ip_input.c revision 1.183
1/* $NetBSD: ip_input.c,v 1.183 2003/11/17 22:34:16 jonathan 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 * 3. All advertising materials mentioning features or use of this software 49 * must display the following acknowledgement: 50 * This product includes software developed by the NetBSD 51 * Foundation, Inc. and its contributors. 52 * 4. Neither the name of The NetBSD Foundation nor the names of its 53 * contributors may be used to endorse or promote products derived 54 * from this software without specific prior written permission. 55 * 56 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 57 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 58 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 59 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 60 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 61 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 62 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 63 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 64 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 65 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 66 * POSSIBILITY OF SUCH DAMAGE. 67 */ 68 69/* 70 * Copyright (c) 1982, 1986, 1988, 1993 71 * The Regents of the University of California. All rights reserved. 72 * 73 * Redistribution and use in source and binary forms, with or without 74 * modification, are permitted provided that the following conditions 75 * are met: 76 * 1. Redistributions of source code must retain the above copyright 77 * notice, this list of conditions and the following disclaimer. 78 * 2. Redistributions in binary form must reproduce the above copyright 79 * notice, this list of conditions and the following disclaimer in the 80 * documentation and/or other materials provided with the distribution. 81 * 3. Neither the name of the University nor the names of its contributors 82 * may be used to endorse or promote products derived from this software 83 * without specific prior written permission. 84 * 85 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 86 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 87 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 88 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 89 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 90 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 91 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 92 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 93 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 94 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 95 * SUCH DAMAGE. 96 * 97 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 98 */ 99 100#include <sys/cdefs.h> 101__KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.183 2003/11/17 22:34:16 jonathan Exp $"); 102 103#include "opt_gateway.h" 104#include "opt_pfil_hooks.h" 105#include "opt_ipsec.h" 106#include "opt_mrouting.h" 107#include "opt_mbuftrace.h" 108#include "opt_inet_csum.h" 109 110#include <sys/param.h> 111#include <sys/systm.h> 112#include <sys/malloc.h> 113#include <sys/mbuf.h> 114#include <sys/domain.h> 115#include <sys/protosw.h> 116#include <sys/socket.h> 117#include <sys/socketvar.h> 118#include <sys/errno.h> 119#include <sys/time.h> 120#include <sys/kernel.h> 121#include <sys/pool.h> 122#include <sys/sysctl.h> 123 124#include <net/if.h> 125#include <net/if_dl.h> 126#include <net/route.h> 127#include <net/pfil.h> 128 129#include <netinet/in.h> 130#include <netinet/in_systm.h> 131#include <netinet/ip.h> 132#include <netinet/in_pcb.h> 133#include <netinet/in_var.h> 134#include <netinet/ip_var.h> 135#include <netinet/ip_icmp.h> 136/* just for gif_ttl */ 137#include <netinet/in_gif.h> 138#include "gif.h" 139#include <net/if_gre.h> 140#include "gre.h" 141 142#ifdef MROUTING 143#include <netinet/ip_mroute.h> 144#endif 145 146#ifdef IPSEC 147#include <netinet6/ipsec.h> 148#include <netkey/key.h> 149#endif 150#ifdef FAST_IPSEC 151#include <netipsec/ipsec.h> 152#include <netipsec/key.h> 153#endif /* FAST_IPSEC*/ 154 155#ifndef IPFORWARDING 156#ifdef GATEWAY 157#define IPFORWARDING 1 /* forward IP packets not for us */ 158#else /* GATEWAY */ 159#define IPFORWARDING 0 /* don't forward IP packets not for us */ 160#endif /* GATEWAY */ 161#endif /* IPFORWARDING */ 162#ifndef IPSENDREDIRECTS 163#define IPSENDREDIRECTS 1 164#endif 165#ifndef IPFORWSRCRT 166#define IPFORWSRCRT 1 /* forward source-routed packets */ 167#endif 168#ifndef IPALLOWSRCRT 169#define IPALLOWSRCRT 1 /* allow source-routed packets */ 170#endif 171#ifndef IPMTUDISC 172#define IPMTUDISC 1 173#endif 174#ifndef IPMTUDISCTIMEOUT 175#define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 176#endif 177 178/* 179 * Note: DIRECTED_BROADCAST is handled this way so that previous 180 * configuration using this option will Just Work. 181 */ 182#ifndef IPDIRECTEDBCAST 183#ifdef DIRECTED_BROADCAST 184#define IPDIRECTEDBCAST 1 185#else 186#define IPDIRECTEDBCAST 0 187#endif /* DIRECTED_BROADCAST */ 188#endif /* IPDIRECTEDBCAST */ 189int ipforwarding = IPFORWARDING; 190int ipsendredirects = IPSENDREDIRECTS; 191int ip_defttl = IPDEFTTL; 192int ip_forwsrcrt = IPFORWSRCRT; 193int ip_directedbcast = IPDIRECTEDBCAST; 194int ip_allowsrcrt = IPALLOWSRCRT; 195int ip_mtudisc = IPMTUDISC; 196int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 197#ifdef DIAGNOSTIC 198int ipprintfs = 0; 199#endif 200/* 201 * XXX - Setting ip_checkinterface mostly implements the receive side of 202 * the Strong ES model described in RFC 1122, but since the routing table 203 * and transmit implementation do not implement the Strong ES model, 204 * setting this to 1 results in an odd hybrid. 205 * 206 * XXX - ip_checkinterface currently must be disabled if you use ipnat 207 * to translate the destination address to another local interface. 208 * 209 * XXX - ip_checkinterface must be disabled if you add IP aliases 210 * to the loopback interface instead of the interface where the 211 * packets for those addresses are received. 212 */ 213int ip_checkinterface = 0; 214 215 216struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 217 218extern struct domain inetdomain; 219int ipqmaxlen = IFQ_MAXLEN; 220u_long in_ifaddrhash; /* size of hash table - 1 */ 221int in_ifaddrentries; /* total number of addrs */ 222struct in_ifaddrhead in_ifaddrhead; 223struct in_ifaddrhashhead *in_ifaddrhashtbl; 224u_long in_multihash; /* size of hash table - 1 */ 225int in_multientries; /* total number of addrs */ 226struct in_multihashhead *in_multihashtbl; 227struct ifqueue ipintrq; 228struct ipstat ipstat; 229uint16_t ip_id; 230 231#ifdef PFIL_HOOKS 232struct pfil_head inet_pfil_hook; 233#endif 234 235struct ipqhead ipq; 236int ipq_locked; 237int ip_nfragpackets = 0; 238int ip_maxfragpackets = 200; 239 240static __inline int ipq_lock_try __P((void)); 241static __inline void ipq_unlock __P((void)); 242 243static __inline int 244ipq_lock_try() 245{ 246 int s; 247 248 /* 249 * Use splvm() -- we're blocking things that would cause 250 * mbuf allocation. 