No empty .Rs/.Re
[netbsd-mini2440.git] / sys / netinet6 / ip6_mroute.c
blob10defe59afa1fcea4312f462cb5b4b8732c5fafc
1 /* $NetBSD: ip6_mroute.c,v 1.97 2009/03/18 16:00:23 cegger Exp $ */
2 /* $KAME: ip6_mroute.c,v 1.49 2001/07/25 09:21:18 jinmei Exp $ */
4 /*
5 * Copyright (C) 1998 WIDE Project.
6 * All rights reserved.
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.
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.
33 /* BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp */
36 * Copyright (c) 1992, 1993
37 * The Regents of the University of California. All rights reserved.
39 * This code is derived from software contributed to Berkeley by
40 * Stephen Deering of Stanford University.
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.
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.
66 * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93
70 * Copyright (c) 1989 Stephen Deering
72 * This code is derived from software contributed to Berkeley by
73 * Stephen Deering of Stanford University.
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.
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.
103 * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93
107 * IP multicast forwarding procedures
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
116 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
119 #include <sys/cdefs.h>
120 __KERNEL_RCSID(0, "$NetBSD: ip6_mroute.c,v 1.97 2009/03/18 16:00:23 cegger Exp $");
122 #include "opt_inet.h"
123 #include "opt_mrouting.h"
125 #include <sys/param.h>
126 #include <sys/systm.h>
127 #include <sys/callout.h>
128 #include <sys/mbuf.h>
129 #include <sys/socket.h>
130 #include <sys/socketvar.h>
131 #include <sys/sockio.h>
132 #include <sys/protosw.h>
133 #include <sys/errno.h>
134 #include <sys/time.h>
135 #include <sys/kernel.h>
136 #include <sys/ioctl.h>
137 #include <sys/sysctl.h>
138 #include <sys/syslog.h>
140 #include <net/if.h>
141 #include <net/route.h>
142 #include <net/raw_cb.h>
143 #include <net/net_stats.h>
145 #include <netinet/in.h>
146 #include <netinet/in_var.h>
147 #include <netinet/icmp6.h>
149 #include <netinet/ip6.h>
150 #include <netinet6/ip6_var.h>
151 #include <netinet6/ip6_private.h>
152 #include <netinet6/ip6_mroute.h>
153 #include <netinet6/scope6_var.h>
154 #include <netinet6/pim6.h>
155 #include <netinet6/pim6_var.h>
156 #include <netinet6/nd6.h>
158 #include <net/net_osdep.h>
160 static int ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
161 static void phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
163 static int set_pim6(int *);
164 static int socket_send(struct socket *, struct mbuf *,
165 struct sockaddr_in6 *);
166 static int register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
169 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static,
170 * except for netstat or debugging purposes.
172 struct socket *ip6_mrouter = NULL;
173 int ip6_mrouter_ver = 0;
174 int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */
175 struct mrt6stat mrt6stat;
177 #define NO_RTE_FOUND 0x1
178 #define RTE_FOUND 0x2
180 struct mf6c *mf6ctable[MF6CTBLSIZ];
181 u_char n6expire[MF6CTBLSIZ];
182 struct mif6 mif6table[MAXMIFS];
183 #ifdef MRT6DEBUG
184 u_int mrt6debug = 0; /* debug level */
185 #define DEBUG_MFC 0x02
186 #define DEBUG_FORWARD 0x04
187 #define DEBUG_EXPIRE 0x08
188 #define DEBUG_XMIT 0x10
189 #define DEBUG_REG 0x20
190 #define DEBUG_PIM 0x40
191 #endif
193 static void expire_upcalls(void *);
194 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */
195 #define UPCALL_EXPIRE 6 /* number of timeouts */
197 #ifdef INET
198 #ifdef MROUTING
199 extern struct socket *ip_mrouter;
200 #endif
201 #endif
204 * 'Interfaces' associated with decapsulator (so we can tell
205 * packets that went through it from ones that get reflected
206 * by a broken gateway). These interfaces are never linked into
207 * the system ifnet list & no routes point to them. I.e., packets
208 * can't be sent this way. They only exist as a placeholder for
209 * multicast source verification.
211 struct ifnet multicast_register_if6;
213 #define ENCAP_HOPS 64
216 * Private variables.
218 static mifi_t nummifs = 0;
219 static mifi_t reg_mif_num = (mifi_t)-1;
221 static percpu_t *pim6stat_percpu;
223 #define PIM6_STATINC(x) _NET_STATINC(pim6stat_percpu, x)
225 static int pim6;
228 * Hash function for a source, group entry
230 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
231 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
232 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
233 (g).s6_addr32[2] ^ (g).s6_addr32[3])
236 * Find a route for a given origin IPv6 address and Multicast group address.
237 * Quality of service parameter to be added in the future!!!
240 #define MF6CFIND(o, g, rt) do { \
241 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
242 rt = NULL; \
243 mrt6stat.mrt6s_mfc_lookups++; \
244 while (_rt) { \
245 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
246 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
247 (_rt->mf6c_stall == NULL)) { \
248 rt = _rt; \
249 break; \
251 _rt = _rt->mf6c_next; \
253 if (rt == NULL) { \
254 mrt6stat.mrt6s_mfc_misses++; \
256 } while (/*CONSTCOND*/ 0)
259 * Macros to compute elapsed time efficiently
260 * Borrowed from Van Jacobson's scheduling code
262 #define TV_DELTA(a, b, delta) do { \
263 int xxs; \
265 delta = (a).tv_usec - (b).tv_usec; \
266 if ((xxs = (a).tv_sec - (b).tv_sec)) { \
267 switch (xxs) { \
268 case 2: \
269 delta += 1000000; \
270 /* FALLTHROUGH */ \
271 case 1: \
272 delta += 1000000; \
273 break; \
274 default: \
275 delta += (1000000 * xxs); \
278 } while (/*CONSTCOND*/ 0)
280 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
281 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
283 #ifdef UPCALL_TIMING
284 #define UPCALL_MAX 50
285 u_long upcall_data[UPCALL_MAX + 1];
286 static void collate();
287 #endif /* UPCALL_TIMING */
289 static int get_sg_cnt(struct sioc_sg_req6 *);
290 static int get_mif6_cnt(struct sioc_mif_req6 *);
291 static int ip6_mrouter_init(struct socket *, int, int);
292 static int add_m6if(struct mif6ctl *);
293 static int del_m6if(mifi_t *);
294 static int add_m6fc(struct mf6cctl *);
295 static int del_m6fc(struct mf6cctl *);
296 static void sysctl_net_inet6_pim6_setup(struct sysctllog **);
298 static callout_t expire_upcalls_ch;
300 void
301 pim6_init(void)
304 sysctl_net_inet6_pim6_setup(NULL);
305 pim6stat_percpu = percpu_alloc(sizeof(uint64_t) * PIM6_NSTATS);
309 * Handle MRT setsockopt commands to modify the multicast routing tables.
