1 /* $NetBSD: ip_input.c,v 1.289 2010/07/19 14:09:45 rmind 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 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 * Redistribution and use in source and binary forms, with or without 41 * modification, are permitted provided that the following conditions 42 * are met: 43 * 1. Redistributions of source code must retain the above copyright 44 * notice, this list of conditions and the following disclaimer. 45 * 2. Redistributions in binary form must reproduce the above copyright 46 * notice, this list of conditions and the following disclaimer in the 47 * documentation and/or other materials provided with the distribution. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 50 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 51 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 52 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 53 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 54 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 55 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 56 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 57 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 58 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 59 * POSSIBILITY OF SUCH DAMAGE. 60 */ 61 62 /* 63 * Copyright (c) 1982, 1986, 1988, 1993 64 * The Regents of the University of California. All rights reserved. 65 * 66 * Redistribution and use in source and binary forms, with or without 67 * modification, are permitted provided that the following conditions 68 * are met: 69 * 1. Redistributions of source code must retain the above copyright 70 * notice, this list of conditions and the following disclaimer. 71 * 2. Redistributions in binary form must reproduce the above copyright 72 * notice, this list of conditions and the following disclaimer in the 73 * documentation and/or other materials provided with the distribution. 74 * 3. Neither the name of the University nor the names of its contributors 75 * may be used to endorse or promote products derived from this software 76 * without specific prior written permission. 77 * 78 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 79 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 80 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 81 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 82 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 83 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 84 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 85 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 86 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 87 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 88 * SUCH DAMAGE. 89 * 90 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 91 */ 92 93 #include <sys/cdefs.h> 94 __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.289 2010/07/19 14:09:45 rmind Exp $"); 95 96 #include "opt_inet.h" 97 #include "opt_compat_netbsd.h" 98 #include "opt_gateway.h" 99 #include "opt_pfil_hooks.h" 100 #include "opt_ipsec.h" 101 #include "opt_mrouting.h" 102 #include "opt_mbuftrace.h" 103 #include "opt_inet_csum.h" 104 105 #include <sys/param.h> 106 #include <sys/systm.h> 107 #include <sys/mbuf.h> 108 #include <sys/domain.h> 109 #include <sys/protosw.h> 110 #include <sys/socket.h> 111 #include <sys/socketvar.h> 112 #include <sys/errno.h> 113 #include <sys/time.h> 114 #include <sys/kernel.h> 115 #include <sys/pool.h> 116 #include <sys/sysctl.h> 117 #include <sys/kauth.h> 118 119 #include <net/if.h> 120 #include <net/if_dl.h> 121 #include <net/route.h> 122 #include <net/pfil.h> 123 124 #include <netinet/in.h> 125 #include <netinet/in_systm.h> 126 #include <netinet/ip.h> 127 #include <netinet/in_pcb.h> 128 #include <netinet/in_proto.h> 129 #include <netinet/in_var.h> 130 #include <netinet/ip_var.h> 131 #include <netinet/ip_private.h> 132 #include <netinet/ip_icmp.h> 133 /* just for gif_ttl */ 134 #include <netinet/in_gif.h> 135 #include "gif.h" 136 #include <net/if_gre.h> 137 #include "gre.h" 138 139 #ifdef MROUTING 140 #include <netinet/ip_mroute.h> 141 #endif 142 143 #ifdef IPSEC 144 #include <netinet6/ipsec.h> 145 #include <netinet6/ipsec_private.h> 146 #include <netkey/key.h> 147 #endif 148 #ifdef FAST_IPSEC 149 #include <netipsec/ipsec.h> 150 #include <netipsec/key.h> 151 #endif /* FAST_IPSEC*/ 152 153 #ifndef IPFORWARDING 154 #ifdef GATEWAY 155 #define IPFORWARDING 1 /* forward IP packets not for us */ 156 #else /* GATEWAY */ 157 #define IPFORWARDING 0 /* don't forward IP packets not for us */ 158 #endif /* GATEWAY */ 159 #endif /* IPFORWARDING */ 160 #ifndef IPSENDREDIRECTS 161 #define IPSENDREDIRECTS 1 162 #endif 163 #ifndef IPFORWSRCRT 164 #define IPFORWSRCRT 1 /* forward source-routed packets */ 165 #endif 166 #ifndef IPALLOWSRCRT 167 #define IPALLOWSRCRT 1 /* allow source-routed packets */ 168 #endif 169 #ifndef IPMTUDISC 170 #define IPMTUDISC 1 171 #endif 172 #ifndef IPMTUDISCTIMEOUT 173 #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 174 #endif 175 176 #ifdef COMPAT_50 177 #include <compat/sys/time.h> 178 #include <compat/sys/socket.h> 179 #endif 180 181 /* 182 * Note: DIRECTED_BROADCAST is handled this way so that previous 183 * configuration using this option will Just Work. 184 */ 185 #ifndef IPDIRECTEDBCAST 186 #ifdef DIRECTED_BROADCAST 187 #define IPDIRECTEDBCAST 1 188 #else 189 #define IPDIRECTEDBCAST 0 190 #endif /* DIRECTED_BROADCAST */ 191 #endif /* IPDIRECTEDBCAST */ 192 int ipforwarding = IPFORWARDING; 193 int ipsendredirects = IPSENDREDIRECTS; 194 int ip_defttl = IPDEFTTL; 195 int ip_forwsrcrt = IPFORWSRCRT; 196 int ip_directedbcast = IPDIRECTEDBCAST; 197 int ip_allowsrcrt = IPALLOWSRCRT; 198 int ip_mtudisc = IPMTUDISC; 199 int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 200 #ifdef DIAGNOSTIC 201 int ipprintfs = 0; 202 #endif 203 204 int ip_do_randomid = 0; 205 206 /* 207 * XXX - Setting ip_checkinterface mostly implements the receive side of 208 * the Strong ES model described in RFC 1122, but since the routing table 209 * and transmit implementation do not implement the Strong ES model, 210 * setting this to 1 results in an odd hybrid. 211 * 212 * XXX - ip_checkinterface currently must be disabled if you use ipnat 213 * to translate the destination address to another local interface. 214 * 215 * XXX - ip_checkinterface must be disabled if you add IP aliases 216 * to the loopback interface instead of the interface where the 217 * packets for those addresses are received. 