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_types.h>
50 #include <net/route.h>
53 #include <netinet/in.h>
54 #include <netinet/in_pcb.h>
55 #include <netinet/in_systm.h>
56 #include <netinet/in_var.h>
57 #include <netinet/ip.h>
58 #include <netinet/ip_icmp.h>
59 #include <netinet/ip_var.h>
60 #include <netinet/ip_options.h>
61 #include <netinet/tcp.h>
62 #include <netinet/tcp_var.h>
63 #include <netinet/tcpip.h>
64 #include <netinet/icmp_var.h>
68 #include <netipsec/ipsec.h>
69 #include <netipsec/key.h>
72 #include <machine/in_cksum.h>
74 #include <security/mac/mac_framework.h>
78 * ICMP routines: error generation, receive packet processing, and
79 * routines to turnaround packets back to the originator, and
80 * host table maintenance routines.
82 static VNET_DEFINE(int, icmplim) = 200;
83 #define V_icmplim VNET(icmplim)
84 SYSCTL_VNET_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RW,
85 &VNET_NAME(icmplim), 0,
86 "Maximum number of ICMP responses per second");
88 static VNET_DEFINE(int, icmplim_output) = 1;
89 #define V_icmplim_output VNET(icmplim_output)
90 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, icmplim_output, CTLFLAG_RW,
91 &VNET_NAME(icmplim_output), 0,
92 "Enable logging of ICMP response rate limiting");
95 VNET_DEFINE(struct icmpstat, icmpstat);
96 SYSCTL_VNET_STRUCT(_net_inet_icmp, ICMPCTL_STATS, stats, CTLFLAG_RW,
97 &VNET_NAME(icmpstat), icmpstat, "");
99 static VNET_DEFINE(int, icmpmaskrepl) = 0;
100 #define V_icmpmaskrepl VNET(icmpmaskrepl)
101 SYSCTL_VNET_INT(_net_inet_icmp, ICMPCTL_MASKREPL, maskrepl, CTLFLAG_RW,
102 &VNET_NAME(icmpmaskrepl), 0,
103 "Reply to ICMP Address Mask Request packets.");
105 static VNET_DEFINE(u_int, icmpmaskfake) = 0;
106 #define V_icmpmaskfake VNET(icmpmaskfake)
107 SYSCTL_VNET_UINT(_net_inet_icmp, OID_AUTO, maskfake, CTLFLAG_RW,
108 &VNET_NAME(icmpmaskfake), 0,
109 "Fake reply to ICMP Address Mask Request packets.");
111 VNET_DEFINE(int, drop_redirect) = 0;
113 static VNET_DEFINE(int, log_redirect) = 0;
114 #define V_log_redirect VNET(log_redirect)
115 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, log_redirect, CTLFLAG_RW,
116 &VNET_NAME(log_redirect), 0,
117 "Log ICMP redirects to the console");
119 static VNET_DEFINE(char, reply_src[IFNAMSIZ]);
120 #define V_reply_src VNET(reply_src)
121 SYSCTL_VNET_STRING(_net_inet_icmp, OID_AUTO, reply_src, CTLFLAG_RW,
122 &VNET_NAME(reply_src), IFNAMSIZ,
123 "icmp reply source for non-local packets.");
125 static VNET_DEFINE(int, icmp_rfi) = 0;
126 #define V_icmp_rfi VNET(icmp_rfi)
127 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, reply_from_interface, CTLFLAG_RW,
128 &VNET_NAME(icmp_rfi), 0,
129 "ICMP reply from incoming interface for non-local packets");
131 static VNET_DEFINE(int, icmp_quotelen) = 8;
132 #define V_icmp_quotelen VNET(icmp_quotelen)
133 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, quotelen, CTLFLAG_RW,
134 &VNET_NAME(icmp_quotelen), 0,
