ip_input.c revision 1.312
1/* $NetBSD: ip_input.c,v 1.312 2014/05/23 19:27:48 rmind Exp $ */ 2 3/* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32/*- 33 * Copyright (c) 1998 The NetBSD Foundation, Inc. 34 * All rights reserved. 35 * 36 * This code is derived from software contributed to The NetBSD Foundation 37 * by Public Access Networks Corporation ("Panix"). It was developed under 38 * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon. 39 * 40 * Redistribution and use in source and binary forms, with or without 41 * modification, are permitted provided that the following conditions 42 * are met: 43 * 1. Redistributions of source code must retain the above copyright 44 * notice, this list of conditions and the following disclaimer. 45 * 2. Redistributions in binary form must reproduce the above copyright 46 * notice, this list of conditions and the following disclaimer in the 47 * documentation and/or other materials provided with the distribution. 48 * 49 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 50 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 51 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 52 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 53 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 54 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 55 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 56 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 57 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 58 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 59 * POSSIBILITY OF SUCH DAMAGE. 60 */ 61 62/* 63 * Copyright (c) 1982, 1986, 1988, 1993 64 * The Regents of the University of California. All rights reserved. 65 * 66 * Redistribution and use in source and binary forms, with or without 67 * modification, are permitted provided that the following conditions 68 * are met: 69 * 1. Redistributions of source code must retain the above copyright 70 * notice, this list of conditions and the following disclaimer. 71 * 2. Redistributions in binary form must reproduce the above copyright 72 * notice, this list of conditions and the following disclaimer in the 73 * documentation and/or other materials provided with the distribution. 74 * 3. Neither the name of the University nor the names of its contributors 75 * may be used to endorse or promote products derived from this software 76 * without specific prior written permission. 77 * 78 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 79 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 80 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 81 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 82 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 83 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 84 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 85 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 86 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 87 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 88 * SUCH DAMAGE. 89 * 90 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 91 */ 92 93#include <sys/cdefs.h> 94__KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.312 2014/05/23 19:27:48 rmind Exp $"); 95 96#include "opt_inet.h" 97#include "opt_compat_netbsd.h" 98#include "opt_gateway.h" 99#include "opt_ipsec.h" 100#include "opt_mrouting.h" 101#include "opt_mbuftrace.h" 102#include "opt_inet_csum.h" 103 104#include <sys/param.h> 105#include <sys/systm.h> 106#include <sys/mbuf.h> 107#include <sys/domain.h> 108#include <sys/protosw.h> 109#include <sys/socket.h> 110#include <sys/socketvar.h> 111#include <sys/errno.h> 112#include <sys/time.h> 113#include <sys/kernel.h> 114#include <sys/pool.h> 115#include <sys/sysctl.h> 116#include <sys/kauth.h> 117 118#include <net/if.h> 119#include <net/if_dl.h> 120#include <net/route.h> 121#include <net/pfil.h> 122 123#include <netinet/in.h> 124#include <netinet/in_systm.h> 125#include <netinet/ip.h> 126#include <netinet/in_pcb.h> 127#include <netinet/in_proto.h> 128#include <netinet/in_var.h> 129#include <netinet/ip_var.h> 130#include <netinet/ip_private.h> 131#include <netinet/ip_icmp.h> 132/* just for gif_ttl */ 133#include <netinet/in_gif.h> 134#include "gif.h" 135#include <net/if_gre.h> 136#include "gre.h" 137 138#ifdef MROUTING 139#include <netinet/ip_mroute.h> 140#endif 141#include <netinet/portalgo.h> 142 143#ifdef IPSEC 144#include <netipsec/ipsec.h> 145#endif 146 147#ifndef IPFORWARDING 148#ifdef GATEWAY 149#define IPFORWARDING 1 /* forward IP packets not for us */ 150#else /* GATEWAY */ 151#define IPFORWARDING 0 /* don't forward IP packets not for us */ 152#endif /* GATEWAY */ 153#endif /* IPFORWARDING */ 154#ifndef IPSENDREDIRECTS 155#define IPSENDREDIRECTS 1 156#endif 157#ifndef IPFORWSRCRT 158#define IPFORWSRCRT 1 /* forward source-routed packets */ 159#endif 160#ifndef IPALLOWSRCRT 161#define IPALLOWSRCRT 1 /* allow source-routed packets */ 162#endif 163#ifndef IPMTUDISC 164#define IPMTUDISC 1 165#endif 166#ifndef IPMTUDISCTIMEOUT 167#define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 168#endif 169 170#ifdef COMPAT_50 171#include <compat/sys/time.h> 172#include <compat/sys/socket.h> 173#endif 174 175/* 176 * Note: DIRECTED_BROADCAST is handled this way so that previous 177 * configuration using this option will Just Work. 178 */ 179#ifndef IPDIRECTEDBCAST 180#ifdef DIRECTED_BROADCAST 181#define IPDIRECTEDBCAST 1 182#else 183#define IPDIRECTEDBCAST 0 184#endif /* DIRECTED_BROADCAST */ 185#endif /* IPDIRECTEDBCAST */ 186int ipforwarding = IPFORWARDING; 187int ipsendredirects = IPSENDREDIRECTS; 188int ip_defttl = IPDEFTTL; 189int ip_forwsrcrt = IPFORWSRCRT; 190int ip_directedbcast = IPDIRECTEDBCAST; 191int ip_allowsrcrt = IPALLOWSRCRT; 192int ip_mtudisc = IPMTUDISC; 193int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 