1 /* $FreeBSD: src/sys/netinet6/in6_src.c,v 1.1.2.3 2002/02/26 18:02:06 ume Exp $ */
2 /* $KAME: in6_src.c,v 1.37 2001/03/29 05:34:31 itojun Exp $ */
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * Copyright (c) 1982, 1986, 1991, 1993
35 * The Regents of the University of California. All rights reserved.
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
40 * 1. Redistributions of source code must retain the above copyright
41 * notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 * notice, this list of conditions and the following disclaimer in the
44 * documentation and/or other materials provided with the distribution.
45 * 3. Neither the name of the University nor the names of its contributors
46 * may be used to endorse or promote products derived from this software
47 * without specific prior written permission.
49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
61 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
65 #include "opt_inet6.h"
67 #include <sys/param.h>
68 #include <sys/systm.h>
70 #include <sys/kernel.h>
71 #include <sys/malloc.h>
73 #include <sys/protosw.h>
74 #include <sys/socket.h>
75 #include <sys/socketvar.h>
76 #include <sys/sockio.h>
77 #include <sys/sysctl.h>
78 #include <sys/errno.h>
84 #include <net/route.h>
86 #include <netinet/in.h>
87 #include <netinet/in_var.h>
88 #include <netinet/in_systm.h>
89 #include <netinet/ip.h>
90 #include <netinet/in_pcb.h>
91 #include <netinet6/in6_var.h>
92 #include <netinet/ip6.h>
93 #include <netinet6/in6_pcb.h>
94 #include <netinet6/ip6_var.h>
95 #include <netinet6/nd6.h>
96 #ifdef ENABLE_DEFAULT_SCOPE
97 #include <netinet6/scope6_var.h>
100 #include <net/net_osdep.h>
102 #define ADDR_LABEL_NOTAPP (-1)
103 struct in6_addrpolicy defaultaddrpolicy;
105 static void init_policy_queue(void);
106 static int add_addrsel_policyent(struct in6_addrpolicy *);
107 static int delete_addrsel_policyent(struct in6_addrpolicy *);
108 static int walk_addrsel_policy(int (*)(struct in6_addrpolicy *, void *),
110 static int dump_addrsel_policyent(struct in6_addrpolicy *, void *);
114 * Return an IPv6 address, which is the most appropriate for a given
115 * destination and user specified options.
116 * If necessary, this function lookups the routing table and returns
117 * an entry to the caller for later use.
120 in6_selectsrc(struct sockaddr_in6 *dstsock, struct ip6_pktopts *opts,
121 struct ip6_moptions *mopts, struct route_in6 *ro,
122 struct in6_addr *laddr, int *errorp, struct thread *td)
124 struct sockaddr_in6 jsin6;
125 struct ucred *cred = NULL;
126 struct in6_addr *dst;
127 struct in6_ifaddr *ia6 = NULL;
128 struct in6_pktinfo *pi = NULL;
131 if (td && td->td_proc && td->td_proc->p_ucred)
132 cred = td->td_proc->p_ucred;
133 if (cred && cred->cr_prison)
135 jsin6.sin6_family = AF_INET6;
136 dst = &dstsock->sin6_addr;
140 * If the source address is explicitly specified by the caller,
143 if (opts && (pi = opts->ip6po_pktinfo) &&
144 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) {
145 jsin6.sin6_addr = pi->ipi6_addr;
146 if (jailed && !jailed_ip(cred->cr_prison,
147 (struct sockaddr *)&jsin6)) {
150 return (&pi->ipi6_addr);
155 * If the source address is not specified but the socket(if any)
156 * is already bound, use the bound address.
