1 /* $NetBSD: ip6_mroute.c,v 1.60 2003/12/10 11:46:33 itojun 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 1.60 2003/12/10 11:46:33 itojun Exp $");
121
122 #include "opt_inet.h"
123 #include "opt_mrouting.h"
124
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/syslog.h>
138
139 #include <net/if.h>
140 #include <net/route.h>
141 #include <net/raw_cb.h>
142
143 #include <netinet/in.h>
144 #include <netinet/in_var.h>
145 #ifdef MULTICAST_PMTUD
146 #include <netinet/icmp6.h>
147 #endif
148
149 #include <netinet/ip6.h>
150 #include <netinet6/ip6_var.h>
151 #include <netinet6/ip6_mroute.h>
152 #include <netinet6/pim6.h>
153 #include <netinet6/pim6_var.h>
154 #include <netinet6/nd6.h>
155
156 #include <net/net_osdep.h>
157
158 static int ip6_mdq __P((struct mbuf *, struct ifnet *, struct mf6c *));
159 static void phyint_send __P((struct ip6_hdr *, struct mif6 *, struct mbuf *));
160
161 static int set_pim6 __P((int *));
162 static int get_pim6 __P((struct mbuf *));
163 static int socket_send __P((struct socket *, struct mbuf *,
164 struct sockaddr_in6 *));
165 static int register_send __P((struct ip6_hdr *, struct mif6 *, struct mbuf *));
166
167 /*
168 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static,
169 * except for netstat or debugging purposes.
170 */
171 struct socket *ip6_mrouter = NULL;
172 int ip6_mrouter_ver = 0;
173 int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */
174 struct mrt6stat mrt6stat;
175
176 #define NO_RTE_FOUND 0x1
177 #define RTE_FOUND 0x2
178
179 struct mf6c *mf6ctable[MF6CTBLSIZ];
180 u_char n6expire[MF6CTBLSIZ];
181 static struct mif6 mif6table[MAXMIFS];
182 #ifdef MRT6DEBUG
183 u_int mrt6debug = 0; /* debug level */
184 #define DEBUG_MFC 0x02
185 #define DEBUG_FORWARD 0x04
186 #define DEBUG_EXPIRE 0x08
187 #define DEBUG_XMIT 0x10
188 #define DEBUG_REG 0x20
189 #define DEBUG_PIM 0x40
190 #endif
191
192 static void expire_upcalls __P((void *));
193 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */
194 #define UPCALL_EXPIRE 6 /* number of timeouts */
195
196 #ifdef INET
197 #ifdef MROUTING
198 extern struct socket *ip_mrouter;
199 #endif
200 #endif
201
202 /*
203 * 'Interfaces' associated with decapsulator (so we can tell
204 * packets that went through it from ones that get reflected
205 * by a broken gateway). These interfaces are never linked into
206 * the system ifnet list & no routes point to them. I.e., packets
207 * can't be sent this way. They only exist as a placeholder for
208 * multicast source verification.
209 */
210 struct ifnet multicast_register_if;
211
212 #define ENCAP_HOPS 64
213
214 /*
215 * Private variables.
216 */
217 static mifi_t nummifs = 0;
218 static mifi_t reg_mif_num = (mifi_t)-1;
219
220 static struct pim6stat pim6stat;
221 static int pim6;
222
223 /*
224 * Hash function for a source, group entry
225 */
226 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
227 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
228 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
229 (g).s6_addr32[2] ^ (g).s6_addr32[3])
230
231 /*
232 * Find a route for a given origin IPv6 address and Multicast group address.
233 * Quality of service parameter to be added in the future!!!
234 */
235
236 #define MF6CFIND(o, g, rt) do { \
237 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
238 rt = NULL; \
239 mrt6stat.mrt6s_mfc_lookups++; \
240 while (_rt) { \
241 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
242 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
243 (_rt->mf6c_stall == NULL)) { \
244 rt = _rt; \
245 break; \
246 } \
247 _rt = _rt->mf6c_next; \
248 } \
249 if (rt == NULL) { \
250 mrt6stat.mrt6s_mfc_misses++; \
251 } \
252 } while (/*CONSTCOND*/ 0)
253
254 /*
255 * Macros to compute elapsed time efficiently
256 * Borrowed from Van Jacobson's scheduling code
257 */
258 #define TV_DELTA(a, b, delta) do { \
259 int xxs; \
260 \
261 delta = (a).tv_usec - (b).tv_usec; \
262 if ((xxs = (a).tv_sec - (b).tv_sec)) { \
263 switch (xxs) { \
264 case 2: \
265 delta += 1000000; \
266 /* FALLTHROUGH */ \
267 case 1: \
268 delta += 1000000; \
269 break; \
270 default: \
271 delta += (1000000 * xxs); \
272 } \
273 } \
274 } while (/*CONSTCOND*/ 0)
275
276 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
277 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
278
279 #ifdef UPCALL_TIMING
280 #define UPCALL_MAX 50
281 u_long upcall_data[UPCALL_MAX + 1];
282 static void collate();
283 #endif /* UPCALL_TIMING */
284
285 static int get_sg_cnt __P((struct sioc_sg_req6 *));
286 static int get_mif6_cnt __P((struct sioc_mif_req6 *));
287 static int ip6_mrouter_init __P((struct socket *, int, int));
288 static int add_m6if __P((struct mif6ctl *));
289 static int del_m6if __P((mifi_t *));
290 static int add_m6fc __P((struct mf6cctl *));
291 static int del_m6fc __P((struct mf6cctl *));
292
293 static struct callout expire_upcalls_ch = CALLOUT_INITIALIZER;
294
295 /*
296 * Handle MRT setsockopt commands to modify the multicast routing tables.
