1/*
2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 * @(#)ip_icmp.c 8.2 (Berkeley) 1/4/94
34 * $FreeBSD: head/sys/netinet/ip_icmp.c 105194 2002-10-16 01:54:46Z sam $
35 */
36
37#include "opt_ipsec.h"
38#include "opt_mac.h"
39
40#include <sys/param.h>
41#include <sys/systm.h>
42#include <sys/mac.h>
43#include <sys/mbuf.h>
44#include <sys/protosw.h>
45#include <sys/socket.h>
46#include <sys/time.h>
47#include <sys/kernel.h>
48#include <sys/sysctl.h>
49
50#include <net/if.h>
51#include <net/if_types.h>
52#include <net/route.h>
53
54#define _IP_VHL
55#include <netinet/in.h>
56#include <netinet/in_systm.h>
57#include <netinet/in_var.h>
58#include <netinet/ip.h>
59#include <netinet/ip_icmp.h>
60#include <netinet/ip_var.h>
61#include <netinet/icmp_var.h>
62
63#ifdef IPSEC
64#include <netinet6/ipsec.h>
65#include <netkey/key.h>
66#endif
67
68#include <machine/in_cksum.h>
69
70/*
71 * ICMP routines: error generation, receive packet processing, and
72 * routines to turnaround packets back to the originator, and
73 * host table maintenance routines.
74 */
75
76static struct icmpstat icmpstat;
77SYSCTL_STRUCT(_net_inet_icmp, ICMPCTL_STATS, stats, CTLFLAG_RW,
78 &icmpstat, icmpstat, "");
79
80static int icmpmaskrepl = 0;
81SYSCTL_INT(_net_inet_icmp, ICMPCTL_MASKREPL, maskrepl, CTLFLAG_RW,
82 &icmpmaskrepl, 0, "");
83
84static int drop_redirect = 0;
85SYSCTL_INT(_net_inet_icmp, OID_AUTO, drop_redirect, CTLFLAG_RW,
86 &drop_redirect, 0, "");
87
88static int log_redirect = 0;
89SYSCTL_INT(_net_inet_icmp, OID_AUTO, log_redirect, CTLFLAG_RW,
90 &log_redirect, 0, "");
91
92static int icmplim = 200;
93SYSCTL_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RW,
94 &icmplim, 0, "");
95
96static int icmplim_output = 1;
97SYSCTL_INT(_net_inet_icmp, OID_AUTO, icmplim_output, CTLFLAG_RW,
98 &icmplim_output, 0, "");
99
100/*
101 * ICMP broadcast echo sysctl
102 */
103
104static int icmpbmcastecho = 0;
105SYSCTL_INT(_net_inet_icmp, OID_AUTO, bmcastecho, CTLFLAG_RW,
106 &icmpbmcastecho, 0, "");
107
108
109#ifdef ICMPPRINTFS
110int icmpprintfs = 0;
111#endif
112
113static void icmp_reflect(struct mbuf *);
114static void icmp_send(struct mbuf *, struct mbuf *, struct route *);
115static int ip_next_mtu(int, int);
116
117extern struct protosw inetsw[];
118
119/*
120 * Generate an error packet of type error
121 * in response to bad packet ip.
122 */
123void
124icmp_error(n, type, code, dest, destifp)
125 struct mbuf *n;
126 int type, code;
127 n_long dest;
128 struct ifnet *destifp;
129{
130 register struct ip *oip = mtod(n, struct ip *), *nip;
131 register unsigned oiplen = IP_VHL_HL(oip->ip_vhl) << 2;
132 register struct icmp *icp;
133 register struct mbuf *m;
134 unsigned icmplen;
135
136#ifdef ICMPPRINTFS
137 if (icmpprintfs)
138 printf("icmp_error(%p, %x, %d)\n", oip, type, code);
139#endif
140 if (type != ICMP_REDIRECT)
141 icmpstat.icps_error++;
142 /*
143 * Don't send error if not the first fragment of message.
144 * Don't error if the old packet protocol was ICMP
145 * error message, only known informational types.
146 */
147 if (oip->ip_off &~ (IP_MF|IP_DF))
148 goto freeit;
149 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT &&
150 n->m_len >= oiplen + ICMP_MINLEN &&
151 !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiplen))->icmp_type)) {
152 icmpstat.icps_oldicmp++;
153 goto freeit;
154 }
155 /* Don't send error in response to a multicast or broadcast packet */
156 if (n->m_flags & (M_BCAST|M_MCAST))
157 goto freeit;
158 /*
159 * First, formulate icmp message
160 */
161 m = m_gethdr(M_DONTWAIT, MT_HEADER);
162 if (m == NULL)
163 goto freeit;
164#ifdef MAC
165 mac_create_mbuf_netlayer(n, m);
166#endif
167 icmplen = min(oiplen + 8, oip->ip_len);
168 if (icmplen < sizeof(struct ip))
169 panic("icmp_error: bad length");
170 m->m_len = icmplen + ICMP_MINLEN;
171 MH_ALIGN(m, m->m_len);
172 icp = mtod(m, struct icmp *);
173 if ((u_int)type > ICMP_MAXTYPE)
174 panic("icmp_error");
175 icmpstat.icps_outhist[type]++;
176 icp->icmp_type = type;
177 if (type == ICMP_REDIRECT)
178 icp->icmp_gwaddr.s_addr = dest;
179 else {
180 icp->icmp_void = 0;
181 /*
182 * The following assignments assume an overlay with the
183 * zeroed icmp_void field.
184 */
185 if (type == ICMP_PARAMPROB) {
186 icp->icmp_pptr = code;
187 code = 0;
188 } else if (type == ICMP_UNREACH &&
189 code == ICMP_UNREACH_NEEDFRAG && destifp) {
190 icp->icmp_nextmtu = htons(destifp->if_mtu);
191 }
192 }
193
194 icp->icmp_code = code;
195 m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip);
196 nip = &icp->icmp_ip;
197
198 /*
199 * Convert fields to network representation.
200 */
201 nip->ip_len = htons(nip->ip_len);
202 nip->ip_off = htons(nip->ip_off);
203
204 /*
205 * Now, copy old ip header (without options)
206 * in front of icmp message.
207 */
208 if (m->m_data - sizeof(struct ip) < m->m_pktdat)
209 panic("icmp len");
210 m->m_data -= sizeof(struct ip);
211 m->m_len += sizeof(struct ip);
212 m->m_pkthdr.len = m->m_len;
213 m->m_pkthdr.rcvif = n->m_pkthdr.rcvif;
214 nip = mtod(m, struct ip *);
215 bcopy((caddr_t)oip, (caddr_t)nip, sizeof(struct ip));
216 nip->ip_len = m->m_len;
217 nip->ip_vhl = IP_VHL_BORING;
218 nip->ip_p = IPPROTO_ICMP;
219 nip->ip_tos = 0;
220 icmp_reflect(m);
221
222freeit:
223 m_freem(n);
224}
225
226static struct sockaddr_in icmpsrc = { sizeof (struct sockaddr_in), AF_INET };
227static struct sockaddr_in icmpdst = { sizeof (struct sockaddr_in), AF_INET };
228static struct sockaddr_in icmpgw = { sizeof (struct sockaddr_in), AF_INET };
229
230/*
231 * Process a received ICMP message.
