summaryrefslogtreecommitdiffstats
path: root/ipc/chromium/src/third_party/libevent/kqueue.c
blob: e5a530f80df0745fac057d79dee481035e98176d (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
/*	$OpenBSD: kqueue.c,v 1.5 2002/07/10 14:41:31 art Exp $	*/

/*
 * Copyright 2000-2007 Niels Provos <provos@citi.umich.edu>
 * Copyright 2007-2012 Niels Provos and Nick Mathewson
 *
 * 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. The name of the author may not be used to endorse or promote products
 *    derived from this software without specific prior written permission.
 *
 * 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 "event2/event-config.h"

#define _GNU_SOURCE

#include <sys/types.h>
#ifdef _EVENT_HAVE_SYS_TIME_H
#include <sys/time.h>
#endif
#include <sys/queue.h>
#include <sys/event.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <errno.h>
#ifdef _EVENT_HAVE_INTTYPES_H
#include <inttypes.h>
#endif

/* Some platforms apparently define the udata field of struct kevent as
 * intptr_t, whereas others define it as void*.  There doesn't seem to be an
 * easy way to tell them apart via autoconf, so we need to use OS macros. */
#if defined(_EVENT_HAVE_INTTYPES_H) && !defined(__OpenBSD__) && !defined(__FreeBSD__) && !defined(__darwin__) && !defined(__APPLE__)
#define PTR_TO_UDATA(x)	((intptr_t)(x))
#define INT_TO_UDATA(x) ((intptr_t)(x))
#else
#define PTR_TO_UDATA(x)	(x)
#define INT_TO_UDATA(x) ((void*)(x))
#endif

#include "event-internal.h"
#include "log-internal.h"
#include "evmap-internal.h"
#include "event2/thread.h"
#include "evthread-internal.h"
#include "changelist-internal.h"

#define NEVENT		64

struct kqop {
	struct kevent *changes;
	int changes_size;

	struct kevent *events;
	int events_size;
	int kq;
	pid_t pid;
};

static void kqop_free(struct kqop *kqop);

static void *kq_init(struct event_base *);
static int kq_sig_add(struct event_base *, int, short, short, void *);
static int kq_sig_del(struct event_base *, int, short, short, void *);
static int kq_dispatch(struct event_base *, struct timeval *);
static void kq_dealloc(struct event_base *);

const struct eventop kqops = {
	"kqueue",
	kq_init,
	event_changelist_add,
	event_changelist_del,
	kq_dispatch,
	kq_dealloc,
	1 /* need reinit */,
    EV_FEATURE_ET|EV_FEATURE_O1|EV_FEATURE_FDS,
	EVENT_CHANGELIST_FDINFO_SIZE
};

static const struct eventop kqsigops = {
	"kqueue_signal",
	NULL,
	kq_sig_add,
	kq_sig_del,
	NULL,
	NULL,
	1 /* need reinit */,
	0,
	0
};

static void *
kq_init(struct event_base *base)
{
	int kq = -1;
	struct kqop *kqueueop = NULL;

	if (!(kqueueop = mm_calloc(1, sizeof(struct kqop))))
		return (NULL);

/* Initialize the kernel queue */

	if ((kq = kqueue()) == -1) {
		event_warn("kqueue");
		goto err;
	}

	kqueueop->kq = kq;

	kqueueop->pid = getpid();

	/* Initialize fields */
	kqueueop->changes = mm_calloc(NEVENT, sizeof(struct kevent));
	if (kqueueop->changes == NULL)
		goto err;
	kqueueop->events = mm_calloc(NEVENT, sizeof(struct kevent));
	if (kqueueop->events == NULL)
		goto err;
	kqueueop->events_size = kqueueop->changes_size = NEVENT;

	base->evsigsel = &kqsigops;

	return (kqueueop);
err:
	if (kqueueop)
		kqop_free(kqueueop);

	return (NULL);
}

#define ADD_UDATA 0x30303

static void
kq_setup_kevent(struct kevent *out, evutil_socket_t fd, int filter, short change)
{
	memset(out, 0, sizeof(struct kevent));
	out->ident = fd;
	out->filter = filter;

