1 /* $NetBSD: ip_input.c,v 1.293 2010/12/13 14:18:50 matt 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 1.293 2010/12/13 14:18:50 matt Exp $"); 95 96 #include "opt_inet.h" 97 #include "opt_compat_netbsd.h" 98 #include "opt_gateway.h" 99 #include "opt_pfil_hooks.h" 100 #include "opt_ipsec.h" 101 #include "opt_mrouting.h" 102 #include "opt_mbuftrace.h" 103 #include "opt_inet_csum.h" 104 105 #include <sys/param.h> 106 #include <sys/systm.h> 107 #include <sys/mbuf.h> 108 #include <sys/domain.h> 109 #include <sys/protosw.h> 110 #include <sys/socket.h> 111 #include <sys/socketvar.h> 112 #include <sys/errno.h> 113 #include <sys/time.h> 114 #include <sys/kernel.h> 115 #include <sys/pool.h> 116 #include <sys/sysctl.h> 117 #include <sys/kauth.h> 118 119 #include <net/if.h> 120 #include <net/if_dl.h> 121 #include <net/route.h> 122 #include <net/pfil.h> 123 124 #include <netinet/in.h> 125 #include <netinet/in_systm.h> 126 #include <netinet/ip.h> 127 #include <netinet/in_pcb.h> 128 #include <netinet/in_proto.h> 129 #include <netinet/in_var.h> 130 #include <netinet/ip_var.h> 131 #include <netinet/ip_private.h> 132 #include <netinet/ip_icmp.h> 133 /* just for gif_ttl */ 134 #include <netinet/in_gif.h> 135 #include "gif.h" 136 #include <net/if_gre.h> 137 #include "gre.h" 138 139 #ifdef MROUTING 140 #include <netinet/ip_mroute.h> 141 #endif 142 143 #ifdef IPSEC 144 #include <netinet6/ipsec.h> 145 #include <netinet6/ipsec_private.h> 146 #include <netkey/key.h> 147 #endif 148 #ifdef FAST_IPSEC 149 #include <netipsec/ipsec.h> 150 #include <netipsec/key.h> 151 #endif /* FAST_IPSEC*/ 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 */ 192 int ipforwarding = IPFORWARDING; 193 int ipsendredirects = IPSENDREDIRECTS; 194 int ip_defttl = IPDEFTTL; 195 int ip_forwsrcrt = IPFORWSRCRT; 196 int ip_directedbcast = IPDIRECTEDBCAST; 197 int ip_allowsrcrt = IPALLOWSRCRT; 198 int ip_mtudisc = IPMTUDISC; 199 int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 200 #ifdef DIAGNOSTIC 201 int ipprintfs = 0; 202 #endif 203 204 int 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 */ 219 int ip_checkinterface = 0; 220 221 222 struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 223 224 int ipqmaxlen = IFQ_MAXLEN; 225 u_long in_ifaddrhash; /* size of hash table - 1 */ 226 int in_ifaddrentries; /* total number of addrs */ 227 struct in_ifaddrhead in_ifaddrhead; 228 struct in_ifaddrhashhead *in_ifaddrhashtbl; 229 u_long in_multihash; /* size of hash table - 1 */ 230 int in_multientries; /* total number of addrs */ 231 struct in_multihashhead *in_multihashtbl; 232 struct ifqueue ipintrq; 233 234 ipid_state_t * ip_ids; 235 uint16_t ip_id; 236 237 percpu_t *ipstat_percpu; 238 239 #ifdef PFIL_HOOKS 240 struct pfil_head inet_pfil_hook; 241 #endif 242 243 struct pool inmulti_pool; 244 245 #ifdef INET_CSUM_COUNTERS 246 #include <sys/device.h> 247 248 struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 249 NULL, "inet", "hwcsum bad"); 250 struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 251 NULL, "inet", "hwcsum ok"); 252 struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 253 NULL, "inet", "swcsum"); 254 255 #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 256 257 EVCNT_ATTACH_STATIC(ip_hwcsum_bad); 258 EVCNT_ATTACH_STATIC(ip_hwcsum_ok); 259 EVCNT_ATTACH_STATIC(ip_swcsum); 260 261 #else 262 263 #define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 264 265 #endif /* INET_CSUM_COUNTERS */ 266 267 /* 268 * We need to save the IP options in case a protocol wants to respond 269 * to an incoming packet over the same route if the packet got here 270 * using IP source routing. This allows connection establishment and 271 * maintenance when the remote end is on a network that is not known 272 * to us. 273 */ 274 int ip_nhops = 0; 275 static struct ip_srcrt { 276 struct in_addr dst; /* final destination */ 277 char nop; /* one NOP to align */ 278 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 279 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 280 } ip_srcrt; 281 282 static void save_rte(u_char *, struct in_addr); 283 284 #ifdef MBUFTRACE 285 struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); 286 struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); 287 #endif 288 289 static void sysctl_net_inet_ip_setup(struct sysctllog **); 290 291 /* 292 * IP initialization: fill in IP protocol switch table. 293 * All protocols not implemented in kernel go to raw IP protocol handler. 294 */ 295 void 296 ip_init(void) 297 { 298 const struct protosw *pr; 299 int i; 300 301 sysctl_net_inet_ip_setup(NULL); 302 303 pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", 304 NULL, IPL_SOFTNET); 305 306 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 307 if (pr == 0) 308 panic("ip_init"); 309 for (i = 0; i < IPPROTO_MAX; i++) 310 ip_protox[i] = pr - inetsw; 311 for (pr = inetdomain.dom_protosw; 312 pr < inetdomain.dom_protoswNPROTOSW; pr++) 313 if (pr->pr_domain->dom_family == PF_INET && 314 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 315 ip_protox[pr->pr_protocol] = pr - inetsw; 316 317 ip_reass_init(); 318 319 ip_ids = ip_id_init(); 320 ip_id = time_second & 0xfffff; 321 322 ipintrq.ifq_maxlen = ipqmaxlen; 323 324 TAILQ_INIT(&in_ifaddrhead); 325 in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true, 326 &in_ifaddrhash); 327 in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true, 328 &in_multihash); 329 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 330 #ifdef GATEWAY 331 ipflow_init(ip_hashsize); 332 #endif 333 334 #ifdef PFIL_HOOKS 335 /* Register our Packet Filter hook. */ 336 inet_pfil_hook.ph_type = PFIL_TYPE_AF; 337 inet_pfil_hook.ph_af = AF_INET; 338 i = pfil_head_register(&inet_pfil_hook); 339 if (i != 0) 340 printf("ip_init: WARNING: unable to register pfil hook, " 341 "error %d\n", i); 342 #endif /* PFIL_HOOKS */ 343 344 #ifdef MBUFTRACE 345 MOWNER_ATTACH(&ip_tx_mowner); 346 MOWNER_ATTACH(&ip_rx_mowner); 347 #endif /* MBUFTRACE */ 348 349 ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); 350 } 351 352 struct sockaddr_in ipaddr = { 353 .sin_len = sizeof(ipaddr), 354 .sin_family = AF_INET, 355 }; 356 struct route ipforward_rt; 357 358 /* 359 * IP software interrupt routine 360 */ 361 void 362 ipintr(void) 363 { 364 int s; 365 struct mbuf *m; 366 struct ifqueue lcl_intrq; 367 368 memset(&lcl_intrq, 0, sizeof(lcl_intrq)); 369 ipintrq.ifq_maxlen = ipqmaxlen; 370 371 mutex_enter(softnet_lock); 372 KERNEL_LOCK(1, NULL); 373 if (!IF_IS_EMPTY(&ipintrq)) { 374 s = splnet(); 375 376 /* Take existing queue onto stack */ 377 lcl_intrq = ipintrq; 378 379 /* Zero out global queue, preserving maxlen and drops */ 380 ipintrq.ifq_head = NULL; 381 ipintrq.ifq_tail = NULL; 382 ipintrq.ifq_len = 0; 383 ipintrq.ifq_maxlen = lcl_intrq.ifq_maxlen; 384 ipintrq.ifq_drops = lcl_intrq.ifq_drops; 385 386 splx(s); 387 } 388 KERNEL_UNLOCK_ONE(NULL); 389 while (!IF_IS_EMPTY(&lcl_intrq)) { 390 IF_DEQUEUE(&lcl_intrq, m); 391 if (m == NULL) 392 break; 393 ip_input(m); 394 } 395 mutex_exit(softnet_lock); 396 } 397 398 /* 399 * Ip input routine. Checksum and byte swap header. If fragmented 400 * try to reassemble. Process options. Pass to next level. 