ip_input.c revision 1.316
1/* $NetBSD: ip_input.c,v 1.316 2014/05/29 23:02:48 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 1.316 2014/05/29 23:02:48 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 void ip_forward(struct mbuf *, int); 276static bool ip_dooptions(struct mbuf *); 277static struct in_ifaddr *ip_rtaddr(struct in_addr); 278static void sysctl_net_inet_ip_setup(struct sysctllog **); 279 280/* 281 * IP initialization: fill in IP protocol switch table. 282 * All protocols not implemented in kernel go to raw IP protocol handler. 283 */ 284void 285ip_init(void) 286{ 287 const struct protosw *pr; 288 289 in_init(); 290 sysctl_net_inet_ip_setup(NULL); 291 292 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 293 KASSERT(pr != NULL); 294 295 for (u_int i = 0; i < IPPROTO_MAX; i++) { 296 ip_protox[i] = pr - inetsw; 297 } 298 for (pr = inetdomain.dom_protosw; 299 pr < inetdomain.dom_protoswNPROTOSW; pr++) 300 if (pr->pr_domain->dom_family == PF_INET && 301 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 302 ip_protox[pr->pr_protocol] = pr - inetsw; 303 304 ip_reass_init(); 305 306 ip_ids = ip_id_init(); 307 ip_id = time_second & 0xfffff; 308 309 ipintrq.ifq_maxlen = IFQ_MAXLEN; 310 311 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 312#ifdef GATEWAY 313 ipflow_init(); 314#endif 315 316 /* Register our Packet Filter hook. */ 317 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 318 KASSERT(inet_pfil_hook != NULL); 319 320#ifdef MBUFTRACE 321 MOWNER_ATTACH(&ip_tx_mowner); 322 MOWNER_ATTACH(&ip_rx_mowner); 323#endif /* MBUFTRACE */ 324 325 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 326} 327 328struct sockaddr_in ipaddr = { 329 .sin_len = sizeof(ipaddr), 330 .sin_family = AF_INET, 331}; 332 333static struct route ipforward_rt; 334 335/* 336 * IP software interrupt routine 337 */ 338void 339ipintr(void) 340{ 341 int s; 342 struct mbuf *m; 343 struct ifqueue lcl_intrq; 344 345 memset(&lcl_intrq, 0, sizeof(lcl_intrq)); 346 347 mutex_enter(softnet_lock); 348 KERNEL_LOCK(1, NULL); 349 if (!IF_IS_EMPTY(&ipintrq)) { 350 s = splnet(); 351 352 /* Take existing queue onto stack */ 353 lcl_intrq = ipintrq; 354 355 /* Zero out global queue, preserving maxlen and drops */ 356 ipintrq.ifq_head = NULL; 357 ipintrq.ifq_tail = NULL; 358 ipintrq.ifq_len = 0; 359 ipintrq.ifq_maxlen = lcl_intrq.ifq_maxlen; 360 ipintrq.ifq_drops = lcl_intrq.ifq_drops; 361 362 splx(s); 363 } 364 KERNEL_UNLOCK_ONE(NULL); 365 while (!IF_IS_EMPTY(&lcl_intrq)) { 366 IF_DEQUEUE(&lcl_intrq, m); 367 if (m == NULL) 368 break; 369 ip_input(m); 370 } 371 mutex_exit(softnet_lock); 372} 373 374/* 375 * Ip input routine. Checksum and byte swap header. If fragmented 376 * try to reassemble. Process options. Pass to next level. 377 */ 378static void 379ip_input(struct mbuf *m) 380{ 381 struct ip *ip = NULL; 382 struct in_ifaddr *ia; 383 struct ifaddr *ifa; 384 int hlen = 0, len; 385 int downmatch; 386 int checkif; 387 int srcrt = 0; 388 ifnet_t *ifp; 389 390 MCLAIM(m, &ip_rx_mowner); 391 KASSERT((m->m_flags & M_PKTHDR) != 0); 392 ifp = m->m_pkthdr.rcvif; 393 394 /* 395 * If no IP addresses have been set yet but the interfaces 396 * are receiving, can't do anything with incoming packets yet. 397 */ 398 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 399 goto bad; 400 IP_STATINC(IP_STAT_TOTAL); 401 /* 402 * If the IP header is not aligned, slurp it up into a new 403 * mbuf with space for link headers, in the event we forward 404 * it. Otherwise, if it is aligned, make sure the entire 405 * base IP header is in the first mbuf of the chain. 406 */ 407 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 408 if ((m = m_copyup(m, sizeof(struct ip), 409 (max_linkhdr + 3) & ~3)) == NULL) { 410 /* XXXJRT new stat, please */ 411 IP_STATINC(IP_STAT_TOOSMALL); 412 return; 413 } 414 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 415 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 416 IP_STATINC(IP_STAT_TOOSMALL); 417 return; 418 } 419 } 420 ip = mtod(m, struct ip *); 421 if (ip->ip_v != IPVERSION) { 422 IP_STATINC(IP_STAT_BADVERS); 423 goto bad; 424 } 425 hlen = ip->ip_hl << 2; 426 if (hlen < sizeof(struct ip)) { /* minimum header length */ 427 IP_STATINC(IP_STAT_BADHLEN); 428 goto bad; 429 } 430 if (hlen > m->m_len) { 431 if ((m = m_pullup(m, hlen)) == NULL) { 432 IP_STATINC(IP_STAT_BADHLEN); 433 return; 434 } 435 ip = mtod(m, struct ip *); 436 } 437 438 /* 439 * RFC1122: packets with a multicast source address are 440 * not allowed. 