1 /* $FreeBSD: releng/6.2/sys/netinet6/ip6_mroute.c 157904 2006-04-20 15:43:45Z suz $ */
2 /* $KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun 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 * 4. Neither the name of the University nor the names of its contributors
52 * may be used to endorse or promote products derived from this software
53 * without specific prior written permission.
54 *
55 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
56 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
57 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
58 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
59 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
60 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
61 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
62 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
63 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
64 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
65 * SUCH DAMAGE.
66 *
67 * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93
68 */
69
70 /*
71 * IP multicast forwarding procedures
72 *
73 * Written by David Waitzman, BBN Labs, August 1988.
74 * Modified by Steve Deering, Stanford, February 1989.
75 * Modified by Mark J. Steiglitz, Stanford, May, 1991
76 * Modified by Van Jacobson, LBL, January 1993
77 * Modified by Ajit Thyagarajan, PARC, August 1993
78 * Modified by Bill Fenner, PARC, April 1994
79 *
80 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
81 */
82
83 #include "opt_inet.h"
84 #include "opt_inet6.h"
85
86 #include <sys/param.h>
87 #include <sys/callout.h>
88 #include <sys/errno.h>
89 #include <sys/kernel.h>
90 #include <sys/lock.h>
91 #include <sys/malloc.h>
92 #include <sys/mbuf.h>
93 #include <sys/protosw.h>
94 #include <sys/signalvar.h>
95 #include <sys/socket.h>
96 #include <sys/socketvar.h>
97 #include <sys/sockio.h>
98 #include <sys/sx.h>
99 #include <sys/syslog.h>
100 #include <sys/systm.h>
101 #include <sys/time.h>
102
103 #include <net/if.h>
104 #include <net/if_types.h>
105 #include <net/raw_cb.h>
106 #include <net/route.h>
107
108 #include <netinet/in.h>
109 #include <netinet/in_var.h>
110 #include <netinet/icmp6.h>
111
112 #include <netinet/ip6.h>
113 #include <netinet6/ip6_var.h>
114 #include <netinet6/scope6_var.h>
115 #include <netinet6/nd6.h>
116 #include <netinet6/ip6_mroute.h>
117 #include <netinet6/pim6.h>
118 #include <netinet6/pim6_var.h>
119
120 #include <net/net_osdep.h>
121
122 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
123
124 #define M_HASCL(m) ((m)->m_flags & M_EXT)
125
126 static int ip6_mdq __P((struct mbuf *, struct ifnet *, struct mf6c *));
127 static void phyint_send __P((struct ip6_hdr *, struct mif6 *, struct mbuf *));
128
129 static int set_pim6 __P((int *));
130 static int socket_send __P((struct socket *, struct mbuf *,
131 struct sockaddr_in6 *));
132 static int register_send __P((struct ip6_hdr *, struct mif6 *,
133 struct mbuf *));
134
135 /*
136 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static,
137 * except for netstat or debugging purposes.
138 */
139 struct socket *ip6_mrouter = NULL;
140 int ip6_mrouter_ver = 0;
141 int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */
142 struct mrt6stat mrt6stat;
143
144 #define NO_RTE_FOUND 0x1
145 #define RTE_FOUND 0x2
146
147 struct mf6c *mf6ctable[MF6CTBLSIZ];
148 u_char n6expire[MF6CTBLSIZ];
149 static struct mif6 mif6table[MAXMIFS];
150 #ifdef MRT6DEBUG
151 u_int mrt6debug = 0; /* debug level */
152 #define DEBUG_MFC 0x02
153 #define DEBUG_FORWARD 0x04
154 #define DEBUG_EXPIRE 0x08
155 #define DEBUG_XMIT 0x10
156 #define DEBUG_REG 0x20
157 #define DEBUG_PIM 0x40
158 #endif
159
160 static void expire_upcalls __P((void *));
161 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */
162 #define UPCALL_EXPIRE 6 /* number of timeouts */
163
164 #ifdef INET
165 #ifdef MROUTING
166 extern struct socket *ip_mrouter;
167 #endif
168 #endif
169
170 /*
171 * 'Interfaces' associated with decapsulator (so we can tell
172 * packets that went through it from ones that get reflected
173 * by a broken gateway). Different from IPv4 register_if,
174 * these interfaces are linked into the system ifnet list,
175 * because per-interface IPv6 statistics are maintained in
176 * ifp->if_afdata. But it does not have any routes point
177 * to them. I.e., packets can't be sent this way. They
178 * only exist as a placeholder for multicast source
179 * verification.
180 */
181 static struct ifnet *multicast_register_if6;
182
183 #define ENCAP_HOPS 64
184
185 /*
186 * Private variables.
187 */
188 static mifi_t nummifs = 0;
189 static mifi_t reg_mif_num = (mifi_t)-1;
190
191 static struct pim6stat pim6stat;
192 static int pim6;
193
194 /*
195 * Hash function for a source, group entry
196 */
197 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
198 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
199 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
200 (g).s6_addr32[2] ^ (g).s6_addr32[3])
201
202 /*
203 * Find a route for a given origin IPv6 address and Multicast group address.
204 * Quality of service parameter to be added in the future!!!
205 */
206
207 #define MF6CFIND(o, g, rt) do { \
208 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
209 rt = NULL; \
210 mrt6stat.mrt6s_mfc_lookups++; \
211 while (_rt) { \
212 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
213 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
214 (_rt->mf6c_stall == NULL)) { \
215 rt = _rt; \
216 break; \
217 } \
218 _rt = _rt->mf6c_next; \
219 } \
220 if (rt == NULL) { \
221 mrt6stat.mrt6s_mfc_misses++; \
222 } \
223 } while (/*CONSTCOND*/ 0)
224
225 /*
226 * Macros to compute elapsed time efficiently
227 * Borrowed from Van Jacobson's scheduling code
228 */
229 #define TV_DELTA(a, b, delta) do { \
230 int xxs; \
231 \
232 delta = (a).tv_usec - (b).tv_usec; \
233 if ((xxs = (a).tv_sec - (b).tv_sec)) { \
234 switch (xxs) { \
235 case 2: \
236 delta += 1000000; \
237 /* FALLTHROUGH */ \
238 case 1: \
239 delta += 1000000; \
240 break; \
241 default: \
242 delta += (1000000 * xxs); \
243 } \
244 } \
245 } while (/*CONSTCOND*/ 0)
246
247 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
248 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
249
250 #ifdef UPCALL_TIMING
251 #define UPCALL_MAX 50
252 u_long upcall_data[UPCALL_MAX + 1];
253 static void collate();
254 #endif /* UPCALL_TIMING */
255
256 static int get_sg_cnt __P((struct sioc_sg_req6 *));
257 static int get_mif6_cnt __P((struct sioc_mif_req6 *));
258 static int ip6_mrouter_init __P((struct socket *, int, int));
259 static int add_m6if __P((struct mif6ctl *));
260 static int del_m6if __P((mifi_t *));
261 static int add_m6fc __P((struct mf6cctl *));
262 static int del_m6fc __P((struct mf6cctl *));
263
264 static struct callout expire_upcalls_ch;
265
266 /*
267 * Handle MRT setsockopt commands to modify the multicast routing tables.