232 */
233void
234icmp_input(m, off)
235 register struct mbuf *m;
236 int off;
237{
238 int hlen = off;
239 register struct icmp *icp;
240 register struct ip *ip = mtod(m, struct ip *);
241 int icmplen = ip->ip_len;
242 register int i;
243 struct in_ifaddr *ia;
244 void (*ctlfunc)(int, struct sockaddr *, void *);
245 int code;
246
247 /*
248 * Locate icmp structure in mbuf, and check
249 * that not corrupted and of at least minimum length.
250 */
251#ifdef ICMPPRINTFS
252 if (icmpprintfs) {
253 char buf[4 * sizeof "123"];
254 strcpy(buf, inet_ntoa(ip->ip_src));
255 printf("icmp_input from %s to %s, len %d\n",
256 buf, inet_ntoa(ip->ip_dst), icmplen);
257 }
258#endif
259 if (icmplen < ICMP_MINLEN) {
260 icmpstat.icps_tooshort++;
261 goto freeit;
262 }
263 i = hlen + min(icmplen, ICMP_ADVLENMIN);
264 if (m->m_len < i && (m = m_pullup(m, i)) == 0) {
265 icmpstat.icps_tooshort++;
266 return;
267 }
268 ip = mtod(m, struct ip *);
269 m->m_len -= hlen;
270 m->m_data += hlen;
271 icp = mtod(m, struct icmp *);
272 if (in_cksum(m, icmplen)) {
273 icmpstat.icps_checksum++;
274 goto freeit;
275 }
276 m->m_len += hlen;
277 m->m_data -= hlen;
278
279 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) {
280 /*
281 * Deliver very specific ICMP type only.
282 */
283 switch (icp->icmp_type) {
284 case ICMP_UNREACH:
285 case ICMP_TIMXCEED:
286 break;
287 default:
288 goto freeit;
289 }
290 }
291
292#ifdef ICMPPRINTFS
293 if (icmpprintfs)
294 printf("icmp_input, type %d code %d\n", icp->icmp_type,
295 icp->icmp_code);
296#endif
297
298 /*
299 * Message type specific processing.
300 */
301 if (icp->icmp_type > ICMP_MAXTYPE)
302 goto raw;
303 icmpstat.icps_inhist[icp->icmp_type]++;
304 code = icp->icmp_code;
305 switch (icp->icmp_type) {
306
307 case ICMP_UNREACH:
308 switch (code) {
309 case ICMP_UNREACH_NET:
310 case ICMP_UNREACH_HOST:
311 case ICMP_UNREACH_SRCFAIL:
312 case ICMP_UNREACH_NET_UNKNOWN:
313 case ICMP_UNREACH_HOST_UNKNOWN:
314 case ICMP_UNREACH_ISOLATED:
315 case ICMP_UNREACH_TOSNET:
316 case ICMP_UNREACH_TOSHOST:
317 case ICMP_UNREACH_HOST_PRECEDENCE:
318 case ICMP_UNREACH_PRECEDENCE_CUTOFF:
319 code = PRC_UNREACH_NET;
320 break;
321
322 case ICMP_UNREACH_NEEDFRAG:
323 code = PRC_MSGSIZE;
324 break;
325
326 /*
327 * RFC 1122, Sections 3.2.2.1 and 4.2.3.9.
328 * Treat subcodes 2,3 as immediate RST
329 */
330 case ICMP_UNREACH_PROTOCOL:
331 case ICMP_UNREACH_PORT:
332 code = PRC_UNREACH_PORT;
333 break;
334
335 case ICMP_UNREACH_NET_PROHIB:
336 case ICMP_UNREACH_HOST_PROHIB:
337 case ICMP_UNREACH_FILTER_PROHIB:
338 code = PRC_UNREACH_ADMIN_PROHIB;
339 break;
340
341 default:
342 goto badcode;
343 }
344 goto deliver;
345
346 case ICMP_TIMXCEED:
347 if (code > 1)
348 goto badcode;
349 code += PRC_TIMXCEED_INTRANS;
350 goto deliver;
351
352 case ICMP_PARAMPROB:
353 if (code > 1)
354 goto badcode;
355 code = PRC_PARAMPROB;
356 goto deliver;
357
358 case ICMP_SOURCEQUENCH:
359 if (code)
360 goto badcode;
361 code = PRC_QUENCH;
362 deliver:
363 /*
364 * Problem with datagram; advise higher level routines.
365 */
366 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
367 IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) {
368 icmpstat.icps_badlen++;
369 goto freeit;
370 }
371 icp->icmp_ip.ip_len = ntohs(icp->icmp_ip.ip_len);
372 /* Discard ICMP's in response to multicast packets */
373 if (IN_MULTICAST(ntohl(icp->icmp_ip.ip_dst.s_addr)))
374 goto badcode;
375#ifdef ICMPPRINTFS
376 if (icmpprintfs)
377 printf("deliver to protocol %d\n", icp->icmp_ip.ip_p);
378#endif
379 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
380#if 1
381 /*
382 * MTU discovery:
383 * If we got a needfrag and there is a host route to the
384 * original destination, and the MTU is not locked, then
385 * set the MTU in the route to the suggested new value
386 * (if given) and then notify as usual. The ULPs will
387 * notice that the MTU has changed and adapt accordingly.
388 * If no new MTU was suggested, then we guess a new one
389 * less than the current value. If the new MTU is
390 * unreasonably small (arbitrarily set at 296), then
391 * we reset the MTU to the interface value and enable the
392 * lock bit, indicating that we are no longer doing MTU
393 * discovery.
394 */
395 if (code == PRC_MSGSIZE) {
396 struct rtentry *rt;
397 int mtu;
398
399 rt = rtalloc1((struct sockaddr *)&icmpsrc, 0,
400 RTF_CLONING | RTF_PRCLONING);
401 if (rt && (rt->rt_flags & RTF_HOST)
402 && !(rt->rt_rmx.rmx_locks & RTV_MTU)) {
403 mtu = ntohs(icp->icmp_nextmtu);
404 if (!mtu)
405 mtu = ip_next_mtu(rt->rt_rmx.rmx_mtu,
406 1);
407#ifdef DEBUG_MTUDISC
408 printf("MTU for %s reduced to %d\n",
409 inet_ntoa(icmpsrc.sin_addr), mtu);
410#endif
411 if (mtu < 296) {
412 /* rt->rt_rmx.rmx_mtu =
413 rt->rt_ifp->if_mtu; */
414 rt->rt_rmx.rmx_locks |= RTV_MTU;
415 } else if (rt->rt_rmx.rmx_mtu > mtu) {
416 rt->rt_rmx.rmx_mtu = mtu;
417 }
418 }
419 if (rt)
420 RTFREE(rt);
421 }
422
423#endif
424 /*
425 * XXX if the packet contains [IPv4 AH TCP], we can't make a
426 * notification to TCP layer.
