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_types.h>
60 #include <net/if_dl.h>
61 #include <net/route.h>
62 #include <net/route_var.h>
63 #include <net/radix.h>
66 #include <netinet/in.h>
67 #include <net/if_llatbl.h>
68 #include <netinet6/in6_var.h>
69 #include <netinet6/in6_ifattach.h>
70 #include <netinet/ip6.h>
71 #include <netinet6/ip6_var.h>
72 #include <netinet6/nd6.h>
73 #include <netinet/icmp6.h>
74 #include <netinet6/scope6_var.h>
76 static struct nd_defrouter *defrtrlist_update(struct nd_defrouter *);
77 static int prelist_update(struct nd_prefixctl *, struct nd_defrouter *,
80 VNET_DEFINE_STATIC(struct nd_drhead, nd6_defrouter);
81 #define V_nd6_defrouter VNET(nd6_defrouter)
83 VNET_DECLARE(int, nd6_recalc_reachtm_interval);
84 #define V_nd6_recalc_reachtm_interval VNET(nd6_recalc_reachtm_interval)
86 VNET_DEFINE_STATIC(struct ifnet *, nd6_defifp);
87 VNET_DEFINE(int, nd6_defifindex);
88 #define V_nd6_defifp VNET(nd6_defifp)
90 VNET_DEFINE(int, ip6_use_tempaddr) = 0;
92 VNET_DEFINE(int, ip6_desync_factor);
93 VNET_DEFINE(u_int32_t, ip6_temp_preferred_lifetime) = DEF_TEMP_PREFERRED_LIFETIME;
94 VNET_DEFINE(u_int32_t, ip6_temp_valid_lifetime) = DEF_TEMP_VALID_LIFETIME;
96 VNET_DEFINE(int, ip6_temp_regen_advance) = TEMPADDR_REGEN_ADVANCE;
98 SYSCTL_DECL(_net_inet6_icmp6);
100 /* RTPREF_MEDIUM has to be 0! */
101 #define RTPREF_HIGH 1
102 #define RTPREF_MEDIUM 0
103 #define RTPREF_LOW (-1)
104 #define RTPREF_RESERVED (-2)
105 #define RTPREF_INVALID (-3) /* internal */
108 defrouter_ref(struct nd_defrouter *dr)
111 refcount_acquire(&dr->refcnt);
115 defrouter_rele(struct nd_defrouter *dr)
118 if (refcount_release(&dr->refcnt))
123 * Remove a router from the global list and optionally stash it in a
124 * caller-supplied queue.
127 defrouter_unlink(struct nd_defrouter *dr, struct nd_drhead *drq)
132 TAILQ_REMOVE(&V_nd6_defrouter, dr, dr_entry);
135 TAILQ_INSERT_TAIL(drq, dr, dr_entry);
139 * Receive Router Solicitation Message - just for routers.
140 * Router solicitation/advertisement is mostly managed by userland program
141 * (rtadvd) so here we have no function like nd6_ra_output().
146 nd6_rs_input(struct mbuf *m, int off, int icmp6len)
150 struct nd_router_solicit *nd_rs;
151 struct in6_addr saddr6;
152 union nd_opts ndopts;
153 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
157 ifp = m->m_pkthdr.rcvif;
160 * Accept RS only when V_ip6_forwarding=1 and the interface has
161 * no ND6_IFF_ACCEPT_RTADV.
163 if (!V_ip6_forwarding || ND_IFINFO(ifp)->flags & ND6_IFF_ACCEPT_RTADV)
166 /* RFC 6980: Nodes MUST silently ignore fragments */
167 if(m->m_flags & M_FRAGMENTED)
171 ip6 = mtod(m, struct ip6_hdr *);
172 if (ip6->ip6_hlim != 255) {
174 "%s: invalid hlim (%d) from %s to %s on %s\n", __func__,
175 ip6->ip6_hlim, ip6_sprintf(ip6bufs, &ip6->ip6_src),
176 ip6_sprintf(ip6bufd, &ip6->ip6_dst), if_name(ifp)));
181 * Don't update the neighbor cache, if src = ::.
182 * This indicates that the src has no IP address assigned yet.
184 saddr6 = ip6->ip6_src;
185 if (IN6_IS_ADDR_UNSPECIFIED(&saddr6))
188 if (m->m_len < off + icmp6len) {
189 m = m_pullup(m, off + icmp6len);
191 IP6STAT_INC(ip6s_exthdrtoolong);
195 ip6 = mtod(m, struct ip6_hdr *);
196 nd_rs = (struct nd_router_solicit *)((caddr_t)ip6 + off);
198 icmp6len -= sizeof(*nd_rs);
199 nd6_option_init(nd_rs + 1, icmp6len, &ndopts);
200 if (nd6_options(&ndopts) < 0) {
202 "%s: invalid ND option, ignored\n", __func__));
203 /* nd6_options have incremented stats */
209 if (ndopts.nd_opts_src_lladdr) {
210 lladdr = (char *)(ndopts.nd_opts_src_lladdr + 1);
211 lladdrlen = ndopts.nd_opts_src_lladdr->nd_opt_len << 3;
214 if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) {
216 "%s: lladdrlen mismatch for %s (if %d, RS packet %d)\n",
217 __func__, ip6_sprintf(ip6bufs, &saddr6),
218 ifp->if_addrlen, lladdrlen - 2));
222 nd6_cache_lladdr(ifp, &saddr6, lladdr, lladdrlen, ND_ROUTER_SOLICIT, 0);
229 ICMP6STAT_INC(icp6s_badrs);
234 * Receive Router Advertisement Message.
237 * TODO: on-link bit on prefix information
238 * TODO: ND_RA_FLAG_{OTHER,MANAGED} processing
241 nd6_ra_input(struct mbuf *m, int off, int icmp6len)
244 struct nd_ifinfo *ndi;
246 struct nd_router_advert *nd_ra;
247 struct in6_addr saddr6;
248 struct nd_defrouter *dr;
249 union nd_opts ndopts;
250 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
254 * We only accept RAs only when the per-interface flag
255 * ND6_IFF_ACCEPT_RTADV is on the receiving interface.
257 ifp = m->m_pkthdr.rcvif;
258 ndi = ND_IFINFO(ifp);
259 if (!(ndi->flags & ND6_IFF_ACCEPT_RTADV))
262 /* RFC 6980: Nodes MUST silently ignore fragments */
263 if(m->m_flags & M_FRAGMENTED)
266 ip6 = mtod(m, struct ip6_hdr *);
267 if (ip6->ip6_hlim != 255) {
269 "%s: invalid hlim (%d) from %s to %s on %s\n", __func__,
270 ip6->ip6_hlim, ip6_sprintf(ip6bufs, &ip6->ip6_src),
271 ip6_sprintf(ip6bufd, &ip6->ip6_dst), if_name(ifp)));
275 saddr6 = ip6->ip6_src;
276 if (!IN6_IS_ADDR_LINKLOCAL(&saddr6)) {
278 "%s: src %s is not link-local\n", __func__,
279 ip6_sprintf(ip6bufs, &saddr6)));
283 if (m->m_len < off + icmp6len) {
284 m = m_pullup(m, off + icmp6len);
286 IP6STAT_INC(ip6s_exthdrtoolong);
290 ip6 = mtod(m, struct ip6_hdr *);
291 nd_ra = (struct nd_router_advert *)((caddr_t)ip6 + off);
293 icmp6len -= sizeof(*nd_ra);
294 nd6_option_init(nd_ra + 1, icmp6len, &ndopts);
295 if (nd6_options(&ndopts) < 0) {
297 "%s: invalid ND option, ignored\n", __func__));
298 /* nd6_options have incremented stats */
305 struct nd_defrouter dr0;
306 u_int32_t advreachable = nd_ra->nd_ra_reachable;
308 /* remember if this is a multicasted advertisement */
309 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst))
312 bzero(&dr0, sizeof(dr0));
314 dr0.raflags = nd_ra->nd_ra_flags_reserved;
316 * Effectively-disable routes from RA messages when
317 * ND6_IFF_NO_RADR enabled on the receiving interface or
318 * (ip6.forwarding == 1 && ip6.rfc6204w3 != 1).
