xref: /netbsd-src/sys/netinet6/ip6_input.c (revision 481fca6e59249d8ffcf24fef7cfbe7b131bfb080)
1 /*	$NetBSD: ip6_input.c,v 1.25 2000/07/06 12:36:19 itojun Exp $	*/
2 /*	$KAME: ip6_input.c,v 1.95 2000/07/02 07:49:37 jinmei 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 IPV6FIREWALL
111 #include <netinet6/ip6_fw.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 IPV6FIREWALL
138 /* firewall hooks */
139 ip6_fw_chk_t *ip6_fw_chk_ptr;
140 ip6_fw_ctl_t *ip6_fw_ctl_ptr;
141 #endif
142 
143 struct ip6stat ip6stat;
144 
145 static void ip6_init2 __P((void *));
146 
147 static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
148 
149 /*
150  * IP6 initialization: fill in IP6 protocol switch table.
151  * All protocols not implemented in kernel go to raw IP6 protocol handler.
152  */
153 void
154 ip6_init()
155 {
156 	register struct ip6protosw *pr;
157 	register int i;
158 	struct timeval tv;
159 
160 	pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
161 	if (pr == 0)
162 		panic("ip6_init");
163 	for (i = 0; i < IPPROTO_MAX; i++)
164 		ip6_protox[i] = pr - inet6sw;
165 	for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
166 	    pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
167 		if (pr->pr_domain->dom_family == PF_INET6 &&
168 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
169 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
170 	ip6intrq.ifq_maxlen = ip6qmaxlen;
171 	nd6_init();
172 	frag6_init();
173 #ifdef IPV6FIREWALL
174 	ip6_fw_init();
175 #endif
176 	/*
177 	 * in many cases, random() here does NOT return random number
178 	 * as initialization during bootstrap time occur in fixed order.
179 	 */
180 	microtime(&tv);
181 	ip6_flow_seq = random() ^ tv.tv_usec;
182 
183 	ip6_init2((void *)0);
184 }
185 
186 static void
187 ip6_init2(dummy)
188 	void *dummy;
189 {
190 	/*
191 	 * to route local address of p2p link to loopback,
192 	 * assign loopback address first.
193 	 */
194 	in6_ifattach(&loif[0], NULL);
195 
196 	/* nd6_timer_init */
197 	callout_init(&nd6_timer_ch);
198 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
199 	/* router renumbering prefix list maintenance */
200 	callout_init(&in6_rr_timer_ch);
201 	callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
202 }
203 
204 /*
205  * IP6 input interrupt handling. Just pass the packet to ip6_input.
206  */
207 void
208 ip6intr()
209 {
210 	int s;
211 	struct mbuf *m;
212 
213 	for (;;) {
214 		s = splimp();
215 		IF_DEQUEUE(&ip6intrq, m);
216 		splx(s);
217 		if (m == 0)
218 			return;
219 		ip6_input(m);
220 	}
221 }
222 
223 extern struct	route_in6 ip6_forward_rt;
224 
225 void
226 ip6_input(m)
227 	struct mbuf *m;
228 {
229 	struct ip6_hdr *ip6;
230 	int off = sizeof(struct ip6_hdr), nest;
231 	u_int32_t plen;
232 	u_int32_t rtalert = ~0;
233 	int nxt, ours = 0;
234 	struct ifnet *deliverifp = NULL;
235 #ifdef	PFIL_HOOKS
236 	struct packet_filter_hook *pfh;
237 	struct mbuf *m0;
238 	int rv;
239 #endif	/* PFIL_HOOKS */
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 		if (m->m_next) {
262 			if (m->m_flags & M_LOOP) {
263 				ip6stat.ip6s_m2m[loif[0].if_index]++;	/*XXX*/
264 			} else if (m->m_pkthdr.rcvif->if_index <= 31)
265 				ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
266 			else
267 				ip6stat.ip6s_m2m[0]++;
268 		} else
269 			ip6stat.ip6s_m1++;
270 	}
271 
272 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
273 	ip6stat.ip6s_total++;
274 
275 #ifndef PULLDOWN_TEST
276 	/* XXX is the line really necessary? */
277 	IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/);
278 #endif
279 
280 	if (m->m_len < sizeof(struct ip6_hdr)) {
281 		struct ifnet *inifp;
282 		inifp = m->m_pkthdr.rcvif;
283 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
284 			ip6stat.ip6s_toosmall++;
285 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
286 			return;
287 		}
288 	}
289 
290 	ip6 = mtod(m, struct ip6_hdr *);
291 
292 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
293 		ip6stat.ip6s_badvers++;
294 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
295 		goto bad;
296 	}
297 
298 #ifdef PFIL_HOOKS
299 	/*
300 	 * Run through list of hooks for input packets.  If there are any
301 	 * filters which require that additional packets in the flow are
302 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
303 	 * Note that filters must _never_ set this flag, as another filter
304 	 * in the list may have previously cleared it.