427 */
428 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput;
429 if (ctlfunc)
430 (*ctlfunc)(code, (struct sockaddr *)&icmpsrc,
431 (void *)&icp->icmp_ip);
432 break;
433
434 badcode:
435 icmpstat.icps_badcode++;
436 break;
437
438 case ICMP_ECHO:
439 if (!icmpbmcastecho
440 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
441 icmpstat.icps_bmcastecho++;
442 break;
443 }
444 icp->icmp_type = ICMP_ECHOREPLY;
445 if (badport_bandlim(BANDLIM_ICMP_ECHO) < 0)
446 goto freeit;
447 else
448 goto reflect;
449
450 case ICMP_TSTAMP:
451 if (!icmpbmcastecho
452 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
453 icmpstat.icps_bmcasttstamp++;
454 break;
455 }
456 if (icmplen < ICMP_TSLEN) {
457 icmpstat.icps_badlen++;
458 break;
459 }
460 icp->icmp_type = ICMP_TSTAMPREPLY;
461 icp->icmp_rtime = iptime();
462 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */
463 if (badport_bandlim(BANDLIM_ICMP_TSTAMP) < 0)
464 goto freeit;
465 else
466 goto reflect;
467
468 case ICMP_MASKREQ:
469 if (icmpmaskrepl == 0)
470 break;
471 /*
472 * We are not able to respond with all ones broadcast
473 * unless we receive it over a point-to-point interface.
474 */
475 if (icmplen < ICMP_MASKLEN)
476 break;
477 switch (ip->ip_dst.s_addr) {
478
479 case INADDR_BROADCAST:
480 case INADDR_ANY:
481 icmpdst.sin_addr = ip->ip_src;
482 break;
483
484 default:
485 icmpdst.sin_addr = ip->ip_dst;
486 }
487 ia = (struct in_ifaddr *)ifaof_ifpforaddr(
488 (struct sockaddr *)&icmpdst, m->m_pkthdr.rcvif);
489 if (ia == 0)
490 break;
491 if (ia->ia_ifp == 0)
492 break;
493 icp->icmp_type = ICMP_MASKREPLY;
494 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr;
495 if (ip->ip_src.s_addr == 0) {
496 if (ia->ia_ifp->if_flags & IFF_BROADCAST)
497 ip->ip_src = satosin(&ia->ia_broadaddr)->sin_addr;
498 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT)
499 ip->ip_src = satosin(&ia->ia_dstaddr)->sin_addr;
500 }
501reflect:
502 ip->ip_len += hlen; /* since ip_input deducts this */
503 icmpstat.icps_reflect++;
504 icmpstat.icps_outhist[icp->icmp_type]++;
505 icmp_reflect(m);
506 return;
507
508 case ICMP_REDIRECT:
509 if (log_redirect) {
510 u_long src, dst, gw;
511
512 src = ntohl(ip->ip_src.s_addr);
513 dst = ntohl(icp->icmp_ip.ip_dst.s_addr);
514 gw = ntohl(icp->icmp_gwaddr.s_addr);
515 printf("icmp redirect from %d.%d.%d.%d: "
516 "%d.%d.%d.%d => %d.%d.%d.%d\n",
517 (int)(src >> 24), (int)((src >> 16) & 0xff),
518 (int)((src >> 8) & 0xff), (int)(src & 0xff),
519 (int)(dst >> 24), (int)((dst >> 16) & 0xff),
520 (int)((dst >> 8) & 0xff), (int)(dst & 0xff),
521 (int)(gw >> 24), (int)((gw >> 16) & 0xff),
522 (int)((gw >> 8) & 0xff), (int)(gw & 0xff));
523 }
524 if (drop_redirect)
525 break;
526 if (code > 3)
527 goto badcode;
528 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
529 IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) {
530 icmpstat.icps_badlen++;
531 break;
532 }
533 /*
534 * Short circuit routing redirects to force
535 * immediate change in the kernel's routing
536 * tables. The message is also handed to anyone
537 * listening on a raw socket (e.g. the routing
538 * daemon for use in updating its tables).
539 */
540 icmpgw.sin_addr = ip->ip_src;
541 icmpdst.sin_addr = icp->icmp_gwaddr;
542#ifdef ICMPPRINTFS
543 if (icmpprintfs) {
544 char buf[4 * sizeof "123"];
545 strcpy(buf, inet_ntoa(icp->icmp_ip.ip_dst));
546
547 printf("redirect dst %s to %s\n",
548 buf, inet_ntoa(icp->icmp_gwaddr));
549 }
550#endif
551 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
552 rtredirect((struct sockaddr *)&icmpsrc,
553 (struct sockaddr *)&icmpdst,
554 (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST,
555 (struct sockaddr *)&icmpgw, (struct rtentry **)0);
556 pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&icmpsrc);
557#ifdef IPSEC
558 key_sa_routechange((struct sockaddr *)&icmpsrc);
559#endif
560 break;
561
562 /*
563 * No kernel processing for the following;
564 * just fall through to send to raw listener.
565 */
566 case ICMP_ECHOREPLY:
567 case ICMP_ROUTERADVERT:
568 case ICMP_ROUTERSOLICIT:
569 case ICMP_TSTAMPREPLY:
570 case ICMP_IREQREPLY:
571 case ICMP_MASKREPLY:
572 default:
573 break;
574 }
575
576raw:
577 rip_input(m, off);
578 return;
579
580freeit:
581 m_freem(m);
582}
583
584/*
585 * Reflect the ip packet back to the source
586 */
587static void
588icmp_reflect(m)
589 struct mbuf *m;
590{
591 struct ip *ip = mtod(m, struct ip *);
592 struct ifaddr *ifa;
593 struct in_ifaddr *ia;
594 struct in_addr t;
595 struct mbuf *opts = 0;
596 int optlen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip);
597 struct route *ro = NULL, rt;
598
599 if (!in_canforward(ip->ip_src) &&
600 ((ntohl(ip->ip_src.s_addr) & IN_CLASSA_NET) !=
601 (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))) {
602 m_freem(m); /* Bad return address */
603 icmpstat.icps_badaddr++;
604 goto done; /* Ip_output() will check for broadcast */
605 }
606 t = ip->ip_dst;
607 ip->ip_dst = ip->ip_src;
608 ro = &rt;
609 bzero(ro, sizeof(*ro));
610 /*
611 * If the incoming packet was addressed directly to us,
612 * use dst as the src for the reply. Otherwise (broadcast
613 * or anonymous), use the address which corresponds
614 * to the incoming interface.
615 */
616 LIST_FOREACH(ia, INADDR_HASH(t.s_addr), ia_hash)
617 if (t.s_addr == IA_SIN(ia)->sin_addr.s_addr)
618 goto match;
619 if (m->m_pkthdr.rcvif != NULL &&
620 m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
621 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrhead, ifa_link) {
622 if (ifa->ifa_addr->sa_family != AF_INET)
623 continue;
624 ia = ifatoia(ifa);
625 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
626 t.s_addr)
627 goto match;
628 }
629 }
630 ia = ip_rtaddr(ip->ip_dst, ro);
631 /* We need a route to do anything useful. */
632 if (ia == NULL) {
633 m_freem(m);
634 icmpstat.icps_noroute++;
635 goto done;
636 }
637match:
638 t = IA_SIN(ia)->sin_addr;
639 ip->ip_src = t;
640 ip->ip_ttl = ip_defttl;
641
642 if (optlen > 0) {
643 register u_char *cp;
644 int opt, cnt;
645 u_int len;
646
647 /*
648 * Retrieve any source routing from the incoming packet;
649 * add on any record-route or timestamp options.
