Home | History | Annotate | Line # | Download | only in netinet6
      1 /*	$NetBSD: ip6_mroute.c,v 1.133 2025/01/02 16:10:22 andvar 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 1.133 2025/01/02 16:10:22 andvar 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 multicasted 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 				ifp = mif6table[mifi].m6_ifp;
    544 				sockaddr_in6_init(&sin6, &in6addr_any, 0, 0, 0);
    545 				if_mcast_op(ifp, SIOCDELMULTI,
    546 				    sin6tocsa(&sin6));
    547 			}
    548 		}
    549 	}
    550 
    551 	memset((void *)mif6table, 0, sizeof(mif6table));
    552 	nummifs = 0;
    553 
    554 	pim6 = 0; /* used to stub out/in pim specific code */
    555 
    556 	callout_stop(&expire_upcalls_ch);
    557 
    558 	/*
    559 	 * Free all multicast forwarding cache entries.
    560 	 */
    561 	for (i = 0; i < MF6CTBLSIZ; i++) {
    562 		rt = mf6ctable[i];
    563 		while (rt) {
    564 			struct mf6c *frt;
    565 
    566 			for (rte = rt->mf6c_stall; rte != NULL; ) {
    567 				struct rtdetq *n = rte->next;
    568 
    569 				m_freem(rte->m);
    570 				free(rte, M_MRTABLE);
    571 				rte = n;
    572 			}
    573 			frt = rt;
    574 			rt = rt->mf6c_next;
    575 			free(frt, M_MRTABLE);
    576 		}
    577 	}
    578 
    579 	memset((void *)mf6ctable, 0, sizeof(mf6ctable));
    580 
    581 	/*
    582 	 * Reset register interface
    583 	 */
    584 	if (reg_mif_num != (mifi_t)-1) {
    585 		if_detach(&multicast_register_if6);
    586 		reg_mif_num = (mifi_t)-1;
    587 	}
    588 
    589 	ip6_mrouter = NULL;
    590 	ip6_mrouter_ver = 0;
    591 
    592 	splx(s);
    593 
    594 #ifdef MRT6DEBUG
    595 	if (mrt6debug)
    596 		log(LOG_DEBUG, "ip6_mrouter_done\n");
    597 #endif
    598 
    599 	return 0;
    600 }
    601 
    602 void
    603 ip6_mrouter_detach(struct ifnet *ifp)
    604 {
    605 	struct rtdetq *rte;
    606 	struct mf6c *mfc;
    607 	mifi_t mifi;
    608 	int i;
    609 
    610 	if (ip6_mrouter == NULL)
    611 		return;
    612 
    613 	/*
    614 	 * Delete a mif which points to ifp.
    615 	 */
    616 	for (mifi = 0; mifi < nummifs; mifi++)
    617 		if (mif6table[mifi].m6_ifp == ifp)
    618 			del_m6if(&mifi);
    619 
    620 	/*
    621 	 * Clear rte->ifp of cache entries received on ifp.
    622 	 */
    623 	for (i = 0; i < MF6CTBLSIZ; i++) {
    624 		if (n6expire[i] == 0)
    625 			continue;
    626 
    627 		for (mfc = mf6ctable[i]; mfc != NULL; mfc = mfc->mf6c_next) {
    628 			for (rte = mfc->mf6c_stall; rte != NULL; rte = rte->next) {
    629 				if (rte->ifp == ifp)
    630 					rte->ifp = NULL;
    631 			}
    632 		}
    633 	}
    634 }
    635 
    636 /*
    637  * Add a mif to the mif table
    638  */
    639 static int
    640 add_m6if(struct mif6ctl *mifcp)
    641 {
    642 	struct mif6 *mifp;
    643 	struct ifnet *ifp;
    644 	struct sockaddr_in6 sin6;
    645 	int error, s;
    646 
    647 	if (mifcp->mif6c_mifi >= MAXMIFS)
    648 		return EINVAL;
    649 	mifp = mif6table + mifcp->mif6c_mifi;
    650 	if (mifp->m6_ifp)
    651 		return EADDRINUSE; /* XXX: is it appropriate? */
    652 	if (!mifcp->mif6c_pifi || (ifp = if_byindex(mifcp->mif6c_pifi)) == NULL)
    653 		return ENXIO;
    654 
    655 	if (mifcp->mif6c_flags & MIFF_REGISTER) {
    656 		ifp = &multicast_register_if6;
    657 
    658 		if (reg_mif_num == (mifi_t)-1) {
    659 			strlcpy(ifp->if_xname, "register_mif",
    660 			    sizeof(ifp->if_xname));
    661 			ifp->if_flags |= IFF_LOOPBACK;
    662 			ifp->if_index = mifcp->mif6c_mifi;
    663 			reg_mif_num = mifcp->mif6c_mifi;
    664 			if_attach(ifp);
    665 		}
    666 	} else {
    667 		/* Make sure the interface supports multicast */
    668 		if ((ifp->if_flags & IFF_MULTICAST) == 0)
    669 			return EOPNOTSUPP;
    670 
    671 		s = splsoftnet();
    672 		/*
    673 		 * Enable promiscuous reception of all IPv6 multicasts
    674 		 * from the interface.
    675 		 */
    676 		sockaddr_in6_init(&sin6, &in6addr_any, 0, 0, 0);
    677 		error = if_mcast_op(ifp, SIOCADDMULTI, sin6tosa(&sin6));
    678 		splx(s);
    679 		if (error)
    680 			return error;
    681 	}
    682 
    683 	s = splsoftnet();
    684 	mifp->m6_flags     = mifcp->mif6c_flags;
    685 	mifp->m6_ifp       = ifp;
    686 	/* initialize per mif pkt counters */
    687 	mifp->m6_pkt_in    = 0;
    688 	mifp->m6_pkt_out   = 0;
    689 	mifp->m6_bytes_in  = 0;
    690 	mifp->m6_bytes_out = 0;
    691 	splx(s);
    692 
    693 	/* Adjust nummifs up if the mifi is higher than nummifs */
    694 	if (nummifs <= mifcp->mif6c_mifi)
    695 		nummifs = mifcp->mif6c_mifi + 1;
    696 
    697 #ifdef MRT6DEBUG
    698 	if (mrt6debug)
    699 		log(LOG_DEBUG,
    700 		    "add_mif #%d, phyint %s\n",
    701 		    mifcp->mif6c_mifi, ifp->if_xname);
    702 #endif
    703 
    704 	return 0;
    705 }
    706 
    707 /*
    708  * Delete a mif from the mif table
    709  */
    710 static int
    711 del_m6if(mifi_t *mifip)
    712 {
    713 	struct mif6 *mifp = mif6table + *mifip;
    714 	mifi_t mifi;
    715 	struct ifnet *ifp;
    716 	struct sockaddr_in6 sin6;
    717 	int s;
    718 
    719 	if (*mifip >= nummifs)
    720 		return EINVAL;
    721 	if (mifp->m6_ifp == NULL)
    722 		return EINVAL;
    723 
    724 	s = splsoftnet();
    725 
    726 	if (!(mifp->m6_flags & MIFF_REGISTER)) {
    727 		/*
    728 		 * XXX: what if there is yet IPv4 multicast daemon
    729 		 *      using the interface?
