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: releng/7.4/sys/netinet6/ip6_mroute.c 182632 2008-09-01 19:23:04Z obrien $");
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/sysctl.h>
101 #include <sys/syslog.h>
102 #include <sys/systm.h>
103 #include <sys/time.h>
104
105 #include <net/if.h>
106 #include <net/if_types.h>
107 #include <net/raw_cb.h>
108 #include <net/route.h>
109
110 #include <netinet/in.h>
111 #include <netinet/in_var.h>
112 #include <netinet/icmp6.h>
113
114 #include <netinet/ip6.h>
115 #include <netinet6/ip6_var.h>
116 #include <netinet6/scope6_var.h>
117 #include <netinet6/nd6.h>
118 #include <netinet6/ip6_mroute.h>
119 #include <netinet6/ip6protosw.h>
120 #include <netinet6/pim6.h>
121 #include <netinet6/pim6_var.h>
122
123 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
124
125 /* XXX: this is a very common idiom; move to <sys/mbuf.h> ? */
126 #define M_HASCL(m) ((m)->m_flags & M_EXT)
127
128 static int ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
129 static void phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
130
131 static int set_pim6(int *);
132 static int socket_send __P((struct socket *, struct mbuf *,
133 struct sockaddr_in6 *));
134 static int register_send __P((struct ip6_hdr *, struct mif6 *,
135 struct mbuf *));
136
137 extern struct domain inet6domain;
138
139 /* XXX: referenced from ip_mroute.c for dynamically loading this code. */
140 struct ip6protosw in6_pim_protosw = {
141 .pr_type = SOCK_RAW,
142 .pr_domain = &inet6domain,
143 .pr_protocol = IPPROTO_PIM,
144 .pr_flags = PR_ATOMIC|PR_ADDR|PR_LASTHDR,
145 .pr_input = pim6_input,
146 .pr_output = rip6_output,
147 .pr_ctloutput = rip6_ctloutput,
148 .pr_usrreqs = &rip6_usrreqs
149 };
150
151 static int ip6_mrouter_ver = 0;
152
153 SYSCTL_DECL(_net_inet6);
154 SYSCTL_DECL(_net_inet6_ip6);
155 SYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim, CTLFLAG_RW, 0, "PIM");
156
157 static struct mrt6stat mrt6stat;
158 SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
159 &mrt6stat, mrt6stat,
160 "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
161
162 #define NO_RTE_FOUND 0x1
163 #define RTE_FOUND 0x2
164
165 static struct mf6c *mf6ctable[MF6CTBLSIZ];
166 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
167 &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
168 "Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
169 "netinet6/ip6_mroute.h)");
170
171 static u_char n6expire[MF6CTBLSIZ];
172
173 static struct mif6 mif6table[MAXMIFS];
174 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mif6table, CTLFLAG_RD,
175 &mif6table, sizeof(mif6table), "S,vif[MAXMIFS]",
176 "Multicast Interfaces (struct mif[MAXMIFS], netinet6/ip6_mroute.h)");
177
178 #ifdef MRT6DEBUG
179 static u_int mrt6debug = 0; /* debug level */
180 #define DEBUG_MFC 0x02
181 #define DEBUG_FORWARD 0x04
182 #define DEBUG_EXPIRE 0x08
183 #define DEBUG_XMIT 0x10
184 #define DEBUG_REG 0x20
185 #define DEBUG_PIM 0x40
186 #endif
187
188 static void expire_upcalls(void *);
189 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */
190 #define UPCALL_EXPIRE 6 /* number of timeouts */
191
192 #ifdef INET
193 #ifdef MROUTING
194 extern struct socket *ip_mrouter;
195 #endif
196 #endif
197
198 /*
199 * 'Interfaces' associated with decapsulator (so we can tell
200 * packets that went through it from ones that get reflected
201 * by a broken gateway). Different from IPv4 register_if,
202 * these interfaces are linked into the system ifnet list,
203 * because per-interface IPv6 statistics are maintained in
204 * ifp->if_afdata. But it does not have any routes point
205 * to them. I.e., packets can't be sent this way. They
206 * only exist as a placeholder for multicast source
207 * verification.
208 */
209 static struct ifnet *multicast_register_if6;
210
211 #define ENCAP_HOPS 64
212
213 /*
214 * Private variables.
215 */
216 static mifi_t nummifs = 0;
217 static mifi_t reg_mif_num = (mifi_t)-1;
218
219 static struct pim6stat pim6stat;
220 SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RD,
221 &pim6stat, pim6stat,
222 "PIM Statistics (struct pim6stat, netinet6/pim_var.h)");
223
224 static int pim6;
225
226 /*
227 * Hash function for a source, group entry
228 */
229 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
230 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
231 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
232 (g).s6_addr32[2] ^ (g).s6_addr32[3])
233
234 /*
235 * Find a route for a given origin IPv6 address and Multicast group address.
236 */
237 #define MF6CFIND(o, g, rt) do { \
238 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
239 rt = NULL; \
240 mrt6stat.mrt6s_mfc_lookups++; \
241 while (_rt) { \
242 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
243 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
244 (_rt->mf6c_stall == NULL)) { \
245 rt = _rt; \
246 break; \
247 } \
248 _rt = _rt->mf6c_next; \
249 } \
250 if (rt == NULL) { \
251 mrt6stat.mrt6s_mfc_misses++; \
252 } \
253 } while (/*CONSTCOND*/ 0)
254
255 /*
256 * Macros to compute elapsed time efficiently
257 * Borrowed from Van Jacobson's scheduling code
258 * XXX: replace with timersub() ?
259 */
260 #define TV_DELTA(a, b, delta) do { \
261 int xxs; \
262 \
263 delta = (a).tv_usec - (b).tv_usec; \
264 if ((xxs = (a).tv_sec - (b).tv_sec)) { \
265 switch (xxs) { \
266 case 2: \
267 delta += 1000000; \
268 /* FALLTHROUGH */ \
269 case 1: \
270 delta += 1000000; \
271 break; \
272 default: \
273 delta += (1000000 * xxs); \
274 } \
275 } \
276 } while (/*CONSTCOND*/ 0)
277
278 /* XXX: replace with timercmp(a, b, <) ? */
279 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
280 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
281
282 #ifdef UPCALL_TIMING
283 #define UPCALL_MAX 50
284 static u_long upcall_data[UPCALL_MAX + 1];
285 static void collate();
286 #endif /* UPCALL_TIMING */
287
288 static int get_sg_cnt(struct sioc_sg_req6 *);
289 static int get_mif6_cnt(struct sioc_mif_req6 *);
290 static int ip6_mrouter_init(struct socket *, int, int);
291 static int add_m6if(struct mif6ctl *);
292 static int del_m6if(mifi_t *);
293 static int add_m6fc(struct mf6cctl *);
294 static int del_m6fc(struct mf6cctl *);
295
296 static struct callout expire_upcalls_ch;
297
298 int X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m);
299 int X_ip6_mrouter_done(void);
300 int X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt);
301 int X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt);
302 int X_mrt6_ioctl(int cmd, caddr_t data);
303
304 /*
305 * Handle MRT setsockopt commands to modify the multicast routing tables.