401 */ 402 void 403 ip_input(struct mbuf *m) 404 { 405 struct ip *ip = NULL; 406 struct in_ifaddr *ia; 407 struct ifaddr *ifa; 408 int hlen = 0, len; 409 int downmatch; 410 int checkif; 411 int srcrt = 0; 412 #ifdef FAST_IPSEC 413 struct m_tag *mtag; 414 struct tdb_ident *tdbi; 415 struct secpolicy *sp; 416 int error, s; 417 #endif /* FAST_IPSEC */ 418 419 MCLAIM(m, &ip_rx_mowner); 420 KASSERT((m->m_flags & M_PKTHDR) != 0); 421 422 /* 423 * If no IP addresses have been set yet but the interfaces 424 * are receiving, can't do anything with incoming packets yet. 425 */ 426 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 427 goto bad; 428 IP_STATINC(IP_STAT_TOTAL); 429 /* 430 * If the IP header is not aligned, slurp it up into a new 431 * mbuf with space for link headers, in the event we forward 432 * it. Otherwise, if it is aligned, make sure the entire 433 * base IP header is in the first mbuf of the chain. 434 */ 435 if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { 436 if ((m = m_copyup(m, sizeof(struct ip), 437 (max_linkhdr + 3) & ~3)) == NULL) { 438 /* XXXJRT new stat, please */ 439 IP_STATINC(IP_STAT_TOOSMALL); 440 return; 441 } 442 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 443 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 444 IP_STATINC(IP_STAT_TOOSMALL); 445 return; 446 } 447 } 448 ip = mtod(m, struct ip *); 449 if (ip->ip_v != IPVERSION) { 450 IP_STATINC(IP_STAT_BADVERS); 451 goto bad; 452 } 453 hlen = ip->ip_hl << 2; 454 if (hlen < sizeof(struct ip)) { /* minimum header length */ 455 IP_STATINC(IP_STAT_BADHLEN); 456 goto bad; 457 } 458 if (hlen > m->m_len) { 459 if ((m = m_pullup(m, hlen)) == 0) { 460 IP_STATINC(IP_STAT_BADHLEN); 461 return; 462 } 463 ip = mtod(m, struct ip *); 464 } 465 466 /* 467 * RFC1122: packets with a multicast source address are 468 * not allowed. 469 */ 470 if (IN_MULTICAST(ip->ip_src.s_addr)) { 471 IP_STATINC(IP_STAT_BADADDR); 472 goto bad; 473 } 474 475 /* 127/8 must not appear on wire - RFC1122 */ 476 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 477 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 478 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 479 IP_STATINC(IP_STAT_BADADDR); 480 goto bad; 481 } 482 } 483 484 switch (m->m_pkthdr.csum_flags & 485 ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) | 486 M_CSUM_IPv4_BAD)) { 487 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 488 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 489 goto badcsum; 490 491 case M_CSUM_IPv4: 492 /* Checksum was okay. */ 493 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 494 break; 495 496 default: 497 /* 498 * Must compute it ourselves. Maybe skip checksum on 499 * loopback interfaces. 500 */ 501 if (__predict_true(!(m->m_pkthdr.rcvif->if_flags & 502 IFF_LOOPBACK) || ip_do_loopback_cksum)) { 503 INET_CSUM_COUNTER_INCR(&ip_swcsum); 504 if (in_cksum(m, hlen) != 0) 505 goto badcsum; 506 } 507 break; 508 } 509 510 /* Retrieve the packet length. */ 511 len = ntohs(ip->ip_len); 512 513 /* 514 * Check for additional length bogosity 515 */ 516 if (len < hlen) { 517 IP_STATINC(IP_STAT_BADLEN); 518 goto bad; 519 } 520 521 /* 522 * Check that the amount of data in the buffers 523 * is as at least much as the IP header would have us expect. 524 * Trim mbufs if longer than we expect. 525 * Drop packet if shorter than we expect. 526 */ 527 if (m->m_pkthdr.len < len) { 528 IP_STATINC(IP_STAT_TOOSHORT); 529 goto bad; 530 } 531 if (m->m_pkthdr.len > len) { 532 if (m->m_len == m->m_pkthdr.len) { 533 m->m_len = len; 534 m->m_pkthdr.len = len; 535 } else 536 m_adj(m, len - m->m_pkthdr.len); 537 } 538 539 #if defined(IPSEC) 540 /* ipflow (IP fast forwarding) is not compatible with IPsec. */ 541 m->m_flags &= ~M_CANFASTFWD; 542 #else 543 /* 544 * Assume that we can create a fast-forward IP flow entry 545 * based on this packet. 546 */ 547 m->m_flags |= M_CANFASTFWD; 548 #endif 549 550 #ifdef PFIL_HOOKS 551 /* 552 * Run through list of hooks for input packets. If there are any 553 * filters which require that additional packets in the flow are 554 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 555 * Note that filters must _never_ set this flag, as another filter 556 * in the list may have previously cleared it. 557 */ 558 /* 559 * let ipfilter look at packet on the wire, 560 * not the decapsulated packet. 561 */ 562 #ifdef IPSEC 563 if (!ipsec_getnhist(m)) 564 #elif defined(FAST_IPSEC) 565 if (!ipsec_indone(m)) 566 #else 567 if (1) 568 #endif 569 { 570 struct in_addr odst; 571 572 odst = ip->ip_dst; 573 if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif, 574 PFIL_IN) != 0) 575 return; 576 if (m == NULL) 577 return; 578 ip = mtod(m, struct ip *); 579 hlen = ip->ip_hl << 2; 580 /* 581 * XXX The setting of "srcrt" here is to prevent ip_forward() 582 * from generating ICMP redirects for packets that have 583 * been redirected by a hook back out on to the same LAN that 584 * they came from and is not an indication that the packet 585 * is being inffluenced by source routing options. This 586 * allows things like 587 * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" 588 * where tlp0 is both on the 1.1.1.0/24 network and is the 589 * default route for hosts on 1.1.1.0/24. Of course this 590 * also requires a "map tlp0 ..." to complete the story. 591 * One might argue whether or not this kind of network config. 592 * should be supported in this manner... 