xref: /netbsd/sys/net/if_stf.c (revision 6550d01e)
1 /*	$NetBSD: if_stf.c,v 1.75 2010/04/05 07:22:24 joerg 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 1.75 2010/04/05 07:22:24 joerg 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_attach(&sc->sc_if, DLT_NULL, sizeof(u_int));
216 	LIST_INSERT_HEAD(&stf_softc_list, sc, sc_list);
217 	return (0);
218 }
219 
220 static int
221 stf_clone_destroy(struct ifnet *ifp)
222 {
223 	struct stf_softc *sc = (void *) ifp;
224 
225 	LIST_REMOVE(sc, sc_list);
226 	encap_detach(sc->encap_cookie);
227 	bpf_detach(ifp);
228 	if_detach(ifp);
229 	rtcache_free(&sc->sc_ro);
230 	free(sc, M_DEVBUF);
231 
232 	return (0);
233 }
234 
235 static int
236 stf_encapcheck(struct mbuf *m, int off, int proto, void *arg)
237 {
238 	struct ip ip;
239 	struct in6_ifaddr *ia6;
240 	struct stf_softc *sc;
241 	struct in_addr a, b;
242 
243 	sc = (struct stf_softc *)arg;
244 	if (sc == NULL)
245 		return 0;
246 
247 	if ((sc->sc_if.if_flags & IFF_UP) == 0)
248 		return 0;
249 
250 	/* IFF_LINK0 means "no decapsulation" */
251 	if ((sc->sc_if.if_flags & IFF_LINK0) != 0)
252 		return 0;
253 
254 	if (proto != IPPROTO_IPV6)
255 		return 0;
256 
257 	m_copydata(m, 0, sizeof(ip), (void *)&ip);
258 
259 	if (ip.ip_v != 4)
260 		return 0;
261 
262 	ia6 = stf_getsrcifa6(&sc->sc_if);
263 	if (ia6 == NULL)
264 		return 0;
265 
266 	/*
267 	 * check if IPv4 dst matches the IPv4 address derived from the
268 	 * local 6to4 address.
269 	 * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
270 	 */
271 	if (memcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst,
272 	    sizeof(ip.ip_dst)) != 0)
273 		return 0;
274 
275 	/*
276 	 * check if IPv4 src matches the IPv4 address derived from the
277 	 * local 6to4 address masked by prefixmask.
278 	 * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
279 	 * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
280 	 */
281 	memset(&a, 0, sizeof(a));
282 	a.s_addr = GET_V4(&ia6->ia_addr.sin6_addr)->s_addr;
283 	a.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
284 	b = ip.ip_src;
285 	b.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
286 	if (a.s_addr != b.s_addr)
287 		return 0;
288 
289 	/* stf interface makes single side match only */
290 	return 32;
291 }
292 
293 static struct in6_ifaddr *
294 stf_getsrcifa6(struct ifnet *ifp)
295 {
296 	struct ifaddr *ifa;
297 	struct in_ifaddr *ia4;
298 	struct sockaddr_in6 *sin6;
299 	struct in_addr in;
300 
301 	IFADDR_FOREACH(ifa, ifp)
302 	{
303 		if (ifa->ifa_addr == NULL)
304 			continue;
305 		if (ifa->ifa_addr->sa_family != AF_INET6)
306 			continue;
307 		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
308 		if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr))
309 			continue;
310 
311 		memcpy(&in, GET_V4(&sin6->sin6_addr), sizeof(in));
312 		INADDR_TO_IA(in, ia4);
313 		if (ia4 == NULL)
314 			continue;
315 
316 		return (struct in6_ifaddr *)ifa;
317 	}
318 
319 	return NULL;
320 }
321 
322 static int
323 stf_output(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *dst,
324     struct rtentry *rt0)
325 {
326 	struct rtentry *rt;
327 	struct stf_softc *sc;
328 	const struct sockaddr_in6 *dst6;
329 	const struct in_addr *in4;
330 	uint8_t tos;
331 	struct ip *ip;
332 	struct ip6_hdr *ip6;
333 	struct in6_ifaddr *ia6;
334 	union {
335 		struct sockaddr		dst;
336 		struct sockaddr_in	dst4;
337 	} u;
338 
339 	sc = (struct stf_softc*)ifp;
340 	dst6 = (const struct sockaddr_in6 *)dst;
341 
342 	/* just in case */
343 	if ((ifp->if_flags & IFF_UP) == 0) {
344 		m_freem(m);
345 		return ENETDOWN;
346 	}
347 
348 	/*
349 	 * If we don't have an ip4 address that match my inner ip6 address,
350 	 * we shouldn't generate output.  Without this check, we'll end up
351 	 * using wrong IPv4 source.