194#ifdef DIAGNOSTIC 195int ipprintfs = 0; 196#endif 197 198int ip_do_randomid = 0; 199 200/* 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 215struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 216 217struct ifqueue ipintrq; 218 219ipid_state_t * ip_ids; 220uint16_t ip_id; 221 222percpu_t *ipstat_percpu; 223 224pfil_head_t *inet_pfil_hook; 225 226#ifdef INET_CSUM_COUNTERS 227#include <sys/device.h> 228 229struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 230 NULL, "inet", "hwcsum bad"); 231struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 232 NULL, "inet", "hwcsum ok"); 233struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 234 NULL, "inet", "swcsum"); 235 236#define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 237 238EVCNT_ATTACH_STATIC(ip_hwcsum_bad); 239EVCNT_ATTACH_STATIC(ip_hwcsum_ok); 240EVCNT_ATTACH_STATIC(ip_swcsum); 241 242#else 243 244#define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 245 246#endif /* INET_CSUM_COUNTERS */ 247 248/* 249 * We need to save the IP options in case a protocol wants to respond 250 * to an incoming packet over the same route if the packet got here 251 * using IP source routing. This allows connection establishment and 252 * maintenance when the remote end is on a network that is not known 253 * to us. 254 */ 255 256static int ip_nhops = 0; 257 258static struct ip_srcrt { 259 struct in_addr dst; /* final destination */ 260 char nop; /* one NOP to align */ 261 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 262 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 263} ip_srcrt; 264 265static int ip_drainwanted; 266 267static void save_rte(u_char *, struct in_addr); 268 269#ifdef MBUFTRACE 270struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 271struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 272#endif 273 274static void ip_input(struct mbuf *); 275static bool ip_dooptions(struct mbuf *); 276static struct in_ifaddr *ip_rtaddr(struct in_addr); 277static void sysctl_net_inet_ip_setup(struct sysctllog **); 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(void) 285{ 286 const struct protosw *pr; 287 288 in_init(); 289 sysctl_net_inet_ip_setup(NULL); 290 291 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 292 KASSERT(pr != NULL); 293 294 for (u_int i = 0; i < IPPROTO_MAX; i++) { 295 ip_protox[i] = pr - inetsw; 296 } 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; 302 303 ip_reass_init(); 304 305 ip_ids = ip_id_init(); 306 ip_id = time_second & 0xfffff; 307 308 ipintrq.ifq_maxlen = IFQ_MAXLEN; 309 310 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 311#ifdef GATEWAY 312 ipflow_init(); 313#endif 314 315 /* Register our Packet Filter hook. */ 316 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 317 KASSERT(inet_pfil_hook != NULL); 318 319#ifdef MBUFTRACE 320 MOWNER_ATTACH(&ip_tx_mowner); 321 MOWNER_ATTACH(&ip_rx_mowner); 322#endif /* MBUFTRACE */ 323 324 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 325} 326 327struct sockaddr_in ipaddr = { 328 .sin_len = sizeof(ipaddr), 329 .sin_family = AF_INET, 330}; 331 332static struct route ipforward_rt; 333 334/* 335 * IP software interrupt routine 336 */ 337void 338ipintr(void) 339{ 340 int s; 341 struct mbuf *m; 342 struct ifqueue lcl_intrq; 343 344 memset(&lcl_intrq, 0, sizeof(lcl_intrq)); 345 346 mutex_enter(softnet_lock); 347 KERNEL_LOCK(1, NULL); 348 if (!IF_IS_EMPTY(&ipintrq)) { 349 s = splnet(); 350 351 /* Take existing queue onto stack */ 352 lcl_intrq = ipintrq; 353 354 /* Zero out global queue, preserving maxlen and drops */ 355 ipintrq.ifq_head = NULL; 356 ipintrq.ifq_tail = NULL; 357 ipintrq.ifq_len = 0; 358 ipintrq.ifq_maxlen = lcl_intrq.ifq_maxlen; 359 ipintrq.ifq_drops = lcl_intrq.ifq_drops; 360 361 splx(s); 362 } 363 KERNEL_UNLOCK_ONE(NULL); 364 while (!IF_IS_EMPTY(&lcl_intrq)) { 365 IF_DEQUEUE(&lcl_intrq, m); 366 if (m == NULL) 367 break; 368 ip_input(m); 369 } 370 mutex_exit(softnet_lock); 371} 372 373/* 374 * Ip input routine. Checksum and byte swap header. If fragmented 375 * try to reassemble. Process options. Pass to next level. 376 */ 377static void 378ip_input(struct mbuf *m) 379{ 380 struct ip *ip = NULL; 381 struct in_ifaddr *ia; 382 struct ifaddr *ifa; 383 int hlen = 0, len; 384 int downmatch; 385 int checkif; 386 int srcrt = 0; 387 388 MCLAIM(m, &ip_rx_mowner); 389 KASSERT((m->m_flags & M_PKTHDR) != 0); 390 391 /* 392 * If no IP addresses have been set yet but the interfaces 393 * are receiving, can't do anything with incoming packets yet. 394 */ 395 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 396 goto bad; 397 IP_STATINC(IP_STAT_TOTAL); 398 /* 399 * If the IP header is not aligned, slurp it up into a new 400 * mbuf with space for link headers, in the event we forward 401 * it. Otherwise, if it is aligned, make sure the entire 402 * base IP header is in the first mbuf of the chain. 403 */ 404 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 405 if ((m = m_copyup(m, sizeof(struct ip), 406 (max_linkhdr + 3) & ~3)) == NULL) { 407 /* XXXJRT new stat, please */ 408 IP_STATINC(IP_STAT_TOOSMALL); 409 return; 410 } 411 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 412 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 413 IP_STATINC(IP_STAT_TOOSMALL); 414 return; 415 } 416 } 417 ip = mtod(m, struct ip *); 418 if (ip->ip_v != IPVERSION) { 419 IP_STATINC(IP_STAT_BADVERS); 420 goto bad; 421 } 422 hlen = ip->ip_hl << 2; 423 if (hlen < sizeof(struct ip)) { /* minimum header length */ 424 IP_STATINC(IP_STAT_BADHLEN); 425 goto bad; 426 } 427 if (hlen > m->m_len) { 428 if ((m = m_pullup(m, hlen)) == NULL) { 429 IP_STATINC(IP_STAT_BADHLEN); 430 return; 431 } 432 ip = mtod(m, struct ip *); 433 } 434 435 /* 436 * RFC1122: packets with a multicast source address are 437 * not allowed. 