1 /* $NetBSD: ip_input.c,v 1.175 2003/08/22 22:00:37 itojun 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 * 3. All advertising materials mentioning features or use of this software 49 * must display the following acknowledgement: 50 * This product includes software developed by the NetBSD 51 * Foundation, Inc. and its contributors. 52 * 4. Neither the name of The NetBSD Foundation nor the names of its 53 * contributors may be used to endorse or promote products derived 54 * from this software without specific prior written permission. 55 * 56 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 57 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 58 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 59 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 60 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 61 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 62 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 63 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 64 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 65 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 66 * POSSIBILITY OF SUCH DAMAGE. 67 */ 68 69 /* 70 * Copyright (c) 1982, 1986, 1988, 1993 71 * The Regents of the University of California. All rights reserved. 72 * 73 * Redistribution and use in source and binary forms, with or without 74 * modification, are permitted provided that the following conditions 75 * are met: 76 * 1. Redistributions of source code must retain the above copyright 77 * notice, this list of conditions and the following disclaimer. 78 * 2. Redistributions in binary form must reproduce the above copyright 79 * notice, this list of conditions and the following disclaimer in the 80 * documentation and/or other materials provided with the distribution. 81 * 3. Neither the name of the University nor the names of its contributors 82 * may be used to endorse or promote products derived from this software 83 * without specific prior written permission. 84 * 85 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 86 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 87 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 88 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 89 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 90 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 91 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 92 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 93 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 94 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 95 * SUCH DAMAGE. 96 * 97 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 98 */ 99 100 #include <sys/cdefs.h> 101 __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.175 2003/08/22 22:00:37 itojun Exp $"); 102 103 #include "opt_gateway.h" 104 #include "opt_pfil_hooks.h" 105 #include "opt_ipsec.h" 106 #include "opt_mrouting.h" 107 #include "opt_mbuftrace.h" 108 #include "opt_inet_csum.h" 109 110 #include <sys/param.h> 111 #include <sys/systm.h> 112 #include <sys/malloc.h> 113 #include <sys/mbuf.h> 114 #include <sys/domain.h> 115 #include <sys/protosw.h> 116 #include <sys/socket.h> 117 #include <sys/socketvar.h> 118 #include <sys/errno.h> 119 #include <sys/time.h> 120 #include <sys/kernel.h> 121 #include <sys/pool.h> 122 #include <sys/sysctl.h> 123 124 #include <net/if.h> 125 #include <net/if_dl.h> 126 #include <net/route.h> 127 #include <net/pfil.h> 128 129 #include <netinet/in.h> 130 #include <netinet/in_systm.h> 131 #include <netinet/ip.h> 132 #include <netinet/in_pcb.h> 133 #include <netinet/in_var.h> 134 #include <netinet/ip_var.h> 135 #include <netinet/ip_icmp.h> 136 /* just for gif_ttl */ 137 #include <netinet/in_gif.h> 138 #include "gif.h" 139 #include <net/if_gre.h> 140 #include "gre.h" 141 142 #ifdef MROUTING 143 #include <netinet/ip_mroute.h> 144 #endif 145 146 #ifdef IPSEC 147 #include <netinet6/ipsec.h> 148 #include <netkey/key.h> 149 #endif 150 #ifdef FAST_IPSEC 151 #include <netipsec/ipsec.h> 152 #include <netipsec/key.h> 153 #endif /* FAST_IPSEC*/ 154 155 #ifndef IPFORWARDING 156 #ifdef GATEWAY 157 #define IPFORWARDING 1 /* forward IP packets not for us */ 158 #else /* GATEWAY */ 159 #define IPFORWARDING 0 /* don't forward IP packets not for us */ 160 #endif /* GATEWAY */ 161 #endif /* IPFORWARDING */ 162 #ifndef IPSENDREDIRECTS 163 #define IPSENDREDIRECTS 1 164 #endif 165 #ifndef IPFORWSRCRT 166 #define IPFORWSRCRT 1 /* forward source-routed packets */ 167 #endif 168 #ifndef IPALLOWSRCRT 169 #define IPALLOWSRCRT 1 /* allow source-routed packets */ 170 #endif 171 #ifndef IPMTUDISC 172 #define IPMTUDISC 1 173 #endif 174 #ifndef IPMTUDISCTIMEOUT 175 #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 176 #endif 177 178 /* 179 * Note: DIRECTED_BROADCAST is handled this way so that previous 180 * configuration using this option will Just Work. 181 */ 182 #ifndef IPDIRECTEDBCAST 183 #ifdef DIRECTED_BROADCAST 