ip_input.c revision 1.353
1/*	$NetBSD: ip_input.c,v 1.353 2017/03/31 06:42:19 ozaki-r Exp $	*/
2
3/*
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 *    notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 *    notice, this list of conditions and the following disclaimer in the
14 *    documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the project nor the names of its contributors
16 *    may be used to endorse or promote products derived from this software
17 *    without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 */
31
32/*-
33 * Copyright (c) 1998 The NetBSD Foundation, Inc.
34 * All rights reserved.
35 *
36 * This code is derived from software contributed to The NetBSD Foundation
37 * by Public Access Networks Corporation ("Panix").  It was developed under
38 * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon.
39 *
40 * Redistribution and use in source and binary forms, with or without
41 * modification, are permitted provided that the following conditions
42 * are met:
43 * 1. Redistributions of source code must retain the above copyright
44 *    notice, this list of conditions and the following disclaimer.
45 * 2. Redistributions in binary form must reproduce the above copyright
46 *    notice, this list of conditions and the following disclaimer in the
47 *    documentation and/or other materials provided with the distribution.
48 *
49 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
50 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
51 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
52 * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
53 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
54 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
55 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
56 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
57 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
58 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
59 * POSSIBILITY OF SUCH DAMAGE.
60 */
61
62/*
63 * Copyright (c) 1982, 1986, 1988, 1993
64 *	The Regents of the University of California.  All rights reserved.
65 *
66 * Redistribution and use in source and binary forms, with or without
67 * modification, are permitted provided that the following conditions
68 * are met:
69 * 1. Redistributions of source code must retain the above copyright
70 *    notice, this list of conditions and the following disclaimer.
71 * 2. Redistributions in binary form must reproduce the above copyright
72 *    notice, this list of conditions and the following disclaimer in the
73 *    documentation and/or other materials provided with the distribution.
74 * 3. Neither the name of the University nor the names of its contributors
75 *    may be used to endorse or promote products derived from this software
76 *    without specific prior written permission.
77 *
78 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
79 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
80 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
81 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
82 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
83 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
84 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
85 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
86 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
87 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
88 * SUCH DAMAGE.
89 *
90 *	@(#)ip_input.c	8.2 (Berkeley) 1/4/94
91 */
92
93#include <sys/cdefs.h>
94__KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.353 2017/03/31 06:42:19 ozaki-r Exp $");
95
96#ifdef _KERNEL_OPT
97#include "opt_inet.h"
98#include "opt_compat_netbsd.h"
99#include "opt_gateway.h"
100#include "opt_ipsec.h"
101#include "opt_mrouting.h"
102#include "opt_mbuftrace.h"
103#include "opt_inet_csum.h"
104#include "opt_net_mpsafe.h"
105#endif
106
107#include "arp.h"
108
109#include <sys/param.h>
110#include <sys/systm.h>
111#include <sys/cpu.h>
112#include <sys/mbuf.h>
113#include <sys/domain.h>
114#include <sys/protosw.h>
115#include <sys/socket.h>
116#include <sys/socketvar.h>
117#include <sys/errno.h>
118#include <sys/time.h>
119#include <sys/kernel.h>
120#include <sys/pool.h>
121#include <sys/sysctl.h>
122#include <sys/kauth.h>
123
124#include <net/if.h>
125#include <net/if_dl.h>
126#include <net/route.h>
127#include <net/pktqueue.h>
128#include <net/pfil.h>
129
130#include <netinet/in.h>
131#include <netinet/in_systm.h>
132#include <netinet/ip.h>
133#include <netinet/in_pcb.h>
134#include <netinet/in_proto.h>
135#include <netinet/in_var.h>
136#include <netinet/ip_var.h>
137#include <netinet/ip_private.h>
138#include <netinet/ip_icmp.h>
139/* just for gif_ttl */
140#include <netinet/in_gif.h>
141#include "gif.h"
142#include <net/if_gre.h>
143#include "gre.h"
144
145#ifdef MROUTING
146#include <netinet/ip_mroute.h>
147#endif
148#include <netinet/portalgo.h>
149
150#ifdef IPSEC
151#include <netipsec/ipsec.h>
152#endif
153
154#ifndef	IPFORWARDING
155#ifdef GATEWAY
156#define	IPFORWARDING	1	/* forward IP packets not for us */
157#else /* GATEWAY */
158#define	IPFORWARDING	0	/* don't forward IP packets not for us */
159#endif /* GATEWAY */
160#endif /* IPFORWARDING */
161#ifndef	IPSENDREDIRECTS
162#define	IPSENDREDIRECTS	1
163#endif
164#ifndef IPFORWSRCRT
165#define	IPFORWSRCRT	1	/* forward source-routed packets */
166#endif
167#ifndef IPALLOWSRCRT
168#define	IPALLOWSRCRT	1	/* allow source-routed packets */
169#endif
170#ifndef IPMTUDISC
171#define IPMTUDISC	1
172#endif
173#ifndef IPMTUDISCTIMEOUT
174#define IPMTUDISCTIMEOUT (10 * 60)	/* as per RFC 1191 */
175#endif
176
177#ifdef COMPAT_50
178#include <compat/sys/time.h>
179#include <compat/sys/socket.h>
180#endif
181
182/*
183 * Note: DIRECTED_BROADCAST is handled this way so that previous
184 * configuration using this option will Just Work.
185 */
186#ifndef IPDIRECTEDBCAST
187#ifdef DIRECTED_BROADCAST
188#define IPDIRECTEDBCAST	1
189#else
190#define	IPDIRECTEDBCAST	0
191#endif /* DIRECTED_BROADCAST */
192#endif /* IPDIRECTEDBCAST */
193int	ipforwarding = IPFORWARDING;
194int	ipsendredirects = IPSENDREDIRECTS;
195int	ip_defttl = IPDEFTTL;
196int	ip_forwsrcrt = IPFORWSRCRT;
197int	ip_directedbcast = IPDIRECTEDBCAST;
198int	ip_allowsrcrt = IPALLOWSRCRT;
199int	ip_mtudisc = IPMTUDISC;
200int	ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
201#ifdef DIAGNOSTIC
202int	ipprintfs = 0;
203#endif
204
205int	ip_do_randomid = 0;
206
207/*
208 * XXX - Setting ip_checkinterface mostly implements the receive side of
209 * the Strong ES model described in RFC 1122, but since the routing table
210 * and transmit implementation do not implement the Strong ES model,
211 * setting this to 1 results in an odd hybrid.
212 *
213 * XXX - ip_checkinterface currently must be disabled if you use ipnat
214 * to translate the destination address to another local interface.
215 *
216 * XXX - ip_checkinterface must be disabled if you add IP aliases
217 * to the loopback interface instead of the interface where the
218 * packets for those addresses are received.
219 */
220static int		ip_checkinterface	__read_mostly = 0;
221
222struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
223
224pktqueue_t *		ip_pktq			__read_mostly;
225pfil_head_t *		inet_pfil_hook		__read_mostly;
226ipid_state_t *		ip_ids			__read_mostly;
227percpu_t *		ipstat_percpu		__read_mostly;
228
229static percpu_t		*ipforward_rt_percpu	__cacheline_aligned;
230
231uint16_t ip_id;
232
233#ifdef INET_CSUM_COUNTERS
234#include <sys/device.h>
235
236struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
237    NULL, "inet", "hwcsum bad");
238struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
239    NULL, "inet", "hwcsum ok");
240struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
241    NULL, "inet", "swcsum");
242
243#define	INET_CSUM_COUNTER_INCR(ev)	(ev)->ev_count++
244
245EVCNT_ATTACH_STATIC(ip_hwcsum_bad);
246EVCNT_ATTACH_STATIC(ip_hwcsum_ok);
247EVCNT_ATTACH_STATIC(ip_swcsum);
248
249#else
250
251#define	INET_CSUM_COUNTER_INCR(ev)	/* nothing */
252
253#endif /* INET_CSUM_COUNTERS */
254
255/*
256 * We need to save the IP options in case a protocol wants to respond
257 * to an incoming packet over the same route if the packet got here
258 * using IP source routing.  This allows connection establishment and
259 * maintenance when the remote end is on a network that is not known
260 * to us.
