2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the project nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * $KAME: nd6_rtr.c,v 1.111 2001/04/27 01:37:15 jinmei Exp $
34 #include <sys/cdefs.h>
35 __FBSDID("$FreeBSD$");
38 #include "opt_inet6.h"
40 #include <sys/param.h>
41 #include <sys/systm.h>
42 #include <sys/malloc.h>
44 #include <sys/refcount.h>
45 #include <sys/socket.h>
46 #include <sys/sockio.h>
48 #include <sys/kernel.h>
50 #include <sys/errno.h>
51 #include <sys/rmlock.h>
52 #include <sys/rwlock.h>
53 #include <sys/sysctl.h>
54 #include <sys/syslog.h>
55 #include <sys/queue.h>
58 #include <net/if_var.h>
59 #include <net/if_private.h>
60 #include <net/if_types.h>
61 #include <net/if_dl.h>
62 #include <net/route.h>
63 #include <net/route/nhop.h>
64 #include <net/route/route_ctl.h>
65 #include <net/radix.h>
68 #include <netinet/in.h>
69 #include <net/if_llatbl.h>
70 #include <netinet6/in6_var.h>
71 #include <netinet6/in6_ifattach.h>
72 #include <netinet/ip6.h>
73 #include <netinet6/ip6_var.h>
74 #include <netinet6/nd6.h>
75 #include <netinet/icmp6.h>
76 #include <netinet6/scope6_var.h>
78 static struct nd_defrouter *defrtrlist_update(struct nd_defrouter *);
79 static int prelist_update(struct nd_prefixctl *, struct nd_defrouter *,
81 static int nd6_prefix_onlink(struct nd_prefix *);
83 TAILQ_HEAD(nd6_drhead, nd_defrouter);
84 VNET_DEFINE_STATIC(struct nd6_drhead, nd6_defrouter);
85 #define V_nd6_defrouter VNET(nd6_defrouter)
87 VNET_DECLARE(int, nd6_recalc_reachtm_interval);
88 #define V_nd6_recalc_reachtm_interval VNET(nd6_recalc_reachtm_interval)
90 VNET_DEFINE_STATIC(struct ifnet *, nd6_defifp);
91 VNET_DEFINE(int, nd6_defifindex);
92 #define V_nd6_defifp VNET(nd6_defifp)
94 VNET_DEFINE(int, ip6_use_tempaddr) = 0;
96 VNET_DEFINE(int, ip6_desync_factor);
97 VNET_DEFINE(u_int32_t, ip6_temp_preferred_lifetime) = DEF_TEMP_PREFERRED_LIFETIME;
98 VNET_DEFINE(u_int32_t, ip6_temp_valid_lifetime) = DEF_TEMP_VALID_LIFETIME;
100 VNET_DEFINE(int, ip6_temp_regen_advance) = TEMPADDR_REGEN_ADVANCE;
103 VNET_DEFINE(int, nd6_ignore_ipv6_only_ra) = 1;
106 SYSCTL_DECL(_net_inet6_icmp6);
108 /* RTPREF_MEDIUM has to be 0! */
109 #define RTPREF_HIGH 1
110 #define RTPREF_MEDIUM 0
111 #define RTPREF_LOW (-1)
112 #define RTPREF_RESERVED (-2)
113 #define RTPREF_INVALID (-3) /* internal */
116 defrouter_ref(struct nd_defrouter *dr)
119 refcount_acquire(&dr->refcnt);
123 defrouter_rele(struct nd_defrouter *dr)
126 if (refcount_release(&dr->refcnt))
131 * Remove a router from the global list and optionally stash it in a
132 * caller-supplied queue.
135 defrouter_unlink(struct nd_defrouter *dr, struct nd6_drhead *drq)
140 TAILQ_REMOVE(&V_nd6_defrouter, dr, dr_entry);
143 TAILQ_INSERT_TAIL(drq, dr, dr_entry);
147 * Receive Router Solicitation Message - just for routers.
148 * Router solicitation/advertisement is mostly managed by userland program
149 * (rtadvd) so here we have no function like nd6_ra_output().
154 nd6_rs_input(struct mbuf *m, int off, int icmp6len)
158 struct nd_router_solicit *nd_rs;
159 struct in6_addr saddr6;
160 union nd_opts ndopts;
161 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
165 ifp = m->m_pkthdr.rcvif;
168 * Accept RS only when V_ip6_forwarding=1 and the interface has
169 * no ND6_IFF_ACCEPT_RTADV.
171 if (!V_ip6_forwarding || ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV)
174 /* RFC 6980: Nodes MUST silently ignore fragments */
175 if(m->m_flags & M_FRAGMENTED)
179 ip6 = mtod(m, struct ip6_hdr *);
180 if (__predict_false(ip6->ip6_hlim != 255)) {
181 ICMP6STAT_INC(icp6s_invlhlim);
183 "%s: invalid hlim (%d) from %s to %s on %s\n", __func__,
184 ip6->ip6_hlim, ip6_sprintf(ip6bufs, &ip6->ip6_src),
185 ip6_sprintf(ip6bufd, &ip6->ip6_dst), if_name(ifp)));
190 * Don't update the neighbor cache, if src = ::.
191 * This indicates that the src has no IP address assigned yet.
193 saddr6 = ip6->ip6_src;
194 if (IN6_IS_ADDR_UNSPECIFIED(&saddr6))
197 if (m->m_len < off + icmp6len) {
198 m = m_pullup(m, off + icmp6len);
200 IP6STAT_INC(ip6s_exthdrtoolong);
204 ip6 = mtod(m, struct ip6_hdr *);
205 nd_rs = (struct nd_router_solicit *)((caddr_t)ip6 + off);
207 icmp6len -= sizeof(*nd_rs);
208 nd6_option_init(nd_rs + 1, icmp6len, &ndopts);
209 if (nd6_options(&ndopts) < 0) {
211 "%s: invalid ND option, ignored\n", __func__));
212 /* nd6_options have incremented stats */
218 if (ndopts.nd_opts_src_lladdr) {
219 lladdr = (char *)(ndopts.nd_opts_src_lladdr + 1);
220 lladdrlen = ndopts.nd_opts_src_lladdr->nd_opt_len << 3;
223 if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) {
225 "%s: lladdrlen mismatch for %s (if %d, RS packet %d)\n",
226 __func__, ip6_sprintf(ip6bufs, &saddr6),
227 ifp->if_addrlen, lladdrlen - 2));
231 nd6_cache_lladdr(ifp, &saddr6, lladdr, lladdrlen, ND_ROUTER_SOLICIT, 0);
238 ICMP6STAT_INC(icp6s_badrs);
244 * An initial update routine for draft-ietf-6man-ipv6only-flag.
245 * We need to iterate over all default routers for the given
246 * interface to see whether they are all advertising the "S"
247 * (IPv6-Only) flag. If they do set, otherwise unset, the
248 * interface flag we later use to filter on.
251 defrtr_ipv6_only_ifp(struct ifnet *ifp)
253 struct nd_defrouter *dr;
254 bool ipv6_only, ipv6_only_old;
256 struct epoch_tracker et;
261 if (V_nd6_ignore_ipv6_only_ra != 0)
266 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry)
267 if (dr->ifp == ifp &&
268 (dr->raflags & ND_RA_FLAG_IPV6_ONLY) == 0)
272 IF_AFDATA_WLOCK(ifp);
273 ipv6_only_old = ND_IFINFO(ifp)->flags & ND6_IFF_IPV6_ONLY;
274 IF_AFDATA_WUNLOCK(ifp);
276 /* If nothing changed, we have an early exit. */
277 if (ipv6_only == ipv6_only_old)
282 * Should we want to set the IPV6-ONLY flag, check if the
283 * interface has a non-0/0 and non-link-local IPv4 address
284 * configured on it. If it has we will assume working
285 * IPv4 operations and will clear the interface flag.
287 has_ipv4_addr = false;
290 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
291 if (ifa->ifa_addr->sa_family != AF_INET)
294 satosin(ifa->ifa_addr)->sin_addr)) {
295 has_ipv4_addr = true;
301 if (ipv6_only && has_ipv4_addr) {
302 log(LOG_NOTICE, "%s rcvd RA w/ IPv6-Only flag set but has IPv4 "
303 "configured, ignoring IPv6-Only flag.\n", ifp->if_xname);
308 IF_AFDATA_WLOCK(ifp);
310 ND_IFINFO(ifp)->flags |= ND6_IFF_IPV6_ONLY;
312 ND_IFINFO(ifp)->flags &= ~ND6_IFF_IPV6_ONLY;
313 IF_AFDATA_WUNLOCK(ifp);
316 /* Send notification of flag change. */
321 defrtr_ipv6_only_ipf_down(struct ifnet *ifp)
324 IF_AFDATA_WLOCK(ifp);
325 ND_IFINFO(ifp)->flags &= ~ND6_IFF_IPV6_ONLY;
326 IF_AFDATA_WUNLOCK(ifp);
328 #endif /* EXPERIMENTAL */
331 nd6_ifnet_link_event(void *arg __unused, struct ifnet *ifp, int linkstate)
335 * XXX-BZ we might want to trigger re-evaluation of our default router
336 * availability. E.g., on link down the default router might be
337 * unreachable but a different interface might still have connectivity.
341 if (linkstate == LINK_STATE_DOWN)
342 defrtr_ipv6_only_ipf_down(ifp);
347 * Receive Router Advertisement Message.
