summaryrefslogtreecommitdiffstats
path: root/sys/kern/kern_clocksource.c
blob: ae64c616c1f665160e9a406d3225fd12f011add1 (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
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
/*-
 * Copyright (c) 2010-2013 Alexander Motin <mav@FreeBSD.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,
 *    without modification, immediately at the beginning of the file.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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/cdefs.h>
__FBSDID("$FreeBSD$");

/*
 * Common routines to manage event timers hardware.
 */

#include "opt_device_polling.h"

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/bus.h>
#include <sys/limits.h>
#include <sys/lock.h>
#include <sys/kdb.h>
#include <sys/ktr.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/kernel.h>
#include <sys/sched.h>
#include <sys/smp.h>
#include <sys/sysctl.h>
#include <sys/timeet.h>
#include <sys/timetc.h>

#include <machine/atomic.h>
#include <machine/clock.h>
#include <machine/cpu.h>
#include <machine/smp.h>

#ifdef KDTRACE_HOOKS
#include <sys/dtrace_bsd.h>
cyclic_clock_func_t	cyclic_clock_func = NULL;
#endif

int			cpu_can_deep_sleep = 0;	/* C3 state is available. */
int			cpu_disable_deep_sleep = 0; /* Timer dies in C3. */

static void		setuptimer(void);
static void		loadtimer(sbintime_t now, int first);
static int		doconfigtimer(void);
static void		configtimer(int start);
static int		round_freq(struct eventtimer *et, int freq);

static sbintime_t	getnextcpuevent(int idle);
static sbintime_t	getnextevent(void);
static int		handleevents(sbintime_t now, int fake);

static struct mtx	et_hw_mtx;

#define	ET_HW_LOCK(state)						\
	{								\
		if (timer->et_flags & ET_FLAGS_PERCPU)			\
			mtx_lock_spin(&(state)->et_hw_mtx);		\
		else							\
			mtx_lock_spin(&et_hw_mtx);			\
	}

#define	ET_HW_UNLOCK(state)						\
	{								\
		if (timer->et_flags & ET_FLAGS_PERCPU)			\
			mtx_unlock_spin(&(state)->et_hw_mtx);		\
		else							\
			mtx_unlock_spin(&et_hw_mtx);			\
	}

static struct eventtimer *timer = NULL;
static sbintime_t	timerperiod;	/* Timer period for periodic mode. */
static sbintime_t	statperiod;	/* statclock() events period. */
static sbintime_t	profperiod;	/* profclock() events period. */
static sbintime_t	nexttick;	/* Next global timer tick time. */
static u_int		busy = 1;	/* Reconfiguration is in progress. */
static int		profiling = 0;	/* Profiling events enabled. */

static char		timername[32];	/* Wanted timer. */
TUNABLE_STR("kern.eventtimer.timer", timername, sizeof(timername));

static int		singlemul = 0;	/* Multiplier for periodic mode. */
TUNABLE_INT("kern.eventtimer.singlemul", &singlemul);
SYSCTL_INT(_kern_eventtimer, OID_AUTO, singlemul, CTLFLAG_RW, &singlemul,
    0, "Multiplier for periodic mode");

static u_int		idletick = 0;	/* Run periodic events when idle. */
TUNABLE_INT("kern.eventtimer.idletick", &idletick);
SYSCTL_UINT(_kern_eventtimer, OID_AUTO, idletick, CTLFLAG_RW, &idletick,
    0, "Run periodic events when idle");

static int		periodic = 0;	/* Periodic or one-shot mode. */
static int		want_periodic = 0; /* What mode to prefer. */
TUNABLE_INT("kern.eventtimer.periodic", &want_periodic);

struct pcpu_state {
	struct mtx	et_hw_mtx;	/* Per-CPU timer mutex. */
	u_int		action;		/* Reconfiguration requests. */
	u_int		handle;		/* Immediate handle resuests. */
	sbintime_t	now;		/* Last tick time. */
	sbintime_t	nextevent;	/* Next scheduled event on this CPU. */
	sbintime_t	nexttick;	/* Next timer tick time. */
	sbintime_t	nexthard;	/* Next hardlock() event. */
	sbintime_t	nextstat;	/* Next statclock() event. */
	sbintime_t	nextprof;	/* Next profclock() event. */
	sbintime_t	nextcall;	/* Next callout event. */
	sbintime_t	nextcallopt;	/* Next optional callout event. */
#ifdef KDTRACE_HOOKS
	sbintime_t	nextcyc;	/* Next OpenSolaris cyclics event. */
#endif
	int		ipi;		/* This CPU needs IPI. */
	int		idle;		/* This CPU is in idle mode. */
};

static DPCPU_DEFINE(struct pcpu_state, timerstate);
DPCPU_DEFINE(sbintime_t, hardclocktime);

