xref: /freebsd/sys/netinet6/ip6_mroute.c (revision 273c26a3c3bea87a241d6879abd4f991db180bf0)
1 /*-
2  * Copyright (C) 1998 WIDE Project.
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. Neither the name of the project nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	$KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $
30  */
31 
32 /*-
33  * Copyright (c) 1989 Stephen Deering
34  * Copyright (c) 1992, 1993
35  *      The Regents of the University of California.  All rights reserved.
36  *
37  * This code is derived from software contributed to Berkeley by
38  * Stephen Deering of Stanford University.
39  *
40  * Redistribution and use in source and binary forms, with or without
41  * modification, are permitted provided that the following conditions
42  * are met:
43  * 1. Redistributions of source code must retain the above copyright
44  *    notice, this list of conditions and the following disclaimer.
45  * 2. Redistributions in binary form must reproduce the above copyright
46  *    notice, this list of conditions and the following disclaimer in the
47  *    documentation and/or other materials provided with the distribution.
48  * 4. Neither the name of the University nor the names of its contributors
49  *    may be used to endorse or promote products derived from this software
50  *    without specific prior written permission.
51  *
52  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62  * SUCH DAMAGE.
63  *
64  *	@(#)ip_mroute.c	8.2 (Berkeley) 11/15/93
65  *	BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp
66  */
67 
68 /*
69  * IP multicast forwarding procedures
70  *
71  * Written by David Waitzman, BBN Labs, August 1988.
72  * Modified by Steve Deering, Stanford, February 1989.
73  * Modified by Mark J. Steiglitz, Stanford, May, 1991
74  * Modified by Van Jacobson, LBL, January 1993
75  * Modified by Ajit Thyagarajan, PARC, August 1993
76  * Modified by Bill Fenner, PARC, April 1994
77  *
78  * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
79  */
80 
81 #include <sys/cdefs.h>
82 __FBSDID("$FreeBSD$");
83 
84 #include "opt_inet6.h"
85 
86 #include <sys/param.h>
87 #include <sys/callout.h>
88 #include <sys/errno.h>
89 #include <sys/kernel.h>
90 #include <sys/lock.h>
91 #include <sys/malloc.h>
92 #include <sys/mbuf.h>
93 #include <sys/module.h>
94 #include <sys/domain.h>
95 #include <sys/protosw.h>
96 #include <sys/sdt.h>
97 #include <sys/signalvar.h>
98 #include <sys/socket.h>
99 #include <sys/socketvar.h>
100 #include <sys/sockio.h>
101 #include <sys/sx.h>
102 #include <sys/sysctl.h>
103 #include <sys/syslog.h>
104 #include <sys/systm.h>
105 #include <sys/time.h>
106 
107 #include <net/if.h>
108 #include <net/if_var.h>
109 #include <net/if_types.h>
110 #include <net/raw_cb.h>
111 #include <net/vnet.h>
112 
113 #include <netinet/in.h>
114 #include <netinet/in_var.h>
115 #include <netinet/icmp6.h>
116 #include <netinet/ip_encap.h>
117 
118 #include <netinet/ip6.h>
119 #include <netinet/in_kdtrace.h>
120 #include <netinet6/ip6_var.h>
121 #include <netinet6/scope6_var.h>
122 #include <netinet6/nd6.h>
123 #include <netinet6/ip6_mroute.h>
124 #include <netinet6/pim6.h>
125 #include <netinet6/pim6_var.h>
126 
127 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
128 
129 static int	ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
130 static void	phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
131 static int	register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
132 static int	set_pim6(int *);
133 static int	socket_send(struct socket *, struct mbuf *,
134 		    struct sockaddr_in6 *);
135 
136 extern int in6_mcast_loop;
137 extern struct domain inet6domain;
138 
139 static const struct encaptab *pim6_encap_cookie;
140 static const struct protosw in6_pim_protosw = {
141 	.pr_type =		SOCK_RAW,
142 	.pr_domain =		&inet6domain,
143 	.pr_protocol =		IPPROTO_PIM,
144 	.pr_flags =		PR_ATOMIC|PR_ADDR|PR_LASTHDR,
145 	.pr_input =		pim6_input,
146 	.pr_output =		rip6_output,
147 	.pr_ctloutput =		rip6_ctloutput,
148 	.pr_usrreqs =		&rip6_usrreqs
149 };
150 static int pim6_encapcheck(const struct mbuf *, int, int, void *);
151 
152 static VNET_DEFINE(int, ip6_mrouter_ver) = 0;
153 #define	V_ip6_mrouter_ver	VNET(ip6_mrouter_ver)
154 
155 SYSCTL_DECL(_net_inet6);
156 SYSCTL_DECL(_net_inet6_ip6);
157 static SYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim, CTLFLAG_RW, 0, "PIM");
158 
159 static struct mrt6stat mrt6stat;
160 SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
161     &mrt6stat, mrt6stat,
162     "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
163 
164 #define	MRT6STAT_INC(name)	mrt6stat.name += 1
165 #define NO_RTE_FOUND	0x1
166 #define RTE_FOUND	0x2
167 
168 static struct mtx mrouter6_mtx;
169 #define	MROUTER6_LOCK()		mtx_lock(&mrouter6_mtx)
170 #define	MROUTER6_UNLOCK()	mtx_unlock(&mrouter6_mtx)
171 #define	MROUTER6_LOCK_ASSERT()	do {					\
172 	mtx_assert(&mrouter6_mtx, MA_OWNED);				\
173 	NET_ASSERT_GIANT();						\
174 } while (0)
175 #define	MROUTER6_LOCK_INIT()	\
176 	mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
177 #define	MROUTER6_LOCK_DESTROY()	mtx_destroy(&mrouter6_mtx)
178 
179 static struct mf6c *mf6ctable[MF6CTBLSIZ];
180 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
181     &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
182     "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
183     "netinet6/ip6_mroute.h)");
184 
185 static struct mtx mfc6_mtx;
186 #define	MFC6_LOCK()		mtx_lock(&mfc6_mtx)
187 #define	MFC6_UNLOCK()		mtx_unlock(&mfc6_mtx)
188 #define	MFC6_LOCK_ASSERT()	do {					\
189 	mtx_assert(&mfc6_mtx, MA_OWNED);				\
190 	NET_ASSERT_GIANT();						\
191 } while (0)
192 #define	MFC6_LOCK_INIT()		\
193 	mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
194 #define	MFC6_LOCK_DESTROY()	mtx_destroy(&mfc6_mtx)
195 
196 static u_char n6expire[MF6CTBLSIZ];
197 
198 static struct mif6 mif6table[MAXMIFS];
199 static int
200 sysctl_mif6table(SYSCTL_HANDLER_ARGS)
201 {
202 	struct mif6_sctl *out;
203 	int error;
204 
205 	out = malloc(sizeof(struct mif6_sctl) * MAXMIFS, M_TEMP, M_WAITOK);
206 	for (int i = 0; i < MAXMIFS; i++) {
207 		out[i].m6_flags		= mif6table[i].m6_flags;
208 		out[i].m6_rate_limit	= mif6table[i].m6_rate_limit;
209 		out[i].m6_lcl_addr	= mif6table[i].m6_lcl_addr;
210 		if (mif6table[i].m6_ifp != NULL)
211 			out[i].m6_ifp	= mif6table[i].m6_ifp->if_index;
212 		else
213 			out[i].m6_ifp	= 0;
214 		out[i].m6_pkt_in	= mif6table[i].m6_pkt_in;
215 		out[i].m6_pkt_out	= mif6table[i].m6_pkt_out;
216 		out[i].m6_bytes_in	= mif6table[i].m6_bytes_in;
217 		out[i].m6_bytes_out	= mif6table[i].m6_bytes_out;
218 	}
219 	error = SYSCTL_OUT(req, out, sizeof(struct mif6_sctl) * MAXMIFS);
220 	free(out, M_TEMP);
221 	return (error);
222 }
223 SYSCTL_PROC(_net_inet6_ip6, OID_AUTO, mif6table, CTLTYPE_OPAQUE | CTLFLAG_RD,
224     NULL, 0, sysctl_mif6table, "S,mif6_sctl[MAXMIFS]",
225     "IPv6 Multicast Interfaces (struct mif6_sctl[MAXMIFS], "
226     "netinet6/ip6_mroute.h)");
227 
228 static struct mtx mif6_mtx;
229 #define	MIF6_LOCK()		mtx_lock(&mif6_mtx)
230 #define	MIF6_UNLOCK()		mtx_unlock(&mif6_mtx)
231 #define	MIF6_LOCK_ASSERT()	mtx_assert(&mif6_mtx, MA_OWNED)
232 #define	MIF6_LOCK_INIT()	\
233 	mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
234 #define	MIF6_LOCK_DESTROY()	mtx_destroy(&mif6_mtx)
235 
236 #ifdef MRT6DEBUG
237 static VNET_DEFINE(u_int, mrt6debug) = 0;	/* debug level */
238 #define	V_mrt6debug		VNET(mrt6debug)
239 #define DEBUG_MFC	0x02
240 #define DEBUG_FORWARD	0x04
241 #define DEBUG_EXPIRE	0x08
242 #define DEBUG_XMIT	0x10
243 #define DEBUG_REG	0x20
244 #define DEBUG_PIM	0x40
245 #define	DEBUG_ERR	0x80
246 #define	DEBUG_ANY	0x7f
247 #define	MRT6_DLOG(m, fmt, ...)	\
248 	if (V_mrt6debug & (m))	\
249 		log(((m) & DEBUG_ERR) ? LOG_ERR: LOG_DEBUG, \
250 		    "%s: " fmt "\n", __func__, ##__VA_ARGS__)
251 #else
252 #define	MRT6_DLOG(m, fmt, ...)