261 */
262
263static int	ip_nhops = 0;
264
265static	struct ip_srcrt {
266	struct	in_addr dst;			/* final destination */
267	char	nop;				/* one NOP to align */
268	char	srcopt[IPOPT_OFFSET + 1];	/* OPTVAL, OLEN and OFFSET */
269	struct	in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
270} ip_srcrt;
271
272static int ip_drainwanted;
273
274static void save_rte(u_char *, struct in_addr);
275
276#ifdef MBUFTRACE
277struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx");
278struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx");
279#endif
280
281static void		ipintr(void *);
282static void		ip_input(struct mbuf *);
283static void		ip_forward(struct mbuf *, int, struct ifnet *);
284static bool		ip_dooptions(struct mbuf *);
285static struct in_ifaddr *ip_rtaddr(struct in_addr, struct psref *);
286static void		sysctl_net_inet_ip_setup(struct sysctllog **);
287
288static struct in_ifaddr	*ip_match_our_address(struct ifnet *, struct ip *,
289			    int *);
290static struct in_ifaddr	*ip_match_our_address_broadcast(struct ifnet *,
291			    struct ip *);
292
293#ifdef NET_MPSAFE
294#define	SOFTNET_LOCK()		mutex_enter(softnet_lock)
295#define	SOFTNET_UNLOCK()	mutex_exit(softnet_lock)
296#else
297#define	SOFTNET_LOCK()		KASSERT(mutex_owned(softnet_lock))
298#define	SOFTNET_UNLOCK()	KASSERT(mutex_owned(softnet_lock))
299#endif
300
301/*
302 * IP initialization: fill in IP protocol switch table.
303 * All protocols not implemented in kernel go to raw IP protocol handler.
304 */
305void
306ip_init(void)
307{
308	const struct protosw *pr;
309
310	in_init();
311	sysctl_net_inet_ip_setup(NULL);
312
313	pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
314	KASSERT(pr != NULL);
315
316	ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL);
317	KASSERT(ip_pktq != NULL);
318
319	for (u_int i = 0; i < IPPROTO_MAX; i++) {
320		ip_protox[i] = pr - inetsw;
321	}
322	for (pr = inetdomain.dom_protosw;
323	    pr < inetdomain.dom_protoswNPROTOSW; pr++)
324		if (pr->pr_domain->dom_family == PF_INET &&
325		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
326			ip_protox[pr->pr_protocol] = pr - inetsw;
327
328	ip_reass_init();
329
330	ip_ids = ip_id_init();
331	ip_id = time_uptime & 0xfffff;
332
333#ifdef GATEWAY
334	ipflow_init();
335#endif
336
337	/* Register our Packet Filter hook. */
338	inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET);
339	KASSERT(inet_pfil_hook != NULL);
340
341#ifdef MBUFTRACE
342	MOWNER_ATTACH(&ip_tx_mowner);
343	MOWNER_ATTACH(&ip_rx_mowner);
344#endif /* MBUFTRACE */
345
346	ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS);
347
348	ipforward_rt_percpu = percpu_alloc(sizeof(struct route));
349	if (ipforward_rt_percpu == NULL)
350		panic("failed to allocate ipforward_rt_percpu");
351
352	ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout);
353}
354
355static struct in_ifaddr *
356ip_match_our_address(struct ifnet *ifp, struct ip *ip, int *downmatch)
357{
358	struct in_ifaddr *ia = NULL;
359	int checkif;
360
361	/*
362	 * Enable a consistency check between the destination address
363	 * and the arrival interface for a unicast packet (the RFC 1122
364	 * strong ES model) if IP forwarding is disabled and the packet
365	 * is not locally generated.
366	 *
367	 * XXX - Checking also should be disabled if the destination
368	 * address is ipnat'ed to a different interface.
369	 *
370	 * XXX - Checking is incompatible with IP aliases added
371	 * to the loopback interface instead of the interface where
372	 * the packets are received.
373	 *
374	 * XXX - We need to add a per ifaddr flag for this so that
375	 * we get finer grain control.
376	 */
377	checkif = ip_checkinterface && (ipforwarding == 0) &&
378	    (ifp->if_flags & IFF_LOOPBACK) == 0;
379
380	IN_ADDRHASH_READER_FOREACH(ia, ip->ip_dst.s_addr) {
381		if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
382			if (ia->ia4_flags & IN_IFF_NOTREADY)
383				continue;
384			if (checkif && ia->ia_ifp != ifp)
385				continue;
386			if ((ia->ia_ifp->if_flags & IFF_UP) != 0 &&
387			    (ia->ia4_flags & IN_IFF_DETACHED) == 0)
388				break;
389			else
390				(*downmatch)++;
391		}
392	}
393
394	return ia;
395}
396
397static struct in_ifaddr *
398ip_match_our_address_broadcast(struct ifnet *ifp, struct ip *ip)
399{
400	struct in_ifaddr *ia = NULL;
401	struct ifaddr *ifa;
402
403	IFADDR_READER_FOREACH(ifa, ifp) {
404		if (ifa->ifa_addr->sa_family != AF_INET)
405			continue;
406		ia = ifatoia(ifa);
407		if (ia->ia4_flags & (IN_IFF_NOTREADY | IN_IFF_DETACHED))
408			continue;
409		if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
410		    in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
411		    /*
412		     * Look for all-0's host part (old broadcast addr),
413		     * either for subnet or net.
414		     */
415		    ip->ip_dst.s_addr == ia->ia_subnet ||
416		    ip->ip_dst.s_addr == ia->ia_net)
417			goto matched;
418		/*
419		 * An interface with IP address zero accepts
420		 * all packets that arrive on that interface.
421		 */
422		if (in_nullhost(ia->ia_addr.sin_addr))
423			goto matched;
424	}
425	ia = NULL;
426
427matched:
428	return ia;
429}
430
431/*
432 * IP software interrupt routine.
433 */
434static void
435ipintr(void *arg __unused)
436{
437	struct mbuf *m;
438
439	KASSERT(cpu_softintr_p());
440
441#ifndef NET_MPSAFE
442	mutex_enter(softnet_lock);
443#endif
444	while ((m = pktq_dequeue(ip_pktq)) != NULL) {
445		ip_input(m);
446	}
447#ifndef NET_MPSAFE
448	mutex_exit(softnet_lock);
449#endif
450}
451
452/*
453 * IP input routine.  Checksum and byte swap header.  If fragmented
454 * try to reassemble.  Process options.  Pass to next level.
455 */
456static void
457ip_input(struct mbuf *m)
458{
459	struct ip *ip = NULL;
460	struct in_ifaddr *ia = NULL;
461	int hlen = 0, len;
462	int downmatch;
463	int srcrt = 0;
464	ifnet_t *ifp;
465	struct psref psref;
466	int s;
467
468	KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software "
469	    "interrupt handler; synchronization assumptions violated");
470
471	MCLAIM(m, &ip_rx_mowner);
472	KASSERT((m->m_flags & M_PKTHDR) != 0);
473
474	ifp = m_get_rcvif_psref(m, &psref);
475	if (__predict_false(ifp == NULL))
476		goto out;
477
478	/*
479	 * If no IP addresses have been set yet but the interfaces
480	 * are receiving, can't do anything with incoming packets yet.
481	 * Note: we pre-check without locks held.