297 */
298 int
299 ip6_mrouter_set(cmd, so, m)
300 int cmd;
301 struct socket *so;
302 struct mbuf *m;
303 {
304 if (cmd != MRT6_INIT && so != ip6_mrouter)
305 return (EACCES);
306
307 switch (cmd) {
308 #ifdef MRT6_OINIT
309 case MRT6_OINIT:
310 #endif
311 case MRT6_INIT:
312 if (m == NULL || m->m_len < sizeof(int))
313 return (EINVAL);
314 return (ip6_mrouter_init(so, *mtod(m, int *), cmd));
315 case MRT6_DONE:
316 return (ip6_mrouter_done());
317 case MRT6_ADD_MIF:
318 if (m == NULL || m->m_len < sizeof(struct mif6ctl))
319 return (EINVAL);
320 return (add_m6if(mtod(m, struct mif6ctl *)));
321 case MRT6_DEL_MIF:
322 if (m == NULL || m->m_len < sizeof(mifi_t))
323 return (EINVAL);
324 return (del_m6if(mtod(m, mifi_t *)));
325 case MRT6_ADD_MFC:
326 if (m == NULL || m->m_len < sizeof(struct mf6cctl))
327 return (EINVAL);
328 return (add_m6fc(mtod(m, struct mf6cctl *)));
329 case MRT6_DEL_MFC:
330 if (m == NULL || m->m_len < sizeof(struct mf6cctl))
331 return (EINVAL);
332 return (del_m6fc(mtod(m, struct mf6cctl *)));
333 case MRT6_PIM:
334 if (m == NULL || m->m_len < sizeof(int))
335 return (EINVAL);
336 return (set_pim6(mtod(m, int *)));
337 default:
338 return (EOPNOTSUPP);
339 }
340 }
341
342 /*
343 * Handle MRT getsockopt commands
344 */
345 int
346 ip6_mrouter_get(cmd, so, m)
347 int cmd;
348 struct socket *so;
349 struct mbuf **m;
350 {
351 struct mbuf *mb;
352
353 if (so != ip6_mrouter) return EACCES;
354
355 *m = mb = m_get(M_WAIT, MT_SOOPTS);
356
357 switch (cmd) {
358 case MRT6_PIM:
359 return get_pim6(mb);
360 default:
361 m_free(mb);
362 return EOPNOTSUPP;
363 }
364 }
365
366 /*
367 * Handle ioctl commands to obtain information from the cache
368 */
369 int
370 mrt6_ioctl(cmd, data)
371 int cmd;
372 caddr_t data;
373 {
374
375 switch (cmd) {
376 case SIOCGETSGCNT_IN6:
377 return (get_sg_cnt((struct sioc_sg_req6 *)data));
378 case SIOCGETMIFCNT_IN6:
379 return (get_mif6_cnt((struct sioc_mif_req6 *)data));
380 default:
381 return (EINVAL);
382 }
383 }
384
385 /*
386 * returns the packet, byte, rpf-failure count for the source group provided
387 */
388 static int
389 get_sg_cnt(req)
390 struct sioc_sg_req6 *req;
391 {
392 struct mf6c *rt;
393 int s;
394
395 s = splsoftnet();
396 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
397 splx(s);
398 if (rt != NULL) {
399 req->pktcnt = rt->mf6c_pkt_cnt;
400 req->bytecnt = rt->mf6c_byte_cnt;
401 req->wrong_if = rt->mf6c_wrong_if;
402 } else
403 return (ESRCH);
404 #if 0
405 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff;
406 #endif
407
408 return 0;
409 }
410
411 /*
412 * returns the input and output packet and byte counts on the mif provided
413 */
414 static int
415 get_mif6_cnt(req)
416 struct sioc_mif_req6 *req;
417 {
418 mifi_t mifi = req->mifi;
419
420 if (mifi >= nummifs)
421 return EINVAL;
422
423 req->icount = mif6table[mifi].m6_pkt_in;
424 req->ocount = mif6table[mifi].m6_pkt_out;
425 req->ibytes = mif6table[mifi].m6_bytes_in;
426 req->obytes = mif6table[mifi].m6_bytes_out;
427
428 return 0;
429 }
430
431 /*
432 * Get PIM processiong global
433 */
434 static int
435 get_pim6(m)
436 struct mbuf *m;
437 {
438 int *i;
439
440 i = mtod(m, int *);
441
442 *i = pim6;
443
444 return 0;
445 }
446
447 static int
448 set_pim6(i)
449 int *i;
450 {
451 if ((*i != 1) && (*i != 0))
452 return EINVAL;
453
454 pim6 = *i;
455
456 return 0;
457 }
458
459 /*
460 * Enable multicast routing
461 */
462 static int
463 ip6_mrouter_init(so, v, cmd)
464 struct socket *so;
465 int v;
466 int cmd;
467 {
468 #ifdef MRT6DEBUG
469 if (mrt6debug)
470 log(LOG_DEBUG,
471 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
472 so->so_type, so->so_proto->pr_protocol);
473 #endif
474
475 if (so->so_type != SOCK_RAW ||
476 so->so_proto->pr_protocol != IPPROTO_ICMPV6)
477 return (EOPNOTSUPP);
478
479 if (v != 1)
480 return (ENOPROTOOPT);
481
482 if (ip6_mrouter != NULL)
483 return (EADDRINUSE);
484
485 ip6_mrouter = so;
486 ip6_mrouter_ver = cmd;
487
488 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
489 bzero((caddr_t)n6expire, sizeof(n6expire));
490
491 pim6 = 0;/* used for stubbing out/in pim stuff */
492
493 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
494 expire_upcalls, NULL);
495
496 #ifdef MRT6DEBUG
497 if (mrt6debug)
498 log(LOG_DEBUG, "ip6_mrouter_init\n");
499 #endif
500
501 return 0;
502 }
503
504 /*
505 * Disable multicast routing
506 */
507 int
508 ip6_mrouter_done()
509 {
510 mifi_t mifi;
511 int i;
512 struct ifnet *ifp;
513 struct in6_ifreq ifr;
514 struct mf6c *rt;
515 struct rtdetq *rte;
516 int s;
517
518 s = splsoftnet();
519
520 /*
521 * For each phyint in use, disable promiscuous reception of all IPv6
522 * multicasts.
523 */
524 #ifdef INET
525 #ifdef MROUTING
526 /*
527 * If there is still IPv4 multicast routing daemon,
528 * we remain interfaces to receive all muliticasted packets.
529 * XXX: there may be an interface in which the IPv4 multicast
530 * daemon is not interested...
531 */
532 if (!ip_mrouter)
533 #endif
534 #endif
535 {
536 for (mifi = 0; mifi < nummifs; mifi++) {
537 if (mif6table[mifi].m6_ifp &&
538 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
539 ifr.ifr_addr.sin6_family = AF_INET6;
540 ifr.ifr_addr.sin6_addr= in6addr_any;
541 ifp = mif6table[mifi].m6_ifp;
542 (*ifp->if_ioctl)(ifp, SIOCDELMULTI,
543 (caddr_t)&ifr);
544 }
545 }
546 }
547 #ifdef notyet
548 bzero((caddr_t)qtable, sizeof(qtable));
549 bzero((caddr_t)tbftable, sizeof(tbftable));
550 #endif
551 bzero((caddr_t)mif6table, 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_free(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 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
580
581 /*
582 * Reset de-encapsulation cache
583 */
584 reg_mif_num = -1;
585
586 ip6_mrouter = NULL;
587 ip6_mrouter_ver = 0;
588
589 splx(s);
590
591 #ifdef MRT6DEBUG
592 if (mrt6debug)
593 log(LOG_DEBUG, "ip6_mrouter_done\n");
594 #endif
595
596 return 0;
597 }
598
599 void
600 ip6_mrouter_detach(ifp)
601 struct ifnet *ifp;
602 {
603 struct rtdetq *rte;
604 struct mf6c *mfc;
605 mifi_t mifi;
606 int i;
607
608 /*
609 * Delete a mif which points to ifp.
