ip_input.c revision 1.358
1/*	$NetBSD: ip_input.c,v 1.358 2017/07/08 22:56:15 christos 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 1.358 2017/07/08 22:56:15 christos Exp $");
95
96#ifdef _KERNEL_OPT
97#include "opt_inet.h"
98#include "opt_gateway.h"
99#include "opt_ipsec.h"
100#include "opt_mrouting.h"
101#include "opt_mbuftrace.h"
102#include "opt_inet_csum.h"
103#include "opt_net_mpsafe.h"
104#endif
105
106#include "arp.h"
107
108#include <sys/param.h>
109#include <sys/systm.h>
110#include <sys/cpu.h>
111#include <sys/mbuf.h>
112#include <sys/domain.h>
113#include <sys/protosw.h>
114#include <sys/socket.h>
115#include <sys/socketvar.h>
116#include <sys/errno.h>
117#include <sys/time.h>
118#include <sys/kernel.h>
119#include <sys/pool.h>
120#include <sys/sysctl.h>
121#include <sys/kauth.h>
122
123#include <net/if.h>
124#include <net/if_dl.h>
125#include <net/route.h>
126#include <net/pktqueue.h>
127#include <net/pfil.h>
128
129#include <netinet/in.h>
130#include <netinet/in_systm.h>
131#include <netinet/ip.h>
132#include <netinet/in_pcb.h>
133#include <netinet/in_proto.h>
134#include <netinet/in_var.h>
135#include <netinet/ip_var.h>
136#include <netinet/ip_private.h>
137#include <netinet/ip_icmp.h>
138/* just for gif_ttl */
139#include <netinet/in_gif.h>
140#include "gif.h"
141#include <net/if_gre.h>
142#include "gre.h"
143
144#ifdef MROUTING
145#include <netinet/ip_mroute.h>
146#endif
147#include <netinet/portalgo.h>
148
149#ifdef IPSEC
150#include <netipsec/ipsec.h>
151#endif
152
153#ifndef	IPFORWARDING
154#ifdef GATEWAY
155#define	IPFORWARDING	1	/* forward IP packets not for us */
156#else /* GATEWAY */
157#define	IPFORWARDING	0	/* don't forward IP packets not for us */
158#endif /* GATEWAY */
159#endif /* IPFORWARDING */
160#ifndef	IPSENDREDIRECTS
161#define	IPSENDREDIRECTS	1
162#endif
163#ifndef IPFORWSRCRT
164#define	IPFORWSRCRT	1	/* forward source-routed packets */
165#endif
166#ifndef IPALLOWSRCRT
167#define	IPALLOWSRCRT	1	/* allow source-routed packets */
168#endif
169#ifndef IPMTUDISC
170#define IPMTUDISC	1
171#endif
172#ifndef IPMTUDISCTIMEOUT
173#define IPMTUDISCTIMEOUT (10 * 60)	/* as per RFC 1191 */
174#endif
175
176/*
177 * Note: DIRECTED_BROADCAST is handled this way so that previous
178 * configuration using this option will Just Work.
179 */
180#ifndef IPDIRECTEDBCAST
181#ifdef DIRECTED_BROADCAST
182#define IPDIRECTEDBCAST	1
183#else
184#define	IPDIRECTEDBCAST	0
185#endif /* DIRECTED_BROADCAST */
186#endif /* IPDIRECTEDBCAST */
187int	ipforwarding = IPFORWARDING;
188int	ipsendredirects = IPSENDREDIRECTS;
189int	ip_defttl = IPDEFTTL;
190int	ip_forwsrcrt = IPFORWSRCRT;
191int	ip_directedbcast = IPDIRECTEDBCAST;
192int	ip_allowsrcrt = IPALLOWSRCRT;
193int	ip_mtudisc = IPMTUDISC;
194int	ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
195#ifdef DIAGNOSTIC
196int	ipprintfs = 0;
197#endif
198
199int	ip_do_randomid = 0;
200
201/*
202 * XXX - Setting ip_checkinterface mostly implements the receive side of
203 * the Strong ES model described in RFC 1122, but since the routing table
204 * and transmit implementation do not implement the Strong ES model,
205 * setting this to 1 results in an odd hybrid.
206 *
207 * XXX - ip_checkinterface currently must be disabled if you use ipnat
208 * to translate the destination address to another local interface.
209 *
210 * XXX - ip_checkinterface must be disabled if you add IP aliases
211 * to the loopback interface instead of the interface where the
212 * packets for those addresses are received.
213 */
214static int		ip_checkinterface	__read_mostly = 0;
215
216struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
217
218pktqueue_t *		ip_pktq			__read_mostly;
219pfil_head_t *		inet_pfil_hook		__read_mostly;
220ipid_state_t *		ip_ids			__read_mostly;
221percpu_t *		ipstat_percpu		__read_mostly;
222
223static percpu_t		*ipforward_rt_percpu	__cacheline_aligned;
224
225uint16_t ip_id;
226
227#ifdef INET_CSUM_COUNTERS
228#include <sys/device.h>
229
230struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
231    NULL, "inet", "hwcsum bad");
232struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
233    NULL, "inet", "hwcsum ok");
234struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
235    NULL, "inet", "swcsum");
236
237#define	INET_CSUM_COUNTER_INCR(ev)	(ev)->ev_count++
238
239EVCNT_ATTACH_STATIC(ip_hwcsum_bad);
240EVCNT_ATTACH_STATIC(ip_hwcsum_ok);
241EVCNT_ATTACH_STATIC(ip_swcsum);
242
243#else
244
245#define	INET_CSUM_COUNTER_INCR(ev)	/* nothing */
246
247#endif /* INET_CSUM_COUNTERS */
248
249/*
250 * Used to save the IP options in case a protocol wants to respond
251 * to an incoming packet over the same route if the packet got here
252 * using IP source routing.  This allows connection establishment and
253 * maintenance when the remote end is on a network that is not known
254 * to us.
255 */
256struct ip_srcrt {
257	int		isr_nhops;		   /* number of hops */
258	struct in_addr	isr_dst;		   /* final destination */
259	char		isr_nop;		   /* one NOP to align */
260	char		isr_hdr[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN & OFFSET */
261	struct in_addr	isr_routes[MAX_IPOPTLEN/sizeof(struct in_addr)];
262};
263
264static int ip_drainwanted;
265
266static void save_rte(struct mbuf *, u_char *, struct in_addr);
267
268#ifdef MBUFTRACE
269struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx");
270struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx");
271#endif
272
273static void		ipintr(void *);
274static void		ip_input(struct mbuf *);
275static void		ip_forward(struct mbuf *, int, struct ifnet *);
276static bool		ip_dooptions(struct mbuf *);
277static struct in_ifaddr *ip_rtaddr(struct in_addr, struct psref *);
278static void		sysctl_net_inet_ip_setup(struct sysctllog **);
279
280static struct in_ifaddr	*ip_match_our_address(struct ifnet *, struct ip *,
281			    int *);
282static struct in_ifaddr	*ip_match_our_address_broadcast(struct ifnet *,
283			    struct ip *);
284
285#ifdef NET_MPSAFE
286#define	SOFTNET_LOCK()		mutex_enter(softnet_lock)
287#define	SOFTNET_UNLOCK()	mutex_exit(softnet_lock)
288#else
289#define	SOFTNET_LOCK()		KASSERT(mutex_owned(softnet_lock))
290#define	SOFTNET_UNLOCK()	KASSERT(mutex_owned(softnet_lock))
291#endif
292
293/*
294 * IP initialization: fill in IP protocol switch table.
295 * All protocols not implemented in kernel go to raw IP protocol handler.