	if (change & EV_CHANGE_ADD) {
		out->flags = EV_ADD;
		/* We set a magic number here so that we can tell 'add'
		 * errors from 'del' errors. */
		out->udata = INT_TO_UDATA(ADD_UDATA);
		if (change & EV_ET)
			out->flags |= EV_CLEAR;
#ifdef NOTE_EOF
		/* Make it behave like select() and poll() */
		if (filter == EVFILT_READ)
			out->fflags = NOTE_EOF;
#endif
	} else {
		EVUTIL_ASSERT(change & EV_CHANGE_DEL);
		out->flags = EV_DELETE;
	}
}

static int
kq_build_changes_list(const struct event_changelist *changelist,
    struct kqop *kqop)
{
	int i;
	int n_changes = 0;

	for (i = 0; i < changelist->n_changes; ++i) {
		struct event_change *in_ch = &changelist->changes[i];
		struct kevent *out_ch;
		if (n_changes >= kqop->changes_size - 1) {
			int newsize = kqop->changes_size * 2;
			struct kevent *newchanges;

			newchanges = mm_realloc(kqop->changes,
			    newsize * sizeof(struct kevent));
			if (newchanges == NULL) {
				event_warn("%s: realloc", __func__);
				return (-1);
			}
			kqop->changes = newchanges;
			kqop->changes_size = newsize;
		}
		if (in_ch->read_change) {
			out_ch = &kqop->changes[n_changes++];
			kq_setup_kevent(out_ch, in_ch->fd, EVFILT_READ,
			    in_ch->read_change);
		}
		if (in_ch->write_change) {
			out_ch = &kqop->changes[n_changes++];
			kq_setup_kevent(out_ch, in_ch->fd, EVFILT_WRITE,
			    in_ch->write_change);
		}
	}
	return n_changes;
}

static int
kq_grow_events(struct kqop *kqop, size_t new_size)
{
	struct kevent *newresult;

	newresult = mm_realloc(kqop->events,
	    new_size * sizeof(struct kevent));

	if (newresult) {
		kqop->events = newresult;
		kqop->events_size = new_size;
		return 0;
	} else {
		return -1;
	}
}

static int
kq_dispatch(struct event_base *base, struct timeval *tv)
{
	struct kqop *kqop = base->evbase;
	struct kevent *events = kqop->events;
	struct kevent *changes;
	struct timespec ts, *ts_p = NULL;
	int i, n_changes, res;

	if (tv != NULL) {
		TIMEVAL_TO_TIMESPEC(tv, &ts);
		ts_p = &ts;
	}

	/* Build "changes" from "base->changes" */
	EVUTIL_ASSERT(kqop->changes);
	n_changes = kq_build_changes_list(&base->changelist, kqop);
	if (n_changes < 0)
		return -1;

	event_changelist_remove_all(&base->changelist, base);

	/* steal the changes array in case some broken code tries to call
	 * dispatch twice at once. */
	changes = kqop->changes;
	kqop->changes = NULL;

	/* Make sure that 'events' is at least as long as the list of changes:
	 * otherwise errors in the changes can get reported as a -1 return
	 * value from kevent() rather than as EV_ERROR events in the events
	 * array.
	 *
	 * (We could instead handle -1 return values from kevent() by
	 * retrying with a smaller changes array or a larger events array,
	 * but this approach seems less risky for now.)
	 */
	if (kqop->events_size < n_changes) {
		int new_size = kqop->events_size;
		do {
			new_size *= 2;
		} while (new_size < n_changes);

		kq_grow_events(kqop, new_size);
		events = kqop->events;
	}

	EVBASE_RELEASE_LOCK(base, th_base_lock);

	res = kevent(kqop->kq, changes, n_changes,
	    events, kqop->events_size, ts_p);

	EVBASE_ACQUIRE_LOCK(base, th_base_lock);

	EVUTIL_ASSERT(kqop->changes == NULL);
	kqop->changes = changes;

	if (res == -1) {
		if (errno != EINTR) {
			event_warn("kevent");
			return (-1);
		}

		return (0);
	}

	event_debug(("%s: kevent reports %d", __func__, res));

	for (i = 0; i < res; i++) {
		int which = 0;

		if (events[i].flags & EV_ERROR) {
			switch (events[i].data) {

			/* Can occur on delete if we are not currently
			 * watching any events on this fd.  That can
			 * happen when the fd was closed and another
			 * file was opened with that fd. */
			case ENOENT:
			/* Can occur for reasons not fully understood
			 * on FreeBSD. */
			case EINVAL:
				continue;

