xref: /freebsd/sys/netinet6/ip6_mroute.c (revision 595e514d0df2bac5b813d35f83e32875dbf16a83)
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("$FreeBSD$");
83 
84 #include "opt_inet6.h"
85 
86 #include <sys/param.h>
87 #include <sys/callout.h>
88 #include <sys/errno.h>
89 #include <sys/kernel.h>
90 #include <sys/lock.h>
91 #include <sys/malloc.h>
92 #include <sys/mbuf.h>
93 #include <sys/module.h>
94 #include <sys/domain.h>
95 #include <sys/protosw.h>
96 #include <sys/signalvar.h>
97 #include <sys/socket.h>
98 #include <sys/socketvar.h>
99 #include <sys/sockio.h>
100 #include <sys/sx.h>
101 #include <sys/sysctl.h>
102 #include <sys/syslog.h>
103 #include <sys/systm.h>
104 #include <sys/time.h>
105 
106 #include <net/if.h>
107 #include <net/if_types.h>
108 #include <net/raw_cb.h>
109 #include <net/vnet.h>
110 
111 #include <netinet/in.h>
112 #include <netinet/in_var.h>
113 #include <netinet/icmp6.h>
114 #include <netinet/ip_encap.h>
115 
116 #include <netinet/ip6.h>
117 #include <netinet6/ip6_var.h>
118 #include <netinet6/scope6_var.h>
119 #include <netinet6/nd6.h>
120 #include <netinet6/ip6_mroute.h>
121 #include <netinet6/ip6protosw.h>
122 #include <netinet6/pim6.h>
123 #include <netinet6/pim6_var.h>
124 
125 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
126 
127 /* XXX: this is a very common idiom; move to <sys/mbuf.h> ? */
128 #define M_HASCL(m) ((m)->m_flags & M_EXT)
129 
130 static int	ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
131 static void	phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
132 static int	register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
133 static int	set_pim6(int *);
134 static int	socket_send(struct socket *, struct mbuf *,
135 		    struct sockaddr_in6 *);
136 
137 extern int in6_mcast_loop;
138 extern struct domain inet6domain;
139 
140 static const struct encaptab *pim6_encap_cookie;
141 static const struct ip6protosw in6_pim_protosw = {
142 	.pr_type =		SOCK_RAW,
143 	.pr_domain =		&inet6domain,
144 	.pr_protocol =		IPPROTO_PIM,
145 	.pr_flags =		PR_ATOMIC|PR_ADDR|PR_LASTHDR,
146 	.pr_input =		pim6_input,
147 	.pr_output =		rip6_output,
148 	.pr_ctloutput =		rip6_ctloutput,
149 	.pr_usrreqs =		&rip6_usrreqs
150 };
151 static int pim6_encapcheck(const struct mbuf *, int, int, void *);
152 
153 static VNET_DEFINE(int, ip6_mrouter_ver) = 0;
154 #define	V_ip6_mrouter_ver	VNET(ip6_mrouter_ver)
155 
156 SYSCTL_DECL(_net_inet6);
157 SYSCTL_DECL(_net_inet6_ip6);
158 static SYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim, CTLFLAG_RW, 0, "PIM");
159 
160 static struct mrt6stat mrt6stat;
161 SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
162     &mrt6stat, mrt6stat,
163     "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
164 
165 #define NO_RTE_FOUND	0x1
166 #define RTE_FOUND	0x2
167 
168 static struct mtx mrouter6_mtx;
169 #define	MROUTER6_LOCK()		mtx_lock(&mrouter6_mtx)
170 #define	MROUTER6_UNLOCK()	mtx_unlock(&mrouter6_mtx)
171 #define	MROUTER6_LOCK_ASSERT()	do {					\
172 	mtx_assert(&mrouter6_mtx, MA_OWNED);				\
173 	NET_ASSERT_GIANT();						\
174 } while (0)
175 #define	MROUTER6_LOCK_INIT()	\
176 	mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
177 #define	MROUTER6_LOCK_DESTROY()	mtx_destroy(&mrouter6_mtx)
178 
179 static struct mf6c *mf6ctable[MF6CTBLSIZ];
180 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
181     &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
182     "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
183     "netinet6/ip6_mroute.h)");
184 
185 static struct mtx mfc6_mtx;
186 #define	MFC6_LOCK()		mtx_lock(&mfc6_mtx)
187 #define	MFC6_UNLOCK()		mtx_unlock(&mfc6_mtx)
188 #define	MFC6_LOCK_ASSERT()	do {					\
189 	mtx_assert(&mfc6_mtx, MA_OWNED);				\
190 	NET_ASSERT_GIANT();						\
191 } while (0)
192 #define	MFC6_LOCK_INIT()		\
193 	mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
194 #define	MFC6_LOCK_DESTROY()	mtx_destroy(&mfc6_mtx)
195 
196 static u_char n6expire[MF6CTBLSIZ];
197 
198 static struct mif6 mif6table[MAXMIFS];
199 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mif6table, CTLFLAG_RD,
200     &mif6table, sizeof(mif6table), "S,mif6[MAXMIFS]",
201     "IPv6 Multicast Interfaces (struct mif6[MAXMIFS], netinet6/ip6_mroute.h)");
202 
203 static struct mtx mif6_mtx;
204 #define	MIF6_LOCK()		mtx_lock(&mif6_mtx)
205 #define	MIF6_UNLOCK()		mtx_unlock(&mif6_mtx)
206 #define	MIF6_LOCK_ASSERT()	mtx_assert(&mif6_mtx, MA_OWNED)
207 #define	MIF6_LOCK_INIT()	\
208 	mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
209 #define	MIF6_LOCK_DESTROY()	mtx_destroy(&mif6_mtx)
210 
211 #ifdef MRT6DEBUG
212 static VNET_DEFINE(u_int, mrt6debug) = 0;	/* debug level */
213 #define	V_mrt6debug		VNET(mrt6debug)
214 #define DEBUG_MFC	0x02
215 #define DEBUG_FORWARD	0x04
216 #define DEBUG_EXPIRE	0x08
217 #define DEBUG_XMIT	0x10
218 #define DEBUG_REG	0x20
219 #define DEBUG_PIM	0x40
220 #endif
221 
222 static void	expire_upcalls(void *);
223 #define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
224 #define	UPCALL_EXPIRE	6		/* number of timeouts */
225 
226 /*
227  * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
228  */
229 
230 /*
231  * 'Interfaces' associated with decapsulator (so we can tell
232  * packets that went through it from ones that get reflected
233  * by a broken gateway).  Different from IPv4 register_if,
234  * these interfaces are linked into the system ifnet list,
235  * because per-interface IPv6 statistics are maintained in
236  * ifp->if_afdata.  But it does not have any routes point
237  * to them.  I.e., packets can't be sent this way.  They
238  * only exist as a placeholder for multicast source
239  * verification.
240  */
241 static struct ifnet *multicast_register_if6;
242 
243 #define ENCAP_HOPS 64
244 
245 /*
246  * Private variables.
247  */
248 static mifi_t nummifs = 0;
249 static mifi_t reg_mif_num = (mifi_t)-1;
250 
251 static struct pim6stat pim6stat;
252 SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RW,
253     &pim6stat, pim6stat,
254     "PIM Statistics (struct pim6stat, netinet6/pim_var.h)");
255 
256 static VNET_DEFINE(int, pim6);
257 #define	V_pim6		VNET(pim6)
258 
259 /*
260  * Hash function for a source, group entry
261  */
262 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
263 				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
264 				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
265 				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
266 
267 /*
268  * Find a route for a given origin IPv6 address and Multicast group address.
269  */
270 #define MF6CFIND(o, g, rt) do { \
271 	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
272 	rt = NULL; \
273 	mrt6stat.mrt6s_mfc_lookups++; \
274 	while (_rt) { \
275 		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
276 		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
277 		    (_rt->mf6c_stall == NULL)) { \
278 			rt = _rt; \
279 			break; \
280 		} \
281 		_rt = _rt->mf6c_next; \
282 	} \
283 	if (rt == NULL) { \
284 		mrt6stat.mrt6s_mfc_misses++; \
285 	} \
286 } while (/*CONSTCOND*/ 0)
287 
288 /*
289  * Macros to compute elapsed time efficiently
290  * Borrowed from Van Jacobson's scheduling code
291  * XXX: replace with timersub() ?
