xref: /netbsd-src/sys/netinet6/ip6_input.c (revision 3b01aba77a7a698587faaae455bbfe740923c1f5)
1 /*	$NetBSD: ip6_input.c,v 1.42 2001/08/06 10:25:01 itojun Exp $	*/
2 /*	$KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $	*/
3 
4 /*
5  * Copyright (C) 1995, 1996, 1997, and 1998 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  * Copyright (c) 1982, 1986, 1988, 1993
35  *	The Regents of the University of California.  All rights reserved.
36  *
37  * Redistribution and use in source and binary forms, with or without
38  * modification, are permitted provided that the following conditions
39  * are met:
40  * 1. Redistributions of source code must retain the above copyright
41  *    notice, this list of conditions and the following disclaimer.
42  * 2. Redistributions in binary form must reproduce the above copyright
43  *    notice, this list of conditions and the following disclaimer in the
44  *    documentation and/or other materials provided with the distribution.
45  * 3. All advertising materials mentioning features or use of this software
46  *    must display the following acknowledgement:
47  *	This product includes software developed by the University of
48  *	California, Berkeley and its contributors.
49  * 4. Neither the name of the University nor the names of its contributors
50  *    may be used to endorse or promote products derived from this software
51  *    without specific prior written permission.
52  *
53  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
63  * SUCH DAMAGE.
64  *
65  *	@(#)ip_input.c	8.2 (Berkeley) 1/4/94
66  */
67 
68 #include "opt_inet.h"
69 #include "opt_ipsec.h"
70 #include "opt_pfil_hooks.h"
71 
72 #include <sys/param.h>
73 #include <sys/systm.h>
74 #include <sys/malloc.h>
75 #include <sys/mbuf.h>
76 #include <sys/domain.h>
77 #include <sys/protosw.h>
78 #include <sys/socket.h>
79 #include <sys/socketvar.h>
80 #include <sys/errno.h>
81 #include <sys/time.h>
82 #include <sys/kernel.h>
83 #include <sys/syslog.h>
84 #include <sys/proc.h>
85 
86 #include <net/if.h>
87 #include <net/if_types.h>
88 #include <net/if_dl.h>
89 #include <net/route.h>
90 #include <net/netisr.h>
91 #ifdef PFIL_HOOKS
92 #include <net/pfil.h>
93 #endif
94 
95 #include <netinet/in.h>
96 #include <netinet/in_systm.h>
97 #ifdef INET
98 #include <netinet/ip.h>
99 #include <netinet/ip_icmp.h>
100 #endif /*INET*/
101 #include <netinet/ip6.h>
102 #include <netinet6/in6_var.h>
103 #include <netinet6/ip6_var.h>
104 #include <netinet6/in6_pcb.h>
105 #include <netinet/icmp6.h>
106 #include <netinet6/in6_ifattach.h>
107 #include <netinet6/nd6.h>
108 #include <netinet6/in6_prefix.h>
109 
110 #ifdef IPSEC
111 #include <netinet6/ipsec.h>
112 #endif
113 
114 #include <netinet6/ip6protosw.h>
115 
116 /* we need it for NLOOP. */
117 #include "loop.h"
118 #include "faith.h"
119 
120 #include "gif.h"
121 #include "bpfilter.h"
122 
123 #include <net/net_osdep.h>
124 
125 extern struct domain inet6domain;
126 
127 u_char ip6_protox[IPPROTO_MAX];
128 static int ip6qmaxlen = IFQ_MAXLEN;
129 struct in6_ifaddr *in6_ifaddr;
130 struct ifqueue ip6intrq;
131 
132 extern struct ifnet loif[NLOOP];
133 int ip6_forward_srcrt;			/* XXX */
134 int ip6_sourcecheck;			/* XXX */
135 int ip6_sourcecheck_interval;		/* XXX */
136 
137 #ifdef PFIL_HOOKS
138 struct pfil_head inet6_pfil_hook;
139 #endif
140 
141 struct ip6stat ip6stat;
142 
143 static void ip6_init2 __P((void *));
144 
145 static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
146 
147 /*
148  * IP6 initialization: fill in IP6 protocol switch table.
149  * All protocols not implemented in kernel go to raw IP6 protocol handler.
150  */
151 void
152 ip6_init()
153 {
154 	struct ip6protosw *pr;
155 	int i;
156 	struct timeval tv;
157 
158 	pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
159 	if (pr == 0)
160 		panic("ip6_init");
161 	for (i = 0; i < IPPROTO_MAX; i++)
162 		ip6_protox[i] = pr - inet6sw;
163 	for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
164 	    pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
165 		if (pr->pr_domain->dom_family == PF_INET6 &&
166 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
167 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
168 	ip6intrq.ifq_maxlen = ip6qmaxlen;
169 	nd6_init();
170 	frag6_init();
171 	/*
172 	 * in many cases, random() here does NOT return random number
173 	 * as initialization during bootstrap time occur in fixed order.
174 	 */
175 	microtime(&tv);
176 	ip6_flow_seq = random() ^ tv.tv_usec;
177 
178 	ip6_init2((void *)0);
179 
180 #ifdef PFIL_HOOKS
181 	/* Register our Packet Filter hook. */
182 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
183 	inet6_pfil_hook.ph_af   = AF_INET6;
184 	i = pfil_head_register(&inet6_pfil_hook);
185 	if (i != 0)
186 		printf("ip6_init: WARNING: unable to register pfil hook, "
187 		    "error %d\n", i);
188 #endif /* PFIL_HOOKS */
189 }
190 
191 static void
192 ip6_init2(dummy)
193 	void *dummy;
194 {
195 	/*
196 	 * to route local address of p2p link to loopback,
197 	 * assign loopback address first.
198 	 */
199 	in6_ifattach(&loif[0], NULL);
200 
201 	/* nd6_timer_init */
202 	callout_init(&nd6_timer_ch);
203 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
204 	/* router renumbering prefix list maintenance */
205 	callout_init(&in6_rr_timer_ch);
206 	callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
207 }
208 
209 /*
210  * IP6 input interrupt handling. Just pass the packet to ip6_input.
