FreeBSD/Linux Kernel Cross Reference
sys/netinet/ip_icmp.c
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 * 4. Neither the name of the University nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * @(#)ip_icmp.c 8.2 (Berkeley) 1/4/94
30 * $FreeBSD: releng/6.2/sys/netinet/ip_icmp.c 155762 2006-02-16 17:50:57Z andre $
31 */
32
33 #include "opt_ipsec.h"
34 #include "opt_mac.h"
35
36 #include <sys/param.h>
37 #include <sys/systm.h>
38 #include <sys/mac.h>
39 #include <sys/mbuf.h>
40 #include <sys/protosw.h>
41 #include <sys/socket.h>
42 #include <sys/time.h>
43 #include <sys/kernel.h>
44 #include <sys/sysctl.h>
45
46 #include <net/if.h>
47 #include <net/if_types.h>
48 #include <net/route.h>
49
50 #include <netinet/in.h>
51 #include <netinet/in_pcb.h>
52 #include <netinet/in_systm.h>
53 #include <netinet/in_var.h>
54 #include <netinet/ip.h>
55 #include <netinet/ip_icmp.h>
56 #include <netinet/ip_var.h>
57 #include <netinet/tcp.h>
58 #include <netinet/tcp_var.h>
59 #include <netinet/tcpip.h>
60 #include <netinet/icmp_var.h>
61
62 #ifdef IPSEC
63 #include <netinet6/ipsec.h>
64 #include <netkey/key.h>
65 #endif
66
67 #ifdef FAST_IPSEC
68 #include <netipsec/ipsec.h>
69 #include <netipsec/key.h>
70 #define IPSEC
71 #endif
72
73 #include <machine/in_cksum.h>
74
75 /*
76 * ICMP routines: error generation, receive packet processing, and
77 * routines to turnaround packets back to the originator, and
78 * host table maintenance routines.
79 */
80
81 struct icmpstat icmpstat;
82 SYSCTL_STRUCT(_net_inet_icmp, ICMPCTL_STATS, stats, CTLFLAG_RW,
83 &icmpstat, icmpstat, "");
84
85 static int icmpmaskrepl = 0;
86 SYSCTL_INT(_net_inet_icmp, ICMPCTL_MASKREPL, maskrepl, CTLFLAG_RW,
87 &icmpmaskrepl, 0, "Reply to ICMP Address Mask Request packets.");
88
89 static u_int icmpmaskfake = 0;
90 SYSCTL_UINT(_net_inet_icmp, OID_AUTO, maskfake, CTLFLAG_RW,
91 &icmpmaskfake, 0, "Fake reply to ICMP Address Mask Request packets.");
92
93 static int drop_redirect = 0;
94 SYSCTL_INT(_net_inet_icmp, OID_AUTO, drop_redirect, CTLFLAG_RW,
95 &drop_redirect, 0, "");
96
97 static int log_redirect = 0;
98 SYSCTL_INT(_net_inet_icmp, OID_AUTO, log_redirect, CTLFLAG_RW,
99 &log_redirect, 0, "");
100
101 static int icmplim = 200;
102 SYSCTL_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RW,
103 &icmplim, 0, "");
104
105 static int icmplim_output = 1;
106 SYSCTL_INT(_net_inet_icmp, OID_AUTO, icmplim_output, CTLFLAG_RW,
107 &icmplim_output, 0, "");
108
109 static char reply_src[IFNAMSIZ];
110 SYSCTL_STRING(_net_inet_icmp, OID_AUTO, reply_src, CTLFLAG_RW,
111 &reply_src, IFNAMSIZ, "icmp reply source for non-local packets.");
112
113 static int icmp_rfi = 0;
114 SYSCTL_INT(_net_inet_icmp, OID_AUTO, reply_from_interface, CTLFLAG_RW,
115 &icmp_rfi, 0, "ICMP reply from incoming interface for "
116 "non-local packets");
117
118 static int icmp_quotelen = 8;
119 SYSCTL_INT(_net_inet_icmp, OID_AUTO, quotelen, CTLFLAG_RW,
120 &icmp_quotelen, 0, "Number of bytes from original packet to "
121 "quote in ICMP reply");
122
123 /*
124 * ICMP broadcast echo sysctl
125 */
126
127 static int icmpbmcastecho = 0;
128 SYSCTL_INT(_net_inet_icmp, OID_AUTO, bmcastecho, CTLFLAG_RW,
129 &icmpbmcastecho, 0, "");
130
131
132 #ifdef ICMPPRINTFS
133 int icmpprintfs = 0;
134 #endif
135
136 static void icmp_reflect(struct mbuf *);
137 static void icmp_send(struct mbuf *, struct mbuf *);
138
139 extern struct protosw inetsw[];
140
141 /*
142 * Generate an error packet of type error
143 * in response to bad packet ip.