251 */ 252 s = splvm(); 253 if (ipq_locked) { 254 splx(s); 255 return (0); 256 } 257 ipq_locked = 1; 258 splx(s); 259 return (1); 260} 261 262static __inline void 263ipq_unlock() 264{ 265 int s; 266 267 s = splvm(); 268 ipq_locked = 0; 269 splx(s); 270} 271 272#ifdef DIAGNOSTIC 273#define IPQ_LOCK() \ 274do { \ 275 if (ipq_lock_try() == 0) { \ 276 printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \ 277 panic("ipq_lock"); \ 278 } \ 279} while (/*CONSTCOND*/ 0) 280#define IPQ_LOCK_CHECK() \ 281do { \ 282 if (ipq_locked == 0) { \ 283 printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \ 284 panic("ipq lock check"); \ 285 } \ 286} while (/*CONSTCOND*/ 0) 287#else 288#define IPQ_LOCK() (void) ipq_lock_try() 289#define IPQ_LOCK_CHECK() /* nothing */ 290#endif 291 292#define IPQ_UNLOCK() ipq_unlock() 293 294struct pool inmulti_pool; 295struct pool ipqent_pool; 296 297#ifdef INET_CSUM_COUNTERS 298#include <sys/device.h> 299 300struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 301 NULL, "inet", "hwcsum bad"); 302struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 303 NULL, "inet", "hwcsum ok"); 304struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 305 NULL, "inet", "swcsum"); 306 307#define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 308 309#else 310 311#define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 312 313#endif /* INET_CSUM_COUNTERS */ 314 315/* 316 * We need to save the IP options in case a protocol wants to respond 317 * to an incoming packet over the same route if the packet got here 318 * using IP source routing. This allows connection establishment and 319 * maintenance when the remote end is on a network that is not known 320 * to us. 321 */ 322int ip_nhops = 0; 323static struct ip_srcrt { 324 struct in_addr dst; /* final destination */ 325 char nop; /* one NOP to align */ 326 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 327 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 328} ip_srcrt; 329 330static void save_rte __P((u_char *, struct in_addr)); 331 332#ifdef MBUFTRACE 333struct mowner ip_rx_mowner = { "internet", "rx" }; 334struct mowner ip_tx_mowner = { "internet", "tx" }; 335#endif 336 337/* 338 * IP initialization: fill in IP protocol switch table. 339 * All protocols not implemented in kernel go to raw IP protocol handler. 340 */ 341void 342ip_init() 343{ 344 struct protosw *pr; 345 int i; 346 347 pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", 348 NULL); 349 pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", 350 NULL); 351 352 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 353 if (pr == 0) 354 panic("ip_init"); 355 for (i = 0; i < IPPROTO_MAX; i++) 356 ip_protox[i] = pr - inetsw; 357 for (pr = inetdomain.dom_protosw; 358 pr < inetdomain.dom_protoswNPROTOSW; pr++) 359 if (pr->pr_domain->dom_family == PF_INET && 360 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 361 ip_protox[pr->pr_protocol] = pr - inetsw; 362 LIST_INIT(&ipq); 363 ip_id = time.tv_sec & 0xfffff; 364 ipintrq.ifq_maxlen = ipqmaxlen; 365 TAILQ_INIT(&in_ifaddrhead); 366 in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IFADDR, 367 M_WAITOK, &in_ifaddrhash); 368 in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IPMADDR, 369 M_WAITOK, &in_multihash); 370 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 371#ifdef GATEWAY 372 ipflow_init(); 373#endif 374 375#ifdef PFIL_HOOKS 376 /* Register our Packet Filter hook. */ 377 inet_pfil_hook.ph_type = PFIL_TYPE_AF; 378 inet_pfil_hook.ph_af = AF_INET; 379 i = pfil_head_register(&inet_pfil_hook); 380 if (i != 0) 381 printf("ip_init: WARNING: unable to register pfil hook, " 382 "error %d\n", i); 383#endif /* PFIL_HOOKS */ 384 385#ifdef INET_CSUM_COUNTERS 386 evcnt_attach_static(&ip_hwcsum_bad); 387 evcnt_attach_static(&ip_hwcsum_ok); 388 evcnt_attach_static(&ip_swcsum); 389#endif /* INET_CSUM_COUNTERS */ 390 391#ifdef MBUFTRACE 392 MOWNER_ATTACH(&ip_tx_mowner); 393 MOWNER_ATTACH(&ip_rx_mowner); 394#endif /* MBUFTRACE */ 395} 396 397struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 398struct route ipforward_rt; 399 400/* 401 * IP software interrupt routine 402 */ 403void 404ipintr() 405{ 406 int s; 407 struct mbuf *m; 408 409 while (1) { 410 s = splnet(); 411 IF_DEQUEUE(&ipintrq, m); 412 splx(s); 413 if (m == 0) 414 return; 415 MCLAIM(m, &ip_rx_mowner); 416 ip_input(m); 417 } 418} 419 420/* 421 * Ip input routine. Checksum and byte swap header. If fragmented 422 * try to reassemble. Process options. Pass to next level. 423 */ 424void 425ip_input(struct mbuf *m) 426{ 427 struct ip *ip = NULL; 428 struct ipq *fp; 429 struct in_ifaddr *ia; 430 struct ifaddr *ifa; 431 struct ipqent *ipqe; 432 int hlen = 0, mff, len; 433 int downmatch; 434 int checkif; 435 int srcrt = 0; 436#ifdef FAST_IPSEC 437 struct m_tag *mtag; 438 struct tdb_ident *tdbi; 439 struct secpolicy *sp; 440 int s, error; 441#endif /* FAST_IPSEC */ 442 443 MCLAIM(m, &ip_rx_mowner); 444#ifdef DIAGNOSTIC 445 if ((m->m_flags & M_PKTHDR) == 0) 446 panic("ipintr no HDR"); 447#endif 448 449 /* 450 * If no IP addresses have been set yet but the interfaces 451 * are receiving, can't do anything with incoming packets yet. 452 */ 453 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 454 goto bad; 455 ipstat.ips_total++; 456 /* 457 * If the IP header is not aligned, slurp it up into a new 458 * mbuf with space for link headers, in the event we forward 459 * it. Otherwise, if it is aligned, make sure the entire 460 * base IP header is in the first mbuf of the chain. 461 */ 462 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 463 if ((m = m_copyup(m, sizeof(struct ip), 464 (max_linkhdr + 3) & ~3)) == NULL) { 465 /* XXXJRT new stat, please */ 466 ipstat.ips_toosmall++; 467 return; 468 } 469 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 470 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 471 ipstat.ips_toosmall++; 472 return; 473 } 474 } 475 ip = mtod(m, struct ip *); 476 if (ip->ip_v != IPVERSION) { 477 ipstat.ips_badvers++; 478 goto bad; 479 } 480 hlen = ip->ip_hl << 2; 481 if (hlen < sizeof(struct ip)) { /* minimum header length */ 482 ipstat.ips_badhlen++; 483 goto bad; 484 } 485 if (hlen > m->m_len) { 486 if ((m = m_pullup(m, hlen)) == 0) { 487 ipstat.ips_badhlen++; 488 return; 489 } 490 ip = mtod(m, struct ip *); 491 } 492 493 /* 494 * RFC1122: packets with a multicast source address are 495 * not allowed. 