ip_icmp.c revision 1.86
1 /* $NetBSD: ip_icmp.c,v 1.86 2004/06/25 15:43:00 itojun 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. Neither the name of the University nor the names of its contributors 85 * may be used to endorse or promote products derived from this software 86 * without specific prior written permission. 87 * 88 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 89 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 90 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 91 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 92 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 93 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 94 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 95 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 96 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 97 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 98 * SUCH DAMAGE. 99 * 100 * @(#)ip_icmp.c 8.2 (Berkeley) 1/4/94 101 */ 102 103 #include <sys/cdefs.h> 104 __KERNEL_RCSID(0, "$NetBSD: ip_icmp.c,v 1.86 2004/06/25 15:43:00 itojun Exp $"); 105 106 #include "opt_ipsec.h" 107 108 #include <sys/param.h> 109 #include <sys/systm.h> 110 #include <sys/malloc.h> 111 #include <sys/mbuf.h> 112 #include <sys/protosw.h> 113 #include <sys/socket.h> 114 #include <sys/time.h> 115 #include <sys/kernel.h> 116 #include <sys/syslog.h> 117 #include <sys/sysctl.h> 118 119 #include <net/if.h> 120 #include <net/route.h> 121 122 #include <netinet/in.h> 123 #include <netinet/in_systm.h> 124 #include <netinet/in_var.h> 125 #include <netinet/ip.h> 126 #include <netinet/ip_icmp.h> 127 #include <netinet/ip_var.h> 128 #include <netinet/in_pcb.h> 129 #include <netinet/icmp_var.h> 130 131 #ifdef IPSEC 132 #include <netinet6/ipsec.h> 133 #include <netkey/key.h> 134 #endif 135 136 #ifdef FAST_IPSEC 137 #include <netipsec/ipsec.h> 138 #include <netipsec/key.h> 139 #endif /* FAST_IPSEC*/ 140 141 #include <machine/stdarg.h> 142 143 /* 144 * ICMP routines: error generation, receive packet processing, and 145 * routines to turnaround packets back to the originator, and 146 * host table maintenance routines. 147 */ 148 149 int icmpmaskrepl = 0; 150 #ifdef ICMPPRINTFS 151 int icmpprintfs = 0; 152 #endif 153 int icmpreturndatabytes = 8; 154 155 /* 156 * List of callbacks to notify when Path MTU changes are made. 157 */ 158 struct icmp_mtudisc_callback { 159 LIST_ENTRY(icmp_mtudisc_callback) mc_list; 160 void (*mc_func) __P((struct in_addr)); 161 }; 162 163 LIST_HEAD(, icmp_mtudisc_callback) icmp_mtudisc_callbacks = 164 LIST_HEAD_INITIALIZER(&icmp_mtudisc_callbacks); 165 166 #if 0 167 static int ip_next_mtu __P((int, int)); 168 #else 169 /*static*/ int ip_next_mtu __P((int, int)); 170 #endif 171 172 extern int icmperrppslim; 173 static int icmperrpps_count = 0; 174 static struct timeval icmperrppslim_last; 175 static int icmp_rediraccept = 1; 176 static int icmp_redirtimeout = 600; 177 static struct rttimer_queue *icmp_redirect_timeout_q = NULL; 178 179 static void icmp_mtudisc_timeout __P((struct rtentry *, struct rttimer *)); 180 static void icmp_redirect_timeout __P((struct rtentry *, struct rttimer *)); 181 182 static int icmp_ratelimit __P((const struct in_addr *, const int, const int)); 183 184 185 void 186 icmp_init() 187 { 188 /* 189 * This is only useful if the user initializes redirtimeout to 190 * something other than zero. 191 */ 192 if (icmp_redirtimeout != 0) { 193 icmp_redirect_timeout_q = 194 rt_timer_queue_create(icmp_redirtimeout); 195 } 196 } 197 198 /* 199 * Register a Path MTU Discovery callback. 