engine RISC core',
 395             'EM_M32C'          : 'Renesas M32C',
 396             'EM_TSK3000'       : 'Altium TSK3000',
 397             'EM_RS08'          : 'Freescale RS08',
 398             'EM_SHARC'         : 'Analog Devices SHARC',
 399             'EM_ECOG2'         : 'Cyan Technology eCOG2',
 400             'EM_SCORE7'        : 'Sunplus S+core7 RISC',
 401             'EM_DSP24'         : 'New Japan Radio (NJR) 24-bit DSP',
 402             'EM_VIDEOCORE3'    : 'Broadcom VideoCore III',
 403             'EM_LATTICEMICO32' : 'Lattice FPGA RISC',
 404             'EM_SE_C17'        : 'Seiko Epson C17',
 405             'EM_TI_C6000'      : 'TI TMS320C6000',
 406             'EM_TI_C2000'      : 'TI TMS320C2000',
 407             'EM_TI_C5500'      : 'TI TMS320C55x',
 408             'EM_TI_ARP32'      : 'TI Application Specific RISC, 32bit',
 409             'EM_TI_PRU'        : 'TI Programmable Realtime Unit',
 410             'EM_MMDSP_PLUS'    : 'STMicroelectronics 64bit VLIW',
 411             'EM_CYPRESS_M8C'   : 'Cypress M8C',
 412             'EM_R32C'          : 'Renesas R32C',
 413             'EM_TRIMEDIA'      : 'NXP Semiconductors TriMedia',
 414             'EM_QDSP6'         : 'QUALCOMM DSP6',
 415             'EM_8051'          : 'Intel 8051',
 416             'EM_STXP7X'        : 'STMicroelectronics STxP7x',
 417             'EM_NDS32'         : 'Andes Technology RISC',
 418             'EM_ECOG1'         : 'Cyan Technology eCOG1X',
 419             'EM_ECOG1X'        : 'Cyan Technology eCOG1X',
 420             'EM_MAXQ30'        : 'Dallas Semiconductor MAXQ30',
 421             'EM_XIMO16'        : 'New Japan Radio (NJR) 16-bit',
 422             'EM_MANIK'         : 'M2000 Reconfigurable RISC',
 423             'EM_CRAYNV2'       : 'Cray Inc. NV2',
 424             'EM_RX'            : 'Renesas RX',
 425             'EM_METAG'         : 'Imagination Technologies META',
 426             'EM_MCST_ELBRUS'   : 'MCST Elbrus',
 427             'EM_ECOG16'        : 'Cyan Technology eCOG16',
 428             'EM_CR16'          : 'National Semiconductor CompactRISC CR16 16-bit',
 429             'EM_ETPU'          : 'Freescale',
 430             'EM_SLE9X'         : 'Infineon Technologies SLE9X',
 431             'EM_L10M'          : 'Intel L10M',
 432             'EM_K10M'          : 'Intel K10M',
 433             'EM_AARCH64'       : 'AArch64',
 434             'EM_AVR32'         : 'Atmel 32-bit',
 435             'EM_STM8'          : 'STMicroeletronics STM8 8-bit',
 436             'EM_TILE64'        : 'Tilera TILE64',
 437             'EM_TILEPRO'       : 'Tilera TILEPro',
 438             'EM_MICROBLAZE'    : 'Xilinx MicroBlaze 32-bit RISC',
 439             'EM_CUDA'          : 'NVIDIA CUDA',
 440             'EM_TILEGX'        : 'Tilera TILE-Gx',
 441             'EM_CLOUDSHIELD'   : 'CloudShield',
 442             'EM_COREA_1ST'     : 'KIPO-KAIST Core-A 1st generation',
 443             'EM_COREA_2ND'     : 'KIPO-KAIST Core-A 2nd generation',
 444             'EM_ARC_COMPACT2'  : 'Synopsys ARCompact V2',
 445             'EM_OPEN8'         : 'Open8 8-bit RISC',
 446             'EM_RL78'          : 'Renesas RL78',
 447             'EM_VIDEOCORE5'    : 'Broadcom VideoCore V',
 448             'EM_78KOR'         : 'Renesas 78KOR',
 449             'EM_56800EX'       : 'Freescale 56800EX',
 450             'EM_BA1'           : 'Beyond BA1',
 451             'EM_BA2'           : 'Beyond BA2',
 452             'EM_XCORE'         : 'XMOS xCORE',
 453             'EM_MCHP_PIC'      : 'Microchip 8-bit PIC',
 454             'EM_INTEL205'      : 'Reserved by Intel',
 455             'EM_INTEL206'      : 'Reserved by Intel',
 456             'EM_INTEL207'      : 'Reserved by Intel',
 457             'EM_INTEL208'      : 'Reserved by Intel',
 458             'EM_INTEL209'      : 'Reserved by Intel',
 459             'EM_KM32'          : 'KM211 KM32 32-bit',
 460             'EM_KMX32'         : 'KM211 KMX32 32-bit',
 461             'EM_KMX16'         : 'KM211 KMX16 16-bit',
 462             'EM_KMX8'          : 'KM211 KMX8 8-bit',
 463             'EM_KVARC'         : 'KM211 KVARC',
 464             'EM_CDP'           : 'Paneve CDP',
 465             'EM_COGE'          : 'Cognitive',
 466             'EM_COOL'          : 'Bluechip Systems CoolEngine',
 467             'EM_NORC'          : 'Nanoradio Optimized RISC',
 468             'EM_CSR_KALIMBA'   : 'CSR Kalimba',