482	 */
483	if (IN_ADDRLIST_READER_EMPTY())
484		goto out;
485	IP_STATINC(IP_STAT_TOTAL);
486
487	/*
488	 * If the IP header is not aligned, slurp it up into a new
489	 * mbuf with space for link headers, in the event we forward
490	 * it.  Otherwise, if it is aligned, make sure the entire
491	 * base IP header is in the first mbuf of the chain.
492	 */
493	if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
494		if ((m = m_copyup(m, sizeof(struct ip),
495				  (max_linkhdr + 3) & ~3)) == NULL) {
496			/* XXXJRT new stat, please */
497			IP_STATINC(IP_STAT_TOOSMALL);
498			goto out;
499		}
500	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
501		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
502			IP_STATINC(IP_STAT_TOOSMALL);
503			goto out;
504		}
505	}
506	ip = mtod(m, struct ip *);
507	if (ip->ip_v != IPVERSION) {
508		IP_STATINC(IP_STAT_BADVERS);
509		goto out;
510	}
511	hlen = ip->ip_hl << 2;
512	if (hlen < sizeof(struct ip)) {	/* minimum header length */
513		IP_STATINC(IP_STAT_BADHLEN);
514		goto out;
515	}
516	if (hlen > m->m_len) {
517		if ((m = m_pullup(m, hlen)) == NULL) {
518			IP_STATINC(IP_STAT_BADHLEN);
519			goto out;
520		}
521		ip = mtod(m, struct ip *);
522	}
523
524	/*
525	 * RFC1122: packets with a multicast source address are
526	 * not allowed.
527	 */
528	if (IN_MULTICAST(ip->ip_src.s_addr)) {
529		IP_STATINC(IP_STAT_BADADDR);
530		goto out;
531	}
532
533	/* 127/8 must not appear on wire - RFC1122 */
534	if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
535	    (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
536		if ((ifp->if_flags & IFF_LOOPBACK) == 0) {
537			IP_STATINC(IP_STAT_BADADDR);
538			goto out;
539		}
540	}
541
542	switch (m->m_pkthdr.csum_flags &
543		((ifp->if_csum_flags_rx & M_CSUM_IPv4) |
544		 M_CSUM_IPv4_BAD)) {
545	case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
546		INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
547		IP_STATINC(IP_STAT_BADSUM);
548		goto out;
549
550	case M_CSUM_IPv4:
551		/* Checksum was okay. */
552		INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
553		break;
554
555	default:
556		/*
557		 * Must compute it ourselves.  Maybe skip checksum on
558		 * loopback interfaces.
559		 */
560		if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) ||
561		    ip_do_loopback_cksum)) {
562			INET_CSUM_COUNTER_INCR(&ip_swcsum);
563			if (in_cksum(m, hlen) != 0) {
564				IP_STATINC(IP_STAT_BADSUM);
565				goto out;
566			}
567		}
568		break;
569	}
570
571	/* Retrieve the packet length. */
572	len = ntohs(ip->ip_len);
573
574	/*
575	 * Check for additional length bogosity
576	 */
577	if (len < hlen) {
578		IP_STATINC(IP_STAT_BADLEN);
579		goto out;
580	}
581
582	/*
583	 * Check that the amount of data in the buffers
584	 * is as at least much as the IP header would have us expect.
585	 * Trim mbufs if longer than we expect.
586	 * Drop packet if shorter than we expect.
587	 */
588	if (m->m_pkthdr.len < len) {
589		IP_STATINC(IP_STAT_TOOSHORT);
590		goto out;
591	}
592	if (m->m_pkthdr.len > len) {
593		if (m->m_len == m->m_pkthdr.len) {
594			m->m_len = len;
595			m->m_pkthdr.len = len;
596		} else
597			m_adj(m, len - m->m_pkthdr.len);
598	}
599
600	/*
601	 * Assume that we can create a fast-forward IP flow entry
602	 * based on this packet.
603	 */
604	m->m_flags |= M_CANFASTFWD;
605
606	/*
607	 * Run through list of hooks for input packets.  If there are any
608	 * filters which require that additional packets in the flow are
609	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
610	 * Note that filters must _never_ set this flag, as another filter
611	 * in the list may have previously cleared it.
612	 */
613#if defined(IPSEC)
614	if (!ipsec_used || !ipsec_indone(m))
615#else
616	if (1)
617#endif
618	{
619		struct in_addr odst = ip->ip_dst;
620		bool freed;
621
622		freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0;
623		if (freed || m == NULL) {
624			m = NULL;
625			goto out;
626		}
627		ip = mtod(m, struct ip *);
628		hlen = ip->ip_hl << 2;
629
630		/*
631		 * XXX The setting of "srcrt" here is to prevent ip_forward()
632		 * from generating ICMP redirects for packets that have
633		 * been redirected by a hook back out on to the same LAN that
634		 * they came from and is not an indication that the packet
635		 * is being inffluenced by source routing options.  This
636		 * allows things like
637		 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
638		 * where tlp0 is both on the 1.1.1.0/24 network and is the
639		 * default route for hosts on 1.1.1.0/24.  Of course this
640		 * also requires a "map tlp0 ..." to complete the story.
641		 * One might argue whether or not this kind of network config.
642		 * should be supported in this manner...
643		 */
644		srcrt = (odst.s_addr != ip->ip_dst.s_addr);
645	}
646
647#ifdef ALTQ
648	/* XXX Temporary until ALTQ is changed to use a pfil hook */
649	if (altq_input) {
650		SOFTNET_LOCK();
651		if ((*altq_input)(m, AF_INET) == 0) {
652			/* Packet dropped by traffic conditioner. */
653			SOFTNET_UNLOCK();
654			m = NULL;
655			goto out;
656		}
657		SOFTNET_UNLOCK();
658	}
659#endif
660
661	/*
662	 * Process options and, if not destined for us,
663	 * ship it on.  ip_dooptions returns 1 when an
664	 * error was detected (causing an icmp message
665	 * to be sent and the original packet to be freed).
666	 */
667	ip_nhops = 0;		/* for source routed packets */
668	if (hlen > sizeof (struct ip) && ip_dooptions(m)) {
669		m = NULL;
670		goto out;
671	}
672
673	/*
674	 * Check our list of addresses, to see if the packet is for us.
675	 *
676	 * Traditional 4.4BSD did not consult IFF_UP at all.
677	 * The behavior here is to treat addresses on !IFF_UP interface
678	 * or IN_IFF_NOTREADY addresses as not mine.
679	 */
680	downmatch = 0;
681	s = pserialize_read_enter();
682	ia = ip_match_our_address(ifp, ip, &downmatch);
683	if (ia != NULL) {
684		pserialize_read_exit(s);
685		goto ours;
686	}
687
688	if (ifp->if_flags & IFF_BROADCAST) {
689		ia = ip_match_our_address_broadcast(ifp, ip);
690		if (ia != NULL) {
691			pserialize_read_exit(s);
692			goto ours;
693		}
694	}
695	pserialize_read_exit(s);
696
697	if (IN_MULTICAST(ip->ip_dst.s_addr)) {
698#ifdef MROUTING
699		extern struct socket *ip_mrouter;
700
701		if (ip_mrouter) {
702			/*
703			 * If we are acting as a multicast router, all
704			 * incoming multicast packets are passed to the
705			 * kernel-level multicast forwarding function.
706			 * The packet is returned (relatively) intact; if
707			 * ip_mforward() returns a non-zero value, the packet
708			 * must be discarded, else it may be accepted below.
709			 *
710			 * (The IP ident field is put in the same byte order
711			 * as expected when ip_mforward() is called from
712			 * ip_output().)
713			 */
714			SOFTNET_LOCK();
715			if (ip_mforward(m, ifp) != 0) {
716				SOFTNET_UNLOCK();
717				IP_STATINC(IP_STAT_CANTFORWARD);
718				goto out;
719			}
720			SOFTNET_UNLOCK();
721
722			/*
723			 * The process-level routing demon needs to receive
724			 * all multicast IGMP packets, whether or not this
725			 * host belongs to their destination groups.
