ip6_mroute.c revision 256108
1/*-
2 * Copyright (C) 1998 WIDE Project.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the project nor the names of its contributors
14 *    may be used to endorse or promote products derived from this software
15 *    without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 *	$KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $
30 */
31
32/*-
33 * Copyright (c) 1989 Stephen Deering
34 * Copyright (c) 1992, 1993
35 *      The Regents of the University of California.  All rights reserved.
36 *
37 * This code is derived from software contributed to Berkeley by
38 * Stephen Deering of Stanford University.
39 *
40 * Redistribution and use in source and binary forms, with or without
41 * modification, are permitted provided that the following conditions
42 * are met:
43 * 1. Redistributions of source code must retain the above copyright
44 *    notice, this list of conditions and the following disclaimer.
45 * 2. Redistributions in binary form must reproduce the above copyright
46 *    notice, this list of conditions and the following disclaimer in the
47 *    documentation and/or other materials provided with the distribution.
48 * 4. Neither the name of the University nor the names of its contributors
49 *    may be used to endorse or promote products derived from this software
50 *    without specific prior written permission.
51 *
52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62 * SUCH DAMAGE.
63 *
64 *	@(#)ip_mroute.c	8.2 (Berkeley) 11/15/93
65 *	BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp
66 */
67
68/*
69 * IP multicast forwarding procedures
70 *
71 * Written by David Waitzman, BBN Labs, August 1988.
72 * Modified by Steve Deering, Stanford, February 1989.
73 * Modified by Mark J. Steiglitz, Stanford, May, 1991
74 * Modified by Van Jacobson, LBL, January 1993
75 * Modified by Ajit Thyagarajan, PARC, August 1993
76 * Modified by Bill Fenner, PARC, April 1994
77 *
78 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
79 */
80
81#include <sys/cdefs.h>
82__FBSDID("$FreeBSD: head/sys/netinet6/ip6_mroute.c 256108 2013-10-07 12:07:40Z glebius $");
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#endif
225
226static void	expire_upcalls(void *);
227#define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
228#define	UPCALL_EXPIRE	6		/* number of timeouts */
229
230/*
231 * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
232 */
233
234/*
235 * 'Interfaces' associated with decapsulator (so we can tell
236 * packets that went through it from ones that get reflected
237 * by a broken gateway).  Different from IPv4 register_if,
238 * these interfaces are linked into the system ifnet list,
239 * because per-interface IPv6 statistics are maintained in
240 * ifp->if_afdata.  But it does not have any routes point
241 * to them.  I.e., packets can't be sent this way.  They
242 * only exist as a placeholder for multicast source
243 * verification.
244 */
245static struct ifnet *multicast_register_if6;
246
247#define ENCAP_HOPS 64
248
249/*
250 * Private variables.
251 */
252static mifi_t nummifs = 0;
253static mifi_t reg_mif_num = (mifi_t)-1;
254
255static struct pim6stat pim6stat;
256SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RW,
257    &pim6stat, pim6stat,
258    "PIM Statistics (struct pim6stat, netinet6/pim6_var.h)");
259
260#define	PIM6STAT_INC(name)	pim6stat.name += 1
261static VNET_DEFINE(int, pim6);
262#define	V_pim6		VNET(pim6)
263
264/*
265 * Hash function for a source, group entry
266 */
267#define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
268				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
269				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
270				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
271
272/*
273 * Find a route for a given origin IPv6 address and Multicast group address.
274 */
275#define MF6CFIND(o, g, rt) do { \
276	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
277	rt = NULL; \
278	MRT6STAT_INC(mrt6s_mfc_lookups); \
279	while (_rt) { \
280		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
281		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
282		    (_rt->mf6c_stall == NULL)) { \
283			rt = _rt; \
284			break; \
285		} \
286		_rt = _rt->mf6c_next; \
287	} \
288	if (rt == NULL) { \
289		MRT6STAT_INC(mrt6s_mfc_misses); \
290	} \
291} while (/*CONSTCOND*/ 0)
292
293/*
294 * Macros to compute elapsed time efficiently
295 * Borrowed from Van Jacobson's scheduling code
296 * XXX: replace with timersub() ?
297 */
298#define TV_DELTA(a, b, delta) do { \
299	    int xxs; \
300		\
301	    delta = (a).tv_usec - (b).tv_usec; \
302	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
303	       switch (xxs) { \
304		      case 2: \
305			  delta += 1000000; \
306			      /* FALLTHROUGH */ \
307		      case 1: \
308			  delta += 1000000; \
309			  break; \
310		      default: \
311			  delta += (1000000 * xxs); \
312	       } \
313	    } \
314} while (/*CONSTCOND*/ 0)
315
316/* XXX: replace with timercmp(a, b, <) ? */
317#define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
318	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
319
320#ifdef UPCALL_TIMING
321#define UPCALL_MAX	50
322static u_long upcall_data[UPCALL_MAX + 1];
323static void collate();
324#endif /* UPCALL_TIMING */
325
326static int ip6_mrouter_init(struct socket *, int, int);
327static int add_m6fc(struct mf6cctl *);
328static int add_m6if(struct mif6ctl *);
329static int del_m6fc(struct mf6cctl *);
330static int del_m6if(mifi_t *);
331static int del_m6if_locked(mifi_t *);
332static int get_mif6_cnt(struct sioc_mif_req6 *);
333static int get_sg_cnt(struct sioc_sg_req6 *);
334
335static struct callout expire_upcalls_ch;
336
337int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
338int X_ip6_mrouter_done(void);
339int X_ip6_mrouter_set(struct socket *, struct sockopt *);
340int X_ip6_mrouter_get(struct socket *, struct sockopt *);
341int X_mrt6_ioctl(u_long, caddr_t);
342
343/*
344 * Handle MRT setsockopt commands to modify the multicast routing tables.
