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  * 3. All advertising materials mentioning features or use of this software
  14  *    must display the following acknowledgement:
  15  *      This product includes software developed by the University of
  16  *      California, Berkeley and its contributors.
  17  * 4. Neither the name of the University nor the names of its contributors
  18  *    may be used to endorse or promote products derived from this software
  19  *    without specific prior written permission.
  20  *
  21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  31  * SUCH DAMAGE.
  32  *
  33  *      @(#)ip_icmp.c   8.2 (Berkeley) 1/4/94
  34  * $FreeBSD: src/sys/netinet/ip_icmp.c,v 1.39.2.19 2003/01/24 05:11:34 sam Exp $
  35  * $DragonFly: src/sys/netinet/ip_icmp.c,v 1.30 2008/06/09 11:24:24 sephe Exp $
  36  */
  37
  38 #include "opt_ipsec.h"
  39
  40 #include <sys/param.h>
  41 #include <sys/systm.h>
  42 #include <sys/mbuf.h>
  43 #include <sys/protosw.h>
  44 #include <sys/socket.h>
  45 #include <sys/time.h>
  46 #include <sys/kernel.h>
  47 #include <sys/sysctl.h>
  48 #include <sys/in_cksum.h>
  49
  50 #include <machine/stdarg.h>
  51
  52 #include <net/if.h>
  53 #include <net/if_types.h>
  54 #include <net/route.h>
  55
  56 #define _IP_VHL
  57 #include <netinet/in.h>
  58 #include <netinet/in_systm.h>
  59 #include <netinet/in_var.h>
  60 #include <netinet/ip.h>
  61 #include <netinet/ip_icmp.h>
  62 #include <netinet/ip_var.h>
  63 #include <netinet/icmp_var.h>
  64
  65 #ifdef IPSEC
  66 #include <netinet6/ipsec.h>
  67 #include <netproto/key/key.h>
  68 #endif
  69
  70 #ifdef FAST_IPSEC
  71 #include <netproto/ipsec/ipsec.h>
  72 #include <netproto/ipsec/key.h>
  73 #define IPSEC
  74 #endif
  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, "");
  89
  90 static int      drop_redirect = 0;
  91 SYSCTL_INT(_net_inet_icmp, OID_AUTO, drop_redirect, CTLFLAG_RW,
  92         &drop_redirect, 0, "");
  93
  94 static int      log_redirect = 0;
  95 SYSCTL_INT(_net_inet_icmp, OID_AUTO, log_redirect, CTLFLAG_RW,
  96         &log_redirect, 0, "");
  97
  98 #ifdef ICMP_BANDLIM
  99
 100 /*
 101  * ICMP error-response bandwidth limiting sysctl.  If not enabled, sysctl
 102  *      variable content is -1 and read-only.
 103  */
 104
 105 static int      icmplim = 200;
 106 SYSCTL_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RW,
 107         &icmplim, 0, "");
 108 #else
 109
 110 static int      icmplim = -1;
 111 SYSCTL_INT(_net_inet_icmp, ICMPCTL_ICMPLIM, icmplim, CTLFLAG_RD,
 112         &icmplim, 0, "");
 113
 114 #endif
 115
 116 static int      icmplim_output = 1;
 117 SYSCTL_INT(_net_inet_icmp, OID_AUTO, icmplim_output, CTLFLAG_RW,
 118         &icmplim_output, 0, "");
 119
 120 /*
 121  * ICMP broadcast echo sysctl
 122  */
 123
 124 static int      icmpbmcastecho = 0;
 125 SYSCTL_INT(_net_inet_icmp, OID_AUTO, bmcastecho, CTLFLAG_RW,
 126         &icmpbmcastecho, 0, "");
 127
 128
 129 #ifdef ICMPPRINTFS
 130 int     icmpprintfs = 0;
 131 #endif
 132
 133 static void     icmp_reflect (struct mbuf *);
 134 static void     icmp_send (struct mbuf *, struct mbuf *, struct route *);
 135
 136 extern  struct protosw inetsw[];
 137
 138 /*
 139  * Generate an error packet of type error
 140  * in response to bad packet ip.
