2 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the project nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * $KAME: in6_src.c,v 1.132 2003/08/26 04:42:27 keiichi Exp $
33 * Copyright (c) 1982, 1986, 1991, 1993
34 * The Regents of the University of California. All rights reserved.
36 * Redistribution and use in source and binary forms, with or without
37 * modification, are permitted provided that the following conditions
39 * 1. Redistributions of source code must retain the above copyright
40 * notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 * notice, this list of conditions and the following disclaimer in the
43 * documentation and/or other materials provided with the distribution.
44 * 4. Neither the name of the University nor the names of its contributors
45 * may be used to endorse or promote products derived from this software
46 * without specific prior written permission.
48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
60 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
63 #include <sys/cdefs.h>
64 __FBSDID("$FreeBSD$");
67 #include "opt_inet6.h"
68 #include "opt_mpath.h"
70 #include <sys/param.h>
71 #include <sys/systm.h>
73 #include <sys/malloc.h>
76 #include <sys/protosw.h>
77 #include <sys/socket.h>
78 #include <sys/socketvar.h>
79 #include <sys/sockio.h>
80 #include <sys/sysctl.h>
81 #include <sys/errno.h>
84 #include <sys/kernel.h>
88 #include <net/if_dl.h>
89 #include <net/route.h>
90 #include <net/if_llatbl.h>
92 #include <net/radix_mpath.h>
95 #include <netinet/in.h>
96 #include <netinet/in_var.h>
97 #include <netinet/in_systm.h>
98 #include <netinet/ip.h>
99 #include <netinet/in_pcb.h>
100 #include <netinet/ip_var.h>
101 #include <netinet/udp.h>
102 #include <netinet/udp_var.h>
104 #include <netinet6/in6_var.h>
105 #include <netinet/ip6.h>
106 #include <netinet6/in6_pcb.h>
107 #include <netinet6/ip6_var.h>
108 #include <netinet6/scope6_var.h>
109 #include <netinet6/nd6.h>
111 static struct mtx addrsel_lock;
112 #define ADDRSEL_LOCK_INIT() mtx_init(&addrsel_lock, "addrsel_lock", NULL, MTX_DEF)
113 #define ADDRSEL_LOCK() mtx_lock(&addrsel_lock)
114 #define ADDRSEL_UNLOCK() mtx_unlock(&addrsel_lock)
115 #define ADDRSEL_LOCK_ASSERT() mtx_assert(&addrsel_lock, MA_OWNED)
117 static struct sx addrsel_sxlock;
118 #define ADDRSEL_SXLOCK_INIT() sx_init(&addrsel_sxlock, "addrsel_sxlock")
119 #define ADDRSEL_SLOCK() sx_slock(&addrsel_sxlock)
120 #define ADDRSEL_SUNLOCK() sx_sunlock(&addrsel_sxlock)
121 #define ADDRSEL_XLOCK() sx_xlock(&addrsel_sxlock)
122 #define ADDRSEL_XUNLOCK() sx_xunlock(&addrsel_sxlock)
124 #define ADDR_LABEL_NOTAPP (-1)
125 static VNET_DEFINE(struct in6_addrpolicy, defaultaddrpolicy);
126 #define V_defaultaddrpolicy VNET(defaultaddrpolicy)
128 VNET_DEFINE(int, ip6_prefer_tempaddr) = 0;
130 static int selectroute __P((struct sockaddr_in6 *, struct ip6_pktopts *,
131 struct ip6_moptions *, struct route_in6 *, struct ifnet **,
132 struct rtentry **, int));
133 static int in6_selectif __P((struct sockaddr_in6 *, struct ip6_pktopts *,
134 struct ip6_moptions *, struct route_in6 *ro, struct ifnet **));
136 static struct in6_addrpolicy *lookup_addrsel_policy(struct sockaddr_in6 *);
138 static void init_policy_queue(void);
139 static int add_addrsel_policyent(struct in6_addrpolicy *);
140 static int delete_addrsel_policyent(struct in6_addrpolicy *);
141 static int walk_addrsel_policy __P((int (*)(struct in6_addrpolicy *, void *),
143 static int dump_addrsel_policyent(struct in6_addrpolicy *, void *);
144 static struct in6_addrpolicy *match_addrsel_policy(struct sockaddr_in6 *);
147 * Return an IPv6 address, which is the most appropriate for a given
148 * destination and user specified options.
