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
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
|
/*****************************************************************************
Copyright (c) 2013, 2015, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2019, MariaDB Corporation.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1335 USA
*****************************************************************************/
/**************************************************//**
@file os/os0event.cc
The interface to the operating system condition variables.
Created 2012-09-23 Sunny Bains
*******************************************************/
#include "os0event.h"
#include "ut0mutex.h"
#include <my_sys.h>
#ifdef _WIN32
#include <windows.h>
#include <synchapi.h>
/** Native condition variable. */
typedef CONDITION_VARIABLE os_cond_t;
#else
/** Native condition variable */
typedef pthread_cond_t os_cond_t;
#endif /* _WIN32 */
/** InnoDB condition variable. */
struct os_event {
os_event() UNIV_NOTHROW;
~os_event() UNIV_NOTHROW;
/**
Destroys a condition variable */
void destroy() UNIV_NOTHROW
{
#ifndef _WIN32
int ret = pthread_cond_destroy(&cond_var);
ut_a(ret == 0);
#endif /* !_WIN32 */
mutex.destroy();
}
/** Set the event */
void set() UNIV_NOTHROW
{
mutex.enter();
if (!m_set) {
broadcast();
}
mutex.exit();
}
int64_t reset() UNIV_NOTHROW
{
mutex.enter();
if (m_set) {
m_set = false;
}
int64_t ret = signal_count;
mutex.exit();
return(ret);
}
/**
Waits for an event object until it is in the signaled state.
Typically, if the event has been signalled after the os_event_reset()
we'll return immediately because event->m_set == true.
There are, however, situations (e.g.: sync_array code) where we may
lose this information. For example:
thread A calls os_event_reset()
thread B calls os_event_set() [event->m_set == true]
thread C calls os_event_reset() [event->m_set == false]
thread A calls os_event_wait() [infinite wait!]
thread C calls os_event_wait() [infinite wait!]
Where such a scenario is possible, to avoid infinite wait, the
value returned by reset() should be passed in as
reset_sig_count. */
void wait_low(int64_t reset_sig_count) UNIV_NOTHROW;
/**
Waits for an event object until it is in the signaled state or
a timeout is exceeded.
@param time_in_usec - timeout in microseconds,
or OS_SYNC_INFINITE_TIME
@param reset_sig_count- zero or the value returned by
previous call of os_event_reset().
@return 0 if success, OS_SYNC_TIME_EXCEEDED if timeout was exceeded */
ulint wait_time_low(
ulint time_in_usec,
int64_t reset_sig_count) UNIV_NOTHROW;
/** @return true if the event is in the signalled state. */
bool is_set() const UNIV_NOTHROW
{
mutex.enter();
bool is_set = m_set;
mutex.exit();
return is_set;
}
private:
/**
Initialize a condition variable */
void init() UNIV_NOTHROW
{
mutex.init();
#ifdef _WIN32
InitializeConditionVariable(&cond_var);
#else
{
int ret;
ret = pthread_cond_init(&cond_var, NULL);
ut_a(ret == 0);
}
#endif /* _WIN32 */
}
/**
Wait on condition variable */
void wait() UNIV_NOTHROW
{
#ifdef _WIN32
if (!SleepConditionVariableCS(&cond_var, mutex, INFINITE)) {
ut_error;
}
#else
{
int ret;
ret = pthread_cond_wait(&cond_var, mutex);
ut_a(ret == 0);
}
#endif /* _WIN32 */
}
/**
Wakes all threads waiting for condition variable */
void broadcast() UNIV_NOTHROW
{
m_set = true;
++signal_count;
#ifdef _WIN32
WakeAllConditionVariable(&cond_var);
#else
{
int ret;
ret = pthread_cond_broadcast(&cond_var);
ut_a(ret == 0);
}
#endif /* _WIN32 */
}
/**
Wakes one thread waiting for condition variable */
void signal() UNIV_NOTHROW
{
#ifdef _WIN32
WakeConditionVariable(&cond_var);
#else
{
int ret;
ret = pthread_cond_signal(&cond_var);
ut_a(ret == 0);
}
#endif /* _WIN32 */
}
/**
Do a timed wait on condition variable.
