1 /* $NetBSD: ip_icmp.c,v 1.71 2002/09/23 05:51:12 simonb Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /*- 33 * Copyright (c) 1998, 2000 The NetBSD Foundation, Inc. 34 * All rights reserved. 35 * 36 * This code is derived from software contributed to The NetBSD Foundation 37 * by Public Access Networks Corporation ("Panix"). It was developed under 38 * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon. 39 * 40 * This code is derived from software contributed to The NetBSD Foundation 41 * by Jason R. Thorpe of Zembu Labs, Inc. 42 * 43 * Redistribution and use in source and binary forms, with or without 44 * modification, are permitted provided that the following conditions 45 * are met: 46 * 1. Redistributions of source code must retain the above copyright 47 * notice, this list of conditions and the following disclaimer. 48 * 2. Redistributions in binary form must reproduce the above copyright 49 * notice, this list of conditions and the following disclaimer in the 50 * documentation and/or other materials provided with the distribution. 51 * 3. All advertising materials mentioning features or use of this software 52 * must display the following acknowledgement: 53 * This product includes software developed by the NetBSD 54 * Foundation, Inc. and its contributors. 55 * 4. Neither the name of The NetBSD Foundation nor the names of its 56 * contributors may be used to endorse or promote products derived 57 * from this software without specific prior written permission. 58 * 59 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 60 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 61 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 62 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 63 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 64 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 65 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 66 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 67 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 68 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 69 * POSSIBILITY OF SUCH DAMAGE. 70 */ 71 72 /* 73 * Copyright (c) 1982, 1986, 1988, 1993 74 * The Regents of the University of California. All rights reserved. 75 * 76 * Redistribution and use in source and binary forms, with or without 77 * modification, are permitted provided that the following conditions 78 * are met: 79 * 1. Redistributions of source code must retain the above copyright 80 * notice, this list of conditions and the following disclaimer. 81 * 2. Redistributions in binary form must reproduce the above copyright 82 * notice, this list of conditions and the following disclaimer in the 83 * documentation and/or other materials provided with the distribution. 84 * 3. All advertising materials mentioning features or use of this software 85 * must display the following acknowledgement: 86 * This product includes software developed by the University of 87 * California, Berkeley and its contributors. 88 * 4. Neither the name of the University nor the names of its contributors 89 * may be used to endorse or promote products derived from this software 90 * without specific prior written permission. 91 * 92 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 93 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 94 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 95 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 96 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 97 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 98 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 99 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 100 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 101 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 102 * SUCH DAMAGE. 