144 */
145 void
146 icmp_error(n, type, code, dest, mtu)
147 struct mbuf *n;
148 int type, code;
149 n_long dest;
150 int mtu;
151 {
152 register struct ip *oip = mtod(n, struct ip *), *nip;
153 register unsigned oiphlen = oip->ip_hl << 2;
154 register struct icmp *icp;
155 register struct mbuf *m;
156 unsigned icmplen, icmpelen, nlen;
157
158 KASSERT((u_int)type <= ICMP_MAXTYPE, ("%s: illegal ICMP type", __func__));
159 #ifdef ICMPPRINTFS
160 if (icmpprintfs)
161 printf("icmp_error(%p, %x, %d)\n", oip, type, code);
162 #endif
163 if (type != ICMP_REDIRECT)
164 icmpstat.icps_error++;
165 /*
166 * Don't send error:
167 * if the original packet was encrypted.
168 * if not the first fragment of message.
169 * in response to a multicast or broadcast packet.
170 * if the old packet protocol was an ICMP error message.
171 */
172 if (n->m_flags & M_DECRYPTED)
173 goto freeit;
174 if (oip->ip_off & ~(IP_MF|IP_DF))
175 goto freeit;
176 if (n->m_flags & (M_BCAST|M_MCAST))
177 goto freeit;
178 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT &&
179 n->m_len >= oiphlen + ICMP_MINLEN &&
180 !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiphlen))->icmp_type)) {
181 icmpstat.icps_oldicmp++;
182 goto freeit;
183 }
184 /* Drop if IP header plus 8 bytes is not contignous in first mbuf. */
185 if (oiphlen + 8 > n->m_len)
186 goto freeit;
187 /*
188 * Calculate length to quote from original packet and
189 * prevent the ICMP mbuf from overflowing.
190 * Unfortunatly this is non-trivial since ip_forward()
191 * sends us truncated packets.
192 */
193 nlen = m_length(n, NULL);
194 if (oip->ip_p == IPPROTO_TCP) {
195 struct tcphdr *th;
196 int tcphlen;
197
198 if (oiphlen + sizeof(struct tcphdr) > n->m_len &&
199 n->m_next == NULL)
200 goto stdreply;
201 if (n->m_len < oiphlen + sizeof(struct tcphdr) &&
202 ((n = m_pullup(n, oiphlen + sizeof(struct tcphdr))) == NULL))
203 goto freeit;
204 th = (struct tcphdr *)((caddr_t)oip + oiphlen);
205 tcphlen = th->th_off << 2;
206 if (tcphlen < sizeof(struct tcphdr))
207 goto freeit;
208 if (oip->ip_len < oiphlen + tcphlen)
209 goto freeit;
210 if (oiphlen + tcphlen > n->m_len && n->m_next == NULL)
211 goto stdreply;
212 if (n->m_len < oiphlen + tcphlen &&
213 ((n = m_pullup(n, oiphlen + tcphlen)) == NULL))
214 goto freeit;
215 icmpelen = max(tcphlen, min(icmp_quotelen, oip->ip_len - oiphlen));
216 } else
217 stdreply: icmpelen = max(8, min(icmp_quotelen, oip->ip_len - oiphlen));
218
219 icmplen = min(oiphlen + icmpelen, nlen);
220 if (icmplen < sizeof(struct ip))
221 goto freeit;
222
223 if (MHLEN > sizeof(struct ip) + ICMP_MINLEN + icmplen)
224 m = m_gethdr(M_DONTWAIT, MT_DATA);
225 else
226 m = m_getcl(M_DONTWAIT, MT_DATA, M_PKTHDR);
227 if (m == NULL)
228 goto freeit;
229 #ifdef MAC
230 mac_create_mbuf_netlayer(n, m);
231 #endif
232 icmplen = min(icmplen, M_TRAILINGSPACE(m) - sizeof(struct ip) - ICMP_MINLEN);
233 m_align(m, ICMP_MINLEN + icmplen);
234 m->m_len = ICMP_MINLEN + icmplen;
235
236 icp = mtod(m, struct icmp *);
237 icmpstat.icps_outhist[type]++;
238 icp->icmp_type = type;
239 if (type == ICMP_REDIRECT)
240 icp->icmp_gwaddr.s_addr = dest;
241 else {
242 icp->icmp_void = 0;
243 /*
244 * The following assignments assume an overlay with the
245 * just zeroed icmp_void field.
