xref: /freebsd/sys/netinet6/ip6_mroute.c (revision 31ba4ce8898f9dfa5e7f054fdbc26e50a599a6e3)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (C) 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  *
31  *	$KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $
32  */
33 
34 /*-
35  * Copyright (c) 1989 Stephen Deering
36  * Copyright (c) 1992, 1993
37  *      The Regents of the University of California.  All rights reserved.
38  *
39  * This code is derived from software contributed to Berkeley by
40  * Stephen Deering of Stanford University.
41  *
42  * Redistribution and use in source and binary forms, with or without
43  * modification, are permitted provided that the following conditions
44  * are met:
45  * 1. Redistributions of source code must retain the above copyright
46  *    notice, this list of conditions and the following disclaimer.
47  * 2. Redistributions in binary form must reproduce the above copyright
48  *    notice, this list of conditions and the following disclaimer in the
49  *    documentation and/or other materials provided with the distribution.
50  * 3. Neither the name of the University nor the names of its contributors
51  *    may be used to endorse or promote products derived from this software
52  *    without specific prior written permission.
53  *
54  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
64  * SUCH DAMAGE.
65  *
66  *	@(#)ip_mroute.c	8.2 (Berkeley) 11/15/93
67  *	BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp
68  */
69 
70 /*
71  * IP multicast forwarding procedures
72  *
73  * Written by David Waitzman, BBN Labs, August 1988.
74  * Modified by Steve Deering, Stanford, February 1989.
75  * Modified by Mark J. Steiglitz, Stanford, May, 1991
76  * Modified by Van Jacobson, LBL, January 1993
77  * Modified by Ajit Thyagarajan, PARC, August 1993
78  * Modified by Bill Fenner, PARC, April 1994
79  *
80  * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
81  */
82 
83 #include <sys/cdefs.h>
84 __FBSDID("$FreeBSD$");
85 
86 #include "opt_inet6.h"
87 
88 #include <sys/param.h>
89 #include <sys/callout.h>
90 #include <sys/errno.h>
91 #include <sys/kernel.h>
92 #include <sys/lock.h>
93 #include <sys/malloc.h>
94 #include <sys/mbuf.h>
95 #include <sys/module.h>
96 #include <sys/domain.h>
97 #include <sys/protosw.h>
98 #include <sys/sdt.h>
99 #include <sys/signalvar.h>
100 #include <sys/socket.h>
101 #include <sys/socketvar.h>
102 #include <sys/sockio.h>
103 #include <sys/sx.h>
104 #include <sys/sysctl.h>
105 #include <sys/syslog.h>
106 #include <sys/systm.h>
107 #include <sys/time.h>
108 
109 #include <net/if.h>
110 #include <net/if_var.h>
111 #include <net/if_types.h>
112 #include <net/vnet.h>
113 
114 #include <netinet/in.h>
115 #include <netinet/in_var.h>
116 #include <netinet/icmp6.h>
117 #include <netinet/ip_encap.h>
118 
119 #include <netinet/ip6.h>
120 #include <netinet/in_kdtrace.h>
121 #include <netinet6/ip6_var.h>
122 #include <netinet6/scope6_var.h>
123 #include <netinet6/nd6.h>
124 #include <netinet6/ip6_mroute.h>
125 #include <netinet6/pim6.h>
126 #include <netinet6/pim6_var.h>
127 
128 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
129 
130 static int	ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
131 static void	phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
132 static int	register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
133 static int	set_pim6(int *);
134 static int	socket_send(struct socket *, struct mbuf *,
135 		    struct sockaddr_in6 *);
136 
137 extern int in6_mcast_loop;
138 extern struct domain inet6domain;
139 
140 static const struct encaptab *pim6_encap_cookie;
141 static int pim6_encapcheck(const struct mbuf *, int, int, void *);
142 static int pim6_input(struct mbuf *, int, int, void *);
143 
144 static const struct encap_config ipv6_encap_cfg = {
145 	.proto = IPPROTO_PIM,
146 	.min_length = sizeof(struct ip6_hdr) + PIM_MINLEN,
147 	.exact_match = 8,
148 	.check = pim6_encapcheck,
149 	.input = pim6_input
150 };
151 
152 VNET_DEFINE_STATIC(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,
158     CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
159     "PIM");
160 
161 static struct mrt6stat mrt6stat;
162 SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
163     &mrt6stat, mrt6stat,
164     "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
165 
166 #define	MRT6STAT_INC(name)	mrt6stat.name += 1
167 #define NO_RTE_FOUND	0x1
168 #define RTE_FOUND	0x2
169 
170 static struct mtx mrouter6_mtx;
171 #define	MROUTER6_LOCK()		mtx_lock(&mrouter6_mtx)
172 #define	MROUTER6_UNLOCK()	mtx_unlock(&mrouter6_mtx)
173 #define	MROUTER6_LOCK_ASSERT()	do {					\
174 	mtx_assert(&mrouter6_mtx, MA_OWNED);				\
175 	NET_ASSERT_GIANT();						\
176 } while (0)
177 #define	MROUTER6_LOCK_INIT()	\
178 	mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
179 #define	MROUTER6_LOCK_DESTROY()	mtx_destroy(&mrouter6_mtx)
180 
181 static struct mf6c *mf6ctable[MF6CTBLSIZ];
182 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
183     &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
184     "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
185     "netinet6/ip6_mroute.h)");
186 
187 static struct mtx mfc6_mtx;
188 #define	MFC6_LOCK()		mtx_lock(&mfc6_mtx)
189 #define	MFC6_UNLOCK()		mtx_unlock(&mfc6_mtx)
190 #define	MFC6_LOCK_ASSERT()	do {					\
191 	mtx_assert(&mfc6_mtx, MA_OWNED);				\
192 	NET_ASSERT_GIANT();						\
193 } while (0)
194 #define	MFC6_LOCK_INIT()		\
195 	mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
196 #define	MFC6_LOCK_DESTROY()	mtx_destroy(&mfc6_mtx)
197 
198 static u_char n6expire[MF6CTBLSIZ];
199 
200 static struct mif6 mif6table[MAXMIFS];
201 static int
202 sysctl_mif6table(SYSCTL_HANDLER_ARGS)
203 {
204 	struct mif6_sctl *out;
205 	int error;
206 
207 	out = malloc(sizeof(struct mif6_sctl) * MAXMIFS, M_TEMP,
208 	    M_WAITOK | M_ZERO);
209 	for (int i = 0; i < MAXMIFS; i++) {
210 		out[i].m6_flags		= mif6table[i].m6_flags;
211 		out[i].m6_rate_limit	= mif6table[i].m6_rate_limit;
212 		out[i].m6_lcl_addr	= mif6table[i].m6_lcl_addr;
213 		if (mif6table[i].m6_ifp != NULL)
214 			out[i].m6_ifp	= mif6table[i].m6_ifp->if_index;
215 		else
216 			out[i].m6_ifp	= 0;
217 		out[i].m6_pkt_in	= mif6table[i].m6_pkt_in;
218 		out[i].m6_pkt_out	= mif6table[i].m6_pkt_out;
219 		out[i].m6_bytes_in	= mif6table[i].m6_bytes_in;
220 		out[i].m6_bytes_out	= mif6table[i].m6_bytes_out;
221 	}
222 	error = SYSCTL_OUT(req, out, sizeof(struct mif6_sctl) * MAXMIFS);
223 	free(out, M_TEMP);
224 	return (error);
225 }
226 SYSCTL_PROC(_net_inet6_ip6, OID_AUTO, mif6table,
227     CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_NEEDGIANT,
228     NULL, 0, sysctl_mif6table, "S,mif6_sctl[MAXMIFS]",
229     "IPv6 Multicast Interfaces (struct mif6_sctl[MAXMIFS], "
230     "netinet6/ip6_mroute.h)");
231 
232 static struct mtx mif6_mtx;
233 #define	MIF6_LOCK()		mtx_lock(&mif6_mtx)
234 #define	MIF6_UNLOCK()		mtx_unlock(&mif6_mtx)
235 #define	MIF6_LOCK_ASSERT()	mtx_assert(&mif6_mtx, MA_OWNED)
236 #define	MIF6_LOCK_INIT()	\
237 	mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
238 #define	MIF6_LOCK_DESTROY()	mtx_destroy(&mif6_mtx)
239 
240 #ifdef MRT6DEBUG
241 VNET_DEFINE_STATIC(u_int, mrt6debug) = 0;	/* debug level */
242 #define	V_mrt6debug		VNET(mrt6debug)
243 #define DEBUG_MFC	0x02
244 #define DEBUG_FORWARD	0x04
245 #define DEBUG_EXPIRE	0x08
246 #define DEBUG_XMIT	0x10
247 #define DEBUG_REG	0x20
248 #define DEBUG_PIM	0x40
249 #define	DEBUG_ERR	0x80
250 #define	DEBUG_ANY	0x7f
251 #define	MRT6_DLOG(m, fmt, ...)	\
252 	if (V_mrt6debug & (m))	\
253 		log(((m) & DEBUG_ERR) ? LOG_ERR: LOG_DEBUG, \
254 		    "%s: " fmt "\n", __func__, ##__VA_ARGS__)
255 #else
256 #define	MRT6_DLOG(m, fmt, ...)
