]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/netinet/in_arp.c
xnu-1504.3.12.tar.gz
[apple/xnu.git] / bsd / netinet / in_arp.c
index cdef93335a661b5746a3997dec7efe0ef99302cb..c553930f1b13fde46a8c263eac47dd2b8837f6f4 100644 (file)
@@ -1,14 +1,19 @@
 /*
- * Copyright (c) 2004 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2004-2009 Apple Inc. All rights reserved.
  *
- * @APPLE_LICENSE_HEADER_START@
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
  * 
  * 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.
+ * 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.
+ * 
+ * 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
@@ -18,7 +23,7 @@
  * 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) 1982, 1989, 1993
 #include <net/if_arp.h>
 #include <net/if_dl.h>
 #include <net/dlil.h>
+#include <net/if_types.h>
 #include <net/route.h>
 #include <netinet/if_ether.h>
 #include <netinet/in_var.h>
+#include <kern/zalloc.h>
 
+#define        SA(p) ((struct sockaddr *)(p))
 #define SIN(s) ((struct sockaddr_in *)s)
 #define CONST_LLADDR(s) ((const u_char*)((s)->sdl_data + (s)->sdl_nlen))
 #define        rt_expire rt_rmx.rmx_expire
+#define        equal(a1, a2) (bcmp((caddr_t)(a1), (caddr_t)(a2), (a1)->sa_len) == 0)
 
 static const size_t MAX_HW_LEN = 10;
 
 SYSCTL_DECL(_net_link_ether);
-SYSCTL_NODE(_net_link_ether, PF_INET, inet, CTLFLAG_RW, 0, "");
+SYSCTL_NODE(_net_link_ether, PF_INET, inet, CTLFLAG_RW|CTLFLAG_LOCKED, 0, "");
 
 /* timer values */
 static int arpt_prune = (5*60*1); /* walk list every 5 minutes */
@@ -103,12 +112,43 @@ SYSCTL_INT(_net_link_ether_inet, OID_AUTO, apple_hwcksum_rx, CTLFLAG_RW,
           &apple_hwcksum_rx, 0, "");
 
 struct llinfo_arp {
+       /*
+        * The following are protected by rnh_lock
+        */
        LIST_ENTRY(llinfo_arp) la_le;
        struct  rtentry *la_rt;
+       /*
+        * The following are protected by rt_lock
+        */
        struct  mbuf *la_hold;          /* last packet until resolved/timeout */
-       long    la_asked;               /* last time we QUERIED for this addr */
+       int32_t la_asked;               /* last time we QUERIED for this addr */
 };
 
+/*
+ * Synchronization notes:
+ *
+ * The global list of ARP entries are stored in llinfo_arp; an entry
+ * gets inserted into the list when the route is created and gets
+ * removed from the list when it is deleted; this is done as part
+ * of RTM_ADD/RTM_RESOLVE/RTM_DELETE in arp_rtrequest().
+ *
+ * Because rnh_lock and rt_lock for the entry are held during those
+ * operations, the same locks (and thus lock ordering) must be used
+ * elsewhere to access the relevant data structure fields:
+ *
+ * la_le.{le_next,le_prev}, la_rt
+ *
+ *     - Routing lock (rnh_lock)
+ *
+ * la_hold, la_asked
+ *
+ *     - Routing entry lock (rt_lock)
+ *
+ * Due to the dependency on rt_lock, llinfo_arp has the same lifetime
+ * as the route entry itself.  When a route is deleted (RTM_DELETE),
+ * it is simply removed from the global list but the memory is not
+ * freed until the route itself is freed.
+ */
 static LIST_HEAD(, llinfo_arp) llinfo_arp;
 
 static int     arp_inuse, arp_allocated;
@@ -116,6 +156,7 @@ static int  arp_inuse, arp_allocated;
 static int     arp_maxtries = 5;
 static int     useloopback = 1; /* use loopback interface for local traffic */
 static int     arp_proxyall = 0;
+static int     arp_sendllconflict = 0;
 
 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, maxtries, CTLFLAG_RW,
           &arp_maxtries, 0, "");
@@ -123,6 +164,8 @@ SYSCTL_INT(_net_link_ether_inet, OID_AUTO, useloopback, CTLFLAG_RW,
           &useloopback, 0, "");
 SYSCTL_INT(_net_link_ether_inet, OID_AUTO, proxyall, CTLFLAG_RW,
           &arp_proxyall, 0, "");
+SYSCTL_INT(_net_link_ether_inet, OID_AUTO, sendllconflict, CTLFLAG_RW,
+          &arp_sendllconflict, 0, "");
 
 static int log_arp_warnings = 0;
 
@@ -130,29 +173,107 @@ SYSCTL_INT(_net_link_ether_inet, OID_AUTO, log_arp_warnings, CTLFLAG_RW,
        &log_arp_warnings, 0,
        "log arp warning messages");
 
+static int keep_announcements = 1;
+SYSCTL_INT(_net_link_ether_inet, OID_AUTO, keep_announcements, CTLFLAG_RW,
+       &keep_announcements, 0,
+       "keep arp announcements");
+
+static int send_conflicting_probes = 1;
+SYSCTL_INT(_net_link_ether_inet, OID_AUTO, send_conflicting_probes, CTLFLAG_RW,
+       &send_conflicting_probes, 0,
+       "send conflicting link-local arp probes");
+
+static errno_t arp_lookup_route(const struct in_addr *, int,
+    int, route_t *, unsigned int);
+static void arptimer(void *);
+static struct llinfo_arp *arp_llinfo_alloc(void);
+static void arp_llinfo_free(void *);
+
 extern u_int32_t       ipv4_ll_arp_aware;
 
+static int arpinit_done;
+
+static struct zone *llinfo_arp_zone;
+#define        LLINFO_ARP_ZONE_MAX     256             /* maximum elements in zone */
+#define        LLINFO_ARP_ZONE_NAME    "llinfo_arp"    /* name for zone */
+
+void
+arp_init(void)
+{
+       if (arpinit_done) {
+               log(LOG_NOTICE, "arp_init called more than once (ignored)\n");
+               return;
+       }
+
+       LIST_INIT(&llinfo_arp);
+
+       llinfo_arp_zone = zinit(sizeof (struct llinfo_arp),
+           LLINFO_ARP_ZONE_MAX * sizeof (struct llinfo_arp), 0,
+           LLINFO_ARP_ZONE_NAME);
+       if (llinfo_arp_zone == NULL)
+               panic("%s: failed allocating llinfo_arp_zone", __func__);
+
+       zone_change(llinfo_arp_zone, Z_EXPAND, TRUE);
+
+       arpinit_done = 1;
+
+       /* start timer */
+       timeout(arptimer, (caddr_t)0, hz);
+}
+
+static struct llinfo_arp *
+arp_llinfo_alloc(void)
+{
+       return (zalloc(llinfo_arp_zone));
+}
+
+static void
+arp_llinfo_free(void *arg)
+{
+       struct llinfo_arp *la = arg;
+
+       if (la->la_le.le_next != NULL || la->la_le.le_prev != NULL) {
+               panic("%s: trying to free %p when it is in use", __func__, la);
+               /* NOTREACHED */
+       }
+
+       /* Just in case there's anything there, free it */
+       if (la->la_hold != NULL) {
+               m_freem(la->la_hold);
+               la->la_hold = NULL;
+       }
+
+       zfree(llinfo_arp_zone, la);
+}
+
 /*
  * Free an arp entry.
  */
 static void
-arptfree(
-       struct llinfo_arp *la)
+arptfree(struct llinfo_arp *la)
 {
        struct rtentry *rt = la->la_rt;
        struct sockaddr_dl *sdl;
-       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
-       if (rt == 0)
-               panic("arptfree");
+
+       lck_mtx_assert(rnh_lock, LCK_MTX_ASSERT_OWNED);
+       RT_LOCK_ASSERT_HELD(rt);
+
        if (rt->rt_refcnt > 0 && (sdl = SDL(rt->rt_gateway)) &&
            sdl->sdl_family == AF_LINK) {
                sdl->sdl_alen = 0;
                la->la_asked = 0;
                rt->rt_flags &= ~RTF_REJECT;
-               return;
+               RT_UNLOCK(rt);
+       } else {
+               /*
+                * Safe to drop rt_lock and use rt_key, since holding
+                * rnh_lock here prevents another thread from calling
+                * rt_setgate() on this route.
+                */
+               RT_UNLOCK(rt);
+               rtrequest_locked(RTM_DELETE, rt_key(rt), NULL, rt_mask(rt),
+                   0, NULL);
        }
-       rtrequest_locked(RTM_DELETE, rt_key(rt), (struct sockaddr *)0, rt_mask(rt),
-                       0, (struct rtentry **)0);
 }
 
