ip_input.c revision 1.64
1/*	$NetBSD: ip_input.c,v 1.64 1998/05/01 03:23:24 thorpej 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	/*
313	 * Assume that we can create a fast-forward IP flow entry
314	 * based on this packet.
315	 */
316	m->m_flags |= M_CANFASTFWD;
317
318#ifdef PFIL_HOOKS
319	/*
320	 * Run through list of hooks for input packets.  If there are any
321	 * filters which require that additional packets in the flow are
322	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
323	 * Note that filters must _never_ set this flag, as another filter
324	 * in the list may have previously cleared it.
325	 */
326	m0 = m;
327	for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next)
328		if (pfh->pfil_func) {
329			rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0);
330			if (rv)
331				goto next;
332			ip = mtod(m = m0, struct ip *);
333		}
334#endif /* PFIL_HOOKS */
335
336	/*
337	 * Process options and, if not destined for us,
338	 * ship it on.  ip_dooptions returns 1 when an
339	 * error was detected (causing an icmp message
340	 * to be sent and the original packet to be freed).
341	 */
342	ip_nhops = 0;		/* for source routed packets */
343	if (hlen > sizeof (struct ip) && ip_dooptions(m))
344		goto next;
345
346	/*
347	 * Check our list of addresses, to see if the packet is for us.
348	 */
349	INADDR_TO_IA(ip->ip_dst, ia);
350	if (ia != NULL) goto ours;
351	if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
352		for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first;
353		    ifa != NULL; ifa = ifa->ifa_list.tqe_next) {
354			if (ifa->ifa_addr->sa_family != AF_INET) continue;
355			ia = ifatoia(ifa);
356			if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
357			    in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
358			    /*
359			     * Look for all-0's host part (old broadcast addr),
360			     * either for subnet or net.
361			     */
362			    ip->ip_dst.s_addr == ia->ia_subnet ||
363			    ip->ip_dst.s_addr == ia->ia_net)
364				goto ours;
365			/*
366			 * An interface with IP address zero accepts
367			 * all packets that arrive on that interface.
368			 */
369			if (in_nullhost(ia->ia_addr.sin_addr))
370				goto ours;
371		}
372	}
373	if (IN_MULTICAST(ip->ip_dst.s_addr)) {
374		struct in_multi *inm;
375#ifdef MROUTING
376		extern struct socket *ip_mrouter;
377
378		if (m->m_flags & M_EXT) {
379			if ((m = m_pullup(m, hlen)) == 0) {
380				ipstat.ips_toosmall++;
381				goto next;
382			}
383			ip = mtod(m, struct ip *);
384		}
385
386		if (ip_mrouter) {
387			/*
388			 * If we are acting as a multicast router, all
389			 * incoming multicast packets are passed to the
390			 * kernel-level multicast forwarding function.
391			 * The packet is returned (relatively) intact; if
392			 * ip_mforward() returns a non-zero value, the packet
393			 * must be discarded, else it may be accepted below.
394			 *
395			 * (The IP ident field is put in the same byte order
396			 * as expected when ip_mforward() is called from
397			 * ip_output().)
398			 */
399			ip->ip_id = htons(ip->ip_id);
400			if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
401				ipstat.ips_cantforward++;
402				m_freem(m);
403				goto next;
404			}
405			ip->ip_id = ntohs(ip->ip_id);
406
407			/*
408			 * The process-level routing demon needs to receive
409			 * all multicast IGMP packets, whether or not this
410			 * host belongs to their destination groups.
411			 */
412			if (ip->ip_p == IPPROTO_IGMP)
413				goto ours;
414			ipstat.ips_forward++;
415		}
416#endif
417		/*
418		 * See if we belong to the destination multicast group on the
419		 * arrival interface.
420		 */
421		IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
422		if (inm == NULL) {
423			ipstat.ips_cantforward++;
424			m_freem(m);
425			goto next;
426		}
427		goto ours;
428	}
429	if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
430	    in_nullhost(ip->ip_dst))
431		goto ours;
432
433	/*
434	 * Not for us; forward if possible and desirable.
435	 */
436	if (ipforwarding == 0) {
437		ipstat.ips_cantforward++;
438		m_freem(m);
439	} else
440		ip_forward(m, 0);
441	goto next;
442
443ours:
444	/*
445	 * If offset or IP_MF are set, must reassemble.
446	 * Otherwise, nothing need be done.
447	 * (We could look in the reassembly queue to see
448	 * if the packet was previously fragmented,
449	 * but it's not worth the time; just let them time out.)
