ip_input.c revision 1.361 1 /* $NetBSD: ip_input.c,v 1.361 2017/09/27 10:05:04 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 1.361 2017/09/27 10:05:04 ozaki-r 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 */
187 int ipforwarding = IPFORWARDING;
188 int ipsendredirects = IPSENDREDIRECTS;
189 int ip_defttl = IPDEFTTL;
190 int ip_forwsrcrt = IPFORWSRCRT;
191 int ip_directedbcast = IPDIRECTEDBCAST;
192 int ip_allowsrcrt = IPALLOWSRCRT;
193 int ip_mtudisc = IPMTUDISC;
194 int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
195 #ifdef DIAGNOSTIC
196 int ipprintfs = 0;
197 #endif
198
199 int 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 */
214 static int ip_checkinterface __read_mostly = 0;
215
216 struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
217
218 pktqueue_t * ip_pktq __read_mostly;
219 pfil_head_t * inet_pfil_hook __read_mostly;
220 ipid_state_t * ip_ids __read_mostly;
221 percpu_t * ipstat_percpu __read_mostly;
222
223 static percpu_t *ipforward_rt_percpu __cacheline_aligned;
224
225 uint16_t ip_id;
226
227 #ifdef INET_CSUM_COUNTERS
228 #include <sys/device.h>
229
230 struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
231 NULL, "inet", "hwcsum bad");
232 struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
233 NULL, "inet", "hwcsum ok");
234 struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
235 NULL, "inet", "swcsum");
236
237 #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
238
239 EVCNT_ATTACH_STATIC(ip_hwcsum_bad);
240 EVCNT_ATTACH_STATIC(ip_hwcsum_ok);
241 EVCNT_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 */
256 struct 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
264 static int ip_drainwanted;
265
266 static void save_rte(struct mbuf *, u_char *, struct in_addr);
267
268 #ifdef MBUFTRACE
269 struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx");
270 struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx");
271 #endif
272
273 static void ipintr(void *);
274 static void ip_input(struct mbuf *);
275 static void ip_forward(struct mbuf *, int, struct ifnet *);
276 static bool ip_dooptions(struct mbuf *);
277 static struct in_ifaddr *ip_rtaddr(struct in_addr, struct psref *);
278 static void sysctl_net_inet_ip_setup(struct sysctllog **);
279
280 static struct in_ifaddr *ip_match_our_address(struct ifnet *, struct ip *,
281 int *);
282 static 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 */
297 void
298 ip_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
343 static struct in_ifaddr *
344 ip_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
385 static struct in_ifaddr *
386 ip_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
415 matched:
416 return ia;
417 }
418
419 /*
420 * IP software interrupt routine.
421 */
422 static void
423 ipintr(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 */
444 static void
445 ip_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 /* Check the security policy (SP) for the packet */
756 if (ipsec_used) {
757 if (ipsec4_input(m, IP_FORWARDING |
758 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) {
759 goto out;
760 }
761 }
762 #endif
763 ip_forward(m, srcrt, ifp);
764 m_put_rcvif_psref(ifp, &psref);
765 }
766 return;
767
768 ours:
769 m_put_rcvif_psref(ifp, &psref);
770 ifp = NULL;
771
772 /*
773 * If offset or IP_MF are set, must reassemble.
774 */
775 if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
776 /*
777 * Pass to IP reassembly mechanism.
778 */
779 if (ip_reass_packet(&m, ip) != 0) {
780 /* Failed; invalid fragment(s) or packet. */
781 goto out;
782 }
783 if (m == NULL) {
784 /* More fragments should come; silently return. */
785 goto out;
786 }
787 /*
788 * Reassembly is done, we have the final packet.
789 * Updated cached data in local variable(s).
790 */
791 ip = mtod(m, struct ip *);
792 hlen = ip->ip_hl << 2;
793 }
794
795 #ifdef IPSEC
796 /*
797 * Enforce IPsec policy checking if we are seeing last header.
798 * Note that we do not visit this with protocols with PCB layer
799 * code - like UDP/TCP/raw IP.
800 */
801 if (ipsec_used &&
802 (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
803 if (ipsec4_input(m, 0) != 0) {
804 goto out;
805 }
806 }
807 #endif
808
809 /*
810 * Switch out to protocol's input routine.
811 */
812 #if IFA_STATS
813 if (ia && ip) {
814 struct in_ifaddr *_ia;
815 /*
816 * Keep a reference from ip_match_our_address with psref
817 * is expensive, so explore ia here again.
818 */
819 s = pserialize_read_enter();
820 _ia = in_get_ia(ip->ip_dst);
821 _ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
822 pserialize_read_exit(s);
823 }
824 #endif
825 IP_STATINC(IP_STAT_DELIVERED);
826
827 const int off = hlen, nh = ip->ip_p;
828
829 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
830 return;
831
832 out:
833 m_put_rcvif_psref(ifp, &psref);
834 if (m != NULL)
835 m_freem(m);
836 }
837
838 /*
839 * IP timer processing.