218 */ 219 int ip_checkinterface = 0; 220 221 222 struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 223 224 int ipqmaxlen = IFQ_MAXLEN; 225 u_long in_ifaddrhash; /* size of hash table - 1 */ 226 int in_ifaddrentries; /* total number of addrs */ 227 struct in_ifaddrhead in_ifaddrhead; 228 struct in_ifaddrhashhead *in_ifaddrhashtbl; 229 u_long in_multihash; /* size of hash table - 1 */ 230 int in_multientries; /* total number of addrs */ 231 struct in_multihashhead *in_multihashtbl; 232 struct ifqueue ipintrq; 233 234 uint16_t ip_id; 235 236 percpu_t *ipstat_percpu; 237 238 #ifdef PFIL_HOOKS 239 struct pfil_head inet_pfil_hook; 240 #endif 241 242 struct pool inmulti_pool; 243 244 #ifdef INET_CSUM_COUNTERS 245 #include <sys/device.h> 246 247 struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 248 NULL, "inet", "hwcsum bad"); 249 struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 250 NULL, "inet", "hwcsum ok"); 251 struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 252 NULL, "inet", "swcsum"); 253 254 #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 255 256 EVCNT_ATTACH_STATIC(ip_hwcsum_bad); 257 EVCNT_ATTACH_STATIC(ip_hwcsum_ok); 258 EVCNT_ATTACH_STATIC(ip_swcsum); 259 260 #else 261 262 #define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 263 264 #endif /* INET_CSUM_COUNTERS */ 265 266 /* 267 * We need to save the IP options in case a protocol wants to respond 268 * to an incoming packet over the same route if the packet got here 269 * using IP source routing. This allows connection establishment and 270 * maintenance when the remote end is on a network that is not known 271 * to us. 272 */ 273 int ip_nhops = 0; 274 static struct ip_srcrt { 275 struct in_addr dst; /* final destination */ 276 char nop; /* one NOP to align */ 277 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 278 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 279 } ip_srcrt; 280 281 static void save_rte(u_char *, struct in_addr); 282 283 #ifdef MBUFTRACE 284 struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 285 struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 286 #endif 287 288 static void sysctl_net_inet_ip_setup(struct sysctllog **); 289 290 /* 291 * IP initialization: fill in IP protocol switch table. 292 * All protocols not implemented in kernel go to raw IP protocol handler. 293 */ 294 void 295 ip_init(void) 296 { 297 const struct protosw *pr; 298 int i; 299 300 sysctl_net_inet_ip_setup(NULL); 301 302 pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", 303 NULL, IPL_SOFTNET); 304 305 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 306 if (pr == 0) 307 panic("ip_init"); 308 for (i = 0; i < IPPROTO_MAX; i++) 309 ip_protox[i] = pr - inetsw; 310 for (pr = inetdomain.dom_protosw; 311 pr < inetdomain.dom_protoswNPROTOSW; pr++) 312 if (pr->pr_domain->dom_family == PF_INET && 313 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 314 ip_protox[pr->pr_protocol] = pr - inetsw; 315 316 ip_reass_init(); 317 318 ip_initid(); 319 ip_id = time_second & 0xfffff; 320 321 ipintrq.ifq_maxlen = ipqmaxlen; 322 323 TAILQ_INIT(&in_ifaddrhead); 324 in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true, 325 &in_ifaddrhash); 326 in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true, 327 &in_multihash); 328 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 329 #ifdef GATEWAY 330 ipflow_init(ip_hashsize); 331 #endif 332 333 #ifdef PFIL_HOOKS 334 /* Register our Packet Filter hook. */ 335 inet_pfil_hook.ph_type = PFIL_TYPE_AF; 336 inet_pfil_hook.ph_af = AF_INET; 337 i = pfil_head_register(&inet_pfil_hook); 338 if (i != 0) 339 printf("ip_init: WARNING: unable to register pfil hook, " 340 "error %d\n", i); 341 #endif /* PFIL_HOOKS */ 342 343 #ifdef MBUFTRACE 344 MOWNER_ATTACH(&ip_tx_mowner); 345 MOWNER_ATTACH(&ip_rx_mowner); 346 #endif /* MBUFTRACE */ 347 348 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 349 } 350 351 struct sockaddr_in ipaddr = { 352 .sin_len = sizeof(ipaddr), 353 .sin_family = AF_INET, 354 }; 355 struct route ipforward_rt; 356 357 /* 358 * IP software interrupt routine 359 */ 360 void 361 ipintr(void) 362 { 363 int s; 364 struct mbuf *m; 365 struct ifqueue lcl_intrq; 366 367 memset(&lcl_intrq, 0, sizeof(lcl_intrq)); 368 ipintrq.ifq_maxlen = ipqmaxlen; 369 370 mutex_enter(softnet_lock); 371 KERNEL_LOCK(1, NULL); 372 if (!IF_IS_EMPTY(&ipintrq)) { 373 s = splnet(); 374 375 /* Take existing queue onto stack */ 376 lcl_intrq = ipintrq; 377 378 /* Zero out global queue, preserving maxlen and drops */ 379 ipintrq.ifq_head = NULL; 380 ipintrq.ifq_tail = NULL; 381 ipintrq.ifq_len = 0; 382 ipintrq.ifq_maxlen = lcl_intrq.ifq_maxlen; 383 ipintrq.ifq_drops = lcl_intrq.ifq_drops; 384 385 splx(s); 386 } 387 KERNEL_UNLOCK_ONE(NULL); 388 while (!IF_IS_EMPTY(&lcl_intrq)) { 389 IF_DEQUEUE(&lcl_intrq, m); 390 if (m == NULL) 391 break; 392 ip_input(m); 393 } 394 mutex_exit(softnet_lock); 395 } 396 397 /* 398 * Ip input routine. Checksum and byte swap header. If fragmented 399 * try to reassemble. Process options. Pass to next level. 400 */ 401 void 402 ip_input(struct mbuf *m) 403 { 404 struct ip *ip = NULL; 405 struct in_ifaddr *ia; 406 struct ifaddr *ifa; 407 int hlen = 0, len; 408 int downmatch; 409 int checkif; 410 int srcrt = 0; 411 #ifdef FAST_IPSEC 412 struct m_tag *mtag; 413 struct tdb_ident *tdbi; 414 struct secpolicy *sp; 415 int error, s; 416 #endif /* FAST_IPSEC */ 417 418 MCLAIM(m, &ip_rx_mowner); 419 KASSERT((m->m_flags & M_PKTHDR) != 0); 420 421 /* 422 * If no IP addresses have been set yet but the interfaces 423 * are receiving, can't do anything with incoming packets yet. 424 */ 425 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 426 goto bad; 427 IP_STATINC(IP_STAT_TOTAL); 428 /* 429 * If the IP header is not aligned, slurp it up into a new 430 * mbuf with space for link headers, in the event we forward 431 * it. Otherwise, if it is aligned, make sure the entire 432 * base IP header is in the first mbuf of the chain. 433 */ 434 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 435 if ((m = m_copyup(m, sizeof(struct ip), 436 (max_linkhdr + 3) & ~3)) == NULL) { 437 /* XXXJRT new stat, please */ 438 IP_STATINC(IP_STAT_TOOSMALL); 439 return; 440 } 441 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 442 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 443 IP_STATINC(IP_STAT_TOOSMALL); 444 return; 445 } 446 } 447 ip = mtod(m, struct ip *); 448 if (ip->ip_v != IPVERSION) { 449 IP_STATINC(IP_STAT_BADVERS); 450 goto bad; 451 } 452 hlen = ip->ip_hl << 2; 453 if (hlen < sizeof(struct ip)) { /* minimum header length */ 454 IP_STATINC(IP_STAT_BADHLEN); 455 goto bad; 456 } 457 if (hlen > m->m_len) { 458 if ((m = m_pullup(m, hlen)) == 0) { 459 IP_STATINC(IP_STAT_BADHLEN); 460 return; 461 } 462 ip = mtod(m, struct ip *); 463 } 464 465 /* 466 * RFC1122: packets with a multicast source address are 467 * not allowed. 468 */ 469 if (IN_MULTICAST(ip->ip_src.s_addr)) { 470 IP_STATINC(IP_STAT_BADADDR); 471 goto bad; 472 } 473 474 /* 127/8 must not appear on wire - RFC1122 */ 475 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 476 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 477 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 478 IP_STATINC(IP_STAT_BADADDR); 479 goto bad; 480 } 481 } 482 483 switch (m->m_pkthdr.csum_flags & 484 ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) | 485 M_CSUM_IPv4_BAD)) { 486 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 487 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 488 goto badcsum; 489 490 case M_CSUM_IPv4: 491 /* Checksum was okay. */ 492 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 493 break; 494 495 default: 496 /* 497 * Must compute it ourselves. Maybe skip checksum on 498 * loopback interfaces. 