aboutsummaryrefslogtreecommitdiffstats
path: root/module/spl/spl-condvar.c
blob: 4778fb256449f03800848b6871d4c68456014290 (plain)
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
/*
 *  Copyright (C) 2007-2010 Lawrence Livermore National Security, LLC.
 *  Copyright (C) 2007 The Regents of the University of California.
 *  Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
 *  Written by Brian Behlendorf <behlendorf1@llnl.gov>.
 *  UCRL-CODE-235197
 *
 *  This file is part of the SPL, Solaris Porting Layer.
 *  For details, see <http://zfsonlinux.org/>.
 *
 *  The SPL 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; either version 2 of the License, or (at your
 *  option) any later version.
 *
 *  The SPL 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 the SPL.  If not, see <http://www.gnu.org/licenses/>.
 *
 *  Solaris Porting Layer (SPL) Credential Implementation.
 */

#include <sys/condvar.h>
#include <sys/time.h>
#include <linux/hrtimer.h>

void
__cv_init(kcondvar_t *cvp, char *name, kcv_type_t type, void *arg)
{
	ASSERT(cvp);
	ASSERT(name == NULL);
	ASSERT(type == CV_DEFAULT);
	ASSERT(arg == NULL);

	cvp->cv_magic = CV_MAGIC;
	init_waitqueue_head(&cvp->cv_event);
	init_waitqueue_head(&cvp->cv_destroy);
	atomic_set(&cvp->cv_waiters, 0);
	atomic_set(&cvp->cv_refs, 1);
	cvp->cv_mutex = NULL;
}
EXPORT_SYMBOL(__cv_init);

static int
cv_destroy_wakeup(kcondvar_t *cvp)
{
	if (!atomic_read(&cvp->cv_waiters) && !atomic_read(&cvp->cv_refs)) {
		ASSERT(cvp->cv_mutex == NULL);
		ASSERT(!waitqueue_active(&cvp->cv_event));
		return (1);
	}

	return (0);
}

void
__cv_destroy(kcondvar_t *cvp)
{
	ASSERT(cvp);
	ASSERT(cvp->cv_magic == CV_MAGIC);

	cvp->cv_magic = CV_DESTROY;
	atomic_dec(&cvp->cv_refs);

	/* Block until all waiters are woken and references dropped. */
	while (cv_destroy_wakeup(cvp) == 0)
		wait_event_timeout(cvp->cv_destroy, cv_destroy_wakeup(cvp), 1);

	ASSERT3P(cvp->cv_mutex, ==, NULL);
	ASSERT3S(atomic_read(&cvp->cv_refs), ==, 0);
	ASSERT3S(atomic_read(&cvp->cv_waiters), ==, 0);
	ASSERT3S(waitqueue_active(&cvp->cv_event), ==, 0);
}
EXPORT_SYMBOL(__cv_destroy);

static void
cv_wait_common(kcondvar_t *cvp, kmutex_t *mp, int state, int io)
{
	DEFINE_WAIT(wait);
	kmutex_t *m;

	ASSERT(cvp);
	ASSERT(mp);
	ASSERT(cvp->cv_magic == CV_MAGIC);
	ASSERT(mutex_owned(mp));
	atomic_inc(&cvp->cv_refs);

	m = ACCESS_ONCE(cvp->cv_mutex);
	if (!m)
		m = xchg(&cvp->cv_mutex, mp);
	/* Ensure the same mutex is used by all callers */
	ASSERT(m == NULL || m == mp);

	prepare_to_wait_exclusive(&cvp->cv_event, &wait, state);
	atomic_inc(&cvp->cv_waiters);

	/*
	 * Mutex should be dropped after prepare_to_wait() this
	 * ensures we're linked in to the waiters list and avoids the
	 * race where 'cvp->cv_waiters > 0' but the list is empty.
	 */
	mutex_exit(mp);
	if (io)
		io_schedule();
	else
		schedule();

	/* No more waiters a different mutex could be used */
	if (atomic_dec_and_test(&cvp->cv_waiters)) {
		/*
		 * This is set without any lock, so it's racy. But this is
		 * just for debug anyway, so make it best-effort
		 */
		cvp->cv_mutex = NULL;
		wake_up(&cvp->cv_destroy);
	}

	finish_wait(&cvp->cv_event, &wait);
	atomic_dec(&cvp->cv_refs);

