1 /* $NetBSD: if_stf.c,v 1.72 2009/04/18 14:58:05 tsutsui Exp $ */
2 /* $KAME: if_stf.c,v 1.62 2001/06/07 22:32:16 itojun Exp $ */
5 * Copyright (C) 2000 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 * 6to4 interface, based on RFC3056.
36 * 6to4 interface is NOT capable of link-layer (I mean, IPv4) multicasting.
37 * There is no address mapping defined from IPv6 multicast address to IPv4
38 * address. Therefore, we do not have IFF_MULTICAST on the interface.
40 * Due to the lack of address mapping for link-local addresses, we cannot
41 * throw packets toward link-local addresses (fe80::x). Also, we cannot throw
42 * packets to link-local multicast addresses (ff02::x).
44 * Here are interesting symptoms due to the lack of link-local address:
46 * Unicast routing exchange:
47 * - RIPng: Impossible. Uses link-local multicast packet toward ff02::9,
48 * and link-local addresses as nexthop.
49 * - OSPFv6: Impossible. OSPFv6 assumes that there's link-local address
50 * assigned to the link, and makes use of them. Also, HELLO packets use
51 * link-local multicast addresses (ff02::5 and ff02::6).
52 * - BGP4+: Maybe. You can only use global address as nexthop, and global
53 * address as TCP endpoint address.
55 * Multicast routing protocols:
56 * - PIM: Hello packet cannot be used to discover adjacent PIM routers.
57 * Adjacent PIM routers must be configured manually (is it really spec-wise
58 * correct thing to do?).
61 * - Redirects cannot be used due to the lack of link-local address.
63 * stf interface does not have, and will not need, a link-local address.
64 * It seems to have no real benefit and does not help the above symptoms much.
65 * Even if we assign link-locals to interface, we cannot really
66 * use link-local unicast/multicast on top of 6to4 cloud (since there's no
67 * encapsulation defined for link-local address), and the above analysis does
68 * not change. RFC3056 does not mandate the assignment of link-local address
71 * 6to4 interface has security issues. Refer to
72 * http://playground.iijlab.net/i-d/draft-itojun-ipv6-transition-abuse-00.txt
73 * for details. The code tries to filter out some of malicious packets.
74 * Note that there is no way to be 100% secure.
77 #include <sys/cdefs.h>
78 __KERNEL_RCSID(0, "$NetBSD: if_stf.c,v 1.72 2009/04/18 14:58:05 tsutsui Exp $");
82 #include <sys/param.h>
83 #include <sys/systm.h>
84 #include <sys/socket.h>
85 #include <sys/sockio.h>
87 #include <sys/errno.h>
88 #include <sys/ioctl.h>
90 #include <sys/protosw.h>
91 #include <sys/queue.h>
92 #include <sys/syslog.h>
93 #include <sys/kauth.h>
98 #include <net/route.h>
99 #include <net/netisr.h>
100 #include <net/if_types.h>
101 #include <net/if_stf.h>
103 #include <netinet/in.h>
104 #include <netinet/in_systm.h>
105 #include <netinet/ip.h>
106 #include <netinet/ip_var.h>
107 #include <netinet/in_var.h>
109 #include <netinet/ip6.h>
110 #include <netinet6/ip6_var.h>
111 #include <netinet6/in6_gif.h>
112 #include <netinet6/in6_var.h>
113 #include <netinet/ip_ecn.h>
115 #include <netinet/ip_encap.h>
117 #include <machine/stdarg.h>
119 #include <net/net_osdep.h>
121 #include "bpfilter.h"
123 #include "gif.h" /*XXX*/
130 #include <net/if_gif.h>
133 #define IN6_IS_ADDR_6TO4(x) (ntohs((x)->s6_addr16[0]) == 0x2002)
