2 * Copyright (c) 2009-2011 Apple Inc. All rights reserved.
4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
6 * This file contains Original Code and/or Modifications of Original Code
7 * as defined in and that are subject to the Apple Public Source License
8 * Version 2.0 (the 'License'). You may not use this file except in
9 * compliance with the License. The rights granted to you under the License
10 * may not be used to create, or enable the creation or redistribution of,
11 * unlawful or unlicensed copies of an Apple operating system, or to
12 * circumvent, violate, or enable the circumvention or violation of, any
13 * terms of an Apple operating system software license agreement.
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
18 * The Original Code and all software distributed under the License are
19 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
22 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
23 * Please see the License for the specific language governing rights and
24 * limitations under the License.
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
29 /* $FreeBSD: src/sys/netinet6/ip6_forward.c,v 1.16 2002/10/16 02:25:05 sam Exp $ */
30 /* $KAME: ip6_forward.c,v 1.69 2001/05/17 03:48:30 itojun Exp $ */
33 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
34 * 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 * 3. Neither the name of the project 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 PROJECT 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 PROJECT 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
62 #include <sys/param.h>
63 #include <sys/systm.h>
64 #include <sys/malloc.h>
66 #include <sys/domain.h>
67 #include <sys/protosw.h>
68 #include <sys/socket.h>
69 #include <sys/errno.h>
71 #include <sys/kernel.h>
72 #include <sys/syslog.h>
75 #include <net/route.h>
77 #include <netinet/in.h>
78 #include <netinet/in_var.h>
79 #include <netinet/in_systm.h>
80 #include <netinet/ip.h>
81 #include <netinet/ip_var.h>
82 #include <netinet6/in6_var.h>
83 #include <netinet/ip6.h>
84 #include <netinet6/ip6_var.h>
85 #include <netinet/icmp6.h>
86 #include <netinet6/nd6.h>
87 #include <netinet6/scope6_var.h>
89 #include <netinet/in_pcb.h>
92 #include <netinet6/ipsec.h>
94 #include <netinet6/ipsec6.h>
96 #include <netkey/key.h>
97 extern int ipsec_bypass
;
100 #include <netinet6/ip6_fw.h>
102 #include <net/net_osdep.h>
105 #include <net/pfvar.h>
109 * Forward a packet. If some error occurs return the sender
110 * an icmp packet. Note we can't always generate a meaningful
111 * icmp message because icmp doesn't have a large enough repertoire
112 * of codes and types.
114 * If not forwarding, just drop the packet. This could be confusing
115 * if ipforwarding was zero but some routing protocol was advancing
116 * us as a gateway to somewhere. However, we must let the routing
117 * protocol deal with that.
122 ip6_forward(struct mbuf
*m
, struct route_in6
*ip6forward_rt
,
125 struct ip6_hdr
*ip6
= mtod(m
, struct ip6_hdr
*);
126 struct sockaddr_in6
*dst
;
128 int error
, type
= 0, code
= 0;
129 boolean_t proxy
= FALSE
;
130 struct mbuf
*mcopy
= NULL
;
131 struct ifnet
*ifp
, *origifp
; /* maybe unnecessary */
132 u_int32_t inzone
, outzone
;
133 struct in6_addr src_in6
, dst_in6
;
135 struct secpolicy
*sp
= NULL
;
137 struct timeval timenow
;
138 unsigned int ifscope
= IFSCOPE_NONE
;
140 struct pf_mtag
*pf_mtag
;
143 getmicrotime(&timenow
);
145 pf_mtag
= pf_find_mtag(m
);
146 if (pf_mtag
!= NULL
&& pf_mtag
->pftag_rtableid
!= IFSCOPE_NONE
)
147 ifscope
= pf_mtag
->pftag_rtableid
;
150 * If the caller provides a route which is on a different interface
151 * than the one specified for scoped forwarding, discard the route
152 * and do a lookup below.
154 if (ifscope
!= IFSCOPE_NONE
&& (rt
= ip6forward_rt
->ro_rt
) != NULL
) {
156 if (rt
->rt_ifp
->if_index
!= ifscope
) {
159 rt
= ip6forward_rt
->ro_rt
= NULL
;
168 * Check AH/ESP integrity.
171 * Don't increment ip6s_cantforward because this is the check
172 * before forwarding packet actually.
