Merge multi-FIB IPv6 support from projects/multi-fibv6/head/:

Extend the so far IPv4-only support for multiple routing tables (FIBs)
introduced in r178888 to IPv6 providing feature parity.

This includes an extended rtalloc(9) KPI for IPv6, the necessary
adjustments to the network stack, and user land support as in netstat.

Sponsored by:	Cisco Systems, Inc.
Reviewed by:	melifaro (basically)
MFC after:	10 days

1 files changed, 278 insertions, 271 deletions

diff --git a/sys/netinet6/in6.c b/sys/netinet6/in6.c
index cf5eb35..0b22a1c 100644
--- a/sys/netinet6/in6.c
+++ b/sys/netinet6/in6.c
@@ -180,6 +180,7 @@ in6_ifaddloop(struct ifaddr *ifa)
 	rt_mask(&rt) = (struct sockaddr *)&mask;
 	rt_key(&rt) = (struct sockaddr *)&addr;
 	rt.rt_flags = RTF_UP | RTF_HOST | RTF_STATIC;
+	/* Announce arrival of local address to all FIBs. */
 	rt_newaddrmsg(RTM_ADD, ifa, 0, &rt);
 }
 
@@ -214,6 +215,7 @@ in6_ifremloop(struct ifaddr *ifa)
 	rt_mask(&rt0) = (struct sockaddr *)&mask;
 	rt_key(&rt0) = (struct sockaddr *)&addr;
 	rt0.rt_flags = RTF_HOST | RTF_STATIC;
+	/* Announce removal of local address to all FIBs. */
 	rt_newaddrmsg(RTM_DELETE, ifa, 0, &rt0);
 }
 
@@ -282,6 +284,11 @@ in6_control(struct socket *so, u_long cmd, caddr_t data,
 	switch (cmd) {
 	case SIOCGETSGCNT_IN6:
 	case SIOCGETMIFCNT_IN6:
+		/*	
+		 * XXX mrt_ioctl has a 3rd, unused, FIB argument in route.c.
+		 * We cannot see how that would be needed, so do not adjust the
+		 * KPI blindly; more likely should clean up the IPv4 variant.
+		 */
 		return (mrt6_ioctl ? mrt6_ioctl(cmd, data) : EOPNOTSUPP);
 	}
 
