kernel - Fix bugs in recent RSS/swap commits
[dragonfly.git] / sys / netinet6 / ip6_mroute.c
blob6d97d660cf552c32674bf9770414fe4baf362be0
1 /* $FreeBSD: src/sys/netinet6/ip6_mroute.c,v 1.2.2.9 2003/01/23 21:06:47 sam Exp $ */
2 /* $KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $ */
4 /*
5 * Copyright (C) 1998 WIDE Project.
6 * All rights reserved.
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.
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.
33 /* BSDI ip_mroute.c,v 2.10 1996/11/14 00:29:52 jch Exp */
36 * Copyright (c) 1989 Stephen Deering
37 * Copyright (c) 1992, 1993
38 * The Regents of the University of California. All rights reserved.
40 * This code is derived from software contributed to Berkeley by
41 * Stephen Deering of Stanford University.
43 * Redistribution and use in source and binary forms, with or without
44 * modification, are permitted provided that the following conditions
45 * are met:
46 * 1. Redistributions of source code must retain the above copyright
47 * notice, this list of conditions and the following disclaimer.
48 * 2. Redistributions in binary form must reproduce the above copyright
49 * notice, this list of conditions and the following disclaimer in the
50 * documentation and/or other materials provided with the distribution.
51 * 3. 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.
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.
67 * @(#)ip_mroute.c 8.2 (Berkeley) 11/15/93
71 * IP multicast forwarding procedures
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
80 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
83 #include "opt_inet.h"
84 #include "opt_inet6.h"
86 #include <sys/param.h>
87 #include <sys/systm.h>
88 #include <sys/callout.h>
89 #include <sys/malloc.h>
90 #include <sys/mbuf.h>
91 #include <sys/socket.h>
92 #include <sys/socketvar.h>
93 #include <sys/sockio.h>
94 #include <sys/protosw.h>
95 #include <sys/errno.h>
96 #include <sys/time.h>
97 #include <sys/kernel.h>
98 #include <sys/syslog.h>
99 #include <sys/thread2.h>
101 #include <net/if.h>
102 #include <net/route.h>
103 #include <net/raw_cb.h>
104 #include <net/netisr2.h>
105 #include <net/netmsg2.h>
107 #include <netinet/in.h>
108 #include <netinet/in_var.h>
110 #include <netinet/ip6.h>
111 #include <netinet6/ip6_var.h>
112 #include <netinet6/ip6_mroute.h>
113 #include <netinet6/nd6.h>
114 #include <netinet6/pim6.h>
115 #include <netinet6/pim6_var.h>
117 #include <net/net_osdep.h>
119 static MALLOC_DEFINE(M_MRTABLE, "mf6c", "multicast forwarding cache entry");
121 #define M_HASCL(m) ((m)->m_flags & M_EXT)
123 static int ip6_mdq (struct mbuf *, struct ifnet *, struct mf6c *);
124 static void phyint_send (struct ip6_hdr *, struct mif6 *, struct mbuf *);
126 static int set_pim6 (int *);
127 static int socket_send(struct socket *, struct mbuf *, struct sockaddr_in6 *);
128 static int register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
131 * Globals. All but ip6_mrouter, ip6_mrtproto and mrt6stat could be static,
132 * except for netstat or debugging purposes.
134 struct socket *ip6_mrouter = NULL;
135 int ip6_mrouter_ver = 0;
136 int ip6_mrtproto = IPPROTO_PIM; /* for netstat only */
137 struct mrt6stat mrt6stat;
139 #define NO_RTE_FOUND 0x1
140 #define RTE_FOUND 0x2
142 struct mf6c *mf6ctable[MF6CTBLSIZ];
143 u_char n6expire[MF6CTBLSIZ];
144 static struct mif6 mif6table[MAXMIFS];
145 #ifdef MRT6DEBUG
146 u_int mrt6debug = 0; /* debug level */
147 #define DEBUG_MFC 0x02
148 #define DEBUG_FORWARD 0x04
149 #define DEBUG_EXPIRE 0x08
150 #define DEBUG_XMIT 0x10
151 #define DEBUG_REG 0x20
152 #define DEBUG_PIM 0x40
153 #endif
155 static void expire_upcalls (void *);
156 static void expire_upcalls_dispatch(netmsg_t);
157 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */
158 #define UPCALL_EXPIRE 6 /* number of timeouts */
160 #ifdef INET
161 #ifdef MROUTING
162 extern struct socket *ip_mrouter;
163 #endif
164 #endif
167 * 'Interfaces' associated with decapsulator (so we can tell
168 * packets that went through it from ones that get reflected
169 * by a broken gateway). These interfaces are never linked into
170 * the system ifnet list & no routes point to them. I.e., packets
171 * can't be sent this way. They only exist as a placeholder for
172 * multicast source verification.
174 struct ifnet multicast_register_if;
176 #define ENCAP_HOPS 64
179 * Private variables.
181 static mifi_t nummifs = 0;
182 static mifi_t reg_mif_num = (mifi_t)-1;
184 static struct pim6stat pim6stat;
185 static int pim6;
188 * Hash function for a source, group entry
190 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
191 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
192 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
193 (g).s6_addr32[2] ^ (g).s6_addr32[3])
196 * Find a route for a given origin IPv6 address and Multicast group address.
197 * Quality of service parameter to be added in the future!!!
200 #define MF6CFIND(o, g, rt) do { \
201 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
202 rt = NULL; \
203 mrt6stat.mrt6s_mfc_lookups++; \
204 while (_rt) { \
205 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
206 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
207 (_rt->mf6c_stall == NULL)) { \
208 rt = _rt; \
209 break; \
211 _rt = _rt->mf6c_next; \
213 if (rt == NULL) { \
214 mrt6stat.mrt6s_mfc_misses++; \
216 } while (0)
219 * Macros to compute elapsed time efficiently
220 * Borrowed from Van Jacobson's scheduling code
222 #define TV_DELTA(a, b, delta) do { \
223 int xxs; \
225 delta = (a).tv_usec - (b).tv_usec; \
226 if ((xxs = (a).tv_sec - (b).tv_sec)) { \
227 switch (xxs) { \
228 case 2: \
229 delta += 1000000; \
230 /* FALLTHROUGH */ \
231 case 1: \
232 delta += 1000000; \
233 break; \
234 default: \
235 delta += (1000000 * xxs); \
238 } while (0)
240 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
241 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
243 #ifdef UPCALL_TIMING
244 #define UPCALL_MAX 50
245 u_long upcall_data[UPCALL_MAX + 1];
246 static void collate();
247 #endif /* UPCALL_TIMING */
249 static int get_sg_cnt (struct sioc_sg_req6 *);
250 static int get_mif6_cnt (struct sioc_mif_req6 *);
251 static int ip6_mrouter_init (struct socket *, struct mbuf *, int);
252 static int add_m6if (struct mif6ctl *);
253 static int del_m6if (mifi_t *);
254 static int add_m6fc (struct mf6cctl *);
255 static int del_m6fc (struct mf6cctl *);
257 static struct callout expire_upcalls_ch;
258 static struct netmsg_base expire_upcalls_nmsg;
261 * Handle MRT setsockopt commands to modify the multicast routing tables.
