1/*-
2 * Copyright (C) 1998 WIDE Project.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the project nor the names of its contributors
14 *    may be used to endorse or promote products derived from this software
15 *    without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 *	$KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $
30 */
31
32/*-
33 * Copyright (c) 1989 Stephen Deering
34 * Copyright (c) 1992, 1993
35 *      The Regents of the University of California.  All rights reserved.
36 *
37 * This code is derived from software contributed to Berkeley by
38 * Stephen Deering of Stanford University.
39 *
40 * Redistribution and use in source and binary forms, with or without
41 * modification, are permitted provided that the following conditions
42 * are met:
43 * 1. Redistributions of source code must retain the above copyright
44 *    notice, this list of conditions and the following disclaimer.
45 * 2. Redistributions in binary form must reproduce the above copyright
46 *    notice, this list of conditions and the following disclaimer in the
47 *    documentation and/or other materials provided with the distribution.
48 * 4. Neither the name of the University nor the names of its contributors
49 *    may be used to endorse or promote products derived from this software
50 *    without specific prior written permission.
51 *
52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62 * SUCH DAMAGE.
63 *
64 *	@(#)ip_mroute.c	8.2 (Berkeley) 11/15/93
65 *	BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp
66 */
67
68/*
69 * IP multicast forwarding procedures
70 *
71 * Written by David Waitzman, BBN Labs, August 1988.
72 * Modified by Steve Deering, Stanford, February 1989.
73 * Modified by Mark J. Steiglitz, Stanford, May, 1991
74 * Modified by Van Jacobson, LBL, January 1993
75 * Modified by Ajit Thyagarajan, PARC, August 1993
76 * Modified by Bill Fenner, PARC, April 1994
77 *
78 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
79 */
80
81#include <sys/cdefs.h>
82__FBSDID("$FreeBSD: stable/11/sys/netinet6/ip6_mroute.c 340923 2018-11-25 18:00:50Z markj $");
83
84#include "opt_inet6.h"
85
86#include <sys/param.h>
87#include <sys/callout.h>
88#include <sys/errno.h>
89#include <sys/kernel.h>
90#include <sys/lock.h>
91#include <sys/malloc.h>
92#include <sys/mbuf.h>
93#include <sys/module.h>
94#include <sys/domain.h>
95#include <sys/protosw.h>
96#include <sys/sdt.h>
97#include <sys/signalvar.h>
98#include <sys/socket.h>
99#include <sys/socketvar.h>
100#include <sys/sockio.h>
101#include <sys/sx.h>
102#include <sys/sysctl.h>
103#include <sys/syslog.h>
104#include <sys/systm.h>
105#include <sys/time.h>
106
107#include <net/if.h>
108#include <net/if_var.h>
109#include <net/if_types.h>
110#include <net/raw_cb.h>
111#include <net/vnet.h>
112
113#include <netinet/in.h>
114#include <netinet/in_var.h>
115#include <netinet/icmp6.h>
116#include <netinet/ip_encap.h>
117
118#include <netinet/ip6.h>
119#include <netinet/in_kdtrace.h>
120#include <netinet6/ip6_var.h>
121#include <netinet6/scope6_var.h>
122#include <netinet6/nd6.h>
123#include <netinet6/ip6_mroute.h>
124#include <netinet6/pim6.h>
125#include <netinet6/pim6_var.h>
126
127static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
128
129static int	ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
130static void	phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
131static int	register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
132static int	set_pim6(int *);
133static int	socket_send(struct socket *, struct mbuf *,
134		    struct sockaddr_in6 *);
135
136extern int in6_mcast_loop;
137extern struct domain inet6domain;
138
139static const struct encaptab *pim6_encap_cookie;
140static const struct protosw in6_pim_protosw = {
141	.pr_type =		SOCK_RAW,
142	.pr_domain =		&inet6domain,
143	.pr_protocol =		IPPROTO_PIM,
144	.pr_flags =		PR_ATOMIC|PR_ADDR|PR_LASTHDR,
145	.pr_input =		pim6_input,
146	.pr_output =		rip6_output,
147	.pr_ctloutput =		rip6_ctloutput,
148	.pr_usrreqs =		&rip6_usrreqs
149};
150static int pim6_encapcheck(const struct mbuf *, int, int, void *);
151
152static VNET_DEFINE(int, ip6_mrouter_ver) = 0;
153#define	V_ip6_mrouter_ver	VNET(ip6_mrouter_ver)
154
155SYSCTL_DECL(_net_inet6);
156SYSCTL_DECL(_net_inet6_ip6);
157static SYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim, CTLFLAG_RW, 0, "PIM");
158
159static struct mrt6stat mrt6stat;
160SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
161    &mrt6stat, mrt6stat,
162    "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
163
164#define	MRT6STAT_INC(name)	mrt6stat.name += 1
165#define NO_RTE_FOUND	0x1
166#define RTE_FOUND	0x2
167
168static struct mtx mrouter6_mtx;
169#define	MROUTER6_LOCK()		mtx_lock(&mrouter6_mtx)
170#define	MROUTER6_UNLOCK()	mtx_unlock(&mrouter6_mtx)
171#define	MROUTER6_LOCK_ASSERT()	do {					\
172	mtx_assert(&mrouter6_mtx, MA_OWNED);				\
173	NET_ASSERT_GIANT();						\
174} while (0)
175#define	MROUTER6_LOCK_INIT()	\
176	mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
177#define	MROUTER6_LOCK_DESTROY()	mtx_destroy(&mrouter6_mtx)
178
179static struct mf6c *mf6ctable[MF6CTBLSIZ];
180SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
181    &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
182    "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
183    "netinet6/ip6_mroute.h)");
184
185static struct mtx mfc6_mtx;
186#define	MFC6_LOCK()		mtx_lock(&mfc6_mtx)
187#define	MFC6_UNLOCK()		mtx_unlock(&mfc6_mtx)
188#define	MFC6_LOCK_ASSERT()	do {					\
189	mtx_assert(&mfc6_mtx, MA_OWNED);				\
190	NET_ASSERT_GIANT();						\
191} while (0)
192#define	MFC6_LOCK_INIT()		\
193	mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
194#define	MFC6_LOCK_DESTROY()	mtx_destroy(&mfc6_mtx)
195
196static u_char n6expire[MF6CTBLSIZ];
197
198static struct mif6 mif6table[MAXMIFS];
199static int
200sysctl_mif6table(SYSCTL_HANDLER_ARGS)
201{
202	struct mif6_sctl *out;
203	int error;
204
205	out = malloc(sizeof(struct mif6_sctl) * MAXMIFS, M_TEMP,
206	    M_WAITOK | M_ZERO);
207	for (int i = 0; i < MAXMIFS; i++) {
208		out[i].m6_flags		= mif6table[i].m6_flags;
209		out[i].m6_rate_limit	= mif6table[i].m6_rate_limit;
210		out[i].m6_lcl_addr	= mif6table[i].m6_lcl_addr;
211		if (mif6table[i].m6_ifp != NULL)
212			out[i].m6_ifp	= mif6table[i].m6_ifp->if_index;
213		else
214			out[i].m6_ifp	= 0;
215		out[i].m6_pkt_in	= mif6table[i].m6_pkt_in;
216		out[i].m6_pkt_out	= mif6table[i].m6_pkt_out;
217		out[i].m6_bytes_in	= mif6table[i].m6_bytes_in;
218		out[i].m6_bytes_out	= mif6table[i].m6_bytes_out;
219	}
220	error = SYSCTL_OUT(req, out, sizeof(struct mif6_sctl) * MAXMIFS);
221	free(out, M_TEMP);
222	return (error);
223}
224SYSCTL_PROC(_net_inet6_ip6, OID_AUTO, mif6table, CTLTYPE_OPAQUE | CTLFLAG_RD,
225    NULL, 0, sysctl_mif6table, "S,mif6_sctl[MAXMIFS]",
226    "IPv6 Multicast Interfaces (struct mif6_sctl[MAXMIFS], "
227    "netinet6/ip6_mroute.h)");
228
229static struct mtx mif6_mtx;
230#define	MIF6_LOCK()		mtx_lock(&mif6_mtx)
231#define	MIF6_UNLOCK()		mtx_unlock(&mif6_mtx)
232#define	MIF6_LOCK_ASSERT()	mtx_assert(&mif6_mtx, MA_OWNED)
233#define	MIF6_LOCK_INIT()	\
234	mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
235#define	MIF6_LOCK_DESTROY()	mtx_destroy(&mif6_mtx)
236
237#ifdef MRT6DEBUG
238static VNET_DEFINE(u_int, mrt6debug) = 0;	/* debug level */
239#define	V_mrt6debug		VNET(mrt6debug)
240#define DEBUG_MFC	0x02
241#define DEBUG_FORWARD	0x04
242#define DEBUG_EXPIRE	0x08
243#define DEBUG_XMIT	0x10
244#define DEBUG_REG	0x20
245#define DEBUG_PIM	0x40
246#define	DEBUG_ERR	0x80
247#define	DEBUG_ANY	0x7f
248#define	MRT6_DLOG(m, fmt, ...)	\
249	if (V_mrt6debug & (m))	\
250		log(((m) & DEBUG_ERR) ? LOG_ERR: LOG_DEBUG, \
251		    "%s: " fmt "\n", __func__, ##__VA_ARGS__)
252#else
253#define	MRT6_DLOG(m, fmt, ...)
