ip_input.c revision 1.323
1/* $NetBSD: ip_input.c,v 1.323 2015/08/07 08:11:33 ozaki-r 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 1.323 2015/08/07 08:11:33 ozaki-r 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/cpu.h> 107#include <sys/mbuf.h> 108#include <sys/domain.h> 109#include <sys/protosw.h> 110#include <sys/socket.h> 111#include <sys/socketvar.h> 112#include <sys/errno.h> 113#include <sys/time.h> 114#include <sys/kernel.h> 115#include <sys/pool.h> 116#include <sys/sysctl.h> 117#include <sys/kauth.h> 118 119#include <net/if.h> 120#include <net/if_dl.h> 121#include <net/route.h> 122#include <net/pktqueue.h> 123#include <net/pfil.h> 124 125#include <netinet/in.h> 126#include <netinet/in_systm.h> 127#include <netinet/ip.h> 128#include <netinet/in_pcb.h> 129#include <netinet/in_proto.h> 130#include <netinet/in_var.h> 131#include <netinet/ip_var.h> 132#include <netinet/ip_private.h> 133#include <netinet/ip_icmp.h> 134/* just for gif_ttl */ 135#include <netinet/in_gif.h> 136#include "gif.h" 137#include <net/if_gre.h> 138#include "gre.h" 139 140#ifdef MROUTING 141#include <netinet/ip_mroute.h> 142#endif 143#include <netinet/portalgo.h> 144 145#ifdef IPSEC 146#include <netipsec/ipsec.h> 147#endif 148 149#ifndef IPFORWARDING 150#ifdef GATEWAY 151#define IPFORWARDING 1 /* forward IP packets not for us */ 152#else /* GATEWAY */ 153#define IPFORWARDING 0 /* don't forward IP packets not for us */ 154#endif /* GATEWAY */ 155#endif /* IPFORWARDING */ 156#ifndef IPSENDREDIRECTS 157#define IPSENDREDIRECTS 1 158#endif 159#ifndef IPFORWSRCRT 160#define IPFORWSRCRT 1 /* forward source-routed packets */ 161#endif 162#ifndef IPALLOWSRCRT 163#define IPALLOWSRCRT 1 /* allow source-routed packets */ 164#endif 165#ifndef IPMTUDISC 166#define IPMTUDISC 1 167#endif 168#ifndef IPMTUDISCTIMEOUT 169#define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 170#endif 171 172#ifdef COMPAT_50 173#include <compat/sys/time.h> 174#include <compat/sys/socket.h> 175#endif 176 177/* 178 * Note: DIRECTED_BROADCAST is handled this way so that previous 179 * configuration using this option will Just Work. 180 */ 181#ifndef IPDIRECTEDBCAST 182#ifdef DIRECTED_BROADCAST 183#define IPDIRECTEDBCAST 1 184#else 185#define IPDIRECTEDBCAST 0 186#endif /* DIRECTED_BROADCAST */ 187#endif /* IPDIRECTEDBCAST */ 188int ipforwarding = IPFORWARDING; 189int ipsendredirects = IPSENDREDIRECTS; 190int ip_defttl = IPDEFTTL; 191int ip_forwsrcrt = IPFORWSRCRT; 192int ip_directedbcast = IPDIRECTEDBCAST; 193int ip_allowsrcrt = IPALLOWSRCRT; 194int ip_mtudisc = IPMTUDISC; 195int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 196#ifdef DIAGNOSTIC 197int ipprintfs = 0; 198#endif 199 200int ip_do_randomid = 0; 201 202/* 203 * XXX - Setting ip_checkinterface mostly implements the receive side of 204 * the Strong ES model described in RFC 1122, but since the routing table 205 * and transmit implementation do not implement the Strong ES model, 206 * setting this to 1 results in an odd hybrid. 207 * 208 * XXX - ip_checkinterface currently must be disabled if you use ipnat 209 * to translate the destination address to another local interface. 210 * 211 * XXX - ip_checkinterface must be disabled if you add IP aliases 212 * to the loopback interface instead of the interface where the 213 * packets for those addresses are received. 214 */ 215static int ip_checkinterface __read_mostly = 0; 216 217struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 218 219pktqueue_t * ip_pktq __read_mostly; 220pfil_head_t * inet_pfil_hook __read_mostly; 221ipid_state_t * ip_ids __read_mostly; 222percpu_t * ipstat_percpu __read_mostly; 223 224static struct route ipforward_rt __cacheline_aligned; 225 226uint16_t ip_id; 227 228#ifdef INET_CSUM_COUNTERS 229#include <sys/device.h> 230 231struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 232 NULL, "inet", "hwcsum bad"); 233struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 234 NULL, "inet", "hwcsum ok"); 235struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 236 NULL, "inet", "swcsum"); 237 238#define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 239 240EVCNT_ATTACH_STATIC(ip_hwcsum_bad); 241EVCNT_ATTACH_STATIC(ip_hwcsum_ok); 242EVCNT_ATTACH_STATIC(ip_swcsum); 243 244#else 245 246#define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 247 248#endif /* INET_CSUM_COUNTERS */ 249 250/* 251 * We need to save the IP options in case a protocol wants to respond 252 * to an incoming packet over the same route if the packet got here 253 * using IP source routing. This allows connection establishment and 254 * maintenance when the remote end is on a network that is not known 255 * to us. 256 */ 257 258static int ip_nhops = 0; 259 260static struct ip_srcrt { 261 struct in_addr dst; /* final destination */ 262 char nop; /* one NOP to align */ 263 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 264 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 265} ip_srcrt; 266 267static int ip_drainwanted; 268 269struct sockaddr_in ipaddr = { 270 .sin_len = sizeof(ipaddr), 271 .sin_family = AF_INET, 272}; 273 274static void save_rte(u_char *, struct in_addr); 275 276#ifdef MBUFTRACE 277struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 278struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 279#endif 280 281static void ipintr(void *); 282static void ip_input(struct mbuf *); 283static void ip_forward(struct mbuf *, int); 284static bool ip_dooptions(struct mbuf *); 285static struct in_ifaddr *ip_rtaddr(struct in_addr); 286static void sysctl_net_inet_ip_setup(struct sysctllog **); 287 288/* XXX: Not yet enabled. */ 289#define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock)) 290#define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock)) 291 292/* 293 * IP initialization: fill in IP protocol switch table. 