summaryrefslogtreecommitdiff
path: root/sys/kern/kern_clockintr.c
blob: 6351a0bb255e2d523d578b06458bea8331b424fb (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
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
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
/* $OpenBSD: kern_clockintr.c,v 1.31 2023/08/11 22:02:50 cheloha Exp $ */
/*
 * Copyright (c) 2003 Dale Rahn <drahn@openbsd.org>
 * Copyright (c) 2020 Mark Kettenis <kettenis@openbsd.org>
 * Copyright (c) 2020-2022 Scott Cheloha <cheloha@openbsd.org>
 *
 * Permission to use, copy, modify, and distribute this software for any
 * purpose with or without fee is hereby granted, provided that the above
 * copyright notice and this permission notice appear in all copies.
 *
 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/atomic.h>
#include <sys/clockintr.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/mutex.h>
#include <sys/resourcevar.h>
#include <sys/queue.h>
#include <sys/sched.h>
#include <sys/stdint.h>
#include <sys/sysctl.h>
#include <sys/time.h>

/*
 * Protection for global variables in this file:
 *
 *	I	Immutable after initialization.
 */
u_int clockintr_flags;			/* [I] global state + behavior flags */
uint32_t hardclock_period;		/* [I] hardclock period (ns) */
uint32_t schedclock_period;		/* [I] schedclock period (ns) */
uint32_t statclock_avg;			/* [I] average statclock period (ns) */
uint32_t statclock_min;			/* [I] minimum statclock period (ns) */
uint32_t statclock_mask;		/* [I] set of allowed offsets */

void clockintr_cancel_locked(struct clockintr *);
uint64_t clockintr_expiration(const struct clockintr *);
void clockintr_hardclock(struct clockintr *, void *);
uint64_t clockintr_nsecuptime(const struct clockintr *);
void clockintr_schedclock(struct clockintr *, void *);
void clockintr_schedule(struct clockintr *, uint64_t);
void clockintr_schedule_locked(struct clockintr *, uint64_t);
void clockintr_statclock(struct clockintr *, void *);
uint64_t clockqueue_next(const struct clockintr_queue *);
void clockqueue_reset_intrclock(struct clockintr_queue *);
uint64_t nsec_advance(uint64_t *, uint64_t, uint64_t);

/*
 * Initialize global state.  Set flags and compute intervals.
 */
void
clockintr_init(u_int flags)
{
	uint32_t half_avg, var;

	KASSERT(CPU_IS_PRIMARY(curcpu()));
	KASSERT(clockintr_flags == 0);
	KASSERT(!ISSET(flags, ~CL_FLAG_MASK));

	KASSERT(hz > 0 && hz <= 1000000000);
	hardclock_period = 1000000000 / hz;
	roundrobin_period = hardclock_period * 10;

	KASSERT(stathz >= 1 && stathz <= 1000000000);

	/*
	 * Compute the average statclock() period.  Then find var, the
	 * largest power of two such that var <= statclock_avg / 2.
	 */
	statclock_avg = 1000000000 / stathz;
	half_avg = statclock_avg / 2;
	for (var = 1U << 31; var > half_avg; var /= 2)
		continue;

	/*
	 * Set a lower bound for the range using statclock_avg and var.
	 * The mask for that range is just (var - 1).
	 */
	statclock_min = statclock_avg - (var / 2);
	statclock_mask = var - 1;

	KASSERT(schedhz >= 0 && schedhz <= 1000000000);
	if (schedhz != 0)
		schedclock_period = 1000000000 / schedhz;

	SET(clockintr_flags, flags | CL_INIT);
}

/*
 * Ready the calling CPU for clockintr_dispatch().  If this is our
 * first time here, install the intrclock, if any, and set necessary
 * flags.  Advance the schedule as needed.
 */
void
clockintr_cpu_init(const struct intrclock *ic)
{
	uint64_t multiplier = 0;
	struct cpu_info *ci = curcpu();
	struct clockintr_queue *cq = &ci->ci_queue;
	struct schedstate_percpu *spc = &ci->ci_schedstate;
	int reset_cq_intrclock = 0;