257 #endif
258 
259 static void	expire_upcalls(void *);
260 #define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
261 #define	UPCALL_EXPIRE	6		/* number of timeouts */
262 
263 /*
264  * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
265  */
266 
267 /*
268  * 'Interfaces' associated with decapsulator (so we can tell
269  * packets that went through it from ones that get reflected
270  * by a broken gateway).  Different from IPv4 register_if,
271  * these interfaces are linked into the system ifnet list,
272  * because per-interface IPv6 statistics are maintained in
273  * ifp->if_afdata.  But it does not have any routes point
274  * to them.  I.e., packets can't be sent this way.  They
275  * only exist as a placeholder for multicast source
276  * verification.
277  */
278 static struct ifnet *multicast_register_if6;
279 
280 #define ENCAP_HOPS 64
281 
282 /*
283  * Private variables.
284  */
285 static mifi_t nummifs = 0;
286 static mifi_t reg_mif_num = (mifi_t)-1;
287 
288 static struct pim6stat pim6stat;
289 SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RW,
290     &pim6stat, pim6stat,
291     "PIM Statistics (struct pim6stat, netinet6/pim6_var.h)");
292 
293 #define	PIM6STAT_INC(name)	pim6stat.name += 1
294 VNET_DEFINE_STATIC(int, pim6);
295 #define	V_pim6		VNET(pim6)
296 
297 /*
298  * Hash function for a source, group entry
299  */
300 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
301 				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
302 				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
303 				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
304 
305 /*
306  * Find a route for a given origin IPv6 address and Multicast group address.
307  */
308 #define MF6CFIND(o, g, rt) do { \
309 	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
310 	rt = NULL; \
311 	while (_rt) { \
312 		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
313 		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
314 		    (_rt->mf6c_stall == NULL)) { \
315 			rt = _rt; \
316 			break; \
317 		} \
318 		_rt = _rt->mf6c_next; \
319 	} \
320 	if (rt == NULL) { \
321 		MRT6STAT_INC(mrt6s_mfc_misses); \
322 	} \
323 } while (/*CONSTCOND*/ 0)
324 
325 /*
326  * Macros to compute elapsed time efficiently
327  * Borrowed from Van Jacobson's scheduling code
328  * XXX: replace with timersub() ?
329  */
330 #define TV_DELTA(a, b, delta) do { \
331 	    int xxs; \
332 		\
333 	    delta = (a).tv_usec - (b).tv_usec; \
334 	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
335 	       switch (xxs) { \
336 		      case 2: \
337 			  delta += 1000000; \
338 			      /* FALLTHROUGH */ \
339 		      case 1: \
340 			  delta += 1000000; \
341 			  break; \
342 		      default: \
343 			  delta += (1000000 * xxs); \
344 	       } \
345 	    } \
346 } while (/*CONSTCOND*/ 0)
347 
348 /* XXX: replace with timercmp(a, b, <) ? */
349 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
350 	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
351 
352 #ifdef UPCALL_TIMING
353 #define UPCALL_MAX	50
354 static u_long upcall_data[UPCALL_MAX + 1];
355 static void collate();
356 #endif /* UPCALL_TIMING */
357 
358 static int ip6_mrouter_init(struct socket *, int, int);
359 static int add_m6fc(struct mf6cctl *);
360 static int add_m6if(struct mif6ctl *);
361 static int del_m6fc(struct mf6cctl *);
362 static int del_m6if(mifi_t *);
363 static int del_m6if_locked(mifi_t *);
364 static int get_mif6_cnt(struct sioc_mif_req6 *);
365 static int get_sg_cnt(struct sioc_sg_req6 *);
366 
367 static struct callout expire_upcalls_ch;
368 
369 int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
370 int X_ip6_mrouter_done(void);
371 int X_ip6_mrouter_set(struct socket *, struct sockopt *);
372 int X_ip6_mrouter_get(struct socket *, struct sockopt *);
373 int X_mrt6_ioctl(u_long, caddr_t);
374 
375 /*
376  * Handle MRT setsockopt commands to modify the multicast routing tables.
377  */
378 int
379 X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
380 {
381 	int error = 0;
382 	int optval;
383 	struct mif6ctl mifc;
384 	struct mf6cctl mfcc;
385 	mifi_t mifi;
386 
387 	if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
388 		return (EPERM);
389 
390 	switch (sopt->sopt_name) {
391 	case MRT6_INIT:
392 #ifdef MRT6_OINIT
393 	case MRT6_OINIT:
394 #endif
395 		error = sooptcopyin(sopt, &optval, sizeof(optval),
396 		    sizeof(optval));
397 		if (error)
398 			break;
399 		error = ip6_mrouter_init(so, optval, sopt->sopt_name);
400 		break;
401 	case MRT6_DONE:
402 		error = X_ip6_mrouter_done();
403 		break;
404 	case MRT6_ADD_MIF:
405 		error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
406 		if (error)
407 			break;
408 		error = add_m6if(&mifc);
409 		break;
410 	case MRT6_ADD_MFC:
411 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
412 		if (error)
413 			break;
414 		error = add_m6fc(&mfcc);
415 		break;
416 	case MRT6_DEL_MFC:
417 		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
418 		if (error)
419 			break;
420 		error = del_m6fc(&mfcc);
421 		break;
422 	case MRT6_DEL_MIF:
423 		error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
424 		if (error)
425 			break;
426 		error = del_m6if(&mifi);
427 		break;
428 	case MRT6_PIM:
429 		error = sooptcopyin(sopt, &optval, sizeof(optval),
430 		    sizeof(optval));
431 		if (error)
432 			break;
433 		error = set_pim6(&optval);
434 		break;
435 	default:
436 		error = EOPNOTSUPP;
437 		break;
438 	}
439 
440 	return (error);
441 }
442 
443 /*
444  * Handle MRT getsockopt commands
445  */
446 int
447 X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
448 {
449 	int error = 0;
450 
451 	if (so != V_ip6_mrouter)
452 		return (EACCES);
453 
454 	switch (sopt->sopt_name) {
455 		case MRT6_PIM:
456 			error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
457 			break;
458 	}
459 	return (error);
460 }
461 
462 /*
463  * Handle ioctl commands to obtain information from the cache
464  */
465 int
466 X_mrt6_ioctl(u_long cmd, caddr_t data)
467 {
468 	int ret;
469 
470 	ret = EINVAL;
471 
472 	switch (cmd) {
473 	case SIOCGETSGCNT_IN6:
474 		ret = get_sg_cnt((struct sioc_sg_req6 *)data);
475 		break;
476 
477 	case SIOCGETMIFCNT_IN6:
478 		ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
479 		break;
480 
481 	default:
482 		break;
483 	}
484 
485 	return (ret);
486 }
487 
488 /*
489  * returns the packet, byte, rpf-failure count for the source group provided
490  */
491 static int
492 get_sg_cnt(struct sioc_sg_req6 *req)
493 {
494 	struct mf6c *rt;
495 	int ret;
496 
497 	ret = 0;
498 
499 	MFC6_LOCK();
500 
501 	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
502 	if (rt == NULL) {
503 		ret = ESRCH;
504 	} else {
505 		req->pktcnt = rt->mf6c_pkt_cnt;
506 		req->bytecnt = rt->mf6c_byte_cnt;
507 		req->wrong_if = rt->mf6c_wrong_if;
508 	}
509 
510 	MFC6_UNLOCK();
511 
512 	return (ret);
513 }
514 
515 /*
516  * returns the input and output packet and byte counts on the mif provided
517  */
518 static int
519 get_mif6_cnt(struct sioc_mif_req6 *req)
520 {
521 	mifi_t mifi;
522 	int ret;
523 
524 	ret = 0;
525 	mifi = req->mifi;
526 
527 	MIF6_LOCK();
528 
529 	if (mifi >= nummifs) {
530 		ret = EINVAL;
531 	} else {
532 		req->icount = mif6table[mifi].m6_pkt_in;
533 		req->ocount = mif6table[mifi].m6_pkt_out;
534 		req->ibytes = mif6table[mifi].m6_bytes_in;
535 		req->obytes = mif6table[mifi].m6_bytes_out;
536 	}
537 
538 	MIF6_UNLOCK();
539 
540 	return (ret);
541 }
542 
543 static int
544 set_pim6(int *i)
545 {
546 	if ((*i != 1) && (*i != 0))
547 		return (EINVAL);
548 
549 	V_pim6 = *i;
550 
551 	return (0);
552 }
553 
554 /*
555  * Enable multicast routing
556  */
557 static int
558 ip6_mrouter_init(struct socket *so, int v, int cmd)
559 {
560 
561 	MRT6_DLOG(DEBUG_ANY, "so_type = %d, pr_protocol = %d",
562 	    so->so_type, so->so_proto->pr_protocol);
563 
564 	if (so->so_type != SOCK_RAW ||
565 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
566 		return (EOPNOTSUPP);
567 
568 	if (v != 1)
569 		return (ENOPROTOOPT);
570 
571 	MROUTER6_LOCK();
572 
573 	if (V_ip6_mrouter != NULL) {
574 		MROUTER6_UNLOCK();
575 		return (EADDRINUSE);
576 	}
577 
578 	V_ip6_mrouter = so;
579 	V_ip6_mrouter_ver = cmd;
580 
581 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
582 	bzero((caddr_t)n6expire, sizeof(n6expire));
583 
584 	V_pim6 = 0;/* used for stubbing out/in pim stuff */
585 
586 	callout_init(&expire_upcalls_ch, 0);
587 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
588 	    expire_upcalls, NULL);
589 
590 	MROUTER6_UNLOCK();
591 	MRT6_DLOG(DEBUG_ANY, "finished");
592 
593 	return (0);
594 }
595 
596 /*
597  * Disable IPv6 multicast forwarding.