294 * All protocols not implemented in kernel go to raw IP protocol handler. 295 */ 296void 297ip_init(void) 298{ 299 const struct protosw *pr; 300 301 in_init(); 302 sysctl_net_inet_ip_setup(NULL); 303 304 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 305 KASSERT(pr != NULL); 306 307 ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL); 308 KASSERT(ip_pktq != NULL); 309 310 for (u_int i = 0; i < IPPROTO_MAX; i++) { 311 ip_protox[i] = pr - inetsw; 312 } 313 for (pr = inetdomain.dom_protosw; 314 pr < inetdomain.dom_protoswNPROTOSW; pr++) 315 if (pr->pr_domain->dom_family == PF_INET && 316 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 317 ip_protox[pr->pr_protocol] = pr - inetsw; 318 319 ip_reass_init(); 320 321 ip_ids = ip_id_init(); 322 ip_id = time_uptime & 0xfffff; 323 324 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 325#ifdef GATEWAY 326 ipflow_init(); 327#endif 328 329 /* Register our Packet Filter hook. */ 330 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 331 KASSERT(inet_pfil_hook != NULL); 332 333#ifdef MBUFTRACE 334 MOWNER_ATTACH(&ip_tx_mowner); 335 MOWNER_ATTACH(&ip_rx_mowner); 336#endif /* MBUFTRACE */ 337 338 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 339} 340 341/* 342 * IP software interrupt routine. 343 */ 344static void 345ipintr(void *arg __unused) 346{ 347 struct mbuf *m; 348 349 KASSERT(cpu_softintr_p()); 350 351 mutex_enter(softnet_lock); 352 while ((m = pktq_dequeue(ip_pktq)) != NULL) { 353 ip_input(m); 354 } 355 mutex_exit(softnet_lock); 356} 357 358/* 359 * IP input routine. Checksum and byte swap header. If fragmented 360 * try to reassemble. Process options. Pass to next level. 361 */ 362static void 363ip_input(struct mbuf *m) 364{ 365 struct ip *ip = NULL; 366 struct in_ifaddr *ia; 367 struct ifaddr *ifa; 368 int hlen = 0, len; 369 int downmatch; 370 int checkif; 371 int srcrt = 0; 372 ifnet_t *ifp; 373 374 KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software " 375 "interrupt handler; synchronization assumptions violated"); 376 377 MCLAIM(m, &ip_rx_mowner); 378 KASSERT((m->m_flags & M_PKTHDR) != 0); 379 ifp = m->m_pkthdr.rcvif; 380 381 /* 382 * If no IP addresses have been set yet but the interfaces 383 * are receiving, can't do anything with incoming packets yet. 384 * Note: we pre-check without locks held. 385 */ 386 if (!TAILQ_FIRST(&in_ifaddrhead)) { 387 goto bad; 388 } 389 IP_STATINC(IP_STAT_TOTAL); 390 391 /* 392 * If the IP header is not aligned, slurp it up into a new 393 * mbuf with space for link headers, in the event we forward 394 * it. Otherwise, if it is aligned, make sure the entire 395 * base IP header is in the first mbuf of the chain. 396 */ 397 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 398 if ((m = m_copyup(m, sizeof(struct ip), 399 (max_linkhdr + 3) & ~3)) == NULL) { 400 /* XXXJRT new stat, please */ 401 IP_STATINC(IP_STAT_TOOSMALL); 402 return; 403 } 404 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 405 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 406 IP_STATINC(IP_STAT_TOOSMALL); 407 return; 408 } 409 } 410 ip = mtod(m, struct ip *); 411 if (ip->ip_v != IPVERSION) { 412 IP_STATINC(IP_STAT_BADVERS); 413 goto bad; 414 } 415 hlen = ip->ip_hl << 2; 416 if (hlen < sizeof(struct ip)) { /* minimum header length */ 417 IP_STATINC(IP_STAT_BADHLEN); 418 goto bad; 419 } 420 if (hlen > m->m_len) { 421 if ((m = m_pullup(m, hlen)) == NULL) { 422 IP_STATINC(IP_STAT_BADHLEN); 423 return; 424 } 425 ip = mtod(m, struct ip *); 426 } 427 428 /* 429 * RFC1122: packets with a multicast source address are 430 * not allowed. 431 */ 432 if (IN_MULTICAST(ip->ip_src.s_addr)) { 433 IP_STATINC(IP_STAT_BADADDR); 434 goto bad; 435 } 436 437 /* 127/8 must not appear on wire - RFC1122 */ 438 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 439 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 440 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 441 IP_STATINC(IP_STAT_BADADDR); 442 goto bad; 443 } 444 } 445 446 switch (m->m_pkthdr.csum_flags & 447 ((ifp->if_csum_flags_rx & M_CSUM_IPv4) | 448 M_CSUM_IPv4_BAD)) { 449 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 450 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 451 goto badcsum; 452 453 case M_CSUM_IPv4: 454 /* Checksum was okay. */ 455 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 456 break; 457 458 default: 459 /* 460 * Must compute it ourselves. Maybe skip checksum on 461 * loopback interfaces. 462 */ 463 if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) || 464 ip_do_loopback_cksum)) { 465 INET_CSUM_COUNTER_INCR(&ip_swcsum); 466 if (in_cksum(m, hlen) != 0) 467 goto badcsum; 468 } 469 break; 470 } 471 472 /* Retrieve the packet length. */ 473 len = ntohs(ip->ip_len); 474 475 /* 476 * Check for additional length bogosity 477 */ 478 if (len < hlen) { 479 IP_STATINC(IP_STAT_BADLEN); 480 goto bad; 481 } 482 483 /* 484 * Check that the amount of data in the buffers 485 * is as at least much as the IP header would have us expect. 486 * Trim mbufs if longer than we expect. 487 * Drop packet if shorter than we expect. 488 */ 489 if (m->m_pkthdr.len < len) { 490 IP_STATINC(IP_STAT_TOOSHORT); 491 goto bad; 492 } 493 if (m->m_pkthdr.len > len) { 494 if (m->m_len == m->m_pkthdr.len) { 495 m->m_len = len; 496 m->m_pkthdr.len = len; 497 } else 498 m_adj(m, len - m->m_pkthdr.len); 499 } 500 501 /* 502 * Assume that we can create a fast-forward IP flow entry 503 * based on this packet. 