ip6_mroute.c revision 260712
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: stable/10/sys/netinet6/ip6_mroute.c 260712 2014-01-16 13:21:32Z ae $");
83
84#include "opt_inet6.h"
85#include "opt_kdtrace.h"
86
87#include <sys/param.h>
88#include <sys/callout.h>
89#include <sys/errno.h>
90#include <sys/kernel.h>
91#include <sys/lock.h>
92#include <sys/malloc.h>
93#include <sys/mbuf.h>
94#include <sys/module.h>
95#include <sys/domain.h>
96#include <sys/protosw.h>
97#include <sys/sdt.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
108#include <net/if.h>
109#include <net/if_types.h>
110#include <net/raw_cb.h>
111#include <net/vnet.h>
112
113#include <netinet/in.h>
114#include <netinet/in_var.h>
115#include <netinet/icmp6.h>
116#include <netinet/ip_encap.h>
117
118#include <netinet/ip6.h>
119#include <netinet/in_kdtrace.h>
120#include <netinet6/ip6_var.h>
121#include <netinet6/scope6_var.h>
122#include <netinet6/nd6.h>
123#include <netinet6/ip6_mroute.h>
124#include <netinet6/ip6protosw.h>
125#include <netinet6/pim6.h>
126#include <netinet6/pim6_var.h>
127
128static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
129
130/* XXX: this is a very common idiom; move to <sys/mbuf.h> ? */
131#define M_HASCL(m) ((m)->m_flags & M_EXT)
132
133static int	ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
134static void	phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
135static int	register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
136static int	set_pim6(int *);
137static int	socket_send(struct socket *, struct mbuf *,
138		    struct sockaddr_in6 *);
139
140extern int in6_mcast_loop;
141extern struct domain inet6domain;
142
143static const struct encaptab *pim6_encap_cookie;
144static const struct ip6protosw in6_pim_protosw = {
145	.pr_type =		SOCK_RAW,
146	.pr_domain =		&inet6domain,
147	.pr_protocol =		IPPROTO_PIM,
148	.pr_flags =		PR_ATOMIC|PR_ADDR|PR_LASTHDR,
149	.pr_input =		pim6_input,
150	.pr_output =		rip6_output,
151	.pr_ctloutput =		rip6_ctloutput,
152	.pr_usrreqs =		&rip6_usrreqs
153};
154static int pim6_encapcheck(const struct mbuf *, int, int, void *);
155
156static VNET_DEFINE(int, ip6_mrouter_ver) = 0;
157#define	V_ip6_mrouter_ver	VNET(ip6_mrouter_ver)
158
159SYSCTL_DECL(_net_inet6);
160SYSCTL_DECL(_net_inet6_ip6);
161static SYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim, CTLFLAG_RW, 0, "PIM");
162
163static struct mrt6stat mrt6stat;
164SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
165    &mrt6stat, mrt6stat,
166    "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
167
168#define	MRT6STAT_INC(name)	mrt6stat.name += 1
169#define NO_RTE_FOUND	0x1
170#define RTE_FOUND	0x2
171
172static struct mtx mrouter6_mtx;
173#define	MROUTER6_LOCK()		mtx_lock(&mrouter6_mtx)
174#define	MROUTER6_UNLOCK()	mtx_unlock(&mrouter6_mtx)
175#define	MROUTER6_LOCK_ASSERT()	do {					\
176	mtx_assert(&mrouter6_mtx, MA_OWNED);				\
177	NET_ASSERT_GIANT();						\
178} while (0)
179#define	MROUTER6_LOCK_INIT()	\
180	mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
181#define	MROUTER6_LOCK_DESTROY()	mtx_destroy(&mrouter6_mtx)
182
183static struct mf6c *mf6ctable[MF6CTBLSIZ];
184SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
185    &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
186    "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
187    "netinet6/ip6_mroute.h)");
188
189static struct mtx mfc6_mtx;
190#define	MFC6_LOCK()		mtx_lock(&mfc6_mtx)
191#define	MFC6_UNLOCK()		mtx_unlock(&mfc6_mtx)
192#define	MFC6_LOCK_ASSERT()	do {					\
193	mtx_assert(&mfc6_mtx, MA_OWNED);				\
194	NET_ASSERT_GIANT();						\
195} while (0)
196#define	MFC6_LOCK_INIT()		\
197	mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
198#define	MFC6_LOCK_DESTROY()	mtx_destroy(&mfc6_mtx)
199
200static u_char n6expire[MF6CTBLSIZ];
201
202static struct mif6 mif6table[MAXMIFS];
203SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mif6table, CTLFLAG_RD,
204    &mif6table, sizeof(mif6table), "S,mif6[MAXMIFS]",
205    "IPv6 Multicast Interfaces (struct mif6[MAXMIFS], netinet6/ip6_mroute.h)");
206
207static struct mtx mif6_mtx;
208#define	MIF6_LOCK()		mtx_lock(&mif6_mtx)
209#define	MIF6_UNLOCK()		mtx_unlock(&mif6_mtx)
210#define	MIF6_LOCK_ASSERT()	mtx_assert(&mif6_mtx, MA_OWNED)
211#define	MIF6_LOCK_INIT()	\
212	mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
213#define	MIF6_LOCK_DESTROY()	mtx_destroy(&mif6_mtx)
214
215#ifdef MRT6DEBUG
216static VNET_DEFINE(u_int, mrt6debug) = 0;	/* debug level */
217#define	V_mrt6debug		VNET(mrt6debug)
218#define DEBUG_MFC	0x02
219#define DEBUG_FORWARD	0x04
220#define DEBUG_EXPIRE	0x08
221#define DEBUG_XMIT	0x10
222#define DEBUG_REG	0x20
223#define DEBUG_PIM	0x40
224#define	DEBUG_ERR	0x80
225#define	DEBUG_ANY	0x7f
226#define	MRT6_DLOG(m, fmt, ...)	\
227	if (V_mrt6debug & (m))	\
228		log(((m) & DEBUG_ERR) ? LOG_ERR: LOG_DEBUG, \
229		    "%s: " fmt "\n", __func__, ##__VA_ARGS__)
230#else
231#define	MRT6_DLOG(m, fmt, ...)
232#endif
233
234static void	expire_upcalls(void *);
235#define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
236#define	UPCALL_EXPIRE	6		/* number of timeouts */
237
238/*
239 * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
240 */
241
242/*
243 * 'Interfaces' associated with decapsulator (so we can tell
244 * packets that went through it from ones that get reflected
245 * by a broken gateway).  Different from IPv4 register_if,
246 * these interfaces are linked into the system ifnet list,
247 * because per-interface IPv6 statistics are maintained in
248 * ifp->if_afdata.  But it does not have any routes point
249 * to them.  I.e., packets can't be sent this way.  They
250 * only exist as a placeholder for multicast source
251 * verification.
252 */
253static struct ifnet *multicast_register_if6;
254
255#define ENCAP_HOPS 64
256
257/*
258 * Private variables.
259 */
260static mifi_t nummifs = 0;
261static mifi_t reg_mif_num = (mifi_t)-1;
262
263static struct pim6stat pim6stat;
264SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RW,
265    &pim6stat, pim6stat,
266    "PIM Statistics (struct pim6stat, netinet6/pim6_var.h)");
267
268#define	PIM6STAT_INC(name)	pim6stat.name += 1
269static VNET_DEFINE(int, pim6);
270#define	V_pim6		VNET(pim6)
271
272/*
273 * Hash function for a source, group entry
274 */
275#define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
276				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
277				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
278				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
279
280/*
281 * Find a route for a given origin IPv6 address and Multicast group address.