598  */
599 int
600 X_ip6_mrouter_done(void)
601 {
602 	mifi_t mifi;
603 	u_long i;
604 	struct mf6c *rt;
605 	struct rtdetq *rte;
606 
607 	MROUTER6_LOCK();
608 
609 	if (V_ip6_mrouter == NULL) {
610 		MROUTER6_UNLOCK();
611 		return (EINVAL);
612 	}
613 
614 	/*
615 	 * For each phyint in use, disable promiscuous reception of all IPv6
616 	 * multicasts.
617 	 */
618 	for (mifi = 0; mifi < nummifs; mifi++) {
619 		if (mif6table[mifi].m6_ifp &&
620 		    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
621 			if_allmulti(mif6table[mifi].m6_ifp, 0);
622 		}
623 	}
624 	bzero((caddr_t)mif6table, sizeof(mif6table));
625 	nummifs = 0;
626 
627 	V_pim6 = 0; /* used to stub out/in pim specific code */
628 
629 	callout_stop(&expire_upcalls_ch);
630 
631 	/*
632 	 * Free all multicast forwarding cache entries.
633 	 */
634 	MFC6_LOCK();
635 	for (i = 0; i < MF6CTBLSIZ; i++) {
636 		rt = mf6ctable[i];
637 		while (rt) {
638 			struct mf6c *frt;
639 
640 			for (rte = rt->mf6c_stall; rte != NULL; ) {
641 				struct rtdetq *n = rte->next;
642 
643 				m_freem(rte->m);
644 				free(rte, M_MRTABLE6);
645 				rte = n;
646 			}
647 			frt = rt;
648 			rt = rt->mf6c_next;
649 			free(frt, M_MRTABLE6);
650 		}
651 	}
652 	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
653 	MFC6_UNLOCK();
654 
655 	/*
656 	 * Reset register interface
657 	 */
658 	if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
659 		if_detach(multicast_register_if6);
660 		if_free(multicast_register_if6);
661 		reg_mif_num = (mifi_t)-1;
662 		multicast_register_if6 = NULL;
663 	}
664 
665 	V_ip6_mrouter = NULL;
666 	V_ip6_mrouter_ver = 0;
667 
668 	MROUTER6_UNLOCK();
669 	MRT6_DLOG(DEBUG_ANY, "finished");
670 
671 	return (0);
672 }
673 
674 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
675 
676 /*
677  * Add a mif to the mif table
678  */
679 static int
680 add_m6if(struct mif6ctl *mifcp)
681 {
682 	struct mif6 *mifp;
683 	struct ifnet *ifp;
684 	int error;
685 
686 	MIF6_LOCK();
687 
688 	if (mifcp->mif6c_mifi >= MAXMIFS) {
689 		MIF6_UNLOCK();
690 		return (EINVAL);
691 	}
692 	mifp = mif6table + mifcp->mif6c_mifi;
693 	if (mifp->m6_ifp != NULL) {
694 		MIF6_UNLOCK();
695 		return (EADDRINUSE); /* XXX: is it appropriate? */
696 	}
697 	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
698 		MIF6_UNLOCK();
699 		return (ENXIO);
700 	}
701 
702 	ifp = ifnet_byindex(mifcp->mif6c_pifi);
703 
704 	if (mifcp->mif6c_flags & MIFF_REGISTER) {
705 		if (reg_mif_num == (mifi_t)-1) {
706 			ifp = if_alloc(IFT_OTHER);
707 
708 			if_initname(ifp, "register_mif", 0);
709 			ifp->if_flags |= IFF_LOOPBACK;
710 			if_attach(ifp);
711 			multicast_register_if6 = ifp;
712 			reg_mif_num = mifcp->mif6c_mifi;
713 			/*
714 			 * it is impossible to guess the ifindex of the
715 			 * register interface.  So mif6c_pifi is automatically
716 			 * calculated.