305 	 */
306 	m0 = m;
307 	pfh = pfil_hook_get(PFIL_IN, &inetsw[ip_protox[IPPROTO_IPV6]].pr_pfh);
308 	for (; pfh; pfh = pfh->pfil_link.tqe_next)
309 		if (pfh->pfil_func) {
310 			rv = pfh->pfil_func(ip6, sizeof(*ip6),
311 					    m->m_pkthdr.rcvif, 0, &m0);
312 			if (rv)
313 				return;
314 			m = m0;
315 			if (m == NULL)
316 				return;
317 			ip6 = mtod(m, struct ip6_hdr *);
318 		}
319 #endif /* PFIL_HOOKS */
320 
321 
322 	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
323 
324 #ifdef IPV6FIREWALL
325 	/*
326 	 * Check with the firewall...
327 	 */
328 	if (ip6_fw_chk_ptr) {
329 		u_short port = 0;
330 		/* If ipfw says divert, we have to just drop packet */
331 		/* use port as a dummy argument */
332 		if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) {
333 			m_freem(m);
334 			m = NULL;
335 		}
336 		if (!m)
337 			return;
338 	}
339 #endif
340 
341 	/*
342 	 * Scope check
343 	 */
344 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
345 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
346 		ip6stat.ip6s_badscope++;
347 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
348 		goto bad;
349 	}
350 	/*
351 	 * The following check is not documented in the spec.  Malicious party
352 	 * may be able to use IPv4 mapped addr to confuse tcp/udp stack and
353 	 * bypass security checks (act as if it was from 127.0.0.1 by using
354 	 * IPv6 src ::ffff:127.0.0.1).	Be cautious.
355 	 */
356 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
357 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
358 		ip6stat.ip6s_badscope++;
359 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
360 		goto bad;
361 	}
362 #if 0
363 	/*
364 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
365 	 *
366 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
367 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
368 	 * is revised to forbid relaying case.
369 	 */
370 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
371 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
372 		ip6stat.ip6s_badscope++;
373 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
374 		goto bad;
375 	}
376 #endif
377 	if (IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
378 	    IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) {
379 		if (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) {
380 			ours = 1;
381 			deliverifp = m->m_pkthdr.rcvif;
382 			goto hbhcheck;
383 		} else {
384 			ip6stat.ip6s_badscope++;
385 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
386 			goto bad;
387 		}
388 	}
389 
390 #ifndef FAKE_LOOPBACK_IF
391 	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0)
392 #else
393 	if (1)
394 #endif
395 	{
396 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
397 			ip6->ip6_src.s6_addr16[1]
398 				= htons(m->m_pkthdr.rcvif->if_index);
399 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
400 			ip6->ip6_dst.s6_addr16[1]
401 				= htons(m->m_pkthdr.rcvif->if_index);
402 	}
403 
404 	/*
405 	 * XXX we need this since we do not have "goto ours" hack route
406 	 * for some of our ifaddrs on loopback interface.
407 	 * we should correct it by changing in6_ifattach to install
408 	 * "goto ours" hack route.
409 	 */
410 	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0) {
411 		if (IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) {
412 			ours = 1;
413 			deliverifp = m->m_pkthdr.rcvif;
414 			goto hbhcheck;
415 		}
416 	}
417 
418 	/*
419 	 * Multicast check
420 	 */
421 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
422 	  	struct	in6_multi *in6m = 0;
423 
424 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
425 		/*
426 		 * See if we belong to the destination multicast group on the
427 		 * arrival interface.
428 		 */
429 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
430 		if (in6m)
431 			ours = 1;
432 		else if (!ip6_mrouter) {
433 			ip6stat.ip6s_notmember++;
434 			ip6stat.ip6s_cantforward++;
435 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
436 			goto bad;
437 		}
438 		deliverifp = m->m_pkthdr.rcvif;
439 		goto hbhcheck;
440 	}
441 
442 	/*
443 	 *  Unicast check
444 	 */
445 	if (ip6_forward_rt.ro_rt != NULL &&
446 	    (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
447 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
448 			       &ip6_forward_rt.ro_dst.sin6_addr))
449 		ip6stat.ip6s_forward_cachehit++;
450 	else {
451 		if (ip6_forward_rt.ro_rt) {
452 			/* route is down or destination is different */
453 			ip6stat.ip6s_forward_cachemiss++;
454 			RTFREE(ip6_forward_rt.ro_rt);
455 			ip6_forward_rt.ro_rt = 0;
456 		}
457 
458 		bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
459 		ip6_forward_rt.ro_dst.sin6_len = sizeof(struct sockaddr_in6);
460 		ip6_forward_rt.ro_dst.sin6_family = AF_INET6;
461 		ip6_forward_rt.ro_dst.sin6_addr = ip6->ip6_dst;
462 
463 		rtalloc((struct route *)&ip6_forward_rt);
464 	}
465 
466 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
467 
468 	/*
469 	 * Accept the packet if the forwarding interface to the destination
470 	 * according to the routing table is the loopback interface,
471 	 * unless the associated route has a gateway.
472 	 * Note that this approach causes to accept a packet if there is a
473 	 * route to the loopback interface for the destination of the packet.
474 	 * But we think it's even useful in some situations, e.g. when using
475 	 * a special daemon which wants to intercept the packet.
476 	 */
477 	if (ip6_forward_rt.ro_rt &&
478 	    (ip6_forward_rt.ro_rt->rt_flags &
479 	     (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
480 #if 0
481 	    /*
482 	     * The check below is redundant since the comparison of
483 	     * the destination and the key of the rtentry has
484 	     * already done through looking up the routing table.