268 */
269 int
270 ip6_mrouter_set(so, sopt)
271 struct socket *so;
272 struct sockopt *sopt;
273 {
274 int error = 0;
275 int optval;
276 struct mif6ctl mifc;
277 struct mf6cctl mfcc;
278 mifi_t mifi;
279
280 if (so != ip6_mrouter && sopt->sopt_name != MRT6_INIT)
281 return (EACCES);
282
283 switch (sopt->sopt_name) {
284 case MRT6_INIT:
285 #ifdef MRT6_OINIT
286 case MRT6_OINIT:
287 #endif
288 error = sooptcopyin(sopt, &optval, sizeof(optval),
289 sizeof(optval));
290 if (error)
291 break;
292 error = ip6_mrouter_init(so, optval, sopt->sopt_name);
293 break;
294 case MRT6_DONE:
295 error = ip6_mrouter_done();
296 break;
297 case MRT6_ADD_MIF:
298 error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
299 if (error)
300 break;
301 error = add_m6if(&mifc);
302 break;
303 case MRT6_ADD_MFC:
304 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
305 if (error)
306 break;
307 error = add_m6fc(&mfcc);
308 break;
309 case MRT6_DEL_MFC:
310 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
311 if (error)
312 break;
313 error = del_m6fc(&mfcc);
314 break;
315 case MRT6_DEL_MIF:
316 error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
317 if (error)
318 break;
319 error = del_m6if(&mifi);
320 break;
321 case MRT6_PIM:
322 error = sooptcopyin(sopt, &optval, sizeof(optval),
323 sizeof(optval));
324 if (error)
325 break;
326 error = set_pim6(&optval);
327 break;
328 default:
329 error = EOPNOTSUPP;
330 break;
331 }
332
333 return (error);
334 }
335
336 /*
337 * Handle MRT getsockopt commands
338 */
339 int
340 ip6_mrouter_get(so, sopt)
341 struct socket *so;
342 struct sockopt *sopt;
343 {
344 int error = 0;
345
346 if (so != ip6_mrouter)
347 return (EACCES);
348
349 switch (sopt->sopt_name) {
350 case MRT6_PIM:
351 error = sooptcopyout(sopt, &pim6, sizeof(pim6));
352 break;
353 }
354 return (error);
355 }
356
357 /*
358 * Handle ioctl commands to obtain information from the cache
359 */
360 int
361 mrt6_ioctl(cmd, data)
362 int cmd;
363 caddr_t data;
364 {
365 switch (cmd) {
366 case SIOCGETSGCNT_IN6:
367 return (get_sg_cnt((struct sioc_sg_req6 *)data));
368 case SIOCGETMIFCNT_IN6:
369 return (get_mif6_cnt((struct sioc_mif_req6 *)data));
370 default:
371 return (EINVAL);
372 }
373 }
374
375 /*
376 * returns the packet, byte, rpf-failure count for the source group provided
377 */
378 static int
379 get_sg_cnt(req)
380 struct sioc_sg_req6 *req;
381 {
382 struct mf6c *rt;
383 int s;
384
385 s = splnet();
386 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
387 splx(s);
388 if (rt != NULL) {
389 req->pktcnt = rt->mf6c_pkt_cnt;
390 req->bytecnt = rt->mf6c_byte_cnt;
391 req->wrong_if = rt->mf6c_wrong_if;
392 } else
393 return (ESRCH);
394 #if 0
395 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff;
396 #endif
397
398 return (0);
399 }
400
401 /*
402 * returns the input and output packet and byte counts on the mif provided
403 */
404 static int
405 get_mif6_cnt(req)
406 struct sioc_mif_req6 *req;
407 {
408 mifi_t mifi = req->mifi;
409
410 if (mifi >= nummifs)
411 return (EINVAL);
412
413 req->icount = mif6table[mifi].m6_pkt_in;
414 req->ocount = mif6table[mifi].m6_pkt_out;
415 req->ibytes = mif6table[mifi].m6_bytes_in;
416 req->obytes = mif6table[mifi].m6_bytes_out;
417
418 return (0);
419 }
420
421 static int
422 set_pim6(i)
423 int *i;
424 {
425 if ((*i != 1) && (*i != 0))
426 return (EINVAL);
427
428 pim6 = *i;
429
430 return (0);
431 }
432
433 /*
434 * Enable multicast routing
435 */
436 static int
437 ip6_mrouter_init(so, v, cmd)
438 struct socket *so;
439 int v;
440 int cmd;
441 {
442 #ifdef MRT6DEBUG
443 if (mrt6debug)
444 log(LOG_DEBUG,
445 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
446 so->so_type, so->so_proto->pr_protocol);
447 #endif
448
449 if (so->so_type != SOCK_RAW ||
450 so->so_proto->pr_protocol != IPPROTO_ICMPV6)
451 return (EOPNOTSUPP);
452
453 if (v != 1)
454 return (ENOPROTOOPT);
455
456 if (ip6_mrouter != NULL)
457 return (EADDRINUSE);
458
459 ip6_mrouter = so;
460 ip6_mrouter_ver = cmd;
461
462 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
463 bzero((caddr_t)n6expire, sizeof(n6expire));
464
465 pim6 = 0;/* used for stubbing out/in pim stuff */
466
467 callout_init(&expire_upcalls_ch, 0);
468 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
469 expire_upcalls, NULL);
470
471 #ifdef MRT6DEBUG
472 if (mrt6debug)
473 log(LOG_DEBUG, "ip6_mrouter_init\n");
474 #endif
475
476 return (0);
477 }
478
479 /*
480 * Disable multicast routing
481 */
482 int
483 ip6_mrouter_done()
484 {
485 mifi_t mifi;
486 int i;
487 struct mf6c *rt;
488 struct rtdetq *rte;
489 int s;
490
491 s = splnet();
492
493 /*
494 * For each phyint in use, disable promiscuous reception of all IPv6
495 * multicasts.
496 */
497 #ifdef INET
498 #ifdef MROUTING
499 /*
500 * If there is still IPv4 multicast routing daemon,
501 * we remain interfaces to receive all muliticasted packets.
502 * XXX: there may be an interface in which the IPv4 multicast
503 * daemon is not interested...
504 */
505 if (!ip_mrouter)
506 #endif
507 #endif
508 {
509 for (mifi = 0; mifi < nummifs; mifi++) {
510 if (mif6table[mifi].m6_ifp &&
511 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
512 if_allmulti(mif6table[mifi].m6_ifp, 0);
513 }
514 }
515 }
516 #ifdef notyet
517 bzero((caddr_t)qtable, sizeof(qtable));
518 bzero((caddr_t)tbftable, sizeof(tbftable));
519 #endif
520 bzero((caddr_t)mif6table, sizeof(mif6table));
521 nummifs = 0;
522
523 pim6 = 0; /* used to stub out/in pim specific code */
524
525 callout_stop(&expire_upcalls_ch);
526
527 /*
528 * Free all multicast forwarding cache entries.