 141  */
 142 void
 143 icmp_error(struct mbuf *n, int type, int code, n_long dest, int destmtu)
 144 {
 145         struct ip *oip = mtod(n, struct ip *), *nip;
 146         unsigned oiplen = IP_VHL_HL(oip->ip_vhl) << 2;
 147         struct icmp *icp;
 148         struct mbuf *m;
 149         unsigned icmplen;
 150
 151 #ifdef ICMPPRINTFS
 152         if (icmpprintfs)
 153                 kprintf("icmp_error(%p, %d, %d)\n", oip, type, code);
 154 #endif
 155         if (type != ICMP_REDIRECT)
 156                 icmpstat.icps_error++;
 157         /*
 158          * Don't send error if the original packet was encrypted.
 159          * Don't send error if not the first fragment of message.
 160          * Don't error if the old packet protocol was ICMP
 161          * error message, only known informational types.
 162          */
 163         if (n->m_flags & M_DECRYPTED)
 164                 goto freeit;
 165         if (oip->ip_off &~ (IP_MF|IP_DF))
 166                 goto freeit;
 167         if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT &&
 168           n->m_len >= oiplen + ICMP_MINLEN &&
 169           !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiplen))->icmp_type)) {
 170                 icmpstat.icps_oldicmp++;
 171                 goto freeit;
 172         }
 173         /* Don't send error in response to a multicast or broadcast packet */
 174         if (n->m_flags & (M_BCAST|M_MCAST))
 175                 goto freeit;
 176         /*
 177          * First, formulate icmp message
 178          */
 179         m = m_gethdr(MB_DONTWAIT, MT_HEADER);
 180         if (m == NULL)
 181                 goto freeit;
 182         icmplen = min(oiplen + 8, oip->ip_len);
 183         if (icmplen < sizeof(struct ip))
 184                 panic("icmp_error: bad length");
 185         m->m_len = icmplen + ICMP_MINLEN;
 186         MH_ALIGN(m, m->m_len);
 187         icp = mtod(m, struct icmp *);
 188         if ((u_int)type > ICMP_MAXTYPE)
 189                 panic("icmp_error");
 190         icmpstat.icps_outhist[type]++;
 191         icp->icmp_type = type;
 192         if (type == ICMP_REDIRECT)
 193                 icp->icmp_gwaddr.s_addr = dest;
 194         else {
 195                 icp->icmp_void = 0;
 196                 /*
 197                  * The following assignments assume an overlay with the
 198                  * zeroed icmp_void field.
 199                  */
 200                 if (type == ICMP_PARAMPROB) {
 201                         icp->icmp_pptr = code;
 202                         code = 0;
 203                 } else if (type == ICMP_UNREACH &&
 204                         code == ICMP_UNREACH_NEEDFRAG && destmtu) {
 205                         icp->icmp_nextmtu = htons(destmtu);
 206                 }
 207         }
 208
 209         icp->icmp_code = code;
 210         m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip);
 211         nip = &icp->icmp_ip;
 212
 213         /*
 214          * Convert fields to network representation.
 215          */
 216         nip->ip_len = htons(nip->ip_len);
 217         nip->ip_off = htons(nip->ip_off);
 218
 219         /*
 220          * Now, copy old ip header (without options)
 221          * in front of icmp message.
 222          */
 223         if (m->m_data - sizeof(struct ip) < m->m_pktdat)
 224                 panic("icmp len");
 225         m->m_data -= sizeof(struct ip);
 226         m->m_len += sizeof(struct ip);
 227         m->m_pkthdr.len = m->m_len;
 228         m->m_pkthdr.rcvif = n->m_pkthdr.rcvif;
 229         nip = mtod(m, struct ip *);
 230         bcopy(oip, nip, sizeof(struct ip));
 231         nip->ip_len = m->m_len;
 232         nip->ip_vhl = IP_VHL_BORING;
 233         nip->ip_p = IPPROTO_ICMP;
 234         nip->ip_tos = 0;
 235         m->m_pkthdr.fw_flags |= n->m_pkthdr.fw_flags & FW_MBUF_GENERATED;
 236         icmp_reflect(m);
 237
 238 freeit:
 239         m_freem(n);
 240 }
 241
 242 /*
 243  * Process a received ICMP message.
 244  */
 245 void
 246 icmp_input(struct mbuf *m, ...)