264 ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
266 int error = 0;
267 struct mbuf *m;
269 if (so != ip6_mrouter && sopt->sopt_name != MRT6_INIT)
270 return (EACCES);
272 if ((error = soopt_getm(sopt, &m)) != 0) /* XXX */
273 return (error);
274 soopt_to_mbuf(sopt, m); /* XXX */
276 switch (sopt->sopt_name) {
277 case MRT6_INIT:
278 #ifdef MRT6_OINIT
279 case MRT6_OINIT:
280 #endif
281 error = ip6_mrouter_init(so, m, sopt->sopt_name);
282 break;
283 case MRT6_DONE:
284 error = ip6_mrouter_done();
285 break;
286 case MRT6_ADD_MIF:
287 error = add_m6if(mtod(m, struct mif6ctl *));
288 break;
289 case MRT6_DEL_MIF:
290 error = del_m6if(mtod(m, mifi_t *));
291 break;
292 case MRT6_ADD_MFC:
293 error = add_m6fc(mtod(m, struct mf6cctl *));
294 break;
295 case MRT6_DEL_MFC:
296 error = del_m6fc(mtod(m, struct mf6cctl *));
297 break;
298 case MRT6_PIM:
299 error = set_pim6(mtod(m, int *));
300 break;
301 default:
302 error = EOPNOTSUPP;
303 break;
306 m_freem(m);
307 return (error);
311 * Handle MRT getsockopt commands
314 ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
316 int error = 0;
318 if (so != ip6_mrouter) return EACCES;
320 switch (sopt->sopt_name) {
321 case MRT6_PIM:
322 soopt_from_kbuf(sopt, &pim6, sizeof(pim6));
323 break;
325 return (error);
329 * Handle ioctl commands to obtain information from the cache
332 mrt6_ioctl(int cmd, caddr_t data)
334 int error = 0;
336 switch (cmd) {
337 case SIOCGETSGCNT_IN6:
338 return (get_sg_cnt((struct sioc_sg_req6 *)data));
339 break; /* for safety */
340 case SIOCGETMIFCNT_IN6:
341 return (get_mif6_cnt((struct sioc_mif_req6 *)data));
342 break; /* for safety */
343 default:
344 return (EINVAL);
345 break;
347 return error;
351 * returns the packet, byte, rpf-failure count for the source group provided
353 static int
354 get_sg_cnt(struct sioc_sg_req6 *req)
356 struct mf6c *rt;
358 crit_enter();
359 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
360 crit_exit();
361 if (rt != NULL) {
362 req->pktcnt = rt->mf6c_pkt_cnt;
363 req->bytecnt = rt->mf6c_byte_cnt;
364 req->wrong_if = rt->mf6c_wrong_if;
365 } else
366 return (ESRCH);
367 #if 0
368 req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff;
369 #endif
371 return 0;
375 * returns the input and output packet and byte counts on the mif provided
377 static int
378 get_mif6_cnt(struct sioc_mif_req6 *req)
380 mifi_t mifi = req->mifi;
382 if (mifi >= nummifs)
383 return EINVAL;
385 req->icount = mif6table[mifi].m6_pkt_in;
386 req->ocount = mif6table[mifi].m6_pkt_out;
387 req->ibytes = mif6table[mifi].m6_bytes_in;
388 req->obytes = mif6table[mifi].m6_bytes_out;
390 return 0;
393 static int
394 set_pim6(int *i)
396 if ((*i != 1) && (*i != 0))
397 return EINVAL;
399 pim6 = *i;
401 return 0;
405 * Enable multicast routing
407 static int
408 ip6_mrouter_init(struct socket *so, struct mbuf *m, int cmd)
410 int *v;
412 ASSERT_IN_NETISR(0);
414 #ifdef MRT6DEBUG
415 if (mrt6debug)
416 log(LOG_DEBUG,
417 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
418 so->so_type, so->so_proto->pr_protocol);
419 #endif
421 if (so->so_type != SOCK_RAW ||
422 so->so_proto->pr_protocol != IPPROTO_ICMPV6)
423 return EOPNOTSUPP;
425 if (!m || (m->m_len != sizeof(int *)))
426 return ENOPROTOOPT;
428 v = mtod(m, int *);
429 if (*v != 1)
430 return ENOPROTOOPT;
432 if (ip6_mrouter != NULL)
433 return EADDRINUSE;
435 ip6_mrouter = so;
436 ip6_mrouter_ver = cmd;
438 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
439 bzero((caddr_t)n6expire, sizeof(n6expire));
441 pim6 = 0;/* used for stubbing out/in pim stuff */
443 callout_init_mp(&expire_upcalls_ch);
444 netmsg_init(&expire_upcalls_nmsg, NULL, &netisr_adone_rport,
445 MSGF_PRIORITY | MSGF_DROPABLE, expire_upcalls_dispatch);
447 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
448 expire_upcalls, NULL);
450 #ifdef MRT6DEBUG
451 if (mrt6debug)
452 log(LOG_DEBUG, "ip6_mrouter_init\n");
453 #endif
455 return 0;
459 * Disable multicast routing
462 ip6_mrouter_done(void)
464 mifi_t mifi;
465 int i;
466 struct ifnet *ifp;
467 struct in6_ifreq ifr;
468 struct mf6c *rt;
469 struct rtdetq *rte;
470 struct lwkt_msg *lmsg = &expire_upcalls_nmsg.lmsg;
472 ASSERT_IN_NETISR(0);
474 if (ip6_mrouter == NULL)
475 return EINVAL;
478 * For each phyint in use, disable promiscuous reception of all IPv6
479 * multicasts.
481 #ifdef INET
482 #ifdef MROUTING
484 * If there is still IPv4 multicast routing daemon,
485 * we remain interfaces to receive all muliticasted packets.
486 * XXX: there may be an interface in which the IPv4 multicast
487 * daemon is not interested...
489 if (!ip_mrouter)
490 #endif
491 #endif
493 for (mifi = 0; mifi < nummifs; mifi++) {
494 if (mif6table[mifi].m6_ifp &&
495 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
496 ifr.ifr_addr.sin6_family = AF_INET6;
497 ifr.ifr_addr.sin6_addr = kin6addr_any;
498 ifp = mif6table[mifi].m6_ifp;
499 ifnet_serialize_all(ifp);
500 ifp->if_ioctl(ifp, SIOCDELMULTI,
501 (caddr_t)&ifr, NULL);
502 ifnet_deserialize_all(ifp);
506 #ifdef notyet
507 bzero((caddr_t)qtable, sizeof(qtable));
508 bzero((caddr_t)tbftable, sizeof(tbftable));
509 #endif
510 bzero((caddr_t)mif6table, sizeof(mif6table));
511 nummifs = 0;
513 pim6 = 0; /* used to stub out/in pim specific code */
515 callout_stop(&expire_upcalls_ch);
516 crit_enter();
517 if ((lmsg->ms_flags & MSGF_DONE) == 0)
518 lwkt_dropmsg(lmsg);
519 crit_exit();
522 * Free all multicast forwarding cache entries.