350 * TODO: on-link bit on prefix information
351 * TODO: ND_RA_FLAG_{OTHER,MANAGED} processing
354 nd6_ra_input(struct mbuf *m, int off, int icmp6len)
357 struct nd_ifinfo *ndi;
359 struct nd_router_advert *nd_ra;
360 struct in6_addr saddr6;
361 struct nd_defrouter *dr;
362 union nd_opts ndopts;
363 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
367 * We only accept RAs only when the per-interface flag
368 * ND6_IFF_ACCEPT_RTADV is on the receiving interface.
370 ifp = m->m_pkthdr.rcvif;
371 ndi = ND_IFINFO(ifp);
372 if (!(ndi->flags & ND6_IFF_ACCEPT_RTADV))
375 /* RFC 6980: Nodes MUST silently ignore fragments */
376 if(m->m_flags & M_FRAGMENTED)
379 ip6 = mtod(m, struct ip6_hdr *);
380 if (__predict_false(ip6->ip6_hlim != 255)) {
381 ICMP6STAT_INC(icp6s_invlhlim);
383 "%s: invalid hlim (%d) from %s to %s on %s\n", __func__,
384 ip6->ip6_hlim, ip6_sprintf(ip6bufs, &ip6->ip6_src),
385 ip6_sprintf(ip6bufd, &ip6->ip6_dst), if_name(ifp)));
389 saddr6 = ip6->ip6_src;
390 if (!IN6_IS_ADDR_LINKLOCAL(&saddr6)) {
392 "%s: src %s is not link-local\n", __func__,
393 ip6_sprintf(ip6bufs, &saddr6)));
397 if (m->m_len < off + icmp6len) {
398 m = m_pullup(m, off + icmp6len);
400 IP6STAT_INC(ip6s_exthdrtoolong);
404 ip6 = mtod(m, struct ip6_hdr *);
405 nd_ra = (struct nd_router_advert *)((caddr_t)ip6 + off);
407 icmp6len -= sizeof(*nd_ra);
408 nd6_option_init(nd_ra + 1, icmp6len, &ndopts);
409 if (nd6_options(&ndopts) < 0) {
411 "%s: invalid ND option, ignored\n", __func__));
412 /* nd6_options have incremented stats */
419 struct nd_defrouter dr0;
420 u_int32_t advreachable = nd_ra->nd_ra_reachable;
422 /* remember if this is a multicasted advertisement */
423 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst))
426 bzero(&dr0, sizeof(dr0));
428 dr0.raflags = nd_ra->nd_ra_flags_reserved;
430 * Effectively-disable routes from RA messages when
431 * ND6_IFF_NO_RADR enabled on the receiving interface or
432 * (ip6.forwarding == 1 && ip6.rfc6204w3 != 1).
434 if (ndi->flags & ND6_IFF_NO_RADR)
436 else if (V_ip6_forwarding && !V_ip6_rfc6204w3)
439 dr0.rtlifetime = ntohs(nd_ra->nd_ra_router_lifetime);
440 dr0.expire = time_uptime + dr0.rtlifetime;
442 /* unspecified or not? (RFC 2461 6.3.4) */
444 advreachable = ntohl(advreachable);
445 if (advreachable <= MAX_REACHABLE_TIME &&
446 ndi->basereachable != advreachable) {
447 ndi->basereachable = advreachable;
448 ndi->reachable = ND_COMPUTE_RTIME(ndi->basereachable);
449 ndi->recalctm = V_nd6_recalc_reachtm_interval; /* reset */
452 if (nd_ra->nd_ra_retransmit)
453 ndi->retrans = ntohl(nd_ra->nd_ra_retransmit);
454 if (nd_ra->nd_ra_curhoplimit) {
455 if (ndi->chlim < nd_ra->nd_ra_curhoplimit)
456 ndi->chlim = nd_ra->nd_ra_curhoplimit;
457 else if (ndi->chlim != nd_ra->nd_ra_curhoplimit) {
458 log(LOG_ERR, "RA with a lower CurHopLimit sent from "
459 "%s on %s (current = %d, received = %d). "
460 "Ignored.\n", ip6_sprintf(ip6bufs, &ip6->ip6_src),
461 if_name(ifp), ndi->chlim, nd_ra->nd_ra_curhoplimit);
464 dr = defrtrlist_update(&dr0);
466 defrtr_ipv6_only_ifp(ifp);
473 if (ndopts.nd_opts_pi) {
474 struct nd_opt_hdr *pt;
475 struct nd_opt_prefix_info *pi = NULL;
476 struct nd_prefixctl pr;
478 for (pt = (struct nd_opt_hdr *)ndopts.nd_opts_pi;
479 pt <= (struct nd_opt_hdr *)ndopts.nd_opts_pi_end;
480 pt = (struct nd_opt_hdr *)((caddr_t)pt +
481 (pt->nd_opt_len << 3))) {
482 if (pt->nd_opt_type != ND_OPT_PREFIX_INFORMATION)
484 pi = (struct nd_opt_prefix_info *)pt;
486 if (pi->nd_opt_pi_len != 4) {
488 "%s: invalid option len %d for prefix "
489 "information option, ignored\n", __func__,
494 if (128 < pi->nd_opt_pi_prefix_len) {
496 "%s: invalid prefix len %d for prefix "
497 "information option, ignored\n", __func__,
498 pi->nd_opt_pi_prefix_len));
502 if (IN6_IS_ADDR_MULTICAST(&pi->nd_opt_pi_prefix)
503 || IN6_IS_ADDR_LINKLOCAL(&pi->nd_opt_pi_prefix)) {
505 "%s: invalid prefix %s, ignored\n",
506 __func__, ip6_sprintf(ip6bufs,
507 &pi->nd_opt_pi_prefix)));
511 bzero(&pr, sizeof(pr));
512 pr.ndpr_prefix.sin6_family = AF_INET6;
513 pr.ndpr_prefix.sin6_len = sizeof(pr.ndpr_prefix);
514 pr.ndpr_prefix.sin6_addr = pi->nd_opt_pi_prefix;
515 pr.ndpr_ifp = (struct ifnet *)m->m_pkthdr.rcvif;
517 pr.ndpr_raf_onlink = (pi->nd_opt_pi_flags_reserved &
518 ND_OPT_PI_FLAG_ONLINK) ? 1 : 0;
519 pr.ndpr_raf_auto = (pi->nd_opt_pi_flags_reserved &
520 ND_OPT_PI_FLAG_AUTO) ? 1 : 0;
521 pr.ndpr_raf_ra_derived = 1;
522 pr.ndpr_plen = pi->nd_opt_pi_prefix_len;
523 pr.ndpr_vltime = ntohl(pi->nd_opt_pi_valid_time);
524 pr.ndpr_pltime = ntohl(pi->nd_opt_pi_preferred_time);
525 (void)prelist_update(&pr, dr, m, mcast);
536 if (ndopts.nd_opts_mtu && ndopts.nd_opts_mtu->nd_opt_mtu_len == 1) {
540 mtu = (u_long)ntohl(ndopts.nd_opts_mtu->nd_opt_mtu_mtu);
543 if (mtu < IPV6_MMTU) {
544 nd6log((LOG_INFO, "%s: bogus mtu option mtu=%lu sent "
545 "from %s, ignoring\n", __func__,
546 mtu, ip6_sprintf(ip6bufs, &ip6->ip6_src)));
551 maxmtu = (ndi->maxmtu && ndi->maxmtu < ifp->if_mtu)
552 ? ndi->maxmtu : ifp->if_mtu;
554 int change = (ndi->linkmtu != mtu);
558 /* in6_maxmtu may change */
563 nd6log((LOG_INFO, "%s: bogus mtu=%lu sent from %s; "
564 "exceeds maxmtu %lu, ignoring\n", __func__,
565 mtu, ip6_sprintf(ip6bufs, &ip6->ip6_src), maxmtu));
572 * Source link layer address
578 if (ndopts.nd_opts_src_lladdr) {
579 lladdr = (char *)(ndopts.nd_opts_src_lladdr + 1);
580 lladdrlen = ndopts.nd_opts_src_lladdr->nd_opt_len << 3;
583 if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) {
585 "%s: lladdrlen mismatch for %s (if %d, RA packet %d)\n",
586 __func__, ip6_sprintf(ip6bufs, &saddr6),
587 ifp->if_addrlen, lladdrlen - 2));
591 nd6_cache_lladdr(ifp, &saddr6, lladdr,
592 lladdrlen, ND_ROUTER_ADVERT, 0);
595 * Installing a link-layer address might change the state of the
596 * router's neighbor cache, which might also affect our on-link
597 * detection of adveritsed prefixes.
599 pfxlist_onlink_check();
607 ICMP6STAT_INC(icp6s_badra);
612 static struct nd_pfxrouter *
613 pfxrtr_lookup(struct nd_prefix *pr, struct nd_defrouter *dr)
615 struct nd_pfxrouter *search;
619 LIST_FOREACH(search, &pr->ndpr_advrtrs, pfr_entry) {
620 if (search->router == dr)
627 pfxrtr_add(struct nd_prefix *pr, struct nd_defrouter *dr)
629 struct nd_pfxrouter *new;
635 if (pfxrtr_lookup(pr, dr) != NULL) {
641 new = malloc(sizeof(*new), M_IP6NDP, M_NOWAIT | M_ZERO);
648 if (pfxrtr_lookup(pr, dr) == NULL) {
649 LIST_INSERT_HEAD(&pr->ndpr_advrtrs, new, pfr_entry);
652 /* We lost a race to add the reference. */
660 pfxlist_onlink_check();
664 pfxrtr_del(struct nd_pfxrouter *pfr)
669 LIST_REMOVE(pfr, pfr_entry);
670 defrouter_rele(pfr->router);
674 /* Default router list processing sub routines. */
676 defrouter_addreq(struct nd_defrouter *new)
678 uint32_t fibnum = new->ifp->if_fib;
679 struct rib_cmd_info rc = {};
684 struct sockaddr_in6 gw = {
685 .sin6_family = AF_INET6,
686 .sin6_len = sizeof(struct sockaddr_in6),
687 .sin6_addr = new->rtaddr,
690 error = rib_add_default_route(fibnum, AF_INET6, new->ifp,
691 (struct sockaddr *)&gw, &rc);
694 struct nhop_object *nh = nhop_select_func(rc.rc_nh_new, 0);
695 rt_routemsg(RTM_ADD, rc.rc_rt, nh, fibnum);
701 * Remove the default route for a given router.
