ip6_input.c revision 1.47
1/*	$NetBSD: ip6_input.c,v 1.47 2001/11/02 08:05:48 itojun Exp $	*/
2/*	$KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $	*/
3
4/*
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 *    notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 *    notice, this list of conditions and the following disclaimer in the
15 *    documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 *    may be used to endorse or promote products derived from this software
18 *    without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 */
32
33/*
34 * Copyright (c) 1982, 1986, 1988, 1993
35 *	The Regents of the University of California.  All rights reserved.
36 *
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
39 * are met:
40 * 1. Redistributions of source code must retain the above copyright
41 *    notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 *    notice, this list of conditions and the following disclaimer in the
44 *    documentation and/or other materials provided with the distribution.
45 * 3. All advertising materials mentioning features or use of this software
46 *    must display the following acknowledgement:
47 *	This product includes software developed by the University of
48 *	California, Berkeley and its contributors.
49 * 4. Neither the name of the University nor the names of its contributors
50 *    may be used to endorse or promote products derived from this software
51 *    without specific prior written permission.
52 *
53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
63 * SUCH DAMAGE.
64 *
65 *	@(#)ip_input.c	8.2 (Berkeley) 1/4/94
66 */
67
68#include "opt_inet.h"
69#include "opt_ipsec.h"
70#include "opt_pfil_hooks.h"
71
72#include <sys/param.h>
73#include <sys/systm.h>
74#include <sys/malloc.h>
75#include <sys/mbuf.h>
76#include <sys/domain.h>
77#include <sys/protosw.h>
78#include <sys/socket.h>
79#include <sys/socketvar.h>
80#include <sys/errno.h>
81#include <sys/time.h>
82#include <sys/kernel.h>
83#include <sys/syslog.h>
84#include <sys/proc.h>
85#include <sys/sysctl.h>
86
87#include <net/if.h>
88#include <net/if_types.h>
89#include <net/if_dl.h>
90#include <net/route.h>
91#include <net/netisr.h>
92#ifdef PFIL_HOOKS
93#include <net/pfil.h>
94#endif
95
96#include <netinet/in.h>
97#include <netinet/in_systm.h>
98#ifdef INET
99#include <netinet/ip.h>
100#include <netinet/ip_icmp.h>
101#endif /* INET */
102#include <netinet/ip6.h>
103#include <netinet6/in6_var.h>
104#include <netinet6/ip6_var.h>
105#include <netinet6/in6_pcb.h>
106#include <netinet/icmp6.h>
107#include <netinet6/in6_ifattach.h>
108#include <netinet6/nd6.h>
109#include <netinet6/in6_prefix.h>
110
111#ifdef IPSEC
112#include <netinet6/ipsec.h>
113#endif
114
115#include <netinet6/ip6protosw.h>
116
117/* we need it for NLOOP. */
118#include "loop.h"
119#include "faith.h"
120#include "gif.h"
121#include "bpfilter.h"
122
123#include <net/net_osdep.h>
124
125extern struct domain inet6domain;
126
127u_char ip6_protox[IPPROTO_MAX];
128static int ip6qmaxlen = IFQ_MAXLEN;
129struct in6_ifaddr *in6_ifaddr;
130struct ifqueue ip6intrq;
131
132extern struct ifnet loif[NLOOP];
133int ip6_forward_srcrt;			/* XXX */
134int ip6_sourcecheck;			/* XXX */
135int ip6_sourcecheck_interval;		/* XXX */
136
137#ifdef PFIL_HOOKS
138struct pfil_head inet6_pfil_hook;
139#endif
140
141struct ip6stat ip6stat;
142
143static void ip6_init2 __P((void *));
144
145static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
146
147/*
148 * IP6 initialization: fill in IP6 protocol switch table.
149 * All protocols not implemented in kernel go to raw IP6 protocol handler.
150 */
151void
152ip6_init()
153{
154	struct ip6protosw *pr;
155	int i;
156	struct timeval tv;
157
158	pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
159	if (pr == 0)
160		panic("ip6_init");
161	for (i = 0; i < IPPROTO_MAX; i++)
162		ip6_protox[i] = pr - inet6sw;
163	for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
164	    pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
165		if (pr->pr_domain->dom_family == PF_INET6 &&
166		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
167			ip6_protox[pr->pr_protocol] = pr - inet6sw;
168	ip6intrq.ifq_maxlen = ip6qmaxlen;
169	nd6_init();
170	frag6_init();
171	/*
172	 * in many cases, random() here does NOT return random number
173	 * as initialization during bootstrap time occur in fixed order.
174	 */
175	microtime(&tv);
176	ip6_flow_seq = random() ^ tv.tv_usec;
177
178	ip6_init2((void *)0);
179
180#ifdef PFIL_HOOKS
181	/* Register our Packet Filter hook. */
182	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
183	inet6_pfil_hook.ph_af   = AF_INET6;
184	i = pfil_head_register(&inet6_pfil_hook);
185	if (i != 0)
186		printf("ip6_init: WARNING: unable to register pfil hook, "
187		    "error %d\n", i);
188#endif /* PFIL_HOOKS */
189}
190
191static void
192ip6_init2(dummy)
193	void *dummy;
194{
195	/*
196	 * to route local address of p2p link to loopback,
197	 * assign loopback address first.
198	 */
199	in6_ifattach(&loif[0], NULL);
200
201	/* nd6_timer_init */
202	callout_init(&nd6_timer_ch);
203	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
204	/* router renumbering prefix list maintenance */
205	callout_init(&in6_rr_timer_ch);
206	callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
207}
208
209/*
210 * IP6 input interrupt handling. Just pass the packet to ip6_input.