254#endif
255
256static void	expire_upcalls(void *);
257#define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
258#define	UPCALL_EXPIRE	6		/* number of timeouts */
259
260/*
261 * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
262 */
263
264/*
265 * 'Interfaces' associated with decapsulator (so we can tell
266 * packets that went through it from ones that get reflected
267 * by a broken gateway).  Different from IPv4 register_if,
268 * these interfaces are linked into the system ifnet list,
269 * because per-interface IPv6 statistics are maintained in
270 * ifp->if_afdata.  But it does not have any routes point
271 * to them.  I.e., packets can't be sent this way.  They
272 * only exist as a placeholder for multicast source
273 * verification.
274 */
275static struct ifnet *multicast_register_if6;
276
277#define ENCAP_HOPS 64
278
279/*
280 * Private variables.
281 */
282static mifi_t nummifs = 0;
283static mifi_t reg_mif_num = (mifi_t)-1;
284
285static struct pim6stat pim6stat;
286SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RW,
287    &pim6stat, pim6stat,
288    "PIM Statistics (struct pim6stat, netinet6/pim6_var.h)");
289
290#define	PIM6STAT_INC(name)	pim6stat.name += 1
291static VNET_DEFINE(int, pim6);
292#define	V_pim6		VNET(pim6)
293
294/*
295 * Hash function for a source, group entry
296 */
297#define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
298				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
299				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
300				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
301
302/*
303 * Find a route for a given origin IPv6 address and Multicast group address.
304 */
305#define MF6CFIND(o, g, rt) do { \
306	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
307	rt = NULL; \
308	while (_rt) { \
309		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
310		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
311		    (_rt->mf6c_stall == NULL)) { \
312			rt = _rt; \
313			break; \
314		} \
315		_rt = _rt->mf6c_next; \
316	} \
317	if (rt == NULL) { \
318		MRT6STAT_INC(mrt6s_mfc_misses); \
319	} \
320} while (/*CONSTCOND*/ 0)
321
322/*
323 * Macros to compute elapsed time efficiently
324 * Borrowed from Van Jacobson's scheduling code
325 * XXX: replace with timersub() ?
326 */
327#define TV_DELTA(a, b, delta) do { \
328	    int xxs; \
329		\
330	    delta = (a).tv_usec - (b).tv_usec; \
331	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
332	       switch (xxs) { \
333		      case 2: \
334			  delta += 1000000; \
335			      /* FALLTHROUGH */ \
336		      case 1: \
337			  delta += 1000000; \
338			  break; \
339		      default: \
340			  delta += (1000000 * xxs); \
341	       } \
342	    } \
343} while (/*CONSTCOND*/ 0)
344
345/* XXX: replace with timercmp(a, b, <) ? */
346#define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
347	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
348
349#ifdef UPCALL_TIMING
350#define UPCALL_MAX	50
351static u_long upcall_data[UPCALL_MAX + 1];
352static void collate();
353#endif /* UPCALL_TIMING */
354
355static int ip6_mrouter_init(struct socket *, int, int);
356static int add_m6fc(struct mf6cctl *);
357static int add_m6if(struct mif6ctl *);
358static int del_m6fc(struct mf6cctl *);
359static int del_m6if(mifi_t *);
360static int del_m6if_locked(mifi_t *);
361static int get_mif6_cnt(struct sioc_mif_req6 *);
362static int get_sg_cnt(struct sioc_sg_req6 *);
363
364static struct callout expire_upcalls_ch;
365
366int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
367int X_ip6_mrouter_done(void);
368int X_ip6_mrouter_set(struct socket *, struct sockopt *);
369int X_ip6_mrouter_get(struct socket *, struct sockopt *);
370int X_mrt6_ioctl(u_long, caddr_t);
371
372/*
373 * Handle MRT setsockopt commands to modify the multicast routing tables.