174 if (ipsec_bypass
== 0) {
175 if (ipsec6_in_reject(m
, NULL
)) {
176 IPSEC_STAT_INCREMENT(ipsec6stat
.in_polvio
);
184 * Do not forward packets to multicast destination (should be handled
186 * Do not forward packets with unspecified source. It was discussed
187 * in July 2000, on ipngwg mailing list.
189 if ((m
->m_flags
& (M_BCAST
|M_MCAST
)) != 0 ||
190 IN6_IS_ADDR_MULTICAST(&ip6
->ip6_dst
) ||
191 IN6_IS_ADDR_UNSPECIFIED(&ip6
->ip6_src
)) {
192 ip6stat
.ip6s_cantforward
++;
193 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
194 if (ip6_log_time
+ ip6_log_interval
< timenow
.tv_sec
) {
195 ip6_log_time
= timenow
.tv_sec
;
198 "from %s to %s nxt %d received on %s\n",
199 ip6_sprintf(&ip6
->ip6_src
),
200 ip6_sprintf(&ip6
->ip6_dst
),
202 if_name(m
->m_pkthdr
.rcvif
));
208 if (ip6
->ip6_hlim
<= IPV6_HLIMDEC
) {
209 /* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
210 icmp6_error(m
, ICMP6_TIME_EXCEEDED
,
211 ICMP6_TIME_EXCEED_TRANSIT
, 0);
216 * See if the destination is a proxied address, and if so pretend
217 * that it's for us. This is mostly to handle NUD probes against
218 * the proxied addresses. We filter for ICMPv6 here and will let
219 * icmp6_input handle the rest.
221 if (!srcrt
&& nd6_prproxy
) {
222 VERIFY(!IN6_IS_ADDR_MULTICAST(&ip6
->ip6_dst
));
223 proxy
= nd6_prproxy_isours(m
, ip6
, ip6forward_rt
, ifscope
);
225 * Don't update hop limit while proxying; RFC 4389 4.1.
226 * Also skip IPsec forwarding path processing as this
227 * packet is not to be forwarded.
233 ip6
->ip6_hlim
-= IPV6_HLIMDEC
;
236 * Save at most ICMPV6_PLD_MAXLEN (= the min IPv6 MTU -
237 * size of IPv6 + ICMPv6 headers) bytes of the packet in case
238 * we need to generate an ICMP6 message to the src.
239 * Thanks to M_EXT, in most cases copy will not occur.
241 * It is important to save it before IPsec processing as IPsec
242 * processing may modify the mbuf.
244 mcopy
= m_copy(m
, 0, imin(m
->m_pkthdr
.len
, ICMPV6_PLD_MAXLEN
));
247 if (ipsec_bypass
!= 0)
249 /* get a security policy for this packet */
250 sp
= ipsec6_getpolicybyaddr(m
, IPSEC_DIR_OUTBOUND
, IP_FORWARDING
,
253 IPSEC_STAT_INCREMENT(ipsec6stat
.out_inval
);
254 ip6stat
.ip6s_cantforward
++;
257 /* XXX: what icmp ? */
269 switch (sp
->policy
) {
270 case IPSEC_POLICY_DISCARD
:
271 case IPSEC_POLICY_GENERATE
:
273 * This packet is just discarded.
275 IPSEC_STAT_INCREMENT(ipsec6stat
.out_polvio
);
276 ip6stat
.ip6s_cantforward
++;
277 key_freesp(sp
, KEY_SADB_UNLOCKED
);
280 /* XXX: what icmp ? */
288 case IPSEC_POLICY_BYPASS
:
289 case IPSEC_POLICY_NONE
:
290 /* no need to do IPsec. */
291 key_freesp(sp
, KEY_SADB_UNLOCKED
);
294 case IPSEC_POLICY_IPSEC
:
295 if (sp
->req
== NULL
) {
296 /* XXX should be panic ? */
297 printf("ip6_forward: No IPsec request specified.\n");
298 ip6stat
.ip6s_cantforward
++;
299 key_freesp(sp
, KEY_SADB_UNLOCKED
);
302 /* XXX: what icmp ? */
313 case IPSEC_POLICY_ENTRUST
:
315 /* should be panic ?? */
316 printf("ip6_forward: Invalid policy found. %d\n", sp
->policy
);
317 key_freesp(sp
, KEY_SADB_UNLOCKED
);
322 struct ipsec_output_state state
;
325 * All the extension headers will become inaccessible
326 * (since they can be encrypted).
