xref: /netbsd-src/sys/net/if_stf.c (revision fad4c9f71477ae11cea2ee75ec82151ac770a534)
1 /*	$NetBSD: if_stf.c,v 1.51 2006/05/14 21:19:33 elad Exp $	*/
2 /*	$KAME: if_stf.c,v 1.62 2001/06/07 22:32:16 itojun Exp $	*/
3 
4 /*
5  * Copyright (C) 2000 WIDE Project.
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
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.
19  *
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
30  * SUCH DAMAGE.
31  */
32 
33 /*
34  * 6to4 interface, based on RFC3056.
35  *
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.
39  *
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).
43  *
44  * Here are interesting symptoms due to the lack of link-local address:
45  *
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.
54  *
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?).
59  *
60  * ICMPv6:
61  * - Redirects cannot be used due to the lack of link-local address.
62  *
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
69  * either.
70  *
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.
75  */
76 
77 #include <sys/cdefs.h>
78 __KERNEL_RCSID(0, "$NetBSD: if_stf.c,v 1.51 2006/05/14 21:19:33 elad Exp $");
79 
80 #include "opt_inet.h"
81 
82 #include <sys/param.h>
83 #include <sys/systm.h>
84 #include <sys/socket.h>
85 #include <sys/sockio.h>
86 #include <sys/mbuf.h>
87 #include <sys/errno.h>
88 #include <sys/ioctl.h>
89 #include <sys/proc.h>
90 #include <sys/protosw.h>
91 #include <sys/queue.h>
92 #include <sys/syslog.h>
93 #include <sys/kauth.h>
94 
95 #include <machine/cpu.h>
96 
97 #include <net/if.h>
98 #include <net/route.h>
99 #include <net/netisr.h>
100 #include <net/if_types.h>
101 #include <net/if_stf.h>
102 
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>
108 
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>
114 
115 #include <netinet/ip_encap.h>
116 
117 #include <machine/stdarg.h>
118 
119 #include <net/net_osdep.h>
120 
121 #include "bpfilter.h"
122 #include "stf.h"
123 #include "gif.h"	/*XXX*/
124 
125 #if NBPFILTER > 0
126 #include <net/bpf.h>
127 #endif
128 
129 #if NGIF > 0
130 #include <net/if_gif.h>
131 #endif
132 
133 #define IN6_IS_ADDR_6TO4(x)	(ntohs((x)->s6_addr16[0]) == 0x2002)
134 #define GET_V4(x)	((struct in_addr *)(&(x)->s6_addr16[1]))
135 
136 struct stf_softc {
137 	struct ifnet	sc_if;	   /* common area */
138 	union {
139 		struct route  __sc_ro4;
140 		struct route_in6 __sc_ro6; /* just for safety */
141 	} __sc_ro46;
142 #define sc_ro	__sc_ro46.__sc_ro4
143 	const struct encaptab *encap_cookie;
144 	LIST_ENTRY(stf_softc) sc_list;
145 };
146 
147 static LIST_HEAD(, stf_softc) stf_softc_list;
148 
149 static int	stf_clone_create(struct if_clone *, int);
150 static int	stf_clone_destroy(struct ifnet *);
151 
152 struct if_clone stf_cloner =
153     IF_CLONE_INITIALIZER("stf", stf_clone_create, stf_clone_destroy);
154 
155 #if NGIF > 0
156 extern int ip_gif_ttl;	/*XXX*/
157 #else
158 static int ip_gif_ttl = 40;	/*XXX*/
159 #endif
160 
161 extern struct domain inetdomain;
162 static const struct protosw in_stf_protosw =
163 { SOCK_RAW,	&inetdomain,	IPPROTO_IPV6,	PR_ATOMIC|PR_ADDR,