282 */
283#define MF6CFIND(o, g, rt) do { \
284	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
285	rt = NULL; \
286	MRT6STAT_INC(mrt6s_mfc_lookups); \
287	while (_rt) { \
288		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
289		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
290		    (_rt->mf6c_stall == NULL)) { \
291			rt = _rt; \
292			break; \
293		} \
294		_rt = _rt->mf6c_next; \
295	} \
296	if (rt == NULL) { \
297		MRT6STAT_INC(mrt6s_mfc_misses); \
298	} \
299} while (/*CONSTCOND*/ 0)
300
301/*
302 * Macros to compute elapsed time efficiently
303 * Borrowed from Van Jacobson's scheduling code
304 * XXX: replace with timersub() ?
305 */
306#define TV_DELTA(a, b, delta) do { \
307	    int xxs; \
308		\
309	    delta = (a).tv_usec - (b).tv_usec; \
310	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
311	       switch (xxs) { \
312		      case 2: \
313			  delta += 1000000; \
314			      /* FALLTHROUGH */ \
315		      case 1: \
316			  delta += 1000000; \
317			  break; \
318		      default: \
319			  delta += (1000000 * xxs); \
320	       } \
321	    } \
322} while (/*CONSTCOND*/ 0)
323
324/* XXX: replace with timercmp(a, b, <) ? */
325#define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
326	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
327
328#ifdef UPCALL_TIMING
329#define UPCALL_MAX	50
330static u_long upcall_data[UPCALL_MAX + 1];
331static void collate();
332#endif /* UPCALL_TIMING */
333
334static int ip6_mrouter_init(struct socket *, int, int);
335static int add_m6fc(struct mf6cctl *);
336static int add_m6if(struct mif6ctl *);
337static int del_m6fc(struct mf6cctl *);
338static int del_m6if(mifi_t *);
339static int del_m6if_locked(mifi_t *);
340static int get_mif6_cnt(struct sioc_mif_req6 *);
341static int get_sg_cnt(struct sioc_sg_req6 *);
342
343static struct callout expire_upcalls_ch;
344
345int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
346int X_ip6_mrouter_done(void);
347int X_ip6_mrouter_set(struct socket *, struct sockopt *);
348int X_ip6_mrouter_get(struct socket *, struct sockopt *);
349int X_mrt6_ioctl(u_long, caddr_t);
350
351/*
352 * Handle MRT setsockopt commands to modify the multicast routing tables.
353 */
354int
355X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
356{
357	int error = 0;
358	int optval;
359	struct mif6ctl mifc;
360	struct mf6cctl mfcc;
361	mifi_t mifi;
362
363	if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
364		return (EACCES);
365
366	switch (sopt->sopt_name) {
367	case MRT6_INIT:
368#ifdef MRT6_OINIT
369	case MRT6_OINIT:
370#endif
371		error = sooptcopyin(sopt, &optval, sizeof(optval),
372		    sizeof(optval));
373		if (error)
374			break;
375		error = ip6_mrouter_init(so, optval, sopt->sopt_name);
376		break;
377	case MRT6_DONE:
378		error = X_ip6_mrouter_done();
379		break;
380	case MRT6_ADD_MIF:
381		error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
382		if (error)
383			break;
384		error = add_m6if(&mifc);
385		break;
386	case MRT6_ADD_MFC:
387		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
388		if (error)
389			break;
390		error = add_m6fc(&mfcc);
391		break;
392	case MRT6_DEL_MFC:
393		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
394		if (error)
395			break;
396		error = del_m6fc(&mfcc);
397		break;
398	case MRT6_DEL_MIF:
399		error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
400		if (error)
401			break;
402		error = del_m6if(&mifi);
403		break;
404	case MRT6_PIM:
405		error = sooptcopyin(sopt, &optval, sizeof(optval),
406		    sizeof(optval));
407		if (error)
408			break;
409		error = set_pim6(&optval);
410		break;
411	default:
412		error = EOPNOTSUPP;
413		break;
414	}
415
416	return (error);
417}
418
419/*
420 * Handle MRT getsockopt commands
421 */
422int
423X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
424{
425	int error = 0;
426
427	if (so != V_ip6_mrouter)
428		return (EACCES);
429
430	switch (sopt->sopt_name) {
431		case MRT6_PIM:
432			error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
433			break;
434	}
435	return (error);
436}
437
438/*
439 * Handle ioctl commands to obtain information from the cache
440 */
441int
442X_mrt6_ioctl(u_long cmd, caddr_t data)
443{
444	int ret;
445
446	ret = EINVAL;
447
448	switch (cmd) {
449	case SIOCGETSGCNT_IN6:
450		ret = get_sg_cnt((struct sioc_sg_req6 *)data);
451		break;
452
453	case SIOCGETMIFCNT_IN6:
454		ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
455		break;
456
457	default:
458		break;
459	}
460
461	return (ret);
462}
463
464/*
465 * returns the packet, byte, rpf-failure count for the source group provided
466 */
467static int
468get_sg_cnt(struct sioc_sg_req6 *req)
469{
470	struct mf6c *rt;
471	int ret;
472
473	ret = 0;
474
475	MFC6_LOCK();
476
477	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
478	if (rt == NULL) {
479		ret = ESRCH;
480	} else {
481		req->pktcnt = rt->mf6c_pkt_cnt;
482		req->bytecnt = rt->mf6c_byte_cnt;
483		req->wrong_if = rt->mf6c_wrong_if;
484	}
485
486	MFC6_UNLOCK();
487
488	return (ret);
489}
490
491/*
492 * returns the input and output packet and byte counts on the mif provided
493 */
494static int
495get_mif6_cnt(struct sioc_mif_req6 *req)
496{
497	mifi_t mifi;
498	int ret;
499
500	ret = 0;
501	mifi = req->mifi;
502
503	MIF6_LOCK();
504
505	if (mifi >= nummifs) {
506		ret = EINVAL;
507	} else {
508		req->icount = mif6table[mifi].m6_pkt_in;
509		req->ocount = mif6table[mifi].m6_pkt_out;
510		req->ibytes = mif6table[mifi].m6_bytes_in;
511		req->obytes = mif6table[mifi].m6_bytes_out;
512	}
513
514	MIF6_UNLOCK();
515
516	return (ret);
517}
518
519static int
520set_pim6(int *i)
521{
522	if ((*i != 1) && (*i != 0))
523		return (EINVAL);
524
525	V_pim6 = *i;
526
527	return (0);
528}
529
530/*
531 * Enable multicast routing
532 */
533static int
534ip6_mrouter_init(struct socket *so, int v, int cmd)
535{
536
537	MRT6_DLOG(DEBUG_ANY, "so_type = %d, pr_protocol = %d",
538	    so->so_type, so->so_proto->pr_protocol);
539
540	if (so->so_type != SOCK_RAW ||
541	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
542		return (EOPNOTSUPP);
543
544	if (v != 1)
545		return (ENOPROTOOPT);
546
547	MROUTER6_LOCK();
548
549	if (V_ip6_mrouter != NULL) {
550		MROUTER6_UNLOCK();
551		return (EADDRINUSE);
552	}
553
554	V_ip6_mrouter = so;
555	V_ip6_mrouter_ver = cmd;
556
557	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
558	bzero((caddr_t)n6expire, sizeof(n6expire));
559
560	V_pim6 = 0;/* used for stubbing out/in pim stuff */
561
562	callout_init(&expire_upcalls_ch, 0);
563	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
564	    expire_upcalls, NULL);
565
566	MROUTER6_UNLOCK();
567	MRT6_DLOG(DEBUG_ANY, "finished");
568
569	return (0);
570}
571
572/*
573 * Disable IPv6 multicast forwarding.
