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 * $FreeBSD$
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 #include <sys/syslog.h>
46
47 #include <net/if.h>
48 #include <net/if_types.h>
49 #include <net/route.h>
50
51 #include <netinet/in.h>
52 #include <netinet/in_pcb.h>
53 #include <netinet/in_systm.h>
54 #include <netinet/in_var.h>
55 #include <netinet/ip.h>
56 #include <netinet/ip_icmp.h>
57 #include <netinet/ip_var.h>
58 #include <netinet/tcp.h>
59 #include <netinet/tcp_var.h>
60 #include <netinet/tcpip.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 #ifdef FAST_IPSEC
69 #include <netipsec/ipsec.h>
70 #include <netipsec/key.h>
71 #define IPSEC
72 #endif
73
74 #include <machine/in_cksum.h>
75
76 /*
77 * ICMP routines: error generation, receive packet processing, and
78 * routines to turnaround packets back to the originator, and
79 * host table maintenance routines.
80 */
81
82 struct icmpstat icmpstat;
83 SYSCTL_STRUCT(_net_inet_icmp, ICMPCTL_STATS, stats, CTLFLAG_RW,
84 &icmpstat, icmpstat, "");
85
86 static int icmpmaskrepl = 0;
87 SYSCTL_INT(_net_inet_icmp, ICMPCTL_MASKREPL, maskrepl, CTLFLAG_RW,
88 &icmpmaskrepl, 0, "Reply to ICMP Address Mask Request packets.");
89
90 static u_int icmpmaskfake = 0;
91 SYSCTL_UINT(_net_inet_icmp, OID_AUTO, maskfake, CTLFLAG_RW,
92 &icmpmaskfake, 0, "Fake reply to ICMP Address Mask Request packets.");
93
94 static int drop_redirect = 0;
95 SYSCTL_INT(_net_inet_icmp, OID_AUTO, drop_redirect, CTLFLAG_RW,
96 &drop_redirect, 0, "");
97
98 static int log_redirect = 0;
99 SYSCTL_INT(_net_inet_icmp, OID_AUTO, log_redirect, CTLFLAG_RW,
100 &log_redirect, 0, "");
101
102 static int icmplim = 200;
103 SYSCTL_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RW,
104 &icmplim, 0, "");
105
106 static int icmplim_output = 1;
107 SYSCTL_INT(_net_inet_icmp, OID_AUTO, icmplim_output, CTLFLAG_RW,
108 &icmplim_output, 0, "");
109
110 static char reply_src[IFNAMSIZ];
111 SYSCTL_STRING(_net_inet_icmp, OID_AUTO, reply_src, CTLFLAG_RW,
112 &reply_src, IFNAMSIZ, "icmp reply source for non-local packets.");
113
114 static int icmp_rfi = 0;
115 SYSCTL_INT(_net_inet_icmp, OID_AUTO, reply_from_interface, CTLFLAG_RW,
116 &icmp_rfi, 0, "ICMP reply from incoming interface for "
117 "non-local packets");
118
119 static int icmp_quotelen = 8;
120 SYSCTL_INT(_net_inet_icmp, OID_AUTO, quotelen, CTLFLAG_RW,
121 &icmp_quotelen, 0, "Number of bytes from original packet to "
122 "quote in ICMP reply");
123
124 /*
125 * ICMP broadcast echo sysctl
126 */
127
128 static int icmpbmcastecho = 0;
129 SYSCTL_INT(_net_inet_icmp, OID_AUTO, bmcastecho, CTLFLAG_RW,
130 &icmpbmcastecho, 0, "");
131
132
133 #ifdef ICMPPRINTFS
134 int icmpprintfs = 0;
135 #endif
136
137 static void icmp_reflect(struct mbuf *);
138 static void icmp_send(struct mbuf *, struct mbuf *);
139
140 extern struct protosw inetsw[];
141
142 /*
143 * Generate an error packet of type error
144 * in response to bad packet ip.
