ip6_input.c revision 121143
1/*	$FreeBSD: head/sys/netinet6/ip6_input.c 121143 2003-10-16 18:57:45Z sam $	*/
2/*	$KAME: ip6_input.c,v 1.259 2002/01/21 04:58:09 jinmei Exp $	*/
3
4/*
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 *    notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 *    notice, this list of conditions and the following disclaimer in the
15 *    documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 *    may be used to endorse or promote products derived from this software
18 *    without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 */
32
33/*
34 * Copyright (c) 1982, 1986, 1988, 1993
35 *	The Regents of the University of California.  All rights reserved.
36 *
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
39 * are met:
40 * 1. Redistributions of source code must retain the above copyright
41 *    notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 *    notice, this list of conditions and the following disclaimer in the
44 *    documentation and/or other materials provided with the distribution.
45 * 3. All advertising materials mentioning features or use of this software
46 *    must display the following acknowledgement:
47 *	This product includes software developed by the University of
48 *	California, Berkeley and its contributors.
49 * 4. Neither the name of the University nor the names of its contributors
50 *    may be used to endorse or promote products derived from this software
51 *    without specific prior written permission.
52 *
53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
63 * SUCH DAMAGE.
64 *
65 *	@(#)ip_input.c	8.2 (Berkeley) 1/4/94
66 */
67
68#include "opt_ip6fw.h"
69#include "opt_inet.h"
70#include "opt_inet6.h"
71#include "opt_ipsec.h"
72#include "opt_pfil_hooks.h"
73#include "opt_random_ip_id.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/proc.h>
80#include <sys/domain.h>
81#include <sys/protosw.h>
82#include <sys/socket.h>
83#include <sys/socketvar.h>
84#include <sys/errno.h>
85#include <sys/time.h>
86#include <sys/kernel.h>
87#include <sys/syslog.h>
88
89#include <net/if.h>
90#include <net/if_types.h>
91#include <net/if_dl.h>
92#include <net/route.h>
93#include <net/netisr.h>
94#ifdef PFIL_HOOKS
95#include <net/pfil.h>
96#endif
97
98#include <netinet/in.h>
99#include <netinet/in_systm.h>
100#ifdef INET
101#include <netinet/ip.h>
102#include <netinet/ip_icmp.h>
103#endif /* INET */
104#include <netinet/ip6.h>
105#include <netinet6/in6_var.h>
106#include <netinet6/ip6_var.h>
107#include <netinet/in_pcb.h>
108#include <netinet/icmp6.h>
109#include <netinet6/in6_ifattach.h>
110#include <netinet6/nd6.h>
111#include <netinet6/in6_prefix.h>
112
113#ifdef IPSEC
114#include <netinet6/ipsec.h>
115#ifdef INET6
116#include <netinet6/ipsec6.h>
117#endif
118#endif
119
120#ifdef FAST_IPSEC
121#include <netipsec/ipsec.h>
122#include <netipsec/ipsec6.h>
123#define	IPSEC
124#endif /* FAST_IPSEC */
125
126#include <netinet6/ip6_fw.h>
127
128#include <netinet6/ip6protosw.h>
129
130#include <net/net_osdep.h>
131
132extern struct domain inet6domain;
133
134u_char ip6_protox[IPPROTO_MAX];
135static struct ifqueue ip6intrq;
136static int ip6qmaxlen = IFQ_MAXLEN;
137struct in6_ifaddr *in6_ifaddr;
138
139extern struct callout in6_tmpaddrtimer_ch;
140
141int ip6_forward_srcrt;			/* XXX */
142int ip6_sourcecheck;			/* XXX */
143int ip6_sourcecheck_interval;		/* XXX */
144
145int ip6_ours_check_algorithm;
146
147#ifdef PFIL_HOOKS
148struct pfil_head inet6_pfil_hook;
149#endif
150
151/* firewall hooks */
152ip6_fw_chk_t *ip6_fw_chk_ptr;
153ip6_fw_ctl_t *ip6_fw_ctl_ptr;
154int ip6_fw_enable = 1;
155
156struct ip6stat ip6stat;
157
158static void ip6_init2 __P((void *));
159static struct ip6aux *ip6_setdstifaddr __P((struct mbuf *, struct in6_ifaddr *));
160static int ip6_hopopts_input __P((u_int32_t *, u_int32_t *, struct mbuf **, int *));
161#ifdef PULLDOWN_TEST
162static struct mbuf *ip6_pullexthdr __P((struct mbuf *, size_t, int));
163#endif
164
165/*
166 * IP6 initialization: fill in IP6 protocol switch table.
167 * All protocols not implemented in kernel go to raw IP6 protocol handler.
168 */
169void
170ip6_init()
171{
172	struct ip6protosw *pr;
173	int i;
174
175#ifdef DIAGNOSTIC
176	if (sizeof(struct protosw) != sizeof(struct ip6protosw))
177		panic("sizeof(protosw) != sizeof(ip6protosw)");
178#endif
179	pr = (struct ip6protosw *)pffindproto(PF_INET6, IPPROTO_RAW, SOCK_RAW);
180	if (pr == 0)
181		panic("ip6_init");
182	for (i = 0; i < IPPROTO_MAX; i++)
183		ip6_protox[i] = pr - inet6sw;
184	for (pr = (struct ip6protosw *)inet6domain.dom_protosw;
185	    pr < (struct ip6protosw *)inet6domain.dom_protoswNPROTOSW; pr++)
186		if (pr->pr_domain->dom_family == PF_INET6 &&
187		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
188			ip6_protox[pr->pr_protocol] = pr - inet6sw;
189#ifdef PFIL_HOOKS
190	inet6_pfil_hook.ph_type = PFIL_TYPE_AF;
191	inet6_pfil_hook.ph_af = AF_INET6;
192	if ((i = pfil_head_register(&inet6_pfil_hook)) != 0)
193		printf("%s: WARNING: unable to register pfil hook, "
194			"error %d\n", __func__, i);
195#endif /* PFIL_HOOKS */
196	ip6intrq.ifq_maxlen = ip6qmaxlen;
197	mtx_init(&ip6intrq.ifq_mtx, "ip6_inq", NULL, MTX_DEF);
198	netisr_register(NETISR_IPV6, ip6_input, &ip6intrq);
199	nd6_init();
200	frag6_init();
201#ifndef RANDOM_IP_ID
202	ip6_flow_seq = arc4random();
203#endif
204	ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR;
205}
206
207static void
208ip6_init2(dummy)
209	void *dummy;
210{
211
212	/*
213	 * to route local address of p2p link to loopback,
214	 * assign loopback address first.