574 */
575int
576X_ip6_mrouter_done(void)
577{
578	mifi_t mifi;
579	u_long i;
580	struct mf6c *rt;
581	struct rtdetq *rte;
582
583	MROUTER6_LOCK();
584
585	if (V_ip6_mrouter == NULL) {
586		MROUTER6_UNLOCK();
587		return (EINVAL);
588	}
589
590	/*
591	 * For each phyint in use, disable promiscuous reception of all IPv6
592	 * multicasts.
593	 */
594	for (mifi = 0; mifi < nummifs; mifi++) {
595		if (mif6table[mifi].m6_ifp &&
596		    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
597			if_allmulti(mif6table[mifi].m6_ifp, 0);
598		}
599	}
600	bzero((caddr_t)mif6table, sizeof(mif6table));
601	nummifs = 0;
602
603	V_pim6 = 0; /* used to stub out/in pim specific code */
604
605	callout_stop(&expire_upcalls_ch);
606
607	/*
608	 * Free all multicast forwarding cache entries.
609	 */
610	MFC6_LOCK();
611	for (i = 0; i < MF6CTBLSIZ; i++) {
612		rt = mf6ctable[i];
613		while (rt) {
614			struct mf6c *frt;
615
616			for (rte = rt->mf6c_stall; rte != NULL; ) {
617				struct rtdetq *n = rte->next;
618
619				m_freem(rte->m);
620				free(rte, M_MRTABLE6);
621				rte = n;
622			}
623			frt = rt;
624			rt = rt->mf6c_next;
625			free(frt, M_MRTABLE6);
626		}
627	}
628	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
629	MFC6_UNLOCK();
630
631	/*
632	 * Reset register interface
633	 */
634	if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
635		if_detach(multicast_register_if6);
636		if_free(multicast_register_if6);
637		reg_mif_num = (mifi_t)-1;
638		multicast_register_if6 = NULL;
639	}
640
641	V_ip6_mrouter = NULL;
642	V_ip6_mrouter_ver = 0;
643
644	MROUTER6_UNLOCK();
645	MRT6_DLOG(DEBUG_ANY, "finished");
646
647	return (0);
648}
649
650static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
651
652/*
653 * Add a mif to the mif table
654 */
655static int
656add_m6if(struct mif6ctl *mifcp)
657{
658	struct mif6 *mifp;
659	struct ifnet *ifp;
660	int error;
661
662	MIF6_LOCK();
663
664	if (mifcp->mif6c_mifi >= MAXMIFS) {
665		MIF6_UNLOCK();
666		return (EINVAL);
667	}
668	mifp = mif6table + mifcp->mif6c_mifi;
669	if (mifp->m6_ifp != NULL) {
670		MIF6_UNLOCK();
671		return (EADDRINUSE); /* XXX: is it appropriate? */
672	}
673	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
674		MIF6_UNLOCK();
675		return (ENXIO);
676	}
677
678	ifp = ifnet_byindex(mifcp->mif6c_pifi);
679
680	if (mifcp->mif6c_flags & MIFF_REGISTER) {
681		if (reg_mif_num == (mifi_t)-1) {
682			ifp = if_alloc(IFT_OTHER);
683
684			if_initname(ifp, "register_mif", 0);
685			ifp->if_flags |= IFF_LOOPBACK;
686			if_attach(ifp);
687			multicast_register_if6 = ifp;
688			reg_mif_num = mifcp->mif6c_mifi;
689			/*
690			 * it is impossible to guess the ifindex of the
691			 * register interface.  So mif6c_pifi is automatically
692			 * calculated.
693			 */
694			mifcp->mif6c_pifi = ifp->if_index;
695		} else {
696			ifp = multicast_register_if6;
697		}
698	} else {
699		/* Make sure the interface supports multicast */
700		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
701			MIF6_UNLOCK();
702			return (EOPNOTSUPP);
703		}
704
705		error = if_allmulti(ifp, 1);
706		if (error) {
707			MIF6_UNLOCK();
708			return (error);
709		}
710	}
711
712	mifp->m6_flags     = mifcp->mif6c_flags;
713	mifp->m6_ifp       = ifp;
714
715	/* initialize per mif pkt counters */
716	mifp->m6_pkt_in    = 0;
717	mifp->m6_pkt_out   = 0;
718	mifp->m6_bytes_in  = 0;
719	mifp->m6_bytes_out = 0;
720
721	/* Adjust nummifs up if the mifi is higher than nummifs */
722	if (nummifs <= mifcp->mif6c_mifi)
723		nummifs = mifcp->mif6c_mifi + 1;
724
725	MIF6_UNLOCK();
726	MRT6_DLOG(DEBUG_ANY, "mif #%d, phyint %s", mifcp->mif6c_mifi,
727	    if_name(ifp));
728
729	return (0);
730}
731
732/*
733 * Delete a mif from the mif table
734 */
735static int
736del_m6if_locked(mifi_t *mifip)
737{
738	struct mif6 *mifp = mif6table + *mifip;
739	mifi_t mifi;
740	struct ifnet *ifp;
741
742	MIF6_LOCK_ASSERT();
743
744	if (*mifip >= nummifs)
745		return (EINVAL);
746	if (mifp->m6_ifp == NULL)
747		return (EINVAL);
748
749	if (!(mifp->m6_flags & MIFF_REGISTER)) {
750		/* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
751		ifp = mifp->m6_ifp;
752		if_allmulti(ifp, 0);
753	} else {
754		if (reg_mif_num != (mifi_t)-1 &&
755		    multicast_register_if6 != NULL) {
756			if_detach(multicast_register_if6);
757			if_free(multicast_register_if6);
758			reg_mif_num = (mifi_t)-1;
759			multicast_register_if6 = NULL;
760		}
761	}
762
763	bzero((caddr_t)mifp, sizeof(*mifp));
764
765	/* Adjust nummifs down */
766	for (mifi = nummifs; mifi > 0; mifi--)
767		if (mif6table[mifi - 1].m6_ifp)
768			break;
769	nummifs = mifi;
770	MRT6_DLOG(DEBUG_ANY, "mif %d, nummifs %d", *mifip, nummifs);
771
772	return (0);
773}
774
775static int
776del_m6if(mifi_t *mifip)
777{
778	int cc;
779
780	MIF6_LOCK();
781	cc = del_m6if_locked(mifip);
782	MIF6_UNLOCK();
783
784	return (cc);
785}
786
787/*
788 * Add an mfc entry
789 */
790static int
791add_m6fc(struct mf6cctl *mfccp)
792{
793	struct mf6c *rt;
794	u_long hash;
795	struct rtdetq *rte;
796	u_short nstl;
797	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
798
799	MFC6_LOCK();
800
801	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
802		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
803
804	/* If an entry already exists, just update the fields */
805	if (rt) {
806		MRT6_DLOG(DEBUG_MFC, "no upcall o %s g %s p %x",
807		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
808		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
809		    mfccp->mf6cc_parent);
810
811		rt->mf6c_parent = mfccp->mf6cc_parent;
812		rt->mf6c_ifset = mfccp->mf6cc_ifset;
813
814		MFC6_UNLOCK();
815		return (0);
816	}
817
818	/*
819	 * Find the entry for which the upcall was made and update
820	 */
821	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
822			mfccp->mf6cc_mcastgrp.sin6_addr);
823	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
824		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
825				       &mfccp->mf6cc_origin.sin6_addr) &&
826		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
827				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
828		    (rt->mf6c_stall != NULL)) {
829
830			if (nstl++)
831				log(LOG_ERR,
832				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
833				    "multiple kernel entries",
834				    ip6_sprintf(ip6bufo,
835					    &mfccp->mf6cc_origin.sin6_addr),
836				    ip6_sprintf(ip6bufg,
837					    &mfccp->mf6cc_mcastgrp.sin6_addr),
838				    mfccp->mf6cc_parent, rt->mf6c_stall);