327 * Don't panic, we need no more updates to extension headers
328 * on inner IPv6 packet (since they are now encapsulated).
330 * IPv6 [ESP|AH] IPv6 [extension headers] payload
332 bzero(&state
, sizeof(state
));
334 state
.dst
= NULL
; /* update at ipsec6_output_tunnel() */
336 error
= ipsec6_output_tunnel(&state
, sp
, 0);
337 key_freesp(sp
, KEY_SADB_UNLOCKED
);
338 if (state
.tunneled
== 4)
339 return (NULL
); /* packet is gone - sent over IPv4 */
342 if (state
.ro
.ro_rt
) {
343 rtfree(state
.ro
.ro_rt
);
344 state
.ro
.ro_rt
= NULL
;
347 /* mbuf is already reclaimed in ipsec6_output_tunnel. */
356 printf("ip6_output (ipsec): error code %d\n", error
);
359 /* don't show these error codes to the user */
362 ip6stat
.ip6s_cantforward
++;
365 /* XXX: what icmp ? */
377 dst
= (struct sockaddr_in6
*)&ip6forward_rt
->ro_dst
;
378 if ((rt
= ip6forward_rt
->ro_rt
) != NULL
) {
380 /* Take an extra ref for ourselves */
381 RT_ADDREF_LOCKED(rt
);
386 * ip6forward_rt->ro_dst.sin6_addr is equal to ip6->ip6_dst
388 if (rt
== NULL
|| !(rt
->rt_flags
& RTF_UP
) ||
389 rt
->generation_id
!= route_generation
) {
391 /* Release extra ref */
392 RT_REMREF_LOCKED(rt
);
395 ip6forward_rt
->ro_rt
= NULL
;
397 /* this probably fails but give it a try again */
398 rtalloc_scoped_ign((struct route
*)ip6forward_rt
,
399 RTF_PRCLONING
, ifscope
);
400 if ((rt
= ip6forward_rt
->ro_rt
) != NULL
) {
402 /* Take an extra ref for ourselves */
403 RT_ADDREF_LOCKED(rt
);
408 ip6stat
.ip6s_noroute
++;
409 in6_ifstat_inc(m
->m_pkthdr
.rcvif
, ifs6_in_noroute
);
411 icmp6_error(mcopy
, ICMP6_DST_UNREACH
,
412 ICMP6_DST_UNREACH_NOROUTE
, 0);
416 RT_LOCK_ASSERT_HELD(rt
);
417 } else if (rt
== NULL
|| !(rt
->rt_flags
& RTF_UP
) ||
418 !IN6_ARE_ADDR_EQUAL(&ip6
->ip6_dst
, &dst
->sin6_addr
) ||
419 rt
->generation_id
!= route_generation
) {
421 /* Release extra ref */
422 RT_REMREF_LOCKED(rt
);
425 ip6forward_rt
->ro_rt
= NULL
;
427 bzero(dst
, sizeof(*dst
));
428 dst
->sin6_len
= sizeof(struct sockaddr_in6
);
429 dst
->sin6_family
= AF_INET6
;
430 dst
->sin6_addr
= ip6
->ip6_dst
;
432 rtalloc_scoped_ign((struct route
*)ip6forward_rt
,
433 RTF_PRCLONING
, ifscope
);
434 if ((rt
= ip6forward_rt
->ro_rt
) == NULL
) {
435 ip6stat
.ip6s_noroute
++;
436 in6_ifstat_inc(m
->m_pkthdr
.rcvif
, ifs6_in_noroute
);
438 icmp6_error(mcopy
, ICMP6_DST_UNREACH
,
439 ICMP6_DST_UNREACH_NOROUTE
, 0);
444 /* Take an extra ref for ourselves */
445 RT_ADDREF_LOCKED(rt
);
449 * Source scope check: if a packet can't be delivered to its
450 * destination for the reason that the destination is beyond the scope
451 * of the source address, discard the packet and return an icmp6
452 * destination unreachable error with Code 2 (beyond scope of source
453 * address) unless we are proxying (source address is link local
454 * for NUDs.) We use a local copy of ip6_src, since in6_setscope()
455 * will possibly modify its first argument.