499 */ 500 if (__predict_true(!(m->m_pkthdr.rcvif->if_flags & 501 IFF_LOOPBACK) || ip_do_loopback_cksum)) { 502 INET_CSUM_COUNTER_INCR(&ip_swcsum); 503 if (in_cksum(m, hlen) != 0) 504 goto badcsum; 505 } 506 break; 507 } 508 509 /* Retrieve the packet length. */ 510 len = ntohs(ip->ip_len); 511 512 /* 513 * Check for additional length bogosity 514 */ 515 if (len < hlen) { 516 IP_STATINC(IP_STAT_BADLEN); 517 goto bad; 518 } 519 520 /* 521 * Check that the amount of data in the buffers 522 * is as at least much as the IP header would have us expect. 523 * Trim mbufs if longer than we expect. 524 * Drop packet if shorter than we expect. 525 */ 526 if (m->m_pkthdr.len < len) { 527 IP_STATINC(IP_STAT_TOOSHORT); 528 goto bad; 529 } 530 if (m->m_pkthdr.len > len) { 531 if (m->m_len == m->m_pkthdr.len) { 532 m->m_len = len; 533 m->m_pkthdr.len = len; 534 } else 535 m_adj(m, len - m->m_pkthdr.len); 536 } 537 538 #if defined(IPSEC) 539 /* ipflow (IP fast forwarding) is not compatible with IPsec. */ 540 m->m_flags &= ~M_CANFASTFWD; 541 #else 542 /* 543 * Assume that we can create a fast-forward IP flow entry 544 * based on this packet. 545 */ 546 m->m_flags |= M_CANFASTFWD; 547 #endif 548 549 #ifdef PFIL_HOOKS 550 /* 551 * Run through list of hooks for input packets. If there are any 552 * filters which require that additional packets in the flow are 553 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 554 * Note that filters must _never_ set this flag, as another filter 555 * in the list may have previously cleared it. 556 */ 557 /* 558 * let ipfilter look at packet on the wire, 559 * not the decapsulated packet. 560 */ 561 #ifdef IPSEC 562 if (!ipsec_getnhist(m)) 563 #elif defined(FAST_IPSEC) 564 if (!ipsec_indone(m)) 565 #else 566 if (1) 567 #endif 568 { 569 struct in_addr odst; 570 571 odst = ip->ip_dst; 572 if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif, 573 PFIL_IN) != 0) 574 return; 575 if (m == NULL) 576 return; 577 ip = mtod(m, struct ip *); 578 hlen = ip->ip_hl << 2; 579 /* 580 * XXX The setting of "srcrt" here is to prevent ip_forward() 581 * from generating ICMP redirects for packets that have 582 * been redirected by a hook back out on to the same LAN that 583 * they came from and is not an indication that the packet 584 * is being inffluenced by source routing options. This 585 * allows things like 586 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 587 * where tlp0 is both on the 1.1.1.0/24 network and is the 588 * default route for hosts on 1.1.1.0/24. Of course this 589 * also requires a "map tlp0 ..." to complete the story. 590 * One might argue whether or not this kind of network config. 591 * should be supported in this manner... 592 */ 593 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 594 } 595 #endif /* PFIL_HOOKS */ 596 597 #ifdef ALTQ 598 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 599 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) { 600 /* packet dropped by traffic conditioner */ 601 return; 602 } 603 #endif 604 605 /* 606 * Process options and, if not destined for us, 607 * ship it on. ip_dooptions returns 1 when an 608 * error was detected (causing an icmp message 609 * to be sent and the original packet to be freed). 610 */ 611 ip_nhops = 0; /* for source routed packets */ 612 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 613 return; 614 615 /* 616 * Enable a consistency check between the destination address 617 * and the arrival interface for a unicast packet (the RFC 1122 618 * strong ES model) if IP forwarding is disabled and the packet 619 * is not locally generated. 620 * 621 * XXX - Checking also should be disabled if the destination 622 * address is ipnat'ed to a different interface. 623 * 624 * XXX - Checking is incompatible with IP aliases added 625 * to the loopback interface instead of the interface where 626 * the packets are received. 627 * 628 * XXX - We need to add a per ifaddr flag for this so that 629 * we get finer grain control. 630 */ 631 checkif = ip_checkinterface && (ipforwarding == 0) && 632 (m->m_pkthdr.rcvif != NULL) && 633 ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0); 634 635 /* 636 * Check our list of addresses, to see if the packet is for us. 637 * 638 * Traditional 4.4BSD did not consult IFF_UP at all. 639 * The behavior here is to treat addresses on !IFF_UP interface 640 * as not mine. 641 */ 642 downmatch = 0; 643 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 644 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 645 if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif) 646 continue; 647 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 648 break; 649 else 650 downmatch++; 651 } 652 } 653 if (ia != NULL) 654 goto ours; 655 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 656 IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) { 657 if (ifa->ifa_addr->sa_family != AF_INET) 658 continue; 659 ia = ifatoia(ifa); 660 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 661 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 662 /* 663 * Look for all-0's host part (old broadcast addr), 664 * either for subnet or net. 665 */ 666 ip->ip_dst.s_addr == ia->ia_subnet || 667 ip->ip_dst.s_addr == ia->ia_net) 668 goto ours; 669 /* 670 * An interface with IP address zero accepts 671 * all packets that arrive on that interface. 672 */ 673 if (in_nullhost(ia->ia_addr.sin_addr)) 674 goto ours; 675 } 676 } 677 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 678 struct in_multi *inm; 679 #ifdef MROUTING 680 extern struct socket *ip_mrouter; 681 682 if (ip_mrouter) { 683 /* 684 * If we are acting as a multicast router, all 685 * incoming multicast packets are passed to the 686 * kernel-level multicast forwarding function. 687 * The packet is returned (relatively) intact; if 688 * ip_mforward() returns a non-zero value, the packet 689 * must be discarded, else it may be accepted below. 690 * 691 * (The IP ident field is put in the same byte order 692 * as expected when ip_mforward() is called from 693 * ip_output().) 694 */ 695 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 696 IP_STATINC(IP_STAT_CANTFORWARD); 697 m_freem(m); 698 return; 699 } 700 701 /* 702 * The process-level routing demon needs to receive 703 * all multicast IGMP packets, whether or not this 704 * host belongs to their destination groups. 705 */ 706 if (ip->ip_p == IPPROTO_IGMP) 707 goto ours; 708 IP_STATINC(IP_STAT_CANTFORWARD); 709 } 710 #endif 711 /* 712 * See if we belong to the destination multicast group on the 713 * arrival interface. 