 /*
@@ -160,22 +281,25 @@ arptfree(
  */
 /* ARGSUSED */
 static void
-arptimer(
-       __unused void *ignored_arg)
+arptimer(void *ignored_arg)
 {
-       struct llinfo_arp *la = llinfo_arp.lh_first;
-       struct llinfo_arp *ola;
+#pragma unused (ignored_arg)
+       struct llinfo_arp *la, *ola;
        struct timeval timenow;
 
-       lck_mtx_lock(rt_mtx);
+       lck_mtx_lock(rnh_lock);
+       la = llinfo_arp.lh_first;
        getmicrotime(&timenow);
        while ((ola = la) != 0) {
                struct rtentry *rt = la->la_rt;
                la = la->la_le.le_next;
+               RT_LOCK(rt);
                if (rt->rt_expire && rt->rt_expire <= timenow.tv_sec)
                        arptfree(ola); /* timer has expired, clear */
+               else
+                       RT_UNLOCK(rt);
        }
-       lck_mtx_unlock(rt_mtx);
+       lck_mtx_unlock(rnh_lock);
        timeout(arptimer, (caddr_t)0, arpt_prune * hz);
 }
 
@@ -189,17 +313,16 @@ arp_rtrequest(
        __unused struct sockaddr *sa)
 {
        struct sockaddr *gate = rt->rt_gateway;
-       struct llinfo_arp *la = (struct llinfo_arp *)rt->rt_llinfo;
+       struct llinfo_arp *la = rt->rt_llinfo;
        static struct sockaddr_dl null_sdl = {sizeof(null_sdl), AF_LINK, 0, 0, 0, 0, 0, {0}};
-       static int arpinit_done;
        struct timeval timenow;
 
        if (!arpinit_done) {
-               arpinit_done = 1;
-               LIST_INIT(&llinfo_arp);
-               timeout(arptimer, (caddr_t)0, hz);
+               panic("%s: ARP has not been initialized", __func__);
+               /* NOTREACHED */
        }
-       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
+       lck_mtx_assert(rnh_lock, LCK_MTX_ASSERT_OWNED);
+       RT_LOCK_ASSERT_HELD(rt);
 
        if (rt->rt_flags & RTF_GATEWAY)
                return;
@@ -219,22 +342,31 @@ arp_rtrequest(
                        /*
                         * Case 1: This route should come from a route to iface.
                         */
-                       rt_setgate(rt, rt_key(rt),
-                                       (struct sockaddr *)&null_sdl);
-                       gate = rt->rt_gateway;
-                       SDL(gate)->sdl_type = rt->rt_ifp->if_type;
-                       SDL(gate)->sdl_index = rt->rt_ifp->if_index;
-                       rt->rt_expire = timenow.tv_sec;
+                       if (rt_setgate(rt, rt_key(rt),
+                           (struct sockaddr *)&null_sdl) == 0) {
+                               gate = rt->rt_gateway;
+                               SDL(gate)->sdl_type = rt->rt_ifp->if_type;
+                               SDL(gate)->sdl_index = rt->rt_ifp->if_index;
+                               /*
+                                * In case we're called before 1.0 sec.
+                                * has elapsed.
+                                */
+                               rt->rt_expire = MAX(timenow.tv_sec, 1);
+                       }
                        break;
                }
                /* Announce a new entry if requested. */
-               if (rt->rt_flags & RTF_ANNOUNCE)
-                       dlil_send_arp(rt->rt_ifp, ARPOP_REQUEST, SDL(gate), rt_key(rt), (struct sockaddr_dl *)rt_key(rt), NULL);
+               if (rt->rt_flags & RTF_ANNOUNCE) {
+                       RT_UNLOCK(rt);
+                       dlil_send_arp(rt->rt_ifp, ARPOP_REQUEST,
+                           SDL(gate), rt_key(rt), NULL, rt_key(rt));
+                       RT_LOCK(rt);
+               }
                /*FALLTHROUGH*/
        case RTM_RESOLVE:
                if (gate->sa_family != AF_LINK ||
                    gate->sa_len < sizeof(null_sdl)) {
-                       if (log_arp_warnings) 
+                       if (log_arp_warnings)
                                log(LOG_DEBUG, "arp_rtrequest: bad gateway value\n");
                        break;
                }
@@ -246,41 +378,44 @@ arp_rtrequest(
                 * Case 2:  This route may come from cloning, or a manual route
                 * add with a LL address.
                 */
-               R_Malloc(la, struct llinfo_arp *, sizeof(*la));
-               rt->rt_llinfo = (caddr_t)la;
-               if (la == 0) {
-                       if ( log_arp_warnings) 
-                               log(LOG_DEBUG, "arp_rtrequest: malloc failed\n");
+               rt->rt_llinfo = la = arp_llinfo_alloc();
+               if (la == NULL) {
+                       if (log_arp_warnings)
+                               log(LOG_DEBUG, "%s: malloc failed\n", __func__);
                        break;
                }
+               rt->rt_llinfo_free = arp_llinfo_free;
+
                arp_inuse++, arp_allocated++;
                Bzero(la, sizeof(*la));
                la->la_rt = rt;
                rt->rt_flags |= RTF_LLINFO;
                LIST_INSERT_HEAD(&llinfo_arp, la, la_le);
 
-#if INET
                /*
                 * This keeps the multicast addresses from showing up
                 * in `arp -a' listings as unresolved.  It's not actually
                 * functional.  Then the same for broadcast.
                 */
                if (IN_MULTICAST(ntohl(SIN(rt_key(rt))->sin_addr.s_addr))) {
-                       dlil_resolve_multi(rt->rt_ifp, rt_key(rt), gate, sizeof(struct sockaddr_dl));
+                       RT_UNLOCK(rt);
+                       dlil_resolve_multi(rt->rt_ifp, rt_key(rt), gate,
+                           sizeof(struct sockaddr_dl));
+                       RT_LOCK(rt);
                        rt->rt_expire = 0;
                }
                else if (in_broadcast(SIN(rt_key(rt))->sin_addr, rt->rt_ifp)) {
                        struct sockaddr_dl      *gate_ll = SDL(gate);
                        size_t  broadcast_len;
-                       ifnet_llbroadcast_copy_bytes(rt->rt_ifp, LLADDR(gate_ll),
-                                                                                sizeof(gate_ll->sdl_data),
-                                                                                &broadcast_len);
+                       ifnet_llbroadcast_copy_bytes(rt->rt_ifp,
+                           LLADDR(gate_ll), sizeof(gate_ll->sdl_data),
+                           &broadcast_len);
                        gate_ll->sdl_alen = broadcast_len;
                        gate_ll->sdl_family = AF_LINK;
                        gate_ll->sdl_len = sizeof(struct sockaddr_dl);
-                       rt->rt_expire = timenow.tv_sec;
+                       /* In case we're called before 1.0 sec. has elapsed */
+                       rt->rt_expire = MAX(timenow.tv_sec, 1);
                }
-#endif
 
                if (SIN(rt_key(rt))->sin_addr.s_addr ==
                    (IA_SIN(rt->rt_ifa))->sin_addr.s_addr) {
@@ -297,7 +432,7 @@ arp_rtrequest(
                        rt->rt_expire = 0;
                        ifnet_lladdr_copy_bytes(rt->rt_ifp, LLADDR(SDL(gate)), SDL(gate)->sdl_alen = 6);
                        if (useloopback)
-                               rt->rt_ifp = loif;
+                               rt->rt_ifp = lo_ifp;
 
