ip6_mroute.c revision 192318
1294464Sdes/*-
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 192318 2009-05-18 17:48:46Z bz $");
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/domain.h>
97#include <sys/protosw.h>
98#include <sys/signalvar.h>
99#include <sys/socket.h>
100#include <sys/socketvar.h>
101#include <sys/sockio.h>
102#include <sys/sx.h>
103#include <sys/sysctl.h>
104#include <sys/syslog.h>
105#include <sys/systm.h>
106#include <sys/time.h>
107#include <sys/vimage.h>
108
109#include <net/if.h>
110#include <net/if_types.h>
111#include <net/raw_cb.h>
112#include <net/route.h>
113#include <net/vnet.h>
114
115#include <netinet/in.h>
116#include <netinet/in_var.h>
117#include <netinet/icmp6.h>
118#include <netinet/vinet.h>
119#include <netinet/ip_encap.h>
120
121#include <netinet/ip6.h>
122#include <netinet6/ip6_var.h>
123#include <netinet6/scope6_var.h>
124#include <netinet6/nd6.h>
125#include <netinet6/ip6_mroute.h>
126#include <netinet6/ip6protosw.h>
127#include <netinet6/pim6.h>
128#include <netinet6/pim6_var.h>
129#include <netinet6/vinet6.h>
130
131static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
132
133/* XXX: this is a very common idiom; move to <sys/mbuf.h> ? */
134#define M_HASCL(m) ((m)->m_flags & M_EXT)
135
136static int	ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
137static void	phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
138static void	pim6_init(void);
139static int	register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
140static int	set_pim6(int *);
141static int	socket_send(struct socket *, struct mbuf *,
142		    struct sockaddr_in6 *);
143
144extern int in6_mcast_loop;
145extern struct domain inet6domain;
146
147static const struct encaptab *pim6_encap_cookie;
148static const struct ip6protosw in6_pim_protosw = {
149	.pr_type =		SOCK_RAW,
150	.pr_domain =		&inet6domain,
151	.pr_protocol =		IPPROTO_PIM,
152	.pr_flags =		PR_ATOMIC|PR_ADDR|PR_LASTHDR,
153	.pr_input =		pim6_input,
154	.pr_output =		rip6_output,
155	.pr_ctloutput =		rip6_ctloutput,
156	.pr_init =		pim6_init,
157	.pr_usrreqs =		&rip6_usrreqs
158};
159static int pim6_encapcheck(const struct mbuf *, int, int, void *);
160
161#ifdef VIMAGE_GLOBALS
162static int ip6_mrouter_ver;
163#endif
164
165SYSCTL_DECL(_net_inet6);
166SYSCTL_DECL(_net_inet6_ip6);
167SYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim, CTLFLAG_RW, 0, "PIM");
168
169static struct mrt6stat mrt6stat;
170SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
171    &mrt6stat, mrt6stat,
172    "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
173
174#define NO_RTE_FOUND	0x1
175#define RTE_FOUND	0x2
176
177static struct mtx mrouter6_mtx;
178#define	MROUTER6_LOCK()		mtx_lock(&mrouter6_mtx)
179#define	MROUTER6_UNLOCK()	mtx_unlock(&mrouter6_mtx)
180#define	MROUTER6_LOCK_ASSERT()	do {					\
181	mtx_assert(&mrouter6_mtx, MA_OWNED);				\
182	NET_ASSERT_GIANT();						\
183} while (0)
184#define	MROUTER6_LOCK_INIT()	\
185	mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
186#define	MROUTER6_LOCK_DESTROY()	mtx_destroy(&mrouter6_mtx)
187
188static struct mf6c *mf6ctable[MF6CTBLSIZ];
189SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
190    &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
191    "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
192    "netinet6/ip6_mroute.h)");
193
194static struct mtx mfc6_mtx;
195#define	MFC6_LOCK()		mtx_lock(&mfc6_mtx)
196#define	MFC6_UNLOCK()		mtx_unlock(&mfc6_mtx)
197#define	MFC6_LOCK_ASSERT()	do {					\
198	mtx_assert(&mfc6_mtx, MA_OWNED);				\
199	NET_ASSERT_GIANT();						\
200} while (0)
201#define	MFC6_LOCK_INIT()		\
202	mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
203#define	MFC6_LOCK_DESTROY()	mtx_destroy(&mfc6_mtx)
204
205static u_char n6expire[MF6CTBLSIZ];
206
207static struct mif6 mif6table[MAXMIFS];
208SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mif6table, CTLFLAG_RD,
209    &mif6table, sizeof(mif6table), "S,mif6[MAXMIFS]",
210    "IPv6 Multicast Interfaces (struct mif6[MAXMIFS], netinet6/ip6_mroute.h)");
211
212static struct mtx mif6_mtx;
213#define	MIF6_LOCK()		mtx_lock(&mif6_mtx)
214#define	MIF6_UNLOCK()		mtx_unlock(&mif6_mtx)
215#define	MIF6_LOCK_ASSERT()	mtx_assert(&mif6_mtx, MA_OWNED)
216#define	MIF6_LOCK_INIT()	\
217	mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
218#define	MIF6_LOCK_DESTROY()	mtx_destroy(&mif6_mtx)
219
220#ifdef MRT6DEBUG
221#ifdef VIMAGE_GLOBALS
222static u_int mrt6debug;		/* debug level */
223#endif
224#define DEBUG_MFC	0x02
225#define DEBUG_FORWARD	0x04
226#define DEBUG_EXPIRE	0x08
227#define DEBUG_XMIT	0x10
228#define DEBUG_REG	0x20
229#define DEBUG_PIM	0x40
230#endif
231
232static void	expire_upcalls(void *);
233#define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
234#define	UPCALL_EXPIRE	6		/* number of timeouts */
235
236/*
237 * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
238 */
239
240/*
241 * 'Interfaces' associated with decapsulator (so we can tell
242 * packets that went through it from ones that get reflected
243 * by a broken gateway).  Different from IPv4 register_if,
244 * these interfaces are linked into the system ifnet list,
245 * because per-interface IPv6 statistics are maintained in
246 * ifp->if_afdata.  But it does not have any routes point
247 * to them.  I.e., packets can't be sent this way.  They
248 * only exist as a placeholder for multicast source
249 * verification.
250 */
251static struct ifnet *multicast_register_if6;
252
253#define ENCAP_HOPS 64
254
255/*
256 * Private variables.
257 */
258static mifi_t nummifs = 0;
259static mifi_t reg_mif_num = (mifi_t)-1;
260
261static struct pim6stat pim6stat;
262SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RD,
263    &pim6stat, pim6stat,
264    "PIM Statistics (struct pim6stat, netinet6/pim_var.h)");
265
266#ifdef VIMAGE_GLOBALS
267static int pim6;
268#endif
269
270/*
271 * Hash function for a source, group entry
272 */
273#define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
274				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
275				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
276				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
277
278/*
279 * Find a route for a given origin IPv6 address and Multicast group address.
280 */
281#define MF6CFIND(o, g, rt) do { \
282	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
283	rt = NULL; \
284	mrt6stat.mrt6s_mfc_lookups++; \
285	while (_rt) { \
286		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
287		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
288		    (_rt->mf6c_stall == NULL)) { \
289			rt = _rt; \
290			break; \
291		} \
292		_rt = _rt->mf6c_next; \
293	} \
294	if (rt == NULL) { \
295		mrt6stat.mrt6s_mfc_misses++; \
296	} \
297} while (/*CONSTCOND*/ 0)
298
299/*
300 * Macros to compute elapsed time efficiently
301 * Borrowed from Van Jacobson's scheduling code
302 * XXX: replace with timersub() ?