441 */ 442 if (IN_MULTICAST(ip->ip_src.s_addr)) { 443 IP_STATINC(IP_STAT_BADADDR); 444 goto bad; 445 } 446 447 /* 127/8 must not appear on wire - RFC1122 */ 448 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 449 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 450 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 451 IP_STATINC(IP_STAT_BADADDR); 452 goto bad; 453 } 454 } 455 456 switch (m->m_pkthdr.csum_flags & 457 ((ifp->if_csum_flags_rx & M_CSUM_IPv4) | 458 M_CSUM_IPv4_BAD)) { 459 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 460 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 461 goto badcsum; 462 463 case M_CSUM_IPv4: 464 /* Checksum was okay. */ 465 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 466 break; 467 468 default: 469 /* 470 * Must compute it ourselves. Maybe skip checksum on 471 * loopback interfaces. 472 */ 473 if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) || 474 ip_do_loopback_cksum)) { 475 INET_CSUM_COUNTER_INCR(&ip_swcsum); 476 if (in_cksum(m, hlen) != 0) 477 goto badcsum; 478 } 479 break; 480 } 481 482 /* Retrieve the packet length. */ 483 len = ntohs(ip->ip_len); 484 485 /* 486 * Check for additional length bogosity 487 */ 488 if (len < hlen) { 489 IP_STATINC(IP_STAT_BADLEN); 490 goto bad; 491 } 492 493 /* 494 * Check that the amount of data in the buffers 495 * is as at least much as the IP header would have us expect. 496 * Trim mbufs if longer than we expect. 497 * Drop packet if shorter than we expect. 498 */ 499 if (m->m_pkthdr.len < len) { 500 IP_STATINC(IP_STAT_TOOSHORT); 501 goto bad; 502 } 503 if (m->m_pkthdr.len > len) { 504 if (m->m_len == m->m_pkthdr.len) { 505 m->m_len = len; 506 m->m_pkthdr.len = len; 507 } else 508 m_adj(m, len - m->m_pkthdr.len); 509 } 510 511 /* 512 * Assume that we can create a fast-forward IP flow entry 513 * based on this packet. 514 */ 515 m->m_flags |= M_CANFASTFWD; 516 517 /* 518 * Run through list of hooks for input packets. If there are any 519 * filters which require that additional packets in the flow are 520 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 521 * Note that filters must _never_ set this flag, as another filter 522 * in the list may have previously cleared it. 523 */ 524#if defined(IPSEC) 525 if (!ipsec_indone(m)) 526#else 527 if (1) 528#endif 529 { 530 struct in_addr odst = ip->ip_dst; 531 532 if (pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0) 533 return; 534 if (m == NULL) 535 return; 536 ip = mtod(m, struct ip *); 537 hlen = ip->ip_hl << 2; 538 /* 539 * XXX The setting of "srcrt" here is to prevent ip_forward() 540 * from generating ICMP redirects for packets that have 541 * been redirected by a hook back out on to the same LAN that 542 * they came from and is not an indication that the packet 543 * is being inffluenced by source routing options. This 544 * allows things like 545 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 546 * where tlp0 is both on the 1.1.1.0/24 network and is the 547 * default route for hosts on 1.1.1.0/24. Of course this 548 * also requires a "map tlp0 ..." to complete the story. 549 * One might argue whether or not this kind of network config. 550 * should be supported in this manner... 551 */ 552 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 553 } 554 555#ifdef ALTQ 556 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 557 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) { 558 /* packet dropped by traffic conditioner */ 559 return; 560 } 561#endif 562 563 /* 564 * Process options and, if not destined for us, 565 * ship it on. ip_dooptions returns 1 when an 566 * error was detected (causing an icmp message 567 * to be sent and the original packet to be freed). 568 */ 569 ip_nhops = 0; /* for source routed packets */ 570 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 571 return; 572 573 /* 574 * Enable a consistency check between the destination address 575 * and the arrival interface for a unicast packet (the RFC 1122 576 * strong ES model) if IP forwarding is disabled and the packet 577 * is not locally generated. 578 * 579 * XXX - Checking also should be disabled if the destination 580 * address is ipnat'ed to a different interface. 581 * 582 * XXX - Checking is incompatible with IP aliases added 583 * to the loopback interface instead of the interface where 584 * the packets are received. 585 * 586 * XXX - We need to add a per ifaddr flag for this so that 587 * we get finer grain control. 588 */ 589 checkif = ip_checkinterface && (ipforwarding == 0) && 590 (ifp->if_flags & IFF_LOOPBACK) == 0; 591 592 /* 593 * Check our list of addresses, to see if the packet is for us. 594 * 595 * Traditional 4.4BSD did not consult IFF_UP at all. 596 * The behavior here is to treat addresses on !IFF_UP interface 597 * as not mine. 598 */ 599 downmatch = 0; 600 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 601 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 602 if (checkif && ia->ia_ifp != ifp) 603 continue; 604 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 605 break; 606 else 607 downmatch++; 608 } 609 } 610 if (ia != NULL) 611 goto ours; 612 if (ifp->if_flags & IFF_BROADCAST) { 613 IFADDR_FOREACH(ifa, ifp) { 614 if (ifa->ifa_addr->sa_family != AF_INET) 615 continue; 616 ia = ifatoia(ifa); 617 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 618 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 619 /* 620 * Look for all-0's host part (old broadcast addr), 621 * either for subnet or net. 622 */ 623 ip->ip_dst.s_addr == ia->ia_subnet || 624 ip->ip_dst.s_addr == ia->ia_net) 625 goto ours; 626 /* 627 * An interface with IP address zero accepts 628 * all packets that arrive on that interface. 