ip_input.c revision 1.63
1/*	$NetBSD: ip_input.c,v 1.63 1998/04/29 21:37:55 matt Exp $	*/
2
3/*-
4 * Copyright (c) 1998 The NetBSD Foundation, Inc.
5 * All rights reserved.
6 *
7 * This code is derived from software contributed to The NetBSD Foundation
8 * by Public Access Networks Corporation ("Panix").  It was developed under
9 * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 *    notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 *    notice, this list of conditions and the following disclaimer in the
18 *    documentation and/or other materials provided with the distribution.
19 * 3. All advertising materials mentioning features or use of this software
20 *    must display the following acknowledgement:
21 *	This product includes software developed by the NetBSD
22 *	Foundation, Inc. and its contributors.
23 * 4. Neither the name of The NetBSD Foundation nor the names of its
24 *    contributors may be used to endorse or promote products derived
25 *    from this software without specific prior written permission.
26 *
27 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
28 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30 * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37 * POSSIBILITY OF SUCH DAMAGE.
38 */
39
40/*
41 * Copyright (c) 1982, 1986, 1988, 1993
42 *	The Regents of the University of California.  All rights reserved.
43 *
44 * Redistribution and use in source and binary forms, with or without
45 * modification, are permitted provided that the following conditions
46 * are met:
47 * 1. Redistributions of source code must retain the above copyright
48 *    notice, this list of conditions and the following disclaimer.
49 * 2. Redistributions in binary form must reproduce the above copyright
50 *    notice, this list of conditions and the following disclaimer in the
51 *    documentation and/or other materials provided with the distribution.
52 * 3. All advertising materials mentioning features or use of this software
53 *    must display the following acknowledgement:
54 *	This product includes software developed by the University of
55 *	California, Berkeley and its contributors.
56 * 4. Neither the name of the University nor the names of its contributors
57 *    may be used to endorse or promote products derived from this software
58 *    without specific prior written permission.
59 *
60 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
61 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
62 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
63 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
64 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
65 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
66 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
67 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
68 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
69 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
70 * SUCH DAMAGE.
71 *
72 *	@(#)ip_input.c	8.2 (Berkeley) 1/4/94
73 */
74
75#include "opt_gateway.h"
76#include "opt_mrouting.h"
77
78#include <sys/param.h>
79#include <sys/systm.h>
80#include <sys/malloc.h>
81#include <sys/mbuf.h>
82#include <sys/domain.h>
83#include <sys/protosw.h>
84#include <sys/socket.h>
85#include <sys/socketvar.h>
86#include <sys/errno.h>
87#include <sys/time.h>
88#include <sys/kernel.h>
89#include <sys/proc.h>
90
91#include <vm/vm.h>
92#include <sys/sysctl.h>
93
94#include <net/if.h>
95#include <net/if_dl.h>
96#include <net/route.h>
97#include <net/pfil.h>
98
99#include <netinet/in.h>
100#include <netinet/in_systm.h>
101#include <netinet/ip.h>
102#include <netinet/in_pcb.h>
103#include <netinet/in_var.h>
104#include <netinet/ip_var.h>
105#include <netinet/ip_icmp.h>
106
107#ifndef	IPFORWARDING
108#ifdef GATEWAY
109#define	IPFORWARDING	1	/* forward IP packets not for us */
110#else /* GATEWAY */
111#define	IPFORWARDING	0	/* don't forward IP packets not for us */
112#endif /* GATEWAY */
113#endif /* IPFORWARDING */
114#ifndef	IPSENDREDIRECTS
115#define	IPSENDREDIRECTS	1
116#endif
117#ifndef IPFORWSRCRT
118#define	IPFORWSRCRT	1	/* forward source-routed packets */
119#endif
120#ifndef IPALLOWSRCRT
121#define	IPALLOWSRCRT	1	/* allow source-routed packets */
122#endif
123#ifndef IPMTUDISC
124#define IPMTUDISC	0
125#endif
126#ifndef IPMTUDISCTIMEOUT
127#define IPMTUDISCTIMEOUT (10 * 60)	/* as per RFC 1191 */
128#endif
129
130/*
131 * Note: DIRECTED_BROADCAST is handled this way so that previous
132 * configuration using this option will Just Work.
133 */
134#ifndef IPDIRECTEDBCAST
135#ifdef DIRECTED_BROADCAST
136#define IPDIRECTEDBCAST	1
137#else
138#define	IPDIRECTEDBCAST	0
139#endif /* DIRECTED_BROADCAST */
140#endif /* IPDIRECTEDBCAST */
141int	ipforwarding = IPFORWARDING;
142int	ipsendredirects = IPSENDREDIRECTS;
143int	ip_defttl = IPDEFTTL;
144int	ip_forwsrcrt = IPFORWSRCRT;
145int	ip_directedbcast = IPDIRECTEDBCAST;
146int	ip_allowsrcrt = IPALLOWSRCRT;
147int	ip_mtudisc = IPMTUDISC;
148u_int	ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
149#ifdef DIAGNOSTIC
150int	ipprintfs = 0;
151#endif
152
153struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
154
155extern	struct domain inetdomain;
156extern	struct protosw inetsw[];
157u_char	ip_protox[IPPROTO_MAX];
158int	ipqmaxlen = IFQ_MAXLEN;
159struct	in_ifaddrhead in_ifaddr;
160struct	in_ifaddrhashhead *in_ifaddrhashtbl;
161struct	ifqueue ipintrq;
162struct	ipstat	ipstat;
163u_int16_t	ip_id;
164int	ip_defttl;
165struct ipqhead ipq;
166
167/*
168 * We need to save the IP options in case a protocol wants to respond
169 * to an incoming packet over the same route if the packet got here
170 * using IP source routing.  This allows connection establishment and
171 * maintenance when the remote end is on a network that is not known
172 * to us.
