1/*
2 * Copyright (c) 2000-2013 Apple Inc. All rights reserved.
3 *
4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
5 *
6 * This file contains Original Code and/or Modifications of Original Code
7 * as defined in and that are subject to the Apple Public Source License
8 * Version 2.0 (the 'License'). You may not use this file except in
9 * compliance with the License. The rights granted to you under the License
10 * may not be used to create, or enable the creation or redistribution of,
11 * unlawful or unlicensed copies of an Apple operating system, or to
12 * circumvent, violate, or enable the circumvention or violation of, any
13 * terms of an Apple operating system software license agreement.
14 *
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
17 *
18 * The Original Code and all software distributed under the License are
19 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
22 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
23 * Please see the License for the specific language governing rights and
24 * limitations under the License.
25 *
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
27 */
28/*
29 * NOTICE: This file was modified by SPARTA, Inc. in 2005 to introduce
30 * support for mandatory and extensible security protections.  This notice
31 * is included in support of clause 2.2 (b) of the Apple Public License,
32 * Version 2.0.
33 */
34/*
35 * IP multicast forwarding procedures
36 *
37 * Written by David Waitzman, BBN Labs, August 1988.
38 * Modified by Steve Deering, Stanford, February 1989.
39 * Modified by Mark J. Steiglitz, Stanford, May, 1991
40 * Modified by Van Jacobson, LBL, January 1993
41 * Modified by Ajit Thyagarajan, PARC, August 1993
42 * Modified by Bill Fenner, PARC, April 1995
43 *
44 * MROUTING Revision: 3.5
45 * $FreeBSD: src/sys/netinet/ip_mroute.c,v 1.56.2.2 2001/07/19 06:37:26 kris Exp $
46 */
47
48
49#include <sys/param.h>
50#include <sys/systm.h>
51#include <sys/malloc.h>
52#include <sys/mbuf.h>
53#include <sys/socket.h>
54#include <sys/socketvar.h>
55#include <sys/protosw.h>
56#include <sys/time.h>
57#include <sys/kernel.h>
58#include <sys/sockio.h>
59#include <sys/syslog.h>
60
61#include <machine/endian.h>
62
63#include <net/if.h>
64#include <net/route.h>
65#include <net/kpi_protocol.h>
66#include <netinet/in.h>
67#include <netinet/in_systm.h>
68#include <netinet/ip.h>
69#include <netinet/ip_var.h>
70#include <netinet/in_var.h>
71#include <netinet/igmp.h>
72#include <netinet/ip_mroute.h>
73#include <netinet/udp.h>
74
75#if CONFIG_MACF_NET
76#include <security/mac_framework.h>
77#endif
78
79
80#if !MROUTING
81extern u_int32_t	_ip_mcast_src(int vifi);
82extern int	_ip_mforward(struct ip *ip, struct ifnet *ifp,
83				  struct mbuf *m, struct ip_moptions *imo);
84extern int	_ip_mrouter_done(void);
85extern int	_ip_mrouter_get(struct socket *so, struct sockopt *sopt);
86extern int	_ip_mrouter_set(struct socket *so, struct sockopt *sopt);
87extern int	_mrt_ioctl(int req, caddr_t data, struct proc *p);
88
89/*
90 * Dummy routines and globals used when multicast routing is not compiled in.
91 */
92
93struct socket  *ip_mrouter  = NULL;
94u_int		rsvpdebug = 0;
95
96int
97_ip_mrouter_set(__unused struct socket *so,
98		__unused struct sockopt *sopt)
99{
100	return(EOPNOTSUPP);
101}
102
103int (*ip_mrouter_set)(struct socket *, struct sockopt *) = _ip_mrouter_set;
104
105
106int
107_ip_mrouter_get(__unused struct socket *so,
108		__unused sockopt *sopt)
109{
110	return(EOPNOTSUPP);
111}
112
113int (*ip_mrouter_get)(struct socket *, struct sockopt *) = _ip_mrouter_get;
114
115int
116_ip_mrouter_done(void)
117{
118	return(0);
119}
120
121int (*ip_mrouter_done)(void) = _ip_mrouter_done;
122
123int
124_ip_mforward(__unused struct ip *ip, __unused struct ifnet *ifp,
125	     __unused struct mbuf *m, __unused ip_moptions *imo)
126{
127	return(0);
128}
129
130int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
131		   struct ip_moptions *) = _ip_mforward;
132
133int
134_mrt_ioctl(__unused u_long req, __unused caddr_t data, __unused struct proc *p)
135{
136	return EOPNOTSUPP;
137}
138
139int (*mrt_ioctl)(u_long, caddr_t, struct proc *) = _mrt_ioctl;
140
141void
142rsvp_input(struct mbuf *m, int iphlen)		/* XXX must fixup manually */
143{
144    /* Can still get packets with rsvp_on = 0 if there is a local member
145     * of the group to which the RSVP packet is addressed.  But in this
146     * case we want to throw the packet away.
147     */
148    if (!rsvp_on) {
149	m_freem(m);
150	return;
151    }
152
153    if (ip_rsvpd != NULL) {
154	if (rsvpdebug)
155	    printf("rsvp_input: Sending packet up old-style socket\n");
156	rip_input(m, iphlen);
157	return;
158    }
159    /* Drop the packet */
160    m_freem(m);
161}
162
163void ipip_input(struct mbuf *m, int iphlen) { /* XXX must fixup manually */
164	rip_input(m, iphlen);
165}
166
167int (*legal_vif_num)(int) = 0;
168
169/*
170 * This should never be called, since IP_MULTICAST_VIF should fail, but
171 * just in case it does get called, the code a little lower in ip_output
172 * will assign the packet a local address.
173 */
174u_int32_t
175_ip_mcast_src(int vifi) { return INADDR_ANY; }
176u_int32_t (*ip_mcast_src)(int) = _ip_mcast_src;
177
178int
179ip_rsvp_vif_init(so, sopt)
180    struct socket *so;
181    struct sockopt *sopt;
182{
183    return(EINVAL);
184}
185
186int
187ip_rsvp_vif_done(so, sopt)
188    struct socket *so;
189    struct sockopt *sopt;
190{
191    return(EINVAL);
192}
193
194void
195ip_rsvp_force_done(so)
196    struct socket *so;
197{
198    return;
199}
200
201#else /* MROUTING */
202
203#define M_HASCL(m)	((m)->m_flags & M_EXT)
204
205#define INSIZ		sizeof(struct in_addr)
206#define	same(a1, a2) \
207	(bcmp((caddr_t)(a1), (caddr_t)(a2), INSIZ) == 0)
208
209
210/*
211 * Globals.  All but ip_mrouter and ip_mrtproto could be static,
212 * except for netstat or debugging purposes.
213 */
214#ifndef MROUTE_LKM
215struct socket  *ip_mrouter  = NULL;
216static struct mrtstat	mrtstat;
217#else /* MROUTE_LKM */
218extern void	X_ipip_input(struct mbuf *m, int iphlen);
219extern struct mrtstat mrtstat;
220static int ip_mrtproto;
221#endif
222
223#define NO_RTE_FOUND 	0x1
224#define RTE_FOUND	0x2
225
226static struct mfc	*mfctable[CONFIG_MFCTBLSIZ];
227static u_char		nexpire[CONFIG_MFCTBLSIZ];
228static struct vif	viftable[CONFIG_MAXVIFS];
229static u_int	mrtdebug = 0;	  /* debug level 	*/
230#define		DEBUG_MFC	0x02
231#define		DEBUG_FORWARD	0x04
232#define		DEBUG_EXPIRE	0x08
233#define		DEBUG_XMIT	0x10
234static u_int  	tbfdebug = 0;     /* tbf debug level 	*/
235static u_int	rsvpdebug = 0;	  /* rsvp debug level   */
236
237#define		EXPIRE_TIMEOUT	(hz / 4)	/* 4x / second		*/
238#define		UPCALL_EXPIRE	6		/* number of timeouts	*/
239
240/*
241 * Define the token bucket filter structures
242 * tbftable -> each vif has one of these for storing info
243 */
244
245static struct tbf tbftable[CONFIG_MAXVIFS];
246#define		TBF_REPROCESS	(hz / 100)	/* 100x / second */
247
248/*
249 * 'Interfaces' associated with decapsulator (so we can tell
250 * packets that went through it from ones that get reflected
251 * by a broken gateway).  These interfaces are never linked into
252 * the system ifnet list & no routes point to them.  I.e., packets
253 * can't be sent this way.  They only exist as a placeholder for
254 * multicast source verification.
255 */
256static struct ifnet multicast_decap_if[CONFIG_MAXVIFS];
257
258#define ENCAP_TTL 64
259#define ENCAP_PROTO IPPROTO_IPIP	/* 4 */
260
261/* prototype IP hdr for encapsulated packets */
262static struct ip multicast_encap_iphdr = {
263#if BYTE_ORDER == LITTLE_ENDIAN
264	sizeof(struct ip) >> 2, IPVERSION,
265#else
266	IPVERSION, sizeof(struct ip) >> 2,
267#endif
268	0,				/* tos */
269	sizeof(struct ip),		/* total length */
270	0,				/* id */
271	0,				/* frag offset */
272	ENCAP_TTL, ENCAP_PROTO,
273	0,				/* checksum */
274	{ 0 }, { 0 }
275};
276
277/*
278 * Private variables.
279 */
280static vifi_t	   numvifs = 0;
281static int have_encap_tunnel = 0;
282
283/*
284 * one-back cache used by ipip_input to locate a tunnel's vif
285 * given a datagram's src ip address.