246 */
247 if (type == ICMP_PARAMPROB) {
248 icp->icmp_pptr = code;
249 code = 0;
250 } else if (type == ICMP_UNREACH &&
251 code == ICMP_UNREACH_NEEDFRAG && mtu) {
252 icp->icmp_nextmtu = htons(mtu);
253 }
254 }
255 icp->icmp_code = code;
256
257 /*
258 * Copy the quotation into ICMP message and
259 * convert quoted IP header back to network representation.
260 */
261 m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip);
262 nip = &icp->icmp_ip;
263 nip->ip_len = htons(nip->ip_len);
264 nip->ip_off = htons(nip->ip_off);
265
266 /*
267 * Set up ICMP message mbuf and copy old IP header (without options
268 * in front of ICMP message.
269 * If the original mbuf was meant to bypass the firewall, the error
270 * reply should bypass as well.
271 */
272 m->m_flags |= n->m_flags & M_SKIP_FIREWALL;
273 m->m_data -= sizeof(struct ip);
274 m->m_len += sizeof(struct ip);
275 m->m_pkthdr.len = m->m_len;
276 m->m_pkthdr.rcvif = n->m_pkthdr.rcvif;
277 nip = mtod(m, struct ip *);
278 bcopy((caddr_t)oip, (caddr_t)nip, sizeof(struct ip));
279 nip->ip_len = m->m_len;
280 nip->ip_v = IPVERSION;
281 nip->ip_hl = 5;
282 nip->ip_p = IPPROTO_ICMP;
283 nip->ip_tos = 0;
284 icmp_reflect(m);
285
286 freeit:
287 m_freem(n);
288 }
289
290 /*
291 * Process a received ICMP message.
292 */
293 void
294 icmp_input(m, off)
295 struct mbuf *m;
296 int off;
297 {
298 struct icmp *icp;
299 struct in_ifaddr *ia;
300 struct ip *ip = mtod(m, struct ip *);
301 struct sockaddr_in icmpsrc, icmpdst, icmpgw;
302 int hlen = off;
303 int icmplen = ip->ip_len;
304 int i, code;
305 void (*ctlfunc)(int, struct sockaddr *, void *);
306
307 /*
308 * Locate icmp structure in mbuf, and check
309 * that not corrupted and of at least minimum length.
310 */
311 #ifdef ICMPPRINTFS
312 if (icmpprintfs) {
313 char buf[4 * sizeof "123"];
314 strcpy(buf, inet_ntoa(ip->ip_src));
315 printf("icmp_input from %s to %s, len %d\n",
316 buf, inet_ntoa(ip->ip_dst), icmplen);
317 }
318 #endif
319 if (icmplen < ICMP_MINLEN) {
320 icmpstat.icps_tooshort++;
321 goto freeit;
322 }
323 i = hlen + min(icmplen, ICMP_ADVLENMIN);
324 if (m->m_len < i && (m = m_pullup(m, i)) == 0) {
325 icmpstat.icps_tooshort++;
326 return;
327 }
328 ip = mtod(m, struct ip *);
329 m->m_len -= hlen;
330 m->m_data += hlen;
331 icp = mtod(m, struct icmp *);
332 if (in_cksum(m, icmplen)) {
333 icmpstat.icps_checksum++;
334 goto freeit;
335 }
336 m->m_len += hlen;
337 m->m_data -= hlen;
338
339 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) {
340 /*
341 * Deliver very specific ICMP type only.