211 */
212void
213ip6intr()
214{
215	int s;
216	struct mbuf *m;
217
218	for (;;) {
219		s = splnet();
220		IF_DEQUEUE(&ip6intrq, m);
221		splx(s);
222		if (m == 0)
223			return;
224		ip6_input(m);
225	}
226}
227
228extern struct	route_in6 ip6_forward_rt;
229
230void
231ip6_input(m)
232	struct mbuf *m;
233{
234	struct ip6_hdr *ip6;
235	int off = sizeof(struct ip6_hdr), nest;
236	u_int32_t plen;
237	u_int32_t rtalert = ~0;
238	int nxt, ours = 0;
239	struct ifnet *deliverifp = NULL;
240
241#ifdef IPSEC
242	/*
243	 * should the inner packet be considered authentic?
244	 * see comment in ah4_input().
245	 */
246	if (m) {
247		m->m_flags &= ~M_AUTHIPHDR;
248		m->m_flags &= ~M_AUTHIPDGM;
249	}
250#endif
251
252	/*
253	 * mbuf statistics
254	 */
255	if (m->m_flags & M_EXT) {
256		if (m->m_next)
257			ip6stat.ip6s_mext2m++;
258		else
259			ip6stat.ip6s_mext1++;
260	} else {
261#define M2MMAX	(sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
262		if (m->m_next) {
263			if (m->m_flags & M_LOOP) {
264				ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
265			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
266				ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
267			else
268				ip6stat.ip6s_m2m[0]++;
269		} else
270			ip6stat.ip6s_m1++;
271#undef M2MMAX
272	}
273
274	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
275	ip6stat.ip6s_total++;
276
277#ifndef PULLDOWN_TEST
278	/* XXX is the line really necessary? */
279	IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /*nothing*/);
280#endif
281
282	if (m->m_len < sizeof(struct ip6_hdr)) {
283		struct ifnet *inifp;
284		inifp = m->m_pkthdr.rcvif;
285		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == 0) {
286			ip6stat.ip6s_toosmall++;
287			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
288			return;
289		}
290	}
291
292	ip6 = mtod(m, struct ip6_hdr *);
293
294	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
295		ip6stat.ip6s_badvers++;
296		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
297		goto bad;
298	}
299
300#ifdef PFIL_HOOKS
301	/*
302	 * Run through list of hooks for input packets.  If there are any
303	 * filters which require that additional packets in the flow are
304	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
305	 * Note that filters must _never_ set this flag, as another filter
306	 * in the list may have previously cleared it.
307	 */
308	/*
309	 * let ipfilter look at packet on the wire,
310	 * not the decapsulated packet.
311	 */
312#ifdef IPSEC
313	if (!ipsec_getnhist(m))
314#else
315	if (1)
316#endif
317	{
318		if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif,
319				   PFIL_IN) != 0)
320			return;
321		if (m == NULL)
322			return;
323		ip6 = mtod(m, struct ip6_hdr *);
324	}
325#endif /* PFIL_HOOKS */
326
327	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
328
329#ifdef ALTQ
330	if (altq_input != NULL && (*altq_input)(m, AF_INET6) == 0) {
331		/* packet is dropped by traffic conditioner */
332		return;
333	}
334#endif
335
336	/*
337	 * Check against address spoofing/corruption.
338	 */
339	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
340	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
341		ip6stat.ip6s_badscope++;
342		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
343		goto bad;
344	}
345	/*
346	 * The following check is not documented in specs.  A malicious
347	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
348	 * and bypass security checks (act as if it was from 127.0.0.1 by using
349	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
350	 *
351	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
352	 * support IPv4-less kernel compilation, we cannot support SIIT
353	 * environment at all.  So, it makes more sense for us to reject any
354	 * malicious packets for non-SIIT environment, than try to do a
355	 * partical support for SIIT environment.
356	 */
357	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
358	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
359		ip6stat.ip6s_badscope++;
360		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
361		goto bad;
362	}
363#if 0
364	/*
365	 * Reject packets with IPv4 compatible addresses (auto tunnel).
366	 *
367	 * The code forbids auto tunnel relay case in RFC1933 (the check is
368	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
369	 * is revised to forbid relaying case.
370	 */
371	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
372	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
373		ip6stat.ip6s_badscope++;
374		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
375		goto bad;
376	}
377#endif
378
379	if (IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
380	    IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) {
381		if (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) {
382			ours = 1;
383			deliverifp = m->m_pkthdr.rcvif;
384			goto hbhcheck;
385		} else {
386			ip6stat.ip6s_badscope++;
387			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
388			goto bad;
389		}
390	}
391
392	/* drop packets if interface ID portion is already filled */
393	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
394		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
395		    ip6->ip6_src.s6_addr16[1]) {
396			ip6stat.ip6s_badscope++;
397			goto bad;
398		}
399		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) &&
400		    ip6->ip6_dst.s6_addr16[1]) {
401			ip6stat.ip6s_badscope++;
402			goto bad;
403		}
404	}
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	 * We use rt->rt_ifp to determine if the address is ours or not.
415	 * If rt_ifp is lo0, the address is ours.
416	 * The problem here is, rt->rt_ifp for fe80::%lo0/64 is set to lo0,
417	 * so any address under fe80::%lo0/64 will be mistakenly considered
418	 * local.  The special case is supplied to handle the case properly
419	 * by actually looking at interface addresses
420	 * (using in6ifa_ifpwithaddr).
421	 */
422	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) != 0 &&
423	    IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst)) {
424		if (!in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, &ip6->ip6_dst)) {
425			icmp6_error(m, ICMP6_DST_UNREACH,
426			    ICMP6_DST_UNREACH_ADDR, 0);
427			/* m is already freed */
428			return;
429		}
430
431		ours = 1;
432		deliverifp = m->m_pkthdr.rcvif;
433		goto hbhcheck;
434	}
435
436	/*
437	 * Multicast check
438	 */
439	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
440	  	struct	in6_multi *in6m = 0;
441
442		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
443		/*
444		 * See if we belong to the destination multicast group on the
445		 * arrival interface.