	KASSERT(ISSET(clockintr_flags, CL_INIT));

	if (ic != NULL && !ISSET(cq->cq_flags, CQ_INTRCLOCK)) {
		cq->cq_intrclock = *ic;
		SET(cq->cq_flags, CQ_INTRCLOCK);
	}

	/* TODO: Remove these from struct clockintr_queue. */
	if (cq->cq_hardclock == NULL) {
		cq->cq_hardclock = clockintr_establish(cq, clockintr_hardclock);
		if (cq->cq_hardclock == NULL)
			panic("%s: failed to establish hardclock", __func__);
	}
	if (cq->cq_statclock == NULL) {
		cq->cq_statclock = clockintr_establish(cq, clockintr_statclock);
		if (cq->cq_statclock == NULL)
			panic("%s: failed to establish statclock", __func__);
	}
	if (schedhz != 0 && cq->cq_schedclock == NULL) {
		cq->cq_schedclock = clockintr_establish(cq,
		    clockintr_schedclock);
		if (cq->cq_schedclock == NULL)
			panic("%s: failed to establish schedclock", __func__);
	}

	/*
	 * Mask CQ_INTRCLOCK while we're advancing the internal clock
	 * interrupts.  We don't want the intrclock to fire until this
	 * thread reaches clockintr_trigger().
	 */
	if (ISSET(cq->cq_flags, CQ_INTRCLOCK)) {
		CLR(cq->cq_flags, CQ_INTRCLOCK);
		reset_cq_intrclock = 1;
	}

	/*
	 * Until we understand scheduler lock contention better, stagger
	 * the hardclock and statclock so they don't all happen at once.
	 * If we have no intrclock it doesn't matter, we have no control
	 * anyway.  The primary CPU's starting offset is always zero, so
	 * leave the multiplier zero.
	 */
	if (!CPU_IS_PRIMARY(ci) && reset_cq_intrclock)
		multiplier = CPU_INFO_UNIT(ci);

	/*
	 * The first time we do this, the primary CPU cannot skip any
	 * hardclocks.  We can skip hardclocks on subsequent calls because
	 * the global tick value is advanced during inittodr(9) on our
	 * behalf.
	 */
	if (CPU_IS_PRIMARY(ci)) {
		if (cq->cq_hardclock->cl_expiration == 0)
			clockintr_schedule(cq->cq_hardclock, 0);
		else
			clockintr_advance(cq->cq_hardclock, hardclock_period);
	} else {
		if (cq->cq_hardclock->cl_expiration == 0) {
			clockintr_stagger(cq->cq_hardclock, hardclock_period,
			     multiplier, MAXCPUS);
		}
		clockintr_advance(cq->cq_hardclock, hardclock_period);
	}

	/*
	 * We can always advance the statclock and schedclock.
	 * There is no reason to stagger a randomized statclock.
	 */
	if (!ISSET(clockintr_flags, CL_RNDSTAT)) {
		if (cq->cq_statclock->cl_expiration == 0) {
			clockintr_stagger(cq->cq_statclock, statclock_avg,
			    multiplier, MAXCPUS);
		}
	}
	clockintr_advance(cq->cq_statclock, statclock_avg);
	if (schedhz != 0) {
		if (cq->cq_schedclock->cl_expiration == 0) {
			clockintr_stagger(cq->cq_schedclock, schedclock_period,
			    multiplier, MAXCPUS);
		}
		clockintr_advance(cq->cq_schedclock, schedclock_period);
	}

	/*
	 * XXX Need to find a better place to do this.  We can't do it in
	 * sched_init_cpu() because initclocks() runs after it.
	 */
	if (spc->spc_itimer->cl_expiration == 0) {
		clockintr_stagger(spc->spc_itimer, hardclock_period,
		    multiplier, MAXCPUS);
	}
	if (spc->spc_profclock->cl_expiration == 0) {
		clockintr_stagger(spc->spc_profclock, profclock_period,
		    multiplier, MAXCPUS);
	}
	if (spc->spc_roundrobin->cl_expiration == 0) {
		clockintr_stagger(spc->spc_roundrobin, hardclock_period,
		    multiplier, MAXCPUS);
	}
	clockintr_advance(spc->spc_roundrobin, roundrobin_period);

	if (reset_cq_intrclock)
		SET(cq->cq_flags, CQ_INTRCLOCK);
}

/*
 * If we have an intrclock, trigger it to start the dispatch cycle.
 */
void
clockintr_trigger(void)
{
	struct clockintr_queue *cq = &curcpu()->ci_queue;

