Home | History | Annotate | Line # | Download | only in netinet6
ip6_mroute.c revision 1.129
      1 /*	$NetBSD: ip6_mroute.c,v 1.129 2018/06/21 10:37:50 knakahara Exp $	*/
      2 /*	$KAME: ip6_mroute.c,v 1.49 2001/07/25 09:21:18 jinmei Exp $	*/
      3 
      4 /*
      5  * Copyright (C) 1998 WIDE Project.
      6  * All rights reserved.
      7  *
      8  * Redistribution and use in source and binary forms, with or without
      9  * modification, are permitted provided that the following conditions
     10  * are met:
     11  * 1. Redistributions of source code must retain the above copyright
     12  *    notice, this list of conditions and the following disclaimer.
     13  * 2. Redistributions in binary form must reproduce the above copyright
     14  *    notice, this list of conditions and the following disclaimer in the
     15  *    documentation and/or other materials provided with the distribution.
     16  * 3. Neither the name of the project nor the names of its contributors
     17  *    may be used to endorse or promote products derived from this software
     18  *    without specific prior written permission.
     19  *
     20  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
     21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
     24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     30  * SUCH DAMAGE.
     31  */
     32 
     33 /*	BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp	*/
     34 
     35 /*
     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  */
     68 
     69 /*
     70  * Copyright (c) 1989 Stephen Deering
     71  *
     72  * This code is derived from software contributed to Berkeley by
     73  * Stephen Deering of Stanford University.
     74  *
     75  * Redistribution and use in source and binary forms, with or without
     76  * modification, are permitted provided that the following conditions
     77  * are met:
     78  * 1. Redistributions of source code must retain the above copyright
     79  *    notice, this list of conditions and the following disclaimer.
     80  * 2. Redistributions in binary form must reproduce the above copyright
     81  *    notice, this list of conditions and the following disclaimer in the
     82  *    documentation and/or other materials provided with the distribution.
     83  * 3. All advertising materials mentioning features or use of this software
     84  *    must display the following acknowledgement:
     85  *      This product includes software developed by the University of
     86  *      California, Berkeley and its contributors.
     87  * 4. Neither the name of the University nor the names of its contributors
     88  *    may be used to endorse or promote products derived from this software
     89  *    without specific prior written permission.
     90  *
     91  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     92  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     93  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     94  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     95  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     96  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     97  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     98  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     99  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
    100  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
    101  * SUCH DAMAGE.
    102  *
    103  *      @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93
    104  */
    105 
    106 /*
    107  * IP multicast forwarding procedures
    108  *
    109  * Written by David Waitzman, BBN Labs, August 1988.
    110  * Modified by Steve Deering, Stanford, February 1989.
    111  * Modified by Mark J. Steiglitz, Stanford, May, 1991
    112  * Modified by Van Jacobson, LBL, January 1993
    113  * Modified by Ajit Thyagarajan, PARC, August 1993
    114  * Modified by Bill Fenner, PARC, April 1994
    115  *
    116  * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
    117  */
    118 
    119 #include <sys/cdefs.h>
    120 __KERNEL_RCSID(0, "$NetBSD: ip6_mroute.c,v 1.129 2018/06/21 10:37:50 knakahara Exp $");
    121 
    122 #ifdef _KERNEL_OPT
    123 #include "opt_inet.h"
    124 #include "opt_mrouting.h"
    125 #endif
    126 
    127 #include <sys/param.h>
    128 #include <sys/systm.h>
    129 #include <sys/callout.h>
    130 #include <sys/mbuf.h>
    131 #include <sys/socket.h>
    132 #include <sys/socketvar.h>
    133 #include <sys/sockio.h>
    134 #include <sys/errno.h>
    135 #include <sys/time.h>
    136 #include <sys/kernel.h>
    137 #include <sys/ioctl.h>
    138 #include <sys/sysctl.h>
    139 #include <sys/syslog.h>
    140 
    141 #include <net/if.h>
    142 #include <net/route.h>
    143 #include <net/raw_cb.h>
    144 #include <net/net_stats.h>
    145 
    146 #include <netinet/in.h>
    147 #include <netinet/in_var.h>
    148 #include <netinet/icmp6.h>
    149 
    150 #include <netinet/ip6.h>
    151 #include <netinet6/ip6_var.h>
    152 #include <netinet6/ip6_private.h>
    153 #include <netinet6/ip6_mroute.h>
    154 #include <netinet6/scope6_var.h>
    155 #include <netinet6/pim6.h>
    156 #include <netinet6/pim6_var.h>
    157 #include <netinet6/nd6.h>
    158 
    159 static int ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
    160 static void phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
    161 
    162 static int set_pim6(int *);
    163 static int socket_send(struct socket *, struct mbuf *, struct sockaddr_in6 *);
    164 static int register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
    165 
    166 /*
    167  * Globals.  All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static,
    168  * except for netstat or debugging purposes.
    169  */
    170 struct socket  *ip6_mrouter = NULL;
    171 int		ip6_mrouter_ver = 0;
    172 int		ip6_mrtproto = IPPROTO_PIM;    /* for netstat only */
    173 struct mrt6stat	mrt6stat;
    174 
    175 #define NO_RTE_FOUND 	0x1
    176 #define RTE_FOUND	0x2
    177 
    178 struct mf6c	*mf6ctable[MF6CTBLSIZ];
    179 u_char		n6expire[MF6CTBLSIZ];
    180 struct mif6 mif6table[MAXMIFS];
    181 #ifdef MRT6DEBUG
    182 u_int		mrt6debug = 0;	  /* debug level 	*/
    183 #define DEBUG_MFC	0x02
    184 #define DEBUG_FORWARD	0x04
    185 #define DEBUG_EXPIRE	0x08
    186 #define DEBUG_XMIT	0x10
    187 #define DEBUG_REG	0x20
    188 #define DEBUG_PIM	0x40
    189 #define __mrt6debugused     /* empty */
    190 #else
    191 #define __mrt6debugused     __unused
    192 #endif
    193 
    194 static void	expire_upcalls(void *);
    195 #define	EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second */
    196 #define	UPCALL_EXPIRE	6		/* number of timeouts */
    197 
    198 #ifdef INET
    199 #ifdef MROUTING
    200 extern struct socket *ip_mrouter;
    201 #endif
    202 #endif
    203 
    204 /*
    205  * 'Interfaces' associated with decapsulator (so we can tell
    206  * packets that went through it from ones that get reflected
    207  * by a broken gateway).  These interfaces are never linked into
    208  * the system ifnet list & no routes point to them.  I.e., packets
    209  * can't be sent this way.  They only exist as a placeholder for
    210  * multicast source verification.
    211  */
    212 struct ifnet multicast_register_if6;
    213 
    214 #define ENCAP_HOPS 64
    215 
    216 /*
    217  * Private variables.
    218  */
    219 static mifi_t nummifs = 0;
    220 static mifi_t reg_mif_num = (mifi_t)-1;
    221 
    222 static percpu_t *pim6stat_percpu;
    223 
    224 #define	PIM6_STATINC(x)		_NET_STATINC(pim6stat_percpu, x)
    225 
    226 static int pim6;
    227 
    228 /*
    229  * Hash function for a source, group entry
    230  */
    231 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
    232 				   (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
    233 				   (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
    234 				   (g).s6_addr32[2] ^ (g).s6_addr32[3])
    235 
    236 /*
    237  * Find a route for a given origin IPv6 address and Multicast group address.
    238  * Quality of service parameter to be added in the future!!!
    239  */
    240 
    241 #define MF6CFIND(o, g, rt) do { \
    242 	struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
    243 	rt = NULL; \
    244 	mrt6stat.mrt6s_mfc_lookups++; \
    245 	while (_rt) { \
    246 		if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
    247 		    IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
    248 		    (_rt->mf6c_stall == NULL)) { \
    249 			rt = _rt; \
    250 			break; \
    251 		} \
    252 		_rt = _rt->mf6c_next; \
    253 	} \
    254 	if (rt == NULL) { \
    255 		mrt6stat.mrt6s_mfc_misses++; \
    256 	} \
    257 } while (/*CONSTCOND*/ 0)
    258 
    259 /*
    260  * Macros to compute elapsed time efficiently
    261  * Borrowed from Van Jacobson's scheduling code
    262  */
    263 #define TV_DELTA(a, b, delta) do { \
    264 	    int xxs; \
    265 		\
    266 	    delta = (a).tv_usec - (b).tv_usec; \
    267 	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
    268 	       switch (xxs) { \
    269 		      case 2: \
    270 			  delta += 1000000; \
    271 			      /* FALLTHROUGH */ \
    272 		      case 1: \
    273 			  delta += 1000000; \
    274 			  break; \
    275 		      default: \
    276 			  delta += (1000000 * xxs); \
    277 	       } \
    278 	    } \
    279 } while (/*CONSTCOND*/ 0)
    280 
    281 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
    282 	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
    283 
    284 #ifdef UPCALL_TIMING
    285 #define UPCALL_MAX	50
    286 u_long upcall_data[UPCALL_MAX + 1];
    287 static void collate();
    288 #endif /* UPCALL_TIMING */
    289 
    290 static int get_sg_cnt(struct sioc_sg_req6 *);
    291 static int get_mif6_cnt(struct sioc_mif_req6 *);
    292 static int ip6_mrouter_init(struct socket *, int, int);
    293 static int add_m6if(struct mif6ctl *);
    294 static int del_m6if(mifi_t *);
    295 static int add_m6fc(struct mf6cctl *);
    296 static int del_m6fc(struct mf6cctl *);
    297 static void sysctl_net_inet6_pim6_setup(struct sysctllog **);
    298 
    299 static callout_t expire_upcalls_ch;
    300 
    301 void
    302 pim6_init(void)
    303 {
    304 
    305 	sysctl_net_inet6_pim6_setup(NULL);
    306 	pim6stat_percpu = percpu_alloc(sizeof(uint64_t) * PIM6_NSTATS);
    307 }
    308 
    309 /*
    310  * Handle MRT setsockopt commands to modify the multicast routing tables.