211  */
212 void
213 ip6intr()
214 {
215 	int s;
216 	struct mbuf *m;
217 
218 	for (;;) {
219 		s = splnet();
220 		IF_DEQUEUE(&ip6intrq, m);
221 		splx(s);
222 		if (m == 0)
223 			return;
224 		ip6_input(m);
225 	}
226 }
227 
228 extern struct	route_in6 ip6_forward_rt;
229 
230 void
231 ip6_input(m)
232 	struct mbuf *m;
233 {
234 	struct ip6_hdr *ip6;
235 	int off = sizeof(struct ip6_hdr), nest;
236 	u_int32_t plen;
237 	u_int32_t rtalert = ~0;
238 	int nxt, ours = 0;
239 	struct ifnet *deliverifp = NULL;
240 
241 #ifdef IPSEC
242 	/*
243 	 * should the inner packet be considered authentic?
244 	 * see comment in ah4_input().
245 	 */
246 	if (m) {
247 		m->m_flags &= ~M_AUTHIPHDR;
248 		m->m_flags &= ~M_AUTHIPDGM;
249 	}
250 #endif
251 
252 	/*
253 	 * mbuf statistics by kazu
254 	 */
255 	if (m->m_flags & M_EXT) {
256 		if (m->m_next)
257 			ip6stat.ip6s_mext2m++;
258 		else
259 			ip6stat.ip6s_mext1++;
260 	} else {
261 #define M2MMAX	(sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
262 		if (m->m_next) {
263 			if (m->m_flags & M_LOOP) {
264 				ip6stat.ip6s_m2m[loif[0].if_index]++;	/*XXX*/
265 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
266 				ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
267 			else
268 				ip6stat.ip6s_m2m[0]++;
269 		} else
270 			ip6stat.ip6s_m1++;
271 #undef M2MMAX
272 	}
273 
274 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
275 	ip6stat.ip6s_total++;
276 
277 #ifndef PULLDOWN_TEST
278 	/* XXX is the line really necessary? */
279 	IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/);
280 #endif
281 
282 	if (m->m_len < sizeof(struct ip6_hdr)) {
283 		struct ifnet *inifp;
284 		inifp = m->m_pkthdr.rcvif;
285 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
286 			ip6stat.ip6s_toosmall++;
287 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
288 			return;
289 		}
290 	}
291 
292 	ip6 = mtod(m, struct ip6_hdr *);
293 
294 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
295 		ip6stat.ip6s_badvers++;
296 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
297 		goto bad;
298 	}
299 
300 #ifdef PFIL_HOOKS
301 	/*
302 	 * Run through list of hooks for input packets.  If there are any
303 	 * filters which require that additional packets in the flow are
304 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
305 	 * Note that filters must _never_ set this flag, as another filter
306 	 * in the list may have previously cleared it.
307 	 */
308 	/*
309 	 * let ipfilter look at packet on the wire,
310 	 * not the decapsulated packet.
311 	 */
312 #ifdef IPSEC
313 	if (!ipsec_getnhist(m))
314 #else
315 	if (1)
316 #endif
317 	{
318 		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
319 				   PFIL_IN) != 0)
320 			return;
321 		if (m == NULL)
322 			return;
323 		ip6 = mtod(m, struct ip6_hdr *);
324 	}
325 #endif /* PFIL_HOOKS */
326 
327 
328 	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
329 
330 #ifdef ALTQ
331 	/* XXX Temporary until ALTQ is changed to use a pfil hook */
332 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
333 		/* packet is dropped by traffic conditioner */
334 		return;
335 	}
336 #endif
337 
338 	/*
339 	 * Scope check
340 	 */
341 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
342 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
343 		ip6stat.ip6s_badscope++;
344 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
345 		goto bad;
346 	}
347 	/*
348 	 * The following check is not documented in the spec.  Malicious party
349 	 * may be able to use IPv4 mapped addr to confuse tcp/udp stack and
350 	 * bypass security checks (act as if it was from 127.0.0.1 by using
351 	 * IPv6 src ::ffff:127.0.0.1).	Be cautious.
352 	 *
353 	 * This check chokes if we are in SIIT cloud.  As none of BSDs support
354 	 * IPv4-less kernel compilation, we cannot support SIIT environment
355 	 * at all.  So, it makes more sense for us to reject any malicious
356 	 * packets for non-SIIT environment, than try to do a partical support
357 	 * for SIIT environment.
358 	 */
359 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
360 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
361 		ip6stat.ip6s_badscope++;
362 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
363 		goto bad;
364 	}
365 #if 0
366 	/*
367 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
368 	 *
369 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
370 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
371 	 * is revised to forbid relaying case.
372 	 */
373 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
374 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
375 		ip6stat.ip6s_badscope++;
376 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
377 		goto bad;
378 	}
379 #endif
380 
381 	if (IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
382 	    IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) {
383 		if (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) {
384 			ours = 1;
385 			deliverifp = m->m_pkthdr.rcvif;
386 			goto hbhcheck;
387 		} else {
388 			ip6stat.ip6s_badscope++;
389 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
390 			goto bad;
391 		}
392 	}
393 
394 	/* drop packets if interface ID portion is already filled */
395 	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
396 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
397 		    ip6->ip6_src.s6_addr16[1]) {
398 			ip6stat.ip6s_badscope++;
399 			goto bad;
400 		}
401 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) &&
402 		    ip6->ip6_dst.s6_addr16[1]) {
403 			ip6stat.ip6s_badscope++;
404 			goto bad;
405 		}
406 	}
407 
408 	if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
409 		ip6->ip6_src.s6_addr16[1]
410 			= htons(m->m_pkthdr.rcvif->if_index);
411 	if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
412 		ip6->ip6_dst.s6_addr16[1]
413 			= htons(m->m_pkthdr.rcvif->if_index);
414 
415 	/*
416 	 * We use rt->rt_ifp to determine if the address is ours or not.
417 	 * If rt_ifp is lo0, the address is ours.
418 	 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0,
419 	 * so any address under fe80::%lo0/64 will be mistakenly considered
420 	 * local.  The special case is supplied to handle the case properly
421 	 * by actually looking at interface addresses
422 	 * (using in6ifa_ifpwithaddr).
