ip_input.c revision 1.327
1/* $NetBSD: ip_input.c,v 1.327 2016/01/21 15:27:48 riastradh 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 1.327 2016/01/21 15:27:48 riastradh 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); 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 const struct ipprotosw *ippr; 305 size_t i; 306 307 in_init(); 308 sysctl_net_inet_ip_setup(NULL); 309 310 for (i = 0; i < inetdomain.dom_nprotosw; i++) { 311 inetdomain.dom_protosw[i] = &inetsw[i].ippr_protosw; 312 } 313 314 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 315 KASSERT(pr != NULL); 316 ippr = container_of(pr, struct ipprotosw, ippr_protosw); 317 318 ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL); 319 KASSERT(ip_pktq != NULL); 320 321 for (i = 0; i < IPPROTO_MAX; i++) { 322 ip_protox[i] = ippr - inetsw; 323 } 324 for (i = 0; i < inetdomain.dom_nprotosw; i++) { 325 pr = inetdomain.dom_protosw[i]; 326 ippr = container_of(pr, struct ipprotosw, ippr_protosw); 327 if (pr->pr_domain->dom_family == PF_INET && 328 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 329 ip_protox[pr->pr_protocol] = ippr - inetsw; 330 } 331 332 ip_reass_init(); 333 334 ip_ids = ip_id_init(); 335 ip_id = time_uptime & 0xfffff; 336 337 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 338#ifdef GATEWAY 339 ipflow_init(); 340#endif 341 342 /* Register our Packet Filter hook. */ 343 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 344 KASSERT(inet_pfil_hook != NULL); 345 346#ifdef MBUFTRACE 347 MOWNER_ATTACH(&ip_tx_mowner); 348 MOWNER_ATTACH(&ip_rx_mowner); 349#endif /* MBUFTRACE */ 350 351 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 352} 353 354/* 355 * IP software interrupt routine. 356 */ 357static void 358ipintr(void *arg __unused) 359{ 360 struct mbuf *m; 361 362 KASSERT(cpu_softintr_p()); 363 364 mutex_enter(softnet_lock); 365 while ((m = pktq_dequeue(ip_pktq)) != NULL) { 366 ip_input(m); 367 } 368 mutex_exit(softnet_lock); 369} 370 371/* 372 * IP input routine. Checksum and byte swap header. If fragmented 373 * try to reassemble. Process options. Pass to next level. 374 */ 375static void 376ip_input(struct mbuf *m) 377{ 378 struct ip *ip = NULL; 379 struct in_ifaddr *ia; 380 struct ifaddr *ifa; 381 int hlen = 0, len; 382 int downmatch; 383 int checkif; 384 int srcrt = 0; 385 ifnet_t *ifp; 386 387 KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software " 388 "interrupt handler; synchronization assumptions violated"); 389 390 MCLAIM(m, &ip_rx_mowner); 391 KASSERT((m->m_flags & M_PKTHDR) != 0); 392 ifp = m->m_pkthdr.rcvif; 393 394 /* 395 * If no IP addresses have been set yet but the interfaces 396 * are receiving, can't do anything with incoming packets yet. 397 * Note: we pre-check without locks held. 398 */ 399 if (!TAILQ_FIRST(&in_ifaddrhead)) { 400 goto bad; 401 } 402 IP_STATINC(IP_STAT_TOTAL); 403 404 /* 405 * If the IP header is not aligned, slurp it up into a new 406 * mbuf with space for link headers, in the event we forward 407 * it. Otherwise, if it is aligned, make sure the entire 408 * base IP header is in the first mbuf of the chain. 409 */ 410 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 411 if ((m = m_copyup(m, sizeof(struct ip), 412 (max_linkhdr + 3) & ~3)) == NULL) { 413 /* XXXJRT new stat, please */ 414 IP_STATINC(IP_STAT_TOOSMALL); 415 return; 416 } 417 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 418 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 419 IP_STATINC(IP_STAT_TOOSMALL); 420 return; 421 } 422 } 423 ip = mtod(m, struct ip *); 424 if (ip->ip_v != IPVERSION) { 425 IP_STATINC(IP_STAT_BADVERS); 426 goto bad; 427 } 428 hlen = ip->ip_hl << 2; 429 if (hlen < sizeof(struct ip)) { /* minimum header length */ 430 IP_STATINC(IP_STAT_BADHLEN); 431 goto bad; 432 } 433 if (hlen > m->m_len) { 434 if ((m = m_pullup(m, hlen)) == NULL) { 435 IP_STATINC(IP_STAT_BADHLEN); 436 return; 437 } 438 ip = mtod(m, struct ip *); 439 } 440 441 /* 442 * RFC1122: packets with a multicast source address are 443 * not allowed. 