1 /*-
2 * Copyright (C) 1998 WIDE Project.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. Neither the name of the project nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * $KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $
30 */
31
32 /*-
33 * Copyright (c) 1989 Stephen Deering
34 * Copyright (c) 1992, 1993
35 * The Regents of the University of California. All rights reserved.
36 *
37 * This code is derived from software contributed to Berkeley by
38 * Stephen Deering of Stanford University.
39 *
40 * Redistribution and use in source and binary forms, with or without
41 * modification, are permitted provided that the following conditions
42 * are met:
43 * 1. Redistributions of source code must retain the above copyright
44 * notice, this list of conditions and the following disclaimer.
45 * 2. Redistributions in binary form must reproduce the above copyright
46 * notice, this list of conditions and the following disclaimer in the
47 * documentation and/or other materials provided with the distribution.
48 * 4. Neither the name of the University nor the names of its contributors
49 * may be used to endorse or promote products derived from this software
50 * without specific prior written permission.
51 *
52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62 * SUCH DAMAGE.
63 *
64 * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93
65 * BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp
66 */
67
68 /*
69 * IP multicast forwarding procedures
70 *
71 * Written by David Waitzman, BBN Labs, August 1988.
72 * Modified by Steve Deering, Stanford, February 1989.
73 * Modified by Mark J. Steiglitz, Stanford, May, 1991
74 * Modified by Van Jacobson, LBL, January 1993
75 * Modified by Ajit Thyagarajan, PARC, August 1993
76 * Modified by Bill Fenner, PARC, April 1994
77 *
78 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
79 */
80
81 #include <sys/cdefs.h>
82 __FBSDID("$FreeBSD$");
83
84 #include "opt_inet.h"
85 #include "opt_inet6.h"
86
87 #include <sys/param.h>
88 #include <sys/callout.h>
89 #include <sys/errno.h>
90 #include <sys/kernel.h>
91 #include <sys/lock.h>
92 #include <sys/malloc.h>
93 #include <sys/mbuf.h>
94 #include <sys/protosw.h>
95 #include <sys/signalvar.h>
96 #include <sys/socket.h>
97 #include <sys/socketvar.h>
98 #include <sys/sockio.h>
99 #include <sys/sx.h>
100 #include <sys/syslog.h>
101 #include <sys/systm.h>
102 #include <sys/time.h>
103
104 #include <net/if.h>
105 #include <net/if_types.h>
106 #include <net/raw_cb.h>
107 #include <net/route.h>
108
109 #include <netinet/in.h>
110 #include <netinet/in_var.h>
111 #include <netinet/icmp6.h>
112
113 #include <netinet/ip6.h>
114 #include <netinet6/ip6_var.h>
115 #include <netinet6/scope6_var.h>
116 #include <netinet6/nd6.h>
117 #include <netinet6/ip6_mroute.h>
118 #include <netinet6/pim6.h>
119 #include <netinet6/pim6_var.h>
120
121 #include <net/net_osdep.h>
122
123 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
124
125 #define M_HASCL(m) ((m)->m_flags & M_EXT)
126
127 static int ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
128 static void phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
129
130 static int set_pim6(int *);
131 static int socket_send __P((struct socket *, struct mbuf *,
132 struct sockaddr_in6 *));
133 static int register_send __P((struct ip6_hdr *, struct mif6 *,
134 struct mbuf *));
135
136 /*
137 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static,
138 * except for netstat or debugging purposes.
139 */
140 struct socket *ip6_mrouter = NULL;
141 int ip6_mrouter_ver = 0;
142 int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */
143 struct mrt6stat mrt6stat;
144
145 #define NO_RTE_FOUND 0x1
146 #define RTE_FOUND 0x2
147
148 struct mf6c *mf6ctable[MF6CTBLSIZ];
149 u_char n6expire[MF6CTBLSIZ];
150 static struct mif6 mif6table[MAXMIFS];
151 #ifdef MRT6DEBUG
152 u_int mrt6debug = 0; /* debug level */
153 #define DEBUG_MFC 0x02
154 #define DEBUG_FORWARD 0x04
155 #define DEBUG_EXPIRE 0x08
156 #define DEBUG_XMIT 0x10
157 #define DEBUG_REG 0x20
158 #define DEBUG_PIM 0x40
159 #endif
160
161 static void expire_upcalls(void *);
162 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */
163 #define UPCALL_EXPIRE 6 /* number of timeouts */
164
165 #ifdef INET
166 #ifdef MROUTING
167 extern struct socket *ip_mrouter;
168 #endif
169 #endif
170
171 /*
172 * 'Interfaces' associated with decapsulator (so we can tell
173 * packets that went through it from ones that get reflected
174 * by a broken gateway). Different from IPv4 register_if,
175 * these interfaces are linked into the system ifnet list,
176 * because per-interface IPv6 statistics are maintained in
177 * ifp->if_afdata. But it does not have any routes point
178 * to them. I.e., packets can't be sent this way. They
179 * only exist as a placeholder for multicast source
180 * verification.
181 */
182 static struct ifnet *multicast_register_if6;
183
184 #define ENCAP_HOPS 64
185
186 /*
187 * Private variables.
188 */
189 static mifi_t nummifs = 0;
190 static mifi_t reg_mif_num = (mifi_t)-1;
191
192 static struct pim6stat pim6stat;
193 static int pim6;
194
195 /*
196 * Hash function for a source, group entry
197 */
198 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
199 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
200 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
201 (g).s6_addr32[2] ^ (g).s6_addr32[3])
202
203 /*
204 * Find a route for a given origin IPv6 address and Multicast group address.
205 * Quality of service parameter to be added in the future!!!
206 */
207
208 #define MF6CFIND(o, g, rt) do { \
209 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
210 rt = NULL; \
211 mrt6stat.mrt6s_mfc_lookups++; \
212 while (_rt) { \
213 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
214 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
215 (_rt->mf6c_stall == NULL)) { \
216 rt = _rt; \
217 break; \
218 } \
219 _rt = _rt->mf6c_next; \
220 } \
221 if (rt == NULL) { \
222 mrt6stat.mrt6s_mfc_misses++; \
223 } \
224 } while (/*CONSTCOND*/ 0)
225
226 /*
227 * Macros to compute elapsed time efficiently
228 * Borrowed from Van Jacobson's scheduling code
229 */
230 #define TV_DELTA(a, b, delta) do { \
231 int xxs; \
232 \
233 delta = (a).tv_usec - (b).tv_usec; \
234 if ((xxs = (a).tv_sec - (b).tv_sec)) { \
235 switch (xxs) { \
236 case 2: \
237 delta += 1000000; \
238 /* FALLTHROUGH */ \
239 case 1: \
240 delta += 1000000; \
241 break; \
242 default: \
243 delta += (1000000 * xxs); \
244 } \
245 } \
246 } while (/*CONSTCOND*/ 0)
247
248 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
249 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
250
251 #ifdef UPCALL_TIMING
252 #define UPCALL_MAX 50
253 u_long upcall_data[UPCALL_MAX + 1];
254 static void collate();
255 #endif /* UPCALL_TIMING */
256
257 static int get_sg_cnt(struct sioc_sg_req6 *);
258 static int get_mif6_cnt(struct sioc_mif_req6 *);
259 static int ip6_mrouter_init(struct socket *, int, int);
260 static int add_m6if(struct mif6ctl *);
261 static int del_m6if(mifi_t *);
262 static int add_m6fc(struct mf6cctl *);
263 static int del_m6fc(struct mf6cctl *);
264
265 static struct callout expire_upcalls_ch;
266
267 /*
268 * Handle MRT setsockopt commands to modify the multicast routing tables.
