coff-dump.py revision 6580a7a6cc637f6d757b8d4fc6faebfa334dbbb0 
1#!/usr/bin/env python
2#===-- coff-dump.py - COFF object file dump utility-------------------------===#
3#
4#                     The LLVM Compiler Infrastructure
5#
6# This file is distributed under the University of Illinois Open Source
7# License. See LICENSE.TXT for details.
8#
9#===------------------------------------------------------------------------===#
10
11#
12# COFF File Definition
13#
14
15def string_table_entry (offset):
16  return ('ptr', '+ + PointerToSymbolTable * NumberOfSymbols 18 %s' % offset, ('scalar', 'cstr', '%s'))
17
18def secname(value):
19  if value[0] == '/':
20    return string_table_entry(value[1:].rstrip('\0'))
21  else:
22    return '%s'
23
24def symname(value):
25  parts = struct.unpack("<2L", value)
26  if parts[0] == 0:
27    return string_table_entry(parts[1])
28  else:
29    return '%s'
30
31file = ('struct', [
32  ('MachineType', ('enum', '<H', '0x%X', {
33    0x0:    'IMAGE_FILE_MACHINE_UNKNOWN',
34    0x1d3:  'IMAGE_FILE_MACHINE_AM33',
35    0x8664: 'IMAGE_FILE_MACHINE_AMD64',
36    0x1c0:  'IMAGE_FILE_MACHINE_ARM',
37    0xebc:  'IMAGE_FILE_MACHINE_EBC',
38    0x14c:  'IMAGE_FILE_MACHINE_I386',
39    0x200:  'IMAGE_FILE_MACHINE_IA64',
40    0x904:  'IMAGE_FILE_MACHINE_M32R',
41    0x266:  'IMAGE_FILE_MACHINE_MIPS16',
42    0x366:  'IMAGE_FILE_MACHINE_MIPSFPU',
43    0x466:  'IMAGE_FILE_MACHINE_MIPSFPU16',
44    0x1f0:  'IMAGE_FILE_MACHINE_POWERPC',
45    0x1f1:  'IMAGE_FILE_MACHINE_POWERPCFP',
46    0x166:  'IMAGE_FILE_MACHINE_R4000',
47    0x1a2:  'IMAGE_FILE_MACHINE_SH3',
48    0x1a3:  'IMAGE_FILE_MACHINE_SH3DSP',
49    0x1a6:  'IMAGE_FILE_MACHINE_SH4',
50    0x1a8:  'IMAGE_FILE_MACHINE_SH5',
51    0x1c2:  'IMAGE_FILE_MACHINE_THUMB',
52    0x169:  'IMAGE_FILE_MACHINE_WCEMIPSV2',
53  })),
54  ('NumberOfSections',     ('scalar',  '<H', '%d')),
55  ('TimeDateStamp',        ('scalar',  '<L', '%d')),
56  ('PointerToSymbolTable', ('scalar',  '<L', '0x%0X')),
57  ('NumberOfSymbols',      ('scalar',  '<L', '%d')),
58  ('SizeOfOptionalHeader', ('scalar',  '<H', '%d')),
59  ('Characteristics',      ('flags',   '<H', '0x%x', [
60    (0x0001,      'IMAGE_FILE_RELOCS_STRIPPED',         ),
61    (0x0002,      'IMAGE_FILE_EXECUTABLE_IMAGE',        ),
62    (0x0004,      'IMAGE_FILE_LINE_NUMS_STRIPPED',      ),
63    (0x0008,      'IMAGE_FILE_LOCAL_SYMS_STRIPPED',     ),
64    (0x0010,      'IMAGE_FILE_AGGRESSIVE_WS_TRIM',      ),
65    (0x0020,      'IMAGE_FILE_LARGE_ADDRESS_AWARE',     ),
66    (0x0080,      'IMAGE_FILE_BYTES_REVERSED_LO',       ),
67    (0x0100,      'IMAGE_FILE_32BIT_MACHINE',           ),
68    (0x0200,      'IMAGE_FILE_DEBUG_STRIPPED',          ),
69    (0x0400,      'IMAGE_FILE_REMOVABLE_RUN_FROM_SWAP', ),
70    (0x0800,      'IMAGE_FILE_NET_RUN_FROM_SWAP',       ),
71    (0x1000,      'IMAGE_FILE_SYSTEM',                  ),