 469             'EM_Z80'           : 'Zilog Z80',
 470             'EM_VISIUM'        : 'VISIUMcore',
 471             'EM_FT32'          : 'FTDI Chip FT32 32-bit RISC',
 472             'EM_MOXIE'         : 'Moxie',
 473             'EM_AMDGPU'        : 'AMD GPU',
 474             'EM_RISCV'         : 'RISC-V',
 475             'EM_BPF'           : 'Linux BPF - in-kernel virtual machine',
 476             'EM_CSKY'          : 'C-SKY',
 477             'EM_FRV'           : 'Fujitsu FR-V'
 478         }
 479
 480         return architectures.get(self['e_machine'], '<unknown>')
 481
 482     def get_shstrndx(self):
 483         """ Find the string table section index for the section header table
 484         """
 485         # From https://refspecs.linuxfoundation.org/elf/gabi4+/ch4.eheader.html:
 486         # If the section name string table section index is greater than or
 487         # equal to SHN_LORESERVE (0xff00), this member has the value SHN_XINDEX
 488         # (0xffff) and the actual index of the section name string table section
 489         # is contained in the sh_link field of the section header at index 0.
 490         if self['e_shstrndx'] != SHN_INDICES.SHN_XINDEX:
 491             return self['e_shstrndx']
 492         else:
 493             return self._get_section_header(0)['sh_link']
 494
 495     #-------------------------------- PRIVATE --------------------------------#
 496
 497     def __getitem__(self, name):
 498         """ Implement dict-like access to header entries
 499         """
 500         return self.header[name]
 501
 502     def _identify_file(self):
 503         """ Verify the ELF file and identify its class and endianness.
 504         """
 505         # Note: this code reads the stream directly, without using ELFStructs,
 506         # since we don't yet know its exact format. ELF was designed to be
 507         # read like this - its e_ident field is word-size and endian agnostic.
 508         self.stream.seek(0)
 509         magic = self.stream.read(4)
 510         elf_assert(magic == b'\x7fELF', 'Magic number does not match')
 511
 512         ei_class = self.stream.read(1)
 513         if ei_class == b'\x01':
 514             self.elfclass = 32
 515         elif ei_class == b'\x02':
 516             self.elfclass = 64
 517         else:
 518             raise ELFError('Invalid EI_CLASS %s' % repr(ei_class))
 519
 520         ei_data = self.stream.read(1)
 521         if ei_data == b'\x01':
 522             self.little_endian = True
 523         elif ei_data == b'\x02':
 524             self.little_endian = False
 525         else:
 526             raise ELFError('Invalid EI_DATA %s' % repr(ei_data))
 527
 528     def _section_offset(self, n):
 529         """ Compute the offset of section #n in the file
 530         """
 531         return self['e_shoff'] + n * self['e_shentsize']
 532
 533     def _segment_offset(self, n):
 534         """ Compute the offset of segment #n in the file
 535         """
 536         return self['e_phoff'] + n * self['e_phentsize']
 537
 538     def _make_segment(self, segment_header):
 539         """ Create a Segment object of the appropriate type
 540         """
 541         segtype = segment_header['p_type']
 542         if segtype == 'PT_INTERP':
 543             return InterpSegment(segment_header, self.stream)
 544         elif segtype == 'PT_DYNAMIC':
 545             return DynamicSegment(segment_header, self.stream, self)
 546         elif segtype == 'PT_NOTE':
 547             return NoteSegment(segment_header, self.stream, self)
 548         else:
 549             return Segment(segment_header, self.stream)
 550
 551     def _get_section_header(self, n):
 552         """ Find the header of section #n, parse it and return the struct
 553         """
 554         return struct_parse(
 555             self.structs.Elf_Shdr,
 556             self.stream,
 557             stream_pos=self._section_offset(n))
 558
 559     def _get_section_name(self, section_header):
 560         """ Given a section header, find this section's name in the file's
 561             string table
 562         """
 563         name_offset = section_header['sh_name']
 564         return self._section_header_stringtable.get_string(name_offset)
 565
 566     def _make_section(self, section_header):
 567         """ Create a section object of the appropriate type
 568         """
 569         name = self._get_section_name(section_header)