444 */ 445 if (IN_MULTICAST(ip->ip_src.s_addr)) { 446 IP_STATINC(IP_STAT_BADADDR); 447 goto bad; 448 } 449 450 /* 127/8 must not appear on wire - RFC1122 */ 451 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 452 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 453 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 454 IP_STATINC(IP_STAT_BADADDR); 455 goto bad; 456 } 457 } 458 459 switch (m->m_pkthdr.csum_flags & 460 ((ifp->if_csum_flags_rx & M_CSUM_IPv4) | 461 M_CSUM_IPv4_BAD)) { 462 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 463 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 464 goto badcsum; 465 466 case M_CSUM_IPv4: 467 /* Checksum was okay. */ 468 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 469 break; 470 471 default: 472 /* 473 * Must compute it ourselves. Maybe skip checksum on 474 * loopback interfaces. 475 */ 476 if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) || 477 ip_do_loopback_cksum)) { 478 INET_CSUM_COUNTER_INCR(&ip_swcsum); 479 if (in_cksum(m, hlen) != 0) 480 goto badcsum; 481 } 482 break; 483 } 484 485 /* Retrieve the packet length. */ 486 len = ntohs(ip->ip_len); 487 488 /* 489 * Check for additional length bogosity 490 */ 491 if (len < hlen) { 492 IP_STATINC(IP_STAT_BADLEN); 493 goto bad; 494 } 495 496 /* 497 * Check that the amount of data in the buffers 498 * is as at least much as the IP header would have us expect. 499 * Trim mbufs if longer than we expect. 500 * Drop packet if shorter than we expect. 501 */ 502 if (m->m_pkthdr.len < len) { 503 IP_STATINC(IP_STAT_TOOSHORT); 504 goto bad; 505 } 506 if (m->m_pkthdr.len > len) { 507 if (m->m_len == m->m_pkthdr.len) { 508 m->m_len = len; 509 m->m_pkthdr.len = len; 510 } else 511 m_adj(m, len - m->m_pkthdr.len); 512 } 513 514 /* 515 * Assume that we can create a fast-forward IP flow entry 516 * based on this packet. 517 */ 518 m->m_flags |= M_CANFASTFWD; 519 520 /* 521 * Run through list of hooks for input packets. If there are any 522 * filters which require that additional packets in the flow are 523 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 524 * Note that filters must _never_ set this flag, as another filter 525 * in the list may have previously cleared it. 526 */ 527#if defined(IPSEC) 528 if (!ipsec_used || !ipsec_indone(m)) 529#else 530 if (1) 531#endif 532 { 533 struct in_addr odst = ip->ip_dst; 534 bool freed; 535 536 SOFTNET_LOCK(); 537 freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0; 538 SOFTNET_UNLOCK(); 539 if (freed || m == NULL) { 540 return; 541 } 542 ip = mtod(m, struct ip *); 543 hlen = ip->ip_hl << 2; 544 545 /* 546 * XXX The setting of "srcrt" here is to prevent ip_forward() 547 * from generating ICMP redirects for packets that have 548 * been redirected by a hook back out on to the same LAN that 549 * they came from and is not an indication that the packet 550 * is being inffluenced by source routing options. This 551 * allows things like 552 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 553 * where tlp0 is both on the 1.1.1.0/24 network and is the 554 * default route for hosts on 1.1.1.0/24. Of course this 555 * also requires a "map tlp0 ..." to complete the story. 556 * One might argue whether or not this kind of network config. 557 * should be supported in this manner... 558 */ 559 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 560 } 561 562#ifdef ALTQ 563 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 564 if (altq_input) { 565 SOFTNET_LOCK(); 566 if ((*altq_input)(m, AF_INET) == 0) { 567 /* Packet dropped by traffic conditioner. */ 568 SOFTNET_UNLOCK(); 569 return; 570 } 571 SOFTNET_UNLOCK(); 572 } 573#endif 574 575 /* 576 * Process options and, if not destined for us, 577 * ship it on. ip_dooptions returns 1 when an 578 * error was detected (causing an icmp message 579 * to be sent and the original packet to be freed). 580 */ 581 ip_nhops = 0; /* for source routed packets */ 582 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 583 return; 584 585 /* 586 * Enable a consistency check between the destination address 587 * and the arrival interface for a unicast packet (the RFC 1122 588 * strong ES model) if IP forwarding is disabled and the packet 589 * is not locally generated. 590 * 591 * XXX - Checking also should be disabled if the destination 592 * address is ipnat'ed to a different interface. 593 * 594 * XXX - Checking is incompatible with IP aliases added 595 * to the loopback interface instead of the interface where 596 * the packets are received. 597 * 598 * XXX - We need to add a per ifaddr flag for this so that 599 * we get finer grain control. 