@@ -820,6 +827,170 @@ out:
 	return (error);
 }
 
+
+/*
+ * Join necessary multicast groups.  Factored out from in6_update_ifa().
+ * This entire work should only be done once, for the default FIB.
+ */
+static int
+in6_update_ifa_join_mc(struct ifnet *ifp, struct in6_aliasreq *ifra,
+    struct in6_ifaddr *ia, int flags, struct in6_multi **in6m_sol)
+{
+	char ip6buf[INET6_ADDRSTRLEN];
+	struct sockaddr_in6 mltaddr, mltmask;
+	struct in6_addr llsol;
+	struct in6_multi_mship *imm;
+	struct rtentry *rt;
+	int delay, error;
+
+	KASSERT(in6m_sol != NULL, ("%s: in6m_sol is NULL", __func__));
+
+	/* Join solicited multicast addr for new host id. */
+	bzero(&llsol, sizeof(struct in6_addr));
+	llsol.s6_addr32[0] = IPV6_ADDR_INT32_MLL;
+	llsol.s6_addr32[1] = 0;
+	llsol.s6_addr32[2] = htonl(1);
+	llsol.s6_addr32[3] = ifra->ifra_addr.sin6_addr.s6_addr32[3];
+	llsol.s6_addr8[12] = 0xff;
+	if ((error = in6_setscope(&llsol, ifp, NULL)) != 0) {
+		/* XXX: should not happen */
+		log(LOG_ERR, "%s: in6_setscope failed\n", __func__);
+		goto cleanup;
+	}
+	delay = 0;
+	if ((flags & IN6_IFAUPDATE_DADDELAY)) {
+		/*
+		 * We need a random delay for DAD on the address being
+		 * configured.  It also means delaying transmission of the
+		 * corresponding MLD report to avoid report collision.
+		 * [RFC 4861, Section 6.3.7]
+		 */
+		delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
+	}
+	imm = in6_joingroup(ifp, &llsol, &error, delay);
+	if (imm == NULL) {
+		nd6log((LOG_WARNING, "%s: addmulti failed for %s on %s "
+		    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf, &llsol),
+		    if_name(ifp), error));
+		goto cleanup;
+	}
+	LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
+	*in6m_sol = imm->i6mm_maddr;
+
+	bzero(&mltmask, sizeof(mltmask));
+	mltmask.sin6_len = sizeof(struct sockaddr_in6);
+	mltmask.sin6_family = AF_INET6;
+	mltmask.sin6_addr = in6mask32;
+#define	MLTMASK_LEN  4	/* mltmask's masklen (=32bit=4octet) */
+
+	/*
+	 * Join link-local all-nodes address.
+	 */
+	bzero(&mltaddr, sizeof(mltaddr));
+	mltaddr.sin6_len = sizeof(struct sockaddr_in6);
+	mltaddr.sin6_family = AF_INET6;
+	mltaddr.sin6_addr = in6addr_linklocal_allnodes;
+	if ((error = in6_setscope(&mltaddr.sin6_addr, ifp, NULL)) != 0)
+		goto cleanup; /* XXX: should not fail */
+
+	/*
+	 * XXX: do we really need this automatic routes?  We should probably
+	 * reconsider this stuff.  Most applications actually do not need the
+	 * routes, since they usually specify the outgoing interface.
+	 */
+	rt = in6_rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL, RT_DEFAULT_FIB);
+	if (rt != NULL) {
+		/* XXX: only works in !SCOPEDROUTING case. */
+		if (memcmp(&mltaddr.sin6_addr,
+		    &((struct sockaddr_in6 *)rt_key(rt))->sin6_addr,
+		    MLTMASK_LEN)) {
+			RTFREE_LOCKED(rt);
+			rt = NULL;
+		}
+	}
+	if (rt == NULL) {
+		error = in6_rtrequest(RTM_ADD, (struct sockaddr *)&mltaddr,
