xref: /netbsd-src/sys/kern/kern_timeout.c (revision 21e37cc72a480a47828990a439cde7ac9ffaf0c6)
1 /*	$NetBSD: kern_timeout.c,v 1.13 2003/10/30 04:32:56 thorpej Exp $	*/
2 
3 /*-
4  * Copyright (c) 2003 The NetBSD Foundation, Inc.
5  * All rights reserved.
6  *
7  * This code is derived from software contributed to The NetBSD Foundation
8  * by Jason R. Thorpe.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  * 3. All advertising materials mentioning features or use of this software
19  *    must display the following acknowledgement:
20  *	This product includes software developed by the NetBSD
21  *	Foundation, Inc. and its contributors.
22  * 4. Neither the name of The NetBSD Foundation nor the names of its
23  *    contributors may be used to endorse or promote products derived
24  *    from this software without specific prior written permission.
25  *
26  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
27  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
28  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
29  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
30  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
31  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
32  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
33  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
34  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
35  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
36  * POSSIBILITY OF SUCH DAMAGE.
37  */
38 
39 /*
40  * Copyright (c) 2001 Thomas Nordin <nordin@openbsd.org>
41  * Copyright (c) 2000-2001 Artur Grabowski <art@openbsd.org>
42  * All rights reserved.
43  *
44  * Redistribution and use in source and binary forms, with or without
45  * modification, are permitted provided that the following conditions
46  * are met:
47  *
48  * 1. Redistributions of source code must retain the above copyright
49  *    notice, this list of conditions and the following disclaimer.
50  * 2. Redistributions in binary form must reproduce the above copyright
51  *    notice, this list of conditions and the following disclaimer in the
52  *    documentation and/or other materials provided with the distribution.
53  * 3. The name of the author may not be used to endorse or promote products
54  *    derived from this software without specific prior written permission.
55  *
56  * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES,
57  * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY
58  * AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL
59  * THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
60  * EXEMPLARY, OR CONSEQUENTIAL  DAMAGES (INCLUDING, BUT NOT LIMITED TO,
61  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
62  * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
63  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
64  * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
65  * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
66  */
67 
68 #include <sys/cdefs.h>
69 __KERNEL_RCSID(0, "$NetBSD: kern_timeout.c,v 1.13 2003/10/30 04:32:56 thorpej Exp $");
70 
71 /*
72  * Adapted from OpenBSD: kern_timeout.c,v 1.15 2002/12/08 04:21:07 art Exp,
73  * modified to match NetBSD's pre-existing callout API.
74  */
75 
76 #include <sys/param.h>
77 #include <sys/systm.h>
78 #include <sys/kernel.h>
79 #include <sys/lock.h>
80 #include <sys/callout.h>
81 
82 #ifdef DDB
83 #include <machine/db_machdep.h>
84 #include <ddb/db_interface.h>
85 #include <ddb/db_access.h>
86 #include <ddb/db_sym.h>
87 #include <ddb/db_output.h>
88 #endif
89 
90 /*
91  * Timeouts are kept in a hierarchical timing wheel. The c_time is the value
92  * of the global variable "hardclock_ticks" when the timeout should be called.
93  * There are four levels with 256 buckets each. See 'Scheme 7' in
94  * "Hashed and Hierarchical Timing Wheels: Efficient Data Structures for
95  * Implementing a Timer Facility" by George Varghese and Tony Lauck.
96  */
97 #define BUCKETS 1024
98 #define WHEELSIZE 256
99 #define WHEELMASK 255
100 #define WHEELBITS 8
101 
102 static struct callout_circq timeout_wheel[BUCKETS];	/* Queues of timeouts */
103 static struct callout_circq timeout_todo;		/* Worklist */
104 
105 #define MASKWHEEL(wheel, time) (((time) >> ((wheel)*WHEELBITS)) & WHEELMASK)
106 
107 #define BUCKET(rel, abs)						\
108     (((rel) <= (1 << (2*WHEELBITS)))					\
109     	? ((rel) <= (1 << WHEELBITS))					\
110             ? &timeout_wheel[MASKWHEEL(0, (abs))]			\
111             : &timeout_wheel[MASKWHEEL(1, (abs)) + WHEELSIZE]		\
112         : ((rel) <= (1 << (3*WHEELBITS)))				\
113             ? &timeout_wheel[MASKWHEEL(2, (abs)) + 2*WHEELSIZE]		\
114             : &timeout_wheel[MASKWHEEL(3, (abs)) + 3*WHEELSIZE])
115 
116 #define MOVEBUCKET(wheel, time)						\
117     CIRCQ_APPEND(&timeout_todo,						\
118         &timeout_wheel[MASKWHEEL((wheel), (time)) + (wheel)*WHEELSIZE])
119 
120 /*
121  * All wheels are locked with the same lock (which must also block out all
122  * interrupts).
