]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/kern/kern_event.c
xnu-792.1.5.tar.gz
[apple/xnu.git] / bsd / kern / kern_event.c
index 0b1425c07dce09a4d712db290c7c6c3f8294b8ac..1bf948822845e820ff8c1deed02cf8bfaa30fead 100644 (file)
@@ -1,24 +1,21 @@
 /*
 /*
- * Copyright (c) 2000-2003 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2000-2005 Apple Computer, Inc. All rights reserved.
  *
  * @APPLE_LICENSE_HEADER_START@
  * 
  *
  * @APPLE_LICENSE_HEADER_START@
  * 
- * Copyright (c) 1999-2003 Apple Computer, Inc.  All Rights Reserved.
+ * 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.
  * 
  * 
- * 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
+ * This 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, QUIET ENJOYMENT OR NON-INFRINGEMENT.
- * Please see the License for the specific language governing rights and
- * limitations under the License.
+ * FITNESS FOR A PARTICULAR PURPOSE 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@
  *
 /*
  *     @(#)kern_event.c       1.0 (3/31/2000)
  */
 /*
  *     @(#)kern_event.c       1.0 (3/31/2000)
  */
+#include <stdint.h>
 
 #include <sys/param.h>
 #include <sys/systm.h>
 #include <sys/filedesc.h>
 #include <sys/kernel.h>
 
 #include <sys/param.h>
 #include <sys/systm.h>
 #include <sys/filedesc.h>
 #include <sys/kernel.h>
-#include <sys/proc.h>
+#include <sys/proc_internal.h>
+#include <sys/kauth.h>
 #include <sys/malloc.h> 
 #include <sys/unistd.h>
 #include <sys/malloc.h> 
 #include <sys/unistd.h>
-#include <sys/file.h>
+#include <sys/file_internal.h>
 #include <sys/fcntl.h>
 #include <sys/select.h>
 #include <sys/queue.h>
 #include <sys/fcntl.h>
 #include <sys/select.h>
 #include <sys/queue.h>
 #include <sys/stat.h>
 #include <sys/sysctl.h>
 #include <sys/uio.h>
 #include <sys/stat.h>
 #include <sys/sysctl.h>
 #include <sys/uio.h>
-
+#include <sys/sysproto.h>
+#include <sys/user.h>
+#include <string.h>
+
+#include <kern/lock.h>
+#include <kern/clock.h>
+#include <kern/thread_call.h>
+#include <kern/sched_prim.h>
 #include <kern/zalloc.h>
 #include <kern/zalloc.h>
+#include <kern/assert.h>
+
+#include <libkern/libkern.h>
+
+extern void unix_syscall_return(int);
 
 MALLOC_DEFINE(M_KQUEUE, "kqueue", "memory for kqueue system");
 
 
 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 inline void kqlock(struct kqueue *kq);
+static inline void kqunlock(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 int     kqlock2knoteuse(struct kqueue *kq, struct knote *kn);
+static int     kqlock2knoteusewait(struct kqueue *kq, struct knote *kn);
+static int     kqlock2knotedrop(struct kqueue *kq, struct knote *kn);
+static int     knoteuse2kqlock(struct kqueue *kq, struct knote *kn);
+
+static void    kqueue_wakeup(struct kqueue *kq);
+static int     kqueue_read(struct fileproc *fp, struct uio *uio,
+                   kauth_cred_t cred, int flags, struct proc *p);
+static int     kqueue_write(struct fileproc *fp, struct uio *uio,
+                   kauth_cred_t cred, int flags, struct proc *p);
+static int     kqueue_ioctl(struct fileproc *fp, u_long com, caddr_t data,
+                   struct proc *p);
+static int     kqueue_select(struct fileproc *fp, int which, void *wql, 
+                   struct proc *p);
+static int     kqueue_close(struct fileglob *fp, struct proc *p);
+static int     kqueue_kqfilter(struct fileproc *fp, struct knote *kn, struct proc *p);
+extern int     kqueue_stat(struct fileproc *fp, struct stat *st, struct proc *p);
 
 static struct fileops kqueueops = {
        kqueue_read,
 
 static struct fileops kqueueops = {
        kqueue_read,
@@ -98,15 +114,28 @@ static struct fileops kqueueops = {
        kqueue_ioctl,
        kqueue_select,
        kqueue_close,
        kqueue_ioctl,
        kqueue_select,
        kqueue_close,
-       kqueue_kqfilter
+       kqueue_kqfilter,
+       0
 };
 
 };
 
-static void    knote_fdpattach(struct knote *kn, struct filedesc *fdp);
+static int kevent_copyin(user_addr_t *addrp, struct kevent *kevp, struct proc *p);
+static int kevent_copyout(struct kevent *kevp, user_addr_t *addrp, struct proc *p);
+
+static int     kevent_callback(struct kqueue *kq, struct kevent *kevp, void *data);
+static void    kevent_continue(struct kqueue *kq, void *data, int error);
+static void    kevent_scan_continue(void *contp, wait_result_t wait_result);
+static int     kevent_process(struct kqueue *kq, kevent_callback_t callback,
+                              void *data, int *countp, struct proc *p);
+static void    knote_put(struct knote *kn);
+static int     knote_fdpattach(struct knote *kn, struct filedesc *fdp, struct proc *p);
 static void    knote_drop(struct knote *kn, struct proc *p);
 static void    knote_drop(struct knote *kn, struct proc *p);
+static void    knote_activate(struct knote *kn);
+static void    knote_deactivate(struct knote *kn);
 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 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);
+extern void    knote_init(void);
 
 static int     filt_fileattach(struct knote *kn);
 static struct filterops file_filtops =
 
 static int     filt_fileattach(struct knote *kn);
 static struct filterops file_filtops =
@@ -118,7 +147,7 @@ static struct filterops kqread_filtops =
        { 1, NULL, filt_kqdetach, filt_kqueue };
 
 /*
        { 1, NULL, filt_kqdetach, filt_kqueue };
 
 /*
- * JMM - placeholder for not-yet-implemented filters
+ * placeholder for not-yet-implemented filters
  */ 
 static int     filt_badattach(struct knote *kn);
 static struct filterops bad_filtops =
  */ 
 static int     filt_badattach(struct knote *kn);
 static struct filterops bad_filtops =
@@ -135,9 +164,10 @@ extern struct filterops fs_filtops;
 
 extern struct filterops sig_filtops;
 
 
 extern struct filterops sig_filtops;
 
-#if 0
-/* JMM - We don't implement these now */
-static void    filt_timerexpire(void *knx);
+
+/* Timer filter */
+static int     filt_timercompute(struct knote *kn, uint64_t *abs_time);
+static void    filt_timerexpire(void *knx, void *param1);
 static int     filt_timerattach(struct knote *kn);
 static void    filt_timerdetach(struct knote *kn);
 static int     filt_timer(struct knote *kn, long hint);
 static int     filt_timerattach(struct knote *kn);
 static void    filt_timerdetach(struct knote *kn);
 static int     filt_timer(struct knote *kn, long hint);
@@ -145,20 +175,21 @@ static int        filt_timer(struct knote *kn, long hint);
 static struct filterops timer_filtops =
        { 0, filt_timerattach, filt_timerdetach, filt_timer };
 
 static struct filterops timer_filtops =
        { 0, filt_timerattach, filt_timerdetach, filt_timer };
 
-static int             kq_ncallouts = 0;
-static int             kq_calloutmax = (4 * 1024);
+/* to avoid arming timers that fire quicker than we can handle */
+static uint64_t        filt_timerfloor = 0; 
 
 
-SYSCTL_INT(_kern, OID_AUTO, kq_calloutmax, CTLFLAG_RW,
-    &kq_calloutmax, 0, "Maximum number of callouts allocated for kqueue");
-#endif /* 0 */
+static lck_mtx_t _filt_timerlock;
+static void    filt_timerlock(void);
+static void    filt_timerunlock(void);
 
 
-static zone_t  knote_zone;
+/*
+ * Sentinel marker for a thread scanning through the list of
+ * active knotes.
+ */
+static struct filterops threadmarker_filtops =
+       { 0, filt_badattach, 0, 0 };
 
 
-#define KNOTE_ACTIVATE(kn) do {                                        \
-       kn->kn_status |= KN_ACTIVE;                                     \
-       if ((kn->kn_status & (KN_QUEUED | KN_DISABLED)) == 0)           \
-               knote_enqueue(kn);                                      \
-} while(0)
+static zone_t  knote_zone;
 
 #define        KN_HASHSIZE             64              /* XXX should be tunable */
 #define KN_HASH(val, mask)     (((val) ^ (val >> 8)) & (mask))
 
 #define        KN_HASHSIZE             64              /* XXX should be tunable */
 #define KN_HASH(val, mask)     (((val) ^ (val >> 8)) & (mask))
@@ -181,15 +212,155 @@ static struct filterops *sysfilt_ops[] = {
        &file_filtops,                  /* EVFILT_VNODE */
        &proc_filtops,                  /* EVFILT_PROC */
        &sig_filtops,                   /* EVFILT_SIGNAL */
        &file_filtops,                  /* EVFILT_VNODE */
        &proc_filtops,                  /* EVFILT_PROC */
        &sig_filtops,                   /* EVFILT_SIGNAL */
-#if 0
        &timer_filtops,                 /* EVFILT_TIMER */
        &timer_filtops,                 /* EVFILT_TIMER */
-#else
-       &bad_filtops,                   /* EVFILT_TIMER */
-#endif
        &bad_filtops,                   /* EVFILT_MACHPORT */
        &bad_filtops,                   /* EVFILT_MACHPORT */
-       &fs_filtops             /* EVFILT_FS */
+       &fs_filtops                     /* EVFILT_FS */
 };
 
 };
 