450	 */
451	if (ip->ip_off & ~(IP_DF|IP_RF)) {
452		/*
453		 * Look for queue of fragments
454		 * of this datagram.
455		 */
456		for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next)
457			if (ip->ip_id == fp->ipq_id &&
458			    in_hosteq(ip->ip_src, fp->ipq_src) &&
459			    in_hosteq(ip->ip_dst, fp->ipq_dst) &&
460			    ip->ip_p == fp->ipq_p)
461				goto found;
462		fp = 0;
463found:
464
465		/*
466		 * Adjust ip_len to not reflect header,
467		 * set ipqe_mff if more fragments are expected,
468		 * convert offset of this to bytes.
469		 */
470		ip->ip_len -= hlen;
471		mff = (ip->ip_off & IP_MF) != 0;
472		if (mff) {
473		        /*
474		         * Make sure that fragments have a data length
475			 * that's a non-zero multiple of 8 bytes.
476		         */
477			if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
478				ipstat.ips_badfrags++;
479				goto bad;
480			}
481		}
482		ip->ip_off <<= 3;
483
484		/*
485		 * If datagram marked as having more fragments
486		 * or if this is not the first fragment,
487		 * attempt reassembly; if it succeeds, proceed.
488		 */
489		if (mff || ip->ip_off) {
490			ipstat.ips_fragments++;
491			MALLOC(ipqe, struct ipqent *, sizeof (struct ipqent),
492			    M_IPQ, M_NOWAIT);
493			if (ipqe == NULL) {
494				ipstat.ips_rcvmemdrop++;
495				goto bad;
496			}
497			ipqe->ipqe_mff = mff;
498			ipqe->ipqe_m = m;
499			ipqe->ipqe_ip = ip;
500			m = ip_reass(ipqe, fp);
501			if (m == 0)
502				goto next;
503			ipstat.ips_reassembled++;
504			ip = mtod(m, struct ip *);
505		} else
506			if (fp)
507				ip_freef(fp);
508	} else
509		ip->ip_len -= hlen;
510
511	/*
512	 * Switch out to protocol's input routine.
513	 */
514	ipstat.ips_delivered++;
515	(*inetsw[ip_protox[ip->ip_p]].pr_input)(m, hlen);
516	goto next;
517bad:
518	m_freem(m);
519	goto next;
520}
521
522/*
523 * Take incoming datagram fragment and try to
524 * reassemble it into whole datagram.  If a chain for
525 * reassembly of this datagram already exists, then it
526 * is given as fp; otherwise have to make a chain.
527 */
528struct mbuf *
529ip_reass(ipqe, fp)
530	register struct ipqent *ipqe;
531	register struct ipq *fp;
532{
533	register struct mbuf *m = ipqe->ipqe_m;
534	register struct ipqent *nq, *p, *q;
535	struct ip *ip;
536	struct mbuf *t;
537	int hlen = ipqe->ipqe_ip->ip_hl << 2;
538	int i, next;
539
540	/*
541	 * Presence of header sizes in mbufs
542	 * would confuse code below.
543	 */
544	m->m_data += hlen;
545	m->m_len -= hlen;
546
547	/*
548	 * If first fragment to arrive, create a reassembly queue.
549	 */
550	if (fp == 0) {
551		MALLOC(fp, struct ipq *, sizeof (struct ipq),
552		    M_FTABLE, M_NOWAIT);
553		if (fp == NULL)
554			goto dropfrag;
555		LIST_INSERT_HEAD(&ipq, fp, ipq_q);
556		fp->ipq_ttl = IPFRAGTTL;
557		fp->ipq_p = ipqe->ipqe_ip->ip_p;
558		fp->ipq_id = ipqe->ipqe_ip->ip_id;
559		LIST_INIT(&fp->ipq_fragq);
560		fp->ipq_src = ipqe->ipqe_ip->ip_src;
561		fp->ipq_dst = ipqe->ipqe_ip->ip_dst;
562		p = NULL;
563		goto insert;
564	}
565
566	/*
567	 * Find a segment which begins after this one does.
568	 */
569	for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
570	    p = q, q = q->ipqe_q.le_next)
571		if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off)
572			break;
573
574	/*
575	 * If there is a preceding segment, it may provide some of
576	 * our data already.  If so, drop the data from the incoming
577	 * segment.  If it provides all of our data, drop us.
578	 */
579	if (p != NULL) {
580		i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len -
581		    ipqe->ipqe_ip->ip_off;
582		if (i > 0) {
583			if (i >= ipqe->ipqe_ip->ip_len)
584				goto dropfrag;
585			m_adj(ipqe->ipqe_m, i);
586			ipqe->ipqe_ip->ip_off += i;
587			ipqe->ipqe_ip->ip_len -= i;
588		}
589	}
590
591	/*
592	 * While we overlap succeeding segments trim them or,
593	 * if they are completely covered, dequeue them.
