ip6_mroute.c revision 190012
1/*-
2 * Copyright (C) 1998 WIDE Project.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the project nor the names of its contributors
14 *    may be used to endorse or promote products derived from this software
15 *    without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 *	$KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $
30 */
31
32/*-
33 * Copyright (c) 1989 Stephen Deering
34 * Copyright (c) 1992, 1993
35 *      The Regents of the University of California.  All rights reserved.
36 *
37 * This code is derived from software contributed to Berkeley by
38 * Stephen Deering of Stanford University.
39 *
40 * Redistribution and use in source and binary forms, with or without
41 * modification, are permitted provided that the following conditions
42 * are met:
43 * 1. Redistributions of source code must retain the above copyright
44 *    notice, this list of conditions and the following disclaimer.
45 * 2. Redistributions in binary form must reproduce the above copyright
46 *    notice, this list of conditions and the following disclaimer in the
47 *    documentation and/or other materials provided with the distribution.
48 * 4. Neither the name of the University nor the names of its contributors
49 *    may be used to endorse or promote products derived from this software
50 *    without specific prior written permission.
51 *
52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62 * SUCH DAMAGE.
63 *
64 *	@(#)ip_mroute.c	8.2 (Berkeley) 11/15/93
65 *	BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp
66 */
67
68/*
69 * IP multicast forwarding procedures
70 *
71 * Written by David Waitzman, BBN Labs, August 1988.
72 * Modified by Steve Deering, Stanford, February 1989.
73 * Modified by Mark J. Steiglitz, Stanford, May, 1991
74 * Modified by Van Jacobson, LBL, January 1993
75 * Modified by Ajit Thyagarajan, PARC, August 1993
76 * Modified by Bill Fenner, PARC, April 1994
77 *
78 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
79 */
80
81#include <sys/cdefs.h>
82__FBSDID("$FreeBSD: head/sys/netinet6/ip6_mroute.c 190012 2009-03-19 01:43:03Z bms $");
83
84#include "opt_inet.h"
85#include "opt_inet6.h"
86#include "opt_route.h"
87
88#include <sys/param.h>
89#include <sys/callout.h>
90#include <sys/errno.h>
91#include <sys/kernel.h>
92#include <sys/lock.h>
93#include <sys/malloc.h>
94#include <sys/mbuf.h>
95#include <sys/module.h>
96#include <sys/protosw.h>
97#include <sys/signalvar.h>
98#include <sys/socket.h>
99#include <sys/socketvar.h>
100#include <sys/sockio.h>
101#include <sys/sx.h>
102#include <sys/sysctl.h>
103#include <sys/syslog.h>
104#include <sys/systm.h>
105#include <sys/time.h>
106#include <sys/vimage.h>
107
108#include <net/if.h>
109#include <net/if_types.h>
110#include <net/raw_cb.h>
111#include <net/route.h>
112#include <net/vnet.h>
113
114#include <netinet/in.h>
115#include <netinet/in_var.h>
116#include <netinet/icmp6.h>
117#include <netinet/vinet.h>
118#include <netinet/ip_encap.h>
119
120#include <netinet/ip6.h>
121#include <netinet6/ip6_var.h>
122#include <netinet6/scope6_var.h>
123#include <netinet6/nd6.h>
124#include <netinet6/ip6_mroute.h>
125#include <netinet6/ip6protosw.h>
126#include <netinet6/pim6.h>
127#include <netinet6/pim6_var.h>
128#include <netinet6/vinet6.h>
129
130static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
131
132/* XXX: this is a very common idiom; move to <sys/mbuf.h> ? */
133#define M_HASCL(m) ((m)->m_flags & M_EXT)
134
135static int	ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
136static void	phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
137static void	pim6_init(void);
138static int	register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
139static int	set_pim6(int *);
140static int	socket_send(struct socket *, struct mbuf *,
141		    struct sockaddr_in6 *);
142
143extern struct domain inet6domain;
144
145static const struct encaptab *pim6_encap_cookie;
146static const struct ip6protosw in6_pim_protosw = {
147	.pr_type =		SOCK_RAW,
148	.pr_domain =		&inet6domain,
149	.pr_protocol =		IPPROTO_PIM,
150	.pr_flags =		PR_ATOMIC|PR_ADDR|PR_LASTHDR,
151	.pr_input =		pim6_input,
152	.pr_output =		rip6_output,
153	.pr_ctloutput =		rip6_ctloutput,
154	.pr_init =		pim6_init,
155	.pr_usrreqs =		&rip6_usrreqs
156};
157static int pim6_encapcheck(const struct mbuf *, int, int, void *);
158
159#ifdef VIMAGE_GLOBALS
160static int ip6_mrouter_ver;
161#endif
162
163SYSCTL_DECL(_net_inet6);
164SYSCTL_DECL(_net_inet6_ip6);
165SYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim, CTLFLAG_RW, 0, "PIM");
166
167static struct mrt6stat mrt6stat;
168SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
169    &mrt6stat, mrt6stat,
170    "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
171
172#define NO_RTE_FOUND	0x1
173#define RTE_FOUND	0x2
174
175static struct mtx mrouter6_mtx;
176#define	MROUTER6_LOCK()		mtx_lock(&mrouter6_mtx)
177#define	MROUTER6_UNLOCK()	mtx_unlock(&mrouter6_mtx)
178#define	MROUTER6_LOCK_ASSERT()	do {					\
179	mtx_assert(&mrouter6_mtx, MA_OWNED);				\
180	NET_ASSERT_GIANT();						\
181} while (0)
182#define	MROUTER6_LOCK_INIT()	\
183	mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
184#define	MROUTER6_LOCK_DESTROY()	mtx_destroy(&mrouter6_mtx)
185
186static struct mf6c *mf6ctable[MF6CTBLSIZ];
187SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
188    &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
189    "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
190    "netinet6/ip6_mroute.h)");
191
192static struct mtx mfc6_mtx;
193#define	MFC6_LOCK()		mtx_lock(&mfc6_mtx)
194#define	MFC6_UNLOCK()		mtx_unlock(&mfc6_mtx)
195#define	MFC6_LOCK_ASSERT()	do {					\
196	mtx_assert(&mfc6_mtx, MA_OWNED);				\
197	NET_ASSERT_GIANT();						\
198} while (0)
199#define	MFC6_LOCK_INIT()		\
200	mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
201#define	MFC6_LOCK_DESTROY()	mtx_destroy(&mfc6_mtx)
202
203static u_char n6expire[MF6CTBLSIZ];
204
205static struct mif6 mif6table[MAXMIFS];
206SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mif6table, CTLFLAG_RD,
207    &mif6table, sizeof(mif6table), "S,mif6[MAXMIFS]",
208    "IPv6 Multicast Interfaces (struct mif6[MAXMIFS], netinet6/ip6_mroute.h)");
209
210static struct mtx mif6_mtx;
211#define	MIF6_LOCK()		mtx_lock(&mif6_mtx)
212#define	MIF6_UNLOCK()		mtx_unlock(&mif6_mtx)
213#define	MIF6_LOCK_ASSERT()	mtx_assert(&mif6_mtx, MA_OWNED)
214#define	MIF6_LOCK_INIT()	\
215	mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
216#define	MIF6_LOCK_DESTROY()	mtx_destroy(&mif6_mtx)
217
218#ifdef MRT6DEBUG
219#ifdef VIMAGE_GLOBALS
220static u_int mrt6debug = 0;		/* debug level */
221#endif
222#define DEBUG_MFC	0x02
223#define DEBUG_FORWARD	0x04
224#define DEBUG_EXPIRE	0x08
225#define DEBUG_XMIT	0x10
226#define DEBUG_REG	0x20
227#define DEBUG_PIM	0x40
228#endif
229
230static void	expire_upcalls(void *);
231#define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
232#define	UPCALL_EXPIRE	6		/* number of timeouts */
233
234/*
235 * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
236 */
237
238/*
239 * 'Interfaces' associated with decapsulator (so we can tell
240 * packets that went through it from ones that get reflected
241 * by a broken gateway).  Different from IPv4 register_if,
242 * these interfaces are linked into the system ifnet list,
243 * because per-interface IPv6 statistics are maintained in
244 * ifp->if_afdata.  But it does not have any routes point
245 * to them.  I.e., packets can't be sent this way.  They
246 * only exist as a placeholder for multicast source
247 * verification.