524 for (i = 0; i < MF6CTBLSIZ; i++) {
525 rt = mf6ctable[i];
526 while (rt) {
527 struct mf6c *frt;
529 for (rte = rt->mf6c_stall; rte != NULL; ) {
530 struct rtdetq *n = rte->next;
532 m_freem(rte->m);
533 kfree(rte, M_MRTABLE);
534 rte = n;
536 frt = rt;
537 rt = rt->mf6c_next;
538 kfree(frt, M_MRTABLE);
542 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
545 * Reset de-encapsulation cache
547 reg_mif_num = -1;
549 ip6_mrouter = NULL;
550 ip6_mrouter_ver = 0;
552 #ifdef MRT6DEBUG
553 if (mrt6debug)
554 log(LOG_DEBUG, "ip6_mrouter_done\n");
555 #endif
557 return 0;
560 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
563 * Add a mif to the mif table
565 static int
566 add_m6if(struct mif6ctl *mifcp)
568 struct mif6 *mifp;
569 struct ifnet *ifp;
570 int error;
571 #ifdef notyet
572 struct tbf *m_tbf = tbftable + mifcp->mif6c_mifi;
573 #endif
575 if (mifcp->mif6c_mifi >= MAXMIFS)
576 return EINVAL;
577 mifp = mif6table + mifcp->mif6c_mifi;
578 if (mifp->m6_ifp)
579 return EADDRINUSE; /* XXX: is it appropriate? */
580 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > if_index)
581 return ENXIO;
582 ifp = ifindex2ifnet[mifcp->mif6c_pifi];
584 if (mifcp->mif6c_flags & MIFF_REGISTER) {
585 if (reg_mif_num == (mifi_t)-1) {
586 strlcpy(multicast_register_if.if_xname, "register_mif",
587 IFNAMSIZ);
588 multicast_register_if.if_flags |= IFF_LOOPBACK;
589 multicast_register_if.if_index = mifcp->mif6c_mifi;
590 reg_mif_num = mifcp->mif6c_mifi;
593 ifp = &multicast_register_if;
595 } /* if REGISTER */
596 else {
597 /* Make sure the interface supports multicast */
598 if ((ifp->if_flags & IFF_MULTICAST) == 0)
599 return EOPNOTSUPP;
601 crit_enter();
602 error = if_allmulti(ifp, 1);
603 crit_exit();
604 if (error)
605 return error;
608 crit_enter();
609 mifp->m6_flags = mifcp->mif6c_flags;
610 mifp->m6_ifp = ifp;
611 #ifdef notyet
612 /* scaling up here allows division by 1024 in critical code */
613 mifp->m6_rate_limit = mifcp->mif6c_rate_limit * 1024 / 1000;
614 #endif
615 /* initialize per mif pkt counters */
616 mifp->m6_pkt_in = 0;
617 mifp->m6_pkt_out = 0;
618 mifp->m6_bytes_in = 0;
619 mifp->m6_bytes_out = 0;
620 crit_exit();
622 /* Adjust nummifs up if the mifi is higher than nummifs */
623 if (nummifs <= mifcp->mif6c_mifi)
624 nummifs = mifcp->mif6c_mifi + 1;
626 #ifdef MRT6DEBUG
627 if (mrt6debug)
628 log(LOG_DEBUG,
629 "add_mif #%d, phyint %s\n",
630 mifcp->mif6c_mifi,
631 ifp->if_xname);
632 #endif
634 return 0;
638 * Delete a mif from the mif table
640 static int
641 del_m6if(mifi_t *mifip)
643 struct mif6 *mifp = mif6table + *mifip;
644 mifi_t mifi;
645 struct ifnet *ifp;
647 if (*mifip >= nummifs)
648 return EINVAL;
649 if (mifp->m6_ifp == NULL)
650 return EINVAL;
652 crit_enter();
654 if (!(mifp->m6_flags & MIFF_REGISTER)) {
656 * XXX: what if there is yet IPv4 multicast daemon
657 * using the interface?
659 ifp = mifp->m6_ifp;
661 if_allmulti(ifp, 0);
664 #ifdef notyet
665 bzero((caddr_t)qtable[*mifip], sizeof(qtable[*mifip]));
666 bzero((caddr_t)mifp->m6_tbf, sizeof(*(mifp->m6_tbf)));
667 #endif
668 bzero((caddr_t)mifp, sizeof (*mifp));
670 /* Adjust nummifs down */
671 for (mifi = nummifs; mifi > 0; mifi--)
672 if (mif6table[mifi - 1].m6_ifp)
673 break;
674 nummifs = mifi;
676 crit_exit();
678 #ifdef MRT6DEBUG
679 if (mrt6debug)
680 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
681 #endif
683 return 0;
687 * Add an mfc entry
689 static int
690 add_m6fc(struct mf6cctl *mfccp)
692 struct mf6c *rt;
693 u_long hash;
694 struct rtdetq *rte;
695 u_short nstl;
697 MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
698 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
700 /* If an entry already exists, just update the fields */
701 if (rt) {
702 #ifdef MRT6DEBUG
703 if (mrt6debug & DEBUG_MFC)
704 log(LOG_DEBUG,
705 "add_m6fc no upcall h %d o %s g %s p %x\n",
706 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
707 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
708 mfccp->mf6cc_parent);
709 #endif
711 crit_enter();
712 rt->mf6c_parent = mfccp->mf6cc_parent;
713 rt->mf6c_ifset = mfccp->mf6cc_ifset;
714 crit_exit();
715 return 0;
719 * Find the entry for which the upcall was made and update
721 crit_enter();
722 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
723 mfccp->mf6cc_mcastgrp.sin6_addr);
724 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
725 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
726 &mfccp->mf6cc_origin.sin6_addr) &&
727 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
728 &mfccp->mf6cc_mcastgrp.sin6_addr) &&
729 (rt->mf6c_stall != NULL)) {
731 if (nstl++)
732 log(LOG_ERR,
733 "add_m6fc: %s o %s g %s p %x dbx %p\n",
734 "multiple kernel entries",
735 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
736 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
737 mfccp->mf6cc_parent, rt->mf6c_stall);
739 #ifdef MRT6DEBUG
740 if (mrt6debug & DEBUG_MFC)
741 log(LOG_DEBUG,
742 "add_m6fc o %s g %s p %x dbg %x\n",
743 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
744 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
745 mfccp->mf6cc_parent, rt->mf6c_stall);
746 #endif
748 rt->mf6c_origin = mfccp->mf6cc_origin;
749 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
750 rt->mf6c_parent = mfccp->mf6cc_parent;
751 rt->mf6c_ifset = mfccp->mf6cc_ifset;
752 /* initialize pkt counters per src-grp */
