xref: /freebsd/sys/netinet6/ip6_mroute.c (revision d0b2dbfa0ecf2bbc9709efc5e20baf8e4b44bbbf)
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 #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_private.h>
110 #include <net/if_types.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 int pim6_encapcheck(const struct mbuf *, int, int, void *);
141 static int pim6_input(struct mbuf *, int, int, void *);
142 
143 static const struct encap_config ipv6_encap_cfg = {
144 	.proto = IPPROTO_PIM,
145 	.min_length = sizeof(struct ip6_hdr) + PIM_MINLEN,
146 	.exact_match = 8,
147 	.check = pim6_encapcheck,
148 	.input = pim6_input
149 };
150 
151 VNET_DEFINE_STATIC(int, ip6_mrouter_ver) = 0;
152 #define	V_ip6_mrouter_ver	VNET(ip6_mrouter_ver)
153 
154 SYSCTL_DECL(_net_inet6);
155 SYSCTL_DECL(_net_inet6_ip6);
156 static SYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim,
157     CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
158     "PIM");
159 
160 static struct mrt6stat mrt6stat;
161 SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
162     &mrt6stat, mrt6stat,
163     "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
164 
165 #define	MRT6STAT_INC(name)	mrt6stat.name += 1
166 #define NO_RTE_FOUND	0x1
167 #define RTE_FOUND	0x2
168 
169 static struct sx mrouter6_mtx;
170 #define	MROUTER6_LOCKPTR()	(&mrouter6_mtx)
171 #define	MROUTER6_LOCK()		sx_xlock(MROUTER6_LOCKPTR())
172 #define	MROUTER6_UNLOCK()	sx_xunlock(MROUTER6_LOCKPTR())
173 #define	MROUTER6_LOCK_ASSERT()	sx_assert(MROUTER6_LOCKPTR(), SA_XLOCKED
174 #define	MROUTER6_LOCK_INIT()	sx_init(MROUTER6_LOCKPTR(), "mrouter6")
175 #define	MROUTER6_LOCK_DESTROY()	sx_destroy(MROUTER6_LOCKPTR())
176 
177 static struct mf6c *mf6ctable[MF6CTBLSIZ];
178 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
179     &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
180     "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
181     "netinet6/ip6_mroute.h)");
182 
183 static struct mtx mfc6_mtx;
184 #define	MFC6_LOCKPTR()		(&mfc6_mtx)
185 #define	MFC6_LOCK()		mtx_lock(MFC6_LOCKPTR())
186 #define	MFC6_UNLOCK()		mtx_unlock(MFC6_LOCKPTR())
187 #define	MFC6_LOCK_ASSERT()	mtx_assert(MFC6_LOCKPTR(), MA_OWNED)
188 #define	MFC6_LOCK_INIT()	mtx_init(MFC6_LOCKPTR(),		\
189 				    "IPv6 multicast forwarding cache",	\
190 				    NULL, MTX_DEF)
191 #define	MFC6_LOCK_DESTROY()	mtx_destroy(MFC6_LOCKPTR())
192 
193 static u_char n6expire[MF6CTBLSIZ];
194 
195 static struct mif6 mif6table[MAXMIFS];
196 static int
197 sysctl_mif6table(SYSCTL_HANDLER_ARGS)
198 {
199 	struct mif6_sctl *out;
200 	int error;
201 
202 	out = malloc(sizeof(struct mif6_sctl) * MAXMIFS, M_TEMP,
203 	    M_WAITOK | M_ZERO);
204 	for (int i = 0; i < MAXMIFS; i++) {
205 		out[i].m6_flags		= mif6table[i].m6_flags;
206 		out[i].m6_rate_limit	= mif6table[i].m6_rate_limit;
207 		out[i].m6_lcl_addr	= mif6table[i].m6_lcl_addr;
208 		if (mif6table[i].m6_ifp != NULL)
209 			out[i].m6_ifp	= mif6table[i].m6_ifp->if_index;
210 		else
211 			out[i].m6_ifp	= 0;
212 		out[i].m6_pkt_in	= mif6table[i].m6_pkt_in;
213 		out[i].m6_pkt_out	= mif6table[i].m6_pkt_out;
214 		out[i].m6_bytes_in	= mif6table[i].m6_bytes_in;
215 		out[i].m6_bytes_out	= mif6table[i].m6_bytes_out;
216 	}
217 	error = SYSCTL_OUT(req, out, sizeof(struct mif6_sctl) * MAXMIFS);
218 	free(out, M_TEMP);
219 	return (error);
220 }
221 SYSCTL_PROC(_net_inet6_ip6, OID_AUTO, mif6table,
222     CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_NEEDGIANT,
223     NULL, 0, sysctl_mif6table, "S,mif6_sctl[MAXMIFS]",
224     "IPv6 Multicast Interfaces (struct mif6_sctl[MAXMIFS], "
225     "netinet6/ip6_mroute.h)");
226 
227 static struct mtx mif6_mtx;
228 #define	MIF6_LOCKPTR()		(&mif6_mtx)
229 #define	MIF6_LOCK()		mtx_lock(MIF6_LOCKPTR())
230 #define	MIF6_UNLOCK()		mtx_unlock(MIF6_LOCKPTR())
231 #define	MIF6_LOCK_ASSERT()	mtx_assert(MIF6_LOCKPTR(), MA_OWNED)
232 #define	MIF6_LOCK_INIT()	\
233 	mtx_init(MIF6_LOCKPTR(), "IPv6 multicast interfaces", NULL, MTX_DEF)
234 #define	MIF6_LOCK_DESTROY()	mtx_destroy(MIF6_LOCKPTR())
235 
236 #ifdef MRT6DEBUG
237 VNET_DEFINE_STATIC(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 VNET_DEFINE_STATIC(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, "%s: socket %p", __func__, so);
558 
559 	if (v != 1)
560 		return (ENOPROTOOPT);
561 
562 	MROUTER6_LOCK();
563 
564 	if (V_ip6_mrouter != NULL) {
565 		MROUTER6_UNLOCK();
566 		return (EADDRINUSE);
567 	}
568 
569 	V_ip6_mrouter = so;
570 	V_ip6_mrouter_ver = cmd;
571 
572 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
573 	bzero((caddr_t)n6expire, sizeof(n6expire));
574 
575 	V_pim6 = 0;/* used for stubbing out/in pim stuff */
576 
577 	callout_init_mtx(&expire_upcalls_ch, MFC6_LOCKPTR(), 0);
578 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
579 	    expire_upcalls, NULL);
580 
581 	MROUTER6_UNLOCK();
582 
583 	MRT6_DLOG(DEBUG_ANY, "finished");
584 
585 	return (0);
586 }
587 
588 /*
589  * Disable IPv6 multicast forwarding.