253 #endif
254 
255 static void	expire_upcalls(void *);
256 #define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
257 #define	UPCALL_EXPIRE	6		/* number of timeouts */
258 
259 /*
260  * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
261  */
262 
263 /*
264  * 'Interfaces' associated with decapsulator (so we can tell
265  * packets that went through it from ones that get reflected
266  * by a broken gateway).  Different from IPv4 register_if,
267  * these interfaces are linked into the system ifnet list,
268  * because per-interface IPv6 statistics are maintained in
269  * ifp->if_afdata.  But it does not have any routes point
270  * to them.  I.e., packets can't be sent this way.  They
271  * only exist as a placeholder for multicast source
272  * verification.
273  */
274 static struct ifnet *multicast_register_if6;
275 
276 #define ENCAP_HOPS 64
277 
278 /*
279  * Private variables.
280  */
281 static mifi_t nummifs = 0;
282 static mifi_t reg_mif_num = (mifi_t)-1;
283 
284 static struct pim6stat pim6stat;
285 SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RW,
286     &pim6stat, pim6stat,
287     "PIM Statistics (struct pim6stat, netinet6/pim6_var.h)");
288 
289 #define	PIM6STAT_INC(name)	pim6stat.name += 1
290 static VNET_DEFINE(int, pim6);
291 #define	V_pim6		VNET(pim6)
292 
293 /*
294  * Hash function for a source, group entry
295  */
296 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
297 				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
298 				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
299 				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
300 
301 /*
302  * Find a route for a given origin IPv6 address and Multicast group address.
303  */
304 #define MF6CFIND(o, g, rt) do { \
305 	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
306 	rt = NULL; \
307 	while (_rt) { \
308 		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
309 		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
310 		    (_rt->mf6c_stall == NULL)) { \
311 			rt = _rt; \
312 			break; \
313 		} \
314 		_rt = _rt->mf6c_next; \
315 	} \
316 	if (rt == NULL) { \
317 		MRT6STAT_INC(mrt6s_mfc_misses); \
318 	} \
319 } while (/*CONSTCOND*/ 0)
320 
321 /*
322  * Macros to compute elapsed time efficiently
323  * Borrowed from Van Jacobson's scheduling code
324  * XXX: replace with timersub() ?
325  */
326 #define TV_DELTA(a, b, delta) do { \
327 	    int xxs; \
328 		\
329 	    delta = (a).tv_usec - (b).tv_usec; \
330 	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
331 	       switch (xxs) { \
332 		      case 2: \
333 			  delta += 1000000; \
334 			      /* FALLTHROUGH */ \
335 		      case 1: \
336 			  delta += 1000000; \
337 			  break; \
338 		      default: \
339 			  delta += (1000000 * xxs); \
340 	       } \
341 	    } \
342 } while (/*CONSTCOND*/ 0)
343 
344 /* XXX: replace with timercmp(a, b, <) ? */
345 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
346 	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
347 
348 #ifdef UPCALL_TIMING
349 #define UPCALL_MAX	50
350 static u_long upcall_data[UPCALL_MAX + 1];
351 static void collate();
352 #endif /* UPCALL_TIMING */
353 
354 static int ip6_mrouter_init(struct socket *, int, int);
355 static int add_m6fc(struct mf6cctl *);
356 static int add_m6if(struct mif6ctl *);
357 static int del_m6fc(struct mf6cctl *);
358 static int del_m6if(mifi_t *);
359 static int del_m6if_locked(mifi_t *);
360 static int get_mif6_cnt(struct sioc_mif_req6 *);
361 static int get_sg_cnt(struct sioc_sg_req6 *);
362 
363 static struct callout expire_upcalls_ch;
364 
365 int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
366 int X_ip6_mrouter_done(void);
367 int X_ip6_mrouter_set(struct socket *, struct sockopt *);
368 int X_ip6_mrouter_get(struct socket *, struct sockopt *);
369 int X_mrt6_ioctl(u_long, caddr_t);
370 
371 /*
372  * Handle MRT setsockopt commands to modify the multicast routing tables.
373  */
374 int
375 X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
376 {
377 	int error = 0;
378 	int optval;
379 	struct mif6ctl mifc;
380 	struct mf6cctl mfcc;
381 	mifi_t mifi;
382 
383 	if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
384 		return (EPERM);
385 
386 	switch (sopt->sopt_name) {
387 	case MRT6_INIT:
388 #ifdef MRT6_OINIT
389 	case MRT6_OINIT:
390 #endif
391 		error = sooptcopyin(sopt, &optval, sizeof(optval),
392 		    sizeof(optval));
393 		if (error)
394 			break;
395 		error = ip6_mrouter_init(so, optval, sopt->sopt_name);
396 		break;
397 	case MRT6_DONE:
398 		error = X_ip6_mrouter_done();
399 		break;
400 	case MRT6_ADD_MIF:
401 		error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
402 		if (error)
403 			break;
404 		error = add_m6if(&mifc);
405 		break;
406 	case MRT6_ADD_MFC:
407 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
408 		if (error)
409 			break;
410 		error = add_m6fc(&mfcc);
411 		break;
412 	case MRT6_DEL_MFC:
413 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
414 		if (error)
415 			break;
416 		error = del_m6fc(&mfcc);
417 		break;
418 	case MRT6_DEL_MIF:
419 		error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
420 		if (error)
421 			break;
422 		error = del_m6if(&mifi);
423 		break;
424 	case MRT6_PIM:
425 		error = sooptcopyin(sopt, &optval, sizeof(optval),
426 		    sizeof(optval));
427 		if (error)
428 			break;
429 		error = set_pim6(&optval);
430 		break;
431 	default:
432 		error = EOPNOTSUPP;
433 		break;
434 	}
435 
436 	return (error);
437 }
438 
439 /*
440  * Handle MRT getsockopt commands
441  */
442 int
443 X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
444 {
445 	int error = 0;
446 
447 	if (so != V_ip6_mrouter)
448 		return (EACCES);
449 
450 	switch (sopt->sopt_name) {
451 		case MRT6_PIM:
452 			error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
453 			break;
454 	}
455 	return (error);
456 }
457 
458 /*
459  * Handle ioctl commands to obtain information from the cache
460  */
461 int
462 X_mrt6_ioctl(u_long cmd, caddr_t data)
463 {
464 	int ret;
465 
466 	ret = EINVAL;
467 
468 	switch (cmd) {
469 	case SIOCGETSGCNT_IN6:
470 		ret = get_sg_cnt((struct sioc_sg_req6 *)data);
471 		break;
472 
473 	case SIOCGETMIFCNT_IN6:
474 		ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
475 		break;
476 
477 	default:
478 		break;
479 	}
480 
481 	return (ret);
482 }
483 
484 /*
485  * returns the packet, byte, rpf-failure count for the source group provided
486  */
487 static int
488 get_sg_cnt(struct sioc_sg_req6 *req)
489 {
490 	struct mf6c *rt;
491 	int ret;
492 
493 	ret = 0;
494 
495 	MFC6_LOCK();
496 
497 	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
498 	if (rt == NULL) {
499 		ret = ESRCH;
500 	} else {
501 		req->pktcnt = rt->mf6c_pkt_cnt;
502 		req->bytecnt = rt->mf6c_byte_cnt;
503 		req->wrong_if = rt->mf6c_wrong_if;
504 	}
505 
506 	MFC6_UNLOCK();
507 
508 	return (ret);
509 }
510 
511 /*
512  * returns the input and output packet and byte counts on the mif provided
513  */
514 static int
515 get_mif6_cnt(struct sioc_mif_req6 *req)
516 {
517 	mifi_t mifi;
518 	int ret;
519 
520 	ret = 0;
521 	mifi = req->mifi;
522 
523 	MIF6_LOCK();
524 
525 	if (mifi >= nummifs) {
526 		ret = EINVAL;
527 	} else {
528 		req->icount = mif6table[mifi].m6_pkt_in;
529 		req->ocount = mif6table[mifi].m6_pkt_out;
530 		req->ibytes = mif6table[mifi].m6_bytes_in;
531 		req->obytes = mif6table[mifi].m6_bytes_out;
532 	}
533 
534 	MIF6_UNLOCK();
535 
536 	return (ret);
537 }
538 
539 static int
540 set_pim6(int *i)
541 {
542 	if ((*i != 1) && (*i != 0))
543 		return (EINVAL);
544 
545 	V_pim6 = *i;
546 
547 	return (0);
548 }
549 
550 /*
551  * Enable multicast routing
552  */
553 static int
554 ip6_mrouter_init(struct socket *so, int v, int cmd)
555 {
556 
557 	MRT6_DLOG(DEBUG_ANY, "so_type = %d, pr_protocol = %d",
558 	    so->so_type, so->so_proto->pr_protocol);
559 
560 	if (so->so_type != SOCK_RAW ||
561 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
562 		return (EOPNOTSUPP);
563 
564 	if (v != 1)
565 		return (ENOPROTOOPT);
566 
567 	MROUTER6_LOCK();
568 
569 	if (V_ip6_mrouter != NULL) {
570 		MROUTER6_UNLOCK();
571 		return (EADDRINUSE);
572 	}
573 
574 	V_ip6_mrouter = so;
575 	V_ip6_mrouter_ver = cmd;
576 
577 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
578 	bzero((caddr_t)n6expire, sizeof(n6expire));
579 
580 	V_pim6 = 0;/* used for stubbing out/in pim stuff */
581 
582 	callout_init(&expire_upcalls_ch, 0);
583 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
584 	    expire_upcalls, NULL);
585 
586 	MROUTER6_UNLOCK();
587 	MRT6_DLOG(DEBUG_ANY, "finished");
588 
589 	return (0);
590 }
591 
592 /*
593  * Disable IPv6 multicast forwarding.
594  */
595 int
596 X_ip6_mrouter_done(void)
597 {
598 	mifi_t mifi;
599 	u_long i;
600 	struct mf6c *rt;
601 	struct rtdetq *rte;
602 
603 	MROUTER6_LOCK();
604 
605 	if (V_ip6_mrouter == NULL) {
606 		MROUTER6_UNLOCK();
607 		return (EINVAL);
608 	}
609 
610 	/*
611 	 * For each phyint in use, disable promiscuous reception of all IPv6
612 	 * multicasts.
613 	 */
614 	for (mifi = 0; mifi < nummifs; mifi++) {
615 		if (mif6table[mifi].m6_ifp &&
616 		    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
617 			if_allmulti(mif6table[mifi].m6_ifp, 0);
618 		}
619 	}
620 	bzero((caddr_t)mif6table, sizeof(mif6table));
621 	nummifs = 0;
622 
623 	V_pim6 = 0; /* used to stub out/in pim specific code */
624 
625 	callout_stop(&expire_upcalls_ch);
626 
627 	/*
628 	 * Free all multicast forwarding cache entries.
629 	 */
630 	MFC6_LOCK();
631 	for (i = 0; i < MF6CTBLSIZ; i++) {
632 		rt = mf6ctable[i];
633 		while (rt) {
634 			struct mf6c *frt;
635 
636 			for (rte = rt->mf6c_stall; rte != NULL; ) {
637 				struct rtdetq *n = rte->next;
638 
639 				m_freem(rte->m);
640 				free(rte, M_MRTABLE6);
641 				rte = n;
642 			}
643 			frt = rt;
644 			rt = rt->mf6c_next;
645 			free(frt, M_MRTABLE6);
646 		}
647 	}
648 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
649 	MFC6_UNLOCK();
650 
651 	/*
652 	 * Reset register interface
653 	 */
654 	if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
655 		if_detach(multicast_register_if6);
656 		if_free(multicast_register_if6);
657 		reg_mif_num = (mifi_t)-1;
658 		multicast_register_if6 = NULL;
659 	}
660 
661 	V_ip6_mrouter = NULL;
662 	V_ip6_mrouter_ver = 0;
663 
664 	MROUTER6_UNLOCK();
665 	MRT6_DLOG(DEBUG_ANY, "finished");
666 
667 	return (0);
668 }
669 
670 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
671 
672 /*
673  * Add a mif to the mif table
674  */
675 static int
676 add_m6if(struct mif6ctl *mifcp)
677 {
678 	struct mif6 *mifp;
679 	struct ifnet *ifp;
680 	int error;
681 
682 	MIF6_LOCK();
683 
684 	if (mifcp->mif6c_mifi >= MAXMIFS) {
685 		MIF6_UNLOCK();
686 		return (EINVAL);
687 	}
688 	mifp = mif6table + mifcp->mif6c_mifi;
689 	if (mifp->m6_ifp != NULL) {
690 		MIF6_UNLOCK();
691 		return (EADDRINUSE); /* XXX: is it appropriate? */
692 	}
693 	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
694 		MIF6_UNLOCK();
695 		return (ENXIO);
696 	}
697 
698 	ifp = ifnet_byindex(mifcp->mif6c_pifi);
699 
700 	if (mifcp->mif6c_flags & MIFF_REGISTER) {
701 		if (reg_mif_num == (mifi_t)-1) {
702 			ifp = if_alloc(IFT_OTHER);
703 
704 			if_initname(ifp, "register_mif", 0);
705 			ifp->if_flags |= IFF_LOOPBACK;
706 			if_attach(ifp);
707 			multicast_register_if6 = ifp;
708 			reg_mif_num = mifcp->mif6c_mifi;
709 			/*
710 			 * it is impossible to guess the ifindex of the
711 			 * register interface.  So mif6c_pifi is automatically
712 			 * calculated.
