ip_input.c revision 1.355
1/* $NetBSD: ip_input.c,v 1.355 2017/06/01 02:45:14 chs 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 1.355 2017/06/01 02:45:14 chs 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#include "opt_net_mpsafe.h" 105#endif 106 107#include "arp.h" 108 109#include <sys/param.h> 110#include <sys/systm.h> 111#include <sys/cpu.h> 112#include <sys/mbuf.h> 113#include <sys/domain.h> 114#include <sys/protosw.h> 115#include <sys/socket.h> 116#include <sys/socketvar.h> 117#include <sys/errno.h> 118#include <sys/time.h> 119#include <sys/kernel.h> 120#include <sys/pool.h> 121#include <sys/sysctl.h> 122#include <sys/kauth.h> 123 124#include <net/if.h> 125#include <net/if_dl.h> 126#include <net/route.h> 127#include <net/pktqueue.h> 128#include <net/pfil.h> 129 130#include <netinet/in.h> 131#include <netinet/in_systm.h> 132#include <netinet/ip.h> 133#include <netinet/in_pcb.h> 134#include <netinet/in_proto.h> 135#include <netinet/in_var.h> 136#include <netinet/ip_var.h> 137#include <netinet/ip_private.h> 138#include <netinet/ip_icmp.h> 139/* just for gif_ttl */ 140#include <netinet/in_gif.h> 141#include "gif.h" 142#include <net/if_gre.h> 143#include "gre.h" 144 145#ifdef MROUTING 146#include <netinet/ip_mroute.h> 147#endif 148#include <netinet/portalgo.h> 149 150#ifdef IPSEC 151#include <netipsec/ipsec.h> 152#endif 153 154#ifndef IPFORWARDING 155#ifdef GATEWAY 156#define IPFORWARDING 1 /* forward IP packets not for us */ 157#else /* GATEWAY */ 158#define IPFORWARDING 0 /* don't forward IP packets not for us */ 159#endif /* GATEWAY */ 160#endif /* IPFORWARDING */ 161#ifndef IPSENDREDIRECTS 162#define IPSENDREDIRECTS 1 163#endif 164#ifndef IPFORWSRCRT 165#define IPFORWSRCRT 1 /* forward source-routed packets */ 166#endif 167#ifndef IPALLOWSRCRT 168#define IPALLOWSRCRT 1 /* allow source-routed packets */ 169#endif 170#ifndef IPMTUDISC 171#define IPMTUDISC 1 172#endif 173#ifndef IPMTUDISCTIMEOUT 174#define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 175#endif 176 177#ifdef COMPAT_50 178#include <compat/sys/time.h> 179#include <compat/sys/socket.h> 180#endif 181 182/* 183 * Note: DIRECTED_BROADCAST is handled this way so that previous 184 * configuration using this option will Just Work. 185 */ 186#ifndef IPDIRECTEDBCAST 187#ifdef DIRECTED_BROADCAST 188#define IPDIRECTEDBCAST 1 189#else 190#define IPDIRECTEDBCAST 0 191#endif /* DIRECTED_BROADCAST */ 192#endif /* IPDIRECTEDBCAST */ 193int ipforwarding = IPFORWARDING; 194int ipsendredirects = IPSENDREDIRECTS; 195int ip_defttl = IPDEFTTL; 196int ip_forwsrcrt = IPFORWSRCRT; 197int ip_directedbcast = IPDIRECTEDBCAST; 198int ip_allowsrcrt = IPALLOWSRCRT; 199int ip_mtudisc = IPMTUDISC; 200int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 201#ifdef DIAGNOSTIC 202int ipprintfs = 0; 203#endif 204 205int ip_do_randomid = 0; 206 207/* 208 * XXX - Setting ip_checkinterface mostly implements the receive side of 209 * the Strong ES model described in RFC 1122, but since the routing table 210 * and transmit implementation do not implement the Strong ES model, 211 * setting this to 1 results in an odd hybrid. 212 * 213 * XXX - ip_checkinterface currently must be disabled if you use ipnat 214 * to translate the destination address to another local interface. 215 * 216 * XXX - ip_checkinterface must be disabled if you add IP aliases 217 * to the loopback interface instead of the interface where the 218 * packets for those addresses are received. 219 */ 220static int ip_checkinterface __read_mostly = 0; 221 222struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 223 224pktqueue_t * ip_pktq __read_mostly; 225pfil_head_t * inet_pfil_hook __read_mostly; 226ipid_state_t * ip_ids __read_mostly; 227percpu_t * ipstat_percpu __read_mostly; 228 229static percpu_t *ipforward_rt_percpu __cacheline_aligned; 230 231uint16_t ip_id; 232 233#ifdef INET_CSUM_COUNTERS 234#include <sys/device.h> 235 236struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 237 NULL, "inet", "hwcsum bad"); 238struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 239 NULL, "inet", "hwcsum ok"); 240struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 241 NULL, "inet", "swcsum"); 242 243#define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 244 245EVCNT_ATTACH_STATIC(ip_hwcsum_bad); 246EVCNT_ATTACH_STATIC(ip_hwcsum_ok); 247EVCNT_ATTACH_STATIC(ip_swcsum); 248 249#else 250 251#define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 252 253#endif /* INET_CSUM_COUNTERS */ 254 255/* 256 * Used to save the IP options in case a protocol wants to respond 257 * to an incoming packet over the same route if the packet got here 258 * using IP source routing. This allows connection establishment and 259 * maintenance when the remote end is on a network that is not known 260 * to us. 261 */ 262struct ip_srcrt { 263 int isr_nhops; /* number of hops */ 264 struct in_addr isr_dst; /* final destination */ 265 char isr_nop; /* one NOP to align */ 266 char isr_hdr[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN & OFFSET */ 267 struct in_addr isr_routes[MAX_IPOPTLEN/sizeof(struct in_addr)]; 268}; 269 270static int ip_drainwanted; 271 272static void save_rte(struct mbuf *, u_char *, struct in_addr); 273 274#ifdef MBUFTRACE 275struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 276struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 277#endif 278 279static void ipintr(void *); 280static void ip_input(struct mbuf *); 281static void ip_forward(struct mbuf *, int, struct ifnet *); 282static bool ip_dooptions(struct mbuf *); 283static struct in_ifaddr *ip_rtaddr(struct in_addr, struct psref *); 284static void sysctl_net_inet_ip_setup(struct sysctllog **); 285 286static struct in_ifaddr *ip_match_our_address(struct ifnet *, struct ip *, 287 int *); 288static struct in_ifaddr *ip_match_our_address_broadcast(struct ifnet *, 289 struct ip *); 290 291#ifdef NET_MPSAFE 292#define SOFTNET_LOCK() mutex_enter(softnet_lock) 293#define SOFTNET_UNLOCK() mutex_exit(softnet_lock) 294#else 295#define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock)) 296#define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock)) 297#endif 298 299/* 300 * IP initialization: fill in IP protocol switch table. 