ip_input.c revision 1.60
1/*	$NetBSD: ip_input.c,v 1.60 1998/04/29 03:44:11 kml 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_mrouting.h"
76
77#include <sys/param.h>
78#include <sys/systm.h>
79#include <sys/malloc.h>
80#include <sys/mbuf.h>
81#include <sys/domain.h>
82#include <sys/protosw.h>
83#include <sys/socket.h>
84#include <sys/socketvar.h>
85#include <sys/errno.h>
86#include <sys/time.h>
87#include <sys/kernel.h>
88#include <sys/proc.h>
89
90#include <vm/vm.h>
91#include <sys/sysctl.h>
92
93#include <net/if.h>
94#include <net/if_dl.h>
95#include <net/route.h>
96#include <net/pfil.h>
97
98#include <netinet/in.h>
99#include <netinet/in_systm.h>
100#include <netinet/ip.h>
101#include <netinet/in_pcb.h>
102#include <netinet/in_var.h>
103#include <netinet/ip_var.h>
104#include <netinet/ip_icmp.h>
105
106#ifndef	IPFORWARDING
107#ifdef GATEWAY
108#define	IPFORWARDING	1	/* forward IP packets not for us */
109#else /* GATEWAY */
110#define	IPFORWARDING	0	/* don't forward IP packets not for us */
111#endif /* GATEWAY */
112#endif /* IPFORWARDING */
113#ifndef	IPSENDREDIRECTS
114#define	IPSENDREDIRECTS	1
115#endif
116#ifndef IPFORWSRCRT
117#define	IPFORWSRCRT	1	/* forward source-routed packets */
118#endif
119#ifndef IPALLOWSRCRT
120#define	IPALLOWSRCRT	1	/* allow source-routed packets */
121#endif
122#ifndef IPMTUDISC
123#define IPMTUDISC	0
124#endif
125#ifndef IPMTUDISCTIMEOUT
126#define IPMTUDISCTIMEOUT (1 * 60)	/* experimental short timeout value! */
127#endif
128
129/*
130 * Note: DIRECTED_BROADCAST is handled this way so that previous
131 * configuration using this option will Just Work.
132 */
133#ifndef IPDIRECTEDBCAST
134#ifdef DIRECTED_BROADCAST
135#define IPDIRECTEDBCAST	1
136#else
137#define	IPDIRECTEDBCAST	0
138#endif /* DIRECTED_BROADCAST */
139#endif /* IPDIRECTEDBCAST */
140int	ipforwarding = IPFORWARDING;
141int	ipsendredirects = IPSENDREDIRECTS;
142int	ip_defttl = IPDEFTTL;
143int	ip_forwsrcrt = IPFORWSRCRT;
144int	ip_directedbcast = IPDIRECTEDBCAST;
145int	ip_allowsrcrt = IPALLOWSRCRT;
146int	ip_mtudisc = IPMTUDISC;
147u_int	ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
148#ifdef DIAGNOSTIC
149int	ipprintfs = 0;
150#endif
151
152struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
153
154extern	struct domain inetdomain;
155extern	struct protosw inetsw[];
156u_char	ip_protox[IPPROTO_MAX];
157int	ipqmaxlen = IFQ_MAXLEN;
158struct	in_ifaddrhead in_ifaddr;
159struct	in_ifaddrhashhead *in_ifaddrhashtbl;
160struct	ifqueue ipintrq;
161
162/*
163 * We need to save the IP options in case a protocol wants to respond
164 * to an incoming packet over the same route if the packet got here
165 * using IP source routing.  This allows connection establishment and
166 * maintenance when the remote end is on a network that is not known
167 * to us.
168 */
169int	ip_nhops = 0;
170static	struct ip_srcrt {
171	struct	in_addr dst;			/* final destination */
172	char	nop;				/* one NOP to align */
173	char	srcopt[IPOPT_OFFSET + 1];	/* OPTVAL, OLEN and OFFSET */
174	struct	in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
175} ip_srcrt;
176
177static void save_rte __P((u_char *, struct in_addr));
178
179/*
180 * IP initialization: fill in IP protocol switch table.
181 * All protocols not implemented in kernel go to raw IP protocol handler.
182 */
183void
184ip_init()
185{
186	register struct protosw *pr;
187	register int i;
188
189	pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
190	if (pr == 0)
191		panic("ip_init");
192	for (i = 0; i < IPPROTO_MAX; i++)
193		ip_protox[i] = pr - inetsw;
194	for (pr = inetdomain.dom_protosw;
195	    pr < inetdomain.dom_protoswNPROTOSW; pr++)
196		if (pr->pr_domain->dom_family == PF_INET &&
197		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
198			ip_protox[pr->pr_protocol] = pr - inetsw;
199	LIST_INIT(&ipq);
200	ip_id = time.tv_sec & 0xffff;
201	ipintrq.ifq_maxlen = ipqmaxlen;
202	TAILQ_INIT(&in_ifaddr);
203	in_ifaddrhashtbl =
204	    hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, M_WAITOK, &in_ifaddrhash);
205	if (ip_mtudisc != 0)
206		ip_mtudisc_timeout_q =
207		    rt_timer_queue_create(ip_mtudisc_timeout);
208}
209
210struct	sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET };
211struct	route ipforward_rt;
212
213/*
214 * Ip input routine.  Checksum and byte swap header.  If fragmented
215 * try to reassemble.  Process options.  Pass to next level.
