2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
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
29 * @(#)ip_icmp.c 8.2 (Berkeley) 1/4/94
32 #include <sys/cdefs.h>
33 __FBSDID("$FreeBSD$");
35 #include "opt_ipsec.h"
37 #include <sys/param.h>
38 #include <sys/systm.h>
40 #include <sys/protosw.h>
41 #include <sys/socket.h>
43 #include <sys/kernel.h>
44 #include <sys/sysctl.h>
47 #include <net/if_types.h>
48 #include <net/route.h>
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/ip_options.h>
59 #include <netinet/tcp.h>
60 #include <netinet/tcp_var.h>
61 #include <netinet/tcpip.h>
62 #include <netinet/icmp_var.h>
65 #include <netipsec/ipsec.h>
66 #include <netipsec/key.h>
69 #include <machine/in_cksum.h>
71 #include <security/mac/mac_framework.h>
74 * ICMP routines: error generation, receive packet processing, and
75 * routines to turnaround packets back to the originator, and
76 * host table maintenance routines.
79 VNET_DEFINE(struct icmpstat, icmpstat);
80 static VNET_DEFINE(int, icmpmaskrepl);
81 static VNET_DEFINE(u_int, icmpmaskfake);
82 static VNET_DEFINE(int, drop_redirect);
83 static VNET_DEFINE(int, log_redirect);
84 static VNET_DEFINE(int, icmplim);
85 static VNET_DEFINE(int, icmplim_output);
86 static VNET_DEFINE(char, reply_src[IFNAMSIZ]);
87 static VNET_DEFINE(int, icmp_rfi);
88 static VNET_DEFINE(int, icmp_quotelen);
89 static VNET_DEFINE(int, icmpbmcastecho);
91 #define V_icmpmaskrepl VNET(icmpmaskrepl)
92 #define V_icmpmaskfake VNET(icmpmaskfake)
93 #define V_drop_redirect VNET(drop_redirect)
94 #define V_log_redirect VNET(log_redirect)
95 #define V_icmplim VNET(icmplim)
96 #define V_icmplim_output VNET(icmplim_output)
97 #define V_reply_src VNET(reply_src)
98 #define V_icmp_rfi VNET(icmp_rfi)
99 #define V_icmp_quotelen VNET(icmp_quotelen)
100 #define V_icmpbmcastecho VNET(icmpbmcastecho)
102 SYSCTL_VNET_STRUCT(_net_inet_icmp, ICMPCTL_STATS, stats, CTLFLAG_RW,
103 &VNET_NAME(icmpstat), icmpstat, "");
105 SYSCTL_VNET_INT(_net_inet_icmp, ICMPCTL_MASKREPL, maskrepl, CTLFLAG_RW,
106 &VNET_NAME(icmpmaskrepl), 0,
107 "Reply to ICMP Address Mask Request packets.");
109 SYSCTL_VNET_UINT(_net_inet_icmp, OID_AUTO, maskfake, CTLFLAG_RW,
110 &VNET_NAME(icmpmaskfake), 0,
111 "Fake reply to ICMP Address Mask Request packets.");
113 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, drop_redirect, CTLFLAG_RW,
114 &VNET_NAME(drop_redirect), 0,
115 "Ignore ICMP redirects");
117 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, log_redirect, CTLFLAG_RW,
118 &VNET_NAME(log_redirect), 0,
119 "Log ICMP redirects to the console");
121 SYSCTL_VNET_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RW,
122 &VNET_NAME(icmplim), 0,
123 "Maximum number of ICMP responses per second");
125 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, icmplim_output, CTLFLAG_RW,
126 &VNET_NAME(icmplim_output), 0,
127 "Enable rate limiting of ICMP responses");
129 SYSCTL_VNET_STRING(_net_inet_icmp, OID_AUTO, reply_src, CTLFLAG_RW,
130 &VNET_NAME(reply_src), IFNAMSIZ,
131 "icmp reply source for non-local packets.");
133 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, reply_from_interface, CTLFLAG_RW,
134 &VNET_NAME(icmp_rfi), 0,
135 "ICMP reply from incoming interface for non-local packets");
137 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, quotelen, CTLFLAG_RW,
138 &VNET_NAME(icmp_quotelen), 0,
139 "Number of bytes from original packet to quote in ICMP reply");
142 * ICMP broadcast echo sysctl
145 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, bmcastecho, CTLFLAG_RW,
146 &VNET_NAME(icmpbmcastecho), 0,
154 static void icmp_reflect(struct mbuf *);
155 static void icmp_send(struct mbuf *, struct mbuf *);
157 extern struct protosw inetsw[];
168 V_icmplim_output = 1;
171 V_icmpbmcastecho = 0;
175 * Kernel module interface for updating icmpstat. The argument is an index
176 * into icmpstat treated as an array of u_long. While this encodes the
177 * general layout of icmpstat into the caller, it doesn't encode its
178 * location, so that future changes to add, for example, per-CPU stats
179 * support won't cause binary compatibility problems for kernel modules.