374 */
375int
376X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
377{
378	int error = 0;
379	int optval;
380	struct mif6ctl mifc;
381	struct mf6cctl mfcc;
382	mifi_t mifi;
383
384	if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
385		return (EPERM);
386
387	switch (sopt->sopt_name) {
388	case MRT6_INIT:
389#ifdef MRT6_OINIT
390	case MRT6_OINIT:
391#endif
392		error = sooptcopyin(sopt, &optval, sizeof(optval),
393		    sizeof(optval));
394		if (error)
395			break;
396		error = ip6_mrouter_init(so, optval, sopt->sopt_name);
397		break;
398	case MRT6_DONE:
399		error = X_ip6_mrouter_done();
400		break;
401	case MRT6_ADD_MIF:
402		error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
403		if (error)
404			break;
405		error = add_m6if(&mifc);
406		break;
407	case MRT6_ADD_MFC:
408		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
409		if (error)
410			break;
411		error = add_m6fc(&mfcc);
412		break;
413	case MRT6_DEL_MFC:
414		error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
415		if (error)
416			break;
417		error = del_m6fc(&mfcc);
418		break;
419	case MRT6_DEL_MIF:
420		error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
421		if (error)
422			break;
423		error = del_m6if(&mifi);
424		break;
425	case MRT6_PIM:
426		error = sooptcopyin(sopt, &optval, sizeof(optval),
427		    sizeof(optval));
428		if (error)
429			break;
430		error = set_pim6(&optval);
431		break;
432	default:
433		error = EOPNOTSUPP;
434		break;
435	}
436
437	return (error);
438}
439
440/*
441 * Handle MRT getsockopt commands
442 */
443int
444X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
445{
446	int error = 0;
447
448	if (so != V_ip6_mrouter)
449		return (EACCES);
450
451	switch (sopt->sopt_name) {
452		case MRT6_PIM:
453			error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
454			break;
455	}
456	return (error);
457}
458
459/*
460 * Handle ioctl commands to obtain information from the cache
461 */
462int
463X_mrt6_ioctl(u_long cmd, caddr_t data)
464{
465	int ret;
466
467	ret = EINVAL;
468
469	switch (cmd) {
470	case SIOCGETSGCNT_IN6:
471		ret = get_sg_cnt((struct sioc_sg_req6 *)data);
472		break;
473
474	case SIOCGETMIFCNT_IN6:
475		ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
476		break;
477
478	default:
479		break;
480	}
481
482	return (ret);
483}
484
485/*
486 * returns the packet, byte, rpf-failure count for the source group provided
487 */
488static int
489get_sg_cnt(struct sioc_sg_req6 *req)
490{
491	struct mf6c *rt;
492	int ret;
493
494	ret = 0;
495
496	MFC6_LOCK();
497
498	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
499	if (rt == NULL) {
500		ret = ESRCH;
501	} else {
502		req->pktcnt = rt->mf6c_pkt_cnt;
503		req->bytecnt = rt->mf6c_byte_cnt;
504		req->wrong_if = rt->mf6c_wrong_if;
505	}
506
507	MFC6_UNLOCK();
508
509	return (ret);
510}
511
512/*
513 * returns the input and output packet and byte counts on the mif provided
514 */
515static int
516get_mif6_cnt(struct sioc_mif_req6 *req)
517{
518	mifi_t mifi;
519	int ret;
520
521	ret = 0;
522	mifi = req->mifi;
523
524	MIF6_LOCK();
525
526	if (mifi >= nummifs) {
527		ret = EINVAL;
528	} else {
529		req->icount = mif6table[mifi].m6_pkt_in;
530		req->ocount = mif6table[mifi].m6_pkt_out;
531		req->ibytes = mif6table[mifi].m6_bytes_in;
532		req->obytes = mif6table[mifi].m6_bytes_out;
533	}
534
535	MIF6_UNLOCK();
536
537	return (ret);
538}
539
540static int
541set_pim6(int *i)
542{
543	if ((*i != 1) && (*i != 0))
544		return (EINVAL);
545
546	V_pim6 = *i;
547
548	return (0);
549}
550
551/*
552 * Enable multicast routing
553 */
554static int
555ip6_mrouter_init(struct socket *so, int v, int cmd)
556{
557
558	MRT6_DLOG(DEBUG_ANY, "so_type = %d, pr_protocol = %d",
559	    so->so_type, so->so_proto->pr_protocol);
560
561	if (so->so_type != SOCK_RAW ||
562	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
563		return (EOPNOTSUPP);
564
565	if (v != 1)
566		return (ENOPROTOOPT);
567
568	MROUTER6_LOCK();
569
570	if (V_ip6_mrouter != NULL) {
571		MROUTER6_UNLOCK();
572		return (EADDRINUSE);
573	}
574
575	V_ip6_mrouter = so;
576	V_ip6_mrouter_ver = cmd;
577
578	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
579	bzero((caddr_t)n6expire, sizeof(n6expire));
580
581	V_pim6 = 0;/* used for stubbing out/in pim stuff */
582
583	callout_init(&expire_upcalls_ch, 0);
584	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
585	    expire_upcalls, NULL);
586
587	MROUTER6_UNLOCK();
588	MRT6_DLOG(DEBUG_ANY, "finished");
589
590	return (0);
591}
592
593/*
594 * Disable IPv6 multicast forwarding.
595 */
596int
597X_ip6_mrouter_done(void)
598{
599	mifi_t mifi;
600	u_long i;
601	struct mf6c *rt;
602	struct rtdetq *rte;
603
604	MROUTER6_LOCK();
605
606	if (V_ip6_mrouter == NULL) {
607		MROUTER6_UNLOCK();
608		return (EINVAL);
609	}
610
611	/*
612	 * For each phyint in use, disable promiscuous reception of all IPv6
613	 * multicasts.
614	 */
615	for (mifi = 0; mifi < nummifs; mifi++) {
616		if (mif6table[mifi].m6_ifp &&
617		    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
618			if_allmulti(mif6table[mifi].m6_ifp, 0);
619		}
620	}
621	bzero((caddr_t)mif6table, sizeof(mif6table));
622	nummifs = 0;
623
624	V_pim6 = 0; /* used to stub out/in pim specific code */
625
626	callout_stop(&expire_upcalls_ch);
627
628	/*
629	 * Free all multicast forwarding cache entries.
630	 */
631	MFC6_LOCK();
632	for (i = 0; i < MF6CTBLSIZ; i++) {
633		rt = mf6ctable[i];
634		while (rt) {
635			struct mf6c *frt;
636
637			for (rte = rt->mf6c_stall; rte != NULL; ) {
638				struct rtdetq *n = rte->next;
639
640				m_freem(rte->m);
641				free(rte, M_MRTABLE6);
642				rte = n;
643			}
644			frt = rt;
645			rt = rt->mf6c_next;
646			free(frt, M_MRTABLE6);
647		}
648	}
649	bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
650	MFC6_UNLOCK();
651
652	/*
653	 * Reset register interface
654	 */
655	if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
656		if_detach(multicast_register_if6);
657		if_free(multicast_register_if6);
658		reg_mif_num = (mifi_t)-1;
659		multicast_register_if6 = NULL;
660	}
661
662	V_ip6_mrouter = NULL;
663	V_ip6_mrouter_ver = 0;
664
665	MROUTER6_UNLOCK();
666	MRT6_DLOG(DEBUG_ANY, "finished");
667
668	return (0);
669}
670
671static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
672
673/*
674 * Add a mif to the mif table
675 */
676static int
677add_m6if(struct mif6ctl *mifcp)
678{
679	struct mif6 *mifp;
680	struct ifnet *ifp;
681	int error;
682
683	MIF6_LOCK();
684
685	if (mifcp->mif6c_mifi >= MAXMIFS) {
686		MIF6_UNLOCK();
687		return (EINVAL);
688	}
689	mifp = mif6table + mifcp->mif6c_mifi;
690	if (mifp->m6_ifp != NULL) {
691		MIF6_UNLOCK();
692		return (EADDRINUSE); /* XXX: is it appropriate? */
693	}
694	if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
695		MIF6_UNLOCK();
696		return (ENXIO);
697	}
698
699	ifp = ifnet_byindex(mifcp->mif6c_pifi);
700
701	if (mifcp->mif6c_flags & MIFF_REGISTER) {
702		if (reg_mif_num == (mifi_t)-1) {
703			ifp = if_alloc(IFT_OTHER);
704
705			if_initname(ifp, "register_mif", 0);
706			ifp->if_flags |= IFF_LOOPBACK;
707			if_attach(ifp);
708			multicast_register_if6 = ifp;
709			reg_mif_num = mifcp->mif6c_mifi;
710			/*
711			 * it is impossible to guess the ifindex of the
712			 * register interface.  So mif6c_pifi is automatically
713			 * calculated.