423 	 */
424 	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 &&
425 	    IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) {
426 		if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) {
427 			icmp6_error(m, ICMP6_DST_UNREACH,
428 			    ICMP6_DST_UNREACH_ADDR, 0);
429 			/* m is already freed */
430 			return;
431 		}
432 
433 		ours = 1;
434 		deliverifp = m->m_pkthdr.rcvif;
435 		goto hbhcheck;
436 	}
437 
438 	/*
439 	 * Multicast check
440 	 */
441 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
442 	  	struct	in6_multi *in6m = 0;
443 
444 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
445 		/*
446 		 * See if we belong to the destination multicast group on the
447 		 * arrival interface.
448 		 */
449 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
450 		if (in6m)
451 			ours = 1;
452 		else if (!ip6_mrouter) {
453 			ip6stat.ip6s_notmember++;
454 			ip6stat.ip6s_cantforward++;
455 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
456 			goto bad;
457 		}
458 		deliverifp = m->m_pkthdr.rcvif;
459 		goto hbhcheck;
460 	}
461 
462 	/*
463 	 *  Unicast check
464 	 */
465 	if (ip6_forward_rt.ro_rt != NULL &&
466 	    (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
467 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
468 			       &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
469 		ip6stat.ip6s_forward_cachehit++;
470 	else {
471 		struct sockaddr_in6 *dst6;
472 
473 		if (ip6_forward_rt.ro_rt) {
474 			/* route is down or destination is different */
475 			ip6stat.ip6s_forward_cachemiss++;
476 			RTFREE(ip6_forward_rt.ro_rt);
477 			ip6_forward_rt.ro_rt = 0;
478 		}
479 
480 		bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
481 		dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
482 		dst6->sin6_len = sizeof(struct sockaddr_in6);
483 		dst6->sin6_family = AF_INET6;
484 		dst6->sin6_addr = ip6->ip6_dst;
485 
486 		rtalloc((struct route *)&ip6_forward_rt);
487 	}
488 
489 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
490 
491 	/*
492 	 * Accept the packet if the forwarding interface to the destination
493 	 * according to the routing table is the loopback interface,
494 	 * unless the associated route has a gateway.
495 	 * Note that this approach causes to accept a packet if there is a
496 	 * route to the loopback interface for the destination of the packet.
497 	 * But we think it's even useful in some situations, e.g. when using
498 	 * a special daemon which wants to intercept the packet.
499 	 */
500 	if (ip6_forward_rt.ro_rt &&
501 	    (ip6_forward_rt.ro_rt->rt_flags &
502 	     (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
503 #if 0
504 	    /*
505 	     * The check below is redundant since the comparison of
506 	     * the destination and the key of the rtentry has
507 	     * already done through looking up the routing table.
508 	     */
509 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
510 			       &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) &&
511 #endif
512 	    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
513 		struct in6_ifaddr *ia6 =
514 			(struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
515 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
516 			m->m_flags |= M_ANYCAST6;
517 		/*
518 		 * packets to a tentative, duplicated, or somehow invalid
519 		 * address must not be accepted.
520 		 */
521 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
522 			/* this address is ready */
523 			ours = 1;
524 			deliverifp = ia6->ia_ifp;	/* correct? */
525 			goto hbhcheck;
526 		} else {
527 			/* address is not ready, so discard the packet. */
528 			nd6log((LOG_INFO,
529 			    "ip6_input: packet to an unready address %s->%s\n",
530 			    ip6_sprintf(&ip6->ip6_src),
531 			    ip6_sprintf(&ip6->ip6_dst)));
532 
533 			goto bad;
534 		}
535 	}
536 
537 	/*
538 	 * FAITH(Firewall Aided Internet Translator)
539 	 */
540 #if defined(NFAITH) && 0 < NFAITH
541 	if (ip6_keepfaith) {
542 		if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
543 		 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
544 			/* XXX do we need more sanity checks? */
545 			ours = 1;
546 			deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /*faith*/
547 			goto hbhcheck;
548 		}
549 	}
550 #endif
551 
552 #if 0
553     {
554 	/*
555 	 * Last resort: check in6_ifaddr for incoming interface.
556 	 * The code is here until I update the "goto ours hack" code above
557 	 * working right.
558 	 */
559 	struct ifaddr *ifa;
560 	for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
561 	     ifa;
562 	     ifa = ifa->ifa_list.tqe_next) {
563 		if (ifa->ifa_addr == NULL)
564 			continue;	/* just for safety */
565 		if (ifa->ifa_addr->sa_family != AF_INET6)
566 			continue;
567 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
568 			ours = 1;
569 			deliverifp = ifa->ifa_ifp;
570 			goto hbhcheck;
571 		}
572 	}
573     }
574 #endif
575 
576 	/*
577 	 * Now there is no reason to process the packet if it's not our own
578 	 * and we're not a router.
579 	 */
580 	if (!ip6_forwarding) {
581 		ip6stat.ip6s_cantforward++;
582 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
583 		goto bad;
584 	}
585 
586   hbhcheck:
587 	/*
588 	 * Process Hop-by-Hop options header if it's contained.
589 	 * m may be modified in ip6_hopopts_input().
590 	 * If a JumboPayload option is included, plen will also be modified.
591 	 */
592 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
593 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
594 		struct ip6_hbh *hbh;
595 
596 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
597 #if 0	/*touches NULL pointer*/
598 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
599 #endif
600 			return;	/* m have already been freed */
601 		}
602 
603 		/* adjust pointer */
604 		ip6 = mtod(m, struct ip6_hdr *);
605 
606 		/*
607 		 * if the payload length field is 0 and the next header field
608 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
609 		 * option MUST be included.
610 		 */
611 		if (ip6->ip6_plen == 0 && plen == 0) {
612 			/*
613 			 * Note that if a valid jumbo payload option is
614 			 * contained, ip6_hoptops_input() must set a valid
615 			 * (non-zero) payload length to the variable plen.