345 */
346int
347X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
348{
349	int error = 0;
350	int optval;
351	struct mif6ctl mifc;
352	struct mf6cctl mfcc;
353	mifi_t mifi;
354
355	if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
356		return (EACCES);
357
358	switch (sopt->sopt_name) {
359	case MRT6_INIT:
360#ifdef MRT6_OINIT
361	case MRT6_OINIT:
362#endif
363		error = sooptcopyin(sopt, &optval, sizeof(optval),
364		    sizeof(optval));
365		if (error)
366			break;
367		error = ip6_mrouter_init(so, optval, sopt->sopt_name);
368		break;
369	case MRT6_DONE:
370		error = X_ip6_mrouter_done();
371		break;
372	case MRT6_ADD_MIF:
373		error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
374		if (error)
375			break;
376		error = add_m6if(&mifc);
377		break;
378	case MRT6_ADD_MFC:
379		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
380		if (error)
381			break;
382		error = add_m6fc(&mfcc);
383		break;
384	case MRT6_DEL_MFC:
385		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
386		if (error)
387			break;
388		error = del_m6fc(&mfcc);
389		break;
390	case MRT6_DEL_MIF:
391		error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
392		if (error)
393			break;
394		error = del_m6if(&mifi);
395		break;
396	case MRT6_PIM:
397		error = sooptcopyin(sopt, &optval, sizeof(optval),
398		    sizeof(optval));
399		if (error)
400			break;
401		error = set_pim6(&optval);
402		break;
403	default:
404		error = EOPNOTSUPP;
405		break;
406	}
407
408	return (error);
409}
410
411/*
412 * Handle MRT getsockopt commands
413 */
414int
415X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
416{
417	int error = 0;
418
419	if (so != V_ip6_mrouter)
420		return (EACCES);
421
422	switch (sopt->sopt_name) {
423		case MRT6_PIM:
424			error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
425			break;
426	}
427	return (error);
428}
429
430/*
431 * Handle ioctl commands to obtain information from the cache
432 */
433int
434X_mrt6_ioctl(u_long cmd, caddr_t data)
435{
436	int ret;
437
438	ret = EINVAL;
439
440	switch (cmd) {
441	case SIOCGETSGCNT_IN6:
442		ret = get_sg_cnt((struct sioc_sg_req6 *)data);
443		break;
444
445	case SIOCGETMIFCNT_IN6:
446		ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
447		break;
448
449	default:
450		break;
451	}
452
453	return (ret);
454}
455
456/*
457 * returns the packet, byte, rpf-failure count for the source group provided
458 */
459static int
460get_sg_cnt(struct sioc_sg_req6 *req)
461{
462	struct mf6c *rt;
463	int ret;
464
465	ret = 0;
466
467	MFC6_LOCK();
468
469	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
470	if (rt == NULL) {
471		ret = ESRCH;
472	} else {
473		req->pktcnt = rt->mf6c_pkt_cnt;
474		req->bytecnt = rt->mf6c_byte_cnt;
475		req->wrong_if = rt->mf6c_wrong_if;
476	}
477
478	MFC6_UNLOCK();
479
480	return (ret);
481}
482
483/*
484 * returns the input and output packet and byte counts on the mif provided
485 */
486static int
487get_mif6_cnt(struct sioc_mif_req6 *req)
488{
489	mifi_t mifi;
490	int ret;
491
492	ret = 0;
493	mifi = req->mifi;
494
495	MIF6_LOCK();
496
497	if (mifi >= nummifs) {
498		ret = EINVAL;
499	} else {
500		req->icount = mif6table[mifi].m6_pkt_in;
501		req->ocount = mif6table[mifi].m6_pkt_out;
502		req->ibytes = mif6table[mifi].m6_bytes_in;
503		req->obytes = mif6table[mifi].m6_bytes_out;
504	}
505
506	MIF6_UNLOCK();
507
508	return (ret);
509}
510
511static int
512set_pim6(int *i)
513{
514	if ((*i != 1) && (*i != 0))
515		return (EINVAL);
516
517	V_pim6 = *i;
518
519	return (0);
520}
521
522/*
523 * Enable multicast routing
524 */
525static int
526ip6_mrouter_init(struct socket *so, int v, int cmd)
527{
528
529#ifdef MRT6DEBUG
530	if (V_mrt6debug)
531		log(LOG_DEBUG,
532		    "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
533		    so->so_type, so->so_proto->pr_protocol);
534#endif
535
536	if (so->so_type != SOCK_RAW ||
537	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
538		return (EOPNOTSUPP);
539
540	if (v != 1)
541		return (ENOPROTOOPT);
542
543	MROUTER6_LOCK();
544
545	if (V_ip6_mrouter != NULL) {
546		MROUTER6_UNLOCK();
547		return (EADDRINUSE);
548	}
549
550	V_ip6_mrouter = so;
551	V_ip6_mrouter_ver = cmd;
552
553	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
554	bzero((caddr_t)n6expire, sizeof(n6expire));
555
556	V_pim6 = 0;/* used for stubbing out/in pim stuff */
557
558	callout_init(&expire_upcalls_ch, 0);
559	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
560	    expire_upcalls, NULL);
561
562	MROUTER6_UNLOCK();
563
564#ifdef MRT6DEBUG
565	if (V_mrt6debug)
566		log(LOG_DEBUG, "ip6_mrouter_init\n");
567#endif
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
646#ifdef MRT6DEBUG
647	if (V_mrt6debug)
648		log(LOG_DEBUG, "ip6_mrouter_done\n");
649#endif
650
651	return (0);
652}
653
654static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
655
656/*
657 * Add a mif to the mif table
658 */
659static int
660add_m6if(struct mif6ctl *mifcp)
661{
662	struct mif6 *mifp;
663	struct ifnet *ifp;
664	int error;
665
666	MIF6_LOCK();
667
668	if (mifcp->mif6c_mifi >= MAXMIFS) {
669		MIF6_UNLOCK();
670		return (EINVAL);
671	}
672	mifp = mif6table + mifcp->mif6c_mifi;
673	if (mifp->m6_ifp != NULL) {
674		MIF6_UNLOCK();
675		return (EADDRINUSE); /* XXX: is it appropriate? */
676	}
677	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
678		MIF6_UNLOCK();
679		return (ENXIO);
680	}
681
682	ifp = ifnet_byindex(mifcp->mif6c_pifi);
683
684	if (mifcp->mif6c_flags & MIFF_REGISTER) {
685		if (reg_mif_num == (mifi_t)-1) {
686			ifp = if_alloc(IFT_OTHER);
687
688			if_initname(ifp, "register_mif", 0);
689			ifp->if_flags |= IFF_LOOPBACK;
690			if_attach(ifp);
691			multicast_register_if6 = ifp;
692			reg_mif_num = mifcp->mif6c_mifi;
693			/*
694			 * it is impossible to guess the ifindex of the
695			 * register interface.  So mif6c_pifi is automatically
696			 * calculated.