702 * This is just a subroutine function for defrouter_select_fib(), and
703 * should not be called from anywhere else.
706 defrouter_delreq(struct nd_defrouter *dr)
708 uint32_t fibnum = dr->ifp->if_fib;
709 struct epoch_tracker et;
710 struct rib_cmd_info rc;
713 struct sockaddr_in6 dst = {
714 .sin6_family = AF_INET6,
715 .sin6_len = sizeof(struct sockaddr_in6),
718 struct sockaddr_in6 gw = {
719 .sin6_family = AF_INET6,
720 .sin6_len = sizeof(struct sockaddr_in6),
721 .sin6_addr = dr->rtaddr,
725 error = rib_del_route_px(fibnum, (struct sockaddr *)&dst, 0,
726 rib_match_gw, (struct sockaddr *)&gw, 0, &rc);
728 struct nhop_object *nh = nhop_select_func(rc.rc_nh_old, 0);
729 rt_routemsg(RTM_DELETE, rc.rc_rt, nh, fibnum);
737 defrouter_del(struct nd_defrouter *dr)
739 struct nd_defrouter *deldr = NULL;
740 struct nd_prefix *pr;
741 struct nd_pfxrouter *pfxrtr;
746 * Flush all the routing table entries that use the router
749 if (ND_IFINFO(dr->ifp)->flags & ND6_IFF_ACCEPT_RTADV)
750 rt6_flush(&dr->rtaddr, dr->ifp);
753 defrtr_ipv6_only_ifp(dr->ifp);
758 defrouter_delreq(dr);
762 * Also delete all the pointers to the router in each prefix lists.
765 LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
766 if ((pfxrtr = pfxrtr_lookup(pr, dr)) != NULL)
771 pfxlist_onlink_check();
774 * If the router is the primary one, choose a new one.
775 * Note that defrouter_select_fib() will remove the current
776 * gateway from the routing table.
779 defrouter_select_fib(deldr->ifp->if_fib);
782 * Release the list reference.
787 struct nd_defrouter *
788 defrouter_lookup_locked(const struct in6_addr *addr, struct ifnet *ifp)
790 struct nd_defrouter *dr;
793 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry)
794 if (dr->ifp == ifp && IN6_ARE_ADDR_EQUAL(addr, &dr->rtaddr)) {
801 struct nd_defrouter *
802 defrouter_lookup(const struct in6_addr *addr, struct ifnet *ifp)
804 struct nd_defrouter *dr;
807 dr = defrouter_lookup_locked(addr, ifp);
813 * Remove all default routes from default router list.
816 defrouter_reset(void)
818 struct nd_defrouter *dr, **dra;
824 * We can't delete routes with the ND lock held, so make a copy of the
825 * current default router list and use that when deleting routes.
828 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry)
832 dra = malloc(count * sizeof(*dra), M_TEMP, M_WAITOK | M_ZERO);
835 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry) {
843 for (i = 0; i < count && dra[i] != NULL; i++) {
844 defrouter_delreq(dra[i]);
845 defrouter_rele(dra[i]);
850 * XXX should we also nuke any default routers in the kernel, by
851 * going through them by rtalloc1()?
856 * Look up a matching default router list entry and remove it. Returns true if a
857 * matching entry was found, false otherwise.
860 defrouter_remove(struct in6_addr *addr, struct ifnet *ifp)
862 struct nd_defrouter *dr;
865 dr = defrouter_lookup_locked(addr, ifp);
871 defrouter_unlink(dr, NULL);
879 * for default router selection
880 * regards router-preference field as a 2-bit signed integer
883 rtpref(struct nd_defrouter *dr)
885 switch (dr->raflags & ND_RA_FLAG_RTPREF_MASK) {
886 case ND_RA_FLAG_RTPREF_HIGH:
887 return (RTPREF_HIGH);
888 case ND_RA_FLAG_RTPREF_MEDIUM:
889 case ND_RA_FLAG_RTPREF_RSV:
890 return (RTPREF_MEDIUM);
891 case ND_RA_FLAG_RTPREF_LOW:
895 * This case should never happen. If it did, it would mean a
896 * serious bug of kernel internal. We thus always bark here.
897 * Or, can we even panic?
899 log(LOG_ERR, "rtpref: impossible RA flag %x\n", dr->raflags);
900 return (RTPREF_INVALID);
906 is_dr_reachable(const struct nd_defrouter *dr) {
907 struct llentry *ln = NULL;
909 ln = nd6_lookup(&dr->rtaddr, LLE_SF(AF_INET6, 0), dr->ifp);
912 bool reachable = ND6_IS_LLINFO_PROBREACH(ln);
918 * Default Router Selection according to Section 6.3.6 of RFC 2461 and
919 * draft-ietf-ipngwg-router-selection:
920 * 1) Routers that are reachable or probably reachable should be preferred.
921 * If we have more than one (probably) reachable router, prefer ones
922 * with the highest router preference.
923 * 2) When no routers on the list are known to be reachable or
924 * probably reachable, routers SHOULD be selected in a round-robin
925 * fashion, regardless of router preference values.
926 * 3) If the Default Router List is empty, assume that all
927 * destinations are on-link.
929 * We assume nd_defrouter is sorted by router preference value.
930 * Since the code below covers both with and without router preference cases,
931 * we do not need to classify the cases by ifdef.
933 * At this moment, we do not try to install more than one default router,
934 * even when the multipath routing is available, because we're not sure about
935 * the benefits for stub hosts comparing to the risk of making the code
936 * complicated and the possibility of introducing bugs.
938 * We maintain a single list of routers for multiple FIBs, only considering one
939 * at a time based on the receiving interface's FIB. If @fibnum is RT_ALL_FIBS,
940 * we do the whole thing multiple times.
943 defrouter_select_fib(int fibnum)
945 struct epoch_tracker et;
946 struct nd_defrouter *dr, *selected_dr, *installed_dr;
948 if (fibnum == RT_ALL_FIBS) {
949 for (fibnum = 0; fibnum < rt_numfibs; fibnum++) {
950 defrouter_select_fib(fibnum);
957 * Let's handle easy case (3) first:
958 * If default router list is empty, there's nothing to be done.
960 if (TAILQ_EMPTY(&V_nd6_defrouter)) {
966 * Search for a (probably) reachable router from the list.
967 * We just pick up the first reachable one (if any), assuming that
968 * the ordering rule of the list described in defrtrlist_update().
970 selected_dr = installed_dr = NULL;
972 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry) {
973 if (dr->ifp->if_fib != fibnum)
976 if (selected_dr == NULL && is_dr_reachable(dr)) {
978 defrouter_ref(selected_dr);
982 if (installed_dr == NULL) {
984 defrouter_ref(installed_dr);
987 * this should not happen.
988 * warn for diagnosis.
990 log(LOG_ERR, "defrouter_select_fib: more than "
991 "one router is installed\n");
997 * If none of the default routers was found to be reachable,
998 * round-robin the list regardless of preference.
999 * Otherwise, if we have an installed router, check if the selected
1000 * (reachable) router should really be preferred to the installed one.
1001 * We only prefer the new router when the old one is not reachable
1002 * or when the new one has a really higher preference value.
1004 if (selected_dr == NULL) {
1005 if (installed_dr == NULL ||
1006 TAILQ_NEXT(installed_dr, dr_entry) == NULL)
1007 dr = TAILQ_FIRST(&V_nd6_defrouter);
1009 dr = TAILQ_NEXT(installed_dr, dr_entry);
1011 /* Ensure we select a router for this FIB. */
1012 TAILQ_FOREACH_FROM(dr, &V_nd6_defrouter, dr_entry) {
1013 if (dr->ifp->if_fib == fibnum) {
1015 defrouter_ref(selected_dr);
1019 } else if (installed_dr != NULL) {
1020 if (is_dr_reachable(installed_dr) &&
1021 rtpref(selected_dr) <= rtpref(installed_dr)) {
1022 defrouter_rele(selected_dr);
1023 selected_dr = installed_dr;
1029 * If we selected a router for this FIB and it's different
1030 * than the installed one, remove the installed router and
1031 * install the selected one in its place.
1033 if (installed_dr != selected_dr) {
1034 if (installed_dr != NULL) {
1035 defrouter_delreq(installed_dr);
1036 defrouter_rele(installed_dr);
1038 if (selected_dr != NULL)
1039 defrouter_addreq(selected_dr);
1041 if (selected_dr != NULL)
1042 defrouter_rele(selected_dr);
1046 static struct nd_defrouter *
1047 defrtrlist_update(struct nd_defrouter *new)
1049 struct nd_defrouter *dr, *n;
1054 if (new->rtlifetime == 0) {
1055 defrouter_remove(&new->rtaddr, new->ifp);
1060 writelocked = false;
1062 dr = defrouter_lookup_locked(&new->rtaddr, new->ifp);
1064 oldpref = rtpref(dr);
1067 dr->raflags = new->raflags; /* XXX flag check */
1068 dr->rtlifetime = new->rtlifetime;
1069 dr->expire = new->expire;
1072 * If the preference does not change, there's no need
1073 * to sort the entries. Also make sure the selected
1074 * router is still installed in the kernel.
