]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/kern/kern_event.c
xnu-517.tar.gz
[apple/xnu.git] / bsd / kern / kern_event.c
index aa0d6326f113ae4b585231622f82ec55a140c09e..0b1425c07dce09a4d712db290c7c6c3f8294b8ac 100644 (file)
 /*
 /*
- * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2000-2003 Apple Computer, Inc. All rights reserved.
  *
  * @APPLE_LICENSE_HEADER_START@
  * 
  *
  * @APPLE_LICENSE_HEADER_START@
  * 
- * The contents of this file constitute Original Code as defined in and
- * are subject to the Apple Public Source License Version 1.1 (the
- * "License").  You may not use this file except in compliance with the
- * License.  Please obtain a copy of the License at
- * http://www.apple.com/publicsource and read it before using this file.
+ * Copyright (c) 1999-2003 Apple Computer, Inc.  All Rights Reserved.
  * 
  * 
- * This Original Code and all software distributed under the License are
- * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
+ * This file contains Original Code and/or Modifications of Original Code
+ * as defined in and that are subject to the Apple Public Source License
+ * Version 2.0 (the 'License'). You may not use this file except in
+ * compliance with the License. Please obtain a copy of the License at
+ * http://www.opensource.apple.com/apsl/ and read it before using this
+ * file.
+ * 
+ * The Original Code and all software distributed under the License are
+ * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
- * License for the specific language governing rights and limitations
- * under the License.
+ * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
+ * Please see the License for the specific language governing rights and
+ * limitations under the License.
  * 
  * @APPLE_LICENSE_HEADER_END@
  *
  */
  * 
  * @APPLE_LICENSE_HEADER_END@
  *
  */
+/*-
+ * Copyright (c) 1999,2000,2001 Jonathan Lemon <jlemon@FreeBSD.org>
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ */
 /*
  *     @(#)kern_event.c       1.0 (3/31/2000)
  */
 
 /*
  *     @(#)kern_event.c       1.0 (3/31/2000)
  */
 
+#include <sys/param.h>
+#include <sys/systm.h>
+#include <sys/filedesc.h>
+#include <sys/kernel.h>
+#include <sys/proc.h>
+#include <sys/malloc.h> 
+#include <sys/unistd.h>
+#include <sys/file.h>
+#include <sys/fcntl.h>
+#include <sys/select.h>
+#include <sys/queue.h>
+#include <sys/event.h>
+#include <sys/eventvar.h>
+#include <sys/protosw.h>
+#include <sys/socket.h>
+#include <sys/socketvar.h>
+#include <sys/stat.h>
+#include <sys/sysctl.h>
+#include <sys/uio.h>
+
+#include <kern/zalloc.h>
+
+MALLOC_DEFINE(M_KQUEUE, "kqueue", "memory for kqueue system");
+
+static int     kqueue_scan(struct file *fp, int maxevents,
+                   struct kevent *ulistp, const struct timespec *timeout,
+                   register_t *retval, struct proc *p);
+static void    kqueue_wakeup(struct kqueue *kq);
+
+static int     kqueue_read __P((struct file *fp, struct uio *uio,
+                   struct ucred *cred, int flags, struct proc *p));
+static int     kqueue_write __P((struct file *fp, struct uio *uio,
+                   struct ucred *cred, int flags, struct proc *p));
+static int     kqueue_ioctl __P((struct file *fp, u_long com, caddr_t data,
+                   struct proc *p));
+static int     kqueue_select __P((struct file *fp, int which, void *wql, 
+                   struct proc *p));
+static int     kqueue_close __P((struct file *fp, struct proc *p));
+static int     kqueue_kqfilter __P((struct file *fp, struct knote *kn, struct proc *p));
+
+static struct fileops kqueueops = {
+       kqueue_read,
+       kqueue_write,
+       kqueue_ioctl,
+       kqueue_select,
+       kqueue_close,
+       kqueue_kqfilter
+};
+
+static void    knote_fdpattach(struct knote *kn, struct filedesc *fdp);
+static void    knote_drop(struct knote *kn, struct proc *p);
+static void    knote_enqueue(struct knote *kn);
+static void    knote_dequeue(struct knote *kn);
+static struct  knote *knote_alloc(void);
+static void    knote_free(struct knote *kn);
+
+static int     filt_fileattach(struct knote *kn);
+static struct filterops file_filtops =
+       { 1, filt_fileattach, NULL, NULL };
+
+static void    filt_kqdetach(struct knote *kn);
+static int     filt_kqueue(struct knote *kn, long hint);
+static struct filterops kqread_filtops =
+       { 1, NULL, filt_kqdetach, filt_kqueue };
+
+/*
+ * JMM - placeholder for not-yet-implemented filters
+ */ 
+static int     filt_badattach(struct knote *kn);
+static struct filterops bad_filtops =
+       { 0, filt_badattach, 0 , 0 };
