xref: /freebsd/sys/netinet6/ip6_mroute.c (revision e796cc77c586c2955b2f3940dbf4991b31e8d289)
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/raw_cb.h>
113 #include <net/vnet.h>
114 
115 #include <netinet/in.h>
116 #include <netinet/in_var.h>
117 #include <netinet/icmp6.h>
118 #include <netinet/ip_encap.h>
119 
120 #include <netinet/ip6.h>
121 #include <netinet/in_kdtrace.h>
122 #include <netinet6/ip6_var.h>
123 #include <netinet6/scope6_var.h>
124 #include <netinet6/nd6.h>
125 #include <netinet6/ip6_mroute.h>
126 #include <netinet6/pim6.h>
127 #include <netinet6/pim6_var.h>
128 
129 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
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 int	register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
134 static int	set_pim6(int *);
135 static int	socket_send(struct socket *, struct mbuf *,
136 		    struct sockaddr_in6 *);
137 
138 extern int in6_mcast_loop;
139 extern struct domain inet6domain;
140 
141 static const struct encaptab *pim6_encap_cookie;
142 static const struct protosw in6_pim_protosw = {
143 	.pr_type =		SOCK_RAW,
144 	.pr_domain =		&inet6domain,
145 	.pr_protocol =		IPPROTO_PIM,
146 	.pr_flags =		PR_ATOMIC|PR_ADDR|PR_LASTHDR,
147 	.pr_input =		pim6_input,
148 	.pr_output =		rip6_output,
149 	.pr_ctloutput =		rip6_ctloutput,
150 	.pr_usrreqs =		&rip6_usrreqs
151 };
152 static int pim6_encapcheck(const struct mbuf *, int, int, void *);
153 
154 static VNET_DEFINE(int, ip6_mrouter_ver) = 0;
155 #define	V_ip6_mrouter_ver	VNET(ip6_mrouter_ver)
156 
157 SYSCTL_DECL(_net_inet6);
158 SYSCTL_DECL(_net_inet6_ip6);
159 static SYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim, CTLFLAG_RW, 0, "PIM");
160 
161 static struct mrt6stat mrt6stat;
162 SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
163     &mrt6stat, mrt6stat,
164     "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
165 
166 #define	MRT6STAT_INC(name)	mrt6stat.name += 1
167 #define NO_RTE_FOUND	0x1
168 #define RTE_FOUND	0x2
169 
170 static struct mtx mrouter6_mtx;
171 #define	MROUTER6_LOCK()		mtx_lock(&mrouter6_mtx)
172 #define	MROUTER6_UNLOCK()	mtx_unlock(&mrouter6_mtx)
173 #define	MROUTER6_LOCK_ASSERT()	do {					\
174 	mtx_assert(&mrouter6_mtx, MA_OWNED);				\
175 	NET_ASSERT_GIANT();						\
176 } while (0)
177 #define	MROUTER6_LOCK_INIT()	\
178 	mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
179 #define	MROUTER6_LOCK_DESTROY()	mtx_destroy(&mrouter6_mtx)
180 
181 static struct mf6c *mf6ctable[MF6CTBLSIZ];
182 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
183     &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
184     "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
185     "netinet6/ip6_mroute.h)");
186 
187 static struct mtx mfc6_mtx;
188 #define	MFC6_LOCK()		mtx_lock(&mfc6_mtx)
189 #define	MFC6_UNLOCK()		mtx_unlock(&mfc6_mtx)
190 #define	MFC6_LOCK_ASSERT()	do {					\
191 	mtx_assert(&mfc6_mtx, MA_OWNED);				\
192 	NET_ASSERT_GIANT();						\
193 } while (0)
194 #define	MFC6_LOCK_INIT()		\
195 	mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
196 #define	MFC6_LOCK_DESTROY()	mtx_destroy(&mfc6_mtx)
197 
198 static u_char n6expire[MF6CTBLSIZ];
199 
200 static struct mif6 mif6table[MAXMIFS];
201 static int
202 sysctl_mif6table(SYSCTL_HANDLER_ARGS)
203 {
204 	struct mif6_sctl *out;
205 	int error;
206 
207 	out = malloc(sizeof(struct mif6_sctl) * MAXMIFS, M_TEMP, M_WAITOK);
208 	for (int i = 0; i < MAXMIFS; i++) {
209 		out[i].m6_flags		= mif6table[i].m6_flags;
210 		out[i].m6_rate_limit	= mif6table[i].m6_rate_limit;
211 		out[i].m6_lcl_addr	= mif6table[i].m6_lcl_addr;
212 		if (mif6table[i].m6_ifp != NULL)
213 			out[i].m6_ifp	= mif6table[i].m6_ifp->if_index;
214 		else
215 			out[i].m6_ifp	= 0;
216 		out[i].m6_pkt_in	= mif6table[i].m6_pkt_in;
217 		out[i].m6_pkt_out	= mif6table[i].m6_pkt_out;
218 		out[i].m6_bytes_in	= mif6table[i].m6_bytes_in;
219 		out[i].m6_bytes_out	= mif6table[i].m6_bytes_out;
220 	}
221 	error = SYSCTL_OUT(req, out, sizeof(struct mif6_sctl) * MAXMIFS);
222 	free(out, M_TEMP);
223 	return (error);
224 }
225 SYSCTL_PROC(_net_inet6_ip6, OID_AUTO, mif6table, CTLTYPE_OPAQUE | CTLFLAG_RD,
226     NULL, 0, sysctl_mif6table, "S,mif6_sctl[MAXMIFS]",
227     "IPv6 Multicast Interfaces (struct mif6_sctl[MAXMIFS], "
228     "netinet6/ip6_mroute.h)");
229 
230 static struct mtx mif6_mtx;
231 #define	MIF6_LOCK()		mtx_lock(&mif6_mtx)
232 #define	MIF6_UNLOCK()		mtx_unlock(&mif6_mtx)
233 #define	MIF6_LOCK_ASSERT()	mtx_assert(&mif6_mtx, MA_OWNED)
234 #define	MIF6_LOCK_INIT()	\
235 	mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
236 #define	MIF6_LOCK_DESTROY()	mtx_destroy(&mif6_mtx)
237 
238 #ifdef MRT6DEBUG
239 static VNET_DEFINE(u_int, mrt6debug) = 0;	/* debug level */
240 #define	V_mrt6debug		VNET(mrt6debug)
241 #define DEBUG_MFC	0x02
242 #define DEBUG_FORWARD	0x04
243 #define DEBUG_EXPIRE	0x08
244 #define DEBUG_XMIT	0x10
245 #define DEBUG_REG	0x20
246 #define DEBUG_PIM	0x40
247 #define	DEBUG_ERR	0x80
248 #define	DEBUG_ANY	0x7f
249 #define	MRT6_DLOG(m, fmt, ...)	\
250 	if (V_mrt6debug & (m))	\
251 		log(((m) & DEBUG_ERR) ? LOG_ERR: LOG_DEBUG, \
252 		    "%s: " fmt "\n", __func__, ##__VA_ARGS__)
253 #else
254 #define	MRT6_DLOG(m, fmt, ...)
255 #endif
256 
257 static void	expire_upcalls(void *);
258 #define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
259 #define	UPCALL_EXPIRE	6		/* number of timeouts */
260 
261 /*
262  * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
263  */
264 
265 /*
266  * 'Interfaces' associated with decapsulator (so we can tell
267  * packets that went through it from ones that get reflected
268  * by a broken gateway).  Different from IPv4 register_if,
269  * these interfaces are linked into the system ifnet list,
270  * because per-interface IPv6 statistics are maintained in
271  * ifp->if_afdata.  But it does not have any routes point
272  * to them.  I.e., packets can't be sent this way.  They
273  * only exist as a placeholder for multicast source
274  * verification.
275  */
276 static struct ifnet *multicast_register_if6;
277 
278 #define ENCAP_HOPS 64
279 
280 /*
281  * Private variables.
