ip_input.c revision 1.93
1/* $NetBSD: ip_input.c,v 1.93 1999/10/17 16:00:00 sommerfeld 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. All advertising materials mentioning features or use of this software 82 * must display the following acknowledgement: 83 * This product includes software developed by the University of 84 * California, Berkeley and its contributors. 85 * 4. Neither the name of the University nor the names of its contributors 86 * may be used to endorse or promote products derived from this software 87 * without specific prior written permission. 88 * 89 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 90 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 91 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 92 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 93 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 94 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 95 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 96 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 97 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 98 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 99 * SUCH DAMAGE. 100 * 101 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 102 */ 103 104#include "opt_gateway.h" 105#include "opt_pfil_hooks.h" 106#include "opt_ipsec.h" 107#include "opt_mrouting.h" 108 109#include <sys/param.h> 110#include <sys/systm.h> 111#include <sys/malloc.h> 112#include <sys/mbuf.h> 113#include <sys/domain.h> 114#include <sys/protosw.h> 115#include <sys/socket.h> 116#include <sys/socketvar.h> 117#include <sys/errno.h> 118#include <sys/time.h> 119#include <sys/kernel.h> 120#include <sys/proc.h> 121#include <sys/pool.h> 122 123#include <vm/vm.h> 124#include <sys/sysctl.h> 125 126#include <net/if.h> 127#include <net/if_dl.h> 128#include <net/route.h> 129#include <net/pfil.h> 130 131#include <netinet/in.h> 132#include <netinet/in_systm.h> 133#include <netinet/ip.h> 134#include <netinet/in_pcb.h> 135#include <netinet/in_var.h> 136#include <netinet/ip_var.h> 137#include <netinet/ip_icmp.h> 138/* just for gif_ttl */ 139#include <netinet/in_gif.h> 140#include "gif.h" 141 142#ifdef IPSEC 143#include <netinet6/ipsec.h> 144#include <netkey/key.h> 145#include <netkey/key_debug.h> 146#endif 147 148#ifndef IPFORWARDING 149#ifdef GATEWAY 150#define IPFORWARDING 1 /* forward IP packets not for us */ 151#else /* GATEWAY */ 152#define IPFORWARDING 0 /* don't forward IP packets not for us */ 153#endif /* GATEWAY */ 154#endif /* IPFORWARDING */ 155#ifndef IPSENDREDIRECTS 156#define IPSENDREDIRECTS 1 157#endif 158#ifndef IPFORWSRCRT 159#define IPFORWSRCRT 1 /* forward source-routed packets */ 160#endif 161#ifndef IPALLOWSRCRT 162#define IPALLOWSRCRT 1 /* allow source-routed packets */ 163#endif 164#ifndef IPMTUDISC 165#define IPMTUDISC 0 166#endif 167#ifndef IPMTUDISCTIMEOUT 168#define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 169#endif 170 171/* 172 * Note: DIRECTED_BROADCAST is handled this way so that previous 173 * configuration using this option will Just Work. 174 */ 175#ifndef IPDIRECTEDBCAST 176#ifdef DIRECTED_BROADCAST 177#define IPDIRECTEDBCAST 1 178#else 179#define IPDIRECTEDBCAST 0 180#endif /* DIRECTED_BROADCAST */ 181#endif /* IPDIRECTEDBCAST */ 182int ipforwarding = IPFORWARDING; 183int ipsendredirects = IPSENDREDIRECTS; 184int ip_defttl = IPDEFTTL; 185int ip_forwsrcrt = IPFORWSRCRT; 186int ip_directedbcast = IPDIRECTEDBCAST; 187int ip_allowsrcrt = IPALLOWSRCRT; 188int ip_mtudisc = IPMTUDISC; 189u_int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 190#ifdef DIAGNOSTIC 191int ipprintfs = 0; 192#endif 193 194struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 195 196extern struct domain inetdomain; 197extern struct protosw inetsw[]; 198u_char ip_protox[IPPROTO_MAX]; 199int ipqmaxlen = IFQ_MAXLEN; 200struct in_ifaddrhead in_ifaddr; 201struct in_ifaddrhashhead *in_ifaddrhashtbl; 202struct ifqueue ipintrq; 203struct ipstat ipstat; 204u_int16_t ip_id; 205int ip_defttl; 206 207struct ipqhead ipq; 208int ipq_locked; 209 210static __inline int ipq_lock_try __P((void)); 211static __inline void ipq_unlock __P((void)); 212 213static __inline int 214ipq_lock_try() 215{ 216 int s; 217 218 s = splimp(); 219 if (ipq_locked) { 220 splx(s); 221 return (0); 222 } 223 ipq_locked = 1; 224 splx(s); 225 return (1); 226} 227 228static __inline void 229ipq_unlock() 230{ 231 int s; 232 233 s = splimp(); 234 ipq_locked = 0; 235 splx(s); 236} 237 238#ifdef DIAGNOSTIC 239#define IPQ_LOCK() \ 240do { \ 241 if (ipq_lock_try() == 0) { \ 242 printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \ 243 panic("ipq_lock"); \ 244 } \ 245} while (0) 246#define IPQ_LOCK_CHECK() \ 247do { \ 248 if (ipq_locked == 0) { \ 249 printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \ 250 panic("ipq lock check"); \ 251 } \ 252} while (0) 253#else 254#define IPQ_LOCK() (void) ipq_lock_try() 255#define IPQ_LOCK_CHECK() /* nothing */ 256#endif 257 258#define IPQ_UNLOCK() ipq_unlock() 259 260struct pool ipqent_pool; 261 262/* 263 * We need to save the IP options in case a protocol wants to respond 264 * to an incoming packet over the same route if the packet got here 265 * using IP source routing. This allows connection establishment and 266 * maintenance when the remote end is on a network that is not known 267 * to us. 268 */ 269int ip_nhops = 0; 270static struct ip_srcrt { 271 struct in_addr dst; /* final destination */ 272 char nop; /* one NOP to align */ 273 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 274 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 275} ip_srcrt; 276 277static void save_rte __P((u_char *, struct in_addr)); 278 279/* 280 * IP initialization: fill in IP protocol switch table. 