301 * All protocols not implemented in kernel go to raw IP protocol handler. 302 */ 303void 304ip_init(void) 305{ 306 const struct protosw *pr; 307 308 in_init(); 309 sysctl_net_inet_ip_setup(NULL); 310 311 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 312 KASSERT(pr != NULL); 313 314 ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL); 315 KASSERT(ip_pktq != NULL); 316 317 for (u_int i = 0; i < IPPROTO_MAX; i++) { 318 ip_protox[i] = pr - inetsw; 319 } 320 for (pr = inetdomain.dom_protosw; 321 pr < inetdomain.dom_protoswNPROTOSW; pr++) 322 if (pr->pr_domain->dom_family == PF_INET && 323 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 324 ip_protox[pr->pr_protocol] = pr - inetsw; 325 326 ip_reass_init(); 327 328 ip_ids = ip_id_init(); 329 ip_id = time_uptime & 0xfffff; 330 331#ifdef GATEWAY 332 ipflow_init(); 333#endif 334 335 /* Register our Packet Filter hook. */ 336 inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET); 337 KASSERT(inet_pfil_hook != NULL); 338 339#ifdef MBUFTRACE 340 MOWNER_ATTACH(&ip_tx_mowner); 341 MOWNER_ATTACH(&ip_rx_mowner); 342#endif /* MBUFTRACE */ 343 344 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 345 ipforward_rt_percpu = percpu_alloc(sizeof(struct route)); 346 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 347} 348 349static struct in_ifaddr * 350ip_match_our_address(struct ifnet *ifp, struct ip *ip, int *downmatch) 351{ 352 struct in_ifaddr *ia = NULL; 353 int checkif; 354 355 /* 356 * Enable a consistency check between the destination address 357 * and the arrival interface for a unicast packet (the RFC 1122 358 * strong ES model) if IP forwarding is disabled and the packet 359 * is not locally generated. 360 * 361 * XXX - Checking also should be disabled if the destination 362 * address is ipnat'ed to a different interface. 363 * 364 * XXX - Checking is incompatible with IP aliases added 365 * to the loopback interface instead of the interface where 366 * the packets are received. 367 * 368 * XXX - We need to add a per ifaddr flag for this so that 369 * we get finer grain control. 370 */ 371 checkif = ip_checkinterface && (ipforwarding == 0) && 372 (ifp->if_flags & IFF_LOOPBACK) == 0; 373 374 IN_ADDRHASH_READER_FOREACH(ia, ip->ip_dst.s_addr) { 375 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 376 if (ia->ia4_flags & IN_IFF_NOTREADY) 377 continue; 378 if (checkif && ia->ia_ifp != ifp) 379 continue; 380 if ((ia->ia_ifp->if_flags & IFF_UP) != 0 && 381 (ia->ia4_flags & IN_IFF_DETACHED) == 0) 382 break; 383 else 384 (*downmatch)++; 385 } 386 } 387 388 return ia; 389} 390 391static struct in_ifaddr * 392ip_match_our_address_broadcast(struct ifnet *ifp, struct ip *ip) 393{ 394 struct in_ifaddr *ia = NULL; 395 struct ifaddr *ifa; 396 397 IFADDR_READER_FOREACH(ifa, ifp) { 398 if (ifa->ifa_addr->sa_family != AF_INET) 399 continue; 400 ia = ifatoia(ifa); 401 if (ia->ia4_flags & (IN_IFF_NOTREADY | IN_IFF_DETACHED)) 402 continue; 403 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 404 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 405 /* 406 * Look for all-0's host part (old broadcast addr), 407 * either for subnet or net. 408 */ 409 ip->ip_dst.s_addr == ia->ia_subnet || 410 ip->ip_dst.s_addr == ia->ia_net) 411 goto matched; 412 /* 413 * An interface with IP address zero accepts 414 * all packets that arrive on that interface. 415 */ 416 if (in_nullhost(ia->ia_addr.sin_addr)) 417 goto matched; 418 } 419 ia = NULL; 420 421matched: 422 return ia; 423} 424 425/* 426 * IP software interrupt routine. 427 */ 428static void 429ipintr(void *arg __unused) 430{ 431 struct mbuf *m; 432 433 KASSERT(cpu_softintr_p()); 434 435#ifndef NET_MPSAFE 436 mutex_enter(softnet_lock); 437#endif 438 while ((m = pktq_dequeue(ip_pktq)) != NULL) { 439 ip_input(m); 440 } 441#ifndef NET_MPSAFE 442 mutex_exit(softnet_lock); 443#endif 444} 445 446/* 447 * IP input routine. Checksum and byte swap header. If fragmented 448 * try to reassemble. Process options. Pass to next level. 449 */ 450static void 451ip_input(struct mbuf *m) 452{ 453 struct ip *ip = NULL; 454 struct in_ifaddr *ia = NULL; 455 int hlen = 0, len; 456 int downmatch; 457 int srcrt = 0; 458 ifnet_t *ifp; 459 struct psref psref; 460 int s; 461 462 KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software " 463 "interrupt handler; synchronization assumptions violated"); 464 465 MCLAIM(m, &ip_rx_mowner); 466 KASSERT((m->m_flags & M_PKTHDR) != 0); 467 468 ifp = m_get_rcvif_psref(m, &psref); 469 if (__predict_false(ifp == NULL)) 470 goto out; 471 472 /* 473 * If no IP addresses have been set yet but the interfaces 474 * are receiving, can't do anything with incoming packets yet. 475 * Note: we pre-check without locks held. 476 */ 477 if (IN_ADDRLIST_READER_EMPTY()) 478 goto out; 479 IP_STATINC(IP_STAT_TOTAL); 480 481 /* 482 * If the IP header is not aligned, slurp it up into a new 483 * mbuf with space for link headers, in the event we forward 484 * it. Otherwise, if it is aligned, make sure the entire 485 * base IP header is in the first mbuf of the chain. 486 */ 487 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 488 if ((m = m_copyup(m, sizeof(struct ip), 489 (max_linkhdr + 3) & ~3)) == NULL) { 490 /* XXXJRT new stat, please */ 491 IP_STATINC(IP_STAT_TOOSMALL); 492 goto out; 493 } 494 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 495 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 496 IP_STATINC(IP_STAT_TOOSMALL); 497 goto out; 498 } 499 } 500 ip = mtod(m, struct ip *); 501 if (ip->ip_v != IPVERSION) { 502 IP_STATINC(IP_STAT_BADVERS); 503 goto out; 504 } 505 hlen = ip->ip_hl << 2; 506 if (hlen < sizeof(struct ip)) { /* minimum header length */ 507 IP_STATINC(IP_STAT_BADHLEN); 508 goto out; 509 } 510 if (hlen > m->m_len) { 511 if ((m = m_pullup(m, hlen)) == NULL) { 512 IP_STATINC(IP_STAT_BADHLEN); 513 goto out; 514 } 515 ip = mtod(m, struct ip *); 516 } 517 518 /* 519 * RFC1122: packets with a multicast source address are 520 * not allowed. 