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("$FreeBSD$");
85 #include "opt_inet6.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>
97 #include <sys/signalvar.h>
98 #include <sys/socket.h>
99 #include <sys/socketvar.h>
100 #include <sys/sockio.h>
102 #include <sys/sysctl.h>
103 #include <sys/syslog.h>
104 #include <sys/systm.h>
105 #include <sys/time.h>
108 #include <net/if_types.h>
109 #include <net/raw_cb.h>
110 #include <net/vnet.h>
112 #include <netinet/in.h>
113 #include <netinet/in_var.h>
114 #include <netinet/icmp6.h>
115 #include <netinet/ip_encap.h>
117 #include <netinet/ip6.h>
118 #include <netinet6/ip6_var.h>
119 #include <netinet6/scope6_var.h>
120 #include <netinet6/nd6.h>
121 #include <netinet6/ip6_mroute.h>
122 #include <netinet6/ip6protosw.h>
123 #include <netinet6/pim6.h>
124 #include <netinet6/pim6_var.h>
126 static MALLOC_DEFINE(M_MRTABLE6, "mf6c", "multicast forwarding cache entry");
128 /* XXX: this is a very common idiom; move to <sys/mbuf.h> ? */
129 #define M_HASCL(m) ((m)->m_flags & M_EXT)
131 static int ip6_mdq(struct mbuf *, struct ifnet *, struct mf6c *);
132 static void phyint_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
133 static void pim6_init(void);
134 static int register_send(struct ip6_hdr *, struct mif6 *, struct mbuf *);
135 static int set_pim6(int *);
136 static int socket_send(struct socket *, struct mbuf *,
137 struct sockaddr_in6 *);
139 extern int in6_mcast_loop;
140 extern struct domain inet6domain;
142 static const struct encaptab *pim6_encap_cookie;
143 static const struct ip6protosw in6_pim_protosw = {
145 .pr_domain = &inet6domain,
146 .pr_protocol = IPPROTO_PIM,
147 .pr_flags = PR_ATOMIC|PR_ADDR|PR_LASTHDR,
148 .pr_input = pim6_input,
149 .pr_output = rip6_output,
150 .pr_ctloutput = rip6_ctloutput,
151 .pr_init = pim6_init,
152 .pr_usrreqs = &rip6_usrreqs
154 static int pim6_encapcheck(const struct mbuf *, int, int, void *);
156 static VNET_DEFINE(int, ip6_mrouter_ver);
157 #define V_ip6_mrouter_ver VNET(ip6_mrouter_ver)
159 SYSCTL_DECL(_net_inet6);
160 SYSCTL_DECL(_net_inet6_ip6);
161 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 NO_RTE_FOUND 0x1
169 #define RTE_FOUND 0x2
171 static struct mtx mrouter6_mtx;
172 #define MROUTER6_LOCK() mtx_lock(&mrouter6_mtx)
173 #define MROUTER6_UNLOCK() mtx_unlock(&mrouter6_mtx)
174 #define MROUTER6_LOCK_ASSERT() do { \
175 mtx_assert(&mrouter6_mtx, MA_OWNED); \
176 NET_ASSERT_GIANT(); \
178 #define MROUTER6_LOCK_INIT() \
179 mtx_init(&mrouter6_mtx, "IPv6 multicast forwarding", NULL, MTX_DEF)
180 #define MROUTER6_LOCK_DESTROY() mtx_destroy(&mrouter6_mtx)
182 static struct mf6c *mf6ctable[MF6CTBLSIZ];
183 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mf6ctable, CTLFLAG_RD,
184 &mf6ctable, sizeof(mf6ctable), "S,*mf6ctable[MF6CTBLSIZ]",
185 "IPv6 Multicast Forwarding Table (struct *mf6ctable[MF6CTBLSIZ], "
186 "netinet6/ip6_mroute.h)");
188 static struct mtx mfc6_mtx;
189 #define MFC6_LOCK() mtx_lock(&mfc6_mtx)
190 #define MFC6_UNLOCK() mtx_unlock(&mfc6_mtx)
191 #define MFC6_LOCK_ASSERT() do { \
192 mtx_assert(&mfc6_mtx, MA_OWNED); \
193 NET_ASSERT_GIANT(); \
195 #define MFC6_LOCK_INIT() \
196 mtx_init(&mfc6_mtx, "IPv6 multicast forwarding cache", NULL, MTX_DEF)
197 #define MFC6_LOCK_DESTROY() mtx_destroy(&mfc6_mtx)
199 static u_char n6expire[MF6CTBLSIZ];
201 static struct mif6 mif6table[MAXMIFS];
202 SYSCTL_OPAQUE(_net_inet6_ip6, OID_AUTO, mif6table, CTLFLAG_RD,
203 &mif6table, sizeof(mif6table), "S,mif6[MAXMIFS]",
204 "IPv6 Multicast Interfaces (struct mif6[MAXMIFS], netinet6/ip6_mroute.h)");
206 static struct mtx mif6_mtx;
207 #define MIF6_LOCK() mtx_lock(&mif6_mtx)
208 #define MIF6_UNLOCK() mtx_unlock(&mif6_mtx)
209 #define MIF6_LOCK_ASSERT() mtx_assert(&mif6_mtx, MA_OWNED)
210 #define MIF6_LOCK_INIT() \
211 mtx_init(&mif6_mtx, "IPv6 multicast interfaces", NULL, MTX_DEF)
212 #define MIF6_LOCK_DESTROY() mtx_destroy(&mif6_mtx)
215 static VNET_DEFINE(u_int, mrt6debug); /* debug level */
216 #define V_mrt6debug VNET(mrt6debug)
217 #define DEBUG_MFC 0x02
218 #define DEBUG_FORWARD 0x04
219 #define DEBUG_EXPIRE 0x08
220 #define DEBUG_XMIT 0x10
221 #define DEBUG_REG 0x20
222 #define DEBUG_PIM 0x40
225 static void expire_upcalls(void *);
226 #define EXPIRE_TIMEOUT (hz / 4) /* 4x / second */
227 #define UPCALL_EXPIRE 6 /* number of timeouts */
230 * XXX TODO: maintain a count to if_allmulti() calls in struct ifnet.