600 */ 601 checkif = ip_checkinterface && (ipforwarding == 0) && 602 (ifp->if_flags & IFF_LOOPBACK) == 0; 603 604 /* 605 * Check our list of addresses, to see if the packet is for us. 606 * 607 * Traditional 4.4BSD did not consult IFF_UP at all. 608 * The behavior here is to treat addresses on !IFF_UP interface 609 * or IN_IFF_NOTREADY addresses as not mine. 610 */ 611 downmatch = 0; 612 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 613 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 614 if (ia->ia4_flags & IN_IFF_NOTREADY) 615 continue; 616 if (checkif && ia->ia_ifp != ifp) 617 continue; 618 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 619 break; 620 else 621 downmatch++; 622 } 623 } 624 if (ia != NULL) 625 goto ours; 626 if (ifp->if_flags & IFF_BROADCAST) { 627 IFADDR_FOREACH(ifa, ifp) { 628 if (ifa->ifa_addr->sa_family != AF_INET) 629 continue; 630 ia = ifatoia(ifa); 631 if (ia->ia4_flags & IN_IFF_NOTREADY) 632 continue; 633 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 634 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 635 /* 636 * Look for all-0's host part (old broadcast addr), 637 * either for subnet or net. 638 */ 639 ip->ip_dst.s_addr == ia->ia_subnet || 640 ip->ip_dst.s_addr == ia->ia_net) 641 goto ours; 642 /* 643 * An interface with IP address zero accepts 644 * all packets that arrive on that interface. 645 */ 646 if (in_nullhost(ia->ia_addr.sin_addr)) 647 goto ours; 648 } 649 } 650 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 651#ifdef MROUTING 652 extern struct socket *ip_mrouter; 653 654 if (ip_mrouter) { 655 /* 656 * If we are acting as a multicast router, all 657 * incoming multicast packets are passed to the 658 * kernel-level multicast forwarding function. 659 * The packet is returned (relatively) intact; if 660 * ip_mforward() returns a non-zero value, the packet 661 * must be discarded, else it may be accepted below. 662 * 663 * (The IP ident field is put in the same byte order 664 * as expected when ip_mforward() is called from 665 * ip_output().) 666 */ 667 SOFTNET_LOCK(); 668 if (ip_mforward(m, ifp) != 0) { 669 SOFTNET_UNLOCK(); 670 IP_STATINC(IP_STAT_CANTFORWARD); 671 m_freem(m); 672 return; 673 } 674 SOFTNET_UNLOCK(); 675 676 /* 677 * The process-level routing demon needs to receive 678 * all multicast IGMP packets, whether or not this 679 * host belongs to their destination groups. 680 */ 681 if (ip->ip_p == IPPROTO_IGMP) { 682 goto ours; 683 } 684 IP_STATINC(IP_STAT_CANTFORWARD); 685 } 686#endif 687 /* 688 * See if we belong to the destination multicast group on the 689 * arrival interface. 690 */ 691 if (!in_multi_group(ip->ip_dst, ifp, 0)) { 692 IP_STATINC(IP_STAT_CANTFORWARD); 693 m_freem(m); 694 return; 695 } 696 goto ours; 697 } 698 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 699 in_nullhost(ip->ip_dst)) 700 goto ours; 701 702 /* 703 * Not for us; forward if possible and desirable. 704 */ 705 if (ipforwarding == 0) { 706 IP_STATINC(IP_STAT_CANTFORWARD); 707 m_freem(m); 708 } else { 709 /* 710 * If ip_dst matched any of my address on !IFF_UP interface, 711 * and there's no IFF_UP interface that matches ip_dst, 712 * send icmp unreach. Forwarding it will result in in-kernel 713 * forwarding loop till TTL goes to 0. 714 */ 715 if (downmatch) { 716 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 717 IP_STATINC(IP_STAT_CANTFORWARD); 718 return; 719 } 720#ifdef IPSEC 721 /* Perform IPsec, if any. */ 722 if (ipsec_used) { 723 SOFTNET_LOCK(); 724 if (ipsec4_input(m, IP_FORWARDING | 725 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) { 726 SOFTNET_UNLOCK(); 727 goto bad; 728 } 729 SOFTNET_UNLOCK(); 730 } 731#endif 732 ip_forward(m, srcrt); 733 } 734 return; 735 736ours: 737 /* 738 * If offset or IP_MF are set, must reassemble. 739 */ 740 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 741 /* 742 * Pass to IP reassembly mechanism. 743 */ 744 if (ip_reass_packet(&m, ip) != 0) { 745 /* Failed; invalid fragment(s) or packet. */ 746 goto bad; 747 } 748 if (m == NULL) { 749 /* More fragments should come; silently return. */ 750 return; 751 } 752 /* 753 * Reassembly is done, we have the final packet. 754 * Updated cached data in local variable(s). 755 */ 756 ip = mtod(m, struct ip *); 757 hlen = ip->ip_hl << 2; 758 } 759 760#ifdef IPSEC 761 /* 762 * Enforce IPsec policy checking if we are seeing last header. 