+
+static int     filt_procattach(struct knote *kn);
+static void    filt_procdetach(struct knote *kn);
+static int     filt_proc(struct knote *kn, long hint);
+
+static struct filterops proc_filtops =
+       { 0, filt_procattach, filt_procdetach, filt_proc };
+
+extern struct filterops fs_filtops;
+
+extern struct filterops sig_filtops;
+
+#if 0
+/* JMM - We don't implement these now */
+static void    filt_timerexpire(void *knx);
+static int     filt_timerattach(struct knote *kn);
+static void    filt_timerdetach(struct knote *kn);
+static int     filt_timer(struct knote *kn, long hint);
+
+static struct filterops timer_filtops =
+       { 0, filt_timerattach, filt_timerdetach, filt_timer };
+
+static int             kq_ncallouts = 0;
+static int             kq_calloutmax = (4 * 1024);
+
+SYSCTL_INT(_kern, OID_AUTO, kq_calloutmax, CTLFLAG_RW,
+    &kq_calloutmax, 0, "Maximum number of callouts allocated for kqueue");
+#endif /* 0 */
+
+static zone_t  knote_zone;
+
+#define KNOTE_ACTIVATE(kn) do {                                        \
+       kn->kn_status |= KN_ACTIVE;                                     \
+       if ((kn->kn_status & (KN_QUEUED | KN_DISABLED)) == 0)           \
+               knote_enqueue(kn);                                      \
+} while(0)
+
+#define        KN_HASHSIZE             64              /* XXX should be tunable */
+#define KN_HASH(val, mask)     (((val) ^ (val >> 8)) & (mask))
+
+#if 0
+extern struct filterops aio_filtops;
+#endif
+
+/*
+ * Table for for all system-defined filters.
+ */
+static struct filterops *sysfilt_ops[] = {
+       &file_filtops,                  /* EVFILT_READ */
+       &file_filtops,                  /* EVFILT_WRITE */
+#if 0
+       &aio_filtops,                   /* EVFILT_AIO */
+#else
+       &bad_filtops,                   /* EVFILT_AIO */
+#endif
+       &file_filtops,                  /* EVFILT_VNODE */
+       &proc_filtops,                  /* EVFILT_PROC */
+       &sig_filtops,                   /* EVFILT_SIGNAL */
+#if 0
+       &timer_filtops,                 /* EVFILT_TIMER */
+#else
+       &bad_filtops,                   /* EVFILT_TIMER */
+#endif
+       &bad_filtops,                   /* EVFILT_MACHPORT */
+       &fs_filtops             /* EVFILT_FS */
+};
+
+static int
+filt_fileattach(struct knote *kn)
+{
+       
+       return (fo_kqfilter(kn->kn_fp, kn, current_proc()));
+}
+
+static void
+filt_kqdetach(struct knote *kn)
+{
+       struct kqueue *kq = (struct kqueue *)kn->kn_fp->f_data;
+
+       if (kq->kq_state & KQ_SEL)
+         return;
+
+       KNOTE_DETACH(&kq->kq_sel.si_note, kn);
+}
+
+/*ARGSUSED*/
+static int
+filt_kqueue(struct knote *kn, long hint)
+{
+       struct kqueue *kq = (struct kqueue *)kn->kn_fp->f_data;
+
+       kn->kn_data = kq->kq_count;
+       return (kn->kn_data > 0);
+}
+
+static int
+filt_procattach(struct knote *kn)
+{
+       struct proc *p;
+
+       p = pfind(kn->kn_id);
+       if (p == NULL)
+               return (ESRCH);
+       if (! PRISON_CHECK(current_proc(), p))
+               return (EACCES);
+
+       kn->kn_ptr.p_proc = p;
+       kn->kn_flags |= EV_CLEAR;               /* automatically set */
+
+       /*
+        * internal flag indicating registration done by kernel
+        */
+       if (kn->kn_flags & EV_FLAG1) {
+               kn->kn_data = kn->kn_sdata;             /* ppid */
+               kn->kn_fflags = NOTE_CHILD;
+               kn->kn_flags &= ~EV_FLAG1;
+       }
+
+       /* XXX lock the proc here while adding to the list? */
+       KNOTE_ATTACH(&p->p_klist, kn);
+
+       return (0);
+}
+
+/*
+ * The knote may be attached to a different process, which may exit,
+ * leaving nothing for the knote to be attached to.  So when the process
+ * exits, the knote is marked as DETACHED and also flagged as ONESHOT so
+ * it will be deleted when read out.  However, as part of the knote deletion,
+ * this routine is called, so a check is needed to avoid actually performing
+ * a detach, because the original process does not exist any more.
+ */
+static void
+filt_procdetach(struct knote *kn)
+{
+       struct proc *p = kn->kn_ptr.p_proc;
+
+       if (kn->kn_status & KN_DETACHED)
+               return;
+
+       /* XXX locking?  this might modify another process. */
+       KNOTE_DETACH(&p->p_klist, kn);
+}
+
+static int
+filt_proc(struct knote *kn, long hint)
+{
+       u_int event;
+
+       /*
+        * mask off extra data
+        */
+       event = (u_int)hint & NOTE_PCTRLMASK;
+
+       /*
+        * if the user is interested in this event, record it.
