xref: /netbsd-src/sys/net/if_stf.c (revision ca453df649ce9db45b64d73678ba06cbccf9aa11)
1 /*	$NetBSD: if_stf.c,v 1.76 2011/07/17 20:54:52 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 1.76 2011/07/17 20:54:52 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 <net/net_osdep.h>
118 
119 #include "stf.h"
120 #include "gif.h"	/*XXX*/
121 
122 #include <net/bpf.h>
123 
124 #if NGIF > 0
125 #include <net/if_gif.h>
126 #endif
127 
128 #define IN6_IS_ADDR_6TO4(x)	(ntohs((x)->s6_addr16[0]) == 0x2002)
129 #define GET_V4(x)	((const struct in_addr *)(&(x)->s6_addr16[1]))
130 
131 struct stf_softc {
132 	struct ifnet	sc_if;	   /* common area */
133 	struct route	sc_ro;
134 	const struct encaptab *encap_cookie;
135 	LIST_ENTRY(stf_softc) sc_list;
136 };
137 
138 static LIST_HEAD(, stf_softc) stf_softc_list;
139 
140 static int	stf_clone_create(struct if_clone *, int);
141 static int	stf_clone_destroy(struct ifnet *);
142 
143 struct if_clone stf_cloner =
144     IF_CLONE_INITIALIZER("stf", stf_clone_create, stf_clone_destroy);
145 
146 #if NGIF > 0
147 extern int ip_gif_ttl;	/*XXX*/
148 #else
149 static int ip_gif_ttl = 40;	/*XXX*/
150 #endif
151 
152 extern struct domain inetdomain;
153 static const struct protosw in_stf_protosw =
154 { SOCK_RAW,	&inetdomain,	IPPROTO_IPV6,	PR_ATOMIC|PR_ADDR,
155   in_stf_input, rip_output,	0,		rip_ctloutput,
156   rip_usrreq,
157   0,            0,              0,              0
158 };
159 
160 void	stfattach(int);
161 
162 static int stf_encapcheck(struct mbuf *, int, int, void *);
163 static struct in6_ifaddr *stf_getsrcifa6(struct ifnet *);
164 static int stf_output(struct ifnet *, struct mbuf *, const struct sockaddr *,
165 	struct rtentry *);
166 static int isrfc1918addr(const struct in_addr *);
167 static int stf_checkaddr4(struct stf_softc *, const struct in_addr *,
168 	struct ifnet *);
169 static int stf_checkaddr6(struct stf_softc *, const struct in6_addr *,
170 	struct ifnet *);
171 static void stf_rtrequest(int, struct rtentry *, const struct rt_addrinfo *);
172 static int stf_ioctl(struct ifnet *, u_long, void *);
173 
174 /* ARGSUSED */
175 void
176 stfattach(int count)
177 {
178 
179 	LIST_INIT(&stf_softc_list);
180 	if_clone_attach(&stf_cloner);
181 }
182 
183 static int
184 stf_clone_create(struct if_clone *ifc, int unit)
185 {
186 	struct stf_softc *sc;
187 
188 	if (LIST_FIRST(&stf_softc_list) != NULL) {
189 		/* Only one stf interface is allowed. */
190 		return (EEXIST);
191 	}
192 
193 	sc = malloc(sizeof(struct stf_softc), M_DEVBUF, M_WAIT|M_ZERO);
194 
195 	if_initname(&sc->sc_if, ifc->ifc_name, unit);
196 
197 	sc->encap_cookie = encap_attach_func(AF_INET, IPPROTO_IPV6,
198 	    stf_encapcheck, &in_stf_protosw, sc);
199 	if (sc->encap_cookie == NULL) {
200 		printf("%s: unable to attach encap\n", if_name(&sc->sc_if));
201 		free(sc, M_DEVBUF);
202 		return (EIO);	/* XXX */
203 	}
204 
205 	sc->sc_if.if_mtu    = STF_MTU;
206 	sc->sc_if.if_flags  = 0;
207 	sc->sc_if.if_ioctl  = stf_ioctl;
208 	sc->sc_if.if_output = stf_output;
209 	sc->sc_if.if_type   = IFT_STF;
210 	sc->sc_if.if_dlt    = DLT_NULL;
211 	if_attach(&sc->sc_if);
212 	if_alloc_sadl(&sc->sc_if);
213 	bpf_attach(&sc->sc_if, DLT_NULL, sizeof(u_int));
214 	LIST_INSERT_HEAD(&stf_softc_list, sc, sc_list);
215 	return (0);
216 }
217 
218 static int
219 stf_clone_destroy(struct ifnet *ifp)
220 {
