summaryrefslogtreecommitdiff
path: root/sys/kern/kern_rwlock.c
blob: 0a2bca5a86175e62d566d13de7cd05c40868ad2e (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
/*	$OpenBSD: kern_rwlock.c,v 1.12 2007/05/04 13:21:03 art Exp $	*/

/*
 * Copyright (c) 2002, 2003 Artur Grabowski <art@openbsd.org>
 * All rights reserved. 
 *
 * Redistribution and use in source and binary forms, with or without 
 * modification, are permitted provided that the following conditions 
 * are met: 
 *
 * 1. Redistributions of source code must retain the above copyright 
 *    notice, this list of conditions and the following disclaimer. 
 * 2. The name of the author may not be used to endorse or promote products
 *    derived from this software without specific prior written permission. 
 *
 * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES,
 * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY
 * AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL
 * THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
 * EXEMPLARY, OR CONSEQUENTIAL  DAMAGES (INCLUDING, BUT NOT LIMITED TO,
 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
 * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
 * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
 * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/rwlock.h>
#include <sys/limits.h>

#include <machine/lock.h>

/* XXX - temporary measure until proc0 is properly aligned */
#define RW_PROC(p) (((long)p) & ~RWLOCK_MASK)

/*
 * Magic wand for lock operations. Every operation checks if certain
 * flags are set and if they aren't, it increments the lock with some
 * value (that might need some computing in a few cases). If the operation
 * fails, we need to set certain flags while waiting for the lock.
 *
 * RW_WRITE	The lock must be completely empty. We increment it with
 *		RWLOCK_WRLOCK and the proc pointer of the holder.
 *		Sets RWLOCK_WAIT|RWLOCK_WRWANT while waiting.
 * RW_READ	RWLOCK_WRLOCK|RWLOCK_WRWANT may not be set. We increment
 *		with RWLOCK_READ_INCR. RWLOCK_WAIT while waiting.
 */
static const struct rwlock_op {
	unsigned long inc;
	unsigned long check;
	unsigned long wait_set;
	long proc_mult;
	int wait_prio;
} rw_ops[] = {
	{	/* RW_WRITE */
		RWLOCK_WRLOCK,
		ULONG_MAX,
		RWLOCK_WAIT | RWLOCK_WRWANT,
		1,
		PLOCK - 4
	},
	{	/* RW_READ */
		RWLOCK_READ_INCR,
		RWLOCK_WRLOCK,
		RWLOCK_WAIT,
		0,
		PLOCK
	},
	{	/* RW_DOWNGRADE */
		RWLOCK_READ_INCR - RWLOCK_WRLOCK,
		0,
		0,
		-1,
		PLOCK
	},
};

#ifndef __HAVE_MD_RWLOCK
/*
 * Simple cases that should be in MD code and atomic.
 */
void
rw_enter_read(struct rwlock *rwl)
{
	unsigned long owner = rwl->rwl_owner;

	if (__predict_false((owner & RWLOCK_WRLOCK) ||
	    rw_cas(&rwl->rwl_owner, owner, owner + RWLOCK_READ_INCR)))
		rw_enter(rwl, RW_READ);
}

void
rw_enter_write(struct rwlock *rwl)
{
	struct proc *p = curproc;

	if (__predict_false(rw_cas(&rwl->rwl_owner, 0,
	    RW_PROC(p) | RWLOCK_WRLOCK)))
		rw_enter(rwl, RW_WRITE);
}

void
rw_exit_read(struct rwlock *rwl)
{
	unsigned long owner = rwl->rwl_owner;

	if (__predict_false((owner & RWLOCK_WAIT) ||
	    rw_cas(&rwl->rwl_owner, owner, owner - RWLOCK_READ_INCR)))
		rw_exit(rwl);
}

void
rw_exit_write(struct rwlock *rwl)
{
	unsigned long owner = rwl->rwl_owner;

	if (__predict_false((owner & RWLOCK_WAIT) ||
	    rw_cas(&rwl->rwl_owner, owner, 0)))
		rw_exit(rwl);
}

#ifndef rw_cas
int
rw_cas(volatile unsigned long *p, unsigned long o, unsigned long n)
{
	if (*p != o)
		return (1);
	*p = n;

	return (0);
}
#endif

#endif

#ifdef DIAGNOSTIC
/*
 * Put the diagnostic functions here to keep the main code free
 * from ifdef clutter.
 */
static void
rw_enter_diag(struct rwlock *rwl, int flags)
{
	switch (flags & RW_OPMASK) {
	case RW_WRITE:
	case RW_READ:
		if (RW_PROC(curproc) == RW_PROC(rwl->rwl_owner))
			panic("rw_enter: locking against myself");
		break;
	case RW_DOWNGRADE:
		/*
		 * If we're downgrading, we must hold the write lock.
		 */
		if ((rwl->rwl_owner & RWLOCK_WRLOCK) == 0)
			panic("rw_enter: downgrade of non-write lock");
		if (RW_PROC(curproc) != RW_PROC(rwl->rwl_owner))
			panic("rw_enter: downgrade, not holder");
		break;

	default:
		panic("rw_enter: unknown op 0x%x", flags);
	}
}

#else
#define rw_enter_diag(r, f)
#endif

void
rw_init(struct rwlock *rwl, const char *name)
{
	rwl->rwl_owner = 0;
	rwl->rwl_name = name;
}

int
rw_enter(struct rwlock *rwl, int flags)
{
	const struct rwlock_op *op;
	struct sleep_state sls;
	unsigned long inc, o;
	int error;

	op = &rw_ops[flags & RW_OPMASK];

	inc = op->inc + RW_PROC(curproc) * op->proc_mult;
retry:
	while (__predict_false(((o = rwl->rwl_owner) & op->check) != 0)) {
		unsigned long set = o | op->wait_set;
		int do_sleep;

		rw_enter_diag(rwl, flags);

		if (flags & RW_NOSLEEP)
			return (EBUSY);

		sleep_setup(&sls, rwl, op->wait_prio, rwl->rwl_name);
		if (flags & RW_INTR)
			sleep_setup_signal(&sls, op->wait_prio | PCATCH);

		do_sleep = !rw_cas(&rwl->rwl_owner, o, set);

		sleep_finish(&sls, do_sleep);
		if ((flags & RW_INTR) &&
		    (error = sleep_finish_signal(&sls)) != 0)
			return (error);
		if (flags & RW_SLEEPFAIL)
			return (EAGAIN);
	}

	if (__predict_false(rw_cas(&rwl->rwl_owner, o, o + inc)))
		goto retry;

	/*
	 * If old lock had RWLOCK_WAIT and RWLOCK_WRLOCK set, it means we
	 * downgraded a write lock and had possible read waiter, wake them
	 * to let them retry the lock.
	 */
	if (__predict_false((o & (RWLOCK_WRLOCK|RWLOCK_WAIT)) ==
	    (RWLOCK_WRLOCK|RWLOCK_WAIT)))
		wakeup(rwl);

	return (0);
}

void
rw_exit(struct rwlock *rwl)
{
	unsigned long owner = rwl->rwl_owner;
	int wrlock = owner & RWLOCK_WRLOCK;
	unsigned long set;

	do {
		owner = rwl->rwl_owner;
		if (wrlock)
			set = 0;
		else
			set = (owner - RWLOCK_READ_INCR) &
				~(RWLOCK_WAIT|RWLOCK_WRWANT);
	} while (rw_cas(&rwl->rwl_owner, owner, set));

	if (owner & RWLOCK_WAIT)
		wakeup(rwl);
}