ip_input.c revision 1.338
1/* $NetBSD: ip_input.c,v 1.338 2016/07/26 08:34:55 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 1.338 2016/07/26 08:34:55 ozaki-r 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 "arp.h" 107 108#include <sys/param.h> 109#include <sys/systm.h> 110#include <sys/cpu.h> 111#include <sys/mbuf.h> 112#include <sys/domain.h> 113#include <sys/protosw.h> 114#include <sys/socket.h> 115#include <sys/socketvar.h> 116#include <sys/errno.h> 117#include <sys/time.h> 118#include <sys/kernel.h> 119#include <sys/pool.h> 120#include <sys/sysctl.h> 121#include <sys/kauth.h> 122 123#include <net/if.h> 124#include <net/if_dl.h> 125#include <net/route.h> 126#include <net/pktqueue.h> 127#include <net/pfil.h> 128 129#include <netinet/in.h> 130#include <netinet/in_systm.h> 131#include <netinet/ip.h> 132#include <netinet/in_pcb.h> 133#include <netinet/in_proto.h> 134#include <netinet/in_var.h> 135#include <netinet/ip_var.h> 136#include <netinet/ip_private.h> 137#include <netinet/ip_icmp.h> 138/* just for gif_ttl */ 139#include <netinet/in_gif.h> 140#include "gif.h" 141#include <net/if_gre.h> 142#include "gre.h" 143 144#ifdef MROUTING 145#include <netinet/ip_mroute.h> 146#endif 147#include <netinet/portalgo.h> 148 149#ifdef IPSEC 150#include <netipsec/ipsec.h> 151#endif 152 153#ifndef IPFORWARDING 154#ifdef GATEWAY 155#define IPFORWARDING 1 /* forward IP packets not for us */ 156#else /* GATEWAY */ 157#define IPFORWARDING 0 /* don't forward IP packets not for us */ 158#endif /* GATEWAY */ 159#endif /* IPFORWARDING */ 160#ifndef IPSENDREDIRECTS 161#define IPSENDREDIRECTS 1 162#endif 163#ifndef IPFORWSRCRT 164#define IPFORWSRCRT 1 /* forward source-routed packets */ 165#endif 166#ifndef IPALLOWSRCRT 167#define IPALLOWSRCRT 1 /* allow source-routed packets */ 168#endif 169#ifndef IPMTUDISC 170#define IPMTUDISC 1 171#endif 172#ifndef IPMTUDISCTIMEOUT 173#define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 174#endif 175 176#ifdef COMPAT_50 177#include <compat/sys/time.h> 178#include <compat/sys/socket.h> 179#endif 180 181/* 182 * Note: DIRECTED_BROADCAST is handled this way so that previous 183 * configuration using this option will Just Work. 184 */ 185#ifndef IPDIRECTEDBCAST 186#ifdef DIRECTED_BROADCAST 187#define IPDIRECTEDBCAST 1 188#else 189#define IPDIRECTEDBCAST 0 190#endif /* DIRECTED_BROADCAST */ 191#endif /* IPDIRECTEDBCAST */ 192int ipforwarding = IPFORWARDING; 193int ipsendredirects = IPSENDREDIRECTS; 194int ip_defttl = IPDEFTTL; 195int ip_forwsrcrt = IPFORWSRCRT; 196int ip_directedbcast = IPDIRECTEDBCAST; 197int ip_allowsrcrt = IPALLOWSRCRT; 198int ip_mtudisc = IPMTUDISC; 199int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 200#ifdef DIAGNOSTIC 201int ipprintfs = 0; 202#endif 203 204int ip_do_randomid = 0; 205 206/* 207 * XXX - Setting ip_checkinterface mostly implements the receive side of 208 * the Strong ES model described in RFC 1122, but since the routing table 209 * and transmit implementation do not implement the Strong ES model, 210 * setting this to 1 results in an odd hybrid. 211 * 212 * XXX - ip_checkinterface currently must be disabled if you use ipnat 213 * to translate the destination address to another local interface. 214 * 215 * XXX - ip_checkinterface must be disabled if you add IP aliases 216 * to the loopback interface instead of the interface where the 217 * packets for those addresses are received. 218 */ 219static int ip_checkinterface __read_mostly = 0; 220 221struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 222 223pktqueue_t * ip_pktq __read_mostly; 224pfil_head_t * inet_pfil_hook __read_mostly; 225ipid_state_t * ip_ids __read_mostly; 226percpu_t * ipstat_percpu __read_mostly; 227 228static struct route ipforward_rt __cacheline_aligned; 229 230uint16_t ip_id; 231 232#ifdef INET_CSUM_COUNTERS 233#include <sys/device.h> 234 235struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 236 NULL, "inet", "hwcsum bad"); 237struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 238 NULL, "inet", "hwcsum ok"); 239struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 240 NULL, "inet", "swcsum"); 241 242#define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 243 244EVCNT_ATTACH_STATIC(ip_hwcsum_bad); 245EVCNT_ATTACH_STATIC(ip_hwcsum_ok); 246EVCNT_ATTACH_STATIC(ip_swcsum); 247 248#else 249 250#define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 251 252#endif /* INET_CSUM_COUNTERS */ 253 254/* 255 * We need to save the IP options in case a protocol wants to respond 256 * to an incoming packet over the same route if the packet got here 257 * using IP source routing. This allows connection establishment and 258 * maintenance when the remote end is on a network that is not known 259 * to us. 260 */ 261 262static int ip_nhops = 0; 263 264static struct ip_srcrt { 265 struct in_addr dst; /* final destination */ 266 char nop; /* one NOP to align */ 267 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 268 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 269} ip_srcrt; 270 271static int ip_drainwanted; 272 273struct sockaddr_in ipaddr = { 274 .sin_len = sizeof(ipaddr), 275 .sin_family = AF_INET, 276}; 277 278static void save_rte(u_char *, struct in_addr); 279 280#ifdef MBUFTRACE 281struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 282struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 283#endif 284 285static void ipintr(void *); 286static void ip_input(struct mbuf *); 287static void ip_forward(struct mbuf *, int, struct ifnet *); 288static bool ip_dooptions(struct mbuf *); 289static struct in_ifaddr *ip_rtaddr(struct in_addr); 290static void sysctl_net_inet_ip_setup(struct sysctllog **); 291 292static struct in_ifaddr *ip_match_our_address(struct ifnet *, struct ip *, 293 int *); 294static struct in_ifaddr *ip_match_our_address_broadcast(struct ifnet *, 295 struct ip *); 296 297/* XXX: Not yet enabled. */ 298#define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock)) 299#define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock)) 300 301/* 302 * IP initialization: fill in IP protocol switch table. 