ip_input.c revision 1.329
1/* $NetBSD: ip_input.c,v 1.329 2016/06/10 13:31:44 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 1.329 2016/06/10 13:31:44 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 **); 291 292/* XXX: Not yet enabled. */ 293#define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock)) 294#define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock)) 295 296/* 297 * IP initialization: fill in IP protocol switch table. 298 * All protocols not implemented in kernel go to raw IP protocol handler. 299 */ 300void 301ip_init(void) 302{ 303 const struct protosw *pr; 304 305 in_init(); 306 sysctl_net_inet_ip_setup(NULL); 307 308 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 309 KASSERT(pr != NULL); 310 311 ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL); 312 KASSERT(ip_pktq != NULL); 313 314 for (u_int i = 0; i < IPPROTO_MAX; i++) { 315 ip_protox[i] = pr - inetsw; 316 } 317 for (pr = inetdomain.dom_protosw; 318 pr < inetdomain.dom_protoswNPROTOSW; pr++) 319 if (pr->pr_domain->dom_family == PF_INET && 320 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 321 ip_protox[pr->pr_protocol] = pr - inetsw; 322 323 ip_reass_init(); 324 325 ip_ids = ip_id_init(); 326 ip_id = time_uptime & 0xfffff; 327 328 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 329#ifdef GATEWAY 330 ipflow_init(); 331#endif 332 333 /* Register our Packet Filter hook. */ 334 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 335 KASSERT(inet_pfil_hook != NULL); 336 337#ifdef MBUFTRACE 338 MOWNER_ATTACH(&ip_tx_mowner); 339 MOWNER_ATTACH(&ip_rx_mowner); 340#endif /* MBUFTRACE */ 341 342 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 343} 344 345/* 346 * IP software interrupt routine. 347 */ 348static void 349ipintr(void *arg __unused) 350{ 351 struct mbuf *m; 352 353 KASSERT(cpu_softintr_p()); 354 355 mutex_enter(softnet_lock); 356 while ((m = pktq_dequeue(ip_pktq)) != NULL) { 357 ip_input(m); 358 } 359 mutex_exit(softnet_lock); 360} 361 362/* 363 * IP input routine. Checksum and byte swap header. If fragmented 364 * try to reassemble. Process options. Pass to next level. 365 */ 366static void 367ip_input(struct mbuf *m) 368{ 369 struct ip *ip = NULL; 370 struct in_ifaddr *ia; 371 struct ifaddr *ifa; 372 int hlen = 0, len; 373 int downmatch; 374 int checkif; 375 int srcrt = 0; 376 ifnet_t *ifp; 377 struct psref psref; 378 379 KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software " 380 "interrupt handler; synchronization assumptions violated"); 381 382 MCLAIM(m, &ip_rx_mowner); 383 KASSERT((m->m_flags & M_PKTHDR) != 0); 384 385 ifp = m_get_rcvif_psref(m, &psref); 386 387 /* 388 * If no IP addresses have been set yet but the interfaces 389 * are receiving, can't do anything with incoming packets yet. 390 * Note: we pre-check without locks held. 391 */ 392 if (!TAILQ_FIRST(&in_ifaddrhead)) { 393 goto bad; 394 } 395 IP_STATINC(IP_STAT_TOTAL); 396 397 /* 398 * If the IP header is not aligned, slurp it up into a new 399 * mbuf with space for link headers, in the event we forward 400 * it. Otherwise, if it is aligned, make sure the entire 401 * base IP header is in the first mbuf of the chain. 402 */ 403 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 404 if ((m = m_copyup(m, sizeof(struct ip), 405 (max_linkhdr + 3) & ~3)) == NULL) { 406 /* XXXJRT new stat, please */ 407 IP_STATINC(IP_STAT_TOOSMALL); 408 goto out; 409 } 410 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 411 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 412 IP_STATINC(IP_STAT_TOOSMALL); 413 goto out; 414 } 415 } 416 ip = mtod(m, struct ip *); 417 if (ip->ip_v != IPVERSION) { 418 IP_STATINC(IP_STAT_BADVERS); 419 goto bad; 420 } 421 hlen = ip->ip_hl << 2; 422 if (hlen < sizeof(struct ip)) { /* minimum header length */ 423 IP_STATINC(IP_STAT_BADHLEN); 424 goto bad; 425 } 426 if (hlen > m->m_len) { 427 if ((m = m_pullup(m, hlen)) == NULL) { 428 IP_STATINC(IP_STAT_BADHLEN); 429 goto out; 430 } 431 ip = mtod(m, struct ip *); 432 } 433 434 /* 435 * RFC1122: packets with a multicast source address are 436 * not allowed. 437 */ 438 if (IN_MULTICAST(ip->ip_src.s_addr)) { 439 IP_STATINC(IP_STAT_BADADDR); 440 goto bad; 441 } 442 443 /* 127/8 must not appear on wire - RFC1122 */ 444 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 445 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 446 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 447 IP_STATINC(IP_STAT_BADADDR); 448 goto bad; 449 } 450 } 451 452 switch (m->m_pkthdr.csum_flags & 453 ((ifp->if_csum_flags_rx & M_CSUM_IPv4) | 454 M_CSUM_IPv4_BAD)) { 455 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 456 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 457 goto badcsum; 458 459 case M_CSUM_IPv4: 460 /* Checksum was okay. */ 461 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 462 break; 463 464 default: 465 /* 466 * Must compute it ourselves. Maybe skip checksum on 467 * loopback interfaces. 