529 */
530 for (i = 0; i < MF6CTBLSIZ; i++) {
531 rt = mf6ctable[i];
532 while (rt) {
533 struct mf6c *frt;
534
535 for (rte = rt->mf6c_stall; rte != NULL; ) {
536 struct rtdetq *n = rte->next;
537
538 m_free(rte->m);
539 free(rte, M_MRTABLE6);
540 rte = n;
541 }
542 frt = rt;
543 rt = rt->mf6c_next;
544 free(frt, M_MRTABLE6);
545 }
546 }
547
548 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
549
550 /*
551 * Reset register interface
552 */
553 if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
554 if_detach(multicast_register_if6);
555 if_free(multicast_register_if6);
556 reg_mif_num = (mifi_t)-1;
557 multicast_register_if6 = NULL;
558 }
559
560 ip6_mrouter = NULL;
561 ip6_mrouter_ver = 0;
562
563 splx(s);
564
565 #ifdef MRT6DEBUG
566 if (mrt6debug)
567 log(LOG_DEBUG, "ip6_mrouter_done\n");
568 #endif
569
570 return (0);
571 }
572
573 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
574
575 /*
576 * Add a mif to the mif table
577 */
578 static int
579 add_m6if(mifcp)
580 struct mif6ctl *mifcp;
581 {
582 struct mif6 *mifp;
583 struct ifnet *ifp;
584 int error, s;
585 #ifdef notyet
586 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi;
587 #endif
588
589 if (mifcp->mif6c_mifi >= MAXMIFS)
590 return (EINVAL);
591 mifp = mif6table + mifcp->mif6c_mifi;
592 if (mifp->m6_ifp)
593 return (EADDRINUSE); /* XXX: is it appropriate? */
594 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > if_index)
595 return (ENXIO);
596 ifp = ifnet_byindex(mifcp->mif6c_pifi);
597
598 if (mifcp->mif6c_flags & MIFF_REGISTER) {
599 if (reg_mif_num == (mifi_t)-1) {
600 ifp = if_alloc(IFT_OTHER);
601
602 if_initname(ifp, "register_mif", 0);
603 ifp->if_flags |= IFF_LOOPBACK;
604 if_attach(ifp);
605 multicast_register_if6 = ifp;
606 reg_mif_num = mifcp->mif6c_mifi;
607 /*
608 * it is impossible to guess the ifindex of the
609 * register interface. So mif6c_pifi is automatically
610 * calculated.
611 */
612 mifcp->mif6c_pifi = ifp->if_index;
613 } else {
614 ifp = multicast_register_if6;
615 }
616
617 } /* if REGISTER */
618 else {
619 /* Make sure the interface supports multicast */
620 if ((ifp->if_flags & IFF_MULTICAST) == 0)
621 return (EOPNOTSUPP);
622
623 s = splnet();
624 error = if_allmulti(ifp, 1);
625 splx(s);
626 if (error)
627 return (error);
628 }
629
630 s = splnet();
631 mifp->m6_flags = mifcp->mif6c_flags;
632 mifp->m6_ifp = ifp;
633 #ifdef notyet
634 /* scaling up here allows division by 1024 in critical code */
635 mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000;
636 #endif
637 /* initialize per mif pkt counters */
638 mifp->m6_pkt_in = 0;
639 mifp->m6_pkt_out = 0;
640 mifp->m6_bytes_in = 0;
641 mifp->m6_bytes_out = 0;
642 splx(s);
643
644 /* Adjust nummifs up if the mifi is higher than nummifs */
645 if (nummifs <= mifcp->mif6c_mifi)
646 nummifs = mifcp->mif6c_mifi + 1;
647
648 #ifdef MRT6DEBUG
649 if (mrt6debug)
650 log(LOG_DEBUG,
651 "add_mif #%d, phyint %s\n",
652 mifcp->mif6c_mifi,
653 ifp->if_xname);
654 #endif
655
656 return (0);
657 }
658
659 /*
660 * Delete a mif from the mif table
661 */
662 static int
663 del_m6if(mifip)
664 mifi_t *mifip;
665 {
666 struct mif6 *mifp = mif6table + *mifip;
667 mifi_t mifi;
668 struct ifnet *ifp;
669 int s;
670
671 if (*mifip >= nummifs)
672 return (EINVAL);
673 if (mifp->m6_ifp == NULL)
674 return (EINVAL);
675
676 s = splnet();
677
678 if (!(mifp->m6_flags & MIFF_REGISTER)) {
679 /*
680 * XXX: what if there is yet IPv4 multicast daemon
681 * using the interface?
682 */
683 ifp = mifp->m6_ifp;
684
685 if_allmulti(ifp, 0);
686 } else {
687 if (reg_mif_num != (mifi_t)-1 &&
688 multicast_register_if6 != NULL) {
689 if_detach(multicast_register_if6);
690 if_free(multicast_register_if6);
691 reg_mif_num = (mifi_t)-1;
692 multicast_register_if6 = NULL;
693 }
694 }
695
696 #ifdef notyet
697 bzero((caddr_t)qtable[*mifip], sizeof(qtable[*mifip]));
698 bzero((caddr_t)mifp->m6_tbf, sizeof(*(mifp->m6_tbf)));
699 #endif
700 bzero((caddr_t)mifp, sizeof(*mifp));
701
702 /* Adjust nummifs down */
703 for (mifi = nummifs; mifi > 0; mifi--)
704 if (mif6table[mifi - 1].m6_ifp)
705 break;
706 nummifs = mifi;
707
708 splx(s);
709
710 #ifdef MRT6DEBUG
711 if (mrt6debug)
712 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
713 #endif
714
715 return (0);
716 }
717
718 /*
719 * Add an mfc entry
720 */
721 static int
722 add_m6fc(mfccp)
723 struct mf6cctl *mfccp;
724 {
725 struct mf6c *rt;
726 u_long hash;
727 struct rtdetq *rte;
728 u_short nstl;
729 int s;
730
731 MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
732 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
733
734 /* If an entry already exists, just update the fields */
735 if (rt) {
736 #ifdef MRT6DEBUG
737 if (mrt6debug & DEBUG_MFC)
738 log(LOG_DEBUG,
739 "add_m6fc no upcall h %d o %s g %s p %x\n",
740 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
741 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
742 mfccp->mf6cc_parent);
743 #endif
744
745 s = splnet();
746 rt->mf6c_parent = mfccp->mf6cc_parent;
747 rt->mf6c_ifset = mfccp->mf6cc_ifset;
748 splx(s);
749 return (0);
750 }
751
752 /*
753 * Find the entry for which the upcall was made and update
754 */
755 s = splnet();
756 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
757 mfccp->mf6cc_mcastgrp.sin6_addr);
758 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
759 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
760 &mfccp->mf6cc_origin.sin6_addr) &&
761 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
762 &mfccp->mf6cc_mcastgrp.sin6_addr) &&
763 (rt->mf6c_stall != NULL)) {
764
765 if (nstl++)
766 log(LOG_ERR,
767 "add_m6fc: %s o %s g %s p %x dbx %p\n",
768 "multiple kernel entries",
769 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
770 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
771 mfccp->mf6cc_parent, rt->mf6c_stall);
772
773 #ifdef MRT6DEBUG
774 if (mrt6debug & DEBUG_MFC)
775 log(LOG_DEBUG,
776 "add_m6fc o %s g %s p %x dbg %x\n",
777 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
778 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
779 mfccp->mf6cc_parent, rt->mf6c_stall);
780 #endif
781
782 rt->mf6c_origin = mfccp->mf6cc_origin;