103 * 104 * @(#)ip_icmp.c 8.2 (Berkeley) 1/4/94 105 */ 106 107 #include <sys/cdefs.h> 108 __KERNEL_RCSID(0, "$NetBSD: ip_icmp.c,v 1.71 2002/09/23 05:51:12 simonb Exp $"); 109 110 #include "opt_ipsec.h" 111 112 #include <sys/param.h> 113 #include <sys/systm.h> 114 #include <sys/malloc.h> 115 #include <sys/mbuf.h> 116 #include <sys/protosw.h> 117 #include <sys/socket.h> 118 #include <sys/time.h> 119 #include <sys/kernel.h> 120 #include <sys/syslog.h> 121 #include <sys/sysctl.h> 122 123 #include <net/if.h> 124 #include <net/route.h> 125 126 #include <netinet/in.h> 127 #include <netinet/in_systm.h> 128 #include <netinet/in_var.h> 129 #include <netinet/ip.h> 130 #include <netinet/ip_icmp.h> 131 #include <netinet/ip_var.h> 132 #include <netinet/in_pcb.h> 133 #include <netinet/icmp_var.h> 134 135 #ifdef IPSEC 136 #include <netinet6/ipsec.h> 137 #include <netkey/key.h> 138 #endif 139 140 #include <machine/stdarg.h> 141 142 /* 143 * ICMP routines: error generation, receive packet processing, and 144 * routines to turnaround packets back to the originator, and 145 * host table maintenance routines. 146 */ 147 148 int icmpmaskrepl = 0; 149 #ifdef ICMPPRINTFS 150 int icmpprintfs = 0; 151 #endif 152 int icmpreturndatabytes = 8; 153 154 /* 155 * List of callbacks to notify when Path MTU changes are made. 156 */ 157 struct icmp_mtudisc_callback { 158 LIST_ENTRY(icmp_mtudisc_callback) mc_list; 159 void (*mc_func) __P((struct in_addr)); 160 }; 161 162 LIST_HEAD(, icmp_mtudisc_callback) icmp_mtudisc_callbacks = 163 LIST_HEAD_INITIALIZER(&icmp_mtudisc_callbacks); 164 165 #if 0 166 static int ip_next_mtu __P((int, int)); 167 #else 168 /*static*/ int ip_next_mtu __P((int, int)); 169 #endif 170 171 extern int icmperrppslim; 172 static int icmperrpps_count = 0; 173 static struct timeval icmperrppslim_last; 174 static int icmp_rediraccept = 1; 175 static int icmp_redirtimeout = 600; 176 static struct rttimer_queue *icmp_redirect_timeout_q = NULL; 177 178 static void icmp_mtudisc_timeout __P((struct rtentry *, struct rttimer *)); 179 static void icmp_redirect_timeout __P((struct rtentry *, struct rttimer *)); 180 181 static int icmp_ratelimit __P((const struct in_addr *, const int, const int)); 182 183 184 void 185 icmp_init() 186 { 187 /* 188 * This is only useful if the user initializes redirtimeout to 189 * something other than zero. 190 */ 191 if (icmp_redirtimeout != 0) { 192 icmp_redirect_timeout_q = 193 rt_timer_queue_create(icmp_redirtimeout); 194 } 195 } 196 197 /* 198 * Register a Path MTU Discovery callback. 199 */ 200 void 201 icmp_mtudisc_callback_register(func) 202 void (*func) __P((struct in_addr)); 203 { 204 struct icmp_mtudisc_callback *mc; 205 206 for (mc = LIST_FIRST(&icmp_mtudisc_callbacks); mc != NULL; 207 mc = LIST_NEXT(mc, mc_list)) { 208 if (mc->mc_func == func) 209 return; 210 } 211 212 mc = malloc(sizeof(*mc), M_PCB, M_NOWAIT); 213 if (mc == NULL) 214 panic("icmp_mtudisc_callback_register"); 215 216 mc->mc_func = func; 217 LIST_INSERT_HEAD(&icmp_mtudisc_callbacks, mc, mc_list); 218 } 219 220 /* 221 * Generate an error packet of type error 222 * in response to bad packet ip. 223 */ 224 void 225 icmp_error(n, type, code, dest, destifp) 226 struct mbuf *n; 227 int type, code; 228 n_long dest; 229 struct ifnet *destifp; 230 { 231 struct ip *oip = mtod(n, struct ip *), *nip; 232 unsigned oiplen = oip->ip_hl << 2; 233 struct icmp *icp; 234 struct mbuf *m; 235 unsigned icmplen, mblen; 236 237 #ifdef ICMPPRINTFS 238 if (icmpprintfs) 239 printf("icmp_error(%x, %d, %d)\n", oip, type, code); 240 #endif 241 if (type != ICMP_REDIRECT) 242 icmpstat.icps_error++; 243 /* 244 * Don't send error if the original packet was encrypted. 245 * Don't send error if not the first fragment of message. 246 * Don't error if the old packet protocol was ICMP 247 * error message, only known informational types. 