456 * [draft-ietf-ipngwg-icmp-v3-04.txt, Section 3.1]
458 src_in6
= ip6
->ip6_src
;
459 if (in6_setscope(&src_in6
, rt
->rt_ifp
, &outzone
)) {
460 /* XXX: this should not happen */
461 ip6stat
.ip6s_cantforward
++;
462 ip6stat
.ip6s_badscope
++;
466 if (in6_setscope(&src_in6
, m
->m_pkthdr
.rcvif
, &inzone
)) {
467 ip6stat
.ip6s_cantforward
++;
468 ip6stat
.ip6s_badscope
++;
473 if (inzone
!= outzone
&& !proxy
) {
474 ip6stat
.ip6s_cantforward
++;
475 ip6stat
.ip6s_badscope
++;
476 in6_ifstat_inc(rt
->rt_ifp
, ifs6_in_discard
);
478 if (ip6_log_time
+ ip6_log_interval
< timenow
.tv_sec
) {
479 ip6_log_time
= timenow
.tv_sec
;
482 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
483 ip6_sprintf(&ip6
->ip6_src
),
484 ip6_sprintf(&ip6
->ip6_dst
),
486 if_name(m
->m_pkthdr
.rcvif
), if_name(rt
->rt_ifp
));
488 /* Release extra ref */
489 RT_REMREF_LOCKED(rt
);
492 icmp6_error(mcopy
, ICMP6_DST_UNREACH
,
493 ICMP6_DST_UNREACH_BEYONDSCOPE
, 0);
500 * Destination scope check: if a packet is going to break the scope
501 * zone of packet's destination address, discard it. This case should
502 * usually be prevented by appropriately-configured routing table, but
503 * we need an explicit check because we may mistakenly forward the
504 * packet to a different zone by (e.g.) a default route.
506 dst_in6
= ip6
->ip6_dst
;
507 if (in6_setscope(&dst_in6
, m
->m_pkthdr
.rcvif
, &inzone
) != 0 ||
508 in6_setscope(&dst_in6
, rt
->rt_ifp
, &outzone
) != 0 ||
510 ip6stat
.ip6s_cantforward
++;
511 ip6stat
.ip6s_badscope
++;
516 if (m
->m_pkthdr
.len
> rt
->rt_ifp
->if_mtu
) {
517 in6_ifstat_inc(rt
->rt_ifp
, ifs6_in_toobig
);
521 struct secpolicy
*sp2
;
526 mtu
= rt
->rt_ifp
->if_mtu
;
529 * When we do IPsec tunnel ingress, we need to play
530 * with the link value (decrement IPsec header size
531 * from mtu value). The code is much simpler than v4
532 * case, as we have the outgoing interface for
533 * encapsulated packet as "rt->rt_ifp".
535 sp2
= ipsec6_getpolicybyaddr(mcopy
, IPSEC_DIR_OUTBOUND
,
536 IP_FORWARDING
, &ipsecerror
);
538 ipsechdrsiz
= ipsec6_hdrsiz(mcopy
,
539 IPSEC_DIR_OUTBOUND
, NULL
);
540 if (ipsechdrsiz
< mtu
)
542 key_freesp(sp2
, KEY_SADB_UNLOCKED
);
545 * if mtu becomes less than minimum MTU,
546 * tell minimum MTU (and I'll need to fragment it).
551 /* Release extra ref */
552 RT_REMREF_LOCKED(rt
);
554 icmp6_error(mcopy
, ICMP6_PACKET_TOO_BIG
, 0, mtu
);
556 /* Release extra ref */
557 RT_REMREF_LOCKED(rt
);
564 if (rt
->rt_flags
& RTF_GATEWAY
)
565 dst
= (struct sockaddr_in6
*)(void *)rt
->rt_gateway
;
568 * If we are to forward the packet using the same interface
569 * as one we got the packet from, perhaps we should send a redirect
570 * to sender to shortcut a hop.
571 * Only send redirect if source is sending directly to us,
572 * and if packet was not source routed (or has any options).
573 * Also, don't send redirect if forwarding using a route
574 * modified by a redirect.
