ip6_mroute.c revision 1.50 1 /* $NetBSD: ip6_mroute.c,v 1.50 2003/07/08 10:20:45 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) 1989 Stephen Deering
37 * Copyright (c) 1992, 1993
38 * The Regents of the University of California. All rights reserved.
39 *
40 * This code is derived from software contributed to Berkeley by
41 * Stephen Deering of Stanford University.
42 *
43 * Redistribution and use in source and binary forms, with or without
44 * modification, are permitted provided that the following conditions
45 * are met:
46 * 1. Redistributions of source code must retain the above copyright
47 * notice, this list of conditions and the following disclaimer.
48 * 2. Redistributions in binary form must reproduce the above copyright
49 * notice, this list of conditions and the following disclaimer in the
50 * documentation and/or other materials provided with the distribution.
51 * 3. All advertising materials mentioning features or use of this software
52 * must display the following acknowledgement:
53 * This product includes software developed by the University of
54 * California, Berkeley and its contributors.
55 * 4. Neither the name of the University nor the names of its contributors
56 * may be used to endorse or promote products derived from this software
57 * without specific prior written permission.
58 *
59 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
60 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
61 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
62 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
63 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
64 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
65 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
66 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
67 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
68 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
69 * SUCH DAMAGE.
70 *
71 * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93
72 */
73
74 /*
75 * IP multicast forwarding procedures
76 *
77 * Written by David Waitzman, BBN Labs, August 1988.
78 * Modified by Steve Deering, Stanford, February 1989.
79 * Modified by Mark J. Steiglitz, Stanford, May, 1991
80 * Modified by Van Jacobson, LBL, January 1993
81 * Modified by Ajit Thyagarajan, PARC, August 1993
82 * Modified by Bill Fenner, PARC, April 1994
83 *
84 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
85 */
86
87 #include <sys/cdefs.h>
88 __KERNEL_RCSID(0, "$NetBSD: ip6_mroute.c,v 1.50 2003/07/08 10:20:45 itojun Exp $");
89
90 #include "opt_inet.h"
91 #include "opt_mrouting.h"
92
93 #ifndef _KERNEL
94 # ifdef KERNEL
95 # define _KERNEL
96 # endif
97 #endif
98
99 #include <sys/param.h>
100 #include <sys/systm.h>
101 #include <sys/callout.h>
102 #include <sys/mbuf.h>
103 #include <sys/socket.h>
104 #include <sys/socketvar.h>
105 #include <sys/sockio.h>
106 #include <sys/protosw.h>
107 #include <sys/errno.h>
108 #include <sys/time.h>
109 #include <sys/kernel.h>
110 #include <sys/ioctl.h>
111 #include <sys/syslog.h>
112
113 #include <net/if.h>
114 #include <net/route.h>
115 #include <net/raw_cb.h>
116
117 #include <netinet/in.h>
118 #include <netinet/in_var.h>
119 #ifdef MULTICAST_PMTUD
120 #include <netinet/icmp6.h>
121 #endif
122
123 #include <netinet/ip6.h>
124 #include <netinet6/ip6_var.h>
125 #include <netinet6/ip6_mroute.h>
126 #include <netinet6/pim6.h>
127 #include <netinet6/pim6_var.h>
128 #include <netinet6/nd6.h>
129
130 #include <net/net_osdep.h>
131
132 static int ip6_mdq __P((struct mbuf *, struct ifnet *, struct mf6c *));
133 static void phyint_send __P((struct ip6_hdr *, struct mif6 *, struct mbuf *));
134
135 static int set_pim6 __P((int *));
136 static int get_pim6 __P((struct mbuf *));
137 static int socket_send __P((struct socket *, struct mbuf *,
138 struct sockaddr_in6 *));
139 static int register_send __P((struct ip6_hdr *, struct mif6 *,
140 struct mbuf *));
141
142 /*
143 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static,
144 * except for netstat or debugging purposes.
145 */
146 struct socket *ip6_mrouter = NULL;
147 int ip6_mrouter_ver = 0;
148 int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */
149 struct mrt6stat mrt6stat;
150
151 #define NO_RTE_FOUND 0x1
152 #define RTE_FOUND 0x2
153
154 struct mf6c *mf6ctable[MF6CTBLSIZ];
155 u_char n6expire[MF6CTBLSIZ];
156 static struct mif6 mif6table[MAXMIFS];
157 #ifdef MRT6DEBUG
158 u_int mrt6debug = 0; /* debug level */
159 #define DEBUG_MFC 0x02
160 #define DEBUG_FORWARD 0x04
161 #define DEBUG_EXPIRE 0x08
162 #define DEBUG_XMIT 0x10
163 #define DEBUG_REG 0x20
164 #define DEBUG_PIM 0x40
165 #endif
166
167 static void expire_upcalls __P((void *));
168 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */
169 #define UPCALL_EXPIRE 6 /* number of timeouts */
170
171 #ifdef INET
172 #ifdef MROUTING
173 extern struct socket *ip_mrouter;
174 #endif
175 #endif
176
177 /*
178 * 'Interfaces' associated with decapsulator (so we can tell
179 * packets that went through it from ones that get reflected
180 * by a broken gateway). These interfaces are never linked into
181 * the system ifnet list & no routes point to them. I.e., packets
182 * can't be sent this way. They only exist as a placeholder for
183 * multicast source verification.
184 */
185 struct ifnet multicast_register_if;
186
187 #define ENCAP_HOPS 64
188
189 /*
190 * Private variables.
191 */
192 static mifi_t nummifs = 0;
193 static mifi_t reg_mif_num = (mifi_t)-1;
194
195 static struct pim6stat pim6stat;
196 static int pim6;
197
198 /*
199 * Hash function for a source, group entry
200 */
201 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
202 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
203 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
204 (g).s6_addr32[2] ^ (g).s6_addr32[3])
205
206 /*
207 * Find a route for a given origin IPv6 address and Multicast group address.
208 * Quality of service parameter to be added in the future!!!
209 */
210
211 #define MF6CFIND(o, g, rt) do { \
212 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
213 rt = NULL; \
214 mrt6stat.mrt6s_mfc_lookups++; \
215 while (_rt) { \
216 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
217 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
218 (_rt->mf6c_stall == NULL)) { \
219 rt = _rt; \
220 break; \
221 } \
222 _rt = _rt->mf6c_next; \
223 } \
224 if (rt == NULL) { \
225 mrt6stat.mrt6s_mfc_misses++; \
226 } \
227 } while (/*CONSTCOND*/ 0)
228
229 /*
230 * Macros to compute elapsed time efficiently
231 * Borrowed from Van Jacobson's scheduling code
232 */
233 #define TV_DELTA(a, b, delta) do { \
234 int xxs; \
235 \
236 delta = (a).tv_usec - (b).tv_usec; \
237 if ((xxs = (a).tv_sec - (b).tv_sec)) { \
238 switch (xxs) { \
239 case 2: \
240 delta += 1000000; \
241 /* FALLTHROUGH */ \
242 case 1: \
243 delta += 1000000; \
244 break; \
245 default: \
246 delta += (1000000 * xxs); \
247 } \
248 } \
249 } while (/*CONSTCOND*/ 0)
250
251 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
252 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
253
254 #ifdef UPCALL_TIMING
255 #define UPCALL_MAX 50
256 u_long upcall_data[UPCALL_MAX + 1];
257 static void collate();
258 #endif /* UPCALL_TIMING */
259
260 static int get_sg_cnt __P((struct sioc_sg_req6 *));
261 static int get_mif6_cnt __P((struct sioc_mif_req6 *));
262 static int ip6_mrouter_init __P((struct socket *, struct mbuf *, int));
263 static int add_m6if __P((struct mif6ctl *));
264 static int del_m6if __P((mifi_t *));
265 static int add_m6fc __P((struct mf6cctl *));
266 static int del_m6fc __P((struct mf6cctl *));
267
268 static struct callout expire_upcalls_ch = CALLOUT_INITIALIZER;
269
270 /*
271 * Handle MRT setsockopt commands to modify the multicast routing tables.