446		 */
447		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
448		if (in6m)
449			ours = 1;
450		else if (!ip6_mrouter) {
451			ip6stat.ip6s_notmember++;
452			ip6stat.ip6s_cantforward++;
453			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
454			goto bad;
455		}
456		deliverifp = m->m_pkthdr.rcvif;
457		goto hbhcheck;
458	}
459
460	/*
461	 *  Unicast check
462	 */
463	if (ip6_forward_rt.ro_rt != NULL &&
464	    (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
465	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
466			       &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
467		ip6stat.ip6s_forward_cachehit++;
468	else {
469		struct sockaddr_in6 *dst6;
470
471		if (ip6_forward_rt.ro_rt) {
472			/* route is down or destination is different */
473			ip6stat.ip6s_forward_cachemiss++;
474			RTFREE(ip6_forward_rt.ro_rt);
475			ip6_forward_rt.ro_rt = 0;
476		}
477
478		bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
479		dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
480		dst6->sin6_len = sizeof(struct sockaddr_in6);
481		dst6->sin6_family = AF_INET6;
482		dst6->sin6_addr = ip6->ip6_dst;
483
484		rtalloc((struct route *)&ip6_forward_rt);
485	}
486
487#define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
488
489	/*
490	 * Accept the packet if the forwarding interface to the destination
491	 * according to the routing table is the loopback interface,
492	 * unless the associated route has a gateway.
493	 * Note that this approach causes to accept a packet if there is a
494	 * route to the loopback interface for the destination of the packet.
495	 * But we think it's even useful in some situations, e.g. when using
496	 * a special daemon which wants to intercept the packet.
497	 */
498	if (ip6_forward_rt.ro_rt &&
499	    (ip6_forward_rt.ro_rt->rt_flags &
500	     (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
501#if 0
502	    /*
503	     * The check below is redundant since the comparison of
504	     * the destination and the key of the rtentry has
505	     * already done through looking up the routing table.
506	     */
507	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
508			       &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr) &&
509#endif
510	    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
511		struct in6_ifaddr *ia6 =
512			(struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
513		if (ia6->ia6_flags & IN6_IFF_ANYCAST)
514			m->m_flags |= M_ANYCAST6;
515		/*
516		 * packets to a tentative, duplicated, or somehow invalid
517		 * address must not be accepted.
518		 */
519		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
520			/* this address is ready */
521			ours = 1;
522			deliverifp = ia6->ia_ifp;	/* correct? */
523			goto hbhcheck;
524		} else {
525			/* address is not ready, so discard the packet. */
526			nd6log((LOG_INFO,
527			    "ip6_input: packet to an unready address %s->%s\n",
528			    ip6_sprintf(&ip6->ip6_src),
529			    ip6_sprintf(&ip6->ip6_dst)));
530
531			goto bad;
532		}
533	}
534
535	/*
536	 * FAITH(Firewall Aided Internet Translator)
537	 */
538#if defined(NFAITH) && 0 < NFAITH
539	if (ip6_keepfaith) {
540		if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
541		 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
542			/* XXX do we need more sanity checks? */
543			ours = 1;
544			deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
545			goto hbhcheck;
546		}
547	}
548#endif
549
550#if 0
551    {
552	/*
553	 * Last resort: check in6_ifaddr for incoming interface.
554	 * The code is here until I update the "goto ours hack" code above
555	 * working right.
556	 */
557	struct ifaddr *ifa;
558	for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
559	     ifa;
560	     ifa = ifa->ifa_list.tqe_next) {
561		if (ifa->ifa_addr == NULL)
562			continue;	/* just for safety */
563		if (ifa->ifa_addr->sa_family != AF_INET6)
564			continue;
565		if (IN6_ARE_ADDR_EQUAL(IFA_IN6(ifa), &ip6->ip6_dst)) {
566			ours = 1;
567			deliverifp = ifa->ifa_ifp;
568			goto hbhcheck;
569		}
570	}
571    }
572#endif
573
574	/*
575	 * Now there is no reason to process the packet if it's not our own
576	 * and we're not a router.
577	 */
578	if (!ip6_forwarding) {
579		ip6stat.ip6s_cantforward++;
580		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
581		goto bad;
582	}
583
584  hbhcheck:
585	/*
586	 * Process Hop-by-Hop options header if it's contained.
587	 * m may be modified in ip6_hopopts_input().
588	 * If a JumboPayload option is included, plen will also be modified.
589	 */
590	plen = (u_int32_t)ntohs(ip6->ip6_plen);
591	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
592		struct ip6_hbh *hbh;
593
594		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
595#if 0	/*touches NULL pointer*/
596			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
597#endif
598			return;	/* m have already been freed */
599		}
600
601		/* adjust pointer */
602		ip6 = mtod(m, struct ip6_hdr *);
603
604		/*
605		 * if the payload length field is 0 and the next header field
606		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
607		 * option MUST be included.
608		 */
609		if (ip6->ip6_plen == 0 && plen == 0) {
610			/*
611			 * Note that if a valid jumbo payload option is
612			 * contained, ip6_hoptops_input() must set a valid
613			 * (non-zero) payload length to the variable plen.
614			 */
615			ip6stat.ip6s_badoptions++;
616			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
617			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
618			icmp6_error(m, ICMP6_PARAM_PROB,
619				    ICMP6_PARAMPROB_HEADER,
620				    (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
621			return;
622		}
623#ifndef PULLDOWN_TEST
624		/* ip6_hopopts_input() ensures that mbuf is contiguous */
625		hbh = (struct ip6_hbh *)(ip6 + 1);
626#else
627		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
628			sizeof(struct ip6_hbh));
629		if (hbh == NULL) {
630			ip6stat.ip6s_tooshort++;
631			return;
632		}
633#endif
634		nxt = hbh->ip6h_nxt;
635
636		/*
637		 * accept the packet if a router alert option is included
638		 * and we act as an IPv6 router.