281 * All protocols not implemented in kernel go to raw IP protocol handler. 282 */ 283void 284ip_init() 285{ 286 register struct protosw *pr; 287 register int i; 288 289 pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", 290 0, NULL, NULL, M_IPQ); 291 292 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 293 if (pr == 0) 294 panic("ip_init"); 295 for (i = 0; i < IPPROTO_MAX; i++) 296 ip_protox[i] = pr - inetsw; 297 for (pr = inetdomain.dom_protosw; 298 pr < inetdomain.dom_protoswNPROTOSW; pr++) 299 if (pr->pr_domain->dom_family == PF_INET && 300 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 301 ip_protox[pr->pr_protocol] = pr - inetsw; 302 LIST_INIT(&ipq); 303 ip_id = time.tv_sec & 0xffff; 304 ipintrq.ifq_maxlen = ipqmaxlen; 305 TAILQ_INIT(&in_ifaddr); 306 in_ifaddrhashtbl = 307 hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, M_WAITOK, &in_ifaddrhash); 308 if (ip_mtudisc != 0) 309 ip_mtudisc_timeout_q = 310 rt_timer_queue_create(ip_mtudisc_timeout); 311#ifdef GATEWAY 312 ipflow_init(); 313#endif 314} 315 316struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 317struct route ipforward_rt; 318 319/* 320 * IP software interrupt routine 321 */ 322void 323ipintr() 324{ 325 int s; 326 struct mbuf *m; 327 328 while (1) { 329 s = splimp(); 330 IF_DEQUEUE(&ipintrq, m); 331 splx(s); 332 if (m == 0) 333 return; 334 ip_input(m); 335 } 336} 337 338/* 339 * Ip input routine. Checksum and byte swap header. If fragmented 340 * try to reassemble. Process options. Pass to next level. 341 */ 342void 343ip_input(struct mbuf *m) 344{ 345 register struct ip *ip = NULL; 346 register struct ipq *fp; 347 register struct in_ifaddr *ia; 348 register struct ifaddr *ifa; 349 struct ipqent *ipqe; 350 int hlen = 0, mff, len; 351#ifdef PFIL_HOOKS 352 struct packet_filter_hook *pfh; 353 struct mbuf *m0; 354 int rv; 355#endif /* PFIL_HOOKS */ 356 357#ifdef DIAGNOSTIC 358 if ((m->m_flags & M_PKTHDR) == 0) 359 panic("ipintr no HDR"); 360#endif 361#ifdef IPSEC 362 /* 363 * should the inner packet be considered authentic? 364 * see comment in ah4_input(). 365 */ 366 if (m) { 367 m->m_flags &= ~M_AUTHIPHDR; 368 m->m_flags &= ~M_AUTHIPDGM; 369 } 370#endif 371 /* 372 * If no IP addresses have been set yet but the interfaces 373 * are receiving, can't do anything with incoming packets yet. 374 */ 375 if (in_ifaddr.tqh_first == 0) 376 goto bad; 377 ipstat.ips_total++; 378 if (m->m_len < sizeof (struct ip) && 379 (m = m_pullup(m, sizeof (struct ip))) == 0) { 380 ipstat.ips_toosmall++; 381 return; 382 } 383 ip = mtod(m, struct ip *); 384 if (ip->ip_v != IPVERSION) { 385 ipstat.ips_badvers++; 386 goto bad; 387 } 388 hlen = ip->ip_hl << 2; 389 if (hlen < sizeof(struct ip)) { /* minimum header length */ 390 ipstat.ips_badhlen++; 391 goto bad; 392 } 393 if (hlen > m->m_len) { 394 if ((m = m_pullup(m, hlen)) == 0) { 395 ipstat.ips_badhlen++; 396 return; 397 } 398 ip = mtod(m, struct ip *); 399 } 400 /* 401 * we drop packets that have a multicast address as source 402 * as wanted by rfc 1112 403 */ 404 if (IN_MULTICAST(ip->ip_src.s_addr)) { 405 goto bad; 406 } 407 408 if (in_cksum(m, hlen) != 0) { 409 ipstat.ips_badsum++; 410 goto bad; 411 } 412 413 /* 414 * Convert fields to host representation. 415 */ 416 NTOHS(ip->ip_len); 417 NTOHS(ip->ip_off); 418 len = ip->ip_len; 419 420 /* 421 * Check for additional length bogosity 422 */ 423 if (len < hlen) { 424 ipstat.ips_badlen++; 425 goto bad; 426 } 427 428 /* 429 * Check that the amount of data in the buffers 430 * is as at least much as the IP header would have us expect. 431 * Trim mbufs if longer than we expect. 432 * Drop packet if shorter than we expect. 433 */ 434 if (m->m_pkthdr.len < len) { 435 ipstat.ips_tooshort++; 436 goto bad; 437 } 438 if (m->m_pkthdr.len > len) { 439 if (m->m_len == m->m_pkthdr.len) { 440 m->m_len = len; 441 m->m_pkthdr.len = len; 442 } else 443 m_adj(m, len - m->m_pkthdr.len); 444 } 445 446 /* 447 * Assume that we can create a fast-forward IP flow entry 448 * based on this packet. 449 */ 450 m->m_flags |= M_CANFASTFWD; 451 452#ifdef PFIL_HOOKS 453 /* 454 * Run through list of hooks for input packets. If there are any 455 * filters which require that additional packets in the flow are 456 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 457 * Note that filters must _never_ set this flag, as another filter 458 * in the list may have previously cleared it. 459 */ 460 m0 = m; 461 for (pfh = pfil_hook_get(PFIL_IN); pfh; pfh = pfh->pfil_link.tqe_next) 462 if (pfh->pfil_func) { 463 rv = pfh->pfil_func(ip, hlen, m->m_pkthdr.rcvif, 0, &m0); 464 if (rv) 465 return; 466 m = m0; 467 if (m == NULL) 468 return; 469 ip = mtod(m, struct ip *); 470 } 471#endif /* PFIL_HOOKS */ 472 473 /* 474 * Process options and, if not destined for us, 475 * ship it on. ip_dooptions returns 1 when an 476 * error was detected (causing an icmp message 477 * to be sent and the original packet to be freed). 