xref: /freebsd/sys/netinet6/ip6_mroute.c (revision 830940567b49bb0c08dfaed40418999e76616909)
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_inet.h"
85 #include "opt_inet6.h"
86 
87 #include <sys/param.h>
88 #include <sys/callout.h>
89 #include <sys/errno.h>
90 #include <sys/kernel.h>
91 #include <sys/lock.h>
92 #include <sys/malloc.h>
93 #include <sys/mbuf.h>
94 #include <sys/module.h>
95 #include <sys/domain.h>
96 #include <sys/protosw.h>
97 #include <sys/signalvar.h>
98 #include <sys/socket.h>
99 #include <sys/socketvar.h>
100 #include <sys/sockio.h>
101 #include <sys/sx.h>
102 #include <sys/sysctl.h>
103 #include <sys/syslog.h>
104 #include <sys/systm.h>
105 #include <sys/time.h>
106 
107 #include <net/if.h>
108 #include <net/if_types.h>
109 #include <net/raw_cb.h>
110 #include <net/vnet.h>
111 
112 #include <netinet/in.h>
113 #include <netinet/in_var.h>
114 #include <netinet/icmp6.h>
115 #include <netinet/ip_encap.h>
116 
117 #include <netinet/ip6.h>
118 #include <netinet6/ip6_var.h>
119 #include <netinet6/scope6_var.h>
120 #include <netinet6/nd6.h>
121 #include <netinet6/ip6_mroute.h>
122 #include <netinet6/ip6protosw.h>
123 #include <netinet6/pim6.h>
124 #include <netinet6/pim6_var.h>
125 
126 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
127 
128 /* XXX: this is a very common idiom; move to <sys/mbuf.h> ? */
129 #define M_HASCL(m) ((m)->m_flags & M_EXT)
130 
131 static int	ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
132 static void	phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
133 static void	pim6_init(void);
134 static int	register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
135 static int	set_pim6(int *);
136 static int	socket_send(struct socket *, struct mbuf *,
137 		    struct sockaddr_in6 *);
138 
139 extern int in6_mcast_loop;
140 extern struct domain inet6domain;
141 
142 static const struct encaptab *pim6_encap_cookie;
143 static const struct ip6protosw in6_pim_protosw = {
144 	.pr_type =		SOCK_RAW,
145 	.pr_domain =		&inet6domain,
146 	.pr_protocol =		IPPROTO_PIM,
147 	.pr_flags =		PR_ATOMIC|PR_ADDR|PR_LASTHDR,
148 	.pr_input =		pim6_input,
149 	.pr_output =		rip6_output,
150 	.pr_ctloutput =		rip6_ctloutput,
151 	.pr_init =		pim6_init,
152 	.pr_usrreqs =		&rip6_usrreqs
153 };
154 static int pim6_encapcheck(const struct mbuf *, int, int, void *);
155 
156 static VNET_DEFINE(int, ip6_mrouter_ver);
157 #define	V_ip6_mrouter_ver	VNET(ip6_mrouter_ver)
158 
159 SYSCTL_DECL(_net_inet6);
160 SYSCTL_DECL(_net_inet6_ip6);
161 SYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim, CTLFLAG_RW, 0, "PIM");
162 
163 static struct mrt6stat mrt6stat;
164 SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
165     &mrt6stat, mrt6stat,
166     "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
167 
168 #define NO_RTE_FOUND	0x1
169 #define RTE_FOUND	0x2
170 
171 static struct mtx mrouter6_mtx;
172 #define	MROUTER6_LOCK()		mtx_lock(&mrouter6_mtx)
173 #define	MROUTER6_UNLOCK()	mtx_unlock(&mrouter6_mtx)
174 #define	MROUTER6_LOCK_ASSERT()	do {					\
175 	mtx_assert(&mrouter6_mtx, MA_OWNED);				\
176 	NET_ASSERT_GIANT();						\
177 } while (0)
178 #define	MROUTER6_LOCK_INIT()	\
179 	mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
180 #define	MROUTER6_LOCK_DESTROY()	mtx_destroy(&mrouter6_mtx)
181 
182 static struct mf6c *mf6ctable[MF6CTBLSIZ];
183 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
184     &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
185     "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
186     "netinet6/ip6_mroute.h)");
187 
188 static struct mtx mfc6_mtx;
189 #define	MFC6_LOCK()		mtx_lock(&mfc6_mtx)
190 #define	MFC6_UNLOCK()		mtx_unlock(&mfc6_mtx)
191 #define	MFC6_LOCK_ASSERT()	do {					\
192 	mtx_assert(&mfc6_mtx, MA_OWNED);				\
193 	NET_ASSERT_GIANT();						\
194 } while (0)
195 #define	MFC6_LOCK_INIT()		\
196 	mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
197 #define	MFC6_LOCK_DESTROY()	mtx_destroy(&mfc6_mtx)
198 
199 static u_char n6expire[MF6CTBLSIZ];
200 
201 static struct mif6 mif6table[MAXMIFS];
202 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mif6table, CTLFLAG_RD,
203     &mif6table, sizeof(mif6table), "S,mif6[MAXMIFS]",
204     "IPv6 Multicast Interfaces (struct mif6[MAXMIFS], netinet6/ip6_mroute.h)");
205 
206 static struct mtx mif6_mtx;
207 #define	MIF6_LOCK()		mtx_lock(&mif6_mtx)
208 #define	MIF6_UNLOCK()		mtx_unlock(&mif6_mtx)
209 #define	MIF6_LOCK_ASSERT()	mtx_assert(&mif6_mtx, MA_OWNED)
210 #define	MIF6_LOCK_INIT()	\
211 	mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
212 #define	MIF6_LOCK_DESTROY()	mtx_destroy(&mif6_mtx)
213 
214 #ifdef MRT6DEBUG
215 static VNET_DEFINE(u_int, mrt6debug);		/* debug level */
216 #define	V_mrt6debug		VNET(mrt6debug)
217 #define DEBUG_MFC	0x02
218 #define DEBUG_FORWARD	0x04
219 #define DEBUG_EXPIRE	0x08
220 #define DEBUG_XMIT	0x10
221 #define DEBUG_REG	0x20
222 #define DEBUG_PIM	0x40
223 #endif
224 
225 static void	expire_upcalls(void *);
226 #define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
227 #define	UPCALL_EXPIRE	6		/* number of timeouts */
228 
229 /*
230  * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
231  */
232 
233 /*
234  * 'Interfaces' associated with decapsulator (so we can tell
235  * packets that went through it from ones that get reflected
236  * by a broken gateway).  Different from IPv4 register_if,
237  * these interfaces are linked into the system ifnet list,
238  * because per-interface IPv6 statistics are maintained in
239  * ifp->if_afdata.  But it does not have any routes point
240  * to them.  I.e., packets can't be sent this way.  They
241  * only exist as a placeholder for multicast source
242  * verification.
243  */
244 static struct ifnet *multicast_register_if6;
245 
246 #define ENCAP_HOPS 64
247 
248 /*
249  * Private variables.
250  */
251 static mifi_t nummifs = 0;
252 static mifi_t reg_mif_num = (mifi_t)-1;
253 
254 static struct pim6stat pim6stat;
255 SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RD,
256     &pim6stat, pim6stat,
257     "PIM Statistics (struct pim6stat, netinet6/pim_var.h)");
258 
259 static VNET_DEFINE(int, pim6);
260 #define	V_pim6		VNET(pim6)
261 
262 /*
263  * Hash function for a source, group entry
264  */
265 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
266 				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
267 				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
268 				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
269 
270 /*
271  * Find a route for a given origin IPv6 address and Multicast group address.
272  */
273 #define MF6CFIND(o, g, rt) do { \
274 	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
275 	rt = NULL; \
276 	mrt6stat.mrt6s_mfc_lookups++; \
277 	while (_rt) { \
278 		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
279 		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
280 		    (_rt->mf6c_stall == NULL)) { \
281 			rt = _rt; \
282 			break; \
283 		} \
284 		_rt = _rt->mf6c_next; \
285 	} \
286 	if (rt == NULL) { \
287 		mrt6stat.mrt6s_mfc_misses++; \
288 	} \
289 } while (/*CONSTCOND*/ 0)
290 
291 /*
292  * Macros to compute elapsed time efficiently
293  * Borrowed from Van Jacobson's scheduling code
294  * XXX: replace with timersub() ?
