ip_icmp.c revision 1.134.4.8
1 /* $NetBSD: ip_icmp.c,v 1.134.4.8 2017/02/05 13:40:59 skrll 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 * 52 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 53 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 54 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 55 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 56 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 57 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 58 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 59 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 60 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 61 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 62 * POSSIBILITY OF SUCH DAMAGE. 63 */ 64 65 /* 66 * Copyright (c) 1982, 1986, 1988, 1993 67 * The Regents of the University of California. All rights reserved. 68 * 69 * Redistribution and use in source and binary forms, with or without 70 * modification, are permitted provided that the following conditions 71 * are met: 72 * 1. Redistributions of source code must retain the above copyright 73 * notice, this list of conditions and the following disclaimer. 74 * 2. Redistributions in binary form must reproduce the above copyright 75 * notice, this list of conditions and the following disclaimer in the 76 * documentation and/or other materials provided with the distribution. 77 * 3. Neither the name of the University nor the names of its contributors 78 * may be used to endorse or promote products derived from this software 79 * without specific prior written permission. 80 * 81 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 82 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 83 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 84 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 85 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 86 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 87 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 88 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 89 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 90 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 91 * SUCH DAMAGE. 92 * 93 * @(#)ip_icmp.c 8.2 (Berkeley) 1/4/94 94 */ 95 96 #include <sys/cdefs.h> 97 __KERNEL_RCSID(0, "$NetBSD: ip_icmp.c,v 1.134.4.8 2017/02/05 13:40:59 skrll Exp $"); 98 99 #ifdef _KERNEL_OPT 100 #include "opt_ipsec.h" 101 #endif 102 103 #include <sys/param.h> 104 #include <sys/systm.h> 105 #include <sys/mbuf.h> 106 #include <sys/protosw.h> 107 #include <sys/socket.h> 108 #include <sys/socketvar.h> /* For softnet_lock */ 109 #include <sys/kmem.h> 110 #include <sys/time.h> 111 #include <sys/kernel.h> 112 #include <sys/syslog.h> 113 #include <sys/sysctl.h> 114 115 #include <net/if.h> 116 #include <net/route.h> 117 118 #include <netinet/in.h> 119 #include <netinet/in_systm.h> 120 #include <netinet/in_var.h> 121 #include <netinet/ip.h> 122 #include <netinet/ip_icmp.h> 123 #include <netinet/ip_var.h> 124 #include <netinet/in_pcb.h> 125 #include <netinet/in_proto.h> 126 #include <netinet/icmp_var.h> 127 #include <netinet/icmp_private.h> 128 #include <netinet/wqinput.h> 129 130 #ifdef IPSEC 131 #include <netipsec/ipsec.h> 132 #include <netipsec/key.h> 133 #endif /* IPSEC*/ 134 135 /* 136 * ICMP routines: error generation, receive packet processing, and 137 * routines to turnaround packets back to the originator, and 138 * host table maintenance routines. 139 */ 140 141 int icmpmaskrepl = 0; 142 int icmpbmcastecho = 0; 143 #ifdef ICMPPRINTFS 144 int icmpprintfs = 0; 145 #endif 146 int icmpreturndatabytes = 8; 147 148 percpu_t *icmpstat_percpu; 149 150 /* 151 * List of callbacks to notify when Path MTU changes are made. 152 */ 153 struct icmp_mtudisc_callback { 154 LIST_ENTRY(icmp_mtudisc_callback) mc_list; 155 void (*mc_func)(struct in_addr); 156 }; 157 158 LIST_HEAD(, icmp_mtudisc_callback) icmp_mtudisc_callbacks = 159 LIST_HEAD_INITIALIZER(&icmp_mtudisc_callbacks); 160 161 #if 0 162 static u_int ip_next_mtu(u_int, int); 163 #else 164 /*static*/ u_int ip_next_mtu(u_int, int); 165 #endif 166 167 extern int icmperrppslim; 168 static int icmperrpps_count = 0; 169 static struct timeval icmperrppslim_last; 170 static int icmp_rediraccept = 1; 171 static int icmp_redirtimeout = 600; 172 static struct rttimer_queue *icmp_redirect_timeout_q = NULL; 173 174 