303 * All protocols not implemented in kernel go to raw IP protocol handler. 304 */ 305void 306ip_init(void) 307{ 308 const struct protosw *pr; 309 310 in_init(); 311 sysctl_net_inet_ip_setup(NULL); 312 313 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 314 KASSERT(pr != NULL); 315 316 ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL); 317 KASSERT(ip_pktq != NULL); 318 319 for (u_int i = 0; i < IPPROTO_MAX; i++) { 320 ip_protox[i] = pr - inetsw; 321 } 322 for (pr = inetdomain.dom_protosw; 323 pr < inetdomain.dom_protoswNPROTOSW; pr++) 324 if (pr->pr_domain->dom_family == PF_INET && 325 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 326 ip_protox[pr->pr_protocol] = pr - inetsw; 327 328 ip_reass_init(); 329 330 ip_ids = ip_id_init(); 331 ip_id = time_uptime & 0xfffff; 332 333 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 334#ifdef GATEWAY 335 ipflow_init(); 336#endif 337 338 /* Register our Packet Filter hook. */ 339 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 340 KASSERT(inet_pfil_hook != NULL); 341 342#ifdef MBUFTRACE 343 MOWNER_ATTACH(&ip_tx_mowner); 344 MOWNER_ATTACH(&ip_rx_mowner); 345#endif /* MBUFTRACE */ 346 347 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 348} 349 350static struct in_ifaddr * 351ip_match_our_address(struct ifnet *ifp, struct ip *ip, int *downmatch) 352{ 353 struct in_ifaddr *ia = NULL; 354 int checkif; 355 356 /* 357 * Enable a consistency check between the destination address 358 * and the arrival interface for a unicast packet (the RFC 1122 359 * strong ES model) if IP forwarding is disabled and the packet 360 * is not locally generated. 361 * 362 * XXX - Checking also should be disabled if the destination 363 * address is ipnat'ed to a different interface. 364 * 365 * XXX - Checking is incompatible with IP aliases added 366 * to the loopback interface instead of the interface where 367 * the packets are received. 368 * 369 * XXX - We need to add a per ifaddr flag for this so that 370 * we get finer grain control. 371 */ 372 checkif = ip_checkinterface && (ipforwarding == 0) && 373 (ifp->if_flags & IFF_LOOPBACK) == 0; 374 375 IN_ADDRHASH_READER_FOREACH(ia, ip->ip_dst.s_addr) { 376 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 377 if (ia->ia4_flags & IN_IFF_NOTREADY) 378 continue; 379 if (checkif && ia->ia_ifp != ifp) 380 continue; 381 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 382 break; 383 else 384 (*downmatch)++; 385 } 386 } 387 388 return ia; 389} 390 391static struct in_ifaddr * 392ip_match_our_address_broadcast(struct ifnet *ifp, struct ip *ip) 393{ 394 struct in_ifaddr *ia = NULL; 395 struct ifaddr *ifa; 396 397 IFADDR_READER_FOREACH(ifa, ifp) { 398 if (ifa->ifa_addr->sa_family != AF_INET) 399 continue; 400 ia = ifatoia(ifa); 401 if (ia->ia4_flags & IN_IFF_NOTREADY) 402 continue; 403 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 404 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 405 /* 406 * Look for all-0's host part (old broadcast addr), 407 * either for subnet or net. 408 */ 409 ip->ip_dst.s_addr == ia->ia_subnet || 410 ip->ip_dst.s_addr == ia->ia_net) 411 goto matched; 412 /* 413 * An interface with IP address zero accepts 414 * all packets that arrive on that interface. 415 */ 416 if (in_nullhost(ia->ia_addr.sin_addr)) 417 goto matched; 418 } 419 ia = NULL; 420 421matched: 422 return ia; 423} 424 425/* 426 * IP software interrupt routine. 427 */ 428static void 429ipintr(void *arg __unused) 430{ 431 struct mbuf *m; 432 433 KASSERT(cpu_softintr_p()); 434 435 mutex_enter(softnet_lock); 436 while ((m = pktq_dequeue(ip_pktq)) != NULL) { 437 ip_input(m); 438 } 439 mutex_exit(softnet_lock); 440} 441 442/* 443 * IP input routine. Checksum and byte swap header. If fragmented 444 * try to reassemble. Process options. Pass to next level. 445 */ 446static void 447ip_input(struct mbuf *m) 448{ 449 struct ip *ip = NULL; 450 struct in_ifaddr *ia; 451 int hlen = 0, len; 452 int downmatch; 453 int srcrt = 0; 454 ifnet_t *ifp; 455 struct psref psref; 456 457 KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software " 458 "interrupt handler; synchronization assumptions violated"); 459 460 MCLAIM(m, &ip_rx_mowner); 461 KASSERT((m->m_flags & M_PKTHDR) != 0); 462 463 ifp = m_get_rcvif_psref(m, &psref); 464 if (__predict_false(ifp == NULL)) 465 goto out; 466 467 /* 468 * If no IP addresses have been set yet but the interfaces 469 * are receiving, can't do anything with incoming packets yet. 470 * Note: we pre-check without locks held. 471 */ 472 if (IN_ADDRLIST_READER_EMPTY()) 473 goto out; 474 IP_STATINC(IP_STAT_TOTAL); 475 476 /* 477 * If the IP header is not aligned, slurp it up into a new 478 * mbuf with space for link headers, in the event we forward 479 * it. Otherwise, if it is aligned, make sure the entire 480 * base IP header is in the first mbuf of the chain. 481 */ 482 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 483 if ((m = m_copyup(m, sizeof(struct ip), 484 (max_linkhdr + 3) & ~3)) == NULL) { 485 /* XXXJRT new stat, please */ 486 IP_STATINC(IP_STAT_TOOSMALL); 487 goto out; 488 } 489 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 490 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 491 IP_STATINC(IP_STAT_TOOSMALL); 492 goto out; 493 } 494 } 495 ip = mtod(m, struct ip *); 496 if (ip->ip_v != IPVERSION) { 497 IP_STATINC(IP_STAT_BADVERS); 498 goto out; 499 } 500 hlen = ip->ip_hl << 2; 501 if (hlen < sizeof(struct ip)) { /* minimum header length */ 502 IP_STATINC(IP_STAT_BADHLEN); 503 goto out; 504 } 505 if (hlen > m->m_len) { 506 if ((m = m_pullup(m, hlen)) == NULL) { 507 IP_STATINC(IP_STAT_BADHLEN); 508 goto out; 509 } 510 ip = mtod(m, struct ip *); 511 } 512 513 /* 514 * RFC1122: packets with a multicast source address are 515 * not allowed. 