269 */
270 int
271 ip6_mrouter_set(so, sopt)
272 struct socket *so;
273 struct sockopt *sopt;
274 {
275 int error = 0;
276 int optval;
277 struct mif6ctl mifc;
278 struct mf6cctl mfcc;
279 mifi_t mifi;
280
281 if (so != ip6_mrouter && sopt->sopt_name != MRT6_INIT)
282 return (EACCES);
283
284 switch (sopt->sopt_name) {
285 case MRT6_INIT:
286 #ifdef MRT6_OINIT
287 case MRT6_OINIT:
288 #endif
289 error = sooptcopyin(sopt, &optval, sizeof(optval),
290 sizeof(optval));
291 if (error)
292 break;
293 error = ip6_mrouter_init(so, optval, sopt->sopt_name);
294 break;
295 case MRT6_DONE:
296 error = ip6_mrouter_done();
297 break;
298 case MRT6_ADD_MIF:
299 error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
300 if (error)
301 break;
302 error = add_m6if(&mifc);
303 break;
304 case MRT6_ADD_MFC:
305 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
306 if (error)
307 break;
308 error = add_m6fc(&mfcc);
309 break;
310 case MRT6_DEL_MFC:
311 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
312 if (error)
313 break;
314 error = del_m6fc(&mfcc);
315 break;
316 case MRT6_DEL_MIF:
317 error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
318 if (error)
319 break;
320 error = del_m6if(&mifi);
321 break;
322 case MRT6_PIM:
323 error = sooptcopyin(sopt, &optval, sizeof(optval),
324 sizeof(optval));
325 if (error)
326 break;
327 error = set_pim6(&optval);
328 break;
329 default:
330 error = EOPNOTSUPP;
331 break;
332 }
333
334 return (error);
335 }
336
337 /*
338 * Handle MRT getsockopt commands
339 */
340 int
341 ip6_mrouter_get(so, sopt)
342 struct socket *so;
343 struct sockopt *sopt;
344 {
345 int error = 0;
346
347 if (so != ip6_mrouter)
348 return (EACCES);
349
350 switch (sopt->sopt_name) {
351 case MRT6_PIM:
352 error = sooptcopyout(sopt, &pim6, sizeof(pim6));
353 break;
354 }
355 return (error);
356 }
357
358 /*
359 * Handle ioctl commands to obtain information from the cache
360 */
361 int
362 mrt6_ioctl(cmd, data)
363 int cmd;
364 caddr_t data;
365 {
366 switch (cmd) {
367 case SIOCGETSGCNT_IN6:
368 return (get_sg_cnt((struct sioc_sg_req6 *)data));
369 case SIOCGETMIFCNT_IN6:
370 return (get_mif6_cnt((struct sioc_mif_req6 *)data));
371 default:
372 return (EINVAL);
373 }
374 }
375
376 /*
377 * returns the packet, byte, rpf-failure count for the source group provided
378 */
379 static int
380 get_sg_cnt(req)
381 struct sioc_sg_req6 *req;
382 {
383 struct mf6c *rt;
384 int s;
385
386 s = splnet();
387 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
388 splx(s);
389 if (rt != NULL) {
390 req->pktcnt = rt->mf6c_pkt_cnt;
391 req->bytecnt = rt->mf6c_byte_cnt;
392 req->wrong_if = rt->mf6c_wrong_if;
393 } else
394 return (ESRCH);
395 #if 0
396 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff;
397 #endif
398
399 return (0);
400 }
401
402 /*
403 * returns the input and output packet and byte counts on the mif provided
404 */
405 static int
406 get_mif6_cnt(req)
407 struct sioc_mif_req6 *req;
408 {
409 mifi_t mifi = req->mifi;
410
411 if (mifi >= nummifs)
412 return (EINVAL);
413
414 req->icount = mif6table[mifi].m6_pkt_in;
415 req->ocount = mif6table[mifi].m6_pkt_out;
416 req->ibytes = mif6table[mifi].m6_bytes_in;
417 req->obytes = mif6table[mifi].m6_bytes_out;
418
419 return (0);
420 }
421
422 static int
423 set_pim6(i)
424 int *i;
425 {
426 if ((*i != 1) && (*i != 0))
427 return (EINVAL);
428
429 pim6 = *i;
430
431 return (0);
432 }
433
434 /*
435 * Enable multicast routing
436 */
437 static int
438 ip6_mrouter_init(so, v, cmd)
439 struct socket *so;
440 int v;
441 int cmd;
442 {
443 #ifdef MRT6DEBUG
444 if (mrt6debug)
445 log(LOG_DEBUG,
446 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
447 so->so_type, so->so_proto->pr_protocol);
448 #endif
449
450 if (so->so_type != SOCK_RAW ||
451 so->so_proto->pr_protocol != IPPROTO_ICMPV6)
452 return (EOPNOTSUPP);
453
454 if (v != 1)
455 return (ENOPROTOOPT);
456
457 if (ip6_mrouter != NULL)
458 return (EADDRINUSE);
459
460 ip6_mrouter = so;
461 ip6_mrouter_ver = cmd;
462
463 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
464 bzero((caddr_t)n6expire, sizeof(n6expire));
465
466 pim6 = 0;/* used for stubbing out/in pim stuff */
467
468 callout_init(&expire_upcalls_ch, 0);
469 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
470 expire_upcalls, NULL);
471
472 #ifdef MRT6DEBUG
473 if (mrt6debug)
474 log(LOG_DEBUG, "ip6_mrouter_init\n");
475 #endif
476
477 return (0);
478 }
479
480 /*
481 * Disable multicast routing
482 */
483 int
484 ip6_mrouter_done()
485 {
486 mifi_t mifi;
487 int i;
488 struct mf6c *rt;
489 struct rtdetq *rte;
490 int s;
491
492 s = splnet();
493
494 /*
495 * For each phyint in use, disable promiscuous reception of all IPv6
496 * multicasts.
497 */
498 #ifdef INET
499 #ifdef MROUTING
500 /*
501 * If there is still IPv4 multicast routing daemon,
502 * we remain interfaces to receive all muliticasted packets.
503 * XXX: there may be an interface in which the IPv4 multicast
504 * daemon is not interested...
505 */
506 if (!ip_mrouter)
507 #endif
508 #endif
509 {
510 for (mifi = 0; mifi < nummifs; mifi++) {
511 if (mif6table[mifi].m6_ifp &&
512 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
513 if_allmulti(mif6table[mifi].m6_ifp, 0);
514 }
515 }
516 }
517 #ifdef notyet
518 bzero((caddr_t)qtable, sizeof(qtable));
519 bzero((caddr_t)tbftable, sizeof(tbftable));
520 #endif
521 bzero((caddr_t)mif6table, sizeof(mif6table));
522 nummifs = 0;
523
524 pim6 = 0; /* used to stub out/in pim specific code */
525
526 callout_stop(&expire_upcalls_ch);
527
528 /*
529 * Free all multicast forwarding cache entries.