496 */ 497 if (IN_MULTICAST(ip->ip_src.s_addr)) { 498 ipstat.ips_badaddr++; 499 goto bad; 500 } 501 502 /* 127/8 must not appear on wire - RFC1122 */ 503 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 504 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 505 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 506 ipstat.ips_badaddr++; 507 goto bad; 508 } 509 } 510 511 switch (m->m_pkthdr.csum_flags & 512 ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) | 513 M_CSUM_IPv4_BAD)) { 514 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 515 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 516 goto badcsum; 517 518 case M_CSUM_IPv4: 519 /* Checksum was okay. */ 520 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 521 break; 522 523 default: 524 /* Must compute it ourselves. */ 525 INET_CSUM_COUNTER_INCR(&ip_swcsum); 526 if (in_cksum(m, hlen) != 0) 527 goto bad; 528 break; 529 } 530 531 /* Retrieve the packet length. */ 532 len = ntohs(ip->ip_len); 533 534 /* 535 * Check for additional length bogosity 536 */ 537 if (len < hlen) { 538 ipstat.ips_badlen++; 539 goto bad; 540 } 541 542 /* 543 * Check that the amount of data in the buffers 544 * is as at least much as the IP header would have us expect. 545 * Trim mbufs if longer than we expect. 546 * Drop packet if shorter than we expect. 547 */ 548 if (m->m_pkthdr.len < len) { 549 ipstat.ips_tooshort++; 550 goto bad; 551 } 552 if (m->m_pkthdr.len > len) { 553 if (m->m_len == m->m_pkthdr.len) { 554 m->m_len = len; 555 m->m_pkthdr.len = len; 556 } else 557 m_adj(m, len - m->m_pkthdr.len); 558 } 559 560#ifdef IPSEC 561 /* ipflow (IP fast forwarding) is not compatible with IPsec. */ 562 m->m_flags &= ~M_CANFASTFWD; 563#else 564 /* 565 * Assume that we can create a fast-forward IP flow entry 566 * based on this packet. 567 */ 568 m->m_flags |= M_CANFASTFWD; 569#endif 570 571#ifdef PFIL_HOOKS 572 /* 573 * Run through list of hooks for input packets. If there are any 574 * filters which require that additional packets in the flow are 575 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 576 * Note that filters must _never_ set this flag, as another filter 577 * in the list may have previously cleared it. 578 */ 579 /* 580 * let ipfilter look at packet on the wire, 581 * not the decapsulated packet. 582 */ 583#ifdef IPSEC 584 if (!ipsec_getnhist(m)) 585#else 586 if (1) 587#endif 588 { 589 struct in_addr odst; 590 591 odst = ip->ip_dst; 592 if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif, 593 PFIL_IN) != 0) 594 return; 595 if (m == NULL) 596 return; 597 ip = mtod(m, struct ip *); 598 hlen = ip->ip_hl << 2; 599 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 600 } 601#endif /* PFIL_HOOKS */ 602 603#ifdef ALTQ 604 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 605 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) { 606 /* packet dropped by traffic conditioner */ 607 return; 608 } 609#endif 610 611 /* 612 * Process options and, if not destined for us, 613 * ship it on. ip_dooptions returns 1 when an 614 * error was detected (causing an icmp message 615 * to be sent and the original packet to be freed). 616 */ 617 ip_nhops = 0; /* for source routed packets */ 618 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 619 return; 620 621 /* 622 * Enable a consistency check between the destination address 623 * and the arrival interface for a unicast packet (the RFC 1122 624 * strong ES model) if IP forwarding is disabled and the packet 625 * is not locally generated. 626 * 627 * XXX - Checking also should be disabled if the destination 628 * address is ipnat'ed to a different interface. 629 * 630 * XXX - Checking is incompatible with IP aliases added 631 * to the loopback interface instead of the interface where 632 * the packets are received. 633 * 634 * XXX - We need to add a per ifaddr flag for this so that 635 * we get finer grain control. 636 */ 637 checkif = ip_checkinterface && (ipforwarding == 0) && 638 (m->m_pkthdr.rcvif != NULL) && 639 ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0); 640 641 /* 642 * Check our list of addresses, to see if the packet is for us. 643 * 644 * Traditional 4.4BSD did not consult IFF_UP at all. 645 * The behavior here is to treat addresses on !IFF_UP interface 646 * as not mine. 647 */ 648 downmatch = 0; 649 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 650 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 651 if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif) 652 continue; 653 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 654 break; 655 else 656 downmatch++; 657 } 658 } 659 if (ia != NULL) 660 goto ours; 661 if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 662 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) { 663 if (ifa->ifa_addr->sa_family != AF_INET) 664 continue; 665 ia = ifatoia(ifa); 666 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 667 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 668 /* 669 * Look for all-0's host part (old broadcast addr), 670 * either for subnet or net. 671 */ 672 ip->ip_dst.s_addr == ia->ia_subnet || 673 ip->ip_dst.s_addr == ia->ia_net) 674 goto ours; 675 /* 676 * An interface with IP address zero accepts 677 * all packets that arrive on that interface. 678 */ 679 if (in_nullhost(ia->ia_addr.sin_addr)) 680 goto ours; 681 } 682 } 683 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 684 struct in_multi *inm; 685#ifdef MROUTING 686 extern struct socket *ip_mrouter; 687 688 if (M_READONLY(m)) { 689 if ((m = m_pullup(m, hlen)) == 0) { 690 ipstat.ips_toosmall++; 691 return; 692 } 693 ip = mtod(m, struct ip *); 694 } 695 696 if (ip_mrouter) { 697 /* 698 * If we are acting as a multicast router, all 699 * incoming multicast packets are passed to the 700 * kernel-level multicast forwarding function. 701 * The packet is returned (relatively) intact; if 702 * ip_mforward() returns a non-zero value, the packet 703 * must be discarded, else it may be accepted below. 704 * 705 * (The IP ident field is put in the same byte order 706 * as expected when ip_mforward() is called from 707 * ip_output().) 708 */ 709 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 710 ipstat.ips_cantforward++; 711 m_freem(m); 712 return; 713 } 714 715 /* 716 * The process-level routing demon needs to receive 717 * all multicast IGMP packets, whether or not this 718 * host belongs to their destination groups. 719 */ 720 if (ip->ip_p == IPPROTO_IGMP) 721 goto ours; 722 ipstat.ips_forward++; 723 } 724#endif 725 /* 726 * See if we belong to the destination multicast group on the 727 * arrival interface. 728 */ 729 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 730 if (inm == NULL) { 731 ipstat.ips_cantforward++; 732 m_freem(m); 733 return; 734 } 735 goto ours; 736 } 737 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 738 in_nullhost(ip->ip_dst)) 739 goto ours; 740 741 /* 742 * Not for us; forward if possible and desirable. 743 */ 744 if (ipforwarding == 0) { 745 ipstat.ips_cantforward++; 746 m_freem(m); 747 } else { 748 /* 749 * If ip_dst matched any of my address on !IFF_UP interface, 750 * and there's no IFF_UP interface that matches ip_dst, 751 * send icmp unreach. Forwarding it will result in in-kernel 752 * forwarding loop till TTL goes to 0. 753 */ 754 if (downmatch) { 755 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 756 ipstat.ips_cantforward++; 757 return; 758 } 759#ifdef IPSEC 760 if (ipsec4_in_reject(m, NULL)) { 761 ipsecstat.in_polvio++; 762 goto bad; 763 } 764#endif 765#ifdef FAST_IPSEC 766 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 767 s = splsoftnet(); 768 if (mtag != NULL) { 769 tdbi = (struct tdb_ident *)(mtag + 1); 770 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 771 } else { 772 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 773 IP_FORWARDING, &error); 774 } 775 if (sp == NULL) { /* NB: can happen if error */ 776 splx(s); 777 /*XXX error stat???