xref: /netbsd-src/sys/netinet/ip_mroute.c (revision ae1bfcddc410612bc8c58b807e1830becb69a24c)
1 /*
2  * Copyright (c) 1989 Stephen Deering
3  * Copyright (c) 1992, 1993
4  *	The Regents of the University of California.  All rights reserved.
5  *
6  * This code is derived from software contributed to Berkeley by
7  * Stephen Deering of Stanford University.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed by the University of
20  *	California, Berkeley and its contributors.
21  * 4. Neither the name of the University nor the names of its contributors
22  *    may be used to endorse or promote products derived from this software
23  *    without specific prior written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35  * SUCH DAMAGE.
36  *
37  *	from: @(#)ip_mroute.c	8.2 (Berkeley) 11/15/93
38  *	$Id: ip_mroute.c,v 1.10 1994/05/13 06:06:23 mycroft Exp $
39  */
40 
41 /*
42  * Procedures for the kernel part of DVMRP,
43  * a Distance-Vector Multicast Routing Protocol.
44  * (See RFC-1075.)
45  *
46  * Written by David Waitzman, BBN Labs, August 1988.
47  * Modified by Steve Deering, Stanford, February 1989.
48  *
49  * MROUTING 1.1
50  */
51 
52 #ifndef MROUTING
53 int	ip_mrtproto;				/* for netstat only */
54 #else
55 
56 #include <sys/param.h>
57 #include <sys/errno.h>
58 #include <sys/ioctl.h>
59 #include <sys/malloc.h>
60 #include <sys/mbuf.h>
61 #include <sys/protosw.h>
62 #include <sys/socket.h>
63 #include <sys/socketvar.h>
64 #include <sys/time.h>
65 
66 #include <net/if.h>
67 #include <net/route.h>
68 #include <net/raw_cb.h>
69 
70 #include <netinet/in.h>
71 #include <netinet/in_systm.h>
72 #include <netinet/ip.h>
73 #include <netinet/in_pcb.h>
74 #include <netinet/in_var.h>
75 #include <netinet/ip_var.h>
76 
77 #include <netinet/igmp.h>
78 #include <netinet/igmp_var.h>
79 #include <netinet/ip_mroute.h>
80 
81 /* Static forwards */
82 static	int ip_mrouter_init __P((struct socket *));
83 static	int add_vif __P((struct vifctl *));
84 static	int del_vif __P((vifi_t *vifip));
85 static	int add_lgrp __P((struct lgrplctl *));
86 static	int del_lgrp __P((struct lgrplctl *));
87 static	int grplst_member __P((struct vif *, struct in_addr));
88 static	u_long nethash __P((struct in_addr in));
89 static	int add_mrt __P((struct mrtctl *));
90 static	int del_mrt __P((struct in_addr *));
91 static	struct mrt *mrtfind __P((struct in_addr));
92 static	void phyint_send __P((struct mbuf *, struct vif *));
93 static	void tunnel_send __P((struct mbuf *, struct vif *));
94 
95 #define INSIZ sizeof(struct in_addr)
96 #define	same(a1, a2) (bcmp((caddr_t)(a1), (caddr_t)(a2), INSIZ) == 0)
97 #define	satosin(sa)	((struct sockaddr_in *)(sa))
98 
99 /*
100  * Globals.  All but ip_mrouter and ip_mrtproto could be static,
101  * except for netstat or debugging purposes.
102  */
103 struct	socket *ip_mrouter = NULL;
104 int	ip_mrtproto = IGMP_DVMRP;		/* for netstat only */
105 
106 struct	mrt *mrttable[MRTHASHSIZ];
107 struct	vif viftable[MAXVIFS];
108 struct	mrtstat	mrtstat;
109 
110 /*
111  * Private variables.
112  */
113 static	vifi_t numvifs = 0;
114 static	struct mrt *cached_mrt = NULL;
115 static	u_long cached_origin;
116 static	u_long cached_originmask;
117 
118 /*
119  * Handle DVMRP setsockopt commands to modify the multicast routing tables.
