xref: /freebsd/sys/kern/kern_event.c (revision a134ebd6e63f658f2d3d04ac0c60d23bcaa86dd7)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
3  *
4  * Copyright (c) 1999,2000,2001 Jonathan Lemon <jlemon@FreeBSD.org>
5  * Copyright 2004 John-Mark Gurney <jmg@FreeBSD.org>
6  * Copyright (c) 2009 Apple, Inc.
7  * All rights reserved.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  *
18  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
19  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
22  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28  * SUCH DAMAGE.
29  */
30 
31 #include <sys/cdefs.h>
32 __FBSDID("$FreeBSD$");
33 
34 #include "opt_ktrace.h"
35 #include "opt_kqueue.h"
36 
37 #ifdef COMPAT_FREEBSD11
38 #define	_WANT_FREEBSD11_KEVENT
39 #endif
40 
41 #include <sys/param.h>
42 #include <sys/systm.h>
43 #include <sys/capsicum.h>
44 #include <sys/kernel.h>
45 #include <sys/limits.h>
46 #include <sys/lock.h>
47 #include <sys/mutex.h>
48 #include <sys/rwlock.h>
49 #include <sys/proc.h>
50 #include <sys/malloc.h>
51 #include <sys/unistd.h>
52 #include <sys/file.h>
53 #include <sys/filedesc.h>
54 #include <sys/filio.h>
55 #include <sys/fcntl.h>
56 #include <sys/kthread.h>
57 #include <sys/selinfo.h>
58 #include <sys/queue.h>
59 #include <sys/event.h>
60 #include <sys/eventvar.h>
61 #include <sys/poll.h>
62 #include <sys/protosw.h>
63 #include <sys/resourcevar.h>
64 #include <sys/sigio.h>
65 #include <sys/signalvar.h>
66 #include <sys/socket.h>
67 #include <sys/socketvar.h>
68 #include <sys/stat.h>
69 #include <sys/sysctl.h>
70 #include <sys/sysproto.h>
71 #include <sys/syscallsubr.h>
72 #include <sys/taskqueue.h>
73 #include <sys/uio.h>
74 #include <sys/user.h>
75 #ifdef KTRACE
76 #include <sys/ktrace.h>
77 #endif
78 #include <machine/atomic.h>
79 
80 #include <vm/uma.h>
81 
82 static MALLOC_DEFINE(M_KQUEUE, "kqueue", "memory for kqueue system");
83 
84 /*
85  * This lock is used if multiple kq locks are required.  This possibly
86  * should be made into a per proc lock.
87  */
88 static struct mtx	kq_global;
89 MTX_SYSINIT(kq_global, &kq_global, "kqueue order", MTX_DEF);
90 #define KQ_GLOBAL_LOCK(lck, haslck)	do {	\
91 	if (!haslck)				\
92 		mtx_lock(lck);			\
93 	haslck = 1;				\
94 } while (0)
95 #define KQ_GLOBAL_UNLOCK(lck, haslck)	do {	\
96 	if (haslck)				\
97 		mtx_unlock(lck);			\
98 	haslck = 0;				\
99 } while (0)
100 
101 TASKQUEUE_DEFINE_THREAD(kqueue_ctx);
102 
103 static int	kevent_copyout(void *arg, struct kevent *kevp, int count);
104 static int	kevent_copyin(void *arg, struct kevent *kevp, int count);
105 static int	kqueue_register(struct kqueue *kq, struct kevent *kev,
106 		    struct thread *td, int mflag);
107 static int	kqueue_acquire(struct file *fp, struct kqueue **kqp);
108 static void	kqueue_release(struct kqueue *kq, int locked);
109 static void	kqueue_destroy(struct kqueue *kq);
110 static void	kqueue_drain(struct kqueue *kq, struct thread *td);
111 static int	kqueue_expand(struct kqueue *kq, struct filterops *fops,
112 		    uintptr_t ident, int mflag);
113 static void	kqueue_task(void *arg, int pending);
114 static int	kqueue_scan(struct kqueue *kq, int maxevents,
115 		    struct kevent_copyops *k_ops,
116 		    const struct timespec *timeout,
117 		    struct kevent *keva, struct thread *td);
118 static void 	kqueue_wakeup(struct kqueue *kq);
119 static struct filterops *kqueue_fo_find(int filt);
120 static void	kqueue_fo_release(int filt);
121 struct g_kevent_args;
122 static int	kern_kevent_generic(struct thread *td,
123 		    struct g_kevent_args *uap,
124 		    struct kevent_copyops *k_ops, const char *struct_name);
125 
126 static fo_ioctl_t	kqueue_ioctl;
127 static fo_poll_t	kqueue_poll;
128 static fo_kqfilter_t	kqueue_kqfilter;
129 static fo_stat_t	kqueue_stat;
130 static fo_close_t	kqueue_close;
131 static fo_fill_kinfo_t	kqueue_fill_kinfo;
132 
133 static struct fileops kqueueops = {
134 	.fo_read = invfo_rdwr,
135 	.fo_write = invfo_rdwr,
136 	.fo_truncate = invfo_truncate,
137 	.fo_ioctl = kqueue_ioctl,
138 	.fo_poll = kqueue_poll,
139 	.fo_kqfilter = kqueue_kqfilter,
140 	.fo_stat = kqueue_stat,
141 	.fo_close = kqueue_close,
142 	.fo_chmod = invfo_chmod,
143 	.fo_chown = invfo_chown,
144 	.fo_sendfile = invfo_sendfile,
145 	.fo_fill_kinfo = kqueue_fill_kinfo,
146 };
147 
148 static int 	knote_attach(struct knote *kn, struct kqueue *kq);
149 static void 	knote_drop(struct knote *kn, struct thread *td);
150 static void 	knote_drop_detached(struct knote *kn, struct thread *td);
151 static void 	knote_enqueue(struct knote *kn);
152 static void 	knote_dequeue(struct knote *kn);
153 static void 	knote_init(void);
154 static struct 	knote *knote_alloc(int mflag);
155 static void 	knote_free(struct knote *kn);
156 
157 static void	filt_kqdetach(struct knote *kn);
158 static int	filt_kqueue(struct knote *kn, long hint);
159 static int	filt_procattach(struct knote *kn);
160 static void	filt_procdetach(struct knote *kn);
161 static int	filt_proc(struct knote *kn, long hint);
162 static int	filt_fileattach(struct knote *kn);
163 static void	filt_timerexpire(void *knx);
164 static int	filt_timerattach(struct knote *kn);
165 static void	filt_timerdetach(struct knote *kn);
166 static void	filt_timerstart(struct knote *kn, sbintime_t to);
167 static void	filt_timertouch(struct knote *kn, struct kevent *kev,
168 		    u_long type);
169 static int	filt_timervalidate(struct knote *kn, sbintime_t *to);
170 static int	filt_timer(struct knote *kn, long hint);
171 static int	filt_userattach(struct knote *kn);
172 static void	filt_userdetach(struct knote *kn);
173 static int	filt_user(struct knote *kn, long hint);
174 static void	filt_usertouch(struct knote *kn, struct kevent *kev,
175 		    u_long type);
176 
177 static struct filterops file_filtops = {
178 	.f_isfd = 1,
179 	.f_attach = filt_fileattach,
180 };
181 static struct filterops kqread_filtops = {
182 	.f_isfd = 1,
183 	.f_detach = filt_kqdetach,
184 	.f_event = filt_kqueue,
185 };
186 /* XXX - move to kern_proc.c?  */
187 static struct filterops proc_filtops = {
188 	.f_isfd = 0,
189 	.f_attach = filt_procattach,
190 	.f_detach = filt_procdetach,
191 	.f_event = filt_proc,
192 };
193 static struct filterops timer_filtops = {
194 	.f_isfd = 0,
195 	.f_attach = filt_timerattach,
196 	.f_detach = filt_timerdetach,
197 	.f_event = filt_timer,
198 	.f_touch = filt_timertouch,
199 };
200 static struct filterops user_filtops = {
201 	.f_attach = filt_userattach,
202 	.f_detach = filt_userdetach,
203 	.f_event = filt_user,
204 	.f_touch = filt_usertouch,
205 };
206 
207 static uma_zone_t	knote_zone;
208 static unsigned int	kq_ncallouts = 0;
209 static unsigned int 	kq_calloutmax = 4 * 1024;
210 SYSCTL_UINT(_kern, OID_AUTO, kq_calloutmax, CTLFLAG_RW,
211     &kq_calloutmax, 0, "Maximum number of callouts allocated for kqueue");
212 
213 /* XXX - ensure not influx ? */
214 #define KNOTE_ACTIVATE(kn, islock) do { 				\
215 	if ((islock))							\
216 		mtx_assert(&(kn)->kn_kq->kq_lock, MA_OWNED);		\
217 	else								\
218 		KQ_LOCK((kn)->kn_kq);					\
219 	(kn)->kn_status |= KN_ACTIVE;					\
220 	if (((kn)->kn_status & (KN_QUEUED | KN_DISABLED)) == 0)		\
221 		knote_enqueue((kn));					\
222 	if (!(islock))							\
223 		KQ_UNLOCK((kn)->kn_kq);					\
224 } while(0)
225 #define KQ_LOCK(kq) do {						\
226 	mtx_lock(&(kq)->kq_lock);					\
227 } while (0)
228 #define KQ_FLUX_WAKEUP(kq) do {						\
229 	if (((kq)->kq_state & KQ_FLUXWAIT) == KQ_FLUXWAIT) {		\
230 		(kq)->kq_state &= ~KQ_FLUXWAIT;				\
231 		wakeup((kq));						\
232 	}								\
233 } while (0)
234 #define KQ_UNLOCK_FLUX(kq) do {						\
235 	KQ_FLUX_WAKEUP(kq);						\
236 	mtx_unlock(&(kq)->kq_lock);					\
237 } while (0)
238 #define KQ_UNLOCK(kq) do {						\
239 	mtx_unlock(&(kq)->kq_lock);					\
240 } while (0)
241 #define KQ_OWNED(kq) do {						\
242 	mtx_assert(&(kq)->kq_lock, MA_OWNED);				\
243 } while (0)
244 #define KQ_NOTOWNED(kq) do {						\
245 	mtx_assert(&(kq)->kq_lock, MA_NOTOWNED);			\
246 } while (0)
247 
248 static struct knlist *
249 kn_list_lock(struct knote *kn)
250 {
251 	struct knlist *knl;
252 
253 	knl = kn->kn_knlist;
254 	if (knl != NULL)
255 		knl->kl_lock(knl->kl_lockarg);
256 	return (knl);
257 }
258 
259 static void
260 kn_list_unlock(struct knlist *knl)
261 {
262 	bool do_free;
263 
264 	if (knl == NULL)
265 		return;
266 	do_free = knl->kl_autodestroy && knlist_empty(knl);
267 	knl->kl_unlock(knl->kl_lockarg);
268 	if (do_free) {
269 		knlist_destroy(knl);
270 		free(knl, M_KQUEUE);
271 	}
272 }
273 
274 static bool
275 kn_in_flux(struct knote *kn)
276 {
277 
278 	return (kn->kn_influx > 0);
279 }
280 
281 static void
282 kn_enter_flux(struct knote *kn)
283 {
284 
285 	KQ_OWNED(kn->kn_kq);
286 	MPASS(kn->kn_influx < INT_MAX);
287 	kn->kn_influx++;
288 }
289 
290 static bool
291 kn_leave_flux(struct knote *kn)
292 {
293 
294 	KQ_OWNED(kn->kn_kq);
295 	MPASS(kn->kn_influx > 0);
296 	kn->kn_influx--;
297 	return (kn->kn_influx == 0);
298 }
299 
300 #define	KNL_ASSERT_LOCK(knl, islocked) do {				\
301 	if (islocked)							\
302 		KNL_ASSERT_LOCKED(knl);				\
303 	else								\
304 		KNL_ASSERT_UNLOCKED(knl);				\
305 } while (0)
306 #ifdef INVARIANTS
307 #define	KNL_ASSERT_LOCKED(knl) do {					\
308 	knl->kl_assert_locked((knl)->kl_lockarg);			\
309 } while (0)
310 #define	KNL_ASSERT_UNLOCKED(knl) do {					\
311 	knl->kl_assert_unlocked((knl)->kl_lockarg);			\
312 } while (0)
313 #else /* !INVARIANTS */
314 #define	KNL_ASSERT_LOCKED(knl) do {} while(0)
315 #define	KNL_ASSERT_UNLOCKED(knl) do {} while (0)
316 #endif /* INVARIANTS */
317 
318 #ifndef	KN_HASHSIZE
319 #define	KN_HASHSIZE		64		/* XXX should be tunable */
320 #endif
321 
322 #define KN_HASH(val, mask)	(((val) ^ (val >> 8)) & (mask))
323 
324 static int
325 filt_nullattach(struct knote *kn)
326 {
327 
328 	return (ENXIO);
329 };
330 
331 struct filterops null_filtops = {
332 	.f_isfd = 0,
333 	.f_attach = filt_nullattach,
334 };
335 
336 /* XXX - make SYSINIT to add these, and move into respective modules. */
337 extern struct filterops sig_filtops;
338 extern struct filterops fs_filtops;
339 
340 /*
341  * Table for for all system-defined filters.