234 * 'Interfaces' associated with decapsulator (so we can tell
235 * packets that went through it from ones that get reflected
236 * by a broken gateway). Different from IPv4 register_if,
237 * these interfaces are linked into the system ifnet list,
238 * because per-interface IPv6 statistics are maintained in
239 * ifp->if_afdata. But it does not have any routes point
240 * to them. I.e., packets can't be sent this way. They
241 * only exist as a placeholder for multicast source
244 static struct ifnet *multicast_register_if6;
246 #define ENCAP_HOPS 64
251 static mifi_t nummifs = 0;
252 static mifi_t reg_mif_num = (mifi_t)-1;
254 static struct pim6stat pim6stat;
255 SYSCTL_STRUCT(_net_inet6_pim, PIM6CTL_STATS, stats, CTLFLAG_RD,
257 "PIM Statistics (struct pim6stat, netinet6/pim_var.h)");
259 static VNET_DEFINE(int, pim6);
260 #define V_pim6 VNET(pim6)
263 * Hash function for a source, group entry
265 #define MF6CHASH(a, g) MF6CHASHMOD((a).s6_addr32[0] ^ (a).s6_addr32[1] ^ \
266 (a).s6_addr32[2] ^ (a).s6_addr32[3] ^ \
267 (g).s6_addr32[0] ^ (g).s6_addr32[1] ^ \
268 (g).s6_addr32[2] ^ (g).s6_addr32[3])
271 * Find a route for a given origin IPv6 address and Multicast group address.
273 #define MF6CFIND(o, g, rt) do { \
274 struct mf6c *_rt = mf6ctable[MF6CHASH(o,g)]; \
276 mrt6stat.mrt6s_mfc_lookups++; \
278 if (IN6_ARE_ADDR_EQUAL(&_rt->mf6c_origin.sin6_addr, &(o)) && \
279 IN6_ARE_ADDR_EQUAL(&_rt->mf6c_mcastgrp.sin6_addr, &(g)) && \
280 (_rt->mf6c_stall == NULL)) { \
284 _rt = _rt->mf6c_next; \
287 mrt6stat.mrt6s_mfc_misses++; \
289 } while (/*CONSTCOND*/ 0)
292 * Macros to compute elapsed time efficiently
293 * Borrowed from Van Jacobson's scheduling code
294 * XXX: replace with timersub() ?
296 #define TV_DELTA(a, b, delta) do { \
299 delta = (a).tv_usec - (b).tv_usec; \
300 if ((xxs = (a).tv_sec - (b).tv_sec)) { \
309 delta += (1000000 * xxs); \
312 } while (/*CONSTCOND*/ 0)
314 /* XXX: replace with timercmp(a, b, <) ? */
315 #define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
316 (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
319 #define UPCALL_MAX 50
320 static u_long upcall_data[UPCALL_MAX + 1];
321 static void collate();
322 #endif /* UPCALL_TIMING */
324 static int ip6_mrouter_init(struct socket *, int, int);
325 static int add_m6fc(struct mf6cctl *);
326 static int add_m6if(struct mif6ctl *);
327 static int del_m6fc(struct mf6cctl *);
328 static int del_m6if(mifi_t *);
329 static int del_m6if_locked(mifi_t *);
330 static int get_mif6_cnt(struct sioc_mif_req6 *);
331 static int get_sg_cnt(struct sioc_sg_req6 *);
333 static struct callout expire_upcalls_ch;
335 int X_ip6_mforward(struct ip6_hdr *, struct ifnet *, struct mbuf *);
336 int X_ip6_mrouter_done(void);
337 int X_ip6_mrouter_set(struct socket *, struct sockopt *);
338 int X_ip6_mrouter_get(struct socket *, struct sockopt *);
339 int X_mrt6_ioctl(u_long, caddr_t);
346 V_mrt6debug = 0; /* debug level */
351 * Handle MRT setsockopt commands to modify the multicast routing tables.
354 X_ip6_mrouter_set(struct socket *so, struct sockopt *sopt)
362 if (so != V_ip6_mrouter && sopt->sopt_name != MRT6_INIT)
365 switch (sopt->sopt_name) {
370 error = sooptcopyin(sopt, &optval, sizeof(optval),
374 error = ip6_mrouter_init(so, optval, sopt->sopt_name);
377 error = X_ip6_mrouter_done();
380 error = sooptcopyin(sopt, &mifc, sizeof(mifc), sizeof(mifc));
383 error = add_m6if(&mifc);
386 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
389 error = add_m6fc(&mfcc);
392 error = sooptcopyin(sopt, &mfcc, sizeof(mfcc), sizeof(mfcc));
395 error = del_m6fc(&mfcc);
398 error = sooptcopyin(sopt, &mifi, sizeof(mifi), sizeof(mifi));
401 error = del_m6if(&mifi);
404 error = sooptcopyin(sopt, &optval, sizeof(optval),
408 error = set_pim6(&optval);
419 * Handle MRT getsockopt commands
422 X_ip6_mrouter_get(struct socket *so, struct sockopt *sopt)
426 if (so != V_ip6_mrouter)
429 switch (sopt->sopt_name) {
431 error = sooptcopyout(sopt, &V_pim6, sizeof(V_pim6));
438 * Handle ioctl commands to obtain information from the cache
441 X_mrt6_ioctl(u_long cmd, caddr_t data)
448 case SIOCGETSGCNT_IN6:
449 ret = get_sg_cnt((struct sioc_sg_req6 *)data);
452 case SIOCGETMIFCNT_IN6:
453 ret = get_mif6_cnt((struct sioc_mif_req6 *)data);
464 * returns the packet, byte, rpf-failure count for the source group provided
467 get_sg_cnt(struct sioc_sg_req6 *req)
476 MF6CFIND(req->src.sin6_addr, req->grp.sin6_addr, rt);
480 req->pktcnt = rt->mf6c_pkt_cnt;
481 req->bytecnt = rt->mf6c_byte_cnt;
482 req->wrong_if = rt->mf6c_wrong_if;
491 * returns the input and output packet and byte counts on the mif provided
494 get_mif6_cnt(struct sioc_mif_req6 *req)
504 if (mifi >= nummifs) {
507 req->icount = mif6table[mifi].m6_pkt_in;
508 req->ocount = mif6table[mifi].m6_pkt_out;
509 req->ibytes = mif6table[mifi].m6_bytes_in;
510 req->obytes = mif6table[mifi].m6_bytes_out;
521 if ((*i != 1) && (*i != 0))
530 * Enable multicast routing
533 ip6_mrouter_init(struct socket *so, int v, int cmd)
536 V_ip6_mrouter_ver = 0;
543 "ip6_mrouter_init: so_type = %d, pr_protocol = %d\n",
544 so->so_type, so->so_proto->pr_protocol);
547 if (so->so_type != SOCK_RAW ||
548 so->so_proto->pr_protocol != IPPROTO_ICMPV6)
552 return (ENOPROTOOPT);
556 if (V_ip6_mrouter != NULL) {
562 V_ip6_mrouter_ver = cmd;
564 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
565 bzero((caddr_t)n6expire, sizeof(n6expire));
567 V_pim6 = 0;/* used for stubbing out/in pim stuff */
569 callout_init(&expire_upcalls_ch, 0);
570 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
571 expire_upcalls, NULL);
577 log(LOG_DEBUG, "ip6_mrouter_init\n");
584 * Disable IPv6 multicast forwarding.