static void icmp_mtudisc_timeout(struct rtentry *, struct rttimer *); 175 static void icmp_redirect_timeout(struct rtentry *, struct rttimer *); 176 177 static void sysctl_netinet_icmp_setup(struct sysctllog **); 178 179 /* workqueue-based pr_input */ 180 static struct wqinput *icmp_wqinput; 181 static void _icmp_input(struct mbuf *, int, int); 182 183 void 184 icmp_init(void) 185 { 186 187 sysctl_netinet_icmp_setup(NULL); 188 189 /* 190 * This is only useful if the user initializes redirtimeout to 191 * something other than zero. 192 */ 193 if (icmp_redirtimeout != 0) { 194 icmp_redirect_timeout_q = 195 rt_timer_queue_create(icmp_redirtimeout); 196 } 197 198 icmpstat_percpu = percpu_alloc(sizeof(uint64_t) * ICMP_NSTATS); 199 icmp_wqinput = wqinput_create("icmp", _icmp_input); 200 } 201 202 /* 203 * Register a Path MTU Discovery callback. 204 */ 205 void 206 icmp_mtudisc_callback_register(void (*func)(struct in_addr)) 207 { 208 struct icmp_mtudisc_callback *mc; 209 210 for (mc = LIST_FIRST(&icmp_mtudisc_callbacks); mc != NULL; 211 mc = LIST_NEXT(mc, mc_list)) { 212 if (mc->mc_func == func) 213 return; 214 } 215 216 mc = kmem_alloc(sizeof(*mc), KM_SLEEP); 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(struct mbuf *n, int type, int code, n_long dest, 227 int destmtu) 228 { 229 struct ip *oip = mtod(n, struct ip *), *nip; 230 unsigned oiplen = oip->ip_hl << 2; 231 struct icmp *icp; 232 struct mbuf *m; 233 struct m_tag *mtag; 234 unsigned icmplen, mblen; 235 236 #ifdef ICMPPRINTFS 237 if (icmpprintfs) 238 printf("icmp_error(%p, type:%d, code:%d)\n", oip, type, code); 239 #endif 240 if (type != ICMP_REDIRECT) 241 ICMP_STATINC(ICMP_STAT_ERROR); 242 /* 243 * Don't send error if the original packet was encrypted. 244 * Don't send error if not the first fragment of message. 245 * Don't error if the old packet protocol was ICMP 246 * error message, only known informational types. 247 */ 248 if (n->m_flags & M_DECRYPTED) 249 goto freeit; 250 if (oip->ip_off &~ htons(IP_MF|IP_DF)) 251 goto freeit; 252 if (oip->ip_p == IPPROTO_ICMP && type != ICMP_REDIRECT && 253 n->m_len >= oiplen + ICMP_MINLEN && 254 !ICMP_INFOTYPE(((struct icmp *)((char *)oip + oiplen))->icmp_type)) { 255 ICMP_STATINC(ICMP_STAT_OLDICMP); 256 goto freeit; 257 } 258 /* Don't send error in response to a multicast or broadcast packet */ 259 if (n->m_flags & (M_BCAST|M_MCAST)) 260 goto freeit; 261 262 /* 263 * First, do a rate limitation check. 264 */ 265 if (icmp_ratelimit(&oip->ip_src, type, code)) { 266 /* XXX stat */ 267 goto freeit; 268 } 269 270 /* 271 * Now, formulate icmp message 272 */ 273 icmplen = oiplen + min(icmpreturndatabytes, 274 ntohs(oip->ip_len) - oiplen); 275 /* 276 * Defend against mbuf chains shorter than oip->ip_len - oiplen: 277 */ 278 mblen = 0; 279 for (m = n; m && (mblen < icmplen); m = m->m_next) 280 mblen += m->m_len; 281 icmplen = min(mblen, icmplen); 282 283 /* 284 * As we are not required to return everything we have, 285 * we return whatever we can return at ease. 286 * 287 * Note that ICMP datagrams longer than 576 octets are out of spec 288 * according to RFC1812; the limit on icmpreturndatabytes below in 289 * icmp_sysctl will keep things below that limit. 290 */ 291 292 KASSERT(ICMP_MINLEN <= MCLBYTES); 293 294 if (icmplen + ICMP_MINLEN > MCLBYTES) 295 icmplen = MCLBYTES - ICMP_MINLEN; 296 297 m = m_gethdr(M_DONTWAIT, MT_HEADER); 298 if (m && (icmplen + ICMP_MINLEN > MHLEN)) { 299 MCLGET(m, M_DONTWAIT); 300 if ((m->m_flags & M_EXT) == 0) { 301 m_freem(m); 302 m = NULL; 303 } 304 } 305 if (m == NULL) 306 goto freeit; 307 MCLAIM(m, n->m_owner); 308 m->m_len = icmplen + ICMP_MINLEN; 309 if ((m->m_flags & M_EXT) == 0) 310 MH_ALIGN(m, m->m_len); 311 else { 312 m->m_data += sizeof(struct ip); 313 m->m_len -= sizeof(struct ip); 314 } 315 icp = mtod(m, struct icmp *); 316 if ((u_int)type > ICMP_MAXTYPE) 317 panic("icmp_error"); 318 ICMP_STATINC(ICMP_STAT_OUTHIST + type); 319 icp->icmp_type = type; 320 if (type == ICMP_REDIRECT) 321 icp->icmp_gwaddr.s_addr = dest; 322 else { 323 icp->icmp_void = 0; 324 /* 325 * The following assignments assume an overlay with the 326 * zeroed icmp_void field. 327 */ 328 if (type == ICMP_PARAMPROB) { 329 icp->icmp_pptr = code; 330 code = 0; 331 } else if (type == ICMP_UNREACH && 332 code == ICMP_UNREACH_NEEDFRAG && destmtu) 333 icp->icmp_nextmtu = htons(destmtu); 334 } 335 336 icp->icmp_code = code; 337 m_copydata(n, 0, icmplen, (void *)&icp->icmp_ip); 338 339 /* 340 * Now, copy old ip header (without options) 341 * in front of icmp message. 