221 	struct stf_softc *sc = (void *) ifp;
222 
223 	LIST_REMOVE(sc, sc_list);
224 	encap_detach(sc->encap_cookie);
225 	bpf_detach(ifp);
226 	if_detach(ifp);
227 	rtcache_free(&sc->sc_ro);
228 	free(sc, M_DEVBUF);
229 
230 	return (0);
231 }
232 
233 static int
234 stf_encapcheck(struct mbuf *m, int off, int proto, void *arg)
235 {
236 	struct ip ip;
237 	struct in6_ifaddr *ia6;
238 	struct stf_softc *sc;
239 	struct in_addr a, b;
240 
241 	sc = (struct stf_softc *)arg;
242 	if (sc == NULL)
243 		return 0;
244 
245 	if ((sc->sc_if.if_flags & IFF_UP) == 0)
246 		return 0;
247 
248 	/* IFF_LINK0 means "no decapsulation" */
249 	if ((sc->sc_if.if_flags & IFF_LINK0) != 0)
250 		return 0;
251 
252 	if (proto != IPPROTO_IPV6)
253 		return 0;
254 
255 	m_copydata(m, 0, sizeof(ip), (void *)&ip);
256 
257 	if (ip.ip_v != 4)
258 		return 0;
259 
260 	ia6 = stf_getsrcifa6(&sc->sc_if);
261 	if (ia6 == NULL)
262 		return 0;
263 
264 	/*
265 	 * check if IPv4 dst matches the IPv4 address derived from the
266 	 * local 6to4 address.
267 	 * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
268 	 */
269 	if (memcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst,
270 	    sizeof(ip.ip_dst)) != 0)
271 		return 0;
272 
273 	/*
274 	 * check if IPv4 src matches the IPv4 address derived from the
275 	 * local 6to4 address masked by prefixmask.
276 	 * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
277 	 * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
278 	 */
279 	memset(&a, 0, sizeof(a));
280 	a.s_addr = GET_V4(&ia6->ia_addr.sin6_addr)->s_addr;
281 	a.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
282 	b = ip.ip_src;
283 	b.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
284 	if (a.s_addr != b.s_addr)
285 		return 0;
286 
287 	/* stf interface makes single side match only */
288 	return 32;
289 }
290 
291 static struct in6_ifaddr *
292 stf_getsrcifa6(struct ifnet *ifp)
293 {
294 	struct ifaddr *ifa;
295 	struct in_ifaddr *ia4;
296 	struct sockaddr_in6 *sin6;
297 	struct in_addr in;
298 
299 	IFADDR_FOREACH(ifa, ifp)
300 	{
301 		if (ifa->ifa_addr == NULL)
302 			continue;
303 		if (ifa->ifa_addr->sa_family != AF_INET6)
304 			continue;
305 		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
306 		if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr))
307 			continue;
308 
309 		memcpy(&in, GET_V4(&sin6->sin6_addr), sizeof(in));
310 		INADDR_TO_IA(in, ia4);
311 		if (ia4 == NULL)
312 			continue;
313 
314 		return (struct in6_ifaddr *)ifa;
315 	}
316 
317 	return NULL;
318 }
319 
320 static int
321 stf_output(struct ifnet *ifp, struct mbuf *m, const struct sockaddr *dst,
322     struct rtentry *rt0)
323 {
324 	struct rtentry *rt;
325 	struct stf_softc *sc;
326 	const struct sockaddr_in6 *dst6;
327 	const struct in_addr *in4;
328 	uint8_t tos;
329 	struct ip *ip;
330 	struct ip6_hdr *ip6;
331 	struct in6_ifaddr *ia6;
332 	union {
333 		struct sockaddr		dst;
334 		struct sockaddr_in	dst4;
335 	} u;
336 
337 	sc = (struct stf_softc*)ifp;
338 	dst6 = (const struct sockaddr_in6 *)dst;
339 
340 	/* just in case */
341 	if ((ifp->if_flags & IFF_UP) == 0) {
342 		m_freem(m);
343 		return ENETDOWN;
344 	}
345 
346 	/*
347 	 * If we don't have an ip4 address that match my inner ip6 address,
348 	 * we shouldn't generate output.  Without this check, we'll end up
349 	 * using wrong IPv4 source.