286 */
287static u_int32_t last_encap_src;
288static struct vif *last_encap_vif;
289
290static u_int32_t	X_ip_mcast_src(int vifi);
291static int	X_ip_mforward(struct ip *ip, struct ifnet *ifp, struct mbuf *m, struct ip_moptions *imo);
292static int	X_ip_mrouter_done(void);
293static int	X_ip_mrouter_get(struct socket *so, struct sockopt *m);
294static int	X_ip_mrouter_set(struct socket *so, struct sockopt *m);
295static int	X_legal_vif_num(int vif);
296static int	X_mrt_ioctl(u_long cmd, caddr_t data);
297
298static int get_sg_cnt(struct sioc_sg_req *);
299static int get_vif_cnt(struct sioc_vif_req *);
300static int ip_mrouter_init(struct socket *, int);
301static int add_vif(struct vifctl *);
302static int del_vif(vifi_t);
303static int add_mfc(struct mfcctl *);
304static int del_mfc(struct mfcctl *);
305static int socket_send(struct socket *, struct mbuf *, struct sockaddr_in *);
306static int set_assert(int);
307static void expire_upcalls(void *);
308static int ip_mdq(struct mbuf *, struct ifnet *, struct mfc *,
309		  vifi_t);
310static void phyint_send(struct ip *, struct vif *, struct mbuf *);
311static void encap_send(struct ip *, struct vif *, struct mbuf *);
312static void tbf_control(struct vif *, struct mbuf *, struct ip *, u_int32_t);
313static void tbf_queue(struct vif *, struct mbuf *);
314static void tbf_process_q(struct vif *);
315static void tbf_reprocess_q(void *);
316static int tbf_dq_sel(struct vif *, struct ip *);
317static void tbf_send_packet(struct vif *, struct mbuf *);
318static void tbf_update_tokens(struct vif *);
319static int priority(struct vif *, struct ip *);
320void multiencap_decap(struct mbuf *);
321
322/*
323 * whether or not special PIM assert processing is enabled.
324 */
325static int pim_assert;
326/*
327 * Rate limit for assert notification messages, in usec
328 */
329#define ASSERT_MSG_TIME		3000000
330
331/*
332 * Hash function for a source, group entry
333 */
334#define MFCHASH(a, g) MFCHASHMOD(((a) >> 20) ^ ((a) >> 10) ^ (a) ^ \
335			((g) >> 20) ^ ((g) >> 10) ^ (g))
336
337/*
338 * Find a route for a given origin IP address and Multicast group address
339 * Type of service parameter to be added in the future!!!
340 */
341
342#define MFCFIND(o, g, rt) { \
343	struct mfc *_rt = mfctable[MFCHASH(o,g)]; \
344	rt = NULL; \
345	++mrtstat.mrts_mfc_lookups; \
346	while (_rt) { \
347		if ((_rt->mfc_origin.s_addr == o) && \
348		    (_rt->mfc_mcastgrp.s_addr == g) && \
349		    (_rt->mfc_stall == NULL)) { \
350			rt = _rt; \
351			break; \
352		} \
353		_rt = _rt->mfc_next; \
354	} \
355	if (rt == NULL) { \
356		++mrtstat.mrts_mfc_misses; \
357	} \
358}
359
360
361/*
362 * Macros to compute elapsed time efficiently
363 * Borrowed from Van Jacobson's scheduling code
364 */
365#define TV_DELTA(a, b, delta) { \
366	    int xxs; \
367		\
368	    delta = (a).tv_usec - (b).tv_usec; \
369	    if ((xxs = (a).tv_sec - (b).tv_sec)) { \
370	       switch (xxs) { \
371		      case 2: \
372			  delta += 1000000; \
373			      /* fall through */ \
374		      case 1: \
375			  delta += 1000000; \
376			  break; \
377		      default: \
378			  delta += (1000000 * xxs); \
379	       } \
380	    } \
381}
382
383#define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
384	      (a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
385
386#if UPCALL_TIMING
387u_int32_t upcall_data[51];
388static void collate(struct timeval *);
389#endif /* UPCALL_TIMING */
390
391
392/*
393 * Handle MRT setsockopt commands to modify the multicast routing tables.
394 */
395static int
396X_ip_mrouter_set(struct socket *so, struct sockopt *sopt)
397{
398	int	error, optval;
399	vifi_t	vifi;
400	struct	vifctl vifc;
401	struct	mfcctl mfc;
402
403	if (so != ip_mrouter && sopt->sopt_name != MRT_INIT)
404		return (EPERM);
405
406	error = 0;
407	switch (sopt->sopt_name) {
408	case MRT_INIT:
409		error = sooptcopyin(sopt, &optval, sizeof optval,
410				    sizeof optval);
411		if (error)
412			break;
413		error = ip_mrouter_init(so, optval);
414		break;
415
416	case MRT_DONE:
417		error = ip_mrouter_done();
418		break;
419
420	case MRT_ADD_VIF:
421		error = sooptcopyin(sopt, &vifc, sizeof vifc, sizeof vifc);
422		if (error)
423			break;
424		error = add_vif(&vifc);
425		break;
426
427	case MRT_DEL_VIF:
428		error = sooptcopyin(sopt, &vifi, sizeof vifi, sizeof vifi);
429		if (error)
430			break;
431		error = del_vif(vifi);
432		break;
433
434	case MRT_ADD_MFC:
435	case MRT_DEL_MFC:
436		error = sooptcopyin(sopt, &mfc, sizeof mfc, sizeof mfc);
437		if (error)
438			break;
439		if (sopt->sopt_name == MRT_ADD_MFC)
440			error = add_mfc(&mfc);
441		else
442			error = del_mfc(&mfc);
443		break;
444
445	case MRT_ASSERT:
446		error = sooptcopyin(sopt, &optval, sizeof optval,
447				    sizeof optval);
448		if (error)
449			break;
450		set_assert(optval);
451		break;
452
453	default:
454		error = EOPNOTSUPP;
455		break;
456	}
457	return (error);
458}
459
460#if !defined(MROUTE_LKM) || !MROUTE_LKM
461int (*ip_mrouter_set)(struct socket *, struct sockopt *) = X_ip_mrouter_set;
462#endif
463
464/*
465 * Handle MRT getsockopt commands
466 */
467static int
468X_ip_mrouter_get(__unused struct socket *so, struct sockopt *sopt)
469{
470	int error;
471	static int vers = 0x0305; /* !!! why is this here? XXX */
472
473	switch (sopt->sopt_name) {
474	case MRT_VERSION:
475		error = sooptcopyout(sopt, &vers, sizeof vers);
476		break;
477
478	case MRT_ASSERT:
479		error = sooptcopyout(sopt, &pim_assert, sizeof pim_assert);
480		break;
481	default:
482		error = EOPNOTSUPP;
483		break;
484	}
485	return (error);
486}
487
488#if !defined(MROUTE_LKM) || !MROUTE_LKM
489int (*ip_mrouter_get)(struct socket *, struct sockopt *) = X_ip_mrouter_get;
490#endif
491
492/*
493 * Handle ioctl commands to obtain information from the cache
494 */
495static int
496X_mrt_ioctl(u_long cmd, caddr_t data)
497{
498    int error = 0;
499
500    switch (cmd) {
501	case (SIOCGETVIFCNT):
502	    return (get_vif_cnt((struct sioc_vif_req *)data));
503	    break;
504	case (SIOCGETSGCNT):
505	    return (get_sg_cnt((struct sioc_sg_req *)data));
506	    break;
507	default:
508	    return (EINVAL);
509	    break;
510    }
511    return error;
512}
513
514#if !defined(MROUTE_LKM) || !MROUTE_LKM
515int (*mrt_ioctl)(u_long, caddr_t) = X_mrt_ioctl;
516#endif
517
518/*
519 * returns the packet, byte, rpf-failure count for the source group provided
520 */
521static int
522get_sg_cnt(struct sioc_sg_req *req)
523{
524    struct mfc *rt;
525
526    MFCFIND(req->src.s_addr, req->grp.s_addr, rt);
527    if (rt != NULL) {
528	req->pktcnt = rt->mfc_pkt_cnt;
529	req->bytecnt = rt->mfc_byte_cnt;
530	req->wrong_if = rt->mfc_wrong_if;
531    } else
532	req->pktcnt = req->bytecnt = req->wrong_if = 0xffffffff;
533
534    return 0;
535}
536
537/*
538 * returns the input and output packet and byte counts on the vif provided
539 */
540static int
541get_vif_cnt(struct sioc_vif_req *req)
542{
543    vifi_t vifi = req->vifi;
544
545    if (vifi >= numvifs) return EINVAL;
546
547    req->icount = viftable[vifi].v_pkt_in;
548    req->ocount = viftable[vifi].v_pkt_out;
549    req->ibytes = viftable[vifi].v_bytes_in;
550    req->obytes = viftable[vifi].v_bytes_out;
551
552    return 0;
553}
554
555/*
556 * Enable multicast routing
557 */
558static int
559ip_mrouter_init(struct socket *so, int vers)
560{
561    if (mrtdebug)
562	log(LOG_DEBUG,"ip_mrouter_init: so_type = %d, pr_protocol = %d\n",
563		so->so_type, so->so_proto->pr_protocol);
564
565    if (so->so_type != SOCK_RAW ||
566	so->so_proto->pr_protocol != IPPROTO_IGMP) return EOPNOTSUPP;
567
568    if (vers != 1)
569	return ENOPROTOOPT;
570
571    if (ip_mrouter != NULL) return EADDRINUSE;
572
573    ip_mrouter = so;
574
575    bzero((caddr_t)mfctable, sizeof(mfctable));
576    bzero((caddr_t)nexpire, sizeof(nexpire));
577
578    pim_assert = 0;
579
580    timeout(expire_upcalls, (caddr_t)NULL, EXPIRE_TIMEOUT);
581
582    if (mrtdebug)
583	log(LOG_DEBUG, "ip_mrouter_init\n");
584
585    return 0;
586}
587
588/*
589 * Disable multicast routing
590 */
591static int
592X_ip_mrouter_done(void)
593{
594    vifi_t vifi;
595    int i;
596    struct ifnet *ifp;
597    struct ifreq ifr;
598    struct mfc *rt;
599    struct rtdetq *rte;
600
601    /*
602     * For each phyint in use, disable promiscuous reception of all IP
603     * multicasts.