629 */ 630 if (in_nullhost(ia->ia_addr.sin_addr)) 631 goto ours; 632 } 633 } 634 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 635#ifdef MROUTING 636 extern struct socket *ip_mrouter; 637 638 if (ip_mrouter) { 639 /* 640 * If we are acting as a multicast router, all 641 * incoming multicast packets are passed to the 642 * kernel-level multicast forwarding function. 643 * The packet is returned (relatively) intact; if 644 * ip_mforward() returns a non-zero value, the packet 645 * must be discarded, else it may be accepted below. 646 * 647 * (The IP ident field is put in the same byte order 648 * as expected when ip_mforward() is called from 649 * ip_output().) 650 */ 651 if (ip_mforward(m, ifp) != 0) { 652 IP_STATINC(IP_STAT_CANTFORWARD); 653 m_freem(m); 654 return; 655 } 656 657 /* 658 * The process-level routing demon needs to receive 659 * all multicast IGMP packets, whether or not this 660 * host belongs to their destination groups. 661 */ 662 if (ip->ip_p == IPPROTO_IGMP) 663 goto ours; 664 IP_STATINC(IP_STAT_CANTFORWARD); 665 } 666#endif 667 /* 668 * See if we belong to the destination multicast group on the 669 * arrival interface. 670 */ 671 if (!in_multi_group(ip->ip_dst, ifp, 0)) { 672 IP_STATINC(IP_STAT_CANTFORWARD); 673 m_freem(m); 674 return; 675 } 676 goto ours; 677 } 678 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 679 in_nullhost(ip->ip_dst)) 680 goto ours; 681 682 /* 683 * Not for us; forward if possible and desirable. 684 */ 685 if (ipforwarding == 0) { 686 IP_STATINC(IP_STAT_CANTFORWARD); 687 m_freem(m); 688 } else { 689 /* 690 * If ip_dst matched any of my address on !IFF_UP interface, 691 * and there's no IFF_UP interface that matches ip_dst, 692 * send icmp unreach. Forwarding it will result in in-kernel 693 * forwarding loop till TTL goes to 0. 694 */ 695 if (downmatch) { 696 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 697 IP_STATINC(IP_STAT_CANTFORWARD); 698 return; 699 } 700#ifdef IPSEC 701 /* Perform IPsec, if any. */ 702 if (ipsec4_input(m, IP_FORWARDING | (ip_directedbcast ? 703 IP_ALLOWBROADCAST : 0)) != 0) { 704 goto bad; 705 } 706#endif 707 ip_forward(m, srcrt); 708 } 709 return; 710 711ours: 712 /* 713 * If offset or IP_MF are set, must reassemble. 714 */ 715 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 716 /* 717 * Pass to IP reassembly mechanism. 718 */ 719 if (ip_reass_packet(&m, ip) != 0) { 720 /* Failed; invalid fragment(s) or packet. */ 721 goto bad; 722 } 723 if (m == NULL) { 724 /* More fragments should come; silently return. */ 725 return; 726 } 727 /* 728 * Reassembly is done, we have the final packet. 729 * Updated cached data in local variable(s). 730 */ 731 ip = mtod(m, struct ip *); 732 hlen = ip->ip_hl << 2; 733 } 734 735#ifdef IPSEC 736 /* 737 * Enforce IPsec policy checking if we are seeing last header. 738 * Note that we do not visit this with protocols with PCB layer 739 * code - like UDP/TCP/raw IP. 740 */ 741 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 742 if (ipsec4_input(m, 0) != 0) { 743 goto bad; 744 } 745 } 746#endif 747 748 /* 749 * Switch out to protocol's input routine. 750 */ 751#if IFA_STATS 752 if (ia && ip) 753 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 754#endif 755 IP_STATINC(IP_STAT_DELIVERED); 756 { 757 int off = hlen, nh = ip->ip_p; 758 759 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 760 return; 761 } 762bad: 763 m_freem(m); 764 return; 765 766badcsum: 767 IP_STATINC(IP_STAT_BADSUM); 768 m_freem(m); 769} 770 771/* 772 * IP timer processing. 773 */ 774void 775ip_slowtimo(void) 776{ 777 778 mutex_enter(softnet_lock); 779 KERNEL_LOCK(1, NULL); 780 781 ip_reass_slowtimo(); 782 783 KERNEL_UNLOCK_ONE(NULL); 784 mutex_exit(softnet_lock); 785} 786 787/* 788 * IP drain processing. 789 */ 790void 791ip_drain(void) 792{ 793 794 KERNEL_LOCK(1, NULL); 795 ip_reass_drain(); 796 KERNEL_UNLOCK_ONE(NULL); 797} 798 799/* 800 * ip_dooptions: perform option processing on a datagram, possibly discarding 801 * it if bad options are encountered, or forwarding it if source-routed. 802 * 803 * => Returns true if packet has been forwarded/freed. 804 * => Returns false if the packet should be processed further. 