295  */
296 #define TV_DELTA(a, b, delta) do { \
297 	    int xxs; \
298 		\
299 	    delta = (a).tv_usec - (b).tv_usec; \
300 	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
301 	       switch (xxs) { \
302 		      case 2: \
303 			  delta += 1000000; \
304 			      /* FALLTHROUGH */ \
305 		      case 1: \
306 			  delta += 1000000; \
307 			  break; \
308 		      default: \
309 			  delta += (1000000 * xxs); \
310 	       } \
311 	    } \
312 } while (/*CONSTCOND*/ 0)
313 
314 /* XXX: replace with timercmp(a, b, <) ? */
315 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
316 	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
317 
318 #ifdef UPCALL_TIMING
319 #define UPCALL_MAX	50
320 static u_long upcall_data[UPCALL_MAX + 1];
321 static void collate();
322 #endif /* UPCALL_TIMING */
323 
324 static int ip6_mrouter_init(struct socket *, int, int);
325 static int add_m6fc(struct mf6cctl *);
326 static int add_m6if(struct mif6ctl *);
327 static int del_m6fc(struct mf6cctl *);
328 static int del_m6if(mifi_t *);
329 static int del_m6if_locked(mifi_t *);
330 static int get_mif6_cnt(struct sioc_mif_req6 *);
331 static int get_sg_cnt(struct sioc_sg_req6 *);
332 
333 static struct callout expire_upcalls_ch;
334 
335 int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
336 int X_ip6_mrouter_done(void);
337 int X_ip6_mrouter_set(struct socket *, struct sockopt *);
338 int X_ip6_mrouter_get(struct socket *, struct sockopt *);
339 int X_mrt6_ioctl(u_long, caddr_t);
340 
341 static void
342 pim6_init(void)
343 {
344 
345 #ifdef MRT6DEBUG
346 	V_mrt6debug = 0;	/* debug level */
347 #endif
348 }
349 
350 /*
351  * Handle MRT setsockopt commands to modify the multicast routing tables.
352  */
353 int
354 X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
355 {
356 	int error = 0;
357 	int optval;
358 	struct mif6ctl mifc;
359 	struct mf6cctl mfcc;
360 	mifi_t mifi;
361 
362 	if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
363 		return (EACCES);
364 
365 	switch (sopt->sopt_name) {
366 	case MRT6_INIT:
367 #ifdef MRT6_OINIT
368 	case MRT6_OINIT:
369 #endif
370 		error = sooptcopyin(sopt, &optval, sizeof(optval),
371 		    sizeof(optval));
372 		if (error)
373 			break;
374 		error = ip6_mrouter_init(so, optval, sopt->sopt_name);
375 		break;
376 	case MRT6_DONE:
377 		error = X_ip6_mrouter_done();
378 		break;
379 	case MRT6_ADD_MIF:
380 		error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
381 		if (error)
382 			break;
383 		error = add_m6if(&mifc);
384 		break;
385 	case MRT6_ADD_MFC:
386 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
387 		if (error)
388 			break;
389 		error = add_m6fc(&mfcc);
390 		break;
391 	case MRT6_DEL_MFC:
392 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
393 		if (error)
394 			break;
395 		error = del_m6fc(&mfcc);
396 		break;
397 	case MRT6_DEL_MIF:
398 		error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
399 		if (error)
400 			break;
401 		error = del_m6if(&mifi);
402 		break;
403 	case MRT6_PIM:
404 		error = sooptcopyin(sopt, &optval, sizeof(optval),
405 		    sizeof(optval));
406 		if (error)
407 			break;
408 		error = set_pim6(&optval);
409 		break;
410 	default:
411 		error = EOPNOTSUPP;
412 		break;
413 	}
414 
415 	return (error);
416 }
417 
418 /*
419  * Handle MRT getsockopt commands
420  */
421 int
422 X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
423 {
424 	int error = 0;
425 
426 	if (so != V_ip6_mrouter)
427 		return (EACCES);
428 
429 	switch (sopt->sopt_name) {
430 		case MRT6_PIM:
431 			error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
432 			break;
433 	}
434 	return (error);
435 }
436 
437 /*
438  * Handle ioctl commands to obtain information from the cache
439  */
440 int
441 X_mrt6_ioctl(u_long cmd, caddr_t data)
442 {
443 	int ret;
444 
445 	ret = EINVAL;
446 
447 	switch (cmd) {
448 	case SIOCGETSGCNT_IN6:
449 		ret = get_sg_cnt((struct sioc_sg_req6 *)data);
450 		break;
451 
452 	case SIOCGETMIFCNT_IN6:
453 		ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
454 		break;
455 
456 	default:
457 		break;
458 	}
459 
460 	return (ret);
461 }
462 
463 /*
464  * returns the packet, byte, rpf-failure count for the source group provided
465  */
466 static int
467 get_sg_cnt(struct sioc_sg_req6 *req)
468 {
469 	struct mf6c *rt;
470 	int ret;
471 
472 	ret = 0;
473 
474 	MFC6_LOCK();
475 
476 	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
477 	if (rt == NULL) {
478 		ret = ESRCH;
479 	} else {
480 		req->pktcnt = rt->mf6c_pkt_cnt;
481 		req->bytecnt = rt->mf6c_byte_cnt;
482 		req->wrong_if = rt->mf6c_wrong_if;
483 	}
484 
485 	MFC6_UNLOCK();
486 
487 	return (ret);
488 }
489 
490 /*
491  * returns the input and output packet and byte counts on the mif provided
492  */
493 static int
494 get_mif6_cnt(struct sioc_mif_req6 *req)
495 {
496 	mifi_t mifi;
497 	int ret;
498 
499 	ret = 0;
500 	mifi = req->mifi;
501 
502 	MIF6_LOCK();
503 
504 	if (mifi >= nummifs) {
505 		ret = EINVAL;
506 	} else {
507 		req->icount = mif6table[mifi].m6_pkt_in;
508 		req->ocount = mif6table[mifi].m6_pkt_out;
509 		req->ibytes = mif6table[mifi].m6_bytes_in;
510 		req->obytes = mif6table[mifi].m6_bytes_out;
511 	}
512 
513 	MIF6_UNLOCK();
514 
515 	return (ret);
516 }
517 
518 static int
519 set_pim6(int *i)
520 {
521 	if ((*i != 1) && (*i != 0))
522 		return (EINVAL);
523 
524 	V_pim6 = *i;
525 
526 	return (0);
527 }
528 
529 /*
530  * Enable multicast routing
531  */
532 static int
533 ip6_mrouter_init(struct socket *so, int v, int cmd)
534 {
535 
536 	V_ip6_mrouter_ver = 0;
537 
538 #ifdef MRT6DEBUG
539 	V_mrt6debug = 0;
540 
541 	if (V_mrt6debug)
542 		log(LOG_DEBUG,
543 		    "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
544 		    so->so_type, so->so_proto->pr_protocol);
545 #endif
546 
547 	if (so->so_type != SOCK_RAW ||
548 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
549 		return (EOPNOTSUPP);
550 
551 	if (v != 1)
552 		return (ENOPROTOOPT);
553 
554 	MROUTER6_LOCK();
555 
556 	if (V_ip6_mrouter != NULL) {
557 		MROUTER6_UNLOCK();
558 		return (EADDRINUSE);
559 	}
560 
561 	V_ip6_mrouter = so;
562 	V_ip6_mrouter_ver = cmd;
563 
564 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
565 	bzero((caddr_t)n6expire, sizeof(n6expire));
566 
567 	V_pim6 = 0;/* used for stubbing out/in pim stuff */
568 
569 	callout_init(&expire_upcalls_ch, 0);
570 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
571 	    expire_upcalls, NULL);
572 
573 	MROUTER6_UNLOCK();
574 
575 #ifdef MRT6DEBUG
576 	if (V_mrt6debug)
577 		log(LOG_DEBUG, "ip6_mrouter_init\n");
578 #endif
579 
580 	return (0);
581 }
582 
583 /*
584  * Disable IPv6 multicast forwarding.
585  */
586 int
587 X_ip6_mrouter_done(void)
588 {
589 	mifi_t mifi;
590 	int i;
591 	struct mf6c *rt;
592 	struct rtdetq *rte;
593 
594 	MROUTER6_LOCK();
595 
596 	if (V_ip6_mrouter == NULL) {
597 		MROUTER6_UNLOCK();
598 		return (EINVAL);
599 	}
600 
601 	/*
602 	 * For each phyint in use, disable promiscuous reception of all IPv6
603 	 * multicasts.