604     */
605    for (vifi = 0; vifi < numvifs; vifi++) {
606	if (viftable[vifi].v_lcl_addr.s_addr != 0 &&
607	    !(viftable[vifi].v_flags & VIFF_TUNNEL)) {
608	    ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_family = AF_INET;
609	    ((struct sockaddr_in *)&(ifr.ifr_addr))->sin_addr.s_addr
610								= INADDR_ANY;
611	    ifp = viftable[vifi].v_ifp;
612	    if_allmulti(ifp, 0);
613	}
614    }
615    bzero((caddr_t)tbftable, sizeof(tbftable));
616    bzero((caddr_t)viftable, sizeof(viftable));
617    numvifs = 0;
618    pim_assert = 0;
619
620    untimeout(expire_upcalls, (caddr_t)NULL);
621
622    /*
623     * Free all multicast forwarding cache entries.
624     */
625    for (i = 0; i < CONFIG_MFCTBLSIZ; i++) {
626	for (rt = mfctable[i]; rt != NULL; ) {
627	    struct mfc *nr = rt->mfc_next;
628
629	    for (rte = rt->mfc_stall; rte != NULL; ) {
630		struct rtdetq *n = rte->next;
631
632		m_freem(rte->m);
633		FREE(rte, M_MRTABLE);
634		rte = n;
635	    }
636	    FREE(rt, M_MRTABLE);
637	    rt = nr;
638	}
639    }
640
641    bzero((caddr_t)mfctable, sizeof(mfctable));
642
643    /*
644     * Reset de-encapsulation cache
645     */
646    last_encap_src = 0;
647    last_encap_vif = NULL;
648    have_encap_tunnel = 0;
649
650    ip_mrouter = NULL;
651
652    if (mrtdebug)
653	log(LOG_DEBUG, "ip_mrouter_done\n");
654
655    return 0;
656}
657
658#if !defined(MROUTE_LKM) || !MROUTE_LKM
659int (*ip_mrouter_done)(void) = X_ip_mrouter_done;
660#endif
661
662/*
663 * Set PIM assert processing global
664 */
665static int
666set_assert(int i)
667{
668    if ((i != 1) && (i != 0))
669	return EINVAL;
670
671    pim_assert = i;
672
673    return 0;
674}
675
676/*
677 * Add a vif to the vif table
678 */
679static int
680add_vif(struct vifctl *vifcp)
681{
682    struct vif *vifp = viftable + vifcp->vifc_vifi;
683    static struct sockaddr_in sin = { sizeof sin, AF_INET,
684										0 , {0}, {0,0,0,0,0,0,0,0,} };
685    struct ifaddr *ifa;
686    struct ifnet *ifp;
687    int error, s;
688    struct tbf *v_tbf = tbftable + vifcp->vifc_vifi;
689
690    if (vifcp->vifc_vifi >= CONFIG_MAXVIFS)  return EINVAL;
691    if (vifp->v_lcl_addr.s_addr != 0) return EADDRINUSE;
692
693    /* Find the interface with an address in AF_INET family */
694    sin.sin_addr = vifcp->vifc_lcl_addr;
695    ifa = ifa_ifwithaddr((struct sockaddr *)&sin);
696    if (ifa == 0) return EADDRNOTAVAIL;
697    ifp = ifa->ifa_ifp;
698    IFA_REMREF(ifa);
699    ifa = NULL;
700
701    if (vifcp->vifc_flags & VIFF_TUNNEL) {
702	if ((vifcp->vifc_flags & VIFF_SRCRT) == 0) {
703		/*
704		 * An encapsulating tunnel is wanted.  Tell ipip_input() to
705		 * start paying attention to encapsulated packets.
706		 */
707		if (have_encap_tunnel == 0) {
708			have_encap_tunnel = 1;
709			for (s = 0; s < CONFIG_MAXVIFS; ++s) {
710				multicast_decap_if[s].if_name = "mdecap";
711				multicast_decap_if[s].if_unit = s;
712				multicast_decap_if[s].if_family = APPLE_IF_FAM_MDECAP;
713			}
714		}
715		/*
716		 * Set interface to fake encapsulator interface
717		 */
718		ifp = &multicast_decap_if[vifcp->vifc_vifi];
719	} else {
720	    log(LOG_ERR, "source routed tunnels not supported\n");
721	    return EOPNOTSUPP;
722	}
723    } else {
724	/* Make sure the interface supports multicast */
725	if ((ifp->if_flags & IFF_MULTICAST) == 0)
726	    return EOPNOTSUPP;
727
728	/* Enable promiscuous reception of all IP multicasts from the if */
729	error = if_allmulti(ifp, 1);
730	if (error)
731	    return error;
732    }
733
734    /* define parameters for the tbf structure */
735    vifp->v_tbf = v_tbf;
736    GET_TIME(vifp->v_tbf->tbf_last_pkt_t);
737    vifp->v_tbf->tbf_n_tok = 0;
738    vifp->v_tbf->tbf_q_len = 0;
739    vifp->v_tbf->tbf_max_q_len = MAXQSIZE;
740    vifp->v_tbf->tbf_q = vifp->v_tbf->tbf_t = NULL;
741
742    vifp->v_flags     = vifcp->vifc_flags;
743    vifp->v_threshold = vifcp->vifc_threshold;
744    vifp->v_lcl_addr  = vifcp->vifc_lcl_addr;
745    vifp->v_rmt_addr  = vifcp->vifc_rmt_addr;
746    vifp->v_ifp       = ifp;
747    /* scaling up here allows division by 1024 in critical code */
748    vifp->v_rate_limit= vifcp->vifc_rate_limit * 1024 / 1000;
749    vifp->v_rsvp_on   = 0;
750    vifp->v_rsvpd     = NULL;
751    /* initialize per vif pkt counters */
752    vifp->v_pkt_in    = 0;
753    vifp->v_pkt_out   = 0;
754    vifp->v_bytes_in  = 0;
755    vifp->v_bytes_out = 0;
756
757    /* Adjust numvifs up if the vifi is higher than numvifs */
758    if (numvifs <= vifcp->vifc_vifi) numvifs = vifcp->vifc_vifi + 1;
759
760    if (mrtdebug)
761	log(LOG_DEBUG, "add_vif #%d, lcladdr %lx, %s %lx, thresh %x, rate %d\n",
762	    vifcp->vifc_vifi,
763	    (u_int32_t)ntohl(vifcp->vifc_lcl_addr.s_addr),
764	    (vifcp->vifc_flags & VIFF_TUNNEL) ? "rmtaddr" : "mask",
765	    (u_int32_t)ntohl(vifcp->vifc_rmt_addr.s_addr),
766	    vifcp->vifc_threshold,
767	    vifcp->vifc_rate_limit);
768
769    return 0;
770}
771
772/*
773 * Delete a vif from the vif table
774 */
775static int
776del_vif(vifi_t vifi)
777{
778    struct vif *vifp = &viftable[vifi];
779    struct mbuf *m;
780    struct ifnet *ifp;
781    struct ifreq ifr;
782
783    if (vifi >= numvifs) return EINVAL;
784    if (vifp->v_lcl_addr.s_addr == 0) return EADDRNOTAVAIL;
785
786    if (!(vifp->v_flags & VIFF_TUNNEL)) {
787	((struct sockaddr_in *)&(ifr.ifr_addr))->sin_family = AF_INET;
788	((struct sockaddr_in *)&(ifr.ifr_addr))->sin_addr.s_addr = INADDR_ANY;
789	ifp = vifp->v_ifp;
790	if_allmulti(ifp, 0);
791    }
792
793    if (vifp == last_encap_vif) {
794	last_encap_vif = 0;
795	last_encap_src = 0;
796    }
797
798    /*
799     * Free packets queued at the interface
800     */
801    while (vifp->v_tbf->tbf_q) {
802	m = vifp->v_tbf->tbf_q;
803	vifp->v_tbf->tbf_q = m->m_act;
804	m_freem(m);
805    }
806
807    bzero((caddr_t)vifp->v_tbf, sizeof(*(vifp->v_tbf)));
808    bzero((caddr_t)vifp, sizeof (*vifp));
809
810    if (mrtdebug)
811      log(LOG_DEBUG, "del_vif %d, numvifs %d\n", vifi, numvifs);
812
813    /* Adjust numvifs down */
814    for (vifi = numvifs; vifi > 0; vifi--)
815	if (viftable[vifi-1].v_lcl_addr.s_addr != 0) break;
816    numvifs = vifi;
817
818    return 0;
819}
820
821/*
822 * Add an mfc entry
823 */
824static int
825add_mfc(struct mfcctl *mfccp)
826{
827    struct mfc *rt;
828    u_int32_t hash;
829    struct rtdetq *rte;
830    u_short nstl;
831    int i;
832
833    MFCFIND(mfccp->mfcc_origin.s_addr, mfccp->mfcc_mcastgrp.s_addr, rt);
834
835    /* If an entry already exists, just update the fields */
836    if (rt) {
837	if (mrtdebug & DEBUG_MFC)
838	    log(LOG_DEBUG,"add_mfc update o %lx g %lx p %x\n",
839		(u_int32_t)ntohl(mfccp->mfcc_origin.s_addr),
840		(u_int32_t)ntohl(mfccp->mfcc_mcastgrp.s_addr),
841		mfccp->mfcc_parent);
842
843	rt->mfc_parent = mfccp->mfcc_parent;
844	for (i = 0; i < numvifs; i++)
845	    rt->mfc_ttls[i] = mfccp->mfcc_ttls[i];
846	return 0;
847    }
848
849    /*
850     * Find the entry for which the upcall was made and update
851     */
852    hash = MFCHASH(mfccp->mfcc_origin.s_addr, mfccp->mfcc_mcastgrp.s_addr);
853    for (rt = mfctable[hash], nstl = 0; rt; rt = rt->mfc_next) {