164   in_stf_input, rip_output,	0,		rip_ctloutput,
165   rip_usrreq,
166   0,            0,              0,              0
167 };
168 
169 void	stfattach(int);
170 
171 static int stf_encapcheck(struct mbuf *, int, int, void *);
172 static struct in6_ifaddr *stf_getsrcifa6(struct ifnet *);
173 static int stf_output(struct ifnet *, struct mbuf *, struct sockaddr *,
174 	struct rtentry *);
175 static int isrfc1918addr(struct in_addr *);
176 static int stf_checkaddr4(struct stf_softc *, struct in_addr *,
177 	struct ifnet *);
178 static int stf_checkaddr6(struct stf_softc *, struct in6_addr *,
179 	struct ifnet *);
180 static void stf_rtrequest(int, struct rtentry *, struct rt_addrinfo *);
181 static int stf_ioctl(struct ifnet *, u_long, caddr_t);
182 
183 /* ARGSUSED */
184 void
185 stfattach(int count)
186 {
187 
188 	LIST_INIT(&stf_softc_list);
189 	if_clone_attach(&stf_cloner);
190 }
191 
192 static int
193 stf_clone_create(struct if_clone *ifc, int unit)
194 {
195 	struct stf_softc *sc;
196 
197 	if (LIST_FIRST(&stf_softc_list) != NULL) {
198 		/* Only one stf interface is allowed. */
199 		return (EEXIST);
200 	}
201 
202 	sc = malloc(sizeof(struct stf_softc), M_DEVBUF, M_WAIT);
203 	memset(sc, 0, sizeof(struct stf_softc));
204 
205 	snprintf(sc->sc_if.if_xname, sizeof(sc->sc_if.if_xname), "%s%d",
206 	    ifc->ifc_name, unit);
207 
208 	sc->encap_cookie = encap_attach_func(AF_INET, IPPROTO_IPV6,
209 	    stf_encapcheck, &in_stf_protosw, sc);
210 	if (sc->encap_cookie == NULL) {
211 		printf("%s: unable to attach encap\n", if_name(&sc->sc_if));
212 		free(sc, M_DEVBUF);
213 		return (EIO);	/* XXX */
214 	}
215 
216 	sc->sc_if.if_mtu    = STF_MTU;
217 	sc->sc_if.if_flags  = 0;
218 	sc->sc_if.if_ioctl  = stf_ioctl;
219 	sc->sc_if.if_output = stf_output;
220 	sc->sc_if.if_type   = IFT_STF;
221 	sc->sc_if.if_dlt    = DLT_NULL;
222 	if_attach(&sc->sc_if);
223 	if_alloc_sadl(&sc->sc_if);
224 #if NBPFILTER > 0
225 	bpfattach(&sc->sc_if, DLT_NULL, sizeof(u_int));
226 #endif
227 	LIST_INSERT_HEAD(&stf_softc_list, sc, sc_list);
228 	return (0);
229 }
230 
231 static int
232 stf_clone_destroy(struct ifnet *ifp)
233 {
234 	struct stf_softc *sc = (void *) ifp;
235 
236 	LIST_REMOVE(sc, sc_list);
237 	encap_detach(sc->encap_cookie);
238 #if NBPFILTER > 0
239 	bpfdetach(ifp);
240 #endif
241 	if_detach(ifp);
242 	free(sc, M_DEVBUF);
243 
244 	return (0);
245 }
246 
247 static int
248 stf_encapcheck(struct mbuf *m, int off, int proto, void *arg)
249 {
250 	struct ip ip;
251 	struct in6_ifaddr *ia6;
252 	struct stf_softc *sc;
253 	struct in_addr a, b;
254 
255 	sc = (struct stf_softc *)arg;
256 	if (sc == NULL)
257 		return 0;
258 
259 	if ((sc->sc_if.if_flags & IFF_UP) == 0)
260 		return 0;
261 
262 	/* IFF_LINK0 means "no decapsulation" */
263 	if ((sc->sc_if.if_flags & IFF_LINK0) != 0)
264 		return 0;
265 
266 	if (proto != IPPROTO_IPV6)
267 		return 0;
268 
269 	m_copydata(m, 0, sizeof(ip), (caddr_t)&ip);
270 
271 	if (ip.ip_v != 4)
272 		return 0;
273 
274 	ia6 = stf_getsrcifa6(&sc->sc_if);
275 	if (ia6 == NULL)
276 		return 0;
277 
278 	/*
279 	 * check if IPv4 dst matches the IPv4 address derived from the
280 	 * local 6to4 address.
