2 * Copyright (C) 1998 WIDE Project.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
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
29 * $KAME: ip6_mroute.c,v 1.58 2001/12/18 02:36:31 itojun Exp $
33 * Copyright (c) 1989 Stephen Deering
34 * Copyright (c) 1992, 1993
35 * The Regents of the University of California. All rights reserved.
37 * This code is derived from software contributed to Berkeley by
38 * Stephen Deering of Stanford University.
40 * Redistribution and use in source and binary forms, with or without
41 * modification, are permitted provided that the following conditions
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.
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
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
69 * IP multicast forwarding procedures
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
78 * MROUTING Revision: 3.5.1.2 + PIM-SMv2 (pimd) Support
81 #include <sys/cdefs.h>
82 __FBSDID("$FreeBSD$");
84 #include "opt_inet6.h"
85 #include "opt_kdtrace.h"
87 #include <sys/param.h>
88 #include <sys/callout.h>
89 #include <sys/errno.h>
90 #include <sys/kernel.h>
92 #include <sys/malloc.h>
94 #include <sys/module.h>
95 #include <sys/domain.h>
96 #include <sys/protosw.h>
98 #include <sys/signalvar.h>
99 #include <sys/socket.h>
100 #include <sys/socketvar.h>
101 #include <sys/sockio.h>
103 #include <sys/sysctl.h>
104 #include <sys/syslog.h>
105 #include <sys/systm.h>
106 #include <sys/time.h>
109 #include <net/if_types.h>
110 #include <net/raw_cb.h>
111 #include <net/vnet.h>
113 #include <netinet/in.h>
114 #include <netinet/in_var.h>
115 #include <netinet/icmp6.h>
116 #include <netinet/ip_encap.h>
118 #include <netinet/ip6.h>
119 #include <netinet/in_kdtrace.h>
120 #include <netinet6/ip6_var.h>
121 #include <netinet6/scope6_var.h>
122 #include <netinet6/nd6.h>
123 #include <netinet6/ip6_mroute.h>
124 #include <netinet6/ip6protosw.h>
125 #include <netinet6/pim6.h>
126 #include <netinet6/pim6_var.h>
128 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
130 /* XXX: this is a very common idiom; move to <sys/mbuf.h> ? */
131 #define M_HASCL(m) ((m)->m_flags & M_EXT)
133 static int ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
134 static void phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
135 static int register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
136 static int set_pim6(int *);
137 static int socket_send(struct socket *, struct mbuf *,
138 struct sockaddr_in6 *);
140 extern int in6_mcast_loop;
141 extern struct domain inet6domain;
143 static const struct encaptab *pim6_encap_cookie;
144 static const struct ip6protosw in6_pim_protosw = {
146 .pr_domain = &inet6domain,
147 .pr_protocol = IPPROTO_PIM,
148 .pr_flags = PR_ATOMIC|PR_ADDR|PR_LASTHDR,
149 .pr_input = pim6_input,
150 .pr_output = rip6_output,
151 .pr_ctloutput = rip6_ctloutput,
152 .pr_usrreqs = &rip6_usrreqs
154 static int pim6_encapcheck(const struct mbuf *, int, int, void *);
156 static VNET_DEFINE(int, ip6_mrouter_ver) = 0;
157 #define V_ip6_mrouter_ver VNET(ip6_mrouter_ver)
159 SYSCTL_DECL(_net_inet6);
160 SYSCTL_DECL(_net_inet6_ip6);
161 static SYSCTL_NODE(_net_inet6, IPPROTO_PIM, pim, CTLFLAG_RW, 0, "PIM");
163 static struct mrt6stat mrt6stat;
164 SYSCTL_STRUCT(_net_inet6_ip6, OID_AUTO, mrt6stat, CTLFLAG_RW,
166 "Multicast Routing Statistics (struct mrt6stat, netinet6/ip6_mroute.h)");
168 #define MRT6STAT_INC(name) mrt6stat.name += 1
169 #define NO_RTE_FOUND 0x1
170 #define RTE_FOUND 0x2
172 static struct mtx mrouter6_mtx;
173 #define MROUTER6_LOCK() mtx_lock(&mrouter6_mtx)
174 #define MROUTER6_UNLOCK() mtx_unlock(&mrouter6_mtx)
175 #define MROUTER6_LOCK_ASSERT() do { \
176 mtx_assert(&mrouter6_mtx, MA_OWNED); \
177 NET_ASSERT_GIANT(); \
179 #define MROUTER6_LOCK_INIT() \
180 mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
181 #define MROUTER6_LOCK_DESTROY() mtx_destroy(&mrouter6_mtx)
183 static struct mf6c *mf6ctable[MF6CTBLSIZ];
184 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
185 &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
186 "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
187 "netinet6/ip6_mroute.h)");
189 static struct mtx mfc6_mtx;
190 #define MFC6_LOCK() mtx_lock(&mfc6_mtx)
191 #define MFC6_UNLOCK() mtx_unlock(&mfc6_mtx)
192 #define MFC6_LOCK_ASSERT() do { \
193 mtx_assert(&mfc6_mtx, MA_OWNED); \
194 NET_ASSERT_GIANT(); \
196 #define MFC6_LOCK_INIT() \
197 mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
198 #define MFC6_LOCK_DESTROY() mtx_destroy(&mfc6_mtx)
200 static u_char n6expire[MF6CTBLSIZ];
202 static struct mif6 mif6table[MAXMIFS];
203 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mif6table, CTLFLAG_RD,
204 &mif6table, sizeof(mif6table), "S,mif6[MAXMIFS]",
205 "IPv6 Multicast Interfaces (struct mif6[MAXMIFS], netinet6/ip6_mroute.h)");
207 static struct mtx mif6_mtx;
208 #define MIF6_LOCK() mtx_lock(&mif6_mtx)
209 #define MIF6_UNLOCK() mtx_unlock(&mif6_mtx)
210 #define MIF6_LOCK_ASSERT() mtx_assert(&mif6_mtx, MA_OWNED)
211 #define MIF6_LOCK_INIT() \
212 mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
213 #define MIF6_LOCK_DESTROY() mtx_destroy(&mif6_mtx)
216 static VNET_DEFINE(u_int, mrt6debug) = 0; /* debug level */
217 #define V_mrt6debug VNET(mrt6debug)
218 #define DEBUG_MFC 0x02
219 #define DEBUG_FORWARD 0x04
220 #define DEBUG_EXPIRE 0x08
221 #define DEBUG_XMIT 0x10
222 #define DEBUG_REG 0x20
223 #define DEBUG_PIM 0x40
224 #define DEBUG_ERR 0x80
225 #define DEBUG_ANY 0x7f
226 #define MRT6_DLOG(m, fmt, ...) \
227 if (V_mrt6debug & (m)) \
228 log(((m) & DEBUG_ERR) ? LOG_ERR: LOG_DEBUG, \
229 "%s: " fmt "\n", __func__, ##__VA_ARGS__)
231 #define MRT6_DLOG(m, fmt, ...)