726			 */
727			if (ip->ip_p == IPPROTO_IGMP) {
728				goto ours;
729			}
730			IP_STATINC(IP_STAT_CANTFORWARD);
731		}
732#endif
733		/*
734		 * See if we belong to the destination multicast group on the
735		 * arrival interface.
736		 */
737		if (!in_multi_group(ip->ip_dst, ifp, 0)) {
738			IP_STATINC(IP_STAT_CANTFORWARD);
739			goto out;
740		}
741		goto ours;
742	}
743	if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
744	    in_nullhost(ip->ip_dst))
745		goto ours;
746
747	/*
748	 * Not for us; forward if possible and desirable.
749	 */
750	if (ipforwarding == 0) {
751		m_put_rcvif_psref(ifp, &psref);
752		IP_STATINC(IP_STAT_CANTFORWARD);
753		m_freem(m);
754	} else {
755		/*
756		 * If ip_dst matched any of my address on !IFF_UP interface,
757		 * and there's no IFF_UP interface that matches ip_dst,
758		 * send icmp unreach.  Forwarding it will result in in-kernel
759		 * forwarding loop till TTL goes to 0.
760		 */
761		if (downmatch) {
762			m_put_rcvif_psref(ifp, &psref);
763			icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
764			IP_STATINC(IP_STAT_CANTFORWARD);
765			return;
766		}
767#ifdef IPSEC
768		/* Perform IPsec, if any. */
769		if (ipsec_used) {
770			SOFTNET_LOCK();
771			if (ipsec4_input(m, IP_FORWARDING |
772			    (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) {
773				SOFTNET_UNLOCK();
774				goto out;
775			}
776			SOFTNET_UNLOCK();
777		}
778#endif
779		ip_forward(m, srcrt, ifp);
780		m_put_rcvif_psref(ifp, &psref);
781	}
782	return;
783
784ours:
785	m_put_rcvif_psref(ifp, &psref);
786	ifp = NULL;
787
788	/*
789	 * If offset or IP_MF are set, must reassemble.
790	 */
791	if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
792		/*
793		 * Pass to IP reassembly mechanism.
794		 */
795		if (ip_reass_packet(&m, ip) != 0) {
796			/* Failed; invalid fragment(s) or packet. */
797			goto out;
798		}
799		if (m == NULL) {
800			/* More fragments should come; silently return. */
801			goto out;
802		}
803		/*
804		 * Reassembly is done, we have the final packet.
805		 * Updated cached data in local variable(s).
806		 */
807		ip = mtod(m, struct ip *);
808		hlen = ip->ip_hl << 2;
809	}
810
811#ifdef IPSEC
812	/*
813	 * Enforce IPsec policy checking if we are seeing last header.
814	 * Note that we do not visit this with protocols with PCB layer
815	 * code - like UDP/TCP/raw IP.
816	 */
817	if (ipsec_used &&
818	    (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
819		SOFTNET_LOCK();
820		if (ipsec4_input(m, 0) != 0) {
821			SOFTNET_UNLOCK();
822			goto out;
823		}
824		SOFTNET_UNLOCK();
825	}
826#endif
827
828	/*
829	 * Switch out to protocol's input routine.
830	 */
831#if IFA_STATS
832	if (ia && ip) {
833		struct in_ifaddr *_ia;
834		/*
835		 * Keep a reference from ip_match_our_address with psref
836		 * is expensive, so explore ia here again.
837		 */
838		s = pserialize_read_enter();
839		_ia = in_get_ia(ip->ip_dst);
840		_ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
841		pserialize_read_exit(s);
842	}
843#endif
844	IP_STATINC(IP_STAT_DELIVERED);
845
846	const int off = hlen, nh = ip->ip_p;
847
848	SOFTNET_LOCK();
849	(*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
850	SOFTNET_UNLOCK();
851	return;
852
853out:
854	m_put_rcvif_psref(ifp, &psref);
855	if (m != NULL)
856		m_freem(m);
857}
858
859/*
860 * IP timer processing.
861 */
862void
863ip_slowtimo(void)
864{
865
866#ifndef NET_MPSAFE
867	mutex_enter(softnet_lock);
868	KERNEL_LOCK(1, NULL);
869#endif
870
871	ip_reass_slowtimo();
872
873#ifndef NET_MPSAFE
874	KERNEL_UNLOCK_ONE(NULL);
875	mutex_exit(softnet_lock);
876#endif
877}
878
879/*
880 * IP drain processing.
881 */
882void
883ip_drain(void)
884{
885
886	KERNEL_LOCK(1, NULL);
887	ip_reass_drain();
888	KERNEL_UNLOCK_ONE(NULL);
889}
890
891/*
892 * ip_dooptions: perform option processing on a datagram, possibly discarding
893 * it if bad options are encountered, or forwarding it if source-routed.
894 *
895 * => Returns true if packet has been forwarded/freed.
896 * => Returns false if the packet should be processed further.
897 */
898static bool
899ip_dooptions(struct mbuf *m)
900{
901	struct ip *ip = mtod(m, struct ip *);
902	u_char *cp, *cp0;
903	struct ip_timestamp *ipt;
904	struct in_ifaddr *ia;
905	int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
906	struct in_addr dst;
907	n_time ntime;
908	struct ifaddr *ifa = NULL;
909	int s;
910
911	dst = ip->ip_dst;
912	cp = (u_char *)(ip + 1);
913	cnt = (ip->ip_hl << 2) - sizeof (struct ip);
914	for (; cnt > 0; cnt -= optlen, cp += optlen) {
915		opt = cp[IPOPT_OPTVAL];
916		if (opt == IPOPT_EOL)
917			break;
918		if (opt == IPOPT_NOP)
919			optlen = 1;
920		else {
921			if (cnt < IPOPT_OLEN + sizeof(*cp)) {
922				code = &cp[IPOPT_OLEN] - (u_char *)ip;
923				goto bad;
924			}
925			optlen = cp[IPOPT_OLEN];
926			if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
927				code = &cp[IPOPT_OLEN] - (u_char *)ip;
928				goto bad;
929			}
930		}
931		switch (opt) {
932
933		default:
934			break;
935
936		/*
937		 * Source routing with record.
938		 * Find interface with current destination address.
939		 * If none on this machine then drop if strictly routed,
940		 * or do nothing if loosely routed.
941		 * Record interface address and bring up next address
942		 * component.  If strictly routed make sure next
943		 * address is on directly accessible net.
944		 */
945		case IPOPT_LSRR:
946		case IPOPT_SSRR: {
947			struct psref psref;
948			struct sockaddr_in ipaddr = {
949			    .sin_len = sizeof(ipaddr),
950			    .sin_family = AF_INET,
951			};
952
953			if (ip_allowsrcrt == 0) {
954				type = ICMP_UNREACH;
955				code = ICMP_UNREACH_NET_PROHIB;
956				goto bad;
957			}
958			if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
959				code = &cp[IPOPT_OLEN] - (u_char *)ip;
960				goto bad;
961			}
962			if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
963				code = &cp[IPOPT_OFFSET] - (u_char *)ip;
964				goto bad;
965			}
966			ipaddr.sin_addr = ip->ip_dst;
967
968			s = pserialize_read_enter();
969			ifa = ifa_ifwithaddr(sintosa(&ipaddr));
970			if (ifa == NULL) {
971				pserialize_read_exit(s);
972				if (opt == IPOPT_SSRR) {
973					type = ICMP_UNREACH;
974					code = ICMP_UNREACH_SRCFAIL;
975					goto bad;
976				}
977				/*
978				 * Loose routing, and not at next destination
979				 * yet; nothing to do except forward.
980				 */
981				break;
982			}
983			pserialize_read_exit(s);
984
985			off--;			/* 0 origin */
986			if ((off + sizeof(struct in_addr)) > optlen) {
987				/*
988				 * End of source route.  Should be for us.