173 */
174int	ip_nhops = 0;
175static	struct ip_srcrt {
176	struct	in_addr dst;			/* final destination */
177	char	nop;				/* one NOP to align */
178	char	srcopt[IPOPT_OFFSET + 1];	/* OPTVAL, OLEN and OFFSET */
179	struct	in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
180} ip_srcrt;
181
182static void save_rte __P((u_char *, struct in_addr));
183
184/*
185 * IP initialization: fill in IP protocol switch table.
186 * All protocols not implemented in kernel go to raw IP protocol handler.
187 */
188void
189ip_init()
190{
191	register struct protosw *pr;
192	register int i;
193
194	pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
195	if (pr == 0)
196		panic("ip_init");
197	for (i = 0; i < IPPROTO_MAX; i++)
198		ip_protox[i] = pr - inetsw;
199	for (pr = inetdomain.dom_protosw;
200	    pr < inetdomain.dom_protoswNPROTOSW; pr++)
201		if (pr->pr_domain->dom_family == PF_INET &&
202		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
203			ip_protox[pr->pr_protocol] = pr - inetsw;
204	LIST_INIT(&ipq);
205	ip_id = time.tv_sec & 0xffff;
206	ipintrq.ifq_maxlen = ipqmaxlen;
207	TAILQ_INIT(&in_ifaddr);
208	in_ifaddrhashtbl =
209	    hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, M_WAITOK, &in_ifaddrhash);
210	if (ip_mtudisc != 0)
211		ip_mtudisc_timeout_q =
212		    rt_timer_queue_create(ip_mtudisc_timeout);
213}
214
215struct	sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
216struct	route ipforward_rt;
217
218/*
219 * Ip input routine.  Checksum and byte swap header.  If fragmented
220 * try to reassemble.  Process options.  Pass to next level.
221 */
222void
223ipintr()
224{
225	register struct ip *ip = NULL;
226	register struct mbuf *m;
227	register struct ipq *fp;
228	register struct in_ifaddr *ia;
229	register struct ifaddr *ifa;
230	struct ipqent *ipqe;
231	int hlen = 0, mff, len, s;
232#ifdef PFIL_HOOKS
233	struct packet_filter_hook *pfh;
234	struct mbuf *m0;
235	int rv;
236#endif /* PFIL_HOOKS */
237
238next:
239	/*
240	 * Get next datagram off input queue and get IP header
241	 * in first mbuf.
242	 */
243	s = splimp();
244	IF_DEQUEUE(&ipintrq, m);
245	splx(s);
246	if (m == 0)
247		return;
248#ifdef	DIAGNOSTIC
249	if ((m->m_flags & M_PKTHDR) == 0)
250		panic("ipintr no HDR");
251#endif
252	/*
253	 * If no IP addresses have been set yet but the interfaces
254	 * are receiving, can't do anything with incoming packets yet.
255	 */
256	if (in_ifaddr.tqh_first == 0)
257		goto bad;
258	ipstat.ips_total++;
259	if (m->m_len < sizeof (struct ip) &&
260	    (m = m_pullup(m, sizeof (struct ip))) == 0) {
261		ipstat.ips_toosmall++;
262		goto next;
263	}
264	ip = mtod(m, struct ip *);
265	if (ip->ip_v != IPVERSION) {
266		ipstat.ips_badvers++;
267		goto bad;
268	}
269	hlen = ip->ip_hl << 2;
270	if (hlen < sizeof(struct ip)) {	/* minimum header length */
271		ipstat.ips_badhlen++;
272		goto bad;
273	}
274	if (hlen > m->m_len) {
275		if ((m = m_pullup(m, hlen)) == 0) {
276			ipstat.ips_badhlen++;
277			goto next;
278		}
279		ip = mtod(m, struct ip *);
280	}
281	if ((ip->ip_sum = in_cksum(m, hlen)) != 0) {
282		ipstat.ips_badsum++;
283		goto bad;
284	}
285
286	/*
287	 * Convert fields to host representation.
288	 */
289	NTOHS(ip->ip_len);
290	NTOHS(ip->ip_id);
291	NTOHS(ip->ip_off);
292	len = ip->ip_len;
293
294	/*
295	 * Check that the amount of data in the buffers
296	 * is as at least much as the IP header would have us expect.
297	 * Trim mbufs if longer than we expect.
298	 * Drop packet if shorter than we expect.
299	 */
300	if (m->m_pkthdr.len < len) {
301		ipstat.ips_tooshort++;
302		goto bad;
303	}
304	if (m->m_pkthdr.len > len) {
305		if (m->m_len == m->m_pkthdr.len) {
306			m->m_len = len;
307			m->m_pkthdr.len = len;
308		} else
309			m_adj(m, len - m->m_pkthdr.len);
310	}
311
312#ifdef PFIL_HOOKS
313	/*
314	 * Run through list of hooks for input packets.
315	 */
316	m0 = m;
317	for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
318		if (pfh->pfil_func) {
319			rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
320			if (rv)
321				goto next;
322			ip = mtod(m = m0, struct ip *);
323		}
324#endif /* PFIL_HOOKS */
325
326	/*
327	 * Process options and, if not destined for us,
328	 * ship it on.  ip_dooptions returns 1 when an
329	 * error was detected (causing an icmp message
330	 * to be sent and the original packet to be freed).
331	 */
332	ip_nhops = 0;		/* for source routed packets */
333	if (hlen > sizeof (struct ip) && ip_dooptions(m))
334		goto next;
335
336	/*
337	 * Check our list of addresses, to see if the packet is for us.