604 	 */
605 	for (mifi = 0; mifi < nummifs; mifi++) {
606 		if (mif6table[mifi].m6_ifp &&
607 		    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
608 			if_allmulti(mif6table[mifi].m6_ifp, 0);
609 		}
610 	}
611 	bzero((caddr_t)mif6table, sizeof(mif6table));
612 	nummifs = 0;
613 
614 	V_pim6 = 0; /* used to stub out/in pim specific code */
615 
616 	callout_stop(&expire_upcalls_ch);
617 
618 	/*
619 	 * Free all multicast forwarding cache entries.
620 	 */
621 	MFC6_LOCK();
622 	for (i = 0; i < MF6CTBLSIZ; i++) {
623 		rt = mf6ctable[i];
624 		while (rt) {
625 			struct mf6c *frt;
626 
627 			for (rte = rt->mf6c_stall; rte != NULL; ) {
628 				struct rtdetq *n = rte->next;
629 
630 				m_free(rte->m);
631 				free(rte, M_MRTABLE6);
632 				rte = n;
633 			}
634 			frt = rt;
635 			rt = rt->mf6c_next;
636 			free(frt, M_MRTABLE6);
637 		}
638 	}
639 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
640 	MFC6_UNLOCK();
641 
642 	/*
643 	 * Reset register interface
644 	 */
645 	if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
646 		if_detach(multicast_register_if6);
647 		if_free(multicast_register_if6);
648 		reg_mif_num = (mifi_t)-1;
649 		multicast_register_if6 = NULL;
650 	}
651 
652 	V_ip6_mrouter = NULL;
653 	V_ip6_mrouter_ver = 0;
654 
655 	MROUTER6_UNLOCK();
656 
657 #ifdef MRT6DEBUG
658 	if (V_mrt6debug)
659 		log(LOG_DEBUG, "ip6_mrouter_done\n");
660 #endif
661 
662 	return (0);
663 }
664 
665 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
666 
667 /*
668  * Add a mif to the mif table
669  */
670 static int
671 add_m6if(struct mif6ctl *mifcp)
672 {
673 	struct mif6 *mifp;
674 	struct ifnet *ifp;
675 	int error;
676 
677 	MIF6_LOCK();
678 
679 	if (mifcp->mif6c_mifi >= MAXMIFS) {
680 		MIF6_UNLOCK();
681 		return (EINVAL);
682 	}
683 	mifp = mif6table + mifcp->mif6c_mifi;
684 	if (mifp->m6_ifp != NULL) {
685 		MIF6_UNLOCK();
686 		return (EADDRINUSE); /* XXX: is it appropriate? */
687 	}
688 	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
689 		MIF6_UNLOCK();
690 		return (ENXIO);
691 	}
692 
693 	ifp = ifnet_byindex(mifcp->mif6c_pifi);
694 
695 	if (mifcp->mif6c_flags & MIFF_REGISTER) {
696 		if (reg_mif_num == (mifi_t)-1) {
697 			ifp = if_alloc(IFT_OTHER);
698 
699 			if_initname(ifp, "register_mif", 0);
700 			ifp->if_flags |= IFF_LOOPBACK;
701 			if_attach(ifp);
702 			multicast_register_if6 = ifp;
703 			reg_mif_num = mifcp->mif6c_mifi;
704 			/*
705 			 * it is impossible to guess the ifindex of the
706 			 * register interface.  So mif6c_pifi is automatically
707 			 * calculated.
708 			 */
709 			mifcp->mif6c_pifi = ifp->if_index;
710 		} else {
711 			ifp = multicast_register_if6;
712 		}
713 	} else {
714 		/* Make sure the interface supports multicast */
715 		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
716 			MIF6_UNLOCK();
717 			return (EOPNOTSUPP);
718 		}
719 
720 		error = if_allmulti(ifp, 1);
721 		if (error) {
722 			MIF6_UNLOCK();
723 			return (error);
724 		}
725 	}
726 
727 	mifp->m6_flags     = mifcp->mif6c_flags;
728 	mifp->m6_ifp       = ifp;
729 
730 	/* initialize per mif pkt counters */
731 	mifp->m6_pkt_in    = 0;
732 	mifp->m6_pkt_out   = 0;
733 	mifp->m6_bytes_in  = 0;
734 	mifp->m6_bytes_out = 0;
735 	bzero(&mifp->m6_route, sizeof(mifp->m6_route));
736 
737 	/* Adjust nummifs up if the mifi is higher than nummifs */
738 	if (nummifs <= mifcp->mif6c_mifi)
739 		nummifs = mifcp->mif6c_mifi + 1;
740 
741 	MIF6_UNLOCK();
742 
743 #ifdef MRT6DEBUG
744 	if (V_mrt6debug)
745 		log(LOG_DEBUG,
746 		    "add_mif #%d, phyint %s\n",
747 		    mifcp->mif6c_mifi,
748 		    ifp->if_xname);
749 #endif
750 
751 	return (0);
752 }
753 
754 /*
755  * Delete a mif from the mif table
756  */
757 static int
758 del_m6if_locked(mifi_t *mifip)
759 {
760 	struct mif6 *mifp = mif6table + *mifip;
761 	mifi_t mifi;
762 	struct ifnet *ifp;
763 
764 	MIF6_LOCK_ASSERT();
765 
766 	if (*mifip >= nummifs)
767 		return (EINVAL);
768 	if (mifp->m6_ifp == NULL)
769 		return (EINVAL);
770 
771 	if (!(mifp->m6_flags & MIFF_REGISTER)) {
772 		/* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
773 		ifp = mifp->m6_ifp;
774 		if_allmulti(ifp, 0);
775 	} else {
776 		if (reg_mif_num != (mifi_t)-1 &&
777 		    multicast_register_if6 != NULL) {
778 			if_detach(multicast_register_if6);
779 			if_free(multicast_register_if6);
780 			reg_mif_num = (mifi_t)-1;
781 			multicast_register_if6 = NULL;
782 		}
783 	}
784 
785 	bzero((caddr_t)mifp, sizeof(*mifp));
786 
787 	/* Adjust nummifs down */
788 	for (mifi = nummifs; mifi > 0; mifi--)
789 		if (mif6table[mifi - 1].m6_ifp)
790 			break;
791 	nummifs = mifi;
792 
793 #ifdef MRT6DEBUG
794 	if (V_mrt6debug)
795 		log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
796 #endif
797 
798 	return (0);
799 }
800 
801 static int
802 del_m6if(mifi_t *mifip)
803 {
804 	int cc;
805 
806 	MIF6_LOCK();
807 	cc = del_m6if_locked(mifip);
808 	MIF6_UNLOCK();
809 
810 	return (cc);
811 }
812 
813 /*
814  * Add an mfc entry
815  */
816 static int
817 add_m6fc(struct mf6cctl *mfccp)
818 {
819 	struct mf6c *rt;
820 	u_long hash;
821 	struct rtdetq *rte;
822 	u_short nstl;
823 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
824 
825 	MFC6_LOCK();
826 
827 	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
828 		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
829 
830 	/* If an entry already exists, just update the fields */
831 	if (rt) {
832 #ifdef MRT6DEBUG
833 		if (V_mrt6debug & DEBUG_MFC) {
834 		    log(LOG_DEBUG,
835 			"add_m6fc no upcall h %d o %s g %s p %x\n",
836 			ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
837 			ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
838 			mfccp->mf6cc_parent);
839 		}
840 #endif
841 
842 		rt->mf6c_parent = mfccp->mf6cc_parent;
843 		rt->mf6c_ifset = mfccp->mf6cc_ifset;
844 
845 		MFC6_UNLOCK();
846 		return (0);
847 	}
848 
849 	/*
850 	 * Find the entry for which the upcall was made and update
851 	 */
852 	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
853 			mfccp->mf6cc_mcastgrp.sin6_addr);
854 	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
855 		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
856 				       &mfccp->mf6cc_origin.sin6_addr) &&
857 		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
858 				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
859 		    (rt->mf6c_stall != NULL)) {
860 
861 			if (nstl++)
862 				log(LOG_ERR,
863 				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
864 				    "multiple kernel entries",
865 				    ip6_sprintf(ip6bufo,
866 					    &mfccp->mf6cc_origin.sin6_addr),
867 				    ip6_sprintf(ip6bufg,
868 					    &mfccp->mf6cc_mcastgrp.sin6_addr),
869 				    mfccp->mf6cc_parent, rt->mf6c_stall);
870 
871 #ifdef MRT6DEBUG
872 			if (V_mrt6debug & DEBUG_MFC)
873 				log(LOG_DEBUG,
874 				    "add_m6fc o %s g %s p %x dbg %x\n",
875 				    ip6_sprintf(ip6bufo,
876 					    &mfccp->mf6cc_origin.sin6_addr),
877 				    ip6_sprintf(ip6bufg,
878 					    &mfccp->mf6cc_mcastgrp.sin6_addr),
879 				    mfccp->mf6cc_parent, rt->mf6c_stall);
880 #endif
881 
882 			rt->mf6c_origin     = mfccp->mf6cc_origin;