248 */
249static struct ifnet *multicast_register_if6;
250
251#define ENCAP_HOPS 64
252
253/*
254 * Private variables.
255 */
256static mifi_t nummifs = 0;
257static mifi_t reg_mif_num = (mifi_t)-1;
258
259static struct pim6stat pim6stat;
260SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RD,
261    &pim6stat, pim6stat,
262    "PIM Statistics (struct pim6stat, netinet6/pim_var.h)");
263
264#ifdef VIMAGE_GLOBALS
265static int pim6;
266#endif
267
268/*
269 * Hash function for a source, group entry
270 */
271#define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
272				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
273				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
274				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
275
276/*
277 * Find a route for a given origin IPv6 address and Multicast group address.
278 */
279#define MF6CFIND(o, g, rt) do { \
280	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
281	rt = NULL; \
282	mrt6stat.mrt6s_mfc_lookups++; \
283	while (_rt) { \
284		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
285		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
286		    (_rt->mf6c_stall == NULL)) { \
287			rt = _rt; \
288			break; \
289		} \
290		_rt = _rt->mf6c_next; \
291	} \
292	if (rt == NULL) { \
293		mrt6stat.mrt6s_mfc_misses++; \
294	} \
295} while (/*CONSTCOND*/ 0)
296
297/*
298 * Macros to compute elapsed time efficiently
299 * Borrowed from Van Jacobson's scheduling code
300 * XXX: replace with timersub() ?
301 */
302#define TV_DELTA(a, b, delta) do { \
303	    int xxs; \
304		\
305	    delta = (a).tv_usec - (b).tv_usec; \
306	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
307	       switch (xxs) { \
308		      case 2: \
309			  delta += 1000000; \
310			      /* FALLTHROUGH */ \
311		      case 1: \
312			  delta += 1000000; \
313			  break; \
314		      default: \
315			  delta += (1000000 * xxs); \
316	       } \
317	    } \
318} while (/*CONSTCOND*/ 0)
319
320/* XXX: replace with timercmp(a, b, <) ? */
321#define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
322	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
323
324#ifdef UPCALL_TIMING
325#define UPCALL_MAX	50
326static u_long upcall_data[UPCALL_MAX + 1];
327static void collate();
328#endif /* UPCALL_TIMING */
329
330static int ip6_mrouter_init(struct socket *, int, int);
331static int add_m6fc(struct mf6cctl *);
332static int add_m6if(struct mif6ctl *);
333static int del_m6fc(struct mf6cctl *);
334static int del_m6if(mifi_t *);
335static int del_m6if_locked(mifi_t *);
336static int get_mif6_cnt(struct sioc_mif_req6 *);
337static int get_sg_cnt(struct sioc_sg_req6 *);
338
339static struct callout expire_upcalls_ch;
340
341int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
342int X_ip6_mrouter_done(void);
343int X_ip6_mrouter_set(struct socket *, struct sockopt *);
344int X_ip6_mrouter_get(struct socket *, struct sockopt *);
345int X_mrt6_ioctl(int, caddr_t);
346
347static void
348pim6_init(void)
349{
350	INIT_VNET_INET6(curvnet);
351
352	V_ip6_mrouter_ver = 0;
353#ifdef MRT6DEBUG
354	V_mrt6debug = 0;	/* debug level */
355#endif
356}
357
358/*
359 * Handle MRT setsockopt commands to modify the multicast routing tables.
360 */
361int
362X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
363{
364	int error = 0;
365	int optval;
366	struct mif6ctl mifc;
367	struct mf6cctl mfcc;
368	mifi_t mifi;
369
370	if (so != ip6_mrouter && sopt->sopt_name != MRT6_INIT)
371		return (EACCES);
372
373	switch (sopt->sopt_name) {
374	case MRT6_INIT:
375#ifdef MRT6_OINIT
376	case MRT6_OINIT:
377#endif
378		error = sooptcopyin(sopt, &optval, sizeof(optval),
379		    sizeof(optval));
380		if (error)
381			break;
382		error = ip6_mrouter_init(so, optval, sopt->sopt_name);
383		break;
384	case MRT6_DONE:
385		error = X_ip6_mrouter_done();
386		break;
387	case MRT6_ADD_MIF:
388		error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
389		if (error)
390			break;
391		error = add_m6if(&mifc);
392		break;
393	case MRT6_ADD_MFC:
394		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
395		if (error)
396			break;
397		error = add_m6fc(&mfcc);
398		break;
399	case MRT6_DEL_MFC:
400		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
401		if (error)
402			break;
403		error = del_m6fc(&mfcc);
404		break;
405	case MRT6_DEL_MIF:
406		error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
407		if (error)
408			break;
409		error = del_m6if(&mifi);
410		break;
411	case MRT6_PIM:
412		error = sooptcopyin(sopt, &optval, sizeof(optval),
413		    sizeof(optval));
414		if (error)
415			break;
416		error = set_pim6(&optval);
417		break;
418	default:
419		error = EOPNOTSUPP;
420		break;
421	}
422
423	return (error);
424}
425
426/*
427 * Handle MRT getsockopt commands
428 */
429int
430X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
431{
432	INIT_VNET_INET6(curvnet);
433	int error = 0;
434
435	if (so != ip6_mrouter)
436		return (EACCES);
437
438	switch (sopt->sopt_name) {
439		case MRT6_PIM:
440			error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
441			break;
442	}
443	return (error);
444}
445
446/*
447 * Handle ioctl commands to obtain information from the cache
448 */
449int
450X_mrt6_ioctl(int cmd, caddr_t data)
451{
452	int ret;
453
454	ret = EINVAL;
455
456	switch (cmd) {
457	case SIOCGETSGCNT_IN6:
458		ret = get_sg_cnt((struct sioc_sg_req6 *)data);
459		break;
460
461	case SIOCGETMIFCNT_IN6:
462		ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
463		break;
464
465	default:
466		break;
467	}
468
469	return (ret);
470}
471
472/*
473 * returns the packet, byte, rpf-failure count for the source group provided
474 */
475static int
476get_sg_cnt(struct sioc_sg_req6 *req)
477{
478	struct mf6c *rt;
479	int ret;
480
481	ret = 0;
482
483	MFC6_LOCK();
484
485	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
486	if (rt == NULL) {
487		ret = ESRCH;
488	} else {
489		req->pktcnt = rt->mf6c_pkt_cnt;
490		req->bytecnt = rt->mf6c_byte_cnt;
491		req->wrong_if = rt->mf6c_wrong_if;
492	}
493
494	MFC6_UNLOCK();
495
496	return (ret);
497}
498
499/*
500 * returns the input and output packet and byte counts on the mif provided
501 */
502static int
503get_mif6_cnt(struct sioc_mif_req6 *req)
504{
505	mifi_t mifi;
506	int ret;
507
508	ret = 0;
509	mifi = req->mifi;
510
511	MIF6_LOCK();
512
513	if (mifi >= nummifs) {
514		ret = EINVAL;
515	} else {
516		req->icount = mif6table[mifi].m6_pkt_in;
517		req->ocount = mif6table[mifi].m6_pkt_out;
518		req->ibytes = mif6table[mifi].m6_bytes_in;
519		req->obytes = mif6table[mifi].m6_bytes_out;
520	}
521
522	MIF6_UNLOCK();
523
524	return (ret);
525}
526
527static int
528set_pim6(int *i)
529{
530	INIT_VNET_INET6(curvnet);
531	if ((*i != 1) && (*i != 0))
532		return (EINVAL);
533
534	V_pim6 = *i;
535
536	return (0);
537}
538
539/*
540 * Enable multicast routing
541 */
542static int
543ip6_mrouter_init(struct socket *so, int v, int cmd)
544{
545	INIT_VNET_INET6(curvnet);
546
547#ifdef MRT6DEBUG
548	if (V_mrt6debug)
549		log(LOG_DEBUG,
550		    "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
551		    so->so_type, so->so_proto->pr_protocol);
552#endif
553
554	if (so->so_type != SOCK_RAW ||
555	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
556		return (EOPNOTSUPP);
557
558	if (v != 1)
559		return (ENOPROTOOPT);
560
561	MROUTER6_LOCK();
562
563	if (ip6_mrouter != NULL) {
564		MROUTER6_UNLOCK();
565		return (EADDRINUSE);
566	}
567
568	ip6_mrouter = so;
569	V_ip6_mrouter_ver = cmd;
570
571	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
572	bzero((caddr_t)n6expire, sizeof(n6expire));
573
574	V_pim6 = 0;/* used for stubbing out/in pim stuff */
575
576	callout_init(&expire_upcalls_ch, 0);
577	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
578	    expire_upcalls, NULL);
579
580	MROUTER6_UNLOCK();
581
582#ifdef MRT6DEBUG
583	if (V_mrt6debug)
584		log(LOG_DEBUG, "ip6_mrouter_init\n");
585#endif
586
587	return (0);
588}
589
590/*
591 * Disable IPv6 multicast forwarding.