504 */ 505 m->m_flags |= M_CANFASTFWD; 506 507 /* 508 * Run through list of hooks for input packets. If there are any 509 * filters which require that additional packets in the flow are 510 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 511 * Note that filters must _never_ set this flag, as another filter 512 * in the list may have previously cleared it. 513 */ 514#if defined(IPSEC) 515 if (!ipsec_used || !ipsec_indone(m)) 516#else 517 if (1) 518#endif 519 { 520 struct in_addr odst = ip->ip_dst; 521 bool freed; 522 523 SOFTNET_LOCK(); 524 freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0; 525 SOFTNET_UNLOCK(); 526 if (freed || m == NULL) { 527 return; 528 } 529 ip = mtod(m, struct ip *); 530 hlen = ip->ip_hl << 2; 531 532 /* 533 * XXX The setting of "srcrt" here is to prevent ip_forward() 534 * from generating ICMP redirects for packets that have 535 * been redirected by a hook back out on to the same LAN that 536 * they came from and is not an indication that the packet 537 * is being inffluenced by source routing options. This 538 * allows things like 539 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 540 * where tlp0 is both on the 1.1.1.0/24 network and is the 541 * default route for hosts on 1.1.1.0/24. Of course this 542 * also requires a "map tlp0 ..." to complete the story. 543 * One might argue whether or not this kind of network config. 544 * should be supported in this manner... 545 */ 546 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 547 } 548 549#ifdef ALTQ 550 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 551 if (altq_input) { 552 SOFTNET_LOCK(); 553 if ((*altq_input)(m, AF_INET) == 0) { 554 /* Packet dropped by traffic conditioner. */ 555 SOFTNET_UNLOCK(); 556 return; 557 } 558 SOFTNET_UNLOCK(); 559 } 560#endif 561 562 /* 563 * Process options and, if not destined for us, 564 * ship it on. ip_dooptions returns 1 when an 565 * error was detected (causing an icmp message 566 * to be sent and the original packet to be freed). 567 */ 568 ip_nhops = 0; /* for source routed packets */ 569 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 570 return; 571 572 /* 573 * Enable a consistency check between the destination address 574 * and the arrival interface for a unicast packet (the RFC 1122 575 * strong ES model) if IP forwarding is disabled and the packet 576 * is not locally generated. 577 * 578 * XXX - Checking also should be disabled if the destination 579 * address is ipnat'ed to a different interface. 580 * 581 * XXX - Checking is incompatible with IP aliases added 582 * to the loopback interface instead of the interface where 583 * the packets are received. 584 * 585 * XXX - We need to add a per ifaddr flag for this so that 586 * we get finer grain control. 587 */ 588 checkif = ip_checkinterface && (ipforwarding == 0) && 589 (ifp->if_flags & IFF_LOOPBACK) == 0; 590 591 /* 592 * Check our list of addresses, to see if the packet is for us. 593 * 594 * Traditional 4.4BSD did not consult IFF_UP at all. 595 * The behavior here is to treat addresses on !IFF_UP interface 596 * or IN_IFF_NOTREADY addresses as not mine. 597 */ 598 downmatch = 0; 599 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 600 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 601 if (ia->ia4_flags & IN_IFF_NOTREADY) 602 continue; 603 if (checkif && ia->ia_ifp != ifp) 604 continue; 605 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 606 break; 607 else 608 downmatch++; 609 } 610 } 611 if (ia != NULL) 612 goto ours; 613 if (ifp->if_flags & IFF_BROADCAST) { 614 IFADDR_FOREACH(ifa, ifp) { 615 if (ifa->ifa_addr->sa_family != AF_INET) 616 continue; 617 ia = ifatoia(ifa); 618 if (ia->ia4_flags & IN_IFF_NOTREADY) 619 continue; 620 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 621 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 622 /* 623 * Look for all-0's host part (old broadcast addr), 624 * either for subnet or net. 625 */ 626 ip->ip_dst.s_addr == ia->ia_subnet || 627 ip->ip_dst.s_addr == ia->ia_net) 628 goto ours; 629 /* 630 * An interface with IP address zero accepts 631 * all packets that arrive on that interface. 632 */ 633 if (in_nullhost(ia->ia_addr.sin_addr)) 634 goto ours; 635 } 636 } 637 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 638#ifdef MROUTING 639 extern struct socket *ip_mrouter; 640 641 if (ip_mrouter) { 642 /* 643 * If we are acting as a multicast router, all 644 * incoming multicast packets are passed to the 645 * kernel-level multicast forwarding function. 646 * The packet is returned (relatively) intact; if 647 * ip_mforward() returns a non-zero value, the packet 648 * must be discarded, else it may be accepted below. 649 * 650 * (The IP ident field is put in the same byte order 651 * as expected when ip_mforward() is called from 652 * ip_output().) 653 */ 654 SOFTNET_LOCK(); 655 if (ip_mforward(m, ifp) != 0) { 656 SOFTNET_UNLOCK(); 657 IP_STATINC(IP_STAT_CANTFORWARD); 658 m_freem(m); 659 return; 660 } 661 SOFTNET_UNLOCK(); 662 663 /* 664 * The process-level routing demon needs to receive 665 * all multicast IGMP packets, whether or not this 666 * host belongs to their destination groups. 667 */ 668 if (ip->ip_p == IPPROTO_IGMP) { 669 goto ours; 670 } 671 IP_STATINC(IP_STAT_CANTFORWARD); 672 } 673#endif 674 /* 675 * See if we belong to the destination multicast group on the 676 * arrival interface. 677 */ 678 if (!in_multi_group(ip->ip_dst, ifp, 0)) { 679 IP_STATINC(IP_STAT_CANTFORWARD); 680 m_freem(m); 681 return; 682 } 683 goto ours; 684 } 685 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 686 in_nullhost(ip->ip_dst)) 687 goto ours; 688 689 /* 690 * Not for us; forward if possible and desirable. 