xref: /netbsd-src/sys/netinet6/ip6_mroute.c (revision 6a493d6bc668897c91594964a732d38505b70cbb)
1 /*	$NetBSD: ip6_mroute.c,v 1.105 2013/11/21 21:55:13 riz Exp $	*/
2 /*	$KAME: ip6_mroute.c,v 1.49 2001/07/25 09:21:18 jinmei Exp $	*/
3 
4 /*
5  * Copyright (C) 1998 WIDE Project.
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the project nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  */
32 
33 /*	BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp	*/
34 
35 /*
36  * Copyright (c) 1992, 1993
37  *      The Regents of the University of California.  All rights reserved.
38  *
39  * This code is derived from software contributed to Berkeley by
40  * Stephen Deering of Stanford University.
41  *
42  * Redistribution and use in source and binary forms, with or without
43  * modification, are permitted provided that the following conditions
44  * are met:
45  * 1. Redistributions of source code must retain the above copyright
46  *    notice, this list of conditions and the following disclaimer.
47  * 2. Redistributions in binary form must reproduce the above copyright
48  *    notice, this list of conditions and the following disclaimer in the
49  *    documentation and/or other materials provided with the distribution.
50  * 3. Neither the name of the University nor the names of its contributors
51  *    may be used to endorse or promote products derived from this software
52  *    without specific prior written permission.
53  *
54  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
64  * SUCH DAMAGE.
65  *
66  *      @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93
67  */
68 
69 /*
70  * Copyright (c) 1989 Stephen Deering
71  *
72  * This code is derived from software contributed to Berkeley by
73  * Stephen Deering of Stanford University.
74  *
75  * Redistribution and use in source and binary forms, with or without
76  * modification, are permitted provided that the following conditions
77  * are met:
78  * 1. Redistributions of source code must retain the above copyright
79  *    notice, this list of conditions and the following disclaimer.
80  * 2. Redistributions in binary form must reproduce the above copyright
81  *    notice, this list of conditions and the following disclaimer in the
82  *    documentation and/or other materials provided with the distribution.
83  * 3. All advertising materials mentioning features or use of this software
84  *    must display the following acknowledgement:
85  *      This product includes software developed by the University of
86  *      California, Berkeley and its contributors.
87  * 4. Neither the name of the University nor the names of its contributors
88  *    may be used to endorse or promote products derived from this software
89  *    without specific prior written permission.
90  *
91  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
92  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
93  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
94  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
95  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
96  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
97  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
98  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
99  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
100  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
101  * SUCH DAMAGE.
102  *
103  *      @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93
104  */
105 
106 /*
107  * IP multicast forwarding procedures
108  *
109  * Written by David Waitzman, BBN Labs, August 1988.
110  * Modified by Steve Deering, Stanford, February 1989.
111  * Modified by Mark J. Steiglitz, Stanford, May, 1991
112  * Modified by Van Jacobson, LBL, January 1993
113  * Modified by Ajit Thyagarajan, PARC, August 1993
114  * Modified by Bill Fenner, PARC, April 1994
115  *
116  * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
117  */
118 
119 #include <sys/cdefs.h>
120 __KERNEL_RCSID(0, "$NetBSD: ip6_mroute.c,v 1.105 2013/11/21 21:55:13 riz Exp $");
121 
122 #include "opt_inet.h"
123 #include "opt_mrouting.h"
124 
125 #include <sys/param.h>
126 #include <sys/systm.h>
127 #include <sys/callout.h>
128 #include <sys/mbuf.h>
129 #include <sys/socket.h>
130 #include <sys/socketvar.h>
131 #include <sys/sockio.h>
132 #include <sys/protosw.h>
133 #include <sys/errno.h>
134 #include <sys/time.h>
135 #include <sys/kernel.h>
136 #include <sys/ioctl.h>
137 #include <sys/sysctl.h>
138 #include <sys/syslog.h>
139 
140 #include <net/if.h>
141 #include <net/route.h>
142 #include <net/raw_cb.h>
143 #include <net/net_stats.h>
144 
145 #include <netinet/in.h>
146 #include <netinet/in_var.h>
147 #include <netinet/icmp6.h>
148 
149 #include <netinet/ip6.h>
150 #include <netinet6/ip6_var.h>
151 #include <netinet6/ip6_private.h>
152 #include <netinet6/ip6_mroute.h>
153 #include <netinet6/scope6_var.h>
154 #include <netinet6/pim6.h>
155 #include <netinet6/pim6_var.h>
156 #include <netinet6/nd6.h>
157 
158 #include <net/net_osdep.h>
159 
160 static int ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
161 static void phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
162 
163 static int set_pim6(int *);
164 static int socket_send(struct socket *, struct mbuf *,
165 	    struct sockaddr_in6 *);
166 static int register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
167 
168 /*
169  * Globals.  All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static,
170  * except for netstat or debugging purposes.
171  */
172 struct socket  *ip6_mrouter = NULL;
173 int		ip6_mrouter_ver = 0;
174 int		ip6_mrtproto = IPPROTO_PIM;    /* for netstat only */
175 struct mrt6stat	mrt6stat;
176 
177 #define NO_RTE_FOUND 	0x1
178 #define RTE_FOUND	0x2
179 
180 struct mf6c	*mf6ctable[MF6CTBLSIZ];
181 u_char		n6expire[MF6CTBLSIZ];
182 struct mif6 mif6table[MAXMIFS];
183 #ifdef MRT6DEBUG
184 u_int		mrt6debug = 0;	  /* debug level 	*/
185 #define DEBUG_MFC	0x02
186 #define DEBUG_FORWARD	0x04
187 #define DEBUG_EXPIRE	0x08
188 #define DEBUG_XMIT	0x10
189 #define DEBUG_REG	0x20
190 #define DEBUG_PIM	0x40
191 #define __mrt6debugused     /* empty */
192 #else
193 #define __mrt6debugused     __unused
194 #endif
195 
196 static void	expire_upcalls(void *);
197 #define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
198 #define	UPCALL_EXPIRE	6		/* number of timeouts */
199 
200 #ifdef INET
201 #ifdef MROUTING
202 extern struct socket *ip_mrouter;
203 #endif
204 #endif
205 
206 /*
207  * 'Interfaces' associated with decapsulator (so we can tell
208  * packets that went through it from ones that get reflected
209  * by a broken gateway).  These interfaces are never linked into
210  * the system ifnet list & no routes point to them.  I.e., packets
211  * can't be sent this way.  They only exist as a placeholder for
212  * multicast source verification.
213  */
214 struct ifnet multicast_register_if6;
215 
216 #define ENCAP_HOPS 64
217 
218 /*
219  * Private variables.
220  */
221 static mifi_t nummifs = 0;
222 static mifi_t reg_mif_num = (mifi_t)-1;
223 
224 static percpu_t *pim6stat_percpu;
225 
226 #define	PIM6_STATINC(x)		_NET_STATINC(pim6stat_percpu, x)
227 
228 static int pim6;
229 
230 /*
231  * Hash function for a source, group entry
232  */
233 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
234 				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
235 				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
236 				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
237 
238 /*
239  * Find a route for a given origin IPv6 address and Multicast group address.
240  * Quality of service parameter to be added in the future!!!