123  */
124 static struct simplelock callout_slock;
125 
126 #define	CALLOUT_LOCK(s)							\
127 do {									\
128 	s = splsched();							\
129 	simple_lock(&callout_slock);					\
130 } while (/*CONSTCOND*/0)
131 
132 #define	CALLOUT_UNLOCK(s)						\
133 do {									\
134 	simple_unlock(&callout_slock);					\
135 	splx((s));							\
136 } while (/*CONSTCOND*/0)
137 
138 /*
139  * Circular queue definitions.
140  */
141 
142 #define CIRCQ_INIT(list)						\
143 do {									\
144         (list)->cq_next_l = (list);					\
145         (list)->cq_prev_l = (list);					\
146 } while (/*CONSTCOND*/0)
147 
148 #define CIRCQ_INSERT(elem, list)					\
149 do {									\
150         (elem)->cq_prev_e = (list)->cq_prev_e;				\
151         (elem)->cq_next_l = (list);					\
152         (list)->cq_prev_l->cq_next_l = (elem);				\
153         (list)->cq_prev_l = (elem);					\
154 } while (/*CONSTCOND*/0)
155 
156 #define CIRCQ_APPEND(fst, snd)						\
157 do {									\
158         if (!CIRCQ_EMPTY(snd)) {					\
159                 (fst)->cq_prev_l->cq_next_l = (snd)->cq_next_l;		\
160                 (snd)->cq_next_l->cq_prev_l = (fst)->cq_prev_l;		\
161                 (snd)->cq_prev_l->cq_next_l = (fst);			\
162                 (fst)->cq_prev_l = (snd)->cq_prev_l;			\
163                 CIRCQ_INIT(snd);					\
164         }								\
165 } while (/*CONSTCOND*/0)
166 
167 #define CIRCQ_REMOVE(elem)						\
168 do {									\
169         (elem)->cq_next_l->cq_prev_e = (elem)->cq_prev_e;		\
170         (elem)->cq_prev_l->cq_next_e = (elem)->cq_next_e;		\
171 } while (/*CONSTCOND*/0)
172 
173 #define CIRCQ_FIRST(list)	((list)->cq_next_e)
174 #define CIRCQ_NEXT(elem)	((elem)->cq_next_e)
175 #define CIRCQ_LAST(elem,list)	((elem)->cq_next_l == (list))
176 #define CIRCQ_EMPTY(list)	((list)->cq_next_l == (list))
177 
178 /*
179  * Some of the "math" in here is a bit tricky.
180  *
181  * We have to beware of wrapping ints.
182  * We use the fact that any element added to the queue must be added with a
183  * positive time. That means that any element `to' on the queue cannot be
184  * scheduled to timeout further in time than INT_MAX, but c->c_time can
185  * be positive or negative so comparing it with anything is dangerous.
186  * The only way we can use the c->c_time value in any predictable way
187  * is when we calculate how far in the future `to' will timeout -
188  * "c->c_time - hardclock_ticks". The result will always be positive for
189  * future timeouts and 0 or negative for due timeouts.
190  */
191 
192 #ifdef CALLOUT_EVENT_COUNTERS
193 static struct evcnt callout_ev_late;
194 #endif
195 
196 /*
197  * callout_startup:
198  *
199  *	Initialize the callout facility, called at system startup time.