135 "Number of bytes from original packet to quote in ICMP reply");
138 * ICMP broadcast echo sysctl
140 static VNET_DEFINE(int, icmpbmcastecho) = 0;
141 #define V_icmpbmcastecho VNET(icmpbmcastecho)
142 SYSCTL_VNET_INT(_net_inet_icmp, OID_AUTO, bmcastecho, CTLFLAG_RW,
143 &VNET_NAME(icmpbmcastecho), 0,
151 static void icmp_reflect(struct mbuf *);
152 static void icmp_send(struct mbuf *, struct mbuf *);
154 extern struct protosw inetsw[];
157 sysctl_net_icmp_drop_redir(SYSCTL_HANDLER_ARGS)
161 struct radix_node_head *rnh;
163 new = V_drop_redirect;
164 error = sysctl_handle_int(oidp, &new, 0, req);
165 if (error == 0 && req->newptr) {
166 new = (new != 0) ? 1 : 0;
168 if (new == V_drop_redirect)
171 for (i = 0; i < rt_numfibs; i++) {
172 if ((rnh = rt_tables_get_rnh(i, AF_INET)) == NULL)
174 RADIX_NODE_HEAD_LOCK(rnh);
175 in_setmatchfunc(rnh, new);
176 RADIX_NODE_HEAD_UNLOCK(rnh);
179 V_drop_redirect = new;
185 SYSCTL_VNET_PROC(_net_inet_icmp, OID_AUTO, drop_redirect,
186 CTLTYPE_INT|CTLFLAG_RW, 0, 0,
187 sysctl_net_icmp_drop_redir, "I", "Ignore ICMP redirects");
190 * Kernel module interface for updating icmpstat. The argument is an index
191 * into icmpstat treated as an array of u_long. While this encodes the
192 * general layout of icmpstat into the caller, it doesn't encode its
193 * location, so that future changes to add, for example, per-CPU stats
194 * support won't cause binary compatibility problems for kernel modules.
197 kmod_icmpstat_inc(int statnum)
200 (*((u_long *)&V_icmpstat + statnum))++;
204 * Generate an error packet of type error
205 * in response to bad packet ip.
208 icmp_error(struct mbuf *n, int type, int code, uint32_t dest, int mtu)
210 register struct ip *oip = mtod(n, struct ip *), *nip;
211 register unsigned oiphlen = oip->ip_hl << 2;
212 register struct icmp *icp;
213 register struct mbuf *m;
214 unsigned icmplen, icmpelen, nlen;
216 KASSERT((u_int)type <= ICMP_MAXTYPE, ("%s: illegal ICMP type", __func__));
219 printf("icmp_error(%p, %x, %d)\n", oip, type, code);
221 if (type != ICMP_REDIRECT)
222 ICMPSTAT_INC(icps_error);
225 * if the original packet was encrypted.
226 * if not the first fragment of message.
227 * in response to a multicast or broadcast packet.
228 * if the old packet protocol was an ICMP error message.
230 if (n->m_flags & M_DECRYPTED)
232 if (oip->ip_off & htons(~(IP_MF|IP_DF)))
234 if (n->m_flags & (M_BCAST|M_MCAST))
236 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT &&
237 n->m_len >= oiphlen + ICMP_MINLEN &&
238 !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiphlen))->icmp_type)) {
239 ICMPSTAT_INC(icps_oldicmp);
242 /* Drop if IP header plus 8 bytes is not contignous in first mbuf. */
243 if (oiphlen + 8 > n->m_len)
246 * Calculate length to quote from original packet and
247 * prevent the ICMP mbuf from overflowing.
248 * Unfortunatly this is non-trivial since ip_forward()
249 * sends us truncated packets.