 247 {
 248         struct sockaddr_in icmpsrc = { sizeof(struct sockaddr_in), AF_INET };
 249         struct sockaddr_in icmpdst = { sizeof(struct sockaddr_in), AF_INET };
 250         struct sockaddr_in icmpgw = { sizeof(struct sockaddr_in), AF_INET };
 251         struct icmp *icp;
 252         struct in_ifaddr *ia;
 253         struct ip *ip = mtod(m, struct ip *);
 254         void (*ctlfunc) (int, struct sockaddr *, void *);
 255         int icmplen = ip->ip_len;
 256         int i, hlen;
 257         int code, off, proto;
 258         __va_list ap;
 259
 260         __va_start(ap, m);
 261         off = __va_arg(ap, int);
 262         proto = __va_arg(ap, int);
 263         __va_end(ap);
 264
 265         hlen = off;
 266
 267         /*
 268          * Locate icmp structure in mbuf, and check
 269          * that not corrupted and of at least minimum length.
 270          */
 271 #ifdef ICMPPRINTFS
 272         if (icmpprintfs) {
 273                 char buf[sizeof "aaa.bbb.ccc.ddd"];
 274
 275                 strcpy(buf, inet_ntoa(ip->ip_src));
 276                 kprintf("icmp_input from %s to %s, len %d\n",
 277                        buf, inet_ntoa(ip->ip_dst), icmplen);
 278         }
 279 #endif
 280         if (icmplen < ICMP_MINLEN) {
 281                 icmpstat.icps_tooshort++;
 282                 goto freeit;
 283         }
 284         i = hlen + min(icmplen, ICMP_ADVLENMIN);
 285         if (m->m_len < i && (m = m_pullup(m, i)) == 0)  {
 286                 icmpstat.icps_tooshort++;
 287                 return;
 288         }
 289         ip = mtod(m, struct ip *);
 290         m->m_len -= hlen;
 291         m->m_data += hlen;
 292         icp = mtod(m, struct icmp *);
 293         if (in_cksum(m, icmplen)) {
 294                 icmpstat.icps_checksum++;
 295                 goto freeit;
 296         }
 297         m->m_len += hlen;
 298         m->m_data -= hlen;
 299
 300         if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) {
 301                 /*
 302                  * Deliver very specific ICMP type only.
 303                  */
 304                 switch (icp->icmp_type) {
 305                 case ICMP_UNREACH:
 306                 case ICMP_TIMXCEED:
 307                         break;
 308                 default:
 309                         goto freeit;
 310                 }
 311         }
 312
 313 #ifdef ICMPPRINTFS
 314         if (icmpprintfs)
 315                 kprintf("icmp_input, type %d code %d\n", icp->icmp_type,
 316                     icp->icmp_code);
 317 #endif
 318
 319         /*
 320          * Message type specific processing.
 321          */
 322         if (icp->icmp_type > ICMP_MAXTYPE)
 323                 goto raw;
 324         icmpstat.icps_inhist[icp->icmp_type]++;
 325         code = icp->icmp_code;
 326         switch (icp->icmp_type) {
 327
 328         case ICMP_UNREACH:
 329                 switch (code) {
 330                         case ICMP_UNREACH_NET:
 331                         case ICMP_UNREACH_HOST:
 332                         case ICMP_UNREACH_SRCFAIL:
 333                         case ICMP_UNREACH_NET_UNKNOWN:
 334                         case ICMP_UNREACH_HOST_UNKNOWN:
 335                         case ICMP_UNREACH_ISOLATED:
 336                         case ICMP_UNREACH_TOSNET:
 337                         case ICMP_UNREACH_TOSHOST:
 338                         case ICMP_UNREACH_HOST_PRECEDENCE:
 339                         case ICMP_UNREACH_PRECEDENCE_CUTOFF:
 340                                 code = PRC_UNREACH_NET;
 341                                 break;
 342
 343                         case ICMP_UNREACH_NEEDFRAG:
 344                                 code = PRC_MSGSIZE;
 345                                 break;
 346
 347                         /*
 348                          * RFC 1122, Sections 3.2.2.1 and 4.2.3.9.