272 */
273 int
274 ip6_mrouter_set(cmd, so, m)
275 int cmd;
276 struct socket *so;
277 struct mbuf *m;
278 {
279 if (cmd != MRT6_INIT && so != ip6_mrouter)
280 return EACCES;
281
282 switch (cmd) {
283 case MRT6_OINIT: return ip6_mrouter_init(so, m, cmd);
284 case MRT6_INIT: return ip6_mrouter_init(so, m, cmd);
285 case MRT6_DONE: return ip6_mrouter_done();
286 case MRT6_ADD_MIF: return add_m6if(mtod(m, struct mif6ctl *));
287 case MRT6_DEL_MIF: return del_m6if(mtod(m, mifi_t *));
288 case MRT6_ADD_MFC: return add_m6fc(mtod(m, struct mf6cctl *));
289 case MRT6_DEL_MFC: return del_m6fc(mtod(m, struct mf6cctl *));
290 case MRT6_PIM: return set_pim6(mtod(m, int *));
291 default: return EOPNOTSUPP;
292 }
293 }
294
295 /*
296 * Handle MRT getsockopt commands
297 */
298 int
299 ip6_mrouter_get(cmd, so, m)
300 int cmd;
301 struct socket *so;
302 struct mbuf **m;
303 {
304 struct mbuf *mb;
305
306 if (so != ip6_mrouter) return EACCES;
307
308 *m = mb = m_get(M_WAIT, MT_SOOPTS);
309
310 switch (cmd) {
311 case MRT6_PIM:
312 return get_pim6(mb);
313 default:
314 m_free(mb);
315 return EOPNOTSUPP;
316 }
317 }
318
319 /*
320 * Handle ioctl commands to obtain information from the cache
321 */
322 int
323 mrt6_ioctl(cmd, data)
324 int cmd;
325 caddr_t data;
326 {
327
328 switch (cmd) {
329 case SIOCGETSGCNT_IN6:
330 return (get_sg_cnt((struct sioc_sg_req6 *)data));
331 case SIOCGETMIFCNT_IN6:
332 return (get_mif6_cnt((struct sioc_mif_req6 *)data));
333 default:
334 return (EINVAL);
335 }
336 }
337
338 /*
339 * returns the packet, byte, rpf-failure count for the source group provided
340 */
341 static int
342 get_sg_cnt(req)
343 struct sioc_sg_req6 *req;
344 {
345 struct mf6c *rt;
346 int s;
347
348 s = splsoftnet();
349 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
350 splx(s);
351 if (rt != NULL) {
352 req->pktcnt = rt->mf6c_pkt_cnt;
353 req->bytecnt = rt->mf6c_byte_cnt;
354 req->wrong_if = rt->mf6c_wrong_if;
355 } else
356 return (ESRCH);
357 #if 0
358 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff;
359 #endif
360
361 return 0;
362 }
363
364 /*
365 * returns the input and output packet and byte counts on the mif provided
366 */
367 static int
368 get_mif6_cnt(req)
369 struct sioc_mif_req6 *req;
370 {
371 mifi_t mifi = req->mifi;
372
373 if (mifi >= nummifs)
374 return EINVAL;
375
376 req->icount = mif6table[mifi].m6_pkt_in;
377 req->ocount = mif6table[mifi].m6_pkt_out;
378 req->ibytes = mif6table[mifi].m6_bytes_in;
379 req->obytes = mif6table[mifi].m6_bytes_out;
380
381 return 0;
382 }
383
384 /*
385 * Get PIM processiong global
386 */
387 static int
388 get_pim6(m)
389 struct mbuf *m;
390 {
391 int *i;
392
393 i = mtod(m, int *);
394
395 *i = pim6;
396
397 return 0;
398 }
399
400 static int
401 set_pim6(i)
402 int *i;
403 {
404 if ((*i != 1) && (*i != 0))
405 return EINVAL;
406
407 pim6 = *i;
408
409 return 0;
410 }
411
412 /*
413 * Enable multicast routing
414 */
415 static int
416 ip6_mrouter_init(so, m, cmd)
417 struct socket *so;
418 struct mbuf *m;
419 int cmd;
420 {
421 int *v;
422
423 #ifdef MRT6DEBUG
424 if (mrt6debug)
425 log(LOG_DEBUG,
426 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
427 so->so_type, so->so_proto->pr_protocol);
428 #endif
429
430 if (so->so_type != SOCK_RAW ||
431 so->so_proto->pr_protocol != IPPROTO_ICMPV6)
432 return EOPNOTSUPP;
433
434 if (!m || (m->m_len != sizeof(int *)))
435 return ENOPROTOOPT;
436
437 v = mtod(m, int *);
438 if (*v != 1)
439 return ENOPROTOOPT;
440
441 if (ip6_mrouter != NULL) return EADDRINUSE;
442
443 ip6_mrouter = so;
444 ip6_mrouter_ver = cmd;
445
446 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
447 bzero((caddr_t)n6expire, sizeof(n6expire));
448
449 pim6 = 0;/* used for stubbing out/in pim stuff */
450
451 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
452 expire_upcalls, NULL);
453
454 #ifdef MRT6DEBUG
455 if (mrt6debug)
456 log(LOG_DEBUG, "ip6_mrouter_init\n");
457 #endif
458
459 return 0;
460 }
461
462 /*
463 * Disable multicast routing
464 */
465 int
466 ip6_mrouter_done()
467 {
468 mifi_t mifi;
469 int i;
470 struct ifnet *ifp;
471 struct in6_ifreq ifr;
472 struct mf6c *rt;
473 struct rtdetq *rte;
474 int s;
475
476 s = splsoftnet();
477
478 /*
479 * For each phyint in use, disable promiscuous reception of all IPv6
480 * multicasts.
481 */
482 #ifdef INET
483 #ifdef MROUTING
484 /*
485 * If there is still IPv4 multicast routing daemon,
486 * we remain interfaces to receive all muliticasted packets.
487 * XXX: there may be an interface in which the IPv4 multicast
488 * daemon is not interested...
489 */
490 if (!ip_mrouter)
491 #endif
492 #endif
493 {
494 for (mifi = 0; mifi < nummifs; mifi++) {
495 if (mif6table[mifi].m6_ifp &&
496 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
497 ifr.ifr_addr.sin6_family = AF_INET6;
498 ifr.ifr_addr.sin6_addr= in6addr_any;
499 ifp = mif6table[mifi].m6_ifp;
500 (*ifp->if_ioctl)(ifp, SIOCDELMULTI,
501 (caddr_t)&ifr);
502 }
503 }
504 }
505 #ifdef notyet
506 bzero((caddr_t)qtable, sizeof(qtable));
507 bzero((caddr_t)tbftable, sizeof(tbftable));
508 #endif
509 bzero((caddr_t)mif6table, sizeof(mif6table));
510 nummifs = 0;
511
512 pim6 = 0; /* used to stub out/in pim specific code */
513
514 callout_stop(&expire_upcalls_ch);
515
516 /*
517 * Free all multicast forwarding cache entries.
