forked from opnsense/src
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathkern_timeout.c
1673 lines (1553 loc) · 47.2 KB
/
kern_timeout.c
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
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/*-
* SPDX-License-Identifier: BSD-3-Clause
*
* Copyright (c) 1982, 1986, 1991, 1993
* The Regents of the University of California. All rights reserved.
* (c) UNIX System Laboratories, Inc.
* All or some portions of this file are derived from material licensed
* to the University of California by American Telephone and Telegraph
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
* the permission of UNIX System Laboratories, Inc.
*
* 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. 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.
* 3. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``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 REGENTS OR CONTRIBUTORS 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.
*
* From: @(#)kern_clock.c 8.5 (Berkeley) 1/21/94
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_callout_profiling.h"
#include "opt_ddb.h"
#if defined(__arm__)
#include "opt_timer.h"
#endif
#include "opt_rss.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/bus.h>
#include <sys/callout.h>
#include <sys/file.h>
#include <sys/interrupt.h>
#include <sys/kernel.h>
#include <sys/ktr.h>
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mutex.h>
#include <sys/proc.h>
#include <sys/sdt.h>
#include <sys/sleepqueue.h>
#include <sys/sysctl.h>
#include <sys/smp.h>
#ifdef DDB
#include <ddb/ddb.h>
#include <machine/_inttypes.h>
#endif
#ifdef SMP
#include <machine/cpu.h>
#endif
#ifndef NO_EVENTTIMERS
DPCPU_DECLARE(sbintime_t, hardclocktime);
#endif
SDT_PROVIDER_DEFINE(callout_execute);
SDT_PROBE_DEFINE1(callout_execute, , , callout__start, "struct callout *");
SDT_PROBE_DEFINE1(callout_execute, , , callout__end, "struct callout *");
#ifdef CALLOUT_PROFILING
static int avg_depth;
SYSCTL_INT(_debug, OID_AUTO, to_avg_depth, CTLFLAG_RD, &avg_depth, 0,
"Average number of items examined per softclock call. Units = 1/1000");
static int avg_gcalls;
SYSCTL_INT(_debug, OID_AUTO, to_avg_gcalls, CTLFLAG_RD, &avg_gcalls, 0,
"Average number of Giant callouts made per softclock call. Units = 1/1000");
static int avg_lockcalls;
SYSCTL_INT(_debug, OID_AUTO, to_avg_lockcalls, CTLFLAG_RD, &avg_lockcalls, 0,
"Average number of lock callouts made per softclock call. Units = 1/1000");
static int avg_mpcalls;
SYSCTL_INT(_debug, OID_AUTO, to_avg_mpcalls, CTLFLAG_RD, &avg_mpcalls, 0,
"Average number of MP callouts made per softclock call. Units = 1/1000");
static int avg_depth_dir;
SYSCTL_INT(_debug, OID_AUTO, to_avg_depth_dir, CTLFLAG_RD, &avg_depth_dir, 0,
"Average number of direct callouts examined per callout_process call. "
"Units = 1/1000");
static int avg_lockcalls_dir;
SYSCTL_INT(_debug, OID_AUTO, to_avg_lockcalls_dir, CTLFLAG_RD,
&avg_lockcalls_dir, 0, "Average number of lock direct callouts made per "
"callout_process call. Units = 1/1000");
static int avg_mpcalls_dir;
SYSCTL_INT(_debug, OID_AUTO, to_avg_mpcalls_dir, CTLFLAG_RD, &avg_mpcalls_dir,
0, "Average number of MP direct callouts made per callout_process call. "
"Units = 1/1000");
#endif
static int ncallout;
SYSCTL_INT(_kern, OID_AUTO, ncallout, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &ncallout, 0,
"Number of entries in callwheel and size of timeout() preallocation");
#ifdef RSS
static int pin_default_swi = 1;
static int pin_pcpu_swi = 1;
#else
static int pin_default_swi = 0;
static int pin_pcpu_swi = 0;
#endif
SYSCTL_INT(_kern, OID_AUTO, pin_default_swi, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &pin_default_swi,
0, "Pin the default (non-per-cpu) swi (shared with PCPU 0 swi)");
SYSCTL_INT(_kern, OID_AUTO, pin_pcpu_swi, CTLFLAG_RDTUN | CTLFLAG_NOFETCH, &pin_pcpu_swi,
0, "Pin the per-CPU swis (except PCPU 0, which is also default");
/*
* TODO:
* allocate more timeout table slots when table overflows.