    730 		 */
    731 		ifp = mifp->m6_ifp;
    732 
    733 		sockaddr_in6_init(&sin6, &in6addr_any, 0, 0, 0);
    734 		if_mcast_op(ifp, SIOCDELMULTI, sin6tosa(&sin6));
    735 	} else {
    736 		if (reg_mif_num != (mifi_t)-1) {
    737 			if_detach(&multicast_register_if6);
    738 			reg_mif_num = (mifi_t)-1;
    739 		}
    740 	}
    741 
    742 	memset((void *)mifp, 0, sizeof (*mifp));
    743 
    744 	/* Adjust nummifs down */
    745 	for (mifi = nummifs; mifi > 0; mifi--)
    746 		if (mif6table[mifi - 1].m6_ifp)
    747 			break;
    748 	nummifs = mifi;
    749 
    750 	splx(s);
    751 
    752 #ifdef MRT6DEBUG
    753 	if (mrt6debug)
    754 		log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
    755 #endif
    756 
    757 	return 0;
    758 }
    759 
    760 /*
    761  * Add an mfc entry
    762  */
    763 static int
    764 add_m6fc(struct mf6cctl *mfccp)
    765 {
    766 	struct mf6c *rt;
    767 	u_long hash;
    768 	struct rtdetq *rte;
    769 	u_short nstl;
    770 	int s;
    771 	char ip6bufo[INET6_ADDRSTRLEN], ip6bufm[INET6_ADDRSTRLEN];
    772 
    773 	MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
    774 		 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
    775 
    776 	/* If an entry already exists, just update the fields */
    777 	if (rt) {
    778 #ifdef MRT6DEBUG
    779 		if (mrt6debug & DEBUG_MFC)
    780 			log(LOG_DEBUG,"add_m6fc update o %s g %s p %x\n",
    781 			    IN6_PRINT(ip6bufo,
    782 			    &mfccp->mf6cc_origin.sin6_addr),
    783 			    IN6_PRINT(ip6bufm,
    784 			    &mfccp->mf6cc_mcastgrp.sin6_addr),
    785 			    mfccp->mf6cc_parent);
    786 #endif
    787 
    788 		s = splsoftnet();
    789 		rt->mf6c_parent = mfccp->mf6cc_parent;
    790 		rt->mf6c_ifset = mfccp->mf6cc_ifset;
    791 		splx(s);
    792 		return 0;
    793 	}
    794 
    795 	/*
    796 	 * Find the entry for which the upcall was made and update
    797 	 */
    798 	s = splsoftnet();
    799 	hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
    800 			mfccp->mf6cc_mcastgrp.sin6_addr);
    801 	for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
    802 		if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
    803 				       &mfccp->mf6cc_origin.sin6_addr) &&
    804 		    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
    805 				       &mfccp->mf6cc_mcastgrp.sin6_addr) &&
    806 		    (rt->mf6c_stall != NULL)) {
    807 
    808 			if (nstl++)
    809 				log(LOG_ERR,
    810 				    "add_m6fc: %s o %s g %s p %x dbx %p\n",
    811 				    "multiple kernel entries",
    812 				    IN6_PRINT(ip6bufo,
    813 				    &mfccp->mf6cc_origin.sin6_addr),
    814 				    IN6_PRINT(ip6bufm,
    815 				    &mfccp->mf6cc_mcastgrp.sin6_addr),
    816 				    mfccp->mf6cc_parent, rt->mf6c_stall);
    817 
    818 #ifdef MRT6DEBUG
    819 			if (mrt6debug & DEBUG_MFC)
    820 				log(LOG_DEBUG,
    821 				    "add_m6fc o %s g %s p %x dbg %p\n",
    822 				    IN6_PRINT(ip6bufo,
    823 				    &mfccp->mf6cc_origin.sin6_addr),
    824 				    IN6_PRINT(ip6bufm,
    825 				    &mfccp->mf6cc_mcastgrp.sin6_addr),
    826 				    mfccp->mf6cc_parent, rt->mf6c_stall);
    827 #endif
    828 
    829 			rt->mf6c_origin     = mfccp->mf6cc_origin;
    830 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
    831 			rt->mf6c_parent     = mfccp->mf6cc_parent;
    832 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
    833 			/* initialize pkt counters per src-grp */
    834 			rt->mf6c_pkt_cnt    = 0;
    835 			rt->mf6c_byte_cnt   = 0;
    836 			rt->mf6c_wrong_if   = 0;
    837 
    838 			rt->mf6c_expire = 0;	/* Don't clean this guy up */
    839 			n6expire[hash]--;
    840 
    841 			/* free packets Qed at the end of this entry */
    842 			for (rte = rt->mf6c_stall; rte != NULL; ) {
    843 				struct rtdetq *n = rte->next;
    844 				if (rte->ifp) {
    845 					ip6_mdq(rte->m, rte->ifp, rt);
    846 				}
    847 				m_freem(rte->m);
    848 #ifdef UPCALL_TIMING
    849 				collate(&(rte->t));
    850 #endif
    851 				free(rte, M_MRTABLE);
    852 				rte = n;
    853 			}
    854 			rt->mf6c_stall = NULL;
    855 		}
    856 	}
    857 
    858 	/*
    859 	 * It is possible that an entry is being inserted without an upcall
    860 	 */
    861 	if (nstl == 0) {
    862 #ifdef MRT6DEBUG
    863 		if (mrt6debug & DEBUG_MFC)
    864 			log(LOG_DEBUG,
    865 			    "add_mfc no upcall h %ld o %s g %s p %x\n",
    866 			    hash,
    867 			    IN6_PRINT(ip6bufo,
    868 			    &mfccp->mf6cc_origin.sin6_addr),
    869 			    IN6_PRINT(ip6bufm,
    870 			    &mfccp->mf6cc_mcastgrp.sin6_addr),
    871 			    mfccp->mf6cc_parent);
    872 #endif
    873 
    874 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
    875 
    876 			if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
    877 					       &mfccp->mf6cc_origin.sin6_addr)&&
    878 			    IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
    879 					       &mfccp->mf6cc_mcastgrp.sin6_addr)) {
    880 
    881 				rt->mf6c_origin     = mfccp->mf6cc_origin;
    882 				rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
    883 				rt->mf6c_parent     = mfccp->mf6cc_parent;
    884 				rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
    885 				/* initialize pkt counters per src-grp */
    886 				rt->mf6c_pkt_cnt    = 0;
    887 				rt->mf6c_byte_cnt   = 0;
    888 				rt->mf6c_wrong_if   = 0;
    889 
    890 				if (rt->mf6c_expire)
    891 					n6expire[hash]--;
    892 				rt->mf6c_expire	   = 0;
    893 			}
    894 		}
    895 		if (rt == NULL) {
    896 			/* no upcall, so make a new entry */
    897 			rt = malloc(sizeof(*rt), M_MRTABLE, M_NOWAIT);
    898 			if (rt == NULL) {
    899 				splx(s);
    900 				return ENOBUFS;
    901 			}
    902 
    903 			/* insert new entry at head of hash chain */
    904 			rt->mf6c_origin     = mfccp->mf6cc_origin;