303 */
304#define TV_DELTA(a, b, delta) do { \
305	    int xxs; \
306		\
307	    delta = (a).tv_usec - (b).tv_usec; \
308	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
309	       switch (xxs) { \
310		      case 2: \
311			  delta += 1000000; \
312			      /* FALLTHROUGH */ \
313		      case 1: \
314			  delta += 1000000; \
315			  break; \
316		      default: \
317			  delta += (1000000 * xxs); \
318	       } \
319	    } \
320} while (/*CONSTCOND*/ 0)
321
322/* XXX: replace with timercmp(a, b, <) ? */
323#define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
324	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
325
326#ifdef UPCALL_TIMING
327#define UPCALL_MAX	50
328static u_long upcall_data[UPCALL_MAX + 1];
329static void collate();
330#endif /* UPCALL_TIMING */
331
332static int ip6_mrouter_init(struct socket *, int, int);
333static int add_m6fc(struct mf6cctl *);
334static int add_m6if(struct mif6ctl *);
335static int del_m6fc(struct mf6cctl *);
336static int del_m6if(mifi_t *);
337static int del_m6if_locked(mifi_t *);
338static int get_mif6_cnt(struct sioc_mif_req6 *);
339static int get_sg_cnt(struct sioc_sg_req6 *);
340
341static struct callout expire_upcalls_ch;
342
343int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
344int X_ip6_mrouter_done(void);
345int X_ip6_mrouter_set(struct socket *, struct sockopt *);
346int X_ip6_mrouter_get(struct socket *, struct sockopt *);
347int X_mrt6_ioctl(int, caddr_t);
348
349static void
350pim6_init(void)
351{
352	INIT_VNET_INET6(curvnet);
353
354	V_ip6_mrouter_ver = 0;
355#ifdef MRT6DEBUG
356	V_mrt6debug = 0;	/* debug level */
357#endif
358}
359
360/*
361 * Handle MRT setsockopt commands to modify the multicast routing tables.
362 */
363int
364X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
365{
366	INIT_VNET_INET6(curvnet);
367	int error = 0;
368	int optval;
369	struct mif6ctl mifc;
370	struct mf6cctl mfcc;
371	mifi_t mifi;
372
373	if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
374		return (EACCES);
375
376	switch (sopt->sopt_name) {
377	case MRT6_INIT:
378#ifdef MRT6_OINIT
379	case MRT6_OINIT:
380#endif
381		error = sooptcopyin(sopt, &optval, sizeof(optval),
382		    sizeof(optval));
383		if (error)
384			break;
385		error = ip6_mrouter_init(so, optval, sopt->sopt_name);
386		break;
387	case MRT6_DONE:
388		error = X_ip6_mrouter_done();
389		break;
390	case MRT6_ADD_MIF:
391		error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
392		if (error)
393			break;
394		error = add_m6if(&mifc);
395		break;
396	case MRT6_ADD_MFC:
397		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
398		if (error)
399			break;
400		error = add_m6fc(&mfcc);
401		break;
402	case MRT6_DEL_MFC:
403		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
404		if (error)
405			break;
406		error = del_m6fc(&mfcc);
407		break;
408	case MRT6_DEL_MIF:
409		error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
410		if (error)
411			break;
412		error = del_m6if(&mifi);
413		break;
414	case MRT6_PIM:
415		error = sooptcopyin(sopt, &optval, sizeof(optval),
416		    sizeof(optval));
417		if (error)
418			break;
419		error = set_pim6(&optval);
420		break;
421	default:
422		error = EOPNOTSUPP;
423		break;
424	}
425
426	return (error);
427}
428
429/*
430 * Handle MRT getsockopt commands
431 */
432int
433X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
434{
435	INIT_VNET_INET6(curvnet);
436	int error = 0;
437
438	if (so != V_ip6_mrouter)
439		return (EACCES);
440
441	switch (sopt->sopt_name) {
442		case MRT6_PIM:
443			error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
444			break;
445	}
446	return (error);
447}
448
449/*
450 * Handle ioctl commands to obtain information from the cache
451 */
452int
453X_mrt6_ioctl(int cmd, caddr_t data)
454{
455	int ret;
456
457	ret = EINVAL;
458
459	switch (cmd) {
460	case SIOCGETSGCNT_IN6:
461		ret = get_sg_cnt((struct sioc_sg_req6 *)data);
462		break;
463
464	case SIOCGETMIFCNT_IN6:
465		ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
466		break;
467
468	default:
469		break;
470	}
471
472	return (ret);
473}
474
475/*
476 * returns the packet, byte, rpf-failure count for the source group provided
477 */
478static int
479get_sg_cnt(struct sioc_sg_req6 *req)
480{
481	struct mf6c *rt;
482	int ret;
483
484	ret = 0;
485
486	MFC6_LOCK();
487
488	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
489	if (rt == NULL) {
490		ret = ESRCH;
491	} else {
492		req->pktcnt = rt->mf6c_pkt_cnt;
493		req->bytecnt = rt->mf6c_byte_cnt;
494		req->wrong_if = rt->mf6c_wrong_if;
495	}
496
497	MFC6_UNLOCK();
498
499	return (ret);
500}
501
502/*
503 * returns the input and output packet and byte counts on the mif provided
504 */
505static int
506get_mif6_cnt(struct sioc_mif_req6 *req)
507{
508	mifi_t mifi;
509	int ret;
510
511	ret = 0;
512	mifi = req->mifi;
513
514	MIF6_LOCK();
515
516	if (mifi >= nummifs) {
517		ret = EINVAL;
518	} else {
519		req->icount = mif6table[mifi].m6_pkt_in;
520		req->ocount = mif6table[mifi].m6_pkt_out;
521		req->ibytes = mif6table[mifi].m6_bytes_in;
522		req->obytes = mif6table[mifi].m6_bytes_out;
523	}
524
525	MIF6_UNLOCK();
526
527	return (ret);
528}
529
530static int
531set_pim6(int *i)
532{
533	INIT_VNET_INET6(curvnet);
534	if ((*i != 1) && (*i != 0))
535		return (EINVAL);
536
537	V_pim6 = *i;
538
539	return (0);
540}
541
542/*
543 * Enable multicast routing
544 */
545static int
546ip6_mrouter_init(struct socket *so, int v, int cmd)
547{
548	INIT_VNET_INET6(curvnet);
549
550	V_ip6_mrouter_ver = 0;
551
552#ifdef MRT6DEBUG
553	V_mrt6debug = 0;
554
555	if (V_mrt6debug)
556		log(LOG_DEBUG,
557		    "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
558		    so->so_type, so->so_proto->pr_protocol);
559#endif
560
561	if (so->so_type != SOCK_RAW ||
562	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
563		return (EOPNOTSUPP);
564
565	if (v != 1)
566		return (ENOPROTOOPT);
567
568	MROUTER6_LOCK();
569
570	if (V_ip6_mrouter != NULL) {
571		MROUTER6_UNLOCK();
572		return (EADDRINUSE);
573	}
574
575	V_ip6_mrouter = so;
576	V_ip6_mrouter_ver = cmd;
577
578	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
579	bzero((caddr_t)n6expire, sizeof(n6expire));
580
581	V_pim6 = 0;/* used for stubbing out/in pim stuff */
582
583	callout_init(&expire_upcalls_ch, 0);
584	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
585	    expire_upcalls, NULL);
586
587	MROUTER6_UNLOCK();
588
589#ifdef MRT6DEBUG
590	if (V_mrt6debug)
591		log(LOG_DEBUG, "ip6_mrouter_init\n");
592#endif
593
594	return (0);
595}
596
597/*
598 * Disable IPv6 multicast forwarding.
599 */
600int
601X_ip6_mrouter_done(void)
602{
603	INIT_VNET_INET6(curvnet);
604	mifi_t mifi;
605	int i;
606	struct mf6c *rt;
607	struct rtdetq *rte;
608
609	MROUTER6_LOCK();
610
611	if (V_ip6_mrouter == NULL) {
612		MROUTER6_UNLOCK();
613		return (EINVAL);
614	}
615
616	/*
617	 * For each phyint in use, disable promiscuous reception of all IPv6
618	 * multicasts.