854
855	if ((rt->mfc_origin.s_addr == mfccp->mfcc_origin.s_addr) &&
856	    (rt->mfc_mcastgrp.s_addr == mfccp->mfcc_mcastgrp.s_addr) &&
857	    (rt->mfc_stall != NULL)) {
858
859	    if (nstl++)
860		log(LOG_ERR, "add_mfc %s o %lx g %lx p %x dbx %p\n",
861		    "multiple kernel entries",
862		    (u_int32_t)ntohl(mfccp->mfcc_origin.s_addr),
863		    (u_int32_t)ntohl(mfccp->mfcc_mcastgrp.s_addr),
864		    mfccp->mfcc_parent, (void *)rt->mfc_stall);
865
866	    if (mrtdebug & DEBUG_MFC)
867		log(LOG_DEBUG,"add_mfc o %lx g %lx p %x dbg %p\n",
868		    (u_int32_t)ntohl(mfccp->mfcc_origin.s_addr),
869		    (u_int32_t)ntohl(mfccp->mfcc_mcastgrp.s_addr),
870		    mfccp->mfcc_parent, (void *)rt->mfc_stall);
871
872	    rt->mfc_origin     = mfccp->mfcc_origin;
873	    rt->mfc_mcastgrp   = mfccp->mfcc_mcastgrp;
874	    rt->mfc_parent     = mfccp->mfcc_parent;
875	    for (i = 0; i < numvifs; i++)
876		rt->mfc_ttls[i] = mfccp->mfcc_ttls[i];
877	    /* initialize pkt counters per src-grp */
878	    rt->mfc_pkt_cnt    = 0;
879	    rt->mfc_byte_cnt   = 0;
880	    rt->mfc_wrong_if   = 0;
881	    rt->mfc_last_assert.tv_sec = rt->mfc_last_assert.tv_usec = 0;
882
883	    rt->mfc_expire = 0;	/* Don't clean this guy up */
884	    nexpire[hash]--;
885
886	    /* free packets Qed at the end of this entry */
887	    for (rte = rt->mfc_stall; rte != NULL; ) {
888		struct rtdetq *n = rte->next;
889
890		ip_mdq(rte->m, rte->ifp, rt, -1);
891		m_freem(rte->m);
892#if UPCALL_TIMING
893		collate(&(rte->t));
894#endif /* UPCALL_TIMING */
895		FREE(rte, M_MRTABLE);
896		rte = n;
897	    }
898	    rt->mfc_stall = NULL;
899	}
900    }
901
902    /*
903     * It is possible that an entry is being inserted without an upcall
904     */
905    if (nstl == 0) {
906	if (mrtdebug & DEBUG_MFC)
907	    log(LOG_DEBUG,"add_mfc no upcall h %lu o %lx g %lx p %x\n",
908		hash, (u_int32_t)ntohl(mfccp->mfcc_origin.s_addr),
909		(u_int32_t)ntohl(mfccp->mfcc_mcastgrp.s_addr),
910		mfccp->mfcc_parent);
911
912	for (rt = mfctable[hash]; rt != NULL; rt = rt->mfc_next) {
913
914	    if ((rt->mfc_origin.s_addr == mfccp->mfcc_origin.s_addr) &&
915		(rt->mfc_mcastgrp.s_addr == mfccp->mfcc_mcastgrp.s_addr)) {
916
917		rt->mfc_origin     = mfccp->mfcc_origin;
918		rt->mfc_mcastgrp   = mfccp->mfcc_mcastgrp;
919		rt->mfc_parent     = mfccp->mfcc_parent;
920		for (i = 0; i < numvifs; i++)
921		    rt->mfc_ttls[i] = mfccp->mfcc_ttls[i];
922		/* initialize pkt counters per src-grp */
923		rt->mfc_pkt_cnt    = 0;
924		rt->mfc_byte_cnt   = 0;
925		rt->mfc_wrong_if   = 0;
926		rt->mfc_last_assert.tv_sec = rt->mfc_last_assert.tv_usec = 0;
927		if (rt->mfc_expire)
928		    nexpire[hash]--;
929		rt->mfc_expire	   = 0;
930	    }
931	}
932	if (rt == NULL) {
933	    /* no upcall, so make a new entry */
934	    rt = (struct mfc *) _MALLOC(sizeof(*rt), M_MRTABLE, M_NOWAIT);
935	    if (rt == NULL) {
936		return ENOBUFS;
937	    }
938
939	    /* insert new entry at head of hash chain */
940	    rt->mfc_origin     = mfccp->mfcc_origin;
941	    rt->mfc_mcastgrp   = mfccp->mfcc_mcastgrp;
942	    rt->mfc_parent     = mfccp->mfcc_parent;
943	    for (i = 0; i < numvifs; i++)
944		    rt->mfc_ttls[i] = mfccp->mfcc_ttls[i];
945	    /* initialize pkt counters per src-grp */
946	    rt->mfc_pkt_cnt    = 0;
947	    rt->mfc_byte_cnt   = 0;
948	    rt->mfc_wrong_if   = 0;
949	    rt->mfc_last_assert.tv_sec = rt->mfc_last_assert.tv_usec = 0;
950	    rt->mfc_expire     = 0;
951	    rt->mfc_stall      = NULL;
952
953	    /* link into table */
954	    rt->mfc_next = mfctable[hash];
955	    mfctable[hash] = rt;
956	}
957    }
958    return 0;
959}
960
961#if UPCALL_TIMING
962/*
963 * collect delay statistics on the upcalls
964 */
965static void
966collate(struct timeval *t)
967{
968    u_int32_t d;
969    struct timeval tp;
970    u_int32_t delta;
971
972    GET_TIME(tp);
973
974    if (TV_LT(*t, tp))
975    {
976	TV_DELTA(tp, *t, delta);
977
978	d = delta >> 10;
979	if (d > 50)
980	    d = 50;
981
982	++upcall_data[d];
983    }
984}
985#endif /* UPCALL_TIMING */
986
987/*
988 * Delete an mfc entry
989 */
990static int
991del_mfc(struct mfcctl *mfccp)
992{
993    struct in_addr 	origin;
994    struct in_addr 	mcastgrp;
995    struct mfc 		*rt;
996    struct mfc	 	**nptr;
997    u_int32_t 		hash;
998
999    origin = mfccp->mfcc_origin;
1000    mcastgrp = mfccp->mfcc_mcastgrp;
1001    hash = MFCHASH(origin.s_addr, mcastgrp.s_addr);
1002
1003    if (mrtdebug & DEBUG_MFC)
1004	log(LOG_DEBUG,"del_mfc orig %lx mcastgrp %lx\n",
1005	    (u_int32_t)ntohl(origin.s_addr), (u_int32_t)ntohl(mcastgrp.s_addr));
1006
1007    nptr = &mfctable[hash];
1008    while ((rt = *nptr) != NULL) {
1009	if (origin.s_addr == rt->mfc_origin.s_addr &&
1010	    mcastgrp.s_addr == rt->mfc_mcastgrp.s_addr &&
1011	    rt->mfc_stall == NULL)
1012	    break;
1013
1014	nptr = &rt->mfc_next;
1015    }
1016    if (rt == NULL) {
1017	return EADDRNOTAVAIL;
1018    }
1019
1020    *nptr = rt->mfc_next;
1021    FREE(rt, M_MRTABLE);
1022
1023    return 0;
1024}
1025
1026/*
1027 * Send a message to mrouted on the multicast routing socket
1028 */
1029static int
1030socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in *src)
1031{
1032	socket_lock(s, 1);
1033	if (s) {
1034		if (sbappendaddr(&s->so_rcv,
1035				 (struct sockaddr *)src,
1036				 mm, (struct mbuf *)0, NULL) != 0) {
1037			sorwakeup(s);
1038			socket_unlock(s, 1);
1039			return 0;
1040		}
1041	}
1042	socket_unlock(s, 1);
1043	m_freem(mm);
1044	return -1;
1045}
1046
1047/*
1048 * IP multicast forwarding function. This function assumes that the packet
1049 * pointed to by "ip" has arrived on (or is about to be sent to) the interface
1050 * pointed to by "ifp", and the packet is to be relayed to other networks
1051 * that have members of the packet's destination IP multicast group.
1052 *
1053 * The packet is returned unscathed to the caller, unless it is
1054 * erroneous, in which case a non-zero return value tells the caller to
1055 * discard it.
1056 */
1057
1058#define IP_HDR_LEN  20	/* # bytes of fixed IP header (excluding options) */
1059#define TUNNEL_LEN  12  /* # bytes of IP option for tunnel encapsulation  */
1060
1061static int
1062X_ip_mforward(struct ip *ip, struct ifnet *ifp, struct mbuf *m,
1063	      struct ip_moptions *imo)
1064{
1065    struct mfc *rt;
1066    u_char *ipoptions;
1067    static struct sockaddr_in 	k_igmpsrc = { sizeof k_igmpsrc, AF_INET,
1068    											0 , {0}, {0,0,0,0,0,0,0,0,} };
1069    static int srctun = 0;
1070    struct mbuf *mm;
1071    vifi_t vifi;
1072    struct vif *vifp;
1073
1074    if (mrtdebug & DEBUG_FORWARD)
1075	log(LOG_DEBUG, "ip_mforward: src %lx, dst %lx, ifp %p\n",
1076	    (u_int32_t)ntohl(ip->ip_src.s_addr), (u_int32_t)ntohl(ip->ip_dst.s_addr),
1077	    (void *)ifp);
1078
1079    if (ip->ip_hl < (IP_HDR_LEN + TUNNEL_LEN) >> 2 ||
1080	(ipoptions = (u_char *)(ip + 1))[1] != IPOPT_LSRR ) {
1081	/*
1082	 * Packet arrived via a physical interface or
1083	 * an encapsulated tunnel.
1084	 */
1085    } else {
1086	/*
1087	 * Packet arrived through a source-route tunnel.
1088	 * Source-route tunnels are no longer supported.