215	 */
216	in6_ifattach(&loif[0], NULL);
217
218	/* nd6_timer_init */
219	callout_init(&nd6_timer_ch, 0);
220	callout_reset(&nd6_timer_ch, hz, nd6_timer, NULL);
221
222	/* router renumbering prefix list maintenance */
223	callout_init(&in6_rr_timer_ch, 0);
224	callout_reset(&in6_rr_timer_ch, hz, in6_rr_timer, NULL);
225
226	/* timer for regeneranation of temporary addresses randomize ID */
227	callout_reset(&in6_tmpaddrtimer_ch,
228		      (ip6_temp_preferred_lifetime - ip6_desync_factor -
229		       ip6_temp_regen_advance) * hz,
230		      in6_tmpaddrtimer, NULL);
231}
232
233/* cheat */
234/* This must be after route_init(), which is now SI_ORDER_THIRD */
235SYSINIT(netinet6init2, SI_SUB_PROTO_DOMAIN, SI_ORDER_MIDDLE, ip6_init2, NULL);
236
237extern struct	route_in6 ip6_forward_rt;
238
239void
240ip6_input(m)
241	struct mbuf *m;
242{
243	struct ip6_hdr *ip6;
244	int off = sizeof(struct ip6_hdr), nest;
245	u_int32_t plen;
246	u_int32_t rtalert = ~0;
247	int nxt, ours = 0;
248	struct ifnet *deliverifp = NULL;
249
250#ifdef PFIL_HOOKS
251	struct in6_addr odst;
252#endif
253	int srcrt = 0;
254#ifdef IPSEC
255	/*
256	 * should the inner packet be considered authentic?
257	 * see comment in ah4_input().
258	 */
259	if (m) {
260		m->m_flags &= ~M_AUTHIPHDR;
261		m->m_flags &= ~M_AUTHIPDGM;
262	}
263#endif
264
265	/*
266	 * make sure we don't have onion peering information into m_aux.
267	 */
268	ip6_delaux(m);
269
270	/*
271	 * mbuf statistics
272	 */
273	if (m->m_flags & M_EXT) {
274		if (m->m_next)
275			ip6stat.ip6s_mext2m++;
276		else
277			ip6stat.ip6s_mext1++;
278	} else {
279#define M2MMAX	(sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0]))
280		if (m->m_next) {
281			if (m->m_flags & M_LOOP) {
282				ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
283			} else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
284				ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
285			else
286				ip6stat.ip6s_m2m[0]++;
287		} else
288			ip6stat.ip6s_m1++;
289#undef M2MMAX
290	}
291
292	in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
293	ip6stat.ip6s_total++;
294
295#ifndef PULLDOWN_TEST
296	/*
297	 * L2 bridge code and some other code can return mbuf chain
298	 * that does not conform to KAME requirement.  too bad.
299	 * XXX: fails to join if interface MTU > MCLBYTES.  jumbogram?
300	 */
301	if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
302		struct mbuf *n;
303
304		MGETHDR(n, M_DONTWAIT, MT_HEADER);
305		if (n)
306			M_MOVE_PKTHDR(n, m);
307		if (n && n->m_pkthdr.len > MHLEN) {
308			MCLGET(n, M_DONTWAIT);
309			if ((n->m_flags & M_EXT) == 0) {
310				m_freem(n);
311				n = NULL;
312			}
313		}
314		if (n == NULL) {
315			m_freem(m);
316			return;	/* ENOBUFS */
317		}
318
319		m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
320		n->m_len = n->m_pkthdr.len;
321		m_freem(m);
322		m = n;
323	}
324	IP6_EXTHDR_CHECK(m, 0, sizeof(struct ip6_hdr), /* nothing */);
325#endif
326
327	if (m->m_len < sizeof(struct ip6_hdr)) {
328		struct ifnet *inifp;
329		inifp = m->m_pkthdr.rcvif;
330		if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
331			ip6stat.ip6s_toosmall++;
332			in6_ifstat_inc(inifp, ifs6_in_hdrerr);
333			return;
334		}
335	}
336
337	ip6 = mtod(m, struct ip6_hdr *);
338
339	if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
340		ip6stat.ip6s_badvers++;
341		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
342		goto bad;
343	}
344
345#ifdef PFIL_HOOKS
346	/*
347	 * Run through list of hooks for input packets.
348	 */
349	if (pfil_run_hooks(&inet6_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN))
350		return;
351	if (m == NULL)			/* consumed by filter */
352		return;
353	 *
354	 * NB: Beware of the destination address changing
355	 *     (e.g. by NAT rewriting).  When this happens,
356	 *     tell ip6_forward to do the right thing.
357	ip6 = mtod(m, struct ip6_hdr *);
358	odst = ip6->ip6_dst;
359#endif /* PFIL_HOOKS */
360
361	ip6stat.ip6s_nxthist[ip6->ip6_nxt]++;
362
363	/*
364	 * Check with the firewall...
365	 */
366	if (ip6_fw_enable && ip6_fw_chk_ptr) {
367		u_short port = 0;
368	srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
369		/* If ipfw says divert, we have to just drop packet */
370		/* use port as a dummy argument */
371		if ((*ip6_fw_chk_ptr)(&ip6, NULL, &port, &m)) {
372			m_freem(m);
373			m = NULL;
374		}
375		if (!m)
376			return;
377	}
378
379	/*
380	 * Check against address spoofing/corruption.
381	 */
382	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_src) ||
383	    IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_dst)) {
384		/*
385		 * XXX: "badscope" is not very suitable for a multicast source.
386		 */
387		ip6stat.ip6s_badscope++;
388		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
389		goto bad;
390	}
391	if ((IN6_IS_ADDR_LOOPBACK(&ip6->ip6_src) ||
392	     IN6_IS_ADDR_LOOPBACK(&ip6->ip6_dst)) &&
393	    (m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
394		ip6stat.ip6s_badscope++;
395		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
396		goto bad;
397	}
398
399	/*
400	 * The following check is not documented in specs.  A malicious
401	 * party may be able to use IPv4 mapped addr to confuse tcp/udp stack
402	 * and bypass security checks (act as if it was from 127.0.0.1 by using
403	 * IPv6 src ::ffff:127.0.0.1).  Be cautious.
404	 *
405	 * This check chokes if we are in an SIIT cloud.  As none of BSDs
406	 * support IPv4-less kernel compilation, we cannot support SIIT
407	 * environment at all.  So, it makes more sense for us to reject any
408	 * malicious packets for non-SIIT environment, than try to do a
409	 * partial support for SIIT environment.
410	 */
411	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
412	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
413		ip6stat.ip6s_badscope++;
414		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
415		goto bad;
416	}
417#if 0
418	/*
419	 * Reject packets with IPv4 compatible addresses (auto tunnel).
420	 *
421	 * The code forbids auto tunnel relay case in RFC1933 (the check is
422	 * stronger than RFC1933).  We may want to re-enable it if mech-xx
423	 * is revised to forbid relaying case.