	/*
	 * Hold mutex after we release the cvp, otherwise we could dead lock
	 * with a thread holding the mutex and call cv_destroy.
	 */
	mutex_enter(mp);
}

void
__cv_wait(kcondvar_t *cvp, kmutex_t *mp)
{
	cv_wait_common(cvp, mp, TASK_UNINTERRUPTIBLE, 0);
}
EXPORT_SYMBOL(__cv_wait);

void
__cv_wait_io(kcondvar_t *cvp, kmutex_t *mp)
{
	cv_wait_common(cvp, mp, TASK_UNINTERRUPTIBLE, 1);
}
EXPORT_SYMBOL(__cv_wait_io);

void
__cv_wait_sig(kcondvar_t *cvp, kmutex_t *mp)
{
	cv_wait_common(cvp, mp, TASK_INTERRUPTIBLE, 0);
}
EXPORT_SYMBOL(__cv_wait_sig);

#if defined(HAVE_IO_SCHEDULE_TIMEOUT)
#define	spl_io_schedule_timeout(t)	io_schedule_timeout(t)
#else
static void
__cv_wakeup(unsigned long data)
{
	wake_up_process((struct task_struct *)data);
}

static long
spl_io_schedule_timeout(long time_left)
{
	long expire_time = jiffies + time_left;
	struct timer_list timer;

	init_timer(&timer);
	setup_timer(&timer, __cv_wakeup, (unsigned long)current);
	timer.expires = expire_time;
	add_timer(&timer);

	io_schedule();

	del_timer_sync(&timer);
	time_left = expire_time - jiffies;

	return (time_left < 0 ? 0 : time_left);
}
#endif

/*
 * 'expire_time' argument is an absolute wall clock time in jiffies.
 * Return value is time left (expire_time - now) or -1 if timeout occurred.
 */
static clock_t
__cv_timedwait_common(kcondvar_t *cvp, kmutex_t *mp, clock_t expire_time,
    int state, int io)
{
	DEFINE_WAIT(wait);
	kmutex_t *m;
	clock_t time_left;

	ASSERT(cvp);
	ASSERT(mp);
	ASSERT(cvp->cv_magic == CV_MAGIC);
	ASSERT(mutex_owned(mp));

	/* XXX - Does not handle jiffie wrap properly */
	time_left = expire_time - jiffies;
	if (time_left <= 0)
		return (-1);

	atomic_inc(&cvp->cv_refs);
	m = ACCESS_ONCE(cvp->cv_mutex);
	if (!m)
		m = xchg(&cvp->cv_mutex, mp);
	/* Ensure the same mutex is used by all callers */
	ASSERT(m == NULL || m == mp);

	prepare_to_wait_exclusive(&cvp->cv_event, &wait, state);
	atomic_inc(&cvp->cv_waiters);

	/*
	 * Mutex should be dropped after prepare_to_wait() this
	 * ensures we're linked in to the waiters list and avoids the
	 * race where 'cvp->cv_waiters > 0' but the list is empty.
	 */
	mutex_exit(mp);
	if (io)
		time_left = spl_io_schedule_timeout(time_left);
	else
		time_left = schedule_timeout(time_left);

	/* No more waiters a different mutex could be used */
	if (atomic_dec_and_test(&cvp->cv_waiters)) {
		/*
		 * This is set without any lock, so it's racy. But this is
		 * just for debug anyway, so make it best-effort
		 */
		cvp->cv_mutex = NULL;
		wake_up(&cvp->cv_destroy);
	}

	finish_wait(&cvp->cv_event, &wait);
	atomic_dec(&cvp->cv_refs);

	/*
	 * Hold mutex after we release the cvp, otherwise we could dead lock
	 * with a thread holding the mutex and call cv_destroy.
	 */
	mutex_enter(mp);
	return (time_left > 0 ? time_left : -1);
}

clock_t
__cv_timedwait(kcondvar_t *cvp, kmutex_t *mp, clock_t exp_time)
{
	return (__cv_timedwait_common(cvp, mp, exp_time,
	    TASK_UNINTERRUPTIBLE, 0));
}
EXPORT_SYMBOL(__cv_timedwait);

clock_t
__cv_timedwait_io(kcondvar_t *cvp, kmutex_t *mp, clock_t exp_time)
{
	return (__cv_timedwait_common(cvp, mp, exp_time,
	    TASK_UNINTERRUPTIBLE, 1));
}
EXPORT_SYMBOL(__cv_timedwait_io);

clock_t
__cv_timedwait_sig(kcondvar_t *cvp, kmutex_t *mp, clock_t exp_time)
{
	return (__cv_timedwait_common(cvp, mp, exp_time,
	    TASK_INTERRUPTIBLE, 0));
}
EXPORT_SYMBOL(__cv_timedwait_sig);

/*
 * 'expire_time' argument is an absolute clock time in nanoseconds.
 * Return value is time left (expire_time - now) or -1 if timeout occurred.
 */
static clock_t
__cv_timedwait_hires(kcondvar_t *cvp, kmutex_t *mp, hrtime_t expire_time,
    int state)
{
	DEFINE_WAIT(wait);
	kmutex_t *m;
	hrtime_t time_left;
	ktime_t ktime_left;