+        */
+       if (kn->kn_sfflags & event)
+               kn->kn_fflags |= event;
+
+       /*
+        * process is gone, so flag the event as finished.
+        */
+       if (event == NOTE_EXIT) {
+               kn->kn_status |= KN_DETACHED;
+               kn->kn_flags |= (EV_EOF | EV_ONESHOT); 
+               return (1);
+       }
+
+       /*
+        * process forked, and user wants to track the new process,
+        * so attach a new knote to it, and immediately report an
+        * event with the parent's pid.
+        */
+       if ((event == NOTE_FORK) && (kn->kn_sfflags & NOTE_TRACK)) {
+               struct kevent kev;
+               int error;
+
+               /*
+                * register knote with new process.
+                */
+               kev.ident = hint & NOTE_PDATAMASK;      /* pid */
+               kev.filter = kn->kn_filter;
+               kev.flags = kn->kn_flags | EV_ADD | EV_ENABLE | EV_FLAG1;
+               kev.fflags = kn->kn_sfflags;
+               kev.data = kn->kn_id;                   /* parent */
+               kev.udata = kn->kn_kevent.udata;        /* preserve udata */
+               error = kqueue_register(kn->kn_kq, &kev, NULL);
+               if (error)
+                       kn->kn_fflags |= NOTE_TRACKERR;
+       }
+
+       return (kn->kn_fflags != 0);
+}
+
+#if 0
+static void
+filt_timerexpire(void *knx)
+{
+       struct knote *kn = knx;
+       struct callout *calloutp;
+       struct timeval tv;
+       int tticks;
+
+       kn->kn_data++;
+       KNOTE_ACTIVATE(kn);
+
+       if ((kn->kn_flags & EV_ONESHOT) == 0) {
+               tv.tv_sec = kn->kn_sdata / 1000;
+               tv.tv_usec = (kn->kn_sdata % 1000) * 1000;
+               tticks = tvtohz(&tv);
+               calloutp = (struct callout *)kn->kn_hook;
+               callout_reset(calloutp, tticks, filt_timerexpire, kn);
+       }
+}
+
+/*
+ * data contains amount of time to sleep, in milliseconds
+ */ 
+static int
+filt_timerattach(struct knote *kn)
+{
+       struct callout *calloutp;
+       struct timeval tv;
+       int tticks;
+
+       if (kq_ncallouts >= kq_calloutmax)
+               return (ENOMEM);
+       kq_ncallouts++;
+
+       tv.tv_sec = kn->kn_sdata / 1000;
+       tv.tv_usec = (kn->kn_sdata % 1000) * 1000;
+       tticks = tvtohz(&tv);
+
+       kn->kn_flags |= EV_CLEAR;               /* automatically set */
+       MALLOC(calloutp, struct callout *, sizeof(*calloutp),
+           M_KQUEUE, M_WAITOK);
+       callout_init(calloutp);
+       callout_reset(calloutp, tticks, filt_timerexpire, kn);
+       kn->kn_hook = (caddr_t)calloutp;
+
+       return (0);
+}
+
+static void
+filt_timerdetach(struct knote *kn)
+{
+       struct callout *calloutp;
+
+       calloutp = (struct callout *)kn->kn_hook;
+       callout_stop(calloutp);
+       FREE(calloutp, M_KQUEUE);
+       kq_ncallouts--;
+}
+
+static int
+filt_timer(struct knote *kn, long hint)
+{
+
+       return (kn->kn_data != 0);
+}
+#endif /* 0 */
+
+/*
+ * JMM - placeholder for not-yet-implemented filters
+ */ 
+static int
+filt_badattach(struct knote *kn)
+{
+       return(EOPNOTSUPP);
+}
+
+#ifndef _SYS_SYSPROTO_H_
+struct kqueue_args {
+       int dummy;
+};
+#endif
+
+int
+kqueue(struct proc *p, struct kqueue_args *uap, register_t *retval)
+{
+       struct filedesc *fdp = p->p_fd;
+       struct kqueue *kq;
+       struct file *fp;
+       int fd, error;
+
+       error = falloc(p, &fp, &fd);
+       if (error)
+               return (error);
+       fp->f_flag = FREAD | FWRITE;
+       fp->f_type = DTYPE_KQUEUE;
+       fp->f_ops = &kqueueops;
+       kq = (struct kqueue *)_MALLOC(sizeof(struct kqueue), M_KQUEUE, M_WAITOK | M_ZERO);
+       TAILQ_INIT(&kq->kq_head);
+       fp->f_data = (caddr_t)kq;
+       *retval = fd;
+       if (fdp->fd_knlistsize < 0)
+               fdp->fd_knlistsize = 0;         /* this process has a kq */
+       kq->kq_fdp = fdp;
+       return (error);
+}
+
+#ifndef _SYS_SYSPROTO_H_