518 */
519 for (i = 0; i < MF6CTBLSIZ; i++) {
520 rt = mf6ctable[i];
521 while (rt) {
522 struct mf6c *frt;
523
524 for (rte = rt->mf6c_stall; rte != NULL; ) {
525 struct rtdetq *n = rte->next;
526
527 m_free(rte->m);
528 free(rte, M_MRTABLE);
529 rte = n;
530 }
531 frt = rt;
532 rt = rt->mf6c_next;
533 free(frt, M_MRTABLE);
534 }
535 }
536
537 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
538
539 /*
540 * Reset de-encapsulation cache
541 */
542 reg_mif_num = -1;
543
544 ip6_mrouter = NULL;
545 ip6_mrouter_ver = 0;
546
547 splx(s);
548
549 #ifdef MRT6DEBUG
550 if (mrt6debug)
551 log(LOG_DEBUG, "ip6_mrouter_done\n");
552 #endif
553
554 return 0;
555 }
556
557 void
558 ip6_mrouter_detach(ifp)
559 struct ifnet *ifp;
560 {
561 struct rtdetq *rte;
562 struct mf6c *mfc;
563 mifi_t mifi;
564 int i;
565
566 /*
567 * Delete a mif which points to ifp.
568 */
569 for (mifi = 0; mifi < nummifs; mifi++)
570 if (mif6table[mifi].m6_ifp == ifp)
571 del_m6if(&mifi);
572
573 /*
574 * Clear rte->ifp of cache entries received on ifp.
575 */
576 for (i = 0; i < MF6CTBLSIZ; i++) {
577 if (n6expire[i] == 0)
578 continue;
579
580 for (mfc = mf6ctable[i]; mfc != NULL; mfc = mfc->mf6c_next) {
581 for (rte = mfc->mf6c_stall; rte != NULL; rte = rte->next) {
582 if (rte->ifp == ifp)
583 rte->ifp = NULL;
584 }
585 }
586 }
587 }
588
589 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
590
591 /*
592 * Add a mif to the mif table
593 */
594 static int
595 add_m6if(mifcp)
596 struct mif6ctl *mifcp;
597 {
598 struct mif6 *mifp;
599 struct ifnet *ifp;
600 struct in6_ifreq ifr;
601 int error, s;
602 #ifdef notyet
603 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi;
604 #endif
605
606 if (mifcp->mif6c_mifi >= MAXMIFS)
607 return EINVAL;
608 mifp = mif6table + mifcp->mif6c_mifi;
609 if (mifp->m6_ifp)
610 return EADDRINUSE; /* XXX: is it appropriate? */
611 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > if_index)
612 return ENXIO;
613 /*
614 * XXX: some OSes can remove ifp and clear ifindex2ifnet[id]
615 * even for id between 0 and if_index.
616 */
617 if ((ifp = ifindex2ifnet[mifcp->mif6c_pifi]) == NULL)
618 return ENXIO;
619
620 if (mifcp->mif6c_flags & MIFF_REGISTER) {
621 if (reg_mif_num == (mifi_t)-1) {
622 strlcpy(multicast_register_if.if_xname,
623 "register_mif",
624 sizeof(multicast_register_if.if_xname));
625 multicast_register_if.if_flags |= IFF_LOOPBACK;
626 multicast_register_if.if_index = mifcp->mif6c_mifi;
627 reg_mif_num = mifcp->mif6c_mifi;
628 }
629
630 ifp = &multicast_register_if;
631
632 } /* if REGISTER */
633 else {
634 /* Make sure the interface supports multicast */
635 if ((ifp->if_flags & IFF_MULTICAST) == 0)
636 return EOPNOTSUPP;
637
638 s = splsoftnet();
639 /*
640 * Enable promiscuous reception of all IPv6 multicasts
641 * from the interface.
642 */
643 ifr.ifr_addr.sin6_family = AF_INET6;
644 ifr.ifr_addr.sin6_addr = in6addr_any;
645 error = (*ifp->if_ioctl)(ifp, SIOCADDMULTI, (caddr_t)&ifr);
646 splx(s);
647 if (error)
648 return error;
649 }
650
651 s = splsoftnet();
652 mifp->m6_flags = mifcp->mif6c_flags;
653 mifp->m6_ifp = ifp;
654 #ifdef notyet
655 /* scaling up here allows division by 1024 in critical code */
656 mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000;
657 #endif
658 /* initialize per mif pkt counters */
659 mifp->m6_pkt_in = 0;
660 mifp->m6_pkt_out = 0;
661 mifp->m6_bytes_in = 0;
662 mifp->m6_bytes_out = 0;
663 splx(s);
664
665 /* Adjust nummifs up if the mifi is higher than nummifs */
666 if (nummifs <= mifcp->mif6c_mifi)
667 nummifs = mifcp->mif6c_mifi + 1;
668
669 #ifdef MRT6DEBUG
670 if (mrt6debug)
671 log(LOG_DEBUG,
672 "add_mif #%d, phyint %s%d\n",
673 mifcp->mif6c_mifi,
674 ifp->if_name, ifp->if_unit);
675 #endif
676
677 return 0;
678 }
679
680 /*
681 * Delete a mif from the mif table
682 */
683 static int
684 del_m6if(mifip)
685 mifi_t *mifip;
686 {
687 struct mif6 *mifp = mif6table + *mifip;
688 mifi_t mifi;
689 struct ifnet *ifp;
690 struct in6_ifreq ifr;
691 int s;
692
693 if (*mifip >= nummifs)
694 return EINVAL;
695 if (mifp->m6_ifp == NULL)
696 return EINVAL;
697
698 s = splsoftnet();
699
700 if (!(mifp->m6_flags & MIFF_REGISTER)) {
701 /*
702 * XXX: what if there is yet IPv4 multicast daemon
703 * using the interface?