200 */ 201 void 202 icmp_mtudisc_callback_register(func) 203 void (*func) __P((struct in_addr)); 204 { 205 struct icmp_mtudisc_callback *mc; 206 207 for (mc = LIST_FIRST(&icmp_mtudisc_callbacks); mc != NULL; 208 mc = LIST_NEXT(mc, mc_list)) { 209 if (mc->mc_func == func) 210 return; 211 } 212 213 mc = malloc(sizeof(*mc), M_PCB, M_NOWAIT); 214 if (mc == NULL) 215 panic("icmp_mtudisc_callback_register"); 216 217 mc->mc_func = func; 218 LIST_INSERT_HEAD(&icmp_mtudisc_callbacks, mc, mc_list); 219 } 220 221 /* 222 * Generate an error packet of type error 223 * in response to bad packet ip. 224 */ 225 void 226 icmp_error(n, type, code, dest, destifp) 227 struct mbuf *n; 228 int type, code; 229 n_long dest; 230 struct ifnet *destifp; 231 { 232 struct ip *oip = mtod(n, struct ip *), *nip; 233 unsigned oiplen = oip->ip_hl << 2; 234 struct icmp *icp; 235 struct mbuf *m; 236 unsigned icmplen, mblen; 237 238 #ifdef ICMPPRINTFS 239 if (icmpprintfs) 240 printf("icmp_error(%x, %d, %d)\n", oip, type, code); 241 #endif 242 if (type != ICMP_REDIRECT) 243 icmpstat.icps_error++; 244 /* 245 * Don't send error if the original packet was encrypted. 246 * Don't send error if not the first fragment of message. 247 * Don't error if the old packet protocol was ICMP 248 * error message, only known informational types. 249 */ 250 if (n->m_flags & M_DECRYPTED) 251 goto freeit; 252 if (oip->ip_off &~ htons(IP_MF|IP_DF)) 253 goto freeit; 254 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT && 255 n->m_len >= oiplen + ICMP_MINLEN && 256 !ICMP_INFOTYPE(((struct icmp *)((caddr_t)oip + oiplen))->icmp_type)) { 257 icmpstat.icps_oldicmp++; 258 goto freeit; 259 } 260 /* Don't send error in response to a multicast or broadcast packet */ 261 if (n->m_flags & (M_BCAST|M_MCAST)) 262 goto freeit; 263 264 /* 265 * First, do a rate limitation check. 266 */ 267 if (icmp_ratelimit(&oip->ip_src, type, code)) { 268 /* XXX stat */ 269 goto freeit; 270 } 271 272 /* 273 * Now, formulate icmp message 274 */ 275 icmplen = oiplen + min(icmpreturndatabytes, 276 ntohs(oip->ip_len) - oiplen); 277 /* 278 * Defend against mbuf chains shorter than oip->ip_len - oiplen: 279 */ 280 mblen = 0; 281 for (m = n; m && (mblen < icmplen); m = m->m_next) 282 mblen += m->m_len; 283 icmplen = min(mblen, icmplen); 284 285 /* 286 * As we are not required to return everything we have, 287 * we return whatever we can return at ease. 288 * 289 * Note that ICMP datagrams longer than 576 octets are out of spec 290 * according to RFC1812; the limit on icmpreturndatabytes below in 291 * icmp_sysctl will keep things below that limit. 292 */ 293 294 KASSERT(ICMP_MINLEN <= MCLBYTES); 295 296 if (icmplen + ICMP_MINLEN > MCLBYTES) 297 icmplen = MCLBYTES - ICMP_MINLEN; 298 299 m = m_gethdr(M_DONTWAIT, MT_HEADER); 300 if (m && (icmplen + ICMP_MINLEN > MHLEN)) { 301 MCLGET(m, M_DONTWAIT); 302 if ((m->m_flags & M_EXT) == 0) { 303 m_freem(m); 304 m = NULL; 305 } 306 } 307 if (m == NULL) 308 goto freeit; 309 MCLAIM(m, n->m_owner); 310 m->m_len = icmplen + ICMP_MINLEN; 311 if ((m->m_flags & M_EXT) == 0) 312 MH_ALIGN(m, m->m_len); 313 icp = mtod(m, struct icmp *); 314 if ((u_int)type > ICMP_MAXTYPE) 315 panic("icmp_error"); 316 icmpstat.icps_outhist[type]++; 317 icp->icmp_type = type; 318 if (type == ICMP_REDIRECT) 319 icp->icmp_gwaddr.s_addr = dest; 320 else { 321 icp->icmp_void = 0; 322 /* 323 * The following assignments assume an overlay with the 324 * zeroed icmp_void field. 325 */ 326 if (type == ICMP_PARAMPROB) { 327 icp->icmp_pptr = code; 328 code = 0; 329 } else if (type == ICMP_UNREACH && 330 code == ICMP_UNREACH_NEEDFRAG && destifp) 331 icp->icmp_nextmtu = htons(destifp->if_mtu); 332 } 333 334 icp->icmp_code = code; 335 m_copydata(n, 0, icmplen, (caddr_t)&icp->icmp_ip); 336 nip = &icp->icmp_ip; 337 338 /* 339 * Now, copy old ip header (without options) 340 * in front of icmp message. 