478 */ 479 ip_nhops = 0; /* for source routed packets */ 480 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 481 return; 482 483 /* 484 * Check our list of addresses, to see if the packet is for us. 485 */ 486 INADDR_TO_IA(ip->ip_dst, ia); 487 if (ia != NULL) 488 goto ours; 489 if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 490 for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first; 491 ifa != NULL; ifa = ifa->ifa_list.tqe_next) { 492 if (ifa->ifa_addr->sa_family != AF_INET) continue; 493 ia = ifatoia(ifa); 494 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 495 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 496 /* 497 * Look for all-0's host part (old broadcast addr), 498 * either for subnet or net. 499 */ 500 ip->ip_dst.s_addr == ia->ia_subnet || 501 ip->ip_dst.s_addr == ia->ia_net) 502 goto ours; 503 /* 504 * An interface with IP address zero accepts 505 * all packets that arrive on that interface. 506 */ 507 if (in_nullhost(ia->ia_addr.sin_addr)) 508 goto ours; 509 } 510 } 511 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 512 struct in_multi *inm; 513#ifdef MROUTING 514 extern struct socket *ip_mrouter; 515 516 if (m->m_flags & M_EXT) { 517 if ((m = m_pullup(m, hlen)) == 0) { 518 ipstat.ips_toosmall++; 519 return; 520 } 521 ip = mtod(m, struct ip *); 522 } 523 524 if (ip_mrouter) { 525 /* 526 * If we are acting as a multicast router, all 527 * incoming multicast packets are passed to the 528 * kernel-level multicast forwarding function. 529 * The packet is returned (relatively) intact; if 530 * ip_mforward() returns a non-zero value, the packet 531 * must be discarded, else it may be accepted below. 532 * 533 * (The IP ident field is put in the same byte order 534 * as expected when ip_mforward() is called from 535 * ip_output().) 536 */ 537 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 538 ipstat.ips_cantforward++; 539 m_freem(m); 540 return; 541 } 542 543 /* 544 * The process-level routing demon needs to receive 545 * all multicast IGMP packets, whether or not this 546 * host belongs to their destination groups. 547 */ 548 if (ip->ip_p == IPPROTO_IGMP) 549 goto ours; 550 ipstat.ips_forward++; 551 } 552#endif 553 /* 554 * See if we belong to the destination multicast group on the 555 * arrival interface. 556 */ 557 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 558 if (inm == NULL) { 559 ipstat.ips_cantforward++; 560 m_freem(m); 561 return; 562 } 563 goto ours; 564 } 565 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 566 in_nullhost(ip->ip_dst)) 567 goto ours; 568 569 /* 570 * Not for us; forward if possible and desirable. 571 */ 572 if (ipforwarding == 0) { 573 ipstat.ips_cantforward++; 574 m_freem(m); 575 } else 576 ip_forward(m, 0); 577 return; 578 579ours: 580 /* 581 * If offset or IP_MF are set, must reassemble. 582 * Otherwise, nothing need be done. 583 * (We could look in the reassembly queue to see 584 * if the packet was previously fragmented, 585 * but it's not worth the time; just let them time out.) 586 */ 587 if (ip->ip_off & ~(IP_DF|IP_RF)) { 588 /* 589 * Look for queue of fragments 590 * of this datagram. 591 */ 592 IPQ_LOCK(); 593 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 594 if (ip->ip_id == fp->ipq_id && 595 in_hosteq(ip->ip_src, fp->ipq_src) && 596 in_hosteq(ip->ip_dst, fp->ipq_dst) && 597 ip->ip_p == fp->ipq_p) 598 goto found; 599 fp = 0; 600found: 601 602 /* 603 * Adjust ip_len to not reflect header, 604 * set ipqe_mff if more fragments are expected, 605 * convert offset of this to bytes. 606 */ 607 ip->ip_len -= hlen; 608 mff = (ip->ip_off & IP_MF) != 0; 609 if (mff) { 610 /* 611 * Make sure that fragments have a data length 612 * that's a non-zero multiple of 8 bytes. 613 */ 614 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 615 ipstat.ips_badfrags++; 616 IPQ_UNLOCK(); 617 goto bad; 618 } 619 } 620 ip->ip_off <<= 3; 621 622 /* 623 * If datagram marked as having more fragments 624 * or if this is not the first fragment, 625 * attempt reassembly; if it succeeds, proceed. 626 */ 627 if (mff || ip->ip_off) { 628 ipstat.ips_fragments++; 629 ipqe = pool_get(&ipqent_pool, PR_NOWAIT); 630 if (ipqe == NULL) { 631 ipstat.ips_rcvmemdrop++; 632 IPQ_UNLOCK(); 633 goto bad; 634 } 635 ipqe->ipqe_mff = mff; 636 ipqe->ipqe_m = m; 637 ipqe->ipqe_ip = ip; 638 m = ip_reass(ipqe, fp); 639 if (m == 0) { 640 IPQ_UNLOCK(); 641 return; 642 } 643 ipstat.ips_reassembled++; 644 ip = mtod(m, struct ip *); 645 hlen = ip->ip_hl << 2; 646 ip->ip_len += hlen; 647 } else 648 if (fp) 649 ip_freef(fp); 650 IPQ_UNLOCK(); 651 } 652 653 /* 654 * Switch out to protocol's input routine. 655 */ 656#if IFA_STATS 657 ia->ia_ifa.ifa_data.ifad_inbytes += ip->ip_len; 658#endif 659 ipstat.ips_delivered++; 660 { 661 int off = hlen, nh = ip->ip_p; 662 663 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 664 return; 665 } 666bad: 667 m_freem(m); 668} 669 670/* 671 * Take incoming datagram fragment and try to 672 * reassemble it into whole datagram. If a chain for 673 * reassembly of this datagram already exists, then it 674 * is given as fp; otherwise have to make a chain. 