forked from ossrs/state-threads
-
Notifications
You must be signed in to change notification settings - Fork 0
/
event.c
1276 lines (1107 loc) · 36.8 KB
/
event.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: MPL-1.1 OR GPL-2.0-or-later */
/*
* The contents of this file are subject to the Mozilla Public
* License Version 1.1 (the "License"); you may not use this file
* except in compliance with the License. You may obtain a copy of
* the License at http://www.mozilla.org/MPL/
*
* Software distributed under the License is distributed on an "AS
* IS" basis, WITHOUT WARRANTY OF ANY KIND, either express or
* implied. See the License for the specific language governing
* rights and limitations under the License.
*
* The Original Code is the Netscape Portable Runtime library.
*
* The Initial Developer of the Original Code is Netscape
* Communications Corporation. Portions created by Netscape are
* Copyright (C) 1994-2000 Netscape Communications Corporation. All
* Rights Reserved.
*
* Contributor(s): Silicon Graphics, Inc.
* Yahoo! Inc.
*
* Alternatively, the contents of this file may be used under the
* terms of the GNU General Public License Version 2 or later (the
* "GPL"), in which case the provisions of the GPL are applicable
* instead of those above. If you wish to allow use of your
* version of this file only under the terms of the GPL and not to
* allow others to use your version of this file under the MPL,
* indicate your decision by deleting the provisions above and
* replace them with the notice and other provisions required by
* the GPL. If you do not delete the provisions above, a recipient
* may use your version of this file under either the MPL or the
* GPL.
*/
#include <stdlib.h>
#include <unistd.h>
#include <fcntl.h>
#include <string.h>
#include <time.h>
#include <errno.h>
#include "common.h"
#ifdef MD_HAVE_KQUEUE
#include <sys/event.h>
#endif
#ifdef MD_HAVE_EPOLL
#include <sys/epoll.h>
#endif
// Global stat.
#if defined(DEBUG) && defined(DEBUG_STATS)
__thread unsigned long long _st_stat_epoll = 0;
__thread unsigned long long _st_stat_epoll_zero = 0;
__thread unsigned long long _st_stat_epoll_shake = 0;
__thread unsigned long long _st_stat_epoll_spin = 0;
#endif
#if !defined(MD_HAVE_KQUEUE) && !defined(MD_HAVE_EPOLL) && !defined(MD_HAVE_SELECT)
#error Only support epoll(for Linux), kqueue(for Darwin) or select(for Cygwin)
#endif
#ifdef MD_HAVE_SELECT
static struct _st_seldata {
fd_set fd_read_set, fd_write_set, fd_exception_set;
int fd_ref_cnts[FD_SETSIZE][3];
int maxfd;
} *_st_select_data;
#define _ST_SELECT_MAX_OSFD (_st_select_data->maxfd)
#define _ST_SELECT_READ_SET (_st_select_data->fd_read_set)
#define _ST_SELECT_WRITE_SET (_st_select_data->fd_write_set)
#define _ST_SELECT_EXCEP_SET (_st_select_data->fd_exception_set)
