xref: /freebsd/sys/netinet6/ip6_mroute.c (revision d3d381b2b194b4d24853e92eecef55f262688d1a)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (C) 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  *
31  *	$KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $
32  */
33 
34 /*-
35  * Copyright (c) 1989 Stephen Deering
36  * Copyright (c) 1992, 1993
37  *      The Regents of the University of California.  All rights reserved.
38  *
39  * This code is derived from software contributed to Berkeley by
40  * Stephen Deering of Stanford University.
41  *
42  * Redistribution and use in source and binary forms, with or without
43  * modification, are permitted provided that the following conditions
44  * are met:
45  * 1. Redistributions of source code must retain the above copyright
46  *    notice, this list of conditions and the following disclaimer.
47  * 2. Redistributions in binary form must reproduce the above copyright
48  *    notice, this list of conditions and the following disclaimer in the
49  *    documentation and/or other materials provided with the distribution.
50  * 3. Neither the name of the University nor the names of its contributors
51  *    may be used to endorse or promote products derived from this software
52  *    without specific prior written permission.
53  *
54  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
64  * SUCH DAMAGE.
65  *
66  *	@(#)ip_mroute.c	8.2 (Berkeley) 11/15/93
67  *	BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp
68  */
69 
70 /*
71  * IP multicast forwarding procedures
72  *
73  * Written by David Waitzman, BBN Labs, August 1988.
74  * Modified by Steve Deering, Stanford, February 1989.
75  * Modified by Mark J. Steiglitz, Stanford, May, 1991
76  * Modified by Van Jacobson, LBL, January 1993
77  * Modified by Ajit Thyagarajan, PARC, August 1993
78  * Modified by Bill Fenner, PARC, April 1994
79  *
80  * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
81  */
82 
83 #include <sys/cdefs.h>
84 __FBSDID("$FreeBSD$");
85 
86 #include "opt_inet6.h"
87 
88 #include <sys/param.h>
89 #include <sys/callout.h>
90 #include <sys/errno.h>
91 #include <sys/kernel.h>
92 #include <sys/lock.h>
93 #include <sys/malloc.h>
94 #include <sys/mbuf.h>
95 #include <sys/module.h>
96 #include <sys/domain.h>
97 #include <sys/protosw.h>
98 #include <sys/sdt.h>
99 #include <sys/signalvar.h>
100 #include <sys/socket.h>
101 #include <sys/socketvar.h>
102 #include <sys/sockio.h>
103 #include <sys/sx.h>
104 #include <sys/sysctl.h>
105 #include <sys/syslog.h>
106 #include <sys/systm.h>
107 #include <sys/time.h>
108 
109 #include <net/if.h>
110 #include <net/if_var.h>
111 #include <net/if_types.h>
112 #include <net/vnet.h>
113 
114 #include <netinet/in.h>
115 #include <netinet/in_var.h>
116 #include <netinet/icmp6.h>
117 #include <netinet/ip_encap.h>
118 
119 #include <netinet/ip6.h>
120 #include <netinet/in_kdtrace.h>
121 #include <netinet6/ip6_var.h>
122 #include <netinet6/scope6_var.h>
123 #include <netinet6/nd6.h>
124 #include <netinet6/ip6_mroute.h>
125 #include <netinet6/pim6.h>
126 #include <netinet6/pim6_var.h>
127 
128 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
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 int pim6_encapcheck(const struct mbuf *, int, int, void *);
142 static int pim6_input(struct mbuf *, int, int, void *);
143 
144 static const struct encap_config ipv6_encap_cfg = {
145 	.proto = IPPROTO_PIM,
146 	.min_length = sizeof(struct ip6_hdr) + PIM_MINLEN,
147 	.exact_match = 8,
148 	.check = pim6_encapcheck,
149 	.input = pim6_input
150 };
151 
152 
153 VNET_DEFINE_STATIC(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	MRT6STAT_INC(name)	mrt6stat.name += 1
166 #define NO_RTE_FOUND	0x1
167 #define RTE_FOUND	0x2
168 
169 static struct mtx mrouter6_mtx;
170 #define	MROUTER6_LOCK()		mtx_lock(&mrouter6_mtx)
171 #define	MROUTER6_UNLOCK()	mtx_unlock(&mrouter6_mtx)
172 #define	MROUTER6_LOCK_ASSERT()	do {					\
173 	mtx_assert(&mrouter6_mtx, MA_OWNED);				\
174 	NET_ASSERT_GIANT();						\
175 } while (0)
176 #define	MROUTER6_LOCK_INIT()	\
177 	mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
178 #define	MROUTER6_LOCK_DESTROY()	mtx_destroy(&mrouter6_mtx)
179 
180 static struct mf6c *mf6ctable[MF6CTBLSIZ];
181 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
182     &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
183     "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
184     "netinet6/ip6_mroute.h)");
185 
186 static struct mtx mfc6_mtx;
187 #define	MFC6_LOCK()		mtx_lock(&mfc6_mtx)
188 #define	MFC6_UNLOCK()		mtx_unlock(&mfc6_mtx)
189 #define	MFC6_LOCK_ASSERT()	do {					\
190 	mtx_assert(&mfc6_mtx, MA_OWNED);				\
191 	NET_ASSERT_GIANT();						\
192 } while (0)
193 #define	MFC6_LOCK_INIT()		\
194 	mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
195 #define	MFC6_LOCK_DESTROY()	mtx_destroy(&mfc6_mtx)
196 
197 static u_char n6expire[MF6CTBLSIZ];
198 
199 static struct mif6 mif6table[MAXMIFS];
200 static int
201 sysctl_mif6table(SYSCTL_HANDLER_ARGS)
202 {
203 	struct mif6_sctl *out;
204 	int error;
205 
206 	out = malloc(sizeof(struct mif6_sctl) * MAXMIFS, M_TEMP, M_WAITOK);
207 	for (int i = 0; i < MAXMIFS; i++) {
208 		out[i].m6_flags		= mif6table[i].m6_flags;
209 		out[i].m6_rate_limit	= mif6table[i].m6_rate_limit;
210 		out[i].m6_lcl_addr	= mif6table[i].m6_lcl_addr;
211 		if (mif6table[i].m6_ifp != NULL)
212 			out[i].m6_ifp	= mif6table[i].m6_ifp->if_index;
213 		else
214 			out[i].m6_ifp	= 0;
215 		out[i].m6_pkt_in	= mif6table[i].m6_pkt_in;
216 		out[i].m6_pkt_out	= mif6table[i].m6_pkt_out;
217 		out[i].m6_bytes_in	= mif6table[i].m6_bytes_in;
218 		out[i].m6_bytes_out	= mif6table[i].m6_bytes_out;
219 	}
220 	error = SYSCTL_OUT(req, out, sizeof(struct mif6_sctl) * MAXMIFS);
221 	free(out, M_TEMP);
222 	return (error);
223 }
224 SYSCTL_PROC(_net_inet6_ip6, OID_AUTO, mif6table, CTLTYPE_OPAQUE | CTLFLAG_RD,
225     NULL, 0, sysctl_mif6table, "S,mif6_sctl[MAXMIFS]",
226     "IPv6 Multicast Interfaces (struct mif6_sctl[MAXMIFS], "
227     "netinet6/ip6_mroute.h)");
228 
229 static struct mtx mif6_mtx;
230 #define	MIF6_LOCK()		mtx_lock(&mif6_mtx)
231 #define	MIF6_UNLOCK()		mtx_unlock(&mif6_mtx)
232 #define	MIF6_LOCK_ASSERT()	mtx_assert(&mif6_mtx, MA_OWNED)
233 #define	MIF6_LOCK_INIT()	\
234 	mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
235 #define	MIF6_LOCK_DESTROY()	mtx_destroy(&mif6_mtx)
236 
237 #ifdef MRT6DEBUG
238 VNET_DEFINE_STATIC(u_int, mrt6debug) = 0;	/* debug level */
239 #define	V_mrt6debug		VNET(mrt6debug)
240 #define DEBUG_MFC	0x02
241 #define DEBUG_FORWARD	0x04
242 #define DEBUG_EXPIRE	0x08
243 #define DEBUG_XMIT	0x10
244 #define DEBUG_REG	0x20
245 #define DEBUG_PIM	0x40
246 #define	DEBUG_ERR	0x80
247 #define	DEBUG_ANY	0x7f
248 #define	MRT6_DLOG(m, fmt, ...)	\
249 	if (V_mrt6debug & (m))	\
250 		log(((m) & DEBUG_ERR) ? LOG_ERR: LOG_DEBUG, \
251 		    "%s: " fmt "\n", __func__, ##__VA_ARGS__)
252 #else
253 #define	MRT6_DLOG(m, fmt, ...)