594	 */
595	for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len >
596	    q->ipqe_ip->ip_off; q = nq) {
597		i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) -
598		    q->ipqe_ip->ip_off;
599		if (i < q->ipqe_ip->ip_len) {
600			q->ipqe_ip->ip_len -= i;
601			q->ipqe_ip->ip_off += i;
602			m_adj(q->ipqe_m, i);
603			break;
604		}
605		nq = q->ipqe_q.le_next;
606		m_freem(q->ipqe_m);
607		LIST_REMOVE(q, ipqe_q);
608		FREE(q, M_IPQ);
609	}
610
611insert:
612	/*
613	 * Stick new segment in its place;
614	 * check for complete reassembly.
615	 */
616	if (p == NULL) {
617		LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q);
618	} else {
619		LIST_INSERT_AFTER(p, ipqe, ipqe_q);
620	}
621	next = 0;
622	for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL;
623	    p = q, q = q->ipqe_q.le_next) {
624		if (q->ipqe_ip->ip_off != next)
625			return (0);
626		next += q->ipqe_ip->ip_len;
627	}
628	if (p->ipqe_mff)
629		return (0);
630
631	/*
632	 * Reassembly is complete.  Check for a bogus message size and
633	 * concatenate fragments.
634	 */
635	q = fp->ipq_fragq.lh_first;
636	ip = q->ipqe_ip;
637	if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) {
638		ipstat.ips_toolong++;
639		ip_freef(fp);
640		return (0);
641	}
642	m = q->ipqe_m;
643	t = m->m_next;
644	m->m_next = 0;
645	m_cat(m, t);
646	nq = q->ipqe_q.le_next;
647	FREE(q, M_IPQ);
648	for (q = nq; q != NULL; q = nq) {
649		t = q->ipqe_m;
650		nq = q->ipqe_q.le_next;
651		FREE(q, M_IPQ);
652		m_cat(m, t);
653	}
654
655	/*
656	 * Create header for new ip packet by
657	 * modifying header of first packet;
658	 * dequeue and discard fragment reassembly header.
659	 * Make header visible.
660	 */
661	ip->ip_len = next;
662	ip->ip_src = fp->ipq_src;
663	ip->ip_dst = fp->ipq_dst;
664	LIST_REMOVE(fp, ipq_q);
665	FREE(fp, M_FTABLE);
666	m->m_len += (ip->ip_hl << 2);
667	m->m_data -= (ip->ip_hl << 2);
668	/* some debugging cruft by sklower, below, will go away soon */
669	if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
670		register int plen = 0;
671		for (t = m; t; t = t->m_next)
672			plen += t->m_len;
673		m->m_pkthdr.len = plen;
674	}
675	return (m);
676
677dropfrag:
678	ipstat.ips_fragdropped++;
679	m_freem(m);
680	FREE(ipqe, M_IPQ);
681	return (0);
682}
683
684/*
685 * Free a fragment reassembly header and all
686 * associated datagrams.
687 */
688void
689ip_freef(fp)
690	struct ipq *fp;
691{
692	register struct ipqent *q, *p;
693
694	for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) {
695		p = q->ipqe_q.le_next;
696		m_freem(q->ipqe_m);
697		LIST_REMOVE(q, ipqe_q);
698		FREE(q, M_IPQ);
699	}
700	LIST_REMOVE(fp, ipq_q);
701	FREE(fp, M_FTABLE);
702}
703
704/*
705 * IP timer processing;
706 * if a timer expires on a reassembly
707 * queue, discard it.
708 */
709void
710ip_slowtimo()
711{
712	register struct ipq *fp, *nfp;
713	int s = splsoftnet();
714
715	for (fp = ipq.lh_first; fp != NULL; fp = nfp) {
716		nfp = fp->ipq_q.le_next;
717		if (--fp->ipq_ttl == 0) {
718			ipstat.ips_fragtimeout++;
719			ip_freef(fp);
720		}
721	}
722#ifdef GATEWAY
723	ipflow_slowtimo();
724#endif
725	splx(s);
726}
727
728/*
729 * Drain off all datagram fragments.
730 */
731void
732ip_drain()
733{
734
735	while (ipq.lh_first != NULL) {
736		ipstat.ips_fragdropped++;
737		ip_freef(ipq.lh_first);
738	}
739}
740
741/*
742 * Do option processing on a datagram,
743 * possibly discarding it if bad options are encountered,
744 * or forwarding it if source-routed.