650 */
651 cp = (u_char *) (ip + 1);
652 if ((opts = ip_srcroute()) == 0 &&
653 (opts = m_gethdr(M_DONTWAIT, MT_HEADER))) {
654 opts->m_len = sizeof(struct in_addr);
655 mtod(opts, struct in_addr *)->s_addr = 0;
656 }
657 if (opts) {
658#ifdef ICMPPRINTFS
659 if (icmpprintfs)
660 printf("icmp_reflect optlen %d rt %d => ",
661 optlen, opts->m_len);
662#endif
663 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) {
664 opt = cp[IPOPT_OPTVAL];
665 if (opt == IPOPT_EOL)
666 break;
667 if (opt == IPOPT_NOP)
668 len = 1;
669 else {
670 if (cnt < IPOPT_OLEN + sizeof(*cp))
671 break;
672 len = cp[IPOPT_OLEN];
673 if (len < IPOPT_OLEN + sizeof(*cp) ||
674 len > cnt)
675 break;
676 }
677 /*
678 * Should check for overflow, but it "can't happen"
679 */
680 if (opt == IPOPT_RR || opt == IPOPT_TS ||
681 opt == IPOPT_SECURITY) {
682 bcopy((caddr_t)cp,
683 mtod(opts, caddr_t) + opts->m_len, len);
684 opts->m_len += len;
685 }
686 }
687 /* Terminate & pad, if necessary */
688 cnt = opts->m_len % 4;
689 if (cnt) {
690 for (; cnt < 4; cnt++) {
691 *(mtod(opts, caddr_t) + opts->m_len) =
692 IPOPT_EOL;
693 opts->m_len++;
694 }
695 }
696#ifdef ICMPPRINTFS
697 if (icmpprintfs)
698 printf("%d\n", opts->m_len);
699#endif
700 }
701 /*
702 * Now strip out original options by copying rest of first
703 * mbuf's data back, and adjust the IP length.
704 */
705 ip->ip_len -= optlen;
706 ip->ip_vhl = IP_VHL_BORING;
707 m->m_len -= optlen;
708 if (m->m_flags & M_PKTHDR)
709 m->m_pkthdr.len -= optlen;
710 optlen += sizeof(struct ip);
711 bcopy((caddr_t)ip + optlen, (caddr_t)(ip + 1),
712 (unsigned)(m->m_len - sizeof(struct ip)));
713 }
714 m->m_flags &= ~(M_BCAST|M_MCAST);
715 icmp_send(m, opts, ro);
716done:
717 if (opts)
718 (void)m_free(opts);
719 if (ro && ro->ro_rt)
720 RTFREE(ro->ro_rt);
721}
722
723/*
724 * Send an icmp packet back to the ip level,
725 * after supplying a checksum.
726 */
727static void
728icmp_send(m, opts, rt)
729 register struct mbuf *m;
730 struct mbuf *opts;
731 struct route *rt;
732{
733 register struct ip *ip = mtod(m, struct ip *);
734 register int hlen;
735 register struct icmp *icp;
736
737 hlen = IP_VHL_HL(ip->ip_vhl) << 2;
738 m->m_data += hlen;
739 m->m_len -= hlen;
740 icp = mtod(m, struct icmp *);
741 icp->icmp_cksum = 0;
742 icp->icmp_cksum = in_cksum(m, ip->ip_len - hlen);
743 m->m_data -= hlen;
744 m->m_len += hlen;
745 m->m_pkthdr.rcvif = (struct ifnet *)0;
746#ifdef ICMPPRINTFS
747 if (icmpprintfs) {
748 char buf[4 * sizeof "123"];
749 strcpy(buf, inet_ntoa(ip->ip_dst));
750 printf("icmp_send dst %s src %s\n",
751 buf, inet_ntoa(ip->ip_src));
752 }
753#endif
754 (void) ip_output(m, opts, rt, 0, NULL, NULL);
755}
756
757n_time
758iptime()
759{
760 struct timeval atv;
761 u_long t;
762
763 getmicrotime(&atv);
764 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000;
765 return (htonl(t));
766}
767
768#if 1
769/*
770 * Return the next larger or smaller MTU plateau (table from RFC 1191)
771 * given current value MTU. If DIR is less than zero, a larger plateau
772 * is returned; otherwise, a smaller value is returned.
773 */
774static int
775ip_next_mtu(mtu, dir)
776 int mtu;
777 int dir;
778{
779 static int mtutab[] = {
780 65535, 32000, 17914, 8166, 4352, 2002, 1492, 1006, 508, 296,
781 68, 0
782 };
783 int i;
784
785 for (i = 0; i < (sizeof mtutab) / (sizeof mtutab[0]); i++) {
786 if (mtu >= mtutab[i])
787 break;
788 }
789
790 if (dir < 0) {
791 if (i == 0) {
792 return 0;
793 } else {
794 return mtutab[i - 1];
795 }
796 } else {
797 if (mtutab[i] == 0) {
798 return 0;
799 } else if(mtu > mtutab[i]) {
800 return mtutab[i];
801 } else {
802 return mtutab[i + 1];
803 }
804 }
805}
806#endif
807
808
809/*
810 * badport_bandlim() - check for ICMP bandwidth limit
811 *
812 * Return 0 if it is ok to send an ICMP error response, -1 if we have
813 * hit our bandwidth limit and it is not ok.
814 *
815 * If icmplim is <= 0, the feature is disabled and 0 is returned.
816 *
817 * For now we separate the TCP and UDP subsystems w/ different 'which'
818 * values. We may eventually remove this separation (and simplify the
819 * code further).
820 *
821 * Note that the printing of the error message is delayed so we can
822 * properly print the icmp error rate that the system was trying to do
823 * (i.e. 22000/100 pps, etc...). This can cause long delays in printing
824 * the 'final' error, but it doesn't make sense to solve the printing
825 * delay with more complex code.
826 */
827
828int
829badport_bandlim(int which)
830{
831 static int lticks[BANDLIM_MAX + 1];
832 static int lpackets[BANDLIM_MAX + 1];
833 int dticks;
834 const char *bandlimittype[] = {
835 "Limiting icmp unreach response",
836 "Limiting icmp ping response",
837 "Limiting icmp tstamp response",
838 "Limiting closed port RST response",
839 "Limiting open port RST response"
840 };
841
842 /*
843 * Return ok status if feature disabled or argument out of
844 * ranage.
845 */
846
847 if (icmplim <= 0 || which > BANDLIM_MAX || which < 0)
848 return(0);
849 dticks = ticks - lticks[which];
850
851 /*
852 * reset stats when cumulative dt exceeds one second.