713 			 */
714 			mifcp->mif6c_pifi = ifp->if_index;
715 		} else {
716 			ifp = multicast_register_if6;
717 		}
718 	} else {
719 		/* Make sure the interface supports multicast */
720 		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
721 			MIF6_UNLOCK();
722 			return (EOPNOTSUPP);
723 		}
724 
725 		error = if_allmulti(ifp, 1);
726 		if (error) {
727 			MIF6_UNLOCK();
728 			return (error);
729 		}
730 	}
731 
732 	mifp->m6_flags     = mifcp->mif6c_flags;
733 	mifp->m6_ifp       = ifp;
734 
735 	/* initialize per mif pkt counters */
736 	mifp->m6_pkt_in    = 0;
737 	mifp->m6_pkt_out   = 0;
738 	mifp->m6_bytes_in  = 0;
739 	mifp->m6_bytes_out = 0;
740 
741 	/* Adjust nummifs up if the mifi is higher than nummifs */
742 	if (nummifs <= mifcp->mif6c_mifi)
743 		nummifs = mifcp->mif6c_mifi + 1;
744 
745 	MIF6_UNLOCK();
746 	MRT6_DLOG(DEBUG_ANY, "mif #%d, phyint %s", mifcp->mif6c_mifi,
747 	    if_name(ifp));
748 
749 	return (0);
750 }
751 
752 /*
753  * Delete a mif from the mif table
754  */
755 static int
756 del_m6if_locked(mifi_t *mifip)
757 {
758 	struct mif6 *mifp = mif6table + *mifip;
759 	mifi_t mifi;
760 	struct ifnet *ifp;
761 
762 	MIF6_LOCK_ASSERT();
763 
764 	if (*mifip >= nummifs)
765 		return (EINVAL);
766 	if (mifp->m6_ifp == NULL)
767 		return (EINVAL);
768 
769 	if (!(mifp->m6_flags & MIFF_REGISTER)) {
770 		/* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
771 		ifp = mifp->m6_ifp;
772 		if_allmulti(ifp, 0);
773 	} else {
774 		if (reg_mif_num != (mifi_t)-1 &&
775 		    multicast_register_if6 != NULL) {
776 			if_detach(multicast_register_if6);
777 			if_free(multicast_register_if6);
778 			reg_mif_num = (mifi_t)-1;
779 			multicast_register_if6 = NULL;
780 		}
781 	}
782 
783 	bzero((caddr_t)mifp, sizeof(*mifp));
784 
785 	/* Adjust nummifs down */
786 	for (mifi = nummifs; mifi > 0; mifi--)
787 		if (mif6table[mifi - 1].m6_ifp)
788 			break;
789 	nummifs = mifi;
790 	MRT6_DLOG(DEBUG_ANY, "mif %d, nummifs %d", *mifip, nummifs);
791 
792 	return (0);
793 }
794 
795 static int
796 del_m6if(mifi_t *mifip)
797 {
798 	int cc;
799 
800 	MIF6_LOCK();
801 	cc = del_m6if_locked(mifip);
802 	MIF6_UNLOCK();
803 
804 	return (cc);
805 }
806 
807 /*
808  * Add an mfc entry
809  */
810 static int
811 add_m6fc(struct mf6cctl *mfccp)
812 {
813 	struct mf6c *rt;
814 	u_long hash;
815 	struct rtdetq *rte;
816 	u_short nstl;
817 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
818 
819 	MFC6_LOCK();
820 
821 	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
822 		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
823 
824 	/* If an entry already exists, just update the fields */
825 	if (rt) {
826 		MRT6_DLOG(DEBUG_MFC, "no upcall o %s g %s p %x",
827 		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
828 		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
829 		    mfccp->mf6cc_parent);
830 
831 		rt->mf6c_parent = mfccp->mf6cc_parent;
832 		rt->mf6c_ifset = mfccp->mf6cc_ifset;
833 
834 		MFC6_UNLOCK();
835 		return (0);
836 	}
837 
838 	/*
839 	 * Find the entry for which the upcall was made and update
840 	 */
841 	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
842 			mfccp->mf6cc_mcastgrp.sin6_addr);
843 	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
844 		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
845 				       &mfccp->mf6cc_origin.sin6_addr) &&
846 		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
847 				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
848 		    (rt->mf6c_stall != NULL)) {
849 
850 			if (nstl++)
851 				log(LOG_ERR,
852 				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
853 				    "multiple kernel entries",
854 				    ip6_sprintf(ip6bufo,
855 					    &mfccp->mf6cc_origin.sin6_addr),
856 				    ip6_sprintf(ip6bufg,
857 					    &mfccp->mf6cc_mcastgrp.sin6_addr),
858 				    mfccp->mf6cc_parent, rt->mf6c_stall);
859 
860 			MRT6_DLOG(DEBUG_MFC, "o %s g %s p %x dbg %p",
861 			    ip6_sprintf(ip6bufo,
862 			    &mfccp->mf6cc_origin.sin6_addr),
863 			    ip6_sprintf(ip6bufg,
864 				&mfccp->mf6cc_mcastgrp.sin6_addr),
865 			    mfccp->mf6cc_parent, rt->mf6c_stall);
866 
867 			rt->mf6c_origin     = mfccp->mf6cc_origin;
868 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
869 			rt->mf6c_parent     = mfccp->mf6cc_parent;
870 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
871 			/* initialize pkt counters per src-grp */
872 			rt->mf6c_pkt_cnt    = 0;
873 			rt->mf6c_byte_cnt   = 0;
874 			rt->mf6c_wrong_if   = 0;
875 
876 			rt->mf6c_expire = 0;	/* Don't clean this guy up */
877 			n6expire[hash]--;
878 
879 			/* free packets Qed at the end of this entry */
880 			for (rte = rt->mf6c_stall; rte != NULL; ) {
881 				struct rtdetq *n = rte->next;
882 				ip6_mdq(rte->m, rte->ifp, rt);
883 				m_freem(rte->m);
884 #ifdef UPCALL_TIMING
885 				collate(&(rte->t));
886 #endif /* UPCALL_TIMING */
887 				free(rte, M_MRTABLE6);
888 				rte = n;
889 			}
890 			rt->mf6c_stall = NULL;
891 		}
892 	}
893 
894 	/*
895 	 * It is possible that an entry is being inserted without an upcall
896 	 */
897 	if (nstl == 0) {
898 		MRT6_DLOG(DEBUG_MFC, "no upcall h %lu o %s g %s p %x", hash,
899 		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
900 		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
901 		    mfccp->mf6cc_parent);
902 
903 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
904 
905 			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
906 					       &mfccp->mf6cc_origin.sin6_addr)&&
907 			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
908 					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
909 
910 				rt->mf6c_origin     = mfccp->mf6cc_origin;
911 				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
912 				rt->mf6c_parent     = mfccp->mf6cc_parent;
913 				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
914 				/* initialize pkt counters per src-grp */
915 				rt->mf6c_pkt_cnt    = 0;
916 				rt->mf6c_byte_cnt   = 0;
917 				rt->mf6c_wrong_if   = 0;
918 
919 				if (rt->mf6c_expire)
920 					n6expire[hash]--;
921 				rt->mf6c_expire	   = 0;
922 			}
923 		}
924 		if (rt == NULL) {
925 			/* no upcall, so make a new entry */
926 			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
927 						  M_NOWAIT);
928 			if (rt == NULL) {
929 				MFC6_UNLOCK();
930 				return (ENOBUFS);
931 			}
932 
933 			/* insert new entry at head of hash chain */
934 			rt->mf6c_origin     = mfccp->mf6cc_origin;
935 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
936 			rt->mf6c_parent     = mfccp->mf6cc_parent;
937 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
938 			/* initialize pkt counters per src-grp */
939 			rt->mf6c_pkt_cnt    = 0;
940 			rt->mf6c_byte_cnt   = 0;
941 			rt->mf6c_wrong_if   = 0;
942 			rt->mf6c_expire     = 0;
943 			rt->mf6c_stall = NULL;
944 
945 			/* link into table */
946 			rt->mf6c_next  = mf6ctable[hash];
947 			mf6ctable[hash] = rt;
948 		}
949 	}
950 
951 	MFC6_UNLOCK();
952 	return (0);
953 }
954 
955 #ifdef UPCALL_TIMING
956 /*
957  * collect delay statistics on the upcalls
958  */
959 static void
960 collate(struct timeval *t)
961 {
962 	u_long d;
963 	struct timeval tp;
964 	u_long delta;
965 
966 	GET_TIME(tp);
967 
968 	if (TV_LT(*t, tp))
969 	{
970 		TV_DELTA(tp, *t, delta);
971 
972 		d = delta >> 10;
973 		if (d > UPCALL_MAX)
974 			d = UPCALL_MAX;
975 
976 		++upcall_data[d];
977 	}
978 }
979 #endif /* UPCALL_TIMING */
980 
981 /*
982  * Delete an mfc entry
983  */
984 static int
985 del_m6fc(struct mf6cctl *mfccp)
986 {
987 #ifdef MRT6DEBUG
988 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
989 #endif
990 	struct sockaddr_in6	origin;
991 	struct sockaddr_in6	mcastgrp;
992 	struct mf6c		*rt;
993 	struct mf6c		**nptr;
994 	u_long		hash;
995 
996 	origin = mfccp->mf6cc_origin;
997 	mcastgrp = mfccp->mf6cc_mcastgrp;
998 	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
999 
1000 	MRT6_DLOG(DEBUG_MFC, "orig %s mcastgrp %s",
1001 	    ip6_sprintf(ip6bufo, &origin.sin6_addr),
1002 	    ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
1003 
1004 	MFC6_LOCK();
1005 
1006 	nptr = &mf6ctable[hash];
1007 	while ((rt = *nptr) != NULL) {
1008 		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
1009 				       &rt->mf6c_origin.sin6_addr) &&
1010 		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1011 				       &rt->mf6c_mcastgrp.sin6_addr) &&
1012 		    rt->mf6c_stall == NULL)
1013 			break;
1014 
1015 		nptr = &rt->mf6c_next;
1016 	}
1017 	if (rt == NULL) {
1018 		MFC6_UNLOCK();
1019 		return (EADDRNOTAVAIL);
1020 	}
1021 
1022 	*nptr = rt->mf6c_next;
1023 	free(rt, M_MRTABLE6);
1024 
1025 	MFC6_UNLOCK();
1026 
1027 	return (0);
1028 }
1029 
1030 static int
1031 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1032 {
1033 
1034 	if (s) {
1035 		if (sbappendaddr(&s->so_rcv,
1036 				 (struct sockaddr *)src,
1037 				 mm, (struct mbuf *)0) != 0) {
1038 			sorwakeup(s);
1039 			return (0);
1040 		}
1041 	}
1042 	m_freem(mm);
1043 	return (-1);
1044 }
1045 
1046 /*
1047  * IPv6 multicast forwarding function. This function assumes that the packet
1048  * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1049  * pointed to by "ifp", and the packet is to be relayed to other networks
1050  * that have members of the packet's destination IPv6 multicast group.
