]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/net/route.c
xnu-1228.0.2.tar.gz
[apple/xnu.git] / bsd / net / route.c
index 22d8678d35503ae663f877b48909a0f64f8c97a6..ffd62033f8be663e111509e365154bbe9d6f8509 100644 (file)
@@ -1,23 +1,29 @@
 /*
- * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2000-2007 Apple Inc. All rights reserved.
  *
- * @APPLE_LICENSE_HEADER_START@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
  * 
- * The contents of this file constitute Original Code as defined in and
- * are subject to the Apple Public Source License Version 1.1 (the
- * "License").  You may not use this file except in compliance with the
- * License.  Please obtain a copy of the License at
- * http://www.apple.com/publicsource and read it before using this file.
+ * 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. The rights granted to you under the License
+ * may not be used to create, or enable the creation or redistribution of,
+ * unlawful or unlicensed copies of an Apple operating system, or to
+ * circumvent, violate, or enable the circumvention or violation of, any
+ * terms of an Apple operating system software license agreement.
  * 
- * This Original Code and all software distributed under the License are
- * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
+ * Please obtain a copy of the License at
+ * http://www.opensource.apple.com/apsl/ and read it before using this file.
+ * 
+ * The Original Code and all software distributed under the License are
+ * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
- * License for the specific language governing rights and limitations
- * under the License.
+ * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
+ * Please see the License for the specific language governing rights and
+ * limitations under the License.
  * 
- * @APPLE_LICENSE_HEADER_END@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  */
 /*
  * Copyright (c) 1980, 1986, 1991, 1993
  * SUCH DAMAGE.
  *
  *     @(#)route.c     8.2 (Berkeley) 11/15/93
+ * $FreeBSD: src/sys/net/route.c,v 1.59.2.3 2001/07/29 19:18:02 ume Exp $
  */
-
-#if NOTFB31
-#include "opt_inet.h"
-#include "opt_mrouting.h"
-#endif
-
 #include <sys/param.h>
 #include <sys/systm.h>
 #include <sys/malloc.h>
 #include <sys/mbuf.h>
 #include <sys/socket.h>
 #include <sys/domain.h>
+#include <sys/syslog.h>
+#include <sys/queue.h>
+#include <kern/lock.h>
+#include <kern/zalloc.h>
 
 #include <net/if.h>
 #include <net/route.h>
 #include <netinet/in.h>
 #include <netinet/ip_mroute.h>
 
+#include <net/if_dl.h>
+
+#include <libkern/OSAtomic.h>
+#include <libkern/OSDebug.h>
+
+#include <pexpert/pexpert.h>
+
+#define        equal(a1, a2) (bcmp((caddr_t)(a1), (caddr_t)(a2), (a1)->sa_len) == 0)
 #define        SA(p) ((struct sockaddr *)(p))
 
+extern void kdp_set_gateway_mac (void *gatewaymac);
+
+extern struct domain routedomain;
 struct route_cb route_cb;
-static struct rtstat rtstat;
+__private_extern__ struct rtstat rtstat  = { 0, 0, 0, 0, 0 };
 struct radix_node_head *rt_tables[AF_MAX+1];
 
-static int     rttrash;                /* routes not in table but not freed */
+lck_mtx_t      *rt_mtx;        /*### global routing tables mutex for now */
+lck_attr_t     *rt_mtx_attr;
+lck_grp_t      *rt_mtx_grp;
+lck_grp_attr_t         *rt_mtx_grp_attr;
+
+lck_mtx_t      *route_domain_mtx;      /*### global routing tables mutex for now */
+int rttrash = 0;               /* routes not in table but not freed */
+
+static unsigned int rte_debug;
+
+/* Possible flags for rte_debug */
+#define        RTD_DEBUG       0x1     /* enable or disable rtentry debug facility */
+#define        RTD_TRACE       0x2     /* trace alloc, free and refcnt */
+#define        RTD_NO_FREE     0x4     /* don't free (good to catch corruptions) */
+
+static struct zone *rte_zone;                  /* special zone for rtentry */
+#define        RTE_ZONE_MAX            65536           /* maximum elements in zone */
+#define        RTE_ZONE_NAME           "rtentry"       /* name of rtentry zone */
+
+#define        RTD_INUSE               0xFEEDFACE      /* entry is in use */
+#define        RTD_FREED               0xDEADBEEF      /* entry is freed */
+
+#define        RTD_TRSTACK_SIZE        8               /* depth of stack trace */
+#define        RTD_REFHIST_SIZE        4               /* refcnt history size */
+
+/*
+ * Debug variant of rtentry structure.
+ */
+struct rtentry_dbg {
+       struct rtentry  rtd_entry;                      /* rtentry */
+       struct rtentry  rtd_entry_saved;                /* saved rtentry */
+       u_int32_t       rtd_inuse;                      /* in use pattern */
+       u_int16_t       rtd_refhold_cnt;                /* # of rtref */
+       u_int16_t       rtd_refrele_cnt;                /* # of rtunref */
+       /*
+        * Thread and PC stack trace up to RTD_TRSTACK_SIZE
+        * deep during alloc and free.
+        */
+       struct thread   *rtd_alloc_thread;
+       void            *rtd_alloc_stk_pc[RTD_TRSTACK_SIZE];
+       struct thread   *rtd_free_thread;
+       void            *rtd_free_stk_pc[RTD_TRSTACK_SIZE];
+       /*
+        * Circular lists of rtref and rtunref callers.
+        */
+       u_int16_t       rtd_refhold_next;
+       u_int16_t       rtd_refrele_next;
+       struct {
+               struct thread *th;
+               void *pc[RTD_TRSTACK_SIZE];
+       } rtd_refhold[RTD_REFHIST_SIZE];
+       struct {
+               struct thread *th;
+               void *pc[RTD_TRSTACK_SIZE];
+       } rtd_refrele[RTD_REFHIST_SIZE];
+       /*
+        * Trash list linkage
+        */
+       TAILQ_ENTRY(rtentry_dbg) rtd_trash_link;
+};
+
+/* List of trash route entries protected by rt_mtx */
+static TAILQ_HEAD(, rtentry_dbg) rttrash_head;
+
+static inline struct rtentry *rte_alloc_debug(void);
+static inline void rte_free_debug(struct rtentry *);
+static void rt_maskedcopy(struct sockaddr *,
+           struct sockaddr *, struct sockaddr *);
+static void rtable_init(void **);
+static inline void rtref_audit(struct rtentry_dbg *);
+static inline void rtunref_audit(struct rtentry_dbg *);
+
+__private_extern__ u_long route_generation = 0;
+extern int use_routegenid;
 