200  */
201 void
202 callout_startup(void)
203 {
204 	int b;
205 
206 	CIRCQ_INIT(&timeout_todo);
207 	for (b = 0; b < BUCKETS; b++)
208 		CIRCQ_INIT(&timeout_wheel[b]);
209 	simple_lock_init(&callout_slock);
210 
211 #ifdef CALLOUT_EVENT_COUNTERS
212 	evcnt_attach_dynamic(&callout_ev_late, EVCNT_TYPE_MISC,
213 	    NULL, "callout", "late");
214 #endif
215 }
216 
217 /*
218  * callout_init:
219  *
220  *	Initialize a callout structure.
221  */
222 void
223 callout_init(struct callout *c)
224 {
225 
226 	memset(c, 0, sizeof(*c));
227 }
228 
229 /*
230  * callout_reset:
231  *
232  *	Reset a callout structure with a new function and argument, and
233  *	schedule it to run.
234  */
235 void
236 callout_reset(struct callout *c, int to_ticks, void (*func)(void *), void *arg)
237 {
238 	int s, old_time;
239 
240 	KASSERT(to_ticks >= 0);
241 
242 	CALLOUT_LOCK(s);
243 
244 	/* Initialize the time here, it won't change. */
245 	old_time = c->c_time;
246 	c->c_time = to_ticks + hardclock_ticks;
247 	c->c_flags &= ~(CALLOUT_FIRED|CALLOUT_INVOKING);
248 
249 	c->c_func = func;
250 	c->c_arg = arg;
251 
252 	/*
253 	 * If this timeout is already scheduled and now is moved
254 	 * earlier, reschedule it now. Otherwise leave it in place
255 	 * and let it be rescheduled later.
256 	 */
257 	if (callout_pending(c)) {
258 		if (c->c_time - old_time < 0) {
259 			CIRCQ_REMOVE(&c->c_list);
260 			CIRCQ_INSERT(&c->c_list, &timeout_todo);
261 		}
262 	} else {
263 		c->c_flags |= CALLOUT_PENDING;
264 		CIRCQ_INSERT(&c->c_list, &timeout_todo);
265 	}
266 
267 	CALLOUT_UNLOCK(s);
268 }
269 
270 /*
271  * callout_schedule:
272  *
273  *	Schedule a callout to run.  The function and argument must
274  *	already be set in the callout structure.
275  */
276 void
277 callout_schedule(struct callout *c, int to_ticks)
278 {
279 	int s, old_time;
280 
281 	KASSERT(to_ticks >= 0);
282 
283 	CALLOUT_LOCK(s);
284 
285 	/* Initialize the time here, it won't change. */
286 	old_time = c->c_time;
287 	c->c_time = to_ticks + hardclock_ticks;
288 	c->c_flags &= ~(CALLOUT_FIRED|CALLOUT_INVOKING);
289 
290 	/*
291 	 * If this timeout is already scheduled and now is moved
292 	 * earlier, reschedule it now. Otherwise leave it in place
293 	 * and let it be rescheduled later.
294 	 */
295 	if (callout_pending(c)) {
296 		if (c->c_time - old_time < 0) {
297 			CIRCQ_REMOVE(&c->c_list);
298 			CIRCQ_INSERT(&c->c_list, &timeout_todo);
299 		}
300 	} else {
301 		c->c_flags |= CALLOUT_PENDING;
302 		CIRCQ_INSERT(&c->c_list, &timeout_todo);
303 	}
304 
305 	CALLOUT_UNLOCK(s);
306 }
307 
308 /*
309  * callout_stop:
310  *
311  *	Cancel a pending callout.
312  */
313 void
314 callout_stop(struct callout *c)
315 {
316 	int s;
317 
318 	CALLOUT_LOCK(s);
319 
320 	if (callout_pending(c))
321 		CIRCQ_REMOVE(&c->c_list);
322 
323 	c->c_flags &= ~(CALLOUT_PENDING|CALLOUT_FIRED);
324 
325 	CALLOUT_UNLOCK(s);
326 }
327 
328 /*
329  * This is called from hardclock() once every tick.
330  * We return !0 if we need to schedule a softclock.