853 */
854
855 if ((unsigned int)dticks > hz) {
856 if (lpackets[which] > icmplim && icmplim_output) {
857 printf("%s from %d to %d packets per second\n",
858 bandlimittype[which],
859 lpackets[which],
860 icmplim
861 );
862 }
863 lticks[which] = ticks;
864 lpackets[which] = 0;
865 }
866
867 /*
868 * bump packet count
869 */
870
871 if (++lpackets[which] > icmplim) {
872 return(-1);
873 }
874 return(0);
875}
876
2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 * @(#)ip_icmp.c 8.2 (Berkeley) 1/4/94
34 * $FreeBSD: head/sys/netinet/ip_icmp.c 105194 2002-10-16 01:54:46Z sam $
35 */
36
37#include "opt_ipsec.h"
38#include "opt_mac.h"
39
40#include <sys/param.h>
41#include <sys/systm.h>
42#include <sys/mac.h>
43#include <sys/mbuf.h>
44#include <sys/protosw.h>
45#include <sys/socket.h>
46#include <sys/time.h>
47#include <sys/kernel.h>
48#include <sys/sysctl.h>
49
50#include <net/if.h>
51#include <net/if_types.h>
52#include <net/route.h>
53
54#define _IP_VHL
55#include <netinet/in.h>
56#include <netinet/in_systm.h>
57#include <netinet/in_var.h>
58#include <netinet/ip.h>
59#include <netinet/ip_icmp.h>
60#include <netinet/ip_var.h>
61#include <netinet/icmp_var.h>
62
63#ifdef IPSEC
64#include <netinet6/ipsec.h>
65#include <netkey/key.h>
66#endif
67
68#include <machine/in_cksum.h>
69
70/*
71 * ICMP routines: error generation, receive packet processing, and
72 * routines to turnaround packets back to the originator, and
73 * host table maintenance routines.
74 */
75
76static struct icmpstat icmpstat;
77SYSCTL_STRUCT(_net_inet_icmp, ICMPCTL_STATS, stats, CTLFLAG_RW,
78 &icmpstat, icmpstat, "");
79
80static int icmpmaskrepl = 0;
81SYSCTL_INT(_net_inet_icmp, ICMPCTL_MASKREPL, maskrepl, CTLFLAG_RW,
82 &icmpmaskrepl, 0, "");
83
84static int drop_redirect = 0;
85SYSCTL_INT(_net_inet_icmp, OID_AUTO, drop_redirect, CTLFLAG_RW,
86 &drop_redirect, 0, "");
87
88static int log_redirect = 0;
89SYSCTL_INT(_net_inet_icmp, OID_AUTO, log_redirect, CTLFLAG_RW,
90 &log_redirect, 0, "");
91
92static int icmplim = 200;
93SYSCTL_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RW,
94 &icmplim, 0, "");
95
96static int icmplim_output = 1;
97SYSCTL_INT(_net_inet_icmp, OID_AUTO, icmplim_output, CTLFLAG_RW,
98 &icmplim_output, 0, "");
99
100/*
101 * ICMP broadcast echo sysctl
102 */
103
104static int icmpbmcastecho = 0;
105SYSCTL_INT(_net_inet_icmp, OID_AUTO, bmcastecho, CTLFLAG_RW,
106 &icmpbmcastecho, 0, "");
107
108
109#ifdef ICMPPRINTFS
110int icmpprintfs = 0;
111#endif
112
113static void icmp_reflect(struct mbuf *);
114static void icmp_send(struct mbuf *, struct mbuf *, struct route *);
115static int ip_next_mtu(int, int);
116
117extern struct protosw inetsw[];
118
119/*
120 * Generate an error packet of type error
121 * in response to bad packet ip.
122 */
123void
124icmp_error(n, type, code, dest, destifp)
125 struct mbuf *n;
126 int type, code;
127 n_long dest;
128 struct ifnet *destifp;
129{
130 register struct ip *oip = mtod(n, struct ip *), *nip;
131 register unsigned oiplen = IP_VHL_HL(oip->ip_vhl) << 2;
132 register struct icmp *icp;
133 register struct mbuf *m;
134 unsigned icmplen;
135
136#ifdef ICMPPRINTFS
137 if (icmpprintfs)
138 printf("icmp_error(%p, %x, %d)\n", oip, type, code);
139#endif
140 if (type != ICMP_REDIRECT)
141 icmpstat.icps_error++;
142 /*
143 * Don't send error if not the first fragment of message.
144 * Don't error if the old packet protocol was ICMP
145 * error message, only known informational types.
146 */
147 if (oip->ip_off &~ (IP_MF|IP_DF))
148 goto freeit;
149 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT &&
150 n->m_len >= oiplen + ICMP_MINLEN &&
151 !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiplen))->icmp_type)) {
152 icmpstat.icps_oldicmp++;
153 goto freeit;
154 }
155 /* Don't send error in response to a multicast or broadcast packet */
156 if (n->m_flags & (M_BCAST|M_MCAST))
157 goto freeit;
158 /*
159 * First, formulate icmp message
160 */
161 m = m_gethdr(M_DONTWAIT, MT_HEADER);
162 if (m == NULL)
163 goto freeit;
164#ifdef MAC
165 mac_create_mbuf_netlayer(n, m);
166#endif
167 icmplen = min(oiplen + 8, oip->ip_len);
168 if (icmplen < sizeof(struct ip))
169 panic("icmp_error: bad length");
170 m->m_len = icmplen + ICMP_MINLEN;
171 MH_ALIGN(m, m->m_len);
172 icp = mtod(m, struct icmp *);
173 if ((u_int)type > ICMP_MAXTYPE)
174 panic("icmp_error");
175 icmpstat.icps_outhist[type]++;
176 icp->icmp_type = type;
177 if (type == ICMP_REDIRECT)
178 icp->icmp_gwaddr.s_addr = dest;
179 else {
180 icp->icmp_void = 0;
181 /*
182 * The following assignments assume an overlay with the
183 * zeroed icmp_void field.
184 */
185 if (type == ICMP_PARAMPROB) {
186 icp->icmp_pptr = code;
187 code = 0;
188 } else if (type == ICMP_UNREACH &&
189 code == ICMP_UNREACH_NEEDFRAG && destifp) {
190 icp->icmp_nextmtu = htons(destifp->if_mtu);
191 }
192 }
193
194 icp->icmp_code = code;
195 m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip);
196 nip = &icp->icmp_ip;
197
198 /*
199 * Convert fields to network representation.
200 */
201 nip->ip_len = htons(nip->ip_len);
202 nip->ip_off = htons(nip->ip_off);
203
204 /*
205 * Now, copy old ip header (without options)
206 * in front of icmp message.
207 */
208 if (m->m_data - sizeof(struct ip) < m->m_pktdat)
209 panic("icmp len");
210 m->m_data -= sizeof(struct ip);
211 m->m_len += sizeof(struct ip);
212 m->m_pkthdr.len = m->m_len;
213 m->m_pkthdr.rcvif = n->m_pkthdr.rcvif;
214 nip = mtod(m, struct ip *);
215 bcopy((caddr_t)oip, (caddr_t)nip, sizeof(struct ip));
216 nip->ip_len = m->m_len;
217 nip->ip_vhl = IP_VHL_BORING;
218 nip->ip_p = IPPROTO_ICMP;
219 nip->ip_tos = 0;
220 icmp_reflect(m);
221
222freeit:
223 m_freem(n);
224}
225
226static struct sockaddr_in icmpsrc = { sizeof (struct sockaddr_in), AF_INET };
227static struct sockaddr_in icmpdst = { sizeof (struct sockaddr_in), AF_INET };
228static struct sockaddr_in icmpgw = { sizeof (struct sockaddr_in), AF_INET };
229
230/*
231 * Process a received ICMP message.