438 */ 439 if (IN_MULTICAST(ip->ip_src.s_addr)) { 440 IP_STATINC(IP_STAT_BADADDR); 441 goto bad; 442 } 443 444 /* 127/8 must not appear on wire - RFC1122 */ 445 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 446 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 447 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 448 IP_STATINC(IP_STAT_BADADDR); 449 goto bad; 450 } 451 } 452 453 switch (m->m_pkthdr.csum_flags & 454 ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) | 455 M_CSUM_IPv4_BAD)) { 456 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 457 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 458 goto badcsum; 459 460 case M_CSUM_IPv4: 461 /* Checksum was okay. */ 462 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 463 break; 464 465 default: 466 /* 467 * Must compute it ourselves. Maybe skip checksum on 468 * loopback interfaces. 469 */ 470 if (__predict_true(!(m->m_pkthdr.rcvif->if_flags & 471 IFF_LOOPBACK) || ip_do_loopback_cksum)) { 472 INET_CSUM_COUNTER_INCR(&ip_swcsum); 473 if (in_cksum(m, hlen) != 0) 474 goto badcsum; 475 } 476 break; 477 } 478 479 /* Retrieve the packet length. */ 480 len = ntohs(ip->ip_len); 481 482 /* 483 * Check for additional length bogosity 484 */ 485 if (len < hlen) { 486 IP_STATINC(IP_STAT_BADLEN); 487 goto bad; 488 } 489 490 /* 491 * Check that the amount of data in the buffers 492 * is as at least much as the IP header would have us expect. 493 * Trim mbufs if longer than we expect. 494 * Drop packet if shorter than we expect. 495 */ 496 if (m->m_pkthdr.len < len) { 497 IP_STATINC(IP_STAT_TOOSHORT); 498 goto bad; 499 } 500 if (m->m_pkthdr.len > len) { 501 if (m->m_len == m->m_pkthdr.len) { 502 m->m_len = len; 503 m->m_pkthdr.len = len; 504 } else 505 m_adj(m, len - m->m_pkthdr.len); 506 } 507 508 /* 509 * Assume that we can create a fast-forward IP flow entry 510 * based on this packet. 511 */ 512 m->m_flags |= M_CANFASTFWD; 513 514 /* 515 * Run through list of hooks for input packets. If there are any 516 * filters which require that additional packets in the flow are 517 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 518 * Note that filters must _never_ set this flag, as another filter 519 * in the list may have previously cleared it. 520 */ 521 /* 522 * let ipfilter look at packet on the wire, 523 * not the decapsulated packet. 524 */ 525#if defined(IPSEC) 526 if (!ipsec_indone(m)) 527#else 528 if (1) 529#endif 530 { 531 struct in_addr odst; 532 533 odst = ip->ip_dst; 534 if (pfil_run_hooks(inet_pfil_hook, &m, m->m_pkthdr.rcvif, 535 PFIL_IN) != 0) 536 return; 537 if (m == NULL) 538 return; 539 ip = mtod(m, struct ip *); 540 hlen = ip->ip_hl << 2; 541 /* 542 * XXX The setting of "srcrt" here is to prevent ip_forward() 543 * from generating ICMP redirects for packets that have 544 * been redirected by a hook back out on to the same LAN that 545 * they came from and is not an indication that the packet 546 * is being inffluenced by source routing options. This 547 * allows things like 548 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 549 * where tlp0 is both on the 1.1.1.0/24 network and is the 550 * default route for hosts on 1.1.1.0/24. Of course this 551 * also requires a "map tlp0 ..." to complete the story. 552 * One might argue whether or not this kind of network config. 553 * should be supported in this manner... 554 */ 555 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 556 } 557 558#ifdef ALTQ 559 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 560 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) { 561 /* packet dropped by traffic conditioner */ 562 return; 563 } 564#endif 565 566 /* 567 * Process options and, if not destined for us, 568 * ship it on. ip_dooptions returns 1 when an 569 * error was detected (causing an icmp message 570 * to be sent and the original packet to be freed). 571 */ 572 ip_nhops = 0; /* for source routed packets */ 573 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 574 return; 575 576 /* 577 * Enable a consistency check between the destination address 578 * and the arrival interface for a unicast packet (the RFC 1122 579 * strong ES model) if IP forwarding is disabled and the packet 580 * is not locally generated. 581 * 582 * XXX - Checking also should be disabled if the destination 583 * address is ipnat'ed to a different interface. 584 * 585 * XXX - Checking is incompatible with IP aliases added 586 * to the loopback interface instead of the interface where 587 * the packets are received. 588 * 589 * XXX - We need to add a per ifaddr flag for this so that 590 * we get finer grain control. 591 */ 592 checkif = ip_checkinterface && (ipforwarding == 0) && 593 (m->m_pkthdr.rcvif != NULL) && 594 ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0); 595 596 /* 597 * Check our list of addresses, to see if the packet is for us. 598 * 599 * Traditional 4.4BSD did not consult IFF_UP at all. 600 * The behavior here is to treat addresses on !IFF_UP interface 601 * as not mine. 602 */ 603 downmatch = 0; 604 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 605 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 606 if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif) 607 continue; 608 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 609 break; 610 else 611 downmatch++; 612 } 613 } 614 if (ia != NULL) 615 goto ours; 616 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 617 IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) { 618 if (ifa->ifa_addr->sa_family != AF_INET) 619 continue; 620 ia = ifatoia(ifa); 621 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 622 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 623 /* 624 * Look for all-0's host part (old broadcast addr), 625 * either for subnet or net. 626 */ 627 ip->ip_dst.s_addr == ia->ia_subnet || 628 ip->ip_dst.s_addr == ia->ia_net) 629 goto ours; 630 /* 631 * An interface with IP address zero accepts 632 * all packets that arrive on that interface. 