ip_input.c revision 1.324
1/* $NetBSD: ip_input.c,v 1.324 2015/08/24 22:21:26 pooka 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 1.324 2015/08/24 22:21:26 pooka Exp $"); 95 96#ifdef _KERNEL_OPT 97#include "opt_inet.h" 98#include "opt_compat_netbsd.h" 99#include "opt_gateway.h" 100#include "opt_ipsec.h" 101#include "opt_mrouting.h" 102#include "opt_mbuftrace.h" 103#include "opt_inet_csum.h" 104#endif 105 106#include <sys/param.h> 107#include <sys/systm.h> 108#include <sys/cpu.h> 109#include <sys/mbuf.h> 110#include <sys/domain.h> 111#include <sys/protosw.h> 112#include <sys/socket.h> 113#include <sys/socketvar.h> 114#include <sys/errno.h> 115#include <sys/time.h> 116#include <sys/kernel.h> 117#include <sys/pool.h> 118#include <sys/sysctl.h> 119#include <sys/kauth.h> 120 121#include <net/if.h> 122#include <net/if_dl.h> 123#include <net/route.h> 124#include <net/pktqueue.h> 125#include <net/pfil.h> 126 127#include <netinet/in.h> 128#include <netinet/in_systm.h> 129#include <netinet/ip.h> 130#include <netinet/in_pcb.h> 131#include <netinet/in_proto.h> 132#include <netinet/in_var.h> 133#include <netinet/ip_var.h> 134#include <netinet/ip_private.h> 135#include <netinet/ip_icmp.h> 136/* just for gif_ttl */ 137#include <netinet/in_gif.h> 138#include "gif.h" 139#include <net/if_gre.h> 140#include "gre.h" 141 142#ifdef MROUTING 143#include <netinet/ip_mroute.h> 144#endif 145#include <netinet/portalgo.h> 146 147#ifdef IPSEC 148#include <netipsec/ipsec.h> 149#endif 150 151#ifndef IPFORWARDING 152#ifdef GATEWAY 153#define IPFORWARDING 1 /* forward IP packets not for us */ 154#else /* GATEWAY */ 155#define IPFORWARDING 0 /* don't forward IP packets not for us */ 156#endif /* GATEWAY */ 157#endif /* IPFORWARDING */ 158#ifndef IPSENDREDIRECTS 159#define IPSENDREDIRECTS 1 160#endif 161#ifndef IPFORWSRCRT 162#define IPFORWSRCRT 1 /* forward source-routed packets */ 163#endif 164#ifndef IPALLOWSRCRT 165#define IPALLOWSRCRT 1 /* allow source-routed packets */ 166#endif 167#ifndef IPMTUDISC 168#define IPMTUDISC 1 169#endif 170#ifndef IPMTUDISCTIMEOUT 171#define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 172#endif 173 174#ifdef COMPAT_50 175#include <compat/sys/time.h> 176#include <compat/sys/socket.h> 177#endif 178 179/* 180 * Note: DIRECTED_BROADCAST is handled this way so that previous 181 * configuration using this option will Just Work. 182 */ 183#ifndef IPDIRECTEDBCAST 184#ifdef DIRECTED_BROADCAST 185#define IPDIRECTEDBCAST 1 186#else 187#define IPDIRECTEDBCAST 0 188#endif /* DIRECTED_BROADCAST */ 189#endif /* IPDIRECTEDBCAST */ 190int ipforwarding = IPFORWARDING; 191int ipsendredirects = IPSENDREDIRECTS; 192int ip_defttl = IPDEFTTL; 193int ip_forwsrcrt = IPFORWSRCRT; 194int ip_directedbcast = IPDIRECTEDBCAST; 195int ip_allowsrcrt = IPALLOWSRCRT; 196int ip_mtudisc = IPMTUDISC; 197int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 198#ifdef DIAGNOSTIC 199int ipprintfs = 0; 200#endif 201 202int ip_do_randomid = 0; 203 204/* 205 * XXX - Setting ip_checkinterface mostly implements the receive side of 206 * the Strong ES model described in RFC 1122, but since the routing table 207 * and transmit implementation do not implement the Strong ES model, 208 * setting this to 1 results in an odd hybrid. 209 * 210 * XXX - ip_checkinterface currently must be disabled if you use ipnat 211 * to translate the destination address to another local interface. 212 * 213 * XXX - ip_checkinterface must be disabled if you add IP aliases 214 * to the loopback interface instead of the interface where the 215 * packets for those addresses are received. 216 */ 217static int ip_checkinterface __read_mostly = 0; 218 219struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 220 221pktqueue_t * ip_pktq __read_mostly; 222pfil_head_t * inet_pfil_hook __read_mostly; 223ipid_state_t * ip_ids __read_mostly; 224percpu_t * ipstat_percpu __read_mostly; 225 226static struct route ipforward_rt __cacheline_aligned; 227 228uint16_t ip_id; 229 230#ifdef INET_CSUM_COUNTERS 231#include <sys/device.h> 232 233struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 234 NULL, "inet", "hwcsum bad"); 235struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 236 NULL, "inet", "hwcsum ok"); 237struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 238 NULL, "inet", "swcsum"); 239 240#define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 241 242EVCNT_ATTACH_STATIC(ip_hwcsum_bad); 243EVCNT_ATTACH_STATIC(ip_hwcsum_ok); 244EVCNT_ATTACH_STATIC(ip_swcsum); 245 246#else 247 248#define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 249 250#endif /* INET_CSUM_COUNTERS */ 251 252/* 253 * We need to save the IP options in case a protocol wants to respond 254 * to an incoming packet over the same route if the packet got here 255 * using IP source routing. This allows connection establishment and 256 * maintenance when the remote end is on a network that is not known 257 * to us. 258 */ 259 260static int ip_nhops = 0; 261 262static struct ip_srcrt { 263 struct in_addr dst; /* final destination */ 264 char nop; /* one NOP to align */ 265 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 266 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 267} ip_srcrt; 268 269static int ip_drainwanted; 270 271struct sockaddr_in ipaddr = { 272 .sin_len = sizeof(ipaddr), 273 .sin_family = AF_INET, 274}; 275 276static void save_rte(u_char *, struct in_addr); 277 278#ifdef MBUFTRACE 279struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 280struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 281#endif 282 283static void ipintr(void *); 284static void ip_input(struct mbuf *); 285static void ip_forward(struct mbuf *, int); 286static bool ip_dooptions(struct mbuf *); 287static struct in_ifaddr *ip_rtaddr(struct in_addr); 288static void sysctl_net_inet_ip_setup(struct sysctllog **); 289 290/* XXX: Not yet enabled. */ 291#define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock)) 292#define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock)) 293 294/* 295 * IP initialization: fill in IP protocol switch table. 