530 */
531 for (i = 0; i < MF6CTBLSIZ; i++) {
532 rt = mf6ctable[i];
533 while (rt) {
534 struct mf6c *frt;
535
536 for (rte = rt->mf6c_stall; rte != NULL; ) {
537 struct rtdetq *n = rte->next;
538
539 m_free(rte->m);
540 free(rte, M_MRTABLE6);
541 rte = n;
542 }
543 frt = rt;
544 rt = rt->mf6c_next;
545 free(frt, M_MRTABLE6);
546 }
547 }
548
549 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
550
551 /*
552 * Reset register interface
553 */
554 if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
555 if_detach(multicast_register_if6);
556 if_free(multicast_register_if6);
557 reg_mif_num = (mifi_t)-1;
558 multicast_register_if6 = NULL;
559 }
560
561 ip6_mrouter = NULL;
562 ip6_mrouter_ver = 0;
563
564 splx(s);
565
566 #ifdef MRT6DEBUG
567 if (mrt6debug)
568 log(LOG_DEBUG, "ip6_mrouter_done\n");
569 #endif
570
571 return (0);
572 }
573
574 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
575
576 /*
577 * Add a mif to the mif table
578 */
579 static int
580 add_m6if(mifcp)
581 struct mif6ctl *mifcp;
582 {
583 struct mif6 *mifp;
584 struct ifnet *ifp;
585 int error, s;
586 #ifdef notyet
587 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi;
588 #endif
589
590 if (mifcp->mif6c_mifi >= MAXMIFS)
591 return (EINVAL);
592 mifp = mif6table + mifcp->mif6c_mifi;
593 if (mifp->m6_ifp)
594 return (EADDRINUSE); /* XXX: is it appropriate? */
595 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > if_index)
596 return (ENXIO);
597 ifp = ifnet_byindex(mifcp->mif6c_pifi);
598
599 if (mifcp->mif6c_flags & MIFF_REGISTER) {
600 if (reg_mif_num == (mifi_t)-1) {
601 ifp = if_alloc(IFT_OTHER);
602
603 if_initname(ifp, "register_mif", 0);
604 ifp->if_flags |= IFF_LOOPBACK;
605 if_attach(ifp);
606 multicast_register_if6 = ifp;
607 reg_mif_num = mifcp->mif6c_mifi;
608 /*
609 * it is impossible to guess the ifindex of the
610 * register interface. So mif6c_pifi is automatically
611 * calculated.
612 */
613 mifcp->mif6c_pifi = ifp->if_index;
614 } else {
615 ifp = multicast_register_if6;
616 }
617
618 } /* if REGISTER */
619 else {
620 /* Make sure the interface supports multicast */
621 if ((ifp->if_flags & IFF_MULTICAST) == 0)
622 return (EOPNOTSUPP);
623
624 s = splnet();
625 error = if_allmulti(ifp, 1);
626 splx(s);
627 if (error)
628 return (error);
629 }
630
631 s = splnet();
632 mifp->m6_flags = mifcp->mif6c_flags;
633 mifp->m6_ifp = ifp;
634 #ifdef notyet
635 /* scaling up here allows division by 1024 in critical code */
636 mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000;
637 #endif
638 /* initialize per mif pkt counters */
639 mifp->m6_pkt_in = 0;
640 mifp->m6_pkt_out = 0;
641 mifp->m6_bytes_in = 0;
642 mifp->m6_bytes_out = 0;
643 splx(s);
644
645 /* Adjust nummifs up if the mifi is higher than nummifs */
646 if (nummifs <= mifcp->mif6c_mifi)
647 nummifs = mifcp->mif6c_mifi + 1;
648
649 #ifdef MRT6DEBUG
650 if (mrt6debug)
651 log(LOG_DEBUG,
652 "add_mif #%d, phyint %s\n",
653 mifcp->mif6c_mifi,
654 ifp->if_xname);
655 #endif
656
657 return (0);
658 }
659
660 /*
661 * Delete a mif from the mif table
662 */
663 static int
664 del_m6if(mifip)
665 mifi_t *mifip;
666 {
667 struct mif6 *mifp = mif6table + *mifip;
668 mifi_t mifi;
669 struct ifnet *ifp;
670 int s;
671
672 if (*mifip >= nummifs)
673 return (EINVAL);
674 if (mifp->m6_ifp == NULL)
675 return (EINVAL);
676
677 s = splnet();
678
679 if (!(mifp->m6_flags & MIFF_REGISTER)) {
680 /*
681 * XXX: what if there is yet IPv4 multicast daemon
682 * using the interface?
683 */
684 ifp = mifp->m6_ifp;
685
686 if_allmulti(ifp, 0);
687 } else {
688 if (reg_mif_num != (mifi_t)-1 &&
689 multicast_register_if6 != NULL) {
690 if_detach(multicast_register_if6);
691 if_free(multicast_register_if6);
692 reg_mif_num = (mifi_t)-1;
693 multicast_register_if6 = NULL;
694 }
695 }
696
697 #ifdef notyet
698 bzero((caddr_t)qtable[*mifip], sizeof(qtable[*mifip]));
699 bzero((caddr_t)mifp->m6_tbf, sizeof(*(mifp->m6_tbf)));
700 #endif
701 bzero((caddr_t)mifp, sizeof(*mifp));
702
703 /* Adjust nummifs down */
704 for (mifi = nummifs; mifi > 0; mifi--)
705 if (mif6table[mifi - 1].m6_ifp)
706 break;
707 nummifs = mifi;
708
709 splx(s);
710
711 #ifdef MRT6DEBUG
712 if (mrt6debug)
713 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
714 #endif
715
716 return (0);
717 }
718
719 /*
720 * Add an mfc entry
721 */
722 static int
723 add_m6fc(mfccp)
724 struct mf6cctl *mfccp;
725 {
726 struct mf6c *rt;
727 u_long hash;
728 struct rtdetq *rte;
729 u_short nstl;
730 int s;
731
732 MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
733 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
734
735 /* If an entry already exists, just update the fields */
736 if (rt) {
737 #ifdef MRT6DEBUG
738 if (mrt6debug & DEBUG_MFC)
739 log(LOG_DEBUG,
740 "add_m6fc no upcall h %d o %s g %s p %x\n",
741 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
742 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
743 mfccp->mf6cc_parent);
744 #endif
745
746 s = splnet();
747 rt->mf6c_parent = mfccp->mf6cc_parent;
748 rt->mf6c_ifset = mfccp->mf6cc_ifset;
749 splx(s);
750 return (0);
751 }
752
753 /*
754 * Find the entry for which the upcall was made and update
755 */
756 s = splnet();
757 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
758 mfccp->mf6cc_mcastgrp.sin6_addr);
759 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
760 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
761 &mfccp->mf6cc_origin.sin6_addr) &&
762 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
763 &mfccp->mf6cc_mcastgrp.sin6_addr) &&
764 (rt->mf6c_stall != NULL)) {
765
766 if (nstl++)
767 log(LOG_ERR,
768 "add_m6fc: %s o %s g %s p %x dbx %p\n",
769 "multiple kernel entries",
770 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
771 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
772 mfccp->mf6cc_parent, rt->mf6c_stall);
773
774 #ifdef MRT6DEBUG
775 if (mrt6debug & DEBUG_MFC)
776 log(LOG_DEBUG,
777 "add_m6fc o %s g %s p %x dbg %x\n",
778 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
779 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
780 mfccp->mf6cc_parent, rt->mf6c_stall);
781 #endif
782
783 rt->mf6c_origin = mfccp->mf6cc_origin;
784 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
785 rt->mf6c_parent = mfccp->mf6cc_parent;