593 */ 594 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 595 } 596 #endif /* PFIL_HOOKS */ 597 598 #ifdef ALTQ 599 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 600 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) { 601 /* packet dropped by traffic conditioner */ 602 return; 603 } 604 #endif 605 606 /* 607 * Process options and, if not destined for us, 608 * ship it on. ip_dooptions returns 1 when an 609 * error was detected (causing an icmp message 610 * to be sent and the original packet to be freed). 611 */ 612 ip_nhops = 0; /* for source routed packets */ 613 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 614 return; 615 616 /* 617 * Enable a consistency check between the destination address 618 * and the arrival interface for a unicast packet (the RFC 1122 619 * strong ES model) if IP forwarding is disabled and the packet 620 * is not locally generated. 621 * 622 * XXX - Checking also should be disabled if the destination 623 * address is ipnat'ed to a different interface. 624 * 625 * XXX - Checking is incompatible with IP aliases added 626 * to the loopback interface instead of the interface where 627 * the packets are received. 628 * 629 * XXX - We need to add a per ifaddr flag for this so that 630 * we get finer grain control. 631 */ 632 checkif = ip_checkinterface && (ipforwarding == 0) && 633 (m->m_pkthdr.rcvif != NULL) && 634 ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0); 635 636 /* 637 * Check our list of addresses, to see if the packet is for us. 638 * 639 * Traditional 4.4BSD did not consult IFF_UP at all. 640 * The behavior here is to treat addresses on !IFF_UP interface 641 * as not mine. 642 */ 643 downmatch = 0; 644 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 645 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 646 if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif) 647 continue; 648 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 649 break; 650 else 651 downmatch++; 652 } 653 } 654 if (ia != NULL) 655 goto ours; 656 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 657 IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) { 658 if (ifa->ifa_addr->sa_family != AF_INET) 659 continue; 660 ia = ifatoia(ifa); 661 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 662 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 663 /* 664 * Look for all-0's host part (old broadcast addr), 665 * either for subnet or net. 666 */ 667 ip->ip_dst.s_addr == ia->ia_subnet || 668 ip->ip_dst.s_addr == ia->ia_net) 669 goto ours; 670 /* 671 * An interface with IP address zero accepts 672 * all packets that arrive on that interface. 673 */ 674 if (in_nullhost(ia->ia_addr.sin_addr)) 675 goto ours; 676 } 677 } 678 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 679 struct in_multi *inm; 680 #ifdef MROUTING 681 extern struct socket *ip_mrouter; 682 683 if (ip_mrouter) { 684 /* 685 * If we are acting as a multicast router, all 686 * incoming multicast packets are passed to the 687 * kernel-level multicast forwarding function. 688 * The packet is returned (relatively) intact; if 689 * ip_mforward() returns a non-zero value, the packet 690 * must be discarded, else it may be accepted below. 691 * 692 * (The IP ident field is put in the same byte order 693 * as expected when ip_mforward() is called from 694 * ip_output().) 695 */ 696 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 697 IP_STATINC(IP_STAT_CANTFORWARD); 698 m_freem(m); 699 return; 700 } 701 702 /* 703 * The process-level routing demon needs to receive 704 * all multicast IGMP packets, whether or not this 705 * host belongs to their destination groups. 706 */ 707 if (ip->ip_p == IPPROTO_IGMP) 708 goto ours; 709 IP_STATINC(IP_STAT_CANTFORWARD); 710 } 711 #endif 712 /* 713 * See if we belong to the destination multicast group on the 714 * arrival interface. 715 */ 716 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 717 if (inm == NULL) { 718 IP_STATINC(IP_STAT_CANTFORWARD); 719 m_freem(m); 720 return; 721 } 722 goto ours; 723 } 724 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 725 in_nullhost(ip->ip_dst)) 726 goto ours; 727 728 /* 729 * Not for us; forward if possible and desirable. 730 */ 731 if (ipforwarding == 0) { 732 IP_STATINC(IP_STAT_CANTFORWARD); 733 m_freem(m); 734 } else { 735 /* 736 * If ip_dst matched any of my address on !IFF_UP interface, 737 * and there's no IFF_UP interface that matches ip_dst, 738 * send icmp unreach. Forwarding it will result in in-kernel 739 * forwarding loop till TTL goes to 0. 740 */ 741 if (downmatch) { 742 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 743 IP_STATINC(IP_STAT_CANTFORWARD); 744 return; 745 } 746 #ifdef IPSEC 747 if (ipsec4_in_reject(m, NULL)) { 748 IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); 749 goto bad; 750 } 751 #endif 752 #ifdef FAST_IPSEC 753 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 754 s = splsoftnet(); 755 if (mtag != NULL) { 756 tdbi = (struct tdb_ident *)(mtag + 1); 757 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 758 } else { 759 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 760 IP_FORWARDING, &error); 761 } 762 if (sp == NULL) { /* NB: can happen if error */ 763 splx(s); 764 /*XXX error stat???*/ 765 DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/ 766 goto bad; 767 } 768 769 /* 770 * Check security policy against packet attributes. 771 */ 772 error = ipsec_in_reject(sp, m); 773 KEY_FREESP(&sp); 774 splx(s); 775 if (error) { 776 IP_STATINC(IP_STAT_CANTFORWARD); 777 goto bad; 778 } 779 780 /* 781 * Peek at the outbound SP for this packet to determine if 782 * it's a Fast Forward candidate. 783 */ 784 mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL); 785 if (mtag != NULL) 786 m->m_flags &= ~M_CANFASTFWD; 787 else { 788 s = splsoftnet(); 789 sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND, 790 (IP_FORWARDING | 791 (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 792 &error, NULL); 793 if (sp != NULL) { 794 m->m_flags &= ~M_CANFASTFWD; 795 KEY_FREESP(&sp); 796 } 797 splx(s); 798 } 799 #endif /* FAST_IPSEC */ 800 801 ip_forward(m, srcrt); 802 } 803 return; 804 805 ours: 806 /* 807 * If offset or IP_MF are set, must reassemble. 808 */ 809 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 810 /* 811 * Pass to IP reassembly mechanism. 