320 if (ndi->flags & ND6_IFF_NO_RADR)
322 else if (V_ip6_forwarding && !V_ip6_rfc6204w3)
325 dr0.rtlifetime = ntohs(nd_ra->nd_ra_router_lifetime);
326 dr0.expire = time_uptime + dr0.rtlifetime;
328 /* unspecified or not? (RFC 2461 6.3.4) */
330 advreachable = ntohl(advreachable);
331 if (advreachable <= MAX_REACHABLE_TIME &&
332 ndi->basereachable != advreachable) {
333 ndi->basereachable = advreachable;
334 ndi->reachable = ND_COMPUTE_RTIME(ndi->basereachable);
335 ndi->recalctm = V_nd6_recalc_reachtm_interval; /* reset */
338 if (nd_ra->nd_ra_retransmit)
339 ndi->retrans = ntohl(nd_ra->nd_ra_retransmit);
340 if (nd_ra->nd_ra_curhoplimit) {
341 if (ndi->chlim < nd_ra->nd_ra_curhoplimit)
342 ndi->chlim = nd_ra->nd_ra_curhoplimit;
343 else if (ndi->chlim != nd_ra->nd_ra_curhoplimit) {
344 log(LOG_ERR, "RA with a lower CurHopLimit sent from "
345 "%s on %s (current = %d, received = %d). "
346 "Ignored.\n", ip6_sprintf(ip6bufs, &ip6->ip6_src),
347 if_name(ifp), ndi->chlim, nd_ra->nd_ra_curhoplimit);
350 dr = defrtrlist_update(&dr0);
356 if (ndopts.nd_opts_pi) {
357 struct nd_opt_hdr *pt;
358 struct nd_opt_prefix_info *pi = NULL;
359 struct nd_prefixctl pr;
361 for (pt = (struct nd_opt_hdr *)ndopts.nd_opts_pi;
362 pt <= (struct nd_opt_hdr *)ndopts.nd_opts_pi_end;
363 pt = (struct nd_opt_hdr *)((caddr_t)pt +
364 (pt->nd_opt_len << 3))) {
365 if (pt->nd_opt_type != ND_OPT_PREFIX_INFORMATION)
367 pi = (struct nd_opt_prefix_info *)pt;
369 if (pi->nd_opt_pi_len != 4) {
371 "%s: invalid option len %d for prefix "
372 "information option, ignored\n", __func__,
377 if (128 < pi->nd_opt_pi_prefix_len) {
379 "%s: invalid prefix len %d for prefix "
380 "information option, ignored\n", __func__,
381 pi->nd_opt_pi_prefix_len));
385 if (IN6_IS_ADDR_MULTICAST(&pi->nd_opt_pi_prefix)
386 || IN6_IS_ADDR_LINKLOCAL(&pi->nd_opt_pi_prefix)) {
388 "%s: invalid prefix %s, ignored\n",
389 __func__, ip6_sprintf(ip6bufs,
390 &pi->nd_opt_pi_prefix)));
394 bzero(&pr, sizeof(pr));
395 pr.ndpr_prefix.sin6_family = AF_INET6;
396 pr.ndpr_prefix.sin6_len = sizeof(pr.ndpr_prefix);
397 pr.ndpr_prefix.sin6_addr = pi->nd_opt_pi_prefix;
398 pr.ndpr_ifp = (struct ifnet *)m->m_pkthdr.rcvif;
400 pr.ndpr_raf_onlink = (pi->nd_opt_pi_flags_reserved &
401 ND_OPT_PI_FLAG_ONLINK) ? 1 : 0;
402 pr.ndpr_raf_auto = (pi->nd_opt_pi_flags_reserved &
403 ND_OPT_PI_FLAG_AUTO) ? 1 : 0;
404 pr.ndpr_plen = pi->nd_opt_pi_prefix_len;
405 pr.ndpr_vltime = ntohl(pi->nd_opt_pi_valid_time);
406 pr.ndpr_pltime = ntohl(pi->nd_opt_pi_preferred_time);
407 (void)prelist_update(&pr, dr, m, mcast);
418 if (ndopts.nd_opts_mtu && ndopts.nd_opts_mtu->nd_opt_mtu_len == 1) {
422 mtu = (u_long)ntohl(ndopts.nd_opts_mtu->nd_opt_mtu_mtu);
425 if (mtu < IPV6_MMTU) {
426 nd6log((LOG_INFO, "%s: bogus mtu option mtu=%lu sent "
427 "from %s, ignoring\n", __func__,
428 mtu, ip6_sprintf(ip6bufs, &ip6->ip6_src)));
433 maxmtu = (ndi->maxmtu && ndi->maxmtu < ifp->if_mtu)
434 ? ndi->maxmtu : ifp->if_mtu;
436 int change = (ndi->linkmtu != mtu);
440 /* in6_maxmtu may change */
445 nd6log((LOG_INFO, "%s: bogus mtu=%lu sent from %s; "
446 "exceeds maxmtu %lu, ignoring\n", __func__,
447 mtu, ip6_sprintf(ip6bufs, &ip6->ip6_src), maxmtu));
454 * Source link layer address
460 if (ndopts.nd_opts_src_lladdr) {
461 lladdr = (char *)(ndopts.nd_opts_src_lladdr + 1);
462 lladdrlen = ndopts.nd_opts_src_lladdr->nd_opt_len << 3;
465 if (lladdr && ((ifp->if_addrlen + 2 + 7) & ~7) != lladdrlen) {
467 "%s: lladdrlen mismatch for %s (if %d, RA packet %d)\n",
468 __func__, ip6_sprintf(ip6bufs, &saddr6),
469 ifp->if_addrlen, lladdrlen - 2));
473 nd6_cache_lladdr(ifp, &saddr6, lladdr,
474 lladdrlen, ND_ROUTER_ADVERT, 0);
477 * Installing a link-layer address might change the state of the
478 * router's neighbor cache, which might also affect our on-link
479 * detection of adveritsed prefixes.
481 pfxlist_onlink_check();
489 ICMP6STAT_INC(icp6s_badra);
493 /* tell the change to user processes watching the routing socket. */
495 nd6_rtmsg(int cmd, struct rtentry *rt)
497 struct rt_addrinfo info;
501 bzero((caddr_t)&info, sizeof(info));
502 info.rti_info[RTAX_DST] = rt_key(rt);
503 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
504 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
508 ifa = CK_STAILQ_FIRST(&ifp->if_addrhead);
509 info.rti_info[RTAX_IFP] = ifa->ifa_addr;
511 IF_ADDR_RUNLOCK(ifp);
512 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr;
516 rt_missmsg_fib(cmd, &info, rt->rt_flags, 0, rt->rt_fibnum);
522 static struct nd_pfxrouter *
523 pfxrtr_lookup(struct nd_prefix *pr, struct nd_defrouter *dr)
525 struct nd_pfxrouter *search;
529 LIST_FOREACH(search, &pr->ndpr_advrtrs, pfr_entry) {
530 if (search->router == dr)
537 pfxrtr_add(struct nd_prefix *pr, struct nd_defrouter *dr)
539 struct nd_pfxrouter *new;
545 if (pfxrtr_lookup(pr, dr) != NULL) {
551 new = malloc(sizeof(*new), M_IP6NDP, M_NOWAIT | M_ZERO);
558 if (pfxrtr_lookup(pr, dr) == NULL) {
559 LIST_INSERT_HEAD(&pr->ndpr_advrtrs, new, pfr_entry);
562 /* We lost a race to add the reference. */
570 pfxlist_onlink_check();
574 pfxrtr_del(struct nd_pfxrouter *pfr)
579 LIST_REMOVE(pfr, pfr_entry);
580 defrouter_rele(pfr->router);
585 /* Default router list processing sub routines. */
587 defrouter_addreq(struct nd_defrouter *new)
589 struct sockaddr_in6 def, mask, gate;
590 struct rtentry *newrt = NULL;
593 bzero(&def, sizeof(def));
594 bzero(&mask, sizeof(mask));
595 bzero(&gate, sizeof(gate));
597 def.sin6_len = mask.sin6_len = gate.sin6_len =
598 sizeof(struct sockaddr_in6);
599 def.sin6_family = gate.sin6_family = AF_INET6;
600 gate.sin6_addr = new->rtaddr;
602 error = in6_rtrequest(RTM_ADD, (struct sockaddr *)&def,
603 (struct sockaddr *)&gate, (struct sockaddr *)&mask,
604 RTF_GATEWAY, &newrt, new->ifp->if_fib);
606 nd6_rtmsg(RTM_ADD, newrt); /* tell user process */
614 * Remove the default route for a given router.
615 * This is just a subroutine function for defrouter_select_fib(), and
616 * should not be called from anywhere else.
619 defrouter_delreq(struct nd_defrouter *dr)
621 struct sockaddr_in6 def, mask, gate;
622 struct rtentry *oldrt = NULL;
624 bzero(&def, sizeof(def));
625 bzero(&mask, sizeof(mask));
626 bzero(&gate, sizeof(gate));
628 def.sin6_len = mask.sin6_len = gate.sin6_len =
629 sizeof(struct sockaddr_in6);
630 def.sin6_family = gate.sin6_family = AF_INET6;
631 gate.sin6_addr = dr->rtaddr;
633 in6_rtrequest(RTM_DELETE, (struct sockaddr *)&def,
634 (struct sockaddr *)&gate,
635 (struct sockaddr *)&mask, RTF_GATEWAY, &oldrt, dr->ifp->if_fib);
637 nd6_rtmsg(RTM_DELETE, oldrt);
645 defrouter_del(struct nd_defrouter *dr)
647 struct nd_defrouter *deldr = NULL;
648 struct nd_prefix *pr;
649 struct nd_pfxrouter *pfxrtr;
654 * Flush all the routing table entries that use the router
657 if (ND_IFINFO(dr->ifp)->flags & ND6_IFF_ACCEPT_RTADV)
658 rt6_flush(&dr->rtaddr, dr->ifp);
662 defrouter_delreq(dr);
666 * Also delete all the pointers to the router in each prefix lists.
669 LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
670 if ((pfxrtr = pfxrtr_lookup(pr, dr)) != NULL)
675 pfxlist_onlink_check();
678 * If the router is the primary one, choose a new one.
679 * Note that defrouter_select_fib() will remove the current
680 * gateway from the routing table.
683 defrouter_select_fib(deldr->ifp->if_fib);
686 * Release the list reference.
692 struct nd_defrouter *
693 defrouter_lookup_locked(struct in6_addr *addr, struct ifnet *ifp)
695 struct nd_defrouter *dr;
698 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry)
699 if (dr->ifp == ifp && IN6_ARE_ADDR_EQUAL(addr, &dr->rtaddr)) {
706 struct nd_defrouter *
707 defrouter_lookup(struct in6_addr *addr, struct ifnet *ifp)
709 struct nd_defrouter *dr;
712 dr = defrouter_lookup_locked(addr, ifp);
718 * Remove all default routes from default router list.