468 */ 469 if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) || 470 ip_do_loopback_cksum)) { 471 INET_CSUM_COUNTER_INCR(&ip_swcsum); 472 if (in_cksum(m, hlen) != 0) 473 goto badcsum; 474 } 475 break; 476 } 477 478 /* Retrieve the packet length. */ 479 len = ntohs(ip->ip_len); 480 481 /* 482 * Check for additional length bogosity 483 */ 484 if (len < hlen) { 485 IP_STATINC(IP_STAT_BADLEN); 486 goto bad; 487 } 488 489 /* 490 * Check that the amount of data in the buffers 491 * is as at least much as the IP header would have us expect. 492 * Trim mbufs if longer than we expect. 493 * Drop packet if shorter than we expect. 494 */ 495 if (m->m_pkthdr.len < len) { 496 IP_STATINC(IP_STAT_TOOSHORT); 497 goto bad; 498 } 499 if (m->m_pkthdr.len > len) { 500 if (m->m_len == m->m_pkthdr.len) { 501 m->m_len = len; 502 m->m_pkthdr.len = len; 503 } else 504 m_adj(m, len - m->m_pkthdr.len); 505 } 506 507 /* 508 * Assume that we can create a fast-forward IP flow entry 509 * based on this packet. 510 */ 511 m->m_flags |= M_CANFASTFWD; 512 513 /* 514 * Run through list of hooks for input packets. If there are any 515 * filters which require that additional packets in the flow are 516 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 517 * Note that filters must _never_ set this flag, as another filter 518 * in the list may have previously cleared it. 519 */ 520#if defined(IPSEC) 521 if (!ipsec_used || !ipsec_indone(m)) 522#else 523 if (1) 524#endif 525 { 526 struct in_addr odst = ip->ip_dst; 527 bool freed; 528 529 SOFTNET_LOCK(); 530 freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0; 531 SOFTNET_UNLOCK(); 532 if (freed || m == NULL) { 533 goto out; 534 } 535 ip = mtod(m, struct ip *); 536 hlen = ip->ip_hl << 2; 537 538 /* 539 * XXX The setting of "srcrt" here is to prevent ip_forward() 540 * from generating ICMP redirects for packets that have 541 * been redirected by a hook back out on to the same LAN that 542 * they came from and is not an indication that the packet 543 * is being inffluenced by source routing options. This 544 * allows things like 545 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 546 * where tlp0 is both on the 1.1.1.0/24 network and is the 547 * default route for hosts on 1.1.1.0/24. Of course this 548 * also requires a "map tlp0 ..." to complete the story. 549 * One might argue whether or not this kind of network config. 550 * should be supported in this manner... 551 */ 552 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 553 } 554 555#ifdef ALTQ 556 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 557 if (altq_input) { 558 SOFTNET_LOCK(); 559 if ((*altq_input)(m, AF_INET) == 0) { 560 /* Packet dropped by traffic conditioner. */ 561 SOFTNET_UNLOCK(); 562 goto out; 563 } 564 SOFTNET_UNLOCK(); 565 } 566#endif 567 568 /* 569 * Process options and, if not destined for us, 570 * ship it on. ip_dooptions returns 1 when an 571 * error was detected (causing an icmp message 572 * to be sent and the original packet to be freed). 573 */ 574 ip_nhops = 0; /* for source routed packets */ 575 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 576 goto out; 577 578 /* 579 * Enable a consistency check between the destination address 580 * and the arrival interface for a unicast packet (the RFC 1122 581 * strong ES model) if IP forwarding is disabled and the packet 582 * is not locally generated. 583 * 584 * XXX - Checking also should be disabled if the destination 585 * address is ipnat'ed to a different interface. 586 * 587 * XXX - Checking is incompatible with IP aliases added 588 * to the loopback interface instead of the interface where 589 * the packets are received. 590 * 591 * XXX - We need to add a per ifaddr flag for this so that 592 * we get finer grain control. 593 */ 594 checkif = ip_checkinterface && (ipforwarding == 0) && 595 (ifp->if_flags & IFF_LOOPBACK) == 0; 596 597 /* 598 * Check our list of addresses, to see if the packet is for us. 599 * 600 * Traditional 4.4BSD did not consult IFF_UP at all. 601 * The behavior here is to treat addresses on !IFF_UP interface 602 * or IN_IFF_NOTREADY addresses as not mine. 603 */ 604 downmatch = 0; 605 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 606 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 607 if (ia->ia4_flags & IN_IFF_NOTREADY) 608 continue; 609 if (checkif && ia->ia_ifp != ifp) 610 continue; 611 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 612 break; 613 else 614 downmatch++; 615 } 616 } 617 if (ia != NULL) 618 goto ours; 619 if (ifp->if_flags & IFF_BROADCAST) { 620 IFADDR_FOREACH(ifa, ifp) { 621 if (ifa->ifa_addr->sa_family != AF_INET) 622 continue; 623 ia = ifatoia(ifa); 624 if (ia->ia4_flags & IN_IFF_NOTREADY) 625 continue; 626 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 627 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 628 /* 629 * Look for all-0's host part (old broadcast addr), 630 * either for subnet or net. 631 */ 632 ip->ip_dst.s_addr == ia->ia_subnet || 633 ip->ip_dst.s_addr == ia->ia_net) 634 goto ours; 635 /* 636 * An interface with IP address zero accepts 637 * all packets that arrive on that interface. 