424	 */
425	if (IN6_IS_ADDR_V4COMPAT(&ip6->ip6_src) ||
426	    IN6_IS_ADDR_V4COMPAT(&ip6->ip6_dst)) {
427		ip6stat.ip6s_badscope++;
428		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
429		goto bad;
430	}
431#endif
432
433	/* drop packets if interface ID portion is already filled */
434	if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) {
435		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src) &&
436		    ip6->ip6_src.s6_addr16[1]) {
437			ip6stat.ip6s_badscope++;
438			goto bad;
439		}
440		if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst) &&
441		    ip6->ip6_dst.s6_addr16[1]) {
442			ip6stat.ip6s_badscope++;
443			goto bad;
444		}
445	}
446
447	if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_src))
448		ip6->ip6_src.s6_addr16[1]
449			= htons(m->m_pkthdr.rcvif->if_index);
450	if (IN6_IS_SCOPE_LINKLOCAL(&ip6->ip6_dst))
451		ip6->ip6_dst.s6_addr16[1]
452			= htons(m->m_pkthdr.rcvif->if_index);
453
454	/*
455	 * Multicast check
456	 */
457	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
458	  	struct in6_multi *in6m = 0;
459
460		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_mcast);
461		/*
462		 * See if we belong to the destination multicast group on the
463		 * arrival interface.
464		 */
465		IN6_LOOKUP_MULTI(ip6->ip6_dst, m->m_pkthdr.rcvif, in6m);
466		if (in6m)
467			ours = 1;
468		else if (!ip6_mrouter) {
469			ip6stat.ip6s_notmember++;
470			ip6stat.ip6s_cantforward++;
471			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
472			goto bad;
473		}
474		deliverifp = m->m_pkthdr.rcvif;
475		goto hbhcheck;
476	}
477
478	/*
479	 *  Unicast check
480	 */
481	if (ip6_forward_rt.ro_rt != NULL &&
482	    (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 &&
483	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
484	    &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr))
485		ip6stat.ip6s_forward_cachehit++;
486	else {
487		struct sockaddr_in6 *dst6;
488
489		if (ip6_forward_rt.ro_rt) {
490			/* route is down or destination is different */
491			ip6stat.ip6s_forward_cachemiss++;
492			RTFREE(ip6_forward_rt.ro_rt);
493			ip6_forward_rt.ro_rt = 0;
494		}
495
496		bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6));
497		dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
498		dst6->sin6_len = sizeof(struct sockaddr_in6);
499		dst6->sin6_family = AF_INET6;
500		dst6->sin6_addr = ip6->ip6_dst;
501
502		rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING);
503	}
504
505#define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key))
506
507	/*
508	 * Accept the packet if the forwarding interface to the destination
509	 * according to the routing table is the loopback interface,
510	 * unless the associated route has a gateway.
511	 * Note that this approach causes to accept a packet if there is a
512	 * route to the loopback interface for the destination of the packet.
513	 * But we think it's even useful in some situations, e.g. when using
514	 * a special daemon which wants to intercept the packet.
515	 *
516	 * XXX: some OSes automatically make a cloned route for the destination
517	 * of an outgoing packet.  If the outgoing interface of the packet
518	 * is a loopback one, the kernel would consider the packet to be
519	 * accepted, even if we have no such address assinged on the interface.
520	 * We check the cloned flag of the route entry to reject such cases,
521	 * assuming that route entries for our own addresses are not made by
522	 * cloning (it should be true because in6_addloop explicitly installs
523	 * the host route).  However, we might have to do an explicit check
524	 * while it would be less efficient.  Or, should we rather install a
525	 * reject route for such a case?
526	 */
527	if (ip6_forward_rt.ro_rt &&
528	    (ip6_forward_rt.ro_rt->rt_flags &
529	     (RTF_HOST|RTF_GATEWAY)) == RTF_HOST &&
530#ifdef RTF_WASCLONED
531	    !(ip6_forward_rt.ro_rt->rt_flags & RTF_WASCLONED) &&
532#endif
533#ifdef RTF_CLONED
534	    !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) &&
535#endif
536#if 0
537	    /*
538	     * The check below is redundant since the comparison of
539	     * the destination and the key of the rtentry has
540	     * already done through looking up the routing table.
541	     */
542	    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
543	    &rt6_key(ip6_forward_rt.ro_rt)->sin6_addr)
544#endif
545	    ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_LOOP) {
546		struct in6_ifaddr *ia6 =
547			(struct in6_ifaddr *)ip6_forward_rt.ro_rt->rt_ifa;
548
549		/*
550		 * record address information into m_aux.
551		 */
552		(void)ip6_setdstifaddr(m, ia6);
553
554		/*
555		 * packets to a tentative, duplicated, or somehow invalid
556		 * address must not be accepted.
557		 */
558		if (!(ia6->ia6_flags & IN6_IFF_NOTREADY)) {
559			/* this address is ready */
560			ours = 1;
561			deliverifp = ia6->ia_ifp;	/* correct? */
562			/* Count the packet in the ip address stats */
563			ia6->ia_ifa.if_ipackets++;
564			ia6->ia_ifa.if_ibytes += m->m_pkthdr.len;
565			goto hbhcheck;
566		} else {
567			/* address is not ready, so discard the packet. */
568			nd6log((LOG_INFO,
569			    "ip6_input: packet to an unready address %s->%s\n",
570			    ip6_sprintf(&ip6->ip6_src),
571			    ip6_sprintf(&ip6->ip6_dst)));
572
573			goto bad;
574		}
575	}
576
577	/*
578	 * FAITH (Firewall Aided Internet Translator)
579	 */
580	if (ip6_keepfaith) {
581		if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp
582		 && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) {
583			/* XXX do we need more sanity checks? */
584			ours = 1;
585			deliverifp = ip6_forward_rt.ro_rt->rt_ifp; /* faith */
586			goto hbhcheck;
587		}
588	}
589
590	/*
591	 * Now there is no reason to process the packet if it's not our own
592	 * and we're not a router.
593	 */
594	if (!ip6_forwarding) {
595		ip6stat.ip6s_cantforward++;
596		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
597		goto bad;
598	}
599
600  hbhcheck:
601	/*
602	 * record address information into m_aux, if we don't have one yet.
603	 * note that we are unable to record it, if the address is not listed
604	 * as our interface address (e.g. multicast addresses, addresses
605	 * within FAITH prefixes and such).
606	 */
607	if (deliverifp && !ip6_getdstifaddr(m)) {
608		struct in6_ifaddr *ia6;
609
610		ia6 = in6_ifawithifp(deliverifp, &ip6->ip6_dst);
611		if (ia6) {
612			if (!ip6_setdstifaddr(m, ia6)) {
613				/*
614				 * XXX maybe we should drop the packet here,
615				 * as we could not provide enough information
616				 * to the upper layers.