521 */ 522 if (IN_MULTICAST(ip->ip_src.s_addr)) { 523 IP_STATINC(IP_STAT_BADADDR); 524 goto out; 525 } 526 527 /* 127/8 must not appear on wire - RFC1122 */ 528 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 529 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 530 if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 531 IP_STATINC(IP_STAT_BADADDR); 532 goto out; 533 } 534 } 535 536 switch (m->m_pkthdr.csum_flags & 537 ((ifp->if_csum_flags_rx & M_CSUM_IPv4) | 538 M_CSUM_IPv4_BAD)) { 539 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 540 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 541 IP_STATINC(IP_STAT_BADSUM); 542 goto out; 543 544 case M_CSUM_IPv4: 545 /* Checksum was okay. */ 546 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 547 break; 548 549 default: 550 /* 551 * Must compute it ourselves. Maybe skip checksum on 552 * loopback interfaces. 553 */ 554 if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) || 555 ip_do_loopback_cksum)) { 556 INET_CSUM_COUNTER_INCR(&ip_swcsum); 557 if (in_cksum(m, hlen) != 0) { 558 IP_STATINC(IP_STAT_BADSUM); 559 goto out; 560 } 561 } 562 break; 563 } 564 565 /* Retrieve the packet length. */ 566 len = ntohs(ip->ip_len); 567 568 /* 569 * Check for additional length bogosity 570 */ 571 if (len < hlen) { 572 IP_STATINC(IP_STAT_BADLEN); 573 goto out; 574 } 575 576 /* 577 * Check that the amount of data in the buffers 578 * is as at least much as the IP header would have us expect. 579 * Trim mbufs if longer than we expect. 580 * Drop packet if shorter than we expect. 581 */ 582 if (m->m_pkthdr.len < len) { 583 IP_STATINC(IP_STAT_TOOSHORT); 584 goto out; 585 } 586 if (m->m_pkthdr.len > len) { 587 if (m->m_len == m->m_pkthdr.len) { 588 m->m_len = len; 589 m->m_pkthdr.len = len; 590 } else 591 m_adj(m, len - m->m_pkthdr.len); 592 } 593 594 /* 595 * Assume that we can create a fast-forward IP flow entry 596 * based on this packet. 597 */ 598 m->m_flags |= M_CANFASTFWD; 599 600 /* 601 * Run through list of hooks for input packets. If there are any 602 * filters which require that additional packets in the flow are 603 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 604 * Note that filters must _never_ set this flag, as another filter 605 * in the list may have previously cleared it. 606 */ 607#if defined(IPSEC) 608 if (!ipsec_used || !ipsec_indone(m)) 609#else 610 if (1) 611#endif 612 { 613 struct in_addr odst = ip->ip_dst; 614 bool freed; 615 616 freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0; 617 if (freed || m == NULL) { 618 m = NULL; 619 goto out; 620 } 621 ip = mtod(m, struct ip *); 622 hlen = ip->ip_hl << 2; 623 624 /* 625 * XXX The setting of "srcrt" here is to prevent ip_forward() 626 * from generating ICMP redirects for packets that have 627 * been redirected by a hook back out on to the same LAN that 628 * they came from and is not an indication that the packet 629 * is being inffluenced by source routing options. This 630 * allows things like 631 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 632 * where tlp0 is both on the 1.1.1.0/24 network and is the 633 * default route for hosts on 1.1.1.0/24. Of course this 634 * also requires a "map tlp0 ..." to complete the story. 635 * One might argue whether or not this kind of network config. 636 * should be supported in this manner... 637 */ 638 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 639 } 640 641#ifdef ALTQ 642 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 643 if (altq_input) { 644 SOFTNET_LOCK(); 645 if ((*altq_input)(m, AF_INET) == 0) { 646 /* Packet dropped by traffic conditioner. */ 647 SOFTNET_UNLOCK(); 648 m = NULL; 649 goto out; 650 } 651 SOFTNET_UNLOCK(); 652 } 653#endif 654 655 /* 656 * Process options and, if not destined for us, 657 * ship it on. ip_dooptions returns 1 when an 658 * error was detected (causing an icmp message 659 * to be sent and the original packet to be freed). 660 */ 661 if (hlen > sizeof (struct ip) && ip_dooptions(m)) { 662 m = NULL; 663 goto out; 664 } 665 666 /* 667 * Check our list of addresses, to see if the packet is for us. 668 * 669 * Traditional 4.4BSD did not consult IFF_UP at all. 670 * The behavior here is to treat addresses on !IFF_UP interface 671 * or IN_IFF_NOTREADY addresses as not mine. 672 */ 673 downmatch = 0; 674 s = pserialize_read_enter(); 675 ia = ip_match_our_address(ifp, ip, &downmatch); 676 if (ia != NULL) { 677 pserialize_read_exit(s); 678 goto ours; 679 } 680 681 if (ifp->if_flags & IFF_BROADCAST) { 682 ia = ip_match_our_address_broadcast(ifp, ip); 683 if (ia != NULL) { 684 pserialize_read_exit(s); 685 goto ours; 686 } 687 } 688 pserialize_read_exit(s); 689 690 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 691#ifdef MROUTING 692 extern struct socket *ip_mrouter; 693 694 if (ip_mrouter) { 695 /* 696 * If we are acting as a multicast router, all 697 * incoming multicast packets are passed to the 698 * kernel-level multicast forwarding function. 699 * The packet is returned (relatively) intact; if 700 * ip_mforward() returns a non-zero value, the packet 701 * must be discarded, else it may be accepted below. 702 * 703 * (The IP ident field is put in the same byte order 704 * as expected when ip_mforward() is called from 705 * ip_output().) 706 */ 707 SOFTNET_LOCK(); 708 if (ip_mforward(m, ifp) != 0) { 709 SOFTNET_UNLOCK(); 710 IP_STATINC(IP_STAT_CANTFORWARD); 711 goto out; 712 } 713 SOFTNET_UNLOCK(); 714 715 /* 716 * The process-level routing demon needs to receive 717 * all multicast IGMP packets, whether or not this 718 * host belongs to their destination groups. 719 */ 720 if (ip->ip_p == IPPROTO_IGMP) { 721 goto ours; 722 } 723 IP_STATINC(IP_STAT_CANTFORWARD); 724 } 725#endif 726 /* 727 * See if we belong to the destination multicast group on the 728 * arrival interface. 729 */ 730 if (!in_multi_group(ip->ip_dst, ifp, 0)) { 731 IP_STATINC(IP_STAT_CANTFORWARD); 732 goto out; 733 } 734 goto ours; 735 } 736 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 737 in_nullhost(ip->ip_dst)) 738 goto ours; 739 740 /* 741 * Not for us; forward if possible and desirable. 742 */ 743 if (ipforwarding == 0) { 744 m_put_rcvif_psref(ifp, &psref); 745 IP_STATINC(IP_STAT_CANTFORWARD); 746 m_freem(m); 747 } else { 748 /* 749 * If ip_dst matched any of my address on !IFF_UP interface, 750 * and there's no IFF_UP interface that matches ip_dst, 751 * send icmp unreach. Forwarding it will result in in-kernel 752 * forwarding loop till TTL goes to 0. 