714			 */
715			mifcp->mif6c_pifi = ifp->if_index;
716		} else {
717			ifp = multicast_register_if6;
718		}
719	} else {
720		/* Make sure the interface supports multicast */
721		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
722			MIF6_UNLOCK();
723			return (EOPNOTSUPP);
724		}
725
726		error = if_allmulti(ifp, 1);
727		if (error) {
728			MIF6_UNLOCK();
729			return (error);
730		}
731	}
732
733	mifp->m6_flags     = mifcp->mif6c_flags;
734	mifp->m6_ifp       = ifp;
735
736	/* initialize per mif pkt counters */
737	mifp->m6_pkt_in    = 0;
738	mifp->m6_pkt_out   = 0;
739	mifp->m6_bytes_in  = 0;
740	mifp->m6_bytes_out = 0;
741
742	/* Adjust nummifs up if the mifi is higher than nummifs */
743	if (nummifs <= mifcp->mif6c_mifi)
744		nummifs = mifcp->mif6c_mifi + 1;
745
746	MIF6_UNLOCK();
747	MRT6_DLOG(DEBUG_ANY, "mif #%d, phyint %s", mifcp->mif6c_mifi,
748	    if_name(ifp));
749
750	return (0);
751}
752
753/*
754 * Delete a mif from the mif table
755 */
756static int
757del_m6if_locked(mifi_t *mifip)
758{
759	struct mif6 *mifp = mif6table + *mifip;
760	mifi_t mifi;
761	struct ifnet *ifp;
762
763	MIF6_LOCK_ASSERT();
764
765	if (*mifip >= nummifs)
766		return (EINVAL);
767	if (mifp->m6_ifp == NULL)
768		return (EINVAL);
769
770	if (!(mifp->m6_flags & MIFF_REGISTER)) {
771		/* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
772		ifp = mifp->m6_ifp;
773		if_allmulti(ifp, 0);
774	} else {
775		if (reg_mif_num != (mifi_t)-1 &&
776		    multicast_register_if6 != NULL) {
777			if_detach(multicast_register_if6);
778			if_free(multicast_register_if6);
779			reg_mif_num = (mifi_t)-1;
780			multicast_register_if6 = NULL;
781		}
782	}
783
784	bzero((caddr_t)mifp, sizeof(*mifp));
785
786	/* Adjust nummifs down */
787	for (mifi = nummifs; mifi > 0; mifi--)
788		if (mif6table[mifi - 1].m6_ifp)
789			break;
790	nummifs = mifi;
791	MRT6_DLOG(DEBUG_ANY, "mif %d, nummifs %d", *mifip, nummifs);
792
793	return (0);
794}
795
796static int
797del_m6if(mifi_t *mifip)
798{
799	int cc;
800
801	MIF6_LOCK();
802	cc = del_m6if_locked(mifip);
803	MIF6_UNLOCK();
804
805	return (cc);
806}
807
808/*
809 * Add an mfc entry
810 */
811static int
812add_m6fc(struct mf6cctl *mfccp)
813{
814	struct mf6c *rt;
815	u_long hash;
816	struct rtdetq *rte;
817	u_short nstl;
818	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
819
820	MFC6_LOCK();
821
822	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
823		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
824
825	/* If an entry already exists, just update the fields */
826	if (rt) {
827		MRT6_DLOG(DEBUG_MFC, "no upcall o %s g %s p %x",
828		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
829		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
830		    mfccp->mf6cc_parent);
831
832		rt->mf6c_parent = mfccp->mf6cc_parent;
833		rt->mf6c_ifset = mfccp->mf6cc_ifset;
834
835		MFC6_UNLOCK();
836		return (0);
837	}
838
839	/*
840	 * Find the entry for which the upcall was made and update
841	 */
842	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
843			mfccp->mf6cc_mcastgrp.sin6_addr);
844	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
845		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
846				       &mfccp->mf6cc_origin.sin6_addr) &&
847		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
848				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
849		    (rt->mf6c_stall != NULL)) {
850
851			if (nstl++)
852				log(LOG_ERR,
853				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
854				    "multiple kernel entries",
855				    ip6_sprintf(ip6bufo,
856					    &mfccp->mf6cc_origin.sin6_addr),
857				    ip6_sprintf(ip6bufg,
858					    &mfccp->mf6cc_mcastgrp.sin6_addr),
859				    mfccp->mf6cc_parent, rt->mf6c_stall);
860
861			MRT6_DLOG(DEBUG_MFC, "o %s g %s p %x dbg %p",
862			    ip6_sprintf(ip6bufo,
863			    &mfccp->mf6cc_origin.sin6_addr),
864			    ip6_sprintf(ip6bufg,
865				&mfccp->mf6cc_mcastgrp.sin6_addr),
866			    mfccp->mf6cc_parent, rt->mf6c_stall);
867
868			rt->mf6c_origin     = mfccp->mf6cc_origin;
869			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
870			rt->mf6c_parent     = mfccp->mf6cc_parent;
871			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
872			/* initialize pkt counters per src-grp */
873			rt->mf6c_pkt_cnt    = 0;
874			rt->mf6c_byte_cnt   = 0;
875			rt->mf6c_wrong_if   = 0;
876
877			rt->mf6c_expire = 0;	/* Don't clean this guy up */
878			n6expire[hash]--;
879
880			/* free packets Qed at the end of this entry */
881			for (rte = rt->mf6c_stall; rte != NULL; ) {
882				struct rtdetq *n = rte->next;
883				ip6_mdq(rte->m, rte->ifp, rt);
884				m_freem(rte->m);
885#ifdef UPCALL_TIMING
886				collate(&(rte->t));
887#endif /* UPCALL_TIMING */
888				free(rte, M_MRTABLE6);
889				rte = n;
890			}
891			rt->mf6c_stall = NULL;
892		}
893	}
894
895	/*
896	 * It is possible that an entry is being inserted without an upcall
897	 */
898	if (nstl == 0) {
899		MRT6_DLOG(DEBUG_MFC, "no upcall h %lu o %s g %s p %x", hash,
900		    ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
901		    ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
902		    mfccp->mf6cc_parent);
903
904		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
905
906			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
907					       &mfccp->mf6cc_origin.sin6_addr)&&
908			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
909					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
910
911				rt->mf6c_origin     = mfccp->mf6cc_origin;
912				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
913				rt->mf6c_parent     = mfccp->mf6cc_parent;
914				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
915				/* initialize pkt counters per src-grp */
916				rt->mf6c_pkt_cnt    = 0;
917				rt->mf6c_byte_cnt   = 0;
918				rt->mf6c_wrong_if   = 0;
919
920				if (rt->mf6c_expire)
921					n6expire[hash]--;
922				rt->mf6c_expire	   = 0;
923			}
924		}
925		if (rt == NULL) {
926			/* no upcall, so make a new entry */
927			rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
928						  M_NOWAIT);
929			if (rt == NULL) {
930				MFC6_UNLOCK();
931				return (ENOBUFS);
932			}
933
934			/* insert new entry at head of hash chain */
935			rt->mf6c_origin     = mfccp->mf6cc_origin;
936			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
937			rt->mf6c_parent     = mfccp->mf6cc_parent;
938			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
939			/* initialize pkt counters per src-grp */
940			rt->mf6c_pkt_cnt    = 0;
941			rt->mf6c_byte_cnt   = 0;
942			rt->mf6c_wrong_if   = 0;
943			rt->mf6c_expire     = 0;
944			rt->mf6c_stall = NULL;
945
946			/* link into table */
947			rt->mf6c_next  = mf6ctable[hash];
948			mf6ctable[hash] = rt;
949		}
950	}
951
952	MFC6_UNLOCK();
953	return (0);
954}
955
956#ifdef UPCALL_TIMING
957/*
958 * collect delay statistics on the upcalls
959 */
960static void
961collate(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 */
985static int
986del_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
1031static int
1032socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1033{
1034
1035	if (s) {
1036		if (sbappendaddr(&s->so_rcv,
1037				 (struct sockaddr *)src,
1038				 mm, (struct mbuf *)0) != 0) {
1039			sorwakeup(s);
1040			return (0);
1041		}
1042	}
1043	m_freem(mm);
1044	return (-1);
1045}
1046
1047/*
1048 * IPv6 multicast forwarding function. This function assumes that the packet
1049 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1050 * pointed to by "ifp", and the packet is to be relayed to other networks
1051 * that have members of the packet's destination IPv6 multicast group.
1052 *
1053 * The packet is returned unscathed to the caller, unless it is
1054 * erroneous, in which case a non-zero return value tells the caller to
1055 * discard it.
1056 *
1057 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1058 * this function is called in the originating context (i.e., not when
1059 * forwarding a packet from other node).  ip6_output(), which is currently the
1060 * only function that calls this function is called in the originating context,
1061 * explicitly ensures this condition.  It is caller's responsibility to ensure
1062 * that if this function is called from somewhere else in the originating
1063 * context in the future.