#define _ST_SELECT_READ_CNT(fd) (_st_select_data->fd_ref_cnts[fd][0])
#define _ST_SELECT_WRITE_CNT(fd) (_st_select_data->fd_ref_cnts[fd][1])
#define _ST_SELECT_EXCEP_CNT(fd) (_st_select_data->fd_ref_cnts[fd][2])
#endif
#ifdef MD_HAVE_KQUEUE
typedef struct _kq_fd_data {
int rd_ref_cnt;
int wr_ref_cnt;
int revents;
} _kq_fd_data_t;
static __thread struct _st_kqdata {
_kq_fd_data_t *fd_data;
struct kevent *evtlist;
struct kevent *addlist;
struct kevent *dellist;
int fd_data_size;
int evtlist_size;
int addlist_size;
int addlist_cnt;
int dellist_size;
int dellist_cnt;
int kq;
pid_t pid;
} *_st_kq_data;
#ifndef ST_KQ_MIN_EVTLIST_SIZE
#define ST_KQ_MIN_EVTLIST_SIZE 64
#endif
#define _ST_KQ_READ_CNT(fd) (_st_kq_data->fd_data[fd].rd_ref_cnt)
#define _ST_KQ_WRITE_CNT(fd) (_st_kq_data->fd_data[fd].wr_ref_cnt)
#define _ST_KQ_REVENTS(fd) (_st_kq_data->fd_data[fd].revents)
#endif /* MD_HAVE_KQUEUE */
#ifdef MD_HAVE_EPOLL
typedef struct _epoll_fd_data {
int rd_ref_cnt;
int wr_ref_cnt;
int ex_ref_cnt;
int revents;
} _epoll_fd_data_t;
static __thread struct _st_epolldata {
_epoll_fd_data_t *fd_data;
struct epoll_event *evtlist;
int fd_data_size;
int evtlist_size;
int evtlist_cnt;
int fd_hint;
int epfd;
} *_st_epoll_data;
#ifndef ST_EPOLL_EVTLIST_SIZE
/* Not a limit, just a hint */
#define ST_EPOLL_EVTLIST_SIZE 4096
#endif
#define _ST_EPOLL_READ_CNT(fd) (_st_epoll_data->fd_data[fd].rd_ref_cnt)
#define _ST_EPOLL_WRITE_CNT(fd) (_st_epoll_data->fd_data[fd].wr_ref_cnt)
#define _ST_EPOLL_EXCEP_CNT(fd) (_st_epoll_data->fd_data[fd].ex_ref_cnt)
#define _ST_EPOLL_REVENTS(fd) (_st_epoll_data->fd_data[fd].revents)
#define _ST_EPOLL_READ_BIT(fd) (_ST_EPOLL_READ_CNT(fd) ? EPOLLIN : 0)
#define _ST_EPOLL_WRITE_BIT(fd) (_ST_EPOLL_WRITE_CNT(fd) ? EPOLLOUT : 0)
#define _ST_EPOLL_EXCEP_BIT(fd) (_ST_EPOLL_EXCEP_CNT(fd) ? EPOLLPRI : 0)
#define _ST_EPOLL_EVENTS(fd) \
(_ST_EPOLL_READ_BIT(fd)|_ST_EPOLL_WRITE_BIT(fd)|_ST_EPOLL_EXCEP_BIT(fd))
#endif /* MD_HAVE_EPOLL */
__thread _st_eventsys_t *_st_eventsys = NULL;
#ifdef MD_HAVE_SELECT
/*****************************************
* select event system
*/
ST_HIDDEN int _st_select_init(void)
{
_st_select_data = (struct _st_seldata *) malloc(sizeof(*_st_select_data));
if (!_st_select_data)
return -1;
memset(_st_select_data, 0, sizeof(*_st_select_data));
_st_select_data->maxfd = -1;
return 0;
}
ST_HIDDEN int _st_select_pollset_add(struct pollfd *pds, int npds)
{
struct pollfd *pd;
struct pollfd *epd = pds + npds;
/* Do checks up front */
for (pd = pds; pd < epd; pd++) {
if (pd->fd < 0 || pd->fd >= FD_SETSIZE || !pd->events ||
(pd->events & ~(POLLIN | POLLOUT | POLLPRI))) {
errno = EINVAL;
return -1;
}
}
for (pd = pds; pd < epd; pd++) {
if (pd->events & POLLIN) {
FD_SET(pd->fd, &_ST_SELECT_READ_SET);
_ST_SELECT_READ_CNT(pd->fd)++;
}
if (pd->events & POLLOUT) {