805 */ 806static bool 807ip_dooptions(struct mbuf *m) 808{ 809 struct ip *ip = mtod(m, struct ip *); 810 u_char *cp, *cp0; 811 struct ip_timestamp *ipt; 812 struct in_ifaddr *ia; 813 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 814 struct in_addr dst; 815 n_time ntime; 816 817 dst = ip->ip_dst; 818 cp = (u_char *)(ip + 1); 819 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 820 for (; cnt > 0; cnt -= optlen, cp += optlen) { 821 opt = cp[IPOPT_OPTVAL]; 822 if (opt == IPOPT_EOL) 823 break; 824 if (opt == IPOPT_NOP) 825 optlen = 1; 826 else { 827 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 828 code = &cp[IPOPT_OLEN] - (u_char *)ip; 829 goto bad; 830 } 831 optlen = cp[IPOPT_OLEN]; 832 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 833 code = &cp[IPOPT_OLEN] - (u_char *)ip; 834 goto bad; 835 } 836 } 837 switch (opt) { 838 839 default: 840 break; 841 842 /* 843 * Source routing with record. 844 * Find interface with current destination address. 845 * If none on this machine then drop if strictly routed, 846 * or do nothing if loosely routed. 847 * Record interface address and bring up next address 848 * component. If strictly routed make sure next 849 * address is on directly accessible net. 850 */ 851 case IPOPT_LSRR: 852 case IPOPT_SSRR: 853 if (ip_allowsrcrt == 0) { 854 type = ICMP_UNREACH; 855 code = ICMP_UNREACH_NET_PROHIB; 856 goto bad; 857 } 858 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 859 code = &cp[IPOPT_OLEN] - (u_char *)ip; 860 goto bad; 861 } 862 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 863 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 864 goto bad; 865 } 866 ipaddr.sin_addr = ip->ip_dst; 867 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 868 if (ia == 0) { 869 if (opt == IPOPT_SSRR) { 870 type = ICMP_UNREACH; 871 code = ICMP_UNREACH_SRCFAIL; 872 goto bad; 873 } 874 /* 875 * Loose routing, and not at next destination 876 * yet; nothing to do except forward. 877 */ 878 break; 879 } 880 off--; /* 0 origin */ 881 if ((off + sizeof(struct in_addr)) > optlen) { 882 /* 883 * End of source route. Should be for us. 884 */ 885 save_rte(cp, ip->ip_src); 886 break; 887 } 888 /* 889 * locate outgoing interface 890 */ 891 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 892 sizeof(ipaddr.sin_addr)); 893 if (opt == IPOPT_SSRR) 894 ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); 895 else 896 ia = ip_rtaddr(ipaddr.sin_addr); 897 if (ia == 0) { 898 type = ICMP_UNREACH; 899 code = ICMP_UNREACH_SRCFAIL; 900 goto bad; 901 } 902 ip->ip_dst = ipaddr.sin_addr; 903 bcopy((void *)&ia->ia_addr.sin_addr, 904 (void *)(cp + off), sizeof(struct in_addr)); 905 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 906 /* 907 * Let ip_intr's mcast routing check handle mcast pkts 908 */ 909 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 910 break; 911 912 case IPOPT_RR: 913 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 914 code = &cp[IPOPT_OLEN] - (u_char *)ip; 915 goto bad; 916 } 917 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 918 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 919 goto bad; 920 } 921 /* 922 * If no space remains, ignore. 923 */ 924 off--; /* 0 origin */ 925 if ((off + sizeof(struct in_addr)) > optlen) 926 break; 927 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 928 sizeof(ipaddr.sin_addr)); 929 /* 930 * locate outgoing interface; if we're the destination, 931 * use the incoming interface (should be same). 932 */ 933 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 934 == NULL && 935 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 936 type = ICMP_UNREACH; 937 code = ICMP_UNREACH_HOST; 938 goto bad; 939 } 940 bcopy((void *)&ia->ia_addr.sin_addr, 941 (void *)(cp + off), sizeof(struct in_addr)); 942 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 943 break; 944 945 case IPOPT_TS: 946 code = cp - (u_char *)ip; 947 ipt = (struct ip_timestamp *)cp; 948 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 949 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 950 goto bad; 951 } 952 if (ipt->ipt_ptr < 5) { 953 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 954 goto bad; 955 } 956 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 957 if (++ipt->ipt_oflw == 0) { 958 code = (u_char *)&ipt->ipt_ptr - 959 (u_char *)ip; 960 goto bad; 961 } 962 break; 963 } 964 cp0 = (cp + ipt->ipt_ptr - 1); 965 switch (ipt->ipt_flg) { 966 967 case IPOPT_TS_TSONLY: 968 break; 969 970 case IPOPT_TS_TSANDADDR: 971 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 972 sizeof(struct in_addr) > ipt->ipt_len) { 973 code = (u_char *)&ipt->ipt_ptr - 974 (u_char *)ip; 975 goto bad; 976 } 977 ipaddr.sin_addr = dst; 978 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 979 m->m_pkthdr.rcvif)); 980 if (ia == 0) 981 continue; 982 bcopy(&ia->ia_addr.sin_addr, 983 cp0, sizeof(struct in_addr)); 984 ipt->ipt_ptr += sizeof(struct in_addr); 985 break; 986 987 case IPOPT_TS_PRESPEC: 988 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 989 sizeof(struct in_addr) > ipt->ipt_len) { 990 code = (u_char *)&ipt->ipt_ptr - 991 (u_char *)ip; 992 goto bad; 993 } 994 