839
840			MRT6_DLOG(DEBUG_MFC, "o %s g %s p %x dbg %p",
841			    ip6_sprintf(ip6bufo,
842			    &mfccp->mf6cc_origin.sin6_addr),
843			    ip6_sprintf(ip6bufg,
844				&mfccp->mf6cc_mcastgrp.sin6_addr),
845			    mfccp->mf6cc_parent, rt->mf6c_stall);
846
847			rt->mf6c_origin     = mfccp->mf6cc_origin;
848			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
849			rt->mf6c_parent     = mfccp->mf6cc_parent;
850			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
851			/* initialize pkt counters per src-grp */
852			rt->mf6c_pkt_cnt    = 0;
853			rt->mf6c_byte_cnt   = 0;
854			rt->mf6c_wrong_if   = 0;
855
856			rt->mf6c_expire = 0;	/* Don't clean this guy up */
857			n6expire[hash]--;
858
859			/* free packets Qed at the end of this entry */
860			for (rte = rt->mf6c_stall; rte != NULL; ) {
861				struct rtdetq *n = rte->next;
862				ip6_mdq(rte->m, rte->ifp, rt);
863				m_freem(rte->m);
864#ifdef UPCALL_TIMING
865				collate(&(rte->t));
866#endif /* UPCALL_TIMING */
867				free(rte, M_MRTABLE6);
868				rte = n;
869			}
870			rt->mf6c_stall = NULL;
871		}
872	}
873
874	/*
875	 * It is possible that an entry is being inserted without an upcall
876	 */
877	if (nstl == 0) {
878		MRT6_DLOG(DEBUG_MFC, "no upcall h %lu o %s g %s p %x", hash,
879		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
880		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
881		    mfccp->mf6cc_parent);
882
883		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
884
885			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
886					       &mfccp->mf6cc_origin.sin6_addr)&&
887			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
888					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
889
890				rt->mf6c_origin     = mfccp->mf6cc_origin;
891				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
892				rt->mf6c_parent     = mfccp->mf6cc_parent;
893				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
894				/* initialize pkt counters per src-grp */
895				rt->mf6c_pkt_cnt    = 0;
896				rt->mf6c_byte_cnt   = 0;
897				rt->mf6c_wrong_if   = 0;
898
899				if (rt->mf6c_expire)
900					n6expire[hash]--;
901				rt->mf6c_expire	   = 0;
902			}
903		}
904		if (rt == NULL) {
905			/* no upcall, so make a new entry */
906			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
907						  M_NOWAIT);
908			if (rt == NULL) {
909				MFC6_UNLOCK();
910				return (ENOBUFS);
911			}
912
913			/* insert new entry at head of hash chain */
914			rt->mf6c_origin     = mfccp->mf6cc_origin;
915			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
916			rt->mf6c_parent     = mfccp->mf6cc_parent;
917			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
918			/* initialize pkt counters per src-grp */
919			rt->mf6c_pkt_cnt    = 0;
920			rt->mf6c_byte_cnt   = 0;
921			rt->mf6c_wrong_if   = 0;
922			rt->mf6c_expire     = 0;
923			rt->mf6c_stall = NULL;
924
925			/* link into table */
926			rt->mf6c_next  = mf6ctable[hash];
927			mf6ctable[hash] = rt;
928		}
929	}
930
931	MFC6_UNLOCK();
932	return (0);
933}
934
935#ifdef UPCALL_TIMING
936/*
937 * collect delay statistics on the upcalls
938 */
939static void
940collate(struct timeval *t)
941{
942	u_long d;
943	struct timeval tp;
944	u_long delta;
945
946	GET_TIME(tp);
947
948	if (TV_LT(*t, tp))
949	{
950		TV_DELTA(tp, *t, delta);
951
952		d = delta >> 10;
953		if (d > UPCALL_MAX)
954			d = UPCALL_MAX;
955
956		++upcall_data[d];
957	}
958}
959#endif /* UPCALL_TIMING */
960
961/*
962 * Delete an mfc entry
963 */
964static int
965del_m6fc(struct mf6cctl *mfccp)
966{
967#ifdef MRT6DEBUG
968	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
969#endif
970	struct sockaddr_in6	origin;
971	struct sockaddr_in6	mcastgrp;
972	struct mf6c		*rt;
973	struct mf6c		**nptr;
974	u_long		hash;
975
976	origin = mfccp->mf6cc_origin;
977	mcastgrp = mfccp->mf6cc_mcastgrp;
978	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
979
980	MRT6_DLOG(DEBUG_MFC, "orig %s mcastgrp %s",
981	    ip6_sprintf(ip6bufo, &origin.sin6_addr),
982	    ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
983
984	MFC6_LOCK();
985
986	nptr = &mf6ctable[hash];
987	while ((rt = *nptr) != NULL) {
988		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
989				       &rt->mf6c_origin.sin6_addr) &&
990		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
991				       &rt->mf6c_mcastgrp.sin6_addr) &&
992		    rt->mf6c_stall == NULL)
993			break;
994
995		nptr = &rt->mf6c_next;
996	}
997	if (rt == NULL) {
998		MFC6_UNLOCK();
999		return (EADDRNOTAVAIL);
1000	}
1001
1002	*nptr = rt->mf6c_next;
1003	free(rt, M_MRTABLE6);
1004
1005	MFC6_UNLOCK();
1006
1007	return (0);
1008}
1009
1010static int
1011socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1012{
1013
1014	if (s) {
1015		if (sbappendaddr(&s->so_rcv,
1016				 (struct sockaddr *)src,
1017				 mm, (struct mbuf *)0) != 0) {
1018			sorwakeup(s);
1019			return (0);
1020		}
1021	}
1022	m_freem(mm);
1023	return (-1);
1024}
1025
1026/*
1027 * IPv6 multicast forwarding function. This function assumes that the packet
1028 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1029 * pointed to by "ifp", and the packet is to be relayed to other networks
1030 * that have members of the packet's destination IPv6 multicast group.
1031 *
1032 * The packet is returned unscathed to the caller, unless it is
1033 * erroneous, in which case a non-zero return value tells the caller to
1034 * discard it.
1035 *
1036 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1037 * this function is called in the originating context (i.e., not when
1038 * forwarding a packet from other node).  ip6_output(), which is currently the
1039 * only function that calls this function is called in the originating context,
1040 * explicitly ensures this condition.  It is caller's responsibility to ensure
1041 * that if this function is called from somewhere else in the originating
1042 * context in the future.
1043 */
1044int
1045X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1046{
1047	struct mf6c *rt;
1048	struct mif6 *mifp;
1049	struct mbuf *mm;
1050	mifi_t mifi;
1051	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1052
1053	MRT6_DLOG(DEBUG_FORWARD, "src %s, dst %s, ifindex %d",
1054	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1055	    ip6_sprintf(ip6bufd, &ip6->ip6_dst), ifp->if_index);
1056
1057	/*
1058	 * Don't forward a packet with Hop limit of zero or one,
1059	 * or a packet destined to a local-only group.