638 */ 639 if (in_nullhost(ia->ia_addr.sin_addr)) 640 goto ours; 641 } 642 } 643 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 644#ifdef MROUTING 645 extern struct socket *ip_mrouter; 646 647 if (ip_mrouter) { 648 /* 649 * If we are acting as a multicast router, all 650 * incoming multicast packets are passed to the 651 * kernel-level multicast forwarding function. 652 * The packet is returned (relatively) intact; if 653 * ip_mforward() returns a non-zero value, the packet 654 * must be discarded, else it may be accepted below. 655 * 656 * (The IP ident field is put in the same byte order 657 * as expected when ip_mforward() is called from 658 * ip_output().) 659 */ 660 SOFTNET_LOCK(); 661 if (ip_mforward(m, ifp) != 0) { 662 SOFTNET_UNLOCK(); 663 IP_STATINC(IP_STAT_CANTFORWARD); 664 goto bad; 665 } 666 SOFTNET_UNLOCK(); 667 668 /* 669 * The process-level routing demon needs to receive 670 * all multicast IGMP packets, whether or not this 671 * host belongs to their destination groups. 672 */ 673 if (ip->ip_p == IPPROTO_IGMP) { 674 goto ours; 675 } 676 IP_STATINC(IP_STAT_CANTFORWARD); 677 } 678#endif 679 /* 680 * See if we belong to the destination multicast group on the 681 * arrival interface. 682 */ 683 if (!in_multi_group(ip->ip_dst, ifp, 0)) { 684 IP_STATINC(IP_STAT_CANTFORWARD); 685 goto bad; 686 } 687 goto ours; 688 } 689 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 690 in_nullhost(ip->ip_dst)) 691 goto ours; 692 693 /* 694 * Not for us; forward if possible and desirable. 695 */ 696 if (ipforwarding == 0) { 697 m_put_rcvif_psref(ifp, &psref); 698 IP_STATINC(IP_STAT_CANTFORWARD); 699 m_freem(m); 700 } else { 701 /* 702 * If ip_dst matched any of my address on !IFF_UP interface, 703 * and there's no IFF_UP interface that matches ip_dst, 704 * send icmp unreach. Forwarding it will result in in-kernel 705 * forwarding loop till TTL goes to 0. 706 */ 707 if (downmatch) { 708 m_put_rcvif_psref(ifp, &psref); 709 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 710 IP_STATINC(IP_STAT_CANTFORWARD); 711 return; 712 } 713#ifdef IPSEC 714 /* Perform IPsec, if any. */ 715 if (ipsec_used) { 716 SOFTNET_LOCK(); 717 if (ipsec4_input(m, IP_FORWARDING | 718 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) { 719 SOFTNET_UNLOCK(); 720 goto bad; 721 } 722 SOFTNET_UNLOCK(); 723 } 724#endif 725 ip_forward(m, srcrt, ifp); 726 m_put_rcvif_psref(ifp, &psref); 727 } 728 return; 729 730ours: 731 m_put_rcvif_psref(ifp, &psref); 732 733 /* 734 * If offset or IP_MF are set, must reassemble. 735 */ 736 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 737 /* 738 * Pass to IP reassembly mechanism. 739 */ 740 if (ip_reass_packet(&m, ip) != 0) { 741 /* Failed; invalid fragment(s) or packet. */ 742 goto bad; 743 } 744 if (m == NULL) { 745 /* More fragments should come; silently return. */ 746 return; 747 } 748 /* 749 * Reassembly is done, we have the final packet. 750 * Updated cached data in local variable(s). 751 */ 752 ip = mtod(m, struct ip *); 753 hlen = ip->ip_hl << 2; 754 } 755 756#ifdef IPSEC 757 /* 758 * Enforce IPsec policy checking if we are seeing last header. 759 * Note that we do not visit this with protocols with PCB layer 760 * code - like UDP/TCP/raw IP. 761 */ 762 if (ipsec_used && 763 (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 764 SOFTNET_LOCK(); 765 if (ipsec4_input(m, 0) != 0) { 766 SOFTNET_UNLOCK(); 767 goto bad; 768 } 769 SOFTNET_UNLOCK(); 770 } 771#endif 772 773 /* 774 * Switch out to protocol's input routine. 775 */ 776#if IFA_STATS 777 if (ia && ip) 778 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 779#endif 780 IP_STATINC(IP_STAT_DELIVERED); 781 782 const int off = hlen, nh = ip->ip_p; 783 784 SOFTNET_LOCK(); 785 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 786 SOFTNET_UNLOCK(); 787 return; 788bad: 789 m_put_rcvif_psref(ifp, &psref); 790 m_freem(m); 791 return; 792 793badcsum: 794 m_put_rcvif_psref(ifp, &psref); 795 IP_STATINC(IP_STAT_BADSUM); 796 m_freem(m); 797 return; 798out: 799 m_put_rcvif_psref(ifp, &psref); 800} 801 802/* 803 * IP timer processing. 804 */ 805void 806ip_slowtimo(void) 807{ 808 809 mutex_enter(softnet_lock); 810 KERNEL_LOCK(1, NULL); 811 812 ip_reass_slowtimo(); 813 814 KERNEL_UNLOCK_ONE(NULL); 815 mutex_exit(softnet_lock); 816} 817 818/* 819 * IP drain processing. 820 */ 821void 822ip_drain(void) 823{ 824 825 KERNEL_LOCK(1, NULL); 826 ip_reass_drain(); 827 KERNEL_UNLOCK_ONE(NULL); 828} 829 830/* 831 * ip_dooptions: perform option processing on a datagram, possibly discarding 832 * it if bad options are encountered, or forwarding it if source-routed. 833 * 834 * => Returns true if packet has been forwarded/freed. 835 * => Returns false if the packet should be processed further. 