312 ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
314 int error, optval;
315 struct mif6ctl mifc;
316 struct mf6cctl mfcc;
317 mifi_t mifi;
319 if (sopt->sopt_name != MRT6_INIT && so != ip6_mrouter)
320 return (EACCES);
322 error = 0;
324 switch (sopt->sopt_name) {
325 #ifdef MRT6_OINIT
326 case MRT6_OINIT:
327 #endif
328 case MRT6_INIT:
329 error = sockopt_getint(sopt, &optval);
330 if (error)
331 break;
332 return (ip6_mrouter_init(so, optval, sopt->sopt_name));
333 case MRT6_DONE:
334 return (ip6_mrouter_done());
335 case MRT6_ADD_MIF:
336 error = sockopt_get(sopt, &mifc, sizeof(mifc));
337 if (error)
338 break;
339 return (add_m6if(&mifc));
340 case MRT6_DEL_MIF:
341 error = sockopt_get(sopt, &mifi, sizeof(mifi));
342 if (error)
343 break;
344 return (del_m6if(&mifi));
345 case MRT6_ADD_MFC:
346 error = sockopt_get(sopt, &mfcc, sizeof(mfcc));
347 if (error)
348 break;
349 return (add_m6fc(&mfcc));
350 case MRT6_DEL_MFC:
351 error = sockopt_get(sopt, &mfcc, sizeof(mfcc));
352 if (error)
353 break;
354 return (del_m6fc(&mfcc));
355 case MRT6_PIM:
356 error = sockopt_getint(sopt, &optval);
357 if (error)
358 break;
359 return (set_pim6(&optval));
360 default:
361 error = EOPNOTSUPP;
364 return (error);
368 * Handle MRT getsockopt commands
371 ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
373 int error;
375 if (so != ip6_mrouter) return EACCES;
377 error = 0;
379 switch (sopt->sopt_name) {
380 case MRT6_PIM:
381 error = sockopt_set(sopt, &pim6, sizeof(pim6));
382 break;
383 default:
384 error = EOPNOTSUPP;
385 break;
388 return (error);
392 * Handle ioctl commands to obtain information from the cache
395 mrt6_ioctl(u_long cmd, void *data)
398 switch (cmd) {
399 case SIOCGETSGCNT_IN6:
400 return (get_sg_cnt((struct sioc_sg_req6 *)data));
401 case SIOCGETMIFCNT_IN6:
402 return (get_mif6_cnt((struct sioc_mif_req6 *)data));
403 default:
404 return (EINVAL);
409 * returns the packet, byte, rpf-failure count for the source group provided
411 static int
412 get_sg_cnt(struct sioc_sg_req6 *req)
414 struct mf6c *rt;
415 int s;
417 s = splsoftnet();
418 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
419 splx(s);
420 if (rt != NULL) {
421 req->pktcnt = rt->mf6c_pkt_cnt;
422 req->bytecnt = rt->mf6c_byte_cnt;
423 req->wrong_if = rt->mf6c_wrong_if;
424 } else
425 return (ESRCH);
426 #if 0
427 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff;
428 #endif
430 return 0;
434 * returns the input and output packet and byte counts on the mif provided
436 static int
437 get_mif6_cnt(struct sioc_mif_req6 *req)
439 mifi_t mifi = req->mifi;
441 if (mifi >= nummifs)
442 return EINVAL;
444 req->icount = mif6table[mifi].m6_pkt_in;
445 req->ocount = mif6table[mifi].m6_pkt_out;
446 req->ibytes = mif6table[mifi].m6_bytes_in;
447 req->obytes = mif6table[mifi].m6_bytes_out;
449 return 0;
452 static int
453 set_pim6(int *i)
455 if ((*i != 1) && (*i != 0))
456 return EINVAL;
458 pim6 = *i;
460 return 0;
464 * Enable multicast routing
466 static int
467 ip6_mrouter_init(struct socket *so, int v, int cmd)
469 #ifdef MRT6DEBUG
470 if (mrt6debug)
471 log(LOG_DEBUG,
472 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
473 so->so_type, so->so_proto->pr_protocol);
474 #endif
476 if (so->so_type != SOCK_RAW ||
477 so->so_proto->pr_protocol != IPPROTO_ICMPV6)
478 return (EOPNOTSUPP);
480 if (v != 1)
481 return (ENOPROTOOPT);
483 if (ip6_mrouter != NULL)
484 return (EADDRINUSE);
486 ip6_mrouter = so;
487 ip6_mrouter_ver = cmd;
489 memset((void *)mf6ctable, 0, sizeof(mf6ctable));
490 memset((void *)n6expire, 0, sizeof(n6expire));
492 pim6 = 0;/* used for stubbing out/in pim stuff */
494 callout_init(&expire_upcalls_ch, CALLOUT_MPSAFE);
495 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
496 expire_upcalls, NULL);
498 #ifdef MRT6DEBUG
499 if (mrt6debug)
500 log(LOG_DEBUG, "ip6_mrouter_init\n");
501 #endif
503 return 0;
507 * Disable multicast routing
510 ip6_mrouter_done(void)
512 mifi_t mifi;
513 int i;
514 struct ifnet *ifp;
515 struct in6_ifreq ifr;
516 struct mf6c *rt;
517 struct rtdetq *rte;
518 int s;
520 s = splsoftnet();
523 * For each phyint in use, disable promiscuous reception of all IPv6
524 * multicasts.
526 #ifdef INET
527 #ifdef MROUTING
529 * If there is still IPv4 multicast routing daemon,
530 * we remain interfaces to receive all muliticasted packets.
531 * XXX: there may be an interface in which the IPv4 multicast
532 * daemon is not interested...
534 if (!ip_mrouter)
535 #endif
536 #endif
538 for (mifi = 0; mifi < nummifs; mifi++) {
539 if (mif6table[mifi].m6_ifp &&
540 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
541 ifr.ifr_addr.sin6_family = AF_INET6;
542 ifr.ifr_addr.sin6_addr= in6addr_any;
543 ifp = mif6table[mifi].m6_ifp;
544 (*ifp->if_ioctl)(ifp, SIOCDELMULTI, &ifr);
548 #ifdef notyet
549 memset((void *)qtable, 0, sizeof(qtable));
550 memset((void *)tbftable, 0, sizeof(tbftable));
551 #endif
552 memset((void *)mif6table, 0, sizeof(mif6table));
553 nummifs = 0;
555 pim6 = 0; /* used to stub out/in pim specific code */
557 callout_stop(&expire_upcalls_ch);
560 * Free all multicast forwarding cache entries.