282  */
283 static mifi_t nummifs = 0;
284 static mifi_t reg_mif_num = (mifi_t)-1;
285 
286 static struct pim6stat pim6stat;
287 SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RW,
288     &pim6stat, pim6stat,
289     "PIM Statistics (struct pim6stat, netinet6/pim6_var.h)");
290 
291 #define	PIM6STAT_INC(name)	pim6stat.name += 1
292 static VNET_DEFINE(int, pim6);
293 #define	V_pim6		VNET(pim6)
294 
295 /*
296  * Hash function for a source, group entry
297  */
298 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
299 				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
300 				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
301 				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
302 
303 /*
304  * Find a route for a given origin IPv6 address and Multicast group address.
305  */
306 #define MF6CFIND(o, g, rt) do { \
307 	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
308 	rt = NULL; \
309 	while (_rt) { \
310 		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
311 		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
312 		    (_rt->mf6c_stall == NULL)) { \
313 			rt = _rt; \
314 			break; \
315 		} \
316 		_rt = _rt->mf6c_next; \
317 	} \
318 	if (rt == NULL) { \
319 		MRT6STAT_INC(mrt6s_mfc_misses); \
320 	} \
321 } while (/*CONSTCOND*/ 0)
322 
323 /*
324  * Macros to compute elapsed time efficiently
325  * Borrowed from Van Jacobson's scheduling code
326  * XXX: replace with timersub() ?
327  */
328 #define TV_DELTA(a, b, delta) do { \
329 	    int xxs; \
330 		\
331 	    delta = (a).tv_usec - (b).tv_usec; \
332 	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
333 	       switch (xxs) { \
334 		      case 2: \
335 			  delta += 1000000; \
336 			      /* FALLTHROUGH */ \
337 		      case 1: \
338 			  delta += 1000000; \
339 			  break; \
340 		      default: \
341 			  delta += (1000000 * xxs); \
342 	       } \
343 	    } \
344 } while (/*CONSTCOND*/ 0)
345 
346 /* XXX: replace with timercmp(a, b, <) ? */
347 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
348 	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
349 
350 #ifdef UPCALL_TIMING
351 #define UPCALL_MAX	50
352 static u_long upcall_data[UPCALL_MAX + 1];
353 static void collate();
354 #endif /* UPCALL_TIMING */
355 
356 static int ip6_mrouter_init(struct socket *, int, int);
357 static int add_m6fc(struct mf6cctl *);
358 static int add_m6if(struct mif6ctl *);
359 static int del_m6fc(struct mf6cctl *);
360 static int del_m6if(mifi_t *);
361 static int del_m6if_locked(mifi_t *);
362 static int get_mif6_cnt(struct sioc_mif_req6 *);
363 static int get_sg_cnt(struct sioc_sg_req6 *);
364 
365 static struct callout expire_upcalls_ch;
366 
367 int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
368 int X_ip6_mrouter_done(void);
369 int X_ip6_mrouter_set(struct socket *, struct sockopt *);
370 int X_ip6_mrouter_get(struct socket *, struct sockopt *);
371 int X_mrt6_ioctl(u_long, caddr_t);
372 
373 /*
374  * Handle MRT setsockopt commands to modify the multicast routing tables.
375  */
376 int
377 X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
378 {
379 	int error = 0;
380 	int optval;
381 	struct mif6ctl mifc;
382 	struct mf6cctl mfcc;
383 	mifi_t mifi;
384 
385 	if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
386 		return (EPERM);
387 
388 	switch (sopt->sopt_name) {
389 	case MRT6_INIT:
390 #ifdef MRT6_OINIT
391 	case MRT6_OINIT:
392 #endif
393 		error = sooptcopyin(sopt, &optval, sizeof(optval),
394 		    sizeof(optval));
395 		if (error)
396 			break;
397 		error = ip6_mrouter_init(so, optval, sopt->sopt_name);
398 		break;
399 	case MRT6_DONE:
400 		error = X_ip6_mrouter_done();
401 		break;
402 	case MRT6_ADD_MIF:
403 		error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
404 		if (error)
405 			break;
406 		error = add_m6if(&mifc);
407 		break;
408 	case MRT6_ADD_MFC:
409 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
410 		if (error)
411 			break;
412 		error = add_m6fc(&mfcc);
413 		break;
414 	case MRT6_DEL_MFC:
415 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
416 		if (error)
417 			break;
418 		error = del_m6fc(&mfcc);
419 		break;
420 	case MRT6_DEL_MIF:
421 		error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
422 		if (error)
423 			break;
424 		error = del_m6if(&mifi);
425 		break;
426 	case MRT6_PIM:
427 		error = sooptcopyin(sopt, &optval, sizeof(optval),
428 		    sizeof(optval));
429 		if (error)
430 			break;
431 		error = set_pim6(&optval);
432 		break;
433 	default:
434 		error = EOPNOTSUPP;
435 		break;
436 	}
437 
438 	return (error);
439 }
440 
441 /*
442  * Handle MRT getsockopt commands
443  */
444 int
445 X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
446 {
447 	int error = 0;
448 
449 	if (so != V_ip6_mrouter)
450 		return (EACCES);
451 
452 	switch (sopt->sopt_name) {
453 		case MRT6_PIM:
454 			error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
455 			break;
456 	}
457 	return (error);
458 }
459 
460 /*
461  * Handle ioctl commands to obtain information from the cache
462  */
463 int
464 X_mrt6_ioctl(u_long cmd, caddr_t data)
465 {
466 	int ret;
467 
468 	ret = EINVAL;
469 
470 	switch (cmd) {
471 	case SIOCGETSGCNT_IN6:
472 		ret = get_sg_cnt((struct sioc_sg_req6 *)data);
473 		break;
474 
475 	case SIOCGETMIFCNT_IN6:
476 		ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
477 		break;
478 
479 	default:
480 		break;
481 	}
482 
483 	return (ret);
484 }
485 
486 /*
487  * returns the packet, byte, rpf-failure count for the source group provided
488  */
489 static int
490 get_sg_cnt(struct sioc_sg_req6 *req)
491 {
492 	struct mf6c *rt;
493 	int ret;
494 
495 	ret = 0;
496 
497 	MFC6_LOCK();
498 
499 	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
500 	if (rt == NULL) {
501 		ret = ESRCH;
502 	} else {
503 		req->pktcnt = rt->mf6c_pkt_cnt;
504 		req->bytecnt = rt->mf6c_byte_cnt;
505 		req->wrong_if = rt->mf6c_wrong_if;
506 	}
507 
508 	MFC6_UNLOCK();
509 
510 	return (ret);
511 }
512 
513 /*
514  * returns the input and output packet and byte counts on the mif provided
515  */
516 static int
517 get_mif6_cnt(struct sioc_mif_req6 *req)
518 {
519 	mifi_t mifi;
520 	int ret;
521 
522 	ret = 0;
523 	mifi = req->mifi;
524 
525 	MIF6_LOCK();
526 
527 	if (mifi >= nummifs) {
528 		ret = EINVAL;
529 	} else {
530 		req->icount = mif6table[mifi].m6_pkt_in;
531 		req->ocount = mif6table[mifi].m6_pkt_out;
532 		req->ibytes = mif6table[mifi].m6_bytes_in;
533 		req->obytes = mif6table[mifi].m6_bytes_out;
534 	}
535 
536 	MIF6_UNLOCK();
537 
538 	return (ret);
539 }
540 
541 static int
542 set_pim6(int *i)
543 {
544 	if ((*i != 1) && (*i != 0))
545 		return (EINVAL);
546 
547 	V_pim6 = *i;
548 
549 	return (0);
550 }
551 
552 /*
553  * Enable multicast routing
554  */
555 static int
556 ip6_mrouter_init(struct socket *so, int v, int cmd)
557 {
558 
559 	MRT6_DLOG(DEBUG_ANY, "so_type = %d, pr_protocol = %d",
560 	    so->so_type, so->so_proto->pr_protocol);
561 
562 	if (so->so_type != SOCK_RAW ||
563 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
564 		return (EOPNOTSUPP);
565 
566 	if (v != 1)
567 		return (ENOPROTOOPT);
568 
569 	MROUTER6_LOCK();
570 
571 	if (V_ip6_mrouter != NULL) {
572 		MROUTER6_UNLOCK();
573 		return (EADDRINUSE);
574 	}
575 
576 	V_ip6_mrouter = so;
577 	V_ip6_mrouter_ver = cmd;
578 
579 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
580 	bzero((caddr_t)n6expire, sizeof(n6expire));
581 
582 	V_pim6 = 0;/* used for stubbing out/in pim stuff */
583 
584 	callout_init(&expire_upcalls_ch, 0);
585 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
586 	    expire_upcalls, NULL);
587 
588 	MROUTER6_UNLOCK();
589 	MRT6_DLOG(DEBUG_ANY, "finished");
590 
591 	return (0);
592 }
593 
594 /*
595  * Disable IPv6 multicast forwarding.