590  */
591 int
592 X_ip6_mrouter_done(void)
593 {
594 	mifi_t mifi;
595 	u_long i;
596 	struct mf6c *rt;
597 	struct rtdetq *rte;
598 
599 	MROUTER6_LOCK();
600 
601 	if (V_ip6_mrouter == NULL) {
602 		MROUTER6_UNLOCK();
603 		return (EINVAL);
604 	}
605 
606 	/*
607 	 * For each phyint in use, disable promiscuous reception of all IPv6
608 	 * multicasts.
609 	 */
610 	for (mifi = 0; mifi < nummifs; mifi++) {
611 		if (mif6table[mifi].m6_ifp &&
612 		    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
613 			if_allmulti(mif6table[mifi].m6_ifp, 0);
614 		}
615 	}
616 	bzero((caddr_t)mif6table, sizeof(mif6table));
617 	nummifs = 0;
618 
619 	V_pim6 = 0; /* used to stub out/in pim specific code */
620 
621 	/*
622 	 * Free all multicast forwarding cache entries.
623 	 */
624 	MFC6_LOCK();
625 	for (i = 0; i < MF6CTBLSIZ; i++) {
626 		rt = mf6ctable[i];
627 		while (rt) {
628 			struct mf6c *frt;
629 
630 			for (rte = rt->mf6c_stall; rte != NULL; ) {
631 				struct rtdetq *n = rte->next;
632 
633 				m_freem(rte->m);
634 				free(rte, M_MRTABLE6);
635 				rte = n;
636 			}
637 			frt = rt;
638 			rt = rt->mf6c_next;
639 			free(frt, M_MRTABLE6);
640 		}
641 	}
642 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
643 	MFC6_UNLOCK();
644 
645 	callout_drain(&expire_upcalls_ch);
646 
647 	/*
648 	 * Reset register interface
649 	 */
650 	if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
651 		if_detach(multicast_register_if6);
652 		if_free(multicast_register_if6);
653 		reg_mif_num = (mifi_t)-1;
654 		multicast_register_if6 = NULL;
655 	}
656 
657 	V_ip6_mrouter = NULL;
658 	V_ip6_mrouter_ver = 0;
659 
660 	MROUTER6_UNLOCK();
661 	MRT6_DLOG(DEBUG_ANY, "finished");
662 
663 	return (0);
664 }
665 
666 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
667 
668 /*
669  * Add a mif to the mif table
670  */
671 static int
672 add_m6if(struct mif6ctl *mifcp)
673 {
674 	struct epoch_tracker et;
675 	struct mif6 *mifp;
676 	struct ifnet *ifp;
677 	int error;
678 
679 	MIF6_LOCK();
680 
681 	if (mifcp->mif6c_mifi >= MAXMIFS) {
682 		MIF6_UNLOCK();
683 		return (EINVAL);
684 	}
685 	mifp = mif6table + mifcp->mif6c_mifi;
686 	if (mifp->m6_ifp != NULL) {
687 		MIF6_UNLOCK();
688 		return (EADDRINUSE); /* XXX: is it appropriate? */
689 	}
690 
691 	NET_EPOCH_ENTER(et);
692 	if ((ifp = ifnet_byindex(mifcp->mif6c_pifi)) == NULL) {
693 		NET_EPOCH_EXIT(et);
694 		MIF6_UNLOCK();
695 		return (ENXIO);
696 	}
697 	NET_EPOCH_EXIT(et);	/* XXXGL: unsafe ifp */
698 
699 	if (mifcp->mif6c_flags & MIFF_REGISTER) {
700 		if (reg_mif_num == (mifi_t)-1) {
701 			ifp = if_alloc(IFT_OTHER);
702 
703 			if_initname(ifp, "register_mif", 0);
704 			ifp->if_flags |= IFF_LOOPBACK;
705 			if_attach(ifp);
706 			multicast_register_if6 = ifp;
707 			reg_mif_num = mifcp->mif6c_mifi;
708 			/*
709 			 * it is impossible to guess the ifindex of the
710 			 * register interface.  So mif6c_pifi is automatically
711 			 * calculated.
712 			 */
713 			mifcp->mif6c_pifi = ifp->if_index;
714 		} else {
715 			ifp = multicast_register_if6;
716 		}
717 	} else {
718 		/* Make sure the interface supports multicast */
719 		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
720 			MIF6_UNLOCK();
721 			return (EOPNOTSUPP);
722 		}
723 
724 		error = if_allmulti(ifp, 1);
725 		if (error) {
726 			MIF6_UNLOCK();
727 			return (error);
728 		}
729 	}
730 
731 	mifp->m6_flags     = mifcp->mif6c_flags;
732 	mifp->m6_ifp       = ifp;
733 
734 	/* initialize per mif pkt counters */
735 	mifp->m6_pkt_in    = 0;
736 	mifp->m6_pkt_out   = 0;
737 	mifp->m6_bytes_in  = 0;
738 	mifp->m6_bytes_out = 0;
739 
740 	/* Adjust nummifs up if the mifi is higher than nummifs */
741 	if (nummifs <= mifcp->mif6c_mifi)
742 		nummifs = mifcp->mif6c_mifi + 1;
743 
744 	MIF6_UNLOCK();
745 	MRT6_DLOG(DEBUG_ANY, "mif #%d, phyint %s", mifcp->mif6c_mifi,
746 	    if_name(ifp));
747 
748 	return (0);
749 }
750 
751 /*
752  * Delete a mif from the mif table
753  */
754 static int
755 del_m6if_locked(mifi_t *mifip)
756 {
757 	struct mif6 *mifp = mif6table + *mifip;
758 	mifi_t mifi;
759 	struct ifnet *ifp;
760 
761 	MIF6_LOCK_ASSERT();
762 
763 	if (*mifip >= nummifs)
764 		return (EINVAL);
765 	if (mifp->m6_ifp == NULL)
766 		return (EINVAL);
767 
768 	if (!(mifp->m6_flags & MIFF_REGISTER)) {
769 		/* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
770 		ifp = mifp->m6_ifp;
771 		if_allmulti(ifp, 0);
772 	} else {
773 		if (reg_mif_num != (mifi_t)-1 &&
774 		    multicast_register_if6 != NULL) {
775 			if_detach(multicast_register_if6);
776 			if_free(multicast_register_if6);
777 			reg_mif_num = (mifi_t)-1;
778 			multicast_register_if6 = NULL;
779 		}
780 	}
781 
782 	bzero((caddr_t)mifp, sizeof(*mifp));
783 
784 	/* Adjust nummifs down */
785 	for (mifi = nummifs; mifi > 0; mifi--)
786 		if (mif6table[mifi - 1].m6_ifp)
787 			break;
788 	nummifs = mifi;
789 	MRT6_DLOG(DEBUG_ANY, "mif %d, nummifs %d", *mifip, nummifs);
790 
791 	return (0);
792 }
793 
794 static int
795 del_m6if(mifi_t *mifip)
796 {
797 	int cc;
798 
799 	MIF6_LOCK();
800 	cc = del_m6if_locked(mifip);
801 	MIF6_UNLOCK();
802 
803 	return (cc);
804 }
805 
806 /*
807  * Add an mfc entry
808  */
809 static int
810 add_m6fc(struct mf6cctl *mfccp)
811 {
812 	struct mf6c *rt;
813 	u_long hash;
814 	struct rtdetq *rte;
815 	u_short nstl;
816 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
817 
818 	MFC6_LOCK();
819 
820 	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
821 		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
822 
823 	/* If an entry already exists, just update the fields */
824 	if (rt) {
825 		MRT6_DLOG(DEBUG_MFC, "no upcall o %s g %s p %x",
826 		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
827 		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
828 		    mfccp->mf6cc_parent);
829 
830 		rt->mf6c_parent = mfccp->mf6cc_parent;
831 		rt->mf6c_ifset = mfccp->mf6cc_ifset;
832 
833 		MFC6_UNLOCK();
834 		return (0);
835 	}
836 
837 	/*
838 	 * Find the entry for which the upcall was made and update
839 	 */
840 	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
841 			mfccp->mf6cc_mcastgrp.sin6_addr);