			/* Can occur on a delete if the fd is closed. */
			case EBADF:
				/* XXXX On NetBSD, we can also get EBADF if we
				 * try to add the write side of a pipe, but
				 * the read side has already been closed.
				 * Other BSDs call this situation 'EPIPE'. It
				 * would be good if we had a way to report
				 * this situation. */
				continue;
			/* These two can occur on an add if the fd was one side
			 * of a pipe, and the other side was closed. */
			case EPERM:
			case EPIPE:
				/* Report read events, if we're listening for
				 * them, so that the user can learn about any
				 * add errors.  (If the operation was a
				 * delete, then udata should be cleared.) */
				if (events[i].udata) {
					/* The operation was an add:
					 * report the error as a read. */
					which |= EV_READ;
					break;
				} else {
					/* The operation was a del:
					 * report nothing. */
					continue;
				}

			/* Other errors shouldn't occur. */
			default:
				errno = events[i].data;
				return (-1);
			}
		} else if (events[i].filter == EVFILT_READ) {
			which |= EV_READ;
		} else if (events[i].filter == EVFILT_WRITE) {
			which |= EV_WRITE;
		} else if (events[i].filter == EVFILT_SIGNAL) {
			which |= EV_SIGNAL;
		}

		if (!which)
			continue;

		if (events[i].filter == EVFILT_SIGNAL) {
			evmap_signal_active(base, events[i].ident, 1);
		} else {
			evmap_io_active(base, events[i].ident, which | EV_ET);
		}
	}

	if (res == kqop->events_size) {
		/* We used all the events space that we have. Maybe we should
		   make it bigger. */
		kq_grow_events(kqop, kqop->events_size * 2);
	}

	return (0);
}

static void
kqop_free(struct kqop *kqop)
{
	if (kqop->changes)
		mm_free(kqop->changes);
	if (kqop->events)
		mm_free(kqop->events);
	if (kqop->kq >= 0 && kqop->pid == getpid())
		close(kqop->kq);
	memset(kqop, 0, sizeof(struct kqop));
	mm_free(kqop);
}

static void
kq_dealloc(struct event_base *base)
{
	struct kqop *kqop = base->evbase;
	evsig_dealloc(base);
	kqop_free(kqop);
}

/* signal handling */
static int
kq_sig_add(struct event_base *base, int nsignal, short old, short events, void *p)
{
	struct kqop *kqop = base->evbase;
	struct kevent kev;
	struct timespec timeout = { 0, 0 };
	(void)p;

	EVUTIL_ASSERT(nsignal >= 0 && nsignal < NSIG);

	memset(&kev, 0, sizeof(kev));
	kev.ident = nsignal;
	kev.filter = EVFILT_SIGNAL;
	kev.flags = EV_ADD;

	/* Be ready for the signal if it is sent any
	 * time between now and the next call to
	 * kq_dispatch. */
	if (kevent(kqop->kq, &kev, 1, NULL, 0, &timeout) == -1)
		return (-1);

	/* Backported from
	 * https://github.com/nmathewson/Libevent/commit/148458e0a1fd25e167aa2ef229d1c9a70b27c3e9 */
	/* We can set the handler for most signals to SIG_IGN and
	 * still have them reported to us in the queue.  However,
	 * if the handler for SIGCHLD is SIG_IGN, the system reaps
	 * zombie processes for us, and we don't get any notification.
	 * This appears to be the only signal with this quirk. */
	if (_evsig_set_handler(base, nsignal,
	                       nsignal == SIGCHLD ? SIG_DFL : SIG_IGN) == -1) {
		return (-1);
	}

	return (0);
}

static int
kq_sig_del(struct event_base *base, int nsignal, short old, short events, void *p)
{
	struct kqop *kqop = base->evbase;
	struct kevent kev;

	struct timespec timeout = { 0, 0 };
	(void)p;

	EVUTIL_ASSERT(nsignal >= 0 && nsignal < NSIG);

	memset(&kev, 0, sizeof(kev));
	kev.ident = nsignal;
	kev.filter = EVFILT_SIGNAL;
	kev.flags = EV_DELETE;

	/* Because we insert signal events
	 * immediately, we need to delete them
	 * immediately, too */
	if (kevent(kqop->kq, &kev, 1, NULL, 0, &timeout) == -1)
		return (-1);

	if (_evsig_restore_handler(base, nsignal) == -1)
		return (-1);

	return (0);
}