1051  *
1052  * The packet is returned unscathed to the caller, unless it is
1053  * erroneous, in which case a non-zero return value tells the caller to
1054  * discard it.
1055  *
1056  * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1057  * this function is called in the originating context (i.e., not when
1058  * forwarding a packet from other node).  ip6_output(), which is currently the
1059  * only function that calls this function is called in the originating context,
1060  * explicitly ensures this condition.  It is caller's responsibility to ensure
1061  * that if this function is called from somewhere else in the originating
1062  * context in the future.
1063  */
1064 int
1065 X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1066 {
1067 	struct rtdetq *rte;
1068 	struct mbuf *mb0;
1069 	struct mf6c *rt;
1070 	struct mif6 *mifp;
1071 	struct mbuf *mm;
1072 	u_long hash;
1073 	mifi_t mifi;
1074 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1075 #ifdef UPCALL_TIMING
1076 	struct timeval tp;
1077 
1078 	GET_TIME(tp);
1079 #endif /* UPCALL_TIMING */
1080 
1081 	MRT6_DLOG(DEBUG_FORWARD, "src %s, dst %s, ifindex %d",
1082 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1083 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst), ifp->if_index);
1084 
1085 	/*
1086 	 * Don't forward a packet with Hop limit of zero or one,
1087 	 * or a packet destined to a local-only group.
1088 	 */
1089 	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1090 	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1091 		return (0);
1092 	ip6->ip6_hlim--;
1093 
1094 	/*
1095 	 * Source address check: do not forward packets with unspecified
1096 	 * source. It was discussed in July 2000, on ipngwg mailing list.
1097 	 * This is rather more serious than unicast cases, because some
1098 	 * MLD packets can be sent with the unspecified source address
1099 	 * (although such packets must normally set 1 to the hop limit field).
1100 	 */
1101 	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1102 		IP6STAT_INC(ip6s_cantforward);
1103 		if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
1104 			V_ip6_log_time = time_uptime;
1105 			log(LOG_DEBUG,
1106 			    "cannot forward "
1107 			    "from %s to %s nxt %d received on %s\n",
1108 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1109 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1110 			    ip6->ip6_nxt,
1111 			    if_name(m->m_pkthdr.rcvif));
1112 		}
1113 		return (0);
1114 	}
1115 
1116 	MFC6_LOCK();
1117 
1118 	/*
1119 	 * Determine forwarding mifs from the forwarding cache table
1120 	 */
1121 	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1122 	MRT6STAT_INC(mrt6s_mfc_lookups);
1123 
1124 	/* Entry exists, so forward if necessary */
1125 	if (rt) {
1126 		MFC6_UNLOCK();
1127 		return (ip6_mdq(m, ifp, rt));
1128 	}
1129 
1130 	/*
1131 	 * If we don't have a route for packet's origin,
1132 	 * Make a copy of the packet & send message to routing daemon.
1133 	 */
1134 	MRT6STAT_INC(mrt6s_no_route);
1135 	MRT6_DLOG(DEBUG_FORWARD | DEBUG_MFC, "no rte s %s g %s",
1136 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1137 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1138 
1139 	/*
1140 	 * Allocate mbufs early so that we don't do extra work if we
1141 	 * are just going to fail anyway.
1142 	 */
1143 	rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6, M_NOWAIT);
1144 	if (rte == NULL) {
1145 		MFC6_UNLOCK();
1146 		return (ENOBUFS);
1147 	}
1148 	mb0 = m_copym(m, 0, M_COPYALL, M_NOWAIT);
1149 	/*
1150 	 * Pullup packet header if needed before storing it,
1151 	 * as other references may modify it in the meantime.
1152 	 */
1153 	if (mb0 && (!M_WRITABLE(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1154 		mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1155 	if (mb0 == NULL) {
1156 		free(rte, M_MRTABLE6);
1157 		MFC6_UNLOCK();
1158 		return (ENOBUFS);
1159 	}
1160 
1161 	/* is there an upcall waiting for this packet? */
1162 	hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1163 	for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1164 		if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1165 		    &rt->mf6c_origin.sin6_addr) &&
1166 		    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1167 		    &rt->mf6c_mcastgrp.sin6_addr) && (rt->mf6c_stall != NULL))
1168 			break;
1169 	}
1170 
1171 	if (rt == NULL) {
1172 		struct mrt6msg *im;
1173 #ifdef MRT6_OINIT
1174 		struct omrt6msg *oim;
1175 #endif
1176 		/* no upcall, so make a new entry */
1177 		rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, M_NOWAIT);
1178 		if (rt == NULL) {
1179 			free(rte, M_MRTABLE6);
1180 			m_freem(mb0);
1181 			MFC6_UNLOCK();
1182 			return (ENOBUFS);
1183 		}
1184 		/*
1185 		 * Make a copy of the header to send to the user
1186 		 * level process
1187 		 */
1188 		mm = m_copym(mb0, 0, sizeof(struct ip6_hdr), M_NOWAIT);
1189 		if (mm == NULL) {
1190 			free(rte, M_MRTABLE6);
1191 			m_freem(mb0);
1192 			free(rt, M_MRTABLE6);
1193 			MFC6_UNLOCK();
1194 			return (ENOBUFS);
1195 		}
1196 
1197 		/*
1198 		 * Send message to routing daemon
1199 		 */
1200 		sin6.sin6_addr = ip6->ip6_src;
1201 		im = NULL;
1202 #ifdef MRT6_OINIT
1203 		oim = NULL;
1204 #endif
1205 		switch (V_ip6_mrouter_ver) {
1206 #ifdef MRT6_OINIT
1207 		case MRT6_OINIT:
1208 			oim = mtod(mm, struct omrt6msg *);
1209 			oim->im6_msgtype = MRT6MSG_NOCACHE;
1210 			oim->im6_mbz = 0;
1211 			break;
1212 #endif
1213 		case MRT6_INIT:
1214 			im = mtod(mm, struct mrt6msg *);
1215 			im->im6_msgtype = MRT6MSG_NOCACHE;
1216 			im->im6_mbz = 0;
1217 			break;
1218 		default:
1219 			free(rte, M_MRTABLE6);
1220 			m_freem(mb0);
1221 			free(rt, M_MRTABLE6);
1222 			MFC6_UNLOCK();
1223 			return (EINVAL);
1224 		}
1225 
1226 		MRT6_DLOG(DEBUG_FORWARD, "getting the iif info in the kernel");
1227 		for (mifp = mif6table, mifi = 0;
1228 		    mifi < nummifs && mifp->m6_ifp != ifp; mifp++, mifi++)
1229 				;
1230 
1231 		switch (V_ip6_mrouter_ver) {
1232 #ifdef MRT6_OINIT
1233 		case MRT6_OINIT:
1234 			oim->im6_mif = mifi;
1235 			break;
1236 #endif
1237 		case MRT6_INIT:
1238 			im->im6_mif = mifi;
1239 			break;
1240 		}
1241 
1242 		if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1243 			log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1244 			    "socket queue full\n");
1245 			MRT6STAT_INC(mrt6s_upq_sockfull);
1246 			free(rte, M_MRTABLE6);
1247 			m_freem(mb0);
1248 			free(rt, M_MRTABLE6);
1249 			MFC6_UNLOCK();
1250 			return (ENOBUFS);
1251 		}
1252 
1253 		MRT6STAT_INC(mrt6s_upcalls);
1254 
1255 		/* insert new entry at head of hash chain */
1256 		bzero(rt, sizeof(*rt));
1257 		rt->mf6c_origin.sin6_family = AF_INET6;
1258 		rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1259 		rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1260 		rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1261 		rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1262 		rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1263 		rt->mf6c_expire = UPCALL_EXPIRE;
1264 		n6expire[hash]++;
1265 		rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1266 
1267 		/* link into table */
1268 		rt->mf6c_next  = mf6ctable[hash];
1269 		mf6ctable[hash] = rt;
1270 		/* Add this entry to the end of the queue */
1271 		rt->mf6c_stall = rte;
1272 	} else {
1273 		/* determine if q has overflowed */
1274 		struct rtdetq **p;
1275 		int npkts = 0;
1276 
1277 		for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1278 			if (++npkts > MAX_UPQ6) {
1279 				MRT6STAT_INC(mrt6s_upq_ovflw);
1280 				free(rte, M_MRTABLE6);
1281 				m_freem(mb0);
1282 				MFC6_UNLOCK();
1283 				return (0);
1284 			}
1285 
1286 		/* Add this entry to the end of the queue */
1287 		*p = rte;
1288 	}
1289 
1290 	rte->next = NULL;
1291 	rte->m = mb0;
1292 	rte->ifp = ifp;
1293 #ifdef UPCALL_TIMING
1294 	rte->t = tp;
1295 #endif /* UPCALL_TIMING */
1296 
1297 	MFC6_UNLOCK();
1298 
1299 	return (0);
1300 }
1301 
1302 /*
1303  * Clean up cache entries if upcalls are not serviced
1304  * Call from the Slow Timeout mechanism, every half second.