610 */
611 for (mifi = 0; mifi < nummifs; mifi++)
612 if (mif6table[mifi].m6_ifp == ifp)
613 del_m6if(&mifi);
614
615 /*
616 * Clear rte->ifp of cache entries received on ifp.
617 */
618 for (i = 0; i < MF6CTBLSIZ; i++) {
619 if (n6expire[i] == 0)
620 continue;
621
622 for (mfc = mf6ctable[i]; mfc != NULL; mfc = mfc->mf6c_next) {
623 for (rte = mfc->mf6c_stall; rte != NULL; rte = rte->next) {
624 if (rte->ifp == ifp)
625 rte->ifp = NULL;
626 }
627 }
628 }
629 }
630
631 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
632
633 /*
634 * Add a mif to the mif table
635 */
636 static int
637 add_m6if(mifcp)
638 struct mif6ctl *mifcp;
639 {
640 struct mif6 *mifp;
641 struct ifnet *ifp;
642 struct in6_ifreq ifr;
643 int error, s;
644 #ifdef notyet
645 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi;
646 #endif
647
648 if (mifcp->mif6c_mifi >= MAXMIFS)
649 return EINVAL;
650 mifp = mif6table + mifcp->mif6c_mifi;
651 if (mifp->m6_ifp)
652 return EADDRINUSE; /* XXX: is it appropriate? */
653 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi >= if_indexlim ||
654 !ifindex2ifnet[mifcp->mif6c_pifi])
655 return ENXIO;
656 /*
657 * XXX: some OSes can remove ifp and clear ifindex2ifnet[id]
658 * even for id between 0 and if_index.
659 */
660 if ((ifp = ifindex2ifnet[mifcp->mif6c_pifi]) == NULL)
661 return ENXIO;
662
663 if (mifcp->mif6c_flags & MIFF_REGISTER) {
664 if (reg_mif_num == (mifi_t)-1) {
665 strlcpy(multicast_register_if.if_xname,
666 "register_mif",
667 sizeof(multicast_register_if.if_xname));
668 multicast_register_if.if_flags |= IFF_LOOPBACK;
669 multicast_register_if.if_index = mifcp->mif6c_mifi;
670 reg_mif_num = mifcp->mif6c_mifi;
671 }
672
673 ifp = &multicast_register_if;
674
675 } /* if REGISTER */
676 else {
677 /* Make sure the interface supports multicast */
678 if ((ifp->if_flags & IFF_MULTICAST) == 0)
679 return EOPNOTSUPP;
680
681 s = splsoftnet();
682 /*
683 * Enable promiscuous reception of all IPv6 multicasts
684 * from the interface.
685 */
686 ifr.ifr_addr.sin6_family = AF_INET6;
687 ifr.ifr_addr.sin6_addr = in6addr_any;
688 error = (*ifp->if_ioctl)(ifp, SIOCADDMULTI, (caddr_t)&ifr);
689 splx(s);
690 if (error)
691 return error;
692 }
693
694 s = splsoftnet();
695 mifp->m6_flags = mifcp->mif6c_flags;
696 mifp->m6_ifp = ifp;
697 #ifdef notyet
698 /* scaling up here allows division by 1024 in critical code */
699 mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000;
700 #endif
701 /* initialize per mif pkt counters */
702 mifp->m6_pkt_in = 0;
703 mifp->m6_pkt_out = 0;
704 mifp->m6_bytes_in = 0;
705 mifp->m6_bytes_out = 0;
706 splx(s);
707
708 /* Adjust nummifs up if the mifi is higher than nummifs */
709 if (nummifs <= mifcp->mif6c_mifi)
710 nummifs = mifcp->mif6c_mifi + 1;
711
712 #ifdef MRT6DEBUG
713 if (mrt6debug)
714 log(LOG_DEBUG,
715 "add_mif #%d, phyint %s%d\n",
716 mifcp->mif6c_mifi,
717 ifp->if_name, ifp->if_unit);
718 #endif
719
720 return 0;
721 }
722
723 /*
724 * Delete a mif from the mif table
725 */
726 static int
727 del_m6if(mifip)
728 mifi_t *mifip;
729 {
730 struct mif6 *mifp = mif6table + *mifip;
731 mifi_t mifi;
732 struct ifnet *ifp;
733 struct in6_ifreq ifr;
734 int s;
735
736 if (*mifip >= nummifs)
737 return EINVAL;
738 if (mifp->m6_ifp == NULL)
739 return EINVAL;
740
741 s = splsoftnet();
742
743 if (!(mifp->m6_flags & MIFF_REGISTER)) {
744 /*
745 * XXX: what if there is yet IPv4 multicast daemon
746 * using the interface?
747 */
748 ifp = mifp->m6_ifp;
749
750 ifr.ifr_addr.sin6_family = AF_INET6;
751 ifr.ifr_addr.sin6_addr = in6addr_any;
752 (*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr);
753 }
754
755 #ifdef notyet
756 bzero((caddr_t)qtable[*mifip], sizeof(qtable[*mifip]));
757 bzero((caddr_t)mifp->m6_tbf, sizeof(*(mifp->m6_tbf)));
758 #endif
759 bzero((caddr_t)mifp, sizeof (*mifp));
760
761 /* Adjust nummifs down */
762 for (mifi = nummifs; mifi > 0; mifi--)
763 if (mif6table[mifi - 1].m6_ifp)
764 break;
765 nummifs = mifi;
766
767 splx(s);
768
769 #ifdef MRT6DEBUG
770 if (mrt6debug)
771 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
772 #endif
773
774 return 0;
775 }
776
777 /*
778 * Add an mfc entry
779 */
780 static int
781 add_m6fc(mfccp)
782 struct mf6cctl *mfccp;
783 {
784 struct mf6c *rt;
785 u_long hash;
786 struct rtdetq *rte;
787 u_short nstl;
788 int s;
789
790 MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
791 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
792
793 /* If an entry already exists, just update the fields */
794 if (rt) {
795 #ifdef MRT6DEBUG
796 if (mrt6debug & DEBUG_MFC)
797 log(LOG_DEBUG,"add_m6fc update o %s g %s p %x\n",
798 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
799 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
800 mfccp->mf6cc_parent);
801 #endif
802
803 s = splsoftnet();
804 rt->mf6c_parent = mfccp->mf6cc_parent;
805 rt->mf6c_ifset = mfccp->mf6cc_ifset;
806 splx(s);
807 return 0;
808 }
809
810 /*
811 * Find the entry for which the upcall was made and update
812 */
813 s = splsoftnet();
814 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
815 mfccp->mf6cc_mcastgrp.sin6_addr);
816 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
817 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
818 &mfccp->mf6cc_origin.sin6_addr) &&
819 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
820 &mfccp->mf6cc_mcastgrp.sin6_addr) &&
821 (rt->mf6c_stall != NULL)) {
822
823 if (nstl++)
824 log(LOG_ERR,
825 "add_m6fc: %s o %s g %s p %x dbx %p\n",
826 "multiple kernel entries",
827 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
828 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
829 mfccp->mf6cc_parent, rt->mf6c_stall);
830