639		 */
640		if (rtalert != ~0 && ip6_forwarding)
641			ours = 1;
642	} else
643		nxt = ip6->ip6_nxt;
644
645	/*
646	 * Check that the amount of data in the buffers
647	 * is as at least much as the IPv6 header would have us expect.
648	 * Trim mbufs if longer than we expect.
649	 * Drop packet if shorter than we expect.
650	 */
651	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
652		ip6stat.ip6s_tooshort++;
653		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
654		goto bad;
655	}
656	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
657		if (m->m_len == m->m_pkthdr.len) {
658			m->m_len = sizeof(struct ip6_hdr) + plen;
659			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
660		} else
661			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
662	}
663
664	/*
665	 * Forward if desirable.
666	 */
667	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
668		/*
669		 * If we are acting as a multicast router, all
670		 * incoming multicast packets are passed to the
671		 * kernel-level multicast forwarding function.
672		 * The packet is returned (relatively) intact; if
673		 * ip6_mforward() returns a non-zero value, the packet
674		 * must be discarded, else it may be accepted below.
675		 */
676		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
677			ip6stat.ip6s_cantforward++;
678			m_freem(m);
679			return;
680		}
681		if (!ours) {
682			m_freem(m);
683			return;
684		}
685	} else if (!ours) {
686		ip6_forward(m, 0);
687		return;
688	}
689
690	ip6 = mtod(m, struct ip6_hdr *);
691
692	/*
693	 * Malicious party may be able to use IPv4 mapped addr to confuse
694	 * tcp/udp stack and bypass security checks (act as if it was from
695	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
696	 *
697	 * For SIIT end node behavior, you may want to disable the check.
698	 * However, you will  become vulnerable to attacks using IPv4 mapped
699	 * source.
700	 */
701	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
702	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
703		ip6stat.ip6s_badscope++;
704		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
705		goto bad;
706	}
707
708	/*
709	 * Tell launch routine the next header
710	 */
711#ifdef IFA_STATS
712	if (deliverifp != NULL) {
713		struct in6_ifaddr *ia6;
714		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
715		if (ia6)
716			ia6->ia_ifa.ifa_data.ifad_inbytes += m->m_pkthdr.len;
717	}
718#endif
719	ip6stat.ip6s_delivered++;
720	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
721	nest = 0;
722
723	while (nxt != IPPROTO_DONE) {
724		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
725			ip6stat.ip6s_toomanyhdr++;
726			goto bad;
727		}
728
729		/*
730		 * protection against faulty packet - there should be
731		 * more sanity checks in header chain processing.
732		 */
733		if (m->m_pkthdr.len < off) {
734			ip6stat.ip6s_tooshort++;
735			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
736			goto bad;
737		}
738
739#ifdef IPSEC
740		/*
741		 * enforce IPsec policy checking if we are seeing last header.
742		 * note that we do not visit this with protocols with pcb layer
743		 * code - like udp/tcp/raw ip.
744		 */
745		if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
746		    ipsec6_in_reject(m, NULL)) {
747			ipsec6stat.in_polvio++;
748			goto bad;
749		}
750#endif
751
752		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
753	}
754	return;
755 bad:
756	m_freem(m);
757}
758
759/*
760 * Hop-by-Hop options header processing. If a valid jumbo payload option is
761 * included, the real payload length will be stored in plenp.
762 */
763static int
764ip6_hopopts_input(plenp, rtalertp, mp, offp)
765	u_int32_t *plenp;
766	u_int32_t *rtalertp;	/* XXX: should be stored more smart way */
767	struct mbuf **mp;
768	int *offp;
769{
770	struct mbuf *m = *mp;
771	int off = *offp, hbhlen;
772	struct ip6_hbh *hbh;
773	u_int8_t *opt;
774
775	/* validation of the length of the header */
776#ifndef PULLDOWN_TEST
777	IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
778	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
779	hbhlen = (hbh->ip6h_len + 1) << 3;
780
781	IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
782	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
783#else
784	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
785		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
786	if (hbh == NULL) {
787		ip6stat.ip6s_tooshort++;
788		return -1;
789	}
790	hbhlen = (hbh->ip6h_len + 1) << 3;
791	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
792		hbhlen);
793	if (hbh == NULL) {
794		ip6stat.ip6s_tooshort++;
795		return -1;
796	}
797#endif
798	off += hbhlen;
799	hbhlen -= sizeof(struct ip6_hbh);
800	opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
801
802	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
803				hbhlen, rtalertp, plenp) < 0)
804		return(-1);
805
806	*offp = off;
807	*mp = m;
808	return(0);
809}
810
811/*
812 * Search header for all Hop-by-hop options and process each option.
813 * This function is separate from ip6_hopopts_input() in order to
814 * handle a case where the sending node itself process its hop-by-hop
815 * options header. In such a case, the function is called from ip6_output().
816 */
817int
818ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
819	struct mbuf *m;
820	u_int8_t *opthead;
821	int hbhlen;
822	u_int32_t *rtalertp;
823	u_int32_t *plenp;
824{
825	struct ip6_hdr *ip6;
826	int optlen = 0;
827	u_int8_t *opt = opthead;
828	u_int16_t rtalert_val;
829	u_int32_t jumboplen;
830
831	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
832		switch (*opt) {
833		case IP6OPT_PAD1:
834			optlen = 1;
835			break;
836		case IP6OPT_PADN:
837			if (hbhlen < IP6OPT_MINLEN) {
838				ip6stat.ip6s_toosmall++;
839				goto bad;
840			}
841			optlen = *(opt + 1) + 2;
842			break;
843		case IP6OPT_RTALERT:
844			/* XXX may need check for alignment */
845			if (hbhlen < IP6OPT_RTALERT_LEN) {
846				ip6stat.ip6s_toosmall++;
847				goto bad;
848			}
849			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
850				 /* XXX: should we discard the packet? */
851				log(LOG_ERR, "length of router alert opt is inconsitent(%d)",
852				    *(opt + 1));
853			}
854			optlen = IP6OPT_RTALERT_LEN;
855			bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
856			*rtalertp = ntohs(rtalert_val);
857			break;
858		case IP6OPT_JUMBO:
859			/* XXX may need check for alignment */
860			if (hbhlen < IP6OPT_JUMBO_LEN) {
861				ip6stat.ip6s_toosmall++;
862				goto bad;
863			}
864			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
865				/* XXX: should we discard the packet? */
866				log(LOG_ERR, "length of jumbopayload opt "
867				    "is inconsistent(%d)\n",
868				    *(opt + 1));
869			}
870			optlen = IP6OPT_JUMBO_LEN;
871
872			/*
873			 * IPv6 packets that have non 0 payload length
874			 * must not contain a jumbo payload option.