617				 */
618			}
619		}
620	}
621
622	/*
623	 * Process Hop-by-Hop options header if it's contained.
624	 * m may be modified in ip6_hopopts_input().
625	 * If a JumboPayload option is included, plen will also be modified.
626	 */
627	plen = (u_int32_t)ntohs(ip6->ip6_plen);
628	if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
629		struct ip6_hbh *hbh;
630
631		if (ip6_hopopts_input(&plen, &rtalert, &m, &off)) {
632#if 0	/*touches NULL pointer*/
633			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
634#endif
635			return;	/* m have already been freed */
636		}
637
638		/* adjust pointer */
639		ip6 = mtod(m, struct ip6_hdr *);
640
641		/*
642		 * if the payload length field is 0 and the next header field
643		 * indicates Hop-by-Hop Options header, then a Jumbo Payload
644		 * option MUST be included.
645		 */
646		if (ip6->ip6_plen == 0 && plen == 0) {
647			/*
648			 * Note that if a valid jumbo payload option is
649			 * contained, ip6_hopopts_input() must set a valid
650			 * (non-zero) payload length to the variable plen.
651			 */
652			ip6stat.ip6s_badoptions++;
653			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_discard);
654			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
655			icmp6_error(m, ICMP6_PARAM_PROB,
656				    ICMP6_PARAMPROB_HEADER,
657				    (caddr_t)&ip6->ip6_plen - (caddr_t)ip6);
658			return;
659		}
660#ifndef PULLDOWN_TEST
661		/* ip6_hopopts_input() ensures that mbuf is contiguous */
662		hbh = (struct ip6_hbh *)(ip6 + 1);
663#else
664		IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
665			sizeof(struct ip6_hbh));
666		if (hbh == NULL) {
667			ip6stat.ip6s_tooshort++;
668			return;
669		}
670#endif
671		nxt = hbh->ip6h_nxt;
672
673		/*
674		 * accept the packet if a router alert option is included
675		 * and we act as an IPv6 router.
676		 */
677		if (rtalert != ~0 && ip6_forwarding)
678			ours = 1;
679	} else
680		nxt = ip6->ip6_nxt;
681
682	/*
683	 * Check that the amount of data in the buffers
684	 * is as at least much as the IPv6 header would have us expect.
685	 * Trim mbufs if longer than we expect.
686	 * Drop packet if shorter than we expect.
687	 */
688	if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
689		ip6stat.ip6s_tooshort++;
690		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
691		goto bad;
692	}
693	if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
694		if (m->m_len == m->m_pkthdr.len) {
695			m->m_len = sizeof(struct ip6_hdr) + plen;
696			m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
697		} else
698			m_adj(m, sizeof(struct ip6_hdr) + plen - m->m_pkthdr.len);
699	}
700
701	/*
702	 * Forward if desirable.
703	 */
704	if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
705		/*
706		 * If we are acting as a multicast router, all
707		 * incoming multicast packets are passed to the
708		 * kernel-level multicast forwarding function.
709		 * The packet is returned (relatively) intact; if
710		 * ip6_mforward() returns a non-zero value, the packet
711		 * must be discarded, else it may be accepted below.
712		 */
713		if (ip6_mrouter && ip6_mforward(ip6, m->m_pkthdr.rcvif, m)) {
714			ip6stat.ip6s_cantforward++;
715			m_freem(m);
716			return;
717		}
718		if (!ours) {
719			m_freem(m);
720			return;
721		}
722	} else if (!ours) {
723		ip6_forward(m, srcrt);
724		return;
725	}
726
727	ip6 = mtod(m, struct ip6_hdr *);
728
729	/*
730	 * Malicious party may be able to use IPv4 mapped addr to confuse
731	 * tcp/udp stack and bypass security checks (act as if it was from
732	 * 127.0.0.1 by using IPv6 src ::ffff:127.0.0.1).  Be cautious.
733	 *
734	 * For SIIT end node behavior, you may want to disable the check.
735	 * However, you will  become vulnerable to attacks using IPv4 mapped
736	 * source.
737	 */
738	if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
739	    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
740		ip6stat.ip6s_badscope++;
741		in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_addrerr);
742		goto bad;
743	}
744
745	/*
746	 * Tell launch routine the next header
747	 */
748	ip6stat.ip6s_delivered++;
749	in6_ifstat_inc(deliverifp, ifs6_in_deliver);
750	nest = 0;
751
752	while (nxt != IPPROTO_DONE) {
753		if (ip6_hdrnestlimit && (++nest > ip6_hdrnestlimit)) {
754			ip6stat.ip6s_toomanyhdr++;
755			goto bad;
756		}
757
758		/*
759		 * protection against faulty packet - there should be
760		 * more sanity checks in header chain processing.
761		 */
762		if (m->m_pkthdr.len < off) {
763			ip6stat.ip6s_tooshort++;
764			in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_truncated);
765			goto bad;
766		}
767
768#if 0
769		/*
770		 * do we need to do it for every header?  yeah, other
771		 * functions can play with it (like re-allocate and copy).
772		 */
773		mhist = ip6_addaux(m);
774		if (mhist && M_TRAILINGSPACE(mhist) >= sizeof(nxt)) {
775			hist = mtod(mhist, caddr_t) + mhist->m_len;
776			bcopy(&nxt, hist, sizeof(nxt));
777			mhist->m_len += sizeof(nxt);
778		} else {
779			ip6stat.ip6s_toomanyhdr++;
780			goto bad;
781		}
782#endif
783
784#ifdef IPSEC
785		/*
786		 * enforce IPsec policy checking if we are seeing last header.
787		 * note that we do not visit this with protocols with pcb layer
788		 * code - like udp/tcp/raw ip.
789		 */
790		if ((inet6sw[ip6_protox[nxt]].pr_flags & PR_LASTHDR) != 0 &&
791		    ipsec6_in_reject(m, NULL)) {
792			ipsec6stat.in_polvio++;
793			goto bad;
794		}
795#endif
796		nxt = (*inet6sw[ip6_protox[nxt]].pr_input)(&m, &off, nxt);
797	}
798	return;
799 bad:
800	m_freem(m);
801}
802
803/*
804 * set/grab in6_ifaddr correspond to IPv6 destination address.
805 * XXX backward compatibility wrapper
806 */
807static struct ip6aux *
808ip6_setdstifaddr(m, ia6)
809	struct mbuf *m;
810	struct in6_ifaddr *ia6;
811{
812	struct ip6aux *n;
813
814	n = ip6_addaux(m);
815	if (n)
816		n->ip6a_dstia6 = ia6;
817	return n;	/* NULL if failed to set */
818}
819
820struct in6_ifaddr *
821ip6_getdstifaddr(m)
822	struct mbuf *m;
823{
824	struct ip6aux *n;
825
826	n = ip6_findaux(m);
827	if (n)
828		return n->ip6a_dstia6;
829	else
830		return NULL;
831}
832
833/*
834 * Hop-by-Hop options header processing. If a valid jumbo payload option is
835 * included, the real payload length will be stored in plenp.