342  */
343 static struct mtx	filterops_lock;
344 MTX_SYSINIT(kqueue_filterops, &filterops_lock, "protect sysfilt_ops",
345 	MTX_DEF);
346 static struct {
347 	struct filterops *for_fop;
348 	int for_nolock;
349 	int for_refcnt;
350 } sysfilt_ops[EVFILT_SYSCOUNT] = {
351 	{ &file_filtops, 1 },			/* EVFILT_READ */
352 	{ &file_filtops, 1 },			/* EVFILT_WRITE */
353 	{ &null_filtops },			/* EVFILT_AIO */
354 	{ &file_filtops, 1 },			/* EVFILT_VNODE */
355 	{ &proc_filtops, 1 },			/* EVFILT_PROC */
356 	{ &sig_filtops, 1 },			/* EVFILT_SIGNAL */
357 	{ &timer_filtops, 1 },			/* EVFILT_TIMER */
358 	{ &file_filtops, 1 },			/* EVFILT_PROCDESC */
359 	{ &fs_filtops, 1 },			/* EVFILT_FS */
360 	{ &null_filtops },			/* EVFILT_LIO */
361 	{ &user_filtops, 1 },			/* EVFILT_USER */
362 	{ &null_filtops },			/* EVFILT_SENDFILE */
363 	{ &file_filtops, 1 },                   /* EVFILT_EMPTY */
364 };
365 
366 /*
367  * Simple redirection for all cdevsw style objects to call their fo_kqfilter
368  * method.
369  */
370 static int
371 filt_fileattach(struct knote *kn)
372 {
373 
374 	return (fo_kqfilter(kn->kn_fp, kn));
375 }
376 
377 /*ARGSUSED*/
378 static int
379 kqueue_kqfilter(struct file *fp, struct knote *kn)
380 {
381 	struct kqueue *kq = kn->kn_fp->f_data;
382 
383 	if (kn->kn_filter != EVFILT_READ)
384 		return (EINVAL);
385 
386 	kn->kn_status |= KN_KQUEUE;
387 	kn->kn_fop = &kqread_filtops;
388 	knlist_add(&kq->kq_sel.si_note, kn, 0);
389 
390 	return (0);
391 }
392 
393 static void
394 filt_kqdetach(struct knote *kn)
395 {
396 	struct kqueue *kq = kn->kn_fp->f_data;
397 
398 	knlist_remove(&kq->kq_sel.si_note, kn, 0);
399 }
400 
401 /*ARGSUSED*/
402 static int
403 filt_kqueue(struct knote *kn, long hint)
404 {
405 	struct kqueue *kq = kn->kn_fp->f_data;
406 
407 	kn->kn_data = kq->kq_count;
408 	return (kn->kn_data > 0);
409 }
410 
411 /* XXX - move to kern_proc.c?  */
412 static int
413 filt_procattach(struct knote *kn)
414 {
415 	struct proc *p;
416 	int error;
417 	bool exiting, immediate;
418 
419 	exiting = immediate = false;
420 	if (kn->kn_sfflags & NOTE_EXIT)
421 		p = pfind_any(kn->kn_id);
422 	else
423 		p = pfind(kn->kn_id);
424 	if (p == NULL)
425 		return (ESRCH);
426 	if (p->p_flag & P_WEXIT)
427 		exiting = true;
428 
429 	if ((error = p_cansee(curthread, p))) {
430 		PROC_UNLOCK(p);
431 		return (error);
432 	}
433 
434 	kn->kn_ptr.p_proc = p;
435 	kn->kn_flags |= EV_CLEAR;		/* automatically set */
436 
437 	/*
438 	 * Internal flag indicating registration done by kernel for the
439 	 * purposes of getting a NOTE_CHILD notification.
440 	 */
441 	if (kn->kn_flags & EV_FLAG2) {
442 		kn->kn_flags &= ~EV_FLAG2;
443 		kn->kn_data = kn->kn_sdata;		/* ppid */
444 		kn->kn_fflags = NOTE_CHILD;
445 		kn->kn_sfflags &= ~(NOTE_EXIT | NOTE_EXEC | NOTE_FORK);
446 		immediate = true; /* Force immediate activation of child note. */
447 	}
448 	/*
449 	 * Internal flag indicating registration done by kernel (for other than
450 	 * NOTE_CHILD).
451 	 */
452 	if (kn->kn_flags & EV_FLAG1) {
453 		kn->kn_flags &= ~EV_FLAG1;
454 	}
455 
456 	knlist_add(p->p_klist, kn, 1);
457 
458 	/*
459 	 * Immediately activate any child notes or, in the case of a zombie
460 	 * target process, exit notes.  The latter is necessary to handle the
461 	 * case where the target process, e.g. a child, dies before the kevent
462 	 * is registered.
463 	 */
464 	if (immediate || (exiting && filt_proc(kn, NOTE_EXIT)))
465 		KNOTE_ACTIVATE(kn, 0);
466 
467 	PROC_UNLOCK(p);
468 
469 	return (0);
470 }
471 
472 /*
473  * The knote may be attached to a different process, which may exit,
474  * leaving nothing for the knote to be attached to.  So when the process
475  * exits, the knote is marked as DETACHED and also flagged as ONESHOT so
476  * it will be deleted when read out.  However, as part of the knote deletion,
477  * this routine is called, so a check is needed to avoid actually performing
478  * a detach, because the original process does not exist any more.
479  */
480 /* XXX - move to kern_proc.c?  */
481 static void
482 filt_procdetach(struct knote *kn)
483 {
484 
485 	knlist_remove(kn->kn_knlist, kn, 0);
486 	kn->kn_ptr.p_proc = NULL;
487 }
488 
489 /* XXX - move to kern_proc.c?  */
490 static int
491 filt_proc(struct knote *kn, long hint)
492 {
493 	struct proc *p;
494 	u_int event;
495 
496 	p = kn->kn_ptr.p_proc;
497 	if (p == NULL) /* already activated, from attach filter */
498 		return (0);
499 
500 	/* Mask off extra data. */
501 	event = (u_int)hint & NOTE_PCTRLMASK;
502 
503 	/* If the user is interested in this event, record it. */
504 	if (kn->kn_sfflags & event)
505 		kn->kn_fflags |= event;
506 
507 	/* Process is gone, so flag the event as finished. */
508 	if (event == NOTE_EXIT) {
509 		kn->kn_flags |= EV_EOF | EV_ONESHOT;
510 		kn->kn_ptr.p_proc = NULL;
511 		if (kn->kn_fflags & NOTE_EXIT)
512 			kn->kn_data = KW_EXITCODE(p->p_xexit, p->p_xsig);
513 		if (kn->kn_fflags == 0)
514 			kn->kn_flags |= EV_DROP;
515 		return (1);
516 	}
517 
518 	return (kn->kn_fflags != 0);
519 }
520 
521 /*
522  * Called when the process forked. It mostly does the same as the
523  * knote(), activating all knotes registered to be activated when the
524  * process forked. Additionally, for each knote attached to the
525  * parent, check whether user wants to track the new process. If so
526  * attach a new knote to it, and immediately report an event with the
527  * child's pid.
528  */
529 void
530 knote_fork(struct knlist *list, int pid)
531 {
532 	struct kqueue *kq;
533 	struct knote *kn;
534 	struct kevent kev;
535 	int error;
536 
537 	MPASS(list != NULL);
538 	KNL_ASSERT_LOCKED(list);
539 	if (SLIST_EMPTY(&list->kl_list))
540 		return;
541 
542 	memset(&kev, 0, sizeof(kev));
543 	SLIST_FOREACH(kn, &list->kl_list, kn_selnext) {
544 		kq = kn->kn_kq;
545 		KQ_LOCK(kq);
546 		if (kn_in_flux(kn) && (kn->kn_status & KN_SCAN) == 0) {
547 			KQ_UNLOCK(kq);
548 			continue;
549 		}
550 
551 		/*
552 		 * The same as knote(), activate the event.
553 		 */
554 		if ((kn->kn_sfflags & NOTE_TRACK) == 0) {
555 			if (kn->kn_fop->f_event(kn, NOTE_FORK))
556 				KNOTE_ACTIVATE(kn, 1);
557 			KQ_UNLOCK(kq);
558 			continue;
559 		}
560 
561 		/*
562 		 * The NOTE_TRACK case. In addition to the activation
563 		 * of the event, we need to register new events to
564 		 * track the child. Drop the locks in preparation for
565 		 * the call to kqueue_register().
566 		 */
567 		kn_enter_flux(kn);
568 		KQ_UNLOCK(kq);
569 		list->kl_unlock(list->kl_lockarg);
570 
571 		/*
572 		 * Activate existing knote and register tracking knotes with
573 		 * new process.
574 		 *
575 		 * First register a knote to get just the child notice. This
576 		 * must be a separate note from a potential NOTE_EXIT
577 		 * notification since both NOTE_CHILD and NOTE_EXIT are defined
578 		 * to use the data field (in conflicting ways).
579 		 */
580 		kev.ident = pid;
581 		kev.filter = kn->kn_filter;
582 		kev.flags = kn->kn_flags | EV_ADD | EV_ENABLE | EV_ONESHOT |
583 		    EV_FLAG2;
584 		kev.fflags = kn->kn_sfflags;
585 		kev.data = kn->kn_id;		/* parent */
586 		kev.udata = kn->kn_kevent.udata;/* preserve udata */
587 		error = kqueue_register(kq, &kev, NULL, M_NOWAIT);
588 		if (error)
589 			kn->kn_fflags |= NOTE_TRACKERR;
590 
591 		/*
592 		 * Then register another knote to track other potential events
593 		 * from the new process.
594 		 */
595 		kev.ident = pid;
596 		kev.filter = kn->kn_filter;
597 		kev.flags = kn->kn_flags | EV_ADD | EV_ENABLE | EV_FLAG1;
598 		kev.fflags = kn->kn_sfflags;
599 		kev.data = kn->kn_id;		/* parent */
600 		kev.udata = kn->kn_kevent.udata;/* preserve udata */
601 		error = kqueue_register(kq, &kev, NULL, M_NOWAIT);
602 		if (error)
603 			kn->kn_fflags |= NOTE_TRACKERR;
604 		if (kn->kn_fop->f_event(kn, NOTE_FORK))
605 			KNOTE_ACTIVATE(kn, 0);
606 		list->kl_lock(list->kl_lockarg);
607 		KQ_LOCK(kq);
608 		kn_leave_flux(kn);
609 		KQ_UNLOCK_FLUX(kq);
610 	}
611 }
612 
613 /*
614  * XXX: EVFILT_TIMER should perhaps live in kern_time.c beside the
615  * interval timer support code.
616  */
617 
618 #define NOTE_TIMER_PRECMASK						\
619     (NOTE_SECONDS | NOTE_MSECONDS | NOTE_USECONDS | NOTE_NSECONDS)
620 
621 static sbintime_t
622 timer2sbintime(int64_t data, int flags)
623 {
624 	int64_t secs;
625 
626         /*
627          * Macros for converting to the fractional second portion of an
628          * sbintime_t using 64bit multiplication to improve precision.
629          */
630 #define NS_TO_SBT(ns) (((ns) * (((uint64_t)1 << 63) / 500000000)) >> 32)
631 #define US_TO_SBT(us) (((us) * (((uint64_t)1 << 63) / 500000)) >> 32)
632 #define MS_TO_SBT(ms) (((ms) * (((uint64_t)1 << 63) / 500)) >> 32)
633 	switch (flags & NOTE_TIMER_PRECMASK) {
634 	case NOTE_SECONDS:
635 #ifdef __LP64__
636 		if (data > (SBT_MAX / SBT_1S))
637 			return (SBT_MAX);
638 #endif
639 		return ((sbintime_t)data << 32);
640 	case NOTE_MSECONDS: /* FALLTHROUGH */
641 	case 0:
642 		if (data >= 1000) {
643 			secs = data / 1000;
644 #ifdef __LP64__
645 			if (secs > (SBT_MAX / SBT_1S))
646 				return (SBT_MAX);
647 #endif
648 			return (secs << 32 | MS_TO_SBT(data % 1000));
649 		}
650 		return (MS_TO_SBT(data));
651 	case NOTE_USECONDS:
652 		if (data >= 1000000) {
653 			secs = data / 1000000;
654 #ifdef __LP64__
655 			if (secs > (SBT_MAX / SBT_1S))
656 				return (SBT_MAX);
657 #endif
658 			return (secs << 32 | US_TO_SBT(data % 1000000));
659 		}
660 		return (US_TO_SBT(data));
661 	case NOTE_NSECONDS:
662 		if (data >= 1000000000) {
663 			secs = data / 1000000000;
664 #ifdef __LP64__
665 			if (secs > (SBT_MAX / SBT_1S))
666 				return (SBT_MAX);
667 #endif
668 			return (secs << 32 | US_TO_SBT(data % 1000000000));
669 		}
670 		return (NS_TO_SBT(data));
671 	default:
672 		break;
673 	}
674 	return (-1);
675 }
676 
677 struct kq_timer_cb_data {
678 	struct callout c;
679 	sbintime_t next;	/* next timer event fires at */
680 	sbintime_t to;		/* precalculated timer period, 0 for abs */
681 };
682 
683 static void
684 filt_timerexpire(void *knx)
685 {
686 	struct knote *kn;
687 	struct kq_timer_cb_data *kc;
688 
689 	kn = knx;
690 	kn->kn_data++;
691 	KNOTE_ACTIVATE(kn, 0);	/* XXX - handle locking */
692 
693 	if ((kn->kn_flags & EV_ONESHOT) != 0)
694 		return;
695 	kc = kn->kn_ptr.p_v;
696 	if (kc->to == 0)
697 		return;
698 	kc->next += kc->to;
699 	callout_reset_sbt_on(&kc->c, kc->next, 0, filt_timerexpire, kn,
700 	    PCPU_GET(cpuid), C_ABSOLUTE);
701 }
702 
703 /*
704  * data contains amount of time to sleep
705  */
706 static int
707 filt_timervalidate(struct knote *kn, sbintime_t *to)
708 {
709 	struct bintime bt;
710 	sbintime_t sbt;
711 
712 	if (kn->kn_sdata < 0)
713 		return (EINVAL);
714 	if (kn->kn_sdata == 0 && (kn->kn_flags & EV_ONESHOT) == 0)
715 		kn->kn_sdata = 1;
716 	/*
717 	 * The only fflags values supported are the timer unit
718 	 * (precision) and the absolute time indicator.