*/ 778 DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/ 779 goto bad; 780 } 781 782 /* 783 * Check security policy against packet attributes. 784 */ 785 error = ipsec_in_reject(sp, m); 786 KEY_FREESP(&sp); 787 splx(s); 788 if (error) { 789 ipstat.ips_cantforward++; 790 goto bad; 791 } 792#endif /* FAST_IPSEC */ 793 794 ip_forward(m, srcrt); 795 } 796 return; 797 798ours: 799 /* 800 * If offset or IP_MF are set, must reassemble. 801 * Otherwise, nothing need be done. 802 * (We could look in the reassembly queue to see 803 * if the packet was previously fragmented, 804 * but it's not worth the time; just let them time out.) 805 */ 806 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 807 if (M_READONLY(m)) { 808 if ((m = m_pullup(m, hlen)) == NULL) { 809 ipstat.ips_toosmall++; 810 goto bad; 811 } 812 ip = mtod(m, struct ip *); 813 } 814 815 /* 816 * Look for queue of fragments 817 * of this datagram. 818 */ 819 IPQ_LOCK(); 820 LIST_FOREACH(fp, &ipq, ipq_q) 821 if (ip->ip_id == fp->ipq_id && 822 in_hosteq(ip->ip_src, fp->ipq_src) && 823 in_hosteq(ip->ip_dst, fp->ipq_dst) && 824 ip->ip_p == fp->ipq_p) 825 goto found; 826 fp = 0; 827found: 828 829 /* 830 * Adjust ip_len to not reflect header, 831 * set ipqe_mff if more fragments are expected, 832 * convert offset of this to bytes. 833 */ 834 ip->ip_len = htons(ntohs(ip->ip_len) - hlen); 835 mff = (ip->ip_off & htons(IP_MF)) != 0; 836 if (mff) { 837 /* 838 * Make sure that fragments have a data length 839 * that's a non-zero multiple of 8 bytes. 840 */ 841 if (ntohs(ip->ip_len) == 0 || 842 (ntohs(ip->ip_len) & 0x7) != 0) { 843 ipstat.ips_badfrags++; 844 IPQ_UNLOCK(); 845 goto bad; 846 } 847 } 848 ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3); 849 850 /* 851 * If datagram marked as having more fragments 852 * or if this is not the first fragment, 853 * attempt reassembly; if it succeeds, proceed. 854 */ 855 if (mff || ip->ip_off != htons(0)) { 856 ipstat.ips_fragments++; 857 ipqe = pool_get(&ipqent_pool, PR_NOWAIT); 858 if (ipqe == NULL) { 859 ipstat.ips_rcvmemdrop++; 860 IPQ_UNLOCK(); 861 goto bad; 862 } 863 ipqe->ipqe_mff = mff; 864 ipqe->ipqe_m = m; 865 ipqe->ipqe_ip = ip; 866 m = ip_reass(ipqe, fp); 867 if (m == 0) { 868 IPQ_UNLOCK(); 869 return; 870 } 871 ipstat.ips_reassembled++; 872 ip = mtod(m, struct ip *); 873 hlen = ip->ip_hl << 2; 874 ip->ip_len = htons(ntohs(ip->ip_len) + hlen); 875 } else 876 if (fp) 877 ip_freef(fp); 878 IPQ_UNLOCK(); 879 } 880 881#if defined(IPSEC) 882 /* 883 * enforce IPsec policy checking if we are seeing last header. 884 * note that we do not visit this with protocols with pcb layer 885 * code - like udp/tcp/raw ip. 886 */ 887 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 && 888 ipsec4_in_reject(m, NULL)) { 889 ipsecstat.in_polvio++; 890 goto bad; 891 } 892#endif 893#if FAST_IPSEC 894 /* 895 * enforce IPsec policy checking if we are seeing last header. 896 * note that we do not visit this with protocols with pcb layer 897 * code - like udp/tcp/raw ip. 898 */ 899 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 900 /* 901 * Check if the packet has already had IPsec processing 902 * done. If so, then just pass it along. This tag gets 903 * set during AH, ESP, etc. input handling, before the 904 * packet is returned to the ip input queue for delivery. 905 */ 906 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 907 s = splsoftnet(); 908 if (mtag != NULL) { 909 tdbi = (struct tdb_ident *)(mtag + 1); 910 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 911 } else { 912 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 913 IP_FORWARDING, &error); 914 } 915 if (sp != NULL) { 916 /* 917 * Check security policy against packet attributes. 918 */ 919 error = ipsec_in_reject(sp, m); 920 KEY_FREESP(&sp); 921 } else { 922 /* XXX error stat??? */ 923 error = EINVAL; 924DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/ 925 goto bad; 926 } 927 splx(s); 928 if (error) 929 goto bad; 930 } 931#endif /* FAST_IPSEC */ 932 933 /* 934 * Switch out to protocol's input routine. 935 */ 936#if IFA_STATS 937 if (ia && ip) 938 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 939#endif 940 ipstat.ips_delivered++; 941 { 942 int off = hlen, nh = ip->ip_p; 943 944 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 945 return; 946 } 947bad: 948 m_freem(m); 949 return; 950 951badcsum: 952 ipstat.ips_badsum++; 953 m_freem(m); 954} 955 956/* 957 * Take incoming datagram fragment and try to 958 * reassemble it into whole datagram. If a chain for 959 * reassembly of this datagram already exists, then it 960 * is given as fp; otherwise have to make a chain. 961 */ 962struct mbuf * 963ip_reass(ipqe, fp) 964 struct ipqent *ipqe; 965 struct ipq *fp; 966{ 967 struct mbuf *m = ipqe->ipqe_m; 968 struct ipqent *nq, *p, *q; 969 struct ip *ip; 970 struct mbuf *t; 971 int hlen = ipqe->ipqe_ip->ip_hl << 2; 972 int i, next; 973 974 IPQ_LOCK_CHECK(); 975 976 /* 977 * Presence of header sizes in mbufs 978 * would confuse code below. 979 */ 980 m->m_data += hlen; 981 m->m_len -= hlen; 982 983 /* 984 * If first fragment to arrive, create a reassembly queue. 985 */ 986 if (fp == 0) { 987 /* 988 * Enforce upper bound on number of fragmented packets 989 * for which we attempt reassembly; 990 * If maxfrag is 0, never accept fragments. 991 * If maxfrag is -1, accept all fragments without limitation. 992 */ 993 if (ip_maxfragpackets < 0) 994 ; 995 else if (ip_nfragpackets >= ip_maxfragpackets) 996 goto dropfrag; 997 ip_nfragpackets++; 998 MALLOC(fp, struct ipq *, sizeof (struct ipq), 999 M_FTABLE, M_NOWAIT); 1000 if (fp == NULL) 1001 goto dropfrag; 1002 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 1003 fp->ipq_ttl = IPFRAGTTL; 1004 fp->ipq_p = ipqe->ipqe_ip->ip_p; 1005 fp->ipq_id = ipqe->ipqe_ip->ip_id; 1006 TAILQ_INIT(&fp->ipq_fragq); 1007 fp->ipq_src = ipqe->ipqe_ip->ip_src; 1008 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 1009 p = NULL; 1010 goto insert; 1011 } 1012 1013 /* 1014 * Find a segment which begins after this one does. 1015 */ 1016 for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; 1017 p = q, q = TAILQ_NEXT(q, ipqe_q)) 1018 if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off)) 1019 break; 1020 1021 /* 1022 * If there is a preceding segment, it may provide some of 1023 * our data already. If so, drop the data from the incoming 1024 * segment. If it provides all of our data, drop us. 1025 */ 1026 if (p != NULL) { 1027 i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) - 1028 ntohs(ipqe->ipqe_ip->ip_off); 1029 if (i > 0) { 1030 if (i >= ntohs(ipqe->ipqe_ip->ip_len)) 1031 goto dropfrag; 1032 m_adj(ipqe->ipqe_m, i); 1033 ipqe->ipqe_ip->ip_off = 1034 htons(ntohs(ipqe->ipqe_ip->ip_off) + i); 1035 ipqe->ipqe_ip->ip_len = 1036 htons(ntohs(ipqe->ipqe_ip->ip_len) - i); 1037 } 1038 } 1039 1040 /* 1041 * While we overlap succeeding segments trim them or, 1042 * if they are completely covered, dequeue them. 1043 */ 1044 for (; q != NULL && 1045 ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) > 1046 ntohs(q->ipqe_ip->ip_off); q = nq) { 1047 i = (ntohs(ipqe->ipqe_ip->ip_off) + 1048 ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off); 1049 if (i < ntohs(q->ipqe_ip->ip_len)) { 1050 q->ipqe_ip->ip_len = 1051 htons(ntohs(q->ipqe_ip->ip_len) - i); 1052 q->ipqe_ip->ip_off = 1053 htons(ntohs(q->ipqe_ip->ip_off) + i); 1054 m_adj(q->ipqe_m, i); 1055 break; 1056 } 1057 nq = TAILQ_NEXT(q, ipqe_q); 1058 m_freem(q->ipqe_m); 1059 TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q); 1060 pool_put(&ipqent_pool, q); 1061 } 1062 1063insert: 1064 /* 1065 * Stick new segment in its place; 1066 * check for complete reassembly. 