836 */
837static int
838ip6_hopopts_input(plenp, rtalertp, mp, offp)
839	u_int32_t *plenp;
840	u_int32_t *rtalertp;	/* XXX: should be stored more smart way */
841	struct mbuf **mp;
842	int *offp;
843{
844	struct mbuf *m = *mp;
845	int off = *offp, hbhlen;
846	struct ip6_hbh *hbh;
847	u_int8_t *opt;
848
849	/* validation of the length of the header */
850#ifndef PULLDOWN_TEST
851	IP6_EXTHDR_CHECK(m, off, sizeof(*hbh), -1);
852	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
853	hbhlen = (hbh->ip6h_len + 1) << 3;
854
855	IP6_EXTHDR_CHECK(m, off, hbhlen, -1);
856	hbh = (struct ip6_hbh *)(mtod(m, caddr_t) + off);
857#else
858	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m,
859		sizeof(struct ip6_hdr), sizeof(struct ip6_hbh));
860	if (hbh == NULL) {
861		ip6stat.ip6s_tooshort++;
862		return -1;
863	}
864	hbhlen = (hbh->ip6h_len + 1) << 3;
865	IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr),
866		hbhlen);
867	if (hbh == NULL) {
868		ip6stat.ip6s_tooshort++;
869		return -1;
870	}
871#endif
872	off += hbhlen;
873	hbhlen -= sizeof(struct ip6_hbh);
874	opt = (u_int8_t *)hbh + sizeof(struct ip6_hbh);
875
876	if (ip6_process_hopopts(m, (u_int8_t *)hbh + sizeof(struct ip6_hbh),
877				hbhlen, rtalertp, plenp) < 0)
878		return (-1);
879
880	*offp = off;
881	*mp = m;
882	return (0);
883}
884
885/*
886 * Search header for all Hop-by-hop options and process each option.
887 * This function is separate from ip6_hopopts_input() in order to
888 * handle a case where the sending node itself process its hop-by-hop
889 * options header. In such a case, the function is called from ip6_output().
890 *
891 * The function assumes that hbh header is located right after the IPv6 header
892 * (RFC2460 p7), opthead is pointer into data content in m, and opthead to
893 * opthead + hbhlen is located in continuous memory region.
894 */
895int
896ip6_process_hopopts(m, opthead, hbhlen, rtalertp, plenp)
897	struct mbuf *m;
898	u_int8_t *opthead;
899	int hbhlen;
900	u_int32_t *rtalertp;
901	u_int32_t *plenp;
902{
903	struct ip6_hdr *ip6;
904	int optlen = 0;
905	u_int8_t *opt = opthead;
906	u_int16_t rtalert_val;
907	u_int32_t jumboplen;
908	const int erroff = sizeof(struct ip6_hdr) + sizeof(struct ip6_hbh);
909
910	for (; hbhlen > 0; hbhlen -= optlen, opt += optlen) {
911		switch (*opt) {
912		case IP6OPT_PAD1:
913			optlen = 1;
914			break;
915		case IP6OPT_PADN:
916			if (hbhlen < IP6OPT_MINLEN) {
917				ip6stat.ip6s_toosmall++;
918				goto bad;
919			}
920			optlen = *(opt + 1) + 2;
921			break;
922		case IP6OPT_RTALERT:
923			/* XXX may need check for alignment */
924			if (hbhlen < IP6OPT_RTALERT_LEN) {
925				ip6stat.ip6s_toosmall++;
926				goto bad;
927			}
928			if (*(opt + 1) != IP6OPT_RTALERT_LEN - 2) {
929				/* XXX stat */
930				icmp6_error(m, ICMP6_PARAM_PROB,
931				    ICMP6_PARAMPROB_HEADER,
932				    erroff + opt + 1 - opthead);
933				return (-1);
934			}
935			optlen = IP6OPT_RTALERT_LEN;
936			bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2);
937			*rtalertp = ntohs(rtalert_val);
938			break;
939		case IP6OPT_JUMBO:
940			/* XXX may need check for alignment */
941			if (hbhlen < IP6OPT_JUMBO_LEN) {
942				ip6stat.ip6s_toosmall++;
943				goto bad;
944			}
945			if (*(opt + 1) != IP6OPT_JUMBO_LEN - 2) {
946				/* XXX stat */
947				icmp6_error(m, ICMP6_PARAM_PROB,
948				    ICMP6_PARAMPROB_HEADER,
949				    erroff + opt + 1 - opthead);
950				return (-1);
951			}
952			optlen = IP6OPT_JUMBO_LEN;
953
954			/*
955			 * IPv6 packets that have non 0 payload length
956			 * must not contain a jumbo payload option.
957			 */
958			ip6 = mtod(m, struct ip6_hdr *);
959			if (ip6->ip6_plen) {
960				ip6stat.ip6s_badoptions++;
961				icmp6_error(m, ICMP6_PARAM_PROB,
962				    ICMP6_PARAMPROB_HEADER,
963				    erroff + opt - opthead);
964				return (-1);
965			}
966
967			/*
968			 * We may see jumbolen in unaligned location, so
969			 * we'd need to perform bcopy().
970			 */
971			bcopy(opt + 2, &jumboplen, sizeof(jumboplen));
972			jumboplen = (u_int32_t)htonl(jumboplen);
973
974#if 1
975			/*
976			 * if there are multiple jumbo payload options,
977			 * *plenp will be non-zero and the packet will be
978			 * rejected.
979			 * the behavior may need some debate in ipngwg -
980			 * multiple options does not make sense, however,
981			 * there's no explicit mention in specification.
982			 */
983			if (*plenp != 0) {
984				ip6stat.ip6s_badoptions++;
985				icmp6_error(m, ICMP6_PARAM_PROB,
986				    ICMP6_PARAMPROB_HEADER,
987				    erroff + opt + 2 - opthead);
988				return (-1);
989			}
990#endif
991
992			/*
993			 * jumbo payload length must be larger than 65535.
994			 */
995			if (jumboplen <= IPV6_MAXPACKET) {
996				ip6stat.ip6s_badoptions++;
997				icmp6_error(m, ICMP6_PARAM_PROB,
998				    ICMP6_PARAMPROB_HEADER,
999				    erroff + opt + 2 - opthead);
1000				return (-1);
1001			}
1002			*plenp = jumboplen;
1003
1004			break;
1005		default:		/* unknown option */
1006			if (hbhlen < IP6OPT_MINLEN) {
1007				ip6stat.ip6s_toosmall++;
1008				goto bad;
1009			}
1010			optlen = ip6_unknown_opt(opt, m,
1011			    erroff + opt - opthead);
1012			if (optlen == -1)
1013				return (-1);
1014			optlen += 2;
1015			break;
1016		}
1017	}
1018
1019	return (0);
1020
1021  bad:
1022	m_freem(m);
1023	return (-1);
1024}
1025
1026/*
1027 * Unknown option processing.