292  */
293 #define TV_DELTA(a, b, delta) do { \
294 	    int xxs; \
295 		\
296 	    delta = (a).tv_usec - (b).tv_usec; \
297 	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
298 	       switch (xxs) { \
299 		      case 2: \
300 			  delta += 1000000; \
301 			      /* FALLTHROUGH */ \
302 		      case 1: \
303 			  delta += 1000000; \
304 			  break; \
305 		      default: \
306 			  delta += (1000000 * xxs); \
307 	       } \
308 	    } \
309 } while (/*CONSTCOND*/ 0)
310 
311 /* XXX: replace with timercmp(a, b, <) ? */
312 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
313 	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
314 
315 #ifdef UPCALL_TIMING
316 #define UPCALL_MAX	50
317 static u_long upcall_data[UPCALL_MAX + 1];
318 static void collate();
319 #endif /* UPCALL_TIMING */
320 
321 static int ip6_mrouter_init(struct socket *, int, int);
322 static int add_m6fc(struct mf6cctl *);
323 static int add_m6if(struct mif6ctl *);
324 static int del_m6fc(struct mf6cctl *);
325 static int del_m6if(mifi_t *);
326 static int del_m6if_locked(mifi_t *);
327 static int get_mif6_cnt(struct sioc_mif_req6 *);
328 static int get_sg_cnt(struct sioc_sg_req6 *);
329 
330 static struct callout expire_upcalls_ch;
331 
332 int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
333 int X_ip6_mrouter_done(void);
334 int X_ip6_mrouter_set(struct socket *, struct sockopt *);
335 int X_ip6_mrouter_get(struct socket *, struct sockopt *);
336 int X_mrt6_ioctl(u_long, caddr_t);
337 
338 /*
339  * Handle MRT setsockopt commands to modify the multicast routing tables.
340  */
341 int
342 X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
343 {
344 	int error = 0;
345 	int optval;
346 	struct mif6ctl mifc;
347 	struct mf6cctl mfcc;
348 	mifi_t mifi;
349 
350 	if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
351 		return (EACCES);
352 
353 	switch (sopt->sopt_name) {
354 	case MRT6_INIT:
355 #ifdef MRT6_OINIT
356 	case MRT6_OINIT:
357 #endif
358 		error = sooptcopyin(sopt, &optval, sizeof(optval),
359 		    sizeof(optval));
360 		if (error)
361 			break;
362 		error = ip6_mrouter_init(so, optval, sopt->sopt_name);
363 		break;
364 	case MRT6_DONE:
365 		error = X_ip6_mrouter_done();
366 		break;
367 	case MRT6_ADD_MIF:
368 		error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
369 		if (error)
370 			break;
371 		error = add_m6if(&mifc);
372 		break;
373 	case MRT6_ADD_MFC:
374 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
375 		if (error)
376 			break;
377 		error = add_m6fc(&mfcc);
378 		break;
379 	case MRT6_DEL_MFC:
380 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
381 		if (error)
382 			break;
383 		error = del_m6fc(&mfcc);
384 		break;
385 	case MRT6_DEL_MIF:
386 		error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
387 		if (error)
388 			break;
389 		error = del_m6if(&mifi);
390 		break;
391 	case MRT6_PIM:
392 		error = sooptcopyin(sopt, &optval, sizeof(optval),
393 		    sizeof(optval));
394 		if (error)
395 			break;
396 		error = set_pim6(&optval);
397 		break;
398 	default:
399 		error = EOPNOTSUPP;
400 		break;
401 	}
402 
403 	return (error);
404 }
405 
406 /*
407  * Handle MRT getsockopt commands
408  */
409 int
410 X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
411 {
412 	int error = 0;
413 
414 	if (so != V_ip6_mrouter)
415 		return (EACCES);
416 
417 	switch (sopt->sopt_name) {
418 		case MRT6_PIM:
419 			error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
420 			break;
421 	}
422 	return (error);
423 }
424 
425 /*
426  * Handle ioctl commands to obtain information from the cache
427  */
428 int
429 X_mrt6_ioctl(u_long cmd, caddr_t data)
430 {
431 	int ret;
432 
433 	ret = EINVAL;
434 
435 	switch (cmd) {
436 	case SIOCGETSGCNT_IN6:
437 		ret = get_sg_cnt((struct sioc_sg_req6 *)data);
438 		break;
439 
440 	case SIOCGETMIFCNT_IN6:
441 		ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
442 		break;
443 
444 	default:
445 		break;
446 	}
447 
448 	return (ret);
449 }
450 
451 /*
452  * returns the packet, byte, rpf-failure count for the source group provided
453  */
454 static int
455 get_sg_cnt(struct sioc_sg_req6 *req)
456 {
457 	struct mf6c *rt;
458 	int ret;
459 
460 	ret = 0;
461 
462 	MFC6_LOCK();
463 
464 	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
465 	if (rt == NULL) {
466 		ret = ESRCH;
467 	} else {
468 		req->pktcnt = rt->mf6c_pkt_cnt;
469 		req->bytecnt = rt->mf6c_byte_cnt;
470 		req->wrong_if = rt->mf6c_wrong_if;
471 	}
472 
473 	MFC6_UNLOCK();
474 
475 	return (ret);
476 }
477 
478 /*
479  * returns the input and output packet and byte counts on the mif provided
480  */
481 static int
482 get_mif6_cnt(struct sioc_mif_req6 *req)
483 {
484 	mifi_t mifi;
485 	int ret;
486 
487 	ret = 0;
488 	mifi = req->mifi;
489 
490 	MIF6_LOCK();
491 
492 	if (mifi >= nummifs) {
493 		ret = EINVAL;
494 	} else {
495 		req->icount = mif6table[mifi].m6_pkt_in;
496 		req->ocount = mif6table[mifi].m6_pkt_out;
497 		req->ibytes = mif6table[mifi].m6_bytes_in;
498 		req->obytes = mif6table[mifi].m6_bytes_out;
499 	}
500 
501 	MIF6_UNLOCK();
502 
503 	return (ret);
504 }
505 
506 static int
507 set_pim6(int *i)
508 {
509 	if ((*i != 1) && (*i != 0))
510 		return (EINVAL);
511 
512 	V_pim6 = *i;
513 
514 	return (0);
515 }
516 
517 /*
518  * Enable multicast routing
519  */
520 static int
521 ip6_mrouter_init(struct socket *so, int v, int cmd)
522 {
523 
524 #ifdef MRT6DEBUG
525 	if (V_mrt6debug)
526 		log(LOG_DEBUG,
527 		    "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
528 		    so->so_type, so->so_proto->pr_protocol);
529 #endif
530 
531 	if (so->so_type != SOCK_RAW ||
532 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
533 		return (EOPNOTSUPP);
534 
535 	if (v != 1)
536 		return (ENOPROTOOPT);
537 
538 	MROUTER6_LOCK();
539 
540 	if (V_ip6_mrouter != NULL) {
541 		MROUTER6_UNLOCK();
542 		return (EADDRINUSE);
543 	}
544 
545 	V_ip6_mrouter = so;
546 	V_ip6_mrouter_ver = cmd;
547 
548 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
549 	bzero((caddr_t)n6expire, sizeof(n6expire));
550 
551 	V_pim6 = 0;/* used for stubbing out/in pim stuff */
552 
553 	callout_init(&expire_upcalls_ch, 0);
554 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
555 	    expire_upcalls, NULL);
556 
557 	MROUTER6_UNLOCK();
558 
559 #ifdef MRT6DEBUG
560 	if (V_mrt6debug)
561 		log(LOG_DEBUG, "ip6_mrouter_init\n");
562 #endif
563 
564 	return (0);
565 }
566 
567 /*
568  * Disable IPv6 multicast forwarding.
569  */
570 int
571 X_ip6_mrouter_done(void)
572 {
573 	mifi_t mifi;
574 	int i;
575 	struct mf6c *rt;
576 	struct rtdetq *rte;
577 
578 	MROUTER6_LOCK();
579 
580 	if (V_ip6_mrouter == NULL) {
581 		MROUTER6_UNLOCK();
582 		return (EINVAL);
583 	}
584 
585 	/*
586 	 * For each phyint in use, disable promiscuous reception of all IPv6
587 	 * multicasts.