341 */ 342 if (m->m_data - sizeof(struct ip) < m->m_pktdat) 343 panic("icmp len"); 344 m->m_data -= sizeof(struct ip); 345 m->m_len += sizeof(struct ip); 346 m->m_pkthdr.len = m->m_len; 347 m->m_pkthdr.rcvif = n->m_pkthdr.rcvif; 348 nip = mtod(m, struct ip *); 349 /* ip_v set in ip_output */ 350 nip->ip_hl = sizeof(struct ip) >> 2; 351 nip->ip_tos = 0; 352 nip->ip_len = htons(m->m_len); 353 /* ip_id set in ip_output */ 354 nip->ip_off = htons(0); 355 /* ip_ttl set in icmp_reflect */ 356 nip->ip_p = IPPROTO_ICMP; 357 nip->ip_src = oip->ip_src; 358 nip->ip_dst = oip->ip_dst; 359 icmp_reflect(m); 360 361 freeit: 362 m_freem(n); 363 } 364 365 static struct sockaddr_in icmpsrc = { sizeof (struct sockaddr_in), AF_INET }; 366 static struct sockaddr_in icmpdst = { sizeof (struct sockaddr_in), AF_INET }; 367 static struct sockaddr_in icmpgw = { sizeof (struct sockaddr_in), AF_INET }; 368 struct sockaddr_in icmpmask = { 8, 0 }; 369 370 /* 371 * Process a received ICMP message. 372 */ 373 void 374 icmp_input(struct mbuf *m, ...) 375 { 376 int proto; 377 struct icmp *icp; 378 struct ip *ip = mtod(m, struct ip *); 379 int icmplen; 380 int i; 381 struct in_ifaddr *ia; 382 void *(*ctlfunc) __P((int, struct sockaddr *, void *)); 383 int code; 384 int hlen; 385 va_list ap; 386 struct rtentry *rt; 387 388 va_start(ap, m); 389 hlen = va_arg(ap, int); 390 proto = va_arg(ap, int); 391 va_end(ap); 392 393 /* 394 * Locate icmp structure in mbuf, and check 395 * that not corrupted and of at least minimum length. 396 */ 397 icmplen = ntohs(ip->ip_len) - hlen; 398 #ifdef ICMPPRINTFS 399 if (icmpprintfs) 400 printf("icmp_input from %x to %x, len %d\n", 401 ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr), 402 icmplen); 403 #endif 404 if (icmplen < ICMP_MINLEN) { 405 icmpstat.icps_tooshort++; 406 goto freeit; 407 } 408 i = hlen + min(icmplen, ICMP_ADVLENMIN); 409 if (m->m_len < i && (m = m_pullup(m, i)) == 0) { 410 icmpstat.icps_tooshort++; 411 return; 412 } 413 ip = mtod(m, struct ip *); 414 m->m_len -= hlen; 415 m->m_data += hlen; 416 icp = mtod(m, struct icmp *); 417 /* Don't need to assert alignment, here. */ 418 if (in_cksum(m, icmplen)) { 419 icmpstat.icps_checksum++; 420 goto freeit; 421 } 422 m->m_len += hlen; 423 m->m_data -= hlen; 424 425 #ifdef ICMPPRINTFS 426 /* 427 * Message type specific processing. 428 */ 429 if (icmpprintfs) 430 printf("icmp_input, type %d code %d\n", icp->icmp_type, 431 icp->icmp_code); 432 #endif 433 if (icp->icmp_type > ICMP_MAXTYPE) 434 goto raw; 435 icmpstat.icps_inhist[icp->icmp_type]++; 436 code = icp->icmp_code; 437 switch (icp->icmp_type) { 438 439 case ICMP_UNREACH: 440 switch (code) { 441 case ICMP_UNREACH_NET: 442 case ICMP_UNREACH_HOST: 443 case ICMP_UNREACH_PROTOCOL: 444 case ICMP_UNREACH_PORT: 445 case ICMP_UNREACH_SRCFAIL: 446 code += PRC_UNREACH_NET; 447 break; 448 449 case ICMP_UNREACH_NEEDFRAG: 450 code = PRC_MSGSIZE; 451 break; 452 453 case ICMP_UNREACH_NET_UNKNOWN: 454 case ICMP_UNREACH_NET_PROHIB: 455 case ICMP_UNREACH_TOSNET: 456 code = PRC_UNREACH_NET; 457 break; 458 459 case ICMP_UNREACH_HOST_UNKNOWN: 460 case ICMP_UNREACH_ISOLATED: 461 case ICMP_UNREACH_HOST_PROHIB: 462 case ICMP_UNREACH_TOSHOST: 463 code = PRC_UNREACH_HOST; 464 break; 465 466 default: 467 goto badcode; 468 } 469 goto deliver; 470 471 case ICMP_TIMXCEED: 472 if (code > 1) 473 goto badcode; 474 code += PRC_TIMXCEED_INTRANS; 475 goto deliver; 476 477 case ICMP_PARAMPROB: 478 if (code > 1) 479 goto badcode; 480 code = PRC_PARAMPROB; 481 goto deliver; 482 483 case ICMP_SOURCEQUENCH: 484 if (code) 485 goto badcode; 486 code = PRC_QUENCH; 487 goto deliver; 488 489 deliver: 490 /* 491 * Problem with datagram; advise higher level routines. 