753 */ 754 if (downmatch) { 755 m_put_rcvif_psref(ifp, &psref); 756 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 757 IP_STATINC(IP_STAT_CANTFORWARD); 758 return; 759 } 760#ifdef IPSEC 761 /* Perform IPsec, if any. */ 762 if (ipsec_used) { 763 SOFTNET_LOCK(); 764 if (ipsec4_input(m, IP_FORWARDING | 765 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) { 766 SOFTNET_UNLOCK(); 767 goto out; 768 } 769 SOFTNET_UNLOCK(); 770 } 771#endif 772 ip_forward(m, srcrt, ifp); 773 m_put_rcvif_psref(ifp, &psref); 774 } 775 return; 776 777ours: 778 m_put_rcvif_psref(ifp, &psref); 779 ifp = NULL; 780 781 /* 782 * If offset or IP_MF are set, must reassemble. 783 */ 784 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 785 /* 786 * Pass to IP reassembly mechanism. 787 */ 788 if (ip_reass_packet(&m, ip) != 0) { 789 /* Failed; invalid fragment(s) or packet. */ 790 goto out; 791 } 792 if (m == NULL) { 793 /* More fragments should come; silently return. */ 794 goto out; 795 } 796 /* 797 * Reassembly is done, we have the final packet. 798 * Updated cached data in local variable(s). 799 */ 800 ip = mtod(m, struct ip *); 801 hlen = ip->ip_hl << 2; 802 } 803 804#ifdef IPSEC 805 /* 806 * Enforce IPsec policy checking if we are seeing last header. 807 * Note that we do not visit this with protocols with PCB layer 808 * code - like UDP/TCP/raw IP. 809 */ 810 if (ipsec_used && 811 (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 812 SOFTNET_LOCK(); 813 if (ipsec4_input(m, 0) != 0) { 814 SOFTNET_UNLOCK(); 815 goto out; 816 } 817 SOFTNET_UNLOCK(); 818 } 819#endif 820 821 /* 822 * Switch out to protocol's input routine. 823 */ 824#if IFA_STATS 825 if (ia && ip) { 826 struct in_ifaddr *_ia; 827 /* 828 * Keep a reference from ip_match_our_address with psref 829 * is expensive, so explore ia here again. 830 */ 831 s = pserialize_read_enter(); 832 _ia = in_get_ia(ip->ip_dst); 833 _ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 834 pserialize_read_exit(s); 835 } 836#endif 837 IP_STATINC(IP_STAT_DELIVERED); 838 839 const int off = hlen, nh = ip->ip_p; 840 841 SOFTNET_LOCK(); 842 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 843 SOFTNET_UNLOCK(); 844 return; 845 846out: 847 m_put_rcvif_psref(ifp, &psref); 848 if (m != NULL) 849 m_freem(m); 850} 851 852/* 853 * IP timer processing. 854 */ 855void 856ip_slowtimo(void) 857{ 858 859#ifndef NET_MPSAFE 860 mutex_enter(softnet_lock); 861 KERNEL_LOCK(1, NULL); 862#endif 863 864 ip_reass_slowtimo(); 865 866#ifndef NET_MPSAFE 867 KERNEL_UNLOCK_ONE(NULL); 868 mutex_exit(softnet_lock); 869#endif 870} 871 872/* 873 * IP drain processing. 874 */ 875void 876ip_drain(void) 877{ 878 879 KERNEL_LOCK(1, NULL); 880 ip_reass_drain(); 881 KERNEL_UNLOCK_ONE(NULL); 882} 883 884/* 885 * ip_dooptions: perform option processing on a datagram, possibly discarding 886 * it if bad options are encountered, or forwarding it if source-routed. 887 * 888 * => Returns true if packet has been forwarded/freed. 889 * => Returns false if the packet should be processed further. 890 */ 891static bool 892ip_dooptions(struct mbuf *m) 893{ 894 struct ip *ip = mtod(m, struct ip *); 895 u_char *cp, *cp0; 896 struct ip_timestamp *ipt; 897 struct in_ifaddr *ia; 898 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 899 struct in_addr dst; 900 n_time ntime; 901 struct ifaddr *ifa = NULL; 902 int s; 903 904 dst = ip->ip_dst; 905 cp = (u_char *)(ip + 1); 906 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 907 for (; cnt > 0; cnt -= optlen, cp += optlen) { 908 opt = cp[IPOPT_OPTVAL]; 909 if (opt == IPOPT_EOL) 910 break; 911 if (opt == IPOPT_NOP) 912 optlen = 1; 913 else { 914 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 915 code = &cp[IPOPT_OLEN] - (u_char *)ip; 916 goto bad; 917 } 918 optlen = cp[IPOPT_OLEN]; 919 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 920 code = &cp[IPOPT_OLEN] - (u_char *)ip; 921 goto bad; 922 } 923 } 924 switch (opt) { 925 926 default: 927 break; 928 929 /* 930 * Source routing with record. 931 * Find interface with current destination address. 932 * If none on this machine then drop if strictly routed, 933 * or do nothing if loosely routed. 934 * Record interface address and bring up next address 935 * component. If strictly routed make sure next 936 * address is on directly accessible net. 937 */ 938 case IPOPT_LSRR: 939 case IPOPT_SSRR: { 940 struct psref psref; 941 struct sockaddr_in ipaddr = { 942 .sin_len = sizeof(ipaddr), 943 .sin_family = AF_INET, 944 }; 945 946 if (ip_allowsrcrt == 0) { 947 type = ICMP_UNREACH; 948 code = ICMP_UNREACH_NET_PROHIB; 949 goto bad; 950 } 951 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 952 code = &cp[IPOPT_OLEN] - (u_char *)ip; 953 goto bad; 954 } 955 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 956 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 957 goto bad; 958 } 959 ipaddr.sin_addr = ip->ip_dst; 960 961 s = pserialize_read_enter(); 962 ifa = ifa_ifwithaddr(sintosa(&ipaddr)); 963 if (ifa == NULL) { 964 pserialize_read_exit(s); 965 if (opt == IPOPT_SSRR) { 966 type = ICMP_UNREACH; 967 code = ICMP_UNREACH_SRCFAIL; 968 goto bad; 969 } 970 /* 971 * Loose routing, and not at next destination 972 * yet; nothing to do except forward. 973 */ 974 break; 975 } 976 pserialize_read_exit(s); 977 978 off--; /* 0 origin */ 979 if ((off + sizeof(struct in_addr)) > optlen) { 980 /* 981 * End of source route. Should be for us. 982 */ 983 save_rte(m, cp, ip->ip_src); 984 break; 985 } 986 /* 987 * locate outgoing interface 988 */ 989 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 990 sizeof(ipaddr.sin_addr)); 991 if (opt == IPOPT_SSRR) { 992 ifa = ifa_ifwithladdr_psref(sintosa(&ipaddr), 993 &psref); 994 if (ifa != NULL) 995 ia = ifatoia(ifa); 996 else 997 ia = NULL; 998 } else { 999 ia = ip_rtaddr(ipaddr.sin_addr, &psref); 1000 } 1001 if (ia == NULL) { 1002 type = ICMP_UNREACH; 1003 code = ICMP_UNREACH_SRCFAIL; 1004 goto bad; 1005 } 1006 ip->ip_dst = ipaddr.sin_addr; 1007 bcopy((void *)&ia->ia_addr.sin_addr, 1008 (void *)(cp + off), sizeof(struct in_addr)); 1009 ia4_release(ia, &psref); 1010 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1011 /* 1012 * Let ip_intr's mcast routing check handle mcast pkts 1013 */ 1014 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1015 break; 1016 } 1017 1018 case IPOPT_RR: { 1019 struct psref psref; 1020 struct sockaddr_in ipaddr = { 1021 .sin_len = sizeof(ipaddr), 1022 .sin_family = AF_INET, 1023 }; 1024 1025 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1026 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1027 goto bad; 1028 } 1029 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1030 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1031 goto bad; 1032 } 1033 /* 1034 * If no space remains, ignore. 