+		    (struct sockaddr *)&ia->ia_addr,
+		    (struct sockaddr *)&mltmask, RTF_UP,
+		    (struct rtentry **)0, RT_DEFAULT_FIB);
+		if (error)
+			goto cleanup;
+	} else
+		RTFREE_LOCKED(rt);
+
+	imm = in6_joingroup(ifp, &mltaddr.sin6_addr, &error, 0);
+	if (imm == NULL) {
+		nd6log((LOG_WARNING, "%s: addmulti failed for %s on %s "
+		    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
+		    &mltaddr.sin6_addr), if_name(ifp), error));
+		goto cleanup;
+	}
+	LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
+
+	/*
+	 * Join node information group address.
+	 */
+	delay = 0;
+	if ((flags & IN6_IFAUPDATE_DADDELAY)) {
+		/*
+		 * The spec does not say anything about delay for this group,
+		 * but the same logic should apply.
+		 */
+		delay = arc4random() % (MAX_RTR_SOLICITATION_DELAY * hz);
+	}
+	if (in6_nigroup(ifp, NULL, -1, &mltaddr.sin6_addr) == 0) {
+		/* XXX jinmei */
+		imm = in6_joingroup(ifp, &mltaddr.sin6_addr, &error, delay);
+		if (imm == NULL)
+			nd6log((LOG_WARNING, "%s: addmulti failed for %s on %s "
+			    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
+			    &mltaddr.sin6_addr), if_name(ifp), error));
+			/* XXX not very fatal, go on... */
+		else
+			LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
+	}
+
+	/*
+	 * Join interface-local all-nodes address.
+	 * (ff01::1%ifN, and ff01::%ifN/32)
+	 */
+	mltaddr.sin6_addr = in6addr_nodelocal_allnodes;
+	if ((error = in6_setscope(&mltaddr.sin6_addr, ifp, NULL)) != 0)
+		goto cleanup; /* XXX: should not fail */
+	/* XXX: again, do we really need the route? */
+	rt = in6_rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL, RT_DEFAULT_FIB);
+	if (rt != NULL) {
+		if (memcmp(&mltaddr.sin6_addr,
+		    &((struct sockaddr_in6 *)rt_key(rt))->sin6_addr,
+		    MLTMASK_LEN)) {
+			RTFREE_LOCKED(rt);
+			rt = NULL;
+		}
+	}
+	if (rt == NULL) {
+		error = in6_rtrequest(RTM_ADD, (struct sockaddr *)&mltaddr,
+		    (struct sockaddr *)&ia->ia_addr,
+		    (struct sockaddr *)&mltmask, RTF_UP,
+		    (struct rtentry **)0, RT_DEFAULT_FIB);
+		if (error)
+			goto cleanup;
+	} else
+		RTFREE_LOCKED(rt);
+
+	imm = in6_joingroup(ifp, &mltaddr.sin6_addr, &error, 0);
+	if (imm == NULL) {
+		nd6log((LOG_WARNING, "%s: addmulti failed for %s on %s "
+		    "(errno=%d)\n", __func__, ip6_sprintf(ip6buf,
+		    &mltaddr.sin6_addr), if_name(ifp), error));
+		goto cleanup;
+	}
+	LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
+#undef	MLTMASK_LEN
+
+cleanup:
+	return (error);
+}
+
 /*
  * Update parameters of an IPv6 interface address.
  * If necessary, a new entry is created and linked into address chains.
@@ -833,9 +1004,7 @@ in6_update_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
 	int error = 0, hostIsNew = 0, plen = -1;
 	struct sockaddr_in6 dst6;
 	struct in6_addrlifetime *lt;
-	struct in6_multi_mship *imm;
 	struct in6_multi *in6m_sol;
-	struct rtentry *rt;
 	int delay;
 	char ip6buf[INET6_ADDRSTRLEN];
 
