ip_input.c revision 1.292
1/* $NetBSD: ip_input.c,v 1.292 2010/12/11 22:37:46 matt 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 1.292 2010/12/11 22:37:46 matt 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 */ 192int ipforwarding = IPFORWARDING; 193int ipsendredirects = IPSENDREDIRECTS; 194int ip_defttl = IPDEFTTL; 195int ip_forwsrcrt = IPFORWSRCRT; 196int ip_directedbcast = IPDIRECTEDBCAST; 197int ip_allowsrcrt = IPALLOWSRCRT; 198int ip_mtudisc = IPMTUDISC; 199int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 200#ifdef DIAGNOSTIC 201int ipprintfs = 0; 202#endif 203 204int 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 */ 219int ip_checkinterface = 0; 220 221 222struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 223 224int ipqmaxlen = IFQ_MAXLEN; 225u_long in_ifaddrhash; /* size of hash table - 1 */ 226int in_ifaddrentries; /* total number of addrs */ 227struct in_ifaddrhead in_ifaddrhead; 228struct in_ifaddrhashhead *in_ifaddrhashtbl; 229u_long in_multihash; /* size of hash table - 1 */ 230int in_multientries; /* total number of addrs */ 231struct in_multihashhead *in_multihashtbl; 232struct ifqueue ipintrq; 233 234ipid_state_t * ip_ids; 235uint16_t ip_id; 236 237percpu_t *ipstat_percpu; 238 239#ifdef PFIL_HOOKS 240struct pfil_head inet_pfil_hook; 241#endif 242 243struct pool inmulti_pool; 244 245#ifdef INET_CSUM_COUNTERS 246#include <sys/device.h> 247 248struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 249 NULL, "inet", "hwcsum bad"); 250struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 251 NULL, "inet", "hwcsum ok"); 252struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 253 NULL, "inet", "swcsum"); 254 255#define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 256 257EVCNT_ATTACH_STATIC(ip_hwcsum_bad); 258EVCNT_ATTACH_STATIC(ip_hwcsum_ok); 259EVCNT_ATTACH_STATIC(ip_swcsum); 260 261#else 262 263#define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 264 265#endif /* INET_CSUM_COUNTERS */ 266 267/* 268 * We need to save the IP options in case a protocol wants to respond 269 * to an incoming packet over the same route if the packet got here 270 * using IP source routing. This allows connection establishment and 271 * maintenance when the remote end is on a network that is not known 272 * to us. 273 */ 274int ip_nhops = 0; 275static struct ip_srcrt { 276 struct in_addr dst; /* final destination */ 277 char nop; /* one NOP to align */ 278 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 279 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 280} ip_srcrt; 281 282static void save_rte(u_char *, struct in_addr); 283 284#ifdef MBUFTRACE 285struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 286struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 287#endif 288 289static void sysctl_net_inet_ip_setup(struct sysctllog **); 290 291/* 292 * IP initialization: fill in IP protocol switch table. 293 * All protocols not implemented in kernel go to raw IP protocol handler. 294 */ 295void 296ip_init(void) 297{ 298 const struct protosw *pr; 299 int i; 300 301 sysctl_net_inet_ip_setup(NULL); 302 303 pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", 304 NULL, IPL_SOFTNET); 305 306 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 307 if (pr == 0) 308 panic("ip_init"); 309 for (i = 0; i < IPPROTO_MAX; i++) 310 ip_protox[i] = pr - inetsw; 311 for (pr = inetdomain.dom_protosw; 312 pr < inetdomain.dom_protoswNPROTOSW; pr++) 313 if (pr->pr_domain->dom_family == PF_INET && 314 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 315 ip_protox[pr->pr_protocol] = pr - inetsw; 316 317 ip_reass_init(); 318 319 ip_ids = ip_id_init(); 320 ip_id = time_second & 0xfffff; 321 322 ipintrq.ifq_maxlen = ipqmaxlen; 323 324 TAILQ_INIT(&in_ifaddrhead); 325 in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true, 326 &in_ifaddrhash); 327 in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true, 328 &in_multihash); 329 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 330#ifdef GATEWAY 331 ipflow_init(ip_hashsize); 332#endif 333 334#ifdef PFIL_HOOKS 335 /* Register our Packet Filter hook. */ 336 inet_pfil_hook.ph_type = PFIL_TYPE_AF; 337 inet_pfil_hook.ph_af = AF_INET; 338 i = pfil_head_register(&inet_pfil_hook); 339 if (i != 0) 340 printf("ip_init: WARNING: unable to register pfil hook, " 341 "error %d\n", i); 342#endif /* PFIL_HOOKS */ 343 344#ifdef MBUFTRACE 345 MOWNER_ATTACH(&ip_tx_mowner); 346 MOWNER_ATTACH(&ip_rx_mowner); 347#endif /* MBUFTRACE */ 348 349 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 350} 351 352struct sockaddr_in ipaddr = { 353 .sin_len = sizeof(ipaddr), 354 .sin_family = AF_INET, 355}; 356struct route ipforward_rt; 357 358/* 359 * IP software interrupt routine 360 */ 361void 362ipintr(void) 363{ 364 int s; 365 struct mbuf *m; 366 struct ifqueue lcl_intrq; 367 368 memset(&lcl_intrq, 0, sizeof(lcl_intrq)); 369 ipintrq.ifq_maxlen = ipqmaxlen; 370 371 mutex_enter(softnet_lock); 372 KERNEL_LOCK(1, NULL); 373 if (!IF_IS_EMPTY(&ipintrq)) { 374 s = splnet(); 375 376 /* Take existing queue onto stack */ 377 lcl_intrq = ipintrq; 378 379 /* Zero out global queue, preserving maxlen and drops */ 380 ipintrq.ifq_head = NULL; 381 ipintrq.ifq_tail = NULL; 382 ipintrq.ifq_len = 0; 383 ipintrq.ifq_maxlen = lcl_intrq.ifq_maxlen; 384 ipintrq.ifq_drops = lcl_intrq.ifq_drops; 385 386 splx(s); 387 } 388 KERNEL_UNLOCK_ONE(NULL); 389 while (!IF_IS_EMPTY(&lcl_intrq)) { 390 IF_DEQUEUE(&lcl_intrq, m); 391 if (m == NULL) 392 break; 393 KERNEL_UNLOCK_ONE(NULL); 394 ip_input(m); 395 KERNEL_LOCK(1, NULL); 396 } 397 mutex_exit(softnet_lock); 398} 399 400/* 401 * Ip input routine. Checksum and byte swap header. If fragmented 402 * try to reassemble. Process options. Pass to next level. 