1035 */ 1036 off--; /* 0 origin */ 1037 if ((off + sizeof(struct in_addr)) > optlen) 1038 break; 1039 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 1040 sizeof(ipaddr.sin_addr)); 1041 /* 1042 * locate outgoing interface; if we're the destination, 1043 * use the incoming interface (should be same). 1044 */ 1045 ifa = ifa_ifwithaddr_psref(sintosa(&ipaddr), &psref); 1046 if (ifa == NULL) { 1047 ia = ip_rtaddr(ipaddr.sin_addr, &psref); 1048 if (ia == NULL) { 1049 type = ICMP_UNREACH; 1050 code = ICMP_UNREACH_HOST; 1051 goto bad; 1052 } 1053 } else { 1054 ia = ifatoia(ifa); 1055 } 1056 bcopy((void *)&ia->ia_addr.sin_addr, 1057 (void *)(cp + off), sizeof(struct in_addr)); 1058 ia4_release(ia, &psref); 1059 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1060 break; 1061 } 1062 1063 case IPOPT_TS: 1064 code = cp - (u_char *)ip; 1065 ipt = (struct ip_timestamp *)cp; 1066 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1067 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1068 goto bad; 1069 } 1070 if (ipt->ipt_ptr < 5) { 1071 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1072 goto bad; 1073 } 1074 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1075 if (++ipt->ipt_oflw == 0) { 1076 code = (u_char *)&ipt->ipt_ptr - 1077 (u_char *)ip; 1078 goto bad; 1079 } 1080 break; 1081 } 1082 cp0 = (cp + ipt->ipt_ptr - 1); 1083 switch (ipt->ipt_flg) { 1084 1085 case IPOPT_TS_TSONLY: 1086 break; 1087 1088 case IPOPT_TS_TSANDADDR: { 1089 struct ifnet *rcvif; 1090 int _s, _ss; 1091 struct sockaddr_in ipaddr = { 1092 .sin_len = sizeof(ipaddr), 1093 .sin_family = AF_INET, 1094 }; 1095 1096 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1097 sizeof(struct in_addr) > ipt->ipt_len) { 1098 code = (u_char *)&ipt->ipt_ptr - 1099 (u_char *)ip; 1100 goto bad; 1101 } 1102 ipaddr.sin_addr = dst; 1103 _ss = pserialize_read_enter(); 1104 rcvif = m_get_rcvif(m, &_s); 1105 if (__predict_true(rcvif != NULL)) { 1106 ifa = ifaof_ifpforaddr(sintosa(&ipaddr), 1107 rcvif); 1108 } 1109 m_put_rcvif(rcvif, &_s); 1110 if (ifa == NULL) { 1111 pserialize_read_exit(_ss); 1112 break; 1113 } 1114 ia = ifatoia(ifa); 1115 bcopy(&ia->ia_addr.sin_addr, 1116 cp0, sizeof(struct in_addr)); 1117 pserialize_read_exit(_ss); 1118 ipt->ipt_ptr += sizeof(struct in_addr); 1119 break; 1120 } 1121 1122 case IPOPT_TS_PRESPEC: { 1123 struct sockaddr_in ipaddr = { 1124 .sin_len = sizeof(ipaddr), 1125 .sin_family = AF_INET, 1126 }; 1127 1128 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1129 sizeof(struct in_addr) > ipt->ipt_len) { 1130 code = (u_char *)&ipt->ipt_ptr - 1131 (u_char *)ip; 1132 goto bad; 1133 } 1134 memcpy(&ipaddr.sin_addr, cp0, 1135 sizeof(struct in_addr)); 1136 s = pserialize_read_enter(); 1137 ifa = ifa_ifwithaddr(sintosa(&ipaddr)); 1138 if (ifa == NULL) { 1139 pserialize_read_exit(s); 1140 continue; 1141 } 1142 pserialize_read_exit(s); 1143 ipt->ipt_ptr += sizeof(struct in_addr); 1144 break; 1145 } 1146 1147 default: 1148 /* XXX can't take &ipt->ipt_flg */ 1149 code = (u_char *)&ipt->ipt_ptr - 1150 (u_char *)ip + 1; 1151 goto bad; 1152 } 1153 ntime = iptime(); 1154 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1155 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1156 sizeof(n_time)); 1157 ipt->ipt_ptr += sizeof(n_time); 1158 } 1159 } 1160 if (forward) { 1161 struct ifnet *rcvif; 1162 struct psref _psref; 1163 1164 if (ip_forwsrcrt == 0) { 1165 type = ICMP_UNREACH; 1166 code = ICMP_UNREACH_SRCFAIL; 1167 goto bad; 1168 } 1169 1170 rcvif = m_get_rcvif_psref(m, &_psref); 1171 if (__predict_false(rcvif == NULL)) { 1172 type = ICMP_UNREACH; 1173 code = ICMP_UNREACH_HOST; 1174 goto bad; 1175 } 1176 ip_forward(m, 1, rcvif); 1177 m_put_rcvif_psref(rcvif, &_psref); 1178 return true; 1179 } 1180 return false; 1181bad: 1182 icmp_error(m, type, code, 0, 0); 1183 IP_STATINC(IP_STAT_BADOPTIONS); 1184 return true; 1185} 1186 1187/* 1188 * ip_rtaddr: given address of next destination (final or next hop), 1189 * return internet address info of interface to be used to get there. 1190 */ 1191static struct in_ifaddr * 1192ip_rtaddr(struct in_addr dst, struct psref *psref) 1193{ 1194 struct rtentry *rt; 1195 union { 1196 struct sockaddr dst; 1197 struct sockaddr_in dst4; 1198 } u; 1199 struct route *ro; 1200 1201 sockaddr_in_init(&u.dst4, &dst, 0); 1202 1203 ro = percpu_getref(ipforward_rt_percpu); 1204 rt = rtcache_lookup(ro, &u.dst); 1205 if (rt == NULL) { 1206 percpu_putref(ipforward_rt_percpu); 1207 return NULL; 1208 } 1209 1210 ia4_acquire(ifatoia(rt->rt_ifa), psref); 1211 rtcache_unref(rt, ro); 1212 percpu_putref(ipforward_rt_percpu); 1213 1214 return ifatoia(rt->rt_ifa); 1215} 1216 1217/* 1218 * save_rte: save incoming source route for use in replies, to be picked 1219 * up later by ip_srcroute if the receiver is interested. 1220 */ 1221static void 1222save_rte(struct mbuf *m, u_char *option, struct in_addr dst) 1223{ 1224 struct ip_srcrt *isr; 1225 struct m_tag *mtag; 1226 unsigned olen; 1227 1228 olen = option[IPOPT_OLEN]; 1229 if (olen > sizeof(isr->isr_hdr) + sizeof(isr->isr_routes)) 1230 return; 1231 1232 mtag = m_tag_get(PACKET_TAG_SRCROUTE, sizeof(*isr), M_NOWAIT); 1233 if (mtag == NULL) 1234 return; 1235 isr = (struct ip_srcrt *)(mtag + 1); 1236 1237 memcpy(isr->isr_hdr, option, olen); 1238 isr->isr_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1239 isr->isr_dst = dst; 1240 m_tag_prepend(m, mtag); 1241} 1242 1243/* 1244 * Retrieve incoming source route for use in replies, 1245 * in the same form used by setsockopt. 