306 */
307 int
308 X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
309 {
310 int error = 0;
311 int optval;
312 struct mif6ctl mifc;
313 struct mf6cctl mfcc;
314 mifi_t mifi;
315
316 if (so != ip6_mrouter && sopt->sopt_name != MRT6_INIT)
317 return (EACCES);
318
319 switch (sopt->sopt_name) {
320 case MRT6_INIT:
321 #ifdef MRT6_OINIT
322 case MRT6_OINIT:
323 #endif
324 error = sooptcopyin(sopt, &optval, sizeof(optval),
325 sizeof(optval));
326 if (error)
327 break;
328 error = ip6_mrouter_init(so, optval, sopt->sopt_name);
329 break;
330 case MRT6_DONE:
331 error = X_ip6_mrouter_done();
332 break;
333 case MRT6_ADD_MIF:
334 error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
335 if (error)
336 break;
337 error = add_m6if(&mifc);
338 break;
339 case MRT6_ADD_MFC:
340 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
341 if (error)
342 break;
343 error = add_m6fc(&mfcc);
344 break;
345 case MRT6_DEL_MFC:
346 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
347 if (error)
348 break;
349 error = del_m6fc(&mfcc);
350 break;
351 case MRT6_DEL_MIF:
352 error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
353 if (error)
354 break;
355 error = del_m6if(&mifi);
356 break;
357 case MRT6_PIM:
358 error = sooptcopyin(sopt, &optval, sizeof(optval),
359 sizeof(optval));
360 if (error)
361 break;
362 error = set_pim6(&optval);
363 break;
364 default:
365 error = EOPNOTSUPP;
366 break;
367 }
368
369 return (error);
370 }
371
372 /*
373 * Handle MRT getsockopt commands
374 */
375 int
376 X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
377 {
378 int error = 0;
379
380 if (so != ip6_mrouter)
381 return (EACCES);
382
383 switch (sopt->sopt_name) {
384 case MRT6_PIM:
385 error = sooptcopyout(sopt, &pim6, sizeof(pim6));
386 break;
387 }
388 return (error);
389 }
390
391 /*
392 * Handle ioctl commands to obtain information from the cache
393 */
394 int
395 X_mrt6_ioctl(int cmd, caddr_t data)
396 {
397 switch (cmd) {
398 case SIOCGETSGCNT_IN6:
399 return (get_sg_cnt((struct sioc_sg_req6 *)data));
400 case SIOCGETMIFCNT_IN6:
401 return (get_mif6_cnt((struct sioc_mif_req6 *)data));
402 default:
403 return (EINVAL);
404 }
405 }
406
407 /*
408 * returns the packet, byte, rpf-failure count for the source group provided
409 */
410 static int
411 get_sg_cnt(struct sioc_sg_req6 *req)
412 {
413 struct mf6c *rt;
414 int s;
415
416 s = splnet();
417 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
418 splx(s);
419 if (rt != NULL) {
420 req->pktcnt = rt->mf6c_pkt_cnt;
421 req->bytecnt = rt->mf6c_byte_cnt;
422 req->wrong_if = rt->mf6c_wrong_if;
423 } else
424 return (ESRCH);
425 #if 0
426 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff;
427 #endif
428
429 return (0);
430 }
431
432 /*
433 * returns the input and output packet and byte counts on the mif provided
434 */
435 static int
436 get_mif6_cnt(struct sioc_mif_req6 *req)
437 {
438 mifi_t mifi = req->mifi;
439
440 if (mifi >= nummifs)
441 return (EINVAL);
442
443 req->icount = mif6table[mifi].m6_pkt_in;
444 req->ocount = mif6table[mifi].m6_pkt_out;
445 req->ibytes = mif6table[mifi].m6_bytes_in;
446 req->obytes = mif6table[mifi].m6_bytes_out;
447
448 return (0);
449 }
450
451 static int
452 set_pim6(int *i)
453 {
454 if ((*i != 1) && (*i != 0))
455 return (EINVAL);
456
457 pim6 = *i;
458
459 return (0);
460 }
461
462 /*
463 * Enable multicast routing
464 */
465 static int
466 ip6_mrouter_init(struct socket *so, int v, int cmd)
467 {
468 #ifdef MRT6DEBUG
469 if (mrt6debug)
470 log(LOG_DEBUG,
471 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
472 so->so_type, so->so_proto->pr_protocol);
473 #endif
474
475 if (so->so_type != SOCK_RAW ||
476 so->so_proto->pr_protocol != IPPROTO_ICMPV6)
477 return (EOPNOTSUPP);
478
479 if (v != 1)
480 return (ENOPROTOOPT);
481
482 if (ip6_mrouter != NULL)
483 return (EADDRINUSE);
484
485 ip6_mrouter = so;
486 ip6_mrouter_ver = cmd;
487
488 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
489 bzero((caddr_t)n6expire, sizeof(n6expire));
490
491 pim6 = 0;/* used for stubbing out/in pim stuff */
492
493 callout_init(&expire_upcalls_ch, 0);
494 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
495 expire_upcalls, NULL);
496
497 #ifdef MRT6DEBUG
498 if (mrt6debug)
499 log(LOG_DEBUG, "ip6_mrouter_init\n");
500 #endif
501
502 return (0);
503 }
504
505 /*
506 * Disable multicast routing
507 */
508 int
509 X_ip6_mrouter_done(void)
510 {
511 mifi_t mifi;
512 int i;
513 struct mf6c *rt;
514 struct rtdetq *rte;
515 int s;
516
517 s = splnet();
518
519 /*
520 * For each phyint in use, disable promiscuous reception of all IPv6
521 * multicasts.
522 */
523 #ifdef INET
524 #ifdef MROUTING
525 /*
526 * If there is still IPv4 multicast routing daemon,
527 * we remain interfaces to receive all muliticasted packets.
528 * XXX: there may be an interface in which the IPv4 multicast
529 * daemon is not interested...
530 */
531 if (!ip_mrouter)
532 #endif
533 #endif
534 {
535 for (mifi = 0; mifi < nummifs; mifi++) {
536 if (mif6table[mifi].m6_ifp &&
537 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
538 if_allmulti(mif6table[mifi].m6_ifp, 0);
539 }
540 }
541 }
542 bzero((caddr_t)mif6table, sizeof(mif6table));
543 nummifs = 0;
544
545 pim6 = 0; /* used to stub out/in pim specific code */
546
547 callout_stop(&expire_upcalls_ch);
548
549 /*
550 * Free all multicast forwarding cache entries.