717 			 */
718 			mifcp->mif6c_pifi = ifp->if_index;
719 		} else {
720 			ifp = multicast_register_if6;
721 		}
722 	} else {
723 		/* Make sure the interface supports multicast */
724 		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
725 			MIF6_UNLOCK();
726 			return (EOPNOTSUPP);
727 		}
728 
729 		error = if_allmulti(ifp, 1);
730 		if (error) {
731 			MIF6_UNLOCK();
732 			return (error);
733 		}
734 	}
735 
736 	mifp->m6_flags     = mifcp->mif6c_flags;
737 	mifp->m6_ifp       = ifp;
738 
739 	/* initialize per mif pkt counters */
740 	mifp->m6_pkt_in    = 0;
741 	mifp->m6_pkt_out   = 0;
742 	mifp->m6_bytes_in  = 0;
743 	mifp->m6_bytes_out = 0;
744 
745 	/* Adjust nummifs up if the mifi is higher than nummifs */
746 	if (nummifs <= mifcp->mif6c_mifi)
747 		nummifs = mifcp->mif6c_mifi + 1;
748 
749 	MIF6_UNLOCK();
750 	MRT6_DLOG(DEBUG_ANY, "mif #%d, phyint %s", mifcp->mif6c_mifi,
751 	    if_name(ifp));
752 
753 	return (0);
754 }
755 
756 /*
757  * Delete a mif from the mif table
758  */
759 static int
760 del_m6if_locked(mifi_t *mifip)
761 {
762 	struct mif6 *mifp = mif6table + *mifip;
763 	mifi_t mifi;
764 	struct ifnet *ifp;
765 
766 	MIF6_LOCK_ASSERT();
767 
768 	if (*mifip >= nummifs)
769 		return (EINVAL);
770 	if (mifp->m6_ifp == NULL)
771 		return (EINVAL);
772 
773 	if (!(mifp->m6_flags & MIFF_REGISTER)) {
774 		/* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
775 		ifp = mifp->m6_ifp;
776 		if_allmulti(ifp, 0);
777 	} else {
778 		if (reg_mif_num != (mifi_t)-1 &&
779 		    multicast_register_if6 != NULL) {
780 			if_detach(multicast_register_if6);
781 			if_free(multicast_register_if6);
782 			reg_mif_num = (mifi_t)-1;
783 			multicast_register_if6 = NULL;
784 		}
785 	}
786 
787 	bzero((caddr_t)mifp, sizeof(*mifp));
788 
789 	/* Adjust nummifs down */
790 	for (mifi = nummifs; mifi > 0; mifi--)
791 		if (mif6table[mifi - 1].m6_ifp)
792 			break;
793 	nummifs = mifi;
794 	MRT6_DLOG(DEBUG_ANY, "mif %d, nummifs %d", *mifip, nummifs);
795 
796 	return (0);
797 }
798 
799 static int
800 del_m6if(mifi_t *mifip)
801 {
802 	int cc;
803 
804 	MIF6_LOCK();
805 	cc = del_m6if_locked(mifip);
806 	MIF6_UNLOCK();
807 
808 	return (cc);
809 }
810 
811 /*
812  * Add an mfc entry
813  */
814 static int
815 add_m6fc(struct mf6cctl *mfccp)
816 {
817 	struct mf6c *rt;
818 	u_long hash;
819 	struct rtdetq *rte;
820 	u_short nstl;
821 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
822 
823 	MFC6_LOCK();
824 
825 	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
826 		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
827 
828 	/* If an entry already exists, just update the fields */
829 	if (rt) {
830 		MRT6_DLOG(DEBUG_MFC, "no upcall o %s g %s p %x",
831 		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
832 		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
833 		    mfccp->mf6cc_parent);
834 
835 		rt->mf6c_parent = mfccp->mf6cc_parent;
836 		rt->mf6c_ifset = mfccp->mf6cc_ifset;
837 
838 		MFC6_UNLOCK();
839 		return (0);
840 	}
841 
842 	/*
843 	 * Find the entry for which the upcall was made and update
844 	 */
845 	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
846 			mfccp->mf6cc_mcastgrp.sin6_addr);
847 	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
848 		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
849 				       &mfccp->mf6cc_origin.sin6_addr) &&
850 		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
851 				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
852 		    (rt->mf6c_stall != NULL)) {
853 			if (nstl++)
854 				log(LOG_ERR,
855 				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
856 				    "multiple kernel entries",
857 				    ip6_sprintf(ip6bufo,
858 					    &mfccp->mf6cc_origin.sin6_addr),
859 				    ip6_sprintf(ip6bufg,
860 					    &mfccp->mf6cc_mcastgrp.sin6_addr),
861 				    mfccp->mf6cc_parent, rt->mf6c_stall);
862 
863 			MRT6_DLOG(DEBUG_MFC, "o %s g %s p %x dbg %p",
864 			    ip6_sprintf(ip6bufo,
865 			    &mfccp->mf6cc_origin.sin6_addr),
866 			    ip6_sprintf(ip6bufg,
867 				&mfccp->mf6cc_mcastgrp.sin6_addr),
868 			    mfccp->mf6cc_parent, rt->mf6c_stall);
869 
870 			rt->mf6c_origin     = mfccp->mf6cc_origin;
871 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
872 			rt->mf6c_parent     = mfccp->mf6cc_parent;
873 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
874 			/* initialize pkt counters per src-grp */
875 			rt->mf6c_pkt_cnt    = 0;
876 			rt->mf6c_byte_cnt   = 0;
877 			rt->mf6c_wrong_if   = 0;
878 
879 			rt->mf6c_expire = 0;	/* Don't clean this guy up */
880 			n6expire[hash]--;
881 
882 			/* free packets Qed at the end of this entry */
883 			for (rte = rt->mf6c_stall; rte != NULL; ) {
884 				struct rtdetq *n = rte->next;
885 				ip6_mdq(rte->m, rte->ifp, rt);
886 				m_freem(rte->m);
887 #ifdef UPCALL_TIMING
888 				collate(&(rte->t));
889 #endif /* UPCALL_TIMING */
890 				free(rte, M_MRTABLE6);
891 				rte = n;
892 			}
893 			rt->mf6c_stall = NULL;
894 		}
895 	}
896 
897 	/*
898 	 * It is possible that an entry is being inserted without an upcall
899 	 */
900 	if (nstl == 0) {
901 		MRT6_DLOG(DEBUG_MFC, "no upcall h %lu o %s g %s p %x", hash,
902 		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
903 		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
904 		    mfccp->mf6cc_parent);
905 
906 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
907 			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
908 					       &mfccp->mf6cc_origin.sin6_addr)&&
909 			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
910 					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
911 				rt->mf6c_origin     = mfccp->mf6cc_origin;
912 				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
913 				rt->mf6c_parent     = mfccp->mf6cc_parent;
914 				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
915 				/* initialize pkt counters per src-grp */
916 				rt->mf6c_pkt_cnt    = 0;
917 				rt->mf6c_byte_cnt   = 0;
918 				rt->mf6c_wrong_if   = 0;
919 
920 				if (rt->mf6c_expire)
921 					n6expire[hash]--;
922 				rt->mf6c_expire	   = 0;
923 			}
924 		}
925 		if (rt == NULL) {
926 			/* no upcall, so make a new entry */
927 			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
928 						  M_NOWAIT);
929 			if (rt == NULL) {
930 				MFC6_UNLOCK();
931 				return (ENOBUFS);
932 			}
933 
934 			/* insert new entry at head of hash chain */
935 			rt->mf6c_origin     = mfccp->mf6cc_origin;
936 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
937 			rt->mf6c_parent     = mfccp->mf6cc_parent;
938 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
939 			/* initialize pkt counters per src-grp */
940 			rt->mf6c_pkt_cnt    = 0;
941 			rt->mf6c_byte_cnt   = 0;
942 			rt->mf6c_wrong_if   = 0;
943 			rt->mf6c_expire     = 0;
944 			rt->mf6c_stall = NULL;
945 
946 			/* link into table */
947 			rt->mf6c_next  = mf6ctable[hash];
948 			mf6ctable[hash] = rt;
949 		}
950 	}
951 
952 	MFC6_UNLOCK();
953 	return (0);
954 }
955 
956 #ifdef UPCALL_TIMING
957 /*
958  * collect delay statistics on the upcalls
959  */
960 static void
961 collate(struct timeval *t)
962 {
963 	u_long d;
964 	struct timeval tp;
965 	u_long delta;
966 
967 	GET_TIME(tp);
968 
969 	if (TV_LT(*t, tp))
970 	{
971 		TV_DELTA(tp, *t, delta);
972 
973 		d = delta >> 10;
974 		if (d > UPCALL_MAX)
975 			d = UPCALL_MAX;
976 
977 		++upcall_data[d];
978 	}
979 }
980 #endif /* UPCALL_TIMING */
981 
982 /*
983  * Delete an mfc entry
984  */
985 static int
986 del_m6fc(struct mf6cctl *mfccp)
987 {
988 #ifdef MRT6DEBUG
989 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
990 #endif
991 	struct sockaddr_in6	origin;
992 	struct sockaddr_in6	mcastgrp;
993 	struct mf6c		*rt;
994 	struct mf6c		**nptr;
995 	u_long		hash;
996 
997 	origin = mfccp->mf6cc_origin;
998 	mcastgrp = mfccp->mf6cc_mcastgrp;
999 	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
1000 
1001 	MRT6_DLOG(DEBUG_MFC, "orig %s mcastgrp %s",
1002 	    ip6_sprintf(ip6bufo, &origin.sin6_addr),
1003 	    ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
1004 
1005 	MFC6_LOCK();
1006 
1007 	nptr = &mf6ctable[hash];
1008 	while ((rt = *nptr) != NULL) {
1009 		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
1010 				       &rt->mf6c_origin.sin6_addr) &&
1011 		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1012 				       &rt->mf6c_mcastgrp.sin6_addr) &&
1013 		    rt->mf6c_stall == NULL)
1014 			break;
1015 
1016 		nptr = &rt->mf6c_next;
1017 	}
1018 	if (rt == NULL) {
1019 		MFC6_UNLOCK();
1020 		return (EADDRNOTAVAIL);
1021 	}
1022 
1023 	*nptr = rt->mf6c_next;
1024 	free(rt, M_MRTABLE6);
1025 
1026 	MFC6_UNLOCK();
1027 
1028 	return (0);
1029 }
1030 
1031 static int
1032 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1033 {
1034 
1035 	if (s) {
1036 		if (sbappendaddr(&s->so_rcv,
1037 				 (struct sockaddr *)src,
1038 				 mm, (struct mbuf *)0) != 0) {
1039 			sorwakeup(s);
1040 			return (0);
1041 		} else
1042 			soroverflow(s);
1043 	}
1044 	m_freem(mm);
1045 	return (-1);
1046 }
1047 
1048 /*
1049  * IPv6 multicast forwarding function. This function assumes that the packet
1050  * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1051  * pointed to by "ifp", and the packet is to be relayed to other networks
1052  * that have members of the packet's destination IPv6 multicast group.