675 */ 676struct mbuf * 677ip_reass(ipqe, fp) 678 register struct ipqent *ipqe; 679 register struct ipq *fp; 680{ 681 register struct mbuf *m = ipqe->ipqe_m; 682 register struct ipqent *nq, *p, *q; 683 struct ip *ip; 684 struct mbuf *t; 685 int hlen = ipqe->ipqe_ip->ip_hl << 2; 686 int i, next; 687 688 IPQ_LOCK_CHECK(); 689 690 /* 691 * Presence of header sizes in mbufs 692 * would confuse code below. 693 */ 694 m->m_data += hlen; 695 m->m_len -= hlen; 696 697 /* 698 * If first fragment to arrive, create a reassembly queue. 699 */ 700 if (fp == 0) { 701 MALLOC(fp, struct ipq *, sizeof (struct ipq), 702 M_FTABLE, M_NOWAIT); 703 if (fp == NULL) 704 goto dropfrag; 705 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 706 fp->ipq_ttl = IPFRAGTTL; 707 fp->ipq_p = ipqe->ipqe_ip->ip_p; 708 fp->ipq_id = ipqe->ipqe_ip->ip_id; 709 LIST_INIT(&fp->ipq_fragq); 710 fp->ipq_src = ipqe->ipqe_ip->ip_src; 711 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 712 p = NULL; 713 goto insert; 714 } 715 716 /* 717 * Find a segment which begins after this one does. 718 */ 719 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 720 p = q, q = q->ipqe_q.le_next) 721 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 722 break; 723 724 /* 725 * If there is a preceding segment, it may provide some of 726 * our data already. If so, drop the data from the incoming 727 * segment. If it provides all of our data, drop us. 728 */ 729 if (p != NULL) { 730 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 731 ipqe->ipqe_ip->ip_off; 732 if (i > 0) { 733 if (i >= ipqe->ipqe_ip->ip_len) 734 goto dropfrag; 735 m_adj(ipqe->ipqe_m, i); 736 ipqe->ipqe_ip->ip_off += i; 737 ipqe->ipqe_ip->ip_len -= i; 738 } 739 } 740 741 /* 742 * While we overlap succeeding segments trim them or, 743 * if they are completely covered, dequeue them. 744 */ 745 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 746 q->ipqe_ip->ip_off; q = nq) { 747 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 748 q->ipqe_ip->ip_off; 749 if (i < q->ipqe_ip->ip_len) { 750 q->ipqe_ip->ip_len -= i; 751 q->ipqe_ip->ip_off += i; 752 m_adj(q->ipqe_m, i); 753 break; 754 } 755 nq = q->ipqe_q.le_next; 756 m_freem(q->ipqe_m); 757 LIST_REMOVE(q, ipqe_q); 758 pool_put(&ipqent_pool, q); 759 } 760 761insert: 762 /* 763 * Stick new segment in its place; 764 * check for complete reassembly. 765 */ 766 if (p == NULL) { 767 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 768 } else { 769 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 770 } 771 next = 0; 772 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 773 p = q, q = q->ipqe_q.le_next) { 774 if (q->ipqe_ip->ip_off != next) 775 return (0); 776 next += q->ipqe_ip->ip_len; 777 } 778 if (p->ipqe_mff) 779 return (0); 780 781 /* 782 * Reassembly is complete. Check for a bogus message size and 783 * concatenate fragments. 784 */ 785 q = fp->ipq_fragq.lh_first; 786 ip = q->ipqe_ip; 787 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 788 ipstat.ips_toolong++; 789 ip_freef(fp); 790 return (0); 791 } 792 m = q->ipqe_m; 793 t = m->m_next; 794 m->m_next = 0; 795 m_cat(m, t); 796 nq = q->ipqe_q.le_next; 797 pool_put(&ipqent_pool, q); 798 for (q = nq; q != NULL; q = nq) { 799 t = q->ipqe_m; 800 nq = q->ipqe_q.le_next; 801 pool_put(&ipqent_pool, q); 802 m_cat(m, t); 803 } 804 805 /* 806 * Create header for new ip packet by 807 * modifying header of first packet; 808 * dequeue and discard fragment reassembly header. 809 * Make header visible. 810 */ 811 ip->ip_len = next; 812 ip->ip_ttl = 0; /* xxx */ 813 ip->ip_sum = 0; 814 ip->ip_src = fp->ipq_src; 815 ip->ip_dst = fp->ipq_dst; 816 LIST_REMOVE(fp, ipq_q); 817 FREE(fp, M_FTABLE); 818 m->m_len += (ip->ip_hl << 2); 819 m->m_data -= (ip->ip_hl << 2); 820 /* some debugging cruft by sklower, below, will go away soon */ 821 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 822 register int plen = 0; 823 for (t = m; t; t = t->m_next) 824 plen += t->m_len; 825 m->m_pkthdr.len = plen; 826 } 827 return (m); 828 829dropfrag: 830 ipstat.ips_fragdropped++; 831 m_freem(m); 832 pool_put(&ipqent_pool, ipqe); 833 return (0); 834} 835 836/* 837 * Free a fragment reassembly header and all 838 * associated datagrams. 839 */ 840void 841ip_freef(fp) 842 struct ipq *fp; 843{ 844 register struct ipqent *q, *p; 845 846 IPQ_LOCK_CHECK(); 847 848 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 849 p = q->ipqe_q.le_next; 850 m_freem(q->ipqe_m); 851 LIST_REMOVE(q, ipqe_q); 852 pool_put(&ipqent_pool, q); 853 } 854 LIST_REMOVE(fp, ipq_q); 855 FREE(fp, M_FTABLE); 856} 857 858/* 859 * IP timer processing; 860 * if a timer expires on a reassembly 861 * queue, discard it. 862 */ 863void 864ip_slowtimo() 865{ 866 register struct ipq *fp, *nfp; 867 int s = splsoftnet(); 868 869 IPQ_LOCK(); 870 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 871 nfp = fp->ipq_q.le_next; 872 if (--fp->ipq_ttl == 0) { 873 ipstat.ips_fragtimeout++; 874 ip_freef(fp); 875 } 876 } 877 IPQ_UNLOCK(); 878#ifdef GATEWAY 879 ipflow_slowtimo(); 880#endif 881 splx(s); 882} 883 884/* 885 * Drain off all datagram fragments. 