158 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr)) {
159 jsin6.sin6_addr = *laddr;
160 if (jailed && !jailed_ip(cred->cr_prison,
161 (struct sockaddr *)&jsin6)) {
169 * If the caller doesn't specify the source address but
170 * the outgoing interface, use an address associated with
173 if (pi && pi->ipi6_ifindex) {
174 /* XXX boundary check is assumed to be already done. */
175 ia6 = in6_ifawithscope(ifindex2ifnet[pi->ipi6_ifindex],
179 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
180 if (!jailed_ip(cred->cr_prison,
181 (struct sockaddr *)&jsin6))
186 *errorp = EADDRNOTAVAIL;
189 return (&satosin6(&ia6->ia_addr)->sin6_addr);
193 * If the destination address is a link-local unicast address or
194 * a multicast address, and if the outgoing interface is specified
195 * by the sin6_scope_id filed, use an address associated with the
197 * XXX: We're now trying to define more specific semantics of
198 * sin6_scope_id field, so this part will be rewritten in
201 if ((IN6_IS_ADDR_LINKLOCAL(dst) || IN6_IS_ADDR_MULTICAST(dst)) &&
202 dstsock->sin6_scope_id) {
204 * I'm not sure if boundary check for scope_id is done
207 if (dstsock->sin6_scope_id < 0 ||
208 if_index < dstsock->sin6_scope_id) {
209 *errorp = ENXIO; /* XXX: better error? */
212 ia6 = in6_ifawithscope(ifindex2ifnet[dstsock->sin6_scope_id],
216 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
217 if (!jailed_ip(cred->cr_prison,
218 (struct sockaddr *)&jsin6))
223 *errorp = EADDRNOTAVAIL;
226 return (&satosin6(&ia6->ia_addr)->sin6_addr);
230 * If the destination address is a multicast address and
231 * the outgoing interface for the address is specified
232 * by the caller, use an address associated with the interface.
233 * There is a sanity check here; if the destination has node-local
234 * scope, the outgoing interfacde should be a loopback address.
235 * Even if the outgoing interface is not specified, we also
236 * choose a loopback interface as the outgoing interface.
238 if (!jailed && IN6_IS_ADDR_MULTICAST(dst)) {
239 struct ifnet *ifp = mopts ? mopts->im6o_multicast_ifp : NULL;
241 if (ifp == NULL && IN6_IS_ADDR_MC_NODELOCAL(dst)) {
246 ia6 = in6_ifawithscope(ifp, dst, cred);
248 *errorp = EADDRNOTAVAIL;
251 return (&satosin6(&ia6->ia_addr)->sin6_addr);
256 * If the next hop address for the packet is specified
257 * by caller, use an address associated with the route
261 struct sockaddr_in6 *sin6_next;
264 if (opts && opts->ip6po_nexthop) {
265 sin6_next = satosin6(opts->ip6po_nexthop);
266 rt = nd6_lookup(&sin6_next->sin6_addr, 1, NULL);
268 ia6 = in6_ifawithscope(rt->rt_ifp, dst, cred);
270 ia6 = ifatoia6(rt->rt_ifa);
273 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
274 if (!jailed_ip(cred->cr_prison,
275 (struct sockaddr *)&jsin6))
280 *errorp = EADDRNOTAVAIL;
283 return (&satosin6(&ia6->ia_addr)->sin6_addr);
288 * If route is known or can be allocated now,
289 * our src addr is taken from the i/f, else punt.
293 (!(ro->ro_rt->rt_flags & RTF_UP) ||
294 satosin6(&ro->ro_dst)->sin6_family != AF_INET6 ||
295 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr,
300 if (ro->ro_rt == NULL || ro->ro_rt->rt_ifp == NULL) {
301 struct sockaddr_in6 *sa6;
303 /* No route yet, so try to acquire one */
304 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
306 sa6->sin6_family = AF_INET6;
307 sa6->sin6_len = sizeof(struct sockaddr_in6);
308 sa6->sin6_addr = *dst;
309 sa6->sin6_scope_id = dstsock->sin6_scope_id;
310 if (!jailed && IN6_IS_ADDR_MULTICAST(dst)) {
312 rtpurelookup((struct sockaddr *)&ro->ro_dst);
314 rtalloc((struct route *)ro);
319 * in_pcbconnect() checks out IFF_LOOPBACK to skip using
320 * the address. But we don't know why it does so.
321 * It is necessary to ensure the scope even for lo0
322 * so doesn't check out IFF_LOOPBACK.