633 */ 634 if (in_nullhost(ia->ia_addr.sin_addr)) 635 goto ours; 636 } 637 } 638 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 639 struct in_multi *inm; 640#ifdef MROUTING 641 extern struct socket *ip_mrouter; 642 643 if (ip_mrouter) { 644 /* 645 * If we are acting as a multicast router, all 646 * incoming multicast packets are passed to the 647 * kernel-level multicast forwarding function. 648 * The packet is returned (relatively) intact; if 649 * ip_mforward() returns a non-zero value, the packet 650 * must be discarded, else it may be accepted below. 651 * 652 * (The IP ident field is put in the same byte order 653 * as expected when ip_mforward() is called from 654 * ip_output().) 655 */ 656 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 657 IP_STATINC(IP_STAT_CANTFORWARD); 658 m_freem(m); 659 return; 660 } 661 662 /* 663 * The process-level routing demon needs to receive 664 * all multicast IGMP packets, whether or not this 665 * host belongs to their destination groups. 666 */ 667 if (ip->ip_p == IPPROTO_IGMP) 668 goto ours; 669 IP_STATINC(IP_STAT_CANTFORWARD); 670 } 671#endif 672 /* 673 * See if we belong to the destination multicast group on the 674 * arrival interface. 675 */ 676 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 677 if (inm == NULL) { 678 IP_STATINC(IP_STAT_CANTFORWARD); 679 m_freem(m); 680 return; 681 } 682 goto ours; 683 } 684 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 685 in_nullhost(ip->ip_dst)) 686 goto ours; 687 688 /* 689 * Not for us; forward if possible and desirable. 690 */ 691 if (ipforwarding == 0) { 692 IP_STATINC(IP_STAT_CANTFORWARD); 693 m_freem(m); 694 } else { 695 /* 696 * If ip_dst matched any of my address on !IFF_UP interface, 697 * and there's no IFF_UP interface that matches ip_dst, 698 * send icmp unreach. Forwarding it will result in in-kernel 699 * forwarding loop till TTL goes to 0. 700 */ 701 if (downmatch) { 702 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 703 IP_STATINC(IP_STAT_CANTFORWARD); 704 return; 705 } 706#ifdef IPSEC 707 /* Perform IPsec, if any. */ 708 if (ipsec4_input(m, IP_FORWARDING | (ip_directedbcast ? 709 IP_ALLOWBROADCAST : 0)) != 0) { 710 goto bad; 711 } 712#endif 713 ip_forward(m, srcrt); 714 } 715 return; 716 717ours: 718 /* 719 * If offset or IP_MF are set, must reassemble. 720 */ 721 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 722 /* 723 * Pass to IP reassembly mechanism. 724 */ 725 if (ip_reass_packet(&m, ip) != 0) { 726 /* Failed; invalid fragment(s) or packet. */ 727 goto bad; 728 } 729 if (m == NULL) { 730 /* More fragments should come; silently return. */ 731 return; 732 } 733 /* 734 * Reassembly is done, we have the final packet. 735 * Updated cached data in local variable(s). 736 */ 737 ip = mtod(m, struct ip *); 738 hlen = ip->ip_hl << 2; 739 } 740 741#ifdef IPSEC 742 /* 743 * Enforce IPsec policy checking if we are seeing last header. 744 * Note that we do not visit this with protocols with PCB layer 745 * code - like UDP/TCP/raw IP. 746 */ 747 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 748 if (ipsec4_input(m, 0) != 0) { 749 goto bad; 750 } 751 } 752#endif 753 754 /* 755 * Switch out to protocol's input routine. 756 */ 757#if IFA_STATS 758 if (ia && ip) 759 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 760#endif 761 IP_STATINC(IP_STAT_DELIVERED); 762 { 763 int off = hlen, nh = ip->ip_p; 764 765 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 766 return; 767 } 768bad: 769 m_freem(m); 770 return; 771 772badcsum: 773 IP_STATINC(IP_STAT_BADSUM); 774 m_freem(m); 775} 776 777/* 778 * IP timer processing. 779 */ 780void 781ip_slowtimo(void) 782{ 783 784 mutex_enter(softnet_lock); 785 KERNEL_LOCK(1, NULL); 786 787 ip_reass_slowtimo(); 788 789 KERNEL_UNLOCK_ONE(NULL); 790 mutex_exit(softnet_lock); 791} 792 793/* 794 * IP drain processing. 795 */ 796void 797ip_drain(void) 798{ 799 800 KERNEL_LOCK(1, NULL); 801 ip_reass_drain(); 802 KERNEL_UNLOCK_ONE(NULL); 803} 804 805/* 806 * ip_dooptions: perform option processing on a datagram, possibly discarding 807 * it if bad options are encountered, or forwarding it if source-routed. 808 * 809 * => Returns true if packet has been forwarded/freed. 810 * => Returns false if the packet should be processed further. 811 */ 812static bool 813ip_dooptions(struct mbuf *m) 814{ 815 struct ip *ip = mtod(m, struct ip *); 816 u_char *cp, *cp0; 817 struct ip_timestamp *ipt; 818 struct in_ifaddr *ia; 819 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 820 struct in_addr dst; 821 n_time ntime; 822 823 dst = ip->ip_dst; 824 cp = (u_char *)(ip + 1); 825 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 826 for (; cnt > 0; cnt -= optlen, cp += optlen) { 827 opt = cp[IPOPT_OPTVAL]; 828 if (opt == IPOPT_EOL) 829 break; 830 if (opt == IPOPT_NOP) 831 optlen = 1; 832 else { 833 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 834 code = &cp[IPOPT_OLEN] - (u_char *)ip; 835 goto bad; 836 } 837 optlen = cp[IPOPT_OLEN]; 838 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 839 code = &cp[IPOPT_OLEN] - (u_char *)ip; 840 goto bad; 841 } 842 } 843 switch (opt) { 844 845 default: 846 break; 847 848 /* 849 * Source routing with record. 850 * Find interface with current destination address. 851 * If none on this machine then drop if strictly routed, 852 * or do nothing if loosely routed. 