619	 */
620	for (mifi = 0; mifi < nummifs; mifi++) {
621		if (mif6table[mifi].m6_ifp &&
622		    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
623			if_allmulti(mif6table[mifi].m6_ifp, 0);
624		}
625	}
626	bzero((caddr_t)mif6table, sizeof(mif6table));
627	nummifs = 0;
628
629	V_pim6 = 0; /* used to stub out/in pim specific code */
630
631	callout_stop(&expire_upcalls_ch);
632
633	/*
634	 * Free all multicast forwarding cache entries.
635	 */
636	MFC6_LOCK();
637	for (i = 0; i < MF6CTBLSIZ; i++) {
638		rt = mf6ctable[i];
639		while (rt) {
640			struct mf6c *frt;
641
642			for (rte = rt->mf6c_stall; rte != NULL; ) {
643				struct rtdetq *n = rte->next;
644
645				m_free(rte->m);
646				free(rte, M_MRTABLE6);
647				rte = n;
648			}
649			frt = rt;
650			rt = rt->mf6c_next;
651			free(frt, M_MRTABLE6);
652		}
653	}
654	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
655	MFC6_UNLOCK();
656
657	/*
658	 * Reset register interface
659	 */
660	if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
661		if_detach(multicast_register_if6);
662		if_free(multicast_register_if6);
663		reg_mif_num = (mifi_t)-1;
664		multicast_register_if6 = NULL;
665	}
666
667	V_ip6_mrouter = NULL;
668	V_ip6_mrouter_ver = 0;
669
670	MROUTER6_UNLOCK();
671
672#ifdef MRT6DEBUG
673	if (V_mrt6debug)
674		log(LOG_DEBUG, "ip6_mrouter_done\n");
675#endif
676
677	return (0);
678}
679
680static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
681
682/*
683 * Add a mif to the mif table
684 */
685static int
686add_m6if(struct mif6ctl *mifcp)
687{
688	INIT_VNET_NET(curvnet);
689	struct mif6 *mifp;
690	struct ifnet *ifp;
691	int error;
692
693	MIF6_LOCK();
694
695	if (mifcp->mif6c_mifi >= MAXMIFS) {
696		MIF6_UNLOCK();
697		return (EINVAL);
698	}
699	mifp = mif6table + mifcp->mif6c_mifi;
700	if (mifp->m6_ifp != NULL) {
701		MIF6_UNLOCK();
702		return (EADDRINUSE); /* XXX: is it appropriate? */
703	}
704	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
705		MIF6_UNLOCK();
706		return (ENXIO);
707	}
708
709	ifp = ifnet_byindex(mifcp->mif6c_pifi);
710
711	if (mifcp->mif6c_flags & MIFF_REGISTER) {
712		if (reg_mif_num == (mifi_t)-1) {
713			ifp = if_alloc(IFT_OTHER);
714
715			if_initname(ifp, "register_mif", 0);
716			ifp->if_flags |= IFF_LOOPBACK;
717			if_attach(ifp);
718			multicast_register_if6 = ifp;
719			reg_mif_num = mifcp->mif6c_mifi;
720			/*
721			 * it is impossible to guess the ifindex of the
722			 * register interface.  So mif6c_pifi is automatically
723			 * calculated.
724			 */
725			mifcp->mif6c_pifi = ifp->if_index;
726		} else {
727			ifp = multicast_register_if6;
728		}
729	} else {
730		/* Make sure the interface supports multicast */
731		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
732			MIF6_UNLOCK();
733			return (EOPNOTSUPP);
734		}
735
736		error = if_allmulti(ifp, 1);
737		if (error) {
738			MIF6_UNLOCK();
739			return (error);
740		}
741	}
742
743	mifp->m6_flags     = mifcp->mif6c_flags;
744	mifp->m6_ifp       = ifp;
745
746	/* initialize per mif pkt counters */
747	mifp->m6_pkt_in    = 0;
748	mifp->m6_pkt_out   = 0;
749	mifp->m6_bytes_in  = 0;
750	mifp->m6_bytes_out = 0;
751	bzero(&mifp->m6_route, sizeof(mifp->m6_route));
752
753	/* Adjust nummifs up if the mifi is higher than nummifs */
754	if (nummifs <= mifcp->mif6c_mifi)
755		nummifs = mifcp->mif6c_mifi + 1;
756
757	MIF6_UNLOCK();
758
759#ifdef MRT6DEBUG
760	if (V_mrt6debug)
761		log(LOG_DEBUG,
762		    "add_mif #%d, phyint %s\n",
763		    mifcp->mif6c_mifi,
764		    ifp->if_xname);
765#endif
766
767	return (0);
768}
769
770/*
771 * Delete a mif from the mif table
772 */
773static int
774del_m6if_locked(mifi_t *mifip)
775{
776	struct mif6 *mifp = mif6table + *mifip;
777	mifi_t mifi;
778	struct ifnet *ifp;
779
780	MIF6_LOCK_ASSERT();
781
782	if (*mifip >= nummifs)
783		return (EINVAL);
784	if (mifp->m6_ifp == NULL)
785		return (EINVAL);
786
787	if (!(mifp->m6_flags & MIFF_REGISTER)) {
788		/* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
789		ifp = mifp->m6_ifp;
790		if_allmulti(ifp, 0);
791	} else {
792		if (reg_mif_num != (mifi_t)-1 &&
793		    multicast_register_if6 != NULL) {
794			if_detach(multicast_register_if6);
795			if_free(multicast_register_if6);
796			reg_mif_num = (mifi_t)-1;
797			multicast_register_if6 = NULL;
798		}
799	}
800
801	bzero((caddr_t)mifp, sizeof(*mifp));
802
803	/* Adjust nummifs down */
804	for (mifi = nummifs; mifi > 0; mifi--)
805		if (mif6table[mifi - 1].m6_ifp)
806			break;
807	nummifs = mifi;
808
809#ifdef MRT6DEBUG
810	if (V_mrt6debug)
811		log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
812#endif
813
814	return (0);
815}
816
817static int
818del_m6if(mifi_t *mifip)
819{
820	int cc;
821
822	MIF6_LOCK();
823	cc = del_m6if_locked(mifip);
824	MIF6_UNLOCK();
825
826	return (cc);
827}
828
829/*
830 * Add an mfc entry
831 */
832static int
833add_m6fc(struct mf6cctl *mfccp)
834{
835	struct mf6c *rt;
836	u_long hash;
837	struct rtdetq *rte;
838	u_short nstl;
839	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
840
841	MFC6_LOCK();
842
843	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
844		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
845
846	/* If an entry already exists, just update the fields */
847	if (rt) {
848#ifdef MRT6DEBUG
849		if (V_mrt6debug & DEBUG_MFC) {
850		    log(LOG_DEBUG,
851			"add_m6fc no upcall h %d o %s g %s p %x\n",
852			ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
853			ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
854			mfccp->mf6cc_parent);
855		}
856#endif
857
858		rt->mf6c_parent = mfccp->mf6cc_parent;
859		rt->mf6c_ifset = mfccp->mf6cc_ifset;
860
861		MFC6_UNLOCK();
862		return (0);
863	}
864
865	/*
866	 * Find the entry for which the upcall was made and update
867	 */
868	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
869			mfccp->mf6cc_mcastgrp.sin6_addr);
870	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
871		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
872				       &mfccp->mf6cc_origin.sin6_addr) &&
873		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
874				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
875		    (rt->mf6c_stall != NULL)) {
876
877			if (nstl++)
878				log(LOG_ERR,
879				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
880				    "multiple kernel entries",
881				    ip6_sprintf(ip6bufo,
882					    &mfccp->mf6cc_origin.sin6_addr),
883				    ip6_sprintf(ip6bufg,
884					    &mfccp->mf6cc_mcastgrp.sin6_addr),
885				    mfccp->mf6cc_parent, rt->mf6c_stall);
886
887#ifdef MRT6DEBUG
888			if (V_mrt6debug & DEBUG_MFC)
889				log(LOG_DEBUG,
890				    "add_m6fc o %s g %s p %x dbg %x\n",
891				    ip6_sprintf(ip6bufo,
892					    &mfccp->mf6cc_origin.sin6_addr),
893				    ip6_sprintf(ip6bufg,
894					    &mfccp->mf6cc_mcastgrp.sin6_addr),
895				    mfccp->mf6cc_parent, rt->mf6c_stall);
896#endif
897
898			rt->mf6c_origin     = mfccp->mf6cc_origin;
899			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