326 ia6 = in6_ifawithscope(ro->ro_rt->rt_ifa->ifa_ifp, dst, cred);
328 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
329 if (!jailed_ip(cred->cr_prison,
330 (struct sockaddr *)&jsin6))
334 if (ia6 == NULL) /* xxx scope error ?*/
335 ia6 = ifatoia6(ro->ro_rt->rt_ifa);
338 jsin6.sin6_addr = (&ia6->ia_addr)->sin6_addr;
339 if (!jailed_ip(cred->cr_prison,
340 (struct sockaddr *)&jsin6))
346 * xxx The followings are necessary? (kazu)
348 * It's for SO_DONTROUTE option in IPv4.(jinmei)
351 struct sockaddr_in6 sin6 = {sizeof(sin6), AF_INET6, 0};
353 sin6->sin6_addr = *dst;
355 ia6 = ifatoia6(ifa_ifwithdstaddr(sin6tosa(&sin6)));
357 ia6 = ifatoia6(ifa_ifwithnet(sin6tosa(&sin6)));
360 return (&satosin6(&ia6->ia_addr)->sin6_addr);
364 *errorp = EHOSTUNREACH; /* no route */
367 return (&satosin6(&ia6->ia_addr)->sin6_addr);
370 *errorp = EADDRNOTAVAIL;
375 * Default hop limit selection. The precedence is as follows:
376 * 1. Hoplimit value specified via ioctl.
377 * 2. (If the outgoing interface is detected) the current
378 * hop limit of the interface specified by router advertisement.
379 * 3. The system default hoplimit.
382 in6_selecthlim(struct in6pcb *in6p, struct ifnet *ifp)
384 if (in6p && in6p->in6p_hops >= 0)
385 return (in6p->in6p_hops);
387 return (ND_IFINFO(ifp)->chlim);
389 return (ip6_defhlim);
393 * XXX: this is borrowed from in6_pcbbind(). If possible, we should
394 * share this function by all *bsd*...
397 in6_pcbsetport(struct in6_addr *laddr, struct inpcb *inp, struct thread *td)
399 struct socket *so = inp->inp_socket;
400 u_int16_t lport = 0, first, last, *lastport;
401 int count, error = 0, wild = 0;
402 struct inpcbportinfo *portinfo = inp->inp_pcbinfo->portinfo;
403 struct ucred *cred = NULL;
405 /* XXX: this is redundant when called from in6_pcbbind */
406 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
407 wild = INPLOOKUP_WILDCARD;
408 if (td->td_proc && td->td_proc->p_ucred)
409 cred = td->td_proc->p_ucred;
411 inp->inp_flags |= INP_ANONPORT;
413 if (inp->inp_flags & INP_HIGHPORT) {
414 first = ipport_hifirstauto; /* sysctl */
415 last = ipport_hilastauto;
416 lastport = &portinfo->lasthi;
417 } else if (inp->inp_flags & INP_LOWPORT) {
418 if ((error = priv_check(td, PRIV_ROOT)) != 0)
420 first = ipport_lowfirstauto; /* 1023 */
421 last = ipport_lowlastauto; /* 600 */
422 lastport = &portinfo->lastlow;
424 first = ipport_firstauto; /* sysctl */
425 last = ipport_lastauto;
426 lastport = &portinfo->lastport;
430 * This has to be atomic. If the porthash is shared across multiple
431 * protocol threads (aka tcp) then the token will be non-NULL.
433 if (portinfo->porttoken)
434 lwkt_gettoken(portinfo->porttoken);
437 * Simple check to ensure all ports are not used up causing
440 * We split the two cases (up and down) so that the direction
441 * is not being tested on each round of the loop.
447 count = first - last;
450 if (count-- < 0) { /* completely used? */
452 * Undo any address bind that may have
455 inp->in6p_laddr = kin6addr_any;
460 if (*lastport > first || *lastport < last)
462 lport = htons(*lastport);
463 } while (in6_pcblookup_local(portinfo, &inp->in6p_laddr,
469 count = last - first;
472 if (count-- < 0) { /* completely used? */
474 * Undo any address bind that may have
477 inp->in6p_laddr = kin6addr_any;
482 if (*lastport < first || *lastport > last)
484 lport = htons(*lastport);
485 } while (in6_pcblookup_local(portinfo, &inp->in6p_laddr,
489 inp->inp_lport = lport;
490 in_pcbinsporthash(portinfo, inp);
493 if (portinfo->porttoken)
494 lwkt_reltoken(portinfo->porttoken);
499 * generate kernel-internal form (scopeid embedded into s6_addr16[1]).