403 */ 404void 405ip_input(struct mbuf *m) 406{ 407 struct ip *ip = NULL; 408 struct in_ifaddr *ia; 409 struct ifaddr *ifa; 410 int hlen = 0, len; 411 int downmatch; 412 int checkif; 413 int srcrt = 0; 414#ifdef FAST_IPSEC 415 struct m_tag *mtag; 416 struct tdb_ident *tdbi; 417 struct secpolicy *sp; 418 int error, s; 419#endif /* FAST_IPSEC */ 420 421 MCLAIM(m, &ip_rx_mowner); 422 KASSERT((m->m_flags & M_PKTHDR) != 0); 423 424 /* 425 * If no IP addresses have been set yet but the interfaces 426 * are receiving, can't do anything with incoming packets yet. 427 */ 428 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 429 goto bad; 430 IP_STATINC(IP_STAT_TOTAL); 431 /* 432 * If the IP header is not aligned, slurp it up into a new 433 * mbuf with space for link headers, in the event we forward 434 * it. Otherwise, if it is aligned, make sure the entire 435 * base IP header is in the first mbuf of the chain. 436 */ 437 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 438 if ((m = m_copyup(m, sizeof(struct ip), 439 (max_linkhdr + 3) & ~3)) == NULL) { 440 /* XXXJRT new stat, please */ 441 IP_STATINC(IP_STAT_TOOSMALL); 442 return; 443 } 444 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 445 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 446 IP_STATINC(IP_STAT_TOOSMALL); 447 return; 448 } 449 } 450 ip = mtod(m, struct ip *); 451 if (ip->ip_v != IPVERSION) { 452 IP_STATINC(IP_STAT_BADVERS); 453 goto bad; 454 } 455 hlen = ip->ip_hl << 2; 456 if (hlen < sizeof(struct ip)) { /* minimum header length */ 457 IP_STATINC(IP_STAT_BADHLEN); 458 goto bad; 459 } 460 if (hlen > m->m_len) { 461 if ((m = m_pullup(m, hlen)) == 0) { 462 IP_STATINC(IP_STAT_BADHLEN); 463 return; 464 } 465 ip = mtod(m, struct ip *); 466 } 467 468 /* 469 * RFC1122: packets with a multicast source address are 470 * not allowed. 471 */ 472 if (IN_MULTICAST(ip->ip_src.s_addr)) { 473 IP_STATINC(IP_STAT_BADADDR); 474 goto bad; 475 } 476 477 /* 127/8 must not appear on wire - RFC1122 */ 478 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 479 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 480 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 481 IP_STATINC(IP_STAT_BADADDR); 482 goto bad; 483 } 484 } 485 486 switch (m->m_pkthdr.csum_flags & 487 ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) | 488 M_CSUM_IPv4_BAD)) { 489 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 490 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 491 goto badcsum; 492 493 case M_CSUM_IPv4: 494 /* Checksum was okay. */ 495 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 496 break; 497 498 default: 499 /* 500 * Must compute it ourselves. Maybe skip checksum on 501 * loopback interfaces. 502 */ 503 if (__predict_true(!(m->m_pkthdr.rcvif->if_flags & 504 IFF_LOOPBACK) || ip_do_loopback_cksum)) { 505 INET_CSUM_COUNTER_INCR(&ip_swcsum); 506 if (in_cksum(m, hlen) != 0) 507 goto badcsum; 508 } 509 break; 510 } 511 512 /* Retrieve the packet length. */ 513 len = ntohs(ip->ip_len); 514 515 /* 516 * Check for additional length bogosity 517 */ 518 if (len < hlen) { 519 IP_STATINC(IP_STAT_BADLEN); 520 goto bad; 521 } 522 523 /* 524 * Check that the amount of data in the buffers 525 * is as at least much as the IP header would have us expect. 526 * Trim mbufs if longer than we expect. 527 * Drop packet if shorter than we expect. 528 */ 529 if (m->m_pkthdr.len < len) { 530 IP_STATINC(IP_STAT_TOOSHORT); 531 goto bad; 532 } 533 if (m->m_pkthdr.len > len) { 534 if (m->m_len == m->m_pkthdr.len) { 535 m->m_len = len; 536 m->m_pkthdr.len = len; 537 } else 538 m_adj(m, len - m->m_pkthdr.len); 539 } 540 541#if defined(IPSEC) 542 /* ipflow (IP fast forwarding) is not compatible with IPsec. */ 543 m->m_flags &= ~M_CANFASTFWD; 544#else 545 /* 546 * Assume that we can create a fast-forward IP flow entry 547 * based on this packet. 548 */ 549 m->m_flags |= M_CANFASTFWD; 550#endif 551 552#ifdef PFIL_HOOKS 553 /* 554 * Run through list of hooks for input packets. If there are any 555 * filters which require that additional packets in the flow are 556 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 557 * Note that filters must _never_ set this flag, as another filter 558 * in the list may have previously cleared it. 559 */ 560 /* 561 * let ipfilter look at packet on the wire, 562 * not the decapsulated packet. 563 */ 564#ifdef IPSEC 565 if (!ipsec_getnhist(m)) 566#elif defined(FAST_IPSEC) 567 if (!ipsec_indone(m)) 568#else 569 if (1) 570#endif 571 { 572 struct in_addr odst; 573 574 odst = ip->ip_dst; 575 if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif, 576 PFIL_IN) != 0) 577 return; 578 if (m == NULL) 579 return; 580 ip = mtod(m, struct ip *); 581 hlen = ip->ip_hl << 2; 582 /* 583 * XXX The setting of "srcrt" here is to prevent ip_forward() 584 * from generating ICMP redirects for packets that have 585 * been redirected by a hook back out on to the same LAN that 586 * they came from and is not an indication that the packet 587 * is being inffluenced by source routing options. This 588 * allows things like 589 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 590 * where tlp0 is both on the 1.1.1.0/24 network and is the 591 * default route for hosts on 1.1.1.0/24. Of course this 592 * also requires a "map tlp0 ..." to complete the story. 593 * One might argue whether or not this kind of network config. 594 * should be supported in this manner... 595 */ 596 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 597 } 598#endif /* PFIL_HOOKS */ 599 600#ifdef ALTQ 601 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 602 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) { 603 /* packet dropped by traffic conditioner */ 604 return; 605 } 606#endif 607 608 /* 609 * Process options and, if not destined for us, 610 * ship it on. ip_dooptions returns 1 when an 611 * error was detected (causing an icmp message 612 * to be sent and the original packet to be freed). 613 */ 614 ip_nhops = 0; /* for source routed packets */ 615 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 616 return; 617 618 /* 619 * Enable a consistency check between the destination address 620 * and the arrival interface for a unicast packet (the RFC 1122 621 * strong ES model) if IP forwarding is disabled and the packet 622 * is not locally generated. 623 * 624 * XXX - Checking also should be disabled if the destination 625 * address is ipnat'ed to a different interface. 