+struct kqueue_portset_np_args {
+       int     fd;
+};
+#endif
+int
+kqueue_portset_np(struct proc *p, struct kqueue_portset_np_args *uap, register_t *retval)
+{
+               /* JMM - Placeholder for now */
+               return (EOPNOTSUPP);
+}
+
+#ifndef _SYS_SYSPROTO_H_
+struct kqueue_from_portset_np_args {
+       int     fd;
+};
+#endif
+int
+kqueue_from_portset_np(struct proc *p, struct kqueue_from_portset_np_args *uap, register_t *retval)
+{
+               /* JMM - Placeholder for now */
+               return (EOPNOTSUPP);
+}
+
+#if !0
+/* JMM - We don't implement this yet */
+#define fhold(fp)
+#define fdrop(fp, p)
+#endif /* !0 */
+
+#ifndef _SYS_SYSPROTO_H_
+struct kevent_args {
+       int     fd;
+       const struct kevent *changelist;
+       int     nchanges;
+       struct  kevent *eventlist;
+       int     nevents;
+       const struct timespec *timeout;
+};
+#endif
+int
+kevent(struct proc *p, struct kevent_args *uap, register_t *retval)
+{
+       struct filedesc* fdp = p->p_fd;
+       struct kqueue *kq;
+       struct file *fp = NULL;
+       struct timespec ts;
+       int i, nerrors, error;
+
+       if (uap->timeout != NULL) {
+               error = copyin((caddr_t)uap->timeout, (caddr_t)&ts, sizeof(ts));
+               if (error)
+                       goto done;
+               uap->timeout = &ts;
+       }
+
+        if (((u_int)uap->fd) >= fdp->fd_nfiles ||
+            (fp = fdp->fd_ofiles[uap->fd]) == NULL ||
+           (fp->f_type != DTYPE_KQUEUE))
+               return (EBADF);
+
+       fhold(fp);
+
+       kq = (struct kqueue *)fp->f_data;
+       nerrors = 0;
+
+       while (uap->nchanges > 0) {
+               int i;
+               int n = uap->nchanges > KQ_NEVENTS ? KQ_NEVENTS : uap->nchanges;
+               struct kevent kq_kev[n];
+
+               error = copyin((caddr_t)uap->changelist, (caddr_t)kq_kev,
+                   n * sizeof(struct kevent));
+               if (error)
+                       goto done;
+               for (i = 0; i < n; i++) {
+                       struct kevent *kevp = &kq_kev[i];
+
+                       kevp->flags &= ~EV_SYSFLAGS;
+                       error = kqueue_register(kq, kevp, p);
+                       if (error) {
+                               if (uap->nevents != 0) {
+                                       kevp->flags = EV_ERROR;
+                                       kevp->data = error;
+                                       (void) copyout((caddr_t)kevp,
+                                           (caddr_t)uap->eventlist,
+                                           sizeof(*kevp));
+                                       uap->eventlist++;
+                                       uap->nevents--;
+                                       nerrors++;
+                               } else {
+                                       goto done;
+                               }
+                       }
+               }
+               uap->nchanges -= n;
+               uap->changelist += n;
+       }
+       if (nerrors) {
+               *retval = nerrors;
+               error = 0;
+               goto done;
+       }
+
+       error = kqueue_scan(fp, uap->nevents, uap->eventlist, uap->timeout, retval, p);
+done:
+       if (fp != NULL)
+               fdrop(fp, p);
+       return (error);
+}
+
+int
+kqueue_register(struct kqueue *kq, struct kevent *kev, struct proc *p)
+{
+       struct filedesc *fdp = kq->kq_fdp;
+       struct filterops *fops;
+       struct file *fp = NULL;
+       struct knote *kn = NULL;
+       int s, error = 0;
+
+       if (kev->filter < 0) {
+               if (kev->filter + EVFILT_SYSCOUNT < 0)
+                       return (EINVAL);
+               fops = sysfilt_ops[~kev->filter];       /* to 0-base index */
+       } else {
+               /*
+                * XXX
+                * filter attach routine is responsible for insuring that
+                * the identifier can be attached to it.