72    (0x2000,      'IMAGE_FILE_DLL',                     ),
73    (0x4000,      'IMAGE_FILE_UP_SYSTEM_ONLY',          ),
74    (0x8000,      'IMAGE_FILE_BYTES_REVERSED_HI',       ),
75  ])),
76  ('Sections', ('array', '1', 'NumberOfSections', ('struct', [
77    ('Name',                 ('scalar',  '<8s', secname)),
78    ('VirtualSize',          ('scalar',  '<L',  '%d'   )),
79    ('VirtualAddress',       ('scalar',  '<L',  '%d'   )),
80    ('SizeOfRawData',        ('scalar',  '<L',  '%d'   )),
81    ('PointerToRawData',     ('scalar',  '<L',  '0x%X' )),
82    ('PointerToRelocations', ('scalar',  '<L',  '0x%X' )),
83    ('PointerToLineNumbers', ('scalar',  '<L',  '0x%X' )),
84    ('NumberOfRelocations',  ('scalar',  '<H',  '%d'   )),
85    ('NumberOfLineNumbers',  ('scalar',  '<H',  '%d'   )),
86    ('Charateristics',       ('flags',   '<L',  '0x%X', [
87      (0x00000008, 'IMAGE_SCN_TYPE_NO_PAD'),
88      (0x00000020, 'IMAGE_SCN_CNT_CODE'),
89      (0x00000040, 'IMAGE_SCN_CNT_INITIALIZED_DATA'),
90      (0x00000080, 'IMAGE_SCN_CNT_UNINITIALIZED_DATA'),
91      (0x00000100, 'IMAGE_SCN_LNK_OTHER'),
92      (0x00000200, 'IMAGE_SCN_LNK_INFO'),
93      (0x00000800, 'IMAGE_SCN_LNK_REMOVE'),
94      (0x00001000, 'IMAGE_SCN_LNK_COMDAT'),
95      (0x00008000, 'IMAGE_SCN_GPREL'),
96      (0x00020000, 'IMAGE_SCN_MEM_PURGEABLE'),
97      (0x00020000, 'IMAGE_SCN_MEM_16BIT'),
98      (0x00040000, 'IMAGE_SCN_MEM_LOCKED'),
99      (0x00080000, 'IMAGE_SCN_MEM_PRELOAD'),
100      (0x00F00000, 'IMAGE_SCN_ALIGN', {
101        0x00100000: 'IMAGE_SCN_ALIGN_1BYTES',
102        0x00200000: 'IMAGE_SCN_ALIGN_2BYTES',
103        0x00300000: 'IMAGE_SCN_ALIGN_4BYTES',
104        0x00400000: 'IMAGE_SCN_ALIGN_8BYTES',
105        0x00500000: 'IMAGE_SCN_ALIGN_16BYTES',
106        0x00600000: 'IMAGE_SCN_ALIGN_32BYTES',
107        0x00700000: 'IMAGE_SCN_ALIGN_64BYTES',
108        0x00800000: 'IMAGE_SCN_ALIGN_128BYTES',
109        0x00900000: 'IMAGE_SCN_ALIGN_256BYTES',
110        0x00A00000: 'IMAGE_SCN_ALIGN_512BYTES',
111        0x00B00000: 'IMAGE_SCN_ALIGN_1024BYTES',
112        0x00C00000: 'IMAGE_SCN_ALIGN_2048BYTES',
113        0x00D00000: 'IMAGE_SCN_ALIGN_4096BYTES',
114        0x00E00000: 'IMAGE_SCN_ALIGN_8192BYTES',
115      }),
116      (0x01000000, 'IMAGE_SCN_LNK_NRELOC_OVFL'),
117      (0x02000000, 'IMAGE_SCN_MEM_DISCARDABLE'),
118      (0x04000000, 'IMAGE_SCN_MEM_NOT_CACHED'),
119      (0x08000000, 'IMAGE_SCN_MEM_NOT_PAGED'),
120      (0x10000000, 'IMAGE_SCN_MEM_SHARED'),
121      (0x20000000, 'IMAGE_SCN_MEM_EXECUTE'),
122      (0x40000000, 'IMAGE_SCN_MEM_READ'),
123      (0x80000000, 'IMAGE_SCN_MEM_WRITE'),
124    ])),
125    ('SectionData', ('ptr', 'PointerToRawData', ('blob', 'SizeOfRawData'))),
126    ('Relocations', ('ptr', 'PointerToRelocations', ('array', '0', 'NumberOfRelocations', ('struct', [
127      ('VirtualAddress',   ('scalar', '<L', '0x%X')),