149 * If necessary, this function lookups the routing table and returns
150 * an entry to the caller for later use.
152 #define REPLACE(r) do {\
153 if ((r) < sizeof(V_ip6stat.ip6s_sources_rule) / \
154 sizeof(V_ip6stat.ip6s_sources_rule[0])) /* check for safety */ \
155 V_ip6stat.ip6s_sources_rule[(r)]++; \
157 char ip6buf[INET6_ADDRSTRLEN], ip6b[INET6_ADDRSTRLEN]; \
158 printf("in6_selectsrc: replace %s with %s by %d\n", ia_best ? ip6_sprintf(ip6buf, &ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(ip6b, &ia->ia_addr.sin6_addr), (r)); \
162 #define NEXT(r) do {\
163 if ((r) < sizeof(V_ip6stat.ip6s_sources_rule) / \
164 sizeof(V_ip6stat.ip6s_sources_rule[0])) /* check for safety */ \
165 V_ip6stat.ip6s_sources_rule[(r)]++; \
167 char ip6buf[INET6_ADDRSTRLEN], ip6b[INET6_ADDRSTRLEN]; \
168 printf("in6_selectsrc: keep %s against %s by %d\n", ia_best ? ip6_sprintf(ip6buf, &ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(ip6b, &ia->ia_addr.sin6_addr), (r)); \
170 goto next; /* XXX: we can't use 'continue' here */ \
172 #define BREAK(r) do { \
173 if ((r) < sizeof(V_ip6stat.ip6s_sources_rule) / \
174 sizeof(V_ip6stat.ip6s_sources_rule[0])) /* check for safety */ \
175 V_ip6stat.ip6s_sources_rule[(r)]++; \
176 goto out; /* XXX: we can't use 'break' here */ \
180 in6_selectsrc(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
181 struct inpcb *inp, struct route_in6 *ro, struct ucred *cred,
182 struct ifnet **ifpp, struct in6_addr *srcp)
184 struct in6_addr dst, tmp;
185 struct ifnet *ifp = NULL;
186 struct in6_ifaddr *ia = NULL, *ia_best = NULL;
187 struct in6_pktinfo *pi = NULL;
188 int dst_scope = -1, best_scope = -1, best_matchlen = -1;
189 struct in6_addrpolicy *dst_policy = NULL, *best_policy = NULL;
193 struct ip6_moptions *mopts;
195 KASSERT(srcp != NULL, ("%s: srcp is NULL", __func__));
197 dst = dstsock->sin6_addr; /* make a copy for local operation */
202 INP_LOCK_ASSERT(inp);
203 mopts = inp->in6p_moptions;
209 * If the source address is explicitly specified by the caller,
210 * check if the requested source address is indeed a unicast address
211 * assigned to the node, and can be used as the packet's source
212 * address. If everything is okay, use the address as source.
214 if (opts && (pi = opts->ip6po_pktinfo) &&
215 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) {
216 struct sockaddr_in6 srcsock;
217 struct in6_ifaddr *ia6;
219 /* get the outgoing interface */
220 if ((error = in6_selectif(dstsock, opts, mopts, ro, &ifp)) != 0)
224 * determine the appropriate zone id of the source based on
225 * the zone of the destination and the outgoing interface.
226 * If the specified address is ambiguous wrt the scope zone,
227 * the interface must be specified; otherwise, ifa_ifwithaddr()
228 * will fail matching the address.