120  */
121 int
122 ip_mrouter_cmd(cmd, so, m)
123 	register int cmd;
124 	register struct socket *so;
125 	register struct mbuf *m;
126 {
127 	register int error = 0;
128 
129 	if (cmd != DVMRP_INIT && so != ip_mrouter)
130 		error = EACCES;
131 	else switch (cmd) {
132 
133 	case DVMRP_INIT:
134 		error = ip_mrouter_init(so);
135 		break;
136 
137 	case DVMRP_DONE:
138 		error = ip_mrouter_done();
139 		break;
140 
141 	case DVMRP_ADD_VIF:
142 		if (m == NULL || m->m_len < sizeof(struct vifctl))
143 			error = EINVAL;
144 		else
145 			error = add_vif(mtod(m, struct vifctl *));
146 		break;
147 
148 	case DVMRP_DEL_VIF:
149 		if (m == NULL || m->m_len < sizeof(short))
150 			error = EINVAL;
151 		else
152 			error = del_vif(mtod(m, vifi_t *));
153 		break;
154 
155 	case DVMRP_ADD_LGRP:
156 		if (m == NULL || m->m_len < sizeof(struct lgrplctl))
157 			error = EINVAL;
158 		else
159 			error = add_lgrp(mtod(m, struct lgrplctl *));
160 		break;
161 
162 	case DVMRP_DEL_LGRP:
163 		if (m == NULL || m->m_len < sizeof(struct lgrplctl))
164 			error = EINVAL;
165 		else
166 			error = del_lgrp(mtod(m, struct lgrplctl *));
167 		break;
168 
169 	case DVMRP_ADD_MRT:
170 		if (m == NULL || m->m_len < sizeof(struct mrtctl))
171 			error = EINVAL;
172 		else
173 			error = add_mrt(mtod(m, struct mrtctl *));
174 		break;
175 
176 	case DVMRP_DEL_MRT:
177 		if (m == NULL || m->m_len < sizeof(struct in_addr))
178 			error = EINVAL;
179 		else
180 			error = del_mrt(mtod(m, struct in_addr *));
181 		break;
182 
183 	default:
184 		error = EOPNOTSUPP;
185 		break;
186 	}
187 	return (error);
188 }
189 
190 /*
191  * Enable multicast routing
192  */
193 static int
194 ip_mrouter_init(so)
195 	register struct socket *so;
196 {
197 	if (so->so_type != SOCK_RAW ||
198 	    so->so_proto->pr_protocol != IPPROTO_IGMP)
199 		return (EOPNOTSUPP);
200 
201 	if (ip_mrouter != NULL)
202 		return (EADDRINUSE);
203 
204 	ip_mrouter = so;
205 
206 	return (0);
207 }
208 
209 /*
210  * Disable multicast routing
211  */
212 int
213 ip_mrouter_done()
214 {
215 	register vifi_t vifi;
216 	register int i;
217 	register struct ifnet *ifp;
218 	register int s;
219 	struct ifreq ifr;
220 
221 	s = splnet();
222 
223 	/*
224 	 * For each phyint in use, free its local group list and
225 	 * disable promiscuous reception of all IP multicasts.
226 	 */
227 	for (vifi = 0; vifi < numvifs; vifi++) {
228 		if (viftable[vifi].v_lcl_addr.s_addr != 0 &&
229 		    !(viftable[vifi].v_flags & VIFF_TUNNEL)) {
230 			if (viftable[vifi].v_lcl_grps)
231 				free(viftable[vifi].v_lcl_grps, M_MRTABLE);
232 			satosin(&ifr.ifr_addr)->sin_family = AF_INET;
233 			satosin(&ifr.ifr_addr)->sin_addr.s_addr = INADDR_ANY;
234 			ifp = viftable[vifi].v_ifp;
235 			(*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr);
236 		}
237 	}
238 	bzero((caddr_t)viftable, sizeof(viftable));
239 	numvifs = 0;
240 
241 	/*
242 	 * Free any multicast route entries.