                }
                break;
@@ -306,14 +441,19 @@ arp_rtrequest(
                if (la == 0)
                        break;
                arp_inuse--;
+               /*
+                * Unchain it but defer the actual freeing until the route
+                * itself is to be freed.  rt->rt_llinfo still points to
+                * llinfo_arp, and likewise, la->la_rt still points to this
+                * route entry, except that RTF_LLINFO is now cleared.
+                */
                LIST_REMOVE(la, la_le);
-               rt->rt_llinfo = 0;
+               la->la_le.le_next = NULL;
+               la->la_le.le_prev = NULL;
                rt->rt_flags &= ~RTF_LLINFO;
-               if (la->la_hold) {
+               if (la->la_hold != NULL)
                        m_freem(la->la_hold);
-               }
                la->la_hold = NULL;
-               R_Free((caddr_t)la);
        }
 }
 
@@ -325,7 +465,7 @@ sdl_addr_to_hex(const struct sockaddr_dl *sdl, char * orig_buf, int buflen)
 {
        char *          buf = orig_buf;
        int             i;
-       const u_char *  lladdr = sdl->sdl_data;
+       const u_char *  lladdr = (u_char *)(size_t)sdl->sdl_data;
        int                     maxbytes = buflen / 3;
        
        if (maxbytes > sdl->sdl_alen) {
@@ -344,71 +484,82 @@ sdl_addr_to_hex(const struct sockaddr_dl *sdl, char * orig_buf, int buflen)
 /*
  * arp_lookup_route will lookup the route for a given address.
  *
- * The routing lock must be held. The address must be for a
- * host on a local network on this interface.
+ * The address must be for a host on a local network on this interface.
+ * If the returned route is non-NULL, the route is locked and the caller
+ * is responsible for unlocking it and releasing its reference.
  */
 static errno_t
-arp_lookup_route(
-       const struct in_addr *addr,
-       int     create,
-       int proxy,
-       route_t *route)
+arp_lookup_route(const struct in_addr *addr, int create, int proxy,
+    route_t *route, unsigned int ifscope)
 {
        struct sockaddr_inarp sin = {sizeof(sin), AF_INET, 0, {0}, {0}, 0, 0};
-       const char *why = 0;
+       const char *why = NULL;
        errno_t error = 0;
-       
-       // Caller is responsible for taking the routing lock
-       lck_mtx_assert(rt_mtx, LCK_MTX_ASSERT_OWNED);
+       route_t rt;
+
+       *route = NULL;
 
        sin.sin_addr.s_addr = addr->s_addr;
        sin.sin_other = proxy ? SIN_PROXY : 0;
-       
-       *route = rtalloc1_locked((const struct sockaddr*)&sin, create, 0);
-       if (*route == NULL)
-               return ENETUNREACH;
-       
-       rtunref(*route);
-       
-       if ((*route)->rt_flags & RTF_GATEWAY) {
+
+       rt = rtalloc1_scoped((struct sockaddr*)&sin, create, 0, ifscope);
+       if (rt == NULL)
+               return (ENETUNREACH);
+
+       RT_LOCK(rt);
+
+       if (rt->rt_flags & RTF_GATEWAY) {
                why = "host is not on local network";
-               
-               /* If there are no references to this route, purge it */
-               if ((*route)->rt_refcnt <= 0 && ((*route)->rt_flags & RTF_WASCLONED) != 0) {
-                       rtrequest_locked(RTM_DELETE,
-                                       (struct sockaddr *)rt_key(*route),
-                                       (*route)->rt_gateway, rt_mask(*route),
-                                       (*route)->rt_flags, 0);
-               }
-               *route = NULL;
                error = ENETUNREACH;
-       }
-       else if (((*route)->rt_flags & RTF_LLINFO) == 0) {
+       } else if (!(rt->rt_flags & RTF_LLINFO)) {
                why = "could not allocate llinfo";
-               *route = NULL;
                error = ENOMEM;
-       }
-       else if ((*route)->rt_gateway->sa_family != AF_LINK) {
+       } else if (rt->rt_gateway->sa_family != AF_LINK) {
                why = "gateway route is not ours";
-               *route = NULL;
                error = EPROTONOSUPPORT;
        }
-       
-       if (why && create && log_arp_warnings) {
-               char    tmp[MAX_IPv4_STR_LEN];
-               log(LOG_DEBUG, "arplookup %s failed: %s\n",
-                       inet_ntop(AF_INET, addr, tmp, sizeof(tmp)), why);
+
+       if (error != 0) {
+               if (create && log_arp_warnings) {
+                       char tmp[MAX_IPv4_STR_LEN];
+                       log(LOG_DEBUG, "arplookup link#%d %s failed: %s\n",
+                           ifscope, inet_ntop(AF_INET, addr, tmp,
+                           sizeof (tmp)), why);
+               }
+
+               /*
+                * If there are no references to this route, and it is
+                * a cloned route, and not static, and ARP had created
+                * the route, then purge it from the routing table as
+                * it is probably bogus.
+                */
+               if (rt->rt_refcnt == 1 &&
+                   (rt->rt_flags & (RTF_WASCLONED | RTF_STATIC)) ==
+                   RTF_WASCLONED) {
+                       /*
+                        * Prevent another thread from modiying rt_key,
+                        * rt_gateway via rt_setgate() after rt_lock is
+                        * dropped by marking the route as defunct.
+                        */
+                       rt->rt_flags |= RTF_CONDEMNED;
+                       RT_UNLOCK(rt);
+                       rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway,
+                           rt_mask(rt), rt->rt_flags, 0);
+                       rtfree(rt);
+               } else {
+                       RT_REMREF_LOCKED(rt);
+                       RT_UNLOCK(rt);
+               }
+               return (error);
        }
-       
-       return error;
-}
 
+       /*
+        * Caller releases reference and does RT_UNLOCK(rt).
+        */
+       *route = rt;
+       return (0);
+}
 