254 #endif
255 
256 static void	expire_upcalls(void *);
257 #define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
258 #define	UPCALL_EXPIRE	6		/* number of timeouts */
259 
260 /*
261  * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
262  */
263 
264 /*
265  * 'Interfaces' associated with decapsulator (so we can tell
266  * packets that went through it from ones that get reflected
267  * by a broken gateway).  Different from IPv4 register_if,
268  * these interfaces are linked into the system ifnet list,
269  * because per-interface IPv6 statistics are maintained in
270  * ifp->if_afdata.  But it does not have any routes point
271  * to them.  I.e., packets can't be sent this way.  They
272  * only exist as a placeholder for multicast source
273  * verification.
274  */
275 static struct ifnet *multicast_register_if6;
276 
277 #define ENCAP_HOPS 64
278 
279 /*
280  * Private variables.
281  */
282 static mifi_t nummifs = 0;
283 static mifi_t reg_mif_num = (mifi_t)-1;
284 
285 static struct pim6stat pim6stat;
286 SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RW,
287     &pim6stat, pim6stat,
288     "PIM Statistics (struct pim6stat, netinet6/pim6_var.h)");
289 
290 #define	PIM6STAT_INC(name)	pim6stat.name += 1
291 VNET_DEFINE_STATIC(int, pim6);
292 #define	V_pim6		VNET(pim6)
293 
294 /*
295  * Hash function for a source, group entry
296  */
297 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
298 				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
299 				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
300 				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
301 
302 /*
303  * Find a route for a given origin IPv6 address and Multicast group address.
304  */
305 #define MF6CFIND(o, g, rt) do { \
306 	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
307 	rt = NULL; \
308 	while (_rt) { \
309 		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
310 		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
311 		    (_rt->mf6c_stall == NULL)) { \
312 			rt = _rt; \
313 			break; \
314 		} \
315 		_rt = _rt->mf6c_next; \
316 	} \
317 	if (rt == NULL) { \
318 		MRT6STAT_INC(mrt6s_mfc_misses); \
319 	} \
320 } while (/*CONSTCOND*/ 0)
321 
322 /*
323  * Macros to compute elapsed time efficiently
324  * Borrowed from Van Jacobson's scheduling code
325  * XXX: replace with timersub() ?
326  */
327 #define TV_DELTA(a, b, delta) do { \
328 	    int xxs; \
329 		\
330 	    delta = (a).tv_usec - (b).tv_usec; \
331 	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
332 	       switch (xxs) { \
333 		      case 2: \
334 			  delta += 1000000; \
335 			      /* FALLTHROUGH */ \
336 		      case 1: \
337 			  delta += 1000000; \
338 			  break; \
339 		      default: \
340 			  delta += (1000000 * xxs); \
341 	       } \
342 	    } \
343 } while (/*CONSTCOND*/ 0)
344 
345 /* XXX: replace with timercmp(a, b, <) ? */
346 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
347 	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
348 
349 #ifdef UPCALL_TIMING
350 #define UPCALL_MAX	50
351 static u_long upcall_data[UPCALL_MAX + 1];
352 static void collate();
353 #endif /* UPCALL_TIMING */
354 
355 static int ip6_mrouter_init(struct socket *, int, int);
356 static int add_m6fc(struct mf6cctl *);
357 static int add_m6if(struct mif6ctl *);
358 static int del_m6fc(struct mf6cctl *);
359 static int del_m6if(mifi_t *);
360 static int del_m6if_locked(mifi_t *);
361 static int get_mif6_cnt(struct sioc_mif_req6 *);
362 static int get_sg_cnt(struct sioc_sg_req6 *);
363 
364 static struct callout expire_upcalls_ch;
365 
366 int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
367 int X_ip6_mrouter_done(void);
368 int X_ip6_mrouter_set(struct socket *, struct sockopt *);
369 int X_ip6_mrouter_get(struct socket *, struct sockopt *);
370 int X_mrt6_ioctl(u_long, caddr_t);
371 
372 /*
373  * Handle MRT setsockopt commands to modify the multicast routing tables.
374  */
375 int
376 X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
377 {
378 	int error = 0;
379 	int optval;
380 	struct mif6ctl mifc;
381 	struct mf6cctl mfcc;
382 	mifi_t mifi;
383 
384 	if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
385 		return (EPERM);
386 
387 	switch (sopt->sopt_name) {
388 	case MRT6_INIT:
389 #ifdef MRT6_OINIT
390 	case MRT6_OINIT:
391 #endif
392 		error = sooptcopyin(sopt, &optval, sizeof(optval),
393 		    sizeof(optval));
394 		if (error)
395 			break;
396 		error = ip6_mrouter_init(so, optval, sopt->sopt_name);
397 		break;
398 	case MRT6_DONE:
399 		error = X_ip6_mrouter_done();
400 		break;
401 	case MRT6_ADD_MIF:
402 		error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
403 		if (error)
404 			break;
405 		error = add_m6if(&mifc);
406 		break;
407 	case MRT6_ADD_MFC:
408 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
409 		if (error)
410 			break;
411 		error = add_m6fc(&mfcc);
412 		break;
413 	case MRT6_DEL_MFC:
414 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
415 		if (error)
416 			break;
417 		error = del_m6fc(&mfcc);
418 		break;
419 	case MRT6_DEL_MIF:
420 		error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
421 		if (error)
422 			break;
423 		error = del_m6if(&mifi);
424 		break;
425 	case MRT6_PIM:
426 		error = sooptcopyin(sopt, &optval, sizeof(optval),
427 		    sizeof(optval));
428 		if (error)
429 			break;
430 		error = set_pim6(&optval);
431 		break;
432 	default:
433 		error = EOPNOTSUPP;
434 		break;
435 	}
436 
437 	return (error);
438 }
439 
440 /*
441  * Handle MRT getsockopt commands
442  */
443 int
444 X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
445 {
446 	int error = 0;
447 
448 	if (so != V_ip6_mrouter)
449 		return (EACCES);
450 
451 	switch (sopt->sopt_name) {
452 		case MRT6_PIM:
453 			error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
454 			break;
455 	}
456 	return (error);
457 }
458 
459 /*
460  * Handle ioctl commands to obtain information from the cache
461  */
462 int
463 X_mrt6_ioctl(u_long cmd, caddr_t data)
464 {
465 	int ret;
466 
467 	ret = EINVAL;
468 
469 	switch (cmd) {
470 	case SIOCGETSGCNT_IN6:
471 		ret = get_sg_cnt((struct sioc_sg_req6 *)data);
472 		break;
473 
474 	case SIOCGETMIFCNT_IN6:
475 		ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
476 		break;
477 
478 	default:
479 		break;
480 	}
481 
482 	return (ret);
483 }
484 
485 /*
486  * returns the packet, byte, rpf-failure count for the source group provided
487  */
488 static int
489 get_sg_cnt(struct sioc_sg_req6 *req)
490 {
491 	struct mf6c *rt;
492 	int ret;
493 
494 	ret = 0;
495 
496 	MFC6_LOCK();
497 
498 	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
499 	if (rt == NULL) {
500 		ret = ESRCH;
501 	} else {
502 		req->pktcnt = rt->mf6c_pkt_cnt;
503 		req->bytecnt = rt->mf6c_byte_cnt;
504 		req->wrong_if = rt->mf6c_wrong_if;
505 	}
506 
507 	MFC6_UNLOCK();
508 
509 	return (ret);
510 }
511 
512 /*
513  * returns the input and output packet and byte counts on the mif provided
514  */
515 static int
516 get_mif6_cnt(struct sioc_mif_req6 *req)
517 {
518 	mifi_t mifi;
519 	int ret;
520 
521 	ret = 0;
522 	mifi = req->mifi;
523 
524 	MIF6_LOCK();
525 
526 	if (mifi >= nummifs) {
527 		ret = EINVAL;
528 	} else {
529 		req->icount = mif6table[mifi].m6_pkt_in;
530 		req->ocount = mif6table[mifi].m6_pkt_out;
531 		req->ibytes = mif6table[mifi].m6_bytes_in;
532 		req->obytes = mif6table[mifi].m6_bytes_out;
533 	}
534 
535 	MIF6_UNLOCK();
536 
537 	return (ret);
538 }
539 
540 static int
541 set_pim6(int *i)
542 {
543 	if ((*i != 1) && (*i != 0))
544 		return (EINVAL);
545 
546 	V_pim6 = *i;
547 
548 	return (0);
549 }
550 
551 /*
552  * Enable multicast routing
553  */
554 static int
555 ip6_mrouter_init(struct socket *so, int v, int cmd)
556 {
557 
558 	MRT6_DLOG(DEBUG_ANY, "so_type = %d, pr_protocol = %d",
559 	    so->so_type, so->so_proto->pr_protocol);
560 
561 	if (so->so_type != SOCK_RAW ||
562 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
563 		return (EOPNOTSUPP);
564 
565 	if (v != 1)
566 		return (ENOPROTOOPT);
567 
568 	MROUTER6_LOCK();
569 
570 	if (V_ip6_mrouter != NULL) {
571 		MROUTER6_UNLOCK();
572 		return (EADDRINUSE);
573 	}
574 
575 	V_ip6_mrouter = so;
576 	V_ip6_mrouter_ver = cmd;
577 
578 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
579 	bzero((caddr_t)n6expire, sizeof(n6expire));
580 
581 	V_pim6 = 0;/* used for stubbing out/in pim stuff */
582 
583 	callout_init(&expire_upcalls_ch, 0);
584 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
585 	    expire_upcalls, NULL);
586 
587 	MROUTER6_UNLOCK();
588 	MRT6_DLOG(DEBUG_ANY, "finished");
589 
590 	return (0);
591 }
592 
593 /*
594  * Disable IPv6 multicast forwarding.