251 nlen = m_length(n, NULL);
252 if (oip->ip_p == IPPROTO_TCP) {
256 if (oiphlen + sizeof(struct tcphdr) > n->m_len &&
259 if (n->m_len < oiphlen + sizeof(struct tcphdr) &&
260 ((n = m_pullup(n, oiphlen + sizeof(struct tcphdr))) == NULL))
262 th = (struct tcphdr *)((caddr_t)oip + oiphlen);
263 tcphlen = th->th_off << 2;
264 if (tcphlen < sizeof(struct tcphdr))
266 if (ntohs(oip->ip_len) < oiphlen + tcphlen)
268 if (oiphlen + tcphlen > n->m_len && n->m_next == NULL)
270 if (n->m_len < oiphlen + tcphlen &&
271 ((n = m_pullup(n, oiphlen + tcphlen)) == NULL))
273 icmpelen = max(tcphlen, min(V_icmp_quotelen,
274 ntohs(oip->ip_len) - oiphlen));
276 stdreply: icmpelen = max(8, min(V_icmp_quotelen, ntohs(oip->ip_len) - oiphlen));
278 icmplen = min(oiphlen + icmpelen, nlen);
279 if (icmplen < sizeof(struct ip))
282 if (MHLEN > sizeof(struct ip) + ICMP_MINLEN + icmplen)
283 m = m_gethdr(M_DONTWAIT, MT_DATA);
285 m = m_getcl(M_DONTWAIT, MT_DATA, M_PKTHDR);
289 mac_netinet_icmp_reply(n, m);
291 icmplen = min(icmplen, M_TRAILINGSPACE(m) - sizeof(struct ip) - ICMP_MINLEN);
292 m_align(m, ICMP_MINLEN + icmplen);
293 m->m_len = ICMP_MINLEN + icmplen;
295 /* XXX MRT make the outgoing packet use the same FIB
296 * that was associated with the incoming packet
298 M_SETFIB(m, M_GETFIB(n));
299 icp = mtod(m, struct icmp *);
300 ICMPSTAT_INC(icps_outhist[type]);
301 icp->icmp_type = type;
302 if (type == ICMP_REDIRECT)
303 icp->icmp_gwaddr.s_addr = dest;
307 * The following assignments assume an overlay with the
308 * just zeroed icmp_void field.
310 if (type == ICMP_PARAMPROB) {
311 icp->icmp_pptr = code;
313 } else if (type == ICMP_UNREACH &&
314 code == ICMP_UNREACH_NEEDFRAG && mtu) {
315 icp->icmp_nextmtu = htons(mtu);
318 icp->icmp_code = code;
321 * Copy the quotation into ICMP message and
322 * convert quoted IP header back to network representation.
324 m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip);
328 * Set up ICMP message mbuf and copy old IP header (without options
329 * in front of ICMP message.
330 * If the original mbuf was meant to bypass the firewall, the error
331 * reply should bypass as well.
333 m->m_flags |= n->m_flags & M_SKIP_FIREWALL;
334 m->m_data -= sizeof(struct ip);
335 m->m_len += sizeof(struct ip);
336 m->m_pkthdr.len = m->m_len;
337 m->m_pkthdr.rcvif = n->m_pkthdr.rcvif;
338 nip = mtod(m, struct ip *);
339 bcopy((caddr_t)oip, (caddr_t)nip, sizeof(struct ip));
340 nip->ip_len = htons(m->m_len);
341 nip->ip_v = IPVERSION;
343 nip->ip_p = IPPROTO_ICMP;
352 * Process a received ICMP message.
355 icmp_input(struct mbuf *m, int off)
358 struct in_ifaddr *ia;
359 struct ip *ip = mtod(m, struct ip *);
360 struct sockaddr_in icmpsrc, icmpdst, icmpgw;
362 int icmplen = ntohs(ip->ip_len);
364 void (*ctlfunc)(int, struct sockaddr *, void *);
368 * Locate icmp structure in mbuf, and check
369 * that not corrupted and of at least minimum length.
373 char buf[4 * sizeof "123"];
374 strcpy(buf, inet_ntoa(ip->ip_src));
375 printf("icmp_input from %s to %s, len %d\n",
376 buf, inet_ntoa(ip->ip_dst), icmplen);
379 if (icmplen < ICMP_MINLEN) {
380 ICMPSTAT_INC(icps_tooshort);
383 i = hlen + min(icmplen, ICMP_ADVLENMIN);
384 if (m->m_len < i && (m = m_pullup(m, i)) == NULL) {
385 ICMPSTAT_INC(icps_tooshort);
388 ip = mtod(m, struct ip *);
391 icp = mtod(m, struct icmp *);
392 if (in_cksum(m, icmplen)) {
393 ICMPSTAT_INC(icps_checksum);
399 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) {
401 * Deliver very specific ICMP type only.