1089	 */
1090	if ((srctun++ % 1000) == 0)
1091	    log(LOG_ERR,
1092		"ip_mforward: received source-routed packet from %lx\n",
1093		(u_int32_t)ntohl(ip->ip_src.s_addr));
1094
1095	return 1;
1096    }
1097
1098    if (imo != NULL)
1099	IMO_LOCK(imo);
1100    if ((imo) && ((vifi = imo->imo_multicast_vif) < numvifs)) {
1101	IMO_UNLOCK(imo);
1102	if (ip->ip_ttl < 255)
1103		ip->ip_ttl++;	/* compensate for -1 in *_send routines */
1104	if (rsvpdebug && ip->ip_p == IPPROTO_RSVP) {
1105	    vifp = viftable + vifi;
1106	    printf("Sending IPPROTO_RSVP from %x to %x on vif %d (%s%s)\n",
1107		ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr), vifi,
1108		(vifp->v_flags & VIFF_TUNNEL) ? "tunnel on " : "",
1109		if_name(vifp->v_ifp));
1110	}
1111	return (ip_mdq(m, ifp, NULL, vifi));
1112    } else if (imo != NULL) {
1113	IMO_UNLOCK(imo);
1114    }
1115    if (rsvpdebug && ip->ip_p == IPPROTO_RSVP) {
1116	printf("Warning: IPPROTO_RSVP from %x to %x without vif option\n",
1117	    ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr));
1118	if(!imo)
1119		printf("In fact, no options were specified at all\n");
1120    }
1121
1122    /*
1123     * Don't forward a packet with time-to-live of zero or one,
1124     * or a packet destined to a local-only group.
1125     */
1126    if (ip->ip_ttl <= 1 ||
1127	ntohl(ip->ip_dst.s_addr) <= INADDR_MAX_LOCAL_GROUP)
1128	return 0;
1129
1130    /*
1131     * Determine forwarding vifs from the forwarding cache table
1132     */
1133    MFCFIND(ip->ip_src.s_addr, ip->ip_dst.s_addr, rt);
1134
1135    /* Entry exists, so forward if necessary */
1136    if (rt != NULL) {
1137	return (ip_mdq(m, ifp, rt, -1));
1138    } else {
1139	/*
1140	 * If we don't have a route for packet's origin,
1141	 * Make a copy of the packet &
1142	 * send message to routing daemon
1143	 */
1144
1145	struct mbuf *mb0;
1146	struct rtdetq *rte;
1147	u_int32_t hash;
1148	int hlen = ip->ip_hl << 2;
1149#if UPCALL_TIMING
1150	struct timeval tp;
1151
1152	GET_TIME(tp);
1153#endif
1154
1155	mrtstat.mrts_no_route++;
1156	if (mrtdebug & (DEBUG_FORWARD | DEBUG_MFC))
1157	    log(LOG_DEBUG, "ip_mforward: no rte s %lx g %lx\n",
1158		(u_int32_t)ntohl(ip->ip_src.s_addr),
1159		(u_int32_t)ntohl(ip->ip_dst.s_addr));
1160
1161	/*
1162	 * Allocate mbufs early so that we don't do extra work if we are
1163	 * just going to fail anyway.  Make sure to pullup the header so
1164	 * that other people can't step on it.
1165	 */
1166	rte = (struct rtdetq *) _MALLOC((sizeof *rte), M_MRTABLE, M_NOWAIT);
1167	if (rte == NULL) {
1168	    return ENOBUFS;
1169	}
1170	mb0 = m_copy(m, 0, M_COPYALL);
1171	if (mb0 && (M_HASCL(mb0) || mb0->m_len < hlen))
1172	    mb0 = m_pullup(mb0, hlen);
1173	if (mb0 == NULL) {
1174	    FREE(rte, M_MRTABLE);
1175	    return ENOBUFS;
1176	}
1177
1178	/* is there an upcall waiting for this packet? */
1179	hash = MFCHASH(ip->ip_src.s_addr, ip->ip_dst.s_addr);
1180	for (rt = mfctable[hash]; rt; rt = rt->mfc_next) {
1181	    if ((ip->ip_src.s_addr == rt->mfc_origin.s_addr) &&
1182		(ip->ip_dst.s_addr == rt->mfc_mcastgrp.s_addr) &&
1183		(rt->mfc_stall != NULL))
1184		break;
1185	}
1186
1187	if (rt == NULL) {
1188	    int i;
1189	    struct igmpmsg *im;
1190
1191	    /* no upcall, so make a new entry */
1192	    rt = (struct mfc *) _MALLOC(sizeof(*rt), M_MRTABLE, M_NOWAIT);
1193	    if (rt == NULL) {
1194		FREE(rte, M_MRTABLE);
1195		m_freem(mb0);
1196		return ENOBUFS;
1197	    }
1198	    /* Make a copy of the header to send to the user level process */
1199	    mm = m_copy(mb0, 0, hlen);
1200	    if (mm == NULL) {
1201		FREE(rte, M_MRTABLE);
1202		m_freem(mb0);
1203		FREE(rt, M_MRTABLE);
1204		return ENOBUFS;
1205	    }
1206
1207	    /*
1208	     * Send message to routing daemon to install
1209	     * a route into the kernel table
1210	     */
1211	    k_igmpsrc.sin_addr = ip->ip_src;
1212
1213	    im = mtod(mm, struct igmpmsg *);
1214	    im->im_msgtype	= IGMPMSG_NOCACHE;
1215	    im->im_mbz		= 0;
1216
1217	    mrtstat.mrts_upcalls++;
1218
1219	    if (socket_send(ip_mrouter, mm, &k_igmpsrc) < 0) {
1220		log(LOG_WARNING, "ip_mforward: ip_mrouter socket queue full\n");
1221		++mrtstat.mrts_upq_sockfull;
1222		FREE(rte, M_MRTABLE);
1223		m_freem(mb0);
1224		FREE(rt, M_MRTABLE);
1225		return ENOBUFS;
1226	    }
1227
1228	    /* insert new entry at head of hash chain */
1229	    rt->mfc_origin.s_addr     = ip->ip_src.s_addr;
1230	    rt->mfc_mcastgrp.s_addr   = ip->ip_dst.s_addr;
1231	    rt->mfc_expire	      = UPCALL_EXPIRE;
1232	    nexpire[hash]++;
1233	    for (i = 0; i < numvifs; i++)
1234		rt->mfc_ttls[i] = 0;
1235	    rt->mfc_parent = -1;
1236
1237	    /* link into table */
1238	    rt->mfc_next   = mfctable[hash];
1239	    mfctable[hash] = rt;
1240	    rt->mfc_stall = rte;
1241
1242	} else {
1243	    /* determine if q has overflowed */
1244	    int npkts = 0;
1245	    struct rtdetq **p;
1246
1247	    for (p = &rt->mfc_stall; *p != NULL; p = &(*p)->next)
1248		npkts++;
1249
1250	    if (npkts > MAX_UPQ) {
1251		mrtstat.mrts_upq_ovflw++;
1252		FREE(rte, M_MRTABLE);
1253		m_freem(mb0);
1254		return 0;
1255	    }
1256
1257	    /* Add this entry to the end of the queue */
1258	    *p = rte;
1259	}
1260
1261	rte->m 			= mb0;
1262	rte->ifp 		= ifp;
1263#if UPCALL_TIMING
1264	rte->t			= tp;
1265#endif
1266	rte->next		= NULL;
1267
1268	return 0;
1269    }
1270}
1271
1272#if !defined(MROUTE_LKM) || !MROUTE_LKM
1273int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
1274		   struct ip_moptions *) = X_ip_mforward;
1275#endif
1276
1277/*
1278 * Clean up the cache entry if upcall is not serviced
1279 */
1280static void
1281expire_upcalls(__unused void *unused)
1282{
1283    struct rtdetq *rte;
1284    struct mfc *mfc, **nptr;
1285    int i;
1286
1287    for (i = 0; i < CONFIG_MFCTBLSIZ; i++) {
1288	if (nexpire[i] == 0)
1289	    continue;
1290	nptr = &mfctable[i];
1291	for (mfc = *nptr; mfc != NULL; mfc = *nptr) {
1292	    /*
1293	     * Skip real cache entries
1294	     * Make sure it wasn't marked to not expire (shouldn't happen)
1295	     * If it expires now
1296	     */
1297	    if (mfc->mfc_stall != NULL &&
1298	        mfc->mfc_expire != 0 &&
1299		--mfc->mfc_expire == 0) {
1300		if (mrtdebug & DEBUG_EXPIRE)
1301		    log(LOG_DEBUG, "expire_upcalls: expiring (%lx %lx)\n",
1302			(u_int32_t)ntohl(mfc->mfc_origin.s_addr),
1303			(u_int32_t)ntohl(mfc->mfc_mcastgrp.s_addr));
1304		/*
1305		 * drop all the packets
1306		 * free the mbuf with the pkt, if, timing info
1307		 */
1308		for (rte = mfc->mfc_stall; rte; ) {
1309		    struct rtdetq *n = rte->next;
1310
1311		    m_freem(rte->m);
1312		    FREE(rte, M_MRTABLE);
1313		    rte = n;
1314		}
1315		++mrtstat.mrts_cache_cleanups;
1316		nexpire[i]--;
1317
1318		*nptr = mfc->mfc_next;
1319		FREE(mfc, M_MRTABLE);
1320	    } else {
1321		nptr = &mfc->mfc_next;
1322	    }
1323	}
1324    }
1325    timeout(expire_upcalls, (caddr_t)NULL, EXPIRE_TIMEOUT);
1326}
1327
1328/*
1329 * Packet forwarding routine once entry in the cache is made
1330 */
1331static int
1332ip_mdq(struct mbuf *m, struct ifnet *ifp, struct mfc *rt,
1333       vifi_t xmt_vif)
1334{
1335    struct ip  *ip = mtod(m, struct ip *);
1336    vifi_t vifi;
1337    struct vif *vifp;
1338    int plen = ip->ip_len;
1339
1340/*
1341 * Macro to send packet on vif.  Since RSVP packets don't get counted on
1342 * input, they shouldn't get counted on output, so statistics keeping is
1343 * seperate.
1344 */
1345#define MC_SEND(ip,vifp,m) {                             \
1346                if ((vifp)->v_flags & VIFF_TUNNEL)  	 \
1347                    encap_send((ip), (vifp), (m));       \
1348                else                                     \
1349                    phyint_send((ip), (vifp), (m));      \
1350}
1351
1352    /*
1353     * If xmt_vif is not -1, send on only the requested vif.