296 * All protocols not implemented in kernel go to raw IP protocol handler. 297 */ 298void 299ip_init(void) 300{ 301 const struct protosw *pr; 302 303 in_init(); 304 sysctl_net_inet_ip_setup(NULL); 305 306 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 307 KASSERT(pr != NULL); 308 309 ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL); 310 KASSERT(ip_pktq != NULL); 311 312 for (u_int i = 0; i < IPPROTO_MAX; i++) { 313 ip_protox[i] = pr - inetsw; 314 } 315 for (pr = inetdomain.dom_protosw; 316 pr < inetdomain.dom_protoswNPROTOSW; pr++) 317 if (pr->pr_domain->dom_family == PF_INET && 318 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 319 ip_protox[pr->pr_protocol] = pr - inetsw; 320 321 ip_reass_init(); 322 323 ip_ids = ip_id_init(); 324 ip_id = time_uptime & 0xfffff; 325 326 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 327#ifdef GATEWAY 328 ipflow_init(); 329#endif 330 331 /* Register our Packet Filter hook. */ 332 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 333 KASSERT(inet_pfil_hook != NULL); 334 335#ifdef MBUFTRACE 336 MOWNER_ATTACH(&ip_tx_mowner); 337 MOWNER_ATTACH(&ip_rx_mowner); 338#endif /* MBUFTRACE */ 339 340 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 341} 342 343/* 344 * IP software interrupt routine. 345 */ 346static void 347ipintr(void *arg __unused) 348{ 349 struct mbuf *m; 350 351 KASSERT(cpu_softintr_p()); 352 353 mutex_enter(softnet_lock); 354 while ((m = pktq_dequeue(ip_pktq)) != NULL) { 355 ip_input(m); 356 } 357 mutex_exit(softnet_lock); 358} 359 360/* 361 * IP input routine. Checksum and byte swap header. If fragmented 362 * try to reassemble. Process options. Pass to next level. 363 */ 364static void 365ip_input(struct mbuf *m) 366{ 367 struct ip *ip = NULL; 368 struct in_ifaddr *ia; 369 struct ifaddr *ifa; 370 int hlen = 0, len; 371 int downmatch; 372 int checkif; 373 int srcrt = 0; 374 ifnet_t *ifp; 375 376 KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software " 377 "interrupt handler; synchronization assumptions violated"); 378 379 MCLAIM(m, &ip_rx_mowner); 380 KASSERT((m->m_flags & M_PKTHDR) != 0); 381 ifp = m->m_pkthdr.rcvif; 382 383 /* 384 * If no IP addresses have been set yet but the interfaces 385 * are receiving, can't do anything with incoming packets yet. 386 * Note: we pre-check without locks held. 387 */ 388 if (!TAILQ_FIRST(&in_ifaddrhead)) { 389 goto bad; 390 } 391 IP_STATINC(IP_STAT_TOTAL); 392 393 /* 394 * If the IP header is not aligned, slurp it up into a new 395 * mbuf with space for link headers, in the event we forward 396 * it. Otherwise, if it is aligned, make sure the entire 397 * base IP header is in the first mbuf of the chain. 398 */ 399 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 400 if ((m = m_copyup(m, sizeof(struct ip), 401 (max_linkhdr + 3) & ~3)) == NULL) { 402 /* XXXJRT new stat, please */ 403 IP_STATINC(IP_STAT_TOOSMALL); 404 return; 405 } 406 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 407 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 408 IP_STATINC(IP_STAT_TOOSMALL); 409 return; 410 } 411 } 412 ip = mtod(m, struct ip *); 413 if (ip->ip_v != IPVERSION) { 414 IP_STATINC(IP_STAT_BADVERS); 415 goto bad; 416 } 417 hlen = ip->ip_hl << 2; 418 if (hlen < sizeof(struct ip)) { /* minimum header length */ 419 IP_STATINC(IP_STAT_BADHLEN); 420 goto bad; 421 } 422 if (hlen > m->m_len) { 423 if ((m = m_pullup(m, hlen)) == NULL) { 424 IP_STATINC(IP_STAT_BADHLEN); 425 return; 426 } 427 ip = mtod(m, struct ip *); 428 } 429 430 /* 431 * RFC1122: packets with a multicast source address are 432 * not allowed. 433 */ 434 if (IN_MULTICAST(ip->ip_src.s_addr)) { 435 IP_STATINC(IP_STAT_BADADDR); 436 goto bad; 437 } 438 439 /* 127/8 must not appear on wire - RFC1122 */ 440 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 441 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 442 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 443 IP_STATINC(IP_STAT_BADADDR); 444 goto bad; 445 } 446 } 447 448 switch (m->m_pkthdr.csum_flags & 449 ((ifp->if_csum_flags_rx & M_CSUM_IPv4) | 450 M_CSUM_IPv4_BAD)) { 451 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 452 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 453 goto badcsum; 454 455 case M_CSUM_IPv4: 456 /* Checksum was okay. */ 457 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 458 break; 459 460 default: 461 /* 462 * Must compute it ourselves. Maybe skip checksum on 463 * loopback interfaces. 464 */ 465 if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) || 466 ip_do_loopback_cksum)) { 467 INET_CSUM_COUNTER_INCR(&ip_swcsum); 468 if (in_cksum(m, hlen) != 0) 469 goto badcsum; 470 } 471 break; 472 } 473 474 /* Retrieve the packet length. */ 475 len = ntohs(ip->ip_len); 476 477 /* 478 * Check for additional length bogosity 479 */ 480 if (len < hlen) { 481 IP_STATINC(IP_STAT_BADLEN); 482 goto bad; 483 } 484 485 /* 486 * Check that the amount of data in the buffers 487 * is as at least much as the IP header would have us expect. 488 * Trim mbufs if longer than we expect. 489 * Drop packet if shorter than we expect. 