691 */ 692 if (ipforwarding == 0) { 693 IP_STATINC(IP_STAT_CANTFORWARD); 694 m_freem(m); 695 } else { 696 /* 697 * If ip_dst matched any of my address on !IFF_UP interface, 698 * and there's no IFF_UP interface that matches ip_dst, 699 * send icmp unreach. Forwarding it will result in in-kernel 700 * forwarding loop till TTL goes to 0. 701 */ 702 if (downmatch) { 703 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 704 IP_STATINC(IP_STAT_CANTFORWARD); 705 return; 706 } 707#ifdef IPSEC 708 /* Perform IPsec, if any. */ 709 if (ipsec_used) { 710 SOFTNET_LOCK(); 711 if (ipsec4_input(m, IP_FORWARDING | 712 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) { 713 SOFTNET_UNLOCK(); 714 goto bad; 715 } 716 SOFTNET_UNLOCK(); 717 } 718#endif 719 ip_forward(m, srcrt); 720 } 721 return; 722 723ours: 724 /* 725 * If offset or IP_MF are set, must reassemble. 726 */ 727 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 728 /* 729 * Pass to IP reassembly mechanism. 730 */ 731 if (ip_reass_packet(&m, ip) != 0) { 732 /* Failed; invalid fragment(s) or packet. */ 733 goto bad; 734 } 735 if (m == NULL) { 736 /* More fragments should come; silently return. */ 737 return; 738 } 739 /* 740 * Reassembly is done, we have the final packet. 741 * Updated cached data in local variable(s). 742 */ 743 ip = mtod(m, struct ip *); 744 hlen = ip->ip_hl << 2; 745 } 746 747#ifdef IPSEC 748 /* 749 * Enforce IPsec policy checking if we are seeing last header. 750 * Note that we do not visit this with protocols with PCB layer 751 * code - like UDP/TCP/raw IP. 752 */ 753 if (ipsec_used && 754 (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 755 SOFTNET_LOCK(); 756 if (ipsec4_input(m, 0) != 0) { 757 SOFTNET_UNLOCK(); 758 goto bad; 759 } 760 SOFTNET_UNLOCK(); 761 } 762#endif 763 764 /* 765 * Switch out to protocol's input routine. 766 */ 767#if IFA_STATS 768 if (ia && ip) 769 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 770#endif 771 IP_STATINC(IP_STAT_DELIVERED); 772 773 const int off = hlen, nh = ip->ip_p; 774 775 SOFTNET_LOCK(); 776 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 777 SOFTNET_UNLOCK(); 778 return; 779bad: 780 m_freem(m); 781 return; 782 783badcsum: 784 IP_STATINC(IP_STAT_BADSUM); 785 m_freem(m); 786} 787 788/* 789 * IP timer processing. 790 */ 791void 792ip_slowtimo(void) 793{ 794 795 mutex_enter(softnet_lock); 796 KERNEL_LOCK(1, NULL); 797 798 ip_reass_slowtimo(); 799 800 KERNEL_UNLOCK_ONE(NULL); 801 mutex_exit(softnet_lock); 802} 803 804/* 805 * IP drain processing. 806 */ 807void 808ip_drain(void) 809{ 810 811 KERNEL_LOCK(1, NULL); 812 ip_reass_drain(); 813 KERNEL_UNLOCK_ONE(NULL); 814} 815 816/* 817 * ip_dooptions: perform option processing on a datagram, possibly discarding 818 * it if bad options are encountered, or forwarding it if source-routed. 819 * 820 * => Returns true if packet has been forwarded/freed. 821 * => Returns false if the packet should be processed further. 822 */ 823static bool 824ip_dooptions(struct mbuf *m) 825{ 826 struct ip *ip = mtod(m, struct ip *); 827 u_char *cp, *cp0; 828 struct ip_timestamp *ipt; 829 struct in_ifaddr *ia; 830 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 831 struct in_addr dst; 832 n_time ntime; 833 834 dst = ip->ip_dst; 835 cp = (u_char *)(ip + 1); 836 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 837 for (; cnt > 0; cnt -= optlen, cp += optlen) { 838 opt = cp[IPOPT_OPTVAL]; 839 if (opt == IPOPT_EOL) 840 break; 841 if (opt == IPOPT_NOP) 842 optlen = 1; 843 else { 844 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 845 code = &cp[IPOPT_OLEN] - (u_char *)ip; 846 goto bad; 847 } 848 optlen = cp[IPOPT_OLEN]; 849 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 850 code = &cp[IPOPT_OLEN] - (u_char *)ip; 851 goto bad; 852 } 853 } 854 switch (opt) { 855 856 default: 857 break; 858 859 /* 860 * Source routing with record. 861 * Find interface with current destination address. 862 * If none on this machine then drop if strictly routed, 863 * or do nothing if loosely routed. 864 * Record interface address and bring up next address 865 * component. If strictly routed make sure next 866 * address is on directly accessible net. 867 */ 868 case IPOPT_LSRR: 869 case IPOPT_SSRR: 870 if (ip_allowsrcrt == 0) { 871 type = ICMP_UNREACH; 872 code = ICMP_UNREACH_NET_PROHIB; 873 goto bad; 874 } 875 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 876 code = &cp[IPOPT_OLEN] - (u_char *)ip; 877 goto bad; 878 } 879 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 880 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 881 goto bad; 882 } 883 ipaddr.sin_addr = ip->ip_dst; 884 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 885 if (ia == 0) { 886 if (opt == IPOPT_SSRR) { 887 type = ICMP_UNREACH; 888 code = ICMP_UNREACH_SRCFAIL; 889 goto bad; 890 } 891 /* 892 * Loose routing, and not at next destination 893 * yet; nothing to do except forward. 894 */ 895 break; 896 } 897 off--; /* 0 origin */ 898 if ((off + sizeof(struct in_addr)) > optlen) { 899 /* 900 * End of source route. Should be for us. 901 */ 902 save_rte(cp, ip->ip_src); 903 break; 904 } 905 /* 906 * locate outgoing interface 907 */ 908 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 909 sizeof(ipaddr.sin_addr)); 910 if (opt == IPOPT_SSRR) 911 ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); 912 else 913 ia = ip_rtaddr(ipaddr.sin_addr); 914 if (ia == 0) { 915 type = ICMP_UNREACH; 916 code = ICMP_UNREACH_SRCFAIL; 917 goto bad; 918 } 919 ip->ip_dst = ipaddr.sin_addr; 920 bcopy((void *)&ia->ia_addr.sin_addr, 921 (void *)(cp + off), sizeof(struct in_addr)); 922 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 923 /* 924 * Let ip_intr's mcast routing check handle mcast pkts 925 */ 926 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 927 break; 928 929 case IPOPT_RR: 930 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 931 code = &cp[IPOPT_OLEN] - (u_char *)ip; 932 goto bad; 933 } 934 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 935 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 936 goto bad; 937 } 938 /* 939 * If no space remains, ignore. 