1076 if (dr->installed && rtpref(new) == oldpref) {
1086 * The router needs to be reinserted into the default router
1087 * list, so upgrade to a write lock. If that fails and the list
1088 * has potentially changed while the lock was dropped, we'll
1089 * redo the lookup with the write lock held.
1093 if (!ND6_TRY_UPGRADE()) {
1094 genid = V_nd6_list_genid;
1097 if (genid != V_nd6_list_genid)
1104 * The preferred router may have changed, so relocate this
1107 TAILQ_REMOVE(&V_nd6_defrouter, dr, dr_entry);
1110 n = malloc(sizeof(*n), M_IP6NDP, M_NOWAIT | M_ZERO);
1115 memcpy(n, new, sizeof(*n));
1116 /* Initialize with an extra reference for the caller. */
1117 refcount_init(&n->refcnt, 2);
1121 * Insert the new router in the Default Router List;
1122 * The Default Router List should be in the descending order
1123 * of router-preferece. Routers with the same preference are
1124 * sorted in the arriving time order.
1127 /* insert at the end of the group */
1128 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry) {
1129 if (rtpref(n) > rtpref(dr))
1133 TAILQ_INSERT_BEFORE(dr, n, dr_entry);
1135 TAILQ_INSERT_TAIL(&V_nd6_defrouter, n, dr_entry);
1139 defrouter_select_fib(new->ifp->if_fib);
1145 in6_init_prefix_ltimes(struct nd_prefix *ndpr)
1147 if (ndpr->ndpr_pltime == ND6_INFINITE_LIFETIME)
1148 ndpr->ndpr_preferred = 0;
1150 ndpr->ndpr_preferred = time_uptime + ndpr->ndpr_pltime;
1151 if (ndpr->ndpr_vltime == ND6_INFINITE_LIFETIME)
1152 ndpr->ndpr_expire = 0;
1154 ndpr->ndpr_expire = time_uptime + ndpr->ndpr_vltime;
1160 in6_init_address_ltimes(struct nd_prefix *new, struct in6_addrlifetime *lt6)
1162 /* init ia6t_expire */
1163 if (lt6->ia6t_vltime == ND6_INFINITE_LIFETIME)
1164 lt6->ia6t_expire = 0;
1166 lt6->ia6t_expire = time_uptime;
1167 lt6->ia6t_expire += lt6->ia6t_vltime;
1170 /* init ia6t_preferred */
1171 if (lt6->ia6t_pltime == ND6_INFINITE_LIFETIME)
1172 lt6->ia6t_preferred = 0;
1174 lt6->ia6t_preferred = time_uptime;
1175 lt6->ia6t_preferred += lt6->ia6t_pltime;
1179 static struct in6_ifaddr *
1180 in6_ifadd(struct nd_prefixctl *pr, int mcast)
1182 struct ifnet *ifp = pr->ndpr_ifp;
1184 struct in6_aliasreq ifra;
1185 struct in6_ifaddr *ia, *ib;
1187 struct in6_addr mask;
1188 int prefixlen = pr->ndpr_plen;
1190 char ip6buf[INET6_ADDRSTRLEN];
1192 in6_prefixlen2mask(&mask, prefixlen);
1195 * find a link-local address (will be interface ID).
1196 * Is it really mandatory? Theoretically, a global or a site-local
1197 * address can be configured without a link-local address, if we
1198 * have a unique interface identifier...
1200 * it is not mandatory to have a link-local address, we can generate
1201 * interface identifier on the fly. we do this because:
1202 * (1) it should be the easiest way to find interface identifier.
1203 * (2) RFC2462 5.4 suggesting the use of the same interface identifier
1204 * for multiple addresses on a single interface, and possible shortcut
1205 * of DAD. we omitted DAD for this reason in the past.
1206 * (3) a user can prevent autoconfiguration of global address
1207 * by removing link-local address by hand (this is partly because we
1208 * don't have other way to control the use of IPv6 on an interface.
1209 * this has been our design choice - cf. NRL's "ifconfig auto").
1210 * (4) it is easier to manage when an interface has addresses
1211 * with the same interface identifier, than to have multiple addresses
1212 * with different interface identifiers.
1214 ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp, 0); /* 0 is OK? */
1216 ib = (struct in6_ifaddr *)ifa;
1220 /* prefixlen + ifidlen must be equal to 128 */
1221 plen0 = in6_mask2len(&ib->ia_prefixmask.sin6_addr, NULL);
1222 if (prefixlen != plen0) {
1225 "%s: wrong prefixlen for %s (prefix=%d ifid=%d)\n",
1226 __func__, if_name(ifp), prefixlen, 128 - plen0));
1231 in6_prepare_ifra(&ifra, &pr->ndpr_prefix.sin6_addr, &mask);
1233 IN6_MASK_ADDR(&ifra.ifra_addr.sin6_addr, &mask);
1235 ifra.ifra_addr.sin6_addr.s6_addr32[0] |=
1236 (ib->ia_addr.sin6_addr.s6_addr32[0] & ~mask.s6_addr32[0]);
1237 ifra.ifra_addr.sin6_addr.s6_addr32[1] |=
1238 (ib->ia_addr.sin6_addr.s6_addr32[1] & ~mask.s6_addr32[1]);
1239 ifra.ifra_addr.sin6_addr.s6_addr32[2] |=
1240 (ib->ia_addr.sin6_addr.s6_addr32[2] & ~mask.s6_addr32[2]);
1241 ifra.ifra_addr.sin6_addr.s6_addr32[3] |=
1242 (ib->ia_addr.sin6_addr.s6_addr32[3] & ~mask.s6_addr32[3]);
1246 ifra.ifra_lifetime.ia6t_vltime = pr->ndpr_vltime;
1247 ifra.ifra_lifetime.ia6t_pltime = pr->ndpr_pltime;
1249 /* XXX: scope zone ID? */
1251 ifra.ifra_flags |= IN6_IFF_AUTOCONF; /* obey autoconf */
1254 * Make sure that we do not have this address already. This should
1255 * usually not happen, but we can still see this case, e.g., if we
1256 * have manually configured the exact address to be configured.
1258 ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp,
1259 &ifra.ifra_addr.sin6_addr);
1262 /* this should be rare enough to make an explicit log */
1263 log(LOG_INFO, "in6_ifadd: %s is already configured\n",
1264 ip6_sprintf(ip6buf, &ifra.ifra_addr.sin6_addr));
1269 * Allocate ifaddr structure, link into chain, etc.
1270 * If we are going to create a new address upon receiving a multicasted
1271 * RA, we need to impose a random delay before starting DAD.
1272 * [draft-ietf-ipv6-rfc2462bis-02.txt, Section 5.4.2]
1276 updateflags |= IN6_IFAUPDATE_DADDELAY;
1277 if ((error = in6_update_ifa(ifp, &ifra, NULL, updateflags)) != 0) {
1279 "%s: failed to make ifaddr %s on %s (errno=%d)\n", __func__,
1280 ip6_sprintf(ip6buf, &ifra.ifra_addr.sin6_addr),
1281 if_name(ifp), error));
1282 return (NULL); /* ifaddr must not have been allocated. */
1285 ia = in6ifa_ifpwithaddr(ifp, &ifra.ifra_addr.sin6_addr);
1287 * XXXRW: Assumption of non-NULLness here might not be true with
1288 * fine-grained locking -- should we validate it? Or just return
1289 * earlier ifa rather than looking it up again?