145 */
146 void
147 icmp_error(n, type, code, dest, mtu)
148 struct mbuf *n;
149 int type, code;
150 n_long dest;
151 int mtu;
152 {
153 register struct ip *oip = mtod(n, struct ip *), *nip;
154 register unsigned oiphlen = oip->ip_hl << 2;
155 register struct icmp *icp;
156 register struct mbuf *m;
157 unsigned icmplen, icmpelen, nlen;
158
159 KASSERT((u_int)type <= ICMP_MAXTYPE, ("%s: illegal ICMP type", __func__));
160 #ifdef ICMPPRINTFS
161 if (icmpprintfs)
162 printf("icmp_error(%p, %x, %d)\n", oip, type, code);
163 #endif
164 if (type != ICMP_REDIRECT)
165 icmpstat.icps_error++;
166 /*
167 * Don't send error:
168 * if the original packet was encrypted.
169 * if not the first fragment of message.
170 * in response to a multicast or broadcast packet.
171 * if the old packet protocol was an ICMP error message.
172 */
173 if (n->m_flags & M_DECRYPTED)
174 goto freeit;
175 if (oip->ip_off & ~(IP_MF|IP_DF))
176 goto freeit;
177 if (n->m_flags & (M_BCAST|M_MCAST))
178 goto freeit;
179 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT &&
180 n->m_len >= oiphlen + ICMP_MINLEN &&
181 !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiphlen))->icmp_type)) {
182 icmpstat.icps_oldicmp++;
183 goto freeit;
184 }
185 /* Drop if IP header plus 8 bytes is not contignous in first mbuf. */
186 if (oiphlen + 8 > n->m_len)
187 goto freeit;
188 /*
189 * Calculate length to quote from original packet and
190 * prevent the ICMP mbuf from overflowing.
191 * Unfortunatly this is non-trivial since ip_forward()
192 * sends us truncated packets.
193 */
194 nlen = m_length(n, NULL);
195 if (oip->ip_p == IPPROTO_TCP) {
196 struct tcphdr *th;
197 int tcphlen;
198
199 if (oiphlen + sizeof(struct tcphdr) > n->m_len &&
200 n->m_next == NULL)
201 goto stdreply;
202 if (n->m_len < oiphlen + sizeof(struct tcphdr) &&
203 ((n = m_pullup(n, oiphlen + sizeof(struct tcphdr))) == NULL))
204 goto freeit;
205 th = (struct tcphdr *)((caddr_t)oip + oiphlen);
206 tcphlen = th->th_off << 2;
207 if (tcphlen < sizeof(struct tcphdr))
208 goto freeit;
209 if (oip->ip_len < oiphlen + tcphlen)
210 goto freeit;
211 if (oiphlen + tcphlen > n->m_len && n->m_next == NULL)
212 goto stdreply;
213 if (n->m_len < oiphlen + tcphlen &&
214 ((n = m_pullup(n, oiphlen + tcphlen)) == NULL))
215 goto freeit;
216 icmpelen = max(tcphlen, min(icmp_quotelen, oip->ip_len - oiphlen));
217 } else
218 stdreply: icmpelen = max(8, min(icmp_quotelen, oip->ip_len - oiphlen));
219
220 icmplen = min(oiphlen + icmpelen, nlen);
221 if (icmplen < sizeof(struct ip))
222 goto freeit;
223
224 if (MHLEN > sizeof(struct ip) + ICMP_MINLEN + icmplen)
225 m = m_gethdr(M_DONTWAIT, MT_DATA);
226 else
227 m = m_getcl(M_DONTWAIT, MT_DATA, M_PKTHDR);
228 if (m == NULL)
229 goto freeit;
230 #ifdef MAC
231 mac_create_mbuf_netlayer(n, m);
232 #endif
233 icmplen = min(icmplen, M_TRAILINGSPACE(m) - sizeof(struct ip) - ICMP_MINLEN);
234 m_align(m, ICMP_MINLEN + icmplen);
235 m->m_len = ICMP_MINLEN + icmplen;
236
237 icp = mtod(m, struct icmp *);
238 icmpstat.icps_outhist[type]++;
239 icp->icmp_type = type;
240 if (type == ICMP_REDIRECT)
241 icp->icmp_gwaddr.s_addr = dest;
242 else {
243 icp->icmp_void = 0;
244 /*
245 * The following assignments assume an overlay with the
246 * just zeroed icmp_void field.