1246 * The first hop is placed before the options, will be removed later. 1247 */ 1248struct mbuf * 1249ip_srcroute(struct mbuf *m0) 1250{ 1251 struct in_addr *p, *q; 1252 struct mbuf *m; 1253 struct ip_srcrt *isr; 1254 struct m_tag *mtag; 1255 1256 mtag = m_tag_find(m0, PACKET_TAG_SRCROUTE, NULL); 1257 if (mtag == NULL) 1258 return NULL; 1259 isr = (struct ip_srcrt *)(mtag + 1); 1260 1261 if (isr->isr_nhops == 0) 1262 return NULL; 1263 1264 m = m_get(M_DONTWAIT, MT_SOOPTS); 1265 if (m == NULL) 1266 return NULL; 1267 1268 MCLAIM(m, &inetdomain.dom_mowner); 1269#define OPTSIZ (sizeof(isr->isr_nop) + sizeof(isr->isr_hdr)) 1270 1271 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + header) */ 1272 m->m_len = (isr->isr_nhops + 1) * sizeof(struct in_addr) + OPTSIZ; 1273 1274 /* 1275 * First save first hop for return route 1276 */ 1277 p = &(isr->isr_routes[isr->isr_nhops - 1]); 1278 *(mtod(m, struct in_addr *)) = *p--; 1279 1280 /* 1281 * Copy option fields and padding (nop) to mbuf. 1282 */ 1283 isr->isr_nop = IPOPT_NOP; 1284 isr->isr_hdr[IPOPT_OFFSET] = IPOPT_MINOFF; 1285 memmove(mtod(m, char *) + sizeof(struct in_addr), &isr->isr_nop, 1286 OPTSIZ); 1287 q = (struct in_addr *)(mtod(m, char *) + 1288 sizeof(struct in_addr) + OPTSIZ); 1289#undef OPTSIZ 1290 /* 1291 * Record return path as an IP source route, 1292 * reversing the path (pointers are now aligned). 1293 */ 1294 while (p >= isr->isr_routes) { 1295 *q++ = *p--; 1296 } 1297 /* 1298 * Last hop goes to final destination. 1299 */ 1300 *q = isr->isr_dst; 1301 m_tag_delete(m0, mtag); 1302 return m; 1303} 1304 1305const int inetctlerrmap[PRC_NCMDS] = { 1306 [PRC_MSGSIZE] = EMSGSIZE, 1307 [PRC_HOSTDEAD] = EHOSTDOWN, 1308 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1309 [PRC_UNREACH_NET] = EHOSTUNREACH, 1310 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1311 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1312 [PRC_UNREACH_PORT] = ECONNREFUSED, 1313 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1314 [PRC_PARAMPROB] = ENOPROTOOPT, 1315}; 1316 1317void 1318ip_fasttimo(void) 1319{ 1320 if (ip_drainwanted) { 1321 ip_drain(); 1322 ip_drainwanted = 0; 1323 } 1324} 1325 1326void 1327ip_drainstub(void) 1328{ 1329 ip_drainwanted = 1; 1330} 1331 1332/* 1333 * Forward a packet. If some error occurs return the sender 1334 * an icmp packet. Note we can't always generate a meaningful 1335 * icmp message because icmp doesn't have a large enough repertoire 1336 * of codes and types. 1337 * 1338 * If not forwarding, just drop the packet. This could be confusing 1339 * if ipforwarding was zero but some routing protocol was advancing 1340 * us as a gateway to somewhere. However, we must let the routing 1341 * protocol deal with that. 1342 * 1343 * The srcrt parameter indicates whether the packet is being forwarded 1344 * via a source route. 1345 */ 1346static void 1347ip_forward(struct mbuf *m, int srcrt, struct ifnet *rcvif) 1348{ 1349 struct ip *ip = mtod(m, struct ip *); 1350 struct rtentry *rt; 1351 int error, type = 0, code = 0, destmtu = 0; 1352 struct mbuf *mcopy; 1353 n_long dest; 1354 union { 1355 struct sockaddr dst; 1356 struct sockaddr_in dst4; 1357 } u; 1358 uint64_t *ips; 1359 struct route *ro; 1360 1361 KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software " 1362 "interrupt handler; synchronization assumptions violated"); 1363 1364 /* 1365 * We are now in the output path. 1366 */ 1367 MCLAIM(m, &ip_tx_mowner); 1368 1369 /* 1370 * Clear any in-bound checksum flags for this packet. 1371 */ 1372 m->m_pkthdr.csum_flags = 0; 1373 1374 dest = 0; 1375 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1376 IP_STATINC(IP_STAT_CANTFORWARD); 1377 m_freem(m); 1378 return; 1379 } 1380 1381 if (ip->ip_ttl <= IPTTLDEC) { 1382 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1383 return; 1384 } 1385 1386 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1387 1388 ro = percpu_getref(ipforward_rt_percpu); 1389 rt = rtcache_lookup(ro, &u.dst); 1390 if (rt == NULL) { 1391 percpu_putref(ipforward_rt_percpu); 1392 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1393 return; 1394 } 1395 1396 /* 1397 * Save at most 68 bytes of the packet in case 1398 * we need to generate an ICMP message to the src. 1399 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1400 */ 1401 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1402 if (mcopy) 1403 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1404 1405 ip->ip_ttl -= IPTTLDEC; 1406 1407 /* 1408 * If forwarding packet using same interface that it came in on, 1409 * perhaps should send a redirect to sender to shortcut a hop. 1410 * Only send redirect if source is sending directly to us, 1411 * and if packet was not source routed (or has any options). 1412 * Also, don't send redirect if forwarding using a default route 1413 * or a route modified by a redirect. 1414 */ 1415 if (rt->rt_ifp == rcvif && 1416 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1417 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1418 ipsendredirects && !srcrt) { 1419 if (rt->rt_ifa && 1420 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1421 ifatoia(rt->rt_ifa)->ia_subnet) { 1422 if (rt->rt_flags & RTF_GATEWAY) 1423 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1424 else 1425 dest = ip->ip_dst.s_addr; 1426 /* 1427 * Router requirements says to only send host 1428 * redirects. 