Deleted Added
full compact
1/* $FreeBSD: head/sys/net/if_stf.c 129880 2004-05-30 20:27:19Z phk $ */
1/* $FreeBSD: head/sys/net/if_stf.c 130933 2004-06-22 20:13:25Z brooks $ */
2/* $KAME: if_stf.c,v 1.73 2001/12/03 11:08:30 keiichi 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 "opt_inet.h"
78#include "opt_inet6.h"
79#include "opt_mac.h"
80
81#include <sys/param.h>
82#include <sys/systm.h>
83#include <sys/socket.h>
84#include <sys/sockio.h>
85#include <sys/mac.h>
86#include <sys/mbuf.h>
87#include <sys/errno.h>
88#include <sys/kernel.h>
89#include <sys/module.h>
90#include <sys/protosw.h>
91#include <sys/queue.h>
92#include <machine/cpu.h>
93
94#include <sys/malloc.h>
95
96#include <net/if.h>
97#include <net/if_clone.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_var.h>
112#include <netinet/ip_ecn.h>
113
114#include <netinet/ip_encap.h>
115
116#include <machine/stdarg.h>
117
118#include <net/net_osdep.h>
119
120#include <net/bpf.h>
121
122#define STFNAME "stf"
123#define STFUNIT 0
124
125#define IN6_IS_ADDR_6TO4(x) (ntohs((x)->s6_addr16[0]) == 0x2002)
126
127/*
128 * XXX: Return a pointer with 16-bit aligned. Don't cast it to
129 * struct in_addr *; use bcopy() instead.
130 */
131#define GET_V4(x) ((caddr_t)(&(x)->s6_addr16[1]))
132
133struct stf_softc {
134 struct ifnet sc_if; /* common area */
135 union {
136 struct route __sc_ro4;
137 struct route_in6 __sc_ro6; /* just for safety */
138 } __sc_ro46;
139#define sc_ro __sc_ro46.__sc_ro4
140 const struct encaptab *encap_cookie;
141 LIST_ENTRY(stf_softc) sc_list; /* all stf's are linked */
142};
143
144/*
145 * All mutable global variables in if_stf.c are protected by stf_mtx.
146 * XXXRW: Note that mutable fields in the softc are not currently locked:
147 * in particular, sc_ro needs to be protected from concurrent entrance
148 * of stf_output().
149 */
150static struct mtx stf_mtx;
151static LIST_HEAD(, stf_softc) stf_softc_list;
152
153static MALLOC_DEFINE(M_STF, STFNAME, "6to4 Tunnel Interface");
154static const int ip_stf_ttl = 40;
155
156extern struct domain inetdomain;
157struct protosw in_stf_protosw =
158{ SOCK_RAW, &inetdomain, IPPROTO_IPV6, PR_ATOMIC|PR_ADDR,
159 in_stf_input, (pr_output_t*)rip_output, 0, rip_ctloutput,
160 0,
161 0, 0, 0, 0,
162 &rip_usrreqs
163};
164
165static char *stfnames[] = {"stf0", "stf", "6to4", NULL};
166
167static int stfmodevent(module_t, int, void *);
168static int stf_encapcheck(const struct mbuf *, int, int, void *);
169static struct in6_ifaddr *stf_getsrcifa6(struct ifnet *);
170static int stf_output(struct ifnet *, struct mbuf *, struct sockaddr *,
171 struct rtentry *);
172static int isrfc1918addr(struct in_addr *);
173static int stf_checkaddr4(struct stf_softc *, struct in_addr *,
174 struct ifnet *);
175static int stf_checkaddr6(struct stf_softc *, struct in6_addr *,
176 struct ifnet *);
177static void stf_rtrequest(int, struct rtentry *, struct rt_addrinfo *);
178static int stf_ioctl(struct ifnet *, u_long, caddr_t);
179
176static int stf_clone_create(struct if_clone *, int);
177static void stf_clone_destroy(struct ifnet *);
180static int stf_clone_match(struct if_clone *, const char *);
181static int stf_clone_create(struct if_clone *, char *, size_t);
182static int stf_clone_destroy(struct if_clone *, struct ifnet *);
183struct if_clone stf_cloner = IFC_CLONE_INITIALIZER(STFNAME, NULL, 0,
184 NULL, stf_clone_match, stf_clone_create, stf_clone_destroy);
185
179/* only one clone is currently allowed */
180struct if_clone stf_cloner =
181 IF_CLONE_INITIALIZER(STFNAME, stf_clone_create, stf_clone_destroy, 0, 0);
186static int
187stf_clone_match(struct if_clone *ifc, const char *name)
188{
189 int i;
190
191 for(i = 0; stfnames[i] != NULL; i++) {
192 if (strcmp(stfnames[i], name) == 0)
193 return (1);
194 }
195
196 return (0);
197}
198
199static int
184stf_clone_create(ifc, unit)
185 struct if_clone *ifc;
186 int unit;
200stf_clone_create(struct if_clone *ifc, char *name, size_t len)
201{
202 int err, unit;
203 struct stf_softc *sc;
204 struct ifnet *ifp;
205
206 /*
207 * We can only have one unit, but since unit allocation is
208 * already locked, we use it to keep from allocating extra
209 * interfaces.