875			 */
876			ip6 = mtod(m, struct ip6_hdr *);
877			if (ip6->ip6_plen) {
878				ip6stat.ip6s_badoptions++;
879				icmp6_error(m, ICMP6_PARAM_PROB,
880					    ICMP6_PARAMPROB_HEADER,
881					    sizeof(struct ip6_hdr) +
882					    sizeof(struct ip6_hbh) +
883					    opt - opthead);
884				return(-1);
885			}
886
887			/*
888			 * We may see jumbolen in unaligned location, so
889			 * we'd need to perform bcopy().
890			 */
891			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
892			jumboplen = (u_int32_t)htonl(jumboplen);
893
894#if 1
895			/*
896			 * if there are multiple jumbo payload options,
897			 * *plenp will be non-zero and the packet will be
898			 * rejected.
899			 * the behavior may need some debate in ipngwg -
900			 * multiple options does not make sense, however,
901			 * there's no explicit mention in specification.
902			 */
903			if (*plenp != 0) {
904				ip6stat.ip6s_badoptions++;
905				icmp6_error(m, ICMP6_PARAM_PROB,
906					    ICMP6_PARAMPROB_HEADER,
907					    sizeof(struct ip6_hdr) +
908					    sizeof(struct ip6_hbh) +
909					    opt + 2 - opthead);
910				return(-1);
911			}
912#endif
913
914			/*
915			 * jumbo payload length must be larger than 65535.
916			 */
917			if (jumboplen <= IPV6_MAXPACKET) {
918				ip6stat.ip6s_badoptions++;
919				icmp6_error(m, ICMP6_PARAM_PROB,
920					    ICMP6_PARAMPROB_HEADER,
921					    sizeof(struct ip6_hdr) +
922					    sizeof(struct ip6_hbh) +
923					    opt + 2 - opthead);
924				return(-1);
925			}
926			*plenp = jumboplen;
927
928			break;
929		default:		/* unknown option */
930			if (hbhlen < IP6OPT_MINLEN) {
931				ip6stat.ip6s_toosmall++;
932				goto bad;
933			}
934			if ((optlen = ip6_unknown_opt(opt, m,
935						      sizeof(struct ip6_hdr) +
936						      sizeof(struct ip6_hbh) +
937						      opt - opthead)) == -1)
938				return(-1);
939			optlen += 2;
940			break;
941		}
942	}
943
944	return(0);
945
946  bad:
947	m_freem(m);
948	return(-1);
949}
950
951/*
952 * Unknown option processing.
953 * The third argument `off' is the offset from the IPv6 header to the option,
954 * which is necessary if the IPv6 header the and option header and IPv6 header
955 * is not continuous in order to return an ICMPv6 error.
956 */
957int
958ip6_unknown_opt(optp, m, off)
959	u_int8_t *optp;
960	struct mbuf *m;
961	int off;
962{
963	struct ip6_hdr *ip6;
964
965	switch (IP6OPT_TYPE(*optp)) {
966	case IP6OPT_TYPE_SKIP: /* ignore the option */
967		return((int)*(optp + 1));
968	case IP6OPT_TYPE_DISCARD:	/* silently discard */
969		m_freem(m);
970		return(-1);
971	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
972		ip6stat.ip6s_badoptions++;
973		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
974		return(-1);
975	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
976		ip6stat.ip6s_badoptions++;
977		ip6 = mtod(m, struct ip6_hdr *);
978		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
979		    (m->m_flags & (M_BCAST|M_MCAST)))
980			m_freem(m);
981		else
982			icmp6_error(m, ICMP6_PARAM_PROB,
983				    ICMP6_PARAMPROB_OPTION, off);
984		return(-1);
985	}
986
987	m_freem(m);		/* XXX: NOTREACHED */
988	return(-1);
989}
990
991/*
992 * Create the "control" list for this pcb.
993 *
994 * The routine will be called from upper layer handlers like tcp6_input().
995 * Thus the routine assumes that the caller (tcp6_input) have already
996 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
997 * very first mbuf on the mbuf chain.
998 * We may want to add some infinite loop prevention or sanity checks for safety.
999 * (This applies only when you are using KAME mbuf chain restriction, i.e.