783 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
784 rt->mf6c_parent = mfccp->mf6cc_parent;
785 rt->mf6c_ifset = mfccp->mf6cc_ifset;
786 /* initialize pkt counters per src-grp */
787 rt->mf6c_pkt_cnt = 0;
788 rt->mf6c_byte_cnt = 0;
789 rt->mf6c_wrong_if = 0;
790
791 rt->mf6c_expire = 0; /* Don't clean this guy up */
792 n6expire[hash]--;
793
794 /* free packets Qed at the end of this entry */
795 for (rte = rt->mf6c_stall; rte != NULL; ) {
796 struct rtdetq *n = rte->next;
797 ip6_mdq(rte->m, rte->ifp, rt);
798 m_freem(rte->m);
799 #ifdef UPCALL_TIMING
800 collate(&(rte->t));
801 #endif /* UPCALL_TIMING */
802 free(rte, M_MRTABLE6);
803 rte = n;
804 }
805 rt->mf6c_stall = NULL;
806 }
807 }
808
809 /*
810 * It is possible that an entry is being inserted without an upcall
811 */
812 if (nstl == 0) {
813 #ifdef MRT6DEBUG
814 if (mrt6debug & DEBUG_MFC)
815 log(LOG_DEBUG,
816 "add_mfc no upcall h %d o %s g %s p %x\n",
817 hash,
818 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
819 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
820 mfccp->mf6cc_parent);
821 #endif
822
823 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
824
825 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
826 &mfccp->mf6cc_origin.sin6_addr)&&
827 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
828 &mfccp->mf6cc_mcastgrp.sin6_addr)) {
829
830 rt->mf6c_origin = mfccp->mf6cc_origin;
831 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
832 rt->mf6c_parent = mfccp->mf6cc_parent;
833 rt->mf6c_ifset = mfccp->mf6cc_ifset;
834 /* initialize pkt counters per src-grp */
835 rt->mf6c_pkt_cnt = 0;
836 rt->mf6c_byte_cnt = 0;
837 rt->mf6c_wrong_if = 0;
838
839 if (rt->mf6c_expire)
840 n6expire[hash]--;
841 rt->mf6c_expire = 0;
842 }
843 }
844 if (rt == NULL) {
845 /* no upcall, so make a new entry */
846 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
847 M_NOWAIT);
848 if (rt == NULL) {
849 splx(s);
850 return (ENOBUFS);
851 }
852
853 /* insert new entry at head of hash chain */
854 rt->mf6c_origin = mfccp->mf6cc_origin;
855 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
856 rt->mf6c_parent = mfccp->mf6cc_parent;
857 rt->mf6c_ifset = mfccp->mf6cc_ifset;
858 /* initialize pkt counters per src-grp */
859 rt->mf6c_pkt_cnt = 0;
860 rt->mf6c_byte_cnt = 0;
861 rt->mf6c_wrong_if = 0;
862 rt->mf6c_expire = 0;
863 rt->mf6c_stall = NULL;
864
865 /* link into table */
866 rt->mf6c_next = mf6ctable[hash];
867 mf6ctable[hash] = rt;
868 }
869 }
870 splx(s);
871 return (0);
872 }
873
874 #ifdef UPCALL_TIMING
875 /*
876 * collect delay statistics on the upcalls
877 */
878 static void
879 collate(t)
880 struct timeval *t;
881 {
882 u_long d;
883 struct timeval tp;
884 u_long delta;
885
886 GET_TIME(tp);
887
888 if (TV_LT(*t, tp))
889 {
890 TV_DELTA(tp, *t, delta);
891
892 d = delta >> 10;
893 if (d > UPCALL_MAX)
894 d = UPCALL_MAX;
895
896 ++upcall_data[d];
897 }
898 }
899 #endif /* UPCALL_TIMING */
900
901 /*
902 * Delete an mfc entry
903 */
904 static int
905 del_m6fc(mfccp)
906 struct mf6cctl *mfccp;
907 {
908 struct sockaddr_in6 origin;
909 struct sockaddr_in6 mcastgrp;
910 struct mf6c *rt;
911 struct mf6c **nptr;
912 u_long hash;
913 int s;
914
915 origin = mfccp->mf6cc_origin;
916 mcastgrp = mfccp->mf6cc_mcastgrp;
917 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
918
919 #ifdef MRT6DEBUG
920 if (mrt6debug & DEBUG_MFC)
921 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
922 ip6_sprintf(&origin.sin6_addr),
923 ip6_sprintf(&mcastgrp.sin6_addr));
924 #endif
925
926 s = splnet();
927
928 nptr = &mf6ctable[hash];
929 while ((rt = *nptr) != NULL) {
930 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
931 &rt->mf6c_origin.sin6_addr) &&
932 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
933 &rt->mf6c_mcastgrp.sin6_addr) &&
934 rt->mf6c_stall == NULL)
935 break;
936
937 nptr = &rt->mf6c_next;
938 }
939 if (rt == NULL) {
940 splx(s);
941 return (EADDRNOTAVAIL);
942 }
943
944 *nptr = rt->mf6c_next;
945 free(rt, M_MRTABLE6);
946
947 splx(s);
948
949 return (0);
950 }
951
952 static int
953 socket_send(s, mm, src)
954 struct socket *s;
955 struct mbuf *mm;
956 struct sockaddr_in6 *src;
957 {
958 if (s) {
959 if (sbappendaddr(&s->so_rcv,
960 (struct sockaddr *)src,
961 mm, (struct mbuf *)0) != 0) {
962 sorwakeup(s);
963 return (0);
964 }
965 }
966 m_freem(mm);
967 return (-1);
968 }
969
970 /*
971 * IPv6 multicast forwarding function. This function assumes that the packet
972 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
973 * pointed to by "ifp", and the packet is to be relayed to other networks
974 * that have members of the packet's destination IPv6 multicast group.
975 *
976 * The packet is returned unscathed to the caller, unless it is
977 * erroneous, in which case a non-zero return value tells the caller to
978 * discard it.
979 *
980 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
981 * this function is called in the originating context (i.e., not when
982 * forwarding a packet from other node). ip6_output(), which is currently the
983 * only function that calls this function is called in the originating context,
984 * explicitly ensures this condition. It is caller's responsibility to ensure
985 * that if this function is called from somewhere else in the originating
986 * context in the future.
987 */
988
989 int
990 ip6_mforward(ip6, ifp, m)
991 struct ip6_hdr *ip6;
992 struct ifnet *ifp;
993 struct mbuf *m;
994 {
995 struct mf6c *rt;
996 struct mif6 *mifp;
997 struct mbuf *mm;
998 int s;
999 mifi_t mifi;
1000
1001 #ifdef MRT6DEBUG
1002 if (mrt6debug & DEBUG_FORWARD)
1003 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1004 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst),
1005 ifp->if_index);
1006 #endif
1007
1008 /*
1009 * Don't forward a packet with Hop limit of zero or one,
1010 * or a packet destined to a local-only group.
1011 */
1012 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1013 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1014 return (0);
1015 ip6->ip6_hlim--;
1016
1017 /*
1018 * Source address check: do not forward packets with unspecified
1019 * source. It was discussed in July 2000, on ipngwg mailing list.