241  */
242 
243 #define MF6CFIND(o, g, rt) do { \
244 	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
245 	rt = NULL; \
246 	mrt6stat.mrt6s_mfc_lookups++; \
247 	while (_rt) { \
248 		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
249 		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
250 		    (_rt->mf6c_stall == NULL)) { \
251 			rt = _rt; \
252 			break; \
253 		} \
254 		_rt = _rt->mf6c_next; \
255 	} \
256 	if (rt == NULL) { \
257 		mrt6stat.mrt6s_mfc_misses++; \
258 	} \
259 } while (/*CONSTCOND*/ 0)
260 
261 /*
262  * Macros to compute elapsed time efficiently
263  * Borrowed from Van Jacobson's scheduling code
264  */
265 #define TV_DELTA(a, b, delta) do { \
266 	    int xxs; \
267 		\
268 	    delta = (a).tv_usec - (b).tv_usec; \
269 	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
270 	       switch (xxs) { \
271 		      case 2: \
272 			  delta += 1000000; \
273 			      /* FALLTHROUGH */ \
274 		      case 1: \
275 			  delta += 1000000; \
276 			  break; \
277 		      default: \
278 			  delta += (1000000 * xxs); \
279 	       } \
280 	    } \
281 } while (/*CONSTCOND*/ 0)
282 
283 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
284 	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
285 
286 #ifdef UPCALL_TIMING
287 #define UPCALL_MAX	50
288 u_long upcall_data[UPCALL_MAX + 1];
289 static void collate();
290 #endif /* UPCALL_TIMING */
291 
292 static int get_sg_cnt(struct sioc_sg_req6 *);
293 static int get_mif6_cnt(struct sioc_mif_req6 *);
294 static int ip6_mrouter_init(struct socket *, int, int);
295 static int add_m6if(struct mif6ctl *);
296 static int del_m6if(mifi_t *);
297 static int add_m6fc(struct mf6cctl *);
298 static int del_m6fc(struct mf6cctl *);
299 static void sysctl_net_inet6_pim6_setup(struct sysctllog **);
300 
301 static callout_t expire_upcalls_ch;
302 
303 void
304 pim6_init(void)
305 {
306 
307 	sysctl_net_inet6_pim6_setup(NULL);
308 	pim6stat_percpu = percpu_alloc(sizeof(uint64_t) * PIM6_NSTATS);
309 }
310 
311 /*
312  * Handle MRT setsockopt commands to modify the multicast routing tables.
313  */
314 int
315 ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
316 {
317 	int error, optval;
318 	struct mif6ctl mifc;
319 	struct mf6cctl mfcc;
320 	mifi_t mifi;
321 
322 	if (sopt->sopt_name != MRT6_INIT && so != ip6_mrouter)
323 		return (EACCES);
324 
325 	error = 0;
326 
327 	switch (sopt->sopt_name) {
328 #ifdef MRT6_OINIT
329 	case MRT6_OINIT:
330 #endif
331 	case MRT6_INIT:
332 		error = sockopt_getint(sopt, &optval);
333 		if (error)
334 			break;
335 		return (ip6_mrouter_init(so, optval, sopt->sopt_name));
336 	case MRT6_DONE:
337 		return (ip6_mrouter_done());
338 	case MRT6_ADD_MIF:
339 		error = sockopt_get(sopt, &mifc, sizeof(mifc));
340 		if (error)
341 			break;
342 		return (add_m6if(&mifc));
343 	case MRT6_DEL_MIF:
344 		error = sockopt_get(sopt, &mifi, sizeof(mifi));
345 		if (error)
346 			break;
347 		return (del_m6if(&mifi));
348 	case MRT6_ADD_MFC:
349 		error = sockopt_get(sopt, &mfcc, sizeof(mfcc));
350 		if (error)
351 			break;
352 		return (add_m6fc(&mfcc));
353 	case MRT6_DEL_MFC:
354 		error = sockopt_get(sopt, &mfcc, sizeof(mfcc));
355 		if (error)
356 			break;
357 		return (del_m6fc(&mfcc));
358 	case MRT6_PIM:
359 		error = sockopt_getint(sopt, &optval);
360 		if (error)
361 			break;
362 		return (set_pim6(&optval));
363 	default:
364 		error = EOPNOTSUPP;
365 	}
366 
367 	return (error);
368 }
369 
370 /*
371  * Handle MRT getsockopt commands
372  */
373 int
374 ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
375 {
376 	int error;
377 
378 	if (so != ip6_mrouter) return EACCES;
379 
380 	error = 0;
381 
382 	switch (sopt->sopt_name) {
383 	case MRT6_PIM:
384 		error = sockopt_set(sopt, &pim6, sizeof(pim6));
385 		break;
386 	default:
387 		error = EOPNOTSUPP;
388 		break;
389 	}
390 
391 	return (error);
392 }
393 
394 /*
395  * Handle ioctl commands to obtain information from the cache
396  */
397 int
398 mrt6_ioctl(u_long cmd, void *data)
399 {
400 
401 	switch (cmd) {
402 	case SIOCGETSGCNT_IN6:
403 		return (get_sg_cnt((struct sioc_sg_req6 *)data));
404 	case SIOCGETMIFCNT_IN6:
405 		return (get_mif6_cnt((struct sioc_mif_req6 *)data));
406 	default:
407 		return (EINVAL);
408 	}
409 }
410 
411 /*
412  * returns the packet, byte, rpf-failure count for the source group provided
413  */
414 static int
415 get_sg_cnt(struct sioc_sg_req6 *req)
416 {
417 	struct mf6c *rt;
418 	int s;
419 
420 	s = splsoftnet();
421 	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
422 	splx(s);
423 	if (rt != NULL) {
424 		req->pktcnt = rt->mf6c_pkt_cnt;
425 		req->bytecnt = rt->mf6c_byte_cnt;
426 		req->wrong_if = rt->mf6c_wrong_if;
427 	} else
428 		return (ESRCH);
429 #if 0
430 		req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff;
431 #endif
432 
433 	return 0;
434 }
435 
436 /*
437  * returns the input and output packet and byte counts on the mif provided
438  */
439 static int
440 get_mif6_cnt(struct sioc_mif_req6 *req)
441 {
442 	mifi_t mifi = req->mifi;
443 
444 	if (mifi >= nummifs)
445 		return EINVAL;
446 
447 	req->icount = mif6table[mifi].m6_pkt_in;
448 	req->ocount = mif6table[mifi].m6_pkt_out;
449 	req->ibytes = mif6table[mifi].m6_bytes_in;
450 	req->obytes = mif6table[mifi].m6_bytes_out;
451 
452 	return 0;
453 }
454 
455 static int
456 set_pim6(int *i)
457 {
458 	if ((*i != 1) && (*i != 0))
459 		return EINVAL;
460 
461 	pim6 = *i;
462 
463 	return 0;
464 }
465 
466 /*
467  * Enable multicast routing
468  */
469 static int
470 ip6_mrouter_init(struct socket *so, int v, int cmd)
471 {
472 #ifdef MRT6DEBUG
473 	if (mrt6debug)
474 		log(LOG_DEBUG,
475 		    "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
476 		    so->so_type, so->so_proto->pr_protocol);
477 #endif
478 
479 	if (so->so_type != SOCK_RAW ||
480 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
481 		return (EOPNOTSUPP);
482 
483 	if (v != 1)
484 		return (ENOPROTOOPT);
485 
486 	if (ip6_mrouter != NULL)
487 		return (EADDRINUSE);
488 
489 	ip6_mrouter = so;
490 	ip6_mrouter_ver = cmd;
491 
492 	memset((void *)mf6ctable, 0, sizeof(mf6ctable));
493 	memset((void *)n6expire, 0, sizeof(n6expire));
494 
495 	pim6 = 0;/* used for stubbing out/in pim stuff */
496 
497 	callout_init(&expire_upcalls_ch, CALLOUT_MPSAFE);
498 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
499 	    expire_upcalls, NULL);
500 
501 #ifdef MRT6DEBUG
502 	if (mrt6debug)
503 		log(LOG_DEBUG, "ip6_mrouter_init\n");
504 #endif
505 
506 	return 0;
507 }
508 
509 /*
510  * Disable multicast routing
511  */
512 int
513 ip6_mrouter_done(void)
514 {
515 	mifi_t mifi;
516 	int i;
517 	struct ifnet *ifp;
518 	struct sockaddr_in6 sin6;
519 	struct mf6c *rt;
520 	struct rtdetq *rte;
521 	int s;
522 
523 	s = splsoftnet();
524 
525 	/*
526 	 * For each phyint in use, disable promiscuous reception of all IPv6
527 	 * multicasts.
528 	 */
529 #ifdef INET
530 #ifdef MROUTING
531 	/*
532 	 * If there is still IPv4 multicast routing daemon,
533 	 * we remain interfaces to receive all muliticasted packets.
534 	 * XXX: there may be an interface in which the IPv4 multicast
535 	 * daemon is not interested...