338	 */
339	INADDR_TO_IA(ip->ip_dst, ia);
340	if (ia != NULL) goto ours;
341	if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
342		for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
343		    ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
344			if (ifa->ifa_addr->sa_family != AF_INET) continue;
345			ia = ifatoia(ifa);
346			if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
347			    in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
348			    /*
349			     * Look for all-0's host part (old broadcast addr),
350			     * either for subnet or net.
351			     */
352			    ip->ip_dst.s_addr == ia->ia_subnet ||
353			    ip->ip_dst.s_addr == ia->ia_net)
354				goto ours;
355			/*
356			 * An interface with IP address zero accepts
357			 * all packets that arrive on that interface.
358			 */
359			if (in_nullhost(ia->ia_addr.sin_addr))
360				goto ours;
361		}
362	}
363	if (IN_MULTICAST(ip->ip_dst.s_addr)) {
364		struct in_multi *inm;
365#ifdef MROUTING
366		extern struct socket *ip_mrouter;
367
368		if (m->m_flags & M_EXT) {
369			if ((m = m_pullup(m, hlen)) == 0) {
370				ipstat.ips_toosmall++;
371				goto next;
372			}
373			ip = mtod(m, struct ip *);
374		}
375
376		if (ip_mrouter) {
377			/*
378			 * If we are acting as a multicast router, all
379			 * incoming multicast packets are passed to the
380			 * kernel-level multicast forwarding function.
381			 * The packet is returned (relatively) intact; if
382			 * ip_mforward() returns a non-zero value, the packet
383			 * must be discarded, else it may be accepted below.
384			 *
385			 * (The IP ident field is put in the same byte order
386			 * as expected when ip_mforward() is called from
387			 * ip_output().)
388			 */
389			ip->ip_id = htons(ip->ip_id);
390			if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
391				ipstat.ips_cantforward++;
392				m_freem(m);
393				goto next;
394			}
395			ip->ip_id = ntohs(ip->ip_id);
396
397			/*
398			 * The process-level routing demon needs to receive
399			 * all multicast IGMP packets, whether or not this
400			 * host belongs to their destination groups.
401			 */
402			if (ip->ip_p == IPPROTO_IGMP)
403				goto ours;
404			ipstat.ips_forward++;
405		}
406#endif
407		/*
408		 * See if we belong to the destination multicast group on the
409		 * arrival interface.
410		 */
411		IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
412		if (inm == NULL) {
413			ipstat.ips_cantforward++;
414			m_freem(m);
415			goto next;
416		}
417		goto ours;
418	}
419	if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
420	    in_nullhost(ip->ip_dst))
421		goto ours;
422
423	/*
424	 * Not for us; forward if possible and desirable.
425	 */
426	if (ipforwarding == 0) {
427		ipstat.ips_cantforward++;
428		m_freem(m);
429	} else
430		ip_forward(m, 0);
431	goto next;
432
433ours:
434	/*
435	 * If offset or IP_MF are set, must reassemble.
436	 * Otherwise, nothing need be done.
437	 * (We could look in the reassembly queue to see
438	 * if the packet was previously fragmented,
439	 * but it's not worth the time; just let them time out.)
440	 */
441	if (ip->ip_off & ~(IP_DF|IP_RF)) {
442		/*
443		 * Look for queue of fragments
444		 * of this datagram.
445		 */
446		for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
447			if (ip->ip_id == fp->ipq_id &&
448			    in_hosteq(ip->ip_src, fp->ipq_src) &&
449			    in_hosteq(ip->ip_dst, fp->ipq_dst) &&
450			    ip->ip_p == fp->ipq_p)
451				goto found;
452		fp = 0;
453found:
454
455		/*
456		 * Adjust ip_len to not reflect header,
457		 * set ipqe_mff if more fragments are expected,
458		 * convert offset of this to bytes.
459		 */
460		ip->ip_len -= hlen;
461		mff = (ip->ip_off & IP_MF) != 0;
462		if (mff) {
463		        /*
464		         * Make sure that fragments have a data length
465			 * that's a non-zero multiple of 8 bytes.
466		         */
467			if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
468				ipstat.ips_badfrags++;
469				goto bad;
470			}
471		}
472		ip->ip_off <<= 3;
473
474		/*
475		 * If datagram marked as having more fragments
476		 * or if this is not the first fragment,
477		 * attempt reassembly; if it succeeds, proceed.
478		 */
479		if (mff || ip->ip_off) {
480			ipstat.ips_fragments++;
481			MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent),
482			    M_IPQ, M_NOWAIT);
483			if (ipqe == NULL) {
484				ipstat.ips_rcvmemdrop++;
485				goto bad;
486			}
487			ipqe->ipqe_mff = mff;
488			ipqe->ipqe_m = m;
489			ipqe->ipqe_ip = ip;
490			m = ip_reass(ipqe, fp);
491			if (m == 0)
492				goto next;
493			ipstat.ips_reassembled++;
494			ip = mtod(m, struct ip *);
495		} else
496			if (fp)
497				ip_freef(fp);
498	} else
499		ip->ip_len -= hlen;
500
501	/*
502	 * Switch out to protocol's input routine.
503	 */
504	ipstat.ips_delivered++;
505	(*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
506	goto next;
507bad:
508	m_freem(m);
509	goto next;
510}
511
512/*
513 * Take incoming datagram fragment and try to
514 * reassemble it into whole datagram.  If a chain for
515 * reassembly of this datagram already exists, then it
516 * is given as fp; otherwise have to make a chain.