182 kmod_icmpstat_inc(int statnum)
185 (*((u_long *)&V_icmpstat + statnum))++;
189 * Generate an error packet of type error
190 * in response to bad packet ip.
193 icmp_error(struct mbuf *n, int type, int code, uint32_t dest, int mtu)
195 register struct ip *oip = mtod(n, struct ip *), *nip;
196 register unsigned oiphlen = oip->ip_hl << 2;
197 register struct icmp *icp;
198 register struct mbuf *m;
199 unsigned icmplen, icmpelen, nlen;
201 KASSERT((u_int)type <= ICMP_MAXTYPE, ("%s: illegal ICMP type", __func__));
204 printf("icmp_error(%p, %x, %d)\n", oip, type, code);
206 if (type != ICMP_REDIRECT)
207 ICMPSTAT_INC(icps_error);
210 * if the original packet was encrypted.
211 * if not the first fragment of message.
212 * in response to a multicast or broadcast packet.
213 * if the old packet protocol was an ICMP error message.
215 if (n->m_flags & M_DECRYPTED)
217 if (oip->ip_off & ~(IP_MF|IP_DF))
219 if (n->m_flags & (M_BCAST|M_MCAST))
221 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT &&
222 n->m_len >= oiphlen + ICMP_MINLEN &&
223 !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiphlen))->icmp_type)) {
224 ICMPSTAT_INC(icps_oldicmp);
227 /* Drop if IP header plus 8 bytes is not contignous in first mbuf. */
228 if (oiphlen + 8 > n->m_len)
231 * Calculate length to quote from original packet and
232 * prevent the ICMP mbuf from overflowing.
233 * Unfortunatly this is non-trivial since ip_forward()
234 * sends us truncated packets.