FD_SET(pd->fd, &_ST_SELECT_WRITE_SET);
_ST_SELECT_WRITE_CNT(pd->fd)++;
}
if (pd->events & POLLPRI) {
FD_SET(pd->fd, &_ST_SELECT_EXCEP_SET);
_ST_SELECT_EXCEP_CNT(pd->fd)++;
}
if (_ST_SELECT_MAX_OSFD < pd->fd)
_ST_SELECT_MAX_OSFD = pd->fd;
}
return 0;
}
ST_HIDDEN void _st_select_pollset_del(struct pollfd *pds, int npds)
{
struct pollfd *pd;
struct pollfd *epd = pds + npds;
for (pd = pds; pd < epd; pd++) {
if (pd->events & POLLIN) {
if (--_ST_SELECT_READ_CNT(pd->fd) == 0)
FD_CLR(pd->fd, &_ST_SELECT_READ_SET);
}
if (pd->events & POLLOUT) {
if (--_ST_SELECT_WRITE_CNT(pd->fd) == 0)
FD_CLR(pd->fd, &_ST_SELECT_WRITE_SET);
}
if (pd->events & POLLPRI) {
if (--_ST_SELECT_EXCEP_CNT(pd->fd) == 0)
FD_CLR(pd->fd, &_ST_SELECT_EXCEP_SET);
}
}
}
ST_HIDDEN void _st_select_find_bad_fd(void)
{
_st_clist_t *q;
_st_pollq_t *pq;
int notify;
struct pollfd *pds, *epds;
int pq_max_osfd, osfd;
short events;
_ST_SELECT_MAX_OSFD = -1;
for (q = _ST_IOQ.next; q != &_ST_IOQ; q = q->next) {
pq = _ST_POLLQUEUE_PTR(q);
notify = 0;
epds = pq->pds + pq->npds;
pq_max_osfd = -1;
for (pds = pq->pds; pds < epds; pds++) {
osfd = pds->fd;
pds->revents = 0;
if (pds->events == 0)
continue;
if (fcntl(osfd, F_GETFL, 0) < 0) {
pds->revents = POLLNVAL;
notify = 1;
}
if (osfd > pq_max_osfd) {
pq_max_osfd = osfd;
}
}
if (notify) {
ST_REMOVE_LINK(&pq->links);
pq->on_ioq = 0;
/*
* Decrement the count of descriptors for each descriptor/event
* because this I/O request is being removed from the ioq
*/
for (pds = pq->pds; pds < epds; pds++) {
osfd = pds->fd;
events = pds->events;
if (events & POLLIN) {
if (--_ST_SELECT_READ_CNT(osfd) == 0) {
FD_CLR(osfd, &_ST_SELECT_READ_SET);
}
}
if (events & POLLOUT) {
if (--_ST_SELECT_WRITE_CNT(osfd) == 0) {
FD_CLR(osfd, &_ST_SELECT_WRITE_SET);
}
}
if (events & POLLPRI) {
if (--_ST_SELECT_EXCEP_CNT(osfd) == 0) {
FD_CLR(osfd, &_ST_SELECT_EXCEP_SET);
}
}
}
if (pq->thread->flags & _ST_FL_ON_SLEEPQ)
_ST_DEL_SLEEPQ(pq->thread);
pq->thread->state = _ST_ST_RUNNABLE;
_ST_ADD_RUNQ(pq->thread);
} else {
if (_ST_SELECT_MAX_OSFD < pq_max_osfd)
_ST_SELECT_MAX_OSFD = pq_max_osfd;
}
}
}
ST_HIDDEN void _st_select_dispatch(void)
{
struct timeval timeout, *tvp;
fd_set r, w, e;
fd_set *rp, *wp, *ep;
int nfd, pq_max_osfd, osfd;
_st_clist_t *q;
st_utime_t min_timeout;
_st_pollq_t *pq;
int notify;
struct pollfd *pds, *epds;
short events, revents;
/*
* Assignment of fd_sets
*/
r = _ST_SELECT_READ_SET;
w = _ST_SELECT_WRITE_SET;
e = _ST_SELECT_EXCEP_SET;
rp = &r;
wp = &w;
ep = &e;
if (_ST_SLEEPQ == NULL) {
tvp = NULL;
} else {
min_timeout = (_ST_SLEEPQ->due <= _ST_LAST_CLOCK) ? 0 :
(_ST_SLEEPQ->due - _ST_LAST_CLOCK);
timeout.tv_sec = (int) (min_timeout / 1000000);
timeout.tv_usec = (int) (min_timeout % 1000000);
tvp = &timeout;
}
/* Check for I/O operations */
nfd = select(_ST_SELECT_MAX_OSFD + 1, rp, wp, ep, tvp);
