xref: /netbsd-src/sys/net/if_stf.c (revision 7f21db1c0118155e0dd40b75182e30c589d9f63e)
1 /*	$NetBSD: if_stf.c,v 1.74 2010/01/19 22:08:01 pooka 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 1.74 2010/01/19 22:08:01 pooka 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 <sys/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 "stf.h"
122 #include "gif.h"	/*XXX*/
123 
124 #include <net/bpf.h>
125 
126 #if NGIF > 0
127 #include <net/if_gif.h>
128 #endif
129 
130 #define IN6_IS_ADDR_6TO4(x)	(ntohs((x)->s6_addr16[0]) == 0x2002)
131 #define GET_V4(x)	((const struct in_addr *)(&(x)->s6_addr16[1]))
132 
133 struct stf_softc {
134 	struct ifnet	sc_if;	   /* common area */
135 	struct route	sc_ro;
136 	const struct encaptab *encap_cookie;
137 	LIST_ENTRY(stf_softc) sc_list;
138 };
139 
140 static LIST_HEAD(, stf_softc) stf_softc_list;
141 
142 static int	stf_clone_create(struct if_clone *, int);
143 static int	stf_clone_destroy(struct ifnet *);
144 
145 struct if_clone stf_cloner =
146     IF_CLONE_INITIALIZER("stf", stf_clone_create, stf_clone_destroy);
147 
148 #if NGIF > 0
149 extern int ip_gif_ttl;	/*XXX*/
150 #else
151 static int ip_gif_ttl = 40;	/*XXX*/
152 #endif
153 
154 extern struct domain inetdomain;
155 static const struct protosw in_stf_protosw =
156 { SOCK_RAW,	&inetdomain,	IPPROTO_IPV6,	PR_ATOMIC|PR_ADDR,
157   in_stf_input, rip_output,	0,		rip_ctloutput,
158   rip_usrreq,
159   0,            0,              0,              0
160 };
161 
162 void	stfattach(int);
163 
164 static int stf_encapcheck(struct mbuf *, int, int, void *);
165 static struct in6_ifaddr *stf_getsrcifa6(struct ifnet *);
166 static int stf_output(struct ifnet *, struct mbuf *, const struct sockaddr *,
167 	struct rtentry *);
168 static int isrfc1918addr(const struct in_addr *);
169 static int stf_checkaddr4(struct stf_softc *, const struct in_addr *,
170 	struct ifnet *);
171 static int stf_checkaddr6(struct stf_softc *, const struct in6_addr *,
172 	struct ifnet *);
173 static void stf_rtrequest(int, struct rtentry *, const struct rt_addrinfo *);
174 static int stf_ioctl(struct ifnet *, u_long, void *);
175 
176 /* ARGSUSED */
177 void
178 stfattach(int count)
179 {
180 
181 	LIST_INIT(&stf_softc_list);
182 	if_clone_attach(&stf_cloner);
183 }
184 
185 static int
186 stf_clone_create(struct if_clone *ifc, int unit)
187 {
188 	struct stf_softc *sc;
189 
190 	if (LIST_FIRST(&stf_softc_list) != NULL) {
191 		/* Only one stf interface is allowed. */
192 		return (EEXIST);
193 	}
194 
195 	sc = malloc(sizeof(struct stf_softc), M_DEVBUF, M_WAIT|M_ZERO);
196 
197 	if_initname(&sc->sc_if, ifc->ifc_name, unit);
198 
199 	sc->encap_cookie = encap_attach_func(AF_INET, IPPROTO_IPV6,
200 	    stf_encapcheck, &in_stf_protosw, sc);
201 	if (sc->encap_cookie == NULL) {
202 		printf("%s: unable to attach encap\n", if_name(&sc->sc_if));
203 		free(sc, M_DEVBUF);
204 		return (EIO);	/* XXX */
205 	}
206 
207 	sc->sc_if.if_mtu    = STF_MTU;
208 	sc->sc_if.if_flags  = 0;
209 	sc->sc_if.if_ioctl  = stf_ioctl;
210 	sc->sc_if.if_output = stf_output;
211 	sc->sc_if.if_type   = IFT_STF;
212 	sc->sc_if.if_dlt    = DLT_NULL;
213 	if_attach(&sc->sc_if);
214 	if_alloc_sadl(&sc->sc_if);
215 	bpf_ops->bpf_attach(&sc->sc_if, DLT_NULL, sizeof(u_int),
216 	    &sc->sc_if.if_bpf);
217 	LIST_INSERT_HEAD(&stf_softc_list, sc, sc_list);
218 	return (0);
219 }
220 
221 static int