236 nlen = m_length(n, NULL);
237 if (oip->ip_p == IPPROTO_TCP) {
241 if (oiphlen + sizeof(struct tcphdr) > n->m_len &&
244 if (n->m_len < oiphlen + sizeof(struct tcphdr) &&
245 ((n = m_pullup(n, oiphlen + sizeof(struct tcphdr))) == NULL))
247 th = (struct tcphdr *)((caddr_t)oip + oiphlen);
248 tcphlen = th->th_off << 2;
249 if (tcphlen < sizeof(struct tcphdr))
251 if (oip->ip_len < oiphlen + tcphlen)
253 if (oiphlen + tcphlen > n->m_len && n->m_next == NULL)
255 if (n->m_len < oiphlen + tcphlen &&
256 ((n = m_pullup(n, oiphlen + tcphlen)) == NULL))
258 icmpelen = max(tcphlen, min(V_icmp_quotelen, oip->ip_len - oiphlen));
260 stdreply: icmpelen = max(8, min(V_icmp_quotelen, oip->ip_len - oiphlen));
262 icmplen = min(oiphlen + icmpelen, nlen);
263 if (icmplen < sizeof(struct ip))
266 if (MHLEN > sizeof(struct ip) + ICMP_MINLEN + icmplen)
267 m = m_gethdr(M_DONTWAIT, MT_DATA);
269 m = m_getcl(M_DONTWAIT, MT_DATA, M_PKTHDR);
273 mac_netinet_icmp_reply(n, m);
275 icmplen = min(icmplen, M_TRAILINGSPACE(m) - sizeof(struct ip) - ICMP_MINLEN);
276 m_align(m, ICMP_MINLEN + icmplen);
277 m->m_len = ICMP_MINLEN + icmplen;
279 /* XXX MRT make the outgoing packet use the same FIB
280 * that was associated with the incoming packet
282 M_SETFIB(m, M_GETFIB(n));
283 icp = mtod(m, struct icmp *);
284 ICMPSTAT_INC(icps_outhist[type]);
285 icp->icmp_type = type;
286 if (type == ICMP_REDIRECT)
287 icp->icmp_gwaddr.s_addr = dest;
291 * The following assignments assume an overlay with the
292 * just zeroed icmp_void field.
294 if (type == ICMP_PARAMPROB) {
295 icp->icmp_pptr = code;
297 } else if (type == ICMP_UNREACH &&
298 code == ICMP_UNREACH_NEEDFRAG && mtu) {
299 icp->icmp_nextmtu = htons(mtu);
302 icp->icmp_code = code;
305 * Copy the quotation into ICMP message and
306 * convert quoted IP header back to network representation.
308 m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip);
310 nip->ip_len = htons(nip->ip_len);
311 nip->ip_off = htons(nip->ip_off);
314 * Set up ICMP message mbuf and copy old IP header (without options
315 * in front of ICMP message.
316 * If the original mbuf was meant to bypass the firewall, the error
317 * reply should bypass as well.
319 m->m_flags |= n->m_flags & M_SKIP_FIREWALL;
320 m->m_data -= sizeof(struct ip);
321 m->m_len += sizeof(struct ip);
322 m->m_pkthdr.len = m->m_len;
323 m->m_pkthdr.rcvif = n->m_pkthdr.rcvif;
324 nip = mtod(m, struct ip *);
325 bcopy((caddr_t)oip, (caddr_t)nip, sizeof(struct ip));
326 nip->ip_len = m->m_len;
327 nip->ip_v = IPVERSION;
329 nip->ip_p = IPPROTO_ICMP;
338 * Process a received ICMP message.
341 icmp_input(struct mbuf *m, int off)
344 struct in_ifaddr *ia;
345 struct ip *ip = mtod(m, struct ip *);
346 struct sockaddr_in icmpsrc, icmpdst, icmpgw;
348 int icmplen = ip->ip_len;
350 void (*ctlfunc)(int, struct sockaddr *, void *);
354 * Locate icmp structure in mbuf, and check
355 * that not corrupted and of at least minimum length.
359 char buf[4 * sizeof "123"];
360 strcpy(buf, inet_ntoa(ip->ip_src));
361 printf("icmp_input from %s to %s, len %d\n",
362 buf, inet_ntoa(ip->ip_dst), icmplen);
365 if (icmplen < ICMP_MINLEN) {
366 ICMPSTAT_INC(icps_tooshort);
369 i = hlen + min(icmplen, ICMP_ADVLENMIN);
370 if (m->m_len < i && (m = m_pullup(m, i)) == 0) {
371 ICMPSTAT_INC(icps_tooshort);
374 ip = mtod(m, struct ip *);
377 icp = mtod(m, struct icmp *);
378 if (in_cksum(m, icmplen)) {
379 ICMPSTAT_INC(icps_checksum);
385 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) {
387 * Deliver very specific ICMP type only.
389 switch (icp->icmp_type) {
400 printf("icmp_input, type %d code %d\n", icp->icmp_type,
405 * Message type specific processing.
