elftools/dwarf/dwarf_expr.py

   1 #-------------------------------------------------------------------------------
   2 # elftools: dwarf/dwarf_expr.py
   3 #
   4 # Decoding DWARF expressions
   5 #
   6 # Eli Bendersky (eliben@gmail.com)
   7 # This code is in the public domain
   8 #-------------------------------------------------------------------------------
   9 from collections import namedtuple
  10 from io import BytesIO
  11
  12 from ..common.utils import struct_parse, bytelist2string, read_blob
  13
  14
  15 # DWARF expression opcodes. name -> opcode mapping
  16 DW_OP_name2opcode = dict(
  17     DW_OP_addr=0x03,
  18     DW_OP_deref=0x06,
  19     DW_OP_const1u=0x08,
  20     DW_OP_const1s=0x09,
  21     DW_OP_const2u=0x0a,
  22     DW_OP_const2s=0x0b,
  23     DW_OP_const4u=0x0c,
  24     DW_OP_const4s=0x0d,
  25     DW_OP_const8u=0x0e,
  26     DW_OP_const8s=0x0f,
  27     DW_OP_constu=0x10,
  28     DW_OP_consts=0x11,
  29     DW_OP_dup=0x12,
  30     DW_OP_drop=0x13,
  31     DW_OP_over=0x14,
  32     DW_OP_pick=0x15,
  33     DW_OP_swap=0x16,
  34     DW_OP_rot=0x17,
  35     DW_OP_xderef=0x18,
  36     DW_OP_abs=0x19,
  37     DW_OP_and=0x1a,
  38     DW_OP_div=0x1b,
  39     DW_OP_minus=0x1c,
  40     DW_OP_mod=0x1d,
  41     DW_OP_mul=0x1e,
  42     DW_OP_neg=0x1f,
  43     DW_OP_not=0x20,
  44     DW_OP_or=0x21,
  45     DW_OP_plus=0x22,
  46     DW_OP_plus_uconst=0x23,
  47     DW_OP_shl=0x24,
  48     DW_OP_shr=0x25,
  49     DW_OP_shra=0x26,
  50     DW_OP_xor=0x27,
  51     DW_OP_bra=0x28,
  52     DW_OP_eq=0x29,
  53     DW_OP_ge=0x2a,
  54     DW_OP_gt=0x2b,
  55     DW_OP_le=0x2c,
  56     DW_OP_lt=0x2d,
  57     DW_OP_ne=0x2e,
  58     DW_OP_skip=0x2f,
  59     DW_OP_regx=0x90,
  60     DW_OP_fbreg=0x91,
  61     DW_OP_bregx=0x92,
  62     DW_OP_piece=0x93,
  63     DW_OP_deref_size=0x94,
  64     DW_OP_xderef_size=0x95,
  65     DW_OP_nop=0x96,
  66     DW_OP_push_object_address=0x97,
  67     DW_OP_call2=0x98,
  68     DW_OP_call4=0x99,
  69     DW_OP_call_ref=0x9a,
  70     DW_OP_form_tls_address=0x9b,
  71     DW_OP_call_frame_cfa=0x9c,
  72     DW_OP_bit_piece=0x9d,
  73     DW_OP_implicit_value=0x9e,
  74     DW_OP_stack_value=0x9f,
  75     DW_OP_implicit_pointer=0xa0,
  76     DW_OP_addrx=0xa1,
  77     DW_OP_constx=0xa2,
  78     DW_OP_entry_value=0xa3,
  79     DW_OP_const_type=0xa4,
  80     DW_OP_regval_type=0xa5,
  81     DW_OP_deref_type=0xa6,
  82     DW_OP_xderef_type=0xa7,
  83     DW_OP_convert=0xa8,
  84     DW_OP_reinterpret=0xa9,
  85     DW_OP_lo_user=0xe0,
  86     DW_OP_GNU_push_tls_address=0xe0,
  87     DW_OP_GNU_implicit_pointer=0xf2,
  88     DW_OP_GNU_entry_value=0xf3,
  89     DW_OP_GNU_const_type=0xf4,
  90     DW_OP_GNU_regval_type=0xf5,
  91     DW_OP_GNU_deref_type=0xf6,
  92     DW_OP_GNU_convert=0xf7,
  93     DW_OP_GNU_parameter_ref=0xfa,
  94     DW_OP_hi_user=0xff,
  95 )
  96
  97 def _generate_dynamic_values(map, prefix, index_start, index_end, value_start):
  98     """ Generate values in a map (dict) dynamically. Each key starts with
  99         a (string) prefix, followed by an index in the inclusive range
 100         [index_start, index_end]. The values start at value_start.