230 bzero(&srcsock, sizeof(srcsock));
231 srcsock.sin6_family = AF_INET6;
232 srcsock.sin6_len = sizeof(srcsock);
233 srcsock.sin6_addr = pi->ipi6_addr;
235 error = in6_setscope(&srcsock.sin6_addr, ifp, NULL);
239 if (cred != NULL && (error = prison_local_ip6(cred,
240 &srcsock.sin6_addr, (inp != NULL &&
241 (inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
244 ia6 = (struct in6_ifaddr *)ifa_ifwithaddr(
245 (struct sockaddr *)&srcsock);
247 (ia6->ia6_flags & (IN6_IFF_ANYCAST | IN6_IFF_NOTREADY))) {
249 ifa_free(&ia6->ia_ifa);
250 return (EADDRNOTAVAIL);
252 pi->ipi6_addr = srcsock.sin6_addr; /* XXX: this overrides pi */
255 bcopy(&ia6->ia_addr.sin6_addr, srcp, sizeof(*srcp));
256 ifa_free(&ia6->ia_ifa);
261 * Otherwise, if the socket has already bound the source, just use it.
263 if (inp != NULL && !IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
265 (error = prison_local_ip6(cred, &inp->in6p_laddr,
266 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0))) != 0)
268 bcopy(&inp->in6p_laddr, srcp, sizeof(*srcp));
273 * Bypass source address selection and use the primary jail IP
276 if (cred != NULL && !prison_saddrsel_ip6(cred, srcp))
280 * If the address is not specified, choose the best one based on
281 * the outgoing interface and the destination address.
283 /* get the outgoing interface */
284 if ((error = in6_selectif(dstsock, opts, mopts, ro, &ifp)) != 0)
288 if (ifp == NULL) /* this should not happen */
289 panic("in6_selectsrc: NULL ifp");
291 error = in6_setscope(&dst, ifp, &odstzone);
296 TAILQ_FOREACH(ia, &V_in6_ifaddrhead, ia_link) {
297 int new_scope = -1, new_matchlen = -1;
298 struct in6_addrpolicy *new_policy = NULL;
299 u_int32_t srczone, osrczone, dstzone;
301 struct ifnet *ifp1 = ia->ia_ifp;
304 * We'll never take an address that breaks the scope zone
305 * of the destination. We also skip an address if its zone
306 * does not contain the outgoing interface.
307 * XXX: we should probably use sin6_scope_id here.
309 if (in6_setscope(&dst, ifp1, &dstzone) ||
310 odstzone != dstzone) {
313 src = ia->ia_addr.sin6_addr;
314 if (in6_setscope(&src, ifp, &osrczone) ||
315 in6_setscope(&src, ifp1, &srczone) ||
316 osrczone != srczone) {
320 /* avoid unusable addresses */
322 (IN6_IFF_NOTREADY | IN6_IFF_ANYCAST | IN6_IFF_DETACHED))) {
325 if (!V_ip6_use_deprecated && IFA6_IS_DEPRECATED(ia))
328 /* If jailed only take addresses of the jail into account. */
330 prison_check_ip6(cred, &ia->ia_addr.sin6_addr) != 0)
333 /* Rule 1: Prefer same address */
334 if (IN6_ARE_ADDR_EQUAL(&dst, &ia->ia_addr.sin6_addr)) {
336 BREAK(1); /* there should be no better candidate */
342 /* Rule 2: Prefer appropriate scope */
344 dst_scope = in6_addrscope(&dst);
345 new_scope = in6_addrscope(&ia->ia_addr.sin6_addr);
346 if (IN6_ARE_SCOPE_CMP(best_scope, new_scope) < 0) {
347 if (IN6_ARE_SCOPE_CMP(best_scope, dst_scope) < 0)
350 } else if (IN6_ARE_SCOPE_CMP(new_scope, best_scope) < 0) {
351 if (IN6_ARE_SCOPE_CMP(new_scope, dst_scope) < 0)
357 * Rule 3: Avoid deprecated addresses. Note that the case of
358 * !ip6_use_deprecated is already rejected above.
360 if (!IFA6_IS_DEPRECATED(ia_best) && IFA6_IS_DEPRECATED(ia))
362 if (IFA6_IS_DEPRECATED(ia_best) && !IFA6_IS_DEPRECATED(ia))
365 /* Rule 4: Prefer home addresses */
367 * XXX: This is a TODO. We should probably merge the MIP6
371 /* Rule 5: Prefer outgoing interface */
372 if (ia_best->ia_ifp == ifp && ia->ia_ifp != ifp)
374 if (ia_best->ia_ifp != ifp && ia->ia_ifp == ifp)
378 * Rule 6: Prefer matching label
379 * Note that best_policy should be non-NULL here.