697			 */
698			mifcp->mif6c_pifi = ifp->if_index;
699		} else {
700			ifp = multicast_register_if6;
701		}
702	} else {
703		/* Make sure the interface supports multicast */
704		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
705			MIF6_UNLOCK();
706			return (EOPNOTSUPP);
707		}
708
709		error = if_allmulti(ifp, 1);
710		if (error) {
711			MIF6_UNLOCK();
712			return (error);
713		}
714	}
715
716	mifp->m6_flags     = mifcp->mif6c_flags;
717	mifp->m6_ifp       = ifp;
718
719	/* initialize per mif pkt counters */
720	mifp->m6_pkt_in    = 0;
721	mifp->m6_pkt_out   = 0;
722	mifp->m6_bytes_in  = 0;
723	mifp->m6_bytes_out = 0;
724
725	/* Adjust nummifs up if the mifi is higher than nummifs */
726	if (nummifs <= mifcp->mif6c_mifi)
727		nummifs = mifcp->mif6c_mifi + 1;
728
729	MIF6_UNLOCK();
730
731#ifdef MRT6DEBUG
732	if (V_mrt6debug)
733		log(LOG_DEBUG,
734		    "add_mif #%d, phyint %s\n",
735		    mifcp->mif6c_mifi,
736		    ifp->if_xname);
737#endif
738
739	return (0);
740}
741
742/*
743 * Delete a mif from the mif table
744 */
745static int
746del_m6if_locked(mifi_t *mifip)
747{
748	struct mif6 *mifp = mif6table + *mifip;
749	mifi_t mifi;
750	struct ifnet *ifp;
751
752	MIF6_LOCK_ASSERT();
753
754	if (*mifip >= nummifs)
755		return (EINVAL);
756	if (mifp->m6_ifp == NULL)
757		return (EINVAL);
758
759	if (!(mifp->m6_flags & MIFF_REGISTER)) {
760		/* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
761		ifp = mifp->m6_ifp;
762		if_allmulti(ifp, 0);
763	} else {
764		if (reg_mif_num != (mifi_t)-1 &&
765		    multicast_register_if6 != NULL) {
766			if_detach(multicast_register_if6);
767			if_free(multicast_register_if6);
768			reg_mif_num = (mifi_t)-1;
769			multicast_register_if6 = NULL;
770		}
771	}
772
773	bzero((caddr_t)mifp, sizeof(*mifp));
774
775	/* Adjust nummifs down */
776	for (mifi = nummifs; mifi > 0; mifi--)
777		if (mif6table[mifi - 1].m6_ifp)
778			break;
779	nummifs = mifi;
780
781#ifdef MRT6DEBUG
782	if (V_mrt6debug)
783		log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
784#endif
785
786	return (0);
787}
788
789static int
790del_m6if(mifi_t *mifip)
791{
792	int cc;
793
794	MIF6_LOCK();
795	cc = del_m6if_locked(mifip);
796	MIF6_UNLOCK();
797
798	return (cc);
799}
800
801/*
802 * Add an mfc entry
803 */
804static int
805add_m6fc(struct mf6cctl *mfccp)
806{
807	struct mf6c *rt;
808	u_long hash;
809	struct rtdetq *rte;
810	u_short nstl;
811	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
812
813	MFC6_LOCK();
814
815	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
816		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
817
818	/* If an entry already exists, just update the fields */
819	if (rt) {
820#ifdef MRT6DEBUG
821		if (V_mrt6debug & DEBUG_MFC) {
822		    log(LOG_DEBUG,
823			"add_m6fc no upcall h %d o %s g %s p %x\n",
824			ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
825			ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
826			mfccp->mf6cc_parent);
827		}
828#endif
829
830		rt->mf6c_parent = mfccp->mf6cc_parent;
831		rt->mf6c_ifset = mfccp->mf6cc_ifset;
832
833		MFC6_UNLOCK();
834		return (0);
835	}
836
837	/*
838	 * Find the entry for which the upcall was made and update
839	 */
840	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
841			mfccp->mf6cc_mcastgrp.sin6_addr);
842	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
843		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
844				       &mfccp->mf6cc_origin.sin6_addr) &&
845		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
846				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
847		    (rt->mf6c_stall != NULL)) {
848
849			if (nstl++)
850				log(LOG_ERR,
851				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
852				    "multiple kernel entries",
853				    ip6_sprintf(ip6bufo,
854					    &mfccp->mf6cc_origin.sin6_addr),
855				    ip6_sprintf(ip6bufg,
856					    &mfccp->mf6cc_mcastgrp.sin6_addr),
857				    mfccp->mf6cc_parent, rt->mf6c_stall);
858
859#ifdef MRT6DEBUG
860			if (V_mrt6debug & DEBUG_MFC)
861				log(LOG_DEBUG,
862				    "add_m6fc o %s g %s p %x dbg %x\n",
863				    ip6_sprintf(ip6bufo,
864					    &mfccp->mf6cc_origin.sin6_addr),
865				    ip6_sprintf(ip6bufg,
866					    &mfccp->mf6cc_mcastgrp.sin6_addr),
867				    mfccp->mf6cc_parent, rt->mf6c_stall);
868#endif
869
870			rt->mf6c_origin     = mfccp->mf6cc_origin;
871			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
872			rt->mf6c_parent     = mfccp->mf6cc_parent;
873			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
874			/* initialize pkt counters per src-grp */
875			rt->mf6c_pkt_cnt    = 0;
876			rt->mf6c_byte_cnt   = 0;
877			rt->mf6c_wrong_if   = 0;
878
879			rt->mf6c_expire = 0;	/* Don't clean this guy up */
880			n6expire[hash]--;
881
882			/* free packets Qed at the end of this entry */
883			for (rte = rt->mf6c_stall; rte != NULL; ) {
884				struct rtdetq *n = rte->next;
885				ip6_mdq(rte->m, rte->ifp, rt);
886				m_freem(rte->m);
887#ifdef UPCALL_TIMING
888				collate(&(rte->t));
889#endif /* UPCALL_TIMING */
890				free(rte, M_MRTABLE6);
891				rte = n;
892			}
893			rt->mf6c_stall = NULL;
894		}
895	}
896
897	/*
898	 * It is possible that an entry is being inserted without an upcall
899	 */
900	if (nstl == 0) {
901#ifdef MRT6DEBUG
902		if (V_mrt6debug & DEBUG_MFC)
903		    log(LOG_DEBUG,
904			"add_mfc no upcall h %d o %s g %s p %x\n",
905			hash,
906			ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
907			ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
908			mfccp->mf6cc_parent);
909#endif
910
911		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
912
913			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
914					       &mfccp->mf6cc_origin.sin6_addr)&&
915			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
916					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
917
918				rt->mf6c_origin     = mfccp->mf6cc_origin;
919				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
920				rt->mf6c_parent     = mfccp->mf6cc_parent;
921				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
922				/* initialize pkt counters per src-grp */
923				rt->mf6c_pkt_cnt    = 0;
924				rt->mf6c_byte_cnt   = 0;
925				rt->mf6c_wrong_if   = 0;
926
927				if (rt->mf6c_expire)
928					n6expire[hash]--;
929				rt->mf6c_expire	   = 0;
930			}
931		}
932		if (rt == NULL) {
933			/* no upcall, so make a new entry */
934			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
935						  M_NOWAIT);
936			if (rt == NULL) {
937				MFC6_UNLOCK();
938				return (ENOBUFS);
939			}
940
941			/* insert new entry at head of hash chain */
942			rt->mf6c_origin     = mfccp->mf6cc_origin;