714 */ 715 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 716 if (inm == NULL) { 717 IP_STATINC(IP_STAT_CANTFORWARD); 718 m_freem(m); 719 return; 720 } 721 goto ours; 722 } 723 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 724 in_nullhost(ip->ip_dst)) 725 goto ours; 726 727 /* 728 * Not for us; forward if possible and desirable. 729 */ 730 if (ipforwarding == 0) { 731 IP_STATINC(IP_STAT_CANTFORWARD); 732 m_freem(m); 733 } else { 734 /* 735 * If ip_dst matched any of my address on !IFF_UP interface, 736 * and there's no IFF_UP interface that matches ip_dst, 737 * send icmp unreach. Forwarding it will result in in-kernel 738 * forwarding loop till TTL goes to 0. 739 */ 740 if (downmatch) { 741 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 742 IP_STATINC(IP_STAT_CANTFORWARD); 743 return; 744 } 745 #ifdef IPSEC 746 if (ipsec4_in_reject(m, NULL)) { 747 IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); 748 goto bad; 749 } 750 #endif 751 #ifdef FAST_IPSEC 752 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 753 s = splsoftnet(); 754 if (mtag != NULL) { 755 tdbi = (struct tdb_ident *)(mtag + 1); 756 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 757 } else { 758 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 759 IP_FORWARDING, &error); 760 } 761 if (sp == NULL) { /* NB: can happen if error */ 762 splx(s); 763 /*XXX error stat???*/ 764 DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/ 765 goto bad; 766 } 767 768 /* 769 * Check security policy against packet attributes. 770 */ 771 error = ipsec_in_reject(sp, m); 772 KEY_FREESP(&sp); 773 splx(s); 774 if (error) { 775 IP_STATINC(IP_STAT_CANTFORWARD); 776 goto bad; 777 } 778 779 /* 780 * Peek at the outbound SP for this packet to determine if 781 * it's a Fast Forward candidate. 782 */ 783 mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL); 784 if (mtag != NULL) 785 m->m_flags &= ~M_CANFASTFWD; 786 else { 787 s = splsoftnet(); 788 sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND, 789 (IP_FORWARDING | 790 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 791 &error, NULL); 792 if (sp != NULL) { 793 m->m_flags &= ~M_CANFASTFWD; 794 KEY_FREESP(&sp); 795 } 796 splx(s); 797 } 798 #endif /* FAST_IPSEC */ 799 800 ip_forward(m, srcrt); 801 } 802 return; 803 804 ours: 805 /* 806 * If offset or IP_MF are set, must reassemble. 807 */ 808 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 809 struct mbuf *m_final; 810 u_int off, flen; 811 bool mff; 812 813 /* 814 * Prevent TCP blind data attacks by not allowing non-initial 815 * fragments to start at less than 68 bytes (minimal fragment 816 * size) and making sure the first fragment is at least 68 817 * bytes. 818 */ 819 off = (ntohs(ip->ip_off) & IP_OFFMASK) << 3; 820 if ((off > 0 ? off + hlen : len) < IP_MINFRAGSIZE - 1) { 821 IP_STATINC(IP_STAT_BADFRAGS); 822 goto bad; 823 } 824 825 /* Fragment length and MF flag. */ 826 flen = ntohs(ip->ip_len) - hlen; 827 mff = (ip->ip_off & htons(IP_MF)) != 0; 828 if (mff) { 829 /* 830 * Make sure that fragments have a data length 831 * which is non-zero and multiple of 8 bytes. 832 */ 833 if (flen == 0 || (flen & 0x7) != 0) { 834 IP_STATINC(IP_STAT_BADFRAGS); 835 goto bad; 836 } 837 } 838 839 /* 840 * Adjust total IP length to not reflect header and convert 841 * offset of this to bytes. XXX: clobbers struct ip. 842 */ 843 ip->ip_len = htons(flen); 844 ip->ip_off = htons(off); 845 846 /* 847 * Pass to IP reassembly mechanism. 848 */ 849 if (ip_reass_packet(m, ip, mff, &m_final) != 0) { 850 /* Failed; invalid fragment(s) or packet. */ 851 goto bad; 852 } 853 if (m_final == NULL) { 854 /* More fragments should come; silently return. */ 855 return; 856 } 857 /* Reassembly is done, we have the final packet. */ 858 m = m_final; 859 860 /* Updated local variable(s). */ 861 ip = mtod(m, struct ip *); 862 hlen = ip->ip_hl << 2; 863 } 864 865 #if defined(IPSEC) 866 /* 867 * enforce IPsec policy checking if we are seeing last header. 868 * note that we do not visit this with protocols with pcb layer 869 * code - like udp/tcp/raw ip. 870 */ 871 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 && 872 ipsec4_in_reject(m, NULL)) { 873 IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); 874 goto bad; 875 } 876 #endif 877 #ifdef FAST_IPSEC 878 /* 879 * enforce IPsec policy checking if we are seeing last header. 880 * note that we do not visit this with protocols with pcb layer 881 * code - like udp/tcp/raw ip. 882 */ 883 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 884 /* 885 * Check if the packet has already had IPsec processing 886 * done. If so, then just pass it along. This tag gets 887 * set during AH, ESP, etc. input handling, before the 888 * packet is returned to the ip input queue for delivery. 889 */ 890 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 891 s = splsoftnet(); 892 if (mtag != NULL) { 893 tdbi = (struct tdb_ident *)(mtag + 1); 894 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 895 } else { 896 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 897 IP_FORWARDING, &error); 898 } 899 if (sp != NULL) { 900 /* 901 * Check security policy against packet attributes. 902 */ 903 error = ipsec_in_reject(sp, m); 904 KEY_FREESP(&sp); 905 } else { 906 /* XXX error stat??? */ 907 error = EINVAL; 908 DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/ 909 } 910 splx(s); 911 if (error) 912 goto bad; 913 } 914 #endif /* FAST_IPSEC */ 915 916 /* 917 * Switch out to protocol's input routine. 918 */ 919 #if IFA_STATS 920 if (ia && ip) 921 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 922 #endif 923 IP_STATINC(IP_STAT_DELIVERED); 924 { 925 int off = hlen, nh = ip->ip_p; 926 927 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 928 return; 929 } 930 bad: 931 m_freem(m); 932 return; 933 934 badcsum: 935 IP_STATINC(IP_STAT_BADSUM); 936 m_freem(m); 937 } 938 939 /* 940 * IP timer processing. 941 */ 942 void 943 ip_slowtimo(void) 944 { 945 946 mutex_enter(softnet_lock); 947 KERNEL_LOCK(1, NULL); 948 949 ip_reass_slowtimo(); 950 951 KERNEL_UNLOCK_ONE(NULL); 952 mutex_exit(softnet_lock); 953 } 954 955 /* 956 * IP drain processing. 957 */ 958 void 959 ip_drain(void) 960 { 961 962 KERNEL_LOCK(1, NULL); 963 ip_reass_drain(); 964 KERNEL_UNLOCK_ONE(NULL); 965 } 966 967 /* 968 * Do option processing on a datagram, 969 * possibly discarding it if bad options are encountered, 970 * or forwarding it if source-routed. 971 * Returns 1 if packet has been forwarded/freed, 972 * 0 if the packet should be processed further. 973 */ 974 int 975 ip_dooptions(struct mbuf *m) 976 { 977 struct ip *ip = mtod(m, struct ip *); 978 u_char *cp, *cp0; 979 struct ip_timestamp *ipt; 980 struct in_ifaddr *ia; 981 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 982 struct in_addr dst; 983 n_time ntime; 984 985 dst = ip->ip_dst; 986 cp = (u_char *)(ip + 1); 987 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 988 for (; cnt > 0; cnt -= optlen, cp += optlen) { 989 opt = cp[IPOPT_OPTVAL]; 990 if (opt == IPOPT_EOL) 991 break; 992 if (opt == IPOPT_NOP) 993 optlen = 1; 994 else { 995 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 996 code = &cp[IPOPT_OLEN] - (u_char *)ip; 997 goto bad; 998 } 999 optlen = cp[IPOPT_OLEN]; 1000 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 1001 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1002 goto bad; 1003 } 1004 } 1005 switch (opt) { 1006 1007 default: 1008 break; 1009 1010 /* 1011 * Source routing with record. 