    905 			rt->mf6c_mcastgrp   = mfccp->mf6cc_mcastgrp;
    906 			rt->mf6c_parent     = mfccp->mf6cc_parent;
    907 			rt->mf6c_ifset	    = mfccp->mf6cc_ifset;
    908 			/* initialize pkt counters per src-grp */
    909 			rt->mf6c_pkt_cnt    = 0;
    910 			rt->mf6c_byte_cnt   = 0;
    911 			rt->mf6c_wrong_if   = 0;
    912 			rt->mf6c_expire     = 0;
    913 			rt->mf6c_stall = NULL;
    914 
    915 			/* link into table */
    916 			rt->mf6c_next  = mf6ctable[hash];
    917 			mf6ctable[hash] = rt;
    918 		}
    919 	}
    920 	splx(s);
    921 	return 0;
    922 }
    923 
    924 #ifdef UPCALL_TIMING
    925 /*
    926  * collect delay statistics on the upcalls
    927  */
    928 static void
    929 collate(struct timeval *t)
    930 {
    931 	u_long d;
    932 	struct timeval tp;
    933 	u_long delta;
    934 
    935 	GET_TIME(tp);
    936 
    937 	if (TV_LT(*t, tp))
    938 	{
    939 		TV_DELTA(tp, *t, delta);
    940 
    941 		d = delta >> 10;
    942 		if (d > UPCALL_MAX)
    943 			d = UPCALL_MAX;
    944 
    945 		++upcall_data[d];
    946 	}
    947 }
    948 #endif /* UPCALL_TIMING */
    949 
    950 /*
    951  * Delete an mfc entry
    952  */
    953 static int
    954 del_m6fc(struct mf6cctl *mfccp)
    955 {
    956 	struct sockaddr_in6 	origin;
    957 	struct sockaddr_in6 	mcastgrp;
    958 	struct mf6c 		*rt;
    959 	struct mf6c	 	**nptr;
    960 	u_long 		hash;
    961 	int s;
    962 
    963 	origin = mfccp->mf6cc_origin;
    964 	mcastgrp = mfccp->mf6cc_mcastgrp;
    965 	hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
    966 
    967 #ifdef MRT6DEBUG
    968 	if (mrt6debug & DEBUG_MFC) {
    969 		char ip6bufo[INET6_ADDRSTRLEN], ip6bufm[INET6_ADDRSTRLEN];
    970 		log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
    971 		    IN6_PRINT(ip6bufo, &origin.sin6_addr),
    972 		    IN6_PRINT(ip6bufm, &mcastgrp.sin6_addr));
    973 	}
    974 #endif
    975 
    976 	s = splsoftnet();
    977 
    978 	nptr = &mf6ctable[hash];
    979 	while ((rt = *nptr) != NULL) {
    980 		if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
    981 				       &rt->mf6c_origin.sin6_addr) &&
    982 		    IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
    983 				       &rt->mf6c_mcastgrp.sin6_addr) &&
    984 		    rt->mf6c_stall == NULL)
    985 			break;
    986 
    987 		nptr = &rt->mf6c_next;
    988 	}
    989 	if (rt == NULL) {
    990 		splx(s);
    991 		return EADDRNOTAVAIL;
    992 	}
    993 
    994 	*nptr = rt->mf6c_next;
    995 	free(rt, M_MRTABLE);
    996 
    997 	splx(s);
    998 
    999 	return 0;
   1000 }
   1001 
   1002 static int
   1003 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
   1004 {
   1005 	if (s) {
   1006 		if (sbappendaddr(&s->so_rcv, sin6tosa(src), mm, NULL) != 0) {
   1007 			sorwakeup(s);
   1008 			return 0;
   1009 		}
   1010 		soroverflow(s);
   1011 	}
   1012 	m_freem(mm);
   1013 	return -1;
   1014 }
   1015 
   1016 /*
   1017  * IPv6 multicast forwarding function. This function assumes that the packet
   1018  * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
   1019  * pointed to by "ifp", and the packet is to be relayed to other networks
   1020  * that have members of the packet's destination IPv6 multicast group.
   1021  *
   1022  * The packet is returned unscathed to the caller, unless it is
   1023  * erroneous, in which case a non-zero return value tells the caller to
   1024  * discard it.
   1025  */
   1026 int
   1027 ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
   1028 {
   1029 	struct mf6c *rt;
   1030 	struct mif6 *mifp;
   1031 	struct mbuf *mm;
   1032 	int s;
   1033 	mifi_t mifi;
   1034 	struct sockaddr_in6 sin6;
   1035 	char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
   1036 
   1037 #ifdef MRT6DEBUG
   1038 	if (mrt6debug & DEBUG_FORWARD)
   1039 		log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
   1040 		    IN6_PRINT(ip6bufs, &ip6->ip6_src),
   1041 		    IN6_PRINT(ip6bufd, &ip6->ip6_dst),
   1042 		    ifp->if_index);
   1043 #endif
   1044 
   1045 	/*
   1046 	 * Don't forward a packet with Hop limit of zero or one,
   1047 	 * or a packet destined to a local-only group.
   1048 	 */
   1049 	if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_NODELOCAL(&ip6->ip6_dst) ||
   1050 	    IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
   1051 		return 0;
   1052 	ip6->ip6_hlim--;
   1053 
   1054 	/*
   1055 	 * Source address check: do not forward packets with unspecified
   1056 	 * source. It was discussed in July 2000, on ipngwg mailing list.
   1057 	 * This is rather more serious than unicast cases, because some
   1058 	 * MLD packets can be sent with the unspecified source address
   1059 	 * (although such packets must normally set the hop limit field to 1).
   1060 	 */
   1061 	if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
   1062 		IP6_STATINC(IP6_STAT_CANTFORWARD);
   1063 		if (ip6_log_time + ip6_log_interval < time_uptime) {
   1064 			ip6_log_time = time_uptime;
   1065 			log(LOG_DEBUG,
   1066 			    "cannot forward "
   1067 			    "from %s to %s nxt %d received on %s\n",
   1068 			    IN6_PRINT(ip6bufs, &ip6->ip6_src),
   1069 			    IN6_PRINT(ip6bufd, &ip6->ip6_dst),
   1070 			    ip6->ip6_nxt,
   1071 			    m->m_pkthdr.rcvif_index ?
   1072 			    if_name(m_get_rcvif_NOMPSAFE(m)) : "?");
   1073 		}
   1074 		return 0;
   1075 	}
   1076 
   1077 	/*
   1078 	 * Determine forwarding mifs from the forwarding cache table
   1079 	 */
   1080 	s = splsoftnet();
   1081 	MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
   1082 
   1083 	/* Entry exists, so forward if necessary */
   1084 	if (rt) {
   1085 		splx(s);
   1086 		return ip6_mdq(m, ifp, rt);
   1087 	} else {
   1088 		/*
   1089 		 * If we don't have a route for packet's origin, make a copy
   1090 		 * of the packet and send message to routing daemon.