1067 */ 1068 if (p == NULL) { 1069 TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 1070 } else { 1071 TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q); 1072 } 1073 next = 0; 1074 for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; 1075 p = q, q = TAILQ_NEXT(q, ipqe_q)) { 1076 if (ntohs(q->ipqe_ip->ip_off) != next) 1077 return (0); 1078 next += ntohs(q->ipqe_ip->ip_len); 1079 } 1080 if (p->ipqe_mff) 1081 return (0); 1082 1083 /* 1084 * Reassembly is complete. Check for a bogus message size and 1085 * concatenate fragments. 1086 */ 1087 q = TAILQ_FIRST(&fp->ipq_fragq); 1088 ip = q->ipqe_ip; 1089 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 1090 ipstat.ips_toolong++; 1091 ip_freef(fp); 1092 return (0); 1093 } 1094 m = q->ipqe_m; 1095 t = m->m_next; 1096 m->m_next = 0; 1097 m_cat(m, t); 1098 nq = TAILQ_NEXT(q, ipqe_q); 1099 pool_put(&ipqent_pool, q); 1100 for (q = nq; q != NULL; q = nq) { 1101 t = q->ipqe_m; 1102 nq = TAILQ_NEXT(q, ipqe_q); 1103 pool_put(&ipqent_pool, q); 1104 m_cat(m, t); 1105 } 1106 1107 /* 1108 * Create header for new ip packet by 1109 * modifying header of first packet; 1110 * dequeue and discard fragment reassembly header. 1111 * Make header visible. 1112 */ 1113 ip->ip_len = htons(next); 1114 ip->ip_src = fp->ipq_src; 1115 ip->ip_dst = fp->ipq_dst; 1116 LIST_REMOVE(fp, ipq_q); 1117 FREE(fp, M_FTABLE); 1118 ip_nfragpackets--; 1119 m->m_len += (ip->ip_hl << 2); 1120 m->m_data -= (ip->ip_hl << 2); 1121 /* some debugging cruft by sklower, below, will go away soon */ 1122 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 1123 int plen = 0; 1124 for (t = m; t; t = t->m_next) 1125 plen += t->m_len; 1126 m->m_pkthdr.len = plen; 1127 } 1128 return (m); 1129 1130dropfrag: 1131 ipstat.ips_fragdropped++; 1132 m_freem(m); 1133 pool_put(&ipqent_pool, ipqe); 1134 return (0); 1135} 1136 1137/* 1138 * Free a fragment reassembly header and all 1139 * associated datagrams. 1140 */ 1141void 1142ip_freef(fp) 1143 struct ipq *fp; 1144{ 1145 struct ipqent *q, *p; 1146 1147 IPQ_LOCK_CHECK(); 1148 1149 for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) { 1150 p = TAILQ_NEXT(q, ipqe_q); 1151 m_freem(q->ipqe_m); 1152 TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q); 1153 pool_put(&ipqent_pool, q); 1154 } 1155 LIST_REMOVE(fp, ipq_q); 1156 FREE(fp, M_FTABLE); 1157 ip_nfragpackets--; 1158} 1159 1160/* 1161 * IP timer processing; 1162 * if a timer expires on a reassembly 1163 * queue, discard it. 1164 */ 1165void 1166ip_slowtimo() 1167{ 1168 struct ipq *fp, *nfp; 1169 int s = splsoftnet(); 1170 1171 IPQ_LOCK(); 1172 for (fp = LIST_FIRST(&ipq); fp != NULL; fp = nfp) { 1173 nfp = LIST_NEXT(fp, ipq_q); 1174 if (--fp->ipq_ttl == 0) { 1175 ipstat.ips_fragtimeout++; 1176 ip_freef(fp); 1177 } 1178 } 1179 /* 1180 * If we are over the maximum number of fragments 1181 * (due to the limit being lowered), drain off 1182 * enough to get down to the new limit. 1183 */ 1184 if (ip_maxfragpackets < 0) 1185 ; 1186 else { 1187 while (ip_nfragpackets > ip_maxfragpackets && LIST_FIRST(&ipq)) 1188 ip_freef(LIST_FIRST(&ipq)); 1189 } 1190 IPQ_UNLOCK(); 1191#ifdef GATEWAY 1192 ipflow_slowtimo(); 1193#endif 1194 splx(s); 1195} 1196 1197/* 1198 * Drain off all datagram fragments. 1199 */ 1200void 1201ip_drain() 1202{ 1203 1204 /* 1205 * We may be called from a device's interrupt context. If 1206 * the ipq is already busy, just bail out now. 1207 */ 1208 if (ipq_lock_try() == 0) 1209 return; 1210 1211 while (LIST_FIRST(&ipq) != NULL) { 1212 ipstat.ips_fragdropped++; 1213 ip_freef(LIST_FIRST(&ipq)); 1214 } 1215 1216 IPQ_UNLOCK(); 1217} 1218 1219/* 1220 * Do option processing on a datagram, 1221 * possibly discarding it if bad options are encountered, 1222 * or forwarding it if source-routed. 1223 * Returns 1 if packet has been forwarded/freed, 1224 * 0 if the packet should be processed further. 1225 */ 1226int 1227ip_dooptions(m) 1228 struct mbuf *m; 1229{ 1230 struct ip *ip = mtod(m, struct ip *); 1231 u_char *cp, *cp0; 1232 struct ip_timestamp *ipt; 1233 struct in_ifaddr *ia; 1234 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 1235 struct in_addr dst; 1236 n_time ntime; 1237 1238 dst = ip->ip_dst; 1239 cp = (u_char *)(ip + 1); 1240 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 1241 for (; cnt > 0; cnt -= optlen, cp += optlen) { 1242 opt = cp[IPOPT_OPTVAL]; 1243 if (opt == IPOPT_EOL) 1244 break; 1245 if (opt == IPOPT_NOP) 1246 optlen = 1; 1247 else { 1248 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 1249 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1250 goto bad; 1251 } 1252 optlen = cp[IPOPT_OLEN]; 1253 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 1254 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1255 goto bad; 1256 } 1257 } 1258 switch (opt) { 1259 1260 default: 1261 break; 1262 1263 /* 1264 * Source routing with record. 1265 * Find interface with current destination address. 1266 * If none on this machine then drop if strictly routed, 1267 * or do nothing if loosely routed. 1268 * Record interface address and bring up next address 1269 * component. If strictly routed make sure next 1270 * address is on directly accessible net. 1271 */ 1272 case IPOPT_LSRR: 1273 case IPOPT_SSRR: 1274 if (ip_allowsrcrt == 0) { 1275 type = ICMP_UNREACH; 1276 code = ICMP_UNREACH_NET_PROHIB; 1277 goto bad; 1278 } 1279 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1280 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1281 goto bad; 1282 } 1283 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1284 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1285 goto bad; 1286 } 1287 ipaddr.sin_addr = ip->ip_dst; 1288 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1289 if (ia == 0) { 1290 if (opt == IPOPT_SSRR) { 1291 type = ICMP_UNREACH; 1292 code = ICMP_UNREACH_SRCFAIL; 1293 goto bad; 1294 } 1295 /* 1296 * Loose routing, and not at next destination 1297 * yet; nothing to do except forward. 