1 /* $FreeBSD: src/sys/net/if_stf.c,v 1.1.2.11 2003/01/23 21:06:44 sam Exp $ */
2 /* $DragonFly: src/sys/net/stf/if_stf.c,v 1.25 2008/10/27 02:56:30 sephe Exp $ */
3 /* $KAME: if_stf.c,v 1.73 2001/12/03 11:08:30 keiichi Exp $ */
6 * Copyright (C) 2000 WIDE Project.
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. Neither the name of the project nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 * 6to4 interface, based on RFC3056.
37 * 6to4 interface is NOT capable of link-layer (I mean, IPv4) multicasting.
38 * There is no address mapping defined from IPv6 multicast address to IPv4
39 * address. Therefore, we do not have IFF_MULTICAST on the interface.
41 * Due to the lack of address mapping for link-local addresses, we cannot
42 * throw packets toward link-local addresses (fe80::x). Also, we cannot throw
43 * packets to link-local multicast addresses (ff02::x).
45 * Here are interesting symptoms due to the lack of link-local address:
47 * Unicast routing exchange:
48 * - RIPng: Impossible. Uses link-local multicast packet toward ff02::9,
49 * and link-local addresses as nexthop.
50 * - OSPFv6: Impossible. OSPFv6 assumes that there's link-local address
51 * assigned to the link, and makes use of them. Also, HELLO packets use
52 * link-local multicast addresses (ff02::5 and ff02::6).
53 * - BGP4+: Maybe. You can only use global address as nexthop, and global
54 * address as TCP endpoint address.
56 * Multicast routing protocols:
57 * - PIM: Hello packet cannot be used to discover adjacent PIM routers.
58 * Adjacent PIM routers must be configured manually (is it really spec-wise
59 * correct thing to do?).
62 * - Redirects cannot be used due to the lack of link-local address.
64 * stf interface does not have, and will not need, a link-local address.
65 * It seems to have no real benefit and does not help the above symptoms much.
66 * Even if we assign link-locals to interface, we cannot really
67 * use link-local unicast/multicast on top of 6to4 cloud (since there's no
68 * encapsulation defined for link-local address), and the above analysis does
69 * not change. RFC3056 does not mandate the assignment of link-local address
72 * 6to4 interface has security issues. Refer to
73 * http://playground.iijlab.net/i-d/draft-itojun-ipv6-transition-abuse-00.txt
74 * for details. The code tries to filter out some of malicious packets.
75 * Note that there is no way to be 100% secure.
79 #include "opt_inet6.h"
81 #include <sys/param.h>
82 #include <sys/systm.h>
83 #include <sys/socket.h>
84 #include <sys/sockio.h>
86 #include <sys/errno.h>
87 #include <sys/protosw.h>
88 #include <sys/kernel.h>
89 #include <machine/cpu.h>
91 #include <sys/malloc.h>
94 #include <net/route.h>
95 #include <net/netisr.h>
96 #include <net/if_types.h>
99 #include <netinet/in.h>
100 #include <netinet/in_systm.h>
101 #include <netinet/ip.h>
102 #include <netinet/ip_var.h>
103 #include <netinet/in_var.h>
105 #include <netinet/ip6.h>
106 #include <netinet6/ip6_var.h>
107 #include <netinet6/in6_var.h>
108 #include <netinet/ip_ecn.h>
110 #include <netinet/ip_encap.h>
112 #include <machine/stdarg.h>
114 #include <net/net_osdep.h>