592 */
593int
594X_ip6_mrouter_done(void)
595{
596	INIT_VNET_INET6(curvnet);
597	mifi_t mifi;
598	int i;
599	struct mf6c *rt;
600	struct rtdetq *rte;
601
602	MROUTER6_LOCK();
603
604	if (ip6_mrouter == NULL) {
605		MROUTER6_UNLOCK();
606		return (EINVAL);
607	}
608
609	/*
610	 * For each phyint in use, disable promiscuous reception of all IPv6
611	 * multicasts.
612	 */
613	for (mifi = 0; mifi < nummifs; mifi++) {
614		if (mif6table[mifi].m6_ifp &&
615		    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
616			if_allmulti(mif6table[mifi].m6_ifp, 0);
617		}
618	}
619	bzero((caddr_t)mif6table, sizeof(mif6table));
620	nummifs = 0;
621
622	V_pim6 = 0; /* used to stub out/in pim specific code */
623
624	callout_stop(&expire_upcalls_ch);
625
626	/*
627	 * Free all multicast forwarding cache entries.
628	 */
629	MFC6_LOCK();
630	for (i = 0; i < MF6CTBLSIZ; i++) {
631		rt = mf6ctable[i];
632		while (rt) {
633			struct mf6c *frt;
634
635			for (rte = rt->mf6c_stall; rte != NULL; ) {
636				struct rtdetq *n = rte->next;
637
638				m_free(rte->m);
639				free(rte, M_MRTABLE6);
640				rte = n;
641			}
642			frt = rt;
643			rt = rt->mf6c_next;
644			free(frt, M_MRTABLE6);
645		}
646	}
647	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
648	MFC6_UNLOCK();
649
650	/*
651	 * Reset register interface
652	 */
653	if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
654		if_detach(multicast_register_if6);
655		if_free(multicast_register_if6);
656		reg_mif_num = (mifi_t)-1;
657		multicast_register_if6 = NULL;
658	}
659
660	ip6_mrouter = NULL;
661	V_ip6_mrouter_ver = 0;
662
663	MROUTER6_UNLOCK();
664
665#ifdef MRT6DEBUG
666	if (V_mrt6debug)
667		log(LOG_DEBUG, "ip6_mrouter_done\n");
668#endif
669
670	return (0);
671}
672
673static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
674
675/*
676 * Add a mif to the mif table
677 */
678static int
679add_m6if(struct mif6ctl *mifcp)
680{
681	INIT_VNET_NET(curvnet);
682	struct mif6 *mifp;
683	struct ifnet *ifp;
684	int error;
685
686	MIF6_LOCK();
687
688	if (mifcp->mif6c_mifi >= MAXMIFS) {
689		MIF6_UNLOCK();
690		return (EINVAL);
691	}
692	mifp = mif6table + mifcp->mif6c_mifi;
693	if (mifp->m6_ifp != NULL) {
694		MIF6_UNLOCK();
695		return (EADDRINUSE); /* XXX: is it appropriate? */
696	}
697	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
698		MIF6_UNLOCK();
699		return (ENXIO);
700	}
701
702	ifp = ifnet_byindex(mifcp->mif6c_pifi);
703
704	if (mifcp->mif6c_flags & MIFF_REGISTER) {
705		if (reg_mif_num == (mifi_t)-1) {
706			ifp = if_alloc(IFT_OTHER);
707
708			if_initname(ifp, "register_mif", 0);
709			ifp->if_flags |= IFF_LOOPBACK;
710			if_attach(ifp);
711			multicast_register_if6 = ifp;
712			reg_mif_num = mifcp->mif6c_mifi;
713			/*
714			 * it is impossible to guess the ifindex of the
715			 * register interface.  So mif6c_pifi is automatically
716			 * calculated.
717			 */
718			mifcp->mif6c_pifi = ifp->if_index;
719		} else {
720			ifp = multicast_register_if6;
721		}
722	} else {
723		/* Make sure the interface supports multicast */
724		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
725			MIF6_UNLOCK();
726			return (EOPNOTSUPP);
727		}
728
729		error = if_allmulti(ifp, 1);
730		if (error) {
731			MIF6_UNLOCK();
732			return (error);
733		}
734	}
735
736	mifp->m6_flags     = mifcp->mif6c_flags;
737	mifp->m6_ifp       = ifp;
738
739	/* initialize per mif pkt counters */
740	mifp->m6_pkt_in    = 0;
741	mifp->m6_pkt_out   = 0;
742	mifp->m6_bytes_in  = 0;
743	mifp->m6_bytes_out = 0;
744	bzero(&mifp->m6_route, sizeof(mifp->m6_route));
745
746	/* Adjust nummifs up if the mifi is higher than nummifs */
747	if (nummifs <= mifcp->mif6c_mifi)
748		nummifs = mifcp->mif6c_mifi + 1;
749
750	MIF6_UNLOCK();
751
752#ifdef MRT6DEBUG
753	if (V_mrt6debug)
754		log(LOG_DEBUG,
755		    "add_mif #%d, phyint %s\n",
756		    mifcp->mif6c_mifi,
757		    ifp->if_xname);
758#endif
759
760	return (0);
761}
762
763/*
764 * Delete a mif from the mif table
765 */
766static int
767del_m6if_locked(mifi_t *mifip)
768{
769	struct mif6 *mifp = mif6table + *mifip;
770	mifi_t mifi;
771	struct ifnet *ifp;
772
773	MIF6_LOCK_ASSERT();
774
775	if (*mifip >= nummifs)
776		return (EINVAL);
777	if (mifp->m6_ifp == NULL)
778		return (EINVAL);
779
780	if (!(mifp->m6_flags & MIFF_REGISTER)) {
781		/* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
782		ifp = mifp->m6_ifp;
783		if_allmulti(ifp, 0);
784	} else {
785		if (reg_mif_num != (mifi_t)-1 &&
786		    multicast_register_if6 != NULL) {
787			if_detach(multicast_register_if6);
788			if_free(multicast_register_if6);
789			reg_mif_num = (mifi_t)-1;
790			multicast_register_if6 = NULL;
791		}
792	}
793
794	bzero((caddr_t)mifp, sizeof(*mifp));
795
796	/* Adjust nummifs down */
797	for (mifi = nummifs; mifi > 0; mifi--)
798		if (mif6table[mifi - 1].m6_ifp)
799			break;
800	nummifs = mifi;
801
802#ifdef MRT6DEBUG
803	if (V_mrt6debug)
804		log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
805#endif
806
807	return (0);
808}
809
810static int
811del_m6if(mifi_t *mifip)
812{
813	int cc;
814
815	MIF6_LOCK();
816	cc = del_m6if_locked(mifip);
817	MIF6_UNLOCK();
818
819	return (cc);
820}
821
822/*
823 * Add an mfc entry
824 */
825static int
826add_m6fc(struct mf6cctl *mfccp)
827{
828	struct mf6c *rt;
829	u_long hash;
830	struct rtdetq *rte;
831	u_short nstl;
832	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
833
834	MFC6_LOCK();
835
836	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
837		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
838
839	/* If an entry already exists, just update the fields */
840	if (rt) {
841#ifdef MRT6DEBUG
842		if (V_mrt6debug & DEBUG_MFC) {
843		    log(LOG_DEBUG,
844			"add_m6fc no upcall h %d o %s g %s p %x\n",
845			ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
846			ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
847			mfccp->mf6cc_parent);
848		}
849#endif
850
851		rt->mf6c_parent = mfccp->mf6cc_parent;
852		rt->mf6c_ifset = mfccp->mf6cc_ifset;
853
854		MFC6_UNLOCK();
855		return (0);
856	}
857
858	/*
859	 * Find the entry for which the upcall was made and update
860	 */
861	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
862			mfccp->mf6cc_mcastgrp.sin6_addr);
863	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
864		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
865				       &mfccp->mf6cc_origin.sin6_addr) &&
866		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
867				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
868		    (rt->mf6c_stall != NULL)) {
869