   1091 		 */
   1092 
   1093 		struct mbuf *mb0;
   1094 		struct rtdetq *rte;
   1095 		u_long hash;
   1096 
   1097 #ifdef UPCALL_TIMING
   1098 		struct timeval tp;
   1099 		GET_TIME(tp);
   1100 #endif
   1101 
   1102 		mrt6stat.mrt6s_no_route++;
   1103 #ifdef MRT6DEBUG
   1104 		if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
   1105 			log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
   1106 			    IN6_PRINT(ip6bufs, &ip6->ip6_src),
   1107 			    IN6_PRINT(ip6bufd, &ip6->ip6_dst));
   1108 #endif
   1109 
   1110 		/*
   1111 		 * Allocate mbufs early so that we don't do extra work if we
   1112 		 * are just going to fail anyway.
   1113 		 */
   1114 		rte = malloc(sizeof(*rte), M_MRTABLE, M_NOWAIT);
   1115 		if (rte == NULL) {
   1116 			splx(s);
   1117 			return ENOBUFS;
   1118 		}
   1119 		mb0 = m_copypacket(m, M_DONTWAIT);
   1120 
   1121 		/*
   1122 		 * Pullup packet header if needed before storing it,
   1123 		 * as other references may modify it in the meantime.
   1124 		 */
   1125 		if (mb0 && M_UNWRITABLE(mb0, sizeof(struct ip6_hdr)))
   1126 			mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
   1127 		if (mb0 == NULL) {
   1128 			free(rte, M_MRTABLE);
   1129 			splx(s);
   1130 			return ENOBUFS;
   1131 		}
   1132 
   1133 		/* is there an upcall waiting for this packet? */
   1134 		hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
   1135 		for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
   1136 			if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
   1137 					       &rt->mf6c_origin.sin6_addr) &&
   1138 			    IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
   1139 					       &rt->mf6c_mcastgrp.sin6_addr) &&
   1140 			    (rt->mf6c_stall != NULL))
   1141 				break;
   1142 		}
   1143 
   1144 		if (rt == NULL) {
   1145 			struct mrt6msg *im;
   1146 			struct omrt6msg *oim;
   1147 
   1148 			/* no upcall, so make a new entry */
   1149 			rt = malloc(sizeof(*rt), M_MRTABLE, M_NOWAIT);
   1150 			if (rt == NULL) {
   1151 				free(rte, M_MRTABLE);
   1152 				m_freem(mb0);
   1153 				splx(s);
   1154 				return ENOBUFS;
   1155 			}
   1156 
   1157 			/*
   1158 			 * Make a copy of the header to send to the user
   1159 			 * level process
   1160 			 */
   1161 			mm = m_copym(mb0, 0, sizeof(struct ip6_hdr), M_DONTWAIT);
   1162 
   1163 			if (mm == NULL) {
   1164 				free(rte, M_MRTABLE);
   1165 				m_freem(mb0);
   1166 				free(rt, M_MRTABLE);
   1167 				splx(s);
   1168 				return ENOBUFS;
   1169 			}
   1170 
   1171 			/*
   1172 			 * Send message to routing daemon
   1173 			 */
   1174 			sockaddr_in6_init(&sin6, &ip6->ip6_src, 0, 0, 0);
   1175 
   1176 			im = NULL;
   1177 			oim = NULL;
   1178 			switch (ip6_mrouter_ver) {
   1179 			case MRT6_OINIT:
   1180 				oim = mtod(mm, struct omrt6msg *);
   1181 				oim->im6_msgtype = MRT6MSG_NOCACHE;
   1182 				oim->im6_mbz = 0;
   1183 				break;
   1184 			case MRT6_INIT:
   1185 				im = mtod(mm, struct mrt6msg *);
   1186 				im->im6_msgtype = MRT6MSG_NOCACHE;
   1187 				im->im6_mbz = 0;
   1188 				break;
   1189 			default:
   1190 				free(rte, M_MRTABLE);
   1191 				m_freem(mb0);
   1192 				free(rt, M_MRTABLE);
   1193 				splx(s);
   1194 				return EINVAL;
   1195 			}
   1196 
   1197 #ifdef MRT6DEBUG
   1198 			if (mrt6debug & DEBUG_FORWARD)
   1199 				log(LOG_DEBUG,
   1200 				    "getting the iif info in the kernel\n");
   1201 #endif
   1202 
   1203 			for (mifp = mif6table, mifi = 0;
   1204 			     mifi < nummifs && mifp->m6_ifp != ifp;
   1205 			     mifp++, mifi++)
   1206 				;
   1207 
   1208 			switch (ip6_mrouter_ver) {
   1209 			case MRT6_OINIT:
   1210 				oim->im6_mif = mifi;
   1211 				break;
   1212 			case MRT6_INIT:
   1213 				im->im6_mif = mifi;
   1214 				break;
   1215 			}
   1216 
   1217 			if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
   1218 				log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
   1219 				    "socket queue full\n");
   1220 				mrt6stat.mrt6s_upq_sockfull++;
   1221 				free(rte, M_MRTABLE);
   1222 				m_freem(mb0);
   1223 				free(rt, M_MRTABLE);
   1224 				splx(s);
   1225 				return ENOBUFS;
   1226 			}
   1227 
   1228 			mrt6stat.mrt6s_upcalls++;
   1229 
   1230 			/* insert new entry at head of hash chain */
   1231 			memset(rt, 0, sizeof(*rt));
   1232 			sockaddr_in6_init(&rt->mf6c_origin, &ip6->ip6_src,
   1233 			    0, 0, 0);
   1234 			sockaddr_in6_init(&rt->mf6c_mcastgrp, &ip6->ip6_dst,
   1235 			    0, 0, 0);
   1236 			rt->mf6c_expire = UPCALL_EXPIRE;
   1237 			n6expire[hash]++;
   1238 			rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
   1239 
   1240 			/* link into table */
   1241 			rt->mf6c_next  = mf6ctable[hash];
   1242 			mf6ctable[hash] = rt;
   1243 			/* Add this entry to the end of the queue */
   1244 			rt->mf6c_stall = rte;
   1245 		} else {
   1246 			/* determine if q has overflowed */
   1247 			struct rtdetq **p;
   1248 			int npkts = 0;
   1249 
   1250 			for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next) {
   1251 				if (++npkts > MAX_UPQ6) {
   1252 					mrt6stat.mrt6s_upq_ovflw++;
   1253 					free(rte, M_MRTABLE);
   1254 					m_freem(mb0);
   1255 					splx(s);
   1256 					return 0;
   1257 				}
   1258 			}
   1259 
   1260 			/* Add this entry to the end of the queue */
   1261 			*p = rte;
   1262 		}
   1263 
   1264 		rte->next = NULL;
   1265 		rte->m = mb0;
   1266 		rte->ifp = ifp;
   1267 #ifdef UPCALL_TIMING
   1268 		rte->t = tp;
   1269 #endif
   1270 
   1271 		splx(s);
   1272 
   1273 		return 0;
   1274 	}
   1275 }
   1276 
   1277 /*
   1278  * Clean up cache entries if upcalls are not serviced
   1279  * Call from the Slow Timeout mechanism, every 0.25 seconds.
