xref: /netbsd-src/sys/netinet6/ip6_input.c (revision 5aefcfdc06931dd97e76246d2fe0302f7b3fe094)
1 /*	$NetBSD: ip6_input.c,v 1.33 2000/12/28 21:41:00 thorpej Exp $	*/
2 /*	$KAME: ip6_input.c,v 1.121 2000/08/31 06:07:29 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 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 #ifdef PFIL_HOOKS
144 struct pfil_head inet6_pfil_hook;
145 #endif
146 
147 struct ip6stat ip6stat;
148 
149 static void ip6_init2 __P((void *));
150 
151 static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
152 
153 /*
154  * IP6 initialization: fill in IP6 protocol switch table.
155  * All protocols not implemented in kernel go to raw IP6 protocol handler.
156  */
157 void
158 ip6_init()
159 {
160 	register struct ip6protosw *pr;
161 	register int i;
162 	struct timeval tv;
163 
164 	pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
165 	if (pr == 0)
166 		panic("ip6_init");
167 	for (i = 0; i < IPPROTO_MAX; i++)
168 		ip6_protox[i] = pr - inet6sw;
169 	for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
170 	    pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
171 		if (pr->pr_domain->dom_family == PF_INET6 &&
172 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
173 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
174 	ip6intrq.ifq_maxlen = ip6qmaxlen;
175 	nd6_init();
176 	frag6_init();
177 #ifdef IPV6FIREWALL
178 	ip6_fw_init();
179 #endif
180 	/*
181 	 * in many cases, random() here does NOT return random number
182 	 * as initialization during bootstrap time occur in fixed order.
183 	 */
184 	microtime(&tv);
185 	ip6_flow_seq = random() ^ tv.tv_usec;
186 
187 	ip6_init2((void *)0);
188 
189 #ifdef PFIL_HOOKS
190 	/* Register our Packet Filter hook. */
191 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
192 	inet6_pfil_hook.ph_af   = AF_INET6;
193 	i = pfil_head_register(&inet6_pfil_hook);
194 	if (i != 0)
195 		printf("ip6_init: WARNING: unable to register pfil hook, "
196 		    "error %d\n", i);
197 #endif /* PFIL_HOOKS */
198 }
199 
200 static void
201 ip6_init2(dummy)
202 	void *dummy;
203 {
204 	/*
205 	 * to route local address of p2p link to loopback,
206 	 * assign loopback address first.
207 	 */
208 	in6_ifattach(&loif[0], NULL);
209 
210 	/* nd6_timer_init */
211 	callout_init(&nd6_timer_ch);
212 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
213 	/* router renumbering prefix list maintenance */
214 	callout_init(&in6_rr_timer_ch);
215 	callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
216 }
217 
218 /*
219  * IP6 input interrupt handling. Just pass the packet to ip6_input.
220  */
221 void
222 ip6intr()
223 {
224 	int s;
225 	struct mbuf *m;
226 
227 	for (;;) {
228 		s = splimp();
229 		IF_DEQUEUE(&ip6intrq, m);
230 		splx(s);
231 		if (m == 0)
232 			return;
233 		ip6_input(m);
234 	}
235 }
236 
237 extern struct	route_in6 ip6_forward_rt;
238 
239 void
240 ip6_input(m)
241 	struct mbuf *m;
242 {
243 	struct ip6_hdr *ip6;
244 	int off = sizeof(struct ip6_hdr), nest;
245 	u_int32_t plen;
246 	u_int32_t rtalert = ~0;
247 	int nxt, ours = 0;
248 	struct ifnet *deliverifp = NULL;
249 
250 #ifdef IPSEC
251 	/*
252 	 * should the inner packet be considered authentic?
253 	 * see comment in ah4_input().
254 	 */
255 	if (m) {
256 		m->m_flags &= ~M_AUTHIPHDR;
257 		m->m_flags &= ~M_AUTHIPDGM;
258 	}
259 #endif
260 
261 	/*
262 	 * mbuf statistics by kazu
263 	 */
264 	if (m->m_flags & M_EXT) {
265 		if (m->m_next)
266 			ip6stat.ip6s_mext2m++;
267 		else
268 			ip6stat.ip6s_mext1++;
269 	} else {
270 		if (m->m_next) {
271 			if (m->m_flags & M_LOOP) {
272 				ip6stat.ip6s_m2m[loif[0].if_index]++;	/*XXX*/
273 			} else if (m->m_pkthdr.rcvif->if_index <= 31)
274 				ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
275 			else
276 				ip6stat.ip6s_m2m[0]++;
277 		} else
278 			ip6stat.ip6s_m1++;
279 	}
280 
281 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
282 	ip6stat.ip6s_total++;
283 
284 #ifndef PULLDOWN_TEST
285 	/* XXX is the line really necessary? */
286 	IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/);
287 #endif
288 
289 	if (m->m_len < sizeof(struct ip6_hdr)) {
290 		struct ifnet *inifp;
291 		inifp = m->m_pkthdr.rcvif;
292 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
293 			ip6stat.ip6s_toosmall++;
294 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
295 			return;
296 		}
297 	}
298 
299 	ip6 = mtod(m, struct ip6_hdr *);
300 
301 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
302 		ip6stat.ip6s_badvers++;
303 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
304 		goto bad;
305 	}
306 
307 #ifdef PFIL_HOOKS
308 	/*
309 	 * Run through list of hooks for input packets.  If there are any
310 	 * filters which require that additional packets in the flow are
311 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
312 	 * Note that filters must _never_ set this flag, as another filter
313 	 * in the list may have previously cleared it.