490 */ 491 if (m->m_pkthdr.len < len) { 492 IP_STATINC(IP_STAT_TOOSHORT); 493 goto bad; 494 } 495 if (m->m_pkthdr.len > len) { 496 if (m->m_len == m->m_pkthdr.len) { 497 m->m_len = len; 498 m->m_pkthdr.len = len; 499 } else 500 m_adj(m, len - m->m_pkthdr.len); 501 } 502 503 /* 504 * Assume that we can create a fast-forward IP flow entry 505 * based on this packet. 506 */ 507 m->m_flags |= M_CANFASTFWD; 508 509 /* 510 * Run through list of hooks for input packets. If there are any 511 * filters which require that additional packets in the flow are 512 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 513 * Note that filters must _never_ set this flag, as another filter 514 * in the list may have previously cleared it. 515 */ 516#if defined(IPSEC) 517 if (!ipsec_used || !ipsec_indone(m)) 518#else 519 if (1) 520#endif 521 { 522 struct in_addr odst = ip->ip_dst; 523 bool freed; 524 525 SOFTNET_LOCK(); 526 freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0; 527 SOFTNET_UNLOCK(); 528 if (freed || m == NULL) { 529 return; 530 } 531 ip = mtod(m, struct ip *); 532 hlen = ip->ip_hl << 2; 533 534 /* 535 * XXX The setting of "srcrt" here is to prevent ip_forward() 536 * from generating ICMP redirects for packets that have 537 * been redirected by a hook back out on to the same LAN that 538 * they came from and is not an indication that the packet 539 * is being inffluenced by source routing options. This 540 * allows things like 541 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 542 * where tlp0 is both on the 1.1.1.0/24 network and is the 543 * default route for hosts on 1.1.1.0/24. Of course this 544 * also requires a "map tlp0 ..." to complete the story. 545 * One might argue whether or not this kind of network config. 546 * should be supported in this manner... 547 */ 548 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 549 } 550 551#ifdef ALTQ 552 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 553 if (altq_input) { 554 SOFTNET_LOCK(); 555 if ((*altq_input)(m, AF_INET) == 0) { 556 /* Packet dropped by traffic conditioner. */ 557 SOFTNET_UNLOCK(); 558 return; 559 } 560 SOFTNET_UNLOCK(); 561 } 562#endif 563 564 /* 565 * Process options and, if not destined for us, 566 * ship it on. ip_dooptions returns 1 when an 567 * error was detected (causing an icmp message 568 * to be sent and the original packet to be freed). 569 */ 570 ip_nhops = 0; /* for source routed packets */ 571 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 572 return; 573 574 /* 575 * Enable a consistency check between the destination address 576 * and the arrival interface for a unicast packet (the RFC 1122 577 * strong ES model) if IP forwarding is disabled and the packet 578 * is not locally generated. 579 * 580 * XXX - Checking also should be disabled if the destination 581 * address is ipnat'ed to a different interface. 582 * 583 * XXX - Checking is incompatible with IP aliases added 584 * to the loopback interface instead of the interface where 585 * the packets are received. 586 * 587 * XXX - We need to add a per ifaddr flag for this so that 588 * we get finer grain control. 589 */ 590 checkif = ip_checkinterface && (ipforwarding == 0) && 591 (ifp->if_flags & IFF_LOOPBACK) == 0; 592 593 /* 594 * Check our list of addresses, to see if the packet is for us. 595 * 596 * Traditional 4.4BSD did not consult IFF_UP at all. 597 * The behavior here is to treat addresses on !IFF_UP interface 598 * or IN_IFF_NOTREADY addresses as not mine. 599 */ 600 downmatch = 0; 601 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 602 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 603 if (ia->ia4_flags & IN_IFF_NOTREADY) 604 continue; 605 if (checkif && ia->ia_ifp != ifp) 606 continue; 607 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 608 break; 609 else 610 downmatch++; 611 } 612 } 613 if (ia != NULL) 614 goto ours; 615 if (ifp->if_flags & IFF_BROADCAST) { 616 IFADDR_FOREACH(ifa, ifp) { 617 if (ifa->ifa_addr->sa_family != AF_INET) 618 continue; 619 ia = ifatoia(ifa); 620 if (ia->ia4_flags & IN_IFF_NOTREADY) 621 continue; 622 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 623 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 624 /* 625 * Look for all-0's host part (old broadcast addr), 626 * either for subnet or net. 627 */ 628 ip->ip_dst.s_addr == ia->ia_subnet || 629 ip->ip_dst.s_addr == ia->ia_net) 630 goto ours; 631 /* 632 * An interface with IP address zero accepts 633 * all packets that arrive on that interface. 634 */ 635 if (in_nullhost(ia->ia_addr.sin_addr)) 636 goto ours; 637 } 638 } 639 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 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 SOFTNET_LOCK(); 657 if (ip_mforward(m, ifp) != 0) { 658 SOFTNET_UNLOCK(); 659 IP_STATINC(IP_STAT_CANTFORWARD); 660 m_freem(m); 661 return; 662 } 663 SOFTNET_UNLOCK(); 664 665 /* 666 * The process-level routing demon needs to receive 667 * all multicast IGMP packets, whether or not this 668 * host belongs to their destination groups. 669 */ 670 if (ip->ip_p == IPPROTO_IGMP) { 671 goto ours; 672 } 673 IP_STATINC(IP_STAT_CANTFORWARD); 674 } 675#endif 676 /* 677 * See if we belong to the destination multicast group on the 678 * arrival interface. 679 */ 680 if (!in_multi_group(ip->ip_dst, ifp, 0)) { 681 IP_STATINC(IP_STAT_CANTFORWARD); 682 m_freem(m); 683 return; 684 } 685 goto ours; 686 } 687 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 688 in_nullhost(ip->ip_dst)) 689 goto ours; 690 691 /* 692 * Not for us; forward if possible and desirable. 693 */ 694 if (ipforwarding == 0) { 695 IP_STATINC(IP_STAT_CANTFORWARD); 696 m_freem(m); 697 } else { 698 /* 699 * If ip_dst matched any of my address on !IFF_UP interface, 700 * and there's no IFF_UP interface that matches ip_dst, 701 * send icmp unreach. Forwarding it will result in in-kernel 702 * forwarding loop till TTL goes to 0. 