+                */
+               printf("unknown filter: %d\n", kev->filter);
+               return (EINVAL);
+       }
+
+       if (fops->f_isfd) {
+               /* validate descriptor */
+               if ((u_int)kev->ident >= fdp->fd_nfiles ||
+                   (fp = fdp->fd_ofiles[kev->ident]) == NULL)
+                       return (EBADF);
+               fhold(fp);
+
+               if (kev->ident < fdp->fd_knlistsize) {
+                       SLIST_FOREACH(kn, &fdp->fd_knlist[kev->ident], kn_link)
+                               if (kq == kn->kn_kq &&
+                                   kev->filter == kn->kn_filter)
+                                       break;
+               }
+       } else {
+               if (fdp->fd_knhashmask != 0) {
+                       struct klist *list;
+                       
+                       list = &fdp->fd_knhash[
+                           KN_HASH((u_long)kev->ident, fdp->fd_knhashmask)];
+                       SLIST_FOREACH(kn, list, kn_link)
+                               if (kev->ident == kn->kn_id &&
+                                   kq == kn->kn_kq &&
+                                   kev->filter == kn->kn_filter)
+                                       break;
+               }
+       }
+
+       if (kn == NULL && ((kev->flags & EV_ADD) == 0)) {
+               error = ENOENT;
+               goto done;
+       }
+
+       /*
+        * kn now contains the matching knote, or NULL if no match
+        */
+       if (kev->flags & EV_ADD) {
+
+               if (kn == NULL) {
+                       kn = knote_alloc();
+                       if (kn == NULL) {
+                               error = ENOMEM;
+                               goto done;
+                       }
+                       kn->kn_fp = fp;
+                       kn->kn_kq = kq;
+                       kn->kn_fop = fops;
+
+                       /*
+                        * apply reference count to knote structure, and
+                        * do not release it at the end of this routine.
+                        */
+                       fp = NULL;
+
+                       kn->kn_sfflags = kev->fflags;
+                       kn->kn_sdata = kev->data;
+                       kev->fflags = 0;
+                       kev->data = 0;
+                       kn->kn_kevent = *kev;
+
+                       knote_fdpattach(kn, fdp);
+                       if ((error = fops->f_attach(kn)) != 0) {
+                               knote_drop(kn, p);
+                               goto done;
+                       }
+               } else {
+                       /*
+                        * The user may change some filter values after the
+                        * initial EV_ADD, but doing so will not reset any 
+                        * filter which have already been triggered.
+                        */
+                       kn->kn_sfflags = kev->fflags;
+                       kn->kn_sdata = kev->data;
+                       kn->kn_kevent.udata = kev->udata;
+               }
+
+               s = splhigh();
+               if (kn->kn_fop->f_event(kn, 0))
+                       KNOTE_ACTIVATE(kn);
+               splx(s);
+
+       } else if (kev->flags & EV_DELETE) {
+               kn->kn_fop->f_detach(kn);
+               knote_drop(kn, p);
+               goto done;
+       }
+
+       if ((kev->flags & EV_DISABLE) &&
+           ((kn->kn_status & KN_DISABLED) == 0)) {
+               s = splhigh();
+               kn->kn_status |= KN_DISABLED;
+               splx(s);
+       }
+
+       if ((kev->flags & EV_ENABLE) && (kn->kn_status & KN_DISABLED)) {
+               s = splhigh();
+               kn->kn_status &= ~KN_DISABLED;
+               if ((kn->kn_status & KN_ACTIVE) &&
+                   ((kn->kn_status & KN_QUEUED) == 0))
+                       knote_enqueue(kn);
+               splx(s);
+       }
+
+done:
+       if (fp != NULL)
+               fdrop(fp, p);
+       return (error);
+}
+
+static int
+kqueue_scan(struct file *fp, int maxevents, struct kevent *ulistp,
+       const struct timespec *tsp, register_t *retval, struct proc *p)
+{
+       struct kqueue *kq = (struct kqueue *)fp->f_data;
+       struct timeval atv, rtv, ttv;
+       int s, count, timeout, error = 0;
+       struct knote marker;
+
+       count = maxevents;
+       if (count == 0)
+               goto done;
+
+       if (tsp != NULL) {
+               TIMESPEC_TO_TIMEVAL(&atv, tsp);
+               if (itimerfix(&atv)) {
+                       error = EINVAL;
+                       goto done;
+               }
+               if (tsp->tv_sec == 0 && tsp->tv_nsec == 0)
+                       timeout = -1;
+               else 
+                       timeout = atv.tv_sec > 24 * 60 * 60 ?
+                           24 * 60 * 60 * hz : tvtohz(&atv);
+               getmicrouptime(&rtv);
+               timevaladd(&atv, &rtv);
+       } else {
+               atv.tv_sec = 0;
+               atv.tv_usec = 0;
+               timeout = 0;
+       }
+       goto start;
+
+retry:
+       if (atv.tv_sec || atv.tv_usec) {
+               getmicrouptime(&rtv);
+               if (timevalcmp(&rtv, &atv, >=))
+                       goto done;
+               ttv = atv;
+               timevalsub(&ttv, &rtv);
+               timeout = ttv.tv_sec > 24 * 60 * 60 ?