883 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
884 			rt->mf6c_parent     = mfccp->mf6cc_parent;
885 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
886 			/* initialize pkt counters per src-grp */
887 			rt->mf6c_pkt_cnt    = 0;
888 			rt->mf6c_byte_cnt   = 0;
889 			rt->mf6c_wrong_if   = 0;
890 
891 			rt->mf6c_expire = 0;	/* Don't clean this guy up */
892 			n6expire[hash]--;
893 
894 			/* free packets Qed at the end of this entry */
895 			for (rte = rt->mf6c_stall; rte != NULL; ) {
896 				struct rtdetq *n = rte->next;
897 				ip6_mdq(rte->m, rte->ifp, rt);
898 				m_freem(rte->m);
899 #ifdef UPCALL_TIMING
900 				collate(&(rte->t));
901 #endif /* UPCALL_TIMING */
902 				free(rte, M_MRTABLE6);
903 				rte = n;
904 			}
905 			rt->mf6c_stall = NULL;
906 		}
907 	}
908 
909 	/*
910 	 * It is possible that an entry is being inserted without an upcall
911 	 */
912 	if (nstl == 0) {
913 #ifdef MRT6DEBUG
914 		if (V_mrt6debug & DEBUG_MFC)
915 		    log(LOG_DEBUG,
916 			"add_mfc no upcall h %d o %s g %s p %x\n",
917 			hash,
918 			ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
919 			ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
920 			mfccp->mf6cc_parent);
921 #endif
922 
923 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
924 
925 			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
926 					       &mfccp->mf6cc_origin.sin6_addr)&&
927 			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
928 					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
929 
930 				rt->mf6c_origin     = mfccp->mf6cc_origin;
931 				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
932 				rt->mf6c_parent     = mfccp->mf6cc_parent;
933 				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
934 				/* initialize pkt counters per src-grp */
935 				rt->mf6c_pkt_cnt    = 0;
936 				rt->mf6c_byte_cnt   = 0;
937 				rt->mf6c_wrong_if   = 0;
938 
939 				if (rt->mf6c_expire)
940 					n6expire[hash]--;
941 				rt->mf6c_expire	   = 0;
942 			}
943 		}
944 		if (rt == NULL) {
945 			/* no upcall, so make a new entry */
946 			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
947 						  M_NOWAIT);
948 			if (rt == NULL) {
949 				MFC6_UNLOCK();
950 				return (ENOBUFS);
951 			}
952 
953 			/* insert new entry at head of hash chain */
954 			rt->mf6c_origin     = mfccp->mf6cc_origin;
955 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
956 			rt->mf6c_parent     = mfccp->mf6cc_parent;
957 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
958 			/* initialize pkt counters per src-grp */
959 			rt->mf6c_pkt_cnt    = 0;
960 			rt->mf6c_byte_cnt   = 0;
961 			rt->mf6c_wrong_if   = 0;
962 			rt->mf6c_expire     = 0;
963 			rt->mf6c_stall = NULL;
964 
965 			/* link into table */
966 			rt->mf6c_next  = mf6ctable[hash];
967 			mf6ctable[hash] = rt;
968 		}
969 	}
970 
971 	MFC6_UNLOCK();
972 	return (0);
973 }
974 
975 #ifdef UPCALL_TIMING
976 /*
977  * collect delay statistics on the upcalls
978  */
979 static void
980 collate(struct timeval *t)
981 {
982 	u_long d;
983 	struct timeval tp;
984 	u_long delta;
985 
986 	GET_TIME(tp);
987 
988 	if (TV_LT(*t, tp))
989 	{
990 		TV_DELTA(tp, *t, delta);
991 
992 		d = delta >> 10;
993 		if (d > UPCALL_MAX)
994 			d = UPCALL_MAX;
995 
996 		++upcall_data[d];
997 	}
998 }
999 #endif /* UPCALL_TIMING */
1000 
1001 /*
1002  * Delete an mfc entry
1003  */
1004 static int
1005 del_m6fc(struct mf6cctl *mfccp)
1006 {
1007 	struct sockaddr_in6	origin;
1008 	struct sockaddr_in6	mcastgrp;
1009 	struct mf6c		*rt;
1010 	struct mf6c		**nptr;
1011 	u_long		hash;
1012 
1013 	origin = mfccp->mf6cc_origin;
1014 	mcastgrp = mfccp->mf6cc_mcastgrp;
1015 	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
1016 
1017 #ifdef MRT6DEBUG
1018 	if (V_mrt6debug & DEBUG_MFC) {
1019 		char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1020 		log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
1021 		    ip6_sprintf(ip6bufo, &origin.sin6_addr),
1022 		    ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
1023 	}
1024 #endif
1025 
1026 	MFC6_LOCK();
1027 
1028 	nptr = &mf6ctable[hash];
1029 	while ((rt = *nptr) != NULL) {
1030 		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
1031 				       &rt->mf6c_origin.sin6_addr) &&
1032 		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1033 				       &rt->mf6c_mcastgrp.sin6_addr) &&
1034 		    rt->mf6c_stall == NULL)
1035 			break;
1036 
1037 		nptr = &rt->mf6c_next;
1038 	}
1039 	if (rt == NULL) {
1040 		MFC6_UNLOCK();
1041 		return (EADDRNOTAVAIL);
1042 	}
1043 
1044 	*nptr = rt->mf6c_next;
1045 	free(rt, M_MRTABLE6);
1046 
1047 	MFC6_UNLOCK();
1048 
1049 	return (0);
1050 }
1051 
1052 static int
1053 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1054 {
1055 
1056 	if (s) {
1057 		if (sbappendaddr(&s->so_rcv,
1058 				 (struct sockaddr *)src,
1059 				 mm, (struct mbuf *)0) != 0) {
1060 			sorwakeup(s);
1061 			return (0);
1062 		}
1063 	}
1064 	m_freem(mm);
1065 	return (-1);
1066 }
1067 
1068 /*
1069  * IPv6 multicast forwarding function. This function assumes that the packet
1070  * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1071  * pointed to by "ifp", and the packet is to be relayed to other networks
1072  * that have members of the packet's destination IPv6 multicast group.
1073  *
1074  * The packet is returned unscathed to the caller, unless it is
1075  * erroneous, in which case a non-zero return value tells the caller to
1076  * discard it.
1077  *
1078  * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1079  * this function is called in the originating context (i.e., not when
1080  * forwarding a packet from other node).  ip6_output(), which is currently the
1081  * only function that calls this function is called in the originating context,
1082  * explicitly ensures this condition.  It is caller's responsibility to ensure
1083  * that if this function is called from somewhere else in the originating
1084  * context in the future.
1085  */
1086 int
1087 X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1088 {
1089 	struct mf6c *rt;
1090 	struct mif6 *mifp;
1091 	struct mbuf *mm;
1092 	mifi_t mifi;
1093 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1094 
1095 #ifdef MRT6DEBUG
1096 	if (V_mrt6debug & DEBUG_FORWARD)
1097 		log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1098 		    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1099 		    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1100 		    ifp->if_index);
1101 #endif
1102 
1103 	/*
1104 	 * Don't forward a packet with Hop limit of zero or one,
1105 	 * or a packet destined to a local-only group.
1106 	 */
1107 	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1108 	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1109 		return (0);
1110 	ip6->ip6_hlim--;
1111 
1112 	/*
1113 	 * Source address check: do not forward packets with unspecified
1114 	 * source. It was discussed in July 2000, on ipngwg mailing list.
1115 	 * This is rather more serious than unicast cases, because some
1116 	 * MLD packets can be sent with the unspecified source address
1117 	 * (although such packets must normally set 1 to the hop limit field).