1291 return (ia); /* this is always non-NULL and referenced. */
1294 static struct nd_prefix *
1295 nd6_prefix_lookup_locked(struct nd_prefixctl *key)
1297 struct nd_prefix *search;
1301 LIST_FOREACH(search, &V_nd_prefix, ndpr_entry) {
1302 if (key->ndpr_ifp == search->ndpr_ifp &&
1303 key->ndpr_plen == search->ndpr_plen &&
1304 in6_are_prefix_equal(&key->ndpr_prefix.sin6_addr,
1305 &search->ndpr_prefix.sin6_addr, key->ndpr_plen)) {
1306 nd6_prefix_ref(search);
1314 nd6_prefix_lookup(struct nd_prefixctl *key)
1316 struct nd_prefix *search;
1319 search = nd6_prefix_lookup_locked(key);
1325 nd6_prefix_ref(struct nd_prefix *pr)
1328 refcount_acquire(&pr->ndpr_refcnt);
1332 nd6_prefix_rele(struct nd_prefix *pr)
1335 if (refcount_release(&pr->ndpr_refcnt)) {
1336 KASSERT(LIST_EMPTY(&pr->ndpr_advrtrs),
1337 ("prefix %p has advertising routers", pr));
1343 nd6_prelist_add(struct nd_prefixctl *pr, struct nd_defrouter *dr,
1344 struct nd_prefix **newp)
1346 struct nd_prefix *new;
1347 char ip6buf[INET6_ADDRSTRLEN];
1350 new = malloc(sizeof(*new), M_IP6NDP, M_NOWAIT | M_ZERO);
1353 refcount_init(&new->ndpr_refcnt, newp != NULL ? 2 : 1);
1354 new->ndpr_ifp = pr->ndpr_ifp;
1355 new->ndpr_prefix = pr->ndpr_prefix;
1356 new->ndpr_plen = pr->ndpr_plen;
1357 new->ndpr_vltime = pr->ndpr_vltime;
1358 new->ndpr_pltime = pr->ndpr_pltime;
1359 new->ndpr_flags = pr->ndpr_flags;
1360 if ((error = in6_init_prefix_ltimes(new)) != 0) {
1361 free(new, M_IP6NDP);
1364 new->ndpr_lastupdate = time_uptime;
1366 /* initialization */
1367 LIST_INIT(&new->ndpr_advrtrs);
1368 in6_prefixlen2mask(&new->ndpr_mask, new->ndpr_plen);
1369 /* make prefix in the canonical form */
1370 IN6_MASK_ADDR(&new->ndpr_prefix.sin6_addr, &new->ndpr_mask);
1373 LIST_INSERT_HEAD(&V_nd_prefix, new, ndpr_entry);
1377 /* ND_OPT_PI_FLAG_ONLINK processing */
1378 if (new->ndpr_raf_onlink) {
1379 struct epoch_tracker et;
1382 NET_EPOCH_ENTER(et);
1383 if ((error = nd6_prefix_onlink(new)) != 0) {
1384 nd6log((LOG_ERR, "%s: failed to make the prefix %s/%d "
1385 "on-link on %s (errno=%d)\n", __func__,
1386 ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
1387 pr->ndpr_plen, if_name(pr->ndpr_ifp), error));
1388 /* proceed anyway. XXX: is it correct? */
1391 ND6_ONLINK_UNLOCK();
1395 pfxrtr_add(new, dr);
1402 * Remove a prefix from the prefix list and optionally stash it in a
1403 * caller-provided list.
1405 * The ND6 lock must be held.
1408 nd6_prefix_unlink(struct nd_prefix *pr, struct nd_prhead *list)
1413 LIST_REMOVE(pr, ndpr_entry);
1416 LIST_INSERT_HEAD(list, pr, ndpr_entry);
1420 * Free an unlinked prefix, first marking it off-link if necessary.
1423 nd6_prefix_del(struct nd_prefix *pr)
1425 struct nd_pfxrouter *pfr, *next;
1427 char ip6buf[INET6_ADDRSTRLEN];
1429 KASSERT(pr->ndpr_addrcnt == 0,
1430 ("prefix %p has referencing addresses", pr));
1431 ND6_UNLOCK_ASSERT();
1434 * Though these flags are now meaningless, we'd rather keep the value
1435 * of pr->ndpr_raf_onlink and pr->ndpr_raf_auto not to confuse users
1436 * when executing "ndp -p".
1438 if ((pr->ndpr_stateflags & NDPRF_ONLINK) != 0) {
1440 if ((e = nd6_prefix_offlink(pr)) != 0) {
1442 "%s: failed to make the prefix %s/%d offlink on %s "
1443 "(errno=%d)\n", __func__,
1444 ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
1445 pr->ndpr_plen, if_name(pr->ndpr_ifp), e));
1446 /* what should we do? */
1448 ND6_ONLINK_UNLOCK();
1451 /* Release references to routers that have advertised this prefix. */
1453 LIST_FOREACH_SAFE(pfr, &pr->ndpr_advrtrs, pfr_entry, next)
1457 nd6_prefix_rele(pr);
1459 pfxlist_onlink_check();
1463 prelist_update(struct nd_prefixctl *new, struct nd_defrouter *dr,
1464 struct mbuf *m, int mcast)
1466 struct in6_ifaddr *ia6 = NULL, *ia6_match = NULL;
1468 struct ifnet *ifp = new->ndpr_ifp;
1469 struct nd_prefix *pr;
1472 struct in6_addrlifetime lt6_tmp;
1473 char ip6buf[INET6_ADDRSTRLEN];
1480 * Authenticity for NA consists authentication for
1481 * both IP header and IP datagrams, doesn't it ?
1483 #if defined(M_AUTHIPHDR) && defined(M_AUTHIPDGM)
1484 auth = ((m->m_flags & M_AUTHIPHDR) &&
1485 (m->m_flags & M_AUTHIPDGM));
1489 if ((pr = nd6_prefix_lookup(new)) != NULL) {
1491 * nd6_prefix_lookup() ensures that pr and new have the same
1492 * prefix on a same interface.
1496 * Update prefix information. Note that the on-link (L) bit
1497 * and the autonomous (A) bit should NOT be changed from 1
1500 if (new->ndpr_raf_onlink == 1)
1501 pr->ndpr_raf_onlink = 1;
1502 if (new->ndpr_raf_auto == 1)
1503 pr->ndpr_raf_auto = 1;
1504 if (new->ndpr_raf_onlink) {
1505 pr->ndpr_vltime = new->ndpr_vltime;
1506 pr->ndpr_pltime = new->ndpr_pltime;
1507 (void)in6_init_prefix_ltimes(pr); /* XXX error case? */
1508 pr->ndpr_lastupdate = time_uptime;
1511 if (new->ndpr_raf_onlink &&
1512 (pr->ndpr_stateflags & NDPRF_ONLINK) == 0) {
1514 if ((error = nd6_prefix_onlink(pr)) != 0) {
1516 "%s: failed to make the prefix %s/%d "
1517 "on-link on %s (errno=%d)\n", __func__,
1519 &pr->ndpr_prefix.sin6_addr),
1520 pr->ndpr_plen, if_name(pr->ndpr_ifp),
1522 /* proceed anyway. XXX: is it correct? */
1524 ND6_ONLINK_UNLOCK();
1530 if (new->ndpr_vltime == 0)
1532 if (new->ndpr_raf_onlink == 0 && new->ndpr_raf_auto == 0)
1535 error = nd6_prelist_add(new, dr, &pr);
1537 nd6log((LOG_NOTICE, "%s: nd6_prelist_add() failed for "
1538 "the prefix %s/%d on %s (errno=%d)\n", __func__,
1539 ip6_sprintf(ip6buf, &new->ndpr_prefix.sin6_addr),
1540 new->ndpr_plen, if_name(new->ndpr_ifp), error));
1541 goto end; /* we should just give up in this case. */
1545 * XXX: from the ND point of view, we can ignore a prefix
1546 * with the on-link bit being zero. However, we need a
1547 * prefix structure for references from autoconfigured
1548 * addresses. Thus, we explicitly make sure that the prefix
1549 * itself expires now.
1551 if (pr->ndpr_raf_onlink == 0) {
1552 pr->ndpr_vltime = 0;
1553 pr->ndpr_pltime = 0;
1554 in6_init_prefix_ltimes(pr);
1559 * Address autoconfiguration based on Section 5.5.3 of RFC 2462.
1560 * Note that pr must be non NULL at this point.
1563 /* 5.5.3 (a). Ignore the prefix without the A bit set. */
1564 if (!new->ndpr_raf_auto)
1568 * 5.5.3 (b). the link-local prefix should have been ignored in
1572 /* 5.5.3 (c). Consistency check on lifetimes: pltime <= vltime. */
1573 if (new->ndpr_pltime > new->ndpr_vltime) {
1574 error = EINVAL; /* XXX: won't be used */
1579 * 5.5.3 (d). If the prefix advertised is not equal to the prefix of
1580 * an address configured by stateless autoconfiguration already in the
1581 * list of addresses associated with the interface, and the Valid
1582 * Lifetime is not 0, form an address. We first check if we have
1583 * a matching prefix.
1584 * Note: we apply a clarification in rfc2462bis-02 here. We only
1585 * consider autoconfigured addresses while RFC2462 simply said
1588 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1589 struct in6_ifaddr *ifa6;
1590 u_int32_t remaininglifetime;
1592 if (ifa->ifa_addr->sa_family != AF_INET6)
1595 ifa6 = (struct in6_ifaddr *)ifa;
1598 * We only consider autoconfigured addresses as per rfc2462bis.
1600 if (!(ifa6->ia6_flags & IN6_IFF_AUTOCONF))
1604 * Spec is not clear here, but I believe we should concentrate
1605 * on unicast (i.e. not anycast) addresses.
1606 * XXX: other ia6_flags? detached or duplicated?
1608 if ((ifa6->ia6_flags & IN6_IFF_ANYCAST) != 0)
1612 * Ignore the address if it is not associated with a prefix
1613 * or is associated with a prefix that is different from this
1614 * one. (pr is never NULL here)
1616 if (ifa6->ia6_ndpr != pr)
1619 if (ia6_match == NULL) /* remember the first one */
1623 * An already autoconfigured address matched. Now that we
1624 * are sure there is at least one matched address, we can
1625 * proceed to 5.5.3. (e): update the lifetimes according to the
1626 * "two hours" rule and the privacy extension.
1627 * We apply some clarifications in rfc2462bis:
1628 * - use remaininglifetime instead of storedlifetime as a
1630 * - remove the dead code in the "two-hour" rule
1632 #define TWOHOUR (120*60)
1633 lt6_tmp = ifa6->ia6_lifetime;
1635 if (lt6_tmp.ia6t_vltime == ND6_INFINITE_LIFETIME)
1636 remaininglifetime = ND6_INFINITE_LIFETIME;
1637 else if (time_uptime - ifa6->ia6_updatetime >
1638 lt6_tmp.ia6t_vltime) {
1640 * The case of "invalid" address. We should usually
1641 * not see this case.