+                       24 * 60 * 60 * hz : tvtohz(&ttv);
+       }
+
+start:
+       s = splhigh();
+       if (kq->kq_count == 0) {
+               if (timeout < 0) { 
+                       error = EWOULDBLOCK;
+               } else {
+                       kq->kq_state |= KQ_SLEEP;
+                       error = tsleep(kq, PSOCK | PCATCH, "kqread", timeout);
+               }
+               splx(s);
+               if (error == 0)
+                       goto retry;
+               /* don't restart after signals... */
+               if (error == ERESTART)
+                       error = EINTR;
+               else if (error == EWOULDBLOCK)
+                       error = 0;
+               goto done;
+       }
+
+       /* JMM - This marker trick doesn't work with multiple threads */
+       TAILQ_INSERT_TAIL(&kq->kq_head, &marker, kn_tqe); 
+       while (count) {
+               int maxkev = (count > KQ_NEVENTS) ? KQ_NEVENTS : count;
+               struct kevent kq_kev[maxkev];
+               struct kevent *kevp = kq_kev;
+               struct knote *kn;
+               int nkev = 0;
+
+               while (nkev < maxkev) {
+                       kn = TAILQ_FIRST(&kq->kq_head);
+                       TAILQ_REMOVE(&kq->kq_head, kn, kn_tqe); 
+                       if (kn == &marker) {
+                               if (count == maxevents)
+                                       goto retry;
+                               break;
+                       } else if (kn->kn_status & KN_DISABLED) {
+                               kn->kn_status &= ~KN_QUEUED;
+                               kq->kq_count--;
+                               continue;
+                       } else if ((kn->kn_flags & EV_ONESHOT) == 0 &&
+                                  kn->kn_fop->f_event(kn, 0) == 0) {
+                               kn->kn_status &= ~(KN_QUEUED | KN_ACTIVE);
+                               kq->kq_count--;
+                               continue;
+                       }
+
+                       *kevp = kn->kn_kevent;
+                       kevp++;
+                       nkev++;
+                       count--;
+
+                       if (kn->kn_flags & EV_ONESHOT) {
+                               kn->kn_status &= ~KN_QUEUED;
+                               kq->kq_count--;
+                               splx(s);
+                               kn->kn_fop->f_detach(kn);
+                               knote_drop(kn, p);
+                               s = splhigh();
+                       } else if (kn->kn_flags & EV_CLEAR) {
+                               kn->kn_data = 0;
+                               kn->kn_fflags = 0;
+                               kn->kn_status &= ~(KN_QUEUED | KN_ACTIVE);
+                               kq->kq_count--;
+                       } else {
+                               TAILQ_INSERT_TAIL(&kq->kq_head, kn, kn_tqe); 
+                       }
+               }
+               splx(s);
+               error = copyout((caddr_t)kq_kev, (caddr_t)ulistp,
+                               sizeof(struct kevent) * nkev);
+               if (kn == &marker)
+                       goto done;
+               ulistp += nkev;
+               s = splhigh();
+               if (error)
+                       break;
+       }
+       TAILQ_REMOVE(&kq->kq_head, &marker, kn_tqe); 
+       splx(s);
+done:
+        *retval = maxevents - count;
+       return (error);
+}
+
+/*
+ * XXX
+ * This could be expanded to call kqueue_scan, if desired.
+ */
+/*ARGSUSED*/
+static int
+kqueue_read(struct file *fp, struct uio *uio, struct ucred *cred,
+       int flags, struct proc *p)
+{
+       return (ENXIO);
+}
+
+/*ARGSUSED*/
+static int
+kqueue_write(struct file *fp, struct uio *uio, struct ucred *cred,
+        int flags, struct proc *p)
+{
+       return (ENXIO);
+}
+
+/*ARGSUSED*/
+static int
+kqueue_ioctl(struct file *fp, u_long com, caddr_t data, struct proc *p)
+{
+       return (ENOTTY);
+}
+
+/*ARGSUSED*/
+static int
+kqueue_select(struct file *fp, int which, void *wql, struct proc *p)
+{
+       struct kqueue *kq = (struct kqueue *)fp->f_data;
+       int retnum = 0;
+       int s = splnet();
+
+       if (which == FREAD) {
+                if (kq->kq_count) {
+                       retnum = 1;
+               } else {
+                        selrecord(p, &kq->kq_sel, wql);
+                       kq->kq_state |= KQ_SEL;
+               }
+       }
+       splx(s);
+       return (retnum);
+}
+
+/*ARGSUSED*/
+static int
+kqueue_close(struct file *fp, struct proc *p)
+{
+       struct kqueue *kq = (struct kqueue *)fp->f_data;
+       struct filedesc *fdp = p->p_fd;
+       struct knote **knp, *kn, *kn0;
+       int i;
+
+       for (i = 0; i < fdp->fd_knlistsize; i++) {
+               knp = &SLIST_FIRST(&fdp->fd_knlist[i]);
+               kn = *knp;
+               while (kn != NULL) {
+                       kn0 = SLIST_NEXT(kn, kn_link);
+                       if (kq == kn->kn_kq) {
+                               kn->kn_fop->f_detach(kn);
+                               fdrop(kn->kn_fp, p);