719 	 */
720 	if ((kn->kn_sfflags & ~(NOTE_TIMER_PRECMASK | NOTE_ABSTIME)) != 0)
721 		return (EINVAL);
722 
723 	*to = timer2sbintime(kn->kn_sdata, kn->kn_sfflags);
724 	if ((kn->kn_sfflags & NOTE_ABSTIME) != 0) {
725 		getboottimebin(&bt);
726 		sbt = bttosbt(bt);
727 		*to -= sbt;
728 	}
729 	if (*to < 0)
730 		return (EINVAL);
731 	return (0);
732 }
733 
734 static int
735 filt_timerattach(struct knote *kn)
736 {
737 	struct kq_timer_cb_data *kc;
738 	sbintime_t to;
739 	unsigned int ncallouts;
740 	int error;
741 
742 	error = filt_timervalidate(kn, &to);
743 	if (error != 0)
744 		return (error);
745 
746 	do {
747 		ncallouts = kq_ncallouts;
748 		if (ncallouts >= kq_calloutmax)
749 			return (ENOMEM);
750 	} while (!atomic_cmpset_int(&kq_ncallouts, ncallouts, ncallouts + 1));
751 
752 	if ((kn->kn_sfflags & NOTE_ABSTIME) == 0)
753 		kn->kn_flags |= EV_CLEAR;	/* automatically set */
754 	kn->kn_status &= ~KN_DETACHED;		/* knlist_add clears it */
755 	kn->kn_ptr.p_v = kc = malloc(sizeof(*kc), M_KQUEUE, M_WAITOK);
756 	callout_init(&kc->c, 1);
757 	filt_timerstart(kn, to);
758 
759 	return (0);
760 }
761 
762 static void
763 filt_timerstart(struct knote *kn, sbintime_t to)
764 {
765 	struct kq_timer_cb_data *kc;
766 
767 	kc = kn->kn_ptr.p_v;
768 	if ((kn->kn_sfflags & NOTE_ABSTIME) != 0) {
769 		kc->next = to;
770 		kc->to = 0;
771 	} else {
772 		kc->next = to + sbinuptime();
773 		kc->to = to;
774 	}
775 	callout_reset_sbt_on(&kc->c, kc->next, 0, filt_timerexpire, kn,
776 	    PCPU_GET(cpuid), C_ABSOLUTE);
777 }
778 
779 static void
780 filt_timerdetach(struct knote *kn)
781 {
782 	struct kq_timer_cb_data *kc;
783 	unsigned int old __unused;
784 
785 	kc = kn->kn_ptr.p_v;
786 	callout_drain(&kc->c);
787 	free(kc, M_KQUEUE);
788 	old = atomic_fetchadd_int(&kq_ncallouts, -1);
789 	KASSERT(old > 0, ("Number of callouts cannot become negative"));
790 	kn->kn_status |= KN_DETACHED;	/* knlist_remove sets it */
791 }
792 
793 static void
794 filt_timertouch(struct knote *kn, struct kevent *kev, u_long type)
795 {
796 	struct kq_timer_cb_data *kc;
797 	struct kqueue *kq;
798 	sbintime_t to;
799 	int error;
800 
801 	switch (type) {
802 	case EVENT_REGISTER:
803 		/* Handle re-added timers that update data/fflags */
804 		if (kev->flags & EV_ADD) {
805 			kc = kn->kn_ptr.p_v;
806 
807 			/* Drain any existing callout. */
808 			callout_drain(&kc->c);
809 
810 			/* Throw away any existing undelivered record
811 			 * of the timer expiration. This is done under
812 			 * the presumption that if a process is
813 			 * re-adding this timer with new parameters,
814 			 * it is no longer interested in what may have
815 			 * happened under the old parameters. If it is
816 			 * interested, it can wait for the expiration,
817 			 * delete the old timer definition, and then
818 			 * add the new one.
819 			 *
820 			 * This has to be done while the kq is locked:
821 			 *   - if enqueued, dequeue
822 			 *   - make it no longer active
823 			 *   - clear the count of expiration events
824 			 */
825 			kq = kn->kn_kq;
826 			KQ_LOCK(kq);
827 			if (kn->kn_status & KN_QUEUED)
828 				knote_dequeue(kn);
829 
830 			kn->kn_status &= ~KN_ACTIVE;
831 			kn->kn_data = 0;
832 			KQ_UNLOCK(kq);
833 
834 			/* Reschedule timer based on new data/fflags */
835 			kn->kn_sfflags = kev->fflags;
836 			kn->kn_sdata = kev->data;
837 			error = filt_timervalidate(kn, &to);
838 			if (error != 0) {
839 			  	kn->kn_flags |= EV_ERROR;
840 				kn->kn_data = error;
841 			} else
842 			  	filt_timerstart(kn, to);
843 		}
844 		break;
845 
846         case EVENT_PROCESS:
847 		*kev = kn->kn_kevent;
848 		if (kn->kn_flags & EV_CLEAR) {
849 			kn->kn_data = 0;
850 			kn->kn_fflags = 0;
851 		}
852 		break;
853 
854 	default:
855 		panic("filt_timertouch() - invalid type (%ld)", type);
856 		break;
857 	}
858 }
859 
860 static int
861 filt_timer(struct knote *kn, long hint)
862 {
863 
864 	return (kn->kn_data != 0);
865 }
866 
867 static int
868 filt_userattach(struct knote *kn)
869 {
870 
871 	/*
872 	 * EVFILT_USER knotes are not attached to anything in the kernel.
873 	 */
874 	kn->kn_hook = NULL;
875 	if (kn->kn_fflags & NOTE_TRIGGER)
876 		kn->kn_hookid = 1;
877 	else
878 		kn->kn_hookid = 0;
879 	return (0);
880 }
881 
882 static void
883 filt_userdetach(__unused struct knote *kn)
884 {
885 
886 	/*
887 	 * EVFILT_USER knotes are not attached to anything in the kernel.
888 	 */
889 }
890 
891 static int
892 filt_user(struct knote *kn, __unused long hint)
893 {
894 
895 	return (kn->kn_hookid);
896 }
897 
898 static void
899 filt_usertouch(struct knote *kn, struct kevent *kev, u_long type)
900 {
901 	u_int ffctrl;
902 
903 	switch (type) {
904 	case EVENT_REGISTER:
905 		if (kev->fflags & NOTE_TRIGGER)
906 			kn->kn_hookid = 1;
907 
908 		ffctrl = kev->fflags & NOTE_FFCTRLMASK;
909 		kev->fflags &= NOTE_FFLAGSMASK;
910 		switch (ffctrl) {
911 		case NOTE_FFNOP:
912 			break;
913 
914 		case NOTE_FFAND:
915 			kn->kn_sfflags &= kev->fflags;
916 			break;
917 
918 		case NOTE_FFOR:
919 			kn->kn_sfflags |= kev->fflags;
920 			break;
921 
922 		case NOTE_FFCOPY:
923 			kn->kn_sfflags = kev->fflags;
924 			break;
925 
926 		default:
927 			/* XXX Return error? */
928 			break;
929 		}
930 		kn->kn_sdata = kev->data;
931 		if (kev->flags & EV_CLEAR) {
932 			kn->kn_hookid = 0;
933 			kn->kn_data = 0;
934 			kn->kn_fflags = 0;
935 		}
936 		break;
937 
938         case EVENT_PROCESS:
939 		*kev = kn->kn_kevent;
940 		kev->fflags = kn->kn_sfflags;
941 		kev->data = kn->kn_sdata;
942 		if (kn->kn_flags & EV_CLEAR) {
943 			kn->kn_hookid = 0;
944 			kn->kn_data = 0;
945 			kn->kn_fflags = 0;
946 		}
947 		break;
948 
949 	default:
950 		panic("filt_usertouch() - invalid type (%ld)", type);
951 		break;
952 	}
953 }
954 
955 int
956 sys_kqueue(struct thread *td, struct kqueue_args *uap)
957 {
958 
959 	return (kern_kqueue(td, 0, NULL));
960 }
961 
962 static void
963 kqueue_init(struct kqueue *kq)
964 {
965 
966 	mtx_init(&kq->kq_lock, "kqueue", NULL, MTX_DEF | MTX_DUPOK);
967 	TAILQ_INIT(&kq->kq_head);
968 	knlist_init_mtx(&kq->kq_sel.si_note, &kq->kq_lock);
969 	TASK_INIT(&kq->kq_task, 0, kqueue_task, kq);
970 }
971 
972 int
973 kern_kqueue(struct thread *td, int flags, struct filecaps *fcaps)
974 {
975 	struct filedesc *fdp;
976 	struct kqueue *kq;
977 	struct file *fp;
978 	struct ucred *cred;
979 	int fd, error;
980 
981 	fdp = td->td_proc->p_fd;
982 	cred = td->td_ucred;
983 	if (!chgkqcnt(cred->cr_ruidinfo, 1, lim_cur(td, RLIMIT_KQUEUES)))
984 		return (ENOMEM);
985 
986 	error = falloc_caps(td, &fp, &fd, flags, fcaps);
987 	if (error != 0) {
988 		chgkqcnt(cred->cr_ruidinfo, -1, 0);
989 		return (error);
990 	}
991 
992 	/* An extra reference on `fp' has been held for us by falloc(). */
993 	kq = malloc(sizeof *kq, M_KQUEUE, M_WAITOK | M_ZERO);
994 	kqueue_init(kq);
995 	kq->kq_fdp = fdp;
996 	kq->kq_cred = crhold(cred);
997 
998 	FILEDESC_XLOCK(fdp);
999 	TAILQ_INSERT_HEAD(&fdp->fd_kqlist, kq, kq_list);
1000 	FILEDESC_XUNLOCK(fdp);
1001 
1002 	finit(fp, FREAD | FWRITE, DTYPE_KQUEUE, kq, &kqueueops);
1003 	fdrop(fp, td);
1004 
1005 	td->td_retval[0] = fd;
1006 	return (0);
1007 }
1008 
1009 struct g_kevent_args {
1010 	int	fd;
1011 	void	*changelist;
1012 	int	nchanges;
1013 	void	*eventlist;
1014 	int	nevents;
1015 	const struct timespec *timeout;
1016 };
1017 
1018 int
1019 sys_kevent(struct thread *td, struct kevent_args *uap)
1020 {
1021 	struct kevent_copyops k_ops = {
1022 		.arg = uap,
1023 		.k_copyout = kevent_copyout,
1024 		.k_copyin = kevent_copyin,
1025 		.kevent_size = sizeof(struct kevent),
1026 	};
1027 	struct g_kevent_args gk_args = {
1028 		.fd = uap->fd,
1029 		.changelist = uap->changelist,
1030 		.nchanges = uap->nchanges,
1031 		.eventlist = uap->eventlist,
1032 		.nevents = uap->nevents,
1033 		.timeout = uap->timeout,
1034 	};
1035 
1036 	return (kern_kevent_generic(td, &gk_args, &k_ops, "kevent"));
1037 }
1038 
1039 static int
1040 kern_kevent_generic(struct thread *td, struct g_kevent_args *uap,
1041     struct kevent_copyops *k_ops, const char *struct_name)
1042 {
1043 	struct timespec ts, *tsp;
1044 #ifdef KTRACE
1045 	struct kevent *eventlist = uap->eventlist;
1046 #endif
1047 	int error;
1048 
1049 	if (uap->timeout != NULL) {
1050 		error = copyin(uap->timeout, &ts, sizeof(ts));
1051 		if (error)
1052 			return (error);
1053 		tsp = &ts;
1054 	} else
1055 		tsp = NULL;
1056 
1057 #ifdef KTRACE
1058 	if (KTRPOINT(td, KTR_STRUCT_ARRAY))
1059 		ktrstructarray(struct_name, UIO_USERSPACE, uap->changelist,
1060 		    uap->nchanges, k_ops->kevent_size);
1061 #endif
1062 
1063 	error = kern_kevent(td, uap->fd, uap->nchanges, uap->nevents,
1064 	    k_ops, tsp);
1065 
1066 #ifdef KTRACE
1067 	if (error == 0 && KTRPOINT(td, KTR_STRUCT_ARRAY))
1068 		ktrstructarray(struct_name, UIO_USERSPACE, eventlist,
1069 		    td->td_retval[0], k_ops->kevent_size);
1070 #endif
1071 
1072 	return (error);
1073 }
1074 
1075 /*
1076  * Copy 'count' items into the destination list pointed to by uap->eventlist.