616 			 */
617 			ip6stat.ip6s_badoptions++;
618 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
619 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
620 			icmp6_error(m, ICMP6_PARAM_PROB,
621 				    ICMP6_PARAMPROB_HEADER,
622 				    (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
623 			return;
624 		}
625 #ifndef PULLDOWN_TEST
626 		/* ip6_hopopts_input() ensures that mbuf is contiguous */
627 		hbh = (struct ip6_hbh *)(ip6 + 1);
628 #else
629 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
630 			sizeof(struct ip6_hbh));
631 		if (hbh == NULL) {
632 			ip6stat.ip6s_tooshort++;
633 			return;
634 		}
635 #endif
636 		nxt = hbh->ip6h_nxt;
637 
638 		/*
639 		 * accept the packet if a router alert option is included
640 		 * and we act as an IPv6 router.
641 		 */
642 		if (rtalert != ~0 && ip6_forwarding)
643 			ours = 1;
644 	} else
645 		nxt = ip6->ip6_nxt;
646 
647 	/*
648 	 * Check that the amount of data in the buffers
649 	 * is as at least much as the IPv6 header would have us expect.
650 	 * Trim mbufs if longer than we expect.
651 	 * Drop packet if shorter than we expect.
652 	 */
653 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
654 		ip6stat.ip6s_tooshort++;
655 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
656 		goto bad;
657 	}
658 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
659 		if (m->m_len == m->m_pkthdr.len) {
660 			m->m_len = sizeof(struct ip6_hdr) + plen;
661 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
662 		} else
663 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
664 	}
665 
666 	/*
667 	 * Forward if desirable.
668 	 */
669 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
670 		/*
671 		 * If we are acting as a multicast router, all
672 		 * incoming multicast packets are passed to the
673 		 * kernel-level multicast forwarding function.
674 		 * The packet is returned (relatively) intact; if
675 		 * ip6_mforward() returns a non-zero value, the packet
676 		 * must be discarded, else it may be accepted below.
677 		 */
678 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
679 			ip6stat.ip6s_cantforward++;
680 			m_freem(m);
681 			return;
682 		}
683 		if (!ours) {
684 			m_freem(m);
685 			return;
686 		}
687 	} else if (!ours) {
688 		ip6_forward(m, 0);
689 		return;
690 	}
691 
692 	ip6 = mtod(m, struct ip6_hdr *);
693 
694 	/*
695 	 * Malicious party may be able to use IPv4 mapped addr to confuse
696 	 * tcp/udp stack and bypass security checks (act as if it was from
697 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
698 	 *
699 	 * For SIIT end node behavior, you may want to disable the check.
700 	 * However, you will  become vulnerable to attacks using IPv4 mapped
701 	 * source.
702 	 */
703 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
704 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
705 		ip6stat.ip6s_badscope++;
706 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
707 		goto bad;
708 	}
709 
710 	/*
711 	 * Tell launch routine the next header
712 	 */
713 #ifdef IFA_STATS
714 	if (deliverifp != NULL) {
715 		struct in6_ifaddr *ia6;
716 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
717 		if (ia6)
718 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
719 	}
720 #endif
721 	ip6stat.ip6s_delivered++;
722 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
723 	nest = 0;
724 
725 	while (nxt != IPPROTO_DONE) {
726 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
727 			ip6stat.ip6s_toomanyhdr++;
728 			goto bad;
729 		}
730 
731 		/*
732 		 * protection against faulty packet - there should be
733 		 * more sanity checks in header chain processing.
734 		 */
735 		if (m->m_pkthdr.len < off) {
736 			ip6stat.ip6s_tooshort++;
737 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
738 			goto bad;
739 		}
740 
741 #ifdef IPSEC
742 		/*
743 		 * enforce IPsec policy checking if we are seeing last header.
744 		 * note that we do not visit this with protocols with pcb layer
745 		 * code - like udp/tcp/raw ip.
746 		 */
747 		if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
748 		    ipsec6_in_reject(m, NULL)) {
749 			ipsec6stat.in_polvio++;
750 			goto bad;
751 		}
752 #endif
753 
754 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
755 	}
756 	return;
757  bad:
758 	m_freem(m);
759 }
760 
761 /*
762  * Hop-by-Hop options header processing. If a valid jumbo payload option is
763  * included, the real payload length will be stored in plenp.
764  */
765 static int
766 ip6_hopopts_input(plenp, rtalertp, mp, offp)
767 	u_int32_t *plenp;
768 	u_int32_t *rtalertp;	/* XXX: should be stored more smart way */
769 	struct mbuf **mp;
770 	int *offp;
771 {
772 	struct mbuf *m = *mp;
773 	int off = *offp, hbhlen;
774 	struct ip6_hbh *hbh;
775 	u_int8_t *opt;
776 
777 	/* validation of the length of the header */
778 #ifndef PULLDOWN_TEST
779 	IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
780 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
781 	hbhlen = (hbh->ip6h_len + 1) << 3;
782 
783 	IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
784 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
785 #else
786 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
787 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
788 	if (hbh == NULL) {
789 		ip6stat.ip6s_tooshort++;
790 		return -1;
791 	}
792 	hbhlen = (hbh->ip6h_len + 1) << 3;
793 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
794 		hbhlen);
795 	if (hbh == NULL) {
796 		ip6stat.ip6s_tooshort++;
797 		return -1;
798 	}
799 #endif
800 	off += hbhlen;
801 	hbhlen -= sizeof(struct ip6_hbh);
802 	opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
803 
804 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
805 				hbhlen, rtalertp, plenp) < 0)
806 		return(-1);
807 
808 	*offp = off;
809 	*mp = m;
810 	return(0);
811 }
812 
813 /*
814  * Search header for all Hop-by-hop options and process each option.
815  * This function is separate from ip6_hopopts_input() in order to
816  * handle a case where the sending node itself process its hop-by-hop
817  * options header. In such a case, the function is called from ip6_output().