216 */
217void
218ipintr()
219{
220	register struct ip *ip = NULL;
221	register struct mbuf *m;
222	register struct ipq *fp;
223	register struct in_ifaddr *ia;
224	register struct ifaddr *ifa;
225	struct ipqent *ipqe;
226	int hlen = 0, mff, len, s;
227#ifdef PFIL_HOOKS
228	struct packet_filter_hook *pfh;
229	struct mbuf *m0;
230	int rv;
231#endif /* PFIL_HOOKS */
232
233next:
234	/*
235	 * Get next datagram off input queue and get IP header
236	 * in first mbuf.
237	 */
238	s = splimp();
239	IF_DEQUEUE(&ipintrq, m);
240	splx(s);
241	if (m == 0)
242		return;
243#ifdef	DIAGNOSTIC
244	if ((m->m_flags & M_PKTHDR) == 0)
245		panic("ipintr no HDR");
246#endif
247	/*
248	 * If no IP addresses have been set yet but the interfaces
249	 * are receiving, can't do anything with incoming packets yet.
250	 */
251	if (in_ifaddr.tqh_first == 0)
252		goto bad;
253	ipstat.ips_total++;
254	if (m->m_len < sizeof (struct ip) &&
255	    (m = m_pullup(m, sizeof (struct ip))) == 0) {
256		ipstat.ips_toosmall++;
257		goto next;
258	}
259	ip = mtod(m, struct ip *);
260	if (ip->ip_v != IPVERSION) {
261		ipstat.ips_badvers++;
262		goto bad;
263	}
264	hlen = ip->ip_hl << 2;
265	if (hlen < sizeof(struct ip)) {	/* minimum header length */
266		ipstat.ips_badhlen++;
267		goto bad;
268	}
269	if (hlen > m->m_len) {
270		if ((m = m_pullup(m, hlen)) == 0) {
271			ipstat.ips_badhlen++;
272			goto next;
273		}
274		ip = mtod(m, struct ip *);
275	}
276	if ((ip->ip_sum = in_cksum(m, hlen)) != 0) {
277		ipstat.ips_badsum++;
278		goto bad;
279	}
280
281	/*
282	 * Convert fields to host representation.
283	 */
284	NTOHS(ip->ip_len);
285	NTOHS(ip->ip_id);
286	NTOHS(ip->ip_off);
287	len = ip->ip_len;
288
289	/*
290	 * Check that the amount of data in the buffers
291	 * is as at least much as the IP header would have us expect.
292	 * Trim mbufs if longer than we expect.
293	 * Drop packet if shorter than we expect.
294	 */
295	if (m->m_pkthdr.len < len) {
296		ipstat.ips_tooshort++;
297		goto bad;
298	}
299	if (m->m_pkthdr.len > len) {
300		if (m->m_len == m->m_pkthdr.len) {
301			m->m_len = len;
302			m->m_pkthdr.len = len;
303		} else
304			m_adj(m, len - m->m_pkthdr.len);
305	}
306
307#ifdef PFIL_HOOKS
308	/*
309	 * Run through list of hooks for input packets.
310	 */
311	m0 = m;
312	for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
313		if (pfh->pfil_func) {
314			rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
315			if (rv)
316				goto next;
317			ip = mtod(m = m0, struct ip *);
318		}
319#endif /* PFIL_HOOKS */
320
321	/*
322	 * Process options and, if not destined for us,
323	 * ship it on.  ip_dooptions returns 1 when an
324	 * error was detected (causing an icmp message
325	 * to be sent and the original packet to be freed).
326	 */
327	ip_nhops = 0;		/* for source routed packets */
328	if (hlen > sizeof (struct ip) && ip_dooptions(m))
329		goto next;
330
331	/*
332	 * Check our list of addresses, to see if the packet is for us.
333	 */
334	INADDR_TO_IA(ip->ip_dst, ia);
335	if (ia != NULL) goto ours;
336	if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
337		for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
338		    ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
339			if (ifa->ifa_addr->sa_family != AF_INET) continue;
340			ia = ifatoia(ifa);
341			if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
342			    in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
343			    /*
344			     * Look for all-0's host part (old broadcast addr),
345			     * either for subnet or net.
346			     */
347			    ip->ip_dst.s_addr == ia->ia_subnet ||
348			    ip->ip_dst.s_addr == ia->ia_net)
349				goto ours;
350			/*
351			 * An interface with IP address zero accepts
352			 * all packets that arrive on that interface.
353			 */
354			if (in_nullhost(ia->ia_addr.sin_addr))
355				goto ours;
356		}
357	}
358	if (IN_MULTICAST(ip->ip_dst.s_addr)) {
359		struct in_multi *inm;
360#ifdef MROUTING
361		extern struct socket *ip_mrouter;
362
363		if (m->m_flags & M_EXT) {
364			if ((m = m_pullup(m, hlen)) == 0) {
365				ipstat.ips_toosmall++;
366				goto next;
367			}
368			ip = mtod(m, struct ip *);
369		}
370
371		if (ip_mrouter) {
372			/*
373			 * If we are acting as a multicast router, all
374			 * incoming multicast packets are passed to the
375			 * kernel-level multicast forwarding function.
376			 * The packet is returned (relatively) intact; if
377			 * ip_mforward() returns a non-zero value, the packet
378			 * must be discarded, else it may be accepted below.