/*
 * Timer broadcast IPI handler.
 */
int
hardclockintr(void)
{
	sbintime_t now;
	struct pcpu_state *state;
	int done;

	if (doconfigtimer() || busy)
		return (FILTER_HANDLED);
	state = DPCPU_PTR(timerstate);
	now = state->now;
	CTR3(KTR_SPARE2, "ipi  at %d:    now  %d.%08x",
	    curcpu, (int)(now >> 32), (u_int)(now & 0xffffffff));
	done = handleevents(now, 0);
	return (done ? FILTER_HANDLED : FILTER_STRAY);
}

/*
 * Handle all events for specified time on this CPU
 */
static int
handleevents(sbintime_t now, int fake)
{
	sbintime_t t, *hct;
	struct trapframe *frame;
	struct pcpu_state *state;
	int usermode;
	int done, runs;

	CTR3(KTR_SPARE2, "handle at %d:  now  %d.%08x",
	    curcpu, (int)(now >> 32), (u_int)(now & 0xffffffff));
	done = 0;
	if (fake) {
		frame = NULL;
		usermode = 0;
	} else {
		frame = curthread->td_intr_frame;
		usermode = TRAPF_USERMODE(frame);
	}

	state = DPCPU_PTR(timerstate);

	runs = 0;
	while (now >= state->nexthard) {
		state->nexthard += tick_sbt;
		runs++;
	}
	if (runs) {
		hct = DPCPU_PTR(hardclocktime);
		*hct = state->nexthard - tick_sbt;
		if (fake < 2) {
			hardclock_cnt(runs, usermode);
			done = 1;
		}
	}
	runs = 0;
	while (now >= state->nextstat) {
		state->nextstat += statperiod;
		runs++;
	}
	if (runs && fake < 2) {
		statclock_cnt(runs, usermode);
		done = 1;
	}
	if (profiling) {
		runs = 0;
		while (now >= state->nextprof) {
			state->nextprof += profperiod;
			runs++;
		}
		if (runs && !fake) {
			profclock_cnt(runs, usermode, TRAPF_PC(frame));
			done = 1;
		}
	} else
		state->nextprof = state->nextstat;
	if (now >= state->nextcallopt) {
		state->nextcall = state->nextcallopt = SBT_MAX;
		callout_process(now);
	}

#ifdef KDTRACE_HOOKS
	if (fake == 0 && now >= state->nextcyc && cyclic_clock_func != NULL) {
		state->nextcyc = SBT_MAX;
		(*cyclic_clock_func)(frame);
	}
#endif

	t = getnextcpuevent(0);
	ET_HW_LOCK(state);
	if (!busy) {
		state->idle = 0;
		state->nextevent = t;
		loadtimer(now, (fake == 2) &&
		    (timer->et_flags & ET_FLAGS_PERCPU));
	}
	ET_HW_UNLOCK(state);
	return (done);
}

/*
 * Schedule binuptime of the next event on current CPU.
 */
static sbintime_t
getnextcpuevent(int idle)
{
	sbintime_t event;
	struct pcpu_state *state;
	u_int hardfreq;

	state = DPCPU_PTR(timerstate);
	/* Handle hardclock() events, skipping some if CPU is idle. */
	event = state->nexthard;
	if (idle) {
		hardfreq = (u_int)hz / 2;
		if (tc_min_ticktock_freq > 2
#ifdef SMP
		    && curcpu == CPU_FIRST()
#endif
		    )
			hardfreq = hz / tc_min_ticktock_freq;
		if (hardfreq > 1)
			event += tick_sbt * (hardfreq - 1);
	}
	/* Handle callout events. */
	if (event > state->nextcall)
		event = state->nextcall;
	if (!idle) { /* If CPU is active - handle other types of events. */
		if (event > state->nextstat)
			event = state->nextstat;
		if (profiling && event > state->nextprof)
			event = state->nextprof;
	}
#ifdef KDTRACE_HOOKS
	if (event > state->nextcyc)
		event = state->nextcyc;
#endif
	return (event);
}