989				 */
990				save_rte(cp, ip->ip_src);
991				break;
992			}
993			/*
994			 * locate outgoing interface
995			 */
996			memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off),
997			    sizeof(ipaddr.sin_addr));
998			if (opt == IPOPT_SSRR) {
999				ifa = ifa_ifwithladdr_psref(sintosa(&ipaddr),
1000				    &psref);
1001				if (ifa != NULL)
1002					ia = ifatoia(ifa);
1003				else
1004					ia = NULL;
1005			} else {
1006				ia = ip_rtaddr(ipaddr.sin_addr, &psref);
1007			}
1008			if (ia == NULL) {
1009				type = ICMP_UNREACH;
1010				code = ICMP_UNREACH_SRCFAIL;
1011				goto bad;
1012			}
1013			ip->ip_dst = ipaddr.sin_addr;
1014			bcopy((void *)&ia->ia_addr.sin_addr,
1015			    (void *)(cp + off), sizeof(struct in_addr));
1016			ia4_release(ia, &psref);
1017			cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1018			/*
1019			 * Let ip_intr's mcast routing check handle mcast pkts
1020			 */
1021			forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1022			break;
1023		    }
1024
1025		case IPOPT_RR: {
1026			struct psref psref;
1027			struct sockaddr_in ipaddr = {
1028			    .sin_len = sizeof(ipaddr),
1029			    .sin_family = AF_INET,
1030			};
1031
1032			if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1033				code = &cp[IPOPT_OLEN] - (u_char *)ip;
1034				goto bad;
1035			}
1036			if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1037				code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1038				goto bad;
1039			}
1040			/*
1041			 * If no space remains, ignore.
1042			 */
1043			off--;			/* 0 origin */
1044			if ((off + sizeof(struct in_addr)) > optlen)
1045				break;
1046			memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst),
1047			    sizeof(ipaddr.sin_addr));
1048			/*
1049			 * locate outgoing interface; if we're the destination,
1050			 * use the incoming interface (should be same).
1051			 */
1052			ifa = ifa_ifwithaddr_psref(sintosa(&ipaddr), &psref);
1053			if (ifa == NULL) {
1054				ia = ip_rtaddr(ipaddr.sin_addr, &psref);
1055				if (ia == NULL) {
1056					type = ICMP_UNREACH;
1057					code = ICMP_UNREACH_HOST;
1058					goto bad;
1059				}
1060			} else {
1061				ia = ifatoia(ifa);
1062			}
1063			bcopy((void *)&ia->ia_addr.sin_addr,
1064			    (void *)(cp + off), sizeof(struct in_addr));
1065			ia4_release(ia, &psref);
1066			cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1067			break;
1068		    }
1069
1070		case IPOPT_TS:
1071			code = cp - (u_char *)ip;
1072			ipt = (struct ip_timestamp *)cp;
1073			if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1074				code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1075				goto bad;
1076			}
1077			if (ipt->ipt_ptr < 5) {
1078				code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1079				goto bad;
1080			}
1081			if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1082				if (++ipt->ipt_oflw == 0) {
1083					code = (u_char *)&ipt->ipt_ptr -
1084					    (u_char *)ip;
1085					goto bad;
1086				}
1087				break;
1088			}
1089			cp0 = (cp + ipt->ipt_ptr - 1);
1090			switch (ipt->ipt_flg) {
1091
1092			case IPOPT_TS_TSONLY:
1093				break;
1094
1095			case IPOPT_TS_TSANDADDR: {
1096				struct ifnet *rcvif;
1097				int _s, _ss;
1098				struct sockaddr_in ipaddr = {
1099				    .sin_len = sizeof(ipaddr),
1100				    .sin_family = AF_INET,
1101				};
1102
1103				if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1104				    sizeof(struct in_addr) > ipt->ipt_len) {
1105					code = (u_char *)&ipt->ipt_ptr -
1106					    (u_char *)ip;
1107					goto bad;
1108				}
1109				ipaddr.sin_addr = dst;
1110				_ss = pserialize_read_enter();
1111				rcvif = m_get_rcvif(m, &_s);
1112				if (__predict_true(rcvif != NULL)) {
1113					ifa = ifaof_ifpforaddr(sintosa(&ipaddr),
1114					    rcvif);
1115				}
1116				m_put_rcvif(rcvif, &_s);
1117				if (ifa == NULL) {
1118					pserialize_read_exit(_ss);
1119					break;
1120				}
1121				ia = ifatoia(ifa);
1122				bcopy(&ia->ia_addr.sin_addr,
1123				    cp0, sizeof(struct in_addr));
1124				pserialize_read_exit(_ss);
1125				ipt->ipt_ptr += sizeof(struct in_addr);
1126				break;
1127			}
1128
1129			case IPOPT_TS_PRESPEC: {
1130				struct sockaddr_in ipaddr = {
1131				    .sin_len = sizeof(ipaddr),
1132				    .sin_family = AF_INET,
1133				};
1134
1135				if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1136				    sizeof(struct in_addr) > ipt->ipt_len) {
1137					code = (u_char *)&ipt->ipt_ptr -
1138					    (u_char *)ip;
1139					goto bad;
1140				}
1141				memcpy(&ipaddr.sin_addr, cp0,
1142				    sizeof(struct in_addr));
1143				s = pserialize_read_enter();
1144				ifa = ifa_ifwithaddr(sintosa(&ipaddr));
1145				if (ifa == NULL) {
1146					pserialize_read_exit(s);
1147					continue;
1148				}
1149				pserialize_read_exit(s);
1150				ipt->ipt_ptr += sizeof(struct in_addr);
1151				break;
1152			    }
1153
1154			default:
1155				/* XXX can't take &ipt->ipt_flg */
1156				code = (u_char *)&ipt->ipt_ptr -
1157				    (u_char *)ip + 1;
1158				goto bad;
1159			}
1160			ntime = iptime();
1161			cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1162			memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1163			    sizeof(n_time));
1164			ipt->ipt_ptr += sizeof(n_time);
1165		}
1166	}
1167	if (forward) {
1168		struct ifnet *rcvif;
1169		struct psref _psref;
1170
1171		if (ip_forwsrcrt == 0) {
1172			type = ICMP_UNREACH;
1173			code = ICMP_UNREACH_SRCFAIL;
1174			goto bad;
1175		}
1176
1177		rcvif = m_get_rcvif_psref(m, &_psref);
1178		if (__predict_false(rcvif == NULL)) {
1179			type = ICMP_UNREACH;
1180			code = ICMP_UNREACH_HOST;
1181			goto bad;
1182		}
1183		ip_forward(m, 1, rcvif);
1184		m_put_rcvif_psref(rcvif, &_psref);
1185		return true;
1186	}
1187	return false;
1188bad:
1189	icmp_error(m, type, code, 0, 0);
1190	IP_STATINC(IP_STAT_BADOPTIONS);
1191	return true;
1192}
1193
1194/*
1195 * ip_rtaddr: given address of next destination (final or next hop),
1196 * return internet address info of interface to be used to get there.
1197 */
1198static struct in_ifaddr *
1199ip_rtaddr(struct in_addr dst, struct psref *psref)
1200{
1201	struct rtentry *rt;
1202	union {
1203		struct sockaddr		dst;
1204		struct sockaddr_in	dst4;
1205	} u;
1206	struct route *ro;
1207
1208	sockaddr_in_init(&u.dst4, &dst, 0);
1209
1210	ro = percpu_getref(ipforward_rt_percpu);
1211	rt = rtcache_lookup(ro, &u.dst);
1212	if (rt == NULL) {
1213		percpu_putref(ipforward_rt_percpu);
1214		return NULL;
1215	}
1216
1217	ia4_acquire(ifatoia(rt->rt_ifa), psref);
1218	rtcache_unref(rt, ro);
1219	percpu_putref(ipforward_rt_percpu);
1220
1221	return ifatoia(rt->rt_ifa);
1222}
1223
1224/*
1225 * save_rte: save incoming source route for use in replies, to be picked
1226 * up later by ip_srcroute if the receiver is interested.