596  */
597 int
598 X_ip6_mrouter_done(void)
599 {
600 	mifi_t mifi;
601 	u_long i;
602 	struct mf6c *rt;
603 	struct rtdetq *rte;
604 
605 	MROUTER6_LOCK();
606 
607 	if (V_ip6_mrouter == NULL) {
608 		MROUTER6_UNLOCK();
609 		return (EINVAL);
610 	}
611 
612 	/*
613 	 * For each phyint in use, disable promiscuous reception of all IPv6
614 	 * multicasts.
615 	 */
616 	for (mifi = 0; mifi < nummifs; mifi++) {
617 		if (mif6table[mifi].m6_ifp &&
618 		    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
619 			if_allmulti(mif6table[mifi].m6_ifp, 0);
620 		}
621 	}
622 	bzero((caddr_t)mif6table, sizeof(mif6table));
623 	nummifs = 0;
624 
625 	V_pim6 = 0; /* used to stub out/in pim specific code */
626 
627 	callout_stop(&expire_upcalls_ch);
628 
629 	/*
630 	 * Free all multicast forwarding cache entries.
631 	 */
632 	MFC6_LOCK();
633 	for (i = 0; i < MF6CTBLSIZ; i++) {
634 		rt = mf6ctable[i];
635 		while (rt) {
636 			struct mf6c *frt;
637 
638 			for (rte = rt->mf6c_stall; rte != NULL; ) {
639 				struct rtdetq *n = rte->next;
640 
641 				m_freem(rte->m);
642 				free(rte, M_MRTABLE6);
643 				rte = n;
644 			}
645 			frt = rt;
646 			rt = rt->mf6c_next;
647 			free(frt, M_MRTABLE6);
648 		}
649 	}
650 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
651 	MFC6_UNLOCK();
652 
653 	/*
654 	 * Reset register interface
655 	 */
656 	if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
657 		if_detach(multicast_register_if6);
658 		if_free(multicast_register_if6);
659 		reg_mif_num = (mifi_t)-1;
660 		multicast_register_if6 = NULL;
661 	}
662 
663 	V_ip6_mrouter = NULL;
664 	V_ip6_mrouter_ver = 0;
665 
666 	MROUTER6_UNLOCK();
667 	MRT6_DLOG(DEBUG_ANY, "finished");
668 
669 	return (0);
670 }
671 
672 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
673 
674 /*
675  * Add a mif to the mif table
676  */
677 static int
678 add_m6if(struct mif6ctl *mifcp)
679 {
680 	struct mif6 *mifp;
681 	struct ifnet *ifp;
682 	int error;
683 
684 	MIF6_LOCK();
685 
686 	if (mifcp->mif6c_mifi >= MAXMIFS) {
687 		MIF6_UNLOCK();
688 		return (EINVAL);
689 	}
690 	mifp = mif6table + mifcp->mif6c_mifi;
691 	if (mifp->m6_ifp != NULL) {
692 		MIF6_UNLOCK();
693 		return (EADDRINUSE); /* XXX: is it appropriate? */
694 	}
695 	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
696 		MIF6_UNLOCK();
697 		return (ENXIO);
698 	}
699 
700 	ifp = ifnet_byindex(mifcp->mif6c_pifi);
701 
702 	if (mifcp->mif6c_flags & MIFF_REGISTER) {
703 		if (reg_mif_num == (mifi_t)-1) {
704 			ifp = if_alloc(IFT_OTHER);
705 
706 			if_initname(ifp, "register_mif", 0);
707 			ifp->if_flags |= IFF_LOOPBACK;
708 			if_attach(ifp);
709 			multicast_register_if6 = ifp;
710 			reg_mif_num = mifcp->mif6c_mifi;
711 			/*
712 			 * it is impossible to guess the ifindex of the
713 			 * register interface.  So mif6c_pifi is automatically
714 			 * calculated.
715 			 */
716 			mifcp->mif6c_pifi = ifp->if_index;
717 		} else {
718 			ifp = multicast_register_if6;
719 		}
720 	} else {
721 		/* Make sure the interface supports multicast */
722 		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
723 			MIF6_UNLOCK();
724 			return (EOPNOTSUPP);
725 		}
726 
727 		error = if_allmulti(ifp, 1);
728 		if (error) {
729 			MIF6_UNLOCK();
730 			return (error);
731 		}
732 	}
733 
734 	mifp->m6_flags     = mifcp->mif6c_flags;
735 	mifp->m6_ifp       = ifp;
736 
737 	/* initialize per mif pkt counters */
738 	mifp->m6_pkt_in    = 0;
739 	mifp->m6_pkt_out   = 0;
740 	mifp->m6_bytes_in  = 0;
741 	mifp->m6_bytes_out = 0;
742 
743 	/* Adjust nummifs up if the mifi is higher than nummifs */
744 	if (nummifs <= mifcp->mif6c_mifi)
745 		nummifs = mifcp->mif6c_mifi + 1;
746 
747 	MIF6_UNLOCK();
748 	MRT6_DLOG(DEBUG_ANY, "mif #%d, phyint %s", mifcp->mif6c_mifi,
749 	    if_name(ifp));
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 	MRT6_DLOG(DEBUG_ANY, "mif %d, nummifs %d", *mifip, nummifs);
793 
794 	return (0);
795 }
796 
797 static int
798 del_m6if(mifi_t *mifip)
799 {
800 	int cc;
801 
802 	MIF6_LOCK();
803 	cc = del_m6if_locked(mifip);
804 	MIF6_UNLOCK();
805 
806 	return (cc);
807 }
808 
809 /*
810  * Add an mfc entry
811  */
812 static int
813 add_m6fc(struct mf6cctl *mfccp)
814 {
815 	struct mf6c *rt;
816 	u_long hash;
817 	struct rtdetq *rte;
818 	u_short nstl;
819 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
820 
821 	MFC6_LOCK();
822 
823 	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
824 		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
825 
826 	/* If an entry already exists, just update the fields */
827 	if (rt) {
828 		MRT6_DLOG(DEBUG_MFC, "no upcall o %s g %s p %x",
829 		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
830 		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
831 		    mfccp->mf6cc_parent);
832 
833 		rt->mf6c_parent = mfccp->mf6cc_parent;
834 		rt->mf6c_ifset = mfccp->mf6cc_ifset;
835 
836 		MFC6_UNLOCK();
837 		return (0);
838 	}
839 
840 	/*
841 	 * Find the entry for which the upcall was made and update
842 	 */
843 	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
844 			mfccp->mf6cc_mcastgrp.sin6_addr);
845 	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
846 		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
847 				       &mfccp->mf6cc_origin.sin6_addr) &&
848 		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
849 				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
850 		    (rt->mf6c_stall != NULL)) {
851 
852 			if (nstl++)
853 				log(LOG_ERR,
854 				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
855 				    "multiple kernel entries",
856 				    ip6_sprintf(ip6bufo,
857 					    &mfccp->mf6cc_origin.sin6_addr),
858 				    ip6_sprintf(ip6bufg,
859 					    &mfccp->mf6cc_mcastgrp.sin6_addr),
860 				    mfccp->mf6cc_parent, rt->mf6c_stall);
861 
862 			MRT6_DLOG(DEBUG_MFC, "o %s g %s p %x dbg %p",
863 			    ip6_sprintf(ip6bufo,
864 			    &mfccp->mf6cc_origin.sin6_addr),
865 			    ip6_sprintf(ip6bufg,
866 				&mfccp->mf6cc_mcastgrp.sin6_addr),
867 			    mfccp->mf6cc_parent, rt->mf6c_stall);
868 
869 			rt->mf6c_origin     = mfccp->mf6cc_origin;
870 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
871 			rt->mf6c_parent     = mfccp->mf6cc_parent;
872 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
873 			/* initialize pkt counters per src-grp */