721 defrouter_reset(void)
723 struct nd_defrouter *dr, **dra;
729 * We can't delete routes with the ND lock held, so make a copy of the
730 * current default router list and use that when deleting routes.
733 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry)
737 dra = malloc(count * sizeof(*dra), M_TEMP, M_WAITOK | M_ZERO);
740 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry) {
748 for (i = 0; i < count && dra[i] != NULL; i++) {
749 defrouter_delreq(dra[i]);
750 defrouter_rele(dra[i]);
755 * XXX should we also nuke any default routers in the kernel, by
756 * going through them by rtalloc1()?
761 * Look up a matching default router list entry and remove it. Returns true if a
762 * matching entry was found, false otherwise.
765 defrouter_remove(struct in6_addr *addr, struct ifnet *ifp)
767 struct nd_defrouter *dr;
770 dr = defrouter_lookup_locked(addr, ifp);
776 defrouter_unlink(dr, NULL);
784 * for default router selection
785 * regards router-preference field as a 2-bit signed integer
788 rtpref(struct nd_defrouter *dr)
790 switch (dr->raflags & ND_RA_FLAG_RTPREF_MASK) {
791 case ND_RA_FLAG_RTPREF_HIGH:
792 return (RTPREF_HIGH);
793 case ND_RA_FLAG_RTPREF_MEDIUM:
794 case ND_RA_FLAG_RTPREF_RSV:
795 return (RTPREF_MEDIUM);
796 case ND_RA_FLAG_RTPREF_LOW:
800 * This case should never happen. If it did, it would mean a
801 * serious bug of kernel internal. We thus always bark here.
802 * Or, can we even panic?
804 log(LOG_ERR, "rtpref: impossible RA flag %x\n", dr->raflags);
805 return (RTPREF_INVALID);
811 * Default Router Selection according to Section 6.3.6 of RFC 2461 and
812 * draft-ietf-ipngwg-router-selection:
813 * 1) Routers that are reachable or probably reachable should be preferred.
814 * If we have more than one (probably) reachable router, prefer ones
815 * with the highest router preference.
816 * 2) When no routers on the list are known to be reachable or
817 * probably reachable, routers SHOULD be selected in a round-robin
818 * fashion, regardless of router preference values.
819 * 3) If the Default Router List is empty, assume that all
820 * destinations are on-link.
822 * We assume nd_defrouter is sorted by router preference value.
823 * Since the code below covers both with and without router preference cases,
824 * we do not need to classify the cases by ifdef.
826 * At this moment, we do not try to install more than one default router,
827 * even when the multipath routing is available, because we're not sure about
828 * the benefits for stub hosts comparing to the risk of making the code
829 * complicated and the possibility of introducing bugs.
831 * We maintain a single list of routers for multiple FIBs, only considering one
832 * at a time based on the receiving interface's FIB. If @fibnum is RT_ALL_FIBS,
833 * we do the whole thing multiple times.
836 defrouter_select_fib(int fibnum)
838 struct nd_defrouter *dr, *selected_dr, *installed_dr;
839 struct llentry *ln = NULL;
841 if (fibnum == RT_ALL_FIBS) {
842 for (fibnum = 0; fibnum < rt_numfibs; fibnum++) {
843 defrouter_select_fib(fibnum);
849 * Let's handle easy case (3) first:
850 * If default router list is empty, there's nothing to be done.
852 if (TAILQ_EMPTY(&V_nd6_defrouter)) {
858 * Search for a (probably) reachable router from the list.
859 * We just pick up the first reachable one (if any), assuming that
860 * the ordering rule of the list described in defrtrlist_update().
862 selected_dr = installed_dr = NULL;
863 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry) {
864 IF_AFDATA_RLOCK(dr->ifp);
865 if (selected_dr == NULL && dr->ifp->if_fib == fibnum &&
866 (ln = nd6_lookup(&dr->rtaddr, 0, dr->ifp)) &&
867 ND6_IS_LLINFO_PROBREACH(ln)) {
869 defrouter_ref(selected_dr);
871 IF_AFDATA_RUNLOCK(dr->ifp);
877 if (dr->installed && dr->ifp->if_fib == fibnum) {
878 if (installed_dr == NULL) {
880 defrouter_ref(installed_dr);
883 * this should not happen.
884 * warn for diagnosis.
886 log(LOG_ERR, "defrouter_select_fib: more than "
887 "one router is installed\n");
892 * If none of the default routers was found to be reachable,
893 * round-robin the list regardless of preference.
894 * Otherwise, if we have an installed router, check if the selected
895 * (reachable) router should really be preferred to the installed one.
896 * We only prefer the new router when the old one is not reachable
897 * or when the new one has a really higher preference value.
899 if (selected_dr == NULL) {
900 if (installed_dr == NULL ||
901 TAILQ_NEXT(installed_dr, dr_entry) == NULL)
902 dr = TAILQ_FIRST(&V_nd6_defrouter);
904 dr = TAILQ_NEXT(installed_dr, dr_entry);
906 /* Ensure we select a router for this FIB. */
907 TAILQ_FOREACH_FROM(dr, &V_nd6_defrouter, dr_entry) {
908 if (dr->ifp->if_fib == fibnum) {
910 defrouter_ref(selected_dr);
914 } else if (installed_dr != NULL) {
915 IF_AFDATA_RLOCK(installed_dr->ifp);
916 if ((ln = nd6_lookup(&installed_dr->rtaddr, 0,
917 installed_dr->ifp)) &&
918 ND6_IS_LLINFO_PROBREACH(ln) &&
919 installed_dr->ifp->if_fib == fibnum &&
920 rtpref(selected_dr) <= rtpref(installed_dr)) {
921 defrouter_rele(selected_dr);
922 selected_dr = installed_dr;
924 IF_AFDATA_RUNLOCK(installed_dr->ifp);
931 * If we selected a router for this FIB and it's different
932 * than the installed one, remove the installed router and
933 * install the selected one in its place.
935 if (installed_dr != selected_dr) {
936 if (installed_dr != NULL) {
937 defrouter_delreq(installed_dr);
938 defrouter_rele(installed_dr);
940 if (selected_dr != NULL)
941 defrouter_addreq(selected_dr);
943 if (selected_dr != NULL)
944 defrouter_rele(selected_dr);
948 * Maintain old KPI for default router selection.
949 * If unspecified, we can re-select routers for all FIBs.
952 defrouter_select(void)
954 defrouter_select_fib(RT_ALL_FIBS);
957 static struct nd_defrouter *
958 defrtrlist_update(struct nd_defrouter *new)
960 struct nd_defrouter *dr, *n;
965 if (new->rtlifetime == 0) {
966 defrouter_remove(&new->rtaddr, new->ifp);
973 dr = defrouter_lookup_locked(&new->rtaddr, new->ifp);
975 oldpref = rtpref(dr);
978 dr->raflags = new->raflags; /* XXX flag check */
979 dr->rtlifetime = new->rtlifetime;
980 dr->expire = new->expire;
983 * If the preference does not change, there's no need
984 * to sort the entries. Also make sure the selected
985 * router is still installed in the kernel.
987 if (dr->installed && rtpref(new) == oldpref) {
997 * The router needs to be reinserted into the default router
998 * list, so upgrade to a write lock. If that fails and the list
999 * has potentially changed while the lock was dropped, we'll
1000 * redo the lookup with the write lock held.
1004 if (!ND6_TRY_UPGRADE()) {
1005 genid = V_nd6_list_genid;
1008 if (genid != V_nd6_list_genid)
1015 * The preferred router may have changed, so relocate this
1018 TAILQ_REMOVE(&V_nd6_defrouter, dr, dr_entry);
1021 n = malloc(sizeof(*n), M_IP6NDP, M_NOWAIT | M_ZERO);
1026 memcpy(n, new, sizeof(*n));
1027 /* Initialize with an extra reference for the caller. */
1028 refcount_init(&n->refcnt, 2);
1032 * Insert the new router in the Default Router List;
1033 * The Default Router List should be in the descending order
1034 * of router-preferece. Routers with the same preference are
1035 * sorted in the arriving time order.
1038 /* insert at the end of the group */
1039 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry) {
1040 if (rtpref(n) > rtpref(dr))
1044 TAILQ_INSERT_BEFORE(dr, n, dr_entry);
1046 TAILQ_INSERT_TAIL(&V_nd6_defrouter, n, dr_entry);
1050 defrouter_select_fib(new->ifp->if_fib);
1056 in6_init_prefix_ltimes(struct nd_prefix *ndpr)
1058 if (ndpr->ndpr_pltime == ND6_INFINITE_LIFETIME)
1059 ndpr->ndpr_preferred = 0;
1061 ndpr->ndpr_preferred = time_uptime + ndpr->ndpr_pltime;
1062 if (ndpr->ndpr_vltime == ND6_INFINITE_LIFETIME)
1063 ndpr->ndpr_expire = 0;
1065 ndpr->ndpr_expire = time_uptime + ndpr->ndpr_vltime;
1071 in6_init_address_ltimes(struct nd_prefix *new, struct in6_addrlifetime *lt6)
1073 /* init ia6t_expire */
1074 if (lt6->ia6t_vltime == ND6_INFINITE_LIFETIME)
1075 lt6->ia6t_expire = 0;
1077 lt6->ia6t_expire = time_uptime;
1078 lt6->ia6t_expire += lt6->ia6t_vltime;
1081 /* init ia6t_preferred */
1082 if (lt6->ia6t_pltime == ND6_INFINITE_LIFETIME)
1083 lt6->ia6t_preferred = 0;
1085 lt6->ia6t_preferred = time_uptime;
1086 lt6->ia6t_preferred += lt6->ia6t_pltime;
1090 static struct in6_ifaddr *
1091 in6_ifadd(struct nd_prefixctl *pr, int mcast)
1093 struct ifnet *ifp = pr->ndpr_ifp;
1095 struct in6_aliasreq ifra;
1096 struct in6_ifaddr *ia, *ib;
1098 struct in6_addr mask;
1099 int prefixlen = pr->ndpr_plen;
1101 char ip6buf[INET6_ADDRSTRLEN];
1103 in6_prefixlen2mask(&mask, prefixlen);
1106 * find a link-local address (will be interface ID).