870			if (nstl++)
871				log(LOG_ERR,
872				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
873				    "multiple kernel entries",
874				    ip6_sprintf(ip6bufo,
875					    &mfccp->mf6cc_origin.sin6_addr),
876				    ip6_sprintf(ip6bufg,
877					    &mfccp->mf6cc_mcastgrp.sin6_addr),
878				    mfccp->mf6cc_parent, rt->mf6c_stall);
879
880#ifdef MRT6DEBUG
881			if (V_mrt6debug & DEBUG_MFC)
882				log(LOG_DEBUG,
883				    "add_m6fc o %s g %s p %x dbg %x\n",
884				    ip6_sprintf(ip6bufo,
885					    &mfccp->mf6cc_origin.sin6_addr),
886				    ip6_sprintf(ip6bufg,
887					    &mfccp->mf6cc_mcastgrp.sin6_addr),
888				    mfccp->mf6cc_parent, rt->mf6c_stall);
889#endif
890
891			rt->mf6c_origin     = mfccp->mf6cc_origin;
892			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
893			rt->mf6c_parent     = mfccp->mf6cc_parent;
894			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
895			/* initialize pkt counters per src-grp */
896			rt->mf6c_pkt_cnt    = 0;
897			rt->mf6c_byte_cnt   = 0;
898			rt->mf6c_wrong_if   = 0;
899
900			rt->mf6c_expire = 0;	/* Don't clean this guy up */
901			n6expire[hash]--;
902
903			/* free packets Qed at the end of this entry */
904			for (rte = rt->mf6c_stall; rte != NULL; ) {
905				struct rtdetq *n = rte->next;
906				ip6_mdq(rte->m, rte->ifp, rt);
907				m_freem(rte->m);
908#ifdef UPCALL_TIMING
909				collate(&(rte->t));
910#endif /* UPCALL_TIMING */
911				free(rte, M_MRTABLE6);
912				rte = n;
913			}
914			rt->mf6c_stall = NULL;
915		}
916	}
917
918	/*
919	 * It is possible that an entry is being inserted without an upcall
920	 */
921	if (nstl == 0) {
922#ifdef MRT6DEBUG
923		if (V_mrt6debug & DEBUG_MFC)
924		    log(LOG_DEBUG,
925			"add_mfc no upcall h %d o %s g %s p %x\n",
926			hash,
927			ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
928			ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
929			mfccp->mf6cc_parent);
930#endif
931
932		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
933
934			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
935					       &mfccp->mf6cc_origin.sin6_addr)&&
936			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
937					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
938
939				rt->mf6c_origin     = mfccp->mf6cc_origin;
940				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
941				rt->mf6c_parent     = mfccp->mf6cc_parent;
942				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
943				/* initialize pkt counters per src-grp */
944				rt->mf6c_pkt_cnt    = 0;
945				rt->mf6c_byte_cnt   = 0;
946				rt->mf6c_wrong_if   = 0;
947
948				if (rt->mf6c_expire)
949					n6expire[hash]--;
950				rt->mf6c_expire	   = 0;
951			}
952		}
953		if (rt == NULL) {
954			/* no upcall, so make a new entry */
955			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
956						  M_NOWAIT);
957			if (rt == NULL) {
958				MFC6_UNLOCK();
959				return (ENOBUFS);
960			}
961
962			/* insert new entry at head of hash chain */
963			rt->mf6c_origin     = mfccp->mf6cc_origin;
964			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
965			rt->mf6c_parent     = mfccp->mf6cc_parent;
966			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
967			/* initialize pkt counters per src-grp */
968			rt->mf6c_pkt_cnt    = 0;
969			rt->mf6c_byte_cnt   = 0;
970			rt->mf6c_wrong_if   = 0;
971			rt->mf6c_expire     = 0;
972			rt->mf6c_stall = NULL;
973
974			/* link into table */
975			rt->mf6c_next  = mf6ctable[hash];
976			mf6ctable[hash] = rt;
977		}
978	}
979
980	MFC6_UNLOCK();
981	return (0);
982}
983
984#ifdef UPCALL_TIMING
985/*
986 * collect delay statistics on the upcalls
987 */
988static void
989collate(struct timeval *t)
990{
991	u_long d;
992	struct timeval tp;
993	u_long delta;
994
995	GET_TIME(tp);
996
997	if (TV_LT(*t, tp))
998	{
999		TV_DELTA(tp, *t, delta);
1000
1001		d = delta >> 10;
1002		if (d > UPCALL_MAX)
1003			d = UPCALL_MAX;
1004
1005		++upcall_data[d];
1006	}
1007}
1008#endif /* UPCALL_TIMING */
1009
1010/*
1011 * Delete an mfc entry
1012 */
1013static int
1014del_m6fc(struct mf6cctl *mfccp)
1015{
1016	struct sockaddr_in6	origin;
1017	struct sockaddr_in6	mcastgrp;
1018	struct mf6c		*rt;
1019	struct mf6c		**nptr;
1020	u_long		hash;
1021
1022	origin = mfccp->mf6cc_origin;
1023	mcastgrp = mfccp->mf6cc_mcastgrp;
1024	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
1025
1026#ifdef MRT6DEBUG
1027	if (V_mrt6debug & DEBUG_MFC) {
1028		char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1029		log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
1030		    ip6_sprintf(ip6bufo, &origin.sin6_addr),
1031		    ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
1032	}
1033#endif
1034
1035	MFC6_LOCK();
1036
1037	nptr = &mf6ctable[hash];
1038	while ((rt = *nptr) != NULL) {
1039		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
1040				       &rt->mf6c_origin.sin6_addr) &&
1041		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1042				       &rt->mf6c_mcastgrp.sin6_addr) &&
1043		    rt->mf6c_stall == NULL)
1044			break;
1045
1046		nptr = &rt->mf6c_next;
1047	}
1048	if (rt == NULL) {
1049		MFC6_UNLOCK();
1050		return (EADDRNOTAVAIL);
1051	}
1052
1053	*nptr = rt->mf6c_next;
1054	free(rt, M_MRTABLE6);
1055
1056	MFC6_UNLOCK();
1057
1058	return (0);
1059}
1060
1061static int
1062socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1063{
1064
1065	if (s) {
1066		if (sbappendaddr(&s->so_rcv,
1067				 (struct sockaddr *)src,
1068				 mm, (struct mbuf *)0) != 0) {
1069			sorwakeup(s);
1070			return (0);
1071		}
1072	}
1073	m_freem(mm);
1074	return (-1);
1075}
1076
1077/*
1078 * IPv6 multicast forwarding function. This function assumes that the packet
1079 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1080 * pointed to by "ifp", and the packet is to be relayed to other networks
1081 * that have members of the packet's destination IPv6 multicast group.
1082 *
1083 * The packet is returned unscathed to the caller, unless it is
1084 * erroneous, in which case a non-zero return value tells the caller to
1085 * discard it.
1086 *
1087 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1088 * this function is called in the originating context (i.e., not when
1089 * forwarding a packet from other node).  ip6_output(), which is currently the
1090 * only function that calls this function is called in the originating context,
1091 * explicitly ensures this condition.  It is caller's responsibility to ensure
1092 * that if this function is called from somewhere else in the originating
1093 * context in the future.
