elftools/dwarf/structs.py

   1 #-------------------------------------------------------------------------------
   2 # elftools: dwarf/structs.py
   3 #
   4 # Encapsulation of Construct structs for parsing DWARF, adjusted for correct
   5 # endianness and word-size.
   6 #
   7 # Eli Bendersky (eliben@gmail.com)
   8 # This code is in the public domain
   9 #-------------------------------------------------------------------------------
  10 from ..construct import (
  11     UBInt8, UBInt16, UBInt32, UBInt64, ULInt8, ULInt16, ULInt32, ULInt64,
  12     SBInt8, SBInt16, SBInt32, SBInt64, SLInt8, SLInt16, SLInt32, SLInt64,
  13     Adapter, Struct, ConstructError, If, RepeatUntil, Field, Rename, Enum,
  14     Array, PrefixedArray, CString, Embed,
  15     )
  16 from ..common.construct_utils import RepeatUntilExcluding
  17
  18 from .enums import *
  19
  20
  21 class DWARFStructs(object):
  22     """ Exposes Construct structs suitable for parsing information from DWARF
  23         sections. Each compile unit in DWARF info can have its own structs
  24         object. Keep in mind that these structs have to be given a name (by
  25         calling them with a name) before being used for parsing (like other
  26         Construct structs). Those that should be used without a name are marked
  27         by (+).
  28
  29         Accessible attributes (mostly as described in chapter 7 of the DWARF
  30         spec v3):
  31
  32             Dwarf_[u]int{8,16,32,64):
  33                 Data chunks of the common sizes
  34
  35             Dwarf_offset:
  36                 32-bit or 64-bit word, depending on dwarf_format
  37
  38             Dwarf_target_addr:
  39                 32-bit or 64-bit word, depending on address size
  40
  41             Dwarf_initial_length:
  42                 "Initial length field" encoding
  43                 section 7.4
  44
  45             Dwarf_{u,s}leb128:
  46                 ULEB128 and SLEB128 variable-length encoding
  47
  48             Dwarf_CU_header (+):
  49                 Compilation unit header
  50
  51             Dwarf_abbrev_declaration (+):
  52                 Abbreviation table declaration - doesn't include the initial
  53                 code, only the contents.
  54
  55             Dwarf_dw_form (+):
  56                 A dictionary mapping 'DW_FORM_*' keys into construct Structs
  57                 that parse such forms. These Structs have already been given
  58                 dummy names.
  59
  60             Dwarf_lineprog_header (+):
  61                 Line program header
  62
  63             Dwarf_lineprog_file_entry (+):
  64                 A single file entry in a line program header or instruction
  65
  66             Dwarf_CIE_header (+):
  67                 A call-frame CIE
  68
  69             Dwarf_FDE_header (+):
  70                 A call-frame FDE
  71
  72         See also the documentation of public methods.
  73     """
  74     def __init__(self, little_endian, dwarf_format, address_size):
  75         """ little_endian:
  76                 True if the file is little endian, False if big
  77
  78             dwarf_format:
  79                 DWARF Format: 32 or 64-bit (see spec section 7.4)
  80
  81             address_size:
  82                 Target machine address size, in bytes (4 or 8). (See spec
  83                 section 7.5.1)
  84         """
  85         assert dwarf_format == 32 or dwarf_format == 64
  86         assert address_size == 8 or address_size == 4
  87         self.little_endian = little_endian
  88         self.dwarf_format = dwarf_format
  89         self.address_size = address_size
  90         self._create_structs()
  91
  92     def initial_length_field_size(self):
  93         """ Size of an initial length field.
  94         """
  95         return 4 if self.dwarf_format == 32 else 12
  96
  97     def _create_structs(self):
  98         if self.little_endian:
  99             self.Dwarf_uint8 = ULInt8
 100             self.Dwarf_uint16 = ULInt16
 101             self.Dwarf_uint32 = ULInt32
 102             self.Dwarf_uint64 = ULInt64
 103             self.Dwarf_offset = ULInt32 if self.dwarf_format == 32 else ULInt64
 104             self.Dwarf_target_addr = (
 105                 ULInt32 if self.address_size == 4 else ULInt64)
 106             self.Dwarf_int8 = SLInt8
 107             self.Dwarf_int16 = SLInt16
 108             self.Dwarf_int32 = SLInt32
 109             self.Dwarf_int64 = SLInt64
 110         else:
 111             self.Dwarf_uint8 = UBInt8
 112             self.Dwarf_uint16 = UBInt16
 113             self.Dwarf_uint32 = UBInt32
 114             self.Dwarf_uint64 = UBInt64
 115             self.Dwarf_offset = UBInt32 if self.dwarf_format == 32 else UBInt64
 116             self.Dwarf_target_addr = (
 117                 UBInt32 if self.address_size == 4 else UBInt64)
 118             self.Dwarf_int8 = SBInt8
 119             self.Dwarf_int16 = SBInt16
 120             self.Dwarf_int32 = SBInt32
 121             self.Dwarf_int64 = SBInt64
 122
 123         self._create_initial_length()
 124         self._create_leb128()
 125         self._create_cu_header()
 126         self._create_abbrev_declaration()
 127         self._create_dw_form()
 128         self._create_lineprog_header()
 129         self._create_callframe_entry_headers()
 130
 131     def _create_initial_length(self):
 132         def _InitialLength(name):
 133             # Adapts a Struct that parses forward a full initial length field.