940 */ 941 off--; /* 0 origin */ 942 if ((off + sizeof(struct in_addr)) > optlen) 943 break; 944 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 945 sizeof(ipaddr.sin_addr)); 946 /* 947 * locate outgoing interface; if we're the destination, 948 * use the incoming interface (should be same). 949 */ 950 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 951 == NULL && 952 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 953 type = ICMP_UNREACH; 954 code = ICMP_UNREACH_HOST; 955 goto bad; 956 } 957 bcopy((void *)&ia->ia_addr.sin_addr, 958 (void *)(cp + off), sizeof(struct in_addr)); 959 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 960 break; 961 962 case IPOPT_TS: 963 code = cp - (u_char *)ip; 964 ipt = (struct ip_timestamp *)cp; 965 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 966 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 967 goto bad; 968 } 969 if (ipt->ipt_ptr < 5) { 970 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 971 goto bad; 972 } 973 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 974 if (++ipt->ipt_oflw == 0) { 975 code = (u_char *)&ipt->ipt_ptr - 976 (u_char *)ip; 977 goto bad; 978 } 979 break; 980 } 981 cp0 = (cp + ipt->ipt_ptr - 1); 982 switch (ipt->ipt_flg) { 983 984 case IPOPT_TS_TSONLY: 985 break; 986 987 case IPOPT_TS_TSANDADDR: 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 ipaddr.sin_addr = dst; 995 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 996 m->m_pkthdr.rcvif)); 997 if (ia == 0) 998 continue; 999 bcopy(&ia->ia_addr.sin_addr, 1000 cp0, sizeof(struct in_addr)); 1001 ipt->ipt_ptr += sizeof(struct in_addr); 1002 break; 1003 1004 case IPOPT_TS_PRESPEC: 1005 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1006 sizeof(struct in_addr) > ipt->ipt_len) { 1007 code = (u_char *)&ipt->ipt_ptr - 1008 (u_char *)ip; 1009 goto bad; 1010 } 1011 memcpy(&ipaddr.sin_addr, cp0, 1012 sizeof(struct in_addr)); 1013 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1014 == NULL) 1015 continue; 1016 ipt->ipt_ptr += sizeof(struct in_addr); 1017 break; 1018 1019 default: 1020 /* XXX can't take &ipt->ipt_flg */ 1021 code = (u_char *)&ipt->ipt_ptr - 1022 (u_char *)ip + 1; 1023 goto bad; 1024 } 1025 ntime = iptime(); 1026 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1027 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1028 sizeof(n_time)); 1029 ipt->ipt_ptr += sizeof(n_time); 1030 } 1031 } 1032 if (forward) { 1033 if (ip_forwsrcrt == 0) { 1034 type = ICMP_UNREACH; 1035 code = ICMP_UNREACH_SRCFAIL; 1036 goto bad; 1037 } 1038 ip_forward(m, 1); 1039 return true; 1040 } 1041 return false; 1042bad: 1043 icmp_error(m, type, code, 0, 0); 1044 IP_STATINC(IP_STAT_BADOPTIONS); 1045 return true; 1046} 1047 1048/* 1049 * ip_rtaddr: given address of next destination (final or next hop), 1050 * return internet address info of interface to be used to get there. 1051 */ 1052static struct in_ifaddr * 1053ip_rtaddr(struct in_addr dst) 1054{ 1055 struct rtentry *rt; 1056 union { 1057 struct sockaddr dst; 1058 struct sockaddr_in dst4; 1059 } u; 1060 1061 sockaddr_in_init(&u.dst4, &dst, 0); 1062 1063 SOFTNET_LOCK(); 1064 rt = rtcache_lookup(&ipforward_rt, &u.dst); 1065 SOFTNET_UNLOCK(); 1066 if (rt == NULL) 1067 return NULL; 1068 1069 return ifatoia(rt->rt_ifa); 1070} 1071 1072/* 1073 * save_rte: save incoming source route for use in replies, to be picked 1074 * up later by ip_srcroute if the receiver is interested. 1075 */ 1076static void 1077save_rte(u_char *option, struct in_addr dst) 1078{ 1079 unsigned olen; 1080 1081 olen = option[IPOPT_OLEN]; 1082 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1083 return; 1084 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1085 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1086 ip_srcrt.dst = dst; 1087} 1088 1089/* 1090 * Retrieve incoming source route for use in replies, 1091 * in the same form used by setsockopt. 1092 * The first hop is placed before the options, will be removed later. 1093 */ 1094struct mbuf * 1095ip_srcroute(void) 1096{ 1097 struct in_addr *p, *q; 1098 struct mbuf *m; 1099 1100 if (ip_nhops == 0) 1101 return NULL; 1102 m = m_get(M_DONTWAIT, MT_SOOPTS); 1103 if (m == 0) 1104 return NULL; 1105 1106 MCLAIM(m, &inetdomain.dom_mowner); 1107#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1108 1109 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1110 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1111 OPTSIZ; 1112 1113 /* 1114 * First save first hop for return route 1115 */ 1116 p = &ip_srcrt.route[ip_nhops - 1]; 1117 *(mtod(m, struct in_addr *)) = *p--; 1118 1119 /* 1120 * Copy option fields and padding (nop) to mbuf. 1121 */ 1122 ip_srcrt.nop = IPOPT_NOP; 1123 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1124 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1125 OPTSIZ); 1126 q = (struct in_addr *)(mtod(m, char *) + 1127 sizeof(struct in_addr) + OPTSIZ); 1128#undef OPTSIZ 1129 /* 1130 * Record return path as an IP source route, 1131 * reversing the path (pointers are now aligned). 1132 */ 1133 while (p >= ip_srcrt.route) { 1134 *q++ = *p--; 1135 } 1136 /* 1137 * Last hop goes to final destination. 