840 */
841 void
842 ip_slowtimo(void)
843 {
844
845 #ifndef NET_MPSAFE
846 mutex_enter(softnet_lock);
847 KERNEL_LOCK(1, NULL);
848 #endif
849
850 ip_reass_slowtimo();
851
852 #ifndef NET_MPSAFE
853 KERNEL_UNLOCK_ONE(NULL);
854 mutex_exit(softnet_lock);
855 #endif
856 }
857
858 /*
859 * IP drain processing.
860 */
861 void
862 ip_drain(void)
863 {
864
865 KERNEL_LOCK(1, NULL);
866 ip_reass_drain();
867 KERNEL_UNLOCK_ONE(NULL);
868 }
869
870 /*
871 * ip_dooptions: perform option processing on a datagram, possibly discarding
872 * it if bad options are encountered, or forwarding it if source-routed.
873 *
874 * => Returns true if packet has been forwarded/freed.
875 * => Returns false if the packet should be processed further.
876 */
877 static bool
878 ip_dooptions(struct mbuf *m)
879 {
880 struct ip *ip = mtod(m, struct ip *);
881 u_char *cp, *cp0;
882 struct ip_timestamp *ipt;
883 struct in_ifaddr *ia;
884 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
885 struct in_addr dst;
886 n_time ntime;
887 struct ifaddr *ifa = NULL;
888 int s;
889
890 dst = ip->ip_dst;
891 cp = (u_char *)(ip + 1);
892 cnt = (ip->ip_hl << 2) - sizeof (struct ip);
893 for (; cnt > 0; cnt -= optlen, cp += optlen) {
894 opt = cp[IPOPT_OPTVAL];
895 if (opt == IPOPT_EOL)
896 break;
897 if (opt == IPOPT_NOP)
898 optlen = 1;
899 else {
900 if (cnt < IPOPT_OLEN + sizeof(*cp)) {
901 code = &cp[IPOPT_OLEN] - (u_char *)ip;
902 goto bad;
903 }
904 optlen = cp[IPOPT_OLEN];
905 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
906 code = &cp[IPOPT_OLEN] - (u_char *)ip;
907 goto bad;
908 }
909 }
910 switch (opt) {
911
912 default:
913 break;
914
915 /*
916 * Source routing with record.
917 * Find interface with current destination address.
918 * If none on this machine then drop if strictly routed,
919 * or do nothing if loosely routed.
920 * Record interface address and bring up next address
921 * component. If strictly routed make sure next
922 * address is on directly accessible net.
923 */
924 case IPOPT_LSRR:
925 case IPOPT_SSRR: {
926 struct psref psref;
927 struct sockaddr_in ipaddr = {
928 .sin_len = sizeof(ipaddr),
929 .sin_family = AF_INET,
930 };
931
932 if (ip_allowsrcrt == 0) {
933 type = ICMP_UNREACH;
934 code = ICMP_UNREACH_NET_PROHIB;
935 goto bad;
936 }
937 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
938 code = &cp[IPOPT_OLEN] - (u_char *)ip;
939 goto bad;
940 }
941 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
942 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
943 goto bad;
944 }
945 ipaddr.sin_addr = ip->ip_dst;
946
947 s = pserialize_read_enter();
948 ifa = ifa_ifwithaddr(sintosa(&ipaddr));
949 if (ifa == NULL) {
950 pserialize_read_exit(s);
951 if (opt == IPOPT_SSRR) {
952 type = ICMP_UNREACH;
953 code = ICMP_UNREACH_SRCFAIL;
954 goto bad;
955 }
956 /*
957 * Loose routing, and not at next destination
958 * yet; nothing to do except forward.
959 */
960 break;
961 }
962 pserialize_read_exit(s);
963
964 off--; /* 0 origin */
965 if ((off + sizeof(struct in_addr)) > optlen) {
966 /*
967 * End of source route. Should be for us.
968 */
969 save_rte(m, cp, ip->ip_src);
970 break;
971 }
972 /*
973 * locate outgoing interface
974 */
975 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off),
976 sizeof(ipaddr.sin_addr));
977 if (opt == IPOPT_SSRR) {
978 ifa = ifa_ifwithladdr_psref(sintosa(&ipaddr),
979 &psref);
980 if (ifa != NULL)
981 ia = ifatoia(ifa);
982 else
983 ia = NULL;
984 } else {
985 ia = ip_rtaddr(ipaddr.sin_addr, &psref);
986 }
987 if (ia == NULL) {
988 type = ICMP_UNREACH;
989 code = ICMP_UNREACH_SRCFAIL;
990 goto bad;
991 }
992 ip->ip_dst = ipaddr.sin_addr;
993 bcopy((void *)&ia->ia_addr.sin_addr,
994 (void *)(cp + off), sizeof(struct in_addr));
995 ia4_release(ia, &psref);
996 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
997 /*
998 * Let ip_intr's mcast routing check handle mcast pkts
999 */
1000 forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1001 break;
1002 }
1003
1004 case IPOPT_RR: {
1005 struct psref psref;
1006 struct sockaddr_in ipaddr = {
1007 .sin_len = sizeof(ipaddr),
1008 .sin_family = AF_INET,
1009 };
1010
1011 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1012 code = &cp[IPOPT_OLEN] - (u_char *)ip;
1013 goto bad;
1014 }
1015 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1016 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1017 goto bad;
1018 }
1019 /*
1020 * If no space remains, ignore.