 349                          * Treat subcodes 2,3 as immediate RST
 350                          */
 351                         case ICMP_UNREACH_PROTOCOL:
 352                         case ICMP_UNREACH_PORT:
 353                                 code = PRC_UNREACH_PORT;
 354                                 break;
 355
 356                         case ICMP_UNREACH_NET_PROHIB:
 357                         case ICMP_UNREACH_HOST_PROHIB:
 358                         case ICMP_UNREACH_FILTER_PROHIB:
 359                                 code = PRC_UNREACH_ADMIN_PROHIB;
 360                                 break;
 361
 362                         default:
 363                                 goto badcode;
 364                 }
 365                 goto deliver;
 366
 367         case ICMP_TIMXCEED:
 368                 if (code > 1)
 369                         goto badcode;
 370                 code += PRC_TIMXCEED_INTRANS;
 371                 goto deliver;
 372
 373         case ICMP_PARAMPROB:
 374                 if (code > 1)
 375                         goto badcode;
 376                 code = PRC_PARAMPROB;
 377                 goto deliver;
 378
 379         case ICMP_SOURCEQUENCH:
 380                 if (code)
 381                         goto badcode;
 382                 code = PRC_QUENCH;
 383 deliver:
 384                 /*
 385                  * Problem with datagram; advise higher level routines.
 386                  */
 387                 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
 388                     IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) {
 389                         icmpstat.icps_badlen++;
 390                         goto freeit;
 391                 }
 392                 icp->icmp_ip.ip_len = ntohs(icp->icmp_ip.ip_len);
 393                 /* Discard ICMP's in response to multicast packets */
 394                 if (IN_MULTICAST(ntohl(icp->icmp_ip.ip_dst.s_addr)))
 395                         goto badcode;
 396 #ifdef ICMPPRINTFS
 397                 if (icmpprintfs)
 398                         kprintf("deliver to protocol %d\n", icp->icmp_ip.ip_p);
 399 #endif
 400                 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
 401 #if 1
 402                 /*
 403                  * MTU discovery:
 404                  * If we got a needfrag and there is a host route to the
 405                  * original destination, and the MTU is not locked, then
 406                  * set the MTU in the route to the suggested new value
 407                  * (if given) and then notify as usual.  The ULPs will
 408                  * notice that the MTU has changed and adapt accordingly.
 409                  * If no new MTU was suggested, then we guess a new one
 410                  * less than the current value.  If the new MTU is
 411                  * unreasonably small (arbitrarily set at 296), then
 412                  * we reset the MTU to the interface value and enable the
 413                  * lock bit, indicating that we are no longer doing MTU
 414                  * discovery.
 415                  */
 416                 if (code == PRC_MSGSIZE) {
 417                         struct rtentry *rt;
 418                         int mtu;
 419
 420                         rt = rtpurelookup((struct sockaddr *)&icmpsrc);
 421                         if (rt != NULL && (rt->rt_flags & RTF_HOST) &&
 422                             !(rt->rt_rmx.rmx_locks & RTV_MTU)) {
 423                                 mtu = ntohs(icp->icmp_nextmtu);
 424                                 if (!mtu)
 425                                         mtu = ip_next_mtu(rt->rt_rmx.rmx_mtu,
 426                                                           1);
 427 #ifdef DEBUG_MTUDISC
 428                                 kprintf("MTU for %s reduced to %d\n",
 429                                         inet_ntoa(icmpsrc.sin_addr), mtu);
 430 #endif
 431                                 if (mtu < 296) {
 432                                         /* rt->rt_rmx.rmx_mtu =
 433                                                 rt->rt_ifp->if_mtu; */
 434                                         rt->rt_rmx.rmx_locks |= RTV_MTU;
 435                                 } else if (rt->rt_rmx.rmx_mtu > mtu) {
 436                                         rt->rt_rmx.rmx_mtu = mtu;
 437                                 }
 438                         }
 439                         if (rt != NULL)
 440                                 --rt->rt_refcnt;
 441                 }
 442 #endif
 443                 /*
 444                  * XXX if the packet contains [IPv4 AH TCP], we can't make a
 445                  * notification to TCP layer.