818  */
819 int
820 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
821 	struct mbuf *m;
822 	u_int8_t *opthead;
823 	int hbhlen;
824 	u_int32_t *rtalertp;
825 	u_int32_t *plenp;
826 {
827 	struct ip6_hdr *ip6;
828 	int optlen = 0;
829 	u_int8_t *opt = opthead;
830 	u_int16_t rtalert_val;
831 	u_int32_t jumboplen;
832 
833 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
834 		switch (*opt) {
835 		case IP6OPT_PAD1:
836 			optlen = 1;
837 			break;
838 		case IP6OPT_PADN:
839 			if (hbhlen < IP6OPT_MINLEN) {
840 				ip6stat.ip6s_toosmall++;
841 				goto bad;
842 			}
843 			optlen = *(opt + 1) + 2;
844 			break;
845 		case IP6OPT_RTALERT:
846 			/* XXX may need check for alignment */
847 			if (hbhlen < IP6OPT_RTALERT_LEN) {
848 				ip6stat.ip6s_toosmall++;
849 				goto bad;
850 			}
851 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
852 				 /* XXX: should we discard the packet? */
853 				log(LOG_ERR, "length of router alert opt is inconsitent(%d)",
854 				    *(opt + 1));
855 			}
856 			optlen = IP6OPT_RTALERT_LEN;
857 			bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
858 			*rtalertp = ntohs(rtalert_val);
859 			break;
860 		case IP6OPT_JUMBO:
861 			/* XXX may need check for alignment */
862 			if (hbhlen < IP6OPT_JUMBO_LEN) {
863 				ip6stat.ip6s_toosmall++;
864 				goto bad;
865 			}
866 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
867 				/* XXX: should we discard the packet? */
868 				log(LOG_ERR, "length of jumbopayload opt "
869 				    "is inconsistent(%d)\n",
870 				    *(opt + 1));
871 			}
872 			optlen = IP6OPT_JUMBO_LEN;
873 
874 			/*
875 			 * IPv6 packets that have non 0 payload length
876 			 * must not contain a jumbo payload option.
877 			 */
878 			ip6 = mtod(m, struct ip6_hdr *);
879 			if (ip6->ip6_plen) {
880 				ip6stat.ip6s_badoptions++;
881 				icmp6_error(m, ICMP6_PARAM_PROB,
882 					    ICMP6_PARAMPROB_HEADER,
883 					    sizeof(struct ip6_hdr) +
884 					    sizeof(struct ip6_hbh) +
885 					    opt - opthead);
886 				return(-1);
887 			}
888 
889 			/*
890 			 * We may see jumbolen in unaligned location, so
891 			 * we'd need to perform bcopy().
892 			 */
893 			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
894 			jumboplen = (u_int32_t)htonl(jumboplen);
895 
896 #if 1
897 			/*
898 			 * if there are multiple jumbo payload options,
899 			 * *plenp will be non-zero and the packet will be
900 			 * rejected.
901 			 * the behavior may need some debate in ipngwg -
902 			 * multiple options does not make sense, however,
903 			 * there's no explicit mention in specification.
904 			 */
905 			if (*plenp != 0) {
906 				ip6stat.ip6s_badoptions++;
907 				icmp6_error(m, ICMP6_PARAM_PROB,
908 					    ICMP6_PARAMPROB_HEADER,
909 					    sizeof(struct ip6_hdr) +
910 					    sizeof(struct ip6_hbh) +
911 					    opt + 2 - opthead);
912 				return(-1);
913 			}
914 #endif
915 
916 			/*
917 			 * jumbo payload length must be larger than 65535.
918 			 */
919 			if (jumboplen <= IPV6_MAXPACKET) {
920 				ip6stat.ip6s_badoptions++;
921 				icmp6_error(m, ICMP6_PARAM_PROB,
922 					    ICMP6_PARAMPROB_HEADER,
923 					    sizeof(struct ip6_hdr) +
924 					    sizeof(struct ip6_hbh) +
925 					    opt + 2 - opthead);
926 				return(-1);
927 			}
928 			*plenp = jumboplen;
929 
930 			break;
931 		default:		/* unknown option */
932 			if (hbhlen < IP6OPT_MINLEN) {
933 				ip6stat.ip6s_toosmall++;
934 				goto bad;
935 			}
936 			if ((optlen = ip6_unknown_opt(opt, m,
937 						      sizeof(struct ip6_hdr) +
938 						      sizeof(struct ip6_hbh) +
939 						      opt - opthead)) == -1)
940 				return(-1);
941 			optlen += 2;
942 			break;
943 		}
944 	}
945 
946 	return(0);
947 
948   bad:
949 	m_freem(m);
950 	return(-1);
951 }
952 
953 /*
954  * Unknown option processing.
955  * The third argument `off' is the offset from the IPv6 header to the option,
956  * which is necessary if the IPv6 header the and option header and IPv6 header
957  * is not continuous in order to return an ICMPv6 error.
958  */
959 int
960 ip6_unknown_opt(optp, m, off)
961 	u_int8_t *optp;
962 	struct mbuf *m;
963 	int off;
964 {
965 	struct ip6_hdr *ip6;
966 
967 	switch (IP6OPT_TYPE(*optp)) {
968 	case IP6OPT_TYPE_SKIP: /* ignore the option */
969 		return((int)*(optp + 1));
970 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
971 		m_freem(m);
972 		return(-1);
973 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
974 		ip6stat.ip6s_badoptions++;
975 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
976 		return(-1);
977 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
978 		ip6stat.ip6s_badoptions++;
979 		ip6 = mtod(m, struct ip6_hdr *);
980 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
981 		    (m->m_flags & (M_BCAST|M_MCAST)))
982 			m_freem(m);
983 		else
984 			icmp6_error(m, ICMP6_PARAM_PROB,
985 				    ICMP6_PARAMPROB_OPTION, off);
986 		return(-1);
987 	}
988 
989 	m_freem(m);		/* XXX: NOTREACHED */
990 	return(-1);
991 }
992 
993 /*
994  * Create the "control" list for this pcb.
995  *
996  * The routine will be called from upper layer handlers like tcp6_input().
997  * Thus the routine assumes that the caller (tcp6_input) have already
998  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
999  * very first mbuf on the mbuf chain.
1000  * We may want to add some infinite loop prevention or sanity checks for safety.