296 */
297void
298ip_init(void)
299{
300	const struct protosw *pr;
301
302	in_init();
303	sysctl_net_inet_ip_setup(NULL);
304
305	pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
306	KASSERT(pr != NULL);
307
308	ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL);
309	KASSERT(ip_pktq != NULL);
310
311	for (u_int i = 0; i < IPPROTO_MAX; i++) {
312		ip_protox[i] = pr - inetsw;
313	}
314	for (pr = inetdomain.dom_protosw;
315	    pr < inetdomain.dom_protoswNPROTOSW; pr++)
316		if (pr->pr_domain->dom_family == PF_INET &&
317		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
318			ip_protox[pr->pr_protocol] = pr - inetsw;
319
320	ip_reass_init();
321
322	ip_ids = ip_id_init();
323	ip_id = time_uptime & 0xfffff;
324
325#ifdef GATEWAY
326	ipflow_init();
327#endif
328
329	/* Register our Packet Filter hook. */
330	inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET);
331	KASSERT(inet_pfil_hook != NULL);
332
333#ifdef MBUFTRACE
334	MOWNER_ATTACH(&ip_tx_mowner);
335	MOWNER_ATTACH(&ip_rx_mowner);
336#endif /* MBUFTRACE */
337
338	ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS);
339	ipforward_rt_percpu = percpu_alloc(sizeof(struct route));
340	ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout);
341}
342
343static struct in_ifaddr *
344ip_match_our_address(struct ifnet *ifp, struct ip *ip, int *downmatch)
345{
346	struct in_ifaddr *ia = NULL;
347	int checkif;
348
349	/*
350	 * Enable a consistency check between the destination address
351	 * and the arrival interface for a unicast packet (the RFC 1122
352	 * strong ES model) if IP forwarding is disabled and the packet
353	 * is not locally generated.
354	 *
355	 * XXX - Checking also should be disabled if the destination
356	 * address is ipnat'ed to a different interface.
357	 *
358	 * XXX - Checking is incompatible with IP aliases added
359	 * to the loopback interface instead of the interface where
360	 * the packets are received.
361	 *
362	 * XXX - We need to add a per ifaddr flag for this so that
363	 * we get finer grain control.
364	 */
365	checkif = ip_checkinterface && (ipforwarding == 0) &&
366	    (ifp->if_flags & IFF_LOOPBACK) == 0;
367
368	IN_ADDRHASH_READER_FOREACH(ia, ip->ip_dst.s_addr) {
369		if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
370			if (ia->ia4_flags & IN_IFF_NOTREADY)
371				continue;
372			if (checkif && ia->ia_ifp != ifp)
373				continue;
374			if ((ia->ia_ifp->if_flags & IFF_UP) != 0 &&
375			    (ia->ia4_flags & IN_IFF_DETACHED) == 0)
376				break;
377			else
378				(*downmatch)++;
379		}
380	}
381
382	return ia;
383}
384
385static struct in_ifaddr *
386ip_match_our_address_broadcast(struct ifnet *ifp, struct ip *ip)
387{
388	struct in_ifaddr *ia = NULL;
389	struct ifaddr *ifa;
390
391	IFADDR_READER_FOREACH(ifa, ifp) {
392		if (ifa->ifa_addr->sa_family != AF_INET)
393			continue;
394		ia = ifatoia(ifa);
395		if (ia->ia4_flags & (IN_IFF_NOTREADY | IN_IFF_DETACHED))
396			continue;
397		if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
398		    in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
399		    /*
400		     * Look for all-0's host part (old broadcast addr),
401		     * either for subnet or net.
402		     */
403		    ip->ip_dst.s_addr == ia->ia_subnet ||
404		    ip->ip_dst.s_addr == ia->ia_net)
405			goto matched;
406		/*
407		 * An interface with IP address zero accepts
408		 * all packets that arrive on that interface.
409		 */
410		if (in_nullhost(ia->ia_addr.sin_addr))
411			goto matched;
412	}
413	ia = NULL;
414
415matched:
416	return ia;
417}
418
419/*
420 * IP software interrupt routine.
421 */
422static void
423ipintr(void *arg __unused)
424{
425	struct mbuf *m;
426
427	KASSERT(cpu_softintr_p());
428
429#ifndef NET_MPSAFE
430	mutex_enter(softnet_lock);
431#endif
432	while ((m = pktq_dequeue(ip_pktq)) != NULL) {
433		ip_input(m);
434	}
435#ifndef NET_MPSAFE
436	mutex_exit(softnet_lock);
437#endif
438}
439
440/*
441 * IP input routine.  Checksum and byte swap header.  If fragmented
442 * try to reassemble.  Process options.  Pass to next level.
443 */
444static void
445ip_input(struct mbuf *m)
446{
447	struct ip *ip = NULL;
448	struct in_ifaddr *ia = NULL;
449	int hlen = 0, len;
450	int downmatch;
451	int srcrt = 0;
452	ifnet_t *ifp;
453	struct psref psref;
454	int s;
455
456	KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software "
457	    "interrupt handler; synchronization assumptions violated");
458
459	MCLAIM(m, &ip_rx_mowner);
460	KASSERT((m->m_flags & M_PKTHDR) != 0);
461
462	ifp = m_get_rcvif_psref(m, &psref);
463	if (__predict_false(ifp == NULL))
464		goto out;
465
466	/*
467	 * If no IP addresses have been set yet but the interfaces
468	 * are receiving, can't do anything with incoming packets yet.
469	 * Note: we pre-check without locks held.
470	 */
471	if (IN_ADDRLIST_READER_EMPTY())
472		goto out;
473	IP_STATINC(IP_STAT_TOTAL);
474
475	/*
476	 * If the IP header is not aligned, slurp it up into a new
477	 * mbuf with space for link headers, in the event we forward
478	 * it.  Otherwise, if it is aligned, make sure the entire
479	 * base IP header is in the first mbuf of the chain.
480	 */
481	if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
482		if ((m = m_copyup(m, sizeof(struct ip),
483				  (max_linkhdr + 3) & ~3)) == NULL) {
484			/* XXXJRT new stat, please */
485			IP_STATINC(IP_STAT_TOOSMALL);
486			goto out;
487		}
488	} else if (__predict_false(m->m_len < sizeof (struct ip))) {
489		if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
490			IP_STATINC(IP_STAT_TOOSMALL);
491			goto out;
492		}
493	}
494	ip = mtod(m, struct ip *);
495	if (ip->ip_v != IPVERSION) {
496		IP_STATINC(IP_STAT_BADVERS);
497		goto out;
498	}
499	hlen = ip->ip_hl << 2;
500	if (hlen < sizeof(struct ip)) {	/* minimum header length */
501		IP_STATINC(IP_STAT_BADHLEN);
502		goto out;
503	}
504	if (hlen > m->m_len) {
505		if ((m = m_pullup(m, hlen)) == NULL) {
506			IP_STATINC(IP_STAT_BADHLEN);
507			goto out;
508		}
509		ip = mtod(m, struct ip *);
510	}
511
512	/*
513	 * RFC1122: packets with a multicast source address are
514	 * not allowed.
515	 */
516	if (IN_MULTICAST(ip->ip_src.s_addr)) {
517		IP_STATINC(IP_STAT_BADADDR);
518		goto out;
519	}
520
521	/* 127/8 must not appear on wire - RFC1122 */
522	if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
523	    (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
524		if ((ifp->if_flags & IFF_LOOPBACK) == 0) {
525			IP_STATINC(IP_STAT_BADADDR);
526			goto out;
527		}
528	}
529
530	switch (m->m_pkthdr.csum_flags &
531		((ifp->if_csum_flags_rx & M_CSUM_IPv4) |
532		 M_CSUM_IPv4_BAD)) {
533	case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
534		INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
535		IP_STATINC(IP_STAT_BADSUM);
536		goto out;
537
538	case M_CSUM_IPv4:
539		/* Checksum was okay. */
540		INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
541		break;
542
543	default:
544		/*
545		 * Must compute it ourselves.  Maybe skip checksum on
546		 * loopback interfaces.
