xref: /netbsd-src/sys/netinet6/ip6_input.c (revision 220b5c059a84c51ea44107ea8951a57ffaecdc8c)
1 /*	$NetBSD: ip6_input.c,v 1.48 2001/11/13 00:57:01 lukem 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 <sys/cdefs.h>
69 __KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.48 2001/11/13 00:57:01 lukem Exp $");
70 
71 #include "opt_inet.h"
72 #include "opt_ipsec.h"
73 #include "opt_pfil_hooks.h"
74 
75 #include <sys/param.h>
76 #include <sys/systm.h>
77 #include <sys/malloc.h>
78 #include <sys/mbuf.h>
79 #include <sys/domain.h>
80 #include <sys/protosw.h>
81 #include <sys/socket.h>
82 #include <sys/socketvar.h>
83 #include <sys/errno.h>
84 #include <sys/time.h>
85 #include <sys/kernel.h>
86 #include <sys/syslog.h>
87 #include <sys/proc.h>
88 #include <sys/sysctl.h>
89 
90 #include <net/if.h>
91 #include <net/if_types.h>
92 #include <net/if_dl.h>
93 #include <net/route.h>
94 #include <net/netisr.h>
95 #ifdef PFIL_HOOKS
96 #include <net/pfil.h>
97 #endif
98 
99 #include <netinet/in.h>
100 #include <netinet/in_systm.h>
101 #ifdef INET
102 #include <netinet/ip.h>
103 #include <netinet/ip_icmp.h>
104 #endif /* INET */
105 #include <netinet/ip6.h>
106 #include <netinet6/in6_var.h>
107 #include <netinet6/ip6_var.h>
108 #include <netinet6/in6_pcb.h>
109 #include <netinet/icmp6.h>
110 #include <netinet6/in6_ifattach.h>
111 #include <netinet6/nd6.h>
112 #include <netinet6/in6_prefix.h>
113 
114 #ifdef IPSEC
115 #include <netinet6/ipsec.h>
116 #endif
117 
118 #include <netinet6/ip6protosw.h>
119 
120 /* we need it for NLOOP. */
121 #include "loop.h"
122 #include "faith.h"
123 #include "gif.h"
124 #include "bpfilter.h"
125 
126 #include <net/net_osdep.h>
127 
128 extern struct domain inet6domain;
129 
130 u_char ip6_protox[IPPROTO_MAX];
131 static int ip6qmaxlen = IFQ_MAXLEN;
132 struct in6_ifaddr *in6_ifaddr;
133 struct ifqueue ip6intrq;
134 
135 extern struct ifnet loif[NLOOP];
136 int ip6_forward_srcrt;			/* XXX */
137 int ip6_sourcecheck;			/* XXX */
138 int ip6_sourcecheck_interval;		/* XXX */
139 
140 #ifdef PFIL_HOOKS
141 struct pfil_head inet6_pfil_hook;
142 #endif
143 
144 struct ip6stat ip6stat;
145 
146 static void ip6_init2 __P((void *));
147 
148 static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
149 
150 /*
151  * IP6 initialization: fill in IP6 protocol switch table.
152  * All protocols not implemented in kernel go to raw IP6 protocol handler.
153  */
154 void
155 ip6_init()
156 {
157 	struct ip6protosw *pr;
158 	int i;
159 	struct timeval tv;
160 
161 	pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
162 	if (pr == 0)
163 		panic("ip6_init");
164 	for (i = 0; i < IPPROTO_MAX; i++)
165 		ip6_protox[i] = pr - inet6sw;
166 	for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
167 	    pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
168 		if (pr->pr_domain->dom_family == PF_INET6 &&
169 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
170 			ip6_protox[pr->pr_protocol] = pr - inet6sw;
171 	ip6intrq.ifq_maxlen = ip6qmaxlen;
172 	nd6_init();
173 	frag6_init();
174 	/*
175 	 * in many cases, random() here does NOT return random number
176 	 * as initialization during bootstrap time occur in fixed order.
177 	 */
178 	microtime(&tv);
179 	ip6_flow_seq = random() ^ tv.tv_usec;
180 
181 	ip6_init2((void *)0);
182 
183 #ifdef PFIL_HOOKS
184 	/* Register our Packet Filter hook. */
185 	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
186 	inet6_pfil_hook.ph_af   = AF_INET6;
187 	i = pfil_head_register(&inet6_pfil_hook);
188 	if (i != 0)
189 		printf("ip6_init: WARNING: unable to register pfil hook, "
190 		    "error %d\n", i);
191 #endif /* PFIL_HOOKS */
192 }
193 
194 static void
195 ip6_init2(dummy)
196 	void *dummy;
197 {
198 	/*
199 	 * to route local address of p2p link to loopback,
200 	 * assign loopback address first.
201 	 */
202 	in6_ifattach(&loif[0], NULL);
203 
204 	/* nd6_timer_init */
205 	callout_init(&nd6_timer_ch);
206 	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
207 	/* router renumbering prefix list maintenance */
208 	callout_init(&in6_rr_timer_ch);
209 	callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
210 }
211 
212 /*
213  * IP6 input interrupt handling. Just pass the packet to ip6_input.
214  */
215 void
216 ip6intr()
217 {
218 	int s;
219 	struct mbuf *m;
220 
221 	for (;;) {
222 		s = splnet();
223 		IF_DEQUEUE(&ip6intrq, m);
224 		splx(s);
225 		if (m == 0)
226 			return;
227 		ip6_input(m);
228 	}
229 }
230 
231 extern struct	route_in6 ip6_forward_rt;
232 
233 void
234 ip6_input(m)
235 	struct mbuf *m;
236 {
237 	struct ip6_hdr *ip6;
238 	int off = sizeof(struct ip6_hdr), nest;
239 	u_int32_t plen;
240 	u_int32_t rtalert = ~0;
241 	int nxt, ours = 0;
242 	struct ifnet *deliverifp = NULL;
243 
244 #ifdef IPSEC
245 	/*
246 	 * should the inner packet be considered authentic?