836 */ 837static bool 838ip_dooptions(struct mbuf *m) 839{ 840 struct ip *ip = mtod(m, struct ip *); 841 u_char *cp, *cp0; 842 struct ip_timestamp *ipt; 843 struct in_ifaddr *ia; 844 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 845 struct in_addr dst; 846 n_time ntime; 847 848 dst = ip->ip_dst; 849 cp = (u_char *)(ip + 1); 850 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 851 for (; cnt > 0; cnt -= optlen, cp += optlen) { 852 opt = cp[IPOPT_OPTVAL]; 853 if (opt == IPOPT_EOL) 854 break; 855 if (opt == IPOPT_NOP) 856 optlen = 1; 857 else { 858 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 859 code = &cp[IPOPT_OLEN] - (u_char *)ip; 860 goto bad; 861 } 862 optlen = cp[IPOPT_OLEN]; 863 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 864 code = &cp[IPOPT_OLEN] - (u_char *)ip; 865 goto bad; 866 } 867 } 868 switch (opt) { 869 870 default: 871 break; 872 873 /* 874 * Source routing with record. 875 * Find interface with current destination address. 876 * If none on this machine then drop if strictly routed, 877 * or do nothing if loosely routed. 878 * Record interface address and bring up next address 879 * component. If strictly routed make sure next 880 * address is on directly accessible net. 881 */ 882 case IPOPT_LSRR: 883 case IPOPT_SSRR: 884 if (ip_allowsrcrt == 0) { 885 type = ICMP_UNREACH; 886 code = ICMP_UNREACH_NET_PROHIB; 887 goto bad; 888 } 889 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 890 code = &cp[IPOPT_OLEN] - (u_char *)ip; 891 goto bad; 892 } 893 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 894 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 895 goto bad; 896 } 897 ipaddr.sin_addr = ip->ip_dst; 898 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 899 if (ia == 0) { 900 if (opt == IPOPT_SSRR) { 901 type = ICMP_UNREACH; 902 code = ICMP_UNREACH_SRCFAIL; 903 goto bad; 904 } 905 /* 906 * Loose routing, and not at next destination 907 * yet; nothing to do except forward. 908 */ 909 break; 910 } 911 off--; /* 0 origin */ 912 if ((off + sizeof(struct in_addr)) > optlen) { 913 /* 914 * End of source route. Should be for us. 915 */ 916 save_rte(cp, ip->ip_src); 917 break; 918 } 919 /* 920 * locate outgoing interface 921 */ 922 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 923 sizeof(ipaddr.sin_addr)); 924 if (opt == IPOPT_SSRR) 925 ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); 926 else 927 ia = ip_rtaddr(ipaddr.sin_addr); 928 if (ia == 0) { 929 type = ICMP_UNREACH; 930 code = ICMP_UNREACH_SRCFAIL; 931 goto bad; 932 } 933 ip->ip_dst = ipaddr.sin_addr; 934 bcopy((void *)&ia->ia_addr.sin_addr, 935 (void *)(cp + off), sizeof(struct in_addr)); 936 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 937 /* 938 * Let ip_intr's mcast routing check handle mcast pkts 939 */ 940 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 941 break; 942 943 case IPOPT_RR: 944 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 945 code = &cp[IPOPT_OLEN] - (u_char *)ip; 946 goto bad; 947 } 948 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 949 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 950 goto bad; 951 } 952 /* 953 * If no space remains, ignore. 954 */ 955 off--; /* 0 origin */ 956 if ((off + sizeof(struct in_addr)) > optlen) 957 break; 958 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 959 sizeof(ipaddr.sin_addr)); 960 /* 961 * locate outgoing interface; if we're the destination, 962 * use the incoming interface (should be same). 963 */ 964 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 965 == NULL && 966 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 967 type = ICMP_UNREACH; 968 code = ICMP_UNREACH_HOST; 969 goto bad; 970 } 971 bcopy((void *)&ia->ia_addr.sin_addr, 972 (void *)(cp + off), sizeof(struct in_addr)); 973 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 974 break; 975 976 case IPOPT_TS: 977 code = cp - (u_char *)ip; 978 ipt = (struct ip_timestamp *)cp; 979 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 980 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 981 goto bad; 982 } 983 if (ipt->ipt_ptr < 5) { 984 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 985 goto bad; 986 } 987 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 988 if (++ipt->ipt_oflw == 0) { 989 code = (u_char *)&ipt->ipt_ptr - 990 (u_char *)ip; 991 goto bad; 992 } 993 break; 994 } 995 cp0 = (cp + ipt->ipt_ptr - 1); 996 switch (ipt->ipt_flg) { 997 998 case IPOPT_TS_TSONLY: 999 break; 1000 1001 case IPOPT_TS_TSANDADDR: { 1002 struct ifnet *rcvif; 1003 int s; 1004 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 ipaddr.sin_addr = dst; 1012 rcvif = m_get_rcvif(m, &s); 1013 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1014 rcvif)); 1015 m_put_rcvif(rcvif, &s); 1016 if (ia == 0) 1017 continue; 1018 bcopy(&ia->ia_addr.sin_addr, 1019 cp0, sizeof(struct in_addr)); 1020 ipt->ipt_ptr += sizeof(struct in_addr); 1021 break; 1022 } 1023 1024 case IPOPT_TS_PRESPEC: 1025 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1026 sizeof(struct in_addr) > ipt->ipt_len) { 1027 code = (u_char *)&ipt->ipt_ptr - 1028 (u_char *)ip; 1029 goto bad; 1030 } 1031 memcpy(&ipaddr.sin_addr, cp0, 1032 sizeof(struct in_addr)); 1033 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1034 == NULL) 1035 continue; 1036 