1012 * Find interface with current destination address. 1013 * If none on this machine then drop if strictly routed, 1014 * or do nothing if loosely routed. 1015 * Record interface address and bring up next address 1016 * component. If strictly routed make sure next 1017 * address is on directly accessible net. 1018 */ 1019 case IPOPT_LSRR: 1020 case IPOPT_SSRR: 1021 if (ip_allowsrcrt == 0) { 1022 type = ICMP_UNREACH; 1023 code = ICMP_UNREACH_NET_PROHIB; 1024 goto bad; 1025 } 1026 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1027 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1028 goto bad; 1029 } 1030 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1031 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1032 goto bad; 1033 } 1034 ipaddr.sin_addr = ip->ip_dst; 1035 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1036 if (ia == 0) { 1037 if (opt == IPOPT_SSRR) { 1038 type = ICMP_UNREACH; 1039 code = ICMP_UNREACH_SRCFAIL; 1040 goto bad; 1041 } 1042 /* 1043 * Loose routing, and not at next destination 1044 * yet; nothing to do except forward. 1045 */ 1046 break; 1047 } 1048 off--; /* 0 origin */ 1049 if ((off + sizeof(struct in_addr)) > optlen) { 1050 /* 1051 * End of source route. Should be for us. 1052 */ 1053 save_rte(cp, ip->ip_src); 1054 break; 1055 } 1056 /* 1057 * locate outgoing interface 1058 */ 1059 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 1060 sizeof(ipaddr.sin_addr)); 1061 if (opt == IPOPT_SSRR) 1062 ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); 1063 else 1064 ia = ip_rtaddr(ipaddr.sin_addr); 1065 if (ia == 0) { 1066 type = ICMP_UNREACH; 1067 code = ICMP_UNREACH_SRCFAIL; 1068 goto bad; 1069 } 1070 ip->ip_dst = ipaddr.sin_addr; 1071 bcopy((void *)&ia->ia_addr.sin_addr, 1072 (void *)(cp + off), sizeof(struct in_addr)); 1073 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1074 /* 1075 * Let ip_intr's mcast routing check handle mcast pkts 1076 */ 1077 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1078 break; 1079 1080 case IPOPT_RR: 1081 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1082 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1083 goto bad; 1084 } 1085 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1086 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1087 goto bad; 1088 } 1089 /* 1090 * If no space remains, ignore. 1091 */ 1092 off--; /* 0 origin */ 1093 if ((off + sizeof(struct in_addr)) > optlen) 1094 break; 1095 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 1096 sizeof(ipaddr.sin_addr)); 1097 /* 1098 * locate outgoing interface; if we're the destination, 1099 * use the incoming interface (should be same). 1100 */ 1101 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 1102 == NULL && 1103 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 1104 type = ICMP_UNREACH; 1105 code = ICMP_UNREACH_HOST; 1106 goto bad; 1107 } 1108 bcopy((void *)&ia->ia_addr.sin_addr, 1109 (void *)(cp + off), sizeof(struct in_addr)); 1110 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1111 break; 1112 1113 case IPOPT_TS: 1114 code = cp - (u_char *)ip; 1115 ipt = (struct ip_timestamp *)cp; 1116 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1117 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1118 goto bad; 1119 } 1120 if (ipt->ipt_ptr < 5) { 1121 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1122 goto bad; 1123 } 1124 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1125 if (++ipt->ipt_oflw == 0) { 1126 code = (u_char *)&ipt->ipt_ptr - 1127 (u_char *)ip; 1128 goto bad; 1129 } 1130 break; 1131 } 1132 cp0 = (cp + ipt->ipt_ptr - 1); 1133 switch (ipt->ipt_flg) { 1134 1135 case IPOPT_TS_TSONLY: 1136 break; 1137 1138 case IPOPT_TS_TSANDADDR: 1139 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1140 sizeof(struct in_addr) > ipt->ipt_len) { 1141 code = (u_char *)&ipt->ipt_ptr - 1142 (u_char *)ip; 1143 goto bad; 1144 } 1145 ipaddr.sin_addr = dst; 1146 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1147 m->m_pkthdr.rcvif)); 1148 if (ia == 0) 1149 continue; 1150 bcopy(&ia->ia_addr.sin_addr, 1151 cp0, sizeof(struct in_addr)); 1152 ipt->ipt_ptr += sizeof(struct in_addr); 1153 break; 1154 1155 case IPOPT_TS_PRESPEC: 1156 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1157 sizeof(struct in_addr) > ipt->ipt_len) { 1158 code = (u_char *)&ipt->ipt_ptr - 1159 (u_char *)ip; 1160 goto bad; 1161 } 1162 memcpy(&ipaddr.sin_addr, cp0, 1163 sizeof(struct in_addr)); 1164 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1165 == NULL) 1166 continue; 1167 ipt->ipt_ptr += sizeof(struct in_addr); 1168 break; 1169 1170 default: 1171 /* XXX can't take &ipt->ipt_flg */ 1172 code = (u_char *)&ipt->ipt_ptr - 1173 (u_char *)ip + 1; 1174 goto bad; 1175 } 1176 ntime = iptime(); 1177 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1178 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1179 sizeof(n_time)); 1180 ipt->ipt_ptr += sizeof(n_time); 1181 } 1182 } 1183 if (forward) { 1184 if (ip_forwsrcrt == 0) { 1185 type = ICMP_UNREACH; 1186 code = ICMP_UNREACH_SRCFAIL; 1187 goto bad; 1188 } 1189 ip_forward(m, 1); 1190 return (1); 1191 } 1192 return (0); 1193 bad: 1194 icmp_error(m, type, code, 0, 0); 1195 IP_STATINC(IP_STAT_BADOPTIONS); 1196 return (1); 1197 } 1198 1199 /* 1200 * Given address of next destination (final or next hop), 1201 * return internet address info of interface to be used to get there. 1202 */ 1203 struct in_ifaddr * 1204 ip_rtaddr(struct in_addr dst) 1205 { 1206 struct rtentry *rt; 1207 union { 1208 struct sockaddr dst; 1209 struct sockaddr_in dst4; 1210 } u; 1211 1212 sockaddr_in_init(&u.dst4, &dst, 0); 1213 1214 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) 1215 return NULL; 1216 1217 return ifatoia(rt->rt_ifa); 1218 } 1219 1220 /* 1221 * Save incoming source route for use in replies, 1222 * to be picked up later by ip_srcroute if the receiver is interested. 1223 */ 1224 void 1225 save_rte(u_char *option, struct in_addr dst) 1226 { 1227 unsigned olen; 1228 1229 olen = option[IPOPT_OLEN]; 1230 #ifdef DIAGNOSTIC 1231 if (ipprintfs) 1232 printf("save_rte: olen %d\n", olen); 1233 #endif /* 0 */ 1234 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1235 return; 1236 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1237 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1238 ip_srcrt.dst = dst; 1239 } 1240 1241 /* 1242 * Retrieve incoming source route for use in replies, 1243 * in the same form used by setsockopt. 1244 * The first hop is placed before the options, will be removed later. 