900			rt->mf6c_parent     = mfccp->mf6cc_parent;
901			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
902			/* initialize pkt counters per src-grp */
903			rt->mf6c_pkt_cnt    = 0;
904			rt->mf6c_byte_cnt   = 0;
905			rt->mf6c_wrong_if   = 0;
906
907			rt->mf6c_expire = 0;	/* Don't clean this guy up */
908			n6expire[hash]--;
909
910			/* free packets Qed at the end of this entry */
911			for (rte = rt->mf6c_stall; rte != NULL; ) {
912				struct rtdetq *n = rte->next;
913				ip6_mdq(rte->m, rte->ifp, rt);
914				m_freem(rte->m);
915#ifdef UPCALL_TIMING
916				collate(&(rte->t));
917#endif /* UPCALL_TIMING */
918				free(rte, M_MRTABLE6);
919				rte = n;
920			}
921			rt->mf6c_stall = NULL;
922		}
923	}
924
925	/*
926	 * It is possible that an entry is being inserted without an upcall
927	 */
928	if (nstl == 0) {
929#ifdef MRT6DEBUG
930		if (V_mrt6debug & DEBUG_MFC)
931		    log(LOG_DEBUG,
932			"add_mfc no upcall h %d o %s g %s p %x\n",
933			hash,
934			ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
935			ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
936			mfccp->mf6cc_parent);
937#endif
938
939		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
940
941			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
942					       &mfccp->mf6cc_origin.sin6_addr)&&
943			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
944					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
945
946				rt->mf6c_origin     = mfccp->mf6cc_origin;
947				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
948				rt->mf6c_parent     = mfccp->mf6cc_parent;
949				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
950				/* initialize pkt counters per src-grp */
951				rt->mf6c_pkt_cnt    = 0;
952				rt->mf6c_byte_cnt   = 0;
953				rt->mf6c_wrong_if   = 0;
954
955				if (rt->mf6c_expire)
956					n6expire[hash]--;
957				rt->mf6c_expire	   = 0;
958			}
959		}
960		if (rt == NULL) {
961			/* no upcall, so make a new entry */
962			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
963						  M_NOWAIT);
964			if (rt == NULL) {
965				MFC6_UNLOCK();
966				return (ENOBUFS);
967			}
968
969			/* insert new entry at head of hash chain */
970			rt->mf6c_origin     = mfccp->mf6cc_origin;
971			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
972			rt->mf6c_parent     = mfccp->mf6cc_parent;
973			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
974			/* initialize pkt counters per src-grp */
975			rt->mf6c_pkt_cnt    = 0;
976			rt->mf6c_byte_cnt   = 0;
977			rt->mf6c_wrong_if   = 0;
978			rt->mf6c_expire     = 0;
979			rt->mf6c_stall = NULL;
980
981			/* link into table */
982			rt->mf6c_next  = mf6ctable[hash];
983			mf6ctable[hash] = rt;
984		}
985	}
986
987	MFC6_UNLOCK();
988	return (0);
989}
990
991#ifdef UPCALL_TIMING
992/*
993 * collect delay statistics on the upcalls
994 */
995static void
996collate(struct timeval *t)
997{
998	u_long d;
999	struct timeval tp;
1000	u_long delta;
1001
1002	GET_TIME(tp);
1003
1004	if (TV_LT(*t, tp))
1005	{
1006		TV_DELTA(tp, *t, delta);
1007
1008		d = delta >> 10;
1009		if (d > UPCALL_MAX)
1010			d = UPCALL_MAX;
1011
1012		++upcall_data[d];
1013	}
1014}
1015#endif /* UPCALL_TIMING */
1016
1017/*
1018 * Delete an mfc entry
1019 */
1020static int
1021del_m6fc(struct mf6cctl *mfccp)
1022{
1023	struct sockaddr_in6	origin;
1024	struct sockaddr_in6	mcastgrp;
1025	struct mf6c		*rt;
1026	struct mf6c		**nptr;
1027	u_long		hash;
1028
1029	origin = mfccp->mf6cc_origin;
1030	mcastgrp = mfccp->mf6cc_mcastgrp;
1031	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
1032
1033#ifdef MRT6DEBUG
1034	if (V_mrt6debug & DEBUG_MFC) {
1035		char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1036		log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
1037		    ip6_sprintf(ip6bufo, &origin.sin6_addr),
1038		    ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
1039	}
1040#endif
1041
1042	MFC6_LOCK();
1043
1044	nptr = &mf6ctable[hash];
1045	while ((rt = *nptr) != NULL) {
1046		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
1047				       &rt->mf6c_origin.sin6_addr) &&
1048		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1049				       &rt->mf6c_mcastgrp.sin6_addr) &&
1050		    rt->mf6c_stall == NULL)
1051			break;
1052
1053		nptr = &rt->mf6c_next;
1054	}
1055	if (rt == NULL) {
1056		MFC6_UNLOCK();
1057		return (EADDRNOTAVAIL);
1058	}
1059
1060	*nptr = rt->mf6c_next;
1061	free(rt, M_MRTABLE6);
1062
1063	MFC6_UNLOCK();
1064
1065	return (0);
1066}
1067
1068static int
1069socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1070{
1071
1072	if (s) {
1073		if (sbappendaddr(&s->so_rcv,
1074				 (struct sockaddr *)src,
1075				 mm, (struct mbuf *)0) != 0) {
1076			sorwakeup(s);
1077			return (0);
1078		}
1079	}
1080	m_freem(mm);
1081	return (-1);
1082}
1083
1084/*
1085 * IPv6 multicast forwarding function. This function assumes that the packet
1086 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1087 * pointed to by "ifp", and the packet is to be relayed to other networks
1088 * that have members of the packet's destination IPv6 multicast group.
1089 *
1090 * The packet is returned unscathed to the caller, unless it is
1091 * erroneous, in which case a non-zero return value tells the caller to
1092 * discard it.
1093 *
1094 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1095 * this function is called in the originating context (i.e., not when
1096 * forwarding a packet from other node).  ip6_output(), which is currently the
1097 * only function that calls this function is called in the originating context,
1098 * explicitly ensures this condition.  It is caller's responsibility to ensure
1099 * that if this function is called from somewhere else in the originating
1100 * context in the future.
1101 */
1102int
1103X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1104{
1105	INIT_VNET_INET6(curvnet);
1106	struct mf6c *rt;
1107	struct mif6 *mifp;
1108	struct mbuf *mm;
1109	mifi_t mifi;
1110	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1111
1112#ifdef MRT6DEBUG
1113	if (V_mrt6debug & DEBUG_FORWARD)
1114		log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1115		    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1116		    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1117		    ifp->if_index);
1118#endif
1119
1120	/*
1121	 * Don't forward a packet with Hop limit of zero or one,
1122	 * or a packet destined to a local-only group.
1123	 */
1124	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1125	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1126		return (0);
1127	ip6->ip6_hlim--;
1128
1129	/*
1130	 * Source address check: do not forward packets with unspecified
1131	 * source. It was discussed in July 2000, on ipngwg mailing list.
1132	 * This is rather more serious than unicast cases, because some
1133	 * MLD packets can be sent with the unspecified source address
1134	 * (although such packets must normally set 1 to the hop limit field).