243 	 */
244 	for (i = 0; i < MRTHASHSIZ; i++)
245 		if (mrttable[i])
246 			free(mrttable[i], M_MRTABLE);
247 	bzero((caddr_t)mrttable, sizeof(mrttable));
248 	cached_mrt = NULL;
249 
250 	ip_mrouter = NULL;
251 
252 	splx(s);
253 	return (0);
254 }
255 
256 /*
257  * Add a vif to the vif table
258  */
259 static int
260 add_vif(vifcp)
261 	register struct vifctl *vifcp;
262 {
263 	register struct vif *vifp = viftable + vifcp->vifc_vifi;
264 	register struct ifaddr *ifa;
265 	register struct ifnet *ifp;
266 	struct ifreq ifr;
267 	register int error, s;
268 	static struct sockaddr_in sin = { sizeof(sin), AF_INET };
269 
270 	if (vifcp->vifc_vifi >= MAXVIFS)
271 		return (EINVAL);
272 	if (vifp->v_lcl_addr.s_addr != 0)
273 		return (EADDRINUSE);
274 
275 	/* Find the interface with an address in AF_INET family */
276 	sin.sin_addr = vifcp->vifc_lcl_addr;
277 	ifa = ifa_ifwithaddr((struct sockaddr *)&sin);
278 	if (ifa == 0)
279 		return (EADDRNOTAVAIL);
280 
281 	s = splnet();
282 
283 	if (vifcp->vifc_flags & VIFF_TUNNEL)
284 		vifp->v_rmt_addr = vifcp->vifc_rmt_addr;
285 	else {
286 		/* Make sure the interface supports multicast */
287 		ifp = ifa->ifa_ifp;
288 		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
289 			splx(s);
290 			return (EOPNOTSUPP);
291 		}
292 		/*
293 		 * Enable promiscuous reception of all IP multicasts
294 		 * from the interface.
295 		 */
296 		satosin(&ifr.ifr_addr)->sin_family = AF_INET;
297 		satosin(&ifr.ifr_addr)->sin_addr.s_addr = INADDR_ANY;
298 		error = (*ifp->if_ioctl)(ifp, SIOCADDMULTI, (caddr_t)&ifr);
299 		if (error) {
300 			splx(s);
301 			return (error);
302 		}
303 	}
304 
305 	vifp->v_flags = vifcp->vifc_flags;
306 	vifp->v_threshold = vifcp->vifc_threshold;
307 	vifp->v_lcl_addr = vifcp->vifc_lcl_addr;
308 	vifp->v_ifp = ifa->ifa_ifp;
309 
310 	/* Adjust numvifs up if the vifi is higher than numvifs */
311 	if (numvifs <= vifcp->vifc_vifi)
312 		numvifs = vifcp->vifc_vifi + 1;
313 
314 	splx(s);
315 	return (0);
316 }
317 
318 /*
319  * Delete a vif from the vif table
320  */
321 static int
322 del_vif(vifip)
323 	register vifi_t *vifip;
324 {
325 	register struct vif *vifp = viftable + *vifip;
326 	register struct ifnet *ifp;
327 	register int i, s;
328 	struct ifreq ifr;
329 
330 	if (*vifip >= numvifs)
331 		return (EINVAL);
332 	if (vifp->v_lcl_addr.s_addr == 0)
333 		return (EADDRNOTAVAIL);
334 
335 	s = splnet();
336 
337 	if (!(vifp->v_flags & VIFF_TUNNEL)) {
338 		if (vifp->v_lcl_grps)
339 			free(vifp->v_lcl_grps, M_MRTABLE);
340 		satosin(&ifr.ifr_addr)->sin_family = AF_INET;
341 		satosin(&ifr.ifr_addr)->sin_addr.s_addr = INADDR_ANY;
342 		ifp = vifp->v_ifp;
343 		(*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr);
344 	}
345 
346 	bzero((caddr_t)vifp, sizeof (*vifp));
347 
348 	/* Adjust numvifs down */
349 	for (i = numvifs - 1; i >= 0; i--)
350 		if (viftable[i].v_lcl_addr.s_addr != 0)
351 			break;
352 	numvifs = i + 1;
353 
354 	splx(s);
355 	return (0);
356 }
357 
358 /*
359  * Add the multicast group in the lgrpctl to the list of local multicast
360  * group memberships associated with the vif indexed by gcp->lgc_vifi.