812 */ 813 if (ip_reass_packet(&m, ip) != 0) { 814 /* Failed; invalid fragment(s) or packet. */ 815 goto bad; 816 } 817 if (m == NULL) { 818 /* More fragments should come; silently return. */ 819 return; 820 } 821 /* 822 * Reassembly is done, we have the final packet. 823 * Updated cached data in local variable(s). 824 */ 825 ip = mtod(m, struct ip *); 826 hlen = ip->ip_hl << 2; 827 } 828 829 #if defined(IPSEC) 830 /* 831 * enforce IPsec policy checking if we are seeing last header. 832 * note that we do not visit this with protocols with pcb layer 833 * code - like udp/tcp/raw ip. 834 */ 835 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 && 836 ipsec4_in_reject(m, NULL)) { 837 IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); 838 goto bad; 839 } 840 #endif 841 #ifdef FAST_IPSEC 842 /* 843 * enforce IPsec policy checking if we are seeing last header. 844 * note that we do not visit this with protocols with pcb layer 845 * code - like udp/tcp/raw ip. 846 */ 847 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 848 /* 849 * Check if the packet has already had IPsec processing 850 * done. If so, then just pass it along. This tag gets 851 * set during AH, ESP, etc. input handling, before the 852 * packet is returned to the ip input queue for delivery. 853 */ 854 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 855 s = splsoftnet(); 856 if (mtag != NULL) { 857 tdbi = (struct tdb_ident *)(mtag + 1); 858 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 859 } else { 860 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 861 IP_FORWARDING, &error); 862 } 863 if (sp != NULL) { 864 /* 865 * Check security policy against packet attributes. 866 */ 867 error = ipsec_in_reject(sp, m); 868 KEY_FREESP(&sp); 869 } else { 870 /* XXX error stat??? */ 871 error = EINVAL; 872 DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/ 873 } 874 splx(s); 875 if (error) 876 goto bad; 877 } 878 #endif /* FAST_IPSEC */ 879 880 /* 881 * Switch out to protocol's input routine. 882 */ 883 #if IFA_STATS 884 if (ia && ip) 885 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 886 #endif 887 IP_STATINC(IP_STAT_DELIVERED); 888 { 889 int off = hlen, nh = ip->ip_p; 890 891 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 892 return; 893 } 894 bad: 895 m_freem(m); 896 return; 897 898 badcsum: 899 IP_STATINC(IP_STAT_BADSUM); 900 m_freem(m); 901 } 902 903 /* 904 * IP timer processing. 905 */ 906 void 907 ip_slowtimo(void) 908 { 909 910 mutex_enter(softnet_lock); 911 KERNEL_LOCK(1, NULL); 912 913 ip_reass_slowtimo(); 914 915 KERNEL_UNLOCK_ONE(NULL); 916 mutex_exit(softnet_lock); 917 } 918 919 /* 920 * IP drain processing. 921 */ 922 void 923 ip_drain(void) 924 { 925 926 KERNEL_LOCK(1, NULL); 927 ip_reass_drain(); 928 KERNEL_UNLOCK_ONE(NULL); 929 } 930 931 /* 932 * Do option processing on a datagram, 933 * possibly discarding it if bad options are encountered, 934 * or forwarding it if source-routed. 935 * Returns 1 if packet has been forwarded/freed, 936 * 0 if the packet should be processed further. 937 */ 938 int 939 ip_dooptions(struct mbuf *m) 940 { 941 struct ip *ip = mtod(m, struct ip *); 942 u_char *cp, *cp0; 943 struct ip_timestamp *ipt; 944 struct in_ifaddr *ia; 945 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 946 struct in_addr dst; 947 n_time ntime; 948 949 dst = ip->ip_dst; 950 cp = (u_char *)(ip + 1); 951 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 952 for (; cnt > 0; cnt -= optlen, cp += optlen) { 953 opt = cp[IPOPT_OPTVAL]; 954 if (opt == IPOPT_EOL) 955 break; 956 if (opt == IPOPT_NOP) 957 optlen = 1; 958 else { 959 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 960 code = &cp[IPOPT_OLEN] - (u_char *)ip; 961 goto bad; 962 } 963 optlen = cp[IPOPT_OLEN]; 964 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 965 code = &cp[IPOPT_OLEN] - (u_char *)ip; 966 goto bad; 967 } 968 } 969 switch (opt) { 970 971 default: 972 break; 973 974 /* 975 * Source routing with record. 976 * Find interface with current destination address. 977 * If none on this machine then drop if strictly routed, 978 * or do nothing if loosely routed. 979 * Record interface address and bring up next address 980 * component. If strictly routed make sure next 981 * address is on directly accessible net. 982 */ 983 case IPOPT_LSRR: 984 case IPOPT_SSRR: 985 if (ip_allowsrcrt == 0) { 986 type = ICMP_UNREACH; 987 code = ICMP_UNREACH_NET_PROHIB; 988 goto bad; 989 } 990 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 991 code = &cp[IPOPT_OLEN] - (u_char *)ip; 992 goto bad; 993 } 994 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 995 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 996 goto bad; 997 } 998 ipaddr.sin_addr = ip->ip_dst; 999 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1000 if (ia == 0) { 1001 if (opt == IPOPT_SSRR) { 1002 type = ICMP_UNREACH; 1003 code = ICMP_UNREACH_SRCFAIL; 1004 goto bad; 1005 } 1006 /* 1007 * Loose routing, and not at next destination 1008 * yet; nothing to do except forward. 1009 */ 1010 break; 1011 } 1012 off--; /* 0 origin */ 1013 if ((off + sizeof(struct in_addr)) > optlen) { 1014 /* 1015 * End of source route. Should be for us. 1016 */ 1017 save_rte(cp, ip->ip_src); 1018 break; 1019 } 1020 /* 1021 * locate outgoing interface 1022 */ 1023 memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off), 1024 sizeof(ipaddr.sin_addr)); 1025 if (opt == IPOPT_SSRR) 1026 ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); 1027 else 1028 ia = ip_rtaddr(ipaddr.sin_addr); 1029 if (ia == 0) { 1030 type = ICMP_UNREACH; 1031 code = ICMP_UNREACH_SRCFAIL; 1032 goto bad; 1033 } 1034 ip->ip_dst = ipaddr.sin_addr; 1035 bcopy((void *)&ia->ia_addr.sin_addr, 1036 (void *)(cp + off), sizeof(struct in_addr)); 1037 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1038 /* 1039 * Let ip_intr's mcast routing check handle mcast pkts 1040 */ 1041 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1042 break; 1043 1044 case IPOPT_RR: 1045 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1046 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1047 goto bad; 1048 } 1049 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1050 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1051 goto bad; 1052 } 1053 /* 1054 * If no space remains, ignore. 