1020 * This is rather more serious than unicast cases, because some
1021 * MLD packets can be sent with the unspecified source address
1022 * (although such packets must normally set 1 to the hop limit field).
1023 */
1024 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1025 ip6stat.ip6s_cantforward++;
1026 if (ip6_log_time + ip6_log_interval < time_second) {
1027 ip6_log_time = time_second;
1028 log(LOG_DEBUG,
1029 "cannot forward "
1030 "from %s to %s nxt %d received on %s\n",
1031 ip6_sprintf(&ip6->ip6_src),
1032 ip6_sprintf(&ip6->ip6_dst),
1033 ip6->ip6_nxt,
1034 if_name(m->m_pkthdr.rcvif));
1035 }
1036 return (0);
1037 }
1038
1039 /*
1040 * Determine forwarding mifs from the forwarding cache table
1041 */
1042 s = splnet();
1043 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1044
1045 /* Entry exists, so forward if necessary */
1046 if (rt) {
1047 splx(s);
1048 return (ip6_mdq(m, ifp, rt));
1049 } else {
1050 /*
1051 * If we don't have a route for packet's origin,
1052 * Make a copy of the packet &
1053 * send message to routing daemon
1054 */
1055
1056 struct mbuf *mb0;
1057 struct rtdetq *rte;
1058 u_long hash;
1059 /* int i, npkts;*/
1060 #ifdef UPCALL_TIMING
1061 struct timeval tp;
1062
1063 GET_TIME(tp);
1064 #endif /* UPCALL_TIMING */
1065
1066 mrt6stat.mrt6s_no_route++;
1067 #ifdef MRT6DEBUG
1068 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1069 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1070 ip6_sprintf(&ip6->ip6_src),
1071 ip6_sprintf(&ip6->ip6_dst));
1072 #endif
1073
1074 /*
1075 * Allocate mbufs early so that we don't do extra work if we
1076 * are just going to fail anyway.
1077 */
1078 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6,
1079 M_NOWAIT);
1080 if (rte == NULL) {
1081 splx(s);
1082 return (ENOBUFS);
1083 }
1084 mb0 = m_copy(m, 0, M_COPYALL);
1085 /*
1086 * Pullup packet header if needed before storing it,
1087 * as other references may modify it in the meantime.
1088 */
1089 if (mb0 &&
1090 (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1091 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1092 if (mb0 == NULL) {
1093 free(rte, M_MRTABLE6);
1094 splx(s);
1095 return (ENOBUFS);
1096 }
1097
1098 /* is there an upcall waiting for this packet? */
1099 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1100 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1101 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1102 &rt->mf6c_origin.sin6_addr) &&
1103 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1104 &rt->mf6c_mcastgrp.sin6_addr) &&
1105 (rt->mf6c_stall != NULL))
1106 break;
1107 }
1108
1109 if (rt == NULL) {
1110 struct mrt6msg *im;
1111 #ifdef MRT6_OINIT
1112 struct omrt6msg *oim;
1113 #endif
1114
1115 /* no upcall, so make a new entry */
1116 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
1117 M_NOWAIT);
1118 if (rt == NULL) {
1119 free(rte, M_MRTABLE6);
1120 m_freem(mb0);
1121 splx(s);
1122 return (ENOBUFS);
1123 }
1124 /*
1125 * Make a copy of the header to send to the user
1126 * level process
1127 */
1128 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1129
1130 if (mm == NULL) {
1131 free(rte, M_MRTABLE6);
1132 m_freem(mb0);
1133 free(rt, M_MRTABLE6);
1134 splx(s);
1135 return (ENOBUFS);
1136 }
1137
1138 /*
1139 * Send message to routing daemon
1140 */
1141 sin6.sin6_addr = ip6->ip6_src;
1142
1143 im = NULL;
1144 #ifdef MRT6_OINIT
1145 oim = NULL;
1146 #endif
1147 switch (ip6_mrouter_ver) {
1148 #ifdef MRT6_OINIT
1149 case MRT6_OINIT:
1150 oim = mtod(mm, struct omrt6msg *);
1151 oim->im6_msgtype = MRT6MSG_NOCACHE;
1152 oim->im6_mbz = 0;
1153 break;
1154 #endif
1155 case MRT6_INIT:
1156 im = mtod(mm, struct mrt6msg *);
1157 im->im6_msgtype = MRT6MSG_NOCACHE;
1158 im->im6_mbz = 0;
1159 break;
1160 default:
1161 free(rte, M_MRTABLE6);
1162 m_freem(mb0);
1163 free(rt, M_MRTABLE6);
1164 splx(s);
1165 return (EINVAL);
1166 }
1167
1168 #ifdef MRT6DEBUG
1169 if (mrt6debug & DEBUG_FORWARD)
1170 log(LOG_DEBUG,
1171 "getting the iif info in the kernel\n");
1172 #endif
1173
1174 for (mifp = mif6table, mifi = 0;
1175 mifi < nummifs && mifp->m6_ifp != ifp;
1176 mifp++, mifi++)
1177 ;
1178
1179 switch (ip6_mrouter_ver) {
1180 #ifdef MRT6_OINIT
1181 case MRT6_OINIT:
1182 oim->im6_mif = mifi;
1183 break;
1184 #endif
1185 case MRT6_INIT:
1186 im->im6_mif = mifi;
1187 break;
1188 }
1189
1190 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1191 log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1192 "socket queue full\n");
1193 mrt6stat.mrt6s_upq_sockfull++;
1194 free(rte, M_MRTABLE6);
1195 m_freem(mb0);
1196 free(rt, M_MRTABLE6);
1197 splx(s);
1198 return (ENOBUFS);
1199 }
1200
1201 mrt6stat.mrt6s_upcalls++;
1202
1203 /* insert new entry at head of hash chain */
1204 bzero(rt, sizeof(*rt));
1205 rt->mf6c_origin.sin6_family = AF_INET6;
1206 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1207 rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1208 rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1209 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1210 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1211 rt->mf6c_expire = UPCALL_EXPIRE;
1212 n6expire[hash]++;
1213 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1214
1215 /* link into table */
1216 rt->mf6c_next = mf6ctable[hash];
1217 mf6ctable[hash] = rt;
1218 /* Add this entry to the end of the queue */
1219 rt->mf6c_stall = rte;
1220 } else {
1221 /* determine if q has overflowed */
1222 struct rtdetq **p;
1223 int npkts = 0;
1224
1225 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1226 if (++npkts > MAX_UPQ6) {
1227 mrt6stat.mrt6s_upq_ovflw++;
1228 free(rte, M_MRTABLE6);
1229 m_freem(mb0);
1230 splx(s);
1231 return (0);
1232 }
1233
1234 /* Add this entry to the end of the queue */
1235 *p = rte;
1236 }
1237
1238 rte->next = NULL;
1239 rte->m = mb0;
1240 rte->ifp = ifp;
1241 #ifdef UPCALL_TIMING
1242 rte->t = tp;
1243 #endif /* UPCALL_TIMING */
1244
1245 splx(s);
1246
1247 return (0);
1248 }
1249 }
1250
1251 /*
1252 * Clean up cache entries if upcalls are not serviced
1253 * Call from the Slow Timeout mechanism, every half second.