403 switch (icp->icmp_type) {
414 printf("icmp_input, type %d code %d\n", icp->icmp_type,
419 * Message type specific processing.
421 if (icp->icmp_type > ICMP_MAXTYPE)
425 bzero(&icmpsrc, sizeof(icmpsrc));
426 icmpsrc.sin_len = sizeof(struct sockaddr_in);
427 icmpsrc.sin_family = AF_INET;
428 bzero(&icmpdst, sizeof(icmpdst));
429 icmpdst.sin_len = sizeof(struct sockaddr_in);
430 icmpdst.sin_family = AF_INET;
431 bzero(&icmpgw, sizeof(icmpgw));
432 icmpgw.sin_len = sizeof(struct sockaddr_in);
433 icmpgw.sin_family = AF_INET;
435 ICMPSTAT_INC(icps_inhist[icp->icmp_type]);
436 code = icp->icmp_code;
437 switch (icp->icmp_type) {
441 case ICMP_UNREACH_NET:
442 case ICMP_UNREACH_HOST:
443 case ICMP_UNREACH_SRCFAIL:
444 case ICMP_UNREACH_NET_UNKNOWN:
445 case ICMP_UNREACH_HOST_UNKNOWN:
446 case ICMP_UNREACH_ISOLATED:
447 case ICMP_UNREACH_TOSNET:
448 case ICMP_UNREACH_TOSHOST:
449 case ICMP_UNREACH_HOST_PRECEDENCE:
450 case ICMP_UNREACH_PRECEDENCE_CUTOFF:
451 code = PRC_UNREACH_NET;
454 case ICMP_UNREACH_NEEDFRAG:
459 * RFC 1122, Sections 3.2.2.1 and 4.2.3.9.
460 * Treat subcodes 2,3 as immediate RST
462 case ICMP_UNREACH_PROTOCOL:
463 case ICMP_UNREACH_PORT:
464 code = PRC_UNREACH_PORT;
467 case ICMP_UNREACH_NET_PROHIB:
468 case ICMP_UNREACH_HOST_PROHIB:
469 case ICMP_UNREACH_FILTER_PROHIB:
470 code = PRC_UNREACH_ADMIN_PROHIB;
481 code += PRC_TIMXCEED_INTRANS;
487 code = PRC_PARAMPROB;
490 case ICMP_SOURCEQUENCH:
496 * Problem with datagram; advise higher level routines.
498 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
499 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) {
500 ICMPSTAT_INC(icps_badlen);
503 /* Discard ICMP's in response to multicast packets */
504 if (IN_MULTICAST(ntohl(icp->icmp_ip.ip_dst.s_addr)))
508 printf("deliver to protocol %d\n", icp->icmp_ip.ip_p);
510 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
512 * XXX if the packet contains [IPv4 AH TCP], we can't make a
513 * notification to TCP layer.
515 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput;
517 (*ctlfunc)(code, (struct sockaddr *)&icmpsrc,
518 (void *)&icp->icmp_ip);
522 ICMPSTAT_INC(icps_badcode);
526 if (!V_icmpbmcastecho
527 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
528 ICMPSTAT_INC(icps_bmcastecho);
531 icp->icmp_type = ICMP_ECHOREPLY;
532 if (badport_bandlim(BANDLIM_ICMP_ECHO) < 0)
538 if (!V_icmpbmcastecho
539 && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
540 ICMPSTAT_INC(icps_bmcasttstamp);
543 if (icmplen < ICMP_TSLEN) {
544 ICMPSTAT_INC(icps_badlen);
547 icp->icmp_type = ICMP_TSTAMPREPLY;
548 icp->icmp_rtime = iptime();
549 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */
550 if (badport_bandlim(BANDLIM_ICMP_TSTAMP) < 0)
556 if (V_icmpmaskrepl == 0)
559 * We are not able to respond with all ones broadcast
560 * unless we receive it over a point-to-point interface.