222 stf_clone_destroy(struct ifnet *ifp)
223 {
224 	struct stf_softc *sc = (void *) ifp;
225 
226 	LIST_REMOVE(sc, sc_list);
227 	encap_detach(sc->encap_cookie);
228 	bpf_ops->bpf_detach(ifp);
229 	if_detach(ifp);
230 	rtcache_free(&sc->sc_ro);
231 	free(sc, M_DEVBUF);
232 
233 	return (0);
234 }
235 
236 static int
237 stf_encapcheck(struct mbuf *m, int off, int proto, void *arg)
238 {
239 	struct ip ip;
240 	struct in6_ifaddr *ia6;
241 	struct stf_softc *sc;
242 	struct in_addr a, b;
243 
244 	sc = (struct stf_softc *)arg;
245 	if (sc == NULL)
246 		return 0;
247 
248 	if ((sc->sc_if.if_flags & IFF_UP) == 0)
249 		return 0;
250 
251 	/* IFF_LINK0 means "no decapsulation" */
252 	if ((sc->sc_if.if_flags & IFF_LINK0) != 0)
253 		return 0;
254 
255 	if (proto != IPPROTO_IPV6)
256 		return 0;
257 
258 	m_copydata(m, 0, sizeof(ip), (void *)&ip);
259 
260 	if (ip.ip_v != 4)
261 		return 0;
262 
263 	ia6 = stf_getsrcifa6(&sc->sc_if);
264 	if (ia6 == NULL)
265 		return 0;
266 
267 	/*
268 	 * check if IPv4 dst matches the IPv4 address derived from the
269 	 * local 6to4 address.
270 	 * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
271 	 */
272 	if (memcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst,
273 	    sizeof(ip.ip_dst)) != 0)
274 		return 0;
275 
276 	/*
277 	 * check if IPv4 src matches the IPv4 address derived from the
278 	 * local 6to4 address masked by prefixmask.
279 	 * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
280 	 * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
281 	 */
282 	memset(&a, 0, sizeof(a));
283 	a.s_addr = GET_V4(&ia6->ia_addr.sin6_addr)->s_addr;
284 	a.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
285 	b = ip.ip_src;
286 	b.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
287 	if (a.s_addr != b.s_addr)
288 		return 0;
289 
290 	/* stf interface makes single side match only */
291 	return 32;
292 }
293 
294 static struct in6_ifaddr *
295 stf_getsrcifa6(struct ifnet *ifp)
296 {
297 	struct ifaddr *ifa;
298 	struct in_ifaddr *ia4;
299 	struct sockaddr_in6 *sin6;
300 	struct in_addr in;
301 
302 	IFADDR_FOREACH(ifa, ifp)
303 	{
304 		if (ifa->ifa_addr == NULL)
305 			continue;
306 		if (ifa->ifa_addr->sa_family != AF_INET6)
307 			continue;
308 		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
309 		if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr))
310 			continue;
311 
312 		memcpy(&in, GET_V4(&sin6->sin6_addr), sizeof(in));
313 		INADDR_TO_IA(in, ia4);
314 		if (ia4 == NULL)
315 			continue;
316 
317 		return (struct in6_ifaddr *)ifa;
318 	}
319 
320 	return NULL;
321 }
322 
323 static int
324 stf_output(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *dst,
325     struct rtentry *rt0)
326 {
327 	struct rtentry *rt;
328 	struct stf_softc *sc;
329 	const struct sockaddr_in6 *dst6;
330 	const struct in_addr *in4;
331 	uint8_t tos;
332 	struct ip *ip;
333 	struct ip6_hdr *ip6;
334 	struct in6_ifaddr *ia6;
335 	union {
336 		struct sockaddr		dst;
337 		struct sockaddr_in	dst4;
338 	} u;
339 
340 	sc = (struct stf_softc*)ifp;
341 	dst6 = (const struct sockaddr_in6 *)dst;
342 
343 	/* just in case */
344 	if ((ifp->if_flags & IFF_UP) == 0) {
345 		m_freem(m);
346 		return ENETDOWN;
347 	}
348 
349 	/*
350 	 * If we don't have an ip4 address that match my inner ip6 address,
351 	 * we shouldn't generate output.  Without this check, we'll end up
352 	 * using wrong IPv4 source.