128      ('SymbolTableIndex', ('scalar', '<L', '%d'  )),
129      ('Type',             ('enum', '<H', '%d', ('MachineType', {
130        0x14c: {
131          0x0000: 'IMAGE_REL_I386_ABSOLUTE',
132          0x0001: 'IMAGE_REL_I386_DIR16',
133          0x0002: 'IMAGE_REL_I386_REL16',
134          0x0006: 'IMAGE_REL_I386_DIR32',
135          0x0007: 'IMAGE_REL_I386_DIR32NB',
136          0x0009: 'IMAGE_REL_I386_SEG12',
137          0x000A: 'IMAGE_REL_I386_SECTION',
138          0x000B: 'IMAGE_REL_I386_SECREL',
139          0x000C: 'IMAGE_REL_I386_TOKEN',
140          0x000D: 'IMAGE_REL_I386_SECREL7',
141          0x0014: 'IMAGE_REL_I386_REL32',
142        },
143        0x8664: {
144          0x0000: 'IMAGE_REL_AMD64_ABSOLUTE',
145          0x0001: 'IMAGE_REL_AMD64_ADDR64',
146          0x0002: 'IMAGE_REL_AMD64_ADDR32',
147          0x0003: 'IMAGE_REL_AMD64_ADDR32NB',
148          0x0004: 'IMAGE_REL_AMD64_REL32',
149          0x0005: 'IMAGE_REL_AMD64_REL32_1',
150          0x0006: 'IMAGE_REL_AMD64_REL32_2',
151          0x0007: 'IMAGE_REL_AMD64_REL32_3',
152          0x0008: 'IMAGE_REL_AMD64_REL32_4',
153          0x0009: 'IMAGE_REL_AMD64_REL32_5',
154          0x000A: 'IMAGE_REL_AMD64_SECTION',
155          0x000B: 'IMAGE_REL_AMD64_SECREL',
156          0x000C: 'IMAGE_REL_AMD64_SECREL7',
157          0x000D: 'IMAGE_REL_AMD64_TOKEN',
158          0x000E: 'IMAGE_REL_AMD64_SREL32',
159          0x000F: 'IMAGE_REL_AMD64_PAIR',
160          0x0010: 'IMAGE_REL_AMD64_SSPAN32',
161        },
162      }))),
163      ('SymbolName',       ('ptr', '+ PointerToSymbolTable * SymbolTableIndex 18', ('scalar',  '<8s', symname)))
164    ])))),
165  ]))),
166  ('Symbols', ('ptr', 'PointerToSymbolTable', ('byte-array', '18', '* NumberOfSymbols 18',  ('struct', [
167    ('Name',                ('scalar',  '<8s', symname)),
168    ('Value',               ('scalar',  '<L',  '%d'   )),
169    ('SectionNumber',       ('scalar',  '<H',  '%d'   )),
170    ('SimpleType',          ('enum',    '<B',  '%d', {
171      0: 'IMAGE_SYM_TYPE_NULL',
172      1: 'IMAGE_SYM_TYPE_VOID',
173      2: 'IMAGE_SYM_TYPE_CHAR',
174      3: 'IMAGE_SYM_TYPE_SHORT',
175      4: 'IMAGE_SYM_TYPE_INT',
176      5: 'IMAGE_SYM_TYPE_LONG',
177      6: 'IMAGE_SYM_TYPE_FLOAT',
178      7: 'IMAGE_SYM_TYPE_DOUBLE',
179      8: 'IMAGE_SYM_TYPE_STRUCT',
180      9: 'IMAGE_SYM_TYPE_UNION',
181      10: 'IMAGE_SYM_TYPE_ENUM',
182      11: 'IMAGE_SYM_TYPE_MOE',
183      12: 'IMAGE_SYM_TYPE_BYTE',
184      13: 'IMAGE_SYM_TYPE_WORD',
185      14: 'IMAGE_SYM_TYPE_UINT',
186      15: 'IMAGE_SYM_TYPE_DWORD',
187    })),
188    ('ComplexType',         ('enum',    '<B',  '%d', {
189      0: 'IMAGE_SYM_DTYPE_NULL',
190      1: 'IMAGE_SYM_DTYPE_POINTER',
191      2: 'IMAGE_SYM_DTYPE_FUNCTION',
192      3: 'IMAGE_SYM_DTYPE_ARRAY',
193    })),
194    ('StorageClass',        ('enum',    '<B',  '%d', {
195      -1:  'IMAGE_SYM_CLASS_END_OF_FUNCTION',