587 X_ip6_mrouter_done(void)
596 if (V_ip6_mrouter == NULL) {
602 * For each phyint in use, disable promiscuous reception of all IPv6
605 for (mifi = 0; mifi < nummifs; mifi++) {
606 if (mif6table[mifi].m6_ifp &&
607 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
608 if_allmulti(mif6table[mifi].m6_ifp, 0);
611 bzero((caddr_t)mif6table, sizeof(mif6table));
614 V_pim6 = 0; /* used to stub out/in pim specific code */
616 callout_stop(&expire_upcalls_ch);
619 * Free all multicast forwarding cache entries.
622 for (i = 0; i < MF6CTBLSIZ; i++) {
627 for (rte = rt->mf6c_stall; rte != NULL; ) {
628 struct rtdetq *n = rte->next;
631 free(rte, M_MRTABLE6);
636 free(frt, M_MRTABLE6);
639 bzero((caddr_t)mf6ctable, sizeof(mf6ctable));
643 * Reset register interface
645 if (reg_mif_num != (mifi_t)-1 && multicast_register_if6 != NULL) {
646 if_detach(multicast_register_if6);
647 if_free(multicast_register_if6);
648 reg_mif_num = (mifi_t)-1;
649 multicast_register_if6 = NULL;
652 V_ip6_mrouter = NULL;
653 V_ip6_mrouter_ver = 0;
659 log(LOG_DEBUG, "ip6_mrouter_done\n");
665 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
668 * Add a mif to the mif table
671 add_m6if(struct mif6ctl *mifcp)
679 if (mifcp->mif6c_mifi >= MAXMIFS) {
683 mifp = mif6table + mifcp->mif6c_mifi;
684 if (mifp->m6_ifp != NULL) {
686 return (EADDRINUSE); /* XXX: is it appropriate? */
688 if (mifcp->mif6c_pifi == 0 || mifcp->mif6c_pifi > V_if_index) {
693 ifp = ifnet_byindex(mifcp->mif6c_pifi);
695 if (mifcp->mif6c_flags & MIFF_REGISTER) {
696 if (reg_mif_num == (mifi_t)-1) {
697 ifp = if_alloc(IFT_OTHER);
699 if_initname(ifp, "register_mif", 0);
700 ifp->if_flags |= IFF_LOOPBACK;
702 multicast_register_if6 = ifp;
703 reg_mif_num = mifcp->mif6c_mifi;
705 * it is impossible to guess the ifindex of the
706 * register interface. So mif6c_pifi is automatically
709 mifcp->mif6c_pifi = ifp->if_index;
711 ifp = multicast_register_if6;
714 /* Make sure the interface supports multicast */
715 if ((ifp->if_flags & IFF_MULTICAST) == 0) {
720 error = if_allmulti(ifp, 1);
727 mifp->m6_flags = mifcp->mif6c_flags;
730 /* initialize per mif pkt counters */
732 mifp->m6_pkt_out = 0;
733 mifp->m6_bytes_in = 0;
734 mifp->m6_bytes_out = 0;
735 bzero(&mifp->m6_route, sizeof(mifp->m6_route));
737 /* Adjust nummifs up if the mifi is higher than nummifs */
738 if (nummifs <= mifcp->mif6c_mifi)
739 nummifs = mifcp->mif6c_mifi + 1;
746 "add_mif #%d, phyint %s\n",
755 * Delete a mif from the mif table
758 del_m6if_locked(mifi_t *mifip)
760 struct mif6 *mifp = mif6table + *mifip;
766 if (*mifip >= nummifs)
768 if (mifp->m6_ifp == NULL)
771 if (!(mifp->m6_flags & MIFF_REGISTER)) {
772 /* XXX: TODO: Maintain an ALLMULTI refcount in struct ifnet. */
776 if (reg_mif_num != (mifi_t)-1 &&
777 multicast_register_if6 != NULL) {
778 if_detach(multicast_register_if6);
779 if_free(multicast_register_if6);
780 reg_mif_num = (mifi_t)-1;
781 multicast_register_if6 = NULL;
785 bzero((caddr_t)mifp, sizeof(*mifp));
787 /* Adjust nummifs down */
788 for (mifi = nummifs; mifi > 0; mifi--)
789 if (mif6table[mifi - 1].m6_ifp)
795 log(LOG_DEBUG, "del_m6if %d, nummifs %d\n", *mifip, nummifs);
802 del_m6if(mifi_t *mifip)
807 cc = del_m6if_locked(mifip);
817 add_m6fc(struct mf6cctl *mfccp)
823 char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
827 MF6CFIND(mfccp->mf6cc_origin.sin6_addr,
828 mfccp->mf6cc_mcastgrp.sin6_addr, rt);
830 /* If an entry already exists, just update the fields */
833 if (V_mrt6debug & DEBUG_MFC) {
835 "add_m6fc no upcall h %d o %s g %s p %x\n",
836 ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
837 ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
838 mfccp->mf6cc_parent);
842 rt->mf6c_parent = mfccp->mf6cc_parent;
843 rt->mf6c_ifset = mfccp->mf6cc_ifset;
850 * Find the entry for which the upcall was made and update
852 hash = MF6CHASH(mfccp->mf6cc_origin.sin6_addr,
853 mfccp->mf6cc_mcastgrp.sin6_addr);
854 for (rt = mf6ctable[hash], nstl = 0; rt; rt = rt->mf6c_next) {
855 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
856 &mfccp->mf6cc_origin.sin6_addr) &&
857 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
858 &mfccp->mf6cc_mcastgrp.sin6_addr) &&
859 (rt->mf6c_stall != NULL)) {
863 "add_m6fc: %s o %s g %s p %x dbx %p\n",
864 "multiple kernel entries",
866 &mfccp->mf6cc_origin.sin6_addr),
868 &mfccp->mf6cc_mcastgrp.sin6_addr),
869 mfccp->mf6cc_parent, rt->mf6c_stall);
872 if (V_mrt6debug & DEBUG_MFC)
874 "add_m6fc o %s g %s p %x dbg %x\n",
876 &mfccp->mf6cc_origin.sin6_addr),
878 &mfccp->mf6cc_mcastgrp.sin6_addr),
879 mfccp->mf6cc_parent, rt->mf6c_stall);
882 rt->mf6c_origin = mfccp->mf6cc_origin;
883 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
884 rt->mf6c_parent = mfccp->mf6cc_parent;
885 rt->mf6c_ifset = mfccp->mf6cc_ifset;
886 /* initialize pkt counters per src-grp */
887 rt->mf6c_pkt_cnt = 0;