874 			rt->mf6c_pkt_cnt    = 0;
875 			rt->mf6c_byte_cnt   = 0;
876 			rt->mf6c_wrong_if   = 0;
877 
878 			rt->mf6c_expire = 0;	/* Don't clean this guy up */
879 			n6expire[hash]--;
880 
881 			/* free packets Qed at the end of this entry */
882 			for (rte = rt->mf6c_stall; rte != NULL; ) {
883 				struct rtdetq *n = rte->next;
884 				ip6_mdq(rte->m, rte->ifp, rt);
885 				m_freem(rte->m);
886 #ifdef UPCALL_TIMING
887 				collate(&(rte->t));
888 #endif /* UPCALL_TIMING */
889 				free(rte, M_MRTABLE6);
890 				rte = n;
891 			}
892 			rt->mf6c_stall = NULL;
893 		}
894 	}
895 
896 	/*
897 	 * It is possible that an entry is being inserted without an upcall
898 	 */
899 	if (nstl == 0) {
900 		MRT6_DLOG(DEBUG_MFC, "no upcall h %lu o %s g %s p %x", hash,
901 		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
902 		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
903 		    mfccp->mf6cc_parent);
904 
905 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
906 
907 			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
908 					       &mfccp->mf6cc_origin.sin6_addr)&&
909 			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
910 					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
911 
912 				rt->mf6c_origin     = mfccp->mf6cc_origin;
913 				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
914 				rt->mf6c_parent     = mfccp->mf6cc_parent;
915 				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
916 				/* initialize pkt counters per src-grp */
917 				rt->mf6c_pkt_cnt    = 0;
918 				rt->mf6c_byte_cnt   = 0;
919 				rt->mf6c_wrong_if   = 0;
920 
921 				if (rt->mf6c_expire)
922 					n6expire[hash]--;
923 				rt->mf6c_expire	   = 0;
924 			}
925 		}
926 		if (rt == NULL) {
927 			/* no upcall, so make a new entry */
928 			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
929 						  M_NOWAIT);
930 			if (rt == NULL) {
931 				MFC6_UNLOCK();
932 				return (ENOBUFS);
933 			}
934 
935 			/* insert new entry at head of hash chain */
936 			rt->mf6c_origin     = mfccp->mf6cc_origin;
937 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
938 			rt->mf6c_parent     = mfccp->mf6cc_parent;
939 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
940 			/* initialize pkt counters per src-grp */
941 			rt->mf6c_pkt_cnt    = 0;
942 			rt->mf6c_byte_cnt   = 0;
943 			rt->mf6c_wrong_if   = 0;
944 			rt->mf6c_expire     = 0;
945 			rt->mf6c_stall = NULL;
946 
947 			/* link into table */
948 			rt->mf6c_next  = mf6ctable[hash];
949 			mf6ctable[hash] = rt;
950 		}
951 	}
952 
953 	MFC6_UNLOCK();
954 	return (0);
955 }
956 
957 #ifdef UPCALL_TIMING
958 /*
959  * collect delay statistics on the upcalls
960  */
961 static void
962 collate(struct timeval *t)
963 {
964 	u_long d;
965 	struct timeval tp;
966 	u_long delta;
967 
968 	GET_TIME(tp);
969 
970 	if (TV_LT(*t, tp))
971 	{
972 		TV_DELTA(tp, *t, delta);
973 
974 		d = delta >> 10;
975 		if (d > UPCALL_MAX)
976 			d = UPCALL_MAX;
977 
978 		++upcall_data[d];
979 	}
980 }
981 #endif /* UPCALL_TIMING */
982 
983 /*
984  * Delete an mfc entry
985  */
986 static int
987 del_m6fc(struct mf6cctl *mfccp)
988 {
989 #ifdef MRT6DEBUG
990 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
991 #endif
992 	struct sockaddr_in6	origin;
993 	struct sockaddr_in6	mcastgrp;
994 	struct mf6c		*rt;
995 	struct mf6c		**nptr;
996 	u_long		hash;
997 
998 	origin = mfccp->mf6cc_origin;
999 	mcastgrp = mfccp->mf6cc_mcastgrp;
1000 	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
1001 
1002 	MRT6_DLOG(DEBUG_MFC, "orig %s mcastgrp %s",
1003 	    ip6_sprintf(ip6bufo, &origin.sin6_addr),
1004 	    ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
1005 
1006 	MFC6_LOCK();
1007 
1008 	nptr = &mf6ctable[hash];
1009 	while ((rt = *nptr) != NULL) {
1010 		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
1011 				       &rt->mf6c_origin.sin6_addr) &&
1012 		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1013 				       &rt->mf6c_mcastgrp.sin6_addr) &&
1014 		    rt->mf6c_stall == NULL)
1015 			break;
1016 
1017 		nptr = &rt->mf6c_next;
1018 	}
1019 	if (rt == NULL) {
1020 		MFC6_UNLOCK();
1021 		return (EADDRNOTAVAIL);
1022 	}
1023 
1024 	*nptr = rt->mf6c_next;
1025 	free(rt, M_MRTABLE6);
1026 
1027 	MFC6_UNLOCK();
1028 
1029 	return (0);
1030 }
1031 
1032 static int
1033 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1034 {
1035 
1036 	if (s) {
1037 		if (sbappendaddr(&s->so_rcv,
1038 				 (struct sockaddr *)src,
1039 				 mm, (struct mbuf *)0) != 0) {
1040 			sorwakeup(s);
1041 			return (0);
1042 		}
1043 	}
1044 	m_freem(mm);
1045 	return (-1);
1046 }
1047 
1048 /*
1049  * IPv6 multicast forwarding function. This function assumes that the packet
1050  * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1051  * pointed to by "ifp", and the packet is to be relayed to other networks
1052  * that have members of the packet's destination IPv6 multicast group.
1053  *
1054  * The packet is returned unscathed to the caller, unless it is
1055  * erroneous, in which case a non-zero return value tells the caller to
1056  * discard it.
1057  *
1058  * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1059  * this function is called in the originating context (i.e., not when
1060  * forwarding a packet from other node).  ip6_output(), which is currently the
1061  * only function that calls this function is called in the originating context,
1062  * explicitly ensures this condition.  It is caller's responsibility to ensure
1063  * that if this function is called from somewhere else in the originating
1064  * context in the future.
1065  */
1066 int
1067 X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1068 {
1069 	struct rtdetq *rte;
1070 	struct mbuf *mb0;
1071 	struct mf6c *rt;
1072 	struct mif6 *mifp;
1073 	struct mbuf *mm;
1074 	u_long hash;
1075 	mifi_t mifi;
1076 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1077 #ifdef UPCALL_TIMING
1078 	struct timeval tp;
1079 
1080 	GET_TIME(tp);
1081 #endif /* UPCALL_TIMING */
1082 
1083 	MRT6_DLOG(DEBUG_FORWARD, "src %s, dst %s, ifindex %d",
1084 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1085 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst), ifp->if_index);
1086 
1087 	/*
1088 	 * Don't forward a packet with Hop limit of zero or one,
1089 	 * or a packet destined to a local-only group.