*/
u_int callwheelsize, callwheelmask;
/*
* The callout cpu exec entities represent informations necessary for
* describing the state of callouts currently running on the CPU and the ones
* necessary for migrating callouts to the new callout cpu. In particular,
* the first entry of the array cc_exec_entity holds informations for callout
* running in SWI thread context, while the second one holds informations
* for callout running directly from hardware interrupt context.
* The cached informations are very important for deferring migration when
* the migrating callout is already running.
*/
struct cc_exec {
struct callout *cc_curr;
void (*cc_drain)(void *);
#ifdef SMP
void (*ce_migration_func)(void *);
void *ce_migration_arg;
int ce_migration_cpu;
sbintime_t ce_migration_time;
sbintime_t ce_migration_prec;
#endif
bool cc_cancel;
bool cc_waiting;
};
/*
* There is one struct callout_cpu per cpu, holding all relevant
* state for the callout processing thread on the individual CPU.
*/
struct callout_cpu {
struct mtx_padalign cc_lock;
struct cc_exec cc_exec_entity[2];
struct callout *cc_next;
struct callout *cc_callout;
struct callout_list *cc_callwheel;
struct callout_tailq cc_expireq;
struct callout_slist cc_callfree;
sbintime_t cc_firstevent;
sbintime_t cc_lastscan;
void *cc_cookie;
u_int cc_bucket;
u_int cc_inited;
char cc_ktr_event_name[20];
};
#define callout_migrating(c) ((c)->c_iflags & CALLOUT_DFRMIGRATION)
#define cc_exec_curr(cc, dir) cc->cc_exec_entity[dir].cc_curr
#define cc_exec_drain(cc, dir) cc->cc_exec_entity[dir].cc_drain
#define cc_exec_next(cc) cc->cc_next
#define cc_exec_cancel(cc, dir) cc->cc_exec_entity[dir].cc_cancel
#define cc_exec_waiting(cc, dir) cc->cc_exec_entity[dir].cc_waiting
#ifdef SMP
#define cc_migration_func(cc, dir) cc->cc_exec_entity[dir].ce_migration_func
#define cc_migration_arg(cc, dir) cc->cc_exec_entity[dir].ce_migration_arg
#define cc_migration_cpu(cc, dir) cc->cc_exec_entity[dir].ce_migration_cpu
#define cc_migration_time(cc, dir) cc->cc_exec_entity[dir].ce_migration_time
#define cc_migration_prec(cc, dir) cc->cc_exec_entity[dir].ce_migration_prec
struct callout_cpu cc_cpu[MAXCPU];
#define CPUBLOCK MAXCPU
#define CC_CPU(cpu) (&cc_cpu[(cpu)])
#define CC_SELF() CC_CPU(PCPU_GET(cpuid))
#else
struct callout_cpu cc_cpu;
#define CC_CPU(cpu) &cc_cpu
#define CC_SELF() &cc_cpu
#endif
#define CC_LOCK(cc) mtx_lock_spin(&(cc)->cc_lock)
#define CC_UNLOCK(cc) mtx_unlock_spin(&(cc)->cc_lock)
#define CC_LOCK_ASSERT(cc) mtx_assert(&(cc)->cc_lock, MA_OWNED)
static int timeout_cpu;
static void callout_cpu_init(struct callout_cpu *cc, int cpu);
static void softclock_call_cc(struct callout *c, struct callout_cpu *cc,
#ifdef CALLOUT_PROFILING
int *mpcalls, int *lockcalls, int *gcalls,
#endif
int direct);
static MALLOC_DEFINE(M_CALLOUT, "callout", "Callout datastructures");
/**
* Locked by cc_lock:
* cc_curr - If a callout is in progress, it is cc_curr.
* If cc_curr is non-NULL, threads waiting in
* callout_drain() will be woken up as soon as the
* relevant callout completes.
* cc_cancel - Changing to 1 with both callout_lock and cc_lock held
* guarantees that the current callout will not run.
* The softclock() function sets this to 0 before it
* drops callout_lock to acquire c_lock, and it calls
* the handler only if curr_cancelled is still 0 after
* cc_lock is successfully acquired.
* cc_waiting - If a thread is waiting in callout_drain(), then
* callout_wait is nonzero. Set only when
* cc_curr is non-NULL.