1107 * Is it really mandatory? Theoretically, a global or a site-local
1108 * address can be configured without a link-local address, if we
1109 * have a unique interface identifier...
1111 * it is not mandatory to have a link-local address, we can generate
1112 * interface identifier on the fly. we do this because:
1113 * (1) it should be the easiest way to find interface identifier.
1114 * (2) RFC2462 5.4 suggesting the use of the same interface identifier
1115 * for multiple addresses on a single interface, and possible shortcut
1116 * of DAD. we omitted DAD for this reason in the past.
1117 * (3) a user can prevent autoconfiguration of global address
1118 * by removing link-local address by hand (this is partly because we
1119 * don't have other way to control the use of IPv6 on an interface.
1120 * this has been our design choice - cf. NRL's "ifconfig auto").
1121 * (4) it is easier to manage when an interface has addresses
1122 * with the same interface identifier, than to have multiple addresses
1123 * with different interface identifiers.
1125 ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp, 0); /* 0 is OK? */
1127 ib = (struct in6_ifaddr *)ifa;
1131 /* prefixlen + ifidlen must be equal to 128 */
1132 plen0 = in6_mask2len(&ib->ia_prefixmask.sin6_addr, NULL);
1133 if (prefixlen != plen0) {
1136 "%s: wrong prefixlen for %s (prefix=%d ifid=%d)\n",
1137 __func__, if_name(ifp), prefixlen, 128 - plen0));
1142 in6_prepare_ifra(&ifra, &pr->ndpr_prefix.sin6_addr, &mask);
1144 IN6_MASK_ADDR(&ifra.ifra_addr.sin6_addr, &mask);
1146 ifra.ifra_addr.sin6_addr.s6_addr32[0] |=
1147 (ib->ia_addr.sin6_addr.s6_addr32[0] & ~mask.s6_addr32[0]);
1148 ifra.ifra_addr.sin6_addr.s6_addr32[1] |=
1149 (ib->ia_addr.sin6_addr.s6_addr32[1] & ~mask.s6_addr32[1]);
1150 ifra.ifra_addr.sin6_addr.s6_addr32[2] |=
1151 (ib->ia_addr.sin6_addr.s6_addr32[2] & ~mask.s6_addr32[2]);
1152 ifra.ifra_addr.sin6_addr.s6_addr32[3] |=
1153 (ib->ia_addr.sin6_addr.s6_addr32[3] & ~mask.s6_addr32[3]);
1157 ifra.ifra_lifetime.ia6t_vltime = pr->ndpr_vltime;
1158 ifra.ifra_lifetime.ia6t_pltime = pr->ndpr_pltime;
1160 /* XXX: scope zone ID? */
1162 ifra.ifra_flags |= IN6_IFF_AUTOCONF; /* obey autoconf */
1165 * Make sure that we do not have this address already. This should
1166 * usually not happen, but we can still see this case, e.g., if we
1167 * have manually configured the exact address to be configured.
1169 ifa = (struct ifaddr *)in6ifa_ifpwithaddr(ifp,
1170 &ifra.ifra_addr.sin6_addr);
1173 /* this should be rare enough to make an explicit log */
1174 log(LOG_INFO, "in6_ifadd: %s is already configured\n",
1175 ip6_sprintf(ip6buf, &ifra.ifra_addr.sin6_addr));
1180 * Allocate ifaddr structure, link into chain, etc.
1181 * If we are going to create a new address upon receiving a multicasted
1182 * RA, we need to impose a random delay before starting DAD.
1183 * [draft-ietf-ipv6-rfc2462bis-02.txt, Section 5.4.2]
1187 updateflags |= IN6_IFAUPDATE_DADDELAY;
1188 if ((error = in6_update_ifa(ifp, &ifra, NULL, updateflags)) != 0) {
1190 "%s: failed to make ifaddr %s on %s (errno=%d)\n", __func__,
1191 ip6_sprintf(ip6buf, &ifra.ifra_addr.sin6_addr),
1192 if_name(ifp), error));
1193 return (NULL); /* ifaddr must not have been allocated. */
1196 ia = in6ifa_ifpwithaddr(ifp, &ifra.ifra_addr.sin6_addr);
1198 * XXXRW: Assumption of non-NULLness here might not be true with
1199 * fine-grained locking -- should we validate it? Or just return
1200 * earlier ifa rather than looking it up again?
1202 return (ia); /* this is always non-NULL and referenced. */
1205 static struct nd_prefix *
1206 nd6_prefix_lookup_locked(struct nd_prefixctl *key)
1208 struct nd_prefix *search;
1212 LIST_FOREACH(search, &V_nd_prefix, ndpr_entry) {
1213 if (key->ndpr_ifp == search->ndpr_ifp &&
1214 key->ndpr_plen == search->ndpr_plen &&
1215 in6_are_prefix_equal(&key->ndpr_prefix.sin6_addr,
1216 &search->ndpr_prefix.sin6_addr, key->ndpr_plen)) {
1217 nd6_prefix_ref(search);
1225 nd6_prefix_lookup(struct nd_prefixctl *key)
1227 struct nd_prefix *search;
1230 search = nd6_prefix_lookup_locked(key);
1236 nd6_prefix_ref(struct nd_prefix *pr)
1239 refcount_acquire(&pr->ndpr_refcnt);
1243 nd6_prefix_rele(struct nd_prefix *pr)
1246 if (refcount_release(&pr->ndpr_refcnt)) {
1247 KASSERT(LIST_EMPTY(&pr->ndpr_advrtrs),
1248 ("prefix %p has advertising routers", pr));
1254 nd6_prelist_add(struct nd_prefixctl *pr, struct nd_defrouter *dr,
1255 struct nd_prefix **newp)
1257 struct nd_prefix *new;
1258 char ip6buf[INET6_ADDRSTRLEN];
1261 new = malloc(sizeof(*new), M_IP6NDP, M_NOWAIT | M_ZERO);
1264 refcount_init(&new->ndpr_refcnt, newp != NULL ? 2 : 1);
1265 new->ndpr_ifp = pr->ndpr_ifp;
1266 new->ndpr_prefix = pr->ndpr_prefix;
1267 new->ndpr_plen = pr->ndpr_plen;
1268 new->ndpr_vltime = pr->ndpr_vltime;
1269 new->ndpr_pltime = pr->ndpr_pltime;
1270 new->ndpr_flags = pr->ndpr_flags;
1271 if ((error = in6_init_prefix_ltimes(new)) != 0) {
1272 free(new, M_IP6NDP);
1275 new->ndpr_lastupdate = time_uptime;
1277 /* initialization */
1278 LIST_INIT(&new->ndpr_advrtrs);
1279 in6_prefixlen2mask(&new->ndpr_mask, new->ndpr_plen);
1280 /* make prefix in the canonical form */
1281 IN6_MASK_ADDR(&new->ndpr_prefix.sin6_addr, &new->ndpr_mask);
1284 LIST_INSERT_HEAD(&V_nd_prefix, new, ndpr_entry);
1288 /* ND_OPT_PI_FLAG_ONLINK processing */
1289 if (new->ndpr_raf_onlink) {
1291 if ((error = nd6_prefix_onlink(new)) != 0) {
1292 nd6log((LOG_ERR, "%s: failed to make the prefix %s/%d "
1293 "on-link on %s (errno=%d)\n", __func__,
1294 ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
1295 pr->ndpr_plen, if_name(pr->ndpr_ifp), error));
1296 /* proceed anyway. XXX: is it correct? */
1298 ND6_ONLINK_UNLOCK();
1302 pfxrtr_add(new, dr);
1309 * Remove a prefix from the prefix list and optionally stash it in a
1310 * caller-provided list.
1312 * The ND6 lock must be held.
1315 nd6_prefix_unlink(struct nd_prefix *pr, struct nd_prhead *list)
1320 LIST_REMOVE(pr, ndpr_entry);
1323 LIST_INSERT_HEAD(list, pr, ndpr_entry);
1327 * Free an unlinked prefix, first marking it off-link if necessary.
1330 nd6_prefix_del(struct nd_prefix *pr)
1332 struct nd_pfxrouter *pfr, *next;
1334 char ip6buf[INET6_ADDRSTRLEN];
1336 KASSERT(pr->ndpr_addrcnt == 0,
1337 ("prefix %p has referencing addresses", pr));
1338 ND6_UNLOCK_ASSERT();
1341 * Though these flags are now meaningless, we'd rather keep the value
1342 * of pr->ndpr_raf_onlink and pr->ndpr_raf_auto not to confuse users
1343 * when executing "ndp -p".