626 * 627 * XXX - Checking is incompatible with IP aliases added 628 * to the loopback interface instead of the interface where 629 * the packets are received. 630 * 631 * XXX - We need to add a per ifaddr flag for this so that 632 * we get finer grain control. 633 */ 634 checkif = ip_checkinterface && (ipforwarding == 0) && 635 (m->m_pkthdr.rcvif != NULL) && 636 ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0); 637 638 /* 639 * Check our list of addresses, to see if the packet is for us. 640 * 641 * Traditional 4.4BSD did not consult IFF_UP at all. 642 * The behavior here is to treat addresses on !IFF_UP interface 643 * as not mine. 644 */ 645 downmatch = 0; 646 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 647 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 648 if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif) 649 continue; 650 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 651 break; 652 else 653 downmatch++; 654 } 655 } 656 if (ia != NULL) 657 goto ours; 658 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 659 IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) { 660 if (ifa->ifa_addr->sa_family != AF_INET) 661 continue; 662 ia = ifatoia(ifa); 663 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 664 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 665 /* 666 * Look for all-0's host part (old broadcast addr), 667 * either for subnet or net. 668 */ 669 ip->ip_dst.s_addr == ia->ia_subnet || 670 ip->ip_dst.s_addr == ia->ia_net) 671 goto ours; 672 /* 673 * An interface with IP address zero accepts 674 * all packets that arrive on that interface. 675 */ 676 if (in_nullhost(ia->ia_addr.sin_addr)) 677 goto ours; 678 } 679 } 680 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 681 struct in_multi *inm; 682#ifdef MROUTING 683 extern struct socket *ip_mrouter; 684 685 if (ip_mrouter) { 686 /* 687 * If we are acting as a multicast router, all 688 * incoming multicast packets are passed to the 689 * kernel-level multicast forwarding function. 690 * The packet is returned (relatively) intact; if 691 * ip_mforward() returns a non-zero value, the packet 692 * must be discarded, else it may be accepted below. 693 * 694 * (The IP ident field is put in the same byte order 695 * as expected when ip_mforward() is called from 696 * ip_output().) 697 */ 698 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 699 IP_STATINC(IP_STAT_CANTFORWARD); 700 m_freem(m); 701 return; 702 } 703 704 /* 705 * The process-level routing demon needs to receive 706 * all multicast IGMP packets, whether or not this 707 * host belongs to their destination groups. 708 */ 709 if (ip->ip_p == IPPROTO_IGMP) 710 goto ours; 711 IP_STATINC(IP_STAT_CANTFORWARD); 712 } 713#endif 714 /* 715 * See if we belong to the destination multicast group on the 716 * arrival interface. 717 */ 718 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 719 if (inm == NULL) { 720 IP_STATINC(IP_STAT_CANTFORWARD); 721 m_freem(m); 722 return; 723 } 724 goto ours; 725 } 726 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 727 in_nullhost(ip->ip_dst)) 728 goto ours; 729 730 /* 731 * Not for us; forward if possible and desirable. 732 */ 733 if (ipforwarding == 0) { 734 IP_STATINC(IP_STAT_CANTFORWARD); 735 m_freem(m); 736 } else { 737 /* 738 * If ip_dst matched any of my address on !IFF_UP interface, 739 * and there's no IFF_UP interface that matches ip_dst, 740 * send icmp unreach. Forwarding it will result in in-kernel 741 * forwarding loop till TTL goes to 0. 742 */ 743 if (downmatch) { 744 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 745 IP_STATINC(IP_STAT_CANTFORWARD); 746 return; 747 } 748#ifdef IPSEC 749 if (ipsec4_in_reject(m, NULL)) { 750 IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); 751 goto bad; 752 } 753#endif 754#ifdef FAST_IPSEC 755 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 756 s = splsoftnet(); 757 if (mtag != NULL) { 758 tdbi = (struct tdb_ident *)(mtag + 1); 759 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 760 } else { 761 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 762 IP_FORWARDING, &error); 763 } 764 if (sp == NULL) { /* NB: can happen if error */ 765 splx(s); 766 /*XXX error stat???*/ 767 DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/ 768 goto bad; 769 } 770 771 /* 772 * Check security policy against packet attributes. 773 */ 774 error = ipsec_in_reject(sp, m); 775 KEY_FREESP(&sp); 776 splx(s); 777 if (error) { 778 IP_STATINC(IP_STAT_CANTFORWARD); 779 goto bad; 780 } 781 782 /* 783 * Peek at the outbound SP for this packet to determine if 784 * it's a Fast Forward candidate. 785 */ 786 mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL); 787 if (mtag != NULL) 788 m->m_flags &= ~M_CANFASTFWD; 789 else { 790 s = splsoftnet(); 791 sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND, 792 (IP_FORWARDING | 793 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 794 &error, NULL); 795 if (sp != NULL) { 796 m->m_flags &= ~M_CANFASTFWD; 797 KEY_FREESP(&sp); 798 } 799 splx(s); 800 } 801#endif /* FAST_IPSEC */ 802 803 ip_forward(m, srcrt); 804 } 805 return; 806 807ours: 808 /* 809 * If offset or IP_MF are set, must reassemble. 810 */ 811 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 812 /* 813 * Pass to IP reassembly mechanism. 814 */ 815 if (ip_reass_packet(&m, ip) != 0) { 816 /* Failed; invalid fragment(s) or packet. */ 817 goto bad; 818 } 819 if (m == NULL) { 820 /* More fragments should come; silently return. */ 821 return; 822 } 823 /* 824 * Reassembly is done, we have the final packet. 825 * Updated cached data in local variable(s). 826 */ 827 ip = mtod(m, struct ip *); 828 hlen = ip->ip_hl << 2; 829 } 830 831#if defined(IPSEC) 832 /* 833 * enforce IPsec policy checking if we are seeing last header. 834 * note that we do not visit this with protocols with pcb layer 835 * code - like udp/tcp/raw ip. 836 */ 837 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 && 838 ipsec4_in_reject(m, NULL)) { 839 IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); 840 goto bad; 841 } 842#endif 843#ifdef FAST_IPSEC 844 /* 845 * enforce IPsec policy checking if we are seeing last header. 