1055 */ 1056 off--; /* 0 origin */ 1057 if ((off + sizeof(struct in_addr)) > optlen) 1058 break; 1059 memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst), 1060 sizeof(ipaddr.sin_addr)); 1061 /* 1062 * locate outgoing interface; if we're the destination, 1063 * use the incoming interface (should be same). 1064 */ 1065 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 1066 == NULL && 1067 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 1068 type = ICMP_UNREACH; 1069 code = ICMP_UNREACH_HOST; 1070 goto bad; 1071 } 1072 bcopy((void *)&ia->ia_addr.sin_addr, 1073 (void *)(cp + off), sizeof(struct in_addr)); 1074 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1075 break; 1076 1077 case IPOPT_TS: 1078 code = cp - (u_char *)ip; 1079 ipt = (struct ip_timestamp *)cp; 1080 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1081 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1082 goto bad; 1083 } 1084 if (ipt->ipt_ptr < 5) { 1085 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1086 goto bad; 1087 } 1088 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1089 if (++ipt->ipt_oflw == 0) { 1090 code = (u_char *)&ipt->ipt_ptr - 1091 (u_char *)ip; 1092 goto bad; 1093 } 1094 break; 1095 } 1096 cp0 = (cp + ipt->ipt_ptr - 1); 1097 switch (ipt->ipt_flg) { 1098 1099 case IPOPT_TS_TSONLY: 1100 break; 1101 1102 case IPOPT_TS_TSANDADDR: 1103 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1104 sizeof(struct in_addr) > ipt->ipt_len) { 1105 code = (u_char *)&ipt->ipt_ptr - 1106 (u_char *)ip; 1107 goto bad; 1108 } 1109 ipaddr.sin_addr = dst; 1110 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1111 m->m_pkthdr.rcvif)); 1112 if (ia == 0) 1113 continue; 1114 bcopy(&ia->ia_addr.sin_addr, 1115 cp0, sizeof(struct in_addr)); 1116 ipt->ipt_ptr += sizeof(struct in_addr); 1117 break; 1118 1119 case IPOPT_TS_PRESPEC: 1120 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1121 sizeof(struct in_addr) > ipt->ipt_len) { 1122 code = (u_char *)&ipt->ipt_ptr - 1123 (u_char *)ip; 1124 goto bad; 1125 } 1126 memcpy(&ipaddr.sin_addr, cp0, 1127 sizeof(struct in_addr)); 1128 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1129 == NULL) 1130 continue; 1131 ipt->ipt_ptr += sizeof(struct in_addr); 1132 break; 1133 1134 default: 1135 /* XXX can't take &ipt->ipt_flg */ 1136 code = (u_char *)&ipt->ipt_ptr - 1137 (u_char *)ip + 1; 1138 goto bad; 1139 } 1140 ntime = iptime(); 1141 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1142 memmove((char *)cp + ipt->ipt_ptr - 1, cp0, 1143 sizeof(n_time)); 1144 ipt->ipt_ptr += sizeof(n_time); 1145 } 1146 } 1147 if (forward) { 1148 if (ip_forwsrcrt == 0) { 1149 type = ICMP_UNREACH; 1150 code = ICMP_UNREACH_SRCFAIL; 1151 goto bad; 1152 } 1153 ip_forward(m, 1); 1154 return (1); 1155 } 1156 return (0); 1157 bad: 1158 icmp_error(m, type, code, 0, 0); 1159 IP_STATINC(IP_STAT_BADOPTIONS); 1160 return (1); 1161 } 1162 1163 /* 1164 * Given address of next destination (final or next hop), 1165 * return internet address info of interface to be used to get there. 1166 */ 1167 struct in_ifaddr * 1168 ip_rtaddr(struct in_addr dst) 1169 { 1170 struct rtentry *rt; 1171 union { 1172 struct sockaddr dst; 1173 struct sockaddr_in dst4; 1174 } u; 1175 1176 sockaddr_in_init(&u.dst4, &dst, 0); 1177 1178 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) 1179 return NULL; 1180 1181 return ifatoia(rt->rt_ifa); 1182 } 1183 1184 /* 1185 * Save incoming source route for use in replies, 1186 * to be picked up later by ip_srcroute if the receiver is interested. 1187 */ 1188 void 1189 save_rte(u_char *option, struct in_addr dst) 1190 { 1191 unsigned olen; 1192 1193 olen = option[IPOPT_OLEN]; 1194 #ifdef DIAGNOSTIC 1195 if (ipprintfs) 1196 printf("save_rte: olen %d\n", olen); 1197 #endif /* 0 */ 1198 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1199 return; 1200 memcpy((void *)ip_srcrt.srcopt, (void *)option, olen); 1201 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1202 ip_srcrt.dst = dst; 1203 } 1204 1205 /* 1206 * Retrieve incoming source route for use in replies, 1207 * in the same form used by setsockopt. 1208 * The first hop is placed before the options, will be removed later. 1209 */ 1210 struct mbuf * 1211 ip_srcroute(void) 1212 { 1213 struct in_addr *p, *q; 1214 struct mbuf *m; 1215 1216 if (ip_nhops == 0) 1217 return NULL; 1218 m = m_get(M_DONTWAIT, MT_SOOPTS); 1219 if (m == 0) 1220 return NULL; 1221 1222 MCLAIM(m, &inetdomain.dom_mowner); 1223 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1224 1225 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1226 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1227 OPTSIZ; 1228 #ifdef DIAGNOSTIC 1229 if (ipprintfs) 1230 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1231 #endif 1232 1233 /* 1234 * First save first hop for return route 1235 */ 1236 p = &ip_srcrt.route[ip_nhops - 1]; 1237 *(mtod(m, struct in_addr *)) = *p--; 1238 #ifdef DIAGNOSTIC 1239 if (ipprintfs) 1240 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1241 #endif 1242 1243 /* 1244 * Copy option fields and padding (nop) to mbuf. 1245 */ 1246 ip_srcrt.nop = IPOPT_NOP; 1247 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1248 memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, 1249 OPTSIZ); 1250 q = (struct in_addr *)(mtod(m, char *) + 1251 sizeof(struct in_addr) + OPTSIZ); 1252 #undef OPTSIZ 1253 /* 1254 * Record return path as an IP source route, 1255 * reversing the path (pointers are now aligned). 1256 */ 1257 while (p >= ip_srcrt.route) { 1258 #ifdef DIAGNOSTIC 1259 if (ipprintfs) 1260 printf(" %x", ntohl(q->s_addr)); 1261 #endif 1262 *q++ = *p--; 1263 } 1264 /* 1265 * Last hop goes to final destination. 1266 */ 1267 *q = ip_srcrt.dst; 1268 #ifdef DIAGNOSTIC 1269 if (ipprintfs) 1270 printf(" %x\n", ntohl(q->s_addr)); 1271 #endif 1272 return (m); 1273 } 1274 1275 const int inetctlerrmap[PRC_NCMDS] = { 1276 [PRC_MSGSIZE] = EMSGSIZE, 1277 [PRC_HOSTDEAD] = EHOSTDOWN, 1278 [PRC_HOSTUNREACH] = EHOSTUNREACH, 1279 [PRC_UNREACH_NET] = EHOSTUNREACH, 1280 [PRC_UNREACH_HOST] = EHOSTUNREACH, 1281 [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, 1282 [PRC_UNREACH_PORT] = ECONNREFUSED, 1283 [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, 1284 [PRC_PARAMPROB] = ENOPROTOOPT, 1285 }; 1286 1287 /* 1288 * Forward a packet. If some error occurs return the sender 1289 * an icmp packet. Note we can't always generate a meaningful 1290 * icmp message because icmp doesn't have a large enough repertoire 1291 * of codes and types. 