-static void rt_maskedcopy __P((struct sockaddr *,
-           struct sockaddr *, struct sockaddr *));
-static void rtable_init __P((void **));
 
 static void
-rtable_init(table)
-       void **table;
+rtable_init(void **table)
 {
        struct domain *dom;
        for (dom = domains; dom; dom = dom->dom_next)
@@ -96,32 +182,77 @@ rtable_init(table)
 }
 
 void
-route_init()
+route_init(void)
 {
+       int size;
+
+       PE_parse_boot_arg("rte_debug", &rte_debug);
+       if (rte_debug != 0)
+               rte_debug |= RTD_DEBUG;
+
+       rt_mtx_grp_attr = lck_grp_attr_alloc_init();
+
+       rt_mtx_grp = lck_grp_alloc_init("route", rt_mtx_grp_attr);
+
+       rt_mtx_attr = lck_attr_alloc_init();
+
+       if ((rt_mtx = lck_mtx_alloc_init(rt_mtx_grp, rt_mtx_attr)) == NULL) {
+               printf("route_init: can't alloc rt_mtx\n");
+               return;
+       }
+
+       lck_mtx_lock(rt_mtx);
        rn_init();      /* initialize all zeroes, all ones, mask table */
+       lck_mtx_unlock(rt_mtx);
        rtable_init((void **)rt_tables);
+       route_domain_mtx = routedomain.dom_mtx;
+
+       if (rte_debug & RTD_DEBUG)
+               size = sizeof (struct rtentry_dbg);
+       else
+               size = sizeof (struct rtentry);
+
+       rte_zone = zinit(size, RTE_ZONE_MAX * size, 0, RTE_ZONE_NAME);
+       if (rte_zone == NULL)
+               panic("route_init: failed allocating rte_zone");
+
+       zone_change(rte_zone, Z_EXPAND, TRUE);
+
+       TAILQ_INIT(&rttrash_head);
 }
 
 /*
  * Packet routing routines.
  */
 void
-rtalloc(ro)
-       register struct route *ro;
+rtalloc(struct route *ro)
 {
-       if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP))
-               return;                          /* XXX */
-       ro->ro_rt = rtalloc1(&ro->ro_dst, 1, 0UL);
+       rtalloc_ign(ro, 0UL);
 }
 
 void
-rtalloc_ign(ro, ignore)
-       register struct route *ro;
-       u_long ignore;
+rtalloc_ign_locked(struct route *ro, u_long ignore)
 {
-       if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP))
-               return;                          /* XXX */
-       ro->ro_rt = rtalloc1(&ro->ro_dst, 1, ignore);
+       struct rtentry *rt;
+
+       if ((rt = ro->ro_rt) != NULL) {
+               if (rt->rt_ifp != NULL && rt->rt_flags & RTF_UP)
+                       return;
+               /* XXX - We are probably always at splnet here already. */
+               rtfree_locked(rt);
+               ro->ro_rt = NULL;
+       }
+       ro->ro_rt = rtalloc1_locked(&ro->ro_dst, 1, ignore);
+       if (ro->ro_rt)
+               ro->ro_rt->generation_id = route_generation;
+}
+void
+rtalloc_ign(struct route *ro, u_long ignore)
+{
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_NOTOWNED);
+       lck_mtx_lock(rt_mtx);
+       rtalloc_ign_locked(ro, ignore);
+       lck_mtx_unlock(rt_mtx);
 }
 
 /*
@@ -129,20 +260,16 @@ rtalloc_ign(ro, ignore)
  * Or, at least try.. Create a cloned route if needed.
  */
 struct rtentry *
-rtalloc1(dst, report, ignflags)
-       register struct sockaddr *dst;
-       int report;
-       u_long ignflags;
+rtalloc1_locked(struct sockaddr *dst, int report, u_long ignflags)
 {
-       register struct radix_node_head *rnh = rt_tables[dst->sa_family];
-       register struct rtentry *rt;
-       register struct radix_node *rn;
+       struct radix_node_head *rnh = rt_tables[dst->sa_family];
+       struct rtentry *rt;
+       struct radix_node *rn;
        struct rtentry *newrt = 0;
        struct rt_addrinfo info;
        u_long nflags;
-       int  s = splnet(), err = 0, msgtype = RTM_MISS;
-
-       /* 
+       int  err = 0, msgtype = RTM_MISS;
+       /*
         * Look up the address in the table for that Address Family
         */
        if (rnh && (rn = rnh->rnh_matchaddr((caddr_t)dst, rnh)) &&
@@ -159,7 +286,7 @@ rtalloc1(dst, report, ignflags)
                         * If it requires that it be cloned, do so.
                         * (This implies it wasn't a HOST route.)
                         */
-                       err = rtrequest(RTM_RESOLVE, dst, SA(0),
+                       err = rtrequest_locked(RTM_RESOLVE, dst, SA(0),
                                              SA(0), 0, &newrt);
                        if (err) {
                                /*
@@ -167,19 +294,19 @@ rtalloc1(dst, report, ignflags)
                                 * what we have will do. Return that.
                                 */
                                newrt = rt;
-                               rt->rt_refcnt++;
+                               rtref(rt);
                                goto miss;
                        }
                        if ((rt = newrt) && (rt->rt_flags & RTF_XRESOLVE)) {
                                /*
-                                * If the new route specifies it be 
+                                * If the new route specifies it be
                                 * externally resolved, then go do that.
                                 */
                                msgtype = RTM_RESOLVE;
                                goto miss;
                        }
                } else
-                       rt->rt_refcnt++;
+                       rtref(rt);
        } else {
                /*
                 * Either we hit the root or couldn't find any match,
@@ -198,66 +325,102 @@ rtalloc1(dst, report, ignflags)
                        rt_missmsg(msgtype, &info, 0, err);
                }
        }
-       splx(s);
        return (newrt);
 }
 
+struct rtentry *
+rtalloc1(struct sockaddr *dst, int report, u_long ignflags)
+{
+       struct rtentry * entry;
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_NOTOWNED);
+       lck_mtx_lock(rt_mtx);
+       entry = rtalloc1_locked(dst, report, ignflags);
+       lck_mtx_unlock(rt_mtx);
+       return (entry);
+}
+
 /*
  * Remove a reference count from an rtentry.
  * If the count gets low enough, take it out of the routing table
  */
 void
-rtfree(rt)
-       register struct rtentry *rt;
+rtfree_locked(struct rtentry *rt)
 {
        /*
         * find the tree for that address family
+        * Note: in the case of igmp packets, there might not be an rnh
         */
-       register struct radix_node_head *rnh =
-               rt_tables[rt_key(rt)->sa_family];
-       register struct ifaddr *ifa;
+       struct radix_node_head *rnh;
 
-       if (rt == 0 || rnh == 0)
-               panic("rtfree");
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
+
+       /* See 3582620 - We hit this during the transition from funnels to locks */
+       if (rt == 0) {
+               printf("rtfree - rt is NULL\n");
+               return;
+       }
+       
+       rnh = rt_tables[rt_key(rt)->sa_family];
 
        /*
         * decrement the reference count by one and if it reaches 0,
         * and there is a close function defined, call the close function
         */
-       rt->rt_refcnt--;
-       if(rnh->rnh_close && rt->rt_refcnt == 0) {
+       rtunref(rt);
+       if (rt->rt_refcnt > 0)
+               return;
+
+       if ((rt->rt_flags & RTF_TRACKREFS) != 0)
+               printf("%s rt(%p)->rt_refcnt(%d), caller=%p\n", __FUNCTION__,
+                       rt, rt->rt_refcnt, __builtin_return_address(0));
+       
+       /*
+        * On last reference give the "close method" a chance to cleanup
+        * private state.  This also permits (for IPv4 and IPv6) a chance
+        * to decide if the routing table entry should be purged immediately
+        * or at a later time.  When an immediate purge is to happen the
+        * close routine typically issues RTM_DELETE which clears the RTF_UP
+        * flag on the entry so that the code below reclaims the storage.
+        */
+       if (rnh->rnh_close && rt->rt_refcnt == 0)
                rnh->rnh_close((struct radix_node *)rt, rnh);
-       }
 