842 	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
843 		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
844 				       &mfccp->mf6cc_origin.sin6_addr) &&
845 		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
846 				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
847 		    (rt->mf6c_stall != NULL)) {
848 			if (nstl++)
849 				log(LOG_ERR,
850 				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
851 				    "multiple kernel entries",
852 				    ip6_sprintf(ip6bufo,
853 					    &mfccp->mf6cc_origin.sin6_addr),
854 				    ip6_sprintf(ip6bufg,
855 					    &mfccp->mf6cc_mcastgrp.sin6_addr),
856 				    mfccp->mf6cc_parent, rt->mf6c_stall);
857 
858 			MRT6_DLOG(DEBUG_MFC, "o %s g %s p %x dbg %p",
859 			    ip6_sprintf(ip6bufo,
860 			    &mfccp->mf6cc_origin.sin6_addr),
861 			    ip6_sprintf(ip6bufg,
862 				&mfccp->mf6cc_mcastgrp.sin6_addr),
863 			    mfccp->mf6cc_parent, rt->mf6c_stall);
864 
865 			rt->mf6c_origin     = mfccp->mf6cc_origin;
866 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
867 			rt->mf6c_parent     = mfccp->mf6cc_parent;
868 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
869 			/* initialize pkt counters per src-grp */
870 			rt->mf6c_pkt_cnt    = 0;
871 			rt->mf6c_byte_cnt   = 0;
872 			rt->mf6c_wrong_if   = 0;
873 
874 			rt->mf6c_expire = 0;	/* Don't clean this guy up */
875 			n6expire[hash]--;
876 
877 			/* free packets Qed at the end of this entry */
878 			for (rte = rt->mf6c_stall; rte != NULL; ) {
879 				struct rtdetq *n = rte->next;
880 				ip6_mdq(rte->m, rte->ifp, rt);
881 				m_freem(rte->m);
882 #ifdef UPCALL_TIMING
883 				collate(&(rte->t));
884 #endif /* UPCALL_TIMING */
885 				free(rte, M_MRTABLE6);
886 				rte = n;
887 			}
888 			rt->mf6c_stall = NULL;
889 		}
890 	}
891 
892 	/*
893 	 * It is possible that an entry is being inserted without an upcall
894 	 */
895 	if (nstl == 0) {
896 		MRT6_DLOG(DEBUG_MFC, "no upcall h %lu o %s g %s p %x", hash,
897 		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
898 		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
899 		    mfccp->mf6cc_parent);
900 
901 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
902 			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
903 					       &mfccp->mf6cc_origin.sin6_addr)&&
904 			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
905 					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
906 				rt->mf6c_origin     = mfccp->mf6cc_origin;
907 				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
908 				rt->mf6c_parent     = mfccp->mf6cc_parent;
909 				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
910 				/* initialize pkt counters per src-grp */
911 				rt->mf6c_pkt_cnt    = 0;
912 				rt->mf6c_byte_cnt   = 0;
913 				rt->mf6c_wrong_if   = 0;
914 
915 				if (rt->mf6c_expire)
916 					n6expire[hash]--;
917 				rt->mf6c_expire	   = 0;
918 			}
919 		}
920 		if (rt == NULL) {
921 			/* no upcall, so make a new entry */
922 			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
923 						  M_NOWAIT);
924 			if (rt == NULL) {
925 				MFC6_UNLOCK();
926 				return (ENOBUFS);
927 			}
928 
929 			/* insert new entry at head of hash chain */
930 			rt->mf6c_origin     = mfccp->mf6cc_origin;
931 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
932 			rt->mf6c_parent     = mfccp->mf6cc_parent;
933 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
934 			/* initialize pkt counters per src-grp */
935 			rt->mf6c_pkt_cnt    = 0;
936 			rt->mf6c_byte_cnt   = 0;
937 			rt->mf6c_wrong_if   = 0;
938 			rt->mf6c_expire     = 0;
939 			rt->mf6c_stall = NULL;
940 
941 			/* link into table */
942 			rt->mf6c_next  = mf6ctable[hash];
943 			mf6ctable[hash] = rt;
944 		}
945 	}
946 
947 	MFC6_UNLOCK();
948 	return (0);
949 }
950 
951 #ifdef UPCALL_TIMING
952 /*
953  * collect delay statistics on the upcalls
954  */
955 static void
956 collate(struct timeval *t)
957 {
958 	u_long d;
959 	struct timeval tp;
960 	u_long delta;
961 
962 	GET_TIME(tp);
963 
964 	if (TV_LT(*t, tp))
965 	{
966 		TV_DELTA(tp, *t, delta);
967 
968 		d = delta >> 10;
969 		if (d > UPCALL_MAX)
970 			d = UPCALL_MAX;
971 
972 		++upcall_data[d];
973 	}
974 }
975 #endif /* UPCALL_TIMING */
976 
977 /*
978  * Delete an mfc entry
979  */
980 static int
981 del_m6fc(struct mf6cctl *mfccp)
982 {
983 #ifdef MRT6DEBUG
984 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
985 #endif
986 	struct sockaddr_in6	origin;
987 	struct sockaddr_in6	mcastgrp;
988 	struct mf6c		*rt;
989 	struct mf6c		**nptr;
990 	u_long		hash;
991 
992 	origin = mfccp->mf6cc_origin;
993 	mcastgrp = mfccp->mf6cc_mcastgrp;
994 	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
995 
996 	MRT6_DLOG(DEBUG_MFC, "orig %s mcastgrp %s",
997 	    ip6_sprintf(ip6bufo, &origin.sin6_addr),
998 	    ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
999 
1000 	MFC6_LOCK();
1001 
1002 	nptr = &mf6ctable[hash];
1003 	while ((rt = *nptr) != NULL) {
1004 		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
1005 				       &rt->mf6c_origin.sin6_addr) &&
1006 		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1007 				       &rt->mf6c_mcastgrp.sin6_addr) &&
1008 		    rt->mf6c_stall == NULL)
1009 			break;
1010 
1011 		nptr = &rt->mf6c_next;
1012 	}
1013 	if (rt == NULL) {
1014 		MFC6_UNLOCK();
1015 		return (EADDRNOTAVAIL);
1016 	}
1017 
1018 	*nptr = rt->mf6c_next;
1019 	free(rt, M_MRTABLE6);
1020 
1021 	MFC6_UNLOCK();
1022 
1023 	return (0);
1024 }
1025 
1026 static int
1027 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1028 {
1029 
1030 	if (s) {
1031 		if (sbappendaddr(&s->so_rcv,
1032 				 (struct sockaddr *)src,
1033 				 mm, (struct mbuf *)0) != 0) {
1034 			sorwakeup(s);
1035 			return (0);
1036 		} else
1037 			soroverflow(s);
1038 	}
1039 	m_freem(mm);
1040 	return (-1);
1041 }
1042 
1043 /*
1044  * IPv6 multicast forwarding function. This function assumes that the packet
1045  * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1046  * pointed to by "ifp", and the packet is to be relayed to other networks
1047  * that have members of the packet's destination IPv6 multicast group.