786 rt->mf6c_ifset = mfccp->mf6cc_ifset;
787 /* initialize pkt counters per src-grp */
788 rt->mf6c_pkt_cnt = 0;
789 rt->mf6c_byte_cnt = 0;
790 rt->mf6c_wrong_if = 0;
791
792 rt->mf6c_expire = 0; /* Don't clean this guy up */
793 n6expire[hash]--;
794
795 /* free packets Qed at the end of this entry */
796 for (rte = rt->mf6c_stall; rte != NULL; ) {
797 struct rtdetq *n = rte->next;
798 ip6_mdq(rte->m, rte->ifp, rt);
799 m_freem(rte->m);
800 #ifdef UPCALL_TIMING
801 collate(&(rte->t));
802 #endif /* UPCALL_TIMING */
803 free(rte, M_MRTABLE6);
804 rte = n;
805 }
806 rt->mf6c_stall = NULL;
807 }
808 }
809
810 /*
811 * It is possible that an entry is being inserted without an upcall
812 */
813 if (nstl == 0) {
814 #ifdef MRT6DEBUG
815 if (mrt6debug & DEBUG_MFC)
816 log(LOG_DEBUG,
817 "add_mfc no upcall h %d o %s g %s p %x\n",
818 hash,
819 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
820 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
821 mfccp->mf6cc_parent);
822 #endif
823
824 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
825
826 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
827 &mfccp->mf6cc_origin.sin6_addr)&&
828 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
829 &mfccp->mf6cc_mcastgrp.sin6_addr)) {
830
831 rt->mf6c_origin = mfccp->mf6cc_origin;
832 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
833 rt->mf6c_parent = mfccp->mf6cc_parent;
834 rt->mf6c_ifset = mfccp->mf6cc_ifset;
835 /* initialize pkt counters per src-grp */
836 rt->mf6c_pkt_cnt = 0;
837 rt->mf6c_byte_cnt = 0;
838 rt->mf6c_wrong_if = 0;
839
840 if (rt->mf6c_expire)
841 n6expire[hash]--;
842 rt->mf6c_expire = 0;
843 }
844 }
845 if (rt == NULL) {
846 /* no upcall, so make a new entry */
847 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
848 M_NOWAIT);
849 if (rt == NULL) {
850 splx(s);
851 return (ENOBUFS);
852 }
853
854 /* insert new entry at head of hash chain */
855 rt->mf6c_origin = mfccp->mf6cc_origin;
856 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
857 rt->mf6c_parent = mfccp->mf6cc_parent;
858 rt->mf6c_ifset = mfccp->mf6cc_ifset;
859 /* initialize pkt counters per src-grp */
860 rt->mf6c_pkt_cnt = 0;
861 rt->mf6c_byte_cnt = 0;
862 rt->mf6c_wrong_if = 0;
863 rt->mf6c_expire = 0;
864 rt->mf6c_stall = NULL;
865
866 /* link into table */
867 rt->mf6c_next = mf6ctable[hash];
868 mf6ctable[hash] = rt;
869 }
870 }
871 splx(s);
872 return (0);
873 }
874
875 #ifdef UPCALL_TIMING
876 /*
877 * collect delay statistics on the upcalls
878 */
879 static void
880 collate(t)
881 struct timeval *t;
882 {
883 u_long d;
884 struct timeval tp;
885 u_long delta;
886
887 GET_TIME(tp);
888
889 if (TV_LT(*t, tp))
890 {
891 TV_DELTA(tp, *t, delta);
892
893 d = delta >> 10;
894 if (d > UPCALL_MAX)
895 d = UPCALL_MAX;
896
897 ++upcall_data[d];
898 }
899 }
900 #endif /* UPCALL_TIMING */
901
902 /*
903 * Delete an mfc entry
904 */
905 static int
906 del_m6fc(mfccp)
907 struct mf6cctl *mfccp;
908 {
909 struct sockaddr_in6 origin;
910 struct sockaddr_in6 mcastgrp;
911 struct mf6c *rt;
912 struct mf6c **nptr;
913 u_long hash;
914 int s;
915
916 origin = mfccp->mf6cc_origin;
917 mcastgrp = mfccp->mf6cc_mcastgrp;
918 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
919
920 #ifdef MRT6DEBUG
921 if (mrt6debug & DEBUG_MFC)
922 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
923 ip6_sprintf(&origin.sin6_addr),
924 ip6_sprintf(&mcastgrp.sin6_addr));
925 #endif
926
927 s = splnet();
928
929 nptr = &mf6ctable[hash];
930 while ((rt = *nptr) != NULL) {
931 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
932 &rt->mf6c_origin.sin6_addr) &&
933 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
934 &rt->mf6c_mcastgrp.sin6_addr) &&
935 rt->mf6c_stall == NULL)
936 break;
937
938 nptr = &rt->mf6c_next;
939 }
940 if (rt == NULL) {
941 splx(s);
942 return (EADDRNOTAVAIL);
943 }
944
945 *nptr = rt->mf6c_next;
946 free(rt, M_MRTABLE6);
947
948 splx(s);
949
950 return (0);
951 }
952
953 static int
954 socket_send(s, mm, src)
955 struct socket *s;
956 struct mbuf *mm;
957 struct sockaddr_in6 *src;
958 {
959 if (s) {
960 if (sbappendaddr(&s->so_rcv,
961 (struct sockaddr *)src,
962 mm, (struct mbuf *)0) != 0) {
963 sorwakeup(s);
964 return (0);
965 }
966 }
967 m_freem(mm);
968 return (-1);
969 }
970
971 /*
972 * IPv6 multicast forwarding function. This function assumes that the packet
973 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
974 * pointed to by "ifp", and the packet is to be relayed to other networks
975 * that have members of the packet's destination IPv6 multicast group.
976 *
977 * The packet is returned unscathed to the caller, unless it is
978 * erroneous, in which case a non-zero return value tells the caller to
979 * discard it.
980 *
981 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
982 * this function is called in the originating context (i.e., not when
983 * forwarding a packet from other node). ip6_output(), which is currently the
984 * only function that calls this function is called in the originating context,
985 * explicitly ensures this condition. It is caller's responsibility to ensure
986 * that if this function is called from somewhere else in the originating
987 * context in the future.
988 */
989
990 int
991 ip6_mforward(ip6, ifp, m)
992 struct ip6_hdr *ip6;
993 struct ifnet *ifp;
994 struct mbuf *m;
995 {
996 struct mf6c *rt;
997 struct mif6 *mifp;
998 struct mbuf *mm;
999 int s;
1000 mifi_t mifi;
1001
1002 #ifdef MRT6DEBUG
1003 if (mrt6debug & DEBUG_FORWARD)
1004 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1005 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst),
1006 ifp->if_index);
1007 #endif
1008
1009 /*
1010 * Don't forward a packet with Hop limit of zero or one,
1011 * or a packet destined to a local-only group.
1012 */
1013 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1014 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1015 return (0);
1016 ip6->ip6_hlim--;
1017
1018 /*
1019 * Source address check: do not forward packets with unspecified
1020 * source. It was discussed in July 2000, on ipngwg mailing list.
1021 * This is rather more serious than unicast cases, because some
1022 * MLD packets can be sent with the unspecified source address
1023 * (although such packets must normally set 1 to the hop limit field).