704 */
705 ifp = mifp->m6_ifp;
706
707 ifr.ifr_addr.sin6_family = AF_INET6;
708 ifr.ifr_addr.sin6_addr = in6addr_any;
709 (*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr);
710 }
711
712 #ifdef notyet
713 bzero((caddr_t)qtable[*mifip], sizeof(qtable[*mifip]));
714 bzero((caddr_t)mifp->m6_tbf, sizeof(*(mifp->m6_tbf)));
715 #endif
716 bzero((caddr_t)mifp, sizeof (*mifp));
717
718 /* Adjust nummifs down */
719 for (mifi = nummifs; mifi > 0; mifi--)
720 if (mif6table[mifi - 1].m6_ifp)
721 break;
722 nummifs = mifi;
723
724 splx(s);
725
726 #ifdef MRT6DEBUG
727 if (mrt6debug)
728 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
729 #endif
730
731 return 0;
732 }
733
734 /*
735 * Add an mfc entry
736 */
737 static int
738 add_m6fc(mfccp)
739 struct mf6cctl *mfccp;
740 {
741 struct mf6c *rt;
742 u_long hash;
743 struct rtdetq *rte;
744 u_short nstl;
745 int s;
746
747 MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
748 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
749
750 /* If an entry already exists, just update the fields */
751 if (rt) {
752 #ifdef MRT6DEBUG
753 if (mrt6debug & DEBUG_MFC)
754 log(LOG_DEBUG,"add_m6fc update o %s g %s p %x\n",
755 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
756 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
757 mfccp->mf6cc_parent);
758 #endif
759
760 s = splsoftnet();
761 rt->mf6c_parent = mfccp->mf6cc_parent;
762 rt->mf6c_ifset = mfccp->mf6cc_ifset;
763 splx(s);
764 return 0;
765 }
766
767 /*
768 * Find the entry for which the upcall was made and update
769 */
770 s = splsoftnet();
771 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
772 mfccp->mf6cc_mcastgrp.sin6_addr);
773 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
774 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
775 &mfccp->mf6cc_origin.sin6_addr) &&
776 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
777 &mfccp->mf6cc_mcastgrp.sin6_addr) &&
778 (rt->mf6c_stall != NULL)) {
779
780 if (nstl++)
781 log(LOG_ERR,
782 "add_m6fc: %s o %s g %s p %x dbx %p\n",
783 "multiple kernel entries",
784 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
785 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
786 mfccp->mf6cc_parent, rt->mf6c_stall);
787
788 #ifdef MRT6DEBUG
789 if (mrt6debug & DEBUG_MFC)
790 log(LOG_DEBUG,
791 "add_m6fc o %s g %s p %x dbg %x\n",
792 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
793 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
794 mfccp->mf6cc_parent, rt->mf6c_stall);
795 #endif
796
797 rt->mf6c_origin = mfccp->mf6cc_origin;
798 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
799 rt->mf6c_parent = mfccp->mf6cc_parent;
800 rt->mf6c_ifset = mfccp->mf6cc_ifset;
801 /* initialize pkt counters per src-grp */
802 rt->mf6c_pkt_cnt = 0;
803 rt->mf6c_byte_cnt = 0;
804 rt->mf6c_wrong_if = 0;
805
806 rt->mf6c_expire = 0; /* Don't clean this guy up */
807 n6expire[hash]--;
808
809 /* free packets Qed at the end of this entry */
810 for (rte = rt->mf6c_stall; rte != NULL; ) {
811 struct rtdetq *n = rte->next;
812 if (rte->ifp) {
813 ip6_mdq(rte->m, rte->ifp, rt);
814 }
815 m_freem(rte->m);
816 #ifdef UPCALL_TIMING
817 collate(&(rte->t));
818 #endif /* UPCALL_TIMING */
819 free(rte, M_MRTABLE);
820 rte = n;
821 }
822 rt->mf6c_stall = NULL;
823 }
824 }
825
826 /*
827 * It is possible that an entry is being inserted without an upcall
828 */
829 if (nstl == 0) {
830 #ifdef MRT6DEBUG
831 if (mrt6debug & DEBUG_MFC)
832 log(LOG_DEBUG,
833 "add_mfc no upcall h %d o %s g %s p %x\n",
834 hash,
835 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
836 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
837 mfccp->mf6cc_parent);
838 #endif
839
840 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
841
842 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
843 &mfccp->mf6cc_origin.sin6_addr)&&
844 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
845 &mfccp->mf6cc_mcastgrp.sin6_addr)) {
846
847 rt->mf6c_origin = mfccp->mf6cc_origin;
848 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
849 rt->mf6c_parent = mfccp->mf6cc_parent;
850 rt->mf6c_ifset = mfccp->mf6cc_ifset;
851 /* initialize pkt counters per src-grp */
852 rt->mf6c_pkt_cnt = 0;
853 rt->mf6c_byte_cnt = 0;
854 rt->mf6c_wrong_if = 0;
855
856 if (rt->mf6c_expire)
857 n6expire[hash]--;
858 rt->mf6c_expire = 0;
859 }
860 }
861 if (rt == NULL) {
862 /* no upcall, so make a new entry */
863 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE,
864 M_NOWAIT);
865 if (rt == NULL) {
866 splx(s);
867 return ENOBUFS;
868 }
869
870 /* insert new entry at head of hash chain */
871 rt->mf6c_origin = mfccp->mf6cc_origin;
872 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
873 rt->mf6c_parent = mfccp->mf6cc_parent;
874 rt->mf6c_ifset = mfccp->mf6cc_ifset;
875 /* initialize pkt counters per src-grp */
876 rt->mf6c_pkt_cnt = 0;
877 rt->mf6c_byte_cnt = 0;
878 rt->mf6c_wrong_if = 0;
879 rt->mf6c_expire = 0;
880 rt->mf6c_stall = NULL;
881
882 /* link into table */
883 rt->mf6c_next = mf6ctable[hash];
884 mf6ctable[hash] = rt;
885 }
886 }
887 splx(s);
888 return 0;
889 }
890
891 #ifdef UPCALL_TIMING
892 /*
893 * collect delay statistics on the upcalls
894 */
895 static void
896 collate(t)
897 struct timeval *t;
898 {
899 u_long d;
900 struct timeval tp;
901 u_long delta;
902
903 GET_TIME(tp);
904
905 if (TV_LT(*t, tp))
906 {
907 TV_DELTA(tp, *t, delta);
908
909 d = delta >> 10;
910 if (d > UPCALL_MAX)
911 d = UPCALL_MAX;
912
913 ++upcall_data[d];
914 }
915 }
916 #endif /* UPCALL_TIMING */
917
918 /*
919 * Delete an mfc entry
920 */
921 static int
922 del_m6fc(mfccp)
923 struct mf6cctl *mfccp;
924 {
925 struct sockaddr_in6 origin;
926 struct sockaddr_in6 mcastgrp;
927 struct mf6c *rt;
928 struct mf6c **nptr;
929 u_long hash;
930 int s;
931
932 origin = mfccp->mf6cc_origin;
933 mcastgrp = mfccp->mf6cc_mcastgrp;
934 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
935
936 #ifdef MRT6DEBUG
937 if (mrt6debug & DEBUG_MFC)
938 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
939 ip6_sprintf(&origin.sin6_addr),
940 ip6_sprintf(&mcastgrp.sin6_addr));
941 #endif
942
943 s = splsoftnet();
944
945 nptr = &mf6ctable[hash];
946 while ((rt = *nptr) != NULL) {
947 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
948 &rt->mf6c_origin.sin6_addr) &&
949 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
950 &rt->mf6c_mcastgrp.sin6_addr) &&
951 rt->mf6c_stall == NULL)
952 break;
953
954 nptr = &rt->mf6c_next;
955 }
956 if (rt == NULL) {
957 splx(s);
958 return EADDRNOTAVAIL;
959 }
960
961 *nptr = rt->mf6c_next;
962 free(rt, M_MRTABLE);
963
964 splx(s);
965
966 return 0;
967 }
968
969 static int
970 socket_send(s, mm, src)
971 struct socket *s;
972 struct mbuf *mm;
973 struct sockaddr_in6 *src;
974 {
975 if (s) {
976 if (sbappendaddr(&s->so_rcv,
977 (struct sockaddr *)src,
978 mm, (struct mbuf *)0) != 0) {
979 sorwakeup(s);
980 return 0;
981 }
982 }
983 m_freem(mm);
984 return -1;
985 }
986
987 /*
988 * IPv6 multicast forwarding function. This function assumes that the packet
989 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
990 * pointed to by "ifp", and the packet is to be relayed to other networks
991 * that have members of the packet's destination IPv6 multicast group.
992 *
993 * The packet is returned unscathed to the caller, unless it is
994 * erroneous, in which case a non-zero return value tells the caller to
995 * discard it.