753 rt->mf6c_pkt_cnt = 0;
754 rt->mf6c_byte_cnt = 0;
755 rt->mf6c_wrong_if = 0;
757 rt->mf6c_expire = 0; /* Don't clean this guy up */
758 n6expire[hash]--;
760 /* free packets Qed at the end of this entry */
761 for (rte = rt->mf6c_stall; rte != NULL; ) {
762 struct rtdetq *n = rte->next;
763 ip6_mdq(rte->m, rte->ifp, rt);
764 m_freem(rte->m);
765 #ifdef UPCALL_TIMING
766 collate(&(rte->t));
767 #endif /* UPCALL_TIMING */
768 kfree(rte, M_MRTABLE);
769 rte = n;
771 rt->mf6c_stall = NULL;
776 * It is possible that an entry is being inserted without an upcall
778 if (nstl == 0) {
779 #ifdef MRT6DEBUG
780 if (mrt6debug & DEBUG_MFC)
781 log(LOG_DEBUG,"add_mfc no upcall h %d o %s g %s p %x\n",
782 hash,
783 ip6_sprintf(&mfccp->mf6cc_origin.sin6_addr),
784 ip6_sprintf(&mfccp->mf6cc_mcastgrp.sin6_addr),
785 mfccp->mf6cc_parent);
786 #endif
788 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
789 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
790 &mfccp->mf6cc_origin.sin6_addr)&&
791 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
792 &mfccp->mf6cc_mcastgrp.sin6_addr)) {
794 rt->mf6c_origin = mfccp->mf6cc_origin;
795 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
796 rt->mf6c_parent = mfccp->mf6cc_parent;
797 rt->mf6c_ifset = mfccp->mf6cc_ifset;
798 /* initialize pkt counters per src-grp */
799 rt->mf6c_pkt_cnt = 0;
800 rt->mf6c_byte_cnt = 0;
801 rt->mf6c_wrong_if = 0;
803 if (rt->mf6c_expire)
804 n6expire[hash]--;
805 rt->mf6c_expire = 0;
808 if (rt == NULL) {
809 /* no upcall, so make a new entry */
810 rt = (struct mf6c *)kmalloc(sizeof(*rt), M_MRTABLE,
811 M_NOWAIT);
812 if (rt == NULL) {
813 crit_exit();
814 return ENOBUFS;
817 /* insert new entry at head of hash chain */
818 rt->mf6c_origin = mfccp->mf6cc_origin;
819 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
820 rt->mf6c_parent = mfccp->mf6cc_parent;
821 rt->mf6c_ifset = mfccp->mf6cc_ifset;
822 /* initialize pkt counters per src-grp */
823 rt->mf6c_pkt_cnt = 0;
824 rt->mf6c_byte_cnt = 0;
825 rt->mf6c_wrong_if = 0;
826 rt->mf6c_expire = 0;
827 rt->mf6c_stall = NULL;
829 /* link into table */
830 rt->mf6c_next = mf6ctable[hash];
831 mf6ctable[hash] = rt;
834 crit_exit();
835 return 0;
838 #ifdef UPCALL_TIMING
840 * collect delay statistics on the upcalls
842 static void
843 collate(struct timeval *t)
845 u_long d;
846 struct timeval tp;
847 u_long delta;
849 GET_TIME(tp);
851 if (TV_LT(*t, tp))
853 TV_DELTA(tp, *t, delta);
855 d = delta >> 10;
856 if (d > UPCALL_MAX)
857 d = UPCALL_MAX;
859 ++upcall_data[d];
862 #endif /* UPCALL_TIMING */
865 * Delete an mfc entry
867 static int
868 del_m6fc(struct mf6cctl *mfccp)
870 struct sockaddr_in6 origin;
871 struct sockaddr_in6 mcastgrp;
872 struct mf6c *rt;
873 struct mf6c **nptr;
874 u_long hash;
876 origin = mfccp->mf6cc_origin;
877 mcastgrp = mfccp->mf6cc_mcastgrp;
878 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
880 #ifdef MRT6DEBUG
881 if (mrt6debug & DEBUG_MFC)
882 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
883 ip6_sprintf(&origin.sin6_addr),
884 ip6_sprintf(&mcastgrp.sin6_addr));
885 #endif
887 crit_enter();
889 nptr = &mf6ctable[hash];
890 while ((rt = *nptr) != NULL) {
891 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
892 &rt->mf6c_origin.sin6_addr) &&
893 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
894 &rt->mf6c_mcastgrp.sin6_addr) &&
895 rt->mf6c_stall == NULL)
896 break;
898 nptr = &rt->mf6c_next;
900 if (rt == NULL) {
901 crit_exit();
902 return EADDRNOTAVAIL;
905 *nptr = rt->mf6c_next;
906 kfree(rt, M_MRTABLE);
908 crit_exit();
910 return 0;
913 static int
914 socket_send(struct socket *so, struct mbuf *mm, struct sockaddr_in6 *src)
916 if (so) {
917 lwkt_gettoken(&so->so_rcv.ssb_token);
918 if (ssb_appendaddr(&so->so_rcv,
919 (struct sockaddr *)src,
920 mm, NULL) != 0) {
921 sorwakeup(so);
922 lwkt_reltoken(&so->so_rcv.ssb_token);
923 return 0;
925 lwkt_reltoken(&so->so_rcv.ssb_token);
927 m_freem(mm);
928 return -1;
932 * IPv6 multicast forwarding function. This function assumes that the packet
933 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
934 * pointed to by "ifp", and the packet is to be relayed to other networks
935 * that have members of the packet's destination IPv6 multicast group.
937 * The packet is returned unscathed to the caller, unless it is
938 * erroneous, in which case a non-zero return value tells the caller to
939 * discard it.
943 ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
945 struct mf6c *rt;
946 struct mif6 *mifp;
947 struct mbuf *mm;
948 mifi_t mifi;
950 #ifdef MRT6DEBUG
951 if (mrt6debug & DEBUG_FORWARD)
952 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
953 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst),
954 ifp->if_index);
955 #endif
958 * Don't forward a packet with Hop limit of zero or one,
959 * or a packet destined to a local-only group.
961 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
962 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
963 return 0;
964 ip6->ip6_hlim--;
967 * Source address check: do not forward packets with unspecified
968 * source. It was discussed in July 2000, on ipngwg mailing list.
969 * This is rather more serious than unicast cases, because some
970 * MLD packets can be sent with the unspecified source address
971 * (although such packets must normally set 1 to the hop limit field).