+/*
+ * kqueue/note lock attributes and implementations
+ *
+ *     kqueues have locks, while knotes have use counts
+ *     Most of the knote state is guarded by the object lock.
+ *     the knote "inuse" count and status use the kqueue lock.
+ */
+lck_grp_attr_t * kq_lck_grp_attr;
+lck_grp_t * kq_lck_grp;
+lck_attr_t * kq_lck_attr;
+
+static inline void
+kqlock(struct kqueue *kq)
+{
+       lck_spin_lock(&kq->kq_lock);
+}
+
+static inline void
+kqunlock(struct kqueue *kq)
+{
+       lck_spin_unlock(&kq->kq_lock);
+}
+
+/* 
+ * Convert a kq lock to a knote use referece.
+ *
+ *     If the knote is being dropped, we can't get
+ *     a use reference, so just return with it
+ *     still locked.
+ *     
+ *     - kq locked at entry
+ *     - unlock on exit if we get the use reference
+ */
+static int
+kqlock2knoteuse(struct kqueue *kq, struct knote *kn)
+{
+       if (kn->kn_status & KN_DROPPING)
+               return 0;
+       kn->kn_inuse++;
+       kqunlock(kq);
+       return 1;
+ }
+
+/* 
+ * Convert a kq lock to a knote use referece.
+ *
+ *     If the knote is being dropped, we can't get
+ *     a use reference, so just return with it
+ *     still locked.
+ *     
+ *     - kq locked at entry
+ *     - kq always unlocked on exit
+ */
+static int
+kqlock2knoteusewait(struct kqueue *kq, struct knote *kn)
+{
+       if (!kqlock2knoteuse(kq, kn)) {
+               kn->kn_status |= KN_DROPWAIT;
+               assert_wait(&kn->kn_status, THREAD_UNINT);
+               kqunlock(kq);
+               thread_block(THREAD_CONTINUE_NULL);
+               return 0;
+       }
+       return 1;
+ }
+
+/* 
+ * Convert from a knote use reference back to kq lock.
+ *
+ *     Drop a use reference and wake any waiters if
+ *     this is the last one.
+ *
+ *     The exit return indicates if the knote is
+ *     still alive - but the kqueue lock is taken
+ *     unconditionally.
+ */
+static int
+knoteuse2kqlock(struct kqueue *kq, struct knote *kn)
+{
+       kqlock(kq);
+       if ((--kn->kn_inuse == 0) &&
+           (kn->kn_status & KN_USEWAIT)) {
+               kn->kn_status &= ~KN_USEWAIT;
+               thread_wakeup(&kn->kn_inuse);
+       }
+       return ((kn->kn_status & KN_DROPPING) == 0);
+ }
+
+/* 
+ * Convert a kq lock to a knote drop referece.
+ *
+ *     If the knote is in use, wait for the use count
+ *     to subside.  We first mark our intention to drop
+ *     it - keeping other users from "piling on."
+ *     If we are too late, we have to wait for the
+ *     other drop to complete.
+ *     
+ *     - kq locked at entry
+ *     - always unlocked on exit.
+ *     - caller can't hold any locks that would prevent
+ *       the other dropper from completing.
+ */
+static int
+kqlock2knotedrop(struct kqueue *kq, struct knote *kn)
+{
+
+       if ((kn->kn_status & KN_DROPPING) == 0) {
+               kn->kn_status |= KN_DROPPING;
+               if (kn->kn_inuse > 0) {
+                       kn->kn_status |= KN_USEWAIT;
+                       assert_wait(&kn->kn_inuse, THREAD_UNINT);
+                       kqunlock(kq);
+                       thread_block(THREAD_CONTINUE_NULL);
+               } else
+                       kqunlock(kq);
+               return 1;
+       } else {
+               kn->kn_status |= KN_DROPWAIT;
+               assert_wait(&kn->kn_status, THREAD_UNINT);
+               kqunlock(kq);
+               thread_block(THREAD_CONTINUE_NULL);
+               return 0;
+       }
+}
+               
+/* 
+ * Release a knote use count reference.
+ */
+static void
+knote_put(struct knote *kn)
+{
+       struct kqueue *kq = kn->kn_kq;
+
+       kqlock(kq);
+       if ((--kn->kn_inuse == 0) && 
+           (kn->kn_status & KN_USEWAIT)) {
+               kn->kn_status &= ~KN_USEWAIT;
+               thread_wakeup(&kn->kn_inuse);
+       }
+       kqunlock(kq);
+ }
+
+
+
 static int
 filt_fileattach(struct knote *kn)
 {
 static int
 filt_fileattach(struct knote *kn)
 {
@@ -197,20 +368,27 @@ filt_fileattach(struct knote *kn)
        return (fo_kqfilter(kn->kn_fp, kn, current_proc()));
 }
 
        return (fo_kqfilter(kn->kn_fp, kn, current_proc()));
 }
 
+#define f_flag f_fglob->fg_flag
+#define f_type f_fglob->fg_type
+#define f_msgcount f_fglob->fg_msgcount
+#define f_cred f_fglob->fg_cred
+#define f_ops f_fglob->fg_ops
+#define f_offset f_fglob->fg_offset
+#define f_data f_fglob->fg_data
+
 static void
 filt_kqdetach(struct knote *kn)
 {
        struct kqueue *kq = (struct kqueue *)kn->kn_fp->f_data;
 
 static void
 filt_kqdetach(struct knote *kn)
 {
        struct kqueue *kq = (struct kqueue *)kn->kn_fp->f_data;
 
-       if (kq->kq_state & KQ_SEL)
-         return;
-
+       kqlock(kq);
        KNOTE_DETACH(&kq->kq_sel.si_note, kn);
        KNOTE_DETACH(&kq->kq_sel.si_note, kn);
+       kqunlock(kq);
 }
 
 /*ARGSUSED*/
 static int
 }
 
 /*ARGSUSED*/
 static int
-filt_kqueue(struct knote *kn, long hint)
+filt_kqueue(struct knote *kn, __unused long hint)
 {
        struct kqueue *kq = (struct kqueue *)kn->kn_fp->f_data;
 
 {
        struct kqueue *kq = (struct kqueue *)kn->kn_fp->f_data;
 
@@ -222,21 +400,23 @@ static int
 filt_procattach(struct knote *kn)
 {
        struct proc *p;
 filt_procattach(struct knote *kn)
 {
        struct proc *p;
+       int funnel_state;
+       
+       funnel_state = thread_funnel_set(kernel_flock, TRUE);
 
        p = pfind(kn->kn_id);
 
        p = pfind(kn->kn_id);
-       if (p == NULL)
+       if (p == NULL) {
+               thread_funnel_set(kernel_flock, funnel_state);
                return (ESRCH);
                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_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_data = (int)kn->kn_sdata;        /* ppid */
                kn->kn_fflags = NOTE_CHILD;
                kn->kn_flags &= ~EV_FLAG1;
        }
                kn->kn_fflags = NOTE_CHILD;
                kn->kn_flags &= ~EV_FLAG1;
        }
@@ -244,6 +424,8 @@ filt_procattach(struct knote *kn)
        /* XXX lock the proc here while adding to the list? */
        KNOTE_ATTACH(&p->p_klist, kn);
 
        /* XXX lock the proc here while adding to the list? */
        KNOTE_ATTACH(&p->p_klist, kn);
 
+       thread_funnel_set(kernel_flock, funnel_state);
+
        return (0);
 }
 
        return (0);
 }
 
@@ -258,19 +440,25 @@ filt_procattach(struct knote *kn)
 static void
 filt_procdetach(struct knote *kn)
 {
 static void
 filt_procdetach(struct knote *kn)
 {
-       struct proc *p = kn->kn_ptr.p_proc;
+       struct proc *p;
+       int funnel_state;
 
 
-       if (kn->kn_status & KN_DETACHED)
-               return;
+       funnel_state = thread_funnel_set(kernel_flock, TRUE);
+       p = pfind(kn->kn_id);
+
+       if (p != (struct proc *)NULL)
+               KNOTE_DETACH(&p->p_klist, kn);
 
 
-       /* XXX locking?  this might modify another process. */
-       KNOTE_DETACH(&p->p_klist, kn);
+       thread_funnel_set(kernel_flock, funnel_state);
 }
 
 static int
 filt_proc(struct knote *kn, long hint)
 {
        u_int event;
 }
 
 static int
 filt_proc(struct knote *kn, long hint)
 {
        u_int event;
+       int funnel_state;
+
+       funnel_state = thread_funnel_set(kernel_flock, TRUE);
 
        /*
         * mask off extra data
 
        /*
         * mask off extra data
@@ -287,8 +475,8 @@ filt_proc(struct knote *kn, long hint)
         * process is gone, so flag the event as finished.
         */
        if (event == NOTE_EXIT) {
         * 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); 
                kn->kn_flags |= (EV_EOF | EV_ONESHOT); 
+               thread_funnel_set(kernel_flock, funnel_state);
                return (1);
        }
 
                return (1);
        }
 
@@ -310,240 +498,654 @@ filt_proc(struct knote *kn, long hint)
                kev.fflags = kn->kn_sfflags;
                kev.data = kn->kn_id;                   /* parent */
                kev.udata = kn->kn_kevent.udata;        /* preserve udata */
                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);
+               error = kevent_register(kn->kn_kq, &kev, NULL);
                if (error)
                        kn->kn_fflags |= NOTE_TRACKERR;
        }
                if (error)
                        kn->kn_fflags |= NOTE_TRACKERR;
        }
+       event = kn->kn_fflags;
+       thread_funnel_set(kernel_flock, funnel_state);
 
 
-       return (kn->kn_fflags != 0);
+       return (event != 0);
 }
 
 }
 
-#if 0
+/*
+ * filt_timercompute - compute absolute timeout
+ *
+ *     The saved-data field in the knote contains the
+ *     time value.  The saved filter-flags indicates
+ *     the unit of measurement.
+ *
+ *     If the timeout is not absolute, adjust it for
+ *     the current time.
+ */
+static int
+filt_timercompute(struct knote *kn, uint64_t *abs_time)
+{
+       uint64_t multiplier;
+       uint64_t raw;
+
+       switch (kn->kn_sfflags & (NOTE_SECONDS|NOTE_USECONDS|NOTE_NSECONDS)) {
+       case NOTE_SECONDS:
+               multiplier = NSEC_PER_SEC;
+               break;
+       case NOTE_USECONDS:
+               multiplier = NSEC_PER_USEC;
+               break;
+       case NOTE_NSECONDS:
+               multiplier = 1;
+               break;
+       case 0: /* milliseconds (default) */
+               multiplier = NSEC_PER_SEC / 1000;
+               break;
+       default:
+               return EINVAL;
+       }
+       nanoseconds_to_absolutetime((uint64_t)kn->kn_sdata * multiplier, &raw);
+       if (raw <= filt_timerfloor) {
+               *abs_time = 0;
+               return 0;
+       }
+       if ((kn->kn_sfflags & NOTE_ABSOLUTE) == NOTE_ABSOLUTE) {
+               uint32_t seconds, nanoseconds;
+               uint64_t now;
+
+               clock_get_calendar_nanotime(&seconds, &nanoseconds);
+               nanoseconds_to_absolutetime((uint64_t)seconds * NSEC_PER_SEC + nanoseconds,
+                                           &now);
+               if (now >= raw + filt_timerfloor) {
+                       *abs_time = 0;
+                       return 0;
+               }
+               raw -= now;
+       } 
+       clock_absolutetime_interval_to_deadline(raw, abs_time);
+       return 0;
+}
+
+/* 
+ * filt_timerexpire - the timer callout routine
+ *
+ *     Just propagate the timer event into the knote
+ *     filter routine (by going through the knote
+ *     synchronization point).  Pass a hint to
+ *     indicate this is a real event, not just a
+ *     query from above.
+ */
 static void
 static void
-filt_timerexpire(void *knx)
+filt_timerexpire(void *knx, __unused void *spare)
 {
 {
+       struct klist timer_list;
        struct knote *kn = 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);
-       }
+
+       /* no "object" for timers, so fake a list */
+       SLIST_INIT(&timer_list);
+       SLIST_INSERT_HEAD(&timer_list, kn, kn_selnext); 
+       KNOTE(&timer_list, 1);
 }
 
 /*
 }
 
 /*
- * data contains amount of time to sleep, in milliseconds
+ * data contains amount of time to sleep, in milliseconds,
+ * or a pointer to a timespec structure.
  */ 
 static int
 filt_timerattach(struct knote *kn)
 {
  */ 
 static int
 filt_timerattach(struct knote *kn)
 {
-       struct callout *calloutp;
-       struct timeval tv;
-       int tticks;
+       thread_call_t callout;
+       uint64_t deadline;
+       int error;
 
 
-       if (kq_ncallouts >= kq_calloutmax)
-               return (ENOMEM);
-       kq_ncallouts++;
+       error = filt_timercompute(kn, &deadline);
+       if (error)
+               return (error);
 