763 * Note that we do not visit this with protocols with PCB layer 764 * code - like UDP/TCP/raw IP. 765 */ 766 if (ipsec_used && 767 ((inetsw[ip_protox[ip->ip_p]].ippr_protosw.pr_flags & PR_LASTHDR) 768 != 0)) { 769 SOFTNET_LOCK(); 770 if (ipsec4_input(m, 0) != 0) { 771 SOFTNET_UNLOCK(); 772 goto bad; 773 } 774 SOFTNET_UNLOCK(); 775 } 776#endif 777 778 /* 779 * Switch out to protocol's input routine. 780 */ 781#if IFA_STATS 782 if (ia && ip) 783 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 784#endif 785 IP_STATINC(IP_STAT_DELIVERED); 786 787 const int off = hlen, nh = ip->ip_p; 788 789 SOFTNET_LOCK(); 790 (*inetsw[ip_protox[nh]].ippr_input)(m, off, nh); 791 SOFTNET_UNLOCK(); 792 return; 793bad: 794 m_freem(m); 795 return; 796 797badcsum: 798 IP_STATINC(IP_STAT_BADSUM); 799 m_freem(m); 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 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1003 sizeof(struct in_addr) > ipt->ipt_len) { 1004 code = (u_char *)&ipt->ipt_ptr - 1005 (u_char *)ip; 1006 goto bad; 1007 } 1008 ipaddr.sin_addr = dst; 1009 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1010 m->m_pkthdr.rcvif)); 1011 if (ia == 0) 1012 continue; 1013 bcopy(&ia->ia_addr.sin_addr, 1014 cp0, sizeof(struct in_addr)); 1015 ipt->ipt_ptr += sizeof(struct in_addr); 1016 break; 1017 1018 case IPOPT_TS_PRESPEC: 1019 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1020 sizeof(struct in_addr) > ipt->ipt_len) { 1021 code = (u_char *)&ipt->ipt_ptr - 1022 (u_char *)ip; 1023 goto bad; 1024 } 1025 memcpy(&ipaddr.sin_addr, cp0, 1026 sizeof(struct in_addr)); 1027 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1028 == NULL) 1029 continue; 1030 ipt->ipt_ptr += sizeof(struct in_addr); 1031 break; 1032 1033 default: 1034 /* XXX can't take &ipt->ipt_flg */ 1035 code = (u_char *)&ipt->ipt_ptr - 1036 (u_char *)ip + 1; 1037 goto bad; 1038 } 1039 ntime = iptime(); 1040 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1041 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1042 sizeof(n_time)); 1043 ipt->ipt_ptr += sizeof(n_time); 1044 } 1045 } 1046 if (forward) { 1047 if (ip_forwsrcrt == 0) { 1048 type = ICMP_UNREACH; 1049 code = ICMP_UNREACH_SRCFAIL; 1050 goto bad; 1051 } 1052 ip_forward(m, 1); 1053 return true; 1054 } 1055 return false; 1056bad: 1057 icmp_error(m, type, code, 0, 0); 1058 IP_STATINC(IP_STAT_BADOPTIONS); 1059 return true; 1060} 1061 1062/* 1063 * ip_rtaddr: given address of next destination (final or next hop), 1064 * return internet address info of interface to be used to get there. 1065 */ 1066static struct in_ifaddr * 1067ip_rtaddr(struct in_addr dst) 1068{ 1069 struct rtentry *rt; 1070 union { 1071 struct sockaddr dst; 1072 struct sockaddr_in dst4; 1073 } u; 1074 1075 sockaddr_in_init(&u.dst4, &dst, 0); 1076 1077 SOFTNET_LOCK(); 1078 rt = rtcache_lookup(&ipforward_rt, &u.dst); 1079 SOFTNET_UNLOCK(); 1080 if (rt == NULL) 1081 return NULL; 1082 1083 return ifatoia(rt->rt_ifa); 1084} 1085 1086/* 1087 * save_rte: save incoming source route for use in replies, to be picked 1088 * up later by ip_srcroute if the receiver is interested. 1089 */ 1090static void 1091save_rte(u_char *option, struct in_addr dst) 1092{ 1093 unsigned olen; 1094 1095 olen = option[IPOPT_OLEN]; 1096 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1097 return; 1098 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1099 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1100 ip_srcrt.dst = dst; 1101} 1102 1103/* 1104 * Retrieve incoming source route for use in replies, 1105 * in the same form used by setsockopt. 1106 * The first hop is placed before the options, will be removed later. 1107 */ 1108struct mbuf * 1109ip_srcroute(void) 1110{ 1111 struct in_addr *p, *q; 1112 struct mbuf *m; 1113 1114 if (ip_nhops == 0) 1115 return NULL; 1116 m = m_get(M_DONTWAIT, MT_SOOPTS); 1117 if (m == 0) 1118 return NULL; 1119 1120 MCLAIM(m, &inetdomain.dom_mowner); 1121#define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1122 1123 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1124 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1125 OPTSIZ; 1126 1127 /* 1128 * First save first hop for return route 1129 */ 1130 p = &ip_srcrt.route[ip_nhops - 1]; 1131 *(mtod(m, struct in_addr *)) = *p--; 1132 1133 /* 1134 * Copy option fields and padding (nop) to mbuf. 1135 */ 1136 ip_srcrt.nop = IPOPT_NOP; 1137 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1138 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1139 OPTSIZ); 1140 q = (struct in_addr *)(mtod(m, char *) + 1141 sizeof(struct in_addr) + OPTSIZ); 1142#undef OPTSIZ 1143 /* 1144 * Record return path as an IP source route, 1145 * reversing the path (pointers are now aligned). 