1118 	 */
1119 	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1120 		V_ip6stat.ip6s_cantforward++;
1121 		if (V_ip6_log_time + V_ip6_log_interval < time_second) {
1122 			V_ip6_log_time = time_second;
1123 			log(LOG_DEBUG,
1124 			    "cannot forward "
1125 			    "from %s to %s nxt %d received on %s\n",
1126 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1127 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1128 			    ip6->ip6_nxt,
1129 			    if_name(m->m_pkthdr.rcvif));
1130 		}
1131 		return (0);
1132 	}
1133 
1134 	MFC6_LOCK();
1135 
1136 	/*
1137 	 * Determine forwarding mifs from the forwarding cache table
1138 	 */
1139 	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1140 
1141 	/* Entry exists, so forward if necessary */
1142 	if (rt) {
1143 		MFC6_UNLOCK();
1144 		return (ip6_mdq(m, ifp, rt));
1145 	} else {
1146 		/*
1147 		 * If we don't have a route for packet's origin,
1148 		 * Make a copy of the packet &
1149 		 * send message to routing daemon
1150 		 */
1151 
1152 		struct mbuf *mb0;
1153 		struct rtdetq *rte;
1154 		u_long hash;
1155 /*		int i, npkts;*/
1156 #ifdef UPCALL_TIMING
1157 		struct timeval tp;
1158 
1159 		GET_TIME(tp);
1160 #endif /* UPCALL_TIMING */
1161 
1162 		mrt6stat.mrt6s_no_route++;
1163 #ifdef MRT6DEBUG
1164 		if (V_mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1165 			log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1166 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1167 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1168 #endif
1169 
1170 		/*
1171 		 * Allocate mbufs early so that we don't do extra work if we
1172 		 * are just going to fail anyway.
1173 		 */
1174 		rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6,
1175 					      M_NOWAIT);
1176 		if (rte == NULL) {
1177 			MFC6_UNLOCK();
1178 			return (ENOBUFS);
1179 		}
1180 		mb0 = m_copy(m, 0, M_COPYALL);
1181 		/*
1182 		 * Pullup packet header if needed before storing it,
1183 		 * as other references may modify it in the meantime.
1184 		 */
1185 		if (mb0 &&
1186 		    (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1187 			mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1188 		if (mb0 == NULL) {
1189 			free(rte, M_MRTABLE6);
1190 			MFC6_UNLOCK();
1191 			return (ENOBUFS);
1192 		}
1193 
1194 		/* is there an upcall waiting for this packet? */
1195 		hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1196 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1197 			if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1198 					       &rt->mf6c_origin.sin6_addr) &&
1199 			    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1200 					       &rt->mf6c_mcastgrp.sin6_addr) &&
1201 			    (rt->mf6c_stall != NULL))
1202 				break;
1203 		}
1204 
1205 		if (rt == NULL) {
1206 			struct mrt6msg *im;
1207 #ifdef MRT6_OINIT
1208 			struct omrt6msg *oim;
1209 #endif
1210 
1211 			/* no upcall, so make a new entry */
1212 			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
1213 						  M_NOWAIT);
1214 			if (rt == NULL) {
1215 				free(rte, M_MRTABLE6);
1216 				m_freem(mb0);
1217 				MFC6_UNLOCK();
1218 				return (ENOBUFS);
1219 			}
1220 			/*
1221 			 * Make a copy of the header to send to the user
1222 			 * level process
1223 			 */
1224 			mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1225 
1226 			if (mm == NULL) {
1227 				free(rte, M_MRTABLE6);
1228 				m_freem(mb0);
1229 				free(rt, M_MRTABLE6);
1230 				MFC6_UNLOCK();
1231 				return (ENOBUFS);
1232 			}
1233 
1234 			/*
1235 			 * Send message to routing daemon
1236 			 */
1237 			sin6.sin6_addr = ip6->ip6_src;
1238 
1239 			im = NULL;
1240 #ifdef MRT6_OINIT
1241 			oim = NULL;
1242 #endif
1243 			switch (V_ip6_mrouter_ver) {
1244 #ifdef MRT6_OINIT
1245 			case MRT6_OINIT:
1246 				oim = mtod(mm, struct omrt6msg *);
1247 				oim->im6_msgtype = MRT6MSG_NOCACHE;
1248 				oim->im6_mbz = 0;
1249 				break;
1250 #endif
1251 			case MRT6_INIT:
1252 				im = mtod(mm, struct mrt6msg *);
1253 				im->im6_msgtype = MRT6MSG_NOCACHE;
1254 				im->im6_mbz = 0;
1255 				break;
1256 			default:
1257 				free(rte, M_MRTABLE6);
1258 				m_freem(mb0);
1259 				free(rt, M_MRTABLE6);
1260 				MFC6_UNLOCK();
1261 				return (EINVAL);
1262 			}
1263 
1264 #ifdef MRT6DEBUG
1265 			if (V_mrt6debug & DEBUG_FORWARD)
1266 				log(LOG_DEBUG,
1267 				    "getting the iif info in the kernel\n");
1268 #endif
1269 
1270 			for (mifp = mif6table, mifi = 0;
1271 			     mifi < nummifs && mifp->m6_ifp != ifp;
1272 			     mifp++, mifi++)
1273 				;
1274 
1275 			switch (V_ip6_mrouter_ver) {
1276 #ifdef MRT6_OINIT
1277 			case MRT6_OINIT:
1278 				oim->im6_mif = mifi;
1279 				break;
1280 #endif
1281 			case MRT6_INIT:
1282 				im->im6_mif = mifi;
1283 				break;
1284 			}
1285 
1286 			if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1287 				log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1288 				    "socket queue full\n");
1289 				mrt6stat.mrt6s_upq_sockfull++;
1290 				free(rte, M_MRTABLE6);
1291 				m_freem(mb0);
1292 				free(rt, M_MRTABLE6);
1293 				MFC6_UNLOCK();
1294 				return (ENOBUFS);
1295 			}
1296 
1297 			mrt6stat.mrt6s_upcalls++;
1298 
1299 			/* insert new entry at head of hash chain */
1300 			bzero(rt, sizeof(*rt));
1301 			rt->mf6c_origin.sin6_family = AF_INET6;
1302 			rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1303 			rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1304 			rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1305 			rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1306 			rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1307 			rt->mf6c_expire = UPCALL_EXPIRE;
1308 			n6expire[hash]++;
1309 			rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1310 
1311 			/* link into table */
1312 			rt->mf6c_next  = mf6ctable[hash];
1313 			mf6ctable[hash] = rt;
1314 			/* Add this entry to the end of the queue */
1315 			rt->mf6c_stall = rte;
1316 		} else {
1317 			/* determine if q has overflowed */
1318 			struct rtdetq **p;
1319 			int npkts = 0;
1320 
1321 			for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1322 				if (++npkts > MAX_UPQ6) {
1323 					mrt6stat.mrt6s_upq_ovflw++;
1324 					free(rte, M_MRTABLE6);
1325 					m_freem(mb0);
1326 					MFC6_UNLOCK();
1327 					return (0);
1328 				}
1329 
1330 			/* Add this entry to the end of the queue */
1331 			*p = rte;
1332 		}
1333 
1334 		rte->next = NULL;
1335 		rte->m = mb0;
1336 		rte->ifp = ifp;
1337 #ifdef UPCALL_TIMING
1338 		rte->t = tp;
1339 #endif /* UPCALL_TIMING */
1340 
1341 		MFC6_UNLOCK();
1342 
1343 		return (0);
1344 	}
1345 }
1346 
1347 /*
1348  * Clean up cache entries if upcalls are not serviced
1349  * Call from the Slow Timeout mechanism, every half second.