1429 */ 1430 type = ICMP_REDIRECT; 1431 code = ICMP_REDIRECT_HOST; 1432 } 1433 } 1434 rtcache_unref(rt, ro); 1435 1436 error = ip_output(m, NULL, ro, 1437 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1438 NULL, NULL); 1439 1440 if (error) { 1441 IP_STATINC(IP_STAT_CANTFORWARD); 1442 goto error; 1443 } 1444 1445 ips = IP_STAT_GETREF(); 1446 ips[IP_STAT_FORWARD]++; 1447 1448 if (type) { 1449 ips[IP_STAT_REDIRECTSENT]++; 1450 IP_STAT_PUTREF(); 1451 goto redirect; 1452 } 1453 1454 IP_STAT_PUTREF(); 1455 if (mcopy) { 1456#ifdef GATEWAY 1457 if (mcopy->m_flags & M_CANFASTFWD) 1458 ipflow_create(ro, mcopy); 1459#endif 1460 m_freem(mcopy); 1461 } 1462 1463 percpu_putref(ipforward_rt_percpu); 1464 return; 1465 1466redirect: 1467error: 1468 if (mcopy == NULL) { 1469 percpu_putref(ipforward_rt_percpu); 1470 return; 1471 } 1472 1473 switch (error) { 1474 1475 case 0: /* forwarded, but need redirect */ 1476 /* type, code set above */ 1477 break; 1478 1479 case ENETUNREACH: /* shouldn't happen, checked above */ 1480 case EHOSTUNREACH: 1481 case ENETDOWN: 1482 case EHOSTDOWN: 1483 default: 1484 type = ICMP_UNREACH; 1485 code = ICMP_UNREACH_HOST; 1486 break; 1487 1488 case EMSGSIZE: 1489 type = ICMP_UNREACH; 1490 code = ICMP_UNREACH_NEEDFRAG; 1491 1492 if ((rt = rtcache_validate(ro)) != NULL) { 1493 destmtu = rt->rt_ifp->if_mtu; 1494 rtcache_unref(rt, ro); 1495 } 1496#ifdef IPSEC 1497 if (ipsec_used) 1498 (void)ipsec4_forward(mcopy, &destmtu); 1499#endif 1500 IP_STATINC(IP_STAT_CANTFRAG); 1501 break; 1502 1503 case ENOBUFS: 1504 /* 1505 * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812, 1506 * Requirements for IP Version 4 Routers. Source quench can 1507 * big problem under DoS attacks or if the underlying 1508 * interface is rate-limited. 1509 */ 1510 if (mcopy) 1511 m_freem(mcopy); 1512 percpu_putref(ipforward_rt_percpu); 1513 return; 1514 } 1515 icmp_error(mcopy, type, code, dest, destmtu); 1516 percpu_putref(ipforward_rt_percpu); 1517} 1518 1519void 1520ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1521 struct mbuf *m) 1522{ 1523 struct socket *so = inp->inp_socket; 1524 ifnet_t *ifp; 1525 int inpflags = inp->inp_flags; 1526 struct psref psref; 1527 1528 ifp = m_get_rcvif_psref(m, &psref); 1529 if (__predict_false(ifp == NULL)) 1530 return; /* XXX should report error? */ 1531 1532 if (so->so_options & SO_TIMESTAMP 1533#ifdef SO_OTIMESTAMP 1534 || so->so_options & SO_OTIMESTAMP 1535#endif 1536 ) { 1537 struct timeval tv; 1538 1539 microtime(&tv); 1540#ifdef SO_OTIMESTAMP 1541 if (so->so_options & SO_OTIMESTAMP) { 1542 struct timeval50 tv50; 1543 timeval_to_timeval50(&tv, &tv50); 1544 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1545 SCM_OTIMESTAMP, SOL_SOCKET); 1546 } else 1547#endif 1548 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1549 SCM_TIMESTAMP, SOL_SOCKET); 1550 if (*mp) 1551 mp = &(*mp)->m_next; 1552 } 1553 if (inpflags & INP_RECVDSTADDR) { 1554 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1555 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1556 if (*mp) 1557 mp = &(*mp)->m_next; 1558 } 1559 if (inpflags & INP_RECVPKTINFO) { 1560 struct in_pktinfo ipi; 1561 ipi.ipi_addr = ip->ip_src; 1562 ipi.ipi_ifindex = ifp->if_index; 1563 *mp = sbcreatecontrol((void *) &ipi, 1564 sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP); 1565 if (*mp) 1566 mp = &(*mp)->m_next; 1567 } 1568 if (inpflags & INP_PKTINFO) { 1569 struct in_pktinfo ipi; 1570 ipi.ipi_addr = ip->ip_dst; 1571 ipi.ipi_ifindex = ifp->if_index; 1572 *mp = sbcreatecontrol((void *) &ipi, 1573 sizeof(ipi), IP_PKTINFO, IPPROTO_IP); 1574 if (*mp) 1575 mp = &(*mp)->m_next; 1576 } 1577 if (inpflags & INP_RECVIF) { 1578 struct sockaddr_dl sdl; 1579 1580 sockaddr_dl_init(&sdl, sizeof(sdl), ifp->if_index, 0, NULL, 0, 1581 NULL, 0); 1582 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1583 if (*mp) 1584 mp = &(*mp)->m_next; 1585 } 1586 if (inpflags & INP_RECVTTL) { 1587 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1588 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1589 if (*mp) 1590 mp = &(*mp)->m_next; 1591 } 1592 m_put_rcvif_psref(ifp, &psref); 1593} 1594 1595/* 1596 * sysctl helper routine for net.inet.ip.forwsrcrt. 1597 */ 1598static int 1599sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1600{ 1601 int error, tmp; 1602 struct sysctlnode node; 1603 1604 node = *rnode; 1605 tmp = ip_forwsrcrt; 1606 node.sysctl_data = &tmp; 1607 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1608 if (error || newp == NULL) 1609 return (error); 1610 1611 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1612 0, NULL, NULL, NULL); 1613 if (error) 1614 return (error); 1615 1616 ip_forwsrcrt = tmp; 1617 1618 return (0); 1619} 1620 1621/* 1622 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1623 * range of the new value and tweaks timers if it changes. 1624 */ 1625static int 1626sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1627{ 1628 int error, tmp; 1629 struct sysctlnode node; 1630 1631 icmp_mtudisc_lock(); 1632 1633 node = *rnode; 1634 tmp = ip_mtudisc_timeout; 1635 node.sysctl_data = &tmp; 1636 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1637 if (error || newp == NULL) 1638 goto out; 1639 if (tmp < 0) { 1640 error = EINVAL; 1641 goto out; 1642 } 1643 1644 ip_mtudisc_timeout = tmp; 1645 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1646 error = 0; 1647out: 1648 icmp_mtudisc_unlock(); 1649 return error; 1650} 1651 1652static int 1653sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1654{ 1655 1656 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1657} 1658 1659static void 1660sysctl_net_inet_ip_setup(struct sysctllog **clog) 1661{ 1662 sysctl_createv(clog, 0, NULL, NULL, 1663 CTLFLAG_PERMANENT, 1664 CTLTYPE_NODE, "inet", 1665 SYSCTL_DESCR("PF_INET related settings"), 1666 NULL, 0, NULL, 0, 1667 CTL_NET, PF_INET, CTL_EOL); 1668 sysctl_createv(clog, 0, NULL, NULL, 1669 CTLFLAG_PERMANENT, 1670 CTLTYPE_NODE, "ip", 1671 SYSCTL_DESCR("IPv4 related settings"), 1672 NULL, 0, NULL, 0, 1673 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1674 1675 sysctl_createv(clog, 0, NULL, NULL, 1676 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1677 CTLTYPE_INT, "forwarding", 1678 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1679 NULL, 0, &ipforwarding, 0, 1680 CTL_NET, PF_INET, IPPROTO_IP, 1681 IPCTL_FORWARDING, CTL_EOL); 1682 sysctl_createv(clog, 0, NULL, NULL, 1683 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1684 CTLTYPE_INT, "redirect", 1685 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1686 NULL, 0, &ipsendredirects, 0, 1687 CTL_NET, PF_INET, IPPROTO_IP, 1688 IPCTL_SENDREDIRECTS, CTL_EOL); 1689 sysctl_createv(clog, 0, NULL, NULL, 1690 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1691 CTLTYPE_INT, "ttl", 1692 SYSCTL_DESCR("Default TTL for an INET datagram"), 1693 NULL, 0, &ip_defttl, 0, 1694 CTL_NET, PF_INET, IPPROTO_IP, 1695 IPCTL_DEFTTL, CTL_EOL); 1696#ifdef IPCTL_DEFMTU 1697 sysctl_createv(clog, 0, NULL, NULL, 1698 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1699 