/* Notify threads that are associated with the selected descriptors */
if (nfd > 0) {
_ST_SELECT_MAX_OSFD = -1;
for (q = _ST_IOQ.next; q != &_ST_IOQ; q = q->next) {
pq = _ST_POLLQUEUE_PTR(q);
notify = 0;
epds = pq->pds + pq->npds;
pq_max_osfd = -1;
for (pds = pq->pds; pds < epds; pds++) {
osfd = pds->fd;
events = pds->events;
revents = 0;
if ((events & POLLIN) && FD_ISSET(osfd, rp)) {
revents |= POLLIN;
}
if ((events & POLLOUT) && FD_ISSET(osfd, wp)) {
revents |= POLLOUT;
}
if ((events & POLLPRI) && FD_ISSET(osfd, ep)) {
revents |= POLLPRI;
}
pds->revents = revents;
if (revents) {
notify = 1;
}
if (osfd > pq_max_osfd) {
pq_max_osfd = osfd;
}
}
if (notify) {
ST_REMOVE_LINK(&pq->links);
pq->on_ioq = 0;
/*
* Decrement the count of descriptors for each descriptor/event
* because this I/O request is being removed from the ioq
*/
for (pds = pq->pds; pds < epds; pds++) {
osfd = pds->fd;
events = pds->events;
if (events & POLLIN) {
if (--_ST_SELECT_READ_CNT(osfd) == 0) {
FD_CLR(osfd, &_ST_SELECT_READ_SET);
}
}
if (events & POLLOUT) {
if (--_ST_SELECT_WRITE_CNT(osfd) == 0) {
FD_CLR(osfd, &_ST_SELECT_WRITE_SET);
}
}
if (events & POLLPRI) {
if (--_ST_SELECT_EXCEP_CNT(osfd) == 0) {
FD_CLR(osfd, &_ST_SELECT_EXCEP_SET);
}
}
}
if (pq->thread->flags & _ST_FL_ON_SLEEPQ)
_ST_DEL_SLEEPQ(pq->thread);
pq->thread->state = _ST_ST_RUNNABLE;
_ST_ADD_RUNQ(pq->thread);
} else {
if (_ST_SELECT_MAX_OSFD < pq_max_osfd)
_ST_SELECT_MAX_OSFD = pq_max_osfd;
}
}
} else if (nfd < 0) {
/*
* It can happen when a thread closes file descriptor
* that is being used by some other thread -- BAD!
*/
if (errno == EBADF)
_st_select_find_bad_fd();
}
}
ST_HIDDEN int _st_select_fd_new(int osfd)
{
if (osfd >= FD_SETSIZE) {
errno = EMFILE;
return -1;
}
return 0;
}
ST_HIDDEN int _st_select_fd_close(int osfd)
{
if (_ST_SELECT_READ_CNT(osfd) || _ST_SELECT_WRITE_CNT(osfd) ||
_ST_SELECT_EXCEP_CNT(osfd)) {
errno = EBUSY;
return -1;
}
return 0;
}
ST_HIDDEN int _st_select_fd_getlimit(void)
{
return FD_SETSIZE;
}
ST_HIDDEN void _st_select_destroy(void)
{
/* TODO: FIXME: Implements it */
}
static _st_eventsys_t _st_select_eventsys = {
"select",
ST_EVENTSYS_SELECT,
_st_select_init,
_st_select_dispatch,
_st_select_pollset_add,
_st_select_pollset_del,
_st_select_fd_new,
_st_select_fd_close,
_st_select_fd_getlimit,
_st_select_destroy
};
#endif
#ifdef MD_HAVE_KQUEUE
/*****************************************
* kqueue event system
*/
ST_HIDDEN int _st_kq_init(void)
{
int err = 0;
int rv = 0;
_st_kq_data = (struct _st_kqdata *) calloc(1, sizeof(*_st_kq_data));
if (!_st_kq_data)
return -1;
if ((_st_kq_data->kq = kqueue()) < 0) {
err = errno;
rv = -1;
goto cleanup_kq;
}
fcntl(_st_kq_data->kq, F_SETFD, FD_CLOEXEC);
_st_kq_data->pid = getpid();
/*
* Allocate file descriptor data array.
* FD_SETSIZE looks like good initial size.