517 */
518struct mbuf *
519ip_reass(ipqe, fp)
520	register struct ipqent *ipqe;
521	register struct ipq *fp;
522{
523	register struct mbuf *m = ipqe->ipqe_m;
524	register struct ipqent *nq, *p, *q;
525	struct ip *ip;
526	struct mbuf *t;
527	int hlen = ipqe->ipqe_ip->ip_hl << 2;
528	int i, next;
529
530	/*
531	 * Presence of header sizes in mbufs
532	 * would confuse code below.
533	 */
534	m->m_data += hlen;
535	m->m_len -= hlen;
536
537	/*
538	 * If first fragment to arrive, create a reassembly queue.
539	 */
540	if (fp == 0) {
541		MALLOC(fp, struct ipq *, sizeof (struct ipq),
542		    M_FTABLE, M_NOWAIT);
543		if (fp == NULL)
544			goto dropfrag;
545		LIST_INSERT_HEAD(&ipq, fp, ipq_q);
546		fp->ipq_ttl = IPFRAGTTL;
547		fp->ipq_p = ipqe->ipqe_ip->ip_p;
548		fp->ipq_id = ipqe->ipqe_ip->ip_id;
549		LIST_INIT(&fp->ipq_fragq);
550		fp->ipq_src = ipqe->ipqe_ip->ip_src;
551		fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
552		p = NULL;
553		goto insert;
554	}
555
556	/*
557	 * Find a segment which begins after this one does.
558	 */
559	for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
560	    p = q, q = q->ipqe_q.le_next)
561		if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
562			break;
563
564	/*
565	 * If there is a preceding segment, it may provide some of
566	 * our data already.  If so, drop the data from the incoming
567	 * segment.  If it provides all of our data, drop us.
568	 */
569	if (p != NULL) {
570		i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
571		    ipqe->ipqe_ip->ip_off;
572		if (i > 0) {
573			if (i >= ipqe->ipqe_ip->ip_len)
574				goto dropfrag;
575			m_adj(ipqe->ipqe_m, i);
576			ipqe->ipqe_ip->ip_off += i;
577			ipqe->ipqe_ip->ip_len -= i;
578		}
579	}
580
581	/*
582	 * While we overlap succeeding segments trim them or,
583	 * if they are completely covered, dequeue them.
584	 */
585	for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
586	    q->ipqe_ip->ip_off; q = nq) {
587		i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
588		    q->ipqe_ip->ip_off;
589		if (i < q->ipqe_ip->ip_len) {
590			q->ipqe_ip->ip_len -= i;
591			q->ipqe_ip->ip_off += i;
592			m_adj(q->ipqe_m, i);
593			break;
594		}
595		nq = q->ipqe_q.le_next;
596		m_freem(q->ipqe_m);
597		LIST_REMOVE(q, ipqe_q);
598		FREE(q, M_IPQ);
599	}
600
601insert:
602	/*
603	 * Stick new segment in its place;
604	 * check for complete reassembly.
605	 */
606	if (p == NULL) {
607		LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
608	} else {
609		LIST_INSERT_AFTER(p, ipqe, ipqe_q);
610	}
611	next = 0;
612	for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
613	    p = q, q = q->ipqe_q.le_next) {
614		if (q->ipqe_ip->ip_off != next)
615			return (0);
616		next += q->ipqe_ip->ip_len;
617	}
618	if (p->ipqe_mff)
619		return (0);
620
621	/*
622	 * Reassembly is complete.  Check for a bogus message size and
623	 * concatenate fragments.
624	 */
625	q = fp->ipq_fragq.lh_first;
626	ip = q->ipqe_ip;
627	if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
628		ipstat.ips_toolong++;
629		ip_freef(fp);
630		return (0);
631	}
632	m = q->ipqe_m;
633	t = m->m_next;
634	m->m_next = 0;
635	m_cat(m, t);
636	nq = q->ipqe_q.le_next;
637	FREE(q, M_IPQ);
638	for (q = nq; q != NULL; q = nq) {
639		t = q->ipqe_m;
640		nq = q->ipqe_q.le_next;
641		FREE(q, M_IPQ);
642		m_cat(m, t);
643	}
644
645	/*
646	 * Create header for new ip packet by
647	 * modifying header of first packet;
648	 * dequeue and discard fragment reassembly header.
649	 * Make header visible.
650	 */
651	ip->ip_len = next;
652	ip->ip_src = fp->ipq_src;
653	ip->ip_dst = fp->ipq_dst;
654	LIST_REMOVE(fp, ipq_q);
655	FREE(fp, M_FTABLE);
656	m->m_len += (ip->ip_hl << 2);
657	m->m_data -= (ip->ip_hl << 2);
658	/* some debugging cruft by sklower, below, will go away soon */
659	if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
660		register int plen = 0;
661		for (t = m; t; t = t->m_next)
662			plen += t->m_len;
663		m->m_pkthdr.len = plen;
664	}
665	return (m);
666
667dropfrag:
668	ipstat.ips_fragdropped++;
669	m_freem(m);
670	FREE(ipqe, M_IPQ);
671	return (0);
672}
673
674/*
675 * Free a fragment reassembly header and all
676 * associated datagrams.
677 */
678void
679ip_freef(fp)
680	struct ipq *fp;
681{
682	register struct ipqent *q, *p;
683
684	for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
685		p = q->ipqe_q.le_next;
686		m_freem(q->ipqe_m);
687		LIST_REMOVE(q, ipqe_q);
688		FREE(q, M_IPQ);
689	}
690	LIST_REMOVE(fp, ipq_q);
691	FREE(fp, M_FTABLE);
692}
693
694/*
695 * IP timer processing;
696 * if a timer expires on a reassembly
697 * queue, discard it.