536 	 */
537 	if (!ip_mrouter)
538 #endif
539 #endif
540 	{
541 		for (mifi = 0; mifi < nummifs; mifi++) {
542 			if (mif6table[mifi].m6_ifp &&
543 			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
544 				sin6.sin6_family = AF_INET6;
545 				sin6.sin6_addr = in6addr_any;
546 				ifp = mif6table[mifi].m6_ifp;
547 				if_mcast_op(ifp, SIOCDELMULTI,
548 				    sin6tocsa(&sin6));
549 			}
550 		}
551 	}
552 #ifdef notyet
553 	memset((void *)qtable, 0, sizeof(qtable));
554 	memset((void *)tbftable, 0, sizeof(tbftable));
555 #endif
556 	memset((void *)mif6table, 0, sizeof(mif6table));
557 	nummifs = 0;
558 
559 	pim6 = 0; /* used to stub out/in pim specific code */
560 
561 	callout_stop(&expire_upcalls_ch);
562 
563 	/*
564 	 * Free all multicast forwarding cache entries.
565 	 */
566 	for (i = 0; i < MF6CTBLSIZ; i++) {
567 		rt = mf6ctable[i];
568 		while (rt) {
569 			struct mf6c *frt;
570 
571 			for (rte = rt->mf6c_stall; rte != NULL; ) {
572 				struct rtdetq *n = rte->next;
573 
574 				m_freem(rte->m);
575 				free(rte, M_MRTABLE);
576 				rte = n;
577 			}
578 			frt = rt;
579 			rt = rt->mf6c_next;
580 			free(frt, M_MRTABLE);
581 		}
582 	}
583 
584 	memset((void *)mf6ctable, 0, sizeof(mf6ctable));
585 
586 	/*
587 	 * Reset register interface
588 	 */
589 	if (reg_mif_num != (mifi_t)-1) {
590 		if_detach(&multicast_register_if6);
591 		reg_mif_num = (mifi_t)-1;
592 	}
593 
594 	ip6_mrouter = NULL;
595 	ip6_mrouter_ver = 0;
596 
597 	splx(s);
598 
599 #ifdef MRT6DEBUG
600 	if (mrt6debug)
601 		log(LOG_DEBUG, "ip6_mrouter_done\n");
602 #endif
603 
604 	return 0;
605 }
606 
607 void
608 ip6_mrouter_detach(struct ifnet *ifp)
609 {
610 	struct rtdetq *rte;
611 	struct mf6c *mfc;
612 	mifi_t mifi;
613 	int i;
614 
615 	if (ip6_mrouter == NULL)
616 		return;
617 
618 	/*
619 	 * Delete a mif which points to ifp.
620 	 */
621 	for (mifi = 0; mifi < nummifs; mifi++)
622 		if (mif6table[mifi].m6_ifp == ifp)
623 			del_m6if(&mifi);
624 
625 	/*
626 	 * Clear rte->ifp of cache entries received on ifp.
627 	 */
628 	for (i = 0; i < MF6CTBLSIZ; i++) {
629 		if (n6expire[i] == 0)
630 			continue;
631 
632 		for (mfc = mf6ctable[i]; mfc != NULL; mfc = mfc->mf6c_next) {
633 			for (rte = mfc->mf6c_stall; rte != NULL; rte = rte->next) {
634 				if (rte->ifp == ifp)
635 					rte->ifp = NULL;
636 			}
637 		}
638 	}
639 }
640 
641 
642 /*
643  * Add a mif to the mif table
644  */
645 static int
646 add_m6if(struct mif6ctl *mifcp)
647 {
648 	struct mif6 *mifp;
649 	struct ifnet *ifp;
650 	struct sockaddr_in6 sin6;
651 	int error, s;
652 #ifdef notyet
653 	struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi;
654 #endif
655 
656 	if (mifcp->mif6c_mifi >= MAXMIFS)
657 		return EINVAL;
658 	mifp = mif6table + mifcp->mif6c_mifi;
659 	if (mifp->m6_ifp)
660 		return EADDRINUSE; /* XXX: is it appropriate? */
661 	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi >= if_indexlim)
662 		return ENXIO;
663 	/*
664 	 * XXX: some OSes can remove ifp and clear ifindex2ifnet[id]
665 	 * even for id between 0 and if_index.
666 	 */
667 	if ((ifp = ifindex2ifnet[mifcp->mif6c_pifi]) == NULL)
668 		return ENXIO;
669 
670 	if (mifcp->mif6c_flags & MIFF_REGISTER) {
671 		ifp = &multicast_register_if6;
672 
673 		if (reg_mif_num == (mifi_t)-1) {
674 			strlcpy(ifp->if_xname, "register_mif",
675 			    sizeof(ifp->if_xname));
676 			ifp->if_flags |= IFF_LOOPBACK;
677 			ifp->if_index = mifcp->mif6c_mifi;
678 			reg_mif_num = mifcp->mif6c_mifi;
679 			if_attach(ifp);
680 		}
681 
682 	} /* if REGISTER */
683 	else {
684 		/* Make sure the interface supports multicast */
685 		if ((ifp->if_flags & IFF_MULTICAST) == 0)
686 			return EOPNOTSUPP;
687 
688 		s = splsoftnet();
689 		/*
690 		 * Enable promiscuous reception of all IPv6 multicasts
691 		 * from the interface.
692 		 */
693 		sin6.sin6_family = AF_INET6;
694 		sin6.sin6_addr = in6addr_any;
695 		error = if_mcast_op(ifp, SIOCADDMULTI, sin6tosa(&sin6));
696 		splx(s);
697 		if (error)
698 			return error;
699 	}
700 
701 	s = splsoftnet();
702 	mifp->m6_flags     = mifcp->mif6c_flags;
703 	mifp->m6_ifp       = ifp;
704 #ifdef notyet
705 	/* scaling up here allows division by 1024 in critical code */
706 	mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000;
707 #endif
708 	/* initialize per mif pkt counters */
709 	mifp->m6_pkt_in    = 0;
710 	mifp->m6_pkt_out   = 0;
711 	mifp->m6_bytes_in  = 0;
712 	mifp->m6_bytes_out = 0;
713 	splx(s);
714 
715 	/* Adjust nummifs up if the mifi is higher than nummifs */
716 	if (nummifs <= mifcp->mif6c_mifi)
717 		nummifs = mifcp->mif6c_mifi + 1;
718 
719 #ifdef MRT6DEBUG
720 	if (mrt6debug)
721 		log(LOG_DEBUG,
722 		    "add_mif #%d, phyint %s\n",
723 		    mifcp->mif6c_mifi, ifp->if_xname);
724 #endif
725 
726 	return 0;
727 }
728 
729 /*
730  * Delete a mif from the mif table
731  */
732 static int
733 del_m6if(mifi_t *mifip)
734 {
735 	struct mif6 *mifp = mif6table + *mifip;
736 	mifi_t mifi;
737 	struct ifnet *ifp;
738 	struct sockaddr_in6 sin6;
739 	int s;
740 
741 	if (*mifip >= nummifs)
742 		return EINVAL;
743 	if (mifp->m6_ifp == NULL)
744 		return EINVAL;
745 
746 	s = splsoftnet();
747 
748 	if (!(mifp->m6_flags & MIFF_REGISTER)) {
749 		/*
750 		 * XXX: what if there is yet IPv4 multicast daemon
751 		 *      using the interface?