588 	 */
589 	for (mifi = 0; mifi < nummifs; mifi++) {
590 		if (mif6table[mifi].m6_ifp &&
591 		    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
592 			if_allmulti(mif6table[mifi].m6_ifp, 0);
593 		}
594 	}
595 	bzero((caddr_t)mif6table, sizeof(mif6table));
596 	nummifs = 0;
597 
598 	V_pim6 = 0; /* used to stub out/in pim specific code */
599 
600 	callout_stop(&expire_upcalls_ch);
601 
602 	/*
603 	 * Free all multicast forwarding cache entries.
604 	 */
605 	MFC6_LOCK();
606 	for (i = 0; i < MF6CTBLSIZ; i++) {
607 		rt = mf6ctable[i];
608 		while (rt) {
609 			struct mf6c *frt;
610 
611 			for (rte = rt->mf6c_stall; rte != NULL; ) {
612 				struct rtdetq *n = rte->next;
613 
614 				m_free(rte->m);
615 				free(rte, M_MRTABLE6);
616 				rte = n;
617 			}
618 			frt = rt;
619 			rt = rt->mf6c_next;
620 			free(frt, M_MRTABLE6);
621 		}
622 	}
623 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
624 	MFC6_UNLOCK();
625 
626 	/*
627 	 * Reset register interface
628 	 */
629 	if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
630 		if_detach(multicast_register_if6);
631 		if_free(multicast_register_if6);
632 		reg_mif_num = (mifi_t)-1;
633 		multicast_register_if6 = NULL;
634 	}
635 
636 	V_ip6_mrouter = NULL;
637 	V_ip6_mrouter_ver = 0;
638 
639 	MROUTER6_UNLOCK();
640 
641 #ifdef MRT6DEBUG
642 	if (V_mrt6debug)
643 		log(LOG_DEBUG, "ip6_mrouter_done\n");
644 #endif
645 
646 	return (0);
647 }
648 
649 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
650 
651 /*
652  * Add a mif to the mif table
653  */
654 static int
655 add_m6if(struct mif6ctl *mifcp)
656 {
657 	struct mif6 *mifp;
658 	struct ifnet *ifp;
659 	int error;
660 
661 	MIF6_LOCK();
662 
663 	if (mifcp->mif6c_mifi >= MAXMIFS) {
664 		MIF6_UNLOCK();
665 		return (EINVAL);
666 	}
667 	mifp = mif6table + mifcp->mif6c_mifi;
668 	if (mifp->m6_ifp != NULL) {
669 		MIF6_UNLOCK();
670 		return (EADDRINUSE); /* XXX: is it appropriate? */
671 	}
672 	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
673 		MIF6_UNLOCK();
674 		return (ENXIO);
675 	}
676 
677 	ifp = ifnet_byindex(mifcp->mif6c_pifi);
678 
679 	if (mifcp->mif6c_flags & MIFF_REGISTER) {
680 		if (reg_mif_num == (mifi_t)-1) {
681 			ifp = if_alloc(IFT_OTHER);
682 
683 			if_initname(ifp, "register_mif", 0);
684 			ifp->if_flags |= IFF_LOOPBACK;
685 			if_attach(ifp);
686 			multicast_register_if6 = ifp;
687 			reg_mif_num = mifcp->mif6c_mifi;
688 			/*
689 			 * it is impossible to guess the ifindex of the
690 			 * register interface.  So mif6c_pifi is automatically
691 			 * calculated.
692 			 */
693 			mifcp->mif6c_pifi = ifp->if_index;
694 		} else {
695 			ifp = multicast_register_if6;
696 		}
697 	} else {
698 		/* Make sure the interface supports multicast */
699 		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
700 			MIF6_UNLOCK();
701 			return (EOPNOTSUPP);
702 		}
703 
704 		error = if_allmulti(ifp, 1);
705 		if (error) {
706 			MIF6_UNLOCK();
707 			return (error);
708 		}
709 	}
710 
711 	mifp->m6_flags     = mifcp->mif6c_flags;
712 	mifp->m6_ifp       = ifp;
713 
714 	/* initialize per mif pkt counters */
715 	mifp->m6_pkt_in    = 0;
716 	mifp->m6_pkt_out   = 0;
717 	mifp->m6_bytes_in  = 0;
718 	mifp->m6_bytes_out = 0;
719 
720 	/* Adjust nummifs up if the mifi is higher than nummifs */
721 	if (nummifs <= mifcp->mif6c_mifi)
722 		nummifs = mifcp->mif6c_mifi + 1;
723 
724 	MIF6_UNLOCK();
725 
726 #ifdef MRT6DEBUG
727 	if (V_mrt6debug)
728 		log(LOG_DEBUG,
729 		    "add_mif #%d, phyint %s\n",
730 		    mifcp->mif6c_mifi,
731 		    ifp->if_xname);
732 #endif
733 
734 	return (0);
735 }
736 
737 /*
738  * Delete a mif from the mif table
739  */
740 static int
741 del_m6if_locked(mifi_t *mifip)
742 {
743 	struct mif6 *mifp = mif6table + *mifip;
744 	mifi_t mifi;
745 	struct ifnet *ifp;
746 
747 	MIF6_LOCK_ASSERT();
748 
749 	if (*mifip >= nummifs)
750 		return (EINVAL);
751 	if (mifp->m6_ifp == NULL)
752 		return (EINVAL);
753 
754 	if (!(mifp->m6_flags & MIFF_REGISTER)) {
755 		/* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
756 		ifp = mifp->m6_ifp;
757 		if_allmulti(ifp, 0);
758 	} else {
759 		if (reg_mif_num != (mifi_t)-1 &&
760 		    multicast_register_if6 != NULL) {
761 			if_detach(multicast_register_if6);
762 			if_free(multicast_register_if6);
763 			reg_mif_num = (mifi_t)-1;
764 			multicast_register_if6 = NULL;
765 		}
766 	}
767 
768 	bzero((caddr_t)mifp, sizeof(*mifp));
769 
770 	/* Adjust nummifs down */
771 	for (mifi = nummifs; mifi > 0; mifi--)
772 		if (mif6table[mifi - 1].m6_ifp)
773 			break;
774 	nummifs = mifi;
775 
776 #ifdef MRT6DEBUG
777 	if (V_mrt6debug)
778 		log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
779 #endif
780 
781 	return (0);
782 }
783 
784 static int
785 del_m6if(mifi_t *mifip)
786 {
787 	int cc;
788 
789 	MIF6_LOCK();
790 	cc = del_m6if_locked(mifip);
791 	MIF6_UNLOCK();
792 
793 	return (cc);
794 }
795 
796 /*
797  * Add an mfc entry
798  */
799 static int
800 add_m6fc(struct mf6cctl *mfccp)
801 {
802 	struct mf6c *rt;
803 	u_long hash;
804 	struct rtdetq *rte;
805 	u_short nstl;
806 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
807 
808 	MFC6_LOCK();
809 
810 	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
811 		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
812 
813 	/* If an entry already exists, just update the fields */
814 	if (rt) {
815 #ifdef MRT6DEBUG
816 		if (V_mrt6debug & DEBUG_MFC) {
817 		    log(LOG_DEBUG,
818 			"add_m6fc no upcall h %d o %s g %s p %x\n",
819 			ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
820 			ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
821 			mfccp->mf6cc_parent);
822 		}
823 #endif
824 
825 		rt->mf6c_parent = mfccp->mf6cc_parent;
826 		rt->mf6c_ifset = mfccp->mf6cc_ifset;
827 
828 		MFC6_UNLOCK();
829 		return (0);
830 	}
831 
832 	/*
833 	 * Find the entry for which the upcall was made and update
834 	 */
835 	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
836 			mfccp->mf6cc_mcastgrp.sin6_addr);
837 	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
838 		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
839 				       &mfccp->mf6cc_origin.sin6_addr) &&
840 		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
841 				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
842 		    (rt->mf6c_stall != NULL)) {
843 
844 			if (nstl++)
845 				log(LOG_ERR,
846 				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
847 				    "multiple kernel entries",
848 				    ip6_sprintf(ip6bufo,
849 					    &mfccp->mf6cc_origin.sin6_addr),
850 				    ip6_sprintf(ip6bufg,
851 					    &mfccp->mf6cc_mcastgrp.sin6_addr),
852 				    mfccp->mf6cc_parent, rt->mf6c_stall);
853 
854 #ifdef MRT6DEBUG
855 			if (V_mrt6debug & DEBUG_MFC)
856 				log(LOG_DEBUG,
857 				    "add_m6fc o %s g %s p %x dbg %x\n",
858 				    ip6_sprintf(ip6bufo,
859 					    &mfccp->mf6cc_origin.sin6_addr),
860 				    ip6_sprintf(ip6bufg,
861 					    &mfccp->mf6cc_mcastgrp.sin6_addr),
862 				    mfccp->mf6cc_parent, rt->mf6c_stall);
863 #endif
864 
865 			rt->mf6c_origin     = mfccp->mf6cc_origin;
866 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
