1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
| """
LLDB AppKit formatters
Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
See https://llvm.org/LICENSE.txt for license information.
SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
"""
# example synthetic children and summary provider for CFString (and related NSString class)
# the real code is part of the LLDB core
import lldb
import lldb.runtime.objc.objc_runtime
import lldb.formatters.Logger
try:
unichr
except NameError:
unichr = chr
def CFString_SummaryProvider(valobj, dict):
logger = lldb.formatters.Logger.Logger()
provider = CFStringSynthProvider(valobj, dict)
if not provider.invalid:
try:
summary = provider.get_child_at_index(
provider.get_child_index("content"))
if isinstance(summary, lldb.SBValue):
summary = summary.GetSummary()
else:
summary = '"' + summary + '"'
except:
summary = None
if summary is None:
summary = '<variable is not NSString>'
return '@' + summary
return ''
def CFAttributedString_SummaryProvider(valobj, dict):
logger = lldb.formatters.Logger.Logger()
offset = valobj.GetTarget().GetProcess().GetAddressByteSize()
pointee = valobj.GetValueAsUnsigned(0)
summary = '<variable is not NSAttributedString>'
if pointee is not None and pointee != 0:
pointee = pointee + offset
child_ptr = valobj.CreateValueFromAddress(
"string_ptr", pointee, valobj.GetType())
child = child_ptr.CreateValueFromAddress(
"string_data",
child_ptr.GetValueAsUnsigned(),
valobj.GetType()).AddressOf()
provider = CFStringSynthProvider(child, dict)
if not provider.invalid:
try:
summary = provider.get_child_at_index(
provider.get_child_index("content")).GetSummary()
except:
summary = '<variable is not NSAttributedString>'
if summary is None:
summary = '<variable is not NSAttributedString>'
return '@' + summary
def __lldb_init_module(debugger, dict):
debugger.HandleCommand(
"type summary add -F CFString.CFString_SummaryProvider NSString CFStringRef CFMutableStringRef")
debugger.HandleCommand(
"type summary add -F CFString.CFAttributedString_SummaryProvider NSAttributedString")
class CFStringSynthProvider:
def __init__(self, valobj, dict):
logger = lldb.formatters.Logger.Logger()
self.valobj = valobj
self.update()
# children other than "content" are for debugging only and must not be
# used in production code
def num_children(self):
logger = lldb.formatters.Logger.Logger()
if self.invalid:
return 0
return 6
def read_unicode(self, pointer, max_len=2048):
logger = lldb.formatters.Logger.Logger()
process = self.valobj.GetTarget().GetProcess()
error = lldb.SBError()
pystr = u''
# cannot do the read at once because the length value has
# a weird encoding. better play it safe here
while max_len > 0:
content = process.ReadMemory(pointer, 2, error)
new_bytes = bytearray(content)
b0 = new_bytes[0]
b1 = new_bytes[1]
pointer = pointer + 2
if b0 == 0 and b1 == 0:
break
# rearrange bytes depending on endianness
# (do we really need this or is Cocoa going to
# use Windows-compatible little-endian even
# if the target is big endian?)
if self.is_little:
value = b1 * 256 + b0
else:
value = b0 * 256 + b1
pystr = pystr + unichr(value)
# read max_len unicode values, not max_len bytes
max_len = max_len - 1
return pystr
# handle the special case strings
# only use the custom code for the tested LP64 case
def handle_special(self):
logger = lldb.formatters.Logger.Logger()
if not self.is_64_bit:
# for 32bit targets, use safe ObjC code
return self.handle_unicode_string_safe()
offset = 12
pointer = self.valobj.GetValueAsUnsigned(0) + offset
pystr = self.read_unicode(pointer)
return self.valobj.CreateValueFromExpression(
"content", "(char*)\"" + pystr.encode('utf-8') + "\"")
# last resort call, use ObjC code to read; the final aim is to
# be able to strip this call away entirely and only do the read
# ourselves
def handle_unicode_string_safe(self):
return self.valobj.CreateValueFromExpression(
"content", "(char*)\"" + self.valobj.GetObjectDescription() + "\"")
def handle_unicode_string(self):
logger = lldb.formatters.Logger.Logger()
# step 1: find offset
if self.inline:
pointer = self.valobj.GetValueAsUnsigned(
0) + self.size_of_cfruntime_base()
if not self.explicit:
# untested, use the safe code path
return self.handle_unicode_string_safe()
else:
# a full pointer is skipped here before getting to the live
# data
pointer = pointer + self.pointer_size
else:
pointer = self.valobj.GetValueAsUnsigned(
0) + self.size_of_cfruntime_base()
# read 8 bytes here and make an address out of them
try:
char_type = self.valobj.GetType().GetBasicType(
lldb.eBasicTypeChar).GetPointerType()
vopointer = self.valobj.CreateValueFromAddress(
"dummy", pointer, char_type)
pointer = vopointer.GetValueAsUnsigned(0)
except:
return self.valobj.CreateValueFromExpression(
"content", '(char*)"@\"invalid NSString\""')
# step 2: read Unicode data at pointer
pystr = self.read_unicode(pointer)
# step 3: return it
return pystr.encode('utf-8')
def handle_inline_explicit(self):
logger = lldb.formatters.Logger.Logger()
offset = 3 * self.pointer_size
offset = offset + self.valobj.GetValueAsUnsigned(0)
return self.valobj.CreateValueFromExpression(
"content", "(char*)(" + str(offset) + ")")
def handle_mutable_string(self):
logger = lldb.formatters.Logger.Logger()
offset = 2 * self.pointer_size
data = self.valobj.CreateChildAtOffset(
"content", offset, self.valobj.GetType().GetBasicType(
lldb.eBasicTypeChar).GetPointerType())
data_value = data.GetValueAsUnsigned(0)
if self.explicit and self.unicode:
return self.read_unicode(data_value).encode('utf-8')
else:
data_value = data_value + 1
return self.valobj.CreateValueFromExpression(
"content", "(char*)(" + str(data_value) + ")")
def handle_UTF8_inline(self):
logger = lldb.formatters.Logger.Logger()
offset = self.valobj.GetValueAsUnsigned(
0) + self.size_of_cfruntime_base()
if not self.explicit:
offset = offset + 1
return self.valobj.CreateValueFromAddress(
"content", offset, self.valobj.GetType().GetBasicType(
lldb.eBasicTypeChar)).AddressOf()
def handle_UTF8_not_inline(self):
logger = lldb.formatters.Logger.Logger()
offset = self.size_of_cfruntime_base()
return self.valobj.CreateChildAtOffset(
"content", offset, self.valobj.GetType().GetBasicType(
lldb.eBasicTypeChar).GetPointerType())
def get_child_at_index(self, index):
logger = lldb.formatters.Logger.Logger()
logger >> "Querying for child [" + str(index) + "]"
if index == 0:
return self.valobj.CreateValueFromExpression(
"mutable", str(int(self.mutable)))
if index == 1:
return self.valobj.CreateValueFromExpression("inline",
str(int(self.inline)))
if index == 2:
return self.valobj.CreateValueFromExpression(
"explicit", str(int(self.explicit)))
if index == 3:
return self.valobj.CreateValueFromExpression(
"unicode", str(int(self.unicode)))
if index == 4:
return self.valobj.CreateValueFromExpression(
"special", str(int(self.special)))
if index == 5:
# we are handling the several possible combinations of flags.
# for each known combination we have a function that knows how to
# go fetch the data from memory instead of running code. if a string is not
# correctly displayed, one should start by finding a combination of flags that
# makes it different from these known cases, and provide a new reader function
# if this is not possible, a new flag might have to be made up (like the "special" flag
# below, which is not a real flag in CFString), or alternatively one might need to use
# the ObjC runtime helper to detect the new class and deal with it accordingly
# print 'mutable = ' + str(self.mutable)
# print 'inline = ' + str(self.inline)
# print 'explicit = ' + str(self.explicit)
# print 'unicode = ' + str(self.unicode)
# print 'special = ' + str(self.special)
if self.mutable:
return self.handle_mutable_string()
elif self.inline and self.explicit and \
self.unicode == False and self.special == False and \
self.mutable == False:
return self.handle_inline_explicit()
elif self.unicode:
return self.handle_unicode_string()
elif self.special:
return self.handle_special()
elif self.inline:
return self.handle_UTF8_inline()
else:
return self.handle_UTF8_not_inline()
def get_child_index(self, name):
logger = lldb.formatters.Logger.Logger()
logger >> "Querying for child ['" + str(name) + "']"
if name == "content":
return self.num_children() - 1
if name == "mutable":
return 0
if name == "inline":
return 1
if name == "explicit":
return 2
if name == "unicode":
return 3
if name == "special":
return 4
# CFRuntimeBase is defined as having an additional
# 4 bytes (padding?) on LP64 architectures
# to get its size we add up sizeof(pointer)+4
# and then add 4 more bytes if we are on a 64bit system
def size_of_cfruntime_base(self):
logger = lldb.formatters.Logger.Logger()
return self.pointer_size + 4 + (4 if self.is_64_bit else 0)
# the info bits are part of the CFRuntimeBase structure
# to get at them we have to skip a uintptr_t and then get
# at the least-significant byte of a 4 byte array. If we are
# on big-endian this means going to byte 3, if we are on
# little endian (OSX & iOS), this means reading byte 0
def offset_of_info_bits(self):
logger = lldb.formatters.Logger.Logger()
offset = self.pointer_size
if not self.is_little:
offset = offset + 3
return offset
def read_info_bits(self):
logger = lldb.formatters.Logger.Logger()
cfinfo = self.valobj.CreateChildAtOffset(
"cfinfo",
self.offset_of_info_bits(),
self.valobj.GetType().GetBasicType(
lldb.eBasicTypeChar))
cfinfo.SetFormat(11)
info = cfinfo.GetValue()
if info is not None:
self.invalid = False
return int(info, 0)
else:
self.invalid = True
return None
# calculating internal flag bits of the CFString object
# this stuff is defined and discussed in CFString.c
def is_mutable(self):
logger = lldb.formatters.Logger.Logger()
return (self.info_bits & 1) == 1
def is_inline(self):
logger = lldb.formatters.Logger.Logger()
return (self.info_bits & 0x60) == 0
# this flag's name is ambiguous, it turns out
# we must skip a length byte to get at the data
# when this flag is False
def has_explicit_length(self):
logger = lldb.formatters.Logger.Logger()
return (self.info_bits & (1 | 4)) != 4
# probably a subclass of NSString. obtained this from [str pathExtension]
# here info_bits = 0 and Unicode data at the start of the padding word
# in the long run using the isa value might be safer as a way to identify this
# instead of reading the info_bits
def is_special_case(self):
logger = lldb.formatters.Logger.Logger()
return self.info_bits == 0
def is_unicode(self):
logger = lldb.formatters.Logger.Logger()
return (self.info_bits & 0x10) == 0x10
# preparing ourselves to read into memory
# by adjusting architecture-specific info
def adjust_for_architecture(self):
logger = lldb.formatters.Logger.Logger()
self.pointer_size = self.valobj.GetTarget().GetProcess().GetAddressByteSize()
self.is_64_bit = self.pointer_size == 8
self.is_little = self.valobj.GetTarget().GetProcess(
).GetByteOrder() == lldb.eByteOrderLittle
# reading info bits out of the CFString and computing
# useful values to get at the real data
def compute_flags(self):
logger = lldb.formatters.Logger.Logger()
self.info_bits = self.read_info_bits()
if self.info_bits is None:
return
self.mutable = self.is_mutable()
self.inline = self.is_inline()
self.explicit = self.has_explicit_length()
self.unicode = self.is_unicode()
self.special = self.is_special_case()
def update(self):
logger = lldb.formatters.Logger.Logger()
self.adjust_for_architecture()
self.compute_flags()
|