/*
 * Schedule binuptime of the next event on all CPUs.
 */
static sbintime_t
getnextevent(void)
{
	struct pcpu_state *state;
	sbintime_t event;
#ifdef SMP
	int	cpu;
#endif
	int	c;

	state = DPCPU_PTR(timerstate);
	event = state->nextevent;
	c = -1;
#ifdef SMP
	if ((timer->et_flags & ET_FLAGS_PERCPU) == 0) {
		CPU_FOREACH(cpu) {
			state = DPCPU_ID_PTR(cpu, timerstate);
			if (event > state->nextevent) {
				event = state->nextevent;
				c = cpu;
			}
		}
	}
#endif
	CTR4(KTR_SPARE2, "next at %d:    next %d.%08x by %d",
	    curcpu, (int)(event >> 32), (u_int)(event & 0xffffffff), c);
	return (event);
}

/* Hardware timer callback function. */
static void
timercb(struct eventtimer *et, void *arg)
{
	sbintime_t now;
	sbintime_t *next;
	struct pcpu_state *state;
#ifdef SMP
	int cpu, bcast;
#endif

	/* Do not touch anything if somebody reconfiguring timers. */
	if (busy)
		return;
	/* Update present and next tick times. */
	state = DPCPU_PTR(timerstate);
	if (et->et_flags & ET_FLAGS_PERCPU) {
		next = &state->nexttick;
	} else
		next = &nexttick;
	now = sbinuptime();
	if (periodic)
		*next = now + timerperiod;
	else
		*next = -1;	/* Next tick is not scheduled yet. */
	state->now = now;
	CTR3(KTR_SPARE2, "intr at %d:    now  %d.%08x",
	    curcpu, (int)(now >> 32), (u_int)(now & 0xffffffff));

#ifdef SMP
	/* Prepare broadcasting to other CPUs for non-per-CPU timers. */
	bcast = 0;
	if ((et->et_flags & ET_FLAGS_PERCPU) == 0 && smp_started) {
		CPU_FOREACH(cpu) {
			state = DPCPU_ID_PTR(cpu, timerstate);
			ET_HW_LOCK(state);
			state->now = now;
			if (now >= state->nextevent) {
				state->nextevent += SBT_1S;
				if (curcpu != cpu) {
					state->ipi = 1;
					bcast = 1;
				}
			}
			ET_HW_UNLOCK(state);
		}
	}
#endif

	/* Handle events for this time on this CPU. */
	handleevents(now, 0);

#ifdef SMP
	/* Broadcast interrupt to other CPUs for non-per-CPU timers. */
	if (bcast) {
		CPU_FOREACH(cpu) {
			if (curcpu == cpu)
				continue;
			state = DPCPU_ID_PTR(cpu, timerstate);
			if (state->ipi) {
				state->ipi = 0;
				ipi_cpu(cpu, IPI_HARDCLOCK);
			}
		}
	}
#endif
}

/*
 * Load new value into hardware timer.
 */
static void
loadtimer(sbintime_t now, int start)
{
	struct pcpu_state *state;
	sbintime_t new;
	sbintime_t *next;
	uint64_t tmp;
	int eq;

	if (timer->et_flags & ET_FLAGS_PERCPU) {
		state = DPCPU_PTR(timerstate);
		next = &state->nexttick;
	} else
		next = &nexttick;
	if (periodic) {
		if (start) {
			/*
			 * Try to start all periodic timers aligned
			 * to period to make events synchronous.
			 */
			tmp = now % timerperiod;
			new = timerperiod - tmp;
			if (new < tmp)		/* Left less then passed. */
				new += timerperiod;
			CTR5(KTR_SPARE2, "load p at %d:   now %d.%08x first in %d.%08x",
			    curcpu, (int)(now >> 32), (u_int)(now & 0xffffffff),
			    (int)(new >> 32), (u_int)(new & 0xffffffff));
			*next = new + now;
			et_start(timer, new, timerperiod);
		}
	} else {
		new = getnextevent();
		eq = (new == *next);
		CTR4(KTR_SPARE2, "load at %d:    next %d.%08x eq %d",
		    curcpu, (int)(new >> 32), (u_int)(new & 0xffffffff), eq);
		if (!eq) {
			*next = new;
			et_start(timer, new - now, 0);
		}
	}
}