516 */ 517 if (IN_MULTICAST(ip->ip_src.s_addr)) { 518 IP_STATINC(IP_STAT_BADADDR); 519 goto out; 520 } 521 522 /* 127/8 must not appear on wire - RFC1122 */ 523 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 524 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 525 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 526 IP_STATINC(IP_STAT_BADADDR); 527 goto out; 528 } 529 } 530 531 switch (m->m_pkthdr.csum_flags & 532 ((ifp->if_csum_flags_rx & M_CSUM_IPv4) | 533 M_CSUM_IPv4_BAD)) { 534 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 535 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 536 IP_STATINC(IP_STAT_BADSUM); 537 goto out; 538 539 case M_CSUM_IPv4: 540 /* Checksum was okay. */ 541 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 542 break; 543 544 default: 545 /* 546 * Must compute it ourselves. Maybe skip checksum on 547 * loopback interfaces. 548 */ 549 if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) || 550 ip_do_loopback_cksum)) { 551 INET_CSUM_COUNTER_INCR(&ip_swcsum); 552 if (in_cksum(m, hlen) != 0) { 553 IP_STATINC(IP_STAT_BADSUM); 554 goto out; 555 } 556 } 557 break; 558 } 559 560 /* Retrieve the packet length. */ 561 len = ntohs(ip->ip_len); 562 563 /* 564 * Check for additional length bogosity 565 */ 566 if (len < hlen) { 567 IP_STATINC(IP_STAT_BADLEN); 568 goto out; 569 } 570 571 /* 572 * Check that the amount of data in the buffers 573 * is as at least much as the IP header would have us expect. 574 * Trim mbufs if longer than we expect. 575 * Drop packet if shorter than we expect. 576 */ 577 if (m->m_pkthdr.len < len) { 578 IP_STATINC(IP_STAT_TOOSHORT); 579 goto out; 580 } 581 if (m->m_pkthdr.len > len) { 582 if (m->m_len == m->m_pkthdr.len) { 583 m->m_len = len; 584 m->m_pkthdr.len = len; 585 } else 586 m_adj(m, len - m->m_pkthdr.len); 587 } 588 589 /* 590 * Assume that we can create a fast-forward IP flow entry 591 * based on this packet. 592 */ 593 m->m_flags |= M_CANFASTFWD; 594 595 /* 596 * Run through list of hooks for input packets. If there are any 597 * filters which require that additional packets in the flow are 598 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 599 * Note that filters must _never_ set this flag, as another filter 600 * in the list may have previously cleared it. 601 */ 602#if defined(IPSEC) 603 if (!ipsec_used || !ipsec_indone(m)) 604#else 605 if (1) 606#endif 607 { 608 struct in_addr odst = ip->ip_dst; 609 bool freed; 610 611 SOFTNET_LOCK(); 612 freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0; 613 SOFTNET_UNLOCK(); 614 if (freed || m == NULL) { 615 goto out; 616 } 617 ip = mtod(m, struct ip *); 618 hlen = ip->ip_hl << 2; 619 620 /* 621 * XXX The setting of "srcrt" here is to prevent ip_forward() 622 * from generating ICMP redirects for packets that have 623 * been redirected by a hook back out on to the same LAN that 624 * they came from and is not an indication that the packet 625 * is being inffluenced by source routing options. This 626 * allows things like 627 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 628 * where tlp0 is both on the 1.1.1.0/24 network and is the 629 * default route for hosts on 1.1.1.0/24. Of course this 630 * also requires a "map tlp0 ..." to complete the story. 631 * One might argue whether or not this kind of network config. 632 * should be supported in this manner... 633 */ 634 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 635 } 636 637#ifdef ALTQ 638 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 639 if (altq_input) { 640 SOFTNET_LOCK(); 641 if ((*altq_input)(m, AF_INET) == 0) { 642 /* Packet dropped by traffic conditioner. */ 643 SOFTNET_UNLOCK(); 644 goto out; 645 } 646 SOFTNET_UNLOCK(); 647 } 648#endif 649 650 /* 651 * Process options and, if not destined for us, 652 * ship it on. ip_dooptions returns 1 when an 653 * error was detected (causing an icmp message 654 * to be sent and the original packet to be freed). 655 */ 656 ip_nhops = 0; /* for source routed packets */ 657 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 658 goto out; 659 660 /* 661 * Check our list of addresses, to see if the packet is for us. 662 * 663 * Traditional 4.4BSD did not consult IFF_UP at all. 664 * The behavior here is to treat addresses on !IFF_UP interface 665 * or IN_IFF_NOTREADY addresses as not mine. 666 */ 667 downmatch = 0; 668 ia = ip_match_our_address(ifp, ip, &downmatch); 669 if (ia != NULL) 670 goto ours; 671 672 if (ifp->if_flags & IFF_BROADCAST) { 673 ia = ip_match_our_address_broadcast(ifp, ip); 674 if (ia != NULL) 675 goto ours; 676 } 677 678 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 679#ifdef MROUTING 680 extern struct socket *ip_mrouter; 681 682 if (ip_mrouter) { 683 /* 684 * If we are acting as a multicast router, all 685 * incoming multicast packets are passed to the 686 * kernel-level multicast forwarding function. 687 * The packet is returned (relatively) intact; if 688 * ip_mforward() returns a non-zero value, the packet 689 * must be discarded, else it may be accepted below. 690 * 691 * (The IP ident field is put in the same byte order 692 * as expected when ip_mforward() is called from 693 * ip_output().) 694 */ 695 SOFTNET_LOCK(); 696 if (ip_mforward(m, ifp) != 0) { 697 SOFTNET_UNLOCK(); 698 IP_STATINC(IP_STAT_CANTFORWARD); 699 goto out; 700 } 701 SOFTNET_UNLOCK(); 702 703 /* 704 * The process-level routing demon needs to receive 705 * all multicast IGMP packets, whether or not this 706 * host belongs to their destination groups. 707 */ 708 if (ip->ip_p == IPPROTO_IGMP) { 709 goto ours; 710 } 711 IP_STATINC(IP_STAT_CANTFORWARD); 712 } 713#endif 714 /* 715 * See if we belong to the destination multicast group on the 716 * arrival interface. 