698 */
699void
700ip_slowtimo()
701{
702	register struct ipq *fp, *nfp;
703	int s = splsoftnet();
704
705	for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
706		nfp = fp->ipq_q.le_next;
707		if (--fp->ipq_ttl == 0) {
708			ipstat.ips_fragtimeout++;
709			ip_freef(fp);
710		}
711	}
712#ifdef GATEWAY
713	ipflow_slowtimo();
714#endif
715	splx(s);
716}
717
718/*
719 * Drain off all datagram fragments.
720 */
721void
722ip_drain()
723{
724
725	while (ipq.lh_first != NULL) {
726		ipstat.ips_fragdropped++;
727		ip_freef(ipq.lh_first);
728	}
729}
730
731/*
732 * Do option processing on a datagram,
733 * possibly discarding it if bad options are encountered,
734 * or forwarding it if source-routed.
735 * Returns 1 if packet has been forwarded/freed,
736 * 0 if the packet should be processed further.
737 */
738int
739ip_dooptions(m)
740	struct mbuf *m;
741{
742	register struct ip *ip = mtod(m, struct ip *);
743	register u_char *cp;
744	register struct ip_timestamp *ipt;
745	register struct in_ifaddr *ia;
746	int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
747	struct in_addr *sin, dst;
748	n_time ntime;
749
750	dst = ip->ip_dst;
751	cp = (u_char *)(ip + 1);
752	cnt = (ip->ip_hl << 2) - sizeof (struct ip);
753	for (; cnt > 0; cnt -= optlen, cp += optlen) {
754		opt = cp[IPOPT_OPTVAL];
755		if (opt == IPOPT_EOL)
756			break;
757		if (opt == IPOPT_NOP)
758			optlen = 1;
759		else {
760			optlen = cp[IPOPT_OLEN];
761			if (optlen <= 0 || optlen > cnt) {
762				code = &cp[IPOPT_OLEN] - (u_char *)ip;
763				goto bad;
764			}
765		}
766		switch (opt) {
767
768		default:
769			break;
770
771		/*
772		 * Source routing with record.
773		 * Find interface with current destination address.
774		 * If none on this machine then drop if strictly routed,
775		 * or do nothing if loosely routed.
776		 * Record interface address and bring up next address
777		 * component.  If strictly routed make sure next
778		 * address is on directly accessible net.
779		 */
780		case IPOPT_LSRR:
781		case IPOPT_SSRR:
782			if (ip_allowsrcrt == 0) {
783				type = ICMP_UNREACH;
784				code = ICMP_UNREACH_NET_PROHIB;
785				goto bad;
786			}
787			if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
788				code = &cp[IPOPT_OFFSET] - (u_char *)ip;
789				goto bad;
790			}
791			ipaddr.sin_addr = ip->ip_dst;
792			ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
793			if (ia == 0) {
794				if (opt == IPOPT_SSRR) {
795					type = ICMP_UNREACH;
796					code = ICMP_UNREACH_SRCFAIL;
797					goto bad;
798				}
799				/*
800				 * Loose routing, and not at next destination
801				 * yet; nothing to do except forward.
802				 */
803				break;
804			}
805			off--;			/* 0 origin */
806			if (off > optlen - sizeof(struct in_addr)) {
807				/*
808				 * End of source route.  Should be for us.
809				 */
810				save_rte(cp, ip->ip_src);
811				break;
812			}
813			/*
814			 * locate outgoing interface
815			 */
816			bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
817			    sizeof(ipaddr.sin_addr));
818			if (opt == IPOPT_SSRR) {
819#define	INA	struct in_ifaddr *
820#define	SA	struct sockaddr *
821			    ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
822			} else
823				ia = ip_rtaddr(ipaddr.sin_addr);
824			if (ia == 0) {
825				type = ICMP_UNREACH;
826				code = ICMP_UNREACH_SRCFAIL;
827				goto bad;
828			}
829			ip->ip_dst = ipaddr.sin_addr;
830			bcopy((caddr_t)&ia->ia_addr.sin_addr,
831			    (caddr_t)(cp + off), sizeof(struct in_addr));
832			cp[IPOPT_OFFSET] += sizeof(struct in_addr);
833			/*
834			 * Let ip_intr's mcast routing check handle mcast pkts
835			 */
836			forward = !IN_MULTICAST(ip->ip_dst.s_addr);
837			break;
838
839		case IPOPT_RR:
840			if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
841				code = &cp[IPOPT_OFFSET] - (u_char *)ip;
842				goto bad;
843			}
844			/*
845			 * If no space remains, ignore.
846			 */
847			off--;			/* 0 origin */
848			if (off > optlen - sizeof(struct in_addr))
849				break;
850			bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
851			    sizeof(ipaddr.sin_addr));
852			/*
853			 * locate outgoing interface; if we're the destination,
854			 * use the incoming interface (should be same).