485 	     */
486 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
487 			       &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) &&
488 #endif
489 	    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
490 		struct in6_ifaddr *ia6 =
491 			(struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
492 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
493 			m->m_flags |= M_ANYCAST6;
494 		/*
495 		 * packets to a tentative, duplicated, or somehow invalid
496 		 * address must not be accepted.
497 		 */
498 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
499 			/* this address is ready */
500 			ours = 1;
501 			deliverifp = ia6->ia_ifp;	/* correct? */
502 			goto hbhcheck;
503 		} else {
504 			/* address is not ready, so discard the packet. */
505 			log(LOG_INFO,
506 			    "ip6_input: packet to an unready address %s->%s",
507 			    ip6_sprintf(&ip6->ip6_src),
508 			    ip6_sprintf(&ip6->ip6_dst));
509 
510 			goto bad;
511 		}
512 	}
513 
514 	/*
515 	 * FAITH(Firewall Aided Internet Translator)
516 	 */
517 #if defined(NFAITH) && 0 < NFAITH
518 	if (ip6_keepfaith) {
519 		if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
520 		 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
521 			/* XXX do we need more sanity checks? */
522 			ours = 1;
523 			deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /*faith*/
524 			goto hbhcheck;
525 		}
526 	}
527 #endif
528 
529 #if 0
530     {
531 	/*
532 	 * Last resort: check in6_ifaddr for incoming interface.
533 	 * The code is here until I update the "goto ours hack" code above
534 	 * working right.
535 	 */
536 	struct ifaddr *ifa;
537 	for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
538 	     ifa;
539 	     ifa = ifa->ifa_list.tqe_next) {
540 		if (ifa->ifa_addr == NULL)
541 			continue;	/* just for safety */
542 		if (ifa->ifa_addr->sa_family != AF_INET6)
543 			continue;
544 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
545 			ours = 1;
546 			deliverifp = ifa->ifa_ifp;
547 			goto hbhcheck;
548 		}
549 	}
550     }
551 #endif
552 
553 	/*
554 	 * Now there is no reason to process the packet if it's not our own
555 	 * and we're not a router.
556 	 */
557 	if (!ip6_forwarding) {
558 		ip6stat.ip6s_cantforward++;
559 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
560 		goto bad;
561 	}
562 
563   hbhcheck:
564 	/*
565 	 * Process Hop-by-Hop options header if it's contained.
566 	 * m may be modified in ip6_hopopts_input().
567 	 * If a JumboPayload option is included, plen will also be modified.
568 	 */
569 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
570 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
571 		struct ip6_hbh *hbh;
572 
573 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
574 #if 0	/*touches NULL pointer*/
575 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
576 #endif
577 			return;	/* m have already been freed */
578 		}
579 
580 		/* adjust pointer */
581 		ip6 = mtod(m, struct ip6_hdr *);
582 
583 		/*
584 		 * if the payload length field is 0 and the next header field
585 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
586 		 * option MUST be included.
587 		 */
588 		if (ip6->ip6_plen == 0 && plen == 0) {
589 			/*
590 			 * Note that if a valid jumbo payload option is
591 			 * contained, ip6_hoptops_input() must set a valid
592 			 * (non-zero) payload length to the variable plen.
593 			 */
594 			ip6stat.ip6s_badoptions++;
595 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
596 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
597 			icmp6_error(m, ICMP6_PARAM_PROB,
598 				    ICMP6_PARAMPROB_HEADER,
599 				    (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
600 			return;
601 		}
602 #ifndef PULLDOWN_TEST
603 		/* ip6_hopopts_input() ensures that mbuf is contiguous */
604 		hbh = (struct ip6_hbh *)(ip6 + 1);
605 #else
606 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
607 			sizeof(struct ip6_hbh));
608 		if (hbh == NULL) {
609 			ip6stat.ip6s_tooshort++;
610 			return;
611 		}
612 #endif
613 		nxt = hbh->ip6h_nxt;
614 
615 		/*
616 		 * accept the packet if a router alert option is included
617 		 * and we act as an IPv6 router.
618 		 */
619 		if (rtalert != ~0 && ip6_forwarding)
620 			ours = 1;
621 	} else
622 		nxt = ip6->ip6_nxt;
623 
624 	/*
625 	 * Check that the amount of data in the buffers
626 	 * is as at least much as the IPv6 header would have us expect.
627 	 * Trim mbufs if longer than we expect.
628 	 * Drop packet if shorter than we expect.
629 	 */
630 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
631 		ip6stat.ip6s_tooshort++;
632 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
633 		goto bad;
634 	}
635 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
636 		if (m->m_len == m->m_pkthdr.len) {
637 			m->m_len = sizeof(struct ip6_hdr) + plen;
638 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
639 		} else
640 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
641 	}
642 
643 	/*
644 	 * Forward if desirable.
645 	 */
646 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
647 		/*
648 		 * If we are acting as a multicast router, all
649 		 * incoming multicast packets are passed to the
650 		 * kernel-level multicast forwarding function.
651 		 * The packet is returned (relatively) intact; if
652 		 * ip6_mforward() returns a non-zero value, the packet
653 		 * must be discarded, else it may be accepted below.