1227 */
1228static void
1229save_rte(u_char *option, struct in_addr dst)
1230{
1231	unsigned olen;
1232
1233	olen = option[IPOPT_OLEN];
1234	if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1235		return;
1236	memcpy((void *)ip_srcrt.srcopt, (void *)option, olen);
1237	ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1238	ip_srcrt.dst = dst;
1239}
1240
1241/*
1242 * Retrieve incoming source route for use in replies,
1243 * in the same form used by setsockopt.
1244 * The first hop is placed before the options, will be removed later.
1245 */
1246struct mbuf *
1247ip_srcroute(void)
1248{
1249	struct in_addr *p, *q;
1250	struct mbuf *m;
1251
1252	if (ip_nhops == 0)
1253		return NULL;
1254	m = m_get(M_DONTWAIT, MT_SOOPTS);
1255	if (m == 0)
1256		return NULL;
1257
1258	MCLAIM(m, &inetdomain.dom_mowner);
1259#define OPTSIZ	(sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1260
1261	/* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1262	m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1263	    OPTSIZ;
1264
1265	/*
1266	 * First save first hop for return route
1267	 */
1268	p = &ip_srcrt.route[ip_nhops - 1];
1269	*(mtod(m, struct in_addr *)) = *p--;
1270
1271	/*
1272	 * Copy option fields and padding (nop) to mbuf.
1273	 */
1274	ip_srcrt.nop = IPOPT_NOP;
1275	ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1276	memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop,
1277	    OPTSIZ);
1278	q = (struct in_addr *)(mtod(m, char *) +
1279	    sizeof(struct in_addr) + OPTSIZ);
1280#undef OPTSIZ
1281	/*
1282	 * Record return path as an IP source route,
1283	 * reversing the path (pointers are now aligned).
1284	 */
1285	while (p >= ip_srcrt.route) {
1286		*q++ = *p--;
1287	}
1288	/*
1289	 * Last hop goes to final destination.
1290	 */
1291	*q = ip_srcrt.dst;
1292	return (m);
1293}
1294
1295const int inetctlerrmap[PRC_NCMDS] = {
1296	[PRC_MSGSIZE] = EMSGSIZE,
1297	[PRC_HOSTDEAD] = EHOSTDOWN,
1298	[PRC_HOSTUNREACH] = EHOSTUNREACH,
1299	[PRC_UNREACH_NET] = EHOSTUNREACH,
1300	[PRC_UNREACH_HOST] = EHOSTUNREACH,
1301	[PRC_UNREACH_PROTOCOL] = ECONNREFUSED,
1302	[PRC_UNREACH_PORT] = ECONNREFUSED,
1303	[PRC_UNREACH_SRCFAIL] = EHOSTUNREACH,
1304	[PRC_PARAMPROB] = ENOPROTOOPT,
1305};
1306
1307void
1308ip_fasttimo(void)
1309{
1310	if (ip_drainwanted) {
1311		ip_drain();
1312		ip_drainwanted = 0;
1313	}
1314}
1315
1316void
1317ip_drainstub(void)
1318{
1319	ip_drainwanted = 1;
1320}
1321
1322/*
1323 * Forward a packet.  If some error occurs return the sender
1324 * an icmp packet.  Note we can't always generate a meaningful
1325 * icmp message because icmp doesn't have a large enough repertoire
1326 * of codes and types.
1327 *
1328 * If not forwarding, just drop the packet.  This could be confusing
1329 * if ipforwarding was zero but some routing protocol was advancing
1330 * us as a gateway to somewhere.  However, we must let the routing
1331 * protocol deal with that.
1332 *
1333 * The srcrt parameter indicates whether the packet is being forwarded
1334 * via a source route.
1335 */
1336static void
1337ip_forward(struct mbuf *m, int srcrt, struct ifnet *rcvif)
1338{
1339	struct ip *ip = mtod(m, struct ip *);
1340	struct rtentry *rt;
1341	int error, type = 0, code = 0, destmtu = 0;
1342	struct mbuf *mcopy;
1343	n_long dest;
1344	union {
1345		struct sockaddr		dst;
1346		struct sockaddr_in	dst4;
1347	} u;
1348	uint64_t *ips;
1349	struct route *ro;
1350
1351	KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software "
1352	    "interrupt handler; synchronization assumptions violated");
1353
1354	/*
1355	 * We are now in the output path.
1356	 */
1357	MCLAIM(m, &ip_tx_mowner);
1358
1359	/*
1360	 * Clear any in-bound checksum flags for this packet.
1361	 */
1362	m->m_pkthdr.csum_flags = 0;
1363
1364	dest = 0;
1365	if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1366		IP_STATINC(IP_STAT_CANTFORWARD);
1367		m_freem(m);
1368		return;
1369	}
1370
1371	if (ip->ip_ttl <= IPTTLDEC) {
1372		icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1373		return;
1374	}
1375
1376	sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1377
1378	ro = percpu_getref(ipforward_rt_percpu);
1379	rt = rtcache_lookup(ro, &u.dst);
1380	if (rt == NULL) {
1381		percpu_putref(ipforward_rt_percpu);
1382		icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1383		return;
1384	}
1385
1386	/*
1387	 * Save at most 68 bytes of the packet in case
1388	 * we need to generate an ICMP message to the src.
1389	 * Pullup to avoid sharing mbuf cluster between m and mcopy.
1390	 */
1391	mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1392	if (mcopy)
1393		mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1394
1395	ip->ip_ttl -= IPTTLDEC;
1396
1397	/*
1398	 * If forwarding packet using same interface that it came in on,
1399	 * perhaps should send a redirect to sender to shortcut a hop.
1400	 * Only send redirect if source is sending directly to us,
1401	 * and if packet was not source routed (or has any options).
1402	 * Also, don't send redirect if forwarding using a default route
1403	 * or a route modified by a redirect.
1404	 */
1405	if (rt->rt_ifp == rcvif &&
1406	    (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1407	    !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1408	    ipsendredirects && !srcrt) {
1409		if (rt->rt_ifa &&
1410		    (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1411		    ifatoia(rt->rt_ifa)->ia_subnet) {
1412			if (rt->rt_flags & RTF_GATEWAY)
1413				dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1414			else
1415				dest = ip->ip_dst.s_addr;
1416			/*
1417			 * Router requirements says to only send host
1418			 * redirects.
1419			 */
1420			type = ICMP_REDIRECT;
1421			code = ICMP_REDIRECT_HOST;
1422		}
1423	}
1424	rtcache_unref(rt, ro);
1425
1426	error = ip_output(m, NULL, ro,
1427	    (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1428	    NULL, NULL);
1429
1430	if (error) {
1431		IP_STATINC(IP_STAT_CANTFORWARD);
1432		goto error;
1433	}
1434
1435	ips = IP_STAT_GETREF();
1436	ips[IP_STAT_FORWARD]++;
1437
1438	if (type) {
1439		ips[IP_STAT_REDIRECTSENT]++;
1440		IP_STAT_PUTREF();
1441		goto redirect;
1442	}
1443
1444	IP_STAT_PUTREF();
1445	if (mcopy) {
1446#ifdef GATEWAY
1447		if (mcopy->m_flags & M_CANFASTFWD)
1448			ipflow_create(ro, mcopy);
1449#endif
1450		m_freem(mcopy);
1451	}
1452
1453	percpu_putref(ipforward_rt_percpu);
1454	return;
1455
1456redirect:
1457error:
1458	if (mcopy == NULL) {
1459		percpu_putref(ipforward_rt_percpu);
1460		return;
1461	}
1462
1463	switch (error) {
1464
1465	case 0:				/* forwarded, but need redirect */
1466		/* type, code set above */
1467		break;
1468
1469	case ENETUNREACH:		/* shouldn't happen, checked above */
1470	case EHOSTUNREACH:
1471	case ENETDOWN:
1472	case EHOSTDOWN:
1473	default:
1474		type = ICMP_UNREACH;
1475		code = ICMP_UNREACH_HOST;
1476		break;
1477
1478	case EMSGSIZE:
1479		type = ICMP_UNREACH;
1480		code = ICMP_UNREACH_NEEDFRAG;
1481
1482		if ((rt = rtcache_validate(ro)) != NULL) {
1483			destmtu = rt->rt_ifp->if_mtu;
1484			rtcache_unref(rt, ro);
1485		}
1486#ifdef IPSEC
1487		if (ipsec_used)
1488			(void)ipsec4_forward(mcopy, &destmtu);
1489#endif
1490		IP_STATINC(IP_STAT_CANTFRAG);
1491		break;
1492
1493	case ENOBUFS:
1494		/*
1495		 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812,
1496		 * Requirements for IP Version 4 Routers.  Source quench can
1497		 * big problem under DoS attacks or if the underlying
1498		 * interface is rate-limited.