577 ip6_sendredirects
&& rt
->rt_ifp
== m
->m_pkthdr
.rcvif
&& !srcrt
&&
578 (rt
->rt_flags
& (RTF_DYNAMIC
|RTF_MODIFIED
)) == 0) {
579 if ((rt
->rt_ifp
->if_flags
& IFF_POINTOPOINT
) != 0) {
581 * If the incoming interface is equal to the outgoing
582 * one, and the link attached to the interface is
583 * point-to-point, then it will be highly probable
584 * that a routing loop occurs. Thus, we immediately
585 * drop the packet and send an ICMPv6 error message.
587 * type/code is based on suggestion by Rich Draves.
588 * not sure if it is the best pick.
590 RT_REMREF_LOCKED(rt
); /* Release extra ref */
592 icmp6_error(mcopy
, ICMP6_DST_UNREACH
,
593 ICMP6_DST_UNREACH_ADDR
, 0);
602 * Check with the firewall...
604 if (ip6_fw_enable
&& ip6_fw_chk_ptr
) {
607 /* Drop the lock but retain the extra ref */
609 /* If ipfw says divert, we have to just drop packet */
610 if (ip6_fw_chk_ptr(&ip6
, ifp
, &port
, &m
)) {
617 /* We still have the extra ref on rt */
623 * Fake scoped addresses. Note that even link-local source or
624 * destinaion can appear, if the originating node just sends the
625 * packet to us (without address resolution for the destination).
626 * Since both icmp6_error and icmp6_redirect_output fill the embedded
627 * link identifiers, we can do this stuff after making a copy for
628 * returning an error.
630 if ((rt
->rt_ifp
->if_flags
& IFF_LOOPBACK
) != 0) {
632 * See corresponding comments in ip6_output.
633 * XXX: but is it possible that ip6_forward() sends a packet
634 * to a loopback interface? I don't think so, and thus
635 * I bark here. (jinmei@kame.net)
636 * XXX: it is common to route invalid packets to loopback.
637 * also, the codepath will be visited on use of ::1 in
643 if ((rt
->rt_flags
& (RTF_BLACKHOLE
|RTF_REJECT
)) == 0)
646 printf("ip6_forward: outgoing interface is loopback. "
647 "src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
648 ip6_sprintf(&ip6
->ip6_src
),
649 ip6_sprintf(&ip6
->ip6_dst
),
650 ip6
->ip6_nxt
, if_name(m
->m_pkthdr
.rcvif
),
651 if_name(rt
->rt_ifp
));
654 /* we can just use rcvif in forwarding. */
655 origifp
= m
->m_pkthdr
.rcvif
;
658 origifp
= rt
->rt_ifp
;
660 * clear embedded scope identifiers if necessary.
661 * in6_clearscope will touch the addresses only when necessary.
663 in6_clearscope(&ip6
->ip6_src
);
664 in6_clearscope(&ip6
->ip6_dst
);
667 /* Drop the lock but retain the extra ref */
671 * If this is to be processed locally, let ip6_input have it.
674 VERIFY(m
->m_pkthdr
.aux_flags
& MAUXF_PROXY_DST
);
675 /* Release extra ref */
683 /* Invoke outbound packet filter */
684 error
= pf_af_hook(ifp
, NULL
, &m
, AF_INET6
, FALSE
, NULL
);
686 if (error
!= 0 || m
== NULL
) {
688 panic("%s: unexpected packet %p\n", __func__
, m
);
691 /* Already freed by callee */
694 ip6
= mtod(m
, struct ip6_hdr
*);
697 error
= nd6_output(ifp
, origifp
, m
, dst
, rt
, NULL
);
699 in6_ifstat_inc(ifp
, ifs6_out_discard
);
700 ip6stat
.ip6s_cantforward
++;
702 ip6stat
.ip6s_forward
++;
703 in6_ifstat_inc(ifp
, ifs6_out_forward
);
705 ip6stat
.ip6s_redirectsent
++;
716 /* Release extra ref */
723 if (type
== ND_REDIRECT
) {
724 icmp6_redirect_output(mcopy
, rt
);
725 /* Release extra ref */
733 /* xxx MTU is constant in PPP? */
737 /* Tell source to slow down like source quench in IP? */
740 case ENETUNREACH
: /* shouldn't happen, checked above */
745 type
= ICMP6_DST_UNREACH
;
746 code
= ICMP6_DST_UNREACH_ADDR
;
749 icmp6_error(mcopy
, type
, code
, 0);
750 /* Release extra ref */
756 /* Release extra ref */