210 */
211 unit = STFUNIT;
212 err = ifc_alloc_unit(ifc, &unit);
213 if (err != 0)
214 return (err);
215
216 sc = malloc(sizeof(struct stf_softc), M_STF, M_WAITOK | M_ZERO);
217 ifp = &sc->sc_if;
193 if_initname(ifp, ifc->ifc_name, unit);
218 /*
219 * Set the name manually rather then using if_initname because
220 * we don't conform to the default naming convention for interfaces.
221 */
222 strlcpy(ifp->if_xname, name, IFNAMSIZ);
223 ifp->if_dname = ifc->ifc_name;
224 ifp->if_dunit = IF_DUNIT_NONE;
225
226 sc->encap_cookie = encap_attach_func(AF_INET, IPPROTO_IPV6,
227 stf_encapcheck, &in_stf_protosw, sc);
228 if (sc->encap_cookie == NULL) {
229 if_printf(ifp, "attach failed\n");
230 free(sc, M_STF);
231 ifc_free_unit(ifc, unit);
232 return (ENOMEM);
233 }
234
235 ifp->if_mtu = IPV6_MMTU;
236 ifp->if_ioctl = stf_ioctl;
237 ifp->if_output = stf_output;
238 ifp->if_type = IFT_STF;
239 ifp->if_snd.ifq_maxlen = IFQ_MAXLEN;
240 if_attach(ifp);
241 bpfattach(ifp, DLT_NULL, sizeof(u_int));
242 mtx_lock(&stf_mtx);
243 LIST_INSERT_HEAD(&stf_softc_list, sc, sc_list);
244 mtx_unlock(&stf_mtx);
245 return (0);
246}
247
248static void
249stf_destroy(struct stf_softc *sc)
250{
251 int err;
252
253 err = encap_detach(sc->encap_cookie);
254 KASSERT(err == 0, ("Unexpected error detaching encap_cookie"));
255 bpfdetach(&sc->sc_if);
256 if_detach(&sc->sc_if);
257
258 free(sc, M_STF);
259}
260
229static void
230stf_clone_destroy(ifp)
231 struct ifnet *ifp;
261static int
262stf_clone_destroy(struct if_clone *ifc, struct ifnet *ifp)
263{
264 struct stf_softc *sc = (void *) ifp;
265
266 mtx_lock(&stf_mtx);
267 LIST_REMOVE(sc, sc_list);
268 mtx_unlock(&stf_mtx);
269
270 stf_destroy(sc);
271 ifc_free_unit(ifc, STFUNIT);
272
273 return (0);
274}
275
276static int
277stfmodevent(mod, type, data)
278 module_t mod;
279 int type;
280 void *data;
281{
282 struct stf_softc *sc;
283
284 switch (type) {
285 case MOD_LOAD:
286 mtx_init(&stf_mtx, "stf_mtx", NULL, MTX_DEF);
287 LIST_INIT(&stf_softc_list);
288 if_clone_attach(&stf_cloner);
289
290 break;
291 case MOD_UNLOAD:
292 if_clone_detach(&stf_cloner);
293
294 mtx_lock(&stf_mtx);
295 while ((sc = LIST_FIRST(&stf_softc_list)) != NULL) {
296 LIST_REMOVE(sc, sc_list);
297 mtx_unlock(&stf_mtx);
298 stf_destroy(sc);
299 mtx_lock(&stf_mtx);
300 }
301 mtx_unlock(&stf_mtx);
302 mtx_destroy(&stf_mtx);
303 break;
304 }
305
306 return (0);
307}
308
309static moduledata_t stf_mod = {
310 "if_stf",
311 stfmodevent,
312 0
313};
314
315DECLARE_MODULE(if_stf, stf_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
316
317static int
318stf_encapcheck(m, off, proto, arg)
319 const struct mbuf *m;
320 int off;
321 int proto;
322 void *arg;
323{
324 struct ip ip;
325 struct in6_ifaddr *ia6;
326 struct stf_softc *sc;
327 struct in_addr a, b, mask;
328
329 sc = (struct stf_softc *)arg;
330 if (sc == NULL)
331 return 0;
332
333 if ((sc->sc_if.if_flags & IFF_UP) == 0)
334 return 0;
335