551 */
552 for (i = 0; i < MF6CTBLSIZ; i++) {
553 rt = mf6ctable[i];
554 while (rt) {
555 struct mf6c *frt;
556
557 for (rte = rt->mf6c_stall; rte != NULL; ) {
558 struct rtdetq *n = rte->next;
559
560 m_free(rte->m);
561 free(rte, M_MRTABLE6);
562 rte = n;
563 }
564 frt = rt;
565 rt = rt->mf6c_next;
566 free(frt, M_MRTABLE6);
567 }
568 }
569
570 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
571
572 /*
573 * Reset register interface
574 */
575 if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
576 if_detach(multicast_register_if6);
577 if_free(multicast_register_if6);
578 reg_mif_num = (mifi_t)-1;
579 multicast_register_if6 = NULL;
580 }
581
582 ip6_mrouter = NULL;
583 ip6_mrouter_ver = 0;
584
585 splx(s);
586
587 #ifdef MRT6DEBUG
588 if (mrt6debug)
589 log(LOG_DEBUG, "ip6_mrouter_done\n");
590 #endif
591
592 return (0);
593 }
594
595 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
596
597 /*
598 * Add a mif to the mif table
599 */
600 static int
601 add_m6if(struct mif6ctl *mifcp)
602 {
603 struct mif6 *mifp;
604 struct ifnet *ifp;
605 int error, s;
606
607 if (mifcp->mif6c_mifi >= MAXMIFS)
608 return (EINVAL);
609 mifp = mif6table + mifcp->mif6c_mifi;
610 if (mifp->m6_ifp)
611 return (EADDRINUSE); /* XXX: is it appropriate? */
612 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > if_index)
613 return (ENXIO);
614 ifp = ifnet_byindex(mifcp->mif6c_pifi);
615
616 if (mifcp->mif6c_flags & MIFF_REGISTER) {
617 if (reg_mif_num == (mifi_t)-1) {
618 ifp = if_alloc(IFT_OTHER);
619
620 if_initname(ifp, "register_mif", 0);
621 ifp->if_flags |= IFF_LOOPBACK;
622 if_attach(ifp);
623 multicast_register_if6 = ifp;
624 reg_mif_num = mifcp->mif6c_mifi;
625 /*
626 * it is impossible to guess the ifindex of the
627 * register interface. So mif6c_pifi is automatically
628 * calculated.
629 */
630 mifcp->mif6c_pifi = ifp->if_index;
631 } else {
632 ifp = multicast_register_if6;
633 }
634
635 } /* if REGISTER */
636 else {
637 /* Make sure the interface supports multicast */
638 if ((ifp->if_flags & IFF_MULTICAST) == 0)
639 return (EOPNOTSUPP);
640
641 s = splnet();
642 error = if_allmulti(ifp, 1);
643 splx(s);
644 if (error)
645 return (error);
646 }
647
648 s = splnet();
649 mifp->m6_flags = mifcp->mif6c_flags;
650 mifp->m6_ifp = ifp;
651
652 /* initialize per mif pkt counters */
653 mifp->m6_pkt_in = 0;
654 mifp->m6_pkt_out = 0;
655 mifp->m6_bytes_in = 0;
656 mifp->m6_bytes_out = 0;
657 splx(s);
658
659 /* Adjust nummifs up if the mifi is higher than nummifs */
660 if (nummifs <= mifcp->mif6c_mifi)
661 nummifs = mifcp->mif6c_mifi + 1;
662
663 #ifdef MRT6DEBUG
664 if (mrt6debug)
665 log(LOG_DEBUG,
666 "add_mif #%d, phyint %s\n",
667 mifcp->mif6c_mifi,
668 ifp->if_xname);
669 #endif
670
671 return (0);
672 }
673
674 /*
675 * Delete a mif from the mif table
676 */
677 static int
678 del_m6if(mifi_t *mifip)
679 {
680 struct mif6 *mifp = mif6table + *mifip;
681 mifi_t mifi;
682 struct ifnet *ifp;
683 int s;
684
685 if (*mifip >= nummifs)
686 return (EINVAL);
687 if (mifp->m6_ifp == NULL)
688 return (EINVAL);
689
690 s = splnet();
691
692 if (!(mifp->m6_flags & MIFF_REGISTER)) {
693 /*
694 * XXX: what if there is yet IPv4 multicast daemon
695 * using the interface?
696 */
697 ifp = mifp->m6_ifp;
698
699 if_allmulti(ifp, 0);
700 } else {
701 if (reg_mif_num != (mifi_t)-1 &&
702 multicast_register_if6 != NULL) {
703 if_detach(multicast_register_if6);
704 if_free(multicast_register_if6);
705 reg_mif_num = (mifi_t)-1;
706 multicast_register_if6 = NULL;
707 }
708 }
709
710 bzero((caddr_t)mifp, sizeof(*mifp));
711
712 /* Adjust nummifs down */
713 for (mifi = nummifs; mifi > 0; mifi--)
714 if (mif6table[mifi - 1].m6_ifp)
715 break;
716 nummifs = mifi;
717
718 splx(s);
719
720 #ifdef MRT6DEBUG
721 if (mrt6debug)
722 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
723 #endif
724
725 return (0);
726 }
727
728 /*
729 * Add an mfc entry
730 */
731 static int
732 add_m6fc(struct mf6cctl *mfccp)
733 {
734 struct mf6c *rt;
735 u_long hash;
736 struct rtdetq *rte;
737 u_short nstl;
738 int s;
739 char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
740
741 MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
742 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
743
744 /* If an entry already exists, just update the fields */
745 if (rt) {
746 #ifdef MRT6DEBUG
747 if (mrt6debug & DEBUG_MFC) {
748 log(LOG_DEBUG,
749 "add_m6fc no upcall h %d o %s g %s p %x\n",
750 ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
751 ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
752 mfccp->mf6cc_parent);
753 }
754 #endif
755
756 s = splnet();
757 rt->mf6c_parent = mfccp->mf6cc_parent;
758 rt->mf6c_ifset = mfccp->mf6cc_ifset;
759 splx(s);
760 return (0);
761 }
762
763 /*
764 * Find the entry for which the upcall was made and update
765 */
766 s = splnet();
767 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
768 mfccp->mf6cc_mcastgrp.sin6_addr);
769 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
770 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
771 &mfccp->mf6cc_origin.sin6_addr) &&
772 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
773 &mfccp->mf6cc_mcastgrp.sin6_addr) &&
774 (rt->mf6c_stall != NULL)) {
775
776 if (nstl++)
777 log(LOG_ERR,
778 "add_m6fc: %s o %s g %s p %x dbx %p\n",
779 "multiple kernel entries",
780 ip6_sprintf(ip6bufo,
781 &mfccp->mf6cc_origin.sin6_addr),
782 ip6_sprintf(ip6bufg,
783 &mfccp->mf6cc_mcastgrp.sin6_addr),
784 mfccp->mf6cc_parent, rt->mf6c_stall);
785
786 #ifdef MRT6DEBUG
787 if (mrt6debug & DEBUG_MFC)
788 log(LOG_DEBUG,
789 "add_m6fc o %s g %s p %x dbg %x\n",
790 ip6_sprintf(ip6bufo,
791 &mfccp->mf6cc_origin.sin6_addr),
792 ip6_sprintf(ip6bufg,
793 &mfccp->mf6cc_mcastgrp.sin6_addr),
794 mfccp->mf6cc_parent, rt->mf6c_stall);
795 #endif
796
797 rt->mf6c_origin = mfccp->mf6cc_origin;
798 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
799 rt->mf6c_parent = mfccp->mf6cc_parent;
800 rt->mf6c_ifset = mfccp->mf6cc_ifset;
801 /* initialize pkt counters per src-grp */
802 rt->mf6c_pkt_cnt = 0;
803 rt->mf6c_byte_cnt = 0;
804 rt->mf6c_wrong_if = 0;
805
806 rt->mf6c_expire = 0; /* Don't clean this guy up */
807 n6expire[hash]--;