   1280  */
   1281 static void
   1282 expire_upcalls(void *unused)
   1283 {
   1284 	struct rtdetq *rte;
   1285 	struct mf6c *mfc, **nptr;
   1286 	int i;
   1287 
   1288 	/* XXX NOMPSAFE still need softnet_lock */
   1289 	mutex_enter(softnet_lock);
   1290 	KERNEL_LOCK(1, NULL);
   1291 
   1292 	for (i = 0; i < MF6CTBLSIZ; i++) {
   1293 		if (n6expire[i] == 0)
   1294 			continue;
   1295 		nptr = &mf6ctable[i];
   1296 		while ((mfc = *nptr) != NULL) {
   1297 			rte = mfc->mf6c_stall;
   1298 			/*
   1299 			 * Skip real cache entries
   1300 			 * Make sure it wasn't marked to not expire (shouldn't happen)
   1301 			 * If it expires now
   1302 			 */
   1303 			if (rte != NULL &&
   1304 			    mfc->mf6c_expire != 0 &&
   1305 			    --mfc->mf6c_expire == 0) {
   1306 #ifdef MRT6DEBUG
   1307 				if (mrt6debug & DEBUG_EXPIRE) {
   1308 					char ip6bufo[INET6_ADDRSTRLEN];
   1309 					char ip6bufm[INET6_ADDRSTRLEN];
   1310 					log(LOG_DEBUG,
   1311 					    "expire_upcalls: expiring (%s %s)\n",
   1312 					    IN6_PRINT(ip6bufo,
   1313 					    &mfc->mf6c_origin.sin6_addr),
   1314 					    IN6_PRINT(ip6bufm,
   1315 					    &mfc->mf6c_mcastgrp.sin6_addr));
   1316 				}
   1317 #endif
   1318 				/*
   1319 				 * drop all the packets
   1320 				 * free the mbuf with the pkt, if, timing info
   1321 				 */
   1322 				do {
   1323 					struct rtdetq *n = rte->next;
   1324 					m_freem(rte->m);
   1325 					free(rte, M_MRTABLE);
   1326 					rte = n;
   1327 				} while (rte != NULL);
   1328 				mrt6stat.mrt6s_cache_cleanups++;
   1329 				n6expire[i]--;
   1330 
   1331 				*nptr = mfc->mf6c_next;
   1332 				free(mfc, M_MRTABLE);
   1333 			} else {
   1334 				nptr = &mfc->mf6c_next;
   1335 			}
   1336 		}
   1337 	}
   1338 	callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
   1339 	    expire_upcalls, NULL);
   1340 
   1341 	KERNEL_UNLOCK_ONE(NULL);
   1342 	mutex_exit(softnet_lock);
   1343 }
   1344 
   1345 /*
   1346  * Macro to send packet on mif.  Since RSVP packets don't get counted on
   1347  * input, they shouldn't get counted on output, so statistics keeping is
   1348  * separate.
   1349  */
   1350 #define MC6_SEND(ip6, mifp, m) do {				\
   1351 	if ((mifp)->m6_flags & MIFF_REGISTER)			\
   1352 		register_send((ip6), (mifp), (m));		\
   1353 	else							\
   1354 		phyint_send((ip6), (mifp), (m));		\
   1355 } while (/*CONSTCOND*/ 0)
   1356 
   1357 /*
   1358  * Packet forwarding routine once entry in the cache is made
   1359  */
   1360 static int
   1361 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
   1362 {
   1363 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
   1364 	mifi_t mifi, iif;
   1365 	struct mif6 *mifp;
   1366 	int plen = m->m_pkthdr.len;
   1367 	struct in6_addr src0, dst0; /* copies for local work */
   1368 	u_int32_t iszone, idzone, oszone, odzone;
   1369 	int error = 0;
   1370 
   1371 	/*
   1372 	 * Don't forward if it didn't arrive from the parent mif
   1373 	 * for its origin.
   1374 	 */
   1375 	mifi = rt->mf6c_parent;
   1376 	if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
   1377 		/* came in the wrong interface */
   1378 #ifdef MRT6DEBUG
   1379 		if (mrt6debug & DEBUG_FORWARD)
   1380 			log(LOG_DEBUG,
   1381 			    "wrong if: ifid %d mifi %d mififid %x\n",
   1382 			    ifp->if_index, mifi,
   1383 			    mif6table[mifi].m6_ifp ?
   1384 			    mif6table[mifi].m6_ifp->if_index : -1);
   1385 #endif
   1386 		mrt6stat.mrt6s_wrong_if++;
   1387 		rt->mf6c_wrong_if++;
   1388 
   1389 		/*
   1390 		 * If we are doing PIM processing, and we are forwarding
   1391 		 * packets on this interface, send a message to the
   1392 		 * routing daemon.
   1393 		 */
   1394 		/* have to make sure this is a valid mif */
   1395 		if (mifi < nummifs && mif6table[mifi].m6_ifp) {
   1396 			if (pim6 && (m->m_flags & M_LOOP) == 0) {
   1397 				/*
   1398 				 * Check the M_LOOP flag to avoid an
   1399 				 * unnecessary PIM assert.
   1400 				 * XXX: M_LOOP is an ad-hoc hack...
   1401 				 */
   1402 				struct sockaddr_in6 sin6;
   1403 
   1404 				struct mbuf *mm;
   1405 				struct mrt6msg *im;
   1406 				struct omrt6msg *oim;
   1407 
   1408 				mm = m_copym(m, 0, sizeof(struct ip6_hdr), M_DONTWAIT);
   1409 				if (mm && M_UNWRITABLE(mm, sizeof(struct ip6_hdr)))
   1410 					mm = m_pullup(mm, sizeof(struct ip6_hdr));
   1411 				if (mm == NULL)
   1412 					return ENOBUFS;
   1413 
   1414 				oim = NULL;
   1415 				im = NULL;
   1416 				switch (ip6_mrouter_ver) {
   1417 				case MRT6_OINIT:
   1418 					oim = mtod(mm, struct omrt6msg *);
   1419 					oim->im6_msgtype = MRT6MSG_WRONGMIF;
   1420 					oim->im6_mbz = 0;
   1421 					break;
   1422 				case MRT6_INIT:
   1423 					im = mtod(mm, struct mrt6msg *);
   1424 					im->im6_msgtype = MRT6MSG_WRONGMIF;
   1425 					im->im6_mbz = 0;
   1426 					break;
   1427 				default:
   1428 					m_freem(mm);
   1429 					return EINVAL;
   1430 				}
   1431 
   1432 				for (mifp = mif6table, iif = 0;
   1433 				     iif < nummifs && mifp &&
   1434 					     mifp->m6_ifp != ifp;
   1435 				     mifp++, iif++)
   1436 					;
   1437 
   1438 				memset(&sin6, 0, sizeof(sin6));
   1439 				sin6.sin6_len = sizeof(sin6);
   1440 				sin6.sin6_family = AF_INET6;
   1441 				switch (ip6_mrouter_ver) {
   1442 				case MRT6_OINIT:
   1443 					oim->im6_mif = iif;
   1444 					sin6.sin6_addr = oim->im6_src;
   1445 					break;
   1446 				case MRT6_INIT:
   1447 					im->im6_mif = iif;
   1448 					sin6.sin6_addr = im->im6_src;
   1449 					break;
   1450 				}
   1451 
   1452 				mrt6stat.mrt6s_upcalls++;
   1453 
   1454 				if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
   1455 #ifdef MRT6DEBUG
   1456 					if (mrt6debug)
   1457 						log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
   1458 #endif
   1459 					++mrt6stat.mrt6s_upq_sockfull;
   1460 					return ENOBUFS;
   1461 				}
   1462 			}
   1463 		}
   1464 
   1465 		return 0;
   1466 	}
   1467 
   1468 	/* If I sourced this packet, it counts as output, else it was input. */
   1469 	if (m->m_pkthdr.rcvif_index == 0) {
   1470 		/* XXX: is rcvif really NULL when output?? */
   1471 		mif6table[mifi].m6_pkt_out++;
   1472 		mif6table[mifi].m6_bytes_out += plen;
   1473 	} else {
   1474 		mif6table[mifi].m6_pkt_in++;
   1475 		mif6table[mifi].m6_bytes_in += plen;
   1476 	}
   1477 	rt->mf6c_pkt_cnt++;
   1478 	rt->mf6c_byte_cnt += plen;
   1479 
   1480 	/*
   1481 	 * For each mif, forward a copy of the packet if there are group
   1482 	 * members downstream on the interface.