ipt->ipt_ptr += sizeof(struct in_addr); 1037 break; 1038 1039 default: 1040 /* XXX can't take &ipt->ipt_flg */ 1041 code = (u_char *)&ipt->ipt_ptr - 1042 (u_char *)ip + 1; 1043 goto bad; 1044 } 1045 ntime = iptime(); 1046 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1047 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1048 sizeof(n_time)); 1049 ipt->ipt_ptr += sizeof(n_time); 1050 } 1051 } 1052 if (forward) { 1053 struct ifnet *rcvif; 1054 struct psref psref; 1055 1056 if (ip_forwsrcrt == 0) { 1057 type = ICMP_UNREACH; 1058 code = ICMP_UNREACH_SRCFAIL; 1059 goto bad; 1060 } 1061 1062 rcvif = m_get_rcvif_psref(m, &psref); 1063 ip_forward(m, 1, rcvif); 1064 m_put_rcvif_psref(rcvif, &psref); 1065 return true; 1066 } 1067 return false; 1068bad: 1069 icmp_error(m, type, code, 0, 0); 1070 IP_STATINC(IP_STAT_BADOPTIONS); 1071 return true; 1072} 1073 1074/* 1075 * ip_rtaddr: given address of next destination (final or next hop), 1076 * return internet address info of interface to be used to get there. 1077 */ 1078static struct in_ifaddr * 1079ip_rtaddr(struct in_addr dst) 1080{ 1081 struct rtentry *rt; 1082 union { 1083 struct sockaddr dst; 1084 struct sockaddr_in dst4; 1085 } u; 1086 1087 sockaddr_in_init(&u.dst4, &dst, 0); 1088 1089 SOFTNET_LOCK(); 1090 rt = rtcache_lookup(&ipforward_rt, &u.dst); 1091 SOFTNET_UNLOCK(); 1092 if (rt == NULL) 1093 return NULL; 1094 1095 return ifatoia(rt->rt_ifa); 1096} 1097 1098/* 1099 * save_rte: save incoming source route for use in replies, to be picked 1100 * up later by ip_srcroute if the receiver is interested. 1101 */ 1102static void 1103save_rte(u_char *option, struct in_addr dst) 1104{ 1105 unsigned olen; 1106 1107 olen = option[IPOPT_OLEN]; 1108 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1109 return; 1110 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1111 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1112 ip_srcrt.dst = dst; 1113} 1114 1115/* 1116 * Retrieve incoming source route for use in replies, 1117 * in the same form used by setsockopt. 1118 * The first hop is placed before the options, will be removed later. 1119 */ 1120struct mbuf * 1121ip_srcroute(void) 1122{ 1123 struct in_addr *p, *q; 1124 struct mbuf *m; 1125 1126 if (ip_nhops == 0) 1127 return NULL; 1128 m = m_get(M_DONTWAIT, MT_SOOPTS); 1129 if (m == 0) 1130 return NULL; 1131 1132 MCLAIM(m, &inetdomain.dom_mowner); 1133#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1134 1135 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1136 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1137 OPTSIZ; 1138 1139 /* 1140 * First save first hop for return route 1141 */ 1142 p = &ip_srcrt.route[ip_nhops - 1]; 1143 *(mtod(m, struct in_addr *)) = *p--; 1144 1145 /* 1146 * Copy option fields and padding (nop) to mbuf. 1147 */ 1148 ip_srcrt.nop = IPOPT_NOP; 1149 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1150 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1151 OPTSIZ); 1152 q = (struct in_addr *)(mtod(m, char *) + 1153 sizeof(struct in_addr) + OPTSIZ); 1154#undef OPTSIZ 1155 /* 1156 * Record return path as an IP source route, 1157 * reversing the path (pointers are now aligned). 1158 */ 1159 while (p >= ip_srcrt.route) { 1160 *q++ = *p--; 1161 } 1162 /* 1163 * Last hop goes to final destination. 1164 */ 1165 *q = ip_srcrt.dst; 1166 return (m); 1167} 1168 1169const int inetctlerrmap[PRC_NCMDS] = { 1170 [PRC_MSGSIZE] = EMSGSIZE, 1171 [PRC_HOSTDEAD] = EHOSTDOWN, 1172 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1173 [PRC_UNREACH_NET] = EHOSTUNREACH, 1174 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1175 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1176 [PRC_UNREACH_PORT] = ECONNREFUSED, 1177 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1178 [PRC_PARAMPROB] = ENOPROTOOPT, 1179}; 1180 1181void 1182ip_fasttimo(void) 1183{ 1184 if (ip_drainwanted) { 1185 ip_drain(); 1186 ip_drainwanted = 0; 1187 } 1188} 1189 1190void 1191ip_drainstub(void) 1192{ 1193 ip_drainwanted = 1; 1194} 1195 1196/* 1197 * Forward a packet. If some error occurs return the sender 1198 * an icmp packet. Note we can't always generate a meaningful 1199 * icmp message because icmp doesn't have a large enough repertoire 1200 * of codes and types. 1201 * 1202 * If not forwarding, just drop the packet. This could be confusing 1203 * if ipforwarding was zero but some routing protocol was advancing 1204 * us as a gateway to somewhere. However, we must let the routing 1205 * protocol deal with that. 1206 * 1207 * The srcrt parameter indicates whether the packet is being forwarded 1208 * via a source route. 1209 */ 1210static void 1211ip_forward(struct mbuf *m, int srcrt, struct ifnet *rcvif) 1212{ 1213 struct ip *ip = mtod(m, struct ip *); 1214 struct rtentry *rt; 1215 int error, type = 0, code = 0, destmtu = 0; 1216 struct mbuf *mcopy; 1217 n_long dest; 1218 union { 1219 struct sockaddr dst; 1220 struct sockaddr_in dst4; 1221 } u; 1222 uint64_t *ips; 1223 1224 KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software " 1225 "interrupt handler; synchronization assumptions violated"); 1226 1227 /* 1228 * We are now in the output path. 1229 */ 1230 MCLAIM(m, &ip_tx_mowner); 1231 1232 /* 1233 * Clear any in-bound checksum flags for this packet. 