    311  */
    312 int
    313 ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
    314 {
    315 	int error, optval;
    316 	struct mif6ctl mifc;
    317 	struct mf6cctl mfcc;
    318 	mifi_t mifi;
    319 
    320 	if (sopt->sopt_name != MRT6_INIT && so != ip6_mrouter)
    321 		return (EACCES);
    322 
    323 	error = 0;
    324 
    325 	switch (sopt->sopt_name) {
    326 #ifdef MRT6_OINIT
    327 	case MRT6_OINIT:
    328 #endif
    329 	case MRT6_INIT:
    330 		error = sockopt_getint(sopt, &optval);
    331 		if (error)
    332 			break;
    333 		return (ip6_mrouter_init(so, optval, sopt->sopt_name));
    334 	case MRT6_DONE:
    335 		return (ip6_mrouter_done());
    336 	case MRT6_ADD_MIF:
    337 		error = sockopt_get(sopt, &mifc, sizeof(mifc));
    338 		if (error)
    339 			break;
    340 		return (add_m6if(&mifc));
    341 	case MRT6_DEL_MIF:
    342 		error = sockopt_get(sopt, &mifi, sizeof(mifi));
    343 		if (error)
    344 			break;
    345 		return (del_m6if(&mifi));
    346 	case MRT6_ADD_MFC:
    347 		error = sockopt_get(sopt, &mfcc, sizeof(mfcc));
    348 		if (error)
    349 			break;
    350 		return (add_m6fc(&mfcc));
    351 	case MRT6_DEL_MFC:
    352 		error = sockopt_get(sopt, &mfcc, sizeof(mfcc));
    353 		if (error)
    354 			break;
    355 		return (del_m6fc(&mfcc));
    356 	case MRT6_PIM:
    357 		error = sockopt_getint(sopt, &optval);
    358 		if (error)
    359 			break;
    360 		return (set_pim6(&optval));
    361 	default:
    362 		error = EOPNOTSUPP;
    363 	}
    364 
    365 	return (error);
    366 }
    367 
    368 /*
    369  * Handle MRT getsockopt commands
    370  */
    371 int
    372 ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
    373 {
    374 	int error;
    375 
    376 	if (so != ip6_mrouter)
    377 		return EACCES;
    378 
    379 	error = 0;
    380 
    381 	switch (sopt->sopt_name) {
    382 	case MRT6_PIM:
    383 		error = sockopt_set(sopt, &pim6, sizeof(pim6));
    384 		break;
    385 	default:
    386 		error = EOPNOTSUPP;
    387 		break;
    388 	}
    389 
    390 	return (error);
    391 }
    392 
    393 /*
    394  * Handle ioctl commands to obtain information from the cache
    395  */
    396 int
    397 mrt6_ioctl(u_long cmd, void *data)
    398 {
    399 
    400 	switch (cmd) {
    401 	case SIOCGETSGCNT_IN6:
    402 		return (get_sg_cnt((struct sioc_sg_req6 *)data));
    403 	case SIOCGETMIFCNT_IN6:
    404 		return (get_mif6_cnt((struct sioc_mif_req6 *)data));
    405 	default:
    406 		return (EINVAL);
    407 	}
    408 }
    409 
    410 /*
    411  * returns the packet, byte, rpf-failure count for the source group provided
    412  */
    413 static int
    414 get_sg_cnt(struct sioc_sg_req6 *req)
    415 {
    416 	struct mf6c *rt;
    417 	int s;
    418 
    419 	s = splsoftnet();
    420 	MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
    421 	splx(s);
    422 	if (rt != NULL) {
    423 		req->pktcnt = rt->mf6c_pkt_cnt;
    424 		req->bytecnt = rt->mf6c_byte_cnt;
    425 		req->wrong_if = rt->mf6c_wrong_if;
    426 	} else
    427 		return (ESRCH);
    428 #if 0
    429 		req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff;
    430 #endif
    431 
    432 	return 0;
    433 }
    434 
    435 /*
    436  * returns the input and output packet and byte counts on the mif provided
    437  */
    438 static int
    439 get_mif6_cnt(struct sioc_mif_req6 *req)
    440 {
    441 	mifi_t mifi = req->mifi;
    442 
    443 	if (mifi >= nummifs)
    444 		return EINVAL;
    445 
    446 	req->icount = mif6table[mifi].m6_pkt_in;
    447 	req->ocount = mif6table[mifi].m6_pkt_out;
    448 	req->ibytes = mif6table[mifi].m6_bytes_in;
    449 	req->obytes = mif6table[mifi].m6_bytes_out;
    450 
    451 	return 0;
    452 }
    453 
    454 static int
    455 set_pim6(int *i)
    456 {
    457 	if ((*i != 1) && (*i != 0))
    458 		return EINVAL;
    459 
    460 	pim6 = *i;
    461 
    462 	return 0;
    463 }
    464 
    465 /*
    466  * Enable multicast routing
    467  */
    468 static int
    469 ip6_mrouter_init(struct socket *so, int v, int cmd)
    470 {
    471 #ifdef MRT6DEBUG
    472 	if (mrt6debug)
    473 		log(LOG_DEBUG,
    474 		    "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
    475 		    so->so_type, so->so_proto->pr_protocol);
    476 #endif
    477 
    478 	if (so->so_type != SOCK_RAW ||
    479 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
    480 		return EOPNOTSUPP;
    481 
    482 	if (v != 1)
    483 		return ENOPROTOOPT;
    484 
    485 	if (ip6_mrouter != NULL)
    486 		return EADDRINUSE;
    487 
    488 	ip6_mrouter = so;
    489 	ip6_mrouter_ver = cmd;
    490 
    491 	memset((void *)mf6ctable, 0, sizeof(mf6ctable));
    492 	memset((void *)n6expire, 0, sizeof(n6expire));
    493 
    494 	pim6 = 0;/* used for stubbing out/in pim stuff */
    495 
    496 	callout_init(&expire_upcalls_ch, CALLOUT_MPSAFE);
    497 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
    498 	    expire_upcalls, NULL);
    499 
    500 #ifdef MRT6DEBUG
    501 	if (mrt6debug)
    502 		log(LOG_DEBUG, "ip6_mrouter_init\n");
    503 #endif
    504 
    505 	return 0;
    506 }
    507 
    508 /*
    509  * Disable multicast routing
    510  */
    511 int
    512 ip6_mrouter_done(void)
    513 {
    514 	mifi_t mifi;
    515 	int i;
    516 	struct ifnet *ifp;
    517 	struct sockaddr_in6 sin6;
    518 	struct mf6c *rt;
    519 	struct rtdetq *rte;
    520 	int s;
    521 
    522 	s = splsoftnet();
    523 
    524 	/*
    525 	 * For each phyint in use, disable promiscuous reception of all IPv6
    526 	 * multicasts.
    527 	 */
    528 #ifdef INET
    529 #ifdef MROUTING
    530 	/*
    531 	 * If there is still IPv4 multicast routing daemon,
    532 	 * we remain interfaces to receive all muliticasted packets.
    533 	 * XXX: there may be an interface in which the IPv4 multicast
    534 	 * daemon is not interested...
    535 	 */
    536 	if (!ip_mrouter)
    537 #endif
    538 #endif
    539 	{
    540 		for (mifi = 0; mifi < nummifs; mifi++) {
    541 			if (mif6table[mifi].m6_ifp &&
    542 			    !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
    543 				sin6.sin6_family = AF_INET6;
    544 				sin6.sin6_addr = in6addr_any;
    545 				ifp = mif6table[mifi].m6_ifp;
    546 				if_mcast_op(ifp, SIOCDELMULTI,
    547 				    sin6tocsa(&sin6));
    548 			}
    549 		}
    550 	}
    551 
    552 	memset((void *)mif6table, 0, sizeof(mif6table));
    553 	nummifs = 0;
    554 
    555 	pim6 = 0; /* used to stub out/in pim specific code */
    556 
    557 	callout_stop(&expire_upcalls_ch);
    558 
    559 	/*
    560 	 * Free all multicast forwarding cache entries.