314 	 */
315 	if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
316 			   PFIL_IN) != 0)
317 		return;
318 	if (m == NULL)
319 		return;
320 	ip6 = mtod(m, struct ip6_hdr *);
321 #endif /* PFIL_HOOKS */
322 
323 
324 	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
325 
326 #ifdef IPV6FIREWALL
327 	/*
328 	 * Check with the firewall...
329 	 */
330 	if (ip6_fw_chk_ptr) {
331 		u_short port = 0;
332 		/* If ipfw says divert, we have to just drop packet */
333 		/* use port as a dummy argument */
334 		if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) {
335 			m_freem(m);
336 			m = NULL;
337 		}
338 		if (!m)
339 			return;
340 	}
341 #endif
342 
343 #ifdef ALTQ
344 	/* XXX Temporary until ALTQ is changed to use a pfil hook */
345 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
346 		/* packet is dropped by traffic conditioner */
347 		return;
348 	}
349 #endif
350 
351 	/*
352 	 * Scope check
353 	 */
354 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
355 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
356 		ip6stat.ip6s_badscope++;
357 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
358 		goto bad;
359 	}
360 	/*
361 	 * The following check is not documented in the spec.  Malicious party
362 	 * may be able to use IPv4 mapped addr to confuse tcp/udp stack and
363 	 * bypass security checks (act as if it was from 127.0.0.1 by using
364 	 * IPv6 src ::ffff:127.0.0.1).	Be cautious.
365 	 */
366 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
367 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
368 		ip6stat.ip6s_badscope++;
369 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
370 		goto bad;
371 	}
372 #if 0
373 	/*
374 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
375 	 *
376 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
377 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
378 	 * is revised to forbid relaying case.
379 	 */
380 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
381 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
382 		ip6stat.ip6s_badscope++;
383 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
384 		goto bad;
385 	}
386 #endif
387 	if (IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
388 	    IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) {
389 		if (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) {
390 			ours = 1;
391 			deliverifp = m->m_pkthdr.rcvif;
392 			goto hbhcheck;
393 		} else {
394 			ip6stat.ip6s_badscope++;
395 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
396 			goto bad;
397 		}
398 	}
399 
400 #ifndef FAKE_LOOPBACK_IF
401 	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0)
402 #else
403 	if (1)
404 #endif
405 	{
406 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
407 			ip6->ip6_src.s6_addr16[1]
408 				= htons(m->m_pkthdr.rcvif->if_index);
409 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
410 			ip6->ip6_dst.s6_addr16[1]
411 				= htons(m->m_pkthdr.rcvif->if_index);
412 	}
413 
414 	/*
415 	 * XXX we need this since we do not have "goto ours" hack route
416 	 * for some of our ifaddrs on loopback interface.
417 	 * we should correct it by changing in6_ifattach to install
418 	 * "goto ours" hack route.
419 	 */
420 	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0) {
421 		if (IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) {
422 			ours = 1;
423 			deliverifp = m->m_pkthdr.rcvif;
424 			goto hbhcheck;
425 		}
426 	}
427 
428 	/*
429 	 * Multicast check
430 	 */
431 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
432 	  	struct	in6_multi *in6m = 0;
433 
434 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
435 		/*
436 		 * See if we belong to the destination multicast group on the
437 		 * arrival interface.
438 		 */
439 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
440 		if (in6m)
441 			ours = 1;
442 		else if (!ip6_mrouter) {
443 			ip6stat.ip6s_notmember++;
444 			ip6stat.ip6s_cantforward++;
445 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
446 			goto bad;
447 		}
448 		deliverifp = m->m_pkthdr.rcvif;
449 		goto hbhcheck;
450 	}
451 
452 	/*
453 	 *  Unicast check
454 	 */
455 	if (ip6_forward_rt.ro_rt != NULL &&
456 	    (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
457 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
458 			       &ip6_forward_rt.ro_dst.sin6_addr))
459 		ip6stat.ip6s_forward_cachehit++;
460 	else {
461 		if (ip6_forward_rt.ro_rt) {
462 			/* route is down or destination is different */
463 			ip6stat.ip6s_forward_cachemiss++;
464 			RTFREE(ip6_forward_rt.ro_rt);
465 			ip6_forward_rt.ro_rt = 0;
466 		}
467 
468 		bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
469 		ip6_forward_rt.ro_dst.sin6_len = sizeof(struct sockaddr_in6);
470 		ip6_forward_rt.ro_dst.sin6_family = AF_INET6;
471 		ip6_forward_rt.ro_dst.sin6_addr = ip6->ip6_dst;
472 
473 		rtalloc((struct route *)&ip6_forward_rt);
474 	}
475 
476 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
477 
478 	/*
479 	 * Accept the packet if the forwarding interface to the destination
480 	 * according to the routing table is the loopback interface,
481 	 * unless the associated route has a gateway.
482 	 * Note that this approach causes to accept a packet if there is a
483 	 * route to the loopback interface for the destination of the packet.
484 	 * But we think it's even useful in some situations, e.g. when using
485 	 * a special daemon which wants to intercept the packet.
486 	 */
487 	if (ip6_forward_rt.ro_rt &&
488 	    (ip6_forward_rt.ro_rt->rt_flags &
489 	     (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
490 #if 0
491 	    /*
492 	     * The check below is redundant since the comparison of
493 	     * the destination and the key of the rtentry has
494 	     * already done through looking up the routing table.
495 	     */
496 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
497 			       &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) &&
498 #endif
499 	    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
500 		struct in6_ifaddr *ia6 =
501 			(struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
502 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
503 			m->m_flags |= M_ANYCAST6;
504 		/*
505 		 * packets to a tentative, duplicated, or somehow invalid
506 		 * address must not be accepted.