118 #define IN6_IS_ADDR_6TO4(x) (ntohs((x)->s6_addr16[0]) == 0x2002)
119 #define GET_V4(x) ((struct in_addr *)(&(x)->s6_addr16[1]))
122 struct ifnet sc_if; /* common area */
124 struct route __sc_ro4;
125 struct route_in6 __sc_ro6; /* just for safety */
127 #define sc_ro __sc_ro46.__sc_ro4
128 const struct encaptab *encap_cookie;
131 static struct stf_softc *stf;
133 static MALLOC_DEFINE(M_STF, "stf", "6to4 Tunnel Interface");
134 static int ip_stf_ttl = 40;
136 extern struct domain inetdomain;
137 struct protosw in_stf_protosw =
140 .pr_domain = &inetdomain,
141 .pr_protocol = IPPROTO_IPV6,
142 .pr_flags = PR_ATOMIC|PR_ADDR,
144 .pr_input = in_stf_input,
145 .pr_output = rip_output,
147 .pr_ctloutput = rip_ctloutput,
149 .pr_usrreqs = &rip_usrreqs
152 static int stfmodevent (module_t, int, void *);
153 static int stf_encapcheck (const struct mbuf *, int, int, void *);
154 static struct in6_ifaddr *stf_getsrcifa6 (struct ifnet *);
155 static int stf_output (struct ifnet *, struct mbuf *, struct sockaddr *,
157 static int stf_checkaddr4 (struct stf_softc *, struct in_addr *,
159 static int stf_checkaddr6 (struct stf_softc *, struct in6_addr *,
161 static void stf_rtrequest (int, struct rtentry *, struct rt_addrinfo *);
162 static int stf_ioctl (struct ifnet *, u_long, caddr_t, struct ucred *);
165 stfmodevent(module_t mod, int type, void *data)
167 struct stf_softc *sc;
169 const struct encaptab *p;
173 stf = kmalloc(sizeof(struct stf_softc), M_STF,
177 bzero(sc, sizeof(*sc));
178 if_initname(&(sc->sc_if), "stf", 0);
180 p = encap_attach_func(AF_INET, IPPROTO_IPV6, stf_encapcheck,
181 (void *)&in_stf_protosw, sc);
183 kprintf("%s: attach failed\n", if_name(&sc->sc_if));
186 sc->encap_cookie = p;
188 sc->sc_if.if_mtu = IPV6_MMTU;
189 sc->sc_if.if_flags = 0;
190 sc->sc_if.if_ioctl = stf_ioctl;
191 sc->sc_if.if_output = stf_output;
192 sc->sc_if.if_type = IFT_STF;
194 /* turn off ingress filter */
195 sc->sc_if.if_flags |= IFF_LINK2;
197 sc->sc_if.if_snd.ifq_maxlen = IFQ_MAXLEN;
198 if_attach(&sc->sc_if, NULL);
199 bpfattach(&sc->sc_if, DLT_NULL, sizeof(u_int));
203 bpfdetach(&sc->sc_if);
204 if_detach(&sc->sc_if);
205 err = encap_detach(sc->encap_cookie);
206 KASSERT(err == 0, ("Unexpected error detaching encap_cookie"));
214 static moduledata_t stf_mod = {
220 DECLARE_MODULE(if_stf, stf_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
223 stf_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
226 struct in6_ifaddr *ia6;
227 struct stf_softc *sc;
230 sc = (struct stf_softc *)arg;
234 if ((sc->sc_if.if_flags & IFF_UP) == 0)
237 /* IFF_LINK0 means "no decapsulation" */
238 if ((sc->sc_if.if_flags & IFF_LINK0) != 0)
241 if (proto != IPPROTO_IPV6)
244 m_copydata(m, 0, sizeof(ip), (caddr_t)&ip);
249 ia6 = stf_getsrcifa6(&sc->sc_if);
254 * check if IPv4 dst matches the IPv4 address derived from the
255 * local 6to4 address.
256 * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
258 if (bcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst,
259 sizeof(ip.ip_dst)) != 0)
263 * check if IPv4 src matches the IPv4 address derived from the
264 * local 6to4 address masked by prefixmask.