1064 */
1065int
1066X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1067{
1068	struct rtdetq *rte;
1069	struct mbuf *mb0;
1070	struct mf6c *rt;
1071	struct mif6 *mifp;
1072	struct mbuf *mm;
1073	u_long hash;
1074	mifi_t mifi;
1075	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1076#ifdef UPCALL_TIMING
1077	struct timeval tp;
1078
1079	GET_TIME(tp);
1080#endif /* UPCALL_TIMING */
1081
1082	MRT6_DLOG(DEBUG_FORWARD, "src %s, dst %s, ifindex %d",
1083	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1084	    ip6_sprintf(ip6bufd, &ip6->ip6_dst), ifp->if_index);
1085
1086	/*
1087	 * Don't forward a packet with Hop limit of zero or one,
1088	 * or a packet destined to a local-only group.
1089	 */
1090	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1091	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1092		return (0);
1093	ip6->ip6_hlim--;
1094
1095	/*
1096	 * Source address check: do not forward packets with unspecified
1097	 * source. It was discussed in July 2000, on ipngwg mailing list.
1098	 * This is rather more serious than unicast cases, because some
1099	 * MLD packets can be sent with the unspecified source address
1100	 * (although such packets must normally set 1 to the hop limit field).
1101	 */
1102	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1103		IP6STAT_INC(ip6s_cantforward);
1104		if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
1105			V_ip6_log_time = time_uptime;
1106			log(LOG_DEBUG,
1107			    "cannot forward "
1108			    "from %s to %s nxt %d received on %s\n",
1109			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1110			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1111			    ip6->ip6_nxt,
1112			    if_name(m->m_pkthdr.rcvif));
1113		}
1114		return (0);
1115	}
1116
1117	MFC6_LOCK();
1118
1119	/*
1120	 * Determine forwarding mifs from the forwarding cache table
1121	 */
1122	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1123	MRT6STAT_INC(mrt6s_mfc_lookups);
1124
1125	/* Entry exists, so forward if necessary */
1126	if (rt) {
1127		MFC6_UNLOCK();
1128		return (ip6_mdq(m, ifp, rt));
1129	}
1130
1131	/*
1132	 * If we don't have a route for packet's origin,
1133	 * Make a copy of the packet & send message to routing daemon.
1134	 */
1135	MRT6STAT_INC(mrt6s_no_route);
1136	MRT6_DLOG(DEBUG_FORWARD | DEBUG_MFC, "no rte s %s g %s",
1137	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1138	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1139
1140	/*
1141	 * Allocate mbufs early so that we don't do extra work if we
1142	 * are just going to fail anyway.
1143	 */
1144	rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6, M_NOWAIT);
1145	if (rte == NULL) {
1146		MFC6_UNLOCK();
1147		return (ENOBUFS);
1148	}
1149	mb0 = m_copy(m, 0, M_COPYALL);
1150	/*
1151	 * Pullup packet header if needed before storing it,
1152	 * as other references may modify it in the meantime.
1153	 */
1154	if (mb0 && (!M_WRITABLE(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1155		mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1156	if (mb0 == NULL) {
1157		free(rte, M_MRTABLE6);
1158		MFC6_UNLOCK();
1159		return (ENOBUFS);
1160	}
1161
1162	/* is there an upcall waiting for this packet? */
1163	hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1164	for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1165		if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1166		    &rt->mf6c_origin.sin6_addr) &&
1167		    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1168		    &rt->mf6c_mcastgrp.sin6_addr) && (rt->mf6c_stall != NULL))
1169			break;
1170	}
1171
1172	if (rt == NULL) {
1173		struct mrt6msg *im;
1174#ifdef MRT6_OINIT
1175		struct omrt6msg *oim;
1176#endif
1177		/* no upcall, so make a new entry */
1178		rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6, M_NOWAIT);
1179		if (rt == NULL) {
1180			free(rte, M_MRTABLE6);
1181			m_freem(mb0);
1182			MFC6_UNLOCK();
1183			return (ENOBUFS);
1184		}
1185		/*
1186		 * Make a copy of the header to send to the user
1187		 * level process
1188		 */
1189		mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1190		if (mm == NULL) {
1191			free(rte, M_MRTABLE6);
1192			m_freem(mb0);
1193			free(rt, M_MRTABLE6);
1194			MFC6_UNLOCK();
1195			return (ENOBUFS);
1196		}
1197
1198		/*
1199		 * Send message to routing daemon
1200		 */
1201		sin6.sin6_addr = ip6->ip6_src;
1202		im = NULL;
1203#ifdef MRT6_OINIT
1204		oim = NULL;
1205#endif
1206		switch (V_ip6_mrouter_ver) {
1207#ifdef MRT6_OINIT
1208		case MRT6_OINIT:
1209			oim = mtod(mm, struct omrt6msg *);
1210			oim->im6_msgtype = MRT6MSG_NOCACHE;
1211			oim->im6_mbz = 0;
1212			break;
1213#endif
1214		case MRT6_INIT:
1215			im = mtod(mm, struct mrt6msg *);
1216			im->im6_msgtype = MRT6MSG_NOCACHE;
1217			im->im6_mbz = 0;
1218			break;
1219		default:
1220			free(rte, M_MRTABLE6);
1221			m_freem(mb0);
1222			free(rt, M_MRTABLE6);
1223			MFC6_UNLOCK();
1224			return (EINVAL);
1225		}
1226
1227		MRT6_DLOG(DEBUG_FORWARD, "getting the iif info in the kernel");
1228		for (mifp = mif6table, mifi = 0;
1229		    mifi < nummifs && mifp->m6_ifp != ifp; mifp++, mifi++)
1230				;
1231
1232		switch (V_ip6_mrouter_ver) {
1233#ifdef MRT6_OINIT
1234		case MRT6_OINIT:
1235			oim->im6_mif = mifi;
1236			break;
1237#endif
1238		case MRT6_INIT:
1239			im->im6_mif = mifi;
1240			break;
1241		}
1242
1243		if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1244			log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1245			    "socket queue full\n");
1246			MRT6STAT_INC(mrt6s_upq_sockfull);
1247			free(rte, M_MRTABLE6);
1248			m_freem(mb0);
1249			free(rt, M_MRTABLE6);
1250			MFC6_UNLOCK();
1251			return (ENOBUFS);
1252		}
1253
1254		MRT6STAT_INC(mrt6s_upcalls);
1255
1256		/* insert new entry at head of hash chain */
1257		bzero(rt, sizeof(*rt));
1258		rt->mf6c_origin.sin6_family = AF_INET6;
1259		rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1260		rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1261		rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1262		rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1263		rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1264		rt->mf6c_expire = UPCALL_EXPIRE;
1265		n6expire[hash]++;
1266		rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1267
1268		/* link into table */
1269		rt->mf6c_next  = mf6ctable[hash];
1270		mf6ctable[hash] = rt;
1271		/* Add this entry to the end of the queue */
1272		rt->mf6c_stall = rte;
1273	} else {
1274		/* determine if q has overflowed */
1275		struct rtdetq **p;
1276		int npkts = 0;
1277
1278		for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1279			if (++npkts > MAX_UPQ6) {
1280				MRT6STAT_INC(mrt6s_upq_ovflw);
1281				free(rte, M_MRTABLE6);
1282				m_freem(mb0);
1283				MFC6_UNLOCK();
1284				return (0);
1285			}
1286
1287		/* Add this entry to the end of the queue */
1288		*p = rte;
1289	}
1290
1291	rte->next = NULL;
1292	rte->m = mb0;
1293	rte->ifp = ifp;
1294#ifdef UPCALL_TIMING
1295	rte->t = tp;
1296#endif /* UPCALL_TIMING */
1297
1298	MFC6_UNLOCK();
1299
1300	return (0);
1301}
1302
1303/*
1304 * Clean up cache entries if upcalls are not serviced
1305 * Call from the Slow Timeout mechanism, every half second.