281 	 * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
282 	 */
283 	if (bcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst,
284 	    sizeof(ip.ip_dst)) != 0)
285 		return 0;
286 
287 	/*
288 	 * check if IPv4 src matches the IPv4 address derived from the
289 	 * local 6to4 address masked by prefixmask.
290 	 * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
291 	 * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
292 	 */
293 	memset(&a, 0, sizeof(a));
294 	a.s_addr = GET_V4(&ia6->ia_addr.sin6_addr)->s_addr;
295 	a.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
296 	b = ip.ip_src;
297 	b.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
298 	if (a.s_addr != b.s_addr)
299 		return 0;
300 
301 	/* stf interface makes single side match only */
302 	return 32;
303 }
304 
305 static struct in6_ifaddr *
306 stf_getsrcifa6(struct ifnet *ifp)
307 {
308 	struct ifaddr *ifa;
309 	struct in_ifaddr *ia4;
310 	struct sockaddr_in6 *sin6;
311 	struct in_addr in;
312 
313 	IFADDR_FOREACH(ifa, ifp)
314 	{
315 		if (ifa->ifa_addr == NULL)
316 			continue;
317 		if (ifa->ifa_addr->sa_family != AF_INET6)
318 			continue;
319 		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
320 		if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr))
321 			continue;
322 
323 		bcopy(GET_V4(&sin6->sin6_addr), &in, sizeof(in));
324 		INADDR_TO_IA(in, ia4);
325 		if (ia4 == NULL)
326 			continue;
327 
328 		return (struct in6_ifaddr *)ifa;
329 	}
330 
331 	return NULL;
332 }
333 
334 static int
335 stf_output(struct ifnet *ifp, struct mbuf *m, struct sockaddr *dst,
336     struct rtentry *rt)
337 {
338 	struct stf_softc *sc;
339 	struct sockaddr_in6 *dst6;
340 	struct in_addr *in4;
341 	struct sockaddr_in *dst4;
342 	u_int8_t tos;
343 	struct ip *ip;
344 	struct ip6_hdr *ip6;
345 	struct in6_ifaddr *ia6;
346 
347 	sc = (struct stf_softc*)ifp;
348 	dst6 = (struct sockaddr_in6 *)dst;
349 
350 	/* just in case */
351 	if ((ifp->if_flags & IFF_UP) == 0) {
352 		m_freem(m);
353 		return ENETDOWN;
354 	}
355 
356 	/*
357 	 * If we don't have an ip4 address that match my inner ip6 address,
358 	 * we shouldn't generate output.  Without this check, we'll end up
359 	 * using wrong IPv4 source.
360 	 */
361 	ia6 = stf_getsrcifa6(ifp);
362 	if (ia6 == NULL) {
363 		m_freem(m);
364 		ifp->if_oerrors++;
365 		return ENETDOWN;
366 	}
367 
368 	if (m->m_len < sizeof(*ip6)) {
369 		m = m_pullup(m, sizeof(*ip6));
370 		if (m == NULL) {
371 			ifp->if_oerrors++;
372 			return ENOBUFS;
373 		}
374 	}
375 	ip6 = mtod(m, struct ip6_hdr *);
376 	tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
377 
378 	/*
379 	 * Pickup the right outer dst addr from the list of candidates.