562 if (icmplen < ICMP_MASKLEN)
564 switch (ip->ip_dst.s_addr) {
566 case INADDR_BROADCAST:
568 icmpdst.sin_addr = ip->ip_src;
572 icmpdst.sin_addr = ip->ip_dst;
574 ia = (struct in_ifaddr *)ifaof_ifpforaddr(
575 (struct sockaddr *)&icmpdst, m->m_pkthdr.rcvif);
578 if (ia->ia_ifp == NULL) {
579 ifa_free(&ia->ia_ifa);
582 icp->icmp_type = ICMP_MASKREPLY;
583 if (V_icmpmaskfake == 0)
584 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr;
586 icp->icmp_mask = V_icmpmaskfake;
587 if (ip->ip_src.s_addr == 0) {
588 if (ia->ia_ifp->if_flags & IFF_BROADCAST)
589 ip->ip_src = satosin(&ia->ia_broadaddr)->sin_addr;
590 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT)
591 ip->ip_src = satosin(&ia->ia_dstaddr)->sin_addr;
593 ifa_free(&ia->ia_ifa);
595 /* Since ip_input() deducts this. */
596 ip->ip_len = htons(ntohs(ip->ip_len) + hlen);
597 ICMPSTAT_INC(icps_reflect);
598 ICMPSTAT_INC(icps_outhist[icp->icmp_type]);
603 if (V_log_redirect) {
606 src = ntohl(ip->ip_src.s_addr);
607 dst = ntohl(icp->icmp_ip.ip_dst.s_addr);
608 gw = ntohl(icp->icmp_gwaddr.s_addr);
609 printf("icmp redirect from %d.%d.%d.%d: "
610 "%d.%d.%d.%d => %d.%d.%d.%d\n",
611 (int)(src >> 24), (int)((src >> 16) & 0xff),
612 (int)((src >> 8) & 0xff), (int)(src & 0xff),
613 (int)(dst >> 24), (int)((dst >> 16) & 0xff),
614 (int)((dst >> 8) & 0xff), (int)(dst & 0xff),
615 (int)(gw >> 24), (int)((gw >> 16) & 0xff),
616 (int)((gw >> 8) & 0xff), (int)(gw & 0xff));
619 * RFC1812 says we must ignore ICMP redirects if we
620 * are acting as router.
622 if (V_drop_redirect || V_ipforwarding)
626 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
627 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) {
628 ICMPSTAT_INC(icps_badlen);
632 * Short circuit routing redirects to force
633 * immediate change in the kernel's routing
634 * tables. The message is also handed to anyone
635 * listening on a raw socket (e.g. the routing
636 * daemon for use in updating its tables).
638 icmpgw.sin_addr = ip->ip_src;
639 icmpdst.sin_addr = icp->icmp_gwaddr;
642 char buf[4 * sizeof "123"];
643 strcpy(buf, inet_ntoa(icp->icmp_ip.ip_dst));
645 printf("redirect dst %s to %s\n",
646 buf, inet_ntoa(icp->icmp_gwaddr));
649 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
650 for ( fibnum = 0; fibnum < rt_numfibs; fibnum++) {
651 in_rtredirect((struct sockaddr *)&icmpsrc,
652 (struct sockaddr *)&icmpdst,
653 (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST,
654 (struct sockaddr *)&icmpgw, fibnum);
656 pfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&icmpsrc);
658 key_sa_routechange((struct sockaddr *)&icmpsrc);
663 * No kernel processing for the following;
664 * just fall through to send to raw listener.
667 case ICMP_ROUTERADVERT:
668 case ICMP_ROUTERSOLICIT:
669 case ICMP_TSTAMPREPLY:
685 * Reflect the ip packet back to the source
688 icmp_reflect(struct mbuf *m)
690 struct ip *ip = mtod(m, struct ip *);
693 struct in_ifaddr *ia;
695 struct mbuf *opts = 0;
696 int optlen = (ip->ip_hl << 2) - sizeof(struct ip);
698 if (IN_MULTICAST(ntohl(ip->ip_src.s_addr)) ||
699 IN_EXPERIMENTAL(ntohl(ip->ip_src.s_addr)) ||
700 IN_ZERONET(ntohl(ip->ip_src.s_addr)) ) {
701 m_freem(m); /* Bad return address */
702 ICMPSTAT_INC(icps_badaddr);
703 goto done; /* Ip_output() will check for broadcast */
707 ip->ip_dst = ip->ip_src;
710 * Source selection for ICMP replies:
712 * If the incoming packet was addressed directly to one of our
713 * own addresses, use dst as the src for the reply.