    561 	 */
    562 	for (i = 0; i < MF6CTBLSIZ; i++) {
    563 		rt = mf6ctable[i];
    564 		while (rt) {
    565 			struct mf6c *frt;
    566 
    567 			for (rte = rt->mf6c_stall; rte != NULL; ) {
    568 				struct rtdetq *n = rte->next;
    569 
    570 				m_freem(rte->m);
    571 				free(rte, M_MRTABLE);
    572 				rte = n;
    573 			}
    574 			frt = rt;
    575 			rt = rt->mf6c_next;
    576 			free(frt, M_MRTABLE);
    577 		}
    578 	}
    579 
    580 	memset((void *)mf6ctable, 0, sizeof(mf6ctable));
    581 
    582 	/*
    583 	 * Reset register interface
    584 	 */
    585 	if (reg_mif_num != (mifi_t)-1) {
    586 		if_detach(&multicast_register_if6);
    587 		reg_mif_num = (mifi_t)-1;
    588 	}
    589 
    590 	ip6_mrouter = NULL;
    591 	ip6_mrouter_ver = 0;
    592 
    593 	splx(s);
    594 
    595 #ifdef MRT6DEBUG
    596 	if (mrt6debug)
    597 		log(LOG_DEBUG, "ip6_mrouter_done\n");
    598 #endif
    599 
    600 	return 0;
    601 }
    602 
    603 void
    604 ip6_mrouter_detach(struct ifnet *ifp)
    605 {
    606 	struct rtdetq *rte;
    607 	struct mf6c *mfc;
    608 	mifi_t mifi;
    609 	int i;
    610 
    611 	if (ip6_mrouter == NULL)
    612 		return;
    613 
    614 	/*
    615 	 * Delete a mif which points to ifp.
    616 	 */
    617 	for (mifi = 0; mifi < nummifs; mifi++)
    618 		if (mif6table[mifi].m6_ifp == ifp)
    619 			del_m6if(&mifi);
    620 
    621 	/*
    622 	 * Clear rte->ifp of cache entries received on ifp.
    623 	 */
    624 	for (i = 0; i < MF6CTBLSIZ; i++) {
    625 		if (n6expire[i] == 0)
    626 			continue;
    627 
    628 		for (mfc = mf6ctable[i]; mfc != NULL; mfc = mfc->mf6c_next) {
    629 			for (rte = mfc->mf6c_stall; rte != NULL; rte = rte->next) {
    630 				if (rte->ifp == ifp)
    631 					rte->ifp = NULL;
    632 			}
    633 		}
    634 	}
    635 }
    636 
    637 /*
    638  * Add a mif to the mif table
    639  */
    640 static int
    641 add_m6if(struct mif6ctl *mifcp)
    642 {
    643 	struct mif6 *mifp;
    644 	struct ifnet *ifp;
    645 	struct sockaddr_in6 sin6;
    646 	int error, s;
    647 
    648 	if (mifcp->mif6c_mifi >= MAXMIFS)
    649 		return EINVAL;
    650 	mifp = mif6table + mifcp->mif6c_mifi;
    651 	if (mifp->m6_ifp)
    652 		return EADDRINUSE; /* XXX: is it appropriate? */
    653 	if (!mifcp->mif6c_pifi || (ifp = if_byindex(mifcp->mif6c_pifi)) == NULL)
    654 		return ENXIO;
    655 
    656 	if (mifcp->mif6c_flags & MIFF_REGISTER) {
    657 		ifp = &multicast_register_if6;
    658 
    659 		if (reg_mif_num == (mifi_t)-1) {
    660 			strlcpy(ifp->if_xname, "register_mif",
    661 			    sizeof(ifp->if_xname));
    662 			ifp->if_flags |= IFF_LOOPBACK;
    663 			ifp->if_index = mifcp->mif6c_mifi;
    664 			reg_mif_num = mifcp->mif6c_mifi;
    665 			if_attach(ifp);
    666 		}
    667 	} else {
    668 		/* Make sure the interface supports multicast */
    669 		if ((ifp->if_flags & IFF_MULTICAST) == 0)
    670 			return EOPNOTSUPP;
    671 
    672 		s = splsoftnet();
    673 		/*
    674 		 * Enable promiscuous reception of all IPv6 multicasts
    675 		 * from the interface.
    676 		 */
    677 		sin6.sin6_family = AF_INET6;
    678 		sin6.sin6_addr = in6addr_any;
    679 		error = if_mcast_op(ifp, SIOCADDMULTI, sin6tosa(&sin6));
    680 		splx(s);
    681 		if (error)
    682 			return error;
    683 	}
    684 
    685 	s = splsoftnet();
    686 	mifp->m6_flags     = mifcp->mif6c_flags;
    687 	mifp->m6_ifp       = ifp;
    688 	/* initialize per mif pkt counters */
    689 	mifp->m6_pkt_in    = 0;
    690 	mifp->m6_pkt_out   = 0;
    691 	mifp->m6_bytes_in  = 0;
    692 	mifp->m6_bytes_out = 0;
    693 	splx(s);
    694 
    695 	/* Adjust nummifs up if the mifi is higher than nummifs */
    696 	if (nummifs <= mifcp->mif6c_mifi)
    697 		nummifs = mifcp->mif6c_mifi + 1;
    698 
    699 #ifdef MRT6DEBUG
    700 	if (mrt6debug)
    701 		log(LOG_DEBUG,
    702 		    "add_mif #%d, phyint %s\n",
    703 		    mifcp->mif6c_mifi, ifp->if_xname);
    704 #endif
    705 
    706 	return 0;
    707 }
    708 
    709 /*
    710  * Delete a mif from the mif table
    711  */
    712 static int
    713 del_m6if(mifi_t *mifip)
    714 {
    715 	struct mif6 *mifp = mif6table + *mifip;
    716 	mifi_t mifi;
    717 	struct ifnet *ifp;
    718 	struct sockaddr_in6 sin6;
    719 	int s;
    720 
    721 	if (*mifip >= nummifs)
    722 		return EINVAL;
    723 	if (mifp->m6_ifp == NULL)
    724 		return EINVAL;
    725 
    726 	s = splsoftnet();
    727 
    728 	if (!(mifp->m6_flags & MIFF_REGISTER)) {
    729 		/*
    730 		 * XXX: what if there is yet IPv4 multicast daemon
    731 		 *      using the interface?
    732 		 */
    733 		ifp = mifp->m6_ifp;
    734 
    735 		sin6.sin6_family = AF_INET6;
    736 		sin6.sin6_addr = in6addr_any;
    737 		if_mcast_op(ifp, SIOCDELMULTI, sin6tosa(&sin6));
    738 	} else {
    739 		if (reg_mif_num != (mifi_t)-1) {
    740 			if_detach(&multicast_register_if6);
    741 			reg_mif_num = (mifi_t)-1;
    742 		}
    743 	}
    744 
    745 	memset((void *)mifp, 0, sizeof (*mifp));
    746 
    747 	/* Adjust nummifs down */
    748 	for (mifi = nummifs; mifi > 0; mifi--)
    749 		if (mif6table[mifi - 1].m6_ifp)
    750 			break;
    751 	nummifs = mifi;
    752 
    753 	splx(s);
    754 
    755 #ifdef MRT6DEBUG
    756 	if (mrt6debug)
    757 		log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
    758 #endif
    759 
    760 	return 0;
    761 }
    762 
    763 /*
    764  * Add an mfc entry
    765  */
    766 static int
    767 add_m6fc(struct mf6cctl *mfccp)
    768 {
    769 	struct mf6c *rt;
    770 	u_long hash;
    771 	struct rtdetq *rte;
    772 	u_short nstl;
    773 	int s;
    774 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufm[INET6_ADDRSTRLEN];
    775 
    776 	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
    777 		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
    778 
    779 	/* If an entry already exists, just update the fields */
    780 	if (rt) {
    781 #ifdef MRT6DEBUG
    782 		if (mrt6debug & DEBUG_MFC)
    783 			log(LOG_DEBUG,"add_m6fc update o %s g %s p %x\n",
    784 			    IN6_PRINT(ip6bufo,
    785 			    &mfccp->mf6cc_origin.sin6_addr),
    786 			    IN6_PRINT(ip6bufm,
    787 			    &mfccp->mf6cc_mcastgrp.sin6_addr),
    788 			    mfccp->mf6cc_parent);
    789 #endif
    790 
    791 		s = splsoftnet();
    792 		rt->mf6c_parent = mfccp->mf6cc_parent;
    793 		rt->mf6c_ifset = mfccp->mf6cc_ifset;
    794 		splx(s);
    795 		return 0;
    796 	}
    797 
    798 	/*
    799 	 * Find the entry for which the upcall was made and update
    800 	 */
    801 	s = splsoftnet();
    802 	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
    803 			mfccp->mf6cc_mcastgrp.sin6_addr);
    804 	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
    805 		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
    806 				       &mfccp->mf6cc_origin.sin6_addr) &&
    807 		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
    808 				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
    809 		    (rt->mf6c_stall != NULL)) {
    810 
    811 			if (nstl++)
    812 				log(LOG_ERR,
    813 				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
    814 				    "multiple kernel entries",
    815 				    IN6_PRINT(ip6bufo,
    816 				    &mfccp->mf6cc_origin.sin6_addr),
    817 				    IN6_PRINT(ip6bufm,
    818 				    &mfccp->mf6cc_mcastgrp.sin6_addr),
    819 				    mfccp->mf6cc_parent, rt->mf6c_stall);
    820 
    821 #ifdef MRT6DEBUG
    822 			if (mrt6debug & DEBUG_MFC)