361  */
362 static int
363 add_lgrp(gcp)
364 	register struct lgrplctl *gcp;
365 {
366 	register struct vif *vifp;
367 	register int s;
368 
369 	if (gcp->lgc_vifi >= numvifs)
370 		return (EINVAL);
371 
372 	vifp = viftable + gcp->lgc_vifi;
373 	if (vifp->v_lcl_addr.s_addr == 0 || (vifp->v_flags & VIFF_TUNNEL))
374 		return (EADDRNOTAVAIL);
375 
376 	/* If not enough space in existing list, allocate a larger one */
377 	s = splnet();
378 	if (vifp->v_lcl_grps_n + 1 >= vifp->v_lcl_grps_max) {
379 		register int num;
380 		register struct in_addr *ip;
381 
382 		num = vifp->v_lcl_grps_max;
383 		if (num <= 0)
384 			num = 32;	/* initial number */
385 		else
386 			num += num;	/* double last number */
387 		ip = (struct in_addr *)malloc(num * sizeof(*ip),
388 		    M_MRTABLE, M_NOWAIT);
389 		if (ip == NULL) {
390 			splx(s);
391 			return (ENOBUFS);
392 		}
393 
394 		bzero((caddr_t)ip, num * sizeof(*ip));	/* XXX paranoid */
395 		bcopy((caddr_t)vifp->v_lcl_grps, (caddr_t)ip,
396 		    vifp->v_lcl_grps_n * sizeof(*ip));
397 
398 		vifp->v_lcl_grps_max = num;
399 		if (vifp->v_lcl_grps)
400 			free(vifp->v_lcl_grps, M_MRTABLE);
401 		vifp->v_lcl_grps = ip;
402 
403 		splx(s);
404 	}
405 
406 	vifp->v_lcl_grps[vifp->v_lcl_grps_n++] = gcp->lgc_gaddr;
407 
408 	if (gcp->lgc_gaddr.s_addr == vifp->v_cached_group)
409 		vifp->v_cached_result = 1;
410 
411 	splx(s);
412 	return (0);
413 }
414 
415 /*
416  * Delete the the local multicast group associated with the vif
417  * indexed by gcp->lgc_vifi.
418  */
419 
420 static int
421 del_lgrp(gcp)
422 	register struct lgrplctl *gcp;
423 {
424 	register struct vif *vifp;
425 	register int i, error, s;
426 
427 	if (gcp->lgc_vifi >= numvifs)
428 		return (EINVAL);
429 	vifp = viftable + gcp->lgc_vifi;
430 	if (vifp->v_lcl_addr.s_addr == 0 || (vifp->v_flags & VIFF_TUNNEL))
431 		return (EADDRNOTAVAIL);
432 
433 	s = splnet();
434 
435 	if (gcp->lgc_gaddr.s_addr == vifp->v_cached_group)
436 		vifp->v_cached_result = 0;
437 
438 	error = EADDRNOTAVAIL;
439 	for (i = 0; i < vifp->v_lcl_grps_n; ++i)
440 		if (same(&gcp->lgc_gaddr, &vifp->v_lcl_grps[i])) {
441 			error = 0;
442 			vifp->v_lcl_grps_n--;
443 			bcopy((caddr_t)&vifp->v_lcl_grps[i + 1],
444 			    (caddr_t)&vifp->v_lcl_grps[i],
445 			    (vifp->v_lcl_grps_n - i) * sizeof(struct in_addr));
446 			error = 0;
447 			break;
448 		}
449 
450 	splx(s);
451 	return (error);
452 }
453 
454 /*
455  * Return 1 if gaddr is a member of the local group list for vifp.
456  */
457 static int
458 grplst_member(vifp, gaddr)
459 	register struct vif *vifp;
460 	struct in_addr gaddr;
461 {
462 	register int i, s;
463 	register u_long addr;
464 
465 	mrtstat.mrts_grp_lookups++;
466 
467 	addr = gaddr.s_addr;
468 	if (addr == vifp->v_cached_group)
469 		return (vifp->v_cached_result);
470 
471 	mrtstat.mrts_grp_misses++;
472 
473 	for (i = 0; i < vifp->v_lcl_grps_n; ++i)
474 		if (addr == vifp->v_lcl_grps[i].s_addr) {
475 			s = splnet();
476 			vifp->v_cached_group = addr;
477 			vifp->v_cached_result = 1;
478 			splx(s);
479 			return (1);
480 		}
481 	s = splnet();
482 	vifp->v_cached_group = addr;
483 	vifp->v_cached_result = 0;
484 	splx(s);
485 	return (0);
486 }
487 
488 /*
489  * A simple hash function: returns MRTHASHMOD of the low-order octet of
490  * the argument's network or subnet number.