867 			rt->mf6c_parent     = mfccp->mf6cc_parent;
868 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
869 			/* initialize pkt counters per src-grp */
870 			rt->mf6c_pkt_cnt    = 0;
871 			rt->mf6c_byte_cnt   = 0;
872 			rt->mf6c_wrong_if   = 0;
873 
874 			rt->mf6c_expire = 0;	/* Don't clean this guy up */
875 			n6expire[hash]--;
876 
877 			/* free packets Qed at the end of this entry */
878 			for (rte = rt->mf6c_stall; rte != NULL; ) {
879 				struct rtdetq *n = rte->next;
880 				ip6_mdq(rte->m, rte->ifp, rt);
881 				m_freem(rte->m);
882 #ifdef UPCALL_TIMING
883 				collate(&(rte->t));
884 #endif /* UPCALL_TIMING */
885 				free(rte, M_MRTABLE6);
886 				rte = n;
887 			}
888 			rt->mf6c_stall = NULL;
889 		}
890 	}
891 
892 	/*
893 	 * It is possible that an entry is being inserted without an upcall
894 	 */
895 	if (nstl == 0) {
896 #ifdef MRT6DEBUG
897 		if (V_mrt6debug & DEBUG_MFC)
898 		    log(LOG_DEBUG,
899 			"add_mfc no upcall h %d o %s g %s p %x\n",
900 			hash,
901 			ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
902 			ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
903 			mfccp->mf6cc_parent);
904 #endif
905 
906 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
907 
908 			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
909 					       &mfccp->mf6cc_origin.sin6_addr)&&
910 			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
911 					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
912 
913 				rt->mf6c_origin     = mfccp->mf6cc_origin;
914 				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
915 				rt->mf6c_parent     = mfccp->mf6cc_parent;
916 				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
917 				/* initialize pkt counters per src-grp */
918 				rt->mf6c_pkt_cnt    = 0;
919 				rt->mf6c_byte_cnt   = 0;
920 				rt->mf6c_wrong_if   = 0;
921 
922 				if (rt->mf6c_expire)
923 					n6expire[hash]--;
924 				rt->mf6c_expire	   = 0;
925 			}
926 		}
927 		if (rt == NULL) {
928 			/* no upcall, so make a new entry */
929 			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
930 						  M_NOWAIT);
931 			if (rt == NULL) {
932 				MFC6_UNLOCK();
933 				return (ENOBUFS);
934 			}
935 
936 			/* insert new entry at head of hash chain */
937 			rt->mf6c_origin     = mfccp->mf6cc_origin;
938 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
939 			rt->mf6c_parent     = mfccp->mf6cc_parent;
940 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
941 			/* initialize pkt counters per src-grp */
942 			rt->mf6c_pkt_cnt    = 0;
943 			rt->mf6c_byte_cnt   = 0;
944 			rt->mf6c_wrong_if   = 0;
945 			rt->mf6c_expire     = 0;
946 			rt->mf6c_stall = NULL;
947 
948 			/* link into table */
949 			rt->mf6c_next  = mf6ctable[hash];
950 			mf6ctable[hash] = rt;
951 		}
952 	}
953 
954 	MFC6_UNLOCK();
955 	return (0);
956 }
957 
958 #ifdef UPCALL_TIMING
959 /*
960  * collect delay statistics on the upcalls
961  */
962 static void
963 collate(struct timeval *t)
964 {
965 	u_long d;
966 	struct timeval tp;
967 	u_long delta;
968 
969 	GET_TIME(tp);
970 
971 	if (TV_LT(*t, tp))
972 	{
973 		TV_DELTA(tp, *t, delta);
974 
975 		d = delta >> 10;
976 		if (d > UPCALL_MAX)
977 			d = UPCALL_MAX;
978 
979 		++upcall_data[d];
980 	}
981 }
982 #endif /* UPCALL_TIMING */
983 
984 /*
985  * Delete an mfc entry
986  */
987 static int
988 del_m6fc(struct mf6cctl *mfccp)
989 {
990 	struct sockaddr_in6	origin;
991 	struct sockaddr_in6	mcastgrp;
992 	struct mf6c		*rt;
993 	struct mf6c		**nptr;
994 	u_long		hash;
995 
996 	origin = mfccp->mf6cc_origin;
997 	mcastgrp = mfccp->mf6cc_mcastgrp;
998 	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
999 
1000 #ifdef MRT6DEBUG
1001 	if (V_mrt6debug & DEBUG_MFC) {
1002 		char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1003 		log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
1004 		    ip6_sprintf(ip6bufo, &origin.sin6_addr),
1005 		    ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
1006 	}
1007 #endif
1008 
1009 	MFC6_LOCK();
1010 
1011 	nptr = &mf6ctable[hash];
1012 	while ((rt = *nptr) != NULL) {
1013 		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
1014 				       &rt->mf6c_origin.sin6_addr) &&
1015 		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1016 				       &rt->mf6c_mcastgrp.sin6_addr) &&
1017 		    rt->mf6c_stall == NULL)
1018 			break;
1019 
1020 		nptr = &rt->mf6c_next;
1021 	}
1022 	if (rt == NULL) {
1023 		MFC6_UNLOCK();
1024 		return (EADDRNOTAVAIL);
1025 	}
1026 
1027 	*nptr = rt->mf6c_next;
1028 	free(rt, M_MRTABLE6);
1029 
1030 	MFC6_UNLOCK();
1031 
1032 	return (0);
1033 }
1034 
1035 static int
1036 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1037 {
1038 
1039 	if (s) {
1040 		if (sbappendaddr(&s->so_rcv,
1041 				 (struct sockaddr *)src,
1042 				 mm, (struct mbuf *)0) != 0) {
1043 			sorwakeup(s);
1044 			return (0);
1045 		}
1046 	}
1047 	m_freem(mm);
1048 	return (-1);
1049 }
1050 
1051 /*
1052  * IPv6 multicast forwarding function. This function assumes that the packet
1053  * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1054  * pointed to by "ifp", and the packet is to be relayed to other networks
1055  * that have members of the packet's destination IPv6 multicast group.
1056  *
1057  * The packet is returned unscathed to the caller, unless it is
1058  * erroneous, in which case a non-zero return value tells the caller to
1059  * discard it.
1060  *
1061  * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1062  * this function is called in the originating context (i.e., not when
1063  * forwarding a packet from other node).  ip6_output(), which is currently the
1064  * only function that calls this function is called in the originating context,
1065  * explicitly ensures this condition.  It is caller's responsibility to ensure
1066  * that if this function is called from somewhere else in the originating
1067  * context in the future.
1068  */
1069 int
1070 X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1071 {
1072 	struct mf6c *rt;
1073 	struct mif6 *mifp;
1074 	struct mbuf *mm;
1075 	mifi_t mifi;
1076 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1077 
1078 #ifdef MRT6DEBUG
1079 	if (V_mrt6debug & DEBUG_FORWARD)
1080 		log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1081 		    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1082 		    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1083 		    ifp->if_index);
1084 #endif
1085 
1086 	/*
1087 	 * Don't forward a packet with Hop limit of zero or one,
1088 	 * or a packet destined to a local-only group.
1089 	 */
1090 	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1091 	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1092 		return (0);
1093 	ip6->ip6_hlim--;
1094 
1095 	/*
1096 	 * Source address check: do not forward packets with unspecified
1097 	 * source. It was discussed in July 2000, on ipngwg mailing list.
1098 	 * This is rather more serious than unicast cases, because some
1099 	 * MLD packets can be sent with the unspecified source address
1100 	 * (although such packets must normally set 1 to the hop limit field).