379			 *
380			 * (The IP ident field is put in the same byte order
381			 * as expected when ip_mforward() is called from
382			 * ip_output().)
383			 */
384			ip->ip_id = htons(ip->ip_id);
385			if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
386				ipstat.ips_cantforward++;
387				m_freem(m);
388				goto next;
389			}
390			ip->ip_id = ntohs(ip->ip_id);
391
392			/*
393			 * The process-level routing demon needs to receive
394			 * all multicast IGMP packets, whether or not this
395			 * host belongs to their destination groups.
396			 */
397			if (ip->ip_p == IPPROTO_IGMP)
398				goto ours;
399			ipstat.ips_forward++;
400		}
401#endif
402		/*
403		 * See if we belong to the destination multicast group on the
404		 * arrival interface.
405		 */
406		IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
407		if (inm == NULL) {
408			ipstat.ips_cantforward++;
409			m_freem(m);
410			goto next;
411		}
412		goto ours;
413	}
414	if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
415	    in_nullhost(ip->ip_dst))
416		goto ours;
417
418	/*
419	 * Not for us; forward if possible and desirable.
420	 */
421	if (ipforwarding == 0) {
422		ipstat.ips_cantforward++;
423		m_freem(m);
424	} else
425		ip_forward(m, 0);
426	goto next;
427
428ours:
429	/*
430	 * If offset or IP_MF are set, must reassemble.
431	 * Otherwise, nothing need be done.
432	 * (We could look in the reassembly queue to see
433	 * if the packet was previously fragmented,
434	 * but it's not worth the time; just let them time out.)
435	 */
436	if (ip->ip_off & ~(IP_DF|IP_RF)) {
437		/*
438		 * Look for queue of fragments
439		 * of this datagram.
440		 */
441		for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
442			if (ip->ip_id == fp->ipq_id &&
443			    in_hosteq(ip->ip_src, fp->ipq_src) &&
444			    in_hosteq(ip->ip_dst, fp->ipq_dst) &&
445			    ip->ip_p == fp->ipq_p)
446				goto found;
447		fp = 0;
448found:
449
450		/*
451		 * Adjust ip_len to not reflect header,
452		 * set ipqe_mff if more fragments are expected,
453		 * convert offset of this to bytes.
454		 */
455		ip->ip_len -= hlen;
456		mff = (ip->ip_off & IP_MF) != 0;
457		if (mff) {
458		        /*
459		         * Make sure that fragments have a data length
460			 * that's a non-zero multiple of 8 bytes.
461		         */
462			if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
463				ipstat.ips_badfrags++;
464				goto bad;
465			}
466		}
467		ip->ip_off <<= 3;
468
469		/*
470		 * If datagram marked as having more fragments
471		 * or if this is not the first fragment,
472		 * attempt reassembly; if it succeeds, proceed.
473		 */
474		if (mff || ip->ip_off) {
475			ipstat.ips_fragments++;
476			MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent),
477			    M_IPQ, M_NOWAIT);
478			if (ipqe == NULL) {
479				ipstat.ips_rcvmemdrop++;
480				goto bad;
481			}
482			ipqe->ipqe_mff = mff;
483			ipqe->ipqe_m = m;
484			ipqe->ipqe_ip = ip;
485			m = ip_reass(ipqe, fp);
486			if (m == 0)
487				goto next;
488			ipstat.ips_reassembled++;
489			ip = mtod(m, struct ip *);
490		} else
491			if (fp)
492				ip_freef(fp);
493	} else
494		ip->ip_len -= hlen;
495
496	/*
497	 * Switch out to protocol's input routine.
498	 */
499	ipstat.ips_delivered++;
500	(*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
501	goto next;
502bad:
503	m_freem(m);
504	goto next;
505}
506
507/*
508 * Take incoming datagram fragment and try to
509 * reassemble it into whole datagram.  If a chain for
510 * reassembly of this datagram already exists, then it
511 * is given as fp; otherwise have to make a chain.
512 */
513struct mbuf *
514ip_reass(ipqe, fp)
515	register struct ipqent *ipqe;
516	register struct ipq *fp;
517{
518	register struct mbuf *m = ipqe->ipqe_m;
519	register struct ipqent *nq, *p, *q;
520	struct ip *ip;
521	struct mbuf *t;
522	int hlen = ipqe->ipqe_ip->ip_hl << 2;
523	int i, next;
524
525	/*
526	 * Presence of header sizes in mbufs
527	 * would confuse code below.
528	 */
529	m->m_data += hlen;
530	m->m_len -= hlen;
531
532	/*
533	 * If first fragment to arrive, create a reassembly queue.
534	 */
535	if (fp == 0) {
536		MALLOC(fp, struct ipq *, sizeof (struct ipq),
537		    M_FTABLE, M_NOWAIT);
538		if (fp == NULL)
539			goto dropfrag;
540		LIST_INSERT_HEAD(&ipq, fp, ipq_q);
541		fp->ipq_ttl = IPFRAGTTL;
542		fp->ipq_p = ipqe->ipqe_ip->ip_p;
543		fp->ipq_id = ipqe->ipqe_ip->ip_id;
544		LIST_INIT(&fp->ipq_fragq);
545		fp->ipq_src = ipqe->ipqe_ip->ip_src;
546		fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
547		p = NULL;
548		goto insert;
549	}
550
551	/*
552	 * Find a segment which begins after this one does.