        /*
         * If we are no longer "up" (and ref == 0)
         * then we can free the resources associated
         * with the route.
         */
-       if (rt->rt_refcnt <= 0 && (rt->rt_flags & RTF_UP) == 0) {
+       if (!(rt->rt_flags & RTF_UP)) {
                if (rt->rt_nodes->rn_flags & (RNF_ACTIVE | RNF_ROOT))
                        panic ("rtfree 2");
-               /* 
+               /*
                 * the rtentry must have been removed from the routing table
                 * so it is represented in rttrash.. remove that now.
                 */
-               rttrash--;
+               (void) OSDecrementAtomic((SInt32 *)&rttrash);
+               if (rte_debug & RTD_DEBUG) {
+                       TAILQ_REMOVE(&rttrash_head, (struct rtentry_dbg *)rt,
+                           rtd_trash_link);
+               }
 
 #ifdef DIAGNOSTIC
                if (rt->rt_refcnt < 0) {
-                       printf("rtfree: %p not freed (neg refs)\n", rt);
+                       printf("rtfree: %p not freed (neg refs) cnt=%d\n",
+                           rt, rt->rt_refcnt);
                        return;
                }
 #endif
 
-               /* 
+               /*
                 * release references on items we hold them on..
                 * e.g other routes and ifaddrs.
                 */
-               if((ifa = rt->rt_ifa))
-                       IFAFREE(ifa);
-               if (rt->rt_parent) {
-                       RTFREE(rt->rt_parent);
+               if (rt->rt_parent)
+                       rtfree_locked(rt->rt_parent);
+
+               if(rt->rt_ifa) {
+                       ifafree(rt->rt_ifa);
+                       rt->rt_ifa = NULL;
                }
 
                /*
@@ -265,25 +428,151 @@ rtfree(rt)
                 * This also frees the gateway, as they are always malloc'd
                 * together.
                 */
-               Free(rt_key(rt));
+               R_Free(rt_key(rt));
 
                /*
                 * and the rtentry itself of course
                 */
-               Free(rt);
+               rte_free(rt);
        }
 }
 
 void
-ifafree(ifa)
-       register struct ifaddr *ifa;
+rtfree(struct rtentry *rt)
 {
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_NOTOWNED);
+       lck_mtx_lock(rt_mtx);   
+       rtfree_locked(rt);
+       lck_mtx_unlock(rt_mtx);
+}
+
+/*
+ * Decrements the refcount but does not free the route when
+ * the refcount reaches zero. Unless you have really good reason,
+ * use rtfree not rtunref.
+ */
+void
+rtunref(struct rtentry *p)
+{
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
+
+       if (p->rt_refcnt <= 0)
+               panic("rtunref: bad refcnt %d for rt=%p\n", p->rt_refcnt, p);
+
+       if (rte_debug & RTD_DEBUG)
+               rtunref_audit((struct rtentry_dbg *)p);
+
+       p->rt_refcnt--;
+}
+
+static inline void
+rtunref_audit(struct rtentry_dbg *rte)
+{
+       if (rte->rtd_inuse != RTD_INUSE)
+               panic("rtunref: on freed rte=%p\n", rte);
+
+       rte->rtd_refrele_cnt++;
+
+       if (rte_debug & RTD_TRACE) {
+               rte->rtd_refrele[rte->rtd_refrele_next].th = current_thread();
+               bzero(rte->rtd_refrele[rte->rtd_refrele_next].pc,
+                   sizeof (rte->rtd_refrele[rte->rtd_refrele_next].pc));
+               (void) OSBacktrace(rte->rtd_refrele[rte->rtd_refrele_next].pc,
+                   RTD_TRSTACK_SIZE);
+
+               rte->rtd_refrele_next =
+                   (rte->rtd_refrele_next + 1) % RTD_REFHIST_SIZE;
+       }
+}
+
+/*
+ * Add a reference count from an rtentry.
+ */
+void
+rtref(struct rtentry *p)
+{
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
+
+       if (p->rt_refcnt < 0)
+               panic("rtref: bad refcnt %d for rt=%p\n", p->rt_refcnt, p);
+
+       if (rte_debug & RTD_DEBUG)
+               rtref_audit((struct rtentry_dbg *)p);
+
+       p->rt_refcnt++;
+       
+       if ((p->rt_flags & RTF_TRACKREFS) != 0)
+               printf("%s rt(%p)->rt_refcnt(%d), caller=%p\n", __FUNCTION__,
+                       p, p->rt_refcnt, __builtin_return_address(0));
+}
+
+static inline void
+rtref_audit(struct rtentry_dbg *rte)
+{
+       if (rte->rtd_inuse != RTD_INUSE)
+               panic("rtref_audit: on freed rte=%p\n", rte);
+
+       rte->rtd_refhold_cnt++;
+
+       if (rte_debug & RTD_TRACE) {
+               rte->rtd_refhold[rte->rtd_refhold_next].th = current_thread();
+               bzero(rte->rtd_refhold[rte->rtd_refhold_next].pc,
+                   sizeof (rte->rtd_refhold[rte->rtd_refhold_next].pc));
+               (void) OSBacktrace(rte->rtd_refhold[rte->rtd_refhold_next].pc,
+                   RTD_TRSTACK_SIZE);
+
+               rte->rtd_refhold_next =
+                   (rte->rtd_refhold_next + 1) % RTD_REFHIST_SIZE;
+       }
+}
+
+void
+rtsetifa(struct rtentry *rt, struct ifaddr* ifa)
+{
+       if (rt == NULL)
+               panic("rtsetifa");
+
+       if (rt->rt_ifa == ifa)
+               return;
+
+       /* Release the old ifa */
+       if (rt->rt_ifa)
+               ifafree(rt->rt_ifa);
+
+       /* Set rt_ifa */
+       rt->rt_ifa = ifa;
+
+       /* Take a reference to the ifa */
+       if (rt->rt_ifa)
+               ifaref(rt->rt_ifa);
+}
+
+void
+ifafree(struct ifaddr *ifa)
+{
+       int oldval;
+
        if (ifa == NULL)
                panic("ifafree");
-       if (ifa->ifa_refcnt == 0)
+
+       oldval = OSAddAtomic(-1, (SInt32 *)&ifa->ifa_refcnt);
+
+       if (oldval == 0) {
+               if  ((ifa->ifa_debug & IFA_ATTACHED) != 0) {
+                       panic("ifa attached to ifp is being freed\n");
+               }
                FREE(ifa, M_IFADDR);
-       else
-               ifa->ifa_refcnt--;
+       }
+}
+
+void
+ifaref(struct ifaddr *ifa)
+{
+       if (ifa == NULL)
+               panic("ifaref");
+
+       if (OSAddAtomic(1, (SInt32 *)&ifa->ifa_refcnt) == 0xffffffff)
+               panic("ifaref - reference count rolled over!");
 }
 