703 */ 704 if (downmatch) { 705 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 706 IP_STATINC(IP_STAT_CANTFORWARD); 707 return; 708 } 709#ifdef IPSEC 710 /* Perform IPsec, if any. */ 711 if (ipsec_used) { 712 SOFTNET_LOCK(); 713 if (ipsec4_input(m, IP_FORWARDING | 714 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) { 715 SOFTNET_UNLOCK(); 716 goto bad; 717 } 718 SOFTNET_UNLOCK(); 719 } 720#endif 721 ip_forward(m, srcrt); 722 } 723 return; 724 725ours: 726 /* 727 * If offset or IP_MF are set, must reassemble. 728 */ 729 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 730 /* 731 * Pass to IP reassembly mechanism. 732 */ 733 if (ip_reass_packet(&m, ip) != 0) { 734 /* Failed; invalid fragment(s) or packet. */ 735 goto bad; 736 } 737 if (m == NULL) { 738 /* More fragments should come; silently return. */ 739 return; 740 } 741 /* 742 * Reassembly is done, we have the final packet. 743 * Updated cached data in local variable(s). 744 */ 745 ip = mtod(m, struct ip *); 746 hlen = ip->ip_hl << 2; 747 } 748 749#ifdef IPSEC 750 /* 751 * Enforce IPsec policy checking if we are seeing last header. 752 * Note that we do not visit this with protocols with PCB layer 753 * code - like UDP/TCP/raw IP. 754 */ 755 if (ipsec_used && 756 (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 757 SOFTNET_LOCK(); 758 if (ipsec4_input(m, 0) != 0) { 759 SOFTNET_UNLOCK(); 760 goto bad; 761 } 762 SOFTNET_UNLOCK(); 763 } 764#endif 765 766 /* 767 * Switch out to protocol's input routine. 768 */ 769#if IFA_STATS 770 if (ia && ip) 771 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 772#endif 773 IP_STATINC(IP_STAT_DELIVERED); 774 775 const int off = hlen, nh = ip->ip_p; 776 777 SOFTNET_LOCK(); 778 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 779 SOFTNET_UNLOCK(); 780 return; 781bad: 782 m_freem(m); 783 return; 784 785badcsum: 786 IP_STATINC(IP_STAT_BADSUM); 787 m_freem(m); 788} 789 790/* 791 * IP timer processing. 792 */ 793void 794ip_slowtimo(void) 795{ 796 797 mutex_enter(softnet_lock); 798 KERNEL_LOCK(1, NULL); 799 800 ip_reass_slowtimo(); 801 802 KERNEL_UNLOCK_ONE(NULL); 803 mutex_exit(softnet_lock); 804} 805 806/* 807 * IP drain processing. 808 */ 809void 810ip_drain(void) 811{ 812 813 KERNEL_LOCK(1, NULL); 814 ip_reass_drain(); 815 KERNEL_UNLOCK_ONE(NULL); 816} 817 818/* 819 * ip_dooptions: perform option processing on a datagram, possibly discarding 820 * it if bad options are encountered, or forwarding it if source-routed. 821 * 822 * => Returns true if packet has been forwarded/freed. 823 * => Returns false if the packet should be processed further. 824 */ 825static bool 826ip_dooptions(struct mbuf *m) 827{ 828 struct ip *ip = mtod(m, struct ip *); 829 u_char *cp, *cp0; 830 struct ip_timestamp *ipt; 831 struct in_ifaddr *ia; 832 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 833 struct in_addr dst; 834 n_time ntime; 835 836 dst = ip->ip_dst; 837 cp = (u_char *)(ip + 1); 838 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 839 for (; cnt > 0; cnt -= optlen, cp += optlen) { 840 opt = cp[IPOPT_OPTVAL]; 841 if (opt == IPOPT_EOL) 842 break; 843 if (opt == IPOPT_NOP) 844 optlen = 1; 845 else { 846 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 847 code = &cp[IPOPT_OLEN] - (u_char *)ip; 848 goto bad; 849 } 850 optlen = cp[IPOPT_OLEN]; 851 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 852 code = &cp[IPOPT_OLEN] - (u_char *)ip; 853 goto bad; 854 } 855 } 856 switch (opt) { 857 858 default: 859 break; 860 861 /* 862 * Source routing with record. 863 * Find interface with current destination address. 864 * If none on this machine then drop if strictly routed, 865 * or do nothing if loosely routed. 866 * Record interface address and bring up next address 867 * component. If strictly routed make sure next 868 * address is on directly accessible net. 869 */ 870 case IPOPT_LSRR: 871 case IPOPT_SSRR: 872 if (ip_allowsrcrt == 0) { 873 type = ICMP_UNREACH; 874 code = ICMP_UNREACH_NET_PROHIB; 875 goto bad; 876 } 877 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 878 code = &cp[IPOPT_OLEN] - (u_char *)ip; 879 goto bad; 880 } 881 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 882 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 883 goto bad; 884 } 885 ipaddr.sin_addr = ip->ip_dst; 886 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 887 if (ia == 0) { 888 if (opt == IPOPT_SSRR) { 889 type = ICMP_UNREACH; 890 code = ICMP_UNREACH_SRCFAIL; 891 goto bad; 892 } 893 /* 894 * Loose routing, and not at next destination 895 * yet; nothing to do except forward. 896 */ 897 break; 898 } 899 off--; /* 0 origin */ 900 if ((off + sizeof(struct in_addr)) > optlen) { 901 /* 902 * End of source route. Should be for us. 903 */ 904 save_rte(cp, ip->ip_src); 905 break; 906 } 907 /* 908 * locate outgoing interface 909 */ 910 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 911 sizeof(ipaddr.sin_addr)); 912 if (opt == IPOPT_SSRR) 913 ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); 914 else 915 ia = ip_rtaddr(ipaddr.sin_addr); 916 if (ia == 0) { 917 type = ICMP_UNREACH; 918 code = ICMP_UNREACH_SRCFAIL; 919 goto bad; 920 } 921 ip->ip_dst = ipaddr.sin_addr; 922 bcopy((void *)&ia->ia_addr.sin_addr, 923 (void *)(cp + off), sizeof(struct in_addr)); 924 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 925 /* 926 * Let ip_intr's mcast routing check handle mcast pkts 927 */ 928 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 929 break; 930 931 case IPOPT_RR: 932 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 933 code = &cp[IPOPT_OLEN] - (u_char *)ip; 934 goto bad; 935 } 936 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 937 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 938 goto bad; 939 } 940 /* 941 * If no space remains, ignore. 