*/
_st_kq_data->fd_data_size = FD_SETSIZE;
_st_kq_data->fd_data = (_kq_fd_data_t *)calloc(_st_kq_data->fd_data_size, sizeof(_kq_fd_data_t));
if (!_st_kq_data->fd_data) {
err = errno;
rv = -1;
goto cleanup_kq;
}
/* Allocate event lists */
_st_kq_data->evtlist_size = ST_KQ_MIN_EVTLIST_SIZE;
_st_kq_data->evtlist = (struct kevent *)malloc(_st_kq_data->evtlist_size * sizeof(struct kevent));
_st_kq_data->addlist_size = ST_KQ_MIN_EVTLIST_SIZE;
_st_kq_data->addlist = (struct kevent *)malloc(_st_kq_data->addlist_size * sizeof(struct kevent));
_st_kq_data->dellist_size = ST_KQ_MIN_EVTLIST_SIZE;
_st_kq_data->dellist = (struct kevent *)malloc(_st_kq_data->dellist_size * sizeof(struct kevent));
if (!_st_kq_data->evtlist || !_st_kq_data->addlist ||
!_st_kq_data->dellist) {
err = ENOMEM;
rv = -1;
}
cleanup_kq:
if (rv < 0) {
if (_st_kq_data->kq >= 0)
close(_st_kq_data->kq);
free(_st_kq_data->fd_data);
free(_st_kq_data->evtlist);
free(_st_kq_data->addlist);
free(_st_kq_data->dellist);
free(_st_kq_data);
_st_kq_data = NULL;
errno = err;
}
return rv;
}
ST_HIDDEN int _st_kq_fd_data_expand(int maxfd)
{
_kq_fd_data_t *ptr;
int n = _st_kq_data->fd_data_size;
while (maxfd >= n)
n <<= 1;
ptr = (_kq_fd_data_t *)realloc(_st_kq_data->fd_data, n * sizeof(_kq_fd_data_t));
if (!ptr)
return -1;
memset(ptr + _st_kq_data->fd_data_size, 0, (n - _st_kq_data->fd_data_size) * sizeof(_kq_fd_data_t));
_st_kq_data->fd_data = ptr;
_st_kq_data->fd_data_size = n;
return 0;
}
ST_HIDDEN int _st_kq_addlist_expand(int avail)
{
struct kevent *ptr;
int n = _st_kq_data->addlist_size;
while (avail > n - _st_kq_data->addlist_cnt)
n <<= 1;
ptr = (struct kevent *)realloc(_st_kq_data->addlist, n * sizeof(struct kevent));
if (!ptr)
return -1;
_st_kq_data->addlist = ptr;
_st_kq_data->addlist_size = n;
/*
* Try to expand the result event list too
* (although we don't have to do it).
*/
ptr = (struct kevent *)realloc(_st_kq_data->evtlist, n * sizeof(struct kevent));
if (ptr) {
_st_kq_data->evtlist = ptr;
_st_kq_data->evtlist_size = n;
}
return 0;
}
ST_HIDDEN void _st_kq_addlist_add(const struct kevent *kev)
{
ST_ASSERT(_st_kq_data->addlist_cnt < _st_kq_data->addlist_size);
memcpy(_st_kq_data->addlist + _st_kq_data->addlist_cnt, kev, sizeof(struct kevent));
_st_kq_data->addlist_cnt++;
}
ST_HIDDEN void _st_kq_dellist_add(const struct kevent *kev)
{
int n = _st_kq_data->dellist_size;
if (_st_kq_data->dellist_cnt >= n) {
struct kevent *ptr;
n <<= 1;
ptr = (struct kevent *)realloc(_st_kq_data->dellist, n * sizeof(struct kevent));
if (!ptr) {
/* See comment in _st_kq_pollset_del() */
return;
}
_st_kq_data->dellist = ptr;
_st_kq_data->dellist_size = n;
}
memcpy(_st_kq_data->dellist + _st_kq_data->dellist_cnt, kev, sizeof(struct kevent));
_st_kq_data->dellist_cnt++;
}
ST_HIDDEN int _st_kq_pollset_add(struct pollfd *pds, int npds)
{
struct kevent kev;
struct pollfd *pd;
struct pollfd *epd = pds + npds;
/*
* Pollset adding is "atomic". That is, either it succeeded for
* all descriptors in the set or it failed. It means that we
* need to do all the checks up front so we don't have to
* "unwind" if adding of one of the descriptors failed.
*/
for (pd = pds; pd < epd; pd++) {
/* POLLIN and/or POLLOUT must be set, but nothing else */
if (pd->fd < 0 || !pd->events || (pd->events & ~(POLLIN | POLLOUT))) {
errno = EINVAL;
return -1;
}
if (pd->fd >= _st_kq_data->fd_data_size &&
_st_kq_fd_data_expand(pd->fd) < 0)
return -1;
}
/*
* Make sure we have enough room in the addlist for twice as many
* descriptors as in the pollset (for both READ and WRITE filters).