1021 */
1022 off--; /* 0 origin */
1023 if ((off + sizeof(struct in_addr)) > optlen)
1024 break;
1025 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst),
1026 sizeof(ipaddr.sin_addr));
1027 /*
1028 * locate outgoing interface; if we're the destination,
1029 * use the incoming interface (should be same).
1030 */
1031 ifa = ifa_ifwithaddr_psref(sintosa(&ipaddr), &psref);
1032 if (ifa == NULL) {
1033 ia = ip_rtaddr(ipaddr.sin_addr, &psref);
1034 if (ia == NULL) {
1035 type = ICMP_UNREACH;
1036 code = ICMP_UNREACH_HOST;
1037 goto bad;
1038 }
1039 } else {
1040 ia = ifatoia(ifa);
1041 }
1042 bcopy((void *)&ia->ia_addr.sin_addr,
1043 (void *)(cp + off), sizeof(struct in_addr));
1044 ia4_release(ia, &psref);
1045 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1046 break;
1047 }
1048
1049 case IPOPT_TS:
1050 code = cp - (u_char *)ip;
1051 ipt = (struct ip_timestamp *)cp;
1052 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1053 code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1054 goto bad;
1055 }
1056 if (ipt->ipt_ptr < 5) {
1057 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1058 goto bad;
1059 }
1060 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1061 if (++ipt->ipt_oflw == 0) {
1062 code = (u_char *)&ipt->ipt_ptr -
1063 (u_char *)ip;
1064 goto bad;
1065 }
1066 break;
1067 }
1068 cp0 = (cp + ipt->ipt_ptr - 1);
1069 switch (ipt->ipt_flg) {
1070
1071 case IPOPT_TS_TSONLY:
1072 break;
1073
1074 case IPOPT_TS_TSANDADDR: {
1075 struct ifnet *rcvif;
1076 int _s, _ss;
1077 struct sockaddr_in ipaddr = {
1078 .sin_len = sizeof(ipaddr),
1079 .sin_family = AF_INET,
1080 };
1081
1082 if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1083 sizeof(struct in_addr) > ipt->ipt_len) {
1084 code = (u_char *)&ipt->ipt_ptr -
1085 (u_char *)ip;
1086 goto bad;
1087 }
1088 ipaddr.sin_addr = dst;
1089 _ss = pserialize_read_enter();
1090 rcvif = m_get_rcvif(m, &_s);
1091 if (__predict_true(rcvif != NULL)) {
1092 ifa = ifaof_ifpforaddr(sintosa(&ipaddr),
1093 rcvif);
1094 }
1095 m_put_rcvif(rcvif, &_s);
1096 if (ifa == NULL) {
1097 pserialize_read_exit(_ss);
1098 break;
1099 }
1100 ia = ifatoia(ifa);
1101 bcopy(&ia->ia_addr.sin_addr,
1102 cp0, sizeof(struct in_addr));
1103 pserialize_read_exit(_ss);
1104 ipt->ipt_ptr += sizeof(struct in_addr);
1105 break;
1106 }
1107
1108 case IPOPT_TS_PRESPEC: {
1109 struct sockaddr_in ipaddr = {
1110 .sin_len = sizeof(ipaddr),
1111 .sin_family = AF_INET,
1112 };
1113
1114 if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1115 sizeof(struct in_addr) > ipt->ipt_len) {
1116 code = (u_char *)&ipt->ipt_ptr -
1117 (u_char *)ip;
1118 goto bad;
1119 }
1120 memcpy(&ipaddr.sin_addr, cp0,
1121 sizeof(struct in_addr));
1122 s = pserialize_read_enter();
1123 ifa = ifa_ifwithaddr(sintosa(&ipaddr));
1124 if (ifa == NULL) {
1125 pserialize_read_exit(s);
1126 continue;
1127 }
1128 pserialize_read_exit(s);
1129 ipt->ipt_ptr += sizeof(struct in_addr);
1130 break;
1131 }
1132
1133 default:
1134 /* XXX can't take &ipt->ipt_flg */
1135 code = (u_char *)&ipt->ipt_ptr -
1136 (u_char *)ip + 1;
1137 goto bad;
1138 }
1139 ntime = iptime();
1140 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1141 memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1142 sizeof(n_time));
1143 ipt->ipt_ptr += sizeof(n_time);
1144 }
1145 }
1146 if (forward) {
1147 struct ifnet *rcvif;
1148 struct psref _psref;
1149
1150 if (ip_forwsrcrt == 0) {
1151 type = ICMP_UNREACH;
1152 code = ICMP_UNREACH_SRCFAIL;
1153 goto bad;
1154 }
1155
1156 rcvif = m_get_rcvif_psref(m, &_psref);
1157 if (__predict_false(rcvif == NULL)) {
1158 type = ICMP_UNREACH;
1159 code = ICMP_UNREACH_HOST;
1160 goto bad;
1161 }
1162 ip_forward(m, 1, rcvif);
1163 m_put_rcvif_psref(rcvif, &_psref);
1164 return true;
1165 }
1166 return false;
1167 bad:
1168 icmp_error(m, type, code, 0, 0);
1169 IP_STATINC(IP_STAT_BADOPTIONS);
1170 return true;
1171 }
1172
1173 /*
1174 * ip_rtaddr: given address of next destination (final or next hop),
1175 * return internet address info of interface to be used to get there.