1245 */ 1246 struct mbuf * 1247 ip_srcroute(void) 1248 { 1249 struct in_addr *p, *q; 1250 struct mbuf *m; 1251 1252 if (ip_nhops == 0) 1253 return NULL; 1254 m = m_get(M_DONTWAIT, MT_SOOPTS); 1255 if (m == 0) 1256 return NULL; 1257 1258 MCLAIM(m, &inetdomain.dom_mowner); 1259 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1260 1261 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1262 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1263 OPTSIZ; 1264 #ifdef DIAGNOSTIC 1265 if (ipprintfs) 1266 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1267 #endif 1268 1269 /* 1270 * First save first hop for return route 1271 */ 1272 p = &ip_srcrt.route[ip_nhops - 1]; 1273 *(mtod(m, struct in_addr *)) = *p--; 1274 #ifdef DIAGNOSTIC 1275 if (ipprintfs) 1276 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1277 #endif 1278 1279 /* 1280 * Copy option fields and padding (nop) to mbuf. 1281 */ 1282 ip_srcrt.nop = IPOPT_NOP; 1283 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1284 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1285 OPTSIZ); 1286 q = (struct in_addr *)(mtod(m, char *) + 1287 sizeof(struct in_addr) + OPTSIZ); 1288 #undef OPTSIZ 1289 /* 1290 * Record return path as an IP source route, 1291 * reversing the path (pointers are now aligned). 1292 */ 1293 while (p >= ip_srcrt.route) { 1294 #ifdef DIAGNOSTIC 1295 if (ipprintfs) 1296 printf(" %x", ntohl(q->s_addr)); 1297 #endif 1298 *q++ = *p--; 1299 } 1300 /* 1301 * Last hop goes to final destination. 1302 */ 1303 *q = ip_srcrt.dst; 1304 #ifdef DIAGNOSTIC 1305 if (ipprintfs) 1306 printf(" %x\n", ntohl(q->s_addr)); 1307 #endif 1308 return (m); 1309 } 1310 1311 const int inetctlerrmap[PRC_NCMDS] = { 1312 [PRC_MSGSIZE] = EMSGSIZE, 1313 [PRC_HOSTDEAD] = EHOSTDOWN, 1314 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1315 [PRC_UNREACH_NET] = EHOSTUNREACH, 1316 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1317 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1318 [PRC_UNREACH_PORT] = ECONNREFUSED, 1319 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1320 [PRC_PARAMPROB] = ENOPROTOOPT, 1321 }; 1322 1323 /* 1324 * Forward a packet. If some error occurs return the sender 1325 * an icmp packet. Note we can't always generate a meaningful 1326 * icmp message because icmp doesn't have a large enough repertoire 1327 * of codes and types. 1328 * 1329 * If not forwarding, just drop the packet. This could be confusing 1330 * if ipforwarding was zero but some routing protocol was advancing 1331 * us as a gateway to somewhere. However, we must let the routing 1332 * protocol deal with that. 1333 * 1334 * The srcrt parameter indicates whether the packet is being forwarded 1335 * via a source route. 1336 */ 1337 void 1338 ip_forward(struct mbuf *m, int srcrt) 1339 { 1340 struct ip *ip = mtod(m, struct ip *); 1341 struct rtentry *rt; 1342 int error, type = 0, code = 0, destmtu = 0; 1343 struct mbuf *mcopy; 1344 n_long dest; 1345 union { 1346 struct sockaddr dst; 1347 struct sockaddr_in dst4; 1348 } u; 1349 1350 /* 1351 * We are now in the output path. 1352 */ 1353 MCLAIM(m, &ip_tx_mowner); 1354 1355 /* 1356 * Clear any in-bound checksum flags for this packet. 1357 */ 1358 m->m_pkthdr.csum_flags = 0; 1359 1360 dest = 0; 1361 #ifdef DIAGNOSTIC 1362 if (ipprintfs) { 1363 printf("forward: src %s ", inet_ntoa(ip->ip_src)); 1364 printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl); 1365 } 1366 #endif 1367 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1368 IP_STATINC(IP_STAT_CANTFORWARD); 1369 m_freem(m); 1370 return; 1371 } 1372 if (ip->ip_ttl <= IPTTLDEC) { 1373 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1374 return; 1375 } 1376 1377 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1378 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1379 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1380 return; 1381 } 1382 1383 /* 1384 * Save at most 68 bytes of the packet in case 1385 * we need to generate an ICMP message to the src. 1386 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1387 */ 1388 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1389 if (mcopy) 1390 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1391 1392 ip->ip_ttl -= IPTTLDEC; 1393 1394 /* 1395 * If forwarding packet using same interface that it came in on, 1396 * perhaps should send a redirect to sender to shortcut a hop. 1397 * Only send redirect if source is sending directly to us, 1398 * and if packet was not source routed (or has any options). 1399 * Also, don't send redirect if forwarding using a default route 1400 * or a route modified by a redirect. 1401 */ 1402 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1403 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1404 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1405 ipsendredirects && !srcrt) { 1406 if (rt->rt_ifa && 1407 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1408 ifatoia(rt->rt_ifa)->ia_subnet) { 1409 if (rt->rt_flags & RTF_GATEWAY) 1410 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1411 else 1412 dest = ip->ip_dst.s_addr; 1413 /* 1414 * Router requirements says to only send host 1415 * redirects. 1416 */ 1417 type = ICMP_REDIRECT; 1418 code = ICMP_REDIRECT_HOST; 1419 #ifdef DIAGNOSTIC 1420 if (ipprintfs) 1421 printf("redirect (%d) to %x\n", code, 1422 (u_int32_t)dest); 1423 #endif 1424 } 1425 } 1426 1427 error = ip_output(m, NULL, &ipforward_rt, 1428 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1429 (struct ip_moptions *)NULL, (struct socket *)NULL); 1430 1431 if (error) 1432 IP_STATINC(IP_STAT_CANTFORWARD); 1433 else { 1434 uint64_t *ips = IP_STAT_GETREF(); 1435 ips[IP_STAT_FORWARD]++; 1436 if (type) { 1437 ips[IP_STAT_REDIRECTSENT]++; 1438 IP_STAT_PUTREF(); 1439 } else { 1440 IP_STAT_PUTREF(); 1441 if (mcopy) { 1442 #ifdef GATEWAY 1443 if (mcopy->m_flags & M_CANFASTFWD) 1444 ipflow_create(&ipforward_rt, mcopy); 1445 #endif 1446 m_freem(mcopy); 1447 } 1448 return; 1449 } 1450 } 1451 if (mcopy == NULL) 1452 return; 1453 1454 switch (error) { 1455 1456 case 0: /* forwarded, but need redirect */ 1457 /* type, code set above */ 1458 break; 1459 1460 case ENETUNREACH: /* shouldn't happen, checked above */ 1461 case EHOSTUNREACH: 1462 case ENETDOWN: 1463 case EHOSTDOWN: 1464 default: 1465 type = ICMP_UNREACH; 1466 code = ICMP_UNREACH_HOST; 1467 break; 1468 1469 case EMSGSIZE: 1470 type = ICMP_UNREACH; 1471 code = ICMP_UNREACH_NEEDFRAG; 1472 1473 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1474 destmtu = rt->rt_ifp->if_mtu; 1475 1476 #if defined(IPSEC) || defined(FAST_IPSEC) 1477 { 1478 /* 1479 * If the packet is routed over IPsec tunnel, tell the 1480 * originator the tunnel MTU. 1481 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz 1482 * XXX quickhack!!! 1483 */ 1484 1485 struct secpolicy *sp; 1486 int ipsecerror; 1487 size_t ipsechdr; 1488 struct route *ro; 1489 1490 sp = ipsec4_getpolicybyaddr(mcopy, 1491 IPSEC_DIR_OUTBOUND, IP_FORWARDING, 1492 &ipsecerror); 1493 1494 if (sp != NULL) { 1495 /* count IPsec header size */ 1496 ipsechdr = ipsec4_hdrsiz(mcopy, 1497 IPSEC_DIR_OUTBOUND, NULL); 1498 1499 /* 1500 * find the correct route for outer IPv4 1501 * header, compute tunnel MTU. 1502 */ 1503 1504 if (sp->req != NULL 1505 && sp->req->sav != NULL 1506 && sp->req->sav->sah != NULL) { 1507 ro = &sp->req->sav->sah->sa_route; 1508 rt = rtcache_validate(ro); 1509 if (rt && rt->rt_ifp) { 1510 destmtu = 1511 rt->rt_rmx.rmx_mtu ? 