381 if (dst_policy == NULL)
382 dst_policy = lookup_addrsel_policy(dstsock);
383 if (dst_policy->label != ADDR_LABEL_NOTAPP) {
384 new_policy = lookup_addrsel_policy(&ia->ia_addr);
385 if (dst_policy->label == best_policy->label &&
386 dst_policy->label != new_policy->label)
388 if (dst_policy->label != best_policy->label &&
389 dst_policy->label == new_policy->label)
394 * Rule 7: Prefer public addresses.
395 * We allow users to reverse the logic by configuring
396 * a sysctl variable, so that privacy conscious users can
397 * always prefer temporary addresses.
400 opts->ip6po_prefer_tempaddr == IP6PO_TEMPADDR_SYSTEM) {
401 prefer_tempaddr = V_ip6_prefer_tempaddr;
402 } else if (opts->ip6po_prefer_tempaddr ==
403 IP6PO_TEMPADDR_NOTPREFER) {
407 if (!(ia_best->ia6_flags & IN6_IFF_TEMPORARY) &&
408 (ia->ia6_flags & IN6_IFF_TEMPORARY)) {
414 if ((ia_best->ia6_flags & IN6_IFF_TEMPORARY) &&
415 !(ia->ia6_flags & IN6_IFF_TEMPORARY)) {
423 * Rule 8: prefer addresses on alive interfaces.
424 * This is a KAME specific rule.
426 if ((ia_best->ia_ifp->if_flags & IFF_UP) &&
427 !(ia->ia_ifp->if_flags & IFF_UP))
429 if (!(ia_best->ia_ifp->if_flags & IFF_UP) &&
430 (ia->ia_ifp->if_flags & IFF_UP))
434 * Rule 14: Use longest matching prefix.
435 * Note: in the address selection draft, this rule is
436 * documented as "Rule 8". However, since it is also
437 * documented that this rule can be overridden, we assign
438 * a large number so that it is easy to assign smaller numbers
439 * to more preferred rules.
441 new_matchlen = in6_matchlen(&ia->ia_addr.sin6_addr, &dst);
442 if (best_matchlen < new_matchlen)
444 if (new_matchlen < best_matchlen)
447 /* Rule 15 is reserved. */
450 * Last resort: just keep the current candidate.
451 * Or, do we need more rules?
457 best_scope = (new_scope >= 0 ? new_scope :
458 in6_addrscope(&ia_best->ia_addr.sin6_addr));
459 best_policy = (new_policy ? new_policy :
460 lookup_addrsel_policy(&ia_best->ia_addr));
461 best_matchlen = (new_matchlen >= 0 ? new_matchlen :
462 in6_matchlen(&ia_best->ia_addr.sin6_addr,
472 if ((ia = ia_best) == NULL) {
473 IN6_IFADDR_RUNLOCK();
474 return (EADDRNOTAVAIL);
478 * At this point at least one of the addresses belonged to the jail
479 * but it could still be, that we want to further restrict it, e.g.
480 * theoratically IN6_IS_ADDR_LOOPBACK.
481 * It must not be IN6_IS_ADDR_UNSPECIFIED anymore.
482 * prison_local_ip6() will fix an IN6_IS_ADDR_LOOPBACK but should
483 * let all others previously selected pass.
484 * Use tmp to not change ::1 on lo0 to the primary jail address.