942 */ 943 off--; /* 0 origin */ 944 if ((off + sizeof(struct in_addr)) > optlen) 945 break; 946 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 947 sizeof(ipaddr.sin_addr)); 948 /* 949 * locate outgoing interface; if we're the destination, 950 * use the incoming interface (should be same). 951 */ 952 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 953 == NULL && 954 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 955 type = ICMP_UNREACH; 956 code = ICMP_UNREACH_HOST; 957 goto bad; 958 } 959 bcopy((void *)&ia->ia_addr.sin_addr, 960 (void *)(cp + off), sizeof(struct in_addr)); 961 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 962 break; 963 964 case IPOPT_TS: 965 code = cp - (u_char *)ip; 966 ipt = (struct ip_timestamp *)cp; 967 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 968 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 969 goto bad; 970 } 971 if (ipt->ipt_ptr < 5) { 972 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 973 goto bad; 974 } 975 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 976 if (++ipt->ipt_oflw == 0) { 977 code = (u_char *)&ipt->ipt_ptr - 978 (u_char *)ip; 979 goto bad; 980 } 981 break; 982 } 983 cp0 = (cp + ipt->ipt_ptr - 1); 984 switch (ipt->ipt_flg) { 985 986 case IPOPT_TS_TSONLY: 987 break; 988 989 case IPOPT_TS_TSANDADDR: 990 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 991 sizeof(struct in_addr) > ipt->ipt_len) { 992 code = (u_char *)&ipt->ipt_ptr - 993 (u_char *)ip; 994 goto bad; 995 } 996 ipaddr.sin_addr = dst; 997 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 998 m->m_pkthdr.rcvif)); 999 if (ia == 0) 1000 continue; 1001 bcopy(&ia->ia_addr.sin_addr, 1002 cp0, sizeof(struct in_addr)); 1003 ipt->ipt_ptr += sizeof(struct in_addr); 1004 break; 1005 1006 case IPOPT_TS_PRESPEC: 1007 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1008 sizeof(struct in_addr) > ipt->ipt_len) { 1009 code = (u_char *)&ipt->ipt_ptr - 1010 (u_char *)ip; 1011 goto bad; 1012 } 1013 memcpy(&ipaddr.sin_addr, cp0, 1014 sizeof(struct in_addr)); 1015 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1016 == NULL) 1017 continue; 1018 ipt->ipt_ptr += sizeof(struct in_addr); 1019 break; 1020 1021 default: 1022 /* XXX can't take &ipt->ipt_flg */ 1023 code = (u_char *)&ipt->ipt_ptr - 1024 (u_char *)ip + 1; 1025 goto bad; 1026 } 1027 ntime = iptime(); 1028 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1029 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1030 sizeof(n_time)); 1031 ipt->ipt_ptr += sizeof(n_time); 1032 } 1033 } 1034 if (forward) { 1035 if (ip_forwsrcrt == 0) { 1036 type = ICMP_UNREACH; 1037 code = ICMP_UNREACH_SRCFAIL; 1038 goto bad; 1039 } 1040 ip_forward(m, 1); 1041 return true; 1042 } 1043 return false; 1044bad: 1045 icmp_error(m, type, code, 0, 0); 1046 IP_STATINC(IP_STAT_BADOPTIONS); 1047 return true; 1048} 1049 1050/* 1051 * ip_rtaddr: given address of next destination (final or next hop), 1052 * return internet address info of interface to be used to get there. 1053 */ 1054static struct in_ifaddr * 1055ip_rtaddr(struct in_addr dst) 1056{ 1057 struct rtentry *rt; 1058 union { 1059 struct sockaddr dst; 1060 struct sockaddr_in dst4; 1061 } u; 1062 1063 sockaddr_in_init(&u.dst4, &dst, 0); 1064 1065 SOFTNET_LOCK(); 1066 rt = rtcache_lookup(&ipforward_rt, &u.dst); 1067 SOFTNET_UNLOCK(); 1068 if (rt == NULL) 1069 return NULL; 1070 1071 return ifatoia(rt->rt_ifa); 1072} 1073 1074/* 1075 * save_rte: save incoming source route for use in replies, to be picked 1076 * up later by ip_srcroute if the receiver is interested. 1077 */ 1078static void 1079save_rte(u_char *option, struct in_addr dst) 1080{ 1081 unsigned olen; 1082 1083 olen = option[IPOPT_OLEN]; 1084 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1085 return; 1086 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1087 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1088 ip_srcrt.dst = dst; 1089} 1090 1091/* 1092 * Retrieve incoming source route for use in replies, 1093 * in the same form used by setsockopt. 1094 * The first hop is placed before the options, will be removed later. 1095 */ 1096struct mbuf * 1097ip_srcroute(void) 1098{ 1099 struct in_addr *p, *q; 1100 struct mbuf *m; 1101 1102 if (ip_nhops == 0) 1103 return NULL; 1104 m = m_get(M_DONTWAIT, MT_SOOPTS); 1105 if (m == 0) 1106 return NULL; 1107 1108 MCLAIM(m, &inetdomain.dom_mowner); 1109#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1110 1111 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1112 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1113 OPTSIZ; 1114 1115 /* 1116 * First save first hop for return route 1117 */ 1118 p = &ip_srcrt.route[ip_nhops - 1]; 1119 *(mtod(m, struct in_addr *)) = *p--; 1120 1121 /* 1122 * Copy option fields and padding (nop) to mbuf. 1123 */ 1124 ip_srcrt.nop = IPOPT_NOP; 1125 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1126 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1127 OPTSIZ); 1128 q = (struct in_addr *)(mtod(m, char *) + 1129 sizeof(struct in_addr) + OPTSIZ); 1130#undef OPTSIZ 1131 /* 1132 * Record return path as an IP source route, 1133 * reversing the path (pointers are now aligned). 1134 */ 1135 while (p >= ip_srcrt.route) { 1136 *q++ = *p--; 1137 } 1138 /* 1139 * Last hop goes to final destination. 