562 for (i = 0; i < MF6CTBLSIZ; i++) {
563 rt = mf6ctable[i];
564 while (rt) {
565 struct mf6c *frt;
567 for (rte = rt->mf6c_stall; rte != NULL; ) {
568 struct rtdetq *n = rte->next;
570 m_free(rte->m);
571 free(rte, M_MRTABLE);
572 rte = n;
574 frt = rt;
575 rt = rt->mf6c_next;
576 free(frt, M_MRTABLE);
580 memset((void *)mf6ctable, 0, sizeof(mf6ctable));
583 * Reset register interface
585 if (reg_mif_num != (mifi_t)-1) {
586 if_detach(&multicast_register_if6);
587 reg_mif_num = (mifi_t)-1;
590 ip6_mrouter = NULL;
591 ip6_mrouter_ver = 0;
593 splx(s);
595 #ifdef MRT6DEBUG
596 if (mrt6debug)
597 log(LOG_DEBUG, "ip6_mrouter_done\n");
598 #endif
600 return 0;
603 void
604 ip6_mrouter_detach(struct ifnet *ifp)
606 struct rtdetq *rte;
607 struct mf6c *mfc;
608 mifi_t mifi;
609 int i;
611 if (ip6_mrouter == NULL)
612 return;
615 * Delete a mif which points to ifp.
617 for (mifi = 0; mifi < nummifs; mifi++)
618 if (mif6table[mifi].m6_ifp == ifp)
619 del_m6if(&mifi);
622 * Clear rte->ifp of cache entries received on ifp.
624 for (i = 0; i < MF6CTBLSIZ; i++) {
625 if (n6expire[i] == 0)
626 continue;
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;
639 * Add a mif to the mif table
641 static int
642 add_m6if(struct mif6ctl *mifcp)
644 struct mif6 *mifp;
645 struct ifnet *ifp;
646 struct in6_ifreq ifr;
647 int error, s;
648 #ifdef notyet
649 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi;
650 #endif
652 if (mifcp->mif6c_mifi >= MAXMIFS)
653 return EINVAL;
654 mifp = mif6table + mifcp->mif6c_mifi;
655 if (mifp->m6_ifp)
656 return EADDRINUSE; /* XXX: is it appropriate? */
657 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi >= if_indexlim)
658 return ENXIO;
660 * XXX: some OSes can remove ifp and clear ifindex2ifnet[id]
661 * even for id between 0 and if_index.
663 if ((ifp = ifindex2ifnet[mifcp->mif6c_pifi]) == NULL)
664 return ENXIO;
666 if (mifcp->mif6c_flags & MIFF_REGISTER) {
667 ifp = &multicast_register_if6;
669 if (reg_mif_num == (mifi_t)-1) {
670 strlcpy(ifp->if_xname, "register_mif",
671 sizeof(ifp->if_xname));
672 ifp->if_flags |= IFF_LOOPBACK;
673 ifp->if_index = mifcp->mif6c_mifi;
674 reg_mif_num = mifcp->mif6c_mifi;
675 if_attach(ifp);
678 } /* if REGISTER */
679 else {
680 /* Make sure the interface supports multicast */
681 if ((ifp->if_flags & IFF_MULTICAST) == 0)
682 return EOPNOTSUPP;
684 s = splsoftnet();
686 * Enable promiscuous reception of all IPv6 multicasts
687 * from the interface.
689 ifr.ifr_addr.sin6_family = AF_INET6;
690 ifr.ifr_addr.sin6_addr = in6addr_any;
691 error = (*ifp->if_ioctl)(ifp, SIOCADDMULTI, &ifr);
692 splx(s);
693 if (error)
694 return error;
697 s = splsoftnet();
698 mifp->m6_flags = mifcp->mif6c_flags;
699 mifp->m6_ifp = ifp;
700 #ifdef notyet
701 /* scaling up here allows division by 1024 in critical code */
702 mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000;
703 #endif
704 /* initialize per mif pkt counters */
705 mifp->m6_pkt_in = 0;
706 mifp->m6_pkt_out = 0;
707 mifp->m6_bytes_in = 0;
708 mifp->m6_bytes_out = 0;
709 splx(s);
711 /* Adjust nummifs up if the mifi is higher than nummifs */
712 if (nummifs <= mifcp->mif6c_mifi)
713 nummifs = mifcp->mif6c_mifi + 1;
715 #ifdef MRT6DEBUG
716 if (mrt6debug)
717 log(LOG_DEBUG,
718 "add_mif #%d, phyint %s\n",
719 mifcp->mif6c_mifi, ifp->if_xname);
720 #endif
722 return 0;
726 * Delete a mif from the mif table
728 static int
729 del_m6if(mifi_t *mifip)
731 struct mif6 *mifp = mif6table + *mifip;
732 mifi_t mifi;
733 struct ifnet *ifp;
734 struct in6_ifreq ifr;
735 int s;
737 if (*mifip >= nummifs)
738 return EINVAL;
739 if (mifp->m6_ifp == NULL)
740 return EINVAL;
742 s = splsoftnet();
744 if (!(mifp->m6_flags & MIFF_REGISTER)) {
746 * XXX: what if there is yet IPv4 multicast daemon
747 * using the interface?
749 ifp = mifp->m6_ifp;
751 ifr.ifr_addr.sin6_family = AF_INET6;
752 ifr.ifr_addr.sin6_addr = in6addr_any;
753 (*ifp->if_ioctl)(ifp, SIOCDELMULTI, &ifr);
754 } else {
755 if (reg_mif_num != (mifi_t)-1) {
756 if_detach(&multicast_register_if6);
757 reg_mif_num = (mifi_t)-1;
761 #ifdef notyet
762 memset((void *)qtable[*mifip], 0, sizeof(qtable[*mifip]));
763 memset((void *)mifp->m6_tbf, 0, sizeof(*(mifp->m6_tbf)));
764 #endif
765 memset((void *)mifp, 0, sizeof (*mifp));
767 /* Adjust nummifs down */
768 for (mifi = nummifs; mifi > 0; mifi--)
769 if (mif6table[mifi - 1].m6_ifp)
770 break;
771 nummifs = mifi;
773 splx(s);
775 #ifdef MRT6DEBUG
776 if (mrt6debug)
777 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
778 #endif
780 return 0;
784 * Add an mfc entry
786 static int
787 add_m6fc(struct mf6cctl *mfccp)
789 struct mf6c *rt;
790 u_long hash;
791 struct rtdetq *rte;
792 u_short nstl;
793 int s;
795 MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
796 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
798 /* If an entry already exists, just update the fields */
799 if (rt) {
800 #ifdef MRT6DEBUG
801 if (mrt6debug & DEBUG_MFC)
802 log(LOG_DEBUG,"add_m6fc update o %s g %s p %x\n",
803 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
804 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
805 mfccp->mf6cc_parent);
806 #endif
808 s = splsoftnet();
809 rt->mf6c_parent = mfccp->mf6cc_parent;
810 rt->mf6c_ifset = mfccp->mf6cc_ifset;
811 splx(s);
812 return 0;
816 * Find the entry for which the upcall was made and update
818 s = splsoftnet();
819 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
820 mfccp->mf6cc_mcastgrp.sin6_addr);
821 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
822 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
823 &mfccp->mf6cc_origin.sin6_addr) &&
824 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
825 &mfccp->mf6cc_mcastgrp.sin6_addr) &&
826 (rt->mf6c_stall != NULL)) {
828 if (nstl++)
829 log(LOG_ERR,
830 "add_m6fc: %s o %s g %s p %x dbx %p\n",
831 "multiple kernel entries",
832 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
833 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