553	 */
554	for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
555	    p = q, q = q->ipqe_q.le_next)
556		if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
557			break;
558
559	/*
560	 * If there is a preceding segment, it may provide some of
561	 * our data already.  If so, drop the data from the incoming
562	 * segment.  If it provides all of our data, drop us.
563	 */
564	if (p != NULL) {
565		i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
566		    ipqe->ipqe_ip->ip_off;
567		if (i > 0) {
568			if (i >= ipqe->ipqe_ip->ip_len)
569				goto dropfrag;
570			m_adj(ipqe->ipqe_m, i);
571			ipqe->ipqe_ip->ip_off += i;
572			ipqe->ipqe_ip->ip_len -= i;
573		}
574	}
575
576	/*
577	 * While we overlap succeeding segments trim them or,
578	 * if they are completely covered, dequeue them.
579	 */
580	for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
581	    q->ipqe_ip->ip_off; q = nq) {
582		i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
583		    q->ipqe_ip->ip_off;
584		if (i < q->ipqe_ip->ip_len) {
585			q->ipqe_ip->ip_len -= i;
586			q->ipqe_ip->ip_off += i;
587			m_adj(q->ipqe_m, i);
588			break;
589		}
590		nq = q->ipqe_q.le_next;
591		m_freem(q->ipqe_m);
592		LIST_REMOVE(q, ipqe_q);
593		FREE(q, M_IPQ);
594	}
595
596insert:
597	/*
598	 * Stick new segment in its place;
599	 * check for complete reassembly.
600	 */
601	if (p == NULL) {
602		LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
603	} else {
604		LIST_INSERT_AFTER(p, ipqe, ipqe_q);
605	}
606	next = 0;
607	for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
608	    p = q, q = q->ipqe_q.le_next) {
609		if (q->ipqe_ip->ip_off != next)
610			return (0);
611		next += q->ipqe_ip->ip_len;
612	}
613	if (p->ipqe_mff)
614		return (0);
615
616	/*
617	 * Reassembly is complete.  Check for a bogus message size and
618	 * concatenate fragments.
619	 */
620	q = fp->ipq_fragq.lh_first;
621	ip = q->ipqe_ip;
622	if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
623		ipstat.ips_toolong++;
624		ip_freef(fp);
625		return (0);
626	}
627	m = q->ipqe_m;
628	t = m->m_next;
629	m->m_next = 0;
630	m_cat(m, t);
631	nq = q->ipqe_q.le_next;
632	FREE(q, M_IPQ);
633	for (q = nq; q != NULL; q = nq) {
634		t = q->ipqe_m;
635		nq = q->ipqe_q.le_next;
636		FREE(q, M_IPQ);
637		m_cat(m, t);
638	}
639
640	/*
641	 * Create header for new ip packet by
642	 * modifying header of first packet;
643	 * dequeue and discard fragment reassembly header.
644	 * Make header visible.
645	 */
646	ip->ip_len = next;
647	ip->ip_src = fp->ipq_src;
648	ip->ip_dst = fp->ipq_dst;
649	LIST_REMOVE(fp, ipq_q);
650	FREE(fp, M_FTABLE);
651	m->m_len += (ip->ip_hl << 2);
652	m->m_data -= (ip->ip_hl << 2);
653	/* some debugging cruft by sklower, below, will go away soon */
654	if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
655		register int plen = 0;
656		for (t = m; t; t = t->m_next)
657			plen += t->m_len;
658		m->m_pkthdr.len = plen;
659	}
660	return (m);
661
662dropfrag:
663	ipstat.ips_fragdropped++;
664	m_freem(m);
665	FREE(ipqe, M_IPQ);
666	return (0);
667}
668
669/*
670 * Free a fragment reassembly header and all
671 * associated datagrams.
672 */
673void
674ip_freef(fp)
675	struct ipq *fp;
676{
677	register struct ipqent *q, *p;
678
679	for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
680		p = q->ipqe_q.le_next;
681		m_freem(q->ipqe_m);
682		LIST_REMOVE(q, ipqe_q);
683		FREE(q, M_IPQ);
684	}
685	LIST_REMOVE(fp, ipq_q);
686	FREE(fp, M_FTABLE);
687}
688
689/*
690 * IP timer processing;
691 * if a timer expires on a reassembly
692 * queue, discard it.
693 */
694void
695ip_slowtimo()
696{
697	register struct ipq *fp, *nfp;
698	int s = splsoftnet();
699
700	for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
701		nfp = fp->ipq_q.le_next;
702		if (--fp->ipq_ttl == 0) {
703			ipstat.ips_fragtimeout++;
704			ip_freef(fp);
705		}
706	}
707	splx(s);
708}
709
710/*
711 * Drain off all datagram fragments.
712 */
713void
714ip_drain()
715{
716
717	while (ipq.lh_first != NULL) {
718		ipstat.ips_fragdropped++;
719		ip_freef(ipq.lh_first);
720	}
721}
722
723/*
724 * Do option processing on a datagram,
725 * possibly discarding it if bad options are encountered,
726 * or forwarding it if source-routed.
727 * Returns 1 if packet has been forwarded/freed,
728 * 0 if the packet should be processed further.