888 rt->mf6c_byte_cnt = 0;
889 rt->mf6c_wrong_if = 0;
891 rt->mf6c_expire = 0; /* Don't clean this guy up */
894 /* free packets Qed at the end of this entry */
895 for (rte = rt->mf6c_stall; rte != NULL; ) {
896 struct rtdetq *n = rte->next;
897 ip6_mdq(rte->m, rte->ifp, rt);
901 #endif /* UPCALL_TIMING */
902 free(rte, M_MRTABLE6);
905 rt->mf6c_stall = NULL;
910 * It is possible that an entry is being inserted without an upcall
914 if (V_mrt6debug & DEBUG_MFC)
916 "add_mfc no upcall h %d o %s g %s p %x\n",
918 ip6_sprintf(ip6bufo, &mfccp->mf6cc_origin.sin6_addr),
919 ip6_sprintf(ip6bufg, &mfccp->mf6cc_mcastgrp.sin6_addr),
920 mfccp->mf6cc_parent);
923 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
925 if (IN6_ARE_ADDR_EQUAL(&rt->mf6c_origin.sin6_addr,
926 &mfccp->mf6cc_origin.sin6_addr)&&
927 IN6_ARE_ADDR_EQUAL(&rt->mf6c_mcastgrp.sin6_addr,
928 &mfccp->mf6cc_mcastgrp.sin6_addr)) {
930 rt->mf6c_origin = mfccp->mf6cc_origin;
931 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
932 rt->mf6c_parent = mfccp->mf6cc_parent;
933 rt->mf6c_ifset = mfccp->mf6cc_ifset;
934 /* initialize pkt counters per src-grp */
935 rt->mf6c_pkt_cnt = 0;
936 rt->mf6c_byte_cnt = 0;
937 rt->mf6c_wrong_if = 0;
945 /* no upcall, so make a new entry */
946 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
953 /* insert new entry at head of hash chain */
954 rt->mf6c_origin = mfccp->mf6cc_origin;
955 rt->mf6c_mcastgrp = mfccp->mf6cc_mcastgrp;
956 rt->mf6c_parent = mfccp->mf6cc_parent;
957 rt->mf6c_ifset = mfccp->mf6cc_ifset;
958 /* initialize pkt counters per src-grp */
959 rt->mf6c_pkt_cnt = 0;
960 rt->mf6c_byte_cnt = 0;
961 rt->mf6c_wrong_if = 0;
963 rt->mf6c_stall = NULL;
965 /* link into table */
966 rt->mf6c_next = mf6ctable[hash];
967 mf6ctable[hash] = rt;
977 * collect delay statistics on the upcalls
980 collate(struct timeval *t)
990 TV_DELTA(tp, *t, delta);
999 #endif /* UPCALL_TIMING */
1002 * Delete an mfc entry
1005 del_m6fc(struct mf6cctl *mfccp)
1007 struct sockaddr_in6 origin;
1008 struct sockaddr_in6 mcastgrp;
1013 origin = mfccp->mf6cc_origin;
1014 mcastgrp = mfccp->mf6cc_mcastgrp;
1015 hash = MF6CHASH(origin.sin6_addr, mcastgrp.sin6_addr);
1018 if (V_mrt6debug & DEBUG_MFC) {
1019 char ip6bufo[INET6_ADDRSTRLEN], ip6bufg[INET6_ADDRSTRLEN];
1020 log(LOG_DEBUG,"del_m6fc orig %s mcastgrp %s\n",
1021 ip6_sprintf(ip6bufo, &origin.sin6_addr),
1022 ip6_sprintf(ip6bufg, &mcastgrp.sin6_addr));
1028 nptr = &mf6ctable[hash];
1029 while ((rt = *nptr) != NULL) {
1030 if (IN6_ARE_ADDR_EQUAL(&origin.sin6_addr,
1031 &rt->mf6c_origin.sin6_addr) &&
1032 IN6_ARE_ADDR_EQUAL(&mcastgrp.sin6_addr,
1033 &rt->mf6c_mcastgrp.sin6_addr) &&
1034 rt->mf6c_stall == NULL)
1037 nptr = &rt->mf6c_next;
1041 return (EADDRNOTAVAIL);
1044 *nptr = rt->mf6c_next;
1045 free(rt, M_MRTABLE6);
1053 socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in6 *src)
1057 if (sbappendaddr(&s->so_rcv,
1058 (struct sockaddr *)src,
1059 mm, (struct mbuf *)0) != 0) {
1069 * IPv6 multicast forwarding function. This function assumes that the packet
1070 * pointed to by "ip6" has arrived on (or is about to be sent to) the interface
1071 * pointed to by "ifp", and the packet is to be relayed to other networks
1072 * that have members of the packet's destination IPv6 multicast group.
1074 * The packet is returned unscathed to the caller, unless it is
1075 * erroneous, in which case a non-zero return value tells the caller to
1078 * NOTE: this implementation assumes that m->m_pkthdr.rcvif is NULL iff
1079 * this function is called in the originating context (i.e., not when
1080 * forwarding a packet from other node). ip6_output(), which is currently the
1081 * only function that calls this function is called in the originating context,
1082 * explicitly ensures this condition. It is caller's responsibility to ensure
1083 * that if this function is called from somewhere else in the originating
1084 * context in the future.
1087 X_ip6_mforward(struct ip6_hdr *ip6, struct ifnet *ifp, struct mbuf *m)
1093 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1096 if (V_mrt6debug & DEBUG_FORWARD)
1097 log(LOG_DEBUG, "ip6_mforward: src %s, dst %s, ifindex %d\n",
1098 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1099 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1104 * Don't forward a packet with Hop limit of zero or one,
1105 * or a packet destined to a local-only group.
1107 if (ip6->ip6_hlim <= 1 || IN6_IS_ADDR_MC_INTFACELOCAL(&ip6->ip6_dst) ||
1108 IN6_IS_ADDR_MC_LINKLOCAL(&ip6->ip6_dst))
1113 * Source address check: do not forward packets with unspecified
1114 * source. It was discussed in July 2000, on ipngwg mailing list.
1115 * This is rather more serious than unicast cases, because some
1116 * MLD packets can be sent with the unspecified source address
1117 * (although such packets must normally set 1 to the hop limit field).