547		 */
548		if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) ||
549		    ip_do_loopback_cksum)) {
550			INET_CSUM_COUNTER_INCR(&ip_swcsum);
551			if (in_cksum(m, hlen) != 0) {
552				IP_STATINC(IP_STAT_BADSUM);
553				goto out;
554			}
555		}
556		break;
557	}
558
559	/* Retrieve the packet length. */
560	len = ntohs(ip->ip_len);
561
562	/*
563	 * Check for additional length bogosity
564	 */
565	if (len < hlen) {
566		IP_STATINC(IP_STAT_BADLEN);
567		goto out;
568	}
569
570	/*
571	 * Check that the amount of data in the buffers
572	 * is as at least much as the IP header would have us expect.
573	 * Trim mbufs if longer than we expect.
574	 * Drop packet if shorter than we expect.
575	 */
576	if (m->m_pkthdr.len < len) {
577		IP_STATINC(IP_STAT_TOOSHORT);
578		goto out;
579	}
580	if (m->m_pkthdr.len > len) {
581		if (m->m_len == m->m_pkthdr.len) {
582			m->m_len = len;
583			m->m_pkthdr.len = len;
584		} else
585			m_adj(m, len - m->m_pkthdr.len);
586	}
587
588	/*
589	 * Assume that we can create a fast-forward IP flow entry
590	 * based on this packet.
591	 */
592	m->m_flags |= M_CANFASTFWD;
593
594	/*
595	 * Run through list of hooks for input packets.  If there are any
596	 * filters which require that additional packets in the flow are
597	 * not fast-forwarded, they must clear the M_CANFASTFWD flag.
598	 * Note that filters must _never_ set this flag, as another filter
599	 * in the list may have previously cleared it.
600	 */
601#if defined(IPSEC)
602	if (!ipsec_used || !ipsec_indone(m))
603#else
604	if (1)
605#endif
606	{
607		struct in_addr odst = ip->ip_dst;
608		bool freed;
609
610		freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0;
611		if (freed || m == NULL) {
612			m = NULL;
613			goto out;
614		}
615		ip = mtod(m, struct ip *);
616		hlen = ip->ip_hl << 2;
617
618		/*
619		 * XXX The setting of "srcrt" here is to prevent ip_forward()
620		 * from generating ICMP redirects for packets that have
621		 * been redirected by a hook back out on to the same LAN that
622		 * they came from and is not an indication that the packet
623		 * is being inffluenced by source routing options.  This
624		 * allows things like
625		 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
626		 * where tlp0 is both on the 1.1.1.0/24 network and is the
627		 * default route for hosts on 1.1.1.0/24.  Of course this
628		 * also requires a "map tlp0 ..." to complete the story.
629		 * One might argue whether or not this kind of network config.
630		 * should be supported in this manner...
631		 */
632		srcrt = (odst.s_addr != ip->ip_dst.s_addr);
633	}
634
635#ifdef ALTQ
636	/* XXX Temporary until ALTQ is changed to use a pfil hook */
637	if (altq_input) {
638		SOFTNET_LOCK();
639		if ((*altq_input)(m, AF_INET) == 0) {
640			/* Packet dropped by traffic conditioner. */
641			SOFTNET_UNLOCK();
642			m = NULL;
643			goto out;
644		}
645		SOFTNET_UNLOCK();
646	}
647#endif
648
649	/*
650	 * Process options and, if not destined for us,
651	 * ship it on.  ip_dooptions returns 1 when an
652	 * error was detected (causing an icmp message
653	 * to be sent and the original packet to be freed).
654	 */
655	if (hlen > sizeof (struct ip) && ip_dooptions(m)) {
656		m = NULL;
657		goto out;
658	}
659
660	/*
661	 * Check our list of addresses, to see if the packet is for us.
662	 *
663	 * Traditional 4.4BSD did not consult IFF_UP at all.
664	 * The behavior here is to treat addresses on !IFF_UP interface
665	 * or IN_IFF_NOTREADY addresses as not mine.
666	 */
667	downmatch = 0;
668	s = pserialize_read_enter();
669	ia = ip_match_our_address(ifp, ip, &downmatch);
670	if (ia != NULL) {
671		pserialize_read_exit(s);
672		goto ours;
673	}
674
675	if (ifp->if_flags & IFF_BROADCAST) {
676		ia = ip_match_our_address_broadcast(ifp, ip);
677		if (ia != NULL) {
678			pserialize_read_exit(s);
679			goto ours;
680		}
681	}
682	pserialize_read_exit(s);
683
684	if (IN_MULTICAST(ip->ip_dst.s_addr)) {
685#ifdef MROUTING
686		extern struct socket *ip_mrouter;
687
688		if (ip_mrouter) {
689			/*
690			 * If we are acting as a multicast router, all
691			 * incoming multicast packets are passed to the
692			 * kernel-level multicast forwarding function.
693			 * The packet is returned (relatively) intact; if
694			 * ip_mforward() returns a non-zero value, the packet
695			 * must be discarded, else it may be accepted below.
696			 *
697			 * (The IP ident field is put in the same byte order
698			 * as expected when ip_mforward() is called from
699			 * ip_output().)
700			 */
701			SOFTNET_LOCK();
702			if (ip_mforward(m, ifp) != 0) {
703				SOFTNET_UNLOCK();
704				IP_STATINC(IP_STAT_CANTFORWARD);
705				goto out;
706			}
707			SOFTNET_UNLOCK();
708
709			/*
710			 * The process-level routing demon needs to receive
711			 * all multicast IGMP packets, whether or not this
712			 * host belongs to their destination groups.
713			 */
714			if (ip->ip_p == IPPROTO_IGMP) {
715				goto ours;
716			}
717			IP_STATINC(IP_STAT_CANTFORWARD);
718		}
719#endif
720		/*
721		 * See if we belong to the destination multicast group on the
722		 * arrival interface.
723		 */
724		if (!in_multi_group(ip->ip_dst, ifp, 0)) {
725			IP_STATINC(IP_STAT_CANTFORWARD);
726			goto out;
727		}
728		goto ours;
729	}
730	if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
731	    in_nullhost(ip->ip_dst))
732		goto ours;
733
734	/*
735	 * Not for us; forward if possible and desirable.
736	 */
737	if (ipforwarding == 0) {
738		m_put_rcvif_psref(ifp, &psref);
739		IP_STATINC(IP_STAT_CANTFORWARD);
740		m_freem(m);
741	} else {
742		/*
743		 * If ip_dst matched any of my address on !IFF_UP interface,
744		 * and there's no IFF_UP interface that matches ip_dst,
745		 * send icmp unreach.  Forwarding it will result in in-kernel
746		 * forwarding loop till TTL goes to 0.
747		 */
748		if (downmatch) {
749			m_put_rcvif_psref(ifp, &psref);
750			icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
751			IP_STATINC(IP_STAT_CANTFORWARD);
752			return;
753		}
754#ifdef IPSEC
755		/* Perform IPsec, if any. */
756		if (ipsec_used) {
757			SOFTNET_LOCK();
758			if (ipsec4_input(m, IP_FORWARDING |
759			    (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) {
760				SOFTNET_UNLOCK();
761				goto out;
762			}
763			SOFTNET_UNLOCK();
764		}
765#endif
766		ip_forward(m, srcrt, ifp);
767		m_put_rcvif_psref(ifp, &psref);
768	}
769	return;
770
771ours:
772	m_put_rcvif_psref(ifp, &psref);
773	ifp = NULL;
774
775	/*
776	 * If offset or IP_MF are set, must reassemble.