	KASSERT(ISSET(cq->cq_flags, CQ_INIT));

	if (ISSET(cq->cq_flags, CQ_INTRCLOCK))
		intrclock_trigger(&cq->cq_intrclock);
}

/*
 * Run all expired events scheduled on the calling CPU.
 */
int
clockintr_dispatch(void *frame)
{
	uint64_t lateness, run = 0, start;
	struct cpu_info *ci = curcpu();
	struct clockintr *cl;
	struct clockintr_queue *cq = &ci->ci_queue;
	u_int ogen;

	if (cq->cq_dispatch != 0)
		panic("%s: recursive dispatch", __func__);
	cq->cq_dispatch = 1;

	splassert(IPL_CLOCK);
	KASSERT(ISSET(cq->cq_flags, CQ_INIT));

	mtx_enter(&cq->cq_mtx);

	/*
	 * If nothing is scheduled or we arrived too early, we have
	 * nothing to do.
	 */
	start = nsecuptime();
	cq->cq_uptime = start;
	if (TAILQ_EMPTY(&cq->cq_pend))
		goto stats;
	if (cq->cq_uptime < clockqueue_next(cq))
		goto rearm;
	lateness = start - clockqueue_next(cq);

	/*
	 * Dispatch expired events.
	 */
	for (;;) {
		cl = TAILQ_FIRST(&cq->cq_pend);
		if (cl == NULL)
			break;
		if (cq->cq_uptime < cl->cl_expiration) {
			/* Double-check the time before giving up. */
			cq->cq_uptime = nsecuptime();
			if (cq->cq_uptime < cl->cl_expiration)
				break;
		}
		clockintr_cancel_locked(cl);
		cq->cq_shadow.cl_expiration = cl->cl_expiration;
		cq->cq_running = cl;
		mtx_leave(&cq->cq_mtx);

		cl->cl_func(&cq->cq_shadow, frame);

		mtx_enter(&cq->cq_mtx);
		cq->cq_running = NULL;
		if (ISSET(cl->cl_flags, CLST_IGNORE_SHADOW)) {
			CLR(cl->cl_flags, CLST_IGNORE_SHADOW);
			CLR(cq->cq_shadow.cl_flags, CLST_SHADOW_PENDING);
		}
		if (ISSET(cq->cq_shadow.cl_flags, CLST_SHADOW_PENDING)) {
			CLR(cq->cq_shadow.cl_flags, CLST_SHADOW_PENDING);
			clockintr_schedule_locked(cl,
			    cq->cq_shadow.cl_expiration);
		}
		run++;
	}

	/*
	 * Dispatch complete.
	 */
rearm:
	/* Rearm the interrupt clock if we have one. */
	if (ISSET(cq->cq_flags, CQ_INTRCLOCK)) {
		if (!TAILQ_EMPTY(&cq->cq_pend)) {
			intrclock_rearm(&cq->cq_intrclock,
			    clockqueue_next(cq) - cq->cq_uptime);
		}
	}
stats:
	/* Update our stats. */
	ogen = cq->cq_gen;
	cq->cq_gen = 0;
	membar_producer();
	cq->cq_stat.cs_dispatched += cq->cq_uptime - start;
	if (run > 0) {
		cq->cq_stat.cs_lateness += lateness;
		cq->cq_stat.cs_prompt++;
		cq->cq_stat.cs_run += run;
	} else if (!TAILQ_EMPTY(&cq->cq_pend)) {
		cq->cq_stat.cs_early++;
		cq->cq_stat.cs_earliness += clockqueue_next(cq) - cq->cq_uptime;
	} else
		cq->cq_stat.cs_spurious++;
	membar_producer();
	cq->cq_gen = MAX(1, ogen + 1);

	mtx_leave(&cq->cq_mtx);

	if (cq->cq_dispatch != 1)
		panic("%s: unexpected value: %u", __func__, cq->cq_dispatch);
	cq->cq_dispatch = 0;