1345 if ((pr->ndpr_stateflags & NDPRF_ONLINK) != 0) {
1347 if ((e = nd6_prefix_offlink(pr)) != 0) {
1349 "%s: failed to make the prefix %s/%d offlink on %s "
1350 "(errno=%d)\n", __func__,
1351 ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
1352 pr->ndpr_plen, if_name(pr->ndpr_ifp), e));
1353 /* what should we do? */
1355 ND6_ONLINK_UNLOCK();
1358 /* Release references to routers that have advertised this prefix. */
1360 LIST_FOREACH_SAFE(pfr, &pr->ndpr_advrtrs, pfr_entry, next)
1364 nd6_prefix_rele(pr);
1366 pfxlist_onlink_check();
1370 prelist_update(struct nd_prefixctl *new, struct nd_defrouter *dr,
1371 struct mbuf *m, int mcast)
1373 struct in6_ifaddr *ia6 = NULL, *ia6_match = NULL;
1375 struct ifnet *ifp = new->ndpr_ifp;
1376 struct nd_prefix *pr;
1379 struct in6_addrlifetime lt6_tmp;
1380 char ip6buf[INET6_ADDRSTRLEN];
1385 * Authenticity for NA consists authentication for
1386 * both IP header and IP datagrams, doesn't it ?
1388 #if defined(M_AUTHIPHDR) && defined(M_AUTHIPDGM)
1389 auth = ((m->m_flags & M_AUTHIPHDR) &&
1390 (m->m_flags & M_AUTHIPDGM));
1394 if ((pr = nd6_prefix_lookup(new)) != NULL) {
1396 * nd6_prefix_lookup() ensures that pr and new have the same
1397 * prefix on a same interface.
1401 * Update prefix information. Note that the on-link (L) bit
1402 * and the autonomous (A) bit should NOT be changed from 1
1405 if (new->ndpr_raf_onlink == 1)
1406 pr->ndpr_raf_onlink = 1;
1407 if (new->ndpr_raf_auto == 1)
1408 pr->ndpr_raf_auto = 1;
1409 if (new->ndpr_raf_onlink) {
1410 pr->ndpr_vltime = new->ndpr_vltime;
1411 pr->ndpr_pltime = new->ndpr_pltime;
1412 (void)in6_init_prefix_ltimes(pr); /* XXX error case? */
1413 pr->ndpr_lastupdate = time_uptime;
1416 if (new->ndpr_raf_onlink &&
1417 (pr->ndpr_stateflags & NDPRF_ONLINK) == 0) {
1419 if ((error = nd6_prefix_onlink(pr)) != 0) {
1421 "%s: failed to make the prefix %s/%d "
1422 "on-link on %s (errno=%d)\n", __func__,
1424 &pr->ndpr_prefix.sin6_addr),
1425 pr->ndpr_plen, if_name(pr->ndpr_ifp),
1427 /* proceed anyway. XXX: is it correct? */
1429 ND6_ONLINK_UNLOCK();
1435 if (new->ndpr_vltime == 0)
1437 if (new->ndpr_raf_onlink == 0 && new->ndpr_raf_auto == 0)
1440 error = nd6_prelist_add(new, dr, &pr);
1442 nd6log((LOG_NOTICE, "%s: nd6_prelist_add() failed for "
1443 "the prefix %s/%d on %s (errno=%d)\n", __func__,
1444 ip6_sprintf(ip6buf, &new->ndpr_prefix.sin6_addr),
1445 new->ndpr_plen, if_name(new->ndpr_ifp), error));
1446 goto end; /* we should just give up in this case. */
1450 * XXX: from the ND point of view, we can ignore a prefix
1451 * with the on-link bit being zero. However, we need a
1452 * prefix structure for references from autoconfigured
1453 * addresses. Thus, we explicitly make sure that the prefix
1454 * itself expires now.
1456 if (pr->ndpr_raf_onlink == 0) {
1457 pr->ndpr_vltime = 0;
1458 pr->ndpr_pltime = 0;
1459 in6_init_prefix_ltimes(pr);
1464 * Address autoconfiguration based on Section 5.5.3 of RFC 2462.
1465 * Note that pr must be non NULL at this point.
1468 /* 5.5.3 (a). Ignore the prefix without the A bit set. */
1469 if (!new->ndpr_raf_auto)
1473 * 5.5.3 (b). the link-local prefix should have been ignored in
1477 /* 5.5.3 (c). Consistency check on lifetimes: pltime <= vltime. */
1478 if (new->ndpr_pltime > new->ndpr_vltime) {
1479 error = EINVAL; /* XXX: won't be used */
1484 * 5.5.3 (d). If the prefix advertised is not equal to the prefix of
1485 * an address configured by stateless autoconfiguration already in the
1486 * list of addresses associated with the interface, and the Valid
1487 * Lifetime is not 0, form an address. We first check if we have
1488 * a matching prefix.
1489 * Note: we apply a clarification in rfc2462bis-02 here. We only
1490 * consider autoconfigured addresses while RFC2462 simply said
1494 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1495 struct in6_ifaddr *ifa6;
1496 u_int32_t remaininglifetime;
1498 if (ifa->ifa_addr->sa_family != AF_INET6)
1501 ifa6 = (struct in6_ifaddr *)ifa;
1504 * We only consider autoconfigured addresses as per rfc2462bis.
1506 if (!(ifa6->ia6_flags & IN6_IFF_AUTOCONF))
1510 * Spec is not clear here, but I believe we should concentrate
1511 * on unicast (i.e. not anycast) addresses.
1512 * XXX: other ia6_flags? detached or duplicated?
1514 if ((ifa6->ia6_flags & IN6_IFF_ANYCAST) != 0)
1518 * Ignore the address if it is not associated with a prefix
1519 * or is associated with a prefix that is different from this
1520 * one. (pr is never NULL here)
1522 if (ifa6->ia6_ndpr != pr)
1525 if (ia6_match == NULL) /* remember the first one */
1529 * An already autoconfigured address matched. Now that we
1530 * are sure there is at least one matched address, we can
1531 * proceed to 5.5.3. (e): update the lifetimes according to the
1532 * "two hours" rule and the privacy extension.
1533 * We apply some clarifications in rfc2462bis:
1534 * - use remaininglifetime instead of storedlifetime as a
1536 * - remove the dead code in the "two-hour" rule
1538 #define TWOHOUR (120*60)
1539 lt6_tmp = ifa6->ia6_lifetime;
1541 if (lt6_tmp.ia6t_vltime == ND6_INFINITE_LIFETIME)
1542 remaininglifetime = ND6_INFINITE_LIFETIME;
1543 else if (time_uptime - ifa6->ia6_updatetime >
1544 lt6_tmp.ia6t_vltime) {
1546 * The case of "invalid" address. We should usually
1547 * not see this case.
1549 remaininglifetime = 0;
1551 remaininglifetime = lt6_tmp.ia6t_vltime -
1552 (time_uptime - ifa6->ia6_updatetime);
1554 /* when not updating, keep the current stored lifetime. */
1555 lt6_tmp.ia6t_vltime = remaininglifetime;
1557 if (TWOHOUR < new->ndpr_vltime ||
1558 remaininglifetime < new->ndpr_vltime) {
1559 lt6_tmp.ia6t_vltime = new->ndpr_vltime;
1560 } else if (remaininglifetime <= TWOHOUR) {
1562 lt6_tmp.ia6t_vltime = new->ndpr_vltime;
1566 * new->ndpr_vltime <= TWOHOUR &&
1567 * TWOHOUR < remaininglifetime
1569 lt6_tmp.ia6t_vltime = TWOHOUR;
1572 /* The 2 hour rule is not imposed for preferred lifetime. */
1573 lt6_tmp.ia6t_pltime = new->ndpr_pltime;
1575 in6_init_address_ltimes(pr, <6_tmp);
1578 * We need to treat lifetimes for temporary addresses
1579 * differently, according to
1580 * draft-ietf-ipv6-privacy-addrs-v2-01.txt 3.3 (1);
1581 * we only update the lifetimes when they are in the maximum
1584 if ((ifa6->ia6_flags & IN6_IFF_TEMPORARY) != 0) {
1585 u_int32_t maxvltime, maxpltime;
1587 if (V_ip6_temp_valid_lifetime >
1588 (u_int32_t)((time_uptime - ifa6->ia6_createtime) +
1589 V_ip6_desync_factor)) {
1590 maxvltime = V_ip6_temp_valid_lifetime -
1591 (time_uptime - ifa6->ia6_createtime) -
1592 V_ip6_desync_factor;
1595 if (V_ip6_temp_preferred_lifetime >
1596 (u_int32_t)((time_uptime - ifa6->ia6_createtime) +
1597 V_ip6_desync_factor)) {
1598 maxpltime = V_ip6_temp_preferred_lifetime -
1599 (time_uptime - ifa6->ia6_createtime) -
1600 V_ip6_desync_factor;
1604 if (lt6_tmp.ia6t_vltime == ND6_INFINITE_LIFETIME ||
1605 lt6_tmp.ia6t_vltime > maxvltime) {
1606 lt6_tmp.ia6t_vltime = maxvltime;
1608 if (lt6_tmp.ia6t_pltime == ND6_INFINITE_LIFETIME ||
1609 lt6_tmp.ia6t_pltime > maxpltime) {
1610 lt6_tmp.ia6t_pltime = maxpltime;
1613 ifa6->ia6_lifetime = lt6_tmp;
1614 ifa6->ia6_updatetime = time_uptime;
1616 IF_ADDR_RUNLOCK(ifp);
1617 if (ia6_match == NULL && new->ndpr_vltime) {
1621 * 5.5.3 (d) (continued)
1622 * No address matched and the valid lifetime is non-zero.