1094 */
1095int
1096X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1097{
1098	INIT_VNET_INET6(curvnet);
1099	struct mf6c *rt;
1100	struct mif6 *mifp;
1101	struct mbuf *mm;
1102	mifi_t mifi;
1103	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1104
1105#ifdef MRT6DEBUG
1106	if (V_mrt6debug & DEBUG_FORWARD)
1107		log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1108		    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1109		    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1110		    ifp->if_index);
1111#endif
1112
1113	/*
1114	 * Don't forward a packet with Hop limit of zero or one,
1115	 * or a packet destined to a local-only group.
1116	 */
1117	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1118	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1119		return (0);
1120	ip6->ip6_hlim--;
1121
1122	/*
1123	 * Source address check: do not forward packets with unspecified
1124	 * source. It was discussed in July 2000, on ipngwg mailing list.
1125	 * This is rather more serious than unicast cases, because some
1126	 * MLD packets can be sent with the unspecified source address
1127	 * (although such packets must normally set 1 to the hop limit field).
1128	 */
1129	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1130		V_ip6stat.ip6s_cantforward++;
1131		if (V_ip6_log_time + V_ip6_log_interval < time_second) {
1132			V_ip6_log_time = time_second;
1133			log(LOG_DEBUG,
1134			    "cannot forward "
1135			    "from %s to %s nxt %d received on %s\n",
1136			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1137			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1138			    ip6->ip6_nxt,
1139			    if_name(m->m_pkthdr.rcvif));
1140		}
1141		return (0);
1142	}
1143
1144	MFC6_LOCK();
1145
1146	/*
1147	 * Determine forwarding mifs from the forwarding cache table
1148	 */
1149	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1150
1151	/* Entry exists, so forward if necessary */
1152	if (rt) {
1153		MFC6_UNLOCK();
1154		return (ip6_mdq(m, ifp, rt));
1155	} else {
1156		/*
1157		 * If we don't have a route for packet's origin,
1158		 * Make a copy of the packet &
1159		 * send message to routing daemon
1160		 */
1161
1162		struct mbuf *mb0;
1163		struct rtdetq *rte;
1164		u_long hash;
1165/*		int i, npkts;*/
1166#ifdef UPCALL_TIMING
1167		struct timeval tp;
1168
1169		GET_TIME(tp);
1170#endif /* UPCALL_TIMING */
1171
1172		mrt6stat.mrt6s_no_route++;
1173#ifdef MRT6DEBUG
1174		if (V_mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1175			log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1176			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1177			    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1178#endif
1179
1180		/*
1181		 * Allocate mbufs early so that we don't do extra work if we
1182		 * are just going to fail anyway.
1183		 */
1184		rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6,
1185					      M_NOWAIT);
1186		if (rte == NULL) {
1187			MFC6_UNLOCK();
1188			return (ENOBUFS);
1189		}
1190		mb0 = m_copy(m, 0, M_COPYALL);
1191		/*
1192		 * Pullup packet header if needed before storing it,
1193		 * as other references may modify it in the meantime.
1194		 */
1195		if (mb0 &&
1196		    (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1197			mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1198		if (mb0 == NULL) {
1199			free(rte, M_MRTABLE6);
1200			MFC6_UNLOCK();
1201			return (ENOBUFS);
1202		}
1203
1204		/* is there an upcall waiting for this packet? */
1205		hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1206		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1207			if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1208					       &rt->mf6c_origin.sin6_addr) &&
1209			    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1210					       &rt->mf6c_mcastgrp.sin6_addr) &&
1211			    (rt->mf6c_stall != NULL))
1212				break;
1213		}
1214
1215		if (rt == NULL) {
1216			struct mrt6msg *im;
1217#ifdef MRT6_OINIT
1218			struct omrt6msg *oim;
1219#endif
1220
1221			/* no upcall, so make a new entry */
1222			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
1223						  M_NOWAIT);
1224			if (rt == NULL) {
1225				free(rte, M_MRTABLE6);
1226				m_freem(mb0);
1227				MFC6_UNLOCK();
1228				return (ENOBUFS);
1229			}
1230			/*
1231			 * Make a copy of the header to send to the user
1232			 * level process
1233			 */
1234			mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1235
1236			if (mm == NULL) {
1237				free(rte, M_MRTABLE6);
1238				m_freem(mb0);
1239				free(rt, M_MRTABLE6);
1240				MFC6_UNLOCK();
1241				return (ENOBUFS);
1242			}
1243
1244			/*
1245			 * Send message to routing daemon
1246			 */
1247			sin6.sin6_addr = ip6->ip6_src;
1248
1249			im = NULL;
1250#ifdef MRT6_OINIT
1251			oim = NULL;
1252#endif
1253			switch (V_ip6_mrouter_ver) {
1254#ifdef MRT6_OINIT
1255			case MRT6_OINIT:
1256				oim = mtod(mm, struct omrt6msg *);
1257				oim->im6_msgtype = MRT6MSG_NOCACHE;
1258				oim->im6_mbz = 0;
1259				break;
1260#endif
1261			case MRT6_INIT:
1262				im = mtod(mm, struct mrt6msg *);
1263				im->im6_msgtype = MRT6MSG_NOCACHE;
1264				im->im6_mbz = 0;
1265				break;
1266			default:
1267				free(rte, M_MRTABLE6);
1268				m_freem(mb0);
1269				free(rt, M_MRTABLE6);
1270				MFC6_UNLOCK();
1271				return (EINVAL);
1272			}
1273
1274#ifdef MRT6DEBUG
1275			if (V_mrt6debug & DEBUG_FORWARD)
1276				log(LOG_DEBUG,
1277				    "getting the iif info in the kernel\n");
1278#endif
1279
1280			for (mifp = mif6table, mifi = 0;
1281			     mifi < nummifs && mifp->m6_ifp != ifp;
1282			     mifp++, mifi++)
1283				;
1284
1285			switch (V_ip6_mrouter_ver) {
1286#ifdef MRT6_OINIT
1287			case MRT6_OINIT:
1288				oim->im6_mif = mifi;
1289				break;
1290#endif
1291			case MRT6_INIT:
1292				im->im6_mif = mifi;
1293				break;
1294			}
1295
1296			if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1297				log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1298				    "socket queue full\n");
1299				mrt6stat.mrt6s_upq_sockfull++;
1300				free(rte, M_MRTABLE6);
1301				m_freem(mb0);
1302				free(rt, M_MRTABLE6);
1303				MFC6_UNLOCK();
1304				return (ENOBUFS);
1305			}
1306
1307			mrt6stat.mrt6s_upcalls++;
1308
1309			/* insert new entry at head of hash chain */
1310			bzero(rt, sizeof(*rt));
1311			rt->mf6c_origin.sin6_family = AF_INET6;
1312			rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1313			rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1314			rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1315			rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1316			rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1317			rt->mf6c_expire = UPCALL_EXPIRE;
1318			n6expire[hash]++;
1319			rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1320
1321			/* link into table */
1322			rt->mf6c_next  = mf6ctable[hash];
1323			mf6ctable[hash] = rt;
1324			/* Add this entry to the end of the queue */
1325			rt->mf6c_stall = rte;
1326		} else {
1327			/* determine if q has overflowed */
1328			struct rtdetq **p;
1329			int npkts = 0;
1330
1331			for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1332				if (++npkts > MAX_UPQ6) {
1333					mrt6stat.mrt6s_upq_ovflw++;
1334					free(rte, M_MRTABLE6);
1335					m_freem(mb0);
1336					MFC6_UNLOCK();
1337					return (0);
1338				}
1339
1340			/* Add this entry to the end of the queue */
1341			*p = rte;
1342		}
1343
1344		rte->next = NULL;
1345		rte->m = mb0;
1346		rte->ifp = ifp;
1347#ifdef UPCALL_TIMING
1348		rte->t = tp;
1349#endif /* UPCALL_TIMING */
1350
1351		MFC6_UNLOCK();
1352
1353		return (0);
1354	}
1355}
1356
1357/*
1358 * Clean up cache entries if upcalls are not serviced
1359 * Call from the Slow Timeout mechanism, every half second.