1146 */ 1147 while (p >= ip_srcrt.route) { 1148 *q++ = *p--; 1149 } 1150 /* 1151 * Last hop goes to final destination. 1152 */ 1153 *q = ip_srcrt.dst; 1154 return (m); 1155} 1156 1157const int inetctlerrmap[PRC_NCMDS] = { 1158 [PRC_MSGSIZE] = EMSGSIZE, 1159 [PRC_HOSTDEAD] = EHOSTDOWN, 1160 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1161 [PRC_UNREACH_NET] = EHOSTUNREACH, 1162 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1163 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1164 [PRC_UNREACH_PORT] = ECONNREFUSED, 1165 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1166 [PRC_PARAMPROB] = ENOPROTOOPT, 1167}; 1168 1169void 1170ip_fasttimo(void) 1171{ 1172 if (ip_drainwanted) { 1173 ip_drain(); 1174 ip_drainwanted = 0; 1175 } 1176} 1177 1178void 1179ip_drainstub(void) 1180{ 1181 ip_drainwanted = 1; 1182} 1183 1184/* 1185 * Forward a packet. If some error occurs return the sender 1186 * an icmp packet. Note we can't always generate a meaningful 1187 * icmp message because icmp doesn't have a large enough repertoire 1188 * of codes and types. 1189 * 1190 * If not forwarding, just drop the packet. This could be confusing 1191 * if ipforwarding was zero but some routing protocol was advancing 1192 * us as a gateway to somewhere. However, we must let the routing 1193 * protocol deal with that. 1194 * 1195 * The srcrt parameter indicates whether the packet is being forwarded 1196 * via a source route. 1197 */ 1198static void 1199ip_forward(struct mbuf *m, int srcrt) 1200{ 1201 struct ip *ip = mtod(m, struct ip *); 1202 struct rtentry *rt; 1203 int error, type = 0, code = 0, destmtu = 0; 1204 struct mbuf *mcopy; 1205 n_long dest; 1206 union { 1207 struct sockaddr dst; 1208 struct sockaddr_in dst4; 1209 } u; 1210 uint64_t *ips; 1211 1212 KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software " 1213 "interrupt handler; synchronization assumptions violated"); 1214 1215 /* 1216 * We are now in the output path. 1217 */ 1218 MCLAIM(m, &ip_tx_mowner); 1219 1220 /* 1221 * Clear any in-bound checksum flags for this packet. 1222 */ 1223 m->m_pkthdr.csum_flags = 0; 1224 1225 dest = 0; 1226 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1227 IP_STATINC(IP_STAT_CANTFORWARD); 1228 m_freem(m); 1229 return; 1230 } 1231 1232 SOFTNET_LOCK(); 1233 1234 if (ip->ip_ttl <= IPTTLDEC) { 1235 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1236 SOFTNET_UNLOCK(); 1237 return; 1238 } 1239 1240 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1241 1242 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1243 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1244 SOFTNET_UNLOCK(); 1245 return; 1246 } 1247 1248 /* 1249 * Save at most 68 bytes of the packet in case 1250 * we need to generate an ICMP message to the src. 1251 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1252 */ 1253 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1254 if (mcopy) 1255 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1256 1257 ip->ip_ttl -= IPTTLDEC; 1258 1259 /* 1260 * If forwarding packet using same interface that it came in on, 1261 * perhaps should send a redirect to sender to shortcut a hop. 1262 * Only send redirect if source is sending directly to us, 1263 * and if packet was not source routed (or has any options). 1264 * Also, don't send redirect if forwarding using a default route 1265 * or a route modified by a redirect. 1266 */ 1267 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1268 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1269 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1270 ipsendredirects && !srcrt) { 1271 if (rt->rt_ifa && 1272 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1273 ifatoia(rt->rt_ifa)->ia_subnet) { 1274 if (rt->rt_flags & RTF_GATEWAY) 1275 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1276 else 1277 dest = ip->ip_dst.s_addr; 1278 /* 1279 * Router requirements says to only send host 1280 * redirects. 