353 	 */
354 	ia6 = stf_getsrcifa6(ifp);
355 	if (ia6 == NULL) {
356 		m_freem(m);
357 		ifp->if_oerrors++;
358 		return ENETDOWN;
359 	}
360 
361 	if (m->m_len < sizeof(*ip6)) {
362 		m = m_pullup(m, sizeof(*ip6));
363 		if (m == NULL) {
364 			ifp->if_oerrors++;
365 			return ENOBUFS;
366 		}
367 	}
368 	ip6 = mtod(m, struct ip6_hdr *);
369 	tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
370 
371 	/*
372 	 * Pickup the right outer dst addr from the list of candidates.
373 	 * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
374 	 */
375 	if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst))
376 		in4 = GET_V4(&ip6->ip6_dst);
377 	else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr))
378 		in4 = GET_V4(&dst6->sin6_addr);
379 	else {
380 		m_freem(m);
381 		ifp->if_oerrors++;
382 		return ENETUNREACH;
383 	}
384 
385 	if (ifp->if_bpf)
386 		bpf_ops->bpf_mtap_af(ifp->if_bpf, AF_INET6, m);
387 
388 	M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
389 	if (m && m->m_len < sizeof(struct ip))
390 		m = m_pullup(m, sizeof(struct ip));
391 	if (m == NULL) {
392 		ifp->if_oerrors++;
393 		return ENOBUFS;
394 	}
395 	ip = mtod(m, struct ip *);
396 
397 	memset(ip, 0, sizeof(*ip));
398 
399 	bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr),
400 	    &ip->ip_src, sizeof(ip->ip_src));
401 	memcpy(&ip->ip_dst, in4, sizeof(ip->ip_dst));
402 	ip->ip_p = IPPROTO_IPV6;
403 	ip->ip_ttl = ip_gif_ttl;	/*XXX*/
404 	ip->ip_len = htons(m->m_pkthdr.len);
405 	if (ifp->if_flags & IFF_LINK1)
406 		ip_ecn_ingress(ECN_ALLOWED, &ip->ip_tos, &tos);
407 	else
408 		ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
409 
410 	sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
411 	if ((rt = rtcache_lookup(&sc->sc_ro, &u.dst)) == NULL) {
412 		m_freem(m);
413 		ifp->if_oerrors++;
414 		return ENETUNREACH;
415 	}
416 
417 	/* If the route constitutes infinite encapsulation, punt. */
418 	if (rt->rt_ifp == ifp) {
419 		rtcache_free(&sc->sc_ro);
420 		m_freem(m);
421 		ifp->if_oerrors++;
422 		return ENETUNREACH;
423 	}
424 
425 	ifp->if_opackets++;
426 	ifp->if_obytes += m->m_pkthdr.len - sizeof(struct ip);
427 	return ip_output(m, NULL, &sc->sc_ro, 0, NULL, NULL);
428 }
429 
430 static int
431 isrfc1918addr(const struct in_addr *in)
432 {
433 	/*
434 	 * returns 1 if private address range:
435 	 * 10.0.0.0/8 172.16.0.0/12 192.168.0.0/16
436 	 */
437 	if ((ntohl(in->s_addr) & 0xff000000) >> 24 == 10 ||
438 	    (ntohl(in->s_addr) & 0xfff00000) >> 16 == 172 * 256 + 16 ||
439 	    (ntohl(in->s_addr) & 0xffff0000) >> 16 == 192 * 256 + 168)
440 		return 1;
441 
442 	return 0;
443 }
444 
445 static int
446 stf_checkaddr4(struct stf_softc *sc, const struct in_addr *in,
447     struct ifnet *inifp /*incoming interface*/)
448 {
449 	struct in_ifaddr *ia4;
450 
451 	/*
452 	 * reject packets with the following address:
453 	 * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
454 	 */
455 	if (IN_MULTICAST(in->s_addr))
456 		return -1;
457 	switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
458 	case 0: case 127: case 255:
459 		return -1;
460 	}
461 
462 	/*
463 	 * reject packets with private address range.