491  */
492 static u_long
493 nethash(in)
494 	struct in_addr in;
495 {
496 	register u_long n;
497 
498 	n = in_netof(in);
499 	while ((n & 0xff) == 0)
500 		n >>= 8;
501 	return (MRTHASHMOD(n));
502 }
503 
504 /*
505  * Add an mrt entry
506  */
507 static int
508 add_mrt(mrtcp)
509 	register struct mrtctl *mrtcp;
510 {
511 	struct mrt *rt;
512 	u_long hash;
513 	int s;
514 
515 	if (rt = mrtfind(mrtcp->mrtc_origin)) {
516 		/* Just update the route */
517 		s = splnet();
518 		rt->mrt_parent = mrtcp->mrtc_parent;
519 		VIFM_COPY(mrtcp->mrtc_children, rt->mrt_children);
520 		VIFM_COPY(mrtcp->mrtc_leaves, rt->mrt_leaves);
521 		splx(s);
522 		return (0);
523 	}
524 
525 	s = splnet();
526 
527 	rt = (struct mrt *)malloc(sizeof(*rt), M_MRTABLE, M_NOWAIT);
528 	if (rt == NULL) {
529 		splx(s);
530 		return (ENOBUFS);
531 	}
532 
533 	/*
534 	 * insert new entry at head of hash chain
535 	 */
536 	rt->mrt_origin = mrtcp->mrtc_origin;
537 	rt->mrt_originmask = mrtcp->mrtc_originmask;
538 	rt->mrt_parent = mrtcp->mrtc_parent;
539 	VIFM_COPY(mrtcp->mrtc_children, rt->mrt_children);
540 	VIFM_COPY(mrtcp->mrtc_leaves, rt->mrt_leaves);
541 	/* link into table */
542 	hash = nethash(mrtcp->mrtc_origin);
543 	rt->mrt_next = mrttable[hash];
544 	mrttable[hash] = rt;
545 
546 	splx(s);
547 	return (0);
548 }
549 
550 /*
551  * Delete an mrt entry
552  */
553 static int
554 del_mrt(origin)
555 	register struct in_addr *origin;
556 {
557 	register struct mrt *rt, *prev_rt;
558 	register u_long hash = nethash(*origin);
559 	register int s;
560 
561 	for (prev_rt = rt = mrttable[hash]; rt; prev_rt = rt, rt = rt->mrt_next)
562 		if (origin->s_addr == rt->mrt_origin.s_addr)
563 			break;
564 	if (!rt)
565 		return (ESRCH);
566 
567 	s = splnet();
568 
569 	if (rt == cached_mrt)
570 		cached_mrt = NULL;
571 
572 	if (prev_rt == rt)
573 		mrttable[hash] = rt->mrt_next;
574 	else
575 		prev_rt->mrt_next = rt->mrt_next;
576 	free(rt, M_MRTABLE);
577 
578 	splx(s);
579 	return (0);
580 }
581 
582 /*
583  * Find a route for a given origin IP address.
584  */
585 static struct mrt *
586 mrtfind(origin)
587 	struct in_addr origin;
588 {
589 	register struct mrt *rt;
590 	register u_int hash;
591 	register int s;
592 
593 	mrtstat.mrts_mrt_lookups++;
594 
595 	if (cached_mrt != NULL &&
596 	    (origin.s_addr & cached_originmask) == cached_origin)
597 		return (cached_mrt);
598 
599 	mrtstat.mrts_mrt_misses++;
600 
601 	hash = nethash(origin);
602 	for (rt = mrttable[hash]; rt; rt = rt->mrt_next)
603 		if ((origin.s_addr & rt->mrt_originmask.s_addr) ==
604 		    rt->mrt_origin.s_addr) {
605 			s = splnet();
606 			cached_mrt = rt;
607 			cached_origin = rt->mrt_origin.s_addr;
608 			cached_originmask = rt->mrt_originmask.s_addr;
609 			splx(s);
610 			return (rt);
611 		}
612 	return (NULL);
613 }
614 
615 /*
616  * IP multicast forwarding function. This function assumes that the packet
617  * pointed to by "ip" has arrived on (or is about to be sent to) the interface
618  * pointed to by "ifp", and the packet is to be relayed to other networks
619  * that have members of the packet's destination IP multicast group.
620  *
621  * The packet is returned unscathed to the caller, unless it is tunneled
622  * or erroneous, in which case a non-zero return value tells the caller to
623  * discard it.