1060	 */
1061	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1062	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1063		return (0);
1064	ip6->ip6_hlim--;
1065
1066	/*
1067	 * Source address check: do not forward packets with unspecified
1068	 * source. It was discussed in July 2000, on ipngwg mailing list.
1069	 * This is rather more serious than unicast cases, because some
1070	 * MLD packets can be sent with the unspecified source address
1071	 * (although such packets must normally set 1 to the hop limit field).
1072	 */
1073	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1074		IP6STAT_INC(ip6s_cantforward);
1075		if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
1076			V_ip6_log_time = time_uptime;
1077			log(LOG_DEBUG,
1078			    "cannot forward "
1079			    "from %s to %s nxt %d received on %s\n",
1080			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1081			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1082			    ip6->ip6_nxt,
1083			    if_name(m->m_pkthdr.rcvif));
1084		}
1085		return (0);
1086	}
1087
1088	MFC6_LOCK();
1089
1090	/*
1091	 * Determine forwarding mifs from the forwarding cache table
1092	 */
1093	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1094
1095	/* Entry exists, so forward if necessary */
1096	if (rt) {
1097		MFC6_UNLOCK();
1098		return (ip6_mdq(m, ifp, rt));
1099	} else {
1100		/*
1101		 * If we don't have a route for packet's origin,
1102		 * Make a copy of the packet &
1103		 * send message to routing daemon
1104		 */
1105
1106		struct mbuf *mb0;
1107		struct rtdetq *rte;
1108		u_long hash;
1109/*		int i, npkts;*/
1110#ifdef UPCALL_TIMING
1111		struct timeval tp;
1112
1113		GET_TIME(tp);
1114#endif /* UPCALL_TIMING */
1115
1116		MRT6STAT_INC(mrt6s_no_route);
1117		MRT6_DLOG(DEBUG_FORWARD | DEBUG_MFC, "no rte s %s g %s",
1118		    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1119		    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1120
1121		/*
1122		 * Allocate mbufs early so that we don't do extra work if we
1123		 * are just going to fail anyway.
1124		 */
1125		rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6,
1126					      M_NOWAIT);
1127		if (rte == NULL) {
1128			MFC6_UNLOCK();
1129			return (ENOBUFS);
1130		}
1131		mb0 = m_copy(m, 0, M_COPYALL);
1132		/*
1133		 * Pullup packet header if needed before storing it,
1134		 * as other references may modify it in the meantime.
1135		 */
1136		if (mb0 &&
1137		    (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1138			mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1139		if (mb0 == NULL) {
1140			free(rte, M_MRTABLE6);
1141			MFC6_UNLOCK();
1142			return (ENOBUFS);
1143		}
1144
1145		/* is there an upcall waiting for this packet? */
1146		hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1147		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1148			if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1149					       &rt->mf6c_origin.sin6_addr) &&
1150			    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1151					       &rt->mf6c_mcastgrp.sin6_addr) &&
1152			    (rt->mf6c_stall != NULL))
1153				break;
1154		}
1155
1156		if (rt == NULL) {
1157			struct mrt6msg *im;
1158#ifdef MRT6_OINIT
1159			struct omrt6msg *oim;
1160#endif
1161
1162			/* no upcall, so make a new entry */
1163			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
1164						  M_NOWAIT);
1165			if (rt == NULL) {
1166				free(rte, M_MRTABLE6);
1167				m_freem(mb0);
1168				MFC6_UNLOCK();
1169				return (ENOBUFS);
1170			}
1171			/*
1172			 * Make a copy of the header to send to the user
1173			 * level process
1174			 */
1175			mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1176
1177			if (mm == NULL) {
1178				free(rte, M_MRTABLE6);
1179				m_freem(mb0);
1180				free(rt, M_MRTABLE6);
1181				MFC6_UNLOCK();
1182				return (ENOBUFS);
1183			}
1184
1185			/*
1186			 * Send message to routing daemon
1187			 */
1188			sin6.sin6_addr = ip6->ip6_src;
1189
1190			im = NULL;
1191#ifdef MRT6_OINIT
1192			oim = NULL;
1193#endif
1194			switch (V_ip6_mrouter_ver) {
1195#ifdef MRT6_OINIT
1196			case MRT6_OINIT:
1197				oim = mtod(mm, struct omrt6msg *);
1198				oim->im6_msgtype = MRT6MSG_NOCACHE;
1199				oim->im6_mbz = 0;
1200				break;
1201#endif
1202			case MRT6_INIT:
1203				im = mtod(mm, struct mrt6msg *);
1204				im->im6_msgtype = MRT6MSG_NOCACHE;
1205				im->im6_mbz = 0;
1206				break;
1207			default:
1208				free(rte, M_MRTABLE6);
1209				m_freem(mb0);
1210				free(rt, M_MRTABLE6);
1211				MFC6_UNLOCK();
1212				return (EINVAL);
1213			}
1214
1215			MRT6_DLOG(DEBUG_FORWARD,
1216			    "getting the iif info in the kernel");
1217
1218			for (mifp = mif6table, mifi = 0;
1219			     mifi < nummifs && mifp->m6_ifp != ifp;
1220			     mifp++, mifi++)
1221				;
1222
1223			switch (V_ip6_mrouter_ver) {
1224#ifdef MRT6_OINIT
1225			case MRT6_OINIT:
1226				oim->im6_mif = mifi;
1227				break;
1228#endif
1229			case MRT6_INIT:
1230				im->im6_mif = mifi;
1231				break;
1232			}
1233
1234			if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1235				log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1236				    "socket queue full\n");
1237				MRT6STAT_INC(mrt6s_upq_sockfull);
1238				free(rte, M_MRTABLE6);
1239				m_freem(mb0);
1240				free(rt, M_MRTABLE6);
1241				MFC6_UNLOCK();
1242				return (ENOBUFS);
1243			}
1244
1245			MRT6STAT_INC(mrt6s_upcalls);
1246
1247			/* insert new entry at head of hash chain */
1248			bzero(rt, sizeof(*rt));
1249			rt->mf6c_origin.sin6_family = AF_INET6;
1250			rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1251			rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1252			rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1253			rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1254			rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1255			rt->mf6c_expire = UPCALL_EXPIRE;
1256			n6expire[hash]++;
1257			rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1258
1259			/* link into table */
1260			rt->mf6c_next  = mf6ctable[hash];
1261			mf6ctable[hash] = rt;
1262			/* Add this entry to the end of the queue */
1263			rt->mf6c_stall = rte;
1264		} else {
1265			/* determine if q has overflowed */
1266			struct rtdetq **p;
1267			int npkts = 0;
1268
1269			for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1270				if (++npkts > MAX_UPQ6) {
1271					MRT6STAT_INC(mrt6s_upq_ovflw);
1272					free(rte, M_MRTABLE6);
1273					m_freem(mb0);
1274					MFC6_UNLOCK();
1275					return (0);
1276				}
1277
1278			/* Add this entry to the end of the queue */
1279			*p = rte;
1280		}
1281
1282		rte->next = NULL;
1283		rte->m = mb0;
1284		rte->ifp = ifp;
1285#ifdef UPCALL_TIMING
1286		rte->t = tp;
1287#endif /* UPCALL_TIMING */
1288
1289		MFC6_UNLOCK();
1290
1291		return (0);
1292	}
1293}
1294
1295/*
1296 * Clean up cache entries if upcalls are not serviced
1297 * Call from the Slow Timeout mechanism, every half second.