1135	 */
1136	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1137		V_ip6stat.ip6s_cantforward++;
1138		if (V_ip6_log_time + V_ip6_log_interval < time_second) {
1139			V_ip6_log_time = time_second;
1140			log(LOG_DEBUG,
1141			    "cannot forward "
1142			    "from %s to %s nxt %d received on %s\n",
1143			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1144			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1145			    ip6->ip6_nxt,
1146			    if_name(m->m_pkthdr.rcvif));
1147		}
1148		return (0);
1149	}
1150
1151	MFC6_LOCK();
1152
1153	/*
1154	 * Determine forwarding mifs from the forwarding cache table
1155	 */
1156	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1157
1158	/* Entry exists, so forward if necessary */
1159	if (rt) {
1160		MFC6_UNLOCK();
1161		return (ip6_mdq(m, ifp, rt));
1162	} else {
1163		/*
1164		 * If we don't have a route for packet's origin,
1165		 * Make a copy of the packet &
1166		 * send message to routing daemon
1167		 */
1168
1169		struct mbuf *mb0;
1170		struct rtdetq *rte;
1171		u_long hash;
1172/*		int i, npkts;*/
1173#ifdef UPCALL_TIMING
1174		struct timeval tp;
1175
1176		GET_TIME(tp);
1177#endif /* UPCALL_TIMING */
1178
1179		mrt6stat.mrt6s_no_route++;
1180#ifdef MRT6DEBUG
1181		if (V_mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1182			log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1183			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1184			    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1185#endif
1186
1187		/*
1188		 * Allocate mbufs early so that we don't do extra work if we
1189		 * are just going to fail anyway.
1190		 */
1191		rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6,
1192					      M_NOWAIT);
1193		if (rte == NULL) {
1194			MFC6_UNLOCK();
1195			return (ENOBUFS);
1196		}
1197		mb0 = m_copy(m, 0, M_COPYALL);
1198		/*
1199		 * Pullup packet header if needed before storing it,
1200		 * as other references may modify it in the meantime.
1201		 */
1202		if (mb0 &&
1203		    (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1204			mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1205		if (mb0 == NULL) {
1206			free(rte, M_MRTABLE6);
1207			MFC6_UNLOCK();
1208			return (ENOBUFS);
1209		}
1210
1211		/* is there an upcall waiting for this packet? */
1212		hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1213		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1214			if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1215					       &rt->mf6c_origin.sin6_addr) &&
1216			    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1217					       &rt->mf6c_mcastgrp.sin6_addr) &&
1218			    (rt->mf6c_stall != NULL))
1219				break;
1220		}
1221
1222		if (rt == NULL) {
1223			struct mrt6msg *im;
1224#ifdef MRT6_OINIT
1225			struct omrt6msg *oim;
1226#endif
1227
1228			/* no upcall, so make a new entry */
1229			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
1230						  M_NOWAIT);
1231			if (rt == NULL) {
1232				free(rte, M_MRTABLE6);
1233				m_freem(mb0);
1234				MFC6_UNLOCK();
1235				return (ENOBUFS);
1236			}
1237			/*
1238			 * Make a copy of the header to send to the user
1239			 * level process
1240			 */
1241			mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1242
1243			if (mm == NULL) {
1244				free(rte, M_MRTABLE6);
1245				m_freem(mb0);
1246				free(rt, M_MRTABLE6);
1247				MFC6_UNLOCK();
1248				return (ENOBUFS);
1249			}
1250
1251			/*
1252			 * Send message to routing daemon
1253			 */
1254			sin6.sin6_addr = ip6->ip6_src;
1255
1256			im = NULL;
1257#ifdef MRT6_OINIT
1258			oim = NULL;
1259#endif
1260			switch (V_ip6_mrouter_ver) {
1261#ifdef MRT6_OINIT
1262			case MRT6_OINIT:
1263				oim = mtod(mm, struct omrt6msg *);
1264				oim->im6_msgtype = MRT6MSG_NOCACHE;
1265				oim->im6_mbz = 0;
1266				break;
1267#endif
1268			case MRT6_INIT:
1269				im = mtod(mm, struct mrt6msg *);
1270				im->im6_msgtype = MRT6MSG_NOCACHE;
1271				im->im6_mbz = 0;
1272				break;
1273			default:
1274				free(rte, M_MRTABLE6);
1275				m_freem(mb0);
1276				free(rt, M_MRTABLE6);
1277				MFC6_UNLOCK();
1278				return (EINVAL);
1279			}
1280
1281#ifdef MRT6DEBUG
1282			if (V_mrt6debug & DEBUG_FORWARD)
1283				log(LOG_DEBUG,
1284				    "getting the iif info in the kernel\n");
1285#endif
1286
1287			for (mifp = mif6table, mifi = 0;
1288			     mifi < nummifs && mifp->m6_ifp != ifp;
1289			     mifp++, mifi++)
1290				;
1291
1292			switch (V_ip6_mrouter_ver) {
1293#ifdef MRT6_OINIT
1294			case MRT6_OINIT:
1295				oim->im6_mif = mifi;
1296				break;
1297#endif
1298			case MRT6_INIT:
1299				im->im6_mif = mifi;
1300				break;
1301			}
1302
1303			if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1304				log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1305				    "socket queue full\n");
1306				mrt6stat.mrt6s_upq_sockfull++;
1307				free(rte, M_MRTABLE6);
1308				m_freem(mb0);
1309				free(rt, M_MRTABLE6);
1310				MFC6_UNLOCK();
1311				return (ENOBUFS);
1312			}
1313
1314			mrt6stat.mrt6s_upcalls++;
1315
1316			/* insert new entry at head of hash chain */
1317			bzero(rt, sizeof(*rt));
1318			rt->mf6c_origin.sin6_family = AF_INET6;
1319			rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1320			rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1321			rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1322			rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1323			rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1324			rt->mf6c_expire = UPCALL_EXPIRE;
1325			n6expire[hash]++;
1326			rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1327
1328			/* link into table */
1329			rt->mf6c_next  = mf6ctable[hash];
1330			mf6ctable[hash] = rt;
1331			/* Add this entry to the end of the queue */
1332			rt->mf6c_stall = rte;
1333		} else {
1334			/* determine if q has overflowed */
1335			struct rtdetq **p;
1336			int npkts = 0;
1337
1338			for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1339				if (++npkts > MAX_UPQ6) {
1340					mrt6stat.mrt6s_upq_ovflw++;
1341					free(rte, M_MRTABLE6);
1342					m_freem(mb0);
1343					MFC6_UNLOCK();
1344					return (0);
1345				}
1346
1347			/* Add this entry to the end of the queue */
1348			*p = rte;
1349		}
1350
1351		rte->next = NULL;
1352		rte->m = mb0;
1353		rte->ifp = ifp;
1354#ifdef UPCALL_TIMING
1355		rte->t = tp;
1356#endif /* UPCALL_TIMING */
1357
1358		MFC6_UNLOCK();
1359
1360		return (0);
1361	}
1362}
1363
1364/*
1365 * Clean up cache entries if upcalls are not serviced
1366 * Call from the Slow Timeout mechanism, every half second.