1024 */
1025 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1026 ip6stat.ip6s_cantforward++;
1027 if (ip6_log_time + ip6_log_interval < time_second) {
1028 ip6_log_time = time_second;
1029 log(LOG_DEBUG,
1030 "cannot forward "
1031 "from %s to %s nxt %d received on %s\n",
1032 ip6_sprintf(&ip6->ip6_src),
1033 ip6_sprintf(&ip6->ip6_dst),
1034 ip6->ip6_nxt,
1035 if_name(m->m_pkthdr.rcvif));
1036 }
1037 return (0);
1038 }
1039
1040 /*
1041 * Determine forwarding mifs from the forwarding cache table
1042 */
1043 s = splnet();
1044 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1045
1046 /* Entry exists, so forward if necessary */
1047 if (rt) {
1048 splx(s);
1049 return (ip6_mdq(m, ifp, rt));
1050 } else {
1051 /*
1052 * If we don't have a route for packet's origin,
1053 * Make a copy of the packet &
1054 * send message to routing daemon
1055 */
1056
1057 struct mbuf *mb0;
1058 struct rtdetq *rte;
1059 u_long hash;
1060 /* int i, npkts;*/
1061 #ifdef UPCALL_TIMING
1062 struct timeval tp;
1063
1064 GET_TIME(tp);
1065 #endif /* UPCALL_TIMING */
1066
1067 mrt6stat.mrt6s_no_route++;
1068 #ifdef MRT6DEBUG
1069 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1070 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1071 ip6_sprintf(&ip6->ip6_src),
1072 ip6_sprintf(&ip6->ip6_dst));
1073 #endif
1074
1075 /*
1076 * Allocate mbufs early so that we don't do extra work if we
1077 * are just going to fail anyway.
1078 */
1079 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6,
1080 M_NOWAIT);
1081 if (rte == NULL) {
1082 splx(s);
1083 return (ENOBUFS);
1084 }
1085 mb0 = m_copy(m, 0, M_COPYALL);
1086 /*
1087 * Pullup packet header if needed before storing it,
1088 * as other references may modify it in the meantime.
1089 */
1090 if (mb0 &&
1091 (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1092 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1093 if (mb0 == NULL) {
1094 free(rte, M_MRTABLE6);
1095 splx(s);
1096 return (ENOBUFS);
1097 }
1098
1099 /* is there an upcall waiting for this packet? */
1100 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1101 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1102 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1103 &rt->mf6c_origin.sin6_addr) &&
1104 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1105 &rt->mf6c_mcastgrp.sin6_addr) &&
1106 (rt->mf6c_stall != NULL))
1107 break;
1108 }
1109
1110 if (rt == NULL) {
1111 struct mrt6msg *im;
1112 #ifdef MRT6_OINIT
1113 struct omrt6msg *oim;
1114 #endif
1115
1116 /* no upcall, so make a new entry */
1117 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
1118 M_NOWAIT);
1119 if (rt == NULL) {
1120 free(rte, M_MRTABLE6);
1121 m_freem(mb0);
1122 splx(s);
1123 return (ENOBUFS);
1124 }
1125 /*
1126 * Make a copy of the header to send to the user
1127 * level process
1128 */
1129 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1130
1131 if (mm == NULL) {
1132 free(rte, M_MRTABLE6);
1133 m_freem(mb0);
1134 free(rt, M_MRTABLE6);
1135 splx(s);
1136 return (ENOBUFS);
1137 }
1138
1139 /*
1140 * Send message to routing daemon
1141 */
1142 sin6.sin6_addr = ip6->ip6_src;
1143
1144 im = NULL;
1145 #ifdef MRT6_OINIT
1146 oim = NULL;
1147 #endif
1148 switch (ip6_mrouter_ver) {
1149 #ifdef MRT6_OINIT
1150 case MRT6_OINIT:
1151 oim = mtod(mm, struct omrt6msg *);
1152 oim->im6_msgtype = MRT6MSG_NOCACHE;
1153 oim->im6_mbz = 0;
1154 break;
1155 #endif
1156 case MRT6_INIT:
1157 im = mtod(mm, struct mrt6msg *);
1158 im->im6_msgtype = MRT6MSG_NOCACHE;
1159 im->im6_mbz = 0;
1160 break;
1161 default:
1162 free(rte, M_MRTABLE6);
1163 m_freem(mb0);
1164 free(rt, M_MRTABLE6);
1165 splx(s);
1166 return (EINVAL);
1167 }
1168
1169 #ifdef MRT6DEBUG
1170 if (mrt6debug & DEBUG_FORWARD)
1171 log(LOG_DEBUG,
1172 "getting the iif info in the kernel\n");
1173 #endif
1174
1175 for (mifp = mif6table, mifi = 0;
1176 mifi < nummifs && mifp->m6_ifp != ifp;
1177 mifp++, mifi++)
1178 ;
1179
1180 switch (ip6_mrouter_ver) {
1181 #ifdef MRT6_OINIT
1182 case MRT6_OINIT:
1183 oim->im6_mif = mifi;
1184 break;
1185 #endif
1186 case MRT6_INIT:
1187 im->im6_mif = mifi;
1188 break;
1189 }
1190
1191 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1192 log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1193 "socket queue full\n");
1194 mrt6stat.mrt6s_upq_sockfull++;
1195 free(rte, M_MRTABLE6);
1196 m_freem(mb0);
1197 free(rt, M_MRTABLE6);
1198 splx(s);
1199 return (ENOBUFS);
1200 }
1201
1202 mrt6stat.mrt6s_upcalls++;
1203
1204 /* insert new entry at head of hash chain */
1205 bzero(rt, sizeof(*rt));
1206 rt->mf6c_origin.sin6_family = AF_INET6;
1207 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1208 rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1209 rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1210 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1211 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1212 rt->mf6c_expire = UPCALL_EXPIRE;
1213 n6expire[hash]++;
1214 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1215
1216 /* link into table */
1217 rt->mf6c_next = mf6ctable[hash];
1218 mf6ctable[hash] = rt;
1219 /* Add this entry to the end of the queue */
1220 rt->mf6c_stall = rte;
1221 } else {
1222 /* determine if q has overflowed */
1223 struct rtdetq **p;
1224 int npkts = 0;
1225
1226 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1227 if (++npkts > MAX_UPQ6) {
1228 mrt6stat.mrt6s_upq_ovflw++;
1229 free(rte, M_MRTABLE6);
1230 m_freem(mb0);
1231 splx(s);
1232 return (0);
1233 }
1234
1235 /* Add this entry to the end of the queue */
1236 *p = rte;
1237 }
1238
1239 rte->next = NULL;
1240 rte->m = mb0;
1241 rte->ifp = ifp;
1242 #ifdef UPCALL_TIMING
1243 rte->t = tp;
1244 #endif /* UPCALL_TIMING */
1245
1246 splx(s);
1247
1248 return (0);
1249 }
1250 }
1251
1252 /*
1253 * Clean up cache entries if upcalls are not serviced
1254 * Call from the Slow Timeout mechanism, every half second.