1101 	 */
1102 	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1103 		IP6STAT_INC(ip6s_cantforward);
1104 		if (V_ip6_log_time + V_ip6_log_interval < time_second) {
1105 			V_ip6_log_time = time_second;
1106 			log(LOG_DEBUG,
1107 			    "cannot forward "
1108 			    "from %s to %s nxt %d received on %s\n",
1109 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1110 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1111 			    ip6->ip6_nxt,
1112 			    if_name(m->m_pkthdr.rcvif));
1113 		}
1114 		return (0);
1115 	}
1116 
1117 	MFC6_LOCK();
1118 
1119 	/*
1120 	 * Determine forwarding mifs from the forwarding cache table
1121 	 */
1122 	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1123 
1124 	/* Entry exists, so forward if necessary */
1125 	if (rt) {
1126 		MFC6_UNLOCK();
1127 		return (ip6_mdq(m, ifp, rt));
1128 	} else {
1129 		/*
1130 		 * If we don't have a route for packet's origin,
1131 		 * Make a copy of the packet &
1132 		 * send message to routing daemon
1133 		 */
1134 
1135 		struct mbuf *mb0;
1136 		struct rtdetq *rte;
1137 		u_long hash;
1138 /*		int i, npkts;*/
1139 #ifdef UPCALL_TIMING
1140 		struct timeval tp;
1141 
1142 		GET_TIME(tp);
1143 #endif /* UPCALL_TIMING */
1144 
1145 		mrt6stat.mrt6s_no_route++;
1146 #ifdef MRT6DEBUG
1147 		if (V_mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1148 			log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1149 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1150 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1151 #endif
1152 
1153 		/*
1154 		 * Allocate mbufs early so that we don't do extra work if we
1155 		 * are just going to fail anyway.
1156 		 */
1157 		rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6,
1158 					      M_NOWAIT);
1159 		if (rte == NULL) {
1160 			MFC6_UNLOCK();
1161 			return (ENOBUFS);
1162 		}
1163 		mb0 = m_copy(m, 0, M_COPYALL);
1164 		/*
1165 		 * Pullup packet header if needed before storing it,
1166 		 * as other references may modify it in the meantime.
1167 		 */
1168 		if (mb0 &&
1169 		    (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1170 			mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1171 		if (mb0 == NULL) {
1172 			free(rte, M_MRTABLE6);
1173 			MFC6_UNLOCK();
1174 			return (ENOBUFS);
1175 		}
1176 
1177 		/* is there an upcall waiting for this packet? */
1178 		hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1179 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1180 			if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1181 					       &rt->mf6c_origin.sin6_addr) &&
1182 			    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1183 					       &rt->mf6c_mcastgrp.sin6_addr) &&
1184 			    (rt->mf6c_stall != NULL))
1185 				break;
1186 		}
1187 
1188 		if (rt == NULL) {
1189 			struct mrt6msg *im;
1190 #ifdef MRT6_OINIT
1191 			struct omrt6msg *oim;
1192 #endif
1193 
1194 			/* no upcall, so make a new entry */
1195 			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
1196 						  M_NOWAIT);
1197 			if (rt == NULL) {
1198 				free(rte, M_MRTABLE6);
1199 				m_freem(mb0);
1200 				MFC6_UNLOCK();
1201 				return (ENOBUFS);
1202 			}
1203 			/*
1204 			 * Make a copy of the header to send to the user
1205 			 * level process
1206 			 */
1207 			mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1208 
1209 			if (mm == NULL) {
1210 				free(rte, M_MRTABLE6);
1211 				m_freem(mb0);
1212 				free(rt, M_MRTABLE6);
1213 				MFC6_UNLOCK();
1214 				return (ENOBUFS);
1215 			}
1216 
1217 			/*
1218 			 * Send message to routing daemon
1219 			 */
1220 			sin6.sin6_addr = ip6->ip6_src;
1221 
1222 			im = NULL;
1223 #ifdef MRT6_OINIT
1224 			oim = NULL;
1225 #endif
1226 			switch (V_ip6_mrouter_ver) {
1227 #ifdef MRT6_OINIT
1228 			case MRT6_OINIT:
1229 				oim = mtod(mm, struct omrt6msg *);
1230 				oim->im6_msgtype = MRT6MSG_NOCACHE;
1231 				oim->im6_mbz = 0;
1232 				break;
1233 #endif
1234 			case MRT6_INIT:
1235 				im = mtod(mm, struct mrt6msg *);
1236 				im->im6_msgtype = MRT6MSG_NOCACHE;
1237 				im->im6_mbz = 0;
1238 				break;
1239 			default:
1240 				free(rte, M_MRTABLE6);
1241 				m_freem(mb0);
1242 				free(rt, M_MRTABLE6);
1243 				MFC6_UNLOCK();
1244 				return (EINVAL);
1245 			}
1246 
1247 #ifdef MRT6DEBUG
1248 			if (V_mrt6debug & DEBUG_FORWARD)
1249 				log(LOG_DEBUG,
1250 				    "getting the iif info in the kernel\n");
1251 #endif
1252 
1253 			for (mifp = mif6table, mifi = 0;
1254 			     mifi < nummifs && mifp->m6_ifp != ifp;
1255 			     mifp++, mifi++)
1256 				;
1257 
1258 			switch (V_ip6_mrouter_ver) {
1259 #ifdef MRT6_OINIT
1260 			case MRT6_OINIT:
1261 				oim->im6_mif = mifi;
1262 				break;
1263 #endif
1264 			case MRT6_INIT:
1265 				im->im6_mif = mifi;
1266 				break;
1267 			}
1268 
1269 			if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1270 				log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1271 				    "socket queue full\n");
1272 				mrt6stat.mrt6s_upq_sockfull++;
1273 				free(rte, M_MRTABLE6);
1274 				m_freem(mb0);
1275 				free(rt, M_MRTABLE6);
1276 				MFC6_UNLOCK();
1277 				return (ENOBUFS);
1278 			}
1279 
1280 			mrt6stat.mrt6s_upcalls++;
1281 
1282 			/* insert new entry at head of hash chain */
1283 			bzero(rt, sizeof(*rt));
1284 			rt->mf6c_origin.sin6_family = AF_INET6;
1285 			rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1286 			rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1287 			rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1288 			rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1289 			rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1290 			rt->mf6c_expire = UPCALL_EXPIRE;
1291 			n6expire[hash]++;
1292 			rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1293 
1294 			/* link into table */
1295 			rt->mf6c_next  = mf6ctable[hash];
1296 			mf6ctable[hash] = rt;
1297 			/* Add this entry to the end of the queue */
1298 			rt->mf6c_stall = rte;
1299 		} else {
1300 			/* determine if q has overflowed */
1301 			struct rtdetq **p;
1302 			int npkts = 0;
1303 
1304 			for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1305 				if (++npkts > MAX_UPQ6) {
1306 					mrt6stat.mrt6s_upq_ovflw++;
1307 					free(rte, M_MRTABLE6);
1308 					m_freem(mb0);
1309 					MFC6_UNLOCK();
1310 					return (0);
1311 				}
1312 
1313 			/* Add this entry to the end of the queue */
1314 			*p = rte;
1315 		}
1316 
1317 		rte->next = NULL;
1318 		rte->m = mb0;
1319 		rte->ifp = ifp;
1320 #ifdef UPCALL_TIMING
1321 		rte->t = tp;
1322 #endif /* UPCALL_TIMING */
1323 
1324 		MFC6_UNLOCK();
1325 
1326 		return (0);
1327 	}
1328 }
1329 
1330 /*
1331  * Clean up cache entries if upcalls are not serviced
1332  * Call from the Slow Timeout mechanism, every half second.