-__private_extern__ errno_t
-arp_route_to_gateway_route(
-       const struct sockaddr *net_dest,
-       route_t hint,
-       route_t *out_route);
 /*
  * arp_route_to_gateway_route will find the gateway route for a given route.
  *
@@ -416,192 +567,356 @@ arp_route_to_gateway_route(
  * If the route goes through a gateway, get the route to the gateway.
  * If the gateway route is down, look it up again.
  * If the route is set to reject, verify it hasn't expired.
+ *
+ * If the returned route is non-NULL, the caller is responsible for
+ * releasing the reference and unlocking the route.
  */
+#define senderr(e) { error = (e); goto bad; }
 __private_extern__ errno_t
-arp_route_to_gateway_route(
-       const struct sockaddr *net_dest,
-       route_t hint,
-       route_t *out_route)
+arp_route_to_gateway_route(const struct sockaddr *net_dest, route_t hint0,
+     route_t *out_route)
 {
-       route_t route = hint;
-       *out_route = NULL;
        struct timeval timenow;
-       
-       /* If we got a hint from the higher layers, check it out */
-       if (route) {
-               lck_mtx_lock(rt_mtx);
-               
-               if ((route->rt_flags & RTF_UP) == 0) {
+       route_t rt = hint0, hint = hint0;
+       errno_t error = 0;
+
+       *out_route = NULL;
+
+       /*
+        * Next hop determination.  Because we may involve the gateway route
+        * in addition to the original route, locking is rather complicated.
+        * The general concept is that regardless of whether the route points
+        * to the original route or to the gateway route, this routine takes
+        * an extra reference on such a route.  This extra reference will be
+        * released at the end.
+        *
+        * Care must be taken to ensure that the "hint0" route never gets freed
+        * via rtfree(), since the caller may have stored it inside a struct
+        * route with a reference held for that placeholder.
+        */
+       if (rt != NULL) {
+               unsigned int ifindex;
+
+               RT_LOCK_SPIN(rt);
+               ifindex = rt->rt_ifp->if_index;
+               RT_ADDREF_LOCKED(rt);
+               if (!(rt->rt_flags & RTF_UP)) {
+                       RT_REMREF_LOCKED(rt);
+                       RT_UNLOCK(rt);
                        /* route is down, find a new one */
-                       hint = route = rtalloc1_locked(net_dest, 1, 0);
-                       if (hint) {
-                               rtunref(hint);
-                       }
-                       else {
-                               /* No route to host */
-                               lck_mtx_unlock(rt_mtx);
-                               return EHOSTUNREACH;
+                       hint = rt = rtalloc1_scoped((struct sockaddr *)
+                           (size_t)net_dest, 1, 0, ifindex);
+                       if (hint != NULL) {
+                               RT_LOCK_SPIN(rt);
+                               ifindex = rt->rt_ifp->if_index;
+                       } else {
+                               senderr(EHOSTUNREACH);
                        }
                }
-               
-               if (route->rt_flags & RTF_GATEWAY) {
+
+               /*
+                * We have a reference to "rt" by now; it will either
+                * be released or freed at the end of this routine.
+                */
+               RT_LOCK_ASSERT_HELD(rt);
+               if (rt->rt_flags & RTF_GATEWAY) {
+                       struct rtentry *gwrt = rt->rt_gwroute;
+                       struct sockaddr_in gw;
+
+                       /* If there's no gateway rt, look it up */
+                       if (gwrt == NULL) {
+                               gw = *((struct sockaddr_in *)rt->rt_gateway);
+                               RT_UNLOCK(rt);
+                               goto lookup;
+                       }
+                       /* Become a regular mutex */
+                       RT_CONVERT_LOCK(rt);
+
                        /*
-                        * We need the gateway route. If it is NULL or down,
-                        * look it up.
+                        * Take gwrt's lock while holding route's lock;
+                        * this is okay since gwrt never points back
+                        * to "rt", so no lock ordering issues.
                         */
-                       if (route->rt_gwroute == 0 ||
-                               (route->rt_gwroute->rt_flags & RTF_UP) == 0) {
-                               if (route->rt_gwroute != 0)
-                                       rtfree_locked(route->rt_gwroute);
-                               
-                               route->rt_gwroute = rtalloc1_locked(route->rt_gateway, 1, 0);
-                               if (route->rt_gwroute == 0) {
-                                       lck_mtx_unlock(rt_mtx);
-                                       return EHOSTUNREACH;
+                       RT_LOCK_SPIN(gwrt);
+                       if (!(gwrt->rt_flags & RTF_UP)) {
+                               struct rtentry *ogwrt;
+
+                               rt->rt_gwroute = NULL;
+                               RT_UNLOCK(gwrt);
+                               gw = *((struct sockaddr_in *)rt->rt_gateway);
+                               RT_UNLOCK(rt);
+                               rtfree(gwrt);
+lookup:
+                               gwrt = rtalloc1_scoped(
+                                   (struct sockaddr *)&gw, 1, 0, ifindex);
+
+                               RT_LOCK(rt);
+                               /*
+                                * Bail out if the route is down, no route
+                                * to gateway, circular route, or if the
+                                * gateway portion of "rt" has changed.
+                                */
+                               if (!(rt->rt_flags & RTF_UP) ||
+                                   gwrt == NULL || gwrt == rt ||
+                                   !equal(SA(&gw), rt->rt_gateway)) {
+                                       if (gwrt == rt) {
+                                               RT_REMREF_LOCKED(gwrt);
+                                               gwrt = NULL;
+                                       }
+                                       RT_UNLOCK(rt);
+                                       if (gwrt != NULL)
+                                               rtfree(gwrt);
+                                       senderr(EHOSTUNREACH);
+                               }
+
+                               /* Remove any existing gwrt */
+                               ogwrt = rt->rt_gwroute;
+                               if ((rt->rt_gwroute = gwrt) != NULL)
+                                       RT_ADDREF(gwrt);
+
+                               /* Clean up "rt" now while we can */
+                               if (rt == hint0) {
+                                       RT_REMREF_LOCKED(rt);
+                                       RT_UNLOCK(rt);
+                               } else {
+                                       RT_UNLOCK(rt);
+                                       rtfree(rt);
+                               }
+                               rt = gwrt;
+                               /* Now free the replaced gwrt */
+                               if (ogwrt != NULL)
+                                       rtfree(ogwrt);
+                               /* If still no route to gateway, bail out */
+                               if (rt == NULL)
+                                       senderr(EHOSTUNREACH);
+                       } else {
+                               RT_ADDREF_LOCKED(gwrt);
+                               RT_UNLOCK(gwrt);
+                               /* Clean up "rt" now while we can */
+                               if (rt == hint0) {
+                                       RT_REMREF_LOCKED(rt);
+                                       RT_UNLOCK(rt);
+                               } else {
+                                       RT_UNLOCK(rt);
+                                       rtfree(rt);
                                }
+                               rt = gwrt;
+                       }
+
+                       /* rt == gwrt; if it is now down, give up */
+                       RT_LOCK_SPIN(rt);
+                       if (!(rt->rt_flags & RTF_UP)) {
+                               RT_UNLOCK(rt);
+                               senderr(EHOSTUNREACH);
                        }
-                       
-                       route = route->rt_gwroute;
                }
-               
-               if (route->rt_flags & RTF_REJECT) {
+
+               if (rt->rt_flags & RTF_REJECT) {
                        getmicrotime(&timenow);
-                       if (route->rt_rmx.rmx_expire == 0 ||
-                               timenow.tv_sec < route->rt_rmx.rmx_expire) {
-                               lck_mtx_unlock(rt_mtx);
-                               return route == hint ? EHOSTDOWN : EHOSTUNREACH;
+                       if (rt->rt_rmx.rmx_expire == 0 ||
+                           timenow.tv_sec < rt->rt_rmx.rmx_expire) {
+                               RT_UNLOCK(rt);
+                               senderr(rt == hint ? EHOSTDOWN : EHOSTUNREACH);
                        }
                }
-               
-               lck_mtx_unlock(rt_mtx);
+
+               /* Become a regular mutex */
+               RT_CONVERT_LOCK(rt);
+
+               /* Caller is responsible for cleaning up "rt" */
+               *out_route = rt;
        }
-       
-       *out_route = route;
-       return 0;
+       return (0);
+
+bad:
+       /* Clean up route (either it is "rt" or "gwrt") */
+       if (rt != NULL) {
+               RT_LOCK_SPIN(rt);
+               if (rt == hint0) {
+                       RT_REMREF_LOCKED(rt);
+                       RT_UNLOCK(rt);
+               } else {
+                       RT_UNLOCK(rt);
+                       rtfree(rt);
+               }
+       }
+       return (error);
 }
+#undef senderr
 
