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("$FreeBSD$");
36 #include "opt_ipsec.h"
38 #include <sys/param.h>
39 #include <sys/systm.h>
41 #include <sys/protosw.h>
42 #include <sys/socket.h>
44 #include <sys/kernel.h>
45 #include <sys/sysctl.h>
46 #include <sys/syslog.h>
49 #include <net/if_var.h>
50 #include <net/if_types.h>
51 #include <net/route.h>
54 #include <netinet/in.h>
55 #include <netinet/in_pcb.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/ip_options.h>
62 #include <netinet/tcp.h>
63 #include <netinet/tcp_var.h>
64 #include <netinet/tcpip.h>
65 #include <netinet/icmp_var.h>
69 #include <netipsec/ipsec.h>
70 #include <netipsec/key.h>
73 #include <machine/in_cksum.h>
75 #include <security/mac/mac_framework.h>
79 * ICMP routines: error generation, receive packet processing, and
80 * routines to turnaround packets back to the originator, and
81 * host table maintenance routines.
83 static VNET_DEFINE(int, icmplim) = 200;
84 #define V_icmplim VNET(icmplim)
85 SYSCTL_VNET_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RW,
86 &VNET_NAME(icmplim), 0,
87 "Maximum number of ICMP responses per second");
89 static VNET_DEFINE(int, icmplim_output) = 1;
90 #define V_icmplim_output VNET(icmplim_output)
91 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, icmplim_output, CTLFLAG_RW,
92 &VNET_NAME(icmplim_output), 0,
93 "Enable logging of ICMP response rate limiting");
96 VNET_PCPUSTAT_DEFINE(struct icmpstat, icmpstat);
97 VNET_PCPUSTAT_SYSINIT(icmpstat);
98 SYSCTL_VNET_PCPUSTAT(_net_inet_icmp, ICMPCTL_STATS, stats, struct icmpstat,
99 icmpstat, "ICMP statistics (struct icmpstat, netinet/icmp_var.h)");
102 VNET_PCPUSTAT_SYSUNINIT(icmpstat);
105 static VNET_DEFINE(int, icmpmaskrepl) = 0;
106 #define V_icmpmaskrepl VNET(icmpmaskrepl)
107 SYSCTL_VNET_INT(_net_inet_icmp, ICMPCTL_MASKREPL, maskrepl, CTLFLAG_RW,
108 &VNET_NAME(icmpmaskrepl), 0,
109 "Reply to ICMP Address Mask Request packets.");
111 static VNET_DEFINE(u_int, icmpmaskfake) = 0;
112 #define V_icmpmaskfake VNET(icmpmaskfake)
113 SYSCTL_VNET_UINT(_net_inet_icmp, OID_AUTO, maskfake, CTLFLAG_RW,
114 &VNET_NAME(icmpmaskfake), 0,
115 "Fake reply to ICMP Address Mask Request packets.");
117 VNET_DEFINE(int, drop_redirect) = 0;
119 static VNET_DEFINE(int, log_redirect) = 0;
120 #define V_log_redirect VNET(log_redirect)
121 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, log_redirect, CTLFLAG_RW,
122 &VNET_NAME(log_redirect), 0,
123 "Log ICMP redirects to the console");
125 static VNET_DEFINE(char, reply_src[IFNAMSIZ]);
126 #define V_reply_src VNET(reply_src)
127 SYSCTL_VNET_STRING(_net_inet_icmp, OID_AUTO, reply_src, CTLFLAG_RW,
128 &VNET_NAME(reply_src), IFNAMSIZ,
129 "icmp reply source for non-local packets.");
131 static VNET_DEFINE(int, icmp_rfi) = 0;
132 #define V_icmp_rfi VNET(icmp_rfi)
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 static VNET_DEFINE(int, icmp_quotelen) = 8;
138 #define V_icmp_quotelen VNET(icmp_quotelen)
139 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, quotelen, CTLFLAG_RW,