855			 */
856			if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
857			    (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
858				type = ICMP_UNREACH;
859				code = ICMP_UNREACH_HOST;
860				goto bad;
861			}
862			bcopy((caddr_t)&ia->ia_addr.sin_addr,
863			    (caddr_t)(cp + off), sizeof(struct in_addr));
864			cp[IPOPT_OFFSET] += sizeof(struct in_addr);
865			break;
866
867		case IPOPT_TS:
868			code = cp - (u_char *)ip;
869			ipt = (struct ip_timestamp *)cp;
870			if (ipt->ipt_len < 5)
871				goto bad;
872			if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
873				if (++ipt->ipt_oflw == 0)
874					goto bad;
875				break;
876			}
877			sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
878			switch (ipt->ipt_flg) {
879
880			case IPOPT_TS_TSONLY:
881				break;
882
883			case IPOPT_TS_TSANDADDR:
884				if (ipt->ipt_ptr + sizeof(n_time) +
885				    sizeof(struct in_addr) > ipt->ipt_len)
886					goto bad;
887				ipaddr.sin_addr = dst;
888				ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
889							    m->m_pkthdr.rcvif);
890				if (ia == 0)
891					continue;
892				bcopy((caddr_t)&ia->ia_addr.sin_addr,
893				    (caddr_t)sin, sizeof(struct in_addr));
894				ipt->ipt_ptr += sizeof(struct in_addr);
895				break;
896
897			case IPOPT_TS_PRESPEC:
898				if (ipt->ipt_ptr + sizeof(n_time) +
899				    sizeof(struct in_addr) > ipt->ipt_len)
900					goto bad;
901				bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
902				    sizeof(struct in_addr));
903				if (ifa_ifwithaddr((SA)&ipaddr) == 0)
904					continue;
905				ipt->ipt_ptr += sizeof(struct in_addr);
906				break;
907
908			default:
909				goto bad;
910			}
911			ntime = iptime();
912			bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
913			    sizeof(n_time));
914			ipt->ipt_ptr += sizeof(n_time);
915		}
916	}
917	if (forward) {
918		if (ip_forwsrcrt == 0) {
919			type = ICMP_UNREACH;
920			code = ICMP_UNREACH_SRCFAIL;
921			goto bad;
922		}
923		ip_forward(m, 1);
924		return (1);
925	}
926	return (0);
927bad:
928	ip->ip_len -= ip->ip_hl << 2;   /* XXX icmp_error adds in hdr length */
929	icmp_error(m, type, code, 0, 0);
930	ipstat.ips_badoptions++;
931	return (1);
932}
933
934/*
935 * Given address of next destination (final or next hop),
936 * return internet address info of interface to be used to get there.
937 */
938struct in_ifaddr *
939ip_rtaddr(dst)
940	 struct in_addr dst;
941{
942	register struct sockaddr_in *sin;
943
944	sin = satosin(&ipforward_rt.ro_dst);
945
946	if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
947		if (ipforward_rt.ro_rt) {
948			RTFREE(ipforward_rt.ro_rt);
949			ipforward_rt.ro_rt = 0;
950		}
951		sin->sin_family = AF_INET;
952		sin->sin_len = sizeof(*sin);
953		sin->sin_addr = dst;
954
955		rtalloc(&ipforward_rt);
956	}
957	if (ipforward_rt.ro_rt == 0)
958		return ((struct in_ifaddr *)0);
959	return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
960}
961
962/*
963 * Save incoming source route for use in replies,
964 * to be picked up later by ip_srcroute if the receiver is interested.
965 */
966void
967save_rte(option, dst)
968	u_char *option;
969	struct in_addr dst;
970{
971	unsigned olen;
972
973	olen = option[IPOPT_OLEN];
974#ifdef DIAGNOSTIC
975	if (ipprintfs)
976		printf("save_rte: olen %d\n", olen);
977#endif
978	if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
979		return;
980	bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
981	ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
982	ip_srcrt.dst = dst;
983}
984
985/*
986 * Retrieve incoming source route for use in replies,
987 * in the same form used by setsockopt.
988 * The first hop is placed before the options, will be removed later.
989 */
990struct mbuf *
991ip_srcroute()
992{
993	register struct in_addr *p, *q;
994	register struct mbuf *m;
995
996	if (ip_nhops == 0)
997		return ((struct mbuf *)0);
998	m = m_get(M_DONTWAIT, MT_SOOPTS);
999	if (m == 0)
1000		return ((struct mbuf *)0);
1001
1002#define OPTSIZ	(sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1003
1004	/* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1005	m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1006	    OPTSIZ;
1007#ifdef DIAGNOSTIC
1008	if (ipprintfs)
1009		printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1010#endif
1011
1012	/*
1013	 * First save first hop for return route
1014	 */
1015	p = &ip_srcrt.route[ip_nhops - 1];
1016	*(mtod(m, struct in_addr *)) = *p--;
1017#ifdef DIAGNOSTIC
1018	if (ipprintfs)
1019		printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1020#endif
1021
1022	/*
1023	 * Copy option fields and padding (nop) to mbuf.
1024	 */
1025	ip_srcrt.nop = IPOPT_NOP;
1026	ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1027	bcopy((caddr_t)&ip_srcrt.nop,
1028	    mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1029	q = (struct in_addr *)(mtod(m, caddr_t) +
1030	    sizeof(struct in_addr) + OPTSIZ);
1031#undef OPTSIZ
1032	/*
1033	 * Record return path as an IP source route,
1034	 * reversing the path (pointers are now aligned).
1035	 */
1036	while (p >= ip_srcrt.route) {
1037#ifdef DIAGNOSTIC
1038		if (ipprintfs)
1039			printf(" %x", ntohl(q->s_addr));
1040#endif
1041		*q++ = *p--;
1042	}
1043	/*
1044	 * Last hop goes to final destination.
1045	 */
1046	*q = ip_srcrt.dst;
1047#ifdef DIAGNOSTIC
1048	if (ipprintfs)
1049		printf(" %x\n", ntohl(q->s_addr));
1050#endif
1051	return (m);
1052}
1053
1054/*
1055 * Strip out IP options, at higher
1056 * level protocol in the kernel.
1057 * Second argument is buffer to which options
1058 * will be moved, and return value is their length.
1059 * XXX should be deleted; last arg currently ignored.