1281 */ 1282 type = ICMP_REDIRECT; 1283 code = ICMP_REDIRECT_HOST; 1284 } 1285 } 1286 1287 error = ip_output(m, NULL, &ipforward_rt, 1288 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1289 NULL, NULL); 1290 1291 if (error) { 1292 IP_STATINC(IP_STAT_CANTFORWARD); 1293 goto error; 1294 } 1295 1296 ips = IP_STAT_GETREF(); 1297 ips[IP_STAT_FORWARD]++; 1298 1299 if (type) { 1300 ips[IP_STAT_REDIRECTSENT]++; 1301 IP_STAT_PUTREF(); 1302 goto redirect; 1303 } 1304 1305 IP_STAT_PUTREF(); 1306 if (mcopy) { 1307#ifdef GATEWAY 1308 if (mcopy->m_flags & M_CANFASTFWD) 1309 ipflow_create(&ipforward_rt, mcopy); 1310#endif 1311 m_freem(mcopy); 1312 } 1313 1314 SOFTNET_UNLOCK(); 1315 return; 1316 1317redirect: 1318error: 1319 if (mcopy == NULL) { 1320 SOFTNET_UNLOCK(); 1321 return; 1322 } 1323 1324 switch (error) { 1325 1326 case 0: /* forwarded, but need redirect */ 1327 /* type, code set above */ 1328 break; 1329 1330 case ENETUNREACH: /* shouldn't happen, checked above */ 1331 case EHOSTUNREACH: 1332 case ENETDOWN: 1333 case EHOSTDOWN: 1334 default: 1335 type = ICMP_UNREACH; 1336 code = ICMP_UNREACH_HOST; 1337 break; 1338 1339 case EMSGSIZE: 1340 type = ICMP_UNREACH; 1341 code = ICMP_UNREACH_NEEDFRAG; 1342 1343 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1344 destmtu = rt->rt_ifp->if_mtu; 1345#ifdef IPSEC 1346 if (ipsec_used) 1347 (void)ipsec4_forward(mcopy, &destmtu); 1348#endif 1349 IP_STATINC(IP_STAT_CANTFRAG); 1350 break; 1351 1352 case ENOBUFS: 1353 /* 1354 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1355 * Requirements for IP Version 4 Routers. Source quench can 1356 * big problem under DoS attacks or if the underlying 1357 * interface is rate-limited. 1358 */ 1359 if (mcopy) 1360 m_freem(mcopy); 1361 SOFTNET_UNLOCK(); 1362 return; 1363 } 1364 icmp_error(mcopy, type, code, dest, destmtu); 1365 SOFTNET_UNLOCK(); 1366} 1367 1368void 1369ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1370 struct mbuf *m) 1371{ 1372 struct socket *so = inp->inp_socket; 1373 ifnet_t *ifp = m->m_pkthdr.rcvif; 1374 int inpflags = inp->inp_flags; 1375 1376 if (so->so_options & SO_TIMESTAMP 1377#ifdef SO_OTIMESTAMP 1378 || so->so_options & SO_OTIMESTAMP 1379#endif 1380 ) { 1381 struct timeval tv; 1382 1383 microtime(&tv); 1384#ifdef SO_OTIMESTAMP 1385 if (so->so_options & SO_OTIMESTAMP) { 1386 struct timeval50 tv50; 1387 timeval_to_timeval50(&tv, &tv50); 1388 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1389 SCM_OTIMESTAMP, SOL_SOCKET); 1390 } else 1391#endif 1392 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1393 SCM_TIMESTAMP, SOL_SOCKET); 1394 if (*mp) 1395 mp = &(*mp)->m_next; 1396 } 1397 if (inpflags & INP_RECVDSTADDR) { 1398 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1399 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1400 if (*mp) 1401 mp = &(*mp)->m_next; 1402 } 1403 if (inpflags & INP_RECVPKTINFO) { 1404 struct in_pktinfo ipi; 1405 ipi.ipi_addr = ip->ip_src; 1406 ipi.ipi_ifindex = ifp->if_index; 1407 *mp = sbcreatecontrol((void *) &ipi, 1408 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1409 if (*mp) 1410 mp = &(*mp)->m_next; 1411 } 1412 if (inpflags & INP_PKTINFO) { 1413 struct in_pktinfo ipi; 1414 ipi.ipi_addr = ip->ip_dst; 1415 ipi.ipi_ifindex = ifp->if_index; 1416 *mp = sbcreatecontrol((void *) &ipi, 1417 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1418 if (*mp) 1419 mp = &(*mp)->m_next; 1420 } 1421 if (inpflags & INP_RECVIF) { 1422 struct sockaddr_dl sdl; 1423 1424 sockaddr_dl_init(&sdl, sizeof(sdl), ifp ? 1425 ifp->if_index : 0, 0, NULL, 0, NULL, 0); 1426 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1427 if (*mp) 1428 mp = &(*mp)->m_next; 1429 } 1430 if (inpflags & INP_RECVTTL) { 1431 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1432 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1433 if (*mp) 1434 mp = &(*mp)->m_next; 1435 } 1436} 1437 1438/* 1439 * sysctl helper routine for net.inet.ip.forwsrcrt. 1440 */ 1441static int 1442sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1443{ 1444 int error, tmp; 1445 struct sysctlnode node; 1446 1447 node = *rnode; 1448 tmp = ip_forwsrcrt; 1449 node.sysctl_data = &tmp; 1450 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1451 if (error || newp == NULL) 1452 return (error); 1453 1454 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1455 0, NULL, NULL, NULL); 1456 if (error) 1457 return (error); 1458 1459 ip_forwsrcrt = tmp; 1460 1461 return (0); 1462} 1463 1464/* 1465 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1466 * range of the new value and tweaks timers if it changes. 