943			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
944			rt->mf6c_parent     = mfccp->mf6cc_parent;
945			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
946			/* initialize pkt counters per src-grp */
947			rt->mf6c_pkt_cnt    = 0;
948			rt->mf6c_byte_cnt   = 0;
949			rt->mf6c_wrong_if   = 0;
950			rt->mf6c_expire     = 0;
951			rt->mf6c_stall = NULL;
952
953			/* link into table */
954			rt->mf6c_next  = mf6ctable[hash];
955			mf6ctable[hash] = rt;
956		}
957	}
958
959	MFC6_UNLOCK();
960	return (0);
961}
962
963#ifdef UPCALL_TIMING
964/*
965 * collect delay statistics on the upcalls
966 */
967static void
968collate(struct timeval *t)
969{
970	u_long d;
971	struct timeval tp;
972	u_long delta;
973
974	GET_TIME(tp);
975
976	if (TV_LT(*t, tp))
977	{
978		TV_DELTA(tp, *t, delta);
979
980		d = delta >> 10;
981		if (d > UPCALL_MAX)
982			d = UPCALL_MAX;
983
984		++upcall_data[d];
985	}
986}
987#endif /* UPCALL_TIMING */
988
989/*
990 * Delete an mfc entry
991 */
992static int
993del_m6fc(struct mf6cctl *mfccp)
994{
995	struct sockaddr_in6	origin;
996	struct sockaddr_in6	mcastgrp;
997	struct mf6c		*rt;
998	struct mf6c		**nptr;
999	u_long		hash;
1000
1001	origin = mfccp->mf6cc_origin;
1002	mcastgrp = mfccp->mf6cc_mcastgrp;
1003	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
1004
1005#ifdef MRT6DEBUG
1006	if (V_mrt6debug & DEBUG_MFC) {
1007		char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1008		log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
1009		    ip6_sprintf(ip6bufo, &origin.sin6_addr),
1010		    ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
1011	}
1012#endif
1013
1014	MFC6_LOCK();
1015
1016	nptr = &mf6ctable[hash];
1017	while ((rt = *nptr) != NULL) {
1018		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
1019				       &rt->mf6c_origin.sin6_addr) &&
1020		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1021				       &rt->mf6c_mcastgrp.sin6_addr) &&
1022		    rt->mf6c_stall == NULL)
1023			break;
1024
1025		nptr = &rt->mf6c_next;
1026	}
1027	if (rt == NULL) {
1028		MFC6_UNLOCK();
1029		return (EADDRNOTAVAIL);
1030	}
1031
1032	*nptr = rt->mf6c_next;
1033	free(rt, M_MRTABLE6);
1034
1035	MFC6_UNLOCK();
1036
1037	return (0);
1038}
1039
1040static int
1041socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1042{
1043
1044	if (s) {
1045		if (sbappendaddr(&s->so_rcv,
1046				 (struct sockaddr *)src,
1047				 mm, (struct mbuf *)0) != 0) {
1048			sorwakeup(s);
1049			return (0);
1050		}
1051	}
1052	m_freem(mm);
1053	return (-1);
1054}
1055
1056/*
1057 * IPv6 multicast forwarding function. This function assumes that the packet
1058 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1059 * pointed to by "ifp", and the packet is to be relayed to other networks
1060 * that have members of the packet's destination IPv6 multicast group.
1061 *
1062 * The packet is returned unscathed to the caller, unless it is
1063 * erroneous, in which case a non-zero return value tells the caller to
1064 * discard it.
1065 *
1066 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1067 * this function is called in the originating context (i.e., not when
1068 * forwarding a packet from other node).  ip6_output(), which is currently the
1069 * only function that calls this function is called in the originating context,
1070 * explicitly ensures this condition.  It is caller's responsibility to ensure
1071 * that if this function is called from somewhere else in the originating
1072 * context in the future.
1073 */
1074int
1075X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1076{
1077	struct mf6c *rt;
1078	struct mif6 *mifp;
1079	struct mbuf *mm;
1080	mifi_t mifi;
1081	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1082
1083#ifdef MRT6DEBUG
1084	if (V_mrt6debug & DEBUG_FORWARD)
1085		log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1086		    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1087		    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1088		    ifp->if_index);
1089#endif
1090
1091	/*
1092	 * Don't forward a packet with Hop limit of zero or one,
1093	 * or a packet destined to a local-only group.
1094	 */
1095	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1096	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1097		return (0);
1098	ip6->ip6_hlim--;
1099
1100	/*
1101	 * Source address check: do not forward packets with unspecified
1102	 * source. It was discussed in July 2000, on ipngwg mailing list.
1103	 * This is rather more serious than unicast cases, because some
1104	 * MLD packets can be sent with the unspecified source address
1105	 * (although such packets must normally set 1 to the hop limit field).
1106	 */
1107	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1108		IP6STAT_INC(ip6s_cantforward);
1109		if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
1110			V_ip6_log_time = time_uptime;
1111			log(LOG_DEBUG,
1112			    "cannot forward "
1113			    "from %s to %s nxt %d received on %s\n",
1114			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1115			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1116			    ip6->ip6_nxt,
1117			    if_name(m->m_pkthdr.rcvif));
1118		}
1119		return (0);
1120	}
1121
1122	MFC6_LOCK();
1123
1124	/*
1125	 * Determine forwarding mifs from the forwarding cache table
1126	 */
1127	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1128
1129	/* Entry exists, so forward if necessary */
1130	if (rt) {
1131		MFC6_UNLOCK();
1132		return (ip6_mdq(m, ifp, rt));
1133	} else {
1134		/*
1135		 * If we don't have a route for packet's origin,
1136		 * Make a copy of the packet &
1137		 * send message to routing daemon
1138		 */
1139
1140		struct mbuf *mb0;
1141		struct rtdetq *rte;
1142		u_long hash;
1143/*		int i, npkts;*/
1144#ifdef UPCALL_TIMING
1145		struct timeval tp;
1146
1147		GET_TIME(tp);
1148#endif /* UPCALL_TIMING */
1149
1150		MRT6STAT_INC(mrt6s_no_route);
1151#ifdef MRT6DEBUG
1152		if (V_mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1153			log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1154			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1155			    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1156#endif
1157
1158		/*
1159		 * Allocate mbufs early so that we don't do extra work if we
1160		 * are just going to fail anyway.
1161		 */
1162		rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6,
1163					      M_NOWAIT);
1164		if (rte == NULL) {
1165			MFC6_UNLOCK();
1166			return (ENOBUFS);
1167		}
1168		mb0 = m_copy(m, 0, M_COPYALL);
1169		/*
1170		 * Pullup packet header if needed before storing it,
1171		 * as other references may modify it in the meantime.