654 		 */
655 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
656 			ip6stat.ip6s_cantforward++;
657 			m_freem(m);
658 			return;
659 		}
660 		if (!ours) {
661 			m_freem(m);
662 			return;
663 		}
664 	} else if (!ours) {
665 		ip6_forward(m, 0);
666 		return;
667 	}
668 
669 	ip6 = mtod(m, struct ip6_hdr *);
670 
671 	/*
672 	 * Malicious party may be able to use IPv4 mapped addr to confuse
673 	 * tcp/udp stack and bypass security checks (act as if it was from
674 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
675 	 *
676 	 * For SIIT end node behavior, you may want to disable the check.
677 	 * However, you will  become vulnerable to attacks using IPv4 mapped
678 	 * source.
679 	 */
680 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
681 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
682 		ip6stat.ip6s_badscope++;
683 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
684 		goto bad;
685 	}
686 
687 	/*
688 	 * Tell launch routine the next header
689 	 */
690 #ifdef IFA_STATS
691 	if (IFA_STATS && deliverifp != NULL) {
692 		struct in6_ifaddr *ia6;
693 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
694 		if (ia6)
695 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
696 	}
697 #endif
698 	ip6stat.ip6s_delivered++;
699 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
700 	nest = 0;
701 	while (nxt != IPPROTO_DONE) {
702 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
703 			ip6stat.ip6s_toomanyhdr++;
704 			goto bad;
705 		}
706 
707 		/*
708 		 * protection against faulty packet - there should be
709 		 * more sanity checks in header chain processing.
710 		 */
711 		if (m->m_pkthdr.len < off) {
712 			ip6stat.ip6s_tooshort++;
713 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
714 			goto bad;
715 		}
716 
717 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
718 	}
719 	return;
720  bad:
721 	m_freem(m);
722 }
723 
724 /*
725  * Hop-by-Hop options header processing. If a valid jumbo payload option is
726  * included, the real payload length will be stored in plenp.
727  */
728 static int
729 ip6_hopopts_input(plenp, rtalertp, mp, offp)
730 	u_int32_t *plenp;
731 	u_int32_t *rtalertp;	/* XXX: should be stored more smart way */
732 	struct mbuf **mp;
733 	int *offp;
734 {
735 	register struct mbuf *m = *mp;
736 	int off = *offp, hbhlen;
737 	struct ip6_hbh *hbh;
738 	u_int8_t *opt;
739 
740 	/* validation of the length of the header */
741 #ifndef PULLDOWN_TEST
742 	IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
743 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
744 	hbhlen = (hbh->ip6h_len + 1) << 3;
745 
746 	IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
747 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
748 #else
749 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
750 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
751 	if (hbh == NULL) {
752 		ip6stat.ip6s_tooshort++;
753 		return -1;
754 	}
755 	hbhlen = (hbh->ip6h_len + 1) << 3;
756 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
757 		hbhlen);
758 	if (hbh == NULL) {
759 		ip6stat.ip6s_tooshort++;
760 		return -1;
761 	}
762 #endif
763 	off += hbhlen;
764 	hbhlen -= sizeof(struct ip6_hbh);
765 	opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
766 
767 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
768 				hbhlen, rtalertp, plenp) < 0)
769 		return(-1);
770 
771 	*offp = off;
772 	*mp = m;
773 	return(0);
774 }
775 
776 /*
777  * Search header for all Hop-by-hop options and process each option.
778  * This function is separate from ip6_hopopts_input() in order to
779  * handle a case where the sending node itself process its hop-by-hop
780  * options header. In such a case, the function is called from ip6_output().
781  */
782 int
783 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
784 	struct mbuf *m;
785 	u_int8_t *opthead;
786 	int hbhlen;
787 	u_int32_t *rtalertp;
788 	u_int32_t *plenp;
789 {
790 	struct ip6_hdr *ip6;
791 	int optlen = 0;
792 	u_int8_t *opt = opthead;
793 	u_int16_t rtalert_val;
794 	u_int32_t jumboplen;
795 
796 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
797 		switch(*opt) {
798 		 case IP6OPT_PAD1:
799 			 optlen = 1;
800 			 break;
801 		 case IP6OPT_PADN:
802 			 if (hbhlen < IP6OPT_MINLEN) {
803 				 ip6stat.ip6s_toosmall++;
804 				 goto bad;
805 			 }
806 			 optlen = *(opt + 1) + 2;
807 			 break;
808 		 case IP6OPT_RTALERT:
809 			 /* XXX may need check for alignment */
810 			 if (hbhlen < IP6OPT_RTALERT_LEN) {
811 				 ip6stat.ip6s_toosmall++;
812 				 goto bad;
813 			 }
814 			 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2)
815 				  /* XXX: should we discard the packet? */
816 				 log(LOG_ERR, "length of router alert opt is inconsitent(%d)",
817 				     *(opt + 1));
818 			 optlen = IP6OPT_RTALERT_LEN;
819 			 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
820 			 *rtalertp = ntohs(rtalert_val);
821 			 break;
822 		 case IP6OPT_JUMBO:
823 			/* XXX may need check for alignment */
824 			if (hbhlen < IP6OPT_JUMBO_LEN) {
825 				ip6stat.ip6s_toosmall++;
826 				goto bad;
827 			}
828 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2)
829 				 /* XXX: should we discard the packet? */
830 				log(LOG_ERR, "length of jumbopayload opt "
831 				    "is inconsistent(%d)",
832 				    *(opt + 1));
833 			optlen = IP6OPT_JUMBO_LEN;
834 
835 			/*
836 			 * IPv6 packets that have non 0 payload length
837 			 * must not contain a jumbo paylod option.