808
809 /* free packets Qed at the end of this entry */
810 for (rte = rt->mf6c_stall; rte != NULL; ) {
811 struct rtdetq *n = rte->next;
812 ip6_mdq(rte->m, rte->ifp, rt);
813 m_freem(rte->m);
814 #ifdef UPCALL_TIMING
815 collate(&(rte->t));
816 #endif /* UPCALL_TIMING */
817 free(rte, M_MRTABLE6);
818 rte = n;
819 }
820 rt->mf6c_stall = NULL;
821 }
822 }
823
824 /*
825 * It is possible that an entry is being inserted without an upcall
826 */
827 if (nstl == 0) {
828 #ifdef MRT6DEBUG
829 if (mrt6debug & DEBUG_MFC)
830 log(LOG_DEBUG,
831 "add_mfc no upcall h %d o %s g %s p %x\n",
832 hash,
833 ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
834 ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
835 mfccp->mf6cc_parent);
836 #endif
837
838 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
839
840 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
841 &mfccp->mf6cc_origin.sin6_addr)&&
842 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
843 &mfccp->mf6cc_mcastgrp.sin6_addr)) {
844
845 rt->mf6c_origin = mfccp->mf6cc_origin;
846 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
847 rt->mf6c_parent = mfccp->mf6cc_parent;
848 rt->mf6c_ifset = mfccp->mf6cc_ifset;
849 /* initialize pkt counters per src-grp */
850 rt->mf6c_pkt_cnt = 0;
851 rt->mf6c_byte_cnt = 0;
852 rt->mf6c_wrong_if = 0;
853
854 if (rt->mf6c_expire)
855 n6expire[hash]--;
856 rt->mf6c_expire = 0;
857 }
858 }
859 if (rt == NULL) {
860 /* no upcall, so make a new entry */
861 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
862 M_NOWAIT);
863 if (rt == NULL) {
864 splx(s);
865 return (ENOBUFS);
866 }
867
868 /* insert new entry at head of hash chain */
869 rt->mf6c_origin = mfccp->mf6cc_origin;
870 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
871 rt->mf6c_parent = mfccp->mf6cc_parent;
872 rt->mf6c_ifset = mfccp->mf6cc_ifset;
873 /* initialize pkt counters per src-grp */
874 rt->mf6c_pkt_cnt = 0;
875 rt->mf6c_byte_cnt = 0;
876 rt->mf6c_wrong_if = 0;
877 rt->mf6c_expire = 0;
878 rt->mf6c_stall = NULL;
879
880 /* link into table */
881 rt->mf6c_next = mf6ctable[hash];
882 mf6ctable[hash] = rt;
883 }
884 }
885 splx(s);
886 return (0);
887 }
888
889 #ifdef UPCALL_TIMING
890 /*
891 * collect delay statistics on the upcalls
892 */
893 static void
894 collate(struct timeval *t)
895 {
896 u_long d;
897 struct timeval tp;
898 u_long delta;
899
900 GET_TIME(tp);
901
902 if (TV_LT(*t, tp))
903 {
904 TV_DELTA(tp, *t, delta);
905
906 d = delta >> 10;
907 if (d > UPCALL_MAX)
908 d = UPCALL_MAX;
909
910 ++upcall_data[d];
911 }
912 }
913 #endif /* UPCALL_TIMING */
914
915 /*
916 * Delete an mfc entry
917 */
918 static int
919 del_m6fc(struct mf6cctl *mfccp)
920 {
921 struct sockaddr_in6 origin;
922 struct sockaddr_in6 mcastgrp;
923 struct mf6c *rt;
924 struct mf6c **nptr;
925 u_long hash;
926 int s;
927
928 origin = mfccp->mf6cc_origin;
929 mcastgrp = mfccp->mf6cc_mcastgrp;
930 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
931
932 #ifdef MRT6DEBUG
933 if (mrt6debug & DEBUG_MFC) {
934 char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
935 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
936 ip6_sprintf(ip6bufo, &origin.sin6_addr),
937 ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
938 }
939 #endif
940
941 s = splnet();
942
943 nptr = &mf6ctable[hash];
944 while ((rt = *nptr) != NULL) {
945 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
946 &rt->mf6c_origin.sin6_addr) &&
947 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
948 &rt->mf6c_mcastgrp.sin6_addr) &&
949 rt->mf6c_stall == NULL)
950 break;
951
952 nptr = &rt->mf6c_next;
953 }
954 if (rt == NULL) {
955 splx(s);
956 return (EADDRNOTAVAIL);
957 }
958
959 *nptr = rt->mf6c_next;
960 free(rt, M_MRTABLE6);
961
962 splx(s);
963
964 return (0);
965 }
966
967 static int
968 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
969 {
970
971 if (s) {
972 if (sbappendaddr(&s->so_rcv,
973 (struct sockaddr *)src,
974 mm, (struct mbuf *)0) != 0) {
975 sorwakeup(s);
976 return (0);
977 }
978 }
979 m_freem(mm);
980 return (-1);
981 }
982
983 /*
984 * IPv6 multicast forwarding function. This function assumes that the packet
985 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
986 * pointed to by "ifp", and the packet is to be relayed to other networks
987 * that have members of the packet's destination IPv6 multicast group.
988 *
989 * The packet is returned unscathed to the caller, unless it is
990 * erroneous, in which case a non-zero return value tells the caller to
991 * discard it.
992 *
993 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
994 * this function is called in the originating context (i.e., not when
995 * forwarding a packet from other node). ip6_output(), which is currently the
996 * only function that calls this function is called in the originating context,
997 * explicitly ensures this condition. It is caller's responsibility to ensure
998 * that if this function is called from somewhere else in the originating
999 * context in the future.
1000 */
1001 int
1002 X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1003 {
1004 struct mf6c *rt;
1005 struct mif6 *mifp;
1006 struct mbuf *mm;
1007 int s;
1008 mifi_t mifi;
1009 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1010
1011 #ifdef MRT6DEBUG
1012 if (mrt6debug & DEBUG_FORWARD)
1013 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1014 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1015 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1016 ifp->if_index);
1017 #endif
1018
1019 /*
1020 * Don't forward a packet with Hop limit of zero or one,
1021 * or a packet destined to a local-only group.
1022 */
1023 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1024 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1025 return (0);
1026 ip6->ip6_hlim--;
1027
1028 /*
1029 * Source address check: do not forward packets with unspecified
1030 * source. It was discussed in July 2000, on ipngwg mailing list.
1031 * This is rather more serious than unicast cases, because some
1032 * MLD packets can be sent with the unspecified source address
1033 * (although such packets must normally set 1 to the hop limit field).