1306 */
1307static void
1308expire_upcalls(void *unused)
1309{
1310#ifdef MRT6DEBUG
1311	char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1312#endif
1313	struct rtdetq *rte;
1314	struct mf6c *mfc, **nptr;
1315	u_long i;
1316
1317	MFC6_LOCK();
1318	for (i = 0; i < MF6CTBLSIZ; i++) {
1319		if (n6expire[i] == 0)
1320			continue;
1321		nptr = &mf6ctable[i];
1322		while ((mfc = *nptr) != NULL) {
1323			rte = mfc->mf6c_stall;
1324			/*
1325			 * Skip real cache entries
1326			 * Make sure it wasn't marked to not expire (shouldn't happen)
1327			 * If it expires now
1328			 */
1329			if (rte != NULL &&
1330			    mfc->mf6c_expire != 0 &&
1331			    --mfc->mf6c_expire == 0) {
1332				MRT6_DLOG(DEBUG_EXPIRE, "expiring (%s %s)",
1333				    ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1334				    ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1335				/*
1336				 * drop all the packets
1337				 * free the mbuf with the pkt, if, timing info
1338				 */
1339				do {
1340					struct rtdetq *n = rte->next;
1341					m_freem(rte->m);
1342					free(rte, M_MRTABLE6);
1343					rte = n;
1344				} while (rte != NULL);
1345				MRT6STAT_INC(mrt6s_cache_cleanups);
1346				n6expire[i]--;
1347
1348				*nptr = mfc->mf6c_next;
1349				free(mfc, M_MRTABLE6);
1350			} else {
1351				nptr = &mfc->mf6c_next;
1352			}
1353		}
1354	}
1355	MFC6_UNLOCK();
1356	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1357	    expire_upcalls, NULL);
1358}
1359
1360/*
1361 * Packet forwarding routine once entry in the cache is made
1362 */
1363static int
1364ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1365{
1366	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1367	mifi_t mifi, iif;
1368	struct mif6 *mifp;
1369	int plen = m->m_pkthdr.len;
1370	struct in6_addr src0, dst0; /* copies for local work */
1371	u_int32_t iszone, idzone, oszone, odzone;
1372	int error = 0;
1373
1374/*
1375 * Macro to send packet on mif.  Since RSVP packets don't get counted on
1376 * input, they shouldn't get counted on output, so statistics keeping is
1377 * separate.
1378 */
1379
1380#define MC6_SEND(ip6, mifp, m) do {				\
1381	if ((mifp)->m6_flags & MIFF_REGISTER)			\
1382		register_send((ip6), (mifp), (m));		\
1383	else							\
1384		phyint_send((ip6), (mifp), (m));		\
1385} while (/*CONSTCOND*/ 0)
1386
1387	/*
1388	 * Don't forward if it didn't arrive from the parent mif
1389	 * for its origin.
1390	 */
1391	mifi = rt->mf6c_parent;
1392	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1393		/* came in the wrong interface */
1394		MRT6_DLOG(DEBUG_FORWARD,
1395		    "wrong if: ifid %d mifi %d mififid %x", ifp->if_index,
1396		    mifi, mif6table[mifi].m6_ifp->if_index);
1397		MRT6STAT_INC(mrt6s_wrong_if);
1398		rt->mf6c_wrong_if++;
1399		/*
1400		 * If we are doing PIM processing, and we are forwarding
1401		 * packets on this interface, send a message to the
1402		 * routing daemon.
1403		 */
1404		/* have to make sure this is a valid mif */
1405		if (mifi < nummifs && mif6table[mifi].m6_ifp)
1406			if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1407				/*
1408				 * Check the M_LOOP flag to avoid an
1409				 * unnecessary PIM assert.
1410				 * XXX: M_LOOP is an ad-hoc hack...
1411				 */
1412				static struct sockaddr_in6 sin6 =
1413				{ sizeof(sin6), AF_INET6 };
1414
1415				struct mbuf *mm;
1416				struct mrt6msg *im;
1417#ifdef MRT6_OINIT
1418				struct omrt6msg *oim;
1419#endif
1420
1421				mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1422				if (mm &&
1423				    (!M_WRITABLE(mm) ||
1424				     mm->m_len < sizeof(struct ip6_hdr)))
1425					mm = m_pullup(mm, sizeof(struct ip6_hdr));
1426				if (mm == NULL)
1427					return (ENOBUFS);
1428
1429#ifdef MRT6_OINIT
1430				oim = NULL;
1431#endif
1432				im = NULL;
1433				switch (V_ip6_mrouter_ver) {
1434#ifdef MRT6_OINIT
1435				case MRT6_OINIT:
1436					oim = mtod(mm, struct omrt6msg *);
1437					oim->im6_msgtype = MRT6MSG_WRONGMIF;
1438					oim->im6_mbz = 0;
1439					break;
1440#endif
1441				case MRT6_INIT:
1442					im = mtod(mm, struct mrt6msg *);
1443					im->im6_msgtype = MRT6MSG_WRONGMIF;
1444					im->im6_mbz = 0;
1445					break;
1446				default:
1447					m_freem(mm);
1448					return (EINVAL);
1449				}
1450
1451				for (mifp = mif6table, iif = 0;
1452				     iif < nummifs && mifp &&
1453					     mifp->m6_ifp != ifp;
1454				     mifp++, iif++)
1455					;
1456
1457				switch (V_ip6_mrouter_ver) {
1458#ifdef MRT6_OINIT
1459				case MRT6_OINIT:
1460					oim->im6_mif = iif;
1461					sin6.sin6_addr = oim->im6_src;
1462					break;
1463#endif
1464				case MRT6_INIT:
1465					im->im6_mif = iif;
1466					sin6.sin6_addr = im->im6_src;
1467					break;
1468				}
1469
1470				MRT6STAT_INC(mrt6s_upcalls);
1471
1472				if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1473					MRT6_DLOG(DEBUG_ANY,
1474					    "ip6_mrouter socket queue full");
1475					MRT6STAT_INC(mrt6s_upq_sockfull);
1476					return (ENOBUFS);
1477				}	/* if socket Q full */
1478			}		/* if PIM */
1479		return (0);
1480	}			/* if wrong iif */
1481
1482	/* If I sourced this packet, it counts as output, else it was input. */
1483	if (m->m_pkthdr.rcvif == NULL) {
1484		/* XXX: is rcvif really NULL when output?? */
1485		mif6table[mifi].m6_pkt_out++;
1486		mif6table[mifi].m6_bytes_out += plen;
1487	} else {
1488		mif6table[mifi].m6_pkt_in++;
1489		mif6table[mifi].m6_bytes_in += plen;
1490	}
1491	rt->mf6c_pkt_cnt++;
1492	rt->mf6c_byte_cnt += plen;
1493
1494	/*
1495	 * For each mif, forward a copy of the packet if there are group
1496	 * members downstream on the interface.
1497	 */
1498	src0 = ip6->ip6_src;
1499	dst0 = ip6->ip6_dst;
1500	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1501	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1502		IP6STAT_INC(ip6s_badscope);
1503		return (error);
1504	}
1505	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1506		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1507			/*
1508			 * check if the outgoing packet is going to break
1509			 * a scope boundary.
1510			 * XXX For packets through PIM register tunnel
1511			 * interface, we believe a routing daemon.