*/
/*
* Resets the execution entity tied to a specific callout cpu.
*/
static void
cc_cce_cleanup(struct callout_cpu *cc, int direct)
{
cc_exec_curr(cc, direct) = NULL;
cc_exec_cancel(cc, direct) = false;
cc_exec_waiting(cc, direct) = false;
#ifdef SMP
cc_migration_cpu(cc, direct) = CPUBLOCK;
cc_migration_time(cc, direct) = 0;
cc_migration_prec(cc, direct) = 0;
cc_migration_func(cc, direct) = NULL;
cc_migration_arg(cc, direct) = NULL;
#endif
}
/*
* Checks if migration is requested by a specific callout cpu.
*/
static int
cc_cce_migrating(struct callout_cpu *cc, int direct)
{
#ifdef SMP
return (cc_migration_cpu(cc, direct) != CPUBLOCK);
#else
return (0);
#endif
}
/*
* Kernel low level callwheel initialization
* called on the BSP during kernel startup.
*/
static void
callout_callwheel_init(void *dummy)
{
struct callout_cpu *cc;
/*
* Calculate the size of the callout wheel and the preallocated
* timeout() structures.
* XXX: Clip callout to result of previous function of maxusers
* maximum 384. This is still huge, but acceptable.
*/
memset(CC_CPU(curcpu), 0, sizeof(cc_cpu));
ncallout = imin(16 + maxproc + maxfiles, 18508);
TUNABLE_INT_FETCH("kern.ncallout", &ncallout);
/*
* Calculate callout wheel size, should be next power of two higher
* than 'ncallout'.
*/
callwheelsize = 1 << fls(ncallout);
callwheelmask = callwheelsize - 1;
/*
* Fetch whether we're pinning the swi's or not.
*/
TUNABLE_INT_FETCH("kern.pin_default_swi", &pin_default_swi);
TUNABLE_INT_FETCH("kern.pin_pcpu_swi", &pin_pcpu_swi);
/*
* Only BSP handles timeout(9) and receives a preallocation.
*
* XXX: Once all timeout(9) consumers are converted this can
* be removed.
*/
timeout_cpu = PCPU_GET(cpuid);
cc = CC_CPU(timeout_cpu);
cc->cc_callout = malloc(ncallout * sizeof(struct callout),
M_CALLOUT, M_WAITOK);
callout_cpu_init(cc, timeout_cpu);
}
SYSINIT(callwheel_init, SI_SUB_CPU, SI_ORDER_ANY, callout_callwheel_init, NULL);
/*
* Initialize the per-cpu callout structures.
*/
static void
callout_cpu_init(struct callout_cpu *cc, int cpu)
{
struct callout *c;
int i;
mtx_init(&cc->cc_lock, "callout", NULL, MTX_SPIN | MTX_RECURSE);
SLIST_INIT(&cc->cc_callfree);
cc->cc_inited = 1;
cc->cc_callwheel = malloc(sizeof(struct callout_list) * callwheelsize,
M_CALLOUT, M_WAITOK);
for (i = 0; i < callwheelsize; i++)
LIST_INIT(&cc->cc_callwheel[i]);
TAILQ_INIT(&cc->cc_expireq);
cc->cc_firstevent = SBT_MAX;
for (i = 0; i < 2; i++)
cc_cce_cleanup(cc, i);
snprintf(cc->cc_ktr_event_name, sizeof(cc->cc_ktr_event_name),
"callwheel cpu %d", cpu);
if (cc->cc_callout == NULL) /* Only BSP handles timeout(9) */
return;
for (i = 0; i < ncallout; i++) {
c = &cc->cc_callout[i];
callout_init(c, 0);
c->c_iflags = CALLOUT_LOCAL_ALLOC;
SLIST_INSERT_HEAD(&cc->cc_callfree, c, c_links.sle);
}
}
#ifdef SMP
/*
* Switches the cpu tied to a specific callout.
* The function expects a locked incoming callout cpu and returns with
* locked outcoming callout cpu.
*/
static struct callout_cpu *
callout_cpu_switch(struct callout *c, struct callout_cpu *cc, int new_cpu)
{
struct callout_cpu *new_cc;
MPASS(c != NULL && cc != NULL);
CC_LOCK_ASSERT(cc);
/*
* Avoid interrupts and preemption firing after the callout cpu
* is blocked in order to avoid deadlocks as the new thread
* may be willing to acquire the callout cpu lock.