memcpy(&ipaddr.sin_addr, cp0, 995 sizeof(struct in_addr)); 996 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 997 == NULL) 998 continue; 999 ipt->ipt_ptr += sizeof(struct in_addr); 1000 break; 1001 1002 default: 1003 /* XXX can't take &ipt->ipt_flg */ 1004 code = (u_char *)&ipt->ipt_ptr - 1005 (u_char *)ip + 1; 1006 goto bad; 1007 } 1008 ntime = iptime(); 1009 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1010 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1011 sizeof(n_time)); 1012 ipt->ipt_ptr += sizeof(n_time); 1013 } 1014 } 1015 if (forward) { 1016 if (ip_forwsrcrt == 0) { 1017 type = ICMP_UNREACH; 1018 code = ICMP_UNREACH_SRCFAIL; 1019 goto bad; 1020 } 1021 ip_forward(m, 1); 1022 return true; 1023 } 1024 return false; 1025bad: 1026 icmp_error(m, type, code, 0, 0); 1027 IP_STATINC(IP_STAT_BADOPTIONS); 1028 return true; 1029} 1030 1031/* 1032 * ip_rtaddr: given address of next destination (final or next hop), 1033 * return internet address info of interface to be used to get there. 1034 */ 1035static struct in_ifaddr * 1036ip_rtaddr(struct in_addr dst) 1037{ 1038 struct rtentry *rt; 1039 union { 1040 struct sockaddr dst; 1041 struct sockaddr_in dst4; 1042 } u; 1043 1044 sockaddr_in_init(&u.dst4, &dst, 0); 1045 1046 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) 1047 return NULL; 1048 1049 return ifatoia(rt->rt_ifa); 1050} 1051 1052/* 1053 * save_rte: save incoming source route for use in replies, to be picked 1054 * up later by ip_srcroute if the receiver is interested. 1055 */ 1056static void 1057save_rte(u_char *option, struct in_addr dst) 1058{ 1059 unsigned olen; 1060 1061 olen = option[IPOPT_OLEN]; 1062 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1063 return; 1064 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1065 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1066 ip_srcrt.dst = dst; 1067} 1068 1069/* 1070 * Retrieve incoming source route for use in replies, 1071 * in the same form used by setsockopt. 1072 * The first hop is placed before the options, will be removed later. 1073 */ 1074struct mbuf * 1075ip_srcroute(void) 1076{ 1077 struct in_addr *p, *q; 1078 struct mbuf *m; 1079 1080 if (ip_nhops == 0) 1081 return NULL; 1082 m = m_get(M_DONTWAIT, MT_SOOPTS); 1083 if (m == 0) 1084 return NULL; 1085 1086 MCLAIM(m, &inetdomain.dom_mowner); 1087#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1088 1089 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1090 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1091 OPTSIZ; 1092 1093 /* 1094 * First save first hop for return route 1095 */ 1096 p = &ip_srcrt.route[ip_nhops - 1]; 1097 *(mtod(m, struct in_addr *)) = *p--; 1098 1099 /* 1100 * Copy option fields and padding (nop) to mbuf. 1101 */ 1102 ip_srcrt.nop = IPOPT_NOP; 1103 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1104 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1105 OPTSIZ); 1106 q = (struct in_addr *)(mtod(m, char *) + 1107 sizeof(struct in_addr) + OPTSIZ); 1108#undef OPTSIZ 1109 /* 1110 * Record return path as an IP source route, 1111 * reversing the path (pointers are now aligned). 1112 */ 1113 while (p >= ip_srcrt.route) { 1114 *q++ = *p--; 1115 } 1116 /* 1117 * Last hop goes to final destination. 1118 */ 1119 *q = ip_srcrt.dst; 1120 return (m); 1121} 1122 1123const int inetctlerrmap[PRC_NCMDS] = { 1124 [PRC_MSGSIZE] = EMSGSIZE, 1125 [PRC_HOSTDEAD] = EHOSTDOWN, 1126 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1127 [PRC_UNREACH_NET] = EHOSTUNREACH, 1128 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1129 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1130 [PRC_UNREACH_PORT] = ECONNREFUSED, 1131 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1132 [PRC_PARAMPROB] = ENOPROTOOPT, 1133}; 1134 1135void 1136ip_fasttimo(void) 1137{ 1138 if (ip_drainwanted) { 1139 ip_drain(); 1140 ip_drainwanted = 0; 1141 } 1142} 1143 1144void 1145ip_drainstub(void) 1146{ 1147 ip_drainwanted = 1; 1148} 1149 1150/* 1151 * Forward a packet. If some error occurs return the sender 1152 * an icmp packet. Note we can't always generate a meaningful 1153 * icmp message because icmp doesn't have a large enough repertoire 1154 * of codes and types. 1155 * 1156 * If not forwarding, just drop the packet. This could be confusing 1157 * if ipforwarding was zero but some routing protocol was advancing 1158 * us as a gateway to somewhere. However, we must let the routing 1159 * protocol deal with that. 1160 * 1161 * The srcrt parameter indicates whether the packet is being forwarded 1162 * via a source route. 1163 */ 1164static void 1165ip_forward(struct mbuf *m, int srcrt) 1166{ 1167 struct ip *ip = mtod(m, struct ip *); 1168 struct rtentry *rt; 1169 int error, type = 0, code = 0, destmtu = 0; 1170 struct mbuf *mcopy; 1171 n_long dest; 1172 union { 1173 struct sockaddr dst; 1174 struct sockaddr_in dst4; 1175 } u; 1176 1177 /* 1178 * We are now in the output path. 1179 */ 1180 MCLAIM(m, &ip_tx_mowner); 1181 1182 /* 1183 * Clear any in-bound checksum flags for this packet. 