1643 remaininglifetime = 0;
1645 remaininglifetime = lt6_tmp.ia6t_vltime -
1646 (time_uptime - ifa6->ia6_updatetime);
1648 /* when not updating, keep the current stored lifetime. */
1649 lt6_tmp.ia6t_vltime = remaininglifetime;
1651 if (TWOHOUR < new->ndpr_vltime ||
1652 remaininglifetime < new->ndpr_vltime) {
1653 lt6_tmp.ia6t_vltime = new->ndpr_vltime;
1654 } else if (remaininglifetime <= TWOHOUR) {
1656 lt6_tmp.ia6t_vltime = new->ndpr_vltime;
1660 * new->ndpr_vltime <= TWOHOUR &&
1661 * TWOHOUR < remaininglifetime
1663 lt6_tmp.ia6t_vltime = TWOHOUR;
1666 /* The 2 hour rule is not imposed for preferred lifetime. */
1667 lt6_tmp.ia6t_pltime = new->ndpr_pltime;
1669 in6_init_address_ltimes(pr, <6_tmp);
1672 * We need to treat lifetimes for temporary addresses
1673 * differently, according to
1674 * draft-ietf-ipv6-privacy-addrs-v2-01.txt 3.3 (1);
1675 * we only update the lifetimes when they are in the maximum
1678 if ((ifa6->ia6_flags & IN6_IFF_TEMPORARY) != 0) {
1679 u_int32_t maxvltime, maxpltime;
1681 if (V_ip6_temp_valid_lifetime >
1682 (u_int32_t)((time_uptime - ifa6->ia6_createtime) +
1683 V_ip6_desync_factor)) {
1684 maxvltime = V_ip6_temp_valid_lifetime -
1685 (time_uptime - ifa6->ia6_createtime) -
1686 V_ip6_desync_factor;
1689 if (V_ip6_temp_preferred_lifetime >
1690 (u_int32_t)((time_uptime - ifa6->ia6_createtime) +
1691 V_ip6_desync_factor)) {
1692 maxpltime = V_ip6_temp_preferred_lifetime -
1693 (time_uptime - ifa6->ia6_createtime) -
1694 V_ip6_desync_factor;
1698 if (lt6_tmp.ia6t_vltime == ND6_INFINITE_LIFETIME ||
1699 lt6_tmp.ia6t_vltime > maxvltime) {
1700 lt6_tmp.ia6t_vltime = maxvltime;
1702 if (lt6_tmp.ia6t_pltime == ND6_INFINITE_LIFETIME ||
1703 lt6_tmp.ia6t_pltime > maxpltime) {
1704 lt6_tmp.ia6t_pltime = maxpltime;
1707 ifa6->ia6_lifetime = lt6_tmp;
1708 ifa6->ia6_updatetime = time_uptime;
1710 if (ia6_match == NULL && new->ndpr_vltime) {
1714 * 5.5.3 (d) (continued)
1715 * No address matched and the valid lifetime is non-zero.
1716 * Create a new address.
1720 * Prefix Length check:
1721 * If the sum of the prefix length and interface identifier
1722 * length does not equal 128 bits, the Prefix Information
1723 * option MUST be ignored. The length of the interface
1724 * identifier is defined in a separate link-type specific
1727 ifidlen = in6_if2idlen(ifp);
1729 /* this should not happen, so we always log it. */
1730 log(LOG_ERR, "prelist_update: IFID undefined (%s)\n",
1734 if (ifidlen + pr->ndpr_plen != 128) {
1736 "%s: invalid prefixlen %d for %s, ignored\n",
1737 __func__, pr->ndpr_plen, if_name(ifp)));
1741 if ((ia6 = in6_ifadd(new, mcast)) != NULL) {
1743 * note that we should use pr (not new) for reference.
1750 * When a new public address is created as described
1751 * in RFC2462, also create a new temporary address.
1754 * When an interface connects to a new link, a new
1755 * randomized interface identifier should be generated
1756 * immediately together with a new set of temporary
1757 * addresses. Thus, we specifiy 1 as the 2nd arg of
1760 if (V_ip6_use_tempaddr) {
1762 if ((e = in6_tmpifadd(ia6, 1, 1)) != 0) {
1763 nd6log((LOG_NOTICE, "%s: failed to "
1764 "create a temporary address "
1765 "(errno=%d)\n", __func__, e));
1768 ifa_free(&ia6->ia_ifa);
1771 * A newly added address might affect the status
1772 * of other addresses, so we check and update it.
1773 * XXX: what if address duplication happens?
1775 pfxlist_onlink_check();
1777 /* just set an error. do not bark here. */
1778 error = EADDRNOTAVAIL; /* XXX: might be unused. */
1784 nd6_prefix_rele(pr);
1789 * A supplement function used in the on-link detection below;
1790 * detect if a given prefix has a (probably) reachable advertising router.
1791 * XXX: lengthy function name...
1793 static struct nd_pfxrouter *
1794 find_pfxlist_reachable_router(struct nd_prefix *pr)
1796 struct epoch_tracker et;
1797 struct nd_pfxrouter *pfxrtr;
1801 NET_EPOCH_ENTER(et);
1802 LIST_FOREACH(pfxrtr, &pr->ndpr_advrtrs, pfr_entry) {
1803 if (is_dr_reachable(pfxrtr->router))
1811 * Check if each prefix in the prefix list has at least one available router
1812 * that advertised the prefix (a router is "available" if its neighbor cache
1813 * entry is reachable or probably reachable).
1814 * If the check fails, the prefix may be off-link, because, for example,
1815 * we have moved from the network but the lifetime of the prefix has not
1816 * expired yet. So we should not use the prefix if there is another prefix
1817 * that has an available router.
1818 * But, if there is no prefix that has an available router, we still regard
1819 * all the prefixes as on-link. This is because we can't tell if all the
1820 * routers are simply dead or if we really moved from the network and there
1821 * is no router around us.
1824 pfxlist_onlink_check(void)
1826 struct nd_prefix *pr;
1827 struct in6_ifaddr *ifa;
1828 struct nd_defrouter *dr;
1829 struct nd_pfxrouter *pfxrtr = NULL;
1830 struct rm_priotracker in6_ifa_tracker;
1838 * Check if there is a prefix that has a reachable advertising
1841 LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
1842 if (pr->ndpr_raf_onlink && find_pfxlist_reachable_router(pr))
1847 * If we have no such prefix, check whether we still have a router
1848 * that does not advertise any prefixes.
1851 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry) {
1852 struct nd_prefix *pr0;
1854 LIST_FOREACH(pr0, &V_nd_prefix, ndpr_entry) {
1855 if ((pfxrtr = pfxrtr_lookup(pr0, dr)) != NULL)
1862 if (pr != NULL || (!TAILQ_EMPTY(&V_nd6_defrouter) && pfxrtr == NULL)) {
1864 * There is at least one prefix that has a reachable router,
1865 * or at least a router which probably does not advertise
1866 * any prefixes. The latter would be the case when we move
1867 * to a new link where we have a router that does not provide
1868 * prefixes and we configure an address by hand.
1869 * Detach prefixes which have no reachable advertising
1870 * router, and attach other prefixes.
1872 LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
1873 /* XXX: a link-local prefix should never be detached */
1874 if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr) ||
1875 pr->ndpr_raf_onlink == 0 ||
1876 pr->ndpr_raf_auto == 0)
1879 if ((pr->ndpr_stateflags & NDPRF_DETACHED) == 0 &&
1880 find_pfxlist_reachable_router(pr) == NULL)
1881 pr->ndpr_stateflags |= NDPRF_DETACHED;
1882 else if ((pr->ndpr_stateflags & NDPRF_DETACHED) != 0 &&
1883 find_pfxlist_reachable_router(pr) != NULL)
1884 pr->ndpr_stateflags &= ~NDPRF_DETACHED;
1887 /* there is no prefix that has a reachable router */
1888 LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
1889 if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr) ||
1890 pr->ndpr_raf_onlink == 0 ||
1891 pr->ndpr_raf_auto == 0)
1893 pr->ndpr_stateflags &= ~NDPRF_DETACHED;
1898 * Remove each interface route associated with a (just) detached
1899 * prefix, and reinstall the interface route for a (just) attached
1900 * prefix. Note that all attempt of reinstallation does not
1901 * necessarily success, when a same prefix is shared among multiple
1902 * interfaces. Such cases will be handled in nd6_prefix_onlink,
1903 * so we don't have to care about them.
1906 LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
1907 char ip6buf[INET6_ADDRSTRLEN];
1910 if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr) ||
1911 pr->ndpr_raf_onlink == 0 ||
1912 pr->ndpr_raf_auto == 0)
1915 flags = pr->ndpr_stateflags & (NDPRF_DETACHED | NDPRF_ONLINK);
1916 if (flags == 0 || flags == (NDPRF_DETACHED | NDPRF_ONLINK)) {
1917 genid = V_nd6_list_genid;
1919 if ((flags & NDPRF_ONLINK) != 0 &&
1920 (e = nd6_prefix_offlink(pr)) != 0) {
1922 "%s: failed to make %s/%d offlink "
1923 "(errno=%d)\n", __func__,
1925 &pr->ndpr_prefix.sin6_addr),
1927 } else if ((flags & NDPRF_ONLINK) == 0 &&
1928 (e = nd6_prefix_onlink(pr)) != 0) {
1930 "%s: failed to make %s/%d onlink "
1931 "(errno=%d)\n", __func__,
1933 &pr->ndpr_prefix.sin6_addr),
1937 if (genid != V_nd6_list_genid)
1943 * Changes on the prefix status might affect address status as well.