1138 */ 1139 *q = ip_srcrt.dst; 1140 return (m); 1141} 1142 1143const int inetctlerrmap[PRC_NCMDS] = { 1144 [PRC_MSGSIZE] = EMSGSIZE, 1145 [PRC_HOSTDEAD] = EHOSTDOWN, 1146 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1147 [PRC_UNREACH_NET] = EHOSTUNREACH, 1148 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1149 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1150 [PRC_UNREACH_PORT] = ECONNREFUSED, 1151 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1152 [PRC_PARAMPROB] = ENOPROTOOPT, 1153}; 1154 1155void 1156ip_fasttimo(void) 1157{ 1158 if (ip_drainwanted) { 1159 ip_drain(); 1160 ip_drainwanted = 0; 1161 } 1162} 1163 1164void 1165ip_drainstub(void) 1166{ 1167 ip_drainwanted = 1; 1168} 1169 1170/* 1171 * Forward a packet. If some error occurs return the sender 1172 * an icmp packet. Note we can't always generate a meaningful 1173 * icmp message because icmp doesn't have a large enough repertoire 1174 * of codes and types. 1175 * 1176 * If not forwarding, just drop the packet. This could be confusing 1177 * if ipforwarding was zero but some routing protocol was advancing 1178 * us as a gateway to somewhere. However, we must let the routing 1179 * protocol deal with that. 1180 * 1181 * The srcrt parameter indicates whether the packet is being forwarded 1182 * via a source route. 1183 */ 1184static void 1185ip_forward(struct mbuf *m, int srcrt) 1186{ 1187 struct ip *ip = mtod(m, struct ip *); 1188 struct rtentry *rt; 1189 int error, type = 0, code = 0, destmtu = 0; 1190 struct mbuf *mcopy; 1191 n_long dest; 1192 union { 1193 struct sockaddr dst; 1194 struct sockaddr_in dst4; 1195 } u; 1196 uint64_t *ips; 1197 1198 KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software " 1199 "interrupt handler; synchronization assumptions violated"); 1200 1201 /* 1202 * We are now in the output path. 1203 */ 1204 MCLAIM(m, &ip_tx_mowner); 1205 1206 /* 1207 * Clear any in-bound checksum flags for this packet. 1208 */ 1209 m->m_pkthdr.csum_flags = 0; 1210 1211 dest = 0; 1212 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1213 IP_STATINC(IP_STAT_CANTFORWARD); 1214 m_freem(m); 1215 return; 1216 } 1217 1218 SOFTNET_LOCK(); 1219 1220 if (ip->ip_ttl <= IPTTLDEC) { 1221 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1222 SOFTNET_UNLOCK(); 1223 return; 1224 } 1225 1226 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1227 1228 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1229 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1230 SOFTNET_UNLOCK(); 1231 return; 1232 } 1233 1234 /* 1235 * Save at most 68 bytes of the packet in case 1236 * we need to generate an ICMP message to the src. 1237 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1238 */ 1239 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1240 if (mcopy) 1241 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1242 1243 ip->ip_ttl -= IPTTLDEC; 1244 1245 /* 1246 * If forwarding packet using same interface that it came in on, 1247 * perhaps should send a redirect to sender to shortcut a hop. 1248 * Only send redirect if source is sending directly to us, 1249 * and if packet was not source routed (or has any options). 1250 * Also, don't send redirect if forwarding using a default route 1251 * or a route modified by a redirect. 1252 */ 1253 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1254 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1255 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1256 ipsendredirects && !srcrt) { 1257 if (rt->rt_ifa && 1258 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1259 ifatoia(rt->rt_ifa)->ia_subnet) { 1260 if (rt->rt_flags & RTF_GATEWAY) 1261 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1262 else 1263 dest = ip->ip_dst.s_addr; 1264 /* 1265 * Router requirements says to only send host 1266 * redirects. 1267 */ 1268 type = ICMP_REDIRECT; 1269 code = ICMP_REDIRECT_HOST; 1270 } 1271 } 1272 1273 error = ip_output(m, NULL, &ipforward_rt, 1274 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1275 NULL, NULL); 1276 1277 if (error) { 1278 IP_STATINC(IP_STAT_CANTFORWARD); 1279 goto error; 1280 } 1281 1282 ips = IP_STAT_GETREF(); 1283 ips[IP_STAT_FORWARD]++; 1284 1285 if (type) { 1286 ips[IP_STAT_REDIRECTSENT]++; 1287 IP_STAT_PUTREF(); 1288 goto redirect; 1289 } 1290 1291 IP_STAT_PUTREF(); 1292 if (mcopy) { 1293#ifdef GATEWAY 1294 if (mcopy->m_flags & M_CANFASTFWD) 1295 ipflow_create(&ipforward_rt, mcopy); 1296#endif 1297 m_freem(mcopy); 1298 } 1299 1300 SOFTNET_UNLOCK(); 1301 return; 1302 1303redirect: 1304error: 1305 if (mcopy == NULL) { 1306 SOFTNET_UNLOCK(); 1307 return; 1308 } 1309 1310 switch (error) { 1311 1312 case 0: /* forwarded, but need redirect */ 1313 /* type, code set above */ 1314 break; 1315 1316 case ENETUNREACH: /* shouldn't happen, checked above */ 1317 case EHOSTUNREACH: 1318 case ENETDOWN: 1319 case EHOSTDOWN: 1320 default: 1321 type = ICMP_UNREACH; 1322 code = ICMP_UNREACH_HOST; 1323 break; 1324 1325 case EMSGSIZE: 1326 type = ICMP_UNREACH; 1327 code = ICMP_UNREACH_NEEDFRAG; 1328 1329 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1330 destmtu = rt->rt_ifp->if_mtu; 1331#ifdef IPSEC 1332 if (ipsec_used) 1333 (void)ipsec4_forward(mcopy, &destmtu); 1334#endif 1335 IP_STATINC(IP_STAT_CANTFRAG); 1336 break; 1337 1338 case ENOBUFS: 1339 /* 1340 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1341 * Requirements for IP Version 4 Routers. Source quench can 1342 * big problem under DoS attacks or if the underlying 1343 * interface is rate-limited. 