1360 */
1361static void
1362expire_upcalls(void *unused)
1363{
1364	struct rtdetq *rte;
1365	struct mf6c *mfc, **nptr;
1366	int i;
1367
1368	MFC6_LOCK();
1369	for (i = 0; i < MF6CTBLSIZ; i++) {
1370		if (n6expire[i] == 0)
1371			continue;
1372		nptr = &mf6ctable[i];
1373		while ((mfc = *nptr) != NULL) {
1374			rte = mfc->mf6c_stall;
1375			/*
1376			 * Skip real cache entries
1377			 * Make sure it wasn't marked to not expire (shouldn't happen)
1378			 * If it expires now
1379			 */
1380			if (rte != NULL &&
1381			    mfc->mf6c_expire != 0 &&
1382			    --mfc->mf6c_expire == 0) {
1383#ifdef MRT6DEBUG
1384				if (V_mrt6debug & DEBUG_EXPIRE) {
1385					char ip6bufo[INET6_ADDRSTRLEN];
1386					char ip6bufg[INET6_ADDRSTRLEN];
1387					log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1388					    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1389					    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1390				}
1391#endif
1392				/*
1393				 * drop all the packets
1394				 * free the mbuf with the pkt, if, timing info
1395				 */
1396				do {
1397					struct rtdetq *n = rte->next;
1398					m_freem(rte->m);
1399					free(rte, M_MRTABLE6);
1400					rte = n;
1401				} while (rte != NULL);
1402				mrt6stat.mrt6s_cache_cleanups++;
1403				n6expire[i]--;
1404
1405				*nptr = mfc->mf6c_next;
1406				free(mfc, M_MRTABLE6);
1407			} else {
1408				nptr = &mfc->mf6c_next;
1409			}
1410		}
1411	}
1412	MFC6_UNLOCK();
1413	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1414	    expire_upcalls, NULL);
1415}
1416
1417/*
1418 * Packet forwarding routine once entry in the cache is made
1419 */
1420static int
1421ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1422{
1423	INIT_VNET_INET6(curvnet);
1424	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1425	mifi_t mifi, iif;
1426	struct mif6 *mifp;
1427	int plen = m->m_pkthdr.len;
1428	struct in6_addr src0, dst0; /* copies for local work */
1429	u_int32_t iszone, idzone, oszone, odzone;
1430	int error = 0;
1431
1432/*
1433 * Macro to send packet on mif.  Since RSVP packets don't get counted on
1434 * input, they shouldn't get counted on output, so statistics keeping is
1435 * separate.
1436 */
1437
1438#define MC6_SEND(ip6, mifp, m) do {				\
1439	if ((mifp)->m6_flags & MIFF_REGISTER)			\
1440		register_send((ip6), (mifp), (m));		\
1441	else							\
1442		phyint_send((ip6), (mifp), (m));		\
1443} while (/*CONSTCOND*/ 0)
1444
1445	/*
1446	 * Don't forward if it didn't arrive from the parent mif
1447	 * for its origin.
1448	 */
1449	mifi = rt->mf6c_parent;
1450	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1451		/* came in the wrong interface */
1452#ifdef MRT6DEBUG
1453		if (V_mrt6debug & DEBUG_FORWARD)
1454			log(LOG_DEBUG,
1455			    "wrong if: ifid %d mifi %d mififid %x\n",
1456			    ifp->if_index, mifi,
1457			    mif6table[mifi].m6_ifp->if_index);
1458#endif
1459		mrt6stat.mrt6s_wrong_if++;
1460		rt->mf6c_wrong_if++;
1461		/*
1462		 * If we are doing PIM processing, and we are forwarding
1463		 * packets on this interface, send a message to the
1464		 * routing daemon.
1465		 */
1466		/* have to make sure this is a valid mif */
1467		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1468			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1469				/*
1470				 * Check the M_LOOP flag to avoid an
1471				 * unnecessary PIM assert.
1472				 * XXX: M_LOOP is an ad-hoc hack...
1473				 */
1474				static struct sockaddr_in6 sin6 =
1475				{ sizeof(sin6), AF_INET6 };
1476
1477				struct mbuf *mm;
1478				struct mrt6msg *im;
1479#ifdef MRT6_OINIT
1480				struct omrt6msg *oim;
1481#endif
1482
1483				mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1484				if (mm &&
1485				    (M_HASCL(mm) ||
1486				     mm->m_len < sizeof(struct ip6_hdr)))
1487					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1488				if (mm == NULL)
1489					return (ENOBUFS);
1490
1491#ifdef MRT6_OINIT
1492				oim = NULL;
1493#endif
1494				im = NULL;
1495				switch (V_ip6_mrouter_ver) {
1496#ifdef MRT6_OINIT
1497				case MRT6_OINIT:
1498					oim = mtod(mm, struct omrt6msg *);
1499					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1500					oim->im6_mbz = 0;
1501					break;
1502#endif
1503				case MRT6_INIT:
1504					im = mtod(mm, struct mrt6msg *);
1505					im->im6_msgtype = MRT6MSG_WRONGMIF;
1506					im->im6_mbz = 0;
1507					break;
1508				default:
1509					m_freem(mm);
1510					return (EINVAL);
1511				}
1512
1513				for (mifp = mif6table, iif = 0;
1514				     iif < nummifs && mifp &&
1515					     mifp->m6_ifp != ifp;
1516				     mifp++, iif++)
1517					;
1518
1519				switch (V_ip6_mrouter_ver) {
1520#ifdef MRT6_OINIT
1521				case MRT6_OINIT:
1522					oim->im6_mif = iif;
1523					sin6.sin6_addr = oim->im6_src;
1524					break;
1525#endif
1526				case MRT6_INIT:
1527					im->im6_mif = iif;
1528					sin6.sin6_addr = im->im6_src;
1529					break;
1530				}
1531
1532				mrt6stat.mrt6s_upcalls++;
1533
1534				if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1535#ifdef MRT6DEBUG
1536					if (V_mrt6debug)
1537						log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1538#endif
1539					++mrt6stat.mrt6s_upq_sockfull;
1540					return (ENOBUFS);
1541				}	/* if socket Q full */
1542			}		/* if PIM */
1543		return (0);
1544	}			/* if wrong iif */
1545
1546	/* If I sourced this packet, it counts as output, else it was input. */
1547	if (m->m_pkthdr.rcvif == NULL) {
1548		/* XXX: is rcvif really NULL when output?? */
1549		mif6table[mifi].m6_pkt_out++;
1550		mif6table[mifi].m6_bytes_out += plen;
1551	} else {
1552		mif6table[mifi].m6_pkt_in++;
1553		mif6table[mifi].m6_bytes_in += plen;
1554	}
1555	rt->mf6c_pkt_cnt++;
1556	rt->mf6c_byte_cnt += plen;
1557
1558	/*
1559	 * For each mif, forward a copy of the packet if there are group
1560	 * members downstream on the interface.
1561	 */
1562	src0 = ip6->ip6_src;
1563	dst0 = ip6->ip6_dst;
1564	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1565	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1566		V_ip6stat.ip6s_badscope++;
1567		return (error);
1568	}
1569	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1570		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1571			/*
1572			 * check if the outgoing packet is going to break
1573			 * a scope boundary.
1574			 * XXX For packets through PIM register tunnel
1575			 * interface, we believe a routing daemon.
1576			 */
1577			if (!(mif6table[rt->mf6c_parent].m6_flags &
1578			      MIFF_REGISTER) &&
1579			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1580				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1581				    &oszone) ||
1582				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1583				    &odzone) ||
1584				    iszone != oszone ||
1585				    idzone != odzone) {
1586					V_ip6stat.ip6s_badscope++;
1587					continue;
1588				}
1589			}
1590
1591			mifp->m6_pkt_out++;
1592			mifp->m6_bytes_out += plen;
1593			MC6_SEND(ip6, mifp, m);
1594		}
1595	}
1596	return (0);
1597}
1598
1599static void
1600phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1601{
1602	INIT_VNET_INET6(curvnet);
1603	struct mbuf *mb_copy;
1604	struct ifnet *ifp = mifp->m6_ifp;
1605	int error = 0;
1606	struct	in6_multi *in6m;
1607	struct sockaddr_in6 *dst6;
1608	u_long linkmtu;
1609
1610	/*
1611	 * Make a new reference to the packet; make sure that
1612	 * the IPv6 header is actually copied, not just referenced,
1613	 * so that ip6_output() only scribbles on the copy.