831 #ifdef MRT6DEBUG
832 if (mrt6debug & DEBUG_MFC)
833 log(LOG_DEBUG,
834 "add_m6fc o %s g %s p %x dbg %x\n",
835 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
836 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
837 mfccp->mf6cc_parent, rt->mf6c_stall);
838 #endif
839
840 rt->mf6c_origin = mfccp->mf6cc_origin;
841 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
842 rt->mf6c_parent = mfccp->mf6cc_parent;
843 rt->mf6c_ifset = mfccp->mf6cc_ifset;
844 /* initialize pkt counters per src-grp */
845 rt->mf6c_pkt_cnt = 0;
846 rt->mf6c_byte_cnt = 0;
847 rt->mf6c_wrong_if = 0;
848
849 rt->mf6c_expire = 0; /* Don't clean this guy up */
850 n6expire[hash]--;
851
852 /* free packets Qed at the end of this entry */
853 for (rte = rt->mf6c_stall; rte != NULL; ) {
854 struct rtdetq *n = rte->next;
855 if (rte->ifp) {
856 ip6_mdq(rte->m, rte->ifp, rt);
857 }
858 m_freem(rte->m);
859 #ifdef UPCALL_TIMING
860 collate(&(rte->t));
861 #endif /* UPCALL_TIMING */
862 free(rte, M_MRTABLE);
863 rte = n;
864 }
865 rt->mf6c_stall = NULL;
866 }
867 }
868
869 /*
870 * It is possible that an entry is being inserted without an upcall
871 */
872 if (nstl == 0) {
873 #ifdef MRT6DEBUG
874 if (mrt6debug & DEBUG_MFC)
875 log(LOG_DEBUG,
876 "add_mfc no upcall h %d o %s g %s p %x\n",
877 hash,
878 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
879 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
880 mfccp->mf6cc_parent);
881 #endif
882
883 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
884
885 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
886 &mfccp->mf6cc_origin.sin6_addr)&&
887 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
888 &mfccp->mf6cc_mcastgrp.sin6_addr)) {
889
890 rt->mf6c_origin = mfccp->mf6cc_origin;
891 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
892 rt->mf6c_parent = mfccp->mf6cc_parent;
893 rt->mf6c_ifset = mfccp->mf6cc_ifset;
894 /* initialize pkt counters per src-grp */
895 rt->mf6c_pkt_cnt = 0;
896 rt->mf6c_byte_cnt = 0;
897 rt->mf6c_wrong_if = 0;
898
899 if (rt->mf6c_expire)
900 n6expire[hash]--;
901 rt->mf6c_expire = 0;
902 }
903 }
904 if (rt == NULL) {
905 /* no upcall, so make a new entry */
906 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE,
907 M_NOWAIT);
908 if (rt == NULL) {
909 splx(s);
910 return ENOBUFS;
911 }
912
913 /* insert new entry at head of hash chain */
914 rt->mf6c_origin = mfccp->mf6cc_origin;
915 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
916 rt->mf6c_parent = mfccp->mf6cc_parent;
917 rt->mf6c_ifset = mfccp->mf6cc_ifset;
918 /* initialize pkt counters per src-grp */
919 rt->mf6c_pkt_cnt = 0;
920 rt->mf6c_byte_cnt = 0;
921 rt->mf6c_wrong_if = 0;
922 rt->mf6c_expire = 0;
923 rt->mf6c_stall = NULL;
924
925 /* link into table */
926 rt->mf6c_next = mf6ctable[hash];
927 mf6ctable[hash] = rt;
928 }
929 }
930 splx(s);
931 return 0;
932 }
933
934 #ifdef UPCALL_TIMING
935 /*
936 * collect delay statistics on the upcalls
937 */
938 static void
939 collate(t)
940 struct timeval *t;
941 {
942 u_long d;
943 struct timeval tp;
944 u_long delta;
945
946 GET_TIME(tp);
947
948 if (TV_LT(*t, tp))
949 {
950 TV_DELTA(tp, *t, delta);
951
952 d = delta >> 10;
953 if (d > UPCALL_MAX)
954 d = UPCALL_MAX;
955
956 ++upcall_data[d];
957 }
958 }
959 #endif /* UPCALL_TIMING */
960
961 /*
962 * Delete an mfc entry
963 */
964 static int
965 del_m6fc(mfccp)
966 struct mf6cctl *mfccp;
967 {
968 struct sockaddr_in6 origin;
969 struct sockaddr_in6 mcastgrp;
970 struct mf6c *rt;
971 struct mf6c **nptr;
972 u_long hash;
973 int s;
974
975 origin = mfccp->mf6cc_origin;
976 mcastgrp = mfccp->mf6cc_mcastgrp;
977 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
978
979 #ifdef MRT6DEBUG
980 if (mrt6debug & DEBUG_MFC)
981 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
982 ip6_sprintf(&origin.sin6_addr),
983 ip6_sprintf(&mcastgrp.sin6_addr));
984 #endif
985
986 s = splsoftnet();
987
988 nptr = &mf6ctable[hash];
989 while ((rt = *nptr) != NULL) {
990 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
991 &rt->mf6c_origin.sin6_addr) &&
992 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
993 &rt->mf6c_mcastgrp.sin6_addr) &&
994 rt->mf6c_stall == NULL)
995 break;
996
997 nptr = &rt->mf6c_next;
998 }
999 if (rt == NULL) {
1000 splx(s);
1001 return EADDRNOTAVAIL;
1002 }
1003
1004 *nptr = rt->mf6c_next;
1005 free(rt, M_MRTABLE);
1006
1007 splx(s);
1008
1009 return 0;
1010 }
1011
1012 static int
1013 socket_send(s, mm, src)
1014 struct socket *s;
1015 struct mbuf *mm;
1016 struct sockaddr_in6 *src;
1017 {
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;
1024 }
1025 }
1026 m_freem(mm);
1027 return -1;
1028 }
1029
1030 /*
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.
1035 *
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.
1039 */
1040
1041 int
1042 ip6_mforward(ip6, ifp, m)
1043 struct ip6_hdr *ip6;
1044 struct ifnet *ifp;
1045 struct mbuf *m;
1046 {
1047 struct mf6c *rt;
1048 struct mif6 *mifp;
1049 struct mbuf *mm;
1050 int s;
1051 mifi_t mifi;
1052
1053 #ifdef MRT6DEBUG
1054 if (mrt6debug & DEBUG_FORWARD)
1055 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1056 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst),
1057 ifp->if_index);
1058 #endif
1059
1060 /*
1061 * Don't forward a packet with Hop limit of zero or one,
1062 * or a packet destined to a local-only group.
1063 */
1064 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_NODELOCAL(&ip6->ip6_dst) ||
1065 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1066 return 0;
1067 ip6->ip6_hlim--;
1068
1069 /*
1070 * Source address check: do not forward packets with unspecified
1071 * source. It was discussed in July 2000, on ipngwg mailing list.