140 &VNET_NAME(icmp_quotelen), 0,
141 "Number of bytes from original packet to quote in ICMP reply");
144 * ICMP broadcast echo sysctl
146 static VNET_DEFINE(int, icmpbmcastecho) = 0;
147 #define V_icmpbmcastecho VNET(icmpbmcastecho)
148 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, bmcastecho, CTLFLAG_RW,
149 &VNET_NAME(icmpbmcastecho), 0,
157 static void icmp_reflect(struct mbuf *);
158 static void icmp_send(struct mbuf *, struct mbuf *);
160 extern struct protosw inetsw[];
163 sysctl_net_icmp_drop_redir(SYSCTL_HANDLER_ARGS)
167 struct radix_node_head *rnh;
169 new = V_drop_redirect;
170 error = sysctl_handle_int(oidp, &new, 0, req);
171 if (error == 0 && req->newptr) {
172 new = (new != 0) ? 1 : 0;
174 if (new == V_drop_redirect)
177 for (i = 0; i < rt_numfibs; i++) {
178 if ((rnh = rt_tables_get_rnh(i, AF_INET)) == NULL)
180 RADIX_NODE_HEAD_LOCK(rnh);
181 in_setmatchfunc(rnh, new);
182 RADIX_NODE_HEAD_UNLOCK(rnh);
185 V_drop_redirect = new;
191 SYSCTL_VNET_PROC(_net_inet_icmp, OID_AUTO, drop_redirect,
192 CTLTYPE_INT|CTLFLAG_RW, 0, 0,
193 sysctl_net_icmp_drop_redir, "I", "Ignore ICMP redirects");
196 * Kernel module interface for updating icmpstat. The argument is an index
197 * into icmpstat treated as an array of u_long. While this encodes the
198 * general layout of icmpstat into the caller, it doesn't encode its
199 * location, so that future changes to add, for example, per-CPU stats
200 * support won't cause binary compatibility problems for kernel modules.
203 kmod_icmpstat_inc(int statnum)
206 counter_u64_add(VNET(icmpstat)[statnum], 1);
210 * Generate an error packet of type error
211 * in response to bad packet ip.
214 icmp_error(struct mbuf *n, int type, int code, uint32_t dest, int mtu)
216 register struct ip *oip = mtod(n, struct ip *), *nip;
217 register unsigned oiphlen = oip->ip_hl << 2;
218 register struct icmp *icp;
219 register struct mbuf *m;
220 unsigned icmplen, icmpelen, nlen;
222 KASSERT((u_int)type <= ICMP_MAXTYPE, ("%s: illegal ICMP type", __func__));
225 printf("icmp_error(%p, %x, %d)\n", oip, type, code);
227 if (type != ICMP_REDIRECT)
228 ICMPSTAT_INC(icps_error);
231 * if the original packet was encrypted.
232 * if not the first fragment of message.
233 * in response to a multicast or broadcast packet.
234 * if the old packet protocol was an ICMP error message.
236 if (n->m_flags & M_DECRYPTED)
238 if (oip->ip_off & htons(~(IP_MF|IP_DF)))
240 if (n->m_flags & (M_BCAST|M_MCAST))
242 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT &&
243 n->m_len >= oiphlen + ICMP_MINLEN &&
244 !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiphlen))->icmp_type)) {
245 ICMPSTAT_INC(icps_oldicmp);
248 /* Drop if IP header plus 8 bytes is not contignous in first mbuf. */
249 if (oiphlen + 8 > n->m_len)
252 * Calculate length to quote from original packet and
253 * prevent the ICMP mbuf from overflowing.
254 * Unfortunatly this is non-trivial since ip_forward()
255 * sends us truncated packets.