 134             # Only if the first word is the continuation value, the second
 135             # word is parsed from the stream.
 136             #
 137             return _InitialLengthAdapter(
 138                 Struct(name,
 139                     self.Dwarf_uint32('first'),
 140                     If(lambda ctx: ctx.first == 0xFFFFFFFF,
 141                         self.Dwarf_uint64('second'),
 142                         elsevalue=None)))
 143         self.Dwarf_initial_length = _InitialLength
 144
 145     def _create_leb128(self):
 146         self.Dwarf_uleb128 = _ULEB128
 147         self.Dwarf_sleb128 = _SLEB128
 148
 149     def _create_cu_header(self):
 150         self.Dwarf_CU_header = Struct('Dwarf_CU_header',
 151             self.Dwarf_initial_length('unit_length'),
 152             self.Dwarf_uint16('version'),
 153             self.Dwarf_offset('debug_abbrev_offset'),
 154             self.Dwarf_uint8('address_size'))
 155
 156     def _create_abbrev_declaration(self):
 157         self.Dwarf_abbrev_declaration = Struct('Dwarf_abbrev_entry',
 158             Enum(self.Dwarf_uleb128('tag'), **ENUM_DW_TAG),
 159             Enum(self.Dwarf_uint8('children_flag'), **ENUM_DW_CHILDREN),
 160             RepeatUntilExcluding(
 161                 lambda obj, ctx:
 162                     obj.name == 'DW_AT_null' and obj.form == 'DW_FORM_null',
 163                 Struct('attr_spec',
 164                     Enum(self.Dwarf_uleb128('name'), **ENUM_DW_AT),
 165                     Enum(self.Dwarf_uleb128('form'), **ENUM_DW_FORM))))
 166
 167     def _create_dw_form(self):
 168         self.Dwarf_dw_form = dict(
 169             DW_FORM_addr=self.Dwarf_target_addr(''),
 170
 171             DW_FORM_block1=self._make_block_struct(self.Dwarf_uint8),
 172             DW_FORM_block2=self._make_block_struct(self.Dwarf_uint16),
 173             DW_FORM_block4=self._make_block_struct(self.Dwarf_uint32),
 174             DW_FORM_block=self._make_block_struct(self.Dwarf_uleb128),
 175
 176             # All DW_FORM_data<n> forms are assumed to be unsigned
 177             DW_FORM_data1=self.Dwarf_uint8(''),
 178             DW_FORM_data2=self.Dwarf_uint16(''),
 179             DW_FORM_data4=self.Dwarf_uint32(''),
 180             DW_FORM_data8=self.Dwarf_uint64(''),
 181             DW_FORM_sdata=self.Dwarf_sleb128(''),
 182             DW_FORM_udata=self.Dwarf_uleb128(''),
 183
 184             DW_FORM_string=CString(''),
 185             DW_FORM_strp=self.Dwarf_offset(''),
 186             DW_FORM_flag=self.Dwarf_uint8(''),
 187
 188             DW_FORM_ref1=self.Dwarf_uint8(''),
 189             DW_FORM_ref2=self.Dwarf_uint16(''),
 190             DW_FORM_ref4=self.Dwarf_uint32(''),
 191             DW_FORM_ref8=self.Dwarf_uint64(''),
 192             DW_FORM_ref_udata=self.Dwarf_uleb128(''),
 193             DW_FORM_ref_addr=self.Dwarf_offset(''),
 194
 195             DW_FORM_indirect=self.Dwarf_uleb128(''),
 196         )
 197
 198     def _create_lineprog_header(self):
 199         # A file entry is terminated by a NULL byte, so we don't want to parse
 200         # past it. Therefore an If is used.