 /*
@@ -296,35 +585,50 @@ ifafree(ifa)
  *
  */
 void
-rtredirect(dst, gateway, netmask, flags, src, rtp)
-       struct sockaddr *dst, *gateway, *netmask, *src;
-       int flags;
-       struct rtentry **rtp;
+rtredirect(struct sockaddr *dst, struct sockaddr *gateway,
+          struct sockaddr *netmask, int flags, struct sockaddr *src,
+          struct rtentry **rtp)
 {
-       register struct rtentry *rt;
+       struct rtentry *rt;
        int error = 0;
        short *stat = 0;
        struct rt_addrinfo info;
-       struct ifaddr *ifa;
+       struct ifaddr *ifa = NULL;
+
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_NOTOWNED);
+       lck_mtx_lock(rt_mtx);
 
        /* verify the gateway is directly reachable */
        if ((ifa = ifa_ifwithnet(gateway)) == 0) {
                error = ENETUNREACH;
                goto out;
        }
-       rt = rtalloc1(dst, 0, 0UL);
+
+       rt = rtalloc1_locked(dst, 0, RTF_CLONING | RTF_PRCLONING);
        /*
         * If the redirect isn't from our current router for this dst,
         * it's either old or wrong.  If it redirects us to ourselves,
         * we have a routing loop, perhaps as a result of an interface
         * going down recently.
         */
-#define        equal(a1, a2) (bcmp((caddr_t)(a1), (caddr_t)(a2), (a1)->sa_len) == 0)
        if (!(flags & RTF_DONE) && rt &&
-            (!equal(src, rt->rt_gateway) || rt->rt_ifa != ifa))
+            (!equal(src, rt->rt_gateway) || !equal(rt->rt_ifa->ifa_addr,
+            ifa->ifa_addr))) {
                error = EINVAL;
-       else if (ifa_ifwithaddr(gateway))
-               error = EHOSTUNREACH;
+       } else {
+               ifafree(ifa);
+               if ((ifa = ifa_ifwithaddr(gateway))) {
+                       ifafree(ifa);
+                       ifa = NULL;
+                       error = EHOSTUNREACH;
+               }
+       }
+       
+       if (ifa) {
+               ifafree(ifa);
+               ifa = NULL;
+       }
+       
        if (error)
                goto done;
        /*
@@ -347,7 +651,7 @@ rtredirect(dst, gateway, netmask, flags, src, rtp)
                         */
                create:
                        flags |=  RTF_GATEWAY | RTF_DYNAMIC;
-                       error = rtrequest((int)RTM_ADD, dst, gateway,
+                       error = rtrequest_locked((int)RTM_ADD, dst, gateway,
                                    netmask, flags,
                                    (struct rtentry **)0);
                        stat = &rtstat.rts_dynamic;
@@ -371,7 +675,7 @@ done:
                if (rtp && !error)
                        *rtp = rt;
                else
-                       rtfree(rt);
+                       rtfree_locked(rt);
        }
 out:
        if (error)
@@ -384,36 +688,50 @@ out:
        info.rti_info[RTAX_NETMASK] = netmask;
        info.rti_info[RTAX_AUTHOR] = src;
        rt_missmsg(RTM_REDIRECT, &info, flags, error);
+       lck_mtx_unlock(rt_mtx);
 }
 
 /*
 * Routing table ioctl interface.
 */
 int
-rtioctl(req, data, p)
-       int req;
-       caddr_t data;
-       struct proc *p;
+rtioctl(int req, caddr_t data, struct proc *p)
 {
-#if INET
-       /* Multicast goop, grrr... */
-#if MROUTING
+#pragma unused(p)
+#if INET && MROUTING
        return mrt_ioctl(req, data);
 #else
-       return mrt_ioctl(req, data, p);
-#endif
-#else /* INET */
        return ENXIO;
-#endif /* INET */
+#endif
+}
+
+struct ifaddr *
+ifa_ifwithroute(
+       int flags,
+       const struct sockaddr   *dst,
+       const struct sockaddr *gateway)
+{
+       struct ifaddr *ifa;
+
+       lck_mtx_lock(rt_mtx);
+       ifa = ifa_ifwithroute_locked(flags, dst, gateway);
+       lck_mtx_unlock(rt_mtx);
+
+       return (ifa);
 }
 
 struct ifaddr *
-ifa_ifwithroute(flags, dst, gateway)
-       int flags;
-       struct sockaddr *dst, *gateway;
+ifa_ifwithroute_locked(
+       int flags,
+       const struct sockaddr *dst,
+       const struct sockaddr *gateway)
 {
-       register struct ifaddr *ifa;
-       if ((flags & RTF_GATEWAY) == 0) {
+       struct ifaddr *ifa = NULL;
+       struct rtentry *rt = NULL;
+
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
+
+       if (!(flags & RTF_GATEWAY)) {
                /*
                 * If we are adding a route to an interface,
                 * and the interface is a pt to pt link
@@ -421,11 +739,10 @@ ifa_ifwithroute(flags, dst, gateway)
                 * as our clue to the interface.  Otherwise
                 * we can use the local address.
                 */
-               ifa = 0;
                if (flags & RTF_HOST) {
                        ifa = ifa_ifwithdstaddr(dst);
                }
-               if (ifa == 0)
+               if (ifa == NULL)
                        ifa = ifa_ifwithaddr(gateway);
        } else {
                /*
@@ -435,21 +752,46 @@ ifa_ifwithroute(flags, dst, gateway)
                 */
                ifa = ifa_ifwithdstaddr(gateway);
        }
-       if (ifa == 0)
+       if (ifa == NULL)
                ifa = ifa_ifwithnet(gateway);
-       if (ifa == 0) {
-               struct rtentry *rt = rtalloc1(dst, 0, 0UL);
-               if (rt == 0)
-                       return (0);
-               rt->rt_refcnt--;
-               if ((ifa = rt->rt_ifa) == 0)
-                       return (0);
+       if (ifa == NULL) {
+               /* Workaround to avoid gcc warning regarding const variable */
+               rt = rtalloc1_locked((struct sockaddr *)(size_t)dst, 0, 0UL);
+               if (rt != NULL) {
+                       ifa = rt->rt_ifa;
+                       if (ifa != NULL)
+                               ifaref(ifa);
+                       rtunref(rt);
+                       rt = NULL;
+               }
        }
-       if (ifa->ifa_addr->sa_family != dst->sa_family) {
-               struct ifaddr *oifa = ifa;
-               ifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp);
-               if (ifa == 0)
-                       ifa = oifa;
+       if (ifa != NULL && ifa->ifa_addr->sa_family != dst->sa_family) {
+               struct ifaddr *newifa;
+               /* Callee adds reference to newifa upon success */
+               newifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp);
+               if (newifa != NULL) {
+                       ifafree(ifa);
+                       ifa = newifa;
+               }
+       }
+       /*
+        * If we are adding a gateway, it is quite possible that the
+        * routing table has a static entry in place for the gateway,
+        * that may not agree with info garnered from the interfaces.
+        * The routing table should carry more precedence than the
+        * interfaces in this matter.  Must be careful not to stomp
+        * on new entries from rtinit, hence (ifa->ifa_addr != gateway).
+        */
+       if ((ifa == NULL ||
+           !equal(ifa->ifa_addr, (struct sockaddr *)(size_t)gateway)) &&
+           (rt = rtalloc1_locked((struct sockaddr *)(size_t)gateway,
+           0, 0UL)) != NULL) {
+               if (ifa != NULL)
+                       ifafree(ifa);
+               ifa = rt->rt_ifa;
+               if (ifa != NULL)
+                       ifaref(ifa);
+               rtunref(rt);
        }
        return (ifa);
 }
@@ -469,19 +811,23 @@ struct rtfc_arg {
  * all the bits of info needed
  */
 int
-rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
-       int req, flags;
-       struct sockaddr *dst, *gateway, *netmask;
-       struct rtentry **ret_nrt;
+rtrequest_locked(
+       int req,
+       struct sockaddr *dst,
+       struct sockaddr *gateway,
+       struct sockaddr *netmask,
+       int flags,
+       struct rtentry **ret_nrt)
 {
-       int s = splnet(); int error = 0;
-       register struct rtentry *rt;
-       register struct radix_node *rn;
-       register struct radix_node_head *rnh;
-       struct ifaddr *ifa;
+       int error = 0;
+       struct rtentry *rt;
+       struct radix_node *rn;
+       struct radix_node_head *rnh;
+       struct ifaddr *ifa = NULL;
        struct sockaddr *ndst;
 #define senderr(x) { error = x ; goto bad; }
 