257 nlen = m_length(n, NULL);
258 if (oip->ip_p == IPPROTO_TCP) {
262 if (oiphlen + sizeof(struct tcphdr) > n->m_len &&
265 if (n->m_len < oiphlen + sizeof(struct tcphdr) &&
266 ((n = m_pullup(n, oiphlen + sizeof(struct tcphdr))) == NULL))
268 th = (struct tcphdr *)((caddr_t)oip + oiphlen);
269 tcphlen = th->th_off << 2;
270 if (tcphlen < sizeof(struct tcphdr))
272 if (ntohs(oip->ip_len) < oiphlen + tcphlen)
274 if (oiphlen + tcphlen > n->m_len && n->m_next == NULL)
276 if (n->m_len < oiphlen + tcphlen &&
277 ((n = m_pullup(n, oiphlen + tcphlen)) == NULL))
279 icmpelen = max(tcphlen, min(V_icmp_quotelen,
280 ntohs(oip->ip_len) - oiphlen));
282 stdreply: icmpelen = max(8, min(V_icmp_quotelen, ntohs(oip->ip_len) - oiphlen));
284 icmplen = min(oiphlen + icmpelen, nlen);
285 if (icmplen < sizeof(struct ip))
288 if (MHLEN > sizeof(struct ip) + ICMP_MINLEN + icmplen)
289 m = m_gethdr(M_NOWAIT, MT_DATA);
291 m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
295 mac_netinet_icmp_reply(n, m);
297 icmplen = min(icmplen, M_TRAILINGSPACE(m) - sizeof(struct ip) - ICMP_MINLEN);
298 m_align(m, ICMP_MINLEN + icmplen);
299 m->m_len = ICMP_MINLEN + icmplen;
301 /* XXX MRT make the outgoing packet use the same FIB
302 * that was associated with the incoming packet
304 M_SETFIB(m, M_GETFIB(n));
305 icp = mtod(m, struct icmp *);
306 ICMPSTAT_INC(icps_outhist[type]);
307 icp->icmp_type = type;
308 if (type == ICMP_REDIRECT)
309 icp->icmp_gwaddr.s_addr = dest;
313 * The following assignments assume an overlay with the
314 * just zeroed icmp_void field.
316 if (type == ICMP_PARAMPROB) {
317 icp->icmp_pptr = code;
319 } else if (type == ICMP_UNREACH &&
320 code == ICMP_UNREACH_NEEDFRAG && mtu) {
321 icp->icmp_nextmtu = htons(mtu);
324 icp->icmp_code = code;
327 * Copy the quotation into ICMP message and
328 * convert quoted IP header back to network representation.
330 m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip);
334 * Set up ICMP message mbuf and copy old IP header (without options
335 * in front of ICMP message.
336 * If the original mbuf was meant to bypass the firewall, the error
337 * reply should bypass as well.
339 m->m_flags |= n->m_flags & M_SKIP_FIREWALL;
340 m->m_data -= sizeof(struct ip);
341 m->m_len += sizeof(struct ip);
342 m->m_pkthdr.len = m->m_len;
343 m->m_pkthdr.rcvif = n->m_pkthdr.rcvif;
344 nip = mtod(m, struct ip *);
345 bcopy((caddr_t)oip, (caddr_t)nip, sizeof(struct ip));
346 nip->ip_len = htons(m->m_len);
347 nip->ip_v = IPVERSION;
349 nip->ip_p = IPPROTO_ICMP;
359 * Process a received ICMP message.
362 icmp_input(struct mbuf *m, int off)
365 struct in_ifaddr *ia;
366 struct ip *ip = mtod(m, struct ip *);
367 struct sockaddr_in icmpsrc, icmpdst, icmpgw;
369 int icmplen = ntohs(ip->ip_len) - off;
371 void (*ctlfunc)(int, struct sockaddr *, void *);
375 * Locate icmp structure in mbuf, and check
376 * that not corrupted and of at least minimum length.