777	 */
778	if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
779		/*
780		 * Pass to IP reassembly mechanism.
781		 */
782		if (ip_reass_packet(&m, ip) != 0) {
783			/* Failed; invalid fragment(s) or packet. */
784			goto out;
785		}
786		if (m == NULL) {
787			/* More fragments should come; silently return. */
788			goto out;
789		}
790		/*
791		 * Reassembly is done, we have the final packet.
792		 * Updated cached data in local variable(s).
793		 */
794		ip = mtod(m, struct ip *);
795		hlen = ip->ip_hl << 2;
796	}
797
798#ifdef IPSEC
799	/*
800	 * Enforce IPsec policy checking if we are seeing last header.
801	 * Note that we do not visit this with protocols with PCB layer
802	 * code - like UDP/TCP/raw IP.
803	 */
804	if (ipsec_used &&
805	    (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
806		SOFTNET_LOCK();
807		if (ipsec4_input(m, 0) != 0) {
808			SOFTNET_UNLOCK();
809			goto out;
810		}
811		SOFTNET_UNLOCK();
812	}
813#endif
814
815	/*
816	 * Switch out to protocol's input routine.
817	 */
818#if IFA_STATS
819	if (ia && ip) {
820		struct in_ifaddr *_ia;
821		/*
822		 * Keep a reference from ip_match_our_address with psref
823		 * is expensive, so explore ia here again.
824		 */
825		s = pserialize_read_enter();
826		_ia = in_get_ia(ip->ip_dst);
827		_ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
828		pserialize_read_exit(s);
829	}
830#endif
831	IP_STATINC(IP_STAT_DELIVERED);
832
833	const int off = hlen, nh = ip->ip_p;
834
835	SOFTNET_LOCK();
836	(*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
837	SOFTNET_UNLOCK();
838	return;
839
840out:
841	m_put_rcvif_psref(ifp, &psref);
842	if (m != NULL)
843		m_freem(m);
844}
845
846/*
847 * IP timer processing.
848 */
849void
850ip_slowtimo(void)
851{
852
853#ifndef NET_MPSAFE
854	mutex_enter(softnet_lock);
855	KERNEL_LOCK(1, NULL);
856#endif
857
858	ip_reass_slowtimo();
859
860#ifndef NET_MPSAFE
861	KERNEL_UNLOCK_ONE(NULL);
862	mutex_exit(softnet_lock);
863#endif
864}
865
866/*
867 * IP drain processing.
868 */
869void
870ip_drain(void)
871{
872
873	KERNEL_LOCK(1, NULL);
874	ip_reass_drain();
875	KERNEL_UNLOCK_ONE(NULL);
876}
877
878/*
879 * ip_dooptions: perform option processing on a datagram, possibly discarding
880 * it if bad options are encountered, or forwarding it if source-routed.
881 *
882 * => Returns true if packet has been forwarded/freed.
883 * => Returns false if the packet should be processed further.
884 */
885static bool
886ip_dooptions(struct mbuf *m)
887{
888	struct ip *ip = mtod(m, struct ip *);
889	u_char *cp, *cp0;
890	struct ip_timestamp *ipt;
891	struct in_ifaddr *ia;
892	int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
893	struct in_addr dst;
894	n_time ntime;
895	struct ifaddr *ifa = NULL;
896	int s;
897
898	dst = ip->ip_dst;
899	cp = (u_char *)(ip + 1);
900	cnt = (ip->ip_hl << 2) - sizeof (struct ip);
901	for (; cnt > 0; cnt -= optlen, cp += optlen) {
902		opt = cp[IPOPT_OPTVAL];
903		if (opt == IPOPT_EOL)
904			break;
905		if (opt == IPOPT_NOP)
906			optlen = 1;
907		else {
908			if (cnt < IPOPT_OLEN + sizeof(*cp)) {
909				code = &cp[IPOPT_OLEN] - (u_char *)ip;
910				goto bad;
911			}
912			optlen = cp[IPOPT_OLEN];
913			if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
914				code = &cp[IPOPT_OLEN] - (u_char *)ip;
915				goto bad;
916			}
917		}
918		switch (opt) {
919
920		default:
921			break;
922
923		/*
924		 * Source routing with record.
925		 * Find interface with current destination address.
926		 * If none on this machine then drop if strictly routed,
927		 * or do nothing if loosely routed.
928		 * Record interface address and bring up next address
929		 * component.  If strictly routed make sure next
930		 * address is on directly accessible net.
931		 */
932		case IPOPT_LSRR:
933		case IPOPT_SSRR: {
934			struct psref psref;
935			struct sockaddr_in ipaddr = {
936			    .sin_len = sizeof(ipaddr),
937			    .sin_family = AF_INET,
938			};
939
940			if (ip_allowsrcrt == 0) {
941				type = ICMP_UNREACH;
942				code = ICMP_UNREACH_NET_PROHIB;
943				goto bad;
944			}
945			if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
946				code = &cp[IPOPT_OLEN] - (u_char *)ip;
947				goto bad;
948			}
949			if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
950				code = &cp[IPOPT_OFFSET] - (u_char *)ip;
951				goto bad;
952			}
953			ipaddr.sin_addr = ip->ip_dst;
954
955			s = pserialize_read_enter();
956			ifa = ifa_ifwithaddr(sintosa(&ipaddr));
957			if (ifa == NULL) {
958				pserialize_read_exit(s);
959				if (opt == IPOPT_SSRR) {
960					type = ICMP_UNREACH;
961					code = ICMP_UNREACH_SRCFAIL;
962					goto bad;
963				}
964				/*
965				 * Loose routing, and not at next destination
966				 * yet; nothing to do except forward.
967				 */
968				break;
969			}
970			pserialize_read_exit(s);
971
972			off--;			/* 0 origin */
973			if ((off + sizeof(struct in_addr)) > optlen) {
974				/*
975				 * End of source route.  Should be for us.
976				 */
977				save_rte(m, cp, ip->ip_src);
978				break;
979			}
980			/*
981			 * locate outgoing interface
982			 */
983			memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off),
984			    sizeof(ipaddr.sin_addr));
985			if (opt == IPOPT_SSRR) {
986				ifa = ifa_ifwithladdr_psref(sintosa(&ipaddr),
987				    &psref);
988				if (ifa != NULL)
989					ia = ifatoia(ifa);
990				else
991					ia = NULL;
992			} else {
993				ia = ip_rtaddr(ipaddr.sin_addr, &psref);
994			}
995			if (ia == NULL) {
996				type = ICMP_UNREACH;
997				code = ICMP_UNREACH_SRCFAIL;
998				goto bad;
999			}
1000			ip->ip_dst = ipaddr.sin_addr;
1001			bcopy((void *)&ia->ia_addr.sin_addr,
1002			    (void *)(cp + off), sizeof(struct in_addr));
1003			ia4_release(ia, &psref);
1004			cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1005			/*
1006			 * Let ip_intr's mcast routing check handle mcast pkts
1007			 */
1008			forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1009			break;
1010		    }
1011
1012		case IPOPT_RR: {
1013			struct psref psref;
1014			struct sockaddr_in ipaddr = {
1015			    .sin_len = sizeof(ipaddr),
1016			    .sin_family = AF_INET,
1017			};
1018
1019			if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1020				code = &cp[IPOPT_OLEN] - (u_char *)ip;
1021				goto bad;
1022			}
1023			if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1024				code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1025				goto bad;
1026			}
1027			/*
1028			 * If no space remains, ignore.