745 * Returns 1 if packet has been forwarded/freed,
746 * 0 if the packet should be processed further.
747 */
748int
749ip_dooptions(m)
750	struct mbuf *m;
751{
752	register struct ip *ip = mtod(m, struct ip *);
753	register u_char *cp;
754	register struct ip_timestamp *ipt;
755	register struct in_ifaddr *ia;
756	int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
757	struct in_addr *sin, dst;
758	n_time ntime;
759
760	dst = ip->ip_dst;
761	cp = (u_char *)(ip + 1);
762	cnt = (ip->ip_hl << 2) - sizeof (struct ip);
763	for (; cnt > 0; cnt -= optlen, cp += optlen) {
764		opt = cp[IPOPT_OPTVAL];
765		if (opt == IPOPT_EOL)
766			break;
767		if (opt == IPOPT_NOP)
768			optlen = 1;
769		else {
770			optlen = cp[IPOPT_OLEN];
771			if (optlen <= 0 || optlen > cnt) {
772				code = &cp[IPOPT_OLEN] - (u_char *)ip;
773				goto bad;
774			}
775		}
776		switch (opt) {
777
778		default:
779			break;
780
781		/*
782		 * Source routing with record.
783		 * Find interface with current destination address.
784		 * If none on this machine then drop if strictly routed,
785		 * or do nothing if loosely routed.
786		 * Record interface address and bring up next address
787		 * component.  If strictly routed make sure next
788		 * address is on directly accessible net.
789		 */
790		case IPOPT_LSRR:
791		case IPOPT_SSRR:
792			if (ip_allowsrcrt == 0) {
793				type = ICMP_UNREACH;
794				code = ICMP_UNREACH_NET_PROHIB;
795				goto bad;
796			}
797			if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
798				code = &cp[IPOPT_OFFSET] - (u_char *)ip;
799				goto bad;
800			}
801			ipaddr.sin_addr = ip->ip_dst;
802			ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)));
803			if (ia == 0) {
804				if (opt == IPOPT_SSRR) {
805					type = ICMP_UNREACH;
806					code = ICMP_UNREACH_SRCFAIL;
807					goto bad;
808				}
809				/*
810				 * Loose routing, and not at next destination
811				 * yet; nothing to do except forward.
812				 */
813				break;
814			}
815			off--;			/* 0 origin */
816			if (off > optlen - sizeof(struct in_addr)) {
817				/*
818				 * End of source route.  Should be for us.
819				 */
820				save_rte(cp, ip->ip_src);
821				break;
822			}
823			/*
824			 * locate outgoing interface
825			 */
826			bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr,
827			    sizeof(ipaddr.sin_addr));
828			if (opt == IPOPT_SSRR) {
829#define	INA	struct in_ifaddr *
830#define	SA	struct sockaddr *
831			    ia = (INA)ifa_ifwithladdr((SA)&ipaddr);
832			} else
833				ia = ip_rtaddr(ipaddr.sin_addr);
834			if (ia == 0) {
835				type = ICMP_UNREACH;
836				code = ICMP_UNREACH_SRCFAIL;
837				goto bad;
838			}
839			ip->ip_dst = ipaddr.sin_addr;
840			bcopy((caddr_t)&ia->ia_addr.sin_addr,
841			    (caddr_t)(cp + off), sizeof(struct in_addr));
842			cp[IPOPT_OFFSET] += sizeof(struct in_addr);
843			/*
844			 * Let ip_intr's mcast routing check handle mcast pkts
845			 */
846			forward = !IN_MULTICAST(ip->ip_dst.s_addr);
847			break;
848
849		case IPOPT_RR:
850			if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
851				code = &cp[IPOPT_OFFSET] - (u_char *)ip;
852				goto bad;
853			}
854			/*
855			 * If no space remains, ignore.
856			 */
857			off--;			/* 0 origin */
858			if (off > optlen - sizeof(struct in_addr))
859				break;
860			bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr,
861			    sizeof(ipaddr.sin_addr));
862			/*
863			 * locate outgoing interface; if we're the destination,
864			 * use the incoming interface (should be same).