380 char buf[4 * sizeof "123"];
381 strcpy(buf, inet_ntoa(ip->ip_src));
382 printf("icmp_input from %s to %s, len %d\n",
383 buf, inet_ntoa(ip->ip_dst), icmplen);
386 if (icmplen < ICMP_MINLEN) {
387 ICMPSTAT_INC(icps_tooshort);
390 i = hlen + min(icmplen, ICMP_ADVLENMIN);
391 if (m->m_len < i && (m = m_pullup(m, i)) == NULL) {
392 ICMPSTAT_INC(icps_tooshort);
395 ip = mtod(m, struct ip *);
398 icp = mtod(m, struct icmp *);
399 if (in_cksum(m, icmplen)) {
400 ICMPSTAT_INC(icps_checksum);
406 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) {
408 * Deliver very specific ICMP type only.
410 switch (icp->icmp_type) {
421 printf("icmp_input, type %d code %d\n", icp->icmp_type,
426 * Message type specific processing.
428 if (icp->icmp_type > ICMP_MAXTYPE)
432 bzero(&icmpsrc, sizeof(icmpsrc));
433 icmpsrc.sin_len = sizeof(struct sockaddr_in);
434 icmpsrc.sin_family = AF_INET;
435 bzero(&icmpdst, sizeof(icmpdst));
436 icmpdst.sin_len = sizeof(struct sockaddr_in);
437 icmpdst.sin_family = AF_INET;
438 bzero(&icmpgw, sizeof(icmpgw));
439 icmpgw.sin_len = sizeof(struct sockaddr_in);
440 icmpgw.sin_family = AF_INET;
442 ICMPSTAT_INC(icps_inhist[icp->icmp_type]);
443 code = icp->icmp_code;
444 switch (icp->icmp_type) {
448 case ICMP_UNREACH_NET:
449 case ICMP_UNREACH_HOST:
450 case ICMP_UNREACH_SRCFAIL:
451 case ICMP_UNREACH_NET_UNKNOWN:
452 case ICMP_UNREACH_HOST_UNKNOWN:
453 case ICMP_UNREACH_ISOLATED:
454 case ICMP_UNREACH_TOSNET:
455 case ICMP_UNREACH_TOSHOST:
456 case ICMP_UNREACH_HOST_PRECEDENCE:
457 case ICMP_UNREACH_PRECEDENCE_CUTOFF:
458 code = PRC_UNREACH_NET;
461 case ICMP_UNREACH_NEEDFRAG:
466 * RFC 1122, Sections 3.2.2.1 and 4.2.3.9.
467 * Treat subcodes 2,3 as immediate RST
469 case ICMP_UNREACH_PROTOCOL:
470 case ICMP_UNREACH_PORT:
471 code = PRC_UNREACH_PORT;
474 case ICMP_UNREACH_NET_PROHIB:
475 case ICMP_UNREACH_HOST_PROHIB:
476 case ICMP_UNREACH_FILTER_PROHIB:
477 code = PRC_UNREACH_ADMIN_PROHIB;
488 code += PRC_TIMXCEED_INTRANS;
494 code = PRC_PARAMPROB;
497 case ICMP_SOURCEQUENCH:
503 * Problem with datagram; advise higher level routines.
505 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
506 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) {
507 ICMPSTAT_INC(icps_badlen);
510 /* Discard ICMP's in response to multicast packets */
511 if (IN_MULTICAST(ntohl(icp->icmp_ip.ip_dst.s_addr)))
515 printf("deliver to protocol %d\n", icp->icmp_ip.ip_p);
517 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
519 * XXX if the packet contains [IPv4 AH TCP], we can't make a
520 * notification to TCP layer.
522 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput;
524 (*ctlfunc)(code, (struct sockaddr *)&icmpsrc,
525 (void *)&icp->icmp_ip);
529 ICMPSTAT_INC(icps_badcode);
533 if (!V_icmpbmcastecho
534 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
535 ICMPSTAT_INC(icps_bmcastecho);
538 icp->icmp_type = ICMP_ECHOREPLY;
539 if (badport_bandlim(BANDLIM_ICMP_ECHO) < 0)
545 if (!V_icmpbmcastecho
546 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
547 ICMPSTAT_INC(icps_bmcasttstamp);
550 if (icmplen < ICMP_TSLEN) {
551 ICMPSTAT_INC(icps_badlen);
554 icp->icmp_type = ICMP_TSTAMPREPLY;
555 icp->icmp_rtime = iptime();
556 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */
557 if (badport_bandlim(BANDLIM_ICMP_TSTAMP) < 0)
563 if (V_icmpmaskrepl == 0)
566 * We are not able to respond with all ones broadcast
567 * unless we receive it over a point-to-point interface.