*/
npds <<= 1;
if (npds > _st_kq_data->addlist_size - _st_kq_data->addlist_cnt && _st_kq_addlist_expand(npds) < 0)
return -1;
for (pd = pds; pd < epd; pd++) {
if ((pd->events & POLLIN) && (_ST_KQ_READ_CNT(pd->fd)++ == 0)) {
memset(&kev, 0, sizeof(kev));
kev.ident = pd->fd;
kev.filter = EVFILT_READ;
#ifdef NOTE_EOF
/* Make it behave like select() and poll() */
kev.fflags = NOTE_EOF;
#endif
kev.flags = (EV_ADD | EV_ONESHOT);
_st_kq_addlist_add(&kev);
}
if ((pd->events & POLLOUT) && (_ST_KQ_WRITE_CNT(pd->fd)++ == 0)) {
memset(&kev, 0, sizeof(kev));
kev.ident = pd->fd;
kev.filter = EVFILT_WRITE;
kev.flags = (EV_ADD | EV_ONESHOT);
_st_kq_addlist_add(&kev);
}
}
return 0;
}
ST_HIDDEN void _st_kq_pollset_del(struct pollfd *pds, int npds)
{
struct kevent kev;
struct pollfd *pd;
struct pollfd *epd = pds + npds;
/*
* It's OK if deleting fails because a descriptor will either be
* closed or fire only once (we set EV_ONESHOT flag).
*/
_st_kq_data->dellist_cnt = 0;
for (pd = pds; pd < epd; pd++) {
if ((pd->events & POLLIN) && (--_ST_KQ_READ_CNT(pd->fd) == 0)) {
memset(&kev, 0, sizeof(kev));
kev.ident = pd->fd;
kev.filter = EVFILT_READ;
kev.flags = EV_DELETE;
_st_kq_dellist_add(&kev);
}
if ((pd->events & POLLOUT) && (--_ST_KQ_WRITE_CNT(pd->fd) == 0)) {
memset(&kev, 0, sizeof(kev));
kev.ident = pd->fd;
kev.filter = EVFILT_WRITE;
kev.flags = EV_DELETE;
_st_kq_dellist_add(&kev);
}
}
if (_st_kq_data->dellist_cnt > 0) {
/*
* We do "synchronous" kqueue deletes to avoid deleting
* closed descriptors and other possible problems.
*/
int rv;
do {
/* This kevent() won't block since result list size is 0 */
rv = kevent(_st_kq_data->kq, _st_kq_data->dellist, _st_kq_data->dellist_cnt, NULL, 0, NULL);
} while (rv < 0 && errno == EINTR);
}
}
ST_HIDDEN void _st_kq_dispatch(void)
{
struct timespec timeout, *tsp;
struct kevent kev;
st_utime_t min_timeout;
_st_clist_t *q;
_st_pollq_t *pq;
struct pollfd *pds, *epds;
int nfd, i, osfd, notify, filter;
short events, revents;
if (_ST_SLEEPQ == NULL) {
tsp = NULL;
} else {
min_timeout = (_ST_SLEEPQ->due <= _ST_LAST_CLOCK) ? 0 : (_ST_SLEEPQ->due - _ST_LAST_CLOCK);
timeout.tv_sec = (time_t) (min_timeout / 1000000);
timeout.tv_nsec = (long) ((min_timeout % 1000000) * 1000);
tsp = &timeout;
}
retry_kevent:
/* Check for I/O operations */
nfd = kevent(_st_kq_data->kq,
_st_kq_data->addlist, _st_kq_data->addlist_cnt,
_st_kq_data->evtlist, _st_kq_data->evtlist_size, tsp);
_st_kq_data->addlist_cnt = 0;
if (nfd > 0) {
for (i = 0; i < nfd; i++) {
osfd = _st_kq_data->evtlist[i].ident;
filter = _st_kq_data->evtlist[i].filter;
if (filter == EVFILT_READ) {
_ST_KQ_REVENTS(osfd) |= POLLIN;
} else if (filter == EVFILT_WRITE) {
_ST_KQ_REVENTS(osfd) |= POLLOUT;
}
if (_st_kq_data->evtlist[i].flags & EV_ERROR) {
if (_st_kq_data->evtlist[i].data == EBADF) {
_ST_KQ_REVENTS(osfd) |= POLLNVAL;
} else {
_ST_KQ_REVENTS(osfd) |= POLLERR;
}
}
}
_st_kq_data->dellist_cnt = 0;
for (q = _ST_IOQ.next; q != &_ST_IOQ; q = q->next) {
pq = _ST_POLLQUEUE_PTR(q);
notify = 0;
epds = pq->pds + pq->npds;
for (pds = pq->pds; pds < epds; pds++) {
osfd = pds->fd;
events = pds->events;
revents = (short)(_ST_KQ_REVENTS(osfd) & ~(POLLIN | POLLOUT));
if ((events & POLLIN) && (_ST_KQ_REVENTS(osfd) & POLLIN)) {
revents |= POLLIN;
}
if ((events & POLLOUT) && (_ST_KQ_REVENTS(osfd) & POLLOUT)) {
revents |= POLLOUT;
}
pds->revents = revents;
if (revents) {
notify = 1;
}
}
if (notify) {
ST_REMOVE_LINK(&pq->links);
pq->on_ioq = 0;
for (pds = pq->pds; pds < epds; pds++) {
osfd = pds->fd;
events = pds->events;
/*
* We set EV_ONESHOT flag so we only need to delete
* descriptor if it didn't fire.