1028 * The third argument `off' is the offset from the IPv6 header to the option,
1029 * which is necessary if the IPv6 header the and option header and IPv6 header
1030 * is not continuous in order to return an ICMPv6 error.
1031 */
1032int
1033ip6_unknown_opt(optp, m, off)
1034	u_int8_t *optp;
1035	struct mbuf *m;
1036	int off;
1037{
1038	struct ip6_hdr *ip6;
1039
1040	switch (IP6OPT_TYPE(*optp)) {
1041	case IP6OPT_TYPE_SKIP: /* ignore the option */
1042		return ((int)*(optp + 1));
1043	case IP6OPT_TYPE_DISCARD:	/* silently discard */
1044		m_freem(m);
1045		return (-1);
1046	case IP6OPT_TYPE_FORCEICMP: /* send ICMP even if multicasted */
1047		ip6stat.ip6s_badoptions++;
1048		icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_OPTION, off);
1049		return (-1);
1050	case IP6OPT_TYPE_ICMP: /* send ICMP if not multicasted */
1051		ip6stat.ip6s_badoptions++;
1052		ip6 = mtod(m, struct ip6_hdr *);
1053		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
1054		    (m->m_flags & (M_BCAST|M_MCAST)))
1055			m_freem(m);
1056		else
1057			icmp6_error(m, ICMP6_PARAM_PROB,
1058				    ICMP6_PARAMPROB_OPTION, off);
1059		return (-1);
1060	}
1061
1062	m_freem(m);		/* XXX: NOTREACHED */
1063	return (-1);
1064}
1065
1066/*
1067 * Create the "control" list for this pcb.
1068 * The function will not modify mbuf chain at all.
1069 *
1070 * with KAME mbuf chain restriction:
1071 * The routine will be called from upper layer handlers like tcp6_input().
1072 * Thus the routine assumes that the caller (tcp6_input) have already
1073 * called IP6_EXTHDR_CHECK() and all the extension headers are located in the
1074 * very first mbuf on the mbuf chain.
1075 */
1076void
1077ip6_savecontrol(in6p, mp, ip6, m)
1078	struct inpcb *in6p;
1079	struct mbuf **mp;
1080	struct ip6_hdr *ip6;
1081	struct mbuf *m;
1082{
1083#if __FreeBSD_version >= 500000
1084	struct thread *td = curthread;	/* XXX */
1085#else
1086	struct proc *td = curproc;	/* XXX */
1087#endif
1088	int privileged = 0;
1089	int rthdr_exist = 0;
1090
1091
1092	if (td && !suser(td))
1093		privileged++;
1094
1095#ifdef SO_TIMESTAMP
1096	if ((in6p->in6p_socket->so_options & SO_TIMESTAMP) != 0) {
1097		struct timeval tv;
1098
1099		microtime(&tv);
1100		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1101		    SCM_TIMESTAMP, SOL_SOCKET);
1102		if (*mp) {
1103			mp = &(*mp)->m_next;
1104		}
1105	}
1106#endif
1107
1108	/* RFC 2292 sec. 5 */
1109	if ((in6p->in6p_flags & IN6P_PKTINFO) != 0) {
1110		struct in6_pktinfo pi6;
1111
1112		bcopy(&ip6->ip6_dst, &pi6.ipi6_addr, sizeof(struct in6_addr));
1113		if (IN6_IS_SCOPE_LINKLOCAL(&pi6.ipi6_addr))
1114			pi6.ipi6_addr.s6_addr16[1] = 0;
1115		pi6.ipi6_ifindex =
1116		    (m && m->m_pkthdr.rcvif) ? m->m_pkthdr.rcvif->if_index : 0;
1117
1118		*mp = sbcreatecontrol((caddr_t) &pi6,
1119		    sizeof(struct in6_pktinfo),
1120		    IPV6_PKTINFO, IPPROTO_IPV6);
1121		if (*mp) {
1122			mp = &(*mp)->m_next;
1123		}
1124	}
1125
1126	if ((in6p->in6p_flags & IN6P_HOPLIMIT) != 0) {
1127		int hlim = ip6->ip6_hlim & 0xff;
1128
1129		*mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int),
1130		    IPV6_HOPLIMIT, IPPROTO_IPV6);
1131		if (*mp) {
1132			mp = &(*mp)->m_next;
1133		}
1134	}
1135
1136	/*
1137	 * IPV6_HOPOPTS socket option. We require super-user privilege
1138	 * for the option, but it might be too strict, since there might
1139	 * be some hop-by-hop options which can be returned to normal user.
1140	 * See RFC 2292 section 6.
1141	 */
1142	if ((in6p->in6p_flags & IN6P_HOPOPTS) != 0 && privileged) {
1143		/*
1144		 * Check if a hop-by-hop options header is contatined in the
1145		 * received packet, and if so, store the options as ancillary
1146		 * data. Note that a hop-by-hop options header must be
1147		 * just after the IPv6 header, which is assured through the
1148		 * IPv6 input processing.
1149		 */
1150		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1151		if (ip6->ip6_nxt == IPPROTO_HOPOPTS) {
1152			struct ip6_hbh *hbh;
1153			int hbhlen = 0;
1154#ifdef PULLDOWN_TEST
1155			struct mbuf *ext;
1156#endif
1157
1158#ifndef PULLDOWN_TEST
1159			hbh = (struct ip6_hbh *)(ip6 + 1);
1160			hbhlen = (hbh->ip6h_len + 1) << 3;
1161#else
1162			ext = ip6_pullexthdr(m, sizeof(struct ip6_hdr),
1163			    ip6->ip6_nxt);
1164			if (ext == NULL) {
1165				ip6stat.ip6s_tooshort++;
1166				return;
1167			}
1168			hbh = mtod(ext, struct ip6_hbh *);
1169			hbhlen = (hbh->ip6h_len + 1) << 3;
1170			if (hbhlen != ext->m_len) {
1171				m_freem(ext);
1172				ip6stat.ip6s_tooshort++;
1173				return;
1174			}
1175#endif
1176
1177			/*
1178			 * XXX: We copy the whole header even if a
1179			 * jumbo payload option is included, the option which
1180			 * is to be removed before returning according to
1181			 * RFC2292.
1182			 * Note: this constraint is removed in 2292bis.