   1483 	 */
   1484 	src0 = ip6->ip6_src;
   1485 	dst0 = ip6->ip6_dst;
   1486 	if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
   1487 	    (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
   1488 		IP6_STATINC(IP6_STAT_BADSCOPE);
   1489 		return error;
   1490 	}
   1491 	for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
   1492 		if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
   1493 			if (mif6table[mifi].m6_ifp == NULL)
   1494 				continue;
   1495 			/*
   1496 			 * check if the outgoing packet is going to break
   1497 			 * a scope boundary.
   1498 			 * XXX: For packets through PIM register tunnel
   1499 			 * interface, we believe the routing daemon.
   1500 			 */
   1501 			if ((mif6table[rt->mf6c_parent].m6_flags &
   1502 			     MIFF_REGISTER) == 0 &&
   1503 			    (mif6table[mifi].m6_flags & MIFF_REGISTER) == 0) {
   1504 				if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
   1505 				    &oszone) ||
   1506 				    in6_setscope(&dst0, mif6table[mifi].m6_ifp,
   1507 				    &odzone) ||
   1508 				    iszone != oszone || idzone != odzone) {
   1509 					IP6_STATINC(IP6_STAT_BADSCOPE);
   1510 					continue;
   1511 				}
   1512 			}
   1513 
   1514 			mifp->m6_pkt_out++;
   1515 			mifp->m6_bytes_out += plen;
   1516 			MC6_SEND(ip6, mifp, m);
   1517 		}
   1518 	}
   1519 
   1520 	return 0;
   1521 }
   1522 
   1523 static void
   1524 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
   1525 {
   1526 	struct mbuf *mb_copy;
   1527 	struct ifnet *ifp = mifp->m6_ifp;
   1528 	int error __mrt6debugused = 0;
   1529 	int s;
   1530 	static struct route ro;
   1531 	bool ingroup;
   1532 	struct sockaddr_in6 dst6;
   1533 
   1534 	s = splsoftnet();
   1535 
   1536 	/*
   1537 	 * Make a new reference to the packet; make sure that
   1538 	 * the IPv6 header is actually copied, not just referenced,
   1539 	 * so that ip6_output() only scribbles on the copy.
   1540 	 */
   1541 	mb_copy = m_copypacket(m, M_DONTWAIT);
   1542 	if (mb_copy && M_UNWRITABLE(mb_copy, sizeof(struct ip6_hdr)))
   1543 		mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
   1544 	if (mb_copy == NULL) {
   1545 		splx(s);
   1546 		return;
   1547 	}
   1548 
   1549 	/* set MCAST flag to the outgoing packet */
   1550 	mb_copy->m_flags |= M_MCAST;
   1551 
   1552 	/*
   1553 	 * If we sourced the packet, call ip6_output since we may divide
   1554 	 * the packet into fragments when the packet is too big for the
   1555 	 * outgoing interface.
   1556 	 * Otherwise, we can simply send the packet to the interface
   1557 	 * sending queue.
   1558 	 */
   1559 	if (m->m_pkthdr.rcvif_index == 0) {
   1560 		struct ip6_moptions im6o;
   1561 
   1562 		im6o.im6o_multicast_if_index = if_get_index(ifp);
   1563 		/* XXX: ip6_output will override ip6->ip6_hlim */
   1564 		im6o.im6o_multicast_hlim = ip6->ip6_hlim;
   1565 		im6o.im6o_multicast_loop = 1;
   1566 		error = ip6_output(mb_copy, NULL, &ro, IPV6_FORWARDING,
   1567 		    &im6o, NULL, NULL);
   1568 
   1569 #ifdef MRT6DEBUG
   1570 		if (mrt6debug & DEBUG_XMIT)
   1571 			log(LOG_DEBUG, "phyint_send on mif %td err %d\n",
   1572 			    mifp - mif6table, error);
   1573 #endif
   1574 		splx(s);
   1575 		return;
   1576 	}
   1577 
   1578 	/*
   1579 	 * If we belong to the destination multicast group
   1580 	 * on the outgoing interface, loop back a copy.
   1581 	 */
   1582 	/*
   1583 	 * Does not have to check source info, as it's already covered by
   1584 	 * ip6_input
   1585 	 */
   1586 	sockaddr_in6_init(&dst6, &ip6->ip6_dst, 0, 0, 0);
   1587 
   1588 	ingroup = in6_multi_group(&ip6->ip6_dst, ifp);
   1589 	if (ingroup) {
   1590 		ip6_mloopback(ifp, m,
   1591 		    satocsin6(rtcache_getdst(&ro)));
   1592 	}
   1593 
   1594 	/*
   1595 	 * Put the packet into the sending queue of the outgoing interface
   1596 	 * if it would fit in the MTU of the interface.
   1597 	 */
   1598 	if (mb_copy->m_pkthdr.len <= ifp->if_mtu || ifp->if_mtu < IPV6_MMTU) {
   1599 		error = ip6_if_output(ifp, ifp, mb_copy, &dst6, NULL);
   1600 #ifdef MRT6DEBUG
   1601 		if (mrt6debug & DEBUG_XMIT)
   1602 			log(LOG_DEBUG, "phyint_send on mif %td err %d\n",
   1603 			    mifp - mif6table, error);
   1604 #endif
   1605 	} else {
   1606 		/*
   1607 		 * pMTU discovery is intentionally disabled by default, since
   1608 		 * various routers may notify pMTU in multicast, which can be
   1609 		 * a DDoS to a router.