717 */ 718 if (!in_multi_group(ip->ip_dst, ifp, 0)) { 719 IP_STATINC(IP_STAT_CANTFORWARD); 720 goto out; 721 } 722 goto ours; 723 } 724 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 725 in_nullhost(ip->ip_dst)) 726 goto ours; 727 728 /* 729 * Not for us; forward if possible and desirable. 730 */ 731 if (ipforwarding == 0) { 732 m_put_rcvif_psref(ifp, &psref); 733 IP_STATINC(IP_STAT_CANTFORWARD); 734 m_freem(m); 735 } else { 736 /* 737 * If ip_dst matched any of my address on !IFF_UP interface, 738 * and there's no IFF_UP interface that matches ip_dst, 739 * send icmp unreach. Forwarding it will result in in-kernel 740 * forwarding loop till TTL goes to 0. 741 */ 742 if (downmatch) { 743 m_put_rcvif_psref(ifp, &psref); 744 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 745 IP_STATINC(IP_STAT_CANTFORWARD); 746 return; 747 } 748#ifdef IPSEC 749 /* Perform IPsec, if any. */ 750 if (ipsec_used) { 751 SOFTNET_LOCK(); 752 if (ipsec4_input(m, IP_FORWARDING | 753 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) { 754 SOFTNET_UNLOCK(); 755 goto out; 756 } 757 SOFTNET_UNLOCK(); 758 } 759#endif 760 ip_forward(m, srcrt, ifp); 761 m_put_rcvif_psref(ifp, &psref); 762 } 763 return; 764 765ours: 766 m_put_rcvif_psref(ifp, &psref); 767 ifp = NULL; 768 769 /* 770 * If offset or IP_MF are set, must reassemble. 771 */ 772 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 773 /* 774 * Pass to IP reassembly mechanism. 775 */ 776 if (ip_reass_packet(&m, ip) != 0) { 777 /* Failed; invalid fragment(s) or packet. */ 778 goto out; 779 } 780 if (m == NULL) { 781 /* More fragments should come; silently return. */ 782 goto out; 783 } 784 /* 785 * Reassembly is done, we have the final packet. 786 * Updated cached data in local variable(s). 787 */ 788 ip = mtod(m, struct ip *); 789 hlen = ip->ip_hl << 2; 790 } 791 792#ifdef IPSEC 793 /* 794 * Enforce IPsec policy checking if we are seeing last header. 795 * Note that we do not visit this with protocols with PCB layer 796 * code - like UDP/TCP/raw IP. 797 */ 798 if (ipsec_used && 799 (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 800 SOFTNET_LOCK(); 801 if (ipsec4_input(m, 0) != 0) { 802 SOFTNET_UNLOCK(); 803 goto out; 804 } 805 SOFTNET_UNLOCK(); 806 } 807#endif 808 809 /* 810 * Switch out to protocol's input routine. 811 */ 812#if IFA_STATS 813 if (ia && ip) 814 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 815#endif 816 IP_STATINC(IP_STAT_DELIVERED); 817 818 const int off = hlen, nh = ip->ip_p; 819 820 SOFTNET_LOCK(); 821 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 822 SOFTNET_UNLOCK(); 823 return; 824 825out: 826 m_put_rcvif_psref(ifp, &psref); 827 if (m != NULL) 828 m_freem(m); 829} 830 831/* 832 * IP timer processing. 833 */ 834void 835ip_slowtimo(void) 836{ 837 838 mutex_enter(softnet_lock); 839 KERNEL_LOCK(1, NULL); 840 841 ip_reass_slowtimo(); 842 843 KERNEL_UNLOCK_ONE(NULL); 844 mutex_exit(softnet_lock); 845} 846 847/* 848 * IP drain processing. 849 */ 850void 851ip_drain(void) 852{ 853 854 KERNEL_LOCK(1, NULL); 855 ip_reass_drain(); 856 KERNEL_UNLOCK_ONE(NULL); 857} 858 859/* 860 * ip_dooptions: perform option processing on a datagram, possibly discarding 861 * it if bad options are encountered, or forwarding it if source-routed. 862 * 863 * => Returns true if packet has been forwarded/freed. 864 * => Returns false if the packet should be processed further. 865 */ 866static bool 867ip_dooptions(struct mbuf *m) 868{ 869 struct ip *ip = mtod(m, struct ip *); 870 u_char *cp, *cp0; 871 struct ip_timestamp *ipt; 872 struct in_ifaddr *ia; 873 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 874 struct in_addr dst; 875 n_time ntime; 876 877 dst = ip->ip_dst; 878 cp = (u_char *)(ip + 1); 879 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 880 for (; cnt > 0; cnt -= optlen, cp += optlen) { 881 opt = cp[IPOPT_OPTVAL]; 882 if (opt == IPOPT_EOL) 883 break; 884 if (opt == IPOPT_NOP) 885 optlen = 1; 886 else { 887 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 888 code = &cp[IPOPT_OLEN] - (u_char *)ip; 889 goto bad; 890 } 891 optlen = cp[IPOPT_OLEN]; 892 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 893 code = &cp[IPOPT_OLEN] - (u_char *)ip; 894 goto bad; 895 } 896 } 897 switch (opt) { 898 899 default: 900 break; 901 902 /* 903 * Source routing with record. 904 * Find interface with current destination address. 905 * If none on this machine then drop if strictly routed, 906 * or do nothing if loosely routed. 907 * Record interface address and bring up next address 908 * component. If strictly routed make sure next 909 * address is on directly accessible net. 910 */ 911 case IPOPT_LSRR: 912 case IPOPT_SSRR: 913 if (ip_allowsrcrt == 0) { 914 type = ICMP_UNREACH; 915 code = ICMP_UNREACH_NET_PROHIB; 916 goto bad; 917 } 918 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 919 code = &cp[IPOPT_OLEN] - (u_char *)ip; 920 goto bad; 921 } 922 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 923 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 924 goto bad; 925 } 926 ipaddr.sin_addr = ip->ip_dst; 927 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 928 if (ia == 0) { 929 if (opt == IPOPT_SSRR) { 930 type = ICMP_UNREACH; 931 code = ICMP_UNREACH_SRCFAIL; 932 goto bad; 933 } 934 /* 935 * Loose routing, and not at next destination 936 * yet; nothing to do except forward. 