342 */ 343 if ((m->m_flags & M_EXT) == 0 && 344 m->m_data - sizeof(struct ip) < m->m_pktdat) 345 panic("icmp len"); 346 m->m_data -= sizeof(struct ip); 347 m->m_len += sizeof(struct ip); 348 m->m_pkthdr.len = m->m_len; 349 m_copy_rcvif(m, n); 350 nip = mtod(m, struct ip *); 351 /* ip_v set in ip_output */ 352 nip->ip_hl = sizeof(struct ip) >> 2; 353 nip->ip_tos = 0; 354 nip->ip_len = htons(m->m_len); 355 /* ip_id set in ip_output */ 356 nip->ip_off = htons(0); 357 /* ip_ttl set in icmp_reflect */ 358 nip->ip_p = IPPROTO_ICMP; 359 nip->ip_src = oip->ip_src; 360 nip->ip_dst = oip->ip_dst; 361 /* move PF m_tag to new packet, if it exists */ 362 mtag = m_tag_find(n, PACKET_TAG_PF, NULL); 363 if (mtag != NULL) { 364 m_tag_unlink(n, mtag); 365 m_tag_prepend(m, mtag); 366 } 367 icmp_reflect(m); 368 369 freeit: 370 m_freem(n); 371 } 372 373 struct sockaddr_in icmpsrc = { 374 .sin_len = sizeof (struct sockaddr_in), 375 .sin_family = AF_INET, 376 }; 377 static struct sockaddr_in icmpdst = { 378 .sin_len = sizeof (struct sockaddr_in), 379 .sin_family = AF_INET, 380 }; 381 static struct sockaddr_in icmpgw = { 382 .sin_len = sizeof (struct sockaddr_in), 383 .sin_family = AF_INET, 384 }; 385 struct sockaddr_in icmpmask = { 386 .sin_len = 8, 387 .sin_family = 0, 388 }; 389 390 /* 391 * Process a received ICMP message. 392 */ 393 static void 394 _icmp_input(struct mbuf *m, int hlen, int proto) 395 { 396 struct icmp *icp; 397 struct ip *ip = mtod(m, struct ip *); 398 int icmplen; 399 int i; 400 struct in_ifaddr *ia; 401 void *(*ctlfunc)(int, const struct sockaddr *, void *); 402 int code; 403 struct rtentry *rt; 404 405 /* 406 * Locate icmp structure in mbuf, and check 407 * that not corrupted and of at least minimum length. 408 */ 409 icmplen = ntohs(ip->ip_len) - hlen; 410 #ifdef ICMPPRINTFS 411 if (icmpprintfs) { 412 char sbuf[INET_ADDRSTRLEN], dbuf[INET_ADDRSTRLEN]; 413 printf("icmp_input from `%s' to `%s', len %d\n", 414 IN_PRINT(sbuf, &ip->ip_src), IN_PRINT(dbuf, &ip->ip_dst), 415 icmplen); 416 } 417 #endif 418 if (icmplen < ICMP_MINLEN) { 419 ICMP_STATINC(ICMP_STAT_TOOSHORT); 420 goto freeit; 421 } 422 i = hlen + min(icmplen, ICMP_ADVLENMIN); 423 if ((m->m_len < i || M_READONLY(m)) && (m = m_pullup(m, i)) == NULL) { 424 ICMP_STATINC(ICMP_STAT_TOOSHORT); 425 return; 426 } 427 ip = mtod(m, struct ip *); 428 m->m_len -= hlen; 429 m->m_data += hlen; 430 icp = mtod(m, struct icmp *); 431 /* Don't need to assert alignment, here. */ 432 if (in_cksum(m, icmplen)) { 433 ICMP_STATINC(ICMP_STAT_CHECKSUM); 434 goto freeit; 435 } 436 m->m_len += hlen; 437 m->m_data -= hlen; 438 439 #ifdef ICMPPRINTFS 440 /* 441 * Message type specific processing. 442 */ 443 if (icmpprintfs) 444 printf("icmp_input(type:%d, code:%d)\n", icp->icmp_type, 445 icp->icmp_code); 446 #endif 447 if (icp->icmp_type > ICMP_MAXTYPE) 448 goto raw; 449 ICMP_STATINC(ICMP_STAT_INHIST + icp->icmp_type); 450 code = icp->icmp_code; 451 switch (icp->icmp_type) { 452 453 case ICMP_UNREACH: 454 switch (code) { 455 case ICMP_UNREACH_PROTOCOL: 456 code = PRC_UNREACH_PROTOCOL; 457 break; 458 459 case ICMP_UNREACH_PORT: 460 code = PRC_UNREACH_PORT; 461 break; 462 463 case ICMP_UNREACH_SRCFAIL: 464 code = PRC_UNREACH_SRCFAIL; 465 break; 466 467 case ICMP_UNREACH_NEEDFRAG: 468 code = PRC_MSGSIZE; 469 break; 470 471 case ICMP_UNREACH_NET: 472 case ICMP_UNREACH_NET_UNKNOWN: 473 case ICMP_UNREACH_NET_PROHIB: 474 case ICMP_UNREACH_TOSNET: 475 code = PRC_UNREACH_NET; 476 break; 477 478 case ICMP_UNREACH_HOST: 479 case ICMP_UNREACH_HOST_UNKNOWN: 480 case ICMP_UNREACH_ISOLATED: 481 case ICMP_UNREACH_HOST_PROHIB: 482 case ICMP_UNREACH_TOSHOST: 483 case ICMP_UNREACH_ADMIN_PROHIBIT: 484 case ICMP_UNREACH_HOST_PREC: 485 case ICMP_UNREACH_PREC_CUTOFF: 486 code = PRC_UNREACH_HOST; 487 break; 488 489 default: 490 goto badcode; 491 } 492 goto deliver; 493 494 case ICMP_TIMXCEED: 495 if (code > 1) 496 goto badcode; 497 code += PRC_TIMXCEED_INTRANS; 498 goto deliver; 499 500 case ICMP_PARAMPROB: 501 if (code > 1) 502 goto badcode; 503 code = PRC_PARAMPROB; 504 goto deliver; 505 506 case ICMP_SOURCEQUENCH: 507 if (code) 508 goto badcode; 509 code = PRC_QUENCH; 510 goto deliver; 511 512 deliver: 513 /* 514 * Problem with datagram; advise higher level routines. 