134 #define GET_V4(x) ((const struct in_addr *)(&(x)->s6_addr16[1]))
137 struct ifnet sc_if
; /* common area */
139 const struct encaptab
*encap_cookie
;
140 LIST_ENTRY(stf_softc
) sc_list
;
143 static LIST_HEAD(, stf_softc
) stf_softc_list
;
145 static int stf_clone_create(struct if_clone
*, int);
146 static int stf_clone_destroy(struct ifnet
*);
148 struct if_clone stf_cloner
=
149 IF_CLONE_INITIALIZER("stf", stf_clone_create
, stf_clone_destroy
);
152 extern int ip_gif_ttl
; /*XXX*/
154 static int ip_gif_ttl
= 40; /*XXX*/
157 extern struct domain inetdomain
;
158 static const struct protosw in_stf_protosw
=
159 { SOCK_RAW
, &inetdomain
, IPPROTO_IPV6
, PR_ATOMIC
|PR_ADDR
,
160 in_stf_input
, rip_output
, 0, rip_ctloutput
,
167 static int stf_encapcheck(struct mbuf
*, int, int, void *);
168 static struct in6_ifaddr
*stf_getsrcifa6(struct ifnet
*);
169 static int stf_output(struct ifnet
*, struct mbuf
*, const struct sockaddr
*,
171 static int isrfc1918addr(const struct in_addr
*);
172 static int stf_checkaddr4(struct stf_softc
*, const struct in_addr
*,
174 static int stf_checkaddr6(struct stf_softc
*, const struct in6_addr
*,
176 static void stf_rtrequest(int, struct rtentry
*, const struct rt_addrinfo
*);
177 static int stf_ioctl(struct ifnet
*, u_long
, void *);
184 LIST_INIT(&stf_softc_list
);
185 if_clone_attach(&stf_cloner
);
189 stf_clone_create(struct if_clone
*ifc
, int unit
)
191 struct stf_softc
*sc
;
193 if (LIST_FIRST(&stf_softc_list
) != NULL
) {
194 /* Only one stf interface is allowed. */
198 sc
= malloc(sizeof(struct stf_softc
), M_DEVBUF
, M_WAIT
|M_ZERO
);
200 if_initname(&sc
->sc_if
, ifc
->ifc_name
, unit
);
202 sc
->encap_cookie
= encap_attach_func(AF_INET
, IPPROTO_IPV6
,
203 stf_encapcheck
, &in_stf_protosw
, sc
);
204 if (sc
->encap_cookie
== NULL
) {
205 printf("%s: unable to attach encap\n", if_name(&sc
->sc_if
));
207 return (EIO
); /* XXX */
210 sc
->sc_if
.if_mtu
= STF_MTU
;
211 sc
->sc_if
.if_flags
= 0;
212 sc
->sc_if
.if_ioctl
= stf_ioctl
;
213 sc
->sc_if
.if_output
= stf_output
;
214 sc
->sc_if
.if_type
= IFT_STF
;
215 sc
->sc_if
.if_dlt
= DLT_NULL
;
216 if_attach(&sc
->sc_if
);
217 if_alloc_sadl(&sc
->sc_if
);
219 bpfattach(&sc
->sc_if
, DLT_NULL
, sizeof(u_int
));
221 LIST_INSERT_HEAD(&stf_softc_list
, sc
, sc_list
);
226 stf_clone_destroy(struct ifnet
*ifp
)
228 struct stf_softc
*sc
= (void *) ifp
;
230 LIST_REMOVE(sc
, sc_list
);
231 encap_detach(sc
->encap_cookie
);
236 rtcache_free(&sc
->sc_ro
);
243 stf_encapcheck(struct mbuf
*m
, int off
, int proto
, void *arg
)
246 struct in6_ifaddr
*ia6
;
247 struct stf_softc
*sc
;
250 sc
= (struct stf_softc
*)arg
;
254 if ((sc
->sc_if
.if_flags
& IFF_UP
) == 0)
257 /* IFF_LINK0 means "no decapsulation" */
258 if ((sc
->sc_if
.if_flags
& IFF_LINK0
) != 0)
261 if (proto
!= IPPROTO_IPV6
)
264 m_copydata(m
, 0, sizeof(ip
), (void *)&ip
);
269 ia6
= stf_getsrcifa6(&sc
->sc_if
);
274 * check if IPv4 dst matches the IPv4 address derived from the
275 * local 6to4 address.
276 * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
278 if (memcmp(GET_V4(&ia6
->ia_addr
.sin6_addr
), &ip
.ip_dst
,
279 sizeof(ip
.ip_dst
)) != 0)
283 * check if IPv4 src matches the IPv4 address derived from the
284 * local 6to4 address masked by prefixmask.