1176 */
1177 static struct in_ifaddr *
1178 ip_rtaddr(struct in_addr dst, struct psref *psref)
1179 {
1180 struct rtentry *rt;
1181 union {
1182 struct sockaddr dst;
1183 struct sockaddr_in dst4;
1184 } u;
1185 struct route *ro;
1186
1187 sockaddr_in_init(&u.dst4, &dst, 0);
1188
1189 ro = percpu_getref(ipforward_rt_percpu);
1190 rt = rtcache_lookup(ro, &u.dst);
1191 if (rt == NULL) {
1192 percpu_putref(ipforward_rt_percpu);
1193 return NULL;
1194 }
1195
1196 ia4_acquire(ifatoia(rt->rt_ifa), psref);
1197 rtcache_unref(rt, ro);
1198 percpu_putref(ipforward_rt_percpu);
1199
1200 return ifatoia(rt->rt_ifa);
1201 }
1202
1203 /*
1204 * save_rte: save incoming source route for use in replies, to be picked
1205 * up later by ip_srcroute if the receiver is interested.
1206 */
1207 static void
1208 save_rte(struct mbuf *m, u_char *option, struct in_addr dst)
1209 {
1210 struct ip_srcrt *isr;
1211 struct m_tag *mtag;
1212 unsigned olen;
1213
1214 olen = option[IPOPT_OLEN];
1215 if (olen > sizeof(isr->isr_hdr) + sizeof(isr->isr_routes))
1216 return;
1217
1218 mtag = m_tag_get(PACKET_TAG_SRCROUTE, sizeof(*isr), M_NOWAIT);
1219 if (mtag == NULL)
1220 return;
1221 isr = (struct ip_srcrt *)(mtag + 1);
1222
1223 memcpy(isr->isr_hdr, option, olen);
1224 isr->isr_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1225 isr->isr_dst = dst;
1226 m_tag_prepend(m, mtag);
1227 }
1228
1229 /*
1230 * Retrieve incoming source route for use in replies,
1231 * in the same form used by setsockopt.
1232 * The first hop is placed before the options, will be removed later.
1233 */
1234 struct mbuf *
1235 ip_srcroute(struct mbuf *m0)
1236 {
1237 struct in_addr *p, *q;
1238 struct mbuf *m;
1239 struct ip_srcrt *isr;
1240 struct m_tag *mtag;
1241
1242 mtag = m_tag_find(m0, PACKET_TAG_SRCROUTE, NULL);
1243 if (mtag == NULL)
1244 return NULL;
1245 isr = (struct ip_srcrt *)(mtag + 1);
1246
1247 if (isr->isr_nhops == 0)
1248 return NULL;
1249
1250 m = m_get(M_DONTWAIT, MT_SOOPTS);
1251 if (m == NULL)
1252 return NULL;
1253
1254 MCLAIM(m, &inetdomain.dom_mowner);
1255 #define OPTSIZ (sizeof(isr->isr_nop) + sizeof(isr->isr_hdr))
1256
1257 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + header) */
1258 m->m_len = (isr->isr_nhops + 1) * sizeof(struct in_addr) + OPTSIZ;
1259
1260 /*
1261 * First save first hop for return route
1262 */
1263 p = &(isr->isr_routes[isr->isr_nhops - 1]);
1264 *(mtod(m, struct in_addr *)) = *p--;
1265
1266 /*
1267 * Copy option fields and padding (nop) to mbuf.
1268 */
1269 isr->isr_nop = IPOPT_NOP;
1270 isr->isr_hdr[IPOPT_OFFSET] = IPOPT_MINOFF;
1271 memmove(mtod(m, char *) + sizeof(struct in_addr), &isr->isr_nop,
1272 OPTSIZ);
1273 q = (struct in_addr *)(mtod(m, char *) +
1274 sizeof(struct in_addr) + OPTSIZ);
1275 #undef OPTSIZ
1276 /*
1277 * Record return path as an IP source route,
1278 * reversing the path (pointers are now aligned).
1279 */
1280 while (p >= isr->isr_routes) {
1281 *q++ = *p--;
1282 }
1283 /*
1284 * Last hop goes to final destination.