716 LIST_FOREACH(ia, INADDR_HASH(t.s_addr), ia_hash) {
717 if (t.s_addr == IA_SIN(ia)->sin_addr.s_addr) {
718 t = IA_SIN(ia)->sin_addr;
726 * If the incoming packet was addressed to one of our broadcast
727 * addresses, use the first non-broadcast address which corresponds
728 * to the incoming interface.
730 ifp = m->m_pkthdr.rcvif;
731 if (ifp != NULL && ifp->if_flags & IFF_BROADCAST) {
733 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
734 if (ifa->ifa_addr->sa_family != AF_INET)
737 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
739 t = IA_SIN(ia)->sin_addr;
740 IF_ADDR_RUNLOCK(ifp);
744 IF_ADDR_RUNLOCK(ifp);
747 * If the packet was transiting through us, use the address of
748 * the interface the packet came through in. If that interface
749 * doesn't have a suitable IP address, the normal selection
752 if (V_icmp_rfi && ifp != NULL) {
754 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
755 if (ifa->ifa_addr->sa_family != AF_INET)
758 t = IA_SIN(ia)->sin_addr;
759 IF_ADDR_RUNLOCK(ifp);
762 IF_ADDR_RUNLOCK(ifp);
765 * If the incoming packet was not addressed directly to us, use
766 * designated interface for icmp replies specified by sysctl
767 * net.inet.icmp.reply_src (default not set). Otherwise continue
768 * with normal source selection.
770 if (V_reply_src[0] != '\0' && (ifp = ifunit(V_reply_src))) {
772 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
773 if (ifa->ifa_addr->sa_family != AF_INET)
776 t = IA_SIN(ia)->sin_addr;
777 IF_ADDR_RUNLOCK(ifp);
780 IF_ADDR_RUNLOCK(ifp);
783 * If the packet was transiting through us, use the address of
784 * the interface that is the closest to the packet source.
785 * When we don't have a route back to the packet source, stop here
786 * and drop the packet.
788 ia = ip_rtaddr(ip->ip_dst, M_GETFIB(m));
791 ICMPSTAT_INC(icps_noroute);
794 t = IA_SIN(ia)->sin_addr;
795 ifa_free(&ia->ia_ifa);
798 mac_netinet_icmp_replyinplace(m);
801 ip->ip_ttl = V_ip_defttl;
809 * Retrieve any source routing from the incoming packet;
810 * add on any record-route or timestamp options.
812 cp = (u_char *) (ip + 1);
813 if ((opts = ip_srcroute(m)) == 0 &&
814 (opts = m_gethdr(M_DONTWAIT, MT_DATA))) {
815 opts->m_len = sizeof(struct in_addr);
816 mtod(opts, struct in_addr *)->s_addr = 0;
821 printf("icmp_reflect optlen %d rt %d => ",
822 optlen, opts->m_len);
824 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) {
825 opt = cp[IPOPT_OPTVAL];
826 if (opt == IPOPT_EOL)
828 if (opt == IPOPT_NOP)
831 if (cnt < IPOPT_OLEN + sizeof(*cp))
833 len = cp[IPOPT_OLEN];
834 if (len < IPOPT_OLEN + sizeof(*cp) ||
839 * Should check for overflow, but it "can't happen"
841 if (opt == IPOPT_RR || opt == IPOPT_TS ||
842 opt == IPOPT_SECURITY) {
844 mtod(opts, caddr_t) + opts->m_len, len);
848 /* Terminate & pad, if necessary */
849 cnt = opts->m_len % 4;
851 for (; cnt < 4; cnt++) {
852 *(mtod(opts, caddr_t) + opts->m_len) =
859 printf("%d\n", opts->m_len);
863 * Now strip out original options by copying rest of first
864 * mbuf's data back, and adjust the IP length.