1119 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1120 V_ip6stat.ip6s_cantforward++;
1121 if (V_ip6_log_time + V_ip6_log_interval < time_second) {
1122 V_ip6_log_time = time_second;
1125 "from %s to %s nxt %d received on %s\n",
1126 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1127 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1129 if_name(m->m_pkthdr.rcvif));
1137 * Determine forwarding mifs from the forwarding cache table
1139 MF6CFIND(ip6->ip6_src, ip6->ip6_dst, rt);
1141 /* Entry exists, so forward if necessary */
1144 return (ip6_mdq(m, ifp, rt));
1147 * If we don't have a route for packet's origin,
1148 * Make a copy of the packet &
1149 * send message to routing daemon
1156 #ifdef UPCALL_TIMING
1160 #endif /* UPCALL_TIMING */
1162 mrt6stat.mrt6s_no_route++;
1164 if (V_mrt6debug & (DEBUG_FORWARD | DEBUG_MFC))
1165 log(LOG_DEBUG, "ip6_mforward: no rte s %s g %s\n",
1166 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1167 ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1171 * Allocate mbufs early so that we don't do extra work if we
1172 * are just going to fail anyway.
1174 rte = (struct rtdetq *)malloc(sizeof(*rte), M_MRTABLE6,
1180 mb0 = m_copy(m, 0, M_COPYALL);
1182 * Pullup packet header if needed before storing it,
1183 * as other references may modify it in the meantime.
1186 (M_HASCL(mb0) || mb0->m_len < sizeof(struct ip6_hdr)))
1187 mb0 = m_pullup(mb0, sizeof(struct ip6_hdr));
1189 free(rte, M_MRTABLE6);
1194 /* is there an upcall waiting for this packet? */
1195 hash = MF6CHASH(ip6->ip6_src, ip6->ip6_dst);
1196 for (rt = mf6ctable[hash]; rt; rt = rt->mf6c_next) {
1197 if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_src,
1198 &rt->mf6c_origin.sin6_addr) &&
1199 IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
1200 &rt->mf6c_mcastgrp.sin6_addr) &&
1201 (rt->mf6c_stall != NULL))
1208 struct omrt6msg *oim;
1211 /* no upcall, so make a new entry */
1212 rt = (struct mf6c *)malloc(sizeof(*rt), M_MRTABLE6,
1215 free(rte, M_MRTABLE6);
1221 * Make a copy of the header to send to the user
1224 mm = m_copy(mb0, 0, sizeof(struct ip6_hdr));
1227 free(rte, M_MRTABLE6);
1229 free(rt, M_MRTABLE6);
1235 * Send message to routing daemon
1237 sin6.sin6_addr = ip6->ip6_src;
1243 switch (V_ip6_mrouter_ver) {
1246 oim = mtod(mm, struct omrt6msg *);
1247 oim->im6_msgtype = MRT6MSG_NOCACHE;
1252 im = mtod(mm, struct mrt6msg *);
1253 im->im6_msgtype = MRT6MSG_NOCACHE;
1257 free(rte, M_MRTABLE6);
1259 free(rt, M_MRTABLE6);
1265 if (V_mrt6debug & DEBUG_FORWARD)
1267 "getting the iif info in the kernel\n");
1270 for (mifp = mif6table, mifi = 0;
1271 mifi < nummifs && mifp->m6_ifp != ifp;
1275 switch (V_ip6_mrouter_ver) {
1278 oim->im6_mif = mifi;
1286 if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1287 log(LOG_WARNING, "ip6_mforward: ip6_mrouter "
1288 "socket queue full\n");
1289 mrt6stat.mrt6s_upq_sockfull++;
1290 free(rte, M_MRTABLE6);
1292 free(rt, M_MRTABLE6);
1297 mrt6stat.mrt6s_upcalls++;
1299 /* insert new entry at head of hash chain */
1300 bzero(rt, sizeof(*rt));
1301 rt->mf6c_origin.sin6_family = AF_INET6;
1302 rt->mf6c_origin.sin6_len = sizeof(struct sockaddr_in6);
1303 rt->mf6c_origin.sin6_addr = ip6->ip6_src;
1304 rt->mf6c_mcastgrp.sin6_family = AF_INET6;
1305 rt->mf6c_mcastgrp.sin6_len = sizeof(struct sockaddr_in6);
1306 rt->mf6c_mcastgrp.sin6_addr = ip6->ip6_dst;
1307 rt->mf6c_expire = UPCALL_EXPIRE;
1309 rt->mf6c_parent = MF6C_INCOMPLETE_PARENT;
1311 /* link into table */
1312 rt->mf6c_next = mf6ctable[hash];
1313 mf6ctable[hash] = rt;
1314 /* Add this entry to the end of the queue */
1315 rt->mf6c_stall = rte;
1317 /* determine if q has overflowed */
1321 for (p = &rt->mf6c_stall; *p != NULL; p = &(*p)->next)
1322 if (++npkts > MAX_UPQ6) {
1323 mrt6stat.mrt6s_upq_ovflw++;
1324 free(rte, M_MRTABLE6);
1330 /* Add this entry to the end of the queue */
1337 #ifdef UPCALL_TIMING
1339 #endif /* UPCALL_TIMING */
1348 * Clean up cache entries if upcalls are not serviced
1349 * Call from the Slow Timeout mechanism, every half second.