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
        /*
         * Find the correct routing tree to use for this Address Family
         */
@@ -505,12 +851,25 @@ rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
                        panic ("rtrequest delete");
                rt = (struct rtentry *)rn;
 
+               /*
+                * Take an extra reference to handle the deletion of a route
+                * entry whose reference count is already 0; e.g. an expiring
+                * cloned route entry or an entry that was added to the table
+                * with 0 reference. If the caller is interested in this route,
+                * we will return it with the reference intact. Otherwise we
+                * will decrement the reference via rtfree_locked() and then
+                * possibly deallocate it.
+                */
+               rtref(rt);
+               rt->rt_flags &= ~RTF_UP;
+
                /*
                 * Now search what's left of the subtree for any cloned
                 * routes which might have been formed from this node.
                 */
-               if ((rt->rt_flags & RTF_PRCLONING) && netmask) {
-                       rnh->rnh_walktree_from(rnh, dst, netmask,
+               if ((rt->rt_flags & (RTF_CLONING | RTF_PRCLONING)) &&
+                   rt_mask(rt)) {
+                       rnh->rnh_walktree_from(rnh, dst, rt_mask(rt),
                                               rt_fixdelete, rt);
                }
 
@@ -521,41 +880,38 @@ rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
                 */
                if (rt->rt_gwroute) {
                        rt = rt->rt_gwroute;
-                       RTFREE(rt);
+                       rtfree_locked(rt);
                        (rt = (struct rtentry *)rn)->rt_gwroute = 0;
                }
 
                /*
-                * NB: RTF_UP must be set during the search above,
-                * because we might delete the last ref, causing
-                * rt to get freed prematurely.
-                *  eh? then why not just add a reference?
-                * I'm not sure how RTF_UP helps matters. (JRE)
-                */
-               rt->rt_flags &= ~RTF_UP;
-
-               /* 
                 * give the protocol a chance to keep things in sync.
                 */
                if ((ifa = rt->rt_ifa) && ifa->ifa_rtrequest)
                        ifa->ifa_rtrequest(RTM_DELETE, rt, SA(0));
+               ifa = NULL;
 
                /*
                 * one more rtentry floating around that is not
                 * linked to the routing table.
                 */
-               rttrash++;
+               (void) OSIncrementAtomic((SInt32 *)&rttrash);
+               if (rte_debug & RTD_DEBUG) {
+                       TAILQ_INSERT_TAIL(&rttrash_head,
+                           (struct rtentry_dbg *)rt, rtd_trash_link);
+               }
 
                /*
                 * If the caller wants it, then it can have it,
                 * but it's up to it to free the rtentry as we won't be
                 * doing it.
                 */
-               if (ret_nrt)
+               if (ret_nrt != NULL) {
+                       /* Return the route to caller with reference intact */
                        *ret_nrt = rt;
-               else if (rt->rt_refcnt <= 0) {
-                       rt->rt_refcnt++;
-                       rtfree(rt);
+               } else {
+                       /* Dereference or deallocate the route */
+                       rtfree_locked(rt);
                }
                break;
 
@@ -563,6 +919,7 @@ rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
                if (ret_nrt == 0 || (rt = *ret_nrt) == 0)
                        senderr(EINVAL);
                ifa = rt->rt_ifa;
+               ifaref(ifa);
                flags = rt->rt_flags &
                    ~(RTF_CLONING | RTF_PRCLONING | RTF_STATIC);
                flags |= RTF_WASCLONED;
@@ -575,12 +932,11 @@ rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
                if ((flags & RTF_GATEWAY) && !gateway)
                        panic("rtrequest: GATEWAY but no gateway");
 
-               if ((ifa = ifa_ifwithroute(flags, dst, gateway)) == 0)
+               if ((ifa = ifa_ifwithroute_locked(flags, dst, gateway)) == 0)
                        senderr(ENETUNREACH);
 
        makeroute:
-               R_Malloc(rt, struct rtentry *, sizeof(*rt));
-               if (rt == 0)
+               if ((rt = rte_alloc()) == NULL)
                        senderr(ENOBUFS);
                Bzero(rt, sizeof(*rt));
                rt->rt_flags = RTF_UP | flags;
@@ -588,8 +944,8 @@ rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
                 * Add the gateway. Possibly re-malloc-ing the storage for it
                 * also add the rt_gwroute if possible.
                 */
-               if (error = rt_setgate(rt, dst, gateway)) {
-                       Free(rt);
+               if ((error = rt_setgate(rt, dst, gateway)) != 0) {
+                       rte_free(rt);
                        senderr(error);
                }
 
@@ -611,10 +967,11 @@ rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
                 * This moved from below so that rnh->rnh_addaddr() can
                 * examine the ifa and  ifa->ifa_ifp if it so desires.
                 */
-               ifa->ifa_refcnt++;
-               rt->rt_ifa = ifa;
-               rt->rt_ifp = ifa->ifa_ifp;
-               rt->rt_dlt = ifa->ifa_dlt;
+               rtsetifa(rt, ifa);
+               rt->rt_ifp = rt->rt_ifa->ifa_ifp;
+
+               /* XXX mtu manipulation will be done in rnh_addaddr -- itojun */
+
                rn = rnh->rnh_addaddr((caddr_t)ndst, (caddr_t)netmask,
                                        rnh, rt->rt_nodes);
                if (rn == 0) {
@@ -626,19 +983,20 @@ rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
                         * mechanism, then we just blow it away and retry
                         * the insertion of the new one.
                         */
-                       rt2 = rtalloc1(dst, 0, RTF_PRCLONING);
+                       rt2 = rtalloc1_locked(dst, 0,
+                           RTF_CLONING | RTF_PRCLONING);
                        if (rt2 && rt2->rt_parent) {
-                               rtrequest(RTM_DELETE, 
+                               rtrequest_locked(RTM_DELETE,
                                          (struct sockaddr *)rt_key(rt2),
                                          rt2->rt_gateway,
                                          rt_mask(rt2), rt2->rt_flags, 0);
-                               RTFREE(rt2);
+                               rtfree_locked(rt2);
                                rn = rnh->rnh_addaddr((caddr_t)ndst,
                                                      (caddr_t)netmask,
                                                      rnh, rt->rt_nodes);
                        } else if (rt2) {
                                /* undo the extra ref we got */
-                               RTFREE(rt2);
+                               rtfree_locked(rt2);
                        }
                }
 