595  */
596 int
597 X_ip6_mrouter_done(void)
598 {
599 	mifi_t mifi;
600 	u_long i;
601 	struct mf6c *rt;
602 	struct rtdetq *rte;
603 
604 	MROUTER6_LOCK();
605 
606 	if (V_ip6_mrouter == NULL) {
607 		MROUTER6_UNLOCK();
608 		return (EINVAL);
609 	}
610 
611 	/*
612 	 * For each phyint in use, disable promiscuous reception of all IPv6
613 	 * multicasts.
614 	 */
615 	for (mifi = 0; mifi < nummifs; mifi++) {
616 		if (mif6table[mifi].m6_ifp &&
617 		    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
618 			if_allmulti(mif6table[mifi].m6_ifp, 0);
619 		}
620 	}
621 	bzero((caddr_t)mif6table, sizeof(mif6table));
622 	nummifs = 0;
623 
624 	V_pim6 = 0; /* used to stub out/in pim specific code */
625 
626 	callout_stop(&expire_upcalls_ch);
627 
628 	/*
629 	 * Free all multicast forwarding cache entries.
630 	 */
631 	MFC6_LOCK();
632 	for (i = 0; i < MF6CTBLSIZ; i++) {
633 		rt = mf6ctable[i];
634 		while (rt) {
635 			struct mf6c *frt;
636 
637 			for (rte = rt->mf6c_stall; rte != NULL; ) {
638 				struct rtdetq *n = rte->next;
639 
640 				m_freem(rte->m);
641 				free(rte, M_MRTABLE6);
642 				rte = n;
643 			}
644 			frt = rt;
645 			rt = rt->mf6c_next;
646 			free(frt, M_MRTABLE6);
647 		}
648 	}
649 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
650 	MFC6_UNLOCK();
651 
652 	/*
653 	 * Reset register interface
654 	 */
655 	if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
656 		if_detach(multicast_register_if6);
657 		if_free(multicast_register_if6);
658 		reg_mif_num = (mifi_t)-1;
659 		multicast_register_if6 = NULL;
660 	}
661 
662 	V_ip6_mrouter = NULL;
663 	V_ip6_mrouter_ver = 0;
664 
665 	MROUTER6_UNLOCK();
666 	MRT6_DLOG(DEBUG_ANY, "finished");
667 
668 	return (0);
669 }
670 
671 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
672 
673 /*
674  * Add a mif to the mif table
675  */
676 static int
677 add_m6if(struct mif6ctl *mifcp)
678 {
679 	struct mif6 *mifp;
680 	struct ifnet *ifp;
681 	int error;
682 
683 	MIF6_LOCK();
684 
685 	if (mifcp->mif6c_mifi >= MAXMIFS) {
686 		MIF6_UNLOCK();
687 		return (EINVAL);
688 	}
689 	mifp = mif6table + mifcp->mif6c_mifi;
690 	if (mifp->m6_ifp != NULL) {
691 		MIF6_UNLOCK();
692 		return (EADDRINUSE); /* XXX: is it appropriate? */
693 	}
694 	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
695 		MIF6_UNLOCK();
696 		return (ENXIO);
697 	}
698 
699 	ifp = ifnet_byindex(mifcp->mif6c_pifi);
700 
701 	if (mifcp->mif6c_flags & MIFF_REGISTER) {
702 		if (reg_mif_num == (mifi_t)-1) {
703 			ifp = if_alloc(IFT_OTHER);
704 
705 			if_initname(ifp, "register_mif", 0);
706 			ifp->if_flags |= IFF_LOOPBACK;
707 			if_attach(ifp);
708 			multicast_register_if6 = ifp;
709 			reg_mif_num = mifcp->mif6c_mifi;
710 			/*
711 			 * it is impossible to guess the ifindex of the
712 			 * register interface.  So mif6c_pifi is automatically
713 			 * calculated.