	ASSERT(cvp);
	ASSERT(mp);
	ASSERT(cvp->cv_magic == CV_MAGIC);
	ASSERT(mutex_owned(mp));

	time_left = expire_time - gethrtime();
	if (time_left <= 0)
		return (-1);

	atomic_inc(&cvp->cv_refs);
	m = ACCESS_ONCE(cvp->cv_mutex);
	if (!m)
		m = xchg(&cvp->cv_mutex, mp);
	/* Ensure the same mutex is used by all callers */
	ASSERT(m == NULL || m == mp);

	prepare_to_wait_exclusive(&cvp->cv_event, &wait, state);
	atomic_inc(&cvp->cv_waiters);

	/*
	 * Mutex should be dropped after prepare_to_wait() this
	 * ensures we're linked in to the waiters list and avoids the
	 * race where 'cvp->cv_waiters > 0' but the list is empty.
	 */
	mutex_exit(mp);
	/*
	 * Allow a 100 us range to give kernel an opportunity to coalesce
	 * interrupts
	 */
	ktime_left = ktime_set(0, time_left);
	schedule_hrtimeout_range(&ktime_left, 100 * NSEC_PER_USEC,
	    HRTIMER_MODE_REL);

	/* No more waiters a different mutex could be used */
	if (atomic_dec_and_test(&cvp->cv_waiters)) {
		/*
		 * This is set without any lock, so it's racy. But this is
		 * just for debug anyway, so make it best-effort
		 */
		cvp->cv_mutex = NULL;
		wake_up(&cvp->cv_destroy);
	}

	finish_wait(&cvp->cv_event, &wait);
	atomic_dec(&cvp->cv_refs);

	mutex_enter(mp);
	time_left = expire_time - gethrtime();
	return (time_left > 0 ? NSEC_TO_TICK(time_left) : -1);
}

/*
 * Compatibility wrapper for the cv_timedwait_hires() Illumos interface.
 */
static clock_t
cv_timedwait_hires_common(kcondvar_t *cvp, kmutex_t *mp, hrtime_t tim, hrtime_t res,
    int flag, int state)
{
	if (res > 1) {
		/*
		 * Align expiration to the specified resolution.
		 */
		if (flag & CALLOUT_FLAG_ROUNDUP)
			tim += res - 1;
		tim = (tim / res) * res;
	}

	if (!(flag & CALLOUT_FLAG_ABSOLUTE))
		tim += gethrtime();

	return (__cv_timedwait_hires(cvp, mp, tim, state));
}

clock_t
cv_timedwait_hires(kcondvar_t *cvp, kmutex_t *mp, hrtime_t tim, hrtime_t res,
    int flag)
{
	return (cv_timedwait_hires_common(cvp, mp, tim, res, flag,
	    TASK_UNINTERRUPTIBLE));
}
EXPORT_SYMBOL(cv_timedwait_hires);

clock_t
cv_timedwait_sig_hires(kcondvar_t *cvp, kmutex_t *mp, hrtime_t tim, hrtime_t res,
    int flag)
{
	return (cv_timedwait_hires_common(cvp, mp, tim, res, flag,
	    TASK_INTERRUPTIBLE));
}
EXPORT_SYMBOL(cv_timedwait_sig_hires);

void
__cv_signal(kcondvar_t *cvp)
{
	ASSERT(cvp);
	ASSERT(cvp->cv_magic == CV_MAGIC);
	atomic_inc(&cvp->cv_refs);

	/*
	 * All waiters are added with WQ_FLAG_EXCLUSIVE so only one
	 * waiter will be set runable with each call to wake_up().
	 * Additionally wake_up() holds a spin_lock assoicated with
	 * the wait queue to ensure we don't race waking up processes.
	 */
	if (atomic_read(&cvp->cv_waiters) > 0)
		wake_up(&cvp->cv_event);

	atomic_dec(&cvp->cv_refs);
}
EXPORT_SYMBOL(__cv_signal);

void
__cv_broadcast(kcondvar_t *cvp)
{
	ASSERT(cvp);
	ASSERT(cvp->cv_magic == CV_MAGIC);
	atomic_inc(&cvp->cv_refs);

	/*
	 * Wake_up_all() will wake up all waiters even those which
	 * have the WQ_FLAG_EXCLUSIVE flag set.
	 */
	if (atomic_read(&cvp->cv_waiters) > 0)
		wake_up_all(&cvp->cv_event);

	atomic_dec(&cvp->cv_refs);
}
EXPORT_SYMBOL(__cv_broadcast);