752 		 */
753 		ifp = mifp->m6_ifp;
754 
755 		sin6.sin6_family = AF_INET6;
756 		sin6.sin6_addr = in6addr_any;
757 		if_mcast_op(ifp, SIOCDELMULTI, sin6tosa(&sin6));
758 	} else {
759 		if (reg_mif_num != (mifi_t)-1) {
760 			if_detach(&multicast_register_if6);
761 			reg_mif_num = (mifi_t)-1;
762 		}
763 	}
764 
765 #ifdef notyet
766 	memset((void *)qtable[*mifip], 0, sizeof(qtable[*mifip]));
767 	memset((void *)mifp->m6_tbf, 0, sizeof(*(mifp->m6_tbf)));
768 #endif
769 	memset((void *)mifp, 0, sizeof (*mifp));
770 
771 	/* Adjust nummifs down */
772 	for (mifi = nummifs; mifi > 0; mifi--)
773 		if (mif6table[mifi - 1].m6_ifp)
774 			break;
775 	nummifs = mifi;
776 
777 	splx(s);
778 
779 #ifdef MRT6DEBUG
780 	if (mrt6debug)
781 		log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
782 #endif
783 
784 	return 0;
785 }
786 
787 /*
788  * Add an mfc entry
789  */
790 static int
791 add_m6fc(struct mf6cctl *mfccp)
792 {
793 	struct mf6c *rt;
794 	u_long hash;
795 	struct rtdetq *rte;
796 	u_short nstl;
797 	int s;
798 
799 	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
800 		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
801 
802 	/* If an entry already exists, just update the fields */
803 	if (rt) {
804 #ifdef MRT6DEBUG
805 		if (mrt6debug & DEBUG_MFC)
806 			log(LOG_DEBUG,"add_m6fc update o %s g %s p %x\n",
807 			    ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
808 			    ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
809 			    mfccp->mf6cc_parent);
810 #endif
811 
812 		s = splsoftnet();
813 		rt->mf6c_parent = mfccp->mf6cc_parent;
814 		rt->mf6c_ifset = mfccp->mf6cc_ifset;
815 		splx(s);
816 		return 0;
817 	}
818 
819 	/*
820 	 * Find the entry for which the upcall was made and update
821 	 */
822 	s = splsoftnet();
823 	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
824 			mfccp->mf6cc_mcastgrp.sin6_addr);
825 	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
826 		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
827 				       &mfccp->mf6cc_origin.sin6_addr) &&
828 		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
829 				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
830 		    (rt->mf6c_stall != NULL)) {
831 
832 			if (nstl++)
833 				log(LOG_ERR,
834 				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
835 				    "multiple kernel entries",
836 				    ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
837 				    ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
838 				    mfccp->mf6cc_parent, rt->mf6c_stall);
839 
840 #ifdef MRT6DEBUG
841 			if (mrt6debug & DEBUG_MFC)
842 				log(LOG_DEBUG,
843 				    "add_m6fc o %s g %s p %x dbg %p\n",
844 				    ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
845 				    ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
846 				    mfccp->mf6cc_parent, rt->mf6c_stall);
847 #endif
848 
849 			rt->mf6c_origin     = mfccp->mf6cc_origin;
850 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
851 			rt->mf6c_parent     = mfccp->mf6cc_parent;
852 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
853 			/* initialize pkt counters per src-grp */
854 			rt->mf6c_pkt_cnt    = 0;
855 			rt->mf6c_byte_cnt   = 0;
856 			rt->mf6c_wrong_if   = 0;
857 
858 			rt->mf6c_expire = 0;	/* Don't clean this guy up */
859 			n6expire[hash]--;
860 
861 			/* free packets Qed at the end of this entry */
862 			for (rte = rt->mf6c_stall; rte != NULL; ) {
863 				struct rtdetq *n = rte->next;
864 				if (rte->ifp) {
865 					ip6_mdq(rte->m, rte->ifp, rt);
866 				}
867 				m_freem(rte->m);
868 #ifdef UPCALL_TIMING
869 				collate(&(rte->t));
870 #endif /* UPCALL_TIMING */
871 				free(rte, M_MRTABLE);
872 				rte = n;
873 			}
874 			rt->mf6c_stall = NULL;
875 		}
876 	}
877 
878 	/*
879 	 * It is possible that an entry is being inserted without an upcall
880 	 */
881 	if (nstl == 0) {
882 #ifdef MRT6DEBUG
883 		if (mrt6debug & DEBUG_MFC)
884 			log(LOG_DEBUG,
885 			    "add_mfc no upcall h %ld o %s g %s p %x\n",
886 			    hash,
887 			    ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
888 			    ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
889 			    mfccp->mf6cc_parent);
890 #endif
891 
892 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
893 
894 			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
895 					       &mfccp->mf6cc_origin.sin6_addr)&&
896 			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
897 					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
898 
899 				rt->mf6c_origin     = mfccp->mf6cc_origin;
900 				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
901 				rt->mf6c_parent     = mfccp->mf6cc_parent;
902 				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
903 				/* initialize pkt counters per src-grp */
904 				rt->mf6c_pkt_cnt    = 0;
905 				rt->mf6c_byte_cnt   = 0;
906 				rt->mf6c_wrong_if   = 0;
907 
908 				if (rt->mf6c_expire)
909 					n6expire[hash]--;
910 				rt->mf6c_expire	   = 0;
911 			}
912 		}
913 		if (rt == NULL) {
914 			/* no upcall, so make a new entry */
915 			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE,
916 						  M_NOWAIT);
917 			if (rt == NULL) {
918 				splx(s);
919 				return ENOBUFS;
920 			}
921 
922 			/* insert new entry at head of hash chain */
923 			rt->mf6c_origin     = mfccp->mf6cc_origin;
924 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
925 			rt->mf6c_parent     = mfccp->mf6cc_parent;
926 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
927 			/* initialize pkt counters per src-grp */
928 			rt->mf6c_pkt_cnt    = 0;
929 			rt->mf6c_byte_cnt   = 0;
930 			rt->mf6c_wrong_if   = 0;
931 			rt->mf6c_expire     = 0;
932 			rt->mf6c_stall = NULL;
933 
934 			/* link into table */
935 			rt->mf6c_next  = mf6ctable[hash];
936 			mf6ctable[hash] = rt;
937 		}
938 	}
939 	splx(s);
940 	return 0;
941 }
942 
943 #ifdef UPCALL_TIMING
944 /*
945  * collect delay statistics on the upcalls
946  */
947 static void
948 collate(struct timeval *t)
949 {
950 	u_long d;
951 	struct timeval tp;
952 	u_long delta;
953 
954 	GET_TIME(tp);
955 
956 	if (TV_LT(*t, tp))
957 	{
958 		TV_DELTA(tp, *t, delta);
959 
960 		d = delta >> 10;
961 		if (d > UPCALL_MAX)
962 			d = UPCALL_MAX;
963 
964 		++upcall_data[d];
965 	}
966 }
967 #endif /* UPCALL_TIMING */
968 
969 /*
970  * Delete an mfc entry
971  */
972 static int
973 del_m6fc(struct mf6cctl *mfccp)
974 {
975 	struct sockaddr_in6 	origin;
976 	struct sockaddr_in6 	mcastgrp;
977 	struct mf6c 		*rt;
978 	struct mf6c	 	**nptr;
979 	u_long 		hash;
980 	int s;
981 
982 	origin = mfccp->mf6cc_origin;
983 	mcastgrp = mfccp->mf6cc_mcastgrp;
984 	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
985 
986 #ifdef MRT6DEBUG
987 	if (mrt6debug & DEBUG_MFC)
988 		log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
989 		    ip6_sprintf(&origin.sin6_addr),
990 		    ip6_sprintf(&mcastgrp.sin6_addr));
991 #endif
992 
993 	s = splsoftnet();
994 
995 	nptr = &mf6ctable[hash];
996 	while ((rt = *nptr) != NULL) {
997 		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
998 				       &rt->mf6c_origin.sin6_addr) &&
999 		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1000 				       &rt->mf6c_mcastgrp.sin6_addr) &&
1001 		    rt->mf6c_stall == NULL)
1002 			break;
1003 
1004 		nptr = &rt->mf6c_next;
1005 	}
1006 	if (rt == NULL) {
1007 		splx(s);
1008 		return EADDRNOTAVAIL;
1009 	}
1010 
1011 	*nptr = rt->mf6c_next;
1012 	free(rt, M_MRTABLE);
1013 
1014 	splx(s);
1015 
1016 	return 0;
1017 }
1018 
1019 static int
1020 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1021 {
1022 	if (s) {
1023 		if (sbappendaddr(&s->so_rcv,
1024 		    (struct sockaddr *)src, mm, NULL) != 0) {
1025 			sorwakeup(s);
1026 			return 0;
1027 		}
1028 	}
1029 	m_freem(mm);
1030 	return -1;
1031 }
1032 
1033 /*
1034  * IPv6 multicast forwarding function. This function assumes that the packet
1035  * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1036  * pointed to by "ifp", and the packet is to be relayed to other networks
1037  * that have members of the packet's destination IPv6 multicast group.