1001  * (This applies only when you are using KAME mbuf chain restriction, i.e.
1002  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
1003  */
1004 void
1005 ip6_savecontrol(in6p, mp, ip6, m)
1006 	struct in6pcb *in6p;
1007 	struct mbuf **mp;
1008 	struct ip6_hdr *ip6;
1009 	struct mbuf *m;
1010 {
1011 	struct proc *p = curproc;	/* XXX */
1012 	int privileged;
1013 
1014 	privileged = 0;
1015 	if (p && !suser(p->p_ucred, &p->p_acflag))
1016 		privileged++;
1017 
1018 #ifdef SO_TIMESTAMP
1019 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP) {
1020 		struct timeval tv;
1021 
1022 		microtime(&tv);
1023 		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1024 			SCM_TIMESTAMP, SOL_SOCKET);
1025 		if (*mp)
1026 			mp = &(*mp)->m_next;
1027 	}
1028 #endif
1029 	if (in6p->in6p_flags & IN6P_RECVDSTADDR) {
1030 		*mp = sbcreatecontrol((caddr_t) &ip6->ip6_dst,
1031 			sizeof(struct in6_addr), IPV6_RECVDSTADDR,
1032 			IPPROTO_IPV6);
1033 		if (*mp)
1034 			mp = &(*mp)->m_next;
1035 	}
1036 
1037 #ifdef noyet
1038 	/* options were tossed above */
1039 	if (in6p->in6p_flags & IN6P_RECVOPTS)
1040 		/* broken */
1041 	/* ip6_srcroute doesn't do what we want here, need to fix */
1042 	if (in6p->in6p_flags & IPV6P_RECVRETOPTS)
1043 		/* broken */
1044 #endif
1045 
1046 	/* RFC 2292 sec. 5 */
1047 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1048 		struct in6_pktinfo pi6;
1049 		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1050 		if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
1051 			pi6.ipi6_addr.s6_addr16[1] = 0;
1052 		pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
1053 					? m->m_pkthdr.rcvif->if_index
1054 					: 0;
1055 		*mp = sbcreatecontrol((caddr_t) &pi6,
1056 			sizeof(struct in6_pktinfo), IPV6_PKTINFO,
1057 			IPPROTO_IPV6);
1058 		if (*mp)
1059 			mp = &(*mp)->m_next;
1060 	}
1061 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1062 		int hlim = ip6->ip6_hlim & 0xff;
1063 		*mp = sbcreatecontrol((caddr_t) &hlim,
1064 			sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6);
1065 		if (*mp)
1066 			mp = &(*mp)->m_next;
1067 	}
1068 	/* IN6P_NEXTHOP - for outgoing packet only */
1069 
1070 	/*
1071 	 * IPV6_HOPOPTS socket option. We require super-user privilege
1072 	 * for the option, but it might be too strict, since there might
1073 	 * be some hop-by-hop options which can be returned to normal user.
1074 	 * See RFC 2292 section 6.
1075 	 */
1076 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) {
1077 		/*
1078 		 * Check if a hop-by-hop options header is contatined in the
1079 		 * received packet, and if so, store the options as ancillary
1080 		 * data. Note that a hop-by-hop options header must be
1081 		 * just after the IPv6 header, which fact is assured through
1082 		 * the IPv6 input processing.
1083 		 */
1084 		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1085 		if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1086 			struct ip6_hbh *hbh;
1087 			int hbhlen;
1088 
1089 #ifndef PULLDOWN_TEST
1090 			hbh = (struct ip6_hbh *)(ip6 + 1);
1091 			hbhlen = (hbh->ip6h_len + 1) << 3;
1092 #else
1093 			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1094 				sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
1095 			if (hbh == NULL) {
1096 				ip6stat.ip6s_tooshort++;
1097 				return;
1098 			}
1099 			hbhlen = (hbh->ip6h_len + 1) << 3;
1100 			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1101 				sizeof(struct ip6_hdr), hbhlen);
1102 			if (hbh == NULL) {
1103 				ip6stat.ip6s_tooshort++;
1104 				return;
1105 			}
1106 #endif
1107 
1108 			/*
1109 			 * XXX: We copy whole the header even if a jumbo
1110 			 * payload option is included, which option is to
1111 			 * be removed before returning in the RFC 2292.
1112 			 * But it's too painful operation...
1113 			 */
1114 			*mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1115 					      IPV6_HOPOPTS, IPPROTO_IPV6);
1116 			if (*mp)
1117 				mp = &(*mp)->m_next;
1118 		}
1119 	}
1120 
1121 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1122 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1123 		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1124 		int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);;
1125 
1126 		/*
1127 		 * Search for destination options headers or routing
1128 		 * header(s) through the header chain, and stores each
1129 		 * header as ancillary data.
1130 		 * Note that the order of the headers remains in
1131 		 * the chain of ancillary data.
1132 		 */
1133 		while (1) {	/* is explicit loop prevention necessary? */
1134 			struct ip6_ext *ip6e;
1135 			int elen;
1136 
1137 #ifndef PULLDOWN_TEST
1138 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1139 			if (nxt == IPPROTO_AH)
1140 				elen = (ip6e->ip6e_len + 2) << 2;
1141 			else
1142 				elen = (ip6e->ip6e_len + 1) << 3;
1143 #else
1144 			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off,
1145 				sizeof(struct ip6_ext));
1146 			if (ip6e == NULL) {
1147 				ip6stat.ip6s_tooshort++;
1148 				return;
1149 			}
1150 			if (nxt == IPPROTO_AH)
1151 				elen = (ip6e->ip6e_len + 2) << 2;
1152 			else
1153 				elen = (ip6e->ip6e_len + 1) << 3;
1154 			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off, elen);
1155 			if (ip6e == NULL) {
1156 				ip6stat.ip6s_tooshort++;
1157 				return;
1158 			}
1159 #endif
1160 
1161 			switch (nxt) {
1162 		        case IPPROTO_DSTOPTS:
1163 				if (!in6p->in6p_flags & IN6P_DSTOPTS)
1164 					break;
1165 
1166 				/*
1167 				 * We also require super-user privilege for
1168 				 * the option.