846 * note that we do not visit this with protocols with pcb layer 847 * code - like udp/tcp/raw ip. 848 */ 849 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 850 /* 851 * Check if the packet has already had IPsec processing 852 * done. If so, then just pass it along. This tag gets 853 * set during AH, ESP, etc. input handling, before the 854 * packet is returned to the ip input queue for delivery. 855 */ 856 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 857 s = splsoftnet(); 858 if (mtag != NULL) { 859 tdbi = (struct tdb_ident *)(mtag + 1); 860 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 861 } else { 862 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 863 IP_FORWARDING, &error); 864 } 865 if (sp != NULL) { 866 /* 867 * Check security policy against packet attributes. 868 */ 869 error = ipsec_in_reject(sp, m); 870 KEY_FREESP(&sp); 871 } else { 872 /* XXX error stat??? */ 873 error = EINVAL; 874DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/ 875 } 876 splx(s); 877 if (error) 878 goto bad; 879 } 880#endif /* FAST_IPSEC */ 881 882 /* 883 * Switch out to protocol's input routine. 884 */ 885#if IFA_STATS 886 if (ia && ip) 887 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 888#endif 889 IP_STATINC(IP_STAT_DELIVERED); 890 { 891 int off = hlen, nh = ip->ip_p; 892 893 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 894 return; 895 } 896bad: 897 m_freem(m); 898 return; 899 900badcsum: 901 IP_STATINC(IP_STAT_BADSUM); 902 m_freem(m); 903} 904 905/* 906 * IP timer processing. 907 */ 908void 909ip_slowtimo(void) 910{ 911 912 mutex_enter(softnet_lock); 913 KERNEL_LOCK(1, NULL); 914 915 ip_reass_slowtimo(); 916 917 KERNEL_UNLOCK_ONE(NULL); 918 mutex_exit(softnet_lock); 919} 920 921/* 922 * IP drain processing. 923 */ 924void 925ip_drain(void) 926{ 927 928 KERNEL_LOCK(1, NULL); 929 ip_reass_drain(); 930 KERNEL_UNLOCK_ONE(NULL); 931} 932 933/* 934 * Do option processing on a datagram, 935 * possibly discarding it if bad options are encountered, 936 * or forwarding it if source-routed. 937 * Returns 1 if packet has been forwarded/freed, 938 * 0 if the packet should be processed further. 939 */ 940int 941ip_dooptions(struct mbuf *m) 942{ 943 struct ip *ip = mtod(m, struct ip *); 944 u_char *cp, *cp0; 945 struct ip_timestamp *ipt; 946 struct in_ifaddr *ia; 947 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 948 struct in_addr dst; 949 n_time ntime; 950 951 dst = ip->ip_dst; 952 cp = (u_char *)(ip + 1); 953 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 954 for (; cnt > 0; cnt -= optlen, cp += optlen) { 955 opt = cp[IPOPT_OPTVAL]; 956 if (opt == IPOPT_EOL) 957 break; 958 if (opt == IPOPT_NOP) 959 optlen = 1; 960 else { 961 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 962 code = &cp[IPOPT_OLEN] - (u_char *)ip; 963 goto bad; 964 } 965 optlen = cp[IPOPT_OLEN]; 966 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 967 code = &cp[IPOPT_OLEN] - (u_char *)ip; 968 goto bad; 969 } 970 } 971 switch (opt) { 972 973 default: 974 break; 975 976 /* 977 * Source routing with record. 978 * Find interface with current destination address. 979 * If none on this machine then drop if strictly routed, 980 * or do nothing if loosely routed. 981 * Record interface address and bring up next address 982 * component. If strictly routed make sure next 983 * address is on directly accessible net. 984 */ 985 case IPOPT_LSRR: 986 case IPOPT_SSRR: 987 if (ip_allowsrcrt == 0) { 988 type = ICMP_UNREACH; 989 code = ICMP_UNREACH_NET_PROHIB; 990 goto bad; 991 } 992 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 993 code = &cp[IPOPT_OLEN] - (u_char *)ip; 994 goto bad; 995 } 996 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 997 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 998 goto bad; 999 } 1000 ipaddr.sin_addr = ip->ip_dst; 1001 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1002 if (ia == 0) { 1003 if (opt == IPOPT_SSRR) { 1004 type = ICMP_UNREACH; 1005 code = ICMP_UNREACH_SRCFAIL; 1006 goto bad; 1007 } 1008 /* 1009 * Loose routing, and not at next destination 1010 * yet; nothing to do except forward. 1011 */ 1012 break; 1013 } 1014 off--; /* 0 origin */ 1015 if ((off + sizeof(struct in_addr)) > optlen) { 1016 /* 1017 * End of source route. Should be for us. 1018 */ 1019 save_rte(cp, ip->ip_src); 1020 break; 1021 } 1022 /* 1023 * locate outgoing interface 1024 */ 1025 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 1026 sizeof(ipaddr.sin_addr)); 1027 if (opt == IPOPT_SSRR) 1028 ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); 1029 else 1030 ia = ip_rtaddr(ipaddr.sin_addr); 1031 if (ia == 0) { 1032 type = ICMP_UNREACH; 1033 code = ICMP_UNREACH_SRCFAIL; 1034 goto bad; 1035 } 1036 ip->ip_dst = ipaddr.sin_addr; 1037 bcopy((void *)&ia->ia_addr.sin_addr, 1038 (void *)(cp + off), sizeof(struct in_addr)); 1039 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1040 /* 1041 * Let ip_intr's mcast routing check handle mcast pkts 1042 */ 1043 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1044 break; 1045 1046 case IPOPT_RR: 1047 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1048 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1049 goto bad; 1050 } 1051 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1052 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1053 goto bad; 1054 } 1055 /* 1056 * If no space remains, ignore. 1057 */ 1058 off--; /* 0 origin */ 1059 if ((off + sizeof(struct in_addr)) > optlen) 1060 break; 1061 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 1062 sizeof(ipaddr.sin_addr)); 1063 /* 1064 * locate outgoing interface; if we're the destination, 1065 * use the incoming interface (should be same). 