1038  *
1039  * The packet is returned unscathed to the caller, unless it is
1040  * erroneous, in which case a non-zero return value tells the caller to
1041  * discard it.
1042  */
1043 
1044 int
1045 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 	int s;
1051 	mifi_t mifi;
1052 	struct sockaddr_in6 sin6;
1053 
1054 #ifdef MRT6DEBUG
1055 	if (mrt6debug & DEBUG_FORWARD)
1056 		log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1057 		    ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst),
1058 		    ifp->if_index);
1059 #endif
1060 
1061 	/*
1062 	 * Don't forward a packet with Hop limit of zero or one,
1063 	 * or a packet destined to a local-only group.
1064 	 */
1065 	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_NODELOCAL(&ip6->ip6_dst) ||
1066 	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1067 		return 0;
1068 	ip6->ip6_hlim--;
1069 
1070 	/*
1071 	 * Source address check: do not forward packets with unspecified
1072 	 * source. It was discussed in July 2000, on ipngwg mailing list.
1073 	 * This is rather more serious than unicast cases, because some
1074 	 * MLD packets can be sent with the unspecified source address
1075 	 * (although such packets must normally set the hop limit field to 1).
1076 	 */
1077 	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1078 		IP6_STATINC(IP6_STAT_CANTFORWARD);
1079 		if (ip6_log_time + ip6_log_interval < time_second) {
1080 			ip6_log_time = time_second;
1081 			log(LOG_DEBUG,
1082 			    "cannot forward "
1083 			    "from %s to %s nxt %d received on %s\n",
1084 			    ip6_sprintf(&ip6->ip6_src),
1085 			    ip6_sprintf(&ip6->ip6_dst),
1086 			    ip6->ip6_nxt,
1087 			    m->m_pkthdr.rcvif ?
1088 			    if_name(m->m_pkthdr.rcvif) : "?");
1089 		}
1090 		return 0;
1091 	}
1092 
1093 	/*
1094 	 * Determine forwarding mifs from the forwarding cache table
1095 	 */
1096 	s = splsoftnet();
1097 	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1098 
1099 	/* Entry exists, so forward if necessary */
1100 	if (rt) {
1101 		splx(s);
1102 		return (ip6_mdq(m, ifp, rt));
1103 	} else {
1104 		/*
1105 		 * If we don't have a route for packet's origin,
1106 		 * Make a copy of the packet &
1107 		 * send message to routing daemon
1108 		 */
1109 
1110 		struct mbuf *mb0;
1111 		struct rtdetq *rte;
1112 		u_long hash;
1113 /*		int i, npkts;*/
1114 #ifdef UPCALL_TIMING
1115 		struct timeval tp;
1116 
1117 		GET_TIME(tp);
1118 #endif /* UPCALL_TIMING */
1119 
1120 		mrt6stat.mrt6s_no_route++;
1121 #ifdef MRT6DEBUG
1122 		if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1123 			log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1124 			    ip6_sprintf(&ip6->ip6_src),
1125 			    ip6_sprintf(&ip6->ip6_dst));
1126 #endif
1127 
1128 		/*
1129 		 * Allocate mbufs early so that we don't do extra work if we
1130 		 * are just going to fail anyway.
1131 		 */
1132 		rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE,
1133 					      M_NOWAIT);
1134 		if (rte == NULL) {
1135 			splx(s);
1136 			return ENOBUFS;
1137 		}
1138 		mb0 = m_copy(m, 0, M_COPYALL);
1139 		/*
1140 		 * Pullup packet header if needed before storing it,
1141 		 * as other references may modify it in the meantime.
1142 		 */
1143 		if (mb0 &&
1144 		    (M_READONLY(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1145 			mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1146 		if (mb0 == NULL) {
1147 			free(rte, M_MRTABLE);
1148 			splx(s);
1149 			return ENOBUFS;
1150 		}
1151 
1152 		/* is there an upcall waiting for this packet? */
1153 		hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1154 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1155 			if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1156 					       &rt->mf6c_origin.sin6_addr) &&
1157 			    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1158 					       &rt->mf6c_mcastgrp.sin6_addr) &&
1159 			    (rt->mf6c_stall != NULL))
1160 				break;
1161 		}
1162 
1163 		if (rt == NULL) {
1164 			struct mrt6msg *im;
1165 			struct omrt6msg *oim;
1166 
1167 			/* no upcall, so make a new entry */
1168 			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE,
1169 						  M_NOWAIT);
1170 			if (rt == NULL) {
1171 				free(rte, M_MRTABLE);
1172 				m_freem(mb0);
1173 				splx(s);
1174 				return ENOBUFS;
1175 			}
1176 			/*
1177 			 * Make a copy of the header to send to the user
1178 			 * level process
1179 			 */
1180 			mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1181 
1182 			if (mm == NULL) {
1183 				free(rte, M_MRTABLE);
1184 				m_freem(mb0);
1185 				free(rt, M_MRTABLE);
1186 				splx(s);
1187 				return ENOBUFS;
1188 			}
1189 
1190 			/*
1191 			 * Send message to routing daemon
1192 			 */
1193 			sockaddr_in6_init(&sin6, &ip6->ip6_src, 0, 0, 0);
1194 
1195 			im = NULL;
1196 			oim = NULL;
1197 			switch (ip6_mrouter_ver) {
1198 			case MRT6_OINIT:
1199 				oim = mtod(mm, struct omrt6msg *);
1200 				oim->im6_msgtype = MRT6MSG_NOCACHE;
1201 				oim->im6_mbz = 0;
1202 				break;
1203 			case MRT6_INIT:
1204 				im = mtod(mm, struct mrt6msg *);
1205 				im->im6_msgtype = MRT6MSG_NOCACHE;
1206 				im->im6_mbz = 0;
1207 				break;
1208 			default:
1209 				free(rte, M_MRTABLE);
1210 				m_freem(mb0);
1211 				free(rt, M_MRTABLE);
1212 				splx(s);
1213 				return EINVAL;
1214 			}
1215 
1216 #ifdef MRT6DEBUG
1217 			if (mrt6debug & DEBUG_FORWARD)
1218 				log(LOG_DEBUG,
1219 				    "getting the iif info in the kernel\n");
1220 #endif
1221 
1222 			for (mifp = mif6table, mifi = 0;
1223 			     mifi < nummifs && mifp->m6_ifp != ifp;
1224 			     mifp++, mifi++)
1225 				;
1226 
1227 			switch (ip6_mrouter_ver) {
1228 			case MRT6_OINIT:
1229 				oim->im6_mif = mifi;
1230 				break;
1231 			case MRT6_INIT:
1232 				im->im6_mif = mifi;
1233 				break;
1234 			}
1235 
1236 			if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1237 				log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1238 				    "socket queue full\n");
1239 				mrt6stat.mrt6s_upq_sockfull++;
1240 				free(rte, M_MRTABLE);
1241 				m_freem(mb0);
1242 				free(rt, M_MRTABLE);
1243 				splx(s);
1244 				return ENOBUFS;
1245 			}
1246 
1247 			mrt6stat.mrt6s_upcalls++;
1248 
1249 			/* insert new entry at head of hash chain */
1250 			memset(rt, 0, sizeof(*rt));
1251 			sockaddr_in6_init(&rt->mf6c_origin, &ip6->ip6_src,
1252 			    0, 0, 0);
1253 			sockaddr_in6_init(&rt->mf6c_mcastgrp, &ip6->ip6_dst,
1254 			    0, 0, 0);
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.mrt6s_upq_ovflw++;
1272 					free(rte, M_MRTABLE);
1273 					m_freem(mb0);
1274 					splx(s);
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 		splx(s);
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 0.25 seconds.