1352 expire_upcalls(void *unused)
1355 struct mf6c *mfc, **nptr;
1359 for (i = 0; i < MF6CTBLSIZ; i++) {
1360 if (n6expire[i] == 0)
1362 nptr = &mf6ctable[i];
1363 while ((mfc = *nptr) != NULL) {
1364 rte = mfc->mf6c_stall;
1366 * Skip real cache entries
1367 * Make sure it wasn't marked to not expire (shouldn't happen)
1371 mfc->mf6c_expire != 0 &&
1372 --mfc->mf6c_expire == 0) {
1374 if (V_mrt6debug & DEBUG_EXPIRE) {
1375 char ip6bufo[INET6_ADDRSTRLEN];
1376 char ip6bufg[INET6_ADDRSTRLEN];
1377 log(LOG_DEBUG, "expire_upcalls: expiring (%s %s)\n",
1378 ip6_sprintf(ip6bufo, &mfc->mf6c_origin.sin6_addr),
1379 ip6_sprintf(ip6bufg, &mfc->mf6c_mcastgrp.sin6_addr));
1383 * drop all the packets
1384 * free the mbuf with the pkt, if, timing info
1387 struct rtdetq *n = rte->next;
1389 free(rte, M_MRTABLE6);
1391 } while (rte != NULL);
1392 mrt6stat.mrt6s_cache_cleanups++;
1395 *nptr = mfc->mf6c_next;
1396 free(mfc, M_MRTABLE6);
1398 nptr = &mfc->mf6c_next;
1403 callout_reset(&expire_upcalls_ch, EXPIRE_TIMEOUT,
1404 expire_upcalls, NULL);
1408 * Packet forwarding routine once entry in the cache is made
1411 ip6_mdq(struct mbuf *m, struct ifnet *ifp, struct mf6c *rt)
1413 struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
1416 int plen = m->m_pkthdr.len;
1417 struct in6_addr src0, dst0; /* copies for local work */
1418 u_int32_t iszone, idzone, oszone, odzone;
1422 * Macro to send packet on mif. Since RSVP packets don't get counted on
1423 * input, they shouldn't get counted on output, so statistics keeping is
1427 #define MC6_SEND(ip6, mifp, m) do { \
1428 if ((mifp)->m6_flags & MIFF_REGISTER) \
1429 register_send((ip6), (mifp), (m)); \
1431 phyint_send((ip6), (mifp), (m)); \
1432 } while (/*CONSTCOND*/ 0)
1435 * Don't forward if it didn't arrive from the parent mif
1438 mifi = rt->mf6c_parent;
1439 if ((mifi >= nummifs) || (mif6table[mifi].m6_ifp != ifp)) {
1440 /* came in the wrong interface */
1442 if (V_mrt6debug & DEBUG_FORWARD)
1444 "wrong if: ifid %d mifi %d mififid %x\n",
1445 ifp->if_index, mifi,
1446 mif6table[mifi].m6_ifp->if_index);
1448 mrt6stat.mrt6s_wrong_if++;
1449 rt->mf6c_wrong_if++;
1451 * If we are doing PIM processing, and we are forwarding
1452 * packets on this interface, send a message to the
1455 /* have to make sure this is a valid mif */
1456 if (mifi < nummifs && mif6table[mifi].m6_ifp)
1457 if (V_pim6 && (m->m_flags & M_LOOP) == 0) {
1459 * Check the M_LOOP flag to avoid an
1460 * unnecessary PIM assert.
1461 * XXX: M_LOOP is an ad-hoc hack...
1463 static struct sockaddr_in6 sin6 =
1464 { sizeof(sin6), AF_INET6 };
1469 struct omrt6msg *oim;
1472 mm = m_copy(m, 0, sizeof(struct ip6_hdr));
1475 mm->m_len < sizeof(struct ip6_hdr)))
1476 mm = m_pullup(mm, sizeof(struct ip6_hdr));
1484 switch (V_ip6_mrouter_ver) {
1487 oim = mtod(mm, struct omrt6msg *);
1488 oim->im6_msgtype = MRT6MSG_WRONGMIF;
1493 im = mtod(mm, struct mrt6msg *);
1494 im->im6_msgtype = MRT6MSG_WRONGMIF;
1502 for (mifp = mif6table, iif = 0;
1503 iif < nummifs && mifp &&
1504 mifp->m6_ifp != ifp;
1508 switch (V_ip6_mrouter_ver) {
1512 sin6.sin6_addr = oim->im6_src;
1517 sin6.sin6_addr = im->im6_src;
1521 mrt6stat.mrt6s_upcalls++;
1523 if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1526 log(LOG_WARNING, "mdq, ip6_mrouter socket queue full\n");
1528 ++mrt6stat.mrt6s_upq_sockfull;
1530 } /* if socket Q full */
1533 } /* if wrong iif */
1535 /* If I sourced this packet, it counts as output, else it was input. */
1536 if (m->m_pkthdr.rcvif == NULL) {
1537 /* XXX: is rcvif really NULL when output?? */
1538 mif6table[mifi].m6_pkt_out++;
1539 mif6table[mifi].m6_bytes_out += plen;
1541 mif6table[mifi].m6_pkt_in++;
1542 mif6table[mifi].m6_bytes_in += plen;
1545 rt->mf6c_byte_cnt += plen;
1548 * For each mif, forward a copy of the packet if there are group
1549 * members downstream on the interface.
1551 src0 = ip6->ip6_src;
1552 dst0 = ip6->ip6_dst;
1553 if ((error = in6_setscope(&src0, ifp, &iszone)) != 0 ||
1554 (error = in6_setscope(&dst0, ifp, &idzone)) != 0) {
1555 V_ip6stat.ip6s_badscope++;
1558 for (mifp = mif6table, mifi = 0; mifi < nummifs; mifp++, mifi++) {
1559 if (IF_ISSET(mifi, &rt->mf6c_ifset)) {
1561 * check if the outgoing packet is going to break
1563 * XXX For packets through PIM register tunnel
1564 * interface, we believe a routing daemon.
1566 if (!(mif6table[rt->mf6c_parent].m6_flags &
1568 !(mif6table[mifi].m6_flags & MIFF_REGISTER)) {
1569 if (in6_setscope(&src0, mif6table[mifi].m6_ifp,
1571 in6_setscope(&dst0, mif6table[mifi].m6_ifp,
1575 V_ip6stat.ip6s_badscope++;
1581 mifp->m6_bytes_out += plen;
1582 MC6_SEND(ip6, mifp, m);
1589 phyint_send(struct ip6_hdr *ip6, struct mif6 *mifp, struct mbuf *m)
1591 struct mbuf *mb_copy;
1592 struct ifnet *ifp = mifp->m6_ifp;
1594 struct sockaddr_in6 *dst6;
1597 dst6 = &mifp->m6_route.ro_dst;
1600 * Make a new reference to the packet; make sure that
1601 * the IPv6 header is actually copied, not just referenced,
1602 * so that ip6_output() only scribbles on the copy.
1604 mb_copy = m_copy(m, 0, M_COPYALL);
1606 (M_HASCL(mb_copy) || mb_copy->m_len < sizeof(struct ip6_hdr)))
1607 mb_copy = m_pullup(mb_copy, sizeof(struct ip6_hdr));
1608 if (mb_copy == NULL) {
1611 /* set MCAST flag to the outgoing packet */
1612 mb_copy->m_flags |= M_MCAST;
1615 * If we sourced the packet, call ip6_output since we may devide
1616 * the packet into fragments when the packet is too big for the
1617 * outgoing interface.
1618 * Otherwise, we can simply send the packet to the interface
1621 if (m->m_pkthdr.rcvif == NULL) {
1622 struct ip6_moptions im6o;
1624 im6o.im6o_multicast_ifp = ifp;
1625 /* XXX: ip6_output will override ip6->ip6_hlim */
1626 im6o.im6o_multicast_hlim = ip6->ip6_hlim;
1627 im6o.im6o_multicast_loop = 1;
1628 error = ip6_output(mb_copy, NULL, &mifp->m6_route,
1629 IPV6_FORWARDING, &im6o, NULL, NULL);
1632 if (V_mrt6debug & DEBUG_XMIT)
1633 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1634 mifp - mif6table, error);
1640 * If configured to loop back multicasts by default,
1641 * loop back a copy now.