1623 * Create a new address.
1627 * Prefix Length check:
1628 * If the sum of the prefix length and interface identifier
1629 * length does not equal 128 bits, the Prefix Information
1630 * option MUST be ignored. The length of the interface
1631 * identifier is defined in a separate link-type specific
1634 ifidlen = in6_if2idlen(ifp);
1636 /* this should not happen, so we always log it. */
1637 log(LOG_ERR, "prelist_update: IFID undefined (%s)\n",
1641 if (ifidlen + pr->ndpr_plen != 128) {
1643 "%s: invalid prefixlen %d for %s, ignored\n",
1644 __func__, pr->ndpr_plen, if_name(ifp)));
1648 if ((ia6 = in6_ifadd(new, mcast)) != NULL) {
1650 * note that we should use pr (not new) for reference.
1657 * When a new public address is created as described
1658 * in RFC2462, also create a new temporary address.
1661 * When an interface connects to a new link, a new
1662 * randomized interface identifier should be generated
1663 * immediately together with a new set of temporary
1664 * addresses. Thus, we specifiy 1 as the 2nd arg of
1667 if (V_ip6_use_tempaddr) {
1669 if ((e = in6_tmpifadd(ia6, 1, 1)) != 0) {
1670 nd6log((LOG_NOTICE, "%s: failed to "
1671 "create a temporary address "
1672 "(errno=%d)\n", __func__, e));
1675 ifa_free(&ia6->ia_ifa);
1678 * A newly added address might affect the status
1679 * of other addresses, so we check and update it.
1680 * XXX: what if address duplication happens?
1682 pfxlist_onlink_check();
1684 /* just set an error. do not bark here. */
1685 error = EADDRNOTAVAIL; /* XXX: might be unused. */
1691 nd6_prefix_rele(pr);
1696 * A supplement function used in the on-link detection below;
1697 * detect if a given prefix has a (probably) reachable advertising router.
1698 * XXX: lengthy function name...
1700 static struct nd_pfxrouter *
1701 find_pfxlist_reachable_router(struct nd_prefix *pr)
1703 struct nd_pfxrouter *pfxrtr;
1709 LIST_FOREACH(pfxrtr, &pr->ndpr_advrtrs, pfr_entry) {
1710 IF_AFDATA_RLOCK(pfxrtr->router->ifp);
1711 ln = nd6_lookup(&pfxrtr->router->rtaddr, 0, pfxrtr->router->ifp);
1712 IF_AFDATA_RUNLOCK(pfxrtr->router->ifp);
1715 canreach = ND6_IS_LLINFO_PROBREACH(ln);
1724 * Check if each prefix in the prefix list has at least one available router
1725 * that advertised the prefix (a router is "available" if its neighbor cache
1726 * entry is reachable or probably reachable).
1727 * If the check fails, the prefix may be off-link, because, for example,
1728 * we have moved from the network but the lifetime of the prefix has not
1729 * expired yet. So we should not use the prefix if there is another prefix
1730 * that has an available router.
1731 * But, if there is no prefix that has an available router, we still regard
1732 * all the prefixes as on-link. This is because we can't tell if all the
1733 * routers are simply dead or if we really moved from the network and there
1734 * is no router around us.
1737 pfxlist_onlink_check(void)
1739 struct nd_prefix *pr;
1740 struct in6_ifaddr *ifa;
1741 struct nd_defrouter *dr;
1742 struct nd_pfxrouter *pfxrtr = NULL;
1743 struct rm_priotracker in6_ifa_tracker;
1751 * Check if there is a prefix that has a reachable advertising
1754 LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
1755 if (pr->ndpr_raf_onlink && find_pfxlist_reachable_router(pr))
1760 * If we have no such prefix, check whether we still have a router
1761 * that does not advertise any prefixes.
1764 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry) {
1765 struct nd_prefix *pr0;
1767 LIST_FOREACH(pr0, &V_nd_prefix, ndpr_entry) {
1768 if ((pfxrtr = pfxrtr_lookup(pr0, dr)) != NULL)
1775 if (pr != NULL || (!TAILQ_EMPTY(&V_nd6_defrouter) && pfxrtr == NULL)) {
1777 * There is at least one prefix that has a reachable router,
1778 * or at least a router which probably does not advertise
1779 * any prefixes. The latter would be the case when we move
1780 * to a new link where we have a router that does not provide
1781 * prefixes and we configure an address by hand.
1782 * Detach prefixes which have no reachable advertising
1783 * router, and attach other prefixes.
1785 LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
1786 /* XXX: a link-local prefix should never be detached */
1787 if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr) ||
1788 pr->ndpr_raf_onlink == 0 ||
1789 pr->ndpr_raf_auto == 0)
1792 if ((pr->ndpr_stateflags & NDPRF_DETACHED) == 0 &&
1793 find_pfxlist_reachable_router(pr) == NULL)
1794 pr->ndpr_stateflags |= NDPRF_DETACHED;
1795 else if ((pr->ndpr_stateflags & NDPRF_DETACHED) != 0 &&
1796 find_pfxlist_reachable_router(pr) != NULL)
1797 pr->ndpr_stateflags &= ~NDPRF_DETACHED;
1800 /* there is no prefix that has a reachable router */
1801 LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
1802 if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr) ||
1803 pr->ndpr_raf_onlink == 0 ||
1804 pr->ndpr_raf_auto == 0)
1806 pr->ndpr_stateflags &= ~NDPRF_DETACHED;
1811 * Remove each interface route associated with a (just) detached
1812 * prefix, and reinstall the interface route for a (just) attached
1813 * prefix. Note that all attempt of reinstallation does not
1814 * necessarily success, when a same prefix is shared among multiple
1815 * interfaces. Such cases will be handled in nd6_prefix_onlink,
1816 * so we don't have to care about them.
1819 LIST_FOREACH(pr, &V_nd_prefix, ndpr_entry) {
1820 char ip6buf[INET6_ADDRSTRLEN];
1823 if (IN6_IS_ADDR_LINKLOCAL(&pr->ndpr_prefix.sin6_addr) ||
1824 pr->ndpr_raf_onlink == 0 ||
1825 pr->ndpr_raf_auto == 0)
1828 flags = pr->ndpr_stateflags & (NDPRF_DETACHED | NDPRF_ONLINK);
1829 if (flags == 0 || flags == (NDPRF_DETACHED | NDPRF_ONLINK)) {
1830 genid = V_nd6_list_genid;
1832 if ((flags & NDPRF_ONLINK) != 0 &&
1833 (e = nd6_prefix_offlink(pr)) != 0) {
1835 "%s: failed to make %s/%d offlink "
1836 "(errno=%d)\n", __func__,
1838 &pr->ndpr_prefix.sin6_addr),
1840 } else if ((flags & NDPRF_ONLINK) == 0 &&
1841 (e = nd6_prefix_onlink(pr)) != 0) {
1843 "%s: failed to make %s/%d onlink "
1844 "(errno=%d)\n", __func__,
1846 &pr->ndpr_prefix.sin6_addr),
1850 if (genid != V_nd6_list_genid)
1856 * Changes on the prefix status might affect address status as well.
1857 * Make sure that all addresses derived from an attached prefix are
1858 * attached, and that all addresses derived from a detached prefix are
1859 * detached. Note, however, that a manually configured address should
1860 * always be attached.
1861 * The precise detection logic is same as the one for prefixes.
1863 IN6_IFADDR_RLOCK(&in6_ifa_tracker);
1864 CK_STAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) {
1865 if (!(ifa->ia6_flags & IN6_IFF_AUTOCONF))
1868 if (ifa->ia6_ndpr == NULL) {
1870 * This can happen when we first configure the address
1871 * (i.e. the address exists, but the prefix does not).
1872 * XXX: complicated relationships...
1877 if (find_pfxlist_reachable_router(ifa->ia6_ndpr))
1881 CK_STAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) {
1882 if ((ifa->ia6_flags & IN6_IFF_AUTOCONF) == 0)
1885 if (ifa->ia6_ndpr == NULL) /* XXX: see above. */
1888 if (find_pfxlist_reachable_router(ifa->ia6_ndpr)) {
1889 if (ifa->ia6_flags & IN6_IFF_DETACHED) {
1890 ifa->ia6_flags &= ~IN6_IFF_DETACHED;
1891 ifa->ia6_flags |= IN6_IFF_TENTATIVE;
1892 nd6_dad_start((struct ifaddr *)ifa, 0);
1895 ifa->ia6_flags |= IN6_IFF_DETACHED;
1899 CK_STAILQ_FOREACH(ifa, &V_in6_ifaddrhead, ia_link) {
1900 if ((ifa->ia6_flags & IN6_IFF_AUTOCONF) == 0)
1903 if (ifa->ia6_flags & IN6_IFF_DETACHED) {
1904 ifa->ia6_flags &= ~IN6_IFF_DETACHED;
1905 ifa->ia6_flags |= IN6_IFF_TENTATIVE;
1906 /* Do we need a delay in this case? */
1907 nd6_dad_start((struct ifaddr *)ifa, 0);
1911 IN6_IFADDR_RUNLOCK(&in6_ifa_tracker);
1913 ND6_ONLINK_UNLOCK();
1917 nd6_prefix_onlink_rtrequest(struct nd_prefix *pr, struct ifaddr *ifa)
1919 static struct sockaddr_dl null_sdl = {sizeof(null_sdl), AF_LINK};
1920 struct rib_head *rnh;
1922 struct sockaddr_in6 mask6;
1924 int error, a_failure, fibnum, maxfib;
1927 * in6_ifinit() sets nd6_rtrequest to ifa_rtrequest for all ifaddrs.