 446                  */
 447                 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput;
 448                 if (ctlfunc)
 449                         (*ctlfunc)(code, (struct sockaddr *)&icmpsrc,
 450                                    &icp->icmp_ip);
 451                 break;
 452
 453 badcode:
 454                 icmpstat.icps_badcode++;
 455                 break;
 456
 457         case ICMP_ECHO:
 458                 if (!icmpbmcastecho
 459                     && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
 460                         icmpstat.icps_bmcastecho++;
 461                         break;
 462                 }
 463                 icp->icmp_type = ICMP_ECHOREPLY;
 464 #ifdef ICMP_BANDLIM
 465                 if (badport_bandlim(BANDLIM_ICMP_ECHO) < 0)
 466                         goto freeit;
 467                 else
 468 #endif
 469                         goto reflect;
 470
 471         case ICMP_TSTAMP:
 472                 if (!icmpbmcastecho
 473                     && (m->m_flags & (M_MCAST | M_BCAST)) != 0) {
 474                         icmpstat.icps_bmcasttstamp++;
 475                         break;
 476                 }
 477                 if (icmplen < ICMP_TSLEN) {
 478                         icmpstat.icps_badlen++;
 479                         break;
 480                 }
 481                 icp->icmp_type = ICMP_TSTAMPREPLY;
 482                 icp->icmp_rtime = iptime();
 483                 icp->icmp_ttime = icp->icmp_rtime;      /* bogus, do later! */
 484 #ifdef ICMP_BANDLIM
 485                 if (badport_bandlim(BANDLIM_ICMP_TSTAMP) < 0)
 486                         goto freeit;
 487                 else
 488 #endif
 489                         goto reflect;
 490
 491         case ICMP_MASKREQ:
 492                 if (icmpmaskrepl == 0)
 493                         break;
 494                 /*
 495                  * We are not able to respond with all ones broadcast
 496                  * unless we receive it over a point-to-point interface.
 497                  */
 498                 if (icmplen < ICMP_MASKLEN)
 499                         break;
 500                 switch (ip->ip_dst.s_addr) {
 501
 502                 case INADDR_BROADCAST:
 503                 case INADDR_ANY:
 504                         icmpdst.sin_addr = ip->ip_src;
 505                         break;
 506
 507                 default:
 508                         icmpdst.sin_addr = ip->ip_dst;
 509                 }
 510                 ia = (struct in_ifaddr *)ifaof_ifpforaddr(
 511                             (struct sockaddr *)&icmpdst, m->m_pkthdr.rcvif);
 512                 if (ia == 0)
 513                         break;
 514                 if (ia->ia_ifp == 0)
 515                         break;
 516                 icp->icmp_type = ICMP_MASKREPLY;
 517                 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr;
 518                 if (ip->ip_src.s_addr == 0) {
 519                         if (ia->ia_ifp->if_flags & IFF_BROADCAST)
 520                             ip->ip_src = satosin(&ia->ia_broadaddr)->sin_addr;
 521                         else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT)
 522                             ip->ip_src = satosin(&ia->ia_dstaddr)->sin_addr;
 523                 }
 524 reflect:
 525                 ip->ip_len += hlen;     /* since ip_input deducts this */
 526                 icmpstat.icps_reflect++;
 527                 icmpstat.icps_outhist[icp->icmp_type]++;
 528                 icmp_reflect(m);
 529                 return;
 530
 531         case ICMP_REDIRECT:
 532                 if (log_redirect) {
 533                         u_long src, dst, gw;
 534
 535                         src = ntohl(ip->ip_src.s_addr);
 536                         dst = ntohl(icp->icmp_ip.ip_dst.s_addr);
 537                         gw = ntohl(icp->icmp_gwaddr.s_addr);
 538                         kprintf("icmp redirect from %d.%d.%d.%d: "
 539                                "%d.%d.%d.%d => %d.%d.%d.%d\n",
 540                                (int)(src >> 24), (int)((src >> 16) & 0xff),
 541                                (int)((src >> 8) & 0xff), (int)(src & 0xff),
 542                                (int)(dst >> 24), (int)((dst >> 16) & 0xff),
 543                                (int)((dst >> 8) & 0xff), (int)(dst & 0xff),
 544                                (int)(gw >> 24), (int)((gw >> 16) & 0xff),
 545                                (int)((gw >> 8) & 0xff), (int)(gw & 0xff));
 546                 }
 547                 if (drop_redirect)
 548                         break;
 549                 if (code > 3)
 550                         goto badcode;
 551                 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) ||
 552                     IP_VHL_HL(icp->icmp_ip.ip_vhl) < (sizeof(struct ip) >> 2)) {
 553                         icmpstat.icps_badlen++;
 554                         break;
 555                 }
 556                 /*
 557                  * Short circuit routing redirects to force
 558                  * immediate change in the kernel's routing
 559                  * tables.  The message is also handed to anyone
 560                  * listening on a raw socket (e.g. the routing
 561                  * daemon for use in updating its tables).
