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
|
#### $Id$
####
#### To use: create a symlink, named .purify, to this file in your home
#### directory:
#### % ln -s $ACE_ROOT/env/purify.solaris2 ~/.purify
#### Please note the leading dot in .purify.
####
#### I use these Purify options:
#### -best-effort -cache-dir=/tmp/purifycache -chain-length=20 \
#### -fds-inuse-at-exit=no -ignore-signals=SIGINT -inuse-at-exit \
#### -max_threads=100
#### These options are hard-coded in ACE wrapper_macros.GNU.
####
#### In addition to what's already in Purify's .purify.solaris2:
####
suppress miu malloc; *alloc; thr_setspecific
suppress miu malloc; realloc; thr_keycreate
suppress mlk malloc; calloc; thr_setspecific
suppress plk malloc; *alloc; thr_setspecific
suppress miu malloc; realloc; putenv
####
#### This looks like a minor problem in the g++ rx library or the Solaris 2.5
#### C library qsort.
####
suppress umr qst; qsort; rx_compactify_nfa; rx_compile; re_compile_pattern; Regex::Regex(const char*,int,int,const char*); Regex::_GLOBAL_.I.(const Regex&)
####
#### A problem in Solaris LWPs? This shows up in the ACE Thread_Manager_Test.
####
suppress umr _lwp_in_syscall; _thrp_suspend
####
#### These look like benign UMRs in the threads library.
####
suppress umr __sigprocmask; _thrp_exit; _t_cancel; _thr_exit_common
suppress umr __sigprocmask; _thrp_exit; _thr_exit_common; _age
####
#### With Solaris 2.5.1, time (and other) functions apparently hold
#### on to memory for the duration of a process. Not unexpected, but
#### the following suppress the Purify MIU reports.
####
suppress miu malloc; calloc; _tzload; _ltzset_u; localtime_u; ctime
suppress miu malloc; calloc; _tzload; _ltzset_u; localtime_u; localtime_r; ctime_r
suppress miu malloc; strdup; _tzload; _ltzset_u; localtime_u; ctime
suppress miu malloc; strdup; _tzload; _ltzset_u; localtime_u; localtime_r; ctime_r
suppress miu malloc; tzcpy; getzname; _ltzset_u; localtime_u; ctime
suppress miu malloc; tzcpy; getzname; _ltzset_u; localtime_u; localtime_r; ctime_r
suppress miu malloc; strdup; getsystemTZ; _ltzset_u; localtime_u; localtime_r; ctime_r
suppress miu malloc; _atfork_append; _ld_concurrency; _init
suppress miu malloc; Balloc; _IO_dtoa; _IO_outfloat; ostream::operator <<(double)
####
#### Another small leak, this doesn't look harmful.
####
suppress miu malloc; process_nls_path; catopen; __STATIC_CONSTRUCTOR(void); _init
####
#### Socket-related allocations, revealed by IOStream_Test or
#### SOCK_Connector_Test.
####
suppress miu malloc; _nss_XbyY_buf_alloc; gethostbyaddr
suppress miu malloc; _nss_XbyY_buf_alloc; gethostbyname
suppress miu malloc; _s_alloc_bufs; _s_alloc; _s_socreate
suppress miu malloc; add_silink; _s_alloc; _s_socreate
suppress miu malloc; calloc; fgetnetconfig; getnetlist; setnetconfig
suppress miu malloc; getnetlist; setnetconfig
suppress miu malloc; *; getnetlist; setnetconfig
suppress miu malloc; *; *; getnetlist; setnetconfig
suppress miu malloc; realloc; rx_compactify_nfa
suppress miu malloc; strdup; __rpc_getconfip; gethostby*_r
suppress miu malloc; strdup; getlookups; fgetnetconfig; getnetlist
suppress miu malloc; *; _nsw_getoneconfig; do_getconfig
suppress miu malloc; add_concell; do_getconfig; __nsw_getconfig
####
#### Allocations during OS accept call. Can be seen in the
#### Cached_Accept_Conn_Test and Cached_Conn_Test
####
suppress mlk malloc; setnetconfig; _s_match_netconf; _s_socreate; __accept; accept