515 */ 516 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || 517 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) { 518 ICMP_STATINC(ICMP_STAT_BADLEN); 519 goto freeit; 520 } 521 if (IN_MULTICAST(icp->icmp_ip.ip_dst.s_addr)) 522 goto badcode; 523 #ifdef ICMPPRINTFS 524 if (icmpprintfs) 525 printf("deliver to protocol %d\n", icp->icmp_ip.ip_p); 526 #endif 527 icmpsrc.sin_addr = icp->icmp_ip.ip_dst; 528 ctlfunc = inetsw[ip_protox[icp->icmp_ip.ip_p]].pr_ctlinput; 529 if (ctlfunc) 530 (void) (*ctlfunc)(code, sintosa(&icmpsrc), 531 &icp->icmp_ip); 532 break; 533 534 badcode: 535 ICMP_STATINC(ICMP_STAT_BADCODE); 536 break; 537 538 case ICMP_ECHO: 539 if (!icmpbmcastecho && 540 (m->m_flags & (M_MCAST | M_BCAST)) != 0) { 541 ICMP_STATINC(ICMP_STAT_BMCASTECHO); 542 break; 543 } 544 icp->icmp_type = ICMP_ECHOREPLY; 545 goto reflect; 546 547 case ICMP_TSTAMP: 548 if (icmplen < ICMP_TSLEN) { 549 ICMP_STATINC(ICMP_STAT_BADLEN); 550 break; 551 } 552 if (!icmpbmcastecho && 553 (m->m_flags & (M_MCAST | M_BCAST)) != 0) { 554 ICMP_STATINC(ICMP_STAT_BMCASTTSTAMP); 555 break; 556 } 557 icp->icmp_type = ICMP_TSTAMPREPLY; 558 icp->icmp_rtime = iptime(); 559 icp->icmp_ttime = icp->icmp_rtime; /* bogus, do later! */ 560 goto reflect; 561 562 case ICMP_MASKREQ: { 563 struct ifnet *rcvif; 564 int s, ss; 565 struct ifaddr *ifa; 566 567 if (icmpmaskrepl == 0) 568 break; 569 /* 570 * We are not able to respond with all ones broadcast 571 * unless we receive it over a point-to-point interface. 572 */ 573 if (icmplen < ICMP_MASKLEN) { 574 ICMP_STATINC(ICMP_STAT_BADLEN); 575 break; 576 } 577 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 578 in_nullhost(ip->ip_dst)) 579 icmpdst.sin_addr = ip->ip_src; 580 else 581 icmpdst.sin_addr = ip->ip_dst; 582 ss = pserialize_read_enter(); 583 rcvif = m_get_rcvif(m, &s); 584 ifa = ifaof_ifpforaddr(sintosa(&icmpdst), rcvif); 585 m_put_rcvif(rcvif, &s); 586 if (ifa == NULL) { 587 pserialize_read_exit(ss); 588 break; 589 } 590 ia = ifatoia(ifa); 591 icp->icmp_type = ICMP_MASKREPLY; 592 icp->icmp_mask = ia->ia_sockmask.sin_addr.s_addr; 593 if (in_nullhost(ip->ip_src)) { 594 if (ia->ia_ifp->if_flags & IFF_BROADCAST) 595 ip->ip_src = ia->ia_broadaddr.sin_addr; 596 else if (ia->ia_ifp->if_flags & IFF_POINTOPOINT) 597 ip->ip_src = ia->ia_dstaddr.sin_addr; 598 } 599 pserialize_read_exit(ss); 600 reflect: 601 { 602 uint64_t *icps = percpu_getref(icmpstat_percpu); 603 icps[ICMP_STAT_REFLECT]++; 604 icps[ICMP_STAT_OUTHIST + icp->icmp_type]++; 605 percpu_putref(icmpstat_percpu); 606 } 607 icmp_reflect(m); 608 return; 609 } 610 611 case ICMP_REDIRECT: 612 if (code > 3) 613 goto badcode; 614 if (icmp_rediraccept == 0) 615 goto freeit; 616 if (icmplen < ICMP_ADVLENMIN || icmplen < ICMP_ADVLEN(icp) || 617 icp->icmp_ip.ip_hl < (sizeof(struct ip) >> 2)) { 618 ICMP_STATINC(ICMP_STAT_BADLEN); 619 break; 620 } 621 /* 622 * Short circuit routing redirects to force 623 * immediate change in the kernel's routing 624 * tables. The message is also handed to anyone 625 * listening on a raw socket (e.g. the routing 626 * daemon for use in updating its tables). 627 */ 628 icmpgw.sin_addr = ip->ip_src; 629 icmpdst.sin_addr = icp->icmp_gwaddr; 630 #ifdef ICMPPRINTFS 631 if (icmpprintfs) { 632 char gbuf[INET_ADDRSTRLEN], dbuf[INET_ADDRSTRLEN]; 633 printf("redirect dst `%s' to `%s'\n", 634 IN_PRINT(dbuf, &icp->icmp_ip.ip_dst), 635 IN_PRINT(gbuf, &icp->icmp_gwaddr)); 636 } 637 #endif 638 icmpsrc.sin_addr = icp->icmp_ip.ip_dst; 639 rt = NULL; 640 rtredirect(sintosa(&icmpsrc), sintosa(&icmpdst), 641 NULL, RTF_GATEWAY | RTF_HOST, sintosa(&icmpgw), &rt); 642 if (rt != NULL && icmp_redirtimeout != 0) { 643 i = rt_timer_add(rt, icmp_redirect_timeout, 644 icmp_redirect_timeout_q); 645 if (i) { 646 char buf[INET_ADDRSTRLEN]; 647 log(LOG_ERR, "ICMP: redirect failed to " 648 "register timeout for route to %s, " 649 "code %d\n", 650 IN_PRINT(buf, &icp->icmp_ip.ip_dst), i); 651 } 652 } 653 if (rt != NULL) 654 rt_unref(rt); 655 656 pfctlinput(PRC_REDIRECT_HOST, sintosa(&icmpsrc)); 657 #if defined(IPSEC) 658 if (ipsec_used) 659 key_sa_routechange((struct sockaddr *)&icmpsrc); 660 #endif 661 break; 662 663 /* 664 * No kernel processing for the following; 665 * just fall through to send to raw listener. 