1255 */
1256 static void
1257 expire_upcalls(unused)
1258 void *unused;
1259 {
1260 struct rtdetq *rte;
1261 struct mf6c *mfc, **nptr;
1262 int i;
1263 int s;
1264
1265 s = splnet();
1266 for (i = 0; i < MF6CTBLSIZ; i++) {
1267 if (n6expire[i] == 0)
1268 continue;
1269 nptr = &mf6ctable[i];
1270 while ((mfc = *nptr) != NULL) {
1271 rte = mfc->mf6c_stall;
1272 /*
1273 * Skip real cache entries
1274 * Make sure it wasn't marked to not expire (shouldn't happen)
1275 * If it expires now
1276 */
1277 if (rte != NULL &&
1278 mfc->mf6c_expire != 0 &&
1279 --mfc->mf6c_expire == 0) {
1280 #ifdef MRT6DEBUG
1281 if (mrt6debug & DEBUG_EXPIRE)
1282 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1283 ip6_sprintf(&mfc->mf6c_origin.sin6_addr),
1284 ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr));
1285 #endif
1286 /*
1287 * drop all the packets
1288 * free the mbuf with the pkt, if, timing info
1289 */
1290 do {
1291 struct rtdetq *n = rte->next;
1292 m_freem(rte->m);
1293 free(rte, M_MRTABLE6);
1294 rte = n;
1295 } while (rte != NULL);
1296 mrt6stat.mrt6s_cache_cleanups++;
1297 n6expire[i]--;
1298
1299 *nptr = mfc->mf6c_next;
1300 free(mfc, M_MRTABLE6);
1301 } else {
1302 nptr = &mfc->mf6c_next;
1303 }
1304 }
1305 }
1306 splx(s);
1307 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1308 expire_upcalls, NULL);
1309 }
1310
1311 /*
1312 * Packet forwarding routine once entry in the cache is made
1313 */
1314 static int
1315 ip6_mdq(m, ifp, rt)
1316 struct mbuf *m;
1317 struct ifnet *ifp;
1318 struct mf6c *rt;
1319 {
1320 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1321 mifi_t mifi, iif;
1322 struct mif6 *mifp;
1323 int plen = m->m_pkthdr.len;
1324 struct in6_addr src0, dst0; /* copies for local work */
1325 u_int32_t iszone, idzone, oszone, odzone;
1326 int error = 0;
1327
1328 /*
1329 * Macro to send packet on mif. Since RSVP packets don't get counted on
1330 * input, they shouldn't get counted on output, so statistics keeping is
1331 * separate.
1332 */
1333
1334 #define MC6_SEND(ip6, mifp, m) do { \
1335 if ((mifp)->m6_flags & MIFF_REGISTER) \
1336 register_send((ip6), (mifp), (m)); \
1337 else \
1338 phyint_send((ip6), (mifp), (m)); \
1339 } while (/*CONSTCOND*/ 0)
1340
1341 /*
1342 * Don't forward if it didn't arrive from the parent mif
1343 * for its origin.
1344 */
1345 mifi = rt->mf6c_parent;
1346 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1347 /* came in the wrong interface */
1348 #ifdef MRT6DEBUG
1349 if (mrt6debug & DEBUG_FORWARD)
1350 log(LOG_DEBUG,
1351 "wrong if: ifid %d mifi %d mififid %x\n",
1352 ifp->if_index, mifi,
1353 mif6table[mifi].m6_ifp->if_index);
1354 #endif
1355 mrt6stat.mrt6s_wrong_if++;
1356 rt->mf6c_wrong_if++;
1357 /*
1358 * If we are doing PIM processing, and we are forwarding
1359 * packets on this interface, send a message to the
1360 * routing daemon.
1361 */
1362 /* have to make sure this is a valid mif */
1363 if (mifi < nummifs && mif6table[mifi].m6_ifp)
1364 if (pim6 && (m->m_flags & M_LOOP) == 0) {
1365 /*
1366 * Check the M_LOOP flag to avoid an
1367 * unnecessary PIM assert.
1368 * XXX: M_LOOP is an ad-hoc hack...
1369 */
1370 static struct sockaddr_in6 sin6 =
1371 { sizeof(sin6), AF_INET6 };
1372
1373 struct mbuf *mm;
1374 struct mrt6msg *im;
1375 #ifdef MRT6_OINIT
1376 struct omrt6msg *oim;
1377 #endif
1378
1379 mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1380 if (mm &&
1381 (M_HASCL(mm) ||
1382 mm->m_len < sizeof(struct ip6_hdr)))
1383 mm = m_pullup(mm, sizeof(struct ip6_hdr));
1384 if (mm == NULL)
1385 return (ENOBUFS);
1386
1387 #ifdef MRT6_OINIT
1388 oim = NULL;
1389 #endif
1390 im = NULL;
1391 switch (ip6_mrouter_ver) {
1392 #ifdef MRT6_OINIT
1393 case MRT6_OINIT:
1394 oim = mtod(mm, struct omrt6msg *);
1395 oim->im6_msgtype = MRT6MSG_WRONGMIF;
1396 oim->im6_mbz = 0;
1397 break;
1398 #endif
1399 case MRT6_INIT:
1400 im = mtod(mm, struct mrt6msg *);
1401 im->im6_msgtype = MRT6MSG_WRONGMIF;
1402 im->im6_mbz = 0;
1403 break;
1404 default:
1405 m_freem(mm);
1406 return (EINVAL);
1407 }
1408
1409 for (mifp = mif6table, iif = 0;
1410 iif < nummifs && mifp &&
1411 mifp->m6_ifp != ifp;
1412 mifp++, iif++)
1413 ;
1414
1415 switch (ip6_mrouter_ver) {
1416 #ifdef MRT6_OINIT
1417 case MRT6_OINIT:
1418 oim->im6_mif = iif;
1419 sin6.sin6_addr = oim->im6_src;
1420 break;
1421 #endif
1422 case MRT6_INIT:
1423 im->im6_mif = iif;
1424 sin6.sin6_addr = im->im6_src;
1425 break;
1426 }
1427
1428 mrt6stat.mrt6s_upcalls++;
1429
1430 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1431 #ifdef MRT6DEBUG
1432 if (mrt6debug)
1433 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1434 #endif
1435 ++mrt6stat.mrt6s_upq_sockfull;
1436 return (ENOBUFS);
1437 } /* if socket Q full */
1438 } /* if PIM */
1439 return (0);
1440 } /* if wrong iif */
1441
1442 /* If I sourced this packet, it counts as output, else it was input. */
1443 if (m->m_pkthdr.rcvif == NULL) {
1444 /* XXX: is rcvif really NULL when output?? */
1445 mif6table[mifi].m6_pkt_out++;
1446 mif6table[mifi].m6_bytes_out += plen;
1447 } else {
1448 mif6table[mifi].m6_pkt_in++;
1449 mif6table[mifi].m6_bytes_in += plen;
1450 }
1451 rt->mf6c_pkt_cnt++;
1452 rt->mf6c_byte_cnt += plen;
1453
1454 /*
1455 * For each mif, forward a copy of the packet if there are group
1456 * members downstream on the interface.
1457 */
1458 src0 = ip6->ip6_src;
1459 dst0 = ip6->ip6_dst;
1460 if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1461 (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1462 ip6stat.ip6s_badscope++;
1463 return (error);
1464 }
1465 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1466 if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1467 /*
1468 * check if the outgoing packet is going to break
1469 * a scope boundary.
1470 * XXX For packets through PIM register tunnel
1471 * interface, we believe a routing daemon.