1254 */
1255 static void
1256 expire_upcalls(unused)
1257 void *unused;
1258 {
1259 struct rtdetq *rte;
1260 struct mf6c *mfc, **nptr;
1261 int i;
1262 int s;
1263
1264 s = splnet();
1265 for (i = 0; i < MF6CTBLSIZ; i++) {
1266 if (n6expire[i] == 0)
1267 continue;
1268 nptr = &mf6ctable[i];
1269 while ((mfc = *nptr) != NULL) {
1270 rte = mfc->mf6c_stall;
1271 /*
1272 * Skip real cache entries
1273 * Make sure it wasn't marked to not expire (shouldn't happen)
1274 * If it expires now
1275 */
1276 if (rte != NULL &&
1277 mfc->mf6c_expire != 0 &&
1278 --mfc->mf6c_expire == 0) {
1279 #ifdef MRT6DEBUG
1280 if (mrt6debug & DEBUG_EXPIRE)
1281 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1282 ip6_sprintf(&mfc->mf6c_origin.sin6_addr),
1283 ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr));
1284 #endif
1285 /*
1286 * drop all the packets
1287 * free the mbuf with the pkt, if, timing info
1288 */
1289 do {
1290 struct rtdetq *n = rte->next;
1291 m_freem(rte->m);
1292 free(rte, M_MRTABLE6);
1293 rte = n;
1294 } while (rte != NULL);
1295 mrt6stat.mrt6s_cache_cleanups++;
1296 n6expire[i]--;
1297
1298 *nptr = mfc->mf6c_next;
1299 free(mfc, M_MRTABLE6);
1300 } else {
1301 nptr = &mfc->mf6c_next;
1302 }
1303 }
1304 }
1305 splx(s);
1306 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1307 expire_upcalls, NULL);
1308 }
1309
1310 /*
1311 * Packet forwarding routine once entry in the cache is made
1312 */
1313 static int
1314 ip6_mdq(m, ifp, rt)
1315 struct mbuf *m;
1316 struct ifnet *ifp;
1317 struct mf6c *rt;
1318 {
1319 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1320 mifi_t mifi, iif;
1321 struct mif6 *mifp;
1322 int plen = m->m_pkthdr.len;
1323 struct in6_addr src0, dst0; /* copies for local work */
1324 u_int32_t iszone, idzone, oszone, odzone;
1325 int error = 0;
1326
1327 /*
1328 * Macro to send packet on mif. Since RSVP packets don't get counted on
1329 * input, they shouldn't get counted on output, so statistics keeping is
1330 * separate.
1331 */
1332
1333 #define MC6_SEND(ip6, mifp, m) do { \
1334 if ((mifp)->m6_flags & MIFF_REGISTER) \
1335 register_send((ip6), (mifp), (m)); \
1336 else \
1337 phyint_send((ip6), (mifp), (m)); \
1338 } while (/*CONSTCOND*/ 0)
1339
1340 /*
1341 * Don't forward if it didn't arrive from the parent mif
1342 * for its origin.
1343 */
1344 mifi = rt->mf6c_parent;
1345 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1346 /* came in the wrong interface */
1347 #ifdef MRT6DEBUG
1348 if (mrt6debug & DEBUG_FORWARD)
1349 log(LOG_DEBUG,
1350 "wrong if: ifid %d mifi %d mififid %x\n",
1351 ifp->if_index, mifi,
1352 mif6table[mifi].m6_ifp->if_index);
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 #ifdef MRT6_OINIT
1375 struct omrt6msg *oim;
1376 #endif
1377
1378 mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1379 if (mm &&
1380 (M_HASCL(mm) ||
1381 mm->m_len < sizeof(struct ip6_hdr)))
1382 mm = m_pullup(mm, sizeof(struct ip6_hdr));
1383 if (mm == NULL)
1384 return (ENOBUFS);
1385
1386 #ifdef MRT6_OINIT
1387 oim = NULL;
1388 #endif
1389 im = NULL;
1390 switch (ip6_mrouter_ver) {
1391 #ifdef MRT6_OINIT
1392 case MRT6_OINIT:
1393 oim = mtod(mm, struct omrt6msg *);
1394 oim->im6_msgtype = MRT6MSG_WRONGMIF;
1395 oim->im6_mbz = 0;
1396 break;
1397 #endif
1398 case MRT6_INIT:
1399 im = mtod(mm, struct mrt6msg *);
1400 im->im6_msgtype = MRT6MSG_WRONGMIF;
1401 im->im6_mbz = 0;
1402 break;
1403 default:
1404 m_freem(mm);
1405 return (EINVAL);
1406 }
1407
1408 for (mifp = mif6table, iif = 0;
1409 iif < nummifs && mifp &&
1410 mifp->m6_ifp != ifp;
1411 mifp++, iif++)
1412 ;
1413
1414 switch (ip6_mrouter_ver) {
1415 #ifdef MRT6_OINIT
1416 case MRT6_OINIT:
1417 oim->im6_mif = iif;
1418 sin6.sin6_addr = oim->im6_src;
1419 break;
1420 #endif
1421 case MRT6_INIT:
1422 im->im6_mif = iif;
1423 sin6.sin6_addr = im->im6_src;
1424 break;
1425 }
1426
1427 mrt6stat.mrt6s_upcalls++;
1428
1429 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1430 #ifdef MRT6DEBUG
1431 if (mrt6debug)
1432 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1433 #endif
1434 ++mrt6stat.mrt6s_upq_sockfull;
1435 return (ENOBUFS);
1436 } /* if socket Q full */
1437 } /* if PIM */
1438 return (0);
1439 } /* if wrong iif */
1440
1441 /* If I sourced this packet, it counts as output, else it was input. */
1442 if (m->m_pkthdr.rcvif == NULL) {
1443 /* XXX: is rcvif really NULL when output?? */
1444 mif6table[mifi].m6_pkt_out++;
1445 mif6table[mifi].m6_bytes_out += plen;
1446 } else {
1447 mif6table[mifi].m6_pkt_in++;
1448 mif6table[mifi].m6_bytes_in += plen;
1449 }
1450 rt->mf6c_pkt_cnt++;
1451 rt->mf6c_byte_cnt += plen;
1452
1453 /*
1454 * For each mif, forward a copy of the packet if there are group
1455 * members downstream on the interface.
1456 */
1457 src0 = ip6->ip6_src;
1458 dst0 = ip6->ip6_dst;
1459 if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1460 (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1461 ip6stat.ip6s_badscope++;
1462 return (error);
1463 }
1464 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1465 if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1466 /*
1467 * check if the outgoing packet is going to break
1468 * a scope boundary.
1469 * XXX For packets through PIM register tunnel
1470 * interface, we believe a routing daemon.