1066 */ 1067 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 1068 == NULL && 1069 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 1070 type = ICMP_UNREACH; 1071 code = ICMP_UNREACH_HOST; 1072 goto bad; 1073 } 1074 bcopy((void *)&ia->ia_addr.sin_addr, 1075 (void *)(cp + off), sizeof(struct in_addr)); 1076 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1077 break; 1078 1079 case IPOPT_TS: 1080 code = cp - (u_char *)ip; 1081 ipt = (struct ip_timestamp *)cp; 1082 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1083 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1084 goto bad; 1085 } 1086 if (ipt->ipt_ptr < 5) { 1087 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1088 goto bad; 1089 } 1090 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1091 if (++ipt->ipt_oflw == 0) { 1092 code = (u_char *)&ipt->ipt_ptr - 1093 (u_char *)ip; 1094 goto bad; 1095 } 1096 break; 1097 } 1098 cp0 = (cp + ipt->ipt_ptr - 1); 1099 switch (ipt->ipt_flg) { 1100 1101 case IPOPT_TS_TSONLY: 1102 break; 1103 1104 case IPOPT_TS_TSANDADDR: 1105 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1106 sizeof(struct in_addr) > ipt->ipt_len) { 1107 code = (u_char *)&ipt->ipt_ptr - 1108 (u_char *)ip; 1109 goto bad; 1110 } 1111 ipaddr.sin_addr = dst; 1112 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1113 m->m_pkthdr.rcvif)); 1114 if (ia == 0) 1115 continue; 1116 bcopy(&ia->ia_addr.sin_addr, 1117 cp0, sizeof(struct in_addr)); 1118 ipt->ipt_ptr += sizeof(struct in_addr); 1119 break; 1120 1121 case IPOPT_TS_PRESPEC: 1122 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1123 sizeof(struct in_addr) > ipt->ipt_len) { 1124 code = (u_char *)&ipt->ipt_ptr - 1125 (u_char *)ip; 1126 goto bad; 1127 } 1128 memcpy(&ipaddr.sin_addr, cp0, 1129 sizeof(struct in_addr)); 1130 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1131 == NULL) 1132 continue; 1133 ipt->ipt_ptr += sizeof(struct in_addr); 1134 break; 1135 1136 default: 1137 /* XXX can't take &ipt->ipt_flg */ 1138 code = (u_char *)&ipt->ipt_ptr - 1139 (u_char *)ip + 1; 1140 goto bad; 1141 } 1142 ntime = iptime(); 1143 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1144 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1145 sizeof(n_time)); 1146 ipt->ipt_ptr += sizeof(n_time); 1147 } 1148 } 1149 if (forward) { 1150 if (ip_forwsrcrt == 0) { 1151 type = ICMP_UNREACH; 1152 code = ICMP_UNREACH_SRCFAIL; 1153 goto bad; 1154 } 1155 ip_forward(m, 1); 1156 return (1); 1157 } 1158 return (0); 1159bad: 1160 icmp_error(m, type, code, 0, 0); 1161 IP_STATINC(IP_STAT_BADOPTIONS); 1162 return (1); 1163} 1164 1165/* 1166 * Given address of next destination (final or next hop), 1167 * return internet address info of interface to be used to get there. 1168 */ 1169struct in_ifaddr * 1170ip_rtaddr(struct in_addr dst) 1171{ 1172 struct rtentry *rt; 1173 union { 1174 struct sockaddr dst; 1175 struct sockaddr_in dst4; 1176 } u; 1177 1178 sockaddr_in_init(&u.dst4, &dst, 0); 1179 1180 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) 1181 return NULL; 1182 1183 return ifatoia(rt->rt_ifa); 1184} 1185 1186/* 1187 * Save incoming source route for use in replies, 1188 * to be picked up later by ip_srcroute if the receiver is interested. 1189 */ 1190void 1191save_rte(u_char *option, struct in_addr dst) 1192{ 1193 unsigned olen; 1194 1195 olen = option[IPOPT_OLEN]; 1196#ifdef DIAGNOSTIC 1197 if (ipprintfs) 1198 printf("save_rte: olen %d\n", olen); 1199#endif /* 0 */ 1200 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1201 return; 1202 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1203 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1204 ip_srcrt.dst = dst; 1205} 1206 1207/* 1208 * Retrieve incoming source route for use in replies, 1209 * in the same form used by setsockopt. 1210 * The first hop is placed before the options, will be removed later. 1211 */ 1212struct mbuf * 1213ip_srcroute(void) 1214{ 1215 struct in_addr *p, *q; 1216 struct mbuf *m; 1217 1218 if (ip_nhops == 0) 1219 return NULL; 1220 m = m_get(M_DONTWAIT, MT_SOOPTS); 1221 if (m == 0) 1222 return NULL; 1223 1224 MCLAIM(m, &inetdomain.dom_mowner); 1225#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1226 1227 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1228 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1229 OPTSIZ; 1230#ifdef DIAGNOSTIC 1231 if (ipprintfs) 1232 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1233#endif 1234 1235 /* 1236 * First save first hop for return route 1237 */ 1238 p = &ip_srcrt.route[ip_nhops - 1]; 1239 *(mtod(m, struct in_addr *)) = *p--; 1240#ifdef DIAGNOSTIC 1241 if (ipprintfs) 1242 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1243#endif 1244 1245 /* 1246 * Copy option fields and padding (nop) to mbuf. 1247 */ 1248 ip_srcrt.nop = IPOPT_NOP; 1249 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1250 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1251 OPTSIZ); 1252 q = (struct in_addr *)(mtod(m, char *) + 1253 sizeof(struct in_addr) + OPTSIZ); 1254#undef OPTSIZ 1255 /* 1256 * Record return path as an IP source route, 1257 * reversing the path (pointers are now aligned). 1258 */ 1259 while (p >= ip_srcrt.route) { 1260#ifdef DIAGNOSTIC 1261 if (ipprintfs) 1262 printf(" %x", ntohl(q->s_addr)); 1263#endif 1264 *q++ = *p--; 1265 } 1266 /* 1267 * Last hop goes to final destination. 1268 */ 1269 *q = ip_srcrt.dst; 1270#ifdef DIAGNOSTIC 1271 if (ipprintfs) 1272 printf(" %x\n", ntohl(q->s_addr)); 1273#endif 1274 return (m); 1275} 1276 1277const int inetctlerrmap[PRC_NCMDS] = { 1278 [PRC_MSGSIZE] = EMSGSIZE, 1279 [PRC_HOSTDEAD] = EHOSTDOWN, 1280 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1281 [PRC_UNREACH_NET] = EHOSTUNREACH, 1282 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1283 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1284 [PRC_UNREACH_PORT] = ECONNREFUSED, 1285 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1286 [PRC_PARAMPROB] = ENOPROTOOPT, 1287}; 1288 1289/* 1290 * Forward a packet. If some error occurs return the sender 1291 * an icmp packet. Note we can't always generate a meaningful 1292 * icmp message because icmp doesn't have a large enough repertoire 1293 * of codes and types. 1294 * 1295 * If not forwarding, just drop the packet. This could be confusing 1296 * if ipforwarding was zero but some routing protocol was advancing 1297 * us as a gateway to somewhere. However, we must let the routing 1298 * protocol deal with that. 1299 * 1300 * The srcrt parameter indicates whether the packet is being forwarded 1301 * via a source route. 