350 	 */
351 	ia6 = stf_getsrcifa6(ifp);
352 	if (ia6 == NULL) {
353 		m_freem(m);
354 		ifp->if_oerrors++;
355 		return ENETDOWN;
356 	}
357 
358 	if (m->m_len < sizeof(*ip6)) {
359 		m = m_pullup(m, sizeof(*ip6));
360 		if (m == NULL) {
361 			ifp->if_oerrors++;
362 			return ENOBUFS;
363 		}
364 	}
365 	ip6 = mtod(m, struct ip6_hdr *);
366 	tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
367 
368 	/*
369 	 * Pickup the right outer dst addr from the list of candidates.
370 	 * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
371 	 */
372 	if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst))
373 		in4 = GET_V4(&ip6->ip6_dst);
374 	else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr))
375 		in4 = GET_V4(&dst6->sin6_addr);
376 	else {
377 		m_freem(m);
378 		ifp->if_oerrors++;
379 		return ENETUNREACH;
380 	}
381 
382 	bpf_mtap_af(ifp, AF_INET6, m);
383 
384 	M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
385 	if (m && m->m_len < sizeof(struct ip))
386 		m = m_pullup(m, sizeof(struct ip));
387 	if (m == NULL) {
388 		ifp->if_oerrors++;
389 		return ENOBUFS;
390 	}
391 	ip = mtod(m, struct ip *);
392 
393 	memset(ip, 0, sizeof(*ip));
394 
395 	bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr),
396 	    &ip->ip_src, sizeof(ip->ip_src));
397 	memcpy(&ip->ip_dst, in4, sizeof(ip->ip_dst));
398 	ip->ip_p = IPPROTO_IPV6;
399 	ip->ip_ttl = ip_gif_ttl;	/*XXX*/
400 	ip->ip_len = htons(m->m_pkthdr.len);
401 	if (ifp->if_flags & IFF_LINK1)
402 		ip_ecn_ingress(ECN_ALLOWED, &ip->ip_tos, &tos);
403 	else
404 		ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
405 
406 	sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
407 	if ((rt = rtcache_lookup(&sc->sc_ro, &u.dst)) == NULL) {
408 		m_freem(m);
409 		ifp->if_oerrors++;
410 		return ENETUNREACH;
411 	}
412 
413 	/* If the route constitutes infinite encapsulation, punt. */
414 	if (rt->rt_ifp == ifp) {
415 		rtcache_free(&sc->sc_ro);
416 		m_freem(m);
417 		ifp->if_oerrors++;
418 		return ENETUNREACH;
419 	}
420 
421 	ifp->if_opackets++;
422 	ifp->if_obytes += m->m_pkthdr.len - sizeof(struct ip);
423 	return ip_output(m, NULL, &sc->sc_ro, 0, NULL, NULL);
424 }
425 
426 static int
427 isrfc1918addr(const struct in_addr *in)
428 {
429 	/*
430 	 * returns 1 if private address range:
431 	 * 10.0.0.0/8 172.16.0.0/12 192.168.0.0/16
432 	 */
433 	if ((ntohl(in->s_addr) & 0xff000000) >> 24 == 10 ||
434 	    (ntohl(in->s_addr) & 0xfff00000) >> 16 == 172 * 256 + 16 ||
435 	    (ntohl(in->s_addr) & 0xffff0000) >> 16 == 192 * 256 + 168)
436 		return 1;
437 
438 	return 0;
439 }
440 
441 static int
442 stf_checkaddr4(struct stf_softc *sc, const struct in_addr *in,
443     struct ifnet *inifp /*incoming interface*/)
444 {
445 	struct in_ifaddr *ia4;
446 
447 	/*
448 	 * reject packets with the following address:
449 	 * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
450 	 */
451 	if (IN_MULTICAST(in->s_addr))
452 		return -1;
453 	switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
454 	case 0: case 127: case 255:
455 		return -1;
456 	}
457 
458 	/*
459 	 * reject packets with private address range.