838 			 */
839 			ip6 = mtod(m, struct ip6_hdr *);
840 			if (ip6->ip6_plen) {
841 				ip6stat.ip6s_badoptions++;
842 				icmp6_error(m, ICMP6_PARAM_PROB,
843 					    ICMP6_PARAMPROB_HEADER,
844 					    sizeof(struct ip6_hdr) +
845 					    sizeof(struct ip6_hbh) +
846 					    opt - opthead);
847 				return(-1);
848 			}
849 
850 			/*
851 			 * We may see jumbolen in unaligned location, so
852 			 * we'd need to perform bcopy().
853 			 */
854 			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
855 			jumboplen = (u_int32_t)htonl(jumboplen);
856 
857 #if 1
858 			/*
859 			 * if there are multiple jumbo payload options,
860 			 * *plenp will be non-zero and the packet will be
861 			 * rejected.
862 			 * the behavior may need some debate in ipngwg -
863 			 * multiple options does not make sense, however,
864 			 * there's no explicit mention in specification.
865 			 */
866 			if (*plenp != 0) {
867 				ip6stat.ip6s_badoptions++;
868 				icmp6_error(m, ICMP6_PARAM_PROB,
869 					    ICMP6_PARAMPROB_HEADER,
870 					    sizeof(struct ip6_hdr) +
871 					    sizeof(struct ip6_hbh) +
872 					    opt + 2 - opthead);
873 				return(-1);
874 			}
875 #endif
876 
877 			/*
878 			 * jumbo payload length must be larger than 65535.
879 			 */
880 			if (jumboplen <= IPV6_MAXPACKET) {
881 				ip6stat.ip6s_badoptions++;
882 				icmp6_error(m, ICMP6_PARAM_PROB,
883 					    ICMP6_PARAMPROB_HEADER,
884 					    sizeof(struct ip6_hdr) +
885 					    sizeof(struct ip6_hbh) +
886 					    opt + 2 - opthead);
887 				return(-1);
888 			}
889 			*plenp = jumboplen;
890 
891 			break;
892 		 default:		/* unknown option */
893 			 if (hbhlen < IP6OPT_MINLEN) {
894 				 ip6stat.ip6s_toosmall++;
895 				 goto bad;
896 			 }
897 			 if ((optlen = ip6_unknown_opt(opt, m,
898 						       sizeof(struct ip6_hdr) +
899 						       sizeof(struct ip6_hbh) +
900 						       opt - opthead)) == -1)
901 				 return(-1);
902 			 optlen += 2;
903 			 break;
904 		}
905 	}
906 
907 	return(0);
908 
909   bad:
910 	m_freem(m);
911 	return(-1);
912 }
913 
914 /*
915  * Unknown option processing.
916  * The third argument `off' is the offset from the IPv6 header to the option,
917  * which is necessary if the IPv6 header the and option header and IPv6 header
918  * is not continuous in order to return an ICMPv6 error.
919  */
920 int
921 ip6_unknown_opt(optp, m, off)
922 	u_int8_t *optp;
923 	struct mbuf *m;
924 	int off;
925 {
926 	struct ip6_hdr *ip6;
927 
928 	switch(IP6OPT_TYPE(*optp)) {
929 	 case IP6OPT_TYPE_SKIP: /* ignore the option */
930 		 return((int)*(optp + 1));
931 	 case IP6OPT_TYPE_DISCARD:	/* silently discard */
932 		 m_freem(m);
933 		 return(-1);
934 	 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
935 		 ip6stat.ip6s_badoptions++;
936 		 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
937 		 return(-1);
938 	 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
939 		 ip6stat.ip6s_badoptions++;
940 		 ip6 = mtod(m, struct ip6_hdr *);
941 		 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
942 		     (m->m_flags & (M_BCAST|M_MCAST)))
943 			 m_freem(m);
944 		 else
945 			 icmp6_error(m, ICMP6_PARAM_PROB,
946 				     ICMP6_PARAMPROB_OPTION, off);
947 		 return(-1);
948 	}
949 
950 	m_freem(m);		/* XXX: NOTREACHED */
951 	return(-1);
952 }
953 
954 /*
955  * Create the "control" list for this pcb.
956  *
957  * The routine will be called from upper layer handlers like tcp6_input().
958  * Thus the routine assumes that the caller (tcp6_input) have already
959  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
960  * very first mbuf on the mbuf chain.
961  * We may want to add some infinite loop prevention or sanity checks for safety.
962  * (This applies only when you are using KAME mbuf chain restriction, i.e.