996 */
997
998 int
999 ip6_mforward(ip6, ifp, m)
1000 struct ip6_hdr *ip6;
1001 struct ifnet *ifp;
1002 struct mbuf *m;
1003 {
1004 struct mf6c *rt;
1005 struct mif6 *mifp;
1006 struct mbuf *mm;
1007 int s;
1008 mifi_t mifi;
1009
1010 #ifdef MRT6DEBUG
1011 if (mrt6debug & DEBUG_FORWARD)
1012 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1013 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst),
1014 ifp->if_index);
1015 #endif
1016
1017 /*
1018 * Don't forward a packet with Hop limit of zero or one,
1019 * or a packet destined to a local-only group.
1020 */
1021 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_NODELOCAL(&ip6->ip6_dst) ||
1022 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1023 return 0;
1024 ip6->ip6_hlim--;
1025
1026 /*
1027 * Source address check: do not forward packets with unspecified
1028 * source. It was discussed in July 2000, on ipngwg mailing list.
1029 * This is rather more serious than unicast cases, because some
1030 * MLD packets can be sent with the unspecified source address
1031 * (although such packets must normally set 1 to the hop limit field).
1032 */
1033 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1034 ip6stat.ip6s_cantforward++;
1035 if (ip6_log_time + ip6_log_interval < time.tv_sec) {
1036 ip6_log_time = time.tv_sec;
1037 log(LOG_DEBUG,
1038 "cannot forward "
1039 "from %s to %s nxt %d received on %s\n",
1040 ip6_sprintf(&ip6->ip6_src),
1041 ip6_sprintf(&ip6->ip6_dst),
1042 ip6->ip6_nxt,
1043 m->m_pkthdr.rcvif ?
1044 if_name(m->m_pkthdr.rcvif) : "?");
1045 }
1046 return 0;
1047 }
1048
1049 /*
1050 * Determine forwarding mifs from the forwarding cache table
1051 */
1052 s = splsoftnet();
1053 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1054
1055 /* Entry exists, so forward if necessary */
1056 if (rt) {
1057 splx(s);
1058 return (ip6_mdq(m, ifp, rt));
1059 } else {
1060 /*
1061 * If we don't have a route for packet's origin,
1062 * Make a copy of the packet &
1063 * send message to routing daemon
1064 */
1065
1066 struct mbuf *mb0;
1067 struct rtdetq *rte;
1068 u_long hash;
1069 /* int i, npkts;*/
1070 #ifdef UPCALL_TIMING
1071 struct timeval tp;
1072
1073 GET_TIME(tp);
1074 #endif /* UPCALL_TIMING */
1075
1076 mrt6stat.mrt6s_no_route++;
1077 #ifdef MRT6DEBUG
1078 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1079 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1080 ip6_sprintf(&ip6->ip6_src),
1081 ip6_sprintf(&ip6->ip6_dst));
1082 #endif
1083
1084 /*
1085 * Allocate mbufs early so that we don't do extra work if we
1086 * are just going to fail anyway.
1087 */
1088 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE,
1089 M_NOWAIT);
1090 if (rte == NULL) {
1091 splx(s);
1092 return ENOBUFS;
1093 }
1094 mb0 = m_copy(m, 0, M_COPYALL);
1095 /*
1096 * Pullup packet header if needed before storing it,
1097 * as other references may modify it in the meantime.
1098 */
1099 if (mb0 &&
1100 (M_READONLY(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1101 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1102 if (mb0 == NULL) {
1103 free(rte, M_MRTABLE);
1104 splx(s);
1105 return ENOBUFS;
1106 }
1107
1108 /* is there an upcall waiting for this packet? */
1109 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1110 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1111 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1112 &rt->mf6c_origin.sin6_addr) &&
1113 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1114 &rt->mf6c_mcastgrp.sin6_addr) &&
1115 (rt->mf6c_stall != NULL))
1116 break;
1117 }
1118
1119 if (rt == NULL) {
1120 struct mrt6msg *im;
1121 struct omrt6msg *oim;
1122
1123 /* no upcall, so make a new entry */
1124 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE,
1125 M_NOWAIT);
1126 if (rt == NULL) {
1127 free(rte, M_MRTABLE);
1128 m_freem(mb0);
1129 splx(s);
1130 return ENOBUFS;
1131 }
1132 /*
1133 * Make a copy of the header to send to the user
1134 * level process
1135 */
1136 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1137
1138 if (mm == NULL) {
1139 free(rte, M_MRTABLE);
1140 m_freem(mb0);
1141 free(rt, M_MRTABLE);
1142 splx(s);
1143 return ENOBUFS;
1144 }
1145
1146 /*
1147 * Send message to routing daemon
1148 */
1149 sin6.sin6_addr = ip6->ip6_src;
1150
1151 im = NULL;
1152 oim = NULL;
1153 switch (ip6_mrouter_ver) {
1154 case MRT6_OINIT:
1155 oim = mtod(mm, struct omrt6msg *);
1156 oim->im6_msgtype = MRT6MSG_NOCACHE;
1157 oim->im6_mbz = 0;
1158 break;
1159 case MRT6_INIT:
1160 im = mtod(mm, struct mrt6msg *);
1161 im->im6_msgtype = MRT6MSG_NOCACHE;
1162 im->im6_mbz = 0;
1163 break;
1164 default:
1165 free(rte, M_MRTABLE);
1166 m_freem(mb0);
1167 free(rt, M_MRTABLE);
1168 splx(s);
1169 return EINVAL;
1170 }
1171
1172 #ifdef MRT6DEBUG
1173 if (mrt6debug & DEBUG_FORWARD)
1174 log(LOG_DEBUG,
1175 "getting the iif info in the kernel\n");
1176 #endif
1177
1178 for (mifp = mif6table, mifi = 0;
1179 mifi < nummifs && mifp->m6_ifp != ifp;
1180 mifp++, mifi++)
1181 ;
1182
1183 switch (ip6_mrouter_ver) {
1184 case MRT6_OINIT:
1185 oim->im6_mif = mifi;
1186 break;
1187 case MRT6_INIT:
1188 im->im6_mif = mifi;
1189 break;
1190 }
1191
1192 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1193 log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1194 "socket queue full\n");
1195 mrt6stat.mrt6s_upq_sockfull++;
1196 free(rte, M_MRTABLE);
1197 m_freem(mb0);
1198 free(rt, M_MRTABLE);
1199 splx(s);
1200 return ENOBUFS;
1201 }
1202
1203 mrt6stat.mrt6s_upcalls++;
1204
1205 /* insert new entry at head of hash chain */
1206 bzero(rt, sizeof(*rt));
1207 rt->mf6c_origin.sin6_family = AF_INET6;
1208 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1209 rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1210 rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1211 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1212 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1213 rt->mf6c_expire = UPCALL_EXPIRE;
1214 n6expire[hash]++;
1215 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1216
1217 /* link into table */
1218 rt->mf6c_next = mf6ctable[hash];
1219 mf6ctable[hash] = rt;
1220 /* Add this entry to the end of the queue */
1221 rt->mf6c_stall = rte;
1222 } else {
1223 /* determine if q has overflowed */
1224 struct rtdetq **p;
1225 int npkts = 0;
1226
1227 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1228 if (++npkts > MAX_UPQ6) {
1229 mrt6stat.mrt6s_upq_ovflw++;
1230 free(rte, M_MRTABLE);
1231 m_freem(mb0);
1232 splx(s);
1233 return 0;
1234 }
1235
1236 /* Add this entry to the end of the queue */
1237 *p = rte;
1238 }
1239
1240 rte->next = NULL;
1241 rte->m = mb0;
1242 rte->ifp = ifp;
1243 #ifdef UPCALL_TIMING
1244 rte->t = tp;
1245 #endif /* UPCALL_TIMING */
1246
1247 splx(s);
1248
1249 return 0;
1250 }
1251 }
1252
1253 /*
1254 * Clean up cache entries if upcalls are not serviced
1255 * Call from the Slow Timeout mechanism, every half second.