352 	 */
353 	ia6 = stf_getsrcifa6(ifp);
354 	if (ia6 == NULL) {
355 		m_freem(m);
356 		ifp->if_oerrors++;
357 		return ENETDOWN;
358 	}
359 
360 	if (m->m_len < sizeof(*ip6)) {
361 		m = m_pullup(m, sizeof(*ip6));
362 		if (m == NULL) {
363 			ifp->if_oerrors++;
364 			return ENOBUFS;
365 		}
366 	}
367 	ip6 = mtod(m, struct ip6_hdr *);
368 	tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
369 
370 	/*
371 	 * Pickup the right outer dst addr from the list of candidates.
372 	 * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
373 	 */
374 	if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst))
375 		in4 = GET_V4(&ip6->ip6_dst);
376 	else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr))
377 		in4 = GET_V4(&dst6->sin6_addr);
378 	else {
379 		m_freem(m);
380 		ifp->if_oerrors++;
381 		return ENETUNREACH;
382 	}
383 
384 	bpf_mtap_af(ifp, AF_INET6, m);
385 
386 	M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
387 	if (m && m->m_len < sizeof(struct ip))
388 		m = m_pullup(m, sizeof(struct ip));
389 	if (m == NULL) {
390 		ifp->if_oerrors++;
391 		return ENOBUFS;
392 	}
393 	ip = mtod(m, struct ip *);
394 
395 	memset(ip, 0, sizeof(*ip));
396 
397 	bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr),
398 	    &ip->ip_src, sizeof(ip->ip_src));
399 	memcpy(&ip->ip_dst, in4, sizeof(ip->ip_dst));
400 	ip->ip_p = IPPROTO_IPV6;
401 	ip->ip_ttl = ip_gif_ttl;	/*XXX*/
402 	ip->ip_len = htons(m->m_pkthdr.len);
403 	if (ifp->if_flags & IFF_LINK1)
404 		ip_ecn_ingress(ECN_ALLOWED, &ip->ip_tos, &tos);
405 	else
406 		ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
407 
408 	sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
409 	if ((rt = rtcache_lookup(&sc->sc_ro, &u.dst)) == NULL) {
410 		m_freem(m);
411 		ifp->if_oerrors++;
412 		return ENETUNREACH;
413 	}
414 
415 	/* If the route constitutes infinite encapsulation, punt. */
416 	if (rt->rt_ifp == ifp) {
417 		rtcache_free(&sc->sc_ro);
418 		m_freem(m);
419 		ifp->if_oerrors++;
420 		return ENETUNREACH;
421 	}
422 
423 	ifp->if_opackets++;
424 	ifp->if_obytes += m->m_pkthdr.len - sizeof(struct ip);
425 	return ip_output(m, NULL, &sc->sc_ro, 0, NULL, NULL);
426 }
427 
428 static int
429 isrfc1918addr(const struct in_addr *in)
430 {
431 	/*
432 	 * returns 1 if private address range:
433 	 * 10.0.0.0/8 172.16.0.0/12 192.168.0.0/16
434 	 */
435 	if ((ntohl(in->s_addr) & 0xff000000) >> 24 == 10 ||
436 	    (ntohl(in->s_addr) & 0xfff00000) >> 16 == 172 * 256 + 16 ||
437 	    (ntohl(in->s_addr) & 0xffff0000) >> 16 == 192 * 256 + 168)
438 		return 1;
439 
440 	return 0;
441 }
442 
443 static int
444 stf_checkaddr4(struct stf_softc *sc, const struct in_addr *in,
445     struct ifnet *inifp /*incoming interface*/)
446 {
447 	struct in_ifaddr *ia4;
448 
449 	/*
450 	 * reject packets with the following address:
451 	 * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
452 	 */
453 	if (IN_MULTICAST(in->s_addr))
454 		return -1;
455 	switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
456 	case 0: case 127: case 255:
457 		return -1;
458 	}
459 
460 	/*
461 	 * reject packets with private address range.