234 static void expire_upcalls(void *);
235 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */
236 #define UPCALL_EXPIRE 6 /* number of timeouts */
239 * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
243 * 'Interfaces' associated with decapsulator (so we can tell
244 * packets that went through it from ones that get reflected
245 * by a broken gateway). Different from IPv4 register_if,
246 * these interfaces are linked into the system ifnet list,
247 * because per-interface IPv6 statistics are maintained in
248 * ifp->if_afdata. But it does not have any routes point
249 * to them. I.e., packets can't be sent this way. They
250 * only exist as a placeholder for multicast source
253 static struct ifnet *multicast_register_if6;
255 #define ENCAP_HOPS 64
260 static mifi_t nummifs = 0;
261 static mifi_t reg_mif_num = (mifi_t)-1;
263 static struct pim6stat pim6stat;
264 SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RW,
266 "PIM Statistics (struct pim6stat, netinet6/pim6_var.h)");
268 #define PIM6STAT_INC(name) pim6stat.name += 1
269 static VNET_DEFINE(int, pim6);
270 #define V_pim6 VNET(pim6)
273 * Hash function for a source, group entry
275 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
276 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
277 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
278 (g).s6_addr32[2] ^ (g).s6_addr32[3])
281 * Find a route for a given origin IPv6 address and Multicast group address.
283 #define MF6CFIND(o, g, rt) do { \
284 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
286 MRT6STAT_INC(mrt6s_mfc_lookups); \
288 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
289 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
290 (_rt->mf6c_stall == NULL)) { \
294 _rt = _rt->mf6c_next; \
297 MRT6STAT_INC(mrt6s_mfc_misses); \
299 } while (/*CONSTCOND*/ 0)
302 * Macros to compute elapsed time efficiently
303 * Borrowed from Van Jacobson's scheduling code
304 * XXX: replace with timersub() ?
306 #define TV_DELTA(a, b, delta) do { \
309 delta = (a).tv_usec - (b).tv_usec; \
310 if ((xxs = (a).tv_sec - (b).tv_sec)) { \
319 delta += (1000000 * xxs); \
322 } while (/*CONSTCOND*/ 0)
324 /* XXX: replace with timercmp(a, b, <) ? */
325 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
326 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
329 #define UPCALL_MAX 50
330 static u_long upcall_data[UPCALL_MAX + 1];
331 static void collate();
332 #endif /* UPCALL_TIMING */
334 static int ip6_mrouter_init(struct socket *, int, int);
335 static int add_m6fc(struct mf6cctl *);
336 static int add_m6if(struct mif6ctl *);
337 static int del_m6fc(struct mf6cctl *);
338 static int del_m6if(mifi_t *);
339 static int del_m6if_locked(mifi_t *);
340 static int get_mif6_cnt(struct sioc_mif_req6 *);
341 static int get_sg_cnt(struct sioc_sg_req6 *);
343 static struct callout expire_upcalls_ch;
345 int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
346 int X_ip6_mrouter_done(void);
347 int X_ip6_mrouter_set(struct socket *, struct sockopt *);
348 int X_ip6_mrouter_get(struct socket *, struct sockopt *);
349 int X_mrt6_ioctl(u_long, caddr_t);
352 * Handle MRT setsockopt commands to modify the multicast routing tables.