666 */ 667 case ICMP_ECHOREPLY: 668 case ICMP_ROUTERADVERT: 669 case ICMP_ROUTERSOLICIT: 670 case ICMP_TSTAMPREPLY: 671 case ICMP_IREQREPLY: 672 case ICMP_MASKREPLY: 673 default: 674 break; 675 } 676 677 raw: 678 rip_input(m, hlen, proto); 679 return; 680 681 freeit: 682 m_freem(m); 683 return; 684 } 685 686 void 687 icmp_input(struct mbuf *m, ...) 688 { 689 int hlen, proto; 690 va_list ap; 691 692 va_start(ap, m); 693 hlen = va_arg(ap, int); 694 proto = va_arg(ap, int); 695 va_end(ap); 696 697 wqinput_input(icmp_wqinput, m, hlen, proto); 698 } 699 700 /* 701 * Reflect the ip packet back to the source 702 */ 703 void 704 icmp_reflect(struct mbuf *m) 705 { 706 struct ip *ip = mtod(m, struct ip *); 707 struct in_ifaddr *ia; 708 struct ifaddr *ifa; 709 struct sockaddr_in *sin; 710 struct in_addr t; 711 struct mbuf *opts = NULL; 712 int optlen = (ip->ip_hl << 2) - sizeof(struct ip); 713 struct ifnet *rcvif; 714 struct psref psref, psref_ia; 715 int s; 716 int bound; 717 718 bound = curlwp_bind(); 719 720 if (!in_canforward(ip->ip_src) && 721 ((ip->ip_src.s_addr & IN_CLASSA_NET) != 722 htonl(IN_LOOPBACKNET << IN_CLASSA_NSHIFT))) { 723 m_freem(m); /* Bad return address */ 724 goto done; /* ip_output() will check for broadcast */ 725 } 726 t = ip->ip_dst; 727 ip->ip_dst = ip->ip_src; 728 /* 729 * If the incoming packet was addressed directly to us, use 730 * dst as the src for the reply. Otherwise (broadcast or 731 * anonymous), use an address which corresponds to the 732 * incoming interface, with a preference for the address which 733 * corresponds to the route to the destination of the ICMP. 734 */ 735 736 /* Look for packet addressed to us */ 737 ia = in_get_ia_psref(t, &psref_ia); 738 if (ia && (ia->ia4_flags & IN_IFF_NOTREADY)) { 739 ia4_release(ia, &psref_ia); 740 ia = NULL; 741 } 742 743 rcvif = m_get_rcvif_psref(m, &psref); 744 745 /* look for packet sent to broadcast address */ 746 if (ia == NULL && rcvif && 747 (rcvif->if_flags & IFF_BROADCAST)) { 748 s = pserialize_read_enter(); 749 IFADDR_READER_FOREACH(ifa, rcvif) { 750 if (ifa->ifa_addr->sa_family != AF_INET) 751 continue; 752 if (in_hosteq(t,ifatoia(ifa)->ia_broadaddr.sin_addr)) { 753 ia = ifatoia(ifa); 754 if ((ia->ia4_flags & IN_IFF_NOTREADY) == 0) 755 break; 756 ia = NULL; 757 } 758 } 759 if (ia != NULL) 760 ia4_acquire(ia, &psref_ia); 761 pserialize_read_exit(s); 762 } 763 764 sin = ia ? &ia->ia_addr : NULL; 765 766 icmpdst.sin_addr = t; 767 768 /* 769 * if the packet is addressed somewhere else, compute the 770 * source address for packets routed back to the source, and 771 * use that, if it's an address on the interface which 772 * received the packet 773 */ 774 if (sin == NULL && rcvif) { 775 struct sockaddr_in sin_dst; 776 struct route icmproute; 777 int errornum; 778 779 sockaddr_in_init(&sin_dst, &ip->ip_dst, 0); 780 memset(&icmproute, 0, sizeof(icmproute)); 781 errornum = 0; 782 ia = in_selectsrc(&sin_dst, &icmproute, 0, NULL, &errornum, 783 &psref_ia); 784 /* errornum is never used */ 785 rtcache_free(&icmproute); 786 /* check to make sure sin is a source address on rcvif */ 787 if (ia != NULL) { 788 sin = &ia->ia_addr; 789 t = sin->sin_addr; 790 sin = NULL; 791 ia4_release(ia, &psref_ia); 792 ia = in_get_ia_on_iface_psref(t, rcvif, &psref_ia); 793 if (ia != NULL) 794 sin = &ia->ia_addr; 795 } 796 } 797 798 /* 799 * if it was not addressed to us, but the route doesn't go out 800 * the source interface, pick an address on the source 801 * interface. This can happen when routing is asymmetric, or 802 * when the incoming packet was encapsulated 803 */ 804 if (sin == NULL && rcvif) { 805 KASSERT(ia == NULL); 806 s = pserialize_read_enter(); 807 IFADDR_READER_FOREACH(ifa, rcvif) { 808 if (ifa->ifa_addr->sa_family != AF_INET) 809 continue; 810 sin = &(ifatoia(ifa)->ia_addr); 811 ia = ifatoia(ifa); 812 ia4_acquire(ia, &psref_ia); 813 break; 814 } 815 pserialize_read_exit(s); 816 } 817 818 m_put_rcvif_psref(rcvif, &psref); 819 820 /* 821 * The following happens if the packet was not addressed to us, 822 * and was received on an interface with no IP address: 823 * We find the first AF_INET address on the first non-loopback 824 * interface. 825 */ 826 if (sin == NULL) { 827 KASSERT(ia == NULL); 828 s = pserialize_read_enter(); 829 IN_ADDRLIST_READER_FOREACH(ia) { 830 if (ia->ia_ifp->if_flags & IFF_LOOPBACK) 831 continue; 832 sin = &ia->ia_addr; 833 ia4_acquire(ia, &psref_ia); 834 break; 835 } 836 pserialize_read_exit(s); 837 } 838 839 /* 840 * If we still didn't find an address, punt. We could have an 841 * interface up (and receiving packets) with no address. 