   1610 		 */
   1611 		if (ip6_mcast_pmtu) {
   1612 			icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0,
   1613 			    ifp->if_mtu);
   1614 		} else {
   1615 			/* simply discard the packet */
   1616 #ifdef MRT6DEBUG
   1617 			if (mrt6debug & DEBUG_XMIT) {
   1618 				char ip6bufs[INET6_ADDRSTRLEN];
   1619 				char ip6bufd[INET6_ADDRSTRLEN];
   1620 				log(LOG_DEBUG,
   1621 				    "phyint_send: packet too big on %s o %s g %s"
   1622 				    " size %d(discarded)\n",
   1623 				    if_name(ifp),
   1624 				    IN6_PRINT(ip6bufs, &ip6->ip6_src),
   1625 				    IN6_PRINT(ip6bufd, &ip6->ip6_dst),
   1626 				    mb_copy->m_pkthdr.len);
   1627 			}
   1628 #endif
   1629 			m_freem(mb_copy);
   1630 		}
   1631 	}
   1632 
   1633 	splx(s);
   1634 }
   1635 
   1636 static int
   1637 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
   1638 {
   1639 	struct mbuf *mm;
   1640 	int i, len = m->m_pkthdr.len;
   1641 	struct sockaddr_in6 sin6;
   1642 	struct mrt6msg *im6;
   1643 
   1644 #ifdef MRT6DEBUG
   1645 	if (mrt6debug) {
   1646 		char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
   1647 		log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
   1648 		    IN6_PRINT(ip6bufs, &ip6->ip6_src),
   1649 		    IN6_PRINT(ip6bufd, &ip6->ip6_dst));
   1650 	}
   1651 #endif
   1652 	PIM6_STATINC(PIM6_STAT_SND_REGISTERS);
   1653 
   1654 	/* Make a copy of the packet to send to the user level process */
   1655 	MGETHDR(mm, M_DONTWAIT, MT_HEADER);
   1656 	if (mm == NULL)
   1657 		return ENOBUFS;
   1658 	mm->m_data += max_linkhdr;
   1659 	mm->m_len = sizeof(struct ip6_hdr);
   1660 
   1661 	if ((mm->m_next = m_copypacket(m, M_DONTWAIT)) == NULL) {
   1662 		m_freem(mm);
   1663 		return ENOBUFS;
   1664 	}
   1665 	i = MHLEN - M_LEADINGSPACE(mm);
   1666 	if (i > len)
   1667 		i = len;
   1668 	mm = m_pullup(mm, i);
   1669 	if (mm == NULL)
   1670 		return ENOBUFS;
   1671 	mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
   1672 
   1673 	/*
   1674 	 * Send message to routing daemon
   1675 	 */
   1676 	sockaddr_in6_init(&sin6, &ip6->ip6_src, 0, 0, 0);
   1677 
   1678 	im6 = mtod(mm, struct mrt6msg *);
   1679 	im6->im6_msgtype = MRT6MSG_WHOLEPKT;
   1680 	im6->im6_mbz = 0;
   1681 	im6->im6_mif = mif - mif6table;
   1682 
   1683 	/* iif info is not given for reg. encap.n */
   1684 	mrt6stat.mrt6s_upcalls++;
   1685 
   1686 	if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
   1687 #ifdef MRT6DEBUG
   1688 		if (mrt6debug)
   1689 			log(LOG_WARNING,
   1690 			    "register_send: ip6_mrouter socket queue full\n");
   1691 #endif
   1692 		++mrt6stat.mrt6s_upq_sockfull;
   1693 		return ENOBUFS;
   1694 	}
   1695 
   1696 	return 0;
   1697 }
   1698 
   1699 /*
   1700  * PIM sparse mode hook. Receives the pim control messages, and passes them up
   1701  * to the listening socket, using rip6_input.
   1702  *
   1703  * The only message processed is the REGISTER pim message; the pim header
   1704  * is stripped off, and the inner packet is passed to register_mforward.
   1705  */
   1706 int
   1707 pim6_input(struct mbuf **mp, int *offp, int proto)
   1708 {
   1709 	struct pim *pim;
   1710 	struct ip6_hdr *ip6 __mrt6debugused;
   1711 	int pimlen;
   1712 	struct mbuf *m = *mp;
   1713 	int minlen;
   1714 	int off = *offp;
   1715 
   1716 	PIM6_STATINC(PIM6_STAT_RCV_TOTAL);
   1717 
   1718 	ip6 = mtod(m, struct ip6_hdr *);
   1719 	pimlen = m->m_pkthdr.len - off;
   1720 
   1721 	/*
   1722 	 * Validate lengths
   1723 	 */
   1724 	if (pimlen < PIM_MINLEN) {
   1725 		PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
   1726 #ifdef MRT6DEBUG
   1727 		if (mrt6debug & DEBUG_PIM)
   1728 			log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
   1729 #endif
   1730 		m_freem(m);
   1731 		return IPPROTO_DONE;
   1732 	}
   1733 
   1734 	/*
   1735 	 * If the packet is at least as big as a REGISTER, go ahead
   1736 	 * and grab the PIM REGISTER header size, to avoid another
   1737 	 * possible m_pullup() later.
   1738 	 *
   1739 	 * PIM_MINLEN       == pimhdr + u_int32 == 8
   1740 	 * PIM6_REG_MINLEN   == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
   1741 	 */
   1742 	minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
   1743 
   1744 	/*
   1745 	 * Make sure that the IP6 and PIM headers in contiguous memory, and
   1746 	 * possibly the PIM REGISTER header
   1747 	 */
   1748 	IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
   1749 	if (pim == NULL) {
   1750 		PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
   1751 		return IPPROTO_DONE;
   1752 	}
   1753 
   1754 	/* PIM version check */
   1755 	if (pim->pim_ver != PIM_VERSION) {
   1756 		PIM6_STATINC(PIM6_STAT_RCV_BADVERSION);
   1757 #ifdef MRT6DEBUG
   1758 		log(LOG_ERR,
   1759 		    "pim6_input: incorrect version %d, expecting %d\n",
   1760 		    pim->pim_ver, PIM_VERSION);
   1761 #endif
   1762 		m_freem(m);
   1763 		return IPPROTO_DONE;
   1764 	}
   1765 
   1766 #define PIM6_CHECKSUM
   1767 #ifdef PIM6_CHECKSUM
   1768 	{
   1769 		int cksumlen;
   1770 
   1771 		/*
   1772 		 * Validate checksum.
   1773 		 * If PIM REGISTER, exclude the data packet
   1774 		 */
   1775 		if (pim->pim_type == PIM_REGISTER)
   1776 			cksumlen = PIM_MINLEN;
   1777 		else
   1778 			cksumlen = pimlen;
   1779 
   1780 		if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
   1781 			PIM6_STATINC(PIM6_STAT_RCV_BADSUM);
   1782 #ifdef MRT6DEBUG
   1783 			if (mrt6debug & DEBUG_PIM)
   1784 				log(LOG_DEBUG,
   1785 				    "pim6_input: invalid checksum\n");
   1786 #endif
   1787 			m_freem(m);
   1788 			return IPPROTO_DONE;
   1789 		}
   1790 	}
   1791 #endif /* PIM_CHECKSUM */
   1792 
   1793 	if (pim->pim_type == PIM_REGISTER) {
   1794 		/*
   1795 		 * since this is a REGISTER, we'll make a copy of the register
   1796 		 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
   1797 		 * routing daemon.