355 X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
363 if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
366 switch (sopt->sopt_name) {
371 error = sooptcopyin(sopt, &optval, sizeof(optval),
375 error = ip6_mrouter_init(so, optval, sopt->sopt_name);
378 error = X_ip6_mrouter_done();
381 error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
384 error = add_m6if(&mifc);
387 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
390 error = add_m6fc(&mfcc);
393 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
396 error = del_m6fc(&mfcc);
399 error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
402 error = del_m6if(&mifi);
405 error = sooptcopyin(sopt, &optval, sizeof(optval),
409 error = set_pim6(&optval);
420 * Handle MRT getsockopt commands
423 X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
427 if (so != V_ip6_mrouter)
430 switch (sopt->sopt_name) {
432 error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
439 * Handle ioctl commands to obtain information from the cache
442 X_mrt6_ioctl(u_long cmd, caddr_t data)
449 case SIOCGETSGCNT_IN6:
450 ret = get_sg_cnt((struct sioc_sg_req6 *)data);
453 case SIOCGETMIFCNT_IN6:
454 ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
465 * returns the packet, byte, rpf-failure count for the source group provided
468 get_sg_cnt(struct sioc_sg_req6 *req)
477 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
481 req->pktcnt = rt->mf6c_pkt_cnt;
482 req->bytecnt = rt->mf6c_byte_cnt;
483 req->wrong_if = rt->mf6c_wrong_if;
492 * returns the input and output packet and byte counts on the mif provided
495 get_mif6_cnt(struct sioc_mif_req6 *req)
505 if (mifi >= nummifs) {
508 req->icount = mif6table[mifi].m6_pkt_in;
509 req->ocount = mif6table[mifi].m6_pkt_out;
510 req->ibytes = mif6table[mifi].m6_bytes_in;
511 req->obytes = mif6table[mifi].m6_bytes_out;
522 if ((*i != 1) && (*i != 0))
531 * Enable multicast routing
534 ip6_mrouter_init(struct socket *so, int v, int cmd)
537 MRT6_DLOG(DEBUG_ANY, "so_type = %d, pr_protocol = %d",
538 so->so_type, so->so_proto->pr_protocol);
540 if (so->so_type != SOCK_RAW ||
541 so->so_proto->pr_protocol != IPPROTO_ICMPV6)
545 return (ENOPROTOOPT);
549 if (V_ip6_mrouter != NULL) {
555 V_ip6_mrouter_ver = cmd;
557 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
558 bzero((caddr_t)n6expire, sizeof(n6expire));
560 V_pim6 = 0;/* used for stubbing out/in pim stuff */
562 callout_init(&expire_upcalls_ch, 0);
563 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
564 expire_upcalls, NULL);
567 MRT6_DLOG(DEBUG_ANY, "finished");
573 * Disable IPv6 multicast forwarding.
576 X_ip6_mrouter_done(void)
585 if (V_ip6_mrouter == NULL) {
591 * For each phyint in use, disable promiscuous reception of all IPv6
594 for (mifi = 0; mifi < nummifs; mifi++) {
595 if (mif6table[mifi].m6_ifp &&
596 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
597 if_allmulti(mif6table[mifi].m6_ifp, 0);
600 bzero((caddr_t)mif6table, sizeof(mif6table));
603 V_pim6 = 0; /* used to stub out/in pim specific code */
605 callout_stop(&expire_upcalls_ch);
608 * Free all multicast forwarding cache entries.
611 for (i = 0; i < MF6CTBLSIZ; i++) {
616 for (rte = rt->mf6c_stall; rte != NULL; ) {
617 struct rtdetq *n = rte->next;
620 free(rte, M_MRTABLE6);
625 free(frt, M_MRTABLE6);
628 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
632 * Reset register interface
634 if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
635 if_detach(multicast_register_if6);
636 if_free(multicast_register_if6);
637 reg_mif_num = (mifi_t)-1;
638 multicast_register_if6 = NULL;
641 V_ip6_mrouter = NULL;
642 V_ip6_mrouter_ver = 0;
645 MRT6_DLOG(DEBUG_ANY, "finished");
650 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
653 * Add a mif to the mif table
656 add_m6if(struct mif6ctl *mifcp)
664 if (mifcp->mif6c_mifi >= MAXMIFS) {
668 mifp = mif6table + mifcp->mif6c_mifi;
669 if (mifp->m6_ifp != NULL) {
671 return (EADDRINUSE); /* XXX: is it appropriate? */
673 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
678 ifp = ifnet_byindex(mifcp->mif6c_pifi);
680 if (mifcp->mif6c_flags & MIFF_REGISTER) {
681 if (reg_mif_num == (mifi_t)-1) {
682 ifp = if_alloc(IFT_OTHER);
684 if_initname(ifp, "register_mif", 0);
685 ifp->if_flags |= IFF_LOOPBACK;
687 multicast_register_if6 = ifp;
688 reg_mif_num = mifcp->mif6c_mifi;
690 * it is impossible to guess the ifindex of the
691 * register interface. So mif6c_pifi is automatically
694 mifcp->mif6c_pifi = ifp->if_index;
696 ifp = multicast_register_if6;
699 /* Make sure the interface supports multicast */
700 if ((ifp->if_flags & IFF_MULTICAST) == 0) {
705 error = if_allmulti(ifp, 1);
712 mifp->m6_flags = mifcp->mif6c_flags;
715 /* initialize per mif pkt counters */
717 mifp->m6_pkt_out = 0;
718 mifp->m6_bytes_in = 0;
719 mifp->m6_bytes_out = 0;
721 /* Adjust nummifs up if the mifi is higher than nummifs */
722 if (nummifs <= mifcp->mif6c_mifi)
723 nummifs = mifcp->mif6c_mifi + 1;
726 MRT6_DLOG(DEBUG_ANY, "mif #%d, phyint %s", mifcp->mif6c_mifi,
733 * Delete a mif from the mif table
736 del_m6if_locked(mifi_t *mifip)
738 struct mif6 *mifp = mif6table + *mifip;
744 if (*mifip >= nummifs)
746 if (mifp->m6_ifp == NULL)
749 if (!(mifp->m6_flags & MIFF_REGISTER)) {
750 /* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