729 */
730int
731ip_dooptions(m)
732	struct mbuf *m;
733{
734	register struct ip *ip = mtod(m, struct ip *);
735	register u_char *cp;
736	register struct ip_timestamp *ipt;
737	register struct in_ifaddr *ia;
738	int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
739	struct in_addr *sin, dst;
740	n_time ntime;
741
742	dst = ip->ip_dst;
743	cp = (u_char *)(ip + 1);
744	cnt = (ip->ip_hl << 2) - sizeof (struct ip);
745	for (; cnt > 0; cnt -= optlen, cp += optlen) {
746		opt = cp[IPOPT_OPTVAL];
747		if (opt == IPOPT_EOL)
748			break;
749		if (opt == IPOPT_NOP)
750			optlen = 1;
751		else {
752			optlen = cp[IPOPT_OLEN];
753			if (optlen <= 0 || optlen > cnt) {
754				code = &cp[IPOPT_OLEN] - (u_char *)ip;
755				goto bad;
756			}
757		}
758		switch (opt) {
759
760		default:
761			break;
762
763		/*
764		 * Source routing with record.
765		 * Find interface with current destination address.
766		 * If none on this machine then drop if strictly routed,
767		 * or do nothing if loosely routed.
768		 * Record interface address and bring up next address
769		 * component.  If strictly routed make sure next
770		 * address is on directly accessible net.
771		 */
772		case IPOPT_LSRR:
773		case IPOPT_SSRR:
774			if (ip_allowsrcrt == 0) {
775				type = ICMP_UNREACH;
776				code = ICMP_UNREACH_NET_PROHIB;
777				goto bad;
778			}
779			if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
780				code = &cp[IPOPT_OFFSET] - (u_char *)ip;
781				goto bad;
782			}
783			ipaddr.sin_addr = ip->ip_dst;
784			ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
785			if (ia == 0) {
786				if (opt == IPOPT_SSRR) {
787					type = ICMP_UNREACH;
788					code = ICMP_UNREACH_SRCFAIL;
789					goto bad;
790				}
791				/*
792				 * Loose routing, and not at next destination
793				 * yet; nothing to do except forward.
794				 */
795				break;
796			}
797			off--;			/* 0 origin */
798			if (off > optlen - sizeof(struct in_addr)) {
799				/*
800				 * End of source route.  Should be for us.
801				 */
802				save_rte(cp, ip->ip_src);
803				break;
804			}
805			/*
806			 * locate outgoing interface
807			 */
808			bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
809			    sizeof(ipaddr.sin_addr));
810			if (opt == IPOPT_SSRR) {
811#define	INA	struct in_ifaddr *
812#define	SA	struct sockaddr *
813			    ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
814			} else
815				ia = ip_rtaddr(ipaddr.sin_addr);
816			if (ia == 0) {
817				type = ICMP_UNREACH;
818				code = ICMP_UNREACH_SRCFAIL;
819				goto bad;
820			}
821			ip->ip_dst = ipaddr.sin_addr;
822			bcopy((caddr_t)&ia->ia_addr.sin_addr,
823			    (caddr_t)(cp + off), sizeof(struct in_addr));
824			cp[IPOPT_OFFSET] += sizeof(struct in_addr);
825			/*
826			 * Let ip_intr's mcast routing check handle mcast pkts
827			 */
828			forward = !IN_MULTICAST(ip->ip_dst.s_addr);
829			break;
830
831		case IPOPT_RR:
832			if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
833				code = &cp[IPOPT_OFFSET] - (u_char *)ip;
834				goto bad;
835			}
836			/*
837			 * If no space remains, ignore.
838			 */
839			off--;			/* 0 origin */
840			if (off > optlen - sizeof(struct in_addr))
841				break;
842			bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
843			    sizeof(ipaddr.sin_addr));
844			/*
845			 * locate outgoing interface; if we're the destination,
846			 * use the incoming interface (should be same).
847			 */
848			if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
849			    (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
850				type = ICMP_UNREACH;
851				code = ICMP_UNREACH_HOST;
852				goto bad;
853			}
854			bcopy((caddr_t)&ia->ia_addr.sin_addr,
855			    (caddr_t)(cp + off), sizeof(struct in_addr));
856			cp[IPOPT_OFFSET] += sizeof(struct in_addr);
857			break;
858
859		case IPOPT_TS:
860			code = cp - (u_char *)ip;
861			ipt = (struct ip_timestamp *)cp;
862			if (ipt->ipt_len < 5)
863				goto bad;
864			if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
865				if (++ipt->ipt_oflw == 0)
866					goto bad;
867				break;
868			}
869			sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
870			switch (ipt->ipt_flg) {
871
872			case IPOPT_TS_TSONLY:
873				break;
874
875			case IPOPT_TS_TSANDADDR:
876				if (ipt->ipt_ptr + sizeof(n_time) +
877				    sizeof(struct in_addr) > ipt->ipt_len)
878					goto bad;
879				ipaddr.sin_addr = dst;
880				ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
881							    m->m_pkthdr.rcvif);
882				if (ia == 0)
883					continue;
884				bcopy((caddr_t)&ia->ia_addr.sin_addr,
885				    (caddr_t)sin, sizeof(struct in_addr));
886				ipt->ipt_ptr += sizeof(struct in_addr);
887				break;
888
889			case IPOPT_TS_PRESPEC:
890				if (ipt->ipt_ptr + sizeof(n_time) +
891				    sizeof(struct in_addr) > ipt->ipt_len)
892					goto bad;
893				bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
894				    sizeof(struct in_addr));
895				if (ifa_ifwithaddr((SA)&ipaddr) == 0)
896					continue;
897				ipt->ipt_ptr += sizeof(struct in_addr);
898				break;
899
900			default:
901				goto bad;
902			}
903			ntime = iptime();
904			bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
905			    sizeof(n_time));
906			ipt->ipt_ptr += sizeof(n_time);
907		}
908	}
909	if (forward) {
910		if (ip_forwsrcrt == 0) {
911			type = ICMP_UNREACH;
912			code = ICMP_UNREACH_SRCFAIL;
913			goto bad;
914		}
915		ip_forward(m, 1);
916		return (1);
917	}
918	return (0);
919bad:
920	ip->ip_len -= ip->ip_hl << 2;   /* XXX icmp_error adds in hdr length */
921	icmp_error(m, type, code, 0, 0);
922	ipstat.ips_badoptions++;
923	return (1);
924}
925
926/*
927 * Given address of next destination (final or next hop),
928 * return internet address info of interface to be used to get there.