569 if (icmplen < ICMP_MASKLEN)
571 switch (ip->ip_dst.s_addr) {
573 case INADDR_BROADCAST:
575 icmpdst.sin_addr = ip->ip_src;
579 icmpdst.sin_addr = ip->ip_dst;
581 ia = (struct in_ifaddr *)ifaof_ifpforaddr(
582 (struct sockaddr *)&icmpdst, m->m_pkthdr.rcvif);
585 if (ia->ia_ifp == NULL) {
586 ifa_free(&ia->ia_ifa);
589 icp->icmp_type = ICMP_MASKREPLY;
590 if (V_icmpmaskfake == 0)
591 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr;
593 icp->icmp_mask = V_icmpmaskfake;
594 if (ip->ip_src.s_addr == 0) {
595 if (ia->ia_ifp->if_flags & IFF_BROADCAST)
596 ip->ip_src = satosin(&ia->ia_broadaddr)->sin_addr;
597 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT)
598 ip->ip_src = satosin(&ia->ia_dstaddr)->sin_addr;
600 ifa_free(&ia->ia_ifa);
602 ICMPSTAT_INC(icps_reflect);
603 ICMPSTAT_INC(icps_outhist[icp->icmp_type]);
608 if (V_log_redirect) {
611 src = ntohl(ip->ip_src.s_addr);
612 dst = ntohl(icp->icmp_ip.ip_dst.s_addr);
613 gw = ntohl(icp->icmp_gwaddr.s_addr);
614 printf("icmp redirect from %d.%d.%d.%d: "
615 "%d.%d.%d.%d => %d.%d.%d.%d\n",
616 (int)(src >> 24), (int)((src >> 16) & 0xff),
617 (int)((src >> 8) & 0xff), (int)(src & 0xff),
618 (int)(dst >> 24), (int)((dst >> 16) & 0xff),
619 (int)((dst >> 8) & 0xff), (int)(dst & 0xff),
620 (int)(gw >> 24), (int)((gw >> 16) & 0xff),
621 (int)((gw >> 8) & 0xff), (int)(gw & 0xff));
624 * RFC1812 says we must ignore ICMP redirects if we
625 * are acting as router.
627 if (V_drop_redirect || V_ipforwarding)
631 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
632 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) {
633 ICMPSTAT_INC(icps_badlen);
637 * Short circuit routing redirects to force
638 * immediate change in the kernel's routing
639 * tables. The message is also handed to anyone
640 * listening on a raw socket (e.g. the routing
641 * daemon for use in updating its tables).
643 icmpgw.sin_addr = ip->ip_src;
644 icmpdst.sin_addr = icp->icmp_gwaddr;
647 char buf[4 * sizeof "123"];
648 strcpy(buf, inet_ntoa(icp->icmp_ip.ip_dst));
650 printf("redirect dst %s to %s\n",
651 buf, inet_ntoa(icp->icmp_gwaddr));
654 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
655 for ( fibnum = 0; fibnum < rt_numfibs; fibnum++) {
656 in_rtredirect((struct sockaddr *)&icmpsrc,
657 (struct sockaddr *)&icmpdst,
658 (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST,
659 (struct sockaddr *)&icmpgw, fibnum);
661 pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&icmpsrc);
663 key_sa_routechange((struct sockaddr *)&icmpsrc);
668 * No kernel processing for the following;
669 * just fall through to send to raw listener.