1172		 */
1173		if (mb0 &&
1174		    (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1175			mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1176		if (mb0 == NULL) {
1177			free(rte, M_MRTABLE6);
1178			MFC6_UNLOCK();
1179			return (ENOBUFS);
1180		}
1181
1182		/* is there an upcall waiting for this packet? */
1183		hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1184		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1185			if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1186					       &rt->mf6c_origin.sin6_addr) &&
1187			    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1188					       &rt->mf6c_mcastgrp.sin6_addr) &&
1189			    (rt->mf6c_stall != NULL))
1190				break;
1191		}
1192
1193		if (rt == NULL) {
1194			struct mrt6msg *im;
1195#ifdef MRT6_OINIT
1196			struct omrt6msg *oim;
1197#endif
1198
1199			/* no upcall, so make a new entry */
1200			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
1201						  M_NOWAIT);
1202			if (rt == NULL) {
1203				free(rte, M_MRTABLE6);
1204				m_freem(mb0);
1205				MFC6_UNLOCK();
1206				return (ENOBUFS);
1207			}
1208			/*
1209			 * Make a copy of the header to send to the user
1210			 * level process
1211			 */
1212			mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1213
1214			if (mm == NULL) {
1215				free(rte, M_MRTABLE6);
1216				m_freem(mb0);
1217				free(rt, M_MRTABLE6);
1218				MFC6_UNLOCK();
1219				return (ENOBUFS);
1220			}
1221
1222			/*
1223			 * Send message to routing daemon
1224			 */
1225			sin6.sin6_addr = ip6->ip6_src;
1226
1227			im = NULL;
1228#ifdef MRT6_OINIT
1229			oim = NULL;
1230#endif
1231			switch (V_ip6_mrouter_ver) {
1232#ifdef MRT6_OINIT
1233			case MRT6_OINIT:
1234				oim = mtod(mm, struct omrt6msg *);
1235				oim->im6_msgtype = MRT6MSG_NOCACHE;
1236				oim->im6_mbz = 0;
1237				break;
1238#endif
1239			case MRT6_INIT:
1240				im = mtod(mm, struct mrt6msg *);
1241				im->im6_msgtype = MRT6MSG_NOCACHE;
1242				im->im6_mbz = 0;
1243				break;
1244			default:
1245				free(rte, M_MRTABLE6);
1246				m_freem(mb0);
1247				free(rt, M_MRTABLE6);
1248				MFC6_UNLOCK();
1249				return (EINVAL);
1250			}
1251
1252#ifdef MRT6DEBUG
1253			if (V_mrt6debug & DEBUG_FORWARD)
1254				log(LOG_DEBUG,
1255				    "getting the iif info in the kernel\n");
1256#endif
1257
1258			for (mifp = mif6table, mifi = 0;
1259			     mifi < nummifs && mifp->m6_ifp != ifp;
1260			     mifp++, mifi++)
1261				;
1262
1263			switch (V_ip6_mrouter_ver) {
1264#ifdef MRT6_OINIT
1265			case MRT6_OINIT:
1266				oim->im6_mif = mifi;
1267				break;
1268#endif
1269			case MRT6_INIT:
1270				im->im6_mif = mifi;
1271				break;
1272			}
1273
1274			if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1275				log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1276				    "socket queue full\n");
1277				MRT6STAT_INC(mrt6s_upq_sockfull);
1278				free(rte, M_MRTABLE6);
1279				m_freem(mb0);
1280				free(rt, M_MRTABLE6);
1281				MFC6_UNLOCK();
1282				return (ENOBUFS);
1283			}
1284
1285			MRT6STAT_INC(mrt6s_upcalls);
1286
1287			/* insert new entry at head of hash chain */
1288			bzero(rt, sizeof(*rt));
1289			rt->mf6c_origin.sin6_family = AF_INET6;
1290			rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1291			rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1292			rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1293			rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1294			rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1295			rt->mf6c_expire = UPCALL_EXPIRE;
1296			n6expire[hash]++;
1297			rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1298
1299			/* link into table */
1300			rt->mf6c_next  = mf6ctable[hash];
1301			mf6ctable[hash] = rt;
1302			/* Add this entry to the end of the queue */
1303			rt->mf6c_stall = rte;
1304		} else {
1305			/* determine if q has overflowed */
1306			struct rtdetq **p;
1307			int npkts = 0;
1308
1309			for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1310				if (++npkts > MAX_UPQ6) {
1311					MRT6STAT_INC(mrt6s_upq_ovflw);
1312					free(rte, M_MRTABLE6);
1313					m_freem(mb0);
1314					MFC6_UNLOCK();
1315					return (0);
1316				}
1317
1318			/* Add this entry to the end of the queue */
1319			*p = rte;
1320		}
1321
1322		rte->next = NULL;
1323		rte->m = mb0;
1324		rte->ifp = ifp;
1325#ifdef UPCALL_TIMING
1326		rte->t = tp;
1327#endif /* UPCALL_TIMING */
1328
1329		MFC6_UNLOCK();
1330
1331		return (0);
1332	}
1333}
1334
1335/*
1336 * Clean up cache entries if upcalls are not serviced
1337 * Call from the Slow Timeout mechanism, every half second.
1338 */
1339static void
1340expire_upcalls(void *unused)
1341{
1342	struct rtdetq *rte;
1343	struct mf6c *mfc, **nptr;
1344	u_long i;
1345
1346	MFC6_LOCK();
1347	for (i = 0; i < MF6CTBLSIZ; i++) {
1348		if (n6expire[i] == 0)
1349			continue;
1350		nptr = &mf6ctable[i];
1351		while ((mfc = *nptr) != NULL) {
1352			rte = mfc->mf6c_stall;
1353			/*
1354			 * Skip real cache entries
1355			 * Make sure it wasn't marked to not expire (shouldn't happen)
1356			 * If it expires now
1357			 */
1358			if (rte != NULL &&
1359			    mfc->mf6c_expire != 0 &&
1360			    --mfc->mf6c_expire == 0) {
1361#ifdef MRT6DEBUG
1362				if (V_mrt6debug & DEBUG_EXPIRE) {
1363					char ip6bufo[INET6_ADDRSTRLEN];
1364					char ip6bufg[INET6_ADDRSTRLEN];
1365					log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1366					    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1367					    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1368				}
1369#endif
1370				/*
1371				 * drop all the packets
1372				 * free the mbuf with the pkt, if, timing info
1373				 */
1374				do {
1375					struct rtdetq *n = rte->next;
1376					m_freem(rte->m);
1377					free(rte, M_MRTABLE6);
1378					rte = n;
1379				} while (rte != NULL);
1380				MRT6STAT_INC(mrt6s_cache_cleanups);
1381				n6expire[i]--;
1382
1383				*nptr = mfc->mf6c_next;
1384				free(mfc, M_MRTABLE6);
1385			} else {
1386				nptr = &mfc->mf6c_next;
1387			}
1388		}
1389	}
1390	MFC6_UNLOCK();
1391	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1392	    expire_upcalls, NULL);
1393}
1394
1395/*
1396 * Packet forwarding routine once entry in the cache is made
1397 */
1398static int
1399ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1400{
1401	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1402	mifi_t mifi, iif;
1403	struct mif6 *mifp;
1404	int plen = m->m_pkthdr.len;
1405	struct in6_addr src0, dst0; /* copies for local work */
1406	u_int32_t iszone, idzone, oszone, odzone;
1407	int error = 0;
1408
1409/*
1410 * Macro to send packet on mif.  Since RSVP packets don't get counted on
1411 * input, they shouldn't get counted on output, so statistics keeping is
1412 * separate.
1413 */
1414
1415#define MC6_SEND(ip6, mifp, m) do {				\
1416	if ((mifp)->m6_flags & MIFF_REGISTER)			\
1417		register_send((ip6), (mifp), (m));		\
1418	else							\
1419		phyint_send((ip6), (mifp), (m));		\
1420} while (/*CONSTCOND*/ 0)
1421
1422	/*
1423	 * Don't forward if it didn't arrive from the parent mif
1424	 * for its origin.