714 			 */
715 			mifcp->mif6c_pifi = ifp->if_index;
716 		} else {
717 			ifp = multicast_register_if6;
718 		}
719 	} else {
720 		/* Make sure the interface supports multicast */
721 		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
722 			MIF6_UNLOCK();
723 			return (EOPNOTSUPP);
724 		}
725 
726 		error = if_allmulti(ifp, 1);
727 		if (error) {
728 			MIF6_UNLOCK();
729 			return (error);
730 		}
731 	}
732 
733 	mifp->m6_flags     = mifcp->mif6c_flags;
734 	mifp->m6_ifp       = ifp;
735 
736 	/* initialize per mif pkt counters */
737 	mifp->m6_pkt_in    = 0;
738 	mifp->m6_pkt_out   = 0;
739 	mifp->m6_bytes_in  = 0;
740 	mifp->m6_bytes_out = 0;
741 
742 	/* Adjust nummifs up if the mifi is higher than nummifs */
743 	if (nummifs <= mifcp->mif6c_mifi)
744 		nummifs = mifcp->mif6c_mifi + 1;
745 
746 	MIF6_UNLOCK();
747 	MRT6_DLOG(DEBUG_ANY, "mif #%d, phyint %s", mifcp->mif6c_mifi,
748 	    if_name(ifp));
749 
750 	return (0);
751 }
752 
753 /*
754  * Delete a mif from the mif table
755  */
756 static int
757 del_m6if_locked(mifi_t *mifip)
758 {
759 	struct mif6 *mifp = mif6table + *mifip;
760 	mifi_t mifi;
761 	struct ifnet *ifp;
762 
763 	MIF6_LOCK_ASSERT();
764 
765 	if (*mifip >= nummifs)
766 		return (EINVAL);
767 	if (mifp->m6_ifp == NULL)
768 		return (EINVAL);
769 
770 	if (!(mifp->m6_flags & MIFF_REGISTER)) {
771 		/* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
772 		ifp = mifp->m6_ifp;
773 		if_allmulti(ifp, 0);
774 	} else {
775 		if (reg_mif_num != (mifi_t)-1 &&
776 		    multicast_register_if6 != NULL) {
777 			if_detach(multicast_register_if6);
778 			if_free(multicast_register_if6);
779 			reg_mif_num = (mifi_t)-1;
780 			multicast_register_if6 = NULL;
781 		}
782 	}
783 
784 	bzero((caddr_t)mifp, sizeof(*mifp));
785 
786 	/* Adjust nummifs down */
787 	for (mifi = nummifs; mifi > 0; mifi--)
788 		if (mif6table[mifi - 1].m6_ifp)
789 			break;
790 	nummifs = mifi;
791 	MRT6_DLOG(DEBUG_ANY, "mif %d, nummifs %d", *mifip, nummifs);
792 
793 	return (0);
794 }
795 
796 static int
797 del_m6if(mifi_t *mifip)
798 {
799 	int cc;
800 
801 	MIF6_LOCK();
802 	cc = del_m6if_locked(mifip);
803 	MIF6_UNLOCK();
804 
805 	return (cc);
806 }
807 
808 /*
809  * Add an mfc entry
810  */
811 static int
812 add_m6fc(struct mf6cctl *mfccp)
813 {
814 	struct mf6c *rt;
815 	u_long hash;
816 	struct rtdetq *rte;
817 	u_short nstl;
818 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
819 
820 	MFC6_LOCK();
821 
822 	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
823 		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
824 
825 	/* If an entry already exists, just update the fields */
826 	if (rt) {
827 		MRT6_DLOG(DEBUG_MFC, "no upcall o %s g %s p %x",
828 		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
829 		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
830 		    mfccp->mf6cc_parent);
831 
832 		rt->mf6c_parent = mfccp->mf6cc_parent;
833 		rt->mf6c_ifset = mfccp->mf6cc_ifset;
834 
835 		MFC6_UNLOCK();
836 		return (0);
837 	}
838 
839 	/*
840 	 * Find the entry for which the upcall was made and update
841 	 */
842 	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
843 			mfccp->mf6cc_mcastgrp.sin6_addr);
844 	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
845 		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
846 				       &mfccp->mf6cc_origin.sin6_addr) &&
847 		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
848 				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
849 		    (rt->mf6c_stall != NULL)) {
850 
851 			if (nstl++)
852 				log(LOG_ERR,
853 				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
854 				    "multiple kernel entries",
855 				    ip6_sprintf(ip6bufo,
856 					    &mfccp->mf6cc_origin.sin6_addr),
857 				    ip6_sprintf(ip6bufg,
858 					    &mfccp->mf6cc_mcastgrp.sin6_addr),
859 				    mfccp->mf6cc_parent, rt->mf6c_stall);
860 
861 			MRT6_DLOG(DEBUG_MFC, "o %s g %s p %x dbg %p",
862 			    ip6_sprintf(ip6bufo,
863 			    &mfccp->mf6cc_origin.sin6_addr),
864 			    ip6_sprintf(ip6bufg,
865 				&mfccp->mf6cc_mcastgrp.sin6_addr),
866 			    mfccp->mf6cc_parent, rt->mf6c_stall);
867 
868 			rt->mf6c_origin     = mfccp->mf6cc_origin;
869 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
870 			rt->mf6c_parent     = mfccp->mf6cc_parent;
871 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
872 			/* initialize pkt counters per src-grp */
873 			rt->mf6c_pkt_cnt    = 0;
874 			rt->mf6c_byte_cnt   = 0;
875 			rt->mf6c_wrong_if   = 0;
876 
877 			rt->mf6c_expire = 0;	/* Don't clean this guy up */
878 			n6expire[hash]--;
879 
880 			/* free packets Qed at the end of this entry */
881 			for (rte = rt->mf6c_stall; rte != NULL; ) {
882 				struct rtdetq *n = rte->next;
883 				ip6_mdq(rte->m, rte->ifp, rt);
884 				m_freem(rte->m);
885 #ifdef UPCALL_TIMING
886 				collate(&(rte->t));
887 #endif /* UPCALL_TIMING */
888 				free(rte, M_MRTABLE6);
889 				rte = n;
890 			}
891 			rt->mf6c_stall = NULL;
892 		}
893 	}
894 
895 	/*
896 	 * It is possible that an entry is being inserted without an upcall
897 	 */
898 	if (nstl == 0) {
899 		MRT6_DLOG(DEBUG_MFC, "no upcall h %lu o %s g %s p %x", hash,
900 		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
901 		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
902 		    mfccp->mf6cc_parent);
903 
904 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
905 
906 			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
907 					       &mfccp->mf6cc_origin.sin6_addr)&&
908 			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
909 					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
910 
911 				rt->mf6c_origin     = mfccp->mf6cc_origin;
912 				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
913 				rt->mf6c_parent     = mfccp->mf6cc_parent;
914 				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
915 				/* initialize pkt counters per src-grp */
916 				rt->mf6c_pkt_cnt    = 0;
917 				rt->mf6c_byte_cnt   = 0;
918 				rt->mf6c_wrong_if   = 0;
919 
920 				if (rt->mf6c_expire)
921 					n6expire[hash]--;
922 				rt->mf6c_expire	   = 0;
923 			}
924 		}
925 		if (rt == NULL) {
926 			/* no upcall, so make a new entry */
927 			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
928 						  M_NOWAIT);
929 			if (rt == NULL) {
930 				MFC6_UNLOCK();
931 				return (ENOBUFS);
932 			}
933 
934 			/* insert new entry at head of hash chain */
935 			rt->mf6c_origin     = mfccp->mf6cc_origin;
936 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
937 			rt->mf6c_parent     = mfccp->mf6cc_parent;
938 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
939 			/* initialize pkt counters per src-grp */
940 			rt->mf6c_pkt_cnt    = 0;
941 			rt->mf6c_byte_cnt   = 0;
942 			rt->mf6c_wrong_if   = 0;
943 			rt->mf6c_expire     = 0;
944 			rt->mf6c_stall = NULL;
945 
946 			/* link into table */
947 			rt->mf6c_next  = mf6ctable[hash];
948 			mf6ctable[hash] = rt;
949 		}
950 	}
951 
952 	MFC6_UNLOCK();
953 	return (0);
954 }
955 
956 #ifdef UPCALL_TIMING
957 /*
958  * collect delay statistics on the upcalls
959  */
960 static void
961 collate(struct timeval *t)
962 {
963 	u_long d;
964 	struct timeval tp;
965 	u_long delta;
966 
967 	GET_TIME(tp);
968 
969 	if (TV_LT(*t, tp))
970 	{
971 		TV_DELTA(tp, *t, delta);
972 
973 		d = delta >> 10;
974 		if (d > UPCALL_MAX)
975 			d = UPCALL_MAX;
976 
977 		++upcall_data[d];
978 	}
979 }
980 #endif /* UPCALL_TIMING */
981 
982 /*
983  * Delete an mfc entry
984  */
985 static int
986 del_m6fc(struct mf6cctl *mfccp)
987 {
988 #ifdef MRT6DEBUG
989 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
990 #endif
991 	struct sockaddr_in6	origin;
992 	struct sockaddr_in6	mcastgrp;
993 	struct mf6c		*rt;
994 	struct mf6c		**nptr;
995 	u_long		hash;
996 
997 	origin = mfccp->mf6cc_origin;
998 	mcastgrp = mfccp->mf6cc_mcastgrp;
999 	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
1000 
1001 	MRT6_DLOG(DEBUG_MFC, "orig %s mcastgrp %s",
1002 	    ip6_sprintf(ip6bufo, &origin.sin6_addr),
1003 	    ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
1004 
1005 	MFC6_LOCK();
1006 
1007 	nptr = &mf6ctable[hash];
1008 	while ((rt = *nptr) != NULL) {
1009 		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
1010 				       &rt->mf6c_origin.sin6_addr) &&
1011 		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1012 				       &rt->mf6c_mcastgrp.sin6_addr) &&
1013 		    rt->mf6c_stall == NULL)
1014 			break;
1015 
1016 		nptr = &rt->mf6c_next;
1017 	}
1018 	if (rt == NULL) {
1019 		MFC6_UNLOCK();
1020 		return (EADDRNOTAVAIL);
1021 	}
1022 
1023 	*nptr = rt->mf6c_next;
1024 	free(rt, M_MRTABLE6);
1025 
1026 	MFC6_UNLOCK();
1027 
1028 	return (0);
1029 }
1030 
1031 static int
1032 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1033 {
1034 
1035 	if (s) {
1036 		if (sbappendaddr(&s->so_rcv,
1037 				 (struct sockaddr *)src,
1038 				 mm, (struct mbuf *)0) != 0) {
1039 			sorwakeup(s);
1040 			return (0);
1041 		}
1042 	}
1043 	m_freem(mm);
1044 	return (-1);
1045 }
1046 
1047 /*
1048  * IPv6 multicast forwarding function. This function assumes that the packet
1049  * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1050  * pointed to by "ifp", and the packet is to be relayed to other networks
1051  * that have members of the packet's destination IPv6 multicast group.