1643 if (in6_mcast_loop) {
1644 dst6->sin6_len = sizeof(struct sockaddr_in6);
1645 dst6->sin6_family = AF_INET6;
1646 dst6->sin6_addr = ip6->ip6_dst;
1647 ip6_mloopback(ifp, m, &mifp->m6_route.ro_dst);
1651 * Put the packet into the sending queue of the outgoing interface
1652 * if it would fit in the MTU of the interface.
1654 linkmtu = IN6_LINKMTU(ifp);
1655 if (mb_copy->m_pkthdr.len <= linkmtu || linkmtu < IPV6_MMTU) {
1656 dst6->sin6_len = sizeof(struct sockaddr_in6);
1657 dst6->sin6_family = AF_INET6;
1658 dst6->sin6_addr = ip6->ip6_dst;
1660 * We just call if_output instead of nd6_output here, since
1661 * we need no ND for a multicast forwarded packet...right?
1663 error = (*ifp->if_output)(ifp, mb_copy,
1664 (struct sockaddr *)&mifp->m6_route.ro_dst, NULL);
1666 if (V_mrt6debug & DEBUG_XMIT)
1667 log(LOG_DEBUG, "phyint_send on mif %d err %d\n",
1668 mifp - mif6table, error);
1672 * pMTU discovery is intentionally disabled by default, since
1673 * various router may notify pMTU in multicast, which can be
1674 * a DDoS to a router
1676 if (V_ip6_mcast_pmtu)
1677 icmp6_error(mb_copy, ICMP6_PACKET_TOO_BIG, 0, linkmtu);
1680 if (V_mrt6debug & DEBUG_XMIT) {
1681 char ip6bufs[INET6_ADDRSTRLEN];
1682 char ip6bufd[INET6_ADDRSTRLEN];
1684 "phyint_send: packet too big on %s o %s "
1685 "g %s size %d(discarded)\n",
1687 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1688 ip6_sprintf(ip6bufd, &ip6->ip6_dst),
1689 mb_copy->m_pkthdr.len);
1691 #endif /* MRT6DEBUG */
1692 m_freem(mb_copy); /* simply discard the packet */
1698 register_send(struct ip6_hdr *ip6, struct mif6 *mif, struct mbuf *m)
1701 int i, len = m->m_pkthdr.len;
1702 static struct sockaddr_in6 sin6 = { sizeof(sin6), AF_INET6 };
1703 struct mrt6msg *im6;
1707 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1708 log(LOG_DEBUG, "** IPv6 register_send **\n src %s dst %s\n",
1709 ip6_sprintf(ip6bufs, &ip6->ip6_src),
1710 ip6_sprintf(ip6bufd, &ip6->ip6_dst));
1713 ++pim6stat.pim6s_snd_registers;
1715 /* Make a copy of the packet to send to the user level process */
1716 MGETHDR(mm, M_DONTWAIT, MT_HEADER);
1719 mm->m_pkthdr.rcvif = NULL;
1720 mm->m_data += max_linkhdr;
1721 mm->m_len = sizeof(struct ip6_hdr);
1723 if ((mm->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1727 i = MHLEN - M_LEADINGSPACE(mm);
1730 mm = m_pullup(mm, i);
1733 /* TODO: check it! */
1734 mm->m_pkthdr.len = len + sizeof(struct ip6_hdr);
1737 * Send message to routing daemon
1739 sin6.sin6_addr = ip6->ip6_src;
1741 im6 = mtod(mm, struct mrt6msg *);
1742 im6->im6_msgtype = MRT6MSG_WHOLEPKT;
1745 im6->im6_mif = mif - mif6table;
1747 /* iif info is not given for reg. encap.n */
1748 mrt6stat.mrt6s_upcalls++;
1750 if (socket_send(V_ip6_mrouter, mm, &sin6) < 0) {
1754 "register_send: ip6_mrouter socket queue full\n");
1756 ++mrt6stat.mrt6s_upq_sockfull;
1763 * pim6_encapcheck() is called by the encap6_input() path at runtime to
1764 * determine if a packet is for PIM; allowing PIM to be dynamically loaded
1768 pim6_encapcheck(const struct mbuf *m, int off, int proto, void *arg)
1772 KASSERT(proto == IPPROTO_PIM, ("not for IPPROTO_PIM"));
1774 if (proto != IPPROTO_PIM)
1775 return 0; /* not for us; reject the datagram. */
1777 return 64; /* claim the datagram. */
1781 * PIM sparse mode hook
1782 * Receives the pim control messages, and passes them up to the listening
1783 * socket, using rip6_input.
1784 * The only message processed is the REGISTER pim message; the pim header
1785 * is stripped off, and the inner packet is passed to register_mforward.
1788 pim6_input(struct mbuf **mp, int *offp, int proto)
1790 struct pim *pim; /* pointer to a pim struct */
1791 struct ip6_hdr *ip6;
1793 struct mbuf *m = *mp;
1797 ++pim6stat.pim6s_rcv_total;
1799 ip6 = mtod(m, struct ip6_hdr *);
1800 pimlen = m->m_pkthdr.len - *offp;
1805 if (pimlen < PIM_MINLEN) {
1806 ++pim6stat.pim6s_rcv_tooshort;
1808 if (V_mrt6debug & DEBUG_PIM)
1809 log(LOG_DEBUG,"pim6_input: PIM packet too short\n");
1812 return (IPPROTO_DONE);
1816 * if the packet is at least as big as a REGISTER, go ahead
1817 * and grab the PIM REGISTER header size, to avoid another
1818 * possible m_pullup() later.
1820 * PIM_MINLEN == pimhdr + u_int32 == 8
1821 * PIM6_REG_MINLEN == pimhdr + reghdr + eip6hdr == 4 + 4 + 40
1823 minlen = (pimlen >= PIM6_REG_MINLEN) ? PIM6_REG_MINLEN : PIM_MINLEN;
1826 * Make sure that the IP6 and PIM headers in contiguous memory, and
1827 * possibly the PIM REGISTER header
1829 #ifndef PULLDOWN_TEST
1830 IP6_EXTHDR_CHECK(m, off, minlen, IPPROTO_DONE);
1831 /* adjust pointer */
1832 ip6 = mtod(m, struct ip6_hdr *);
1834 /* adjust mbuf to point to the PIM header */
1835 pim = (struct pim *)((caddr_t)ip6 + off);
1837 IP6_EXTHDR_GET(pim, struct pim *, m, off, minlen);
1839 pim6stat.pim6s_rcv_tooshort++;
1840 return (IPPROTO_DONE);
1844 #define PIM6_CHECKSUM
1845 #ifdef PIM6_CHECKSUM
1850 * Validate checksum.