380 	 * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
381 	 */
382 	if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst))
383 		in4 = GET_V4(&ip6->ip6_dst);
384 	else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr))
385 		in4 = GET_V4(&dst6->sin6_addr);
386 	else {
387 		m_freem(m);
388 		ifp->if_oerrors++;
389 		return ENETUNREACH;
390 	}
391 
392 #if NBPFILTER > 0
393 	if (ifp->if_bpf)
394 		bpf_mtap_af(ifp->if_bpf, AF_INET6, m);
395 #endif /*NBPFILTER > 0*/
396 
397 	M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
398 	if (m && m->m_len < sizeof(struct ip))
399 		m = m_pullup(m, sizeof(struct ip));
400 	if (m == NULL) {
401 		ifp->if_oerrors++;
402 		return ENOBUFS;
403 	}
404 	ip = mtod(m, struct ip *);
405 
406 	memset(ip, 0, sizeof(*ip));
407 
408 	bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr),
409 	    &ip->ip_src, sizeof(ip->ip_src));
410 	bcopy(in4, &ip->ip_dst, sizeof(ip->ip_dst));
411 	ip->ip_p = IPPROTO_IPV6;
412 	ip->ip_ttl = ip_gif_ttl;	/*XXX*/
413 	ip->ip_len = htons(m->m_pkthdr.len);
414 	if (ifp->if_flags & IFF_LINK1)
415 		ip_ecn_ingress(ECN_ALLOWED, &ip->ip_tos, &tos);
416 	else
417 		ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
418 
419 	dst4 = (struct sockaddr_in *)&sc->sc_ro.ro_dst;
420 	if (dst4->sin_family != AF_INET ||
421 	    bcmp(&dst4->sin_addr, &ip->ip_dst, sizeof(ip->ip_dst)) != 0) {
422 		/* cache route doesn't match */
423 		dst4->sin_family = AF_INET;
424 		dst4->sin_len = sizeof(struct sockaddr_in);
425 		bcopy(&ip->ip_dst, &dst4->sin_addr, sizeof(dst4->sin_addr));
426 		if (sc->sc_ro.ro_rt) {
427 			RTFREE(sc->sc_ro.ro_rt);
428 			sc->sc_ro.ro_rt = NULL;
429 		}
430 	}
431 
432 	if (sc->sc_ro.ro_rt == NULL) {
433 		rtalloc(&sc->sc_ro);
434 		if (sc->sc_ro.ro_rt == NULL) {
435 			m_freem(m);
436 			ifp->if_oerrors++;
437 			return ENETUNREACH;
438 		}
439 	}
440 
441 	ifp->if_opackets++;
442 	return ip_output(m, NULL, &sc->sc_ro, 0,
443 	    (struct ip_moptions *)NULL, (struct socket *)NULL);
444 }
445 
446 static int
447 isrfc1918addr(struct in_addr *in)
448 {
449 	/*
450 	 * returns 1 if private address range:
451 	 * 10.0.0.0/8 172.16.0.0/12 192.168.0.0/16
452 	 */
453 	if ((ntohl(in->s_addr) & 0xff000000) >> 24 == 10 ||
454 	    (ntohl(in->s_addr) & 0xfff00000) >> 16 == 172 * 256 + 16 ||
455 	    (ntohl(in->s_addr) & 0xffff0000) >> 16 == 192 * 256 + 168)
456 		return 1;
457 
458 	return 0;
459 }
460 
461 static int
462 stf_checkaddr4(struct stf_softc *sc, struct in_addr *in,
463     struct ifnet *inifp /*incoming interface*/)
464 {
465 	struct in_ifaddr *ia4;
466 
467 	/*
468 	 * reject packets with the following address:
469 	 * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
470 	 */
471 	if (IN_MULTICAST(in->s_addr))
472 		return -1;
473 	switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
474 	case 0: case 127: case 255:
475 		return -1;
476 	}
477 
478 	/*
479 	 * reject packets with private address range.
