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
|
#!/usr/bin/python3
# -*- tab-width: 4; indent-tabs-mode: nil; py-indent-offset: 4 -*-
#
# This file is part of the LibreOffice project.
#
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this
# file, You can obtain one at http://mozilla.org/MPL/2.0/.
#
### script to help debug leaks of reference counted objects
## I. to use it, first override acquire() and release()
# Foo * g_pTrackedFoo = 0;
# Foo::Foo()
# static int nFoos = 0;
# if (++nFoos == 42) // track instance #42
# g_pTrackedFoo = this;
# void Foo::acquire()
# if (this == g_pTrackedFoo)
# ; // set gdb breakpoint here
# Foo_Base::acquire()
# void Foo::release()
# if (this == g_pTrackedFoo)
# ; // set gdb breakpoint here
# Foo_Base::release()
## II. run test/soffice in gdb and set breakpoints in acquire/release
## with a command to print the backtrace
# set logging enabled on
# break foo.cxx:123
# break foo.cxx:234
# command 1 2
# bt
# c
# end
# run
## III. now feed logfile gdb.txt into this script
# bin/refcount_leak.py < gdb.txt
###
from operator import itemgetter
import re
import sys
threshold = 2
class Trace:
clock = 0 # global counter
# frames: list of stack frames, beginning with outermost
def __init__(self, lines):
lines.reverse()
self.frames = lines
Trace.clock += 1
self.clock = Trace.clock
def addTrace(traces, lines):
if traces is not None and len(lines) > 0:
traces.append(Trace(lines))
def readGdbLog(infile):
traces_acquire = []
traces_release = []
current = None
lines = []
apattern = re.compile("^Breakpoint.*::acquire")
rpattern = re.compile("^Breakpoint.*::release")
for line in infile:
if apattern.match(line):
addTrace(current, lines)
lines = []
current = traces_acquire
if rpattern.match(line):
addTrace(current, lines)
lines = []
current = traces_release
if line.startswith("#"):
# strip #123 stack frame number, and newline
lines.append(line[line.index("0x"):-1])
addTrace(current, lines)
print("# parsed traces acquire: ", len(traces_acquire))
print("# parsed traces release: ", len(traces_release))
return (traces_acquire, traces_release)
def getFunction(frame):
start = frame.index(" in ") + len(" in ")
try:
end = frame.index(" at ", start)
except ValueError:
# argh... stack frames may be split across multiple lines if
# a parameter has a fancy pretty printer
return frame[start:]
return frame[start:end]
def matchStack(trace_acquire, trace_release):
if trace_release.clock < trace_acquire.clock:
return None # acquire must precede release
common = 0
refpattern = re.compile(r"::Reference<.*>::Reference\(")
for (frame1, frame2) in zip(trace_release.frames, trace_acquire.frames):
if frame1 == frame2:
common += 1
else:
if getFunction(frame1) == getFunction(frame2):
common += 1
acquireframes = len(trace_acquire.frames)
# there is sometimes a dozen frames of UNO type related junk
# on the stack where the acquire() happens, which breaks the
# matching; try to avoid that
for i in range(common, acquireframes):
if refpattern.search(trace_acquire.frames[i]):
acquireframes = i+1 # cut off junk above Reference ctor
break
score = max(len(trace_release.frames), acquireframes) - common
# smaller score is better
return (score, trace_release.clock - trace_acquire.clock)
# brute force greedy n^2 matching
def matchStacks(traces_acquire, traces_release):
matches = []
for release in traces_release:
for acquire in traces_acquire:
score = matchStack(acquire, release)
if score is not None:
matches.append((score, acquire, release))
matches.sort(key=itemgetter(0))
return matches
def bestMatches(traces_acquire, traces_release, matches):
traces_aunmatched = traces_acquire
traces_runmatched = traces_release
bestmatches = []
for (score,acquire,release) in matches:
if not(acquire in traces_aunmatched and release in traces_runmatched):
continue
traces_aunmatched.remove(acquire)
traces_runmatched.remove(release)
bestmatches.append((score,acquire,release))
print("# unmatched acquire: ", len(traces_aunmatched))
print("# unmatched release: ", len(traces_runmatched))
return (bestmatches,traces_aunmatched,traces_runmatched)
def printTrace(trace):
for frame in reversed(trace.frames):
print(" ", frame)
def printMatched(bestmatches):
for (score,acquire,release) in reversed(bestmatches):
print("\n*** Matched trace with score: ", score)
print(" acquire: ")
printTrace(acquire)
print(" release: ")
printTrace(release)
def printUnmatched(traces, prefix):
for trace in traces:
print("\n*** Unmatched trace (", prefix, "):")
printTrace(trace)
if __name__ == "__main__":
(traces_acquire, traces_release) = readGdbLog(sys.stdin)
matches = matchStacks(traces_acquire, traces_release)
(bestmatches,traces_au,traces_ru) = bestMatches(traces_acquire, traces_release, matches)
# print output, sorted with the most suspicious stuff first:
printUnmatched(traces_au, "acquire")
printUnmatched(traces_ru, "release")
printMatched(bestmatches)
# vim:set shiftwidth=4 softtabstop=4 expandtab:
|