 
-       tv.tv_sec = kn->kn_sdata / 1000;
-       tv.tv_usec = (kn->kn_sdata % 1000) * 1000;
-       tticks = tvtohz(&tv);
+       if (deadline) {
+               callout = thread_call_allocate(filt_timerexpire, kn);
+               if (NULL == callout)
+                       return (ENOMEM);
+       } else {  
+               /* handle as immediate */
+               kn->kn_sdata = 0;
+               callout = NULL;
+       }
 
 
-       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;
+       filt_timerlock();
+       kn->kn_hook = (caddr_t)callout;
 
 
+       /* absolute=EV_ONESHOT */
+       if (kn->kn_sfflags & NOTE_ABSOLUTE)
+               kn->kn_flags |= EV_ONESHOT; 
+
+       if (deadline) {
+               /* all others - if not faking immediate */
+               kn->kn_flags |= EV_CLEAR;
+               thread_call_enter_delayed(callout, deadline);
+               kn->kn_hookid = 0;
+       } else {
+               /* fake immediate */
+               kn->kn_hookid = 1;
+       }
+       filt_timerunlock();
        return (0);
 }
 
 static void
 filt_timerdetach(struct knote *kn)
 {
        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--;
+       thread_call_t callout;
+
+       filt_timerlock();
+       callout = (thread_call_t)kn->kn_hook;
+       if (callout != NULL) {
+               boolean_t cancelled;
+
+               /* cancel the callout if we can */
+               cancelled = thread_call_cancel(callout);
+               if (cancelled) {
+                       /* got it, just free it */
+                       kn->kn_hook = NULL;
+                       filt_timerunlock();
+                       thread_call_free(callout);
+                       return;
+               }
+               /* we have to wait for the expire routine.  */
+               kn->kn_hookid = -1;     /* we are detaching */
+               assert_wait(&kn->kn_hook, THREAD_UNINT);
+               filt_timerunlock();
+               thread_block(THREAD_CONTINUE_NULL);
+               assert(kn->kn_hook == NULL);
+               return;
+       }
+       /* nothing to do */
+       filt_timerunlock(); 
 }
 
 }
 
+
+
 static int
 static int
-filt_timer(struct knote *kn, long hint)
+filt_timer(struct knote *kn, __unused long hint)
 {
 {
+       int result;
+       
+       if (hint) {
+               /* real timer pop */
+               thread_call_t callout;
+               boolean_t detaching;
+
+               filt_timerlock();
+               
+               kn->kn_data++;
+
+               detaching = (kn->kn_hookid < 0);
+               callout = (thread_call_t)kn->kn_hook;
 
 
-       return (kn->kn_data != 0);
+               if (!detaching && (kn->kn_flags & EV_ONESHOT) == 0) {
+                       uint64_t deadline;
+                       int error;
+
+                       /* user input data may have changed - deal */
+                       error = filt_timercompute(kn, &deadline);
+                       if (error) {
+                               kn->kn_flags |= EV_ERROR;
+                               kn->kn_data = error;
+                       } else if (deadline == 0) {
+                               /* revert to fake immediate */
+                               kn->kn_flags &= ~EV_CLEAR;
+                               kn->kn_sdata = 0;
+                               kn->kn_hookid = 1;
+                       } else {
+                               /* keep the callout and re-arm */
+                               thread_call_enter_delayed(callout, deadline);
+                               filt_timerunlock();
+                               return 1;
+                       }
+               }
+               kn->kn_hook = NULL;
+               filt_timerunlock();
+               thread_call_free(callout);
+
+               /* if someone is waiting for timer to pop */
+               if (detaching)
+                       thread_wakeup(&kn->kn_hook);
+
+               return 1;
+       } 
+
+       /* user-query */
+       filt_timerlock();
+
+       /* change fake timer to real if needed */
+       while (kn->kn_hookid > 0 && kn->kn_sdata > 0) {
+               int error;
+
+               /* update the fake timer (make real) */
+               kn->kn_hookid = 0;
+               kn->kn_data = 0;
+               filt_timerunlock();
+               error = filt_timerattach(kn);
+               filt_timerlock();
+               if (error) {
+                       kn->kn_flags |= EV_ERROR;
+                       kn->kn_data = error;
+                       filt_timerunlock();
+                       return 1;
+               }
+       }
+
+       /* if still fake, pretend it fired */
+       if (kn->kn_hookid > 0)
+               kn->kn_data = 1;
+
+       result = (kn->kn_data != 0);
+       filt_timerunlock();
+       return result;
+}
+
+static void
+filt_timerlock(void)
+{
+       lck_mtx_lock(&_filt_timerlock);
+}
+
+static void
+filt_timerunlock(void)
+{
+       lck_mtx_unlock(&_filt_timerlock);
 }
 }
-#endif /* 0 */
 
 /*
  * JMM - placeholder for not-yet-implemented filters
  */ 
 static int
 
 /*
  * JMM - placeholder for not-yet-implemented filters
  */ 
 static int
-filt_badattach(struct knote *kn)
+filt_badattach(__unused struct knote *kn)
 {
 {
-       return(EOPNOTSUPP);
+       return(ENOTSUP);
 }
 
 }
 
-#ifndef _SYS_SYSPROTO_H_
-struct kqueue_args {
-       int dummy;
-};
-#endif
 
 
-int
-kqueue(struct proc *p, struct kqueue_args *uap, register_t *retval)
+struct kqueue *
+kqueue_alloc(struct proc *p)
 {
        struct filedesc *fdp = p->p_fd;
        struct kqueue *kq;
 {
        struct filedesc *fdp = p->p_fd;
        struct kqueue *kq;
-       struct file *fp;
+
+       MALLOC_ZONE(kq, struct kqueue *, sizeof(struct kqueue), M_KQUEUE, M_WAITOK);
+       if (kq != NULL) {
+               bzero(kq, sizeof(struct kqueue));
+               lck_spin_init(&kq->kq_lock, kq_lck_grp, kq_lck_attr);
+               TAILQ_INIT(&kq->kq_head);
+               TAILQ_INIT(&kq->kq_inprocess);
+               kq->kq_fdp = fdp;
+       }
+
+       if (fdp->fd_knlistsize < 0) {
+               proc_fdlock(p);
+               if (fdp->fd_knlistsize < 0)
+                       fdp->fd_knlistsize = 0;         /* this process has had a kq */
+               proc_fdunlock(p);
+       }
+
+       return kq;
+}
+
+
+/*
+ * kqueue_dealloc - detach all knotes from a kqueue and free it
+ *
+ *     We walk each list looking for knotes referencing this
+ *     this kqueue.  If we find one, we try to drop it.  But
+ *     if we fail to get a drop reference, that will wait
+ *     until it is dropped.  So, we can just restart again
+ *     safe in the assumption that the list will eventually
+ *     not contain any more references to this kqueue (either
+ *     we dropped them all, or someone else did).
+ *
+ *     Assumes no new events are being added to the kqueue.
+ *     Nothing locked on entry or exit.
+ */
+void
+kqueue_dealloc(struct kqueue *kq, struct proc *p)
+{
+       struct filedesc *fdp = p->p_fd;
+       struct knote *kn;
+       int i;
+
+       proc_fdlock(p);
+       for (i = 0; i < fdp->fd_knlistsize; i++) {
+               kn = SLIST_FIRST(&fdp->fd_knlist[i]);
+               while (kn != NULL) {
+                       if (kq == kn->kn_kq) {
+                               kqlock(kq);
+                               proc_fdunlock(p);
+                               /* drop it ourselves or wait */
+                               if (kqlock2knotedrop(kq, kn)) {
+                                       kn->kn_fop->f_detach(kn);
+                                       knote_drop(kn, p);
+                               }
+                               proc_fdlock(p);
+                               /* start over at beginning of list */
+                               kn = SLIST_FIRST(&fdp->fd_knlist[i]);
+                               continue;
+                       }
+                       kn = SLIST_NEXT(kn, kn_link);
+               }
+       }
+       if (fdp->fd_knhashmask != 0) {
+               for (i = 0; i < (int)fdp->fd_knhashmask + 1; i++) {
+                       kn = SLIST_FIRST(&fdp->fd_knhash[i]);
+                       while (kn != NULL) {
+                               if (kq == kn->kn_kq) {
+                                       kqlock(kq);
+                                       proc_fdunlock(p);
+                                       /* drop it ourselves or wait */
+                                       if (kqlock2knotedrop(kq, kn)) {
+                                               kn->kn_fop->f_detach(kn);
+                                               knote_drop(kn, p);
+                                       }
+                                       proc_fdlock(p);
+                                       /* start over at beginning of list */
+                                       kn = SLIST_FIRST(&fdp->fd_knhash[i]);
+                                       continue;
+                               }
+                               kn = SLIST_NEXT(kn, kn_link);
+                       }
+               }
+       }
+       proc_fdunlock(p);
+       lck_spin_destroy(&kq->kq_lock, kq_lck_grp);
+       FREE_ZONE(kq, sizeof(struct kqueue), M_KQUEUE);
+}
+
+int
+kqueue(struct proc *p, __unused struct kqueue_args *uap, register_t *retval)
+{
+       struct kqueue *kq;
+       struct fileproc *fp;
        int fd, error;
 
        error = falloc(p, &fp, &fd);
        int fd, error;
 
        error = falloc(p, &fp, &fd);
-       if (error)
+       if (error) {
                return (error);
                return (error);
+       }
+
+       kq = kqueue_alloc(p);
+       if (kq == NULL) {
+               fp_free(p, fd, fp);
+               return (ENOMEM);
+       }
+
        fp->f_flag = FREAD | FWRITE;
        fp->f_type = DTYPE_KQUEUE;
        fp->f_ops = &kqueueops;
        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;
        fp->f_data = (caddr_t)kq;
+
+       proc_fdlock(p);
+       *fdflags(p, fd) &= ~UF_RESERVED;
+       fp_drop(p, fd, fp, 1);
+       proc_fdunlock(p);
+
        *retval = fd;
        *retval = fd;
-       if (fdp->fd_knlistsize < 0)
-               fdp->fd_knlistsize = 0;         /* this process has a kq */
-       kq->kq_fdp = fdp;
        return (error);
 }
 
        return (error);
 }
 
-#ifndef _SYS_SYSPROTO_H_
-struct kqueue_portset_np_args {
-       int     fd;
-};
-#endif
 int
 int
-kqueue_portset_np(struct proc *p, struct kqueue_portset_np_args *uap, register_t *retval)
+kqueue_portset_np(__unused struct proc *p, 
+                                 __unused struct kqueue_portset_np_args *uap, 
+                                 __unused register_t *retval)
 {
                /* JMM - Placeholder for now */
 {
                /* JMM - Placeholder for now */
-               return (EOPNOTSUPP);
+               return (ENOTSUP);
 }
 
 }
 
-#ifndef _SYS_SYSPROTO_H_
-struct kqueue_from_portset_np_args {
-       int     fd;
-};
-#endif
 int
 int
-kqueue_from_portset_np(struct proc *p, struct kqueue_from_portset_np_args *uap, register_t *retval)
+kqueue_from_portset_np(__unused struct proc *p, 
+                                          __unused struct kqueue_from_portset_np_args *uap, 
+                                          __unused register_t *retval)
 {
                /* JMM - Placeholder for now */
 {
                /* JMM - Placeholder for now */
-               return (EOPNOTSUPP);
+               return (ENOTSUP);
 }
 