865			 */
866			if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 &&
867			    (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) {
868				type = ICMP_UNREACH;
869				code = ICMP_UNREACH_HOST;
870				goto bad;
871			}
872			bcopy((caddr_t)&ia->ia_addr.sin_addr,
873			    (caddr_t)(cp + off), sizeof(struct in_addr));
874			cp[IPOPT_OFFSET] += sizeof(struct in_addr);
875			break;
876
877		case IPOPT_TS:
878			code = cp - (u_char *)ip;
879			ipt = (struct ip_timestamp *)cp;
880			if (ipt->ipt_len < 5)
881				goto bad;
882			if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
883				if (++ipt->ipt_oflw == 0)
884					goto bad;
885				break;
886			}
887			sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1);
888			switch (ipt->ipt_flg) {
889
890			case IPOPT_TS_TSONLY:
891				break;
892
893			case IPOPT_TS_TSANDADDR:
894				if (ipt->ipt_ptr + sizeof(n_time) +
895				    sizeof(struct in_addr) > ipt->ipt_len)
896					goto bad;
897				ipaddr.sin_addr = dst;
898				ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
899							    m->m_pkthdr.rcvif);
900				if (ia == 0)
901					continue;
902				bcopy((caddr_t)&ia->ia_addr.sin_addr,
903				    (caddr_t)sin, sizeof(struct in_addr));
904				ipt->ipt_ptr += sizeof(struct in_addr);
905				break;
906
907			case IPOPT_TS_PRESPEC:
908				if (ipt->ipt_ptr + sizeof(n_time) +
909				    sizeof(struct in_addr) > ipt->ipt_len)
910					goto bad;
911				bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr,
912				    sizeof(struct in_addr));
913				if (ifa_ifwithaddr((SA)&ipaddr) == 0)
914					continue;
915				ipt->ipt_ptr += sizeof(struct in_addr);
916				break;
917
918			default:
919				goto bad;
920			}
921			ntime = iptime();
922			bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1,
923			    sizeof(n_time));
924			ipt->ipt_ptr += sizeof(n_time);
925		}
926	}
927	if (forward) {
928		if (ip_forwsrcrt == 0) {
929			type = ICMP_UNREACH;
930			code = ICMP_UNREACH_SRCFAIL;
931			goto bad;
932		}
933		ip_forward(m, 1);
934		return (1);
935	}
936	return (0);
937bad:
938	ip->ip_len -= ip->ip_hl << 2;   /* XXX icmp_error adds in hdr length */
939	icmp_error(m, type, code, 0, 0);
940	ipstat.ips_badoptions++;
941	return (1);
942}
943
944/*
945 * Given address of next destination (final or next hop),
946 * return internet address info of interface to be used to get there.
947 */
948struct in_ifaddr *
949ip_rtaddr(dst)
950	 struct in_addr dst;
951{
952	register struct sockaddr_in *sin;
953
954	sin = satosin(&ipforward_rt.ro_dst);
955
956	if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) {
957		if (ipforward_rt.ro_rt) {
958			RTFREE(ipforward_rt.ro_rt);
959			ipforward_rt.ro_rt = 0;
960		}
961		sin->sin_family = AF_INET;
962		sin->sin_len = sizeof(*sin);
963		sin->sin_addr = dst;
964
965		rtalloc(&ipforward_rt);
966	}
967	if (ipforward_rt.ro_rt == 0)
968		return ((struct in_ifaddr *)0);
969	return (ifatoia(ipforward_rt.ro_rt->rt_ifa));
970}
971
972/*
973 * Save incoming source route for use in replies,
974 * to be picked up later by ip_srcroute if the receiver is interested.
975 */
976void
977save_rte(option, dst)
978	u_char *option;
979	struct in_addr dst;
980{
981	unsigned olen;
982
983	olen = option[IPOPT_OLEN];
984#ifdef DIAGNOSTIC
985	if (ipprintfs)
986		printf("save_rte: olen %d\n", olen);
987#endif
988	if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
989		return;
990	bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen);
991	ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
992	ip_srcrt.dst = dst;
993}
994
995/*
996 * Retrieve incoming source route for use in replies,
997 * in the same form used by setsockopt.
998 * The first hop is placed before the options, will be removed later.
999 */
1000struct mbuf *
1001ip_srcroute()
1002{
1003	register struct in_addr *p, *q;
1004	register struct mbuf *m;
1005
1006	if (ip_nhops == 0)
1007		return ((struct mbuf *)0);
1008	m = m_get(M_DONTWAIT, MT_SOOPTS);
1009	if (m == 0)
1010		return ((struct mbuf *)0);
1011
1012#define OPTSIZ	(sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1013
1014	/* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1015	m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1016	    OPTSIZ;
1017#ifdef DIAGNOSTIC
1018	if (ipprintfs)
1019		printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len);
1020#endif
1021
1022	/*
1023	 * First save first hop for return route
1024	 */
1025	p = &ip_srcrt.route[ip_nhops - 1];
1026	*(mtod(m, struct in_addr *)) = *p--;
1027#ifdef DIAGNOSTIC
1028	if (ipprintfs)
1029		printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1030#endif
1031
1032	/*
1033	 * Copy option fields and padding (nop) to mbuf.