973 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
974 ip6stat.ip6s_cantforward++;
975 if (ip6_log_time + ip6_log_interval < time_uptime) {
976 ip6_log_time = time_uptime;
977 log(LOG_DEBUG,
978 "cannot forward "
979 "from %s to %s nxt %d received on %s\n",
980 ip6_sprintf(&ip6->ip6_src),
981 ip6_sprintf(&ip6->ip6_dst),
982 ip6->ip6_nxt,
983 if_name(m->m_pkthdr.rcvif));
985 return 0;
989 * Determine forwarding mifs from the forwarding cache table
991 crit_enter();
992 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
994 /* Entry exists, so forward if necessary */
995 if (rt) {
996 crit_exit();
997 return (ip6_mdq(m, ifp, rt));
998 } else {
1000 * If we don't have a route for packet's origin,
1001 * Make a copy of the packet &
1002 * send message to routing daemon
1005 struct mbuf *mb0;
1006 struct rtdetq *rte;
1007 u_long hash;
1008 /* int i, npkts;*/
1009 #ifdef UPCALL_TIMING
1010 struct timeval tp;
1012 GET_TIME(tp);
1013 #endif /* UPCALL_TIMING */
1015 mrt6stat.mrt6s_no_route++;
1016 #ifdef MRT6DEBUG
1017 if (mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1018 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1019 ip6_sprintf(&ip6->ip6_src),
1020 ip6_sprintf(&ip6->ip6_dst));
1021 #endif
1024 * Allocate mbufs early so that we don't do extra work if we
1025 * are just going to fail anyway.
1027 rte = (struct rtdetq *)kmalloc(sizeof(*rte), M_MRTABLE,
1028 M_NOWAIT);
1029 if (rte == NULL) {
1030 crit_exit();
1031 return ENOBUFS;
1033 mb0 = m_copy(m, 0, M_COPYALL);
1035 * Pullup packet header if needed before storing it,
1036 * as other references may modify it in the meantime.
1038 if (mb0 &&
1039 (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1040 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1041 if (mb0 == NULL) {
1042 kfree(rte, M_MRTABLE);
1043 crit_exit();
1044 return ENOBUFS;
1047 /* is there an upcall waiting for this packet? */
1048 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1049 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1050 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1051 &rt->mf6c_origin.sin6_addr) &&
1052 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1053 &rt->mf6c_mcastgrp.sin6_addr) &&
1054 (rt->mf6c_stall != NULL))
1055 break;
1058 if (rt == NULL) {
1059 struct mrt6msg *im;
1060 #ifdef MRT6_OINIT
1061 struct omrt6msg *oim;
1062 #endif
1064 /* no upcall, so make a new entry */
1065 rt = (struct mf6c *)kmalloc(sizeof(*rt), M_MRTABLE,
1066 M_NOWAIT);
1067 if (rt == NULL) {
1068 kfree(rte, M_MRTABLE);
1069 m_freem(mb0);
1070 crit_exit();
1071 return ENOBUFS;
1074 * Make a copy of the header to send to the user
1075 * level process
1077 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1079 if (mm == NULL) {
1080 kfree(rte, M_MRTABLE);
1081 m_freem(mb0);
1082 kfree(rt, M_MRTABLE);
1083 crit_exit();
1084 return ENOBUFS;
1088 * Send message to routing daemon
1090 sin6.sin6_addr = ip6->ip6_src;
1092 im = NULL;
1093 #ifdef MRT6_OINIT
1094 oim = NULL;
1095 #endif
1096 switch (ip6_mrouter_ver) {
1097 #ifdef MRT6_OINIT
1098 case MRT6_OINIT:
1099 oim = mtod(mm, struct omrt6msg *);
1100 oim->im6_msgtype = MRT6MSG_NOCACHE;
1101 oim->im6_mbz = 0;
1102 break;
1103 #endif
1104 case MRT6_INIT:
1105 im = mtod(mm, struct mrt6msg *);
1106 im->im6_msgtype = MRT6MSG_NOCACHE;
1107 im->im6_mbz = 0;
1108 break;
1109 default:
1110 kfree(rte, M_MRTABLE);
1111 m_freem(mb0);
1112 kfree(rt, M_MRTABLE);
1113 crit_exit();
1114 return EINVAL;
1117 #ifdef MRT6DEBUG
1118 if (mrt6debug & DEBUG_FORWARD)
1119 log(LOG_DEBUG,
1120 "getting the iif info in the kernel\n");
1121 #endif
1123 for (mifp = mif6table, mifi = 0;
1124 mifi < nummifs && mifp->m6_ifp != ifp;
1125 mifp++, mifi++)
1128 switch (ip6_mrouter_ver) {
1129 #ifdef MRT6_OINIT
1130 case MRT6_OINIT:
1131 oim->im6_mif = mifi;
1132 break;
1133 #endif
1134 case MRT6_INIT:
1135 im->im6_mif = mifi;
1136 break;
1139 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1140 log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1141 "socket queue full\n");
1142 mrt6stat.mrt6s_upq_sockfull++;
1143 kfree(rte, M_MRTABLE);
1144 m_freem(mb0);
1145 kfree(rt, M_MRTABLE);
1146 crit_exit();
1147 return ENOBUFS;
1150 mrt6stat.mrt6s_upcalls++;
1152 /* insert new entry at head of hash chain */
1153 bzero(rt, sizeof(*rt));
1154 rt->mf6c_origin.sin6_family = AF_INET6;
1155 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1156 rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1157 rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1158 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1159 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1160 rt->mf6c_expire = UPCALL_EXPIRE;
1161 n6expire[hash]++;
1162 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1164 /* link into table */
1165 rt->mf6c_next = mf6ctable[hash];
1166 mf6ctable[hash] = rt;
1167 /* Add this entry to the end of the queue */
1168 rt->mf6c_stall = rte;
1169 } else {
1170 /* determine if q has overflowed */
1171 struct rtdetq **p;
1172 int npkts = 0;
1174 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1175 if (++npkts > MAX_UPQ6) {
1176 mrt6stat.mrt6s_upq_ovflw++;
1177 kfree(rte, M_MRTABLE);
1178 m_freem(mb0);
1179 crit_exit();
1180 return 0;
1183 /* Add this entry to the end of the queue */
1184 *p = rte;
1187 rte->next = NULL;
1188 rte->m = mb0;
1189 rte->ifp = ifp;
1190 #ifdef UPCALL_TIMING
1191 rte->t = tp;
1192 #endif /* UPCALL_TIMING */
1194 crit_exit();
1196 return 0;
1201 * Clean up cache entries if upcalls are not serviced
1202 * Call from the Slow Timeout mechanism, every half second.