929 */
930struct in_ifaddr *
931ip_rtaddr(dst)
932	 struct in_addr dst;
933{
934	register struct sockaddr_in *sin;
935
936	sin = satosin(&ipforward_rt.ro_dst);
937
938	if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
939		if (ipforward_rt.ro_rt) {
940			RTFREE(ipforward_rt.ro_rt);
941			ipforward_rt.ro_rt = 0;
942		}
943		sin->sin_family = AF_INET;
944		sin->sin_len = sizeof(*sin);
945		sin->sin_addr = dst;
946
947		rtalloc(&ipforward_rt);
948	}
949	if (ipforward_rt.ro_rt == 0)
950		return ((struct in_ifaddr *)0);
951	return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
952}
953
954/*
955 * Save incoming source route for use in replies,
956 * to be picked up later by ip_srcroute if the receiver is interested.
957 */
958void
959save_rte(option, dst)
960	u_char *option;
961	struct in_addr dst;
962{
963	unsigned olen;
964
965	olen = option[IPOPT_OLEN];
966#ifdef DIAGNOSTIC
967	if (ipprintfs)
968		printf("save_rte: olen %d\n", olen);
969#endif
970	if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
971		return;
972	bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
973	ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
974	ip_srcrt.dst = dst;
975}
976
977/*
978 * Retrieve incoming source route for use in replies,
979 * in the same form used by setsockopt.
980 * The first hop is placed before the options, will be removed later.
981 */
982struct mbuf *
983ip_srcroute()
984{
985	register struct in_addr *p, *q;
986	register struct mbuf *m;
987
988	if (ip_nhops == 0)
989		return ((struct mbuf *)0);
990	m = m_get(M_DONTWAIT, MT_SOOPTS);
991	if (m == 0)
992		return ((struct mbuf *)0);
993
994#define OPTSIZ	(sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
995
996	/* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
997	m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
998	    OPTSIZ;
999#ifdef DIAGNOSTIC
1000	if (ipprintfs)
1001		printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1002#endif
1003
1004	/*
1005	 * First save first hop for return route
1006	 */
1007	p = &ip_srcrt.route[ip_nhops - 1];
1008	*(mtod(m, struct in_addr *)) = *p--;
1009#ifdef DIAGNOSTIC
1010	if (ipprintfs)
1011		printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1012#endif
1013
1014	/*
1015	 * Copy option fields and padding (nop) to mbuf.
1016	 */
1017	ip_srcrt.nop = IPOPT_NOP;
1018	ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1019	bcopy((caddr_t)&ip_srcrt.nop,
1020	    mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1021	q = (struct in_addr *)(mtod(m, caddr_t) +
1022	    sizeof(struct in_addr) + OPTSIZ);
1023#undef OPTSIZ
1024	/*
1025	 * Record return path as an IP source route,
1026	 * reversing the path (pointers are now aligned).
1027	 */
1028	while (p >= ip_srcrt.route) {
1029#ifdef DIAGNOSTIC
1030		if (ipprintfs)
1031			printf(" %x", ntohl(q->s_addr));
1032#endif
1033		*q++ = *p--;
1034	}
1035	/*
1036	 * Last hop goes to final destination.
1037	 */
1038	*q = ip_srcrt.dst;
1039#ifdef DIAGNOSTIC
1040	if (ipprintfs)
1041		printf(" %x\n", ntohl(q->s_addr));
1042#endif
1043	return (m);
1044}
1045
1046/*
1047 * Strip out IP options, at higher
1048 * level protocol in the kernel.
1049 * Second argument is buffer to which options
1050 * will be moved, and return value is their length.
1051 * XXX should be deleted; last arg currently ignored.