1072 * This is rather more serious than unicast cases, because some
1073 * MLD packets can be sent with the unspecified source address
1074 * (although such packets must normally set 1 to the hop limit field).
1075 */
1076 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1077 ip6stat.ip6s_cantforward++;
1078 if (ip6_log_time + ip6_log_interval < time.tv_sec) {
1079 ip6_log_time = time.tv_sec;
1080 log(LOG_DEBUG,
1081 "cannot forward "
1082 "from %s to %s nxt %d received on %s\n",
1083 ip6_sprintf(&ip6->ip6_src),
1084 ip6_sprintf(&ip6->ip6_dst),
1085 ip6->ip6_nxt,
1086 m->m_pkthdr.rcvif ?
1087 if_name(m->m_pkthdr.rcvif) : "?");
1088 }
1089 return 0;
1090 }
1091
1092 /*
1093 * Determine forwarding mifs from the forwarding cache table
1094 */
1095 s = splsoftnet();
1096 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1097
1098 /* Entry exists, so forward if necessary */
1099 if (rt) {
1100 splx(s);
1101 return (ip6_mdq(m, ifp, rt));
1102 } else {
1103 /*
1104 * If we don't have a route for packet's origin,
1105 * Make a copy of the packet &
1106 * send message to routing daemon
1107 */
1108
1109 struct mbuf *mb0;
1110 struct rtdetq *rte;
1111 u_long hash;
1112 /* int i, npkts;*/
1113 #ifdef UPCALL_TIMING
1114 struct timeval tp;
1115
1116 GET_TIME(tp);
1117 #endif /* UPCALL_TIMING */
1118
1119 mrt6stat.mrt6s_no_route++;
1120 #ifdef MRT6DEBUG
1121 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1122 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1123 ip6_sprintf(&ip6->ip6_src),
1124 ip6_sprintf(&ip6->ip6_dst));
1125 #endif
1126
1127 /*
1128 * Allocate mbufs early so that we don't do extra work if we
1129 * are just going to fail anyway.
1130 */
1131 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE,
1132 M_NOWAIT);
1133 if (rte == NULL) {
1134 splx(s);
1135 return ENOBUFS;
1136 }
1137 mb0 = m_copy(m, 0, M_COPYALL);
1138 /*
1139 * Pullup packet header if needed before storing it,
1140 * as other references may modify it in the meantime.
1141 */
1142 if (mb0 &&
1143 (M_READONLY(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1144 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1145 if (mb0 == NULL) {
1146 free(rte, M_MRTABLE);
1147 splx(s);
1148 return ENOBUFS;
1149 }
1150
1151 /* is there an upcall waiting for this packet? */
1152 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1153 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1154 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1155 &rt->mf6c_origin.sin6_addr) &&
1156 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1157 &rt->mf6c_mcastgrp.sin6_addr) &&
1158 (rt->mf6c_stall != NULL))
1159 break;
1160 }
1161
1162 if (rt == NULL) {
1163 struct mrt6msg *im;
1164 struct omrt6msg *oim;
1165
1166 /* no upcall, so make a new entry */
1167 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE,
1168 M_NOWAIT);
1169 if (rt == NULL) {
1170 free(rte, M_MRTABLE);
1171 m_freem(mb0);
1172 splx(s);
1173 return ENOBUFS;
1174 }
1175 /*
1176 * Make a copy of the header to send to the user
1177 * level process
1178 */
1179 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1180
1181 if (mm == NULL) {
1182 free(rte, M_MRTABLE);
1183 m_freem(mb0);
1184 free(rt, M_MRTABLE);
1185 splx(s);
1186 return ENOBUFS;
1187 }
1188
1189 /*
1190 * Send message to routing daemon
1191 */
1192 sin6.sin6_addr = ip6->ip6_src;
1193
1194 im = NULL;
1195 oim = NULL;
1196 switch (ip6_mrouter_ver) {
1197 case MRT6_OINIT:
1198 oim = mtod(mm, struct omrt6msg *);
1199 oim->im6_msgtype = MRT6MSG_NOCACHE;
1200 oim->im6_mbz = 0;
1201 break;
1202 case MRT6_INIT:
1203 im = mtod(mm, struct mrt6msg *);
1204 im->im6_msgtype = MRT6MSG_NOCACHE;
1205 im->im6_mbz = 0;
1206 break;
1207 default:
1208 free(rte, M_MRTABLE);
1209 m_freem(mb0);
1210 free(rt, M_MRTABLE);
1211 splx(s);
1212 return EINVAL;
1213 }
1214
1215 #ifdef MRT6DEBUG
1216 if (mrt6debug & DEBUG_FORWARD)
1217 log(LOG_DEBUG,
1218 "getting the iif info in the kernel\n");
1219 #endif
1220
1221 for (mifp = mif6table, mifi = 0;
1222 mifi < nummifs && mifp->m6_ifp != ifp;
1223 mifp++, mifi++)
1224 ;
1225
1226 switch (ip6_mrouter_ver) {
1227 case MRT6_OINIT:
1228 oim->im6_mif = mifi;
1229 break;
1230 case MRT6_INIT:
1231 im->im6_mif = mifi;
1232 break;
1233 }
1234
1235 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1236 log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1237 "socket queue full\n");
1238 mrt6stat.mrt6s_upq_sockfull++;
1239 free(rte, M_MRTABLE);
1240 m_freem(mb0);
1241 free(rt, M_MRTABLE);
1242 splx(s);
1243 return ENOBUFS;
1244 }
1245
1246 mrt6stat.mrt6s_upcalls++;
1247
1248 /* insert new entry at head of hash chain */
1249 bzero(rt, sizeof(*rt));
1250 rt->mf6c_origin.sin6_family = AF_INET6;
1251 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1252 rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1253 rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1254 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1255 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1256 rt->mf6c_expire = UPCALL_EXPIRE;
1257 n6expire[hash]++;
1258 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1259
1260 /* link into table */
1261 rt->mf6c_next = mf6ctable[hash];
1262 mf6ctable[hash] = rt;
1263 /* Add this entry to the end of the queue */
1264 rt->mf6c_stall = rte;
1265 } else {
1266 /* determine if q has overflowed */
1267 struct rtdetq **p;
1268 int npkts = 0;
1269
1270 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1271 if (++npkts > MAX_UPQ6) {
1272 mrt6stat.mrt6s_upq_ovflw++;
1273 free(rte, M_MRTABLE);
1274 m_freem(mb0);
1275 splx(s);
1276 return 0;
1277 }
1278
1279 /* Add this entry to the end of the queue */
1280 *p = rte;
1281 }
1282
1283 rte->next = NULL;
1284 rte->m = mb0;
1285 rte->ifp = ifp;
1286 #ifdef UPCALL_TIMING
1287 rte->t = tp;
1288 #endif /* UPCALL_TIMING */
1289
1290 splx(s);
1291
1292 return 0;
1293 }
1294 }
1295
1296 /*
1297 * Clean up cache entries if upcalls are not serviced
1298 * Call from the Slow Timeout mechanism, every half second.