1354     *
1355     * (since vifi_t is u_short, -1 becomes MAXUSHORT, which > numvifs.)
1356     */
1357    if (xmt_vif < numvifs) {
1358	MC_SEND(ip, viftable + xmt_vif, m);
1359	return 1;
1360    }
1361
1362    /*
1363     * Don't forward if it didn't arrive from the parent vif for its origin.
1364     */
1365    vifi = rt->mfc_parent;
1366    if ((vifi >= numvifs) || (viftable[vifi].v_ifp != ifp)) {
1367	/* came in the wrong interface */
1368	if (mrtdebug & DEBUG_FORWARD)
1369	    log(LOG_DEBUG, "wrong if: ifp %p vifi %d vififp %p\n",
1370		(void *)ifp, vifi, (void *)viftable[vifi].v_ifp);
1371	++mrtstat.mrts_wrong_if;
1372	++rt->mfc_wrong_if;
1373	/*
1374	 * If we are doing PIM assert processing, and we are forwarding
1375	 * packets on this interface, and it is a broadcast medium
1376	 * interface (and not a tunnel), send a message to the routing daemon.
1377	 */
1378	if (pim_assert && rt->mfc_ttls[vifi] &&
1379		(ifp->if_flags & IFF_BROADCAST) &&
1380		!(viftable[vifi].v_flags & VIFF_TUNNEL)) {
1381	    struct sockaddr_in k_igmpsrc;
1382	    struct mbuf *mm;
1383	    struct igmpmsg *im;
1384	    int hlen = ip->ip_hl << 2;
1385	    struct timeval now;
1386	    u_int32_t delta;
1387
1388	    GET_TIME(now);
1389
1390	    TV_DELTA(rt->mfc_last_assert, now, delta);
1391
1392	    if (delta > ASSERT_MSG_TIME) {
1393		mm = m_copy(m, 0, hlen);
1394		if (mm && (M_HASCL(mm) || mm->m_len < hlen))
1395		    mm = m_pullup(mm, hlen);
1396		if (mm == NULL) {
1397		    return ENOBUFS;
1398		}
1399
1400		rt->mfc_last_assert = now;
1401
1402		im = mtod(mm, struct igmpmsg *);
1403		im->im_msgtype	= IGMPMSG_WRONGVIF;
1404		im->im_mbz		= 0;
1405		im->im_vif		= vifi;
1406
1407		k_igmpsrc.sin_addr = im->im_src;
1408
1409		socket_send(ip_mrouter, mm, &k_igmpsrc);
1410	    }
1411	}
1412	return 0;
1413    }
1414
1415    /* If I sourced this packet, it counts as output, else it was input. */
1416    if (ip->ip_src.s_addr == viftable[vifi].v_lcl_addr.s_addr) {
1417	viftable[vifi].v_pkt_out++;
1418	viftable[vifi].v_bytes_out += plen;
1419    } else {
1420	viftable[vifi].v_pkt_in++;
1421	viftable[vifi].v_bytes_in += plen;
1422    }
1423    rt->mfc_pkt_cnt++;
1424    rt->mfc_byte_cnt += plen;
1425
1426    /*
1427     * For each vif, decide if a copy of the packet should be forwarded.
1428     * Forward if:
1429     *		- the ttl exceeds the vif's threshold
1430     *		- there are group members downstream on interface
1431     */
1432    for (vifp = viftable, vifi = 0; vifi < numvifs; vifp++, vifi++)
1433	if ((rt->mfc_ttls[vifi] > 0) &&
1434	    (ip->ip_ttl > rt->mfc_ttls[vifi])) {
1435	    vifp->v_pkt_out++;
1436	    vifp->v_bytes_out += plen;
1437	    MC_SEND(ip, vifp, m);
1438	}
1439
1440    return 0;
1441}
1442
1443/*
1444 * check if a vif number is legal/ok. This is used by ip_output, to export
1445 * numvifs there,
1446 */
1447static int
1448X_legal_vif_num(int vif)
1449{
1450    if (vif >= 0 && vif < numvifs)
1451       return(1);
1452    else
1453       return(0);
1454}
1455
1456#if !defined(MROUTE_LKM) || !MROUTE_LKM
1457int (*legal_vif_num)(int) = X_legal_vif_num;
1458#endif
1459
1460/*
1461 * Return the local address used by this vif
1462 */
1463static u_int32_t
1464X_ip_mcast_src(int vifi)
1465{
1466    if (vifi >= 0 && vifi < numvifs)
1467	return viftable[vifi].v_lcl_addr.s_addr;
1468    else
1469	return INADDR_ANY;
1470}
1471
1472#if !defined(MROUTE_LKM) || !MROUTE_LKM
1473u_int32_t (*ip_mcast_src)(int) = X_ip_mcast_src;
1474#endif
1475
1476static void
1477phyint_send(struct ip *ip, struct vif *vifp, struct mbuf *m)
1478{
1479    struct mbuf *mb_copy;
1480    int hlen = ip->ip_hl << 2;
1481
1482    /*
1483     * Make a new reference to the packet; make sure that
1484     * the IP header is actually copied, not just referenced,
1485     * so that ip_output() only scribbles on the copy.
1486     */
1487    mb_copy = m_copy(m, 0, M_COPYALL);
1488    if (mb_copy && (M_HASCL(mb_copy) || mb_copy->m_len < hlen))
1489	mb_copy = m_pullup(mb_copy, hlen);
1490    if (mb_copy == NULL)
1491	return;
1492
1493    if (vifp->v_rate_limit == 0)
1494	tbf_send_packet(vifp, mb_copy);
1495    else
1496	tbf_control(vifp, mb_copy, mtod(mb_copy, struct ip *), ip->ip_len);
1497}
1498
1499static void
1500encap_send(struct ip *ip, struct vif *vifp, struct mbuf *m)
1501{
1502    struct mbuf *mb_copy;
1503    struct ip *ip_copy;
1504    int i, len = ip->ip_len;
1505
1506    /*
1507     * copy the old packet & pullup its IP header into the
1508     * new mbuf so we can modify it.  Try to fill the new
1509     * mbuf since if we don't the ethernet driver will.
1510     */
1511    MGETHDR(mb_copy, M_DONTWAIT, MT_HEADER);
1512    if (mb_copy == NULL)
1513	return;
1514#if CONFIG_MACF_NET
1515    mac_mbuf_label_associate_multicast_encap(m, vifp->v_ifp, mb_copy);
1516#endif
1517    mb_copy->m_data += max_linkhdr;
1518    mb_copy->m_len = sizeof(multicast_encap_iphdr);
1519
1520    if ((mb_copy->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
1521	m_freem(mb_copy);
1522	return;
1523    }
1524    i = MHLEN - M_LEADINGSPACE(mb_copy);
1525    if (i > len)
1526	i = len;
1527    mb_copy = m_pullup(mb_copy, i);
1528    if (mb_copy == NULL)
1529	return;
1530    mb_copy->m_pkthdr.len = len + sizeof(multicast_encap_iphdr);
1531
1532    /*
1533     * fill in the encapsulating IP header.
1534     */
1535    ip_copy = mtod(mb_copy, struct ip *);
1536    *ip_copy = multicast_encap_iphdr;
1537    ip_copy->ip_id = ip_randomid();
1538    ip_copy->ip_len += len;
1539    ip_copy->ip_src = vifp->v_lcl_addr;
1540    ip_copy->ip_dst = vifp->v_rmt_addr;
1541
1542    /*
1543     * turn the encapsulated IP header back into a valid one.
1544     */
1545    ip = (struct ip *)((caddr_t)ip_copy + sizeof(multicast_encap_iphdr));
1546    --ip->ip_ttl;
1547
1548#if BYTE_ORDER != BIG_ENDIAN
1549    HTONS(ip->ip_len);
1550    HTONS(ip->ip_off);
1551#endif
1552
1553    ip->ip_sum = 0;
1554    mb_copy->m_data += sizeof(multicast_encap_iphdr);
1555    ip->ip_sum = in_cksum(mb_copy, ip->ip_hl << 2);
1556    mb_copy->m_data -= sizeof(multicast_encap_iphdr);
1557
1558    if (vifp->v_rate_limit == 0)
1559	tbf_send_packet(vifp, mb_copy);
1560    else
1561	tbf_control(vifp, mb_copy, ip, ip_copy->ip_len);
1562}
1563
1564/*
1565 * De-encapsulate a packet and feed it back through ip input (this
1566 * routine is called whenever IP gets a packet with proto type
1567 * ENCAP_PROTO and a local destination address).
1568 */
1569void
1570#if MROUTE_LKM
1571X_ipip_input(struct mbuf *m, int iphlen)
1572#else
1573ipip_input(struct mbuf *m, int iphlen)
1574#endif
1575{
1576    struct ifnet *ifp = m->m_pkthdr.rcvif;
1577    struct ip *ip = mtod(m, struct ip *);
1578    int hlen = ip->ip_hl << 2;
1579    struct vif *vifp;
1580
1581    if (!have_encap_tunnel) {
1582	    rip_input(m, iphlen);
1583	    return;
1584    }
1585    /*
1586     * dump the packet if it's not to a multicast destination or if
1587     * we don't have an encapsulating tunnel with the source.
1588     * Note:  This code assumes that the remote site IP address
1589     * uniquely identifies the tunnel (i.e., that this site has
1590     * at most one tunnel with the remote site).