CTLTYPE_INT, "mtu", 1700 SYSCTL_DESCR("Default MTA for an INET route"), 1701 NULL, 0, &ip_mtu, 0, 1702 CTL_NET, PF_INET, IPPROTO_IP, 1703 IPCTL_DEFMTU, CTL_EOL); 1704#endif /* IPCTL_DEFMTU */ 1705 sysctl_createv(clog, 0, NULL, NULL, 1706 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1707 CTLTYPE_INT, "forwsrcrt", 1708 SYSCTL_DESCR("Enable forwarding of source-routed " 1709 "datagrams"), 1710 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1711 CTL_NET, PF_INET, IPPROTO_IP, 1712 IPCTL_FORWSRCRT, CTL_EOL); 1713 sysctl_createv(clog, 0, NULL, NULL, 1714 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1715 CTLTYPE_INT, "directed-broadcast", 1716 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1717 NULL, 0, &ip_directedbcast, 0, 1718 CTL_NET, PF_INET, IPPROTO_IP, 1719 IPCTL_DIRECTEDBCAST, CTL_EOL); 1720 sysctl_createv(clog, 0, NULL, NULL, 1721 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1722 CTLTYPE_INT, "allowsrcrt", 1723 SYSCTL_DESCR("Accept source-routed datagrams"), 1724 NULL, 0, &ip_allowsrcrt, 0, 1725 CTL_NET, PF_INET, IPPROTO_IP, 1726 IPCTL_ALLOWSRCRT, CTL_EOL); 1727 1728 sysctl_createv(clog, 0, NULL, NULL, 1729 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1730 CTLTYPE_INT, "mtudisc", 1731 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1732 NULL, 0, &ip_mtudisc, 0, 1733 CTL_NET, PF_INET, IPPROTO_IP, 1734 IPCTL_MTUDISC, CTL_EOL); 1735 sysctl_createv(clog, 0, NULL, NULL, 1736 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1737 CTLTYPE_INT, "anonportmin", 1738 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1739 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1740 CTL_NET, PF_INET, IPPROTO_IP, 1741 IPCTL_ANONPORTMIN, CTL_EOL); 1742 sysctl_createv(clog, 0, NULL, NULL, 1743 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1744 CTLTYPE_INT, "anonportmax", 1745 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1746 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1747 CTL_NET, PF_INET, IPPROTO_IP, 1748 IPCTL_ANONPORTMAX, CTL_EOL); 1749 sysctl_createv(clog, 0, NULL, NULL, 1750 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1751 CTLTYPE_INT, "mtudisctimeout", 1752 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1753 sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0, 1754 CTL_NET, PF_INET, IPPROTO_IP, 1755 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1756#ifndef IPNOPRIVPORTS 1757 sysctl_createv(clog, 0, NULL, NULL, 1758 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1759 CTLTYPE_INT, "lowportmin", 1760 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1761 "to assign"), 1762 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1763 CTL_NET, PF_INET, IPPROTO_IP, 1764 IPCTL_LOWPORTMIN, CTL_EOL); 1765 sysctl_createv(clog, 0, NULL, NULL, 1766 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1767 CTLTYPE_INT, "lowportmax", 1768 SYSCTL_DESCR("Highest privileged ephemeral port number " 1769 "to assign"), 1770 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1771 CTL_NET, PF_INET, IPPROTO_IP, 1772 IPCTL_LOWPORTMAX, CTL_EOL); 1773#endif /* IPNOPRIVPORTS */ 1774#if NGRE > 0 1775 sysctl_createv(clog, 0, NULL, NULL, 1776 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1777 CTLTYPE_INT, "grettl", 1778 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1779 NULL, 0, &ip_gre_ttl, 0, 1780 CTL_NET, PF_INET, IPPROTO_IP, 1781 IPCTL_GRE_TTL, CTL_EOL); 1782#endif /* NGRE */ 1783 sysctl_createv(clog, 0, NULL, NULL, 1784 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1785 CTLTYPE_INT, "checkinterface", 1786 SYSCTL_DESCR("Enable receive side of Strong ES model " 1787 "from RFC1122"), 1788 NULL, 0, &ip_checkinterface, 0, 1789 CTL_NET, PF_INET, IPPROTO_IP, 1790 IPCTL_CHECKINTERFACE, CTL_EOL); 1791 sysctl_createv(clog, 0, NULL, NULL, 1792 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1793 CTLTYPE_INT, "random_id", 1794 SYSCTL_DESCR("Assign random ip_id values"), 1795 NULL, 0, &ip_do_randomid, 0, 1796 CTL_NET, PF_INET, IPPROTO_IP, 1797 IPCTL_RANDOMID, CTL_EOL); 1798 sysctl_createv(clog, 0, NULL, NULL, 1799 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1800 CTLTYPE_INT, "do_loopback_cksum", 1801 SYSCTL_DESCR("Perform IP checksum on loopback"), 1802 NULL, 0, &ip_do_loopback_cksum, 0, 1803 CTL_NET, PF_INET, IPPROTO_IP, 1804 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1805 sysctl_createv(clog, 0, NULL, NULL, 1806 CTLFLAG_PERMANENT, 1807 CTLTYPE_STRUCT, "stats", 1808 SYSCTL_DESCR("IP statistics"), 1809 sysctl_net_inet_ip_stats, 0, NULL, 0, 1810 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1811 CTL_EOL); 1812#if NARP 1813 sysctl_createv(clog, 0, NULL, NULL, 1814 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1815 CTLTYPE_INT, "dad_count", 1816 SYSCTL_DESCR("Number of Duplicate Address Detection " 1817 "probes to send"), 1818 NULL, 0, &ip_dad_count, 0, 1819 CTL_NET, PF_INET, IPPROTO_IP, 1820 IPCTL_DAD_COUNT, CTL_EOL); 1821#endif 1822 1823 /* anonportalgo RFC6056 subtree */ 1824 const struct sysctlnode *portalgo_node; 1825 sysctl_createv(clog, 0, NULL, &portalgo_node, 1826 CTLFLAG_PERMANENT, 1827 CTLTYPE_NODE, "anonportalgo", 1828 SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"), 1829 NULL, 0, NULL, 0, 1830 CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL); 1831 sysctl_createv(clog, 0, &portalgo_node, NULL, 1832 CTLFLAG_PERMANENT, 1833 CTLTYPE_STRING, "available", 1834 SYSCTL_DESCR("available algorithms"), 1835 sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN, 1836 CTL_CREATE, CTL_EOL); 1837 sysctl_createv(clog, 0, &portalgo_node, NULL, 1838 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1839 CTLTYPE_STRING, "selected", 1840 SYSCTL_DESCR("selected algorithm"), 1841 sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN, 1842 CTL_CREATE, CTL_EOL); 1843 sysctl_createv(clog, 0, &portalgo_node, NULL, 1844 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1845 CTLTYPE_STRUCT, "reserve", 1846 SYSCTL_DESCR("bitmap of reserved ports"), 1847 sysctl_portalgo_reserve4, 0, NULL, 0, 1848 CTL_CREATE, CTL_EOL); 1849} 1850 1851void 1852ip_statinc(u_int stat) 1853{ 1854 1855 KASSERT(stat < IP_NSTATS); 1856 IP_STATINC(stat); 1857} 1858