842 */ 843 if (sin == NULL) { 844 KASSERT(ia == NULL); 845 m_freem(m); 846 goto done; 847 } 848 849 ip->ip_src = sin->sin_addr; 850 ip->ip_ttl = MAXTTL; 851 852 if (ia != NULL) 853 ia4_release(ia, &psref_ia); 854 855 if (optlen > 0) { 856 u_char *cp; 857 int opt, cnt; 858 u_int len; 859 860 /* 861 * Retrieve any source routing from the incoming packet; 862 * add on any record-route or timestamp options. 863 */ 864 cp = (u_char *) (ip + 1); 865 if ((opts = ip_srcroute()) == NULL && 866 (opts = m_gethdr(M_DONTWAIT, MT_HEADER))) { 867 MCLAIM(opts, m->m_owner); 868 opts->m_len = sizeof(struct in_addr); 869 *mtod(opts, struct in_addr *) = zeroin_addr; 870 } 871 if (opts) { 872 #ifdef ICMPPRINTFS 873 if (icmpprintfs) 874 printf("icmp_reflect optlen %d rt %d => ", 875 optlen, opts->m_len); 876 #endif 877 for (cnt = optlen; cnt > 0; cnt -= len, cp += len) { 878 opt = cp[IPOPT_OPTVAL]; 879 if (opt == IPOPT_EOL) 880 break; 881 if (opt == IPOPT_NOP) 882 len = 1; 883 else { 884 if (cnt < IPOPT_OLEN + sizeof(*cp)) 885 break; 886 len = cp[IPOPT_OLEN]; 887 if (len < IPOPT_OLEN + sizeof(*cp) || 888 len > cnt) 889 break; 890 } 891 /* 892 * Should check for overflow, but it "can't happen" 893 */ 894 if (opt == IPOPT_RR || opt == IPOPT_TS || 895 opt == IPOPT_SECURITY) { 896 memmove(mtod(opts, char *) + opts->m_len, 897 cp, len); 898 opts->m_len += len; 899 } 900 } 901 /* Terminate & pad, if necessary */ 902 if ((cnt = opts->m_len % 4) != 0) { 903 for (; cnt < 4; cnt++) { 904 *(mtod(opts, char *) + opts->m_len) = 905 IPOPT_EOL; 906 opts->m_len++; 907 } 908 } 909 #ifdef ICMPPRINTFS 910 if (icmpprintfs) 911 printf("%d\n", opts->m_len); 912 #endif 913 } 914 /* 915 * Now strip out original options by copying rest of first 916 * mbuf's data back, and adjust the IP length. 917 */ 918 ip->ip_len = htons(ntohs(ip->ip_len) - optlen); 919 ip->ip_hl = sizeof(struct ip) >> 2; 920 m->m_len -= optlen; 921 if (m->m_flags & M_PKTHDR) 922 m->m_pkthdr.len -= optlen; 923 optlen += sizeof(struct ip); 924 memmove(ip + 1, (char *)ip + optlen, 925 (unsigned)(m->m_len - sizeof(struct ip))); 926 } 927 m_tag_delete_nonpersistent(m); 928 m->m_flags &= ~(M_BCAST|M_MCAST); 929 930 /* 931 * Clear any in-bound checksum flags for this packet. 932 */ 933 if (m->m_flags & M_PKTHDR) 934 m->m_pkthdr.csum_flags = 0; 935 936 icmp_send(m, opts); 937 done: 938 curlwp_bindx(bound); 939 if (opts) 940 (void)m_free(opts); 941 } 942 943 /* 944 * Send an icmp packet back to the ip level, 945 * after supplying a checksum. 946 */ 947 void 948 icmp_send(struct mbuf *m, struct mbuf *opts) 949 { 950 struct ip *ip = mtod(m, struct ip *); 951 int hlen; 952 struct icmp *icp; 953 954 hlen = ip->ip_hl << 2; 955 m->m_data += hlen; 956 m->m_len -= hlen; 957 icp = mtod(m, struct icmp *); 958 icp->icmp_cksum = 0; 959 icp->icmp_cksum = in_cksum(m, ntohs(ip->ip_len) - hlen); 960 m->m_data -= hlen; 961 m->m_len += hlen; 962 #ifdef ICMPPRINTFS 963 if (icmpprintfs) { 964 char sbuf[INET_ADDRSTRLEN], dbuf[INET_ADDRSTRLEN]; 965 printf("icmp_send to destination `%s' from `%s'\n", 966 IN_PRINT(dbuf, &ip->ip_dst), IN_PRINT(sbuf, &ip->ip_src)); 967 } 968 #endif 969 (void)ip_output(m, opts, NULL, 0, NULL, NULL); 970 } 971 972 n_time 973 iptime(void) 974 { 975 struct timeval atv; 976 u_long t; 977 978 microtime(&atv); 979 t = (atv.tv_sec % (24*60*60)) * 1000 + atv.tv_usec / 1000; 980 return (htonl(t)); 981 } 982 983 /* 984 * sysctl helper routine for net.inet.icmp.returndatabytes. ensures 985 * that the new value is in the correct range. 986 */ 987 static int 988 sysctl_net_inet_icmp_returndatabytes(SYSCTLFN_ARGS) 989 { 990 int error, t; 991 struct sysctlnode node; 992 993 node = *rnode; 994 node.sysctl_data = &t; 995 t = icmpreturndatabytes; 996 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 997 if (error || newp == NULL) 998 return (error); 999 1000 if (t < 8 || t > 512) 1001 return (EINVAL); 1002 icmpreturndatabytes = t; 1003 1004 return (0); 1005 } 1006 1007 /* 1008 * sysctl helper routine for net.inet.icmp.redirtimeout. ensures that 1009 * the given value is not less than zero and then resets the timeout 1010 * queue. 