*/
c->c_cpu = CPUBLOCK;
spinlock_enter();
CC_UNLOCK(cc);
new_cc = CC_CPU(new_cpu);
CC_LOCK(new_cc);
spinlock_exit();
c->c_cpu = new_cpu;
return (new_cc);
}
#endif
/*
* Start standard softclock thread.
*/
static void
start_softclock(void *dummy)
{
struct callout_cpu *cc;
char name[MAXCOMLEN];
#ifdef SMP
int cpu;
struct intr_event *ie;
#endif
cc = CC_CPU(timeout_cpu);
snprintf(name, sizeof(name), "clock (%d)", timeout_cpu);
if (swi_add(&clk_intr_event, name, softclock, cc, SWI_CLOCK,
INTR_MPSAFE, &cc->cc_cookie))
panic("died while creating standard software ithreads");
if (pin_default_swi &&
(intr_event_bind(clk_intr_event, timeout_cpu) != 0)) {
printf("%s: timeout clock couldn't be pinned to cpu %d\n",
__func__,
timeout_cpu);
}
#ifdef SMP
CPU_FOREACH(cpu) {
if (cpu == timeout_cpu)
continue;
cc = CC_CPU(cpu);
cc->cc_callout = NULL; /* Only BSP handles timeout(9). */
callout_cpu_init(cc, cpu);
snprintf(name, sizeof(name), "clock (%d)", cpu);
ie = NULL;
if (swi_add(&ie, name, softclock, cc, SWI_CLOCK,
INTR_MPSAFE, &cc->cc_cookie))
panic("died while creating standard software ithreads");
if (pin_pcpu_swi && (intr_event_bind(ie, cpu) != 0)) {
printf("%s: per-cpu clock couldn't be pinned to "
"cpu %d\n",
__func__,
cpu);
}
}
#endif
}
SYSINIT(start_softclock, SI_SUB_SOFTINTR, SI_ORDER_FIRST, start_softclock, NULL);
#define CC_HASH_SHIFT 8
static inline u_int
callout_hash(sbintime_t sbt)
{
return (sbt >> (32 - CC_HASH_SHIFT));
}
static inline u_int
callout_get_bucket(sbintime_t sbt)
{
return (callout_hash(sbt) & callwheelmask);
}
void
callout_process(sbintime_t now)
{
struct callout *tmp, *tmpn;
struct callout_cpu *cc;
struct callout_list *sc;
sbintime_t first, last, max, tmp_max;
uint32_t lookahead;
u_int firstb, lastb, nowb;
#ifdef CALLOUT_PROFILING
int depth_dir = 0, mpcalls_dir = 0, lockcalls_dir = 0;
#endif
cc = CC_SELF();
mtx_lock_spin_flags(&cc->cc_lock, MTX_QUIET);
/* Compute the buckets of the last scan and present times. */
firstb = callout_hash(cc->cc_lastscan);
cc->cc_lastscan = now;
nowb = callout_hash(now);
/* Compute the last bucket and minimum time of the bucket after it. */
if (nowb == firstb)
lookahead = (SBT_1S / 16);
else if (nowb - firstb == 1)
lookahead = (SBT_1S / 8);
else
lookahead = (SBT_1S / 2);
first = last = now;
first += (lookahead / 2);
last += lookahead;
last &= (0xffffffffffffffffLLU << (32 - CC_HASH_SHIFT));
lastb = callout_hash(last) - 1;
max = last;
/*
* Check if we wrapped around the entire wheel from the last scan.
* In case, we need to scan entirely the wheel for pending callouts.
*/
if (lastb - firstb >= callwheelsize) {
lastb = firstb + callwheelsize - 1;
if (nowb - firstb >= callwheelsize)
nowb = lastb;
}
/* Iterate callwheel from firstb to nowb and then up to lastb. */
do {
sc = &cc->cc_callwheel[firstb & callwheelmask];
tmp = LIST_FIRST(sc);
while (tmp != NULL) {
/* Run the callout if present time within allowed. */
if (tmp->c_time <= now) {
/*
* Consumer told us the callout may be run
* directly from hardware interrupt context.