 570         sectype = section_header['sh_type']
 571
 572         if sectype == 'SHT_STRTAB':
 573             return StringTableSection(section_header, name, self)
 574         elif sectype == 'SHT_NULL':
 575             return NullSection(section_header, name, self)
 576         elif sectype in ('SHT_SYMTAB', 'SHT_DYNSYM', 'SHT_SUNW_LDYNSYM'):
 577             return self._make_symbol_table_section(section_header, name)
 578         elif sectype == 'SHT_SYMTAB_SHNDX':
 579             return self._make_symbol_table_index_section(section_header, name)
 580         elif sectype == 'SHT_SUNW_syminfo':
 581             return self._make_sunwsyminfo_table_section(section_header, name)
 582         elif sectype == 'SHT_GNU_verneed':
 583             return self._make_gnu_verneed_section(section_header, name)
 584         elif sectype == 'SHT_GNU_verdef':
 585             return self._make_gnu_verdef_section(section_header, name)
 586         elif sectype == 'SHT_GNU_versym':
 587             return self._make_gnu_versym_section(section_header, name)
 588         elif sectype in ('SHT_REL', 'SHT_RELA'):
 589             return RelocationSection(section_header, name, self)
 590         elif sectype == 'SHT_DYNAMIC':
 591             return DynamicSection(section_header, name, self)
 592         elif sectype == 'SHT_NOTE':
 593             return NoteSection(section_header, name, self)
 594         elif sectype == 'SHT_PROGBITS' and name == '.stab':
 595             return StabSection(section_header, name, self)
 596         elif sectype == 'SHT_ARM_ATTRIBUTES':
 597             return ARMAttributesSection(section_header, name, self)
 598         elif sectype == 'SHT_HASH':
 599             return self._make_elf_hash_section(section_header, name)
 600         elif sectype == 'SHT_GNU_HASH':
 601             return self._make_gnu_hash_section(section_header, name)
 602         elif sectype == 'SHT_RELR':
 603             return RelrRelocationSection(section_header, name, self)
 604         else:
 605             return Section(section_header, name, self)
 606
 607     def _make_section_name_map(self):
 608         self._section_name_map = {}
 609         for i, sec in enumerate(self.iter_sections()):
 610             self._section_name_map[sec.name] = i
 611
 612     def _make_symbol_table_section(self, section_header, name):
 613         """ Create a SymbolTableSection
 614         """
 615         linked_strtab_index = section_header['sh_link']
 616         strtab_section = self.get_section(linked_strtab_index)
 617         return SymbolTableSection(
 618             section_header, name,
 619             elffile=self,
 620             stringtable=strtab_section)
 621
 622     def _make_symbol_table_index_section(self, section_header, name):
 623         """ Create a SymbolTableIndexSection object
 624         """
 625         linked_symtab_index = section_header['sh_link']
 626         return SymbolTableIndexSection(
 627             section_header, name, elffile=self,
 628             symboltable=linked_symtab_index)
 629
 630     def _make_sunwsyminfo_table_section(self, section_header, name):
 631         """ Create a SUNWSyminfoTableSection
 632         """
 633         linked_strtab_index = section_header['sh_link']
 634         strtab_section = self.get_section(linked_strtab_index)
 635         return SUNWSyminfoTableSection(
 636             section_header, name,
 637             elffile=self,
 638             symboltable=strtab_section)
 639
 640     def _make_gnu_verneed_section(self, section_header, name):
 641         """ Create a GNUVerNeedSection
 642         """
 643         linked_strtab_index = section_header['sh_link']
 644         strtab_section = self.get_section(linked_strtab_index)
 645         return GNUVerNeedSection(
 646             section_header, name,
 647             elffile=self,
 648             stringtable=strtab_section)
 649
 650     def _make_gnu_verdef_section(self, section_header, name):
 651         """ Create a GNUVerDefSection
 652         """
 653         linked_strtab_index = section_header['sh_link']
 654         strtab_section = self.get_section(linked_strtab_index)
 655         return GNUVerDefSection(
 656             section_header, name,
 657             elffile=self,
 658             stringtable=strtab_section)
 659
 660     def _make_gnu_versym_section(self, section_header, name):
 661         """ Create a GNUVerSymSection
 662         """