486 tmp = ia->ia_addr.sin6_addr;
487 if (cred != NULL && prison_local_ip6(cred, &tmp, (inp != NULL &&
488 (inp->inp_flags & IN6P_IPV6_V6ONLY) != 0)) != 0) {
489 IN6_IFADDR_RUNLOCK();
490 return (EADDRNOTAVAIL);
496 bcopy(&tmp, srcp, sizeof(*srcp));
497 IN6_IFADDR_RUNLOCK();
502 * clone - meaningful only for bsdi and freebsd
505 selectroute(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
506 struct ip6_moptions *mopts, struct route_in6 *ro,
507 struct ifnet **retifp, struct rtentry **retrt, int norouteok)
510 struct ifnet *ifp = NULL;
511 struct rtentry *rt = NULL;
512 struct sockaddr_in6 *sin6_next;
513 struct in6_pktinfo *pi = NULL;
514 struct in6_addr *dst = &dstsock->sin6_addr;
516 char ip6buf[INET6_ADDRSTRLEN];
518 if (dstsock->sin6_addr.s6_addr32[0] == 0 &&
519 dstsock->sin6_addr.s6_addr32[1] == 0 &&
520 !IN6_IS_ADDR_LOOPBACK(&dstsock->sin6_addr)) {
521 printf("in6_selectroute: strange destination %s\n",
522 ip6_sprintf(ip6buf, &dstsock->sin6_addr));
524 printf("in6_selectroute: destination = %s%%%d\n",
525 ip6_sprintf(ip6buf, &dstsock->sin6_addr),
526 dstsock->sin6_scope_id); /* for debug */
530 /* If the caller specify the outgoing interface explicitly, use it. */
531 if (opts && (pi = opts->ip6po_pktinfo) != NULL && pi->ipi6_ifindex) {
532 /* XXX boundary check is assumed to be already done. */
533 ifp = ifnet_byindex(pi->ipi6_ifindex);
535 (norouteok || retrt == NULL ||
536 IN6_IS_ADDR_MULTICAST(dst))) {
538 * we do not have to check or get the route for
547 * If the destination address is a multicast address and the outgoing
548 * interface for the address is specified by the caller, use it.
550 if (IN6_IS_ADDR_MULTICAST(dst) &&
551 mopts != NULL && (ifp = mopts->im6o_multicast_ifp) != NULL) {
552 goto done; /* we do not need a route for multicast. */
557 * If the next hop address for the packet is specified by the caller,
558 * use it as the gateway.
560 if (opts && opts->ip6po_nexthop) {
561 struct route_in6 *ron;
564 sin6_next = satosin6(opts->ip6po_nexthop);
566 /* at this moment, we only support AF_INET6 next hops */
567 if (sin6_next->sin6_family != AF_INET6) {
568 error = EAFNOSUPPORT; /* or should we proceed? */
573 * If the next hop is an IPv6 address, then the node identified
574 * by that address must be a neighbor of the sending host.
576 ron = &opts->ip6po_nextroute;
578 * XXX what do we do here?
583 if (ron->ro_rt == NULL) {
584 rtalloc((struct route *)ron); /* multi path case? */
585 if (ron->ro_rt == NULL) {
590 error = EHOSTUNREACH;
598 la = lla_lookup(LLTABLE6(ifp), 0, (struct sockaddr *)&sin6_next->sin6_addr);
599 IF_AFDATA_UNLOCK(ifp);
603 error = EHOSTUNREACH;
608 (ron->ro_rt->rt_flags & (RTF_UP | RTF_LLINFO)) !=
609 (RTF_UP | RTF_LLINFO)) ||
610 !IN6_ARE_ADDR_EQUAL(&satosin6(&ron->ro_dst)->sin6_addr,
611 &sin6_next->sin6_addr)) {
616 *satosin6(&ron->ro_dst) = *sin6_next;
618 if (ron->ro_rt == NULL) {
619 rtalloc((struct route *)ron); /* multi path case? */
620 if (ron->ro_rt == NULL ||
621 !(ron->ro_rt->rt_flags & RTF_LLINFO)) {
626 error = EHOSTUNREACH;
633 * When cloning is required, try to allocate a route to the
634 * destination so that the caller can store path MTU
641 * Use a cached route if it exists and is valid, else try to allocate
642 * a new one. Note that we should check the address family of the
643 * cached destination, in case of sharing the cache with IPv4.