 }
 
-#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)
+static int
+kevent_copyin(user_addr_t *addrp, struct kevent *kevp, struct proc *p)
 {
 {
-       struct filedesc* fdp = p->p_fd;
-       struct kqueue *kq;
-       struct file *fp = NULL;
-       struct timespec ts;
-       int i, nerrors, error;
+       int advance;
+       int error;
 
 
-       if (uap->timeout != NULL) {
-               error = copyin((caddr_t)uap->timeout, (caddr_t)&ts, sizeof(ts));
+       if (IS_64BIT_PROCESS(p)) {
+               struct user_kevent kev64;
+
+               advance = sizeof(kev64);
+               error = copyin(*addrp, (caddr_t)&kev64, advance);
                if (error)
                if (error)
-                       goto done;
-               uap->timeout = &ts;
+                       return error;
+               kevp->ident = CAST_DOWN(uintptr_t, kev64.ident);
+               kevp->filter = kev64.filter;
+               kevp->flags = kev64.flags;
+               kevp->fflags = kev64.fflags;
+               kevp->data = CAST_DOWN(intptr_t, kev64.data);
+               kevp->udata = kev64.udata;
+       } else {
+               /*
+                * compensate for legacy in-kernel kevent layout
+                * where the udata field is alredy 64-bit.
+                */
+               advance = sizeof(*kevp) + sizeof(void *) - sizeof(user_addr_t);
+               error = copyin(*addrp, (caddr_t)kevp, advance);
        }
        }
+       if (!error)
+               *addrp += advance;
+       return error;
+}
 
 
-        if (((u_int)uap->fd) >= fdp->fd_nfiles ||
-            (fp = fdp->fd_ofiles[uap->fd]) == NULL ||
-           (fp->f_type != DTYPE_KQUEUE))
-               return (EBADF);
+static int
+kevent_copyout(struct kevent *kevp, user_addr_t *addrp, struct proc *p)
+{
+       int advance;
+       int error;
+
+       if (IS_64BIT_PROCESS(p)) {
+               struct user_kevent kev64;
+
+               kev64.ident = (uint64_t) kevp->ident;
+               kev64.filter = kevp->filter;
+               kev64.flags = kevp->flags;
+               kev64.fflags = kevp->fflags;
+               kev64.data = (int64_t) kevp->data;
+               kev64.udata = kevp->udata;
+               advance = sizeof(kev64);
+               error = copyout((caddr_t)&kev64, *addrp, advance);
+       } else {
+               /*
+                * compensate for legacy in-kernel kevent layout
+                * where the udata field is alredy 64-bit.
+                */
+               advance = sizeof(*kevp) + sizeof(void *) - sizeof(user_addr_t);
+               error = copyout((caddr_t)kevp, *addrp, advance);
+       }
+       if (!error)
+               *addrp += advance;
+       return error;
+}
 
 
-       fhold(fp);
+/*
+ * kevent_continue - continue a kevent syscall after blocking
+ *
+ *     assume we inherit a use count on the kq fileglob.
+ */
 
 
-       kq = (struct kqueue *)fp->f_data;
-       nerrors = 0;
+static void
+kevent_continue(__unused struct kqueue *kq, void *data, int error)
+{
+       struct _kevent *cont_args;
+       struct fileproc *fp;
+       register_t *retval;
+       int noutputs;
+       int fd;
+       struct proc *p = current_proc();
+
+       cont_args = (struct _kevent *)data;
+       noutputs = cont_args->eventout;
+       retval = cont_args->retval;
+       fd = cont_args->fd;
+       fp = cont_args->fp;
+
+       fp_drop(p, fd, fp, 0);
+
+       /* don't restart after signals... */
+       if (error == ERESTART)
+               error = EINTR;
+       else if (error == EWOULDBLOCK)
+               error = 0;
+       if (error == 0)
+               *retval = noutputs;
+       unix_syscall_return(error);
+}
 
 
-       while (uap->nchanges > 0) {
-               int i;
-               int n = uap->nchanges > KQ_NEVENTS ? KQ_NEVENTS : uap->nchanges;
-               struct kevent kq_kev[n];
+/*
+ * kevent - [syscall] register and wait for kernel events
+ *
+ */
 
 
-               error = copyin((caddr_t)uap->changelist, (caddr_t)kq_kev,
-                   n * sizeof(struct kevent));
+int
+kevent(struct proc *p, struct kevent_args *uap, register_t *retval)
+{
+       user_addr_t changelist = uap->changelist;
+       user_addr_t ueventlist = uap->eventlist;
+       int nchanges = uap->nchanges;
+       int nevents = uap->nevents;
+       int fd = uap->fd;
+
+       struct _kevent *cont_args;
+       uthread_t ut;
+       struct kqueue *kq;
+       struct fileproc *fp;
+       struct kevent kev; 
+       int error, noutputs;
+       struct timeval atv;
+
+       /* convert timeout to absolute - if we have one */
+       if (uap->timeout != USER_ADDR_NULL) {
+               struct timeval rtv;
+               if ( IS_64BIT_PROCESS(p) ) {
+                       struct user_timespec ts;
+                       error = copyin( uap->timeout, &ts, sizeof(ts) );
+                       if ((ts.tv_sec & 0xFFFFFFFF00000000ull) != 0)
+                               error = EINVAL;
+                       else
+                               TIMESPEC_TO_TIMEVAL(&rtv, &ts);
+               } else {
+                       struct timespec ts;
+                       error = copyin( uap->timeout, &ts, sizeof(ts) );
+                       TIMESPEC_TO_TIMEVAL(&rtv, &ts);
+               }
                if (error)
                if (error)
-                       goto done;
-               for (i = 0; i < n; i++) {
-                       struct kevent *kevp = &kq_kev[i];
+                       return error;
+               if (itimerfix(&rtv))
+                       return EINVAL;
+               getmicrouptime(&atv);
+               timevaladd(&atv, &rtv);
+       } else {
+               atv.tv_sec = 0;
+               atv.tv_usec = 0;
+       }
 
 
-                       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;
-                               }
-                       }
+       /* get a usecount for the kq itself */
+       if ((error = fp_getfkq(p, fd, &fp, &kq)) != 0)
+               return(error);
+
+       /* register all the change requests the user provided... */
+       noutputs = 0;
+       while (nchanges > 0) {
+               error = kevent_copyin(&changelist, &kev, p);
+               if (error)
+                       break;
+                               
+               kev.flags &= ~EV_SYSFLAGS;
+               error = kevent_register(kq, &kev, p);
+               if (error) {
+                       if (nevents == 0)
+                               break;
+                       kev.flags = EV_ERROR;
+                       kev.data = error;
+                       (void) kevent_copyout(&kev, &ueventlist, p);
+                       nevents--;
+                       noutputs++;
                }
                }
-               uap->nchanges -= n;
-               uap->changelist += n;
-       }
-       if (nerrors) {
-               *retval = nerrors;
-               error = 0;
-               goto done;
+               nchanges--;
        }
 
        }
 
-       error = kqueue_scan(fp, uap->nevents, uap->eventlist, uap->timeout, retval, p);
-done:
-       if (fp != NULL)
-               fdrop(fp, p);
-       return (error);
+       /* store the continuation/completion data in the uthread */
+       ut = (uthread_t)get_bsdthread_info(current_thread());
+       cont_args = (struct _kevent *)&ut->uu_state.ss_kevent;
+       cont_args->fp = fp;
+       cont_args->fd = fd;
+       cont_args->retval = retval;
+       cont_args->eventlist = ueventlist;
+       cont_args->eventcount = nevents;
+       cont_args->eventout = noutputs;
+
+       if (nevents > 0 && noutputs == 0 && error == 0)
+               error = kevent_scan(kq, kevent_callback,
+                                   kevent_continue, cont_args,
+                                   &atv, p);
+       kevent_continue(kq, cont_args, error);
+       /* NOTREACHED */
+       return error;
+}
+
+
+/*
+ * kevent_callback - callback for each individual event
+ *
+ *     called with nothing locked
+ *     caller holds a reference on the kqueue
+ */
+
+static int
+kevent_callback(__unused struct kqueue *kq, struct kevent *kevp, void *data)
+{
+       struct _kevent *cont_args;
+       int error;
+
+       cont_args = (struct _kevent *)data;
+       assert(cont_args->eventout < cont_arg->eventcount);
+
+       /*
+        * Copy out the appropriate amount of event data for this user.
+        */
+       error = kevent_copyout(kevp, &cont_args->eventlist, current_proc());
+
+       /*
+        * If there isn't space for additional events, return
+        * a harmless error to stop the processing here
+        */
+       if (error == 0 && ++cont_args->eventout == cont_args->eventcount)
+                       error = EWOULDBLOCK;
+       return error;
 }
 
 }
 
+/*
+ * kevent_register - add a new event to a kqueue
+ *
+ *     Creates a mapping between the event source and
+ *     the kqueue via a knote data structure.
+ *
+ *     Because many/most the event sources are file
+ *     descriptor related, the knote is linked off
+ *     the filedescriptor table for quick access.
+ *
+ *     called with nothing locked
+ *     caller holds a reference on the kqueue
+ */
+
 int
 int
-kqueue_register(struct kqueue *kq, struct kevent *kev, struct proc *p)
+kevent_register(struct kqueue *kq, struct kevent *kev, struct proc *p)
 {
        struct filedesc *fdp = kq->kq_fdp;
        struct filterops *fops;
 {
        struct filedesc *fdp = kq->kq_fdp;
        struct filterops *fops;
-       struct file *fp = NULL;
+       struct fileproc *fp = NULL;
        struct knote *kn = NULL;
        struct knote *kn = NULL;
-       int s, error = 0;
+       int error = 0;
 
        if (kev->filter < 0) {
                if (kev->filter + EVFILT_SYSCOUNT < 0)
 
        if (kev->filter < 0) {
                if (kev->filter + EVFILT_SYSCOUNT < 0)
@@ -559,20 +1161,22 @@ kqueue_register(struct kqueue *kq, struct kevent *kev, struct proc *p)
                return (EINVAL);
        }
 
                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);
+       /* this iocount needs to be dropped if it is not registered */
+       if (fops->f_isfd && (error = fp_lookup(p, kev->ident, &fp, 0)) != 0)
+               return(error);
 
 
-               if (kev->ident < fdp->fd_knlistsize) {
+ restart:
+       proc_fdlock(p);
+       if (fops->f_isfd) {
+               /* fd-based knotes are linked off the fd table */
+               if (kev->ident < (u_int)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 {
                        SLIST_FOREACH(kn, &fdp->fd_knlist[kev->ident], kn_link)
                                if (kq == kn->kn_kq &&
                                    kev->filter == kn->kn_filter)
                                        break;
                }
        } else {
+               /* hash non-fd knotes here too */
                if (fdp->fd_knhashmask != 0) {
                        struct klist *list;
                        
                if (fdp->fd_knhashmask != 0) {
                        struct klist *list;
                        
@@ -586,329 +1190,497 @@ kqueue_register(struct kqueue *kq, struct kevent *kev, struct proc *p)
                }
        }
 