+/*
+ * This is the ARP pre-output routine; care must be taken to ensure that
+ * the "hint" route never gets freed via rtfree(), since the caller may
+ * have stored it inside a struct route with a reference held for that
+ * placeholder.
+ */
 errno_t
-arp_lookup_ip(
-       ifnet_t ifp,
-       const struct sockaddr_in *net_dest,
-       struct sockaddr_dl *ll_dest,
-       size_t  ll_dest_len,
-       route_t hint,
-       mbuf_t packet)
+arp_lookup_ip(ifnet_t ifp, const struct sockaddr_in *net_dest,
+    struct sockaddr_dl *ll_dest, size_t        ll_dest_len, route_t hint,
+    mbuf_t packet)
 {
-       route_t route = NULL;
+       route_t route = NULL;   /* output route */
        errno_t result = 0;
        struct sockaddr_dl      *gateway;
        struct llinfo_arp       *llinfo;
        struct timeval timenow;
-       
+
        if (net_dest->sin_family != AF_INET)
-               return EAFNOSUPPORT;
-       
+               return (EAFNOSUPPORT);
+
        if ((ifp->if_flags & (IFF_UP|IFF_RUNNING)) != (IFF_UP|IFF_RUNNING))
-               return ENETDOWN;
-       
+               return (ENETDOWN);
+
        /*
         * If we were given a route, verify the route and grab the gateway
         */
-       if (hint) {
-               result = arp_route_to_gateway_route((const struct sockaddr*)net_dest,
-                                                                                       hint, &route);
+       if (hint != NULL) {
+               /*
+                * Callee holds a reference on the route and returns
+                * with the route entry locked, upon success.
+                */
+               result = arp_route_to_gateway_route((const struct sockaddr*)
+                   net_dest, hint, &route);
                if (result != 0)
-                       return result;
+                       return (result);
+               if (route != NULL)
+                       RT_LOCK_ASSERT_HELD(route);
        }
-       
+
        if (packet->m_flags & M_BCAST) {
-               u_long  broadcast_len;
+               size_t  broadcast_len;
                bzero(ll_dest, ll_dest_len);
-               result = ifnet_llbroadcast_copy_bytes(ifp, LLADDR(ll_dest), ll_dest_len
-                                                                                         - offsetof(struct sockaddr_dl,
-                                                                                         sdl_data), &broadcast_len);
-               if (result != 0) {
-                       return result;
+               result = ifnet_llbroadcast_copy_bytes(ifp, LLADDR(ll_dest),
+                   ll_dest_len - offsetof(struct sockaddr_dl, sdl_data),
+                   &broadcast_len);
+               if (result == 0) {
+                       ll_dest->sdl_alen = broadcast_len;
+                       ll_dest->sdl_family = AF_LINK;
+                       ll_dest->sdl_len = sizeof(struct sockaddr_dl);
                }
-               
-               ll_dest->sdl_alen = broadcast_len;
-               ll_dest->sdl_family = AF_LINK;
-               ll_dest->sdl_len = sizeof(struct sockaddr_dl);
-               
-               return 0;
+               goto release;
        }
        if (packet->m_flags & M_MCAST) {
-               return dlil_resolve_multi(ifp, (const struct sockaddr*)net_dest,
-                                                                  (struct sockaddr*)ll_dest, ll_dest_len);
+               if (route != NULL)
+                       RT_UNLOCK(route);
+               result = dlil_resolve_multi(ifp,
+                   (const struct sockaddr*)net_dest,
+                   (struct sockaddr*)ll_dest, ll_dest_len);
+               if (route != NULL)
+                       RT_LOCK(route);
+               goto release;
        }
-       
-       lck_mtx_lock(rt_mtx);
-       
+
        /*
         * If we didn't find a route, or the route doesn't have
         * link layer information, trigger the creation of the
         * route and link layer information.
         */
-       if (route == NULL || route->rt_llinfo == NULL)
-               result = arp_lookup_route(&net_dest->sin_addr, 1, 0, &route);
-       
+       if (route == NULL || route->rt_llinfo == NULL) {
+               /* Clean up now while we can */
+               if (route != NULL) {
+                       if (route == hint) {
+                               RT_REMREF_LOCKED(route);
+                               RT_UNLOCK(route);
+                       } else {
+                               RT_UNLOCK(route);
+                               rtfree(route);
+                       }
+               }
+               /*
+                * Callee holds a reference on the route and returns
+                * with the route entry locked, upon success.
+                */
+               result = arp_lookup_route(&net_dest->sin_addr, 1, 0, &route,
+                   ifp->if_index);
+               if (result == 0)
+                       RT_LOCK_ASSERT_HELD(route);
+       }
+
        if (result || route == NULL || route->rt_llinfo == NULL) {
                char    tmp[MAX_IPv4_STR_LEN];
-               lck_mtx_unlock(rt_mtx);
-               if (log_arp_warnings)
-                       log(LOG_DEBUG, "arpresolve: can't allocate llinfo for %s\n",
-                               inet_ntop(AF_INET, &net_dest->sin_addr, tmp, sizeof(tmp)));
-               return result;
+
+               /* In case result is 0 but no route, return an error */
+               if (result == 0)
+                       result = EHOSTUNREACH;
+
+               if (log_arp_warnings &&
+                   route != NULL && route->rt_llinfo == NULL)
+                       log(LOG_DEBUG, "arpresolve: can't allocate llinfo "
+                           "for %s\n", inet_ntop(AF_INET, &net_dest->sin_addr,
+                           tmp, sizeof(tmp)));
+               goto release;
        }
-       
+
        /*
         * Now that we have the right route, is it filled in?
         */
        gateway = SDL(route->rt_gateway);
        getmicrotime(&timenow);
-       if ((route->rt_rmx.rmx_expire == 0 || route->rt_rmx.rmx_expire > timenow.tv_sec) &&
-               gateway != NULL && gateway->sdl_family == AF_LINK && gateway->sdl_alen != 0) {
+       if ((route->rt_rmx.rmx_expire == 0 ||
+           route->rt_rmx.rmx_expire > timenow.tv_sec) && gateway != NULL &&
+           gateway->sdl_family == AF_LINK && gateway->sdl_alen != 0) {
                bcopy(gateway, ll_dest, MIN(gateway->sdl_len, ll_dest_len));
-               lck_mtx_unlock(rt_mtx);
-               return 0;
+               result = 0;
+               goto release;
        }
-       
+
+       if (ifp->if_flags & IFF_NOARP) {
+               result = ENOTSUP;
+               goto release;
+       }
+
        /*
         * Route wasn't complete/valid. We need to arp.
         */
-       if (ifp->if_flags & IFF_NOARP) {
-               lck_mtx_unlock(rt_mtx);
-               return ENOTSUP;
-       }
-       
-       llinfo = (struct llinfo_arp*)route->rt_llinfo;
-       if (packet) {
-               if (llinfo->la_hold) {
+       llinfo = route->rt_llinfo;
+       if (packet != NULL) {
+               if (llinfo->la_hold != NULL)
                        m_freem(llinfo->la_hold);
-               }
                llinfo->la_hold = packet;
        }
-       
+
        if (route->rt_rmx.rmx_expire) {
                route->rt_flags &= ~RTF_REJECT;
-               if (llinfo->la_asked == 0 || route->rt_rmx.rmx_expire != timenow.tv_sec) {
+               if (llinfo->la_asked == 0 ||
+                   route->rt_rmx.rmx_expire != timenow.tv_sec) {
                        route->rt_rmx.rmx_expire = timenow.tv_sec;
                        if (llinfo->la_asked++ < arp_maxtries) {
-                               lck_mtx_unlock(rt_mtx);
-                               dlil_send_arp(ifp, ARPOP_REQUEST, NULL, route->rt_ifa->ifa_addr,
-                                                         NULL, (const struct sockaddr*)net_dest);
-                               return EJUSTRETURN;
-                       }
-                       else {
+                               struct ifaddr *rt_ifa = route->rt_ifa;
+                               ifaref(rt_ifa);
+                               RT_UNLOCK(route);
+                               dlil_send_arp(ifp, ARPOP_REQUEST, NULL,
+                                   rt_ifa->ifa_addr, NULL,
+                                   (const struct sockaddr*)net_dest);
+                               ifafree(rt_ifa);
+                               RT_LOCK(route);
+                               result = EJUSTRETURN;
+                               goto release;
+                       } else {
                                route->rt_flags |= RTF_REJECT;
                                route->rt_rmx.rmx_expire += arpt_down;
                                llinfo->la_asked = 0;
-                               llinfo->la_hold = 0;
-                               lck_mtx_unlock(rt_mtx);
-                               return EHOSTUNREACH;
+                               llinfo->la_hold = NULL;
+                               result = EHOSTUNREACH;
+                               goto release;
                        }
                }
        }
-       lck_mtx_unlock(rt_mtx);
-       
-       return EJUSTRETURN;
+
+       /* The packet is now held inside la_hold (can "packet" be NULL?) */
+       result = EJUSTRETURN;
+
+release:
+       if (route != NULL) {
+               if (route == hint) {
+                       RT_REMREF_LOCKED(route);
+                       RT_UNLOCK(route);
+               } else {
+                       RT_UNLOCK(route);
+                       rtfree(route);
+               }
+       }
+       return (result);
 }
 