1234 */ 1235 m->m_pkthdr.csum_flags = 0; 1236 1237 dest = 0; 1238 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1239 IP_STATINC(IP_STAT_CANTFORWARD); 1240 m_freem(m); 1241 return; 1242 } 1243 1244 SOFTNET_LOCK(); 1245 1246 if (ip->ip_ttl <= IPTTLDEC) { 1247 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1248 SOFTNET_UNLOCK(); 1249 return; 1250 } 1251 1252 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1253 1254 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1255 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1256 SOFTNET_UNLOCK(); 1257 return; 1258 } 1259 1260 /* 1261 * Save at most 68 bytes of the packet in case 1262 * we need to generate an ICMP message to the src. 1263 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1264 */ 1265 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1266 if (mcopy) 1267 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1268 1269 ip->ip_ttl -= IPTTLDEC; 1270 1271 /* 1272 * If forwarding packet using same interface that it came in on, 1273 * perhaps should send a redirect to sender to shortcut a hop. 1274 * Only send redirect if source is sending directly to us, 1275 * and if packet was not source routed (or has any options). 1276 * Also, don't send redirect if forwarding using a default route 1277 * or a route modified by a redirect. 1278 */ 1279 if (rt->rt_ifp == rcvif && 1280 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1281 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1282 ipsendredirects && !srcrt) { 1283 if (rt->rt_ifa && 1284 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1285 ifatoia(rt->rt_ifa)->ia_subnet) { 1286 if (rt->rt_flags & RTF_GATEWAY) 1287 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1288 else 1289 dest = ip->ip_dst.s_addr; 1290 /* 1291 * Router requirements says to only send host 1292 * redirects. 1293 */ 1294 type = ICMP_REDIRECT; 1295 code = ICMP_REDIRECT_HOST; 1296 } 1297 } 1298 1299 error = ip_output(m, NULL, &ipforward_rt, 1300 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1301 NULL, NULL); 1302 1303 if (error) { 1304 IP_STATINC(IP_STAT_CANTFORWARD); 1305 goto error; 1306 } 1307 1308 ips = IP_STAT_GETREF(); 1309 ips[IP_STAT_FORWARD]++; 1310 1311 if (type) { 1312 ips[IP_STAT_REDIRECTSENT]++; 1313 IP_STAT_PUTREF(); 1314 goto redirect; 1315 } 1316 1317 IP_STAT_PUTREF(); 1318 if (mcopy) { 1319#ifdef GATEWAY 1320 if (mcopy->m_flags & M_CANFASTFWD) 1321 ipflow_create(&ipforward_rt, mcopy); 1322#endif 1323 m_freem(mcopy); 1324 } 1325 1326 SOFTNET_UNLOCK(); 1327 return; 1328 1329redirect: 1330error: 1331 if (mcopy == NULL) { 1332 SOFTNET_UNLOCK(); 1333 return; 1334 } 1335 1336 switch (error) { 1337 1338 case 0: /* forwarded, but need redirect */ 1339 /* type, code set above */ 1340 break; 1341 1342 case ENETUNREACH: /* shouldn't happen, checked above */ 1343 case EHOSTUNREACH: 1344 case ENETDOWN: 1345 case EHOSTDOWN: 1346 default: 1347 type = ICMP_UNREACH; 1348 code = ICMP_UNREACH_HOST; 1349 break; 1350 1351 case EMSGSIZE: 1352 type = ICMP_UNREACH; 1353 code = ICMP_UNREACH_NEEDFRAG; 1354 1355 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1356 destmtu = rt->rt_ifp->if_mtu; 1357#ifdef IPSEC 1358 if (ipsec_used) 1359 (void)ipsec4_forward(mcopy, &destmtu); 1360#endif 1361 IP_STATINC(IP_STAT_CANTFRAG); 1362 break; 1363 1364 case ENOBUFS: 1365 /* 1366 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1367 * Requirements for IP Version 4 Routers. Source quench can 1368 * big problem under DoS attacks or if the underlying 1369 * interface is rate-limited. 1370 */ 1371 if (mcopy) 1372 m_freem(mcopy); 1373 SOFTNET_UNLOCK(); 1374 return; 1375 } 1376 icmp_error(mcopy, type, code, dest, destmtu); 1377 SOFTNET_UNLOCK(); 1378} 1379 1380void 1381ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1382 struct mbuf *m) 1383{ 1384 struct socket *so = inp->inp_socket; 1385 ifnet_t *ifp; 1386 int inpflags = inp->inp_flags; 1387 struct psref psref; 1388 1389 ifp = m_get_rcvif_psref(m, &psref); 1390 if (so->so_options & SO_TIMESTAMP 1391#ifdef SO_OTIMESTAMP 1392 || so->so_options & SO_OTIMESTAMP 1393#endif 1394 ) { 1395 struct timeval tv; 1396 1397 microtime(&tv); 1398#ifdef SO_OTIMESTAMP 1399 if (so->so_options & SO_OTIMESTAMP) { 1400 struct timeval50 tv50; 1401 timeval_to_timeval50(&tv, &tv50); 1402 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1403 SCM_OTIMESTAMP, SOL_SOCKET); 1404 } else 1405#endif 1406 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1407 SCM_TIMESTAMP, SOL_SOCKET); 1408 if (*mp) 1409 mp = &(*mp)->m_next; 1410 } 1411 if (inpflags & INP_RECVDSTADDR) { 1412 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1413 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1414 if (*mp) 1415 mp = &(*mp)->m_next; 1416 } 1417 if (inpflags & INP_RECVPKTINFO) { 1418 struct in_pktinfo ipi; 1419 ipi.ipi_addr = ip->ip_src; 1420 ipi.ipi_ifindex = ifp->if_index; 1421 *mp = sbcreatecontrol((void *) &ipi, 1422 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1423 if (*mp) 1424 mp = &(*mp)->m_next; 1425 } 1426 if (inpflags & INP_PKTINFO) { 1427 struct in_pktinfo ipi; 1428 ipi.ipi_addr = ip->ip_dst; 1429 ipi.ipi_ifindex = ifp->if_index; 1430 *mp = sbcreatecontrol((void *) &ipi, 1431 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1432 if (*mp) 1433 mp = &(*mp)->m_next; 1434 } 1435 if (inpflags & INP_RECVIF) { 1436 struct sockaddr_dl sdl; 1437 1438 sockaddr_dl_init(&sdl, sizeof(sdl), ifp ? 