1090 	 */
1091 	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1092 	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1093 		return (0);
1094 	ip6->ip6_hlim--;
1095 
1096 	/*
1097 	 * Source address check: do not forward packets with unspecified
1098 	 * source. It was discussed in July 2000, on ipngwg mailing list.
1099 	 * This is rather more serious than unicast cases, because some
1100 	 * MLD packets can be sent with the unspecified source address
1101 	 * (although such packets must normally set 1 to the hop limit field).
1102 	 */
1103 	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1104 		IP6STAT_INC(ip6s_cantforward);
1105 		if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
1106 			V_ip6_log_time = time_uptime;
1107 			log(LOG_DEBUG,
1108 			    "cannot forward "
1109 			    "from %s to %s nxt %d received on %s\n",
1110 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1111 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1112 			    ip6->ip6_nxt,
1113 			    if_name(m->m_pkthdr.rcvif));
1114 		}
1115 		return (0);
1116 	}
1117 
1118 	MFC6_LOCK();
1119 
1120 	/*
1121 	 * Determine forwarding mifs from the forwarding cache table
1122 	 */
1123 	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1124 	MRT6STAT_INC(mrt6s_mfc_lookups);
1125 
1126 	/* Entry exists, so forward if necessary */
1127 	if (rt) {
1128 		MFC6_UNLOCK();
1129 		return (ip6_mdq(m, ifp, rt));
1130 	}
1131 
1132 	/*
1133 	 * If we don't have a route for packet's origin,
1134 	 * Make a copy of the packet & send message to routing daemon.
1135 	 */
1136 	MRT6STAT_INC(mrt6s_no_route);
1137 	MRT6_DLOG(DEBUG_FORWARD | DEBUG_MFC, "no rte s %s g %s",
1138 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1139 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1140 
1141 	/*
1142 	 * Allocate mbufs early so that we don't do extra work if we
1143 	 * are just going to fail anyway.
1144 	 */
1145 	rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6, M_NOWAIT);
1146 	if (rte == NULL) {
1147 		MFC6_UNLOCK();
1148 		return (ENOBUFS);
1149 	}
1150 	mb0 = m_copym(m, 0, M_COPYALL, M_NOWAIT);
1151 	/*
1152 	 * Pullup packet header if needed before storing it,
1153 	 * as other references may modify it in the meantime.
1154 	 */
1155 	if (mb0 && (!M_WRITABLE(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1156 		mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1157 	if (mb0 == NULL) {
1158 		free(rte, M_MRTABLE6);
1159 		MFC6_UNLOCK();
1160 		return (ENOBUFS);
1161 	}
1162 
1163 	/* is there an upcall waiting for this packet? */
1164 	hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1165 	for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1166 		if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1167 		    &rt->mf6c_origin.sin6_addr) &&
1168 		    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1169 		    &rt->mf6c_mcastgrp.sin6_addr) && (rt->mf6c_stall != NULL))
1170 			break;
1171 	}
1172 
1173 	if (rt == NULL) {
1174 		struct mrt6msg *im;
1175 #ifdef MRT6_OINIT
1176 		struct omrt6msg *oim;
1177 #endif
1178 		/* no upcall, so make a new entry */
1179 		rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, M_NOWAIT);
1180 		if (rt == NULL) {
1181 			free(rte, M_MRTABLE6);
1182 			m_freem(mb0);
1183 			MFC6_UNLOCK();
1184 			return (ENOBUFS);
1185 		}
1186 		/*
1187 		 * Make a copy of the header to send to the user
1188 		 * level process
1189 		 */
1190 		mm = m_copym(mb0, 0, sizeof(struct ip6_hdr), M_NOWAIT);
1191 		if (mm == NULL) {
1192 			free(rte, M_MRTABLE6);
1193 			m_freem(mb0);
1194 			free(rt, M_MRTABLE6);
1195 			MFC6_UNLOCK();
1196 			return (ENOBUFS);
1197 		}
1198 
1199 		/*
1200 		 * Send message to routing daemon
1201 		 */
1202 		sin6.sin6_addr = ip6->ip6_src;
1203 		im = NULL;
1204 #ifdef MRT6_OINIT
1205 		oim = NULL;
1206 #endif
1207 		switch (V_ip6_mrouter_ver) {
1208 #ifdef MRT6_OINIT
1209 		case MRT6_OINIT:
1210 			oim = mtod(mm, struct omrt6msg *);
1211 			oim->im6_msgtype = MRT6MSG_NOCACHE;
1212 			oim->im6_mbz = 0;
1213 			break;
1214 #endif
1215 		case MRT6_INIT:
1216 			im = mtod(mm, struct mrt6msg *);
1217 			im->im6_msgtype = MRT6MSG_NOCACHE;
1218 			im->im6_mbz = 0;
1219 			break;
1220 		default:
1221 			free(rte, M_MRTABLE6);
1222 			m_freem(mb0);
1223 			free(rt, M_MRTABLE6);
1224 			MFC6_UNLOCK();
1225 			return (EINVAL);
1226 		}
1227 
1228 		MRT6_DLOG(DEBUG_FORWARD, "getting the iif info in the kernel");
1229 		for (mifp = mif6table, mifi = 0;
1230 		    mifi < nummifs && mifp->m6_ifp != ifp; mifp++, mifi++)
1231 				;
1232 
1233 		switch (V_ip6_mrouter_ver) {
1234 #ifdef MRT6_OINIT
1235 		case MRT6_OINIT:
1236 			oim->im6_mif = mifi;
1237 			break;
1238 #endif
1239 		case MRT6_INIT:
1240 			im->im6_mif = mifi;
1241 			break;
1242 		}
1243 
1244 		if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1245 			log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1246 			    "socket queue full\n");
1247 			MRT6STAT_INC(mrt6s_upq_sockfull);
1248 			free(rte, M_MRTABLE6);
1249 			m_freem(mb0);
1250 			free(rt, M_MRTABLE6);
1251 			MFC6_UNLOCK();
1252 			return (ENOBUFS);
1253 		}
1254 
1255 		MRT6STAT_INC(mrt6s_upcalls);
1256 
1257 		/* insert new entry at head of hash chain */
1258 		bzero(rt, sizeof(*rt));
1259 		rt->mf6c_origin.sin6_family = AF_INET6;
1260 		rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1261 		rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1262 		rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1263 		rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1264 		rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1265 		rt->mf6c_expire = UPCALL_EXPIRE;
1266 		n6expire[hash]++;
1267 		rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1268 
1269 		/* link into table */
1270 		rt->mf6c_next  = mf6ctable[hash];
1271 		mf6ctable[hash] = rt;
1272 		/* Add this entry to the end of the queue */
1273 		rt->mf6c_stall = rte;
1274 	} else {
1275 		/* determine if q has overflowed */
1276 		struct rtdetq **p;
1277 		int npkts = 0;
1278 
1279 		for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1280 			if (++npkts > MAX_UPQ6) {
1281 				MRT6STAT_INC(mrt6s_upq_ovflw);
1282 				free(rte, M_MRTABLE6);
1283 				m_freem(mb0);
1284 				MFC6_UNLOCK();
1285 				return (0);
1286 			}
1287 
1288 		/* Add this entry to the end of the queue */
1289 		*p = rte;
1290 	}
1291 
1292 	rte->next = NULL;
1293 	rte->m = mb0;
1294 	rte->ifp = ifp;
1295 #ifdef UPCALL_TIMING
1296 	rte->t = tp;
1297 #endif /* UPCALL_TIMING */
1298 
1299 	MFC6_UNLOCK();
1300 
1301 	return (0);
1302 }
1303 
1304 /*
1305  * Clean up cache entries if upcalls are not serviced
1306  * Call from the Slow Timeout mechanism, every half second.