285 * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
286 * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
288 memset(&a
, 0, sizeof(a
));
289 a
.s_addr
= GET_V4(&ia6
->ia_addr
.sin6_addr
)->s_addr
;
290 a
.s_addr
&= GET_V4(&ia6
->ia_prefixmask
.sin6_addr
)->s_addr
;
292 b
.s_addr
&= GET_V4(&ia6
->ia_prefixmask
.sin6_addr
)->s_addr
;
293 if (a
.s_addr
!= b
.s_addr
)
296 /* stf interface makes single side match only */
300 static struct in6_ifaddr
*
301 stf_getsrcifa6(struct ifnet
*ifp
)
304 struct in_ifaddr
*ia4
;
305 struct sockaddr_in6
*sin6
;
308 IFADDR_FOREACH(ifa
, ifp
)
310 if (ifa
->ifa_addr
== NULL
)
312 if (ifa
->ifa_addr
->sa_family
!= AF_INET6
)
314 sin6
= (struct sockaddr_in6
*)ifa
->ifa_addr
;
315 if (!IN6_IS_ADDR_6TO4(&sin6
->sin6_addr
))
318 memcpy(&in
, GET_V4(&sin6
->sin6_addr
), sizeof(in
));
319 INADDR_TO_IA(in
, ia4
);
323 return (struct in6_ifaddr
*)ifa
;
330 stf_output(struct ifnet
*ifp
, struct mbuf
*m
, const struct sockaddr
*dst
,
334 struct stf_softc
*sc
;
335 const struct sockaddr_in6
*dst6
;
336 const struct in_addr
*in4
;
340 struct in6_ifaddr
*ia6
;
343 struct sockaddr_in dst4
;
346 sc
= (struct stf_softc
*)ifp
;
347 dst6
= (const struct sockaddr_in6
*)dst
;
350 if ((ifp
->if_flags
& IFF_UP
) == 0) {
356 * If we don't have an ip4 address that match my inner ip6 address,
357 * we shouldn't generate output. Without this check, we'll end up
358 * using wrong IPv4 source.
360 ia6
= stf_getsrcifa6(ifp
);
367 if (m
->m_len
< sizeof(*ip6
)) {
368 m
= m_pullup(m
, sizeof(*ip6
));
374 ip6
= mtod(m
, struct ip6_hdr
*);
375 tos
= (ntohl(ip6
->ip6_flow
) >> 20) & 0xff;
378 * Pickup the right outer dst addr from the list of candidates.
379 * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
381 if (IN6_IS_ADDR_6TO4(&ip6
->ip6_dst
))
382 in4
= GET_V4(&ip6
->ip6_dst
);
383 else if (IN6_IS_ADDR_6TO4(&dst6
->sin6_addr
))
384 in4
= GET_V4(&dst6
->sin6_addr
);
393 bpf_mtap_af(ifp
->if_bpf
, AF_INET6
, m
);
394 #endif /*NBPFILTER > 0*/
396 M_PREPEND(m
, sizeof(struct ip
), M_DONTWAIT
);
397 if (m
&& m
->m_len
< sizeof(struct ip
))
398 m
= m_pullup(m
, sizeof(struct ip
));
403 ip
= mtod(m
, struct ip
*);
405 memset(ip
, 0, sizeof(*ip
));
407 bcopy(GET_V4(&((struct sockaddr_in6
*)&ia6
->ia_addr
)->sin6_addr
),
408 &ip
->ip_src
, sizeof(ip
->ip_src
));
409 memcpy(&ip
->ip_dst
, in4
, sizeof(ip
->ip_dst
));
410 ip
->ip_p
= IPPROTO_IPV6
;
411 ip
->ip_ttl
= ip_gif_ttl
; /*XXX*/
412 ip
->ip_len
= htons(m
->m_pkthdr
.len
);
413 if (ifp
->if_flags
& IFF_LINK1
)
414 ip_ecn_ingress(ECN_ALLOWED
, &ip
->ip_tos
, &tos
);
416 ip_ecn_ingress(ECN_NOCARE
, &ip
->ip_tos
, &tos
);
418 sockaddr_in_init(&u
.dst4
, &ip
->ip_dst
, 0);
419 if ((rt
= rtcache_lookup(&sc
->sc_ro
, &u
.dst
)) == NULL
) {
425 /* If the route constitutes infinite encapsulation, punt. */
426 if (rt
->rt_ifp
== ifp
) {