1472 */
1473 if (!(mif6table[rt->mf6c_parent].m6_flags &
1474 MIFF_REGISTER) &&
1475 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1476 if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1477 &oszone) ||
1478 in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1479 &odzone) ||
1480 iszone != oszone ||
1481 idzone != odzone) {
1482 ip6stat.ip6s_badscope++;
1483 continue;
1484 }
1485 }
1486
1487 mifp->m6_pkt_out++;
1488 mifp->m6_bytes_out += plen;
1489 MC6_SEND(ip6, mifp, m);
1490 }
1491 }
1492 return (0);
1493 }
1494
1495 static void
1496 phyint_send(ip6, mifp, m)
1497 struct ip6_hdr *ip6;
1498 struct mif6 *mifp;
1499 struct mbuf *m;
1500 {
1501 struct mbuf *mb_copy;
1502 struct ifnet *ifp = mifp->m6_ifp;
1503 int error = 0;
1504 int s = splnet(); /* needs to protect static "ro" below. */
1505 static struct route_in6 ro;
1506 struct in6_multi *in6m;
1507 struct sockaddr_in6 *dst6;
1508 u_long linkmtu;
1509
1510 /*
1511 * Make a new reference to the packet; make sure that
1512 * the IPv6 header is actually copied, not just referenced,
1513 * so that ip6_output() only scribbles on the copy.
1514 */
1515 mb_copy = m_copy(m, 0, M_COPYALL);
1516 if (mb_copy &&
1517 (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1518 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1519 if (mb_copy == NULL) {
1520 splx(s);
1521 return;
1522 }
1523 /* set MCAST flag to the outgoing packet */
1524 mb_copy->m_flags |= M_MCAST;
1525
1526 /*
1527 * If we sourced the packet, call ip6_output since we may devide
1528 * the packet into fragments when the packet is too big for the
1529 * outgoing interface.
1530 * Otherwise, we can simply send the packet to the interface
1531 * sending queue.
1532 */
1533 if (m->m_pkthdr.rcvif == NULL) {
1534 struct ip6_moptions im6o;
1535
1536 im6o.im6o_multicast_ifp = ifp;
1537 /* XXX: ip6_output will override ip6->ip6_hlim */
1538 im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1539 im6o.im6o_multicast_loop = 1;
1540 error = ip6_output(mb_copy, NULL, &ro,
1541 IPV6_FORWARDING, &im6o, NULL, NULL);
1542
1543 #ifdef MRT6DEBUG
1544 if (mrt6debug & DEBUG_XMIT)
1545 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1546 mifp - mif6table, error);
1547 #endif
1548 splx(s);
1549 return;
1550 }
1551
1552 /*
1553 * If we belong to the destination multicast group
1554 * on the outgoing interface, loop back a copy.
1555 */
1556 dst6 = (struct sockaddr_in6 *)&ro.ro_dst;
1557 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m);
1558 if (in6m != NULL) {
1559 dst6->sin6_len = sizeof(struct sockaddr_in6);
1560 dst6->sin6_family = AF_INET6;
1561 dst6->sin6_addr = ip6->ip6_dst;
1562 ip6_mloopback(ifp, m, (struct sockaddr_in6 *)&ro.ro_dst);
1563 }
1564 /*
1565 * Put the packet into the sending queue of the outgoing interface
1566 * if it would fit in the MTU of the interface.
1567 */
1568 linkmtu = IN6_LINKMTU(ifp);
1569 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1570 dst6->sin6_len = sizeof(struct sockaddr_in6);
1571 dst6->sin6_family = AF_INET6;
1572 dst6->sin6_addr = ip6->ip6_dst;
1573 /*
1574 * We just call if_output instead of nd6_output here, since
1575 * we need no ND for a multicast forwarded packet...right?
1576 */
1577 error = (*ifp->if_output)(ifp, mb_copy,
1578 (struct sockaddr *)&ro.ro_dst, NULL);
1579 #ifdef MRT6DEBUG
1580 if (mrt6debug & DEBUG_XMIT)
1581 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1582 mifp - mif6table, error);
1583 #endif
1584 } else {
1585 /*
1586 * pMTU discovery is intentionally disabled by default, since
1587 * various router may notify pMTU in multicast, which can be
1588 * a DDoS to a router
1589 */
1590 if (ip6_mcast_pmtu)
1591 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1592 else {
1593 #ifdef MRT6DEBUG
1594 if (mrt6debug & DEBUG_XMIT)
1595 log(LOG_DEBUG,
1596 "phyint_send: packet too big on %s o %s "
1597 "g %s size %d(discarded)\n",
1598 if_name(ifp),
1599 ip6_sprintf(&ip6->ip6_src),
1600 ip6_sprintf(&ip6->ip6_dst),
1601 mb_copy->m_pkthdr.len);
1602 #endif /* MRT6DEBUG */
1603 m_freem(mb_copy); /* simply discard the packet */
1604 }
1605 }
1606
1607 splx(s);
1608 }
1609
1610 static int
1611 register_send(ip6, mif, m)
1612 struct ip6_hdr *ip6;
1613 struct mif6 *mif;
1614 struct mbuf *m;
1615 {
1616 struct mbuf *mm;
1617 int i, len = m->m_pkthdr.len;
1618 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1619 struct mrt6msg *im6;
1620
1621 #ifdef MRT6DEBUG
1622 if (mrt6debug)
1623 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1624 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst));
1625 #endif
1626 ++pim6stat.pim6s_snd_registers;
1627
1628 /* Make a copy of the packet to send to the user level process */
1629 MGETHDR(mm, M_DONTWAIT, MT_HEADER);
1630 if (mm == NULL)
1631 return (ENOBUFS);
1632 mm->m_pkthdr.rcvif = NULL;
1633 mm->m_data += max_linkhdr;
1634 mm->m_len = sizeof(struct ip6_hdr);
1635
1636 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1637 m_freem(mm);
1638 return (ENOBUFS);
1639 }
1640 i = MHLEN - M_LEADINGSPACE(mm);
1641 if (i > len)
1642 i = len;
1643 mm = m_pullup(mm, i);
1644 if (mm == NULL)
1645 return (ENOBUFS);
1646 /* TODO: check it! */
1647 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1648
1649 /*
1650 * Send message to routing daemon
1651 */
1652 sin6.sin6_addr = ip6->ip6_src;
1653
1654 im6 = mtod(mm, struct mrt6msg *);
1655 im6->im6_msgtype = MRT6MSG_WHOLEPKT;
1656 im6->im6_mbz = 0;
1657
1658 im6->im6_mif = mif - mif6table;
1659
1660 /* iif info is not given for reg. encap.n */
1661 mrt6stat.mrt6s_upcalls++;
1662
1663 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1664 #ifdef MRT6DEBUG
1665 if (mrt6debug)
1666 log(LOG_WARNING,
1667 "register_send: ip6_mrouter socket queue full\n");
1668 #endif
1669 ++mrt6stat.mrt6s_upq_sockfull;
1670 return (ENOBUFS);
1671 }
1672 return (0);
1673 }
1674
1675 /*
1676 * PIM sparse mode hook
1677 * Receives the pim control messages, and passes them up to the listening
1678 * socket, using rip6_input.
1679 * The only message processed is the REGISTER pim message; the pim header
1680 * is stripped off, and the inner packet is passed to register_mforward.