/*
 * Prepare event timer parameters after configuration changes.
 */
static void
setuptimer(void)
{
	int freq;

	if (periodic && (timer->et_flags & ET_FLAGS_PERIODIC) == 0)
		periodic = 0;
	else if (!periodic && (timer->et_flags & ET_FLAGS_ONESHOT) == 0)
		periodic = 1;
	singlemul = MIN(MAX(singlemul, 1), 20);
	freq = hz * singlemul;
	while (freq < (profiling ? profhz : stathz))
		freq += hz;
	freq = round_freq(timer, freq);
	timerperiod = SBT_1S / freq;
}

/*
 * Reconfigure specified per-CPU timer on other CPU. Called from IPI handler.
 */
static int
doconfigtimer(void)
{
	sbintime_t now;
	struct pcpu_state *state;

	state = DPCPU_PTR(timerstate);
	switch (atomic_load_acq_int(&state->action)) {
	case 1:
		now = sbinuptime();
		ET_HW_LOCK(state);
		loadtimer(now, 1);
		ET_HW_UNLOCK(state);
		state->handle = 0;
		atomic_store_rel_int(&state->action, 0);
		return (1);
	case 2:
		ET_HW_LOCK(state);
		et_stop(timer);
		ET_HW_UNLOCK(state);
		state->handle = 0;
		atomic_store_rel_int(&state->action, 0);
		return (1);
	}
	if (atomic_readandclear_int(&state->handle) && !busy) {
		now = sbinuptime();
		handleevents(now, 0);
		return (1);
	}
	return (0);
}

/*
 * Reconfigure specified timer.
 * For per-CPU timers use IPI to make other CPUs to reconfigure.
 */
static void
configtimer(int start)
{
	sbintime_t now, next;
	struct pcpu_state *state;
	int cpu;

	if (start) {
		setuptimer();
		now = sbinuptime();
	} else
		now = 0;
	critical_enter();
	ET_HW_LOCK(DPCPU_PTR(timerstate));
	if (start) {
		/* Initialize time machine parameters. */
		next = now + timerperiod;
		if (periodic)
			nexttick = next;
		else
			nexttick = -1;
		CPU_FOREACH(cpu) {
			state = DPCPU_ID_PTR(cpu, timerstate);
			state->now = now;
			if (!smp_started && cpu != CPU_FIRST())
				state->nextevent = SBT_MAX;
			else
				state->nextevent = next;
			if (periodic)
				state->nexttick = next;
			else
				state->nexttick = -1;
			state->nexthard = next;
			state->nextstat = next;
			state->nextprof = next;
			state->nextcall = next;
			state->nextcallopt = next;
			hardclock_sync(cpu);
		}
		busy = 0;
		/* Start global timer or per-CPU timer of this CPU. */
		loadtimer(now, 1);
	} else {
		busy = 1;
		/* Stop global timer or per-CPU timer of this CPU. */
		et_stop(timer);
	}
	ET_HW_UNLOCK(DPCPU_PTR(timerstate));
#ifdef SMP
	/* If timer is global or there is no other CPUs yet - we are done. */
	if ((timer->et_flags & ET_FLAGS_PERCPU) == 0 || !smp_started) {
		critical_exit();
		return;
	}
	/* Set reconfigure flags for other CPUs. */
	CPU_FOREACH(cpu) {
		state = DPCPU_ID_PTR(cpu, timerstate);
		atomic_store_rel_int(&state->action,
		    (cpu == curcpu) ? 0 : ( start ? 1 : 2));
	}
	/* Broadcast reconfigure IPI. */
	ipi_all_but_self(IPI_HARDCLOCK);
	/* Wait for reconfiguration completed. */
restart:
	cpu_spinwait();
	CPU_FOREACH(cpu) {
		if (cpu == curcpu)
			continue;
		state = DPCPU_ID_PTR(cpu, timerstate);
		if (atomic_load_acq_int(&state->action))
			goto restart;
	}
#endif
	critical_exit();
}