1077  */
1078 static int
1079 kevent_copyout(void *arg, struct kevent *kevp, int count)
1080 {
1081 	struct kevent_args *uap;
1082 	int error;
1083 
1084 	KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count));
1085 	uap = (struct kevent_args *)arg;
1086 
1087 	error = copyout(kevp, uap->eventlist, count * sizeof *kevp);
1088 	if (error == 0)
1089 		uap->eventlist += count;
1090 	return (error);
1091 }
1092 
1093 /*
1094  * Copy 'count' items from the list pointed to by uap->changelist.
1095  */
1096 static int
1097 kevent_copyin(void *arg, struct kevent *kevp, int count)
1098 {
1099 	struct kevent_args *uap;
1100 	int error;
1101 
1102 	KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count));
1103 	uap = (struct kevent_args *)arg;
1104 
1105 	error = copyin(uap->changelist, kevp, count * sizeof *kevp);
1106 	if (error == 0)
1107 		uap->changelist += count;
1108 	return (error);
1109 }
1110 
1111 #ifdef COMPAT_FREEBSD11
1112 static int
1113 kevent11_copyout(void *arg, struct kevent *kevp, int count)
1114 {
1115 	struct freebsd11_kevent_args *uap;
1116 	struct kevent_freebsd11 kev11;
1117 	int error, i;
1118 
1119 	KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count));
1120 	uap = (struct freebsd11_kevent_args *)arg;
1121 
1122 	for (i = 0; i < count; i++) {
1123 		kev11.ident = kevp->ident;
1124 		kev11.filter = kevp->filter;
1125 		kev11.flags = kevp->flags;
1126 		kev11.fflags = kevp->fflags;
1127 		kev11.data = kevp->data;
1128 		kev11.udata = kevp->udata;
1129 		error = copyout(&kev11, uap->eventlist, sizeof(kev11));
1130 		if (error != 0)
1131 			break;
1132 		uap->eventlist++;
1133 		kevp++;
1134 	}
1135 	return (error);
1136 }
1137 
1138 /*
1139  * Copy 'count' items from the list pointed to by uap->changelist.
1140  */
1141 static int
1142 kevent11_copyin(void *arg, struct kevent *kevp, int count)
1143 {
1144 	struct freebsd11_kevent_args *uap;
1145 	struct kevent_freebsd11 kev11;
1146 	int error, i;
1147 
1148 	KASSERT(count <= KQ_NEVENTS, ("count (%d) > KQ_NEVENTS", count));
1149 	uap = (struct freebsd11_kevent_args *)arg;
1150 
1151 	for (i = 0; i < count; i++) {
1152 		error = copyin(uap->changelist, &kev11, sizeof(kev11));
1153 		if (error != 0)
1154 			break;
1155 		kevp->ident = kev11.ident;
1156 		kevp->filter = kev11.filter;
1157 		kevp->flags = kev11.flags;
1158 		kevp->fflags = kev11.fflags;
1159 		kevp->data = (uintptr_t)kev11.data;
1160 		kevp->udata = kev11.udata;
1161 		bzero(&kevp->ext, sizeof(kevp->ext));
1162 		uap->changelist++;
1163 		kevp++;
1164 	}
1165 	return (error);
1166 }
1167 
1168 int
1169 freebsd11_kevent(struct thread *td, struct freebsd11_kevent_args *uap)
1170 {
1171 	struct kevent_copyops k_ops = {
1172 		.arg = uap,
1173 		.k_copyout = kevent11_copyout,
1174 		.k_copyin = kevent11_copyin,
1175 		.kevent_size = sizeof(struct kevent_freebsd11),
1176 	};
1177 	struct g_kevent_args gk_args = {
1178 		.fd = uap->fd,
1179 		.changelist = uap->changelist,
1180 		.nchanges = uap->nchanges,
1181 		.eventlist = uap->eventlist,
1182 		.nevents = uap->nevents,
1183 		.timeout = uap->timeout,
1184 	};
1185 
1186 	return (kern_kevent_generic(td, &gk_args, &k_ops, "kevent_freebsd11"));
1187 }
1188 #endif
1189 
1190 int
1191 kern_kevent(struct thread *td, int fd, int nchanges, int nevents,
1192     struct kevent_copyops *k_ops, const struct timespec *timeout)
1193 {
1194 	cap_rights_t rights;
1195 	struct file *fp;
1196 	int error;
1197 
1198 	cap_rights_init_zero(&rights);
1199 	if (nchanges > 0)
1200 		cap_rights_set_one(&rights, CAP_KQUEUE_CHANGE);
1201 	if (nevents > 0)
1202 		cap_rights_set_one(&rights, CAP_KQUEUE_EVENT);
1203 	error = fget(td, fd, &rights, &fp);
1204 	if (error != 0)
1205 		return (error);
1206 
1207 	error = kern_kevent_fp(td, fp, nchanges, nevents, k_ops, timeout);
1208 	fdrop(fp, td);
1209 
1210 	return (error);
1211 }
1212 
1213 static int
1214 kqueue_kevent(struct kqueue *kq, struct thread *td, int nchanges, int nevents,
1215     struct kevent_copyops *k_ops, const struct timespec *timeout)
1216 {
1217 	struct kevent keva[KQ_NEVENTS];
1218 	struct kevent *kevp, *changes;
1219 	int i, n, nerrors, error;
1220 
1221 	nerrors = 0;
1222 	while (nchanges > 0) {
1223 		n = nchanges > KQ_NEVENTS ? KQ_NEVENTS : nchanges;
1224 		error = k_ops->k_copyin(k_ops->arg, keva, n);
1225 		if (error)
1226 			return (error);
1227 		changes = keva;
1228 		for (i = 0; i < n; i++) {
1229 			kevp = &changes[i];
1230 			if (!kevp->filter)
1231 				continue;
1232 			kevp->flags &= ~EV_SYSFLAGS;
1233 			error = kqueue_register(kq, kevp, td, M_WAITOK);
1234 			if (error || (kevp->flags & EV_RECEIPT)) {
1235 				if (nevents == 0)
1236 					return (error);
1237 				kevp->flags = EV_ERROR;
1238 				kevp->data = error;
1239 				(void)k_ops->k_copyout(k_ops->arg, kevp, 1);
1240 				nevents--;
1241 				nerrors++;
1242 			}
1243 		}
1244 		nchanges -= n;
1245 	}
1246 	if (nerrors) {
1247 		td->td_retval[0] = nerrors;
1248 		return (0);
1249 	}
1250 
1251 	return (kqueue_scan(kq, nevents, k_ops, timeout, keva, td));
1252 }
1253 
1254 int
1255 kern_kevent_fp(struct thread *td, struct file *fp, int nchanges, int nevents,
1256     struct kevent_copyops *k_ops, const struct timespec *timeout)
1257 {
1258 	struct kqueue *kq;
1259 	int error;
1260 
1261 	error = kqueue_acquire(fp, &kq);
1262 	if (error != 0)
1263 		return (error);
1264 	error = kqueue_kevent(kq, td, nchanges, nevents, k_ops, timeout);
1265 	kqueue_release(kq, 0);
1266 	return (error);
1267 }
1268 
1269 /*
1270  * Performs a kevent() call on a temporarily created kqueue. This can be
1271  * used to perform one-shot polling, similar to poll() and select().
1272  */
1273 int
1274 kern_kevent_anonymous(struct thread *td, int nevents,
1275     struct kevent_copyops *k_ops)
1276 {
1277 	struct kqueue kq = {};
1278 	int error;
1279 
1280 	kqueue_init(&kq);
1281 	kq.kq_refcnt = 1;
1282 	error = kqueue_kevent(&kq, td, nevents, nevents, k_ops, NULL);
1283 	kqueue_drain(&kq, td);
1284 	kqueue_destroy(&kq);
1285 	return (error);
1286 }
1287 
1288 int
1289 kqueue_add_filteropts(int filt, struct filterops *filtops)
1290 {
1291 	int error;
1292 
1293 	error = 0;
1294 	if (filt > 0 || filt + EVFILT_SYSCOUNT < 0) {
1295 		printf(
1296 "trying to add a filterop that is out of range: %d is beyond %d\n",
1297 		    ~filt, EVFILT_SYSCOUNT);
1298 		return EINVAL;
1299 	}
1300 	mtx_lock(&filterops_lock);
1301 	if (sysfilt_ops[~filt].for_fop != &null_filtops &&
1302 	    sysfilt_ops[~filt].for_fop != NULL)
1303 		error = EEXIST;
1304 	else {
1305 		sysfilt_ops[~filt].for_fop = filtops;
1306 		sysfilt_ops[~filt].for_refcnt = 0;
1307 	}
1308 	mtx_unlock(&filterops_lock);
1309 
1310 	return (error);
1311 }
1312 
1313 int
1314 kqueue_del_filteropts(int filt)
1315 {
1316 	int error;
1317 
1318 	error = 0;
1319 	if (filt > 0 || filt + EVFILT_SYSCOUNT < 0)
1320 		return EINVAL;
1321 
1322 	mtx_lock(&filterops_lock);
1323 	if (sysfilt_ops[~filt].for_fop == &null_filtops ||
1324 	    sysfilt_ops[~filt].for_fop == NULL)
1325 		error = EINVAL;
1326 	else if (sysfilt_ops[~filt].for_refcnt != 0)
1327 		error = EBUSY;
1328 	else {
1329 		sysfilt_ops[~filt].for_fop = &null_filtops;
1330 		sysfilt_ops[~filt].for_refcnt = 0;
1331 	}
1332 	mtx_unlock(&filterops_lock);
1333 
1334 	return error;
1335 }
1336 
1337 static struct filterops *
1338 kqueue_fo_find(int filt)
1339 {
1340 
1341 	if (filt > 0 || filt + EVFILT_SYSCOUNT < 0)
1342 		return NULL;
1343 
1344 	if (sysfilt_ops[~filt].for_nolock)
1345 		return sysfilt_ops[~filt].for_fop;
1346 
1347 	mtx_lock(&filterops_lock);
1348 	sysfilt_ops[~filt].for_refcnt++;
1349 	if (sysfilt_ops[~filt].for_fop == NULL)
1350 		sysfilt_ops[~filt].for_fop = &null_filtops;
1351 	mtx_unlock(&filterops_lock);
1352 
1353 	return sysfilt_ops[~filt].for_fop;
1354 }
1355 
1356 static void
1357 kqueue_fo_release(int filt)
1358 {
1359 
1360 	if (filt > 0 || filt + EVFILT_SYSCOUNT < 0)
1361 		return;
1362 
1363 	if (sysfilt_ops[~filt].for_nolock)
1364 		return;
1365 
1366 	mtx_lock(&filterops_lock);
1367 	KASSERT(sysfilt_ops[~filt].for_refcnt > 0,
1368 	    ("filter object refcount not valid on release"));
1369 	sysfilt_ops[~filt].for_refcnt--;
1370 	mtx_unlock(&filterops_lock);
1371 }
1372 
1373 /*
1374  * A ref to kq (obtained via kqueue_acquire) must be held.
1375  */
1376 static int
1377 kqueue_register(struct kqueue *kq, struct kevent *kev, struct thread *td,
1378     int mflag)
1379 {
1380 	struct filterops *fops;
1381 	struct file *fp;
1382 	struct knote *kn, *tkn;
1383 	struct knlist *knl;
1384 	int error, filt, event;
1385 	int haskqglobal, filedesc_unlock;
1386 
1387 	if ((kev->flags & (EV_ENABLE | EV_DISABLE)) == (EV_ENABLE | EV_DISABLE))
1388 		return (EINVAL);
1389 
1390 	fp = NULL;
1391 	kn = NULL;
1392 	knl = NULL;
1393 	error = 0;
1394 	haskqglobal = 0;
1395 	filedesc_unlock = 0;
1396 
1397 	filt = kev->filter;
1398 	fops = kqueue_fo_find(filt);
1399 	if (fops == NULL)
1400 		return EINVAL;
1401 
1402 	if (kev->flags & EV_ADD) {
1403 		/*
1404 		 * Prevent waiting with locks.  Non-sleepable
1405 		 * allocation failures are handled in the loop, only
1406 		 * if the spare knote appears to be actually required.