462 	 * (requirement from RFC3056 section 2 1st paragraph)
463 	 */
464 	if (isrfc1918addr(in))
465 		return -1;
466 
467 	/*
468 	 * reject packet with IPv4 link-local (169.254.0.0/16),
469 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
470 	 */
471 	if (((ntohl(in->s_addr) & 0xff000000) >> 24) == 169 &&
472 	    ((ntohl(in->s_addr) & 0x00ff0000) >> 16) == 254)
473 		return -1;
474 
475 	/*
476 	 * reject packets with broadcast
477 	 */
478 	TAILQ_FOREACH(ia4, &in_ifaddrhead, ia_list)
479 	{
480 		if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0)
481 			continue;
482 		if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr)
483 			return -1;
484 	}
485 
486 	/*
487 	 * perform ingress filter
488 	 */
489 	if (sc && (sc->sc_if.if_flags & IFF_LINK2) == 0 && inifp) {
490 		struct sockaddr_in sin;
491 		struct rtentry *rt;
492 
493 		memset(&sin, 0, sizeof(sin));
494 		sin.sin_family = AF_INET;
495 		sin.sin_len = sizeof(struct sockaddr_in);
496 		sin.sin_addr = *in;
497 		rt = rtalloc1((struct sockaddr *)&sin, 0);
498 		if (!rt || rt->rt_ifp != inifp) {
499 #if 0
500 			log(LOG_WARNING, "%s: packet from 0x%x dropped "
501 			    "due to ingress filter\n", if_name(&sc->sc_if),
502 			    (uint32_t)ntohl(sin.sin_addr.s_addr));
503 #endif
504 			if (rt)
505 				rtfree(rt);
506 			return -1;
507 		}
508 		rtfree(rt);
509 	}
510 
511 	return 0;
512 }
513 
514 static int
515 stf_checkaddr6(struct stf_softc *sc, const struct in6_addr *in6,
516     struct ifnet *inifp /*incoming interface*/)
517 {
518 
519 	/*
520 	 * check 6to4 addresses
521 	 */
522 	if (IN6_IS_ADDR_6TO4(in6))
523 		return stf_checkaddr4(sc, GET_V4(in6), inifp);
524 
525 	/*
526 	 * reject anything that look suspicious.  the test is implemented
527 	 * in ip6_input too, but we check here as well to
528 	 * (1) reject bad packets earlier, and
529 	 * (2) to be safe against future ip6_input change.
530 	 */
531 	if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6))
532 		return -1;
533 
534 	/*
535 	 * reject link-local and site-local unicast
536 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
537 	 */
538 	if (IN6_IS_ADDR_LINKLOCAL(in6) || IN6_IS_ADDR_SITELOCAL(in6))
539 		return -1;
540 
541 	/*
542 	 * reject node-local and link-local multicast
543 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
544 	 */
545 	if (IN6_IS_ADDR_MC_NODELOCAL(in6) || IN6_IS_ADDR_MC_LINKLOCAL(in6))
546 		return -1;
547 
548 	return 0;
549 }
550 
551 void
552 in_stf_input(struct mbuf *m, ...)
553 {
554 	int off, proto;
555 	struct stf_softc *sc;
556 	struct ip *ip;
557 	struct ip6_hdr *ip6;
558 	uint8_t otos, itos;
559 	int s, isr;
560 	struct ifqueue *ifq = NULL;
561 	struct ifnet *ifp;
562 	va_list ap;
563 
564 	va_start(ap, m);
565 	off = va_arg(ap, int);
566 	proto = va_arg(ap, int);
567 	va_end(ap);
568 
569 	if (proto != IPPROTO_IPV6) {
570 		m_freem(m);
571 		return;
572 	}
573 
574 	ip = mtod(m, struct ip *);
575 
576 	sc = (struct stf_softc *)encap_getarg(m);
577 
578 	if (sc == NULL || (sc->sc_if.if_flags & IFF_UP) == 0) {
579 		m_freem(m);
580 		return;
581 	}
582 
583 	ifp = &sc->sc_if;
584 
585 	/*
586 	 * perform sanity check against outer src/dst.
587 	 * for source, perform ingress filter as well.
588 	 */
589 	if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
590 	    stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
591 		m_freem(m);
592 		return;
593 	}
594 
595 	otos = ip->ip_tos;
596 	m_adj(m, off);
597 
598 	if (m->m_len < sizeof(*ip6)) {
599 		m = m_pullup(m, sizeof(*ip6));
600 		if (!m)
601 			return;
602 	}
603 	ip6 = mtod(m, struct ip6_hdr *);
604 
605 	/*
606 	 * perform sanity check against inner src/dst.