754 if (reg_mif_num != (mifi_t)-1 &&
755 multicast_register_if6 != NULL) {
756 if_detach(multicast_register_if6);
757 if_free(multicast_register_if6);
758 reg_mif_num = (mifi_t)-1;
759 multicast_register_if6 = NULL;
763 bzero((caddr_t)mifp, sizeof(*mifp));
765 /* Adjust nummifs down */
766 for (mifi = nummifs; mifi > 0; mifi--)
767 if (mif6table[mifi - 1].m6_ifp)
770 MRT6_DLOG(DEBUG_ANY, "mif %d, nummifs %d", *mifip, nummifs);
776 del_m6if(mifi_t *mifip)
781 cc = del_m6if_locked(mifip);
791 add_m6fc(struct mf6cctl *mfccp)
797 char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
801 MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
802 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
804 /* If an entry already exists, just update the fields */
806 MRT6_DLOG(DEBUG_MFC, "no upcall o %s g %s p %x",
807 ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
808 ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
809 mfccp->mf6cc_parent);
811 rt->mf6c_parent = mfccp->mf6cc_parent;
812 rt->mf6c_ifset = mfccp->mf6cc_ifset;
819 * Find the entry for which the upcall was made and update
821 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
822 mfccp->mf6cc_mcastgrp.sin6_addr);
823 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
824 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
825 &mfccp->mf6cc_origin.sin6_addr) &&
826 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
827 &mfccp->mf6cc_mcastgrp.sin6_addr) &&
828 (rt->mf6c_stall != NULL)) {
832 "add_m6fc: %s o %s g %s p %x dbx %p\n",
833 "multiple kernel entries",
835 &mfccp->mf6cc_origin.sin6_addr),
837 &mfccp->mf6cc_mcastgrp.sin6_addr),
838 mfccp->mf6cc_parent, rt->mf6c_stall);
840 MRT6_DLOG(DEBUG_MFC, "o %s g %s p %x dbg %p",
842 &mfccp->mf6cc_origin.sin6_addr),
844 &mfccp->mf6cc_mcastgrp.sin6_addr),
845 mfccp->mf6cc_parent, rt->mf6c_stall);
847 rt->mf6c_origin = mfccp->mf6cc_origin;
848 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
849 rt->mf6c_parent = mfccp->mf6cc_parent;
850 rt->mf6c_ifset = mfccp->mf6cc_ifset;
851 /* initialize pkt counters per src-grp */
852 rt->mf6c_pkt_cnt = 0;
853 rt->mf6c_byte_cnt = 0;
854 rt->mf6c_wrong_if = 0;
856 rt->mf6c_expire = 0; /* Don't clean this guy up */
859 /* free packets Qed at the end of this entry */
860 for (rte = rt->mf6c_stall; rte != NULL; ) {
861 struct rtdetq *n = rte->next;
862 ip6_mdq(rte->m, rte->ifp, rt);
866 #endif /* UPCALL_TIMING */
867 free(rte, M_MRTABLE6);
870 rt->mf6c_stall = NULL;
875 * It is possible that an entry is being inserted without an upcall
878 MRT6_DLOG(DEBUG_MFC, "no upcall h %lu o %s g %s p %x", hash,
879 ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
880 ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
881 mfccp->mf6cc_parent);
883 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
885 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
886 &mfccp->mf6cc_origin.sin6_addr)&&
887 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
888 &mfccp->mf6cc_mcastgrp.sin6_addr)) {
890 rt->mf6c_origin = mfccp->mf6cc_origin;
891 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
892 rt->mf6c_parent = mfccp->mf6cc_parent;
893 rt->mf6c_ifset = mfccp->mf6cc_ifset;
894 /* initialize pkt counters per src-grp */
895 rt->mf6c_pkt_cnt = 0;
896 rt->mf6c_byte_cnt = 0;
897 rt->mf6c_wrong_if = 0;
905 /* no upcall, so make a new entry */
906 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
913 /* insert new entry at head of hash chain */
914 rt->mf6c_origin = mfccp->mf6cc_origin;
915 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
916 rt->mf6c_parent = mfccp->mf6cc_parent;
917 rt->mf6c_ifset = mfccp->mf6cc_ifset;
918 /* initialize pkt counters per src-grp */
919 rt->mf6c_pkt_cnt = 0;
920 rt->mf6c_byte_cnt = 0;
921 rt->mf6c_wrong_if = 0;
923 rt->mf6c_stall = NULL;
925 /* link into table */
926 rt->mf6c_next = mf6ctable[hash];
927 mf6ctable[hash] = rt;
937 * collect delay statistics on the upcalls
940 collate(struct timeval *t)
950 TV_DELTA(tp, *t, delta);
959 #endif /* UPCALL_TIMING */
962 * Delete an mfc entry
965 del_m6fc(struct mf6cctl *mfccp)
968 char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
970 struct sockaddr_in6 origin;
971 struct sockaddr_in6 mcastgrp;
976 origin = mfccp->mf6cc_origin;
977 mcastgrp = mfccp->mf6cc_mcastgrp;
978 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
980 MRT6_DLOG(DEBUG_MFC, "orig %s mcastgrp %s",
981 ip6_sprintf(ip6bufo, &origin.sin6_addr),
982 ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
986 nptr = &mf6ctable[hash];
987 while ((rt = *nptr) != NULL) {
988 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
989 &rt->mf6c_origin.sin6_addr) &&
990 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
991 &rt->mf6c_mcastgrp.sin6_addr) &&
992 rt->mf6c_stall == NULL)
995 nptr = &rt->mf6c_next;
999 return (EADDRNOTAVAIL);
1002 *nptr = rt->mf6c_next;
1003 free(rt, M_MRTABLE6);
1011 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1015 if (sbappendaddr(&s->so_rcv,
1016 (struct sockaddr *)src,
1017 mm, (struct mbuf *)0) != 0) {
1027 * IPv6 multicast forwarding function. This function assumes that the packet
1028 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1029 * pointed to by "ifp", and the packet is to be relayed to other networks
1030 * that have members of the packet's destination IPv6 multicast group.