1302 */ 1303void 1304ip_forward(struct mbuf *m, int srcrt) 1305{ 1306 struct ip *ip = mtod(m, struct ip *); 1307 struct rtentry *rt; 1308 int error, type = 0, code = 0, destmtu = 0; 1309 struct mbuf *mcopy; 1310 n_long dest; 1311 union { 1312 struct sockaddr dst; 1313 struct sockaddr_in dst4; 1314 } u; 1315 1316 /* 1317 * We are now in the output path. 1318 */ 1319 MCLAIM(m, &ip_tx_mowner); 1320 1321 /* 1322 * Clear any in-bound checksum flags for this packet. 1323 */ 1324 m->m_pkthdr.csum_flags = 0; 1325 1326 dest = 0; 1327#ifdef DIAGNOSTIC 1328 if (ipprintfs) { 1329 printf("forward: src %s ", inet_ntoa(ip->ip_src)); 1330 printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl); 1331 } 1332#endif 1333 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1334 IP_STATINC(IP_STAT_CANTFORWARD); 1335 m_freem(m); 1336 return; 1337 } 1338 if (ip->ip_ttl <= IPTTLDEC) { 1339 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1340 return; 1341 } 1342 1343 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1344 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1345 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1346 return; 1347 } 1348 1349 /* 1350 * Save at most 68 bytes of the packet in case 1351 * we need to generate an ICMP message to the src. 1352 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1353 */ 1354 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1355 if (mcopy) 1356 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1357 1358 ip->ip_ttl -= IPTTLDEC; 1359 1360 /* 1361 * If forwarding packet using same interface that it came in on, 1362 * perhaps should send a redirect to sender to shortcut a hop. 1363 * Only send redirect if source is sending directly to us, 1364 * and if packet was not source routed (or has any options). 1365 * Also, don't send redirect if forwarding using a default route 1366 * or a route modified by a redirect. 1367 */ 1368 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1369 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1370 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1371 ipsendredirects && !srcrt) { 1372 if (rt->rt_ifa && 1373 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1374 ifatoia(rt->rt_ifa)->ia_subnet) { 1375 if (rt->rt_flags & RTF_GATEWAY) 1376 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1377 else 1378 dest = ip->ip_dst.s_addr; 1379 /* 1380 * Router requirements says to only send host 1381 * redirects. 1382 */ 1383 type = ICMP_REDIRECT; 1384 code = ICMP_REDIRECT_HOST; 1385#ifdef DIAGNOSTIC 1386 if (ipprintfs) 1387 printf("redirect (%d) to %x\n", code, 1388 (u_int32_t)dest); 1389#endif 1390 } 1391 } 1392 1393 error = ip_output(m, NULL, &ipforward_rt, 1394 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1395 (struct ip_moptions *)NULL, (struct socket *)NULL); 1396 1397 if (error) 1398 IP_STATINC(IP_STAT_CANTFORWARD); 1399 else { 1400 uint64_t *ips = IP_STAT_GETREF(); 1401 ips[IP_STAT_FORWARD]++; 1402 if (type) { 1403 ips[IP_STAT_REDIRECTSENT]++; 1404 IP_STAT_PUTREF(); 1405 } else { 1406 IP_STAT_PUTREF(); 1407 if (mcopy) { 1408#ifdef GATEWAY 1409 if (mcopy->m_flags & M_CANFASTFWD) 1410 ipflow_create(&ipforward_rt, mcopy); 1411#endif 1412 m_freem(mcopy); 1413 } 1414 return; 1415 } 1416 } 1417 if (mcopy == NULL) 1418 return; 1419 1420 switch (error) { 1421 1422 case 0: /* forwarded, but need redirect */ 1423 /* type, code set above */ 1424 break; 1425 1426 case ENETUNREACH: /* shouldn't happen, checked above */ 1427 case EHOSTUNREACH: 1428 case ENETDOWN: 1429 case EHOSTDOWN: 1430 default: 1431 type = ICMP_UNREACH; 1432 code = ICMP_UNREACH_HOST; 1433 break; 1434 1435 case EMSGSIZE: 1436 type = ICMP_UNREACH; 1437 code = ICMP_UNREACH_NEEDFRAG; 1438 1439 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1440 destmtu = rt->rt_ifp->if_mtu; 1441 1442#if defined(IPSEC) || defined(FAST_IPSEC) 1443 { 1444 /* 1445 * If the packet is routed over IPsec tunnel, tell the 1446 * originator the tunnel MTU. 1447 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz 1448 * XXX quickhack!!! 1449 */ 1450 1451 struct secpolicy *sp; 1452 int ipsecerror; 1453 size_t ipsechdr; 1454 struct route *ro; 1455 1456 sp = ipsec4_getpolicybyaddr(mcopy, 1457 IPSEC_DIR_OUTBOUND, IP_FORWARDING, 1458 &ipsecerror); 1459 1460 if (sp != NULL) { 1461 /* count IPsec header size */ 1462 ipsechdr = ipsec4_hdrsiz(mcopy, 1463 IPSEC_DIR_OUTBOUND, NULL); 1464 1465 /* 1466 * find the correct route for outer IPv4 1467 * header, compute tunnel MTU. 1468 */ 1469 1470 if (sp->req != NULL 1471 && sp->req->sav != NULL 1472 && sp->req->sav->sah != NULL) { 1473 ro = &sp->req->sav->sah->sa_route; 1474 rt = rtcache_validate(ro); 1475 if (rt && rt->rt_ifp) { 1476 destmtu = 1477 rt->rt_rmx.rmx_mtu ? 1478 rt->rt_rmx.rmx_mtu : 1479 rt->rt_ifp->if_mtu; 1480 destmtu -= ipsechdr; 1481 } 1482 } 1483 1484#ifdef IPSEC 1485 key_freesp(sp); 1486#else 1487 KEY_FREESP(&sp); 1488#endif 1489 } 1490 } 1491#endif /*defined(IPSEC) || defined(FAST_IPSEC)*/ 1492 IP_STATINC(IP_STAT_CANTFRAG); 1493 break; 1494 1495 case ENOBUFS: 1496#if 1 1497 /* 1498 * a router should not generate ICMP_SOURCEQUENCH as 1499 * required in RFC1812 Requirements for IP Version 4 Routers. 1500 * source quench could be a big problem under DoS attacks, 1501 * or if the underlying interface is rate-limited. 