@@ -1083,178 +1252,17 @@ in6_update_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
 	 * not just go to unlink.
 	 */
 
-	/* Join necessary multicast groups */
+	/* Join necessary multicast groups. */
 	in6m_sol = NULL;
 	if ((ifp->if_flags & IFF_MULTICAST) != 0) {
-		struct sockaddr_in6 mltaddr, mltmask;
-		struct in6_addr llsol;
-
-		/* join solicited multicast addr for new host id */
-		bzero(&llsol, sizeof(struct in6_addr));
-		llsol.s6_addr32[0] = IPV6_ADDR_INT32_MLL;
-		llsol.s6_addr32[1] = 0;
-		llsol.s6_addr32[2] = htonl(1);
-		llsol.s6_addr32[3] = ifra->ifra_addr.sin6_addr.s6_addr32[3];
-		llsol.s6_addr8[12] = 0xff;
-		if ((error = in6_setscope(&llsol, ifp, NULL)) != 0) {
-			/* XXX: should not happen */
-			log(LOG_ERR, "in6_update_ifa: "
-			    "in6_setscope failed\n");
-			goto cleanup;
-		}
-		delay = 0;
-		if ((flags & IN6_IFAUPDATE_DADDELAY)) {
-			/*
-			 * We need a random delay for DAD on the address
-			 * being configured.  It also means delaying
-			 * transmission of the corresponding MLD report to
-			 * avoid report collision.
-			 * [RFC 4861, Section 6.3.7]
-			 */
-			delay = arc4random() %
-			    (MAX_RTR_SOLICITATION_DELAY * hz);
-		}
-		imm = in6_joingroup(ifp, &llsol, &error, delay);
-		if (imm == NULL) {
-			nd6log((LOG_WARNING,
-			    "in6_update_ifa: addmulti failed for "
-			    "%s on %s (errno=%d)\n",
-			    ip6_sprintf(ip6buf, &llsol), if_name(ifp),
-			    error));
+		error = in6_update_ifa_join_mc(ifp, ifra, ia, flags, &in6m_sol);
+		if (error)
 			goto cleanup;
-		}
-		LIST_INSERT_HEAD(&ia->ia6_memberships,
-		    imm, i6mm_chain);
-		in6m_sol = imm->i6mm_maddr;
-
-		bzero(&mltmask, sizeof(mltmask));
-		mltmask.sin6_len = sizeof(struct sockaddr_in6);
-		mltmask.sin6_family = AF_INET6;
-		mltmask.sin6_addr = in6mask32;
-#define	MLTMASK_LEN  4	/* mltmask's masklen (=32bit=4octet) */
-
-		/*
-		 * join link-local all-nodes address
-		 */
-		bzero(&mltaddr, sizeof(mltaddr));
-		mltaddr.sin6_len = sizeof(struct sockaddr_in6);
-		mltaddr.sin6_family = AF_INET6;
-		mltaddr.sin6_addr = in6addr_linklocal_allnodes;
-		if ((error = in6_setscope(&mltaddr.sin6_addr, ifp, NULL)) !=
-		    0)
-			goto cleanup; /* XXX: should not fail */
-
-		/*
-		 * XXX: do we really need this automatic routes?
-		 * We should probably reconsider this stuff.  Most applications
-		 * actually do not need the routes, since they usually specify
-		 * the outgoing interface.
-		 */
-		rt = rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL);
-		if (rt) {
-			/* XXX: only works in !SCOPEDROUTING case. */
-			if (memcmp(&mltaddr.sin6_addr,
-			    &((struct sockaddr_in6 *)rt_key(rt))->sin6_addr,
-			    MLTMASK_LEN)) {
-				RTFREE_LOCKED(rt);
-				rt = NULL;
-			}
-		}
-		if (!rt) {
-			error = rtrequest(RTM_ADD, (struct sockaddr *)&mltaddr,
-			    (struct sockaddr *)&ia->ia_addr,
-			    (struct sockaddr *)&mltmask, RTF_UP,
-			    (struct rtentry **)0);
-			if (error)
-				goto cleanup;
-		} else {
-			RTFREE_LOCKED(rt);
-		}
-
-		imm = in6_joingroup(ifp, &mltaddr.sin6_addr, &error, 0);
-		if (!imm) {
-			nd6log((LOG_WARNING,
-			    "in6_update_ifa: addmulti failed for "
-			    "%s on %s (errno=%d)\n",
-			    ip6_sprintf(ip6buf, &mltaddr.sin6_addr),
-			    if_name(ifp), error));
-			goto cleanup;
-		}
-		LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
-
-		/*
-		 * join node information group address
-		 */
-		delay = 0;
-		if ((flags & IN6_IFAUPDATE_DADDELAY)) {
-			/*
-			 * The spec doesn't say anything about delay for this
-			 * group, but the same logic should apply.
-			 */
-			delay = arc4random() %
-			    (MAX_RTR_SOLICITATION_DELAY * hz);
-		}
-		if (in6_nigroup(ifp, NULL, -1, &mltaddr.sin6_addr) == 0) {
-			imm = in6_joingroup(ifp, &mltaddr.sin6_addr, &error,
-			    delay); /* XXX jinmei */
-			if (!imm) {
-				nd6log((LOG_WARNING, "in6_update_ifa: "
-				    "addmulti failed for %s on %s "
-				    "(errno=%d)\n",
-				    ip6_sprintf(ip6buf, &mltaddr.sin6_addr),
-				    if_name(ifp), error));
-				/* XXX not very fatal, go on... */
-			} else {
-				LIST_INSERT_HEAD(&ia->ia6_memberships,
-				    imm, i6mm_chain);
-			}
-		}
-
-		/*
-		 * join interface-local all-nodes address.
-		 * (ff01::1%ifN, and ff01::%ifN/32)
-		 */
-		mltaddr.sin6_addr = in6addr_nodelocal_allnodes;
-		if ((error = in6_setscope(&mltaddr.sin6_addr, ifp, NULL))
-		    != 0)
-			goto cleanup; /* XXX: should not fail */
-		/* XXX: again, do we really need the route? */
-		rt = rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL);
-		if (rt) {
-			if (memcmp(&mltaddr.sin6_addr,
-			    &((struct sockaddr_in6 *)rt_key(rt))->sin6_addr,
-			    MLTMASK_LEN)) {
-				RTFREE_LOCKED(rt);
-				rt = NULL;
-			}
-		}
-		if (!rt) {
-			error = rtrequest(RTM_ADD, (struct sockaddr *)&mltaddr,
-			    (struct sockaddr *)&ia->ia_addr,
-			    (struct sockaddr *)&mltmask, RTF_UP,
-			    (struct rtentry **)0);
-			if (error)
-				goto cleanup;
-		} else
-			RTFREE_LOCKED(rt);
-
-		imm = in6_joingroup(ifp, &mltaddr.sin6_addr, &error, 0);
-		if (!imm) {
-			nd6log((LOG_WARNING, "in6_update_ifa: "
-			    "addmulti failed for %s on %s "
-			    "(errno=%d)\n",
-			    ip6_sprintf(ip6buf, &mltaddr.sin6_addr),
-			    if_name(ifp), error));
-			goto cleanup;
-		}
-		LIST_INSERT_HEAD(&ia->ia6_memberships, imm, i6mm_chain);
-#undef	MLTMASK_LEN
 	}
 