247 	 * see comment in ah4_input().
248 	 */
249 	if (m) {
250 		m->m_flags &= ~M_AUTHIPHDR;
251 		m->m_flags &= ~M_AUTHIPDGM;
252 	}
253 #endif
254 
255 	/*
256 	 * mbuf statistics
257 	 */
258 	if (m->m_flags & M_EXT) {
259 		if (m->m_next)
260 			ip6stat.ip6s_mext2m++;
261 		else
262 			ip6stat.ip6s_mext1++;
263 	} else {
264 #define M2MMAX	(sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
265 		if (m->m_next) {
266 			if (m->m_flags & M_LOOP) {
267 				ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
268 			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
269 				ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
270 			else
271 				ip6stat.ip6s_m2m[0]++;
272 		} else
273 			ip6stat.ip6s_m1++;
274 #undef M2MMAX
275 	}
276 
277 	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
278 	ip6stat.ip6s_total++;
279 
280 #ifndef PULLDOWN_TEST
281 	/* XXX is the line really necessary? */
282 	IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/);
283 #endif
284 
285 	if (m->m_len < sizeof(struct ip6_hdr)) {
286 		struct ifnet *inifp;
287 		inifp = m->m_pkthdr.rcvif;
288 		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
289 			ip6stat.ip6s_toosmall++;
290 			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
291 			return;
292 		}
293 	}
294 
295 	ip6 = mtod(m, struct ip6_hdr *);
296 
297 	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
298 		ip6stat.ip6s_badvers++;
299 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
300 		goto bad;
301 	}
302 
303 #ifdef PFIL_HOOKS
304 	/*
305 	 * Run through list of hooks for input packets.  If there are any
306 	 * filters which require that additional packets in the flow are
307 	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
308 	 * Note that filters must _never_ set this flag, as another filter
309 	 * in the list may have previously cleared it.
310 	 */
311 	/*
312 	 * let ipfilter look at packet on the wire,
313 	 * not the decapsulated packet.
314 	 */
315 #ifdef IPSEC
316 	if (!ipsec_getnhist(m))
317 #else
318 	if (1)
319 #endif
320 	{
321 		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
322 				   PFIL_IN) != 0)
323 			return;
324 		if (m == NULL)
325 			return;
326 		ip6 = mtod(m, struct ip6_hdr *);
327 	}
328 #endif /* PFIL_HOOKS */
329 
330 	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
331 
332 #ifdef ALTQ
333 	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
334 		/* packet is dropped by traffic conditioner */
335 		return;
336 	}
337 #endif
338 
339 	/*
340 	 * Check against address spoofing/corruption.
341 	 */
342 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
343 	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
344 		ip6stat.ip6s_badscope++;
345 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
346 		goto bad;
347 	}
348 	/*
349 	 * The following check is not documented in specs.  A malicious
350 	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
351 	 * and bypass security checks (act as if it was from 127.0.0.1 by using
352 	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
353 	 *
354 	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
355 	 * support IPv4-less kernel compilation, we cannot support SIIT
356 	 * environment at all.  So, it makes more sense for us to reject any
357 	 * malicious packets for non-SIIT environment, than try to do a
358 	 * partical support for SIIT environment.
359 	 */
360 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
361 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
362 		ip6stat.ip6s_badscope++;
363 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
364 		goto bad;
365 	}
366 #if 0
367 	/*
368 	 * Reject packets with IPv4 compatible addresses (auto tunnel).
369 	 *
370 	 * The code forbids auto tunnel relay case in RFC1933 (the check is
371 	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
372 	 * is revised to forbid relaying case.
373 	 */
374 	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
375 	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
376 		ip6stat.ip6s_badscope++;
377 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
378 		goto bad;
379 	}
380 #endif
381 
382 	if (IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
383 	    IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) {
384 		if (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) {
385 			ours = 1;
386 			deliverifp = m->m_pkthdr.rcvif;
387 			goto hbhcheck;
388 		} else {
389 			ip6stat.ip6s_badscope++;
390 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
391 			goto bad;
392 		}
393 	}
394 
395 	/* drop packets if interface ID portion is already filled */
396 	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
397 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
398 		    ip6->ip6_src.s6_addr16[1]) {
399 			ip6stat.ip6s_badscope++;
400 			goto bad;
401 		}
402 		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) &&
403 		    ip6->ip6_dst.s6_addr16[1]) {
404 			ip6stat.ip6s_badscope++;
405 			goto bad;
406 		}
407 	}
408 
409 	if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
410 		ip6->ip6_src.s6_addr16[1]
411 			= htons(m->m_pkthdr.rcvif->if_index);
412 	if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
413 		ip6->ip6_dst.s6_addr16[1]
414 			= htons(m->m_pkthdr.rcvif->if_index);
415 
416 	/*
417 	 * We use rt->rt_ifp to determine if the address is ours or not.
418 	 * If rt_ifp is lo0, the address is ours.
419 	 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0,
420 	 * so any address under fe80::%lo0/64 will be mistakenly considered
421 	 * local.  The special case is supplied to handle the case properly
422 	 * by actually looking at interface addresses
423 	 * (using in6ifa_ifpwithaddr).
424 	 */
425 	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 &&
426 	    IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) {
427 		if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) {
428 			icmp6_error(m, ICMP6_DST_UNREACH,
429 			    ICMP6_DST_UNREACH_ADDR, 0);
430 			/* m is already freed */
431 			return;
432 		}
433 
434 		ours = 1;
435 		deliverifp = m->m_pkthdr.rcvif;
436 		goto hbhcheck;
437 	}
438 
439 	/*
440 	 * Multicast check
441 	 */
442 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
443 	  	struct	in6_multi *in6m = 0;
444 
445 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
446 		/*
447 		 * See if we belong to the destination multicast group on the
448 		 * arrival interface.