672 case ICMP_ROUTERADVERT:
673 case ICMP_ROUTERSOLICIT:
674 case ICMP_TSTAMPREPLY:
690 * Reflect the ip packet back to the source
693 icmp_reflect(struct mbuf *m)
695 struct ip *ip = mtod(m, struct ip *);
698 struct in_ifaddr *ia;
700 struct mbuf *opts = 0;
701 int optlen = (ip->ip_hl << 2) - sizeof(struct ip);
703 if (IN_MULTICAST(ntohl(ip->ip_src.s_addr)) ||
704 IN_EXPERIMENTAL(ntohl(ip->ip_src.s_addr)) ||
705 IN_ZERONET(ntohl(ip->ip_src.s_addr)) ) {
706 m_freem(m); /* Bad return address */
707 ICMPSTAT_INC(icps_badaddr);
708 goto done; /* Ip_output() will check for broadcast */
712 ip->ip_dst = ip->ip_src;
715 * Source selection for ICMP replies:
717 * If the incoming packet was addressed directly to one of our
718 * own addresses, use dst as the src for the reply.
721 LIST_FOREACH(ia, INADDR_HASH(t.s_addr), ia_hash) {
722 if (t.s_addr == IA_SIN(ia)->sin_addr.s_addr) {
723 t = IA_SIN(ia)->sin_addr;
731 * If the incoming packet was addressed to one of our broadcast
732 * addresses, use the first non-broadcast address which corresponds
733 * to the incoming interface.
735 ifp = m->m_pkthdr.rcvif;
736 if (ifp != NULL && ifp->if_flags & IFF_BROADCAST) {
738 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
739 if (ifa->ifa_addr->sa_family != AF_INET)
742 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
744 t = IA_SIN(ia)->sin_addr;
745 IF_ADDR_RUNLOCK(ifp);
749 IF_ADDR_RUNLOCK(ifp);
752 * If the packet was transiting through us, use the address of
753 * the interface the packet came through in. If that interface
754 * doesn't have a suitable IP address, the normal selection
757 if (V_icmp_rfi && ifp != NULL) {
759 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
760 if (ifa->ifa_addr->sa_family != AF_INET)
763 t = IA_SIN(ia)->sin_addr;
764 IF_ADDR_RUNLOCK(ifp);
767 IF_ADDR_RUNLOCK(ifp);
770 * If the incoming packet was not addressed directly to us, use
771 * designated interface for icmp replies specified by sysctl
772 * net.inet.icmp.reply_src (default not set). Otherwise continue
773 * with normal source selection.
775 if (V_reply_src[0] != '\0' && (ifp = ifunit(V_reply_src))) {
777 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
778 if (ifa->ifa_addr->sa_family != AF_INET)
781 t = IA_SIN(ia)->sin_addr;
782 IF_ADDR_RUNLOCK(ifp);
785 IF_ADDR_RUNLOCK(ifp);
788 * If the packet was transiting through us, use the address of
789 * the interface that is the closest to the packet source.
790 * When we don't have a route back to the packet source, stop here
791 * and drop the packet.
793 ia = ip_rtaddr(ip->ip_dst, M_GETFIB(m));
796 ICMPSTAT_INC(icps_noroute);
799 t = IA_SIN(ia)->sin_addr;
800 ifa_free(&ia->ia_ifa);
803 mac_netinet_icmp_replyinplace(m);
806 ip->ip_ttl = V_ip_defttl;
814 * Retrieve any source routing from the incoming packet;
815 * add on any record-route or timestamp options.