1305  */
1306 static void
1307 expire_upcalls(void *unused)
1308 {
1309 #ifdef MRT6DEBUG
1310 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1311 #endif
1312 	struct rtdetq *rte;
1313 	struct mf6c *mfc, **nptr;
1314 	u_long i;
1315 
1316 	MFC6_LOCK();
1317 	for (i = 0; i < MF6CTBLSIZ; i++) {
1318 		if (n6expire[i] == 0)
1319 			continue;
1320 		nptr = &mf6ctable[i];
1321 		while ((mfc = *nptr) != NULL) {
1322 			rte = mfc->mf6c_stall;
1323 			/*
1324 			 * Skip real cache entries
1325 			 * Make sure it wasn't marked to not expire (shouldn't happen)
1326 			 * If it expires now
1327 			 */
1328 			if (rte != NULL &&
1329 			    mfc->mf6c_expire != 0 &&
1330 			    --mfc->mf6c_expire == 0) {
1331 				MRT6_DLOG(DEBUG_EXPIRE, "expiring (%s %s)",
1332 				    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1333 				    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1334 				/*
1335 				 * drop all the packets
1336 				 * free the mbuf with the pkt, if, timing info
1337 				 */
1338 				do {
1339 					struct rtdetq *n = rte->next;
1340 					m_freem(rte->m);
1341 					free(rte, M_MRTABLE6);
1342 					rte = n;
1343 				} while (rte != NULL);
1344 				MRT6STAT_INC(mrt6s_cache_cleanups);
1345 				n6expire[i]--;
1346 
1347 				*nptr = mfc->mf6c_next;
1348 				free(mfc, M_MRTABLE6);
1349 			} else {
1350 				nptr = &mfc->mf6c_next;
1351 			}
1352 		}
1353 	}
1354 	MFC6_UNLOCK();
1355 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1356 	    expire_upcalls, NULL);
1357 }
1358 
1359 /*
1360  * Packet forwarding routine once entry in the cache is made
1361  */
1362 static int
1363 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1364 {
1365 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1366 	mifi_t mifi, iif;
1367 	struct mif6 *mifp;
1368 	int plen = m->m_pkthdr.len;
1369 	struct in6_addr src0, dst0; /* copies for local work */
1370 	u_int32_t iszone, idzone, oszone, odzone;
1371 	int error = 0;
1372 
1373 /*
1374  * Macro to send packet on mif.  Since RSVP packets don't get counted on
1375  * input, they shouldn't get counted on output, so statistics keeping is
1376  * separate.
1377  */
1378 
1379 #define MC6_SEND(ip6, mifp, m) do {				\
1380 	if ((mifp)->m6_flags & MIFF_REGISTER)			\
1381 		register_send((ip6), (mifp), (m));		\
1382 	else							\
1383 		phyint_send((ip6), (mifp), (m));		\
1384 } while (/*CONSTCOND*/ 0)
1385 
1386 	/*
1387 	 * Don't forward if it didn't arrive from the parent mif
1388 	 * for its origin.
1389 	 */
1390 	mifi = rt->mf6c_parent;
1391 	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1392 		/* came in the wrong interface */
1393 		MRT6_DLOG(DEBUG_FORWARD,
1394 		    "wrong if: ifid %d mifi %d mififid %x", ifp->if_index,
1395 		    mifi, mif6table[mifi].m6_ifp->if_index);
1396 		MRT6STAT_INC(mrt6s_wrong_if);
1397 		rt->mf6c_wrong_if++;
1398 		/*
1399 		 * If we are doing PIM processing, and we are forwarding
1400 		 * packets on this interface, send a message to the
1401 		 * routing daemon.
1402 		 */
1403 		/* have to make sure this is a valid mif */
1404 		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1405 			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1406 				/*
1407 				 * Check the M_LOOP flag to avoid an
1408 				 * unnecessary PIM assert.
1409 				 * XXX: M_LOOP is an ad-hoc hack...
1410 				 */
1411 				static struct sockaddr_in6 sin6 =
1412 				{ sizeof(sin6), AF_INET6 };
1413 
1414 				struct mbuf *mm;
1415 				struct mrt6msg *im;
1416 #ifdef MRT6_OINIT
1417 				struct omrt6msg *oim;
1418 #endif
1419 
1420 				mm = m_copym(m, 0, sizeof(struct ip6_hdr),
1421 				    M_NOWAIT);
1422 				if (mm &&
1423 				    (!M_WRITABLE(mm) ||
1424 				     mm->m_len < sizeof(struct ip6_hdr)))
1425 					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1426 				if (mm == NULL)
1427 					return (ENOBUFS);
1428 
1429 #ifdef MRT6_OINIT
1430 				oim = NULL;
1431 #endif
1432 				im = NULL;
1433 				switch (V_ip6_mrouter_ver) {
1434 #ifdef MRT6_OINIT
1435 				case MRT6_OINIT:
1436 					oim = mtod(mm, struct omrt6msg *);
1437 					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1438 					oim->im6_mbz = 0;
1439 					break;
1440 #endif
1441 				case MRT6_INIT:
1442 					im = mtod(mm, struct mrt6msg *);
1443 					im->im6_msgtype = MRT6MSG_WRONGMIF;
1444 					im->im6_mbz = 0;
1445 					break;
1446 				default:
1447 					m_freem(mm);
1448 					return (EINVAL);
1449 				}
1450 
1451 				for (mifp = mif6table, iif = 0;
1452 				     iif < nummifs && mifp &&
1453 					     mifp->m6_ifp != ifp;
1454 				     mifp++, iif++)
1455 					;
1456 
1457 				switch (V_ip6_mrouter_ver) {
1458 #ifdef MRT6_OINIT
1459 				case MRT6_OINIT:
1460 					oim->im6_mif = iif;
1461 					sin6.sin6_addr = oim->im6_src;
1462 					break;
1463 #endif
1464 				case MRT6_INIT:
1465 					im->im6_mif = iif;
1466 					sin6.sin6_addr = im->im6_src;
1467 					break;
1468 				}
1469 
1470 				MRT6STAT_INC(mrt6s_upcalls);
1471 
1472 				if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1473 					MRT6_DLOG(DEBUG_ANY,
1474 					    "ip6_mrouter socket queue full");
1475 					MRT6STAT_INC(mrt6s_upq_sockfull);
1476 					return (ENOBUFS);
1477 				}	/* if socket Q full */
1478 			}		/* if PIM */
1479 		return (0);
1480 	}			/* if wrong iif */
1481 
1482 	/* If I sourced this packet, it counts as output, else it was input. */
1483 	if (m->m_pkthdr.rcvif == NULL) {
1484 		/* XXX: is rcvif really NULL when output?? */
1485 		mif6table[mifi].m6_pkt_out++;
1486 		mif6table[mifi].m6_bytes_out += plen;
1487 	} else {
1488 		mif6table[mifi].m6_pkt_in++;
1489 		mif6table[mifi].m6_bytes_in += plen;
1490 	}
1491 	rt->mf6c_pkt_cnt++;
1492 	rt->mf6c_byte_cnt += plen;
1493 
1494 	/*
1495 	 * For each mif, forward a copy of the packet if there are group
1496 	 * members downstream on the interface.