1053  *
1054  * The packet is returned unscathed to the caller, unless it is
1055  * erroneous, in which case a non-zero return value tells the caller to
1056  * discard it.
1057  *
1058  * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1059  * this function is called in the originating context (i.e., not when
1060  * forwarding a packet from other node).  ip6_output(), which is currently the
1061  * only function that calls this function is called in the originating context,
1062  * explicitly ensures this condition.  It is caller's responsibility to ensure
1063  * that if this function is called from somewhere else in the originating
1064  * context in the future.
1065  */
1066 int
1067 X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1068 {
1069 	struct rtdetq *rte;
1070 	struct mbuf *mb0;
1071 	struct mf6c *rt;
1072 	struct mif6 *mifp;
1073 	struct mbuf *mm;
1074 	u_long hash;
1075 	mifi_t mifi;
1076 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1077 #ifdef UPCALL_TIMING
1078 	struct timeval tp;
1079 
1080 	GET_TIME(tp);
1081 #endif /* UPCALL_TIMING */
1082 
1083 	MRT6_DLOG(DEBUG_FORWARD, "src %s, dst %s, ifindex %d",
1084 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1085 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst), ifp->if_index);
1086 
1087 	/*
1088 	 * Don't forward a packet with Hop limit of zero or one,
1089 	 * or a packet destined to a local-only group.
1090 	 */
1091 	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1092 	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1093 		return (0);
1094 	ip6->ip6_hlim--;
1095 
1096 	/*
1097 	 * Source address check: do not forward packets with unspecified
1098 	 * source. It was discussed in July 2000, on ipngwg mailing list.
1099 	 * This is rather more serious than unicast cases, because some
1100 	 * MLD packets can be sent with the unspecified source address
1101 	 * (although such packets must normally set 1 to the hop limit field).
1102 	 */
1103 	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1104 		IP6STAT_INC(ip6s_cantforward);
1105 		if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
1106 			V_ip6_log_time = time_uptime;
1107 			log(LOG_DEBUG,
1108 			    "cannot forward "
1109 			    "from %s to %s nxt %d received on %s\n",
1110 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1111 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1112 			    ip6->ip6_nxt,
1113 			    if_name(m->m_pkthdr.rcvif));
1114 		}
1115 		return (0);
1116 	}
1117 
1118 	MFC6_LOCK();
1119 
1120 	/*
1121 	 * Determine forwarding mifs from the forwarding cache table
1122 	 */
1123 	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1124 	MRT6STAT_INC(mrt6s_mfc_lookups);
1125 
1126 	/* Entry exists, so forward if necessary */
1127 	if (rt) {
1128 		MFC6_UNLOCK();
1129 		return (ip6_mdq(m, ifp, rt));
1130 	}
1131 
1132 	/*
1133 	 * If we don't have a route for packet's origin,
1134 	 * Make a copy of the packet & send message to routing daemon.
1135 	 */
1136 	MRT6STAT_INC(mrt6s_no_route);
1137 	MRT6_DLOG(DEBUG_FORWARD | DEBUG_MFC, "no rte s %s g %s",
1138 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1139 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1140 
1141 	/*
1142 	 * Allocate mbufs early so that we don't do extra work if we
1143 	 * are just going to fail anyway.
1144 	 */
1145 	rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6, M_NOWAIT);
1146 	if (rte == NULL) {
1147 		MFC6_UNLOCK();
1148 		return (ENOBUFS);
1149 	}
1150 	mb0 = m_copym(m, 0, M_COPYALL, M_NOWAIT);
1151 	/*
1152 	 * Pullup packet header if needed before storing it,
1153 	 * as other references may modify it in the meantime.
1154 	 */
1155 	if (mb0 && (!M_WRITABLE(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1156 		mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1157 	if (mb0 == NULL) {
1158 		free(rte, M_MRTABLE6);
1159 		MFC6_UNLOCK();
1160 		return (ENOBUFS);
1161 	}
1162 
1163 	/* is there an upcall waiting for this packet? */
1164 	hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1165 	for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1166 		if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1167 		    &rt->mf6c_origin.sin6_addr) &&
1168 		    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1169 		    &rt->mf6c_mcastgrp.sin6_addr) && (rt->mf6c_stall != NULL))
1170 			break;
1171 	}
1172 
1173 	if (rt == NULL) {
1174 		struct mrt6msg *im;
1175 #ifdef MRT6_OINIT
1176 		struct omrt6msg *oim;
1177 #endif
1178 		/* no upcall, so make a new entry */
1179 		rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, M_NOWAIT);
1180 		if (rt == NULL) {
1181 			free(rte, M_MRTABLE6);
1182 			m_freem(mb0);
1183 			MFC6_UNLOCK();
1184 			return (ENOBUFS);
1185 		}
1186 		/*
1187 		 * Make a copy of the header to send to the user
1188 		 * level process
1189 		 */
1190 		mm = m_copym(mb0, 0, sizeof(struct ip6_hdr), M_NOWAIT);
1191 		if (mm == NULL) {
1192 			free(rte, M_MRTABLE6);
1193 			m_freem(mb0);
1194 			free(rt, M_MRTABLE6);
1195 			MFC6_UNLOCK();
1196 			return (ENOBUFS);
1197 		}
1198 
1199 		/*
1200 		 * Send message to routing daemon
1201 		 */
1202 		sin6.sin6_addr = ip6->ip6_src;
1203 		im = NULL;
1204 #ifdef MRT6_OINIT
1205 		oim = NULL;
1206 #endif
1207 		switch (V_ip6_mrouter_ver) {
1208 #ifdef MRT6_OINIT
1209 		case MRT6_OINIT:
1210 			oim = mtod(mm, struct omrt6msg *);
1211 			oim->im6_msgtype = MRT6MSG_NOCACHE;
1212 			oim->im6_mbz = 0;
1213 			break;
1214 #endif
1215 		case MRT6_INIT:
1216 			im = mtod(mm, struct mrt6msg *);
1217 			im->im6_msgtype = MRT6MSG_NOCACHE;
1218 			im->im6_mbz = 0;
1219 			break;
1220 		default:
1221 			free(rte, M_MRTABLE6);
1222 			m_freem(mb0);
1223 			free(rt, M_MRTABLE6);
1224 			MFC6_UNLOCK();
1225 			return (EINVAL);
1226 		}
1227 
1228 		MRT6_DLOG(DEBUG_FORWARD, "getting the iif info in the kernel");
1229 		for (mifp = mif6table, mifi = 0;
1230 		    mifi < nummifs && mifp->m6_ifp != ifp; mifp++, mifi++)
1231 				;
1232 
1233 		switch (V_ip6_mrouter_ver) {
1234 #ifdef MRT6_OINIT
1235 		case MRT6_OINIT:
1236 			oim->im6_mif = mifi;
1237 			break;
1238 #endif
1239 		case MRT6_INIT:
1240 			im->im6_mif = mifi;
1241 			break;
1242 		}
1243 
1244 		if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1245 			log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1246 			    "socket queue full\n");
1247 			MRT6STAT_INC(mrt6s_upq_sockfull);
1248 			free(rte, M_MRTABLE6);
1249 			m_freem(mb0);
1250 			free(rt, M_MRTABLE6);
1251 			MFC6_UNLOCK();
1252 			return (ENOBUFS);
1253 		}
1254 
1255 		MRT6STAT_INC(mrt6s_upcalls);
1256 
1257 		/* insert new entry at head of hash chain */
1258 		bzero(rt, sizeof(*rt));
1259 		rt->mf6c_origin.sin6_family = AF_INET6;
1260 		rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1261 		rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1262 		rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1263 		rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1264 		rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1265 		rt->mf6c_expire = UPCALL_EXPIRE;
1266 		n6expire[hash]++;
1267 		rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1268 
1269 		/* link into table */
1270 		rt->mf6c_next  = mf6ctable[hash];
1271 		mf6ctable[hash] = rt;
1272 		/* Add this entry to the end of the queue */
1273 		rt->mf6c_stall = rte;
1274 	} else {
1275 		/* determine if q has overflowed */
1276 		struct rtdetq **p;
1277 		int npkts = 0;
1278 
1279 		for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1280 			if (++npkts > MAX_UPQ6) {
1281 				MRT6STAT_INC(mrt6s_upq_ovflw);
1282 				free(rte, M_MRTABLE6);
1283 				m_freem(mb0);
1284 				MFC6_UNLOCK();
1285 				return (0);
1286 			}
1287 
1288 		/* Add this entry to the end of the queue */
1289 		*p = rte;
1290 	}
1291 
1292 	rte->next = NULL;
1293 	rte->m = mb0;
1294 	rte->ifp = ifp;
1295 #ifdef UPCALL_TIMING
1296 	rte->t = tp;
1297 #endif /* UPCALL_TIMING */
1298 
1299 	MFC6_UNLOCK();
1300 
1301 	return (0);
1302 }
1303 
1304 /*
1305  * Clean up cache entries if upcalls are not serviced
1306  * Call from the Slow Timeout mechanism, every half second.