1256 */
1257 static void
1258 expire_upcalls(unused)
1259 void *unused;
1260 {
1261 struct rtdetq *rte;
1262 struct mf6c *mfc, **nptr;
1263 int i;
1264 int s;
1265
1266 s = splsoftnet();
1267 for (i = 0; i < MF6CTBLSIZ; i++) {
1268 if (n6expire[i] == 0)
1269 continue;
1270 nptr = &mf6ctable[i];
1271 while ((mfc = *nptr) != NULL) {
1272 rte = mfc->mf6c_stall;
1273 /*
1274 * Skip real cache entries
1275 * Make sure it wasn't marked to not expire (shouldn't happen)
1276 * If it expires now
1277 */
1278 if (rte != NULL &&
1279 mfc->mf6c_expire != 0 &&
1280 --mfc->mf6c_expire == 0) {
1281 #ifdef MRT6DEBUG
1282 if (mrt6debug & DEBUG_EXPIRE)
1283 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1284 ip6_sprintf(&mfc->mf6c_origin.sin6_addr),
1285 ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr));
1286 #endif
1287 /*
1288 * drop all the packets
1289 * free the mbuf with the pkt, if, timing info
1290 */
1291 do {
1292 struct rtdetq *n = rte->next;
1293 m_freem(rte->m);
1294 free(rte, M_MRTABLE);
1295 rte = n;
1296 } while (rte != NULL);
1297 mrt6stat.mrt6s_cache_cleanups++;
1298 n6expire[i]--;
1299
1300 *nptr = mfc->mf6c_next;
1301 free(mfc, M_MRTABLE);
1302 } else {
1303 nptr = &mfc->mf6c_next;
1304 }
1305 }
1306 }
1307 splx(s);
1308 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1309 expire_upcalls, NULL);
1310 }
1311
1312 /*
1313 * Packet forwarding routine once entry in the cache is made
1314 */
1315 static int
1316 ip6_mdq(m, ifp, rt)
1317 struct mbuf *m;
1318 struct ifnet *ifp;
1319 struct mf6c *rt;
1320 {
1321 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1322 mifi_t mifi, iif;
1323 struct mif6 *mifp;
1324 int plen = m->m_pkthdr.len;
1325
1326 /*
1327 * Macro to send packet on mif. Since RSVP packets don't get counted on
1328 * input, they shouldn't get counted on output, so statistics keeping is
1329 * separate.
1330 */
1331
1332 #define MC6_SEND(ip6, mifp, m) do { \
1333 if ((mifp)->m6_flags & MIFF_REGISTER) \
1334 register_send((ip6), (mifp), (m)); \
1335 else \
1336 phyint_send((ip6), (mifp), (m)); \
1337 } while (/*CONSTCOND*/ 0)
1338
1339 /*
1340 * Don't forward if it didn't arrive from the parent mif
1341 * for its origin.
1342 */
1343 mifi = rt->mf6c_parent;
1344 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1345 /* came in the wrong interface */
1346 #ifdef MRT6DEBUG
1347 if (mrt6debug & DEBUG_FORWARD)
1348 log(LOG_DEBUG,
1349 "wrong if: ifid %d mifi %d mififid %x\n",
1350 ifp->if_index, mifi,
1351 mif6table[mifi].m6_ifp ?
1352 mif6table[mifi].m6_ifp->if_index : -1);
1353 #endif
1354 mrt6stat.mrt6s_wrong_if++;
1355 rt->mf6c_wrong_if++;
1356 /*
1357 * If we are doing PIM processing, and we are forwarding
1358 * packets on this interface, send a message to the
1359 * routing daemon.
1360 */
1361 /* have to make sure this is a valid mif */
1362 if (mifi < nummifs && mif6table[mifi].m6_ifp)
1363 if (pim6 && (m->m_flags & M_LOOP) == 0) {
1364 /*
1365 * Check the M_LOOP flag to avoid an
1366 * unnecessary PIM assert.
1367 * XXX: M_LOOP is an ad-hoc hack...
1368 */
1369 static struct sockaddr_in6 sin6 =
1370 { sizeof(sin6), AF_INET6 };
1371
1372 struct mbuf *mm;
1373 struct mrt6msg *im;
1374 struct omrt6msg *oim;
1375
1376 mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1377 if (mm &&
1378 (M_READONLY(mm) ||
1379 mm->m_len < sizeof(struct ip6_hdr)))
1380 mm = m_pullup(mm, sizeof(struct ip6_hdr));
1381 if (mm == NULL)
1382 return ENOBUFS;
1383
1384 oim = NULL;
1385 im = NULL;
1386 switch (ip6_mrouter_ver) {
1387 case MRT6_OINIT:
1388 oim = mtod(mm, struct omrt6msg *);
1389 oim->im6_msgtype = MRT6MSG_WRONGMIF;
1390 oim->im6_mbz = 0;
1391 break;
1392 case MRT6_INIT:
1393 im = mtod(mm, struct mrt6msg *);
1394 im->im6_msgtype = MRT6MSG_WRONGMIF;
1395 im->im6_mbz = 0;
1396 break;
1397 default:
1398 m_freem(mm);
1399 return EINVAL;
1400 }
1401
1402 for (mifp = mif6table, iif = 0;
1403 iif < nummifs && mifp &&
1404 mifp->m6_ifp != ifp;
1405 mifp++, iif++)
1406 ;
1407
1408 switch (ip6_mrouter_ver) {
1409 case MRT6_OINIT:
1410 oim->im6_mif = iif;
1411 sin6.sin6_addr = oim->im6_src;
1412 break;
1413 case MRT6_INIT:
1414 im->im6_mif = iif;
1415 sin6.sin6_addr = im->im6_src;
1416 break;
1417 }
1418
1419 mrt6stat.mrt6s_upcalls++;
1420
1421 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1422 #ifdef MRT6DEBUG
1423 if (mrt6debug)
1424 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1425 #endif
1426 ++mrt6stat.mrt6s_upq_sockfull;
1427 return ENOBUFS;
1428 } /* if socket Q full */
1429 } /* if PIM */
1430 return 0;
1431 } /* if wrong iif */
1432
1433 /* If I sourced this packet, it counts as output, else it was input. */
1434 if (m->m_pkthdr.rcvif == NULL) {
1435 /* XXX: is rcvif really NULL when output?? */
1436 mif6table[mifi].m6_pkt_out++;
1437 mif6table[mifi].m6_bytes_out += plen;
1438 } else {
1439 mif6table[mifi].m6_pkt_in++;
1440 mif6table[mifi].m6_bytes_in += plen;
1441 }
1442 rt->mf6c_pkt_cnt++;
1443 rt->mf6c_byte_cnt += plen;
1444
1445 /*
1446 * For each mif, forward a copy of the packet if there are group
1447 * members downstream on the interface.
1448 */
1449 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++)
1450 if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1451 if (mif6table[mifi].m6_ifp == NULL)
1452 continue;
1453 #ifdef notyet
1454 /*
1455 * check if the outgoing packet is going to break
1456 * a scope boundary.
1457 * XXX For packets through PIM register tunnel
1458 * interface, we believe a routing daemon.