1681 */
1682 int
1683 pim6_input(mp, offp, proto)
1684 struct mbuf **mp;
1685 int *offp, proto;
1686 {
1687 struct pim *pim; /* pointer to a pim struct */
1688 struct ip6_hdr *ip6;
1689 int pimlen;
1690 struct mbuf *m = *mp;
1691 int minlen;
1692 int off = *offp;
1693
1694 ++pim6stat.pim6s_rcv_total;
1695
1696 ip6 = mtod(m, struct ip6_hdr *);
1697 pimlen = m->m_pkthdr.len - *offp;
1698
1699 /*
1700 * Validate lengths
1701 */
1702 if (pimlen < PIM_MINLEN) {
1703 ++pim6stat.pim6s_rcv_tooshort;
1704 #ifdef MRT6DEBUG
1705 if (mrt6debug & DEBUG_PIM)
1706 log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1707 #endif
1708 m_freem(m);
1709 return (IPPROTO_DONE);
1710 }
1711
1712 /*
1713 * if the packet is at least as big as a REGISTER, go ahead
1714 * and grab the PIM REGISTER header size, to avoid another
1715 * possible m_pullup() later.
1716 *
1717 * PIM_MINLEN == pimhdr + u_int32 == 8
1718 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1719 */
1720 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1721
1722 /*
1723 * Make sure that the IP6 and PIM headers in contiguous memory, and
1724 * possibly the PIM REGISTER header
1725 */
1726 #ifndef PULLDOWN_TEST
1727 IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1728 /* adjust pointer */
1729 ip6 = mtod(m, struct ip6_hdr *);
1730
1731 /* adjust mbuf to point to the PIM header */
1732 pim = (struct pim *)((caddr_t)ip6 + off);
1733 #else
1734 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1735 if (pim == NULL) {
1736 pim6stat.pim6s_rcv_tooshort++;
1737 return (IPPROTO_DONE);
1738 }
1739 #endif
1740
1741 #define PIM6_CHECKSUM
1742 #ifdef PIM6_CHECKSUM
1743 {
1744 int cksumlen;
1745
1746 /*
1747 * Validate checksum.
1748 * If PIM REGISTER, exclude the data packet
1749 */
1750 if (pim->pim_type == PIM_REGISTER)
1751 cksumlen = PIM_MINLEN;
1752 else
1753 cksumlen = pimlen;
1754
1755 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1756 ++pim6stat.pim6s_rcv_badsum;
1757 #ifdef MRT6DEBUG
1758 if (mrt6debug & DEBUG_PIM)
1759 log(LOG_DEBUG,
1760 "pim6_input: invalid checksum\n");
1761 #endif
1762 m_freem(m);
1763 return (IPPROTO_DONE);
1764 }
1765 }
1766 #endif /* PIM_CHECKSUM */
1767
1768 /* PIM version check */
1769 if (pim->pim_ver != PIM_VERSION) {
1770 ++pim6stat.pim6s_rcv_badversion;
1771 #ifdef MRT6DEBUG
1772 log(LOG_ERR,
1773 "pim6_input: incorrect version %d, expecting %d\n",
1774 pim->pim_ver, PIM_VERSION);
1775 #endif
1776 m_freem(m);
1777 return (IPPROTO_DONE);
1778 }
1779
1780 if (pim->pim_type == PIM_REGISTER) {
1781 /*
1782 * since this is a REGISTER, we'll make a copy of the register
1783 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1784 * routing daemon.
1785 */
1786 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1787
1788 struct mbuf *mcp;
1789 struct ip6_hdr *eip6;
1790 u_int32_t *reghdr;
1791 int rc;
1792
1793 ++pim6stat.pim6s_rcv_registers;
1794
1795 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1796 #ifdef MRT6DEBUG
1797 if (mrt6debug & DEBUG_PIM)
1798 log(LOG_DEBUG,
1799 "pim6_input: register mif not set: %d\n",
1800 reg_mif_num);
1801 #endif
1802 m_freem(m);
1803 return (IPPROTO_DONE);
1804 }
1805
1806 reghdr = (u_int32_t *)(pim + 1);
1807
1808 if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1809 goto pim6_input_to_daemon;
1810
1811 /*
1812 * Validate length
1813 */
1814 if (pimlen < PIM6_REG_MINLEN) {
1815 ++pim6stat.pim6s_rcv_tooshort;
1816 ++pim6stat.pim6s_rcv_badregisters;
1817 #ifdef MRT6DEBUG
1818 log(LOG_ERR,
1819 "pim6_input: register packet size too "
1820 "small %d from %s\n",
1821 pimlen, ip6_sprintf(&ip6->ip6_src));
1822 #endif
1823 m_freem(m);
1824 return (IPPROTO_DONE);
1825 }
1826
1827 eip6 = (struct ip6_hdr *) (reghdr + 1);
1828 #ifdef MRT6DEBUG
1829 if (mrt6debug & DEBUG_PIM)
1830 log(LOG_DEBUG,
1831 "pim6_input[register], eip6: %s -> %s, "
1832 "eip6 plen %d\n",
1833 ip6_sprintf(&eip6->ip6_src),
1834 ip6_sprintf(&eip6->ip6_dst),
1835 ntohs(eip6->ip6_plen));
1836 #endif
1837
1838 /* verify the version number of the inner packet */
1839 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1840 ++pim6stat.pim6s_rcv_badregisters;
1841 #ifdef MRT6DEBUG
1842 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1843 "of the inner packet\n",
1844 (eip6->ip6_vfc & IPV6_VERSION));
1845 #endif
1846 m_freem(m);
1847 return (IPPROTO_NONE);
1848 }
1849
1850 /* verify the inner packet is destined to a mcast group */
1851 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1852 ++pim6stat.pim6s_rcv_badregisters;
1853 #ifdef MRT6DEBUG
1854 if (mrt6debug & DEBUG_PIM)
1855 log(LOG_DEBUG,
1856 "pim6_input: inner packet of register "
1857 "is not multicast %s\n",
1858 ip6_sprintf(&eip6->ip6_dst));
1859 #endif
1860 m_freem(m);
1861 return (IPPROTO_DONE);
1862 }
1863
1864 /*
1865 * make a copy of the whole header to pass to the daemon later.
1866 */
1867 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1868 if (mcp == NULL) {
1869 #ifdef MRT6DEBUG
1870 log(LOG_ERR,
1871 "pim6_input: pim register: "
1872 "could not copy register head\n");
1873 #endif
1874 m_freem(m);
1875 return (IPPROTO_DONE);
1876 }
1877
1878 /*
1879 * forward the inner ip6 packet; point m_data at the inner ip6.
1880 */
1881 m_adj(m, off + PIM_MINLEN);
1882 #ifdef MRT6DEBUG
1883 if (mrt6debug & DEBUG_PIM) {
1884 log(LOG_DEBUG,
1885 "pim6_input: forwarding decapsulated register: "
1886 "src %s, dst %s, mif %d\n",
1887 ip6_sprintf(&eip6->ip6_src),
1888 ip6_sprintf(&eip6->ip6_dst),
1889 reg_mif_num);
1890 }
1891 #endif
1892
1893 rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1894 dst.sin6_family, 0);
1895
1896 /* prepare the register head to send to the mrouting daemon */
1897 m = mcp;
1898 }
1899
1900 /*
1901 * Pass the PIM message up to the daemon; if it is a register message
1902 * pass the 'head' only up to the daemon. This includes the
1903 * encapsulator ip6 header, pim header, register header and the
1904 * encapsulated ip6 header.
1905 */
1906 pim6_input_to_daemon:
1907 rip6_input(&m, offp, proto);
1908 return (IPPROTO_DONE);
1909 }
Cache object: d38280caf92119dc6d85a855f5490d53
|