427 rtcache_free(&sc
->sc_ro
);
434 ifp
->if_obytes
+= m
->m_pkthdr
.len
- sizeof(struct ip
);
435 return ip_output(m
, NULL
, &sc
->sc_ro
, 0, NULL
, NULL
);
439 isrfc1918addr(const struct in_addr
*in
)
442 * returns 1 if private address range:
443 * 10.0.0.0/8 172.16.0.0/12 192.168.0.0/16
445 if ((ntohl(in
->s_addr
) & 0xff000000) >> 24 == 10 ||
446 (ntohl(in
->s_addr
) & 0xfff00000) >> 16 == 172 * 256 + 16 ||
447 (ntohl(in
->s_addr
) & 0xffff0000) >> 16 == 192 * 256 + 168)
454 stf_checkaddr4(struct stf_softc
*sc
, const struct in_addr
*in
,
455 struct ifnet
*inifp
/*incoming interface*/)
457 struct in_ifaddr
*ia4
;
460 * reject packets with the following address:
461 * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
463 if (IN_MULTICAST(in
->s_addr
))
465 switch ((ntohl(in
->s_addr
) & 0xff000000) >> 24) {
466 case 0: case 127: case 255:
471 * reject packets with private address range.
472 * (requirement from RFC3056 section 2 1st paragraph)
474 if (isrfc1918addr(in
))
478 * reject packet with IPv4 link-local (169.254.0.0/16),
479 * as suggested in draft-savola-v6ops-6to4-security-00.txt
481 if (((ntohl(in
->s_addr
) & 0xff000000) >> 24) == 169 &&
482 ((ntohl(in
->s_addr
) & 0x00ff0000) >> 16) == 254)
486 * reject packets with broadcast
488 TAILQ_FOREACH(ia4
, &in_ifaddrhead
, ia_list
)
490 if ((ia4
->ia_ifa
.ifa_ifp
->if_flags
& IFF_BROADCAST
) == 0)
492 if (in
->s_addr
== ia4
->ia_broadaddr
.sin_addr
.s_addr
)
497 * perform ingress filter
499 if (sc
&& (sc
->sc_if
.if_flags
& IFF_LINK2
) == 0 && inifp
) {
500 struct sockaddr_in sin
;
503 memset(&sin
, 0, sizeof(sin
));
504 sin
.sin_family
= AF_INET
;
505 sin
.sin_len
= sizeof(struct sockaddr_in
);
507 rt
= rtalloc1((struct sockaddr
*)&sin
, 0);
508 if (!rt
|| rt
->rt_ifp
!= inifp
) {
510 log(LOG_WARNING
, "%s: packet from 0x%x dropped "
511 "due to ingress filter\n", if_name(&sc
->sc_if
),
512 (uint32_t)ntohl(sin
.sin_addr
.s_addr
));
525 stf_checkaddr6(struct stf_softc
*sc
, const struct in6_addr
*in6
,
526 struct ifnet
*inifp
/*incoming interface*/)
530 * check 6to4 addresses
532 if (IN6_IS_ADDR_6TO4(in6
))
533 return stf_checkaddr4(sc
, GET_V4(in6
), inifp
);
536 * reject anything that look suspicious. the test is implemented
537 * in ip6_input too, but we check here as well to
538 * (1) reject bad packets earlier, and
539 * (2) to be safe against future ip6_input change.
541 if (IN6_IS_ADDR_V4COMPAT(in6
) || IN6_IS_ADDR_V4MAPPED(in6
))
545 * reject link-local and site-local unicast
546 * as suggested in draft-savola-v6ops-6to4-security-00.txt
548 if (IN6_IS_ADDR_LINKLOCAL(in6
) || IN6_IS_ADDR_SITELOCAL(in6
))
552 * reject node-local and link-local multicast
553 * as suggested in draft-savola-v6ops-6to4-security-00.txt
555 if (IN6_IS_ADDR_MC_NODELOCAL(in6
) || IN6_IS_ADDR_MC_LINKLOCAL(in6
))
562 in_stf_input(struct mbuf
*m
, ...)