1285 */
1286 *q = isr->isr_dst;
1287 m_tag_delete(m0, mtag);
1288 return m;
1289 }
1290
1291 const int inetctlerrmap[PRC_NCMDS] = {
1292 [PRC_MSGSIZE] = EMSGSIZE,
1293 [PRC_HOSTDEAD] = EHOSTDOWN,
1294 [PRC_HOSTUNREACH] = EHOSTUNREACH,
1295 [PRC_UNREACH_NET] = EHOSTUNREACH,
1296 [PRC_UNREACH_HOST] = EHOSTUNREACH,
1297 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED,
1298 [PRC_UNREACH_PORT] = ECONNREFUSED,
1299 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH,
1300 [PRC_PARAMPROB] = ENOPROTOOPT,
1301 };
1302
1303 void
1304 ip_fasttimo(void)
1305 {
1306 if (ip_drainwanted) {
1307 ip_drain();
1308 ip_drainwanted = 0;
1309 }
1310 }
1311
1312 void
1313 ip_drainstub(void)
1314 {
1315 ip_drainwanted = 1;
1316 }
1317
1318 /*
1319 * Forward a packet. If some error occurs return the sender
1320 * an icmp packet. Note we can't always generate a meaningful
1321 * icmp message because icmp doesn't have a large enough repertoire
1322 * of codes and types.
1323 *
1324 * If not forwarding, just drop the packet. This could be confusing
1325 * if ipforwarding was zero but some routing protocol was advancing
1326 * us as a gateway to somewhere. However, we must let the routing
1327 * protocol deal with that.
1328 *
1329 * The srcrt parameter indicates whether the packet is being forwarded
1330 * via a source route.
1331 */
1332 static void
1333 ip_forward(struct mbuf *m, int srcrt, struct ifnet *rcvif)
1334 {
1335 struct ip *ip = mtod(m, struct ip *);
1336 struct rtentry *rt;
1337 int error, type = 0, code = 0, destmtu = 0;
1338 struct mbuf *mcopy;
1339 n_long dest;
1340 union {
1341 struct sockaddr dst;
1342 struct sockaddr_in dst4;
1343 } u;
1344 uint64_t *ips;
1345 struct route *ro;
1346
1347 KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software "
1348 "interrupt handler; synchronization assumptions violated");
1349
1350 /*
1351 * We are now in the output path.
1352 */
1353 MCLAIM(m, &ip_tx_mowner);
1354
1355 /*
1356 * Clear any in-bound checksum flags for this packet.
1357 */
1358 m->m_pkthdr.csum_flags = 0;
1359
1360 dest = 0;
1361 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1362 IP_STATINC(IP_STAT_CANTFORWARD);
1363 m_freem(m);
1364 return;
1365 }
1366
1367 if (ip->ip_ttl <= IPTTLDEC) {
1368 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1369 return;
1370 }
1371
1372 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1373
1374 ro = percpu_getref(ipforward_rt_percpu);
1375 rt = rtcache_lookup(ro, &u.dst);
1376 if (rt == NULL) {
1377 percpu_putref(ipforward_rt_percpu);
1378 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1379 return;
1380 }
1381
1382 /*
1383 * Save at most 68 bytes of the packet in case
1384 * we need to generate an ICMP message to the src.
1385 * Pullup to avoid sharing mbuf cluster between m and mcopy.
1386 */
1387 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1388 if (mcopy)
1389 mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1390
1391 ip->ip_ttl -= IPTTLDEC;
1392
1393 /*
1394 * If forwarding packet using same interface that it came in on,
1395 * perhaps should send a redirect to sender to shortcut a hop.
1396 * Only send redirect if source is sending directly to us,
1397 * and if packet was not source routed (or has any options).
1398 * Also, don't send redirect if forwarding using a default route
1399 * or a route modified by a redirect.
1400 */
1401 if (rt->rt_ifp == rcvif &&
1402 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1403 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1404 ipsendredirects && !srcrt) {
1405 if (rt->rt_ifa &&
1406 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1407 ifatoia(rt->rt_ifa)->ia_subnet) {
1408 if (rt->rt_flags & RTF_GATEWAY)
1409 dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1410 else
1411 dest = ip->ip_dst.s_addr;
1412 /*
1413 * Router requirements says to only send host
1414 * redirects.