1292 * 1293 * If not forwarding, just drop the packet. This could be confusing 1294 * if ipforwarding was zero but some routing protocol was advancing 1295 * us as a gateway to somewhere. However, we must let the routing 1296 * protocol deal with that. 1297 * 1298 * The srcrt parameter indicates whether the packet is being forwarded 1299 * via a source route. 1300 */ 1301 void 1302 ip_forward(struct mbuf *m, int srcrt) 1303 { 1304 struct ip *ip = mtod(m, struct ip *); 1305 struct rtentry *rt; 1306 int error, type = 0, code = 0, destmtu = 0; 1307 struct mbuf *mcopy; 1308 n_long dest; 1309 union { 1310 struct sockaddr dst; 1311 struct sockaddr_in dst4; 1312 } u; 1313 1314 /* 1315 * We are now in the output path. 1316 */ 1317 MCLAIM(m, &ip_tx_mowner); 1318 1319 /* 1320 * Clear any in-bound checksum flags for this packet. 1321 */ 1322 m->m_pkthdr.csum_flags = 0; 1323 1324 dest = 0; 1325 #ifdef DIAGNOSTIC 1326 if (ipprintfs) { 1327 printf("forward: src %s ", inet_ntoa(ip->ip_src)); 1328 printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl); 1329 } 1330 #endif 1331 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1332 IP_STATINC(IP_STAT_CANTFORWARD); 1333 m_freem(m); 1334 return; 1335 } 1336 if (ip->ip_ttl <= IPTTLDEC) { 1337 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1338 return; 1339 } 1340 1341 sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); 1342 if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { 1343 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); 1344 return; 1345 } 1346 1347 /* 1348 * Save at most 68 bytes of the packet in case 1349 * we need to generate an ICMP message to the src. 1350 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1351 */ 1352 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1353 if (mcopy) 1354 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1355 1356 ip->ip_ttl -= IPTTLDEC; 1357 1358 /* 1359 * If forwarding packet using same interface that it came in on, 1360 * perhaps should send a redirect to sender to shortcut a hop. 1361 * Only send redirect if source is sending directly to us, 1362 * and if packet was not source routed (or has any options). 1363 * Also, don't send redirect if forwarding using a default route 1364 * or a route modified by a redirect. 1365 */ 1366 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1367 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1368 !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && 1369 ipsendredirects && !srcrt) { 1370 if (rt->rt_ifa && 1371 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1372 ifatoia(rt->rt_ifa)->ia_subnet) { 1373 if (rt->rt_flags & RTF_GATEWAY) 1374 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1375 else 1376 dest = ip->ip_dst.s_addr; 1377 /* 1378 * Router requirements says to only send host 1379 * redirects. 1380 */ 1381 type = ICMP_REDIRECT; 1382 code = ICMP_REDIRECT_HOST; 1383 #ifdef DIAGNOSTIC 1384 if (ipprintfs) 1385 printf("redirect (%d) to %x\n", code, 1386 (u_int32_t)dest); 1387 #endif 1388 } 1389 } 1390 1391 error = ip_output(m, NULL, &ipforward_rt, 1392 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1393 (struct ip_moptions *)NULL, (struct socket *)NULL); 1394 1395 if (error) 1396 IP_STATINC(IP_STAT_CANTFORWARD); 1397 else { 1398 uint64_t *ips = IP_STAT_GETREF(); 1399 ips[IP_STAT_FORWARD]++; 1400 if (type) { 1401 ips[IP_STAT_REDIRECTSENT]++; 1402 IP_STAT_PUTREF(); 1403 } else { 1404 IP_STAT_PUTREF(); 1405 if (mcopy) { 1406 #ifdef GATEWAY 1407 if (mcopy->m_flags & M_CANFASTFWD) 1408 ipflow_create(&ipforward_rt, mcopy); 1409 #endif 1410 m_freem(mcopy); 1411 } 1412 return; 1413 } 1414 } 1415 if (mcopy == NULL) 1416 return; 1417 1418 switch (error) { 1419 1420 case 0: /* forwarded, but need redirect */ 1421 /* type, code set above */ 1422 break; 1423 1424 case ENETUNREACH: /* shouldn't happen, checked above */ 1425 case EHOSTUNREACH: 1426 case ENETDOWN: 1427 case EHOSTDOWN: 1428 default: 1429 type = ICMP_UNREACH; 1430 code = ICMP_UNREACH_HOST; 1431 break; 1432 1433 case EMSGSIZE: 1434 type = ICMP_UNREACH; 1435 code = ICMP_UNREACH_NEEDFRAG; 1436 1437 if ((rt = rtcache_validate(&ipforward_rt)) != NULL) 1438 destmtu = rt->rt_ifp->if_mtu; 1439 1440 #if defined(IPSEC) || defined(FAST_IPSEC) 1441 { 1442 /* 1443 * If the packet is routed over IPsec tunnel, tell the 1444 * originator the tunnel MTU. 1445 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz 1446 * XXX quickhack!!! 1447 */ 1448 1449 struct secpolicy *sp; 1450 int ipsecerror; 1451 size_t ipsechdr; 1452 struct route *ro; 1453 1454 sp = ipsec4_getpolicybyaddr(mcopy, 1455 IPSEC_DIR_OUTBOUND, IP_FORWARDING, 1456 &ipsecerror); 1457 1458 if (sp != NULL) { 1459 /* count IPsec header size */ 1460 ipsechdr = ipsec4_hdrsiz(mcopy, 1461 IPSEC_DIR_OUTBOUND, NULL); 1462 1463 /* 1464 * find the correct route for outer IPv4 1465 * header, compute tunnel MTU. 1466 */ 1467 1468 if (sp->req != NULL 1469 && sp->req->sav != NULL 1470 && sp->req->sav->sah != NULL) { 1471 ro = &sp->req->sav->sah->sa_route; 1472 rt = rtcache_validate(ro); 1473 if (rt && rt->rt_ifp) { 1474 destmtu = 1475 rt->rt_rmx.rmx_mtu ? 1476 rt->rt_rmx.rmx_mtu : 1477 rt->rt_ifp->if_mtu; 1478 destmtu -= ipsechdr; 1479 } 1480 } 1481 1482 #ifdef IPSEC 1483 key_freesp(sp); 1484 #else 1485 KEY_FREESP(&sp); 1486 #endif 1487 } 1488 } 1489 #endif /*defined(IPSEC) || defined(FAST_IPSEC)*/ 1490 IP_STATINC(IP_STAT_CANTFRAG); 1491 break; 1492 1493 case ENOBUFS: 1494 #if 1 1495 /* 1496 * a router should not generate ICMP_SOURCEQUENCH as 1497 * required in RFC1812 Requirements for IP Version 4 Routers. 1498 * source quench could be a big problem under DoS attacks, 1499 * or if the underlying interface is rate-limited. 