1052 */
1053void
1054ip_stripoptions(m, mopt)
1055	register struct mbuf *m;
1056	struct mbuf *mopt;
1057{
1058	register int i;
1059	struct ip *ip = mtod(m, struct ip *);
1060	register caddr_t opts;
1061	int olen;
1062
1063	olen = (ip->ip_hl<<2) - sizeof (struct ip);
1064	opts = (caddr_t)(ip + 1);
1065	i = m->m_len - (sizeof (struct ip) + olen);
1066	bcopy(opts  + olen, opts, (unsigned)i);
1067	m->m_len -= olen;
1068	if (m->m_flags & M_PKTHDR)
1069		m->m_pkthdr.len -= olen;
1070	ip->ip_hl = sizeof(struct ip) >> 2;
1071}
1072
1073int inetctlerrmap[PRC_NCMDS] = {
1074	0,		0,		0,		0,
1075	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1076	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1077	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1078	0,		0,		0,		0,
1079	ENOPROTOOPT
1080};
1081
1082/*
1083 * Forward a packet.  If some error occurs return the sender
1084 * an icmp packet.  Note we can't always generate a meaningful
1085 * icmp message because icmp doesn't have a large enough repertoire
1086 * of codes and types.
1087 *
1088 * If not forwarding, just drop the packet.  This could be confusing
1089 * if ipforwarding was zero but some routing protocol was advancing
1090 * us as a gateway to somewhere.  However, we must let the routing
1091 * protocol deal with that.
1092 *
1093 * The srcrt parameter indicates whether the packet is being forwarded
1094 * via a source route.
1095 */
1096void
1097ip_forward(m, srcrt)
1098	struct mbuf *m;
1099	int srcrt;
1100{
1101	register struct ip *ip = mtod(m, struct ip *);
1102	register struct sockaddr_in *sin;
1103	register struct rtentry *rt;
1104	int error, type = 0, code = 0;
1105	struct mbuf *mcopy;
1106	n_long dest;
1107	struct ifnet *destifp;
1108
1109	dest = 0;
1110#ifdef DIAGNOSTIC
1111	if (ipprintfs)
1112		printf("forward: src %x dst %x ttl %x\n",
1113		    ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl);
1114#endif
1115	if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1116		ipstat.ips_cantforward++;
1117		m_freem(m);
1118		return;
1119	}
1120	HTONS(ip->ip_id);
1121	if (ip->ip_ttl <= IPTTLDEC) {
1122		icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1123		return;
1124	}
1125	ip->ip_ttl -= IPTTLDEC;
1126
1127	sin = satosin(&ipforward_rt.ro_dst);
1128	if ((rt = ipforward_rt.ro_rt) == 0 ||
1129	    !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1130		if (ipforward_rt.ro_rt) {
1131			RTFREE(ipforward_rt.ro_rt);
1132			ipforward_rt.ro_rt = 0;
1133		}
1134		sin->sin_family = AF_INET;
1135		sin->sin_len = sizeof(struct sockaddr_in);
1136		sin->sin_addr = ip->ip_dst;
1137
1138		rtalloc(&ipforward_rt);
1139		if (ipforward_rt.ro_rt == 0) {
1140			icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1141			return;
1142		}
1143		rt = ipforward_rt.ro_rt;
1144	}
1145
1146	/*
1147	 * Save at most 68 bytes of the packet in case
1148	 * we need to generate an ICMP message to the src.
1149	 */
1150	mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1151
1152	/*
1153	 * If forwarding packet using same interface that it came in on,
1154	 * perhaps should send a redirect to sender to shortcut a hop.
1155	 * Only send redirect if source is sending directly to us,
1156	 * and if packet was not source routed (or has any options).
1157	 * Also, don't send redirect if forwarding using a default route
1158	 * or a route modified by a redirect.
1159	 */
1160	if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1161	    (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1162	    !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1163	    ipsendredirects && !srcrt) {
1164		if (rt->rt_ifa &&
1165		    (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1166		    ifatoia(rt->rt_ifa)->ia_subnet) {
1167		    if (rt->rt_flags & RTF_GATEWAY)
1168			dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1169		    else
1170			dest = ip->ip_dst.s_addr;
1171		    /* Router requirements says to only send host redirects */
1172		    type = ICMP_REDIRECT;
1173		    code = ICMP_REDIRECT_HOST;
1174#ifdef DIAGNOSTIC
1175		    if (ipprintfs)
1176		    	printf("redirect (%d) to %x\n", code, (u_int32_t)dest);
1177#endif
1178		}
1179	}
1180
1181	error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1182	    (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1183	if (error)
1184		ipstat.ips_cantforward++;
1185	else {
1186		ipstat.ips_forward++;
1187		if (type)
1188			ipstat.ips_redirectsent++;
1189		else {
1190			if (mcopy)
1191				m_freem(mcopy);
1192			return;
1193		}
1194	}
1195	if (mcopy == NULL)
1196		return;
1197	destifp = NULL;
1198
1199	switch (error) {
1200
1201	case 0:				/* forwarded, but need redirect */
1202		/* type, code set above */
1203		break;
1204
1205	case ENETUNREACH:		/* shouldn't happen, checked above */
1206	case EHOSTUNREACH:
1207	case ENETDOWN:
1208	case EHOSTDOWN:
1209	default:
1210		type = ICMP_UNREACH;
1211		code = ICMP_UNREACH_HOST;
1212		break;
1213
1214	case EMSGSIZE:
1215		type = ICMP_UNREACH;
1216		code = ICMP_UNREACH_NEEDFRAG;
1217		if (ipforward_rt.ro_rt)
1218			destifp = ipforward_rt.ro_rt->rt_ifp;
1219		ipstat.ips_cantfrag++;
1220		break;
1221
1222	case ENOBUFS:
1223		type = ICMP_SOURCEQUENCH;
1224		code = 0;
1225		break;
1226	}
1227	icmp_error(mcopy, type, code, dest, destifp);
1228}
1229
1230void
1231ip_savecontrol(inp, mp, ip, m)
1232	register struct inpcb *inp;
1233	register struct mbuf **mp;
1234	register struct ip *ip;
1235	register struct mbuf *m;
1236{
1237
1238	if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1239		struct timeval tv;
1240
1241		microtime(&tv);
1242		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1243		    SCM_TIMESTAMP, SOL_SOCKET);
1244		if (*mp)
1245			mp = &(*mp)->m_next;
1246	}
1247	if (inp->inp_flags & INP_RECVDSTADDR) {
1248		*mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1249		    sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1250		if (*mp)
1251			mp = &(*mp)->m_next;
1252	}
1253#ifdef notyet
1254	/*
1255	 * XXX
1256	 * Moving these out of udp_input() made them even more broken
1257	 * than they already were.