247 */
248 if (type == ICMP_PARAMPROB) {
249 icp->icmp_pptr = code;
250 code = 0;
251 } else if (type == ICMP_UNREACH &&
252 code == ICMP_UNREACH_NEEDFRAG && mtu) {
253 icp->icmp_nextmtu = htons(mtu);
254 }
255 }
256 icp->icmp_code = code;
257
258 /*
259 * Copy the quotation into ICMP message and
260 * convert quoted IP header back to network representation.
261 */
262 m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip);
263 nip = &icp->icmp_ip;
264 nip->ip_len = htons(nip->ip_len);
265 nip->ip_off = htons(nip->ip_off);
266
267 /*
268 * Set up ICMP message mbuf and copy old IP header (without options
269 * in front of ICMP message.
270 * If the original mbuf was meant to bypass the firewall, the error
271 * reply should bypass as well.
272 */
273 m->m_flags |= n->m_flags & M_SKIP_FIREWALL;
274 m->m_data -= sizeof(struct ip);
275 m->m_len += sizeof(struct ip);
276 m->m_pkthdr.len = m->m_len;
277 m->m_pkthdr.rcvif = n->m_pkthdr.rcvif;
278 nip = mtod(m, struct ip *);
279 bcopy((caddr_t)oip, (caddr_t)nip, sizeof(struct ip));
280 nip->ip_len = m->m_len;
281 nip->ip_v = IPVERSION;
282 nip->ip_hl = 5;
283 nip->ip_p = IPPROTO_ICMP;
284 nip->ip_tos = 0;
285 icmp_reflect(m);
286
287 freeit:
288 m_freem(n);
289 }
290
291 /*
292 * Process a received ICMP message.
293 */
294 void
295 icmp_input(m, off)
296 struct mbuf *m;
297 int off;
298 {
299 struct icmp *icp;
300 struct in_ifaddr *ia;
301 struct ip *ip = mtod(m, struct ip *);
302 struct sockaddr_in icmpsrc, icmpdst, icmpgw;
303 int hlen = off;
304 int icmplen = ip->ip_len;
305 int i, code;
306 void (*ctlfunc)(int, struct sockaddr *, void *);
307
308 /*
309 * Locate icmp structure in mbuf, and check
310 * that not corrupted and of at least minimum length.
311 */
312 #ifdef ICMPPRINTFS
313 if (icmpprintfs) {
314 char buf[4 * sizeof "123"];
315 strcpy(buf, inet_ntoa(ip->ip_src));
316 printf("icmp_input from %s to %s, len %d\n",
317 buf, inet_ntoa(ip->ip_dst), icmplen);
318 }
319 #endif
320 if (icmplen < ICMP_MINLEN) {
321 icmpstat.icps_tooshort++;
322 goto freeit;
323 }
324 i = hlen + min(icmplen, ICMP_ADVLENMIN);
325 if (m->m_len < i && (m = m_pullup(m, i)) == 0) {
326 icmpstat.icps_tooshort++;
327 return;
328 }
329 ip = mtod(m, struct ip *);
330 m->m_len -= hlen;
331 m->m_data += hlen;
332 icp = mtod(m, struct icmp *);
333 if (in_cksum(m, icmplen)) {
334 icmpstat.icps_checksum++;
335 goto freeit;
336 }
337 m->m_len += hlen;
338 m->m_data -= hlen;
339
340 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) {
341 /*
342 * Deliver very specific ICMP type only.
343 */
344 switch (icp->icmp_type) {
345 case ICMP_UNREACH:
346 case ICMP_TIMXCEED:
347 break;
348 default:
349 goto freeit;
350 }
351 }
352
353 #ifdef ICMPPRINTFS
354 if (icmpprintfs)
355 printf("icmp_input, type %d code %d\n", icp->icmp_type,
356 icp->icmp_code);
357 #endif
358
359 /*
360 * Message type specific processing.