336 /* IFF_LINK0 means "no decapsulation" */
337 if ((sc->sc_if.if_flags & IFF_LINK0) != 0)
338 return 0;
339
340 if (proto != IPPROTO_IPV6)
341 return 0;
342
343 /* LINTED const cast */
344 m_copydata((struct mbuf *)(uintptr_t)m, 0, sizeof(ip), (caddr_t)&ip);
345
346 if (ip.ip_v != 4)
347 return 0;
348
349 ia6 = stf_getsrcifa6(&sc->sc_if);
350 if (ia6 == NULL)
351 return 0;
352
353 /*
354 * check if IPv4 dst matches the IPv4 address derived from the
355 * local 6to4 address.
356 * success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
357 */
358 if (bcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst,
359 sizeof(ip.ip_dst)) != 0)
360 return 0;
361
362 /*
363 * check if IPv4 src matches the IPv4 address derived from the
364 * local 6to4 address masked by prefixmask.
365 * success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
366 * fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
367 */
368 bzero(&a, sizeof(a));
369 bcopy(GET_V4(&ia6->ia_addr.sin6_addr), &a, sizeof(a));
370 bcopy(GET_V4(&ia6->ia_prefixmask.sin6_addr), &mask, sizeof(mask));
371 a.s_addr &= mask.s_addr;
372 b = ip.ip_src;
373 b.s_addr &= mask.s_addr;
374 if (a.s_addr != b.s_addr)
375 return 0;
376
377 /* stf interface makes single side match only */
378 return 32;
379}
380
381static struct in6_ifaddr *
382stf_getsrcifa6(ifp)
383 struct ifnet *ifp;
384{
385 struct ifaddr *ia;
386 struct in_ifaddr *ia4;
387 struct sockaddr_in6 *sin6;
388 struct in_addr in;
389
390 for (ia = TAILQ_FIRST(&ifp->if_addrlist);
391 ia;
392 ia = TAILQ_NEXT(ia, ifa_list))
393 {
394 if (ia->ifa_addr == NULL)
395 continue;
396 if (ia->ifa_addr->sa_family != AF_INET6)
397 continue;
398 sin6 = (struct sockaddr_in6 *)ia->ifa_addr;
399 if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr))
400 continue;
401
402 bcopy(GET_V4(&sin6->sin6_addr), &in, sizeof(in));
403 LIST_FOREACH(ia4, INADDR_HASH(in.s_addr), ia_hash)
404 if (ia4->ia_addr.sin_addr.s_addr == in.s_addr)
405 break;
406 if (ia4 == NULL)
407 continue;
408
409 return (struct in6_ifaddr *)ia;
410 }
411
412 return NULL;
413}
414
415static int
416stf_output(ifp, m, dst, rt)
417 struct ifnet *ifp;
418 struct mbuf *m;
419 struct sockaddr *dst;
420 struct rtentry *rt;
421{
422 struct stf_softc *sc;
423 struct sockaddr_in6 *dst6;
424 struct in_addr in4;
425 caddr_t ptr;
426 struct sockaddr_in *dst4;
427 u_int8_t tos;
428 struct ip *ip;
429 struct ip6_hdr *ip6;
430 struct in6_ifaddr *ia6;
431#ifdef MAC
432 int error;
433
434 error = mac_check_ifnet_transmit(ifp, m);
435 if (error) {
436 m_freem(m);
437 return (error);
438 }
439#endif
440
441 sc = (struct stf_softc*)ifp;
442 dst6 = (struct sockaddr_in6 *)dst;
443
444 /* just in case */
445 if ((ifp->if_flags & IFF_UP) == 0) {
446 m_freem(m);
447 ifp->if_oerrors++;
448 return ENETDOWN;
449 }
450
451 /*
452 * If we don't have an ip4 address that match my inner ip6 address,
453 * we shouldn't generate output. Without this check, we'll end up
454 * using wrong IPv4 source.