1000 * you are using IP6_EXTHDR_CHECK() not m_pulldown())
1001 */
1002void
1003ip6_savecontrol(in6p, mp, ip6, m)
1004	struct in6pcb *in6p;
1005	struct mbuf **mp;
1006	struct ip6_hdr *ip6;
1007	struct mbuf *m;
1008{
1009	struct proc *p = curproc;	/* XXX */
1010	int privileged;
1011
1012	privileged = 0;
1013	if (p && !suser(p->p_ucred, &p->p_acflag))
1014		privileged++;
1015
1016#ifdef SO_TIMESTAMP
1017	if (in6p->in6p_socket->so_options & SO_TIMESTAMP) {
1018		struct timeval tv;
1019
1020		microtime(&tv);
1021		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1022			SCM_TIMESTAMP, SOL_SOCKET);
1023		if (*mp)
1024			mp = &(*mp)->m_next;
1025	}
1026#endif
1027	if (in6p->in6p_flags & IN6P_RECVDSTADDR) {
1028		*mp = sbcreatecontrol((caddr_t) &ip6->ip6_dst,
1029			sizeof(struct in6_addr), IPV6_RECVDSTADDR,
1030			IPPROTO_IPV6);
1031		if (*mp)
1032			mp = &(*mp)->m_next;
1033	}
1034
1035#ifdef noyet
1036	/* options were tossed above */
1037	if (in6p->in6p_flags & IN6P_RECVOPTS)
1038		/* broken */
1039	/* ip6_srcroute doesn't do what we want here, need to fix */
1040	if (in6p->in6p_flags & IPV6P_RECVRETOPTS)
1041		/* broken */
1042#endif
1043
1044	/* RFC 2292 sec. 5 */
1045	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1046		struct in6_pktinfo pi6;
1047		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1048		if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
1049			pi6.ipi6_addr.s6_addr16[1] = 0;
1050		pi6.ipi6_ifindex = (m && m->m_pkthdr.rcvif)
1051					? m->m_pkthdr.rcvif->if_index
1052					: 0;
1053		*mp = sbcreatecontrol((caddr_t) &pi6,
1054			sizeof(struct in6_pktinfo), IPV6_PKTINFO,
1055			IPPROTO_IPV6);
1056		if (*mp)
1057			mp = &(*mp)->m_next;
1058	}
1059	if (in6p->in6p_flags & IN6P_HOPLIMIT) {
1060		int hlim = ip6->ip6_hlim & 0xff;
1061		*mp = sbcreatecontrol((caddr_t) &hlim,
1062			sizeof(int), IPV6_HOPLIMIT, IPPROTO_IPV6);
1063		if (*mp)
1064			mp = &(*mp)->m_next;
1065	}
1066	/* IN6P_NEXTHOP - for outgoing packet only */
1067
1068	/*
1069	 * IPV6_HOPOPTS socket option. We require super-user privilege
1070	 * for the option, but it might be too strict, since there might
1071	 * be some hop-by-hop options which can be returned to normal user.
1072	 * See RFC 2292 section 6.
1073	 */
1074	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) {
1075		/*
1076		 * Check if a hop-by-hop options header is contatined in the
1077		 * received packet, and if so, store the options as ancillary
1078		 * data. Note that a hop-by-hop options header must be
1079		 * just after the IPv6 header, which fact is assured through
1080		 * the IPv6 input processing.
1081		 */
1082		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1083		if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1084			struct ip6_hbh *hbh;
1085			int hbhlen;
1086
1087#ifndef PULLDOWN_TEST
1088			hbh = (struct ip6_hbh *)(ip6 + 1);
1089			hbhlen = (hbh->ip6h_len + 1) << 3;
1090#else
1091			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1092				sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
1093			if (hbh == NULL) {
1094				ip6stat.ip6s_tooshort++;
1095				return;
1096			}
1097			hbhlen = (hbh->ip6h_len + 1) << 3;
1098			IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
1099				sizeof(struct ip6_hdr), hbhlen);
1100			if (hbh == NULL) {
1101				ip6stat.ip6s_tooshort++;
1102				return;
1103			}
1104#endif
1105
1106			/*
1107			 * XXX: We copy whole the header even if a jumbo
1108			 * payload option is included, which option is to
1109			 * be removed before returning in the RFC 2292.
1110			 * But it's too painful operation...
1111			 */
1112			*mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1113					      IPV6_HOPOPTS, IPPROTO_IPV6);
1114			if (*mp)
1115				mp = &(*mp)->m_next;
1116		}
1117	}
1118
1119	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1120	if (in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDR)) {
1121		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1122		int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);;
1123
1124		/*
1125		 * Search for destination options headers or routing
1126		 * header(s) through the header chain, and stores each
1127		 * header as ancillary data.
1128		 * Note that the order of the headers remains in
1129		 * the chain of ancillary data.
1130		 */
1131		while (1) {	/* is explicit loop prevention necessary? */
1132			struct ip6_ext *ip6e;
1133			int elen;
1134
1135#ifndef PULLDOWN_TEST
1136			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1137			if (nxt == IPPROTO_AH)
1138				elen = (ip6e->ip6e_len + 2) << 2;
1139			else
1140				elen = (ip6e->ip6e_len + 1) << 3;
1141#else
1142			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off,
1143				sizeof(struct ip6_ext));
1144			if (ip6e == NULL) {
1145				ip6stat.ip6s_tooshort++;
1146				return;
1147			}
1148			if (nxt == IPPROTO_AH)
1149				elen = (ip6e->ip6e_len + 2) << 2;
1150			else
1151				elen = (ip6e->ip6e_len + 1) << 3;
1152			IP6_EXTHDR_GET(ip6e, struct ip6_ext *, m, off, elen);
1153			if (ip6e == NULL) {
1154				ip6stat.ip6s_tooshort++;
1155				return;
1156			}
1157#endif
1158
1159			switch (nxt) {
1160		        case IPPROTO_DSTOPTS:
1161				if (!in6p->in6p_flags & IN6P_DSTOPTS)
1162					break;
1163
1164				/*
1165				 * We also require super-user privilege for
1166				 * the option.
1167				 * See the comments on IN6_HOPOPTS.
