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
|
/*-
* Copyright (c) 2014-present MongoDB, Inc.
* Copyright (c) 2008-2014 WiredTiger, Inc.
* All rights reserved.
*
* See the file LICENSE for redistribution information.
*/
#include "wt_internal.h"
/*
* __wt_cond_alloc --
* Allocate and initialize a condition variable.
*/
int
__wt_cond_alloc(WT_SESSION_IMPL *session, const char *name, WT_CONDVAR **condp)
{
WT_CONDVAR *cond;
WT_DECL_RET;
WT_RET(__wt_calloc_one(session, &cond));
WT_ERR(pthread_mutex_init(&cond->mtx, NULL));
#ifdef HAVE_PTHREAD_COND_MONOTONIC
{
pthread_condattr_t condattr;
WT_ERR(pthread_condattr_init(&condattr));
ret = pthread_condattr_setclock(&condattr, CLOCK_MONOTONIC);
if (ret == 0)
ret = pthread_cond_init(&cond->cond, &condattr);
WT_TRET(pthread_condattr_destroy(&condattr));
WT_ERR(ret);
}
#else
WT_ERR(pthread_cond_init(&cond->cond, NULL));
#endif
cond->name = name;
cond->waiters = 0;
*condp = cond;
return (0);
err:
__wt_free(session, cond);
return (ret);
}
/*
* __wt_cond_wait_signal --
* Wait on a mutex, optionally timing out. If we get it before the time out period expires, let
* the caller know.
*/
void
__wt_cond_wait_signal(WT_SESSION_IMPL *session, WT_CONDVAR *cond, uint64_t usecs,
bool (*run_func)(WT_SESSION_IMPL *), bool *signalled)
{
struct timespec ts;
WT_DECL_RET;
WT_TRACK_OP_DECL;
bool locked;
WT_TRACK_OP_INIT(session);
locked = false;
/* Fast path if already signalled. */
*signalled = true;
if (__wt_atomic_addi32(&cond->waiters, 1) == 0) {
WT_TRACK_OP_END(session);
return;
}
__wt_verbose_debug2(session, WT_VERB_MUTEX, "wait %s", cond->name);
WT_STAT_CONN_INCR(session, cond_wait);
WT_ERR(pthread_mutex_lock(&cond->mtx));
locked = true;
/*
* It's possible to race with threads waking us up. That's not a problem if there are multiple
* wakeups because the next wakeup will get us, or if we're only pausing for a short period.
* It's a problem if there's only a single wakeup, our waker is likely waiting for us to exit.
* After acquiring the mutex (so we're guaranteed to be awakened by any future wakeup call),
* optionally check if we're OK to keep running. This won't ensure our caller won't just loop
* and call us again, but at least it's not our fault.
*
* Assert we're not waiting longer than a second if not checking the run status.
*/
WT_ASSERT(session, run_func != NULL || usecs <= WT_MILLION);
if (run_func != NULL && !run_func(session))
goto skipping;
if (usecs > 0) {
/*
* Get the current time as the basis for calculating when the wait should end. Prefer a monotonic
* clock source to avoid unexpectedly long sleeps when the system clock is adjusted.
*
* Failing that, query the time directly and don't attempt to correct for the clock moving
* backwards, which would result in a sleep that is too long by however much the clock is updated.
* This isn't as good as a monotonic clock source but makes the window of vulnerability smaller
* (i.e., the calculated time is only incorrect if the system clock changes in between us querying
* it and waiting).
*/
#ifdef HAVE_PTHREAD_COND_MONOTONIC
WT_SYSCALL_RETRY(clock_gettime(CLOCK_MONOTONIC, &ts), ret);
if (ret != 0)
WT_IGNORE_RET(__wt_panic(session, ret, "clock_gettime"));
#else
__wt_epoch_raw(session, &ts);
#endif
ts.tv_sec += (time_t)(((uint64_t)ts.tv_nsec + WT_THOUSAND * usecs) / WT_BILLION);
ts.tv_nsec = (long)(((uint64_t)ts.tv_nsec + WT_THOUSAND * usecs) % WT_BILLION);
ret = pthread_cond_timedwait(&cond->cond, &cond->mtx, &ts);
} else
ret = pthread_cond_wait(&cond->cond, &cond->mtx);
/*
* Check pthread_cond_wait() return for EINTR, ETIME and ETIMEDOUT, some systems return these
* errors.
*/
if (ret == EINTR ||
#ifdef ETIME
ret == ETIME ||
#endif
ret == ETIMEDOUT) {
skipping:
*signalled = false;
ret = 0;
}
err:
(void)__wt_atomic_subi32(&cond->waiters, 1);
if (locked)
WT_TRET(pthread_mutex_unlock(&cond->mtx));
WT_TRACK_OP_END(session);
if (ret == 0)
return;
WT_IGNORE_RET(__wt_panic(session, ret, "pthread_cond_wait: %s", cond->name));
}
/*
* __wt_cond_signal --
* Signal a waiting thread.
*/
void
__wt_cond_signal(WT_SESSION_IMPL *session, WT_CONDVAR *cond)
{
WT_DECL_RET;
__wt_verbose_debug2(session, WT_VERB_MUTEX, "signal %s", cond->name);
/*
* Our callers often set flags to cause a thread to exit. Add a barrier to ensure exit flags are
* seen by the sleeping threads, otherwise we can wake up a thread, it immediately goes back to
* sleep, and we'll hang. Use a full barrier (we may not write before waiting on thread join).
*/
WT_FULL_BARRIER();
/*
* Fast path if we are in (or can enter), a state where the next waiter will return immediately
* as already signaled.
*/
if (cond->waiters == -1 || (cond->waiters == 0 && __wt_atomic_casi32(&cond->waiters, 0, -1)))
return;
WT_ERR(pthread_mutex_lock(&cond->mtx));
ret = pthread_cond_broadcast(&cond->cond);
WT_TRET(pthread_mutex_unlock(&cond->mtx));
if (ret == 0)
return;
err:
WT_IGNORE_RET(__wt_panic(session, ret, "pthread_cond_broadcast: %s", cond->name));
}
/*
* __wt_cond_destroy --
* Destroy a condition variable.
*/
void
__wt_cond_destroy(WT_SESSION_IMPL *session, WT_CONDVAR **condp)
{
WT_CONDVAR *cond;
WT_DECL_RET;
cond = *condp;
if (cond == NULL)
return;
if ((ret = pthread_cond_destroy(&cond->cond)) != 0)
WT_IGNORE_RET(__wt_panic(session, ret, "pthread_cond_destroy: %s", cond->name));
if ((ret = pthread_mutex_destroy(&cond->mtx)) != 0)
WT_IGNORE_RET(__wt_panic(session, ret, "pthread_mutex_destroy: %s", cond->name));
__wt_free(session, *condp);
}
|