232 */
233void
234icmp_input(m, off)
235 register struct mbuf *m;
236 int off;
237{
238 int hlen = off;
239 register struct icmp *icp;
240 register struct ip *ip = mtod(m, struct ip *);
241 int icmplen = ip->ip_len;
242 register int i;
243 struct in_ifaddr *ia;
244 void (*ctlfunc)(int, struct sockaddr *, void *);
245 int code;
246
247 /*
248 * Locate icmp structure in mbuf, and check
249 * that not corrupted and of at least minimum length.
250 */
251#ifdef ICMPPRINTFS
252 if (icmpprintfs) {
253 char buf[4 * sizeof "123"];
254 strcpy(buf, inet_ntoa(ip->ip_src));
255 printf("icmp_input from %s to %s, len %d\n",
256 buf, inet_ntoa(ip->ip_dst), icmplen);
257 }
258#endif
259 if (icmplen < ICMP_MINLEN) {
260 icmpstat.icps_tooshort++;
261 goto freeit;
262 }
263 i = hlen + min(icmplen, ICMP_ADVLENMIN);
264 if (m->m_len < i && (m = m_pullup(m, i)) == 0) {
265 icmpstat.icps_tooshort++;
266 return;
267 }
268 ip = mtod(m, struct ip *);
269 m->m_len -= hlen;
270 m->m_data += hlen;
271 icp = mtod(m, struct icmp *);
272 if (in_cksum(m, icmplen)) {
273 icmpstat.icps_checksum++;
274 goto freeit;
275 }
276 m->m_len += hlen;
277 m->m_data -= hlen;
278
279 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) {
280 /*
281 * Deliver very specific ICMP type only.
282 */
283 switch (icp->icmp_type) {
284 case ICMP_UNREACH:
285 case ICMP_TIMXCEED:
286 break;
287 default:
288 goto freeit;
289 }
290 }
291
292#ifdef ICMPPRINTFS
293 if (icmpprintfs)
294 printf("icmp_input, type %d code %d\n", icp->icmp_type,
295 icp->icmp_code);
296#endif
297
298 /*
299 * Message type specific processing.
300 */
301 if (icp->icmp_type > ICMP_MAXTYPE)
302 goto raw;
303 icmpstat.icps_inhist[icp->icmp_type]++;
304 code = icp->icmp_code;
305 switch (icp->icmp_type) {
306
307 case ICMP_UNREACH:
308 switch (code) {
309 case ICMP_UNREACH_NET:
310 case ICMP_UNREACH_HOST:
311 case ICMP_UNREACH_SRCFAIL:
312 case ICMP_UNREACH_NET_UNKNOWN:
313 case ICMP_UNREACH_HOST_UNKNOWN:
314 case ICMP_UNREACH_ISOLATED:
315 case ICMP_UNREACH_TOSNET:
316 case ICMP_UNREACH_TOSHOST:
317 case ICMP_UNREACH_HOST_PRECEDENCE:
318 case ICMP_UNREACH_PRECEDENCE_CUTOFF:
319 code = PRC_UNREACH_NET;
320 break;
321
322 case ICMP_UNREACH_NEEDFRAG:
323 code = PRC_MSGSIZE;
324 break;
325
326 /*
327 * RFC 1122, Sections 3.2.2.1 and 4.2.3.9.
328 * Treat subcodes 2,3 as immediate RST
329 */
330 case ICMP_UNREACH_PROTOCOL:
331 case ICMP_UNREACH_PORT:
332 code = PRC_UNREACH_PORT;
333 break;
334
335 case ICMP_UNREACH_NET_PROHIB:
336 case ICMP_UNREACH_HOST_PROHIB:
337 case ICMP_UNREACH_FILTER_PROHIB:
338 code = PRC_UNREACH_ADMIN_PROHIB;
339 break;
340
341 default:
342 goto badcode;
343 }
344 goto deliver;
345
346 case ICMP_TIMXCEED:
347 if (code > 1)
348 goto badcode;
349 code += PRC_TIMXCEED_INTRANS;
350 goto deliver;
351
352 case ICMP_PARAMPROB:
353 if (code > 1)
354 goto badcode;
355 code = PRC_PARAMPROB;
356 goto deliver;
357
358 case ICMP_SOURCEQUENCH:
359 if (code)
360 goto badcode;
361 code = PRC_QUENCH;
362 deliver:
363 /*
364 * Problem with datagram; advise higher level routines.
365 */
366 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
367 IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) {
368 icmpstat.icps_badlen++;
369 goto freeit;
370 }
371 icp->icmp_ip.ip_len = ntohs(icp->icmp_ip.ip_len);
372 /* Discard ICMP's in response to multicast packets */
373 if (IN_MULTICAST(ntohl(icp->icmp_ip.ip_dst.s_addr)))
374 goto badcode;
375#ifdef ICMPPRINTFS
376 if (icmpprintfs)
377 printf("deliver to protocol %d\n", icp->icmp_ip.ip_p);
378#endif
379 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
380#if 1
381 /*
382 * MTU discovery:
383 * If we got a needfrag and there is a host route to the
384 * original destination, and the MTU is not locked, then
385 * set the MTU in the route to the suggested new value
386 * (if given) and then notify as usual. The ULPs will
387 * notice that the MTU has changed and adapt accordingly.
388 * If no new MTU was suggested, then we guess a new one
389 * less than the current value. If the new MTU is
390 * unreasonably small (arbitrarily set at 296), then
391 * we reset the MTU to the interface value and enable the
392 * lock bit, indicating that we are no longer doing MTU
393 * discovery.
394 */
395 if (code == PRC_MSGSIZE) {
396 struct rtentry *rt;
397 int mtu;
398
399 rt = rtalloc1((struct sockaddr *)&icmpsrc, 0,
400 RTF_CLONING | RTF_PRCLONING);
401 if (rt && (rt->rt_flags & RTF_HOST)
402 && !(rt->rt_rmx.rmx_locks & RTV_MTU)) {
403 mtu = ntohs(icp->icmp_nextmtu);
404 if (!mtu)
405 mtu = ip_next_mtu(rt->rt_rmx.rmx_mtu,
406 1);
407#ifdef DEBUG_MTUDISC
408 printf("MTU for %s reduced to %d\n",
409 inet_ntoa(icmpsrc.sin_addr), mtu);
410#endif
411 if (mtu < 296) {
412 /* rt->rt_rmx.rmx_mtu =
413 rt->rt_ifp->if_mtu; */
414 rt->rt_rmx.rmx_locks |= RTV_MTU;
415 } else if (rt->rt_rmx.rmx_mtu > mtu) {
416 rt->rt_rmx.rmx_mtu = mtu;
417 }
418 }
419 if (rt)
420 RTFREE(rt);
421 }
422
423#endif
424 /*
425 * XXX if the packet contains [IPv4 AH TCP], we can't make a
426 * notification to TCP layer.