 562                  */
 563                 icmpgw.sin_addr = ip->ip_src;
 564                 icmpdst.sin_addr = icp->icmp_gwaddr;
 565 #ifdef  ICMPPRINTFS
 566                 if (icmpprintfs) {
 567                         char buf[sizeof "aaa.bbb.ccc.ddd"];
 568
 569                         strcpy(buf, inet_ntoa(icp->icmp_ip.ip_dst));
 570                         kprintf("redirect dst %s to %s\n",
 571                                buf, inet_ntoa(icp->icmp_gwaddr));
 572                 }
 573 #endif
 574                 icmpsrc.sin_addr = icp->icmp_ip.ip_dst;
 575                 rtredirect((struct sockaddr *)&icmpsrc,
 576                   (struct sockaddr *)&icmpdst,
 577                   (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST,
 578                   (struct sockaddr *)&icmpgw);
 579                 kpfctlinput(PRC_REDIRECT_HOST, (struct sockaddr *)&icmpsrc);
 580 #ifdef IPSEC
 581                 key_sa_routechange((struct sockaddr *)&icmpsrc);
 582 #endif
 583                 break;
 584
 585         /*
 586          * No kernel processing for the following;
 587          * just fall through to send to raw listener.
 588          */
 589         case ICMP_ECHOREPLY:
 590         case ICMP_ROUTERADVERT:
 591         case ICMP_ROUTERSOLICIT:
 592         case ICMP_TSTAMPREPLY:
 593         case ICMP_IREQREPLY:
 594         case ICMP_MASKREPLY:
 595         default:
 596                 break;
 597         }
 598
 599 raw:
 600         rip_input(m, off, proto);
 601         return;
 602
 603 freeit:
 604         m_freem(m);
 605 }
 606
 607 /*
 608  * Reflect the ip packet back to the source
 609  */
 610 static void
 611 icmp_reflect(struct mbuf *m)
 612 {
 613         struct ip *ip = mtod(m, struct ip *);
 614         struct in_ifaddr *ia;
 615         struct in_ifaddr_container *iac;
 616         struct in_addr t;
 617         struct mbuf *opts = 0;
 618         int optlen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip);
 619         struct route *ro = NULL, rt;
 620
 621         if (!in_canforward(ip->ip_src) &&
 622             ((ntohl(ip->ip_src.s_addr) & IN_CLASSA_NET) !=
 623              (IN_LOOPBACKNET << IN_CLASSA_NSHIFT))) {
 624                 m_freem(m);     /* Bad return address */
 625                 icmpstat.icps_badaddr++;
 626                 goto done;      /* Ip_output() will check for broadcast */
 627         }
 628         t = ip->ip_dst;
 629         ip->ip_dst = ip->ip_src;
 630         ro = &rt;
 631         bzero(ro, sizeof *ro);
 632         /*
 633          * If the incoming packet was addressed directly to us,
 634          * use dst as the src for the reply.  Otherwise (broadcast
 635          * or anonymous), use the address which corresponds
 636          * to the incoming interface.
 637          */
 638         ia = NULL;
 639         LIST_FOREACH(iac, INADDR_HASH(t.s_addr), ia_hash) {
 640                 if (t.s_addr == IA_SIN(iac->ia)->sin_addr.s_addr) {
 641                         ia = iac->ia;
 642                         goto match;
 643                 }
 644         }
 645         if (m->m_pkthdr.rcvif != NULL &&
 646             m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
 647                 struct ifaddr_container *ifac;
 648
 649                 TAILQ_FOREACH(ifac, &m->m_pkthdr.rcvif->if_addrheads[mycpuid],
 650                               ifa_link) {
 651                         struct ifaddr *ifa = ifac->ifa;
 652
 653                         if (ifa->ifa_addr->sa_family != AF_INET)
 654                                 continue;
 655                         ia = ifatoia(ifa);
 656                         if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
 657                             t.s_addr)
 658                                 goto match;
 659                 }
 660         }
 661         ia = ip_rtaddr(ip->ip_dst, ro);
 662         /* We need a route to do anything useful. */
 663         if (ia == NULL) {
 664                 m_freem(m);
 665                 icmpstat.icps_noroute++;
 666                 goto done;
 667         }
 668 match:
 669         t = IA_SIN(ia)->sin_addr;
 670         ip->ip_src = t;
 671         ip->ip_ttl = ip_defttl;
 672
 673         if (optlen > 0) {
 674                 u_char *cp;
 675                 int opt, cnt;
 676                 u_int len;
 677
 678                 /*
 679                  * Retrieve any source routing from the incoming packet;
 680                  * add on any record-route or timestamp options.