1459 */
1460 if ((mif6table[rt->mf6c_parent].m6_flags &
1461 MIFF_REGISTER) == 0 &&
1462 (mif6table[mifi].m6_flags & MIFF_REGISTER) == 0 &&
1463 (in6_addr2scopeid(ifp, &ip6->ip6_dst) !=
1464 in6_addr2scopeid(mif6table[mifi].m6_ifp,
1465 &ip6->ip6_dst) ||
1466 in6_addr2scopeid(ifp, &ip6->ip6_src) !=
1467 in6_addr2scopeid(mif6table[mifi].m6_ifp,
1468 &ip6->ip6_src))) {
1469 ip6stat.ip6s_badscope++;
1470 continue;
1471 }
1472 #endif
1473
1474 mifp->m6_pkt_out++;
1475 mifp->m6_bytes_out += plen;
1476 MC6_SEND(ip6, mifp, m);
1477 }
1478 return 0;
1479 }
1480
1481 static void
1482 phyint_send(ip6, mifp, m)
1483 struct ip6_hdr *ip6;
1484 struct mif6 *mifp;
1485 struct mbuf *m;
1486 {
1487 struct mbuf *mb_copy;
1488 struct ifnet *ifp = mifp->m6_ifp;
1489 int error = 0;
1490 int s = splsoftnet();
1491 static struct route_in6 ro;
1492 struct in6_multi *in6m;
1493 struct sockaddr_in6 *dst6;
1494 u_long linkmtu;
1495
1496 /*
1497 * Make a new reference to the packet; make sure that
1498 * the IPv6 header is actually copied, not just referenced,
1499 * so that ip6_output() only scribbles on the copy.
1500 */
1501 mb_copy = m_copy(m, 0, M_COPYALL);
1502 if (mb_copy &&
1503 (M_READONLY(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1504 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1505 if (mb_copy == NULL) {
1506 splx(s);
1507 return;
1508 }
1509 /* set MCAST flag to the outgoing packet */
1510 mb_copy->m_flags |= M_MCAST;
1511
1512 /*
1513 * If we sourced the packet, call ip6_output since we may devide
1514 * the packet into fragments when the packet is too big for the
1515 * outgoing interface.
1516 * Otherwise, we can simply send the packet to the interface
1517 * sending queue.
1518 */
1519 if (m->m_pkthdr.rcvif == NULL) {
1520 struct ip6_moptions im6o;
1521
1522 im6o.im6o_multicast_ifp = ifp;
1523 /* XXX: ip6_output will override ip6->ip6_hlim */
1524 im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1525 im6o.im6o_multicast_loop = 1;
1526 error = ip6_output(mb_copy, NULL, &ro,
1527 IPV6_FORWARDING, &im6o, NULL);
1528
1529 #ifdef MRT6DEBUG
1530 if (mrt6debug & DEBUG_XMIT)
1531 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1532 mifp - mif6table, error);
1533 #endif
1534 splx(s);
1535 return;
1536 }
1537
1538 /*
1539 * If we belong to the destination multicast group
1540 * on the outgoing interface, loop back a copy.
1541 */
1542 dst6 = (struct sockaddr_in6 *)&ro.ro_dst;
1543 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m);
1544 if (in6m != NULL) {
1545 dst6->sin6_len = sizeof(struct sockaddr_in6);
1546 dst6->sin6_family = AF_INET6;
1547 dst6->sin6_addr = ip6->ip6_dst;
1548 ip6_mloopback(ifp, m, (struct sockaddr_in6 *)&ro.ro_dst);
1549 }
1550 /*
1551 * Put the packet into the sending queue of the outgoing interface
1552 * if it would fit in the MTU of the interface.
1553 */
1554 linkmtu = IN6_LINKMTU(ifp);
1555 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1556 dst6->sin6_len = sizeof(struct sockaddr_in6);
1557 dst6->sin6_family = AF_INET6;
1558 dst6->sin6_addr = ip6->ip6_dst;
1559 /*
1560 * We just call if_output instead of nd6_output here, since
1561 * we need no ND for a multicast forwarded packet...right?
1562 */
1563 error = (*ifp->if_output)(ifp, mb_copy,
1564 (struct sockaddr *)&ro.ro_dst, NULL);
1565 #ifdef MRT6DEBUG
1566 if (mrt6debug & DEBUG_XMIT)
1567 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1568 mifp - mif6table, error);
1569 #endif
1570 } else {
1571 #ifdef MULTICAST_PMTUD
1572 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1573 #else
1574 #ifdef MRT6DEBUG
1575 if (mrt6debug & DEBUG_XMIT)
1576 log(LOG_DEBUG,
1577 "phyint_send: packet too big on %s o %s g %s"
1578 " size %d(discarded)\n",
1579 if_name(ifp),
1580 ip6_sprintf(&ip6->ip6_src),
1581 ip6_sprintf(&ip6->ip6_dst),
1582 mb_copy->m_pkthdr.len);
1583 #endif /* MRT6DEBUG */
1584 m_freem(mb_copy); /* simply discard the packet */
1585 #endif
1586 }
1587
1588 splx(s);
1589 }
1590
1591 static int
1592 register_send(ip6, mif, m)
1593 struct ip6_hdr *ip6;
1594 struct mif6 *mif;
1595 struct mbuf *m;
1596 {
1597 struct mbuf *mm;
1598 int i, len = m->m_pkthdr.len;
1599 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1600 struct mrt6msg *im6;
1601
1602 #ifdef MRT6DEBUG
1603 if (mrt6debug)
1604 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1605 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst));
1606 #endif
1607 ++pim6stat.pim6s_snd_registers;
1608
1609 /* Make a copy of the packet to send to the user level process */
1610 MGETHDR(mm, M_DONTWAIT, MT_HEADER);
1611 if (mm == NULL)
1612 return ENOBUFS;
1613 mm->m_data += max_linkhdr;
1614 mm->m_len = sizeof(struct ip6_hdr);
1615
1616 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1617 m_freem(mm);
1618 return ENOBUFS;
1619 }
1620 i = MHLEN - M_LEADINGSPACE(mm);
1621 if (i > len)
1622 i = len;
1623 mm = m_pullup(mm, i);
1624 if (mm == NULL)
1625 return ENOBUFS;
1626 /* TODO: check it! */
1627 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1628
1629 /*
1630 * Send message to routing daemon
1631 */
1632 sin6.sin6_addr = ip6->ip6_src;
1633
1634 im6 = mtod(mm, struct mrt6msg *);
1635 im6->im6_msgtype = MRT6MSG_WHOLEPKT;
1636 im6->im6_mbz = 0;
1637
1638 im6->im6_mif = mif - mif6table;
1639
1640 /* iif info is not given for reg. encap.n */
1641 mrt6stat.mrt6s_upcalls++;
1642
1643 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1644 #ifdef MRT6DEBUG
1645 if (mrt6debug)
1646 log(LOG_WARNING,
1647 "register_send: ip6_mrouter socket queue full\n");
1648 #endif
1649 ++mrt6stat.mrt6s_upq_sockfull;
1650 return ENOBUFS;
1651 }
1652 return 0;
1653 }
1654
1655 /*
1656 * PIM sparse mode hook
1657 * Receives the pim control messages, and passes them up to the listening
1658 * socket, using rip6_input.
1659 * The only message processed is the REGISTER pim message; the pim header
1660 * is stripped off, and the inner packet is passed to register_mforward.