 	/*
 	 * Perform DAD, if needed.
-	 * XXX It may be of use, if we can administratively
-	 * disable DAD.
+	 * XXX It may be of use, if we can administratively disable DAD.
 	 */
 	if (in6if_do_dad(ifp) && ((ifra->ifra_flags & IN6_IFF_NODAD) == 0) &&
 	    (ia->ia6_flags & IN6_IFF_TENTATIVE))
@@ -1312,58 +1320,20 @@ in6_update_ifa(struct ifnet *ifp, struct in6_aliasreq *ifra,
 	return error;
 }
 
-void
-in6_purgeaddr(struct ifaddr *ifa)
+/*
+ * Leave multicast groups.  Factored out from in6_purgeaddr().
+ * This entire work should only be done once, for the default FIB.
+ */
+static int
+in6_purgeaddr_mc(struct ifnet *ifp, struct in6_ifaddr *ia, struct ifaddr *ifa0)
 {
-	struct ifnet *ifp = ifa->ifa_ifp;
-	struct in6_ifaddr *ia = (struct in6_ifaddr *) ifa;
-	struct in6_multi_mship *imm;
 	struct sockaddr_in6 mltaddr, mltmask;
-	int plen, error;
+	struct in6_multi_mship *imm;
 	struct rtentry *rt;
-	struct ifaddr *ifa0;
-
-	if (ifa->ifa_carp)
-		(*carp_detach_p)(ifa);
-
-	/*
-	 * find another IPv6 address as the gateway for the
-	 * link-local and node-local all-nodes multicast
-	 * address routes
-	 */
-	IF_ADDR_RLOCK(ifp);
-	TAILQ_FOREACH(ifa0, &ifp->if_addrhead, ifa_link) {
-		if ((ifa0->ifa_addr->sa_family != AF_INET6) ||
-		    memcmp(&satosin6(ifa0->ifa_addr)->sin6_addr,
-			   &ia->ia_addr.sin6_addr, 
-			   sizeof(struct in6_addr)) == 0)
-			continue;
-		else
-			break;
-	}
-	if (ifa0 != NULL)
-		ifa_ref(ifa0);
-	IF_ADDR_RUNLOCK(ifp);
-
-	/*
-	 * Remove the loopback route to the interface address.
-	 * The check for the current setting of "nd6_useloopback" 
-	 * is not needed.
-	 */
-	if (ia->ia_flags & IFA_RTSELF) {
-		error = ifa_del_loopback_route((struct ifaddr *)ia,
-				       (struct sockaddr *)&ia->ia_addr);
-		if (error == 0)
-			ia->ia_flags &= ~IFA_RTSELF;
-	}
-
-	/* stop DAD processing */
-	nd6_dad_stop(ifa);
-
-	in6_ifremloop(ifa);
+	int error;
 