*/
if (tmp->c_iflags & CALLOUT_DIRECT) {
#ifdef CALLOUT_PROFILING
++depth_dir;
#endif
cc_exec_next(cc) =
LIST_NEXT(tmp, c_links.le);
cc->cc_bucket = firstb & callwheelmask;
LIST_REMOVE(tmp, c_links.le);
softclock_call_cc(tmp, cc,
#ifdef CALLOUT_PROFILING
&mpcalls_dir, &lockcalls_dir, NULL,
#endif
1);
tmp = cc_exec_next(cc);
cc_exec_next(cc) = NULL;
} else {
tmpn = LIST_NEXT(tmp, c_links.le);
LIST_REMOVE(tmp, c_links.le);
TAILQ_INSERT_TAIL(&cc->cc_expireq,
tmp, c_links.tqe);
tmp->c_iflags |= CALLOUT_PROCESSED;
tmp = tmpn;
}
continue;
}
/* Skip events from distant future. */
if (tmp->c_time >= max)
goto next;
/*
* Event minimal time is bigger than present maximal
* time, so it cannot be aggregated.
*/
if (tmp->c_time > last) {
lastb = nowb;
goto next;
}
/* Update first and last time, respecting this event. */
if (tmp->c_time < first)
first = tmp->c_time;
tmp_max = tmp->c_time + tmp->c_precision;
if (tmp_max < last)
last = tmp_max;
next:
tmp = LIST_NEXT(tmp, c_links.le);
}
/* Proceed with the next bucket. */
firstb++;
/*
* Stop if we looked after present time and found
* some event we can't execute at now.
* Stop if we looked far enough into the future.
*/
} while (((int)(firstb - lastb)) <= 0);
cc->cc_firstevent = last;
#ifndef NO_EVENTTIMERS
cpu_new_callout(curcpu, last, first);
#endif
#ifdef CALLOUT_PROFILING
avg_depth_dir += (depth_dir * 1000 - avg_depth_dir) >> 8;
avg_mpcalls_dir += (mpcalls_dir * 1000 - avg_mpcalls_dir) >> 8;
avg_lockcalls_dir += (lockcalls_dir * 1000 - avg_lockcalls_dir) >> 8;
#endif
mtx_unlock_spin_flags(&cc->cc_lock, MTX_QUIET);
/*
* swi_sched acquires the thread lock, so we don't want to call it
* with cc_lock held; incorrect locking order.
*/
if (!TAILQ_EMPTY(&cc->cc_expireq))
swi_sched(cc->cc_cookie, 0);
}
static struct callout_cpu *
callout_lock(struct callout *c)
{
struct callout_cpu *cc;
int cpu;
for (;;) {
cpu = c->c_cpu;
#ifdef SMP
if (cpu == CPUBLOCK) {
while (c->c_cpu == CPUBLOCK)
cpu_spinwait();
continue;
}
#endif
cc = CC_CPU(cpu);
CC_LOCK(cc);
if (cpu == c->c_cpu)
break;
CC_UNLOCK(cc);
}
return (cc);
}
static void
callout_cc_add(struct callout *c, struct callout_cpu *cc,
sbintime_t sbt, sbintime_t precision, void (*func)(void *),
void *arg, int cpu, int flags)
{
int bucket;
CC_LOCK_ASSERT(cc);
if (sbt < cc->cc_lastscan)
sbt = cc->cc_lastscan;
c->c_arg = arg;
c->c_iflags |= CALLOUT_PENDING;
c->c_iflags &= ~CALLOUT_PROCESSED;
c->c_flags |= CALLOUT_ACTIVE;
if (flags & C_DIRECT_EXEC)
c->c_iflags |= CALLOUT_DIRECT;
c->c_func = func;
c->c_time = sbt;
c->c_precision = precision;
bucket = callout_get_bucket(c->c_time);
CTR3(KTR_CALLOUT, "precision set for %p: %d.%08x",
c, (int)(c->c_precision >> 32),
(u_int)(c->c_precision & 0xffffffff));
LIST_INSERT_HEAD(&cc->cc_callwheel[bucket], c, c_links.le);
if (cc->cc_bucket == bucket)
cc_exec_next(cc) = c;
#ifndef NO_EVENTTIMERS
/*
* Inform the eventtimers(4) subsystem there's a new callout
* that has been inserted, but only if really required.