1614	 */
1615	mb_copy = m_copy(m, 0, M_COPYALL);
1616	if (mb_copy &&
1617	    (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1618		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1619	if (mb_copy == NULL) {
1620		return;
1621	}
1622	/* set MCAST flag to the outgoing packet */
1623	mb_copy->m_flags |= M_MCAST;
1624
1625	/*
1626	 * If we sourced the packet, call ip6_output since we may devide
1627	 * the packet into fragments when the packet is too big for the
1628	 * outgoing interface.
1629	 * Otherwise, we can simply send the packet to the interface
1630	 * sending queue.
1631	 */
1632	if (m->m_pkthdr.rcvif == NULL) {
1633		struct ip6_moptions im6o;
1634
1635		im6o.im6o_multicast_ifp = ifp;
1636		/* XXX: ip6_output will override ip6->ip6_hlim */
1637		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1638		im6o.im6o_multicast_loop = 1;
1639		error = ip6_output(mb_copy, NULL, &mifp->m6_route,
1640				   IPV6_FORWARDING, &im6o, NULL, NULL);
1641
1642#ifdef MRT6DEBUG
1643		if (V_mrt6debug & DEBUG_XMIT)
1644			log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1645			    mifp - mif6table, error);
1646#endif
1647		return;
1648	}
1649
1650	/*
1651	 * If we belong to the destination multicast group
1652	 * on the outgoing interface, loop back a copy.
1653	 */
1654	dst6 = &mifp->m6_route.ro_dst;
1655	IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m);
1656	if (in6m != NULL) {
1657		dst6->sin6_len = sizeof(struct sockaddr_in6);
1658		dst6->sin6_family = AF_INET6;
1659		dst6->sin6_addr = ip6->ip6_dst;
1660		ip6_mloopback(ifp, m, &mifp->m6_route.ro_dst);
1661	}
1662	/*
1663	 * Put the packet into the sending queue of the outgoing interface
1664	 * if it would fit in the MTU of the interface.
1665	 */
1666	linkmtu = IN6_LINKMTU(ifp);
1667	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1668		dst6->sin6_len = sizeof(struct sockaddr_in6);
1669		dst6->sin6_family = AF_INET6;
1670		dst6->sin6_addr = ip6->ip6_dst;
1671		/*
1672		 * We just call if_output instead of nd6_output here, since
1673		 * we need no ND for a multicast forwarded packet...right?
1674		 */
1675		error = (*ifp->if_output)(ifp, mb_copy,
1676		    (struct sockaddr *)&mifp->m6_route.ro_dst, NULL);
1677#ifdef MRT6DEBUG
1678		if (V_mrt6debug & DEBUG_XMIT)
1679			log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1680			    mifp - mif6table, error);
1681#endif
1682	} else {
1683		/*
1684		 * pMTU discovery is intentionally disabled by default, since
1685		 * various router may notify pMTU in multicast, which can be
1686		 * a DDoS to a router
1687		 */
1688		if (V_ip6_mcast_pmtu)
1689			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1690		else {
1691#ifdef MRT6DEBUG
1692			if (V_mrt6debug & DEBUG_XMIT) {
1693				char ip6bufs[INET6_ADDRSTRLEN];
1694				char ip6bufd[INET6_ADDRSTRLEN];
1695				log(LOG_DEBUG,
1696				    "phyint_send: packet too big on %s o %s "
1697				    "g %s size %d(discarded)\n",
1698				    if_name(ifp),
1699				    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1700				    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1701				    mb_copy->m_pkthdr.len);
1702			}
1703#endif /* MRT6DEBUG */
1704			m_freem(mb_copy); /* simply discard the packet */
1705		}
1706	}
1707}
1708
1709static int
1710register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1711{
1712	struct mbuf *mm;
1713	int i, len = m->m_pkthdr.len;
1714	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1715	struct mrt6msg *im6;
1716
1717#ifdef MRT6DEBUG
1718	if (V_mrt6debug) {
1719		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1720		log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1721		    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1722		    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1723	}
1724#endif
1725	++pim6stat.pim6s_snd_registers;
1726
1727	/* Make a copy of the packet to send to the user level process */
1728	MGETHDR(mm, M_DONTWAIT, MT_HEADER);
1729	if (mm == NULL)
1730		return (ENOBUFS);
1731	mm->m_pkthdr.rcvif = NULL;
1732	mm->m_data += max_linkhdr;
1733	mm->m_len = sizeof(struct ip6_hdr);
1734
1735	if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1736		m_freem(mm);
1737		return (ENOBUFS);
1738	}
1739	i = MHLEN - M_LEADINGSPACE(mm);
1740	if (i > len)
1741		i = len;
1742	mm = m_pullup(mm, i);
1743	if (mm == NULL)
1744		return (ENOBUFS);
1745/* TODO: check it! */
1746	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1747
1748	/*
1749	 * Send message to routing daemon
1750	 */
1751	sin6.sin6_addr = ip6->ip6_src;
1752
1753	im6 = mtod(mm, struct mrt6msg *);
1754	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1755	im6->im6_mbz          = 0;
1756
1757	im6->im6_mif = mif - mif6table;
1758
1759	/* iif info is not given for reg. encap.n */
1760	mrt6stat.mrt6s_upcalls++;
1761
1762	if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1763#ifdef MRT6DEBUG
1764		if (V_mrt6debug)
1765			log(LOG_WARNING,
1766			    "register_send: ip6_mrouter socket queue full\n");
1767#endif
1768		++mrt6stat.mrt6s_upq_sockfull;
1769		return (ENOBUFS);
1770	}
1771	return (0);
1772}
1773
1774/*
1775 * pim6_encapcheck() is called by the encap6_input() path at runtime to
1776 * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1777 * into the kernel.
1778 */
1779static int
1780pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
1781{
1782
1783#ifdef DIAGNOSTIC
1784    KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1785#endif
1786    if (proto != IPPROTO_PIM)
1787	return 0;	/* not for us; reject the datagram. */
1788
1789    return 64;		/* claim the datagram. */
1790}
1791
1792/*
1793 * PIM sparse mode hook
1794 * Receives the pim control messages, and passes them up to the listening
1795 * socket, using rip6_input.
1796 * The only message processed is the REGISTER pim message; the pim header
1797 * is stripped off, and the inner packet is passed to register_mforward.
1798 */
1799int
1800pim6_input(struct mbuf **mp, int *offp, int proto)
1801{
1802	INIT_VNET_INET6(curvnet);
1803	struct pim *pim; /* pointer to a pim struct */
1804	struct ip6_hdr *ip6;
1805	int pimlen;
1806	struct mbuf *m = *mp;
1807	int minlen;
1808	int off = *offp;
1809
1810	++pim6stat.pim6s_rcv_total;
1811
1812	ip6 = mtod(m, struct ip6_hdr *);
1813	pimlen = m->m_pkthdr.len - *offp;
1814
1815	/*
1816	 * Validate lengths
1817	 */
1818	if (pimlen < PIM_MINLEN) {
1819		++pim6stat.pim6s_rcv_tooshort;
1820#ifdef MRT6DEBUG
1821		if (V_mrt6debug & DEBUG_PIM)
1822			log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1823#endif
1824		m_freem(m);
1825		return (IPPROTO_DONE);
1826	}
1827
1828	/*
1829	 * if the packet is at least as big as a REGISTER, go ahead
1830	 * and grab the PIM REGISTER header size, to avoid another
1831	 * possible m_pullup() later.