1307  */
1308 static void
1309 expire_upcalls(void *unused)
1310 {
1311 #ifdef MRT6DEBUG
1312 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1313 #endif
1314 	struct rtdetq *rte;
1315 	struct mf6c *mfc, **nptr;
1316 	u_long i;
1317 
1318 	MFC6_LOCK();
1319 	for (i = 0; i < MF6CTBLSIZ; i++) {
1320 		if (n6expire[i] == 0)
1321 			continue;
1322 		nptr = &mf6ctable[i];
1323 		while ((mfc = *nptr) != NULL) {
1324 			rte = mfc->mf6c_stall;
1325 			/*
1326 			 * Skip real cache entries
1327 			 * Make sure it wasn't marked to not expire (shouldn't happen)
1328 			 * If it expires now
1329 			 */
1330 			if (rte != NULL &&
1331 			    mfc->mf6c_expire != 0 &&
1332 			    --mfc->mf6c_expire == 0) {
1333 				MRT6_DLOG(DEBUG_EXPIRE, "expiring (%s %s)",
1334 				    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1335 				    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1336 				/*
1337 				 * drop all the packets
1338 				 * free the mbuf with the pkt, if, timing info
1339 				 */
1340 				do {
1341 					struct rtdetq *n = rte->next;
1342 					m_freem(rte->m);
1343 					free(rte, M_MRTABLE6);
1344 					rte = n;
1345 				} while (rte != NULL);
1346 				MRT6STAT_INC(mrt6s_cache_cleanups);
1347 				n6expire[i]--;
1348 
1349 				*nptr = mfc->mf6c_next;
1350 				free(mfc, M_MRTABLE6);
1351 			} else {
1352 				nptr = &mfc->mf6c_next;
1353 			}
1354 		}
1355 	}
1356 	MFC6_UNLOCK();
1357 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1358 	    expire_upcalls, NULL);
1359 }
1360 
1361 /*
1362  * Packet forwarding routine once entry in the cache is made
1363  */
1364 static int
1365 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1366 {
1367 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1368 	mifi_t mifi, iif;
1369 	struct mif6 *mifp;
1370 	int plen = m->m_pkthdr.len;
1371 	struct in6_addr src0, dst0; /* copies for local work */
1372 	u_int32_t iszone, idzone, oszone, odzone;
1373 	int error = 0;
1374 
1375 /*
1376  * Macro to send packet on mif.  Since RSVP packets don't get counted on
1377  * input, they shouldn't get counted on output, so statistics keeping is
1378  * separate.
1379  */
1380 
1381 #define MC6_SEND(ip6, mifp, m) do {				\
1382 	if ((mifp)->m6_flags & MIFF_REGISTER)			\
1383 		register_send((ip6), (mifp), (m));		\
1384 	else							\
1385 		phyint_send((ip6), (mifp), (m));		\
1386 } while (/*CONSTCOND*/ 0)
1387 
1388 	/*
1389 	 * Don't forward if it didn't arrive from the parent mif
1390 	 * for its origin.
1391 	 */
1392 	mifi = rt->mf6c_parent;
1393 	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1394 		/* came in the wrong interface */
1395 		MRT6_DLOG(DEBUG_FORWARD,
1396 		    "wrong if: ifid %d mifi %d mififid %x", ifp->if_index,
1397 		    mifi, mif6table[mifi].m6_ifp->if_index);
1398 		MRT6STAT_INC(mrt6s_wrong_if);
1399 		rt->mf6c_wrong_if++;
1400 		/*
1401 		 * If we are doing PIM processing, and we are forwarding
1402 		 * packets on this interface, send a message to the
1403 		 * routing daemon.
1404 		 */
1405 		/* have to make sure this is a valid mif */
1406 		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1407 			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1408 				/*
1409 				 * Check the M_LOOP flag to avoid an
1410 				 * unnecessary PIM assert.
1411 				 * XXX: M_LOOP is an ad-hoc hack...
1412 				 */
1413 				static struct sockaddr_in6 sin6 =
1414 				{ sizeof(sin6), AF_INET6 };
1415 
1416 				struct mbuf *mm;
1417 				struct mrt6msg *im;
1418 #ifdef MRT6_OINIT
1419 				struct omrt6msg *oim;
1420 #endif
1421 
1422 				mm = m_copym(m, 0, sizeof(struct ip6_hdr),
1423 				    M_NOWAIT);
1424 				if (mm &&
1425 				    (!M_WRITABLE(mm) ||
1426 				     mm->m_len < sizeof(struct ip6_hdr)))
1427 					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1428 				if (mm == NULL)
1429 					return (ENOBUFS);
1430 
1431 #ifdef MRT6_OINIT
1432 				oim = NULL;
1433 #endif
1434 				im = NULL;
1435 				switch (V_ip6_mrouter_ver) {
1436 #ifdef MRT6_OINIT
1437 				case MRT6_OINIT:
1438 					oim = mtod(mm, struct omrt6msg *);
1439 					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1440 					oim->im6_mbz = 0;
1441 					break;
1442 #endif
1443 				case MRT6_INIT:
1444 					im = mtod(mm, struct mrt6msg *);
1445 					im->im6_msgtype = MRT6MSG_WRONGMIF;
1446 					im->im6_mbz = 0;
1447 					break;
1448 				default:
1449 					m_freem(mm);
1450 					return (EINVAL);
1451 				}
1452 
1453 				for (mifp = mif6table, iif = 0;
1454 				     iif < nummifs && mifp &&
1455 					     mifp->m6_ifp != ifp;
1456 				     mifp++, iif++)
1457 					;
1458 
1459 				switch (V_ip6_mrouter_ver) {
1460 #ifdef MRT6_OINIT
1461 				case MRT6_OINIT:
1462 					oim->im6_mif = iif;
1463 					sin6.sin6_addr = oim->im6_src;
1464 					break;
1465 #endif
1466 				case MRT6_INIT:
1467 					im->im6_mif = iif;
1468 					sin6.sin6_addr = im->im6_src;
1469 					break;
1470 				}
1471 
1472 				MRT6STAT_INC(mrt6s_upcalls);
1473 
1474 				if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1475 					MRT6_DLOG(DEBUG_ANY,
1476 					    "ip6_mrouter socket queue full");
1477 					MRT6STAT_INC(mrt6s_upq_sockfull);
1478 					return (ENOBUFS);
1479 				}	/* if socket Q full */
1480 			}		/* if PIM */
1481 		return (0);
1482 	}			/* if wrong iif */
1483 
1484 	/* If I sourced this packet, it counts as output, else it was input. */
1485 	if (m->m_pkthdr.rcvif == NULL) {
1486 		/* XXX: is rcvif really NULL when output?? */
1487 		mif6table[mifi].m6_pkt_out++;
1488 		mif6table[mifi].m6_bytes_out += plen;
1489 	} else {
1490 		mif6table[mifi].m6_pkt_in++;
1491 		mif6table[mifi].m6_bytes_in += plen;
1492 	}
1493 	rt->mf6c_pkt_cnt++;
1494 	rt->mf6c_byte_cnt += plen;
1495 
1496 	/*
1497 	 * For each mif, forward a copy of the packet if there are group
1498 	 * members downstream on the interface.
1499 	 */
1500 	src0 = ip6->ip6_src;
1501 	dst0 = ip6->ip6_dst;
1502 	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1503 	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1504 		IP6STAT_INC(ip6s_badscope);
1505 		return (error);
1506 	}
1507 	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1508 		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1509 			/*
1510 			 * check if the outgoing packet is going to break
1511 			 * a scope boundary.
1512 			 * XXX For packets through PIM register tunnel
1513 			 * interface, we believe a routing daemon.
1514 			 */
1515 			if (!(mif6table[rt->mf6c_parent].m6_flags &
1516 			      MIFF_REGISTER) &&
1517 			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1518 				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1519 				    &oszone) ||
1520 				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1521 				    &odzone) ||
1522 				    iszone != oszone ||
1523 				    idzone != odzone) {
1524 					IP6STAT_INC(ip6s_badscope);
1525 					continue;
1526 				}
1527 			}
1528 
1529 			mifp->m6_pkt_out++;
1530 			mifp->m6_bytes_out += plen;
1531 			MC6_SEND(ip6, mifp, m);
1532 		}
1533 	}
1534 	return (0);
1535 }
1536 
1537 static void
1538 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1539 {
1540 #ifdef MRT6DEBUG
1541 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1542 #endif
1543 	struct mbuf *mb_copy;
1544 	struct ifnet *ifp = mifp->m6_ifp;
1545 	int error = 0;
1546 	u_long linkmtu;
1547 
1548 	/*
1549 	 * Make a new reference to the packet; make sure that
1550 	 * the IPv6 header is actually copied, not just referenced,
1551 	 * so that ip6_output() only scribbles on the copy.