1298 */ 1299 break; 1300 } 1301 off--; /* 0 origin */ 1302 if ((off + sizeof(struct in_addr)) > optlen) { 1303 /* 1304 * End of source route. Should be for us. 1305 */ 1306 save_rte(cp, ip->ip_src); 1307 break; 1308 } 1309 /* 1310 * locate outgoing interface 1311 */ 1312 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 1313 sizeof(ipaddr.sin_addr)); 1314 if (opt == IPOPT_SSRR) 1315 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1316 else 1317 ia = ip_rtaddr(ipaddr.sin_addr); 1318 if (ia == 0) { 1319 type = ICMP_UNREACH; 1320 code = ICMP_UNREACH_SRCFAIL; 1321 goto bad; 1322 } 1323 ip->ip_dst = ipaddr.sin_addr; 1324 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1325 (caddr_t)(cp + off), sizeof(struct in_addr)); 1326 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1327 /* 1328 * Let ip_intr's mcast routing check handle mcast pkts 1329 */ 1330 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1331 break; 1332 1333 case IPOPT_RR: 1334 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1335 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1336 goto bad; 1337 } 1338 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1339 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1340 goto bad; 1341 } 1342 /* 1343 * If no space remains, ignore. 1344 */ 1345 off--; /* 0 origin */ 1346 if ((off + sizeof(struct in_addr)) > optlen) 1347 break; 1348 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 1349 sizeof(ipaddr.sin_addr)); 1350 /* 1351 * locate outgoing interface; if we're the destination, 1352 * use the incoming interface (should be same). 1353 */ 1354 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 1355 == NULL && 1356 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 1357 type = ICMP_UNREACH; 1358 code = ICMP_UNREACH_HOST; 1359 goto bad; 1360 } 1361 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1362 (caddr_t)(cp + off), sizeof(struct in_addr)); 1363 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1364 break; 1365 1366 case IPOPT_TS: 1367 code = cp - (u_char *)ip; 1368 ipt = (struct ip_timestamp *)cp; 1369 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1370 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1371 goto bad; 1372 } 1373 if (ipt->ipt_ptr < 5) { 1374 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1375 goto bad; 1376 } 1377 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1378 if (++ipt->ipt_oflw == 0) { 1379 code = (u_char *)&ipt->ipt_ptr - 1380 (u_char *)ip; 1381 goto bad; 1382 } 1383 break; 1384 } 1385 cp0 = (cp + ipt->ipt_ptr - 1); 1386 switch (ipt->ipt_flg) { 1387 1388 case IPOPT_TS_TSONLY: 1389 break; 1390 1391 case IPOPT_TS_TSANDADDR: 1392 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1393 sizeof(struct in_addr) > ipt->ipt_len) { 1394 code = (u_char *)&ipt->ipt_ptr - 1395 (u_char *)ip; 1396 goto bad; 1397 } 1398 ipaddr.sin_addr = dst; 1399 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1400 m->m_pkthdr.rcvif)); 1401 if (ia == 0) 1402 continue; 1403 bcopy(&ia->ia_addr.sin_addr, 1404 cp0, sizeof(struct in_addr)); 1405 ipt->ipt_ptr += sizeof(struct in_addr); 1406 break; 1407 1408 case IPOPT_TS_PRESPEC: 1409 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1410 sizeof(struct in_addr) > ipt->ipt_len) { 1411 code = (u_char *)&ipt->ipt_ptr - 1412 (u_char *)ip; 1413 goto bad; 1414 } 1415 bcopy(cp0, &ipaddr.sin_addr, 1416 sizeof(struct in_addr)); 1417 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1418 == NULL) 1419 continue; 1420 ipt->ipt_ptr += sizeof(struct in_addr); 1421 break; 1422 1423 default: 1424 /* XXX can't take &ipt->ipt_flg */ 1425 code = (u_char *)&ipt->ipt_ptr - 1426 (u_char *)ip + 1; 1427 goto bad; 1428 } 1429 ntime = iptime(); 1430 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1431 bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1, 1432 sizeof(n_time)); 1433 ipt->ipt_ptr += sizeof(n_time); 1434 } 1435 } 1436 if (forward) { 1437 if (ip_forwsrcrt == 0) { 1438 type = ICMP_UNREACH; 1439 code = ICMP_UNREACH_SRCFAIL; 1440 goto bad; 1441 } 1442 ip_forward(m, 1); 1443 return (1); 1444 } 1445 return (0); 1446bad: 1447 icmp_error(m, type, code, 0, 0); 1448 ipstat.ips_badoptions++; 1449 return (1); 1450} 1451 1452/* 1453 * Given address of next destination (final or next hop), 1454 * return internet address info of interface to be used to get there. 1455 */ 1456struct in_ifaddr * 1457ip_rtaddr(dst) 1458 struct in_addr dst; 1459{ 1460 struct sockaddr_in *sin; 1461 1462 sin = satosin(&ipforward_rt.ro_dst); 1463 1464 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 1465 if (ipforward_rt.ro_rt) { 1466 RTFREE(ipforward_rt.ro_rt); 1467 ipforward_rt.ro_rt = 0; 1468 } 1469 sin->sin_family = AF_INET; 1470 sin->sin_len = sizeof(*sin); 1471 sin->sin_addr = dst; 1472 1473 rtalloc(&ipforward_rt); 1474 } 1475 if (ipforward_rt.ro_rt == 0) 1476 return ((struct in_ifaddr *)0); 1477 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 1478} 1479 1480/* 1481 * Save incoming source route for use in replies, 1482 * to be picked up later by ip_srcroute if the receiver is interested. 1483 */ 1484void 1485save_rte(option, dst) 1486 u_char *option; 1487 struct in_addr dst; 1488{ 1489 unsigned olen; 1490 1491 olen = option[IPOPT_OLEN]; 1492#ifdef DIAGNOSTIC 1493 if (ipprintfs) 1494 printf("save_rte: olen %d\n", olen); 1495#endif /* 0 */ 1496 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1497 return; 1498 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 1499 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1500 ip_srcrt.dst = dst; 1501} 1502 1503/* 1504 * Retrieve incoming source route for use in replies, 1505 * in the same form used by setsockopt. 1506 * The first hop is placed before the options, will be removed later. 1507 */ 1508struct mbuf * 1509ip_srcroute() 1510{ 1511 struct in_addr *p, *q; 1512 struct mbuf *m; 1513 1514 if (ip_nhops == 0) 1515 return ((struct mbuf *)0); 1516 m = m_get(M_DONTWAIT, MT_SOOPTS); 1517 if (m == 0) 1518 return ((struct mbuf *)0); 1519 1520 MCLAIM(m, &inetdomain.dom_mowner); 1521#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1522 1523 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1524 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1525 OPTSIZ; 1526#ifdef DIAGNOSTIC 1527 if (ipprintfs) 1528 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1529#endif 1530 1531 /* 1532 * First save first hop for return route 1533 */ 1534 p = &ip_srcrt.route[ip_nhops - 1]; 1535 *(mtod(m, struct in_addr *)) = *p--; 1536#ifdef DIAGNOSTIC 1537 if (ipprintfs) 1538 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1539#endif 1540 1541 /* 1542 * Copy option fields and padding (nop) to mbuf. 1543 */ 1544 ip_srcrt.nop = IPOPT_NOP; 1545 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1546 bcopy((caddr_t)&ip_srcrt.nop, 1547 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 1548 q = (struct in_addr *)(mtod(m, caddr_t) + 1549 sizeof(struct in_addr) + OPTSIZ); 1550#undef OPTSIZ 1551 /* 1552 * Record return path as an IP source route, 1553 * reversing the path (pointers are now aligned). 1554 */ 1555 while (p >= ip_srcrt.route) { 1556#ifdef DIAGNOSTIC 1557 if (ipprintfs) 1558 printf(" %x", ntohl(q->s_addr)); 1559#endif 1560 *q++ = *p--; 1561 } 1562 /* 1563 * Last hop goes to final destination. 