1512 rt->rt_rmx.rmx_mtu : 1513 rt->rt_ifp->if_mtu; 1514 destmtu -= ipsechdr; 1515 } 1516 } 1517 1518 #ifdef IPSEC 1519 key_freesp(sp); 1520 #else 1521 KEY_FREESP(&sp); 1522 #endif 1523 } 1524 } 1525 #endif /*defined(IPSEC) || defined(FAST_IPSEC)*/ 1526 IP_STATINC(IP_STAT_CANTFRAG); 1527 break; 1528 1529 case ENOBUFS: 1530 #if 1 1531 /* 1532 * a router should not generate ICMP_SOURCEQUENCH as 1533 * required in RFC1812 Requirements for IP Version 4 Routers. 1534 * source quench could be a big problem under DoS attacks, 1535 * or if the underlying interface is rate-limited. 1536 */ 1537 if (mcopy) 1538 m_freem(mcopy); 1539 return; 1540 #else 1541 type = ICMP_SOURCEQUENCH; 1542 code = 0; 1543 break; 1544 #endif 1545 } 1546 icmp_error(mcopy, type, code, dest, destmtu); 1547 } 1548 1549 void 1550 ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1551 struct mbuf *m) 1552 { 1553 1554 if (inp->inp_socket->so_options & SO_TIMESTAMP 1555 #ifdef SO_OTIMESTAMP 1556 || inp->inp_socket->so_options & SO_OTIMESTAMP 1557 #endif 1558 ) { 1559 struct timeval tv; 1560 1561 microtime(&tv); 1562 #ifdef SO_OTIMESTAMP 1563 if (inp->inp_socket->so_options & SO_OTIMESTAMP) { 1564 struct timeval50 tv50; 1565 timeval_to_timeval50(&tv, &tv50); 1566 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1567 SCM_OTIMESTAMP, SOL_SOCKET); 1568 } else 1569 #endif 1570 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1571 SCM_TIMESTAMP, SOL_SOCKET); 1572 if (*mp) 1573 mp = &(*mp)->m_next; 1574 } 1575 if (inp->inp_flags & INP_RECVDSTADDR) { 1576 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1577 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1578 if (*mp) 1579 mp = &(*mp)->m_next; 1580 } 1581 #ifdef notyet 1582 /* 1583 * XXX 1584 * Moving these out of udp_input() made them even more broken 1585 * than they already were. 1586 * - fenner@parc.xerox.com 1587 */ 1588 /* options were tossed already */ 1589 if (inp->inp_flags & INP_RECVOPTS) { 1590 *mp = sbcreatecontrol((void *) opts_deleted_above, 1591 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1592 if (*mp) 1593 mp = &(*mp)->m_next; 1594 } 1595 /* ip_srcroute doesn't do what we want here, need to fix */ 1596 if (inp->inp_flags & INP_RECVRETOPTS) { 1597 *mp = sbcreatecontrol((void *) ip_srcroute(), 1598 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1599 if (*mp) 1600 mp = &(*mp)->m_next; 1601 } 1602 #endif 1603 if (inp->inp_flags & INP_RECVIF) { 1604 struct sockaddr_dl sdl; 1605 1606 sockaddr_dl_init(&sdl, sizeof(sdl), 1607 (m->m_pkthdr.rcvif != NULL) 1608 ? m->m_pkthdr.rcvif->if_index 1609 : 0, 1610 0, NULL, 0, NULL, 0); 1611 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1612 if (*mp) 1613 mp = &(*mp)->m_next; 1614 } 1615 if (inp->inp_flags & INP_RECVTTL) { 1616 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1617 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1618 if (*mp) 1619 mp = &(*mp)->m_next; 1620 } 1621 } 1622 1623 /* 1624 * sysctl helper routine for net.inet.ip.forwsrcrt. 1625 */ 1626 static int 1627 sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1628 { 1629 int error, tmp; 1630 struct sysctlnode node; 1631 1632 node = *rnode; 1633 tmp = ip_forwsrcrt; 1634 node.sysctl_data = &tmp; 1635 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1636 if (error || newp == NULL) 1637 return (error); 1638 1639 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1640 0, NULL, NULL, NULL); 1641 if (error) 1642 return (error); 1643 1644 ip_forwsrcrt = tmp; 1645 1646 return (0); 1647 } 1648 1649 /* 1650 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1651 * range of the new value and tweaks timers if it changes. 1652 */ 1653 static int 1654 sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1655 { 1656 int error, tmp; 1657 struct sysctlnode node; 1658 1659 node = *rnode; 1660 tmp = ip_mtudisc_timeout; 1661 node.sysctl_data = &tmp; 1662 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1663 if (error || newp == NULL) 1664 return (error); 1665 if (tmp < 0) 1666 return (EINVAL); 1667 1668 mutex_enter(softnet_lock); 1669 1670 ip_mtudisc_timeout = tmp; 1671 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1672 1673 mutex_exit(softnet_lock); 1674 1675 return (0); 1676 } 1677 1678 #ifdef GATEWAY 1679 /* 1680 * sysctl helper routine for net.inet.ip.maxflows. 1681 */ 1682 static int 1683 sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS) 1684 { 1685 int error; 1686 1687 error = sysctl_lookup(SYSCTLFN_CALL(rnode)); 1688 if (error || newp == NULL) 1689 return (error); 1690 1691 mutex_enter(softnet_lock); 1692 KERNEL_LOCK(1, NULL); 1693 1694 ipflow_prune(); 1695 1696 KERNEL_UNLOCK_ONE(NULL); 1697 mutex_exit(softnet_lock); 1698 1699 return (0); 1700 } 1701 1702 static int 1703 sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS) 1704 { 1705 int error, tmp; 1706 struct sysctlnode node; 1707 1708 node = *rnode; 1709 tmp = ip_hashsize; 1710 node.sysctl_data = &tmp; 1711 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1712 if (error || newp == NULL) 1713 return (error); 1714 1715 if ((tmp & (tmp - 1)) == 0 && tmp != 0) { 1716 /* 1717 * Can only fail due to malloc() 1718 */ 1719 mutex_enter(softnet_lock); 1720 KERNEL_LOCK(1, NULL); 1721 1722 error = ipflow_invalidate_all(tmp); 1723 1724 KERNEL_UNLOCK_ONE(NULL); 1725 mutex_exit(softnet_lock); 1726 1727 } else { 1728 /* 1729 * EINVAL if not a power of 2 1730 */ 1731 error = EINVAL; 1732 } 1733 1734 return error; 1735 } 1736 #endif /* GATEWAY */ 1737 1738 static int 1739 sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1740 { 1741 1742 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1743 } 1744 1745 static void 1746 sysctl_net_inet_ip_setup(struct sysctllog **clog) 1747 { 1748 extern int subnetsarelocal, hostzeroisbroadcast; 1749 1750 sysctl_createv(clog, 0, NULL, NULL, 1751 CTLFLAG_PERMANENT, 1752 CTLTYPE_NODE, "net", NULL, 1753 NULL, 0, NULL, 0, 1754 CTL_NET, CTL_EOL); 1755 sysctl_createv(clog, 0, NULL, NULL, 1756 CTLFLAG_PERMANENT, 1757 CTLTYPE_NODE, "inet", 1758 SYSCTL_DESCR("PF_INET related settings"), 1759 NULL, 0, NULL, 0, 1760 CTL_NET, PF_INET, CTL_EOL); 1761 sysctl_createv(clog, 0, NULL, NULL, 1762 CTLFLAG_PERMANENT, 1763 CTLTYPE_NODE, "ip", 1764 SYSCTL_DESCR("IPv4 related settings"), 1765 NULL, 0, NULL, 0, 1766 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1767 1768 sysctl_createv(clog, 0, NULL, NULL, 1769 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1770 CTLTYPE_INT, "forwarding", 1771 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1772 NULL, 0, &ipforwarding, 0, 1773 CTL_NET, PF_INET, IPPROTO_IP, 1774 IPCTL_FORWARDING, CTL_EOL); 1775 sysctl_createv(clog, 0, NULL, NULL, 1776 