1298 */
1299static void
1300expire_upcalls(void *unused)
1301{
1302#ifdef MRT6DEBUG
1303	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1304#endif
1305	struct rtdetq *rte;
1306	struct mf6c *mfc, **nptr;
1307	u_long i;
1308
1309	MFC6_LOCK();
1310	for (i = 0; i < MF6CTBLSIZ; i++) {
1311		if (n6expire[i] == 0)
1312			continue;
1313		nptr = &mf6ctable[i];
1314		while ((mfc = *nptr) != NULL) {
1315			rte = mfc->mf6c_stall;
1316			/*
1317			 * Skip real cache entries
1318			 * Make sure it wasn't marked to not expire (shouldn't happen)
1319			 * If it expires now
1320			 */
1321			if (rte != NULL &&
1322			    mfc->mf6c_expire != 0 &&
1323			    --mfc->mf6c_expire == 0) {
1324				MRT6_DLOG(DEBUG_EXPIRE, "expiring (%s %s)",
1325				    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1326				    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1327				/*
1328				 * drop all the packets
1329				 * free the mbuf with the pkt, if, timing info
1330				 */
1331				do {
1332					struct rtdetq *n = rte->next;
1333					m_freem(rte->m);
1334					free(rte, M_MRTABLE6);
1335					rte = n;
1336				} while (rte != NULL);
1337				MRT6STAT_INC(mrt6s_cache_cleanups);
1338				n6expire[i]--;
1339
1340				*nptr = mfc->mf6c_next;
1341				free(mfc, M_MRTABLE6);
1342			} else {
1343				nptr = &mfc->mf6c_next;
1344			}
1345		}
1346	}
1347	MFC6_UNLOCK();
1348	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1349	    expire_upcalls, NULL);
1350}
1351
1352/*
1353 * Packet forwarding routine once entry in the cache is made
1354 */
1355static int
1356ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1357{
1358	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1359	mifi_t mifi, iif;
1360	struct mif6 *mifp;
1361	int plen = m->m_pkthdr.len;
1362	struct in6_addr src0, dst0; /* copies for local work */
1363	u_int32_t iszone, idzone, oszone, odzone;
1364	int error = 0;
1365
1366/*
1367 * Macro to send packet on mif.  Since RSVP packets don't get counted on
1368 * input, they shouldn't get counted on output, so statistics keeping is
1369 * separate.
1370 */
1371
1372#define MC6_SEND(ip6, mifp, m) do {				\
1373	if ((mifp)->m6_flags & MIFF_REGISTER)			\
1374		register_send((ip6), (mifp), (m));		\
1375	else							\
1376		phyint_send((ip6), (mifp), (m));		\
1377} while (/*CONSTCOND*/ 0)
1378
1379	/*
1380	 * Don't forward if it didn't arrive from the parent mif
1381	 * for its origin.
1382	 */
1383	mifi = rt->mf6c_parent;
1384	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1385		/* came in the wrong interface */
1386		MRT6_DLOG(DEBUG_FORWARD,
1387		    "wrong if: ifid %d mifi %d mififid %x", ifp->if_index,
1388		    mifi, mif6table[mifi].m6_ifp->if_index);
1389		MRT6STAT_INC(mrt6s_wrong_if);
1390		rt->mf6c_wrong_if++;
1391		/*
1392		 * If we are doing PIM processing, and we are forwarding
1393		 * packets on this interface, send a message to the
1394		 * routing daemon.
1395		 */
1396		/* have to make sure this is a valid mif */
1397		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1398			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1399				/*
1400				 * Check the M_LOOP flag to avoid an
1401				 * unnecessary PIM assert.
1402				 * XXX: M_LOOP is an ad-hoc hack...
1403				 */
1404				static struct sockaddr_in6 sin6 =
1405				{ sizeof(sin6), AF_INET6 };
1406
1407				struct mbuf *mm;
1408				struct mrt6msg *im;
1409#ifdef MRT6_OINIT
1410				struct omrt6msg *oim;
1411#endif
1412
1413				mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1414				if (mm &&
1415				    (M_HASCL(mm) ||
1416				     mm->m_len < sizeof(struct ip6_hdr)))
1417					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1418				if (mm == NULL)
1419					return (ENOBUFS);
1420
1421#ifdef MRT6_OINIT
1422				oim = NULL;
1423#endif
1424				im = NULL;
1425				switch (V_ip6_mrouter_ver) {
1426#ifdef MRT6_OINIT
1427				case MRT6_OINIT:
1428					oim = mtod(mm, struct omrt6msg *);
1429					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1430					oim->im6_mbz = 0;
1431					break;
1432#endif
1433				case MRT6_INIT:
1434					im = mtod(mm, struct mrt6msg *);
1435					im->im6_msgtype = MRT6MSG_WRONGMIF;
1436					im->im6_mbz = 0;
1437					break;
1438				default:
1439					m_freem(mm);
1440					return (EINVAL);
1441				}
1442
1443				for (mifp = mif6table, iif = 0;
1444				     iif < nummifs && mifp &&
1445					     mifp->m6_ifp != ifp;
1446				     mifp++, iif++)
1447					;
1448
1449				switch (V_ip6_mrouter_ver) {
1450#ifdef MRT6_OINIT
1451				case MRT6_OINIT:
1452					oim->im6_mif = iif;
1453					sin6.sin6_addr = oim->im6_src;
1454					break;
1455#endif
1456				case MRT6_INIT:
1457					im->im6_mif = iif;
1458					sin6.sin6_addr = im->im6_src;
1459					break;
1460				}
1461
1462				MRT6STAT_INC(mrt6s_upcalls);
1463
1464				if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1465					MRT6_DLOG(DEBUG_ANY,
1466					    "ip6_mrouter socket queue full");
1467					MRT6STAT_INC(mrt6s_upq_sockfull);
1468					return (ENOBUFS);
1469				}	/* if socket Q full */
1470			}		/* if PIM */
1471		return (0);
1472	}			/* if wrong iif */
1473
1474	/* If I sourced this packet, it counts as output, else it was input. */
1475	if (m->m_pkthdr.rcvif == NULL) {
1476		/* XXX: is rcvif really NULL when output?? */
1477		mif6table[mifi].m6_pkt_out++;
1478		mif6table[mifi].m6_bytes_out += plen;
1479	} else {
1480		mif6table[mifi].m6_pkt_in++;
1481		mif6table[mifi].m6_bytes_in += plen;
1482	}
1483	rt->mf6c_pkt_cnt++;
1484	rt->mf6c_byte_cnt += plen;
1485
1486	/*
1487	 * For each mif, forward a copy of the packet if there are group
1488	 * members downstream on the interface.
1489	 */
1490	src0 = ip6->ip6_src;
1491	dst0 = ip6->ip6_dst;
1492	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1493	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1494		IP6STAT_INC(ip6s_badscope);
1495		return (error);
1496	}
1497	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1498		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1499			/*
1500			 * check if the outgoing packet is going to break
1501			 * a scope boundary.
1502			 * XXX For packets through PIM register tunnel
1503			 * interface, we believe a routing daemon.
1504			 */
1505			if (!(mif6table[rt->mf6c_parent].m6_flags &
1506			      MIFF_REGISTER) &&
1507			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1508				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1509				    &oszone) ||
1510				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1511				    &odzone) ||
1512				    iszone != oszone ||
1513				    idzone != odzone) {
1514					IP6STAT_INC(ip6s_badscope);
1515					continue;
1516				}
1517			}
1518
1519			mifp->m6_pkt_out++;
1520			mifp->m6_bytes_out += plen;
1521			MC6_SEND(ip6, mifp, m);
1522		}
1523	}
1524	return (0);
1525}
1526
1527static void
1528phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1529{
1530#ifdef MRT6DEBUG
1531	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1532#endif
1533	struct mbuf *mb_copy;
1534	struct ifnet *ifp = mifp->m6_ifp;
1535	int error = 0;
1536	u_long linkmtu;
1537
1538	/*
1539	 * Make a new reference to the packet; make sure that
1540	 * the IPv6 header is actually copied, not just referenced,
1541	 * so that ip6_output() only scribbles on the copy.