500 * If the address scope of is link-local, embed the interface index in the
501 * address. The routine determines our precedence
502 * between advanced API scope/interface specification and basic API
505 * this function should be nuked in the future, when we get rid of
506 * embedded scopeid thing.
508 * XXX actually, it is over-specification to return ifp against sin6_scope_id.
509 * there can be multiple interfaces that belong to a particular scope zone
510 * (in specification, we have 1:N mapping between a scope zone and interfaces).
511 * we may want to change the function to return something other than ifp.
514 in6_embedscope(struct in6_addr *in6,
515 const struct sockaddr_in6 *sin6,
516 #ifdef HAVE_NRL_INPCB
518 #define in6p_outputopts inp_outputopts6
519 #define in6p_moptions inp_moptions6
525 struct ifnet *ifp = NULL;
528 *in6 = sin6->sin6_addr;
529 scopeid = sin6->sin6_scope_id;
534 * don't try to read sin6->sin6_addr beyond here, since the caller may
535 * ask us to overwrite existing sockaddr_in6
538 #ifdef ENABLE_DEFAULT_SCOPE
540 scopeid = scope6_addr2default(in6);
543 if (IN6_IS_SCOPE_LINKLOCAL(in6)) {
544 struct in6_pktinfo *pi;
547 * KAME assumption: link id == interface id
550 if (in6p && in6p->in6p_outputopts &&
551 (pi = in6p->in6p_outputopts->ip6po_pktinfo) &&
553 ifp = ifindex2ifnet[pi->ipi6_ifindex];
554 in6->s6_addr16[1] = htons(pi->ipi6_ifindex);
555 } else if (in6p && IN6_IS_ADDR_MULTICAST(in6) &&
556 in6p->in6p_moptions &&
557 in6p->in6p_moptions->im6o_multicast_ifp) {
558 ifp = in6p->in6p_moptions->im6o_multicast_ifp;
559 in6->s6_addr16[1] = htons(ifp->if_index);
560 } else if (scopeid) {
562 if (scopeid < 0 || if_index < scopeid)
563 return ENXIO; /* XXX EINVAL? */
564 ifp = ifindex2ifnet[scopeid];
565 /*XXX assignment to 16bit from 32bit variable */
566 in6->s6_addr16[1] = htons(scopeid & 0xffff);
575 #ifdef HAVE_NRL_INPCB
576 #undef in6p_outputopts
581 * generate standard sockaddr_in6 from embedded form.
582 * touches sin6_addr and sin6_scope_id only.
584 * this function should be nuked in the future, when we get rid of
585 * embedded scopeid thing.
588 in6_recoverscope(struct sockaddr_in6 *sin6, const struct in6_addr *in6,
593 sin6->sin6_addr = *in6;
596 * don't try to read *in6 beyond here, since the caller may
597 * ask us to overwrite existing sockaddr_in6
600 sin6->sin6_scope_id = 0;
601 if (IN6_IS_SCOPE_LINKLOCAL(in6)) {
603 * KAME assumption: link id == interface id
605 scopeid = ntohs(sin6->sin6_addr.s6_addr16[1]);
608 if (scopeid < 0 || if_index < scopeid)
610 if (ifp && ifp->if_index != scopeid)
612 sin6->sin6_addr.s6_addr16[1] = 0;
613 sin6->sin6_scope_id = scopeid;
621 * just clear the embedded scope identifer.
622 * XXX: currently used for bsdi4 only as a supplement function.
625 in6_clearscope(struct in6_addr *addr)
627 if (IN6_IS_SCOPE_LINKLOCAL(addr))
628 addr->s6_addr16[1] = 0;
632 addrsel_policy_init(void)
637 /* initialize the "last resort" policy */
638 bzero(&defaultaddrpolicy, sizeof(defaultaddrpolicy));
639 defaultaddrpolicy.label = ADDR_LABEL_NOTAPP;
643 * Subroutines to manage the address selection policy table via sysctl.