 201         self.Dwarf_lineprog_file_entry = Struct('file_entry',
 202             CString('name'),
 203             If(lambda ctx: len(ctx.name) != 0,
 204                 Embed(Struct('',
 205                     self.Dwarf_uleb128('dir_index'),
 206                     self.Dwarf_uleb128('mtime'),
 207                     self.Dwarf_uleb128('length')))))
 208
 209         self.Dwarf_lineprog_header = Struct('Dwarf_lineprog_header',
 210             self.Dwarf_initial_length('unit_length'),
 211             self.Dwarf_uint16('version'),
 212             self.Dwarf_offset('header_length'),
 213             self.Dwarf_uint8('minimum_instruction_length'),
 214             self.Dwarf_uint8('default_is_stmt'),
 215             self.Dwarf_int8('line_base'),
 216             self.Dwarf_uint8('line_range'),
 217             self.Dwarf_uint8('opcode_base'),
 218             Array(lambda ctx: ctx['opcode_base'] - 1,
 219                   self.Dwarf_uint8('standard_opcode_lengths')),
 220             RepeatUntilExcluding(
 221                 lambda obj, ctx: obj == b'',
 222                 CString('include_directory')),
 223             RepeatUntilExcluding(
 224                 lambda obj, ctx: len(obj.name) == 0,
 225                 self.Dwarf_lineprog_file_entry),
 226             )
 227
 228     def _create_callframe_entry_headers(self):
 229         self.Dwarf_CIE_header = Struct('Dwarf_CIE_header',
 230             self.Dwarf_initial_length('length'),
 231             self.Dwarf_offset('CIE_id'),
 232             self.Dwarf_uint8('version'),
 233             CString('augmentation'),
 234             self.Dwarf_uleb128('code_alignment_factor'),
 235             self.Dwarf_sleb128('data_alignment_factor'),
 236             self.Dwarf_uleb128('return_address_register'))
 237
 238         self.Dwarf_FDE_header = Struct('Dwarf_FDE_header',
 239             self.Dwarf_initial_length('length'),
 240             self.Dwarf_offset('CIE_pointer'),
 241             self.Dwarf_target_addr('initial_location'),
 242             self.Dwarf_target_addr('address_range'))
 243
 244     def _make_block_struct(self, length_field):
 245         """ Create a struct for DW_FORM_block<size>
 246         """
 247         return PrefixedArray(
 248                     subcon=self.Dwarf_uint8('elem'),
 249                     length_field=length_field(''))
 250
 251
 252 class _InitialLengthAdapter(Adapter):
 253     """ A standard Construct adapter that expects a sub-construct
 254         as a struct with one or two values (first, second).
 255     """
 256     def _decode(self, obj, context):
 257         if obj.first < 0xFFFFFF00:
 258             return obj.first
 259         else:
 260             if obj.first == 0xFFFFFFFF:
 261                 return obj.second
 262             else:
 263                 raise ConstructError("Failed decoding initial length for %X" % (
 264                     obj.first))
 265
 266
 267 def _LEB128_reader():
 268     """ Read LEB128 variable-length data from the stream. The data is terminated
 269         by a byte with 0 in its highest bit.
 270     """
 271     return RepeatUntil(
 272                 lambda obj, ctx: ord(obj) < 0x80,
 273                 Field(None, 1))
 274
 275
 276 class _ULEB128Adapter(Adapter):
 277     """ An adapter for ULEB128, given a sequence of bytes in a sub-construct.
 278     """
 279     def _decode(self, obj, context):
 280         value = 0
 281         for b in reversed(obj):
 282             value = (value << 7) + (ord(b) & 0x7F)
 283         return value
 284
 285
 286 class _SLEB128Adapter(Adapter):
 287     """ An adapter for SLEB128, given a sequence of bytes in a sub-construct.
 288     """
 289     def _decode(self, obj, context):
 290         value = 0
 291         for b in reversed(obj):
 292             value = (value << 7) + (ord(b) & 0x7F)
 293         if ord(obj[-1]) & 0x40:
 294             # negative -> sign extend
 295             #
 296             value |= - (1 << (7 * len(obj)))
 297         return value
 298
 299
 300 def _ULEB128(name):
 301     """ A construct creator for ULEB128 encoding.
 302     """
 303     return Rename(name, _ULEB128Adapter(_LEB128_reader()))
 304
 305
 306 def _SLEB128(name):
 307     """ A construct creator for SLEB128 encoding.
 308     """
 309     return Rename(name, _SLEB128Adapter(_LEB128_reader()))
 310
 311