                }
        }
 
-       if (kn == NULL && ((kev->flags & EV_ADD) == 0)) {
-               error = ENOENT;
-               goto done;
+       /*
+        * kn now contains the matching knote, or NULL if no match
+        */
+       if (kn == NULL) {
+               if ((kev->flags & (EV_ADD|EV_DELETE)) == EV_ADD) {
+                       kn = knote_alloc();
+                       if (kn == NULL) {
+                               proc_fdunlock(p);
+                               error = ENOMEM;
+                               goto done;
+                       }
+                       kn->kn_fp = fp;
+                       kn->kn_kq = kq;
+                       kn->kn_tq = &kq->kq_head;
+                       kn->kn_fop = fops;
+                       kn->kn_sfflags = kev->fflags;
+                       kn->kn_sdata = kev->data;
+                       kev->fflags = 0;
+                       kev->data = 0;
+                       kn->kn_kevent = *kev;
+                       kn->kn_inuse = 1;  /* for f_attach() */
+                       kn->kn_status = 0;
+
+                       /* before anyone can find it */
+                       if (kev->flags & EV_DISABLE)
+                               kn->kn_status |= KN_DISABLED;
+
+                       error = knote_fdpattach(kn, fdp, p);
+                       proc_fdunlock(p);
+
+                       if (error) {
+                               knote_free(kn);
+                               goto done;
+                       }
+
+                       /*
+                        * apply reference count to knote structure, and
+                        * do not release it at the end of this routine.
+                        */
+                       fp = NULL;
+
+                       /*
+                        * If the attach fails here, we can drop it knowing
+                        * that nobody else has a reference to the knote.
+                        */
+                       if ((error = fops->f_attach(kn)) != 0) {
+                               knote_drop(kn, p);
+                               goto done;
+                       }
+               } else {
+                       proc_fdunlock(p);
+                       error = ENOENT;
+                       goto done;
+               }
+       } else {
+               /* existing knote - get kqueue lock */
+               kqlock(kq);
+               proc_fdunlock(p);
+               
+               if (kev->flags & EV_DELETE) {
+                       knote_dequeue(kn);
+                       kn->kn_status |= KN_DISABLED;
+                       if (kqlock2knotedrop(kq, kn)) {
+                               kn->kn_fop->f_detach(kn);
+                               knote_drop(kn, p);
+                       }
+                       goto done;
+               }
+                       
+               /* update status flags for existing knote */
+               if (kev->flags & EV_DISABLE) {
+                       knote_dequeue(kn);
+                       kn->kn_status |= KN_DISABLED;
+               } else if (kev->flags & EV_ENABLE) {
+                       kn->kn_status &= ~KN_DISABLED;
+                       if (kn->kn_status & KN_ACTIVE)
+                               knote_enqueue(kn);
+               }
+
+               /*
+                * If somebody is in the middle of dropping this
+                * knote - go find/insert a new one.  But we have
+                * wait for this one to go away first.
+                */
+               if (!kqlock2knoteusewait(kq, kn))
+                       /* kqueue unlocked */
+                       goto restart;
+
+               /*
+                * 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;
+       }
+                       
+       /* still have use ref on knote */
+       if (kn->kn_fop->f_event(kn, 0)) {
+               if (knoteuse2kqlock(kq, kn))
+                       knote_activate(kn);
+               kqunlock(kq);
+       } else {
+               knote_put(kn);
+       }
+
+done:
+       if (fp != NULL)
+               fp_drop(p, kev->ident, fp, 0);
+       return (error);
+}
+
+/*
+ * kevent_process - process the triggered events in a kqueue
+ *
+ *     Walk the queued knotes and validate that they are
+ *     really still triggered events by calling the filter
+ *     routines (if necessary).  Hold a use reference on
+ *     the knote to avoid it being detached. For each event
+ *     that is still considered triggered, invoke the
+ *     callback routine provided.
+ *
+ *     caller holds a reference on the kqueue.
+ *     kqueue locked on entry and exit - but may be dropped
+ */
+
+static int
+kevent_process(struct kqueue *kq,
+              kevent_callback_t callback,
+              void *data, 
+              int *countp,
+              struct proc *p)
+{
+       struct knote *kn;
+       struct kevent kev;
+       int nevents;
+       int error;
+
+ restart:
+       if (kq->kq_count == 0) {
+               *countp = 0;
+               return 0;
+       }
+
+       /* if someone else is processing the queue, wait */
+       if (!TAILQ_EMPTY(&kq->kq_inprocess)) {
+               assert_wait(&kq->kq_inprocess, THREAD_UNINT);
+               kq->kq_state |= KQ_PROCWAIT;
+               kqunlock(kq);
+               thread_block(THREAD_CONTINUE_NULL);
+               kqlock(kq);
+               goto restart;
        }
 
        }
 
-       /*
-        * kn now contains the matching knote, or NULL if no match
-        */
-       if (kev->flags & EV_ADD) {
+       error = 0;
+       nevents = 0;
+       while (error == 0 &&
+              (kn = TAILQ_FIRST(&kq->kq_head)) != NULL) {
 
 
-               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;
+               /*
+                * move knote to the processed queue.
+                * this is also protected by the kq lock.
+                */
+               assert(kn->kn_tq == &kq->kq_head);
+               TAILQ_REMOVE(&kq->kq_head, kn, kn_tqe);
+               kn->kn_tq = &kq->kq_inprocess;
+               TAILQ_INSERT_TAIL(&kq->kq_inprocess, kn, kn_tqe);
 
 
-                       /*
-                        * apply reference count to knote structure, and
-                        * do not release it at the end of this routine.
-                        */
-                       fp = NULL;
+               /*
+                * Non-EV_ONESHOT events must be re-validated.
+                *
+                * Convert our lock to a use-count and call the event's
+                * filter routine to update.
+                *
+                * If the event is dropping (or no longer valid), we
+                * already have it off the active queue, so just
+                * finish the job of deactivating it.
+                */
+               if ((kn->kn_flags & EV_ONESHOT) == 0) {
+                       int result;
 
 
-                       kn->kn_sfflags = kev->fflags;
-                       kn->kn_sdata = kev->data;
-                       kev->fflags = 0;
-                       kev->data = 0;
-                       kn->kn_kevent = *kev;
+                       if (kqlock2knoteuse(kq, kn)) {
+                               
+                               /* call the filter with just a ref */
+                               result = kn->kn_fop->f_event(kn, 0);
 
 
-                       knote_fdpattach(kn, fdp);
-                       if ((error = fops->f_attach(kn)) != 0) {
+                               if (!knoteuse2kqlock(kq, kn) || result == 0) {
+                                       knote_deactivate(kn);
+                                       continue;
+                               }
+                       } else {
+                               knote_deactivate(kn);
+                               continue;
+                       }
+               }
+
+               /*
+                * Got a valid triggered knote with the kqueue
+                * still locked.  Snapshot the data, and determine
+                * how to dispatch the knote for future events.
+                */
+               kev = kn->kn_kevent;
+
+               /* now what happens to it? */
+               if (kn->kn_flags & EV_ONESHOT) {
+                       knote_deactivate(kn);
+                       if (kqlock2knotedrop(kq, kn)) {
+                               kn->kn_fop->f_detach(kn);
                                knote_drop(kn, p);
                                knote_drop(kn, p);
-                               goto done;
                        }
                        }
+               } else if (kn->kn_flags & EV_CLEAR) {
+                       knote_deactivate(kn);
+                       kn->kn_data = 0;
+                       kn->kn_fflags = 0;
+                       kqunlock(kq);
                } else {
                        /*
                } 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.
+                        * leave on in-process queue.  We'll
+                        * move all the remaining ones back
+                        * the kq queue and wakeup any
+                        * waiters when we are done.
                         */
                         */
-                       kn->kn_sfflags = kev->fflags;
-                       kn->kn_sdata = kev->data;
-                       kn->kn_kevent.udata = kev->udata;
+                       kqunlock(kq);
                }
 
                }
 
-               s = splhigh();
-               if (kn->kn_fop->f_event(kn, 0))
-                       KNOTE_ACTIVATE(kn);
-               splx(s);
+               /* callback to handle each event as we find it */
+               error = (callback)(kq, &kev, data);
+               nevents++;
 
 
-       } else if (kev->flags & EV_DELETE) {
-               kn->kn_fop->f_detach(kn);
-               knote_drop(kn, p);
-               goto done;
+               kqlock(kq);
        }
 
        }
 
-       if ((kev->flags & EV_DISABLE) &&
-           ((kn->kn_status & KN_DISABLED) == 0)) {
-               s = splhigh();
-               kn->kn_status |= KN_DISABLED;
-               splx(s);
+       /*
+        * With the kqueue still locked, move any knotes
+        * remaining on the in-process queue back to the
+        * kq's queue and wake up any waiters.
+        */
+       while ((kn = TAILQ_FIRST(&kq->kq_inprocess)) != NULL) {
+               assert(kn->kn_tq == &kq->kq_inprocess);
+               TAILQ_REMOVE(&kq->kq_inprocess, kn, kn_tqe);
+               kn->kn_tq = &kq->kq_head;
+               TAILQ_INSERT_TAIL(&kq->kq_head, kn, kn_tqe);
        }
        }
-
-       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);
+       if (kq->kq_state & KQ_PROCWAIT) {
+               kq->kq_state &= ~KQ_PROCWAIT;
+               thread_wakeup(&kq->kq_inprocess);
        }
 
        }
 
-done:
-       if (fp != NULL)
-               fdrop(fp, p);
-       return (error);
+       *countp = nevents;
+       return error;
 }
 
 }
 
-static int
-kqueue_scan(struct file *fp, int maxevents, struct kevent *ulistp,
-       const struct timespec *tsp, register_t *retval, struct proc *p)
+
+static void
+kevent_scan_continue(void *data, wait_result_t wait_result)
 {
 {
-       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;
+       uthread_t ut = (uthread_t)get_bsdthread_info(current_thread());
+       struct _kevent_scan * cont_args = &ut->uu_state.ss_kevent_scan;
+       struct kqueue *kq = (struct kqueue *)data;
+       int error;
+       int count;
+
+       /* convert the (previous) wait_result to a proper error */
+       switch (wait_result) {
+       case THREAD_AWAKENED:
+               kqlock(kq);
+               error = kevent_process(kq, cont_args->call, cont_args, &count, current_proc());
+               if (error == 0 && count == 0) {
+                       assert_wait_deadline(kq, THREAD_ABORTSAFE, cont_args->deadline);
+                       kq->kq_state |= KQ_SLEEP;
+                       kqunlock(kq);
+                       thread_block_parameter(kevent_scan_continue, kq);
+                       /* NOTREACHED */
                }
                }
-               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;
+               kqunlock(kq);
+               break;
+       case THREAD_TIMED_OUT:
+               error = EWOULDBLOCK; 
+               break;
+       case THREAD_INTERRUPTED:
+               error = EINTR;
+               break;
+       default:
+               panic("kevent_scan_cont() - invalid wait_result (%d)", wait_result);
+               error = 0;
        }
        }
-       goto start;
+       
+       /* call the continuation with the results */
+       assert(cont_args->cont != NULL);
+       (cont_args->cont)(kq, cont_args->data, error);
+}
 