342 */
343 switch (icp->icmp_type) {
344 case ICMP_UNREACH:
345 case ICMP_TIMXCEED:
346 break;
347 default:
348 goto freeit;
349 }
350 }
351
352 #ifdef ICMPPRINTFS
353 if (icmpprintfs)
354 printf("icmp_input, type %d code %d\n", icp->icmp_type,
355 icp->icmp_code);
356 #endif
357
358 /*
359 * Message type specific processing.
360 */
361 if (icp->icmp_type > ICMP_MAXTYPE)
362 goto raw;
363
364 /* Initialize */
365 bzero(&icmpsrc, sizeof(icmpsrc));
366 icmpsrc.sin_len = sizeof(struct sockaddr_in);
367 icmpsrc.sin_family = AF_INET;
368 bzero(&icmpdst, sizeof(icmpdst));
369 icmpdst.sin_len = sizeof(struct sockaddr_in);
370 icmpdst.sin_family = AF_INET;
371 bzero(&icmpgw, sizeof(icmpgw));
372 icmpgw.sin_len = sizeof(struct sockaddr_in);
373 icmpgw.sin_family = AF_INET;
374
375 icmpstat.icps_inhist[icp->icmp_type]++;
376 code = icp->icmp_code;
377 switch (icp->icmp_type) {
378
379 case ICMP_UNREACH:
380 switch (code) {
381 case ICMP_UNREACH_NET:
382 case ICMP_UNREACH_HOST:
383 case ICMP_UNREACH_SRCFAIL:
384 case ICMP_UNREACH_NET_UNKNOWN:
385 case ICMP_UNREACH_HOST_UNKNOWN:
386 case ICMP_UNREACH_ISOLATED:
387 case ICMP_UNREACH_TOSNET:
388 case ICMP_UNREACH_TOSHOST:
389 case ICMP_UNREACH_HOST_PRECEDENCE:
390 case ICMP_UNREACH_PRECEDENCE_CUTOFF:
391 code = PRC_UNREACH_NET;
392 break;
393
394 case ICMP_UNREACH_NEEDFRAG:
395 code = PRC_MSGSIZE;
396 break;
397
398 /*
399 * RFC 1122, Sections 3.2.2.1 and 4.2.3.9.
400 * Treat subcodes 2,3 as immediate RST
401 */
402 case ICMP_UNREACH_PROTOCOL:
403 case ICMP_UNREACH_PORT:
404 code = PRC_UNREACH_PORT;
405 break;
406
407 case ICMP_UNREACH_NET_PROHIB:
408 case ICMP_UNREACH_HOST_PROHIB:
409 case ICMP_UNREACH_FILTER_PROHIB:
410 code = PRC_UNREACH_ADMIN_PROHIB;
411 break;
412
413 default:
414 goto badcode;
415 }
416 goto deliver;
417
418 case ICMP_TIMXCEED:
419 if (code > 1)
420 goto badcode;
421 code += PRC_TIMXCEED_INTRANS;
422 goto deliver;
423
424 case ICMP_PARAMPROB:
425 if (code > 1)
426 goto badcode;
427 code = PRC_PARAMPROB;
428 goto deliver;
429
430 case ICMP_SOURCEQUENCH:
431 if (code)
432 goto badcode;
433 code = PRC_QUENCH;
434 deliver:
435 /*
436 * Problem with datagram; advise higher level routines.
437 */
438 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
439 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) {
440 icmpstat.icps_badlen++;
441 goto freeit;
442 }
443 icp->icmp_ip.ip_len = ntohs(icp->icmp_ip.ip_len);
444 /* Discard ICMP's in response to multicast packets */
445 if (IN_MULTICAST(ntohl(icp->icmp_ip.ip_dst.s_addr)))
446 goto badcode;
447 #ifdef ICMPPRINTFS
448 if (icmpprintfs)
449 printf("deliver to protocol %d\n", icp->icmp_ip.ip_p);
450 #endif
451 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
452 /*
453 * XXX if the packet contains [IPv4 AH TCP], we can't make a
454 * notification to TCP layer.