492 */ 493 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || 494 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) { 495 icmpstat.icps_badlen++; 496 goto freeit; 497 } 498 if (IN_MULTICAST(icp->icmp_ip.ip_dst.s_addr)) 499 goto badcode; 500 #ifdef ICMPPRINTFS 501 if (icmpprintfs) 502 printf("deliver to protocol %d\n", icp->icmp_ip.ip_p); 503 #endif 504 icmpsrc.sin_addr = icp->icmp_ip.ip_dst; 505 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput; 506 if (ctlfunc) 507 (void) (*ctlfunc)(code, sintosa(&icmpsrc), 508 &icp->icmp_ip); 509 break; 510 511 badcode: 512 icmpstat.icps_badcode++; 513 break; 514 515 case ICMP_ECHO: 516 icp->icmp_type = ICMP_ECHOREPLY; 517 goto reflect; 518 519 case ICMP_TSTAMP: 520 if (icmplen < ICMP_TSLEN) { 521 icmpstat.icps_badlen++; 522 break; 523 } 524 icp->icmp_type = ICMP_TSTAMPREPLY; 525 icp->icmp_rtime = iptime(); 526 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */ 527 goto reflect; 528 529 case ICMP_MASKREQ: 530 if (icmpmaskrepl == 0) 531 break; 532 /* 533 * We are not able to respond with all ones broadcast 534 * unless we receive it over a point-to-point interface. 535 */ 536 if (icmplen < ICMP_MASKLEN) { 537 icmpstat.icps_badlen++; 538 break; 539 } 540 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 541 in_nullhost(ip->ip_dst)) 542 icmpdst.sin_addr = ip->ip_src; 543 else 544 icmpdst.sin_addr = ip->ip_dst; 545 ia = ifatoia(ifaof_ifpforaddr(sintosa(&icmpdst), 546 m->m_pkthdr.rcvif)); 547 if (ia == 0) 548 break; 549 icp->icmp_type = ICMP_MASKREPLY; 550 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr; 551 if (in_nullhost(ip->ip_src)) { 552 if (ia->ia_ifp->if_flags & IFF_BROADCAST) 553 ip->ip_src = ia->ia_broadaddr.sin_addr; 554 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT) 555 ip->ip_src = ia->ia_dstaddr.sin_addr; 556 } 557 reflect: 558 icmpstat.icps_reflect++; 559 icmpstat.icps_outhist[icp->icmp_type]++; 560 icmp_reflect(m); 561 return; 562 563 case ICMP_REDIRECT: 564 if (code > 3) 565 goto badcode; 566 if (icmp_rediraccept == 0) 567 goto freeit; 568 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || 569 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) { 570 icmpstat.icps_badlen++; 571 break; 572 } 573 /* 574 * Short circuit routing redirects to force 575 * immediate change in the kernel's routing 576 * tables. The message is also handed to anyone 577 * listening on a raw socket (e.g. the routing 578 * daemon for use in updating its tables). 579 */ 580 icmpgw.sin_addr = ip->ip_src; 581 icmpdst.sin_addr = icp->icmp_gwaddr; 582 #ifdef ICMPPRINTFS 583 if (icmpprintfs) 584 printf("redirect dst %x to %x\n", icp->icmp_ip.ip_dst, 585 icp->icmp_gwaddr); 586 #endif 587 icmpsrc.sin_addr = icp->icmp_ip.ip_dst; 588 rt = NULL; 589 rtredirect(sintosa(&icmpsrc), sintosa(&icmpdst), 590 (struct sockaddr *)0, RTF_GATEWAY | RTF_HOST, 591 sintosa(&icmpgw), (struct rtentry **)&rt); 592 if (rt != NULL && icmp_redirtimeout != 0) { 593 i = rt_timer_add(rt, icmp_redirect_timeout, 594 icmp_redirect_timeout_q); 595 if (i) 596 log(LOG_ERR, "ICMP: redirect failed to " 597 "register timeout for route to %x, " 598 "code %d\n", 599 icp->icmp_ip.ip_dst.s_addr, i); 600 } 601 if (rt != NULL) 602 rtfree(rt); 603 604 pfctlinput(PRC_REDIRECT_HOST, sintosa(&icmpsrc)); 605 #if defined(IPSEC) || defined(FAST_IPSEC) 606 key_sa_routechange((struct sockaddr *)&icmpsrc); 607 #endif 608 break; 609 610 /* 611 * No kernel processing for the following; 612 * just fall through to send to raw listener. 613 */ 614 case ICMP_ECHOREPLY: 615 case ICMP_ROUTERADVERT: 616 case ICMP_ROUTERSOLICIT: 617 case ICMP_TSTAMPREPLY: 618 case ICMP_IREQREPLY: 619 case ICMP_MASKREPLY: 620 default: 621 break; 622 } 623 624 raw: 625 rip_input(m, hlen, proto); 626 return; 627 628 freeit: 629 m_freem(m); 630 return; 631 } 632 633 /* 634 * Reflect the ip packet back to the source 635 */ 636 void 637 icmp_reflect(m) 638 struct mbuf *m; 639 { 640 struct ip *ip = mtod(m, struct ip *); 641 struct in_ifaddr *ia; 642 struct ifaddr *ifa; 643 struct sockaddr_in *sin = 0; 644 struct in_addr t; 645 struct mbuf *opts = 0; 646 int optlen = (ip->ip_hl << 2) - sizeof(struct ip); 647 648 if (!in_canforward(ip->ip_src) && 649 ((ip->ip_src.s_addr & IN_CLASSA_NET) != 650 htonl(IN_LOOPBACKNET << IN_CLASSA_NSHIFT))) { 651 m_freem(m); /* Bad return address */ 652 goto done; /* ip_output() will check for broadcast */ 653 } 654 t = ip->ip_dst; 655 ip->ip_dst = ip->ip_src; 656 /* 657 * If the incoming packet was addressed directly to us, use 658 * dst as the src for the reply. Otherwise (broadcast or 659 * anonymous), use an address which corresponds to the 660 * incoming interface, with a preference for the address which 661 * corresponds to the route to the destination of the ICMP. 662 */ 663 664 /* Look for packet addressed to us */ 665 INADDR_TO_IA(t, ia); 666 667 /* look for packet sent to broadcast address */ 668 if (ia == NULL && m->m_pkthdr.rcvif && 669 (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST)) { 670 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) { 671 if (ifa->ifa_addr->sa_family != AF_INET) 672 continue; 673 if (in_hosteq(t,ifatoia(ifa)->ia_broadaddr.sin_addr)) { 674 ia = ifatoia(ifa); 675 break; 676 } 677 } 678 } 679 680 if (ia) 681 sin = &ia->ia_addr; 682 683 icmpdst.sin_addr = t; 684 685 /* 686 * if the packet is addressed somewhere else, compute the 687 * source address for packets routed back to the source, and 688 * use that, if it's an address on the interface which 689 * received the packet 690 */ 691 if (sin == (struct sockaddr_in *)0 && m->m_pkthdr.rcvif) { 692 struct sockaddr_in sin_dst; 693 struct route icmproute; 694 int errornum; 695 696 sin_dst.sin_family = AF_INET; 697 sin_dst.sin_len = sizeof(struct sockaddr_in); 698 sin_dst.sin_addr = ip->ip_dst; 699 bzero(&icmproute, sizeof(icmproute)); 700 errornum = 0; 701 sin = in_selectsrc(&sin_dst, &icmproute, 0, NULL, &errornum); 702 /* errornum is never used */ 703 if (icmproute.ro_rt) 704 RTFREE(icmproute.ro_rt); 705 /* check to make sure sin is a source address on rcvif */ 706 if (sin) { 707 t = sin->sin_addr; 708 sin = (struct sockaddr_in *)0; 709 INADDR_TO_IA(t, ia); 710 while (ia) { 711 if (ia->ia_ifp == m->m_pkthdr.rcvif) { 712 sin = &ia->ia_addr; 713 break; 714 } 715 NEXT_IA_WITH_SAME_ADDR(ia); 716 } 717 } 718 } 719 720 /* 721 * if it was not addressed to us, but the route doesn't go out 722 * the source interface, pick an address on the source 723 * interface. This can happen when routing is asymmetric, or 724 * when the incoming packet was encapsulated 725 */ 726 if (sin == (struct sockaddr_in *)0 && m->m_pkthdr.rcvif) { 727 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) { 728 if (ifa->ifa_addr->sa_family != AF_INET) 729 continue; 730 sin = &(ifatoia(ifa)->ia_addr); 731 break; 732 } 733 } 734 735 /* 736 * The following happens if the packet was not addressed to us, 737 * and was received on an interface with no IP address: 738 * We find the first AF_INET address on the first non-loopback 739 * interface. 740 */ 741 if (sin == (struct sockaddr_in *)0) 742 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_list) { 743 if (ia->ia_ifp->if_flags & IFF_LOOPBACK) 744 continue; 745 sin = &ia->ia_addr; 746 break; 747 } 748 749 /* 750 * If we still didn't find an address, punt. We could have an 751 * interface up (and receiving packets) with no address. 752 */ 753 if (sin == (struct sockaddr_in *)0) { 754 m_freem(m); 755 goto done; 756 } 757 758 ip->ip_src = sin->sin_addr; 759 ip->ip_ttl = MAXTTL; 760 761 if (optlen > 0) { 762 u_char *cp; 763 int opt, cnt; 764 u_int len; 765 766 /* 767 * Retrieve any source routing from the incoming packet; 768 * add on any record-route or timestamp options. 