@param abstime - timeout
@param time_in_ms - timeout in milliseconds.
@return true if timed out, false otherwise */
bool timed_wait(
#ifndef _WIN32
const timespec* abstime
#else
DWORD time_in_ms
#endif /* !_WIN32 */
);
private:
bool m_set; /*!< this is true when the
event is in the signaled
state, i.e., a thread does
not stop if it tries to wait
for this event */
int64_t signal_count; /*!< this is incremented
each time the event becomes
signaled */
mutable OSMutex mutex; /*!< this mutex protects
the next fields */
os_cond_t cond_var; /*!< condition variable is
used in waiting for the event */
protected:
// Disable copying
os_event(const os_event&);
os_event& operator=(const os_event&);
};
/**
Do a timed wait on condition variable.
@param abstime - absolute time to wait
@param time_in_ms - timeout in milliseconds
@return true if timed out */
bool
os_event::timed_wait(
#ifndef _WIN32
const timespec* abstime
#else
DWORD time_in_ms
#endif /* !_WIN32 */
)
{
#ifdef _WIN32
BOOL ret;
ret = SleepConditionVariableCS(&cond_var, mutex, time_in_ms);
if (!ret) {
DWORD err = GetLastError();
/* FQDN=msdn.microsoft.com
@see http://$FQDN/en-us/library/ms686301%28VS.85%29.aspx,
"Condition variables are subject to spurious wakeups
(those not associated with an explicit wake) and stolen wakeups
(another thread manages to run before the woken thread)."
Check for both types of timeouts.
Conditions are checked by the caller.*/
if (err == WAIT_TIMEOUT || err == ERROR_TIMEOUT) {
return(true);
}
}
ut_a(ret);
return(false);
#else
int ret;
ret = pthread_cond_timedwait(&cond_var, mutex, abstime);
switch (ret) {
case 0:
case ETIMEDOUT:
/* We play it safe by checking for EINTR even though
according to the POSIX documentation it can't return EINTR. */
case EINTR:
break;
default:
ib::error() << "pthread_cond_timedwait() returned: " << ret
<< ": abstime={" << abstime->tv_sec << ","
<< abstime->tv_nsec << "}";
ut_error;
}
return(ret == ETIMEDOUT);
#endif /* _WIN32 */
}
/**
Waits for an event object until it is in the signaled state.
Typically, if the event has been signalled after the os_event_reset()
we'll return immediately because event->m_set == true.
There are, however, situations (e.g.: sync_array code) where we may
lose this information. For example:
thread A calls os_event_reset()
thread B calls os_event_set() [event->m_set == true]
thread C calls os_event_reset() [event->m_set == false]
thread A calls os_event_wait() [infinite wait!]
thread C calls os_event_wait() [infinite wait!]
Where such a scenario is possible, to avoid infinite wait, the
value returned by reset() should be passed in as
reset_sig_count. */
void
os_event::wait_low(
int64_t reset_sig_count) UNIV_NOTHROW
{
mutex.enter();
if (!reset_sig_count) {
reset_sig_count = signal_count;
}
while (!m_set && signal_count == reset_sig_count) {
wait();
/* Spurious wakeups may occur: we have to check if the
event really has been signaled after we came here to wait. */
}
mutex.exit();
}
/**
Waits for an event object until it is in the signaled state or
a timeout is exceeded.
@param time_in_usec - timeout in microseconds, or OS_SYNC_INFINITE_TIME
@param reset_sig_count - zero or the value returned by previous call
of os_event_reset().