607 	 * for source, perform ingress filter as well.
608 	 */
609 	if (stf_checkaddr6(sc, &ip6->ip6_dst, NULL) < 0 ||
610 	    stf_checkaddr6(sc, &ip6->ip6_src, m->m_pkthdr.rcvif) < 0) {
611 		m_freem(m);
612 		return;
613 	}
614 
615 	itos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
616 	if ((ifp->if_flags & IFF_LINK1) != 0)
617 		ip_ecn_egress(ECN_ALLOWED, &otos, &itos);
618 	else
619 		ip_ecn_egress(ECN_NOCARE, &otos, &itos);
620 	ip6->ip6_flow &= ~htonl(0xff << 20);
621 	ip6->ip6_flow |= htonl((uint32_t)itos << 20);
622 
623 	m->m_pkthdr.rcvif = ifp;
624 
625 	bpf_mtap_af(ifp, AF_INET6, m);
626 
627 	/*
628 	 * Put the packet to the network layer input queue according to the
629 	 * specified address family.
630 	 * See net/if_gif.c for possible issues with packet processing
631 	 * reorder due to extra queueing.
632 	 */
633 	ifq = &ip6intrq;
634 	isr = NETISR_IPV6;
635 
636 	s = splnet();
637 	if (IF_QFULL(ifq)) {
638 		IF_DROP(ifq);	/* update statistics */
639 		m_freem(m);
640 		splx(s);
641 		return;
642 	}
643 	IF_ENQUEUE(ifq, m);
644 	schednetisr(isr);
645 	ifp->if_ipackets++;
646 	ifp->if_ibytes += m->m_pkthdr.len;
647 	splx(s);
648 }
649 
650 /* ARGSUSED */
651 static void
652 stf_rtrequest(int cmd, struct rtentry *rt,
653     const struct rt_addrinfo *info)
654 {
655 	if (rt != NULL) {
656 		struct stf_softc *sc;
657 
658 		sc = LIST_FIRST(&stf_softc_list);
659 		rt->rt_rmx.rmx_mtu = (sc != NULL) ? sc->sc_if.if_mtu : STF_MTU;
660 	}
661 }
662 
663 static int
664 stf_ioctl(struct ifnet *ifp, u_long cmd, void *data)
665 {
666 	struct lwp		*l = curlwp;	/* XXX */
667 	struct ifaddr		*ifa;
668 	struct ifreq		*ifr = data;
669 	struct sockaddr_in6	*sin6;
670 	int			error;
671 
672 	error = 0;
673 	switch (cmd) {
674 	case SIOCINITIFADDR:
675 		ifa = (struct ifaddr *)data;
676 		if (ifa == NULL || ifa->ifa_addr->sa_family != AF_INET6) {
677 			error = EAFNOSUPPORT;
678 			break;
679 		}
680 		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
681 		if (IN6_IS_ADDR_6TO4(&sin6->sin6_addr) &&
682 		    !isrfc1918addr(GET_V4(&sin6->sin6_addr))) {
683 			ifa->ifa_rtrequest = stf_rtrequest;
684 			ifp->if_flags |= IFF_UP;
685 		} else
686 			error = EINVAL;
687 		break;
688 
689 	case SIOCADDMULTI:
690 	case SIOCDELMULTI:
691 		if (ifr != NULL &&
692 		    ifreq_getaddr(cmd, ifr)->sa_family == AF_INET6)
693 			;
694 		else
695 			error = EAFNOSUPPORT;
696 		break;
697 
698 	case SIOCSIFMTU:
699 		error = kauth_authorize_network(l->l_cred,
700 		    KAUTH_NETWORK_INTERFACE,
701 		    KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, KAUTH_ARG(cmd),
702 		    NULL);
703 		if (error)
704 			break;
705 		if (ifr->ifr_mtu < STF_MTU_MIN || ifr->ifr_mtu > STF_MTU_MAX)
706 			return EINVAL;
707 		else if ((error = ifioctl_common(ifp, cmd, data)) == ENETRESET)
708 			error = 0;
709 		break;
710 
711 	default:
712 		error = ifioctl_common(ifp, cmd, data);
713 		break;
714 	}
715 
716 	return error;
717 }
718