1048  *
1049  * The packet is returned unscathed to the caller, unless it is
1050  * erroneous, in which case a non-zero return value tells the caller to
1051  * discard it.
1052  *
1053  * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1054  * this function is called in the originating context (i.e., not when
1055  * forwarding a packet from other node).  ip6_output(), which is currently the
1056  * only function that calls this function is called in the originating context,
1057  * explicitly ensures this condition.  It is caller's responsibility to ensure
1058  * that if this function is called from somewhere else in the originating
1059  * context in the future.
1060  */
1061 int
1062 X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1063 {
1064 	struct rtdetq *rte;
1065 	struct mbuf *mb0;
1066 	struct mf6c *rt;
1067 	struct mif6 *mifp;
1068 	struct mbuf *mm;
1069 	u_long hash;
1070 	mifi_t mifi;
1071 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1072 #ifdef UPCALL_TIMING
1073 	struct timeval tp;
1074 
1075 	GET_TIME(tp);
1076 #endif /* UPCALL_TIMING */
1077 
1078 	MRT6_DLOG(DEBUG_FORWARD, "src %s, dst %s, ifindex %d",
1079 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1080 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst), ifp->if_index);
1081 
1082 	/*
1083 	 * Don't forward a packet with Hop limit of zero or one,
1084 	 * or a packet destined to a local-only group.
1085 	 */
1086 	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1087 	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1088 		return (0);
1089 	ip6->ip6_hlim--;
1090 
1091 	/*
1092 	 * Source address check: do not forward packets with unspecified
1093 	 * source. It was discussed in July 2000, on ipngwg mailing list.
1094 	 * This is rather more serious than unicast cases, because some
1095 	 * MLD packets can be sent with the unspecified source address
1096 	 * (although such packets must normally set 1 to the hop limit field).
1097 	 */
1098 	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1099 		IP6STAT_INC(ip6s_cantforward);
1100 		if (V_ip6_log_cannot_forward && ip6_log_ratelimit()) {
1101 			log(LOG_DEBUG,
1102 			    "cannot forward "
1103 			    "from %s to %s nxt %d received on %s\n",
1104 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1105 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1106 			    ip6->ip6_nxt,
1107 			    if_name(m->m_pkthdr.rcvif));
1108 		}
1109 		return (0);
1110 	}
1111 
1112 	MFC6_LOCK();
1113 
1114 	/*
1115 	 * Determine forwarding mifs from the forwarding cache table
1116 	 */
1117 	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1118 	MRT6STAT_INC(mrt6s_mfc_lookups);
1119 
1120 	/* Entry exists, so forward if necessary */
1121 	if (rt) {
1122 		MFC6_UNLOCK();
1123 		return (ip6_mdq(m, ifp, rt));
1124 	}
1125 
1126 	/*
1127 	 * If we don't have a route for packet's origin,
1128 	 * Make a copy of the packet & send message to routing daemon.
1129 	 */
1130 	MRT6STAT_INC(mrt6s_no_route);
1131 	MRT6_DLOG(DEBUG_FORWARD | DEBUG_MFC, "no rte s %s g %s",
1132 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1133 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1134 
1135 	/*
1136 	 * Allocate mbufs early so that we don't do extra work if we
1137 	 * are just going to fail anyway.
1138 	 */
1139 	rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6, M_NOWAIT);
1140 	if (rte == NULL) {
1141 		MFC6_UNLOCK();
1142 		return (ENOBUFS);
1143 	}
1144 	mb0 = m_copym(m, 0, M_COPYALL, M_NOWAIT);
1145 	/*
1146 	 * Pullup packet header if needed before storing it,
1147 	 * as other references may modify it in the meantime.