1307  */
1308 static void
1309 expire_upcalls(void *unused)
1310 {
1311 #ifdef MRT6DEBUG
1312 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1313 #endif
1314 	struct rtdetq *rte;
1315 	struct mf6c *mfc, **nptr;
1316 	u_long i;
1317 
1318 	MFC6_LOCK();
1319 	for (i = 0; i < MF6CTBLSIZ; i++) {
1320 		if (n6expire[i] == 0)
1321 			continue;
1322 		nptr = &mf6ctable[i];
1323 		while ((mfc = *nptr) != NULL) {
1324 			rte = mfc->mf6c_stall;
1325 			/*
1326 			 * Skip real cache entries
1327 			 * Make sure it wasn't marked to not expire (shouldn't happen)
1328 			 * If it expires now
1329 			 */
1330 			if (rte != NULL &&
1331 			    mfc->mf6c_expire != 0 &&
1332 			    --mfc->mf6c_expire == 0) {
1333 				MRT6_DLOG(DEBUG_EXPIRE, "expiring (%s %s)",
1334 				    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1335 				    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1336 				/*
1337 				 * drop all the packets
1338 				 * free the mbuf with the pkt, if, timing info
1339 				 */
1340 				do {
1341 					struct rtdetq *n = rte->next;
1342 					m_freem(rte->m);
1343 					free(rte, M_MRTABLE6);
1344 					rte = n;
1345 				} while (rte != NULL);
1346 				MRT6STAT_INC(mrt6s_cache_cleanups);
1347 				n6expire[i]--;
1348 
1349 				*nptr = mfc->mf6c_next;
1350 				free(mfc, M_MRTABLE6);
1351 			} else {
1352 				nptr = &mfc->mf6c_next;
1353 			}
1354 		}
1355 	}
1356 	MFC6_UNLOCK();
1357 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1358 	    expire_upcalls, NULL);
1359 }
1360 
1361 /*
1362  * Packet forwarding routine once entry in the cache is made
1363  */
1364 static int
1365 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1366 {
1367 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1368 	mifi_t mifi, iif;
1369 	struct mif6 *mifp;
1370 	int plen = m->m_pkthdr.len;
1371 	struct in6_addr src0, dst0; /* copies for local work */
1372 	u_int32_t iszone, idzone, oszone, odzone;
1373 	int error = 0;
1374 
1375 	/*
1376 	 * Don't forward if it didn't arrive from the parent mif
1377 	 * for its origin.
1378 	 */
1379 	mifi = rt->mf6c_parent;
1380 	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1381 		/* came in the wrong interface */
1382 		MRT6_DLOG(DEBUG_FORWARD,
1383 		    "wrong if: ifid %d mifi %d mififid %x", ifp->if_index,
1384 		    mifi, mif6table[mifi].m6_ifp->if_index);
1385 		MRT6STAT_INC(mrt6s_wrong_if);
1386 		rt->mf6c_wrong_if++;
1387 		/*
1388 		 * If we are doing PIM processing, and we are forwarding
1389 		 * packets on this interface, send a message to the
1390 		 * routing daemon.
1391 		 */
1392 		/* have to make sure this is a valid mif */
1393 		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1394 			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1395 				/*
1396 				 * Check the M_LOOP flag to avoid an
1397 				 * unnecessary PIM assert.
1398 				 * XXX: M_LOOP is an ad-hoc hack...
1399 				 */
1400 				static struct sockaddr_in6 sin6 =
1401 				{ sizeof(sin6), AF_INET6 };
1402 
1403 				struct mbuf *mm;
1404 				struct mrt6msg *im;
1405 #ifdef MRT6_OINIT
1406 				struct omrt6msg *oim;
1407 #endif
1408 
1409 				mm = m_copym(m, 0, sizeof(struct ip6_hdr),
1410 				    M_NOWAIT);
1411 				if (mm &&
1412 				    (!M_WRITABLE(mm) ||
1413 				     mm->m_len < sizeof(struct ip6_hdr)))
1414 					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1415 				if (mm == NULL)
1416 					return (ENOBUFS);
1417 
1418 #ifdef MRT6_OINIT
1419 				oim = NULL;
1420 #endif
1421 				im = NULL;
1422 				switch (V_ip6_mrouter_ver) {
1423 #ifdef MRT6_OINIT
1424 				case MRT6_OINIT:
1425 					oim = mtod(mm, struct omrt6msg *);
1426 					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1427 					oim->im6_mbz = 0;
1428 					break;
1429 #endif
1430 				case MRT6_INIT:
1431 					im = mtod(mm, struct mrt6msg *);
1432 					im->im6_msgtype = MRT6MSG_WRONGMIF;
1433 					im->im6_mbz = 0;
1434 					break;
1435 				default:
1436 					m_freem(mm);
1437 					return (EINVAL);
1438 				}
1439 
1440 				for (mifp = mif6table, iif = 0;
1441 				     iif < nummifs && mifp &&
1442 					     mifp->m6_ifp != ifp;
1443 				     mifp++, iif++)
1444 					;
1445 
1446 				switch (V_ip6_mrouter_ver) {
1447 #ifdef MRT6_OINIT
1448 				case MRT6_OINIT:
1449 					oim->im6_mif = iif;
1450 					sin6.sin6_addr = oim->im6_src;
1451 					break;
1452 #endif
1453 				case MRT6_INIT:
1454 					im->im6_mif = iif;
1455 					sin6.sin6_addr = im->im6_src;
1456 					break;
1457 				}
1458 
1459 				MRT6STAT_INC(mrt6s_upcalls);
1460 
1461 				if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1462 					MRT6_DLOG(DEBUG_ANY,
1463 					    "ip6_mrouter socket queue full");
1464 					MRT6STAT_INC(mrt6s_upq_sockfull);
1465 					return (ENOBUFS);
1466 				}	/* if socket Q full */
1467 			}		/* if PIM */
1468 		return (0);
1469 	}			/* if wrong iif */
1470 
1471 	/* If I sourced this packet, it counts as output, else it was input. */
1472 	if (m->m_pkthdr.rcvif == NULL) {
1473 		/* XXX: is rcvif really NULL when output?? */
1474 		mif6table[mifi].m6_pkt_out++;
1475 		mif6table[mifi].m6_bytes_out += plen;
1476 	} else {
1477 		mif6table[mifi].m6_pkt_in++;
1478 		mif6table[mifi].m6_bytes_in += plen;
1479 	}
1480 	rt->mf6c_pkt_cnt++;
1481 	rt->mf6c_byte_cnt += plen;
1482 
1483 	/*
1484 	 * For each mif, forward a copy of the packet if there are group
1485 	 * members downstream on the interface.
1486 	 */
1487 	src0 = ip6->ip6_src;
1488 	dst0 = ip6->ip6_dst;
1489 	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1490 	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1491 		IP6STAT_INC(ip6s_badscope);
1492 		return (error);
1493 	}
1494 	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1495 		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1496 			/*
1497 			 * check if the outgoing packet is going to break
1498 			 * a scope boundary.