1299 */
1300 static void
1301 expire_upcalls(unused)
1302 void *unused;
1303 {
1304 struct rtdetq *rte;
1305 struct mf6c *mfc, **nptr;
1306 int i;
1307 int s;
1308
1309 s = splsoftnet();
1310 for (i = 0; i < MF6CTBLSIZ; i++) {
1311 if (n6expire[i] == 0)
1312 continue;
1313 nptr = &mf6ctable[i];
1314 while ((mfc = *nptr) != NULL) {
1315 rte = mfc->mf6c_stall;
1316 /*
1317 * Skip real cache entries
1318 * Make sure it wasn't marked to not expire (shouldn't happen)
1319 * If it expires now
1320 */
1321 if (rte != NULL &&
1322 mfc->mf6c_expire != 0 &&
1323 --mfc->mf6c_expire == 0) {
1324 #ifdef MRT6DEBUG
1325 if (mrt6debug & DEBUG_EXPIRE)
1326 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1327 ip6_sprintf(&mfc->mf6c_origin.sin6_addr),
1328 ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr));
1329 #endif
1330 /*
1331 * drop all the packets
1332 * free the mbuf with the pkt, if, timing info
1333 */
1334 do {
1335 struct rtdetq *n = rte->next;
1336 m_freem(rte->m);
1337 free(rte, M_MRTABLE);
1338 rte = n;
1339 } while (rte != NULL);
1340 mrt6stat.mrt6s_cache_cleanups++;
1341 n6expire[i]--;
1342
1343 *nptr = mfc->mf6c_next;
1344 free(mfc, M_MRTABLE);
1345 } else {
1346 nptr = &mfc->mf6c_next;
1347 }
1348 }
1349 }
1350 splx(s);
1351 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1352 expire_upcalls, NULL);
1353 }
1354
1355 /*
1356 * Packet forwarding routine once entry in the cache is made
1357 */
1358 static int
1359 ip6_mdq(m, ifp, rt)
1360 struct mbuf *m;
1361 struct ifnet *ifp;
1362 struct mf6c *rt;
1363 {
1364 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1365 mifi_t mifi, iif;
1366 struct mif6 *mifp;
1367 int plen = m->m_pkthdr.len;
1368
1369 /*
1370 * Macro to send packet on mif. Since RSVP packets don't get counted on
1371 * input, they shouldn't get counted on output, so statistics keeping is
1372 * separate.
1373 */
1374
1375 #define MC6_SEND(ip6, mifp, m) do { \
1376 if ((mifp)->m6_flags & MIFF_REGISTER) \
1377 register_send((ip6), (mifp), (m)); \
1378 else \
1379 phyint_send((ip6), (mifp), (m)); \
1380 } while (/*CONSTCOND*/ 0)
1381
1382 /*
1383 * Don't forward if it didn't arrive from the parent mif
1384 * for its origin.
1385 */
1386 mifi = rt->mf6c_parent;
1387 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1388 /* came in the wrong interface */
1389 #ifdef MRT6DEBUG
1390 if (mrt6debug & DEBUG_FORWARD)
1391 log(LOG_DEBUG,
1392 "wrong if: ifid %d mifi %d mififid %x\n",
1393 ifp->if_index, mifi,
1394 mif6table[mifi].m6_ifp ?
1395 mif6table[mifi].m6_ifp->if_index : -1);
1396 #endif
1397 mrt6stat.mrt6s_wrong_if++;
1398 rt->mf6c_wrong_if++;
1399 /*
1400 * If we are doing PIM processing, and we are forwarding
1401 * packets on this interface, send a message to the
1402 * routing daemon.
1403 */
1404 /* have to make sure this is a valid mif */
1405 if (mifi < nummifs && mif6table[mifi].m6_ifp)
1406 if (pim6 && (m->m_flags & M_LOOP) == 0) {
1407 /*
1408 * Check the M_LOOP flag to avoid an
1409 * unnecessary PIM assert.
1410 * XXX: M_LOOP is an ad-hoc hack...
1411 */
1412 static struct sockaddr_in6 sin6 =
1413 { sizeof(sin6), AF_INET6 };
1414
1415 struct mbuf *mm;
1416 struct mrt6msg *im;
1417 struct omrt6msg *oim;
1418
1419 mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1420 if (mm &&
1421 (M_READONLY(mm) ||
1422 mm->m_len < sizeof(struct ip6_hdr)))
1423 mm = m_pullup(mm, sizeof(struct ip6_hdr));
1424 if (mm == NULL)
1425 return ENOBUFS;
1426
1427 oim = NULL;
1428 im = NULL;
1429 switch (ip6_mrouter_ver) {
1430 case MRT6_OINIT:
1431 oim = mtod(mm, struct omrt6msg *);
1432 oim->im6_msgtype = MRT6MSG_WRONGMIF;
1433 oim->im6_mbz = 0;
1434 break;
1435 case MRT6_INIT:
1436 im = mtod(mm, struct mrt6msg *);
1437 im->im6_msgtype = MRT6MSG_WRONGMIF;
1438 im->im6_mbz = 0;
1439 break;
1440 default:
1441 m_freem(mm);
1442 return EINVAL;
1443 }
1444
1445 for (mifp = mif6table, iif = 0;
1446 iif < nummifs && mifp &&
1447 mifp->m6_ifp != ifp;
1448 mifp++, iif++)
1449 ;
1450
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;
1460 }
1461
1462 mrt6stat.mrt6s_upcalls++;
1463
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 */
1475
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;
1484 }
1485 rt->mf6c_pkt_cnt++;
1486 rt->mf6c_byte_cnt += plen;
1487
1488 /*
1489 * For each mif, forward a copy of the packet if there are group
1490 * members downstream on the interface.
1491 */
1492 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++)
1493 if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1494 if (mif6table[mifi].m6_ifp == NULL)
1495 continue;
1496 #ifdef notyet
1497 /*
1498 * check if the outgoing packet is going to break
1499 * a scope boundary.
1500 * XXX For packets through PIM register tunnel
1501 * interface, we believe a routing daemon.