624  */
625 
626 #define IP_HDR_LEN  20	/* # bytes of fixed IP header (excluding options) */
627 #define TUNNEL_LEN  12  /* # bytes of IP option for tunnel encapsulation  */
628 
629 int
630 ip_mforward(m, ifp)
631 	register struct mbuf *m;
632 	register struct ifnet *ifp;
633 {
634 	register struct ip *ip = mtod(m, struct ip *);
635 	register struct mrt *rt;
636 	register struct vif *vifp;
637 	register int vifi;
638 	register u_char *ipoptions;
639 	u_long tunnel_src;
640 
641 	if (ip->ip_hl < (IP_HDR_LEN + TUNNEL_LEN) >> 2 ||
642 	    (ipoptions = (u_char *)(ip + 1))[1] != IPOPT_LSRR ) {
643 		/*
644 		 * Packet arrived via a physical interface.
645 		 */
646 		tunnel_src = 0;
647 	} else {
648 		/*
649 		 * Packet arrived through a tunnel.
650 		 *
651 		 * A tunneled packet has a single NOP option and a
652 		 * two-element loose-source-and-record-route (LSRR)
653 		 * option immediately following the fixed-size part of
654 		 * the IP header.  At this point in processing, the IP
655 		 * header should contain the following IP addresses:
656 		 *
657 		 * original source          - in the source address field
658 		 * destination group        - in the destination address field
659 		 * remote tunnel end-point  - in the first  element of LSRR
660 		 * one of this host's addrs - in the second element of LSRR
661 		 *
662 		 * NOTE: RFC-1075 would have the original source and
663 		 * remote tunnel end-point addresses swapped.  However,
664 		 * that could cause delivery of ICMP error messages to
665 		 * innocent applications on intermediate routing
666 		 * hosts!  Therefore, we hereby change the spec.
667 		 */
668 
669 		/*
670 		 * Verify that the tunnel options are well-formed.
671 		 */
672 		if (ipoptions[0] != IPOPT_NOP ||
673 		    ipoptions[2] != 11 ||	/* LSRR option length   */
674 		    ipoptions[3] != 12 ||	/* LSRR address pointer */
675 		    (tunnel_src = *(u_long *)(&ipoptions[4])) == 0) {
676 			mrtstat.mrts_bad_tunnel++;
677 			return (1);
678 		}
679 
680 		/*
681 		 * Delete the tunnel options from the packet.
682 		 */
683 		ovbcopy((caddr_t)(ipoptions + TUNNEL_LEN), (caddr_t)ipoptions,
684 		    (unsigned)(m->m_len - (IP_HDR_LEN + TUNNEL_LEN)));
685 		m->m_len -= TUNNEL_LEN;
686 		ip->ip_len -= TUNNEL_LEN;
687 		ip->ip_hl -= TUNNEL_LEN >> 2;
688 	}
689 
690 	/*
691 	 * Don't forward a packet with time-to-live of zero or one,
692 	 * or a packet destined to a local-only group.
693 	 */
694 	if (ip->ip_ttl <= 1 ||
695 	    ntohl(ip->ip_dst.s_addr) <= INADDR_MAX_LOCAL_GROUP)
696 		return ((int)tunnel_src);
697 
698 	/*
699 	 * Don't forward if we don't have a route for the packet's origin.
700 	 */
701 	if (!(rt = mrtfind(ip->ip_src))) {
702 		mrtstat.mrts_no_route++;
703 		return ((int)tunnel_src);
704 	}
705 
706 	/*
707 	 * Don't forward if it didn't arrive from the parent vif for its origin.
708 	 */
709 	vifi = rt->mrt_parent;
710 	if (tunnel_src == 0 ) {
711 		if ((viftable[vifi].v_flags & VIFF_TUNNEL) ||
712 		    viftable[vifi].v_ifp != ifp )
713 			return ((int)tunnel_src);
714 	} else {
715 		if (!(viftable[vifi].v_flags & VIFF_TUNNEL) ||
716 		    viftable[vifi].v_rmt_addr.s_addr != tunnel_src )
717 			return ((int)tunnel_src);
718 	}
719 
720 	/*
721 	 * For each vif, decide if a copy of the packet should be forwarded.
722 	 * Forward if:
723 	 *		- the ttl exceeds the vif's threshold AND
724 	 *		- the vif is a child in the origin's route AND
725 	 *		- ( the vif is not a leaf in the origin's route OR
726 	 *		    the destination group has members on the vif )
727 	 *
728 	 * (This might be speeded up with some sort of cache -- someday.)