1591     */
1592    if (! IN_MULTICAST(ntohl(((struct ip *)((char *)ip + hlen))->ip_dst.s_addr))) {
1593	++mrtstat.mrts_bad_tunnel;
1594	m_freem(m);
1595	return;
1596    }
1597    if (ip->ip_src.s_addr != last_encap_src) {
1598	struct vif *vife;
1599
1600	vifp = viftable;
1601	vife = vifp + numvifs;
1602	last_encap_src = ip->ip_src.s_addr;
1603	last_encap_vif = 0;
1604	for ( ; vifp < vife; ++vifp)
1605	    if (vifp->v_rmt_addr.s_addr == ip->ip_src.s_addr) {
1606		if ((vifp->v_flags & (VIFF_TUNNEL|VIFF_SRCRT))
1607		    == VIFF_TUNNEL)
1608		    last_encap_vif = vifp;
1609		break;
1610	    }
1611    }
1612    if ((vifp = last_encap_vif) == 0) {
1613	last_encap_src = 0;
1614	mrtstat.mrts_cant_tunnel++; /*XXX*/
1615	m_freem(m);
1616	if (mrtdebug)
1617	  log(LOG_DEBUG, "ip_mforward: no tunnel with %lx\n",
1618		(u_int32_t)ntohl(ip->ip_src.s_addr));
1619	return;
1620    }
1621    ifp = vifp->v_ifp;
1622
1623    if (hlen > IP_HDR_LEN)
1624      ip_stripoptions(m, (struct mbuf *) 0);
1625    m->m_data += IP_HDR_LEN;
1626    m->m_len -= IP_HDR_LEN;
1627    m->m_pkthdr.len -= IP_HDR_LEN;
1628    m->m_pkthdr.rcvif = ifp;
1629
1630    proto_inject(PF_INET, m);
1631}
1632
1633/*
1634 * Token bucket filter module
1635 */
1636
1637static void
1638tbf_control(struct vif *vifp, struct mbuf *m, struct ip *ip,
1639	    u_int32_t p_len)
1640{
1641    struct tbf *t = vifp->v_tbf;
1642
1643    if (p_len > MAX_BKT_SIZE) {
1644	/* drop if packet is too large */
1645	mrtstat.mrts_pkt2large++;
1646	m_freem(m);
1647	return;
1648    }
1649
1650    tbf_update_tokens(vifp);
1651
1652    /* if there are enough tokens,
1653     * and the queue is empty,
1654     * send this packet out
1655     */
1656
1657    if (t->tbf_q_len == 0) {
1658	/* queue empty, send packet if enough tokens */
1659	if (p_len <= t->tbf_n_tok) {
1660	    t->tbf_n_tok -= p_len;
1661	    tbf_send_packet(vifp, m);
1662	} else {
1663	    /* queue packet and timeout till later */
1664	    tbf_queue(vifp, m);
1665	    timeout(tbf_reprocess_q, (caddr_t)vifp, TBF_REPROCESS);
1666	}
1667    } else if (t->tbf_q_len < t->tbf_max_q_len) {
1668	/* finite queue length, so queue pkts and process queue */
1669	tbf_queue(vifp, m);
1670	tbf_process_q(vifp);
1671    } else {
1672	/* queue length too much, try to dq and queue and process */
1673	if (!tbf_dq_sel(vifp, ip)) {
1674	    mrtstat.mrts_q_overflow++;
1675	    m_freem(m);
1676	    return;
1677	} else {
1678	    tbf_queue(vifp, m);
1679	    tbf_process_q(vifp);
1680	}
1681    }
1682    return;
1683}
1684
1685/*
1686 * adds a packet to the queue at the interface
1687 */
1688static void
1689tbf_queue(struct vif *vifp, struct mbuf *m)
1690{
1691    struct tbf *t = vifp->v_tbf;
1692
1693    if (t->tbf_t == NULL) {
1694	/* Queue was empty */
1695	t->tbf_q = m;
1696    } else {
1697	/* Insert at tail */
1698	t->tbf_t->m_act = m;
1699    }
1700
1701    /* Set new tail pointer */
1702    t->tbf_t = m;
1703
1704#if DIAGNOSTIC
1705    /* Make sure we didn't get fed a bogus mbuf */
1706    if (m->m_act)
1707	panic("tbf_queue: m_act");
1708#endif
1709    m->m_act = NULL;
1710
1711    t->tbf_q_len++;
1712}
1713
1714
1715/*
1716 * processes the queue at the interface
1717 */
1718static void
1719tbf_process_q(struct vif *vifp)
1720{
1721    struct mbuf *m;
1722    int len;
1723    struct tbf *t = vifp->v_tbf;
1724
1725    /* loop through the queue at the interface and send as many packets
1726     * as possible
1727     */
1728    while (t->tbf_q_len > 0) {
1729	m = t->tbf_q;
1730
1731	len = mtod(m, struct ip *)->ip_len;
1732
1733	/* determine if the packet can be sent */
1734	if (len <= t->tbf_n_tok) {
1735	    /* if so,
1736	     * reduce no of tokens, dequeue the packet,
1737	     * send the packet.
1738	     */
1739	    t->tbf_n_tok -= len;
1740
1741	    t->tbf_q = m->m_act;
1742	    if (--t->tbf_q_len == 0)
1743		t->tbf_t = NULL;
1744
1745	    m->m_act = NULL;
1746	    tbf_send_packet(vifp, m);
1747
1748	} else break;
1749    }
1750}
1751
1752static void
1753tbf_reprocess_q(void *xvifp)
1754{
1755    struct vif *vifp = xvifp;
1756
1757    if (ip_mrouter == NULL)  {
1758	return;
1759     }
1760
1761    tbf_update_tokens(vifp);
1762
1763    tbf_process_q(vifp);
1764
1765    if (vifp->v_tbf->tbf_q_len)
1766	timeout(tbf_reprocess_q, (caddr_t)vifp, TBF_REPROCESS);
1767}
1768
1769/* function that will selectively discard a member of the queue
1770 * based on the precedence value and the priority
1771 */
1772static int
1773tbf_dq_sel(struct vif *vifp, struct ip *ip)
1774{
1775    u_int p;
1776    struct mbuf *m, *last;
1777    struct mbuf **np;
1778    struct tbf *t = vifp->v_tbf;
1779
1780    p = priority(vifp, ip);
1781
1782    np = &t->tbf_q;
1783    last = NULL;
1784    while ((m = *np) != NULL) {
1785	if (p > priority(vifp, mtod(m, struct ip *))) {
1786	    *np = m->m_act;
1787	    /* If we're removing the last packet, fix the tail pointer */
1788	    if (m == t->tbf_t)
1789		t->tbf_t = last;
1790	    m_freem(m);
1791	    /* it's impossible for the queue to be empty, but
1792	     * we check anyway. */
1793	    if (--t->tbf_q_len == 0)
1794		t->tbf_t = NULL;
1795	    mrtstat.mrts_drop_sel++;
1796	    return(1);
1797	}
1798	np = &m->m_act;
1799	last = m;
1800    }
1801    return(0);
1802}
1803
1804static void
1805tbf_send_packet(struct vif *vifp, struct mbuf *m)
1806{
1807    int error;
1808    struct route ro;
1809
1810    bzero(&ro, sizeof (ro));
1811    if (vifp->v_flags & VIFF_TUNNEL) {
1812	/* If tunnel options */
1813	ip_output(m, (struct mbuf *)0, &ro,
1814		  IP_FORWARDING, (struct ip_moptions *)0, NULL);
1815    } else {
1816	struct ip_moptions *imo;
1817
1818	imo = ip_allocmoptions(M_DONTWAIT);
1819	if (imo == NULL) {
1820		error = ENOMEM;
1821		goto done;
1822	}
1823
1824	imo->imo_multicast_ifp  = vifp->v_ifp;
1825	imo->imo_multicast_ttl  = mtod(m, struct ip *)->ip_ttl - 1;
1826	imo->imo_multicast_loop = 1;
1827	imo->imo_multicast_vif  = -1;
1828
1829	/*
1830	 * Re-entrancy should not be a problem here, because
1831	 * the packets that we send out and are looped back at us
1832	 * should get rejected because they appear to come from
1833	 * the loopback interface, thus preventing looping.
1834	 */
1835	error = ip_output(m, (struct mbuf *)0, &ro,
1836			  IP_FORWARDING, imo, NULL);
1837
1838	IMO_REMREF(imo);
1839done:
1840	if (mrtdebug & DEBUG_XMIT)
1841	    log(LOG_DEBUG, "phyint_send on vif %d err %d\n",
1842		vifp - viftable, error);
1843    }
1844    ROUTE_RELEASE(&ro);
1845}
1846
1847/* determine the current time and then
1848 * the elapsed time (between the last time and time now)
1849 * in milliseconds & update the no. of tokens in the bucket
1850 */
1851static void
1852tbf_update_tokens(struct vif *vifp)
1853{
1854    struct timeval tp;
1855    u_int32_t tm;
1856    struct tbf *t = vifp->v_tbf;
1857
1858    GET_TIME(tp);
1859
1860    TV_DELTA(tp, t->tbf_last_pkt_t, tm);
1861
1862    /*
1863     * This formula is actually
1864     * "time in seconds" * "bytes/second".
1865     *
1866     * (tm / 1000000) * (v_rate_limit * 1000 * (1000/1024) / 8)
1867     *
1868     * The (1000/1024) was introduced in add_vif to optimize
1869     * this divide into a shift.