1407 		 */
1408 		tkn = knote_alloc(mflag);
1409 	} else {
1410 		tkn = NULL;
1411 	}
1412 
1413 findkn:
1414 	if (fops->f_isfd) {
1415 		KASSERT(td != NULL, ("td is NULL"));
1416 		if (kev->ident > INT_MAX)
1417 			error = EBADF;
1418 		else
1419 			error = fget(td, kev->ident, &cap_event_rights, &fp);
1420 		if (error)
1421 			goto done;
1422 
1423 		if ((kev->flags & EV_ADD) == EV_ADD && kqueue_expand(kq, fops,
1424 		    kev->ident, M_NOWAIT) != 0) {
1425 			/* try again */
1426 			fdrop(fp, td);
1427 			fp = NULL;
1428 			error = kqueue_expand(kq, fops, kev->ident, mflag);
1429 			if (error)
1430 				goto done;
1431 			goto findkn;
1432 		}
1433 
1434 		if (fp->f_type == DTYPE_KQUEUE) {
1435 			/*
1436 			 * If we add some intelligence about what we are doing,
1437 			 * we should be able to support events on ourselves.
1438 			 * We need to know when we are doing this to prevent
1439 			 * getting both the knlist lock and the kq lock since
1440 			 * they are the same thing.
1441 			 */
1442 			if (fp->f_data == kq) {
1443 				error = EINVAL;
1444 				goto done;
1445 			}
1446 
1447 			/*
1448 			 * Pre-lock the filedesc before the global
1449 			 * lock mutex, see the comment in
1450 			 * kqueue_close().
1451 			 */
1452 			FILEDESC_XLOCK(td->td_proc->p_fd);
1453 			filedesc_unlock = 1;
1454 			KQ_GLOBAL_LOCK(&kq_global, haskqglobal);
1455 		}
1456 
1457 		KQ_LOCK(kq);
1458 		if (kev->ident < kq->kq_knlistsize) {
1459 			SLIST_FOREACH(kn, &kq->kq_knlist[kev->ident], kn_link)
1460 				if (kev->filter == kn->kn_filter)
1461 					break;
1462 		}
1463 	} else {
1464 		if ((kev->flags & EV_ADD) == EV_ADD) {
1465 			error = kqueue_expand(kq, fops, kev->ident, mflag);
1466 			if (error != 0)
1467 				goto done;
1468 		}
1469 
1470 		KQ_LOCK(kq);
1471 
1472 		/*
1473 		 * If possible, find an existing knote to use for this kevent.
1474 		 */
1475 		if (kev->filter == EVFILT_PROC &&
1476 		    (kev->flags & (EV_FLAG1 | EV_FLAG2)) != 0) {
1477 			/* This is an internal creation of a process tracking
1478 			 * note. Don't attempt to coalesce this with an
1479 			 * existing note.
1480 			 */
1481 			;
1482 		} else if (kq->kq_knhashmask != 0) {
1483 			struct klist *list;
1484 
1485 			list = &kq->kq_knhash[
1486 			    KN_HASH((u_long)kev->ident, kq->kq_knhashmask)];
1487 			SLIST_FOREACH(kn, list, kn_link)
1488 				if (kev->ident == kn->kn_id &&
1489 				    kev->filter == kn->kn_filter)
1490 					break;
1491 		}
1492 	}
1493 
1494 	/* knote is in the process of changing, wait for it to stabilize. */
1495 	if (kn != NULL && kn_in_flux(kn)) {
1496 		KQ_GLOBAL_UNLOCK(&kq_global, haskqglobal);
1497 		if (filedesc_unlock) {
1498 			FILEDESC_XUNLOCK(td->td_proc->p_fd);
1499 			filedesc_unlock = 0;
1500 		}
1501 		kq->kq_state |= KQ_FLUXWAIT;
1502 		msleep(kq, &kq->kq_lock, PSOCK | PDROP, "kqflxwt", 0);
1503 		if (fp != NULL) {
1504 			fdrop(fp, td);
1505 			fp = NULL;
1506 		}
1507 		goto findkn;
1508 	}
1509 
1510 	/*
1511 	 * kn now contains the matching knote, or NULL if no match
1512 	 */
1513 	if (kn == NULL) {
1514 		if (kev->flags & EV_ADD) {
1515 			kn = tkn;
1516 			tkn = NULL;
1517 			if (kn == NULL) {
1518 				KQ_UNLOCK(kq);
1519 				error = ENOMEM;
1520 				goto done;
1521 			}
1522 			kn->kn_fp = fp;
1523 			kn->kn_kq = kq;
1524 			kn->kn_fop = fops;
1525 			/*
1526 			 * apply reference counts to knote structure, and
1527 			 * do not release it at the end of this routine.
1528 			 */
1529 			fops = NULL;
1530 			fp = NULL;
1531 
1532 			kn->kn_sfflags = kev->fflags;
1533 			kn->kn_sdata = kev->data;
1534 			kev->fflags = 0;
1535 			kev->data = 0;
1536 			kn->kn_kevent = *kev;
1537 			kn->kn_kevent.flags &= ~(EV_ADD | EV_DELETE |
1538 			    EV_ENABLE | EV_DISABLE | EV_FORCEONESHOT);
1539 			kn->kn_status = KN_DETACHED;
1540 			if ((kev->flags & EV_DISABLE) != 0)
1541 				kn->kn_status |= KN_DISABLED;
1542 			kn_enter_flux(kn);
1543 
1544 			error = knote_attach(kn, kq);
1545 			KQ_UNLOCK(kq);
1546 			if (error != 0) {
1547 				tkn = kn;
1548 				goto done;
1549 			}
1550 
1551 			if ((error = kn->kn_fop->f_attach(kn)) != 0) {
1552 				knote_drop_detached(kn, td);
1553 				goto done;
1554 			}
1555 			knl = kn_list_lock(kn);
1556 			goto done_ev_add;
1557 		} else {
1558 			/* No matching knote and the EV_ADD flag is not set. */
1559 			KQ_UNLOCK(kq);
1560 			error = ENOENT;
1561 			goto done;
1562 		}
1563 	}
1564 
1565 	if (kev->flags & EV_DELETE) {
1566 		kn_enter_flux(kn);
1567 		KQ_UNLOCK(kq);
1568 		knote_drop(kn, td);
1569 		goto done;
1570 	}
1571 
1572 	if (kev->flags & EV_FORCEONESHOT) {
1573 		kn->kn_flags |= EV_ONESHOT;
1574 		KNOTE_ACTIVATE(kn, 1);
1575 	}
1576 
1577 	if ((kev->flags & EV_ENABLE) != 0)
1578 		kn->kn_status &= ~KN_DISABLED;
1579 	else if ((kev->flags & EV_DISABLE) != 0)
1580 		kn->kn_status |= KN_DISABLED;
1581 
1582 	/*
1583 	 * The user may change some filter values after the initial EV_ADD,
1584 	 * but doing so will not reset any filter which has already been
1585 	 * triggered.
1586 	 */
1587 	kn->kn_status |= KN_SCAN;
1588 	kn_enter_flux(kn);
1589 	KQ_UNLOCK(kq);
1590 	knl = kn_list_lock(kn);
1591 	kn->kn_kevent.udata = kev->udata;
1592 	if (!fops->f_isfd && fops->f_touch != NULL) {
1593 		fops->f_touch(kn, kev, EVENT_REGISTER);
1594 	} else {
1595 		kn->kn_sfflags = kev->fflags;
1596 		kn->kn_sdata = kev->data;
1597 	}
1598 
1599 done_ev_add:
1600 	/*
1601 	 * We can get here with kn->kn_knlist == NULL.  This can happen when
1602 	 * the initial attach event decides that the event is "completed"
1603 	 * already, e.g., filt_procattach() is called on a zombie process.  It
1604 	 * will call filt_proc() which will remove it from the list, and NULL
1605 	 * kn_knlist.
1606 	 *
1607 	 * KN_DISABLED will be stable while the knote is in flux, so the
1608 	 * unlocked read will not race with an update.
1609 	 */
1610 	if ((kn->kn_status & KN_DISABLED) == 0)
1611 		event = kn->kn_fop->f_event(kn, 0);
1612 	else
1613 		event = 0;
1614 
1615 	KQ_LOCK(kq);
1616 	if (event)
1617 		kn->kn_status |= KN_ACTIVE;
1618 	if ((kn->kn_status & (KN_ACTIVE | KN_DISABLED | KN_QUEUED)) ==
1619 	    KN_ACTIVE)
1620 		knote_enqueue(kn);
1621 	kn->kn_status &= ~KN_SCAN;
1622 	kn_leave_flux(kn);
1623 	kn_list_unlock(knl);
1624 	KQ_UNLOCK_FLUX(kq);
1625 
1626 done:
1627 	KQ_GLOBAL_UNLOCK(&kq_global, haskqglobal);
1628 	if (filedesc_unlock)
1629 		FILEDESC_XUNLOCK(td->td_proc->p_fd);
1630 	if (fp != NULL)
1631 		fdrop(fp, td);
1632 	knote_free(tkn);
1633 	if (fops != NULL)
1634 		kqueue_fo_release(filt);
1635 	return (error);
1636 }
1637 
1638 static int
1639 kqueue_acquire(struct file *fp, struct kqueue **kqp)
1640 {
1641 	int error;
1642 	struct kqueue *kq;
1643 
1644 	error = 0;
1645 
1646 	kq = fp->f_data;
1647 	if (fp->f_type != DTYPE_KQUEUE || kq == NULL)
1648 		return (EBADF);
1649 	*kqp = kq;
1650 	KQ_LOCK(kq);
1651 	if ((kq->kq_state & KQ_CLOSING) == KQ_CLOSING) {
1652 		KQ_UNLOCK(kq);
1653 		return (EBADF);
1654 	}
1655 	kq->kq_refcnt++;
1656 	KQ_UNLOCK(kq);
1657 
1658 	return error;
1659 }
1660 
1661 static void
1662 kqueue_release(struct kqueue *kq, int locked)
1663 {
1664 	if (locked)
1665 		KQ_OWNED(kq);
1666 	else
1667 		KQ_LOCK(kq);
1668 	kq->kq_refcnt--;
1669 	if (kq->kq_refcnt == 1)
1670 		wakeup(&kq->kq_refcnt);
1671 	if (!locked)
1672 		KQ_UNLOCK(kq);
1673 }
1674 
1675 static void
1676 kqueue_schedtask(struct kqueue *kq)
1677 {
1678 
1679 	KQ_OWNED(kq);
1680 	KASSERT(((kq->kq_state & KQ_TASKDRAIN) != KQ_TASKDRAIN),
1681 	    ("scheduling kqueue task while draining"));
1682 
1683 	if ((kq->kq_state & KQ_TASKSCHED) != KQ_TASKSCHED) {
1684 		taskqueue_enqueue(taskqueue_kqueue_ctx, &kq->kq_task);
1685 		kq->kq_state |= KQ_TASKSCHED;
1686 	}
1687 }
1688 
1689 /*
1690  * Expand the kq to make sure we have storage for fops/ident pair.
1691  *
1692  * Return 0 on success (or no work necessary), return errno on failure.
1693  */
1694 static int
1695 kqueue_expand(struct kqueue *kq, struct filterops *fops, uintptr_t ident,
1696     int mflag)
1697 {
1698 	struct klist *list, *tmp_knhash, *to_free;
1699 	u_long tmp_knhashmask;
1700 	int error, fd, size;
1701 
1702 	KQ_NOTOWNED(kq);
1703 
1704 	error = 0;
1705 	to_free = NULL;
1706 	if (fops->f_isfd) {
1707 		fd = ident;
1708 		if (kq->kq_knlistsize <= fd) {
1709 			size = kq->kq_knlistsize;
1710 			while (size <= fd)
1711 				size += KQEXTENT;
1712 			list = malloc(size * sizeof(*list), M_KQUEUE, mflag);
1713 			if (list == NULL)
1714 				return ENOMEM;
1715 			KQ_LOCK(kq);
1716 			if ((kq->kq_state & KQ_CLOSING) != 0) {
1717 				to_free = list;
1718 				error = EBADF;
1719 			} else if (kq->kq_knlistsize > fd) {
1720 				to_free = list;
1721 			} else {
1722 				if (kq->kq_knlist != NULL) {
1723 					bcopy(kq->kq_knlist, list,
1724 					    kq->kq_knlistsize * sizeof(*list));
1725 					to_free = kq->kq_knlist;
1726 					kq->kq_knlist = NULL;
1727 				}
1728 				bzero((caddr_t)list +
1729 				    kq->kq_knlistsize * sizeof(*list),
1730 				    (size - kq->kq_knlistsize) * sizeof(*list));
1731 				kq->kq_knlistsize = size;
1732 				kq->kq_knlist = list;
1733 			}
1734 			KQ_UNLOCK(kq);
1735 		}
1736 	} else {
1737 		if (kq->kq_knhashmask == 0) {
1738 			tmp_knhash = hashinit_flags(KN_HASHSIZE, M_KQUEUE,
1739 			    &tmp_knhashmask, (mflag & M_WAITOK) != 0 ?