    823 				log(LOG_DEBUG,
    824 				    "add_m6fc o %s g %s p %x dbg %p\n",
    825 				    IN6_PRINT(ip6bufo,
    826 				    &mfccp->mf6cc_origin.sin6_addr),
    827 				    IN6_PRINT(ip6bufm,
    828 				    &mfccp->mf6cc_mcastgrp.sin6_addr),
    829 				    mfccp->mf6cc_parent, rt->mf6c_stall);
    830 #endif
    831 
    832 			rt->mf6c_origin     = mfccp->mf6cc_origin;
    833 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
    834 			rt->mf6c_parent     = mfccp->mf6cc_parent;
    835 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
    836 			/* initialize pkt counters per src-grp */
    837 			rt->mf6c_pkt_cnt    = 0;
    838 			rt->mf6c_byte_cnt   = 0;
    839 			rt->mf6c_wrong_if   = 0;
    840 
    841 			rt->mf6c_expire = 0;	/* Don't clean this guy up */
    842 			n6expire[hash]--;
    843 
    844 			/* free packets Qed at the end of this entry */
    845 			for (rte = rt->mf6c_stall; rte != NULL; ) {
    846 				struct rtdetq *n = rte->next;
    847 				if (rte->ifp) {
    848 					ip6_mdq(rte->m, rte->ifp, rt);
    849 				}
    850 				m_freem(rte->m);
    851 #ifdef UPCALL_TIMING
    852 				collate(&(rte->t));
    853 #endif
    854 				free(rte, M_MRTABLE);
    855 				rte = n;
    856 			}
    857 			rt->mf6c_stall = NULL;
    858 		}
    859 	}
    860 
    861 	/*
    862 	 * It is possible that an entry is being inserted without an upcall
    863 	 */
    864 	if (nstl == 0) {
    865 #ifdef MRT6DEBUG
    866 		if (mrt6debug & DEBUG_MFC)
    867 			log(LOG_DEBUG,
    868 			    "add_mfc no upcall h %ld o %s g %s p %x\n",
    869 			    hash,
    870 			    IN6_PRINT(ip6bufo,
    871 			    &mfccp->mf6cc_origin.sin6_addr),
    872 			    IN6_PRINT(ip6bufm,
    873 			    &mfccp->mf6cc_mcastgrp.sin6_addr),
    874 			    mfccp->mf6cc_parent);
    875 #endif
    876 
    877 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
    878 
    879 			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
    880 					       &mfccp->mf6cc_origin.sin6_addr)&&
    881 			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
    882 					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
    883 
    884 				rt->mf6c_origin     = mfccp->mf6cc_origin;
    885 				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
    886 				rt->mf6c_parent     = mfccp->mf6cc_parent;
    887 				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
    888 				/* initialize pkt counters per src-grp */
    889 				rt->mf6c_pkt_cnt    = 0;
    890 				rt->mf6c_byte_cnt   = 0;
    891 				rt->mf6c_wrong_if   = 0;
    892 
    893 				if (rt->mf6c_expire)
    894 					n6expire[hash]--;
    895 				rt->mf6c_expire	   = 0;
    896 			}
    897 		}
    898 		if (rt == NULL) {
    899 			/* no upcall, so make a new entry */
    900 			rt = malloc(sizeof(*rt), M_MRTABLE, M_NOWAIT);
    901 			if (rt == NULL) {
    902 				splx(s);
    903 				return ENOBUFS;
    904 			}
    905 
    906 			/* insert new entry at head of hash chain */
    907 			rt->mf6c_origin     = mfccp->mf6cc_origin;
    908 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
    909 			rt->mf6c_parent     = mfccp->mf6cc_parent;
    910 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
    911 			/* initialize pkt counters per src-grp */
    912 			rt->mf6c_pkt_cnt    = 0;
    913 			rt->mf6c_byte_cnt   = 0;
    914 			rt->mf6c_wrong_if   = 0;
    915 			rt->mf6c_expire     = 0;
    916 			rt->mf6c_stall = NULL;
    917 
    918 			/* link into table */
    919 			rt->mf6c_next  = mf6ctable[hash];
    920 			mf6ctable[hash] = rt;
    921 		}
    922 	}
    923 	splx(s);
    924 	return 0;
    925 }
    926 
    927 #ifdef UPCALL_TIMING
    928 /*
    929  * collect delay statistics on the upcalls
    930  */
    931 static void
    932 collate(struct timeval *t)
    933 {
    934 	u_long d;
    935 	struct timeval tp;
    936 	u_long delta;
    937 
    938 	GET_TIME(tp);
    939 
    940 	if (TV_LT(*t, tp))
    941 	{
    942 		TV_DELTA(tp, *t, delta);
    943 
    944 		d = delta >> 10;
    945 		if (d > UPCALL_MAX)
    946 			d = UPCALL_MAX;
    947 
    948 		++upcall_data[d];
    949 	}
    950 }
    951 #endif /* UPCALL_TIMING */
    952 
    953 /*
    954  * Delete an mfc entry
    955  */
    956 static int
    957 del_m6fc(struct mf6cctl *mfccp)
    958 {
    959 	struct sockaddr_in6 	origin;
    960 	struct sockaddr_in6 	mcastgrp;
    961 	struct mf6c 		*rt;
    962 	struct mf6c	 	**nptr;
    963 	u_long 		hash;
    964 	int s;
    965 
    966 	origin = mfccp->mf6cc_origin;
    967 	mcastgrp = mfccp->mf6cc_mcastgrp;
    968 	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
    969 
    970 #ifdef MRT6DEBUG
    971 	if (mrt6debug & DEBUG_MFC) {
    972 		char ip6bufo[INET6_ADDRSTRLEN], ip6bufm[INET6_ADDRSTRLEN];
    973 		log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
    974 		    IN6_PRINT(ip6bufo, &origin.sin6_addr),
    975 		    IN6_PRINT(ip6bufm, &mcastgrp.sin6_addr));
    976 	}
    977 #endif
    978 
    979 	s = splsoftnet();
    980 
    981 	nptr = &mf6ctable[hash];
    982 	while ((rt = *nptr) != NULL) {
    983 		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
    984 				       &rt->mf6c_origin.sin6_addr) &&
    985 		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
    986 				       &rt->mf6c_mcastgrp.sin6_addr) &&
    987 		    rt->mf6c_stall == NULL)
    988 			break;
    989 
    990 		nptr = &rt->mf6c_next;
    991 	}
    992 	if (rt == NULL) {
    993 		splx(s);
    994 		return EADDRNOTAVAIL;
    995 	}
    996 
    997 	*nptr = rt->mf6c_next;
    998 	free(rt, M_MRTABLE);
    999 
   1000 	splx(s);
   1001 
   1002 	return 0;
   1003 }
   1004 
   1005 static int
   1006 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
   1007 {
   1008 	if (s) {
   1009 		if (sbappendaddr(&s->so_rcv, sin6tosa(src), mm, NULL) != 0) {
   1010 			sorwakeup(s);
   1011 			return 0;
   1012 		}
   1013 		soroverflow(s);
   1014 	}
   1015 	m_freem(mm);
   1016 	return -1;
   1017 }
   1018 
   1019 /*
   1020  * IPv6 multicast forwarding function. This function assumes that the packet
   1021  * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
   1022  * pointed to by "ifp", and the packet is to be relayed to other networks
   1023  * that have members of the packet's destination IPv6 multicast group.
   1024  *
   1025  * The packet is returned unscathed to the caller, unless it is
   1026  * erroneous, in which case a non-zero return value tells the caller to
   1027  * discard it.
   1028  */
   1029 int
   1030 ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
   1031 {
   1032 	struct mf6c *rt;
   1033 	struct mif6 *mifp;
   1034 	struct mbuf *mm;
   1035 	int s;
   1036 	mifi_t mifi;
   1037 	struct sockaddr_in6 sin6;
   1038 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
   1039 
   1040 #ifdef MRT6DEBUG
   1041 	if (mrt6debug & DEBUG_FORWARD)
   1042 		log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
   1043 		    IN6_PRINT(ip6bufs, &ip6->ip6_src),
   1044 		    IN6_PRINT(ip6bufd, &ip6->ip6_dst),
   1045 		    ifp->if_index);
   1046 #endif
   1047 
   1048 	/*
   1049 	 * Don't forward a packet with Hop limit of zero or one,
   1050 	 * or a packet destined to a local-only group.
   1051 	 */
   1052 	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_NODELOCAL(&ip6->ip6_dst) ||
   1053 	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
   1054 		return 0;
   1055 	ip6->ip6_hlim--;
   1056 
   1057 	/*
   1058 	 * Source address check: do not forward packets with unspecified
   1059 	 * source. It was discussed in July 2000, on ipngwg mailing list.