1184 */ 1185 m->m_pkthdr.csum_flags = 0; 1186 1187 dest = 0; 1188 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1189 IP_STATINC(IP_STAT_CANTFORWARD); 1190 m_freem(m); 1191 return; 1192 } 1193 if (ip->ip_ttl <= IPTTLDEC) { 1194 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1195 return; 1196 } 1197 1198 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1199 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1200 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1201 return; 1202 } 1203 1204 /* 1205 * Save at most 68 bytes of the packet in case 1206 * we need to generate an ICMP message to the src. 1207 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1208 */ 1209 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1210 if (mcopy) 1211 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1212 1213 ip->ip_ttl -= IPTTLDEC; 1214 1215 /* 1216 * If forwarding packet using same interface that it came in on, 1217 * perhaps should send a redirect to sender to shortcut a hop. 1218 * Only send redirect if source is sending directly to us, 1219 * and if packet was not source routed (or has any options). 1220 * Also, don't send redirect if forwarding using a default route 1221 * or a route modified by a redirect. 1222 */ 1223 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1224 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1225 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1226 ipsendredirects && !srcrt) { 1227 if (rt->rt_ifa && 1228 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1229 ifatoia(rt->rt_ifa)->ia_subnet) { 1230 if (rt->rt_flags & RTF_GATEWAY) 1231 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1232 else 1233 dest = ip->ip_dst.s_addr; 1234 /* 1235 * Router requirements says to only send host 1236 * redirects. 1237 */ 1238 type = ICMP_REDIRECT; 1239 code = ICMP_REDIRECT_HOST; 1240 } 1241 } 1242 1243 error = ip_output(m, NULL, &ipforward_rt, 1244 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1245 NULL, NULL); 1246 1247 if (error) 1248 IP_STATINC(IP_STAT_CANTFORWARD); 1249 else { 1250 uint64_t *ips = IP_STAT_GETREF(); 1251 ips[IP_STAT_FORWARD]++; 1252 if (type) { 1253 ips[IP_STAT_REDIRECTSENT]++; 1254 IP_STAT_PUTREF(); 1255 } else { 1256 IP_STAT_PUTREF(); 1257 if (mcopy) { 1258#ifdef GATEWAY 1259 if (mcopy->m_flags & M_CANFASTFWD) 1260 ipflow_create(&ipforward_rt, mcopy); 1261#endif 1262 m_freem(mcopy); 1263 } 1264 return; 1265 } 1266 } 1267 if (mcopy == NULL) 1268 return; 1269 1270 switch (error) { 1271 1272 case 0: /* forwarded, but need redirect */ 1273 /* type, code set above */ 1274 break; 1275 1276 case ENETUNREACH: /* shouldn't happen, checked above */ 1277 case EHOSTUNREACH: 1278 case ENETDOWN: 1279 case EHOSTDOWN: 1280 default: 1281 type = ICMP_UNREACH; 1282 code = ICMP_UNREACH_HOST; 1283 break; 1284 1285 case EMSGSIZE: 1286 type = ICMP_UNREACH; 1287 code = ICMP_UNREACH_NEEDFRAG; 1288 1289 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1290 destmtu = rt->rt_ifp->if_mtu; 1291#ifdef IPSEC 1292 (void)ipsec4_forward(mcopy, &destmtu); 1293#endif 1294 IP_STATINC(IP_STAT_CANTFRAG); 1295 break; 1296 1297 case ENOBUFS: 1298 /* 1299 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1300 * Requirements for IP Version 4 Routers. Source quench can 1301 * big problem under DoS attacks or if the underlying 1302 * interface is rate-limited. 1303 */ 1304 if (mcopy) 1305 m_freem(mcopy); 1306 return; 1307 } 1308 icmp_error(mcopy, type, code, dest, destmtu); 1309} 1310 1311void 1312ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1313 struct mbuf *m) 1314{ 1315 struct socket *so = inp->inp_socket; 1316 ifnet_t *ifp = m->m_pkthdr.rcvif; 1317 int inpflags = inp->inp_flags; 1318 1319 if (so->so_options & SO_TIMESTAMP 1320#ifdef SO_OTIMESTAMP 1321 || so->so_options & SO_OTIMESTAMP 1322#endif 1323 ) { 1324 struct timeval tv; 1325 1326 microtime(&tv); 1327#ifdef SO_OTIMESTAMP 1328 if (so->so_options & SO_OTIMESTAMP) { 1329 struct timeval50 tv50; 1330 timeval_to_timeval50(&tv, &tv50); 1331 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1332 SCM_OTIMESTAMP, SOL_SOCKET); 1333 } else 1334#endif 1335 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1336 SCM_TIMESTAMP, SOL_SOCKET); 1337 if (*mp) 1338 mp = &(*mp)->m_next; 1339 } 1340 if (inpflags & INP_RECVDSTADDR) { 1341 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1342 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1343 if (*mp) 1344 mp = &(*mp)->m_next; 1345 } 1346 if (inpflags & INP_RECVPKTINFO) { 1347 struct in_pktinfo ipi; 1348 ipi.ipi_addr = ip->ip_src; 1349 ipi.ipi_ifindex = ifp->if_index; 1350 *mp = sbcreatecontrol((void *) &ipi, 1351 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1352 if (*mp) 1353 mp = &(*mp)->m_next; 1354 } 1355 if (inpflags & INP_PKTINFO) { 1356 struct in_pktinfo ipi; 1357 ipi.ipi_addr = ip->ip_dst; 1358 ipi.ipi_ifindex = ifp->if_index; 1359 *mp = sbcreatecontrol((void *) &ipi, 1360 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1361 if (*mp) 1362 mp = &(*mp)->m_next; 1363 } 1364 if (inpflags & INP_RECVIF) { 1365 struct sockaddr_dl sdl; 1366 1367 sockaddr_dl_init(&sdl, sizeof(sdl), ifp ? 