407 if (icp->icmp_type > ICMP_MAXTYPE)
411 bzero(&icmpsrc, sizeof(icmpsrc));
412 icmpsrc.sin_len = sizeof(struct sockaddr_in);
413 icmpsrc.sin_family = AF_INET;
414 bzero(&icmpdst, sizeof(icmpdst));
415 icmpdst.sin_len = sizeof(struct sockaddr_in);
416 icmpdst.sin_family = AF_INET;
417 bzero(&icmpgw, sizeof(icmpgw));
418 icmpgw.sin_len = sizeof(struct sockaddr_in);
419 icmpgw.sin_family = AF_INET;
421 ICMPSTAT_INC(icps_inhist[icp->icmp_type]);
422 code = icp->icmp_code;
423 switch (icp->icmp_type) {
427 case ICMP_UNREACH_NET:
428 case ICMP_UNREACH_HOST:
429 case ICMP_UNREACH_SRCFAIL:
430 case ICMP_UNREACH_NET_UNKNOWN:
431 case ICMP_UNREACH_HOST_UNKNOWN:
432 case ICMP_UNREACH_ISOLATED:
433 case ICMP_UNREACH_TOSNET:
434 case ICMP_UNREACH_TOSHOST:
435 case ICMP_UNREACH_HOST_PRECEDENCE:
436 case ICMP_UNREACH_PRECEDENCE_CUTOFF:
437 code = PRC_UNREACH_NET;
440 case ICMP_UNREACH_NEEDFRAG:
445 * RFC 1122, Sections 3.2.2.1 and 4.2.3.9.
446 * Treat subcodes 2,3 as immediate RST
448 case ICMP_UNREACH_PROTOCOL:
449 case ICMP_UNREACH_PORT:
450 code = PRC_UNREACH_PORT;
453 case ICMP_UNREACH_NET_PROHIB:
454 case ICMP_UNREACH_HOST_PROHIB:
455 case ICMP_UNREACH_FILTER_PROHIB:
456 code = PRC_UNREACH_ADMIN_PROHIB;
467 code += PRC_TIMXCEED_INTRANS;
473 code = PRC_PARAMPROB;
476 case ICMP_SOURCEQUENCH:
482 * Problem with datagram; advise higher level routines.
484 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
485 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) {
486 ICMPSTAT_INC(icps_badlen);
489 icp->icmp_ip.ip_len = ntohs(icp->icmp_ip.ip_len);
490 /* Discard ICMP's in response to multicast packets */
491 if (IN_MULTICAST(ntohl(icp->icmp_ip.ip_dst.s_addr)))
495 printf("deliver to protocol %d\n", icp->icmp_ip.ip_p);
497 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
499 * XXX if the packet contains [IPv4 AH TCP], we can't make a
500 * notification to TCP layer.
502 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput;
504 (*ctlfunc)(code, (struct sockaddr *)&icmpsrc,
505 (void *)&icp->icmp_ip);
509 ICMPSTAT_INC(icps_badcode);
513 if (!V_icmpbmcastecho
514 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
515 ICMPSTAT_INC(icps_bmcastecho);
518 icp->icmp_type = ICMP_ECHOREPLY;
519 if (badport_bandlim(BANDLIM_ICMP_ECHO) < 0)
525 if (!V_icmpbmcastecho
526 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
527 ICMPSTAT_INC(icps_bmcasttstamp);
530 if (icmplen < ICMP_TSLEN) {
531 ICMPSTAT_INC(icps_badlen);
534 icp->icmp_type = ICMP_TSTAMPREPLY;
535 icp->icmp_rtime = iptime();
536 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */
537 if (badport_bandlim(BANDLIM_ICMP_TSTAMP) < 0)
543 if (V_icmpmaskrepl == 0)
546 * We are not able to respond with all ones broadcast
547 * unless we receive it over a point-to-point interface.