455 */
456 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput;
457 if (ctlfunc)
458 (*ctlfunc)(code, (struct sockaddr *)&icmpsrc,
459 (void *)&icp->icmp_ip);
460 break;
461
462 badcode:
463 icmpstat.icps_badcode++;
464 break;
465
466 case ICMP_ECHO:
467 if (!icmpbmcastecho
468 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
469 icmpstat.icps_bmcastecho++;
470 break;
471 }
472 icp->icmp_type = ICMP_ECHOREPLY;
473 if (badport_bandlim(BANDLIM_ICMP_ECHO) < 0)
474 goto freeit;
475 else
476 goto reflect;
477
478 case ICMP_TSTAMP:
479 if (!icmpbmcastecho
480 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
481 icmpstat.icps_bmcasttstamp++;
482 break;
483 }
484 if (icmplen < ICMP_TSLEN) {
485 icmpstat.icps_badlen++;
486 break;
487 }
488 icp->icmp_type = ICMP_TSTAMPREPLY;
489 icp->icmp_rtime = iptime();
490 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */
491 if (badport_bandlim(BANDLIM_ICMP_TSTAMP) < 0)
492 goto freeit;
493 else
494 goto reflect;
495
496 case ICMP_MASKREQ:
497 if (icmpmaskrepl == 0)
498 break;
499 /*
500 * We are not able to respond with all ones broadcast
501 * unless we receive it over a point-to-point interface.
502 */
503 if (icmplen < ICMP_MASKLEN)
504 break;
505 switch (ip->ip_dst.s_addr) {
506
507 case INADDR_BROADCAST:
508 case INADDR_ANY:
509 icmpdst.sin_addr = ip->ip_src;
510 break;
511
512 default:
513 icmpdst.sin_addr = ip->ip_dst;
514 }
515 ia = (struct in_ifaddr *)ifaof_ifpforaddr(
516 (struct sockaddr *)&icmpdst, m->m_pkthdr.rcvif);
517 if (ia == 0)
518 break;
519 if (ia->ia_ifp == 0)
520 break;
521 icp->icmp_type = ICMP_MASKREPLY;
522 if (icmpmaskfake == 0)
523 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr;
524 else
525 icp->icmp_mask = icmpmaskfake;
526 if (ip->ip_src.s_addr == 0) {
527 if (ia->ia_ifp->if_flags & IFF_BROADCAST)
528 ip->ip_src = satosin(&ia->ia_broadaddr)->sin_addr;
529 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT)
530 ip->ip_src = satosin(&ia->ia_dstaddr)->sin_addr;
531 }
532 reflect:
533 ip->ip_len += hlen; /* since ip_input deducts this */
534 icmpstat.icps_reflect++;
535 icmpstat.icps_outhist[icp->icmp_type]++;
536 icmp_reflect(m);
537 return;
538
539 case ICMP_REDIRECT:
540 if (log_redirect) {
541 u_long src, dst, gw;
542
543 src = ntohl(ip->ip_src.s_addr);
544 dst = ntohl(icp->icmp_ip.ip_dst.s_addr);
545 gw = ntohl(icp->icmp_gwaddr.s_addr);
546 printf("icmp redirect from %d.%d.%d.%d: "
547 "%d.%d.%d.%d => %d.%d.%d.%d\n",
548 (int)(src >> 24), (int)((src >> 16) & 0xff),
549 (int)((src >> 8) & 0xff), (int)(src & 0xff),
550 (int)(dst >> 24), (int)((dst >> 16) & 0xff),
551 (int)((dst >> 8) & 0xff), (int)(dst & 0xff),
552 (int)(gw >> 24), (int)((gw >> 16) & 0xff),
553 (int)((gw >> 8) & 0xff), (int)(gw & 0xff));
554 }
555 /*
556 * RFC1812 says we must ignore ICMP redirects if we
557 * are acting as router.