1148 	 */
1149 	if (mb0 && (!M_WRITABLE(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1150 		mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1151 	if (mb0 == NULL) {
1152 		free(rte, M_MRTABLE6);
1153 		MFC6_UNLOCK();
1154 		return (ENOBUFS);
1155 	}
1156 
1157 	/* is there an upcall waiting for this packet? */
1158 	hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1159 	for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1160 		if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1161 		    &rt->mf6c_origin.sin6_addr) &&
1162 		    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1163 		    &rt->mf6c_mcastgrp.sin6_addr) && (rt->mf6c_stall != NULL))
1164 			break;
1165 	}
1166 
1167 	if (rt == NULL) {
1168 		struct mrt6msg *im;
1169 #ifdef MRT6_OINIT
1170 		struct omrt6msg *oim;
1171 #endif
1172 		/* no upcall, so make a new entry */
1173 		rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, M_NOWAIT);
1174 		if (rt == NULL) {
1175 			free(rte, M_MRTABLE6);
1176 			m_freem(mb0);
1177 			MFC6_UNLOCK();
1178 			return (ENOBUFS);
1179 		}
1180 		/*
1181 		 * Make a copy of the header to send to the user
1182 		 * level process
1183 		 */
1184 		mm = m_copym(mb0, 0, sizeof(struct ip6_hdr), M_NOWAIT);
1185 		if (mm == NULL) {
1186 			free(rte, M_MRTABLE6);
1187 			m_freem(mb0);
1188 			free(rt, M_MRTABLE6);
1189 			MFC6_UNLOCK();
1190 			return (ENOBUFS);
1191 		}
1192 
1193 		/*
1194 		 * Send message to routing daemon
1195 		 */
1196 		sin6.sin6_addr = ip6->ip6_src;
1197 		im = NULL;
1198 #ifdef MRT6_OINIT
1199 		oim = NULL;
1200 #endif
1201 		switch (V_ip6_mrouter_ver) {
1202 #ifdef MRT6_OINIT
1203 		case MRT6_OINIT:
1204 			oim = mtod(mm, struct omrt6msg *);
1205 			oim->im6_msgtype = MRT6MSG_NOCACHE;
1206 			oim->im6_mbz = 0;
1207 			break;
1208 #endif
1209 		case MRT6_INIT:
1210 			im = mtod(mm, struct mrt6msg *);
1211 			im->im6_msgtype = MRT6MSG_NOCACHE;
1212 			im->im6_mbz = 0;
1213 			break;
1214 		default:
1215 			free(rte, M_MRTABLE6);
1216 			m_freem(mb0);
1217 			free(rt, M_MRTABLE6);
1218 			MFC6_UNLOCK();
1219 			return (EINVAL);
1220 		}
1221 
1222 		MRT6_DLOG(DEBUG_FORWARD, "getting the iif info in the kernel");
1223 		for (mifp = mif6table, mifi = 0;
1224 		    mifi < nummifs && mifp->m6_ifp != ifp; mifp++, mifi++)
1225 				;
1226 
1227 		switch (V_ip6_mrouter_ver) {
1228 #ifdef MRT6_OINIT
1229 		case MRT6_OINIT:
1230 			oim->im6_mif = mifi;
1231 			break;
1232 #endif
1233 		case MRT6_INIT:
1234 			im->im6_mif = mifi;
1235 			break;
1236 		}
1237 
1238 		if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1239 			log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1240 			    "socket queue full\n");
1241 			MRT6STAT_INC(mrt6s_upq_sockfull);
1242 			free(rte, M_MRTABLE6);
1243 			m_freem(mb0);
1244 			free(rt, M_MRTABLE6);
1245 			MFC6_UNLOCK();
1246 			return (ENOBUFS);
1247 		}
1248 
1249 		MRT6STAT_INC(mrt6s_upcalls);
1250 
1251 		/* insert new entry at head of hash chain */
1252 		bzero(rt, sizeof(*rt));
1253 		rt->mf6c_origin.sin6_family = AF_INET6;
1254 		rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1255 		rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1256 		rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1257 		rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1258 		rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1259 		rt->mf6c_expire = UPCALL_EXPIRE;
1260 		n6expire[hash]++;
1261 		rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1262 
1263 		/* link into table */
1264 		rt->mf6c_next  = mf6ctable[hash];
1265 		mf6ctable[hash] = rt;
1266 		/* Add this entry to the end of the queue */
1267 		rt->mf6c_stall = rte;
1268 	} else {
1269 		/* determine if q has overflowed */
1270 		struct rtdetq **p;
1271 		int npkts = 0;
1272 
1273 		for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1274 			if (++npkts > MAX_UPQ6) {
1275 				MRT6STAT_INC(mrt6s_upq_ovflw);
1276 				free(rte, M_MRTABLE6);
1277 				m_freem(mb0);
1278 				MFC6_UNLOCK();
1279 				return (0);
1280 			}
1281 
1282 		/* Add this entry to the end of the queue */
1283 		*p = rte;
1284 	}
1285 
1286 	rte->next = NULL;
1287 	rte->m = mb0;
1288 	rte->ifp = ifp;
1289 #ifdef UPCALL_TIMING
1290 	rte->t = tp;
1291 #endif /* UPCALL_TIMING */
1292 
1293 	MFC6_UNLOCK();
1294 
1295 	return (0);
1296 }
1297 
1298 /*
1299  * Clean up cache entries if upcalls are not serviced
1300  * Call from the Slow Timeout mechanism, every half second.
1301  */
1302 static void
1303 expire_upcalls(void *unused)
1304 {
1305 #ifdef MRT6DEBUG
1306 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1307 #endif
1308 	struct rtdetq *rte;
1309 	struct mf6c *mfc, **nptr;
1310 	u_long i;
1311 
1312 	MFC6_LOCK_ASSERT();
1313 
1314 	for (i = 0; i < MF6CTBLSIZ; i++) {
1315 		if (n6expire[i] == 0)
1316 			continue;
1317 		nptr = &mf6ctable[i];
1318 		while ((mfc = *nptr) != NULL) {
1319 			rte = mfc->mf6c_stall;
1320 			/*
1321 			 * Skip real cache entries
1322 			 * Make sure it wasn't marked to not expire (shouldn't happen)
1323 			 * If it expires now
1324 			 */
1325 			if (rte != NULL &&
1326 			    mfc->mf6c_expire != 0 &&
1327 			    --mfc->mf6c_expire == 0) {
1328 				MRT6_DLOG(DEBUG_EXPIRE, "expiring (%s %s)",
1329 				    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1330 				    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1331 				/*
1332 				 * drop all the packets
1333 				 * free the mbuf with the pkt, if, timing info
1334 				 */
1335 				do {
1336 					struct rtdetq *n = rte->next;
1337 					m_freem(rte->m);
1338 					free(rte, M_MRTABLE6);
1339 					rte = n;
1340 				} while (rte != NULL);
1341 				MRT6STAT_INC(mrt6s_cache_cleanups);
1342 				n6expire[i]--;
1343 
1344 				*nptr = mfc->mf6c_next;
1345 				free(mfc, M_MRTABLE6);
1346 			} else {
1347 				nptr = &mfc->mf6c_next;
1348 			}
1349 		}
1350 	}
1351 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1352 	    expire_upcalls, NULL);
1353 }
1354 
1355 /*
1356  * Packet forwarding routine once entry in the cache is made
1357  */
1358 static int
1359 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1360 {
1361 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1362 	mifi_t mifi, iif;
1363 	struct mif6 *mifp;
1364 	int plen = m->m_pkthdr.len;
1365 	struct in6_addr src0, dst0; /* copies for local work */
1366 	u_int32_t iszone, idzone, oszone, odzone;
1367 	int error = 0;
1368 
1369 	/*
1370 	 * Don't forward if it didn't arrive from the parent mif
1371 	 * for its origin.
1372 	 */
1373 	mifi = rt->mf6c_parent;
1374 	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1375 		/* came in the wrong interface */
1376 		MRT6_DLOG(DEBUG_FORWARD,
1377 		    "wrong if: ifid %d mifi %d mififid %x", ifp->if_index,
1378 		    mifi, mif6table[mifi].m6_ifp->if_index);
1379 		MRT6STAT_INC(mrt6s_wrong_if);
1380 		rt->mf6c_wrong_if++;
1381 		/*
1382 		 * If we are doing PIM processing, and we are forwarding
1383 		 * packets on this interface, send a message to the
1384 		 * routing daemon.