1333  */
1334 static void
1335 expire_upcalls(void *unused)
1336 {
1337 	struct rtdetq *rte;
1338 	struct mf6c *mfc, **nptr;
1339 	int i;
1340 
1341 	MFC6_LOCK();
1342 	for (i = 0; i < MF6CTBLSIZ; i++) {
1343 		if (n6expire[i] == 0)
1344 			continue;
1345 		nptr = &mf6ctable[i];
1346 		while ((mfc = *nptr) != NULL) {
1347 			rte = mfc->mf6c_stall;
1348 			/*
1349 			 * Skip real cache entries
1350 			 * Make sure it wasn't marked to not expire (shouldn't happen)
1351 			 * If it expires now
1352 			 */
1353 			if (rte != NULL &&
1354 			    mfc->mf6c_expire != 0 &&
1355 			    --mfc->mf6c_expire == 0) {
1356 #ifdef MRT6DEBUG
1357 				if (V_mrt6debug & DEBUG_EXPIRE) {
1358 					char ip6bufo[INET6_ADDRSTRLEN];
1359 					char ip6bufg[INET6_ADDRSTRLEN];
1360 					log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1361 					    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1362 					    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1363 				}
1364 #endif
1365 				/*
1366 				 * drop all the packets
1367 				 * free the mbuf with the pkt, if, timing info
1368 				 */
1369 				do {
1370 					struct rtdetq *n = rte->next;
1371 					m_freem(rte->m);
1372 					free(rte, M_MRTABLE6);
1373 					rte = n;
1374 				} while (rte != NULL);
1375 				mrt6stat.mrt6s_cache_cleanups++;
1376 				n6expire[i]--;
1377 
1378 				*nptr = mfc->mf6c_next;
1379 				free(mfc, M_MRTABLE6);
1380 			} else {
1381 				nptr = &mfc->mf6c_next;
1382 			}
1383 		}
1384 	}
1385 	MFC6_UNLOCK();
1386 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1387 	    expire_upcalls, NULL);
1388 }
1389 
1390 /*
1391  * Packet forwarding routine once entry in the cache is made
1392  */
1393 static int
1394 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1395 {
1396 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1397 	mifi_t mifi, iif;
1398 	struct mif6 *mifp;
1399 	int plen = m->m_pkthdr.len;
1400 	struct in6_addr src0, dst0; /* copies for local work */
1401 	u_int32_t iszone, idzone, oszone, odzone;
1402 	int error = 0;
1403 
1404 /*
1405  * Macro to send packet on mif.  Since RSVP packets don't get counted on
1406  * input, they shouldn't get counted on output, so statistics keeping is
1407  * separate.
1408  */
1409 
1410 #define MC6_SEND(ip6, mifp, m) do {				\
1411 	if ((mifp)->m6_flags & MIFF_REGISTER)			\
1412 		register_send((ip6), (mifp), (m));		\
1413 	else							\
1414 		phyint_send((ip6), (mifp), (m));		\
1415 } while (/*CONSTCOND*/ 0)
1416 
1417 	/*
1418 	 * Don't forward if it didn't arrive from the parent mif
1419 	 * for its origin.
1420 	 */
1421 	mifi = rt->mf6c_parent;
1422 	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1423 		/* came in the wrong interface */
1424 #ifdef MRT6DEBUG
1425 		if (V_mrt6debug & DEBUG_FORWARD)
1426 			log(LOG_DEBUG,
1427 			    "wrong if: ifid %d mifi %d mififid %x\n",
1428 			    ifp->if_index, mifi,
1429 			    mif6table[mifi].m6_ifp->if_index);
1430 #endif
1431 		mrt6stat.mrt6s_wrong_if++;
1432 		rt->mf6c_wrong_if++;
1433 		/*
1434 		 * If we are doing PIM processing, and we are forwarding
1435 		 * packets on this interface, send a message to the
1436 		 * routing daemon.
1437 		 */
1438 		/* have to make sure this is a valid mif */
1439 		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1440 			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1441 				/*
1442 				 * Check the M_LOOP flag to avoid an
1443 				 * unnecessary PIM assert.
1444 				 * XXX: M_LOOP is an ad-hoc hack...
1445 				 */
1446 				static struct sockaddr_in6 sin6 =
1447 				{ sizeof(sin6), AF_INET6 };
1448 
1449 				struct mbuf *mm;
1450 				struct mrt6msg *im;
1451 #ifdef MRT6_OINIT
1452 				struct omrt6msg *oim;
1453 #endif
1454 
1455 				mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1456 				if (mm &&
1457 				    (M_HASCL(mm) ||
1458 				     mm->m_len < sizeof(struct ip6_hdr)))
1459 					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1460 				if (mm == NULL)
1461 					return (ENOBUFS);
1462 
1463 #ifdef MRT6_OINIT
1464 				oim = NULL;
1465 #endif
1466 				im = NULL;
1467 				switch (V_ip6_mrouter_ver) {
1468 #ifdef MRT6_OINIT
1469 				case MRT6_OINIT:
1470 					oim = mtod(mm, struct omrt6msg *);
1471 					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1472 					oim->im6_mbz = 0;
1473 					break;
1474 #endif
1475 				case MRT6_INIT:
1476 					im = mtod(mm, struct mrt6msg *);
1477 					im->im6_msgtype = MRT6MSG_WRONGMIF;
1478 					im->im6_mbz = 0;
1479 					break;
1480 				default:
1481 					m_freem(mm);
1482 					return (EINVAL);
1483 				}
1484 
1485 				for (mifp = mif6table, iif = 0;
1486 				     iif < nummifs && mifp &&
1487 					     mifp->m6_ifp != ifp;
1488 				     mifp++, iif++)
1489 					;
1490 
1491 				switch (V_ip6_mrouter_ver) {
1492 #ifdef MRT6_OINIT
1493 				case MRT6_OINIT:
1494 					oim->im6_mif = iif;
1495 					sin6.sin6_addr = oim->im6_src;
1496 					break;
1497 #endif
1498 				case MRT6_INIT:
1499 					im->im6_mif = iif;
1500 					sin6.sin6_addr = im->im6_src;
1501 					break;
1502 				}
1503 
1504 				mrt6stat.mrt6s_upcalls++;
1505 
1506 				if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1507 #ifdef MRT6DEBUG
1508 					if (V_mrt6debug)
1509 						log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1510 #endif
1511 					++mrt6stat.mrt6s_upq_sockfull;
1512 					return (ENOBUFS);
1513 				}	/* if socket Q full */
1514 			}		/* if PIM */
1515 		return (0);
1516 	}			/* if wrong iif */
1517 
1518 	/* If I sourced this packet, it counts as output, else it was input. */
1519 	if (m->m_pkthdr.rcvif == NULL) {
1520 		/* XXX: is rcvif really NULL when output?? */
1521 		mif6table[mifi].m6_pkt_out++;
1522 		mif6table[mifi].m6_bytes_out += plen;
1523 	} else {
1524 		mif6table[mifi].m6_pkt_in++;
1525 		mif6table[mifi].m6_bytes_in += plen;
1526 	}
1527 	rt->mf6c_pkt_cnt++;
1528 	rt->mf6c_byte_cnt += plen;
1529 
1530 	/*
1531 	 * For each mif, forward a copy of the packet if there are group
1532 	 * members downstream on the interface.
1533 	 */
1534 	src0 = ip6->ip6_src;
1535 	dst0 = ip6->ip6_dst;
1536 	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1537 	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1538 		IP6STAT_INC(ip6s_badscope);
1539 		return (error);
1540 	}
1541 	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1542 		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1543 			/*
1544 			 * check if the outgoing packet is going to break
1545 			 * a scope boundary.
1546 			 * XXX For packets through PIM register tunnel
1547 			 * interface, we believe a routing daemon.
1548 			 */
1549 			if (!(mif6table[rt->mf6c_parent].m6_flags &
1550 			      MIFF_REGISTER) &&
1551 			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1552 				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1553 				    &oszone) ||
1554 				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1555 				    &odzone) ||
1556 				    iszone != oszone ||
1557 				    idzone != odzone) {
1558 					IP6STAT_INC(ip6s_badscope);
1559 					continue;
1560 				}
1561 			}
1562 
1563 			mifp->m6_pkt_out++;
1564 			mifp->m6_bytes_out += plen;
1565 			MC6_SEND(ip6, mifp, m);
1566 		}
1567 	}
1568 	return (0);
1569 }
1570 
1571 static void
1572 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1573 {
1574 	struct mbuf *mb_copy;
1575 	struct ifnet *ifp = mifp->m6_ifp;
1576 	int error = 0;
1577 	u_long linkmtu;
1578 
1579 	/*
1580 	 * Make a new reference to the packet; make sure that
1581 	 * the IPv6 header is actually copied, not just referenced,
1582 	 * so that ip6_output() only scribbles on the copy.
1583 	 */
1584 	mb_copy = m_copy(m, 0, M_COPYALL);
1585 	if (mb_copy &&
1586 	    (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1587 		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1588 	if (mb_copy == NULL) {
1589 		return;
1590 	}
1591 	/* set MCAST flag to the outgoing packet */
1592 	mb_copy->m_flags |= M_MCAST;
1593 
1594 	/*
1595 	 * If we sourced the packet, call ip6_output since we may devide
1596 	 * the packet into fragments when the packet is too big for the
1597 	 * outgoing interface.