1140 */ 1141 *q = ip_srcrt.dst; 1142 return (m); 1143} 1144 1145const int inetctlerrmap[PRC_NCMDS] = { 1146 [PRC_MSGSIZE] = EMSGSIZE, 1147 [PRC_HOSTDEAD] = EHOSTDOWN, 1148 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1149 [PRC_UNREACH_NET] = EHOSTUNREACH, 1150 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1151 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1152 [PRC_UNREACH_PORT] = ECONNREFUSED, 1153 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1154 [PRC_PARAMPROB] = ENOPROTOOPT, 1155}; 1156 1157void 1158ip_fasttimo(void) 1159{ 1160 if (ip_drainwanted) { 1161 ip_drain(); 1162 ip_drainwanted = 0; 1163 } 1164} 1165 1166void 1167ip_drainstub(void) 1168{ 1169 ip_drainwanted = 1; 1170} 1171 1172/* 1173 * Forward a packet. If some error occurs return the sender 1174 * an icmp packet. Note we can't always generate a meaningful 1175 * icmp message because icmp doesn't have a large enough repertoire 1176 * of codes and types. 1177 * 1178 * If not forwarding, just drop the packet. This could be confusing 1179 * if ipforwarding was zero but some routing protocol was advancing 1180 * us as a gateway to somewhere. However, we must let the routing 1181 * protocol deal with that. 1182 * 1183 * The srcrt parameter indicates whether the packet is being forwarded 1184 * via a source route. 1185 */ 1186static void 1187ip_forward(struct mbuf *m, int srcrt) 1188{ 1189 struct ip *ip = mtod(m, struct ip *); 1190 struct rtentry *rt; 1191 int error, type = 0, code = 0, destmtu = 0; 1192 struct mbuf *mcopy; 1193 n_long dest; 1194 union { 1195 struct sockaddr dst; 1196 struct sockaddr_in dst4; 1197 } u; 1198 uint64_t *ips; 1199 1200 KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software " 1201 "interrupt handler; synchronization assumptions violated"); 1202 1203 /* 1204 * We are now in the output path. 1205 */ 1206 MCLAIM(m, &ip_tx_mowner); 1207 1208 /* 1209 * Clear any in-bound checksum flags for this packet. 1210 */ 1211 m->m_pkthdr.csum_flags = 0; 1212 1213 dest = 0; 1214 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1215 IP_STATINC(IP_STAT_CANTFORWARD); 1216 m_freem(m); 1217 return; 1218 } 1219 1220 SOFTNET_LOCK(); 1221 1222 if (ip->ip_ttl <= IPTTLDEC) { 1223 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1224 SOFTNET_UNLOCK(); 1225 return; 1226 } 1227 1228 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1229 1230 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1231 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1232 SOFTNET_UNLOCK(); 1233 return; 1234 } 1235 1236 /* 1237 * Save at most 68 bytes of the packet in case 1238 * we need to generate an ICMP message to the src. 1239 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1240 */ 1241 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1242 if (mcopy) 1243 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1244 1245 ip->ip_ttl -= IPTTLDEC; 1246 1247 /* 1248 * If forwarding packet using same interface that it came in on, 1249 * perhaps should send a redirect to sender to shortcut a hop. 1250 * Only send redirect if source is sending directly to us, 1251 * and if packet was not source routed (or has any options). 1252 * Also, don't send redirect if forwarding using a default route 1253 * or a route modified by a redirect. 1254 */ 1255 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1256 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1257 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1258 ipsendredirects && !srcrt) { 1259 if (rt->rt_ifa && 1260 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1261 ifatoia(rt->rt_ifa)->ia_subnet) { 1262 if (rt->rt_flags & RTF_GATEWAY) 1263 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1264 else 1265 dest = ip->ip_dst.s_addr; 1266 /* 1267 * Router requirements says to only send host 1268 * redirects. 1269 */ 1270 type = ICMP_REDIRECT; 1271 code = ICMP_REDIRECT_HOST; 1272 } 1273 } 1274 1275 error = ip_output(m, NULL, &ipforward_rt, 1276 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1277 NULL, NULL); 1278 1279 if (error) { 1280 IP_STATINC(IP_STAT_CANTFORWARD); 1281 goto error; 1282 } 1283 1284 ips = IP_STAT_GETREF(); 1285 ips[IP_STAT_FORWARD]++; 1286 1287 if (type) { 1288 ips[IP_STAT_REDIRECTSENT]++; 1289 IP_STAT_PUTREF(); 1290 goto redirect; 1291 } 1292 1293 IP_STAT_PUTREF(); 1294 if (mcopy) { 1295#ifdef GATEWAY 1296 if (mcopy->m_flags & M_CANFASTFWD) 1297 ipflow_create(&ipforward_rt, mcopy); 1298#endif 1299 m_freem(mcopy); 1300 } 1301 1302 SOFTNET_UNLOCK(); 1303 return; 1304 1305redirect: 1306error: 1307 if (mcopy == NULL) { 1308 SOFTNET_UNLOCK(); 1309 return; 1310 } 1311 1312 switch (error) { 1313 1314 case 0: /* forwarded, but need redirect */ 1315 /* type, code set above */ 1316 break; 1317 1318 case ENETUNREACH: /* shouldn't happen, checked above */ 1319 case EHOSTUNREACH: 1320 case ENETDOWN: 1321 case EHOSTDOWN: 1322 default: 1323 type = ICMP_UNREACH; 1324 code = ICMP_UNREACH_HOST; 1325 break; 1326 1327 case EMSGSIZE: 1328 type = ICMP_UNREACH; 1329 code = ICMP_UNREACH_NEEDFRAG; 1330 1331 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1332 destmtu = rt->rt_ifp->if_mtu; 1333#ifdef IPSEC 1334 if (ipsec_used) 1335 (void)ipsec4_forward(mcopy, &destmtu); 1336#endif 1337 IP_STATINC(IP_STAT_CANTFRAG); 1338 break; 1339 1340 case ENOBUFS: 1341 /* 1342 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1343 * Requirements for IP Version 4 Routers. Source quench can 1344 * big problem under DoS attacks or if the underlying 1345 * interface is rate-limited. 