1661 */
1662 int
1663 pim6_input(mp, offp, proto)
1664 struct mbuf **mp;
1665 int *offp, proto;
1666 {
1667 struct pim *pim; /* pointer to a pim struct */
1668 struct ip6_hdr *ip6;
1669 int pimlen;
1670 struct mbuf *m = *mp;
1671 int minlen;
1672 int off = *offp;
1673
1674 ++pim6stat.pim6s_rcv_total;
1675
1676 ip6 = mtod(m, struct ip6_hdr *);
1677 pimlen = m->m_pkthdr.len - *offp;
1678
1679 /*
1680 * Validate lengths
1681 */
1682 if (pimlen < PIM_MINLEN) {
1683 ++pim6stat.pim6s_rcv_tooshort;
1684 #ifdef MRT6DEBUG
1685 if (mrt6debug & DEBUG_PIM)
1686 log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1687 #endif
1688 m_freem(m);
1689 return (IPPROTO_DONE);
1690 }
1691
1692 /*
1693 * if the packet is at least as big as a REGISTER, go ahead
1694 * and grab the PIM REGISTER header size, to avoid another
1695 * possible m_pullup() later.
1696 *
1697 * PIM_MINLEN == pimhdr + u_int32 == 8
1698 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1699 */
1700 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1701
1702 /*
1703 * Make sure that the IP6 and PIM headers in contiguous memory, and
1704 * possibly the PIM REGISTER header
1705 */
1706 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1707 if (pim == NULL) {
1708 pim6stat.pim6s_rcv_tooshort++;
1709 return IPPROTO_DONE;
1710 }
1711
1712 /* PIM version check */
1713 if (pim->pim_ver != PIM_VERSION) {
1714 ++pim6stat.pim6s_rcv_badversion;
1715 #ifdef MRT6DEBUG
1716 log(LOG_ERR,
1717 "pim6_input: incorrect version %d, expecting %d\n",
1718 pim->pim_ver, PIM_VERSION);
1719 #endif
1720 m_freem(m);
1721 return (IPPROTO_DONE);
1722 }
1723
1724 #define PIM6_CHECKSUM
1725 #ifdef PIM6_CHECKSUM
1726 {
1727 int cksumlen;
1728
1729 /*
1730 * Validate checksum.
1731 * If PIM REGISTER, exclude the data packet
1732 */
1733 if (pim->pim_type == PIM_REGISTER)
1734 cksumlen = PIM_MINLEN;
1735 else
1736 cksumlen = pimlen;
1737
1738 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1739 ++pim6stat.pim6s_rcv_badsum;
1740 #ifdef MRT6DEBUG
1741 if (mrt6debug & DEBUG_PIM)
1742 log(LOG_DEBUG,
1743 "pim6_input: invalid checksum\n");
1744 #endif
1745 m_freem(m);
1746 return (IPPROTO_DONE);
1747 }
1748 }
1749 #endif /* PIM_CHECKSUM */
1750
1751 if (pim->pim_type == PIM_REGISTER) {
1752 /*
1753 * since this is a REGISTER, we'll make a copy of the register
1754 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1755 * routing daemon.
1756 */
1757 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1758
1759 struct mbuf *mcp;
1760 struct ip6_hdr *eip6;
1761 u_int32_t *reghdr;
1762 int rc;
1763
1764 ++pim6stat.pim6s_rcv_registers;
1765
1766 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1767 #ifdef MRT6DEBUG
1768 if (mrt6debug & DEBUG_PIM)
1769 log(LOG_DEBUG,
1770 "pim6_input: register mif not set: %d\n",
1771 reg_mif_num);
1772 #endif
1773 m_freem(m);
1774 return (IPPROTO_DONE);
1775 }
1776
1777 reghdr = (u_int32_t *)(pim + 1);
1778
1779 if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1780 goto pim6_input_to_daemon;
1781
1782 /*
1783 * Validate length
1784 */
1785 if (pimlen < PIM6_REG_MINLEN) {
1786 ++pim6stat.pim6s_rcv_tooshort;
1787 ++pim6stat.pim6s_rcv_badregisters;
1788 #ifdef MRT6DEBUG
1789 log(LOG_ERR,
1790 "pim6_input: register packet size too "
1791 "small %d from %s\n",
1792 pimlen, ip6_sprintf(&ip6->ip6_src));
1793 #endif
1794 m_freem(m);
1795 return (IPPROTO_DONE);
1796 }
1797
1798 eip6 = (struct ip6_hdr *) (reghdr + 1);
1799 #ifdef MRT6DEBUG
1800 if (mrt6debug & DEBUG_PIM)
1801 log(LOG_DEBUG,
1802 "pim6_input[register], eip6: %s -> %s, "
1803 "eip6 plen %d\n",
1804 ip6_sprintf(&eip6->ip6_src),
1805 ip6_sprintf(&eip6->ip6_dst),
1806 ntohs(eip6->ip6_plen));
1807 #endif
1808
1809 /* verify the version number of the inner packet */
1810 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1811 ++pim6stat.pim6s_rcv_badregisters;
1812 #ifdef MRT6DEBUG
1813 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1814 "of the inner packet\n",
1815 (eip6->ip6_vfc & IPV6_VERSION));
1816 #endif
1817 m_freem(m);
1818 return (IPPROTO_NONE);
1819 }
1820
1821 /* verify the inner packet is destined to a mcast group */
1822 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1823 ++pim6stat.pim6s_rcv_badregisters;
1824 #ifdef MRT6DEBUG
1825 if (mrt6debug & DEBUG_PIM)
1826 log(LOG_DEBUG,
1827 "pim6_input: inner packet of register "
1828 "is not multicast %s\n",
1829 ip6_sprintf(&eip6->ip6_dst));
1830 #endif
1831 m_freem(m);
1832 return (IPPROTO_DONE);
1833 }
1834
1835 /*
1836 * make a copy of the whole header to pass to the daemon later.
1837 */
1838 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1839 if (mcp == NULL) {
1840 #ifdef MRT6DEBUG
1841 log(LOG_ERR,
1842 "pim6_input: pim register: "
1843 "could not copy register head\n");
1844 #endif
1845 m_freem(m);
1846 return (IPPROTO_DONE);
1847 }
1848
1849 /*
1850 * forward the inner ip6 packet; point m_data at the inner ip6.
1851 */
1852 m_adj(m, off + PIM_MINLEN);
1853 #ifdef MRT6DEBUG
1854 if (mrt6debug & DEBUG_PIM) {
1855 log(LOG_DEBUG,
1856 "pim6_input: forwarding decapsulated register: "
1857 "src %s, dst %s, mif %d\n",
1858 ip6_sprintf(&eip6->ip6_src),
1859 ip6_sprintf(&eip6->ip6_dst),
1860 reg_mif_num);
1861 }
1862 #endif
1863
1864 rc = looutput(mif6table[reg_mif_num].m6_ifp, m,
1865 (struct sockaddr *) &dst,
1866 (struct rtentry *) NULL);
1867
1868 /* prepare the register head to send to the mrouting daemon */
1869 m = mcp;
1870 }
1871
1872 /*
1873 * Pass the PIM message up to the daemon; if it is a register message
1874 * pass the 'head' only up to the daemon. This includes the
1875 * encapsulator ip6 header, pim header, register header and the
1876 * encapsulated ip6 header.
1877 */
1878 pim6_input_to_daemon:
1879 rip6_input(&m, offp, proto);
1880 return (IPPROTO_DONE);
1881 }
1882