 errno_t
@@ -613,49 +928,109 @@ arp_ip_handle_input(
        const struct sockaddr_in *target_ip)
 {
        char    ipv4str[MAX_IPv4_STR_LEN];
-       struct sockaddr_dl *gateway;
+       struct sockaddr_dl proxied;
+       struct sockaddr_dl *gateway, *target_hw = NULL;
+       struct ifaddr *ifa;
        struct in_ifaddr *ia;
        struct in_ifaddr *best_ia = NULL;
        route_t route = NULL;
        char buf[3 * MAX_HW_LEN]; // enough for MAX_HW_LEN byte hw address
        struct llinfo_arp *llinfo;
-       struct timeval timenow;
        errno_t error;
+       int created_announcement = 0;
+       int bridged = 0, is_bridge = 0;
        
        /* Do not respond to requests for 0.0.0.0 */
-       if (target_ip->sin_addr.s_addr == 0 && arpop == ARPOP_REQUEST) {
-               return 0;
-       }
+       if (target_ip->sin_addr.s_addr == 0 && arpop == ARPOP_REQUEST)
+               goto done;
        
+       if (ifp->if_bridge)
+               bridged = 1;
+       if (ifp->if_type == IFT_BRIDGE)
+               is_bridge = 1;
+
        /*
         * Determine if this ARP is for us
+        * For a bridge, we want to check the address irrespective 
+        * of the receive interface.
+        */
+       lck_rw_lock_shared(in_ifaddr_rwlock);
+       TAILQ_FOREACH(ia, INADDR_HASH(target_ip->sin_addr.s_addr), ia_hash) {
+               if (((bridged && ia->ia_ifp->if_bridge != NULL) ||
+                       (ia->ia_ifp == ifp)) &&
+                   ia->ia_addr.sin_addr.s_addr == target_ip->sin_addr.s_addr) {
+                               best_ia = ia;
+                               ifaref(&best_ia->ia_ifa);
+                               lck_rw_done(in_ifaddr_rwlock);
+                               goto match;
+               }
+       }
+
+       TAILQ_FOREACH(ia, INADDR_HASH(sender_ip->sin_addr.s_addr), ia_hash) {
+               if (((bridged && ia->ia_ifp->if_bridge != NULL) ||
+                       (ia->ia_ifp == ifp)) &&
+                   ia->ia_addr.sin_addr.s_addr == sender_ip->sin_addr.s_addr) {
+                               best_ia = ia;
+                               ifaref(&best_ia->ia_ifa);
+                               lck_rw_done(in_ifaddr_rwlock);
+                               goto match;
+               }
+       }
+
+#define BDG_MEMBER_MATCHES_ARP(addr, ifp, ia)                                                          \
+       (ia->ia_ifp->if_bridge == ifp->if_softc &&                                                              \
+       !bcmp(ifnet_lladdr(ia->ia_ifp), ifnet_lladdr(ifp), ifp->if_addrlen) &&  \
+       addr == ia->ia_addr.sin_addr.s_addr)
+       /*
+        * Check the case when bridge shares its MAC address with
+        * some of its children, so packets are claimed by bridge
+        * itself (bridge_input() does it first), but they are really
+        * meant to be destined to the bridge member.
         */
-       lck_mtx_lock(rt_mtx);
-       for (ia = in_ifaddrhead.tqh_first; ia; ia = ia->ia_link.tqe_next) {
-               /* do_bridge should be tested here for bridging */
-               if (ia->ia_ifp == ifp) {
-                       best_ia = ia;
-                       if (target_ip->sin_addr.s_addr == ia->ia_addr.sin_addr.s_addr ||
-                               sender_ip->sin_addr.s_addr == ia->ia_addr.sin_addr.s_addr) {
-                               break;
+       if (is_bridge) {
+               TAILQ_FOREACH(ia, INADDR_HASH(target_ip->sin_addr.s_addr), ia_hash) {
+                       if (BDG_MEMBER_MATCHES_ARP(target_ip->sin_addr.s_addr, ifp, ia)) {
+                               ifp = ia->ia_ifp;
+                               best_ia = ia;
+                               ifaref(&best_ia->ia_ifa);
+                               lck_rw_done(in_ifaddr_rwlock);
+                               goto match;
                        }
                }
        }
-       
-       /* If we don't have an IP address on this interface, ignore the packet */
-       if (best_ia == 0) {
-               lck_mtx_unlock(rt_mtx);
-               return 0;
+       lck_rw_done(in_ifaddr_rwlock);
+
+       /*
+        * No match, use the first inet address on the receive interface
+        * as a dummy address for the rest of the function; we may be
+        * proxying for another address.
+        */
+       ifnet_lock_shared(ifp);
+       TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
+               if (ifa->ifa_addr->sa_family != AF_INET)
+                       continue;
+               best_ia = (struct in_ifaddr *)ifa;
+               ifaref(&best_ia->ia_ifa);
+               ifnet_lock_done(ifp);
+               goto match;
        }
-       
+       ifnet_lock_done(ifp);
+
+       /*
+        * If we're not a bridge member, or if we are but there's no
+        * IPv4 address to use for the interface, drop the packet.
+        */
+       if (!bridged || best_ia == NULL)
+               goto done;
+
+match:
        /* If the packet is from this interface, ignore the packet */
        if (!bcmp(CONST_LLADDR(sender_hw), ifnet_lladdr(ifp), sender_hw->sdl_len)) {
-               lck_mtx_unlock(rt_mtx);
-               return 0;
+               goto done;
        }
-       
+
        /* Check for a conflict */
-       if (sender_ip->sin_addr.s_addr == best_ia->ia_addr.sin_addr.s_addr) {
+       if (!bridged && sender_ip->sin_addr.s_addr == best_ia->ia_addr.sin_addr.s_addr) {
                struct kev_msg        ev_msg;
                struct kev_in_collision *in_collision;
                u_char  storage[sizeof(struct kev_in_collision) + MAX_HW_LEN];
@@ -664,7 +1039,7 @@ arp_ip_handle_input(
                        ifp->if_name, ifp->if_unit,
                        inet_ntop(AF_INET, &sender_ip->sin_addr, ipv4str, sizeof(ipv4str)),
                        sdl_addr_to_hex(sender_hw, buf, sizeof(buf)));
-               
+
                /* Send a kernel event so anyone can learn of the conflict */
                in_collision->link_data.if_family = ifp->if_family;
                in_collision->link_data.if_unit = ifp->if_unit;
@@ -680,49 +1055,80 @@ arp_ip_handle_input(
                ev_msg.dv[0].data_length = sizeof(struct kev_in_collision) + in_collision->hw_len;
                ev_msg.dv[1].data_length = 0;
                kev_post_msg(&ev_msg);
-               
+
                goto respond;
        }
-       
+
        /*
         * Look up the routing entry. If it doesn't exist and we are the
-        * target, go ahead and create one.
+        * target, and the sender isn't 0.0.0.0, go ahead and create one.
+        * Callee holds a reference on the route and returns with the route
+        * entry locked, upon success.
         */
-       error = arp_lookup_route(&sender_ip->sin_addr, (target_ip->sin_addr.s_addr ==
-                               best_ia->ia_addr.sin_addr.s_addr), 0, &route);
-       
+       error = arp_lookup_route(&sender_ip->sin_addr,
+           (target_ip->sin_addr.s_addr == best_ia->ia_addr.sin_addr.s_addr &&
+           sender_ip->sin_addr.s_addr != 0), 0, &route, ifp->if_index);
+
+       if (error == 0)
+               RT_LOCK_ASSERT_HELD(route);
+
        if (error || route == 0 || route->rt_gateway == 0) {
-               if (ipv4_ll_arp_aware != 0 && IN_LINKLOCAL(target_ip->sin_addr.s_addr)
-                       && arpop == ARPOP_REQUEST && sender_ip->sin_addr.s_addr == 0) {
+               if (arpop != ARPOP_REQUEST) {
+                       goto respond;
+               }
+               if (arp_sendllconflict
+                   && send_conflicting_probes != 0
+                   && (ifp->if_eflags & IFEF_ARPLL) != 0 
+                   && IN_LINKLOCAL(ntohl(target_ip->sin_addr.s_addr))
+                   && sender_ip->sin_addr.s_addr == 0) {
                        /*
                         * Verify this ARP probe doesn't conflict with an IPv4LL we know of
                         * on another interface.
                         */
-                       error = arp_lookup_route(&target_ip->sin_addr, 0, 0, &route);
+                       if (route != NULL) {
+                               RT_REMREF_LOCKED(route);
+                               RT_UNLOCK(route);
+                               route = NULL;
+                       }
+                       /*
+                        * Callee holds a reference on the route and returns
+                        * with the route entry locked, upon success.
+                        */
+                       error = arp_lookup_route(&target_ip->sin_addr, 0, 0,
+                           &route, ifp->if_index);
+
+                       if (error == 0)