549 if (icmplen < ICMP_MASKLEN)
551 switch (ip->ip_dst.s_addr) {
553 case INADDR_BROADCAST:
555 icmpdst.sin_addr = ip->ip_src;
559 icmpdst.sin_addr = ip->ip_dst;
561 ia = (struct in_ifaddr *)ifaof_ifpforaddr(
562 (struct sockaddr *)&icmpdst, m->m_pkthdr.rcvif);
565 if (ia->ia_ifp == NULL) {
566 ifa_free(&ia->ia_ifa);
569 icp->icmp_type = ICMP_MASKREPLY;
570 if (V_icmpmaskfake == 0)
571 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr;
573 icp->icmp_mask = V_icmpmaskfake;
574 if (ip->ip_src.s_addr == 0) {
575 if (ia->ia_ifp->if_flags & IFF_BROADCAST)
576 ip->ip_src = satosin(&ia->ia_broadaddr)->sin_addr;
577 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT)
578 ip->ip_src = satosin(&ia->ia_dstaddr)->sin_addr;
580 ifa_free(&ia->ia_ifa);
582 ip->ip_len += hlen; /* since ip_input deducts this */
583 ICMPSTAT_INC(icps_reflect);
584 ICMPSTAT_INC(icps_outhist[icp->icmp_type]);
589 if (V_log_redirect) {
592 src = ntohl(ip->ip_src.s_addr);
593 dst = ntohl(icp->icmp_ip.ip_dst.s_addr);
594 gw = ntohl(icp->icmp_gwaddr.s_addr);
595 printf("icmp redirect from %d.%d.%d.%d: "
596 "%d.%d.%d.%d => %d.%d.%d.%d\n",
597 (int)(src >> 24), (int)((src >> 16) & 0xff),
598 (int)((src >> 8) & 0xff), (int)(src & 0xff),
599 (int)(dst >> 24), (int)((dst >> 16) & 0xff),
600 (int)((dst >> 8) & 0xff), (int)(dst & 0xff),
601 (int)(gw >> 24), (int)((gw >> 16) & 0xff),
602 (int)((gw >> 8) & 0xff), (int)(gw & 0xff));
605 * RFC1812 says we must ignore ICMP redirects if we
606 * are acting as router.
608 if (V_drop_redirect || V_ipforwarding)
612 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
613 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) {
614 ICMPSTAT_INC(icps_badlen);
618 * Short circuit routing redirects to force
619 * immediate change in the kernel's routing
620 * tables. The message is also handed to anyone
621 * listening on a raw socket (e.g. the routing
622 * daemon for use in updating its tables).
624 icmpgw.sin_addr = ip->ip_src;
625 icmpdst.sin_addr = icp->icmp_gwaddr;
628 char buf[4 * sizeof "123"];
629 strcpy(buf, inet_ntoa(icp->icmp_ip.ip_dst));
631 printf("redirect dst %s to %s\n",
632 buf, inet_ntoa(icp->icmp_gwaddr));
635 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
636 for ( fibnum = 0; fibnum < rt_numfibs; fibnum++) {
637 in_rtredirect((struct sockaddr *)&icmpsrc,
638 (struct sockaddr *)&icmpdst,
639 (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST,
640 (struct sockaddr *)&icmpgw, fibnum);
642 pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&icmpsrc);
644 key_sa_routechange((struct sockaddr *)&icmpsrc);
649 * No kernel processing for the following;
650 * just fall through to send to raw listener.