455 */
456 ia6 = stf_getsrcifa6(ifp);
457 if (ia6 == NULL) {
458 m_freem(m);
459 ifp->if_oerrors++;
460 return ENETDOWN;
461 }
462
463 if (m->m_len < sizeof(*ip6)) {
464 m = m_pullup(m, sizeof(*ip6));
465 if (!m) {
466 ifp->if_oerrors++;
467 return ENOBUFS;
468 }
469 }
470 ip6 = mtod(m, struct ip6_hdr *);
471 tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
472
473 /*
474 * Pickup the right outer dst addr from the list of candidates.
475 * ip6_dst has priority as it may be able to give us shorter IPv4 hops.
476 */
477 ptr = NULL;
478 if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst))
479 ptr = GET_V4(&ip6->ip6_dst);
480 else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr))
481 ptr = GET_V4(&dst6->sin6_addr);
482 else {
483 m_freem(m);
484 ifp->if_oerrors++;
485 return ENETUNREACH;
486 }
487 bcopy(ptr, &in4, sizeof(in4));
488
489#if NBPFILTER > 0
490 if (ifp->if_bpf) {
491 /*
492 * We need to prepend the address family as
493 * a four byte field. Cons up a dummy header
494 * to pacify bpf. This is safe because bpf
495 * will only read from the mbuf (i.e., it won't
496 * try to free it or keep a pointer a to it).
497 */
498 u_int32_t af = AF_INET6;
499#ifdef HAVE_OLD_BPF
500 struct mbuf m0;
501
502 m0.m_next = m;
503 m0.m_len = 4;
504 m0.m_data = (char *)&af;
505
506 BPF_MTAP(ifp, &m0);
507#else
508 bpf_mtap2(ifp->if_bpf, &af, sizeof(af), m);
509#endif
510 }
511#endif /*NBPFILTER > 0*/
512
513 M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
514 if (m && m->m_len < sizeof(struct ip))
515 m = m_pullup(m, sizeof(struct ip));
516 if (m == NULL) {
517 ifp->if_oerrors++;
518 return ENOBUFS;
519 }
520 ip = mtod(m, struct ip *);
521
522 bzero(ip, sizeof(*ip));
523
524 bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr),
525 &ip->ip_src, sizeof(ip->ip_src));
526 bcopy(&in4, &ip->ip_dst, sizeof(ip->ip_dst));
527 ip->ip_p = IPPROTO_IPV6;
528 ip->ip_ttl = ip_stf_ttl;
529 ip->ip_len = m->m_pkthdr.len; /*host order*/
530 if (ifp->if_flags & IFF_LINK1)
531 ip_ecn_ingress(ECN_ALLOWED, &ip->ip_tos, &tos);
532 else
533 ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
534
535 /*
536 * XXXRW: Locking of sc_ro required.