*/
if ((events & POLLIN) && (--_ST_KQ_READ_CNT(osfd) == 0) && ((_ST_KQ_REVENTS(osfd) & POLLIN) == 0)) {
memset(&kev, 0, sizeof(kev));
kev.ident = osfd;
kev.filter = EVFILT_READ;
kev.flags = EV_DELETE;
_st_kq_dellist_add(&kev);
}
if ((events & POLLOUT) && (--_ST_KQ_WRITE_CNT(osfd) == 0) && ((_ST_KQ_REVENTS(osfd) & POLLOUT) == 0)) {
memset(&kev, 0, sizeof(kev));
kev.ident = osfd;
kev.filter = EVFILT_WRITE;
kev.flags = EV_DELETE;
_st_kq_dellist_add(&kev);
}
}
if (pq->thread->flags & _ST_FL_ON_SLEEPQ)
_ST_DEL_SLEEPQ(pq->thread);
pq->thread->state = _ST_ST_RUNNABLE;
_ST_ADD_RUNQ(pq->thread);
}
}
if (_st_kq_data->dellist_cnt > 0) {
int rv;
do {
/* This kevent() won't block since result list size is 0 */
rv = kevent(_st_kq_data->kq, _st_kq_data->dellist, _st_kq_data->dellist_cnt, NULL, 0, NULL);
} while (rv < 0 && errno == EINTR);
}
for (i = 0; i < nfd; i++) {
osfd = _st_kq_data->evtlist[i].ident;
_ST_KQ_REVENTS(osfd) = 0;
}
} else if (nfd < 0) {
if (errno == EBADF && _st_kq_data->pid != getpid()) {
/* We probably forked, reinitialize kqueue */
if ((_st_kq_data->kq = kqueue()) < 0) {
/* There is nothing we can do here, will retry later */
return;
}
fcntl(_st_kq_data->kq, F_SETFD, FD_CLOEXEC);
_st_kq_data->pid = getpid();
/* Re-register all descriptors on ioq with new kqueue */
memset(_st_kq_data->fd_data, 0, _st_kq_data->fd_data_size * sizeof(_kq_fd_data_t));
for (q = _ST_IOQ.next; q != &_ST_IOQ; q = q->next) {
pq = _ST_POLLQUEUE_PTR(q);
_st_kq_pollset_add(pq->pds, pq->npds);
}
goto retry_kevent;
}
}
}
ST_HIDDEN int _st_kq_fd_new(int osfd)
{
if (osfd >= _st_kq_data->fd_data_size && _st_kq_fd_data_expand(osfd) < 0)
return -1;
return 0;
}
ST_HIDDEN int _st_kq_fd_close(int osfd)
{
if (_ST_KQ_READ_CNT(osfd) || _ST_KQ_WRITE_CNT(osfd)) {
errno = EBUSY;
return -1;
}
return 0;
}
ST_HIDDEN int _st_kq_fd_getlimit(void)
{
/* zero means no specific limit */
return 0;
}
ST_HIDDEN void _st_kq_destroy(void)
{
/* TODO: FIXME: Implements it */
}
static _st_eventsys_t _st_kq_eventsys = {
"kqueue",
ST_EVENTSYS_ALT,
_st_kq_init,
_st_kq_dispatch,
_st_kq_pollset_add,
_st_kq_pollset_del,
_st_kq_fd_new,
_st_kq_fd_close,
_st_kq_fd_getlimit,
_st_kq_destroy
};
#endif /* MD_HAVE_KQUEUE */
#ifdef MD_HAVE_EPOLL
/*****************************************
* epoll event system
*/
ST_HIDDEN int _st_epoll_init(void)
{
int fdlim;
int err = 0;
int rv = 0;
_st_epoll_data = (struct _st_epolldata *) calloc(1, sizeof(*_st_epoll_data));
if (!_st_epoll_data)
return -1;
fdlim = st_getfdlimit();
_st_epoll_data->fd_hint = (fdlim > 0 && fdlim < ST_EPOLL_EVTLIST_SIZE) ? fdlim : ST_EPOLL_EVTLIST_SIZE;
if ((_st_epoll_data->epfd = epoll_create(_st_epoll_data->fd_hint)) < 0) {
err = errno;
rv = -1;
goto cleanup_epoll;
}