1204 static void
1205 expire_upcalls_dispatch(netmsg_t nmsg)
1207 struct rtdetq *rte;
1208 struct mf6c *mfc, **nptr;
1209 int i;
1211 ASSERT_IN_NETISR(0);
1213 /* Reply ASAP */
1214 crit_enter();
1215 lwkt_replymsg(&nmsg->lmsg, 0);
1216 crit_exit();
1218 for (i = 0; i < MF6CTBLSIZ; i++) {
1219 if (n6expire[i] == 0)
1220 continue;
1221 nptr = &mf6ctable[i];
1222 while ((mfc = *nptr) != NULL) {
1223 rte = mfc->mf6c_stall;
1225 * Skip real cache entries
1226 * Make sure it wasn't marked to not expire (shouldn't happen)
1227 * If it expires now
1229 if (rte != NULL &&
1230 mfc->mf6c_expire != 0 &&
1231 --mfc->mf6c_expire == 0) {
1232 #ifdef MRT6DEBUG
1233 if (mrt6debug & DEBUG_EXPIRE)
1234 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1235 ip6_sprintf(&mfc->mf6c_origin.sin6_addr),
1236 ip6_sprintf(&mfc->mf6c_mcastgrp.sin6_addr));
1237 #endif
1239 * drop all the packets
1240 * free the mbuf with the pkt, if, timing info
1242 do {
1243 struct rtdetq *n = rte->next;
1244 m_freem(rte->m);
1245 kfree(rte, M_MRTABLE);
1246 rte = n;
1247 } while (rte != NULL);
1248 mrt6stat.mrt6s_cache_cleanups++;
1249 n6expire[i]--;
1251 *nptr = mfc->mf6c_next;
1252 kfree(mfc, M_MRTABLE);
1253 } else {
1254 nptr = &mfc->mf6c_next;
1258 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1259 expire_upcalls, NULL);
1262 static void
1263 expire_upcalls(void *arg __unused)
1265 struct lwkt_msg *lmsg = &expire_upcalls_nmsg.lmsg;
1267 KASSERT(mycpuid == 0, ("expire upcalls timer not on cpu0"));
1269 crit_enter();
1270 if (lmsg->ms_flags & MSGF_DONE)
1271 lwkt_sendmsg_oncpu(netisr_cpuport(0), lmsg);
1272 crit_exit();
1276 * Packet forwarding routine once entry in the cache is made
1278 static int
1279 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1281 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1282 mifi_t mifi, iif;
1283 struct mif6 *mifp;
1284 int plen = m->m_pkthdr.len;
1285 u_int32_t dscopein, sscopein;
1288 * Macro to send packet on mif. Since RSVP packets don't get counted on
1289 * input, they shouldn't get counted on output, so statistics keeping is
1290 * separate.
1293 #define MC6_SEND(ip6, mifp, m) do { \
1294 if ((mifp)->m6_flags & MIFF_REGISTER) \
1295 register_send((ip6), (mifp), (m)); \
1296 else \
1297 phyint_send((ip6), (mifp), (m)); \
1298 } while (0)
1301 * Don't forward if it didn't arrive from the parent mif
1302 * for its origin.
1304 mifi = rt->mf6c_parent;
1305 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1306 /* came in the wrong interface */
1307 #ifdef MRT6DEBUG
1308 if (mrt6debug & DEBUG_FORWARD)
1309 log(LOG_DEBUG,
1310 "wrong if: ifid %d mifi %d mififid %x\n",
1311 ifp->if_index, mifi,
1312 mif6table[mifi].m6_ifp->if_index);
1313 #endif
1314 mrt6stat.mrt6s_wrong_if++;
1315 rt->mf6c_wrong_if++;
1317 * If we are doing PIM processing, and we are forwarding
1318 * packets on this interface, send a message to the
1319 * routing daemon.
1321 /* have to make sure this is a valid mif */
1322 if (mifi < nummifs && mif6table[mifi].m6_ifp)
1323 if (pim6 && (m->m_flags & M_LOOP) == 0) {
1325 * Check the M_LOOP flag to avoid an
1326 * unnecessary PIM assert.
1327 * XXX: M_LOOP is an ad-hoc hack...
1329 static struct sockaddr_in6 sin6 =
1330 { sizeof(sin6), AF_INET6 };
1332 struct mbuf *mm;
1333 struct mrt6msg *im;
1334 #ifdef MRT6_OINIT
1335 struct omrt6msg *oim;
1336 #endif
1338 mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1339 if (mm &&
1340 (M_HASCL(mm) ||
1341 mm->m_len < sizeof(struct ip6_hdr)))
1342 mm = m_pullup(mm, sizeof(struct ip6_hdr));
1343 if (mm == NULL)
1344 return ENOBUFS;
1346 #ifdef MRT6_OINIT
1347 oim = NULL;
1348 #endif
1349 im = NULL;
1350 switch (ip6_mrouter_ver) {
1351 #ifdef MRT6_OINIT
1352 case MRT6_OINIT:
1353 oim = mtod(mm, struct omrt6msg *);
1354 oim->im6_msgtype = MRT6MSG_WRONGMIF;
1355 oim->im6_mbz = 0;
1356 break;
1357 #endif
1358 case MRT6_INIT:
1359 im = mtod(mm, struct mrt6msg *);
1360 im->im6_msgtype = MRT6MSG_WRONGMIF;
1361 im->im6_mbz = 0;
1362 break;
1363 default:
1364 m_freem(mm);
1365 return EINVAL;
1368 for (mifp = mif6table, iif = 0;
1369 iif < nummifs && mifp &&
1370 mifp->m6_ifp != ifp;
1371 mifp++, iif++)
1374 switch (ip6_mrouter_ver) {
1375 #ifdef MRT6_OINIT
1376 case MRT6_OINIT:
1377 oim->im6_mif = iif;
1378 sin6.sin6_addr = oim->im6_src;
1379 break;
1380 #endif
1381 case MRT6_INIT:
1382 im->im6_mif = iif;
1383 sin6.sin6_addr = im->im6_src;
1384 break;
1387 mrt6stat.mrt6s_upcalls++;
1389 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1390 #ifdef MRT6DEBUG
1391 if (mrt6debug)
1392 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1393 #endif
1394 ++mrt6stat.mrt6s_upq_sockfull;
1395 return ENOBUFS;
1396 } /* if socket Q full */
1397 } /* if PIM */
1398 return 0;
1399 } /* if wrong iif */
1401 /* If I sourced this packet, it counts as output, else it was input. */
1402 if (m->m_pkthdr.rcvif == NULL) {
1403 /* XXX: is rcvif really NULL when output?? */
1404 mif6table[mifi].m6_pkt_out++;
1405 mif6table[mifi].m6_bytes_out += plen;
1406 } else {
1407 mif6table[mifi].m6_pkt_in++;
1408 mif6table[mifi].m6_bytes_in += plen;
1410 rt->mf6c_pkt_cnt++;
1411 rt->mf6c_byte_cnt += plen;
1414 * For each mif, forward a copy of the packet if there are group
1415 * members downstream on the interface.
1417 if (in6_addr2zoneid(ifp, &ip6->ip6_dst, &dscopein) ||
1418 in6_addr2zoneid(ifp, &ip6->ip6_src, &sscopein))
1419 return (EINVAL);
1420 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1421 if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1422 u_int32_t dscopeout, sscopeout;
1425 * check if the outgoing packet is going to break
1426 * a scope boundary.