507 		 */
508 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
509 			/* this address is ready */
510 			ours = 1;
511 			deliverifp = ia6->ia_ifp;	/* correct? */
512 			goto hbhcheck;
513 		} else {
514 			/* address is not ready, so discard the packet. */
515 			log(LOG_INFO,
516 			    "ip6_input: packet to an unready address %s->%s\n",
517 			    ip6_sprintf(&ip6->ip6_src),
518 			    ip6_sprintf(&ip6->ip6_dst));
519 
520 			goto bad;
521 		}
522 	}
523 
524 	/*
525 	 * FAITH(Firewall Aided Internet Translator)
526 	 */
527 #if defined(NFAITH) && 0 < NFAITH
528 	if (ip6_keepfaith) {
529 		if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
530 		 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
531 			/* XXX do we need more sanity checks? */
532 			ours = 1;
533 			deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /*faith*/
534 			goto hbhcheck;
535 		}
536 	}
537 #endif
538 
539 #if 0
540     {
541 	/*
542 	 * Last resort: check in6_ifaddr for incoming interface.
543 	 * The code is here until I update the "goto ours hack" code above
544 	 * working right.
545 	 */
546 	struct ifaddr *ifa;
547 	for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
548 	     ifa;
549 	     ifa = ifa->ifa_list.tqe_next) {
550 		if (ifa->ifa_addr == NULL)
551 			continue;	/* just for safety */
552 		if (ifa->ifa_addr->sa_family != AF_INET6)
553 			continue;
554 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
555 			ours = 1;
556 			deliverifp = ifa->ifa_ifp;
557 			goto hbhcheck;
558 		}
559 	}
560     }
561 #endif
562 
563 	/*
564 	 * Now there is no reason to process the packet if it's not our own
565 	 * and we're not a router.
566 	 */
567 	if (!ip6_forwarding) {
568 		ip6stat.ip6s_cantforward++;
569 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
570 		goto bad;
571 	}
572 
573   hbhcheck:
574 	/*
575 	 * Process Hop-by-Hop options header if it's contained.
576 	 * m may be modified in ip6_hopopts_input().
577 	 * If a JumboPayload option is included, plen will also be modified.
578 	 */
579 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
580 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
581 		struct ip6_hbh *hbh;
582 
583 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
584 #if 0	/*touches NULL pointer*/
585 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
586 #endif
587 			return;	/* m have already been freed */
588 		}
589 
590 		/* adjust pointer */
591 		ip6 = mtod(m, struct ip6_hdr *);
592 
593 		/*
594 		 * if the payload length field is 0 and the next header field
595 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
596 		 * option MUST be included.
597 		 */
598 		if (ip6->ip6_plen == 0 && plen == 0) {
599 			/*
600 			 * Note that if a valid jumbo payload option is
601 			 * contained, ip6_hoptops_input() must set a valid
602 			 * (non-zero) payload length to the variable plen.
603 			 */
604 			ip6stat.ip6s_badoptions++;
605 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
606 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
607 			icmp6_error(m, ICMP6_PARAM_PROB,
608 				    ICMP6_PARAMPROB_HEADER,
609 				    (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
610 			return;
611 		}
612 #ifndef PULLDOWN_TEST
613 		/* ip6_hopopts_input() ensures that mbuf is contiguous */
614 		hbh = (struct ip6_hbh *)(ip6 + 1);
615 #else
616 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
617 			sizeof(struct ip6_hbh));
618 		if (hbh == NULL) {
619 			ip6stat.ip6s_tooshort++;
620 			return;
621 		}
622 #endif
623 		nxt = hbh->ip6h_nxt;
624 
625 		/*
626 		 * accept the packet if a router alert option is included
627 		 * and we act as an IPv6 router.
628 		 */
629 		if (rtalert != ~0 && ip6_forwarding)
630 			ours = 1;
631 	} else
632 		nxt = ip6->ip6_nxt;
633 
634 	/*
635 	 * Check that the amount of data in the buffers
636 	 * is as at least much as the IPv6 header would have us expect.
637 	 * Trim mbufs if longer than we expect.
638 	 * Drop packet if shorter than we expect.
639 	 */
640 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
641 		ip6stat.ip6s_tooshort++;
642 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
643 		goto bad;
644 	}
645 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
646 		if (m->m_len == m->m_pkthdr.len) {
647 			m->m_len = sizeof(struct ip6_hdr) + plen;
648 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
649 		} else
650 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
651 	}
652 
653 	/*
654 	 * Forward if desirable.
655 	 */
656 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
657 		/*
658 		 * If we are acting as a multicast router, all
659 		 * incoming multicast packets are passed to the
660 		 * kernel-level multicast forwarding function.
661 		 * The packet is returned (relatively) intact; if
662 		 * ip6_mforward() returns a non-zero value, the packet
663 		 * must be discarded, else it may be accepted below.
664 		 */
665 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
666 			ip6stat.ip6s_cantforward++;
667 			m_freem(m);
668 			return;
669 		}
670 		if (!ours) {
671 			m_freem(m);
672 			return;
673 		}
674 	} else if (!ours) {
675 		ip6_forward(m, 0);
676 		return;
677 	}
678 
679 	ip6 = mtod(m, struct ip6_hdr *);
680 
681 	/*
682 	 * Malicious party may be able to use IPv4 mapped addr to confuse
683 	 * tcp/udp stack and bypass security checks (act as if it was from
684 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
685 	 *
686 	 * For SIIT end node behavior, you may want to disable the check.