460 	 * (requirement from RFC3056 section 2 1st paragraph)
461 	 */
462 	if (isrfc1918addr(in))
463 		return -1;
464 
465 	/*
466 	 * reject packet with IPv4 link-local (169.254.0.0/16),
467 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
468 	 */
469 	if (((ntohl(in->s_addr) & 0xff000000) >> 24) == 169 &&
470 	    ((ntohl(in->s_addr) & 0x00ff0000) >> 16) == 254)
471 		return -1;
472 
473 	/*
474 	 * reject packets with broadcast
475 	 */
476 	TAILQ_FOREACH(ia4, &in_ifaddrhead, ia_list)
477 	{
478 		if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0)
479 			continue;
480 		if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr)
481 			return -1;
482 	}
483 
484 	/*
485 	 * perform ingress filter
486 	 */
487 	if (sc && (sc->sc_if.if_flags & IFF_LINK2) == 0 && inifp) {
488 		struct sockaddr_in sin;
489 		struct rtentry *rt;
490 
491 		memset(&sin, 0, sizeof(sin));
492 		sin.sin_family = AF_INET;
493 		sin.sin_len = sizeof(struct sockaddr_in);
494 		sin.sin_addr = *in;
495 		rt = rtalloc1((struct sockaddr *)&sin, 0);
496 		if (!rt || rt->rt_ifp != inifp) {
497 #if 0
498 			log(LOG_WARNING, "%s: packet from 0x%x dropped "
499 			    "due to ingress filter\n", if_name(&sc->sc_if),
500 			    (uint32_t)ntohl(sin.sin_addr.s_addr));
501 #endif
502 			if (rt)
503 				rtfree(rt);
504 			return -1;
505 		}
506 		rtfree(rt);
507 	}
508 
509 	return 0;
510 }
511 
512 static int
513 stf_checkaddr6(struct stf_softc *sc, const struct in6_addr *in6,
514     struct ifnet *inifp /*incoming interface*/)
515 {
516 
517 	/*
518 	 * check 6to4 addresses
519 	 */
520 	if (IN6_IS_ADDR_6TO4(in6))
521 		return stf_checkaddr4(sc, GET_V4(in6), inifp);
522 
523 	/*
524 	 * reject anything that look suspicious.  the test is implemented
525 	 * in ip6_input too, but we check here as well to
526 	 * (1) reject bad packets earlier, and
527 	 * (2) to be safe against future ip6_input change.
528 	 */
529 	if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6))
530 		return -1;
531 
532 	/*
533 	 * reject link-local and site-local unicast
534 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
535 	 */
536 	if (IN6_IS_ADDR_LINKLOCAL(in6) || IN6_IS_ADDR_SITELOCAL(in6))
537 		return -1;
538 
539 	/*
540 	 * reject node-local and link-local multicast
541 	 * as suggested in draft-savola-v6ops-6to4-security-00.txt
542 	 */
543 	if (IN6_IS_ADDR_MC_NODELOCAL(in6) || IN6_IS_ADDR_MC_LINKLOCAL(in6))
544 		return -1;
545 
546 	return 0;
547 }
548 
549 void
550 in_stf_input(struct mbuf *m, ...)
551 {
552 	int off, proto;
553 	struct stf_softc *sc;
554 	struct ip *ip;
555 	struct ip6_hdr *ip6;
556 	uint8_t otos, itos;
557 	int s, isr;
558 	struct ifqueue *ifq = NULL;
559 	struct ifnet *ifp;
560 	va_list ap;
561 
562 	va_start(ap, m);
563 	off = va_arg(ap, int);
564 	proto = va_arg(ap, int);
565 	va_end(ap);
566 
567 	if (proto != IPPROTO_IPV6) {
568 		m_freem(m);
569 		return;
570 	}
571 
572 	ip = mtod(m, struct ip *);
573 
574 	sc = (struct stf_softc *)encap_getarg(m);
575 
576 	if (sc == NULL || (sc->sc_if.if_flags & IFF_UP) == 0) {
577 		m_freem(m);
578 		return;
579 	}
580 
581 	ifp = &sc->sc_if;
582 
583 	/*
584 	 * perform sanity check against outer src/dst.
585 	 * for source, perform ingress filter as well.
586 	 */
587 	if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
588 	    stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
589 		m_freem(m);
590 		return;
591 	}
592 
593 	otos = ip->ip_tos;
594 	m_adj(m, off);
595 
596 	if (m->m_len < sizeof(*ip6)) {
597 		m = m_pullup(m, sizeof(*ip6));
598 		if (!m)
599 			return;
600 	}
601 	ip6 = mtod(m, struct ip6_hdr *);
602 
603 	/*
604 	 * perform sanity check against inner src/dst.