1258	 *	- fenner@parc.xerox.com
1259	 */
1260	/* options were tossed already */
1261	if (inp->inp_flags & INP_RECVOPTS) {
1262		*mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1263		    sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1264		if (*mp)
1265			mp = &(*mp)->m_next;
1266	}
1267	/* ip_srcroute doesn't do what we want here, need to fix */
1268	if (inp->inp_flags & INP_RECVRETOPTS) {
1269		*mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1270		    sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1271		if (*mp)
1272			mp = &(*mp)->m_next;
1273	}
1274#endif
1275	if (inp->inp_flags & INP_RECVIF) {
1276		struct sockaddr_dl sdl;
1277
1278		sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1279		sdl.sdl_family = AF_LINK;
1280		sdl.sdl_index = m->m_pkthdr.rcvif ?
1281		    m->m_pkthdr.rcvif->if_index : 0;
1282		sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1283		*mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1284		    IP_RECVIF, IPPROTO_IP);
1285		if (*mp)
1286			mp = &(*mp)->m_next;
1287	}
1288}
1289
1290int
1291ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1292	int *name;
1293	u_int namelen;
1294	void *oldp;
1295	size_t *oldlenp;
1296	void *newp;
1297	size_t newlen;
1298{
1299	extern int subnetsarelocal;
1300
1301	int error, old;
1302
1303	/* All sysctl names at this level are terminal. */
1304	if (namelen != 1)
1305		return (ENOTDIR);
1306
1307	switch (name[0]) {
1308	case IPCTL_FORWARDING:
1309		return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1310	case IPCTL_SENDREDIRECTS:
1311		return (sysctl_int(oldp, oldlenp, newp, newlen,
1312			&ipsendredirects));
1313	case IPCTL_DEFTTL:
1314		return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1315#ifdef notyet
1316	case IPCTL_DEFMTU:
1317		return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1318#endif
1319	case IPCTL_FORWSRCRT:
1320		/* Don't allow this to change in a secure environment.  */
1321		if (securelevel > 0)
1322			return (sysctl_rdint(oldp, oldlenp, newp,
1323			    ip_forwsrcrt));
1324		else
1325			return (sysctl_int(oldp, oldlenp, newp, newlen,
1326			    &ip_forwsrcrt));
1327	case IPCTL_DIRECTEDBCAST:
1328		return (sysctl_int(oldp, oldlenp, newp, newlen,
1329		    &ip_directedbcast));
1330	case IPCTL_ALLOWSRCRT:
1331		return (sysctl_int(oldp, oldlenp, newp, newlen,
1332		    &ip_allowsrcrt));
1333	case IPCTL_SUBNETSARELOCAL:
1334		return (sysctl_int(oldp, oldlenp, newp, newlen,
1335		    &subnetsarelocal));
1336	case IPCTL_MTUDISC:
1337		error = sysctl_int(oldp, oldlenp, newp, newlen,
1338		    &ip_mtudisc);
1339		if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1340			ip_mtudisc_timeout_q =
1341			    rt_timer_queue_create(ip_mtudisc_timeout);
1342		} else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1343			rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1344			ip_mtudisc_timeout_q = NULL;
1345		}
1346		return error;
1347	case IPCTL_ANONPORTMIN:
1348		old = anonportmin;
1349		error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1350		if (anonportmin >= anonportmax || anonportmin > 65535
1351#ifndef IPNOPRIVPORTS
1352		    || anonportmin < IPPORT_RESERVED
1353#endif
1354		    ) {
1355			anonportmin = old;
1356			return (EINVAL);
1357		}
1358		return (error);
1359	case IPCTL_ANONPORTMAX:
1360		old = anonportmax;
1361		error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1362		if (anonportmin >= anonportmax || anonportmax > 65535
1363#ifndef IPNOPRIVPORTS
1364		    || anonportmax < IPPORT_RESERVED
1365#endif
1366		    ) {
1367			anonportmax = old;
1368			return (EINVAL);
1369		}
1370		return (error);
1371	case IPCTL_MTUDISCTIMEOUT:
1372		error = sysctl_int(oldp, oldlenp, newp, newlen,
1373		   &ip_mtudisc_timeout);
1374		if (ip_mtudisc_timeout_q != NULL)
1375			rt_timer_queue_change(ip_mtudisc_timeout_q,
1376					      ip_mtudisc_timeout);
1377		return (error);
1378	default:
1379		return (EOPNOTSUPP);
1380	}
1381	/* NOTREACHED */
1382}
1383