834 mfccp->mf6cc_parent, rt->mf6c_stall);
836 #ifdef MRT6DEBUG
837 if (mrt6debug & DEBUG_MFC)
838 log(LOG_DEBUG,
839 "add_m6fc o %s g %s p %x dbg %p\n",
840 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
841 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
842 mfccp->mf6cc_parent, rt->mf6c_stall);
843 #endif
845 rt->mf6c_origin = mfccp->mf6cc_origin;
846 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
847 rt->mf6c_parent = mfccp->mf6cc_parent;
848 rt->mf6c_ifset = mfccp->mf6cc_ifset;
849 /* initialize pkt counters per src-grp */
850 rt->mf6c_pkt_cnt = 0;
851 rt->mf6c_byte_cnt = 0;
852 rt->mf6c_wrong_if = 0;
854 rt->mf6c_expire = 0; /* Don't clean this guy up */
855 n6expire[hash]--;
857 /* free packets Qed at the end of this entry */
858 for (rte = rt->mf6c_stall; rte != NULL; ) {
859 struct rtdetq *n = rte->next;
860 if (rte->ifp) {
861 ip6_mdq(rte->m, rte->ifp, rt);
863 m_freem(rte->m);
864 #ifdef UPCALL_TIMING
865 collate(&(rte->t));
866 #endif /* UPCALL_TIMING */
867 free(rte, M_MRTABLE);
868 rte = n;
870 rt->mf6c_stall = NULL;
875 * It is possible that an entry is being inserted without an upcall
877 if (nstl == 0) {
878 #ifdef MRT6DEBUG
879 if (mrt6debug & DEBUG_MFC)
880 log(LOG_DEBUG,
881 "add_mfc no upcall h %ld o %s g %s p %x\n",
882 hash,
883 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
884 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
885 mfccp->mf6cc_parent);
886 #endif
888 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
890 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
891 &mfccp->mf6cc_origin.sin6_addr)&&
892 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
893 &mfccp->mf6cc_mcastgrp.sin6_addr)) {
895 rt->mf6c_origin = mfccp->mf6cc_origin;
896 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
897 rt->mf6c_parent = mfccp->mf6cc_parent;
898 rt->mf6c_ifset = mfccp->mf6cc_ifset;
899 /* initialize pkt counters per src-grp */
900 rt->mf6c_pkt_cnt = 0;
901 rt->mf6c_byte_cnt = 0;
902 rt->mf6c_wrong_if = 0;
904 if (rt->mf6c_expire)
905 n6expire[hash]--;
906 rt->mf6c_expire = 0;
909 if (rt == NULL) {
910 /* no upcall, so make a new entry */
911 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE,
912 M_NOWAIT);
913 if (rt == NULL) {
914 splx(s);
915 return ENOBUFS;
918 /* insert new entry at head of hash chain */
919 rt->mf6c_origin = mfccp->mf6cc_origin;
920 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
921 rt->mf6c_parent = mfccp->mf6cc_parent;
922 rt->mf6c_ifset = mfccp->mf6cc_ifset;
923 /* initialize pkt counters per src-grp */
924 rt->mf6c_pkt_cnt = 0;
925 rt->mf6c_byte_cnt = 0;
926 rt->mf6c_wrong_if = 0;
927 rt->mf6c_expire = 0;
928 rt->mf6c_stall = NULL;
930 /* link into table */
931 rt->mf6c_next = mf6ctable[hash];
932 mf6ctable[hash] = rt;
935 splx(s);
936 return 0;
939 #ifdef UPCALL_TIMING
941 * collect delay statistics on the upcalls
943 static void
944 collate(struct timeval *t)
946 u_long d;
947 struct timeval tp;
948 u_long delta;
950 GET_TIME(tp);
952 if (TV_LT(*t, tp))
954 TV_DELTA(tp, *t, delta);
956 d = delta >> 10;
957 if (d > UPCALL_MAX)
958 d = UPCALL_MAX;
960 ++upcall_data[d];
963 #endif /* UPCALL_TIMING */
966 * Delete an mfc entry
968 static int
969 del_m6fc(struct mf6cctl *mfccp)
971 struct sockaddr_in6 origin;
972 struct sockaddr_in6 mcastgrp;
973 struct mf6c *rt;
974 struct mf6c **nptr;
975 u_long hash;
976 int s;
978 origin = mfccp->mf6cc_origin;
979 mcastgrp = mfccp->mf6cc_mcastgrp;
980 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
982 #ifdef MRT6DEBUG
983 if (mrt6debug & DEBUG_MFC)
984 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
985 ip6_sprintf(&origin.sin6_addr),
986 ip6_sprintf(&mcastgrp.sin6_addr));
987 #endif
989 s = splsoftnet();
991 nptr = &mf6ctable[hash];
992 while ((rt = *nptr) != NULL) {
993 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
994 &rt->mf6c_origin.sin6_addr) &&
995 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
996 &rt->mf6c_mcastgrp.sin6_addr) &&
997 rt->mf6c_stall == NULL)
998 break;
1000 nptr = &rt->mf6c_next;
1002 if (rt == NULL) {
1003 splx(s);
1004 return EADDRNOTAVAIL;
1007 *nptr = rt->mf6c_next;
1008 free(rt, M_MRTABLE);
1010 splx(s);
1012 return 0;
1015 static int
1016 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1018 if (s) {
1019 if (sbappendaddr(&s->so_rcv,
1020 (struct sockaddr *)src,
1021 mm, (struct mbuf *)0) != 0) {
1022 sorwakeup(s);
1023 return 0;
1026 m_freem(mm);
1027 return -1;
1031 * IPv6 multicast forwarding function. This function assumes that the packet
1032 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1033 * pointed to by "ifp", and the packet is to be relayed to other networks
1034 * that have members of the packet's destination IPv6 multicast group.
1036 * The packet is returned unscathed to the caller, unless it is
1037 * erroneous, in which case a non-zero return value tells the caller to
1038 * discard it.
1042 ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1044 struct mf6c *rt;
1045 struct mif6 *mifp;
1046 struct mbuf *mm;
1047 int s;
1048 mifi_t mifi;
1049 struct sockaddr_in6 sin6;
1051 #ifdef MRT6DEBUG
1052 if (mrt6debug & DEBUG_FORWARD)
1053 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1054 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst),
1055 ifp->if_index);
1056 #endif
1059 * Don't forward a packet with Hop limit of zero or one,
1060 * or a packet destined to a local-only group.
1062 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_NODELOCAL(&ip6->ip6_dst) ||
1063 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1064 return 0;
1065 ip6->ip6_hlim--;
1068 * Source address check: do not forward packets with unspecified
1069 * source. It was discussed in July 2000, on ipngwg mailing list.
1070 * This is rather more serious than unicast cases, because some
1071 * MLD packets can be sent with the unspecified source address
1072 * (although such packets must normally set the hop limit field to 1).