*/
if (SBT_MAX - c->c_time < c->c_precision)
c->c_precision = SBT_MAX - c->c_time;
sbt = c->c_time + c->c_precision;
if (sbt < cc->cc_firstevent) {
cc->cc_firstevent = sbt;
cpu_new_callout(cpu, sbt, c->c_time);
}
#endif
}
static void
callout_cc_del(struct callout *c, struct callout_cpu *cc)
{
if ((c->c_iflags & CALLOUT_LOCAL_ALLOC) == 0)
return;
c->c_func = NULL;
SLIST_INSERT_HEAD(&cc->cc_callfree, c, c_links.sle);
}
static void
softclock_call_cc(struct callout *c, struct callout_cpu *cc,
#ifdef CALLOUT_PROFILING
int *mpcalls, int *lockcalls, int *gcalls,
#endif
int direct)
{
struct rm_priotracker tracker;
void (*c_func)(void *);
void *c_arg;
struct lock_class *class;
struct lock_object *c_lock;
uintptr_t lock_status;
int c_iflags;
#ifdef SMP
struct callout_cpu *new_cc;
void (*new_func)(void *);
void *new_arg;
int flags, new_cpu;
sbintime_t new_prec, new_time;
#endif
#if defined(DIAGNOSTIC) || defined(CALLOUT_PROFILING)
sbintime_t sbt1, sbt2;
struct timespec ts2;
static sbintime_t maxdt = 2 * SBT_1MS; /* 2 msec */
static timeout_t *lastfunc;
#endif
KASSERT((c->c_iflags & CALLOUT_PENDING) == CALLOUT_PENDING,
("softclock_call_cc: pend %p %x", c, c->c_iflags));
KASSERT((c->c_flags & CALLOUT_ACTIVE) == CALLOUT_ACTIVE,
("softclock_call_cc: act %p %x", c, c->c_flags));
class = (c->c_lock != NULL) ? LOCK_CLASS(c->c_lock) : NULL;
lock_status = 0;
if (c->c_flags & CALLOUT_SHAREDLOCK) {
if (class == &lock_class_rm)
lock_status = (uintptr_t)&tracker;
else
lock_status = 1;
}
c_lock = c->c_lock;
c_func = c->c_func;
c_arg = c->c_arg;
c_iflags = c->c_iflags;
if (c->c_iflags & CALLOUT_LOCAL_ALLOC)
c->c_iflags = CALLOUT_LOCAL_ALLOC;
else
c->c_iflags &= ~CALLOUT_PENDING;
cc_exec_curr(cc, direct) = c;
cc_exec_cancel(cc, direct) = false;
cc_exec_drain(cc, direct) = NULL;
CC_UNLOCK(cc);
if (c_lock != NULL) {
class->lc_lock(c_lock, lock_status);
/*
* The callout may have been cancelled
* while we switched locks.
*/
if (cc_exec_cancel(cc, direct)) {
class->lc_unlock(c_lock);
goto skip;
}
/* The callout cannot be stopped now. */
cc_exec_cancel(cc, direct) = true;
if (c_lock == &Giant.lock_object) {
#ifdef CALLOUT_PROFILING
(*gcalls)++;
#endif
CTR3(KTR_CALLOUT, "callout giant %p func %p arg %p",
c, c_func, c_arg);
} else {
#ifdef CALLOUT_PROFILING
(*lockcalls)++;
#endif
CTR3(KTR_CALLOUT, "callout lock %p func %p arg %p",
c, c_func, c_arg);
}
} else {
#ifdef CALLOUT_PROFILING
(*mpcalls)++;
#endif
CTR3(KTR_CALLOUT, "callout %p func %p arg %p",
c, c_func, c_arg);
}
KTR_STATE3(KTR_SCHED, "callout", cc->cc_ktr_event_name, "running",
"func:%p", c_func, "arg:%p", c_arg, "direct:%d", direct);
#if defined(DIAGNOSTIC) || defined(CALLOUT_PROFILING)
sbt1 = sbinuptime();
#endif
THREAD_NO_SLEEPING();
SDT_PROBE1(callout_execute, , , callout__start, c);
c_func(c_arg);
SDT_PROBE1(callout_execute, , , callout__end, c);
THREAD_SLEEPING_OK();
#if defined(DIAGNOSTIC) || defined(CALLOUT_PROFILING)
sbt2 = sbinuptime();
sbt2 -= sbt1;
if (sbt2 > maxdt) {
if (lastfunc != c_func || sbt2 > maxdt * 2) {
ts2 = sbttots(sbt2);
printf(
"Expensive timeout(9) function: %p(%p) %jd.%09ld s\n",
c_func, c_arg, (intmax_t)ts2.tv_sec, ts2.tv_nsec);
}
maxdt = sbt2;
lastfunc = c_func;
}
#endif
KTR_STATE0(KTR_SCHED, "callout", cc->cc_ktr_event_name, "idle");
CTR1(KTR_CALLOUT, "callout %p finished", c);
if ((c_iflags & CALLOUT_RETURNUNLOCKED) == 0)
class->lc_unlock(c_lock);
skip:
CC_LOCK(cc);
KASSERT(cc_exec_curr(cc, direct) == c, ("mishandled cc_curr"));
cc_exec_curr(cc, direct) = NULL;
if (cc_exec_drain(cc, direct)) {
void (*drain)(void *);
drain = cc_exec_drain(cc, direct);
cc_exec_drain(cc, direct) = NULL;
CC_UNLOCK(cc);
drain(c_arg);
CC_LOCK(cc);
}
if (cc_exec_waiting(cc, direct)) {
/*
* There is someone waiting for the
* callout to complete.