449 		 */
450 		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
451 		if (in6m)
452 			ours = 1;
453 		else if (!ip6_mrouter) {
454 			ip6stat.ip6s_notmember++;
455 			ip6stat.ip6s_cantforward++;
456 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
457 			goto bad;
458 		}
459 		deliverifp = m->m_pkthdr.rcvif;
460 		goto hbhcheck;
461 	}
462 
463 	/*
464 	 *  Unicast check
465 	 */
466 	if (ip6_forward_rt.ro_rt != NULL &&
467 	    (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
468 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
469 			       &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
470 		ip6stat.ip6s_forward_cachehit++;
471 	else {
472 		struct sockaddr_in6 *dst6;
473 
474 		if (ip6_forward_rt.ro_rt) {
475 			/* route is down or destination is different */
476 			ip6stat.ip6s_forward_cachemiss++;
477 			RTFREE(ip6_forward_rt.ro_rt);
478 			ip6_forward_rt.ro_rt = 0;
479 		}
480 
481 		bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
482 		dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
483 		dst6->sin6_len = sizeof(struct sockaddr_in6);
484 		dst6->sin6_family = AF_INET6;
485 		dst6->sin6_addr = ip6->ip6_dst;
486 
487 		rtalloc((struct route *)&ip6_forward_rt);
488 	}
489 
490 #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
491 
492 	/*
493 	 * Accept the packet if the forwarding interface to the destination
494 	 * according to the routing table is the loopback interface,
495 	 * unless the associated route has a gateway.
496 	 * Note that this approach causes to accept a packet if there is a
497 	 * route to the loopback interface for the destination of the packet.
498 	 * But we think it's even useful in some situations, e.g. when using
499 	 * a special daemon which wants to intercept the packet.
500 	 */
501 	if (ip6_forward_rt.ro_rt &&
502 	    (ip6_forward_rt.ro_rt->rt_flags &
503 	     (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
504 #if 0
505 	    /*
506 	     * The check below is redundant since the comparison of
507 	     * the destination and the key of the rtentry has
508 	     * already done through looking up the routing table.
509 	     */
510 	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
511 			       &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) &&
512 #endif
513 	    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
514 		struct in6_ifaddr *ia6 =
515 			(struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
516 		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
517 			m->m_flags |= M_ANYCAST6;
518 		/*
519 		 * packets to a tentative, duplicated, or somehow invalid
520 		 * address must not be accepted.
521 		 */
522 		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
523 			/* this address is ready */
524 			ours = 1;
525 			deliverifp = ia6->ia_ifp;	/* correct? */
526 			goto hbhcheck;
527 		} else {
528 			/* address is not ready, so discard the packet. */
529 			nd6log((LOG_INFO,
530 			    "ip6_input: packet to an unready address %s->%s\n",
531 			    ip6_sprintf(&ip6->ip6_src),
532 			    ip6_sprintf(&ip6->ip6_dst)));
533 
534 			goto bad;
535 		}
536 	}
537 
538 	/*
539 	 * FAITH(Firewall Aided Internet Translator)
540 	 */
541 #if defined(NFAITH) && 0 < NFAITH
542 	if (ip6_keepfaith) {
543 		if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
544 		 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
545 			/* XXX do we need more sanity checks? */
546 			ours = 1;
547 			deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
548 			goto hbhcheck;
549 		}
550 	}
551 #endif
552 
553 #if 0
554     {
555 	/*
556 	 * Last resort: check in6_ifaddr for incoming interface.
557 	 * The code is here until I update the "goto ours hack" code above
558 	 * working right.
559 	 */
560 	struct ifaddr *ifa;
561 	for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
562 	     ifa;
563 	     ifa = ifa->ifa_list.tqe_next) {
564 		if (ifa->ifa_addr == NULL)
565 			continue;	/* just for safety */
566 		if (ifa->ifa_addr->sa_family != AF_INET6)
567 			continue;
568 		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
569 			ours = 1;
570 			deliverifp = ifa->ifa_ifp;
571 			goto hbhcheck;
572 		}
573 	}
574     }
575 #endif
576 
577 	/*
578 	 * Now there is no reason to process the packet if it's not our own
579 	 * and we're not a router.
580 	 */
581 	if (!ip6_forwarding) {
582 		ip6stat.ip6s_cantforward++;
583 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
584 		goto bad;
585 	}
586 
587   hbhcheck:
588 	/*
589 	 * Process Hop-by-Hop options header if it's contained.
590 	 * m may be modified in ip6_hopopts_input().
591 	 * If a JumboPayload option is included, plen will also be modified.
592 	 */
593 	plen = (u_int32_t)ntohs(ip6->ip6_plen);
594 	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
595 		struct ip6_hbh *hbh;
596 
597 		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
598 #if 0	/*touches NULL pointer*/
599 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
600 #endif
601 			return;	/* m have already been freed */
602 		}
603 
604 		/* adjust pointer */
605 		ip6 = mtod(m, struct ip6_hdr *);
606 
607 		/*
608 		 * if the payload length field is 0 and the next header field
609 		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
610 		 * option MUST be included.
611 		 */
612 		if (ip6->ip6_plen == 0 && plen == 0) {
613 			/*
614 			 * Note that if a valid jumbo payload option is
615 			 * contained, ip6_hoptops_input() must set a valid
616 			 * (non-zero) payload length to the variable plen.
617 			 */
618 			ip6stat.ip6s_badoptions++;
619 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
620 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
621 			icmp6_error(m, ICMP6_PARAM_PROB,
622 				    ICMP6_PARAMPROB_HEADER,
623 				    (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
624 			return;
625 		}
626 #ifndef PULLDOWN_TEST
627 		/* ip6_hopopts_input() ensures that mbuf is contiguous */
628 		hbh = (struct ip6_hbh *)(ip6 + 1);
629 #else
630 		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
631 			sizeof(struct ip6_hbh));
632 		if (hbh == NULL) {
633 			ip6stat.ip6s_tooshort++;
634 			return;
635 		}
636 #endif
637 		nxt = hbh->ip6h_nxt;
638 
639 		/*
640 		 * accept the packet if a router alert option is included
641 		 * and we act as an IPv6 router.