1367 */
1368static void
1369expire_upcalls(void *unused)
1370{
1371	struct rtdetq *rte;
1372	struct mf6c *mfc, **nptr;
1373	int i;
1374
1375	MFC6_LOCK();
1376	for (i = 0; i < MF6CTBLSIZ; i++) {
1377		if (n6expire[i] == 0)
1378			continue;
1379		nptr = &mf6ctable[i];
1380		while ((mfc = *nptr) != NULL) {
1381			rte = mfc->mf6c_stall;
1382			/*
1383			 * Skip real cache entries
1384			 * Make sure it wasn't marked to not expire (shouldn't happen)
1385			 * If it expires now
1386			 */
1387			if (rte != NULL &&
1388			    mfc->mf6c_expire != 0 &&
1389			    --mfc->mf6c_expire == 0) {
1390#ifdef MRT6DEBUG
1391				if (V_mrt6debug & DEBUG_EXPIRE) {
1392					char ip6bufo[INET6_ADDRSTRLEN];
1393					char ip6bufg[INET6_ADDRSTRLEN];
1394					log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1395					    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1396					    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1397				}
1398#endif
1399				/*
1400				 * drop all the packets
1401				 * free the mbuf with the pkt, if, timing info
1402				 */
1403				do {
1404					struct rtdetq *n = rte->next;
1405					m_freem(rte->m);
1406					free(rte, M_MRTABLE6);
1407					rte = n;
1408				} while (rte != NULL);
1409				mrt6stat.mrt6s_cache_cleanups++;
1410				n6expire[i]--;
1411
1412				*nptr = mfc->mf6c_next;
1413				free(mfc, M_MRTABLE6);
1414			} else {
1415				nptr = &mfc->mf6c_next;
1416			}
1417		}
1418	}
1419	MFC6_UNLOCK();
1420	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1421	    expire_upcalls, NULL);
1422}
1423
1424/*
1425 * Packet forwarding routine once entry in the cache is made
1426 */
1427static int
1428ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1429{
1430	INIT_VNET_INET6(curvnet);
1431	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1432	mifi_t mifi, iif;
1433	struct mif6 *mifp;
1434	int plen = m->m_pkthdr.len;
1435	struct in6_addr src0, dst0; /* copies for local work */
1436	u_int32_t iszone, idzone, oszone, odzone;
1437	int error = 0;
1438
1439/*
1440 * Macro to send packet on mif.  Since RSVP packets don't get counted on
1441 * input, they shouldn't get counted on output, so statistics keeping is
1442 * separate.
1443 */
1444
1445#define MC6_SEND(ip6, mifp, m) do {				\
1446	if ((mifp)->m6_flags & MIFF_REGISTER)			\
1447		register_send((ip6), (mifp), (m));		\
1448	else							\
1449		phyint_send((ip6), (mifp), (m));		\
1450} while (/*CONSTCOND*/ 0)
1451
1452	/*
1453	 * Don't forward if it didn't arrive from the parent mif
1454	 * for its origin.
1455	 */
1456	mifi = rt->mf6c_parent;
1457	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1458		/* came in the wrong interface */
1459#ifdef MRT6DEBUG
1460		if (V_mrt6debug & DEBUG_FORWARD)
1461			log(LOG_DEBUG,
1462			    "wrong if: ifid %d mifi %d mififid %x\n",
1463			    ifp->if_index, mifi,
1464			    mif6table[mifi].m6_ifp->if_index);
1465#endif
1466		mrt6stat.mrt6s_wrong_if++;
1467		rt->mf6c_wrong_if++;
1468		/*
1469		 * If we are doing PIM processing, and we are forwarding
1470		 * packets on this interface, send a message to the
1471		 * routing daemon.
1472		 */
1473		/* have to make sure this is a valid mif */
1474		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1475			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1476				/*
1477				 * Check the M_LOOP flag to avoid an
1478				 * unnecessary PIM assert.
1479				 * XXX: M_LOOP is an ad-hoc hack...
1480				 */
1481				static struct sockaddr_in6 sin6 =
1482				{ sizeof(sin6), AF_INET6 };
1483
1484				struct mbuf *mm;
1485				struct mrt6msg *im;
1486#ifdef MRT6_OINIT
1487				struct omrt6msg *oim;
1488#endif
1489
1490				mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1491				if (mm &&
1492				    (M_HASCL(mm) ||
1493				     mm->m_len < sizeof(struct ip6_hdr)))
1494					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1495				if (mm == NULL)
1496					return (ENOBUFS);
1497
1498#ifdef MRT6_OINIT
1499				oim = NULL;
1500#endif
1501				im = NULL;
1502				switch (V_ip6_mrouter_ver) {
1503#ifdef MRT6_OINIT
1504				case MRT6_OINIT:
1505					oim = mtod(mm, struct omrt6msg *);
1506					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1507					oim->im6_mbz = 0;
1508					break;
1509#endif
1510				case MRT6_INIT:
1511					im = mtod(mm, struct mrt6msg *);
1512					im->im6_msgtype = MRT6MSG_WRONGMIF;
1513					im->im6_mbz = 0;
1514					break;
1515				default:
1516					m_freem(mm);
1517					return (EINVAL);
1518				}
1519
1520				for (mifp = mif6table, iif = 0;
1521				     iif < nummifs && mifp &&
1522					     mifp->m6_ifp != ifp;
1523				     mifp++, iif++)
1524					;
1525
1526				switch (V_ip6_mrouter_ver) {
1527#ifdef MRT6_OINIT
1528				case MRT6_OINIT:
1529					oim->im6_mif = iif;
1530					sin6.sin6_addr = oim->im6_src;
1531					break;
1532#endif
1533				case MRT6_INIT:
1534					im->im6_mif = iif;
1535					sin6.sin6_addr = im->im6_src;
1536					break;
1537				}
1538
1539				mrt6stat.mrt6s_upcalls++;
1540
1541				if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1542#ifdef MRT6DEBUG
1543					if (V_mrt6debug)
1544						log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1545#endif
1546					++mrt6stat.mrt6s_upq_sockfull;
1547					return (ENOBUFS);
1548				}	/* if socket Q full */
1549			}		/* if PIM */
1550		return (0);
1551	}			/* if wrong iif */
1552
1553	/* If I sourced this packet, it counts as output, else it was input. */
1554	if (m->m_pkthdr.rcvif == NULL) {
1555		/* XXX: is rcvif really NULL when output?? */
1556		mif6table[mifi].m6_pkt_out++;
1557		mif6table[mifi].m6_bytes_out += plen;
1558	} else {
1559		mif6table[mifi].m6_pkt_in++;
1560		mif6table[mifi].m6_bytes_in += plen;
1561	}
1562	rt->mf6c_pkt_cnt++;
1563	rt->mf6c_byte_cnt += plen;
1564
1565	/*
1566	 * For each mif, forward a copy of the packet if there are group
1567	 * members downstream on the interface.
1568	 */
1569	src0 = ip6->ip6_src;
1570	dst0 = ip6->ip6_dst;
1571	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1572	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1573		V_ip6stat.ip6s_badscope++;
1574		return (error);
1575	}
1576	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1577		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1578			/*
1579			 * check if the outgoing packet is going to break
1580			 * a scope boundary.
1581			 * XXX For packets through PIM register tunnel
1582			 * interface, we believe a routing daemon.
1583			 */
1584			if (!(mif6table[rt->mf6c_parent].m6_flags &
1585			      MIFF_REGISTER) &&
1586			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1587				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1588				    &oszone) ||
1589				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1590				    &odzone) ||
1591				    iszone != oszone ||
1592				    idzone != odzone) {
1593					V_ip6stat.ip6s_badscope++;
1594					continue;
1595				}
1596			}
1597
1598			mifp->m6_pkt_out++;
1599			mifp->m6_bytes_out += plen;
1600			MC6_SEND(ip6, mifp, m);
1601		}
1602	}
1603	return (0);
1604}
1605
1606static void
1607phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1608{
1609	INIT_VNET_INET6(curvnet);
1610	struct mbuf *mb_copy;
1611	struct ifnet *ifp = mifp->m6_ifp;
1612	int error = 0;
1613	struct sockaddr_in6 *dst6;
1614	u_long linkmtu;
1615
1616	dst6 = &mifp->m6_route.ro_dst;
1617
1618	/*
1619	 * Make a new reference to the packet; make sure that
1620	 * the IPv6 header is actually copied, not just referenced,
1621	 * so that ip6_output() only scribbles on the copy.