464 	 * (requirement from RFC3056 section 2 1st paragraph)
465 	 */
466 	if (isrfc1918addr(in))
467 		return -1;
468 
469 	/*
470 	 * reject packet with IPv4 link-local (169.254.0.0/16),
471 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
472 	 */
473 	if (((ntohl(in->s_addr) & 0xff000000) >> 24) == 169 &&
474 	    ((ntohl(in->s_addr) & 0x00ff0000) >> 16) == 254)
475 		return -1;
476 
477 	/*
478 	 * reject packets with broadcast
479 	 */
480 	TAILQ_FOREACH(ia4, &in_ifaddrhead, ia_list)
481 	{
482 		if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0)
483 			continue;
484 		if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr)
485 			return -1;
486 	}
487 
488 	/*
489 	 * perform ingress filter
490 	 */
491 	if (sc && (sc->sc_if.if_flags & IFF_LINK2) == 0 && inifp) {
492 		struct sockaddr_in sin;
493 		struct rtentry *rt;
494 
495 		memset(&sin, 0, sizeof(sin));
496 		sin.sin_family = AF_INET;
497 		sin.sin_len = sizeof(struct sockaddr_in);
498 		sin.sin_addr = *in;
499 		rt = rtalloc1((struct sockaddr *)&sin, 0);
500 		if (!rt || rt->rt_ifp != inifp) {
501 #if 0
502 			log(LOG_WARNING, "%s: packet from 0x%x dropped "
503 			    "due to ingress filter\n", if_name(&sc->sc_if),
504 			    (uint32_t)ntohl(sin.sin_addr.s_addr));
505 #endif
506 			if (rt)
507 				rtfree(rt);
508 			return -1;
509 		}
510 		rtfree(rt);
511 	}
512 
513 	return 0;
514 }
515 
516 static int
517 stf_checkaddr6(struct stf_softc *sc, const struct in6_addr *in6,
518     struct ifnet *inifp /*incoming interface*/)
519 {
520 
521 	/*
522 	 * check 6to4 addresses
523 	 */
524 	if (IN6_IS_ADDR_6TO4(in6))
525 		return stf_checkaddr4(sc, GET_V4(in6), inifp);
526 
527 	/*
528 	 * reject anything that look suspicious.  the test is implemented
529 	 * in ip6_input too, but we check here as well to
530 	 * (1) reject bad packets earlier, and
531 	 * (2) to be safe against future ip6_input change.
532 	 */
533 	if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6))
534 		return -1;
535 
536 	/*
537 	 * reject link-local and site-local unicast
538 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
539 	 */
540 	if (IN6_IS_ADDR_LINKLOCAL(in6) || IN6_IS_ADDR_SITELOCAL(in6))
541 		return -1;
542 
543 	/*
544 	 * reject node-local and link-local multicast
545 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
546 	 */
547 	if (IN6_IS_ADDR_MC_NODELOCAL(in6) || IN6_IS_ADDR_MC_LINKLOCAL(in6))
548 		return -1;
549 
550 	return 0;
551 }
552 
553 void
554 in_stf_input(struct mbuf *m, ...)
555 {
556 	int off, proto;
557 	struct stf_softc *sc;
558 	struct ip *ip;
559 	struct ip6_hdr *ip6;
560 	uint8_t otos, itos;
561 	int s, isr;
562 	struct ifqueue *ifq = NULL;
563 	struct ifnet *ifp;
564 	va_list ap;
565 
566 	va_start(ap, m);
567 	off = va_arg(ap, int);
568 	proto = va_arg(ap, int);
569 	va_end(ap);
570 
571 	if (proto != IPPROTO_IPV6) {
572 		m_freem(m);
573 		return;
574 	}
575 
576 	ip = mtod(m, struct ip *);
577 
578 	sc = (struct stf_softc *)encap_getarg(m);
579 
580 	if (sc == NULL || (sc->sc_if.if_flags & IFF_UP) == 0) {
581 		m_freem(m);
582 		return;
583 	}
584 
585 	ifp = &sc->sc_if;
586 
587 	/*
588 	 * perform sanity check against outer src/dst.
589 	 * for source, perform ingress filter as well.
590 	 */
591 	if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
592 	    stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
593 		m_freem(m);
594 		return;
595 	}
596 
597 	otos = ip->ip_tos;
598 	m_adj(m, off);
599 
600 	if (m->m_len < sizeof(*ip6)) {
601 		m = m_pullup(m, sizeof(*ip6));
602 		if (!m)
603 			return;
604 	}
605 	ip6 = mtod(m, struct ip6_hdr *);
606 
607 	/*
608 	 * perform sanity check against inner src/dst.