769 */ 770 cp = (u_char *) (ip + 1); 771 if ((opts = ip_srcroute()) == 0 && 772 (opts = m_gethdr(M_DONTWAIT, MT_HEADER))) { 773 MCLAIM(opts, m->m_owner); 774 opts->m_len = sizeof(struct in_addr); 775 *mtod(opts, struct in_addr *) = zeroin_addr; 776 } 777 if (opts) { 778 #ifdef ICMPPRINTFS 779 if (icmpprintfs) 780 printf("icmp_reflect optlen %d rt %d => ", 781 optlen, opts->m_len); 782 #endif 783 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) { 784 opt = cp[IPOPT_OPTVAL]; 785 if (opt == IPOPT_EOL) 786 break; 787 if (opt == IPOPT_NOP) 788 len = 1; 789 else { 790 if (cnt < IPOPT_OLEN + sizeof(*cp)) 791 break; 792 len = cp[IPOPT_OLEN]; 793 if (len < IPOPT_OLEN + sizeof(*cp) || 794 len > cnt) 795 break; 796 } 797 /* 798 * Should check for overflow, but it "can't happen" 799 */ 800 if (opt == IPOPT_RR || opt == IPOPT_TS || 801 opt == IPOPT_SECURITY) { 802 bcopy((caddr_t)cp, 803 mtod(opts, caddr_t) + opts->m_len, len); 804 opts->m_len += len; 805 } 806 } 807 /* Terminate & pad, if necessary */ 808 if ((cnt = opts->m_len % 4) != 0) { 809 for (; cnt < 4; cnt++) { 810 *(mtod(opts, caddr_t) + opts->m_len) = 811 IPOPT_EOL; 812 opts->m_len++; 813 } 814 } 815 #ifdef ICMPPRINTFS 816 if (icmpprintfs) 817 printf("%d\n", opts->m_len); 818 #endif 819 } 820 /* 821 * Now strip out original options by copying rest of first 822 * mbuf's data back, and adjust the IP length. 823 */ 824 ip->ip_len = htons(ntohs(ip->ip_len) - optlen); 825 ip->ip_hl = sizeof(struct ip) >> 2; 826 m->m_len -= optlen; 827 if (m->m_flags & M_PKTHDR) 828 m->m_pkthdr.len -= optlen; 829 optlen += sizeof(struct ip); 830 bcopy((caddr_t)ip + optlen, (caddr_t)(ip + 1), 831 (unsigned)(m->m_len - sizeof(struct ip))); 832 } 833 m_tag_delete_nonpersistent(m); 834 m->m_flags &= ~(M_BCAST|M_MCAST); 835 836 /* 837 * Clear any in-bound checksum flags for this packet. 838 */ 839 if (m->m_flags & M_PKTHDR) 840 m->m_pkthdr.csum_flags = 0; 841 842 icmp_send(m, opts); 843 done: 844 if (opts) 845 (void)m_free(opts); 846 } 847 848 /* 849 * Send an icmp packet back to the ip level, 850 * after supplying a checksum. 851 */ 852 void 853 icmp_send(m, opts) 854 struct mbuf *m; 855 struct mbuf *opts; 856 { 857 struct ip *ip = mtod(m, struct ip *); 858 int hlen; 859 struct icmp *icp; 860 861 hlen = ip->ip_hl << 2; 862 m->m_data += hlen; 863 m->m_len -= hlen; 864 icp = mtod(m, struct icmp *); 865 icp->icmp_cksum = 0; 866 icp->icmp_cksum = in_cksum(m, ntohs(ip->ip_len) - hlen); 867 m->m_data -= hlen; 868 m->m_len += hlen; 869 #ifdef ICMPPRINTFS 870 if (icmpprintfs) 871 printf("icmp_send dst %x src %x\n", ip->ip_dst, ip->ip_src); 872 #endif 873 (void) ip_output(m, opts, NULL, 0, 874 (struct ip_moptions *)NULL, (struct socket *)NULL); 875 } 876 877 n_time 878 iptime() 879 { 880 struct timeval atv; 881 u_long t; 882 883 microtime(&atv); 884 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000; 885 return (htonl(t)); 886 } 887 888 /* 889 * sysctl helper routine for net.inet.icmp.returndatabytes. ensures 890 * that the new value is in the correct range. 891 */ 892 static int 893 sysctl_net_inet_icmp_returndatabytes(SYSCTLFN_ARGS) 894 { 895 int error, t; 896 struct sysctlnode node; 897 898 node = *rnode; 899 node.sysctl_data = &t; 900 t = icmpreturndatabytes; 901 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 902 if (error || newp == NULL) 903 return (error); 904 905 if (t < 8 || t > 512) 906 return (EINVAL); 907 icmpreturndatabytes = t; 908 909 return (0); 910 } 911 912 /* 913 * sysctl helper routine for net.inet.icmp.redirtimeout. ensures that 914 * the given value is not less than zero and then resets the timeout 915 * queue. 916 */ 917 static int 918 sysctl_net_inet_icmp_redirtimeout(SYSCTLFN_ARGS) 919 { 920 int error, tmp; 921 struct sysctlnode node; 922 923 node = *rnode; 924 node.sysctl_data = &tmp; 925 tmp = icmp_redirtimeout; 926 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 927 if (error || newp == NULL) 928 return (error); 929 if (tmp < 0) 930 return (EINVAL); 931 icmp_redirtimeout = tmp; 932 933 /* 934 * was it a *defined* side-effect that anyone even *reading* 935 * this value causes these things to happen? 