1499 			 * XXX For packets through PIM register tunnel
1500 			 * interface, we believe a routing daemon.
1501 			 */
1502 			if (!(mif6table[rt->mf6c_parent].m6_flags &
1503 			      MIFF_REGISTER) &&
1504 			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1505 				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1506 				    &oszone) ||
1507 				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1508 				    &odzone) ||
1509 				    iszone != oszone ||
1510 				    idzone != odzone) {
1511 					IP6STAT_INC(ip6s_badscope);
1512 					continue;
1513 				}
1514 			}
1515 
1516 			mifp->m6_pkt_out++;
1517 			mifp->m6_bytes_out += plen;
1518 			if (mifp->m6_flags & MIFF_REGISTER)
1519 				register_send(ip6, mifp, m);
1520 			else
1521 				phyint_send(ip6, mifp, m);
1522 		}
1523 	}
1524 	return (0);
1525 }
1526 
1527 static void
1528 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1529 {
1530 #ifdef MRT6DEBUG
1531 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1532 #endif
1533 	struct mbuf *mb_copy;
1534 	struct ifnet *ifp = mifp->m6_ifp;
1535 	int error = 0;
1536 	u_long linkmtu;
1537 
1538 	/*
1539 	 * Make a new reference to the packet; make sure that
1540 	 * the IPv6 header is actually copied, not just referenced,
1541 	 * so that ip6_output() only scribbles on the copy.
1542 	 */
1543 	mb_copy = m_copym(m, 0, M_COPYALL, M_NOWAIT);
1544 	if (mb_copy &&
1545 	    (!M_WRITABLE(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1546 		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1547 	if (mb_copy == NULL) {
1548 		return;
1549 	}
1550 	/* set MCAST flag to the outgoing packet */
1551 	mb_copy->m_flags |= M_MCAST;
1552 
1553 	/*
1554 	 * If we sourced the packet, call ip6_output since we may devide
1555 	 * the packet into fragments when the packet is too big for the
1556 	 * outgoing interface.
1557 	 * Otherwise, we can simply send the packet to the interface
1558 	 * sending queue.
1559 	 */
1560 	if (m->m_pkthdr.rcvif == NULL) {
1561 		struct ip6_moptions im6o;
1562 		struct epoch_tracker et;
1563 
1564 		im6o.im6o_multicast_ifp = ifp;
1565 		/* XXX: ip6_output will override ip6->ip6_hlim */
1566 		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1567 		im6o.im6o_multicast_loop = 1;
1568 		NET_EPOCH_ENTER(et);
1569 		error = ip6_output(mb_copy, NULL, NULL, IPV6_FORWARDING, &im6o,
1570 		    NULL, NULL);
1571 		NET_EPOCH_EXIT(et);
1572 
1573 		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1574 		    (uint16_t)(mifp - mif6table), error);
1575 		return;
1576 	}
1577 
1578 	/*
1579 	 * If configured to loop back multicasts by default,
1580 	 * loop back a copy now.
1581 	 */
1582 	if (in6_mcast_loop)
1583 		ip6_mloopback(ifp, m);
1584 
1585 	/*
1586 	 * Put the packet into the sending queue of the outgoing interface
1587 	 * if it would fit in the MTU of the interface.
1588 	 */
1589 	linkmtu = IN6_LINKMTU(ifp);
1590 	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1591 		struct sockaddr_in6 dst6;
1592 
1593 		bzero(&dst6, sizeof(dst6));
1594 		dst6.sin6_len = sizeof(struct sockaddr_in6);
1595 		dst6.sin6_family = AF_INET6;
1596 		dst6.sin6_addr = ip6->ip6_dst;
1597 
1598 		IP_PROBE(send, NULL, NULL, ip6, ifp, NULL, ip6);
1599 		/*
1600 		 * We just call if_output instead of nd6_output here, since
1601 		 * we need no ND for a multicast forwarded packet...right?
1602 		 */
1603 		m_clrprotoflags(m);	/* Avoid confusing lower layers. */
1604 		error = (*ifp->if_output)(ifp, mb_copy,
1605 		    (struct sockaddr *)&dst6, NULL);
1606 		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1607 		    (uint16_t)(mifp - mif6table), error);
1608 	} else {
1609 		/*
1610 		 * pMTU discovery is intentionally disabled by default, since
1611 		 * various router may notify pMTU in multicast, which can be
1612 		 * a DDoS to a router
1613 		 */
1614 		if (V_ip6_mcast_pmtu)
1615 			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1616 		else {
1617 			MRT6_DLOG(DEBUG_XMIT, " packet too big on %s o %s "
1618 			    "g %s size %d (discarded)", if_name(ifp),
1619 			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1620 			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1621 			    mb_copy->m_pkthdr.len);
1622 			m_freem(mb_copy); /* simply discard the packet */
1623 		}
1624 	}
1625 }
1626 
1627 static int
1628 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1629 {
1630 #ifdef MRT6DEBUG
1631 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1632 #endif
1633 	struct mbuf *mm;
1634 	int i, len = m->m_pkthdr.len;
1635 	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1636 	struct mrt6msg *im6;
1637 
1638 	MRT6_DLOG(DEBUG_ANY, "src %s dst %s",
1639 	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1640 	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1641 	PIM6STAT_INC(pim6s_snd_registers);
1642 
1643 	/* Make a copy of the packet to send to the user level process. */
1644 	mm = m_gethdr(M_NOWAIT, MT_DATA);
1645 	if (mm == NULL)
1646 		return (ENOBUFS);
1647 	mm->m_data += max_linkhdr;
1648 	mm->m_len = sizeof(struct ip6_hdr);
1649 
1650 	if ((mm->m_next = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL) {
1651 		m_freem(mm);
1652 		return (ENOBUFS);
1653 	}
1654 	i = MHLEN - M_LEADINGSPACE(mm);
1655 	if (i > len)
1656 		i = len;
1657 	mm = m_pullup(mm, i);
1658 	if (mm == NULL)
1659 		return (ENOBUFS);
1660 /* TODO: check it! */
1661 	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1662 
1663 	/*
1664 	 * Send message to routing daemon
1665 	 */
1666 	sin6.sin6_addr = ip6->ip6_src;
1667 
1668 	im6 = mtod(mm, struct mrt6msg *);
1669 	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1670 	im6->im6_mbz          = 0;
1671 
1672 	im6->im6_mif = mif - mif6table;
1673 
1674 	/* iif info is not given for reg. encap.n */
1675 	MRT6STAT_INC(mrt6s_upcalls);
1676 
1677 	if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1678 		MRT6_DLOG(DEBUG_ANY, "ip6_mrouter socket queue full");
1679 		MRT6STAT_INC(mrt6s_upq_sockfull);
1680 		return (ENOBUFS);
1681 	}
1682 	return (0);
1683 }
1684 
1685 /*
1686  * pim6_encapcheck() is called by the encap6_input() path at runtime to
1687  * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1688  * into the kernel.
1689  */
1690 static int
1691 pim6_encapcheck(const struct mbuf *m __unused, int off __unused,
1692     int proto __unused, void *arg __unused)
1693 {
1694 
1695     KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1696     return (8);		/* claim the datagram. */
1697 }
1698 
1699 /*
1700  * PIM sparse mode hook
1701  * Receives the pim control messages, and passes them up to the listening
1702  * socket, using rip6_input.
1703  * The only message processed is the REGISTER pim message; the pim header
1704  * is stripped off, and the inner packet is passed to register_mforward.
1705  */
1706 static int
1707 pim6_input(struct mbuf *m, int off, int proto, void *arg __unused)
1708 {
1709 	struct pim *pim; /* pointer to a pim struct */
1710 	struct ip6_hdr *ip6;
1711 	int pimlen;
1712 	int minlen;
1713 
1714 	PIM6STAT_INC(pim6s_rcv_total);
1715 
1716 	/*
1717 	 * Validate lengths
1718 	 */
1719 	pimlen = m->m_pkthdr.len - off;
1720 	if (pimlen < PIM_MINLEN) {
1721 		PIM6STAT_INC(pim6s_rcv_tooshort);
1722 		MRT6_DLOG(DEBUG_PIM, "PIM packet too short");
1723 		m_freem(m);
1724 		return (IPPROTO_DONE);
1725 	}
1726 
1727 	/*
1728 	 * if the packet is at least as big as a REGISTER, go ahead
1729 	 * and grab the PIM REGISTER header size, to avoid another
1730 	 * possible m_pullup() later.