1928 * ifa->ifa_rtrequest = nd6_rtrequest;
1930 bzero(&mask6, sizeof(mask6));
1931 mask6.sin6_len = sizeof(mask6);
1932 mask6.sin6_addr = pr->ndpr_mask;
1933 rtflags = (ifa->ifa_flags & ~IFA_RTSELF) | RTF_UP;
1935 if(V_rt_add_addr_allfibs) {
1937 maxfib = rt_numfibs;
1939 fibnum = ifa->ifa_ifp->if_fib;
1940 maxfib = fibnum + 1;
1943 for (; fibnum < maxfib; fibnum++) {
1946 error = in6_rtrequest(RTM_ADD,
1947 (struct sockaddr *)&pr->ndpr_prefix, ifa->ifa_addr,
1948 (struct sockaddr *)&mask6, rtflags, &rt, fibnum);
1950 KASSERT(rt != NULL, ("%s: in6_rtrequest return no "
1951 "error(%d) but rt is NULL, pr=%p, ifa=%p", __func__,
1954 rnh = rt_tables_get_rnh(rt->rt_fibnum, AF_INET6);
1955 /* XXX what if rhn == NULL? */
1958 if (rt_setgate(rt, rt_key(rt),
1959 (struct sockaddr *)&null_sdl) == 0) {
1960 struct sockaddr_dl *dl;
1962 dl = (struct sockaddr_dl *)rt->rt_gateway;
1963 dl->sdl_type = rt->rt_ifp->if_type;
1964 dl->sdl_index = rt->rt_ifp->if_index;
1967 nd6_rtmsg(RTM_ADD, rt);
1969 pr->ndpr_stateflags |= NDPRF_ONLINK;
1971 char ip6buf[INET6_ADDRSTRLEN];
1972 char ip6bufg[INET6_ADDRSTRLEN];
1973 char ip6bufm[INET6_ADDRSTRLEN];
1974 struct sockaddr_in6 *sin6;
1976 sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
1977 nd6log((LOG_ERR, "%s: failed to add "
1978 "route for a prefix (%s/%d) on %s, gw=%s, mask=%s, "
1979 "flags=%lx errno = %d\n", __func__,
1980 ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
1981 pr->ndpr_plen, if_name(pr->ndpr_ifp),
1982 ip6_sprintf(ip6bufg, &sin6->sin6_addr),
1983 ip6_sprintf(ip6bufm, &mask6.sin6_addr),
1986 /* Save last error to return, see rtinit(). */
1997 /* Return the last error we got. */
2002 nd6_prefix_onlink(struct nd_prefix *pr)
2005 struct ifnet *ifp = pr->ndpr_ifp;
2006 struct nd_prefix *opr;
2007 char ip6buf[INET6_ADDRSTRLEN];
2010 ND6_ONLINK_LOCK_ASSERT();
2011 ND6_UNLOCK_ASSERT();
2013 if ((pr->ndpr_stateflags & NDPRF_ONLINK) != 0)
2017 * Add the interface route associated with the prefix. Before
2018 * installing the route, check if there's the same prefix on another
2019 * interface, and the prefix has already installed the interface route.
2020 * Although such a configuration is expected to be rare, we explicitly
2024 LIST_FOREACH(opr, &V_nd_prefix, ndpr_entry) {
2028 if ((opr->ndpr_stateflags & NDPRF_ONLINK) == 0)
2031 if (!V_rt_add_addr_allfibs &&
2032 opr->ndpr_ifp->if_fib != pr->ndpr_ifp->if_fib)
2035 if (opr->ndpr_plen == pr->ndpr_plen &&
2036 in6_are_prefix_equal(&pr->ndpr_prefix.sin6_addr,
2037 &opr->ndpr_prefix.sin6_addr, pr->ndpr_plen)) {
2045 * We prefer link-local addresses as the associated interface address.
2047 /* search for a link-local addr */
2048 ifa = (struct ifaddr *)in6ifa_ifpforlinklocal(ifp,
2049 IN6_IFF_NOTREADY | IN6_IFF_ANYCAST);
2051 /* XXX: freebsd does not have ifa_ifwithaf */
2053 CK_STAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
2054 if (ifa->ifa_addr->sa_family == AF_INET6) {
2059 IF_ADDR_RUNLOCK(ifp);
2060 /* should we care about ia6_flags? */
2064 * This can still happen, when, for example, we receive an RA
2065 * containing a prefix with the L bit set and the A bit clear,
2066 * after removing all IPv6 addresses on the receiving
2067 * interface. This should, of course, be rare though.
2070 "%s: failed to find any ifaddr to add route for a "
2071 "prefix(%s/%d) on %s\n", __func__,
2072 ip6_sprintf(ip6buf, &pr->ndpr_prefix.sin6_addr),
2073 pr->ndpr_plen, if_name(ifp)));
2077 error = nd6_prefix_onlink_rtrequest(pr, ifa);
2086 nd6_prefix_offlink(struct nd_prefix *pr)
2089 struct ifnet *ifp = pr->ndpr_ifp;
2090 struct nd_prefix *opr;
2091 struct sockaddr_in6 sa6, mask6;
2093 char ip6buf[INET6_ADDRSTRLEN];
2095 int fibnum, maxfib, a_failure;
2097 ND6_ONLINK_LOCK_ASSERT();
2098 ND6_UNLOCK_ASSERT();
2100 if ((pr->ndpr_stateflags & NDPRF_ONLINK) == 0)
2103 bzero(&sa6, sizeof(sa6));
2104 sa6.sin6_family = AF_INET6;
2105 sa6.sin6_len = sizeof(sa6);
2106 bcopy(&pr->ndpr_prefix.sin6_addr, &sa6.sin6_addr,
2107 sizeof(struct in6_addr));
2108 bzero(&mask6, sizeof(mask6));
2109 mask6.sin6_family = AF_INET6;
2110 mask6.sin6_len = sizeof(sa6);
2111 bcopy(&pr->ndpr_mask, &mask6.sin6_addr, sizeof(struct in6_addr));
2113 if (V_rt_add_addr_allfibs) {
2115 maxfib = rt_numfibs;
2117 fibnum = ifp->if_fib;
2118 maxfib = fibnum + 1;
2122 for (; fibnum < maxfib; fibnum++) {
2124 error = in6_rtrequest(RTM_DELETE, (struct sockaddr *)&sa6, NULL,
2125 (struct sockaddr *)&mask6, 0, &rt, fibnum);
2127 /* report the route deletion to the routing socket. */
2129 nd6_rtmsg(RTM_DELETE, rt);
2131 /* Save last error to return, see rtinit(). */
2141 pr->ndpr_stateflags &= ~NDPRF_ONLINK;
2144 * There might be the same prefix on another interface,
2145 * the prefix which could not be on-link just because we have
2146 * the interface route (see comments in nd6_prefix_onlink).
2147 * If there's one, try to make the prefix on-link on the
2152 LIST_FOREACH(opr, &V_nd_prefix, ndpr_entry) {
2154 * KAME specific: detached prefixes should not be
2157 if (opr == pr || (opr->ndpr_stateflags &
2158 (NDPRF_ONLINK | NDPRF_DETACHED)) != 0)
2161 if (opr->ndpr_plen == pr->ndpr_plen &&
2162 in6_are_prefix_equal(&pr->ndpr_prefix.sin6_addr,
2163 &opr->ndpr_prefix.sin6_addr, pr->ndpr_plen)) {
2166 genid = V_nd6_list_genid;
2168 if ((e = nd6_prefix_onlink(opr)) != 0) {
2170 "%s: failed to recover a prefix "
2171 "%s/%d from %s to %s (errno=%d)\n",
2172 __func__, ip6_sprintf(ip6buf,
2173 &opr->ndpr_prefix.sin6_addr),
2174 opr->ndpr_plen, if_name(ifp),
2175 if_name(opr->ndpr_ifp), e));
2179 if (genid != V_nd6_list_genid)
2185 /* XXX: can we still set the NDPRF_ONLINK flag? */
2187 "%s: failed to delete route: %s/%d on %s (errno=%d)\n",
2188 __func__, ip6_sprintf(ip6buf, &sa6.sin6_addr),
2189 pr->ndpr_plen, if_name(ifp), error));
2193 lltable_prefix_free(AF_INET6, (struct sockaddr *)&sa6,
2194 (struct sockaddr *)&mask6, LLE_STATIC);
2200 * ia0 - corresponding public address
2203 in6_tmpifadd(const struct in6_ifaddr *ia0, int forcegen, int delay)
2205 struct ifnet *ifp = ia0->ia_ifa.ifa_ifp;
2206 struct in6_ifaddr *newia;
2207 struct in6_aliasreq ifra;
2209 int trylimit = 3; /* XXX: adhoc value */
2211 u_int32_t randid[2];
2212 time_t vltime0, pltime0;
2214 in6_prepare_ifra(&ifra, &ia0->ia_addr.sin6_addr,
2215 &ia0->ia_prefixmask.sin6_addr);
2217 ifra.ifra_addr = ia0->ia_addr; /* XXX: do we need this ? */
2218 /* clear the old IFID */
2219 IN6_MASK_ADDR(&ifra.ifra_addr.sin6_addr,
2220 &ifra.ifra_prefixmask.sin6_addr);
2223 if (in6_get_tmpifid(ifp, (u_int8_t *)randid,
2224 (const u_int8_t *)&ia0->ia_addr.sin6_addr.s6_addr[8], forcegen)) {
2225 nd6log((LOG_NOTICE, "%s: failed to find a good random IFID\n",
2229 ifra.ifra_addr.sin6_addr.s6_addr32[2] |=
2230 (randid[0] & ~(ifra.ifra_prefixmask.sin6_addr.s6_addr32[2]));
2231 ifra.ifra_addr.sin6_addr.s6_addr32[3] |=
2232 (randid[1] & ~(ifra.ifra_prefixmask.sin6_addr.s6_addr32[3]));
2235 * in6_get_tmpifid() quite likely provided a unique interface ID.