936 */ 937 if (icmp_redirect_timeout_q != NULL) { 938 if (icmp_redirtimeout == 0) { 939 rt_timer_queue_destroy(icmp_redirect_timeout_q, 940 TRUE); 941 icmp_redirect_timeout_q = NULL; 942 } else { 943 rt_timer_queue_change(icmp_redirect_timeout_q, 944 icmp_redirtimeout); 945 } 946 } else if (icmp_redirtimeout > 0) { 947 icmp_redirect_timeout_q = 948 rt_timer_queue_create(icmp_redirtimeout); 949 } 950 951 return (0); 952 } 953 954 SYSCTL_SETUP(sysctl_net_inet_icmp_setup, "sysctl net.inet.icmp subtree setup") 955 { 956 957 sysctl_createv(clog, 0, NULL, NULL, 958 CTLFLAG_PERMANENT, 959 CTLTYPE_NODE, "net", NULL, 960 NULL, 0, NULL, 0, 961 CTL_NET, CTL_EOL); 962 sysctl_createv(clog, 0, NULL, NULL, 963 CTLFLAG_PERMANENT, 964 CTLTYPE_NODE, "inet", NULL, 965 NULL, 0, NULL, 0, 966 CTL_NET, PF_INET, CTL_EOL); 967 sysctl_createv(clog, 0, NULL, NULL, 968 CTLFLAG_PERMANENT, 969 CTLTYPE_NODE, "icmp", 970 SYSCTL_DESCR("ICMPv4 related settings"), 971 NULL, 0, NULL, 0, 972 CTL_NET, PF_INET, IPPROTO_ICMP, CTL_EOL); 973 974 sysctl_createv(clog, 0, NULL, NULL, 975 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 976 CTLTYPE_INT, "maskrepl", 977 SYSCTL_DESCR("Respond to ICMP_MASKREQ messages"), 978 NULL, 0, &icmpmaskrepl, 0, 979 CTL_NET, PF_INET, IPPROTO_ICMP, 980 ICMPCTL_MASKREPL, CTL_EOL); 981 sysctl_createv(clog, 0, NULL, NULL, 982 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 983 CTLTYPE_INT, "returndatabytes", 984 SYSCTL_DESCR("Number of bytes to return in an ICMP " 985 "error message"), 986 sysctl_net_inet_icmp_returndatabytes, 0, 987 &icmpreturndatabytes, 0, 988 CTL_NET, PF_INET, IPPROTO_ICMP, 989 ICMPCTL_RETURNDATABYTES, CTL_EOL); 990 sysctl_createv(clog, 0, NULL, NULL, 991 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 992 CTLTYPE_INT, "errppslimit", 993 SYSCTL_DESCR("Maximum number of outgoing ICMP error " 994 "messages per second"), 995 NULL, 0, &icmperrppslim, 0, 996 CTL_NET, PF_INET, IPPROTO_ICMP, 997 ICMPCTL_ERRPPSLIMIT, CTL_EOL); 998 sysctl_createv(clog, 0, NULL, NULL, 999 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1000 CTLTYPE_INT, "rediraccept", 1001 SYSCTL_DESCR("Accept ICMP_REDIRECT messages"), 1002 NULL, 0, &icmp_rediraccept, 0, 1003 CTL_NET, PF_INET, IPPROTO_ICMP, 1004 ICMPCTL_REDIRACCEPT, CTL_EOL); 1005 sysctl_createv(clog, 0, NULL, NULL, 1006 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1007 CTLTYPE_INT, "redirtimeout", 1008 SYSCTL_DESCR("Lifetime of ICMP_REDIRECT generated " 1009 "routes"), 1010 sysctl_net_inet_icmp_redirtimeout, 0, 1011 &icmp_redirtimeout, 0, 1012 CTL_NET, PF_INET, IPPROTO_ICMP, 1013 ICMPCTL_REDIRTIMEOUT, CTL_EOL); 1014 } 1015 1016 /* Table of common MTUs: */ 1017 1018 static const u_int mtu_table[] = { 1019 65535, 65280, 32000, 17914, 9180, 8166, 1020 4352, 2002, 1492, 1006, 508, 296, 68, 0 1021 }; 1022 1023 void 1024 icmp_mtudisc(icp, faddr) 1025 struct icmp *icp; 1026 struct in_addr faddr; 1027 { 1028 struct icmp_mtudisc_callback *mc; 1029 struct sockaddr *dst = sintosa(&icmpsrc); 1030 struct rtentry *rt; 1031 u_long mtu = ntohs(icp->icmp_nextmtu); /* Why a long? IPv6 */ 1032 int error; 1033 1034 rt = rtalloc1(dst, 1); 1035 if (rt == 0) 1036 return; 1037 1038 /* If we didn't get a host route, allocate one */ 1039 1040 if ((rt->rt_flags & RTF_HOST) == 0) { 1041 struct rtentry *nrt; 1042 1043 error = rtrequest((int) RTM_ADD, dst, 1044 (struct sockaddr *) rt->rt_gateway, 1045 (struct sockaddr *) 0, 1046 RTF_GATEWAY | RTF_HOST | RTF_DYNAMIC, &nrt); 1047 if (error) { 1048 rtfree(rt); 1049 return; 1050 } 1051 nrt->rt_rmx = rt->rt_rmx; 1052 rtfree(rt); 1053 rt = nrt; 1054 } 1055 error = rt_timer_add(rt, icmp_mtudisc_timeout, ip_mtudisc_timeout_q); 1056 if (error) { 1057 rtfree(rt); 1058 return; 1059 } 1060 1061 if (mtu == 0) { 1062 int i = 0; 1063 1064 mtu = ntohs(icp->icmp_ip.ip_len); 