1598 	 * Otherwise, we can simply send the packet to the interface
1599 	 * sending queue.
1600 	 */
1601 	if (m->m_pkthdr.rcvif == NULL) {
1602 		struct ip6_moptions im6o;
1603 
1604 		im6o.im6o_multicast_ifp = ifp;
1605 		/* XXX: ip6_output will override ip6->ip6_hlim */
1606 		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1607 		im6o.im6o_multicast_loop = 1;
1608 		error = ip6_output(mb_copy, NULL, NULL, IPV6_FORWARDING, &im6o,
1609 		    NULL, NULL);
1610 
1611 #ifdef MRT6DEBUG
1612 		if (V_mrt6debug & DEBUG_XMIT)
1613 			log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1614 			    mifp - mif6table, error);
1615 #endif
1616 		return;
1617 	}
1618 
1619 	/*
1620 	 * If configured to loop back multicasts by default,
1621 	 * loop back a copy now.
1622 	 */
1623 	if (in6_mcast_loop) {
1624 		struct sockaddr_in6 dst6;
1625 
1626 		bzero(&dst6, sizeof(dst6));
1627 		dst6.sin6_len = sizeof(struct sockaddr_in6);
1628 		dst6.sin6_family = AF_INET6;
1629 		dst6.sin6_addr = ip6->ip6_dst;
1630 		ip6_mloopback(ifp, m, &dst6);
1631 	}
1632 
1633 	/*
1634 	 * Put the packet into the sending queue of the outgoing interface
1635 	 * if it would fit in the MTU of the interface.
1636 	 */
1637 	linkmtu = IN6_LINKMTU(ifp);
1638 	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1639 		struct sockaddr_in6 dst6;
1640 
1641 		bzero(&dst6, sizeof(dst6));
1642 		dst6.sin6_len = sizeof(struct sockaddr_in6);
1643 		dst6.sin6_family = AF_INET6;
1644 		dst6.sin6_addr = ip6->ip6_dst;
1645 		/*
1646 		 * We just call if_output instead of nd6_output here, since
1647 		 * we need no ND for a multicast forwarded packet...right?
1648 		 */
1649 		error = (*ifp->if_output)(ifp, mb_copy,
1650 		    (struct sockaddr *)&dst6, NULL);
1651 #ifdef MRT6DEBUG
1652 		if (V_mrt6debug & DEBUG_XMIT)
1653 			log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1654 			    mifp - mif6table, error);
1655 #endif
1656 	} else {
1657 		/*
1658 		 * pMTU discovery is intentionally disabled by default, since
1659 		 * various router may notify pMTU in multicast, which can be
1660 		 * a DDoS to a router
1661 		 */
1662 		if (V_ip6_mcast_pmtu)
1663 			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1664 		else {
1665 #ifdef MRT6DEBUG
1666 			if (V_mrt6debug & DEBUG_XMIT) {
1667 				char ip6bufs[INET6_ADDRSTRLEN];
1668 				char ip6bufd[INET6_ADDRSTRLEN];
1669 				log(LOG_DEBUG,
1670 				    "phyint_send: packet too big on %s o %s "
1671 				    "g %s size %d(discarded)\n",
1672 				    if_name(ifp),
1673 				    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1674 				    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1675 				    mb_copy->m_pkthdr.len);
1676 			}
1677 #endif /* MRT6DEBUG */
1678 			m_freem(mb_copy); /* simply discard the packet */
1679 		}
1680 	}
1681 }
1682 
1683 static int
1684 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1685 {
1686 	struct mbuf *mm;
1687 	int i, len = m->m_pkthdr.len;
1688 	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1689 	struct mrt6msg *im6;
1690 
1691 #ifdef MRT6DEBUG
1692 	if (V_mrt6debug) {
1693 		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1694 		log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1695 		    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1696 		    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1697 	}
1698 #endif
1699 	++pim6stat.pim6s_snd_registers;
1700 
1701 	/* Make a copy of the packet to send to the user level process. */
1702 	mm = m_gethdr(M_NOWAIT, MT_DATA);
1703 	if (mm == NULL)
1704 		return (ENOBUFS);
1705 	mm->m_data += max_linkhdr;
1706 	mm->m_len = sizeof(struct ip6_hdr);
1707 
1708 	if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1709 		m_freem(mm);
1710 		return (ENOBUFS);
1711 	}
1712 	i = MHLEN - M_LEADINGSPACE(mm);
1713 	if (i > len)
1714 		i = len;
1715 	mm = m_pullup(mm, i);
1716 	if (mm == NULL)
1717 		return (ENOBUFS);
1718 /* TODO: check it! */
1719 	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1720 
1721 	/*
1722 	 * Send message to routing daemon
1723 	 */
1724 	sin6.sin6_addr = ip6->ip6_src;
1725 
1726 	im6 = mtod(mm, struct mrt6msg *);
1727 	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1728 	im6->im6_mbz          = 0;
1729 
1730 	im6->im6_mif = mif - mif6table;
1731 
1732 	/* iif info is not given for reg. encap.n */
1733 	mrt6stat.mrt6s_upcalls++;
1734 
1735 	if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1736 #ifdef MRT6DEBUG
1737 		if (V_mrt6debug)
1738 			log(LOG_WARNING,
1739 			    "register_send: ip6_mrouter socket queue full\n");
1740 #endif
1741 		++mrt6stat.mrt6s_upq_sockfull;
1742 		return (ENOBUFS);
1743 	}
1744 	return (0);
1745 }
1746 
1747 /*
1748  * pim6_encapcheck() is called by the encap6_input() path at runtime to
1749  * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1750  * into the kernel.
1751  */
1752 static int
1753 pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
1754 {
1755 
1756 #ifdef DIAGNOSTIC
1757     KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1758 #endif
1759     if (proto != IPPROTO_PIM)
1760 	return 0;	/* not for us; reject the datagram. */
1761 
1762     return 64;		/* claim the datagram. */
1763 }
1764 
1765 /*
1766  * PIM sparse mode hook
1767  * Receives the pim control messages, and passes them up to the listening
1768  * socket, using rip6_input.
1769  * The only message processed is the REGISTER pim message; the pim header
1770  * is stripped off, and the inner packet is passed to register_mforward.
1771  */
1772 int
1773 pim6_input(struct mbuf **mp, int *offp, int proto)
1774 {
1775 	struct pim *pim; /* pointer to a pim struct */
1776 	struct ip6_hdr *ip6;
1777 	int pimlen;
1778 	struct mbuf *m = *mp;
1779 	int minlen;
1780 	int off = *offp;
1781 
1782 	++pim6stat.pim6s_rcv_total;
1783 
1784 	ip6 = mtod(m, struct ip6_hdr *);
1785 	pimlen = m->m_pkthdr.len - *offp;
1786 
1787 	/*
1788 	 * Validate lengths
1789 	 */
1790 	if (pimlen < PIM_MINLEN) {
1791 		++pim6stat.pim6s_rcv_tooshort;
1792 #ifdef MRT6DEBUG
1793 		if (V_mrt6debug & DEBUG_PIM)
1794 			log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1795 #endif
1796 		m_freem(m);
1797 		return (IPPROTO_DONE);
1798 	}
1799 
1800 	/*
1801 	 * if the packet is at least as big as a REGISTER, go ahead
1802 	 * and grab the PIM REGISTER header size, to avoid another
1803 	 * possible m_pullup() later.
1804 	 *
1805 	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1806 	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1807 	 */
1808 	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1809 
1810 	/*
1811 	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1812 	 * possibly the PIM REGISTER header
1813 	 */
1814 #ifndef PULLDOWN_TEST
1815 	IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1816 	/* adjust pointer */
1817 	ip6 = mtod(m, struct ip6_hdr *);
1818 
1819 	/* adjust mbuf to point to the PIM header */
1820 	pim = (struct pim *)((caddr_t)ip6 + off);
1821 #else
1822 	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1823 	if (pim == NULL) {
1824 		pim6stat.pim6s_rcv_tooshort++;
1825 		return (IPPROTO_DONE);
1826 	}
1827 #endif
1828 
1829 #define PIM6_CHECKSUM
1830 #ifdef PIM6_CHECKSUM
1831 	{
1832 		int cksumlen;
1833 
1834 		/*
1835 		 * Validate checksum.