642 		 */
643 		if (rtalert != ~0 && ip6_forwarding)
644 			ours = 1;
645 	} else
646 		nxt = ip6->ip6_nxt;
647 
648 	/*
649 	 * Check that the amount of data in the buffers
650 	 * is as at least much as the IPv6 header would have us expect.
651 	 * Trim mbufs if longer than we expect.
652 	 * Drop packet if shorter than we expect.
653 	 */
654 	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
655 		ip6stat.ip6s_tooshort++;
656 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
657 		goto bad;
658 	}
659 	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
660 		if (m->m_len == m->m_pkthdr.len) {
661 			m->m_len = sizeof(struct ip6_hdr) + plen;
662 			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
663 		} else
664 			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
665 	}
666 
667 	/*
668 	 * Forward if desirable.
669 	 */
670 	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
671 		/*
672 		 * If we are acting as a multicast router, all
673 		 * incoming multicast packets are passed to the
674 		 * kernel-level multicast forwarding function.
675 		 * The packet is returned (relatively) intact; if
676 		 * ip6_mforward() returns a non-zero value, the packet
677 		 * must be discarded, else it may be accepted below.
678 		 */
679 		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
680 			ip6stat.ip6s_cantforward++;
681 			m_freem(m);
682 			return;
683 		}
684 		if (!ours) {
685 			m_freem(m);
686 			return;
687 		}
688 	} else if (!ours) {
689 		ip6_forward(m, 0);
690 		return;
691 	}
692 
693 	ip6 = mtod(m, struct ip6_hdr *);
694 
695 	/*
696 	 * Malicious party may be able to use IPv4 mapped addr to confuse
697 	 * tcp/udp stack and bypass security checks (act as if it was from
698 	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
699 	 *
700 	 * For SIIT end node behavior, you may want to disable the check.
701 	 * However, you will  become vulnerable to attacks using IPv4 mapped
702 	 * source.
703 	 */
704 	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
705 	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
706 		ip6stat.ip6s_badscope++;
707 		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
708 		goto bad;
709 	}
710 
711 	/*
712 	 * Tell launch routine the next header
713 	 */
714 #ifdef IFA_STATS
715 	if (deliverifp != NULL) {
716 		struct in6_ifaddr *ia6;
717 		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
718 		if (ia6)
719 			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
720 	}
721 #endif
722 	ip6stat.ip6s_delivered++;
723 	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
724 	nest = 0;
725 
726 	while (nxt != IPPROTO_DONE) {
727 		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
728 			ip6stat.ip6s_toomanyhdr++;
729 			goto bad;
730 		}
731 
732 		/*
733 		 * protection against faulty packet - there should be
734 		 * more sanity checks in header chain processing.
735 		 */
736 		if (m->m_pkthdr.len < off) {
737 			ip6stat.ip6s_tooshort++;
738 			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
739 			goto bad;
740 		}
741 
742 #ifdef IPSEC
743 		/*
744 		 * enforce IPsec policy checking if we are seeing last header.
745 		 * note that we do not visit this with protocols with pcb layer
746 		 * code - like udp/tcp/raw ip.
747 		 */
748 		if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
749 		    ipsec6_in_reject(m, NULL)) {
750 			ipsec6stat.in_polvio++;
751 			goto bad;
752 		}
753 #endif
754 
755 		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
756 	}
757 	return;
758  bad:
759 	m_freem(m);
760 }
761 
762 /*
763  * Hop-by-Hop options header processing. If a valid jumbo payload option is
764  * included, the real payload length will be stored in plenp.
765  */
766 static int
767 ip6_hopopts_input(plenp, rtalertp, mp, offp)
768 	u_int32_t *plenp;
769 	u_int32_t *rtalertp;	/* XXX: should be stored more smart way */
770 	struct mbuf **mp;
771 	int *offp;
772 {
773 	struct mbuf *m = *mp;
774 	int off = *offp, hbhlen;
775 	struct ip6_hbh *hbh;
776 	u_int8_t *opt;
777 
778 	/* validation of the length of the header */
779 #ifndef PULLDOWN_TEST
780 	IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
781 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
782 	hbhlen = (hbh->ip6h_len + 1) << 3;
783 
784 	IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
785 	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
786 #else
787 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
788 		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
789 	if (hbh == NULL) {
790 		ip6stat.ip6s_tooshort++;
791 		return -1;
792 	}
793 	hbhlen = (hbh->ip6h_len + 1) << 3;
794 	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
795 		hbhlen);
796 	if (hbh == NULL) {
797 		ip6stat.ip6s_tooshort++;
798 		return -1;
799 	}
800 #endif
801 	off += hbhlen;
802 	hbhlen -= sizeof(struct ip6_hbh);
803 	opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
804 
805 	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
806 				hbhlen, rtalertp, plenp) < 0)
807 		return(-1);
808 
809 	*offp = off;
810 	*mp = m;
811 	return(0);
812 }
813 
814 /*
815  * Search header for all Hop-by-hop options and process each option.
816  * This function is separate from ip6_hopopts_input() in order to
817  * handle a case where the sending node itself process its hop-by-hop
818  * options header. In such a case, the function is called from ip6_output().