1032 * The packet is returned unscathed to the caller, unless it is
1033 * erroneous, in which case a non-zero return value tells the caller to
1036 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1037 * this function is called in the originating context (i.e., not when
1038 * forwarding a packet from other node). ip6_output(), which is currently the
1039 * only function that calls this function is called in the originating context,
1040 * explicitly ensures this condition. It is caller's responsibility to ensure
1041 * that if this function is called from somewhere else in the originating
1042 * context in the future.
1045 X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1051 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1053 MRT6_DLOG(DEBUG_FORWARD, "src %s, dst %s, ifindex %d",
1054 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1055 ip6_sprintf(ip6bufd, &ip6->ip6_dst), ifp->if_index);
1058 * Don't forward a packet with Hop limit of zero or one,
1059 * or a packet destined to a local-only group.
1061 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1062 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1067 * Source address check: do not forward packets with unspecified
1068 * source. It was discussed in July 2000, on ipngwg mailing list.
1069 * This is rather more serious than unicast cases, because some
1070 * MLD packets can be sent with the unspecified source address
1071 * (although such packets must normally set 1 to the hop limit field).
1073 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1074 IP6STAT_INC(ip6s_cantforward);
1075 if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
1076 V_ip6_log_time = time_uptime;
1079 "from %s to %s nxt %d received on %s\n",
1080 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1081 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1083 if_name(m->m_pkthdr.rcvif));
1091 * Determine forwarding mifs from the forwarding cache table
1093 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1095 /* Entry exists, so forward if necessary */
1098 return (ip6_mdq(m, ifp, rt));
1101 * If we don't have a route for packet's origin,
1102 * Make a copy of the packet &
1103 * send message to routing daemon
1110 #ifdef UPCALL_TIMING
1114 #endif /* UPCALL_TIMING */
1116 MRT6STAT_INC(mrt6s_no_route);
1117 MRT6_DLOG(DEBUG_FORWARD | DEBUG_MFC, "no rte s %s g %s",
1118 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1119 ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1122 * Allocate mbufs early so that we don't do extra work if we
1123 * are just going to fail anyway.
1125 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6,
1131 mb0 = m_copy(m, 0, M_COPYALL);
1133 * Pullup packet header if needed before storing it,
1134 * as other references may modify it in the meantime.
1137 (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1138 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1140 free(rte, M_MRTABLE6);
1145 /* is there an upcall waiting for this packet? */
1146 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1147 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1148 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1149 &rt->mf6c_origin.sin6_addr) &&
1150 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1151 &rt->mf6c_mcastgrp.sin6_addr) &&
1152 (rt->mf6c_stall != NULL))
1159 struct omrt6msg *oim;
1162 /* no upcall, so make a new entry */
1163 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
1166 free(rte, M_MRTABLE6);
1172 * Make a copy of the header to send to the user
1175 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1178 free(rte, M_MRTABLE6);
1180 free(rt, M_MRTABLE6);
1186 * Send message to routing daemon
1188 sin6.sin6_addr = ip6->ip6_src;
1194 switch (V_ip6_mrouter_ver) {
1197 oim = mtod(mm, struct omrt6msg *);
1198 oim->im6_msgtype = MRT6MSG_NOCACHE;
1203 im = mtod(mm, struct mrt6msg *);
1204 im->im6_msgtype = MRT6MSG_NOCACHE;
1208 free(rte, M_MRTABLE6);
1210 free(rt, M_MRTABLE6);
1215 MRT6_DLOG(DEBUG_FORWARD,
1216 "getting the iif info in the kernel");
1218 for (mifp = mif6table, mifi = 0;
1219 mifi < nummifs && mifp->m6_ifp != ifp;
1223 switch (V_ip6_mrouter_ver) {
1226 oim->im6_mif = mifi;
1234 if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1235 log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1236 "socket queue full\n");
1237 MRT6STAT_INC(mrt6s_upq_sockfull);
1238 free(rte, M_MRTABLE6);
1240 free(rt, M_MRTABLE6);
1245 MRT6STAT_INC(mrt6s_upcalls);
1247 /* insert new entry at head of hash chain */
1248 bzero(rt, sizeof(*rt));
1249 rt->mf6c_origin.sin6_family = AF_INET6;
1250 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1251 rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1252 rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1253 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1254 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1255 rt->mf6c_expire = UPCALL_EXPIRE;
1257 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1259 /* link into table */
1260 rt->mf6c_next = mf6ctable[hash];
1261 mf6ctable[hash] = rt;
1262 /* Add this entry to the end of the queue */
1263 rt->mf6c_stall = rte;
1265 /* determine if q has overflowed */
1269 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1270 if (++npkts > MAX_UPQ6) {
1271 MRT6STAT_INC(mrt6s_upq_ovflw);
1272 free(rte, M_MRTABLE6);
1278 /* Add this entry to the end of the queue */
1285 #ifdef UPCALL_TIMING
1287 #endif /* UPCALL_TIMING */
1296 * Clean up cache entries if upcalls are not serviced
1297 * Call from the Slow Timeout mechanism, every half second.