1346 */ 1347 if (mcopy) 1348 m_freem(mcopy); 1349 SOFTNET_UNLOCK(); 1350 return; 1351 } 1352 icmp_error(mcopy, type, code, dest, destmtu); 1353 SOFTNET_UNLOCK(); 1354} 1355 1356void 1357ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1358 struct mbuf *m) 1359{ 1360 struct socket *so = inp->inp_socket; 1361 ifnet_t *ifp = m->m_pkthdr.rcvif; 1362 int inpflags = inp->inp_flags; 1363 1364 if (so->so_options & SO_TIMESTAMP 1365#ifdef SO_OTIMESTAMP 1366 || so->so_options & SO_OTIMESTAMP 1367#endif 1368 ) { 1369 struct timeval tv; 1370 1371 microtime(&tv); 1372#ifdef SO_OTIMESTAMP 1373 if (so->so_options & SO_OTIMESTAMP) { 1374 struct timeval50 tv50; 1375 timeval_to_timeval50(&tv, &tv50); 1376 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1377 SCM_OTIMESTAMP, SOL_SOCKET); 1378 } else 1379#endif 1380 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1381 SCM_TIMESTAMP, SOL_SOCKET); 1382 if (*mp) 1383 mp = &(*mp)->m_next; 1384 } 1385 if (inpflags & INP_RECVDSTADDR) { 1386 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1387 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1388 if (*mp) 1389 mp = &(*mp)->m_next; 1390 } 1391 if (inpflags & INP_RECVPKTINFO) { 1392 struct in_pktinfo ipi; 1393 ipi.ipi_addr = ip->ip_src; 1394 ipi.ipi_ifindex = ifp->if_index; 1395 *mp = sbcreatecontrol((void *) &ipi, 1396 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1397 if (*mp) 1398 mp = &(*mp)->m_next; 1399 } 1400 if (inpflags & INP_PKTINFO) { 1401 struct in_pktinfo ipi; 1402 ipi.ipi_addr = ip->ip_dst; 1403 ipi.ipi_ifindex = ifp->if_index; 1404 *mp = sbcreatecontrol((void *) &ipi, 1405 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1406 if (*mp) 1407 mp = &(*mp)->m_next; 1408 } 1409 if (inpflags & INP_RECVIF) { 1410 struct sockaddr_dl sdl; 1411 1412 sockaddr_dl_init(&sdl, sizeof(sdl), ifp ? 1413 ifp->if_index : 0, 0, NULL, 0, NULL, 0); 1414 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1415 if (*mp) 1416 mp = &(*mp)->m_next; 1417 } 1418 if (inpflags & INP_RECVTTL) { 1419 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1420 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1421 if (*mp) 1422 mp = &(*mp)->m_next; 1423 } 1424} 1425 1426/* 1427 * sysctl helper routine for net.inet.ip.forwsrcrt. 1428 */ 1429static int 1430sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1431{ 1432 int error, tmp; 1433 struct sysctlnode node; 1434 1435 node = *rnode; 1436 tmp = ip_forwsrcrt; 1437 node.sysctl_data = &tmp; 1438 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1439 if (error || newp == NULL) 1440 return (error); 1441 1442 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1443 0, NULL, NULL, NULL); 1444 if (error) 1445 return (error); 1446 1447 ip_forwsrcrt = tmp; 1448 1449 return (0); 1450} 1451 1452/* 1453 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1454 * range of the new value and tweaks timers if it changes. 1455 */ 1456static int 1457sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1458{ 1459 int error, tmp; 1460 struct sysctlnode node; 1461 1462 node = *rnode; 1463 tmp = ip_mtudisc_timeout; 1464 node.sysctl_data = &tmp; 1465 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1466 if (error || newp == NULL) 1467 return (error); 1468 if (tmp < 0) 1469 return (EINVAL); 1470 1471 mutex_enter(softnet_lock); 1472 1473 ip_mtudisc_timeout = tmp; 1474 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1475 1476 mutex_exit(softnet_lock); 1477 1478 return (0); 1479} 1480 1481static int 1482sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1483{ 1484 1485 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1486} 1487 1488static void 1489sysctl_net_inet_ip_setup(struct sysctllog **clog) 1490{ 1491 sysctl_createv(clog, 0, NULL, NULL, 1492 CTLFLAG_PERMANENT, 1493 CTLTYPE_NODE, "inet", 1494 SYSCTL_DESCR("PF_INET related settings"), 1495 NULL, 0, NULL, 0, 1496 CTL_NET, PF_INET, CTL_EOL); 1497 sysctl_createv(clog, 0, NULL, NULL, 1498 CTLFLAG_PERMANENT, 1499 CTLTYPE_NODE, "ip", 1500 SYSCTL_DESCR("IPv4 related settings"), 1501 NULL, 0, NULL, 0, 1502 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1503 1504 sysctl_createv(clog, 0, NULL, NULL, 1505 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1506 CTLTYPE_INT, "forwarding", 1507 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1508 NULL, 0, &ipforwarding, 0, 1509 CTL_NET, PF_INET, IPPROTO_IP, 1510 IPCTL_FORWARDING, CTL_EOL); 1511 sysctl_createv(clog, 0, NULL, NULL, 1512 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1513 CTLTYPE_INT, "redirect", 1514 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1515 NULL, 0, &ipsendredirects, 0, 1516 CTL_NET, PF_INET, IPPROTO_IP, 1517 IPCTL_SENDREDIRECTS, CTL_EOL); 1518 sysctl_createv(clog, 0, NULL, NULL, 1519 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1520 CTLTYPE_INT, "ttl", 1521 SYSCTL_DESCR("Default TTL for an INET datagram"), 1522 NULL, 0, &ip_defttl, 0, 1523 CTL_NET, PF_INET, IPPROTO_IP, 1524 IPCTL_DEFTTL, CTL_EOL); 1525#ifdef IPCTL_DEFMTU 1526 sysctl_createv(clog, 0, NULL, NULL, 1527 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1528 CTLTYPE_INT, "mtu", 1529 SYSCTL_DESCR("Default MTA for an INET route"), 1530 NULL, 0, &ip_mtu, 0, 1531 CTL_NET, PF_INET, IPPROTO_IP, 1532 IPCTL_DEFMTU, CTL_EOL); 1533#endif /* IPCTL_DEFMTU */ 1534 sysctl_createv(clog, 0, NULL, NULL, 1535 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1536 CTLTYPE_INT, "forwsrcrt", 1537 SYSCTL_DESCR("Enable forwarding of source-routed " 1538 "datagrams"), 1539 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1540 CTL_NET, PF_INET, IPPROTO_IP, 1541 IPCTL_FORWSRCRT, CTL_EOL); 1542 sysctl_createv(clog, 0, NULL, NULL, 1543 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1544 CTLTYPE_INT, "directed-broadcast", 1545 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1546 NULL, 0, &ip_directedbcast, 0, 1547 CTL_NET, PF_INET, IPPROTO_IP, 1548 IPCTL_DIRECTEDBCAST, CTL_EOL); 