1425	 */
1426	mifi = rt->mf6c_parent;
1427	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1428		/* came in the wrong interface */
1429#ifdef MRT6DEBUG
1430		if (V_mrt6debug & DEBUG_FORWARD)
1431			log(LOG_DEBUG,
1432			    "wrong if: ifid %d mifi %d mififid %x\n",
1433			    ifp->if_index, mifi,
1434			    mif6table[mifi].m6_ifp->if_index);
1435#endif
1436		MRT6STAT_INC(mrt6s_wrong_if);
1437		rt->mf6c_wrong_if++;
1438		/*
1439		 * If we are doing PIM processing, and we are forwarding
1440		 * packets on this interface, send a message to the
1441		 * routing daemon.
1442		 */
1443		/* have to make sure this is a valid mif */
1444		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1445			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1446				/*
1447				 * Check the M_LOOP flag to avoid an
1448				 * unnecessary PIM assert.
1449				 * XXX: M_LOOP is an ad-hoc hack...
1450				 */
1451				static struct sockaddr_in6 sin6 =
1452				{ sizeof(sin6), AF_INET6 };
1453
1454				struct mbuf *mm;
1455				struct mrt6msg *im;
1456#ifdef MRT6_OINIT
1457				struct omrt6msg *oim;
1458#endif
1459
1460				mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1461				if (mm &&
1462				    (M_HASCL(mm) ||
1463				     mm->m_len < sizeof(struct ip6_hdr)))
1464					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1465				if (mm == NULL)
1466					return (ENOBUFS);
1467
1468#ifdef MRT6_OINIT
1469				oim = NULL;
1470#endif
1471				im = NULL;
1472				switch (V_ip6_mrouter_ver) {
1473#ifdef MRT6_OINIT
1474				case MRT6_OINIT:
1475					oim = mtod(mm, struct omrt6msg *);
1476					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1477					oim->im6_mbz = 0;
1478					break;
1479#endif
1480				case MRT6_INIT:
1481					im = mtod(mm, struct mrt6msg *);
1482					im->im6_msgtype = MRT6MSG_WRONGMIF;
1483					im->im6_mbz = 0;
1484					break;
1485				default:
1486					m_freem(mm);
1487					return (EINVAL);
1488				}
1489
1490				for (mifp = mif6table, iif = 0;
1491				     iif < nummifs && mifp &&
1492					     mifp->m6_ifp != ifp;
1493				     mifp++, iif++)
1494					;
1495
1496				switch (V_ip6_mrouter_ver) {
1497#ifdef MRT6_OINIT
1498				case MRT6_OINIT:
1499					oim->im6_mif = iif;
1500					sin6.sin6_addr = oim->im6_src;
1501					break;
1502#endif
1503				case MRT6_INIT:
1504					im->im6_mif = iif;
1505					sin6.sin6_addr = im->im6_src;
1506					break;
1507				}
1508
1509				MRT6STAT_INC(mrt6s_upcalls);
1510
1511				if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1512#ifdef MRT6DEBUG
1513					if (V_mrt6debug)
1514						log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1515#endif
1516					MRT6STAT_INC(mrt6s_upq_sockfull);
1517					return (ENOBUFS);
1518				}	/* if socket Q full */
1519			}		/* if PIM */
1520		return (0);
1521	}			/* if wrong iif */
1522
1523	/* If I sourced this packet, it counts as output, else it was input. */
1524	if (m->m_pkthdr.rcvif == NULL) {
1525		/* XXX: is rcvif really NULL when output?? */
1526		mif6table[mifi].m6_pkt_out++;
1527		mif6table[mifi].m6_bytes_out += plen;
1528	} else {
1529		mif6table[mifi].m6_pkt_in++;
1530		mif6table[mifi].m6_bytes_in += plen;
1531	}
1532	rt->mf6c_pkt_cnt++;
1533	rt->mf6c_byte_cnt += plen;
1534
1535	/*
1536	 * For each mif, forward a copy of the packet if there are group
1537	 * members downstream on the interface.
1538	 */
1539	src0 = ip6->ip6_src;
1540	dst0 = ip6->ip6_dst;
1541	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1542	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1543		IP6STAT_INC(ip6s_badscope);
1544		return (error);
1545	}
1546	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1547		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1548			/*
1549			 * check if the outgoing packet is going to break
1550			 * a scope boundary.
1551			 * XXX For packets through PIM register tunnel
1552			 * interface, we believe a routing daemon.
1553			 */
1554			if (!(mif6table[rt->mf6c_parent].m6_flags &
1555			      MIFF_REGISTER) &&
1556			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1557				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1558				    &oszone) ||
1559				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1560				    &odzone) ||
1561				    iszone != oszone ||
1562				    idzone != odzone) {
1563					IP6STAT_INC(ip6s_badscope);
1564					continue;
1565				}
1566			}
1567
1568			mifp->m6_pkt_out++;
1569			mifp->m6_bytes_out += plen;
1570			MC6_SEND(ip6, mifp, m);
1571		}
1572	}
1573	return (0);
1574}
1575
1576static void
1577phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1578{
1579	struct mbuf *mb_copy;
1580	struct ifnet *ifp = mifp->m6_ifp;
1581	int error = 0;
1582	u_long linkmtu;
1583
1584	/*
1585	 * Make a new reference to the packet; make sure that
1586	 * the IPv6 header is actually copied, not just referenced,
1587	 * so that ip6_output() only scribbles on the copy.
1588	 */
1589	mb_copy = m_copy(m, 0, M_COPYALL);
1590	if (mb_copy &&
1591	    (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1592		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1593	if (mb_copy == NULL) {
1594		return;
1595	}
1596	/* set MCAST flag to the outgoing packet */
1597	mb_copy->m_flags |= M_MCAST;
1598
1599	/*
1600	 * If we sourced the packet, call ip6_output since we may devide
1601	 * the packet into fragments when the packet is too big for the
1602	 * outgoing interface.
1603	 * Otherwise, we can simply send the packet to the interface
1604	 * sending queue.
1605	 */
1606	if (m->m_pkthdr.rcvif == NULL) {
1607		struct ip6_moptions im6o;
1608
1609		im6o.im6o_multicast_ifp = ifp;
1610		/* XXX: ip6_output will override ip6->ip6_hlim */
1611		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1612		im6o.im6o_multicast_loop = 1;
1613		error = ip6_output(mb_copy, NULL, NULL, IPV6_FORWARDING, &im6o,
1614		    NULL, NULL);
1615
1616#ifdef MRT6DEBUG
1617		if (V_mrt6debug & DEBUG_XMIT)
1618			log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1619			    mifp - mif6table, error);
1620#endif
1621		return;
1622	}
1623
1624	/*
1625	 * If configured to loop back multicasts by default,
1626	 * loop back a copy now.
1627	 */
1628	if (in6_mcast_loop) {
1629		struct sockaddr_in6 dst6;
1630
1631		bzero(&dst6, sizeof(dst6));
1632		dst6.sin6_len = sizeof(struct sockaddr_in6);
1633		dst6.sin6_family = AF_INET6;
1634		dst6.sin6_addr = ip6->ip6_dst;
1635		ip6_mloopback(ifp, m, &dst6);
1636	}
1637
1638	/*
1639	 * Put the packet into the sending queue of the outgoing interface
1640	 * if it would fit in the MTU of the interface.