248 */ 249 if (n->m_flags & M_DECRYPTED) 250 goto freeit; 251 if (oip->ip_off &~ htons(IP_MF|IP_DF)) 252 goto freeit; 253 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT && 254 n->m_len >= oiplen + ICMP_MINLEN && 255 !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiplen))->icmp_type)) { 256 icmpstat.icps_oldicmp++; 257 goto freeit; 258 } 259 /* Don't send error in response to a multicast or broadcast packet */ 260 if (n->m_flags & (M_BCAST|M_MCAST)) 261 goto freeit; 262 263 /* 264 * First, do a rate limitation check. 265 */ 266 if (icmp_ratelimit(&oip->ip_src, type, code)) { 267 /* XXX stat */ 268 goto freeit; 269 } 270 271 /* 272 * Now, formulate icmp message 273 */ 274 icmplen = oiplen + min(icmpreturndatabytes, 275 ntohs(oip->ip_len) - oiplen); 276 /* 277 * Defend against mbuf chains shorter than oip->ip_len: 278 */ 279 mblen = 0; 280 for (m = n; m && (mblen < icmplen); m = m->m_next) 281 mblen += m->m_len; 282 icmplen = min(mblen, icmplen); 283 284 /* 285 * As we are not required to return everything we have, 286 * we return whatever we can return at ease. 287 * 288 * Note that ICMP datagrams longer than 576 octets are out of spec 289 * according to RFC1812; the limit on icmpreturndatabytes below in 290 * icmp_sysctl will keep things below that limit. 291 */ 292 293 KASSERT(ICMP_MINLEN <= MCLBYTES); 294 295 if (icmplen + ICMP_MINLEN > MCLBYTES) 296 icmplen = MCLBYTES - ICMP_MINLEN; 297 298 m = m_gethdr(M_DONTWAIT, MT_HEADER); 299 if (m && (icmplen + ICMP_MINLEN > MHLEN)) { 300 MCLGET(m, M_DONTWAIT); 301 if ((m->m_flags & M_EXT) == 0) { 302 m_freem(m); 303 m = NULL; 304 } 305 } 306 if (m == NULL) 307 goto freeit; 308 m->m_len = icmplen + ICMP_MINLEN; 309 if ((m->m_flags & M_EXT) == 0) 310 MH_ALIGN(m, m->m_len); 311 icp = mtod(m, struct icmp *); 312 if ((u_int)type > ICMP_MAXTYPE) 313 panic("icmp_error"); 314 icmpstat.icps_outhist[type]++; 315 icp->icmp_type = type; 316 if (type == ICMP_REDIRECT) 317 icp->icmp_gwaddr.s_addr = dest; 318 else { 319 icp->icmp_void = 0; 320 /* 321 * The following assignments assume an overlay with the 322 * zeroed icmp_void field. 323 */ 324 if (type == ICMP_PARAMPROB) { 325 icp->icmp_pptr = code; 326 code = 0; 327 } else if (type == ICMP_UNREACH && 328 code == ICMP_UNREACH_NEEDFRAG && destifp) 329 icp->icmp_nextmtu = htons(destifp->if_mtu); 330 } 331 332 icp->icmp_code = code; 333 m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip); 334 nip = &icp->icmp_ip; 335 336 /* 337 * Now, copy old ip header (without options) 338 * in front of icmp message. 339 */ 340 if (m->m_data - sizeof(struct ip) < m->m_pktdat) 341 panic("icmp len"); 342 m->m_data -= sizeof(struct ip); 343 m->m_len += sizeof(struct ip); 344 m->m_pkthdr.len = m->m_len; 345 m->m_pkthdr.rcvif = n->m_pkthdr.rcvif; 346 nip = mtod(m, struct ip *); 347 /* ip_v set in ip_output */ 348 nip->ip_hl = sizeof(struct ip) >> 2; 349 nip->ip_tos = 0; 350 nip->ip_len = htons(m->m_len); 351 /* ip_id set in ip_output */ 352 nip->ip_off = htons(0); 353 /* ip_ttl set in icmp_reflect */ 354 nip->ip_p = IPPROTO_ICMP; 355 nip->ip_src = oip->ip_src; 356 nip->ip_dst = oip->ip_dst; 357 icmp_reflect(m); 358 359 freeit: 360 m_freem(n); 361 } 362 363 static struct sockaddr_in icmpsrc = { sizeof (struct sockaddr_in), AF_INET }; 364 static struct sockaddr_in icmpdst = { sizeof (struct sockaddr_in), AF_INET }; 365 static struct sockaddr_in icmpgw = { sizeof (struct sockaddr_in), AF_INET }; 366 struct sockaddr_in icmpmask = { 8, 0 }; 367 368 /* 369 * Process a received ICMP message. 370 */ 371 void 372 #if __STDC__ 373 icmp_input(struct mbuf *m, ...) 374 #else 375 icmp_input(m, va_alist) 376 struct mbuf *m; 377 va_dcl 378 #endif 379 { 380 int proto; 381 struct icmp *icp; 382 struct ip *ip = mtod(m, struct ip *); 383 int icmplen; 384 int i; 385 struct in_ifaddr *ia; 386 void *(*ctlfunc) __P((int, struct sockaddr *, void *)); 387 int code; 388 int hlen; 389 va_list ap; 390 struct rtentry *rt; 391 392 va_start(ap, m); 393 hlen = va_arg(ap, int); 394 proto = va_arg(ap, int); 395 va_end(ap); 396 397 /* 398 * Locate icmp structure in mbuf, and check 399 * that not corrupted and of at least minimum length. 