1564 */ 1565 *q = ip_srcrt.dst; 1566#ifdef DIAGNOSTIC 1567 if (ipprintfs) 1568 printf(" %x\n", ntohl(q->s_addr)); 1569#endif 1570 return (m); 1571} 1572 1573/* 1574 * Strip out IP options, at higher 1575 * level protocol in the kernel. 1576 * Second argument is buffer to which options 1577 * will be moved, and return value is their length. 1578 * XXX should be deleted; last arg currently ignored. 1579 */ 1580void 1581ip_stripoptions(m, mopt) 1582 struct mbuf *m; 1583 struct mbuf *mopt; 1584{ 1585 int i; 1586 struct ip *ip = mtod(m, struct ip *); 1587 caddr_t opts; 1588 int olen; 1589 1590 olen = (ip->ip_hl << 2) - sizeof (struct ip); 1591 opts = (caddr_t)(ip + 1); 1592 i = m->m_len - (sizeof (struct ip) + olen); 1593 bcopy(opts + olen, opts, (unsigned)i); 1594 m->m_len -= olen; 1595 if (m->m_flags & M_PKTHDR) 1596 m->m_pkthdr.len -= olen; 1597 ip->ip_len = htons(ntohs(ip->ip_len) - olen); 1598 ip->ip_hl = sizeof (struct ip) >> 2; 1599} 1600 1601const int inetctlerrmap[PRC_NCMDS] = { 1602 0, 0, 0, 0, 1603 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1604 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1605 EMSGSIZE, EHOSTUNREACH, 0, 0, 1606 0, 0, 0, 0, 1607 ENOPROTOOPT 1608}; 1609 1610/* 1611 * Forward a packet. If some error occurs return the sender 1612 * an icmp packet. Note we can't always generate a meaningful 1613 * icmp message because icmp doesn't have a large enough repertoire 1614 * of codes and types. 1615 * 1616 * If not forwarding, just drop the packet. This could be confusing 1617 * if ipforwarding was zero but some routing protocol was advancing 1618 * us as a gateway to somewhere. However, we must let the routing 1619 * protocol deal with that. 1620 * 1621 * The srcrt parameter indicates whether the packet is being forwarded 1622 * via a source route. 1623 */ 1624void 1625ip_forward(m, srcrt) 1626 struct mbuf *m; 1627 int srcrt; 1628{ 1629 struct ip *ip = mtod(m, struct ip *); 1630 struct sockaddr_in *sin; 1631 struct rtentry *rt; 1632 int error, type = 0, code = 0; 1633 struct mbuf *mcopy; 1634 n_long dest; 1635 struct ifnet *destifp; 1636#if defined(IPSEC) || defined(FAST_IPSEC) 1637 struct ifnet dummyifp; 1638#endif 1639 1640 /* 1641 * We are now in the output path. 1642 */ 1643 MCLAIM(m, &ip_tx_mowner); 1644 1645 /* 1646 * Clear any in-bound checksum flags for this packet. 1647 */ 1648 m->m_pkthdr.csum_flags = 0; 1649 1650 dest = 0; 1651#ifdef DIAGNOSTIC 1652 if (ipprintfs) 1653 printf("forward: src %2.2x dst %2.2x ttl %x\n", 1654 ntohl(ip->ip_src.s_addr), 1655 ntohl(ip->ip_dst.s_addr), ip->ip_ttl); 1656#endif 1657 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1658 ipstat.ips_cantforward++; 1659 m_freem(m); 1660 return; 1661 } 1662 if (ip->ip_ttl <= IPTTLDEC) { 1663 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1664 return; 1665 } 1666 ip->ip_ttl -= IPTTLDEC; 1667 1668 sin = satosin(&ipforward_rt.ro_dst); 1669 if ((rt = ipforward_rt.ro_rt) == 0 || 1670 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1671 if (ipforward_rt.ro_rt) { 1672 RTFREE(ipforward_rt.ro_rt); 1673 ipforward_rt.ro_rt = 0; 1674 } 1675 sin->sin_family = AF_INET; 1676 sin->sin_len = sizeof(struct sockaddr_in); 1677 sin->sin_addr = ip->ip_dst; 1678 1679 rtalloc(&ipforward_rt); 1680 if (ipforward_rt.ro_rt == 0) { 1681 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1682 return; 1683 } 1684 rt = ipforward_rt.ro_rt; 1685 } 1686 1687 /* 1688 * Save at most 68 bytes of the packet in case 1689 * we need to generate an ICMP message to the src. 1690 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1691 */ 1692 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1693 if (mcopy) 1694 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1695 1696 /* 1697 * If forwarding packet using same interface that it came in on, 1698 * perhaps should send a redirect to sender to shortcut a hop. 1699 * Only send redirect if source is sending directly to us, 1700 * and if packet was not source routed (or has any options). 1701 * Also, don't send redirect if forwarding using a default route 1702 * or a route modified by a redirect. 1703 */ 1704 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1705 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1706 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1707 ipsendredirects && !srcrt) { 1708 if (rt->rt_ifa && 1709 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1710 ifatoia(rt->rt_ifa)->ia_subnet) { 1711 if (rt->rt_flags & RTF_GATEWAY) 1712 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1713 else 1714 dest = ip->ip_dst.s_addr; 1715 /* 1716 * Router requirements says to only send host 1717 * redirects. 1718 */ 1719 type = ICMP_REDIRECT; 1720 code = ICMP_REDIRECT_HOST; 1721#ifdef DIAGNOSTIC 1722 if (ipprintfs) 1723 printf("redirect (%d) to %x\n", code, 1724 (u_int32_t)dest); 1725#endif 1726 } 1727 } 1728 1729 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1730 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1731 (struct ip_moptions *)NULL, (struct socket *)NULL); 1732 1733 if (error) 1734 ipstat.ips_cantforward++; 1735 else { 1736 ipstat.ips_forward++; 1737 if (type) 1738 ipstat.ips_redirectsent++; 1739 else { 1740 if (mcopy) { 1741#ifdef GATEWAY 1742 if (mcopy->m_flags & M_CANFASTFWD) 1743 ipflow_create(&ipforward_rt, mcopy); 1744#endif 1745 m_freem(mcopy); 1746 } 1747 return; 1748 } 1749 } 1750 if (mcopy == NULL) 1751 return; 1752 destifp = NULL; 1753 1754 switch (error) { 1755 1756 case 0: /* forwarded, but need redirect */ 1757 /* type, code set above */ 1758 break; 1759 1760 case ENETUNREACH: /* shouldn't happen, checked above */ 1761 case EHOSTUNREACH: 1762 case ENETDOWN: 1763 case EHOSTDOWN: 1764 default: 1765 type = ICMP_UNREACH; 1766 code = ICMP_UNREACH_HOST; 1767 break; 1768 1769 case EMSGSIZE: 1770 type = ICMP_UNREACH; 1771 code = ICMP_UNREACH_NEEDFRAG; 1772#if !defined(IPSEC) && !defined(FAST_IPSEC) 1773 if (ipforward_rt.ro_rt) 1774 destifp = ipforward_rt.ro_rt->rt_ifp; 1775#else 1776 /* 1777 * If the packet is routed over IPsec tunnel, tell the 1778 * originator the tunnel MTU. 1779 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz 1780 * XXX quickhack!!! 1781 */ 1782 if (ipforward_rt.ro_rt) { 1783 struct secpolicy *sp; 1784 int ipsecerror; 1785 size_t ipsechdr; 1786 struct route *ro; 1787 1788 sp = ipsec4_getpolicybyaddr(mcopy, 1789 IPSEC_DIR_OUTBOUND, IP_FORWARDING, 1790 &ipsecerror); 1791 1792 if (sp == NULL) 1793 destifp = ipforward_rt.ro_rt->rt_ifp; 1794 else { 1795 /* count IPsec header size */ 1796 ipsechdr = ipsec4_hdrsiz(mcopy, 1797 IPSEC_DIR_OUTBOUND, NULL); 1798 1799 /* 1800 * find the correct route for outer IPv4 1801 * header, compute tunnel MTU. 1802 * 1803 * XXX BUG ALERT 1804 * The "dummyifp" code relies upon the fact 1805 * that icmp_error() touches only ifp->if_mtu. 1806 */ 1807 /*XXX*/ 1808 destifp = NULL; 1809 if (sp->req != NULL 1810 && sp->req->sav != NULL 1811 && sp->req->sav->sah != NULL) { 1812 ro = &sp->req->sav->sah->sa_route; 1813 if (ro->ro_rt && ro->ro_rt->rt_ifp) { 1814 dummyifp.if_mtu = 1815 ro->ro_rt->rt_rmx.rmx_mtu ? 