653 case ICMP_ROUTERADVERT:
654 case ICMP_ROUTERSOLICIT:
655 case ICMP_TSTAMPREPLY:
671 * Reflect the ip packet back to the source
674 icmp_reflect(struct mbuf *m)
676 struct ip *ip = mtod(m, struct ip *);
679 struct in_ifaddr *ia;
681 struct mbuf *opts = 0;
682 int optlen = (ip->ip_hl << 2) - sizeof(struct ip);
684 if (IN_MULTICAST(ntohl(ip->ip_src.s_addr)) ||
685 IN_EXPERIMENTAL(ntohl(ip->ip_src.s_addr)) ||
686 IN_ZERONET(ntohl(ip->ip_src.s_addr)) ) {
687 m_freem(m); /* Bad return address */
688 ICMPSTAT_INC(icps_badaddr);
689 goto done; /* Ip_output() will check for broadcast */
693 ip->ip_dst = ip->ip_src;
696 * Source selection for ICMP replies:
698 * If the incoming packet was addressed directly to one of our
699 * own addresses, use dst as the src for the reply.
702 LIST_FOREACH(ia, INADDR_HASH(t.s_addr), ia_hash) {
703 if (t.s_addr == IA_SIN(ia)->sin_addr.s_addr) {
704 t = IA_SIN(ia)->sin_addr;
712 * If the incoming packet was addressed to one of our broadcast
713 * addresses, use the first non-broadcast address which corresponds
714 * to the incoming interface.
716 ifp = m->m_pkthdr.rcvif;
717 if (ifp != NULL && ifp->if_flags & IFF_BROADCAST) {
719 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
720 if (ifa->ifa_addr->sa_family != AF_INET)
723 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
725 t = IA_SIN(ia)->sin_addr;
733 * If the packet was transiting through us, use the address of
734 * the interface the packet came through in. If that interface
735 * doesn't have a suitable IP address, the normal selection
738 if (V_icmp_rfi && ifp != NULL) {
740 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
741 if (ifa->ifa_addr->sa_family != AF_INET)
744 t = IA_SIN(ia)->sin_addr;
751 * If the incoming packet was not addressed directly to us, use
752 * designated interface for icmp replies specified by sysctl
753 * net.inet.icmp.reply_src (default not set). Otherwise continue
754 * with normal source selection.
756 if (V_reply_src[0] != '\0' && (ifp = ifunit(V_reply_src))) {
758 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
759 if (ifa->ifa_addr->sa_family != AF_INET)
762 t = IA_SIN(ia)->sin_addr;
769 * If the packet was transiting through us, use the address of
770 * the interface that is the closest to the packet source.
771 * When we don't have a route back to the packet source, stop here
772 * and drop the packet.
774 ia = ip_rtaddr(ip->ip_dst, M_GETFIB(m));
777 ICMPSTAT_INC(icps_noroute);
780 t = IA_SIN(ia)->sin_addr;
781 ifa_free(&ia->ia_ifa);
784 mac_netinet_icmp_replyinplace(m);
787 ip->ip_ttl = V_ip_defttl;
795 * Retrieve any source routing from the incoming packet;
796 * add on any record-route or timestamp options.
798 cp = (u_char *) (ip + 1);
799 if ((opts = ip_srcroute(m)) == 0 &&
800 (opts = m_gethdr(M_DONTWAIT, MT_DATA))) {
801 opts->m_len = sizeof(struct in_addr);
802 mtod(opts, struct in_addr *)->s_addr = 0;
807 printf("icmp_reflect optlen %d rt %d => ",
808 optlen, opts->m_len);
810 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) {
811 opt = cp[IPOPT_OPTVAL];
812 if (opt == IPOPT_EOL)
814 if (opt == IPOPT_NOP)
817 if (cnt < IPOPT_OLEN + sizeof(*cp))
819 len = cp[IPOPT_OLEN];
820 if (len < IPOPT_OLEN + sizeof(*cp) ||
825 * Should check for overflow, but it "can't happen"
827 if (opt == IPOPT_RR || opt == IPOPT_TS ||
828 opt == IPOPT_SECURITY) {
830 mtod(opts, caddr_t) + opts->m_len, len);
834 /* Terminate & pad, if necessary */
835 cnt = opts->m_len % 4;
837 for (; cnt < 4; cnt++) {
838 *(mtod(opts, caddr_t) + opts->m_len) =
845 printf("%d\n", opts->m_len);
849 * Now strip out original options by copying rest of first
850 * mbuf's data back, and adjust the IP length.