1512			 */
1513			if (!(mif6table[rt->mf6c_parent].m6_flags &
1514			      MIFF_REGISTER) &&
1515			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1516				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1517				    &oszone) ||
1518				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1519				    &odzone) ||
1520				    iszone != oszone ||
1521				    idzone != odzone) {
1522					IP6STAT_INC(ip6s_badscope);
1523					continue;
1524				}
1525			}
1526
1527			mifp->m6_pkt_out++;
1528			mifp->m6_bytes_out += plen;
1529			MC6_SEND(ip6, mifp, m);
1530		}
1531	}
1532	return (0);
1533}
1534
1535static void
1536phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1537{
1538#ifdef MRT6DEBUG
1539	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1540#endif
1541	struct mbuf *mb_copy;
1542	struct ifnet *ifp = mifp->m6_ifp;
1543	int error = 0;
1544	u_long linkmtu;
1545
1546	/*
1547	 * Make a new reference to the packet; make sure that
1548	 * the IPv6 header is actually copied, not just referenced,
1549	 * so that ip6_output() only scribbles on the copy.
1550	 */
1551	mb_copy = m_copy(m, 0, M_COPYALL);
1552	if (mb_copy &&
1553	    (!M_WRITABLE(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1554		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1555	if (mb_copy == NULL) {
1556		return;
1557	}
1558	/* set MCAST flag to the outgoing packet */
1559	mb_copy->m_flags |= M_MCAST;
1560
1561	/*
1562	 * If we sourced the packet, call ip6_output since we may devide
1563	 * the packet into fragments when the packet is too big for the
1564	 * outgoing interface.
1565	 * Otherwise, we can simply send the packet to the interface
1566	 * sending queue.
1567	 */
1568	if (m->m_pkthdr.rcvif == NULL) {
1569		struct ip6_moptions im6o;
1570
1571		im6o.im6o_multicast_ifp = ifp;
1572		/* XXX: ip6_output will override ip6->ip6_hlim */
1573		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1574		im6o.im6o_multicast_loop = 1;
1575		error = ip6_output(mb_copy, NULL, NULL, IPV6_FORWARDING, &im6o,
1576		    NULL, NULL);
1577
1578		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1579		    (uint16_t)(mifp - mif6table), error);
1580		return;
1581	}
1582
1583	/*
1584	 * If configured to loop back multicasts by default,
1585	 * loop back a copy now.
1586	 */
1587	if (in6_mcast_loop)
1588		ip6_mloopback(ifp, m);
1589
1590	/*
1591	 * Put the packet into the sending queue of the outgoing interface
1592	 * if it would fit in the MTU of the interface.
1593	 */
1594	linkmtu = IN6_LINKMTU(ifp);
1595	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1596		struct sockaddr_in6 dst6;
1597
1598		bzero(&dst6, sizeof(dst6));
1599		dst6.sin6_len = sizeof(struct sockaddr_in6);
1600		dst6.sin6_family = AF_INET6;
1601		dst6.sin6_addr = ip6->ip6_dst;
1602
1603		IP_PROBE(send, NULL, NULL, ip6, ifp, NULL, ip6);
1604		/*
1605		 * We just call if_output instead of nd6_output here, since
1606		 * we need no ND for a multicast forwarded packet...right?
1607		 */
1608		m_clrprotoflags(m);	/* Avoid confusing lower layers. */
1609		error = (*ifp->if_output)(ifp, mb_copy,
1610		    (struct sockaddr *)&dst6, NULL);
1611		MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1612		    (uint16_t)(mifp - mif6table), error);
1613	} else {
1614		/*
1615		 * pMTU discovery is intentionally disabled by default, since
1616		 * various router may notify pMTU in multicast, which can be
1617		 * a DDoS to a router
1618		 */
1619		if (V_ip6_mcast_pmtu)
1620			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1621		else {
1622			MRT6_DLOG(DEBUG_XMIT, " packet too big on %s o %s "
1623			    "g %s size %d (discarded)", if_name(ifp),
1624			    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1625			    ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1626			    mb_copy->m_pkthdr.len);
1627			m_freem(mb_copy); /* simply discard the packet */
1628		}
1629	}
1630}
1631
1632static int
1633register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1634{
1635#ifdef MRT6DEBUG
1636	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1637#endif
1638	struct mbuf *mm;
1639	int i, len = m->m_pkthdr.len;
1640	static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1641	struct mrt6msg *im6;
1642
1643	MRT6_DLOG(DEBUG_ANY, "src %s dst %s",
1644	    ip6_sprintf(ip6bufs, &ip6->ip6_src),
1645	    ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1646	PIM6STAT_INC(pim6s_snd_registers);
1647
1648	/* Make a copy of the packet to send to the user level process. */
1649	mm = m_gethdr(M_NOWAIT, MT_DATA);
1650	if (mm == NULL)
1651		return (ENOBUFS);
1652	mm->m_data += max_linkhdr;
1653	mm->m_len = sizeof(struct ip6_hdr);
1654
1655	if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1656		m_freem(mm);
1657		return (ENOBUFS);
1658	}
1659	i = MHLEN - M_LEADINGSPACE(mm);
1660	if (i > len)
1661		i = len;
1662	mm = m_pullup(mm, i);
1663	if (mm == NULL)
1664		return (ENOBUFS);
1665/* TODO: check it! */
1666	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1667
1668	/*
1669	 * Send message to routing daemon
1670	 */
1671	sin6.sin6_addr = ip6->ip6_src;
1672
1673	im6 = mtod(mm, struct mrt6msg *);
1674	im6->im6_msgtype      = MRT6MSG_WHOLEPKT;
1675	im6->im6_mbz          = 0;
1676
1677	im6->im6_mif = mif - mif6table;
1678
1679	/* iif info is not given for reg. encap.n */
1680	MRT6STAT_INC(mrt6s_upcalls);
1681
1682	if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1683		MRT6_DLOG(DEBUG_ANY, "ip6_mrouter socket queue full");
1684		MRT6STAT_INC(mrt6s_upq_sockfull);
1685		return (ENOBUFS);
1686	}
1687	return (0);
1688}
1689
1690/*
1691 * pim6_encapcheck() is called by the encap6_input() path at runtime to
1692 * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1693 * into the kernel.
1694 */
1695static int
1696pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
1697{
1698
1699#ifdef DIAGNOSTIC
1700    KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1701#endif
1702    if (proto != IPPROTO_PIM)
1703	return 0;	/* not for us; reject the datagram. */
1704
1705    return 64;		/* claim the datagram. */
1706}
1707
1708/*
1709 * PIM sparse mode hook
1710 * Receives the pim control messages, and passes them up to the listening
1711 * socket, using rip6_input.
1712 * The only message processed is the REGISTER pim message; the pim header
1713 * is stripped off, and the inner packet is passed to register_mforward.
1714 */
1715int
1716pim6_input(struct mbuf **mp, int *offp, int proto)
1717{
1718	struct pim *pim; /* pointer to a pim struct */
1719	struct ip6_hdr *ip6;
1720	int pimlen;
1721	struct mbuf *m = *mp;
1722	int minlen;
1723	int off = *offp;
1724
1725	PIM6STAT_INC(pim6s_rcv_total);
1726
1727	ip6 = mtod(m, struct ip6_hdr *);
1728	pimlen = m->m_pkthdr.len - *offp;
1729
1730	/*
1731	 * Validate lengths
1732	 */
1733	if (pimlen < PIM_MINLEN) {
1734		PIM6STAT_INC(pim6s_rcv_tooshort);
1735		MRT6_DLOG(DEBUG_PIM, "PIM packet too short");
1736		m_freem(m);
1737		return (IPPROTO_DONE);
1738	}
1739
1740	/*
1741	 * if the packet is at least as big as a REGISTER, go ahead
1742	 * and grab the PIM REGISTER header size, to avoid another
1743	 * possible m_pullup() later.