1034 */
1035 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1036 ip6stat.ip6s_cantforward++;
1037 if (ip6_log_time + ip6_log_interval < time_second) {
1038 ip6_log_time = time_second;
1039 log(LOG_DEBUG,
1040 "cannot forward "
1041 "from %s to %s nxt %d received on %s\n",
1042 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1043 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1044 ip6->ip6_nxt,
1045 if_name(m->m_pkthdr.rcvif));
1046 }
1047 return (0);
1048 }
1049
1050 /*
1051 * Determine forwarding mifs from the forwarding cache table
1052 */
1053 s = splnet();
1054 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1055
1056 /* Entry exists, so forward if necessary */
1057 if (rt) {
1058 splx(s);
1059 return (ip6_mdq(m, ifp, rt));
1060 } else {
1061 /*
1062 * If we don't have a route for packet's origin,
1063 * Make a copy of the packet &
1064 * send message to routing daemon
1065 */
1066
1067 struct mbuf *mb0;
1068 struct rtdetq *rte;
1069 u_long hash;
1070 /* int i, npkts;*/
1071 #ifdef UPCALL_TIMING
1072 struct timeval tp;
1073
1074 GET_TIME(tp);
1075 #endif /* UPCALL_TIMING */
1076
1077 mrt6stat.mrt6s_no_route++;
1078 #ifdef MRT6DEBUG
1079 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1080 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1081 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1082 ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1083 #endif
1084
1085 /*
1086 * Allocate mbufs early so that we don't do extra work if we
1087 * are just going to fail anyway.
1088 */
1089 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6,
1090 M_NOWAIT);
1091 if (rte == NULL) {
1092 splx(s);
1093 return (ENOBUFS);
1094 }
1095 mb0 = m_copy(m, 0, M_COPYALL);
1096 /*
1097 * Pullup packet header if needed before storing it,
1098 * as other references may modify it in the meantime.
1099 */
1100 if (mb0 &&
1101 (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1102 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1103 if (mb0 == NULL) {
1104 free(rte, M_MRTABLE6);
1105 splx(s);
1106 return (ENOBUFS);
1107 }
1108
1109 /* is there an upcall waiting for this packet? */
1110 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1111 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1112 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1113 &rt->mf6c_origin.sin6_addr) &&
1114 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1115 &rt->mf6c_mcastgrp.sin6_addr) &&
1116 (rt->mf6c_stall != NULL))
1117 break;
1118 }
1119
1120 if (rt == NULL) {
1121 struct mrt6msg *im;
1122 #ifdef MRT6_OINIT
1123 struct omrt6msg *oim;
1124 #endif
1125
1126 /* no upcall, so make a new entry */
1127 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
1128 M_NOWAIT);
1129 if (rt == NULL) {
1130 free(rte, M_MRTABLE6);
1131 m_freem(mb0);
1132 splx(s);
1133 return (ENOBUFS);
1134 }
1135 /*
1136 * Make a copy of the header to send to the user
1137 * level process
1138 */
1139 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1140
1141 if (mm == NULL) {
1142 free(rte, M_MRTABLE6);
1143 m_freem(mb0);
1144 free(rt, M_MRTABLE6);
1145 splx(s);
1146 return (ENOBUFS);
1147 }
1148
1149 /*
1150 * Send message to routing daemon
1151 */
1152 sin6.sin6_addr = ip6->ip6_src;
1153
1154 im = NULL;
1155 #ifdef MRT6_OINIT
1156 oim = NULL;
1157 #endif
1158 switch (ip6_mrouter_ver) {
1159 #ifdef MRT6_OINIT
1160 case MRT6_OINIT:
1161 oim = mtod(mm, struct omrt6msg *);
1162 oim->im6_msgtype = MRT6MSG_NOCACHE;
1163 oim->im6_mbz = 0;
1164 break;
1165 #endif
1166 case MRT6_INIT:
1167 im = mtod(mm, struct mrt6msg *);
1168 im->im6_msgtype = MRT6MSG_NOCACHE;
1169 im->im6_mbz = 0;
1170 break;
1171 default:
1172 free(rte, M_MRTABLE6);
1173 m_freem(mb0);
1174 free(rt, M_MRTABLE6);
1175 splx(s);
1176 return (EINVAL);
1177 }
1178
1179 #ifdef MRT6DEBUG
1180 if (mrt6debug & DEBUG_FORWARD)
1181 log(LOG_DEBUG,
1182 "getting the iif info in the kernel\n");
1183 #endif
1184
1185 for (mifp = mif6table, mifi = 0;
1186 mifi < nummifs && mifp->m6_ifp != ifp;
1187 mifp++, mifi++)
1188 ;
1189
1190 switch (ip6_mrouter_ver) {
1191 #ifdef MRT6_OINIT
1192 case MRT6_OINIT:
1193 oim->im6_mif = mifi;
1194 break;
1195 #endif
1196 case MRT6_INIT:
1197 im->im6_mif = mifi;
1198 break;
1199 }
1200
1201 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1202 log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1203 "socket queue full\n");
1204 mrt6stat.mrt6s_upq_sockfull++;
1205 free(rte, M_MRTABLE6);
1206 m_freem(mb0);
1207 free(rt, M_MRTABLE6);
1208 splx(s);
1209 return (ENOBUFS);
1210 }
1211
1212 mrt6stat.mrt6s_upcalls++;
1213
1214 /* insert new entry at head of hash chain */
1215 bzero(rt, sizeof(*rt));
1216 rt->mf6c_origin.sin6_family = AF_INET6;
1217 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1218 rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1219 rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1220 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1221 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1222 rt->mf6c_expire = UPCALL_EXPIRE;
1223 n6expire[hash]++;
1224 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1225
1226 /* link into table */
1227 rt->mf6c_next = mf6ctable[hash];
1228 mf6ctable[hash] = rt;
1229 /* Add this entry to the end of the queue */
1230 rt->mf6c_stall = rte;
1231 } else {
1232 /* determine if q has overflowed */
1233 struct rtdetq **p;
1234 int npkts = 0;
1235
1236 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1237 if (++npkts > MAX_UPQ6) {
1238 mrt6stat.mrt6s_upq_ovflw++;
1239 free(rte, M_MRTABLE6);
1240 m_freem(mb0);
1241 splx(s);
1242 return (0);
1243 }
1244
1245 /* Add this entry to the end of the queue */
1246 *p = rte;
1247 }
1248
1249 rte->next = NULL;
1250 rte->m = mb0;
1251 rte->ifp = ifp;
1252 #ifdef UPCALL_TIMING
1253 rte->t = tp;
1254 #endif /* UPCALL_TIMING */
1255
1256 splx(s);
1257
1258 return (0);
1259 }
1260 }
1261
1262 /*
1263 * Clean up cache entries if upcalls are not serviced
1264 * Call from the Slow Timeout mechanism, every half second.