558 */
559 if (drop_redirect || ipforwarding)
560 break;
561 if (code > 3)
562 goto badcode;
563 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
564 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) {
565 icmpstat.icps_badlen++;
566 break;
567 }
568 /*
569 * Short circuit routing redirects to force
570 * immediate change in the kernel's routing
571 * tables. The message is also handed to anyone
572 * listening on a raw socket (e.g. the routing
573 * daemon for use in updating its tables).
574 */
575 icmpgw.sin_addr = ip->ip_src;
576 icmpdst.sin_addr = icp->icmp_gwaddr;
577 #ifdef ICMPPRINTFS
578 if (icmpprintfs) {
579 char buf[4 * sizeof "123"];
580 strcpy(buf, inet_ntoa(icp->icmp_ip.ip_dst));
581
582 printf("redirect dst %s to %s\n",
583 buf, inet_ntoa(icp->icmp_gwaddr));
584 }
585 #endif
586 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
587 rtredirect((struct sockaddr *)&icmpsrc,
588 (struct sockaddr *)&icmpdst,
589 (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST,
590 (struct sockaddr *)&icmpgw);
591 pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&icmpsrc);
592 #ifdef IPSEC
593 key_sa_routechange((struct sockaddr *)&icmpsrc);
594 #endif
595 break;
596
597 /*
598 * No kernel processing for the following;
599 * just fall through to send to raw listener.
600 */
601 case ICMP_ECHOREPLY:
602 case ICMP_ROUTERADVERT:
603 case ICMP_ROUTERSOLICIT:
604 case ICMP_TSTAMPREPLY:
605 case ICMP_IREQREPLY:
606 case ICMP_MASKREPLY:
607 default:
608 break;
609 }
610
611 raw:
612 rip_input(m, off);
613 return;
614
615 freeit:
616 m_freem(m);
617 }
618
619 /*
620 * Reflect the ip packet back to the source
621 */
622 static void
623 icmp_reflect(m)
624 struct mbuf *m;
625 {
626 struct ip *ip = mtod(m, struct ip *);
627 struct ifaddr *ifa;
628 struct ifnet *ifn;
629 struct in_ifaddr *ia;
630 struct in_addr t;
631 struct mbuf *opts = 0;
632 int optlen = (ip->ip_hl << 2) - sizeof(struct ip);
633
634 if (!in_canforward(ip->ip_src) &&
635 ((ntohl(ip->ip_src.s_addr) & IN_CLASSA_NET) !=
636 (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))) {
637 m_freem(m); /* Bad return address */
638 icmpstat.icps_badaddr++;
639 goto done; /* Ip_output() will check for broadcast */
640 }
641 t = ip->ip_dst;
642 ip->ip_dst = ip->ip_src;
643
644 /*
645 * Source selection for ICMP replies:
646 *
647 * If the incoming packet was addressed directly to one of our
648 * own addresses, use dst as the src for the reply.
649 */
650 LIST_FOREACH(ia, INADDR_HASH(t.s_addr), ia_hash)
651 if (t.s_addr == IA_SIN(ia)->sin_addr.s_addr)
652 goto match;
653 /*
654 * If the incoming packet was addressed to one of our broadcast
655 * addresses, use the first non-broadcast address which corresponds
656 * to the incoming interface.
657 */
658 if (m->m_pkthdr.rcvif != NULL &&
659 m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
660 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrhead, ifa_link) {
661 if (ifa->ifa_addr->sa_family != AF_INET)
662 continue;
663 ia = ifatoia(ifa);
664 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
665 t.s_addr)
666 goto match;
667 }
668 }
669 /*
670 * If the packet was transiting through us, use the address of
671 * the interface the packet came through in. If that interface
672 * doesn't have a suitable IP address, the normal selection
673 * criteria apply.
674 */
675 if (icmp_rfi && m->m_pkthdr.rcvif != NULL) {
676 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrhead, ifa_link) {
677 if (ifa->ifa_addr->sa_family != AF_INET)
678 continue;
679 ia = ifatoia(ifa);
680 goto match;
681 }
682 }
683 /*
684 * If the incoming packet was not addressed directly to us, use
685 * designated interface for icmp replies specified by sysctl
686 * net.inet.icmp.reply_src (default not set). Otherwise continue
687 * with normal source selection.