886 */ 887void 888ip_drain() 889{ 890 891 /* 892 * We may be called from a device's interrupt context. If 893 * the ipq is already busy, just bail out now. 894 */ 895 if (ipq_lock_try() == 0) 896 return; 897 898 while (ipq.lh_first != NULL) { 899 ipstat.ips_fragdropped++; 900 ip_freef(ipq.lh_first); 901 } 902 903 IPQ_UNLOCK(); 904} 905 906/* 907 * Do option processing on a datagram, 908 * possibly discarding it if bad options are encountered, 909 * or forwarding it if source-routed. 910 * Returns 1 if packet has been forwarded/freed, 911 * 0 if the packet should be processed further. 912 */ 913int 914ip_dooptions(m) 915 struct mbuf *m; 916{ 917 register struct ip *ip = mtod(m, struct ip *); 918 register u_char *cp; 919 register struct ip_timestamp *ipt; 920 register struct in_ifaddr *ia; 921 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 922 struct in_addr *sin, dst; 923 n_time ntime; 924 925 dst = ip->ip_dst; 926 cp = (u_char *)(ip + 1); 927 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 928 for (; cnt > 0; cnt -= optlen, cp += optlen) { 929 opt = cp[IPOPT_OPTVAL]; 930 if (opt == IPOPT_EOL) 931 break; 932 if (opt == IPOPT_NOP) 933 optlen = 1; 934 else { 935 optlen = cp[IPOPT_OLEN]; 936 if (optlen <= 0 || optlen > cnt) { 937 code = &cp[IPOPT_OLEN] - (u_char *)ip; 938 goto bad; 939 } 940 } 941 switch (opt) { 942 943 default: 944 break; 945 946 /* 947 * Source routing with record. 948 * Find interface with current destination address. 949 * If none on this machine then drop if strictly routed, 950 * or do nothing if loosely routed. 951 * Record interface address and bring up next address 952 * component. If strictly routed make sure next 953 * address is on directly accessible net. 954 */ 955 case IPOPT_LSRR: 956 case IPOPT_SSRR: 957 if (ip_allowsrcrt == 0) { 958 type = ICMP_UNREACH; 959 code = ICMP_UNREACH_NET_PROHIB; 960 goto bad; 961 } 962 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 963 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 964 goto bad; 965 } 966 ipaddr.sin_addr = ip->ip_dst; 967 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 968 if (ia == 0) { 969 if (opt == IPOPT_SSRR) { 970 type = ICMP_UNREACH; 971 code = ICMP_UNREACH_SRCFAIL; 972 goto bad; 973 } 974 /* 975 * Loose routing, and not at next destination 976 * yet; nothing to do except forward. 977 */ 978 break; 979 } 980 off--; /* 0 origin */ 981 if (off > optlen - sizeof(struct in_addr)) { 982 /* 983 * End of source route. Should be for us. 984 */ 985 save_rte(cp, ip->ip_src); 986 break; 987 } 988 /* 989 * locate outgoing interface 990 */ 991 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 992 sizeof(ipaddr.sin_addr)); 993 if (opt == IPOPT_SSRR) { 994#define INA struct in_ifaddr * 995#define SA struct sockaddr * 996 ia = (INA)ifa_ifwithladdr((SA)&ipaddr); 997 } else 998 ia = ip_rtaddr(ipaddr.sin_addr); 999 if (ia == 0) { 1000 type = ICMP_UNREACH; 1001 code = ICMP_UNREACH_SRCFAIL; 1002 goto bad; 1003 } 1004 ip->ip_dst = ipaddr.sin_addr; 1005 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1006 (caddr_t)(cp + off), sizeof(struct in_addr)); 1007 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1008 /* 1009 * Let ip_intr's mcast routing check handle mcast pkts 1010 */ 1011 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1012 break; 1013 1014 case IPOPT_RR: 1015 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1016 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1017 goto bad; 1018 } 1019 /* 1020 * If no space remains, ignore. 1021 */ 1022 off--; /* 0 origin */ 1023 if (off > optlen - sizeof(struct in_addr)) 1024 break; 1025 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 1026 sizeof(ipaddr.sin_addr)); 1027 /* 1028 * locate outgoing interface; if we're the destination, 1029 * use the incoming interface (should be same). 1030 */ 1031 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == 0 && 1032 (ia = ip_rtaddr(ipaddr.sin_addr)) == 0) { 1033 type = ICMP_UNREACH; 1034 code = ICMP_UNREACH_HOST; 1035 goto bad; 1036 } 1037 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1038 (caddr_t)(cp + off), sizeof(struct in_addr)); 1039 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1040 break; 1041 1042 case IPOPT_TS: 1043 code = cp - (u_char *)ip; 1044 ipt = (struct ip_timestamp *)cp; 1045 if (ipt->ipt_len < 5) 1046 goto bad; 1047 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1048 if (++ipt->ipt_oflw == 0) 1049 goto bad; 1050 break; 1051 } 1052 sin = (struct in_addr *)(cp + ipt->ipt_ptr - 1); 1053 switch (ipt->ipt_flg) { 1054 1055 case IPOPT_TS_TSONLY: 1056 break; 1057 1058 case IPOPT_TS_TSANDADDR: 1059 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1060 sizeof(struct in_addr) > ipt->ipt_len) 1061 goto bad; 1062 ipaddr.sin_addr = dst; 1063 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr, 1064 m->m_pkthdr.rcvif); 1065 if (ia == 0) 1066 continue; 1067 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1068 (caddr_t)sin, sizeof(struct in_addr)); 1069 ipt->ipt_ptr += sizeof(struct in_addr); 1070 break; 1071 1072 case IPOPT_TS_PRESPEC: 1073 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1074 sizeof(struct in_addr) > ipt->ipt_len) 1075 goto bad; 1076 bcopy((caddr_t)sin, (caddr_t)&ipaddr.sin_addr, 1077 sizeof(struct in_addr)); 1078 if (ifa_ifwithaddr((SA)&ipaddr) == 0) 1079 continue; 1080 ipt->ipt_ptr += sizeof(struct in_addr); 1081 break; 1082 1083 default: 1084 goto bad; 1085 } 1086 ntime = iptime(); 1087 bcopy((caddr_t)&ntime, (caddr_t)cp + ipt->ipt_ptr - 1, 1088 sizeof(n_time)); 1089 ipt->ipt_ptr += sizeof(n_time); 1090 } 1091 } 1092 if (forward) { 1093 if (ip_forwsrcrt == 0) { 1094 type = ICMP_UNREACH; 1095 code = ICMP_UNREACH_SRCFAIL; 1096 goto bad; 1097 } 1098 ip_forward(m, 1); 1099 return (1); 1100 } 1101 return (0); 1102bad: 1103 icmp_error(m, type, code, 0, 0); 1104 ipstat.ips_badoptions++; 1105 return (1); 1106} 1107 1108/* 1109 * Given address of next destination (final or next hop), 1110 * return internet address info of interface to be used to get there. 