609 	 * for source, perform ingress filter as well.
610 	 */
611 	if (stf_checkaddr6(sc, &ip6->ip6_dst, NULL) < 0 ||
612 	    stf_checkaddr6(sc, &ip6->ip6_src, m->m_pkthdr.rcvif) < 0) {
613 		m_freem(m);
614 		return;
615 	}
616 
617 	itos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
618 	if ((ifp->if_flags & IFF_LINK1) != 0)
619 		ip_ecn_egress(ECN_ALLOWED, &otos, &itos);
620 	else
621 		ip_ecn_egress(ECN_NOCARE, &otos, &itos);
622 	ip6->ip6_flow &= ~htonl(0xff << 20);
623 	ip6->ip6_flow |= htonl((uint32_t)itos << 20);
624 
625 	m->m_pkthdr.rcvif = ifp;
626 
627 	if (ifp->if_bpf)
628 		bpf_ops->bpf_mtap_af(ifp->if_bpf, AF_INET6, m);
629 
630 	/*
631 	 * Put the packet to the network layer input queue according to the
632 	 * specified address family.
633 	 * See net/if_gif.c for possible issues with packet processing
634 	 * reorder due to extra queueing.
635 	 */
636 	ifq = &ip6intrq;
637 	isr = NETISR_IPV6;
638 
639 	s = splnet();
640 	if (IF_QFULL(ifq)) {
641 		IF_DROP(ifq);	/* update statistics */
642 		m_freem(m);
643 		splx(s);
644 		return;
645 	}
646 	IF_ENQUEUE(ifq, m);
647 	schednetisr(isr);
648 	ifp->if_ipackets++;
649 	ifp->if_ibytes += m->m_pkthdr.len;
650 	splx(s);
651 }
652 
653 /* ARGSUSED */
654 static void
655 stf_rtrequest(int cmd, struct rtentry *rt,
656     const struct rt_addrinfo *info)
657 {
658 	if (rt != NULL) {
659 		struct stf_softc *sc;
660 
661 		sc = LIST_FIRST(&stf_softc_list);
662 		rt->rt_rmx.rmx_mtu = (sc != NULL) ? sc->sc_if.if_mtu : STF_MTU;
663 	}
664 }
665 
666 static int
667 stf_ioctl(struct ifnet *ifp, u_long cmd, void *data)
668 {
669 	struct lwp		*l = curlwp;	/* XXX */
670 	struct ifaddr		*ifa;
671 	struct ifreq		*ifr = data;
672 	struct sockaddr_in6	*sin6;
673 	int			error;
674 
675 	error = 0;
676 	switch (cmd) {
677 	case SIOCINITIFADDR:
678 		ifa = (struct ifaddr *)data;
679 		if (ifa == NULL || ifa->ifa_addr->sa_family != AF_INET6) {
680 			error = EAFNOSUPPORT;
681 			break;
682 		}
683 		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
684 		if (IN6_IS_ADDR_6TO4(&sin6->sin6_addr) &&
685 		    !isrfc1918addr(GET_V4(&sin6->sin6_addr))) {
686 			ifa->ifa_rtrequest = stf_rtrequest;
687 			ifp->if_flags |= IFF_UP;
688 		} else
689 			error = EINVAL;
690 		break;
691 
692 	case SIOCADDMULTI:
693 	case SIOCDELMULTI:
694 		if (ifr != NULL &&
695 		    ifreq_getaddr(cmd, ifr)->sa_family == AF_INET6)
696 			;
697 		else
698 			error = EAFNOSUPPORT;
699 		break;
700 
701 	case SIOCSIFMTU:
702 		error = kauth_authorize_network(l->l_cred,
703 		    KAUTH_NETWORK_INTERFACE,
704 		    KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, KAUTH_ARG(cmd),
705 		    NULL);
706 		if (error)
707 			break;
708 		if (ifr->ifr_mtu < STF_MTU_MIN || ifr->ifr_mtu > STF_MTU_MAX)
709 			return EINVAL;
710 		else if ((error = ifioctl_common(ifp, cmd, data)) == ENETRESET)
711 			error = 0;
712 		break;
713 
714 	default:
715 		error = ifioctl_common(ifp, cmd, data);
716 		break;
717 	}
718 
719 	return error;
720 }
721