/*
 * Calculate nearest frequency supported by hardware timer.
 */
static int
round_freq(struct eventtimer *et, int freq)
{
	uint64_t div;

	if (et->et_frequency != 0) {
		div = lmax((et->et_frequency + freq / 2) / freq, 1);
		if (et->et_flags & ET_FLAGS_POW2DIV)
			div = 1 << (flsl(div + div / 2) - 1);
		freq = (et->et_frequency + div / 2) / div;
	}
	if (et->et_min_period > SBT_1S)
		panic("Event timer \"%s\" doesn't support sub-second periods!",
		    et->et_name);
	else if (et->et_min_period != 0)
		freq = min(freq, SBT2FREQ(et->et_min_period));
	if (et->et_max_period < SBT_1S && et->et_max_period != 0)
		freq = max(freq, SBT2FREQ(et->et_max_period));
	return (freq);
}

/*
 * Configure and start event timers (BSP part).
 */
void
cpu_initclocks_bsp(void)
{
	struct pcpu_state *state;
	int base, div, cpu;

	mtx_init(&et_hw_mtx, "et_hw_mtx", NULL, MTX_SPIN);
	CPU_FOREACH(cpu) {
		state = DPCPU_ID_PTR(cpu, timerstate);
		mtx_init(&state->et_hw_mtx, "et_hw_mtx", NULL, MTX_SPIN);
#ifdef KDTRACE_HOOKS
		state->nextcyc = SBT_MAX;
#endif
		state->nextcall = SBT_MAX;
		state->nextcallopt = SBT_MAX;
	}
	periodic = want_periodic;
	/* Grab requested timer or the best of present. */
	if (timername[0])
		timer = et_find(timername, 0, 0);
	if (timer == NULL && periodic) {
		timer = et_find(NULL,
		    ET_FLAGS_PERIODIC, ET_FLAGS_PERIODIC);
	}
	if (timer == NULL) {
		timer = et_find(NULL,
		    ET_FLAGS_ONESHOT, ET_FLAGS_ONESHOT);
	}
	if (timer == NULL && !periodic) {
		timer = et_find(NULL,
		    ET_FLAGS_PERIODIC, ET_FLAGS_PERIODIC);
	}
	if (timer == NULL)
		panic("No usable event timer found!");
	et_init(timer, timercb, NULL, NULL);

	/* Adapt to timer capabilities. */
	if (periodic && (timer->et_flags & ET_FLAGS_PERIODIC) == 0)
		periodic = 0;
	else if (!periodic && (timer->et_flags & ET_FLAGS_ONESHOT) == 0)
		periodic = 1;
	if (timer->et_flags & ET_FLAGS_C3STOP)
		cpu_disable_deep_sleep++;

	/*
	 * We honor the requested 'hz' value.
	 * We want to run stathz in the neighborhood of 128hz.
	 * We would like profhz to run as often as possible.
	 */
	if (singlemul <= 0 || singlemul > 20) {
		if (hz >= 1500 || (hz % 128) == 0)
			singlemul = 1;
		else if (hz >= 750)
			singlemul = 2;
		else
			singlemul = 4;
	}
	if (periodic) {
		base = round_freq(timer, hz * singlemul);
		singlemul = max((base + hz / 2) / hz, 1);
		hz = (base + singlemul / 2) / singlemul;
		if (base <= 128)
			stathz = base;
		else {
			div = base / 128;
			if (div >= singlemul && (div % singlemul) == 0)
				div++;
			stathz = base / div;
		}
		profhz = stathz;
		while ((profhz + stathz) <= 128 * 64)
			profhz += stathz;
		profhz = round_freq(timer, profhz);
	} else {
		hz = round_freq(timer, hz);
		stathz = round_freq(timer, 127);
		profhz = round_freq(timer, stathz * 64);
	}
	tick = 1000000 / hz;
	tick_sbt = SBT_1S / hz;
	tick_bt = sbttobt(tick_sbt);
	statperiod = SBT_1S / stathz;
	profperiod = SBT_1S / profhz;
	ET_LOCK();
	configtimer(1);
	ET_UNLOCK();
}