687 	 * However, you will  become vulnerable to attacks using IPv4 mapped
688 	 * source.
689 	 */
690 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
691 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
692 		ip6stat.ip6s_badscope++;
693 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
694 		goto bad;
695 	}
696 
697 	/*
698 	 * Tell launch routine the next header
699 	 */
700 #ifdef IFA_STATS
701 	if (deliverifp != NULL) {
702 		struct in6_ifaddr *ia6;
703 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
704 		if (ia6)
705 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
706 	}
707 #endif
708 	ip6stat.ip6s_delivered++;
709 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
710 	nest = 0;
711 	while (nxt != IPPROTO_DONE) {
712 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
713 			ip6stat.ip6s_toomanyhdr++;
714 			goto bad;
715 		}
716 
717 		/*
718 		 * protection against faulty packet - there should be
719 		 * more sanity checks in header chain processing.
720 		 */
721 		if (m->m_pkthdr.len < off) {
722 			ip6stat.ip6s_tooshort++;
723 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
724 			goto bad;
725 		}
726 
727 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
728 	}
729 	return;
730  bad:
731 	m_freem(m);
732 }
733 
734 /*
735  * Hop-by-Hop options header processing. If a valid jumbo payload option is
736  * included, the real payload length will be stored in plenp.
737  */
738 static int
739 ip6_hopopts_input(plenp, rtalertp, mp, offp)
740 	u_int32_t *plenp;
741 	u_int32_t *rtalertp;	/* XXX: should be stored more smart way */
742 	struct mbuf **mp;
743 	int *offp;
744 {
745 	register struct mbuf *m = *mp;
746 	int off = *offp, hbhlen;
747 	struct ip6_hbh *hbh;
748 	u_int8_t *opt;
749 
750 	/* validation of the length of the header */
751 #ifndef PULLDOWN_TEST
752 	IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
753 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
754 	hbhlen = (hbh->ip6h_len + 1) << 3;
755 
756 	IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
757 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
758 #else
759 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
760 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
761 	if (hbh == NULL) {
762 		ip6stat.ip6s_tooshort++;
763 		return -1;
764 	}
765 	hbhlen = (hbh->ip6h_len + 1) << 3;
766 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
767 		hbhlen);
768 	if (hbh == NULL) {
769 		ip6stat.ip6s_tooshort++;
770 		return -1;
771 	}
772 #endif
773 	off += hbhlen;
774 	hbhlen -= sizeof(struct ip6_hbh);
775 	opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
776 
777 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
778 				hbhlen, rtalertp, plenp) < 0)
779 		return(-1);
780 
781 	*offp = off;
782 	*mp = m;
783 	return(0);
784 }
785 
786 /*
787  * Search header for all Hop-by-hop options and process each option.
788  * This function is separate from ip6_hopopts_input() in order to
789  * handle a case where the sending node itself process its hop-by-hop
790  * options header. In such a case, the function is called from ip6_output().
791  */
792 int
793 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
794 	struct mbuf *m;
795 	u_int8_t *opthead;
796 	int hbhlen;
797 	u_int32_t *rtalertp;
798 	u_int32_t *plenp;
799 {
800 	struct ip6_hdr *ip6;
801 	int optlen = 0;
802 	u_int8_t *opt = opthead;
803 	u_int16_t rtalert_val;
804 	u_int32_t jumboplen;
805 
806 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
807 		switch(*opt) {
808 		 case IP6OPT_PAD1:
809 			 optlen = 1;
810 			 break;
811 		 case IP6OPT_PADN:
812 			 if (hbhlen < IP6OPT_MINLEN) {
813 				 ip6stat.ip6s_toosmall++;
814 				 goto bad;
815 			 }
816 			 optlen = *(opt + 1) + 2;
817 			 break;
818 		 case IP6OPT_RTALERT:
819 			 /* XXX may need check for alignment */
820 			 if (hbhlen < IP6OPT_RTALERT_LEN) {
821 				 ip6stat.ip6s_toosmall++;
822 				 goto bad;
823 			 }
824 			 if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2)
825 				  /* XXX: should we discard the packet? */
826 				 log(LOG_ERR, "length of router alert opt is inconsitent(%d)",
827 				     *(opt + 1));
828 			 optlen = IP6OPT_RTALERT_LEN;
829 			 bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
830 			 *rtalertp = ntohs(rtalert_val);
831 			 break;
832 		 case IP6OPT_JUMBO:
833 			/* XXX may need check for alignment */
834 			if (hbhlen < IP6OPT_JUMBO_LEN) {
835 				ip6stat.ip6s_toosmall++;
836 				goto bad;
837 			}
838 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2)
839 				 /* XXX: should we discard the packet? */
840 				log(LOG_ERR, "length of jumbopayload opt "
841 				    "is inconsistent(%d)\n",
842 				    *(opt + 1));
843 			optlen = IP6OPT_JUMBO_LEN;
844 
845 			/*
846 			 * IPv6 packets that have non 0 payload length
847 			 * must not contain a jumbo paylod option.
848 			 */
849 			ip6 = mtod(m, struct ip6_hdr *);
850 			if (ip6->ip6_plen) {
851 				ip6stat.ip6s_badoptions++;
852 				icmp6_error(m, ICMP6_PARAM_PROB,
853 					    ICMP6_PARAMPROB_HEADER,
854 					    sizeof(struct ip6_hdr) +
855 					    sizeof(struct ip6_hbh) +
856 					    opt - opthead);
857 				return(-1);
858 			}
859 
860 			/*
861 			 * We may see jumbolen in unaligned location, so
862 			 * we'd need to perform bcopy().