1111 */ 1112struct in_ifaddr * 1113ip_rtaddr(dst) 1114 struct in_addr dst; 1115{ 1116 register struct sockaddr_in *sin; 1117 1118 sin = satosin(&ipforward_rt.ro_dst); 1119 1120 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 1121 if (ipforward_rt.ro_rt) { 1122 RTFREE(ipforward_rt.ro_rt); 1123 ipforward_rt.ro_rt = 0; 1124 } 1125 sin->sin_family = AF_INET; 1126 sin->sin_len = sizeof(*sin); 1127 sin->sin_addr = dst; 1128 1129 rtalloc(&ipforward_rt); 1130 } 1131 if (ipforward_rt.ro_rt == 0) 1132 return ((struct in_ifaddr *)0); 1133 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 1134} 1135 1136/* 1137 * Save incoming source route for use in replies, 1138 * to be picked up later by ip_srcroute if the receiver is interested. 1139 */ 1140void 1141save_rte(option, dst) 1142 u_char *option; 1143 struct in_addr dst; 1144{ 1145 unsigned olen; 1146 1147 olen = option[IPOPT_OLEN]; 1148#ifdef DIAGNOSTIC 1149 if (ipprintfs) 1150 printf("save_rte: olen %d\n", olen); 1151#endif /* 0 */ 1152 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1153 return; 1154 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 1155 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1156 ip_srcrt.dst = dst; 1157} 1158 1159/* 1160 * Retrieve incoming source route for use in replies, 1161 * in the same form used by setsockopt. 1162 * The first hop is placed before the options, will be removed later. 1163 */ 1164struct mbuf * 1165ip_srcroute() 1166{ 1167 register struct in_addr *p, *q; 1168 register struct mbuf *m; 1169 1170 if (ip_nhops == 0) 1171 return ((struct mbuf *)0); 1172 m = m_get(M_DONTWAIT, MT_SOOPTS); 1173 if (m == 0) 1174 return ((struct mbuf *)0); 1175 1176#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1177 1178 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1179 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1180 OPTSIZ; 1181#ifdef DIAGNOSTIC 1182 if (ipprintfs) 1183 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1184#endif 1185 1186 /* 1187 * First save first hop for return route 1188 */ 1189 p = &ip_srcrt.route[ip_nhops - 1]; 1190 *(mtod(m, struct in_addr *)) = *p--; 1191#ifdef DIAGNOSTIC 1192 if (ipprintfs) 1193 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1194#endif 1195 1196 /* 1197 * Copy option fields and padding (nop) to mbuf. 1198 */ 1199 ip_srcrt.nop = IPOPT_NOP; 1200 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1201 bcopy((caddr_t)&ip_srcrt.nop, 1202 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 1203 q = (struct in_addr *)(mtod(m, caddr_t) + 1204 sizeof(struct in_addr) + OPTSIZ); 1205#undef OPTSIZ 1206 /* 1207 * Record return path as an IP source route, 1208 * reversing the path (pointers are now aligned). 1209 */ 1210 while (p >= ip_srcrt.route) { 1211#ifdef DIAGNOSTIC 1212 if (ipprintfs) 1213 printf(" %x", ntohl(q->s_addr)); 1214#endif 1215 *q++ = *p--; 1216 } 1217 /* 1218 * Last hop goes to final destination. 1219 */ 1220 *q = ip_srcrt.dst; 1221#ifdef DIAGNOSTIC 1222 if (ipprintfs) 1223 printf(" %x\n", ntohl(q->s_addr)); 1224#endif 1225 return (m); 1226} 1227 1228/* 1229 * Strip out IP options, at higher 1230 * level protocol in the kernel. 1231 * Second argument is buffer to which options 1232 * will be moved, and return value is their length. 1233 * XXX should be deleted; last arg currently ignored. 1234 */ 1235void 1236ip_stripoptions(m, mopt) 1237 register struct mbuf *m; 1238 struct mbuf *mopt; 1239{ 1240 register int i; 1241 struct ip *ip = mtod(m, struct ip *); 1242 register caddr_t opts; 1243 int olen; 1244 1245 olen = (ip->ip_hl << 2) - sizeof (struct ip); 1246 opts = (caddr_t)(ip + 1); 1247 i = m->m_len - (sizeof (struct ip) + olen); 1248 bcopy(opts + olen, opts, (unsigned)i); 1249 m->m_len -= olen; 1250 if (m->m_flags & M_PKTHDR) 1251 m->m_pkthdr.len -= olen; 1252 ip->ip_len -= olen; 1253 ip->ip_hl = sizeof (struct ip) >> 2; 1254} 1255 1256int inetctlerrmap[PRC_NCMDS] = { 1257 0, 0, 0, 0, 1258 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1259 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1260 EMSGSIZE, EHOSTUNREACH, 0, 0, 1261 0, 0, 0, 0, 1262 ENOPROTOOPT 1263}; 1264 1265/* 1266 * Forward a packet. If some error occurs return the sender 1267 * an icmp packet. Note we can't always generate a meaningful 1268 * icmp message because icmp doesn't have a large enough repertoire 1269 * of codes and types. 