   1060 	 * This is rather more serious than unicast cases, because some
   1061 	 * MLD packets can be sent with the unspecified source address
   1062 	 * (although such packets must normally set the hop limit field to 1).
   1063 	 */
   1064 	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
   1065 		IP6_STATINC(IP6_STAT_CANTFORWARD);
   1066 		if (ip6_log_time + ip6_log_interval < time_uptime) {
   1067 			ip6_log_time = time_uptime;
   1068 			log(LOG_DEBUG,
   1069 			    "cannot forward "
   1070 			    "from %s to %s nxt %d received on %s\n",
   1071 			    IN6_PRINT(ip6bufs, &ip6->ip6_src),
   1072 			    IN6_PRINT(ip6bufd, &ip6->ip6_dst),
   1073 			    ip6->ip6_nxt,
   1074 			    m->m_pkthdr.rcvif_index ?
   1075 			    if_name(m_get_rcvif_NOMPSAFE(m)) : "?");
   1076 		}
   1077 		return 0;
   1078 	}
   1079 
   1080 	/*
   1081 	 * Determine forwarding mifs from the forwarding cache table
   1082 	 */
   1083 	s = splsoftnet();
   1084 	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
   1085 
   1086 	/* Entry exists, so forward if necessary */
   1087 	if (rt) {
   1088 		splx(s);
   1089 		return ip6_mdq(m, ifp, rt);
   1090 	} else {
   1091 		/*
   1092 		 * If we don't have a route for packet's origin, make a copy
   1093 		 * of the packet and send message to routing daemon.
   1094 		 */
   1095 
   1096 		struct mbuf *mb0;
   1097 		struct rtdetq *rte;
   1098 		u_long hash;
   1099 
   1100 #ifdef UPCALL_TIMING
   1101 		struct timeval tp;
   1102 		GET_TIME(tp);
   1103 #endif
   1104 
   1105 		mrt6stat.mrt6s_no_route++;
   1106 #ifdef MRT6DEBUG
   1107 		if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
   1108 			log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
   1109 			    IN6_PRINT(ip6bufs, &ip6->ip6_src),
   1110 			    IN6_PRINT(ip6bufd, &ip6->ip6_dst));
   1111 #endif
   1112 
   1113 		/*
   1114 		 * Allocate mbufs early so that we don't do extra work if we
   1115 		 * are just going to fail anyway.
   1116 		 */
   1117 		rte = malloc(sizeof(*rte), M_MRTABLE, M_NOWAIT);
   1118 		if (rte == NULL) {
   1119 			splx(s);
   1120 			return ENOBUFS;
   1121 		}
   1122 		mb0 = m_copypacket(m, M_DONTWAIT);
   1123 
   1124 		/*
   1125 		 * Pullup packet header if needed before storing it,
   1126 		 * as other references may modify it in the meantime.
   1127 		 */
   1128 		if (mb0 && M_UNWRITABLE(mb0, sizeof(struct ip6_hdr)))
   1129 			mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
   1130 		if (mb0 == NULL) {
   1131 			free(rte, M_MRTABLE);
   1132 			splx(s);
   1133 			return ENOBUFS;
   1134 		}
   1135 
   1136 		/* is there an upcall waiting for this packet? */
   1137 		hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
   1138 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
   1139 			if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
   1140 					       &rt->mf6c_origin.sin6_addr) &&
   1141 			    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
   1142 					       &rt->mf6c_mcastgrp.sin6_addr) &&
   1143 			    (rt->mf6c_stall != NULL))
   1144 				break;
   1145 		}
   1146 
   1147 		if (rt == NULL) {
   1148 			struct mrt6msg *im;
   1149 			struct omrt6msg *oim;
   1150 
   1151 			/* no upcall, so make a new entry */
   1152 			rt = malloc(sizeof(*rt), M_MRTABLE, M_NOWAIT);
   1153 			if (rt == NULL) {
   1154 				free(rte, M_MRTABLE);
   1155 				m_freem(mb0);
   1156 				splx(s);
   1157 				return ENOBUFS;
   1158 			}
   1159 
   1160 			/*
   1161 			 * Make a copy of the header to send to the user
   1162 			 * level process
   1163 			 */
   1164 			mm = m_copym(mb0, 0, sizeof(struct ip6_hdr), M_DONTWAIT);
   1165 
   1166 			if (mm == NULL) {
   1167 				free(rte, M_MRTABLE);
   1168 				m_freem(mb0);
   1169 				free(rt, M_MRTABLE);
   1170 				splx(s);
   1171 				return ENOBUFS;
   1172 			}
   1173 
   1174 			/*
   1175 			 * Send message to routing daemon
   1176 			 */
   1177 			sockaddr_in6_init(&sin6, &ip6->ip6_src, 0, 0, 0);
   1178 
   1179 			im = NULL;
   1180 			oim = NULL;
   1181 			switch (ip6_mrouter_ver) {
   1182 			case MRT6_OINIT:
   1183 				oim = mtod(mm, struct omrt6msg *);
   1184 				oim->im6_msgtype = MRT6MSG_NOCACHE;
   1185 				oim->im6_mbz = 0;
   1186 				break;
   1187 			case MRT6_INIT:
   1188 				im = mtod(mm, struct mrt6msg *);
   1189 				im->im6_msgtype = MRT6MSG_NOCACHE;
   1190 				im->im6_mbz = 0;
   1191 				break;
   1192 			default:
   1193 				free(rte, M_MRTABLE);
   1194 				m_freem(mb0);
   1195 				free(rt, M_MRTABLE);
   1196 				splx(s);
   1197 				return EINVAL;
   1198 			}
   1199 
   1200 #ifdef MRT6DEBUG
   1201 			if (mrt6debug & DEBUG_FORWARD)
   1202 				log(LOG_DEBUG,
   1203 				    "getting the iif info in the kernel\n");
   1204 #endif
   1205 
   1206 			for (mifp = mif6table, mifi = 0;
   1207 			     mifi < nummifs && mifp->m6_ifp != ifp;
   1208 			     mifp++, mifi++)
   1209 				;
   1210 
   1211 			switch (ip6_mrouter_ver) {
   1212 			case MRT6_OINIT:
   1213 				oim->im6_mif = mifi;
   1214 				break;
   1215 			case MRT6_INIT:
   1216 				im->im6_mif = mifi;
   1217 				break;
   1218 			}
   1219 
   1220 			if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
   1221 				log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
   1222 				    "socket queue full\n");
   1223 				mrt6stat.mrt6s_upq_sockfull++;
   1224 				free(rte, M_MRTABLE);
   1225 				m_freem(mb0);
   1226 				free(rt, M_MRTABLE);
   1227 				splx(s);
   1228 				return ENOBUFS;
   1229 			}
   1230 
   1231 			mrt6stat.mrt6s_upcalls++;
   1232 
   1233 			/* insert new entry at head of hash chain */
   1234 			memset(rt, 0, sizeof(*rt));
   1235 			sockaddr_in6_init(&rt->mf6c_origin, &ip6->ip6_src,
   1236 			    0, 0, 0);
   1237 			sockaddr_in6_init(&rt->mf6c_mcastgrp, &ip6->ip6_dst,
   1238 			    0, 0, 0);
   1239 			rt->mf6c_expire = UPCALL_EXPIRE;
   1240 			n6expire[hash]++;
   1241 			rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
   1242 
   1243 			/* link into table */
   1244 			rt->mf6c_next  = mf6ctable[hash];
   1245 			mf6ctable[hash] = rt;
   1246 			/* Add this entry to the end of the queue */
   1247 			rt->mf6c_stall = rte;
   1248 		} else {
   1249 			/* determine if q has overflowed */
   1250 			struct rtdetq **p;
   1251 			int npkts = 0;
   1252 
   1253 			for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next) {
   1254 				if (++npkts > MAX_UPQ6) {
   1255 					mrt6stat.mrt6s_upq_ovflw++;
   1256 					free(rte, M_MRTABLE);
   1257 					m_freem(mb0);
   1258 					splx(s);
   1259 					return 0;
   1260 				}
   1261 			}
   1262 
   1263 			/* Add this entry to the end of the queue */
   1264 			*p = rte;
   1265 		}
   1266 
   1267 		rte->next = NULL;
   1268 		rte->m = mb0;
   1269 		rte->ifp = ifp;
   1270 #ifdef UPCALL_TIMING
   1271 		rte->t = tp;
   1272 #endif
   1273 
   1274 		splx(s);
   1275 
   1276 		return 0;
   1277 	}
   1278 }
   1279 
   1280 /*
   1281  * Clean up cache entries if upcalls are not serviced
   1282  * Call from the Slow Timeout mechanism, every 0.25 seconds.