/*
 * Start per-CPU event timers on APs.
 */
void
cpu_initclocks_ap(void)
{
	sbintime_t now;
	struct pcpu_state *state;
	struct thread *td;

	state = DPCPU_PTR(timerstate);
	now = sbinuptime();
	ET_HW_LOCK(state);
	state->now = now;
	hardclock_sync(curcpu);
	spinlock_enter();
	ET_HW_UNLOCK(state);
	td = curthread;
	td->td_intr_nesting_level++;
	handleevents(state->now, 2);
	td->td_intr_nesting_level--;
	spinlock_exit();
}

/*
 * Switch to profiling clock rates.
 */
void
cpu_startprofclock(void)
{

	ET_LOCK();
	if (profiling == 0) {
		if (periodic) {
			configtimer(0);
			profiling = 1;
			configtimer(1);
		} else
			profiling = 1;
	} else
		profiling++;
	ET_UNLOCK();
}

/*
 * Switch to regular clock rates.
 */
void
cpu_stopprofclock(void)
{

	ET_LOCK();
	if (profiling == 1) {
		if (periodic) {
			configtimer(0);
			profiling = 0;
			configtimer(1);
		} else
		profiling = 0;
	} else
		profiling--;
	ET_UNLOCK();
}

/*
 * Switch to idle mode (all ticks handled).
 */
sbintime_t
cpu_idleclock(void)
{
	sbintime_t now, t;
	struct pcpu_state *state;

	if (idletick || busy ||
	    (periodic && (timer->et_flags & ET_FLAGS_PERCPU))
#ifdef DEVICE_POLLING
	    || curcpu == CPU_FIRST()
#endif
	    )
		return (-1);
	state = DPCPU_PTR(timerstate);
	if (periodic)
		now = state->now;
	else
		now = sbinuptime();
	CTR3(KTR_SPARE2, "idle at %d:    now  %d.%08x",
	    curcpu, (int)(now >> 32), (u_int)(now & 0xffffffff));
	t = getnextcpuevent(1);
	ET_HW_LOCK(state);
	state->idle = 1;
	state->nextevent = t;
	if (!periodic)
		loadtimer(now, 0);
	ET_HW_UNLOCK(state);
	return (MAX(t - now, 0));
}

/*
 * Switch to active mode (skip empty ticks).
 */
void
cpu_activeclock(void)
{
	sbintime_t now;
	struct pcpu_state *state;
	struct thread *td;

	state = DPCPU_PTR(timerstate);
	if (state->idle == 0 || busy)
		return;
	if (periodic)
		now = state->now;
	else
		now = sbinuptime();
	CTR3(KTR_SPARE2, "active at %d:  now  %d.%08x",
	    curcpu, (int)(now >> 32), (u_int)(now & 0xffffffff));
	spinlock_enter();
	td = curthread;
	td->td_intr_nesting_level++;
	handleevents(now, 1);
	td->td_intr_nesting_level--;
	spinlock_exit();
}

/*
 * Change the frequency of the given timer.  This changes et->et_frequency and
 * if et is the active timer it reconfigures the timer on all CPUs.  This is
 * intended to be a private interface for the use of et_change_frequency() only.
 */
void
cpu_et_frequency(struct eventtimer *et, uint64_t newfreq)
{

	ET_LOCK();
	if (et == timer) {
		configtimer(0);
		et->et_frequency = newfreq;
		configtimer(1);
	} else
		et->et_frequency = newfreq;
	ET_UNLOCK();
}

#ifdef KDTRACE_HOOKS
void
clocksource_cyc_set(const struct bintime *bt)
{
	sbintime_t now, t;
	struct pcpu_state *state;

	/* Do not touch anything if somebody reconfiguring timers. */
	if (busy)
		return;
	t = bttosbt(*bt);
	state = DPCPU_PTR(timerstate);
	if (periodic)
		now = state->now;
	else
		now = sbinuptime();

	CTR5(KTR_SPARE2, "set_cyc at %d:  now  %d.%08x  t  %d.%08x",
	    curcpu, (int)(now >> 32), (u_int)(now & 0xffffffff),
	    (int)(t >> 32), (u_int)(t & 0xffffffff));