863 			 */
864 			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
865 			jumboplen = (u_int32_t)htonl(jumboplen);
866 
867 #if 1
868 			/*
869 			 * if there are multiple jumbo payload options,
870 			 * *plenp will be non-zero and the packet will be
871 			 * rejected.
872 			 * the behavior may need some debate in ipngwg -
873 			 * multiple options does not make sense, however,
874 			 * there's no explicit mention in specification.
875 			 */
876 			if (*plenp != 0) {
877 				ip6stat.ip6s_badoptions++;
878 				icmp6_error(m, ICMP6_PARAM_PROB,
879 					    ICMP6_PARAMPROB_HEADER,
880 					    sizeof(struct ip6_hdr) +
881 					    sizeof(struct ip6_hbh) +
882 					    opt + 2 - opthead);
883 				return(-1);
884 			}
885 #endif
886 
887 			/*
888 			 * jumbo payload length must be larger than 65535.
889 			 */
890 			if (jumboplen <= IPV6_MAXPACKET) {
891 				ip6stat.ip6s_badoptions++;
892 				icmp6_error(m, ICMP6_PARAM_PROB,
893 					    ICMP6_PARAMPROB_HEADER,
894 					    sizeof(struct ip6_hdr) +
895 					    sizeof(struct ip6_hbh) +
896 					    opt + 2 - opthead);
897 				return(-1);
898 			}
899 			*plenp = jumboplen;
900 
901 			break;
902 		 default:		/* unknown option */
903 			 if (hbhlen < IP6OPT_MINLEN) {
904 				 ip6stat.ip6s_toosmall++;
905 				 goto bad;
906 			 }
907 			 if ((optlen = ip6_unknown_opt(opt, m,
908 						       sizeof(struct ip6_hdr) +
909 						       sizeof(struct ip6_hbh) +
910 						       opt - opthead)) == -1)
911 				 return(-1);
912 			 optlen += 2;
913 			 break;
914 		}
915 	}
916 
917 	return(0);
918 
919   bad:
920 	m_freem(m);
921 	return(-1);
922 }
923 
924 /*
925  * Unknown option processing.
926  * The third argument `off' is the offset from the IPv6 header to the option,
927  * which is necessary if the IPv6 header the and option header and IPv6 header
928  * is not continuous in order to return an ICMPv6 error.
929  */
930 int
931 ip6_unknown_opt(optp, m, off)
932 	u_int8_t *optp;
933 	struct mbuf *m;
934 	int off;
935 {
936 	struct ip6_hdr *ip6;
937 
938 	switch(IP6OPT_TYPE(*optp)) {
939 	 case IP6OPT_TYPE_SKIP: /* ignore the option */
940 		 return((int)*(optp + 1));
941 	 case IP6OPT_TYPE_DISCARD:	/* silently discard */
942 		 m_freem(m);
943 		 return(-1);
944 	 case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
945 		 ip6stat.ip6s_badoptions++;
946 		 icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
947 		 return(-1);
948 	 case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
949 		 ip6stat.ip6s_badoptions++;
950 		 ip6 = mtod(m, struct ip6_hdr *);
951 		 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
952 		     (m->m_flags & (M_BCAST|M_MCAST)))
953 			 m_freem(m);
954 		 else
955 			 icmp6_error(m, ICMP6_PARAM_PROB,
956 				     ICMP6_PARAMPROB_OPTION, off);
957 		 return(-1);
958 	}
959 
960 	m_freem(m);		/* XXX: NOTREACHED */
961 	return(-1);
962 }
963 
964 /*
965  * Create the "control" list for this pcb.
966  *
967  * The routine will be called from upper layer handlers like tcp6_input().
968  * Thus the routine assumes that the caller (tcp6_input) have already
969  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
970  * very first mbuf on the mbuf chain.
971  * We may want to add some infinite loop prevention or sanity checks for safety.
972  * (This applies only when you are using KAME mbuf chain restriction, i.e.
973  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
974  */
975 void
976 ip6_savecontrol(in6p, mp, ip6, m)
977 	register struct in6pcb *in6p;
978 	register struct mbuf **mp;
979 	register struct ip6_hdr *ip6;
980 	register struct mbuf *m;
981 {
982 	struct proc *p = curproc;	/* XXX */
983 	int privileged;
984 
985 	privileged = 0;
986 	if (p && !suser(p->p_ucred, &p->p_acflag))
987 		privileged++;
988 
989 #ifdef SO_TIMESTAMP
990 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP) {
991 		struct timeval tv;
992 
993 		microtime(&tv);
994 		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
995 			SCM_TIMESTAMP, SOL_SOCKET);
996 		if (*mp)
997 			mp = &(*mp)->m_next;
998 	}
999 #endif
1000 	if (in6p->in6p_flags & IN6P_RECVDSTADDR) {
1001 		*mp = sbcreatecontrol((caddr_t) &ip6->ip6_dst,
1002 			sizeof(struct in6_addr), IPV6_RECVDSTADDR,
1003 			IPPROTO_IPV6);
1004 		if (*mp)
1005 			mp = &(*mp)->m_next;
1006 	}
1007 
1008 #ifdef noyet
1009 	/* options were tossed above */
1010 	if (in6p->in6p_flags & IN6P_RECVOPTS)
1011 		/* broken */
1012 	/* ip6_srcroute doesn't do what we want here, need to fix */
1013 	if (in6p->in6p_flags & IPV6P_RECVRETOPTS)
1014 		/* broken */
1015 #endif
1016 
1017 	/* RFC 2292 sec. 5 */
1018 	if (in6p->in6p_flags & IN6P_PKTINFO) {
1019 		struct in6_pktinfo pi6;
1020 		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1021 		if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
1022 			pi6.ipi6_addr.s6_addr16[1] = 0;
1023 		pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
1024 					? m->m_pkthdr.rcvif->if_index
1025 					: 0;
1026 		*mp = sbcreatecontrol((caddr_t) &pi6,
1027 			sizeof(struct in6_pktinfo), IPV6_PKTINFO,
1028 			IPPROTO_IPV6);
1029 		if (*mp)
1030 			mp = &(*mp)->m_next;
1031 	}
1032 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1033 		int hlim = ip6->ip6_hlim & 0xff;
1034 		*mp = sbcreatecontrol((caddr_t) &hlim,
1035 			sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6);
1036 		if (*mp)
1037 			mp = &(*mp)->m_next;
1038 	}
1039 	/* IN6P_NEXTHOP - for outgoing packet only */
1040 
1041 	/*
1042 	 * IPV6_HOPOPTS socket option. We require super-user privilege
1043 	 * for the option, but it might be too strict, since there might
1044 	 * be some hop-by-hop options which can be returned to normal user.