817 cp = (u_char *) (ip + 1);
818 if ((opts = ip_srcroute(m)) == 0 &&
819 (opts = m_gethdr(M_NOWAIT, MT_DATA))) {
820 opts->m_len = sizeof(struct in_addr);
821 mtod(opts, struct in_addr *)->s_addr = 0;
826 printf("icmp_reflect optlen %d rt %d => ",
827 optlen, opts->m_len);
829 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) {
830 opt = cp[IPOPT_OPTVAL];
831 if (opt == IPOPT_EOL)
833 if (opt == IPOPT_NOP)
836 if (cnt < IPOPT_OLEN + sizeof(*cp))
838 len = cp[IPOPT_OLEN];
839 if (len < IPOPT_OLEN + sizeof(*cp) ||
844 * Should check for overflow, but it "can't happen"
846 if (opt == IPOPT_RR || opt == IPOPT_TS ||
847 opt == IPOPT_SECURITY) {
849 mtod(opts, caddr_t) + opts->m_len, len);
853 /* Terminate & pad, if necessary */
854 cnt = opts->m_len % 4;
856 for (; cnt < 4; cnt++) {
857 *(mtod(opts, caddr_t) + opts->m_len) =
864 printf("%d\n", opts->m_len);
869 m_tag_delete_nonpersistent(m);
870 m->m_flags &= ~(M_BCAST|M_MCAST);
878 * Send an icmp packet back to the ip level,
879 * after supplying a checksum.
882 icmp_send(struct mbuf *m, struct mbuf *opts)
884 register struct ip *ip = mtod(m, struct ip *);
886 register struct icmp *icp;
888 hlen = ip->ip_hl << 2;
891 icp = mtod(m, struct icmp *);
893 icp->icmp_cksum = in_cksum(m, ntohs(ip->ip_len) - hlen);
896 m->m_pkthdr.rcvif = (struct ifnet *)0;
899 char buf[4 * sizeof "123"];
900 strcpy(buf, inet_ntoa(ip->ip_dst));
901 printf("icmp_send dst %s src %s\n",
902 buf, inet_ntoa(ip->ip_src));
905 (void) ip_output(m, opts, NULL, 0, NULL, NULL);
909 * Return milliseconds since 00:00 GMT in network format.
918 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000;
923 * Return the next larger or smaller MTU plateau (table from RFC 1191)
924 * given current value MTU. If DIR is less than zero, a larger plateau
925 * is returned; otherwise, a smaller value is returned.
928 ip_next_mtu(int mtu, int dir)
930 static int mtutab[] = {
931 65535, 32000, 17914, 8166, 4352, 2002, 1492, 1280, 1006, 508,
936 size = (sizeof mtutab) / (sizeof mtutab[0]);
938 for (i = 0; i < size; i++)
942 for (i = size - 1; i >= 0; i--)
945 if (mtu == mtutab[0])
954 * badport_bandlim() - check for ICMP bandwidth limit
956 * Return 0 if it is ok to send an ICMP error response, -1 if we have
957 * hit our bandwidth limit and it is not ok.
959 * If icmplim is <= 0, the feature is disabled and 0 is returned.
961 * For now we separate the TCP and UDP subsystems w/ different 'which'
962 * values. We may eventually remove this separation (and simplify the
965 * Note that the printing of the error message is delayed so we can
966 * properly print the icmp error rate that the system was trying to do
967 * (i.e. 22000/100 pps, etc...). This can cause long delays in printing
968 * the 'final' error, but it doesn't make sense to solve the printing
969 * delay with more complex code.
973 badport_bandlim(int which)
976 #define N(a) (sizeof (a) / sizeof (a[0]))
979 struct timeval lasttime;
981 } rates[BANDLIM_MAX+1] = {
982 { "icmp unreach response" },
983 { "icmp ping response" },
984 { "icmp tstamp response" },
985 { "closed port RST response" },
986 { "open port RST response" },
987 { "icmp6 unreach response" },
988 { "sctp ootb response" }
992 * Return ok status if feature disabled or argument out of range.
994 if (V_icmplim > 0 && (u_int) which < N(rates)) {
995 struct rate *r = &rates[which];
996 int opps = r->curpps;
998 if (!ppsratecheck(&r->lasttime, &r->curpps, V_icmplim))
999 return -1; /* discard packet */
1001 * If we've dropped below the threshold after having
1002 * rate-limited traffic print the message. This preserves
1003 * the previous behaviour at the expense of added complexity.
1005 if (V_icmplim_output && opps > V_icmplim)
1006 log(LOG_NOTICE, "Limiting %s from %d to %d packets/sec\n",
1007 r->type, opps, V_icmplim);
1009 return 0; /* okay to send packet */