 	/*
-	 * leave from multicast groups we have joined for the interface
+	 * Leave from multicast groups we have joined for the interface.
 	 */
 	while ((imm = LIST_FIRST(&ia->ia6_memberships)) != NULL) {
 		LIST_REMOVE(imm, i6mm_chain);
@@ -1371,7 +1341,7 @@ in6_purgeaddr(struct ifaddr *ifa)
 	}
 
 	/*
-	 * remove the link-local all-nodes address
+	 * Remove the link-local all-nodes address.
 	 */
 	bzero(&mltmask, sizeof(mltmask));
 	mltmask.sin6_len = sizeof(struct sockaddr_in6);
@@ -1384,31 +1354,33 @@ in6_purgeaddr(struct ifaddr *ifa)
 	mltaddr.sin6_addr = in6addr_linklocal_allnodes;
 
 	if ((error = in6_setscope(&mltaddr.sin6_addr, ifp, NULL)) != 0)
-		goto cleanup; 
+		return (error);
 
-	rt = rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL);
+	rt = in6_rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL, RT_DEFAULT_FIB);
 	if (rt != NULL && rt->rt_gateway != NULL &&
 	    (memcmp(&satosin6(rt->rt_gateway)->sin6_addr, 
 		    &ia->ia_addr.sin6_addr,
 		    sizeof(ia->ia_addr.sin6_addr)) == 0)) {
 		/* 
-		 * if no more IPv6 address exists on this interface
-		 * then remove the multicast address route
+		 * If no more IPv6 address exists on this interface then
+		 * remove the multicast address route.
 		 */
 		if (ifa0 == NULL) {
 			memcpy(&mltaddr.sin6_addr, &satosin6(rt_key(rt))->sin6_addr, 
 			       sizeof(mltaddr.sin6_addr));
 			RTFREE_LOCKED(rt);
-			error = rtrequest(RTM_DELETE, (struct sockaddr *)&mltaddr,
-					  (struct sockaddr *)&ia->ia_addr,
-					  (struct sockaddr *)&mltmask, RTF_UP,
-					  (struct rtentry **)0);
+			error = in6_rtrequest(RTM_DELETE,
+			    (struct sockaddr *)&mltaddr,
+			    (struct sockaddr *)&ia->ia_addr,
+			    (struct sockaddr *)&mltmask, RTF_UP,
+			    (struct rtentry **)0, RT_DEFAULT_FIB);
 			if (error)
-				log(LOG_INFO, "in6_purgeaddr: link-local all-nodes"
-				    "multicast address deletion error\n");
+				log(LOG_INFO, "%s: link-local all-nodes "
+				    "multicast address deletion error\n",
+				    __func__);
 		} else {
 			/*
-			 * replace the gateway of the route
+			 * Replace the gateway of the route.
 			 */
 			struct sockaddr_in6 sa;
 
@@ -1427,38 +1399,38 @@ in6_purgeaddr(struct ifaddr *ifa)
 	}
 
 	/*
-	 * remove the node-local all-nodes address
+	 * Remove the node-local all-nodes address.
 	 */
 	mltaddr.sin6_addr = in6addr_nodelocal_allnodes;
-	if ((error = in6_setscope(&mltaddr.sin6_addr, ifp, NULL)) !=
-	    0)
-		goto cleanup;
+	if ((error = in6_setscope(&mltaddr.sin6_addr, ifp, NULL)) != 0)
+		return (error);
 