1740 			    HASH_WAITOK : HASH_NOWAIT);
1741 			if (tmp_knhash == NULL)
1742 				return (ENOMEM);
1743 			KQ_LOCK(kq);
1744 			if ((kq->kq_state & KQ_CLOSING) != 0) {
1745 				to_free = tmp_knhash;
1746 				error = EBADF;
1747 			} else if (kq->kq_knhashmask == 0) {
1748 				kq->kq_knhash = tmp_knhash;
1749 				kq->kq_knhashmask = tmp_knhashmask;
1750 			} else {
1751 				to_free = tmp_knhash;
1752 			}
1753 			KQ_UNLOCK(kq);
1754 		}
1755 	}
1756 	free(to_free, M_KQUEUE);
1757 
1758 	KQ_NOTOWNED(kq);
1759 	return (error);
1760 }
1761 
1762 static void
1763 kqueue_task(void *arg, int pending)
1764 {
1765 	struct kqueue *kq;
1766 	int haskqglobal;
1767 
1768 	haskqglobal = 0;
1769 	kq = arg;
1770 
1771 	KQ_GLOBAL_LOCK(&kq_global, haskqglobal);
1772 	KQ_LOCK(kq);
1773 
1774 	KNOTE_LOCKED(&kq->kq_sel.si_note, 0);
1775 
1776 	kq->kq_state &= ~KQ_TASKSCHED;
1777 	if ((kq->kq_state & KQ_TASKDRAIN) == KQ_TASKDRAIN) {
1778 		wakeup(&kq->kq_state);
1779 	}
1780 	KQ_UNLOCK(kq);
1781 	KQ_GLOBAL_UNLOCK(&kq_global, haskqglobal);
1782 }
1783 
1784 /*
1785  * Scan, update kn_data (if not ONESHOT), and copyout triggered events.
1786  * We treat KN_MARKER knotes as if they are in flux.
1787  */
1788 static int
1789 kqueue_scan(struct kqueue *kq, int maxevents, struct kevent_copyops *k_ops,
1790     const struct timespec *tsp, struct kevent *keva, struct thread *td)
1791 {
1792 	struct kevent *kevp;
1793 	struct knote *kn, *marker;
1794 	struct knlist *knl;
1795 	sbintime_t asbt, rsbt;
1796 	int count, error, haskqglobal, influx, nkev, touch;
1797 
1798 	count = maxevents;
1799 	nkev = 0;
1800 	error = 0;
1801 	haskqglobal = 0;
1802 
1803 	if (maxevents == 0)
1804 		goto done_nl;
1805 
1806 	rsbt = 0;
1807 	if (tsp != NULL) {
1808 		if (tsp->tv_sec < 0 || tsp->tv_nsec < 0 ||
1809 		    tsp->tv_nsec >= 1000000000) {
1810 			error = EINVAL;
1811 			goto done_nl;
1812 		}
1813 		if (timespecisset(tsp)) {
1814 			if (tsp->tv_sec <= INT32_MAX) {
1815 				rsbt = tstosbt(*tsp);
1816 				if (TIMESEL(&asbt, rsbt))
1817 					asbt += tc_tick_sbt;
1818 				if (asbt <= SBT_MAX - rsbt)
1819 					asbt += rsbt;
1820 				else
1821 					asbt = 0;
1822 				rsbt >>= tc_precexp;
1823 			} else
1824 				asbt = 0;
1825 		} else
1826 			asbt = -1;
1827 	} else
1828 		asbt = 0;
1829 	marker = knote_alloc(M_WAITOK);
1830 	marker->kn_status = KN_MARKER;
1831 	KQ_LOCK(kq);
1832 
1833 retry:
1834 	kevp = keva;
1835 	if (kq->kq_count == 0) {
1836 		if (asbt == -1) {
1837 			error = EWOULDBLOCK;
1838 		} else {
1839 			kq->kq_state |= KQ_SLEEP;
1840 			error = msleep_sbt(kq, &kq->kq_lock, PSOCK | PCATCH,
1841 			    "kqread", asbt, rsbt, C_ABSOLUTE);
1842 		}
1843 		if (error == 0)
1844 			goto retry;
1845 		/* don't restart after signals... */
1846 		if (error == ERESTART)
1847 			error = EINTR;
1848 		else if (error == EWOULDBLOCK)
1849 			error = 0;
1850 		goto done;
1851 	}
1852 
1853 	TAILQ_INSERT_TAIL(&kq->kq_head, marker, kn_tqe);
1854 	influx = 0;
1855 	while (count) {
1856 		KQ_OWNED(kq);
1857 		kn = TAILQ_FIRST(&kq->kq_head);
1858 
1859 		if ((kn->kn_status == KN_MARKER && kn != marker) ||
1860 		    kn_in_flux(kn)) {
1861 			if (influx) {
1862 				influx = 0;
1863 				KQ_FLUX_WAKEUP(kq);
1864 			}
1865 			kq->kq_state |= KQ_FLUXWAIT;
1866 			error = msleep(kq, &kq->kq_lock, PSOCK,
1867 			    "kqflxwt", 0);
1868 			continue;
1869 		}
1870 
1871 		TAILQ_REMOVE(&kq->kq_head, kn, kn_tqe);
1872 		if ((kn->kn_status & KN_DISABLED) == KN_DISABLED) {
1873 			kn->kn_status &= ~KN_QUEUED;
1874 			kq->kq_count--;
1875 			continue;
1876 		}
1877 		if (kn == marker) {
1878 			KQ_FLUX_WAKEUP(kq);
1879 			if (count == maxevents)
1880 				goto retry;
1881 			goto done;
1882 		}
1883 		KASSERT(!kn_in_flux(kn),
1884 		    ("knote %p is unexpectedly in flux", kn));
1885 
1886 		if ((kn->kn_flags & EV_DROP) == EV_DROP) {
1887 			kn->kn_status &= ~KN_QUEUED;
1888 			kn_enter_flux(kn);
1889 			kq->kq_count--;
1890 			KQ_UNLOCK(kq);
1891 			/*
1892 			 * We don't need to lock the list since we've
1893 			 * marked it as in flux.
1894 			 */
1895 			knote_drop(kn, td);
1896 			KQ_LOCK(kq);
1897 			continue;
1898 		} else if ((kn->kn_flags & EV_ONESHOT) == EV_ONESHOT) {
1899 			kn->kn_status &= ~KN_QUEUED;
1900 			kn_enter_flux(kn);
1901 			kq->kq_count--;
1902 			KQ_UNLOCK(kq);
1903 			/*
1904 			 * We don't need to lock the list since we've
1905 			 * marked the knote as being in flux.
1906 			 */
1907 			*kevp = kn->kn_kevent;
1908 			knote_drop(kn, td);
1909 			KQ_LOCK(kq);
1910 			kn = NULL;
1911 		} else {
1912 			kn->kn_status |= KN_SCAN;
1913 			kn_enter_flux(kn);
1914 			KQ_UNLOCK(kq);
1915 			if ((kn->kn_status & KN_KQUEUE) == KN_KQUEUE)
1916 				KQ_GLOBAL_LOCK(&kq_global, haskqglobal);
1917 			knl = kn_list_lock(kn);
1918 			if (kn->kn_fop->f_event(kn, 0) == 0) {
1919 				KQ_LOCK(kq);
1920 				KQ_GLOBAL_UNLOCK(&kq_global, haskqglobal);
1921 				kn->kn_status &= ~(KN_QUEUED | KN_ACTIVE |
1922 				    KN_SCAN);
1923 				kn_leave_flux(kn);
1924 				kq->kq_count--;
1925 				kn_list_unlock(knl);
1926 				influx = 1;
1927 				continue;
1928 			}
1929 			touch = (!kn->kn_fop->f_isfd &&
1930 			    kn->kn_fop->f_touch != NULL);
1931 			if (touch)
1932 				kn->kn_fop->f_touch(kn, kevp, EVENT_PROCESS);
1933 			else
1934 				*kevp = kn->kn_kevent;
1935 			KQ_LOCK(kq);
1936 			KQ_GLOBAL_UNLOCK(&kq_global, haskqglobal);
1937 			if (kn->kn_flags & (EV_CLEAR | EV_DISPATCH)) {
1938 				/*
1939 				 * Manually clear knotes who weren't
1940 				 * 'touch'ed.
1941 				 */
1942 				if (touch == 0 && kn->kn_flags & EV_CLEAR) {
1943 					kn->kn_data = 0;
1944 					kn->kn_fflags = 0;
1945 				}
1946 				if (kn->kn_flags & EV_DISPATCH)
1947 					kn->kn_status |= KN_DISABLED;
1948 				kn->kn_status &= ~(KN_QUEUED | KN_ACTIVE);
1949 				kq->kq_count--;
1950 			} else
1951 				TAILQ_INSERT_TAIL(&kq->kq_head, kn, kn_tqe);
1952 
1953 			kn->kn_status &= ~KN_SCAN;
1954 			kn_leave_flux(kn);
1955 			kn_list_unlock(knl);
1956 			influx = 1;
1957 		}
1958 
1959 		/* we are returning a copy to the user */
1960 		kevp++;
1961 		nkev++;
1962 		count--;
1963 
1964 		if (nkev == KQ_NEVENTS) {
1965 			influx = 0;
1966 			KQ_UNLOCK_FLUX(kq);
1967 			error = k_ops->k_copyout(k_ops->arg, keva, nkev);
1968 			nkev = 0;
1969 			kevp = keva;
1970 			KQ_LOCK(kq);
1971 			if (error)
1972 				break;
1973 		}
1974 	}
1975 	TAILQ_REMOVE(&kq->kq_head, marker, kn_tqe);
1976 done:
1977 	KQ_OWNED(kq);
1978 	KQ_UNLOCK_FLUX(kq);
1979 	knote_free(marker);
1980 done_nl:
1981 	KQ_NOTOWNED(kq);
1982 	if (nkev != 0)
1983 		error = k_ops->k_copyout(k_ops->arg, keva, nkev);
1984 	td->td_retval[0] = maxevents - count;
1985 	return (error);
1986 }
1987 
1988 /*ARGSUSED*/
1989 static int
1990 kqueue_ioctl(struct file *fp, u_long cmd, void *data,
1991 	struct ucred *active_cred, struct thread *td)
1992 {
1993 	/*
1994 	 * Enabling sigio causes two major problems:
1995 	 * 1) infinite recursion:
1996 	 * Synopsys: kevent is being used to track signals and have FIOASYNC
1997 	 * set.  On receipt of a signal this will cause a kqueue to recurse
1998 	 * into itself over and over.  Sending the sigio causes the kqueue
1999 	 * to become ready, which in turn posts sigio again, forever.
2000 	 * Solution: this can be solved by setting a flag in the kqueue that
2001 	 * we have a SIGIO in progress.
2002 	 * 2) locking problems:
2003 	 * Synopsys: Kqueue is a leaf subsystem, but adding signalling puts
2004 	 * us above the proc and pgrp locks.
2005 	 * Solution: Post a signal using an async mechanism, being sure to
2006 	 * record a generation count in the delivery so that we do not deliver
2007 	 * a signal to the wrong process.
2008 	 *
2009 	 * Note, these two mechanisms are somewhat mutually exclusive!
2010 	 */
2011 #if 0
2012 	struct kqueue *kq;
2013 
2014 	kq = fp->f_data;
2015 	switch (cmd) {
2016 	case FIOASYNC:
2017 		if (*(int *)data) {
2018 			kq->kq_state |= KQ_ASYNC;
2019 		} else {
2020 			kq->kq_state &= ~KQ_ASYNC;
2021 		}
2022 		return (0);
2023 
2024 	case FIOSETOWN:
2025 		return (fsetown(*(int *)data, &kq->kq_sigio));
2026 
2027 	case FIOGETOWN:
2028 		*(int *)data = fgetown(&kq->kq_sigio);
2029 		return (0);
2030 	}
2031 #endif
2032 
2033 	return (ENOTTY);
2034 }
2035 
2036 /*ARGSUSED*/
2037 static int
2038 kqueue_poll(struct file *fp, int events, struct ucred *active_cred,
2039 	struct thread *td)
2040 {
2041 	struct kqueue *kq;
2042 	int revents = 0;
2043 	int error;
2044 
2045 	if ((error = kqueue_acquire(fp, &kq)))
2046 		return POLLERR;
2047 
2048 	KQ_LOCK(kq);
2049 	if (events & (POLLIN | POLLRDNORM)) {
2050 		if (kq->kq_count) {
2051 			revents |= events & (POLLIN | POLLRDNORM);
2052 		} else {
2053 			selrecord(td, &kq->kq_sel);
2054 			if (SEL_WAITING(&kq->kq_sel))
2055 				kq->kq_state |= KQ_SEL;
2056 		}
2057 	}
2058 	kqueue_release(kq, 1);
2059 	KQ_UNLOCK(kq);
2060 	return (revents);
2061 }
2062 
2063 /*ARGSUSED*/
2064 static int
2065 kqueue_stat(struct file *fp, struct stat *st, struct ucred *active_cred,
2066 	struct thread *td)
2067 {
2068 
2069 	bzero((void *)st, sizeof *st);
2070 	/*
2071 	 * We no longer return kq_count because the unlocked value is useless.