1467 */ 1468static int 1469sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1470{ 1471 int error, tmp; 1472 struct sysctlnode node; 1473 1474 node = *rnode; 1475 tmp = ip_mtudisc_timeout; 1476 node.sysctl_data = &tmp; 1477 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1478 if (error || newp == NULL) 1479 return (error); 1480 if (tmp < 0) 1481 return (EINVAL); 1482 1483 mutex_enter(softnet_lock); 1484 1485 ip_mtudisc_timeout = tmp; 1486 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1487 1488 mutex_exit(softnet_lock); 1489 1490 return (0); 1491} 1492 1493static int 1494sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1495{ 1496 1497 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1498} 1499 1500static void 1501sysctl_net_inet_ip_setup(struct sysctllog **clog) 1502{ 1503 sysctl_createv(clog, 0, NULL, NULL, 1504 CTLFLAG_PERMANENT, 1505 CTLTYPE_NODE, "inet", 1506 SYSCTL_DESCR("PF_INET related settings"), 1507 NULL, 0, NULL, 0, 1508 CTL_NET, PF_INET, CTL_EOL); 1509 sysctl_createv(clog, 0, NULL, NULL, 1510 CTLFLAG_PERMANENT, 1511 CTLTYPE_NODE, "ip", 1512 SYSCTL_DESCR("IPv4 related settings"), 1513 NULL, 0, NULL, 0, 1514 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1515 1516 sysctl_createv(clog, 0, NULL, NULL, 1517 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1518 CTLTYPE_INT, "forwarding", 1519 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1520 NULL, 0, &ipforwarding, 0, 1521 CTL_NET, PF_INET, IPPROTO_IP, 1522 IPCTL_FORWARDING, CTL_EOL); 1523 sysctl_createv(clog, 0, NULL, NULL, 1524 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1525 CTLTYPE_INT, "redirect", 1526 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1527 NULL, 0, &ipsendredirects, 0, 1528 CTL_NET, PF_INET, IPPROTO_IP, 1529 IPCTL_SENDREDIRECTS, CTL_EOL); 1530 sysctl_createv(clog, 0, NULL, NULL, 1531 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1532 CTLTYPE_INT, "ttl", 1533 SYSCTL_DESCR("Default TTL for an INET datagram"), 1534 NULL, 0, &ip_defttl, 0, 1535 CTL_NET, PF_INET, IPPROTO_IP, 1536 IPCTL_DEFTTL, CTL_EOL); 1537#ifdef IPCTL_DEFMTU 1538 sysctl_createv(clog, 0, NULL, NULL, 1539 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1540 CTLTYPE_INT, "mtu", 1541 SYSCTL_DESCR("Default MTA for an INET route"), 1542 NULL, 0, &ip_mtu, 0, 1543 CTL_NET, PF_INET, IPPROTO_IP, 1544 IPCTL_DEFMTU, CTL_EOL); 1545#endif /* IPCTL_DEFMTU */ 1546 sysctl_createv(clog, 0, NULL, NULL, 1547 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1548 CTLTYPE_INT, "forwsrcrt", 1549 SYSCTL_DESCR("Enable forwarding of source-routed " 1550 "datagrams"), 1551 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1552 CTL_NET, PF_INET, IPPROTO_IP, 1553 IPCTL_FORWSRCRT, CTL_EOL); 1554 sysctl_createv(clog, 0, NULL, NULL, 1555 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1556 CTLTYPE_INT, "directed-broadcast", 1557 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1558 NULL, 0, &ip_directedbcast, 0, 1559 CTL_NET, PF_INET, IPPROTO_IP, 1560 IPCTL_DIRECTEDBCAST, CTL_EOL); 1561 sysctl_createv(clog, 0, NULL, NULL, 1562 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1563 CTLTYPE_INT, "allowsrcrt", 1564 SYSCTL_DESCR("Accept source-routed datagrams"), 1565 NULL, 0, &ip_allowsrcrt, 0, 1566 CTL_NET, PF_INET, IPPROTO_IP, 1567 IPCTL_ALLOWSRCRT, CTL_EOL); 1568 1569 sysctl_createv(clog, 0, NULL, NULL, 1570 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1571 CTLTYPE_INT, "mtudisc", 1572 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1573 NULL, 0, &ip_mtudisc, 0, 1574 CTL_NET, PF_INET, IPPROTO_IP, 1575 IPCTL_MTUDISC, CTL_EOL); 1576 sysctl_createv(clog, 0, NULL, NULL, 1577 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1578 CTLTYPE_INT, "anonportmin", 1579 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1580 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1581 CTL_NET, PF_INET, IPPROTO_IP, 1582 IPCTL_ANONPORTMIN, CTL_EOL); 1583 sysctl_createv(clog, 0, NULL, NULL, 1584 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1585 