853 * Record interface address and bring up next address 854 * component. If strictly routed make sure next 855 * address is on directly accessible net. 856 */ 857 case IPOPT_LSRR: 858 case IPOPT_SSRR: 859 if (ip_allowsrcrt == 0) { 860 type = ICMP_UNREACH; 861 code = ICMP_UNREACH_NET_PROHIB; 862 goto bad; 863 } 864 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 865 code = &cp[IPOPT_OLEN] - (u_char *)ip; 866 goto bad; 867 } 868 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 869 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 870 goto bad; 871 } 872 ipaddr.sin_addr = ip->ip_dst; 873 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 874 if (ia == 0) { 875 if (opt == IPOPT_SSRR) { 876 type = ICMP_UNREACH; 877 code = ICMP_UNREACH_SRCFAIL; 878 goto bad; 879 } 880 /* 881 * Loose routing, and not at next destination 882 * yet; nothing to do except forward. 883 */ 884 break; 885 } 886 off--; /* 0 origin */ 887 if ((off + sizeof(struct in_addr)) > optlen) { 888 /* 889 * End of source route. Should be for us. 890 */ 891 save_rte(cp, ip->ip_src); 892 break; 893 } 894 /* 895 * locate outgoing interface 896 */ 897 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 898 sizeof(ipaddr.sin_addr)); 899 if (opt == IPOPT_SSRR) 900 ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); 901 else 902 ia = ip_rtaddr(ipaddr.sin_addr); 903 if (ia == 0) { 904 type = ICMP_UNREACH; 905 code = ICMP_UNREACH_SRCFAIL; 906 goto bad; 907 } 908 ip->ip_dst = ipaddr.sin_addr; 909 bcopy((void *)&ia->ia_addr.sin_addr, 910 (void *)(cp + off), sizeof(struct in_addr)); 911 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 912 /* 913 * Let ip_intr's mcast routing check handle mcast pkts 914 */ 915 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 916 break; 917 918 case IPOPT_RR: 919 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 920 code = &cp[IPOPT_OLEN] - (u_char *)ip; 921 goto bad; 922 } 923 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 924 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 925 goto bad; 926 } 927 /* 928 * If no space remains, ignore. 929 */ 930 off--; /* 0 origin */ 931 if ((off + sizeof(struct in_addr)) > optlen) 932 break; 933 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 934 sizeof(ipaddr.sin_addr)); 935 /* 936 * locate outgoing interface; if we're the destination, 937 * use the incoming interface (should be same). 938 */ 939 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 940 == NULL && 941 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 942 type = ICMP_UNREACH; 943 code = ICMP_UNREACH_HOST; 944 goto bad; 945 } 946 bcopy((void *)&ia->ia_addr.sin_addr, 947 (void *)(cp + off), sizeof(struct in_addr)); 948 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 949 break; 950 951 case IPOPT_TS: 952 code = cp - (u_char *)ip; 953 ipt = (struct ip_timestamp *)cp; 954 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 955 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 956 goto bad; 957 } 958 if (ipt->ipt_ptr < 5) { 959 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 960 goto bad; 961 } 962 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 963 if (++ipt->ipt_oflw == 0) { 964 code = (u_char *)&ipt->ipt_ptr - 965 (u_char *)ip; 966 goto bad; 967 } 968 break; 969 } 970 cp0 = (cp + ipt->ipt_ptr - 1); 971 switch (ipt->ipt_flg) { 972 973 case IPOPT_TS_TSONLY: 974 break; 975 976 case IPOPT_TS_TSANDADDR: 977 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 978 sizeof(struct in_addr) > ipt->ipt_len) { 979 code = (u_char *)&ipt->ipt_ptr - 980 (u_char *)ip; 981 goto bad; 982 } 983 ipaddr.sin_addr = dst; 984 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 985 m->m_pkthdr.rcvif)); 986 if (ia == 0) 987 continue; 988 bcopy(&ia->ia_addr.sin_addr, 989 cp0, sizeof(struct in_addr)); 990 ipt->ipt_ptr += sizeof(struct in_addr); 991 break; 992 993 case IPOPT_TS_PRESPEC: 994 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 995 sizeof(struct in_addr) > ipt->ipt_len) { 996 code = (u_char *)&ipt->ipt_ptr - 997 (u_char *)ip; 998 goto bad; 999 } 1000 memcpy(&ipaddr.sin_addr, cp0, 1001 sizeof(struct in_addr)); 1002 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1003 == NULL) 1004 continue; 1005 ipt->ipt_ptr += sizeof(struct in_addr); 1006 break; 1007 1008 default: 1009 /* XXX can't take &ipt->ipt_flg */ 1010 code = (u_char *)&ipt->ipt_ptr - 1011 (u_char *)ip + 1; 1012 goto bad; 1013 } 1014 ntime = iptime(); 1015 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1016 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1017 sizeof(n_time)); 1018 ipt->ipt_ptr += sizeof(n_time); 1019 } 1020 } 1021 if (forward) { 1022 if (ip_forwsrcrt == 0) { 1023 type = ICMP_UNREACH; 1024 code = ICMP_UNREACH_SRCFAIL; 1025 goto bad; 1026 } 1027 ip_forward(m, 1); 1028 return true; 1029 } 1030 return false; 1031bad: 1032 icmp_error(m, type, code, 0, 0); 1033 IP_STATINC(IP_STAT_BADOPTIONS); 1034 return true; 1035} 1036 1037/* 1038 * ip_rtaddr: given address of next destination (final or next hop), 1039 * return internet address info of interface to be used to get there. 1040 */ 1041static struct in_ifaddr * 1042ip_rtaddr(struct in_addr dst) 1043{ 1044 struct rtentry *rt; 1045 union { 1046 struct sockaddr dst; 1047 struct sockaddr_in dst4; 1048 } u; 1049 1050 sockaddr_in_init(&u.dst4, &dst, 0); 1051 1052 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) 1053 return NULL; 1054 1055 return ifatoia(rt->rt_ifa); 1056} 1057 1058/* 1059 * save_rte: save incoming source route for use in replies, to be picked 1060 * up later by ip_srcroute if the receiver is interested. 1061 */ 1062static void 1063save_rte(u_char *option, struct in_addr dst) 1064{ 1065 unsigned olen; 1066 1067 olen = option[IPOPT_OLEN]; 1068 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1069 return; 1070 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1071 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1072 ip_srcrt.dst = dst; 1073} 1074 1075/* 1076 * Retrieve incoming source route for use in replies, 1077 * in the same form used by setsockopt. 1078 * The first hop is placed before the options, will be removed later. 