963  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
964  */
965 void
966 ip6_savecontrol(in6p, mp, ip6, m)
967 	register struct in6pcb *in6p;
968 	register struct mbuf **mp;
969 	register struct ip6_hdr *ip6;
970 	register struct mbuf *m;
971 {
972 	struct proc *p = curproc;	/* XXX */
973 	int privileged;
974 
975 	privileged = 0;
976 	if (p && !suser(p->p_ucred, &p->p_acflag))
977 		privileged++;
978 
979 #ifdef SO_TIMESTAMP
980 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP) {
981 		struct timeval tv;
982 
983 		microtime(&tv);
984 		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
985 			SCM_TIMESTAMP, SOL_SOCKET);
986 		if (*mp)
987 			mp = &(*mp)->m_next;
988 	}
989 #endif
990 	if (in6p->in6p_flags & IN6P_RECVDSTADDR) {
991 		*mp = sbcreatecontrol((caddr_t) &ip6->ip6_dst,
992 			sizeof(struct in6_addr), IPV6_RECVDSTADDR,
993 			IPPROTO_IPV6);
994 		if (*mp)
995 			mp = &(*mp)->m_next;
996 	}
997 
998 #ifdef noyet
999 	/* options were tossed above */
1000 	if (in6p->in6p_flags & IN6P_RECVOPTS)
1001 		/* broken */
1002 	/* ip6_srcroute doesn't do what we want here, need to fix */
1003 	if (in6p->in6p_flags & IPV6P_RECVRETOPTS)
1004 		/* broken */
1005 #endif
1006 
1007 	/* RFC 2292 sec. 5 */
1008 	if (in6p->in6p_flags & IN6P_PKTINFO) {
1009 		struct in6_pktinfo pi6;
1010 		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1011 		if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
1012 			pi6.ipi6_addr.s6_addr16[1] = 0;
1013 		pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
1014 					? m->m_pkthdr.rcvif->if_index
1015 					: 0;
1016 		*mp = sbcreatecontrol((caddr_t) &pi6,
1017 			sizeof(struct in6_pktinfo), IPV6_PKTINFO,
1018 			IPPROTO_IPV6);
1019 		if (*mp)
1020 			mp = &(*mp)->m_next;
1021 	}
1022 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1023 		int hlim = ip6->ip6_hlim & 0xff;
1024 		*mp = sbcreatecontrol((caddr_t) &hlim,
1025 			sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6);
1026 		if (*mp)
1027 			mp = &(*mp)->m_next;
1028 	}
1029 	/* IN6P_NEXTHOP - for outgoing packet only */
1030 
1031 	/*
1032 	 * IPV6_HOPOPTS socket option. We require super-user privilege
1033 	 * for the option, but it might be too strict, since there might
1034 	 * be some hop-by-hop options which can be returned to normal user.
1035 	 * See RFC 2292 section 6.
1036 	 */
1037 	if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) {
1038 		/*
1039 		 * Check if a hop-by-hop options header is contatined in the
1040 		 * received packet, and if so, store the options as ancillary
1041 		 * data. Note that a hop-by-hop options header must be
1042 		 * just after the IPv6 header, which fact is assured through
1043 		 * the IPv6 input processing.
1044 		 */
1045 		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1046 		if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1047 			struct ip6_hbh *hbh;
1048 			int hbhlen;
1049 
1050 #ifndef PULLDOWN_TEST
1051 			hbh = (struct ip6_hbh *)(ip6 + 1);
1052 			hbhlen = (hbh->ip6h_len + 1) << 3;
1053 #else
1054 			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1055 				sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
1056 			if (hbh == NULL) {
1057 				ip6stat.ip6s_tooshort++;
1058 				return;
1059 			}
1060 			hbhlen = (hbh->ip6h_len + 1) << 3;
1061 			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1062 				sizeof(struct ip6_hdr), hbhlen);
1063 			if (hbh == NULL) {
1064 				ip6stat.ip6s_tooshort++;
1065 				return;
1066 			}
1067 #endif
1068 
1069 			/*
1070 			 * XXX: We copy whole the header even if a jumbo
1071 			 * payload option is included, which option is to
1072 			 * be removed before returning in the RFC 2292.
1073 			 * But it's too painful operation...
1074 			 */
1075 			*mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1076 					      IPV6_HOPOPTS, IPPROTO_IPV6);
1077 			if (*mp)
1078 				mp = &(*mp)->m_next;
1079 		}
1080 	}
1081 
1082 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1083 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1084 		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1085 		int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);;
1086 
1087 		/*
1088 		 * Search for destination options headers or routing
1089 		 * header(s) through the header chain, and stores each
1090 		 * header as ancillary data.
1091 		 * Note that the order of the headers remains in
1092 		 * the chain of ancillary data.
1093 		 */
1094 		while(1) {	/* is explicit loop prevention necessary? */
1095 			struct ip6_ext *ip6e;
1096 			int elen;
1097 
1098 #ifndef PULLDOWN_TEST
1099 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1100 			if (nxt == IPPROTO_AH)
1101 				elen = (ip6e->ip6e_len + 2) << 2;
1102 			else
1103 				elen = (ip6e->ip6e_len + 1) << 3;
1104 #else
1105 			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off,
1106 				sizeof(struct ip6_ext));
1107 			if (ip6e == NULL) {
1108 				ip6stat.ip6s_tooshort++;
1109 				return;
1110 			}
1111 			if (nxt == IPPROTO_AH)
1112 				elen = (ip6e->ip6e_len + 2) << 2;
1113 			else
1114 				elen = (ip6e->ip6e_len + 1) << 3;
1115 			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off, elen);
1116 			if (ip6e == NULL) {
1117 				ip6stat.ip6s_tooshort++;
1118 				return;
1119 			}
1120 #endif
1121 
1122 			switch(nxt) {
1123 		         case IPPROTO_DSTOPTS:
1124 				 if (!in6p->in6p_flags & IN6P_DSTOPTS)
1125 					 break;
1126 
1127 				 /*
1128 				  * We also require super-user privilege for
1129 				  * the option.