1415 */
1416 type = ICMP_REDIRECT;
1417 code = ICMP_REDIRECT_HOST;
1418 }
1419 }
1420 rtcache_unref(rt, ro);
1421
1422 error = ip_output(m, NULL, ro,
1423 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1424 NULL, NULL);
1425
1426 if (error) {
1427 IP_STATINC(IP_STAT_CANTFORWARD);
1428 goto error;
1429 }
1430
1431 ips = IP_STAT_GETREF();
1432 ips[IP_STAT_FORWARD]++;
1433
1434 if (type) {
1435 ips[IP_STAT_REDIRECTSENT]++;
1436 IP_STAT_PUTREF();
1437 goto redirect;
1438 }
1439
1440 IP_STAT_PUTREF();
1441 if (mcopy) {
1442 #ifdef GATEWAY
1443 if (mcopy->m_flags & M_CANFASTFWD)
1444 ipflow_create(ro, mcopy);
1445 #endif
1446 m_freem(mcopy);
1447 }
1448
1449 percpu_putref(ipforward_rt_percpu);
1450 return;
1451
1452 redirect:
1453 error:
1454 if (mcopy == NULL) {
1455 percpu_putref(ipforward_rt_percpu);
1456 return;
1457 }
1458
1459 switch (error) {
1460
1461 case 0: /* forwarded, but need redirect */
1462 /* type, code set above */
1463 break;
1464
1465 case ENETUNREACH: /* shouldn't happen, checked above */
1466 case EHOSTUNREACH:
1467 case ENETDOWN:
1468 case EHOSTDOWN:
1469 default:
1470 type = ICMP_UNREACH;
1471 code = ICMP_UNREACH_HOST;
1472 break;
1473
1474 case EMSGSIZE:
1475 type = ICMP_UNREACH;
1476 code = ICMP_UNREACH_NEEDFRAG;
1477
1478 if ((rt = rtcache_validate(ro)) != NULL) {
1479 destmtu = rt->rt_ifp->if_mtu;
1480 rtcache_unref(rt, ro);
1481 }
1482 #ifdef IPSEC
1483 if (ipsec_used)
1484 (void)ipsec4_forward(mcopy, &destmtu);
1485 #endif
1486 IP_STATINC(IP_STAT_CANTFRAG);
1487 break;
1488
1489 case ENOBUFS:
1490 /*
1491 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812,
1492 * Requirements for IP Version 4 Routers. Source quench can
1493 * big problem under DoS attacks or if the underlying
1494 * interface is rate-limited.
1495 */
1496 if (mcopy)
1497 m_freem(mcopy);
1498 percpu_putref(ipforward_rt_percpu);
1499 return;
1500 }
1501 icmp_error(mcopy, type, code, dest, destmtu);
1502 percpu_putref(ipforward_rt_percpu);
1503 }
1504
1505 void
1506 ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
1507 struct mbuf *m)
1508 {
1509 struct socket *so = inp->inp_socket;
1510 int inpflags = inp->inp_flags;
1511
1512 if (SOOPT_TIMESTAMP(so->so_options))
1513 mp = sbsavetimestamp(so->so_options, m, mp);
1514
1515 if (inpflags & INP_RECVDSTADDR) {
1516 *mp = sbcreatecontrol(&ip->ip_dst,
1517 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1518 if (*mp)
1519 mp = &(*mp)->m_next;
1520 }
1521
1522 if (inpflags & INP_RECVTTL) {
1523 *mp = sbcreatecontrol(&ip->ip_ttl,
1524 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP);
1525 if (*mp)
1526 mp = &(*mp)->m_next;
1527 }
1528
1529 struct psref psref;
1530 ifnet_t *ifp = m_get_rcvif_psref(m, &psref);
1531 if (__predict_false(ifp == NULL)) {
1532 #ifdef DIAGNOSTIC
1533 printf("%s: missing receive interface\n", __func__);
1534 #endif
1535 return; /* XXX should report error? */
1536 }
1537
1538 if (inpflags & INP_RECVPKTINFO) {
1539 struct in_pktinfo ipi;
1540 ipi.ipi_addr = ip->ip_src;
1541 ipi.ipi_ifindex = ifp->if_index;
1542 *mp = sbcreatecontrol(&ipi,
1543 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP);
1544 if (*mp)
1545 mp = &(*mp)->m_next;
1546 }
1547 if (inpflags & INP_PKTINFO) {
1548 struct in_pktinfo ipi;
1549 ipi.ipi_addr = ip->ip_dst;
1550 ipi.ipi_ifindex = ifp->if_index;
1551 *mp = sbcreatecontrol(&ipi,
1552 sizeof(ipi), IP_PKTINFO, IPPROTO_IP);
1553 if (*mp)
1554 mp = &(*mp)->m_next;
1555 }
1556 if (inpflags & INP_RECVIF) {
1557 struct sockaddr_dl sdl;
1558
1559 sockaddr_dl_init(&sdl, sizeof(sdl), ifp->if_index, 0, NULL, 0,
1560 NULL, 0);
1561 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1562 if (*mp)
1563 mp = &(*mp)->m_next;
1564 }
1565 m_put_rcvif_psref(ifp, &psref);
1566 }
1567
1568 /*
1569 * sysctl helper routine for net.inet.ip.forwsrcrt.
1570 */
1571 static int
1572 sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
1573 {
1574 int error, tmp;
1575 struct sysctlnode node;
1576
1577 node = *rnode;
1578 tmp = ip_forwsrcrt;
1579 node.sysctl_data = &tmp;
1580 error = sysctl_lookup(SYSCTLFN_CALL(&node));
1581 if (error || newp == NULL)
1582 return (error);
1583
1584 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
1585 0, NULL, NULL, NULL);
1586 if (error)
1587 return (error);
1588
1589 ip_forwsrcrt = tmp;
1590
1591 return (0);
1592 }
1593
1594 /*
1595 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
1596 * range of the new value and tweaks timers if it changes.