2236 * However, we may still have a chance to see collision, because
2237 * there may be a time lag between generation of the ID and generation
2238 * of the address. So, we'll do one more sanity check.
2241 if (in6_localip(&ifra.ifra_addr.sin6_addr) != 0) {
2242 if (trylimit-- > 0) {
2247 /* Give up. Something strange should have happened. */
2248 nd6log((LOG_NOTICE, "%s: failed to find a unique random IFID\n",
2254 * The Valid Lifetime is the lower of the Valid Lifetime of the
2255 * public address or TEMP_VALID_LIFETIME.
2256 * The Preferred Lifetime is the lower of the Preferred Lifetime
2257 * of the public address or TEMP_PREFERRED_LIFETIME -
2260 if (ia0->ia6_lifetime.ia6t_vltime != ND6_INFINITE_LIFETIME) {
2261 vltime0 = IFA6_IS_INVALID(ia0) ? 0 :
2262 (ia0->ia6_lifetime.ia6t_vltime -
2263 (time_uptime - ia0->ia6_updatetime));
2264 if (vltime0 > V_ip6_temp_valid_lifetime)
2265 vltime0 = V_ip6_temp_valid_lifetime;
2267 vltime0 = V_ip6_temp_valid_lifetime;
2268 if (ia0->ia6_lifetime.ia6t_pltime != ND6_INFINITE_LIFETIME) {
2269 pltime0 = IFA6_IS_DEPRECATED(ia0) ? 0 :
2270 (ia0->ia6_lifetime.ia6t_pltime -
2271 (time_uptime - ia0->ia6_updatetime));
2272 if (pltime0 > V_ip6_temp_preferred_lifetime - V_ip6_desync_factor){
2273 pltime0 = V_ip6_temp_preferred_lifetime -
2274 V_ip6_desync_factor;
2277 pltime0 = V_ip6_temp_preferred_lifetime - V_ip6_desync_factor;
2278 ifra.ifra_lifetime.ia6t_vltime = vltime0;
2279 ifra.ifra_lifetime.ia6t_pltime = pltime0;
2282 * A temporary address is created only if this calculated Preferred
2283 * Lifetime is greater than REGEN_ADVANCE time units.
2285 if (ifra.ifra_lifetime.ia6t_pltime <= V_ip6_temp_regen_advance)
2288 /* XXX: scope zone ID? */
2290 ifra.ifra_flags |= (IN6_IFF_AUTOCONF|IN6_IFF_TEMPORARY);
2292 /* allocate ifaddr structure, link into chain, etc. */
2295 updateflags |= IN6_IFAUPDATE_DADDELAY;
2296 if ((error = in6_update_ifa(ifp, &ifra, NULL, updateflags)) != 0)
2299 newia = in6ifa_ifpwithaddr(ifp, &ifra.ifra_addr.sin6_addr);
2300 if (newia == NULL) { /* XXX: can it happen? */
2302 "%s: ifa update succeeded, but we got no ifaddr\n",
2304 return (EINVAL); /* XXX */
2306 newia->ia6_ndpr = ia0->ia6_ndpr;
2307 newia->ia6_ndpr->ndpr_addrcnt++;
2308 ifa_free(&newia->ia_ifa);
2311 * A newly added address might affect the status of other addresses.
2312 * XXX: when the temporary address is generated with a new public
2313 * address, the onlink check is redundant. However, it would be safe
2314 * to do the check explicitly everywhere a new address is generated,
2315 * and, in fact, we surely need the check when we create a new
2316 * temporary address due to deprecation of an old temporary address.
2318 pfxlist_onlink_check();
2324 rt6_deleteroute(const struct rtentry *rt, void *arg)
2326 #define SIN6(s) ((struct sockaddr_in6 *)s)
2327 struct in6_addr *gate = (struct in6_addr *)arg;
2329 if (rt->rt_gateway == NULL || rt->rt_gateway->sa_family != AF_INET6)
2332 if (!IN6_ARE_ADDR_EQUAL(gate, &SIN6(rt->rt_gateway)->sin6_addr)) {
2337 * Do not delete a static route.
2338 * XXX: this seems to be a bit ad-hoc. Should we consider the
2339 * 'cloned' bit instead?
2341 if ((rt->rt_flags & RTF_STATIC) != 0)
2345 * We delete only host route. This means, in particular, we don't
2346 * delete default route.
2348 if ((rt->rt_flags & RTF_HOST) == 0)
2356 * Delete all the routing table entries that use the specified gateway.
2357 * XXX: this function causes search through all entries of routing table, so
2358 * it shouldn't be called when acting as a router.
2361 rt6_flush(struct in6_addr *gateway, struct ifnet *ifp)
2364 /* We'll care only link-local addresses */
2365 if (!IN6_IS_ADDR_LINKLOCAL(gateway))
2368 /* XXX Do we really need to walk any but the default FIB? */
2369 rt_foreach_fib_walk_del(AF_INET6, rt6_deleteroute, (void *)gateway);
2373 nd6_setdefaultiface(int ifindex)
2377 if (ifindex < 0 || V_if_index < ifindex)
2379 if (ifindex != 0 && !ifnet_byindex(ifindex))
2382 if (V_nd6_defifindex != ifindex) {
2383 V_nd6_defifindex = ifindex;
2384 if (V_nd6_defifindex > 0)
2385 V_nd6_defifp = ifnet_byindex(V_nd6_defifindex);
2387 V_nd6_defifp = NULL;
2390 * Our current implementation assumes one-to-one maping between
2391 * interfaces and links, so it would be natural to use the
2392 * default interface as the default link.
2394 scope6_setdefault(V_nd6_defifp);
2401 nd6_defrouter_list_empty(void)
2404 return (TAILQ_EMPTY(&V_nd6_defrouter));
2408 nd6_defrouter_timer(void)
2410 struct nd_defrouter *dr, *ndr;
2411 struct nd_drhead drq;
2416 TAILQ_FOREACH_SAFE(dr, &V_nd6_defrouter, dr_entry, ndr)
2417 if (dr->expire && dr->expire < time_uptime)
2418 defrouter_unlink(dr, &drq);
2421 while ((dr = TAILQ_FIRST(&drq)) != NULL) {
2422 TAILQ_REMOVE(&drq, dr, dr_entry);
2428 * Nuke default router list entries toward ifp.
2429 * We defer removal of default router list entries that is installed in the
2430 * routing table, in order to keep additional side effects as small as possible.
2433 nd6_defrouter_purge(struct ifnet *ifp)
2435 struct nd_defrouter *dr, *ndr;
2436 struct nd_drhead drq;
2441 TAILQ_FOREACH_SAFE(dr, &V_nd6_defrouter, dr_entry, ndr) {
2445 defrouter_unlink(dr, &drq);
2447 TAILQ_FOREACH_SAFE(dr, &V_nd6_defrouter, dr_entry, ndr) {
2451 defrouter_unlink(dr, &drq);
2455 /* Delete the unlinked router objects. */
2456 while ((dr = TAILQ_FIRST(&drq)) != NULL) {
2457 TAILQ_REMOVE(&drq, dr, dr_entry);
2463 nd6_defrouter_flush_all(void)
2465 struct nd_defrouter *dr;
2466 struct nd_drhead drq;
2471 while ((dr = TAILQ_FIRST(&V_nd6_defrouter)) != NULL)
2472 defrouter_unlink(dr, &drq);
2475 while ((dr = TAILQ_FIRST(&drq)) != NULL) {
2476 TAILQ_REMOVE(&drq, dr, dr_entry);
2482 nd6_defrouter_init(void)
2485 TAILQ_INIT(&V_nd6_defrouter);
2489 nd6_sysctl_drlist(SYSCTL_HANDLER_ARGS)
2491 struct in6_defrouter d;
2492 struct nd_defrouter *dr;
2495 if (req->newptr != NULL)
2498 error = sysctl_wire_old_buffer(req, 0);
2502 bzero(&d, sizeof(d));
2503 d.rtaddr.sin6_family = AF_INET6;
2504 d.rtaddr.sin6_len = sizeof(d.rtaddr);
2507 TAILQ_FOREACH(dr, &V_nd6_defrouter, dr_entry) {
2508 d.rtaddr.sin6_addr = dr->rtaddr;
2509 error = sa6_recoverscope(&d.rtaddr);
2512 d.flags = dr->raflags;
2513 d.rtlifetime = dr->rtlifetime;
2514 d.expire = dr->expire + (time_second - time_uptime);
2515 d.if_index = dr->ifp->if_index;
2516 error = SYSCTL_OUT(req, &d, sizeof(d));
2523 SYSCTL_PROC(_net_inet6_icmp6, ICMPV6CTL_ND6_DRLIST, nd6_drlist,
2524 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_MPSAFE,
2525 NULL, 0, nd6_sysctl_drlist, "S,in6_defrouter",
2526 "NDP default router list");