688 */
689 if (reply_src[0] != '\0' && (ifn = ifunit(reply_src))) {
690 TAILQ_FOREACH(ifa, &ifn->if_addrhead, ifa_link) {
691 if (ifa->ifa_addr->sa_family != AF_INET)
692 continue;
693 ia = ifatoia(ifa);
694 goto match;
695 }
696 }
697 /*
698 * If the packet was transiting through us, use the address of
699 * the interface that is the closest to the packet source.
700 * When we don't have a route back to the packet source, stop here
701 * and drop the packet.
702 */
703 ia = ip_rtaddr(ip->ip_dst);
704 if (ia == NULL) {
705 m_freem(m);
706 icmpstat.icps_noroute++;
707 goto done;
708 }
709 match:
710 #ifdef MAC
711 mac_reflect_mbuf_icmp(m);
712 #endif
713 t = IA_SIN(ia)->sin_addr;
714 ip->ip_src = t;
715 ip->ip_ttl = ip_defttl;
716
717 if (optlen > 0) {
718 register u_char *cp;
719 int opt, cnt;
720 u_int len;
721
722 /*
723 * Retrieve any source routing from the incoming packet;
724 * add on any record-route or timestamp options.
725 */
726 cp = (u_char *) (ip + 1);
727 if ((opts = ip_srcroute(m)) == 0 &&
728 (opts = m_gethdr(M_DONTWAIT, MT_HEADER))) {
729 opts->m_len = sizeof(struct in_addr);
730 mtod(opts, struct in_addr *)->s_addr = 0;
731 }
732 if (opts) {
733 #ifdef ICMPPRINTFS
734 if (icmpprintfs)
735 printf("icmp_reflect optlen %d rt %d => ",
736 optlen, opts->m_len);
737 #endif
738 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) {
739 opt = cp[IPOPT_OPTVAL];
740 if (opt == IPOPT_EOL)
741 break;
742 if (opt == IPOPT_NOP)
743 len = 1;
744 else {
745 if (cnt < IPOPT_OLEN + sizeof(*cp))
746 break;
747 len = cp[IPOPT_OLEN];
748 if (len < IPOPT_OLEN + sizeof(*cp) ||
749 len > cnt)
750 break;
751 }
752 /*
753 * Should check for overflow, but it "can't happen"
754 */
755 if (opt == IPOPT_RR || opt == IPOPT_TS ||
756 opt == IPOPT_SECURITY) {
757 bcopy((caddr_t)cp,
758 mtod(opts, caddr_t) + opts->m_len, len);
759 opts->m_len += len;
760 }
761 }
762 /* Terminate & pad, if necessary */
763 cnt = opts->m_len % 4;
764 if (cnt) {
765 for (; cnt < 4; cnt++) {
766 *(mtod(opts, caddr_t) + opts->m_len) =
767 IPOPT_EOL;
768 opts->m_len++;
769 }
770 }
771 #ifdef ICMPPRINTFS
772 if (icmpprintfs)
773 printf("%d\n", opts->m_len);
774 #endif
775 }
776 /*
777 * Now strip out original options by copying rest of first
778 * mbuf's data back, and adjust the IP length.
779 */
780 ip->ip_len -= optlen;
781 ip->ip_v = IPVERSION;
782 ip->ip_hl = 5;
783 m->m_len -= optlen;
784 if (m->m_flags & M_PKTHDR)
785 m->m_pkthdr.len -= optlen;
786 optlen += sizeof(struct ip);
787 bcopy((caddr_t)ip + optlen, (caddr_t)(ip + 1),
788 (unsigned)(m->m_len - sizeof(struct ip)));
789 }
790 m_tag_delete_nonpersistent(m);
791 m->m_flags &= ~(M_BCAST|M_MCAST);
792 icmp_send(m, opts);
793 done:
794 if (opts)
795 (void)m_free(opts);
796 }
797
798 /*
799 * Send an icmp packet back to the ip level,
800 * after supplying a checksum.