 681                  */
 682                 cp = (u_char *) (ip + 1);
 683                 if ((opts = ip_srcroute()) == 0 &&
 684                     (opts = m_gethdr(MB_DONTWAIT, MT_HEADER))) {
 685                         opts->m_len = sizeof(struct in_addr);
 686                         mtod(opts, struct in_addr *)->s_addr = 0;
 687                 }
 688                 if (opts) {
 689 #ifdef ICMPPRINTFS
 690                         if (icmpprintfs)
 691                                 kprintf("icmp_reflect optlen %d rt %d => ",
 692                                        optlen, opts->m_len);
 693 #endif
 694                         for (cnt = optlen; cnt > 0; cnt -= len, cp += len) {
 695                                 opt = cp[IPOPT_OPTVAL];
 696                                 if (opt == IPOPT_EOL)
 697                                         break;
 698                                 if (opt == IPOPT_NOP)
 699                                         len = 1;
 700                                 else {
 701                                         if (cnt < IPOPT_OLEN + sizeof *cp)
 702                                                 break;
 703                                         len = cp[IPOPT_OLEN];
 704                                         if (len < IPOPT_OLEN + sizeof *cp ||
 705                                             len > cnt)
 706                                         break;
 707                                 }
 708                                 /*
 709                                  * Should check for overflow, but it
 710                                  * "can't happen".
 711                                  */
 712                                 if (opt == IPOPT_RR || opt == IPOPT_TS ||
 713                                     opt == IPOPT_SECURITY) {
 714                                         bcopy(cp,
 715                                               mtod(opts, caddr_t) + opts->m_len,
 716                                               len);
 717                                         opts->m_len += len;
 718                                 }
 719                         }
 720                         /* Terminate & pad, if necessary */
 721                         cnt = opts->m_len % 4;
 722                         if (cnt) {
 723                                 for (; cnt < 4; cnt++) {
 724                                         *(mtod(opts, caddr_t) + opts->m_len) =
 725                                             IPOPT_EOL;
 726                                         opts->m_len++;
 727                                 }
 728                         }
 729 #ifdef ICMPPRINTFS
 730                         if (icmpprintfs)
 731                                 kprintf("%d\n", opts->m_len);
 732 #endif
 733                 }
 734                 /*
 735                  * Now strip out original options by copying rest of first
 736                  * mbuf's data back, and adjust the IP length.
 737                  */
 738                 ip->ip_len -= optlen;
 739                 ip->ip_vhl = IP_VHL_BORING;
 740                 m->m_len -= optlen;
 741                 if (m->m_flags & M_PKTHDR)
 742                         m->m_pkthdr.len -= optlen;
 743                 optlen += sizeof(struct ip);
 744                 bcopy((caddr_t)ip + optlen, ip + 1,
 745                       m->m_len - sizeof(struct ip));
 746         }
 747         m->m_pkthdr.fw_flags &= FW_MBUF_GENERATED;
 748         m->m_flags &= ~(M_BCAST|M_MCAST);
 749         icmp_send(m, opts, ro);
 750 done:
 751         if (opts)
 752                 m_free(opts);
 753         if (ro && ro->ro_rt)
 754                 RTFREE(ro->ro_rt);
 755 }
 756
 757 /*
 758  * Send an icmp packet back to the ip level,
 759  * after supplying a checksum.