1427 * XXX For packets through PIM register tunnel
1428 * interface, we believe a routing daemon.
1430 if (!(mif6table[rt->mf6c_parent].m6_flags &
1431 MIFF_REGISTER) &&
1432 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1433 if (in6_addr2zoneid(mif6table[mifi].m6_ifp,
1434 &ip6->ip6_dst,
1435 &dscopeout) ||
1436 in6_addr2zoneid(mif6table[mifi].m6_ifp,
1437 &ip6->ip6_src,
1438 &sscopeout) ||
1439 dscopein != dscopeout ||
1440 sscopein != sscopeout) {
1441 ip6stat.ip6s_badscope++;
1442 continue;
1446 mifp->m6_pkt_out++;
1447 mifp->m6_bytes_out += plen;
1448 MC6_SEND(ip6, mifp, m);
1451 return 0;
1454 static void
1455 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1457 struct mbuf *mb_copy;
1458 struct ifnet *ifp = mifp->m6_ifp;
1459 int error = 0;
1460 static struct route_in6 ro;
1461 struct in6_multi *in6m;
1462 struct sockaddr_in6 *dst6;
1463 u_long linkmtu;
1465 crit_enter(); /* needs to protect static "ro" below. */
1468 * Make a new reference to the packet; make sure that
1469 * the IPv6 header is actually copied, not just referenced,
1470 * so that ip6_output() only scribbles on the copy.
1472 mb_copy = m_copy(m, 0, M_COPYALL);
1473 if (mb_copy &&
1474 (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1475 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1476 if (mb_copy == NULL) {
1477 crit_exit();
1478 return;
1480 /* set MCAST flag to the outgoing packet */
1481 mb_copy->m_flags |= M_MCAST;
1484 * If we sourced the packet, call ip6_output since we may devide
1485 * the packet into fragments when the packet is too big for the
1486 * outgoing interface.
1487 * Otherwise, we can simply send the packet to the interface
1488 * sending queue.
1490 if (m->m_pkthdr.rcvif == NULL) {
1491 struct ip6_moptions im6o;
1493 im6o.im6o_multicast_ifp = ifp;
1494 /* XXX: ip6_output will override ip6->ip6_hlim */
1495 im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1496 im6o.im6o_multicast_loop = 1;
1497 error = ip6_output(mb_copy, NULL, &ro,
1498 IPV6_FORWARDING, &im6o, NULL, NULL);
1500 #ifdef MRT6DEBUG
1501 if (mrt6debug & DEBUG_XMIT)
1502 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1503 mifp - mif6table, error);
1504 #endif
1505 crit_exit();
1506 return;
1510 * If we belong to the destination multicast group
1511 * on the outgoing interface, loop back a copy.
1513 dst6 = (struct sockaddr_in6 *)&ro.ro_dst;
1514 in6m = IN6_LOOKUP_MULTI(&ip6->ip6_dst, ifp);
1515 if (in6m != NULL) {
1516 dst6->sin6_len = sizeof(struct sockaddr_in6);
1517 dst6->sin6_family = AF_INET6;
1518 dst6->sin6_addr = ip6->ip6_dst;
1519 ip6_mloopback(ifp, m, (struct sockaddr_in6 *)&ro.ro_dst);
1522 * Put the packet into the sending queue of the outgoing interface
1523 * if it would fit in the MTU of the interface.
1525 linkmtu = IN6_LINKMTU(ifp);
1526 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1527 dst6->sin6_len = sizeof(struct sockaddr_in6);
1528 dst6->sin6_family = AF_INET6;
1529 dst6->sin6_addr = ip6->ip6_dst;
1531 * We just call if_output instead of nd6_output here, since
1532 * we need no ND for a multicast forwarded packet...right?
1534 error = ifp->if_output(ifp, mb_copy,
1535 (struct sockaddr *)&ro.ro_dst, NULL);
1536 #ifdef MRT6DEBUG
1537 if (mrt6debug & DEBUG_XMIT)
1538 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1539 mifp - mif6table, error);
1540 #endif
1541 } else {
1542 #ifdef MULTICAST_PMTUD
1543 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1544 #else
1545 #ifdef MRT6DEBUG
1546 if (mrt6debug & DEBUG_XMIT)
1547 log(LOG_DEBUG,
1548 "phyint_send: packet too big on %s o %s g %s"
1549 " size %d(discarded)\n",
1550 if_name(ifp),
1551 ip6_sprintf(&ip6->ip6_src),
1552 ip6_sprintf(&ip6->ip6_dst),
1553 mb_copy->m_pkthdr.len);
1554 #endif /* MRT6DEBUG */
1555 m_freem(mb_copy); /* simply discard the packet */
1556 #endif
1559 crit_exit();
1562 static int
1563 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1565 struct mbuf *mm;
1566 int i, len = m->m_pkthdr.len;
1567 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1568 struct mrt6msg *im6;
1570 #ifdef MRT6DEBUG
1571 if (mrt6debug)
1572 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1573 ip6_sprintf(&ip6->ip6_src), ip6_sprintf(&ip6->ip6_dst));
1574 #endif
1575 ++pim6stat.pim6s_snd_registers;
1577 /* Make a copy of the packet to send to the user level process */
1578 MGETHDR(mm, M_NOWAIT, MT_HEADER);
1579 if (mm == NULL)
1580 return ENOBUFS;
1581 mm->m_pkthdr.rcvif = NULL;
1582 mm->m_data += max_linkhdr;
1583 mm->m_len = sizeof(struct ip6_hdr);
1585 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1586 m_freem(mm);
1587 return ENOBUFS;
1589 i = MHLEN - M_LEADINGSPACE(mm);
1590 if (i > len)
1591 i = len;
1592 mm = m_pullup(mm, i);
1593 if (mm == NULL)
1594 return ENOBUFS;
1595 /* TODO: check it! */
1596 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1599 * Send message to routing daemon
1601 sin6.sin6_addr = ip6->ip6_src;
1603 im6 = mtod(mm, struct mrt6msg *);
1604 im6->im6_msgtype = MRT6MSG_WHOLEPKT;
1605 im6->im6_mbz = 0;
1607 im6->im6_mif = mif - mif6table;
1609 /* iif info is not given for reg. encap.n */
1610 mrt6stat.mrt6s_upcalls++;
1612 if (socket_send(ip6_mrouter, mm, &sin6) < 0) {
1613 #ifdef MRT6DEBUG
1614 if (mrt6debug)
1615 log(LOG_WARNING,
1616 "register_send: ip6_mrouter socket queue full\n");
1617 #endif
1618 ++mrt6stat.mrt6s_upq_sockfull;
1619 return ENOBUFS;
1621 return 0;
1625 * PIM sparse mode hook
1626 * Receives the pim control messages, and passes them up to the listening
1627 * socket, using rip6_input.
1628 * The only message processed is the REGISTER pim message; the pim header
1629 * is stripped off, and the inner packet is passed to register_mforward.