1052  *
1053  * The packet is returned unscathed to the caller, unless it is
1054  * erroneous, in which case a non-zero return value tells the caller to
1055  * discard it.
1056  *
1057  * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1058  * this function is called in the originating context (i.e., not when
1059  * forwarding a packet from other node).  ip6_output(), which is currently the
1060  * only function that calls this function is called in the originating context,
1061  * explicitly ensures this condition.  It is caller's responsibility to ensure
1062  * that if this function is called from somewhere else in the originating
1063  * context in the future.
1064  */
1065 int
1066 X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1067 {
1068 	struct rtdetq *rte;
1069 	struct mbuf *mb0;
1070 	struct mf6c *rt;
1071 	struct mif6 *mifp;
1072 	struct mbuf *mm;
1073 	u_long hash;
1074 	mifi_t mifi;
1075 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1076 #ifdef UPCALL_TIMING
1077 	struct timeval tp;
1078 
1079 	GET_TIME(tp);
1080 #endif /* UPCALL_TIMING */
1081 
1082 	MRT6_DLOG(DEBUG_FORWARD, "src %s, dst %s, ifindex %d",
1083 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1084 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst), ifp->if_index);
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_uptime) {
1105 			V_ip6_log_time = time_uptime;
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 	MRT6STAT_INC(mrt6s_mfc_lookups);
1124 
1125 	/* Entry exists, so forward if necessary */
1126 	if (rt) {
1127 		MFC6_UNLOCK();
1128 		return (ip6_mdq(m, ifp, rt));
1129 	}
1130 
1131 	/*
1132 	 * If we don't have a route for packet's origin,
1133 	 * Make a copy of the packet & send message to routing daemon.
1134 	 */
1135 	MRT6STAT_INC(mrt6s_no_route);
1136 	MRT6_DLOG(DEBUG_FORWARD | DEBUG_MFC, "no rte s %s g %s",
1137 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1138 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1139 
1140 	/*
1141 	 * Allocate mbufs early so that we don't do extra work if we
1142 	 * are just going to fail anyway.
1143 	 */
1144 	rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6, M_NOWAIT);
1145 	if (rte == NULL) {
1146 		MFC6_UNLOCK();
1147 		return (ENOBUFS);
1148 	}
1149 	mb0 = m_copym(m, 0, M_COPYALL, M_NOWAIT);
1150 	/*
1151 	 * Pullup packet header if needed before storing it,
1152 	 * as other references may modify it in the meantime.
1153 	 */
1154 	if (mb0 && (!M_WRITABLE(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1155 		mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1156 	if (mb0 == NULL) {
1157 		free(rte, M_MRTABLE6);
1158 		MFC6_UNLOCK();
1159 		return (ENOBUFS);
1160 	}
1161 
1162 	/* is there an upcall waiting for this packet? */
1163 	hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1164 	for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1165 		if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1166 		    &rt->mf6c_origin.sin6_addr) &&
1167 		    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1168 		    &rt->mf6c_mcastgrp.sin6_addr) && (rt->mf6c_stall != NULL))
1169 			break;
1170 	}
1171 
1172 	if (rt == NULL) {
1173 		struct mrt6msg *im;
1174 #ifdef MRT6_OINIT
1175 		struct omrt6msg *oim;
1176 #endif
1177 		/* no upcall, so make a new entry */
1178 		rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, M_NOWAIT);
1179 		if (rt == NULL) {
1180 			free(rte, M_MRTABLE6);
1181 			m_freem(mb0);
1182 			MFC6_UNLOCK();
1183 			return (ENOBUFS);
1184 		}
1185 		/*
1186 		 * Make a copy of the header to send to the user
1187 		 * level process
1188 		 */
1189 		mm = m_copym(mb0, 0, sizeof(struct ip6_hdr), M_NOWAIT);
1190 		if (mm == NULL) {
1191 			free(rte, M_MRTABLE6);
1192 			m_freem(mb0);
1193 			free(rt, M_MRTABLE6);
1194 			MFC6_UNLOCK();
1195 			return (ENOBUFS);
1196 		}
1197 
1198 		/*
1199 		 * Send message to routing daemon
1200 		 */
1201 		sin6.sin6_addr = ip6->ip6_src;
1202 		im = NULL;
1203 #ifdef MRT6_OINIT
1204 		oim = NULL;
1205 #endif
1206 		switch (V_ip6_mrouter_ver) {
1207 #ifdef MRT6_OINIT
1208 		case MRT6_OINIT:
1209 			oim = mtod(mm, struct omrt6msg *);
1210 			oim->im6_msgtype = MRT6MSG_NOCACHE;
1211 			oim->im6_mbz = 0;
1212 			break;
1213 #endif
1214 		case MRT6_INIT:
1215 			im = mtod(mm, struct mrt6msg *);
1216 			im->im6_msgtype = MRT6MSG_NOCACHE;
1217 			im->im6_mbz = 0;
1218 			break;
1219 		default:
1220 			free(rte, M_MRTABLE6);
1221 			m_freem(mb0);
1222 			free(rt, M_MRTABLE6);
1223 			MFC6_UNLOCK();
1224 			return (EINVAL);
1225 		}
1226 
1227 		MRT6_DLOG(DEBUG_FORWARD, "getting the iif info in the kernel");
1228 		for (mifp = mif6table, mifi = 0;
1229 		    mifi < nummifs && mifp->m6_ifp != ifp; mifp++, mifi++)
1230 				;
1231 
1232 		switch (V_ip6_mrouter_ver) {
1233 #ifdef MRT6_OINIT
1234 		case MRT6_OINIT:
1235 			oim->im6_mif = mifi;
1236 			break;
1237 #endif
1238 		case MRT6_INIT:
1239 			im->im6_mif = mifi;
1240 			break;
1241 		}
1242 
1243 		if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1244 			log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1245 			    "socket queue full\n");
1246 			MRT6STAT_INC(mrt6s_upq_sockfull);
1247 			free(rte, M_MRTABLE6);
1248 			m_freem(mb0);
1249 			free(rt, M_MRTABLE6);
1250 			MFC6_UNLOCK();
1251 			return (ENOBUFS);
1252 		}
1253 
1254 		MRT6STAT_INC(mrt6s_upcalls);
1255 
1256 		/* insert new entry at head of hash chain */
1257 		bzero(rt, sizeof(*rt));
1258 		rt->mf6c_origin.sin6_family = AF_INET6;
1259 		rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1260 		rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1261 		rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1262 		rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1263 		rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1264 		rt->mf6c_expire = UPCALL_EXPIRE;
1265 		n6expire[hash]++;
1266 		rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1267 
1268 		/* link into table */
1269 		rt->mf6c_next  = mf6ctable[hash];
1270 		mf6ctable[hash] = rt;
1271 		/* Add this entry to the end of the queue */
1272 		rt->mf6c_stall = rte;
1273 	} else {
1274 		/* determine if q has overflowed */
1275 		struct rtdetq **p;
1276 		int npkts = 0;
1277 
1278 		for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1279 			if (++npkts > MAX_UPQ6) {
1280 				MRT6STAT_INC(mrt6s_upq_ovflw);
1281 				free(rte, M_MRTABLE6);
1282 				m_freem(mb0);
1283 				MFC6_UNLOCK();
1284 				return (0);
1285 			}
1286 
1287 		/* Add this entry to the end of the queue */
1288 		*p = rte;
1289 	}
1290 
1291 	rte->next = NULL;
1292 	rte->m = mb0;
1293 	rte->ifp = ifp;
1294 #ifdef UPCALL_TIMING
1295 	rte->t = tp;
1296 #endif /* UPCALL_TIMING */
1297 
1298 	MFC6_UNLOCK();
1299 
1300 	return (0);
1301 }
1302 
1303 /*
1304  * Clean up cache entries if upcalls are not serviced
1305  * Call from the Slow Timeout mechanism, every half second.