1385 		 */
1386 		/* have to make sure this is a valid mif */
1387 		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1388 			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1389 				/*
1390 				 * Check the M_LOOP flag to avoid an
1391 				 * unnecessary PIM assert.
1392 				 * XXX: M_LOOP is an ad-hoc hack...
1393 				 */
1394 				static struct sockaddr_in6 sin6 =
1395 				{ sizeof(sin6), AF_INET6 };
1396 
1397 				struct mbuf *mm;
1398 				struct mrt6msg *im;
1399 #ifdef MRT6_OINIT
1400 				struct omrt6msg *oim;
1401 #endif
1402 
1403 				mm = m_copym(m, 0, sizeof(struct ip6_hdr),
1404 				    M_NOWAIT);
1405 				if (mm &&
1406 				    (!M_WRITABLE(mm) ||
1407 				     mm->m_len < sizeof(struct ip6_hdr)))
1408 					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1409 				if (mm == NULL)
1410 					return (ENOBUFS);
1411 
1412 #ifdef MRT6_OINIT
1413 				oim = NULL;
1414 #endif
1415 				im = NULL;
1416 				switch (V_ip6_mrouter_ver) {
1417 #ifdef MRT6_OINIT
1418 				case MRT6_OINIT:
1419 					oim = mtod(mm, struct omrt6msg *);
1420 					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1421 					oim->im6_mbz = 0;
1422 					break;
1423 #endif
1424 				case MRT6_INIT:
1425 					im = mtod(mm, struct mrt6msg *);
1426 					im->im6_msgtype = MRT6MSG_WRONGMIF;
1427 					im->im6_mbz = 0;
1428 					break;
1429 				default:
1430 					m_freem(mm);
1431 					return (EINVAL);
1432 				}
1433 
1434 				for (mifp = mif6table, iif = 0;
1435 				     iif < nummifs && mifp &&
1436 					     mifp->m6_ifp != ifp;
1437 				     mifp++, iif++)
1438 					;
1439 
1440 				switch (V_ip6_mrouter_ver) {
1441 #ifdef MRT6_OINIT
1442 				case MRT6_OINIT:
1443 					oim->im6_mif = iif;
1444 					sin6.sin6_addr = oim->im6_src;
1445 					break;
1446 #endif
1447 				case MRT6_INIT:
1448 					im->im6_mif = iif;
1449 					sin6.sin6_addr = im->im6_src;
1450 					break;
1451 				}
1452 
1453 				MRT6STAT_INC(mrt6s_upcalls);
1454 
1455 				if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1456 					MRT6_DLOG(DEBUG_ANY,
1457 					    "ip6_mrouter socket queue full");
1458 					MRT6STAT_INC(mrt6s_upq_sockfull);
1459 					return (ENOBUFS);
1460 				}	/* if socket Q full */
1461 			}		/* if PIM */
1462 		return (0);
1463 	}			/* if wrong iif */
1464 
1465 	/* If I sourced this packet, it counts as output, else it was input. */
1466 	if (m->m_pkthdr.rcvif == NULL) {
1467 		/* XXX: is rcvif really NULL when output?? */
1468 		mif6table[mifi].m6_pkt_out++;
1469 		mif6table[mifi].m6_bytes_out += plen;
1470 	} else {
1471 		mif6table[mifi].m6_pkt_in++;
1472 		mif6table[mifi].m6_bytes_in += plen;
1473 	}
1474 	rt->mf6c_pkt_cnt++;
1475 	rt->mf6c_byte_cnt += plen;
1476 
1477 	/*
1478 	 * For each mif, forward a copy of the packet if there are group
1479 	 * members downstream on the interface.
1480 	 */
1481 	src0 = ip6->ip6_src;
1482 	dst0 = ip6->ip6_dst;
1483 	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1484 	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1485 		IP6STAT_INC(ip6s_badscope);
1486 		return (error);
1487 	}
1488 	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1489 		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1490 			/*
1491 			 * check if the outgoing packet is going to break
1492 			 * a scope boundary.
1493 			 * XXX For packets through PIM register tunnel
1494 			 * interface, we believe a routing daemon.
1495 			 */
1496 			if (!(mif6table[rt->mf6c_parent].m6_flags &
1497 			      MIFF_REGISTER) &&
1498 			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1499 				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1500 				    &oszone) ||
1501 				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1502 				    &odzone) ||
1503 				    iszone != oszone ||
1504 				    idzone != odzone) {
1505 					IP6STAT_INC(ip6s_badscope);
1506 					continue;
1507 				}
1508 			}
1509 
1510 			mifp->m6_pkt_out++;
1511 			mifp->m6_bytes_out += plen;
1512 			if (mifp->m6_flags & MIFF_REGISTER)
1513 				register_send(ip6, mifp, m);
1514 			else
1515 				phyint_send(ip6, mifp, m);
1516 		}
1517 	}
1518 	return (0);
1519 }
1520 
1521 static void
1522 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1523 {
1524 #ifdef MRT6DEBUG
1525 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1526 #endif
1527 	struct mbuf *mb_copy;
1528 	struct ifnet *ifp = mifp->m6_ifp;
1529 	int error __unused = 0;
1530 	u_long linkmtu;
1531 
1532 	/*
1533 	 * Make a new reference to the packet; make sure that
1534 	 * the IPv6 header is actually copied, not just referenced,
1535 	 * so that ip6_output() only scribbles on the copy.
1536 	 */
1537 	mb_copy = m_copym(m, 0, M_COPYALL, M_NOWAIT);
1538 	if (mb_copy &&
1539 	    (!M_WRITABLE(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1540 		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1541 	if (mb_copy == NULL) {
1542 		return;
1543 	}
1544 	/* set MCAST flag to the outgoing packet */
1545 	mb_copy->m_flags |= M_MCAST;
1546 
1547 	/*
1548 	 * If we sourced the packet, call ip6_output since we may devide
1549 	 * the packet into fragments when the packet is too big for the
1550 	 * outgoing interface.
1551 	 * Otherwise, we can simply send the packet to the interface
1552 	 * sending queue.
1553 	 */
1554 	if (m->m_pkthdr.rcvif == NULL) {
1555 		struct ip6_moptions im6o;
1556 		struct epoch_tracker et;
1557 
1558 		im6o.im6o_multicast_ifp = ifp;
1559 		/* XXX: ip6_output will override ip6->ip6_hlim */
1560 		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1561 		im6o.im6o_multicast_loop = 1;
1562 		NET_EPOCH_ENTER(et);
1563 		error = ip6_output(mb_copy, NULL, NULL, IPV6_FORWARDING, &im6o,
1564 		    NULL, NULL);
1565 		NET_EPOCH_EXIT(et);
1566 
1567 		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1568 		    (uint16_t)(mifp - mif6table), error);
1569 		return;
1570 	}
1571 
1572 	/*
1573 	 * If configured to loop back multicasts by default,
1574 	 * loop back a copy now.