537 */
538 dst4 = (struct sockaddr_in *)&sc->sc_ro.ro_dst;
539 if (dst4->sin_family != AF_INET ||
540 bcmp(&dst4->sin_addr, &ip->ip_dst, sizeof(ip->ip_dst)) != 0) {
541 /* cache route doesn't match */
542 dst4->sin_family = AF_INET;
543 dst4->sin_len = sizeof(struct sockaddr_in);
544 bcopy(&ip->ip_dst, &dst4->sin_addr, sizeof(dst4->sin_addr));
545 if (sc->sc_ro.ro_rt) {
546 RTFREE(sc->sc_ro.ro_rt);
547 sc->sc_ro.ro_rt = NULL;
548 }
549 }
550
551 if (sc->sc_ro.ro_rt == NULL) {
552 rtalloc(&sc->sc_ro);
553 if (sc->sc_ro.ro_rt == NULL) {
554 m_freem(m);
555 ifp->if_oerrors++;
556 return ENETUNREACH;
557 }
558 }
559
560 ifp->if_opackets++;
561 return ip_output(m, NULL, &sc->sc_ro, 0, NULL, NULL);
562}
563
564static int
565isrfc1918addr(in)
566 struct in_addr *in;
567{
568 /*
569 * returns 1 if private address range:
570 * 10.0.0.0/8 172.16.0.0/12 192.168.0.0/16
571 */
572 if ((ntohl(in->s_addr) & 0xff000000) >> 24 == 10 ||
573 (ntohl(in->s_addr) & 0xfff00000) >> 16 == 172 * 256 + 16 ||
574 (ntohl(in->s_addr) & 0xffff0000) >> 16 == 192 * 256 + 168)
575 return 1;
576
577 return 0;
578}
579
580static int
581stf_checkaddr4(sc, in, inifp)
582 struct stf_softc *sc;
583 struct in_addr *in;
584 struct ifnet *inifp; /* incoming interface */
585{
586 struct in_ifaddr *ia4;
587
588 /*
589 * reject packets with the following address:
590 * 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
591 */
592 if (IN_MULTICAST(ntohl(in->s_addr)))
593 return -1;
594 switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
595 case 0: case 127: case 255:
596 return -1;
597 }
598
599 /*
600 * reject packets with private address range.
601 * (requirement from RFC3056 section 2 1st paragraph)
602 */
603 if (isrfc1918addr(in))
604 return -1;
605
606 /*
607 * reject packets with broadcast
608 */
609 for (ia4 = TAILQ_FIRST(&in_ifaddrhead);
610 ia4;
611 ia4 = TAILQ_NEXT(ia4, ia_link))
612 {
613 if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0)
614 continue;
615 if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr)
616 return -1;
617 }
618
619 /*
620 * perform ingress filter
621 */
622 if (sc && (sc->sc_if.if_flags & IFF_LINK2) == 0 && inifp) {
623 struct sockaddr_in sin;
624 struct rtentry *rt;
625
626 bzero(&sin, sizeof(sin));
627 sin.sin_family = AF_INET;
628 sin.sin_len = sizeof(struct sockaddr_in);
629 sin.sin_addr = *in;
630 rt = rtalloc1((struct sockaddr *)&sin, 0, 0UL);
631 if (!rt || rt->rt_ifp != inifp) {
632#if 0
633 log(LOG_WARNING, "%s: packet from 0x%x dropped "
634 "due to ingress filter\n", if_name(&sc->sc_if),
635 (u_int32_t)ntohl(sin.sin_addr.s_addr));
636#endif
637 if (rt)
638 rtfree(rt);
639 return -1;
640 }
641 rtfree(rt);
642 }
643
644 return 0;
645}
646
647static int
648stf_checkaddr6(sc, in6, inifp)
649 struct stf_softc *sc;
650 struct in6_addr *in6;
651 struct ifnet *inifp; /* incoming interface */
652{
653 /*
654 * check 6to4 addresses
655 */
656 if (IN6_IS_ADDR_6TO4(in6)) {
657 struct in_addr in4;
658 bcopy(GET_V4(in6), &in4, sizeof(in4));
659 return stf_checkaddr4(sc, &in4, inifp);
660 }
661
662 /*
663 * reject anything that look suspicious. the test is implemented
664 * in ip6_input too, but we check here as well to
665 * (1) reject bad packets earlier, and
666 * (2) to be safe against future ip6_input change.
667 */
668 if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6))
669 return -1;
670
671 return 0;
672}
673
674void
675in_stf_input(m, off)
676 struct mbuf *m;
677 int off;
678{
679 int proto;
680 struct stf_softc *sc;
681 struct ip *ip;
682 struct ip6_hdr *ip6;
683 u_int8_t otos, itos;
684 struct ifnet *ifp;
685
686 proto = mtod(m, struct ip *)->ip_p;
687
688 if (proto != IPPROTO_IPV6) {
689 m_freem(m);
690 return;
691 }
692
693 ip = mtod(m, struct ip *);
694
695 sc = (struct stf_softc *)encap_getarg(m);
696
697 if (sc == NULL || (sc->sc_if.if_flags & IFF_UP) == 0) {
698 m_freem(m);
699 return;
700 }
701
702 ifp = &sc->sc_if;
703
704#ifdef MAC
705 mac_create_mbuf_from_ifnet(ifp, m);
706#endif
707
708 /*
709 * perform sanity check against outer src/dst.