 
-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;
-       }
+/*
+ * kevent_scan - scan and wait for events in a kqueue
+ *
+ *     Process the triggered events in a kqueue.
+ *
+ *     If there are no events triggered arrange to
+ *     wait for them. If the caller provided a
+ *     continuation routine, then kevent_scan will
+ *     also.
+ *
+ *     The callback routine must be valid.
+ *     The caller must hold a use-count reference on the kq.
+ */
 
 
-       /* 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;
-                       }
+int
+kevent_scan(struct kqueue *kq, 
+           kevent_callback_t callback,
+           kevent_continue_t continuation,
+           void *data,
+           struct timeval *atvp,
+           struct proc *p)
+{
+       thread_continue_t cont = THREAD_CONTINUE_NULL;
+       uint64_t deadline;
+       int error;
+       int first;
 
 
-                       *kevp = kn->kn_kevent;
-                       kevp++;
-                       nkev++;
-                       count--;
+       assert(callback != NULL);
 
 
-                       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--;
+       first = 1;
+       for (;;) {
+               wait_result_t wait_result;
+               int count;
+
+               /*
+                * Make a pass through the kq to find events already
+                * triggered.
+                */
+               kqlock(kq);
+               error = kevent_process(kq, callback, data, &count, p);
+               if (error || count)
+                       break; /* lock still held */
+
+               /* looks like we have to consider blocking */
+               if (first) {
+                       first = 0;
+                       /* convert the timeout to a deadline once */
+                       if (atvp->tv_sec || atvp->tv_usec) {
+                               uint32_t seconds, nanoseconds;
+                               uint64_t now;
+                               
+                               clock_get_uptime(&now);
+                               nanoseconds_to_absolutetime((uint64_t)atvp->tv_sec * NSEC_PER_SEC +
+                                                           atvp->tv_usec * NSEC_PER_USEC,
+                                                           &deadline);
+                               if (now >= deadline) {
+                                       /* non-blocking call */
+                                       error = EWOULDBLOCK;
+                                       break; /* lock still held */
+                               }
+                               deadline -= now;
+                               clock_absolutetime_interval_to_deadline(deadline, &deadline);
                        } else {
                        } else {
-                               TAILQ_INSERT_TAIL(&kq->kq_head, kn, kn_tqe); 
+                               deadline = 0;   /* block forever */
+                       }
+
+                       if (continuation) {
+                               uthread_t ut = (uthread_t)get_bsdthread_info(current_thread());
+                               struct _kevent_scan *cont_args = &ut->uu_state.ss_kevent_scan;
+                               
+                               cont_args->call = callback;
+                               cont_args->cont = continuation;
+                               cont_args->deadline = deadline;
+                               cont_args->data = data;
+                               cont = kevent_scan_continue;
                        }
                }
                        }
                }
-               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;
+
+               /* go ahead and wait */
+               assert_wait_deadline(kq, THREAD_ABORTSAFE, deadline);
+               kq->kq_state |= KQ_SLEEP;
+               kqunlock(kq);
+               wait_result = thread_block_parameter(cont, kq);
+               /* NOTREACHED if (continuation != NULL) */
+
+               switch (wait_result) {
+               case THREAD_AWAKENED:
+                       continue;
+               case THREAD_TIMED_OUT:
+                       return EWOULDBLOCK; 
+               case THREAD_INTERRUPTED:
+                       return EINTR;
+               default:
+                       panic("kevent_scan - bad wait_result (%d)",
+                             wait_result);
+                       error = 0;
+               }
        }
        }
-       TAILQ_REMOVE(&kq->kq_head, &marker, kn_tqe); 
-       splx(s);
-done:
-        *retval = maxevents - count;
-       return (error);
+       kqunlock(kq);
+       return error;
 }
 
 }
 
+
 /*
  * XXX
  * This could be expanded to call kqueue_scan, if desired.
  */
 /*ARGSUSED*/
 static int
 /*
  * 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)
+kqueue_read(__unused struct fileproc *fp, 
+                       __unused struct uio *uio, 
+                       __unused kauth_cred_t cred,
+                       __unused int flags, 
+                       __unused struct proc *p)
 {
        return (ENXIO);
 }
 
 /*ARGSUSED*/
 static int
 {
        return (ENXIO);
 }
 
 /*ARGSUSED*/
 static int
-kqueue_write(struct file *fp, struct uio *uio, struct ucred *cred,
-        int flags, struct proc *p)
+kqueue_write(__unused struct fileproc *fp, 
+                        __unused struct uio *uio, 
+                        __unused kauth_cred_t cred,
+                        __unused int flags, 
+                        __unused struct proc *p)
 {
        return (ENXIO);
 }
 
 /*ARGSUSED*/
 static int
 {
        return (ENXIO);
 }
 
 /*ARGSUSED*/
 static int
-kqueue_ioctl(struct file *fp, u_long com, caddr_t data, struct proc *p)
+kqueue_ioctl(__unused struct fileproc *fp, 
+                        __unused u_long com, 
+                        __unused caddr_t data, 
+                        __unused struct proc *p)
 {
        return (ENOTTY);
 }
 
 /*ARGSUSED*/
 static int
 {
        return (ENOTTY);
 }
 
 /*ARGSUSED*/
 static int
-kqueue_select(struct file *fp, int which, void *wql, struct proc *p)
+kqueue_select(struct fileproc *fp, int which, void *wql, struct proc *p)
 {
        struct kqueue *kq = (struct kqueue *)fp->f_data;
        int retnum = 0;
 {
        struct kqueue *kq = (struct kqueue *)fp->f_data;
        int retnum = 0;
-       int s = splnet();
 
        if (which == FREAD) {
 
        if (which == FREAD) {
+               kqlock(kq);
                 if (kq->kq_count) {
                        retnum = 1;
                } else {
                 if (kq->kq_count) {
                        retnum = 1;
                } else {
-                        selrecord(p, &kq->kq_sel, wql);
+                       selrecord(p, &kq->kq_sel, wql);
                        kq->kq_state |= KQ_SEL;
                }
                        kq->kq_state |= KQ_SEL;
                }
+               kqunlock(kq);
        }
        }
-       splx(s);
        return (retnum);
 }
 
        return (retnum);
 }
 
+/*
+ * kqueue_close -
+ */
 /*ARGSUSED*/
 static int
 /*ARGSUSED*/
 static int
-kqueue_close(struct file *fp, struct proc *p)
+kqueue_close(struct fileglob *fg, 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;
+       struct kqueue *kq = (struct kqueue *)fg->fg_data;
 
 
+       kqueue_dealloc(kq, p);
+       fg->fg_data = NULL;
        return (0);
 }
 
 /*ARGSUSED*/
        return (0);
 }
 
 /*ARGSUSED*/
+/*
+ * The callers has taken a use-count reference on this kqueue and will donate it
+ * to the kqueue we are being added to.  This keeps the kqueue from closing until
+ * that relationship is torn down.
+ */
 static int
 static int
-kqueue_kqfilter(struct file *fp, struct knote *kn, struct proc *p)
+kqueue_kqfilter(__unused struct fileproc *fp, struct knote *kn, __unused struct proc *p)
 {
        struct kqueue *kq = (struct kqueue *)kn->kn_fp->f_data;
 
 {
        struct kqueue *kq = (struct kqueue *)kn->kn_fp->f_data;
 
-       if (kn->kn_filter != EVFILT_READ || (kq->kq_state & KQ_SEL))
+       if (kn->kn_filter != EVFILT_READ)
                return (1);
 
        kn->kn_fop = &kqread_filtops;
                return (1);
 
        kn->kn_fop = &kqread_filtops;
+       kqlock(kq);
        KNOTE_ATTACH(&kq->kq_sel.si_note, kn);
        KNOTE_ATTACH(&kq->kq_sel.si_note, kn);
+       kqunlock(kq);
        return (0);
 }
 
 /*ARGSUSED*/
 int
        return (0);
 }
 
 /*ARGSUSED*/
 int
-kqueue_stat(struct file *fp, struct stat *st, struct proc *p)
+kqueue_stat(struct fileproc *fp, struct stat *st, __unused struct proc *p)
 {
        struct kqueue *kq = (struct kqueue *)fp->f_data;
 
 {
        struct kqueue *kq = (struct kqueue *)fp->f_data;
 
@@ -919,19 +1691,22 @@ kqueue_stat(struct file *fp, struct stat *st, struct proc *p)
        return (0);
 }
 
        return (0);
 }
 
+/*
+ * Called with the kqueue locked
+ */
 static void
 kqueue_wakeup(struct kqueue *kq)
 {
 
        if (kq->kq_state & KQ_SLEEP) {
                kq->kq_state &= ~KQ_SLEEP;
 static void
 kqueue_wakeup(struct kqueue *kq)
 {
 
        if (kq->kq_state & KQ_SLEEP) {
                kq->kq_state &= ~KQ_SLEEP;
-               wakeup(kq);
+               thread_wakeup(kq);
        }
        if (kq->kq_state & KQ_SEL) {
        }
        if (kq->kq_state & KQ_SEL) {
-         // kq->kq_state &= ~KQ_SEL; /* remove for now */
+               kq->kq_state &= ~KQ_SEL;
                selwakeup(&kq->kq_sel);
                selwakeup(&kq->kq_sel);
-       } else
-               KNOTE(&kq->kq_sel.si_note, 0);
+       }
+       KNOTE(&kq->kq_sel.si_note, 0);
 }
 
 void
 }
 
 void
@@ -940,21 +1715,46 @@ klist_init(struct klist *list)
        SLIST_INIT(list);
 }
 
        SLIST_INIT(list);
 }
 
+
 /*
 /*
- * walk down a list of knotes, activating them if their event has triggered.
+ * Query/Post each knote in the object's list
+ *
+ *     The object lock protects the list. It is assumed
+ *     that the filter/event routine for the object can
+ *     determine that the object is already locked (via
+ *     the hind) and not deadlock itself.
+ *
+ *     The object lock should also hold off pending
+ *     detach/drop operations.  But we'll prevent it here
+ *     too - just in case.
  */
 void
 knote(struct klist *list, long hint)
 {
        struct knote *kn;
 
  */
 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);
+       SLIST_FOREACH(kn, list, kn_selnext) {
+               struct kqueue *kq = kn->kn_kq;
+
+               kqlock(kq);
+               if (kqlock2knoteuse(kq, kn)) {
+                       int result;
+
+                       /* call the event with only a use count */
+                       result = kn->kn_fop->f_event(kn, hint);
+
+                       /* if its not going away and triggered */
+                       if (knoteuse2kqlock(kq, kn) && result)
+                               knote_activate(kn);
+                       /* lock held again */
+               }
+               kqunlock(kq);
+       }
 }
 
 /*
  * attach a knote to the specified list.  Return true if this is the first entry.
 }
 
 /*
  * attach a knote to the specified list.  Return true if this is the first entry.
+ * The list is protected by whatever lock the object it is associated with uses.
  */
 int
 knote_attach(struct klist *list, struct knote *kn)
  */
 int
 knote_attach(struct klist *list, struct knote *kn)
@@ -966,6 +1766,7 @@ knote_attach(struct klist *list, struct knote *kn)
 
 /*
  * detach a knote from the specified list.  Return true if that was the last entry.
 