-	rt = rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL);
+	rt = in6_rtalloc1((struct sockaddr *)&mltaddr, 0, 0UL, RT_DEFAULT_FIB);
 	if (rt != NULL && rt->rt_gateway != NULL &&
 	    (memcmp(&satosin6(rt->rt_gateway)->sin6_addr, 
 		    &ia->ia_addr.sin6_addr,
 		    sizeof(ia->ia_addr.sin6_addr)) == 0)) {
 		/* 
-		 * if no more IPv6 address exists on this interface
-		 * then remove the multicast address route
+		 * If no more IPv6 address exists on this interface then
+		 * remove the multicast address route.
 		 */
 		if (ifa0 == NULL) {
 			memcpy(&mltaddr.sin6_addr, &satosin6(rt_key(rt))->sin6_addr, 
 			       sizeof(mltaddr.sin6_addr));
 
 			RTFREE_LOCKED(rt);
-			error = rtrequest(RTM_DELETE, (struct sockaddr *)&mltaddr,
-					  (struct sockaddr *)&ia->ia_addr,
-					  (struct sockaddr *)&mltmask, RTF_UP,
-					  (struct rtentry **)0);
-
+			error = in6_rtrequest(RTM_DELETE,
+			    (struct sockaddr *)&mltaddr,
+			    (struct sockaddr *)&ia->ia_addr,
+			    (struct sockaddr *)&mltmask, RTF_UP,
+			    (struct rtentry **)0, RT_DEFAULT_FIB);
 			if (error)
-				log(LOG_INFO, "in6_purgeaddr: node-local all-nodes"
-				    "multicast address deletion error\n");
+				log(LOG_INFO, "%s: node-local all-nodes"
+				    "multicast address deletion error\n",
+				    __func__);
 		} else {
 			/*
-			 * replace the gateway of the route
+			 * Replace the gateway of the route.
 			 */
 			struct sockaddr_in6 sa;
 
@@ -1476,31 +1448,70 @@ in6_purgeaddr(struct ifaddr *ifa)
 			RTFREE_LOCKED(rt);
 	}
 
-cleanup:
+	return (0);
+}
+
+void
+in6_purgeaddr(struct ifaddr *ifa)
+{
+	struct ifnet *ifp = ifa->ifa_ifp;
+	struct in6_ifaddr *ia = (struct in6_ifaddr *) ifa;
+	int plen, error;
+	struct ifaddr *ifa0;
+
+	if (ifa->ifa_carp)
+		(*carp_detach_p)(ifa);
+
+	/*
+	 * find another IPv6 address as the gateway for the
+	 * link-local and node-local all-nodes multicast
+	 * address routes
+	 */
+	IF_ADDR_RLOCK(ifp);
+	TAILQ_FOREACH(ifa0, &ifp->if_addrhead, ifa_link) {
+		if ((ifa0->ifa_addr->sa_family != AF_INET6) ||
+		    memcmp(&satosin6(ifa0->ifa_addr)->sin6_addr,
+			   &ia->ia_addr.sin6_addr, 
+			   sizeof(struct in6_addr)) == 0)
+			continue;
+		else
+			break;
+	}
+	if (ifa0 != NULL)
+		ifa_ref(ifa0);
+	IF_ADDR_RUNLOCK(ifp);
+
+	/*
+	 * Remove the loopback route to the interface address.
+	 * The check for the current setting of "nd6_useloopback" 
+	 * is not needed.
+	 */
+	if (ia->ia_flags & IFA_RTSELF) {
+		error = ifa_del_loopback_route((struct ifaddr *)ia,
+				       (struct sockaddr *)&ia->ia_addr);
+		if (error == 0)
+			ia->ia_flags &= ~IFA_RTSELF;
+	}
+
+	/* stop DAD processing */
+	nd6_dad_stop(ifa);
+
+	/* Remove local address entry from lltable. */
+	in6_ifremloop(ifa);
+
+	/* Leave multicast groups. */
+	error = in6_purgeaddr_mc(ifp, ia, ifa0);
+
 	if (ifa0 != NULL)
 		ifa_free(ifa0);
 