819  */
820 int
821 ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
822 	struct mbuf *m;
823 	u_int8_t *opthead;
824 	int hbhlen;
825 	u_int32_t *rtalertp;
826 	u_int32_t *plenp;
827 {
828 	struct ip6_hdr *ip6;
829 	int optlen = 0;
830 	u_int8_t *opt = opthead;
831 	u_int16_t rtalert_val;
832 	u_int32_t jumboplen;
833 
834 	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
835 		switch (*opt) {
836 		case IP6OPT_PAD1:
837 			optlen = 1;
838 			break;
839 		case IP6OPT_PADN:
840 			if (hbhlen < IP6OPT_MINLEN) {
841 				ip6stat.ip6s_toosmall++;
842 				goto bad;
843 			}
844 			optlen = *(opt + 1) + 2;
845 			break;
846 		case IP6OPT_RTALERT:
847 			/* XXX may need check for alignment */
848 			if (hbhlen < IP6OPT_RTALERT_LEN) {
849 				ip6stat.ip6s_toosmall++;
850 				goto bad;
851 			}
852 			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
853 				 /* XXX: should we discard the packet? */
854 				log(LOG_ERR, "length of router alert opt is inconsitent(%d)",
855 				    *(opt + 1));
856 			}
857 			optlen = IP6OPT_RTALERT_LEN;
858 			bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
859 			*rtalertp = ntohs(rtalert_val);
860 			break;
861 		case IP6OPT_JUMBO:
862 			/* XXX may need check for alignment */
863 			if (hbhlen < IP6OPT_JUMBO_LEN) {
864 				ip6stat.ip6s_toosmall++;
865 				goto bad;
866 			}
867 			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
868 				/* XXX: should we discard the packet? */
869 				log(LOG_ERR, "length of jumbopayload opt "
870 				    "is inconsistent(%d)\n",
871 				    *(opt + 1));
872 			}
873 			optlen = IP6OPT_JUMBO_LEN;
874 
875 			/*
876 			 * IPv6 packets that have non 0 payload length
877 			 * must not contain a jumbo payload option.
878 			 */
879 			ip6 = mtod(m, struct ip6_hdr *);
880 			if (ip6->ip6_plen) {
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 - opthead);
887 				return(-1);
888 			}
889 
890 			/*
891 			 * We may see jumbolen in unaligned location, so
892 			 * we'd need to perform bcopy().
893 			 */
894 			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
895 			jumboplen = (u_int32_t)htonl(jumboplen);
896 
897 #if 1
898 			/*
899 			 * if there are multiple jumbo payload options,
900 			 * *plenp will be non-zero and the packet will be
901 			 * rejected.
902 			 * the behavior may need some debate in ipngwg -
903 			 * multiple options does not make sense, however,
904 			 * there's no explicit mention in specification.
905 			 */
906 			if (*plenp != 0) {
907 				ip6stat.ip6s_badoptions++;
908 				icmp6_error(m, ICMP6_PARAM_PROB,
909 					    ICMP6_PARAMPROB_HEADER,
910 					    sizeof(struct ip6_hdr) +
911 					    sizeof(struct ip6_hbh) +
912 					    opt + 2 - opthead);
913 				return(-1);
914 			}
915 #endif
916 
917 			/*
918 			 * jumbo payload length must be larger than 65535.
919 			 */
920 			if (jumboplen <= IPV6_MAXPACKET) {
921 				ip6stat.ip6s_badoptions++;
922 				icmp6_error(m, ICMP6_PARAM_PROB,
923 					    ICMP6_PARAMPROB_HEADER,
924 					    sizeof(struct ip6_hdr) +
925 					    sizeof(struct ip6_hbh) +
926 					    opt + 2 - opthead);
927 				return(-1);
928 			}
929 			*plenp = jumboplen;
930 
931 			break;
932 		default:		/* unknown option */
933 			if (hbhlen < IP6OPT_MINLEN) {
934 				ip6stat.ip6s_toosmall++;
935 				goto bad;
936 			}
937 			if ((optlen = ip6_unknown_opt(opt, m,
938 						      sizeof(struct ip6_hdr) +
939 						      sizeof(struct ip6_hbh) +
940 						      opt - opthead)) == -1)
941 				return(-1);
942 			optlen += 2;
943 			break;
944 		}
945 	}
946 
947 	return(0);
948 
949   bad:
950 	m_freem(m);
951 	return(-1);
952 }
953 
954 /*
955  * Unknown option processing.
956  * The third argument `off' is the offset from the IPv6 header to the option,
957  * which is necessary if the IPv6 header the and option header and IPv6 header
958  * is not continuous in order to return an ICMPv6 error.
959  */
960 int
961 ip6_unknown_opt(optp, m, off)
962 	u_int8_t *optp;
963 	struct mbuf *m;
964 	int off;
965 {
966 	struct ip6_hdr *ip6;
967 
968 	switch (IP6OPT_TYPE(*optp)) {
969 	case IP6OPT_TYPE_SKIP: /* ignore the option */
970 		return((int)*(optp + 1));
971 	case IP6OPT_TYPE_DISCARD:	/* silently discard */
972 		m_freem(m);
973 		return(-1);
974 	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
975 		ip6stat.ip6s_badoptions++;
976 		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
977 		return(-1);
978 	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
979 		ip6stat.ip6s_badoptions++;
980 		ip6 = mtod(m, struct ip6_hdr *);
981 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
982 		    (m->m_flags & (M_BCAST|M_MCAST)))
983 			m_freem(m);
984 		else
985 			icmp6_error(m, ICMP6_PARAM_PROB,
986 				    ICMP6_PARAMPROB_OPTION, off);
987 		return(-1);
988 	}
989 
990 	m_freem(m);		/* XXX: NOTREACHED */
991 	return(-1);
992 }
993 
994 /*
995  * Create the "control" list for this pcb.
996  *
997  * The routine will be called from upper layer handlers like tcp6_input().
998  * Thus the routine assumes that the caller (tcp6_input) have already
999  * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1000  * very first mbuf on the mbuf chain.
1001  * We may want to add some infinite loop prevention or sanity checks for safety.
1002  * (This applies only when you are using KAME mbuf chain restriction, i.e.