* If the callout was scheduled for
* migration just cancel it.
*/
if (cc_cce_migrating(cc, direct)) {
cc_cce_cleanup(cc, direct);
/*
* It should be assert here that the callout is not
* destroyed but that is not easy.
*/
c->c_iflags &= ~CALLOUT_DFRMIGRATION;
}
cc_exec_waiting(cc, direct) = false;
CC_UNLOCK(cc);
wakeup(&cc_exec_waiting(cc, direct));
CC_LOCK(cc);
} else if (cc_cce_migrating(cc, direct)) {
KASSERT((c_iflags & CALLOUT_LOCAL_ALLOC) == 0,
("Migrating legacy callout %p", c));
#ifdef SMP
/*
* If the callout was scheduled for
* migration just perform it now.
*/
new_cpu = cc_migration_cpu(cc, direct);
new_time = cc_migration_time(cc, direct);
new_prec = cc_migration_prec(cc, direct);
new_func = cc_migration_func(cc, direct);
new_arg = cc_migration_arg(cc, direct);
cc_cce_cleanup(cc, direct);
/*
* It should be assert here that the callout is not destroyed
* but that is not easy.
*
* As first thing, handle deferred callout stops.
*/
if (!callout_migrating(c)) {
CTR3(KTR_CALLOUT,
"deferred cancelled %p func %p arg %p",
c, new_func, new_arg);
callout_cc_del(c, cc);
return;
}
c->c_iflags &= ~CALLOUT_DFRMIGRATION;
new_cc = callout_cpu_switch(c, cc, new_cpu);
flags = (direct) ? C_DIRECT_EXEC : 0;
callout_cc_add(c, new_cc, new_time, new_prec, new_func,
new_arg, new_cpu, flags);
CC_UNLOCK(new_cc);
CC_LOCK(cc);
#else
panic("migration should not happen");
#endif
}
/*
* If the current callout is locally allocated (from
* timeout(9)) then put it on the freelist.
*
* Note: we need to check the cached copy of c_iflags because
* if it was not local, then it's not safe to deref the
* callout pointer.
*/
KASSERT((c_iflags & CALLOUT_LOCAL_ALLOC) == 0 ||
c->c_iflags == CALLOUT_LOCAL_ALLOC,
("corrupted callout"));
if (c_iflags & CALLOUT_LOCAL_ALLOC)
callout_cc_del(c, cc);
}
/*
* The callout mechanism is based on the work of Adam M. Costello and
* George Varghese, published in a technical report entitled "Redesigning
* the BSD Callout and Timer Facilities" and modified slightly for inclusion
* in FreeBSD by Justin T. Gibbs. The original work on the data structures
* used in this implementation was published by G. Varghese and T. Lauck in
* the paper "Hashed and Hierarchical Timing Wheels: Data Structures for
* the Efficient Implementation of a Timer Facility" in the Proceedings of
* the 11th ACM Annual Symposium on Operating Systems Principles,
* Austin, Texas Nov 1987.
*/
/*
* Software (low priority) clock interrupt.
* Run periodic events from timeout queue.