1265 */
1266 static void
1267 expire_upcalls(void *unused)
1268 {
1269 struct rtdetq *rte;
1270 struct mf6c *mfc, **nptr;
1271 int i;
1272 int s;
1273
1274 s = splnet();
1275 for (i = 0; i < MF6CTBLSIZ; i++) {
1276 if (n6expire[i] == 0)
1277 continue;
1278 nptr = &mf6ctable[i];
1279 while ((mfc = *nptr) != NULL) {
1280 rte = mfc->mf6c_stall;
1281 /*
1282 * Skip real cache entries
1283 * Make sure it wasn't marked to not expire (shouldn't happen)
1284 * If it expires now
1285 */
1286 if (rte != NULL &&
1287 mfc->mf6c_expire != 0 &&
1288 --mfc->mf6c_expire == 0) {
1289 #ifdef MRT6DEBUG
1290 if (mrt6debug & DEBUG_EXPIRE) {
1291 char ip6bufo[INET6_ADDRSTRLEN];
1292 char ip6bufg[INET6_ADDRSTRLEN];
1293 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1294 ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1295 ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1296 }
1297 #endif
1298 /*
1299 * drop all the packets
1300 * free the mbuf with the pkt, if, timing info
1301 */
1302 do {
1303 struct rtdetq *n = rte->next;
1304 m_freem(rte->m);
1305 free(rte, M_MRTABLE6);
1306 rte = n;
1307 } while (rte != NULL);
1308 mrt6stat.mrt6s_cache_cleanups++;
1309 n6expire[i]--;
1310
1311 *nptr = mfc->mf6c_next;
1312 free(mfc, M_MRTABLE6);
1313 } else {
1314 nptr = &mfc->mf6c_next;
1315 }
1316 }
1317 }
1318 splx(s);
1319 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1320 expire_upcalls, NULL);
1321 }
1322
1323 /*
1324 * Packet forwarding routine once entry in the cache is made
1325 */
1326 static int
1327 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1328 {
1329 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1330 mifi_t mifi, iif;
1331 struct mif6 *mifp;
1332 int plen = m->m_pkthdr.len;
1333 struct in6_addr src0, dst0; /* copies for local work */
1334 u_int32_t iszone, idzone, oszone, odzone;
1335 int error = 0;
1336
1337 /*
1338 * Macro to send packet on mif. Since RSVP packets don't get counted on
1339 * input, they shouldn't get counted on output, so statistics keeping is
1340 * separate.
1341 */
1342
1343 #define MC6_SEND(ip6, mifp, m) do { \
1344 if ((mifp)->m6_flags & MIFF_REGISTER) \
1345 register_send((ip6), (mifp), (m)); \
1346 else \
1347 phyint_send((ip6), (mifp), (m)); \
1348 } while (/*CONSTCOND*/ 0)
1349
1350 /*
1351 * Don't forward if it didn't arrive from the parent mif
1352 * for its origin.
1353 */
1354 mifi = rt->mf6c_parent;
1355 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1356 /* came in the wrong interface */
1357 #ifdef MRT6DEBUG
1358 if (mrt6debug & DEBUG_FORWARD)
1359 log(LOG_DEBUG,
1360 "wrong if: ifid %d mifi %d mififid %x\n",
1361 ifp->if_index, mifi,
1362 mif6table[mifi].m6_ifp->if_index);
1363 #endif
1364 mrt6stat.mrt6s_wrong_if++;
1365 rt->mf6c_wrong_if++;
1366 /*
1367 * If we are doing PIM processing, and we are forwarding
1368 * packets on this interface, send a message to the
1369 * routing daemon.
1370 */
1371 /* have to make sure this is a valid mif */
1372 if (mifi < nummifs && mif6table[mifi].m6_ifp)
1373 if (pim6 && (m->m_flags & M_LOOP) == 0) {
1374 /*
1375 * Check the M_LOOP flag to avoid an
1376 * unnecessary PIM assert.
1377 * XXX: M_LOOP is an ad-hoc hack...
1378 */
1379 static struct sockaddr_in6 sin6 =
1380 { sizeof(sin6), AF_INET6 };
1381
1382 struct mbuf *mm;
1383 struct mrt6msg *im;
1384 #ifdef MRT6_OINIT
1385 struct omrt6msg *oim;
1386 #endif
1387
1388 mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1389 if (mm &&
1390 (M_HASCL(mm) ||
1391 mm->m_len < sizeof(struct ip6_hdr)))
1392 mm = m_pullup(mm, sizeof(struct ip6_hdr));
1393 if (mm == NULL)
1394 return (ENOBUFS);
1395
1396 #ifdef MRT6_OINIT
1397 oim = NULL;
1398 #endif
1399 im = NULL;
1400 switch (ip6_mrouter_ver) {
1401 #ifdef MRT6_OINIT
1402 case MRT6_OINIT:
1403 oim = mtod(mm, struct omrt6msg *);
1404 oim->im6_msgtype = MRT6MSG_WRONGMIF;
1405 oim->im6_mbz = 0;
1406 break;
1407 #endif
1408 case MRT6_INIT:
1409 im = mtod(mm, struct mrt6msg *);
1410 im->im6_msgtype = MRT6MSG_WRONGMIF;
1411 im->im6_mbz = 0;
1412 break;
1413 default:
1414 m_freem(mm);
1415 return (EINVAL);
1416 }
1417
1418 for (mifp = mif6table, iif = 0;
1419 iif < nummifs && mifp &&
1420 mifp->m6_ifp != ifp;
1421 mifp++, iif++)
1422 ;
1423
1424 switch (ip6_mrouter_ver) {
1425 #ifdef MRT6_OINIT
1426 case MRT6_OINIT:
1427 oim->im6_mif = iif;
1428 sin6.sin6_addr = oim->im6_src;
1429 break;
1430 #endif
1431 case MRT6_INIT:
1432 im->im6_mif = iif;
1433 sin6.sin6_addr = im->im6_src;
1434 break;
1435 }
1436
1437 mrt6stat.mrt6s_upcalls++;
1438
1439 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1440 #ifdef MRT6DEBUG
1441 if (mrt6debug)
1442 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1443 #endif
1444 ++mrt6stat.mrt6s_upq_sockfull;
1445 return (ENOBUFS);
1446 } /* if socket Q full */
1447 } /* if PIM */
1448 return (0);
1449 } /* if wrong iif */
1450
1451 /* If I sourced this packet, it counts as output, else it was input. */
1452 if (m->m_pkthdr.rcvif == NULL) {
1453 /* XXX: is rcvif really NULL when output?? */
1454 mif6table[mifi].m6_pkt_out++;
1455 mif6table[mifi].m6_bytes_out += plen;
1456 } else {
1457 mif6table[mifi].m6_pkt_in++;
1458 mif6table[mifi].m6_bytes_in += plen;
1459 }
1460 rt->mf6c_pkt_cnt++;
1461 rt->mf6c_byte_cnt += plen;
1462
1463 /*
1464 * For each mif, forward a copy of the packet if there are group
1465 * members downstream on the interface.
1466 */
1467 src0 = ip6->ip6_src;
1468 dst0 = ip6->ip6_dst;
1469 if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1470 (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1471 ip6stat.ip6s_badscope++;
1472 return (error);
1473 }
1474 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1475 if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1476 /*
1477 * check if the outgoing packet is going to break
1478 * a scope boundary.
1479 * XXX For packets through PIM register tunnel
1480 * interface, we believe a routing daemon.