937 */ 938 break; 939 } 940 off--; /* 0 origin */ 941 if ((off + sizeof(struct in_addr)) > optlen) { 942 /* 943 * End of source route. Should be for us. 944 */ 945 save_rte(cp, ip->ip_src); 946 break; 947 } 948 /* 949 * locate outgoing interface 950 */ 951 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 952 sizeof(ipaddr.sin_addr)); 953 if (opt == IPOPT_SSRR) 954 ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); 955 else 956 ia = ip_rtaddr(ipaddr.sin_addr); 957 if (ia == 0) { 958 type = ICMP_UNREACH; 959 code = ICMP_UNREACH_SRCFAIL; 960 goto bad; 961 } 962 ip->ip_dst = ipaddr.sin_addr; 963 bcopy((void *)&ia->ia_addr.sin_addr, 964 (void *)(cp + off), sizeof(struct in_addr)); 965 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 966 /* 967 * Let ip_intr's mcast routing check handle mcast pkts 968 */ 969 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 970 break; 971 972 case IPOPT_RR: 973 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 974 code = &cp[IPOPT_OLEN] - (u_char *)ip; 975 goto bad; 976 } 977 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 978 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 979 goto bad; 980 } 981 /* 982 * If no space remains, ignore. 983 */ 984 off--; /* 0 origin */ 985 if ((off + sizeof(struct in_addr)) > optlen) 986 break; 987 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 988 sizeof(ipaddr.sin_addr)); 989 /* 990 * locate outgoing interface; if we're the destination, 991 * use the incoming interface (should be same). 992 */ 993 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 994 == NULL && 995 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 996 type = ICMP_UNREACH; 997 code = ICMP_UNREACH_HOST; 998 goto bad; 999 } 1000 bcopy((void *)&ia->ia_addr.sin_addr, 1001 (void *)(cp + off), sizeof(struct in_addr)); 1002 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1003 break; 1004 1005 case IPOPT_TS: 1006 code = cp - (u_char *)ip; 1007 ipt = (struct ip_timestamp *)cp; 1008 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1009 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1010 goto bad; 1011 } 1012 if (ipt->ipt_ptr < 5) { 1013 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1014 goto bad; 1015 } 1016 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1017 if (++ipt->ipt_oflw == 0) { 1018 code = (u_char *)&ipt->ipt_ptr - 1019 (u_char *)ip; 1020 goto bad; 1021 } 1022 break; 1023 } 1024 cp0 = (cp + ipt->ipt_ptr - 1); 1025 switch (ipt->ipt_flg) { 1026 1027 case IPOPT_TS_TSONLY: 1028 break; 1029 1030 case IPOPT_TS_TSANDADDR: { 1031 struct ifnet *rcvif; 1032 int s; 1033 1034 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1035 sizeof(struct in_addr) > ipt->ipt_len) { 1036 code = (u_char *)&ipt->ipt_ptr - 1037 (u_char *)ip; 1038 goto bad; 1039 } 1040 ipaddr.sin_addr = dst; 1041 rcvif = m_get_rcvif(m, &s); 1042 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1043 rcvif)); 1044 m_put_rcvif(rcvif, &s); 1045 if (ia == 0) 1046 continue; 1047 bcopy(&ia->ia_addr.sin_addr, 1048 cp0, sizeof(struct in_addr)); 1049 ipt->ipt_ptr += sizeof(struct in_addr); 1050 break; 1051 } 1052 1053 case IPOPT_TS_PRESPEC: 1054 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1055 sizeof(struct in_addr) > ipt->ipt_len) { 1056 code = (u_char *)&ipt->ipt_ptr - 1057 (u_char *)ip; 1058 goto bad; 1059 } 1060 memcpy(&ipaddr.sin_addr, cp0, 1061 sizeof(struct in_addr)); 1062 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1063 == NULL) 1064 continue; 1065 ipt->ipt_ptr += sizeof(struct in_addr); 1066 break; 1067 1068 default: 1069 /* XXX can't take &ipt->ipt_flg */ 1070 code = (u_char *)&ipt->ipt_ptr - 1071 (u_char *)ip + 1; 1072 goto bad; 1073 } 1074 ntime = iptime(); 1075 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1076 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1077 sizeof(n_time)); 1078 ipt->ipt_ptr += sizeof(n_time); 1079 } 1080 } 1081 if (forward) { 1082 struct ifnet *rcvif; 1083 struct psref psref; 1084 1085 if (ip_forwsrcrt == 0) { 1086 type = ICMP_UNREACH; 1087 code = ICMP_UNREACH_SRCFAIL; 1088 goto bad; 1089 } 1090 1091 rcvif = m_get_rcvif_psref(m, &psref); 1092 if (__predict_false(rcvif == NULL)) { 1093 type = ICMP_UNREACH; 1094 code = ICMP_UNREACH_HOST; 1095 goto bad; 1096 } 1097 ip_forward(m, 1, rcvif); 1098 m_put_rcvif_psref(rcvif, &psref); 1099 return true; 1100 } 1101 return false; 1102bad: 1103 icmp_error(m, type, code, 0, 0); 1104 IP_STATINC(IP_STAT_BADOPTIONS); 1105 return true; 1106} 1107 1108/* 1109 * ip_rtaddr: given address of next destination (final or next hop), 1110 * return internet address info of interface to be used to get there. 1111 */ 1112static struct in_ifaddr * 1113ip_rtaddr(struct in_addr dst) 1114{ 1115 struct rtentry *rt; 1116 union { 1117 struct sockaddr dst; 1118 struct sockaddr_in dst4; 1119 } u; 1120 1121 sockaddr_in_init(&u.dst4, &dst, 0); 1122 1123 SOFTNET_LOCK(); 1124 rt = rtcache_lookup(&ipforward_rt, &u.dst); 1125 SOFTNET_UNLOCK(); 1126 if (rt == NULL) 1127 return NULL; 1128 1129 return ifatoia(rt->rt_ifa); 1130} 1131 1132/* 1133 * save_rte: save incoming source route for use in replies, to be picked 1134 * up later by ip_srcroute if the receiver is interested. 1135 */ 1136static void 1137save_rte(u_char *option, struct in_addr dst) 1138{ 1139 unsigned olen; 1140 1141 olen = option[IPOPT_OLEN]; 1142 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1143 return; 1144 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1145 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1146 ip_srcrt.dst = dst; 1147} 1148 1149/* 1150 * Retrieve incoming source route for use in replies, 1151 * in the same form used by setsockopt. 1152 * The first hop is placed before the options, will be removed later. 1153 */ 1154struct mbuf * 1155ip_srcroute(void) 1156{ 1157 struct in_addr *p, *q; 1158 struct mbuf *m; 1159 1160 if (ip_nhops == 0) 1161 return NULL; 1162 m = m_get(M_DONTWAIT, MT_SOOPTS); 1163 if (m == 0) 1164 return NULL; 1165 1166 MCLAIM(m, &inetdomain.dom_mowner); 1167#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1168 1169 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1170 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1171 OPTSIZ; 1172 1173 /* 1174 * First save first hop for return route 1175 */ 1176 p = &ip_srcrt.route[ip_nhops - 1]; 1177 *(mtod(m, struct in_addr *)) = *p--; 1178 1179 /* 1180 * Copy option fields and padding (nop) to mbuf. 1181 */ 1182 ip_srcrt.nop = IPOPT_NOP; 1183 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1184 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1185 OPTSIZ); 1186 q = (struct in_addr *)(mtod(m, char *) + 1187 sizeof(struct in_addr) + OPTSIZ); 1188#undef OPTSIZ 1189 /* 1190 * Record return path as an IP source route, 1191 * reversing the path (pointers are now aligned). 