866 ip->ip_len = htons(ntohs(ip->ip_len) - optlen);
867 ip->ip_v = IPVERSION;
870 if (m->m_flags & M_PKTHDR)
871 m->m_pkthdr.len -= optlen;
872 optlen += sizeof(struct ip);
873 bcopy((caddr_t)ip + optlen, (caddr_t)(ip + 1),
874 (unsigned)(m->m_len - sizeof(struct ip)));
876 m_tag_delete_nonpersistent(m);
877 m->m_flags &= ~(M_BCAST|M_MCAST);
885 * Send an icmp packet back to the ip level,
886 * after supplying a checksum.
889 icmp_send(struct mbuf *m, struct mbuf *opts)
891 register struct ip *ip = mtod(m, struct ip *);
893 register struct icmp *icp;
895 hlen = ip->ip_hl << 2;
898 icp = mtod(m, struct icmp *);
900 icp->icmp_cksum = in_cksum(m, ntohs(ip->ip_len) - hlen);
903 m->m_pkthdr.rcvif = (struct ifnet *)0;
906 char buf[4 * sizeof "123"];
907 strcpy(buf, inet_ntoa(ip->ip_dst));
908 printf("icmp_send dst %s src %s\n",
909 buf, inet_ntoa(ip->ip_src));
912 (void) ip_output(m, opts, NULL, 0, NULL, NULL);
916 * Return milliseconds since 00:00 GMT in network format.
925 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000;
930 * Return the next larger or smaller MTU plateau (table from RFC 1191)
931 * given current value MTU. If DIR is less than zero, a larger plateau
932 * is returned; otherwise, a smaller value is returned.
935 ip_next_mtu(int mtu, int dir)
937 static int mtutab[] = {
938 65535, 32000, 17914, 8166, 4352, 2002, 1492, 1280, 1006, 508,
943 size = (sizeof mtutab) / (sizeof mtutab[0]);
945 for (i = 0; i < size; i++)
949 for (i = size - 1; i >= 0; i--)
952 if (mtu == mtutab[0])
961 * badport_bandlim() - check for ICMP bandwidth limit
963 * Return 0 if it is ok to send an ICMP error response, -1 if we have
964 * hit our bandwidth limit and it is not ok.
966 * If icmplim is <= 0, the feature is disabled and 0 is returned.
968 * For now we separate the TCP and UDP subsystems w/ different 'which'
969 * values. We may eventually remove this separation (and simplify the
972 * Note that the printing of the error message is delayed so we can
973 * properly print the icmp error rate that the system was trying to do
974 * (i.e. 22000/100 pps, etc...). This can cause long delays in printing
975 * the 'final' error, but it doesn't make sense to solve the printing
976 * delay with more complex code.
980 badport_bandlim(int which)
983 #define N(a) (sizeof (a) / sizeof (a[0]))
986 struct timeval lasttime;
988 } rates[BANDLIM_MAX+1] = {
989 { "icmp unreach response" },
990 { "icmp ping response" },
991 { "icmp tstamp response" },
992 { "closed port RST response" },
993 { "open port RST response" },
994 { "icmp6 unreach response" },
995 { "sctp ootb response" }
999 * Return ok status if feature disabled or argument out of range.
1001 if (V_icmplim > 0 && (u_int) which < N(rates)) {
1002 struct rate *r = &rates[which];
1003 int opps = r->curpps;
1005 if (!ppsratecheck(&r->lasttime, &r->curpps, V_icmplim))
1006 return -1; /* discard packet */
1008 * If we've dropped below the threshold after having
1009 * rate-limited traffic print the message. This preserves
1010 * the previous behaviour at the expense of added complexity.
1012 if (V_icmplim_output && opps > V_icmplim)
1013 log(LOG_NOTICE, "Limiting %s from %d to %d packets/sec\n",
1014 r->type, opps, V_icmplim);
1016 return 0; /* okay to send packet */