1502 */ 1503 if (mcopy) 1504 m_freem(mcopy); 1505 return; 1506#else 1507 type = ICMP_SOURCEQUENCH; 1508 code = 0; 1509 break; 1510#endif 1511 } 1512 icmp_error(mcopy, type, code, dest, destmtu); 1513} 1514 1515void 1516ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1517 struct mbuf *m) 1518{ 1519 1520 if (inp->inp_socket->so_options & SO_TIMESTAMP 1521#ifdef SO_OTIMESTAMP 1522 || inp->inp_socket->so_options & SO_OTIMESTAMP 1523#endif 1524 ) { 1525 struct timeval tv; 1526 1527 microtime(&tv); 1528#ifdef SO_OTIMESTAMP 1529 if (inp->inp_socket->so_options & SO_OTIMESTAMP) { 1530 struct timeval50 tv50; 1531 timeval_to_timeval50(&tv, &tv50); 1532 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1533 SCM_OTIMESTAMP, SOL_SOCKET); 1534 } else 1535#endif 1536 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1537 SCM_TIMESTAMP, SOL_SOCKET); 1538 if (*mp) 1539 mp = &(*mp)->m_next; 1540 } 1541 if (inp->inp_flags & INP_RECVDSTADDR) { 1542 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1543 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1544 if (*mp) 1545 mp = &(*mp)->m_next; 1546 } 1547#ifdef notyet 1548 /* 1549 * XXX 1550 * Moving these out of udp_input() made them even more broken 1551 * than they already were. 1552 * - fenner@parc.xerox.com 1553 */ 1554 /* options were tossed already */ 1555 if (inp->inp_flags & INP_RECVOPTS) { 1556 *mp = sbcreatecontrol((void *) opts_deleted_above, 1557 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1558 if (*mp) 1559 mp = &(*mp)->m_next; 1560 } 1561 /* ip_srcroute doesn't do what we want here, need to fix */ 1562 if (inp->inp_flags & INP_RECVRETOPTS) { 1563 *mp = sbcreatecontrol((void *) ip_srcroute(), 1564 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1565 if (*mp) 1566 mp = &(*mp)->m_next; 1567 } 1568#endif 1569 if (inp->inp_flags & INP_RECVIF) { 1570 struct sockaddr_dl sdl; 1571 1572 sockaddr_dl_init(&sdl, sizeof(sdl), 1573 (m->m_pkthdr.rcvif != NULL) 1574 ? m->m_pkthdr.rcvif->if_index 1575 : 0, 1576 0, NULL, 0, NULL, 0); 1577 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1578 if (*mp) 1579 mp = &(*mp)->m_next; 1580 } 1581 if (inp->inp_flags & INP_RECVTTL) { 1582 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1583 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1584 if (*mp) 1585 mp = &(*mp)->m_next; 1586 } 1587} 1588 1589/* 1590 * sysctl helper routine for net.inet.ip.forwsrcrt. 1591 */ 1592static int 1593sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1594{ 1595 int error, tmp; 1596 struct sysctlnode node; 1597 1598 node = *rnode; 1599 tmp = ip_forwsrcrt; 1600 node.sysctl_data = &tmp; 1601 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1602 if (error || newp == NULL) 1603 return (error); 1604 1605 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1606 0, NULL, NULL, NULL); 1607 if (error) 1608 return (error); 1609 1610 ip_forwsrcrt = tmp; 1611 1612 return (0); 1613} 1614 1615/* 1616 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1617 * range of the new value and tweaks timers if it changes. 1618 */ 1619static int 1620sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1621{ 1622 int error, tmp; 1623 struct sysctlnode node; 1624 1625 node = *rnode; 1626 tmp = ip_mtudisc_timeout; 1627 node.sysctl_data = &tmp; 1628 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1629 if (error || newp == NULL) 1630 return (error); 1631 if (tmp < 0) 1632 return (EINVAL); 1633 1634 mutex_enter(softnet_lock); 1635 1636 ip_mtudisc_timeout = tmp; 1637 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1638 1639 mutex_exit(softnet_lock); 1640 1641 return (0); 1642} 1643 1644#ifdef GATEWAY 1645/* 1646 * sysctl helper routine for net.inet.ip.maxflows. 1647 */ 1648static int 1649sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS) 1650{ 1651 int error; 1652 1653 error = sysctl_lookup(SYSCTLFN_CALL(rnode)); 1654 if (error || newp == NULL) 1655 return (error); 1656 1657 mutex_enter(softnet_lock); 1658 KERNEL_LOCK(1, NULL); 1659 1660 ipflow_prune(); 1661 1662 KERNEL_UNLOCK_ONE(NULL); 1663 mutex_exit(softnet_lock); 1664 1665 return (0); 1666} 1667 1668static int 1669sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS) 1670{ 1671 int error, tmp; 1672 struct sysctlnode node; 1673 1674 node = *rnode; 1675 tmp = ip_hashsize; 1676 node.sysctl_data = &tmp; 1677 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1678 if (error || newp == NULL) 1679 return (error); 1680 1681 if ((tmp & (tmp - 1)) == 0 && tmp != 0) { 1682 /* 1683 * Can only fail due to malloc() 1684 */ 1685 mutex_enter(softnet_lock); 1686 KERNEL_LOCK(1, NULL); 1687 1688 error = ipflow_invalidate_all(tmp); 1689 1690 KERNEL_UNLOCK_ONE(NULL); 1691 mutex_exit(softnet_lock); 1692 1693 } else { 1694 /* 1695 * EINVAL if not a power of 2 1696 */ 1697 error = EINVAL; 1698 } 1699 1700 return error; 1701} 1702#endif /* GATEWAY */ 1703 1704static int 1705sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1706{ 1707 1708 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1709} 1710 1711static void 1712sysctl_net_inet_ip_setup(struct sysctllog **clog) 1713{ 1714 extern int subnetsarelocal, hostzeroisbroadcast; 1715 1716 sysctl_createv(clog, 0, NULL, NULL, 1717 CTLFLAG_PERMANENT, 1718 CTLTYPE_NODE, "net", NULL, 1719 NULL, 0, NULL, 0, 1720 CTL_NET, CTL_EOL); 1721 sysctl_createv(clog, 0, NULL, NULL, 1722 CTLFLAG_PERMANENT, 1723 CTLTYPE_NODE, "inet", 1724 SYSCTL_DESCR("PF_INET related settings"), 1725 NULL, 0, NULL, 0, 1726 CTL_NET, PF_INET, CTL_EOL); 1727 sysctl_createv(clog, 0, NULL, NULL, 1728 CTLFLAG_PERMANENT, 1729 CTLTYPE_NODE, "ip", 1730 SYSCTL_DESCR("IPv4 related settings"), 1731 NULL, 0, NULL, 0, 1732 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1733 1734 sysctl_createv(clog, 0, NULL, NULL, 1735 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1736 CTLTYPE_INT, "forwarding", 1737 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1738 NULL, 0, &ipforwarding, 0, 1739 CTL_NET, PF_INET, IPPROTO_IP, 1740 IPCTL_FORWARDING, CTL_EOL); 1741 sysctl_createv(clog, 0, NULL, NULL, 1742 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1743 CTLTYPE_INT, "redirect", 1744 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1745 NULL, 0, &ipsendredirects, 0, 1746 CTL_NET, PF_INET, IPPROTO_IP, 1747 IPCTL_SENDREDIRECTS, CTL_EOL); 1748 sysctl_createv(clog, 0, NULL, NULL, 1749 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1750 CTLTYPE_INT, "ttl", 1751 SYSCTL_DESCR("Default TTL for an INET datagram"), 1752 NULL, 0, &ip_defttl, 0, 1753 CTL_NET, PF_INET, IPPROTO_IP, 1754 IPCTL_DEFTTL, CTL_EOL); 1755#ifdef IPCTL_DEFMTU 1756 sysctl_createv(clog, 0, NULL, NULL, 1757 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1758 CTLTYPE_INT, "mtu", 1759 SYSCTL_DESCR("Default MTA for an INET route"), 1760 NULL, 0, &ip_mtu, 0, 1761 CTL_NET, PF_INET, IPPROTO_IP, 1762 IPCTL_DEFMTU, CTL_EOL); 1763#endif /* IPCTL_DEFMTU */ 1764 sysctl_createv(clog, 