1045 	 * See RFC 2292 section 6.
1046 	 */
1047 	if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) {
1048 		/*
1049 		 * Check if a hop-by-hop options header is contatined in the
1050 		 * received packet, and if so, store the options as ancillary
1051 		 * data. Note that a hop-by-hop options header must be
1052 		 * just after the IPv6 header, which fact is assured through
1053 		 * the IPv6 input processing.
1054 		 */
1055 		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1056 		if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1057 			struct ip6_hbh *hbh;
1058 			int hbhlen;
1059 
1060 #ifndef PULLDOWN_TEST
1061 			hbh = (struct ip6_hbh *)(ip6 + 1);
1062 			hbhlen = (hbh->ip6h_len + 1) << 3;
1063 #else
1064 			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1065 				sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
1066 			if (hbh == NULL) {
1067 				ip6stat.ip6s_tooshort++;
1068 				return;
1069 			}
1070 			hbhlen = (hbh->ip6h_len + 1) << 3;
1071 			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1072 				sizeof(struct ip6_hdr), hbhlen);
1073 			if (hbh == NULL) {
1074 				ip6stat.ip6s_tooshort++;
1075 				return;
1076 			}
1077 #endif
1078 
1079 			/*
1080 			 * XXX: We copy whole the header even if a jumbo
1081 			 * payload option is included, which option is to
1082 			 * be removed before returning in the RFC 2292.
1083 			 * But it's too painful operation...
1084 			 */
1085 			*mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1086 					      IPV6_HOPOPTS, IPPROTO_IPV6);
1087 			if (*mp)
1088 				mp = &(*mp)->m_next;
1089 		}
1090 	}
1091 
1092 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1093 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1094 		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1095 		int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);;
1096 
1097 		/*
1098 		 * Search for destination options headers or routing
1099 		 * header(s) through the header chain, and stores each
1100 		 * header as ancillary data.
1101 		 * Note that the order of the headers remains in
1102 		 * the chain of ancillary data.
1103 		 */
1104 		while(1) {	/* is explicit loop prevention necessary? */
1105 			struct ip6_ext *ip6e;
1106 			int elen;
1107 
1108 #ifndef PULLDOWN_TEST
1109 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1110 			if (nxt == IPPROTO_AH)
1111 				elen = (ip6e->ip6e_len + 2) << 2;
1112 			else
1113 				elen = (ip6e->ip6e_len + 1) << 3;
1114 #else
1115 			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off,
1116 				sizeof(struct ip6_ext));
1117 			if (ip6e == NULL) {
1118 				ip6stat.ip6s_tooshort++;
1119 				return;
1120 			}
1121 			if (nxt == IPPROTO_AH)
1122 				elen = (ip6e->ip6e_len + 2) << 2;
1123 			else
1124 				elen = (ip6e->ip6e_len + 1) << 3;
1125 			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off, elen);
1126 			if (ip6e == NULL) {
1127 				ip6stat.ip6s_tooshort++;
1128 				return;
1129 			}
1130 #endif
1131 
1132 			switch(nxt) {
1133 		         case IPPROTO_DSTOPTS:
1134 				 if (!in6p->in6p_flags & IN6P_DSTOPTS)
1135 					 break;
1136 
1137 				 /*
1138 				  * We also require super-user privilege for
1139 				  * the option.
1140 				  * See the comments on IN6_HOPOPTS.
1141 				  */
1142 				 if (!privileged)
1143 					 break;
1144 
1145 				 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1146 						       IPV6_DSTOPTS,
1147 						       IPPROTO_IPV6);
1148 				 if (*mp)
1149 					 mp = &(*mp)->m_next;
1150 				 break;
1151 
1152 			 case IPPROTO_ROUTING:
1153 				 if (!in6p->in6p_flags & IN6P_RTHDR)
1154 					 break;
1155 
1156 				 *mp = sbcreatecontrol((caddr_t)ip6e, elen,
1157 						       IPV6_RTHDR,
1158 						       IPPROTO_IPV6);
1159 				 if (*mp)
1160 					 mp = &(*mp)->m_next;
1161 				 break;
1162 
1163 			 case IPPROTO_UDP:
1164 			 case IPPROTO_TCP:
1165 			 case IPPROTO_ICMPV6:
1166 			 default:
1167 				 /*
1168 				  * stop search if we encounter an upper
1169 				  * layer protocol headers.