1500 */ 1501 if (mcopy) 1502 m_freem(mcopy); 1503 return; 1504 #else 1505 type = ICMP_SOURCEQUENCH; 1506 code = 0; 1507 break; 1508 #endif 1509 } 1510 icmp_error(mcopy, type, code, dest, destmtu); 1511 } 1512 1513 void 1514 ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, 1515 struct mbuf *m) 1516 { 1517 1518 if (inp->inp_socket->so_options & SO_TIMESTAMP 1519 #ifdef SO_OTIMESTAMP 1520 || inp->inp_socket->so_options & SO_OTIMESTAMP 1521 #endif 1522 ) { 1523 struct timeval tv; 1524 1525 microtime(&tv); 1526 #ifdef SO_OTIMESTAMP 1527 if (inp->inp_socket->so_options & SO_OTIMESTAMP) { 1528 struct timeval50 tv50; 1529 timeval_to_timeval50(&tv, &tv50); 1530 *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50), 1531 SCM_OTIMESTAMP, SOL_SOCKET); 1532 } else 1533 #endif 1534 *mp = sbcreatecontrol((void *) &tv, sizeof(tv), 1535 SCM_TIMESTAMP, SOL_SOCKET); 1536 if (*mp) 1537 mp = &(*mp)->m_next; 1538 } 1539 if (inp->inp_flags & INP_RECVDSTADDR) { 1540 *mp = sbcreatecontrol((void *) &ip->ip_dst, 1541 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1542 if (*mp) 1543 mp = &(*mp)->m_next; 1544 } 1545 #ifdef notyet 1546 /* 1547 * XXX 1548 * Moving these out of udp_input() made them even more broken 1549 * than they already were. 1550 * - fenner@parc.xerox.com 1551 */ 1552 /* options were tossed already */ 1553 if (inp->inp_flags & INP_RECVOPTS) { 1554 *mp = sbcreatecontrol((void *) opts_deleted_above, 1555 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1556 if (*mp) 1557 mp = &(*mp)->m_next; 1558 } 1559 /* ip_srcroute doesn't do what we want here, need to fix */ 1560 if (inp->inp_flags & INP_RECVRETOPTS) { 1561 *mp = sbcreatecontrol((void *) ip_srcroute(), 1562 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1563 if (*mp) 1564 mp = &(*mp)->m_next; 1565 } 1566 #endif 1567 if (inp->inp_flags & INP_RECVIF) { 1568 struct sockaddr_dl sdl; 1569 1570 sockaddr_dl_init(&sdl, sizeof(sdl), 1571 (m->m_pkthdr.rcvif != NULL) 1572 ? m->m_pkthdr.rcvif->if_index 1573 : 0, 1574 0, NULL, 0, NULL, 0); 1575 *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); 1576 if (*mp) 1577 mp = &(*mp)->m_next; 1578 } 1579 if (inp->inp_flags & INP_RECVTTL) { 1580 *mp = sbcreatecontrol((void *) &ip->ip_ttl, 1581 sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP); 1582 if (*mp) 1583 mp = &(*mp)->m_next; 1584 } 1585 } 1586 1587 /* 1588 * sysctl helper routine for net.inet.ip.forwsrcrt. 1589 */ 1590 static int 1591 sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) 1592 { 1593 int error, tmp; 1594 struct sysctlnode node; 1595 1596 node = *rnode; 1597 tmp = ip_forwsrcrt; 1598 node.sysctl_data = &tmp; 1599 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1600 if (error || newp == NULL) 1601 return (error); 1602 1603 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, 1604 0, NULL, NULL, NULL); 1605 if (error) 1606 return (error); 1607 1608 ip_forwsrcrt = tmp; 1609 1610 return (0); 1611 } 1612 1613 /* 1614 * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the 1615 * range of the new value and tweaks timers if it changes. 1616 */ 1617 static int 1618 sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) 1619 { 1620 int error, tmp; 1621 struct sysctlnode node; 1622 1623 node = *rnode; 1624 tmp = ip_mtudisc_timeout; 1625 node.sysctl_data = &tmp; 1626 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1627 if (error || newp == NULL) 1628 return (error); 1629 if (tmp < 0) 1630 return (EINVAL); 1631 1632 mutex_enter(softnet_lock); 1633 1634 ip_mtudisc_timeout = tmp; 1635 rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); 1636 1637 mutex_exit(softnet_lock); 1638 1639 return (0); 1640 } 1641 1642 #ifdef GATEWAY 1643 /* 1644 * sysctl helper routine for net.inet.ip.maxflows. 1645 */ 1646 static int 1647 sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS) 1648 { 1649 int error; 1650 1651 error = sysctl_lookup(SYSCTLFN_CALL(rnode)); 1652 if (error || newp == NULL) 1653 return (error); 1654 1655 mutex_enter(softnet_lock); 1656 KERNEL_LOCK(1, NULL); 1657 1658 ipflow_prune(); 1659 1660 KERNEL_UNLOCK_ONE(NULL); 1661 mutex_exit(softnet_lock); 1662 1663 return (0); 1664 } 1665 1666 static int 1667 sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS) 1668 { 1669 int error, tmp; 1670 struct sysctlnode node; 1671 1672 node = *rnode; 1673 tmp = ip_hashsize; 1674 node.sysctl_data = &tmp; 1675 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1676 if (error || newp == NULL) 1677 return (error); 1678 1679 if ((tmp & (tmp - 1)) == 0 && tmp != 0) { 1680 /* 1681 * Can only fail due to malloc() 1682 */ 1683 mutex_enter(softnet_lock); 1684 KERNEL_LOCK(1, NULL); 1685 1686 error = ipflow_invalidate_all(tmp); 1687 1688 KERNEL_UNLOCK_ONE(NULL); 1689 mutex_exit(softnet_lock); 1690 1691 } else { 1692 /* 1693 * EINVAL if not a power of 2 1694 */ 1695 error = EINVAL; 1696 } 1697 1698 return error; 1699 } 1700 #endif /* GATEWAY */ 1701 1702 static int 1703 sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) 1704 { 1705 1706 return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); 1707 } 1708 1709 static void 1710 sysctl_net_inet_ip_setup(struct sysctllog **clog) 1711 { 1712 extern int subnetsarelocal, hostzeroisbroadcast; 1713 1714 sysctl_createv(clog, 0, NULL, NULL, 1715 CTLFLAG_PERMANENT, 1716 CTLTYPE_NODE, "net", NULL, 1717 NULL, 0, NULL, 0, 1718 CTL_NET, CTL_EOL); 1719 sysctl_createv(clog, 0, NULL, NULL, 1720 CTLFLAG_PERMANENT, 1721 CTLTYPE_NODE, "inet", 1722 SYSCTL_DESCR("PF_INET related settings"), 1723 NULL, 0, NULL, 0, 1724 CTL_NET, PF_INET, CTL_EOL); 1725 sysctl_createv(clog, 0, NULL, NULL, 1726 CTLFLAG_PERMANENT, 1727 CTLTYPE_NODE, "ip", 1728 SYSCTL_DESCR("IPv4 related settings"), 1729 NULL, 0, NULL, 0, 1730 CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); 1731 1732 sysctl_createv(clog, 0, NULL, NULL, 1733 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1734 CTLTYPE_INT, "forwarding", 1735 SYSCTL_DESCR("Enable forwarding of INET datagrams"), 1736 NULL, 0, &ipforwarding, 0, 1737 CTL_NET, PF_INET, IPPROTO_IP, 1738 IPCTL_FORWARDING, CTL_EOL); 1739 sysctl_createv(clog, 0, NULL, NULL, 1740 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1741 CTLTYPE_INT, "redirect", 1742 SYSCTL_DESCR("Enable sending of ICMP redirect messages"), 1743 NULL, 0, &ipsendredirects, 0, 1744 CTL_NET, PF_INET, IPPROTO_IP, 1745 IPCTL_SENDREDIRECTS, CTL_EOL); 1746 sysctl_createv(clog, 0, NULL, NULL, 1747 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1748 CTLTYPE_INT, "ttl", 1749 SYSCTL_DESCR("Default TTL for an INET datagram"), 1750 NULL, 0, &ip_defttl, 0, 1751 CTL_NET, PF_INET, IPPROTO_IP, 1752 IPCTL_DEFTTL, CTL_EOL); 1753 #ifdef IPCTL_DEFMTU 1754 sysctl_createv(clog, 0, NULL, NULL, 1755 CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, 1756 