1029			 */
1030			off--;			/* 0 origin */
1031			if ((off + sizeof(struct in_addr)) > optlen)
1032				break;
1033			memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst),
1034			    sizeof(ipaddr.sin_addr));
1035			/*
1036			 * locate outgoing interface; if we're the destination,
1037			 * use the incoming interface (should be same).
1038			 */
1039			ifa = ifa_ifwithaddr_psref(sintosa(&ipaddr), &psref);
1040			if (ifa == NULL) {
1041				ia = ip_rtaddr(ipaddr.sin_addr, &psref);
1042				if (ia == NULL) {
1043					type = ICMP_UNREACH;
1044					code = ICMP_UNREACH_HOST;
1045					goto bad;
1046				}
1047			} else {
1048				ia = ifatoia(ifa);
1049			}
1050			bcopy((void *)&ia->ia_addr.sin_addr,
1051			    (void *)(cp + off), sizeof(struct in_addr));
1052			ia4_release(ia, &psref);
1053			cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1054			break;
1055		    }
1056
1057		case IPOPT_TS:
1058			code = cp - (u_char *)ip;
1059			ipt = (struct ip_timestamp *)cp;
1060			if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1061				code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1062				goto bad;
1063			}
1064			if (ipt->ipt_ptr < 5) {
1065				code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1066				goto bad;
1067			}
1068			if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1069				if (++ipt->ipt_oflw == 0) {
1070					code = (u_char *)&ipt->ipt_ptr -
1071					    (u_char *)ip;
1072					goto bad;
1073				}
1074				break;
1075			}
1076			cp0 = (cp + ipt->ipt_ptr - 1);
1077			switch (ipt->ipt_flg) {
1078
1079			case IPOPT_TS_TSONLY:
1080				break;
1081
1082			case IPOPT_TS_TSANDADDR: {
1083				struct ifnet *rcvif;
1084				int _s, _ss;
1085				struct sockaddr_in ipaddr = {
1086				    .sin_len = sizeof(ipaddr),
1087				    .sin_family = AF_INET,
1088				};
1089
1090				if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1091				    sizeof(struct in_addr) > ipt->ipt_len) {
1092					code = (u_char *)&ipt->ipt_ptr -
1093					    (u_char *)ip;
1094					goto bad;
1095				}
1096				ipaddr.sin_addr = dst;
1097				_ss = pserialize_read_enter();
1098				rcvif = m_get_rcvif(m, &_s);
1099				if (__predict_true(rcvif != NULL)) {
1100					ifa = ifaof_ifpforaddr(sintosa(&ipaddr),
1101					    rcvif);
1102				}
1103				m_put_rcvif(rcvif, &_s);
1104				if (ifa == NULL) {
1105					pserialize_read_exit(_ss);
1106					break;
1107				}
1108				ia = ifatoia(ifa);
1109				bcopy(&ia->ia_addr.sin_addr,
1110				    cp0, sizeof(struct in_addr));
1111				pserialize_read_exit(_ss);
1112				ipt->ipt_ptr += sizeof(struct in_addr);
1113				break;
1114			}
1115
1116			case IPOPT_TS_PRESPEC: {
1117				struct sockaddr_in ipaddr = {
1118				    .sin_len = sizeof(ipaddr),
1119				    .sin_family = AF_INET,
1120				};
1121
1122				if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1123				    sizeof(struct in_addr) > ipt->ipt_len) {
1124					code = (u_char *)&ipt->ipt_ptr -
1125					    (u_char *)ip;
1126					goto bad;
1127				}
1128				memcpy(&ipaddr.sin_addr, cp0,
1129				    sizeof(struct in_addr));
1130				s = pserialize_read_enter();
1131				ifa = ifa_ifwithaddr(sintosa(&ipaddr));
1132				if (ifa == NULL) {
1133					pserialize_read_exit(s);
1134					continue;
1135				}
1136				pserialize_read_exit(s);
1137				ipt->ipt_ptr += sizeof(struct in_addr);
1138				break;
1139			    }
1140
1141			default:
1142				/* XXX can't take &ipt->ipt_flg */
1143				code = (u_char *)&ipt->ipt_ptr -
1144				    (u_char *)ip + 1;
1145				goto bad;
1146			}
1147			ntime = iptime();
1148			cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1149			memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1150			    sizeof(n_time));
1151			ipt->ipt_ptr += sizeof(n_time);
1152		}
1153	}
1154	if (forward) {
1155		struct ifnet *rcvif;
1156		struct psref _psref;
1157
1158		if (ip_forwsrcrt == 0) {
1159			type = ICMP_UNREACH;
1160			code = ICMP_UNREACH_SRCFAIL;
1161			goto bad;
1162		}
1163
1164		rcvif = m_get_rcvif_psref(m, &_psref);
1165		if (__predict_false(rcvif == NULL)) {
1166			type = ICMP_UNREACH;
1167			code = ICMP_UNREACH_HOST;
1168			goto bad;
1169		}
1170		ip_forward(m, 1, rcvif);
1171		m_put_rcvif_psref(rcvif, &_psref);
1172		return true;
1173	}
1174	return false;
1175bad:
1176	icmp_error(m, type, code, 0, 0);
1177	IP_STATINC(IP_STAT_BADOPTIONS);
1178	return true;
1179}
1180
1181/*
1182 * ip_rtaddr: given address of next destination (final or next hop),
1183 * return internet address info of interface to be used to get there.
1184 */
1185static struct in_ifaddr *
1186ip_rtaddr(struct in_addr dst, struct psref *psref)
1187{
1188	struct rtentry *rt;
1189	union {
1190		struct sockaddr		dst;
1191		struct sockaddr_in	dst4;
1192	} u;
1193	struct route *ro;
1194
1195	sockaddr_in_init(&u.dst4, &dst, 0);
1196
1197	ro = percpu_getref(ipforward_rt_percpu);
1198	rt = rtcache_lookup(ro, &u.dst);
1199	if (rt == NULL) {
1200		percpu_putref(ipforward_rt_percpu);
1201		return NULL;
1202	}
1203
1204	ia4_acquire(ifatoia(rt->rt_ifa), psref);
1205	rtcache_unref(rt, ro);
1206	percpu_putref(ipforward_rt_percpu);
1207
1208	return ifatoia(rt->rt_ifa);
1209}
1210
1211/*
1212 * save_rte: save incoming source route for use in replies, to be picked
1213 * up later by ip_srcroute if the receiver is interested.
1214 */
1215static void
1216save_rte(struct mbuf *m, u_char *option, struct in_addr dst)
1217{
1218	struct ip_srcrt *isr;
1219	struct m_tag *mtag;
1220	unsigned olen;
1221
1222	olen = option[IPOPT_OLEN];
1223	if (olen > sizeof(isr->isr_hdr) + sizeof(isr->isr_routes))
1224		return;
1225
1226	mtag = m_tag_get(PACKET_TAG_SRCROUTE, sizeof(*isr), M_NOWAIT);
1227	if (mtag == NULL)
1228		return;
1229	isr = (struct ip_srcrt *)(mtag + 1);
1230
1231	memcpy(isr->isr_hdr, option, olen);
1232	isr->isr_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1233	isr->isr_dst = dst;
1234	m_tag_prepend(m, mtag);
1235}
1236
1237/*
1238 * Retrieve incoming source route for use in replies,
1239 * in the same form used by setsockopt.
1240 * The first hop is placed before the options, will be removed later.