1641	 */
1642	linkmtu = IN6_LINKMTU(ifp);
1643	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1644		struct sockaddr_in6 dst6;
1645
1646		bzero(&dst6, sizeof(dst6));
1647		dst6.sin6_len = sizeof(struct sockaddr_in6);
1648		dst6.sin6_family = AF_INET6;
1649		dst6.sin6_addr = ip6->ip6_dst;
1650
1651		IP_PROBE(send, NULL, NULL, ip6, ifp, NULL, ip6);
1652		/*
1653		 * We just call if_output instead of nd6_output here, since
1654		 * we need no ND for a multicast forwarded packet...right?
1655		 */
1656		m_clrprotoflags(m);	/* Avoid confusing lower layers. */
1657		error = (*ifp->if_output)(ifp, mb_copy,
1658		    (struct sockaddr *)&dst6, NULL);
1659#ifdef MRT6DEBUG
1660		if (V_mrt6debug & DEBUG_XMIT)
1661			log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1662			    mifp - mif6table, error);
1663#endif
1664	} else {
1665		/*
1666		 * pMTU discovery is intentionally disabled by default, since
1667		 * various router may notify pMTU in multicast, which can be
1668		 * a DDoS to a router
1669		 */
1670		if (V_ip6_mcast_pmtu)
1671			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1672		else {
1673#ifdef MRT6DEBUG
1674			if (V_mrt6debug & DEBUG_XMIT) {
1675				char ip6bufs[INET6_ADDRSTRLEN];
1676				char ip6bufd[INET6_ADDRSTRLEN];
1677				log(LOG_DEBUG,
1678				    "phyint_send: packet too big on %s o %s "
1679				    "g %s size %d(discarded)\n",
1680				    if_name(ifp),
1681				    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1682				    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1683				    mb_copy->m_pkthdr.len);
1684			}
1685#endif /* MRT6DEBUG */
1686			m_freem(mb_copy); /* simply discard the packet */
1687		}
1688	}
1689}
1690
1691static int
1692register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1693{
1694	struct mbuf *mm;
1695	int i, len = m->m_pkthdr.len;
1696	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1697	struct mrt6msg *im6;
1698
1699#ifdef MRT6DEBUG
1700	if (V_mrt6debug) {
1701		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1702		log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1703		    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1704		    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1705	}
1706#endif
1707	PIM6STAT_INC(pim6s_snd_registers);
1708
1709	/* Make a copy of the packet to send to the user level process. */
1710	mm = m_gethdr(M_NOWAIT, MT_DATA);
1711	if (mm == NULL)
1712		return (ENOBUFS);
1713	mm->m_data += max_linkhdr;
1714	mm->m_len = sizeof(struct ip6_hdr);
1715
1716	if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1717		m_freem(mm);
1718		return (ENOBUFS);
1719	}
1720	i = MHLEN - M_LEADINGSPACE(mm);
1721	if (i > len)
1722		i = len;
1723	mm = m_pullup(mm, i);
1724	if (mm == NULL)
1725		return (ENOBUFS);
1726/* TODO: check it! */
1727	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1728
1729	/*
1730	 * Send message to routing daemon
1731	 */
1732	sin6.sin6_addr = ip6->ip6_src;
1733
1734	im6 = mtod(mm, struct mrt6msg *);
1735	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1736	im6->im6_mbz          = 0;
1737
1738	im6->im6_mif = mif - mif6table;
1739
1740	/* iif info is not given for reg. encap.n */
1741	MRT6STAT_INC(mrt6s_upcalls);
1742
1743	if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1744#ifdef MRT6DEBUG
1745		if (V_mrt6debug)
1746			log(LOG_WARNING,
1747			    "register_send: ip6_mrouter socket queue full\n");
1748#endif
1749		MRT6STAT_INC(mrt6s_upq_sockfull);
1750		return (ENOBUFS);
1751	}
1752	return (0);
1753}
1754
1755/*
1756 * pim6_encapcheck() is called by the encap6_input() path at runtime to
1757 * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1758 * into the kernel.
1759 */
1760static int
1761pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
1762{
1763
1764#ifdef DIAGNOSTIC
1765    KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1766#endif
1767    if (proto != IPPROTO_PIM)
1768	return 0;	/* not for us; reject the datagram. */
1769
1770    return 64;		/* claim the datagram. */
1771}
1772
1773/*
1774 * PIM sparse mode hook
1775 * Receives the pim control messages, and passes them up to the listening
1776 * socket, using rip6_input.
1777 * The only message processed is the REGISTER pim message; the pim header
1778 * is stripped off, and the inner packet is passed to register_mforward.
1779 */
1780int
1781pim6_input(struct mbuf **mp, int *offp, int proto)
1782{
1783	struct pim *pim; /* pointer to a pim struct */
1784	struct ip6_hdr *ip6;
1785	int pimlen;
1786	struct mbuf *m = *mp;
1787	int minlen;
1788	int off = *offp;
1789
1790	PIM6STAT_INC(pim6s_rcv_total);
1791
1792	ip6 = mtod(m, struct ip6_hdr *);
1793	pimlen = m->m_pkthdr.len - *offp;
1794
1795	/*
1796	 * Validate lengths
1797	 */
1798	if (pimlen < PIM_MINLEN) {
1799		PIM6STAT_INC(pim6s_rcv_tooshort);
1800#ifdef MRT6DEBUG
1801		if (V_mrt6debug & DEBUG_PIM)
1802			log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1803#endif
1804		m_freem(m);
1805		return (IPPROTO_DONE);
1806	}
1807
1808	/*
1809	 * if the packet is at least as big as a REGISTER, go ahead
1810	 * and grab the PIM REGISTER header size, to avoid another
1811	 * possible m_pullup() later.
1812	 *
1813	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1814	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1815	 */
1816	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1817
1818	/*
1819	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1820	 * possibly the PIM REGISTER header
1821	 */
1822#ifndef PULLDOWN_TEST
1823	IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1824	/* adjust pointer */
1825	ip6 = mtod(m, struct ip6_hdr *);
1826
1827	/* adjust mbuf to point to the PIM header */
1828	pim = (struct pim *)((caddr_t)ip6 + off);
1829#else
1830	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1831	if (pim == NULL) {
1832		PIM6STAT_INC(pim6s_rcv_tooshort);
1833		return (IPPROTO_DONE);
1834	}
1835#endif
1836
1837#define PIM6_CHECKSUM
1838#ifdef PIM6_CHECKSUM
1839	{
1840		int cksumlen;
1841
1842		/*
1843		 * Validate checksum.