361 */
362 if (icp->icmp_type > ICMP_MAXTYPE)
363 goto raw;
364
365 /* Initialize */
366 bzero(&icmpsrc, sizeof(icmpsrc));
367 icmpsrc.sin_len = sizeof(struct sockaddr_in);
368 icmpsrc.sin_family = AF_INET;
369 bzero(&icmpdst, sizeof(icmpdst));
370 icmpdst.sin_len = sizeof(struct sockaddr_in);
371 icmpdst.sin_family = AF_INET;
372 bzero(&icmpgw, sizeof(icmpgw));
373 icmpgw.sin_len = sizeof(struct sockaddr_in);
374 icmpgw.sin_family = AF_INET;
375
376 icmpstat.icps_inhist[icp->icmp_type]++;
377 code = icp->icmp_code;
378 switch (icp->icmp_type) {
379
380 case ICMP_UNREACH:
381 switch (code) {
382 case ICMP_UNREACH_NET:
383 case ICMP_UNREACH_HOST:
384 case ICMP_UNREACH_SRCFAIL:
385 case ICMP_UNREACH_NET_UNKNOWN:
386 case ICMP_UNREACH_HOST_UNKNOWN:
387 case ICMP_UNREACH_ISOLATED:
388 case ICMP_UNREACH_TOSNET:
389 case ICMP_UNREACH_TOSHOST:
390 case ICMP_UNREACH_HOST_PRECEDENCE:
391 case ICMP_UNREACH_PRECEDENCE_CUTOFF:
392 code = PRC_UNREACH_NET;
393 break;
394
395 case ICMP_UNREACH_NEEDFRAG:
396 code = PRC_MSGSIZE;
397 break;
398
399 /*
400 * RFC 1122, Sections 3.2.2.1 and 4.2.3.9.
401 * Treat subcodes 2,3 as immediate RST
402 */
403 case ICMP_UNREACH_PROTOCOL:
404 case ICMP_UNREACH_PORT:
405 code = PRC_UNREACH_PORT;
406 break;
407
408 case ICMP_UNREACH_NET_PROHIB:
409 case ICMP_UNREACH_HOST_PROHIB:
410 case ICMP_UNREACH_FILTER_PROHIB:
411 code = PRC_UNREACH_ADMIN_PROHIB;
412 break;
413
414 default:
415 goto badcode;
416 }
417 goto deliver;
418
419 case ICMP_TIMXCEED:
420 if (code > 1)
421 goto badcode;
422 code += PRC_TIMXCEED_INTRANS;
423 goto deliver;
424
425 case ICMP_PARAMPROB:
426 if (code > 1)
427 goto badcode;
428 code = PRC_PARAMPROB;
429 goto deliver;
430
431 case ICMP_SOURCEQUENCH:
432 if (code)
433 goto badcode;
434 code = PRC_QUENCH;
435 deliver:
436 /*
437 * Problem with datagram; advise higher level routines.
438 */
439 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
440 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) {
441 icmpstat.icps_badlen++;
442 goto freeit;
443 }
444 icp->icmp_ip.ip_len = ntohs(icp->icmp_ip.ip_len);
445 /* Discard ICMP's in response to multicast packets */
446 if (IN_MULTICAST(ntohl(icp->icmp_ip.ip_dst.s_addr)))
447 goto badcode;
448 #ifdef ICMPPRINTFS
449 if (icmpprintfs)
450 printf("deliver to protocol %d\n", icp->icmp_ip.ip_p);
451 #endif
452 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
453 /*
454 * XXX if the packet contains [IPv4 AH TCP], we can't make a
455 * notification to TCP layer.