1632 pim6_input(struct mbuf **mp, int *offp, int proto)
1634 struct pim *pim; /* pointer to a pim struct */
1635 struct ip6_hdr *ip6;
1636 int pimlen;
1637 struct mbuf *m = *mp;
1638 int minlen;
1639 int off = *offp;
1641 ++pim6stat.pim6s_rcv_total;
1643 ip6 = mtod(m, struct ip6_hdr *);
1644 pimlen = m->m_pkthdr.len - *offp;
1647 * Validate lengths
1649 if (pimlen < PIM_MINLEN) {
1650 ++pim6stat.pim6s_rcv_tooshort;
1651 #ifdef MRT6DEBUG
1652 if (mrt6debug & DEBUG_PIM)
1653 log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1654 #endif
1655 m_freem(m);
1656 return (IPPROTO_DONE);
1660 * if the packet is at least as big as a REGISTER, go ahead
1661 * and grab the PIM REGISTER header size, to avoid another
1662 * possible m_pullup() later.
1664 * PIM_MINLEN == pimhdr + u_int32 == 8
1665 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1667 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1670 * Make sure that the IP6 and PIM headers in contiguous memory, and
1671 * possibly the PIM REGISTER header
1673 #ifndef PULLDOWN_TEST
1674 IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1675 /* adjust pointer */
1676 ip6 = mtod(m, struct ip6_hdr *);
1678 /* adjust mbuf to point to the PIM header */
1679 pim = (struct pim *)((caddr_t)ip6 + off);
1680 #else
1681 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1682 if (pim == NULL) {
1683 pim6stat.pim6s_rcv_tooshort++;
1684 return IPPROTO_DONE;
1686 #endif
1688 #define PIM6_CHECKSUM
1689 #ifdef PIM6_CHECKSUM
1691 int cksumlen;
1694 * Validate checksum.
1695 * If PIM REGISTER, exclude the data packet
1697 if (pim->pim_type == PIM_REGISTER)
1698 cksumlen = PIM_MINLEN;
1699 else
1700 cksumlen = pimlen;
1702 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1703 ++pim6stat.pim6s_rcv_badsum;
1704 #ifdef MRT6DEBUG
1705 if (mrt6debug & DEBUG_PIM)
1706 log(LOG_DEBUG,
1707 "pim6_input: invalid checksum\n");
1708 #endif
1709 m_freem(m);
1710 return (IPPROTO_DONE);
1713 #endif /* PIM_CHECKSUM */
1715 /* PIM version check */
1716 if (pim->pim_ver != PIM_VERSION) {
1717 ++pim6stat.pim6s_rcv_badversion;
1718 #ifdef MRT6DEBUG
1719 log(LOG_ERR,
1720 "pim6_input: incorrect version %d, expecting %d\n",
1721 pim->pim_ver, PIM_VERSION);
1722 #endif
1723 m_freem(m);
1724 return (IPPROTO_DONE);
1727 if (pim->pim_type == PIM_REGISTER) {
1729 * since this is a REGISTER, we'll make a copy of the register
1730 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1731 * routing daemon.
1733 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1735 struct mbuf *mcp;
1736 struct ip6_hdr *eip6;
1737 u_int32_t *reghdr;
1739 ++pim6stat.pim6s_rcv_registers;
1741 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1742 #ifdef MRT6DEBUG
1743 if (mrt6debug & DEBUG_PIM)
1744 log(LOG_DEBUG,
1745 "pim6_input: register mif not set: %d\n",
1746 reg_mif_num);
1747 #endif
1748 m_freem(m);
1749 return (IPPROTO_DONE);
1752 reghdr = (u_int32_t *)(pim + 1);
1754 if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1755 goto pim6_input_to_daemon;
1758 * Validate length
1760 if (pimlen < PIM6_REG_MINLEN) {
1761 ++pim6stat.pim6s_rcv_tooshort;
1762 ++pim6stat.pim6s_rcv_badregisters;
1763 #ifdef MRT6DEBUG
1764 log(LOG_ERR,
1765 "pim6_input: register packet size too "
1766 "small %d from %s\n",
1767 pimlen, ip6_sprintf(&ip6->ip6_src));
1768 #endif
1769 m_freem(m);
1770 return (IPPROTO_DONE);
1773 eip6 = (struct ip6_hdr *) (reghdr + 1);
1774 #ifdef MRT6DEBUG
1775 if (mrt6debug & DEBUG_PIM)
1776 log(LOG_DEBUG,
1777 "pim6_input[register], eip6: %s -> %s, "
1778 "eip6 plen %d\n",
1779 ip6_sprintf(&eip6->ip6_src),
1780 ip6_sprintf(&eip6->ip6_dst),
1781 ntohs(eip6->ip6_plen));
1782 #endif
1784 /* verify the version number of the inner packet */
1785 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1786 ++pim6stat.pim6s_rcv_badregisters;
1787 #ifdef MRT6DEBUG
1788 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1789 "of the inner packet\n",
1790 (eip6->ip6_vfc & IPV6_VERSION));
1791 #endif
1792 m_freem(m);
1793 return (IPPROTO_NONE);
1796 /* verify the inner packet is destined to a mcast group */
1797 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1798 ++pim6stat.pim6s_rcv_badregisters;
1799 #ifdef MRT6DEBUG
1800 if (mrt6debug & DEBUG_PIM)
1801 log(LOG_DEBUG,
1802 "pim6_input: inner packet of register "
1803 "is not multicast %s\n",
1804 ip6_sprintf(&eip6->ip6_dst));
1805 #endif
1806 m_freem(m);
1807 return (IPPROTO_DONE);
1811 * make a copy of the whole header to pass to the daemon later.
1813 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1814 if (mcp == NULL) {
1815 #ifdef MRT6DEBUG
1816 log(LOG_ERR,
1817 "pim6_input: pim register: "
1818 "could not copy register head\n");
1819 #endif
1820 m_freem(m);
1821 return (IPPROTO_DONE);
1825 * forward the inner ip6 packet; point m_data at the inner ip6.
1827 m_adj(m, off + PIM_MINLEN);
1828 #ifdef MRT6DEBUG
1829 if (mrt6debug & DEBUG_PIM) {
1830 log(LOG_DEBUG,
1831 "pim6_input: forwarding decapsulated register: "
1832 "src %s, dst %s, mif %d\n",
1833 ip6_sprintf(&eip6->ip6_src),
1834 ip6_sprintf(&eip6->ip6_dst),
1835 reg_mif_num);
1837 #endif
1839 if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1840 dst.sin6_family, 0);
1842 /* prepare the register head to send to the mrouting daemon */
1843 m = mcp;
1847 * Pass the PIM message up to the daemon; if it is a register message
1848 * pass the 'head' only up to the daemon. This includes the
1849 * encapsulator ip6 header, pim header, register header and the
1850 * encapsulated ip6 header.
1852 pim6_input_to_daemon:
1853 rip6_input(&m, offp, proto);
1854 return (IPPROTO_DONE);