647 (!(ro->ro_rt->rt_flags & RTF_UP) ||
648 ((struct sockaddr *)(&ro->ro_dst))->sa_family != AF_INET6 ||
649 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr,
652 ro->ro_rt = (struct rtentry *)NULL;
654 if (ro->ro_rt == (struct rtentry *)NULL) {
655 struct sockaddr_in6 *sa6;
657 /* No route yet, so try to acquire one */
658 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
659 sa6 = (struct sockaddr_in6 *)&ro->ro_dst;
661 sa6->sin6_scope_id = 0;
664 rtalloc_mpath((struct route *)ro,
665 ntohl(sa6->sin6_addr.s6_addr32[3]));
667 ro->ro_rt = rtalloc1(&((struct route *)ro)
670 RT_UNLOCK(ro->ro_rt);
675 * do not care about the result if we have the nexthop
676 * explicitly specified.
678 if (opts && opts->ip6po_nexthop)
682 ifp = ro->ro_rt->rt_ifp;
684 if (ifp == NULL) { /* can this really happen? */
689 if (ro->ro_rt == NULL)
690 error = EHOSTUNREACH;
694 * Check if the outgoing interface conflicts with
695 * the interface specified by ipi6_ifindex (if specified).
696 * Note that loopback interface is always okay.
697 * (this may happen when we are sending a packet to one of
698 * our own addresses.)
700 if (ifp && opts && opts->ip6po_pktinfo &&
701 opts->ip6po_pktinfo->ipi6_ifindex) {
702 if (!(ifp->if_flags & IFF_LOOPBACK) &&
704 opts->ip6po_pktinfo->ipi6_ifindex) {
705 error = EHOSTUNREACH;
712 if (ifp == NULL && rt == NULL) {
714 * This can happen if the caller did not pass a cached route
715 * nor any other hints. We treat this case an error.
717 error = EHOSTUNREACH;
719 if (error == EHOSTUNREACH)
720 V_ip6stat.ip6s_noroute++;
722 if (retifp != NULL) {
726 * Adjust the "outgoing" interface. If we're going to loop
727 * the packet back to ourselves, the ifp would be the loopback
728 * interface. However, we'd rather know the interface associated
729 * to the destination address (which should probably be one of
730 * our own addresses.)
733 if ((rt->rt_ifp->if_flags & IFF_LOOPBACK) &&
734 (rt->rt_gateway->sa_family == AF_LINK))
736 ifnet_byindex(((struct sockaddr_dl *)
737 rt->rt_gateway)->sdl_index);
742 *retrt = rt; /* rt may be NULL */
748 in6_selectif(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
749 struct ip6_moptions *mopts, struct route_in6 *ro, struct ifnet **retifp)
752 struct route_in6 sro;
753 struct rtentry *rt = NULL;
756 bzero(&sro, sizeof(sro));
760 if ((error = selectroute(dstsock, opts, mopts, ro, retifp,
762 if (ro == &sro && rt && rt == sro.ro_rt)
768 * do not use a rejected or black hole route.
769 * XXX: this check should be done in the L2 output routine.
770 * However, if we skipped this check here, we'd see the following
772 * - install a rejected route for a scoped address prefix
774 * - send a packet to a destination that matches the scoped prefix,
775 * with ambiguity about the scope zone.
776 * - pick the outgoing interface from the route, and disambiguate the
777 * scope zone with the interface.
778 * - ip6_output() would try to get another route with the "new"
779 * destination, which may be valid.
780 * - we'd see no error on output.
781 * Although this may not be very harmful, it should still be confusing.
782 * We thus reject the case here.
784 if (rt && (rt->rt_flags & (RTF_REJECT | RTF_BLACKHOLE))) {
785 int flags = (rt->rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH);
787 if (ro == &sro && rt && rt == sro.ro_rt)
792 if (ro == &sro && rt && rt == sro.ro_rt)
798 * clone - meaningful only for bsdi and freebsd
801 in6_selectroute(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
802 struct ip6_moptions *mopts, struct route_in6 *ro,
803 struct ifnet **retifp, struct rtentry **retrt)
806 return (selectroute(dstsock, opts, mopts, ro, retifp,
811 * Default hop limit selection. The precedence is as follows:
812 * 1. Hoplimit value specified via ioctl.