@return 0 if success, OS_SYNC_TIME_EXCEEDED if timeout was exceeded */
ulint
os_event::wait_time_low(
ulint time_in_usec,
int64_t reset_sig_count) UNIV_NOTHROW
{
bool timed_out = false;
#ifdef _WIN32
DWORD time_in_ms;
if (time_in_usec != OS_SYNC_INFINITE_TIME) {
time_in_ms = DWORD(time_in_usec / 1000);
} else {
time_in_ms = INFINITE;
}
#else
struct timespec abstime;
if (time_in_usec != OS_SYNC_INFINITE_TIME) {
ulonglong usec = ulonglong(time_in_usec) + my_hrtime().val;
abstime.tv_sec = static_cast<time_t>(usec / 1000000);
abstime.tv_nsec = static_cast<uint>((usec % 1000000) * 1000);
} else {
abstime.tv_nsec = 999999999;
abstime.tv_sec = (time_t) ULINT_MAX;
}
ut_a(abstime.tv_nsec <= 999999999);
#endif /* _WIN32 */
mutex.enter();
if (!reset_sig_count) {
reset_sig_count = signal_count;
}
do {
if (m_set || signal_count != reset_sig_count) {
break;
}
#ifndef _WIN32
timed_out = timed_wait(&abstime);
#else
timed_out = timed_wait(time_in_ms);
#endif /* !_WIN32 */
} while (!timed_out);
mutex.exit();
return(timed_out ? OS_SYNC_TIME_EXCEEDED : 0);
}
/** Constructor */
os_event::os_event() UNIV_NOTHROW
{
init();
m_set = false;
/* We return this value in os_event_reset(),
which can then be be used to pass to the
os_event_wait_low(). The value of zero is
reserved in os_event_wait_low() for the case
when the caller does not want to pass any
signal_count value. To distinguish between
the two cases we initialize signal_count
to 1 here. */
signal_count = 1;
}
/** Destructor */
os_event::~os_event() UNIV_NOTHROW
{
destroy();
}
/**
Creates an event semaphore, i.e., a semaphore which may just have two
states: signaled and nonsignaled. The created event is manual reset: it
must be reset explicitly by calling sync_os_reset_event.
@return the event handle */
os_event_t os_event_create(const char*)
{
return(UT_NEW_NOKEY(os_event()));
}
/**
Check if the event is set.
@return true if set */
bool
os_event_is_set(
/*============*/
const os_event_t event) /*!< in: event to test */
{
return(event->is_set());
}
/**
Sets an event semaphore to the signaled state: lets waiting threads
proceed. */
void
os_event_set(
/*=========*/
os_event_t event) /*!< in/out: event to set */
{
event->set();
}
/**
Resets an event semaphore to the nonsignaled state. Waiting threads will
stop to wait for the event.
The return value should be passed to os_even_wait_low() if it is desired
that this thread should not wait in case of an intervening call to
os_event_set() between this os_event_reset() and the
os_event_wait_low() call. See comments for os_event_wait_low().
@return current signal_count. */
int64_t
os_event_reset(
/*===========*/
os_event_t event) /*!< in/out: event to reset */
{
return(event->reset());
}
/**
Waits for an event object until it is in the signaled state or
a timeout is exceeded.
@return 0 if success, OS_SYNC_TIME_EXCEEDED if timeout was exceeded */
ulint
os_event_wait_time_low(
/*===================*/
os_event_t event, /*!< in/out: event to wait */
ulint time_in_usec, /*!< in: timeout in
microseconds, or
OS_SYNC_INFINITE_TIME */
int64_t reset_sig_count) /*!< in: zero or the value
returned by previous call of
os_event_reset(). */
{
return(event->wait_time_low(time_in_usec, reset_sig_count));
}
/**
Waits for an event object until it is in the signaled state.
Where such a scenario is possible, to avoid infinite wait, the
value returned by os_event_reset() should be passed in as
reset_sig_count. */
void
os_event_wait_low(
/*==============*/
os_event_t event, /*!< in: event to wait */
int64_t reset_sig_count) /*!< in: zero or the value
returned by previous call of
os_event_reset(). */
{
event->wait_low(reset_sig_count);
}
/**
Frees an event object. */
void
os_event_destroy(
/*=============*/
os_event_t& event) /*!< in/own: event to free */
{
UT_DELETE(event);
event = NULL;
}
|