1870     */
1871    t->tbf_n_tok += tm * vifp->v_rate_limit / 1024 / 8;
1872    t->tbf_last_pkt_t = tp;
1873
1874    if (t->tbf_n_tok > MAX_BKT_SIZE)
1875	t->tbf_n_tok = MAX_BKT_SIZE;
1876}
1877
1878static int
1879priority(__unused struct vif *vifp, struct ip *ip)
1880{
1881    int prio;
1882
1883    /* temporary hack; may add general packet classifier some day */
1884
1885    /*
1886     * The UDP port space is divided up into four priority ranges:
1887     * [0, 16384)     : unclassified - lowest priority
1888     * [16384, 32768) : audio - highest priority
1889     * [32768, 49152) : whiteboard - medium priority
1890     * [49152, 65536) : video - low priority
1891     */
1892    if (ip->ip_p == IPPROTO_UDP) {
1893	struct udphdr *udp = (struct udphdr *)(((char *)ip) + (ip->ip_hl << 2));
1894	switch (ntohs(udp->uh_dport) & 0xc000) {
1895	    case 0x4000:
1896		prio = 70;
1897		break;
1898	    case 0x8000:
1899		prio = 60;
1900		break;
1901	    case 0xc000:
1902		prio = 55;
1903		break;
1904	    default:
1905		prio = 50;
1906		break;
1907	}
1908	if (tbfdebug > 1)
1909		log(LOG_DEBUG, "port %x prio%d\n", ntohs(udp->uh_dport), prio);
1910    } else {
1911	    prio = 50;
1912    }
1913    return prio;
1914}
1915
1916/*
1917 * End of token bucket filter modifications
1918 */
1919
1920int
1921ip_rsvp_vif_init(struct socket *so, struct sockopt *sopt)
1922{
1923    int error, i;
1924
1925    if (rsvpdebug)
1926	printf("ip_rsvp_vif_init: so_type = %d, pr_protocol = %d\n",
1927	       so->so_type, so->so_proto->pr_protocol);
1928
1929    if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP)
1930	return EOPNOTSUPP;
1931
1932    /* Check mbuf. */
1933    error = sooptcopyin(sopt, &i, sizeof i, sizeof i);
1934    if (error)
1935	    return (error);
1936
1937    if (rsvpdebug)
1938	printf("ip_rsvp_vif_init: vif = %d rsvp_on = %d\n", i, rsvp_on);
1939
1940    /* Check vif. */
1941    if (!legal_vif_num(i)) {
1942	return EADDRNOTAVAIL;
1943    }
1944
1945    /* Check if socket is available. */
1946    if (viftable[i].v_rsvpd != NULL) {
1947	return EADDRINUSE;
1948    }
1949
1950    viftable[i].v_rsvpd = so;
1951    /* This may seem silly, but we need to be sure we don't over-increment
1952     * the RSVP counter, in case something slips up.
1953     */
1954    if (!viftable[i].v_rsvp_on) {
1955	viftable[i].v_rsvp_on = 1;
1956	rsvp_on++;
1957    }
1958
1959    return 0;
1960}
1961
1962int
1963ip_rsvp_vif_done(struct socket *so, struct sockopt *sopt)
1964{
1965	int error, i;
1966
1967	if (rsvpdebug)
1968		printf("ip_rsvp_vif_done: so_type = %d, pr_protocol = %d\n",
1969		       so->so_type, so->so_proto->pr_protocol);
1970
1971	if (so->so_type != SOCK_RAW ||
1972	    so->so_proto->pr_protocol != IPPROTO_RSVP)
1973		return EOPNOTSUPP;
1974
1975	error = sooptcopyin(sopt, &i, sizeof i, sizeof i);
1976	if (error)
1977		return (error);
1978
1979	/* Check vif. */
1980	if (!legal_vif_num(i)) {
1981		return EADDRNOTAVAIL;
1982	}
1983
1984	if (rsvpdebug)
1985		printf("ip_rsvp_vif_done: v_rsvpd = %p so = %p\n",
1986		       viftable[i].v_rsvpd, so);
1987
1988	viftable[i].v_rsvpd = NULL;
1989	/*
1990	 * This may seem silly, but we need to be sure we don't over-decrement
1991	 * the RSVP counter, in case something slips up.
1992	 */
1993	if (viftable[i].v_rsvp_on) {
1994		viftable[i].v_rsvp_on = 0;
1995		rsvp_on--;
1996	}
1997
1998	return 0;
1999}
2000
2001void
2002ip_rsvp_force_done(struct socket *so)
2003{
2004    int vifi;
2005
2006    /* Don't bother if it is not the right type of socket. */
2007    if (so->so_type != SOCK_RAW || so->so_proto->pr_protocol != IPPROTO_RSVP)
2008	return;
2009
2010    /* The socket may be attached to more than one vif...this
2011     * is perfectly legal.
2012     */
2013    for (vifi = 0; vifi < numvifs; vifi++) {
2014	if (viftable[vifi].v_rsvpd == so) {
2015	    viftable[vifi].v_rsvpd = NULL;
2016	    /* This may seem silly, but we need to be sure we don't
2017	     * over-decrement the RSVP counter, in case something slips up.
2018	     */
2019	    if (viftable[vifi].v_rsvp_on) {
2020		viftable[vifi].v_rsvp_on = 0;
2021		rsvp_on--;
2022	    }
2023	}
2024    }
2025
2026    return;
2027}
2028
2029void
2030rsvp_input(struct mbuf *m, int iphlen)
2031{
2032    int vifi;
2033    struct ip *ip = mtod(m, struct ip *);
2034    static struct sockaddr_in rsvp_src = { sizeof rsvp_src, AF_INET,
2035    										0 , {0}, {0,0,0,0,0,0,0,0,} };
2036    struct ifnet *ifp;
2037
2038    if (rsvpdebug)
2039	printf("rsvp_input: rsvp_on %d\n",rsvp_on);
2040
2041    /* Can still get packets with rsvp_on = 0 if there is a local member
2042     * of the group to which the RSVP packet is addressed.  But in this
2043     * case we want to throw the packet away.
2044     */
2045    if (!rsvp_on) {
2046	m_freem(m);
2047	return;
2048    }
2049
2050    if (rsvpdebug)
2051	printf("rsvp_input: check vifs\n");
2052
2053#if DIAGNOSTIC
2054    if (!(m->m_flags & M_PKTHDR))
2055	    panic("rsvp_input no hdr");
2056#endif
2057
2058    ifp = m->m_pkthdr.rcvif;
2059    /* Find which vif the packet arrived on. */
2060    for (vifi = 0; vifi < numvifs; vifi++)
2061	if (viftable[vifi].v_ifp == ifp)
2062	    break;
2063
2064    if (vifi == numvifs || viftable[vifi].v_rsvpd == NULL) {
2065	/*
2066	 * If the old-style non-vif-associated socket is set,
2067	 * then use it.  Otherwise, drop packet since there
2068	 * is no specific socket for this vif.
2069	 */
2070	if (ip_rsvpd != NULL) {
2071	    if (rsvpdebug)
2072		printf("rsvp_input: Sending packet up old-style socket\n");
2073	    rip_input(m, iphlen);  /* xxx */
2074	} else {
2075	    if (rsvpdebug && vifi == numvifs)
2076		printf("rsvp_input: Can't find vif for packet.\n");
2077	    else if (rsvpdebug && viftable[vifi].v_rsvpd == NULL)
2078		printf("rsvp_input: No socket defined for vif %d\n",vifi);
2079	    m_freem(m);
2080	}
2081	return;
2082    }
2083    rsvp_src.sin_addr = ip->ip_src;
2084
2085    if (rsvpdebug && m)
2086	printf("rsvp_input: m->m_len = %d, sbspace() = %d\n",
2087	       m->m_len,sbspace(&(viftable[vifi].v_rsvpd->so_rcv)));
2088
2089    if (socket_send(viftable[vifi].v_rsvpd, m, &rsvp_src) < 0) {
2090	if (rsvpdebug)
2091	    printf("rsvp_input: Failed to append to socket\n");
2092    } else {
2093	if (rsvpdebug)
2094	    printf("rsvp_input: send packet up\n");
2095    }
2096
2097}
2098
2099#if MROUTE_LKM
2100#include <sys/conf.h>
2101#include <sys/exec.h>
2102#include <sys/sysent.h>
2103#include <sys/lkm.h>
2104
2105MOD_MISC("ip_mroute_mod")
2106
2107static int
2108ip_mroute_mod_handle(struct lkm_table *lkmtp, int cmd)
2109{
2110	int i;
2111	struct lkm_misc	*args = lkmtp->private.lkm_misc;
2112	int err = 0;
2113
2114	switch(cmd) {
2115		static int (*old_ip_mrouter_cmd)();
2116		static int (*old_ip_mrouter_done)();
2117		static int (*old_ip_mforward)();
2118		static int (*old_mrt_ioctl)();
2119		static void (*old_proto4_input)();
2120		static int (*old_legal_vif_num)();
2121
2122	case LKM_E_LOAD:
2123		if(lkmexists(lkmtp) || ip_mrtproto)
2124		  return(EEXIST);
2125		old_ip_mrouter_cmd = ip_mrouter_cmd;
2126		ip_mrouter_cmd = X_ip_mrouter_cmd;
2127		old_ip_mrouter_done = ip_mrouter_done;
2128		ip_mrouter_done = X_ip_mrouter_done;
2129		old_ip_mforward = ip_mforward;
2130		ip_mforward = X_ip_mforward;
2131		old_mrt_ioctl = mrt_ioctl;
2132		mrt_ioctl = X_mrt_ioctl;
2133              old_proto4_input = ip_protox[ENCAP_PROTO]->pr_input;
2134              ip_protox[ENCAP_PROTO]->pr_input = X_ipip_input;
2135		old_legal_vif_num = legal_vif_num;
2136		legal_vif_num = X_legal_vif_num;
2137		ip_mrtproto = IGMP_DVMRP;
2138
2139		printf("\nIP multicast routing loaded\n");
2140		break;
2141
2142	case LKM_E_UNLOAD:
2143		if (ip_mrouter)
2144		  return EINVAL;
2145
2146		ip_mrouter_cmd = old_ip_mrouter_cmd;
2147		ip_mrouter_done = old_ip_mrouter_done;
2148		ip_mforward = old_ip_mforward;
2149		mrt_ioctl = old_mrt_ioctl;
2150              ip_protox[ENCAP_PROTO]->pr_input = old_proto4_input;
2151		legal_vif_num = old_legal_vif_num;
2152		ip_mrtproto = 0;
2153		break;
2154
2155	default:
2156		err = EINVAL;
2157		break;
2158	}
2159
2160	return(err);
2161}
2162
2163int
2164ip_mroute_mod(struct lkm_table *lkmtp, int cmd, int ver) {
2165	DISPATCH(lkmtp, cmd, ver, ip_mroute_mod_handle, ip_mroute_mod_handle,
2166		 nosys);
2167}
2168
2169#endif /* MROUTE_LKM */
2170#endif /* MROUTING */
2171