1816 ro->ro_rt->rt_rmx.rmx_mtu : 1817 ro->ro_rt->rt_ifp->if_mtu; 1818 dummyifp.if_mtu -= ipsechdr; 1819 destifp = &dummyifp; 1820 } 1821 } 1822 1823#ifdef IPSEC 1824 key_freesp(sp); 1825#else 1826 KEY_FREESP(&sp); 1827#endif 1828 } 1829 } 1830#endif /*IPSEC*/ 1831 ipstat.ips_cantfrag++; 1832 break; 1833 1834 case ENOBUFS: 1835#if 1 1836 /* 1837 * a router should not generate ICMP_SOURCEQUENCH as 1838 * required in RFC1812 Requirements for IP Version 4 Routers. 1839 * source quench could be a big problem under DoS attacks, 1840 * or if the underlying interface is rate-limited. 1841 */ 1842 if (mcopy) 1843 m_freem(mcopy); 1844 return; 1845#else 1846 type = ICMP_SOURCEQUENCH; 1847 code = 0; 1848 break; 1849#endif 1850 } 1851 icmp_error(mcopy, type, code, dest, destifp); 1852} 1853 1854void 1855ip_savecontrol(inp, mp, ip, m) 1856 struct inpcb *inp; 1857 struct mbuf **mp; 1858 struct ip *ip; 1859 struct mbuf *m; 1860{ 1861 1862 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1863 struct timeval tv; 1864 1865 microtime(&tv); 1866 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1867 SCM_TIMESTAMP, SOL_SOCKET); 1868 if (*mp) 1869 mp = &(*mp)->m_next; 1870 } 1871 if (inp->inp_flags & INP_RECVDSTADDR) { 1872 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1873 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1874 if (*mp) 1875 mp = &(*mp)->m_next; 1876 } 1877#ifdef notyet 1878 /* 1879 * XXX 1880 * Moving these out of udp_input() made them even more broken 1881 * than they already were. 1882 * - fenner@parc.xerox.com 1883 */ 1884 /* options were tossed already */ 1885 if (inp->inp_flags & INP_RECVOPTS) { 1886 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1887 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1888 if (*mp) 1889 mp = &(*mp)->m_next; 1890 } 1891 /* ip_srcroute doesn't do what we want here, need to fix */ 1892 if (inp->inp_flags & INP_RECVRETOPTS) { 1893 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1894 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1895 if (*mp) 1896 mp = &(*mp)->m_next; 1897 } 1898#endif 1899 if (inp->inp_flags & INP_RECVIF) { 1900 struct sockaddr_dl sdl; 1901 1902 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1903 sdl.sdl_family = AF_LINK; 1904 sdl.sdl_index = m->m_pkthdr.rcvif ? 1905 m->m_pkthdr.rcvif->if_index : 0; 1906 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1907 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1908 IP_RECVIF, IPPROTO_IP); 1909 if (*mp) 1910 mp = &(*mp)->m_next; 1911 } 1912} 1913 1914int 1915ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1916 int *name; 1917 u_int namelen; 1918 void *oldp; 1919 size_t *oldlenp; 1920 void *newp; 1921 size_t newlen; 1922{ 1923 extern int subnetsarelocal, hostzeroisbroadcast; 1924 1925 int error, old; 1926 1927 /* All sysctl names (except ifq.*) at this level are terminal. */ 1928 if ((namelen != 1) && !(namelen == 2 && name[0] == IPCTL_IFQ)) 1929 return (ENOTDIR); 1930 1931 switch (name[0]) { 1932 case IPCTL_FORWARDING: 1933 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1934 case IPCTL_SENDREDIRECTS: 1935 return (sysctl_int(oldp, oldlenp, newp, newlen, 1936 &ipsendredirects)); 1937 case IPCTL_DEFTTL: 1938 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1939#ifdef notyet 1940 case IPCTL_DEFMTU: 1941 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1942#endif 1943 case IPCTL_FORWSRCRT: 1944 /* Don't allow this to change in a secure environment. */ 1945 if (securelevel > 0) 1946 return (sysctl_rdint(oldp, oldlenp, newp, 1947 ip_forwsrcrt)); 1948 else 1949 return (sysctl_int(oldp, oldlenp, newp, newlen, 1950 &ip_forwsrcrt)); 1951 case IPCTL_DIRECTEDBCAST: 1952 return (sysctl_int(oldp, oldlenp, newp, newlen, 1953 &ip_directedbcast)); 1954 case IPCTL_ALLOWSRCRT: 1955 return (sysctl_int(oldp, oldlenp, newp, newlen, 1956 &ip_allowsrcrt)); 1957 case IPCTL_SUBNETSARELOCAL: 1958 return (sysctl_int(oldp, oldlenp, newp, newlen, 1959 &subnetsarelocal)); 1960 case IPCTL_MTUDISC: 1961 error = sysctl_int(oldp, oldlenp, newp, newlen, 1962 &ip_mtudisc); 1963 if (error == 0 && ip_mtudisc == 0) 1964 rt_timer_queue_remove_all(ip_mtudisc_timeout_q, TRUE); 1965 return error; 1966 case IPCTL_ANONPORTMIN: 1967 old = anonportmin; 1968 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1969 if (anonportmin >= anonportmax || anonportmin < 0 1970 || anonportmin > 65535 1971#ifndef IPNOPRIVPORTS 1972 || anonportmin < IPPORT_RESERVED 1973#endif 1974 ) { 1975 anonportmin = old; 1976 return (EINVAL); 1977 } 1978 return (error); 1979 case IPCTL_ANONPORTMAX: 1980 old = anonportmax; 1981 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1982 if (anonportmin >= anonportmax || anonportmax < 0 1983 || anonportmax > 65535 1984#ifndef IPNOPRIVPORTS 1985 || anonportmax < IPPORT_RESERVED 1986#endif 1987 ) { 1988 anonportmax = old; 1989 return (EINVAL); 1990 } 1991 return (error); 1992 case IPCTL_MTUDISCTIMEOUT: 1993 old = ip_mtudisc_timeout; 1994 error = sysctl_int(oldp, oldlenp, newp, newlen, 1995 &ip_mtudisc_timeout); 1996 if (ip_mtudisc_timeout < 0) { 1997 ip_mtudisc_timeout = old; 1998 return (EINVAL); 1999 } 2000 if (error == 0) 2001 rt_timer_queue_change(ip_mtudisc_timeout_q, 2002 ip_mtudisc_timeout); 2003 return (error); 2004#ifdef GATEWAY 2005 case IPCTL_MAXFLOWS: 2006 { 2007 int s; 2008 2009 error = sysctl_int(oldp, oldlenp, newp, newlen, 2010 &ip_maxflows); 2011 s = splsoftnet(); 2012 ipflow_reap(0); 2013 splx(s); 2014 return (error); 2015 } 2016#endif 2017 case IPCTL_HOSTZEROBROADCAST: 2018 return (sysctl_int(oldp, oldlenp, newp, newlen, 2019 &hostzeroisbroadcast)); 2020#if NGIF > 0 2021 case IPCTL_GIF_TTL: 2022 return (sysctl_int(oldp, oldlenp, newp, newlen, 2023 &ip_gif_ttl)); 2024#endif 2025 2026#if NGRE > 0 2027 case IPCTL_GRE_TTL: 2028 return (sysctl_int(oldp, oldlenp, newp, newlen, 2029 &ip_gre_ttl)); 2030#endif 2031 2032#ifndef IPNOPRIVPORTS 2033 case IPCTL_LOWPORTMIN: 2034 old = lowportmin; 2035 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin); 2036 if (lowportmin >= lowportmax 2037 || lowportmin > IPPORT_RESERVEDMAX 2038 || lowportmin < IPPORT_RESERVEDMIN 2039 ) { 2040 lowportmin = old; 2041 return (EINVAL); 2042 } 2043 return (error); 2044 case IPCTL_LOWPORTMAX: 2045 old = lowportmax; 2046 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax); 2047 if (lowportmin >= lowportmax 2048 || lowportmax > IPPORT_RESERVEDMAX 2049 || lowportmax < IPPORT_RESERVEDMIN 2050 ) { 2051 lowportmax = old; 2052 return (EINVAL); 2053 } 2054 return (error); 2055#endif 2056 2057 case IPCTL_MAXFRAGPACKETS: 2058 return (sysctl_int(oldp, oldlenp, newp, newlen, 2059 &ip_maxfragpackets)); 2060 2061 case IPCTL_CHECKINTERFACE: 2062 return (sysctl_int(oldp, oldlenp, newp, newlen, 2063 &ip_checkinterface)); 2064 2065 case IPCTL_IFQ: 2066 return (sysctl_ifq(name + 1, namelen - 1, oldp, oldlenp, 2067 newp, newlen, &ipintrq)); 2068 2069 default: 2070 return (EOPNOTSUPP); 2071 } 2072 /* NOTREACHED */ 2073} 2074