813 * 2. (If the outgoing interface is detected) the current
814 * hop limit of the interface specified by router advertisement.
815 * 3. The system default hoplimit.
818 in6_selecthlim(struct inpcb *in6p, struct ifnet *ifp)
821 if (in6p && in6p->in6p_hops >= 0)
822 return (in6p->in6p_hops);
824 return (ND_IFINFO(ifp)->chlim);
825 else if (in6p && !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) {
826 struct route_in6 ro6;
829 bzero(&ro6, sizeof(ro6));
830 ro6.ro_dst.sin6_family = AF_INET6;
831 ro6.ro_dst.sin6_len = sizeof(struct sockaddr_in6);
832 ro6.ro_dst.sin6_addr = in6p->in6p_faddr;
833 rtalloc((struct route *)&ro6);
835 lifp = ro6.ro_rt->rt_ifp;
838 return (ND_IFINFO(lifp)->chlim);
840 return (V_ip6_defhlim);
842 return (V_ip6_defhlim);
846 * XXX: this is borrowed from in6_pcbbind(). If possible, we should
847 * share this function by all *bsd*...
850 in6_pcbsetport(struct in6_addr *laddr, struct inpcb *inp, struct ucred *cred)
852 struct socket *so = inp->inp_socket;
854 int error, lookupflags = 0;
856 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
859 INP_WLOCK_ASSERT(inp);
860 INP_HASH_WLOCK_ASSERT(pcbinfo);
862 error = prison_local_ip6(cred, laddr,
863 ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0));
867 /* XXX: this is redundant when called from in6_pcbbind */
868 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
869 lookupflags = INPLOOKUP_WILDCARD;
871 inp->inp_flags |= INP_ANONPORT;
873 error = in_pcb_lport(inp, NULL, &lport, cred, lookupflags);
877 inp->inp_lport = lport;
878 if (in_pcbinshash(inp) != 0) {
879 inp->in6p_laddr = in6addr_any;
888 addrsel_policy_init(void)
893 /* initialize the "last resort" policy */
894 bzero(&V_defaultaddrpolicy, sizeof(V_defaultaddrpolicy));
895 V_defaultaddrpolicy.label = ADDR_LABEL_NOTAPP;
897 if (!IS_DEFAULT_VNET(curvnet))
901 ADDRSEL_SXLOCK_INIT();
904 static struct in6_addrpolicy *
905 lookup_addrsel_policy(struct sockaddr_in6 *key)
907 struct in6_addrpolicy *match = NULL;
910 match = match_addrsel_policy(key);
913 match = &V_defaultaddrpolicy;
922 * Subroutines to manage the address selection policy table via sysctl.
925 struct sysctl_req *w_req;
928 static int in6_src_sysctl(SYSCTL_HANDLER_ARGS);
929 SYSCTL_DECL(_net_inet6_ip6);
930 SYSCTL_NODE(_net_inet6_ip6, IPV6CTL_ADDRCTLPOLICY, addrctlpolicy,
931 CTLFLAG_RD, in6_src_sysctl, "");
934 in6_src_sysctl(SYSCTL_HANDLER_ARGS)
941 bzero(&w, sizeof(w));
944 return (walk_addrsel_policy(dump_addrsel_policyent, &w));
948 in6_src_ioctl(u_long cmd, caddr_t data)
951 struct in6_addrpolicy ent0;
953 if (cmd != SIOCAADDRCTL_POLICY && cmd != SIOCDADDRCTL_POLICY)
954 return (EOPNOTSUPP); /* check for safety */
956 ent0 = *(struct in6_addrpolicy *)data;
958 if (ent0.label == ADDR_LABEL_NOTAPP)
960 /* check if the prefix mask is consecutive. */
961 if (in6_mask2len(&ent0.addrmask.sin6_addr, NULL) < 0)
963 /* clear trailing garbages (if any) of the prefix address. */
964 for (i = 0; i < 4; i++) {
965 ent0.addr.sin6_addr.s6_addr32[i] &=
966 ent0.addrmask.sin6_addr.s6_addr32[i];
971 case SIOCAADDRCTL_POLICY:
972 return (add_addrsel_policyent(&ent0));
973 case SIOCDADDRCTL_POLICY:
974 return (delete_addrsel_policyent(&ent0));
977 return (0); /* XXX: compromise compilers */
981 * The followings are implementation of the policy table using a
983 * XXX such details should be hidden.