 101     """
 102     for index in range(index_start, index_end + 1):
 103         name = '%s%s' % (prefix, index)
 104         value = value_start + index - index_start
 105         map[name] = value
 106
 107 _generate_dynamic_values(DW_OP_name2opcode, 'DW_OP_lit', 0, 31, 0x30)
 108 _generate_dynamic_values(DW_OP_name2opcode, 'DW_OP_reg', 0, 31, 0x50)
 109 _generate_dynamic_values(DW_OP_name2opcode, 'DW_OP_breg', 0, 31, 0x70)
 110
 111 # opcode -> name mapping
 112 DW_OP_opcode2name = dict((v, k) for k, v in DW_OP_name2opcode.items())
 113
 114
 115 # Each parsed DWARF expression is returned as this type with its numeric opcode,
 116 # op name (as a string) and a list of arguments.
 117 DWARFExprOp = namedtuple('DWARFExprOp', 'op op_name args offset')
 118
 119
 120 class DWARFExprParser(object):
 121     """DWARF expression parser.
 122
 123     When initialized, requires structs to cache a dispatch table. After that,
 124     parse_expr can be called repeatedly - it's stateless.
 125     """
 126
 127     def __init__(self, structs):
 128         self._dispatch_table = _init_dispatch_table(structs)
 129
 130     def parse_expr(self, expr):
 131         """ Parses expr (a list of integers) into a list of DWARFExprOp.
 132
 133         The list can potentially be nested.
 134         """
 135         stream = BytesIO(bytelist2string(expr))
 136         parsed = []
 137
 138         while True:
 139             # Get the next opcode from the stream. If nothing is left in the
 140             # stream, we're done.
 141             offset = stream.tell()
 142             byte = stream.read(1)
 143             if len(byte) == 0:
 144                 break
 145
 146             # Decode the opcode and its name.
 147             op = ord(byte)
 148             op_name = DW_OP_opcode2name.get(op, 'OP:0x%x' % op)
 149
 150             # Use dispatch table to parse args.
 151             arg_parser = self._dispatch_table[op]
 152             args = arg_parser(stream)
 153
 154             parsed.append(DWARFExprOp(op=op, op_name=op_name, args=args, offset=offset))
 155
 156         return parsed
 157
 158
 159 def _init_dispatch_table(structs):
 160     """Creates a dispatch table for parsing args of an op.
 161
 162     Returns a dict mapping opcode to a function. The function accepts a stream
 163     and return a list of parsed arguments for the opcode from the stream;
 164     the stream is advanced by the function as needed.
 165     """
 166     table = {}
 167     def add(opcode_name, func):
 168         table[DW_OP_name2opcode[opcode_name]] = func
 169
 170     def parse_noargs():
 171         return lambda stream: []
 172
 173     def parse_op_addr():
 174         return lambda stream: [struct_parse(structs.Dwarf_target_addr(''),
 175                                             stream)]
 176
 177     def parse_arg_struct(arg_struct):
 178         return lambda stream: [struct_parse(arg_struct, stream)]
 179
 180     def parse_arg_struct2(arg1_struct, arg2_struct):
 181         return lambda stream: [struct_parse(arg1_struct, stream),
 182                                struct_parse(arg2_struct, stream)]
 183
 184     # ULEB128, then an expression of that length
 185     def parse_nestedexpr():
 186         def parse(stream):
 187             size = struct_parse(structs.Dwarf_uleb128(''), stream)
 188             nested_expr_blob = read_blob(stream, size)
 189             return [DWARFExprParser(structs).parse_expr(nested_expr_blob)]
 190         return parse
 191
 192     # ULEB128, then a blob of that size
 193     def parse_blob():
 194         return lambda stream: [read_blob(stream, struct_parse(structs.Dwarf_uleb128(''), stream))]
 195
 196     # ULEB128 with datatype DIE offset, then byte, then a blob of that size
 197     def parse_typedblob():
 198         return lambda stream: [struct_parse(structs.Dwarf_uleb128(''), stream), read_blob(stream, struct_parse(structs.Dwarf_uint8(''), stream))]
 199
 200     add('DW_OP_addr', parse_op_addr())
 201     add('DW_OP_addrx', parse_arg_struct(structs.Dwarf_uleb128('')))
 202     add('DW_OP_const1u', parse_arg_struct(structs.Dwarf_uint8('')))
 203     add('DW_OP_const1s', parse_arg_struct(structs.Dwarf_int8('')))
 204     add('DW_OP_const2u', parse_arg_struct(structs.Dwarf_uint16('')))