CTLTYPE_INT, "mtu", 1757 SYSCTL_DESCR("Default MTA for an INET route"), 1758 NULL, 0, &ip_mtu, 0, 1759 CTL_NET, PF_INET, IPPROTO_IP, 1760 IPCTL_DEFMTU, CTL_EOL); 1761 #endif /* IPCTL_DEFMTU */ 1762 sysctl_createv(clog, 0, NULL, NULL, 1763 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1764 CTLTYPE_INT, "forwsrcrt", 1765 SYSCTL_DESCR("Enable forwarding of source-routed " 1766 "datagrams"), 1767 sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, 1768 CTL_NET, PF_INET, IPPROTO_IP, 1769 IPCTL_FORWSRCRT, CTL_EOL); 1770 sysctl_createv(clog, 0, NULL, NULL, 1771 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1772 CTLTYPE_INT, "directed-broadcast", 1773 SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), 1774 NULL, 0, &ip_directedbcast, 0, 1775 CTL_NET, PF_INET, IPPROTO_IP, 1776 IPCTL_DIRECTEDBCAST, CTL_EOL); 1777 sysctl_createv(clog, 0, NULL, NULL, 1778 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1779 CTLTYPE_INT, "allowsrcrt", 1780 SYSCTL_DESCR("Accept source-routed datagrams"), 1781 NULL, 0, &ip_allowsrcrt, 0, 1782 CTL_NET, PF_INET, IPPROTO_IP, 1783 IPCTL_ALLOWSRCRT, CTL_EOL); 1784 sysctl_createv(clog, 0, NULL, NULL, 1785 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1786 CTLTYPE_INT, "subnetsarelocal", 1787 SYSCTL_DESCR("Whether logical subnets are considered " 1788 "local"), 1789 NULL, 0, &subnetsarelocal, 0, 1790 CTL_NET, PF_INET, IPPROTO_IP, 1791 IPCTL_SUBNETSARELOCAL, CTL_EOL); 1792 sysctl_createv(clog, 0, NULL, NULL, 1793 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1794 CTLTYPE_INT, "mtudisc", 1795 SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), 1796 NULL, 0, &ip_mtudisc, 0, 1797 CTL_NET, PF_INET, IPPROTO_IP, 1798 IPCTL_MTUDISC, CTL_EOL); 1799 sysctl_createv(clog, 0, NULL, NULL, 1800 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1801 CTLTYPE_INT, "anonportmin", 1802 SYSCTL_DESCR("Lowest ephemeral port number to assign"), 1803 sysctl_net_inet_ip_ports, 0, &anonportmin, 0, 1804 CTL_NET, PF_INET, IPPROTO_IP, 1805 IPCTL_ANONPORTMIN, CTL_EOL); 1806 sysctl_createv(clog, 0, NULL, NULL, 1807 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1808 CTLTYPE_INT, "anonportmax", 1809 SYSCTL_DESCR("Highest ephemeral port number to assign"), 1810 sysctl_net_inet_ip_ports, 0, &anonportmax, 0, 1811 CTL_NET, PF_INET, IPPROTO_IP, 1812 IPCTL_ANONPORTMAX, CTL_EOL); 1813 sysctl_createv(clog, 0, NULL, NULL, 1814 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1815 CTLTYPE_INT, "mtudisctimeout", 1816 SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), 1817 sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0, 1818 CTL_NET, PF_INET, IPPROTO_IP, 1819 IPCTL_MTUDISCTIMEOUT, CTL_EOL); 1820 #ifdef GATEWAY 1821 sysctl_createv(clog, 0, NULL, NULL, 1822 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1823 CTLTYPE_INT, "maxflows", 1824 SYSCTL_DESCR("Number of flows for fast forwarding"), 1825 sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0, 1826 CTL_NET, PF_INET, IPPROTO_IP, 1827 IPCTL_MAXFLOWS, CTL_EOL); 1828 sysctl_createv(clog, 0, NULL, NULL, 1829 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1830 CTLTYPE_INT, "hashsize", 1831 SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"), 1832 sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0, 1833 CTL_NET, PF_INET, IPPROTO_IP, 1834 CTL_CREATE, CTL_EOL); 1835 #endif /* GATEWAY */ 1836 sysctl_createv(clog, 0, NULL, NULL, 1837 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1838 CTLTYPE_INT, "hostzerobroadcast", 1839 SYSCTL_DESCR("All zeroes address is broadcast address"), 1840 NULL, 0, &hostzeroisbroadcast, 0, 1841 CTL_NET, PF_INET, IPPROTO_IP, 1842 IPCTL_HOSTZEROBROADCAST, CTL_EOL); 1843 #if NGIF > 0 1844 sysctl_createv(clog, 0, NULL, NULL, 1845 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1846 CTLTYPE_INT, "gifttl", 1847 SYSCTL_DESCR("Default TTL for a gif tunnel datagram"), 1848 NULL, 0, &ip_gif_ttl, 0, 1849 CTL_NET, PF_INET, IPPROTO_IP, 1850 IPCTL_GIF_TTL, CTL_EOL); 1851 #endif /* NGIF */ 1852 #ifndef IPNOPRIVPORTS 1853 sysctl_createv(clog, 0, NULL, NULL, 1854 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1855 CTLTYPE_INT, "lowportmin", 1856 SYSCTL_DESCR("Lowest privileged ephemeral port number " 1857 "to assign"), 1858 sysctl_net_inet_ip_ports, 0, &lowportmin, 0, 1859 CTL_NET, PF_INET, IPPROTO_IP, 1860 IPCTL_LOWPORTMIN, CTL_EOL); 1861 sysctl_createv(clog, 0, NULL, NULL, 1862 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1863 CTLTYPE_INT, "lowportmax", 1864 SYSCTL_DESCR("Highest privileged ephemeral port number " 1865 "to assign"), 1866 sysctl_net_inet_ip_ports, 0, &lowportmax, 0, 1867 CTL_NET, PF_INET, IPPROTO_IP, 1868 IPCTL_LOWPORTMAX, CTL_EOL); 1869 #endif /* IPNOPRIVPORTS */ 1870 #if NGRE > 0 1871 sysctl_createv(clog, 0, NULL, NULL, 1872 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1873 CTLTYPE_INT, "grettl", 1874 SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), 1875 NULL, 0, &ip_gre_ttl, 0, 1876 CTL_NET, PF_INET, IPPROTO_IP, 1877 IPCTL_GRE_TTL, CTL_EOL); 1878 #endif /* NGRE */ 1879 sysctl_createv(clog, 0, NULL, NULL, 1880 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1881 CTLTYPE_INT, "checkinterface", 1882 SYSCTL_DESCR("Enable receive side of Strong ES model " 1883 "from RFC1122"), 1884 NULL, 0, &ip_checkinterface, 0, 1885 CTL_NET, PF_INET, IPPROTO_IP, 1886 IPCTL_CHECKINTERFACE, CTL_EOL); 1887 sysctl_createv(clog, 0, NULL, NULL, 1888 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1889 CTLTYPE_INT, "random_id", 1890 SYSCTL_DESCR("Assign random ip_id values"), 1891 NULL, 0, &ip_do_randomid, 0, 1892 CTL_NET, PF_INET, IPPROTO_IP, 1893 IPCTL_RANDOMID, CTL_EOL); 1894 sysctl_createv(clog, 0, NULL, NULL, 1895 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1896 CTLTYPE_INT, "do_loopback_cksum", 1897 SYSCTL_DESCR("Perform IP checksum on loopback"), 1898 NULL, 0, &ip_do_loopback_cksum, 0, 1899 CTL_NET, PF_INET, IPPROTO_IP, 1900 IPCTL_LOOPBACKCKSUM, CTL_EOL); 1901 sysctl_createv(clog, 0, NULL, NULL, 1902 CTLFLAG_PERMANENT, 1903 CTLTYPE_STRUCT, "stats", 1904 SYSCTL_DESCR("IP statistics"), 1905 sysctl_net_inet_ip_stats, 0, NULL, 0, 1906 CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, 1907 CTL_EOL); 1908 } 1909 1910 void 1911 ip_statinc(u_int stat) 1912 { 1913 1914 KASSERT(stat < IP_NSTATS); 1915 IP_STATINC(stat); 1916 } 1917