1130 				  * See the comments on IN6_HOPOPTS.
1131 				  */
1132 				 if (!privileged)
1133 					 break;
1134 
1135 				 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1136 						       IPV6_DSTOPTS,
1137 						       IPPROTO_IPV6);
1138 				 if (*mp)
1139 					 mp = &(*mp)->m_next;
1140 				 break;
1141 
1142 			 case IPPROTO_ROUTING:
1143 				 if (!in6p->in6p_flags & IN6P_RTHDR)
1144 					 break;
1145 
1146 				 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1147 						       IPV6_RTHDR,
1148 						       IPPROTO_IPV6);
1149 				 if (*mp)
1150 					 mp = &(*mp)->m_next;
1151 				 break;
1152 
1153 			 case IPPROTO_UDP:
1154 			 case IPPROTO_TCP:
1155 			 case IPPROTO_ICMPV6:
1156 			 default:
1157 				 /*
1158 				  * stop search if we encounter an upper
1159 				  * layer protocol headers.
1160 				  */
1161 				 goto loopend;
1162 
1163 			 case IPPROTO_HOPOPTS:
1164 			 case IPPROTO_AH: /* is it possible? */
1165 				 break;
1166 			}
1167 
1168 			/* proceed with the next header. */
1169 			off += elen;
1170 			nxt = ip6e->ip6e_nxt;
1171 		}
1172 	  loopend:
1173 	}
1174 	if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) {
1175 		/* to be done */
1176 	}
1177 	if ((in6p->in6p_flags & IN6P_DSTOPTS) && privileged) {
1178 		/* to be done */
1179 	}
1180 	/* IN6P_RTHDR - to be done */
1181 
1182 }
1183 
1184 /*
1185  * Get pointer to the previous header followed by the header
1186  * currently processed.
1187  * XXX: This function supposes that
1188  *	M includes all headers,
1189  *	the next header field and the header length field of each header
1190  *	are valid, and
1191  *	the sum of each header length equals to OFF.
1192  * Because of these assumptions, this function must be called very
1193  * carefully. Moreover, it will not be used in the near future when
1194  * we develop `neater' mechanism to process extension headers.
1195  */
1196 char *
1197 ip6_get_prevhdr(m, off)
1198 	struct mbuf *m;
1199 	int off;
1200 {
1201 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1202 
1203 	if (off == sizeof(struct ip6_hdr))
1204 		return(&ip6->ip6_nxt);
1205 	else {
1206 		int len, nxt;
1207 		struct ip6_ext *ip6e = NULL;
1208 
1209 		nxt = ip6->ip6_nxt;
1210 		len = sizeof(struct ip6_hdr);
1211 		while (len < off) {
1212 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1213 
1214 			switch(nxt) {
1215 			case IPPROTO_FRAGMENT:
1216 				len += sizeof(struct ip6_frag);
1217 				break;
1218 			case IPPROTO_AH:
1219 				len += (ip6e->ip6e_len + 2) << 2;
1220 				break;
1221 			default:
1222 				len += (ip6e->ip6e_len + 1) << 3;
1223 				break;
1224 			}
1225 			nxt = ip6e->ip6e_nxt;
1226 		}
1227 		if (ip6e)
1228 			return(&ip6e->ip6e_nxt);
1229 		else
1230 			return NULL;
1231 	}
1232 }
1233 
1234 /*
1235  * get next header offset.  m will be retained.
1236  */
1237 int
1238 ip6_nexthdr(m, off, proto, nxtp)
1239 	struct mbuf *m;
1240 	int off;
1241 	int proto;
1242 	int *nxtp;
1243 {
1244 	struct ip6_hdr ip6;
1245 	struct ip6_ext ip6e;
1246 	struct ip6_frag fh;
1247 
1248 	/* just in case */
1249 	if (m == NULL)
1250 		panic("ip6_nexthdr: m == NULL");
1251 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1252 		return -1;
1253 
1254 	switch (proto) {
1255 	case IPPROTO_IPV6:
1256 		if (m->m_pkthdr.len < off + sizeof(ip6))
1257 			return -1;
1258 		m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1259 		if (nxtp)
1260 			*nxtp = ip6.ip6_nxt;
1261 		off += sizeof(ip6);
1262 		return off;
1263 
1264 	case IPPROTO_FRAGMENT:
1265 		/*
1266 		 * terminate parsing if it is not the first fragment,
1267 		 * it does not make sense to parse through it.