456 */
457 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput;
458 if (ctlfunc)
459 (*ctlfunc)(code, (struct sockaddr *)&icmpsrc,
460 (void *)&icp->icmp_ip);
461 break;
462
463 badcode:
464 icmpstat.icps_badcode++;
465 break;
466
467 case ICMP_ECHO:
468 if (!icmpbmcastecho
469 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
470 icmpstat.icps_bmcastecho++;
471 break;
472 }
473 icp->icmp_type = ICMP_ECHOREPLY;
474 if (badport_bandlim(BANDLIM_ICMP_ECHO) < 0)
475 goto freeit;
476 else
477 goto reflect;
478
479 case ICMP_TSTAMP:
480 if (!icmpbmcastecho
481 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
482 icmpstat.icps_bmcasttstamp++;
483 break;
484 }
485 if (icmplen < ICMP_TSLEN) {
486 icmpstat.icps_badlen++;
487 break;
488 }
489 icp->icmp_type = ICMP_TSTAMPREPLY;
490 icp->icmp_rtime = iptime();
491 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */
492 if (badport_bandlim(BANDLIM_ICMP_TSTAMP) < 0)
493 goto freeit;
494 else
495 goto reflect;
496
497 case ICMP_MASKREQ:
498 if (icmpmaskrepl == 0)
499 break;
500 /*
501 * We are not able to respond with all ones broadcast
502 * unless we receive it over a point-to-point interface.
503 */
504 if (icmplen < ICMP_MASKLEN)
505 break;
506 switch (ip->ip_dst.s_addr) {
507
508 case INADDR_BROADCAST:
509 case INADDR_ANY:
510 icmpdst.sin_addr = ip->ip_src;
511 break;
512
513 default:
514 icmpdst.sin_addr = ip->ip_dst;
515 }
516 ia = (struct in_ifaddr *)ifaof_ifpforaddr(
517 (struct sockaddr *)&icmpdst, m->m_pkthdr.rcvif);
518 if (ia == 0)
519 break;
520 if (ia->ia_ifp == 0)
521 break;
522 icp->icmp_type = ICMP_MASKREPLY;
523 if (icmpmaskfake == 0)
524 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr;
525 else
526 icp->icmp_mask = icmpmaskfake;
527 if (ip->ip_src.s_addr == 0) {
528 if (ia->ia_ifp->if_flags & IFF_BROADCAST)
529 ip->ip_src = satosin(&ia->ia_broadaddr)->sin_addr;
530 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT)
531 ip->ip_src = satosin(&ia->ia_dstaddr)->sin_addr;
532 }
533 reflect:
534 ip->ip_len += hlen; /* since ip_input deducts this */
535 icmpstat.icps_reflect++;
536 icmpstat.icps_outhist[icp->icmp_type]++;
537 icmp_reflect(m);
538 return;
539
540 case ICMP_REDIRECT:
541 if (log_redirect) {
542 u_long src, dst, gw;
543
544 src = ntohl(ip->ip_src.s_addr);
545 dst = ntohl(icp->icmp_ip.ip_dst.s_addr);
546 gw = ntohl(icp->icmp_gwaddr.s_addr);
547 printf("icmp redirect from %d.%d.%d.%d: "
548 "%d.%d.%d.%d => %d.%d.%d.%d\n",
549 (int)(src >> 24), (int)((src >> 16) & 0xff),
550 (int)((src >> 8) & 0xff), (int)(src & 0xff),
551 (int)(dst >> 24), (int)((dst >> 16) & 0xff),
552 (int)((dst >> 8) & 0xff), (int)(dst & 0xff),
553 (int)(gw >> 24), (int)((gw >> 16) & 0xff),
554 (int)((gw >> 8) & 0xff), (int)(gw & 0xff));
555 }
556 /*
557 * RFC1812 says we must ignore ICMP redirects if we
558 * are acting as router.
559 */
560 if (drop_redirect || ipforwarding)
561 break;
562 if (code > 3)
563 goto badcode;
564 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
565 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) {
566 icmpstat.icps_badlen++;
567 break;
568 }
569 /*
570 * Short circuit routing redirects to force
571 * immediate change in the kernel's routing
572 * tables. The message is also handed to anyone
573 * listening on a raw socket (e.g. the routing
574 * daemon for use in updating its tables).