1074 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1075 IP6_STATINC(IP6_STAT_CANTFORWARD);
1076 if (ip6_log_time + ip6_log_interval < time_second) {
1077 ip6_log_time = time_second;
1078 log(LOG_DEBUG,
1079 "cannot forward "
1080 "from %s to %s nxt %d received on %s\n",
1081 ip6_sprintf(&ip6->ip6_src),
1082 ip6_sprintf(&ip6->ip6_dst),
1083 ip6->ip6_nxt,
1084 m->m_pkthdr.rcvif ?
1085 if_name(m->m_pkthdr.rcvif) : "?");
1087 return 0;
1091 * Determine forwarding mifs from the forwarding cache table
1093 s = splsoftnet();
1094 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1096 /* Entry exists, so forward if necessary */
1097 if (rt) {
1098 splx(s);
1099 return (ip6_mdq(m, ifp, rt));
1100 } else {
1102 * If we don't have a route for packet's origin,
1103 * Make a copy of the packet &
1104 * send message to routing daemon
1107 struct mbuf *mb0;
1108 struct rtdetq *rte;
1109 u_long hash;
1110 /* int i, npkts;*/
1111 #ifdef UPCALL_TIMING
1112 struct timeval tp;
1114 GET_TIME(tp);
1115 #endif /* UPCALL_TIMING */
1117 mrt6stat.mrt6s_no_route++;
1118 #ifdef MRT6DEBUG
1119 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1120 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1121 ip6_sprintf(&ip6->ip6_src),
1122 ip6_sprintf(&ip6->ip6_dst));
1123 #endif
1126 * Allocate mbufs early so that we don't do extra work if we
1127 * are just going to fail anyway.
1129 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE,
1130 M_NOWAIT);
1131 if (rte == NULL) {
1132 splx(s);
1133 return ENOBUFS;
1135 mb0 = m_copy(m, 0, M_COPYALL);
1137 * Pullup packet header if needed before storing it,
1138 * as other references may modify it in the meantime.
1140 if (mb0 &&
1141 (M_READONLY(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1142 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1143 if (mb0 == NULL) {
1144 free(rte, M_MRTABLE);
1145 splx(s);
1146 return ENOBUFS;
1149 /* is there an upcall waiting for this packet? */
1150 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1151 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1152 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1153 &rt->mf6c_origin.sin6_addr) &&
1154 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1155 &rt->mf6c_mcastgrp.sin6_addr) &&
1156 (rt->mf6c_stall != NULL))
1157 break;
1160 if (rt == NULL) {
1161 struct mrt6msg *im;
1162 struct omrt6msg *oim;
1164 /* no upcall, so make a new entry */
1165 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE,
1166 M_NOWAIT);
1167 if (rt == NULL) {
1168 free(rte, M_MRTABLE);
1169 m_freem(mb0);
1170 splx(s);
1171 return ENOBUFS;
1174 * Make a copy of the header to send to the user
1175 * level process
1177 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1179 if (mm == NULL) {
1180 free(rte, M_MRTABLE);
1181 m_freem(mb0);
1182 free(rt, M_MRTABLE);
1183 splx(s);
1184 return ENOBUFS;
1188 * Send message to routing daemon
1190 sockaddr_in6_init(&sin6, &ip6->ip6_src, 0, 0, 0);
1192 im = NULL;
1193 oim = NULL;
1194 switch (ip6_mrouter_ver) {
1195 case MRT6_OINIT:
1196 oim = mtod(mm, struct omrt6msg *);
1197 oim->im6_msgtype = MRT6MSG_NOCACHE;
1198 oim->im6_mbz = 0;
1199 break;
1200 case MRT6_INIT:
1201 im = mtod(mm, struct mrt6msg *);
1202 im->im6_msgtype = MRT6MSG_NOCACHE;
1203 im->im6_mbz = 0;
1204 break;
1205 default:
1206 free(rte, M_MRTABLE);
1207 m_freem(mb0);
1208 free(rt, M_MRTABLE);
1209 splx(s);
1210 return EINVAL;
1213 #ifdef MRT6DEBUG
1214 if (mrt6debug & DEBUG_FORWARD)
1215 log(LOG_DEBUG,
1216 "getting the iif info in the kernel\n");
1217 #endif
1219 for (mifp = mif6table, mifi = 0;
1220 mifi < nummifs && mifp->m6_ifp != ifp;
1221 mifp++, mifi++)
1224 switch (ip6_mrouter_ver) {
1225 case MRT6_OINIT:
1226 oim->im6_mif = mifi;
1227 break;
1228 case MRT6_INIT:
1229 im->im6_mif = mifi;
1230 break;
1233 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1234 log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1235 "socket queue full\n");
1236 mrt6stat.mrt6s_upq_sockfull++;
1237 free(rte, M_MRTABLE);
1238 m_freem(mb0);
1239 free(rt, M_MRTABLE);
1240 splx(s);
1241 return ENOBUFS;
1244 mrt6stat.mrt6s_upcalls++;
1246 /* insert new entry at head of hash chain */
1247 memset(rt, 0, sizeof(*rt));
1248 sockaddr_in6_init(&rt->mf6c_origin, &ip6->ip6_src,
1249 0, 0, 0);
1250 sockaddr_in6_init(&rt->mf6c_mcastgrp, &ip6->ip6_dst,
1251 0, 0, 0);
1252 rt->mf6c_expire = UPCALL_EXPIRE;
1253 n6expire[hash]++;
1254 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1256 /* link into table */
1257 rt->mf6c_next = mf6ctable[hash];
1258 mf6ctable[hash] = rt;
1259 /* Add this entry to the end of the queue */
1260 rt->mf6c_stall = rte;
1261 } else {
1262 /* determine if q has overflowed */
1263 struct rtdetq **p;
1264 int npkts = 0;
1266 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1267 if (++npkts > MAX_UPQ6) {
1268 mrt6stat.mrt6s_upq_ovflw++;
1269 free(rte, M_MRTABLE);
1270 m_freem(mb0);
1271 splx(s);
1272 return 0;
1275 /* Add this entry to the end of the queue */
1276 *p = rte;
1279 rte->next = NULL;
1280 rte->m = mb0;
1281 rte->ifp = ifp;
1282 #ifdef UPCALL_TIMING
1283 rte->t = tp;
1284 #endif /* UPCALL_TIMING */
1286 splx(s);
1288 return 0;
1293 * Clean up cache entries if upcalls are not serviced
1294 * Call from the Slow Timeout mechanism, every 0.25 seconds.