1034	 */
1035	ip_srcrt.nop = IPOPT_NOP;
1036	ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1037	bcopy((caddr_t)&ip_srcrt.nop,
1038	    mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ);
1039	q = (struct in_addr *)(mtod(m, caddr_t) +
1040	    sizeof(struct in_addr) + OPTSIZ);
1041#undef OPTSIZ
1042	/*
1043	 * Record return path as an IP source route,
1044	 * reversing the path (pointers are now aligned).
1045	 */
1046	while (p >= ip_srcrt.route) {
1047#ifdef DIAGNOSTIC
1048		if (ipprintfs)
1049			printf(" %x", ntohl(q->s_addr));
1050#endif
1051		*q++ = *p--;
1052	}
1053	/*
1054	 * Last hop goes to final destination.
1055	 */
1056	*q = ip_srcrt.dst;
1057#ifdef DIAGNOSTIC
1058	if (ipprintfs)
1059		printf(" %x\n", ntohl(q->s_addr));
1060#endif
1061	return (m);
1062}
1063
1064/*
1065 * Strip out IP options, at higher
1066 * level protocol in the kernel.
1067 * Second argument is buffer to which options
1068 * will be moved, and return value is their length.
1069 * XXX should be deleted; last arg currently ignored.
1070 */
1071void
1072ip_stripoptions(m, mopt)
1073	register struct mbuf *m;
1074	struct mbuf *mopt;
1075{
1076	register int i;
1077	struct ip *ip = mtod(m, struct ip *);
1078	register caddr_t opts;
1079	int olen;
1080
1081	olen = (ip->ip_hl<<2) - sizeof (struct ip);
1082	opts = (caddr_t)(ip + 1);
1083	i = m->m_len - (sizeof (struct ip) + olen);
1084	bcopy(opts  + olen, opts, (unsigned)i);
1085	m->m_len -= olen;
1086	if (m->m_flags & M_PKTHDR)
1087		m->m_pkthdr.len -= olen;
1088	ip->ip_hl = sizeof(struct ip) >> 2;
1089}
1090
1091int inetctlerrmap[PRC_NCMDS] = {
1092	0,		0,		0,		0,
1093	0,		EMSGSIZE,	EHOSTDOWN,	EHOSTUNREACH,
1094	EHOSTUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
1095	EMSGSIZE,	EHOSTUNREACH,	0,		0,
1096	0,		0,		0,		0,
1097	ENOPROTOOPT
1098};
1099
1100/*
1101 * Forward a packet.  If some error occurs return the sender
1102 * an icmp packet.  Note we can't always generate a meaningful
1103 * icmp message because icmp doesn't have a large enough repertoire
1104 * of codes and types.
1105 *
1106 * If not forwarding, just drop the packet.  This could be confusing
1107 * if ipforwarding was zero but some routing protocol was advancing
1108 * us as a gateway to somewhere.  However, we must let the routing
1109 * protocol deal with that.
1110 *
1111 * The srcrt parameter indicates whether the packet is being forwarded
1112 * via a source route.
1113 */
1114void
1115ip_forward(m, srcrt)
1116	struct mbuf *m;
1117	int srcrt;
1118{
1119	register struct ip *ip = mtod(m, struct ip *);
1120	register struct sockaddr_in *sin;
1121	register struct rtentry *rt;
1122	int error, type = 0, code = 0;
1123	struct mbuf *mcopy;
1124	n_long dest;
1125	struct ifnet *destifp;
1126
1127	dest = 0;
1128#ifdef DIAGNOSTIC
1129	if (ipprintfs)
1130		printf("forward: src %x dst %x ttl %x\n",
1131		    ip->ip_src.s_addr, ip->ip_dst.s_addr, ip->ip_ttl);
1132#endif
1133	if (m->m_flags & M_BCAST || in_canforward(ip->ip_dst) == 0) {
1134		ipstat.ips_cantforward++;
1135		m_freem(m);
1136		return;
1137	}
1138	HTONS(ip->ip_id);
1139	if (ip->ip_ttl <= IPTTLDEC) {
1140		icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1141		return;
1142	}
1143	ip->ip_ttl -= IPTTLDEC;
1144
1145	sin = satosin(&ipforward_rt.ro_dst);
1146	if ((rt = ipforward_rt.ro_rt) == 0 ||
1147	    !in_hosteq(ip->ip_dst, sin->sin_addr)) {
1148		if (ipforward_rt.ro_rt) {
1149			RTFREE(ipforward_rt.ro_rt);
1150			ipforward_rt.ro_rt = 0;
1151		}
1152		sin->sin_family = AF_INET;
1153		sin->sin_len = sizeof(struct sockaddr_in);
1154		sin->sin_addr = ip->ip_dst;
1155
1156		rtalloc(&ipforward_rt);
1157		if (ipforward_rt.ro_rt == 0) {
1158			icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1159			return;
1160		}
1161		rt = ipforward_rt.ro_rt;
1162	}
1163
1164	/*
1165	 * Save at most 68 bytes of the packet in case
1166	 * we need to generate an ICMP message to the src.