1344 */ 1345 if (mcopy) 1346 m_freem(mcopy); 1347 SOFTNET_UNLOCK(); 1348 return; 1349 } 1350 icmp_error(mcopy, type, code, dest, destmtu); 1351 SOFTNET_UNLOCK(); 1352} 1353 1354void 1355ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1356 struct mbuf *m) 1357{ 1358 struct socket *so = inp->inp_socket; 1359 ifnet_t *ifp = m->m_pkthdr.rcvif; 1360 int inpflags = inp->inp_flags; 1361 1362 if (so->so_options & SO_TIMESTAMP 1363#ifdef SO_OTIMESTAMP 1364 || so->so_options & SO_OTIMESTAMP 1365#endif 1366 ) { 1367 struct timeval tv; 1368 1369 microtime(&tv); 1370#ifdef SO_OTIMESTAMP 1371 if (so->so_options & SO_OTIMESTAMP) { 1372 struct timeval50 tv50; 1373 timeval_to_timeval50(&tv, &tv50); 1374 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1375 SCM_OTIMESTAMP, SOL_SOCKET); 1376 } else 1377#endif 1378 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1379 SCM_TIMESTAMP, SOL_SOCKET); 1380 if (*mp) 1381 mp = &(*mp)->m_next; 1382 } 1383 if (inpflags & INP_RECVDSTADDR) { 1384 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1385 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1386 if (*mp) 1387 mp = &(*mp)->m_next; 1388 } 1389 if (inpflags & INP_RECVPKTINFO) { 1390 struct in_pktinfo ipi; 1391 ipi.ipi_addr = ip->ip_src; 1392 ipi.ipi_ifindex = ifp->if_index; 1393 *mp = sbcreatecontrol((void *) &ipi, 1394 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1395 if (*mp) 1396 mp = &(*mp)->m_next; 1397 } 1398 if (inpflags & INP_PKTINFO) { 1399 struct in_pktinfo ipi; 1400 ipi.ipi_addr = ip->ip_dst; 1401 ipi.ipi_ifindex = ifp->if_index; 1402 *mp = sbcreatecontrol((void *) &ipi, 1403 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1404 if (*mp) 1405 mp = &(*mp)->m_next; 1406 } 1407 if (inpflags & INP_RECVIF) { 1408 struct sockaddr_dl sdl; 1409 1410 sockaddr_dl_init(&sdl, sizeof(sdl), ifp ? 1411 ifp->if_index : 0, 0, NULL, 0, NULL, 0); 1412 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1413 if (*mp) 1414 mp = &(*mp)->m_next; 1415 } 1416 if (inpflags & INP_RECVTTL) { 1417 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1418 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1419 if (*mp) 1420 mp = &(*mp)->m_next; 1421 } 1422} 1423 1424/* 1425 * sysctl helper routine for net.inet.ip.forwsrcrt. 1426 */ 1427static int 1428sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1429{ 1430 int error, tmp; 1431 struct sysctlnode node; 1432 1433 node = *rnode; 1434 tmp = ip_forwsrcrt; 1435 node.sysctl_data = &tmp; 1436 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1437 if (error || newp == NULL) 1438 return (error); 1439 1440 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1441 0, NULL, NULL, NULL); 1442 if (error) 1443 return (error); 1444 1445 ip_forwsrcrt = tmp; 1446 1447 return (0); 1448} 1449 1450/* 1451 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1452 * range of the new value and tweaks timers if it changes. 1453 */ 1454static int 1455sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1456{ 1457 int error, tmp; 1458 struct sysctlnode node; 1459 1460 node = *rnode; 1461 tmp = ip_mtudisc_timeout; 1462 node.sysctl_data = &tmp; 1463 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1464 if (error || newp == NULL) 1465 return (error); 1466 if (tmp < 0) 1467 return (EINVAL); 1468 1469 mutex_enter(softnet_lock); 1470 1471 ip_mtudisc_timeout = tmp; 1472 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1473 1474 mutex_exit(softnet_lock); 1475 1476 return (0); 1477} 1478 1479static int 1480sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1481{ 1482 1483 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1484} 1485 1486static void 1487sysctl_net_inet_ip_setup(struct sysctllog **clog) 1488{ 1489 sysctl_createv(clog, 0, NULL, NULL, 1490 CTLFLAG_PERMANENT, 1491 CTLTYPE_NODE, "inet", 1492 SYSCTL_DESCR("PF_INET related settings"), 1493 NULL, 0, NULL, 0, 1494 CTL_NET, PF_INET, CTL_EOL); 1495 sysctl_createv(clog, 0, NULL, NULL, 1496 CTLFLAG_PERMANENT, 1497 CTLTYPE_NODE, "ip", 1498 SYSCTL_DESCR("IPv4 related settings"), 1499 NULL, 0, NULL, 0, 1500 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1501 1502 sysctl_createv(clog, 0, NULL, NULL, 1503 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1504 CTLTYPE_INT, "forwarding", 1505 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1506 NULL, 0, &ipforwarding, 0, 1507 CTL_NET, PF_INET, IPPROTO_IP, 1508 IPCTL_FORWARDING, CTL_EOL); 1509 sysctl_createv(clog, 0, NULL, NULL, 1510 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1511 CTLTYPE_INT, "redirect", 1512 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1513 NULL, 0, &ipsendredirects, 0, 1514 CTL_NET, PF_INET, IPPROTO_IP, 1515 IPCTL_SENDREDIRECTS, CTL_EOL); 1516 sysctl_createv(clog, 0, NULL, NULL, 1517 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1518 CTLTYPE_INT, "ttl", 1519 SYSCTL_DESCR("Default TTL for an INET datagram"), 1520 NULL, 0, &ip_defttl, 0, 1521 CTL_NET, PF_INET, IPPROTO_IP, 1522 IPCTL_DEFTTL, CTL_EOL); 1523#ifdef IPCTL_DEFMTU 1524 sysctl_createv(clog, 0, NULL, NULL, 1525 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1526 CTLTYPE_INT, "mtu", 1527 SYSCTL_DESCR("Default MTA for an INET route"), 1528 NULL, 0, &ip_mtu, 0, 1529 CTL_NET, PF_INET, IPPROTO_IP, 1530 IPCTL_DEFMTU, CTL_EOL); 1531#endif /* IPCTL_DEFMTU */ 1532 sysctl_createv(clog, 0, NULL, NULL, 1533 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1534 CTLTYPE_INT, "forwsrcrt", 1535 SYSCTL_DESCR("Enable forwarding of source-routed " 1536 "datagrams"), 1537 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1538 CTL_NET, PF_INET, IPPROTO_IP, 1539 IPCTL_FORWSRCRT, CTL_EOL); 1540 sysctl_createv(clog, 0, NULL, NULL, 1541 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1542 CTLTYPE_INT, "directed-broadcast", 1543 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1544 NULL, 0, &ip_directedbcast, 0, 1545 CTL_NET, PF_INET, IPPROTO_IP, 1546 