1844		 * If PIM REGISTER, exclude the data packet
1845		 */
1846		if (pim->pim_type == PIM_REGISTER)
1847			cksumlen = PIM_MINLEN;
1848		else
1849			cksumlen = pimlen;
1850
1851		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1852			PIM6STAT_INC(pim6s_rcv_badsum);
1853#ifdef MRT6DEBUG
1854			if (V_mrt6debug & DEBUG_PIM)
1855				log(LOG_DEBUG,
1856				    "pim6_input: invalid checksum\n");
1857#endif
1858			m_freem(m);
1859			return (IPPROTO_DONE);
1860		}
1861	}
1862#endif /* PIM_CHECKSUM */
1863
1864	/* PIM version check */
1865	if (pim->pim_ver != PIM_VERSION) {
1866		PIM6STAT_INC(pim6s_rcv_badversion);
1867#ifdef MRT6DEBUG
1868		log(LOG_ERR,
1869		    "pim6_input: incorrect version %d, expecting %d\n",
1870		    pim->pim_ver, PIM_VERSION);
1871#endif
1872		m_freem(m);
1873		return (IPPROTO_DONE);
1874	}
1875
1876	if (pim->pim_type == PIM_REGISTER) {
1877		/*
1878		 * since this is a REGISTER, we'll make a copy of the register
1879		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1880		 * routing daemon.
1881		 */
1882		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1883
1884		struct mbuf *mcp;
1885		struct ip6_hdr *eip6;
1886		u_int32_t *reghdr;
1887		int rc;
1888#ifdef MRT6DEBUG
1889		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1890#endif
1891
1892		PIM6STAT_INC(pim6s_rcv_registers);
1893
1894		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1895#ifdef MRT6DEBUG
1896			if (V_mrt6debug & DEBUG_PIM)
1897				log(LOG_DEBUG,
1898				    "pim6_input: register mif not set: %d\n",
1899				    reg_mif_num);
1900#endif
1901			m_freem(m);
1902			return (IPPROTO_DONE);
1903		}
1904
1905		reghdr = (u_int32_t *)(pim + 1);
1906
1907		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1908			goto pim6_input_to_daemon;
1909
1910		/*
1911		 * Validate length
1912		 */
1913		if (pimlen < PIM6_REG_MINLEN) {
1914			PIM6STAT_INC(pim6s_rcv_tooshort);
1915			PIM6STAT_INC(pim6s_rcv_badregisters);
1916#ifdef MRT6DEBUG
1917			log(LOG_ERR,
1918			    "pim6_input: register packet size too "
1919			    "small %d from %s\n",
1920			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1921#endif
1922			m_freem(m);
1923			return (IPPROTO_DONE);
1924		}
1925
1926		eip6 = (struct ip6_hdr *) (reghdr + 1);
1927#ifdef MRT6DEBUG
1928		if (V_mrt6debug & DEBUG_PIM)
1929			log(LOG_DEBUG,
1930			    "pim6_input[register], eip6: %s -> %s, "
1931			    "eip6 plen %d\n",
1932			    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1933			    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1934			    ntohs(eip6->ip6_plen));
1935#endif
1936
1937		/* verify the version number of the inner packet */
1938		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1939			PIM6STAT_INC(pim6s_rcv_badregisters);
1940#ifdef MRT6DEBUG
1941			log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1942			    "of the inner packet\n",
1943			    (eip6->ip6_vfc & IPV6_VERSION));
1944#endif
1945			m_freem(m);
1946			return (IPPROTO_NONE);
1947		}
1948
1949		/* verify the inner packet is destined to a mcast group */
1950		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1951			PIM6STAT_INC(pim6s_rcv_badregisters);
1952#ifdef MRT6DEBUG
1953			if (V_mrt6debug & DEBUG_PIM)
1954				log(LOG_DEBUG,
1955				    "pim6_input: inner packet of register "
1956				    "is not multicast %s\n",
1957				    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1958#endif
1959			m_freem(m);
1960			return (IPPROTO_DONE);
1961		}
1962
1963		/*
1964		 * make a copy of the whole header to pass to the daemon later.
1965		 */
1966		mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1967		if (mcp == NULL) {
1968#ifdef MRT6DEBUG
1969			log(LOG_ERR,
1970			    "pim6_input: pim register: "
1971			    "could not copy register head\n");
1972#endif
1973			m_freem(m);
1974			return (IPPROTO_DONE);
1975		}
1976
1977		/*
1978		 * forward the inner ip6 packet; point m_data at the inner ip6.
1979		 */
1980		m_adj(m, off + PIM_MINLEN);
1981#ifdef MRT6DEBUG
1982		if (V_mrt6debug & DEBUG_PIM) {
1983			log(LOG_DEBUG,
1984			    "pim6_input: forwarding decapsulated register: "
1985			    "src %s, dst %s, mif %d\n",
1986			    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1987			    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1988			    reg_mif_num);
1989		}
1990#endif
1991
1992		rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1993				dst.sin6_family, 0);
1994
1995		/* prepare the register head to send to the mrouting daemon */
1996		m = mcp;
1997	}
1998
1999	/*
2000	 * Pass the PIM message up to the daemon; if it is a register message
2001	 * pass the 'head' only up to the daemon. This includes the
2002	 * encapsulator ip6 header, pim header, register header and the
2003	 * encapsulated ip6 header.
2004	 */
2005  pim6_input_to_daemon:
2006	rip6_input(&m, offp, proto);
2007	return (IPPROTO_DONE);
2008}
2009
2010static int
2011ip6_mroute_modevent(module_t mod, int type, void *unused)
2012{
2013
2014	switch (type) {
2015	case MOD_LOAD:
2016		MROUTER6_LOCK_INIT();
2017		MFC6_LOCK_INIT();
2018		MIF6_LOCK_INIT();
2019
2020		pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM,
2021			pim6_encapcheck,
2022			(const struct protosw *)&in6_pim_protosw, NULL);
2023		if (pim6_encap_cookie == NULL) {
2024			printf("ip6_mroute: unable to attach pim6 encap\n");
2025			MIF6_LOCK_DESTROY();
2026			MFC6_LOCK_DESTROY();
2027			MROUTER6_LOCK_DESTROY();
2028			return (EINVAL);
2029		}
2030
2031		ip6_mforward = X_ip6_mforward;
2032		ip6_mrouter_done = X_ip6_mrouter_done;
2033		ip6_mrouter_get = X_ip6_mrouter_get;
2034		ip6_mrouter_set = X_ip6_mrouter_set;
2035		mrt6_ioctl = X_mrt6_ioctl;
2036		break;
2037
2038	case MOD_UNLOAD:
2039		if (V_ip6_mrouter != NULL)
2040			return EINVAL;
2041
2042		if (pim6_encap_cookie) {
2043			encap_detach(pim6_encap_cookie);
2044			pim6_encap_cookie = NULL;
2045		}
2046		X_ip6_mrouter_done();
2047		ip6_mforward = NULL;
2048		ip6_mrouter_done = NULL;
2049		ip6_mrouter_get = NULL;
2050		ip6_mrouter_set = NULL;
2051		mrt6_ioctl = NULL;
2052
2053		MIF6_LOCK_DESTROY();
2054		MFC6_LOCK_DESTROY();
2055		MROUTER6_LOCK_DESTROY();
2056		break;
2057
2058	default:
2059		return (EOPNOTSUPP);
2060	}
2061
2062	return (0);
2063}
2064
2065static moduledata_t ip6_mroutemod = {
2066	"ip6_mroute",
2067	ip6_mroute_modevent,
2068	0
2069};
2070
2071DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PSEUDO, SI_ORDER_ANY);
2072