1003  * you are using IP6_EXTHDR_CHECK() not m_pulldown())
1004  */
1005 void
1006 ip6_savecontrol(in6p, mp, ip6, m)
1007 	struct in6pcb *in6p;
1008 	struct mbuf **mp;
1009 	struct ip6_hdr *ip6;
1010 	struct mbuf *m;
1011 {
1012 	struct proc *p = curproc;	/* XXX */
1013 	int privileged;
1014 
1015 	privileged = 0;
1016 	if (p && !suser(p->p_ucred, &p->p_acflag))
1017 		privileged++;
1018 
1019 #ifdef SO_TIMESTAMP
1020 	if (in6p->in6p_socket->so_options & SO_TIMESTAMP) {
1021 		struct timeval tv;
1022 
1023 		microtime(&tv);
1024 		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1025 			SCM_TIMESTAMP, SOL_SOCKET);
1026 		if (*mp)
1027 			mp = &(*mp)->m_next;
1028 	}
1029 #endif
1030 	if (in6p->in6p_flags & IN6P_RECVDSTADDR) {
1031 		*mp = sbcreatecontrol((caddr_t) &ip6->ip6_dst,
1032 			sizeof(struct in6_addr), IPV6_RECVDSTADDR,
1033 			IPPROTO_IPV6);
1034 		if (*mp)
1035 			mp = &(*mp)->m_next;
1036 	}
1037 
1038 #ifdef noyet
1039 	/* options were tossed above */
1040 	if (in6p->in6p_flags & IN6P_RECVOPTS)
1041 		/* broken */
1042 	/* ip6_srcroute doesn't do what we want here, need to fix */
1043 	if (in6p->in6p_flags & IPV6P_RECVRETOPTS)
1044 		/* broken */
1045 #endif
1046 
1047 	/* RFC 2292 sec. 5 */
1048 	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1049 		struct in6_pktinfo pi6;
1050 		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1051 		if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
1052 			pi6.ipi6_addr.s6_addr16[1] = 0;
1053 		pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
1054 					? m->m_pkthdr.rcvif->if_index
1055 					: 0;
1056 		*mp = sbcreatecontrol((caddr_t) &pi6,
1057 			sizeof(struct in6_pktinfo), IPV6_PKTINFO,
1058 			IPPROTO_IPV6);
1059 		if (*mp)
1060 			mp = &(*mp)->m_next;
1061 	}
1062 	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1063 		int hlim = ip6->ip6_hlim & 0xff;
1064 		*mp = sbcreatecontrol((caddr_t) &hlim,
1065 			sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6);
1066 		if (*mp)
1067 			mp = &(*mp)->m_next;
1068 	}
1069 	/* IN6P_NEXTHOP - for outgoing packet only */
1070 
1071 	/*
1072 	 * IPV6_HOPOPTS socket option. We require super-user privilege
1073 	 * for the option, but it might be too strict, since there might
1074 	 * be some hop-by-hop options which can be returned to normal user.
1075 	 * See RFC 2292 section 6.
1076 	 */
1077 	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) {
1078 		/*
1079 		 * Check if a hop-by-hop options header is contatined in the
1080 		 * received packet, and if so, store the options as ancillary
1081 		 * data. Note that a hop-by-hop options header must be
1082 		 * just after the IPv6 header, which fact is assured through
1083 		 * the IPv6 input processing.
1084 		 */
1085 		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1086 		if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1087 			struct ip6_hbh *hbh;
1088 			int hbhlen;
1089 
1090 #ifndef PULLDOWN_TEST
1091 			hbh = (struct ip6_hbh *)(ip6 + 1);
1092 			hbhlen = (hbh->ip6h_len + 1) << 3;
1093 #else
1094 			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1095 				sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
1096 			if (hbh == NULL) {
1097 				ip6stat.ip6s_tooshort++;
1098 				return;
1099 			}
1100 			hbhlen = (hbh->ip6h_len + 1) << 3;
1101 			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1102 				sizeof(struct ip6_hdr), hbhlen);
1103 			if (hbh == NULL) {
1104 				ip6stat.ip6s_tooshort++;
1105 				return;
1106 			}
1107 #endif
1108 
1109 			/*
1110 			 * XXX: We copy whole the header even if a jumbo
1111 			 * payload option is included, which option is to
1112 			 * be removed before returning in the RFC 2292.
1113 			 * But it's too painful operation...
1114 			 */
1115 			*mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1116 					      IPV6_HOPOPTS, IPPROTO_IPV6);
1117 			if (*mp)
1118 				mp = &(*mp)->m_next;
1119 		}
1120 	}
1121 
1122 	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1123 	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1124 		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1125 		int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);;
1126 
1127 		/*
1128 		 * Search for destination options headers or routing
1129 		 * header(s) through the header chain, and stores each
1130 		 * header as ancillary data.
1131 		 * Note that the order of the headers remains in
1132 		 * the chain of ancillary data.
1133 		 */
1134 		while (1) {	/* is explicit loop prevention necessary? */
1135 			struct ip6_ext *ip6e;
1136 			int elen;
1137 
1138 #ifndef PULLDOWN_TEST
1139 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1140 			if (nxt == IPPROTO_AH)
1141 				elen = (ip6e->ip6e_len + 2) << 2;
1142 			else
1143 				elen = (ip6e->ip6e_len + 1) << 3;
1144 #else
1145 			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off,
1146 				sizeof(struct ip6_ext));
1147 			if (ip6e == NULL) {
1148 				ip6stat.ip6s_tooshort++;
1149 				return;
1150 			}
1151 			if (nxt == IPPROTO_AH)
1152 				elen = (ip6e->ip6e_len + 2) << 2;
1153 			else
1154 				elen = (ip6e->ip6e_len + 1) << 3;
1155 			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off, elen);
1156 			if (ip6e == NULL) {
1157 				ip6stat.ip6s_tooshort++;
1158 				return;
1159 			}
1160 #endif
1161 
1162 			switch (nxt) {
1163 		        case IPPROTO_DSTOPTS:
1164 				if (!in6p->in6p_flags & IN6P_DSTOPTS)
1165 					break;
1166 
1167 				/*
1168 				 * We also require super-user privilege for
1169 				 * the option.
1170 				 * See the comments on IN6_HOPOPTS.