427 */
428 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput;
429 if (ctlfunc)
430 (*ctlfunc)(code, (struct sockaddr *)&icmpsrc,
431 (void *)&icp->icmp_ip);
432 break;
433
434 badcode:
435 icmpstat.icps_badcode++;
436 break;
437
438 case ICMP_ECHO:
439 if (!icmpbmcastecho
440 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
441 icmpstat.icps_bmcastecho++;
442 break;
443 }
444 icp->icmp_type = ICMP_ECHOREPLY;
445 if (badport_bandlim(BANDLIM_ICMP_ECHO) < 0)
446 goto freeit;
447 else
448 goto reflect;
449
450 case ICMP_TSTAMP:
451 if (!icmpbmcastecho
452 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
453 icmpstat.icps_bmcasttstamp++;
454 break;
455 }
456 if (icmplen < ICMP_TSLEN) {
457 icmpstat.icps_badlen++;
458 break;
459 }
460 icp->icmp_type = ICMP_TSTAMPREPLY;
461 icp->icmp_rtime = iptime();
462 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */
463 if (badport_bandlim(BANDLIM_ICMP_TSTAMP) < 0)
464 goto freeit;
465 else
466 goto reflect;
467
468 case ICMP_MASKREQ:
469 if (icmpmaskrepl == 0)
470 break;
471 /*
472 * We are not able to respond with all ones broadcast
473 * unless we receive it over a point-to-point interface.
474 */
475 if (icmplen < ICMP_MASKLEN)
476 break;
477 switch (ip->ip_dst.s_addr) {
478
479 case INADDR_BROADCAST:
480 case INADDR_ANY:
481 icmpdst.sin_addr = ip->ip_src;
482 break;
483
484 default:
485 icmpdst.sin_addr = ip->ip_dst;
486 }
487 ia = (struct in_ifaddr *)ifaof_ifpforaddr(
488 (struct sockaddr *)&icmpdst, m->m_pkthdr.rcvif);
489 if (ia == 0)
490 break;
491 if (ia->ia_ifp == 0)
492 break;
493 icp->icmp_type = ICMP_MASKREPLY;
494 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr;
495 if (ip->ip_src.s_addr == 0) {
496 if (ia->ia_ifp->if_flags & IFF_BROADCAST)
497 ip->ip_src = satosin(&ia->ia_broadaddr)->sin_addr;
498 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT)
499 ip->ip_src = satosin(&ia->ia_dstaddr)->sin_addr;
500 }
501reflect:
502 ip->ip_len += hlen; /* since ip_input deducts this */
503 icmpstat.icps_reflect++;
504 icmpstat.icps_outhist[icp->icmp_type]++;
505 icmp_reflect(m);
506 return;
507
508 case ICMP_REDIRECT:
509 if (log_redirect) {
510 u_long src, dst, gw;
511
512 src = ntohl(ip->ip_src.s_addr);
513 dst = ntohl(icp->icmp_ip.ip_dst.s_addr);
514 gw = ntohl(icp->icmp_gwaddr.s_addr);
515 printf("icmp redirect from %d.%d.%d.%d: "
516 "%d.%d.%d.%d => %d.%d.%d.%d\n",
517 (int)(src >> 24), (int)((src >> 16) & 0xff),
518 (int)((src >> 8) & 0xff), (int)(src & 0xff),
519 (int)(dst >> 24), (int)((dst >> 16) & 0xff),
520 (int)((dst >> 8) & 0xff), (int)(dst & 0xff),
521 (int)(gw >> 24), (int)((gw >> 16) & 0xff),
522 (int)((gw >> 8) & 0xff), (int)(gw & 0xff));
523 }
524 if (drop_redirect)
525 break;
526 if (code > 3)
527 goto badcode;
528 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
529 IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) {
530 icmpstat.icps_badlen++;
531 break;
532 }
533 /*
534 * Short circuit routing redirects to force
535 * immediate change in the kernel's routing
536 * tables. The message is also handed to anyone
537 * listening on a raw socket (e.g. the routing
538 * daemon for use in updating its tables).
539 */
540 icmpgw.sin_addr = ip->ip_src;
541 icmpdst.sin_addr = icp->icmp_gwaddr;
542#ifdef ICMPPRINTFS
543 if (icmpprintfs) {
544 char buf[4 * sizeof "123"];
545 strcpy(buf, inet_ntoa(icp->icmp_ip.ip_dst));
546
547 printf("redirect dst %s to %s\n",
548 buf, inet_ntoa(icp->icmp_gwaddr));
549 }
550#endif
551 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
552 rtredirect((struct sockaddr *)&icmpsrc,
553 (struct sockaddr *)&icmpdst,
554 (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST,
555 (struct sockaddr *)&icmpgw, (struct rtentry **)0);
556 pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&icmpsrc);
557#ifdef IPSEC
558 key_sa_routechange((struct sockaddr *)&icmpsrc);
559#endif
560 break;
561
562 /*
563 * No kernel processing for the following;
564 * just fall through to send to raw listener.
565 */
566 case ICMP_ECHOREPLY:
567 case ICMP_ROUTERADVERT:
568 case ICMP_ROUTERSOLICIT:
569 case ICMP_TSTAMPREPLY:
570 case ICMP_IREQREPLY:
571 case ICMP_MASKREPLY:
572 default:
573 break;
574 }
575
576raw:
577 rip_input(m, off);
578 return;
579
580freeit:
581 m_freem(m);
582}
583
584/*
585 * Reflect the ip packet back to the source
586 */
587static void
588icmp_reflect(m)
589 struct mbuf *m;
590{
591 struct ip *ip = mtod(m, struct ip *);
592 struct ifaddr *ifa;
593 struct in_ifaddr *ia;
594 struct in_addr t;
595 struct mbuf *opts = 0;
596 int optlen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip);
597 struct route *ro = NULL, rt;
598
599 if (!in_canforward(ip->ip_src) &&
600 ((ntohl(ip->ip_src.s_addr) & IN_CLASSA_NET) !=
601 (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))) {
602 m_freem(m); /* Bad return address */
603 icmpstat.icps_badaddr++;
604 goto done; /* Ip_output() will check for broadcast */
605 }
606 t = ip->ip_dst;
607 ip->ip_dst = ip->ip_src;
608 ro = &rt;
609 bzero(ro, sizeof(*ro));
610 /*
611 * If the incoming packet was addressed directly to us,
612 * use dst as the src for the reply. Otherwise (broadcast
613 * or anonymous), use the address which corresponds
614 * to the incoming interface.
615 */
616 LIST_FOREACH(ia, INADDR_HASH(t.s_addr), ia_hash)
617 if (t.s_addr == IA_SIN(ia)->sin_addr.s_addr)
618 goto match;
619 if (m->m_pkthdr.rcvif != NULL &&
620 m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
621 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrhead, ifa_link) {
622 if (ifa->ifa_addr->sa_family != AF_INET)
623 continue;
624 ia = ifatoia(ifa);
625 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
626 t.s_addr)
627 goto match;
628 }
629 }
630 ia = ip_rtaddr(ip->ip_dst, ro);
631 /* We need a route to do anything useful. */
632 if (ia == NULL) {
633 m_freem(m);
634 icmpstat.icps_noroute++;
635 goto done;
636 }
637match:
638 t = IA_SIN(ia)->sin_addr;
639 ip->ip_src = t;
640 ip->ip_ttl = ip_defttl;
641
642 if (optlen > 0) {
643 register u_char *cp;
644 int opt, cnt;
645 u_int len;
646
647 /*
648 * Retrieve any source routing from the incoming packet;
649 * add on any record-route or timestamp options.