1502 */
1503 if ((mif6table[rt->mf6c_parent].m6_flags &
1504 MIFF_REGISTER) == 0 &&
1505 (mif6table[mifi].m6_flags & MIFF_REGISTER) == 0 &&
1506 (in6_addr2scopeid(ifp, &ip6->ip6_dst) !=
1507 in6_addr2scopeid(mif6table[mifi].m6_ifp,
1508 &ip6->ip6_dst) ||
1509 in6_addr2scopeid(ifp, &ip6->ip6_src) !=
1510 in6_addr2scopeid(mif6table[mifi].m6_ifp,
1511 &ip6->ip6_src))) {
1512 ip6stat.ip6s_badscope++;
1513 continue;
1514 }
1515 #endif
1516
1517 mifp->m6_pkt_out++;
1518 mifp->m6_bytes_out += plen;
1519 MC6_SEND(ip6, mifp, m);
1520 }
1521 return 0;
1522 }
1523
1524 static void
1525 phyint_send(ip6, mifp, m)
1526 struct ip6_hdr *ip6;
1527 struct mif6 *mifp;
1528 struct mbuf *m;
1529 {
1530 struct mbuf *mb_copy;
1531 struct ifnet *ifp = mifp->m6_ifp;
1532 int error = 0;
1533 int s = splsoftnet();
1534 static struct route_in6 ro;
1535 struct in6_multi *in6m;
1536 struct sockaddr_in6 *dst6;
1537 u_long linkmtu;
1538
1539 /*
1540 * Make a new reference to the packet; make sure that
1541 * the IPv6 header is actually copied, not just referenced,
1542 * so that ip6_output() only scribbles on the copy.
1543 */
1544 mb_copy = m_copy(m, 0, M_COPYALL);
1545 if (mb_copy &&
1546 (M_READONLY(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1547 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1548 if (mb_copy == NULL) {
1549 splx(s);
1550 return;
1551 }
1552 /* set MCAST flag to the outgoing packet */
1553 mb_copy->m_flags |= M_MCAST;
1554
1555 /*
1556 * If we sourced the packet, call ip6_output since we may devide
1557 * the packet into fragments when the packet is too big for the
1558 * outgoing interface.
1559 * Otherwise, we can simply send the packet to the interface
1560 * sending queue.
1561 */
1562 if (m->m_pkthdr.rcvif == NULL) {
1563 struct ip6_moptions im6o;
1564
1565 im6o.im6o_multicast_ifp = ifp;
1566 /* XXX: ip6_output will override ip6->ip6_hlim */
1567 im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1568 im6o.im6o_multicast_loop = 1;
1569 error = ip6_output(mb_copy, NULL, &ro, IPV6_FORWARDING,
1570 &im6o, (struct socket *)0, NULL);
1571
1572 #ifdef MRT6DEBUG
1573 if (mrt6debug & DEBUG_XMIT)
1574 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1575 mifp - mif6table, error);
1576 #endif
1577 splx(s);
1578 return;
1579 }
1580
1581 /*
1582 * If we belong to the destination multicast group
1583 * on the outgoing interface, loop back a copy.
1584 */
1585 dst6 = (struct sockaddr_in6 *)&ro.ro_dst;
1586 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m);
1587 if (in6m != NULL) {
1588 dst6->sin6_len = sizeof(struct sockaddr_in6);
1589 dst6->sin6_family = AF_INET6;
1590 dst6->sin6_addr = ip6->ip6_dst;
1591 ip6_mloopback(ifp, m, (struct sockaddr_in6 *)&ro.ro_dst);
1592 }
1593 /*
1594 * Put the packet into the sending queue of the outgoing interface
1595 * if it would fit in the MTU of the interface.
1596 */
1597 linkmtu = IN6_LINKMTU(ifp);
1598 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1599 dst6->sin6_len = sizeof(struct sockaddr_in6);
1600 dst6->sin6_family = AF_INET6;
1601 dst6->sin6_addr = ip6->ip6_dst;
1602 /*
1603 * We just call if_output instead of nd6_output here, since
1604 * we need no ND for a multicast forwarded packet...right?
1605 */
1606 error = (*ifp->if_output)(ifp, mb_copy,
1607 (struct sockaddr *)&ro.ro_dst, NULL);
1608 #ifdef MRT6DEBUG
1609 if (mrt6debug & DEBUG_XMIT)
1610 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1611 mifp - mif6table, error);
1612 #endif
1613 } else {
1614 #ifdef MULTICAST_PMTUD
1615 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1616 #else
1617 #ifdef MRT6DEBUG
1618 if (mrt6debug & DEBUG_XMIT)
1619 log(LOG_DEBUG,
1620 "phyint_send: packet too big on %s o %s g %s"
1621 " size %d(discarded)\n",
1622 if_name(ifp),
1623 ip6_sprintf(&ip6->ip6_src),
1624 ip6_sprintf(&ip6->ip6_dst),
1625 mb_copy->m_pkthdr.len);
1626 #endif /* MRT6DEBUG */
1627 m_freem(mb_copy); /* simply discard the packet */
1628 #endif
1629 }
1630
1631 splx(s);
1632 }
1633
1634 static int
1635 register_send(ip6, mif, m)
1636 struct ip6_hdr *ip6;
1637 struct mif6 *mif;
1638 struct mbuf *m;
1639 {
1640 struct mbuf *mm;
1641 int i, len = m->m_pkthdr.len;
1642 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1643 struct mrt6msg *im6;
1644
1645 #ifdef MRT6DEBUG
1646 if (mrt6debug)
1647 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1648 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst));
1649 #endif
1650 ++pim6stat.pim6s_snd_registers;
1651
1652 /* Make a copy of the packet to send to the user level process */
1653 MGETHDR(mm, M_DONTWAIT, MT_HEADER);
1654 if (mm == NULL)
1655 return ENOBUFS;
1656 mm->m_data += max_linkhdr;
1657 mm->m_len = sizeof(struct ip6_hdr);
1658
1659 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1660 m_freem(mm);
1661 return ENOBUFS;
1662 }
1663 i = MHLEN - M_LEADINGSPACE(mm);
1664 if (i > len)
1665 i = len;
1666 mm = m_pullup(mm, i);
1667 if (mm == NULL)
1668 return ENOBUFS;
1669 /* TODO: check it! */
1670 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1671
1672 /*
1673 * Send message to routing daemon
1674 */
1675 sin6.sin6_addr = ip6->ip6_src;
1676
1677 im6 = mtod(mm, struct mrt6msg *);
1678 im6->im6_msgtype = MRT6MSG_WHOLEPKT;
1679 im6->im6_mbz = 0;
1680
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 return 0;
1696 }
1697
1698 /*
1699 * PIM sparse mode hook
1700 * Receives the pim control messages, and passes them up to the listening
1701 * socket, using rip6_input.
1702 * The only message processed is the REGISTER pim message; the pim header
1703 * is stripped off, and the inner packet is passed to register_mforward.
1704 */
1705 int
1706 pim6_input(mp, offp, proto)
1707 struct mbuf **mp;
1708 int *offp, proto;
1709 {
1710 struct pim *pim; /* pointer to a pim struct */
1711 struct ip6_hdr *ip6;
1712 int pimlen;
1713 struct mbuf *m = *mp;
1714 int minlen;
1715 int off = *offp;
1716
1717 ++pim6stat.pim6s_rcv_total;
1718
1719 ip6 = mtod(m, struct ip6_hdr *);
1720 pimlen = m->m_pkthdr.len - *offp;
1721
1722 /*
1723 * Validate lengths
1724 */
1725 if (pimlen < PIM_MINLEN) {
1726 ++pim6stat.pim6s_rcv_tooshort;
1727 #ifdef MRT6DEBUG
1728 if (mrt6debug & DEBUG_PIM)
1729 log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1730 #endif
1731 m_freem(m);
1732 return (IPPROTO_DONE);
1733 }
1734
1735 /*
1736 * if the packet is at least as big as a REGISTER, go ahead
1737 * and grab the PIM REGISTER header size, to avoid another
1738 * possible m_pullup() later.