852 ip->ip_len -= optlen;
853 ip->ip_v = IPVERSION;
856 if (m->m_flags & M_PKTHDR)
857 m->m_pkthdr.len -= optlen;
858 optlen += sizeof(struct ip);
859 bcopy((caddr_t)ip + optlen, (caddr_t)(ip + 1),
860 (unsigned)(m->m_len - sizeof(struct ip)));
862 m_tag_delete_nonpersistent(m);
863 m->m_flags &= ~(M_BCAST|M_MCAST);
871 * Send an icmp packet back to the ip level,
872 * after supplying a checksum.
875 icmp_send(struct mbuf *m, struct mbuf *opts)
877 register struct ip *ip = mtod(m, struct ip *);
879 register struct icmp *icp;
881 hlen = ip->ip_hl << 2;
884 icp = mtod(m, struct icmp *);
886 icp->icmp_cksum = in_cksum(m, ip->ip_len - hlen);
889 m->m_pkthdr.rcvif = (struct ifnet *)0;
892 char buf[4 * sizeof "123"];
893 strcpy(buf, inet_ntoa(ip->ip_dst));
894 printf("icmp_send dst %s src %s\n",
895 buf, inet_ntoa(ip->ip_src));
898 (void) ip_output(m, opts, NULL, 0, NULL, NULL);
902 * Return milliseconds since 00:00 GMT in network format.
911 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000;
916 * Return the next larger or smaller MTU plateau (table from RFC 1191)
917 * given current value MTU. If DIR is less than zero, a larger plateau
918 * is returned; otherwise, a smaller value is returned.
921 ip_next_mtu(int mtu, int dir)
923 static int mtutab[] = {
924 65535, 32000, 17914, 8166, 4352, 2002, 1492, 1280, 1006, 508,
929 size = (sizeof mtutab) / (sizeof mtutab[0]);
931 for (i = 0; i < size; i++)
935 for (i = size - 1; i >= 0; i--)
938 if (mtu == mtutab[0])
946 * badport_bandlim() - check for ICMP bandwidth limit
948 * Return 0 if it is ok to send an ICMP error response, -1 if we have
949 * hit our bandwidth limit and it is not ok.
951 * If icmplim is <= 0, the feature is disabled and 0 is returned.
953 * For now we separate the TCP and UDP subsystems w/ different 'which'
954 * values. We may eventually remove this separation (and simplify the
957 * Note that the printing of the error message is delayed so we can
958 * properly print the icmp error rate that the system was trying to do
959 * (i.e. 22000/100 pps, etc...). This can cause long delays in printing
960 * the 'final' error, but it doesn't make sense to solve the printing
961 * delay with more complex code.
965 badport_bandlim(int which)
968 #define N(a) (sizeof (a) / sizeof (a[0]))
971 struct timeval lasttime;
973 } rates[BANDLIM_MAX+1] = {
974 { "icmp unreach response" },
975 { "icmp ping response" },
976 { "icmp tstamp response" },
977 { "closed port RST response" },
978 { "open port RST response" },
979 { "icmp6 unreach response" }
983 * Return ok status if feature disabled or argument out of range.
985 if (V_icmplim > 0 && (u_int) which < N(rates)) {
986 struct rate *r = &rates[which];
987 int opps = r->curpps;
989 if (!ppsratecheck(&r->lasttime, &r->curpps, V_icmplim))
990 return -1; /* discard packet */
992 * If we've dropped below the threshold after having
993 * rate-limited traffic print the message. This preserves
994 * the previous behaviour at the expense of added complexity.
996 if (V_icmplim_output && opps > V_icmplim)
997 printf("Limiting %s from %d to %d packets/sec\n",
998 r->type, opps, V_icmplim);
1000 return 0; /* okay to send packet */