1306  */
1307 static void
1308 expire_upcalls(void *unused)
1309 {
1310 #ifdef MRT6DEBUG
1311 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1312 #endif
1313 	struct rtdetq *rte;
1314 	struct mf6c *mfc, **nptr;
1315 	u_long i;
1316 
1317 	MFC6_LOCK();
1318 	for (i = 0; i < MF6CTBLSIZ; i++) {
1319 		if (n6expire[i] == 0)
1320 			continue;
1321 		nptr = &mf6ctable[i];
1322 		while ((mfc = *nptr) != NULL) {
1323 			rte = mfc->mf6c_stall;
1324 			/*
1325 			 * Skip real cache entries
1326 			 * Make sure it wasn't marked to not expire (shouldn't happen)
1327 			 * If it expires now
1328 			 */
1329 			if (rte != NULL &&
1330 			    mfc->mf6c_expire != 0 &&
1331 			    --mfc->mf6c_expire == 0) {
1332 				MRT6_DLOG(DEBUG_EXPIRE, "expiring (%s %s)",
1333 				    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1334 				    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1335 				/*
1336 				 * drop all the packets
1337 				 * free the mbuf with the pkt, if, timing info
1338 				 */
1339 				do {
1340 					struct rtdetq *n = rte->next;
1341 					m_freem(rte->m);
1342 					free(rte, M_MRTABLE6);
1343 					rte = n;
1344 				} while (rte != NULL);
1345 				MRT6STAT_INC(mrt6s_cache_cleanups);
1346 				n6expire[i]--;
1347 
1348 				*nptr = mfc->mf6c_next;
1349 				free(mfc, M_MRTABLE6);
1350 			} else {
1351 				nptr = &mfc->mf6c_next;
1352 			}
1353 		}
1354 	}
1355 	MFC6_UNLOCK();
1356 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1357 	    expire_upcalls, NULL);
1358 }
1359 
1360 /*
1361  * Packet forwarding routine once entry in the cache is made
1362  */
1363 static int
1364 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1365 {
1366 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1367 	mifi_t mifi, iif;
1368 	struct mif6 *mifp;
1369 	int plen = m->m_pkthdr.len;
1370 	struct in6_addr src0, dst0; /* copies for local work */
1371 	u_int32_t iszone, idzone, oszone, odzone;
1372 	int error = 0;
1373 
1374 /*
1375  * Macro to send packet on mif.  Since RSVP packets don't get counted on
1376  * input, they shouldn't get counted on output, so statistics keeping is
1377  * separate.
1378  */
1379 
1380 #define MC6_SEND(ip6, mifp, m) do {				\
1381 	if ((mifp)->m6_flags & MIFF_REGISTER)			\
1382 		register_send((ip6), (mifp), (m));		\
1383 	else							\
1384 		phyint_send((ip6), (mifp), (m));		\
1385 } while (/*CONSTCOND*/ 0)
1386 
1387 	/*
1388 	 * Don't forward if it didn't arrive from the parent mif
1389 	 * for its origin.
1390 	 */
1391 	mifi = rt->mf6c_parent;
1392 	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1393 		/* came in the wrong interface */
1394 		MRT6_DLOG(DEBUG_FORWARD,
1395 		    "wrong if: ifid %d mifi %d mififid %x", ifp->if_index,
1396 		    mifi, mif6table[mifi].m6_ifp->if_index);
1397 		MRT6STAT_INC(mrt6s_wrong_if);
1398 		rt->mf6c_wrong_if++;
1399 		/*
1400 		 * If we are doing PIM processing, and we are forwarding
1401 		 * packets on this interface, send a message to the
1402 		 * routing daemon.
1403 		 */
1404 		/* have to make sure this is a valid mif */
1405 		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1406 			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1407 				/*
1408 				 * Check the M_LOOP flag to avoid an
1409 				 * unnecessary PIM assert.
1410 				 * XXX: M_LOOP is an ad-hoc hack...
1411 				 */
1412 				static struct sockaddr_in6 sin6 =
1413 				{ sizeof(sin6), AF_INET6 };
1414 
1415 				struct mbuf *mm;
1416 				struct mrt6msg *im;
1417 #ifdef MRT6_OINIT
1418 				struct omrt6msg *oim;
1419 #endif
1420 
1421 				mm = m_copym(m, 0, sizeof(struct ip6_hdr),
1422 				    M_NOWAIT);
1423 				if (mm &&
1424 				    (!M_WRITABLE(mm) ||
1425 				     mm->m_len < sizeof(struct ip6_hdr)))
1426 					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1427 				if (mm == NULL)
1428 					return (ENOBUFS);
1429 
1430 #ifdef MRT6_OINIT
1431 				oim = NULL;
1432 #endif
1433 				im = NULL;
1434 				switch (V_ip6_mrouter_ver) {
1435 #ifdef MRT6_OINIT
1436 				case MRT6_OINIT:
1437 					oim = mtod(mm, struct omrt6msg *);
1438 					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1439 					oim->im6_mbz = 0;
1440 					break;
1441 #endif
1442 				case MRT6_INIT:
1443 					im = mtod(mm, struct mrt6msg *);
1444 					im->im6_msgtype = MRT6MSG_WRONGMIF;
1445 					im->im6_mbz = 0;
1446 					break;
1447 				default:
1448 					m_freem(mm);
1449 					return (EINVAL);
1450 				}
1451 
1452 				for (mifp = mif6table, iif = 0;
1453 				     iif < nummifs && mifp &&
1454 					     mifp->m6_ifp != ifp;
1455 				     mifp++, iif++)
1456 					;
1457 
1458 				switch (V_ip6_mrouter_ver) {
1459 #ifdef MRT6_OINIT
1460 				case MRT6_OINIT:
1461 					oim->im6_mif = iif;
1462 					sin6.sin6_addr = oim->im6_src;
1463 					break;
1464 #endif
1465 				case MRT6_INIT:
1466 					im->im6_mif = iif;
1467 					sin6.sin6_addr = im->im6_src;
1468 					break;
1469 				}
1470 
1471 				MRT6STAT_INC(mrt6s_upcalls);
1472 
1473 				if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1474 					MRT6_DLOG(DEBUG_ANY,
1475 					    "ip6_mrouter socket queue full");
1476 					MRT6STAT_INC(mrt6s_upq_sockfull);
1477 					return (ENOBUFS);
1478 				}	/* if socket Q full */
1479 			}		/* if PIM */
1480 		return (0);
1481 	}			/* if wrong iif */
1482 
1483 	/* If I sourced this packet, it counts as output, else it was input. */
1484 	if (m->m_pkthdr.rcvif == NULL) {
1485 		/* XXX: is rcvif really NULL when output?? */
1486 		mif6table[mifi].m6_pkt_out++;
1487 		mif6table[mifi].m6_bytes_out += plen;
1488 	} else {
1489 		mif6table[mifi].m6_pkt_in++;
1490 		mif6table[mifi].m6_bytes_in += plen;
1491 	}
1492 	rt->mf6c_pkt_cnt++;
1493 	rt->mf6c_byte_cnt += plen;
1494 
1495 	/*
1496 	 * For each mif, forward a copy of the packet if there are group
1497 	 * members downstream on the interface.
1498 	 */
1499 	src0 = ip6->ip6_src;
1500 	dst0 = ip6->ip6_dst;
1501 	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1502 	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1503 		IP6STAT_INC(ip6s_badscope);
1504 		return (error);
1505 	}
1506 	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1507 		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1508 			/*
1509 			 * check if the outgoing packet is going to break
1510 			 * a scope boundary.
1511 			 * XXX For packets through PIM register tunnel
1512 			 * interface, we believe a routing daemon.