1192 */ 1193 while (p >= ip_srcrt.route) { 1194 *q++ = *p--; 1195 } 1196 /* 1197 * Last hop goes to final destination. 1198 */ 1199 *q = ip_srcrt.dst; 1200 return (m); 1201} 1202 1203const int inetctlerrmap[PRC_NCMDS] = { 1204 [PRC_MSGSIZE] = EMSGSIZE, 1205 [PRC_HOSTDEAD] = EHOSTDOWN, 1206 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1207 [PRC_UNREACH_NET] = EHOSTUNREACH, 1208 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1209 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1210 [PRC_UNREACH_PORT] = ECONNREFUSED, 1211 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1212 [PRC_PARAMPROB] = ENOPROTOOPT, 1213}; 1214 1215void 1216ip_fasttimo(void) 1217{ 1218 if (ip_drainwanted) { 1219 ip_drain(); 1220 ip_drainwanted = 0; 1221 } 1222} 1223 1224void 1225ip_drainstub(void) 1226{ 1227 ip_drainwanted = 1; 1228} 1229 1230/* 1231 * Forward a packet. If some error occurs return the sender 1232 * an icmp packet. Note we can't always generate a meaningful 1233 * icmp message because icmp doesn't have a large enough repertoire 1234 * of codes and types. 1235 * 1236 * If not forwarding, just drop the packet. This could be confusing 1237 * if ipforwarding was zero but some routing protocol was advancing 1238 * us as a gateway to somewhere. However, we must let the routing 1239 * protocol deal with that. 1240 * 1241 * The srcrt parameter indicates whether the packet is being forwarded 1242 * via a source route. 1243 */ 1244static void 1245ip_forward(struct mbuf *m, int srcrt, struct ifnet *rcvif) 1246{ 1247 struct ip *ip = mtod(m, struct ip *); 1248 struct rtentry *rt; 1249 int error, type = 0, code = 0, destmtu = 0; 1250 struct mbuf *mcopy; 1251 n_long dest; 1252 union { 1253 struct sockaddr dst; 1254 struct sockaddr_in dst4; 1255 } u; 1256 uint64_t *ips; 1257 1258 KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software " 1259 "interrupt handler; synchronization assumptions violated"); 1260 1261 /* 1262 * We are now in the output path. 1263 */ 1264 MCLAIM(m, &ip_tx_mowner); 1265 1266 /* 1267 * Clear any in-bound checksum flags for this packet. 1268 */ 1269 m->m_pkthdr.csum_flags = 0; 1270 1271 dest = 0; 1272 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1273 IP_STATINC(IP_STAT_CANTFORWARD); 1274 m_freem(m); 1275 return; 1276 } 1277 1278 SOFTNET_LOCK(); 1279 1280 if (ip->ip_ttl <= IPTTLDEC) { 1281 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1282 SOFTNET_UNLOCK(); 1283 return; 1284 } 1285 1286 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1287 1288 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1289 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1290 SOFTNET_UNLOCK(); 1291 return; 1292 } 1293 1294 /* 1295 * Save at most 68 bytes of the packet in case 1296 * we need to generate an ICMP message to the src. 1297 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1298 */ 1299 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1300 if (mcopy) 1301 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1302 1303 ip->ip_ttl -= IPTTLDEC; 1304 1305 /* 1306 * If forwarding packet using same interface that it came in on, 1307 * perhaps should send a redirect to sender to shortcut a hop. 1308 * Only send redirect if source is sending directly to us, 1309 * and if packet was not source routed (or has any options). 1310 * Also, don't send redirect if forwarding using a default route 1311 * or a route modified by a redirect. 1312 */ 1313 if (rt->rt_ifp == rcvif && 1314 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1315 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1316 ipsendredirects && !srcrt) { 1317 if (rt->rt_ifa && 1318 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1319 ifatoia(rt->rt_ifa)->ia_subnet) { 1320 if (rt->rt_flags & RTF_GATEWAY) 1321 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1322 else 1323 dest = ip->ip_dst.s_addr; 1324 /* 1325 * Router requirements says to only send host 1326 * redirects. 1327 */ 1328 type = ICMP_REDIRECT; 1329 code = ICMP_REDIRECT_HOST; 1330 } 1331 } 1332 1333 error = ip_output(m, NULL, &ipforward_rt, 1334 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1335 NULL, NULL); 1336 1337 if (error) { 1338 IP_STATINC(IP_STAT_CANTFORWARD); 1339 goto error; 1340 } 1341 1342 ips = IP_STAT_GETREF(); 1343 ips[IP_STAT_FORWARD]++; 1344 1345 if (type) { 1346 ips[IP_STAT_REDIRECTSENT]++; 1347 IP_STAT_PUTREF(); 1348 goto redirect; 1349 } 1350 1351 IP_STAT_PUTREF(); 1352 if (mcopy) { 1353#ifdef GATEWAY 1354 if (mcopy->m_flags & M_CANFASTFWD) 1355 ipflow_create(&ipforward_rt, mcopy); 1356#endif 1357 m_freem(mcopy); 1358 } 1359 1360 SOFTNET_UNLOCK(); 1361 return; 1362 1363redirect: 1364error: 1365 if (mcopy == NULL) { 1366 SOFTNET_UNLOCK(); 1367 return; 1368 } 1369 1370 switch (error) { 1371 1372 case 0: /* forwarded, but need redirect */ 1373 /* type, code set above */ 1374 break; 1375 1376 case ENETUNREACH: /* shouldn't happen, checked above */ 1377 case EHOSTUNREACH: 1378 case ENETDOWN: 1379 case EHOSTDOWN: 1380 default: 1381 type = ICMP_UNREACH; 1382 code = ICMP_UNREACH_HOST; 1383 break; 1384 1385 case EMSGSIZE: 1386 type = ICMP_UNREACH; 1387 code = ICMP_UNREACH_NEEDFRAG; 1388 1389 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1390 destmtu = rt->rt_ifp->if_mtu; 1391#ifdef IPSEC 1392 if (ipsec_used) 1393 (void)ipsec4_forward(mcopy, &destmtu); 1394#endif 1395 IP_STATINC(IP_STAT_CANTFRAG); 1396 break; 1397 1398 case ENOBUFS: 1399 /* 1400 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1401 * Requirements for IP Version 4 Routers. Source quench can 1402 * big problem under DoS attacks or if the underlying 1403 * interface is rate-limited. 