1542	 */
1543	mb_copy = m_copy(m, 0, M_COPYALL);
1544	if (mb_copy &&
1545	    (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1546		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1547	if (mb_copy == NULL) {
1548		return;
1549	}
1550	/* set MCAST flag to the outgoing packet */
1551	mb_copy->m_flags |= M_MCAST;
1552
1553	/*
1554	 * If we sourced the packet, call ip6_output since we may devide
1555	 * the packet into fragments when the packet is too big for the
1556	 * outgoing interface.
1557	 * Otherwise, we can simply send the packet to the interface
1558	 * sending queue.
1559	 */
1560	if (m->m_pkthdr.rcvif == NULL) {
1561		struct ip6_moptions im6o;
1562
1563		im6o.im6o_multicast_ifp = ifp;
1564		/* XXX: ip6_output will override ip6->ip6_hlim */
1565		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1566		im6o.im6o_multicast_loop = 1;
1567		error = ip6_output(mb_copy, NULL, NULL, IPV6_FORWARDING, &im6o,
1568		    NULL, NULL);
1569
1570		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1571		    (uint16_t)(mifp - mif6table), error);
1572		return;
1573	}
1574
1575	/*
1576	 * If configured to loop back multicasts by default,
1577	 * loop back a copy now.
1578	 */
1579	if (in6_mcast_loop) {
1580		struct sockaddr_in6 dst6;
1581
1582		bzero(&dst6, sizeof(dst6));
1583		dst6.sin6_len = sizeof(struct sockaddr_in6);
1584		dst6.sin6_family = AF_INET6;
1585		dst6.sin6_addr = ip6->ip6_dst;
1586		ip6_mloopback(ifp, m, &dst6);
1587	}
1588
1589	/*
1590	 * Put the packet into the sending queue of the outgoing interface
1591	 * if it would fit in the MTU of the interface.
1592	 */
1593	linkmtu = IN6_LINKMTU(ifp);
1594	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1595		struct sockaddr_in6 dst6;
1596
1597		bzero(&dst6, sizeof(dst6));
1598		dst6.sin6_len = sizeof(struct sockaddr_in6);
1599		dst6.sin6_family = AF_INET6;
1600		dst6.sin6_addr = ip6->ip6_dst;
1601
1602		IP_PROBE(send, NULL, NULL, ip6, ifp, NULL, ip6);
1603		/*
1604		 * We just call if_output instead of nd6_output here, since
1605		 * we need no ND for a multicast forwarded packet...right?
1606		 */
1607		m_clrprotoflags(m);	/* Avoid confusing lower layers. */
1608		error = (*ifp->if_output)(ifp, mb_copy,
1609		    (struct sockaddr *)&dst6, NULL);
1610		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1611		    (uint16_t)(mifp - mif6table), error);
1612	} else {
1613		/*
1614		 * pMTU discovery is intentionally disabled by default, since
1615		 * various router may notify pMTU in multicast, which can be
1616		 * a DDoS to a router
1617		 */
1618		if (V_ip6_mcast_pmtu)
1619			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1620		else {
1621			MRT6_DLOG(DEBUG_XMIT, " packet too big on %s o %s "
1622			    "g %s size %d (discarded)", if_name(ifp),
1623			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1624			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1625			    mb_copy->m_pkthdr.len);
1626			m_freem(mb_copy); /* simply discard the packet */
1627		}
1628	}
1629}
1630
1631static int
1632register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1633{
1634#ifdef MRT6DEBUG
1635	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1636#endif
1637	struct mbuf *mm;
1638	int i, len = m->m_pkthdr.len;
1639	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1640	struct mrt6msg *im6;
1641
1642	MRT6_DLOG(DEBUG_ANY, "src %s dst %s",
1643	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1644	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1645	PIM6STAT_INC(pim6s_snd_registers);
1646
1647	/* Make a copy of the packet to send to the user level process. */
1648	mm = m_gethdr(M_NOWAIT, MT_DATA);
1649	if (mm == NULL)
1650		return (ENOBUFS);
1651	mm->m_data += max_linkhdr;
1652	mm->m_len = sizeof(struct ip6_hdr);
1653
1654	if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1655		m_freem(mm);
1656		return (ENOBUFS);
1657	}
1658	i = MHLEN - M_LEADINGSPACE(mm);
1659	if (i > len)
1660		i = len;
1661	mm = m_pullup(mm, i);
1662	if (mm == NULL)
1663		return (ENOBUFS);
1664/* TODO: check it! */
1665	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1666
1667	/*
1668	 * Send message to routing daemon
1669	 */
1670	sin6.sin6_addr = ip6->ip6_src;
1671
1672	im6 = mtod(mm, struct mrt6msg *);
1673	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1674	im6->im6_mbz          = 0;
1675
1676	im6->im6_mif = mif - mif6table;
1677
1678	/* iif info is not given for reg. encap.n */
1679	MRT6STAT_INC(mrt6s_upcalls);
1680
1681	if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1682		MRT6_DLOG(DEBUG_ANY, "ip6_mrouter socket queue full");
1683		MRT6STAT_INC(mrt6s_upq_sockfull);
1684		return (ENOBUFS);
1685	}
1686	return (0);
1687}
1688
1689/*
1690 * pim6_encapcheck() is called by the encap6_input() path at runtime to
1691 * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1692 * into the kernel.
1693 */
1694static int
1695pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
1696{
1697
1698#ifdef DIAGNOSTIC
1699    KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1700#endif
1701    if (proto != IPPROTO_PIM)
1702	return 0;	/* not for us; reject the datagram. */
1703
1704    return 64;		/* claim the datagram. */
1705}
1706
1707/*
1708 * PIM sparse mode hook
1709 * Receives the pim control messages, and passes them up to the listening
1710 * socket, using rip6_input.
1711 * The only message processed is the REGISTER pim message; the pim header
1712 * is stripped off, and the inner packet is passed to register_mforward.
1713 */
1714int
1715pim6_input(struct mbuf **mp, int *offp, int proto)
1716{
1717	struct pim *pim; /* pointer to a pim struct */
1718	struct ip6_hdr *ip6;
1719	int pimlen;
1720	struct mbuf *m = *mp;
1721	int minlen;
1722	int off = *offp;
1723
1724	PIM6STAT_INC(pim6s_rcv_total);
1725
1726	ip6 = mtod(m, struct ip6_hdr *);
1727	pimlen = m->m_pkthdr.len - *offp;
1728
1729	/*
1730	 * Validate lengths
1731	 */
1732	if (pimlen < PIM_MINLEN) {
1733		PIM6STAT_INC(pim6s_rcv_tooshort);
1734		MRT6_DLOG(DEBUG_PIM, "PIM packet too short");
1735		m_freem(m);
1736		return (IPPROTO_DONE);
1737	}
1738
1739	/*
1740	 * if the packet is at least as big as a REGISTER, go ahead
1741	 * and grab the PIM REGISTER header size, to avoid another
1742	 * possible m_pullup() later.