1300 expire_upcalls(void *unused)
1303 char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1306 struct mf6c *mfc, **nptr;
1310 for (i = 0; i < MF6CTBLSIZ; i++) {
1311 if (n6expire[i] == 0)
1313 nptr = &mf6ctable[i];
1314 while ((mfc = *nptr) != NULL) {
1315 rte = mfc->mf6c_stall;
1317 * Skip real cache entries
1318 * Make sure it wasn't marked to not expire (shouldn't happen)
1322 mfc->mf6c_expire != 0 &&
1323 --mfc->mf6c_expire == 0) {
1324 MRT6_DLOG(DEBUG_EXPIRE, "expiring (%s %s)",
1325 ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1326 ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1328 * drop all the packets
1329 * free the mbuf with the pkt, if, timing info
1332 struct rtdetq *n = rte->next;
1334 free(rte, M_MRTABLE6);
1336 } while (rte != NULL);
1337 MRT6STAT_INC(mrt6s_cache_cleanups);
1340 *nptr = mfc->mf6c_next;
1341 free(mfc, M_MRTABLE6);
1343 nptr = &mfc->mf6c_next;
1348 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1349 expire_upcalls, NULL);
1353 * Packet forwarding routine once entry in the cache is made
1356 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1358 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1361 int plen = m->m_pkthdr.len;
1362 struct in6_addr src0, dst0; /* copies for local work */
1363 u_int32_t iszone, idzone, oszone, odzone;
1367 * Macro to send packet on mif. Since RSVP packets don't get counted on
1368 * input, they shouldn't get counted on output, so statistics keeping is
1372 #define MC6_SEND(ip6, mifp, m) do { \
1373 if ((mifp)->m6_flags & MIFF_REGISTER) \
1374 register_send((ip6), (mifp), (m)); \
1376 phyint_send((ip6), (mifp), (m)); \
1377 } while (/*CONSTCOND*/ 0)
1380 * Don't forward if it didn't arrive from the parent mif
1383 mifi = rt->mf6c_parent;
1384 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1385 /* came in the wrong interface */
1386 MRT6_DLOG(DEBUG_FORWARD,
1387 "wrong if: ifid %d mifi %d mififid %x", ifp->if_index,
1388 mifi, mif6table[mifi].m6_ifp->if_index);
1389 MRT6STAT_INC(mrt6s_wrong_if);
1390 rt->mf6c_wrong_if++;
1392 * If we are doing PIM processing, and we are forwarding
1393 * packets on this interface, send a message to the
1396 /* have to make sure this is a valid mif */
1397 if (mifi < nummifs && mif6table[mifi].m6_ifp)
1398 if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1400 * Check the M_LOOP flag to avoid an
1401 * unnecessary PIM assert.
1402 * XXX: M_LOOP is an ad-hoc hack...
1404 static struct sockaddr_in6 sin6 =
1405 { sizeof(sin6), AF_INET6 };
1410 struct omrt6msg *oim;
1413 mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1416 mm->m_len < sizeof(struct ip6_hdr)))
1417 mm = m_pullup(mm, sizeof(struct ip6_hdr));
1425 switch (V_ip6_mrouter_ver) {
1428 oim = mtod(mm, struct omrt6msg *);
1429 oim->im6_msgtype = MRT6MSG_WRONGMIF;
1434 im = mtod(mm, struct mrt6msg *);
1435 im->im6_msgtype = MRT6MSG_WRONGMIF;
1443 for (mifp = mif6table, iif = 0;
1444 iif < nummifs && mifp &&
1445 mifp->m6_ifp != ifp;
1449 switch (V_ip6_mrouter_ver) {
1453 sin6.sin6_addr = oim->im6_src;
1458 sin6.sin6_addr = im->im6_src;
1462 MRT6STAT_INC(mrt6s_upcalls);
1464 if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1465 MRT6_DLOG(DEBUG_ANY,
1466 "ip6_mrouter socket queue full");
1467 MRT6STAT_INC(mrt6s_upq_sockfull);
1469 } /* if socket Q full */
1472 } /* if wrong iif */
1474 /* If I sourced this packet, it counts as output, else it was input. */
1475 if (m->m_pkthdr.rcvif == NULL) {
1476 /* XXX: is rcvif really NULL when output?? */
1477 mif6table[mifi].m6_pkt_out++;
1478 mif6table[mifi].m6_bytes_out += plen;
1480 mif6table[mifi].m6_pkt_in++;
1481 mif6table[mifi].m6_bytes_in += plen;
1484 rt->mf6c_byte_cnt += plen;
1487 * For each mif, forward a copy of the packet if there are group
1488 * members downstream on the interface.
1490 src0 = ip6->ip6_src;
1491 dst0 = ip6->ip6_dst;
1492 if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1493 (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1494 IP6STAT_INC(ip6s_badscope);
1497 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1498 if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1500 * check if the outgoing packet is going to break
1502 * XXX For packets through PIM register tunnel
1503 * interface, we believe a routing daemon.
1505 if (!(mif6table[rt->mf6c_parent].m6_flags &
1507 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1508 if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1510 in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1514 IP6STAT_INC(ip6s_badscope);
1520 mifp->m6_bytes_out += plen;
1521 MC6_SEND(ip6, mifp, m);
1528 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1531 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1533 struct mbuf *mb_copy;
1534 struct ifnet *ifp = mifp->m6_ifp;
1539 * Make a new reference to the packet; make sure that
1540 * the IPv6 header is actually copied, not just referenced,
1541 * so that ip6_output() only scribbles on the copy.
1543 mb_copy = m_copy(m, 0, M_COPYALL);
1545 (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1546 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1547 if (mb_copy == NULL) {
1550 /* set MCAST flag to the outgoing packet */
1551 mb_copy->m_flags |= M_MCAST;
1554 * If we sourced the packet, call ip6_output since we may devide
1555 * the packet into fragments when the packet is too big for the
1556 * outgoing interface.