1832	 *
1833	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1834	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1835	 */
1836	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1837
1838	/*
1839	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1840	 * possibly the PIM REGISTER header
1841	 */
1842#ifndef PULLDOWN_TEST
1843	IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1844	/* adjust pointer */
1845	ip6 = mtod(m, struct ip6_hdr *);
1846
1847	/* adjust mbuf to point to the PIM header */
1848	pim = (struct pim *)((caddr_t)ip6 + off);
1849#else
1850	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1851	if (pim == NULL) {
1852		pim6stat.pim6s_rcv_tooshort++;
1853		return (IPPROTO_DONE);
1854	}
1855#endif
1856
1857#define PIM6_CHECKSUM
1858#ifdef PIM6_CHECKSUM
1859	{
1860		int cksumlen;
1861
1862		/*
1863		 * Validate checksum.
1864		 * If PIM REGISTER, exclude the data packet
1865		 */
1866		if (pim->pim_type == PIM_REGISTER)
1867			cksumlen = PIM_MINLEN;
1868		else
1869			cksumlen = pimlen;
1870
1871		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1872			++pim6stat.pim6s_rcv_badsum;
1873#ifdef MRT6DEBUG
1874			if (V_mrt6debug & DEBUG_PIM)
1875				log(LOG_DEBUG,
1876				    "pim6_input: invalid checksum\n");
1877#endif
1878			m_freem(m);
1879			return (IPPROTO_DONE);
1880		}
1881	}
1882#endif /* PIM_CHECKSUM */
1883
1884	/* PIM version check */
1885	if (pim->pim_ver != PIM_VERSION) {
1886		++pim6stat.pim6s_rcv_badversion;
1887#ifdef MRT6DEBUG
1888		log(LOG_ERR,
1889		    "pim6_input: incorrect version %d, expecting %d\n",
1890		    pim->pim_ver, PIM_VERSION);
1891#endif
1892		m_freem(m);
1893		return (IPPROTO_DONE);
1894	}
1895
1896	if (pim->pim_type == PIM_REGISTER) {
1897		/*
1898		 * since this is a REGISTER, we'll make a copy of the register
1899		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1900		 * routing daemon.
1901		 */
1902		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1903
1904		struct mbuf *mcp;
1905		struct ip6_hdr *eip6;
1906		u_int32_t *reghdr;
1907		int rc;
1908#ifdef MRT6DEBUG
1909		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1910#endif
1911
1912		++pim6stat.pim6s_rcv_registers;
1913
1914		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1915#ifdef MRT6DEBUG
1916			if (V_mrt6debug & DEBUG_PIM)
1917				log(LOG_DEBUG,
1918				    "pim6_input: register mif not set: %d\n",
1919				    reg_mif_num);
1920#endif
1921			m_freem(m);
1922			return (IPPROTO_DONE);
1923		}
1924
1925		reghdr = (u_int32_t *)(pim + 1);
1926
1927		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1928			goto pim6_input_to_daemon;
1929
1930		/*
1931		 * Validate length
1932		 */
1933		if (pimlen < PIM6_REG_MINLEN) {
1934			++pim6stat.pim6s_rcv_tooshort;
1935			++pim6stat.pim6s_rcv_badregisters;
1936#ifdef MRT6DEBUG
1937			log(LOG_ERR,
1938			    "pim6_input: register packet size too "
1939			    "small %d from %s\n",
1940			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1941#endif
1942			m_freem(m);
1943			return (IPPROTO_DONE);
1944		}
1945
1946		eip6 = (struct ip6_hdr *) (reghdr + 1);
1947#ifdef MRT6DEBUG
1948		if (V_mrt6debug & DEBUG_PIM)
1949			log(LOG_DEBUG,
1950			    "pim6_input[register], eip6: %s -> %s, "
1951			    "eip6 plen %d\n",
1952			    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1953			    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1954			    ntohs(eip6->ip6_plen));
1955#endif
1956
1957		/* verify the version number of the inner packet */
1958		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1959			++pim6stat.pim6s_rcv_badregisters;
1960#ifdef MRT6DEBUG
1961			log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1962			    "of the inner packet\n",
1963			    (eip6->ip6_vfc & IPV6_VERSION));
1964#endif
1965			m_freem(m);
1966			return (IPPROTO_NONE);
1967		}
1968
1969		/* verify the inner packet is destined to a mcast group */
1970		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1971			++pim6stat.pim6s_rcv_badregisters;
1972#ifdef MRT6DEBUG
1973			if (V_mrt6debug & DEBUG_PIM)
1974				log(LOG_DEBUG,
1975				    "pim6_input: inner packet of register "
1976				    "is not multicast %s\n",
1977				    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1978#endif
1979			m_freem(m);
1980			return (IPPROTO_DONE);
1981		}
1982
1983		/*
1984		 * make a copy of the whole header to pass to the daemon later.
1985		 */
1986		mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1987		if (mcp == NULL) {
1988#ifdef MRT6DEBUG
1989			log(LOG_ERR,
1990			    "pim6_input: pim register: "
1991			    "could not copy register head\n");
1992#endif
1993			m_freem(m);
1994			return (IPPROTO_DONE);
1995		}
1996
1997		/*
1998		 * forward the inner ip6 packet; point m_data at the inner ip6.
1999		 */
2000		m_adj(m, off + PIM_MINLEN);
2001#ifdef MRT6DEBUG
2002		if (V_mrt6debug & DEBUG_PIM) {
2003			log(LOG_DEBUG,
2004			    "pim6_input: forwarding decapsulated register: "
2005			    "src %s, dst %s, mif %d\n",
2006			    ip6_sprintf(ip6bufs, &eip6->ip6_src),
2007			    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
2008			    reg_mif_num);
2009		}
2010#endif
2011
2012		rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
2013				dst.sin6_family, 0);
2014
2015		/* prepare the register head to send to the mrouting daemon */
2016		m = mcp;
2017	}
2018
2019	/*
2020	 * Pass the PIM message up to the daemon; if it is a register message
2021	 * pass the 'head' only up to the daemon. This includes the
2022	 * encapsulator ip6 header, pim header, register header and the
2023	 * encapsulated ip6 header.
2024	 */
2025  pim6_input_to_daemon:
2026	rip6_input(&m, offp, proto);
2027	return (IPPROTO_DONE);
2028}
2029
2030static int
2031ip6_mroute_modevent(module_t mod, int type, void *unused)
2032{
2033
2034	switch (type) {
2035	case MOD_LOAD:
2036		MROUTER6_LOCK_INIT();
2037		MFC6_LOCK_INIT();
2038		MIF6_LOCK_INIT();
2039
2040		pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM,
2041			pim6_encapcheck,
2042			(const struct protosw *)&in6_pim_protosw, NULL);
2043		if (pim6_encap_cookie == NULL) {
2044			printf("ip6_mroute: unable to attach pim6 encap\n");
2045			MIF6_LOCK_DESTROY();
2046			MFC6_LOCK_DESTROY();
2047			MROUTER6_LOCK_DESTROY();
2048			return (EINVAL);
2049		}
2050
2051		ip6_mforward = X_ip6_mforward;
2052		ip6_mrouter_done = X_ip6_mrouter_done;
2053		ip6_mrouter_get = X_ip6_mrouter_get;
2054		ip6_mrouter_set = X_ip6_mrouter_set;
2055		mrt6_ioctl = X_mrt6_ioctl;
2056		break;
2057
2058	case MOD_UNLOAD:
2059		if (ip6_mrouter != NULL)
2060			return EINVAL;
2061
2062		if (pim6_encap_cookie) {
2063			encap_detach(pim6_encap_cookie);
2064			pim6_encap_cookie = NULL;
2065		}
2066		X_ip6_mrouter_done();
2067		ip6_mforward = NULL;
2068		ip6_mrouter_done = NULL;
2069		ip6_mrouter_get = NULL;
2070		ip6_mrouter_set = NULL;
2071		mrt6_ioctl = NULL;
2072
2073		MIF6_LOCK_DESTROY();
2074		MFC6_LOCK_DESTROY();
2075		MROUTER6_LOCK_DESTROY();
2076		break;
2077
2078	default:
2079		return (EOPNOTSUPP);
2080	}
2081
2082	return (0);
2083}
2084
2085static moduledata_t ip6_mroutemod = {
2086	"ip6_mroute",
2087	ip6_mroute_modevent,
2088	0
2089};
2090
2091DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PSEUDO, SI_ORDER_ANY);
2092