	return run > 0;
}

uint64_t
clockintr_advance(struct clockintr *cl, uint64_t period)
{
	uint64_t count, expiration;
	struct clockintr_queue *cq = cl->cl_queue;

	if (cl == &cq->cq_shadow) {
		count = nsec_advance(&cl->cl_expiration, period, cq->cq_uptime);
		SET(cl->cl_flags, CLST_SHADOW_PENDING);
		return count;
	}

	mtx_enter(&cq->cq_mtx);
	expiration = cl->cl_expiration;
	count = nsec_advance(&expiration, period, nsecuptime());
	if (ISSET(cl->cl_flags, CLST_PENDING))
		clockintr_cancel_locked(cl);
	clockintr_schedule_locked(cl, expiration);
	if (ISSET(cq->cq_flags, CQ_INTRCLOCK)) {
		if (cl == TAILQ_FIRST(&cq->cq_pend)) {
			if (cq == &curcpu()->ci_queue)
				clockqueue_reset_intrclock(cq);
		}
	}
	if (cl == cq->cq_running)
		SET(cl->cl_flags, CLST_IGNORE_SHADOW);
	mtx_leave(&cq->cq_mtx);
	return count;
}

void
clockintr_cancel(struct clockintr *cl)
{
	struct clockintr_queue *cq = cl->cl_queue;
	int was_next;

	if (cl == &cq->cq_shadow) {
		CLR(cl->cl_flags, CLST_SHADOW_PENDING);
		return;
	}

	mtx_enter(&cq->cq_mtx);
	if (ISSET(cl->cl_flags, CLST_PENDING)) {
		was_next = cl == TAILQ_FIRST(&cq->cq_pend);
		clockintr_cancel_locked(cl);
		if (ISSET(cq->cq_flags, CQ_INTRCLOCK)) {
			if (was_next && !TAILQ_EMPTY(&cq->cq_pend)) {
				if (cq == &curcpu()->ci_queue)
					clockqueue_reset_intrclock(cq);
			}
		}
	}
	if (cl == cq->cq_running)
		SET(cl->cl_flags, CLST_IGNORE_SHADOW);
	mtx_leave(&cq->cq_mtx);
}

void
clockintr_cancel_locked(struct clockintr *cl)
{
	struct clockintr_queue *cq = cl->cl_queue;

	MUTEX_ASSERT_LOCKED(&cq->cq_mtx);
	KASSERT(ISSET(cl->cl_flags, CLST_PENDING));

	TAILQ_REMOVE(&cq->cq_pend, cl, cl_plink);
	CLR(cl->cl_flags, CLST_PENDING);
}

struct clockintr *
clockintr_establish(struct clockintr_queue *cq,
    void (*func)(struct clockintr *, void *))
{
	struct clockintr *cl;

	cl = malloc(sizeof *cl, M_DEVBUF, M_NOWAIT | M_ZERO);
	if (cl == NULL)
		return NULL;
	cl->cl_func = func;
	cl->cl_queue = cq;

	mtx_enter(&cq->cq_mtx);
	TAILQ_INSERT_TAIL(&cq->cq_est, cl, cl_elink);
	mtx_leave(&cq->cq_mtx);
	return cl;
}

uint64_t
clockintr_expiration(const struct clockintr *cl)
{
	uint64_t expiration;
	struct clockintr_queue *cq = cl->cl_queue;

	if (cl == &cq->cq_shadow)
		return cl->cl_expiration;

	mtx_enter(&cq->cq_mtx);
	expiration = cl->cl_expiration;
	mtx_leave(&cq->cq_mtx);
	return expiration;
}

void
clockintr_schedule(struct clockintr *cl, uint64_t expiration)
{
	struct clockintr_queue *cq = cl->cl_queue;

	if (cl == &cq->cq_shadow) {
		cl->cl_expiration = expiration;
		SET(cl->cl_flags, CLST_SHADOW_PENDING);
		return;
	}

	mtx_enter(&cq->cq_mtx);
	if (ISSET(cl->cl_flags, CLST_PENDING))
		clockintr_cancel_locked(cl);
	clockintr_schedule_locked(cl, expiration);
	if (ISSET(cq->cq_flags, CQ_INTRCLOCK)) {
		if (cl == TAILQ_FIRST(&cq->cq_pend)) {
			if (cq == &curcpu()->ci_queue)
				clockqueue_reset_intrclock(cq);
		}
	}
	if (cl == cq->cq_running)
		SET(cl->cl_flags, CLST_IGNORE_SHADOW);
	mtx_leave(&cq->cq_mtx);
}

void
clockintr_schedule_locked(struct clockintr *cl, uint64_t expiration)
{
	struct clockintr *elm;
	struct clockintr_queue *cq = cl->cl_queue;