1552 	 */
1553 	mb_copy = m_copym(m, 0, M_COPYALL, M_NOWAIT);
1554 	if (mb_copy &&
1555 	    (!M_WRITABLE(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1556 		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1557 	if (mb_copy == NULL) {
1558 		return;
1559 	}
1560 	/* set MCAST flag to the outgoing packet */
1561 	mb_copy->m_flags |= M_MCAST;
1562 
1563 	/*
1564 	 * If we sourced the packet, call ip6_output since we may devide
1565 	 * the packet into fragments when the packet is too big for the
1566 	 * outgoing interface.
1567 	 * Otherwise, we can simply send the packet to the interface
1568 	 * sending queue.
1569 	 */
1570 	if (m->m_pkthdr.rcvif == NULL) {
1571 		struct ip6_moptions im6o;
1572 
1573 		im6o.im6o_multicast_ifp = ifp;
1574 		/* XXX: ip6_output will override ip6->ip6_hlim */
1575 		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1576 		im6o.im6o_multicast_loop = 1;
1577 		error = ip6_output(mb_copy, NULL, NULL, IPV6_FORWARDING, &im6o,
1578 		    NULL, NULL);
1579 
1580 		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1581 		    (uint16_t)(mifp - mif6table), error);
1582 		return;
1583 	}
1584 
1585 	/*
1586 	 * If configured to loop back multicasts by default,
1587 	 * loop back a copy now.
1588 	 */
1589 	if (in6_mcast_loop)
1590 		ip6_mloopback(ifp, m);
1591 
1592 	/*
1593 	 * Put the packet into the sending queue of the outgoing interface
1594 	 * if it would fit in the MTU of the interface.
1595 	 */
1596 	linkmtu = IN6_LINKMTU(ifp);
1597 	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1598 		struct sockaddr_in6 dst6;
1599 
1600 		bzero(&dst6, sizeof(dst6));
1601 		dst6.sin6_len = sizeof(struct sockaddr_in6);
1602 		dst6.sin6_family = AF_INET6;
1603 		dst6.sin6_addr = ip6->ip6_dst;
1604 
1605 		IP_PROBE(send, NULL, NULL, ip6, ifp, NULL, ip6);
1606 		/*
1607 		 * We just call if_output instead of nd6_output here, since
1608 		 * we need no ND for a multicast forwarded packet...right?
1609 		 */
1610 		m_clrprotoflags(m);	/* Avoid confusing lower layers. */
1611 		error = (*ifp->if_output)(ifp, mb_copy,
1612 		    (struct sockaddr *)&dst6, NULL);
1613 		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1614 		    (uint16_t)(mifp - mif6table), error);
1615 	} else {
1616 		/*
1617 		 * pMTU discovery is intentionally disabled by default, since
1618 		 * various router may notify pMTU in multicast, which can be
1619 		 * a DDoS to a router
1620 		 */
1621 		if (V_ip6_mcast_pmtu)
1622 			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1623 		else {
1624 			MRT6_DLOG(DEBUG_XMIT, " packet too big on %s o %s "
1625 			    "g %s size %d (discarded)", if_name(ifp),
1626 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1627 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1628 			    mb_copy->m_pkthdr.len);
1629 			m_freem(mb_copy); /* simply discard the packet */
1630 		}
1631 	}
1632 }
1633 
1634 static int
1635 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1636 {
1637 #ifdef MRT6DEBUG
1638 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1639 #endif
1640 	struct mbuf *mm;
1641 	int i, len = m->m_pkthdr.len;
1642 	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1643 	struct mrt6msg *im6;
1644 
1645 	MRT6_DLOG(DEBUG_ANY, "src %s dst %s",
1646 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1647 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1648 	PIM6STAT_INC(pim6s_snd_registers);
1649 
1650 	/* Make a copy of the packet to send to the user level process. */
1651 	mm = m_gethdr(M_NOWAIT, MT_DATA);
1652 	if (mm == NULL)
1653 		return (ENOBUFS);
1654 	mm->m_data += max_linkhdr;
1655 	mm->m_len = sizeof(struct ip6_hdr);
1656 
1657 	if ((mm->m_next = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL) {
1658 		m_freem(mm);
1659 		return (ENOBUFS);
1660 	}
1661 	i = MHLEN - M_LEADINGSPACE(mm);
1662 	if (i > len)
1663 		i = len;
1664 	mm = m_pullup(mm, i);
1665 	if (mm == NULL)
1666 		return (ENOBUFS);
1667 /* TODO: check it! */
1668 	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1669 
1670 	/*
1671 	 * Send message to routing daemon
1672 	 */
1673 	sin6.sin6_addr = ip6->ip6_src;
1674 
1675 	im6 = mtod(mm, struct mrt6msg *);
1676 	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1677 	im6->im6_mbz          = 0;
1678 
1679 	im6->im6_mif = mif - mif6table;
1680 
1681 	/* iif info is not given for reg. encap.n */
1682 	MRT6STAT_INC(mrt6s_upcalls);
1683 
1684 	if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1685 		MRT6_DLOG(DEBUG_ANY, "ip6_mrouter socket queue full");
1686 		MRT6STAT_INC(mrt6s_upq_sockfull);
1687 		return (ENOBUFS);
1688 	}
1689 	return (0);
1690 }
1691 
1692 /*
1693  * pim6_encapcheck() is called by the encap6_input() path at runtime to
1694  * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1695  * into the kernel.
1696  */
1697 static int
1698 pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
1699 {
1700 
1701 #ifdef DIAGNOSTIC
1702     KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1703 #endif
1704     if (proto != IPPROTO_PIM)
1705 	return 0;	/* not for us; reject the datagram. */
1706 
1707     return 64;		/* claim the datagram. */
1708 }
1709 
1710 /*
1711  * PIM sparse mode hook
1712  * Receives the pim control messages, and passes them up to the listening
1713  * socket, using rip6_input.
1714  * The only message processed is the REGISTER pim message; the pim header
1715  * is stripped off, and the inner packet is passed to register_mforward.
1716  */
1717 int
1718 pim6_input(struct mbuf **mp, int *offp, int proto)
1719 {
1720 	struct pim *pim; /* pointer to a pim struct */
1721 	struct ip6_hdr *ip6;
1722 	int pimlen;
1723 	struct mbuf *m = *mp;
1724 	int minlen;
1725 	int off = *offp;
1726 
1727 	PIM6STAT_INC(pim6s_rcv_total);
1728 
1729 	ip6 = mtod(m, struct ip6_hdr *);
1730 	pimlen = m->m_pkthdr.len - *offp;
1731 
1732 	/*
1733 	 * Validate lengths
1734 	 */
1735 	if (pimlen < PIM_MINLEN) {
1736 		PIM6STAT_INC(pim6s_rcv_tooshort);
1737 		MRT6_DLOG(DEBUG_PIM, "PIM packet too short");
1738 		m_freem(m);
1739 		return (IPPROTO_DONE);
1740 	}
1741 
1742 	/*
1743 	 * if the packet is at least as big as a REGISTER, go ahead
1744 	 * and grab the PIM REGISTER header size, to avoid another
1745 	 * possible m_pullup() later.