2072 	 * If you spent all this time getting the count, why not spend your
2073 	 * syscall better by calling kevent?
2074 	 *
2075 	 * XXX - This is needed for libc_r.
2076 	 */
2077 	st->st_mode = S_IFIFO;
2078 	return (0);
2079 }
2080 
2081 static void
2082 kqueue_drain(struct kqueue *kq, struct thread *td)
2083 {
2084 	struct knote *kn;
2085 	int i;
2086 
2087 	KQ_LOCK(kq);
2088 
2089 	KASSERT((kq->kq_state & KQ_CLOSING) != KQ_CLOSING,
2090 	    ("kqueue already closing"));
2091 	kq->kq_state |= KQ_CLOSING;
2092 	if (kq->kq_refcnt > 1)
2093 		msleep(&kq->kq_refcnt, &kq->kq_lock, PSOCK, "kqclose", 0);
2094 
2095 	KASSERT(kq->kq_refcnt == 1, ("other refs are out there!"));
2096 
2097 	KASSERT(knlist_empty(&kq->kq_sel.si_note),
2098 	    ("kqueue's knlist not empty"));
2099 
2100 	for (i = 0; i < kq->kq_knlistsize; i++) {
2101 		while ((kn = SLIST_FIRST(&kq->kq_knlist[i])) != NULL) {
2102 			if (kn_in_flux(kn)) {
2103 				kq->kq_state |= KQ_FLUXWAIT;
2104 				msleep(kq, &kq->kq_lock, PSOCK, "kqclo1", 0);
2105 				continue;
2106 			}
2107 			kn_enter_flux(kn);
2108 			KQ_UNLOCK(kq);
2109 			knote_drop(kn, td);
2110 			KQ_LOCK(kq);
2111 		}
2112 	}
2113 	if (kq->kq_knhashmask != 0) {
2114 		for (i = 0; i <= kq->kq_knhashmask; i++) {
2115 			while ((kn = SLIST_FIRST(&kq->kq_knhash[i])) != NULL) {
2116 				if (kn_in_flux(kn)) {
2117 					kq->kq_state |= KQ_FLUXWAIT;
2118 					msleep(kq, &kq->kq_lock, PSOCK,
2119 					       "kqclo2", 0);
2120 					continue;
2121 				}
2122 				kn_enter_flux(kn);
2123 				KQ_UNLOCK(kq);
2124 				knote_drop(kn, td);
2125 				KQ_LOCK(kq);
2126 			}
2127 		}
2128 	}
2129 
2130 	if ((kq->kq_state & KQ_TASKSCHED) == KQ_TASKSCHED) {
2131 		kq->kq_state |= KQ_TASKDRAIN;
2132 		msleep(&kq->kq_state, &kq->kq_lock, PSOCK, "kqtqdr", 0);
2133 	}
2134 
2135 	if ((kq->kq_state & KQ_SEL) == KQ_SEL) {
2136 		selwakeuppri(&kq->kq_sel, PSOCK);
2137 		if (!SEL_WAITING(&kq->kq_sel))
2138 			kq->kq_state &= ~KQ_SEL;
2139 	}
2140 
2141 	KQ_UNLOCK(kq);
2142 }
2143 
2144 static void
2145 kqueue_destroy(struct kqueue *kq)
2146 {
2147 
2148 	KASSERT(kq->kq_fdp == NULL,
2149 	    ("kqueue still attached to a file descriptor"));
2150 	seldrain(&kq->kq_sel);
2151 	knlist_destroy(&kq->kq_sel.si_note);
2152 	mtx_destroy(&kq->kq_lock);
2153 
2154 	if (kq->kq_knhash != NULL)
2155 		free(kq->kq_knhash, M_KQUEUE);
2156 	if (kq->kq_knlist != NULL)
2157 		free(kq->kq_knlist, M_KQUEUE);
2158 
2159 	funsetown(&kq->kq_sigio);
2160 }
2161 
2162 /*ARGSUSED*/
2163 static int
2164 kqueue_close(struct file *fp, struct thread *td)
2165 {
2166 	struct kqueue *kq = fp->f_data;
2167 	struct filedesc *fdp;
2168 	int error;
2169 	int filedesc_unlock;
2170 
2171 	if ((error = kqueue_acquire(fp, &kq)))
2172 		return error;
2173 	kqueue_drain(kq, td);
2174 
2175 	/*
2176 	 * We could be called due to the knote_drop() doing fdrop(),
2177 	 * called from kqueue_register().  In this case the global
2178 	 * lock is owned, and filedesc sx is locked before, to not
2179 	 * take the sleepable lock after non-sleepable.
2180 	 */
2181 	fdp = kq->kq_fdp;
2182 	kq->kq_fdp = NULL;
2183 	if (!sx_xlocked(FILEDESC_LOCK(fdp))) {
2184 		FILEDESC_XLOCK(fdp);
2185 		filedesc_unlock = 1;
2186 	} else
2187 		filedesc_unlock = 0;
2188 	TAILQ_REMOVE(&fdp->fd_kqlist, kq, kq_list);
2189 	if (filedesc_unlock)
2190 		FILEDESC_XUNLOCK(fdp);
2191 
2192 	kqueue_destroy(kq);
2193 	chgkqcnt(kq->kq_cred->cr_ruidinfo, -1, 0);
2194 	crfree(kq->kq_cred);
2195 	free(kq, M_KQUEUE);
2196 	fp->f_data = NULL;
2197 
2198 	return (0);
2199 }
2200 
2201 static int
2202 kqueue_fill_kinfo(struct file *fp, struct kinfo_file *kif, struct filedesc *fdp)
2203 {
2204 
2205 	kif->kf_type = KF_TYPE_KQUEUE;
2206 	return (0);
2207 }
2208 
2209 static void
2210 kqueue_wakeup(struct kqueue *kq)
2211 {
2212 	KQ_OWNED(kq);
2213 
2214 	if ((kq->kq_state & KQ_SLEEP) == KQ_SLEEP) {
2215 		kq->kq_state &= ~KQ_SLEEP;
2216 		wakeup(kq);
2217 	}
2218 	if ((kq->kq_state & KQ_SEL) == KQ_SEL) {
2219 		selwakeuppri(&kq->kq_sel, PSOCK);
2220 		if (!SEL_WAITING(&kq->kq_sel))
2221 			kq->kq_state &= ~KQ_SEL;
2222 	}
2223 	if (!knlist_empty(&kq->kq_sel.si_note))
2224 		kqueue_schedtask(kq);
2225 	if ((kq->kq_state & KQ_ASYNC) == KQ_ASYNC) {
2226 		pgsigio(&kq->kq_sigio, SIGIO, 0);
2227 	}
2228 }
2229 
2230 /*
2231  * Walk down a list of knotes, activating them if their event has triggered.
2232  *
2233  * There is a possibility to optimize in the case of one kq watching another.
2234  * Instead of scheduling a task to wake it up, you could pass enough state
2235  * down the chain to make up the parent kqueue.  Make this code functional
2236  * first.
2237  */
2238 void
2239 knote(struct knlist *list, long hint, int lockflags)
2240 {
2241 	struct kqueue *kq;
2242 	struct knote *kn, *tkn;
2243 	int error;
2244 
2245 	if (list == NULL)
2246 		return;
2247 
2248 	KNL_ASSERT_LOCK(list, lockflags & KNF_LISTLOCKED);
2249 
2250 	if ((lockflags & KNF_LISTLOCKED) == 0)
2251 		list->kl_lock(list->kl_lockarg);
2252 
2253 	/*
2254 	 * If we unlock the list lock (and enter influx), we can
2255 	 * eliminate the kqueue scheduling, but this will introduce
2256 	 * four lock/unlock's for each knote to test.  Also, marker
2257 	 * would be needed to keep iteration position, since filters
2258 	 * or other threads could remove events.
2259 	 */
2260 	SLIST_FOREACH_SAFE(kn, &list->kl_list, kn_selnext, tkn) {
2261 		kq = kn->kn_kq;
2262 		KQ_LOCK(kq);
2263 		if (kn_in_flux(kn) && (kn->kn_status & KN_SCAN) == 0) {
2264 			/*
2265 			 * Do not process the influx notes, except for
2266 			 * the influx coming from the kq unlock in the
2267 			 * kqueue_scan().  In the later case, we do
2268 			 * not interfere with the scan, since the code
2269 			 * fragment in kqueue_scan() locks the knlist,
2270 			 * and cannot proceed until we finished.
2271 			 */
2272 			KQ_UNLOCK(kq);
2273 		} else if ((lockflags & KNF_NOKQLOCK) != 0) {
2274 			kn_enter_flux(kn);
2275 			KQ_UNLOCK(kq);
2276 			error = kn->kn_fop->f_event(kn, hint);
2277 			KQ_LOCK(kq);
2278 			kn_leave_flux(kn);
2279 			if (error)
2280 				KNOTE_ACTIVATE(kn, 1);
2281 			KQ_UNLOCK_FLUX(kq);
2282 		} else {
2283 			if (kn->kn_fop->f_event(kn, hint))
2284 				KNOTE_ACTIVATE(kn, 1);
2285 			KQ_UNLOCK(kq);
2286 		}
2287 	}
2288 	if ((lockflags & KNF_LISTLOCKED) == 0)
2289 		list->kl_unlock(list->kl_lockarg);
2290 }
2291 
2292 /*
2293  * add a knote to a knlist
2294  */
2295 void
2296 knlist_add(struct knlist *knl, struct knote *kn, int islocked)
2297 {
2298 
2299 	KNL_ASSERT_LOCK(knl, islocked);
2300 	KQ_NOTOWNED(kn->kn_kq);
2301 	KASSERT(kn_in_flux(kn), ("knote %p not in flux", kn));
2302 	KASSERT((kn->kn_status & KN_DETACHED) != 0,
2303 	    ("knote %p was not detached", kn));
2304 	if (!islocked)
2305 		knl->kl_lock(knl->kl_lockarg);
2306 	SLIST_INSERT_HEAD(&knl->kl_list, kn, kn_selnext);
2307 	if (!islocked)
2308 		knl->kl_unlock(knl->kl_lockarg);
2309 	KQ_LOCK(kn->kn_kq);
2310 	kn->kn_knlist = knl;
2311 	kn->kn_status &= ~KN_DETACHED;
2312 	KQ_UNLOCK(kn->kn_kq);
2313 }
2314 
2315 static void
2316 knlist_remove_kq(struct knlist *knl, struct knote *kn, int knlislocked,
2317     int kqislocked)
2318 {
2319 
2320 	KASSERT(!kqislocked || knlislocked, ("kq locked w/o knl locked"));
2321 	KNL_ASSERT_LOCK(knl, knlislocked);
2322 	mtx_assert(&kn->kn_kq->kq_lock, kqislocked ? MA_OWNED : MA_NOTOWNED);
2323 	KASSERT(kqislocked || kn_in_flux(kn), ("knote %p not in flux", kn));
2324 	KASSERT((kn->kn_status & KN_DETACHED) == 0,
2325 	    ("knote %p was already detached", kn));
2326 	if (!knlislocked)
2327 		knl->kl_lock(knl->kl_lockarg);
2328 	SLIST_REMOVE(&knl->kl_list, kn, knote, kn_selnext);
2329 	kn->kn_knlist = NULL;
2330 	if (!knlislocked)
2331 		kn_list_unlock(knl);
2332 	if (!kqislocked)
2333 		KQ_LOCK(kn->kn_kq);
2334 	kn->kn_status |= KN_DETACHED;
2335 	if (!kqislocked)
2336 		KQ_UNLOCK(kn->kn_kq);
2337 }
2338 
2339 /*
2340  * remove knote from the specified knlist
2341  */
2342 void
2343 knlist_remove(struct knlist *knl, struct knote *kn, int islocked)
2344 {
2345 
2346 	knlist_remove_kq(knl, kn, islocked, 0);
2347 }
2348 
2349 int
2350 knlist_empty(struct knlist *knl)
2351 {
2352 
2353 	KNL_ASSERT_LOCKED(knl);
2354 	return (SLIST_EMPTY(&knl->kl_list));
2355 }
2356 
2357 static struct mtx knlist_lock;
2358 MTX_SYSINIT(knlist_lock, &knlist_lock, "knlist lock for lockless objects",
2359     MTX_DEF);
2360 static void knlist_mtx_lock(void *arg);
2361 static void knlist_mtx_unlock(void *arg);
2362 
2363 static void
2364 knlist_mtx_lock(void *arg)
2365 {
2366 
2367 	mtx_lock((struct mtx *)arg);
2368 }
2369 
2370 static void
2371 knlist_mtx_unlock(void *arg)
2372 {
2373 
2374 	mtx_unlock((struct mtx *)arg);
2375 }
2376 
2377 static void
2378 knlist_mtx_assert_locked(void *arg)
2379 {
2380 
2381 	mtx_assert((struct mtx *)arg, MA_OWNED);
2382 }
2383 
2384 static void
2385 knlist_mtx_assert_unlocked(void *arg)
2386 {
2387 
2388 	mtx_assert((struct mtx *)arg, MA_NOTOWNED);
2389 }
2390 
2391 static void
2392 knlist_rw_rlock(void *arg)
2393 {
2394 
2395 	rw_rlock((struct rwlock *)arg);
2396 }
2397 
2398 static void
2399 knlist_rw_runlock(void *arg)
2400 {
2401 
2402 	rw_runlock((struct rwlock *)arg);
2403 }
2404 
2405 static void
2406 knlist_rw_assert_locked(void *arg)
2407 {
2408 
2409 	rw_assert((struct rwlock *)arg, RA_LOCKED);
2410 }
2411 
2412 static void
2413 knlist_rw_assert_unlocked(void *arg)
2414 {
2415 
2416 	rw_assert((struct rwlock *)arg, RA_UNLOCKED);
2417 }
2418 
2419 void
2420 knlist_init(struct knlist *knl, void *lock, void (*kl_lock)(void *),
2421     void (*kl_unlock)(void *),
2422     void (*kl_assert_locked)(void *), void (*kl_assert_unlocked)(void *))
2423 {
2424 
2425 	if (lock == NULL)
2426 		knl->kl_lockarg = &knlist_lock;
2427 	else
2428 		knl->kl_lockarg = lock;
2429 
2430 	if (kl_lock == NULL)
2431 		knl->kl_lock = knlist_mtx_lock;
2432 	else
2433 		knl->kl_lock = kl_lock;
2434 	if (kl_unlock == NULL)
2435 		knl->kl_unlock = knlist_mtx_unlock;
2436 	else
2437 		knl->kl_unlock = kl_unlock;
2438 	if (kl_assert_locked == NULL)
2439 		knl->kl_assert_locked = knlist_mtx_assert_locked;
2440 	else
2441 		knl->kl_assert_locked = kl_assert_locked;
2442 	if (kl_assert_unlocked == NULL)
2443 		knl->kl_assert_unlocked = knlist_mtx_assert_unlocked;
2444 	else
2445 		knl->kl_assert_unlocked = kl_assert_unlocked;
2446 
2447 	knl->kl_autodestroy = 0;
2448 	SLIST_INIT(&knl->kl_list);
2449 }
2450 
2451 void
2452 knlist_init_mtx(struct knlist *knl, struct mtx *lock)
2453 {
2454 
2455 	knlist_init(knl, lock, NULL, NULL, NULL, NULL);
2456 }
2457 
2458 struct knlist *
2459 knlist_alloc(struct mtx *lock)
2460 {
2461 	struct knlist *knl;
2462 
2463 	knl = malloc(sizeof(struct knlist), M_KQUEUE, M_WAITOK);
2464 	knlist_init_mtx(knl, lock);
2465 	return (knl);
2466 }
2467 
2468 void
2469 knlist_init_rw_reader(struct knlist *knl, struct rwlock *lock)
2470 {
2471 
2472 	knlist_init(knl, lock, knlist_rw_rlock, knlist_rw_runlock,
2473 	    knlist_rw_assert_locked, knlist_rw_assert_unlocked);
2474 }
2475 
2476 void
2477 knlist_destroy(struct knlist *knl)
2478 {
2479 
2480 	KASSERT(KNLIST_EMPTY(knl),
2481 	    ("destroying knlist %p with knotes on it", knl));
2482 }
2483 
2484 void
2485 knlist_detach(struct knlist *knl)
2486 {
2487 
2488 	KNL_ASSERT_LOCKED(knl);
2489 	knl->kl_autodestroy = 1;
2490 	if (knlist_empty(knl)) {
2491 		knlist_destroy(knl);
2492 		free(knl, M_KQUEUE);
2493 	}
2494 }
2495 
2496 /*
2497  * Even if we are locked, we may need to drop the lock to allow any influx
2498  * knotes time to "settle".