1170 				  */
1171 				 goto loopend;
1172 
1173 			 case IPPROTO_HOPOPTS:
1174 			 case IPPROTO_AH: /* is it possible? */
1175 				 break;
1176 			}
1177 
1178 			/* proceed with the next header. */
1179 			off += elen;
1180 			nxt = ip6e->ip6e_nxt;
1181 		}
1182 	  loopend:
1183 	}
1184 	if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) {
1185 		/* to be done */
1186 	}
1187 	if ((in6p->in6p_flags & IN6P_DSTOPTS) && privileged) {
1188 		/* to be done */
1189 	}
1190 	/* IN6P_RTHDR - to be done */
1191 
1192 }
1193 
1194 /*
1195  * Get pointer to the previous header followed by the header
1196  * currently processed.
1197  * XXX: This function supposes that
1198  *	M includes all headers,
1199  *	the next header field and the header length field of each header
1200  *	are valid, and
1201  *	the sum of each header length equals to OFF.
1202  * Because of these assumptions, this function must be called very
1203  * carefully. Moreover, it will not be used in the near future when
1204  * we develop `neater' mechanism to process extension headers.
1205  */
1206 char *
1207 ip6_get_prevhdr(m, off)
1208 	struct mbuf *m;
1209 	int off;
1210 {
1211 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1212 
1213 	if (off == sizeof(struct ip6_hdr))
1214 		return(&ip6->ip6_nxt);
1215 	else {
1216 		int len, nxt;
1217 		struct ip6_ext *ip6e = NULL;
1218 
1219 		nxt = ip6->ip6_nxt;
1220 		len = sizeof(struct ip6_hdr);
1221 		while (len < off) {
1222 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1223 
1224 			switch(nxt) {
1225 			case IPPROTO_FRAGMENT:
1226 				len += sizeof(struct ip6_frag);
1227 				break;
1228 			case IPPROTO_AH:
1229 				len += (ip6e->ip6e_len + 2) << 2;
1230 				break;
1231 			default:
1232 				len += (ip6e->ip6e_len + 1) << 3;
1233 				break;
1234 			}
1235 			nxt = ip6e->ip6e_nxt;
1236 		}
1237 		if (ip6e)
1238 			return(&ip6e->ip6e_nxt);
1239 		else
1240 			return NULL;
1241 	}
1242 }
1243 
1244 /*
1245  * get next header offset.  m will be retained.
1246  */
1247 int
1248 ip6_nexthdr(m, off, proto, nxtp)
1249 	struct mbuf *m;
1250 	int off;
1251 	int proto;
1252 	int *nxtp;
1253 {
1254 	struct ip6_hdr ip6;
1255 	struct ip6_ext ip6e;
1256 	struct ip6_frag fh;
1257 
1258 	/* just in case */
1259 	if (m == NULL)
1260 		panic("ip6_nexthdr: m == NULL");
1261 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1262 		return -1;
1263 
1264 	switch (proto) {
1265 	case IPPROTO_IPV6:
1266 		if (m->m_pkthdr.len < off + sizeof(ip6))
1267 			return -1;
1268 		m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1269 		if (nxtp)
1270 			*nxtp = ip6.ip6_nxt;
1271 		off += sizeof(ip6);
1272 		return off;
1273 
1274 	case IPPROTO_FRAGMENT:
1275 		/*
1276 		 * terminate parsing if it is not the first fragment,
1277 		 * it does not make sense to parse through it.
1278 		 */
1279 		if (m->m_pkthdr.len < off + sizeof(fh))
1280 			return -1;
1281 		m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1282 		if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0)
1283 			return -1;
1284 		if (nxtp)
1285 			*nxtp = fh.ip6f_nxt;
1286 		off += sizeof(struct ip6_frag);
1287 		return off;
1288 
1289 	case IPPROTO_AH:
1290 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1291 			return -1;
1292 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1293 		if (nxtp)
1294 			*nxtp = ip6e.ip6e_nxt;
1295 		off += (ip6e.ip6e_len + 2) << 2;
1296 		return off;
1297 
1298 	case IPPROTO_HOPOPTS:
1299 	case IPPROTO_ROUTING:
1300 	case IPPROTO_DSTOPTS:
1301 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1302 			return -1;
1303 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1304 		if (nxtp)
1305 			*nxtp = ip6e.ip6e_nxt;
1306 		off += (ip6e.ip6e_len + 1) << 3;
1307 		return off;
1308 
1309 	case IPPROTO_NONE:
1310 	case IPPROTO_ESP:
1311 	case IPPROTO_IPCOMP:
1312 		/* give up */
1313 		return -1;
1314 
1315 	default:
1316 		return -1;
1317 	}
1318 
1319 	return -1;
1320 }
1321 
1322 /*
1323  * get offset for the last header in the chain.  m will be kept untainted.