+                               knote_free(kn);
+                               *knp = kn0;
+                       } else {
+                               knp = &SLIST_NEXT(kn, kn_link);
+                       }
+                       kn = kn0;
+               }
+       }
+       if (fdp->fd_knhashmask != 0) {
+               for (i = 0; i < fdp->fd_knhashmask + 1; i++) {
+                       knp = &SLIST_FIRST(&fdp->fd_knhash[i]);
+                       kn = *knp;
+                       while (kn != NULL) {
+                               kn0 = SLIST_NEXT(kn, kn_link);
+                               if (kq == kn->kn_kq) {
+                                       kn->kn_fop->f_detach(kn);
+               /* XXX non-fd release of kn->kn_ptr */
+                                       knote_free(kn);
+                                       *knp = kn0;
+                               } else {
+                                       knp = &SLIST_NEXT(kn, kn_link);
+                               }
+                               kn = kn0;
+                       }
+               }
+       }
+       _FREE(kq, M_KQUEUE);
+       fp->f_data = NULL;
+
+       return (0);
+}
+
+/*ARGSUSED*/
+static int
+kqueue_kqfilter(struct file *fp, struct knote *kn, struct proc *p)
+{
+       struct kqueue *kq = (struct kqueue *)kn->kn_fp->f_data;
+
+       if (kn->kn_filter != EVFILT_READ || (kq->kq_state & KQ_SEL))
+               return (1);
+
+       kn->kn_fop = &kqread_filtops;
+       KNOTE_ATTACH(&kq->kq_sel.si_note, kn);
+       return (0);
+}
+
+/*ARGSUSED*/
+int
+kqueue_stat(struct file *fp, struct stat *st, struct proc *p)
+{
+       struct kqueue *kq = (struct kqueue *)fp->f_data;
+
+       bzero((void *)st, sizeof(*st));
+       st->st_size = kq->kq_count;
+       st->st_blksize = sizeof(struct kevent);
+       st->st_mode = S_IFIFO;
+       return (0);
+}
+
+static void
+kqueue_wakeup(struct kqueue *kq)
+{
+
+       if (kq->kq_state & KQ_SLEEP) {
+               kq->kq_state &= ~KQ_SLEEP;
+               wakeup(kq);
+       }
+       if (kq->kq_state & KQ_SEL) {
+         // kq->kq_state &= ~KQ_SEL; /* remove for now */
+               selwakeup(&kq->kq_sel);
+       } else
+               KNOTE(&kq->kq_sel.si_note, 0);
+}
+
+void
+klist_init(struct klist *list)
+{
+       SLIST_INIT(list);
+}
+
+/*
+ * walk down a list of knotes, activating them if their event has triggered.
+ */
+void
+knote(struct klist *list, long hint)
+{
+       struct knote *kn;
+
+       SLIST_FOREACH(kn, list, kn_selnext)
+               if (kn->kn_fop->f_event(kn, hint))
+                       KNOTE_ACTIVATE(kn);
+}
+
+/*
+ * attach a knote to the specified list.  Return true if this is the first entry.
+ */
+int
+knote_attach(struct klist *list, struct knote *kn)
+{
+       int ret = SLIST_EMPTY(list);
+       SLIST_INSERT_HEAD(list, kn, kn_selnext);
+       return ret;
+}
+
+/*
+ * detach a knote from the specified list.  Return true if that was the last entry.
+ */
+int
+knote_detach(struct klist *list, struct knote *kn)
+{
+       SLIST_REMOVE(list, kn, knote, kn_selnext);
+       return SLIST_EMPTY(list);
+}
+
+/*
+ * remove all knotes from a specified klist
+ */
+void
+knote_remove(struct proc *p, struct klist *list)
+{
+       struct knote *kn;
+
+       while ((kn = SLIST_FIRST(list)) != NULL) {
+               kn->kn_fop->f_detach(kn);
+               knote_drop(kn, p);
+       }
+}
+
+/*
+ * remove all knotes referencing a specified fd
+ */
+void
+knote_fdclose(struct proc *p, int fd)
+{
+       struct filedesc *fdp = p->p_fd;
+       struct klist *list = &fdp->fd_knlist[fd];
+
+       knote_remove(p, list);
+}
+
+static void
+knote_fdpattach(struct knote *kn, struct filedesc *fdp)
+{
+       struct klist *list;
+       int size;
+
+       if (! kn->kn_fop->f_isfd) {
+               if (fdp->fd_knhashmask == 0)
+                       fdp->fd_knhash = hashinit(KN_HASHSIZE, M_KQUEUE,
+                           &fdp->fd_knhashmask);
+               list = &fdp->fd_knhash[KN_HASH(kn->kn_id, fdp->fd_knhashmask)];
+               goto done;
+       }
+
+       if (fdp->fd_knlistsize <= kn->kn_id) {
+               size = fdp->fd_knlistsize;
+               while (size <= kn->kn_id)
+                       size += KQEXTENT;
+               MALLOC(list, struct klist *,
+                   size * sizeof(struct klist *), M_KQUEUE, M_WAITOK);
+               bcopy((caddr_t)fdp->fd_knlist, (caddr_t)list,
+                   fdp->fd_knlistsize * sizeof(struct klist *));
+               bzero((caddr_t)list +
+                   fdp->fd_knlistsize * sizeof(struct klist *),
+                   (size - fdp->fd_knlistsize) * sizeof(struct klist *));
+               if (fdp->fd_knlist != NULL)
+                       FREE(fdp->fd_knlist, M_KQUEUE);
+               fdp->fd_knlistsize = size;
+               fdp->fd_knlist = list;
+       }
+       list = &fdp->fd_knlist[kn->kn_id];
+done:
+       SLIST_INSERT_HEAD(list, kn, kn_link);
+       kn->kn_status = 0;
+}
+
+/*
+ * should be called at spl == 0, since we don't want to hold spl
+ * while calling fdrop and free.