1060 */
1061void
1062ip_stripoptions(m, mopt)
1063	register struct mbuf *m;
1064	struct mbuf *mopt;
1065{
1066	register int i;
1067	struct ip *ip = mtod(m, struct ip *);
1068	register caddr_t opts;
1069	int olen;
1070
1071	olen = (ip->ip_hl<<2) - sizeof (struct ip);
1072	opts = (caddr_t)(ip + 1);
1073	i = m->m_len - (sizeof (struct ip) + olen);
1074	bcopy(opts  + olen, opts, (unsigned)i);
1075	m->m_len -= olen;
1076	if (m->m_flags & M_PKTHDR)
1077		m->m_pkthdr.len -= olen;
1078	ip->ip_hl = sizeof(struct ip) >> 2;
1079}
1080
1081int inetctlerrmap[PRC_NCMDS] = {
1082	0,		0,		0,		0,
1083	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1084	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1085	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1086	0,		0,		0,		0,
1087	ENOPROTOOPT
1088};
1089
1090/*
1091 * Forward a packet.  If some error occurs return the sender
1092 * an icmp packet.  Note we can't always generate a meaningful
1093 * icmp message because icmp doesn't have a large enough repertoire
1094 * of codes and types.
1095 *
1096 * If not forwarding, just drop the packet.  This could be confusing
1097 * if ipforwarding was zero but some routing protocol was advancing
1098 * us as a gateway to somewhere.  However, we must let the routing
1099 * protocol deal with that.
1100 *
1101 * The srcrt parameter indicates whether the packet is being forwarded
1102 * via a source route.
1103 */
1104void
1105ip_forward(m, srcrt)
1106	struct mbuf *m;
1107	int srcrt;
1108{
1109	register struct ip *ip = mtod(m, struct ip *);
1110	register struct sockaddr_in *sin;
1111	register struct rtentry *rt;
1112	int error, type = 0, code = 0;
1113	struct mbuf *mcopy;
1114	n_long dest;
1115	struct ifnet *destifp;
1116
1117	dest = 0;
1118#ifdef DIAGNOSTIC
1119	if (ipprintfs)
1120		printf("forward: src %x dst %x ttl %x\n",
1121		    ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl);
1122#endif
1123	if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1124		ipstat.ips_cantforward++;
1125		m_freem(m);
1126		return;
1127	}
1128	HTONS(ip->ip_id);
1129	if (ip->ip_ttl <= IPTTLDEC) {
1130		icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1131		return;
1132	}
1133	ip->ip_ttl -= IPTTLDEC;
1134
1135	sin = satosin(&ipforward_rt.ro_dst);
1136	if ((rt = ipforward_rt.ro_rt) == 0 ||
1137	    !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1138		if (ipforward_rt.ro_rt) {
1139			RTFREE(ipforward_rt.ro_rt);
1140			ipforward_rt.ro_rt = 0;
1141		}
1142		sin->sin_family = AF_INET;
1143		sin->sin_len = sizeof(struct sockaddr_in);
1144		sin->sin_addr = ip->ip_dst;
1145
1146		rtalloc(&ipforward_rt);
1147		if (ipforward_rt.ro_rt == 0) {
1148			icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1149			return;
1150		}
1151		rt = ipforward_rt.ro_rt;
1152	}
1153
1154	/*
1155	 * Save at most 68 bytes of the packet in case
1156	 * we need to generate an ICMP message to the src.
1157	 */
1158	mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1159
1160	/*
1161	 * If forwarding packet using same interface that it came in on,
1162	 * perhaps should send a redirect to sender to shortcut a hop.
1163	 * Only send redirect if source is sending directly to us,
1164	 * and if packet was not source routed (or has any options).
1165	 * Also, don't send redirect if forwarding using a default route
1166	 * or a route modified by a redirect.
1167	 */
1168	if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1169	    (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1170	    !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1171	    ipsendredirects && !srcrt) {
1172		if (rt->rt_ifa &&
1173		    (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1174		    ifatoia(rt->rt_ifa)->ia_subnet) {
1175		    if (rt->rt_flags & RTF_GATEWAY)
1176			dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1177		    else
1178			dest = ip->ip_dst.s_addr;
1179		    /* Router requirements says to only send host redirects */
1180		    type = ICMP_REDIRECT;
1181		    code = ICMP_REDIRECT_HOST;
1182#ifdef DIAGNOSTIC
1183		    if (ipprintfs)
1184		    	printf("redirect (%d) to %x\n", code, (u_int32_t)dest);
1185#endif
1186		}
1187	}
1188
1189	error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1190	    (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1191	if (error)
1192		ipstat.ips_cantforward++;
1193	else {
1194		ipstat.ips_forward++;
1195		if (type)
1196			ipstat.ips_redirectsent++;
1197		else {
1198			if (mcopy) {
1199#ifdef GATEWAY
1200				ipflow_create(&ipforward_rt, mcopy);
1201#endif
1202				m_freem(mcopy);
1203			}
1204			return;
1205		}
1206	}
1207	if (mcopy == NULL)
1208		return;
1209	destifp = NULL;
1210
1211	switch (error) {
1212
1213	case 0:				/* forwarded, but need redirect */
1214		/* type, code set above */
1215		break;
1216
1217	case ENETUNREACH:		/* shouldn't happen, checked above */
1218	case EHOSTUNREACH:
1219	case ENETDOWN:
1220	case EHOSTDOWN:
1221	default:
1222		type = ICMP_UNREACH;
1223		code = ICMP_UNREACH_HOST;
1224		break;
1225
1226	case EMSGSIZE:
1227		type = ICMP_UNREACH;
1228		code = ICMP_UNREACH_NEEDFRAG;
1229		if (ipforward_rt.ro_rt)
1230			destifp = ipforward_rt.ro_rt->rt_ifp;
1231		ipstat.ips_cantfrag++;
1232		break;
1233
1234	case ENOBUFS:
1235		type = ICMP_SOURCEQUENCH;
1236		code = 0;
1237		break;
1238	}
1239	icmp_error(mcopy, type, code, dest, destifp);
1240}
1241
1242void
1243ip_savecontrol(inp, mp, ip, m)
1244	register struct inpcb *inp;
1245	register struct mbuf **mp;
1246	register struct ip *ip;
1247	register struct mbuf *m;
1248{
1249
1250	if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1251		struct timeval tv;
1252
1253		microtime(&tv);
1254		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1255		    SCM_TIMESTAMP, SOL_SOCKET);
1256		if (*mp)
1257			mp = &(*mp)->m_next;
1258	}
1259	if (inp->inp_flags & INP_RECVDSTADDR) {
1260		*mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1261		    sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1262		if (*mp)
1263			mp = &(*mp)->m_next;
1264	}
1265#ifdef notyet
1266	/*
1267	 * XXX
1268	 * Moving these out of udp_input() made them even more broken
1269	 * than they already were.