2499  */
2500 void
2501 knlist_cleardel(struct knlist *knl, struct thread *td, int islocked, int killkn)
2502 {
2503 	struct knote *kn, *kn2;
2504 	struct kqueue *kq;
2505 
2506 	KASSERT(!knl->kl_autodestroy, ("cleardel for autodestroy %p", knl));
2507 	if (islocked)
2508 		KNL_ASSERT_LOCKED(knl);
2509 	else {
2510 		KNL_ASSERT_UNLOCKED(knl);
2511 again:		/* need to reacquire lock since we have dropped it */
2512 		knl->kl_lock(knl->kl_lockarg);
2513 	}
2514 
2515 	SLIST_FOREACH_SAFE(kn, &knl->kl_list, kn_selnext, kn2) {
2516 		kq = kn->kn_kq;
2517 		KQ_LOCK(kq);
2518 		if (kn_in_flux(kn)) {
2519 			KQ_UNLOCK(kq);
2520 			continue;
2521 		}
2522 		knlist_remove_kq(knl, kn, 1, 1);
2523 		if (killkn) {
2524 			kn_enter_flux(kn);
2525 			KQ_UNLOCK(kq);
2526 			knote_drop_detached(kn, td);
2527 		} else {
2528 			/* Make sure cleared knotes disappear soon */
2529 			kn->kn_flags |= EV_EOF | EV_ONESHOT;
2530 			KQ_UNLOCK(kq);
2531 		}
2532 		kq = NULL;
2533 	}
2534 
2535 	if (!SLIST_EMPTY(&knl->kl_list)) {
2536 		/* there are still in flux knotes remaining */
2537 		kn = SLIST_FIRST(&knl->kl_list);
2538 		kq = kn->kn_kq;
2539 		KQ_LOCK(kq);
2540 		KASSERT(kn_in_flux(kn), ("knote removed w/o list lock"));
2541 		knl->kl_unlock(knl->kl_lockarg);
2542 		kq->kq_state |= KQ_FLUXWAIT;
2543 		msleep(kq, &kq->kq_lock, PSOCK | PDROP, "kqkclr", 0);
2544 		kq = NULL;
2545 		goto again;
2546 	}
2547 
2548 	if (islocked)
2549 		KNL_ASSERT_LOCKED(knl);
2550 	else {
2551 		knl->kl_unlock(knl->kl_lockarg);
2552 		KNL_ASSERT_UNLOCKED(knl);
2553 	}
2554 }
2555 
2556 /*
2557  * Remove all knotes referencing a specified fd must be called with FILEDESC
2558  * lock.  This prevents a race where a new fd comes along and occupies the
2559  * entry and we attach a knote to the fd.
2560  */
2561 void
2562 knote_fdclose(struct thread *td, int fd)
2563 {
2564 	struct filedesc *fdp = td->td_proc->p_fd;
2565 	struct kqueue *kq;
2566 	struct knote *kn;
2567 	int influx;
2568 
2569 	FILEDESC_XLOCK_ASSERT(fdp);
2570 
2571 	/*
2572 	 * We shouldn't have to worry about new kevents appearing on fd
2573 	 * since filedesc is locked.
2574 	 */
2575 	TAILQ_FOREACH(kq, &fdp->fd_kqlist, kq_list) {
2576 		KQ_LOCK(kq);
2577 
2578 again:
2579 		influx = 0;
2580 		while (kq->kq_knlistsize > fd &&
2581 		    (kn = SLIST_FIRST(&kq->kq_knlist[fd])) != NULL) {
2582 			if (kn_in_flux(kn)) {
2583 				/* someone else might be waiting on our knote */
2584 				if (influx)
2585 					wakeup(kq);
2586 				kq->kq_state |= KQ_FLUXWAIT;
2587 				msleep(kq, &kq->kq_lock, PSOCK, "kqflxwt", 0);
2588 				goto again;
2589 			}
2590 			kn_enter_flux(kn);
2591 			KQ_UNLOCK(kq);
2592 			influx = 1;
2593 			knote_drop(kn, td);
2594 			KQ_LOCK(kq);
2595 		}
2596 		KQ_UNLOCK_FLUX(kq);
2597 	}
2598 }
2599 
2600 static int
2601 knote_attach(struct knote *kn, struct kqueue *kq)
2602 {
2603 	struct klist *list;
2604 
2605 	KASSERT(kn_in_flux(kn), ("knote %p not marked influx", kn));
2606 	KQ_OWNED(kq);
2607 
2608 	if ((kq->kq_state & KQ_CLOSING) != 0)
2609 		return (EBADF);
2610 	if (kn->kn_fop->f_isfd) {
2611 		if (kn->kn_id >= kq->kq_knlistsize)
2612 			return (ENOMEM);
2613 		list = &kq->kq_knlist[kn->kn_id];
2614 	} else {
2615 		if (kq->kq_knhash == NULL)
2616 			return (ENOMEM);
2617 		list = &kq->kq_knhash[KN_HASH(kn->kn_id, kq->kq_knhashmask)];
2618 	}
2619 	SLIST_INSERT_HEAD(list, kn, kn_link);
2620 	return (0);
2621 }
2622 
2623 static void
2624 knote_drop(struct knote *kn, struct thread *td)
2625 {
2626 
2627 	if ((kn->kn_status & KN_DETACHED) == 0)
2628 		kn->kn_fop->f_detach(kn);
2629 	knote_drop_detached(kn, td);
2630 }
2631 
2632 static void
2633 knote_drop_detached(struct knote *kn, struct thread *td)
2634 {
2635 	struct kqueue *kq;
2636 	struct klist *list;
2637 
2638 	kq = kn->kn_kq;
2639 
2640 	KASSERT((kn->kn_status & KN_DETACHED) != 0,
2641 	    ("knote %p still attached", kn));
2642 	KQ_NOTOWNED(kq);
2643 
2644 	KQ_LOCK(kq);
2645 	KASSERT(kn->kn_influx == 1,
2646 	    ("knote_drop called on %p with influx %d", kn, kn->kn_influx));
2647 
2648 	if (kn->kn_fop->f_isfd)
2649 		list = &kq->kq_knlist[kn->kn_id];
2650 	else
2651 		list = &kq->kq_knhash[KN_HASH(kn->kn_id, kq->kq_knhashmask)];
2652 
2653 	if (!SLIST_EMPTY(list))
2654 		SLIST_REMOVE(list, kn, knote, kn_link);
2655 	if (kn->kn_status & KN_QUEUED)
2656 		knote_dequeue(kn);
2657 	KQ_UNLOCK_FLUX(kq);
2658 
2659 	if (kn->kn_fop->f_isfd) {
2660 		fdrop(kn->kn_fp, td);
2661 		kn->kn_fp = NULL;
2662 	}
2663 	kqueue_fo_release(kn->kn_kevent.filter);
2664 	kn->kn_fop = NULL;
2665 	knote_free(kn);
2666 }
2667 
2668 static void
2669 knote_enqueue(struct knote *kn)
2670 {
2671 	struct kqueue *kq = kn->kn_kq;
2672 
2673 	KQ_OWNED(kn->kn_kq);
2674 	KASSERT((kn->kn_status & KN_QUEUED) == 0, ("knote already queued"));
2675 
2676 	TAILQ_INSERT_TAIL(&kq->kq_head, kn, kn_tqe);
2677 	kn->kn_status |= KN_QUEUED;
2678 	kq->kq_count++;
2679 	kqueue_wakeup(kq);
2680 }
2681 
2682 static void
2683 knote_dequeue(struct knote *kn)
2684 {
2685 	struct kqueue *kq = kn->kn_kq;
2686 
2687 	KQ_OWNED(kn->kn_kq);
2688 	KASSERT(kn->kn_status & KN_QUEUED, ("knote not queued"));
2689 
2690 	TAILQ_REMOVE(&kq->kq_head, kn, kn_tqe);
2691 	kn->kn_status &= ~KN_QUEUED;
2692 	kq->kq_count--;
2693 }
2694 
2695 static void
2696 knote_init(void)
2697 {
2698 
2699 	knote_zone = uma_zcreate("KNOTE", sizeof(struct knote), NULL, NULL,
2700 	    NULL, NULL, UMA_ALIGN_PTR, 0);
2701 }
2702 SYSINIT(knote, SI_SUB_PSEUDO, SI_ORDER_ANY, knote_init, NULL);
2703 
2704 static struct knote *
2705 knote_alloc(int mflag)
2706 {
2707 
2708 	return (uma_zalloc(knote_zone, mflag | M_ZERO));
2709 }
2710 
2711 static void
2712 knote_free(struct knote *kn)
2713 {
2714 
2715 	uma_zfree(knote_zone, kn);
2716 }
2717 
2718 /*
2719  * Register the kev w/ the kq specified by fd.
2720  */
2721 int
2722 kqfd_register(int fd, struct kevent *kev, struct thread *td, int mflag)
2723 {
2724 	struct kqueue *kq;
2725 	struct file *fp;
2726 	cap_rights_t rights;
2727 	int error;
2728 
2729 	error = fget(td, fd, cap_rights_init(&rights, CAP_KQUEUE_CHANGE), &fp);
2730 	if (error != 0)
2731 		return (error);
2732 	if ((error = kqueue_acquire(fp, &kq)) != 0)
2733 		goto noacquire;
2734 
2735 	error = kqueue_register(kq, kev, td, mflag);
2736 	kqueue_release(kq, 0);
2737 
2738 noacquire:
2739 	fdrop(fp, td);
2740 	return (error);
2741 }
2742