1079 */ 1080struct mbuf * 1081ip_srcroute(void) 1082{ 1083 struct in_addr *p, *q; 1084 struct mbuf *m; 1085 1086 if (ip_nhops == 0) 1087 return NULL; 1088 m = m_get(M_DONTWAIT, MT_SOOPTS); 1089 if (m == 0) 1090 return NULL; 1091 1092 MCLAIM(m, &inetdomain.dom_mowner); 1093#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1094 1095 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1096 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1097 OPTSIZ; 1098 1099 /* 1100 * First save first hop for return route 1101 */ 1102 p = &ip_srcrt.route[ip_nhops - 1]; 1103 *(mtod(m, struct in_addr *)) = *p--; 1104 1105 /* 1106 * Copy option fields and padding (nop) to mbuf. 1107 */ 1108 ip_srcrt.nop = IPOPT_NOP; 1109 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1110 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1111 OPTSIZ); 1112 q = (struct in_addr *)(mtod(m, char *) + 1113 sizeof(struct in_addr) + OPTSIZ); 1114#undef OPTSIZ 1115 /* 1116 * Record return path as an IP source route, 1117 * reversing the path (pointers are now aligned). 1118 */ 1119 while (p >= ip_srcrt.route) { 1120 *q++ = *p--; 1121 } 1122 /* 1123 * Last hop goes to final destination. 1124 */ 1125 *q = ip_srcrt.dst; 1126 return (m); 1127} 1128 1129const int inetctlerrmap[PRC_NCMDS] = { 1130 [PRC_MSGSIZE] = EMSGSIZE, 1131 [PRC_HOSTDEAD] = EHOSTDOWN, 1132 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1133 [PRC_UNREACH_NET] = EHOSTUNREACH, 1134 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1135 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1136 [PRC_UNREACH_PORT] = ECONNREFUSED, 1137 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1138 [PRC_PARAMPROB] = ENOPROTOOPT, 1139}; 1140 1141void 1142ip_fasttimo(void) 1143{ 1144 if (ip_drainwanted) { 1145 ip_drain(); 1146 ip_drainwanted = 0; 1147 } 1148} 1149 1150void 1151ip_drainstub(void) 1152{ 1153 ip_drainwanted = 1; 1154} 1155 1156/* 1157 * Forward a packet. If some error occurs return the sender 1158 * an icmp packet. Note we can't always generate a meaningful 1159 * icmp message because icmp doesn't have a large enough repertoire 1160 * of codes and types. 1161 * 1162 * If not forwarding, just drop the packet. This could be confusing 1163 * if ipforwarding was zero but some routing protocol was advancing 1164 * us as a gateway to somewhere. However, we must let the routing 1165 * protocol deal with that. 1166 * 1167 * The srcrt parameter indicates whether the packet is being forwarded 1168 * via a source route. 1169 */ 1170void 1171ip_forward(struct mbuf *m, int srcrt) 1172{ 1173 struct ip *ip = mtod(m, struct ip *); 1174 struct rtentry *rt; 1175 int error, type = 0, code = 0, destmtu = 0; 1176 struct mbuf *mcopy; 1177 n_long dest; 1178 union { 1179 struct sockaddr dst; 1180 struct sockaddr_in dst4; 1181 } u; 1182 1183 /* 1184 * We are now in the output path. 1185 */ 1186 MCLAIM(m, &ip_tx_mowner); 1187 1188 /* 1189 * Clear any in-bound checksum flags for this packet. 1190 */ 1191 m->m_pkthdr.csum_flags = 0; 1192 1193 dest = 0; 1194 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1195 IP_STATINC(IP_STAT_CANTFORWARD); 1196 m_freem(m); 1197 return; 1198 } 1199 if (ip->ip_ttl <= IPTTLDEC) { 1200 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1201 return; 1202 } 1203 1204 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1205 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1206 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1207 return; 1208 } 1209 1210 /* 1211 * Save at most 68 bytes of the packet in case 1212 * we need to generate an ICMP message to the src. 1213 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1214 */ 1215 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1216 if (mcopy) 1217 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1218 1219 ip->ip_ttl -= IPTTLDEC; 1220 1221 /* 1222 * If forwarding packet using same interface that it came in on, 1223 * perhaps should send a redirect to sender to shortcut a hop. 1224 * Only send redirect if source is sending directly to us, 1225 * and if packet was not source routed (or has any options). 1226 * Also, don't send redirect if forwarding using a default route 1227 * or a route modified by a redirect. 1228 */ 1229 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1230 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1231 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1232 ipsendredirects && !srcrt) { 1233 if (rt->rt_ifa && 1234 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1235 ifatoia(rt->rt_ifa)->ia_subnet) { 1236 if (rt->rt_flags & RTF_GATEWAY) 1237 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1238 else 1239 dest = ip->ip_dst.s_addr; 1240 /* 1241 * Router requirements says to only send host 1242 * redirects. 1243 */ 1244 type = ICMP_REDIRECT; 1245 code = ICMP_REDIRECT_HOST; 1246 } 1247 } 1248 1249 error = ip_output(m, NULL, &ipforward_rt, 1250 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1251 NULL, NULL); 1252 1253 if (error) 1254 IP_STATINC(IP_STAT_CANTFORWARD); 1255 else { 1256 uint64_t *ips = IP_STAT_GETREF(); 1257 ips[IP_STAT_FORWARD]++; 1258 if (type) { 1259 ips[IP_STAT_REDIRECTSENT]++; 1260 IP_STAT_PUTREF(); 1261 } else { 1262 IP_STAT_PUTREF(); 1263 if (mcopy) { 1264#ifdef GATEWAY 1265 if (mcopy->m_flags & M_CANFASTFWD) 1266 ipflow_create(&ipforward_rt, mcopy); 1267#endif 1268 m_freem(mcopy); 1269 } 1270 return; 1271 } 1272 } 1273 if (mcopy == NULL) 1274 return; 1275 1276 switch (error) { 1277 1278 case 0: /* forwarded, but need redirect */ 1279 /* type, code set above */ 1280 break; 1281 1282 case ENETUNREACH: /* shouldn't happen, checked above */ 1283 case EHOSTUNREACH: 1284 case ENETDOWN: 1285 case EHOSTDOWN: 1286 default: 1287 type = ICMP_UNREACH; 1288 code = ICMP_UNREACH_HOST; 1289 break; 1290 1291 case EMSGSIZE: 1292 type = ICMP_UNREACH; 1293 code = ICMP_UNREACH_NEEDFRAG; 1294 1295 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1296 destmtu = rt->rt_ifp->if_mtu; 1297#ifdef IPSEC 1298 (void)ipsec4_forward(mcopy, &destmtu); 1299#endif 1300 IP_STATINC(IP_STAT_CANTFRAG); 1301 break; 1302 1303 case ENOBUFS: 1304 /* 1305 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1306 * Requirements for IP Version 4 Routers. Source quench can 1307 * big problem under DoS attacks or if the underlying 1308 * interface is rate-limited. 