1167	 */
1168	mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68));
1169
1170	/*
1171	 * If forwarding packet using same interface that it came in on,
1172	 * perhaps should send a redirect to sender to shortcut a hop.
1173	 * Only send redirect if source is sending directly to us,
1174	 * and if packet was not source routed (or has any options).
1175	 * Also, don't send redirect if forwarding using a default route
1176	 * or a route modified by a redirect.
1177	 */
1178	if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1179	    (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1180	    !in_nullhost(satosin(rt_key(rt))->sin_addr) &&
1181	    ipsendredirects && !srcrt) {
1182		if (rt->rt_ifa &&
1183		    (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1184		    ifatoia(rt->rt_ifa)->ia_subnet) {
1185		    if (rt->rt_flags & RTF_GATEWAY)
1186			dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1187		    else
1188			dest = ip->ip_dst.s_addr;
1189		    /* Router requirements says to only send host redirects */
1190		    type = ICMP_REDIRECT;
1191		    code = ICMP_REDIRECT_HOST;
1192#ifdef DIAGNOSTIC
1193		    if (ipprintfs)
1194		    	printf("redirect (%d) to %x\n", code, (u_int32_t)dest);
1195#endif
1196		}
1197	}
1198
1199	error = ip_output(m, (struct mbuf *)0, &ipforward_rt,
1200	    (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0);
1201	if (error)
1202		ipstat.ips_cantforward++;
1203	else {
1204		ipstat.ips_forward++;
1205		if (type)
1206			ipstat.ips_redirectsent++;
1207		else {
1208			if (mcopy) {
1209#ifdef GATEWAY
1210				if (mcopy->m_flags & M_CANFASTFWD)
1211					ipflow_create(&ipforward_rt, mcopy);
1212#endif
1213				m_freem(mcopy);
1214			}
1215			return;
1216		}
1217	}
1218	if (mcopy == NULL)
1219		return;
1220	destifp = NULL;
1221
1222	switch (error) {
1223
1224	case 0:				/* forwarded, but need redirect */
1225		/* type, code set above */
1226		break;
1227
1228	case ENETUNREACH:		/* shouldn't happen, checked above */
1229	case EHOSTUNREACH:
1230	case ENETDOWN:
1231	case EHOSTDOWN:
1232	default:
1233		type = ICMP_UNREACH;
1234		code = ICMP_UNREACH_HOST;
1235		break;
1236
1237	case EMSGSIZE:
1238		type = ICMP_UNREACH;
1239		code = ICMP_UNREACH_NEEDFRAG;
1240		if (ipforward_rt.ro_rt)
1241			destifp = ipforward_rt.ro_rt->rt_ifp;
1242		ipstat.ips_cantfrag++;
1243		break;
1244
1245	case ENOBUFS:
1246		type = ICMP_SOURCEQUENCH;
1247		code = 0;
1248		break;
1249	}
1250	icmp_error(mcopy, type, code, dest, destifp);
1251}
1252
1253void
1254ip_savecontrol(inp, mp, ip, m)
1255	register struct inpcb *inp;
1256	register struct mbuf **mp;
1257	register struct ip *ip;
1258	register struct mbuf *m;
1259{
1260
1261	if (inp->inp_socket->so_options & SO_TIMESTAMP) {
1262		struct timeval tv;
1263
1264		microtime(&tv);
1265		*mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
1266		    SCM_TIMESTAMP, SOL_SOCKET);
1267		if (*mp)
1268			mp = &(*mp)->m_next;
1269	}
1270	if (inp->inp_flags & INP_RECVDSTADDR) {
1271		*mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
1272		    sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1273		if (*mp)
1274			mp = &(*mp)->m_next;
1275	}
1276#ifdef notyet
1277	/*
1278	 * XXX
1279	 * Moving these out of udp_input() made them even more broken
1280	 * than they already were.