   1283  */
   1284 static void
   1285 expire_upcalls(void *unused)
   1286 {
   1287 	struct rtdetq *rte;
   1288 	struct mf6c *mfc, **nptr;
   1289 	int i;
   1290 
   1291 	/* XXX NOMPSAFE still need softnet_lock */
   1292 	mutex_enter(softnet_lock);
   1293 	KERNEL_LOCK(1, NULL);
   1294 
   1295 	for (i = 0; i < MF6CTBLSIZ; i++) {
   1296 		if (n6expire[i] == 0)
   1297 			continue;
   1298 		nptr = &mf6ctable[i];
   1299 		while ((mfc = *nptr) != NULL) {
   1300 			rte = mfc->mf6c_stall;
   1301 			/*
   1302 			 * Skip real cache entries
   1303 			 * Make sure it wasn't marked to not expire (shouldn't happen)
   1304 			 * If it expires now
   1305 			 */
   1306 			if (rte != NULL &&
   1307 			    mfc->mf6c_expire != 0 &&
   1308 			    --mfc->mf6c_expire == 0) {
   1309 #ifdef MRT6DEBUG
   1310 				if (mrt6debug & DEBUG_EXPIRE) {
   1311 					char ip6bufo[INET6_ADDRSTRLEN];
   1312 					char ip6bufm[INET6_ADDRSTRLEN];
   1313 					log(LOG_DEBUG,
   1314 					    "expire_upcalls: expiring (%s %s)\n",
   1315 					    IN6_PRINT(ip6bufo,
   1316 					    &mfc->mf6c_origin.sin6_addr),
   1317 					    IN6_PRINT(ip6bufm,
   1318 					    &mfc->mf6c_mcastgrp.sin6_addr));
   1319 				}
   1320 #endif
   1321 				/*
   1322 				 * drop all the packets
   1323 				 * free the mbuf with the pkt, if, timing info
   1324 				 */
   1325 				do {
   1326 					struct rtdetq *n = rte->next;
   1327 					m_freem(rte->m);
   1328 					free(rte, M_MRTABLE);
   1329 					rte = n;
   1330 				} while (rte != NULL);
   1331 				mrt6stat.mrt6s_cache_cleanups++;
   1332 				n6expire[i]--;
   1333 
   1334 				*nptr = mfc->mf6c_next;
   1335 				free(mfc, M_MRTABLE);
   1336 			} else {
   1337 				nptr = &mfc->mf6c_next;
   1338 			}
   1339 		}
   1340 	}
   1341 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
   1342 	    expire_upcalls, NULL);
   1343 
   1344 	KERNEL_UNLOCK_ONE(NULL);
   1345 	mutex_exit(softnet_lock);
   1346 }
   1347 
   1348 /*
   1349  * Macro to send packet on mif.  Since RSVP packets don't get counted on
   1350  * input, they shouldn't get counted on output, so statistics keeping is
   1351  * separate.
   1352  */
   1353 #define MC6_SEND(ip6, mifp, m) do {				\
   1354 	if ((mifp)->m6_flags & MIFF_REGISTER)			\
   1355 		register_send((ip6), (mifp), (m));		\
   1356 	else							\
   1357 		phyint_send((ip6), (mifp), (m));		\
   1358 } while (/*CONSTCOND*/ 0)
   1359 
   1360 /*
   1361  * Packet forwarding routine once entry in the cache is made
   1362  */
   1363 static int
   1364 ip6_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 	 * Don't forward if it didn't arrive from the parent mif
   1376 	 * for its origin.
   1377 	 */
   1378 	mifi = rt->mf6c_parent;
   1379 	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
   1380 		/* came in the wrong interface */
   1381 #ifdef MRT6DEBUG
   1382 		if (mrt6debug & DEBUG_FORWARD)
   1383 			log(LOG_DEBUG,
   1384 			    "wrong if: ifid %d mifi %d mififid %x\n",
   1385 			    ifp->if_index, mifi,
   1386 			    mif6table[mifi].m6_ifp ?
   1387 			    mif6table[mifi].m6_ifp->if_index : -1);
   1388 #endif
   1389 		mrt6stat.mrt6s_wrong_if++;
   1390 		rt->mf6c_wrong_if++;
   1391 
   1392 		/*
   1393 		 * If we are doing PIM processing, and we are forwarding
   1394 		 * packets on this interface, send a message to the
   1395 		 * routing daemon.
   1396 		 */
   1397 		/* have to make sure this is a valid mif */
   1398 		if (mifi < nummifs && mif6table[mifi].m6_ifp) {
   1399 			if (pim6 && (m->m_flags & M_LOOP) == 0) {
   1400 				/*
   1401 				 * Check the M_LOOP flag to avoid an
   1402 				 * unnecessary PIM assert.
   1403 				 * XXX: M_LOOP is an ad-hoc hack...
   1404 				 */
   1405 				struct sockaddr_in6 sin6;
   1406 
   1407 				struct mbuf *mm;
   1408 				struct mrt6msg *im;
   1409 				struct omrt6msg *oim;
   1410 
   1411 				mm = m_copym(m, 0, sizeof(struct ip6_hdr), M_DONTWAIT);
   1412 				if (mm && M_UNWRITABLE(mm, sizeof(struct ip6_hdr)))
   1413 					mm = m_pullup(mm, sizeof(struct ip6_hdr));
   1414 				if (mm == NULL)
   1415 					return ENOBUFS;
   1416 
   1417 				oim = NULL;
   1418 				im = NULL;
   1419 				switch (ip6_mrouter_ver) {
   1420 				case MRT6_OINIT:
   1421 					oim = mtod(mm, struct omrt6msg *);
   1422 					oim->im6_msgtype = MRT6MSG_WRONGMIF;
   1423 					oim->im6_mbz = 0;
   1424 					break;
   1425 				case MRT6_INIT:
   1426 					im = mtod(mm, struct mrt6msg *);
   1427 					im->im6_msgtype = MRT6MSG_WRONGMIF;
   1428 					im->im6_mbz = 0;
   1429 					break;
   1430 				default:
   1431 					m_freem(mm);
   1432 					return EINVAL;
   1433 				}
   1434 
   1435 				for (mifp = mif6table, iif = 0;
   1436 				     iif < nummifs && mifp &&
   1437 					     mifp->m6_ifp != ifp;
   1438 				     mifp++, iif++)
   1439 					;
   1440 
   1441 				memset(&sin6, 0, sizeof(sin6));
   1442 				sin6.sin6_len = sizeof(sin6);
   1443 				sin6.sin6_family = AF_INET6;
   1444 				switch (ip6_mrouter_ver) {
   1445 				case MRT6_OINIT:
   1446 					oim->im6_mif = iif;
   1447 					sin6.sin6_addr = oim->im6_src;
   1448 					break;
   1449 				case MRT6_INIT:
   1450 					im->im6_mif = iif;
   1451 					sin6.sin6_addr = im->im6_src;
   1452 					break;
   1453 				}
   1454 
   1455 				mrt6stat.mrt6s_upcalls++;
   1456 
   1457 				if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
   1458 #ifdef MRT6DEBUG
   1459 					if (mrt6debug)
   1460 						log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
   1461 #endif
   1462 					++mrt6stat.mrt6s_upq_sockfull;
   1463 					return ENOBUFS;
   1464 				}
   1465 			}
   1466 		}
   1467 
   1468 		return 0;
   1469 	}
   1470 
   1471 	/* If I sourced this packet, it counts as output, else it was input. */
   1472 	if (m->m_pkthdr.rcvif_index == 0) {
   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 		IP6_STATINC(IP6_STAT_BADSCOPE);
   1492 		return error;
   1493 	}
   1494 	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
   1495 		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
   1496 			if (mif6table[mifi].m6_ifp == NULL)
   1497 				continue;
   1498 			/*
   1499 			 * check if the outgoing packet is going to break
   1500 			 * a scope boundary.
   1501 			 * XXX: For packets through PIM register tunnel
   1502 			 * interface, we believe the routing daemon.
   1503 			 */
   1504 			if ((mif6table[rt->mf6c_parent].m6_flags &
   1505 			     MIFF_REGISTER) == 0 &&
   1506 			    (mif6table[mifi].m6_flags & MIFF_REGISTER) == 0) {
   1507 				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
   1508 				    &oszone) ||
   1509 				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
   1510 				    &odzone) ||
   1511 				    iszone != oszone || idzone != odzone) {
   1512 					IP6_STATINC(IP6_STAT_BADSCOPE);
   1513 					continue;
   1514 				}
   1515 			}
   1516 
   1517 			mifp->m6_pkt_out++;
   1518 			mifp->m6_bytes_out += plen;
   1519 			MC6_SEND(ip6, mifp, m);
   1520 		}
   1521 	}
   1522 
   1523 	return 0;
   1524 }
   1525 
   1526 static void
   1527 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
   1528 {
   1529 	struct mbuf *mb_copy;
   1530 	struct ifnet *ifp = mifp->m6_ifp;
   1531 	int error __mrt6debugused = 0;
   1532 	int s;
   1533 	static struct route ro;
   1534 	bool ingroup;
   1535 	struct sockaddr_in6 dst6;
   1536 	u_long linkmtu;
   1537 
   1538 	s = splsoftnet();
   1539 
   1540 	/*
   1541 	 * Make a new reference to the packet; make sure that
   1542 	 * the IPv6 header is actually copied, not just referenced,
   1543 	 * so that ip6_output() only scribbles on the copy.