1296 static void
1297 expire_upcalls(void *unused)
1299 struct rtdetq *rte;
1300 struct mf6c *mfc, **nptr;
1301 int i;
1303 mutex_enter(softnet_lock);
1304 KERNEL_LOCK(1, NULL);
1306 for (i = 0; i < MF6CTBLSIZ; i++) {
1307 if (n6expire[i] == 0)
1308 continue;
1309 nptr = &mf6ctable[i];
1310 while ((mfc = *nptr) != NULL) {
1311 rte = mfc->mf6c_stall;
1313 * Skip real cache entries
1314 * Make sure it wasn't marked to not expire (shouldn't happen)
1315 * If it expires now
1317 if (rte != NULL &&
1318 mfc->mf6c_expire != 0 &&
1319 --mfc->mf6c_expire == 0) {
1320 #ifdef MRT6DEBUG
1321 if (mrt6debug & DEBUG_EXPIRE)
1322 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1323 ip6_sprintf(&mfc->mf6c_origin.sin6_addr),
1324 ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr));
1325 #endif
1327 * drop all the packets
1328 * free the mbuf with the pkt, if, timing info
1330 do {
1331 struct rtdetq *n = rte->next;
1332 m_freem(rte->m);
1333 free(rte, M_MRTABLE);
1334 rte = n;
1335 } while (rte != NULL);
1336 mrt6stat.mrt6s_cache_cleanups++;
1337 n6expire[i]--;
1339 *nptr = mfc->mf6c_next;
1340 free(mfc, M_MRTABLE);
1341 } else {
1342 nptr = &mfc->mf6c_next;
1346 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1347 expire_upcalls, NULL);
1349 KERNEL_UNLOCK_ONE(NULL);
1350 mutex_exit(softnet_lock);
1354 * Packet forwarding routine once entry in the cache is made
1356 static int
1357 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1359 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1360 mifi_t mifi, iif;
1361 struct mif6 *mifp;
1362 int plen = m->m_pkthdr.len;
1363 struct in6_addr src0, dst0; /* copies for local work */
1364 u_int32_t iszone, idzone, oszone, odzone;
1365 int error = 0;
1368 * Macro to send packet on mif. Since RSVP packets don't get counted on
1369 * input, they shouldn't get counted on output, so statistics keeping is
1370 * separate.
1373 #define MC6_SEND(ip6, mifp, m) do { \
1374 if ((mifp)->m6_flags & MIFF_REGISTER) \
1375 register_send((ip6), (mifp), (m)); \
1376 else \
1377 phyint_send((ip6), (mifp), (m)); \
1378 } while (/*CONSTCOND*/ 0)
1381 * Don't forward if it didn't arrive from the parent mif
1382 * for its origin.
1384 mifi = rt->mf6c_parent;
1385 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1386 /* came in the wrong interface */
1387 #ifdef MRT6DEBUG
1388 if (mrt6debug & DEBUG_FORWARD)
1389 log(LOG_DEBUG,
1390 "wrong if: ifid %d mifi %d mififid %x\n",
1391 ifp->if_index, mifi,
1392 mif6table[mifi].m6_ifp ?
1393 mif6table[mifi].m6_ifp->if_index : -1);
1394 #endif
1395 mrt6stat.mrt6s_wrong_if++;
1396 rt->mf6c_wrong_if++;
1398 * If we are doing PIM processing, and we are forwarding
1399 * packets on this interface, send a message to the
1400 * routing daemon.
1402 /* have to make sure this is a valid mif */
1403 if (mifi < nummifs && mif6table[mifi].m6_ifp)
1404 if (pim6 && (m->m_flags & M_LOOP) == 0) {
1406 * Check the M_LOOP flag to avoid an
1407 * unnecessary PIM assert.
1408 * XXX: M_LOOP is an ad-hoc hack...
1410 struct sockaddr_in6 sin6;
1412 struct mbuf *mm;
1413 struct mrt6msg *im;
1414 struct omrt6msg *oim;
1416 mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1417 if (mm &&
1418 (M_READONLY(mm) ||
1419 mm->m_len < sizeof(struct ip6_hdr)))
1420 mm = m_pullup(mm, sizeof(struct ip6_hdr));
1421 if (mm == NULL)
1422 return ENOBUFS;
1424 oim = NULL;
1425 im = NULL;
1426 switch (ip6_mrouter_ver) {
1427 case MRT6_OINIT:
1428 oim = mtod(mm, struct omrt6msg *);
1429 oim->im6_msgtype = MRT6MSG_WRONGMIF;
1430 oim->im6_mbz = 0;
1431 break;
1432 case MRT6_INIT:
1433 im = mtod(mm, struct mrt6msg *);
1434 im->im6_msgtype = MRT6MSG_WRONGMIF;
1435 im->im6_mbz = 0;
1436 break;
1437 default:
1438 m_freem(mm);
1439 return EINVAL;
1442 for (mifp = mif6table, iif = 0;
1443 iif < nummifs && mifp &&
1444 mifp->m6_ifp != ifp;
1445 mifp++, iif++)
1448 memset(&sin6, 0, sizeof(sin6));
1449 sin6.sin6_len = sizeof(sin6);
1450 sin6.sin6_family = AF_INET6;
1451 switch (ip6_mrouter_ver) {
1452 case MRT6_OINIT:
1453 oim->im6_mif = iif;
1454 sin6.sin6_addr = oim->im6_src;
1455 break;
1456 case MRT6_INIT:
1457 im->im6_mif = iif;
1458 sin6.sin6_addr = im->im6_src;
1459 break;
1462 mrt6stat.mrt6s_upcalls++;
1464 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1465 #ifdef MRT6DEBUG
1466 if (mrt6debug)
1467 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1468 #endif
1469 ++mrt6stat.mrt6s_upq_sockfull;
1470 return ENOBUFS;
1471 } /* if socket Q full */
1472 } /* if PIM */
1473 return 0;
1474 } /* if wrong iif */
1476 /* If I sourced this packet, it counts as output, else it was input. */
1477 if (m->m_pkthdr.rcvif == NULL) {
1478 /* XXX: is rcvif really NULL when output?? */
1479 mif6table[mifi].m6_pkt_out++;
1480 mif6table[mifi].m6_bytes_out += plen;
1481 } else {
1482 mif6table[mifi].m6_pkt_in++;
1483 mif6table[mifi].m6_bytes_in += plen;
1485 rt->mf6c_pkt_cnt++;
1486 rt->mf6c_byte_cnt += plen;
1489 * For each mif, forward a copy of the packet if there are group
1490 * members downstream on the interface.
1492 src0 = ip6->ip6_src;
1493 dst0 = ip6->ip6_dst;
1494 if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1495 (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1496 IP6_STATINC(IP6_STAT_BADSCOPE);
1497 return (error);
1499 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++)
1500 if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1501 if (mif6table[mifi].m6_ifp == NULL)
1502 continue;
1504 * check if the outgoing packet is going to break
1505 * a scope boundary.
1506 * XXX: For packets through PIM register tunnel
1507 * interface, we believe the routing daemon.