565 struct stf_softc
*sc
;
570 struct ifqueue
*ifq
= NULL
;
575 off
= va_arg(ap
, int);
576 proto
= va_arg(ap
, int);
579 if (proto
!= IPPROTO_IPV6
) {
584 ip
= mtod(m
, struct ip
*);
586 sc
= (struct stf_softc
*)encap_getarg(m
);
588 if (sc
== NULL
|| (sc
->sc_if
.if_flags
& IFF_UP
) == 0) {
596 * perform sanity check against outer src/dst.
597 * for source, perform ingress filter as well.
599 if (stf_checkaddr4(sc
, &ip
->ip_dst
, NULL
) < 0 ||
600 stf_checkaddr4(sc
, &ip
->ip_src
, m
->m_pkthdr
.rcvif
) < 0) {
608 if (m
->m_len
< sizeof(*ip6
)) {
609 m
= m_pullup(m
, sizeof(*ip6
));
613 ip6
= mtod(m
, struct ip6_hdr
*);
616 * perform sanity check against inner src/dst.
617 * for source, perform ingress filter as well.
619 if (stf_checkaddr6(sc
, &ip6
->ip6_dst
, NULL
) < 0 ||
620 stf_checkaddr6(sc
, &ip6
->ip6_src
, m
->m_pkthdr
.rcvif
) < 0) {
625 itos
= (ntohl(ip6
->ip6_flow
) >> 20) & 0xff;
626 if ((ifp
->if_flags
& IFF_LINK1
) != 0)
627 ip_ecn_egress(ECN_ALLOWED
, &otos
, &itos
);
629 ip_ecn_egress(ECN_NOCARE
, &otos
, &itos
);
630 ip6
->ip6_flow
&= ~htonl(0xff << 20);
631 ip6
->ip6_flow
|= htonl((uint32_t)itos
<< 20);
633 m
->m_pkthdr
.rcvif
= ifp
;
637 bpf_mtap_af(ifp
->if_bpf
, AF_INET6
, m
);
638 #endif /*NBPFILTER > 0*/
641 * Put the packet to the network layer input queue according to the
642 * specified address family.
643 * See net/if_gif.c for possible issues with packet processing
644 * reorder due to extra queueing.
651 IF_DROP(ifq
); /* update statistics */
659 ifp
->if_ibytes
+= m
->m_pkthdr
.len
;
665 stf_rtrequest(int cmd
, struct rtentry
*rt
,
666 const struct rt_addrinfo
*info
)
669 struct stf_softc
*sc
;
671 sc
= LIST_FIRST(&stf_softc_list
);
672 rt
->rt_rmx
.rmx_mtu
= (sc
!= NULL
) ? sc
->sc_if
.if_mtu
: STF_MTU
;
677 stf_ioctl(struct ifnet
*ifp
, u_long cmd
, void *data
)
679 struct lwp
*l
= curlwp
; /* XXX */
681 struct ifreq
*ifr
= data
;
682 struct sockaddr_in6
*sin6
;
688 ifa
= (struct ifaddr
*)data
;
689 if (ifa
== NULL
|| ifa
->ifa_addr
->sa_family
!= AF_INET6
) {
690 error
= EAFNOSUPPORT
;
693 sin6
= (struct sockaddr_in6
*)ifa
->ifa_addr
;
694 if (IN6_IS_ADDR_6TO4(&sin6
->sin6_addr
) &&
695 !isrfc1918addr(GET_V4(&sin6
->sin6_addr
))) {
696 ifa
->ifa_rtrequest
= stf_rtrequest
;
697 ifp
->if_flags
|= IFF_UP
;
705 ifreq_getaddr(cmd
, ifr
)->sa_family
== AF_INET6
)
708 error
= EAFNOSUPPORT
;
712 error
= kauth_authorize_network(l
->l_cred
,
713 KAUTH_NETWORK_INTERFACE
,
714 KAUTH_REQ_NETWORK_INTERFACE_SETPRIV
, ifp
, KAUTH_ARG(cmd
),
718 if (ifr
->ifr_mtu
< STF_MTU_MIN
|| ifr
->ifr_mtu
> STF_MTU_MAX
)
720 else if ((error
= ifioctl_common(ifp
, cmd
, data
)) == ENETRESET
)
725 error
= ifioctl_common(ifp
, cmd
, data
);