400 */ 401 icmplen = ntohs(ip->ip_len) - hlen; 402 #ifdef ICMPPRINTFS 403 if (icmpprintfs) 404 printf("icmp_input from %x to %x, len %d\n", 405 ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr), 406 icmplen); 407 #endif 408 if (icmplen < ICMP_MINLEN) { 409 icmpstat.icps_tooshort++; 410 goto freeit; 411 } 412 i = hlen + min(icmplen, ICMP_ADVLENMIN); 413 if (m->m_len < i && (m = m_pullup(m, i)) == 0) { 414 icmpstat.icps_tooshort++; 415 return; 416 } 417 ip = mtod(m, struct ip *); 418 m->m_len -= hlen; 419 m->m_data += hlen; 420 icp = mtod(m, struct icmp *); 421 /* Don't need to assert alignment, here. */ 422 if (in_cksum(m, icmplen)) { 423 icmpstat.icps_checksum++; 424 goto freeit; 425 } 426 m->m_len += hlen; 427 m->m_data -= hlen; 428 429 #ifdef ICMPPRINTFS 430 /* 431 * Message type specific processing. 432 */ 433 if (icmpprintfs) 434 printf("icmp_input, type %d code %d\n", icp->icmp_type, 435 icp->icmp_code); 436 #endif 437 if (icp->icmp_type > ICMP_MAXTYPE) 438 goto raw; 439 icmpstat.icps_inhist[icp->icmp_type]++; 440 code = icp->icmp_code; 441 switch (icp->icmp_type) { 442 443 case ICMP_UNREACH: 444 switch (code) { 445 case ICMP_UNREACH_NET: 446 case ICMP_UNREACH_HOST: 447 case ICMP_UNREACH_PROTOCOL: 448 case ICMP_UNREACH_PORT: 449 case ICMP_UNREACH_SRCFAIL: 450 code += PRC_UNREACH_NET; 451 break; 452 453 case ICMP_UNREACH_NEEDFRAG: 454 code = PRC_MSGSIZE; 455 break; 456 457 case ICMP_UNREACH_NET_UNKNOWN: 458 case ICMP_UNREACH_NET_PROHIB: 459 case ICMP_UNREACH_TOSNET: 460 code = PRC_UNREACH_NET; 461 break; 462 463 case ICMP_UNREACH_HOST_UNKNOWN: 464 case ICMP_UNREACH_ISOLATED: 465 case ICMP_UNREACH_HOST_PROHIB: 466 case ICMP_UNREACH_TOSHOST: 467 code = PRC_UNREACH_HOST; 468 break; 469 470 default: 471 goto badcode; 472 } 473 goto deliver; 474 475 case ICMP_TIMXCEED: 476 if (code > 1) 477 goto badcode; 478 code += PRC_TIMXCEED_INTRANS; 479 goto deliver; 480 481 case ICMP_PARAMPROB: 482 if (code > 1) 483 goto badcode; 484 code = PRC_PARAMPROB; 485 goto deliver; 486 487 case ICMP_SOURCEQUENCH: 488 if (code) 489 goto badcode; 490 code = PRC_QUENCH; 491 goto deliver; 492 493 deliver: 494 /* 495 * Problem with datagram; advise higher level routines. 496 */ 497 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || 498 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) { 499 icmpstat.icps_badlen++; 500 goto freeit; 501 } 502 if (IN_MULTICAST(icp->icmp_ip.ip_dst.s_addr)) 503 goto badcode; 504 #ifdef ICMPPRINTFS 505 if (icmpprintfs) 506 printf("deliver to protocol %d\n", icp->icmp_ip.ip_p); 507 #endif 508 icmpsrc.sin_addr = icp->icmp_ip.ip_dst; 509 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput; 510 if (ctlfunc) 511 (void) (*ctlfunc)(code, sintosa(&icmpsrc), 512 &icp->icmp_ip); 513 break; 514 515 badcode: 516 icmpstat.icps_badcode++; 517 break; 518 519 case ICMP_ECHO: 520 icp->icmp_type = ICMP_ECHOREPLY; 521 goto reflect; 522 523 case ICMP_TSTAMP: 524 if (icmplen < ICMP_TSLEN) { 525 icmpstat.icps_badlen++; 526 break; 527 } 528 icp->icmp_type = ICMP_TSTAMPREPLY; 529 icp->icmp_rtime = iptime(); 530 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */ 531 goto reflect; 532 533 case ICMP_MASKREQ: 534 if (icmpmaskrepl == 0) 535 break; 536 /* 537 * We are not able to respond with all ones broadcast 538 * unless we receive it over a point-to-point interface. 