1011 */ 1012 static int 1013 sysctl_net_inet_icmp_redirtimeout(SYSCTLFN_ARGS) 1014 { 1015 int error, tmp; 1016 struct sysctlnode node; 1017 1018 node = *rnode; 1019 node.sysctl_data = &tmp; 1020 tmp = icmp_redirtimeout; 1021 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1022 if (error || newp == NULL) 1023 return (error); 1024 if (tmp < 0) 1025 return (EINVAL); 1026 icmp_redirtimeout = tmp; 1027 1028 /* XXX NOMPSAFE still need softnet_lock */ 1029 mutex_enter(softnet_lock); 1030 1031 /* 1032 * was it a *defined* side-effect that anyone even *reading* 1033 * this value causes these things to happen? 1034 */ 1035 if (icmp_redirect_timeout_q != NULL) { 1036 if (icmp_redirtimeout == 0) { 1037 rt_timer_queue_destroy(icmp_redirect_timeout_q); 1038 icmp_redirect_timeout_q = NULL; 1039 } else { 1040 rt_timer_queue_change(icmp_redirect_timeout_q, 1041 icmp_redirtimeout); 1042 } 1043 } else if (icmp_redirtimeout > 0) { 1044 icmp_redirect_timeout_q = 1045 rt_timer_queue_create(icmp_redirtimeout); 1046 } 1047 1048 mutex_exit(softnet_lock); 1049 1050 return (0); 1051 } 1052 1053 static int 1054 sysctl_net_inet_icmp_stats(SYSCTLFN_ARGS) 1055 { 1056 1057 return (NETSTAT_SYSCTL(icmpstat_percpu, ICMP_NSTATS)); 1058 } 1059 1060 static void 1061 sysctl_netinet_icmp_setup(struct sysctllog **clog) 1062 { 1063 1064 sysctl_createv(clog, 0, NULL, NULL, 1065 CTLFLAG_PERMANENT, 1066 CTLTYPE_NODE, "inet", NULL, 1067 NULL, 0, NULL, 0, 1068 CTL_NET, PF_INET, CTL_EOL); 1069 sysctl_createv(clog, 0, NULL, NULL, 1070 CTLFLAG_PERMANENT, 1071 CTLTYPE_NODE, "icmp", 1072 SYSCTL_DESCR("ICMPv4 related settings"), 1073 NULL, 0, NULL, 0, 1074 CTL_NET, PF_INET, IPPROTO_ICMP, CTL_EOL); 1075 1076 sysctl_createv(clog, 0, NULL, NULL, 1077 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1078 CTLTYPE_INT, "maskrepl", 1079 SYSCTL_DESCR("Respond to ICMP_MASKREQ messages"), 1080 NULL, 0, &icmpmaskrepl, 0, 1081 CTL_NET, PF_INET, IPPROTO_ICMP, 1082 ICMPCTL_MASKREPL, CTL_EOL); 1083 sysctl_createv(clog, 0, NULL, NULL, 1084 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1085 CTLTYPE_INT, "returndatabytes", 1086 SYSCTL_DESCR("Number of bytes to return in an ICMP " 1087 "error message"), 1088 sysctl_net_inet_icmp_returndatabytes, 0, 1089 &icmpreturndatabytes, 0, 1090 CTL_NET, PF_INET, IPPROTO_ICMP, 1091 ICMPCTL_RETURNDATABYTES, CTL_EOL); 1092 sysctl_createv(clog, 0, NULL, NULL, 1093 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1094 CTLTYPE_INT, "errppslimit", 1095 SYSCTL_DESCR("Maximum number of outgoing ICMP error " 1096 "messages per second"), 1097 NULL, 0, &icmperrppslim, 0, 1098 CTL_NET, PF_INET, IPPROTO_ICMP, 1099 ICMPCTL_ERRPPSLIMIT, CTL_EOL); 1100 sysctl_createv(clog, 0, NULL, NULL, 1101 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1102 CTLTYPE_INT, "rediraccept", 1103 SYSCTL_DESCR("Accept ICMP_REDIRECT messages"), 1104 NULL, 0, &icmp_rediraccept, 0, 1105 CTL_NET, PF_INET, IPPROTO_ICMP, 1106 ICMPCTL_REDIRACCEPT, CTL_EOL); 1107 sysctl_createv(clog, 0, NULL, NULL, 1108 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1109 CTLTYPE_INT, "redirtimeout", 1110 SYSCTL_DESCR("Lifetime of ICMP_REDIRECT generated " 1111 "routes"), 1112 sysctl_net_inet_icmp_redirtimeout, 0, 1113 &icmp_redirtimeout, 0, 1114 CTL_NET, PF_INET, IPPROTO_ICMP, 1115 ICMPCTL_REDIRTIMEOUT, CTL_EOL); 1116 sysctl_createv(clog, 0, NULL, NULL, 1117 CTLFLAG_PERMANENT, 1118 CTLTYPE_STRUCT, "stats", 1119 SYSCTL_DESCR("ICMP statistics"), 1120 sysctl_net_inet_icmp_stats, 0, NULL, 0, 1121 CTL_NET, PF_INET, IPPROTO_ICMP, ICMPCTL_STATS, 1122 CTL_EOL); 1123 sysctl_createv(clog, 0, NULL, NULL, 1124 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1125 CTLTYPE_INT, "bmcastecho", 1126 SYSCTL_DESCR("Respond to ICMP_ECHO or ICMP_TIMESTAMP " 1127 "message to the broadcast or multicast"), 1128 NULL, 0, &icmpbmcastecho, 0, 1129 CTL_NET, PF_INET, IPPROTO_ICMP, ICMPCTL_BMCASTECHO, 1130 CTL_EOL); 1131 } 1132 1133 void 1134 icmp_statinc(u_int stat) 1135 { 1136 1137 KASSERT(stat < ICMP_NSTATS); 1138 ICMP_STATINC(stat); 1139 } 1140 1141 /* Table of common MTUs: */ 1142 1143 static const u_int mtu_table[] = { 1144 65535, 65280, 32000, 17914, 9180, 8166, 1145 4352, 2002, 1492, 1006, 508, 296, 68, 0 1146 }; 1147 1148 void 1149 icmp_mtudisc(struct icmp *icp, struct in_addr faddr) 1150 { 1151 struct icmp_mtudisc_callback *mc; 1152 struct sockaddr *dst = sintosa(&icmpsrc); 1153 struct rtentry *rt; 1154 u_long mtu = ntohs(icp->icmp_nextmtu); /* Why a long? IPv6 */ 1155 int error; 1156 1157 