   1544 	 */
   1545 	mb_copy = m_copypacket(m, M_DONTWAIT);
   1546 	if (mb_copy && M_UNWRITABLE(mb_copy, sizeof(struct ip6_hdr)))
   1547 		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
   1548 	if (mb_copy == NULL) {
   1549 		splx(s);
   1550 		return;
   1551 	}
   1552 
   1553 	/* set MCAST flag to the outgoing packet */
   1554 	mb_copy->m_flags |= M_MCAST;
   1555 
   1556 	/*
   1557 	 * If we sourced the packet, call ip6_output since we may divide
   1558 	 * the packet into fragments when the packet is too big for the
   1559 	 * outgoing interface.
   1560 	 * Otherwise, we can simply send the packet to the interface
   1561 	 * sending queue.
   1562 	 */
   1563 	if (m->m_pkthdr.rcvif_index == 0) {
   1564 		struct ip6_moptions im6o;
   1565 
   1566 		im6o.im6o_multicast_if_index = if_get_index(ifp);
   1567 		/* XXX: ip6_output will override ip6->ip6_hlim */
   1568 		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
   1569 		im6o.im6o_multicast_loop = 1;
   1570 		error = ip6_output(mb_copy, NULL, &ro, IPV6_FORWARDING,
   1571 		    &im6o, NULL, NULL);
   1572 
   1573 #ifdef MRT6DEBUG
   1574 		if (mrt6debug & DEBUG_XMIT)
   1575 			log(LOG_DEBUG, "phyint_send on mif %td err %d\n",
   1576 			    mifp - mif6table, error);
   1577 #endif
   1578 		splx(s);
   1579 		return;
   1580 	}
   1581 
   1582 	/*
   1583 	 * If we belong to the destination multicast group
   1584 	 * on the outgoing interface, loop back a copy.
   1585 	 */
   1586 	/*
   1587 	 * Does not have to check source info, as it's alreay covered by
   1588 	 * ip6_input
   1589 	 */
   1590 	sockaddr_in6_init(&dst6, &ip6->ip6_dst, 0, 0, 0);
   1591 
   1592 	ingroup = in6_multi_group(&ip6->ip6_dst, ifp);
   1593 	if (ingroup) {
   1594 		ip6_mloopback(ifp, m,
   1595 		    satocsin6(rtcache_getdst(&ro)));
   1596 	}
   1597 
   1598 	/*
   1599 	 * Put the packet into the sending queue of the outgoing interface
   1600 	 * if it would fit in the MTU of the interface.
   1601 	 */
   1602 	linkmtu = IN6_LINKMTU(ifp);
   1603 	if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
   1604 		error = ip6_if_output(ifp, ifp, mb_copy, &dst6, NULL);
   1605 #ifdef MRT6DEBUG
   1606 		if (mrt6debug & DEBUG_XMIT)
   1607 			log(LOG_DEBUG, "phyint_send on mif %td err %d\n",
   1608 			    mifp - mif6table, error);
   1609 #endif
   1610 	} else {
   1611 		/*
   1612 		 * pMTU discovery is intentionally disabled by default, since
   1613 		 * various routers may notify pMTU in multicast, which can be
   1614 		 * a DDoS to a router.
   1615 		 */
   1616 		if (ip6_mcast_pmtu) {
   1617 			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
   1618 		} else {
   1619 			/* simply discard the packet */
   1620 #ifdef MRT6DEBUG
   1621 			if (mrt6debug & DEBUG_XMIT) {
   1622 				char ip6bufs[INET6_ADDRSTRLEN];
   1623 				char ip6bufd[INET6_ADDRSTRLEN];
   1624 				log(LOG_DEBUG,
   1625 				    "phyint_send: packet too big on %s o %s g %s"
   1626 				    " size %d(discarded)\n",
   1627 				    if_name(ifp),
   1628 				    IN6_PRINT(ip6bufs, &ip6->ip6_src),
   1629 				    IN6_PRINT(ip6bufd, &ip6->ip6_dst),
   1630 				    mb_copy->m_pkthdr.len);
   1631 			}
   1632 #endif
   1633 			m_freem(mb_copy);
   1634 		}
   1635 	}
   1636 
   1637 	splx(s);
   1638 }
   1639 
   1640 static int
   1641 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
   1642 {
   1643 	struct mbuf *mm;
   1644 	int i, len = m->m_pkthdr.len;
   1645 	struct sockaddr_in6 sin6;
   1646 	struct mrt6msg *im6;
   1647 
   1648 #ifdef MRT6DEBUG
   1649 	if (mrt6debug) {
   1650 		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
   1651 		log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
   1652 		    IN6_PRINT(ip6bufs, &ip6->ip6_src),
   1653 		    IN6_PRINT(ip6bufd, &ip6->ip6_dst));
   1654 	}
   1655 #endif
   1656 	PIM6_STATINC(PIM6_STAT_SND_REGISTERS);
   1657 
   1658 	/* Make a copy of the packet to send to the user level process */
   1659 	MGETHDR(mm, M_DONTWAIT, MT_HEADER);
   1660 	if (mm == NULL)
   1661 		return ENOBUFS;
   1662 	mm->m_data += max_linkhdr;
   1663 	mm->m_len = sizeof(struct ip6_hdr);
   1664 
   1665 	if ((mm->m_next = m_copypacket(m, M_DONTWAIT)) == NULL) {
   1666 		m_freem(mm);
   1667 		return ENOBUFS;
   1668 	}
   1669 	i = MHLEN - M_LEADINGSPACE(mm);
   1670 	if (i > len)
   1671 		i = len;
   1672 	mm = m_pullup(mm, i);
   1673 	if (mm == NULL)
   1674 		return ENOBUFS;
   1675 	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
   1676 
   1677 	/*
   1678 	 * Send message to routing daemon
   1679 	 */
   1680 	sockaddr_in6_init(&sin6, &ip6->ip6_src, 0, 0, 0);
   1681 
   1682 	im6 = mtod(mm, struct mrt6msg *);
   1683 	im6->im6_msgtype = MRT6MSG_WHOLEPKT;
   1684 	im6->im6_mbz = 0;
   1685 	im6->im6_mif = mif - mif6table;
   1686 
   1687 	/* iif info is not given for reg. encap.n */
   1688 	mrt6stat.mrt6s_upcalls++;
   1689 
   1690 	if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
   1691 #ifdef MRT6DEBUG
   1692 		if (mrt6debug)
   1693 			log(LOG_WARNING,
   1694 			    "register_send: ip6_mrouter socket queue full\n");
   1695 #endif
   1696 		++mrt6stat.mrt6s_upq_sockfull;
   1697 		return ENOBUFS;
   1698 	}
   1699 
   1700 	return 0;
   1701 }
   1702 
   1703 /*
   1704  * PIM sparse mode hook. Receives the pim control messages, and passes them up
   1705  * to the listening socket, using rip6_input.
   1706  *
   1707  * The only message processed is the REGISTER pim message; the pim header
   1708  * is stripped off, and the inner packet is passed to register_mforward.
   1709  */
   1710 int
   1711 pim6_input(struct mbuf **mp, int *offp, int proto)
   1712 {
   1713 	struct pim *pim;
   1714 	struct ip6_hdr *ip6 __mrt6debugused;
   1715 	int pimlen;
   1716 	struct mbuf *m = *mp;
   1717 	int minlen;
   1718 	int off = *offp;
   1719 
   1720 	PIM6_STATINC(PIM6_STAT_RCV_TOTAL);
   1721 
   1722 	ip6 = mtod(m, struct ip6_hdr *);
   1723 	pimlen = m->m_pkthdr.len - off;
   1724 
   1725 	/*
   1726 	 * Validate lengths
   1727 	 */
   1728 	if (pimlen < PIM_MINLEN) {
   1729 		PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
   1730 #ifdef MRT6DEBUG
   1731 		if (mrt6debug & DEBUG_PIM)
   1732 			log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
   1733 #endif
   1734 		m_freem(m);
   1735 		return IPPROTO_DONE;
   1736 	}
   1737 
   1738 	/*
   1739 	 * If the packet is at least as big as a REGISTER, go ahead
   1740 	 * and grab the PIM REGISTER header size, to avoid another
   1741 	 * possible m_pullup() later.
   1742 	 *
   1743 	 * PIM_MINLEN       == pimhdr + u_int32 == 8
   1744 	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
   1745 	 */
   1746 	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
   1747 
   1748 	/*
   1749 	 * Make sure that the IP6 and PIM headers in contiguous memory, and
   1750 	 * possibly the PIM REGISTER header
   1751 	 */
   1752 	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
   1753 	if (pim == NULL) {
   1754 		PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
   1755 		return IPPROTO_DONE;
   1756 	}
   1757 
   1758 	/* PIM version check */
   1759 	if (pim->pim_ver != PIM_VERSION) {
   1760 		PIM6_STATINC(PIM6_STAT_RCV_BADVERSION);
   1761 #ifdef MRT6DEBUG
   1762 		log(LOG_ERR,
   1763 		    "pim6_input: incorrect version %d, expecting %d\n",
   1764 		    pim->pim_ver, PIM_VERSION);
   1765 #endif
   1766 		m_freem(m);
   1767 		return IPPROTO_DONE;
   1768 	}
   1769 
   1770 #define PIM6_CHECKSUM
   1771 #ifdef PIM6_CHECKSUM
   1772 	{
   1773 		int cksumlen;
   1774 
   1775 		/*
   1776 		 * Validate checksum.