1739 *
1740 * PIM_MINLEN == pimhdr + u_int32 == 8
1741 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1742 */
1743 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1744
1745 /*
1746 * Make sure that the IP6 and PIM headers in contiguous memory, and
1747 * possibly the PIM REGISTER header
1748 */
1749 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1750 if (pim == NULL) {
1751 pim6stat.pim6s_rcv_tooshort++;
1752 return IPPROTO_DONE;
1753 }
1754
1755 /* PIM version check */
1756 if (pim->pim_ver != PIM_VERSION) {
1757 ++pim6stat.pim6s_rcv_badversion;
1758 #ifdef MRT6DEBUG
1759 log(LOG_ERR,
1760 "pim6_input: incorrect version %d, expecting %d\n",
1761 pim->pim_ver, PIM_VERSION);
1762 #endif
1763 m_freem(m);
1764 return (IPPROTO_DONE);
1765 }
1766
1767 #define PIM6_CHECKSUM
1768 #ifdef PIM6_CHECKSUM
1769 {
1770 int cksumlen;
1771
1772 /*
1773 * Validate checksum.
1774 * If PIM REGISTER, exclude the data packet
1775 */
1776 if (pim->pim_type == PIM_REGISTER)
1777 cksumlen = PIM_MINLEN;
1778 else
1779 cksumlen = pimlen;
1780
1781 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1782 ++pim6stat.pim6s_rcv_badsum;
1783 #ifdef MRT6DEBUG
1784 if (mrt6debug & DEBUG_PIM)
1785 log(LOG_DEBUG,
1786 "pim6_input: invalid checksum\n");
1787 #endif
1788 m_freem(m);
1789 return (IPPROTO_DONE);
1790 }
1791 }
1792 #endif /* PIM_CHECKSUM */
1793
1794 if (pim->pim_type == PIM_REGISTER) {
1795 /*
1796 * since this is a REGISTER, we'll make a copy of the register
1797 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1798 * routing daemon.
1799 */
1800 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1801
1802 struct mbuf *mcp;
1803 struct ip6_hdr *eip6;
1804 u_int32_t *reghdr;
1805
1806 ++pim6stat.pim6s_rcv_registers;
1807
1808 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1809 #ifdef MRT6DEBUG
1810 if (mrt6debug & DEBUG_PIM)
1811 log(LOG_DEBUG,
1812 "pim6_input: register mif not set: %d\n",
1813 reg_mif_num);
1814 #endif
1815 m_freem(m);
1816 return (IPPROTO_DONE);
1817 }
1818
1819 reghdr = (u_int32_t *)(pim + 1);
1820
1821 if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1822 goto pim6_input_to_daemon;
1823
1824 /*
1825 * Validate length
1826 */
1827 if (pimlen < PIM6_REG_MINLEN) {
1828 ++pim6stat.pim6s_rcv_tooshort;
1829 ++pim6stat.pim6s_rcv_badregisters;
1830 #ifdef MRT6DEBUG
1831 log(LOG_ERR,
1832 "pim6_input: register packet size too "
1833 "small %d from %s\n",
1834 pimlen, ip6_sprintf(&ip6->ip6_src));
1835 #endif
1836 m_freem(m);
1837 return (IPPROTO_DONE);
1838 }
1839
1840 eip6 = (struct ip6_hdr *) (reghdr + 1);
1841 #ifdef MRT6DEBUG
1842 if (mrt6debug & DEBUG_PIM)
1843 log(LOG_DEBUG,
1844 "pim6_input[register], eip6: %s -> %s, "
1845 "eip6 plen %d\n",
1846 ip6_sprintf(&eip6->ip6_src),
1847 ip6_sprintf(&eip6->ip6_dst),
1848 ntohs(eip6->ip6_plen));
1849 #endif
1850
1851 /* verify the version number of the inner packet */
1852 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1853 ++pim6stat.pim6s_rcv_badregisters;
1854 #ifdef MRT6DEBUG
1855 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1856 "of the inner packet\n",
1857 (eip6->ip6_vfc & IPV6_VERSION));
1858 #endif
1859 m_freem(m);
1860 return (IPPROTO_NONE);
1861 }
1862
1863 /* verify the inner packet is destined to a mcast group */
1864 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1865 ++pim6stat.pim6s_rcv_badregisters;
1866 #ifdef MRT6DEBUG
1867 if (mrt6debug & DEBUG_PIM)
1868 log(LOG_DEBUG,
1869 "pim6_input: inner packet of register "
1870 "is not multicast %s\n",
1871 ip6_sprintf(&eip6->ip6_dst));
1872 #endif
1873 m_freem(m);
1874 return (IPPROTO_DONE);
1875 }
1876
1877 /*
1878 * make a copy of the whole header to pass to the daemon later.
1879 */
1880 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1881 if (mcp == NULL) {
1882 #ifdef MRT6DEBUG
1883 log(LOG_ERR,
1884 "pim6_input: pim register: "
1885 "could not copy register head\n");
1886 #endif
1887 m_freem(m);
1888 return (IPPROTO_DONE);
1889 }
1890
1891 /*
1892 * forward the inner ip6 packet; point m_data at the inner ip6.
1893 */
1894 m_adj(m, off + PIM_MINLEN);
1895 #ifdef MRT6DEBUG
1896 if (mrt6debug & DEBUG_PIM) {
1897 log(LOG_DEBUG,
1898 "pim6_input: forwarding decapsulated register: "
1899 "src %s, dst %s, mif %d\n",
1900 ip6_sprintf(&eip6->ip6_src),
1901 ip6_sprintf(&eip6->ip6_dst),
1902 reg_mif_num);
1903 }
1904 #endif
1905
1906 looutput(mif6table[reg_mif_num].m6_ifp, m,
1907 (struct sockaddr *) &dst,
1908 (struct rtentry *) NULL);
1909
1910 /* prepare the register head to send to the mrouting daemon */
1911 m = mcp;
1912 }
1913
1914 /*
1915 * Pass the PIM message up to the daemon; if it is a register message
1916 * pass the 'head' only up to the daemon. This includes the
1917 * encapsulator ip6 header, pim header, register header and the
1918 * encapsulated ip6 header.
1919 */
1920 pim6_input_to_daemon:
1921 rip6_input(&m, offp, proto);
1922 return (IPPROTO_DONE);
1923 }
Cache object: d98d62c1727930cb0981928ee1609f17
|