1481 */
1482 if (!(mif6table[rt->mf6c_parent].m6_flags &
1483 MIFF_REGISTER) &&
1484 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1485 if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1486 &oszone) ||
1487 in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1488 &odzone) ||
1489 iszone != oszone ||
1490 idzone != odzone) {
1491 ip6stat.ip6s_badscope++;
1492 continue;
1493 }
1494 }
1495
1496 mifp->m6_pkt_out++;
1497 mifp->m6_bytes_out += plen;
1498 MC6_SEND(ip6, mifp, m);
1499 }
1500 }
1501 return (0);
1502 }
1503
1504 static void
1505 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1506 {
1507 struct mbuf *mb_copy;
1508 struct ifnet *ifp = mifp->m6_ifp;
1509 int error = 0;
1510 int s = splnet(); /* needs to protect static "ro" below. */
1511 static struct route_in6 ro;
1512 struct in6_multi *in6m;
1513 struct sockaddr_in6 *dst6;
1514 u_long linkmtu;
1515
1516 /*
1517 * Make a new reference to the packet; make sure that
1518 * the IPv6 header is actually copied, not just referenced,
1519 * so that ip6_output() only scribbles on the copy.
1520 */
1521 mb_copy = m_copy(m, 0, M_COPYALL);
1522 if (mb_copy &&
1523 (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1524 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1525 if (mb_copy == NULL) {
1526 splx(s);
1527 return;
1528 }
1529 /* set MCAST flag to the outgoing packet */
1530 mb_copy->m_flags |= M_MCAST;
1531
1532 /*
1533 * If we sourced the packet, call ip6_output since we may devide
1534 * the packet into fragments when the packet is too big for the
1535 * outgoing interface.
1536 * Otherwise, we can simply send the packet to the interface
1537 * sending queue.
1538 */
1539 if (m->m_pkthdr.rcvif == NULL) {
1540 struct ip6_moptions im6o;
1541
1542 im6o.im6o_multicast_ifp = ifp;
1543 /* XXX: ip6_output will override ip6->ip6_hlim */
1544 im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1545 im6o.im6o_multicast_loop = 1;
1546 error = ip6_output(mb_copy, NULL, &ro,
1547 IPV6_FORWARDING, &im6o, NULL, NULL);
1548
1549 #ifdef MRT6DEBUG
1550 if (mrt6debug & DEBUG_XMIT)
1551 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1552 mifp - mif6table, error);
1553 #endif
1554 splx(s);
1555 return;
1556 }
1557
1558 /*
1559 * If we belong to the destination multicast group
1560 * on the outgoing interface, loop back a copy.
1561 */
1562 dst6 = (struct sockaddr_in6 *)&ro.ro_dst;
1563 IN6_LOOKUP_MULTI(ip6->ip6_dst, ifp, in6m);
1564 if (in6m != NULL) {
1565 dst6->sin6_len = sizeof(struct sockaddr_in6);
1566 dst6->sin6_family = AF_INET6;
1567 dst6->sin6_addr = ip6->ip6_dst;
1568 ip6_mloopback(ifp, m, (struct sockaddr_in6 *)&ro.ro_dst);
1569 }
1570 /*
1571 * Put the packet into the sending queue of the outgoing interface
1572 * if it would fit in the MTU of the interface.
1573 */
1574 linkmtu = IN6_LINKMTU(ifp);
1575 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1576 dst6->sin6_len = sizeof(struct sockaddr_in6);
1577 dst6->sin6_family = AF_INET6;
1578 dst6->sin6_addr = ip6->ip6_dst;
1579 /*
1580 * We just call if_output instead of nd6_output here, since
1581 * we need no ND for a multicast forwarded packet...right?
1582 */
1583 error = (*ifp->if_output)(ifp, mb_copy,
1584 (struct sockaddr *)&ro.ro_dst, NULL);
1585 #ifdef MRT6DEBUG
1586 if (mrt6debug & DEBUG_XMIT)
1587 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1588 mifp - mif6table, error);
1589 #endif
1590 } else {
1591 /*
1592 * pMTU discovery is intentionally disabled by default, since
1593 * various router may notify pMTU in multicast, which can be
1594 * a DDoS to a router
1595 */
1596 if (ip6_mcast_pmtu)
1597 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1598 else {
1599 #ifdef MRT6DEBUG
1600 if (mrt6debug & DEBUG_XMIT) {
1601 char ip6bufs[INET6_ADDRSTRLEN];
1602 char ip6bufd[INET6_ADDRSTRLEN];
1603 log(LOG_DEBUG,
1604 "phyint_send: packet too big on %s o %s "
1605 "g %s size %d(discarded)\n",
1606 if_name(ifp),
1607 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1608 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1609 mb_copy->m_pkthdr.len);
1610 }
1611 #endif /* MRT6DEBUG */
1612 m_freem(mb_copy); /* simply discard the packet */
1613 }
1614 }
1615
1616 splx(s);
1617 }
1618
1619 static int
1620 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1621 {
1622 struct mbuf *mm;
1623 int i, len = m->m_pkthdr.len;
1624 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1625 struct mrt6msg *im6;
1626
1627 #ifdef MRT6DEBUG
1628 if (mrt6debug) {
1629 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1630 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1631 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1632 ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1633 }
1634 #endif
1635 ++pim6stat.pim6s_snd_registers;
1636
1637 /* Make a copy of the packet to send to the user level process */
1638 MGETHDR(mm, M_DONTWAIT, MT_HEADER);
1639 if (mm == NULL)
1640 return (ENOBUFS);
1641 mm->m_pkthdr.rcvif = NULL;
1642 mm->m_data += max_linkhdr;
1643 mm->m_len = sizeof(struct ip6_hdr);
1644
1645 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1646 m_freem(mm);
1647 return (ENOBUFS);
1648 }
1649 i = MHLEN - M_LEADINGSPACE(mm);
1650 if (i > len)
1651 i = len;
1652 mm = m_pullup(mm, i);
1653 if (mm == NULL)
1654 return (ENOBUFS);
1655 /* TODO: check it! */
1656 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1657
1658 /*
1659 * Send message to routing daemon
1660 */
1661 sin6.sin6_addr = ip6->ip6_src;
1662
1663 im6 = mtod(mm, struct mrt6msg *);
1664 im6->im6_msgtype = MRT6MSG_WHOLEPKT;
1665 im6->im6_mbz = 0;
1666
1667 im6->im6_mif = mif - mif6table;
1668
1669 /* iif info is not given for reg. encap.n */
1670 mrt6stat.mrt6s_upcalls++;
1671
1672 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1673 #ifdef MRT6DEBUG
1674 if (mrt6debug)
1675 log(LOG_WARNING,
1676 "register_send: ip6_mrouter socket queue full\n");
1677 #endif
1678 ++mrt6stat.mrt6s_upq_sockfull;
1679 return (ENOBUFS);
1680 }
1681 return (0);
1682 }
1683
1684 /*
1685 * PIM sparse mode hook
1686 * Receives the pim control messages, and passes them up to the listening
1687 * socket, using rip6_input.
1688 * The only message processed is the REGISTER pim message; the pim header
1689 * is stripped off, and the inner packet is passed to register_mforward.