1281	 *	- fenner@parc.xerox.com
1282	 */
1283	/* options were tossed already */
1284	if (inp->inp_flags & INP_RECVOPTS) {
1285		*mp = sbcreatecontrol((caddr_t) opts_deleted_above,
1286		    sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
1287		if (*mp)
1288			mp = &(*mp)->m_next;
1289	}
1290	/* ip_srcroute doesn't do what we want here, need to fix */
1291	if (inp->inp_flags & INP_RECVRETOPTS) {
1292		*mp = sbcreatecontrol((caddr_t) ip_srcroute(),
1293		    sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
1294		if (*mp)
1295			mp = &(*mp)->m_next;
1296	}
1297#endif
1298	if (inp->inp_flags & INP_RECVIF) {
1299		struct sockaddr_dl sdl;
1300
1301		sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]);
1302		sdl.sdl_family = AF_LINK;
1303		sdl.sdl_index = m->m_pkthdr.rcvif ?
1304		    m->m_pkthdr.rcvif->if_index : 0;
1305		sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0;
1306		*mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len,
1307		    IP_RECVIF, IPPROTO_IP);
1308		if (*mp)
1309			mp = &(*mp)->m_next;
1310	}
1311}
1312
1313int
1314ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen)
1315	int *name;
1316	u_int namelen;
1317	void *oldp;
1318	size_t *oldlenp;
1319	void *newp;
1320	size_t newlen;
1321{
1322	extern int subnetsarelocal;
1323
1324	int error, old;
1325
1326	/* All sysctl names at this level are terminal. */
1327	if (namelen != 1)
1328		return (ENOTDIR);
1329
1330	switch (name[0]) {
1331	case IPCTL_FORWARDING:
1332		return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding));
1333	case IPCTL_SENDREDIRECTS:
1334		return (sysctl_int(oldp, oldlenp, newp, newlen,
1335			&ipsendredirects));
1336	case IPCTL_DEFTTL:
1337		return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl));
1338#ifdef notyet
1339	case IPCTL_DEFMTU:
1340		return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu));
1341#endif
1342	case IPCTL_FORWSRCRT:
1343		/* Don't allow this to change in a secure environment.  */
1344		if (securelevel > 0)
1345			return (sysctl_rdint(oldp, oldlenp, newp,
1346			    ip_forwsrcrt));
1347		else
1348			return (sysctl_int(oldp, oldlenp, newp, newlen,
1349			    &ip_forwsrcrt));
1350	case IPCTL_DIRECTEDBCAST:
1351		return (sysctl_int(oldp, oldlenp, newp, newlen,
1352		    &ip_directedbcast));
1353	case IPCTL_ALLOWSRCRT:
1354		return (sysctl_int(oldp, oldlenp, newp, newlen,
1355		    &ip_allowsrcrt));
1356	case IPCTL_SUBNETSARELOCAL:
1357		return (sysctl_int(oldp, oldlenp, newp, newlen,
1358		    &subnetsarelocal));
1359	case IPCTL_MTUDISC:
1360		error = sysctl_int(oldp, oldlenp, newp, newlen,
1361		    &ip_mtudisc);
1362		if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) {
1363			ip_mtudisc_timeout_q =
1364			    rt_timer_queue_create(ip_mtudisc_timeout);
1365		} else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) {
1366			rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE);
1367			ip_mtudisc_timeout_q = NULL;
1368		}
1369		return error;
1370	case IPCTL_ANONPORTMIN:
1371		old = anonportmin;
1372		error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin);
1373		if (anonportmin >= anonportmax || anonportmin > 65535
1374#ifndef IPNOPRIVPORTS
1375		    || anonportmin < IPPORT_RESERVED
1376#endif
1377		    ) {
1378			anonportmin = old;
1379			return (EINVAL);
1380		}
1381		return (error);
1382	case IPCTL_ANONPORTMAX:
1383		old = anonportmax;
1384		error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax);
1385		if (anonportmin >= anonportmax || anonportmax > 65535
1386#ifndef IPNOPRIVPORTS
1387		    || anonportmax < IPPORT_RESERVED
1388#endif
1389		    ) {
1390			anonportmax = old;
1391			return (EINVAL);
1392		}
1393		return (error);
1394	case IPCTL_MTUDISCTIMEOUT:
1395		error = sysctl_int(oldp, oldlenp, newp, newlen,
1396		   &ip_mtudisc_timeout);
1397		if (ip_mtudisc_timeout_q != NULL)
1398			rt_timer_queue_change(ip_mtudisc_timeout_q,
1399					      ip_mtudisc_timeout);
1400		return (error);
1401	default:
1402		return (EOPNOTSUPP);
1403	}
1404	/* NOTREACHED */
1405}
1406