1497 	 */
1498 	src0 = ip6->ip6_src;
1499 	dst0 = ip6->ip6_dst;
1500 	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1501 	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1502 		IP6STAT_INC(ip6s_badscope);
1503 		return (error);
1504 	}
1505 	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1506 		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1507 			/*
1508 			 * check if the outgoing packet is going to break
1509 			 * a scope boundary.
1510 			 * XXX For packets through PIM register tunnel
1511 			 * interface, we believe a routing daemon.
1512 			 */
1513 			if (!(mif6table[rt->mf6c_parent].m6_flags &
1514 			      MIFF_REGISTER) &&
1515 			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1516 				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1517 				    &oszone) ||
1518 				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1519 				    &odzone) ||
1520 				    iszone != oszone ||
1521 				    idzone != odzone) {
1522 					IP6STAT_INC(ip6s_badscope);
1523 					continue;
1524 				}
1525 			}
1526 
1527 			mifp->m6_pkt_out++;
1528 			mifp->m6_bytes_out += plen;
1529 			MC6_SEND(ip6, mifp, m);
1530 		}
1531 	}
1532 	return (0);
1533 }
1534 
1535 static void
1536 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1537 {
1538 #ifdef MRT6DEBUG
1539 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1540 #endif
1541 	struct mbuf *mb_copy;
1542 	struct ifnet *ifp = mifp->m6_ifp;
1543 	int error = 0;
1544 	u_long linkmtu;
1545 
1546 	/*
1547 	 * Make a new reference to the packet; make sure that
1548 	 * the IPv6 header is actually copied, not just referenced,
1549 	 * so that ip6_output() only scribbles on the copy.
1550 	 */
1551 	mb_copy = m_copym(m, 0, M_COPYALL, M_NOWAIT);
1552 	if (mb_copy &&
1553 	    (!M_WRITABLE(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1554 		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1555 	if (mb_copy == NULL) {
1556 		return;
1557 	}
1558 	/* set MCAST flag to the outgoing packet */
1559 	mb_copy->m_flags |= M_MCAST;
1560 
1561 	/*
1562 	 * If we sourced the packet, call ip6_output since we may devide
1563 	 * the packet into fragments when the packet is too big for the
1564 	 * outgoing interface.
1565 	 * Otherwise, we can simply send the packet to the interface
1566 	 * sending queue.
1567 	 */
1568 	if (m->m_pkthdr.rcvif == NULL) {
1569 		struct ip6_moptions im6o;
1570 
1571 		im6o.im6o_multicast_ifp = ifp;
1572 		/* XXX: ip6_output will override ip6->ip6_hlim */
1573 		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1574 		im6o.im6o_multicast_loop = 1;
1575 		error = ip6_output(mb_copy, NULL, NULL, IPV6_FORWARDING, &im6o,
1576 		    NULL, NULL);
1577 
1578 		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1579 		    (uint16_t)(mifp - mif6table), error);
1580 		return;
1581 	}
1582 
1583 	/*
1584 	 * If configured to loop back multicasts by default,
1585 	 * loop back a copy now.
1586 	 */
1587 	if (in6_mcast_loop)
1588 		ip6_mloopback(ifp, m);
1589 
1590 	/*
1591 	 * Put the packet into the sending queue of the outgoing interface
1592 	 * if it would fit in the MTU of the interface.
1593 	 */
1594 	linkmtu = IN6_LINKMTU(ifp);
1595 	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1596 		struct sockaddr_in6 dst6;
1597 
1598 		bzero(&dst6, sizeof(dst6));
1599 		dst6.sin6_len = sizeof(struct sockaddr_in6);
1600 		dst6.sin6_family = AF_INET6;
1601 		dst6.sin6_addr = ip6->ip6_dst;
1602 
1603 		IP_PROBE(send, NULL, NULL, ip6, ifp, NULL, ip6);
1604 		/*
1605 		 * We just call if_output instead of nd6_output here, since
1606 		 * we need no ND for a multicast forwarded packet...right?
1607 		 */
1608 		m_clrprotoflags(m);	/* Avoid confusing lower layers. */
1609 		error = (*ifp->if_output)(ifp, mb_copy,
1610 		    (struct sockaddr *)&dst6, NULL);
1611 		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1612 		    (uint16_t)(mifp - mif6table), error);
1613 	} else {
1614 		/*
1615 		 * pMTU discovery is intentionally disabled by default, since
1616 		 * various router may notify pMTU in multicast, which can be
1617 		 * a DDoS to a router
1618 		 */
1619 		if (V_ip6_mcast_pmtu)
1620 			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1621 		else {
1622 			MRT6_DLOG(DEBUG_XMIT, " packet too big on %s o %s "
1623 			    "g %s size %d (discarded)", if_name(ifp),
1624 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1625 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1626 			    mb_copy->m_pkthdr.len);
1627 			m_freem(mb_copy); /* simply discard the packet */
1628 		}
1629 	}
1630 }
1631 
1632 static int
1633 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1634 {
1635 #ifdef MRT6DEBUG
1636 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1637 #endif
1638 	struct mbuf *mm;
1639 	int i, len = m->m_pkthdr.len;
1640 	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1641 	struct mrt6msg *im6;
1642 
1643 	MRT6_DLOG(DEBUG_ANY, "src %s dst %s",
1644 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1645 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1646 	PIM6STAT_INC(pim6s_snd_registers);
1647 
1648 	/* Make a copy of the packet to send to the user level process. */
1649 	mm = m_gethdr(M_NOWAIT, MT_DATA);
1650 	if (mm == NULL)
1651 		return (ENOBUFS);
1652 	mm->m_data += max_linkhdr;
1653 	mm->m_len = sizeof(struct ip6_hdr);
1654 
1655 	if ((mm->m_next = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL) {
1656 		m_freem(mm);
1657 		return (ENOBUFS);
1658 	}
1659 	i = MHLEN - M_LEADINGSPACE(mm);
1660 	if (i > len)
1661 		i = len;
1662 	mm = m_pullup(mm, i);
1663 	if (mm == NULL)
1664 		return (ENOBUFS);
1665 /* TODO: check it! */
1666 	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1667 
1668 	/*
1669 	 * Send message to routing daemon
1670 	 */
1671 	sin6.sin6_addr = ip6->ip6_src;
1672 
1673 	im6 = mtod(mm, struct mrt6msg *);
1674 	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1675 	im6->im6_mbz          = 0;
1676 
1677 	im6->im6_mif = mif - mif6table;
1678 
1679 	/* iif info is not given for reg. encap.n */
1680 	MRT6STAT_INC(mrt6s_upcalls);
1681 
1682 	if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1683 		MRT6_DLOG(DEBUG_ANY, "ip6_mrouter socket queue full");
1684 		MRT6STAT_INC(mrt6s_upq_sockfull);
1685 		return (ENOBUFS);
1686 	}
1687 	return (0);
1688 }
1689 
1690 /*
1691  * pim6_encapcheck() is called by the encap6_input() path at runtime to
1692  * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1693  * into the kernel.
1694  */
1695 static int
1696 pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
1697 {
1698 
1699 #ifdef DIAGNOSTIC
1700     KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1701 #endif
1702     if (proto != IPPROTO_PIM)
1703 	return 0;	/* not for us; reject the datagram. */
1704 
1705     return 64;		/* claim the datagram. */
1706 }
1707 
1708 /*
1709  * PIM sparse mode hook
1710  * Receives the pim control messages, and passes them up to the listening
1711  * socket, using rip6_input.
1712  * The only message processed is the REGISTER pim message; the pim header
1713  * is stripped off, and the inner packet is passed to register_mforward.
1714  */
1715 int
1716 pim6_input(struct mbuf **mp, int *offp, int proto)
1717 {
1718 	struct pim *pim; /* pointer to a pim struct */
1719 	struct ip6_hdr *ip6;
1720 	int pimlen;
1721 	struct mbuf *m = *mp;
1722 	int minlen;
1723 	int off = *offp;
1724 
1725 	PIM6STAT_INC(pim6s_rcv_total);
1726 
1727 	ip6 = mtod(m, struct ip6_hdr *);
1728 	pimlen = m->m_pkthdr.len - *offp;
1729 
1730 	/*
1731 	 * Validate lengths
1732 	 */
1733 	if (pimlen < PIM_MINLEN) {
1734 		PIM6STAT_INC(pim6s_rcv_tooshort);
1735 		MRT6_DLOG(DEBUG_PIM, "PIM packet too short");
1736 		m_freem(m);
1737 		return (IPPROTO_DONE);
1738 	}
1739 
1740 	/*
1741 	 * if the packet is at least as big as a REGISTER, go ahead
1742 	 * and grab the PIM REGISTER header size, to avoid another
1743 	 * possible m_pullup() later.