1270 * 1271 * If not forwarding, just drop the packet. This could be confusing 1272 * if ipforwarding was zero but some routing protocol was advancing 1273 * us as a gateway to somewhere. However, we must let the routing 1274 * protocol deal with that. 1275 * 1276 * The srcrt parameter indicates whether the packet is being forwarded 1277 * via a source route. 1278 */ 1279void 1280ip_forward(m, srcrt) 1281 struct mbuf *m; 1282 int srcrt; 1283{ 1284 register struct ip *ip = mtod(m, struct ip *); 1285 register struct sockaddr_in *sin; 1286 register struct rtentry *rt; 1287 int error, type = 0, code = 0; 1288 struct mbuf *mcopy; 1289 n_long dest; 1290 struct ifnet *destifp; 1291#ifdef IPSEC 1292 struct ifnet dummyifp; 1293#endif 1294 1295 dest = 0; 1296#ifdef DIAGNOSTIC 1297 if (ipprintfs) 1298 printf("forward: src %2.2x dst %2.2x ttl %x\n", 1299 ntohl(ip->ip_src.s_addr), 1300 ntohl(ip->ip_dst.s_addr), ip->ip_ttl); 1301#endif 1302 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1303 ipstat.ips_cantforward++; 1304 m_freem(m); 1305 return; 1306 } 1307 if (ip->ip_ttl <= IPTTLDEC) { 1308 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1309 return; 1310 } 1311 ip->ip_ttl -= IPTTLDEC; 1312 1313 sin = satosin(&ipforward_rt.ro_dst); 1314 if ((rt = ipforward_rt.ro_rt) == 0 || 1315 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1316 if (ipforward_rt.ro_rt) { 1317 RTFREE(ipforward_rt.ro_rt); 1318 ipforward_rt.ro_rt = 0; 1319 } 1320 sin->sin_family = AF_INET; 1321 sin->sin_len = sizeof(struct sockaddr_in); 1322 sin->sin_addr = ip->ip_dst; 1323 1324 rtalloc(&ipforward_rt); 1325 if (ipforward_rt.ro_rt == 0) { 1326 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1327 return; 1328 } 1329 rt = ipforward_rt.ro_rt; 1330 } 1331 1332 /* 1333 * Save at most 68 bytes of the packet in case 1334 * we need to generate an ICMP message to the src. 1335 */ 1336 mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); 1337 1338 /* 1339 * If forwarding packet using same interface that it came in on, 1340 * perhaps should send a redirect to sender to shortcut a hop. 1341 * Only send redirect if source is sending directly to us, 1342 * and if packet was not source routed (or has any options). 1343 * Also, don't send redirect if forwarding using a default route 1344 * or a route modified by a redirect. 1345 */ 1346 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1347 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1348 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1349 ipsendredirects && !srcrt) { 1350 if (rt->rt_ifa && 1351 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1352 ifatoia(rt->rt_ifa)->ia_subnet) { 1353 if (rt->rt_flags & RTF_GATEWAY) 1354 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1355 else 1356 dest = ip->ip_dst.s_addr; 1357 /* 1358 * Router requirements says to only send host 1359 * redirects. 1360 */ 1361 type = ICMP_REDIRECT; 1362 code = ICMP_REDIRECT_HOST; 1363#ifdef DIAGNOSTIC 1364 if (ipprintfs) 1365 printf("redirect (%d) to %x\n", code, 1366 (u_int32_t)dest); 1367#endif 1368 } 1369 } 1370 1371#ifdef IPSEC 1372 m->m_pkthdr.rcvif = NULL; 1373#endif /*IPSEC*/ 1374 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1375 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1376 if (error) 1377 ipstat.ips_cantforward++; 1378 else { 1379 ipstat.ips_forward++; 1380 if (type) 1381 ipstat.ips_redirectsent++; 1382 else { 1383 if (mcopy) { 1384#ifdef GATEWAY 1385 if (mcopy->m_flags & M_CANFASTFWD) 1386 ipflow_create(&ipforward_rt, mcopy); 1387#endif 1388 m_freem(mcopy); 1389 } 1390 return; 1391 } 1392 } 1393 if (mcopy == NULL) 1394 return; 1395 destifp = NULL; 1396 1397 switch (error) { 1398 1399 case 0: /* forwarded, but need redirect */ 1400 /* type, code set above */ 1401 break; 1402 1403 case ENETUNREACH: /* shouldn't happen, checked above */ 1404 case EHOSTUNREACH: 1405 case ENETDOWN: 1406 case EHOSTDOWN: 1407 default: 1408 type = ICMP_UNREACH; 1409 code = ICMP_UNREACH_HOST; 1410 break; 1411 1412 case EMSGSIZE: 1413 type = ICMP_UNREACH; 1414 code = ICMP_UNREACH_NEEDFRAG; 1415#ifndef IPSEC 1416 if (ipforward_rt.ro_rt) 1417 destifp = ipforward_rt.ro_rt->rt_ifp; 1418#else 1419 /* 1420 * If the packet is routed over IPsec tunnel, tell the 1421 * originator the tunnel MTU. 1422 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz 1423 * XXX quickhack!!! 1424 */ 1425 if (ipforward_rt.ro_rt) { 1426 struct secpolicy *sp; 1427 int ipsecerror; 1428 int ipsechdr; 1429 struct route *ro; 1430 1431 sp = ipsec4_getpolicybyaddr(mcopy, 1432 IP_FORWARDING, 1433 &ipsecerror); 1434 1435 if (sp == NULL) 1436 destifp = ipforward_rt.ro_rt->rt_ifp; 1437 else { 1438 /* count IPsec header size */ 1439 ipsechdr = ipsec4_hdrsiz(mcopy, NULL); 1440 1441 /* 1442 * find the correct route for outer IPv4 1443 * header, compute tunnel MTU. 1444 * 1445 * XXX BUG ALERT 1446 * The "dummyifp" code relies upon the fact 1447 * that icmp_error() touches only ifp->if_mtu. 1448 */ 1449 /*XXX*/ 1450 destifp = NULL; 1451 if (sp->req != NULL 1452 && sp->req->sa != NULL) { 1453 ro = &sp->req->sa->saidx->sa_route; 1454 if (ro->ro_rt && ro->ro_rt->rt_ifp) { 1455 dummyifp.if_mtu = 1456 ro->ro_rt->rt_ifp->if_mtu; 1457 dummyifp.if_mtu -= ipsechdr; 1458 destifp = &dummyifp; 1459 } 1460 } 1461 1462 key_freesp(sp); 1463 } 1464 } 1465#endif /*IPSEC*/ 1466 