 663         linked_strtab_index = section_header['sh_link']
 664         strtab_section = self.get_section(linked_strtab_index)
 665         return GNUVerSymSection(
 666             section_header, name,
 667             elffile=self,
 668             symboltable=strtab_section)
 669
 670     def _make_elf_hash_section(self, section_header, name):
 671         linked_symtab_index = section_header['sh_link']
 672         symtab_section = self.get_section(linked_symtab_index)
 673         return ELFHashSection(
 674             section_header, name, self, symtab_section
 675         )
 676
 677     def _make_gnu_hash_section(self, section_header, name):
 678         linked_symtab_index = section_header['sh_link']
 679         symtab_section = self.get_section(linked_symtab_index)
 680         return GNUHashSection(
 681             section_header, name, self, symtab_section
 682         )
 683
 684     def _get_segment_header(self, n):
 685         """ Find the header of segment #n, parse it and return the struct
 686         """
 687         return struct_parse(
 688             self.structs.Elf_Phdr,
 689             self.stream,
 690             stream_pos=self._segment_offset(n))
 691
 692     def _get_section_header_stringtable(self):
 693         """ Get the string table section corresponding to the section header
 694             table.
 695         """
 696         stringtable_section_num = self.get_shstrndx()
 697         return StringTableSection(
 698                 header=self._get_section_header(stringtable_section_num),
 699                 name='',
 700                 elffile=self)
 701
 702     def _parse_elf_header(self):
 703         """ Parses the ELF file header and assigns the result to attributes
 704             of this object.
 705         """
 706         return struct_parse(self.structs.Elf_Ehdr, self.stream, stream_pos=0)
 707
 708     def _read_dwarf_section(self, section, relocate_dwarf_sections):
 709         """ Read the contents of a DWARF section from the stream and return a
 710             DebugSectionDescriptor. Apply relocations if asked to.
 711         """
 712         # The section data is read into a new stream, for processing
 713         section_stream = BytesIO()
 714         section_stream.write(section.data())
 715
 716         if relocate_dwarf_sections:
 717             reloc_handler = RelocationHandler(self)
 718             reloc_section = reloc_handler.find_relocations_for_section(section)
 719             if reloc_section is not None:
 720                 reloc_handler.apply_section_relocations(
 721                         section_stream, reloc_section)
 722
 723         return DebugSectionDescriptor(
 724                 stream=section_stream,
 725                 name=section.name,
 726                 global_offset=section['sh_offset'],
 727                 size=section.data_size,
 728                 address=section['sh_addr'])
 729
 730     @staticmethod
 731     def _decompress_dwarf_section(section):
 732         """ Returns the uncompressed contents of the provided DWARF section.
 733         """
 734         # TODO: support other compression formats from readelf.c
 735         assert section.size > 12, 'Unsupported compression format.'
 736
 737         section.stream.seek(0)
 738         # According to readelf.c the content should contain "ZLIB"
 739         # followed by the uncompressed section size - 8 bytes in
 740         # big-endian order
 741         compression_type = section.stream.read(4)
 742         assert compression_type == b'ZLIB', \
 743             'Invalid compression type: %r' % (compression_type)
 744
 745         uncompressed_size = struct.unpack('>Q', section.stream.read(8))[0]
 746
 747         decompressor = zlib.decompressobj()
 748         uncompressed_stream = BytesIO()
 749         while True:
 750             chunk = section.stream.read(PAGESIZE)
 751             if not chunk:
 752                 break
 753             uncompressed_stream.write(decompressor.decompress(chunk))
 754         uncompressed_stream.write(decompressor.flush())
 755
 756         uncompressed_stream.seek(0, io.SEEK_END)
 757         size = uncompressed_stream.tell()
 758         assert uncompressed_size == size, \
 759                 'Wrong uncompressed size: expected %r, but got %r' % (
 760                     uncompressed_size, size,
 761                 )
 762
 763         return section._replace(stream=uncompressed_stream, size=size)