1368 ifp->if_index : 0, 0, NULL, 0, NULL, 0); 1369 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1370 if (*mp) 1371 mp = &(*mp)->m_next; 1372 } 1373 if (inpflags & INP_RECVTTL) { 1374 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1375 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1376 if (*mp) 1377 mp = &(*mp)->m_next; 1378 } 1379} 1380 1381/* 1382 * sysctl helper routine for net.inet.ip.forwsrcrt. 1383 */ 1384static int 1385sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1386{ 1387 int error, tmp; 1388 struct sysctlnode node; 1389 1390 node = *rnode; 1391 tmp = ip_forwsrcrt; 1392 node.sysctl_data = &tmp; 1393 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1394 if (error || newp == NULL) 1395 return (error); 1396 1397 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1398 0, NULL, NULL, NULL); 1399 if (error) 1400 return (error); 1401 1402 ip_forwsrcrt = tmp; 1403 1404 return (0); 1405} 1406 1407/* 1408 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1409 * range of the new value and tweaks timers if it changes. 1410 */ 1411static int 1412sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1413{ 1414 int error, tmp; 1415 struct sysctlnode node; 1416 1417 node = *rnode; 1418 tmp = ip_mtudisc_timeout; 1419 node.sysctl_data = &tmp; 1420 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1421 if (error || newp == NULL) 1422 return (error); 1423 if (tmp < 0) 1424 return (EINVAL); 1425 1426 mutex_enter(softnet_lock); 1427 1428 ip_mtudisc_timeout = tmp; 1429 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1430 1431 mutex_exit(softnet_lock); 1432 1433 return (0); 1434} 1435 1436static int 1437sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1438{ 1439 1440 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1441} 1442 1443static void 1444sysctl_net_inet_ip_setup(struct sysctllog **clog) 1445{ 1446 sysctl_createv(clog, 0, NULL, NULL, 1447 CTLFLAG_PERMANENT, 1448 CTLTYPE_NODE, "inet", 1449 SYSCTL_DESCR("PF_INET related settings"), 1450 NULL, 0, NULL, 0, 1451 CTL_NET, PF_INET, CTL_EOL); 1452 sysctl_createv(clog, 0, NULL, NULL, 1453 CTLFLAG_PERMANENT, 1454 CTLTYPE_NODE, "ip", 1455 SYSCTL_DESCR("IPv4 related settings"), 1456 NULL, 0, NULL, 0, 1457 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1458 1459 sysctl_createv(clog, 0, NULL, NULL, 1460 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1461 CTLTYPE_INT, "forwarding", 1462 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1463 NULL, 0, &ipforwarding, 0, 1464 CTL_NET, PF_INET, IPPROTO_IP, 1465 IPCTL_FORWARDING, CTL_EOL); 1466 sysctl_createv(clog, 0, NULL, NULL, 1467 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1468 CTLTYPE_INT, "redirect", 1469 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1470 NULL, 0, &ipsendredirects, 0, 1471 CTL_NET, PF_INET, IPPROTO_IP, 1472 IPCTL_SENDREDIRECTS, CTL_EOL); 1473 sysctl_createv(clog, 0, NULL, NULL, 1474 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1475 CTLTYPE_INT, "ttl", 1476 SYSCTL_DESCR("Default TTL for an INET datagram"), 1477 NULL, 0, &ip_defttl, 0, 1478 CTL_NET, PF_INET, IPPROTO_IP, 1479 IPCTL_DEFTTL, CTL_EOL); 1480#ifdef IPCTL_DEFMTU 1481 sysctl_createv(clog, 0, NULL, NULL, 1482 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1483 CTLTYPE_INT, "mtu", 1484 SYSCTL_DESCR("Default MTA for an INET route"), 1485 NULL, 0, &ip_mtu, 0, 1486 CTL_NET, PF_INET, IPPROTO_IP, 1487 IPCTL_DEFMTU, CTL_EOL); 1488#endif /* IPCTL_DEFMTU */ 1489 sysctl_createv(clog, 0, NULL, NULL, 1490 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1491 CTLTYPE_INT, "forwsrcrt", 1492 SYSCTL_DESCR("Enable forwarding of source-routed " 1493 "datagrams"), 1494 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1495 CTL_NET, PF_INET, IPPROTO_IP, 1496 IPCTL_FORWSRCRT, CTL_EOL); 1497 sysctl_createv(clog, 0, NULL, NULL, 1498 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1499 CTLTYPE_INT, "directed-broadcast", 1500 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1501 NULL, 0, &ip_directedbcast, 0, 1502 CTL_NET, PF_INET, IPPROTO_IP, 1503 IPCTL_DIRECTEDBCAST, CTL_EOL); 1504 sysctl_createv(clog, 0, NULL, NULL, 1505 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1506 CTLTYPE_INT, "allowsrcrt", 1507 SYSCTL_DESCR("Accept source-routed datagrams"), 1508 NULL, 0, &ip_allowsrcrt, 0, 1509 CTL_NET, PF_INET, IPPROTO_IP, 1510 IPCTL_ALLOWSRCRT, CTL_EOL); 1511 1512 sysctl_createv(clog, 0, NULL, NULL, 1513 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1514 CTLTYPE_INT, "mtudisc", 1515 