265 * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
266 * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
268 bzero(&a, sizeof(a));
269 a.s_addr = GET_V4(&ia6->ia_addr.sin6_addr)->s_addr;
270 a.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
272 b.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
273 if (a.s_addr != b.s_addr)
276 /* stf interface makes single side match only */
280 static struct in6_ifaddr *
281 stf_getsrcifa6(struct ifnet *ifp)
283 struct ifaddr_container *ifac;
284 struct sockaddr_in6 *sin6;
287 TAILQ_FOREACH(ifac, &ifp->if_addrheads[mycpuid], ifa_link) {
288 struct ifaddr *ia = ifac->ifa;
289 struct in_ifaddr_container *iac;
291 if (ia->ifa_addr == NULL)
293 if (ia->ifa_addr->sa_family != AF_INET6)
295 sin6 = (struct sockaddr_in6 *)ia->ifa_addr;
296 if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr))
299 bcopy(GET_V4(&sin6->sin6_addr), &in, sizeof(in));
300 LIST_FOREACH(iac, INADDR_HASH(in.s_addr), ia_hash) {
301 if (iac->ia->ia_addr.sin_addr.s_addr == in.s_addr)
307 return (struct in6_ifaddr *)ia;
314 stf_output_serialized(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst,
317 struct stf_softc *sc;
318 struct sockaddr_in6 *dst6;
320 struct sockaddr_in *dst4;
324 struct in6_ifaddr *ia6;
325 static const uint32_t af = AF_INET6;
327 sc = (struct stf_softc*)ifp;
328 dst6 = (struct sockaddr_in6 *)dst;
331 if ((ifp->if_flags & IFF_UP) == 0) {
337 * If we don't have an ip4 address that match my inner ip6 address,
338 * we shouldn't generate output. Without this check, we'll end up
339 * using wrong IPv4 source.
341 ia6 = stf_getsrcifa6(ifp);
347 if (m->m_len < sizeof(*ip6)) {
348 m = m_pullup(m, sizeof(*ip6));
352 ip6 = mtod(m, struct ip6_hdr *);
353 tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
356 * Pickup the right outer dst addr from the list of candidates.
357 * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
359 if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst))
360 in4 = GET_V4(&ip6->ip6_dst);
361 else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr))
362 in4 = GET_V4(&dst6->sin6_addr);
369 bpf_ptap(ifp->if_bpf, m, &af, sizeof(af));
371 M_PREPEND(m, sizeof(struct ip), MB_DONTWAIT);
372 if (m && m->m_len < sizeof(struct ip))
373 m = m_pullup(m, sizeof(struct ip));
376 ip = mtod(m, struct ip *);
378 bzero(ip, sizeof(*ip));
380 bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr),
381 &ip->ip_src, sizeof(ip->ip_src));
382 bcopy(in4, &ip->ip_dst, sizeof(ip->ip_dst));
383 ip->ip_p = IPPROTO_IPV6;
384 ip->ip_ttl = ip_stf_ttl;
385 ip->ip_len = m->m_pkthdr.len; /*host order*/
386 if (ifp->if_flags & IFF_LINK1)
387 ip_ecn_ingress(ECN_ALLOWED, &ip->ip_tos, &tos);
389 ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
391 dst4 = (struct sockaddr_in *)&sc->sc_ro.ro_dst;
392 if (dst4->sin_family != AF_INET ||
393 bcmp(&dst4->sin_addr, &ip->ip_dst, sizeof(ip->ip_dst)) != 0) {
394 /* cache route doesn't match */
395 dst4->sin_family = AF_INET;
396 dst4->sin_len = sizeof(struct sockaddr_in);
397 bcopy(&ip->ip_dst, &dst4->sin_addr, sizeof(dst4->sin_addr));
398 if (sc->sc_ro.ro_rt) {
399 RTFREE(sc->sc_ro.ro_rt);
400 sc->sc_ro.ro_rt = NULL;
404 if (sc->sc_ro.ro_rt == NULL) {
406 if (sc->sc_ro.ro_rt == NULL) {
412 return ip_output(m, NULL, &sc->sc_ro, 0, NULL, NULL);
416 stf_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst,
421 ifnet_serialize_tx(ifp);
422 error = stf_output_serialized(ifp, m, dst, rt);
423 ifnet_deserialize_tx(ifp);
430 * inifp: incoming interface
433 stf_checkaddr4(struct stf_softc *sc, struct in_addr *in, struct ifnet *inifp)
435 struct in_ifaddr_container *iac;
438 * reject packets with the following address:
439 * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
441 if (IN_MULTICAST(ntohl(in->s_addr)))
443 switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
444 case 0: case 127: case 255:
449 * reject packets with broadcast
451 TAILQ_FOREACH(iac, &in_ifaddrheads[mycpuid], ia_link) {
452 struct in_ifaddr *ia4 = iac->ia;
454 if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0)
456 if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr)
461 * perform ingress filter
463 if (sc && (sc->sc_if.if_flags & IFF_LINK2) == 0 && inifp) {
464 struct sockaddr_in sin;
467 bzero(&sin, sizeof(sin));
468 sin.sin_family = AF_INET;
469 sin.sin_len = sizeof(struct sockaddr_in);
471 rt = rtpurelookup((struct sockaddr *)&sin);
472 if (!rt || rt->rt_ifp != inifp) {
474 log(LOG_WARNING, "%s: packet from 0x%x dropped "
475 "due to ingress filter\n", if_name(&sc->sc_if),
476 (u_int32_t)ntohl(sin.sin_addr.s_addr));
490 * inifp: incoming interface
493 stf_checkaddr6(struct stf_softc *sc, struct in6_addr *in6, struct ifnet *inifp)
496 * check 6to4 addresses
498 if (IN6_IS_ADDR_6TO4(in6))
499 return stf_checkaddr4(sc, GET_V4(in6), inifp);
502 * reject anything that look suspicious. the test is implemented
503 * in ip6_input too, but we check here as well to
504 * (1) reject bad packets earlier, and
505 * (2) to be safe against future ip6_input change.
507 if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6))
514 in_stf_input(struct mbuf **mp, int *offp, int proto)
517 struct stf_softc *sc;
523 static const uint32_t af = AF_INET6;
528 if (proto != IPPROTO_IPV6) {
530 return(IPPROTO_DONE);
533 ip = mtod(m, struct ip *);
535 sc = (struct stf_softc *)encap_getarg(m);
537 if (sc == NULL || (sc->sc_if.if_flags & IFF_UP) == 0) {
539 return(IPPROTO_DONE);
545 * perform sanity check against outer src/dst.
546 * for source, perform ingress filter as well.
548 if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
549 stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
551 return(IPPROTO_DONE);
557 if (m->m_len < sizeof(*ip6)) {
558 m = m_pullup(m, sizeof(*ip6));
560 return(IPPROTO_DONE);
562 ip6 = mtod(m, struct ip6_hdr *);
565 * perform sanity check against inner src/dst.
566 * for source, perform ingress filter as well.
568 if (stf_checkaddr6(sc, &ip6->ip6_dst, NULL) < 0 ||
569 stf_checkaddr6(sc, &ip6->ip6_src, m->m_pkthdr.rcvif) < 0) {
571 return(IPPROTO_DONE);
574 itos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
575 if ((ifp->if_flags & IFF_LINK1) != 0)
576 ip_ecn_egress(ECN_ALLOWED, &otos, &itos);
578 ip_ecn_egress(ECN_NOCARE, &otos, &itos);
579 ip6->ip6_flow &= ~htonl(0xff << 20);
580 ip6->ip6_flow |= htonl((u_int32_t)itos << 20);
582 m->m_pkthdr.rcvif = ifp;
585 bpf_ptap(ifp->if_bpf, m, &af, sizeof(af));
588 * Put the packet to the network layer input queue according to the
589 * specified address family.
590 * See net/if_gif.c for possible issues with packet processing
591 * reorder due to extra queueing.
594 ifp->if_ibytes += m->m_pkthdr.len;
595 netisr_queue(NETISR_IPV6, m);
596 return(IPPROTO_DONE);
601 stf_rtrequest(int cmd, struct rtentry *rt, struct rt_addrinfo *info)
605 rt->rt_rmx.rmx_mtu = IPV6_MMTU;
609 stf_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data, struct ucred *cr)
613 struct sockaddr_in6 *sin6;
619 ifa = (struct ifaddr *)data;
620 if (ifa == NULL || ifa->ifa_addr->sa_family != AF_INET6) {
621 error = EAFNOSUPPORT;
624 sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
625 if (IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
626 ifa->ifa_rtrequest = stf_rtrequest;
627 ifp->if_flags |= IFF_UP;
634 ifr = (struct ifreq *)data;
635 if (ifr && ifr->ifr_addr.sa_family == AF_INET6)
638 error = EAFNOSUPPORT;