 /*
  * detach a knote from the specified list.  Return true if that was the last entry.
+ * The list is protected by whatever lock the object it is associated with uses.
  */
 int
 knote_detach(struct klist *list, struct knote *kn)
  */
 int
 knote_detach(struct klist *list, struct knote *kn)
@@ -975,67 +1776,88 @@ knote_detach(struct klist *list, struct knote *kn)
 }
 
 /*
 }
 
 /*
- * remove all knotes from a specified klist
+ * remove all knotes referencing a specified fd
+ *
+ * Essentially an inlined knote_remove & knote_drop
+ * when we know for sure that the thing is a file
+ * 
+ * Entered with the proc_fd lock already held.
+ * It returns the same way, but may drop it temporarily.
  */
 void
  */
 void
-knote_remove(struct proc *p, struct klist *list)
+knote_fdclose(struct proc *p, int fd)
 {
 {
+       struct filedesc *fdp = p->p_fd;
+       struct klist *list;
        struct knote *kn;
 
        struct knote *kn;
 
+       list = &fdp->fd_knlist[fd];
        while ((kn = SLIST_FIRST(list)) != NULL) {
        while ((kn = SLIST_FIRST(list)) != NULL) {
-               kn->kn_fop->f_detach(kn);
-               knote_drop(kn, p);
-       }
-}
+               struct kqueue *kq = kn->kn_kq;
 
 
-/*
- * 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];
+               kqlock(kq);
+               proc_fdunlock(p);
+
+               /*
+                * Convert the lock to a drop ref.
+                * If we get it, go ahead and drop it.
+                * Otherwise, we waited for it to
+                * be dropped by the other guy, so
+                * it is safe to move on in the list.
+                */
+               if (kqlock2knotedrop(kq, kn)) {
+                       kn->kn_fop->f_detach(kn);
+                       knote_drop(kn, p);
+               }
+                       
+               proc_fdlock(p);
 
 
-       knote_remove(p, list);
+               /* the fd tables may have changed - start over */
+               list = &fdp->fd_knlist[fd];
+       }
 }
 
 }
 
-static void
-knote_fdpattach(struct knote *kn, struct filedesc *fdp)
+/* proc_fdlock held on entry (and exit) */
+static int
+knote_fdpattach(struct knote *kn, struct filedesc *fdp, __unused struct proc *p)
 {
 {
-       struct klist *list;
-       int size;
+       struct klist *list = NULL;
 
        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)];
 
        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)
+       } else {
+               if ((u_int)fdp->fd_knlistsize <= kn->kn_id) {
+                       u_int size = 0;
+
+                       /* have to grow the fd_knlist */
+                       size = fdp->fd_knlistsize;
+                       while (size <= kn->kn_id)
+                               size += KQEXTENT;
+                       MALLOC(list, struct klist *,
+                              size * sizeof(struct klist *), M_KQUEUE, M_WAITOK);
+                       if (list == NULL)
+                               return (ENOMEM);
+                       
+                       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 *));
                        FREE(fdp->fd_knlist, M_KQUEUE);
                        FREE(fdp->fd_knlist, M_KQUEUE);
-               fdp->fd_knlistsize = size;
-               fdp->fd_knlist = list;
+                       fdp->fd_knlist = list;
+                       fdp->fd_knlistsize = size;
+               }
+               list = &fdp->fd_knlist[kn->kn_id];
        }
        }
-       list = &fdp->fd_knlist[kn->kn_id];
-done:
        SLIST_INSERT_HEAD(list, kn, kn_link);
        SLIST_INSERT_HEAD(list, kn, kn_link);
-       kn->kn_status = 0;
+       return (0);
 }
 
 }
 
+
+
 /*
  * should be called at spl == 0, since we don't want to hold spl
  * while calling fdrop and free.
 /*
  * should be called at spl == 0, since we don't want to hold spl
  * while calling fdrop and free.
@@ -1044,55 +1866,97 @@ static void
 knote_drop(struct knote *kn, struct proc *p)
 {
         struct filedesc *fdp = p->p_fd;
 knote_drop(struct knote *kn, struct proc *p)
 {
         struct filedesc *fdp = p->p_fd;
+       struct kqueue *kq = kn->kn_kq;
        struct klist *list;
 
        struct klist *list;
 
+       proc_fdlock(p);
        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_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);
+       kqlock(kq);
+       knote_dequeue(kn);
+       if (kn->kn_status & KN_DROPWAIT)
+               thread_wakeup(&kn->kn_status);
+       kqunlock(kq);
+       proc_fdunlock(p);
+
        if (kn->kn_fop->f_isfd)
        if (kn->kn_fop->f_isfd)
-               fdrop(kn->kn_fp, p);
+               fp_drop(p, kn->kn_id, kn->kn_fp, 0);
+
        knote_free(kn);
 }
 
        knote_free(kn);
 }
 
+/* called with kqueue lock held */
+static void
+knote_activate(struct knote *kn)
+{
+       struct kqueue *kq = kn->kn_kq;
+
+       kn->kn_status |= KN_ACTIVE;
+       knote_enqueue(kn);
+       kqueue_wakeup(kq);
+ }
+
+/* called with kqueue lock held */
+static void
+knote_deactivate(struct knote *kn)
+{      
+       kn->kn_status &= ~KN_ACTIVE;
+       knote_dequeue(kn);
+}
 
 
+/* called with kqueue lock held */
 static void
 knote_enqueue(struct knote *kn)
 {
        struct kqueue *kq = kn->kn_kq;
 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"));
+       if ((kn->kn_status & (KN_QUEUED | KN_DISABLED)) == 0) {
+               struct kqtailq *tq = kn->kn_tq;
 
 
-       TAILQ_INSERT_TAIL(&kq->kq_head, kn, kn_tqe); 
-       kn->kn_status |= KN_QUEUED;
-       kq->kq_count++;
-       splx(s);
-       kqueue_wakeup(kq);
+               TAILQ_INSERT_TAIL(tq, kn, kn_tqe); 
+               kn->kn_status |= KN_QUEUED;
+               kq->kq_count++;
+       }
 }
 
 }
 
+/* called with kqueue lock held */
 static void
 knote_dequeue(struct knote *kn)
 {
        struct kqueue *kq = kn->kn_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"));
+       assert((kn->kn_status & KN_DISABLED) == 0);
+       if ((kn->kn_status & KN_QUEUED) == KN_QUEUED) {
+               struct kqtailq *tq = kn->kn_tq;
 
 
-       TAILQ_REMOVE(&kq->kq_head, kn, kn_tqe); 
-       kn->kn_status &= ~KN_QUEUED;
-       kq->kq_count--;
-       splx(s);
+               TAILQ_REMOVE(tq, kn, kn_tqe); 
+               kn->kn_tq = &kq->kq_head;
+               kn->kn_status &= ~KN_QUEUED;
+               kq->kq_count--;
+       }
 }
 
 void
 knote_init(void)
 {
        knote_zone = zinit(sizeof(struct knote), 8192*sizeof(struct knote), 8192, "knote zone");
 }
 
 void
 knote_init(void)
 {
        knote_zone = zinit(sizeof(struct knote), 8192*sizeof(struct knote), 8192, "knote zone");
+
+       /* allocate kq lock group attribute and group */
+       kq_lck_grp_attr= lck_grp_attr_alloc_init();
+       lck_grp_attr_setstat(kq_lck_grp_attr);
+
+       kq_lck_grp = lck_grp_alloc_init("kqueue",  kq_lck_grp_attr);
+
+       /* Allocate kq lock attribute */
+       kq_lck_attr = lck_attr_alloc_init();
+       lck_attr_setdefault(kq_lck_attr);
+
+       /* Initialize the timer filter lock */
+       lck_mtx_init(&_filt_timerlock, kq_lck_grp, kq_lck_attr);
 }
 SYSINIT(knote, SI_SUB_PSEUDO, SI_ORDER_ANY, knote_init, NULL)
 
 }
 SYSINIT(knote, SI_SUB_PSEUDO, SI_ORDER_ANY, knote_init, NULL)
 
@@ -1105,7 +1969,7 @@ knote_alloc(void)
 static void
 knote_free(struct knote *kn)
 {
 static void
 knote_free(struct knote *kn)
 {
-       zfree(knote_zone, (vm_offset_t)kn);
+       zfree(knote_zone, kn);
 }
 
 #include <sys/param.h>
 }
 
 #include <sys/param.h>
@@ -1119,8 +1983,17 @@ knote_free(struct knote *kn)
 #include <sys/syslog.h>
 
 
 #include <sys/syslog.h>
 
 
-int    raw_usrreq();
-struct pr_usrreqs event_usrreqs;
+static int kev_attach(struct socket *so, int proto, struct proc *p);
+static int kev_detach(struct socket *so);
+static int kev_control(struct socket *so, u_long cmd, caddr_t data, struct ifnet *ifp, struct proc *p);
+
+struct pr_usrreqs event_usrreqs = {
+     pru_abort_notsupp, pru_accept_notsupp, kev_attach, pru_bind_notsupp, pru_connect_notsupp,
+     pru_connect2_notsupp, kev_control, kev_detach, pru_disconnect_notsupp,
+     pru_listen_notsupp, pru_peeraddr_notsupp, pru_rcvd_notsupp, pru_rcvoob_notsupp,
+     pru_send_notsupp, pru_sense_null, pru_shutdown_notsupp, pru_sockaddr_notsupp,
+     pru_sosend_notsupp, soreceive, pru_sopoll_notsupp
+};
 
 struct protosw eventsw[] = {
      {
 
 struct protosw eventsw[] = {
      {
@@ -1128,7 +2001,14 @@ struct protosw eventsw[] = {
          0,            0,              0,              0,
          0,
          0,            0,              0,              0,
          0,            0,              0,              0,
          0,
          0,            0,              0,              0,
-         0,            &event_usrreqs
+#if __APPLE__
+         0,
+#endif
+         &event_usrreqs,
+         0,            0,              0,
+#if __APPLE__
+         {0, 0},       0,              {0}
+#endif
      }
 };
 
      }
 };
 
@@ -1136,7 +2016,12 @@ static
 struct kern_event_head kern_event_head;
 
 static u_long static_event_id = 0;
 struct kern_event_head kern_event_head;
 
 static u_long static_event_id = 0;
+struct domain *sysdom = &systemdomain;
 
 
+static lck_grp_t               *evt_mtx_grp;
+static lck_attr_t              *evt_mtx_attr;
+static lck_grp_attr_t  *evt_mtx_grp_attr;
+lck_mtx_t                              *evt_mutex;
 /*
  * Install the protosw's for the NKE manager.  Invoked at
  *  extension load time
 /*
  * Install the protosw's for the NKE manager.  Invoked at
  *  extension load time
@@ -1146,14 +2031,32 @@ kern_event_init(void)
 {
     int retval;
 
 {
     int retval;
 
-    if ((retval = net_add_proto(eventsw, &systemdomain)) == 0)
-            return(KERN_SUCCESS);
+    if ((retval = net_add_proto(eventsw, &systemdomain)) != 0) {
+           log(LOG_WARNING, "Can't install kernel events protocol (%d)\n", retval);
+            return(retval);
+       }
     