 760  */
 761 static void
 762 icmp_send(struct mbuf *m, struct mbuf *opts, struct route *rt)
 763 {
 764         struct ip *ip = mtod(m, struct ip *);
 765         int hlen;
 766         struct icmp *icp;
 767
 768         hlen = IP_VHL_HL(ip->ip_vhl) << 2;
 769         m->m_data += hlen;
 770         m->m_len -= hlen;
 771         icp = mtod(m, struct icmp *);
 772         icp->icmp_cksum = 0;
 773         icp->icmp_cksum = in_cksum(m, ip->ip_len - hlen);
 774         m->m_data -= hlen;
 775         m->m_len += hlen;
 776         m->m_pkthdr.rcvif = (struct ifnet *)0;
 777 #ifdef ICMPPRINTFS
 778         if (icmpprintfs) {
 779                 char buf[sizeof "aaa.bbb.ccc.ddd"];
 780
 781                 strcpy(buf, inet_ntoa(ip->ip_dst));
 782                 kprintf("icmp_send dst %s src %s\n", buf, inet_ntoa(ip->ip_src));
 783         }
 784 #endif
 785         ip_output(m, opts, rt, 0, NULL, NULL);
 786 }
 787
 788 n_time
 789 iptime(void)
 790 {
 791         struct timeval atv;
 792         u_long t;
 793
 794         getmicrotime(&atv);
 795         t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000;
 796         return (htonl(t));
 797 }
 798
 799 #if 1
 800 /*
 801  * Return the next larger or smaller MTU plateau (table from RFC 1191)
 802  * given current value MTU.  If DIR is less than zero, a larger plateau
 803  * is returned; otherwise, a smaller value is returned.
 804  */
 805 int
 806 ip_next_mtu(int mtu, int dir)
 807 {
 808         static int mtutab[] = {
 809                 65535, 32000, 17914, 8166, 4352, 2002, 1492, 1006, 508, 296,
 810                 68, 0
 811         };
 812         int i;
 813
 814         for (i = 0; i < (sizeof mtutab) / (sizeof mtutab[0]); i++) {
 815                 if (mtu >= mtutab[i])
 816                         break;
 817         }
 818
 819         if (dir < 0) {
 820                 if (i == 0) {
 821                         return 0;
 822                 } else {
 823                         return mtutab[i - 1];
 824                 }
 825         } else {
 826                 if (mtutab[i] == 0) {
 827                         return 0;
 828                 } else if(mtu > mtutab[i]) {
 829                         return mtutab[i];
 830                 } else {
 831                         return mtutab[i + 1];
 832                 }
 833         }
 834 }
 835 #endif
 836
 837 #ifdef ICMP_BANDLIM
 838 /*
 839  * badport_bandlim() - check for ICMP bandwidth limit
 840  *
 841  *      Return 0 if it is ok to send an ICMP error response, -1 if we have
 842  *      hit our bandwidth limit and it is not ok.
 843  *
 844  *      If icmplim is <= 0, the feature is disabled and 0 is returned.
 845  *
 846  *      For now we separate the TCP and UDP subsystems w/ different 'which'
 847  *      values.  We may eventually remove this separation (and simplify the
 848  *      code further).
 849  *
 850  *      Note that the printing of the error message is delayed so we can
 851  *      properly print the icmp error rate that the system was trying to do
 852  *      (i.e. 22000/100 pps, etc...).  This can cause long delays in printing
 853  *      the 'final' error, but it doesn't make sense to solve the printing
 854  *      delay with more complex code.
 855  */
 856 int
 857 badport_bandlim(int which)
 858 {
 859         static int lticks[BANDLIM_MAX + 1];
 860         static int lpackets[BANDLIM_MAX + 1];
 861         int dticks;
 862         const char *bandlimittype[] = {
 863                 "Limiting icmp unreach response",
 864                 "Limiting icmp ping response",
 865                 "Limiting icmp tstamp response",
 866                 "Limiting closed port RST response",
 867                 "Limiting open port RST response"
 868                 };
 869
 870         /*
 871          * Return ok status if feature disabled or argument out of
 872          * ranage.
 873          */
 874
 875         if (icmplim <= 0 || which > BANDLIM_MAX || which < 0)
 876                 return(0);
 877         dticks = ticks - lticks[which];
 878
 879         /*
 880          * reset stats when cumulative dt exceeds one second.
 881          */
 882
 883         if ((unsigned int)dticks > hz) {
 884                 if (lpackets[which] > icmplim && icmplim_output) {
 885                         kprintf("%s from %d to %d packets per second\n",
 886                                 bandlimittype[which],
 887                                 lpackets[which],
 888                                 icmplim
 889                         );
 890                 }
 891                 lticks[which] = ticks;
 892                 lpackets[which] = 0;
 893         }
 894
 895         /*
 896          * bump packet count
 897          */
 898
 899         if (++lpackets[which] > icmplim) {
 900                 return(-1);
 901         }
 902         return(0);
 903 }
 904 #endif