1513 			 */
1514 			if (!(mif6table[rt->mf6c_parent].m6_flags &
1515 			      MIFF_REGISTER) &&
1516 			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1517 				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1518 				    &oszone) ||
1519 				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1520 				    &odzone) ||
1521 				    iszone != oszone ||
1522 				    idzone != odzone) {
1523 					IP6STAT_INC(ip6s_badscope);
1524 					continue;
1525 				}
1526 			}
1527 
1528 			mifp->m6_pkt_out++;
1529 			mifp->m6_bytes_out += plen;
1530 			MC6_SEND(ip6, mifp, m);
1531 		}
1532 	}
1533 	return (0);
1534 }
1535 
1536 static void
1537 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1538 {
1539 #ifdef MRT6DEBUG
1540 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1541 #endif
1542 	struct mbuf *mb_copy;
1543 	struct ifnet *ifp = mifp->m6_ifp;
1544 	int error = 0;
1545 	u_long linkmtu;
1546 
1547 	/*
1548 	 * Make a new reference to the packet; make sure that
1549 	 * the IPv6 header is actually copied, not just referenced,
1550 	 * so that ip6_output() only scribbles on the copy.
1551 	 */
1552 	mb_copy = m_copym(m, 0, M_COPYALL, M_NOWAIT);
1553 	if (mb_copy &&
1554 	    (!M_WRITABLE(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1555 		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1556 	if (mb_copy == NULL) {
1557 		return;
1558 	}
1559 	/* set MCAST flag to the outgoing packet */
1560 	mb_copy->m_flags |= M_MCAST;
1561 
1562 	/*
1563 	 * If we sourced the packet, call ip6_output since we may devide
1564 	 * the packet into fragments when the packet is too big for the
1565 	 * outgoing interface.
1566 	 * Otherwise, we can simply send the packet to the interface
1567 	 * sending queue.
1568 	 */
1569 	if (m->m_pkthdr.rcvif == NULL) {
1570 		struct ip6_moptions im6o;
1571 
1572 		im6o.im6o_multicast_ifp = ifp;
1573 		/* XXX: ip6_output will override ip6->ip6_hlim */
1574 		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1575 		im6o.im6o_multicast_loop = 1;
1576 		error = ip6_output(mb_copy, NULL, NULL, IPV6_FORWARDING, &im6o,
1577 		    NULL, NULL);
1578 
1579 		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1580 		    (uint16_t)(mifp - mif6table), error);
1581 		return;
1582 	}
1583 
1584 	/*
1585 	 * If configured to loop back multicasts by default,
1586 	 * loop back a copy now.
1587 	 */
1588 	if (in6_mcast_loop)
1589 		ip6_mloopback(ifp, m);
1590 
1591 	/*
1592 	 * Put the packet into the sending queue of the outgoing interface
1593 	 * if it would fit in the MTU of the interface.
1594 	 */
1595 	linkmtu = IN6_LINKMTU(ifp);
1596 	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1597 		struct sockaddr_in6 dst6;
1598 
1599 		bzero(&dst6, sizeof(dst6));
1600 		dst6.sin6_len = sizeof(struct sockaddr_in6);
1601 		dst6.sin6_family = AF_INET6;
1602 		dst6.sin6_addr = ip6->ip6_dst;
1603 
1604 		IP_PROBE(send, NULL, NULL, ip6, ifp, NULL, ip6);
1605 		/*
1606 		 * We just call if_output instead of nd6_output here, since
1607 		 * we need no ND for a multicast forwarded packet...right?
1608 		 */
1609 		m_clrprotoflags(m);	/* Avoid confusing lower layers. */
1610 		error = (*ifp->if_output)(ifp, mb_copy,
1611 		    (struct sockaddr *)&dst6, NULL);
1612 		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1613 		    (uint16_t)(mifp - mif6table), error);
1614 	} else {
1615 		/*
1616 		 * pMTU discovery is intentionally disabled by default, since
1617 		 * various router may notify pMTU in multicast, which can be
1618 		 * a DDoS to a router
1619 		 */
1620 		if (V_ip6_mcast_pmtu)
1621 			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1622 		else {
1623 			MRT6_DLOG(DEBUG_XMIT, " packet too big on %s o %s "
1624 			    "g %s size %d (discarded)", if_name(ifp),
1625 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1626 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1627 			    mb_copy->m_pkthdr.len);
1628 			m_freem(mb_copy); /* simply discard the packet */
1629 		}
1630 	}
1631 }
1632 
1633 static int
1634 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1635 {
1636 #ifdef MRT6DEBUG
1637 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1638 #endif
1639 	struct mbuf *mm;
1640 	int i, len = m->m_pkthdr.len;
1641 	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1642 	struct mrt6msg *im6;
1643 
1644 	MRT6_DLOG(DEBUG_ANY, "src %s dst %s",
1645 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1646 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1647 	PIM6STAT_INC(pim6s_snd_registers);
1648 
1649 	/* Make a copy of the packet to send to the user level process. */
1650 	mm = m_gethdr(M_NOWAIT, MT_DATA);
1651 	if (mm == NULL)
1652 		return (ENOBUFS);
1653 	mm->m_data += max_linkhdr;
1654 	mm->m_len = sizeof(struct ip6_hdr);
1655 
1656 	if ((mm->m_next = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL) {
1657 		m_freem(mm);
1658 		return (ENOBUFS);
1659 	}
1660 	i = MHLEN - M_LEADINGSPACE(mm);
1661 	if (i > len)
1662 		i = len;
1663 	mm = m_pullup(mm, i);
1664 	if (mm == NULL)
1665 		return (ENOBUFS);
1666 /* TODO: check it! */
1667 	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1668 
1669 	/*
1670 	 * Send message to routing daemon
1671 	 */
1672 	sin6.sin6_addr = ip6->ip6_src;
1673 
1674 	im6 = mtod(mm, struct mrt6msg *);
1675 	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1676 	im6->im6_mbz          = 0;
1677 
1678 	im6->im6_mif = mif - mif6table;
1679 
1680 	/* iif info is not given for reg. encap.n */
1681 	MRT6STAT_INC(mrt6s_upcalls);
1682 
1683 	if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1684 		MRT6_DLOG(DEBUG_ANY, "ip6_mrouter socket queue full");
1685 		MRT6STAT_INC(mrt6s_upq_sockfull);
1686 		return (ENOBUFS);
1687 	}
1688 	return (0);
1689 }
1690 
1691 /*
1692  * pim6_encapcheck() is called by the encap6_input() path at runtime to
1693  * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1694  * into the kernel.
1695  */
1696 static int
1697 pim6_encapcheck(const struct mbuf *m __unused, int off __unused,
1698     int proto __unused, void *arg __unused)
1699 {
1700 
1701     KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1702     return (8);		/* claim the datagram. */
1703 }
1704 
1705 /*
1706  * PIM sparse mode hook
1707  * Receives the pim control messages, and passes them up to the listening
1708  * socket, using rip6_input.
1709  * The only message processed is the REGISTER pim message; the pim header
1710  * is stripped off, and the inner packet is passed to register_mforward.
1711  */
1712 static int
1713 pim6_input(struct mbuf *m, int off, int proto, void *arg __unused)
1714 {
1715 	struct pim *pim; /* pointer to a pim struct */
1716 	struct ip6_hdr *ip6;
1717 	int pimlen;
1718 	int minlen;
1719 
1720 	PIM6STAT_INC(pim6s_rcv_total);
1721 
1722 	ip6 = mtod(m, struct ip6_hdr *);
1723 	pimlen = m->m_pkthdr.len - off;
1724 
1725 	/*
1726 	 * Validate lengths
1727 	 */
1728 	if (pimlen < PIM_MINLEN) {
1729 		PIM6STAT_INC(pim6s_rcv_tooshort);
1730 		MRT6_DLOG(DEBUG_PIM, "PIM packet too short");
1731 		m_freem(m);
1732 		return (IPPROTO_DONE);
1733 	}
1734 
1735 	/*
1736 	 * if the packet is at least as big as a REGISTER, go ahead
1737 	 * and grab the PIM REGISTER header size, to avoid another
1738 	 * possible m_pullup() later.