1744	 *
1745	 * PIM_MINLEN       == pimhdr + u_int32 == 8
1746	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1747	 */
1748	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1749
1750	/*
1751	 * Make sure that the IP6 and PIM headers in contiguous memory, and
1752	 * possibly the PIM REGISTER header
1753	 */
1754#ifndef PULLDOWN_TEST
1755	IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1756	/* adjust pointer */
1757	ip6 = mtod(m, struct ip6_hdr *);
1758
1759	/* adjust mbuf to point to the PIM header */
1760	pim = (struct pim *)((caddr_t)ip6 + off);
1761#else
1762	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1763	if (pim == NULL) {
1764		PIM6STAT_INC(pim6s_rcv_tooshort);
1765		return (IPPROTO_DONE);
1766	}
1767#endif
1768
1769#define PIM6_CHECKSUM
1770#ifdef PIM6_CHECKSUM
1771	{
1772		int cksumlen;
1773
1774		/*
1775		 * Validate checksum.
1776		 * If PIM REGISTER, exclude the data packet
1777		 */
1778		if (pim->pim_type == PIM_REGISTER)
1779			cksumlen = PIM_MINLEN;
1780		else
1781			cksumlen = pimlen;
1782
1783		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1784			PIM6STAT_INC(pim6s_rcv_badsum);
1785			MRT6_DLOG(DEBUG_PIM, "invalid checksum");
1786			m_freem(m);
1787			return (IPPROTO_DONE);
1788		}
1789	}
1790#endif /* PIM_CHECKSUM */
1791
1792	/* PIM version check */
1793	if (pim->pim_ver != PIM_VERSION) {
1794		PIM6STAT_INC(pim6s_rcv_badversion);
1795		MRT6_DLOG(DEBUG_ANY | DEBUG_ERR,
1796		    "incorrect version %d, expecting %d",
1797		    pim->pim_ver, PIM_VERSION);
1798		m_freem(m);
1799		return (IPPROTO_DONE);
1800	}
1801
1802	if (pim->pim_type == PIM_REGISTER) {
1803		/*
1804		 * since this is a REGISTER, we'll make a copy of the register
1805		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1806		 * routing daemon.
1807		 */
1808		static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1809
1810		struct mbuf *mcp;
1811		struct ip6_hdr *eip6;
1812		u_int32_t *reghdr;
1813		int rc;
1814#ifdef MRT6DEBUG
1815		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1816#endif
1817
1818		PIM6STAT_INC(pim6s_rcv_registers);
1819
1820		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1821			MRT6_DLOG(DEBUG_PIM, "register mif not set: %d",
1822			    reg_mif_num);
1823			m_freem(m);
1824			return (IPPROTO_DONE);
1825		}
1826
1827		reghdr = (u_int32_t *)(pim + 1);
1828
1829		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1830			goto pim6_input_to_daemon;
1831
1832		/*
1833		 * Validate length
1834		 */
1835		if (pimlen < PIM6_REG_MINLEN) {
1836			PIM6STAT_INC(pim6s_rcv_tooshort);
1837			PIM6STAT_INC(pim6s_rcv_badregisters);
1838			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "register packet "
1839			    "size too small %d from %s",
1840			    pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1841			m_freem(m);
1842			return (IPPROTO_DONE);
1843		}
1844
1845		eip6 = (struct ip6_hdr *) (reghdr + 1);
1846		MRT6_DLOG(DEBUG_PIM, "eip6: %s -> %s, eip6 plen %d",
1847		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1848		    ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1849		    ntohs(eip6->ip6_plen));
1850
1851		/* verify the version number of the inner packet */
1852		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1853			PIM6STAT_INC(pim6s_rcv_badregisters);
1854			MRT6_DLOG(DEBUG_ANY, "invalid IP version (%d) "
1855			    "of the inner packet",
1856			    (eip6->ip6_vfc & IPV6_VERSION));
1857			m_freem(m);
1858			return (IPPROTO_DONE);
1859		}
1860
1861		/* verify the inner packet is destined to a mcast group */
1862		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1863			PIM6STAT_INC(pim6s_rcv_badregisters);
1864			MRT6_DLOG(DEBUG_PIM, "inner packet of register "
1865			    "is not multicast %s",
1866			    ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1867			m_freem(m);
1868			return (IPPROTO_DONE);
1869		}
1870
1871		/*
1872		 * make a copy of the whole header to pass to the daemon later.
1873		 */
1874		mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1875		if (mcp == NULL) {
1876			MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "pim register: "
1877			    "could not copy register head");
1878			m_freem(m);
1879			return (IPPROTO_DONE);
1880		}
1881
1882		/*
1883		 * forward the inner ip6 packet; point m_data at the inner ip6.
1884		 */
1885		m_adj(m, off + PIM_MINLEN);
1886		MRT6_DLOG(DEBUG_PIM, "forwarding decapsulated register: "
1887		    "src %s, dst %s, mif %d",
1888		    ip6_sprintf(ip6bufs, &eip6->ip6_src),
1889		    ip6_sprintf(ip6bufd, &eip6->ip6_dst), reg_mif_num);
1890
1891		rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1892				dst.sin6_family, 0);
1893
1894		/* prepare the register head to send to the mrouting daemon */
1895		m = mcp;
1896	}
1897
1898	/*
1899	 * Pass the PIM message up to the daemon; if it is a register message
1900	 * pass the 'head' only up to the daemon. This includes the
1901	 * encapsulator ip6 header, pim header, register header and the
1902	 * encapsulated ip6 header.
1903	 */
1904  pim6_input_to_daemon:
1905	rip6_input(&m, offp, proto);
1906	return (IPPROTO_DONE);
1907}
1908
1909static int
1910ip6_mroute_modevent(module_t mod, int type, void *unused)
1911{
1912
1913	switch (type) {
1914	case MOD_LOAD:
1915		MROUTER6_LOCK_INIT();
1916		MFC6_LOCK_INIT();
1917		MIF6_LOCK_INIT();
1918
1919		pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM,
1920			pim6_encapcheck,
1921			(const struct protosw *)&in6_pim_protosw, NULL);
1922		if (pim6_encap_cookie == NULL) {
1923			printf("ip6_mroute: unable to attach pim6 encap\n");
1924			MIF6_LOCK_DESTROY();
1925			MFC6_LOCK_DESTROY();
1926			MROUTER6_LOCK_DESTROY();
1927			return (EINVAL);
1928		}
1929
1930		ip6_mforward = X_ip6_mforward;
1931		ip6_mrouter_done = X_ip6_mrouter_done;
1932		ip6_mrouter_get = X_ip6_mrouter_get;
1933		ip6_mrouter_set = X_ip6_mrouter_set;
1934		mrt6_ioctl = X_mrt6_ioctl;
1935		break;
1936
1937	case MOD_UNLOAD:
1938		if (V_ip6_mrouter != NULL)
1939			return EINVAL;
1940
1941		if (pim6_encap_cookie) {
1942			encap_detach(pim6_encap_cookie);
1943			pim6_encap_cookie = NULL;
1944		}
1945		X_ip6_mrouter_done();
1946		ip6_mforward = NULL;
1947		ip6_mrouter_done = NULL;
1948		ip6_mrouter_get = NULL;
1949		ip6_mrouter_set = NULL;
1950		mrt6_ioctl = NULL;
1951
1952		MIF6_LOCK_DESTROY();
1953		MFC6_LOCK_DESTROY();
1954		MROUTER6_LOCK_DESTROY();
1955		break;
1956
1957	default:
1958		return (EOPNOTSUPP);
1959	}
1960
1961	return (0);
1962}
1963
1964static moduledata_t ip6_mroutemod = {
1965	"ip6_mroute",
1966	ip6_mroute_modevent,
1967	0
1968};
1969
1970DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PROTO_MC, SI_ORDER_ANY);
1971