1944 * Make sure that all addresses derived from an attached prefix are
1945 * attached, and that all addresses derived from a detached prefix are
1946 * detached. Note, however, that a manually configured address should
1947 * always be attached.
1948 * The precise detection logic is same as the one for prefixes.
1950 IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1951 CK_STAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) {
1952 if (!(ifa->ia6_flags & IN6_IFF_AUTOCONF))
1955 if (ifa->ia6_ndpr == NULL) {
1957 * This can happen when we first configure the address
1958 * (i.e. the address exists, but the prefix does not).
1959 * XXX: complicated relationships...
1964 if (find_pfxlist_reachable_router(ifa->ia6_ndpr))
1968 CK_STAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) {
1969 if ((ifa->ia6_flags & IN6_IFF_AUTOCONF) == 0)
1972 if (ifa->ia6_ndpr == NULL) /* XXX: see above. */
1975 if (find_pfxlist_reachable_router(ifa->ia6_ndpr)) {
1976 if (ifa->ia6_flags & IN6_IFF_DETACHED) {
1977 ifa->ia6_flags &= ~IN6_IFF_DETACHED;
1978 ifa->ia6_flags |= IN6_IFF_TENTATIVE;
1979 nd6_dad_start((struct ifaddr *)ifa, 0);
1982 ifa->ia6_flags |= IN6_IFF_DETACHED;
1986 CK_STAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) {
1987 if ((ifa->ia6_flags & IN6_IFF_AUTOCONF) == 0)
1990 if (ifa->ia6_flags & IN6_IFF_DETACHED) {
1991 ifa->ia6_flags &= ~IN6_IFF_DETACHED;
1992 ifa->ia6_flags |= IN6_IFF_TENTATIVE;
1993 /* Do we need a delay in this case? */
1994 nd6_dad_start((struct ifaddr *)ifa, 0);
1998 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
2000 ND6_ONLINK_UNLOCK();
2004 * Add or remove interface route specified by @dst, @netmask and @ifp.
2006 * Returns 0 on success
2009 nd6_prefix_rtrequest(uint32_t fibnum, int cmd, struct sockaddr_in6 *dst,
2010 struct sockaddr_in6 *netmask, struct ifnet *ifp, struct ifaddr *ifa)
2012 struct epoch_tracker et;
2015 /* Prepare gateway */
2016 struct sockaddr_dl_short sdl = {
2017 .sdl_family = AF_LINK,
2018 .sdl_len = sizeof(struct sockaddr_dl_short),
2019 .sdl_type = ifp->if_type,
2020 .sdl_index = ifp->if_index,
2023 struct rt_addrinfo info = {
2026 .rti_flags = RTF_PINNED | ((netmask != NULL) ? 0 : RTF_HOST),
2028 [RTAX_DST] = (struct sockaddr *)dst,
2029 [RTAX_NETMASK] = (struct sockaddr *)netmask,
2030 [RTAX_GATEWAY] = (struct sockaddr *)&sdl,
2033 /* Don't set additional per-gw filters on removal */
2035 NET_EPOCH_ENTER(et);
2036 error = rib_handle_ifaddr_info(fibnum, cmd, &info);
2042 nd6_prefix_onlink_rtrequest(struct nd_prefix *pr, struct ifaddr *ifa)
2046 struct sockaddr_in6 mask6 = {
2047 .sin6_family = AF_INET6,
2048 .sin6_len = sizeof(struct sockaddr_in6),
2049 .sin6_addr = pr->ndpr_mask,
2051 struct sockaddr_in6 *pmask6 = (pr->ndpr_plen != 128) ? &mask6 : NULL;
2053 error = nd6_prefix_rtrequest(pr->ndpr_ifp->if_fib, RTM_ADD,
2054 &pr->ndpr_prefix, pmask6, pr->ndpr_ifp, ifa);
2056 pr->ndpr_stateflags |= NDPRF_ONLINK;
2062 nd6_prefix_onlink(struct nd_prefix *pr)
2064 struct epoch_tracker et;
2066 struct ifnet *ifp = pr->ndpr_ifp;
2067 struct nd_prefix *opr;
2068 char ip6buf[INET6_ADDRSTRLEN];
2071 ND6_ONLINK_LOCK_ASSERT();
2072 ND6_UNLOCK_ASSERT();
2074 if ((pr->ndpr_stateflags & NDPRF_ONLINK) != 0)
2078 * Add the interface route associated with the prefix. Before
2079 * installing the route, check if there's the same prefix on another
2080 * interface, and the prefix has already installed the interface route.
2081 * Although such a configuration is expected to be rare, we explicitly
2085 LIST_FOREACH(opr, &V_nd_prefix, ndpr_entry) {
2089 if ((opr->ndpr_stateflags & NDPRF_ONLINK) == 0)
2092 if (!V_rt_add_addr_allfibs &&
2093 opr->ndpr_ifp->if_fib != pr->ndpr_ifp->if_fib)
2096 if (opr->ndpr_plen == pr->ndpr_plen &&
2097 in6_are_prefix_equal(&pr->ndpr_prefix.sin6_addr,
2098 &opr->ndpr_prefix.sin6_addr, pr->ndpr_plen)) {
2106 * We prefer link-local addresses as the associated interface address.
2108 /* search for a link-local addr */
2109 NET_EPOCH_ENTER(et);
2110 ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp,
2111 IN6_IFF_NOTREADY | IN6_IFF_ANYCAST);
2113 /* XXX: freebsd does not have ifa_ifwithaf */
2114 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
2115 if (ifa->ifa_addr->sa_family == AF_INET6) {
2120 /* should we care about ia6_flags? */
2124 * This can still happen, when, for example, we receive an RA
2125 * containing a prefix with the L bit set and the A bit clear,
2126 * after removing all IPv6 addresses on the receiving
2127 * interface. This should, of course, be rare though.
2130 "%s: failed to find any ifaddr to add route for a "
2131 "prefix(%s/%d) on %s\n", __func__,
2132 ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
2133 pr->ndpr_plen, if_name(ifp)));
2136 error = nd6_prefix_onlink_rtrequest(pr, ifa);
2145 nd6_prefix_offlink(struct nd_prefix *pr)
2148 struct ifnet *ifp = pr->ndpr_ifp;
2149 struct nd_prefix *opr;
2150 char ip6buf[INET6_ADDRSTRLEN];
2154 ND6_ONLINK_LOCK_ASSERT();
2155 ND6_UNLOCK_ASSERT();
2157 if ((pr->ndpr_stateflags & NDPRF_ONLINK) == 0)
2160 struct sockaddr_in6 mask6 = {
2161 .sin6_family = AF_INET6,
2162 .sin6_len = sizeof(struct sockaddr_in6),
2163 .sin6_addr = pr->ndpr_mask,
2165 struct sockaddr_in6 *pmask6 = (pr->ndpr_plen != 128) ? &mask6 : NULL;
2167 error = nd6_prefix_rtrequest(ifp->if_fib, RTM_DELETE,
2168 &pr->ndpr_prefix, pmask6, ifp, NULL);
2172 pr->ndpr_stateflags &= ~NDPRF_ONLINK;
2175 * There might be the same prefix on another interface,
2176 * the prefix which could not be on-link just because we have
2177 * the interface route (see comments in nd6_prefix_onlink).
2178 * If there's one, try to make the prefix on-link on the
2183 LIST_FOREACH(opr, &V_nd_prefix, ndpr_entry) {
2185 * KAME specific: detached prefixes should not be
2188 if (opr == pr || (opr->ndpr_stateflags &
2189 (NDPRF_ONLINK | NDPRF_DETACHED)) != 0)
2192 if (opr->ndpr_plen == pr->ndpr_plen &&
2193 in6_are_prefix_equal(&pr->ndpr_prefix.sin6_addr,
2194 &opr->ndpr_prefix.sin6_addr, pr->ndpr_plen)) {
2197 genid = V_nd6_list_genid;
2199 if ((e = nd6_prefix_onlink(opr)) != 0) {
2201 "%s: failed to recover a prefix "
2202 "%s/%d from %s to %s (errno=%d)\n",
2203 __func__, ip6_sprintf(ip6buf,
2204 &opr->ndpr_prefix.sin6_addr),
2205 opr->ndpr_plen, if_name(ifp),
2206 if_name(opr->ndpr_ifp), e));
2210 if (genid != V_nd6_list_genid)
2216 /* XXX: can we still set the NDPRF_ONLINK flag? */
2218 "%s: failed to delete route: %s/%d on %s (errno=%d)\n",
2219 __func__, ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
2220 pr->ndpr_plen, if_name(ifp), error));
2224 lltable_prefix_free(AF_INET6,
2225 (struct sockaddr *)&pr->ndpr_prefix,
2226 (struct sockaddr *)&mask6, LLE_STATIC);
2232 * ia0 - corresponding public address
2235 in6_tmpifadd(const struct in6_ifaddr *ia0, int forcegen, int delay)
2237 struct ifnet *ifp = ia0->ia_ifa.ifa_ifp;
2238 struct in6_ifaddr *newia;
2239 struct in6_aliasreq ifra;
2241 int trylimit = 3; /* XXX: adhoc value */
2243 u_int32_t randid[2];
2244 time_t vltime0, pltime0;
2246 in6_prepare_ifra(&ifra, &ia0->ia_addr.sin6_addr,
2247 &ia0->ia_prefixmask.sin6_addr);
2249 ifra.ifra_addr = ia0->ia_addr; /* XXX: do we need this ? */
2250 /* clear the old IFID */
2251 IN6_MASK_ADDR(&ifra.ifra_addr.sin6_addr,
2252 &ifra.ifra_prefixmask.sin6_addr);
2255 if (in6_get_tmpifid(ifp, (u_int8_t *)randid,
2256 (const u_int8_t *)&ia0->ia_addr.sin6_addr.s6_addr[8], forcegen)) {
2257 nd6log((LOG_NOTICE, "%s: failed to find a good random IFID\n",
2261 ifra.ifra_addr.sin6_addr.s6_addr32[2] |=
2262 (randid[0] & ~(ifra.ifra_prefixmask.sin6_addr.s6_addr32[2]));
2263 ifra.ifra_addr.sin6_addr.s6_addr32[3] |=
2264 (randid[1] & ~(ifra.ifra_prefixmask.sin6_addr.s6_addr32[3]));
2267 * in6_get_tmpifid() quite likely provided a unique interface ID.