1509 if ((mif6table[rt->mf6c_parent].m6_flags &
1510 MIFF_REGISTER) == 0 &&
1511 (mif6table[mifi].m6_flags & MIFF_REGISTER) == 0) {
1512 if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1513 &oszone) ||
1514 in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1515 &odzone) ||
1516 iszone != oszone || idzone != odzone) {
1517 IP6_STATINC(IP6_STAT_BADSCOPE);
1518 continue;
1522 mifp->m6_pkt_out++;
1523 mifp->m6_bytes_out += plen;
1524 MC6_SEND(ip6, mifp, m);
1526 return 0;
1529 static void
1530 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1532 struct mbuf *mb_copy;
1533 struct ifnet *ifp = mifp->m6_ifp;
1534 int error = 0;
1535 int s;
1536 static struct route ro;
1537 struct in6_multi *in6m;
1538 struct sockaddr_in6 dst6;
1539 u_long linkmtu;
1541 s = splsoftnet();
1543 * Make a new reference to the packet; make sure that
1544 * the IPv6 header is actually copied, not just referenced,
1545 * so that ip6_output() only scribbles on the copy.
1547 mb_copy = m_copy(m, 0, M_COPYALL);
1548 if (mb_copy &&
1549 (M_READONLY(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1550 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1551 if (mb_copy == NULL) {
1552 splx(s);
1553 return;
1555 /* set MCAST flag to the outgoing packet */
1556 mb_copy->m_flags |= M_MCAST;
1559 * If we sourced the packet, call ip6_output since we may divide
1560 * the packet into fragments when the packet is too big for the
1561 * outgoing interface.
1562 * Otherwise, we can simply send the packet to the interface
1563 * sending queue.
1565 if (m->m_pkthdr.rcvif == NULL) {
1566 struct ip6_moptions im6o;
1568 im6o.im6o_multicast_ifp = ifp;
1569 /* XXX: ip6_output will override ip6->ip6_hlim */
1570 im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1571 im6o.im6o_multicast_loop = 1;
1572 error = ip6_output(mb_copy, NULL, &ro, IPV6_FORWARDING,
1573 &im6o, NULL, NULL);
1575 #ifdef MRT6DEBUG
1576 if (mrt6debug & DEBUG_XMIT)
1577 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1578 mifp - mif6table, error);
1579 #endif
1580 splx(s);
1581 return;
1585 * If we belong to the destination multicast group
1586 * on the outgoing interface, loop back a copy.
1589 * Does not have to check source info, as it's alreay covered by
1590 * ip6_input
1592 sockaddr_in6_init(&dst6, &ip6->ip6_dst, 0, 0, 0);
1594 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m);
1595 if (in6m != NULL) {
1596 ip6_mloopback(ifp, m,
1597 satocsin6(rtcache_getdst(&ro)));
1601 * Put the packet into the sending queue of the outgoing interface
1602 * if it would fit in the MTU of the interface.
1604 linkmtu = IN6_LINKMTU(ifp);
1605 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1607 * We could call if_output directly here, but we use
1608 * nd6_output on purpose to see if IPv6 operation is allowed
1609 * on the interface.
1611 error = nd6_output(ifp, ifp, mb_copy, &dst6, NULL);
1612 #ifdef MRT6DEBUG
1613 if (mrt6debug & DEBUG_XMIT)
1614 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1615 mifp - mif6table, error);
1616 #endif
1617 } else {
1619 * pMTU discovery is intentionally disabled by default, since
1620 * various router may notify pMTU in multicast, which can be
1621 * a DDoS to a router
1623 if (ip6_mcast_pmtu)
1624 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1625 else {
1626 #ifdef MRT6DEBUG
1627 if (mrt6debug & DEBUG_XMIT)
1628 log(LOG_DEBUG,
1629 "phyint_send: packet too big on %s o %s g %s"
1630 " size %d(discarded)\n",
1631 if_name(ifp),
1632 ip6_sprintf(&ip6->ip6_src),
1633 ip6_sprintf(&ip6->ip6_dst),
1634 mb_copy->m_pkthdr.len);
1635 #endif /* MRT6DEBUG */
1636 m_freem(mb_copy); /* simply discard the packet */
1640 splx(s);
1643 static int
1644 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1646 struct mbuf *mm;
1647 int i, len = m->m_pkthdr.len;
1648 struct sockaddr_in6 sin6;
1649 struct mrt6msg *im6;
1651 #ifdef MRT6DEBUG
1652 if (mrt6debug)
1653 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1654 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst));
1655 #endif
1656 PIM6_STATINC(PIM6_STAT_SND_REGISTERS);
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);
1665 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1666 m_freem(mm);
1667 return ENOBUFS;
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 /* TODO: check it! */
1676 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1679 * Send message to routing daemon
1681 sockaddr_in6_init(&sin6, &ip6->ip6_src, 0, 0, 0);
1683 im6 = mtod(mm, struct mrt6msg *);
1684 im6->im6_msgtype = MRT6MSG_WHOLEPKT;
1685 im6->im6_mbz = 0;
1687 im6->im6_mif = mif - mif6table;
1689 /* iif info is not given for reg. encap.n */
1690 mrt6stat.mrt6s_upcalls++;
1692 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1693 #ifdef MRT6DEBUG
1694 if (mrt6debug)
1695 log(LOG_WARNING,
1696 "register_send: ip6_mrouter socket queue full\n");
1697 #endif
1698 ++mrt6stat.mrt6s_upq_sockfull;
1699 return ENOBUFS;
1701 return 0;
1705 * PIM sparse mode hook
1706 * Receives the pim control messages, and passes them up to the listening
1707 * socket, using rip6_input.
1708 * The only message processed is the REGISTER pim message; the pim header
1709 * is stripped off, and the inner packet is passed to register_mforward.
1712 pim6_input(struct mbuf **mp, int *offp, int proto)
1714 struct pim *pim; /* pointer to a pim struct */
1715 struct ip6_hdr *ip6;
1716 int pimlen;
1717 struct mbuf *m = *mp;
1718 int minlen;
1719 int off = *offp;
1721 PIM6_STATINC(PIM6_STAT_RCV_TOTAL);
1723 ip6 = mtod(m, struct ip6_hdr *);
1724 pimlen = m->m_pkthdr.len - *offp;
1727 * Validate lengths
1729 if (pimlen < PIM_MINLEN) {
1730 PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
1731 #ifdef MRT6DEBUG
1732 if (mrt6debug & DEBUG_PIM)
1733 log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1734 #endif
1735 m_freem(m);
1736 return (IPPROTO_DONE);
1740 * if the packet is at least as big as a REGISTER, go ahead
1741 * and grab the PIM REGISTER header size, to avoid another
1742 * possible m_pullup() later.
1744 * PIM_MINLEN == pimhdr + u_int32 == 8
1745 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1747 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1750 * Make sure that the IP6 and PIM headers in contiguous memory, and
1751 * possibly the PIM REGISTER header
1753 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1754 if (pim == NULL) {
1755 PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
1756 return IPPROTO_DONE;
1759 /* PIM version check */
1760 if (pim->pim_ver != PIM_VERSION) {
1761 PIM6_STATINC(PIM6_STAT_RCV_BADVERSION);
1762 #ifdef MRT6DEBUG
1763 log(LOG_ERR,
1764 "pim6_input: incorrect version %d, expecting %d\n",
1765 pim->pim_ver, PIM_VERSION);
1766 #endif
1767 m_freem(m);
1768 return (IPPROTO_DONE);
1771 #define PIM6_CHECKSUM
1772 #ifdef PIM6_CHECKSUM
1774 int cksumlen;
1777 * Validate checksum.