1298  */
1299 static void
1300 expire_upcalls(void *unused)
1301 {
1302 	struct rtdetq *rte;
1303 	struct mf6c *mfc, **nptr;
1304 	int i;
1305 
1306 	mutex_enter(softnet_lock);
1307 	KERNEL_LOCK(1, NULL);
1308 
1309 	for (i = 0; i < MF6CTBLSIZ; i++) {
1310 		if (n6expire[i] == 0)
1311 			continue;
1312 		nptr = &mf6ctable[i];
1313 		while ((mfc = *nptr) != NULL) {
1314 			rte = mfc->mf6c_stall;
1315 			/*
1316 			 * Skip real cache entries
1317 			 * Make sure it wasn't marked to not expire (shouldn't happen)
1318 			 * If it expires now
1319 			 */
1320 			if (rte != NULL &&
1321 			    mfc->mf6c_expire != 0 &&
1322 			    --mfc->mf6c_expire == 0) {
1323 #ifdef MRT6DEBUG
1324 				if (mrt6debug & DEBUG_EXPIRE)
1325 					log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1326 					    ip6_sprintf(&mfc->mf6c_origin.sin6_addr),
1327 					    ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr));
1328 #endif
1329 				/*
1330 				 * drop all the packets
1331 				 * free the mbuf with the pkt, if, timing info
1332 				 */
1333 				do {
1334 					struct rtdetq *n = rte->next;
1335 					m_freem(rte->m);
1336 					free(rte, M_MRTABLE);
1337 					rte = n;
1338 				} while (rte != NULL);
1339 				mrt6stat.mrt6s_cache_cleanups++;
1340 				n6expire[i]--;
1341 
1342 				*nptr = mfc->mf6c_next;
1343 				free(mfc, M_MRTABLE);
1344 			} else {
1345 				nptr = &mfc->mf6c_next;
1346 			}
1347 		}
1348 	}
1349 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1350 	    expire_upcalls, NULL);
1351 
1352 	KERNEL_UNLOCK_ONE(NULL);
1353 	mutex_exit(softnet_lock);
1354 }
1355 
1356 /*
1357  * Packet forwarding routine once entry in the cache is made
1358  */
1359 static int
1360 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1361 {
1362 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1363 	mifi_t mifi, iif;
1364 	struct mif6 *mifp;
1365 	int plen = m->m_pkthdr.len;
1366 	struct in6_addr src0, dst0; /* copies for local work */
1367 	u_int32_t iszone, idzone, oszone, odzone;
1368 	int error = 0;
1369 
1370 /*
1371  * Macro to send packet on mif.  Since RSVP packets don't get counted on
1372  * input, they shouldn't get counted on output, so statistics keeping is
1373  * separate.
1374  */
1375 
1376 #define MC6_SEND(ip6, mifp, m) do {				\
1377 	if ((mifp)->m6_flags & MIFF_REGISTER)			\
1378 		register_send((ip6), (mifp), (m));		\
1379 	else							\
1380 		phyint_send((ip6), (mifp), (m));		\
1381 } while (/*CONSTCOND*/ 0)
1382 
1383 	/*
1384 	 * Don't forward if it didn't arrive from the parent mif
1385 	 * for its origin.
1386 	 */
1387 	mifi = rt->mf6c_parent;
1388 	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1389 		/* came in the wrong interface */
1390 #ifdef MRT6DEBUG
1391 		if (mrt6debug & DEBUG_FORWARD)
1392 			log(LOG_DEBUG,
1393 			    "wrong if: ifid %d mifi %d mififid %x\n",
1394 			    ifp->if_index, mifi,
1395 			    mif6table[mifi].m6_ifp ?
1396 			    mif6table[mifi].m6_ifp->if_index : -1);
1397 #endif
1398 		mrt6stat.mrt6s_wrong_if++;
1399 		rt->mf6c_wrong_if++;
1400 		/*
1401 		 * If we are doing PIM processing, and we are forwarding
1402 		 * packets on this interface, send a message to the
1403 		 * routing daemon.
1404 		 */
1405 		/* have to make sure this is a valid mif */
1406 		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1407 			if (pim6 && (m->m_flags & M_LOOP) == 0) {
1408 				/*
1409 				 * Check the M_LOOP flag to avoid an
1410 				 * unnecessary PIM assert.
1411 				 * XXX: M_LOOP is an ad-hoc hack...
1412 				 */
1413 				struct sockaddr_in6 sin6;
1414 
1415 				struct mbuf *mm;
1416 				struct mrt6msg *im;
1417 				struct omrt6msg *oim;
1418 
1419 				mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1420 				if (mm &&
1421 				    (M_READONLY(mm) ||
1422 				     mm->m_len < sizeof(struct ip6_hdr)))
1423 					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1424 				if (mm == NULL)
1425 					return ENOBUFS;
1426 
1427 				oim = NULL;
1428 				im = NULL;
1429 				switch (ip6_mrouter_ver) {
1430 				case MRT6_OINIT:
1431 					oim = mtod(mm, struct omrt6msg *);
1432 					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1433 					oim->im6_mbz = 0;
1434 					break;
1435 				case MRT6_INIT:
1436 					im = mtod(mm, struct mrt6msg *);
1437 					im->im6_msgtype = MRT6MSG_WRONGMIF;
1438 					im->im6_mbz = 0;
1439 					break;
1440 				default:
1441 					m_freem(mm);
1442 					return EINVAL;
1443 				}
1444 
1445 				for (mifp = mif6table, iif = 0;
1446 				     iif < nummifs && mifp &&
1447 					     mifp->m6_ifp != ifp;
1448 				     mifp++, iif++)
1449 					;
1450 
1451 				memset(&sin6, 0, sizeof(sin6));
1452 				sin6.sin6_len = sizeof(sin6);
1453 				sin6.sin6_family = AF_INET6;
1454 				switch (ip6_mrouter_ver) {
1455 				case MRT6_OINIT:
1456 					oim->im6_mif = iif;
1457 					sin6.sin6_addr = oim->im6_src;
1458 					break;
1459 				case MRT6_INIT:
1460 					im->im6_mif = iif;
1461 					sin6.sin6_addr = im->im6_src;
1462 					break;
1463 				}
1464 
1465 				mrt6stat.mrt6s_upcalls++;
1466 
1467 				if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1468 #ifdef MRT6DEBUG
1469 					if (mrt6debug)
1470 						log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1471 #endif
1472 					++mrt6stat.mrt6s_upq_sockfull;
1473 					return ENOBUFS;
1474 				}	/* if socket Q full */
1475 			}		/* if PIM */
1476 		return 0;
1477 	}			/* if wrong iif */
1478 
1479 	/* If I sourced this packet, it counts as output, else it was input. */
1480 	if (m->m_pkthdr.rcvif == NULL) {
1481 		/* XXX: is rcvif really NULL when output?? */
1482 		mif6table[mifi].m6_pkt_out++;
1483 		mif6table[mifi].m6_bytes_out += plen;
1484 	} else {
1485 		mif6table[mifi].m6_pkt_in++;
1486 		mif6table[mifi].m6_bytes_in += plen;
1487 	}
1488 	rt->mf6c_pkt_cnt++;
1489 	rt->mf6c_byte_cnt += plen;
1490 
1491 	/*
1492 	 * For each mif, forward a copy of the packet if there are group
1493 	 * members downstream on the interface.
1494 	 */
1495 	src0 = ip6->ip6_src;
1496 	dst0 = ip6->ip6_dst;
1497 	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1498 	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1499 		IP6_STATINC(IP6_STAT_BADSCOPE);
1500 		return (error);
1501 	}
1502 	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++)
1503 		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1504 			if (mif6table[mifi].m6_ifp == NULL)
1505 				continue;
1506 			/*
1507 			 * check if the outgoing packet is going to break
1508 			 * a scope boundary.
1509 			 * XXX: For packets through PIM register tunnel
1510 			 * interface, we believe the routing daemon.