1575 	 */
1576 	if (in6_mcast_loop)
1577 		ip6_mloopback(ifp, m);
1578 
1579 	/*
1580 	 * Put the packet into the sending queue of the outgoing interface
1581 	 * if it would fit in the MTU of the interface.
1582 	 */
1583 	linkmtu = IN6_LINKMTU(ifp);
1584 	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1585 		struct sockaddr_in6 dst6;
1586 
1587 		bzero(&dst6, sizeof(dst6));
1588 		dst6.sin6_len = sizeof(struct sockaddr_in6);
1589 		dst6.sin6_family = AF_INET6;
1590 		dst6.sin6_addr = ip6->ip6_dst;
1591 
1592 		IP_PROBE(send, NULL, NULL, ip6, ifp, NULL, ip6);
1593 		/*
1594 		 * We just call if_output instead of nd6_output here, since
1595 		 * we need no ND for a multicast forwarded packet...right?
1596 		 */
1597 		m_clrprotoflags(m);	/* Avoid confusing lower layers. */
1598 		error = (*ifp->if_output)(ifp, mb_copy,
1599 		    (struct sockaddr *)&dst6, NULL);
1600 		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1601 		    (uint16_t)(mifp - mif6table), error);
1602 	} else {
1603 		/*
1604 		 * pMTU discovery is intentionally disabled by default, since
1605 		 * various router may notify pMTU in multicast, which can be
1606 		 * a DDoS to a router
1607 		 */
1608 		if (V_ip6_mcast_pmtu)
1609 			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1610 		else {
1611 			MRT6_DLOG(DEBUG_XMIT, " packet too big on %s o %s "
1612 			    "g %s size %d (discarded)", if_name(ifp),
1613 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1614 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1615 			    mb_copy->m_pkthdr.len);
1616 			m_freem(mb_copy); /* simply discard the packet */
1617 		}
1618 	}
1619 }
1620 
1621 static int
1622 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1623 {
1624 #ifdef MRT6DEBUG
1625 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1626 #endif
1627 	struct mbuf *mm;
1628 	int i, len = m->m_pkthdr.len;
1629 	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1630 	struct mrt6msg *im6;
1631 
1632 	MRT6_DLOG(DEBUG_ANY, "src %s dst %s",
1633 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1634 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1635 	PIM6STAT_INC(pim6s_snd_registers);
1636 
1637 	/* Make a copy of the packet to send to the user level process. */
1638 	mm = m_gethdr(M_NOWAIT, MT_DATA);
1639 	if (mm == NULL)
1640 		return (ENOBUFS);
1641 	mm->m_data += max_linkhdr;
1642 	mm->m_len = sizeof(struct ip6_hdr);
1643 
1644 	if ((mm->m_next = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL) {
1645 		m_freem(mm);
1646 		return (ENOBUFS);
1647 	}
1648 	i = MHLEN - M_LEADINGSPACE(mm);
1649 	if (i > len)
1650 		i = len;
1651 	mm = m_pullup(mm, i);
1652 	if (mm == NULL)
1653 		return (ENOBUFS);
1654 /* TODO: check it! */
1655 	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1656 
1657 	/*
1658 	 * Send message to routing daemon
1659 	 */
1660 	sin6.sin6_addr = ip6->ip6_src;
1661 
1662 	im6 = mtod(mm, struct mrt6msg *);
1663 	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1664 	im6->im6_mbz          = 0;
1665 
1666 	im6->im6_mif = mif - mif6table;
1667 
1668 	/* iif info is not given for reg. encap.n */
1669 	MRT6STAT_INC(mrt6s_upcalls);
1670 
1671 	if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1672 		MRT6_DLOG(DEBUG_ANY, "ip6_mrouter socket queue full");
1673 		MRT6STAT_INC(mrt6s_upq_sockfull);
1674 		return (ENOBUFS);
1675 	}
1676 	return (0);
1677 }
1678 
1679 /*
1680  * pim6_encapcheck() is called by the encap6_input() path at runtime to
1681  * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1682  * into the kernel.
1683  */
1684 static int
1685 pim6_encapcheck(const struct mbuf *m __unused, int off __unused,
1686     int proto __unused, void *arg __unused)
1687 {
1688 
1689     KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1690     return (8);		/* claim the datagram. */
1691 }
1692 
1693 /*
1694  * PIM sparse mode hook
1695  * Receives the pim control messages, and passes them up to the listening
1696  * socket, using rip6_input.
1697  * The only message processed is the REGISTER pim message; the pim header
1698  * is stripped off, and the inner packet is passed to register_mforward.
1699  */
1700 static int
1701 pim6_input(struct mbuf *m, int off, int proto, void *arg __unused)
1702 {
1703 	struct pim *pim; /* pointer to a pim struct */
1704 	struct ip6_hdr *ip6;
1705 	int pimlen;
1706 	int minlen;
1707 
1708 	PIM6STAT_INC(pim6s_rcv_total);
1709 
1710 	/*
1711 	 * Validate lengths
1712 	 */
1713 	pimlen = m->m_pkthdr.len - off;
1714 	if (pimlen < PIM_MINLEN) {
1715 		PIM6STAT_INC(pim6s_rcv_tooshort);
1716 		MRT6_DLOG(DEBUG_PIM, "PIM packet too short");
1717 		m_freem(m);
1718 		return (IPPROTO_DONE);
1719 	}
1720 
1721 	/*
1722 	 * if the packet is at least as big as a REGISTER, go ahead
1723 	 * and grab the PIM REGISTER header size, to avoid another
1724 	 * possible m_pullup() later.
1725 	 *
1726 	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1727 	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1728 	 */
1729 	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1730 
1731 	/*
1732 	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1733 	 * possibly the PIM REGISTER header
1734 	 */
1735 	if (m->m_len < off + minlen) {
1736 		m = m_pullup(m, off + minlen);
1737 		if (m == NULL) {
1738 			IP6STAT_INC(ip6s_exthdrtoolong);
1739 			return (IPPROTO_DONE);
1740 		}
1741 	}
1742 	ip6 = mtod(m, struct ip6_hdr *);
1743 	pim = (struct pim *)((caddr_t)ip6 + off);
1744 
1745 #define PIM6_CHECKSUM
1746 #ifdef PIM6_CHECKSUM
1747 	{
1748 		int cksumlen;
1749 
1750 		/*
1751 		 * Validate checksum.