+                               RT_LOCK_ASSERT_HELD(route);
+
                        if (error == 0 && route && route->rt_gateway) {
                                gateway = SDL(route->rt_gateway);
-                               if (route->rt_ifp != ifp &&
-                                       (gateway->sdl_alen != sender_hw->sdl_alen ||
-                                        bcmp(CONST_LLADDR(gateway), CONST_LLADDR(sender_hw),
-                                        gateway->sdl_alen) != 0)) {
+                               if (route->rt_ifp != ifp && gateway->sdl_alen != 0 
+                                   && (gateway->sdl_alen != sender_hw->sdl_alen 
+                                       || bcmp(CONST_LLADDR(gateway), CONST_LLADDR(sender_hw),
+                                               gateway->sdl_alen) != 0)) {
                                        /*
                                         * A node is probing for an IPv4LL we know exists on a
                                         * different interface. We respond with a conflicting probe
                                         * to force the new device to pick a different IPv4LL
                                         * address.
                                         */
-                                       log(LOG_INFO,
+                                       if (log_arp_warnings) {
+                                           log(LOG_INFO,
                                                "arp: %s on %s%d sent probe for %s, already on %s%d\n",
                                                sdl_addr_to_hex(sender_hw, buf, sizeof(buf)),
                                                ifp->if_name, ifp->if_unit,
                                                inet_ntop(AF_INET, &target_ip->sin_addr, ipv4str,
                                                                  sizeof(ipv4str)),
                                                route->rt_ifp->if_name, route->rt_ifp->if_unit);
-                                       log(LOG_INFO,
+                                           log(LOG_INFO,
                                                "arp: sending conflicting probe to %s on %s%d\n",
                                                sdl_addr_to_hex(sender_hw, buf, sizeof(buf)),
                                                ifp->if_name, ifp->if_unit);
-                                       
+                                       }
+                                       /* We're done with the route */
+                                       RT_REMREF_LOCKED(route);
+                                       RT_UNLOCK(route);
+                                       route = NULL;
                                        /*
                                         * Send a conservative unicast "ARP probe".
                                         * This should force the other device to pick a new number.
@@ -730,20 +1136,59 @@ arp_ip_handle_input(
                                         * has already assigned that number.
                                         * This will not imply to the device that we own that address.
                                         */
+                                       ifnet_lock_shared(ifp);
+                                       ifa = TAILQ_FIRST(&ifp->if_addrhead);
+                                       if (ifa != NULL)
+                                               ifaref(ifa);
+                                       ifnet_lock_done(ifp);
                                        dlil_send_arp_internal(ifp, ARPOP_REQUEST,
-                                               (struct sockaddr_dl*)TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr,
+                                               ifa != NULL ? SDL(ifa->ifa_addr) : NULL,
                                                (const struct sockaddr*)sender_ip, sender_hw,
                                                (const struct sockaddr*)target_ip);
+                                       if (ifa != NULL) {
+                                               ifafree(ifa);
+                                               ifa = NULL;
+                                       }
                                }
                        }
+                       goto respond;
+               } else if (keep_announcements != 0
+                          && target_ip->sin_addr.s_addr == sender_ip->sin_addr.s_addr) {
+                       /* don't create entry if link-local address and link-local is disabled */
+                       if (!IN_LINKLOCAL(ntohl(sender_ip->sin_addr.s_addr)) 
+                           || (ifp->if_eflags & IFEF_ARPLL) != 0) {
+                               if (route != NULL) {
+                                       RT_REMREF_LOCKED(route);
+                                       RT_UNLOCK(route);
+                                       route = NULL;
+                               }
+                               /*
+                                * Callee holds a reference on the route and
+                                * returns with the route entry locked, upon
+                                * success.
+                                */
+                               error = arp_lookup_route(&sender_ip->sin_addr,
+                                   1, 0, &route, ifp->if_index);
+
+                               if (error == 0)
+                                       RT_LOCK_ASSERT_HELD(route);
+
+                               if (error == 0 && route != NULL && route->rt_gateway != NULL) {
+                                       created_announcement = 1;
+                               }
+                       }
+                       if (created_announcement == 0) {
+                               goto respond;
+                       }
+               } else {
+                       goto respond;
                }
-               
-               goto respond;
        }
-       
+
+       RT_LOCK_ASSERT_HELD(route);
        gateway = SDL(route->rt_gateway);
-       if (route->rt_ifp != ifp) {
-               if (!IN_LINKLOCAL(sender_ip->sin_addr.s_addr) || (ifp->if_eflags & IFEF_ARPLL) == 0) {
+       if (!bridged && route->rt_ifp != ifp) {
+               if (!IN_LINKLOCAL(ntohl(sender_ip->sin_addr.s_addr)) || (ifp->if_eflags & IFEF_ARPLL) == 0) {
                        if (log_arp_warnings)
                                log(LOG_ERR, "arp: %s is on %s%d but got reply from %s on %s%d\n",
                                        inet_ntop(AF_INET, &sender_ip->sin_addr, ipv4str,
@@ -759,110 +1204,177 @@ arp_ip_handle_input(
                        if (route->rt_rmx.rmx_expire == 0) {
                                goto respond;
                        }
-                       
+
                        /*
-                        * Don't change the cloned route away from the parent's interface
-                        * if the address did resolve.
+                        * We're about to check and/or change the route's ifp
+                        * and ifa, so do the lock dance: drop rt_lock, hold
+                        * rnh_lock and re-hold rt_lock to avoid violating the
+                        * lock ordering.  We have an extra reference on the
+                        * route, so it won't go away while we do this.
                         */
-                       if (gateway->sdl_alen != 0 && route->rt_parent &&
-                               route->rt_parent->rt_ifp == route->rt_ifp) {
+                       RT_UNLOCK(route);
+                       lck_mtx_lock(rnh_lock);
+                       RT_LOCK(route);
+                       /*
+                        * Don't change the cloned route away from the
+                        * parent's interface if the address did resolve
+                        * or if the route is defunct.  rt_ifp on both
+                        * the parent and the clone can now be freely
+                        * accessed now that we have acquired rnh_lock.
+                        */
+                       gateway = SDL(route->rt_gateway);
+                       if ((gateway->sdl_alen != 0 && route->rt_parent &&
+                           route->rt_parent->rt_ifp == route->rt_ifp) ||
+                           (route->rt_flags & RTF_CONDEMNED)) {
+                               RT_REMREF_LOCKED(route);
+                               RT_UNLOCK(route);
+                               route = NULL;
+                               lck_mtx_unlock(rnh_lock);
                                goto respond;
                        }
-                       