1169 				 * See the comments on IN6_HOPOPTS.
1170 				 */
1171 				if (!privileged)
1172 					break;
1173 
1174 				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
1175 						      IPV6_DSTOPTS,
1176 						      IPPROTO_IPV6);
1177 				if (*mp)
1178 					mp = &(*mp)->m_next;
1179 				break;
1180 
1181 			case IPPROTO_ROUTING:
1182 				if (!in6p->in6p_flags & IN6P_RTHDR)
1183 					break;
1184 
1185 				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
1186 						      IPV6_RTHDR,
1187 						      IPPROTO_IPV6);
1188 				if (*mp)
1189 					mp = &(*mp)->m_next;
1190 				break;
1191 
1192 			case IPPROTO_UDP:
1193 			case IPPROTO_TCP:
1194 			case IPPROTO_ICMPV6:
1195 			default:
1196 				/*
1197 				 * stop search if we encounter an upper
1198 				 * layer protocol headers.
1199 				 */
1200 				goto loopend;
1201 
1202 			case IPPROTO_HOPOPTS:
1203 			case IPPROTO_AH: /* is it possible? */
1204 				break;
1205 			}
1206 
1207 			/* proceed with the next header. */
1208 			off += elen;
1209 			nxt = ip6e->ip6e_nxt;
1210 		}
1211 	  loopend:
1212 		;
1213 	}
1214 	if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) {
1215 		/* to be done */
1216 	}
1217 	if ((in6p->in6p_flags & IN6P_DSTOPTS) && privileged) {
1218 		/* to be done */
1219 	}
1220 	/* IN6P_RTHDR - to be done */
1221 
1222 }
1223 
1224 /*
1225  * Get pointer to the previous header followed by the header
1226  * currently processed.
1227  * XXX: This function supposes that
1228  *	M includes all headers,
1229  *	the next header field and the header length field of each header
1230  *	are valid, and
1231  *	the sum of each header length equals to OFF.
1232  * Because of these assumptions, this function must be called very
1233  * carefully. Moreover, it will not be used in the near future when
1234  * we develop `neater' mechanism to process extension headers.
1235  */
1236 char *
1237 ip6_get_prevhdr(m, off)
1238 	struct mbuf *m;
1239 	int off;
1240 {
1241 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1242 
1243 	if (off == sizeof(struct ip6_hdr))
1244 		return(&ip6->ip6_nxt);
1245 	else {
1246 		int len, nxt;
1247 		struct ip6_ext *ip6e = NULL;
1248 
1249 		nxt = ip6->ip6_nxt;
1250 		len = sizeof(struct ip6_hdr);
1251 		while (len < off) {
1252 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1253 
1254 			switch (nxt) {
1255 			case IPPROTO_FRAGMENT:
1256 				len += sizeof(struct ip6_frag);
1257 				break;
1258 			case IPPROTO_AH:
1259 				len += (ip6e->ip6e_len + 2) << 2;
1260 				break;
1261 			default:
1262 				len += (ip6e->ip6e_len + 1) << 3;
1263 				break;
1264 			}
1265 			nxt = ip6e->ip6e_nxt;
1266 		}
1267 		if (ip6e)
1268 			return(&ip6e->ip6e_nxt);
1269 		else
1270 			return NULL;
1271 	}
1272 }
1273 
1274 /*
1275  * get next header offset.  m will be retained.
1276  */
1277 int
1278 ip6_nexthdr(m, off, proto, nxtp)
1279 	struct mbuf *m;
1280 	int off;
1281 	int proto;
1282 	int *nxtp;
1283 {
1284 	struct ip6_hdr ip6;
1285 	struct ip6_ext ip6e;
1286 	struct ip6_frag fh;
1287 
1288 	/* just in case */
1289 	if (m == NULL)
1290 		panic("ip6_nexthdr: m == NULL");
1291 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1292 		return -1;
1293 
1294 	switch (proto) {
1295 	case IPPROTO_IPV6:
1296 		if (m->m_pkthdr.len < off + sizeof(ip6))
1297 			return -1;
1298 		m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1299 		if (nxtp)
1300 			*nxtp = ip6.ip6_nxt;
1301 		off += sizeof(ip6);
1302 		return off;
1303 
1304 	case IPPROTO_FRAGMENT:
1305 		/*
1306 		 * terminate parsing if it is not the first fragment,
1307 		 * it does not make sense to parse through it.
1308 		 */
1309 		if (m->m_pkthdr.len < off + sizeof(fh))
1310 			return -1;
1311 		m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1312 		if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0)
1313 			return -1;
1314 		if (nxtp)
1315 			*nxtp = fh.ip6f_nxt;
1316 		off += sizeof(struct ip6_frag);
1317 		return off;
1318 
1319 	case IPPROTO_AH:
1320 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1321 			return -1;
1322 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1323 		if (nxtp)
1324 			*nxtp = ip6e.ip6e_nxt;
1325 		off += (ip6e.ip6e_len + 2) << 2;
1326 		return off;
1327 
1328 	case IPPROTO_HOPOPTS:
1329 	case IPPROTO_ROUTING:
1330 	case IPPROTO_DSTOPTS:
1331 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1332 			return -1;
1333 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1334 		if (nxtp)
1335 			*nxtp = ip6e.ip6e_nxt;
1336 		off += (ip6e.ip6e_len + 1) << 3;
1337 		return off;
1338 
1339 	case IPPROTO_NONE:
1340 	case IPPROTO_ESP:
1341 	case IPPROTO_IPCOMP:
1342 		/* give up */
1343 		return -1;
1344 
1345 	default:
1346 		return -1;
1347 	}
1348 
1349 	return -1;
1350 }
1351 
1352 /*
1353  * get offset for the last header in the chain.  m will be kept untainted.