 205     add('DW_OP_const2s', parse_arg_struct(structs.Dwarf_int16('')))
 206     add('DW_OP_const4u', parse_arg_struct(structs.Dwarf_uint32('')))
 207     add('DW_OP_const4s', parse_arg_struct(structs.Dwarf_int32('')))
 208     add('DW_OP_const8u', parse_arg_struct(structs.Dwarf_uint64('')))
 209     add('DW_OP_const8s', parse_arg_struct(structs.Dwarf_int64('')))
 210     add('DW_OP_constu', parse_arg_struct(structs.Dwarf_uleb128('')))
 211     add('DW_OP_consts', parse_arg_struct(structs.Dwarf_sleb128('')))
 212     add('DW_OP_pick', parse_arg_struct(structs.Dwarf_uint8('')))
 213     add('DW_OP_plus_uconst', parse_arg_struct(structs.Dwarf_uleb128('')))
 214     add('DW_OP_bra', parse_arg_struct(structs.Dwarf_int16('')))
 215     add('DW_OP_skip', parse_arg_struct(structs.Dwarf_int16('')))
 216
 217     for opname in [ 'DW_OP_deref', 'DW_OP_dup', 'DW_OP_drop', 'DW_OP_over',
 218                     'DW_OP_swap', 'DW_OP_swap', 'DW_OP_rot', 'DW_OP_xderef',
 219                     'DW_OP_abs', 'DW_OP_and', 'DW_OP_div', 'DW_OP_minus',
 220                     'DW_OP_mod', 'DW_OP_mul', 'DW_OP_neg', 'DW_OP_not',
 221                     'DW_OP_or', 'DW_OP_plus', 'DW_OP_shl', 'DW_OP_shr',
 222                     'DW_OP_shra', 'DW_OP_xor', 'DW_OP_eq', 'DW_OP_ge',
 223                     'DW_OP_gt', 'DW_OP_le', 'DW_OP_lt', 'DW_OP_ne', 'DW_OP_nop',
 224                     'DW_OP_push_object_address', 'DW_OP_form_tls_address',
 225                     'DW_OP_call_frame_cfa', 'DW_OP_stack_value',
 226                     'DW_OP_GNU_push_tls_address']:
 227         add(opname, parse_noargs())
 228
 229     for n in range(0, 32):
 230         add('DW_OP_lit%s' % n, parse_noargs())
 231         add('DW_OP_reg%s' % n, parse_noargs())
 232         add('DW_OP_breg%s' % n, parse_arg_struct(structs.Dwarf_sleb128('')))
 233
 234     add('DW_OP_fbreg', parse_arg_struct(structs.Dwarf_sleb128('')))
 235     add('DW_OP_regx', parse_arg_struct(structs.Dwarf_uleb128('')))
 236     add('DW_OP_bregx', parse_arg_struct2(structs.Dwarf_uleb128(''),
 237                                          structs.Dwarf_sleb128('')))
 238     add('DW_OP_piece', parse_arg_struct(structs.Dwarf_uleb128('')))
 239     add('DW_OP_bit_piece', parse_arg_struct2(structs.Dwarf_uleb128(''),
 240                                              structs.Dwarf_uleb128('')))
 241     add('DW_OP_deref_size', parse_arg_struct(structs.Dwarf_int8('')))
 242     add('DW_OP_xderef_size', parse_arg_struct(structs.Dwarf_int8('')))
 243     add('DW_OP_call2', parse_arg_struct(structs.Dwarf_uint16('')))
 244     add('DW_OP_call4', parse_arg_struct(structs.Dwarf_uint32('')))
 245     add('DW_OP_call_ref', parse_arg_struct(structs.Dwarf_offset('')))
 246     add('DW_OP_implicit_value', parse_blob())
 247     add('DW_OP_entry_value', parse_nestedexpr())
 248     add('DW_OP_const_type', parse_typedblob())
 249     add('DW_OP_regval_type', parse_arg_struct2(structs.Dwarf_uleb128(''),
 250                                                    structs.Dwarf_uleb128('')))
 251     add('DW_OP_deref_type', parse_arg_struct2(structs.Dwarf_uint8(''),
 252                                               structs.Dwarf_uleb128('')))
 253     add('DW_OP_implicit_pointer', parse_arg_struct2(structs.Dwarf_offset(''),
 254                                                         structs.Dwarf_sleb128('')))
 255     add('DW_OP_convert', parse_arg_struct(structs.Dwarf_uleb128('')))
 256     add('DW_OP_GNU_entry_value', parse_nestedexpr())
 257     add('DW_OP_GNU_const_type', parse_typedblob())
 258     add('DW_OP_GNU_regval_type', parse_arg_struct2(structs.Dwarf_uleb128(''),
 259                                                    structs.Dwarf_uleb128('')))
 260     add('DW_OP_GNU_deref_type', parse_arg_struct2(structs.Dwarf_uint8(''),
 261                                                    structs.Dwarf_uleb128('')))
 262     add('DW_OP_GNU_implicit_pointer', parse_arg_struct2(structs.Dwarf_offset(''),
 263                                                         structs.Dwarf_sleb128('')))
 264     add('DW_OP_GNU_parameter_ref', parse_arg_struct(structs.Dwarf_offset('')))
 265     add('DW_OP_GNU_convert', parse_arg_struct(structs.Dwarf_uleb128('')))
 266
 267     return table