1597 */
1598 static int
1599 sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1600 {
1601 int error, tmp;
1602 struct sysctlnode node;
1603
1604 icmp_mtudisc_lock();
1605
1606 node = *rnode;
1607 tmp = ip_mtudisc_timeout;
1608 node.sysctl_data = &tmp;
1609 error = sysctl_lookup(SYSCTLFN_CALL(&node));
1610 if (error || newp == NULL)
1611 goto out;
1612 if (tmp < 0) {
1613 error = EINVAL;
1614 goto out;
1615 }
1616
1617 ip_mtudisc_timeout = tmp;
1618 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
1619 error = 0;
1620 out:
1621 icmp_mtudisc_unlock();
1622 return error;
1623 }
1624
1625 static int
1626 sysctl_net_inet_ip_stats(SYSCTLFN_ARGS)
1627 {
1628
1629 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS));
1630 }
1631
1632 static void
1633 sysctl_net_inet_ip_setup(struct sysctllog **clog)
1634 {
1635 sysctl_createv(clog, 0, NULL, NULL,
1636 CTLFLAG_PERMANENT,
1637 CTLTYPE_NODE, "inet",
1638 SYSCTL_DESCR("PF_INET related settings"),
1639 NULL, 0, NULL, 0,
1640 CTL_NET, PF_INET, CTL_EOL);
1641 sysctl_createv(clog, 0, NULL, NULL,
1642 CTLFLAG_PERMANENT,
1643 CTLTYPE_NODE, "ip",
1644 SYSCTL_DESCR("IPv4 related settings"),
1645 NULL, 0, NULL, 0,
1646 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1647
1648 sysctl_createv(clog, 0, NULL, NULL,
1649 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1650 CTLTYPE_INT, "forwarding",
1651 SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1652 NULL, 0, &ipforwarding, 0,
1653 CTL_NET, PF_INET, IPPROTO_IP,
1654 IPCTL_FORWARDING, CTL_EOL);
1655 sysctl_createv(clog, 0, NULL, NULL,
1656 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1657 CTLTYPE_INT, "redirect",
1658 SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1659 NULL, 0, &ipsendredirects, 0,
1660 CTL_NET, PF_INET, IPPROTO_IP,
1661 IPCTL_SENDREDIRECTS, CTL_EOL);
1662 sysctl_createv(clog, 0, NULL, NULL,
1663 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1664 CTLTYPE_INT, "ttl",
1665 SYSCTL_DESCR("Default TTL for an INET datagram"),
1666 NULL, 0, &ip_defttl, 0,
1667 CTL_NET, PF_INET, IPPROTO_IP,
1668 IPCTL_DEFTTL, CTL_EOL);
1669 #ifdef IPCTL_DEFMTU
1670 sysctl_createv(clog, 0, NULL, NULL,
1671 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1672 CTLTYPE_INT, "mtu",
1673 SYSCTL_DESCR("Default MTA for an INET route"),
1674 NULL, 0, &ip_mtu, 0,
1675 CTL_NET, PF_INET, IPPROTO_IP,
1676 IPCTL_DEFMTU, CTL_EOL);
1677 #endif /* IPCTL_DEFMTU */
1678 sysctl_createv(clog, 0, NULL, NULL,
1679 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1680 CTLTYPE_INT, "forwsrcrt",
1681 SYSCTL_DESCR("Enable forwarding of source-routed "
1682 "datagrams"),
1683 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1684 CTL_NET, PF_INET, IPPROTO_IP,
1685 IPCTL_FORWSRCRT, CTL_EOL);
1686 sysctl_createv(clog, 0, NULL, NULL,
1687 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1688 CTLTYPE_INT, "directed-broadcast",
1689 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1690 NULL, 0, &ip_directedbcast, 0,
1691 CTL_NET, PF_INET, IPPROTO_IP,
1692 IPCTL_DIRECTEDBCAST, CTL_EOL);
1693 sysctl_createv(clog, 0, NULL, NULL,
1694 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1695 CTLTYPE_INT, "allowsrcrt",
1696 SYSCTL_DESCR("Accept source-routed datagrams"),
1697 NULL, 0, &ip_allowsrcrt, 0,
1698 CTL_NET, PF_INET, IPPROTO_IP,
1699 IPCTL_ALLOWSRCRT, CTL_EOL);
1700
1701 sysctl_createv(clog, 0, NULL, NULL,
1702 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1703 CTLTYPE_INT, "mtudisc",
1704 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1705 NULL, 0, &ip_mtudisc, 0,
1706 CTL_NET, PF_INET, IPPROTO_IP,
1707 IPCTL_MTUDISC, CTL_EOL);
1708 sysctl_createv(clog, 0, NULL, NULL,
1709 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1710 CTLTYPE_INT, "anonportmin",
1711 SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1712 sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
1713 CTL_NET, PF_INET, IPPROTO_IP,
1714 IPCTL_ANONPORTMIN, CTL_EOL);
1715 sysctl_createv(clog, 0, NULL, NULL,
1716 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1717 CTLTYPE_INT, "anonportmax",
1718 SYSCTL_DESCR("Highest ephemeral port number to assign"),
1719 sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