480 	 * (requirement from RFC3056 section 2 1st paragraph)
481 	 */
482 	if (isrfc1918addr(in))
483 		return -1;
484 
485 	/*
486 	 * reject packet with IPv4 link-local (169.254.0.0/16),
487 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
488 	 */
489 	if (((ntohl(in->s_addr) & 0xff000000) >> 24) == 169 &&
490 	    ((ntohl(in->s_addr) & 0x00ff0000) >> 16) == 254)
491 		return -1;
492 
493 	/*
494 	 * reject packets with broadcast
495 	 */
496 	TAILQ_FOREACH(ia4, &in_ifaddrhead, ia_list)
497 	{
498 		if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0)
499 			continue;
500 		if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr)
501 			return -1;
502 	}
503 
504 	/*
505 	 * perform ingress filter
506 	 */
507 	if (sc && (sc->sc_if.if_flags & IFF_LINK2) == 0 && inifp) {
508 		struct sockaddr_in sin;
509 		struct rtentry *rt;
510 
511 		memset(&sin, 0, sizeof(sin));
512 		sin.sin_family = AF_INET;
513 		sin.sin_len = sizeof(struct sockaddr_in);
514 		sin.sin_addr = *in;
515 		rt = rtalloc1((struct sockaddr *)&sin, 0);
516 		if (!rt || rt->rt_ifp != inifp) {
517 #if 0
518 			log(LOG_WARNING, "%s: packet from 0x%x dropped "
519 			    "due to ingress filter\n", if_name(&sc->sc_if),
520 			    (u_int32_t)ntohl(sin.sin_addr.s_addr));
521 #endif
522 			if (rt)
523 				rtfree(rt);
524 			return -1;
525 		}
526 		rtfree(rt);
527 	}
528 
529 	return 0;
530 }
531 
532 static int
533 stf_checkaddr6(struct stf_softc *sc, struct in6_addr *in6,
534     struct ifnet *inifp /*incoming interface*/)
535 {
536 
537 	/*
538 	 * check 6to4 addresses
539 	 */
540 	if (IN6_IS_ADDR_6TO4(in6))
541 		return stf_checkaddr4(sc, GET_V4(in6), inifp);
542 
543 	/*
544 	 * reject anything that look suspicious.  the test is implemented
545 	 * in ip6_input too, but we check here as well to
546 	 * (1) reject bad packets earlier, and
547 	 * (2) to be safe against future ip6_input change.
548 	 */
549 	if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6))
550 		return -1;
551 
552 	/*
553 	 * reject link-local and site-local unicast
554 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
555 	 */
556 	if (IN6_IS_ADDR_LINKLOCAL(in6) || IN6_IS_ADDR_SITELOCAL(in6))
557 		return -1;
558 
559 	/*
560 	 * reject node-local and link-local multicast
561 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
562 	 */
563 	if (IN6_IS_ADDR_MC_NODELOCAL(in6) || IN6_IS_ADDR_MC_LINKLOCAL(in6))
564 		return -1;
565 
566 	return 0;
567 }
568 
569 void
570 in_stf_input(struct mbuf *m, ...)
571 {
572 	int off, proto;
573 	struct stf_softc *sc;
574 	struct ip *ip;
575 	struct ip6_hdr *ip6;
576 	u_int8_t otos, itos;
577 	int s, isr;
578 	struct ifqueue *ifq = NULL;
579 	struct ifnet *ifp;
580 	va_list ap;
581 
582 	va_start(ap, m);
583 	off = va_arg(ap, int);
584 	proto = va_arg(ap, int);
585 	va_end(ap);
586 
587 	if (proto != IPPROTO_IPV6) {
588 		m_freem(m);
589 		return;
590 	}
591 
592 	ip = mtod(m, struct ip *);
593 
594 	sc = (struct stf_softc *)encap_getarg(m);
595 
596 	if (sc == NULL || (sc->sc_if.if_flags & IFF_UP) == 0) {
597 		m_freem(m);
598 		return;
599 	}
600 
601 	ifp = &sc->sc_if;
602 
603 	/*
604 	 * perform sanity check against outer src/dst.
605 	 * for source, perform ingress filter as well.
606 	 */
607 	if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
608 	    stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
609 		m_freem(m);
610 		return;
611 	}
612 
613 	otos = ip->ip_tos;
614 	m_adj(m, off);
615 
616 	if (m->m_len < sizeof(*ip6)) {
617 		m = m_pullup(m, sizeof(*ip6));
618 		if (!m)
619 			return;
620 	}
621 	ip6 = mtod(m, struct ip6_hdr *);
622 
623 	/*
624 	 * perform sanity check against inner src/dst.
625 	 * for source, perform ingress filter as well.