2268 * However, we may still have a chance to see collision, because
2269 * there may be a time lag between generation of the ID and generation
2270 * of the address. So, we'll do one more sanity check.
2273 if (in6_localip(&ifra.ifra_addr.sin6_addr) != 0) {
2274 if (trylimit-- > 0) {
2279 /* Give up. Something strange should have happened. */
2280 nd6log((LOG_NOTICE, "%s: failed to find a unique random IFID\n",
2286 * The Valid Lifetime is the lower of the Valid Lifetime of the
2287 * public address or TEMP_VALID_LIFETIME.
2288 * The Preferred Lifetime is the lower of the Preferred Lifetime
2289 * of the public address or TEMP_PREFERRED_LIFETIME -
2292 if (ia0->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
2293 vltime0 = IFA6_IS_INVALID(ia0) ? 0 :
2294 (ia0->ia6_lifetime.ia6t_vltime -
2295 (time_uptime - ia0->ia6_updatetime));
2296 if (vltime0 > V_ip6_temp_valid_lifetime)
2297 vltime0 = V_ip6_temp_valid_lifetime;
2299 vltime0 = V_ip6_temp_valid_lifetime;
2300 if (ia0->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
2301 pltime0 = IFA6_IS_DEPRECATED(ia0) ? 0 :
2302 (ia0->ia6_lifetime.ia6t_pltime -
2303 (time_uptime - ia0->ia6_updatetime));
2304 if (pltime0 > V_ip6_temp_preferred_lifetime - V_ip6_desync_factor){
2305 pltime0 = V_ip6_temp_preferred_lifetime -
2306 V_ip6_desync_factor;
2309 pltime0 = V_ip6_temp_preferred_lifetime - V_ip6_desync_factor;
2310 ifra.ifra_lifetime.ia6t_vltime = vltime0;
2311 ifra.ifra_lifetime.ia6t_pltime = pltime0;
2314 * A temporary address is created only if this calculated Preferred
2315 * Lifetime is greater than REGEN_ADVANCE time units.
2317 if (ifra.ifra_lifetime.ia6t_pltime <= V_ip6_temp_regen_advance)
2320 /* XXX: scope zone ID? */
2322 ifra.ifra_flags |= (IN6_IFF_AUTOCONF|IN6_IFF_TEMPORARY);
2324 /* allocate ifaddr structure, link into chain, etc. */
2327 updateflags |= IN6_IFAUPDATE_DADDELAY;
2328 if ((error = in6_update_ifa(ifp, &ifra, NULL, updateflags)) != 0)
2331 newia = in6ifa_ifpwithaddr(ifp, &ifra.ifra_addr.sin6_addr);
2332 if (newia == NULL) { /* XXX: can it happen? */
2334 "%s: ifa update succeeded, but we got no ifaddr\n",
2336 return (EINVAL); /* XXX */
2338 newia->ia6_ndpr = ia0->ia6_ndpr;
2339 newia->ia6_ndpr->ndpr_addrcnt++;
2340 ifa_free(&newia->ia_ifa);
2343 * A newly added address might affect the status of other addresses.
2344 * XXX: when the temporary address is generated with a new public
2345 * address, the onlink check is redundant. However, it would be safe
2346 * to do the check explicitly everywhere a new address is generated,
2347 * and, in fact, we surely need the check when we create a new
2348 * temporary address due to deprecation of an old temporary address.
2350 pfxlist_onlink_check();
2356 rt6_deleteroute(const struct rtentry *rt, const struct nhop_object *nh,
2359 struct in6_addr *gate = (struct in6_addr *)arg;
2362 if (nh->gw_sa.sa_family != AF_INET6)
2365 if (!IN6_ARE_ADDR_EQUAL(gate, &nh->gw6_sa.sin6_addr)) {
2370 * Do not delete a static route.
2371 * XXX: this seems to be a bit ad-hoc. Should we consider the
2372 * 'cloned' bit instead?
2374 nh_rt_flags = nhop_get_rtflags(nh);
2375 if ((nh_rt_flags & RTF_STATIC) != 0)
2379 * We delete only host route. This means, in particular, we don't
2380 * delete default route.
2382 if ((nh_rt_flags & RTF_HOST) == 0)
2390 * Delete all the routing table entries that use the specified gateway.
2391 * XXX: this function causes search through all entries of routing table, so
2392 * it shouldn't be called when acting as a router.
2395 rt6_flush(struct in6_addr *gateway, struct ifnet *ifp)
2398 /* We'll care only link-local addresses */
2399 if (!IN6_IS_ADDR_LINKLOCAL(gateway))
2402 /* XXX Do we really need to walk any but the default FIB? */
2403 rib_foreach_table_walk_del(AF_INET6, rt6_deleteroute, (void *)gateway);
2407 nd6_setdefaultiface(int ifindex)
2410 if (V_nd6_defifindex != ifindex) {
2411 V_nd6_defifindex = ifindex;
2412 if (V_nd6_defifindex != 0) {
2413 struct epoch_tracker et;
2416 * XXXGL: this function should use ifnet_byindex_ref!
2418 NET_EPOCH_ENTER(et);
2419 V_nd6_defifp = ifnet_byindex(V_nd6_defifindex);
2421 if (V_nd6_defifp == NULL)
2424 V_nd6_defifp = NULL;
2427 * Our current implementation assumes one-to-one mapping between
2428 * interfaces and links, so it would be natural to use the
2429 * default interface as the default link.
2431 scope6_setdefault(V_nd6_defifp);
2438 nd6_defrouter_list_empty(void)
2441 return (TAILQ_EMPTY(&V_nd6_defrouter));
2445 nd6_defrouter_timer(void)
2447 struct nd_defrouter *dr, *ndr;
2448 struct nd6_drhead drq;
2453 TAILQ_FOREACH_SAFE(dr, &V_nd6_defrouter, dr_entry, ndr)
2454 if (dr->expire && dr->expire < time_uptime)
2455 defrouter_unlink(dr, &drq);
2458 while ((dr = TAILQ_FIRST(&drq)) != NULL) {
2459 TAILQ_REMOVE(&drq, dr, dr_entry);
2465 * Nuke default router list entries toward ifp.
2466 * We defer removal of default router list entries that is installed in the
2467 * routing table, in order to keep additional side effects as small as possible.
2470 nd6_defrouter_purge(struct ifnet *ifp)
2472 struct nd_defrouter *dr, *ndr;
2473 struct nd6_drhead drq;
2478 TAILQ_FOREACH_SAFE(dr, &V_nd6_defrouter, dr_entry, ndr) {
2482 defrouter_unlink(dr, &drq);
2484 TAILQ_FOREACH_SAFE(dr, &V_nd6_defrouter, dr_entry, ndr) {
2488 defrouter_unlink(dr, &drq);
2492 /* Delete the unlinked router objects. */
2493 while ((dr = TAILQ_FIRST(&drq)) != NULL) {
2494 TAILQ_REMOVE(&drq, dr, dr_entry);
2500 nd6_defrouter_flush_all(void)
2502 struct nd_defrouter *dr;
2503 struct nd6_drhead drq;
2508 while ((dr = TAILQ_FIRST(&V_nd6_defrouter)) != NULL)
2509 defrouter_unlink(dr, &drq);
2512 while ((dr = TAILQ_FIRST(&drq)) != NULL) {
2513 TAILQ_REMOVE(&drq, dr, dr_entry);
2519 nd6_defrouter_init(void)
2522 TAILQ_INIT(&V_nd6_defrouter);
2526 nd6_sysctl_drlist(SYSCTL_HANDLER_ARGS)
2528 struct in6_defrouter d;
2529 struct nd_defrouter *dr;
2532 if (req->newptr != NULL)
2535 error = sysctl_wire_old_buffer(req, 0);
2539 bzero(&d, sizeof(d));
2540 d.rtaddr.sin6_family = AF_INET6;
2541 d.rtaddr.sin6_len = sizeof(d.rtaddr);
2544 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry) {
2545 d.rtaddr.sin6_addr = dr->rtaddr;
2546 error = sa6_recoverscope(&d.rtaddr);
2549 d.flags = dr->raflags;
2550 d.rtlifetime = dr->rtlifetime;
2551 d.expire = dr->expire + (time_second - time_uptime);
2552 d.if_index = dr->ifp->if_index;
2553 error = SYSCTL_OUT(req, &d, sizeof(d));
2560 SYSCTL_PROC(_net_inet6_icmp6, ICMPV6CTL_ND6_DRLIST, nd6_drlist,
2561 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_MPSAFE,
2562 NULL, 0, nd6_sysctl_drlist, "S,in6_defrouter",
2563 "NDP default router list");