1309 */ 1310 if (mcopy) 1311 m_freem(mcopy); 1312 return; 1313 } 1314 icmp_error(mcopy, type, code, dest, destmtu); 1315} 1316 1317void 1318ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1319 struct mbuf *m) 1320{ 1321 struct socket *so = inp->inp_socket; 1322 int inpflags = inp->inp_flags; 1323 1324 if (so->so_options & SO_TIMESTAMP 1325#ifdef SO_OTIMESTAMP 1326 || so->so_options & SO_OTIMESTAMP 1327#endif 1328 ) { 1329 struct timeval tv; 1330 1331 microtime(&tv); 1332#ifdef SO_OTIMESTAMP 1333 if (so->so_options & SO_OTIMESTAMP) { 1334 struct timeval50 tv50; 1335 timeval_to_timeval50(&tv, &tv50); 1336 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1337 SCM_OTIMESTAMP, SOL_SOCKET); 1338 } else 1339#endif 1340 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1341 SCM_TIMESTAMP, SOL_SOCKET); 1342 if (*mp) 1343 mp = &(*mp)->m_next; 1344 } 1345 if (inpflags & INP_RECVDSTADDR) { 1346 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1347 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1348 if (*mp) 1349 mp = &(*mp)->m_next; 1350 } 1351 if (inpflags & INP_RECVPKTINFO) { 1352 struct in_pktinfo ipi; 1353 ipi.ipi_addr = ip->ip_src; 1354 ipi.ipi_ifindex = m->m_pkthdr.rcvif->if_index; 1355 *mp = sbcreatecontrol((void *) &ipi, 1356 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1357 if (*mp) 1358 mp = &(*mp)->m_next; 1359 } 1360 if (inpflags & INP_PKTINFO) { 1361 struct in_pktinfo ipi; 1362 ipi.ipi_addr = ip->ip_dst; 1363 ipi.ipi_ifindex = m->m_pkthdr.rcvif->if_index; 1364 *mp = sbcreatecontrol((void *) &ipi, 1365 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1366 if (*mp) 1367 mp = &(*mp)->m_next; 1368 } 1369 if (inpflags & INP_RECVIF) { 1370 struct sockaddr_dl sdl; 1371 1372 sockaddr_dl_init(&sdl, sizeof(sdl), m->m_pkthdr.rcvif ? 1373 m->m_pkthdr.rcvif->if_index : 0, 0, NULL, 0, NULL, 0); 1374 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1375 if (*mp) 1376 mp = &(*mp)->m_next; 1377 } 1378 if (inpflags & INP_RECVTTL) { 1379 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1380 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1381 if (*mp) 1382 mp = &(*mp)->m_next; 1383 } 1384} 1385 1386/* 1387 * sysctl helper routine for net.inet.ip.forwsrcrt. 1388 */ 1389static int 1390sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1391{ 1392 int error, tmp; 1393 struct sysctlnode node; 1394 1395 node = *rnode; 1396 tmp = ip_forwsrcrt; 1397 node.sysctl_data = &tmp; 1398 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1399 if (error || newp == NULL) 1400 return (error); 1401 1402 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1403 0, NULL, NULL, NULL); 1404 if (error) 1405 return (error); 1406 1407 ip_forwsrcrt = tmp; 1408 1409 return (0); 1410} 1411 1412/* 1413 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1414 * range of the new value and tweaks timers if it changes. 1415 */ 1416static int 1417sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1418{ 1419 int error, tmp; 1420 struct sysctlnode node; 1421 1422 node = *rnode; 1423 tmp = ip_mtudisc_timeout; 1424 node.sysctl_data = &tmp; 1425 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1426 if (error || newp == NULL) 1427 return (error); 1428 if (tmp < 0) 1429 return (EINVAL); 1430 1431 mutex_enter(softnet_lock); 1432 1433 ip_mtudisc_timeout = tmp; 1434 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1435 1436 mutex_exit(softnet_lock); 1437 1438 return (0); 1439} 1440 1441static int 1442sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1443{ 1444 1445 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1446} 1447 1448static void 1449sysctl_net_inet_ip_setup(struct sysctllog **clog) 1450{ 1451 sysctl_createv(clog, 0, NULL, NULL, 1452 CTLFLAG_PERMANENT, 1453 CTLTYPE_NODE, "inet", 1454 SYSCTL_DESCR("PF_INET related settings"), 1455 NULL, 0, NULL, 0, 1456 CTL_NET, PF_INET, CTL_EOL); 1457 sysctl_createv(clog, 0, NULL, NULL, 1458 CTLFLAG_PERMANENT, 1459 CTLTYPE_NODE, "ip", 1460 SYSCTL_DESCR("IPv4 related settings"), 1461 NULL, 0, NULL, 0, 1462 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1463 1464 sysctl_createv(clog, 0, NULL, NULL, 1465 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1466 CTLTYPE_INT, "forwarding", 1467 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1468 NULL, 0, &ipforwarding, 0, 1469 CTL_NET, PF_INET, IPPROTO_IP, 1470 IPCTL_FORWARDING, CTL_EOL); 1471 sysctl_createv(clog, 0, NULL, NULL, 1472 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1473 CTLTYPE_INT, "redirect", 1474 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1475 NULL, 0, &ipsendredirects, 0, 1476 CTL_NET, PF_INET, IPPROTO_IP, 1477 IPCTL_SENDREDIRECTS, CTL_EOL); 1478 sysctl_createv(clog, 0, NULL, NULL, 1479 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1480 CTLTYPE_INT, "ttl", 1481 SYSCTL_DESCR("Default TTL for an INET datagram"), 1482 NULL, 0, &ip_defttl, 0, 1483 CTL_NET, PF_INET, IPPROTO_IP, 1484 IPCTL_DEFTTL, CTL_EOL); 1485#ifdef IPCTL_DEFMTU 1486 sysctl_createv(clog, 0, NULL, NULL, 1487 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1488 CTLTYPE_INT, "mtu", 1489 SYSCTL_DESCR("Default MTA for an INET route"), 1490 NULL, 0, &ip_mtu, 0, 1491 CTL_NET, PF_INET, IPPROTO_IP, 1492 IPCTL_DEFMTU, CTL_EOL); 1493#endif /* IPCTL_DEFMTU */ 1494 sysctl_createv(clog, 0, NULL, NULL, 1495 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1496 CTLTYPE_INT, "forwsrcrt", 