                        /* Change the interface when the existing route is on */
                        route->rt_ifp = ifp;
                        rtsetifa(route, &best_ia->ia_ifa);
                        gateway->sdl_index = ifp->if_index;
+                       RT_UNLOCK(route);
+                       lck_mtx_unlock(rnh_lock);
+                       RT_LOCK(route);
+                       /* Don't bother if the route is down */
+                       if (!(route->rt_flags & RTF_UP))
+                               goto respond;
+                       /* Refresh gateway pointer */
+                       gateway = SDL(route->rt_gateway);
                }
+               RT_LOCK_ASSERT_HELD(route);
        }
-       
+
        if (gateway->sdl_alen && bcmp(LLADDR(gateway), CONST_LLADDR(sender_hw), gateway->sdl_alen)) {
-               if (route->rt_rmx.rmx_expire) {
+               if (route->rt_rmx.rmx_expire && log_arp_warnings) {
                        char buf2[3 * MAX_HW_LEN];
                        log(LOG_INFO, "arp: %s moved from %s to %s on %s%d\n",
-                               inet_ntop(AF_INET, &sender_ip->sin_addr, ipv4str,
-                                                 sizeof(ipv4str)),
-                               sdl_addr_to_hex(gateway, buf, sizeof(buf)),
-                               sdl_addr_to_hex(sender_hw, buf2, sizeof(buf2)), ifp->if_name,
-                               ifp->if_unit);
+                           inet_ntop(AF_INET, &sender_ip->sin_addr, ipv4str,
+                           sizeof(ipv4str)),
+                           sdl_addr_to_hex(gateway, buf, sizeof(buf)),
+                           sdl_addr_to_hex(sender_hw, buf2, sizeof(buf2)),
+                           ifp->if_name, ifp->if_unit);
                }
-               else {
-                       log(LOG_ERR,
-                               "arp: %s attempts to modify permanent entry for %s on %s%d\n",
-                               sdl_addr_to_hex(sender_hw, buf, sizeof(buf)),
-                               inet_ntop(AF_INET, &sender_ip->sin_addr, ipv4str,
-                                                 sizeof(ipv4str)),
-                               ifp->if_name, ifp->if_unit);
+               else if (route->rt_rmx.rmx_expire == 0) {
+                       if (log_arp_warnings) {
+                               log(LOG_ERR, "arp: %s attempts to modify "
+                                   "permanent entry for %s on %s%d\n",
+                                   sdl_addr_to_hex(sender_hw, buf,
+                                   sizeof(buf)),
+                                   inet_ntop(AF_INET, &sender_ip->sin_addr,
+                                   ipv4str, sizeof(ipv4str)),
+                                   ifp->if_name, ifp->if_unit);
+                       }
                        goto respond;
                }
        }
-       
+
        /* Copy the sender hardware address in to the route's gateway address */
        gateway->sdl_alen = sender_hw->sdl_alen;
        bcopy(CONST_LLADDR(sender_hw), LLADDR(gateway), gateway->sdl_alen);
-       
+
        /* Update the expire time for the route and clear the reject flag */
-       getmicrotime(&timenow);
-       if (route->rt_rmx.rmx_expire)
+       if (route->rt_rmx.rmx_expire) {
+               struct timeval timenow;
+
+               getmicrotime(&timenow);
                route->rt_rmx.rmx_expire = timenow.tv_sec + arpt_keep;
+       }
        route->rt_flags &= ~RTF_REJECT;
-       
+
        /* update the llinfo, send a queued packet if there is one */
-       llinfo = (struct llinfo_arp*)route->rt_llinfo;
+       llinfo = route->rt_llinfo;
        llinfo->la_asked = 0;
        if (llinfo->la_hold) {
                struct mbuf *m0;
                m0 = llinfo->la_hold;
                llinfo->la_hold = 0;
-               
-               /* Should we a reference on the route first? */
-               lck_mtx_unlock(rt_mtx);
+
+               RT_UNLOCK(route);
                dlil_output(ifp, PF_INET, m0, (caddr_t)route, rt_key(route), 0);
-               lck_mtx_lock(rt_mtx);
+               RT_REMREF(route);
+               route = NULL;
        }
-       
+
 respond:
-       if (arpop != ARPOP_REQUEST) {
-               lck_mtx_unlock(rt_mtx);
-               return 0;
+       if (route != NULL) {
+               RT_REMREF_LOCKED(route);
+               RT_UNLOCK(route);
+               route = NULL;
        }
-       
+
+       if (arpop != ARPOP_REQUEST)
+               goto done;
+
        /* If we are not the target, check if we should proxy */
        if (target_ip->sin_addr.s_addr != best_ia->ia_addr.sin_addr.s_addr) {
-       
-               /* Find a proxy route */
-               error = arp_lookup_route(&target_ip->sin_addr, 0, SIN_PROXY, &route);
-               if (error || route == NULL) {
-                       
-                       /* We don't have a route entry indicating we should use proxy */
-                       /* If we aren't supposed to proxy all, we are done */
-                       if (!arp_proxyall) {
-                               lck_mtx_unlock(rt_mtx);
-                               return 0;
-                       }
-                       
-                       /* See if we have a route to the target ip before we proxy it */
-                       route = rtalloc1_locked((const struct sockaddr*)target_ip, 0, 0);
-                       if (!route) {
-                               lck_mtx_unlock(rt_mtx);
-                               return 0;
-                       }
-                       
+               /*
+                * Find a proxy route; callee holds a reference on the
+                * route and returns with the route entry locked, upon
+                * success.
+                */
+               error = arp_lookup_route(&target_ip->sin_addr, 0, SIN_PROXY,
+                   &route, ifp->if_index);
+
+               if (error == 0) {
+                       RT_LOCK_ASSERT_HELD(route);
+                       /*
+                        * Return proxied ARP replies only on the interface
+                        * or bridge cluster where this network resides.
+                        * Otherwise we may conflict with the host we are
+                        * proxying for.
+                        */
+                       if (route->rt_ifp != ifp &&
+                               (route->rt_ifp->if_bridge != ifp->if_bridge ||
+                                ifp->if_bridge == NULL)) {
+                                       RT_REMREF_LOCKED(route);
+                                       RT_UNLOCK(route);
+                                       goto done;
+                               }
+                       proxied = *SDL(route->rt_gateway);
+                       target_hw = &proxied;
+               } else {
+                       /*
+                        * We don't have a route entry indicating we should
+                        * use proxy.  If we aren't supposed to proxy all,
+                        * we are done.
+                        */
+                       if (!arp_proxyall)
+                               goto done;
+
+                       /*
+                        * See if we have a route to the target ip before
+                        * we proxy it.
+                        */
+                       route = rtalloc1_scoped((struct sockaddr *)
+                           (size_t)target_ip, 0, 0, ifp->if_index);
+                       if (!route)
+                               goto done;
+
                        /*
                         * Don't proxy for hosts already on the same interface.
                         */
+                       RT_LOCK(route);
                        if (route->rt_ifp == ifp) {
-                               rtfree_locked(route);
-                               lck_mtx_unlock(rt_mtx);
-                               return 0;
+                               RT_UNLOCK(route);
+                               rtfree(route);
+                               goto done;
                        }
                }
+               RT_REMREF_LOCKED(route);
+               RT_UNLOCK(route);
        }
-       lck_mtx_unlock(rt_mtx);
-       
-       dlil_send_arp(ifp, ARPOP_REPLY, NULL, (const struct sockaddr*)target_ip,
-               sender_hw, (const struct sockaddr*)sender_ip);
-       
+
+       dlil_send_arp(ifp, ARPOP_REPLY,
+           target_hw, (const struct sockaddr*)target_ip,
+           sender_hw, (const struct sockaddr*)sender_ip);
+
+done:
+       if (best_ia != NULL)
+               ifafree(&best_ia->ia_ifa);
        return 0;
 }