1268 		 */
1269 		if (m->m_pkthdr.len < off + sizeof(fh))
1270 			return -1;
1271 		m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1272 		if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0)
1273 			return -1;
1274 		if (nxtp)
1275 			*nxtp = fh.ip6f_nxt;
1276 		off += sizeof(struct ip6_frag);
1277 		return off;
1278 
1279 	case IPPROTO_AH:
1280 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1281 			return -1;
1282 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1283 		if (nxtp)
1284 			*nxtp = ip6e.ip6e_nxt;
1285 		off += (ip6e.ip6e_len + 2) << 2;
1286 		return off;
1287 
1288 	case IPPROTO_HOPOPTS:
1289 	case IPPROTO_ROUTING:
1290 	case IPPROTO_DSTOPTS:
1291 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1292 			return -1;
1293 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1294 		if (nxtp)
1295 			*nxtp = ip6e.ip6e_nxt;
1296 		off += (ip6e.ip6e_len + 1) << 3;
1297 		return off;
1298 
1299 	case IPPROTO_NONE:
1300 	case IPPROTO_ESP:
1301 	case IPPROTO_IPCOMP:
1302 		/* give up */
1303 		return -1;
1304 
1305 	default:
1306 		return -1;
1307 	}
1308 
1309 	return -1;
1310 }
1311 
1312 /*
1313  * get offset for the last header in the chain.  m will be kept untainted.
1314  */
1315 int
1316 ip6_lasthdr(m, off, proto, nxtp)
1317 	struct mbuf *m;
1318 	int off;
1319 	int proto;
1320 	int *nxtp;
1321 {
1322 	int newoff;
1323 	int nxt;
1324 
1325 	if (!nxtp) {
1326 		nxt = -1;
1327 		nxtp = &nxt;
1328 	}
1329 	while (1) {
1330 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1331 		if (newoff < 0)
1332 			return off;
1333 		else if (newoff < off)
1334 			return -1;	/* invalid */
1335 		else if (newoff == off)
1336 			return newoff;
1337 
1338 		off = newoff;
1339 		proto = *nxtp;
1340 	}
1341 }
1342 
1343 /*
1344  * System control for IP6
1345  */
1346 
1347 u_char	inet6ctlerrmap[PRC_NCMDS] = {
1348 	0,		0,		0,		0,
1349 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1350 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1351 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1352 	0,		0,		0,		0,
1353 	ENOPROTOOPT
1354 };
1355 
1356 #include <uvm/uvm_extern.h>
1357 #include <sys/sysctl.h>
1358 
1359 int
1360 ip6_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1361 	int *name;
1362 	u_int namelen;
1363 	void *oldp;
1364 	size_t *oldlenp;
1365 	void *newp;
1366 	size_t newlen;
1367 {
1368 	/* All sysctl names at this level are terminal. */
1369 	if (namelen != 1)
1370 		return ENOTDIR;
1371 
1372 	switch (name[0]) {
1373 
1374 	case IPV6CTL_FORWARDING:
1375 		return sysctl_int(oldp, oldlenp, newp, newlen,
1376 				  &ip6_forwarding);
1377 	case IPV6CTL_SENDREDIRECTS:
1378 		return sysctl_int(oldp, oldlenp, newp, newlen,
1379 				&ip6_sendredirects);
1380 	case IPV6CTL_DEFHLIM:
1381 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_defhlim);
1382 	case IPV6CTL_MAXFRAGPACKETS:
1383 		return sysctl_int(oldp, oldlenp, newp, newlen,
1384 				&ip6_maxfragpackets);
1385 	case IPV6CTL_ACCEPT_RTADV:
1386 		return sysctl_int(oldp, oldlenp, newp, newlen,
1387 				&ip6_accept_rtadv);
1388 	case IPV6CTL_KEEPFAITH:
1389 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_keepfaith);
1390 	case IPV6CTL_LOG_INTERVAL:
1391 		return sysctl_int(oldp, oldlenp, newp, newlen,
1392 				&ip6_log_interval);
1393 	case IPV6CTL_HDRNESTLIMIT:
1394 		return sysctl_int(oldp, oldlenp, newp, newlen,
1395 				&ip6_hdrnestlimit);
1396 	case IPV6CTL_DAD_COUNT:
1397 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_dad_count);
1398 	case IPV6CTL_AUTO_FLOWLABEL:
1399 		return sysctl_int(oldp, oldlenp, newp, newlen,
1400 				&ip6_auto_flowlabel);
1401 	case IPV6CTL_DEFMCASTHLIM:
1402 		return sysctl_int(oldp, oldlenp, newp, newlen,
1403 				&ip6_defmcasthlim);
1404 	case IPV6CTL_GIF_HLIM:
1405 		return sysctl_int(oldp, oldlenp, newp, newlen,
1406 				&ip6_gif_hlim);
1407 	case IPV6CTL_KAME_VERSION:
1408 		return sysctl_rdstring(oldp, oldlenp, newp, __KAME_VERSION);
1409 	case IPV6CTL_USE_DEPRECATED:
1410 		return sysctl_int(oldp, oldlenp, newp, newlen,
1411 				&ip6_use_deprecated);
1412 	case IPV6CTL_RR_PRUNE:
1413 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_rr_prune);
1414 #ifndef INET6_BINDV6ONLY
1415 	case IPV6CTL_BINDV6ONLY:
1416 		return sysctl_int(oldp, oldlenp, newp, newlen,
1417 				&ip6_bindv6only);
1418 #endif
1419 	default:
1420 		return EOPNOTSUPP;
1421 	}
1422 	/* NOTREACHED */
1423 }
1424