1171 				 */
1172 				if (!privileged)
1173 					break;
1174 
1175 				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
1176 						      IPV6_DSTOPTS,
1177 						      IPPROTO_IPV6);
1178 				if (*mp)
1179 					mp = &(*mp)->m_next;
1180 				break;
1181 
1182 			case IPPROTO_ROUTING:
1183 				if (!in6p->in6p_flags & IN6P_RTHDR)
1184 					break;
1185 
1186 				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
1187 						      IPV6_RTHDR,
1188 						      IPPROTO_IPV6);
1189 				if (*mp)
1190 					mp = &(*mp)->m_next;
1191 				break;
1192 
1193 			case IPPROTO_UDP:
1194 			case IPPROTO_TCP:
1195 			case IPPROTO_ICMPV6:
1196 			default:
1197 				/*
1198 				 * stop search if we encounter an upper
1199 				 * layer protocol headers.
1200 				 */
1201 				goto loopend;
1202 
1203 			case IPPROTO_HOPOPTS:
1204 			case IPPROTO_AH: /* is it possible? */
1205 				break;
1206 			}
1207 
1208 			/* proceed with the next header. */
1209 			off += elen;
1210 			nxt = ip6e->ip6e_nxt;
1211 		}
1212 	  loopend:
1213 		;
1214 	}
1215 	if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) {
1216 		/* to be done */
1217 	}
1218 	if ((in6p->in6p_flags & IN6P_DSTOPTS) && privileged) {
1219 		/* to be done */
1220 	}
1221 	/* IN6P_RTHDR - to be done */
1222 
1223 }
1224 
1225 /*
1226  * Get pointer to the previous header followed by the header
1227  * currently processed.
1228  * XXX: This function supposes that
1229  *	M includes all headers,
1230  *	the next header field and the header length field of each header
1231  *	are valid, and
1232  *	the sum of each header length equals to OFF.
1233  * Because of these assumptions, this function must be called very
1234  * carefully. Moreover, it will not be used in the near future when
1235  * we develop `neater' mechanism to process extension headers.
1236  */
1237 char *
1238 ip6_get_prevhdr(m, off)
1239 	struct mbuf *m;
1240 	int off;
1241 {
1242 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1243 
1244 	if (off == sizeof(struct ip6_hdr))
1245 		return(&ip6->ip6_nxt);
1246 	else {
1247 		int len, nxt;
1248 		struct ip6_ext *ip6e = NULL;
1249 
1250 		nxt = ip6->ip6_nxt;
1251 		len = sizeof(struct ip6_hdr);
1252 		while (len < off) {
1253 			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1254 
1255 			switch (nxt) {
1256 			case IPPROTO_FRAGMENT:
1257 				len += sizeof(struct ip6_frag);
1258 				break;
1259 			case IPPROTO_AH:
1260 				len += (ip6e->ip6e_len + 2) << 2;
1261 				break;
1262 			default:
1263 				len += (ip6e->ip6e_len + 1) << 3;
1264 				break;
1265 			}
1266 			nxt = ip6e->ip6e_nxt;
1267 		}
1268 		if (ip6e)
1269 			return(&ip6e->ip6e_nxt);
1270 		else
1271 			return NULL;
1272 	}
1273 }
1274 
1275 /*
1276  * get next header offset.  m will be retained.
1277  */
1278 int
1279 ip6_nexthdr(m, off, proto, nxtp)
1280 	struct mbuf *m;
1281 	int off;
1282 	int proto;
1283 	int *nxtp;
1284 {
1285 	struct ip6_hdr ip6;
1286 	struct ip6_ext ip6e;
1287 	struct ip6_frag fh;
1288 
1289 	/* just in case */
1290 	if (m == NULL)
1291 		panic("ip6_nexthdr: m == NULL");
1292 	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1293 		return -1;
1294 
1295 	switch (proto) {
1296 	case IPPROTO_IPV6:
1297 		if (m->m_pkthdr.len < off + sizeof(ip6))
1298 			return -1;
1299 		m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1300 		if (nxtp)
1301 			*nxtp = ip6.ip6_nxt;
1302 		off += sizeof(ip6);
1303 		return off;
1304 
1305 	case IPPROTO_FRAGMENT:
1306 		/*
1307 		 * terminate parsing if it is not the first fragment,
1308 		 * it does not make sense to parse through it.
1309 		 */
1310 		if (m->m_pkthdr.len < off + sizeof(fh))
1311 			return -1;
1312 		m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1313 		if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0)
1314 			return -1;
1315 		if (nxtp)
1316 			*nxtp = fh.ip6f_nxt;
1317 		off += sizeof(struct ip6_frag);
1318 		return off;
1319 
1320 	case IPPROTO_AH:
1321 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1322 			return -1;
1323 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1324 		if (nxtp)
1325 			*nxtp = ip6e.ip6e_nxt;
1326 		off += (ip6e.ip6e_len + 2) << 2;
1327 		if (m->m_pkthdr.len < off)
1328 			return -1;
1329 		return off;
1330 
1331 	case IPPROTO_HOPOPTS:
1332 	case IPPROTO_ROUTING:
1333 	case IPPROTO_DSTOPTS:
1334 		if (m->m_pkthdr.len < off + sizeof(ip6e))
1335 			return -1;
1336 		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1337 		if (nxtp)
1338 			*nxtp = ip6e.ip6e_nxt;
1339 		off += (ip6e.ip6e_len + 1) << 3;
1340 		if (m->m_pkthdr.len < off)
1341 			return -1;
1342 		return off;
1343 
1344 	case IPPROTO_NONE:
1345 	case IPPROTO_ESP:
1346 	case IPPROTO_IPCOMP:
1347 		/* give up */
1348 		return -1;
1349 
1350 	default:
1351 		return -1;
1352 	}
1353 
1354 	return -1;
1355 }
1356 
1357 /*
1358  * get offset for the last header in the chain.  m will be kept untainted.