1739 	 *
1740 	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1741 	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1742 	 */
1743 	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1744 
1745 	/*
1746 	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1747 	 * possibly the PIM REGISTER header
1748 	 */
1749 #ifndef PULLDOWN_TEST
1750 	IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1751 	/* adjust pointer */
1752 	ip6 = mtod(m, struct ip6_hdr *);
1753 
1754 	/* adjust mbuf to point to the PIM header */
1755 	pim = (struct pim *)((caddr_t)ip6 + off);
1756 #else
1757 	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1758 	if (pim == NULL) {
1759 		PIM6STAT_INC(pim6s_rcv_tooshort);
1760 		return (IPPROTO_DONE);
1761 	}
1762 #endif
1763 
1764 #define PIM6_CHECKSUM
1765 #ifdef PIM6_CHECKSUM
1766 	{
1767 		int cksumlen;
1768 
1769 		/*
1770 		 * Validate checksum.
1771 		 * If PIM REGISTER, exclude the data packet
1772 		 */
1773 		if (pim->pim_type == PIM_REGISTER)
1774 			cksumlen = PIM_MINLEN;
1775 		else
1776 			cksumlen = pimlen;
1777 
1778 		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1779 			PIM6STAT_INC(pim6s_rcv_badsum);
1780 			MRT6_DLOG(DEBUG_PIM, "invalid checksum");
1781 			m_freem(m);
1782 			return (IPPROTO_DONE);
1783 		}
1784 	}
1785 #endif /* PIM_CHECKSUM */
1786 
1787 	/* PIM version check */
1788 	if (pim->pim_ver != PIM_VERSION) {
1789 		PIM6STAT_INC(pim6s_rcv_badversion);
1790 		MRT6_DLOG(DEBUG_ANY | DEBUG_ERR,
1791 		    "incorrect version %d, expecting %d",
1792 		    pim->pim_ver, PIM_VERSION);
1793 		m_freem(m);
1794 		return (IPPROTO_DONE);
1795 	}
1796 
1797 	if (pim->pim_type == PIM_REGISTER) {
1798 		/*
1799 		 * since this is a REGISTER, we'll make a copy of the register
1800 		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1801 		 * routing daemon.
1802 		 */
1803 		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1804 
1805 		struct mbuf *mcp;
1806 		struct ip6_hdr *eip6;
1807 		u_int32_t *reghdr;
1808 		int rc;
1809 #ifdef MRT6DEBUG
1810 		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1811 #endif
1812 
1813 		PIM6STAT_INC(pim6s_rcv_registers);
1814 
1815 		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1816 			MRT6_DLOG(DEBUG_PIM, "register mif not set: %d",
1817 			    reg_mif_num);
1818 			m_freem(m);
1819 			return (IPPROTO_DONE);
1820 		}
1821 
1822 		reghdr = (u_int32_t *)(pim + 1);
1823 
1824 		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1825 			goto pim6_input_to_daemon;
1826 
1827 		/*
1828 		 * Validate length
1829 		 */
1830 		if (pimlen < PIM6_REG_MINLEN) {
1831 			PIM6STAT_INC(pim6s_rcv_tooshort);
1832 			PIM6STAT_INC(pim6s_rcv_badregisters);
1833 			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "register packet "
1834 			    "size too small %d from %s",
1835 			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1836 			m_freem(m);
1837 			return (IPPROTO_DONE);
1838 		}
1839 
1840 		eip6 = (struct ip6_hdr *) (reghdr + 1);
1841 		MRT6_DLOG(DEBUG_PIM, "eip6: %s -> %s, eip6 plen %d",
1842 		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1843 		    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1844 		    ntohs(eip6->ip6_plen));
1845 
1846 		/* verify the version number of the inner packet */
1847 		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1848 			PIM6STAT_INC(pim6s_rcv_badregisters);
1849 			MRT6_DLOG(DEBUG_ANY, "invalid IP version (%d) "
1850 			    "of the inner packet",
1851 			    (eip6->ip6_vfc & IPV6_VERSION));
1852 			m_freem(m);
1853 			return (IPPROTO_DONE);
1854 		}
1855 
1856 		/* verify the inner packet is destined to a mcast group */
1857 		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1858 			PIM6STAT_INC(pim6s_rcv_badregisters);
1859 			MRT6_DLOG(DEBUG_PIM, "inner packet of register "
1860 			    "is not multicast %s",
1861 			    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1862 			m_freem(m);
1863 			return (IPPROTO_DONE);
1864 		}
1865 
1866 		/*
1867 		 * make a copy of the whole header to pass to the daemon later.
1868 		 */
1869 		mcp = m_copym(m, 0, off + PIM6_REG_MINLEN, M_NOWAIT);
1870 		if (mcp == NULL) {
1871 			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "pim register: "
1872 			    "could not copy register head");
1873 			m_freem(m);
1874 			return (IPPROTO_DONE);
1875 		}
1876 
1877 		/*
1878 		 * forward the inner ip6 packet; point m_data at the inner ip6.
1879 		 */
1880 		m_adj(m, off + PIM_MINLEN);
1881 		MRT6_DLOG(DEBUG_PIM, "forwarding decapsulated register: "
1882 		    "src %s, dst %s, mif %d",
1883 		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1884 		    ip6_sprintf(ip6bufd, &eip6->ip6_dst), reg_mif_num);
1885 
1886 		rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1887 				dst.sin6_family, 0);
1888 
1889 		/* prepare the register head to send to the mrouting daemon */
1890 		m = mcp;
1891 	}
1892 
1893 	/*
1894 	 * Pass the PIM message up to the daemon; if it is a register message
1895 	 * pass the 'head' only up to the daemon. This includes the
1896 	 * encapsulator ip6 header, pim header, register header and the
1897 	 * encapsulated ip6 header.
1898 	 */
1899   pim6_input_to_daemon:
1900 	return (rip6_input(&m, &off, proto));
1901 }
1902 
1903 static int
1904 ip6_mroute_modevent(module_t mod, int type, void *unused)
1905 {
1906 
1907 	switch (type) {
1908 	case MOD_LOAD:
1909 		MROUTER6_LOCK_INIT();
1910 		MFC6_LOCK_INIT();
1911 		MIF6_LOCK_INIT();
1912 
1913 		pim6_encap_cookie = ip6_encap_attach(&ipv6_encap_cfg,
1914 		    NULL, M_WAITOK);
1915 		if (pim6_encap_cookie == NULL) {
1916 			printf("ip6_mroute: unable to attach pim6 encap\n");
1917 			MIF6_LOCK_DESTROY();
1918 			MFC6_LOCK_DESTROY();
1919 			MROUTER6_LOCK_DESTROY();
1920 			return (EINVAL);
1921 		}
1922 
1923 		ip6_mforward = X_ip6_mforward;
1924 		ip6_mrouter_done = X_ip6_mrouter_done;
1925 		ip6_mrouter_get = X_ip6_mrouter_get;
1926 		ip6_mrouter_set = X_ip6_mrouter_set;
1927 		mrt6_ioctl = X_mrt6_ioctl;
1928 		break;
1929 
1930 	case MOD_UNLOAD:
1931 		if (V_ip6_mrouter != NULL)
1932 			return EINVAL;
1933 
1934 		if (pim6_encap_cookie) {
1935 			ip6_encap_detach(pim6_encap_cookie);
1936 			pim6_encap_cookie = NULL;
1937 		}
1938 		X_ip6_mrouter_done();
1939 		ip6_mforward = NULL;
1940 		ip6_mrouter_done = NULL;
1941 		ip6_mrouter_get = NULL;
1942 		ip6_mrouter_set = NULL;
1943 		mrt6_ioctl = NULL;
1944 
1945 		MIF6_LOCK_DESTROY();
1946 		MFC6_LOCK_DESTROY();
1947 		MROUTER6_LOCK_DESTROY();
1948 		break;
1949 
1950 	default:
1951 		return (EOPNOTSUPP);
1952 	}
1953 
1954 	return (0);
1955 }
1956 
1957 static moduledata_t ip6_mroutemod = {
1958 	"ip6_mroute",
1959 	ip6_mroute_modevent,
1960 	0
1961 };
1962 
1963 DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PROTO_MC, SI_ORDER_ANY);
1964