@@ -648,27 +1006,27 @@ rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
                 */
                if (rn == 0) {
                        if (rt->rt_gwroute)
-                               rtfree(rt->rt_gwroute);
+                               rtfree_locked(rt->rt_gwroute);
                        if (rt->rt_ifa) {
-                               IFAFREE(rt->rt_ifa);
+                               ifafree(rt->rt_ifa);
                        }
-                       Free(rt_key(rt));
-                       Free(rt);
+                       R_Free(rt_key(rt));
+                       rte_free(rt);
                        senderr(EEXIST);
                }
 
                rt->rt_parent = 0;
 
-               /* 
+               /*
                 * If we got here from RESOLVE, then we are cloning
-                * so clone the rest, and note that we 
+                * so clone the rest, and note that we
                 * are a clone (and increment the parent's references)
                 */
                if (req == RTM_RESOLVE) {
                        rt->rt_rmx = (*ret_nrt)->rt_rmx; /* copy metrics */
-                       if ((*ret_nrt)->rt_flags & RTF_PRCLONING) {
+                       if ((*ret_nrt)->rt_flags & (RTF_CLONING | RTF_PRCLONING)) {
                                rt->rt_parent = (*ret_nrt);
-                               (*ret_nrt)->rt_refcnt++;
+                               rtref(*ret_nrt);
                        }
                }
 
@@ -678,6 +1036,8 @@ rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
                 */
                if (ifa->ifa_rtrequest)
                        ifa->ifa_rtrequest(req, rt, SA(ret_nrt ? *ret_nrt : 0));
+               ifafree(ifa);
+               ifa = 0;
 
                /*
                 * We repeat the same procedure from rt_setgate() here because
@@ -698,15 +1058,32 @@ rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
                 */
                if (ret_nrt) {
                        *ret_nrt = rt;
-                       rt->rt_refcnt++;
+                       rtref(rt);
                }
                break;
        }
 bad:
-       splx(s);
+       if (ifa)
+               ifafree(ifa);
        return (error);
 }
 
+int
+rtrequest(
+       int req,
+       struct sockaddr *dst,
+       struct sockaddr *gateway,
+       struct sockaddr *netmask,
+       int flags,
+       struct rtentry **ret_nrt)
+{
+       int error;
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_NOTOWNED);
+       lck_mtx_lock(rt_mtx);
+       error = rtrequest_locked(req, dst, gateway, netmask, flags, ret_nrt);
+       lck_mtx_unlock(rt_mtx);
+       return (error);
+}
 /*
  * Called from rtrequest(RTM_DELETE, ...) to fix up the route's ``family''
  * (i.e., the routes related to it by the operation of cloning).  This
@@ -715,15 +1092,16 @@ bad:
  * the late parent (passed in as VP here) are themselves deleted.
  */
 static int
-rt_fixdelete(rn, vp)
-       struct radix_node *rn;
-       void *vp;
+rt_fixdelete(struct radix_node *rn, void *vp)
 {
        struct rtentry *rt = (struct rtentry *)rn;
        struct rtentry *rt0 = vp;
 
-       if (rt->rt_parent == rt0 && !(rt->rt_flags & RTF_PINNED)) {
-               return rtrequest(RTM_DELETE, rt_key(rt),
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
+
+       if (rt->rt_parent == rt0 &&
+           !(rt->rt_flags & (RTF_PINNED | RTF_CLONING | RTF_PRCLONING))) {
+               return rtrequest_locked(RTM_DELETE, rt_key(rt),
                                 (struct sockaddr *)0, rt_mask(rt),
                                 rt->rt_flags, (struct rtentry **)0);
        }
@@ -737,9 +1115,7 @@ rt_fixdelete(rn, vp)
  * network route and (cloned) host routes.  For this reason, a simple check
  * of rt->rt_parent is insufficient; each candidate route must be tested
  * against the (mask, value) of the new route (passed as before in vp)
- * to see if the new route matches it.  Unfortunately, this has the obnoxious
- * property of also triggering for insertion /above/ a pre-existing network
- * route and clones.  Sigh.  This may be fixed some day.
+ * to see if the new route matches it.
  *
  * XXX - it may be possible to do fixdelete() for changes and reserve this
  * routine just for adds.  I'm not sure why I thought it was necessary to do
@@ -750,23 +1126,24 @@ static int rtfcdebug = 0;
 #endif
 
 static int
-rt_fixchange(rn, vp)
-       struct radix_node *rn;
-       void *vp;
+rt_fixchange(struct radix_node *rn, void *vp)
 {
        struct rtentry *rt = (struct rtentry *)rn;
        struct rtfc_arg *ap = vp;
        struct rtentry *rt0 = ap->rt0;
        struct radix_node_head *rnh = ap->rnh;
-       u_char *xk1, *xm1, *xk2;
-       int i, len;
+       u_char *xk1, *xm1, *xk2, *xmp;
+       int i, len, mlen;
 
 #ifdef DEBUG
        if (rtfcdebug)
                printf("rt_fixchange: rt %p, rt0 %p\n", rt, rt0);
 #endif
 
-       if (!rt->rt_parent || (rt->rt_flags & RTF_PINNED)) {
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
+
+       if (!rt->rt_parent ||
+           (rt->rt_flags & (RTF_PINNED | RTF_CLONING | RTF_PRCLONING))) {
 #ifdef DEBUG
                if(rtfcdebug) printf("no parent or pinned\n");
 #endif
@@ -777,7 +1154,7 @@ rt_fixchange(rn, vp)
 #ifdef DEBUG
                if(rtfcdebug) printf("parent match\n");
 #endif
-               return rtrequest(RTM_DELETE, rt_key(rt),
+               return rtrequest_locked(RTM_DELETE, rt_key(rt),
                                 (struct sockaddr *)0, rt_mask(rt),
                                 rt->rt_flags, (struct rtentry **)0);
        }
@@ -793,7 +1170,29 @@ rt_fixchange(rn, vp)
        xm1 = (u_char *)rt_mask(rt0);
        xk2 = (u_char *)rt_key(rt);
 