1731 	 *
1732 	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1733 	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1734 	 */
1735 	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1736 
1737 	/*
1738 	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1739 	 * possibly the PIM REGISTER header
1740 	 */
1741 	if (m->m_len < off + minlen) {
1742 		m = m_pullup(m, off + minlen);
1743 		if (m == NULL) {
1744 			IP6STAT_INC(ip6s_exthdrtoolong);
1745 			return (IPPROTO_DONE);
1746 		}
1747 	}
1748 	ip6 = mtod(m, struct ip6_hdr *);
1749 	pim = (struct pim *)((caddr_t)ip6 + off);
1750 
1751 #define PIM6_CHECKSUM
1752 #ifdef PIM6_CHECKSUM
1753 	{
1754 		int cksumlen;
1755 
1756 		/*
1757 		 * Validate checksum.
1758 		 * If PIM REGISTER, exclude the data packet
1759 		 */
1760 		if (pim->pim_type == PIM_REGISTER)
1761 			cksumlen = PIM_MINLEN;
1762 		else
1763 			cksumlen = pimlen;
1764 
1765 		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1766 			PIM6STAT_INC(pim6s_rcv_badsum);
1767 			MRT6_DLOG(DEBUG_PIM, "invalid checksum");
1768 			m_freem(m);
1769 			return (IPPROTO_DONE);
1770 		}
1771 	}
1772 #endif /* PIM_CHECKSUM */
1773 
1774 	/* PIM version check */
1775 	if (pim->pim_ver != PIM_VERSION) {
1776 		PIM6STAT_INC(pim6s_rcv_badversion);
1777 		MRT6_DLOG(DEBUG_ANY | DEBUG_ERR,
1778 		    "incorrect version %d, expecting %d",
1779 		    pim->pim_ver, PIM_VERSION);
1780 		m_freem(m);
1781 		return (IPPROTO_DONE);
1782 	}
1783 
1784 	if (pim->pim_type == PIM_REGISTER) {
1785 		/*
1786 		 * since this is a REGISTER, we'll make a copy of the register
1787 		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1788 		 * routing daemon.
1789 		 */
1790 		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1791 
1792 		struct mbuf *mcp;
1793 		struct ip6_hdr *eip6;
1794 		u_int32_t *reghdr;
1795 		int rc;
1796 #ifdef MRT6DEBUG
1797 		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1798 #endif
1799 
1800 		PIM6STAT_INC(pim6s_rcv_registers);
1801 
1802 		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1803 			MRT6_DLOG(DEBUG_PIM, "register mif not set: %d",
1804 			    reg_mif_num);
1805 			m_freem(m);
1806 			return (IPPROTO_DONE);
1807 		}
1808 
1809 		reghdr = (u_int32_t *)(pim + 1);
1810 
1811 		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1812 			goto pim6_input_to_daemon;
1813 
1814 		/*
1815 		 * Validate length
1816 		 */
1817 		if (pimlen < PIM6_REG_MINLEN) {
1818 			PIM6STAT_INC(pim6s_rcv_tooshort);
1819 			PIM6STAT_INC(pim6s_rcv_badregisters);
1820 			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "register packet "
1821 			    "size too small %d from %s",
1822 			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1823 			m_freem(m);
1824 			return (IPPROTO_DONE);
1825 		}
1826 
1827 		eip6 = (struct ip6_hdr *) (reghdr + 1);
1828 		MRT6_DLOG(DEBUG_PIM, "eip6: %s -> %s, eip6 plen %d",
1829 		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1830 		    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1831 		    ntohs(eip6->ip6_plen));
1832 
1833 		/* verify the version number of the inner packet */
1834 		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1835 			PIM6STAT_INC(pim6s_rcv_badregisters);
1836 			MRT6_DLOG(DEBUG_ANY, "invalid IP version (%d) "
1837 			    "of the inner packet",
1838 			    (eip6->ip6_vfc & IPV6_VERSION));
1839 			m_freem(m);
1840 			return (IPPROTO_DONE);
1841 		}
1842 
1843 		/* verify the inner packet is destined to a mcast group */
1844 		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1845 			PIM6STAT_INC(pim6s_rcv_badregisters);
1846 			MRT6_DLOG(DEBUG_PIM, "inner packet of register "
1847 			    "is not multicast %s",
1848 			    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1849 			m_freem(m);
1850 			return (IPPROTO_DONE);
1851 		}
1852 
1853 		/*
1854 		 * make a copy of the whole header to pass to the daemon later.
1855 		 */
1856 		mcp = m_copym(m, 0, off + PIM6_REG_MINLEN, M_NOWAIT);
1857 		if (mcp == NULL) {
1858 			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "pim register: "
1859 			    "could not copy register head");
1860 			m_freem(m);
1861 			return (IPPROTO_DONE);
1862 		}
1863 
1864 		/*
1865 		 * forward the inner ip6 packet; point m_data at the inner ip6.
1866 		 */
1867 		m_adj(m, off + PIM_MINLEN);
1868 		MRT6_DLOG(DEBUG_PIM, "forwarding decapsulated register: "
1869 		    "src %s, dst %s, mif %d",
1870 		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1871 		    ip6_sprintf(ip6bufd, &eip6->ip6_dst), reg_mif_num);
1872 
1873 		rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1874 				dst.sin6_family, 0);
1875 
1876 		/* prepare the register head to send to the mrouting daemon */
1877 		m = mcp;
1878 	}
1879 
1880 	/*
1881 	 * Pass the PIM message up to the daemon; if it is a register message
1882 	 * pass the 'head' only up to the daemon. This includes the
1883 	 * encapsulator ip6 header, pim header, register header and the
1884 	 * encapsulated ip6 header.
1885 	 */
1886   pim6_input_to_daemon:
1887 	return (rip6_input(&m, &off, proto));
1888 }
1889 
1890 static int
1891 ip6_mroute_modevent(module_t mod, int type, void *unused)
1892 {
1893 
1894 	switch (type) {
1895 	case MOD_LOAD:
1896 		MROUTER6_LOCK_INIT();
1897 		MFC6_LOCK_INIT();
1898 		MIF6_LOCK_INIT();
1899 
1900 		pim6_encap_cookie = ip6_encap_attach(&ipv6_encap_cfg,
1901 		    NULL, M_WAITOK);
1902 		if (pim6_encap_cookie == NULL) {
1903 			printf("ip6_mroute: unable to attach pim6 encap\n");
1904 			MIF6_LOCK_DESTROY();
1905 			MFC6_LOCK_DESTROY();
1906 			MROUTER6_LOCK_DESTROY();
1907 			return (EINVAL);
1908 		}
1909 
1910 		ip6_mforward = X_ip6_mforward;
1911 		ip6_mrouter_done = X_ip6_mrouter_done;
1912 		ip6_mrouter_get = X_ip6_mrouter_get;
1913 		ip6_mrouter_set = X_ip6_mrouter_set;
1914 		mrt6_ioctl = X_mrt6_ioctl;
1915 		break;
1916 
1917 	case MOD_UNLOAD:
1918 		if (V_ip6_mrouter != NULL)
1919 			return EINVAL;
1920 
1921 		if (pim6_encap_cookie) {
1922 			ip6_encap_detach(pim6_encap_cookie);
1923 			pim6_encap_cookie = NULL;
1924 		}
1925 		X_ip6_mrouter_done();
1926 		ip6_mforward = NULL;
1927 		ip6_mrouter_done = NULL;
1928 		ip6_mrouter_get = NULL;
1929 		ip6_mrouter_set = NULL;
1930 		mrt6_ioctl = NULL;
1931 
1932 		MIF6_LOCK_DESTROY();
1933 		MFC6_LOCK_DESTROY();
1934 		MROUTER6_LOCK_DESTROY();
1935 		break;
1936 
1937 	default:
1938 		return (EOPNOTSUPP);
1939 	}
1940 
1941 	return (0);
1942 }
1943 
1944 static moduledata_t ip6_mroutemod = {
1945 	"ip6_mroute",
1946 	ip6_mroute_modevent,
1947 	0
1948 };
1949 
1950 DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PROTO_MC, SI_ORDER_ANY);
1951