1183			 */
1184			*mp = sbcreatecontrol((caddr_t)hbh, hbhlen,
1185			    IPV6_HOPOPTS, IPPROTO_IPV6);
1186			if (*mp) {
1187				mp = &(*mp)->m_next;
1188			}
1189#ifdef PULLDOWN_TEST
1190			m_freem(ext);
1191#endif
1192		}
1193	}
1194
1195	/* IPV6_DSTOPTS and IPV6_RTHDR socket options */
1196	if ((in6p->in6p_flags & (IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1197		int proto, off, nxt;
1198
1199		/*
1200		 * go through the header chain to see if a routing header is
1201		 * contained in the packet. We need this information to store
1202		 * destination options headers (if any) properly.
1203		 * XXX: performance issue. We should record this info when
1204		 * processing extension headers in incoming routine.
1205		 * (todo) use m_aux?
1206		 */
1207		proto = IPPROTO_IPV6;
1208		off = 0;
1209		nxt = -1;
1210		while (1) {
1211			int newoff;
1212
1213			newoff = ip6_nexthdr(m, off, proto, &nxt);
1214			if (newoff < 0)
1215				break;
1216			if (newoff < off) /* invalid, check for safety */
1217				break;
1218			if ((proto = nxt) == IPPROTO_ROUTING) {
1219				rthdr_exist = 1;
1220				break;
1221			}
1222			off = newoff;
1223		}
1224	}
1225
1226	if ((in6p->in6p_flags &
1227	     (IN6P_RTHDR | IN6P_DSTOPTS | IN6P_RTHDRDSTOPTS)) != 0) {
1228		struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1229		int nxt = ip6->ip6_nxt, off = sizeof(struct ip6_hdr);
1230
1231		/*
1232		 * Search for destination options headers or routing
1233		 * header(s) through the header chain, and stores each
1234		 * header as ancillary data.
1235		 * Note that the order of the headers remains in
1236		 * the chain of ancillary data.
1237		 */
1238		while (1) {	/* is explicit loop prevention necessary? */
1239			struct ip6_ext *ip6e = NULL;
1240			int elen;
1241#ifdef PULLDOWN_TEST
1242			struct mbuf *ext = NULL;
1243#endif
1244
1245			/*
1246			 * if it is not an extension header, don't try to
1247			 * pull it from the chain.
1248			 */
1249			switch (nxt) {
1250			case IPPROTO_DSTOPTS:
1251			case IPPROTO_ROUTING:
1252			case IPPROTO_HOPOPTS:
1253			case IPPROTO_AH: /* is it possible? */
1254				break;
1255			default:
1256				goto loopend;
1257			}
1258
1259#ifndef PULLDOWN_TEST
1260			if (off + sizeof(*ip6e) > m->m_len)
1261				goto loopend;
1262			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + off);
1263			if (nxt == IPPROTO_AH)
1264				elen = (ip6e->ip6e_len + 2) << 2;
1265			else
1266				elen = (ip6e->ip6e_len + 1) << 3;
1267			if (off + elen > m->m_len)
1268				goto loopend;
1269#else
1270			ext = ip6_pullexthdr(m, off, nxt);
1271			if (ext == NULL) {
1272				ip6stat.ip6s_tooshort++;
1273				return;
1274			}
1275			ip6e = mtod(ext, struct ip6_ext *);
1276			if (nxt == IPPROTO_AH)
1277				elen = (ip6e->ip6e_len + 2) << 2;
1278			else
1279				elen = (ip6e->ip6e_len + 1) << 3;
1280			if (elen != ext->m_len) {
1281				m_freem(ext);
1282				ip6stat.ip6s_tooshort++;
1283				return;
1284			}
1285#endif
1286
1287			switch (nxt) {
1288			case IPPROTO_DSTOPTS:
1289				if (!(in6p->in6p_flags & IN6P_DSTOPTS))
1290					break;
1291
1292				/*
1293				 * We also require super-user privilege for
1294				 * the option.  See comments on IN6_HOPOPTS.
1295				 */
1296				if (!privileged)
1297					break;
1298
1299				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
1300				    IPV6_DSTOPTS,
1301				    IPPROTO_IPV6);
1302				if (*mp)
1303					mp = &(*mp)->m_next;
1304				break;
1305			case IPPROTO_ROUTING:
1306				if (!in6p->in6p_flags & IN6P_RTHDR)
1307					break;
1308
1309				*mp = sbcreatecontrol((caddr_t)ip6e, elen,
1310				    IPV6_RTHDR,
1311				    IPPROTO_IPV6);
1312				if (*mp)
1313					mp = &(*mp)->m_next;
1314				break;
1315			case IPPROTO_HOPOPTS:
1316			case IPPROTO_AH: /* is it possible? */
1317				break;
1318
1319			default:
1320				/*
1321			 	 * other cases have been filtered in the above.
1322				 * none will visit this case.  here we supply
1323				 * the code just in case (nxt overwritten or
1324				 * other cases).
1325				 */
1326#ifdef PULLDOWN_TEST
1327				m_freem(ext);
1328#endif
1329				goto loopend;
1330
1331			}
1332
1333			/* proceed with the next header. */
1334			off += elen;
1335			nxt = ip6e->ip6e_nxt;
1336			ip6e = NULL;
1337#ifdef PULLDOWN_TEST
1338			m_freem(ext);
1339			ext = NULL;
1340#endif
1341		}
1342	  loopend:
1343		;
1344	}
1345
1346}
1347
1348#ifdef PULLDOWN_TEST
1349/*
1350 * pull single extension header from mbuf chain.  returns single mbuf that
1351 * contains the result, or NULL on error.
1352 */
1353static struct mbuf *
1354ip6_pullexthdr(m, off, nxt)
1355	struct mbuf *m;
1356	size_t off;
1357	int nxt;
1358{
1359	struct ip6_ext ip6e;
1360	size_t elen;
1361	struct mbuf *n;
1362
1363#ifdef DIAGNOSTIC
1364	switch (nxt) {
1365	case IPPROTO_DSTOPTS:
1366	case IPPROTO_ROUTING:
1367	case IPPROTO_HOPOPTS:
1368	case IPPROTO_AH: /* is it possible? */
1369		break;
1370	default:
1371		printf("ip6_pullexthdr: invalid nxt=%d\n", nxt);
1372	}
1373#endif
1374
1375	m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1376	if (nxt == IPPROTO_AH)
1377		elen = (ip6e.ip6e_len + 2) << 2;
1378	else
1379		elen = (ip6e.ip6e_len + 1) << 3;
1380
1381	MGET(n, M_DONTWAIT, MT_DATA);
1382	if (n && elen >= MLEN) {
1383		MCLGET(n, M_DONTWAIT);
1384		if ((n->m_flags & M_EXT) == 0) {
1385			m_free(n);
1386			n = NULL;
1387		}
1388	}
1389	if (!n)
1390		return NULL;
1391
1392	n->m_len = 0;
1393	if (elen >= M_TRAILINGSPACE(n)) {
1394		m_free(n);
1395		return NULL;
1396	}
1397
1398	m_copydata(m, off, elen, mtod(n, caddr_t));
1399	n->m_len = elen;
1400	return n;
1401}
1402#endif
1403
1404/*
1405 * Get pointer to the previous header followed by the header
1406 * currently processed.