1557 * Otherwise, we can simply send the packet to the interface
1560 if (m->m_pkthdr.rcvif == NULL) {
1561 struct ip6_moptions im6o;
1563 im6o.im6o_multicast_ifp = ifp;
1564 /* XXX: ip6_output will override ip6->ip6_hlim */
1565 im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1566 im6o.im6o_multicast_loop = 1;
1567 error = ip6_output(mb_copy, NULL, NULL, IPV6_FORWARDING, &im6o,
1570 MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1571 (uint16_t)(mifp - mif6table), error);
1576 * If configured to loop back multicasts by default,
1577 * loop back a copy now.
1579 if (in6_mcast_loop) {
1580 struct sockaddr_in6 dst6;
1582 bzero(&dst6, sizeof(dst6));
1583 dst6.sin6_len = sizeof(struct sockaddr_in6);
1584 dst6.sin6_family = AF_INET6;
1585 dst6.sin6_addr = ip6->ip6_dst;
1586 ip6_mloopback(ifp, m, &dst6);
1590 * Put the packet into the sending queue of the outgoing interface
1591 * if it would fit in the MTU of the interface.
1593 linkmtu = IN6_LINKMTU(ifp);
1594 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1595 struct sockaddr_in6 dst6;
1597 bzero(&dst6, sizeof(dst6));
1598 dst6.sin6_len = sizeof(struct sockaddr_in6);
1599 dst6.sin6_family = AF_INET6;
1600 dst6.sin6_addr = ip6->ip6_dst;
1602 IP_PROBE(send, NULL, NULL, ip6, ifp, NULL, ip6);
1604 * We just call if_output instead of nd6_output here, since
1605 * we need no ND for a multicast forwarded packet...right?
1607 m_clrprotoflags(m); /* Avoid confusing lower layers. */
1608 error = (*ifp->if_output)(ifp, mb_copy,
1609 (struct sockaddr *)&dst6, NULL);
1610 MRT6_DLOG(DEBUG_XMIT, "mif %u err %d",
1611 (uint16_t)(mifp - mif6table), error);
1614 * pMTU discovery is intentionally disabled by default, since
1615 * various router may notify pMTU in multicast, which can be
1616 * a DDoS to a router
1618 if (V_ip6_mcast_pmtu)
1619 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1621 MRT6_DLOG(DEBUG_XMIT, " packet too big on %s o %s "
1622 "g %s size %d (discarded)", if_name(ifp),
1623 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1624 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1625 mb_copy->m_pkthdr.len);
1626 m_freem(mb_copy); /* simply discard the packet */
1632 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1635 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1638 int i, len = m->m_pkthdr.len;
1639 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1640 struct mrt6msg *im6;
1642 MRT6_DLOG(DEBUG_ANY, "src %s dst %s",
1643 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1644 ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1645 PIM6STAT_INC(pim6s_snd_registers);
1647 /* Make a copy of the packet to send to the user level process. */
1648 mm = m_gethdr(M_NOWAIT, MT_DATA);
1651 mm->m_data += max_linkhdr;
1652 mm->m_len = sizeof(struct ip6_hdr);
1654 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1658 i = MHLEN - M_LEADINGSPACE(mm);
1661 mm = m_pullup(mm, i);
1664 /* TODO: check it! */
1665 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1668 * Send message to routing daemon
1670 sin6.sin6_addr = ip6->ip6_src;
1672 im6 = mtod(mm, struct mrt6msg *);
1673 im6->im6_msgtype = MRT6MSG_WHOLEPKT;
1676 im6->im6_mif = mif - mif6table;
1678 /* iif info is not given for reg. encap.n */
1679 MRT6STAT_INC(mrt6s_upcalls);
1681 if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1682 MRT6_DLOG(DEBUG_ANY, "ip6_mrouter socket queue full");
1683 MRT6STAT_INC(mrt6s_upq_sockfull);
1690 * pim6_encapcheck() is called by the encap6_input() path at runtime to
1691 * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1695 pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
1699 KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1701 if (proto != IPPROTO_PIM)
1702 return 0; /* not for us; reject the datagram. */
1704 return 64; /* claim the datagram. */
1708 * PIM sparse mode hook
1709 * Receives the pim control messages, and passes them up to the listening
1710 * socket, using rip6_input.
1711 * The only message processed is the REGISTER pim message; the pim header
1712 * is stripped off, and the inner packet is passed to register_mforward.
1715 pim6_input(struct mbuf **mp, int *offp, int proto)
1717 struct pim *pim; /* pointer to a pim struct */
1718 struct ip6_hdr *ip6;
1720 struct mbuf *m = *mp;
1724 PIM6STAT_INC(pim6s_rcv_total);
1726 ip6 = mtod(m, struct ip6_hdr *);
1727 pimlen = m->m_pkthdr.len - *offp;
1732 if (pimlen < PIM_MINLEN) {
1733 PIM6STAT_INC(pim6s_rcv_tooshort);
1734 MRT6_DLOG(DEBUG_PIM, "PIM packet too short");
1736 return (IPPROTO_DONE);
1740 * if the packet is at least as big as a REGISTER, go ahead
1741 * and grab the PIM REGISTER header size, to avoid another
1742 * possible m_pullup() later.
1744 * PIM_MINLEN == pimhdr + u_int32 == 8
1745 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1747 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1750 * Make sure that the IP6 and PIM headers in contiguous memory, and
1751 * possibly the PIM REGISTER header
1753 #ifndef PULLDOWN_TEST
1754 IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1755 /* adjust pointer */
1756 ip6 = mtod(m, struct ip6_hdr *);
1758 /* adjust mbuf to point to the PIM header */
1759 pim = (struct pim *)((caddr_t)ip6 + off);
1761 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1763 PIM6STAT_INC(pim6s_rcv_tooshort);
1764 return (IPPROTO_DONE);
1768 #define PIM6_CHECKSUM
1769 #ifdef PIM6_CHECKSUM
1774 * Validate checksum.