IPCTL_DIRECTEDBCAST, CTL_EOL); 1547 sysctl_createv(clog, 0, NULL, NULL, 1548 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1549 CTLTYPE_INT, "allowsrcrt", 1550 SYSCTL_DESCR("Accept source-routed datagrams"), 1551 NULL, 0, &ip_allowsrcrt, 0, 1552 CTL_NET, PF_INET, IPPROTO_IP, 1553 IPCTL_ALLOWSRCRT, CTL_EOL); 1554 1555 sysctl_createv(clog, 0, NULL, NULL, 1556 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1557 CTLTYPE_INT, "mtudisc", 1558 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1559 NULL, 0, &ip_mtudisc, 0, 1560 CTL_NET, PF_INET, IPPROTO_IP, 1561 IPCTL_MTUDISC, CTL_EOL); 1562 sysctl_createv(clog, 0, NULL, NULL, 1563 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1564 CTLTYPE_INT, "anonportmin", 1565 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1566 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1567 CTL_NET, PF_INET, IPPROTO_IP, 1568 IPCTL_ANONPORTMIN, CTL_EOL); 1569 sysctl_createv(clog, 0, NULL, NULL, 1570 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1571 CTLTYPE_INT, "anonportmax", 1572 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1573 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1574 CTL_NET, PF_INET, IPPROTO_IP, 1575 IPCTL_ANONPORTMAX, CTL_EOL); 1576 sysctl_createv(clog, 0, NULL, NULL, 1577 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1578 CTLTYPE_INT, "mtudisctimeout", 1579 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1580 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1581 CTL_NET, PF_INET, IPPROTO_IP, 1582 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1583#if NGIF > 0 1584 sysctl_createv(clog, 0, NULL, NULL, 1585 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1586 CTLTYPE_INT, "gifttl", 1587 SYSCTL_DESCR("Default TTL for a gif tunnel datagram"), 1588 NULL, 0, &ip_gif_ttl, 0, 1589 CTL_NET, PF_INET, IPPROTO_IP, 1590 IPCTL_GIF_TTL, CTL_EOL); 1591#endif /* NGIF */ 1592#ifndef IPNOPRIVPORTS 1593 sysctl_createv(clog, 0, NULL, NULL, 1594 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1595 CTLTYPE_INT, "lowportmin", 1596 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1597 "to assign"), 1598 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1599 CTL_NET, PF_INET, IPPROTO_IP, 1600 IPCTL_LOWPORTMIN, CTL_EOL); 1601 sysctl_createv(clog, 0, NULL, NULL, 1602 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1603 CTLTYPE_INT, "lowportmax", 1604 SYSCTL_DESCR("Highest privileged ephemeral port number " 1605 "to assign"), 1606 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1607 CTL_NET, PF_INET, IPPROTO_IP, 1608 IPCTL_LOWPORTMAX, CTL_EOL); 1609#endif /* IPNOPRIVPORTS */ 1610#if NGRE > 0 1611 sysctl_createv(clog, 0, NULL, NULL, 1612 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1613 CTLTYPE_INT, "grettl", 1614 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1615 NULL, 0, &ip_gre_ttl, 0, 1616 CTL_NET, PF_INET, IPPROTO_IP, 1617 IPCTL_GRE_TTL, CTL_EOL); 1618#endif /* NGRE */ 1619 sysctl_createv(clog, 0, NULL, NULL, 1620 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1621 CTLTYPE_INT, "checkinterface", 1622 SYSCTL_DESCR("Enable receive side of Strong ES model " 1623 "from RFC1122"), 1624 NULL, 0, &ip_checkinterface, 0, 1625 CTL_NET, PF_INET, IPPROTO_IP, 1626 IPCTL_CHECKINTERFACE, CTL_EOL); 1627 sysctl_createv(clog, 0, NULL, NULL, 1628 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1629 CTLTYPE_INT, "random_id", 1630 SYSCTL_DESCR("Assign random ip_id values"), 1631 NULL, 0, &ip_do_randomid, 0, 1632 CTL_NET, PF_INET, IPPROTO_IP, 1633 IPCTL_RANDOMID, CTL_EOL); 1634 sysctl_createv(clog, 0, NULL, NULL, 1635 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1636 CTLTYPE_INT, "do_loopback_cksum", 1637 SYSCTL_DESCR("Perform IP checksum on loopback"), 1638 NULL, 0, &ip_do_loopback_cksum, 0, 1639 CTL_NET, PF_INET, IPPROTO_IP, 1640 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1641 sysctl_createv(clog, 0, NULL, NULL, 1642 CTLFLAG_PERMANENT, 1643 CTLTYPE_STRUCT, "stats", 1644 SYSCTL_DESCR("IP statistics"), 1645 sysctl_net_inet_ip_stats, 0, NULL, 0, 1646 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1647 CTL_EOL); 1648#if NARP 1649 sysctl_createv(clog, 0, NULL, NULL, 1650 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1651 CTLTYPE_INT, "dad_count", 1652 SYSCTL_DESCR("Number of Duplicate Address Detection " 1653 "probes to send"), 1654 NULL, 0, &ip_dad_count, 0, 1655 CTL_NET, PF_INET, IPPROTO_IP, 1656 IPCTL_DAD_COUNT, CTL_EOL); 1657#endif 1658 1659 /* anonportalgo RFC6056 subtree */ 1660 const struct sysctlnode *portalgo_node; 1661 sysctl_createv(clog, 0, NULL, &portalgo_node, 1662 CTLFLAG_PERMANENT, 1663 CTLTYPE_NODE, "anonportalgo", 1664 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1665 NULL, 0, NULL, 0, 1666 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1667 sysctl_createv(clog, 0, &portalgo_node, NULL, 1668 CTLFLAG_PERMANENT, 1669 CTLTYPE_STRING, "available", 1670 SYSCTL_DESCR("available algorithms"), 1671 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1672 CTL_CREATE, CTL_EOL); 1673 sysctl_createv(clog, 0, &portalgo_node, NULL, 1674 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1675 CTLTYPE_STRING, "selected", 1676 SYSCTL_DESCR("selected algorithm"), 1677 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1678 CTL_CREATE, CTL_EOL); 1679 sysctl_createv(clog, 0, &portalgo_node, NULL, 1680 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1681 CTLTYPE_STRUCT, "reserve", 1682 SYSCTL_DESCR("bitmap of reserved ports"), 1683 sysctl_portalgo_reserve4, 0, NULL, 0, 1684 CTL_CREATE, CTL_EOL); 1685} 1686 1687void 1688ip_statinc(u_int stat) 1689{ 1690 1691 KASSERT(stat < IP_NSTATS); 1692 IP_STATINC(stat); 1693} 1694