605 	 * for source, perform ingress filter as well.
606 	 */
607 	if (stf_checkaddr6(sc, &ip6->ip6_dst, NULL) < 0 ||
608 	    stf_checkaddr6(sc, &ip6->ip6_src, m->m_pkthdr.rcvif) < 0) {
609 		m_freem(m);
610 		return;
611 	}
612 
613 	itos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
614 	if ((ifp->if_flags & IFF_LINK1) != 0)
615 		ip_ecn_egress(ECN_ALLOWED, &otos, &itos);
616 	else
617 		ip_ecn_egress(ECN_NOCARE, &otos, &itos);
618 	ip6->ip6_flow &= ~htonl(0xff << 20);
619 	ip6->ip6_flow |= htonl((uint32_t)itos << 20);
620 
621 	m->m_pkthdr.rcvif = ifp;
622 
623 	bpf_mtap_af(ifp, AF_INET6, m);
624 
625 	/*
626 	 * Put the packet to the network layer input queue according to the
627 	 * specified address family.
628 	 * See net/if_gif.c for possible issues with packet processing
629 	 * reorder due to extra queueing.
630 	 */
631 	ifq = &ip6intrq;
632 	isr = NETISR_IPV6;
633 
634 	s = splnet();
635 	if (IF_QFULL(ifq)) {
636 		IF_DROP(ifq);	/* update statistics */
637 		m_freem(m);
638 		splx(s);
639 		return;
640 	}
641 	IF_ENQUEUE(ifq, m);
642 	schednetisr(isr);
643 	ifp->if_ipackets++;
644 	ifp->if_ibytes += m->m_pkthdr.len;
645 	splx(s);
646 }
647 
648 /* ARGSUSED */
649 static void
650 stf_rtrequest(int cmd, struct rtentry *rt,
651     const struct rt_addrinfo *info)
652 {
653 	if (rt != NULL) {
654 		struct stf_softc *sc;
655 
656 		sc = LIST_FIRST(&stf_softc_list);
657 		rt->rt_rmx.rmx_mtu = (sc != NULL) ? sc->sc_if.if_mtu : STF_MTU;
658 	}
659 }
660 
661 static int
662 stf_ioctl(struct ifnet *ifp, u_long cmd, void *data)
663 {
664 	struct lwp		*l = curlwp;	/* XXX */
665 	struct ifaddr		*ifa;
666 	struct ifreq		*ifr = data;
667 	struct sockaddr_in6	*sin6;
668 	int			error;
669 
670 	error = 0;
671 	switch (cmd) {
672 	case SIOCINITIFADDR:
673 		ifa = (struct ifaddr *)data;
674 		if (ifa == NULL || ifa->ifa_addr->sa_family != AF_INET6) {
675 			error = EAFNOSUPPORT;
676 			break;
677 		}
678 		sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
679 		if (IN6_IS_ADDR_6TO4(&sin6->sin6_addr) &&
680 		    !isrfc1918addr(GET_V4(&sin6->sin6_addr))) {
681 			ifa->ifa_rtrequest = stf_rtrequest;
682 			ifp->if_flags |= IFF_UP;
683 		} else
684 			error = EINVAL;
685 		break;
686 
687 	case SIOCADDMULTI:
688 	case SIOCDELMULTI:
689 		if (ifr != NULL &&
690 		    ifreq_getaddr(cmd, ifr)->sa_family == AF_INET6)
691 			;
692 		else
693 			error = EAFNOSUPPORT;
694 		break;
695 
696 	case SIOCSIFMTU:
697 		error = kauth_authorize_network(l->l_cred,
698 		    KAUTH_NETWORK_INTERFACE,
699 		    KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, KAUTH_ARG(cmd),
700 		    NULL);
701 		if (error)
702 			break;
703 		if (ifr->ifr_mtu < STF_MTU_MIN || ifr->ifr_mtu > STF_MTU_MAX)
704 			return EINVAL;
705 		else if ((error = ifioctl_common(ifp, cmd, data)) == ENETRESET)
706 			error = 0;
707 		break;
708 
709 	default:
710 		error = ifioctl_common(ifp, cmd, data);
711 		break;
712 	}
713 
714 	return error;
715 }
716