1241 */
1242struct mbuf *
1243ip_srcroute(struct mbuf *m0)
1244{
1245	struct in_addr *p, *q;
1246	struct mbuf *m;
1247	struct ip_srcrt *isr;
1248	struct m_tag *mtag;
1249
1250	mtag = m_tag_find(m0, PACKET_TAG_SRCROUTE, NULL);
1251	if (mtag == NULL)
1252		return NULL;
1253	isr = (struct ip_srcrt *)(mtag + 1);
1254
1255	if (isr->isr_nhops == 0)
1256		return NULL;
1257
1258	m = m_get(M_DONTWAIT, MT_SOOPTS);
1259	if (m == NULL)
1260		return NULL;
1261
1262	MCLAIM(m, &inetdomain.dom_mowner);
1263#define OPTSIZ	(sizeof(isr->isr_nop) + sizeof(isr->isr_hdr))
1264
1265	/* length is (nhops+1)*sizeof(addr) + sizeof(nop + header) */
1266	m->m_len = (isr->isr_nhops + 1) * sizeof(struct in_addr) + OPTSIZ;
1267
1268	/*
1269	 * First save first hop for return route
1270	 */
1271	p = &(isr->isr_routes[isr->isr_nhops - 1]);
1272	*(mtod(m, struct in_addr *)) = *p--;
1273
1274	/*
1275	 * Copy option fields and padding (nop) to mbuf.
1276	 */
1277	isr->isr_nop = IPOPT_NOP;
1278	isr->isr_hdr[IPOPT_OFFSET] = IPOPT_MINOFF;
1279	memmove(mtod(m, char *) + sizeof(struct in_addr), &isr->isr_nop,
1280	    OPTSIZ);
1281	q = (struct in_addr *)(mtod(m, char *) +
1282	    sizeof(struct in_addr) + OPTSIZ);
1283#undef OPTSIZ
1284	/*
1285	 * Record return path as an IP source route,
1286	 * reversing the path (pointers are now aligned).
1287	 */
1288	while (p >= isr->isr_routes) {
1289		*q++ = *p--;
1290	}
1291	/*
1292	 * Last hop goes to final destination.
1293	 */
1294	*q = isr->isr_dst;
1295	m_tag_delete(m0, mtag);
1296	return m;
1297}
1298
1299const int inetctlerrmap[PRC_NCMDS] = {
1300	[PRC_MSGSIZE] = EMSGSIZE,
1301	[PRC_HOSTDEAD] = EHOSTDOWN,
1302	[PRC_HOSTUNREACH] = EHOSTUNREACH,
1303	[PRC_UNREACH_NET] = EHOSTUNREACH,
1304	[PRC_UNREACH_HOST] = EHOSTUNREACH,
1305	[PRC_UNREACH_PROTOCOL] = ECONNREFUSED,
1306	[PRC_UNREACH_PORT] = ECONNREFUSED,
1307	[PRC_UNREACH_SRCFAIL] = EHOSTUNREACH,
1308	[PRC_PARAMPROB] = ENOPROTOOPT,
1309};
1310
1311void
1312ip_fasttimo(void)
1313{
1314	if (ip_drainwanted) {
1315		ip_drain();
1316		ip_drainwanted = 0;
1317	}
1318}
1319
1320void
1321ip_drainstub(void)
1322{
1323	ip_drainwanted = 1;
1324}
1325
1326/*
1327 * Forward a packet.  If some error occurs return the sender
1328 * an icmp packet.  Note we can't always generate a meaningful
1329 * icmp message because icmp doesn't have a large enough repertoire
1330 * of codes and types.
1331 *
1332 * If not forwarding, just drop the packet.  This could be confusing
1333 * if ipforwarding was zero but some routing protocol was advancing
1334 * us as a gateway to somewhere.  However, we must let the routing
1335 * protocol deal with that.
1336 *
1337 * The srcrt parameter indicates whether the packet is being forwarded
1338 * via a source route.
1339 */
1340static void
1341ip_forward(struct mbuf *m, int srcrt, struct ifnet *rcvif)
1342{
1343	struct ip *ip = mtod(m, struct ip *);
1344	struct rtentry *rt;
1345	int error, type = 0, code = 0, destmtu = 0;
1346	struct mbuf *mcopy;
1347	n_long dest;
1348	union {
1349		struct sockaddr		dst;
1350		struct sockaddr_in	dst4;
1351	} u;
1352	uint64_t *ips;
1353	struct route *ro;
1354
1355	KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software "
1356	    "interrupt handler; synchronization assumptions violated");
1357
1358	/*
1359	 * We are now in the output path.
1360	 */
1361	MCLAIM(m, &ip_tx_mowner);
1362
1363	/*
1364	 * Clear any in-bound checksum flags for this packet.
1365	 */
1366	m->m_pkthdr.csum_flags = 0;
1367
1368	dest = 0;
1369	if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1370		IP_STATINC(IP_STAT_CANTFORWARD);
1371		m_freem(m);
1372		return;
1373	}
1374
1375	if (ip->ip_ttl <= IPTTLDEC) {
1376		icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1377		return;
1378	}
1379
1380	sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1381
1382	ro = percpu_getref(ipforward_rt_percpu);
1383	rt = rtcache_lookup(ro, &u.dst);
1384	if (rt == NULL) {
1385		percpu_putref(ipforward_rt_percpu);
1386		icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1387		return;
1388	}
1389
1390	/*
1391	 * Save at most 68 bytes of the packet in case
1392	 * we need to generate an ICMP message to the src.
1393	 * Pullup to avoid sharing mbuf cluster between m and mcopy.
1394	 */
1395	mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1396	if (mcopy)
1397		mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1398
1399	ip->ip_ttl -= IPTTLDEC;
1400
1401	/*
1402	 * If forwarding packet using same interface that it came in on,
1403	 * perhaps should send a redirect to sender to shortcut a hop.
1404	 * Only send redirect if source is sending directly to us,
1405	 * and if packet was not source routed (or has any options).
1406	 * Also, don't send redirect if forwarding using a default route
1407	 * or a route modified by a redirect.
1408	 */
1409	if (rt->rt_ifp == rcvif &&
1410	    (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1411	    !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1412	    ipsendredirects && !srcrt) {
1413		if (rt->rt_ifa &&
1414		    (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1415		    ifatoia(rt->rt_ifa)->ia_subnet) {
1416			if (rt->rt_flags & RTF_GATEWAY)
1417				dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1418			else
1419				dest = ip->ip_dst.s_addr;
1420			/*
1421			 * Router requirements says to only send host
1422			 * redirects.