1511 			 */
1512 			if ((mif6table[rt->mf6c_parent].m6_flags &
1513 			     MIFF_REGISTER) == 0 &&
1514 			    (mif6table[mifi].m6_flags & MIFF_REGISTER) == 0) {
1515 				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1516 				    &oszone) ||
1517 				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1518 				    &odzone) ||
1519 				    iszone != oszone || idzone != odzone) {
1520 					IP6_STATINC(IP6_STAT_BADSCOPE);
1521 					continue;
1522 				}
1523 			}
1524 
1525 			mifp->m6_pkt_out++;
1526 			mifp->m6_bytes_out += plen;
1527 			MC6_SEND(ip6, mifp, m);
1528 		}
1529 	return 0;
1530 }
1531 
1532 static void
1533 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1534 {
1535 	struct mbuf *mb_copy;
1536 	struct ifnet *ifp = mifp->m6_ifp;
1537 	int error __mrt6debugused = 0;
1538 	int s;
1539 	static struct route ro;
1540 	struct in6_multi *in6m;
1541 	struct sockaddr_in6 dst6;
1542 	u_long linkmtu;
1543 
1544 	s = splsoftnet();
1545 	/*
1546 	 * Make a new reference to the packet; make sure that
1547 	 * the IPv6 header is actually copied, not just referenced,
1548 	 * so that ip6_output() only scribbles on the copy.
1549 	 */
1550 	mb_copy = m_copy(m, 0, M_COPYALL);
1551 	if (mb_copy &&
1552 	    (M_READONLY(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1553 		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1554 	if (mb_copy == NULL) {
1555 		splx(s);
1556 		return;
1557 	}
1558 	/* set MCAST flag to the outgoing packet */
1559 	mb_copy->m_flags |= M_MCAST;
1560 
1561 	/*
1562 	 * If we sourced the packet, call ip6_output since we may divide
1563 	 * the packet into fragments when the packet is too big for the
1564 	 * outgoing interface.
1565 	 * Otherwise, we can simply send the packet to the interface
1566 	 * sending queue.
1567 	 */
1568 	if (m->m_pkthdr.rcvif == NULL) {
1569 		struct ip6_moptions im6o;
1570 
1571 		im6o.im6o_multicast_ifp = ifp;
1572 		/* XXX: ip6_output will override ip6->ip6_hlim */
1573 		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1574 		im6o.im6o_multicast_loop = 1;
1575 		error = ip6_output(mb_copy, NULL, &ro, IPV6_FORWARDING,
1576 				   &im6o, NULL, NULL);
1577 
1578 #ifdef MRT6DEBUG
1579 		if (mrt6debug & DEBUG_XMIT)
1580 			log(LOG_DEBUG, "phyint_send on mif %td err %d\n",
1581 			    mifp - mif6table, error);
1582 #endif
1583 		splx(s);
1584 		return;
1585 	}
1586 
1587 	/*
1588 	 * If we belong to the destination multicast group
1589 	 * on the outgoing interface, loop back a copy.
1590 	 */
1591 	/*
1592 	 * Does not have to check source info, as it's alreay covered by
1593 	 * ip6_input
1594 	 */
1595 	sockaddr_in6_init(&dst6, &ip6->ip6_dst, 0, 0, 0);
1596 
1597 	IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m);
1598 	if (in6m != NULL) {
1599 		ip6_mloopback(ifp, m,
1600 		    satocsin6(rtcache_getdst(&ro)));
1601 	}
1602 
1603 	/*
1604 	 * Put the packet into the sending queue of the outgoing interface
1605 	 * if it would fit in the MTU of the interface.
1606 	 */
1607 	linkmtu = IN6_LINKMTU(ifp);
1608 	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1609 		/*
1610 		 * We could call if_output directly here, but we use
1611 		 * nd6_output on purpose to see if IPv6 operation is allowed
1612 		 * on the interface.
1613 		 */
1614 		error = nd6_output(ifp, ifp, mb_copy, &dst6, NULL);
1615 #ifdef MRT6DEBUG
1616 		if (mrt6debug & DEBUG_XMIT)
1617 			log(LOG_DEBUG, "phyint_send on mif %td err %d\n",
1618 			    mifp - mif6table, error);
1619 #endif
1620 	} else {
1621 		/*
1622 		 * pMTU discovery is intentionally disabled by default, since
1623 		 * various router may notify pMTU in multicast, which can be
1624 		 * a DDoS to a router
1625 		 */
1626 		if (ip6_mcast_pmtu)
1627 			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1628 		else {
1629 #ifdef MRT6DEBUG
1630 			if (mrt6debug & DEBUG_XMIT)
1631 				log(LOG_DEBUG,
1632 				    "phyint_send: packet too big on %s o %s g %s"
1633 				    " size %d(discarded)\n",
1634 				    if_name(ifp),
1635 				    ip6_sprintf(&ip6->ip6_src),
1636 				    ip6_sprintf(&ip6->ip6_dst),
1637 				    mb_copy->m_pkthdr.len);
1638 #endif /* MRT6DEBUG */
1639 			m_freem(mb_copy); /* simply discard the packet */
1640 		}
1641 	}
1642 
1643 	splx(s);
1644 }
1645 
1646 static int
1647 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1648 {
1649 	struct mbuf *mm;
1650 	int i, len = m->m_pkthdr.len;
1651 	struct sockaddr_in6 sin6;
1652 	struct mrt6msg *im6;
1653 
1654 #ifdef MRT6DEBUG
1655 	if (mrt6debug)
1656 		log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1657 		    ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst));
1658 #endif
1659 	PIM6_STATINC(PIM6_STAT_SND_REGISTERS);
1660 
1661 	/* Make a copy of the packet to send to the user level process */
1662 	MGETHDR(mm, M_DONTWAIT, MT_HEADER);
1663 	if (mm == NULL)
1664 		return ENOBUFS;
1665 	mm->m_data += max_linkhdr;
1666 	mm->m_len = sizeof(struct ip6_hdr);
1667 
1668 	if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1669 		m_freem(mm);
1670 		return ENOBUFS;
1671 	}
1672 	i = MHLEN - M_LEADINGSPACE(mm);
1673 	if (i > len)
1674 		i = len;
1675 	mm = m_pullup(mm, i);
1676 	if (mm == NULL)
1677 		return ENOBUFS;
1678 /* TODO: check it! */
1679 	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1680 
1681 	/*
1682 	 * Send message to routing daemon
1683 	 */
1684 	sockaddr_in6_init(&sin6, &ip6->ip6_src, 0, 0, 0);
1685 
1686 	im6 = mtod(mm, struct mrt6msg *);
1687 	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1688 	im6->im6_mbz          = 0;
1689 
1690 	im6->im6_mif = mif - mif6table;
1691 
1692 	/* iif info is not given for reg. encap.n */
1693 	mrt6stat.mrt6s_upcalls++;
1694 
1695 	if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1696 #ifdef MRT6DEBUG
1697 		if (mrt6debug)
1698 			log(LOG_WARNING,
1699 			    "register_send: ip6_mrouter socket queue full\n");
1700 #endif
1701 		++mrt6stat.mrt6s_upq_sockfull;
1702 		return ENOBUFS;
1703 	}
1704 	return 0;
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  */
1714 int
1715 pim6_input(struct mbuf **mp, int *offp, int proto)
1716 {
1717 	struct pim *pim; /* pointer to a pim struct */
1718 	struct ip6_hdr *ip6 __mrt6debugused;
1719 	int pimlen;
1720 	struct mbuf *m = *mp;
1721 	int minlen;
1722 	int off = *offp;
1723 
1724 	PIM6_STATINC(PIM6_STAT_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 		PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
1734 #ifdef MRT6DEBUG
1735 		if (mrt6debug & DEBUG_PIM)
1736 			log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1737 #endif
1738 		m_freem(m);
1739 		return (IPPROTO_DONE);
1740 	}
1741 
1742 	/*
1743 	 * if the packet is at least as big as a REGISTER, go ahead
1744 	 * and grab the PIM REGISTER header size, to avoid another
1745 	 * possible m_pullup() later.