   1777 		 * If PIM REGISTER, exclude the data packet
   1778 		 */
   1779 		if (pim->pim_type == PIM_REGISTER)
   1780 			cksumlen = PIM_MINLEN;
   1781 		else
   1782 			cksumlen = pimlen;
   1783 
   1784 		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
   1785 			PIM6_STATINC(PIM6_STAT_RCV_BADSUM);
   1786 #ifdef MRT6DEBUG
   1787 			if (mrt6debug & DEBUG_PIM)
   1788 				log(LOG_DEBUG,
   1789 				    "pim6_input: invalid checksum\n");
   1790 #endif
   1791 			m_freem(m);
   1792 			return IPPROTO_DONE;
   1793 		}
   1794 	}
   1795 #endif /* PIM_CHECKSUM */
   1796 
   1797 	if (pim->pim_type == PIM_REGISTER) {
   1798 		/*
   1799 		 * since this is a REGISTER, we'll make a copy of the register
   1800 		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
   1801 		 * routing daemon.
   1802 		 */
   1803 		static const struct sockaddr_in6 dst = {
   1804 			.sin6_len = sizeof(dst),
   1805 			.sin6_family = AF_INET6,
   1806 		};
   1807 
   1808 		struct mbuf *mcp;
   1809 		struct ip6_hdr *eip6;
   1810 		u_int32_t *reghdr;
   1811 
   1812 		PIM6_STATINC(PIM6_STAT_RCV_REGISTERS);
   1813 
   1814 		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
   1815 #ifdef MRT6DEBUG
   1816 			if (mrt6debug & DEBUG_PIM)
   1817 				log(LOG_DEBUG,
   1818 				    "pim6_input: register mif not set: %d\n",
   1819 				    reg_mif_num);
   1820 #endif
   1821 			m_freem(m);
   1822 			return IPPROTO_DONE;
   1823 		}
   1824 
   1825 		reghdr = (u_int32_t *)(pim + 1);
   1826 
   1827 		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
   1828 			goto pim6_input_to_daemon;
   1829 
   1830 		/*
   1831 		 * Validate length
   1832 		 */
   1833 		if (pimlen < PIM6_REG_MINLEN) {
   1834 #ifdef MRT6DEBUG
   1835 			char ip6buf[INET6_ADDRSTRLEN];
   1836 			log(LOG_ERR,
   1837 			    "pim6_input: register packet size too "
   1838 			    "small %d from %s\n",
   1839 			    pimlen, IN6_PRINT(ip6buf, &ip6->ip6_src));
   1840 #endif
   1841 			PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
   1842 			PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
   1843 			m_freem(m);
   1844 			return IPPROTO_DONE;
   1845 		}
   1846 
   1847 		eip6 = (struct ip6_hdr *)(reghdr + 1);
   1848 #ifdef MRT6DEBUG
   1849 		if (mrt6debug & DEBUG_PIM) {
   1850 			char ip6bufs[INET6_ADDRSTRLEN];
   1851 			char ip6bufd[INET6_ADDRSTRLEN];
   1852 			log(LOG_DEBUG,
   1853 			    "pim6_input[register], eip6: %s -> %s, "
   1854 			    "eip6 plen %d\n",
   1855 			    IN6_PRINT(ip6bufs, &eip6->ip6_src),
   1856 			    IN6_PRINT(ip6bufd, &eip6->ip6_dst),
   1857 			    ntohs(eip6->ip6_plen));
   1858 		}
   1859 #endif
   1860 
   1861 		/* verify the version number of the inner packet */
   1862 		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
   1863 			PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
   1864 #ifdef MRT6DEBUG
   1865 			log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
   1866 			    "of the inner packet\n",
   1867 			    (eip6->ip6_vfc & IPV6_VERSION));
   1868 #endif
   1869 			m_freem(m);
   1870 			return IPPROTO_DONE;
   1871 		}
   1872 
   1873 		/* verify the inner packet is destined to a mcast group */
   1874 		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
   1875 			PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
   1876 #ifdef MRT6DEBUG
   1877 			if (mrt6debug & DEBUG_PIM) {
   1878 				char ip6buf[INET6_ADDRSTRLEN];
   1879 				log(LOG_DEBUG,
   1880 				    "pim6_input: inner packet of register "
   1881 				    "is not multicast %s\n",
   1882 				    IN6_PRINT(ip6buf, &eip6->ip6_dst));
   1883 			}
   1884 #endif
   1885 			m_freem(m);
   1886 			return IPPROTO_DONE;
   1887 		}
   1888 
   1889 		/*
   1890 		 * make a copy of the whole header to pass to the daemon later.
   1891 		 */
   1892 		mcp = m_copym(m, 0, off + PIM6_REG_MINLEN, M_DONTWAIT);
   1893 		if (mcp == NULL) {
   1894 #ifdef MRT6DEBUG
   1895 			log(LOG_ERR,
   1896 			    "pim6_input: pim register: "
   1897 			    "could not copy register head\n");
   1898 #endif
   1899 			m_freem(m);
   1900 			return IPPROTO_DONE;
   1901 		}
   1902 
   1903 		/*
   1904 		 * forward the inner ip6 packet; point m_data at the inner ip6.
   1905 		 */
   1906 		m_adj(m, off + PIM_MINLEN);
   1907 #ifdef MRT6DEBUG
   1908 		if (mrt6debug & DEBUG_PIM) {
   1909 			char ip6bufs[INET6_ADDRSTRLEN];
   1910 			char ip6bufd[INET6_ADDRSTRLEN];
   1911 			log(LOG_DEBUG,
   1912 			    "pim6_input: forwarding decapsulated register: "
   1913 			    "src %s, dst %s, mif %d\n",
   1914 			    IN6_PRINT(ip6bufs, &eip6->ip6_src),
   1915 			    IN6_PRINT(ip6bufd, &eip6->ip6_dst),
   1916 			    reg_mif_num);
   1917 		}
   1918 #endif
   1919 
   1920 		looutput(mif6table[reg_mif_num].m6_ifp, m, sin6tocsa(&dst),
   1921 		    NULL);
   1922 
   1923 		/* prepare the register head to send to the mrouting daemon */
   1924 		m = mcp;
   1925 	}
   1926 
   1927 	/*
   1928 	 * Pass the PIM message up to the daemon; if it is a register message
   1929 	 * pass the 'head' only up to the daemon. This includes the
   1930 	 * encapsulator ip6 header, pim header, register header and the
   1931 	 * encapsulated ip6 header.
   1932 	 */
   1933 pim6_input_to_daemon:
   1934 	/*
   1935 	 * Currently, rip6_input() is always called holding softnet_lock
   1936 	 * by ipintr()(!NET_MPSAFE) or PR_INPUT_WRAP()(NET_MPSAFE).
   1937 	 */
   1938 	KASSERT(mutex_owned(softnet_lock));
   1939 	rip6_input(&m, offp, proto);
   1940 	return IPPROTO_DONE;
   1941 }
   1942 
   1943 static int
   1944 sysctl_net_inet6_pim6_stats(SYSCTLFN_ARGS)
   1945 {
   1946 
   1947 	return (NETSTAT_SYSCTL(pim6stat_percpu, PIM6_NSTATS));
   1948 }
   1949 
   1950 static void
   1951 sysctl_net_inet6_pim6_setup(struct sysctllog **clog)
   1952 {
   1953 
   1954 	sysctl_createv(clog, 0, NULL, NULL,
   1955 		       CTLFLAG_PERMANENT,
   1956 		       CTLTYPE_NODE, "inet6", NULL,
   1957 		       NULL, 0, NULL, 0,
   1958 		       CTL_NET, PF_INET6, CTL_EOL);
   1959 	sysctl_createv(clog, 0, NULL, NULL,
   1960 		       CTLFLAG_PERMANENT,
   1961 		       CTLTYPE_NODE, "pim6",
   1962 		       SYSCTL_DESCR("PIMv6 settings"),
   1963 		       NULL, 0, NULL, 0,
   1964 		       CTL_NET, PF_INET6, IPPROTO_PIM, CTL_EOL);
   1965 
   1966 	sysctl_createv(clog, 0, NULL, NULL,
   1967 		       CTLFLAG_PERMANENT,
   1968 		       CTLTYPE_STRUCT, "stats",
   1969 		       SYSCTL_DESCR("PIMv6 statistics"),
   1970 		       sysctl_net_inet6_pim6_stats, 0, NULL, 0,
   1971 		       CTL_NET, PF_INET6, IPPROTO_PIM, PIM6CTL_STATS,
   1972 		       CTL_EOL);
   1973 }
   1974