1404 */ 1405 if (mcopy) 1406 m_freem(mcopy); 1407 SOFTNET_UNLOCK(); 1408 return; 1409 } 1410 icmp_error(mcopy, type, code, dest, destmtu); 1411 SOFTNET_UNLOCK(); 1412} 1413 1414void 1415ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1416 struct mbuf *m) 1417{ 1418 struct socket *so = inp->inp_socket; 1419 ifnet_t *ifp; 1420 int inpflags = inp->inp_flags; 1421 struct psref psref; 1422 1423 ifp = m_get_rcvif_psref(m, &psref); 1424 if (__predict_false(ifp == NULL)) 1425 return; /* XXX should report error? */ 1426 1427 if (so->so_options & SO_TIMESTAMP 1428#ifdef SO_OTIMESTAMP 1429 || so->so_options & SO_OTIMESTAMP 1430#endif 1431 ) { 1432 struct timeval tv; 1433 1434 microtime(&tv); 1435#ifdef SO_OTIMESTAMP 1436 if (so->so_options & SO_OTIMESTAMP) { 1437 struct timeval50 tv50; 1438 timeval_to_timeval50(&tv, &tv50); 1439 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1440 SCM_OTIMESTAMP, SOL_SOCKET); 1441 } else 1442#endif 1443 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1444 SCM_TIMESTAMP, SOL_SOCKET); 1445 if (*mp) 1446 mp = &(*mp)->m_next; 1447 } 1448 if (inpflags & INP_RECVDSTADDR) { 1449 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1450 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1451 if (*mp) 1452 mp = &(*mp)->m_next; 1453 } 1454 if (inpflags & INP_RECVPKTINFO) { 1455 struct in_pktinfo ipi; 1456 ipi.ipi_addr = ip->ip_src; 1457 ipi.ipi_ifindex = ifp->if_index; 1458 *mp = sbcreatecontrol((void *) &ipi, 1459 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1460 if (*mp) 1461 mp = &(*mp)->m_next; 1462 } 1463 if (inpflags & INP_PKTINFO) { 1464 struct in_pktinfo ipi; 1465 ipi.ipi_addr = ip->ip_dst; 1466 ipi.ipi_ifindex = ifp->if_index; 1467 *mp = sbcreatecontrol((void *) &ipi, 1468 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1469 if (*mp) 1470 mp = &(*mp)->m_next; 1471 } 1472 if (inpflags & INP_RECVIF) { 1473 struct sockaddr_dl sdl; 1474 1475 sockaddr_dl_init(&sdl, sizeof(sdl), ifp->if_index, 0, NULL, 0, 1476 NULL, 0); 1477 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1478 if (*mp) 1479 mp = &(*mp)->m_next; 1480 } 1481 if (inpflags & INP_RECVTTL) { 1482 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1483 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1484 if (*mp) 1485 mp = &(*mp)->m_next; 1486 } 1487 m_put_rcvif_psref(ifp, &psref); 1488} 1489 1490/* 1491 * sysctl helper routine for net.inet.ip.forwsrcrt. 1492 */ 1493static int 1494sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1495{ 1496 int error, tmp; 1497 struct sysctlnode node; 1498 1499 node = *rnode; 1500 tmp = ip_forwsrcrt; 1501 node.sysctl_data = &tmp; 1502 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1503 if (error || newp == NULL) 1504 return (error); 1505 1506 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1507 0, NULL, NULL, NULL); 1508 if (error) 1509 return (error); 1510 1511 ip_forwsrcrt = tmp; 1512 1513 return (0); 1514} 1515 1516/* 1517 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1518 * range of the new value and tweaks timers if it changes. 1519 */ 1520static int 1521sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1522{ 1523 int error, tmp; 1524 struct sysctlnode node; 1525 1526 node = *rnode; 1527 tmp = ip_mtudisc_timeout; 1528 node.sysctl_data = &tmp; 1529 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1530 if (error || newp == NULL) 1531 return (error); 1532 if (tmp < 0) 1533 return (EINVAL); 1534 1535 mutex_enter(softnet_lock); 1536 1537 ip_mtudisc_timeout = tmp; 1538 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1539 1540 mutex_exit(softnet_lock); 1541 1542 return (0); 1543} 1544 1545static int 1546sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1547{ 1548 1549 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1550} 1551 1552static void 1553sysctl_net_inet_ip_setup(struct sysctllog **clog) 1554{ 1555 sysctl_createv(clog, 0, NULL, NULL, 1556 CTLFLAG_PERMANENT, 1557 CTLTYPE_NODE, "inet", 1558 SYSCTL_DESCR("PF_INET related settings"), 1559 NULL, 0, NULL, 0, 1560 CTL_NET, PF_INET, CTL_EOL); 1561 sysctl_createv(clog, 0, NULL, NULL, 1562 CTLFLAG_PERMANENT, 1563 CTLTYPE_NODE, "ip", 1564 SYSCTL_DESCR("IPv4 related settings"), 1565 NULL, 0, NULL, 0, 1566 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1567 1568 sysctl_createv(clog, 0, NULL, NULL, 1569 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1570 CTLTYPE_INT, "forwarding", 1571 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1572 NULL, 0, &ipforwarding, 0, 1573 CTL_NET, PF_INET, IPPROTO_IP, 1574 IPCTL_FORWARDING, CTL_EOL); 1575 sysctl_createv(clog, 0, NULL, NULL, 1576 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1577 CTLTYPE_INT, "redirect", 1578 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1579 NULL, 0, &ipsendredirects, 0, 1580 CTL_NET, PF_INET, IPPROTO_IP, 1581 IPCTL_SENDREDIRECTS, CTL_EOL); 1582 sysctl_createv(clog, 0, NULL, NULL, 1583 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1584 CTLTYPE_INT, "ttl", 1585 SYSCTL_DESCR("Default TTL for an INET datagram"), 1586 NULL, 0, &ip_defttl, 0, 1587 CTL_NET, PF_INET, IPPROTO_IP, 1588 IPCTL_DEFTTL, CTL_EOL); 1589#ifdef IPCTL_DEFMTU 1590 sysctl_createv(clog, 0, NULL, NULL, 1591 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1592 CTLTYPE_INT, "mtu", 1593 SYSCTL_DESCR("Default MTA for an INET route"), 1594 NULL, 0, &ip_mtu, 0, 1595 CTL_NET, PF_INET, IPPROTO_IP, 1596 IPCTL_DEFMTU, CTL_EOL); 1597#endif /* IPCTL_DEFMTU */ 1598 sysctl_createv(clog, 0, NULL, NULL, 1599 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1600 