-       for (i = rnh->rnh_treetop->rn_off; i < len; i++) {
+       /* avoid applying a less specific route */
+       xmp = (u_char *)rt_mask(rt->rt_parent);
+       mlen = ((struct sockaddr *)rt_key(rt->rt_parent))->sa_len;
+       if (mlen > ((struct sockaddr *)rt_key(rt0))->sa_len) {
+#if DEBUG
+               if (rtfcdebug)
+                       printf("rt_fixchange: inserting a less "
+                              "specific route\n");
+#endif
+               return 0;
+       }
+       for (i = rnh->rnh_treetop->rn_offset; i < mlen; i++) {
+               if ((xmp[i] & ~(xmp[i] ^ xm1[i])) != xmp[i]) {
+#if DEBUG
+                       if (rtfcdebug)
+                               printf("rt_fixchange: inserting a less "
+                                      "specific route\n");
+#endif
+                       return 0;
+               }
+       }
+
+       for (i = rnh->rnh_treetop->rn_offset; i < len; i++) {
                if ((xk2[i] & xm1[i]) != xk1[i]) {
 #ifdef DEBUG
                        if(rtfcdebug) printf("no match\n");
@@ -809,25 +1208,25 @@ rt_fixchange(rn, vp)
 #ifdef DEBUG
        if(rtfcdebug) printf("deleting\n");
 #endif
-       return rtrequest(RTM_DELETE, rt_key(rt), (struct sockaddr *)0,
+       return rtrequest_locked(RTM_DELETE, rt_key(rt), (struct sockaddr *)0,
                         rt_mask(rt), rt->rt_flags, (struct rtentry **)0);
 }
 
 int
-rt_setgate(rt0, dst, gate)
-       struct rtentry *rt0;
-       struct sockaddr *dst, *gate;
+rt_setgate(struct rtentry *rt0, struct sockaddr *dst, struct sockaddr *gate)
 {
        caddr_t new, old;
        int dlen = ROUNDUP(dst->sa_len), glen = ROUNDUP(gate->sa_len);
-       register struct rtentry *rt = rt0;
+       struct rtentry *rt = rt0;
        struct radix_node_head *rnh = rt_tables[dst->sa_family];
-
        /*
         * A host route with the destination equal to the gateway
         * will interfere with keeping LLINFO in the routing
         * table, so disallow it.
         */
+       
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
+
        if (((rt0->rt_flags & (RTF_HOST|RTF_GATEWAY|RTF_LLINFO)) ==
                                        (RTF_HOST|RTF_GATEWAY)) &&
            (dst->sa_len == gate->sa_len) &&
@@ -837,7 +1236,7 @@ rt_setgate(rt0, dst, gate)
                 * or a routing redirect, so try to delete it.
                 */
                if (rt_key(rt0))
-                       rtrequest(RTM_DELETE, (struct sockaddr *)rt_key(rt0),
+                       rtrequest_locked(RTM_DELETE, (struct sockaddr *)rt_key(rt0),
                            rt0->rt_gateway, rt_mask(rt0), rt0->rt_flags, 0);
                return EADDRNOTAVAIL;
        }
@@ -867,13 +1266,13 @@ rt_setgate(rt0, dst, gate)
         */
        Bcopy(gate, (rt->rt_gateway = (struct sockaddr *)(new + dlen)), glen);
 
-       /* 
-        * if we are replacing the chunk (or it's new) we need to 
+       /*
+        * if we are replacing the chunk (or it's new) we need to
         * replace the dst as well
         */
        if (old) {
                Bcopy(dst, new, dlen);
-               Free(old);
+               R_Free(old);
        }
 
        /*
@@ -881,7 +1280,7 @@ rt_setgate(rt0, dst, gate)
         * so drop it.
         */
        if (rt->rt_gwroute) {
-               rt = rt->rt_gwroute; RTFREE(rt);
+               rt = rt->rt_gwroute; rtfree_locked(rt);
                rt = rt0; rt->rt_gwroute = 0;
        }
        /*
@@ -895,12 +1294,18 @@ rt_setgate(rt0, dst, gate)
         * This is obviously mandatory when we get rt->rt_output().
         */
        if (rt->rt_flags & RTF_GATEWAY) {
-               rt->rt_gwroute = rtalloc1(gate, 1, RTF_PRCLONING);
+               rt->rt_gwroute = rtalloc1_locked(gate, 1, RTF_PRCLONING);
                if (rt->rt_gwroute == rt) {
-                       RTFREE(rt->rt_gwroute);
+                       rtfree_locked(rt->rt_gwroute);
                        rt->rt_gwroute = 0;
                        return EDQUOT; /* failure */
                }
+               /* Tell the kernel debugger about the new default gateway */
+               if ((AF_INET == rt->rt_gateway->sa_family) && 
+                   rt->rt_gwroute && rt->rt_gwroute->rt_gateway && 
+                   (AF_LINK == rt->rt_gwroute->rt_gateway->sa_family)) {
+                 kdp_set_gateway_mac(((struct sockaddr_dl *)rt0->rt_gwroute->rt_gateway)->sdl_data);
+               }
        }
 
        /*
@@ -920,12 +1325,12 @@ rt_setgate(rt0, dst, gate)
 }
 
 static void
-rt_maskedcopy(src, dst, netmask)
-       struct sockaddr *src, *dst, *netmask;
+rt_maskedcopy(struct sockaddr *src, struct sockaddr *dst,
+             struct sockaddr *netmask)
 {
-       register u_char *cp1 = (u_char *)src;
-       register u_char *cp2 = (u_char *)dst;
-       register u_char *cp3 = (u_char *)netmask;
+       u_char *cp1 = (u_char *)src;
+       u_char *cp2 = (u_char *)dst;
+       u_char *cp3 = (u_char *)netmask;
        u_char *cplim = cp2 + *cp3;
        u_char *cplim2 = cp2 + *cp1;
 