rt = rtalloc1(dst, 1); 1158 if (rt == NULL) 1159 return; 1160 1161 /* If we didn't get a host route, allocate one */ 1162 1163 if ((rt->rt_flags & RTF_HOST) == 0) { 1164 struct rtentry *nrt; 1165 1166 error = rtrequest(RTM_ADD, dst, rt->rt_gateway, NULL, 1167 RTF_GATEWAY | RTF_HOST | RTF_DYNAMIC, &nrt); 1168 if (error) { 1169 rt_unref(rt); 1170 return; 1171 } 1172 nrt->rt_rmx = rt->rt_rmx; 1173 rt_unref(rt); 1174 rt = nrt; 1175 } 1176 1177 if (ip_mtudisc_timeout_q == NULL) 1178 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 1179 error = rt_timer_add(rt, icmp_mtudisc_timeout, ip_mtudisc_timeout_q); 1180 if (error) { 1181 rt_unref(rt); 1182 return; 1183 } 1184 1185 if (mtu == 0) { 1186 int i = 0; 1187 1188 mtu = ntohs(icp->icmp_ip.ip_len); 1189 /* Some 4.2BSD-based routers incorrectly adjust the ip_len */ 1190 if (mtu > rt->rt_rmx.rmx_mtu && rt->rt_rmx.rmx_mtu != 0) 1191 mtu -= (icp->icmp_ip.ip_hl << 2); 1192 1193 /* If we still can't guess a value, try the route */ 1194 1195 if (mtu == 0) { 1196 mtu = rt->rt_rmx.rmx_mtu; 1197 1198 /* If no route mtu, default to the interface mtu */ 1199 1200 if (mtu == 0) 1201 mtu = rt->rt_ifp->if_mtu; 1202 } 1203 1204 for (i = 0; i < sizeof(mtu_table) / sizeof(mtu_table[0]); i++) 1205 if (mtu > mtu_table[i]) { 1206 mtu = mtu_table[i]; 1207 break; 1208 } 1209 } 1210 1211 /* 1212 * XXX: RTV_MTU is overloaded, since the admin can set it 1213 * to turn off PMTU for a route, and the kernel can 1214 * set it to indicate a serious problem with PMTU 1215 * on a route. We should be using a separate flag 1216 * for the kernel to indicate this. 1217 */ 1218 1219 if ((rt->rt_rmx.rmx_locks & RTV_MTU) == 0) { 1220 if (mtu < 296 || mtu > rt->rt_ifp->if_mtu) 1221 rt->rt_rmx.rmx_locks |= RTV_MTU; 1222 else if (rt->rt_rmx.rmx_mtu > mtu || 1223 rt->rt_rmx.rmx_mtu == 0) { 1224 ICMP_STATINC(ICMP_STAT_PMTUCHG); 1225 rt->rt_rmx.rmx_mtu = mtu; 1226 } 1227 } 1228 1229 if (rt != NULL) 1230 rt_unref(rt); 1231 1232 /* 1233 * Notify protocols that the MTU for this destination 1234 * has changed. 1235 */ 1236 for (mc = LIST_FIRST(&icmp_mtudisc_callbacks); mc != NULL; 1237 mc = LIST_NEXT(mc, mc_list)) 1238 (*mc->mc_func)(faddr); 1239 } 1240 1241 /* 1242 * Return the next larger or smaller MTU plateau (table from RFC 1191) 1243 * given current value MTU. If DIR is less than zero, a larger plateau 1244 * is returned; otherwise, a smaller value is returned. 1245 */ 1246 u_int 1247 ip_next_mtu(u_int mtu, int dir) /* XXX */ 1248 { 1249 int i; 1250 1251 for (i = 0; i < (sizeof mtu_table) / (sizeof mtu_table[0]); i++) { 1252 if (mtu >= mtu_table[i]) 1253 break; 1254 } 1255 1256 if (dir < 0) { 1257 if (i == 0) { 1258 return 0; 1259 } else { 1260 return mtu_table[i - 1]; 1261 } 1262 } else { 1263 if (mtu_table[i] == 0) { 1264 return 0; 1265 } else if (mtu > mtu_table[i]) { 1266 return mtu_table[i]; 1267 } else { 1268 return mtu_table[i + 1]; 1269 } 1270 } 1271 } 1272 1273 static void 1274 icmp_mtudisc_timeout(struct rtentry *rt, struct rttimer *r) 1275 { 1276 1277 KASSERT(rt != NULL); 1278 rt_assert_referenced(rt); 1279 1280 if ((rt->rt_flags & (RTF_DYNAMIC | RTF_HOST)) == 1281 (RTF_DYNAMIC | RTF_HOST)) { 1282 rtrequest(RTM_DELETE, rt_getkey(rt), 1283 rt->rt_gateway, rt_mask(rt), rt->rt_flags, NULL); 1284 } else { 1285 if ((rt->rt_rmx.rmx_locks & RTV_MTU) == 0) { 1286 rt->rt_rmx.rmx_mtu = 0; 1287 } 1288 } 1289 } 1290 1291 static void 1292 icmp_redirect_timeout(struct rtentry *rt, struct rttimer *r) 1293 { 1294 1295 KASSERT(rt != NULL); 1296 rt_assert_referenced(rt); 1297 1298 if ((rt->rt_flags & (RTF_DYNAMIC | RTF_HOST)) == 1299 (RTF_DYNAMIC | RTF_HOST)) { 1300 rtrequest(RTM_DELETE, rt_getkey(rt), 1301 rt->rt_gateway, rt_mask(rt), rt->rt_flags, NULL); 1302 } 1303 } 1304 1305 /* 1306 * Perform rate limit check. 1307 * Returns 0 if it is okay to send the icmp packet. 1308 * Returns 1 if the router SHOULD NOT send this icmp packet due to rate 1309 * limitation. 1310 * 1311 * XXX per-destination/type check necessary? 1312 */ 1313 int 1314 icmp_ratelimit(const struct in_addr *dst, const int type, 1315 const int code) 1316 { 1317 1318 /* PPS limit */ 1319 if (!ppsratecheck(&icmperrppslim_last, &icmperrpps_count, 1320 icmperrppslim)) { 1321 /* The packet is subject to rate limit */ 1322 return 1; 1323 } 1324 1325 /* okay to send */ 1326 return 0; 1327 } 1328
Indexes created Fri Oct 03 02:09:57 GMT 2025