1168				 */
1169				if (!privileged)
1170					break;
1171
1172				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
1173						      IPV6_DSTOPTS,
1174						      IPPROTO_IPV6);
1175				if (*mp)
1176					mp = &(*mp)->m_next;
1177				break;
1178
1179			case IPPROTO_ROUTING:
1180				if (!in6p->in6p_flags & IN6P_RTHDR)
1181					break;
1182
1183				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
1184						      IPV6_RTHDR,
1185						      IPPROTO_IPV6);
1186				if (*mp)
1187					mp = &(*mp)->m_next;
1188				break;
1189
1190			case IPPROTO_UDP:
1191			case IPPROTO_TCP:
1192			case IPPROTO_ICMPV6:
1193			default:
1194				/*
1195				 * stop search if we encounter an upper
1196				 * layer protocol headers.
1197				 */
1198				goto loopend;
1199
1200			case IPPROTO_HOPOPTS:
1201			case IPPROTO_AH: /* is it possible? */
1202				break;
1203			}
1204
1205			/* proceed with the next header. */
1206			off += elen;
1207			nxt = ip6e->ip6e_nxt;
1208		}
1209	  loopend:
1210		;
1211	}
1212	if ((in6p->in6p_flags & IN6P_HOPOPTS) && privileged) {
1213		/* to be done */
1214	}
1215	if ((in6p->in6p_flags & IN6P_DSTOPTS) && privileged) {
1216		/* to be done */
1217	}
1218	/* IN6P_RTHDR - to be done */
1219
1220}
1221
1222/*
1223 * Get pointer to the previous header followed by the header
1224 * currently processed.
1225 * XXX: This function supposes that
1226 *	M includes all headers,
1227 *	the next header field and the header length field of each header
1228 *	are valid, and
1229 *	the sum of each header length equals to OFF.
1230 * Because of these assumptions, this function must be called very
1231 * carefully. Moreover, it will not be used in the near future when
1232 * we develop `neater' mechanism to process extension headers.
1233 */
1234char *
1235ip6_get_prevhdr(m, off)
1236	struct mbuf *m;
1237	int off;
1238{
1239	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1240
1241	if (off == sizeof(struct ip6_hdr))
1242		return(&ip6->ip6_nxt);
1243	else {
1244		int len, nxt;
1245		struct ip6_ext *ip6e = NULL;
1246
1247		nxt = ip6->ip6_nxt;
1248		len = sizeof(struct ip6_hdr);
1249		while (len < off) {
1250			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1251
1252			switch (nxt) {
1253			case IPPROTO_FRAGMENT:
1254				len += sizeof(struct ip6_frag);
1255				break;
1256			case IPPROTO_AH:
1257				len += (ip6e->ip6e_len + 2) << 2;
1258				break;
1259			default:
1260				len += (ip6e->ip6e_len + 1) << 3;
1261				break;
1262			}
1263			nxt = ip6e->ip6e_nxt;
1264		}
1265		if (ip6e)
1266			return(&ip6e->ip6e_nxt);
1267		else
1268			return NULL;
1269	}
1270}
1271
1272/*
1273 * get next header offset.  m will be retained.
1274 */
1275int
1276ip6_nexthdr(m, off, proto, nxtp)
1277	struct mbuf *m;
1278	int off;
1279	int proto;
1280	int *nxtp;
1281{
1282	struct ip6_hdr ip6;
1283	struct ip6_ext ip6e;
1284	struct ip6_frag fh;
1285
1286	/* just in case */
1287	if (m == NULL)
1288		panic("ip6_nexthdr: m == NULL");
1289	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1290		return -1;
1291
1292	switch (proto) {
1293	case IPPROTO_IPV6:
1294		if (m->m_pkthdr.len < off + sizeof(ip6))
1295			return -1;
1296		m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1297		if (nxtp)
1298			*nxtp = ip6.ip6_nxt;
1299		off += sizeof(ip6);
1300		return off;
1301
1302	case IPPROTO_FRAGMENT:
1303		/*
1304		 * terminate parsing if it is not the first fragment,
1305		 * it does not make sense to parse through it.
1306		 */
1307		if (m->m_pkthdr.len < off + sizeof(fh))
1308			return -1;
1309		m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1310		if ((ntohs(fh.ip6f_offlg) & IP6F_OFF_MASK) != 0)
1311			return -1;
1312		if (nxtp)
1313			*nxtp = fh.ip6f_nxt;
1314		off += sizeof(struct ip6_frag);
1315		return off;
1316
1317	case IPPROTO_AH:
1318		if (m->m_pkthdr.len < off + sizeof(ip6e))
1319			return -1;
1320		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1321		if (nxtp)
1322			*nxtp = ip6e.ip6e_nxt;
1323		off += (ip6e.ip6e_len + 2) << 2;
1324		if (m->m_pkthdr.len < off)
1325			return -1;
1326		return off;
1327
1328	case IPPROTO_HOPOPTS:
1329	case IPPROTO_ROUTING:
1330	case IPPROTO_DSTOPTS:
1331		if (m->m_pkthdr.len < off + sizeof(ip6e))
1332			return -1;
1333		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1334		if (nxtp)
1335			*nxtp = ip6e.ip6e_nxt;
1336		off += (ip6e.ip6e_len + 1) << 3;
1337		if (m->m_pkthdr.len < off)
1338			return -1;
1339		return off;
1340
1341	case IPPROTO_NONE:
1342	case IPPROTO_ESP:
1343	case IPPROTO_IPCOMP:
1344		/* give up */
1345		return -1;
1346
1347	default:
1348		return -1;
1349	}
1350
1351	return -1;
1352}
1353
1354/*
1355 * get offset for the last header in the chain.  m will be kept untainted.