1359  */
1360 int
1361 ip6_lasthdr(m, off, proto, nxtp)
1362 	struct mbuf *m;
1363 	int off;
1364 	int proto;
1365 	int *nxtp;
1366 {
1367 	int newoff;
1368 	int nxt;
1369 
1370 	if (!nxtp) {
1371 		nxt = -1;
1372 		nxtp = &nxt;
1373 	}
1374 	while (1) {
1375 		newoff = ip6_nexthdr(m, off, proto, nxtp);
1376 		if (newoff < 0)
1377 			return off;
1378 		else if (newoff < off)
1379 			return -1;	/* invalid */
1380 		else if (newoff == off)
1381 			return newoff;
1382 
1383 		off = newoff;
1384 		proto = *nxtp;
1385 	}
1386 }
1387 
1388 /*
1389  * System control for IP6
1390  */
1391 
1392 u_char	inet6ctlerrmap[PRC_NCMDS] = {
1393 	0,		0,		0,		0,
1394 	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1395 	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1396 	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1397 	0,		0,		0,		0,
1398 	ENOPROTOOPT
1399 };
1400 
1401 int
1402 ip6_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1403 	int *name;
1404 	u_int namelen;
1405 	void *oldp;
1406 	size_t *oldlenp;
1407 	void *newp;
1408 	size_t newlen;
1409 {
1410 	int old, error;
1411 
1412 	/* All sysctl names at this level are terminal. */
1413 	if (namelen != 1)
1414 		return ENOTDIR;
1415 
1416 	switch (name[0]) {
1417 
1418 	case IPV6CTL_FORWARDING:
1419 		return sysctl_int(oldp, oldlenp, newp, newlen,
1420 				  &ip6_forwarding);
1421 	case IPV6CTL_SENDREDIRECTS:
1422 		return sysctl_int(oldp, oldlenp, newp, newlen,
1423 				&ip6_sendredirects);
1424 	case IPV6CTL_DEFHLIM:
1425 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_defhlim);
1426 	case IPV6CTL_MAXFRAGPACKETS:
1427 		return sysctl_int(oldp, oldlenp, newp, newlen,
1428 				&ip6_maxfragpackets);
1429 	case IPV6CTL_ACCEPT_RTADV:
1430 		return sysctl_int(oldp, oldlenp, newp, newlen,
1431 				&ip6_accept_rtadv);
1432 	case IPV6CTL_KEEPFAITH:
1433 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_keepfaith);
1434 	case IPV6CTL_LOG_INTERVAL:
1435 		return sysctl_int(oldp, oldlenp, newp, newlen,
1436 				&ip6_log_interval);
1437 	case IPV6CTL_HDRNESTLIMIT:
1438 		return sysctl_int(oldp, oldlenp, newp, newlen,
1439 				&ip6_hdrnestlimit);
1440 	case IPV6CTL_DAD_COUNT:
1441 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_dad_count);
1442 	case IPV6CTL_AUTO_FLOWLABEL:
1443 		return sysctl_int(oldp, oldlenp, newp, newlen,
1444 				&ip6_auto_flowlabel);
1445 	case IPV6CTL_DEFMCASTHLIM:
1446 		return sysctl_int(oldp, oldlenp, newp, newlen,
1447 				&ip6_defmcasthlim);
1448 	case IPV6CTL_GIF_HLIM:
1449 		return sysctl_int(oldp, oldlenp, newp, newlen,
1450 				&ip6_gif_hlim);
1451 	case IPV6CTL_KAME_VERSION:
1452 		return sysctl_rdstring(oldp, oldlenp, newp, __KAME_VERSION);
1453 	case IPV6CTL_USE_DEPRECATED:
1454 		return sysctl_int(oldp, oldlenp, newp, newlen,
1455 				&ip6_use_deprecated);
1456 	case IPV6CTL_RR_PRUNE:
1457 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_rr_prune);
1458 	case IPV6CTL_V6ONLY:
1459 #ifdef INET6_BINDV6ONLY
1460 		return sysctl_rdint(oldp, oldlenp, newp, ip6_v6only);
1461 #else
1462 		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_v6only);
1463 #endif
1464 	case IPV6CTL_ANONPORTMIN:
1465 		old = ip6_anonportmin;
1466 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1467 		    &ip6_anonportmin);
1468 		if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmin < 0 ||
1469 		    ip6_anonportmin > 65535
1470 #ifndef IPNOPRIVPORTS
1471 		    || ip6_anonportmin < IPV6PORT_RESERVED
1472 #endif
1473 		    ) {
1474 			ip6_anonportmin = old;
1475 			return (EINVAL);
1476 		}
1477 		return (error);
1478 	case IPV6CTL_ANONPORTMAX:
1479 		old = ip6_anonportmax;
1480 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1481 		    &ip6_anonportmax);
1482 		if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmax < 0 ||
1483 		    ip6_anonportmax > 65535
1484 #ifndef IPNOPRIVPORTS
1485 		    || ip6_anonportmax < IPV6PORT_RESERVED
1486 #endif
1487 		    ) {
1488 			ip6_anonportmax = old;
1489 			return (EINVAL);
1490 		}
1491 		return (error);
1492 #ifndef IPNOPRIVPORTS
1493 	case IPV6CTL_LOWPORTMIN:
1494 		old = ip6_lowportmin;
1495 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1496 		    &ip6_lowportmin);
1497 		if (ip6_lowportmin >= ip6_lowportmax ||
1498 		    ip6_lowportmin > IPV6PORT_RESERVEDMAX ||
1499 		    ip6_lowportmin < IPV6PORT_RESERVEDMIN) {
1500 			ip6_lowportmin = old;
1501 			return (EINVAL);
1502 		}
1503 		return (error);
1504 	case IPV6CTL_LOWPORTMAX:
1505 		old = ip6_lowportmax;
1506 		error = sysctl_int(oldp, oldlenp, newp, newlen,
1507 		    &ip6_lowportmax);
1508 		if (ip6_lowportmin >= ip6_lowportmax ||
1509 		    ip6_lowportmax > IPV6PORT_RESERVEDMAX ||
1510 		    ip6_lowportmax < IPV6PORT_RESERVEDMIN) {
1511 			ip6_lowportmax = old;
1512 			return (EINVAL);
1513 		}
1514 		return (error);
1515 #endif
1516 	default:
1517 		return EOPNOTSUPP;
1518 	}
1519 	/* NOTREACHED */
1520 }
1521