fcntl(_st_epoll_data->epfd, F_SETFD, FD_CLOEXEC);
/* Allocate file descriptor data array */
_st_epoll_data->fd_data_size = _st_epoll_data->fd_hint;
_st_epoll_data->fd_data = (_epoll_fd_data_t *)calloc(_st_epoll_data->fd_data_size, sizeof(_epoll_fd_data_t));
if (!_st_epoll_data->fd_data) {
err = errno;
rv = -1;
goto cleanup_epoll;
}
/* Allocate event lists */
_st_epoll_data->evtlist_size = _st_epoll_data->fd_hint;
_st_epoll_data->evtlist = (struct epoll_event *)malloc(_st_epoll_data->evtlist_size * sizeof(struct epoll_event));
if (!_st_epoll_data->evtlist) {
err = errno;
rv = -1;
}
cleanup_epoll:
if (rv < 0) {
if (_st_epoll_data->epfd >= 0)
close(_st_epoll_data->epfd);
free(_st_epoll_data->fd_data);
free(_st_epoll_data->evtlist);
free(_st_epoll_data);
_st_epoll_data = NULL;
errno = err;
}
return rv;
}
ST_HIDDEN int _st_epoll_fd_data_expand(int maxfd)
{
_epoll_fd_data_t *ptr;
int n = _st_epoll_data->fd_data_size;
while (maxfd >= n)
n <<= 1;
ptr = (_epoll_fd_data_t *)realloc(_st_epoll_data->fd_data, n * sizeof(_epoll_fd_data_t));
if (!ptr)
return -1;
memset(ptr + _st_epoll_data->fd_data_size, 0, (n - _st_epoll_data->fd_data_size) * sizeof(_epoll_fd_data_t));
_st_epoll_data->fd_data = ptr;
_st_epoll_data->fd_data_size = n;
return 0;
}
ST_HIDDEN void _st_epoll_evtlist_expand(void)
{
struct epoll_event *ptr;
int n = _st_epoll_data->evtlist_size;
while (_st_epoll_data->evtlist_cnt > n)
n <<= 1;
ptr = (struct epoll_event *)realloc(_st_epoll_data->evtlist, n * sizeof(struct epoll_event));
if (ptr) {
_st_epoll_data->evtlist = ptr;
_st_epoll_data->evtlist_size = n;
}
}
ST_HIDDEN void _st_epoll_pollset_del(struct pollfd *pds, int npds)
{
struct epoll_event ev;
struct pollfd *pd;
struct pollfd *epd = pds + npds;
int old_events, events, op;
/*
* It's more or less OK if deleting fails because a descriptor
* will either be closed or deleted in dispatch function after
* it fires.
*/
for (pd = pds; pd < epd; pd++) {
old_events = _ST_EPOLL_EVENTS(pd->fd);
if (pd->events & POLLIN)
_ST_EPOLL_READ_CNT(pd->fd)--;
if (pd->events & POLLOUT)
_ST_EPOLL_WRITE_CNT(pd->fd)--;
if (pd->events & POLLPRI)
_ST_EPOLL_EXCEP_CNT(pd->fd)--;
events = _ST_EPOLL_EVENTS(pd->fd);
/*
* The _ST_EPOLL_REVENTS check below is needed so we can use
* this function inside dispatch(). Outside of dispatch()
* _ST_EPOLL_REVENTS is always zero for all descriptors.
*/
if (events != old_events && _ST_EPOLL_REVENTS(pd->fd) == 0) {
op = events ? EPOLL_CTL_MOD : EPOLL_CTL_DEL;
ev.events = events;
ev.data.fd = pd->fd;
if (epoll_ctl(_st_epoll_data->epfd, op, pd->fd, &ev) == 0 && op == EPOLL_CTL_DEL) {
_st_epoll_data->evtlist_cnt--;
}
}
}
}
ST_HIDDEN int _st_epoll_pollset_add(struct pollfd *pds, int npds)
{
struct epoll_event ev;
int i, fd;
int old_events, events, op;