CTLTYPE_INT, "anonportmax", 1586 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1587 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1588 CTL_NET, PF_INET, IPPROTO_IP, 1589 IPCTL_ANONPORTMAX, CTL_EOL); 1590 sysctl_createv(clog, 0, NULL, NULL, 1591 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1592 CTLTYPE_INT, "mtudisctimeout", 1593 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1594 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1595 CTL_NET, PF_INET, IPPROTO_IP, 1596 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1597#ifndef IPNOPRIVPORTS 1598 sysctl_createv(clog, 0, NULL, NULL, 1599 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1600 CTLTYPE_INT, "lowportmin", 1601 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1602 "to assign"), 1603 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1604 CTL_NET, PF_INET, IPPROTO_IP, 1605 IPCTL_LOWPORTMIN, CTL_EOL); 1606 sysctl_createv(clog, 0, NULL, NULL, 1607 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1608 CTLTYPE_INT, "lowportmax", 1609 SYSCTL_DESCR("Highest privileged ephemeral port number " 1610 "to assign"), 1611 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1612 CTL_NET, PF_INET, IPPROTO_IP, 1613 IPCTL_LOWPORTMAX, CTL_EOL); 1614#endif /* IPNOPRIVPORTS */ 1615#if NGRE > 0 1616 sysctl_createv(clog, 0, NULL, NULL, 1617 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1618 CTLTYPE_INT, "grettl", 1619 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1620 NULL, 0, &ip_gre_ttl, 0, 1621 CTL_NET, PF_INET, IPPROTO_IP, 1622 IPCTL_GRE_TTL, CTL_EOL); 1623#endif /* NGRE */ 1624 sysctl_createv(clog, 0, NULL, NULL, 1625 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1626 CTLTYPE_INT, "checkinterface", 1627 SYSCTL_DESCR("Enable receive side of Strong ES model " 1628 "from RFC1122"), 1629 NULL, 0, &ip_checkinterface, 0, 1630 CTL_NET, PF_INET, IPPROTO_IP, 1631 IPCTL_CHECKINTERFACE, CTL_EOL); 1632 sysctl_createv(clog, 0, NULL, NULL, 1633 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1634 CTLTYPE_INT, "random_id", 1635 SYSCTL_DESCR("Assign random ip_id values"), 1636 NULL, 0, &ip_do_randomid, 0, 1637 CTL_NET, PF_INET, IPPROTO_IP, 1638 IPCTL_RANDOMID, CTL_EOL); 1639 sysctl_createv(clog, 0, NULL, NULL, 1640 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1641 CTLTYPE_INT, "do_loopback_cksum", 1642 SYSCTL_DESCR("Perform IP checksum on loopback"), 1643 NULL, 0, &ip_do_loopback_cksum, 0, 1644 CTL_NET, PF_INET, IPPROTO_IP, 1645 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1646 sysctl_createv(clog, 0, NULL, NULL, 1647 CTLFLAG_PERMANENT, 1648 CTLTYPE_STRUCT, "stats", 1649 SYSCTL_DESCR("IP statistics"), 1650 sysctl_net_inet_ip_stats, 0, NULL, 0, 1651 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1652 CTL_EOL); 1653#if NARP 1654 sysctl_createv(clog, 0, NULL, NULL, 1655 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1656 CTLTYPE_INT, "dad_count", 1657 SYSCTL_DESCR("Number of Duplicate Address Detection " 1658 "probes to send"), 1659 NULL, 0, &ip_dad_count, 0, 1660 CTL_NET, PF_INET, IPPROTO_IP, 1661 IPCTL_DAD_COUNT, CTL_EOL); 1662#endif 1663 1664 /* anonportalgo RFC6056 subtree */ 1665 const struct sysctlnode *portalgo_node; 1666 sysctl_createv(clog, 0, NULL, &portalgo_node, 1667 CTLFLAG_PERMANENT, 1668 CTLTYPE_NODE, "anonportalgo", 1669 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1670 NULL, 0, NULL, 0, 1671 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1672 sysctl_createv(clog, 0, &portalgo_node, NULL, 1673 CTLFLAG_PERMANENT, 1674 CTLTYPE_STRING, "available", 1675 SYSCTL_DESCR("available algorithms"), 1676 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1677 CTL_CREATE, CTL_EOL); 1678 sysctl_createv(clog, 0, &portalgo_node, NULL, 1679 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1680 CTLTYPE_STRING, "selected", 1681 SYSCTL_DESCR("selected algorithm"), 1682 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1683 CTL_CREATE, CTL_EOL); 1684 sysctl_createv(clog, 0, &portalgo_node, NULL, 1685 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1686 CTLTYPE_STRUCT, "reserve", 1687 SYSCTL_DESCR("bitmap of reserved ports"), 1688 sysctl_portalgo_reserve4, 0, NULL, 0, 1689 CTL_CREATE, CTL_EOL); 1690} 1691 1692void 1693ip_statinc(u_int stat) 1694{ 1695 1696 KASSERT(stat < IP_NSTATS); 1697 IP_STATINC(stat); 1698} 1699