539 */ 540 if (icmplen < ICMP_MASKLEN) { 541 icmpstat.icps_badlen++; 542 break; 543 } 544 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 545 in_nullhost(ip->ip_dst)) 546 icmpdst.sin_addr = ip->ip_src; 547 else 548 icmpdst.sin_addr = ip->ip_dst; 549 ia = ifatoia(ifaof_ifpforaddr(sintosa(&icmpdst), 550 m->m_pkthdr.rcvif)); 551 if (ia == 0) 552 break; 553 icp->icmp_type = ICMP_MASKREPLY; 554 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr; 555 if (in_nullhost(ip->ip_src)) { 556 if (ia->ia_ifp->if_flags & IFF_BROADCAST) 557 ip->ip_src = ia->ia_broadaddr.sin_addr; 558 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT) 559 ip->ip_src = ia->ia_dstaddr.sin_addr; 560 } 561 reflect: 562 icmpstat.icps_reflect++; 563 icmpstat.icps_outhist[icp->icmp_type]++; 564 icmp_reflect(m); 565 return; 566 567 case ICMP_REDIRECT: 568 if (code > 3) 569 goto badcode; 570 if (icmp_rediraccept == 0) 571 goto freeit; 572 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || 573 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) { 574 icmpstat.icps_badlen++; 575 break; 576 } 577 /* 578 * Short circuit routing redirects to force 579 * immediate change in the kernel's routing 580 * tables. The message is also handed to anyone 581 * listening on a raw socket (e.g. the routing 582 * daemon for use in updating its tables). 583 */ 584 icmpgw.sin_addr = ip->ip_src; 585 icmpdst.sin_addr = icp->icmp_gwaddr; 586 #ifdef ICMPPRINTFS 587 if (icmpprintfs) 588 printf("redirect dst %x to %x\n", icp->icmp_ip.ip_dst, 589 icp->icmp_gwaddr); 590 #endif 591 icmpsrc.sin_addr = icp->icmp_ip.ip_dst; 592 rt = NULL; 593 rtredirect(sintosa(&icmpsrc), sintosa(&icmpdst), 594 (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST, 595 sintosa(&icmpgw), (struct rtentry **)&rt); 596 if (rt != NULL && icmp_redirtimeout != 0) { 597 i = rt_timer_add(rt, icmp_redirect_timeout, 598 icmp_redirect_timeout_q); 599 if (i) 600 log(LOG_ERR, "ICMP: redirect failed to " 601 "register timeout for route to %x, " 602 "code %d\n", 603 icp->icmp_ip.ip_dst.s_addr, i); 604 } 605 if (rt != NULL) 606 rtfree(rt); 607 608 pfctlinput(PRC_REDIRECT_HOST, sintosa(&icmpsrc)); 609 #ifdef IPSEC 610 key_sa_routechange((struct sockaddr *)&icmpsrc); 611 #endif 612 break; 613 614 /* 615 * No kernel processing for the following; 616 * just fall through to send to raw listener. 617 */ 618 case ICMP_ECHOREPLY: 619 case ICMP_ROUTERADVERT: 620 case ICMP_ROUTERSOLICIT: 621 case ICMP_TSTAMPREPLY: 622 case ICMP_IREQREPLY: 623 case ICMP_MASKREPLY: 624 default: 625 break; 626 } 627 628 raw: 629 rip_input(m, hlen, proto); 630 return; 631 632 freeit: 633 m_freem(m); 634 return; 635 } 636 637 /* 638 * Reflect the ip packet back to the source 639 */ 640 void 641 icmp_reflect(m) 642 struct mbuf *m; 643 { 644 struct ip *ip = mtod(m, struct ip *); 645 struct in_ifaddr *ia; 646 struct ifaddr *ifa; 647 struct sockaddr_in *sin = 0; 648 struct in_addr t; 649 struct mbuf *opts = 0; 650 int optlen = (ip->ip_hl << 2) - sizeof(struct ip); 651 652 if (!in_canforward(ip->ip_src) && 653 ((ip->ip_src.s_addr & IN_CLASSA_NET) != 654 htonl(IN_LOOPBACKNET << IN_CLASSA_NSHIFT))) { 655 m_freem(m); /* Bad return address */ 656 goto done; /* ip_output() will check for broadcast */ 657 } 658 t = ip->ip_dst; 659 ip->ip_dst = ip->ip_src; 660 /* 661 * If the incoming packet was addressed directly to us, use 662 * dst as the src for the reply. Otherwise (broadcast or 663 * anonymous), use an address which corresponds to the 664 * incoming interface, with a preference for the address which 665 * corresponds to the route to the destination of the ICMP. 666 */ 667 668 /* Look for packet addressed to us */ 669 INADDR_TO_IA(t, ia); 670 671 /* look for packet sent to broadcast address */ 672 if (ia == NULL && (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST)) { 673 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) { 674 if (ifa->ifa_addr->sa_family != AF_INET) 675 continue; 676 if (in_hosteq(t,ifatoia(ifa)->ia_broadaddr.sin_addr)) { 677 ia = ifatoia(ifa); 678 break; 679 } 680 } 681 } 682 683 if (ia) 684 sin = &ia->ia_addr; 685 686 icmpdst.sin_addr = t; 687 688 /* 689 * if the packet is addressed somewhere else, compute the 690 * source address for packets routed back to the source, and 691 * use that, if it's an address on the interface which 692 * received the packet 693 */ 694 if (sin == (struct sockaddr_in *)0) { 695 struct sockaddr_in sin_dst; 696 struct route icmproute; 697 int errornum; 698 699 sin_dst.sin_family = AF_INET; 700 sin_dst.sin_len = sizeof(struct sockaddr_in); 701 sin_dst.sin_addr = ip->ip_dst; 702 bzero(&icmproute, sizeof(icmproute)); 703 errornum = 0; 704 sin = in_selectsrc(&sin_dst, &icmproute, 0, NULL, &errornum); 705 /* errornum is never used */ 706 if (icmproute.ro_rt) 707 RTFREE(icmproute.ro_rt); 708 /* check to make sure sin is a source address on rcvif */ 709 if (sin) { 710 t = sin->sin_addr; 711 sin = (struct sockaddr_in *)0; 712 INADDR_TO_IA(t, ia); 713 while (ia) { 714 if (ia->ia_ifp == m->m_pkthdr.rcvif) { 715 sin = &ia->ia_addr; 716 break; 717 } 718 NEXT_IA_WITH_SAME_ADDR(ia); 719 } 720 } 721 } 722 723 /* 724 * if it was not addressed to us, but the route doesn't go out 725 * the source interface, pick an address on the source 726 * interface. This can happen when routing is asymmetric, or 727 * when the incoming packet was encapsulated 728 */ 729 if (sin == (struct sockaddr_in *)0) { 730 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) { 731 if (ifa->ifa_addr->sa_family != AF_INET) 732 continue; 733 sin = &(ifatoia(ifa)->ia_addr); 734 break; 735 } 736 } 737 738 /* 739 * The following happens if the packet was not addressed to us, 740 * and was received on an interface with no IP address: 741 * We find the first AF_INET address on the first non-loopback 742 * interface. 743 */ 744 if (sin == (struct sockaddr_in *)0) 745 TAILQ_FOREACH(ia, &in_ifaddr, ia_list) { 746 if (ia->ia_ifp->if_flags & IFF_LOOPBACK) 747 continue; 748 sin = &ia->ia_addr; 749 break; 750 } 751 752 /* 753 * If we still didn't find an address, punt. We could have an 754 * interface up (and receiving packets) with no address. 755 */ 756 if (sin == (struct sockaddr_in *)0) { 757 m_freem(m); 758 goto done; 759 } 760 761 ip->ip_src = sin->sin_addr; 762 ip->ip_ttl = MAXTTL; 763 764 if (optlen > 0) { 765 u_char *cp; 766 int opt, cnt; 767 u_int len; 768 769 /* 770 * Retrieve any source routing from the incoming packet; 771 * add on any record-route or timestamp options. 772 */ 773 cp = (u_char *) (ip + 1); 774 if ((opts = ip_srcroute()) == 0 && 775 (opts = m_gethdr(M_DONTWAIT, MT_HEADER))) { 776 opts->m_len = sizeof(struct in_addr); 777 *mtod(opts, struct in_addr *) = zeroin_addr; 778 } 779 if (opts) { 780 #ifdef ICMPPRINTFS 781 if (icmpprintfs) 782 printf("icmp_reflect optlen %d rt %d => ", 783 optlen, opts->m_len); 784 #endif 785 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) { 786 opt = cp[IPOPT_OPTVAL]; 787 if (opt == IPOPT_EOL) 788 break; 789 if (opt == IPOPT_NOP) 790 len = 1; 791 else { 792 if (cnt < IPOPT_OLEN + sizeof(*cp)) 793 break; 794 len = cp[IPOPT_OLEN]; 795 if (len < IPOPT_OLEN + sizeof(*cp) || 796 len > cnt) 797 break; 798 } 799 /* 800 * Should check for overflow, but it "can't happen" 801 */ 802 if (opt == IPOPT_RR || opt == IPOPT_TS || 803 opt == IPOPT_SECURITY) { 804 bcopy((caddr_t)cp, 805 mtod(opts, caddr_t) + opts->m_len, len); 806 opts->m_len += len; 807 } 808 } 809 /* Terminate & pad, if necessary */ 810 if ((cnt = opts->m_len % 4) != 0) { 811 for (; cnt < 4; cnt++) { 812 *(mtod(opts, caddr_t) + opts->m_len) = 813 IPOPT_EOL; 814 opts->m_len++; 815 } 816 } 817 #ifdef ICMPPRINTFS 818 if (icmpprintfs) 819 printf("%d\n", opts->m_len); 820 #endif 821 } 822 /* 823 * Now strip out original options by copying rest of first 824 * mbuf's data back, and adjust the IP length. 