1439 ifp->if_index : 0, 0, NULL, 0, NULL, 0); 1440 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1441 if (*mp) 1442 mp = &(*mp)->m_next; 1443 } 1444 if (inpflags & INP_RECVTTL) { 1445 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1446 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1447 if (*mp) 1448 mp = &(*mp)->m_next; 1449 } 1450 m_put_rcvif_psref(ifp, &psref); 1451} 1452 1453/* 1454 * sysctl helper routine for net.inet.ip.forwsrcrt. 1455 */ 1456static int 1457sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1458{ 1459 int error, tmp; 1460 struct sysctlnode node; 1461 1462 node = *rnode; 1463 tmp = ip_forwsrcrt; 1464 node.sysctl_data = &tmp; 1465 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1466 if (error || newp == NULL) 1467 return (error); 1468 1469 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1470 0, NULL, NULL, NULL); 1471 if (error) 1472 return (error); 1473 1474 ip_forwsrcrt = tmp; 1475 1476 return (0); 1477} 1478 1479/* 1480 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1481 * range of the new value and tweaks timers if it changes. 1482 */ 1483static int 1484sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1485{ 1486 int error, tmp; 1487 struct sysctlnode node; 1488 1489 node = *rnode; 1490 tmp = ip_mtudisc_timeout; 1491 node.sysctl_data = &tmp; 1492 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1493 if (error || newp == NULL) 1494 return (error); 1495 if (tmp < 0) 1496 return (EINVAL); 1497 1498 mutex_enter(softnet_lock); 1499 1500 ip_mtudisc_timeout = tmp; 1501 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1502 1503 mutex_exit(softnet_lock); 1504 1505 return (0); 1506} 1507 1508static int 1509sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1510{ 1511 1512 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1513} 1514 1515static void 1516sysctl_net_inet_ip_setup(struct sysctllog **clog) 1517{ 1518 sysctl_createv(clog, 0, NULL, NULL, 1519 CTLFLAG_PERMANENT, 1520 CTLTYPE_NODE, "inet", 1521 SYSCTL_DESCR("PF_INET related settings"), 1522 NULL, 0, NULL, 0, 1523 CTL_NET, PF_INET, CTL_EOL); 1524 sysctl_createv(clog, 0, NULL, NULL, 1525 CTLFLAG_PERMANENT, 1526 CTLTYPE_NODE, "ip", 1527 SYSCTL_DESCR("IPv4 related settings"), 1528 NULL, 0, NULL, 0, 1529 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1530 1531 sysctl_createv(clog, 0, NULL, NULL, 1532 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1533 CTLTYPE_INT, "forwarding", 1534 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1535 NULL, 0, &ipforwarding, 0, 1536 CTL_NET, PF_INET, IPPROTO_IP, 1537 IPCTL_FORWARDING, CTL_EOL); 1538 sysctl_createv(clog, 0, NULL, NULL, 1539 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1540 CTLTYPE_INT, "redirect", 1541 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1542 NULL, 0, &ipsendredirects, 0, 1543 CTL_NET, PF_INET, IPPROTO_IP, 1544 IPCTL_SENDREDIRECTS, CTL_EOL); 1545 sysctl_createv(clog, 0, NULL, NULL, 1546 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1547 CTLTYPE_INT, "ttl", 1548 SYSCTL_DESCR("Default TTL for an INET datagram"), 1549 NULL, 0, &ip_defttl, 0, 1550 CTL_NET, PF_INET, IPPROTO_IP, 1551 IPCTL_DEFTTL, CTL_EOL); 1552#ifdef IPCTL_DEFMTU 1553 sysctl_createv(clog, 0, NULL, NULL, 1554 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1555 CTLTYPE_INT, "mtu", 1556 SYSCTL_DESCR("Default MTA for an INET route"), 1557 NULL, 0, &ip_mtu, 0, 1558 CTL_NET, PF_INET, IPPROTO_IP, 1559 IPCTL_DEFMTU, CTL_EOL); 1560#endif /* IPCTL_DEFMTU */ 1561 sysctl_createv(clog, 0, NULL, NULL, 1562 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1563 CTLTYPE_INT, "forwsrcrt", 1564 SYSCTL_DESCR("Enable forwarding of source-routed " 1565 "datagrams"), 1566 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1567 CTL_NET, PF_INET, IPPROTO_IP, 1568 IPCTL_FORWSRCRT, CTL_EOL); 1569 sysctl_createv(clog, 0, NULL, NULL, 1570 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1571 CTLTYPE_INT, "directed-broadcast", 1572 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1573 NULL, 0, &ip_directedbcast, 0, 1574 CTL_NET, PF_INET, IPPROTO_IP, 1575 IPCTL_DIRECTEDBCAST, CTL_EOL); 1576 sysctl_createv(clog, 0, NULL, NULL, 1577 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1578 CTLTYPE_INT, "allowsrcrt", 1579 SYSCTL_DESCR("Accept source-routed datagrams"), 1580 NULL, 0, &ip_allowsrcrt, 0, 1581 CTL_NET, PF_INET, IPPROTO_IP, 1582 IPCTL_ALLOWSRCRT, CTL_EOL); 1583 1584 