1407 * XXX: This function supposes that
1408 *	M includes all headers,
1409 *	the next header field and the header length field of each header
1410 *	are valid, and
1411 *	the sum of each header length equals to OFF.
1412 * Because of these assumptions, this function must be called very
1413 * carefully. Moreover, it will not be used in the near future when
1414 * we develop `neater' mechanism to process extension headers.
1415 */
1416char *
1417ip6_get_prevhdr(m, off)
1418	struct mbuf *m;
1419	int off;
1420{
1421	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1422
1423	if (off == sizeof(struct ip6_hdr))
1424		return (&ip6->ip6_nxt);
1425	else {
1426		int len, nxt;
1427		struct ip6_ext *ip6e = NULL;
1428
1429		nxt = ip6->ip6_nxt;
1430		len = sizeof(struct ip6_hdr);
1431		while (len < off) {
1432			ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len);
1433
1434			switch (nxt) {
1435			case IPPROTO_FRAGMENT:
1436				len += sizeof(struct ip6_frag);
1437				break;
1438			case IPPROTO_AH:
1439				len += (ip6e->ip6e_len + 2) << 2;
1440				break;
1441			default:
1442				len += (ip6e->ip6e_len + 1) << 3;
1443				break;
1444			}
1445			nxt = ip6e->ip6e_nxt;
1446		}
1447		if (ip6e)
1448			return (&ip6e->ip6e_nxt);
1449		else
1450			return NULL;
1451	}
1452}
1453
1454/*
1455 * get next header offset.  m will be retained.
1456 */
1457int
1458ip6_nexthdr(m, off, proto, nxtp)
1459	struct mbuf *m;
1460	int off;
1461	int proto;
1462	int *nxtp;
1463{
1464	struct ip6_hdr ip6;
1465	struct ip6_ext ip6e;
1466	struct ip6_frag fh;
1467
1468	/* just in case */
1469	if (m == NULL)
1470		panic("ip6_nexthdr: m == NULL");
1471	if ((m->m_flags & M_PKTHDR) == 0 || m->m_pkthdr.len < off)
1472		return -1;
1473
1474	switch (proto) {
1475	case IPPROTO_IPV6:
1476		if (m->m_pkthdr.len < off + sizeof(ip6))
1477			return -1;
1478		m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6);
1479		if (nxtp)
1480			*nxtp = ip6.ip6_nxt;
1481		off += sizeof(ip6);
1482		return off;
1483
1484	case IPPROTO_FRAGMENT:
1485		/*
1486		 * terminate parsing if it is not the first fragment,
1487		 * it does not make sense to parse through it.
1488		 */
1489		if (m->m_pkthdr.len < off + sizeof(fh))
1490			return -1;
1491		m_copydata(m, off, sizeof(fh), (caddr_t)&fh);
1492		/* IP6F_OFF_MASK = 0xfff8(BigEndian), 0xf8ff(LittleEndian) */
1493		if (fh.ip6f_offlg & IP6F_OFF_MASK)
1494			return -1;
1495		if (nxtp)
1496			*nxtp = fh.ip6f_nxt;
1497		off += sizeof(struct ip6_frag);
1498		return off;
1499
1500	case IPPROTO_AH:
1501		if (m->m_pkthdr.len < off + sizeof(ip6e))
1502			return -1;
1503		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1504		if (nxtp)
1505			*nxtp = ip6e.ip6e_nxt;
1506		off += (ip6e.ip6e_len + 2) << 2;
1507		return off;
1508
1509	case IPPROTO_HOPOPTS:
1510	case IPPROTO_ROUTING:
1511	case IPPROTO_DSTOPTS:
1512		if (m->m_pkthdr.len < off + sizeof(ip6e))
1513			return -1;
1514		m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e);
1515		if (nxtp)
1516			*nxtp = ip6e.ip6e_nxt;
1517		off += (ip6e.ip6e_len + 1) << 3;
1518		return off;
1519
1520	case IPPROTO_NONE:
1521	case IPPROTO_ESP:
1522	case IPPROTO_IPCOMP:
1523		/* give up */
1524		return -1;
1525
1526	default:
1527		return -1;
1528	}
1529
1530	return -1;
1531}
1532
1533/*
1534 * get offset for the last header in the chain.  m will be kept untainted.
1535 */
1536int
1537ip6_lasthdr(m, off, proto, nxtp)
1538	struct mbuf *m;
1539	int off;
1540	int proto;
1541	int *nxtp;
1542{
1543	int newoff;
1544	int nxt;
1545
1546	if (!nxtp) {
1547		nxt = -1;
1548		nxtp = &nxt;
1549	}
1550	while (1) {
1551		newoff = ip6_nexthdr(m, off, proto, nxtp);
1552		if (newoff < 0)
1553			return off;
1554		else if (newoff < off)
1555			return -1;	/* invalid */
1556		else if (newoff == off)
1557			return newoff;
1558
1559		off = newoff;
1560		proto = *nxtp;
1561	}
1562}
1563
1564struct ip6aux *
1565ip6_addaux(m)
1566	struct mbuf *m;
1567{
1568	struct m_tag *mtag;
1569
1570	mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1571	if (!mtag) {
1572		mtag = m_tag_get(PACKET_TAG_IPV6_INPUT, sizeof(struct ip6aux),
1573		    M_NOWAIT);
1574		if (mtag)
1575			m_tag_prepend(m, mtag);
1576	}
1577	if (mtag)
1578		bzero(mtag+1, sizeof (struct ip6aux));
1579	return mtag ? (struct ip6aux*)(mtag+1) : NULL;
1580}
1581
1582struct ip6aux *
1583ip6_findaux(m)
1584	struct mbuf *m;
1585{
1586	struct m_tag *mtag;
1587
1588	mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1589	return mtag ? (struct ip6aux*)(mtag+1) : NULL;
1590}
1591
1592void
1593ip6_delaux(m)
1594	struct mbuf *m;
1595{
1596	struct m_tag *mtag;
1597
1598	mtag = m_tag_find(m, PACKET_TAG_IPV6_INPUT, NULL);
1599	if (mtag)
1600		m_tag_delete(m, mtag);
1601}
1602
1603/*
1604 * System control for IP6
1605 */
1606
1607u_char	inet6ctlerrmap[PRC_NCMDS] = {
1608	0,		0,		0,		0,
1609	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1610	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1611	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1612	0,		0,		0,		0,
1613	ENOPROTOOPT
1614};
1615