650 */
651 cp = (u_char *) (ip + 1);
652 if ((opts = ip_srcroute()) == 0 &&
653 (opts = m_gethdr(M_DONTWAIT, MT_HEADER))) {
654 opts->m_len = sizeof(struct in_addr);
655 mtod(opts, struct in_addr *)->s_addr = 0;
656 }
657 if (opts) {
658#ifdef ICMPPRINTFS
659 if (icmpprintfs)
660 printf("icmp_reflect optlen %d rt %d => ",
661 optlen, opts->m_len);
662#endif
663 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) {
664 opt = cp[IPOPT_OPTVAL];
665 if (opt == IPOPT_EOL)
666 break;
667 if (opt == IPOPT_NOP)
668 len = 1;
669 else {
670 if (cnt < IPOPT_OLEN + sizeof(*cp))
671 break;
672 len = cp[IPOPT_OLEN];
673 if (len < IPOPT_OLEN + sizeof(*cp) ||
674 len > cnt)
675 break;
676 }
677 /*
678 * Should check for overflow, but it "can't happen"
679 */
680 if (opt == IPOPT_RR || opt == IPOPT_TS ||
681 opt == IPOPT_SECURITY) {
682 bcopy((caddr_t)cp,
683 mtod(opts, caddr_t) + opts->m_len, len);
684 opts->m_len += len;
685 }
686 }
687 /* Terminate & pad, if necessary */
688 cnt = opts->m_len % 4;
689 if (cnt) {
690 for (; cnt < 4; cnt++) {
691 *(mtod(opts, caddr_t) + opts->m_len) =
692 IPOPT_EOL;
693 opts->m_len++;
694 }
695 }
696#ifdef ICMPPRINTFS
697 if (icmpprintfs)
698 printf("%d\n", opts->m_len);
699#endif
700 }
701 /*
702 * Now strip out original options by copying rest of first
703 * mbuf's data back, and adjust the IP length.
704 */
705 ip->ip_len -= optlen;
706 ip->ip_vhl = IP_VHL_BORING;
707 m->m_len -= optlen;
708 if (m->m_flags & M_PKTHDR)
709 m->m_pkthdr.len -= optlen;
710 optlen += sizeof(struct ip);
711 bcopy((caddr_t)ip + optlen, (caddr_t)(ip + 1),
712 (unsigned)(m->m_len - sizeof(struct ip)));
713 }
714 m->m_flags &= ~(M_BCAST|M_MCAST);
715 icmp_send(m, opts, ro);
716done:
717 if (opts)
718 (void)m_free(opts);
719 if (ro && ro->ro_rt)
720 RTFREE(ro->ro_rt);
721}
722
723/*
724 * Send an icmp packet back to the ip level,
725 * after supplying a checksum.
726 */
727static void
728icmp_send(m, opts, rt)
729 register struct mbuf *m;
730 struct mbuf *opts;
731 struct route *rt;
732{
733 register struct ip *ip = mtod(m, struct ip *);
734 register int hlen;
735 register struct icmp *icp;
736
737 hlen = IP_VHL_HL(ip->ip_vhl) << 2;
738 m->m_data += hlen;
739 m->m_len -= hlen;
740 icp = mtod(m, struct icmp *);
741 icp->icmp_cksum = 0;
742 icp->icmp_cksum = in_cksum(m, ip->ip_len - hlen);
743 m->m_data -= hlen;
744 m->m_len += hlen;
745 m->m_pkthdr.rcvif = (struct ifnet *)0;
746#ifdef ICMPPRINTFS
747 if (icmpprintfs) {
748 char buf[4 * sizeof "123"];
749 strcpy(buf, inet_ntoa(ip->ip_dst));
750 printf("icmp_send dst %s src %s\n",
751 buf, inet_ntoa(ip->ip_src));
752 }
753#endif
754 (void) ip_output(m, opts, rt, 0, NULL, NULL);
755}
756
757n_time
758iptime()
759{
760 struct timeval atv;
761 u_long t;
762
763 getmicrotime(&atv);
764 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000;
765 return (htonl(t));
766}
767
768#if 1
769/*
770 * Return the next larger or smaller MTU plateau (table from RFC 1191)
771 * given current value MTU. If DIR is less than zero, a larger plateau
772 * is returned; otherwise, a smaller value is returned.
773 */
774static int
775ip_next_mtu(mtu, dir)
776 int mtu;
777 int dir;
778{
779 static int mtutab[] = {
780 65535, 32000, 17914, 8166, 4352, 2002, 1492, 1006, 508, 296,
781 68, 0
782 };
783 int i;
784
785 for (i = 0; i < (sizeof mtutab) / (sizeof mtutab[0]); i++) {
786 if (mtu >= mtutab[i])
787 break;
788 }
789
790 if (dir < 0) {
791 if (i == 0) {
792 return 0;
793 } else {
794 return mtutab[i - 1];
795 }
796 } else {
797 if (mtutab[i] == 0) {
798 return 0;
799 } else if(mtu > mtutab[i]) {
800 return mtutab[i];
801 } else {
802 return mtutab[i + 1];
803 }
804 }
805}
806#endif
807
808
809/*
810 * badport_bandlim() - check for ICMP bandwidth limit
811 *
812 * Return 0 if it is ok to send an ICMP error response, -1 if we have
813 * hit our bandwidth limit and it is not ok.
814 *
815 * If icmplim is <= 0, the feature is disabled and 0 is returned.
816 *
817 * For now we separate the TCP and UDP subsystems w/ different 'which'
818 * values. We may eventually remove this separation (and simplify the
819 * code further).
820 *
821 * Note that the printing of the error message is delayed so we can
822 * properly print the icmp error rate that the system was trying to do
823 * (i.e. 22000/100 pps, etc...). This can cause long delays in printing
824 * the 'final' error, but it doesn't make sense to solve the printing
825 * delay with more complex code.
826 */
827
828int
829badport_bandlim(int which)
830{
831 static int lticks[BANDLIM_MAX + 1];
832 static int lpackets[BANDLIM_MAX + 1];
833 int dticks;
834 const char *bandlimittype[] = {
835 "Limiting icmp unreach response",
836 "Limiting icmp ping response",
837 "Limiting icmp tstamp response",
838 "Limiting closed port RST response",
839 "Limiting open port RST response"
840 };
841
842 /*
843 * Return ok status if feature disabled or argument out of
844 * ranage.
845 */
846
847 if (icmplim <= 0 || which > BANDLIM_MAX || which < 0)
848 return(0);
849 dticks = ticks - lticks[which];
850
851 /*
852 * reset stats when cumulative dt exceeds one second.
853 */
854
855 if ((unsigned int)dticks > hz) {
856 if (lpackets[which] > icmplim && icmplim_output) {
857 printf("%s from %d to %d packets per second\n",
858 bandlimittype[which],
859 lpackets[which],
860 icmplim
861 );
862 }
863 lticks[which] = ticks;
864 lpackets[which] = 0;
865 }
866
867 /*
868 * bump packet count
869 */
870
871 if (++lpackets[which] > icmplim) {
872 return(-1);
873 }
874 return(0);
875}
876