1350  */
1351 static void
1352 expire_upcalls(void *unused)
1353 {
1354 	struct rtdetq *rte;
1355 	struct mf6c *mfc, **nptr;
1356 	int i;
1357 
1358 	MFC6_LOCK();
1359 	for (i = 0; i < MF6CTBLSIZ; i++) {
1360 		if (n6expire[i] == 0)
1361 			continue;
1362 		nptr = &mf6ctable[i];
1363 		while ((mfc = *nptr) != NULL) {
1364 			rte = mfc->mf6c_stall;
1365 			/*
1366 			 * Skip real cache entries
1367 			 * Make sure it wasn't marked to not expire (shouldn't happen)
1368 			 * If it expires now
1369 			 */
1370 			if (rte != NULL &&
1371 			    mfc->mf6c_expire != 0 &&
1372 			    --mfc->mf6c_expire == 0) {
1373 #ifdef MRT6DEBUG
1374 				if (V_mrt6debug & DEBUG_EXPIRE) {
1375 					char ip6bufo[INET6_ADDRSTRLEN];
1376 					char ip6bufg[INET6_ADDRSTRLEN];
1377 					log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1378 					    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1379 					    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1380 				}
1381 #endif
1382 				/*
1383 				 * drop all the packets
1384 				 * free the mbuf with the pkt, if, timing info
1385 				 */
1386 				do {
1387 					struct rtdetq *n = rte->next;
1388 					m_freem(rte->m);
1389 					free(rte, M_MRTABLE6);
1390 					rte = n;
1391 				} while (rte != NULL);
1392 				mrt6stat.mrt6s_cache_cleanups++;
1393 				n6expire[i]--;
1394 
1395 				*nptr = mfc->mf6c_next;
1396 				free(mfc, M_MRTABLE6);
1397 			} else {
1398 				nptr = &mfc->mf6c_next;
1399 			}
1400 		}
1401 	}
1402 	MFC6_UNLOCK();
1403 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1404 	    expire_upcalls, NULL);
1405 }
1406 
1407 /*
1408  * Packet forwarding routine once entry in the cache is made
1409  */
1410 static int
1411 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1412 {
1413 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1414 	mifi_t mifi, iif;
1415 	struct mif6 *mifp;
1416 	int plen = m->m_pkthdr.len;
1417 	struct in6_addr src0, dst0; /* copies for local work */
1418 	u_int32_t iszone, idzone, oszone, odzone;
1419 	int error = 0;
1420 
1421 /*
1422  * Macro to send packet on mif.  Since RSVP packets don't get counted on
1423  * input, they shouldn't get counted on output, so statistics keeping is
1424  * separate.
1425  */
1426 
1427 #define MC6_SEND(ip6, mifp, m) do {				\
1428 	if ((mifp)->m6_flags & MIFF_REGISTER)			\
1429 		register_send((ip6), (mifp), (m));		\
1430 	else							\
1431 		phyint_send((ip6), (mifp), (m));		\
1432 } while (/*CONSTCOND*/ 0)
1433 
1434 	/*
1435 	 * Don't forward if it didn't arrive from the parent mif
1436 	 * for its origin.
1437 	 */
1438 	mifi = rt->mf6c_parent;
1439 	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1440 		/* came in the wrong interface */
1441 #ifdef MRT6DEBUG
1442 		if (V_mrt6debug & DEBUG_FORWARD)
1443 			log(LOG_DEBUG,
1444 			    "wrong if: ifid %d mifi %d mififid %x\n",
1445 			    ifp->if_index, mifi,
1446 			    mif6table[mifi].m6_ifp->if_index);
1447 #endif
1448 		mrt6stat.mrt6s_wrong_if++;
1449 		rt->mf6c_wrong_if++;
1450 		/*
1451 		 * If we are doing PIM processing, and we are forwarding
1452 		 * packets on this interface, send a message to the
1453 		 * routing daemon.
1454 		 */
1455 		/* have to make sure this is a valid mif */
1456 		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1457 			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1458 				/*
1459 				 * Check the M_LOOP flag to avoid an
1460 				 * unnecessary PIM assert.
1461 				 * XXX: M_LOOP is an ad-hoc hack...
1462 				 */
1463 				static struct sockaddr_in6 sin6 =
1464 				{ sizeof(sin6), AF_INET6 };
1465 
1466 				struct mbuf *mm;
1467 				struct mrt6msg *im;
1468 #ifdef MRT6_OINIT
1469 				struct omrt6msg *oim;
1470 #endif
1471 
1472 				mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1473 				if (mm &&
1474 				    (M_HASCL(mm) ||
1475 				     mm->m_len < sizeof(struct ip6_hdr)))
1476 					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1477 				if (mm == NULL)
1478 					return (ENOBUFS);
1479 
1480 #ifdef MRT6_OINIT
1481 				oim = NULL;
1482 #endif
1483 				im = NULL;
1484 				switch (V_ip6_mrouter_ver) {
1485 #ifdef MRT6_OINIT
1486 				case MRT6_OINIT:
1487 					oim = mtod(mm, struct omrt6msg *);
1488 					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1489 					oim->im6_mbz = 0;
1490 					break;
1491 #endif
1492 				case MRT6_INIT:
1493 					im = mtod(mm, struct mrt6msg *);
1494 					im->im6_msgtype = MRT6MSG_WRONGMIF;
1495 					im->im6_mbz = 0;
1496 					break;
1497 				default:
1498 					m_freem(mm);
1499 					return (EINVAL);
1500 				}
1501 
1502 				for (mifp = mif6table, iif = 0;
1503 				     iif < nummifs && mifp &&
1504 					     mifp->m6_ifp != ifp;
1505 				     mifp++, iif++)
1506 					;
1507 
1508 				switch (V_ip6_mrouter_ver) {
1509 #ifdef MRT6_OINIT
1510 				case MRT6_OINIT:
1511 					oim->im6_mif = iif;
1512 					sin6.sin6_addr = oim->im6_src;
1513 					break;
1514 #endif
1515 				case MRT6_INIT:
1516 					im->im6_mif = iif;
1517 					sin6.sin6_addr = im->im6_src;
1518 					break;
1519 				}
1520 
1521 				mrt6stat.mrt6s_upcalls++;
1522 
1523 				if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1524 #ifdef MRT6DEBUG
1525 					if (V_mrt6debug)
1526 						log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1527 #endif
1528 					++mrt6stat.mrt6s_upq_sockfull;
1529 					return (ENOBUFS);
1530 				}	/* if socket Q full */
1531 			}		/* if PIM */
1532 		return (0);
1533 	}			/* if wrong iif */
1534 
1535 	/* If I sourced this packet, it counts as output, else it was input. */
1536 	if (m->m_pkthdr.rcvif == NULL) {
1537 		/* XXX: is rcvif really NULL when output?? */
1538 		mif6table[mifi].m6_pkt_out++;
1539 		mif6table[mifi].m6_bytes_out += plen;
1540 	} else {
1541 		mif6table[mifi].m6_pkt_in++;
1542 		mif6table[mifi].m6_bytes_in += plen;
1543 	}
1544 	rt->mf6c_pkt_cnt++;
1545 	rt->mf6c_byte_cnt += plen;
1546 
1547 	/*
1548 	 * For each mif, forward a copy of the packet if there are group
1549 	 * members downstream on the interface.
1550 	 */
1551 	src0 = ip6->ip6_src;
1552 	dst0 = ip6->ip6_dst;
1553 	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1554 	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1555 		V_ip6stat.ip6s_badscope++;
1556 		return (error);
1557 	}
1558 	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1559 		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1560 			/*
1561 			 * check if the outgoing packet is going to break
1562 			 * a scope boundary.
1563 			 * XXX For packets through PIM register tunnel
1564 			 * interface, we believe a routing daemon.
1565 			 */
1566 			if (!(mif6table[rt->mf6c_parent].m6_flags &
1567 			      MIFF_REGISTER) &&
1568 			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1569 				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1570 				    &oszone) ||
1571 				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1572 				    &odzone) ||
1573 				    iszone != oszone ||
1574 				    idzone != odzone) {
1575 					V_ip6stat.ip6s_badscope++;
1576 					continue;
1577 				}
1578 			}
1579 
1580 			mifp->m6_pkt_out++;
1581 			mifp->m6_bytes_out += plen;
1582 			MC6_SEND(ip6, mifp, m);
1583 		}
1584 	}
1585 	return (0);
1586 }
1587 
1588 static void
1589 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1590 {
1591 	struct mbuf *mb_copy;
1592 	struct ifnet *ifp = mifp->m6_ifp;
1593 	int error = 0;
1594 	struct sockaddr_in6 *dst6;
1595 	u_long linkmtu;
1596 
1597 	dst6 = &mifp->m6_route.ro_dst;
1598 
1599 	/*
1600 	 * Make a new reference to the packet; make sure that
1601 	 * the IPv6 header is actually copied, not just referenced,
1602 	 * so that ip6_output() only scribbles on the copy.