1499		 */
1500		if (mcopy)
1501			m_freem(mcopy);
1502		percpu_putref(ipforward_rt_percpu);
1503		return;
1504	}
1505	icmp_error(mcopy, type, code, dest, destmtu);
1506	percpu_putref(ipforward_rt_percpu);
1507}
1508
1509void
1510ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
1511    struct mbuf *m)
1512{
1513	struct socket *so = inp->inp_socket;
1514	ifnet_t *ifp;
1515	int inpflags = inp->inp_flags;
1516	struct psref psref;
1517
1518	ifp = m_get_rcvif_psref(m, &psref);
1519	if (__predict_false(ifp == NULL))
1520		return; /* XXX should report error? */
1521
1522	if (so->so_options & SO_TIMESTAMP
1523#ifdef SO_OTIMESTAMP
1524	    || so->so_options & SO_OTIMESTAMP
1525#endif
1526	    ) {
1527		struct timeval tv;
1528
1529		microtime(&tv);
1530#ifdef SO_OTIMESTAMP
1531		if (so->so_options & SO_OTIMESTAMP) {
1532			struct timeval50 tv50;
1533			timeval_to_timeval50(&tv, &tv50);
1534			*mp = sbcreatecontrol((void *) &tv50, sizeof(tv50),
1535			    SCM_OTIMESTAMP, SOL_SOCKET);
1536		} else
1537#endif
1538		*mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1539		    SCM_TIMESTAMP, SOL_SOCKET);
1540		if (*mp)
1541			mp = &(*mp)->m_next;
1542	}
1543	if (inpflags & INP_RECVDSTADDR) {
1544		*mp = sbcreatecontrol((void *) &ip->ip_dst,
1545		    sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1546		if (*mp)
1547			mp = &(*mp)->m_next;
1548	}
1549	if (inpflags & INP_RECVPKTINFO) {
1550		struct in_pktinfo ipi;
1551		ipi.ipi_addr = ip->ip_src;
1552		ipi.ipi_ifindex = ifp->if_index;
1553		*mp = sbcreatecontrol((void *) &ipi,
1554		    sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP);
1555		if (*mp)
1556			mp = &(*mp)->m_next;
1557	}
1558	if (inpflags & INP_PKTINFO) {
1559		struct in_pktinfo ipi;
1560		ipi.ipi_addr = ip->ip_dst;
1561		ipi.ipi_ifindex = ifp->if_index;
1562		*mp = sbcreatecontrol((void *) &ipi,
1563		    sizeof(ipi), IP_PKTINFO, IPPROTO_IP);
1564		if (*mp)
1565			mp = &(*mp)->m_next;
1566	}
1567	if (inpflags & INP_RECVIF) {
1568		struct sockaddr_dl sdl;
1569
1570		sockaddr_dl_init(&sdl, sizeof(sdl), ifp->if_index, 0, NULL, 0,
1571		    NULL, 0);
1572		*mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1573		if (*mp)
1574			mp = &(*mp)->m_next;
1575	}
1576	if (inpflags & INP_RECVTTL) {
1577		*mp = sbcreatecontrol((void *) &ip->ip_ttl,
1578		    sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP);
1579		if (*mp)
1580			mp = &(*mp)->m_next;
1581	}
1582	m_put_rcvif_psref(ifp, &psref);
1583}
1584
1585/*
1586 * sysctl helper routine for net.inet.ip.forwsrcrt.
1587 */
1588static int
1589sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
1590{
1591	int error, tmp;
1592	struct sysctlnode node;
1593
1594	node = *rnode;
1595	tmp = ip_forwsrcrt;
1596	node.sysctl_data = &tmp;
1597	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1598	if (error || newp == NULL)
1599		return (error);
1600
1601	error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
1602	    0, NULL, NULL, NULL);
1603	if (error)
1604		return (error);
1605
1606	ip_forwsrcrt = tmp;
1607
1608	return (0);
1609}
1610
1611/*
1612 * sysctl helper routine for net.inet.ip.mtudisctimeout.  checks the
1613 * range of the new value and tweaks timers if it changes.
1614 */
1615static int
1616sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1617{
1618	int error, tmp;
1619	struct sysctlnode node;
1620
1621	icmp_mtudisc_lock();
1622
1623	node = *rnode;
1624	tmp = ip_mtudisc_timeout;
1625	node.sysctl_data = &tmp;
1626	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1627	if (error || newp == NULL)
1628		goto out;
1629	if (tmp < 0) {
1630		error = EINVAL;
1631		goto out;
1632	}
1633
1634	ip_mtudisc_timeout = tmp;
1635	rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
1636	error = 0;
1637out:
1638	icmp_mtudisc_unlock();
1639	return error;
1640}
1641
1642static int
1643sysctl_net_inet_ip_stats(SYSCTLFN_ARGS)
1644{
1645
1646	return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS));
1647}
1648
1649static void
1650sysctl_net_inet_ip_setup(struct sysctllog **clog)
1651{
1652	sysctl_createv(clog, 0, NULL, NULL,
1653		       CTLFLAG_PERMANENT,
1654		       CTLTYPE_NODE, "inet",
1655		       SYSCTL_DESCR("PF_INET related settings"),
1656		       NULL, 0, NULL, 0,
1657		       CTL_NET, PF_INET, CTL_EOL);
1658	sysctl_createv(clog, 0, NULL, NULL,
1659		       CTLFLAG_PERMANENT,
1660		       CTLTYPE_NODE, "ip",
1661		       SYSCTL_DESCR("IPv4 related settings"),
1662		       NULL, 0, NULL, 0,
1663		       CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1664
1665	sysctl_createv(clog, 0, NULL, NULL,
1666		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1667		       CTLTYPE_INT, "forwarding",
1668		       SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1669		       NULL, 0, &ipforwarding, 0,
1670		       CTL_NET, PF_INET, IPPROTO_IP,
1671		       IPCTL_FORWARDING, CTL_EOL);
1672	sysctl_createv(clog, 0, NULL, NULL,
1673		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1674		       CTLTYPE_INT, "redirect",
1675		       SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1676		       NULL, 0, &ipsendredirects, 0,
1677		       CTL_NET, PF_INET, IPPROTO_IP,
1678		       IPCTL_SENDREDIRECTS, CTL_EOL);
1679	sysctl_createv(clog, 0, NULL, NULL,
1680		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1681		       CTLTYPE_INT, "ttl",
1682		       SYSCTL_DESCR("Default TTL for an INET datagram"),
1683		       NULL, 0, &ip_defttl, 0,
1684		       CTL_NET, PF_INET, IPPROTO_IP,
1685		       IPCTL_DEFTTL, CTL_EOL);
1686#ifdef IPCTL_DEFMTU
1687	sysctl_createv(clog, 0, NULL, NULL,
1688		       CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1689		       CTLTYPE_INT, "mtu",
1690		       SYSCTL_DESCR("Default MTA for an INET route"),
1691		       NULL, 0, &ip_mtu, 0,
1692		       CTL_NET, PF_INET, IPPROTO_IP,
1693		       IPCTL_DEFMTU, CTL_EOL);
1694#endif /* IPCTL_DEFMTU */
1695	sysctl_createv(clog, 0, NULL, NULL,
1696		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1697		       CTLTYPE_INT, "forwsrcrt",
1698		       SYSCTL_DESCR("Enable forwarding of source-routed "
1699				    "datagrams"),
1700		       sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1701		       CTL_NET, PF_INET, IPPROTO_IP,
1702		       IPCTL_FORWSRCRT, CTL_EOL);
1703	sysctl_createv(clog, 0, NULL, NULL,