1549 sysctl_createv(clog, 0, NULL, NULL, 1550 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1551 CTLTYPE_INT, "allowsrcrt", 1552 SYSCTL_DESCR("Accept source-routed datagrams"), 1553 NULL, 0, &ip_allowsrcrt, 0, 1554 CTL_NET, PF_INET, IPPROTO_IP, 1555 IPCTL_ALLOWSRCRT, CTL_EOL); 1556 1557 sysctl_createv(clog, 0, NULL, NULL, 1558 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1559 CTLTYPE_INT, "mtudisc", 1560 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1561 NULL, 0, &ip_mtudisc, 0, 1562 CTL_NET, PF_INET, IPPROTO_IP, 1563 IPCTL_MTUDISC, CTL_EOL); 1564 sysctl_createv(clog, 0, NULL, NULL, 1565 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1566 CTLTYPE_INT, "anonportmin", 1567 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1568 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1569 CTL_NET, PF_INET, IPPROTO_IP, 1570 IPCTL_ANONPORTMIN, CTL_EOL); 1571 sysctl_createv(clog, 0, NULL, NULL, 1572 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1573 CTLTYPE_INT, "anonportmax", 1574 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1575 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1576 CTL_NET, PF_INET, IPPROTO_IP, 1577 IPCTL_ANONPORTMAX, CTL_EOL); 1578 sysctl_createv(clog, 0, NULL, NULL, 1579 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1580 CTLTYPE_INT, "mtudisctimeout", 1581 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1582 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1583 CTL_NET, PF_INET, IPPROTO_IP, 1584 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1585#if NGIF > 0 1586 sysctl_createv(clog, 0, NULL, NULL, 1587 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1588 CTLTYPE_INT, "gifttl", 1589 SYSCTL_DESCR("Default TTL for a gif tunnel datagram"), 1590 NULL, 0, &ip_gif_ttl, 0, 1591 CTL_NET, PF_INET, IPPROTO_IP, 1592 IPCTL_GIF_TTL, CTL_EOL); 1593#endif /* NGIF */ 1594#ifndef IPNOPRIVPORTS 1595 sysctl_createv(clog, 0, NULL, NULL, 1596 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1597 CTLTYPE_INT, "lowportmin", 1598 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1599 "to assign"), 1600 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1601 CTL_NET, PF_INET, IPPROTO_IP, 1602 IPCTL_LOWPORTMIN, CTL_EOL); 1603 sysctl_createv(clog, 0, NULL, NULL, 1604 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1605 CTLTYPE_INT, "lowportmax", 1606 SYSCTL_DESCR("Highest privileged ephemeral port number " 1607 "to assign"), 1608 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1609 CTL_NET, PF_INET, IPPROTO_IP, 1610 IPCTL_LOWPORTMAX, CTL_EOL); 1611#endif /* IPNOPRIVPORTS */ 1612#if NGRE > 0 1613 sysctl_createv(clog, 0, NULL, NULL, 1614 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1615 CTLTYPE_INT, "grettl", 1616 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1617 NULL, 0, &ip_gre_ttl, 0, 1618 CTL_NET, PF_INET, IPPROTO_IP, 1619 IPCTL_GRE_TTL, CTL_EOL); 1620#endif /* NGRE */ 1621 sysctl_createv(clog, 0, NULL, NULL, 1622 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1623 CTLTYPE_INT, "checkinterface", 1624 SYSCTL_DESCR("Enable receive side of Strong ES model " 1625 "from RFC1122"), 1626 NULL, 0, &ip_checkinterface, 0, 1627 CTL_NET, PF_INET, IPPROTO_IP, 1628 IPCTL_CHECKINTERFACE, CTL_EOL); 1629 sysctl_createv(clog, 0, NULL, NULL, 1630 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1631 CTLTYPE_INT, "random_id", 1632 SYSCTL_DESCR("Assign random ip_id values"), 1633 NULL, 0, &ip_do_randomid, 0, 1634 CTL_NET, PF_INET, IPPROTO_IP, 1635 IPCTL_RANDOMID, CTL_EOL); 1636 sysctl_createv(clog, 0, NULL, NULL, 1637 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1638 CTLTYPE_INT, "do_loopback_cksum", 1639 SYSCTL_DESCR("Perform IP checksum on loopback"), 1640 NULL, 0, &ip_do_loopback_cksum, 0, 1641 CTL_NET, PF_INET, IPPROTO_IP, 1642 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1643 sysctl_createv(clog, 0, NULL, NULL, 1644 CTLFLAG_PERMANENT, 1645 CTLTYPE_STRUCT, "stats", 1646 SYSCTL_DESCR("IP statistics"), 1647 sysctl_net_inet_ip_stats, 0, NULL, 0, 1648 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1649 CTL_EOL); 1650#if NARP 1651 sysctl_createv(clog, 0, NULL, NULL, 1652 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1653 CTLTYPE_INT, "dad_count", 1654 SYSCTL_DESCR("Number of Duplicate Address Detection " 1655 "probes to send"), 1656 NULL, 0, &ip_dad_count, 0, 1657 CTL_NET, PF_INET, IPPROTO_IP, 1658 IPCTL_DAD_COUNT, CTL_EOL); 1659#endif 1660 1661 /* anonportalgo RFC6056 subtree */ 1662 const struct sysctlnode *portalgo_node; 1663 sysctl_createv(clog, 0, NULL, &portalgo_node, 1664 CTLFLAG_PERMANENT, 1665 CTLTYPE_NODE, "anonportalgo", 1666 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1667 NULL, 0, NULL, 0, 1668 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1669 sysctl_createv(clog, 0, &portalgo_node, NULL, 1670 CTLFLAG_PERMANENT, 1671 CTLTYPE_STRING, "available", 1672 SYSCTL_DESCR("available algorithms"), 1673 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1674 CTL_CREATE, CTL_EOL); 1675 sysctl_createv(clog, 0, &portalgo_node, NULL, 1676 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1677 CTLTYPE_STRING, "selected", 1678 SYSCTL_DESCR("selected algorithm"), 1679 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1680 CTL_CREATE, CTL_EOL); 1681 sysctl_createv(clog, 0, &portalgo_node, NULL, 1682 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1683 CTLTYPE_STRUCT, "reserve", 1684 SYSCTL_DESCR("bitmap of reserved ports"), 1685 sysctl_portalgo_reserve4, 0, NULL, 0, 1686 CTL_CREATE, CTL_EOL); 1687} 1688 1689void 1690ip_statinc(u_int stat) 1691{ 1692 1693 KASSERT(stat < IP_NSTATS); 1694 IP_STATINC(stat); 1695} 1696