1720 CTL_NET, PF_INET, IPPROTO_IP,
1721 IPCTL_ANONPORTMAX, CTL_EOL);
1722 sysctl_createv(clog, 0, NULL, NULL,
1723 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1724 CTLTYPE_INT, "mtudisctimeout",
1725 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1726 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0,
1727 CTL_NET, PF_INET, IPPROTO_IP,
1728 IPCTL_MTUDISCTIMEOUT, CTL_EOL);
1729 #ifndef IPNOPRIVPORTS
1730 sysctl_createv(clog, 0, NULL, NULL,
1731 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1732 CTLTYPE_INT, "lowportmin",
1733 SYSCTL_DESCR("Lowest privileged ephemeral port number "
1734 "to assign"),
1735 sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
1736 CTL_NET, PF_INET, IPPROTO_IP,
1737 IPCTL_LOWPORTMIN, CTL_EOL);
1738 sysctl_createv(clog, 0, NULL, NULL,
1739 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1740 CTLTYPE_INT, "lowportmax",
1741 SYSCTL_DESCR("Highest privileged ephemeral port number "
1742 "to assign"),
1743 sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
1744 CTL_NET, PF_INET, IPPROTO_IP,
1745 IPCTL_LOWPORTMAX, CTL_EOL);
1746 #endif /* IPNOPRIVPORTS */
1747 #if NGRE > 0
1748 sysctl_createv(clog, 0, NULL, NULL,
1749 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1750 CTLTYPE_INT, "grettl",
1751 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1752 NULL, 0, &ip_gre_ttl, 0,
1753 CTL_NET, PF_INET, IPPROTO_IP,
1754 IPCTL_GRE_TTL, CTL_EOL);
1755 #endif /* NGRE */
1756 sysctl_createv(clog, 0, NULL, NULL,
1757 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1758 CTLTYPE_INT, "checkinterface",
1759 SYSCTL_DESCR("Enable receive side of Strong ES model "
1760 "from RFC1122"),
1761 NULL, 0, &ip_checkinterface, 0,
1762 CTL_NET, PF_INET, IPPROTO_IP,
1763 IPCTL_CHECKINTERFACE, CTL_EOL);
1764 sysctl_createv(clog, 0, NULL, NULL,
1765 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1766 CTLTYPE_INT, "random_id",
1767 SYSCTL_DESCR("Assign random ip_id values"),
1768 NULL, 0, &ip_do_randomid, 0,
1769 CTL_NET, PF_INET, IPPROTO_IP,
1770 IPCTL_RANDOMID, CTL_EOL);
1771 sysctl_createv(clog, 0, NULL, NULL,
1772 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1773 CTLTYPE_INT, "do_loopback_cksum",
1774 SYSCTL_DESCR("Perform IP checksum on loopback"),
1775 NULL, 0, &ip_do_loopback_cksum, 0,
1776 CTL_NET, PF_INET, IPPROTO_IP,
1777 IPCTL_LOOPBACKCKSUM, CTL_EOL);
1778 sysctl_createv(clog, 0, NULL, NULL,
1779 CTLFLAG_PERMANENT,
1780 CTLTYPE_STRUCT, "stats",
1781 SYSCTL_DESCR("IP statistics"),
1782 sysctl_net_inet_ip_stats, 0, NULL, 0,
1783 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
1784 CTL_EOL);
1785 #if NARP
1786 sysctl_createv(clog, 0, NULL, NULL,
1787 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1788 CTLTYPE_INT, "dad_count",
1789 SYSCTL_DESCR("Number of Duplicate Address Detection "
1790 "probes to send"),
1791 NULL, 0, &ip_dad_count, 0,
1792 CTL_NET, PF_INET, IPPROTO_IP,
1793 IPCTL_DAD_COUNT, CTL_EOL);
1794 #endif
1795
1796 /* anonportalgo RFC6056 subtree */
1797 const struct sysctlnode *portalgo_node;
1798 sysctl_createv(clog, 0, NULL, &portalgo_node,
1799 CTLFLAG_PERMANENT,
1800 CTLTYPE_NODE, "anonportalgo",
1801 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"),
1802 NULL, 0, NULL, 0,
1803 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL);
1804 sysctl_createv(clog, 0, &portalgo_node, NULL,
1805 CTLFLAG_PERMANENT,
1806 CTLTYPE_STRING, "available",
1807 SYSCTL_DESCR("available algorithms"),
1808 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
1809 CTL_CREATE, CTL_EOL);
1810 sysctl_createv(clog, 0, &portalgo_node, NULL,
1811 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1812 CTLTYPE_STRING, "selected",
1813 SYSCTL_DESCR("selected algorithm"),
1814 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN,
1815 CTL_CREATE, CTL_EOL);
1816 sysctl_createv(clog, 0, &portalgo_node, NULL,
1817 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1818 CTLTYPE_STRUCT, "reserve",
1819 SYSCTL_DESCR("bitmap of reserved ports"),
1820 sysctl_portalgo_reserve4, 0, NULL, 0,
1821 CTL_CREATE, CTL_EOL);
1822 }
1823
1824 void
1825 ip_statinc(u_int stat)
1826 {
1827
1828 KASSERT(stat < IP_NSTATS);
1829 IP_STATINC(stat);
1830 }
1831