*/
void
softclock(void *arg)
{
struct callout_cpu *cc;
struct callout *c;
#ifdef CALLOUT_PROFILING
int depth = 0, gcalls = 0, lockcalls = 0, mpcalls = 0;
#endif
cc = (struct callout_cpu *)arg;
CC_LOCK(cc);
while ((c = TAILQ_FIRST(&cc->cc_expireq)) != NULL) {
TAILQ_REMOVE(&cc->cc_expireq, c, c_links.tqe);
softclock_call_cc(c, cc,
#ifdef CALLOUT_PROFILING
&mpcalls, &lockcalls, &gcalls,
#endif
0);
#ifdef CALLOUT_PROFILING
++depth;
#endif
}
#ifdef CALLOUT_PROFILING
avg_depth += (depth * 1000 - avg_depth) >> 8;
avg_mpcalls += (mpcalls * 1000 - avg_mpcalls) >> 8;
avg_lockcalls += (lockcalls * 1000 - avg_lockcalls) >> 8;
avg_gcalls += (gcalls * 1000 - avg_gcalls) >> 8;
#endif
CC_UNLOCK(cc);
}
/*
* timeout --
* Execute a function after a specified length of time.
*
* untimeout --
* Cancel previous timeout function call.
*
* callout_handle_init --
* Initialize a handle so that using it with untimeout is benign.
*
* See AT&T BCI Driver Reference Manual for specification. This
* implementation differs from that one in that although an
* identification value is returned from timeout, the original
* arguments to timeout as well as the identifier are used to
* identify entries for untimeout.
*/
struct callout_handle
timeout(timeout_t *ftn, void *arg, int to_ticks)
{
struct callout_cpu *cc;
struct callout *new;
struct callout_handle handle;
cc = CC_CPU(timeout_cpu);
CC_LOCK(cc);
/* Fill in the next free callout structure. */
new = SLIST_FIRST(&cc->cc_callfree);
if (new == NULL)
/* XXX Attempt to malloc first */
panic("timeout table full");
SLIST_REMOVE_HEAD(&cc->cc_callfree, c_links.sle);
callout_reset(new, to_ticks, ftn, arg);
handle.callout = new;
CC_UNLOCK(cc);
return (handle);
}
void
untimeout(timeout_t *ftn, void *arg, struct callout_handle handle)
{
struct callout_cpu *cc;
/*
* Check for a handle that was initialized
* by callout_handle_init, but never used
* for a real timeout.
*/
if (handle.callout == NULL)
return;
cc = callout_lock(handle.callout);
if (handle.callout->c_func == ftn && handle.callout->c_arg == arg)
callout_stop(handle.callout);
CC_UNLOCK(cc);
}
void
callout_handle_init(struct callout_handle *handle)
{
handle->callout = NULL;
}
void
callout_when(sbintime_t sbt, sbintime_t precision, int flags,
sbintime_t *res, sbintime_t *prec_res)
{
sbintime_t to_sbt, to_pr;
if ((flags & (C_ABSOLUTE | C_PRECALC)) != 0) {
*res = sbt;
*prec_res = precision;
return;
}
if ((flags & C_HARDCLOCK) != 0 && sbt < tick_sbt)
sbt = tick_sbt;
if ((flags & C_HARDCLOCK) != 0 ||
#ifdef NO_EVENTTIMERS
sbt >= sbt_timethreshold) {
to_sbt = getsbinuptime();
/* Add safety belt for the case of hz > 1000. */
to_sbt += tc_tick_sbt - tick_sbt;
#else
sbt >= sbt_tickthreshold) {
/*
* Obtain the time of the last hardclock() call on
* this CPU directly from the kern_clocksource.c.
* This value is per-CPU, but it is equal for all
* active ones.
*/
#ifdef __LP64__
to_sbt = DPCPU_GET(hardclocktime);
#else
spinlock_enter();
to_sbt = DPCPU_GET(hardclocktime);
spinlock_exit();
#endif
#endif
if (cold && to_sbt == 0)
to_sbt = sbinuptime();
if ((flags & C_HARDCLOCK) == 0)
to_sbt += tick_sbt;
} else
to_sbt = sbinuptime();
if (SBT_MAX - to_sbt < sbt)
to_sbt = SBT_MAX;
else
to_sbt += sbt;
*res = to_sbt;
to_pr = ((C_PRELGET(flags) < 0) ? sbt >> tc_precexp :
sbt >> C_PRELGET(flags));
*prec_res = to_pr > precision ? to_pr : precision;
}