1423			 */
1424			type = ICMP_REDIRECT;
1425			code = ICMP_REDIRECT_HOST;
1426		}
1427	}
1428	rtcache_unref(rt, ro);
1429
1430	error = ip_output(m, NULL, ro,
1431	    (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1432	    NULL, NULL);
1433
1434	if (error) {
1435		IP_STATINC(IP_STAT_CANTFORWARD);
1436		goto error;
1437	}
1438
1439	ips = IP_STAT_GETREF();
1440	ips[IP_STAT_FORWARD]++;
1441
1442	if (type) {
1443		ips[IP_STAT_REDIRECTSENT]++;
1444		IP_STAT_PUTREF();
1445		goto redirect;
1446	}
1447
1448	IP_STAT_PUTREF();
1449	if (mcopy) {
1450#ifdef GATEWAY
1451		if (mcopy->m_flags & M_CANFASTFWD)
1452			ipflow_create(ro, mcopy);
1453#endif
1454		m_freem(mcopy);
1455	}
1456
1457	percpu_putref(ipforward_rt_percpu);
1458	return;
1459
1460redirect:
1461error:
1462	if (mcopy == NULL) {
1463		percpu_putref(ipforward_rt_percpu);
1464		return;
1465	}
1466
1467	switch (error) {
1468
1469	case 0:				/* forwarded, but need redirect */
1470		/* type, code set above */
1471		break;
1472
1473	case ENETUNREACH:		/* shouldn't happen, checked above */
1474	case EHOSTUNREACH:
1475	case ENETDOWN:
1476	case EHOSTDOWN:
1477	default:
1478		type = ICMP_UNREACH;
1479		code = ICMP_UNREACH_HOST;
1480		break;
1481
1482	case EMSGSIZE:
1483		type = ICMP_UNREACH;
1484		code = ICMP_UNREACH_NEEDFRAG;
1485
1486		if ((rt = rtcache_validate(ro)) != NULL) {
1487			destmtu = rt->rt_ifp->if_mtu;
1488			rtcache_unref(rt, ro);
1489		}
1490#ifdef IPSEC
1491		if (ipsec_used)
1492			(void)ipsec4_forward(mcopy, &destmtu);
1493#endif
1494		IP_STATINC(IP_STAT_CANTFRAG);
1495		break;
1496
1497	case ENOBUFS:
1498		/*
1499		 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812,
1500		 * Requirements for IP Version 4 Routers.  Source quench can
1501		 * big problem under DoS attacks or if the underlying
1502		 * interface is rate-limited.
1503		 */
1504		if (mcopy)
1505			m_freem(mcopy);
1506		percpu_putref(ipforward_rt_percpu);
1507		return;
1508	}
1509	icmp_error(mcopy, type, code, dest, destmtu);
1510	percpu_putref(ipforward_rt_percpu);
1511}
1512
1513void
1514ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
1515    struct mbuf *m)
1516{
1517	struct socket *so = inp->inp_socket;
1518	int inpflags = inp->inp_flags;
1519
1520	if (SOOPT_TIMESTAMP(so->so_options))
1521		mp = sbsavetimestamp(so->so_options, m, mp);
1522
1523	if (inpflags & INP_RECVDSTADDR) {
1524		*mp = sbcreatecontrol(&ip->ip_dst,
1525		    sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1526		if (*mp)
1527			mp = &(*mp)->m_next;
1528	}
1529
1530	if (inpflags & INP_RECVTTL) {
1531		*mp = sbcreatecontrol(&ip->ip_ttl,
1532		    sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP);
1533		if (*mp)
1534			mp = &(*mp)->m_next;
1535	}
1536
1537	struct psref psref;
1538	ifnet_t *ifp = m_get_rcvif_psref(m, &psref);
1539	if (__predict_false(ifp == NULL)) {
1540#ifdef DIAGNOSTIC
1541		printf("%s: missing receive interface\n", __func__);
1542#endif
1543		return; /* XXX should report error? */
1544	}
1545
1546	if (inpflags & INP_RECVPKTINFO) {
1547		struct in_pktinfo ipi;
1548		ipi.ipi_addr = ip->ip_src;
1549		ipi.ipi_ifindex = ifp->if_index;
1550		*mp = sbcreatecontrol(&ipi,
1551		    sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP);
1552		if (*mp)
1553			mp = &(*mp)->m_next;
1554	}
1555	if (inpflags & INP_PKTINFO) {
1556		struct in_pktinfo ipi;
1557		ipi.ipi_addr = ip->ip_dst;
1558		ipi.ipi_ifindex = ifp->if_index;
1559		*mp = sbcreatecontrol(&ipi,
1560		    sizeof(ipi), IP_PKTINFO, IPPROTO_IP);
1561		if (*mp)
1562			mp = &(*mp)->m_next;
1563	}
1564	if (inpflags & INP_RECVIF) {
1565		struct sockaddr_dl sdl;
1566
1567		sockaddr_dl_init(&sdl, sizeof(sdl), ifp->if_index, 0, NULL, 0,
1568		    NULL, 0);
1569		*mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1570		if (*mp)
1571			mp = &(*mp)->m_next;
1572	}
1573	m_put_rcvif_psref(ifp, &psref);
1574}
1575
1576/*
1577 * sysctl helper routine for net.inet.ip.forwsrcrt.
1578 */
1579static int
1580sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
1581{
1582	int error, tmp;
1583	struct sysctlnode node;
1584
1585	node = *rnode;
1586	tmp = ip_forwsrcrt;
1587	node.sysctl_data = &tmp;
1588	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1589	if (error || newp == NULL)
1590		return (error);
1591
1592	error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
1593	    0, NULL, NULL, NULL);
1594	if (error)
1595		return (error);
1596
1597	ip_forwsrcrt = tmp;
1598
1599	return (0);
1600}
1601
1602/*
1603 * sysctl helper routine for net.inet.ip.mtudisctimeout.  checks the
1604 * range of the new value and tweaks timers if it changes.
1605 */
1606static int
1607sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1608{
1609	int error, tmp;
1610	struct sysctlnode node;
1611
1612	icmp_mtudisc_lock();
1613
1614	node = *rnode;
1615	tmp = ip_mtudisc_timeout;
1616	node.sysctl_data = &tmp;
1617	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1618	if (error || newp == NULL)
1619		goto out;
1620	if (tmp < 0) {
1621		error = EINVAL;
1622		goto out;
1623	}
1624
1625	ip_mtudisc_timeout = tmp;
1626	rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
1627	error = 0;
1628out:
1629	icmp_mtudisc_unlock();
1630	return error;
1631}
1632
1633static int
1634sysctl_net_inet_ip_stats(SYSCTLFN_ARGS)
1635{
1636
1637	return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS));
1638}
1639
1640static void
1641sysctl_net_inet_ip_setup(struct sysctllog **clog)
1642{
1643	sysctl_createv(clog, 0, NULL, NULL,
1644		       CTLFLAG_PERMANENT,
1645		       CTLTYPE_NODE, "inet",
1646		       SYSCTL_DESCR("PF_INET related settings"),
1647		       NULL, 0, NULL, 0,
1648		       CTL_NET, PF_INET, CTL_EOL);
1649	sysctl_createv(clog, 0, NULL, NULL,
1650		       CTLFLAG_PERMANENT,
1651		       CTLTYPE_NODE, "ip",
1652		       SYSCTL_DESCR("IPv4 related settings"),
1653		       NULL, 0, NULL, 0,
1654		       CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1655
1656	sysctl_createv(clog, 0, NULL, NULL,
1657		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1658		       CTLTYPE_INT, "forwarding",
1659		       SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1660		       NULL, 0, &ipforwarding, 0,
1661		       CTL_NET, PF_INET, IPPROTO_IP,
1662		       IPCTL_FORWARDING, CTL_EOL);
1663	sysctl_createv(clog, 0, NULL, NULL,
1664		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1665		       CTLTYPE_INT, "redirect",
1666		       SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1667		       NULL, 0, &ipsendredirects, 0,
1668		       CTL_NET, PF_INET, IPPROTO_IP,
1669		       IPCTL_SENDREDIRECTS, CTL_EOL);
1670	sysctl_createv(clog, 0, NULL, NULL,
1671		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1672		       CTLTYPE_INT, "ttl",
1673		       SYSCTL_DESCR("Default TTL for an INET datagram"),
1674		       NULL, 0, &ip_defttl, 0,
1675		       CTL_NET, PF_INET, IPPROTO_IP,
1676		       IPCTL_DEFTTL, CTL_EOL);
1677#ifdef IPCTL_DEFMTU
1678	sysctl_createv(clog, 0, NULL, NULL,
1679		       CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1680		       CTLTYPE_INT, "mtu",