1622	 */
1623	mb_copy = m_copy(m, 0, M_COPYALL);
1624	if (mb_copy &&
1625	    (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1626		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1627	if (mb_copy == NULL) {
1628		return;
1629	}
1630	/* set MCAST flag to the outgoing packet */
1631	mb_copy->m_flags |= M_MCAST;
1632
1633	/*
1634	 * If we sourced the packet, call ip6_output since we may devide
1635	 * the packet into fragments when the packet is too big for the
1636	 * outgoing interface.
1637	 * Otherwise, we can simply send the packet to the interface
1638	 * sending queue.
1639	 */
1640	if (m->m_pkthdr.rcvif == NULL) {
1641		struct ip6_moptions im6o;
1642
1643		im6o.im6o_multicast_ifp = ifp;
1644		/* XXX: ip6_output will override ip6->ip6_hlim */
1645		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1646		im6o.im6o_multicast_loop = 1;
1647		error = ip6_output(mb_copy, NULL, &mifp->m6_route,
1648				   IPV6_FORWARDING, &im6o, NULL, NULL);
1649
1650#ifdef MRT6DEBUG
1651		if (V_mrt6debug & DEBUG_XMIT)
1652			log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1653			    mifp - mif6table, error);
1654#endif
1655		return;
1656	}
1657
1658	/*
1659	 * If configured to loop back multicasts by default,
1660	 * loop back a copy now.
1661	 */
1662	if (in6_mcast_loop) {
1663		dst6->sin6_len = sizeof(struct sockaddr_in6);
1664		dst6->sin6_family = AF_INET6;
1665		dst6->sin6_addr = ip6->ip6_dst;
1666		ip6_mloopback(ifp, m, &mifp->m6_route.ro_dst);
1667	}
1668
1669	/*
1670	 * Put the packet into the sending queue of the outgoing interface
1671	 * if it would fit in the MTU of the interface.
1672	 */
1673	linkmtu = IN6_LINKMTU(ifp);
1674	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1675		dst6->sin6_len = sizeof(struct sockaddr_in6);
1676		dst6->sin6_family = AF_INET6;
1677		dst6->sin6_addr = ip6->ip6_dst;
1678		/*
1679		 * We just call if_output instead of nd6_output here, since
1680		 * we need no ND for a multicast forwarded packet...right?
1681		 */
1682		error = (*ifp->if_output)(ifp, mb_copy,
1683		    (struct sockaddr *)&mifp->m6_route.ro_dst, NULL);
1684#ifdef MRT6DEBUG
1685		if (V_mrt6debug & DEBUG_XMIT)
1686			log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1687			    mifp - mif6table, error);
1688#endif
1689	} else {
1690		/*
1691		 * pMTU discovery is intentionally disabled by default, since
1692		 * various router may notify pMTU in multicast, which can be
1693		 * a DDoS to a router
1694		 */
1695		if (V_ip6_mcast_pmtu)
1696			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1697		else {
1698#ifdef MRT6DEBUG
1699			if (V_mrt6debug & DEBUG_XMIT) {
1700				char ip6bufs[INET6_ADDRSTRLEN];
1701				char ip6bufd[INET6_ADDRSTRLEN];
1702				log(LOG_DEBUG,
1703				    "phyint_send: packet too big on %s o %s "
1704				    "g %s size %d(discarded)\n",
1705				    if_name(ifp),
1706				    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1707				    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1708				    mb_copy->m_pkthdr.len);
1709			}
1710#endif /* MRT6DEBUG */
1711			m_freem(mb_copy); /* simply discard the packet */
1712		}
1713	}
1714}
1715
1716static int
1717register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1718{
1719	INIT_VNET_INET6(curvnet);
1720	struct mbuf *mm;
1721	int i, len = m->m_pkthdr.len;
1722	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1723	struct mrt6msg *im6;
1724
1725#ifdef MRT6DEBUG
1726	if (V_mrt6debug) {
1727		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1728		log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1729		    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1730		    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1731	}
1732#endif
1733	++pim6stat.pim6s_snd_registers;
1734
1735	/* Make a copy of the packet to send to the user level process */
1736	MGETHDR(mm, M_DONTWAIT, MT_HEADER);
1737	if (mm == NULL)
1738		return (ENOBUFS);
1739	mm->m_pkthdr.rcvif = NULL;
1740	mm->m_data += max_linkhdr;
1741	mm->m_len = sizeof(struct ip6_hdr);
1742
1743	if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1744		m_freem(mm);
1745		return (ENOBUFS);
1746	}
1747	i = MHLEN - M_LEADINGSPACE(mm);
1748	if (i > len)
1749		i = len;
1750	mm = m_pullup(mm, i);
1751	if (mm == NULL)
1752		return (ENOBUFS);
1753/* TODO: check it! */
1754	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1755
1756	/*
1757	 * Send message to routing daemon
1758	 */
1759	sin6.sin6_addr = ip6->ip6_src;
1760
1761	im6 = mtod(mm, struct mrt6msg *);
1762	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1763	im6->im6_mbz          = 0;
1764
1765	im6->im6_mif = mif - mif6table;
1766
1767	/* iif info is not given for reg. encap.n */
1768	mrt6stat.mrt6s_upcalls++;
1769
1770	if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1771#ifdef MRT6DEBUG
1772		if (V_mrt6debug)
1773			log(LOG_WARNING,
1774			    "register_send: ip6_mrouter socket queue full\n");
1775#endif
1776		++mrt6stat.mrt6s_upq_sockfull;
1777		return (ENOBUFS);
1778	}
1779	return (0);
1780}
1781
1782/*
1783 * pim6_encapcheck() is called by the encap6_input() path at runtime to
1784 * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1785 * into the kernel.
1786 */
1787static int
1788pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
1789{
1790
1791#ifdef DIAGNOSTIC
1792    KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1793#endif
1794    if (proto != IPPROTO_PIM)
1795	return 0;	/* not for us; reject the datagram. */
1796
1797    return 64;		/* claim the datagram. */
1798}
1799
1800/*
1801 * PIM sparse mode hook
1802 * Receives the pim control messages, and passes them up to the listening
1803 * socket, using rip6_input.
1804 * The only message processed is the REGISTER pim message; the pim header
1805 * is stripped off, and the inner packet is passed to register_mforward.
1806 */
1807int
1808pim6_input(struct mbuf **mp, int *offp, int proto)
1809{
1810	INIT_VNET_INET6(curvnet);
1811	struct pim *pim; /* pointer to a pim struct */
1812	struct ip6_hdr *ip6;
1813	int pimlen;
1814	struct mbuf *m = *mp;
1815	int minlen;
1816	int off = *offp;
1817
1818	++pim6stat.pim6s_rcv_total;
1819
1820	ip6 = mtod(m, struct ip6_hdr *);
1821	pimlen = m->m_pkthdr.len - *offp;
1822
1823	/*
1824	 * Validate lengths
1825	 */
1826	if (pimlen < PIM_MINLEN) {
1827		++pim6stat.pim6s_rcv_tooshort;
1828#ifdef MRT6DEBUG
1829		if (V_mrt6debug & DEBUG_PIM)
1830			log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1831#endif
1832		m_freem(m);
1833		return (IPPROTO_DONE);
1834	}
1835
1836	/*
1837	 * if the packet is at least as big as a REGISTER, go ahead
1838	 * and grab the PIM REGISTER header size, to avoid another
1839	 * possible m_pullup() later.
1840	 *
1841	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1842	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1843	 */
1844	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1845
1846	/*
1847	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1848	 * possibly the PIM REGISTER header
1849	 */
1850#ifndef PULLDOWN_TEST
1851	IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1852	/* adjust pointer */
1853	ip6 = mtod(m, struct ip6_hdr *);
1854
1855	/* adjust mbuf to point to the PIM header */
1856	pim = (struct pim *)((caddr_t)ip6 + off);
1857#else
1858	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1859	if (pim == NULL) {
1860		pim6stat.pim6s_rcv_tooshort++;
1861		return (IPPROTO_DONE);
1862	}
1863#endif
1864
1865#define PIM6_CHECKSUM
1866#ifdef PIM6_CHECKSUM
1867	{
1868		int cksumlen;
1869
1870		/*
1871		 * Validate checksum.