1743	 *
1744	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1745	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1746	 */
1747	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1748
1749	/*
1750	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1751	 * possibly the PIM REGISTER header
1752	 */
1753#ifndef PULLDOWN_TEST
1754	IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1755	/* adjust pointer */
1756	ip6 = mtod(m, struct ip6_hdr *);
1757
1758	/* adjust mbuf to point to the PIM header */
1759	pim = (struct pim *)((caddr_t)ip6 + off);
1760#else
1761	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1762	if (pim == NULL) {
1763		PIM6STAT_INC(pim6s_rcv_tooshort);
1764		return (IPPROTO_DONE);
1765	}
1766#endif
1767
1768#define PIM6_CHECKSUM
1769#ifdef PIM6_CHECKSUM
1770	{
1771		int cksumlen;
1772
1773		/*
1774		 * Validate checksum.
1775		 * If PIM REGISTER, exclude the data packet
1776		 */
1777		if (pim->pim_type == PIM_REGISTER)
1778			cksumlen = PIM_MINLEN;
1779		else
1780			cksumlen = pimlen;
1781
1782		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1783			PIM6STAT_INC(pim6s_rcv_badsum);
1784			MRT6_DLOG(DEBUG_PIM, "invalid checksum");
1785			m_freem(m);
1786			return (IPPROTO_DONE);
1787		}
1788	}
1789#endif /* PIM_CHECKSUM */
1790
1791	/* PIM version check */
1792	if (pim->pim_ver != PIM_VERSION) {
1793		PIM6STAT_INC(pim6s_rcv_badversion);
1794		MRT6_DLOG(DEBUG_ANY | DEBUG_ERR,
1795		    "incorrect version %d, expecting %d",
1796		    pim->pim_ver, PIM_VERSION);
1797		m_freem(m);
1798		return (IPPROTO_DONE);
1799	}
1800
1801	if (pim->pim_type == PIM_REGISTER) {
1802		/*
1803		 * since this is a REGISTER, we'll make a copy of the register
1804		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1805		 * routing daemon.
1806		 */
1807		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1808
1809		struct mbuf *mcp;
1810		struct ip6_hdr *eip6;
1811		u_int32_t *reghdr;
1812		int rc;
1813#ifdef MRT6DEBUG
1814		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1815#endif
1816
1817		PIM6STAT_INC(pim6s_rcv_registers);
1818
1819		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1820			MRT6_DLOG(DEBUG_PIM, "register mif not set: %d",
1821			    reg_mif_num);
1822			m_freem(m);
1823			return (IPPROTO_DONE);
1824		}
1825
1826		reghdr = (u_int32_t *)(pim + 1);
1827
1828		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1829			goto pim6_input_to_daemon;
1830
1831		/*
1832		 * Validate length
1833		 */
1834		if (pimlen < PIM6_REG_MINLEN) {
1835			PIM6STAT_INC(pim6s_rcv_tooshort);
1836			PIM6STAT_INC(pim6s_rcv_badregisters);
1837			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "register packet "
1838			    "size too small %d from %s",
1839			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1840			m_freem(m);
1841			return (IPPROTO_DONE);
1842		}
1843
1844		eip6 = (struct ip6_hdr *) (reghdr + 1);
1845		MRT6_DLOG(DEBUG_PIM, "eip6: %s -> %s, eip6 plen %d",
1846		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1847		    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1848		    ntohs(eip6->ip6_plen));
1849
1850		/* verify the version number of the inner packet */
1851		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1852			PIM6STAT_INC(pim6s_rcv_badregisters);
1853			MRT6_DLOG(DEBUG_ANY, "invalid IP version (%d) "
1854			    "of the inner packet",
1855			    (eip6->ip6_vfc & IPV6_VERSION));
1856			m_freem(m);
1857			return (IPPROTO_NONE);
1858		}
1859
1860		/* verify the inner packet is destined to a mcast group */
1861		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1862			PIM6STAT_INC(pim6s_rcv_badregisters);
1863			MRT6_DLOG(DEBUG_PIM, "inner packet of register "
1864			    "is not multicast %s",
1865			    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1866			m_freem(m);
1867			return (IPPROTO_DONE);
1868		}
1869
1870		/*
1871		 * make a copy of the whole header to pass to the daemon later.
1872		 */
1873		mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1874		if (mcp == NULL) {
1875			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "pim register: "
1876			    "could not copy register head");
1877			m_freem(m);
1878			return (IPPROTO_DONE);
1879		}
1880
1881		/*
1882		 * forward the inner ip6 packet; point m_data at the inner ip6.
1883		 */
1884		m_adj(m, off + PIM_MINLEN);
1885		MRT6_DLOG(DEBUG_PIM, "forwarding decapsulated register: "
1886		    "src %s, dst %s, mif %d",
1887		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1888		    ip6_sprintf(ip6bufd, &eip6->ip6_dst), reg_mif_num);
1889
1890		rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1891				dst.sin6_family, 0);
1892
1893		/* prepare the register head to send to the mrouting daemon */
1894		m = mcp;
1895	}
1896
1897	/*
1898	 * Pass the PIM message up to the daemon; if it is a register message
1899	 * pass the 'head' only up to the daemon. This includes the
1900	 * encapsulator ip6 header, pim header, register header and the
1901	 * encapsulated ip6 header.
1902	 */
1903  pim6_input_to_daemon:
1904	rip6_input(&m, offp, proto);
1905	return (IPPROTO_DONE);
1906}
1907
1908static int
1909ip6_mroute_modevent(module_t mod, int type, void *unused)
1910{
1911
1912	switch (type) {
1913	case MOD_LOAD:
1914		MROUTER6_LOCK_INIT();
1915		MFC6_LOCK_INIT();
1916		MIF6_LOCK_INIT();
1917
1918		pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM,
1919			pim6_encapcheck,
1920			(const struct protosw *)&in6_pim_protosw, NULL);
1921		if (pim6_encap_cookie == NULL) {
1922			printf("ip6_mroute: unable to attach pim6 encap\n");
1923			MIF6_LOCK_DESTROY();
1924			MFC6_LOCK_DESTROY();
1925			MROUTER6_LOCK_DESTROY();
1926			return (EINVAL);
1927		}
1928
1929		ip6_mforward = X_ip6_mforward;
1930		ip6_mrouter_done = X_ip6_mrouter_done;
1931		ip6_mrouter_get = X_ip6_mrouter_get;
1932		ip6_mrouter_set = X_ip6_mrouter_set;
1933		mrt6_ioctl = X_mrt6_ioctl;
1934		break;
1935
1936	case MOD_UNLOAD:
1937		if (V_ip6_mrouter != NULL)
1938			return EINVAL;
1939
1940		if (pim6_encap_cookie) {
1941			encap_detach(pim6_encap_cookie);
1942			pim6_encap_cookie = NULL;
1943		}
1944		X_ip6_mrouter_done();
1945		ip6_mforward = NULL;
1946		ip6_mrouter_done = NULL;
1947		ip6_mrouter_get = NULL;
1948		ip6_mrouter_set = NULL;
1949		mrt6_ioctl = NULL;
1950
1951		MIF6_LOCK_DESTROY();
1952		MFC6_LOCK_DESTROY();
1953		MROUTER6_LOCK_DESTROY();
1954		break;
1955
1956	default:
1957		return (EOPNOTSUPP);
1958	}
1959
1960	return (0);
1961}
1962
1963static moduledata_t ip6_mroutemod = {
1964	"ip6_mroute",
1965	ip6_mroute_modevent,
1966	0
1967};
1968
1969DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PSEUDO, SI_ORDER_ANY);
1970