1681		       SYSCTL_DESCR("Default MTA for an INET route"),
1682		       NULL, 0, &ip_mtu, 0,
1683		       CTL_NET, PF_INET, IPPROTO_IP,
1684		       IPCTL_DEFMTU, CTL_EOL);
1685#endif /* IPCTL_DEFMTU */
1686	sysctl_createv(clog, 0, NULL, NULL,
1687		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1688		       CTLTYPE_INT, "forwsrcrt",
1689		       SYSCTL_DESCR("Enable forwarding of source-routed "
1690				    "datagrams"),
1691		       sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1692		       CTL_NET, PF_INET, IPPROTO_IP,
1693		       IPCTL_FORWSRCRT, CTL_EOL);
1694	sysctl_createv(clog, 0, NULL, NULL,
1695		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1696		       CTLTYPE_INT, "directed-broadcast",
1697		       SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1698		       NULL, 0, &ip_directedbcast, 0,
1699		       CTL_NET, PF_INET, IPPROTO_IP,
1700		       IPCTL_DIRECTEDBCAST, CTL_EOL);
1701	sysctl_createv(clog, 0, NULL, NULL,
1702		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1703		       CTLTYPE_INT, "allowsrcrt",
1704		       SYSCTL_DESCR("Accept source-routed datagrams"),
1705		       NULL, 0, &ip_allowsrcrt, 0,
1706		       CTL_NET, PF_INET, IPPROTO_IP,
1707		       IPCTL_ALLOWSRCRT, CTL_EOL);
1708
1709	sysctl_createv(clog, 0, NULL, NULL,
1710		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1711		       CTLTYPE_INT, "mtudisc",
1712		       SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1713		       NULL, 0, &ip_mtudisc, 0,
1714		       CTL_NET, PF_INET, IPPROTO_IP,
1715		       IPCTL_MTUDISC, CTL_EOL);
1716	sysctl_createv(clog, 0, NULL, NULL,
1717		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1718		       CTLTYPE_INT, "anonportmin",
1719		       SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1720		       sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
1721		       CTL_NET, PF_INET, IPPROTO_IP,
1722		       IPCTL_ANONPORTMIN, CTL_EOL);
1723	sysctl_createv(clog, 0, NULL, NULL,
1724		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1725		       CTLTYPE_INT, "anonportmax",
1726		       SYSCTL_DESCR("Highest ephemeral port number to assign"),
1727		       sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
1728		       CTL_NET, PF_INET, IPPROTO_IP,
1729		       IPCTL_ANONPORTMAX, CTL_EOL);
1730	sysctl_createv(clog, 0, NULL, NULL,
1731		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1732		       CTLTYPE_INT, "mtudisctimeout",
1733		       SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1734		       sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0,
1735		       CTL_NET, PF_INET, IPPROTO_IP,
1736		       IPCTL_MTUDISCTIMEOUT, CTL_EOL);
1737#ifndef IPNOPRIVPORTS
1738	sysctl_createv(clog, 0, NULL, NULL,
1739		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1740		       CTLTYPE_INT, "lowportmin",
1741		       SYSCTL_DESCR("Lowest privileged ephemeral port number "
1742				    "to assign"),
1743		       sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
1744		       CTL_NET, PF_INET, IPPROTO_IP,
1745		       IPCTL_LOWPORTMIN, CTL_EOL);
1746	sysctl_createv(clog, 0, NULL, NULL,
1747		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1748		       CTLTYPE_INT, "lowportmax",
1749		       SYSCTL_DESCR("Highest privileged ephemeral port number "
1750				    "to assign"),
1751		       sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
1752		       CTL_NET, PF_INET, IPPROTO_IP,
1753		       IPCTL_LOWPORTMAX, CTL_EOL);
1754#endif /* IPNOPRIVPORTS */
1755#if NGRE > 0
1756	sysctl_createv(clog, 0, NULL, NULL,
1757		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1758		       CTLTYPE_INT, "grettl",
1759		       SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1760		       NULL, 0, &ip_gre_ttl, 0,
1761		       CTL_NET, PF_INET, IPPROTO_IP,
1762		       IPCTL_GRE_TTL, CTL_EOL);
1763#endif /* NGRE */
1764	sysctl_createv(clog, 0, NULL, NULL,
1765		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1766		       CTLTYPE_INT, "checkinterface",
1767		       SYSCTL_DESCR("Enable receive side of Strong ES model "
1768				    "from RFC1122"),
1769		       NULL, 0, &ip_checkinterface, 0,
1770		       CTL_NET, PF_INET, IPPROTO_IP,
1771		       IPCTL_CHECKINTERFACE, CTL_EOL);
1772	sysctl_createv(clog, 0, NULL, NULL,
1773		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1774		       CTLTYPE_INT, "random_id",
1775		       SYSCTL_DESCR("Assign random ip_id values"),
1776		       NULL, 0, &ip_do_randomid, 0,
1777		       CTL_NET, PF_INET, IPPROTO_IP,
1778		       IPCTL_RANDOMID, CTL_EOL);
1779	sysctl_createv(clog, 0, NULL, NULL,
1780		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1781		       CTLTYPE_INT, "do_loopback_cksum",
1782		       SYSCTL_DESCR("Perform IP checksum on loopback"),
1783		       NULL, 0, &ip_do_loopback_cksum, 0,
1784		       CTL_NET, PF_INET, IPPROTO_IP,
1785		       IPCTL_LOOPBACKCKSUM, CTL_EOL);
1786	sysctl_createv(clog, 0, NULL, NULL,
1787		       CTLFLAG_PERMANENT,
1788		       CTLTYPE_STRUCT, "stats",
1789		       SYSCTL_DESCR("IP statistics"),
1790		       sysctl_net_inet_ip_stats, 0, NULL, 0,
1791		       CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
1792		       CTL_EOL);
1793#if NARP
1794	sysctl_createv(clog, 0, NULL, NULL,
1795		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1796		       CTLTYPE_INT, "dad_count",
1797		       SYSCTL_DESCR("Number of Duplicate Address Detection "
1798				    "probes to send"),
1799		       NULL, 0, &ip_dad_count, 0,
1800		       CTL_NET, PF_INET, IPPROTO_IP,
1801		       IPCTL_DAD_COUNT, CTL_EOL);
1802#endif
1803
1804	/* anonportalgo RFC6056 subtree */
1805	const struct sysctlnode *portalgo_node;
1806	sysctl_createv(clog, 0, NULL, &portalgo_node,
1807		       CTLFLAG_PERMANENT,
1808		       CTLTYPE_NODE, "anonportalgo",
1809		       SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"),
1810	    	       NULL, 0, NULL, 0,
1811		       CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL);
1812	sysctl_createv(clog, 0, &portalgo_node, NULL,
1813		       CTLFLAG_PERMANENT,
1814		       CTLTYPE_STRING, "available",
1815		       SYSCTL_DESCR("available algorithms"),
1816		       sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
1817		       CTL_CREATE, CTL_EOL);
1818	sysctl_createv(clog, 0, &portalgo_node, NULL,
1819		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1820		       CTLTYPE_STRING, "selected",
1821		       SYSCTL_DESCR("selected algorithm"),
1822		       sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN,
1823		       CTL_CREATE, CTL_EOL);
1824	sysctl_createv(clog, 0, &portalgo_node, NULL,
1825		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1826		       CTLTYPE_STRUCT, "reserve",
1827		       SYSCTL_DESCR("bitmap of reserved ports"),
1828		       sysctl_portalgo_reserve4, 0, NULL, 0,
1829		       CTL_CREATE, CTL_EOL);
1830}
1831
1832void
1833ip_statinc(u_int stat)
1834{
1835
1836	KASSERT(stat < IP_NSTATS);
1837	IP_STATINC(stat);
1838}
1839