1746 	 *
1747 	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1748 	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1749 	 */
1750 	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1751 
1752 	/*
1753 	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1754 	 * possibly the PIM REGISTER header
1755 	 */
1756 #ifndef PULLDOWN_TEST
1757 	IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1758 	/* adjust pointer */
1759 	ip6 = mtod(m, struct ip6_hdr *);
1760 
1761 	/* adjust mbuf to point to the PIM header */
1762 	pim = (struct pim *)((caddr_t)ip6 + off);
1763 #else
1764 	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1765 	if (pim == NULL) {
1766 		PIM6STAT_INC(pim6s_rcv_tooshort);
1767 		return (IPPROTO_DONE);
1768 	}
1769 #endif
1770 
1771 #define PIM6_CHECKSUM
1772 #ifdef PIM6_CHECKSUM
1773 	{
1774 		int cksumlen;
1775 
1776 		/*
1777 		 * Validate checksum.
1778 		 * If PIM REGISTER, exclude the data packet
1779 		 */
1780 		if (pim->pim_type == PIM_REGISTER)
1781 			cksumlen = PIM_MINLEN;
1782 		else
1783 			cksumlen = pimlen;
1784 
1785 		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1786 			PIM6STAT_INC(pim6s_rcv_badsum);
1787 			MRT6_DLOG(DEBUG_PIM, "invalid checksum");
1788 			m_freem(m);
1789 			return (IPPROTO_DONE);
1790 		}
1791 	}
1792 #endif /* PIM_CHECKSUM */
1793 
1794 	/* PIM version check */
1795 	if (pim->pim_ver != PIM_VERSION) {
1796 		PIM6STAT_INC(pim6s_rcv_badversion);
1797 		MRT6_DLOG(DEBUG_ANY | DEBUG_ERR,
1798 		    "incorrect version %d, expecting %d",
1799 		    pim->pim_ver, PIM_VERSION);
1800 		m_freem(m);
1801 		return (IPPROTO_DONE);
1802 	}
1803 
1804 	if (pim->pim_type == PIM_REGISTER) {
1805 		/*
1806 		 * since this is a REGISTER, we'll make a copy of the register
1807 		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1808 		 * routing daemon.
1809 		 */
1810 		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1811 
1812 		struct mbuf *mcp;
1813 		struct ip6_hdr *eip6;
1814 		u_int32_t *reghdr;
1815 		int rc;
1816 #ifdef MRT6DEBUG
1817 		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1818 #endif
1819 
1820 		PIM6STAT_INC(pim6s_rcv_registers);
1821 
1822 		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1823 			MRT6_DLOG(DEBUG_PIM, "register mif not set: %d",
1824 			    reg_mif_num);
1825 			m_freem(m);
1826 			return (IPPROTO_DONE);
1827 		}
1828 
1829 		reghdr = (u_int32_t *)(pim + 1);
1830 
1831 		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1832 			goto pim6_input_to_daemon;
1833 
1834 		/*
1835 		 * Validate length
1836 		 */
1837 		if (pimlen < PIM6_REG_MINLEN) {
1838 			PIM6STAT_INC(pim6s_rcv_tooshort);
1839 			PIM6STAT_INC(pim6s_rcv_badregisters);
1840 			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "register packet "
1841 			    "size too small %d from %s",
1842 			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1843 			m_freem(m);
1844 			return (IPPROTO_DONE);
1845 		}
1846 
1847 		eip6 = (struct ip6_hdr *) (reghdr + 1);
1848 		MRT6_DLOG(DEBUG_PIM, "eip6: %s -> %s, eip6 plen %d",
1849 		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1850 		    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1851 		    ntohs(eip6->ip6_plen));
1852 
1853 		/* verify the version number of the inner packet */
1854 		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1855 			PIM6STAT_INC(pim6s_rcv_badregisters);
1856 			MRT6_DLOG(DEBUG_ANY, "invalid IP version (%d) "
1857 			    "of the inner packet",
1858 			    (eip6->ip6_vfc & IPV6_VERSION));
1859 			m_freem(m);
1860 			return (IPPROTO_DONE);
1861 		}
1862 
1863 		/* verify the inner packet is destined to a mcast group */
1864 		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1865 			PIM6STAT_INC(pim6s_rcv_badregisters);
1866 			MRT6_DLOG(DEBUG_PIM, "inner packet of register "
1867 			    "is not multicast %s",
1868 			    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1869 			m_freem(m);
1870 			return (IPPROTO_DONE);
1871 		}
1872 
1873 		/*
1874 		 * make a copy of the whole header to pass to the daemon later.
1875 		 */
1876 		mcp = m_copym(m, 0, off + PIM6_REG_MINLEN, M_NOWAIT);
1877 		if (mcp == NULL) {
1878 			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "pim register: "
1879 			    "could not copy register head");
1880 			m_freem(m);
1881 			return (IPPROTO_DONE);
1882 		}
1883 
1884 		/*
1885 		 * forward the inner ip6 packet; point m_data at the inner ip6.
1886 		 */
1887 		m_adj(m, off + PIM_MINLEN);
1888 		MRT6_DLOG(DEBUG_PIM, "forwarding decapsulated register: "
1889 		    "src %s, dst %s, mif %d",
1890 		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1891 		    ip6_sprintf(ip6bufd, &eip6->ip6_dst), reg_mif_num);
1892 
1893 		rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1894 				dst.sin6_family, 0);
1895 
1896 		/* prepare the register head to send to the mrouting daemon */
1897 		m = mcp;
1898 	}
1899 
1900 	/*
1901 	 * Pass the PIM message up to the daemon; if it is a register message
1902 	 * pass the 'head' only up to the daemon. This includes the
1903 	 * encapsulator ip6 header, pim header, register header and the
1904 	 * encapsulated ip6 header.
1905 	 */
1906   pim6_input_to_daemon:
1907 	rip6_input(&m, offp, proto);
1908 	return (IPPROTO_DONE);
1909 }
1910 
1911 static int
1912 ip6_mroute_modevent(module_t mod, int type, void *unused)
1913 {
1914 
1915 	switch (type) {
1916 	case MOD_LOAD:
1917 		MROUTER6_LOCK_INIT();
1918 		MFC6_LOCK_INIT();
1919 		MIF6_LOCK_INIT();
1920 
1921 		pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM,
1922 			pim6_encapcheck,
1923 			(const struct protosw *)&in6_pim_protosw, NULL);
1924 		if (pim6_encap_cookie == NULL) {
1925 			printf("ip6_mroute: unable to attach pim6 encap\n");
1926 			MIF6_LOCK_DESTROY();
1927 			MFC6_LOCK_DESTROY();
1928 			MROUTER6_LOCK_DESTROY();
1929 			return (EINVAL);
1930 		}
1931 
1932 		ip6_mforward = X_ip6_mforward;
1933 		ip6_mrouter_done = X_ip6_mrouter_done;
1934 		ip6_mrouter_get = X_ip6_mrouter_get;
1935 		ip6_mrouter_set = X_ip6_mrouter_set;
1936 		mrt6_ioctl = X_mrt6_ioctl;
1937 		break;
1938 
1939 	case MOD_UNLOAD:
1940 		if (V_ip6_mrouter != NULL)
1941 			return EINVAL;
1942 
1943 		if (pim6_encap_cookie) {
1944 			encap_detach(pim6_encap_cookie);
1945 			pim6_encap_cookie = NULL;
1946 		}
1947 		X_ip6_mrouter_done();
1948 		ip6_mforward = NULL;
1949 		ip6_mrouter_done = NULL;
1950 		ip6_mrouter_get = NULL;
1951 		ip6_mrouter_set = NULL;
1952 		mrt6_ioctl = NULL;
1953 
1954 		MIF6_LOCK_DESTROY();
1955 		MFC6_LOCK_DESTROY();
1956 		MROUTER6_LOCK_DESTROY();
1957 		break;
1958 
1959 	default:
1960 		return (EOPNOTSUPP);
1961 	}
1962 
1963 	return (0);
1964 }
1965 
1966 static moduledata_t ip6_mroutemod = {
1967 	"ip6_mroute",
1968 	ip6_mroute_modevent,
1969 	0
1970 };
1971 
1972 DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PROTO_MC, SI_ORDER_ANY);
1973