	MUTEX_ASSERT_LOCKED(&cq->cq_mtx);
	KASSERT(!ISSET(cl->cl_flags, CLST_PENDING));

	cl->cl_expiration = expiration;
	TAILQ_FOREACH(elm, &cq->cq_pend, cl_plink) {
		if (cl->cl_expiration < elm->cl_expiration)
			break;
	}
	if (elm == NULL)
		TAILQ_INSERT_TAIL(&cq->cq_pend, cl, cl_plink);
	else
		TAILQ_INSERT_BEFORE(elm, cl, cl_plink);
	SET(cl->cl_flags, CLST_PENDING);
}

void
clockintr_stagger(struct clockintr *cl, uint64_t period, u_int n, u_int count)
{
	struct clockintr_queue *cq = cl->cl_queue;

	KASSERT(n < count);

	mtx_enter(&cq->cq_mtx);
	if (ISSET(cl->cl_flags, CLST_PENDING))
		panic("%s: clock interrupt pending", __func__);
	cl->cl_expiration = period / count * n;
	mtx_leave(&cq->cq_mtx);
}

uint64_t
clockintr_nsecuptime(const struct clockintr *cl)
{
	KASSERT(cl == &cl->cl_queue->cq_shadow);
	return cl->cl_queue->cq_uptime;
}

void
clockintr_hardclock(struct clockintr *cl, void *frame)
{
	uint64_t count, i;

	count = clockintr_advance(cl, hardclock_period);
	for (i = 0; i < count; i++)
		hardclock(frame);
}

void
clockintr_schedclock(struct clockintr *cl, void *unused)
{
	uint64_t count, i;
	struct proc *p = curproc;

	count = clockintr_advance(cl, schedclock_period);
	if (p != NULL) {
		for (i = 0; i < count; i++)
			schedclock(p);
	}
}

void
clockintr_statclock(struct clockintr *cl, void *frame)
{
	uint64_t count, expiration, i, uptime;
	uint32_t off;

	if (ISSET(clockintr_flags, CL_RNDSTAT)) {
		count = 0;
		expiration = clockintr_expiration(cl);
		uptime = clockintr_nsecuptime(cl);
		while (expiration <= uptime) {
			while ((off = (random() & statclock_mask)) == 0)
				continue;
			expiration += statclock_min + off;
			count++;
		}
		clockintr_schedule(cl, expiration);
	} else {
		count = clockintr_advance(cl, statclock_avg);
	}
	for (i = 0; i < count; i++)
		statclock(frame);
}

void
clockqueue_init(struct clockintr_queue *cq)
{
	if (ISSET(cq->cq_flags, CQ_INIT))
		return;

	cq->cq_shadow.cl_queue = cq;
	mtx_init(&cq->cq_mtx, IPL_CLOCK);
	TAILQ_INIT(&cq->cq_est);
	TAILQ_INIT(&cq->cq_pend);
	cq->cq_gen = 1;
	SET(cq->cq_flags, CQ_INIT);
}

uint64_t
clockqueue_next(const struct clockintr_queue *cq)
{
	MUTEX_ASSERT_LOCKED(&cq->cq_mtx);
	return TAILQ_FIRST(&cq->cq_pend)->cl_expiration;
}

void
clockqueue_reset_intrclock(struct clockintr_queue *cq)
{
	uint64_t exp, now;

	MUTEX_ASSERT_LOCKED(&cq->cq_mtx);
	KASSERT(ISSET(cq->cq_flags, CQ_INTRCLOCK));

	exp = clockqueue_next(cq);
	now = nsecuptime();
	if (now < exp)
		intrclock_rearm(&cq->cq_intrclock, exp - now);
	else
		intrclock_trigger(&cq->cq_intrclock);
}