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1777 CTLTYPE_INT, "redirect", 1778 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1779 NULL, 0, &ipsendredirects, 0, 1780 CTL_NET, PF_INET, IPPROTO_IP, 1781 IPCTL_SENDREDIRECTS, CTL_EOL); 1782 sysctl_createv(clog, 0, NULL, NULL, 1783 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1784 CTLTYPE_INT, "ttl", 1785 SYSCTL_DESCR("Default TTL for an INET datagram"), 1786 NULL, 0, &ip_defttl, 0, 1787 CTL_NET, PF_INET, IPPROTO_IP, 1788 IPCTL_DEFTTL, CTL_EOL); 1789 #ifdef IPCTL_DEFMTU 1790 sysctl_createv(clog, 0, NULL, NULL, 1791 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1792 CTLTYPE_INT, "mtu", 1793 SYSCTL_DESCR("Default MTA for an INET route"), 1794 NULL, 0, &ip_mtu, 0, 1795 CTL_NET, PF_INET, IPPROTO_IP, 1796 IPCTL_DEFMTU, CTL_EOL); 1797 #endif /* IPCTL_DEFMTU */ 1798 sysctl_createv(clog, 0, NULL, NULL, 1799 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1800 CTLTYPE_INT, "forwsrcrt", 1801 SYSCTL_DESCR("Enable forwarding of source-routed " 1802 "datagrams"), 1803 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1804 CTL_NET, PF_INET, IPPROTO_IP, 1805 IPCTL_FORWSRCRT, CTL_EOL); 1806 sysctl_createv(clog, 0, NULL, NULL, 1807 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1808 CTLTYPE_INT, "directed-broadcast", 1809 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1810 NULL, 0, &ip_directedbcast, 0, 1811 CTL_NET, PF_INET, IPPROTO_IP, 1812 IPCTL_DIRECTEDBCAST, CTL_EOL); 1813 sysctl_createv(clog, 0, NULL, NULL, 1814 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1815 CTLTYPE_INT, "allowsrcrt", 1816 SYSCTL_DESCR("Accept source-routed datagrams"), 1817 NULL, 0, &ip_allowsrcrt, 0, 1818 CTL_NET, PF_INET, IPPROTO_IP, 1819 IPCTL_ALLOWSRCRT, CTL_EOL); 1820 sysctl_createv(clog, 0, NULL, NULL, 1821 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1822 CTLTYPE_INT, "subnetsarelocal", 1823 SYSCTL_DESCR("Whether logical subnets are considered " 1824 "local"), 1825 NULL, 0, &subnetsarelocal, 0, 1826 CTL_NET, PF_INET, IPPROTO_IP, 1827 IPCTL_SUBNETSARELOCAL, CTL_EOL); 1828 sysctl_createv(clog, 0, NULL, NULL, 1829 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1830 CTLTYPE_INT, "mtudisc", 1831 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1832 NULL, 0, &ip_mtudisc, 0, 1833 CTL_NET, PF_INET, IPPROTO_IP, 1834 IPCTL_MTUDISC, CTL_EOL); 1835 sysctl_createv(clog, 0, NULL, NULL, 1836 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1837 CTLTYPE_INT, "anonportmin", 1838 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1839 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1840 CTL_NET, PF_INET, IPPROTO_IP, 1841 IPCTL_ANONPORTMIN, CTL_EOL); 1842 sysctl_createv(clog, 0, NULL, NULL, 1843 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1844 CTLTYPE_INT, "anonportmax", 1845 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1846 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1847 CTL_NET, PF_INET, IPPROTO_IP, 1848 IPCTL_ANONPORTMAX, CTL_EOL); 1849 sysctl_createv(clog, 0, NULL, NULL, 1850 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1851 CTLTYPE_INT, "mtudisctimeout", 1852 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1853 sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0, 1854 CTL_NET, PF_INET, IPPROTO_IP, 1855 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1856 #ifdef GATEWAY 1857 sysctl_createv(clog, 0, NULL, NULL, 1858 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1859 CTLTYPE_INT, "maxflows", 1860 SYSCTL_DESCR("Number of flows for fast forwarding"), 1861 sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0, 1862 CTL_NET, PF_INET, IPPROTO_IP, 1863 IPCTL_MAXFLOWS, CTL_EOL); 1864 sysctl_createv(clog, 0, NULL, NULL, 1865 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1866 CTLTYPE_INT, "hashsize", 1867 SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"), 1868 sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0, 1869 CTL_NET, PF_INET, IPPROTO_IP, 1870 CTL_CREATE, CTL_EOL); 1871 #endif /* GATEWAY */ 1872 sysctl_createv(clog, 0, NULL, NULL, 1873 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1874 CTLTYPE_INT, "hostzerobroadcast", 1875 SYSCTL_DESCR("All zeroes address is broadcast address"), 1876 NULL, 0, &hostzeroisbroadcast, 0, 1877 CTL_NET, PF_INET, IPPROTO_IP, 1878 IPCTL_HOSTZEROBROADCAST, CTL_EOL); 1879 #if NGIF > 0 1880 sysctl_createv(clog, 0, NULL, NULL, 1881 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1882 CTLTYPE_INT, "gifttl", 1883 SYSCTL_DESCR("Default TTL for a gif tunnel datagram"), 1884 NULL, 0, &ip_gif_ttl, 0, 1885 CTL_NET, PF_INET, IPPROTO_IP, 1886 IPCTL_GIF_TTL, CTL_EOL); 1887 #endif /* NGIF */ 1888 #ifndef IPNOPRIVPORTS 1889 sysctl_createv(clog, 0, NULL, NULL, 1890 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1891 CTLTYPE_INT, "lowportmin", 1892 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1893 "to assign"), 1894 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1895 CTL_NET, PF_INET, IPPROTO_IP, 1896 IPCTL_LOWPORTMIN, CTL_EOL); 1897 sysctl_createv(clog, 0, NULL, NULL, 1898 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1899 CTLTYPE_INT, "lowportmax", 1900 SYSCTL_DESCR("Highest privileged ephemeral port number " 1901 "to assign"), 1902 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1903 CTL_NET, PF_INET, IPPROTO_IP, 1904 IPCTL_LOWPORTMAX, CTL_EOL); 1905 #endif /* IPNOPRIVPORTS */ 1906 #if NGRE > 0 1907 sysctl_createv(clog, 0, NULL, NULL, 1908 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1909 CTLTYPE_INT, "grettl", 1910 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1911 NULL, 0, &ip_gre_ttl, 0, 1912 CTL_NET, PF_INET, IPPROTO_IP, 1913 IPCTL_GRE_TTL, CTL_EOL); 1914 #endif /* NGRE */ 1915 sysctl_createv(clog, 0, NULL, NULL, 1916 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1917 CTLTYPE_INT, "checkinterface", 1918 SYSCTL_DESCR("Enable receive side of Strong ES model " 1919 "from RFC1122"), 1920 NULL, 0, &ip_checkinterface, 0, 1921 CTL_NET, PF_INET, IPPROTO_IP, 1922 IPCTL_CHECKINTERFACE, CTL_EOL); 1923 sysctl_createv(clog, 0, NULL, NULL, 1924 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1925 CTLTYPE_INT, "random_id", 1926 SYSCTL_DESCR("Assign random ip_id values"), 1927 NULL, 0, &ip_do_randomid, 0, 1928 CTL_NET, PF_INET, IPPROTO_IP, 1929 IPCTL_RANDOMID, CTL_EOL); 1930 sysctl_createv(clog, 0, NULL, NULL, 1931 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1932 CTLTYPE_INT, "do_loopback_cksum", 1933 SYSCTL_DESCR("Perform IP checksum on loopback"), 1934 NULL, 0, &ip_do_loopback_cksum, 0, 1935 CTL_NET, PF_INET, IPPROTO_IP, 1936 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1937 sysctl_createv(clog, 0, NULL, NULL, 1938 CTLFLAG_PERMANENT, 1939 CTLTYPE_STRUCT, "stats", 1940 SYSCTL_DESCR("IP statistics"), 1941 sysctl_net_inet_ip_stats, 0, NULL, 0, 1942 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1943 CTL_EOL); 1944 } 1945 1946 void 1947 ip_statinc(u_int stat) 1948 { 1949 1950 KASSERT(stat < IP_NSTATS); 1951 IP_STATINC(stat); 1952 } 1953