####
#### With Sun C++ 4.2, there's an FMR within the ostream deletion.
####
suppress fmr unsafe_ostream::flush(void); unsafe_ostream::~unsafe_ostream(void); ostream::~ostream(void); iostream::~iostream(void)
####
#### Problem with FMR in thread-specific storage? The Purify
#### trace doesn't show where the allocated storage was freed;
#### it usually does for FMRs.
####
#### **** Purify instrumented ./TSS_Test (pid 16650) ****
#### FMR: Free memory read:
#### * This is occurring while in thread 5:
#### thr_getspecific [libthread.so.1]
#### worker(void*) [OS.i:3975]
#### _thread_start [libthread.so.1]
#### * Reading 4 bytes from 0x7e8c8 in the heap.
#### * Address 0x7e8c8 is 24 bytes into a freed block at 0x7e8b0 of 32 bytes.
#### * This block was allocated from thread 4:
#### malloc [rtlib.o]
#### realloc [rtlib.o]
#### thr_keycreate [libthread.so.1]
#### ACE_OS::thr_keycreate(unsigned int*,void(*)(void*),void*) [OS.cpp:1938]
#### worker(void*) [TSS_Test.cpp:125]
#### _thread_start [libthread.so.1]
#### * There have been 0 frees since this block was freed thread 4.
####
#### **** Purify instrumented ./TSS_Test (pid 16650) ****
#### FMR: Free memory read:
#### * This is occurring while in thread 5:
#### thr_setspecific [libthread.so.1]
#### ACE_OS::thr_setspecific(unsigned int,void*) [OS.cpp:1866]
#### worker(void*) [TSS_Test.cpp:98]
#### _thread_start [libthread.so.1]
#### * Reading 4 bytes from 0x7e8c8 in the heap.
#### * Address 0x7e8c8 is 24 bytes into a freed block at 0x7e8b0 of 32 bytes.
#### * This block was allocated from thread 4:
#### malloc [rtlib.o]
#### realloc [rtlib.o]
#### thr_keycreate [libthread.so.1]
#### ACE_OS::thr_keycreate(unsigned int*,void(*)(void*),void*) [OS.cpp:1938]
#### worker(void*) [TSS_Test.cpp:125]
#### _thread_start [libthread.so.1]
#### * There have been 0 frees since this block was freed thread 4.
####
suppress fmr thr_getspecific
suppress fmr thr_setspecific
####
#### Write of one byte beyond the stack in ::pthread_setcanceltype ().
#### This code fragment exhibits the BSW error:
#### int result;
#### return ::pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, &result);
####
suppress bsw pthread_setcanceltype
####
#### TSS leaks: on Solaris, TSS isn't cleaned up very well.
####
suppress mlk malloc; __0OnWuI; operator new(unsigned int); ACE_TSS<*>::make_TSS_TYPE(void) const
# Don't seem to need these any more:
# suppress miu malloc; __bUiLtIn_nEw; __builtin_new; ACE_Task_Exit::instance ()
# suppress miu malloc; __bUiLtIn_nEw; __builtin_new; ACE_TSS<*>::make_TSS_TYPE () const
# suppress mlk malloc; __bUiLtIn_nEw; __builtin_new; ACE_TSS<*>::make_TSS_TYPE () const
# suppress plk malloc; __bUiLtIn_nEw; __builtin_new; ACE_TSS<*>::make_TSS_TYPE () const
####
#### SunOS 5.7's thread library has these leaks.
####
suppress miu malloc; _atfork_append; _t0init; _init
suppress miu malloc; _atfork_append; _ld_concurrency; _t0init; _init
suppress umr *; get_info_by_class; sched_getparam
suppress umr *; *; get_info_by_class; sched_getparam
suppress umr get_local_info; order_haddrlist; _get_hostserv_inetnetdir_byname; gethostbyname_r
#### EOF
|