1836 		 * If PIM REGISTER, exclude the data packet
1837 		 */
1838 		if (pim->pim_type == PIM_REGISTER)
1839 			cksumlen = PIM_MINLEN;
1840 		else
1841 			cksumlen = pimlen;
1842 
1843 		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1844 			++pim6stat.pim6s_rcv_badsum;
1845 #ifdef MRT6DEBUG
1846 			if (V_mrt6debug & DEBUG_PIM)
1847 				log(LOG_DEBUG,
1848 				    "pim6_input: invalid checksum\n");
1849 #endif
1850 			m_freem(m);
1851 			return (IPPROTO_DONE);
1852 		}
1853 	}
1854 #endif /* PIM_CHECKSUM */
1855 
1856 	/* PIM version check */
1857 	if (pim->pim_ver != PIM_VERSION) {
1858 		++pim6stat.pim6s_rcv_badversion;
1859 #ifdef MRT6DEBUG
1860 		log(LOG_ERR,
1861 		    "pim6_input: incorrect version %d, expecting %d\n",
1862 		    pim->pim_ver, PIM_VERSION);
1863 #endif
1864 		m_freem(m);
1865 		return (IPPROTO_DONE);
1866 	}
1867 
1868 	if (pim->pim_type == PIM_REGISTER) {
1869 		/*
1870 		 * since this is a REGISTER, we'll make a copy of the register
1871 		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1872 		 * routing daemon.
1873 		 */
1874 		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1875 
1876 		struct mbuf *mcp;
1877 		struct ip6_hdr *eip6;
1878 		u_int32_t *reghdr;
1879 		int rc;
1880 #ifdef MRT6DEBUG
1881 		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1882 #endif
1883 
1884 		++pim6stat.pim6s_rcv_registers;
1885 
1886 		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1887 #ifdef MRT6DEBUG
1888 			if (V_mrt6debug & DEBUG_PIM)
1889 				log(LOG_DEBUG,
1890 				    "pim6_input: register mif not set: %d\n",
1891 				    reg_mif_num);
1892 #endif
1893 			m_freem(m);
1894 			return (IPPROTO_DONE);
1895 		}
1896 
1897 		reghdr = (u_int32_t *)(pim + 1);
1898 
1899 		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1900 			goto pim6_input_to_daemon;
1901 
1902 		/*
1903 		 * Validate length
1904 		 */
1905 		if (pimlen < PIM6_REG_MINLEN) {
1906 			++pim6stat.pim6s_rcv_tooshort;
1907 			++pim6stat.pim6s_rcv_badregisters;
1908 #ifdef MRT6DEBUG
1909 			log(LOG_ERR,
1910 			    "pim6_input: register packet size too "
1911 			    "small %d from %s\n",
1912 			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1913 #endif
1914 			m_freem(m);
1915 			return (IPPROTO_DONE);
1916 		}
1917 
1918 		eip6 = (struct ip6_hdr *) (reghdr + 1);
1919 #ifdef MRT6DEBUG
1920 		if (V_mrt6debug & DEBUG_PIM)
1921 			log(LOG_DEBUG,
1922 			    "pim6_input[register], eip6: %s -> %s, "
1923 			    "eip6 plen %d\n",
1924 			    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1925 			    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1926 			    ntohs(eip6->ip6_plen));
1927 #endif
1928 
1929 		/* verify the version number of the inner packet */
1930 		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1931 			++pim6stat.pim6s_rcv_badregisters;
1932 #ifdef MRT6DEBUG
1933 			log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1934 			    "of the inner packet\n",
1935 			    (eip6->ip6_vfc & IPV6_VERSION));
1936 #endif
1937 			m_freem(m);
1938 			return (IPPROTO_NONE);
1939 		}
1940 
1941 		/* verify the inner packet is destined to a mcast group */
1942 		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1943 			++pim6stat.pim6s_rcv_badregisters;
1944 #ifdef MRT6DEBUG
1945 			if (V_mrt6debug & DEBUG_PIM)
1946 				log(LOG_DEBUG,
1947 				    "pim6_input: inner packet of register "
1948 				    "is not multicast %s\n",
1949 				    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1950 #endif
1951 			m_freem(m);
1952 			return (IPPROTO_DONE);
1953 		}
1954 
1955 		/*
1956 		 * make a copy of the whole header to pass to the daemon later.
1957 		 */
1958 		mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1959 		if (mcp == NULL) {
1960 #ifdef MRT6DEBUG
1961 			log(LOG_ERR,
1962 			    "pim6_input: pim register: "
1963 			    "could not copy register head\n");
1964 #endif
1965 			m_freem(m);
1966 			return (IPPROTO_DONE);
1967 		}
1968 
1969 		/*
1970 		 * forward the inner ip6 packet; point m_data at the inner ip6.
1971 		 */
1972 		m_adj(m, off + PIM_MINLEN);
1973 #ifdef MRT6DEBUG
1974 		if (V_mrt6debug & DEBUG_PIM) {
1975 			log(LOG_DEBUG,
1976 			    "pim6_input: forwarding decapsulated register: "
1977 			    "src %s, dst %s, mif %d\n",
1978 			    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1979 			    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1980 			    reg_mif_num);
1981 		}
1982 #endif
1983 
1984 		rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1985 				dst.sin6_family, 0);
1986 
1987 		/* prepare the register head to send to the mrouting daemon */
1988 		m = mcp;
1989 	}
1990 
1991 	/*
1992 	 * Pass the PIM message up to the daemon; if it is a register message
1993 	 * pass the 'head' only up to the daemon. This includes the
1994 	 * encapsulator ip6 header, pim header, register header and the
1995 	 * encapsulated ip6 header.
1996 	 */
1997   pim6_input_to_daemon:
1998 	rip6_input(&m, offp, proto);
1999 	return (IPPROTO_DONE);
2000 }
2001 
2002 static int
2003 ip6_mroute_modevent(module_t mod, int type, void *unused)
2004 {
2005 
2006 	switch (type) {
2007 	case MOD_LOAD:
2008 		MROUTER6_LOCK_INIT();
2009 		MFC6_LOCK_INIT();
2010 		MIF6_LOCK_INIT();
2011 
2012 		pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM,
2013 			pim6_encapcheck,
2014 			(const struct protosw *)&in6_pim_protosw, NULL);
2015 		if (pim6_encap_cookie == NULL) {
2016 			printf("ip6_mroute: unable to attach pim6 encap\n");
2017 			MIF6_LOCK_DESTROY();
2018 			MFC6_LOCK_DESTROY();
2019 			MROUTER6_LOCK_DESTROY();
2020 			return (EINVAL);
2021 		}
2022 
2023 		ip6_mforward = X_ip6_mforward;
2024 		ip6_mrouter_done = X_ip6_mrouter_done;
2025 		ip6_mrouter_get = X_ip6_mrouter_get;
2026 		ip6_mrouter_set = X_ip6_mrouter_set;
2027 		mrt6_ioctl = X_mrt6_ioctl;
2028 		break;
2029 
2030 	case MOD_UNLOAD:
2031 		if (V_ip6_mrouter != NULL)
2032 			return EINVAL;
2033 
2034 		if (pim6_encap_cookie) {
2035 			encap_detach(pim6_encap_cookie);
2036 			pim6_encap_cookie = NULL;
2037 		}
2038 		X_ip6_mrouter_done();
2039 		ip6_mforward = NULL;
2040 		ip6_mrouter_done = NULL;
2041 		ip6_mrouter_get = NULL;
2042 		ip6_mrouter_set = NULL;
2043 		mrt6_ioctl = NULL;
2044 
2045 		MIF6_LOCK_DESTROY();
2046 		MFC6_LOCK_DESTROY();
2047 		MROUTER6_LOCK_DESTROY();
2048 		break;
2049 
2050 	default:
2051 		return (EOPNOTSUPP);
2052 	}
2053 
2054 	return (0);
2055 }
2056 
2057 static moduledata_t ip6_mroutemod = {
2058 	"ip6_mroute",
2059 	ip6_mroute_modevent,
2060 	0
2061 };
2062 
2063 DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PSEUDO, SI_ORDER_ANY);
2064