575 */
576 icmpgw.sin_addr = ip->ip_src;
577 icmpdst.sin_addr = icp->icmp_gwaddr;
578 #ifdef ICMPPRINTFS
579 if (icmpprintfs) {
580 char buf[4 * sizeof "123"];
581 strcpy(buf, inet_ntoa(icp->icmp_ip.ip_dst));
582
583 printf("redirect dst %s to %s\n",
584 buf, inet_ntoa(icp->icmp_gwaddr));
585 }
586 #endif
587 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
588 rtredirect((struct sockaddr *)&icmpsrc,
589 (struct sockaddr *)&icmpdst,
590 (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST,
591 (struct sockaddr *)&icmpgw);
592 pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&icmpsrc);
593 #ifdef IPSEC
594 key_sa_routechange((struct sockaddr *)&icmpsrc);
595 #endif
596 break;
597
598 /*
599 * No kernel processing for the following;
600 * just fall through to send to raw listener.
601 */
602 case ICMP_ECHOREPLY:
603 case ICMP_ROUTERADVERT:
604 case ICMP_ROUTERSOLICIT:
605 case ICMP_TSTAMPREPLY:
606 case ICMP_IREQREPLY:
607 case ICMP_MASKREPLY:
608 default:
609 break;
610 }
611
612 raw:
613 rip_input(m, off);
614 return;
615
616 freeit:
617 m_freem(m);
618 }
619
620 /*
621 * Reflect the ip packet back to the source
622 */
623 static void
624 icmp_reflect(m)
625 struct mbuf *m;
626 {
627 struct ip *ip = mtod(m, struct ip *);
628 struct ifaddr *ifa;
629 struct ifnet *ifn;
630 struct in_ifaddr *ia;
631 struct in_addr t;
632 struct mbuf *opts = 0;
633 int optlen = (ip->ip_hl << 2) - sizeof(struct ip);
634
635 if (IN_MULTICAST(ntohl(ip->ip_src.s_addr)) ||
636 IN_EXPERIMENTAL(ntohl(ip->ip_src.s_addr)) ||
637 IN_ZERONET(ntohl(ip->ip_src.s_addr)) ) {
638 m_freem(m); /* Bad return address */
639 icmpstat.icps_badaddr++;
640 goto done; /* Ip_output() will check for broadcast */
641 }
642
643 t = ip->ip_dst;
644 ip->ip_dst = ip->ip_src;
645
646 /*
647 * Source selection for ICMP replies:
648 *
649 * If the incoming packet was addressed directly to one of our
650 * own addresses, use dst as the src for the reply.
651 */
652 LIST_FOREACH(ia, INADDR_HASH(t.s_addr), ia_hash)
653 if (t.s_addr == IA_SIN(ia)->sin_addr.s_addr)
654 goto match;
655 /*
656 * If the incoming packet was addressed to one of our broadcast
657 * addresses, use the first non-broadcast address which corresponds
658 * to the incoming interface.
659 */
660 if (m->m_pkthdr.rcvif != NULL &&
661 m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
662 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrhead, ifa_link) {
663 if (ifa->ifa_addr->sa_family != AF_INET)
664 continue;
665 ia = ifatoia(ifa);
666 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
667 t.s_addr)
668 goto match;
669 }
670 }
671 /*
672 * If the packet was transiting through us, use the address of
673 * the interface the packet came through in. If that interface
674 * doesn't have a suitable IP address, the normal selection
675 * criteria apply.
676 */
677 if (icmp_rfi && m->m_pkthdr.rcvif != NULL) {
678 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrhead, ifa_link) {
679 if (ifa->ifa_addr->sa_family != AF_INET)
680 continue;
681 ia = ifatoia(ifa);
682 goto match;
683 }
684 }
685 /*
686 * If the incoming packet was not addressed directly to us, use
687 * designated interface for icmp replies specified by sysctl
688 * net.inet.icmp.reply_src (default not set). Otherwise continue
689 * with normal source selection.