1603 	 */
1604 	mb_copy = m_copy(m, 0, M_COPYALL);
1605 	if (mb_copy &&
1606 	    (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1607 		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1608 	if (mb_copy == NULL) {
1609 		return;
1610 	}
1611 	/* set MCAST flag to the outgoing packet */
1612 	mb_copy->m_flags |= M_MCAST;
1613 
1614 	/*
1615 	 * If we sourced the packet, call ip6_output since we may devide
1616 	 * the packet into fragments when the packet is too big for the
1617 	 * outgoing interface.
1618 	 * Otherwise, we can simply send the packet to the interface
1619 	 * sending queue.
1620 	 */
1621 	if (m->m_pkthdr.rcvif == NULL) {
1622 		struct ip6_moptions im6o;
1623 
1624 		im6o.im6o_multicast_ifp = ifp;
1625 		/* XXX: ip6_output will override ip6->ip6_hlim */
1626 		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1627 		im6o.im6o_multicast_loop = 1;
1628 		error = ip6_output(mb_copy, NULL, &mifp->m6_route,
1629 				   IPV6_FORWARDING, &im6o, NULL, NULL);
1630 
1631 #ifdef MRT6DEBUG
1632 		if (V_mrt6debug & DEBUG_XMIT)
1633 			log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1634 			    mifp - mif6table, error);
1635 #endif
1636 		return;
1637 	}
1638 
1639 	/*
1640 	 * If configured to loop back multicasts by default,
1641 	 * loop back a copy now.
1642 	 */
1643 	if (in6_mcast_loop) {
1644 		dst6->sin6_len = sizeof(struct sockaddr_in6);
1645 		dst6->sin6_family = AF_INET6;
1646 		dst6->sin6_addr = ip6->ip6_dst;
1647 		ip6_mloopback(ifp, m, &mifp->m6_route.ro_dst);
1648 	}
1649 
1650 	/*
1651 	 * Put the packet into the sending queue of the outgoing interface
1652 	 * if it would fit in the MTU of the interface.
1653 	 */
1654 	linkmtu = IN6_LINKMTU(ifp);
1655 	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1656 		dst6->sin6_len = sizeof(struct sockaddr_in6);
1657 		dst6->sin6_family = AF_INET6;
1658 		dst6->sin6_addr = ip6->ip6_dst;
1659 		/*
1660 		 * We just call if_output instead of nd6_output here, since
1661 		 * we need no ND for a multicast forwarded packet...right?
1662 		 */
1663 		error = (*ifp->if_output)(ifp, mb_copy,
1664 		    (struct sockaddr *)&mifp->m6_route.ro_dst, NULL);
1665 #ifdef MRT6DEBUG
1666 		if (V_mrt6debug & DEBUG_XMIT)
1667 			log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1668 			    mifp - mif6table, error);
1669 #endif
1670 	} else {
1671 		/*
1672 		 * pMTU discovery is intentionally disabled by default, since
1673 		 * various router may notify pMTU in multicast, which can be
1674 		 * a DDoS to a router
1675 		 */
1676 		if (V_ip6_mcast_pmtu)
1677 			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1678 		else {
1679 #ifdef MRT6DEBUG
1680 			if (V_mrt6debug & DEBUG_XMIT) {
1681 				char ip6bufs[INET6_ADDRSTRLEN];
1682 				char ip6bufd[INET6_ADDRSTRLEN];
1683 				log(LOG_DEBUG,
1684 				    "phyint_send: packet too big on %s o %s "
1685 				    "g %s size %d(discarded)\n",
1686 				    if_name(ifp),
1687 				    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1688 				    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1689 				    mb_copy->m_pkthdr.len);
1690 			}
1691 #endif /* MRT6DEBUG */
1692 			m_freem(mb_copy); /* simply discard the packet */
1693 		}
1694 	}
1695 }
1696 
1697 static int
1698 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1699 {
1700 	struct mbuf *mm;
1701 	int i, len = m->m_pkthdr.len;
1702 	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1703 	struct mrt6msg *im6;
1704 
1705 #ifdef MRT6DEBUG
1706 	if (V_mrt6debug) {
1707 		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1708 		log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1709 		    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1710 		    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1711 	}
1712 #endif
1713 	++pim6stat.pim6s_snd_registers;
1714 
1715 	/* Make a copy of the packet to send to the user level process */
1716 	MGETHDR(mm, M_DONTWAIT, MT_HEADER);
1717 	if (mm == NULL)
1718 		return (ENOBUFS);
1719 	mm->m_pkthdr.rcvif = NULL;
1720 	mm->m_data += max_linkhdr;
1721 	mm->m_len = sizeof(struct ip6_hdr);
1722 
1723 	if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1724 		m_freem(mm);
1725 		return (ENOBUFS);
1726 	}
1727 	i = MHLEN - M_LEADINGSPACE(mm);
1728 	if (i > len)
1729 		i = len;
1730 	mm = m_pullup(mm, i);
1731 	if (mm == NULL)
1732 		return (ENOBUFS);
1733 /* TODO: check it! */
1734 	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1735 
1736 	/*
1737 	 * Send message to routing daemon
1738 	 */
1739 	sin6.sin6_addr = ip6->ip6_src;
1740 
1741 	im6 = mtod(mm, struct mrt6msg *);
1742 	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1743 	im6->im6_mbz          = 0;
1744 
1745 	im6->im6_mif = mif - mif6table;
1746 
1747 	/* iif info is not given for reg. encap.n */
1748 	mrt6stat.mrt6s_upcalls++;
1749 
1750 	if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1751 #ifdef MRT6DEBUG
1752 		if (V_mrt6debug)
1753 			log(LOG_WARNING,
1754 			    "register_send: ip6_mrouter socket queue full\n");
1755 #endif
1756 		++mrt6stat.mrt6s_upq_sockfull;
1757 		return (ENOBUFS);
1758 	}
1759 	return (0);
1760 }
1761 
1762 /*
1763  * pim6_encapcheck() is called by the encap6_input() path at runtime to
1764  * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1765  * into the kernel.
1766  */
1767 static int
1768 pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
1769 {
1770 
1771 #ifdef DIAGNOSTIC
1772     KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1773 #endif
1774     if (proto != IPPROTO_PIM)
1775 	return 0;	/* not for us; reject the datagram. */
1776 
1777     return 64;		/* claim the datagram. */
1778 }
1779 
1780 /*
1781  * PIM sparse mode hook
1782  * Receives the pim control messages, and passes them up to the listening
1783  * socket, using rip6_input.
1784  * The only message processed is the REGISTER pim message; the pim header
1785  * is stripped off, and the inner packet is passed to register_mforward.
1786  */
1787 int
1788 pim6_input(struct mbuf **mp, int *offp, int proto)
1789 {
1790 	struct pim *pim; /* pointer to a pim struct */
1791 	struct ip6_hdr *ip6;
1792 	int pimlen;
1793 	struct mbuf *m = *mp;
1794 	int minlen;
1795 	int off = *offp;
1796 
1797 	++pim6stat.pim6s_rcv_total;
1798 
1799 	ip6 = mtod(m, struct ip6_hdr *);
1800 	pimlen = m->m_pkthdr.len - *offp;
1801 
1802 	/*
1803 	 * Validate lengths
1804 	 */
1805 	if (pimlen < PIM_MINLEN) {
1806 		++pim6stat.pim6s_rcv_tooshort;
1807 #ifdef MRT6DEBUG
1808 		if (V_mrt6debug & DEBUG_PIM)
1809 			log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1810 #endif
1811 		m_freem(m);
1812 		return (IPPROTO_DONE);
1813 	}
1814 
1815 	/*
1816 	 * if the packet is at least as big as a REGISTER, go ahead
1817 	 * and grab the PIM REGISTER header size, to avoid another
1818 	 * possible m_pullup() later.
1819 	 *
1820 	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1821 	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1822 	 */
1823 	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1824 
1825 	/*
1826 	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1827 	 * possibly the PIM REGISTER header
1828 	 */
1829 #ifndef PULLDOWN_TEST
1830 	IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1831 	/* adjust pointer */
1832 	ip6 = mtod(m, struct ip6_hdr *);
1833 
1834 	/* adjust mbuf to point to the PIM header */
1835 	pim = (struct pim *)((caddr_t)ip6 + off);
1836 #else
1837 	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1838 	if (pim == NULL) {
1839 		pim6stat.pim6s_rcv_tooshort++;
1840 		return (IPPROTO_DONE);
1841 	}
1842 #endif
1843 
1844 #define PIM6_CHECKSUM
1845 #ifdef PIM6_CHECKSUM
1846 	{
1847 		int cksumlen;
1848 
1849 		/*
1850 		 * Validate checksum.