825 */ 826 ip->ip_len = htons(ntohs(ip->ip_len) - optlen); 827 ip->ip_hl = sizeof(struct ip) >> 2; 828 m->m_len -= optlen; 829 if (m->m_flags & M_PKTHDR) 830 m->m_pkthdr.len -= optlen; 831 optlen += sizeof(struct ip); 832 bcopy((caddr_t)ip + optlen, (caddr_t)(ip + 1), 833 (unsigned)(m->m_len - sizeof(struct ip))); 834 } 835 m->m_flags &= ~(M_BCAST|M_MCAST); 836 icmp_send(m, opts); 837 done: 838 if (opts) 839 (void)m_free(opts); 840 } 841 842 /* 843 * Send an icmp packet back to the ip level, 844 * after supplying a checksum. 845 */ 846 void 847 icmp_send(m, opts) 848 struct mbuf *m; 849 struct mbuf *opts; 850 { 851 struct ip *ip = mtod(m, struct ip *); 852 int hlen; 853 struct icmp *icp; 854 855 hlen = ip->ip_hl << 2; 856 m->m_data += hlen; 857 m->m_len -= hlen; 858 icp = mtod(m, struct icmp *); 859 icp->icmp_cksum = 0; 860 icp->icmp_cksum = in_cksum(m, ntohs(ip->ip_len) - hlen); 861 m->m_data -= hlen; 862 m->m_len += hlen; 863 #ifdef ICMPPRINTFS 864 if (icmpprintfs) 865 printf("icmp_send dst %x src %x\n", ip->ip_dst, ip->ip_src); 866 #endif 867 #ifdef IPSEC 868 /* Don't lookup socket */ 869 (void)ipsec_setsocket(m, NULL); 870 #endif 871 (void) ip_output(m, opts, NULL, 0, NULL); 872 } 873 874 n_time 875 iptime() 876 { 877 struct timeval atv; 878 u_long t; 879 880 microtime(&atv); 881 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000; 882 return (htonl(t)); 883 } 884 885 int 886 icmp_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 887 int *name; 888 u_int namelen; 889 void *oldp; 890 size_t *oldlenp; 891 void *newp; 892 size_t newlen; 893 { 894 int arg, error; 895 896 /* All sysctl names at this level are terminal. */ 897 if (namelen != 1) 898 return (ENOTDIR); 899 900 switch (name[0]) 901 { 902 case ICMPCTL_MASKREPL: 903 error = sysctl_int(oldp, oldlenp, newp, newlen, &icmpmaskrepl); 904 break; 905 case ICMPCTL_RETURNDATABYTES: 906 arg = icmpreturndatabytes; 907 error = sysctl_int(oldp, oldlenp, newp, newlen, &arg); 908 if (error) 909 break; 910 if ((arg >= 8) || (arg <= 512)) 911 icmpreturndatabytes = arg; 912 else 913 error = EINVAL; 914 break; 915 case ICMPCTL_ERRPPSLIMIT: 916 error = sysctl_int(oldp, oldlenp, newp, newlen, &icmperrppslim); 917 break; 918 case ICMPCTL_REDIRACCEPT: 919 error = sysctl_int(oldp, oldlenp, newp, newlen, 920 &icmp_rediraccept); 921 break; 922 case ICMPCTL_REDIRTIMEOUT: 923 error = sysctl_int(oldp, oldlenp, newp, newlen, 924 &icmp_redirtimeout); 925 if (icmp_redirect_timeout_q != NULL) { 926 if (icmp_redirtimeout == 0) { 927 rt_timer_queue_destroy(icmp_redirect_timeout_q, 928 TRUE); 929 icmp_redirect_timeout_q = NULL; 930 } else { 931 rt_timer_queue_change(icmp_redirect_timeout_q, 932 icmp_redirtimeout); 933 } 934 } else if (icmp_redirtimeout > 0) { 935 icmp_redirect_timeout_q = 936 rt_timer_queue_create(icmp_redirtimeout); 937 } 938 return (error); 939 default: 940 error = ENOPROTOOPT; 941 break; 942 } 943 return error; 944 } 945 946 /* Table of common MTUs: */ 947 948 static const u_int mtu_table[] = { 949 65535, 65280, 32000, 17914, 9180, 8166, 950 4352, 2002, 1492, 1006, 508, 296, 68, 0 951 }; 952 953 void 954 icmp_mtudisc(icp, faddr) 955 struct icmp *icp; 956 struct in_addr faddr; 957 { 958 struct icmp_mtudisc_callback *mc; 959 struct sockaddr *dst = sintosa(&icmpsrc); 960 struct rtentry *rt; 961 u_long mtu = ntohs(icp->icmp_nextmtu); /* Why a long? IPv6 */ 962 int error; 963 964 rt = rtalloc1(dst, 1); 965 if (rt == 0) 966 return; 967 968 /* If we didn't get a host route, allocate one */ 969 970 if ((rt->rt_flags & RTF_HOST) == 0) { 971 struct rtentry *nrt; 972 973 error = rtrequest((int) RTM_ADD, dst, 974 (struct sockaddr *) rt->rt_gateway, 975 (struct sockaddr *) 0, 976 RTF_GATEWAY | RTF_HOST | RTF_DYNAMIC, &nrt); 977 if (error) { 978 rtfree(rt); 979 return; 980 } 981 nrt->rt_rmx = rt->rt_rmx; 982 rtfree(rt); 983 rt = nrt; 984 } 985 error = rt_timer_add(rt, icmp_mtudisc_timeout, ip_mtudisc_timeout_q); 986 if (error) { 987 rtfree(rt); 988 return; 989 } 990 991 if (mtu == 0) { 992 int i = 0; 993 994 mtu = ntohs(icp->icmp_ip.ip_len); 995 /* Some 4.2BSD-based routers incorrectly adjust the ip_len */ 996 if (mtu > rt->rt_rmx.rmx_mtu && rt->rt_rmx.rmx_mtu != 0) 997 mtu -= (icp->icmp_ip.ip_hl << 2); 998 999 /* If we still can't guess a value, try the route */ 1000 1001 if (mtu == 0) { 1002 mtu = rt->rt_rmx.rmx_mtu; 1003 1004 /* If no route mtu, default to the interface mtu */ 1005 1006 if (mtu == 0) 1007 mtu = rt->rt_ifp->if_mtu; 1008 } 1009 1010 for (i = 0; i < sizeof(mtu_table) / sizeof(mtu_table[0]); i++) 1011 if (mtu > mtu_table[i]) { 1012 mtu = mtu_table[i]; 1013 break; 1014 } 1015 } 1016 1017 /* 1018 * XXX: RTV_MTU is overloaded, since the admin can set it 1019 * to turn off PMTU for a route, and the kernel can 1020 * set it to indicate a serious problem with PMTU 1021 * on a route. We should be using a separate flag 1022 * for the kernel to indicate this. 1023 */ 1024 1025 if ((rt->rt_rmx.rmx_locks & RTV_MTU) == 0) { 1026 if (mtu < 296 || mtu > rt->rt_ifp->if_mtu) 1027 rt->rt_rmx.rmx_locks |= RTV_MTU; 1028 else if (rt->rt_rmx.rmx_mtu > mtu || 1029 rt->rt_rmx.rmx_mtu == 0) { 1030 icmpstat.icps_pmtuchg++; 1031 rt->rt_rmx.rmx_mtu = mtu; 1032 } 1033 } 1034 1035 if (rt) 1036 rtfree(rt); 1037 1038 /* 1039 * Notify protocols that the MTU for this destination 1040 * has changed. 1041 */ 1042 for (mc = LIST_FIRST(&icmp_mtudisc_callbacks); mc != NULL; 1043 mc = LIST_NEXT(mc, mc_list)) 1044 (*mc->mc_func)(faddr); 1045 } 1046 1047 /* 1048 * Return the next larger or smaller MTU plateau (table from RFC 1191) 1049 * given current value MTU. If DIR is less than zero, a larger plateau 1050 * is returned; otherwise, a smaller value is returned. 1051 */ 1052 int 1053 ip_next_mtu(mtu, dir) /* XXX */ 1054 int mtu; 1055 int dir; 1056 { 1057 int i; 1058 1059 for (i = 0; i < (sizeof mtu_table) / (sizeof mtu_table[0]); i++) { 1060 if (mtu >= mtu_table[i]) 1061 break; 1062 } 1063 1064 if (dir < 0) { 1065 if (i == 0) { 1066 return 0; 1067 } else { 1068 return mtu_table[i - 1]; 1069 } 1070 } else { 1071 if (mtu_table[i] == 0) { 1072 return 0; 1073 } else if (mtu > mtu_table[i]) { 1074 return mtu_table[i]; 1075 } else { 1076 return mtu_table[i + 1]; 1077 } 1078 } 1079 } 1080 1081 static void 1082 icmp_mtudisc_timeout(rt, r) 1083 struct rtentry *rt; 1084 struct rttimer *r; 1085 { 1086 if (rt == NULL) 1087 panic("icmp_mtudisc_timeout: bad route to timeout"); 1088 if ((rt->rt_flags & (RTF_DYNAMIC | RTF_HOST)) == 1089 (RTF_DYNAMIC | RTF_HOST)) { 1090 rtrequest((int) RTM_DELETE, (struct sockaddr *)rt_key(rt), 1091 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 0); 1092 } else { 1093 if ((rt->rt_rmx.rmx_locks & RTV_MTU) == 0) { 1094 rt->rt_rmx.rmx_mtu = 0; 1095 } 1096 } 1097 } 1098 1099 static void 1100 icmp_redirect_timeout(rt, r) 1101 struct rtentry *rt; 1102 struct rttimer *r; 1103 { 1104 if (rt == NULL) 1105 panic("icmp_redirect_timeout: bad route to timeout"); 1106 if ((rt->rt_flags & (RTF_DYNAMIC | RTF_HOST)) == 1107 (RTF_DYNAMIC | RTF_HOST)) { 1108 rtrequest((int) RTM_DELETE, (struct sockaddr *)rt_key(rt), 1109 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 0); 1110 } 1111 } 1112 1113 /* 1114 * Perform rate limit check. 1115 * Returns 0 if it is okay to send the icmp packet. 1116 * Returns 1 if the router SHOULD NOT send this icmp packet due to rate 1117 * limitation. 1118 * 1119 * XXX per-destination/type check necessary? 1120 */ 1121 static int 1122 icmp_ratelimit(dst, type, code) 1123 const struct in_addr *dst; 1124 const int type; /* not used at this moment */ 1125 const int code; /* not used at this moment */ 1126 { 1127 1128 /* PPS limit */ 1129 if (!ppsratecheck(&icmperrppslim_last, &icmperrpps_count, 1130 icmperrppslim)) { 1131 /* The packet is subject to rate limit */ 1132 return 1; 1133 } 1134 1135 /*okay to send*/ 1136 return 0; 1137 } 1138