1356 */
1357int
1358ip6_lasthdr(m, off, proto, nxtp)
1359	struct mbuf *m;
1360	int off;
1361	int proto;
1362	int *nxtp;
1363{
1364	int newoff;
1365	int nxt;
1366
1367	if (!nxtp) {
1368		nxt = -1;
1369		nxtp = &nxt;
1370	}
1371	while (1) {
1372		newoff = ip6_nexthdr(m, off, proto, nxtp);
1373		if (newoff < 0)
1374			return off;
1375		else if (newoff < off)
1376			return -1;	/* invalid */
1377		else if (newoff == off)
1378			return newoff;
1379
1380		off = newoff;
1381		proto = *nxtp;
1382	}
1383}
1384
1385/*
1386 * System control for IP6
1387 */
1388
1389u_char	inet6ctlerrmap[PRC_NCMDS] = {
1390	0,		0,		0,		0,
1391	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1392	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1393	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1394	0,		0,		0,		0,
1395	ENOPROTOOPT
1396};
1397
1398int
1399ip6_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1400	int *name;
1401	u_int namelen;
1402	void *oldp;
1403	size_t *oldlenp;
1404	void *newp;
1405	size_t newlen;
1406{
1407	int old, error;
1408
1409	/* All sysctl names at this level are terminal. */
1410	if (namelen != 1)
1411		return ENOTDIR;
1412
1413	switch (name[0]) {
1414
1415	case IPV6CTL_FORWARDING:
1416		return sysctl_int(oldp, oldlenp, newp, newlen,
1417				  &ip6_forwarding);
1418	case IPV6CTL_SENDREDIRECTS:
1419		return sysctl_int(oldp, oldlenp, newp, newlen,
1420				&ip6_sendredirects);
1421	case IPV6CTL_DEFHLIM:
1422		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_defhlim);
1423	case IPV6CTL_MAXFRAGPACKETS:
1424		return sysctl_int(oldp, oldlenp, newp, newlen,
1425				&ip6_maxfragpackets);
1426	case IPV6CTL_ACCEPT_RTADV:
1427		return sysctl_int(oldp, oldlenp, newp, newlen,
1428				&ip6_accept_rtadv);
1429	case IPV6CTL_KEEPFAITH:
1430		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_keepfaith);
1431	case IPV6CTL_LOG_INTERVAL:
1432		return sysctl_int(oldp, oldlenp, newp, newlen,
1433				&ip6_log_interval);
1434	case IPV6CTL_HDRNESTLIMIT:
1435		return sysctl_int(oldp, oldlenp, newp, newlen,
1436				&ip6_hdrnestlimit);
1437	case IPV6CTL_DAD_COUNT:
1438		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_dad_count);
1439	case IPV6CTL_AUTO_FLOWLABEL:
1440		return sysctl_int(oldp, oldlenp, newp, newlen,
1441				&ip6_auto_flowlabel);
1442	case IPV6CTL_DEFMCASTHLIM:
1443		return sysctl_int(oldp, oldlenp, newp, newlen,
1444				&ip6_defmcasthlim);
1445	case IPV6CTL_GIF_HLIM:
1446		return sysctl_int(oldp, oldlenp, newp, newlen,
1447				&ip6_gif_hlim);
1448	case IPV6CTL_KAME_VERSION:
1449		return sysctl_rdstring(oldp, oldlenp, newp, __KAME_VERSION);
1450	case IPV6CTL_USE_DEPRECATED:
1451		return sysctl_int(oldp, oldlenp, newp, newlen,
1452				&ip6_use_deprecated);
1453	case IPV6CTL_RR_PRUNE:
1454		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_rr_prune);
1455	case IPV6CTL_V6ONLY:
1456#ifdef INET6_BINDV6ONLY
1457		return sysctl_rdint(oldp, oldlenp, newp, ip6_v6only);
1458#else
1459		return sysctl_int(oldp, oldlenp, newp, newlen, &ip6_v6only);
1460#endif
1461	case IPV6CTL_ANONPORTMIN:
1462		old = ip6_anonportmin;
1463		error = sysctl_int(oldp, oldlenp, newp, newlen,
1464		    &ip6_anonportmin);
1465		if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmin < 0 ||
1466		    ip6_anonportmin > 65535
1467#ifndef IPNOPRIVPORTS
1468		    || ip6_anonportmin < IPV6PORT_RESERVED
1469#endif
1470		    ) {
1471			ip6_anonportmin = old;
1472			return (EINVAL);
1473		}
1474		return (error);
1475	case IPV6CTL_ANONPORTMAX:
1476		old = ip6_anonportmax;
1477		error = sysctl_int(oldp, oldlenp, newp, newlen,
1478		    &ip6_anonportmax);
1479		if (ip6_anonportmin >= ip6_anonportmax || ip6_anonportmax < 0 ||
1480		    ip6_anonportmax > 65535
1481#ifndef IPNOPRIVPORTS
1482		    || ip6_anonportmax < IPV6PORT_RESERVED
1483#endif
1484		    ) {
1485			ip6_anonportmax = old;
1486			return (EINVAL);
1487		}
1488		return (error);
1489#ifndef IPNOPRIVPORTS
1490	case IPV6CTL_LOWPORTMIN:
1491		old = ip6_lowportmin;
1492		error = sysctl_int(oldp, oldlenp, newp, newlen,
1493		    &ip6_lowportmin);
1494		if (ip6_lowportmin >= ip6_lowportmax ||
1495		    ip6_lowportmin > IPV6PORT_RESERVEDMAX ||
1496		    ip6_lowportmin < IPV6PORT_RESERVEDMIN) {
1497			ip6_lowportmin = old;
1498			return (EINVAL);
1499		}
1500		return (error);
1501	case IPV6CTL_LOWPORTMAX:
1502		old = ip6_lowportmax;
1503		error = sysctl_int(oldp, oldlenp, newp, newlen,
1504		    &ip6_lowportmax);
1505		if (ip6_lowportmin >= ip6_lowportmax ||
1506		    ip6_lowportmax > IPV6PORT_RESERVEDMAX ||
1507		    ip6_lowportmax < IPV6PORT_RESERVEDMIN) {
1508			ip6_lowportmax = old;
1509			return (EINVAL);
1510		}
1511		return (error);
1512#endif
1513	default:
1514		return EOPNOTSUPP;
1515	}
1516	/* NOTREACHED */
1517}
1518