1752 		 * If PIM REGISTER, exclude the data packet
1753 		 */
1754 		if (pim->pim_type == PIM_REGISTER)
1755 			cksumlen = PIM_MINLEN;
1756 		else
1757 			cksumlen = pimlen;
1758 
1759 		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1760 			PIM6STAT_INC(pim6s_rcv_badsum);
1761 			MRT6_DLOG(DEBUG_PIM, "invalid checksum");
1762 			m_freem(m);
1763 			return (IPPROTO_DONE);
1764 		}
1765 	}
1766 #endif /* PIM_CHECKSUM */
1767 
1768 	/* PIM version check */
1769 	if (pim->pim_ver != PIM_VERSION) {
1770 		PIM6STAT_INC(pim6s_rcv_badversion);
1771 		MRT6_DLOG(DEBUG_ANY | DEBUG_ERR,
1772 		    "incorrect version %d, expecting %d",
1773 		    pim->pim_ver, PIM_VERSION);
1774 		m_freem(m);
1775 		return (IPPROTO_DONE);
1776 	}
1777 
1778 	if (pim->pim_type == PIM_REGISTER) {
1779 		/*
1780 		 * since this is a REGISTER, we'll make a copy of the register
1781 		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1782 		 * routing daemon.
1783 		 */
1784 		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1785 
1786 		struct mbuf *mcp;
1787 		struct ip6_hdr *eip6;
1788 		u_int32_t *reghdr;
1789 #ifdef MRT6DEBUG
1790 		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1791 #endif
1792 
1793 		PIM6STAT_INC(pim6s_rcv_registers);
1794 
1795 		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1796 			MRT6_DLOG(DEBUG_PIM, "register mif not set: %d",
1797 			    reg_mif_num);
1798 			m_freem(m);
1799 			return (IPPROTO_DONE);
1800 		}
1801 
1802 		reghdr = (u_int32_t *)(pim + 1);
1803 
1804 		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1805 			goto pim6_input_to_daemon;
1806 
1807 		/*
1808 		 * Validate length
1809 		 */
1810 		if (pimlen < PIM6_REG_MINLEN) {
1811 			PIM6STAT_INC(pim6s_rcv_tooshort);
1812 			PIM6STAT_INC(pim6s_rcv_badregisters);
1813 			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "register packet "
1814 			    "size too small %d from %s",
1815 			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1816 			m_freem(m);
1817 			return (IPPROTO_DONE);
1818 		}
1819 
1820 		eip6 = (struct ip6_hdr *) (reghdr + 1);
1821 		MRT6_DLOG(DEBUG_PIM, "eip6: %s -> %s, eip6 plen %d",
1822 		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1823 		    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1824 		    ntohs(eip6->ip6_plen));
1825 
1826 		/* verify the version number of the inner packet */
1827 		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1828 			PIM6STAT_INC(pim6s_rcv_badregisters);
1829 			MRT6_DLOG(DEBUG_ANY, "invalid IP version (%d) "
1830 			    "of the inner packet",
1831 			    (eip6->ip6_vfc & IPV6_VERSION));
1832 			m_freem(m);
1833 			return (IPPROTO_DONE);
1834 		}
1835 
1836 		/* verify the inner packet is destined to a mcast group */
1837 		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1838 			PIM6STAT_INC(pim6s_rcv_badregisters);
1839 			MRT6_DLOG(DEBUG_PIM, "inner packet of register "
1840 			    "is not multicast %s",
1841 			    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1842 			m_freem(m);
1843 			return (IPPROTO_DONE);
1844 		}
1845 
1846 		/*
1847 		 * make a copy of the whole header to pass to the daemon later.
1848 		 */
1849 		mcp = m_copym(m, 0, off + PIM6_REG_MINLEN, M_NOWAIT);
1850 		if (mcp == NULL) {
1851 			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "pim register: "
1852 			    "could not copy register head");
1853 			m_freem(m);
1854 			return (IPPROTO_DONE);
1855 		}
1856 
1857 		/*
1858 		 * forward the inner ip6 packet; point m_data at the inner ip6.
1859 		 */
1860 		m_adj(m, off + PIM_MINLEN);
1861 		MRT6_DLOG(DEBUG_PIM, "forwarding decapsulated register: "
1862 		    "src %s, dst %s, mif %d",
1863 		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1864 		    ip6_sprintf(ip6bufd, &eip6->ip6_dst), reg_mif_num);
1865 
1866 		if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1867 				dst.sin6_family, 0);
1868 
1869 		/* prepare the register head to send to the mrouting daemon */
1870 		m = mcp;
1871 	}
1872 
1873 	/*
1874 	 * Pass the PIM message up to the daemon; if it is a register message
1875 	 * pass the 'head' only up to the daemon. This includes the
1876 	 * encapsulator ip6 header, pim header, register header and the
1877 	 * encapsulated ip6 header.
1878 	 */
1879   pim6_input_to_daemon:
1880 	return (rip6_input(&m, &off, proto));
1881 }
1882 
1883 static int
1884 ip6_mroute_modevent(module_t mod, int type, void *unused)
1885 {
1886 
1887 	switch (type) {
1888 	case MOD_LOAD:
1889 		MROUTER6_LOCK_INIT();
1890 		MFC6_LOCK_INIT();
1891 		MIF6_LOCK_INIT();
1892 
1893 		pim6_encap_cookie = ip6_encap_attach(&ipv6_encap_cfg,
1894 		    NULL, M_WAITOK);
1895 		if (pim6_encap_cookie == NULL) {
1896 			printf("ip6_mroute: unable to attach pim6 encap\n");
1897 			MIF6_LOCK_DESTROY();
1898 			MFC6_LOCK_DESTROY();
1899 			MROUTER6_LOCK_DESTROY();
1900 			return (EINVAL);
1901 		}
1902 
1903 		ip6_mforward = X_ip6_mforward;
1904 		ip6_mrouter_done = X_ip6_mrouter_done;
1905 		ip6_mrouter_get = X_ip6_mrouter_get;
1906 		ip6_mrouter_set = X_ip6_mrouter_set;
1907 		mrt6_ioctl = X_mrt6_ioctl;
1908 		break;
1909 
1910 	case MOD_UNLOAD:
1911 		if (V_ip6_mrouter != NULL)
1912 			return EINVAL;
1913 
1914 		if (pim6_encap_cookie) {
1915 			ip6_encap_detach(pim6_encap_cookie);
1916 			pim6_encap_cookie = NULL;
1917 		}
1918 		X_ip6_mrouter_done();
1919 		ip6_mforward = NULL;
1920 		ip6_mrouter_done = NULL;
1921 		ip6_mrouter_get = NULL;
1922 		ip6_mrouter_set = NULL;
1923 		mrt6_ioctl = NULL;
1924 
1925 		MIF6_LOCK_DESTROY();
1926 		MFC6_LOCK_DESTROY();
1927 		MROUTER6_LOCK_DESTROY();
1928 		break;
1929 
1930 	default:
1931 		return (EOPNOTSUPP);
1932 	}
1933 
1934 	return (0);
1935 }
1936 
1937 static moduledata_t ip6_mroutemod = {
1938 	"ip6_mroute",
1939 	ip6_mroute_modevent,
1940 	0
1941 };
1942 
1943 DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PROTO_MC, SI_ORDER_ANY);
1944