1354  */
1355 int
1356 ip6_lasthdr(m, off, proto, nxtp)
1357 	struct mbuf *m;
1358 	int off;
1359 	int proto;
1360 	int *nxtp;
1361 {
1362 	int newoff;
1363 	int nxt;
1364 
1365 	if (!nxtp) {
1366 		nxt = -1;
1367 		nxtp = &nxt;
1368 	}
1369 	while (1) {
1370 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1371 		if (newoff < 0)
1372 			return off;
1373 		else if (newoff < off)
1374 			return -1;	/* invalid */
1375 		else if (newoff == off)
1376 			return newoff;
1377 
1378 		off = newoff;
1379 		proto = *nxtp;
1380 	}
1381 }
1382 
1383 /*
1384  * System control for IP6
1385  */
1386 
1387 u_char	inet6ctlerrmap[PRC_NCMDS] = {
1388 	0,		0,		0,		0,
1389 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1390 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1391 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1392 	0,		0,		0,		0,
1393 	ENOPROTOOPT
1394 };
1395 
1396 #include <uvm/uvm_extern.h>
1397 #include <sys/sysctl.h>
1398 
1399 int
1400 ip6_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1401 	int *name;
1402 	u_int namelen;
1403 	void *oldp;
1404 	size_t *oldlenp;
1405 	void *newp;
1406 	size_t newlen;
1407 {
1408 	int old, error;
1409 
1410 	/* All sysctl names at this level are terminal. */
1411 	if (namelen != 1)
1412 		return ENOTDIR;
1413 
1414 	switch (name[0]) {
1415 
1416 	case IPV6CTL_FORWARDING:
1417 		return sysctl_int(oldp, oldlenp, newp, newlen,
1418 				  &ip6_forwarding);
1419 	case IPV6CTL_SENDREDIRECTS:
1420 		return sysctl_int(oldp, oldlenp, newp, newlen,
1421 				&ip6_sendredirects);
1422 	case IPV6CTL_DEFHLIM:
1423 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_defhlim);
1424 	case IPV6CTL_MAXFRAGPACKETS:
1425 		return sysctl_int(oldp, oldlenp, newp, newlen,
1426 				&ip6_maxfragpackets);
1427 	case IPV6CTL_ACCEPT_RTADV:
1428 		return sysctl_int(oldp, oldlenp, newp, newlen,
1429 				&ip6_accept_rtadv);
1430 	case IPV6CTL_KEEPFAITH:
1431 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_keepfaith);
1432 	case IPV6CTL_LOG_INTERVAL:
1433 		return sysctl_int(oldp, oldlenp, newp, newlen,
1434 				&ip6_log_interval);
1435 	case IPV6CTL_HDRNESTLIMIT:
1436 		return sysctl_int(oldp, oldlenp, newp, newlen,
1437 				&ip6_hdrnestlimit);
1438 	case IPV6CTL_DAD_COUNT:
1439 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_dad_count);
1440 	case IPV6CTL_AUTO_FLOWLABEL:
1441 		return sysctl_int(oldp, oldlenp, newp, newlen,
1442 				&ip6_auto_flowlabel);
1443 	case IPV6CTL_DEFMCASTHLIM:
1444 		return sysctl_int(oldp, oldlenp, newp, newlen,
1445 				&ip6_defmcasthlim);
1446 	case IPV6CTL_GIF_HLIM:
1447 		return sysctl_int(oldp, oldlenp, newp, newlen,
1448 				&ip6_gif_hlim);
1449 	case IPV6CTL_KAME_VERSION:
1450 		return sysctl_rdstring(oldp, oldlenp, newp, __KAME_VERSION);
1451 	case IPV6CTL_USE_DEPRECATED:
1452 		return sysctl_int(oldp, oldlenp, newp, newlen,
1453 				&ip6_use_deprecated);
1454 	case IPV6CTL_RR_PRUNE:
1455 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_rr_prune);
1456 #ifndef INET6_BINDV6ONLY
1457 	case IPV6CTL_BINDV6ONLY:
1458 		return sysctl_int(oldp, oldlenp, newp, newlen,
1459 				&ip6_bindv6only);
1460 #endif
1461 	case IPV6CTL_ANONPORTMIN:
1462 		old = ip6_anonportmin;
1463 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1464 		    &ip6_anonportmin);
1465 		if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmin < 0 ||
1466 		    ip6_anonportmin > 65535
1467 #ifndef IPNOPRIVPORTS
1468 		    || ip6_anonportmin < IPV6PORT_RESERVED
1469 #endif
1470 		    ) {
1471 			ip6_anonportmin = old;
1472 			return (EINVAL);
1473 		}
1474 		return (error);
1475 	case IPV6CTL_ANONPORTMAX:
1476 		old = ip6_anonportmax;
1477 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1478 		    &ip6_anonportmax);
1479 		if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmax < 0 ||
1480 		    ip6_anonportmax > 65535
1481 #ifndef IPNOPRIVPORTS
1482 		    || ip6_anonportmax < IPV6PORT_RESERVED
1483 #endif
1484 		    ) {
1485 			ip6_anonportmax = old;
1486 			return (EINVAL);
1487 		}
1488 		return (error);
1489 #ifndef IPNOPRIVPORTS
1490 	case IPV6CTL_LOWPORTMIN:
1491 		old = ip6_lowportmin;
1492 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1493 		    &ip6_lowportmin);
1494 		if (ip6_lowportmin >= ip6_lowportmax ||
1495 		    ip6_lowportmin > IPV6PORT_RESERVEDMAX ||
1496 		    ip6_lowportmin < IPV6PORT_RESERVEDMIN) {
1497 			ip6_lowportmin = old;
1498 			return (EINVAL);
1499 		}
1500 		return (error);
1501 	case IPV6CTL_LOWPORTMAX:
1502 		old = ip6_lowportmax;
1503 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1504 		    &ip6_lowportmax);
1505 		if (ip6_lowportmin >= ip6_lowportmax ||
1506 		    ip6_lowportmax > IPV6PORT_RESERVEDMAX ||
1507 		    ip6_lowportmax < IPV6PORT_RESERVEDMIN) {
1508 			ip6_lowportmax = old;
1509 			return (EINVAL);
1510 		}
1511 		return (error);
1512 #endif
1513 	default:
1514 		return EOPNOTSUPP;
1515 	}
1516 	/* NOTREACHED */
1517 }
1518