0, NULL, NULL, 1765 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1766 CTLTYPE_INT, "forwsrcrt", 1767 SYSCTL_DESCR("Enable forwarding of source-routed " 1768 "datagrams"), 1769 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1770 CTL_NET, PF_INET, IPPROTO_IP, 1771 IPCTL_FORWSRCRT, CTL_EOL); 1772 sysctl_createv(clog, 0, NULL, NULL, 1773 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1774 CTLTYPE_INT, "directed-broadcast", 1775 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1776 NULL, 0, &ip_directedbcast, 0, 1777 CTL_NET, PF_INET, IPPROTO_IP, 1778 IPCTL_DIRECTEDBCAST, CTL_EOL); 1779 sysctl_createv(clog, 0, NULL, NULL, 1780 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1781 CTLTYPE_INT, "allowsrcrt", 1782 SYSCTL_DESCR("Accept source-routed datagrams"), 1783 NULL, 0, &ip_allowsrcrt, 0, 1784 CTL_NET, PF_INET, IPPROTO_IP, 1785 IPCTL_ALLOWSRCRT, CTL_EOL); 1786 sysctl_createv(clog, 0, NULL, NULL, 1787 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1788 CTLTYPE_INT, "subnetsarelocal", 1789 SYSCTL_DESCR("Whether logical subnets are considered " 1790 "local"), 1791 NULL, 0, &subnetsarelocal, 0, 1792 CTL_NET, PF_INET, IPPROTO_IP, 1793 IPCTL_SUBNETSARELOCAL, CTL_EOL); 1794 sysctl_createv(clog, 0, NULL, NULL, 1795 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1796 CTLTYPE_INT, "mtudisc", 1797 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1798 NULL, 0, &ip_mtudisc, 0, 1799 CTL_NET, PF_INET, IPPROTO_IP, 1800 IPCTL_MTUDISC, CTL_EOL); 1801 sysctl_createv(clog, 0, NULL, NULL, 1802 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1803 CTLTYPE_INT, "anonportmin", 1804 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1805 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1806 CTL_NET, PF_INET, IPPROTO_IP, 1807 IPCTL_ANONPORTMIN, CTL_EOL); 1808 sysctl_createv(clog, 0, NULL, NULL, 1809 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1810 CTLTYPE_INT, "anonportmax", 1811 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1812 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1813 CTL_NET, PF_INET, IPPROTO_IP, 1814 IPCTL_ANONPORTMAX, CTL_EOL); 1815 sysctl_createv(clog, 0, NULL, NULL, 1816 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1817 CTLTYPE_INT, "mtudisctimeout", 1818 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1819 sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0, 1820 CTL_NET, PF_INET, IPPROTO_IP, 1821 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1822#ifdef GATEWAY 1823 sysctl_createv(clog, 0, NULL, NULL, 1824 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1825 CTLTYPE_INT, "maxflows", 1826 SYSCTL_DESCR("Number of flows for fast forwarding"), 1827 sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0, 1828 CTL_NET, PF_INET, IPPROTO_IP, 1829 IPCTL_MAXFLOWS, CTL_EOL); 1830 sysctl_createv(clog, 0, NULL, NULL, 1831 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1832 CTLTYPE_INT, "hashsize", 1833 SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"), 1834 sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0, 1835 CTL_NET, PF_INET, IPPROTO_IP, 1836 CTL_CREATE, CTL_EOL); 1837#endif /* GATEWAY */ 1838 sysctl_createv(clog, 0, NULL, NULL, 1839 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1840 CTLTYPE_INT, "hostzerobroadcast", 1841 SYSCTL_DESCR("All zeroes address is broadcast address"), 1842 NULL, 0, &hostzeroisbroadcast, 0, 1843 CTL_NET, PF_INET, IPPROTO_IP, 1844 IPCTL_HOSTZEROBROADCAST, CTL_EOL); 1845#if NGIF > 0 1846 sysctl_createv(clog, 0, NULL, NULL, 1847 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1848 CTLTYPE_INT, "gifttl", 1849 SYSCTL_DESCR("Default TTL for a gif tunnel datagram"), 1850 NULL, 0, &ip_gif_ttl, 0, 1851 CTL_NET, PF_INET, IPPROTO_IP, 1852 IPCTL_GIF_TTL, CTL_EOL); 1853#endif /* NGIF */ 1854#ifndef IPNOPRIVPORTS 1855 sysctl_createv(clog, 0, NULL, NULL, 1856 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1857 CTLTYPE_INT, "lowportmin", 1858 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1859 "to assign"), 1860 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1861 CTL_NET, PF_INET, IPPROTO_IP, 1862 IPCTL_LOWPORTMIN, CTL_EOL); 1863 sysctl_createv(clog, 0, NULL, NULL, 1864 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1865 CTLTYPE_INT, "lowportmax", 1866 SYSCTL_DESCR("Highest privileged ephemeral port number " 1867 "to assign"), 1868 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1869 CTL_NET, PF_INET, IPPROTO_IP, 1870 IPCTL_LOWPORTMAX, CTL_EOL); 1871#endif /* IPNOPRIVPORTS */ 1872#if NGRE > 0 1873 sysctl_createv(clog, 0, NULL, NULL, 1874 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1875 CTLTYPE_INT, "grettl", 1876 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1877 NULL, 0, &ip_gre_ttl, 0, 1878 CTL_NET, PF_INET, IPPROTO_IP, 1879 IPCTL_GRE_TTL, CTL_EOL); 1880#endif /* NGRE */ 1881 sysctl_createv(clog, 0, NULL, NULL, 1882 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1883 CTLTYPE_INT, "checkinterface", 1884 SYSCTL_DESCR("Enable receive side of Strong ES model " 1885 "from RFC1122"), 1886 NULL, 0, &ip_checkinterface, 0, 1887 CTL_NET, PF_INET, IPPROTO_IP, 1888 IPCTL_CHECKINTERFACE, CTL_EOL); 1889 sysctl_createv(clog, 0, NULL, NULL, 1890 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1891 CTLTYPE_INT, "random_id", 1892 SYSCTL_DESCR("Assign random ip_id values"), 1893 NULL, 0, &ip_do_randomid, 0, 1894 CTL_NET, PF_INET, IPPROTO_IP, 1895 IPCTL_RANDOMID, CTL_EOL); 1896 sysctl_createv(clog, 0, NULL, NULL, 1897 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1898 CTLTYPE_INT, "do_loopback_cksum", 1899 SYSCTL_DESCR("Perform IP checksum on loopback"), 1900 NULL, 0, &ip_do_loopback_cksum, 0, 1901 CTL_NET, PF_INET, IPPROTO_IP, 1902 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1903 sysctl_createv(clog, 0, NULL, NULL, 1904 CTLFLAG_PERMANENT, 1905 CTLTYPE_STRUCT, "stats", 1906 SYSCTL_DESCR("IP statistics"), 1907 sysctl_net_inet_ip_stats, 0, NULL, 0, 1908 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1909 CTL_EOL); 1910} 1911 1912void 1913ip_statinc(u_int stat) 1914{ 1915 1916 KASSERT(stat < IP_NSTATS); 1917 IP_STATINC(stat); 1918} 1919