984 * XXX implementation using binary tree should be more efficient.
986 struct addrsel_policyent {
987 TAILQ_ENTRY(addrsel_policyent) ape_entry;
988 struct in6_addrpolicy ape_policy;
991 TAILQ_HEAD(addrsel_policyhead, addrsel_policyent);
993 static VNET_DEFINE(struct addrsel_policyhead, addrsel_policytab);
994 #define V_addrsel_policytab VNET(addrsel_policytab)
997 init_policy_queue(void)
1000 TAILQ_INIT(&V_addrsel_policytab);
1004 add_addrsel_policyent(struct in6_addrpolicy *newpolicy)
1006 struct addrsel_policyent *new, *pol;
1008 new = malloc(sizeof(*new), M_IFADDR,
1013 /* duplication check */
1014 TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) {
1015 if (IN6_ARE_ADDR_EQUAL(&newpolicy->addr.sin6_addr,
1016 &pol->ape_policy.addr.sin6_addr) &&
1017 IN6_ARE_ADDR_EQUAL(&newpolicy->addrmask.sin6_addr,
1018 &pol->ape_policy.addrmask.sin6_addr)) {
1021 free(new, M_IFADDR);
1022 return (EEXIST); /* or override it? */
1026 bzero(new, sizeof(*new));
1028 /* XXX: should validate entry */
1029 new->ape_policy = *newpolicy;
1031 TAILQ_INSERT_TAIL(&V_addrsel_policytab, new, ape_entry);
1039 delete_addrsel_policyent(struct in6_addrpolicy *key)
1041 struct addrsel_policyent *pol;
1046 /* search for the entry in the table */
1047 TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) {
1048 if (IN6_ARE_ADDR_EQUAL(&key->addr.sin6_addr,
1049 &pol->ape_policy.addr.sin6_addr) &&
1050 IN6_ARE_ADDR_EQUAL(&key->addrmask.sin6_addr,
1051 &pol->ape_policy.addrmask.sin6_addr)) {
1061 TAILQ_REMOVE(&V_addrsel_policytab, pol, ape_entry);
1069 walk_addrsel_policy(int (*callback)(struct in6_addrpolicy *, void *),
1072 struct addrsel_policyent *pol;
1076 TAILQ_FOREACH(pol, &V_addrsel_policytab, ape_entry) {
1077 if ((error = (*callback)(&pol->ape_policy, w)) != 0) {
1087 dump_addrsel_policyent(struct in6_addrpolicy *pol, void *arg)
1090 struct walkarg *w = arg;
1092 error = SYSCTL_OUT(w->w_req, pol, sizeof(*pol));
1097 static struct in6_addrpolicy *
1098 match_addrsel_policy(struct sockaddr_in6 *key)
1100 struct addrsel_policyent *pent;
1101 struct in6_addrpolicy *bestpol = NULL, *pol;
1102 int matchlen, bestmatchlen = -1;
1103 u_char *mp, *ep, *k, *p, m;
1105 TAILQ_FOREACH(pent, &V_addrsel_policytab, ape_entry) {
1108 pol = &pent->ape_policy;
1109 mp = (u_char *)&pol->addrmask.sin6_addr;
1110 ep = mp + 16; /* XXX: scope field? */
1111 k = (u_char *)&key->sin6_addr;
1112 p = (u_char *)&pol->addr.sin6_addr;
1113 for (; mp < ep && *mp; mp++, k++, p++) {
1116 goto next; /* not match */
1117 if (m == 0xff) /* short cut for a typical case */
1127 /* matched. check if this is better than the current best. */
1128 if (bestpol == NULL ||
1129 matchlen > bestmatchlen) {
1131 bestmatchlen = matchlen;