690 */
691 if (reply_src[0] != '\0' && (ifn = ifunit(reply_src))) {
692 TAILQ_FOREACH(ifa, &ifn->if_addrhead, ifa_link) {
693 if (ifa->ifa_addr->sa_family != AF_INET)
694 continue;
695 ia = ifatoia(ifa);
696 goto match;
697 }
698 }
699 /*
700 * If the packet was transiting through us, use the address of
701 * the interface that is the closest to the packet source.
702 * When we don't have a route back to the packet source, stop here
703 * and drop the packet.
704 */
705 ia = ip_rtaddr(ip->ip_dst);
706 if (ia == NULL) {
707 m_freem(m);
708 icmpstat.icps_noroute++;
709 goto done;
710 }
711 match:
712 #ifdef MAC
713 mac_reflect_mbuf_icmp(m);
714 #endif
715 t = IA_SIN(ia)->sin_addr;
716 ip->ip_src = t;
717 ip->ip_ttl = ip_defttl;
718
719 if (optlen > 0) {
720 register u_char *cp;
721 int opt, cnt;
722 u_int len;
723
724 /*
725 * Retrieve any source routing from the incoming packet;
726 * add on any record-route or timestamp options.
727 */
728 cp = (u_char *) (ip + 1);
729 if ((opts = ip_srcroute(m)) == 0 &&
730 (opts = m_gethdr(M_DONTWAIT, MT_HEADER))) {
731 opts->m_len = sizeof(struct in_addr);
732 mtod(opts, struct in_addr *)->s_addr = 0;
733 }
734 if (opts) {
735 #ifdef ICMPPRINTFS
736 if (icmpprintfs)
737 printf("icmp_reflect optlen %d rt %d => ",
738 optlen, opts->m_len);
739 #endif
740 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) {
741 opt = cp[IPOPT_OPTVAL];
742 if (opt == IPOPT_EOL)
743 break;
744 if (opt == IPOPT_NOP)
745 len = 1;
746 else {
747 if (cnt < IPOPT_OLEN + sizeof(*cp))
748 break;
749 len = cp[IPOPT_OLEN];
750 if (len < IPOPT_OLEN + sizeof(*cp) ||
751 len > cnt)
752 break;
753 }
754 /*
755 * Should check for overflow, but it "can't happen"
756 */
757 if (opt == IPOPT_RR || opt == IPOPT_TS ||
758 opt == IPOPT_SECURITY) {
759 bcopy((caddr_t)cp,
760 mtod(opts, caddr_t) + opts->m_len, len);
761 opts->m_len += len;
762 }
763 }
764 /* Terminate & pad, if necessary */
765 cnt = opts->m_len % 4;
766 if (cnt) {
767 for (; cnt < 4; cnt++) {
768 *(mtod(opts, caddr_t) + opts->m_len) =
769 IPOPT_EOL;
770 opts->m_len++;
771 }
772 }
773 #ifdef ICMPPRINTFS
774 if (icmpprintfs)
775 printf("%d\n", opts->m_len);
776 #endif
777 }
778 /*
779 * Now strip out original options by copying rest of first
780 * mbuf's data back, and adjust the IP length.
781 */
782 ip->ip_len -= optlen;
783 ip->ip_v = IPVERSION;
784 ip->ip_hl = 5;
785 m->m_len -= optlen;
786 if (m->m_flags & M_PKTHDR)
787 m->m_pkthdr.len -= optlen;
788 optlen += sizeof(struct ip);
789 bcopy((caddr_t)ip + optlen, (caddr_t)(ip + 1),
790 (unsigned)(m->m_len - sizeof(struct ip)));
791 }
792 m_tag_delete_nonpersistent(m);
793 m->m_flags &= ~(M_BCAST|M_MCAST);
794 icmp_send(m, opts);
795 done:
796 if (opts)
797 (void)m_free(opts);
798 }
799
800 /*
801 * Send an icmp packet back to the ip level,
802 * after supplying a checksum.