331  */
332 int
333 callout_hardclock(void)
334 {
335 	int s;
336 	int needsoftclock;
337 
338 	CALLOUT_LOCK(s);
339 
340 	MOVEBUCKET(0, hardclock_ticks);
341 	if (MASKWHEEL(0, hardclock_ticks) == 0) {
342 		MOVEBUCKET(1, hardclock_ticks);
343 		if (MASKWHEEL(1, hardclock_ticks) == 0) {
344 			MOVEBUCKET(2, hardclock_ticks);
345 			if (MASKWHEEL(2, hardclock_ticks) == 0)
346 				MOVEBUCKET(3, hardclock_ticks);
347 		}
348 	}
349 
350 	needsoftclock = !CIRCQ_EMPTY(&timeout_todo);
351 	CALLOUT_UNLOCK(s);
352 
353 	return needsoftclock;
354 }
355 
356 /* ARGSUSED */
357 void
358 softclock(void *v)
359 {
360 	struct callout *c;
361 	void (*func)(void *);
362 	void *arg;
363 	int s;
364 
365 	CALLOUT_LOCK(s);
366 
367 	while (!CIRCQ_EMPTY(&timeout_todo)) {
368 		c = CIRCQ_FIRST(&timeout_todo);
369 		CIRCQ_REMOVE(&c->c_list);
370 
371 		/* If due run it, otherwise insert it into the right bucket. */
372 		if (c->c_time - hardclock_ticks > 0) {
373 			CIRCQ_INSERT(&c->c_list,
374 			    BUCKET((c->c_time - hardclock_ticks), c->c_time));
375 		} else {
376 #ifdef CALLOUT_EVENT_COUNTERS
377 			if (c->c_time - hardclock_ticks < 0)
378 				callout_ev_late.ev_count++;
379 #endif
380 			c->c_flags = (c->c_flags  & ~CALLOUT_PENDING) |
381 			    (CALLOUT_FIRED|CALLOUT_INVOKING);
382 
383 			func = c->c_func;
384 			arg = c->c_arg;
385 
386 			CALLOUT_UNLOCK(s);
387 			(*func)(arg);
388 			CALLOUT_LOCK(s);
389 		}
390 	}
391 
392 	CALLOUT_UNLOCK(s);
393 }
394 
395 #ifdef DDB
396 static void
397 db_show_callout_bucket(struct callout_circq *bucket)
398 {
399 	struct callout *c;
400 	db_expr_t offset;
401 	char *name;
402 
403 	if (CIRCQ_EMPTY(bucket))
404 		return;
405 
406 	for (c = CIRCQ_FIRST(bucket); /*nothing*/; c = CIRCQ_NEXT(&c->c_list)) {
407 		db_find_sym_and_offset((db_addr_t)(intptr_t)c->c_func, &name,
408 		    &offset);
409 		name = name ? name : "?";
410 #ifdef _LP64
411 #define	POINTER_WIDTH	"%16lx"
412 #else
413 #define	POINTER_WIDTH	"%8lx"
414 #endif
415 		db_printf("%9d %2d/%-4d " POINTER_WIDTH "  %s\n",
416 		    c->c_time - hardclock_ticks,
417 		    (int)((bucket - timeout_wheel) / WHEELSIZE),
418 		    (int)(bucket - timeout_wheel), (u_long) c->c_arg, name);
419 
420 		if (CIRCQ_LAST(&c->c_list, bucket))
421 			break;
422 	}
423 }
424 
425 void
426 db_show_callout(db_expr_t addr, int haddr, db_expr_t count, char *modif)
427 {
428 	int b;
429 
430 	db_printf("hardclock_ticks now: %d\n", hardclock_ticks);
431 #ifdef _LP64
432 	db_printf("    ticks  wheel               arg  func\n");
433 #else
434 	db_printf("    ticks  wheel       arg  func\n");
435 #endif
436 
437 	/*
438 	 * Don't lock the callwheel; all the other CPUs are paused
439 	 * anyhow, and we might be called in a circumstance where
440 	 * some other CPU was paused while holding the lock.
441 	 */
442 
443 	db_show_callout_bucket(&timeout_todo);
444 	for (b = 0; b < BUCKETS; b++)
445 		db_show_callout_bucket(&timeout_wheel[b]);
446 }
447 #endif /* DDB */
448