1851 * If PIM REGISTER, exclude the data packet
1853 if (pim->pim_type == PIM_REGISTER)
1854 cksumlen = PIM_MINLEN;
1858 if (in6_cksum(m, IPPROTO_PIM, off, cksumlen)) {
1859 ++pim6stat.pim6s_rcv_badsum;
1861 if (V_mrt6debug & DEBUG_PIM)
1863 "pim6_input: invalid checksum\n");
1866 return (IPPROTO_DONE);
1869 #endif /* PIM_CHECKSUM */
1871 /* PIM version check */
1872 if (pim->pim_ver != PIM_VERSION) {
1873 ++pim6stat.pim6s_rcv_badversion;
1876 "pim6_input: incorrect version %d, expecting %d\n",
1877 pim->pim_ver, PIM_VERSION);
1880 return (IPPROTO_DONE);
1883 if (pim->pim_type == PIM_REGISTER) {
1885 * since this is a REGISTER, we'll make a copy of the register
1886 * headers ip6+pim+u_int32_t+encap_ip6, to be passed up to the
1889 static struct sockaddr_in6 dst = { sizeof(dst), AF_INET6 };
1892 struct ip6_hdr *eip6;
1896 char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
1899 ++pim6stat.pim6s_rcv_registers;
1901 if ((reg_mif_num >= nummifs) || (reg_mif_num == (mifi_t) -1)) {
1903 if (V_mrt6debug & DEBUG_PIM)
1905 "pim6_input: register mif not set: %d\n",
1909 return (IPPROTO_DONE);
1912 reghdr = (u_int32_t *)(pim + 1);
1914 if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
1915 goto pim6_input_to_daemon;
1920 if (pimlen < PIM6_REG_MINLEN) {
1921 ++pim6stat.pim6s_rcv_tooshort;
1922 ++pim6stat.pim6s_rcv_badregisters;
1925 "pim6_input: register packet size too "
1926 "small %d from %s\n",
1927 pimlen, ip6_sprintf(ip6bufs, &ip6->ip6_src));
1930 return (IPPROTO_DONE);
1933 eip6 = (struct ip6_hdr *) (reghdr + 1);
1935 if (V_mrt6debug & DEBUG_PIM)
1937 "pim6_input[register], eip6: %s -> %s, "
1939 ip6_sprintf(ip6bufs, &eip6->ip6_src),
1940 ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1941 ntohs(eip6->ip6_plen));
1944 /* verify the version number of the inner packet */
1945 if ((eip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
1946 ++pim6stat.pim6s_rcv_badregisters;
1948 log(LOG_DEBUG, "pim6_input: invalid IP version (%d) "
1949 "of the inner packet\n",
1950 (eip6->ip6_vfc & IPV6_VERSION));
1953 return (IPPROTO_NONE);
1956 /* verify the inner packet is destined to a mcast group */
1957 if (!IN6_IS_ADDR_MULTICAST(&eip6->ip6_dst)) {
1958 ++pim6stat.pim6s_rcv_badregisters;
1960 if (V_mrt6debug & DEBUG_PIM)
1962 "pim6_input: inner packet of register "
1963 "is not multicast %s\n",
1964 ip6_sprintf(ip6bufd, &eip6->ip6_dst));
1967 return (IPPROTO_DONE);
1971 * make a copy of the whole header to pass to the daemon later.
1973 mcp = m_copy(m, 0, off + PIM6_REG_MINLEN);
1977 "pim6_input: pim register: "
1978 "could not copy register head\n");
1981 return (IPPROTO_DONE);
1985 * forward the inner ip6 packet; point m_data at the inner ip6.
1987 m_adj(m, off + PIM_MINLEN);
1989 if (V_mrt6debug & DEBUG_PIM) {
1991 "pim6_input: forwarding decapsulated register: "
1992 "src %s, dst %s, mif %d\n",
1993 ip6_sprintf(ip6bufs, &eip6->ip6_src),
1994 ip6_sprintf(ip6bufd, &eip6->ip6_dst),
1999 rc = if_simloop(mif6table[reg_mif_num].m6_ifp, m,
2000 dst.sin6_family, 0);
2002 /* prepare the register head to send to the mrouting daemon */
2007 * Pass the PIM message up to the daemon; if it is a register message
2008 * pass the 'head' only up to the daemon. This includes the
2009 * encapsulator ip6 header, pim header, register header and the
2010 * encapsulated ip6 header.
2012 pim6_input_to_daemon:
2013 rip6_input(&m, offp, proto);
2014 return (IPPROTO_DONE);
2018 ip6_mroute_modevent(module_t mod, int type, void *unused)
2023 MROUTER6_LOCK_INIT();
2027 pim6_encap_cookie = encap_attach_func(AF_INET6, IPPROTO_PIM,
2029 (const struct protosw *)&in6_pim_protosw, NULL);
2030 if (pim6_encap_cookie == NULL) {
2031 printf("ip6_mroute: unable to attach pim6 encap\n");
2032 MIF6_LOCK_DESTROY();
2033 MFC6_LOCK_DESTROY();
2034 MROUTER6_LOCK_DESTROY();
2038 ip6_mforward = X_ip6_mforward;
2039 ip6_mrouter_done = X_ip6_mrouter_done;
2040 ip6_mrouter_get = X_ip6_mrouter_get;
2041 ip6_mrouter_set = X_ip6_mrouter_set;
2042 mrt6_ioctl = X_mrt6_ioctl;
2046 if (V_ip6_mrouter != NULL)
2049 if (pim6_encap_cookie) {
2050 encap_detach(pim6_encap_cookie);
2051 pim6_encap_cookie = NULL;
2053 X_ip6_mrouter_done();
2054 ip6_mforward = NULL;
2055 ip6_mrouter_done = NULL;
2056 ip6_mrouter_get = NULL;
2057 ip6_mrouter_set = NULL;
2060 MIF6_LOCK_DESTROY();
2061 MFC6_LOCK_DESTROY();
2062 MROUTER6_LOCK_DESTROY();
2066 return (EOPNOTSUPP);
2072 static moduledata_t ip6_mroutemod = {
2074 ip6_mroute_modevent,
2078 DECLARE_MODULE(ip6_mroute, ip6_mroutemod, SI_SUB_PSEUDO, SI_ORDER_ANY);