1471 */
1472 if (!(mif6table[rt->mf6c_parent].m6_flags &
1473 MIFF_REGISTER) &&
1474 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1475 if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1476 &oszone) ||
1477 in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1478 &odzone) ||
1479 iszone != oszone ||
1480 idzone != odzone) {
1481 ip6stat.ip6s_badscope++;
1482 continue;
1483 }
1484 }
1485
1486 mifp->m6_pkt_out++;
1487 mifp->m6_bytes_out += plen;
1488 MC6_SEND(ip6, mifp, m);
1489 }
1490 }
1491 return (0);
1492 }
1493
1494 static void
1495 phyint_send(ip6, mifp, m)
1496 struct ip6_hdr *ip6;
1497 struct mif6 *mifp;
1498 struct mbuf *m;
1499 {
1500 struct mbuf *mb_copy;
1501 struct ifnet *ifp = mifp->m6_ifp;
1502 int error = 0;
1503 int s = splnet(); /* needs to protect static "ro" below. */
1504 static struct route_in6 ro;
1505 struct in6_multi *in6m;
1506 struct sockaddr_in6 *dst6;
1507 u_long linkmtu;
1508
1509 /*
1510 * Make a new reference to the packet; make sure that
1511 * the IPv6 header is actually copied, not just referenced,
1512 * so that ip6_output() only scribbles on the copy.
1513 */
1514 mb_copy = m_copy(m, 0, M_COPYALL);
1515 if (mb_copy &&
1516 (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1517 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1518 if (mb_copy == NULL) {
1519 splx(s);
1520 return;
1521 }
1522 /* set MCAST flag to the outgoing packet */
1523 mb_copy->m_flags |= M_MCAST;
1524
1525 /*
1526 * If we sourced the packet, call ip6_output since we may devide
1527 * the packet into fragments when the packet is too big for the
1528 * outgoing interface.
1529 * Otherwise, we can simply send the packet to the interface
1530 * sending queue.
1531 */
1532 if (m->m_pkthdr.rcvif == NULL) {
1533 struct ip6_moptions im6o;
1534
1535 im6o.im6o_multicast_ifp = ifp;
1536 /* XXX: ip6_output will override ip6->ip6_hlim */
1537 im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1538 im6o.im6o_multicast_loop = 1;
1539 error = ip6_output(mb_copy, NULL, &ro,
1540 IPV6_FORWARDING, &im6o, NULL, NULL);
1541
1542 #ifdef MRT6DEBUG
1543 if (mrt6debug & DEBUG_XMIT)
1544 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1545 mifp - mif6table, error);
1546 #endif
1547 splx(s);
1548 return;
1549 }
1550
1551 /*
1552 * If we belong to the destination multicast group
1553 * on the outgoing interface, loop back a copy.
1554 */
1555 dst6 = (struct sockaddr_in6 *)&ro.ro_dst;
1556 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m);
1557 if (in6m != NULL) {
1558 dst6->sin6_len = sizeof(struct sockaddr_in6);
1559 dst6->sin6_family = AF_INET6;
1560 dst6->sin6_addr = ip6->ip6_dst;
1561 ip6_mloopback(ifp, m, (struct sockaddr_in6 *)&ro.ro_dst);
1562 }
1563 /*
1564 * Put the packet into the sending queue of the outgoing interface
1565 * if it would fit in the MTU of the interface.
1566 */
1567 linkmtu = IN6_LINKMTU(ifp);
1568 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1569 dst6->sin6_len = sizeof(struct sockaddr_in6);
1570 dst6->sin6_family = AF_INET6;
1571 dst6->sin6_addr = ip6->ip6_dst;
1572 /*
1573 * We just call if_output instead of nd6_output here, since
1574 * we need no ND for a multicast forwarded packet...right?
1575 */
1576 error = (*ifp->if_output)(ifp, mb_copy,
1577 (struct sockaddr *)&ro.ro_dst, NULL);
1578 #ifdef MRT6DEBUG
1579 if (mrt6debug & DEBUG_XMIT)
1580 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1581 mifp - mif6table, error);
1582 #endif
1583 } else {
1584 /*
1585 * pMTU discovery is intentionally disabled by default, since
1586 * various router may notify pMTU in multicast, which can be
1587 * a DDoS to a router
1588 */
1589 if (ip6_mcast_pmtu)
1590 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1591 else {
1592 #ifdef MRT6DEBUG
1593 if (mrt6debug & DEBUG_XMIT)
1594 log(LOG_DEBUG,
1595 "phyint_send: packet too big on %s o %s "
1596 "g %s size %d(discarded)\n",
1597 if_name(ifp),
1598 ip6_sprintf(&ip6->ip6_src),
1599 ip6_sprintf(&ip6->ip6_dst),
1600 mb_copy->m_pkthdr.len);
1601 #endif /* MRT6DEBUG */
1602 m_freem(mb_copy); /* simply discard the packet */
1603 }
1604 }
1605
1606 splx(s);
1607 }
1608
1609 static int
1610 register_send(ip6, mif, m)
1611 struct ip6_hdr *ip6;
1612 struct mif6 *mif;
1613 struct mbuf *m;
1614 {
1615 struct mbuf *mm;
1616 int i, len = m->m_pkthdr.len;
1617 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1618 struct mrt6msg *im6;
1619
1620 #ifdef MRT6DEBUG
1621 if (mrt6debug)
1622 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1623 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst));
1624 #endif
1625 ++pim6stat.pim6s_snd_registers;
1626
1627 /* Make a copy of the packet to send to the user level process */
1628 MGETHDR(mm, M_DONTWAIT, MT_HEADER);
1629 if (mm == NULL)
1630 return (ENOBUFS);
1631 mm->m_pkthdr.rcvif = NULL;
1632 mm->m_data += max_linkhdr;
1633 mm->m_len = sizeof(struct ip6_hdr);
1634
1635 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1636 m_freem(mm);
1637 return (ENOBUFS);
1638 }
1639 i = MHLEN - M_LEADINGSPACE(mm);
1640 if (i > len)
1641 i = len;
1642 mm = m_pullup(mm, i);
1643 if (mm == NULL)
1644 return (ENOBUFS);
1645 /* TODO: check it! */
1646 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1647
1648 /*
1649 * Send message to routing daemon
1650 */
1651 sin6.sin6_addr = ip6->ip6_src;
1652
1653 im6 = mtod(mm, struct mrt6msg *);
1654 im6->im6_msgtype = MRT6MSG_WHOLEPKT;
1655 im6->im6_mbz = 0;
1656
1657 im6->im6_mif = mif - mif6table;
1658
1659 /* iif info is not given for reg. encap.n */
1660 mrt6stat.mrt6s_upcalls++;
1661
1662 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1663 #ifdef MRT6DEBUG
1664 if (mrt6debug)
1665 log(LOG_WARNING,
1666 "register_send: ip6_mrouter socket queue full\n");
1667 #endif
1668 ++mrt6stat.mrt6s_upq_sockfull;
1669 return (ENOBUFS);
1670 }
1671 return (0);
1672 }
1673
1674 /*
1675 * PIM sparse mode hook
1676 * Receives the pim control messages, and passes them up to the listening
1677 * socket, using rip6_input.
1678 * The only message processed is the REGISTER pim message; the pim header
1679 * is stripped off, and the inner packet is passed to register_mforward.