1744 	 *
1745 	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1746 	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1747 	 */
1748 	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1749 
1750 	/*
1751 	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1752 	 * possibly the PIM REGISTER header
1753 	 */
1754 #ifndef PULLDOWN_TEST
1755 	IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1756 	/* adjust pointer */
1757 	ip6 = mtod(m, struct ip6_hdr *);
1758 
1759 	/* adjust mbuf to point to the PIM header */
1760 	pim = (struct pim *)((caddr_t)ip6 + off);
1761 #else
1762 	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1763 	if (pim == NULL) {
1764 		PIM6STAT_INC(pim6s_rcv_tooshort);
1765 		return (IPPROTO_DONE);
1766 	}
1767 #endif
1768 
1769 #define PIM6_CHECKSUM
1770 #ifdef PIM6_CHECKSUM
1771 	{
1772 		int cksumlen;
1773 
1774 		/*
1775 		 * Validate checksum.
1776 		 * If PIM REGISTER, exclude the data packet
1777 		 */
1778 		if (pim->pim_type == PIM_REGISTER)
1779 			cksumlen = PIM_MINLEN;
1780 		else
1781 			cksumlen = pimlen;
1782 
1783 		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1784 			PIM6STAT_INC(pim6s_rcv_badsum);
1785 			MRT6_DLOG(DEBUG_PIM, "invalid checksum");
1786 			m_freem(m);
1787 			return (IPPROTO_DONE);
1788 		}
1789 	}
1790 #endif /* PIM_CHECKSUM */
1791 
1792 	/* PIM version check */
1793 	if (pim->pim_ver != PIM_VERSION) {
1794 		PIM6STAT_INC(pim6s_rcv_badversion);
1795 		MRT6_DLOG(DEBUG_ANY | DEBUG_ERR,
1796 		    "incorrect version %d, expecting %d",
1797 		    pim->pim_ver, PIM_VERSION);
1798 		m_freem(m);
1799 		return (IPPROTO_DONE);
1800 	}
1801 
1802 	if (pim->pim_type == PIM_REGISTER) {
1803 		/*
1804 		 * since this is a REGISTER, we'll make a copy of the register
1805 		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1806 		 * routing daemon.
1807 		 */
1808 		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1809 
1810 		struct mbuf *mcp;
1811 		struct ip6_hdr *eip6;
1812 		u_int32_t *reghdr;
1813 		int rc;
1814 #ifdef MRT6DEBUG
1815 		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1816 #endif
1817 
1818 		PIM6STAT_INC(pim6s_rcv_registers);
1819 
1820 		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1821 			MRT6_DLOG(DEBUG_PIM, "register mif not set: %d",
1822 			    reg_mif_num);
1823 			m_freem(m);
1824 			return (IPPROTO_DONE);
1825 		}
1826 
1827 		reghdr = (u_int32_t *)(pim + 1);
1828 
1829 		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1830 			goto pim6_input_to_daemon;
1831 
1832 		/*
1833 		 * Validate length
1834 		 */
1835 		if (pimlen < PIM6_REG_MINLEN) {
1836 			PIM6STAT_INC(pim6s_rcv_tooshort);
1837 			PIM6STAT_INC(pim6s_rcv_badregisters);
1838 			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "register packet "
1839 			    "size too small %d from %s",
1840 			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1841 			m_freem(m);
1842 			return (IPPROTO_DONE);
1843 		}
1844 
1845 		eip6 = (struct ip6_hdr *) (reghdr + 1);
1846 		MRT6_DLOG(DEBUG_PIM, "eip6: %s -> %s, eip6 plen %d",
1847 		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1848 		    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1849 		    ntohs(eip6->ip6_plen));
1850 
1851 		/* verify the version number of the inner packet */
1852 		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1853 			PIM6STAT_INC(pim6s_rcv_badregisters);
1854 			MRT6_DLOG(DEBUG_ANY, "invalid IP version (%d) "
1855 			    "of the inner packet",
1856 			    (eip6->ip6_vfc & IPV6_VERSION));
1857 			m_freem(m);
1858 			return (IPPROTO_NONE);
1859 		}
1860 
1861 		/* verify the inner packet is destined to a mcast group */
1862 		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1863 			PIM6STAT_INC(pim6s_rcv_badregisters);
1864 			MRT6_DLOG(DEBUG_PIM, "inner packet of register "
1865 			    "is not multicast %s",
1866 			    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1867 			m_freem(m);
1868 			return (IPPROTO_DONE);
1869 		}
1870 
1871 		/*
1872 		 * make a copy of the whole header to pass to the daemon later.
1873 		 */
1874 		mcp = m_copym(m, 0, off + PIM6_REG_MINLEN, M_NOWAIT);
1875 		if (mcp == NULL) {
1876 			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "pim register: "
1877 			    "could not copy register head");
1878 			m_freem(m);
1879 			return (IPPROTO_DONE);
1880 		}
1881 
1882 		/*
1883 		 * forward the inner ip6 packet; point m_data at the inner ip6.
1884 		 */
1885 		m_adj(m, off + PIM_MINLEN);
1886 		MRT6_DLOG(DEBUG_PIM, "forwarding decapsulated register: "
1887 		    "src %s, dst %s, mif %d",
1888 		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1889 		    ip6_sprintf(ip6bufd, &eip6->ip6_dst), reg_mif_num);
1890 
1891 		rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1892 				dst.sin6_family, 0);
1893 
1894 		/* prepare the register head to send to the mrouting daemon */
1895 		m = mcp;
1896 	}
1897 
1898 	/*
1899 	 * Pass the PIM message up to the daemon; if it is a register message
1900 	 * pass the 'head' only up to the daemon. This includes the
1901 	 * encapsulator ip6 header, pim header, register header and the
1902 	 * encapsulated ip6 header.
1903 	 */
1904   pim6_input_to_daemon:
1905 	rip6_input(&m, offp, proto);
1906 	return (IPPROTO_DONE);
1907 }
1908 
1909 static int
1910 ip6_mroute_modevent(module_t mod, int type, void *unused)
1911 {
1912 
1913 	switch (type) {
1914 	case MOD_LOAD:
1915 		MROUTER6_LOCK_INIT();
1916 		MFC6_LOCK_INIT();
1917 		MIF6_LOCK_INIT();
1918 
1919 		pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM,
1920 			pim6_encapcheck,
1921 			(const struct protosw *)&in6_pim_protosw, NULL);
1922 		if (pim6_encap_cookie == NULL) {
1923 			printf("ip6_mroute: unable to attach pim6 encap\n");
1924 			MIF6_LOCK_DESTROY();
1925 			MFC6_LOCK_DESTROY();
1926 			MROUTER6_LOCK_DESTROY();
1927 			return (EINVAL);
1928 		}
1929 
1930 		ip6_mforward = X_ip6_mforward;
1931 		ip6_mrouter_done = X_ip6_mrouter_done;
1932 		ip6_mrouter_get = X_ip6_mrouter_get;
1933 		ip6_mrouter_set = X_ip6_mrouter_set;
1934 		mrt6_ioctl = X_mrt6_ioctl;
1935 		break;
1936 
1937 	case MOD_UNLOAD:
1938 		if (V_ip6_mrouter != NULL)
1939 			return EINVAL;
1940 
1941 		if (pim6_encap_cookie) {
1942 			encap_detach(pim6_encap_cookie);
1943 			pim6_encap_cookie = NULL;
1944 		}
1945 		X_ip6_mrouter_done();
1946 		ip6_mforward = NULL;
1947 		ip6_mrouter_done = NULL;
1948 		ip6_mrouter_get = NULL;
1949 		ip6_mrouter_set = NULL;
1950 		mrt6_ioctl = NULL;
1951 
1952 		MIF6_LOCK_DESTROY();
1953 		MFC6_LOCK_DESTROY();
1954 		MROUTER6_LOCK_DESTROY();
1955 		break;
1956 
1957 	default:
1958 		return (EOPNOTSUPP);
1959 	}
1960 
1961 	return (0);
1962 }
1963 
1964 static moduledata_t ip6_mroutemod = {
1965 	"ip6_mroute",
1966 	ip6_mroute_modevent,
1967 	0
1968 };
1969 
1970 DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PROTO_MC, SI_ORDER_ANY);
1971