1775 * If PIM REGISTER, exclude the data packet
1777 if (pim->pim_type == PIM_REGISTER)
1778 cksumlen = PIM_MINLEN;
1782 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1783 PIM6STAT_INC(pim6s_rcv_badsum);
1784 MRT6_DLOG(DEBUG_PIM, "invalid checksum");
1786 return (IPPROTO_DONE);
1789 #endif /* PIM_CHECKSUM */
1791 /* PIM version check */
1792 if (pim->pim_ver != PIM_VERSION) {
1793 PIM6STAT_INC(pim6s_rcv_badversion);
1794 MRT6_DLOG(DEBUG_ANY | DEBUG_ERR,
1795 "incorrect version %d, expecting %d",
1796 pim->pim_ver, PIM_VERSION);
1798 return (IPPROTO_DONE);
1801 if (pim->pim_type == PIM_REGISTER) {
1803 * since this is a REGISTER, we'll make a copy of the register
1804 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1807 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1810 struct ip6_hdr *eip6;
1814 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1817 PIM6STAT_INC(pim6s_rcv_registers);
1819 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1820 MRT6_DLOG(DEBUG_PIM, "register mif not set: %d",
1823 return (IPPROTO_DONE);
1826 reghdr = (u_int32_t *)(pim + 1);
1828 if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1829 goto pim6_input_to_daemon;
1834 if (pimlen < PIM6_REG_MINLEN) {
1835 PIM6STAT_INC(pim6s_rcv_tooshort);
1836 PIM6STAT_INC(pim6s_rcv_badregisters);
1837 MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "register packet "
1838 "size too small %d from %s",
1839 pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1841 return (IPPROTO_DONE);
1844 eip6 = (struct ip6_hdr *) (reghdr + 1);
1845 MRT6_DLOG(DEBUG_PIM, "eip6: %s -> %s, eip6 plen %d",
1846 ip6_sprintf(ip6bufs, &eip6->ip6_src),
1847 ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1848 ntohs(eip6->ip6_plen));
1850 /* verify the version number of the inner packet */
1851 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1852 PIM6STAT_INC(pim6s_rcv_badregisters);
1853 MRT6_DLOG(DEBUG_ANY, "invalid IP version (%d) "
1854 "of the inner packet",
1855 (eip6->ip6_vfc & IPV6_VERSION));
1857 return (IPPROTO_NONE);
1860 /* verify the inner packet is destined to a mcast group */
1861 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1862 PIM6STAT_INC(pim6s_rcv_badregisters);
1863 MRT6_DLOG(DEBUG_PIM, "inner packet of register "
1864 "is not multicast %s",
1865 ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1867 return (IPPROTO_DONE);
1871 * make a copy of the whole header to pass to the daemon later.
1873 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1875 MRT6_DLOG(DEBUG_ANY | DEBUG_ERR, "pim register: "
1876 "could not copy register head");
1878 return (IPPROTO_DONE);
1882 * forward the inner ip6 packet; point m_data at the inner ip6.
1884 m_adj(m, off + PIM_MINLEN);
1885 MRT6_DLOG(DEBUG_PIM, "forwarding decapsulated register: "
1886 "src %s, dst %s, mif %d",
1887 ip6_sprintf(ip6bufs, &eip6->ip6_src),
1888 ip6_sprintf(ip6bufd, &eip6->ip6_dst), reg_mif_num);
1890 rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
1891 dst.sin6_family, 0);
1893 /* prepare the register head to send to the mrouting daemon */
1898 * Pass the PIM message up to the daemon; if it is a register message
1899 * pass the 'head' only up to the daemon. This includes the
1900 * encapsulator ip6 header, pim header, register header and the
1901 * encapsulated ip6 header.
1903 pim6_input_to_daemon:
1904 rip6_input(&m, offp, proto);
1905 return (IPPROTO_DONE);
1909 ip6_mroute_modevent(module_t mod, int type, void *unused)
1914 MROUTER6_LOCK_INIT();
1918 pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM,
1920 (const struct protosw *)&in6_pim_protosw, NULL);
1921 if (pim6_encap_cookie == NULL) {
1922 printf("ip6_mroute: unable to attach pim6 encap\n");
1923 MIF6_LOCK_DESTROY();
1924 MFC6_LOCK_DESTROY();
1925 MROUTER6_LOCK_DESTROY();
1929 ip6_mforward = X_ip6_mforward;
1930 ip6_mrouter_done = X_ip6_mrouter_done;
1931 ip6_mrouter_get = X_ip6_mrouter_get;
1932 ip6_mrouter_set = X_ip6_mrouter_set;
1933 mrt6_ioctl = X_mrt6_ioctl;
1937 if (V_ip6_mrouter != NULL)
1940 if (pim6_encap_cookie) {
1941 encap_detach(pim6_encap_cookie);
1942 pim6_encap_cookie = NULL;
1944 X_ip6_mrouter_done();
1945 ip6_mforward = NULL;
1946 ip6_mrouter_done = NULL;
1947 ip6_mrouter_get = NULL;
1948 ip6_mrouter_set = NULL;
1951 MIF6_LOCK_DESTROY();
1952 MFC6_LOCK_DESTROY();
1953 MROUTER6_LOCK_DESTROY();
1957 return (EOPNOTSUPP);
1963 static moduledata_t ip6_mroutemod = {
1965 ip6_mroute_modevent,
1969 DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PSEUDO, SI_ORDER_ANY);