 	plen = in6_mask2len(&ia->ia_prefixmask.sin6_addr, NULL); /* XXX */
 	if ((ia->ia_flags & IFA_ROUTE) && plen == 128) {
-		int error;
-		struct sockaddr *dstaddr;
-
-		/* 
-		 * use the interface address if configuring an
-		 * interface address with a /128 prefix len
-		 */
-		if (ia->ia_dstaddr.sin6_family == AF_INET6)
-			dstaddr = (struct sockaddr *)&ia->ia_dstaddr;
-		else
-			dstaddr = (struct sockaddr *)&ia->ia_addr;
-
-		error = rtrequest(RTM_DELETE,
-		    (struct sockaddr *)dstaddr,
-		    (struct sockaddr *)&ia->ia_addr,
-		    (struct sockaddr *)&ia->ia_prefixmask,
-		    ia->ia_flags | RTF_HOST, NULL);
+		error = rtinit(&(ia->ia_ifa), RTM_DELETE, ia->ia_flags |
+		    (ia->ia_dstaddr.sin6_family == AF_INET6) ? RTF_HOST : 0);
 		if (error != 0)
-			return;
+			log(LOG_INFO, "%s: err=%d, destination address delete "
+			    "failed\n", __func__, error);
 		ia->ia_flags &= ~IFA_ROUTE;
 	}
 
@@ -1832,8 +1843,7 @@ in6_lifaddr_ioctl(struct socket *so, u_long cmd, caddr_t data,
 }
 
 /*
- * Initialize an interface's intetnet6 address
- * and routing table entry.
+ * Initialize an interface's IPv6 address and routing table entry.
  */
 static int
 in6_ifinit(struct ifnet *ifp, struct in6_ifaddr *ia,
@@ -1883,13 +1893,8 @@ in6_ifinit(struct ifnet *ifp, struct in6_ifaddr *ia,
 	if (!(ia->ia_flags & IFA_ROUTE) && plen == 128 &&
 	    ia->ia_dstaddr.sin6_family == AF_INET6) {
 		int rtflags = RTF_UP | RTF_HOST;
-
-		error = rtrequest(RTM_ADD,
-		    (struct sockaddr *)&ia->ia_dstaddr,
-		    (struct sockaddr *)&ia->ia_addr,
-		    (struct sockaddr *)&ia->ia_prefixmask,
-		    ia->ia_flags | rtflags, NULL);
-		if (error != 0)
+		error = rtinit(&ia->ia_ifa, RTM_ADD, ia->ia_flags | rtflags);
+		if (error)
 			return (error);
 		ia->ia_flags |= IFA_ROUTE;
 		/*
@@ -1909,7 +1914,7 @@ in6_ifinit(struct ifnet *ifp, struct in6_ifaddr *ia,
 			ia->ia_flags |= IFA_RTSELF;
 	}
 
-	/* Add ownaddr as loopback rtentry, if necessary (ex. on p2p link). */
+	/* Add local address to lltable, if necessary (ex. on p2p link). */
 	if (newhost)
 		in6_ifaddloop(&(ia->ia_ifa));
 
@@ -2512,8 +2517,10 @@ in6_lltable_rtcheck(struct ifnet *ifp,
 	KASSERT(l3addr->sa_family == AF_INET6,
 	    ("sin_family %d", l3addr->sa_family));
 
+	/* Our local addresses are always only installed on the default FIB. */
 	/* XXX rtalloc1 should take a const param */
-	rt = rtalloc1(__DECONST(struct sockaddr *, l3addr), 0, 0);
+	rt = in6_rtalloc1(__DECONST(struct sockaddr *, l3addr), 0, 0,
+	    RT_DEFAULT_FIB);
 	if (rt == NULL || (rt->rt_flags & RTF_GATEWAY) || rt->rt_ifp != ifp) {
 		struct ifaddr *ifa;
 		/*