1851 		 * If PIM REGISTER, exclude the data packet
1852 		 */
1853 		if (pim->pim_type == PIM_REGISTER)
1854 			cksumlen = PIM_MINLEN;
1855 		else
1856 			cksumlen = pimlen;
1857 
1858 		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1859 			++pim6stat.pim6s_rcv_badsum;
1860 #ifdef MRT6DEBUG
1861 			if (V_mrt6debug & DEBUG_PIM)
1862 				log(LOG_DEBUG,
1863 				    "pim6_input: invalid checksum\n");
1864 #endif
1865 			m_freem(m);
1866 			return (IPPROTO_DONE);
1867 		}
1868 	}
1869 #endif /* PIM_CHECKSUM */
1870 
1871 	/* PIM version check */
1872 	if (pim->pim_ver != PIM_VERSION) {
1873 		++pim6stat.pim6s_rcv_badversion;
1874 #ifdef MRT6DEBUG
1875 		log(LOG_ERR,
1876 		    "pim6_input: incorrect version %d, expecting %d\n",
1877 		    pim->pim_ver, PIM_VERSION);
1878 #endif
1879 		m_freem(m);
1880 		return (IPPROTO_DONE);
1881 	}
1882 
1883 	if (pim->pim_type == PIM_REGISTER) {
1884 		/*
1885 		 * since this is a REGISTER, we'll make a copy of the register
1886 		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1887 		 * routing daemon.
1888 		 */
1889 		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1890 
1891 		struct mbuf *mcp;
1892 		struct ip6_hdr *eip6;
1893 		u_int32_t *reghdr;
1894 		int rc;
1895 #ifdef MRT6DEBUG
1896 		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1897 #endif
1898 
1899 		++pim6stat.pim6s_rcv_registers;
1900 
1901 		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1902 #ifdef MRT6DEBUG
1903 			if (V_mrt6debug & DEBUG_PIM)
1904 				log(LOG_DEBUG,
1905 				    "pim6_input: register mif not set: %d\n",
1906 				    reg_mif_num);
1907 #endif
1908 			m_freem(m);
1909 			return (IPPROTO_DONE);
1910 		}
1911 
1912 		reghdr = (u_int32_t *)(pim + 1);
1913 
1914 		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1915 			goto pim6_input_to_daemon;
1916 
1917 		/*
1918 		 * Validate length
1919 		 */
1920 		if (pimlen < PIM6_REG_MINLEN) {
1921 			++pim6stat.pim6s_rcv_tooshort;
1922 			++pim6stat.pim6s_rcv_badregisters;
1923 #ifdef MRT6DEBUG
1924 			log(LOG_ERR,
1925 			    "pim6_input: register packet size too "
1926 			    "small %d from %s\n",
1927 			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1928 #endif
1929 			m_freem(m);
1930 			return (IPPROTO_DONE);
1931 		}
1932 
1933 		eip6 = (struct ip6_hdr *) (reghdr + 1);
1934 #ifdef MRT6DEBUG
1935 		if (V_mrt6debug & DEBUG_PIM)
1936 			log(LOG_DEBUG,
1937 			    "pim6_input[register], eip6: %s -> %s, "
1938 			    "eip6 plen %d\n",
1939 			    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1940 			    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1941 			    ntohs(eip6->ip6_plen));
1942 #endif
1943 
1944 		/* verify the version number of the inner packet */
1945 		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1946 			++pim6stat.pim6s_rcv_badregisters;
1947 #ifdef MRT6DEBUG
1948 			log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1949 			    "of the inner packet\n",
1950 			    (eip6->ip6_vfc & IPV6_VERSION));
1951 #endif
1952 			m_freem(m);
1953 			return (IPPROTO_NONE);
1954 		}
1955 
1956 		/* verify the inner packet is destined to a mcast group */
1957 		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1958 			++pim6stat.pim6s_rcv_badregisters;
1959 #ifdef MRT6DEBUG
1960 			if (V_mrt6debug & DEBUG_PIM)
1961 				log(LOG_DEBUG,
1962 				    "pim6_input: inner packet of register "
1963 				    "is not multicast %s\n",
1964 				    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1965 #endif
1966 			m_freem(m);
1967 			return (IPPROTO_DONE);
1968 		}
1969 
1970 		/*
1971 		 * make a copy of the whole header to pass to the daemon later.
1972 		 */
1973 		mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1974 		if (mcp == NULL) {
1975 #ifdef MRT6DEBUG
1976 			log(LOG_ERR,
1977 			    "pim6_input: pim register: "
1978 			    "could not copy register head\n");
1979 #endif
1980 			m_freem(m);
1981 			return (IPPROTO_DONE);
1982 		}
1983 
1984 		/*
1985 		 * forward the inner ip6 packet; point m_data at the inner ip6.
1986 		 */
1987 		m_adj(m, off + PIM_MINLEN);
1988 #ifdef MRT6DEBUG
1989 		if (V_mrt6debug & DEBUG_PIM) {
1990 			log(LOG_DEBUG,
1991 			    "pim6_input: forwarding decapsulated register: "
1992 			    "src %s, dst %s, mif %d\n",
1993 			    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1994 			    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1995 			    reg_mif_num);
1996 		}
1997 #endif
1998 
1999 		rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
2000 				dst.sin6_family, 0);
2001 
2002 		/* prepare the register head to send to the mrouting daemon */
2003 		m = mcp;
2004 	}
2005 
2006 	/*
2007 	 * Pass the PIM message up to the daemon; if it is a register message
2008 	 * pass the 'head' only up to the daemon. This includes the
2009 	 * encapsulator ip6 header, pim header, register header and the
2010 	 * encapsulated ip6 header.
2011 	 */
2012   pim6_input_to_daemon:
2013 	rip6_input(&m, offp, proto);
2014 	return (IPPROTO_DONE);
2015 }
2016 
2017 static int
2018 ip6_mroute_modevent(module_t mod, int type, void *unused)
2019 {
2020 
2021 	switch (type) {
2022 	case MOD_LOAD:
2023 		MROUTER6_LOCK_INIT();
2024 		MFC6_LOCK_INIT();
2025 		MIF6_LOCK_INIT();
2026 
2027 		pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM,
2028 			pim6_encapcheck,
2029 			(const struct protosw *)&in6_pim_protosw, NULL);
2030 		if (pim6_encap_cookie == NULL) {
2031 			printf("ip6_mroute: unable to attach pim6 encap\n");
2032 			MIF6_LOCK_DESTROY();
2033 			MFC6_LOCK_DESTROY();
2034 			MROUTER6_LOCK_DESTROY();
2035 			return (EINVAL);
2036 		}
2037 
2038 		ip6_mforward = X_ip6_mforward;
2039 		ip6_mrouter_done = X_ip6_mrouter_done;
2040 		ip6_mrouter_get = X_ip6_mrouter_get;
2041 		ip6_mrouter_set = X_ip6_mrouter_set;
2042 		mrt6_ioctl = X_mrt6_ioctl;
2043 		break;
2044 
2045 	case MOD_UNLOAD:
2046 		if (V_ip6_mrouter != NULL)
2047 			return EINVAL;
2048 
2049 		if (pim6_encap_cookie) {
2050 			encap_detach(pim6_encap_cookie);
2051 			pim6_encap_cookie = NULL;
2052 		}
2053 		X_ip6_mrouter_done();
2054 		ip6_mforward = NULL;
2055 		ip6_mrouter_done = NULL;
2056 		ip6_mrouter_get = NULL;
2057 		ip6_mrouter_set = NULL;
2058 		mrt6_ioctl = NULL;
2059 
2060 		MIF6_LOCK_DESTROY();
2061 		MFC6_LOCK_DESTROY();
2062 		MROUTER6_LOCK_DESTROY();
2063 		break;
2064 
2065 	default:
2066 		return (EOPNOTSUPP);
2067 	}
2068 
2069 	return (0);
2070 }
2071 
2072 static moduledata_t ip6_mroutemod = {
2073 	"ip6_mroute",
2074 	ip6_mroute_modevent,
2075 	0
2076 };
2077 
2078 DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PSEUDO, SI_ORDER_ANY);
2079