1690 */
1691 int
1692 pim6_input(struct mbuf **mp, int *offp, int proto)
1693 {
1694 struct pim *pim; /* pointer to a pim struct */
1695 struct ip6_hdr *ip6;
1696 int pimlen;
1697 struct mbuf *m = *mp;
1698 int minlen;
1699 int off = *offp;
1700
1701 ++pim6stat.pim6s_rcv_total;
1702
1703 ip6 = mtod(m, struct ip6_hdr *);
1704 pimlen = m->m_pkthdr.len - *offp;
1705
1706 /*
1707 * Validate lengths
1708 */
1709 if (pimlen < PIM_MINLEN) {
1710 ++pim6stat.pim6s_rcv_tooshort;
1711 #ifdef MRT6DEBUG
1712 if (mrt6debug & DEBUG_PIM)
1713 log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1714 #endif
1715 m_freem(m);
1716 return (IPPROTO_DONE);
1717 }
1718
1719 /*
1720 * if the packet is at least as big as a REGISTER, go ahead
1721 * and grab the PIM REGISTER header size, to avoid another
1722 * possible m_pullup() later.
1723 *
1724 * PIM_MINLEN == pimhdr + u_int32 == 8
1725 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1726 */
1727 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1728
1729 /*
1730 * Make sure that the IP6 and PIM headers in contiguous memory, and
1731 * possibly the PIM REGISTER header
1732 */
1733 #ifndef PULLDOWN_TEST
1734 IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1735 /* adjust pointer */
1736 ip6 = mtod(m, struct ip6_hdr *);
1737
1738 /* adjust mbuf to point to the PIM header */
1739 pim = (struct pim *)((caddr_t)ip6 + off);
1740 #else
1741 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1742 if (pim == NULL) {
1743 pim6stat.pim6s_rcv_tooshort++;
1744 return (IPPROTO_DONE);
1745 }
1746 #endif
1747
1748 #define PIM6_CHECKSUM
1749 #ifdef PIM6_CHECKSUM
1750 {
1751 int cksumlen;
1752
1753 /*
1754 * Validate checksum.
1755 * If PIM REGISTER, exclude the data packet
1756 */
1757 if (pim->pim_type == PIM_REGISTER)
1758 cksumlen = PIM_MINLEN;
1759 else
1760 cksumlen = pimlen;
1761
1762 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1763 ++pim6stat.pim6s_rcv_badsum;
1764 #ifdef MRT6DEBUG
1765 if (mrt6debug & DEBUG_PIM)
1766 log(LOG_DEBUG,
1767 "pim6_input: invalid checksum\n");
1768 #endif
1769 m_freem(m);
1770 return (IPPROTO_DONE);
1771 }
1772 }
1773 #endif /* PIM_CHECKSUM */
1774
1775 /* PIM version check */
1776 if (pim->pim_ver != PIM_VERSION) {
1777 ++pim6stat.pim6s_rcv_badversion;
1778 #ifdef MRT6DEBUG
1779 log(LOG_ERR,
1780 "pim6_input: incorrect version %d, expecting %d\n",
1781 pim->pim_ver, PIM_VERSION);
1782 #endif
1783 m_freem(m);
1784 return (IPPROTO_DONE);
1785 }
1786
1787 if (pim->pim_type == PIM_REGISTER) {
1788 /*
1789 * since this is a REGISTER, we'll make a copy of the register
1790 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1791 * routing daemon.
1792 */
1793 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1794
1795 struct mbuf *mcp;
1796 struct ip6_hdr *eip6;
1797 u_int32_t *reghdr;
1798 int rc;
1799 #ifdef MRT6DEBUG
1800 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1801 #endif
1802
1803 ++pim6stat.pim6s_rcv_registers;
1804
1805 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1806 #ifdef MRT6DEBUG
1807 if (mrt6debug & DEBUG_PIM)
1808 log(LOG_DEBUG,
1809 "pim6_input: register mif not set: %d\n",
1810 reg_mif_num);
1811 #endif
1812 m_freem(m);
1813 return (IPPROTO_DONE);
1814 }
1815
1816 reghdr = (u_int32_t *)(pim + 1);
1817
1818 if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1819 goto pim6_input_to_daemon;
1820
1821 /*
1822 * Validate length
1823 */
1824 if (pimlen < PIM6_REG_MINLEN) {
1825 ++pim6stat.pim6s_rcv_tooshort;
1826 ++pim6stat.pim6s_rcv_badregisters;
1827 #ifdef MRT6DEBUG
1828 log(LOG_ERR,
1829 "pim6_input: register packet size too "
1830 "small %d from %s\n",
1831 pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1832 #endif
1833 m_freem(m);
1834 return (IPPROTO_DONE);
1835 }
1836
1837 eip6 = (struct ip6_hdr *) (reghdr + 1);
1838 #ifdef MRT6DEBUG
1839 if (mrt6debug & DEBUG_PIM)
1840 log(LOG_DEBUG,
1841 "pim6_input[register], eip6: %s -> %s, "
1842 "eip6 plen %d\n",
1843 ip6_sprintf(ip6bufs, &eip6->ip6_src),
1844 ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1845 ntohs(eip6->ip6_plen));
1846 #endif
1847
1848 /* verify the version number of the inner packet */
1849 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1850 ++pim6stat.pim6s_rcv_badregisters;
1851 #ifdef MRT6DEBUG
1852 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1853 "of the inner packet\n",
1854 (eip6->ip6_vfc & IPV6_VERSION));
1855 #endif
1856 m_freem(m);
1857 return (IPPROTO_NONE);
1858 }
1859
1860 /* verify the inner packet is destined to a mcast group */
1861 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1862 ++pim6stat.pim6s_rcv_badregisters;
1863 #ifdef MRT6DEBUG
1864 if (mrt6debug & DEBUG_PIM)
1865 log(LOG_DEBUG,
1866 "pim6_input: inner packet of register "
1867 "is not multicast %s\n",
1868 ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1869 #endif
1870 m_freem(m);
1871 return (IPPROTO_DONE);
1872 }
1873
1874 /*
1875 * make a copy of the whole header to pass to the daemon later.
1876 */
1877 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1878 if (mcp == NULL) {
1879 #ifdef MRT6DEBUG
1880 log(LOG_ERR,
1881 "pim6_input: pim register: "
1882 "could not copy register head\n");
1883 #endif
1884 m_freem(m);
1885 return (IPPROTO_DONE);
1886 }
1887
1888 /*
1889 * forward the inner ip6 packet; point m_data at the inner ip6.
1890 */
1891 m_adj(m, off + PIM_MINLEN);
1892 #ifdef MRT6DEBUG
1893 if (mrt6debug & DEBUG_PIM) {
1894 log(LOG_DEBUG,
1895 "pim6_input: forwarding decapsulated register: "
1896 "src %s, dst %s, mif %d\n",
1897 ip6_sprintf(ip6bufs, &eip6->ip6_src),
1898 ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1899 reg_mif_num);
1900 }
1901 #endif
1902
1903 rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1904 dst.sin6_family, 0);
1905
1906 /* prepare the register head to send to the mrouting daemon */
1907 m = mcp;
1908 }
1909
1910 /*
1911 * Pass the PIM message up to the daemon; if it is a register message
1912 * pass the 'head' only up to the daemon. This includes the
1913 * encapsulator ip6 header, pim header, register header and the
1914 * encapsulated ip6 header.
1915 */
1916 pim6_input_to_daemon:
1917 rip6_input(&m, offp, proto);
1918 return (IPPROTO_DONE);
1919 }
Cache object: 2e715cb66ac1705608e0a2b8e2a5d750
|