1680 */
1681 int
1682 pim6_input(mp, offp, proto)
1683 struct mbuf **mp;
1684 int *offp, proto;
1685 {
1686 struct pim *pim; /* pointer to a pim struct */
1687 struct ip6_hdr *ip6;
1688 int pimlen;
1689 struct mbuf *m = *mp;
1690 int minlen;
1691 int off = *offp;
1692
1693 ++pim6stat.pim6s_rcv_total;
1694
1695 ip6 = mtod(m, struct ip6_hdr *);
1696 pimlen = m->m_pkthdr.len - *offp;
1697
1698 /*
1699 * Validate lengths
1700 */
1701 if (pimlen < PIM_MINLEN) {
1702 ++pim6stat.pim6s_rcv_tooshort;
1703 #ifdef MRT6DEBUG
1704 if (mrt6debug & DEBUG_PIM)
1705 log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1706 #endif
1707 m_freem(m);
1708 return (IPPROTO_DONE);
1709 }
1710
1711 /*
1712 * if the packet is at least as big as a REGISTER, go ahead
1713 * and grab the PIM REGISTER header size, to avoid another
1714 * possible m_pullup() later.
1715 *
1716 * PIM_MINLEN == pimhdr + u_int32 == 8
1717 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1718 */
1719 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1720
1721 /*
1722 * Make sure that the IP6 and PIM headers in contiguous memory, and
1723 * possibly the PIM REGISTER header
1724 */
1725 #ifndef PULLDOWN_TEST
1726 IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1727 /* adjust pointer */
1728 ip6 = mtod(m, struct ip6_hdr *);
1729
1730 /* adjust mbuf to point to the PIM header */
1731 pim = (struct pim *)((caddr_t)ip6 + off);
1732 #else
1733 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1734 if (pim == NULL) {
1735 pim6stat.pim6s_rcv_tooshort++;
1736 return (IPPROTO_DONE);
1737 }
1738 #endif
1739
1740 #define PIM6_CHECKSUM
1741 #ifdef PIM6_CHECKSUM
1742 {
1743 int cksumlen;
1744
1745 /*
1746 * Validate checksum.
1747 * If PIM REGISTER, exclude the data packet
1748 */
1749 if (pim->pim_type == PIM_REGISTER)
1750 cksumlen = PIM_MINLEN;
1751 else
1752 cksumlen = pimlen;
1753
1754 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1755 ++pim6stat.pim6s_rcv_badsum;
1756 #ifdef MRT6DEBUG
1757 if (mrt6debug & DEBUG_PIM)
1758 log(LOG_DEBUG,
1759 "pim6_input: invalid checksum\n");
1760 #endif
1761 m_freem(m);
1762 return (IPPROTO_DONE);
1763 }
1764 }
1765 #endif /* PIM_CHECKSUM */
1766
1767 /* PIM version check */
1768 if (pim->pim_ver != PIM_VERSION) {
1769 ++pim6stat.pim6s_rcv_badversion;
1770 #ifdef MRT6DEBUG
1771 log(LOG_ERR,
1772 "pim6_input: incorrect version %d, expecting %d\n",
1773 pim->pim_ver, PIM_VERSION);
1774 #endif
1775 m_freem(m);
1776 return (IPPROTO_DONE);
1777 }
1778
1779 if (pim->pim_type == PIM_REGISTER) {
1780 /*
1781 * since this is a REGISTER, we'll make a copy of the register
1782 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1783 * routing daemon.
1784 */
1785 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1786
1787 struct mbuf *mcp;
1788 struct ip6_hdr *eip6;
1789 u_int32_t *reghdr;
1790 int rc;
1791
1792 ++pim6stat.pim6s_rcv_registers;
1793
1794 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1795 #ifdef MRT6DEBUG
1796 if (mrt6debug & DEBUG_PIM)
1797 log(LOG_DEBUG,
1798 "pim6_input: register mif not set: %d\n",
1799 reg_mif_num);
1800 #endif
1801 m_freem(m);
1802 return (IPPROTO_DONE);
1803 }
1804
1805 reghdr = (u_int32_t *)(pim + 1);
1806
1807 if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1808 goto pim6_input_to_daemon;
1809
1810 /*
1811 * Validate length
1812 */
1813 if (pimlen < PIM6_REG_MINLEN) {
1814 ++pim6stat.pim6s_rcv_tooshort;
1815 ++pim6stat.pim6s_rcv_badregisters;
1816 #ifdef MRT6DEBUG
1817 log(LOG_ERR,
1818 "pim6_input: register packet size too "
1819 "small %d from %s\n",
1820 pimlen, ip6_sprintf(&ip6->ip6_src));
1821 #endif
1822 m_freem(m);
1823 return (IPPROTO_DONE);
1824 }
1825
1826 eip6 = (struct ip6_hdr *) (reghdr + 1);
1827 #ifdef MRT6DEBUG
1828 if (mrt6debug & DEBUG_PIM)
1829 log(LOG_DEBUG,
1830 "pim6_input[register], eip6: %s -> %s, "
1831 "eip6 plen %d\n",
1832 ip6_sprintf(&eip6->ip6_src),
1833 ip6_sprintf(&eip6->ip6_dst),
1834 ntohs(eip6->ip6_plen));
1835 #endif
1836
1837 /* verify the version number of the inner packet */
1838 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1839 ++pim6stat.pim6s_rcv_badregisters;
1840 #ifdef MRT6DEBUG
1841 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1842 "of the inner packet\n",
1843 (eip6->ip6_vfc & IPV6_VERSION));
1844 #endif
1845 m_freem(m);
1846 return (IPPROTO_NONE);
1847 }
1848
1849 /* verify the inner packet is destined to a mcast group */
1850 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1851 ++pim6stat.pim6s_rcv_badregisters;
1852 #ifdef MRT6DEBUG
1853 if (mrt6debug & DEBUG_PIM)
1854 log(LOG_DEBUG,
1855 "pim6_input: inner packet of register "
1856 "is not multicast %s\n",
1857 ip6_sprintf(&eip6->ip6_dst));
1858 #endif
1859 m_freem(m);
1860 return (IPPROTO_DONE);
1861 }
1862
1863 /*
1864 * make a copy of the whole header to pass to the daemon later.
1865 */
1866 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1867 if (mcp == NULL) {
1868 #ifdef MRT6DEBUG
1869 log(LOG_ERR,
1870 "pim6_input: pim register: "
1871 "could not copy register head\n");
1872 #endif
1873 m_freem(m);
1874 return (IPPROTO_DONE);
1875 }
1876
1877 /*
1878 * forward the inner ip6 packet; point m_data at the inner ip6.
1879 */
1880 m_adj(m, off + PIM_MINLEN);
1881 #ifdef MRT6DEBUG
1882 if (mrt6debug & DEBUG_PIM) {
1883 log(LOG_DEBUG,
1884 "pim6_input: forwarding decapsulated register: "
1885 "src %s, dst %s, mif %d\n",
1886 ip6_sprintf(&eip6->ip6_src),
1887 ip6_sprintf(&eip6->ip6_dst),
1888 reg_mif_num);
1889 }
1890 #endif
1891
1892 rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1893 dst.sin6_family, 0);
1894
1895 /* prepare the register head to send to the mrouting daemon */
1896 m = mcp;
1897 }
1898
1899 /*
1900 * Pass the PIM message up to the daemon; if it is a register message
1901 * pass the 'head' only up to the daemon. This includes the
1902 * encapsulator ip6 header, pim header, register header and the
1903 * encapsulated ip6 header.
1904 */
1905 pim6_input_to_daemon:
1906 rip6_input(&m, offp, proto);
1907 return (IPPROTO_DONE);
1908 }
Cache object: 0b4673a35bae82ad099fdab0bd9edb77
|