   1798 		 */
   1799 		static const struct sockaddr_in6 dst = {
   1800 			.sin6_len = sizeof(dst),
   1801 			.sin6_family = AF_INET6,
   1802 		};
   1803 
   1804 		struct mbuf *mcp;
   1805 		struct ip6_hdr *eip6;
   1806 		u_int32_t *reghdr;
   1807 
   1808 		PIM6_STATINC(PIM6_STAT_RCV_REGISTERS);
   1809 
   1810 		if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
   1811 #ifdef MRT6DEBUG
   1812 			if (mrt6debug & DEBUG_PIM)
   1813 				log(LOG_DEBUG,
   1814 				    "pim6_input: register mif not set: %d\n",
   1815 				    reg_mif_num);
   1816 #endif
   1817 			m_freem(m);
   1818 			return IPPROTO_DONE;
   1819 		}
   1820 
   1821 		reghdr = (u_int32_t *)(pim + 1);
   1822 
   1823 		if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
   1824 			goto pim6_input_to_daemon;
   1825 
   1826 		/*
   1827 		 * Validate length
   1828 		 */
   1829 		if (pimlen < PIM6_REG_MINLEN) {
   1830 #ifdef MRT6DEBUG
   1831 			char ip6buf[INET6_ADDRSTRLEN];
   1832 			log(LOG_ERR,
   1833 			    "pim6_input: register packet size too "
   1834 			    "small %d from %s\n",
   1835 			    pimlen, IN6_PRINT(ip6buf, &ip6->ip6_src));
   1836 #endif
   1837 			PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
   1838 			PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
   1839 			m_freem(m);
   1840 			return IPPROTO_DONE;
   1841 		}
   1842 
   1843 		eip6 = (struct ip6_hdr *)(reghdr + 1);
   1844 #ifdef MRT6DEBUG
   1845 		if (mrt6debug & DEBUG_PIM) {
   1846 			char ip6bufs[INET6_ADDRSTRLEN];
   1847 			char ip6bufd[INET6_ADDRSTRLEN];
   1848 			log(LOG_DEBUG,
   1849 			    "pim6_input[register], eip6: %s -> %s, "
   1850 			    "eip6 plen %d\n",
   1851 			    IN6_PRINT(ip6bufs, &eip6->ip6_src),
   1852 			    IN6_PRINT(ip6bufd, &eip6->ip6_dst),
   1853 			    ntohs(eip6->ip6_plen));
   1854 		}
   1855 #endif
   1856 
   1857 		/* verify the version number of the inner packet */
   1858 		if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
   1859 			PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
   1860 #ifdef MRT6DEBUG
   1861 			log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
   1862 			    "of the inner packet\n",
   1863 			    (eip6->ip6_vfc & IPV6_VERSION));
   1864 #endif
   1865 			m_freem(m);
   1866 			return IPPROTO_DONE;
   1867 		}
   1868 
   1869 		/* verify the inner packet is destined to a mcast group */
   1870 		if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
   1871 			PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
   1872 #ifdef MRT6DEBUG
   1873 			if (mrt6debug & DEBUG_PIM) {
   1874 				char ip6buf[INET6_ADDRSTRLEN];
   1875 				log(LOG_DEBUG,
   1876 				    "pim6_input: inner packet of register "
   1877 				    "is not multicast %s\n",
   1878 				    IN6_PRINT(ip6buf, &eip6->ip6_dst));
   1879 			}
   1880 #endif
   1881 			m_freem(m);
   1882 			return IPPROTO_DONE;
   1883 		}
   1884 
   1885 		/*
   1886 		 * make a copy of the whole header to pass to the daemon later.
   1887 		 */
   1888 		mcp = m_copym(m, 0, off + PIM6_REG_MINLEN, M_DONTWAIT);
   1889 		if (mcp == NULL) {
   1890 #ifdef MRT6DEBUG
   1891 			log(LOG_ERR,
   1892 			    "pim6_input: pim register: "
   1893 			    "could not copy register head\n");
   1894 #endif
   1895 			m_freem(m);
   1896 			return IPPROTO_DONE;
   1897 		}
   1898 
   1899 		/*
   1900 		 * forward the inner ip6 packet; point m_data at the inner ip6.
   1901 		 */
   1902 		m_adj(m, off + PIM_MINLEN);
   1903 #ifdef MRT6DEBUG
   1904 		if (mrt6debug & DEBUG_PIM) {
   1905 			char ip6bufs[INET6_ADDRSTRLEN];
   1906 			char ip6bufd[INET6_ADDRSTRLEN];
   1907 			log(LOG_DEBUG,
   1908 			    "pim6_input: forwarding decapsulated register: "
   1909 			    "src %s, dst %s, mif %d\n",
   1910 			    IN6_PRINT(ip6bufs, &eip6->ip6_src),
   1911 			    IN6_PRINT(ip6bufd, &eip6->ip6_dst),
   1912 			    reg_mif_num);
   1913 		}
   1914 #endif
   1915 
   1916 		looutput(mif6table[reg_mif_num].m6_ifp, m, sin6tocsa(&dst),
   1917 		    NULL);
   1918 
   1919 		/* prepare the register head to send to the mrouting daemon */
   1920 		m = mcp;
   1921 	}
   1922 
   1923 	/*
   1924 	 * Pass the PIM message up to the daemon; if it is a register message
   1925 	 * pass the 'head' only up to the daemon. This includes the
   1926 	 * encapsulator ip6 header, pim header, register header and the
   1927 	 * encapsulated ip6 header.
   1928 	 */
   1929 pim6_input_to_daemon:
   1930 	/*
   1931 	 * Currently, rip6_input() is always called holding softnet_lock
   1932 	 * by ipintr()(!NET_MPSAFE) or PR_INPUT_WRAP()(NET_MPSAFE).
   1933 	 */
   1934 	KASSERT(mutex_owned(softnet_lock));
   1935 	rip6_input(&m, offp, proto);
   1936 	return IPPROTO_DONE;
   1937 }
   1938 
   1939 static int
   1940 sysctl_net_inet6_pim6_stats(SYSCTLFN_ARGS)
   1941 {
   1942 
   1943 	return (NETSTAT_SYSCTL(pim6stat_percpu, PIM6_NSTATS));
   1944 }
   1945 
   1946 static void
   1947 sysctl_net_inet6_pim6_setup(struct sysctllog **clog)
   1948 {
   1949 
   1950 	sysctl_createv(clog, 0, NULL, NULL,
   1951 		       CTLFLAG_PERMANENT,
   1952 		       CTLTYPE_NODE, "inet6", NULL,
   1953 		       NULL, 0, NULL, 0,
   1954 		       CTL_NET, PF_INET6, CTL_EOL);
   1955 	sysctl_createv(clog, 0, NULL, NULL,
   1956 		       CTLFLAG_PERMANENT,
   1957 		       CTLTYPE_NODE, "pim6",
   1958 		       SYSCTL_DESCR("PIMv6 settings"),
   1959 		       NULL, 0, NULL, 0,
   1960 		       CTL_NET, PF_INET6, IPPROTO_PIM, CTL_EOL);
   1961 
   1962 	sysctl_createv(clog, 0, NULL, NULL,
   1963 		       CTLFLAG_PERMANENT,
   1964 		       CTLTYPE_STRUCT, "stats",
   1965 		       SYSCTL_DESCR("PIMv6 statistics"),
   1966 		       sysctl_net_inet6_pim6_stats, 0, NULL, 0,
   1967 		       CTL_NET, PF_INET6, IPPROTO_PIM, PIM6CTL_STATS,
   1968 		       CTL_EOL);
   1969 }
   1970