801 */
802 static void
803 icmp_send(m, opts)
804 register struct mbuf *m;
805 struct mbuf *opts;
806 {
807 register struct ip *ip = mtod(m, struct ip *);
808 register int hlen;
809 register struct icmp *icp;
810
811 hlen = ip->ip_hl << 2;
812 m->m_data += hlen;
813 m->m_len -= hlen;
814 icp = mtod(m, struct icmp *);
815 icp->icmp_cksum = 0;
816 icp->icmp_cksum = in_cksum(m, ip->ip_len - hlen);
817 m->m_data -= hlen;
818 m->m_len += hlen;
819 m->m_pkthdr.rcvif = (struct ifnet *)0;
820 #ifdef ICMPPRINTFS
821 if (icmpprintfs) {
822 char buf[4 * sizeof "123"];
823 strcpy(buf, inet_ntoa(ip->ip_dst));
824 printf("icmp_send dst %s src %s\n",
825 buf, inet_ntoa(ip->ip_src));
826 }
827 #endif
828 (void) ip_output(m, opts, NULL, 0, NULL, NULL);
829 }
830
831 n_time
832 iptime()
833 {
834 struct timeval atv;
835 u_long t;
836
837 getmicrotime(&atv);
838 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000;
839 return (htonl(t));
840 }
841
842 /*
843 * Return the next larger or smaller MTU plateau (table from RFC 1191)
844 * given current value MTU. If DIR is less than zero, a larger plateau
845 * is returned; otherwise, a smaller value is returned.
846 */
847 int
848 ip_next_mtu(mtu, dir)
849 int mtu;
850 int dir;
851 {
852 static int mtutab[] = {
853 65535, 32000, 17914, 8166, 4352, 2002, 1492, 1280, 1006, 508,
854 296, 68, 0
855 };
856 int i, size;
857
858 size = (sizeof mtutab) / (sizeof mtutab[0]);
859 if (dir >= 0) {
860 for (i = 0; i < size; i++)
861 if (mtu > mtutab[i])
862 return mtutab[i];
863 } else {
864 for (i = size - 1; i >= 0; i--)
865 if (mtu < mtutab[i])
866 return mtutab[i];
867 if (mtu == mtutab[0])
868 return mtutab[0];
869 }
870 return 0;
871 }
872
873
874 /*
875 * badport_bandlim() - check for ICMP bandwidth limit
876 *
877 * Return 0 if it is ok to send an ICMP error response, -1 if we have
878 * hit our bandwidth limit and it is not ok.
879 *
880 * If icmplim is <= 0, the feature is disabled and 0 is returned.
881 *
882 * For now we separate the TCP and UDP subsystems w/ different 'which'
883 * values. We may eventually remove this separation (and simplify the
884 * code further).
885 *
886 * Note that the printing of the error message is delayed so we can
887 * properly print the icmp error rate that the system was trying to do
888 * (i.e. 22000/100 pps, etc...). This can cause long delays in printing
889 * the 'final' error, but it doesn't make sense to solve the printing
890 * delay with more complex code.
891 */
892
893 int
894 badport_bandlim(int which)
895 {
896 #define N(a) (sizeof (a) / sizeof (a[0]))
897 static struct rate {
898 const char *type;
899 struct timeval lasttime;
900 int curpps;
901 } rates[BANDLIM_MAX+1] = {
902 { "icmp unreach response" },
903 { "icmp ping response" },
904 { "icmp tstamp response" },
905 { "closed port RST response" },
906 { "open port RST response" }
907 };
908
909 /*
910 * Return ok status if feature disabled or argument out of range.
911 */
912 if (icmplim > 0 && (u_int) which < N(rates)) {
913 struct rate *r = &rates[which];
914 int opps = r->curpps;
915
916 if (!ppsratecheck(&r->lasttime, &r->curpps, icmplim))
917 return -1; /* discard packet */
918 /*
919 * If we've dropped below the threshold after having
920 * rate-limited traffic print the message. This preserves
921 * the previous behaviour at the expense of added complexity.
922 */
923 if (icmplim_output && opps > icmplim)
924 printf("Limiting %s from %d to %d packets/sec\n",
925 r->type, opps, icmplim);
926 }
927 return 0; /* okay to send packet */
928 #undef N
929 }
Cache object: a0b89fb504278a878f028248a1d90e78
|