1497 SYSCTL_DESCR("Enable forwarding of source-routed " 1498 "datagrams"), 1499 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1500 CTL_NET, PF_INET, IPPROTO_IP, 1501 IPCTL_FORWSRCRT, CTL_EOL); 1502 sysctl_createv(clog, 0, NULL, NULL, 1503 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1504 CTLTYPE_INT, "directed-broadcast", 1505 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1506 NULL, 0, &ip_directedbcast, 0, 1507 CTL_NET, PF_INET, IPPROTO_IP, 1508 IPCTL_DIRECTEDBCAST, CTL_EOL); 1509 sysctl_createv(clog, 0, NULL, NULL, 1510 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1511 CTLTYPE_INT, "allowsrcrt", 1512 SYSCTL_DESCR("Accept source-routed datagrams"), 1513 NULL, 0, &ip_allowsrcrt, 0, 1514 CTL_NET, PF_INET, IPPROTO_IP, 1515 IPCTL_ALLOWSRCRT, CTL_EOL); 1516 1517 sysctl_createv(clog, 0, NULL, NULL, 1518 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1519 CTLTYPE_INT, "mtudisc", 1520 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1521 NULL, 0, &ip_mtudisc, 0, 1522 CTL_NET, PF_INET, IPPROTO_IP, 1523 IPCTL_MTUDISC, CTL_EOL); 1524 sysctl_createv(clog, 0, NULL, NULL, 1525 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1526 CTLTYPE_INT, "anonportmin", 1527 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1528 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1529 CTL_NET, PF_INET, IPPROTO_IP, 1530 IPCTL_ANONPORTMIN, CTL_EOL); 1531 sysctl_createv(clog, 0, NULL, NULL, 1532 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1533 CTLTYPE_INT, "anonportmax", 1534 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1535 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1536 CTL_NET, PF_INET, IPPROTO_IP, 1537 IPCTL_ANONPORTMAX, CTL_EOL); 1538 sysctl_createv(clog, 0, NULL, NULL, 1539 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1540 CTLTYPE_INT, "mtudisctimeout", 1541 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1542 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1543 CTL_NET, PF_INET, IPPROTO_IP, 1544 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1545#if NGIF > 0 1546 sysctl_createv(clog, 0, NULL, NULL, 1547 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1548 CTLTYPE_INT, "gifttl", 1549 SYSCTL_DESCR("Default TTL for a gif tunnel datagram"), 1550 NULL, 0, &ip_gif_ttl, 0, 1551 CTL_NET, PF_INET, IPPROTO_IP, 1552 IPCTL_GIF_TTL, CTL_EOL); 1553#endif /* NGIF */ 1554#ifndef IPNOPRIVPORTS 1555 sysctl_createv(clog, 0, NULL, NULL, 1556 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1557 CTLTYPE_INT, "lowportmin", 1558 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1559 "to assign"), 1560 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1561 CTL_NET, PF_INET, IPPROTO_IP, 1562 IPCTL_LOWPORTMIN, CTL_EOL); 1563 sysctl_createv(clog, 0, NULL, NULL, 1564 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1565 CTLTYPE_INT, "lowportmax", 1566 SYSCTL_DESCR("Highest privileged ephemeral port number " 1567 "to assign"), 1568 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1569 CTL_NET, PF_INET, IPPROTO_IP, 1570 IPCTL_LOWPORTMAX, CTL_EOL); 1571#endif /* IPNOPRIVPORTS */ 1572#if NGRE > 0 1573 sysctl_createv(clog, 0, NULL, NULL, 1574 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1575 CTLTYPE_INT, "grettl", 1576 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1577 NULL, 0, &ip_gre_ttl, 0, 1578 CTL_NET, PF_INET, IPPROTO_IP, 1579 IPCTL_GRE_TTL, CTL_EOL); 1580#endif /* NGRE */ 1581 sysctl_createv(clog, 0, NULL, NULL, 1582 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1583 CTLTYPE_INT, "checkinterface", 1584 SYSCTL_DESCR("Enable receive side of Strong ES model " 1585 "from RFC1122"), 1586 NULL, 0, &ip_checkinterface, 0, 1587 CTL_NET, PF_INET, IPPROTO_IP, 1588 IPCTL_CHECKINTERFACE, CTL_EOL); 1589 sysctl_createv(clog, 0, NULL, NULL, 1590 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1591 CTLTYPE_INT, "random_id", 1592 SYSCTL_DESCR("Assign random ip_id values"), 1593 NULL, 0, &ip_do_randomid, 0, 1594 CTL_NET, PF_INET, IPPROTO_IP, 1595 IPCTL_RANDOMID, CTL_EOL); 1596 sysctl_createv(clog, 0, NULL, NULL, 1597 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1598 CTLTYPE_INT, "do_loopback_cksum", 1599 SYSCTL_DESCR("Perform IP checksum on loopback"), 1600 NULL, 0, &ip_do_loopback_cksum, 0, 1601 CTL_NET, PF_INET, IPPROTO_IP, 1602 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1603 sysctl_createv(clog, 0, NULL, NULL, 1604 CTLFLAG_PERMANENT, 1605 CTLTYPE_STRUCT, "stats", 1606 SYSCTL_DESCR("IP statistics"), 1607 sysctl_net_inet_ip_stats, 0, NULL, 0, 1608 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1609 CTL_EOL); 1610 1611 /* anonportalgo RFC6056 subtree */ 1612 const struct sysctlnode *portalgo_node; 1613 sysctl_createv(clog, 0, NULL, &portalgo_node, 1614 CTLFLAG_PERMANENT, 1615 CTLTYPE_NODE, "anonportalgo", 1616 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1617 NULL, 0, NULL, 0, 1618 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1619 sysctl_createv(clog, 0, &portalgo_node, NULL, 1620 CTLFLAG_PERMANENT, 1621 CTLTYPE_STRING, "available", 1622 SYSCTL_DESCR("available algorithms"), 1623 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1624 CTL_CREATE, CTL_EOL); 1625 sysctl_createv(clog, 0, &portalgo_node, NULL, 1626 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1627 CTLTYPE_STRING, "selected", 1628 SYSCTL_DESCR("selected algorithm"), 1629 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1630 CTL_CREATE, CTL_EOL); 1631 sysctl_createv(clog, 0, &portalgo_node, NULL, 1632 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1633 CTLTYPE_STRUCT, "reserve", 1634 SYSCTL_DESCR("bitmap of reserved ports"), 1635 sysctl_portalgo_reserve4, 0, NULL, 0, 1636 CTL_CREATE, CTL_EOL); 1637} 1638 1639void 1640ip_statinc(u_int stat) 1641{ 1642 1643 KASSERT(stat < IP_NSTATS); 1644 IP_STATINC(stat); 1645} 1646