ipstat.ips_cantfrag++; 1467 break; 1468 1469 case ENOBUFS: 1470 type = ICMP_SOURCEQUENCH; 1471 code = 0; 1472 break; 1473 } 1474 icmp_error(mcopy, type, code, dest, destifp); 1475} 1476 1477void 1478ip_savecontrol(inp, mp, ip, m) 1479 register struct inpcb *inp; 1480 register struct mbuf **mp; 1481 register struct ip *ip; 1482 register struct mbuf *m; 1483{ 1484 1485 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1486 struct timeval tv; 1487 1488 microtime(&tv); 1489 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1490 SCM_TIMESTAMP, SOL_SOCKET); 1491 if (*mp) 1492 mp = &(*mp)->m_next; 1493 } 1494 if (inp->inp_flags & INP_RECVDSTADDR) { 1495 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1496 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1497 if (*mp) 1498 mp = &(*mp)->m_next; 1499 } 1500#ifdef notyet 1501 /* 1502 * XXX 1503 * Moving these out of udp_input() made them even more broken 1504 * than they already were. 1505 * - fenner@parc.xerox.com 1506 */ 1507 /* options were tossed already */ 1508 if (inp->inp_flags & INP_RECVOPTS) { 1509 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1510 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1511 if (*mp) 1512 mp = &(*mp)->m_next; 1513 } 1514 /* ip_srcroute doesn't do what we want here, need to fix */ 1515 if (inp->inp_flags & INP_RECVRETOPTS) { 1516 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1517 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1518 if (*mp) 1519 mp = &(*mp)->m_next; 1520 } 1521#endif 1522 if (inp->inp_flags & INP_RECVIF) { 1523 struct sockaddr_dl sdl; 1524 1525 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1526 sdl.sdl_family = AF_LINK; 1527 sdl.sdl_index = m->m_pkthdr.rcvif ? 1528 m->m_pkthdr.rcvif->if_index : 0; 1529 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1530 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1531 IP_RECVIF, IPPROTO_IP); 1532 if (*mp) 1533 mp = &(*mp)->m_next; 1534 } 1535} 1536 1537int 1538ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1539 int *name; 1540 u_int namelen; 1541 void *oldp; 1542 size_t *oldlenp; 1543 void *newp; 1544 size_t newlen; 1545{ 1546 extern int subnetsarelocal, hostzeroisbroadcast; 1547 1548 int error, old; 1549 1550 /* All sysctl names at this level are terminal. */ 1551 if (namelen != 1) 1552 return (ENOTDIR); 1553 1554 switch (name[0]) { 1555 case IPCTL_FORWARDING: 1556 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1557 case IPCTL_SENDREDIRECTS: 1558 return (sysctl_int(oldp, oldlenp, newp, newlen, 1559 &ipsendredirects)); 1560 case IPCTL_DEFTTL: 1561 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1562#ifdef notyet 1563 case IPCTL_DEFMTU: 1564 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1565#endif 1566 case IPCTL_FORWSRCRT: 1567 /* Don't allow this to change in a secure environment. */ 1568 if (securelevel > 0) 1569 return (sysctl_rdint(oldp, oldlenp, newp, 1570 ip_forwsrcrt)); 1571 else 1572 return (sysctl_int(oldp, oldlenp, newp, newlen, 1573 &ip_forwsrcrt)); 1574 case IPCTL_DIRECTEDBCAST: 1575 return (sysctl_int(oldp, oldlenp, newp, newlen, 1576 &ip_directedbcast)); 1577 case IPCTL_ALLOWSRCRT: 1578 return (sysctl_int(oldp, oldlenp, newp, newlen, 1579 &ip_allowsrcrt)); 1580 case IPCTL_SUBNETSARELOCAL: 1581 return (sysctl_int(oldp, oldlenp, newp, newlen, 1582 &subnetsarelocal)); 1583 case IPCTL_MTUDISC: 1584 error = sysctl_int(oldp, oldlenp, newp, newlen, 1585 &ip_mtudisc); 1586 if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) { 1587 ip_mtudisc_timeout_q = 1588 rt_timer_queue_create(ip_mtudisc_timeout); 1589 } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) { 1590 rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE); 1591 ip_mtudisc_timeout_q = NULL; 1592 } 1593 return error; 1594 case IPCTL_ANONPORTMIN: 1595 old = anonportmin; 1596 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1597 if (anonportmin >= anonportmax || anonportmin > 65535 1598#ifndef IPNOPRIVPORTS 1599 || anonportmin < IPPORT_RESERVED 1600#endif 1601 ) { 1602 anonportmin = old; 1603 return (EINVAL); 1604 } 1605 return (error); 1606 case IPCTL_ANONPORTMAX: 1607 old = anonportmax; 1608 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1609 if (anonportmin >= anonportmax || anonportmax > 65535 1610#ifndef IPNOPRIVPORTS 1611 || anonportmax < IPPORT_RESERVED 1612#endif 1613 ) { 1614 anonportmax = old; 1615 return (EINVAL); 1616 } 1617 return (error); 1618 case IPCTL_MTUDISCTIMEOUT: 1619 error = sysctl_int(oldp, oldlenp, newp, newlen, 1620 &ip_mtudisc_timeout); 1621 if (ip_mtudisc_timeout_q != NULL) 1622 rt_timer_queue_change(ip_mtudisc_timeout_q, 1623 ip_mtudisc_timeout); 1624 return (error); 1625#ifdef GATEWAY 1626 case IPCTL_MAXFLOWS: 1627 { 1628 int s; 1629 1630 error = sysctl_int(oldp, oldlenp, newp, newlen, 1631 &ip_maxflows); 1632 s = splsoftnet(); 1633 ipflow_reap(0); 1634 splx(s); 1635 return (error); 1636 } 1637#endif 1638 case IPCTL_HOSTZEROBROADCAST: 1639 return (sysctl_int(oldp, oldlenp, newp, newlen, 1640 &hostzeroisbroadcast)); 1641#if NGIF > 0 1642 case IPCTL_GIF_TTL: 1643 return(sysctl_int(oldp, oldlenp, newp, newlen, 1644 &ip_gif_ttl)); 1645#endif 1646 1647 default: 1648 return (EOPNOTSUPP); 1649 } 1650 /* NOTREACHED */ 1651} 1652