196      0: 'IMAGE_SYM_CLASS_NULL',
197      1: 'IMAGE_SYM_CLASS_AUTOMATIC',
198      2: 'IMAGE_SYM_CLASS_EXTERNAL',
199      3: 'IMAGE_SYM_CLASS_STATIC',
200      4: 'IMAGE_SYM_CLASS_REGISTER',
201      5: 'IMAGE_SYM_CLASS_EXTERNAL_DEF',
202      6: 'IMAGE_SYM_CLASS_LABEL',
203      7: 'IMAGE_SYM_CLASS_UNDEFINED_LABEL',
204      8: 'IMAGE_SYM_CLASS_MEMBER_OF_STRUCT',
205      9: 'IMAGE_SYM_CLASS_ARGUMENT',
206      10: 'IMAGE_SYM_CLASS_STRUCT_TAG',
207      11: 'IMAGE_SYM_CLASS_MEMBER_OF_UNION',
208      12: 'IMAGE_SYM_CLASS_UNION_TAG',
209      13: 'IMAGE_SYM_CLASS_TYPE_DEFINITION',
210      14: 'IMAGE_SYM_CLASS_UNDEFINED_STATIC',
211      15: 'IMAGE_SYM_CLASS_ENUM_TAG',
212      16: 'IMAGE_SYM_CLASS_MEMBER_OF_ENUM',
213      17: 'IMAGE_SYM_CLASS_REGISTER_PARAM',
214      18: 'IMAGE_SYM_CLASS_BIT_FIELD',
215      100: 'IMAGE_SYM_CLASS_BLOCK',
216      101: 'IMAGE_SYM_CLASS_FUNCTION',
217      102: 'IMAGE_SYM_CLASS_END_OF_STRUCT',
218      103: 'IMAGE_SYM_CLASS_FILE',
219      104: 'IMAGE_SYM_CLASS_SECTION',
220      105: 'IMAGE_SYM_CLASS_WEAK_EXTERNAL',
221      107: 'IMAGE_SYM_CLASS_CLR_TOKEN',
222    })),
223    ('NumberOfAuxSymbols',  ('scalar',  '<B',  '%d'  )),
224    ('AuxillaryData', ('blob', '* NumberOfAuxSymbols 18')),
225  ])))),
226])
227
228#
229# Definition Interpreter
230#
231
232import sys, types, struct, re
233
234Input = None
235Stack = []
236Fields = {}
237
238Indent = 0
239NewLine = True
240
241def indent():
242  global Indent
243  Indent += 1
244
245def dedent():
246  global Indent
247  Indent -= 1
248
249def write(input):
250  global NewLine
251  output = ""
252
253  for char in input:
254
255    if NewLine:
256      output += Indent * '  '
257      NewLine = False
258
259    output += char
260
261    if char == '\n':
262      NewLine = True
263
264  sys.stdout.write(output)
265
266def read(format):
267  return struct.unpack(format, Input.read(struct.calcsize(format)))
268
269def read_cstr():
270  output = ""
271  while True:
272    char = Input.read(1)
273    if len(char) == 0:
274      raise RuntimeError ("EOF while reading cstr")
275    if char == '\0':
276      break
277    output += char
278  return output
279
280def push_pos(seek_to = None):
281  Stack [0:0] = [Input.tell()]
282  if seek_to:
283    Input.seek(seek_to)
284
285def pop_pos():
286  assert(len(Stack) > 0)
287  Input.seek(Stack[0])
288  del Stack[0]
289
290def print_binary_data(size):
291  value = ""
292  while size > 0:
293    if size >= 16:
294      data = Input.read(16)
295      size -= 16
296    else:
297      data = Input.read(size)
298      size = 0
299    value += data
300    bytes = ""
301    text = ""
302    for index in xrange(16):
303      if index < len(data):
304        if index == 8:
305          bytes += "- "
306        ch = ord(data[index])
307        bytes += "%02X " % ch
308        if ch >= 0x20 and ch <= 0x7F:
309          text += data[index]
310        else:
311          text += "."
312      else:
313        if index == 8:
314          bytes += "  "
315        bytes += "   "
316
317    write("%s|%s|\n" % (bytes, text))
318  return value
319
320idlit = re.compile("[a-zA-Z][a-zA-Z0-9_-]*")
321numlit = re.compile("[0-9]+")
322
323def read_value(expr):
324
325  input = iter(expr.split())
326
327  def eval():
328
329    token = input.next()
330
331    if expr == 'cstr':
332      return read_cstr()
333    if expr == 'true':
334      return True
335    if expr == 'false':
336      return False
337
338    if len(token) > 1 and token[0] in ('=', '@', '<', '!', '>'):
339      val = read(expr)
340      assert(len(val) == 1)
341      return val[0]
342
343    if token == '+':
344      return eval() + eval()
345    if token == '-':
346      return eval() - eval()
347    if token == '*':
348      return eval() * eval()
349    if token == '/':
350      return eval() / eval()
351
352    if idlit.match(token):
353      return Fields[token]
354    if numlit.match(token):
355      return int(token)
356
357    raise RuntimeError("unexpected token %s" % repr(token))
358
359  value = eval ()
360
361  try:
362    input.next ()
363  except StopIteration:
364    return value
365  raise RuntimeError("unexpected input at end of expression")
366
367def write_value(format,value):
368  format_type = type(format)
369  if format_type is types.StringType:
370    write(format % value)
371  elif format_type is types.FunctionType:
372    write_value(format(value), value)
373  elif format_type is types.TupleType:
374    Fields['this'] = value
375    handle_element(format)
376  else:
377    raise RuntimeError("unexpected type: %s" % repr(format_type))
378
379def handle_scalar(entry):
380  iformat = entry[1]
381  oformat = entry[2]
382
383  value = read_value(iformat)
384
385  write_value(oformat, value)
386
387  return value
388
389def handle_enum(entry):
390  iformat = entry[1]
391  oformat = entry[2]
392  definitions = entry[3]
393
394  value = read_value(iformat)
395
396  if type(definitions) is types.TupleType:
397    selector = read_value(definitions[0])
398    definitions = definitions[1][selector]
399
400  if value in definitions:
401    description = definitions[value]
402  else:
403    description = "unknown"
404
405  write("%s (" % description)
406  write_value(oformat, value)
407  write(")")
408
409  return value
410
411def handle_flags(entry):
412  iformat = entry[1]
413  oformat = entry[2]
414  definitions = entry[3]
415
416  value = read_value(iformat)
417
418  write_value(oformat, value)
419
420  indent()
421  for entry in definitions:
422    mask = entry[0]
423    name = entry[1]
424    if len (entry) == 3:
425      map = entry[2]
426      selection = value & mask
427      if selection in map:
428        write("\n%s" % map[selection])
429      else:
430        write("\n%s <%d>" % (name, selection))
431    elif len(entry) == 2:
432      if value & mask != 0:
433        write("\n%s" % name)
434  dedent()
435
436  return value
437
438def handle_struct(entry):
439  global Fields
440  members = entry[1]
441
442  newFields = {}
443
444  write("{\n");
445  indent()
446
447  for member in members:
448    name = member[0]
449    type = member[1]
450
451    write("%s = "%name.ljust(24))
452
453    value = handle_element(type)
454
455    write("\n")
456
457    Fields[name] = value
458    newFields[name] = value
459
460  dedent()
461  write("}")
462
463  return newFields
464
465def handle_array(entry):
466  start_index = entry[1]
467  length = entry[2]
468  element = entry[3]
469
470  newItems = []
471
472  write("[\n")
473  indent()
474
475  start_index = read_value(start_index)
476  value = read_value(length)
477
478  for index in xrange(value):
479    write("%d = " % (index + start_index))
480    value = handle_element(element)
481    write("\n")
482    newItems.append(value)
483
484  dedent()
485  write("]")
486
487  return newItems
488
489def handle_byte_array(entry):
490  ent_size = entry[1]
491  length = entry[2]
492  element = entry[3]
493
494  newItems = []
495
496  write("[\n")
497  indent()
498
499  item_size = read_value(ent_size)
500  value = read_value(length)
501  end_of_array = Input.tell() + value
502
503  prev_loc = Input.tell()
504  index = 0
505  while Input.tell() < end_of_array:
506    write("%d = " % index)
507    value = handle_element(element)
508    write("\n")
509    newItems.append(value)
510    index += (Input.tell() - prev_loc) / item_size
511    prev_loc = Input.tell()
512
513  dedent()
514  write("]")
515
516  return newItems
517
518def handle_ptr(entry):
519  offset = entry[1]
520  element = entry[2]
521
522  value = None
523  offset = read_value(offset)
524
525  if offset != 0:
526
527    push_pos(offset)
528
529    value = handle_element(element)
530
531    pop_pos()
532
533  else:
534    write("None")
535
536  return value
537
538def handle_blob(entry):
539  length = entry[1]
540
541  write("\n")
542  indent()
543
544  value = print_binary_data(read_value(length))
545
546  dedent()
547
548  return value
549
550def handle_element(entry):
551  handlers = {
552    'struct':      handle_struct,
553    'scalar':      handle_scalar,
554    'enum':        handle_enum,
555    'flags':       handle_flags,
556    'ptr':         handle_ptr,
557    'blob':        handle_blob,
558    'array':       handle_array,
559    'byte-array':  handle_byte_array,
560  }
561
562  if not entry[0] in handlers:
563    raise RuntimeError ("unexpected type '%s'" % str (entry[0]))
564
565  return handlers[entry[0]](entry)
566
567if len(sys.argv) <= 1 or sys.argv[1] == '-':
568  import StringIO
569  Input = StringIO.StringIO(sys.stdin.read())
570else:
571  Input = open (sys.argv[1], "rb")
572
573try:
574  handle_element(file)
575finally:
576  Input.close()
577  Input = None
578