1324  */
1325 int
1326 ip6_lasthdr(m, off, proto, nxtp)
1327 	struct mbuf *m;
1328 	int off;
1329 	int proto;
1330 	int *nxtp;
1331 {
1332 	int newoff;
1333 	int nxt;
1334 
1335 	if (!nxtp) {
1336 		nxt = -1;
1337 		nxtp = &nxt;
1338 	}
1339 	while (1) {
1340 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1341 		if (newoff < 0)
1342 			return off;
1343 		else if (newoff < off)
1344 			return -1;	/* invalid */
1345 		else if (newoff == off)
1346 			return newoff;
1347 
1348 		off = newoff;
1349 		proto = *nxtp;
1350 	}
1351 }
1352 
1353 /*
1354  * System control for IP6
1355  */
1356 
1357 u_char	inet6ctlerrmap[PRC_NCMDS] = {
1358 	0,		0,		0,		0,
1359 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1360 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1361 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1362 	0,		0,		0,		0,
1363 	ENOPROTOOPT
1364 };
1365 
1366 #include <uvm/uvm_extern.h>
1367 #include <sys/sysctl.h>
1368 
1369 int
1370 ip6_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1371 	int *name;
1372 	u_int namelen;
1373 	void *oldp;
1374 	size_t *oldlenp;
1375 	void *newp;
1376 	size_t newlen;
1377 {
1378 	int old, error;
1379 
1380 	/* All sysctl names at this level are terminal. */
1381 	if (namelen != 1)
1382 		return ENOTDIR;
1383 
1384 	switch (name[0]) {
1385 
1386 	case IPV6CTL_FORWARDING:
1387 		return sysctl_int(oldp, oldlenp, newp, newlen,
1388 				  &ip6_forwarding);
1389 	case IPV6CTL_SENDREDIRECTS:
1390 		return sysctl_int(oldp, oldlenp, newp, newlen,
1391 				&ip6_sendredirects);
1392 	case IPV6CTL_DEFHLIM:
1393 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_defhlim);
1394 	case IPV6CTL_MAXFRAGPACKETS:
1395 		return sysctl_int(oldp, oldlenp, newp, newlen,
1396 				&ip6_maxfragpackets);
1397 	case IPV6CTL_ACCEPT_RTADV:
1398 		return sysctl_int(oldp, oldlenp, newp, newlen,
1399 				&ip6_accept_rtadv);
1400 	case IPV6CTL_KEEPFAITH:
1401 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_keepfaith);
1402 	case IPV6CTL_LOG_INTERVAL:
1403 		return sysctl_int(oldp, oldlenp, newp, newlen,
1404 				&ip6_log_interval);
1405 	case IPV6CTL_HDRNESTLIMIT:
1406 		return sysctl_int(oldp, oldlenp, newp, newlen,
1407 				&ip6_hdrnestlimit);
1408 	case IPV6CTL_DAD_COUNT:
1409 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_dad_count);
1410 	case IPV6CTL_AUTO_FLOWLABEL:
1411 		return sysctl_int(oldp, oldlenp, newp, newlen,
1412 				&ip6_auto_flowlabel);
1413 	case IPV6CTL_DEFMCASTHLIM:
1414 		return sysctl_int(oldp, oldlenp, newp, newlen,
1415 				&ip6_defmcasthlim);
1416 	case IPV6CTL_GIF_HLIM:
1417 		return sysctl_int(oldp, oldlenp, newp, newlen,
1418 				&ip6_gif_hlim);
1419 	case IPV6CTL_KAME_VERSION:
1420 		return sysctl_rdstring(oldp, oldlenp, newp, __KAME_VERSION);
1421 	case IPV6CTL_USE_DEPRECATED:
1422 		return sysctl_int(oldp, oldlenp, newp, newlen,
1423 				&ip6_use_deprecated);
1424 	case IPV6CTL_RR_PRUNE:
1425 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_rr_prune);
1426 #ifndef INET6_BINDV6ONLY
1427 	case IPV6CTL_BINDV6ONLY:
1428 		return sysctl_int(oldp, oldlenp, newp, newlen,
1429 				&ip6_bindv6only);
1430 #endif
1431 	case IPV6CTL_ANONPORTMIN:
1432 		old = ip6_anonportmin;
1433 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1434 		    &ip6_anonportmin);
1435 		if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmin < 0 ||
1436 		    ip6_anonportmin > 65535
1437 #ifndef IPNOPRIVPORTS
1438 		    || ip6_anonportmin < IPV6PORT_RESERVED
1439 #endif
1440 		    ) {
1441 			ip6_anonportmin = old;
1442 			return (EINVAL);
1443 		}
1444 		return (error);
1445 	case IPV6CTL_ANONPORTMAX:
1446 		old = ip6_anonportmax;
1447 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1448 		    &ip6_anonportmax);
1449 		if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmax < 0 ||
1450 		    ip6_anonportmax > 65535
1451 #ifndef IPNOPRIVPORTS
1452 		    || ip6_anonportmax < IPV6PORT_RESERVED
1453 #endif
1454 		    ) {
1455 			ip6_anonportmax = old;
1456 			return (EINVAL);
1457 		}
1458 		return (error);
1459 #ifndef IPNOPRIVPORTS
1460 	case IPV6CTL_LOWPORTMIN:
1461 		old = ip6_lowportmin;
1462 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1463 		    &ip6_lowportmin);
1464 		if (ip6_lowportmin >= ip6_lowportmax ||
1465 		    ip6_lowportmin > IPV6PORT_RESERVEDMAX ||
1466 		    ip6_lowportmin < IPV6PORT_RESERVEDMIN) {
1467 			ip6_lowportmin = old;
1468 			return (EINVAL);
1469 		}
1470 		return (error);
1471 	case IPV6CTL_LOWPORTMAX:
1472 		old = ip6_lowportmax;
1473 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1474 		    &ip6_lowportmax);
1475 		if (ip6_lowportmin >= ip6_lowportmax ||
1476 		    ip6_lowportmax > IPV6PORT_RESERVEDMAX ||
1477 		    ip6_lowportmax < IPV6PORT_RESERVEDMIN) {
1478 			ip6_lowportmax = old;
1479 			return (EINVAL);
1480 		}
1481 		return (error);
1482 #endif
1483 	default:
1484 		return EOPNOTSUPP;
1485 	}
1486 	/* NOTREACHED */
1487 }
1488