1270	 *	- fenner@parc.xerox.com
1271	 */
1272	/* options were tossed already */
1273	if (inp->inp_flags & INP_RECVOPTS) {
1274		*mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1275		    sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1276		if (*mp)
1277			mp = &(*mp)->m_next;
1278	}
1279	/* ip_srcroute doesn't do what we want here, need to fix */
1280	if (inp->inp_flags & INP_RECVRETOPTS) {
1281		*mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1282		    sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1283		if (*mp)
1284			mp = &(*mp)->m_next;
1285	}
1286#endif
1287	if (inp->inp_flags & INP_RECVIF) {
1288		struct sockaddr_dl sdl;
1289
1290		sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1291		sdl.sdl_family = AF_LINK;
1292		sdl.sdl_index = m->m_pkthdr.rcvif ?
1293		    m->m_pkthdr.rcvif->if_index : 0;
1294		sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1295		*mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1296		    IP_RECVIF, IPPROTO_IP);
1297		if (*mp)
1298			mp = &(*mp)->m_next;
1299	}
1300}
1301
1302int
1303ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1304	int *name;
1305	u_int namelen;
1306	void *oldp;
1307	size_t *oldlenp;
1308	void *newp;
1309	size_t newlen;
1310{
1311	extern int subnetsarelocal;
1312
1313	int error, old;
1314
1315	/* All sysctl names at this level are terminal. */
1316	if (namelen != 1)
1317		return (ENOTDIR);
1318
1319	switch (name[0]) {
1320	case IPCTL_FORWARDING:
1321		return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1322	case IPCTL_SENDREDIRECTS:
1323		return (sysctl_int(oldp, oldlenp, newp, newlen,
1324			&ipsendredirects));
1325	case IPCTL_DEFTTL:
1326		return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1327#ifdef notyet
1328	case IPCTL_DEFMTU:
1329		return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1330#endif
1331	case IPCTL_FORWSRCRT:
1332		/* Don't allow this to change in a secure environment.  */
1333		if (securelevel > 0)
1334			return (sysctl_rdint(oldp, oldlenp, newp,
1335			    ip_forwsrcrt));
1336		else
1337			return (sysctl_int(oldp, oldlenp, newp, newlen,
1338			    &ip_forwsrcrt));
1339	case IPCTL_DIRECTEDBCAST:
1340		return (sysctl_int(oldp, oldlenp, newp, newlen,
1341		    &ip_directedbcast));
1342	case IPCTL_ALLOWSRCRT:
1343		return (sysctl_int(oldp, oldlenp, newp, newlen,
1344		    &ip_allowsrcrt));
1345	case IPCTL_SUBNETSARELOCAL:
1346		return (sysctl_int(oldp, oldlenp, newp, newlen,
1347		    &subnetsarelocal));
1348	case IPCTL_MTUDISC:
1349		error = sysctl_int(oldp, oldlenp, newp, newlen,
1350		    &ip_mtudisc);
1351		if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1352			ip_mtudisc_timeout_q =
1353			    rt_timer_queue_create(ip_mtudisc_timeout);
1354		} else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1355			rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1356			ip_mtudisc_timeout_q = NULL;
1357		}
1358		return error;
1359	case IPCTL_ANONPORTMIN:
1360		old = anonportmin;
1361		error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1362		if (anonportmin >= anonportmax || anonportmin > 65535
1363#ifndef IPNOPRIVPORTS
1364		    || anonportmin < IPPORT_RESERVED
1365#endif
1366		    ) {
1367			anonportmin = old;
1368			return (EINVAL);
1369		}
1370		return (error);
1371	case IPCTL_ANONPORTMAX:
1372		old = anonportmax;
1373		error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1374		if (anonportmin >= anonportmax || anonportmax > 65535
1375#ifndef IPNOPRIVPORTS
1376		    || anonportmax < IPPORT_RESERVED
1377#endif
1378		    ) {
1379			anonportmax = old;
1380			return (EINVAL);
1381		}
1382		return (error);
1383	case IPCTL_MTUDISCTIMEOUT:
1384		error = sysctl_int(oldp, oldlenp, newp, newlen,
1385		   &ip_mtudisc_timeout);
1386		if (ip_mtudisc_timeout_q != NULL)
1387			rt_timer_queue_change(ip_mtudisc_timeout_q,
1388					      ip_mtudisc_timeout);
1389		return (error);
1390	default:
1391		return (EOPNOTSUPP);
1392	}
1393	/* NOTREACHED */
1394}
1395