729 	 */
730 	for (vifp = viftable, vifi = 0; vifi < numvifs; vifp++, vifi++) {
731 		if (ip->ip_ttl > vifp->v_threshold &&
732 		    VIFM_ISSET(vifi, rt->mrt_children) &&
733 		    (!VIFM_ISSET(vifi, rt->mrt_leaves) ||
734 		    grplst_member(vifp, ip->ip_dst))) {
735 			if (vifp->v_flags & VIFF_TUNNEL)
736 				tunnel_send(m, vifp);
737 			else
738 				phyint_send(m, vifp);
739 		}
740 	}
741 
742 	return ((int)tunnel_src);
743 }
744 
745 static void
746 phyint_send(m, vifp)
747 	register struct mbuf *m;
748 	register struct vif *vifp;
749 {
750 	register struct ip *ip = mtod(m, struct ip *);
751 	register struct mbuf *mb_copy;
752 	register struct ip_moptions *imo;
753 	register int error;
754 	struct ip_moptions simo;
755 
756 	mb_copy = m_copy(m, 0, M_COPYALL);
757 	if (mb_copy == NULL)
758 		return;
759 
760 	imo = &simo;
761 	imo->imo_multicast_ifp = vifp->v_ifp;
762 	imo->imo_multicast_ttl = ip->ip_ttl - 1;
763 	imo->imo_multicast_loop = 1;
764 
765 	error = ip_output(mb_copy, NULL, NULL, IP_FORWARDING, imo);
766 }
767 
768 static void
769 tunnel_send(m, vifp)
770 	register struct mbuf *m;
771 	register struct vif *vifp;
772 {
773 	register struct ip *ip = mtod(m, struct ip *);
774 	register struct mbuf *mb_copy, *mb_opts;
775 	register struct ip *ip_copy;
776 	register int error;
777 	register u_char *cp;
778 
779 	/*
780 	 * Make sure that adding the tunnel options won't exceed the
781 	 * maximum allowed number of option bytes.
782 	 */
783 	if (ip->ip_hl > (60 - TUNNEL_LEN) >> 2) {
784 		mrtstat.mrts_cant_tunnel++;
785 		return;
786 	}
787 
788 	/*
789 	 * Get a private copy of the IP header so that changes to some
790 	 * of the IP fields don't damage the original header, which is
791 	 * examined later in ip_input.c.
792 	 */
793 	mb_copy = m_copy(m, IP_HDR_LEN, M_COPYALL);
794 	if (mb_copy == NULL)
795 		return;
796 	MGETHDR(mb_opts, M_DONTWAIT, MT_HEADER);
797 	if (mb_opts == NULL) {
798 		m_freem(mb_copy);
799 		return;
800 	}
801 	/*
802 	 * Make mb_opts be the new head of the packet chain.
803 	 * Any options of the packet were left in the old packet chain head
804 	 */
805 	mb_opts->m_next = mb_copy;
806 	mb_opts->m_len = IP_HDR_LEN + TUNNEL_LEN;
807 	mb_opts->m_data += MSIZE - mb_opts->m_len;
808 
809 	ip_copy = mtod(mb_opts, struct ip *);
810 	/*
811 	 * Copy the base ip header to the new head mbuf.
812 	 */
813 	*ip_copy = *ip;
814 	ip_copy->ip_ttl--;
815 	ip_copy->ip_dst = vifp->v_rmt_addr;	/* remote tunnel end-point */
816 	/*
817 	 * Adjust the ip header length to account for the tunnel options.
818 	 */
819 	ip_copy->ip_hl += TUNNEL_LEN >> 2;
820 	ip_copy->ip_len += TUNNEL_LEN;
821 	/*
822 	 * Add the NOP and LSRR after the base ip header
823 	 */
824 	cp = (u_char *)(ip_copy + 1);
825 	*cp++ = IPOPT_NOP;
826 	*cp++ = IPOPT_LSRR;
827 	*cp++ = 11;		/* LSRR option length */
828 	*cp++ = 8;		/* LSSR pointer to second element */
829 	*(u_long*)cp = vifp->v_lcl_addr.s_addr;	/* local tunnel end-point */
830 	cp += 4;
831 	*(u_long*)cp = ip->ip_dst.s_addr;		/* destination group */
832 
833 	error = ip_output(mb_opts, NULL, NULL, IP_FORWARDING, NULL);
834 }
835 #endif
836