646 struct sysctl_req *w_req;
649 static int in6_src_sysctl(SYSCTL_HANDLER_ARGS);
650 SYSCTL_DECL(_net_inet6_ip6);
651 SYSCTL_NODE(_net_inet6_ip6, IPV6CTL_ADDRCTLPOLICY, addrctlpolicy,
652 CTLFLAG_RD, in6_src_sysctl, "Address selection policy");
655 in6_src_sysctl(SYSCTL_HANDLER_ARGS)
662 bzero(&w, sizeof(w));
665 return (walk_addrsel_policy(dump_addrsel_policyent, &w));
669 in6_src_ioctl(u_long cmd, caddr_t data)
672 struct in6_addrpolicy ent0;
674 if (cmd != SIOCAADDRCTL_POLICY && cmd != SIOCDADDRCTL_POLICY)
675 return (EOPNOTSUPP); /* check for safety */
677 ent0 = *(struct in6_addrpolicy *)data;
679 if (ent0.label == ADDR_LABEL_NOTAPP)
681 /* check if the prefix mask is consecutive. */
682 if (in6_mask2len(&ent0.addrmask.sin6_addr, NULL) < 0)
684 /* clear trailing garbages (if any) of the prefix address. */
685 for (i = 0; i < 4; i++) {
686 ent0.addr.sin6_addr.s6_addr32[i] &=
687 ent0.addrmask.sin6_addr.s6_addr32[i];
692 case SIOCAADDRCTL_POLICY:
693 return (add_addrsel_policyent(&ent0));
694 case SIOCDADDRCTL_POLICY:
695 return (delete_addrsel_policyent(&ent0));
698 return (0); /* XXX: compromise compilers */
702 * The followings are implementation of the policy table using a
704 * XXX such details should be hidden.
705 * XXX implementation using binary tree should be more efficient.
707 struct addrsel_policyent {
708 TAILQ_ENTRY(addrsel_policyent) ape_entry;
709 struct in6_addrpolicy ape_policy;
712 TAILQ_HEAD(addrsel_policyhead, addrsel_policyent);
714 struct addrsel_policyhead addrsel_policytab;
717 init_policy_queue(void)
719 TAILQ_INIT(&addrsel_policytab);
723 add_addrsel_policyent(struct in6_addrpolicy *newpolicy)
725 struct addrsel_policyent *new, *pol;
727 /* duplication check */
728 for (pol = TAILQ_FIRST(&addrsel_policytab); pol;
729 pol = TAILQ_NEXT(pol, ape_entry)) {
730 if (SA6_ARE_ADDR_EQUAL(&newpolicy->addr,
731 &pol->ape_policy.addr) &&
732 SA6_ARE_ADDR_EQUAL(&newpolicy->addrmask,
733 &pol->ape_policy.addrmask)) {
734 return (EEXIST); /* or override it? */
738 new = kmalloc(sizeof(*new), M_IFADDR, M_WAITOK | M_ZERO);
740 /* XXX: should validate entry */
741 new->ape_policy = *newpolicy;
743 TAILQ_INSERT_TAIL(&addrsel_policytab, new, ape_entry);
749 delete_addrsel_policyent(struct in6_addrpolicy *key)
751 struct addrsel_policyent *pol;
753 /* search for the entry in the table */
754 for (pol = TAILQ_FIRST(&addrsel_policytab); pol;
755 pol = TAILQ_NEXT(pol, ape_entry)) {
756 if (SA6_ARE_ADDR_EQUAL(&key->addr, &pol->ape_policy.addr) &&
757 SA6_ARE_ADDR_EQUAL(&key->addrmask,
758 &pol->ape_policy.addrmask)) {
765 TAILQ_REMOVE(&addrsel_policytab, pol, ape_entry);
766 kfree(pol, M_IFADDR);
772 walk_addrsel_policy(int(*callback)(struct in6_addrpolicy *, void *), void *w)
774 struct addrsel_policyent *pol;
777 for (pol = TAILQ_FIRST(&addrsel_policytab); pol;
778 pol = TAILQ_NEXT(pol, ape_entry)) {
779 if ((error = (*callback)(&pol->ape_policy, w)) != 0)
787 dump_addrsel_policyent(struct in6_addrpolicy *pol, void *arg)
790 struct walkarg *w = arg;
792 error = SYSCTL_OUT(w->w_req, pol, sizeof(*pol));