	ET_HW_LOCK(state);
	if (t == state->nextcyc)
		goto done;
	state->nextcyc = t;
	if (t >= state->nextevent)
		goto done;
	state->nextevent = t;
	if (!periodic)
		loadtimer(now, 0);
done:
	ET_HW_UNLOCK(state);
}
#endif

void
cpu_new_callout(int cpu, sbintime_t bt, sbintime_t bt_opt)
{
	struct pcpu_state *state;

	/* Do not touch anything if somebody reconfiguring timers. */
	if (busy)
		return;
	CTR6(KTR_SPARE2, "new co at %d:    on %d at %d.%08x - %d.%08x",
	    curcpu, cpu, (int)(bt_opt >> 32), (u_int)(bt_opt & 0xffffffff),
	    (int)(bt >> 32), (u_int)(bt & 0xffffffff));
	state = DPCPU_ID_PTR(cpu, timerstate);
	ET_HW_LOCK(state);

	/*
	 * If there is callout time already set earlier -- do nothing.
	 * This check may appear redundant because we check already in
	 * callout_process() but this double check guarantees we're safe
	 * with respect to race conditions between interrupts execution
	 * and scheduling.
	 */
	state->nextcallopt = bt_opt;
	if (bt >= state->nextcall)
		goto done;
	state->nextcall = bt;
	/* If there is some other event set earlier -- do nothing. */
	if (bt >= state->nextevent)
		goto done;
	state->nextevent = bt;
	/* If timer is periodic -- there is nothing to reprogram. */
	if (periodic)
		goto done;
	/* If timer is global or of the current CPU -- reprogram it. */
	if ((timer->et_flags & ET_FLAGS_PERCPU) == 0 || cpu == curcpu) {
		loadtimer(sbinuptime(), 0);
done:
		ET_HW_UNLOCK(state);
		return;
	}
	/* Otherwise make other CPU to reprogram it. */
	state->handle = 1;
	ET_HW_UNLOCK(state);
#ifdef SMP
	ipi_cpu(cpu, IPI_HARDCLOCK);
#endif
}

/*
 * Report or change the active event timers hardware.
 */
static int
sysctl_kern_eventtimer_timer(SYSCTL_HANDLER_ARGS)
{
	char buf[32];
	struct eventtimer *et;
	int error;

	ET_LOCK();
	et = timer;
	snprintf(buf, sizeof(buf), "%s", et->et_name);
	ET_UNLOCK();
	error = sysctl_handle_string(oidp, buf, sizeof(buf), req);
	ET_LOCK();
	et = timer;
	if (error != 0 || req->newptr == NULL ||
	    strcasecmp(buf, et->et_name) == 0) {
		ET_UNLOCK();
		return (error);
	}
	et = et_find(buf, 0, 0);
	if (et == NULL) {
		ET_UNLOCK();
		return (ENOENT);
	}
	configtimer(0);
	et_free(timer);
	if (et->et_flags & ET_FLAGS_C3STOP)
		cpu_disable_deep_sleep++;
	if (timer->et_flags & ET_FLAGS_C3STOP)
		cpu_disable_deep_sleep--;
	periodic = want_periodic;
	timer = et;
	et_init(timer, timercb, NULL, NULL);
	configtimer(1);
	ET_UNLOCK();
	return (error);
}
SYSCTL_PROC(_kern_eventtimer, OID_AUTO, timer,
    CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_MPSAFE,
    0, 0, sysctl_kern_eventtimer_timer, "A", "Chosen event timer");

/*
 * Report or change the active event timer periodicity.
 */
static int
sysctl_kern_eventtimer_periodic(SYSCTL_HANDLER_ARGS)
{
	int error, val;

	val = periodic;
	error = sysctl_handle_int(oidp, &val, 0, req);
	if (error != 0 || req->newptr == NULL)
		return (error);
	ET_LOCK();
	configtimer(0);
	periodic = want_periodic = val;
	configtimer(1);
	ET_UNLOCK();
	return (error);
}
SYSCTL_PROC(_kern_eventtimer, OID_AUTO, periodic,
    CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE,
    0, 0, sysctl_kern_eventtimer_periodic, "I", "Enable event timer periodic mode");
OpenPOWER on IntegriCloud