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1516 NULL, 0, &ip_mtudisc, 0, 1517 CTL_NET, PF_INET, IPPROTO_IP, 1518 IPCTL_MTUDISC, CTL_EOL); 1519 sysctl_createv(clog, 0, NULL, NULL, 1520 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1521 CTLTYPE_INT, "anonportmin", 1522 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1523 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1524 CTL_NET, PF_INET, IPPROTO_IP, 1525 IPCTL_ANONPORTMIN, CTL_EOL); 1526 sysctl_createv(clog, 0, NULL, NULL, 1527 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1528 CTLTYPE_INT, "anonportmax", 1529 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1530 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1531 CTL_NET, PF_INET, IPPROTO_IP, 1532 IPCTL_ANONPORTMAX, CTL_EOL); 1533 sysctl_createv(clog, 0, NULL, NULL, 1534 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1535 CTLTYPE_INT, "mtudisctimeout", 1536 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1537 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1538 CTL_NET, PF_INET, IPPROTO_IP, 1539 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1540#if NGIF > 0 1541 sysctl_createv(clog, 0, NULL, NULL, 1542 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1543 CTLTYPE_INT, "gifttl", 1544 SYSCTL_DESCR("Default TTL for a gif tunnel datagram"), 1545 NULL, 0, &ip_gif_ttl, 0, 1546 CTL_NET, PF_INET, IPPROTO_IP, 1547 IPCTL_GIF_TTL, CTL_EOL); 1548#endif /* NGIF */ 1549#ifndef IPNOPRIVPORTS 1550 sysctl_createv(clog, 0, NULL, NULL, 1551 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1552 CTLTYPE_INT, "lowportmin", 1553 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1554 "to assign"), 1555 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1556 CTL_NET, PF_INET, IPPROTO_IP, 1557 IPCTL_LOWPORTMIN, CTL_EOL); 1558 sysctl_createv(clog, 0, NULL, NULL, 1559 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1560 CTLTYPE_INT, "lowportmax", 1561 SYSCTL_DESCR("Highest privileged ephemeral port number " 1562 "to assign"), 1563 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1564 CTL_NET, PF_INET, IPPROTO_IP, 1565 IPCTL_LOWPORTMAX, CTL_EOL); 1566#endif /* IPNOPRIVPORTS */ 1567#if NGRE > 0 1568 sysctl_createv(clog, 0, NULL, NULL, 1569 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1570 CTLTYPE_INT, "grettl", 1571 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1572 NULL, 0, &ip_gre_ttl, 0, 1573 CTL_NET, PF_INET, IPPROTO_IP, 1574 IPCTL_GRE_TTL, CTL_EOL); 1575#endif /* NGRE */ 1576 sysctl_createv(clog, 0, NULL, NULL, 1577 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1578 CTLTYPE_INT, "checkinterface", 1579 SYSCTL_DESCR("Enable receive side of Strong ES model " 1580 "from RFC1122"), 1581 NULL, 0, &ip_checkinterface, 0, 1582 CTL_NET, PF_INET, IPPROTO_IP, 1583 IPCTL_CHECKINTERFACE, CTL_EOL); 1584 sysctl_createv(clog, 0, NULL, NULL, 1585 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1586 CTLTYPE_INT, "random_id", 1587 SYSCTL_DESCR("Assign random ip_id values"), 1588 NULL, 0, &ip_do_randomid, 0, 1589 CTL_NET, PF_INET, IPPROTO_IP, 1590 IPCTL_RANDOMID, CTL_EOL); 1591 sysctl_createv(clog, 0, NULL, NULL, 1592 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1593 CTLTYPE_INT, "do_loopback_cksum", 1594 SYSCTL_DESCR("Perform IP checksum on loopback"), 1595 NULL, 0, &ip_do_loopback_cksum, 0, 1596 CTL_NET, PF_INET, IPPROTO_IP, 1597 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1598 sysctl_createv(clog, 0, NULL, NULL, 1599 CTLFLAG_PERMANENT, 1600 CTLTYPE_STRUCT, "stats", 1601 SYSCTL_DESCR("IP statistics"), 1602 sysctl_net_inet_ip_stats, 0, NULL, 0, 1603 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1604 CTL_EOL); 1605 1606 /* anonportalgo RFC6056 subtree */ 1607 const struct sysctlnode *portalgo_node; 1608 sysctl_createv(clog, 0, NULL, &portalgo_node, 1609 CTLFLAG_PERMANENT, 1610 CTLTYPE_NODE, "anonportalgo", 1611 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1612 NULL, 0, NULL, 0, 1613 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1614 sysctl_createv(clog, 0, &portalgo_node, NULL, 1615 CTLFLAG_PERMANENT, 1616 CTLTYPE_STRING, "available", 1617 SYSCTL_DESCR("available algorithms"), 1618 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1619 CTL_CREATE, CTL_EOL); 1620 sysctl_createv(clog, 0, &portalgo_node, NULL, 1621 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1622 CTLTYPE_STRING, "selected", 1623 SYSCTL_DESCR("selected algorithm"), 1624 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1625 CTL_CREATE, CTL_EOL); 1626 sysctl_createv(clog, 0, &portalgo_node, NULL, 1627 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1628 CTLTYPE_STRUCT, "reserve", 1629 SYSCTL_DESCR("bitmap of reserved ports"), 1630 sysctl_portalgo_reserve4, 0, NULL, 0, 1631 CTL_CREATE, CTL_EOL); 1632} 1633 1634void 1635ip_statinc(u_int stat) 1636{ 1637 1638 KASSERT(stat < IP_NSTATS); 1639 IP_STATINC(stat); 1640} 1641