+ */
+static void
+knote_drop(struct knote *kn, struct proc *p)
+{
+        struct filedesc *fdp = p->p_fd;
+       struct klist *list;
+
+       if (kn->kn_fop->f_isfd)
+               list = &fdp->fd_knlist[kn->kn_id];
+       else
+               list = &fdp->fd_knhash[KN_HASH(kn->kn_id, fdp->fd_knhashmask)];
+
+       SLIST_REMOVE(list, kn, knote, kn_link);
+       if (kn->kn_status & KN_QUEUED)
+               knote_dequeue(kn);
+       if (kn->kn_fop->f_isfd)
+               fdrop(kn->kn_fp, p);
+       knote_free(kn);
+}
+
+
+static void
+knote_enqueue(struct knote *kn)
+{
+       struct kqueue *kq = kn->kn_kq;
+       int s = splhigh();
+
+       KASSERT((kn->kn_status & KN_QUEUED) == 0, ("knote already queued"));
+
+       TAILQ_INSERT_TAIL(&kq->kq_head, kn, kn_tqe); 
+       kn->kn_status |= KN_QUEUED;
+       kq->kq_count++;
+       splx(s);
+       kqueue_wakeup(kq);
+}
+
+static void
+knote_dequeue(struct knote *kn)
+{
+       struct kqueue *kq = kn->kn_kq;
+       int s = splhigh();
+
+       KASSERT(kn->kn_status & KN_QUEUED, ("knote not queued"));
+
+       TAILQ_REMOVE(&kq->kq_head, kn, kn_tqe); 
+       kn->kn_status &= ~KN_QUEUED;
+       kq->kq_count--;
+       splx(s);
+}
+
+void
+knote_init(void)
+{
+       knote_zone = zinit(sizeof(struct knote), 8192*sizeof(struct knote), 8192, "knote zone");
+}
+SYSINIT(knote, SI_SUB_PSEUDO, SI_ORDER_ANY, knote_init, NULL)
+
+static struct knote *
+knote_alloc(void)
+{
+       return ((struct knote *)zalloc(knote_zone));
+}
+
+static void
+knote_free(struct knote *kn)
+{
+       zfree(knote_zone, (vm_offset_t)kn);
+}
+
 #include <sys/param.h>
 #include <sys/socket.h>
 #include <sys/protosw.h>
 #include <sys/param.h>
 #include <sys/socket.h>
 #include <sys/protosw.h>
@@ -74,6 +1158,10 @@ int kev_attach(struct socket *so, int proto, struct proc *p)
      int error;
      struct kern_event_pcb  *ev_pcb;
 
      int error;
      struct kern_event_pcb  *ev_pcb;
 
+     error = soreserve(so, KEV_SNDSPACE, KEV_RECVSPACE);
+     if (error)
+          return error;
+
      ev_pcb = _MALLOC(sizeof(struct kern_event_pcb), M_PCB, M_WAITOK);
      if (ev_pcb == 0)
          return ENOBUFS;
      ev_pcb = _MALLOC(sizeof(struct kern_event_pcb), M_PCB, M_WAITOK);
      if (ev_pcb == 0)
          return ENOBUFS;
@@ -83,9 +1171,6 @@ int kev_attach(struct socket *so, int proto, struct proc *p)
 
      so->so_pcb = (caddr_t) ev_pcb;
      LIST_INSERT_HEAD(&kern_event_head, ev_pcb, ev_link);
 
      so->so_pcb = (caddr_t) ev_pcb;
      LIST_INSERT_HEAD(&kern_event_head, ev_pcb, ev_link);
-     error = soreserve(so, KEV_SNDSPACE, KEV_RECVSPACE);
-     if (error)
-         return error;
 
      return 0;
 }
 
      return 0;
 }
@@ -95,9 +1180,11 @@ int kev_detach(struct socket *so)
 {
      struct kern_event_pcb *ev_pcb = (struct kern_event_pcb *) so->so_pcb;
 
 {
      struct kern_event_pcb *ev_pcb = (struct kern_event_pcb *) so->so_pcb;
 
-     LIST_REMOVE(ev_pcb, ev_link);
-     if (ev_pcb)
-         FREE(ev_pcb, M_PCB);
+     if (ev_pcb != 0) {
+          LIST_REMOVE(ev_pcb, ev_link);
+          FREE(ev_pcb, M_PCB);
+          so->so_pcb = 0;
+     }
 
      return 0;
 }
 
      return 0;
 }