1778 * If PIM REGISTER, exclude the data packet
1780 if (pim->pim_type == PIM_REGISTER)
1781 cksumlen = PIM_MINLEN;
1782 else
1783 cksumlen = pimlen;
1785 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1786 PIM6_STATINC(PIM6_STAT_RCV_BADSUM);
1787 #ifdef MRT6DEBUG
1788 if (mrt6debug & DEBUG_PIM)
1789 log(LOG_DEBUG,
1790 "pim6_input: invalid checksum\n");
1791 #endif
1792 m_freem(m);
1793 return (IPPROTO_DONE);
1796 #endif /* PIM_CHECKSUM */
1798 if (pim->pim_type == PIM_REGISTER) {
1800 * since this is a REGISTER, we'll make a copy of the register
1801 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1802 * routing daemon.
1804 static const struct sockaddr_in6 dst = {
1805 .sin6_len = sizeof(dst),
1806 .sin6_family = AF_INET6,
1809 struct mbuf *mcp;
1810 struct ip6_hdr *eip6;
1811 u_int32_t *reghdr;
1813 PIM6_STATINC(PIM6_STAT_RCV_REGISTERS);
1815 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1816 #ifdef MRT6DEBUG
1817 if (mrt6debug & DEBUG_PIM)
1818 log(LOG_DEBUG,
1819 "pim6_input: register mif not set: %d\n",
1820 reg_mif_num);
1821 #endif
1822 m_freem(m);
1823 return (IPPROTO_DONE);
1826 reghdr = (u_int32_t *)(pim + 1);
1828 if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1829 goto pim6_input_to_daemon;
1832 * Validate length
1834 if (pimlen < PIM6_REG_MINLEN) {
1835 PIM6_STATINC(PIM6_STAT_RCV_TOOSHORT);
1836 PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
1837 #ifdef MRT6DEBUG
1838 log(LOG_ERR,
1839 "pim6_input: register packet size too "
1840 "small %d from %s\n",
1841 pimlen, ip6_sprintf(&ip6->ip6_src));
1842 #endif
1843 m_freem(m);
1844 return (IPPROTO_DONE);
1847 eip6 = (struct ip6_hdr *) (reghdr + 1);
1848 #ifdef MRT6DEBUG
1849 if (mrt6debug & DEBUG_PIM)
1850 log(LOG_DEBUG,
1851 "pim6_input[register], eip6: %s -> %s, "
1852 "eip6 plen %d\n",
1853 ip6_sprintf(&eip6->ip6_src),
1854 ip6_sprintf(&eip6->ip6_dst),
1855 ntohs(eip6->ip6_plen));
1856 #endif
1858 /* verify the version number of the inner packet */
1859 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1860 PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
1861 #ifdef MRT6DEBUG
1862 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1863 "of the inner packet\n",
1864 (eip6->ip6_vfc & IPV6_VERSION));
1865 #endif
1866 m_freem(m);
1867 return (IPPROTO_NONE);
1870 /* verify the inner packet is destined to a mcast group */
1871 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1872 PIM6_STATINC(PIM6_STAT_RCV_BADREGISTERS);
1873 #ifdef MRT6DEBUG
1874 if (mrt6debug & DEBUG_PIM)
1875 log(LOG_DEBUG,
1876 "pim6_input: inner packet of register "
1877 "is not multicast %s\n",
1878 ip6_sprintf(&eip6->ip6_dst));
1879 #endif
1880 m_freem(m);
1881 return (IPPROTO_DONE);
1885 * make a copy of the whole header to pass to the daemon later.
1887 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1888 if (mcp == NULL) {
1889 #ifdef MRT6DEBUG
1890 log(LOG_ERR,
1891 "pim6_input: pim register: "
1892 "could not copy register head\n");
1893 #endif
1894 m_freem(m);
1895 return (IPPROTO_DONE);
1899 * forward the inner ip6 packet; point m_data at the inner ip6.
1901 m_adj(m, off + PIM_MINLEN);
1902 #ifdef MRT6DEBUG
1903 if (mrt6debug & DEBUG_PIM) {
1904 log(LOG_DEBUG,
1905 "pim6_input: forwarding decapsulated register: "
1906 "src %s, dst %s, mif %d\n",
1907 ip6_sprintf(&eip6->ip6_src),
1908 ip6_sprintf(&eip6->ip6_dst),
1909 reg_mif_num);
1911 #endif
1913 looutput(mif6table[reg_mif_num].m6_ifp, m,
1914 (struct sockaddr *)__UNCONST(&dst),
1915 (struct rtentry *) NULL);
1917 /* prepare the register head to send to the mrouting daemon */
1918 m = mcp;
1922 * Pass the PIM message up to the daemon; if it is a register message
1923 * pass the 'head' only up to the daemon. This includes the
1924 * encapsulator ip6 header, pim header, register header and the
1925 * encapsulated ip6 header.
1927 pim6_input_to_daemon:
1928 rip6_input(&m, offp, proto);
1929 return (IPPROTO_DONE);
1932 static int
1933 sysctl_net_inet6_pim6_stats(SYSCTLFN_ARGS)
1936 return (NETSTAT_SYSCTL(pim6stat_percpu, PIM6_NSTATS));
1939 static void
1940 sysctl_net_inet6_pim6_setup(struct sysctllog **clog)
1942 sysctl_createv(clog, 0, NULL, NULL,
1943 CTLFLAG_PERMANENT,
1944 CTLTYPE_NODE, "net", NULL,
1945 NULL, 0, NULL, 0,
1946 CTL_NET, CTL_EOL);
1947 sysctl_createv(clog, 0, NULL, NULL,
1948 CTLFLAG_PERMANENT,
1949 CTLTYPE_NODE, "inet6", NULL,
1950 NULL, 0, NULL, 0,
1951 CTL_NET, PF_INET6, CTL_EOL);
1952 sysctl_createv(clog, 0, NULL, NULL,
1953 CTLFLAG_PERMANENT,
1954 CTLTYPE_NODE, "pim6",
1955 SYSCTL_DESCR("PIMv6 settings"),
1956 NULL, 0, NULL, 0,
1957 CTL_NET, PF_INET6, IPPROTO_PIM, CTL_EOL);
1959 sysctl_createv(clog, 0, NULL, NULL,
1960 CTLFLAG_PERMANENT,
1961 CTLTYPE_STRUCT, "stats",
1962 SYSCTL_DESCR("PIMv6 statistics"),
1963 sysctl_net_inet6_pim6_stats, 0, NULL, 0,
1964 CTL_NET, PF_INET6, IPPROTO_PIM, PIM6CTL_STATS,
1965 CTL_EOL);