sysctl_createv(clog, 0, NULL, NULL, 1585 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1586 CTLTYPE_INT, "mtudisc", 1587 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1588 NULL, 0, &ip_mtudisc, 0, 1589 CTL_NET, PF_INET, IPPROTO_IP, 1590 IPCTL_MTUDISC, CTL_EOL); 1591 sysctl_createv(clog, 0, NULL, NULL, 1592 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1593 CTLTYPE_INT, "anonportmin", 1594 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1595 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1596 CTL_NET, PF_INET, IPPROTO_IP, 1597 IPCTL_ANONPORTMIN, CTL_EOL); 1598 sysctl_createv(clog, 0, NULL, NULL, 1599 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1600 CTLTYPE_INT, "anonportmax", 1601 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1602 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1603 CTL_NET, PF_INET, IPPROTO_IP, 1604 IPCTL_ANONPORTMAX, CTL_EOL); 1605 sysctl_createv(clog, 0, NULL, NULL, 1606 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1607 CTLTYPE_INT, "mtudisctimeout", 1608 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1609 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1610 CTL_NET, PF_INET, IPPROTO_IP, 1611 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1612#ifndef IPNOPRIVPORTS 1613 sysctl_createv(clog, 0, NULL, NULL, 1614 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1615 CTLTYPE_INT, "lowportmin", 1616 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1617 "to assign"), 1618 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1619 CTL_NET, PF_INET, IPPROTO_IP, 1620 IPCTL_LOWPORTMIN, CTL_EOL); 1621 sysctl_createv(clog, 0, NULL, NULL, 1622 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1623 CTLTYPE_INT, "lowportmax", 1624 SYSCTL_DESCR("Highest privileged ephemeral port number " 1625 "to assign"), 1626 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1627 CTL_NET, PF_INET, IPPROTO_IP, 1628 IPCTL_LOWPORTMAX, CTL_EOL); 1629#endif /* IPNOPRIVPORTS */ 1630#if NGRE > 0 1631 sysctl_createv(clog, 0, NULL, NULL, 1632 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1633 CTLTYPE_INT, "grettl", 1634 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1635 NULL, 0, &ip_gre_ttl, 0, 1636 CTL_NET, PF_INET, IPPROTO_IP, 1637 IPCTL_GRE_TTL, CTL_EOL); 1638#endif /* NGRE */ 1639 sysctl_createv(clog, 0, NULL, NULL, 1640 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1641 CTLTYPE_INT, "checkinterface", 1642 SYSCTL_DESCR("Enable receive side of Strong ES model " 1643 "from RFC1122"), 1644 NULL, 0, &ip_checkinterface, 0, 1645 CTL_NET, PF_INET, IPPROTO_IP, 1646 IPCTL_CHECKINTERFACE, CTL_EOL); 1647 sysctl_createv(clog, 0, NULL, NULL, 1648 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1649 CTLTYPE_INT, "random_id", 1650 SYSCTL_DESCR("Assign random ip_id values"), 1651 NULL, 0, &ip_do_randomid, 0, 1652 CTL_NET, PF_INET, IPPROTO_IP, 1653 IPCTL_RANDOMID, CTL_EOL); 1654 sysctl_createv(clog, 0, NULL, NULL, 1655 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1656 CTLTYPE_INT, "do_loopback_cksum", 1657 SYSCTL_DESCR("Perform IP checksum on loopback"), 1658 NULL, 0, &ip_do_loopback_cksum, 0, 1659 CTL_NET, PF_INET, IPPROTO_IP, 1660 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1661 sysctl_createv(clog, 0, NULL, NULL, 1662 CTLFLAG_PERMANENT, 1663 CTLTYPE_STRUCT, "stats", 1664 SYSCTL_DESCR("IP statistics"), 1665 sysctl_net_inet_ip_stats, 0, NULL, 0, 1666 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1667 CTL_EOL); 1668#if NARP 1669 sysctl_createv(clog, 0, NULL, NULL, 1670 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1671 CTLTYPE_INT, "dad_count", 1672 SYSCTL_DESCR("Number of Duplicate Address Detection " 1673 "probes to send"), 1674 NULL, 0, &ip_dad_count, 0, 1675 CTL_NET, PF_INET, IPPROTO_IP, 1676 IPCTL_DAD_COUNT, CTL_EOL); 1677#endif 1678 1679 /* anonportalgo RFC6056 subtree */ 1680 const struct sysctlnode *portalgo_node; 1681 sysctl_createv(clog, 0, NULL, &portalgo_node, 1682 CTLFLAG_PERMANENT, 1683 CTLTYPE_NODE, "anonportalgo", 1684 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1685 NULL, 0, NULL, 0, 1686 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1687 sysctl_createv(clog, 0, &portalgo_node, NULL, 1688 CTLFLAG_PERMANENT, 1689 CTLTYPE_STRING, "available", 1690 SYSCTL_DESCR("available algorithms"), 1691 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1692 CTL_CREATE, CTL_EOL); 1693 sysctl_createv(clog, 0, &portalgo_node, NULL, 1694 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1695 CTLTYPE_STRING, "selected", 1696 SYSCTL_DESCR("selected algorithm"), 1697 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1698 CTL_CREATE, CTL_EOL); 1699 sysctl_createv(clog, 0, &portalgo_node, NULL, 1700 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1701 CTLTYPE_STRUCT, "reserve", 1702 SYSCTL_DESCR("bitmap of reserved ports"), 1703 sysctl_portalgo_reserve4, 0, NULL, 0, 1704 CTL_CREATE, CTL_EOL); 1705} 1706 1707void 1708ip_statinc(u_int stat) 1709{ 1710 1711 KASSERT(stat < IP_NSTATS); 1712 IP_STATINC(stat); 1713} 1714