710 * for source, perform ingress filter as well.
711 */
712 if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
713 stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
714 m_freem(m);
715 return;
716 }
717
718 otos = ip->ip_tos;
719 m_adj(m, off);
720
721 if (m->m_len < sizeof(*ip6)) {
722 m = m_pullup(m, sizeof(*ip6));
723 if (!m)
724 return;
725 }
726 ip6 = mtod(m, struct ip6_hdr *);
727
728 /*
729 * perform sanity check against inner src/dst.
730 * for source, perform ingress filter as well.
731 */
732 if (stf_checkaddr6(sc, &ip6->ip6_dst, NULL) < 0 ||
733 stf_checkaddr6(sc, &ip6->ip6_src, m->m_pkthdr.rcvif) < 0) {
734 m_freem(m);
735 return;
736 }
737
738 itos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
739 if ((ifp->if_flags & IFF_LINK1) != 0)
740 ip_ecn_egress(ECN_ALLOWED, &otos, &itos);
741 else
742 ip_ecn_egress(ECN_NOCARE, &otos, &itos);
743 ip6->ip6_flow &= ~htonl(0xff << 20);
744 ip6->ip6_flow |= htonl((u_int32_t)itos << 20);
745
746 m->m_pkthdr.rcvif = ifp;
747
748 if (ifp->if_bpf) {
749 /*
750 * We need to prepend the address family as
751 * a four byte field. Cons up a dummy header
752 * to pacify bpf. This is safe because bpf
753 * will only read from the mbuf (i.e., it won't
754 * try to free it or keep a pointer a to it).
755 */
756 u_int32_t af = AF_INET6;
757#ifdef HAVE_OLD_BPF
758 struct mbuf m0;
759
760 m0.m_next = m;
761 m0.m_len = 4;
762 m0.m_data = (char *)&af;
763
764 BPF_MTAP(ifp, &m0);
765#else
766 bpf_mtap2(ifp->if_bpf, &af, sizeof(ah), m);
767#endif
768 }
769
770 /*
771 * Put the packet to the network layer input queue according to the
772 * specified address family.
773 * See net/if_gif.c for possible issues with packet processing
774 * reorder due to extra queueing.
775 */
776 ifp->if_ipackets++;
777 ifp->if_ibytes += m->m_pkthdr.len;
778 netisr_dispatch(NETISR_IPV6, m);
779}
780
781/* ARGSUSED */
782static void
783stf_rtrequest(cmd, rt, info)
784 int cmd;
785 struct rtentry *rt;
786 struct rt_addrinfo *info;
787{
788 RT_LOCK_ASSERT(rt);
789
790 if (rt)
791 rt->rt_rmx.rmx_mtu = IPV6_MMTU;
792}
793
794static int
795stf_ioctl(ifp, cmd, data)
796 struct ifnet *ifp;
797 u_long cmd;
798 caddr_t data;
799{
800 struct ifaddr *ifa;
801 struct ifreq *ifr;
802 struct sockaddr_in6 *sin6;
803 struct in_addr addr;
804 int error;
805
806 error = 0;
807 switch (cmd) {
808 case SIOCSIFADDR:
809 ifa = (struct ifaddr *)data;
810 if (ifa == NULL || ifa->ifa_addr->sa_family != AF_INET6) {
811 error = EAFNOSUPPORT;
812 break;
813 }
814 sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
815 if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
816 error = EINVAL;
817 break;
818 }
819 bcopy(GET_V4(&sin6->sin6_addr), &addr, sizeof(addr));
820 if (isrfc1918addr(&addr)) {
821 error = EINVAL;
822 break;
823 }
824
825 ifa->ifa_rtrequest = stf_rtrequest;
826 ifp->if_flags |= IFF_UP;
827 break;
828
829 case SIOCADDMULTI:
830 case SIOCDELMULTI:
831 ifr = (struct ifreq *)data;
832 if (ifr && ifr->ifr_addr.sa_family == AF_INET6)
833 ;
834 else
835 error = EAFNOSUPPORT;
836 break;
837
838 default:
839 error = EINVAL;
840 break;
841 }
842
843 return error;
844}