/*
 * Advance *next in increments of period until it exceeds now.
 * Returns the number of increments *next was advanced.
 *
 * We check the common cases first to avoid division if possible.
 * This does no overflow checking.
 */
uint64_t
nsec_advance(uint64_t *next, uint64_t period, uint64_t now)
{
	uint64_t elapsed;

	if (now < *next)
		return 0;

	if (now < *next + period) {
		*next += period;
		return 1;
	}

	elapsed = (now - *next) / period + 1;
	*next += period * elapsed;
	return elapsed;
}

int
sysctl_clockintr(int *name, u_int namelen, void *oldp, size_t *oldlenp,
    void *newp, size_t newlen)
{
	struct clockintr_stat sum, tmp;
	struct clockintr_queue *cq;
	struct cpu_info *ci;
	CPU_INFO_ITERATOR cii;
	u_int gen;

	if (namelen != 1)
		return ENOTDIR;

	switch (name[0]) {
	case KERN_CLOCKINTR_STATS:
		memset(&sum, 0, sizeof sum);
		CPU_INFO_FOREACH(cii, ci) {
			cq = &ci->ci_queue;
			if (!ISSET(cq->cq_flags, CQ_INIT))
				continue;
			do {
				gen = cq->cq_gen;
				membar_consumer();
				tmp = cq->cq_stat;
				membar_consumer();
			} while (gen == 0 || gen != cq->cq_gen);
			sum.cs_dispatched += tmp.cs_dispatched;
			sum.cs_early += tmp.cs_early;
			sum.cs_earliness += tmp.cs_earliness;
			sum.cs_lateness += tmp.cs_lateness;
			sum.cs_prompt += tmp.cs_prompt;
			sum.cs_run += tmp.cs_run;
			sum.cs_spurious += tmp.cs_spurious;
		}
		return sysctl_rdstruct(oldp, oldlenp, newp, &sum, sizeof sum);
	default:
		break;
	}

	return EINVAL;
}

#ifdef DDB

#include <machine/db_machdep.h>

#include <ddb/db_interface.h>
#include <ddb/db_output.h>
#include <ddb/db_sym.h>

void db_show_clockintr(const struct clockintr *, const char *, u_int);
void db_show_clockintr_cpu(struct cpu_info *);

void
db_show_all_clockintr(db_expr_t addr, int haddr, db_expr_t count, char *modif)
{
	struct timespec now;
	struct cpu_info *ci;
	CPU_INFO_ITERATOR cii;

	nanouptime(&now);
	db_printf("%20s\n", "UPTIME");
	db_printf("%10lld.%09ld\n", now.tv_sec, now.tv_nsec);
	db_printf("\n");
	db_printf("%20s  %5s  %3s  %s\n", "EXPIRATION", "STATE", "CPU", "NAME");
	CPU_INFO_FOREACH(cii, ci) {
		if (ISSET(ci->ci_queue.cq_flags, CQ_INIT))
			db_show_clockintr_cpu(ci);
	}
}

void
db_show_clockintr_cpu(struct cpu_info *ci)
{
	struct clockintr *elm;
	struct clockintr_queue *cq = &ci->ci_queue;
	u_int cpu = CPU_INFO_UNIT(ci);

	if (cq->cq_running != NULL)
		db_show_clockintr(cq->cq_running, "run", cpu);
	TAILQ_FOREACH(elm, &cq->cq_pend, cl_plink)
		db_show_clockintr(elm, "pend", cpu);
	TAILQ_FOREACH(elm, &cq->cq_est, cl_elink) {
		if (!ISSET(elm->cl_flags, CLST_PENDING))
			db_show_clockintr(elm, "idle", cpu);
	}
}

void
db_show_clockintr(const struct clockintr *cl, const char *state, u_int cpu)
{
	struct timespec ts;
	char *name;
	db_expr_t offset;

	NSEC_TO_TIMESPEC(cl->cl_expiration, &ts);
	db_find_sym_and_offset((vaddr_t)cl->cl_func, &name, &offset);
	if (name == NULL)
		name = "?";
	db_printf("%10lld.%09ld  %5s  %3u  %s\n",
	    ts.tv_sec, ts.tv_nsec, state, cpu, name);
}

#endif /* DDB */