CTLTYPE_INT, "forwsrcrt", 1601 SYSCTL_DESCR("Enable forwarding of source-routed " 1602 "datagrams"), 1603 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1604 CTL_NET, PF_INET, IPPROTO_IP, 1605 IPCTL_FORWSRCRT, CTL_EOL); 1606 sysctl_createv(clog, 0, NULL, NULL, 1607 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1608 CTLTYPE_INT, "directed-broadcast", 1609 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1610 NULL, 0, &ip_directedbcast, 0, 1611 CTL_NET, PF_INET, IPPROTO_IP, 1612 IPCTL_DIRECTEDBCAST, CTL_EOL); 1613 sysctl_createv(clog, 0, NULL, NULL, 1614 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1615 CTLTYPE_INT, "allowsrcrt", 1616 SYSCTL_DESCR("Accept source-routed datagrams"), 1617 NULL, 0, &ip_allowsrcrt, 0, 1618 CTL_NET, PF_INET, IPPROTO_IP, 1619 IPCTL_ALLOWSRCRT, CTL_EOL); 1620 1621 sysctl_createv(clog, 0, NULL, NULL, 1622 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1623 CTLTYPE_INT, "mtudisc", 1624 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1625 NULL, 0, &ip_mtudisc, 0, 1626 CTL_NET, PF_INET, IPPROTO_IP, 1627 IPCTL_MTUDISC, CTL_EOL); 1628 sysctl_createv(clog, 0, NULL, NULL, 1629 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1630 CTLTYPE_INT, "anonportmin", 1631 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1632 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1633 CTL_NET, PF_INET, IPPROTO_IP, 1634 IPCTL_ANONPORTMIN, CTL_EOL); 1635 sysctl_createv(clog, 0, NULL, NULL, 1636 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1637 CTLTYPE_INT, "anonportmax", 1638 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1639 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1640 CTL_NET, PF_INET, IPPROTO_IP, 1641 IPCTL_ANONPORTMAX, CTL_EOL); 1642 sysctl_createv(clog, 0, NULL, NULL, 1643 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1644 CTLTYPE_INT, "mtudisctimeout", 1645 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1646 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1647 CTL_NET, PF_INET, IPPROTO_IP, 1648 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1649#ifndef IPNOPRIVPORTS 1650 sysctl_createv(clog, 0, NULL, NULL, 1651 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1652 CTLTYPE_INT, "lowportmin", 1653 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1654 "to assign"), 1655 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1656 CTL_NET, PF_INET, IPPROTO_IP, 1657 IPCTL_LOWPORTMIN, CTL_EOL); 1658 sysctl_createv(clog, 0, NULL, NULL, 1659 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1660 CTLTYPE_INT, "lowportmax", 1661 SYSCTL_DESCR("Highest privileged ephemeral port number " 1662 "to assign"), 1663 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1664 CTL_NET, PF_INET, IPPROTO_IP, 1665 IPCTL_LOWPORTMAX, CTL_EOL); 1666#endif /* IPNOPRIVPORTS */ 1667#if NGRE > 0 1668 sysctl_createv(clog, 0, NULL, NULL, 1669 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1670 CTLTYPE_INT, "grettl", 1671 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1672 NULL, 0, &ip_gre_ttl, 0, 1673 CTL_NET, PF_INET, IPPROTO_IP, 1674 IPCTL_GRE_TTL, CTL_EOL); 1675#endif /* NGRE */ 1676 sysctl_createv(clog, 0, NULL, NULL, 1677 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1678 CTLTYPE_INT, "checkinterface", 1679 SYSCTL_DESCR("Enable receive side of Strong ES model " 1680 "from RFC1122"), 1681 NULL, 0, &ip_checkinterface, 0, 1682 CTL_NET, PF_INET, IPPROTO_IP, 1683 IPCTL_CHECKINTERFACE, CTL_EOL); 1684 sysctl_createv(clog, 0, NULL, NULL, 1685 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1686 CTLTYPE_INT, "random_id", 1687 SYSCTL_DESCR("Assign random ip_id values"), 1688 NULL, 0, &ip_do_randomid, 0, 1689 CTL_NET, PF_INET, IPPROTO_IP, 1690 IPCTL_RANDOMID, CTL_EOL); 1691 sysctl_createv(clog, 0, NULL, NULL, 1692 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1693 CTLTYPE_INT, "do_loopback_cksum", 1694 SYSCTL_DESCR("Perform IP checksum on loopback"), 1695 NULL, 0, &ip_do_loopback_cksum, 0, 1696 CTL_NET, PF_INET, IPPROTO_IP, 1697 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1698 sysctl_createv(clog, 0, NULL, NULL, 1699 CTLFLAG_PERMANENT, 1700 CTLTYPE_STRUCT, "stats", 1701 SYSCTL_DESCR("IP statistics"), 1702 sysctl_net_inet_ip_stats, 0, NULL, 0, 1703 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1704 CTL_EOL); 1705#if NARP 1706 sysctl_createv(clog, 0, NULL, NULL, 1707 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1708 CTLTYPE_INT, "dad_count", 1709 SYSCTL_DESCR("Number of Duplicate Address Detection " 1710 "probes to send"), 1711 NULL, 0, &ip_dad_count, 0, 1712 CTL_NET, PF_INET, IPPROTO_IP, 1713 IPCTL_DAD_COUNT, CTL_EOL); 1714#endif 1715 1716 /* anonportalgo RFC6056 subtree */ 1717 const struct sysctlnode *portalgo_node; 1718 sysctl_createv(clog, 0, NULL, &portalgo_node, 1719 CTLFLAG_PERMANENT, 1720 CTLTYPE_NODE, "anonportalgo", 1721 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1722 NULL, 0, NULL, 0, 1723 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1724 sysctl_createv(clog, 0, &portalgo_node, NULL, 1725 CTLFLAG_PERMANENT, 1726 CTLTYPE_STRING, "available", 1727 SYSCTL_DESCR("available algorithms"), 1728 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1729 CTL_CREATE, CTL_EOL); 1730 sysctl_createv(clog, 0, &portalgo_node, NULL, 1731 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1732 CTLTYPE_STRING, "selected", 1733 SYSCTL_DESCR("selected algorithm"), 1734 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1735 CTL_CREATE, CTL_EOL); 1736 sysctl_createv(clog, 0, &portalgo_node, NULL, 1737 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1738 CTLTYPE_STRUCT, "reserve", 1739 SYSCTL_DESCR("bitmap of reserved ports"), 1740 sysctl_portalgo_reserve4, 0, NULL, 0, 1741 CTL_CREATE, CTL_EOL); 1742} 1743 1744void 1745ip_statinc(u_int stat) 1746{ 1747 1748 KASSERT(stat < IP_NSTATS); 1749 IP_STATINC(stat); 1750} 1751