1065 /* Some 4.2BSD-based routers incorrectly adjust the ip_len */ 1066 if (mtu > rt->rt_rmx.rmx_mtu && rt->rt_rmx.rmx_mtu != 0) 1067 mtu -= (icp->icmp_ip.ip_hl << 2); 1068 1069 /* If we still can't guess a value, try the route */ 1070 1071 if (mtu == 0) { 1072 mtu = rt->rt_rmx.rmx_mtu; 1073 1074 /* If no route mtu, default to the interface mtu */ 1075 1076 if (mtu == 0) 1077 mtu = rt->rt_ifp->if_mtu; 1078 } 1079 1080 for (i = 0; i < sizeof(mtu_table) / sizeof(mtu_table[0]); i++) 1081 if (mtu > mtu_table[i]) { 1082 mtu = mtu_table[i]; 1083 break; 1084 } 1085 } 1086 1087 /* 1088 * XXX: RTV_MTU is overloaded, since the admin can set it 1089 * to turn off PMTU for a route, and the kernel can 1090 * set it to indicate a serious problem with PMTU 1091 * on a route. We should be using a separate flag 1092 * for the kernel to indicate this. 1093 */ 1094 1095 if ((rt->rt_rmx.rmx_locks & RTV_MTU) == 0) { 1096 if (mtu < 296 || mtu > rt->rt_ifp->if_mtu) 1097 rt->rt_rmx.rmx_locks |= RTV_MTU; 1098 else if (rt->rt_rmx.rmx_mtu > mtu || 1099 rt->rt_rmx.rmx_mtu == 0) { 1100 icmpstat.icps_pmtuchg++; 1101 rt->rt_rmx.rmx_mtu = mtu; 1102 } 1103 } 1104 1105 if (rt) 1106 rtfree(rt); 1107 1108 /* 1109 * Notify protocols that the MTU for this destination 1110 * has changed. 1111 */ 1112 for (mc = LIST_FIRST(&icmp_mtudisc_callbacks); mc != NULL; 1113 mc = LIST_NEXT(mc, mc_list)) 1114 (*mc->mc_func)(faddr); 1115 } 1116 1117 /* 1118 * Return the next larger or smaller MTU plateau (table from RFC 1191) 1119 * given current value MTU. If DIR is less than zero, a larger plateau 1120 * is returned; otherwise, a smaller value is returned. 1121 */ 1122 int 1123 ip_next_mtu(mtu, dir) /* XXX */ 1124 int mtu; 1125 int dir; 1126 { 1127 int i; 1128 1129 for (i = 0; i < (sizeof mtu_table) / (sizeof mtu_table[0]); i++) { 1130 if (mtu >= mtu_table[i]) 1131 break; 1132 } 1133 1134 if (dir < 0) { 1135 if (i == 0) { 1136 return 0; 1137 } else { 1138 return mtu_table[i - 1]; 1139 } 1140 } else { 1141 if (mtu_table[i] == 0) { 1142 return 0; 1143 } else if (mtu > mtu_table[i]) { 1144 return mtu_table[i]; 1145 } else { 1146 return mtu_table[i + 1]; 1147 } 1148 } 1149 } 1150 1151 static void 1152 icmp_mtudisc_timeout(rt, r) 1153 struct rtentry *rt; 1154 struct rttimer *r; 1155 { 1156 if (rt == NULL) 1157 panic("icmp_mtudisc_timeout: bad route to timeout"); 1158 if ((rt->rt_flags & (RTF_DYNAMIC | RTF_HOST)) == 1159 (RTF_DYNAMIC | RTF_HOST)) { 1160 rtrequest((int) RTM_DELETE, (struct sockaddr *)rt_key(rt), 1161 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 0); 1162 } else { 1163 if ((rt->rt_rmx.rmx_locks & RTV_MTU) == 0) { 1164 rt->rt_rmx.rmx_mtu = 0; 1165 } 1166 } 1167 } 1168 1169 static void 1170 icmp_redirect_timeout(rt, r) 1171 struct rtentry *rt; 1172 struct rttimer *r; 1173 { 1174 if (rt == NULL) 1175 panic("icmp_redirect_timeout: bad route to timeout"); 1176 if ((rt->rt_flags & (RTF_DYNAMIC | RTF_HOST)) == 1177 (RTF_DYNAMIC | RTF_HOST)) { 1178 rtrequest((int) RTM_DELETE, (struct sockaddr *)rt_key(rt), 1179 rt->rt_gateway, rt_mask(rt), rt->rt_flags, 0); 1180 } 1181 } 1182 1183 /* 1184 * Perform rate limit check. 1185 * Returns 0 if it is okay to send the icmp packet. 1186 * Returns 1 if the router SHOULD NOT send this icmp packet due to rate 1187 * limitation. 1188 * 1189 * XXX per-destination/type check necessary? 1190 */ 1191 static int 1192 icmp_ratelimit(dst, type, code) 1193 const struct in_addr *dst; 1194 const int type; /* not used at this moment */ 1195 const int code; /* not used at this moment */ 1196 { 1197 1198 /* PPS limit */ 1199 if (!ppsratecheck(&icmperrppslim_last, &icmperrpps_count, 1200 icmperrppslim)) { 1201 /* The packet is subject to rate limit */ 1202 return 1; 1203 } 1204 1205 /*okay to send*/ 1206 return 0; 1207 } 1208
Indexes created Wed Sep 24 14:09:57 GMT 2025