1704		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1705		       CTLTYPE_INT, "directed-broadcast",
1706		       SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1707		       NULL, 0, &ip_directedbcast, 0,
1708		       CTL_NET, PF_INET, IPPROTO_IP,
1709		       IPCTL_DIRECTEDBCAST, CTL_EOL);
1710	sysctl_createv(clog, 0, NULL, NULL,
1711		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1712		       CTLTYPE_INT, "allowsrcrt",
1713		       SYSCTL_DESCR("Accept source-routed datagrams"),
1714		       NULL, 0, &ip_allowsrcrt, 0,
1715		       CTL_NET, PF_INET, IPPROTO_IP,
1716		       IPCTL_ALLOWSRCRT, CTL_EOL);
1717
1718	sysctl_createv(clog, 0, NULL, NULL,
1719		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1720		       CTLTYPE_INT, "mtudisc",
1721		       SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1722		       NULL, 0, &ip_mtudisc, 0,
1723		       CTL_NET, PF_INET, IPPROTO_IP,
1724		       IPCTL_MTUDISC, CTL_EOL);
1725	sysctl_createv(clog, 0, NULL, NULL,
1726		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1727		       CTLTYPE_INT, "anonportmin",
1728		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1729		       sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
1730		       CTL_NET, PF_INET, IPPROTO_IP,
1731		       IPCTL_ANONPORTMIN, CTL_EOL);
1732	sysctl_createv(clog, 0, NULL, NULL,
1733		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1734		       CTLTYPE_INT, "anonportmax",
1735		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
1736		       sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
1737		       CTL_NET, PF_INET, IPPROTO_IP,
1738		       IPCTL_ANONPORTMAX, CTL_EOL);
1739	sysctl_createv(clog, 0, NULL, NULL,
1740		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1741		       CTLTYPE_INT, "mtudisctimeout",
1742		       SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1743		       sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0,
1744		       CTL_NET, PF_INET, IPPROTO_IP,
1745		       IPCTL_MTUDISCTIMEOUT, CTL_EOL);
1746#ifndef IPNOPRIVPORTS
1747	sysctl_createv(clog, 0, NULL, NULL,
1748		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1749		       CTLTYPE_INT, "lowportmin",
1750		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
1751				    "to assign"),
1752		       sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
1753		       CTL_NET, PF_INET, IPPROTO_IP,
1754		       IPCTL_LOWPORTMIN, CTL_EOL);
1755	sysctl_createv(clog, 0, NULL, NULL,
1756		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1757		       CTLTYPE_INT, "lowportmax",
1758		       SYSCTL_DESCR("Highest privileged ephemeral port number "
1759				    "to assign"),
1760		       sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
1761		       CTL_NET, PF_INET, IPPROTO_IP,
1762		       IPCTL_LOWPORTMAX, CTL_EOL);
1763#endif /* IPNOPRIVPORTS */
1764#if NGRE > 0
1765	sysctl_createv(clog, 0, NULL, NULL,
1766		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1767		       CTLTYPE_INT, "grettl",
1768		       SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1769		       NULL, 0, &ip_gre_ttl, 0,
1770		       CTL_NET, PF_INET, IPPROTO_IP,
1771		       IPCTL_GRE_TTL, CTL_EOL);
1772#endif /* NGRE */
1773	sysctl_createv(clog, 0, NULL, NULL,
1774		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1775		       CTLTYPE_INT, "checkinterface",
1776		       SYSCTL_DESCR("Enable receive side of Strong ES model "
1777				    "from RFC1122"),
1778		       NULL, 0, &ip_checkinterface, 0,
1779		       CTL_NET, PF_INET, IPPROTO_IP,
1780		       IPCTL_CHECKINTERFACE, CTL_EOL);
1781	sysctl_createv(clog, 0, NULL, NULL,
1782		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1783		       CTLTYPE_INT, "random_id",
1784		       SYSCTL_DESCR("Assign random ip_id values"),
1785		       NULL, 0, &ip_do_randomid, 0,
1786		       CTL_NET, PF_INET, IPPROTO_IP,
1787		       IPCTL_RANDOMID, CTL_EOL);
1788	sysctl_createv(clog, 0, NULL, NULL,
1789		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1790		       CTLTYPE_INT, "do_loopback_cksum",
1791		       SYSCTL_DESCR("Perform IP checksum on loopback"),
1792		       NULL, 0, &ip_do_loopback_cksum, 0,
1793		       CTL_NET, PF_INET, IPPROTO_IP,
1794		       IPCTL_LOOPBACKCKSUM, CTL_EOL);
1795	sysctl_createv(clog, 0, NULL, NULL,
1796		       CTLFLAG_PERMANENT,
1797		       CTLTYPE_STRUCT, "stats",
1798		       SYSCTL_DESCR("IP statistics"),
1799		       sysctl_net_inet_ip_stats, 0, NULL, 0,
1800		       CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
1801		       CTL_EOL);
1802#if NARP
1803	sysctl_createv(clog, 0, NULL, NULL,
1804		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1805		       CTLTYPE_INT, "dad_count",
1806		       SYSCTL_DESCR("Number of Duplicate Address Detection "
1807				    "probes to send"),
1808		       NULL, 0, &ip_dad_count, 0,
1809		       CTL_NET, PF_INET, IPPROTO_IP,
1810		       IPCTL_DAD_COUNT, CTL_EOL);
1811#endif
1812
1813	/* anonportalgo RFC6056 subtree */
1814	const struct sysctlnode *portalgo_node;
1815	sysctl_createv(clog, 0, NULL, &portalgo_node,
1816		       CTLFLAG_PERMANENT,
1817		       CTLTYPE_NODE, "anonportalgo",
1818		       SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"),
1819	    	       NULL, 0, NULL, 0,
1820		       CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL);
1821	sysctl_createv(clog, 0, &portalgo_node, NULL,
1822		       CTLFLAG_PERMANENT,
1823		       CTLTYPE_STRING, "available",
1824		       SYSCTL_DESCR("available algorithms"),
1825		       sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
1826		       CTL_CREATE, CTL_EOL);
1827	sysctl_createv(clog, 0, &portalgo_node, NULL,
1828		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1829		       CTLTYPE_STRING, "selected",
1830		       SYSCTL_DESCR("selected algorithm"),
1831		       sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN,
1832		       CTL_CREATE, CTL_EOL);
1833	sysctl_createv(clog, 0, &portalgo_node, NULL,
1834		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1835		       CTLTYPE_STRUCT, "reserve",
1836		       SYSCTL_DESCR("bitmap of reserved ports"),
1837		       sysctl_portalgo_reserve4, 0, NULL, 0,
1838		       CTL_CREATE, CTL_EOL);
1839}
1840
1841void
1842ip_statinc(u_int stat)
1843{
1844
1845	KASSERT(stat < IP_NSTATS);
1846	IP_STATINC(stat);
1847}
1848