     
-    log(LOG_WARNING, "Can't install kernel events protocol (%d)\n", retval);
-    return(retval);
+       /*
+        * allocate lock group attribute and group for kern event 
+        */
+       evt_mtx_grp_attr = lck_grp_attr_alloc_init();
+
+       evt_mtx_grp = lck_grp_alloc_init("eventlist", evt_mtx_grp_attr);
+               
+       /*
+        * allocate the lock attribute for mutexes
+        */
+       evt_mtx_attr = lck_attr_alloc_init();
+       lck_attr_setdefault(evt_mtx_attr);
+       evt_mutex = lck_mtx_alloc_init(evt_mtx_grp, evt_mtx_attr);
+       if (evt_mutex == NULL)
+                       return (ENOMEM);
+       
+    return(KERN_SUCCESS);
 }
 
 }
 
-int kev_attach(struct socket *so, int proto, struct proc *p)
+static int
+kev_attach(struct socket *so, __unused int proto, __unused struct proc *p)
 {
      int error;
      struct kern_event_pcb  *ev_pcb;
 {
      int error;
      struct kern_event_pcb  *ev_pcb;
@@ -1162,7 +2065,7 @@ int kev_attach(struct socket *so, int proto, struct proc *p)
      if (error)
           return error;
 
      if (error)
           return error;
 
-     ev_pcb = _MALLOC(sizeof(struct kern_event_pcb), M_PCB, M_WAITOK);
+     MALLOC(ev_pcb, struct kern_event_pcb *, sizeof(struct kern_event_pcb), M_PCB, M_WAITOK);
      if (ev_pcb == 0)
          return ENOBUFS;
 
      if (ev_pcb == 0)
          return ENOBUFS;
 
@@ -1170,25 +2073,69 @@ int kev_attach(struct socket *so, int proto, struct proc *p)
      ev_pcb->vendor_code_filter = 0xffffffff;
 
      so->so_pcb = (caddr_t) ev_pcb;
      ev_pcb->vendor_code_filter = 0xffffffff;
 
      so->so_pcb = (caddr_t) ev_pcb;
+        lck_mtx_lock(evt_mutex);
      LIST_INSERT_HEAD(&kern_event_head, ev_pcb, ev_link);
      LIST_INSERT_HEAD(&kern_event_head, ev_pcb, ev_link);
+        lck_mtx_unlock(evt_mutex);
 
      return 0;
 }
 
 
 
      return 0;
 }
 
 
-int kev_detach(struct socket *so)
+static int
+kev_detach(struct socket *so)
 {
      struct kern_event_pcb *ev_pcb = (struct kern_event_pcb *) so->so_pcb;
 
      if (ev_pcb != 0) {
 {
      struct kern_event_pcb *ev_pcb = (struct kern_event_pcb *) so->so_pcb;
 
      if (ev_pcb != 0) {
-          LIST_REMOVE(ev_pcb, ev_link);
-          FREE(ev_pcb, M_PCB);
-          so->so_pcb = 0;
+               lck_mtx_lock(evt_mutex);
+               LIST_REMOVE(ev_pcb, ev_link);
+               lck_mtx_unlock(evt_mutex);
+               FREE(ev_pcb, M_PCB);
+               so->so_pcb = 0;
+               so->so_flags |= SOF_PCBCLEARING;
      }
 
      return 0;
 }
 
      }
 
      return 0;
 }
 
+/*
+ * For now, kev_vender_code and mbuf_tags use the same
+ * mechanism.
+ */
+extern errno_t mbuf_tag_id_find_internal(const char *string, u_long *out_id,
+                                                                                int create);
+
+errno_t kev_vendor_code_find(
+       const char      *string,
+       u_long          *out_vender_code)
+{
+       if (strlen(string) >= KEV_VENDOR_CODE_MAX_STR_LEN) {
+               return EINVAL;
+       }
+       return mbuf_tag_id_find_internal(string, out_vender_code, 1);
+}
+
+extern void mbuf_tag_id_first_last(u_long *first, u_long *last);
+
+errno_t  kev_msg_post(struct kev_msg *event_msg)
+{
+       u_long  min_vendor, max_vendor;
+       
+       mbuf_tag_id_first_last(&min_vendor, &max_vendor);
+       
+       if (event_msg == NULL)
+               return EINVAL;
+       
+       /* Limit third parties to posting events for registered vendor codes only */
+       if (event_msg->vendor_code < min_vendor ||
+               event_msg->vendor_code > max_vendor)
+       {
+               return EINVAL;
+       }
+       
+       return kev_post_msg(event_msg);
+}
+       
 
 int  kev_post_msg(struct kev_msg *event_msg)
 {
 
 int  kev_post_msg(struct kev_msg *event_msg)
 {
@@ -1196,9 +2143,21 @@ int  kev_post_msg(struct kev_msg *event_msg)
      struct kern_event_pcb  *ev_pcb;
      struct kern_event_msg  *ev;
      char              *tmp;
      struct kern_event_pcb  *ev_pcb;
      struct kern_event_msg  *ev;
      char              *tmp;
-     int               total_size;
+     unsigned long     total_size;
      int               i;
 
      int               i;
 
+       /* Verify the message is small enough to fit in one mbuf w/o cluster */
+       total_size = KEV_MSG_HEADER_SIZE;
+       
+       for (i = 0; i < 5; i++) {
+               if (event_msg->dv[i].data_length == 0)
+                       break;
+               total_size += event_msg->dv[i].data_length;
+       }
+       
+       if (total_size > MLEN) {
+               return EMSGSIZE;
+       }
 
      m = m_get(M_DONTWAIT, MT_DATA);
      if (m == 0)
 
      m = m_get(M_DONTWAIT, MT_DATA);
      if (m == 0)
@@ -1218,7 +2177,6 @@ int  kev_post_msg(struct kev_msg *event_msg)
          tmp += event_msg->dv[i].data_length;
      }
 
          tmp += event_msg->dv[i].data_length;
      }
 
-
      ev->id = ++static_event_id;
      ev->total_size   = total_size;
      ev->vendor_code  = event_msg->vendor_code;
      ev->id = ++static_event_id;
      ev->total_size   = total_size;
      ev->vendor_code  = event_msg->vendor_code;
@@ -1227,6 +2185,7 @@ int  kev_post_msg(struct kev_msg *event_msg)
      ev->event_code   = event_msg->event_code;
 
      m->m_len = total_size;
      ev->event_code   = event_msg->event_code;
 
      m->m_len = total_size;
+     lck_mtx_lock(evt_mutex);
      for (ev_pcb = LIST_FIRST(&kern_event_head); 
          ev_pcb; 
          ev_pcb = LIST_NEXT(ev_pcb, ev_link)) {
      for (ev_pcb = LIST_FIRST(&kern_event_head); 
          ev_pcb; 
          ev_pcb = LIST_NEXT(ev_pcb, ev_link)) {
@@ -1248,67 +2207,69 @@ int  kev_post_msg(struct kev_msg *event_msg)
          m2 = m_copym(m, 0, m->m_len, M_NOWAIT);
          if (m2 == 0) {
               m_free(m);
          m2 = m_copym(m, 0, m->m_len, M_NOWAIT);
          if (m2 == 0) {
               m_free(m);
+                  lck_mtx_unlock(evt_mutex);
               return ENOBUFS;
          }
               return ENOBUFS;
          }
-
-         sbappendrecord(&ev_pcb->ev_socket->so_rcv, m2);
-         sorwakeup(ev_pcb->ev_socket);
+         socket_lock(ev_pcb->ev_socket, 1);
+         if (sbappendrecord(&ev_pcb->ev_socket->so_rcv, m2))
+                 sorwakeup(ev_pcb->ev_socket);
+         socket_unlock(ev_pcb->ev_socket, 1);
      }
 
      }
 
-
      m_free(m);
      m_free(m);
+     lck_mtx_unlock(evt_mutex);
      return 0;
 }
 
      return 0;
 }
 
-
-int kev_control(so, cmd, data, ifp, p)
-    struct socket *so;
-    u_long cmd;
-    caddr_t data;
-    register struct ifnet *ifp;
-    struct proc *p;
+static int
+kev_control(struct socket *so, 
+                       u_long cmd, 
+                       caddr_t data, 
+                       __unused struct ifnet *ifp, 
+                       __unused struct proc *p)
 {
 {
-     struct kev_request *kev_req = (struct kev_request *) data;
-     int  stat = 0;
-     struct kern_event_pcb  *ev_pcb;
-     u_long  *id_value = (u_long *) data;
-
-
-     switch (cmd) {
-
-     case SIOCGKEVID:
-         *id_value = static_event_id;
-         break;
-
-     case SIOCSKEVFILT:
-         ev_pcb = (struct kern_event_pcb *) so->so_pcb;
-         ev_pcb->vendor_code_filter = kev_req->vendor_code;
-         ev_pcb->class_filter     = kev_req->kev_class;
-         ev_pcb->subclass_filter  = kev_req->kev_subclass;
-         break;
-
-     case SIOCGKEVFILT:
-         ev_pcb = (struct kern_event_pcb *) so->so_pcb;
-         kev_req->vendor_code = ev_pcb->vendor_code_filter;
-         kev_req->kev_class   = ev_pcb->class_filter;
-         kev_req->kev_subclass = ev_pcb->subclass_filter;
-         break;
-
-     default:
-         return EOPNOTSUPP;
-     }
-
-     return 0;
+       struct kev_request *kev_req = (struct kev_request *) data;
+       struct kern_event_pcb  *ev_pcb;
+       struct kev_vendor_code *kev_vendor;
+       u_long  *id_value = (u_long *) data;
+       
+       
+       switch (cmd) {
+               
+               case SIOCGKEVID:
+                       *id_value = static_event_id;
+                       break;
+               
+               case SIOCSKEVFILT:
+                       ev_pcb = (struct kern_event_pcb *) so->so_pcb;
+                       ev_pcb->vendor_code_filter = kev_req->vendor_code;
+                       ev_pcb->class_filter     = kev_req->kev_class;
+                       ev_pcb->subclass_filter  = kev_req->kev_subclass;
+                       break;
+               
+               case SIOCGKEVFILT:
+                       ev_pcb = (struct kern_event_pcb *) so->so_pcb;
+                       kev_req->vendor_code = ev_pcb->vendor_code_filter;
+                       kev_req->kev_class   = ev_pcb->class_filter;
+                       kev_req->kev_subclass = ev_pcb->subclass_filter;
+                       break;
+               
+               case SIOCGKEVVENDOR:
+                       kev_vendor = (struct kev_vendor_code*)data;
+                       
+                       /* Make sure string is NULL terminated */
+                       kev_vendor->vendor_string[KEV_VENDOR_CODE_MAX_STR_LEN-1] = 0;
+                       
+                       return mbuf_tag_id_find_internal(kev_vendor->vendor_string,
+                                                                                        &kev_vendor->vendor_code, 0);
+               
+               default:
+                       return ENOTSUP;
+       }
+       
+       return 0;
 }
 
 
 }
 
 
-struct pr_usrreqs event_usrreqs = {
-     pru_abort_notsupp, pru_accept_notsupp, kev_attach, pru_bind_notsupp, pru_connect_notsupp,
-     pru_connect2_notsupp, kev_control, kev_detach, pru_disconnect_notsupp,
-     pru_listen_notsupp, pru_peeraddr_notsupp, pru_rcvd_notsupp, pru_rcvoob_notsupp,
-     pru_send_notsupp, pru_sense_null, pru_shutdown_notsupp, pru_sockaddr_notsupp,
-     pru_sosend_notsupp, soreceive, sopoll
-};
-