1872		 * If PIM REGISTER, exclude the data packet
1873		 */
1874		if (pim->pim_type == PIM_REGISTER)
1875			cksumlen = PIM_MINLEN;
1876		else
1877			cksumlen = pimlen;
1878
1879		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1880			++pim6stat.pim6s_rcv_badsum;
1881#ifdef MRT6DEBUG
1882			if (V_mrt6debug & DEBUG_PIM)
1883				log(LOG_DEBUG,
1884				    "pim6_input: invalid checksum\n");
1885#endif
1886			m_freem(m);
1887			return (IPPROTO_DONE);
1888		}
1889	}
1890#endif /* PIM_CHECKSUM */
1891
1892	/* PIM version check */
1893	if (pim->pim_ver != PIM_VERSION) {
1894		++pim6stat.pim6s_rcv_badversion;
1895#ifdef MRT6DEBUG
1896		log(LOG_ERR,
1897		    "pim6_input: incorrect version %d, expecting %d\n",
1898		    pim->pim_ver, PIM_VERSION);
1899#endif
1900		m_freem(m);
1901		return (IPPROTO_DONE);
1902	}
1903
1904	if (pim->pim_type == PIM_REGISTER) {
1905		/*
1906		 * since this is a REGISTER, we'll make a copy of the register
1907		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1908		 * routing daemon.
1909		 */
1910		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1911
1912		struct mbuf *mcp;
1913		struct ip6_hdr *eip6;
1914		u_int32_t *reghdr;
1915		int rc;
1916#ifdef MRT6DEBUG
1917		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1918#endif
1919
1920		++pim6stat.pim6s_rcv_registers;
1921
1922		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1923#ifdef MRT6DEBUG
1924			if (V_mrt6debug & DEBUG_PIM)
1925				log(LOG_DEBUG,
1926				    "pim6_input: register mif not set: %d\n",
1927				    reg_mif_num);
1928#endif
1929			m_freem(m);
1930			return (IPPROTO_DONE);
1931		}
1932
1933		reghdr = (u_int32_t *)(pim + 1);
1934
1935		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1936			goto pim6_input_to_daemon;
1937
1938		/*
1939		 * Validate length
1940		 */
1941		if (pimlen < PIM6_REG_MINLEN) {
1942			++pim6stat.pim6s_rcv_tooshort;
1943			++pim6stat.pim6s_rcv_badregisters;
1944#ifdef MRT6DEBUG
1945			log(LOG_ERR,
1946			    "pim6_input: register packet size too "
1947			    "small %d from %s\n",
1948			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1949#endif
1950			m_freem(m);
1951			return (IPPROTO_DONE);
1952		}
1953
1954		eip6 = (struct ip6_hdr *) (reghdr + 1);
1955#ifdef MRT6DEBUG
1956		if (V_mrt6debug & DEBUG_PIM)
1957			log(LOG_DEBUG,
1958			    "pim6_input[register], eip6: %s -> %s, "
1959			    "eip6 plen %d\n",
1960			    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1961			    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1962			    ntohs(eip6->ip6_plen));
1963#endif
1964
1965		/* verify the version number of the inner packet */
1966		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1967			++pim6stat.pim6s_rcv_badregisters;
1968#ifdef MRT6DEBUG
1969			log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1970			    "of the inner packet\n",
1971			    (eip6->ip6_vfc & IPV6_VERSION));
1972#endif
1973			m_freem(m);
1974			return (IPPROTO_NONE);
1975		}
1976
1977		/* verify the inner packet is destined to a mcast group */
1978		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1979			++pim6stat.pim6s_rcv_badregisters;
1980#ifdef MRT6DEBUG
1981			if (V_mrt6debug & DEBUG_PIM)
1982				log(LOG_DEBUG,
1983				    "pim6_input: inner packet of register "
1984				    "is not multicast %s\n",
1985				    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1986#endif
1987			m_freem(m);
1988			return (IPPROTO_DONE);
1989		}
1990
1991		/*
1992		 * make a copy of the whole header to pass to the daemon later.
1993		 */
1994		mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1995		if (mcp == NULL) {
1996#ifdef MRT6DEBUG
1997			log(LOG_ERR,
1998			    "pim6_input: pim register: "
1999			    "could not copy register head\n");
2000#endif
2001			m_freem(m);
2002			return (IPPROTO_DONE);
2003		}
2004
2005		/*
2006		 * forward the inner ip6 packet; point m_data at the inner ip6.
2007		 */
2008		m_adj(m, off + PIM_MINLEN);
2009#ifdef MRT6DEBUG
2010		if (V_mrt6debug & DEBUG_PIM) {
2011			log(LOG_DEBUG,
2012			    "pim6_input: forwarding decapsulated register: "
2013			    "src %s, dst %s, mif %d\n",
2014			    ip6_sprintf(ip6bufs, &eip6->ip6_src),
2015			    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
2016			    reg_mif_num);
2017		}
2018#endif
2019
2020		rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
2021				dst.sin6_family, 0);
2022
2023		/* prepare the register head to send to the mrouting daemon */
2024		m = mcp;
2025	}
2026
2027	/*
2028	 * Pass the PIM message up to the daemon; if it is a register message
2029	 * pass the 'head' only up to the daemon. This includes the
2030	 * encapsulator ip6 header, pim header, register header and the
2031	 * encapsulated ip6 header.
2032	 */
2033  pim6_input_to_daemon:
2034	rip6_input(&m, offp, proto);
2035	return (IPPROTO_DONE);
2036}
2037
2038static int
2039ip6_mroute_modevent(module_t mod, int type, void *unused)
2040{
2041	INIT_VNET_INET6(curvnet);
2042
2043	switch (type) {
2044	case MOD_LOAD:
2045		MROUTER6_LOCK_INIT();
2046		MFC6_LOCK_INIT();
2047		MIF6_LOCK_INIT();
2048
2049		pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM,
2050			pim6_encapcheck,
2051			(const struct protosw *)&in6_pim_protosw, NULL);
2052		if (pim6_encap_cookie == NULL) {
2053			printf("ip6_mroute: unable to attach pim6 encap\n");
2054			MIF6_LOCK_DESTROY();
2055			MFC6_LOCK_DESTROY();
2056			MROUTER6_LOCK_DESTROY();
2057			return (EINVAL);
2058		}
2059
2060		ip6_mforward = X_ip6_mforward;
2061		ip6_mrouter_done = X_ip6_mrouter_done;
2062		ip6_mrouter_get = X_ip6_mrouter_get;
2063		ip6_mrouter_set = X_ip6_mrouter_set;
2064		mrt6_ioctl = X_mrt6_ioctl;
2065		break;
2066
2067	case MOD_UNLOAD:
2068		if (V_ip6_mrouter != NULL)
2069			return EINVAL;
2070
2071		if (pim6_encap_cookie) {
2072			encap_detach(pim6_encap_cookie);
2073			pim6_encap_cookie = NULL;
2074		}
2075		X_ip6_mrouter_done();
2076		ip6_mforward = NULL;
2077		ip6_mrouter_done = NULL;
2078		ip6_mrouter_get = NULL;
2079		ip6_mrouter_set = NULL;
2080		mrt6_ioctl = NULL;
2081
2082		MIF6_LOCK_DESTROY();
2083		MFC6_LOCK_DESTROY();
2084		MROUTER6_LOCK_DESTROY();
2085		break;
2086
2087	default:
2088		return (EOPNOTSUPP);
2089	}
2090
2091	return (0);
2092}
2093
2094static moduledata_t ip6_mroutemod = {
2095	"ip6_mroute",
2096	ip6_mroute_modevent,
2097	0
2098};
2099
2100DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PSEUDO, SI_ORDER_ANY);
2101