1746 	 *
1747 	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1748 	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1749 	 */
1750 	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1751 
1752 	/*
1753 	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1754 	 * possibly the PIM REGISTER header
1755 	 */
1756 	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1757 	if (pim == NULL) {
1758 		PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
1759 		return IPPROTO_DONE;
1760 	}
1761 
1762 	/* PIM version check */
1763 	if (pim->pim_ver != PIM_VERSION) {
1764 		PIM6_STATINC(PIM6_STAT_RCV_BADVERSION);
1765 #ifdef MRT6DEBUG
1766 		log(LOG_ERR,
1767 		    "pim6_input: incorrect version %d, expecting %d\n",
1768 		    pim->pim_ver, PIM_VERSION);
1769 #endif
1770 		m_freem(m);
1771 		return (IPPROTO_DONE);
1772 	}
1773 
1774 #define PIM6_CHECKSUM
1775 #ifdef PIM6_CHECKSUM
1776 	{
1777 		int cksumlen;
1778 
1779 		/*
1780 		 * Validate checksum.
1781 		 * If PIM REGISTER, exclude the data packet
1782 		 */
1783 		if (pim->pim_type == PIM_REGISTER)
1784 			cksumlen = PIM_MINLEN;
1785 		else
1786 			cksumlen = pimlen;
1787 
1788 		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1789 			PIM6_STATINC(PIM6_STAT_RCV_BADSUM);
1790 #ifdef MRT6DEBUG
1791 			if (mrt6debug & DEBUG_PIM)
1792 				log(LOG_DEBUG,
1793 				    "pim6_input: invalid checksum\n");
1794 #endif
1795 			m_freem(m);
1796 			return (IPPROTO_DONE);
1797 		}
1798 	}
1799 #endif /* PIM_CHECKSUM */
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 const struct sockaddr_in6 dst = {
1808 			.sin6_len = sizeof(dst),
1809 			.sin6_family = AF_INET6,
1810 		};
1811 
1812 		struct mbuf *mcp;
1813 		struct ip6_hdr *eip6;
1814 		u_int32_t *reghdr;
1815 
1816 		PIM6_STATINC(PIM6_STAT_RCV_REGISTERS);
1817 
1818 		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1819 #ifdef MRT6DEBUG
1820 			if (mrt6debug & DEBUG_PIM)
1821 				log(LOG_DEBUG,
1822 				    "pim6_input: register mif not set: %d\n",
1823 				    reg_mif_num);
1824 #endif
1825 			m_freem(m);
1826 			return (IPPROTO_DONE);
1827 		}
1828 
1829 		reghdr = (u_int32_t *)(pim + 1);
1830 
1831 		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1832 			goto pim6_input_to_daemon;
1833 
1834 		/*
1835 		 * Validate length
1836 		 */
1837 		if (pimlen < PIM6_REG_MINLEN) {
1838 			PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
1839 			PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
1840 #ifdef MRT6DEBUG
1841 			log(LOG_ERR,
1842 			    "pim6_input: register packet size too "
1843 			    "small %d from %s\n",
1844 			    pimlen, ip6_sprintf(&ip6->ip6_src));
1845 #endif
1846 			m_freem(m);
1847 			return (IPPROTO_DONE);
1848 		}
1849 
1850 		eip6 = (struct ip6_hdr *) (reghdr + 1);
1851 #ifdef MRT6DEBUG
1852 		if (mrt6debug & DEBUG_PIM)
1853 			log(LOG_DEBUG,
1854 			    "pim6_input[register], eip6: %s -> %s, "
1855 			    "eip6 plen %d\n",
1856 			    ip6_sprintf(&eip6->ip6_src),
1857 			    ip6_sprintf(&eip6->ip6_dst),
1858 			    ntohs(eip6->ip6_plen));
1859 #endif
1860 
1861 		/* verify the version number of the inner packet */
1862 		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1863 			PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
1864 #ifdef MRT6DEBUG
1865 			log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1866 			    "of the inner packet\n",
1867 			    (eip6->ip6_vfc & IPV6_VERSION));
1868 #endif
1869 			m_freem(m);
1870 			return (IPPROTO_NONE);
1871 		}
1872 
1873 		/* verify the inner packet is destined to a mcast group */
1874 		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1875 			PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
1876 #ifdef MRT6DEBUG
1877 			if (mrt6debug & DEBUG_PIM)
1878 				log(LOG_DEBUG,
1879 				    "pim6_input: inner packet of register "
1880 				    "is not multicast %s\n",
1881 				    ip6_sprintf(&eip6->ip6_dst));
1882 #endif
1883 			m_freem(m);
1884 			return (IPPROTO_DONE);
1885 		}
1886 
1887 		/*
1888 		 * make a copy of the whole header to pass to the daemon later.
1889 		 */
1890 		mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1891 		if (mcp == NULL) {
1892 #ifdef MRT6DEBUG
1893 			log(LOG_ERR,
1894 			    "pim6_input: pim register: "
1895 			    "could not copy register head\n");
1896 #endif
1897 			m_freem(m);
1898 			return (IPPROTO_DONE);
1899 		}
1900 
1901 		/*
1902 		 * forward the inner ip6 packet; point m_data at the inner ip6.
1903 		 */
1904 		m_adj(m, off + PIM_MINLEN);
1905 #ifdef MRT6DEBUG
1906 		if (mrt6debug & DEBUG_PIM) {
1907 			log(LOG_DEBUG,
1908 			    "pim6_input: forwarding decapsulated register: "
1909 			    "src %s, dst %s, mif %d\n",
1910 			    ip6_sprintf(&eip6->ip6_src),
1911 			    ip6_sprintf(&eip6->ip6_dst),
1912 			    reg_mif_num);
1913 		}
1914 #endif
1915 
1916 		looutput(mif6table[reg_mif_num].m6_ifp, m,
1917 			      (struct sockaddr *)__UNCONST(&dst), NULL);
1918 
1919 		/* prepare the register head to send to the mrouting daemon */
1920 		m = mcp;
1921 	}
1922 
1923 	/*
1924 	 * Pass the PIM message up to the daemon; if it is a register message
1925 	 * pass the 'head' only up to the daemon. This includes the
1926 	 * encapsulator ip6 header, pim header, register header and the
1927 	 * encapsulated ip6 header.
1928 	 */
1929   pim6_input_to_daemon:
1930 	rip6_input(&m, offp, proto);
1931 	return (IPPROTO_DONE);
1932 }
1933 
1934 static int
1935 sysctl_net_inet6_pim6_stats(SYSCTLFN_ARGS)
1936 {
1937 
1938 	return (NETSTAT_SYSCTL(pim6stat_percpu, PIM6_NSTATS));
1939 }
1940 
1941 static void
1942 sysctl_net_inet6_pim6_setup(struct sysctllog **clog)
1943 {
1944 	sysctl_createv(clog, 0, NULL, NULL,
1945 		       CTLFLAG_PERMANENT,
1946 		       CTLTYPE_NODE, "net", NULL,
1947 		       NULL, 0, NULL, 0,
1948 		       CTL_NET, CTL_EOL);
1949 	sysctl_createv(clog, 0, NULL, NULL,
1950 		       CTLFLAG_PERMANENT,
1951 		       CTLTYPE_NODE, "inet6", NULL,
1952 		       NULL, 0, NULL, 0,
1953 		       CTL_NET, PF_INET6, CTL_EOL);
1954 	sysctl_createv(clog, 0, NULL, NULL,
1955 		       CTLFLAG_PERMANENT,
1956 		       CTLTYPE_NODE, "pim6",
1957 		       SYSCTL_DESCR("PIMv6 settings"),
1958 		       NULL, 0, NULL, 0,
1959 		       CTL_NET, PF_INET6, IPPROTO_PIM, CTL_EOL);
1960 
1961 	sysctl_createv(clog, 0, NULL, NULL,
1962 		       CTLFLAG_PERMANENT,
1963 		       CTLTYPE_STRUCT, "stats",
1964 		       SYSCTL_DESCR("PIMv6 statistics"),
1965 		       sysctl_net_inet6_pim6_stats, 0, NULL, 0,
1966 		       CTL_NET, PF_INET6, IPPROTO_PIM, PIM6CTL_STATS,
1967 		       CTL_EOL);
1968 }
1969