@@ -944,13 +1349,22 @@ rt_maskedcopy(src, dst, netmask)
  * for an interface.
  */
 int
-rtinit(ifa, cmd, flags)
-       register struct ifaddr *ifa;
-       int cmd, flags;
+rtinit(struct ifaddr *ifa, int cmd, int flags)
 {
-       register struct rtentry *rt;
-       register struct sockaddr *dst;
-       register struct sockaddr *deldst;
+       int error;
+       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_NOTOWNED);
+       lck_mtx_lock(rt_mtx);
+       error = rtinit_locked(ifa, cmd, flags);
+       lck_mtx_unlock(rt_mtx);
+       return (error);
+}
+
+int
+rtinit_locked(struct ifaddr *ifa, int cmd, int flags)
+{
+       struct rtentry *rt;
+       struct sockaddr *dst;
+       struct sockaddr *deldst;
        struct mbuf *m = 0;
        struct rtentry *nrt = 0;
        int error;
@@ -962,13 +1376,16 @@ rtinit(ifa, cmd, flags)
         * be confusing at best and possibly worse.
         */
        if (cmd == RTM_DELETE) {
-               /* 
+               /*
                 * It's a delete, so it should already exist..
                 * If it's a net, mask off the host bits
                 * (Assuming we have a mask)
                 */
                if ((flags & RTF_HOST) == 0 && ifa->ifa_netmask) {
-                       m = m_get(M_WAIT, MT_SONAME);
+                       m = m_get(M_DONTWAIT, MT_SONAME);
+                       if (m == NULL) {
+                               return(ENOBUFS);
+                       }
                        deldst = mtod(m, struct sockaddr *);
                        rt_maskedcopy(dst, deldst, ifa->ifa_netmask);
                        dst = deldst;
@@ -979,20 +1396,20 @@ rtinit(ifa, cmd, flags)
                 * We set "report" to FALSE so that if it doesn't exist,
                 * it doesn't report an error or clone a route, etc. etc.
                 */
-               rt = rtalloc1(dst, 0, 0UL);
+               rt = rtalloc1_locked(dst, 0, 0UL);
                if (rt) {
                        /*
                         * Ok so we found the rtentry. it has an extra reference
                         * for us at this stage. we won't need that so
                         * lop that off now.
                         */
-                       rt->rt_refcnt--;
+                       rtunref(rt);
                        if (rt->rt_ifa != ifa) {
                                /*
                                 * If the interface in the rtentry doesn't match
                                 * the interface we are using, then we don't
                                 * want to delete it, so return an error.
-                                * This seems to be the only point of 
+                                * This seems to be the only point of
                                 * this whole RTM_DELETE clause.
                                 */
                                if (m)
@@ -1004,11 +1421,12 @@ rtinit(ifa, cmd, flags)
                /* XXX */
 #if 0
                else {
-                       /* 
+                       /*
                         * One would think that as we are deleting, and we know
                         * it doesn't exist, we could just return at this point
                         * with an "ELSE" clause, but apparently not..
                         */
+                       lck_mtx_unlock(rt_mtx);
                        return (flags & RTF_HOST ? EHOSTUNREACH
                                                        : ENETUNREACH);
                }
@@ -1017,7 +1435,7 @@ rtinit(ifa, cmd, flags)
        /*
         * Do the actual request
         */
-       error = rtrequest(cmd, dst, ifa->ifa_addr, ifa->ifa_netmask,
+       error = rtrequest_locked(cmd, dst, ifa->ifa_addr, ifa->ifa_netmask,
                        flags | ifa->ifa_flags, &nrt);
        if (m)
                (void) m_free(m);
@@ -1030,10 +1448,9 @@ rtinit(ifa, cmd, flags)
                 * notify any listenning routing agents of the change
                 */
                rt_newaddrmsg(cmd, ifa, error, nrt);
-               if (rt->rt_refcnt <= 0) {
-                       rt->rt_refcnt++;
-                       rtfree(rt);
-               }
+               if (use_routegenid)
+                       route_generation++;
+               rtfree_locked(rt);
        }
 
        /*
@@ -1042,16 +1459,14 @@ rtinit(ifa, cmd, flags)
         */
        if (cmd == RTM_ADD && error == 0 && (rt = nrt)) {
                /*
-                * We just wanted to add it.. we don't actually need a reference
-                */
-               rt->rt_refcnt--;
-               /*
-                * If it came back with an unexpected interface, then it must 
+                * If it came back with an unexpected interface, then it must
                 * have already existed or something. (XXX)
                 */
                if (rt->rt_ifa != ifa) {
-                       printf("rtinit: wrong ifa (%p) was (%p)\n", ifa,
-                               rt->rt_ifa);
+                       if (!(rt->rt_ifa->ifa_ifp->if_flags &
+                           (IFF_POINTOPOINT|IFF_LOOPBACK)))
+                               printf("rtinit: wrong ifa (%p) was (%p)\n",
+                                   ifa, rt->rt_ifa);
                        /*
                         * Ask that the protocol in question
                         * remove anything it has associated with
@@ -1059,19 +1474,16 @@ rtinit(ifa, cmd, flags)
                         */
                        if (rt->rt_ifa->ifa_rtrequest)
                            rt->rt_ifa->ifa_rtrequest(RTM_DELETE, rt, SA(0));
-                       /* 
-                        * Remove the referenve to the it's ifaddr.
+                       /*
+                        * Set the route's ifa.
                         */
-                       IFAFREE(rt->rt_ifa);
+                       rtsetifa(rt, ifa);
                        /*
                         * And substitute in references to the ifaddr
                         * we are adding.
                         */
-                       rt->rt_ifa = ifa;
                        rt->rt_ifp = ifa->ifa_ifp;
-                       rt->rt_dlt = ifa->ifa_dlt;
-                       rt->rt_rmx.rmx_mtu = ifa->ifa_ifp->if_mtu;
-                       ifa->ifa_refcnt++;
+                       rt->rt_rmx.rmx_mtu = ifa->ifa_ifp->if_mtu;      /*XXX*/
                        /*
                         * Now ask the protocol to check if it needs
                         * any special processing in its new form.
@@ -1083,6 +1495,84 @@ rtinit(ifa, cmd, flags)
                 * notify any listenning routing agents of the change
                 */
                rt_newaddrmsg(cmd, ifa, error, nrt);
+               if (use_routegenid)
+                       route_generation++;
+               /*
+                * We just wanted to add it; we don't actually need a
+                * reference.  This will result in a route that's added
+                * to the routing table without a reference count.  The
+                * RTM_DELETE code will do the necessary step to adjust
+                * the reference count at deletion time.
+                */
+               rtunref(rt);
        }
        return (error);
 }
+
+struct rtentry *
+rte_alloc(void)
+{
+       if (rte_debug & RTD_DEBUG)
+               return (rte_alloc_debug());
+
+       return ((struct rtentry *)zalloc(rte_zone));
+}
+
+void
+rte_free(struct rtentry *p)
+{
+       if (rte_debug & RTD_DEBUG) {
+               rte_free_debug(p);
+               return;
+       }
+
+       if (p->rt_refcnt != 0)
+               panic("rte_free: rte=%p refcnt=%d non-zero\n", p, p->rt_refcnt);
+
+       zfree(rte_zone, p);
+}
+
+static inline struct rtentry *
+rte_alloc_debug(void)
+{
+       struct rtentry_dbg *rte;
+
+       rte = ((struct rtentry_dbg *)zalloc(rte_zone));
+       if (rte != NULL) {
+               bzero(rte, sizeof (*rte));
+               if (rte_debug & RTD_TRACE) {
+                       rte->rtd_alloc_thread = current_thread();
+                       (void) OSBacktrace(rte->rtd_alloc_stk_pc,
+                           RTD_TRSTACK_SIZE);
+               }
+               rte->rtd_inuse = RTD_INUSE;
+       }
+       return ((struct rtentry *)rte);
+}
+
+static inline void
+rte_free_debug(struct rtentry *p)
+{
+       struct rtentry_dbg *rte = (struct rtentry_dbg *)p;
+
+       if (p->rt_refcnt != 0)
+               panic("rte_free: rte=%p refcnt=%d\n", p, p->rt_refcnt);
+
+       if (rte->rtd_inuse == RTD_FREED)
+               panic("rte_free: double free rte=%p\n", rte);
+       else if (rte->rtd_inuse != RTD_INUSE)
+               panic("rte_free: corrupted rte=%p\n", rte);
+
+       bcopy((caddr_t)p, (caddr_t)&rte->rtd_entry_saved, sizeof (*p));
+       bzero((caddr_t)p, sizeof (*p));
+
+       rte->rtd_inuse = RTD_FREED;
+
+       if (rte_debug & RTD_TRACE) {
+               rte->rtd_free_thread = current_thread();
+               (void) OSBacktrace(rte->rtd_free_stk_pc, RTD_TRSTACK_SIZE);
+       }
+
+       if (!(rte_debug & RTD_NO_FREE))
+               zfree(rte_zone, p);
+}