803 */
804 static void
805 icmp_send(m, opts)
806 register struct mbuf *m;
807 struct mbuf *opts;
808 {
809 register struct ip *ip = mtod(m, struct ip *);
810 register int hlen;
811 register struct icmp *icp;
812
813 hlen = ip->ip_hl << 2;
814 m->m_data += hlen;
815 m->m_len -= hlen;
816 icp = mtod(m, struct icmp *);
817 icp->icmp_cksum = 0;
818 icp->icmp_cksum = in_cksum(m, ip->ip_len - hlen);
819 m->m_data -= hlen;
820 m->m_len += hlen;
821 m->m_pkthdr.rcvif = (struct ifnet *)0;
822 #ifdef ICMPPRINTFS
823 if (icmpprintfs) {
824 char buf[4 * sizeof "123"];
825 strcpy(buf, inet_ntoa(ip->ip_dst));
826 printf("icmp_send dst %s src %s\n",
827 buf, inet_ntoa(ip->ip_src));
828 }
829 #endif
830 (void) ip_output(m, opts, NULL, 0, NULL, NULL);
831 }
832
833 n_time
834 iptime()
835 {
836 struct timeval atv;
837 u_long t;
838
839 getmicrotime(&atv);
840 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000;
841 return (htonl(t));
842 }
843
844 /*
845 * Return the next larger or smaller MTU plateau (table from RFC 1191)
846 * given current value MTU. If DIR is less than zero, a larger plateau
847 * is returned; otherwise, a smaller value is returned.
848 */
849 int
850 ip_next_mtu(mtu, dir)
851 int mtu;
852 int dir;
853 {
854 static int mtutab[] = {
855 65535, 32000, 17914, 8166, 4352, 2002, 1492, 1280, 1006, 508,
856 296, 68, 0
857 };
858 int i, size;
859
860 size = (sizeof mtutab) / (sizeof mtutab[0]);
861 if (dir >= 0) {
862 for (i = 0; i < size; i++)
863 if (mtu > mtutab[i])
864 return mtutab[i];
865 } else {
866 for (i = size - 1; i >= 0; i--)
867 if (mtu < mtutab[i])
868 return mtutab[i];
869 if (mtu == mtutab[0])
870 return mtutab[0];
871 }
872 return 0;
873 }
874
875
876 /*
877 * badport_bandlim() - check for ICMP bandwidth limit
878 *
879 * Return 0 if it is ok to send an ICMP error response, -1 if we have
880 * hit our bandwidth limit and it is not ok.
881 *
882 * If icmplim is <= 0, the feature is disabled and 0 is returned.
883 *
884 * For now we separate the TCP and UDP subsystems w/ different 'which'
885 * values. We may eventually remove this separation (and simplify the
886 * code further).
887 *
888 * Note that the printing of the error message is delayed so we can
889 * properly print the icmp error rate that the system was trying to do
890 * (i.e. 22000/100 pps, etc...). This can cause long delays in printing
891 * the 'final' error, but it doesn't make sense to solve the printing
892 * delay with more complex code.
893 */
894
895 int
896 badport_bandlim(int which)
897 {
898 #define N(a) (sizeof (a) / sizeof (a[0]))
899 static struct rate {
900 const char *type;
901 struct timeval lasttime;
902 int curpps;
903 } rates[BANDLIM_MAX+1] = {
904 { "icmp unreach response" },
905 { "icmp ping response" },
906 { "icmp tstamp response" },
907 { "closed port RST response" },
908 { "open port RST response" }
909 };
910
911 /*
912 * Return ok status if feature disabled or argument out of range.
913 */
914 if (icmplim > 0 && (u_int) which < N(rates)) {
915 struct rate *r = &rates[which];
916 int opps = r->curpps;
917
918 if (!ppsratecheck(&r->lasttime, &r->curpps, icmplim))
919 return -1; /* discard packet */
920 /*
921 * If we've dropped below the threshold after having
922 * rate-limited traffic print the message. This preserves
923 * the previous behaviour at the expense of added complexity.
924 */
925 if (icmplim_output && opps > icmplim)
926 log(LOG_NOTICE, "Limiting %s from %d to %d packets/sec\n",
927 r->type, opps, icmplim);
928 }
929 return 0; /* okay to send packet */
930 #undef N
931 }
Cache object: d2da9d90197c42face9ed683f5878792
|