626 	 */
627 	if (stf_checkaddr6(sc, &ip6->ip6_dst, NULL) < 0 ||
628 	    stf_checkaddr6(sc, &ip6->ip6_src, m->m_pkthdr.rcvif) < 0) {
629 		m_freem(m);
630 		return;
631 	}
632 
633 	itos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
634 	if ((ifp->if_flags & IFF_LINK1) != 0)
635 		ip_ecn_egress(ECN_ALLOWED, &otos, &itos);
636 	else
637 		ip_ecn_egress(ECN_NOCARE, &otos, &itos);
638 	ip6->ip6_flow &= ~htonl(0xff << 20);
639 	ip6->ip6_flow |= htonl((u_int32_t)itos << 20);
640 
641 	m->m_pkthdr.rcvif = ifp;
642 
643 #if NBPFILTER > 0
644 	if (ifp->if_bpf)
645 		bpf_mtap_af(ifp->if_bpf, AF_INET6, m);
646 #endif /*NBPFILTER > 0*/
647 
648 	/*
649 	 * Put the packet to the network layer input queue according to the
650 	 * specified address family.
651 	 * See net/if_gif.c for possible issues with packet processing
652 	 * reorder due to extra queueing.
653 	 */
654 	ifq = &ip6intrq;
655 	isr = NETISR_IPV6;
656 
657 	s = splnet();
658 	if (IF_QFULL(ifq)) {
659 		IF_DROP(ifq);	/* update statistics */
660 		m_freem(m);
661 		splx(s);
662 		return;
663 	}
664 	IF_ENQUEUE(ifq, m);
665 	schednetisr(isr);
666 	ifp->if_ipackets++;
667 	ifp->if_ibytes += m->m_pkthdr.len;
668 	splx(s);
669 }
670 
671 /* ARGSUSED */
672 static void
673 stf_rtrequest(int cmd, struct rtentry *rt, struct rt_addrinfo *info)
674 {
675 	if (rt != NULL) {
676 		struct stf_softc *sc;
677 
678 		sc = LIST_FIRST(&stf_softc_list);
679 		rt->rt_rmx.rmx_mtu = (sc != NULL) ? sc->sc_if.if_mtu : STF_MTU;
680 	}
681 }
682 
683 static int
684 stf_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
685 {
686 	struct proc 		*p = curproc;	/* XXX */
687 	struct ifaddr		*ifa;
688 	struct ifreq		*ifr;
689 	struct sockaddr_in6	*sin6;
690 	int			error;
691 	u_long			mtu;
692 
693 	error = 0;
694 	switch (cmd) {
695 	case SIOCSIFADDR:
696 		ifa = (struct ifaddr *)data;
697 		if (ifa == NULL || ifa->ifa_addr->sa_family != AF_INET6) {
698 			error = EAFNOSUPPORT;
699 			break;
700 		}
701 		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
702 		if (IN6_IS_ADDR_6TO4(&sin6->sin6_addr) &&
703 		    !isrfc1918addr(GET_V4(&sin6->sin6_addr))) {
704 			ifa->ifa_rtrequest = stf_rtrequest;
705 			ifp->if_flags |= IFF_UP;
706 		} else
707 			error = EINVAL;
708 		break;
709 
710 	case SIOCADDMULTI:
711 	case SIOCDELMULTI:
712 		ifr = (struct ifreq *)data;
713 		if (ifr != NULL && ifr->ifr_addr.sa_family == AF_INET6)
714 			;
715 		else
716 			error = EAFNOSUPPORT;
717 		break;
718 
719 	case SIOCSIFMTU:
720 		if ((error = kauth_authorize_generic(p->p_cred, KAUTH_GENERIC_ISSUSER, &p->p_acflag)) != 0)
721 			break;
722 		ifr = (struct ifreq *)data;
723 		mtu = ifr->ifr_mtu;
724 		if (mtu < STF_MTU_MIN || mtu > STF_MTU_MAX)
725 			return EINVAL;
726 		ifp->if_mtu = mtu;
727 		break;
728 
729 	default:
730 		error = EINVAL;
731 		break;
732 	}
733 
734 	return error;
735 }
736