184 #define IPDIRECTEDBCAST 1 185 #else 186 #define IPDIRECTEDBCAST 0 187 #endif /* DIRECTED_BROADCAST */ 188 #endif /* IPDIRECTEDBCAST */ 189 int ipforwarding = IPFORWARDING; 190 int ipsendredirects = IPSENDREDIRECTS; 191 int ip_defttl = IPDEFTTL; 192 int ip_forwsrcrt = IPFORWSRCRT; 193 int ip_directedbcast = IPDIRECTEDBCAST; 194 int ip_allowsrcrt = IPALLOWSRCRT; 195 int ip_mtudisc = IPMTUDISC; 196 int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 197 #ifdef DIAGNOSTIC 198 int ipprintfs = 0; 199 #endif 200 /* 201 * XXX - Setting ip_checkinterface mostly implements the receive side of 202 * the Strong ES model described in RFC 1122, but since the routing table 203 * and transmit implementation do not implement the Strong ES model, 204 * setting this to 1 results in an odd hybrid. 205 * 206 * XXX - ip_checkinterface currently must be disabled if you use ipnat 207 * to translate the destination address to another local interface. 208 * 209 * XXX - ip_checkinterface must be disabled if you add IP aliases 210 * to the loopback interface instead of the interface where the 211 * packets for those addresses are received. 212 */ 213 int ip_checkinterface = 0; 214 215 216 struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 217 218 extern struct domain inetdomain; 219 int ipqmaxlen = IFQ_MAXLEN; 220 u_long in_ifaddrhash; /* size of hash table - 1 */ 221 int in_ifaddrentries; /* total number of addrs */ 222 struct in_ifaddrhead in_ifaddr; 223 struct in_ifaddrhashhead *in_ifaddrhashtbl; 224 u_long in_multihash; /* size of hash table - 1 */ 225 int in_multientries; /* total number of addrs */ 226 struct in_multihashhead *in_multihashtbl; 227 struct ifqueue ipintrq; 228 struct ipstat ipstat; 229 u_int16_t ip_id; 230 231 #ifdef PFIL_HOOKS 232 struct pfil_head inet_pfil_hook; 233 #endif 234 235 struct ipqhead ipq; 236 int ipq_locked; 237 int ip_nfragpackets = 0; 238 int ip_maxfragpackets = 200; 239 240 static __inline int ipq_lock_try __P((void)); 241 static __inline void ipq_unlock __P((void)); 242 243 static __inline int 244 ipq_lock_try() 245 { 246 int s; 247 248 /* 249 * Use splvm() -- we're blocking things that would cause 250 * mbuf allocation. 251 */ 252 s = splvm(); 253 if (ipq_locked) { 254 splx(s); 255 return (0); 256 } 257 ipq_locked = 1; 258 splx(s); 259 return (1); 260 } 261 262 static __inline void 263 ipq_unlock() 264 { 265 int s; 266 267 s = splvm(); 268 ipq_locked = 0; 269 splx(s); 270 } 271 272 #ifdef DIAGNOSTIC 273 #define IPQ_LOCK() \ 274 do { \ 275 if (ipq_lock_try() == 0) { \ 276 printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \ 277 panic("ipq_lock"); \ 278 } \ 279 } while (/*CONSTCOND*/ 0) 280 #define IPQ_LOCK_CHECK() \ 281 do { \ 282 if (ipq_locked == 0) { \ 283 printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \ 284 panic("ipq lock check"); \ 285 } \ 286 } while (/*CONSTCOND*/ 0) 287 #else 288 #define IPQ_LOCK() (void) ipq_lock_try() 289 #define IPQ_LOCK_CHECK() /* nothing */ 290 #endif 291 292 #define IPQ_UNLOCK() ipq_unlock() 293 294 struct pool inmulti_pool; 295 struct pool ipqent_pool; 296 297 #ifdef INET_CSUM_COUNTERS 298 #include <sys/device.h> 299 300 struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 301 NULL, "inet", "hwcsum bad"); 302 struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 303 NULL, "inet", "hwcsum ok"); 304 struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 305 NULL, "inet", "swcsum"); 306 307 #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 308 309 #else 310 311 #define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 312 313 #endif /* INET_CSUM_COUNTERS */ 314 315 /* 316 * We need to save the IP options in case a protocol wants to respond 317 * to an incoming packet over the same route if the packet got here 318 * using IP source routing. This allows connection establishment and 319 * maintenance when the remote end is on a network that is not known 320 * to us. 321 */ 322 int ip_nhops = 0; 323 static struct ip_srcrt { 324 struct in_addr dst; /* final destination */ 325 char nop; /* one NOP to align */ 326 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 327 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 328 } ip_srcrt; 329 330 static void save_rte __P((u_char *, struct in_addr)); 331 332 #ifdef MBUFTRACE 333 struct mowner ip_rx_mowner = { "internet", "rx" }; 334 struct mowner ip_tx_mowner = { "internet", "tx" }; 335 #endif 336 337 /* 338 * IP initialization: fill in IP protocol switch table. 339 * All protocols not implemented in kernel go to raw IP protocol handler. 340 */ 341 void 342 ip_init() 343 { 344 struct protosw *pr; 345 int i; 346 347 pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", 348 NULL); 349 pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", 350 NULL); 351 352 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 353 if (pr == 0) 354 panic("ip_init"); 355 for (i = 0; i < IPPROTO_MAX; i++) 356 ip_protox[i] = pr - inetsw; 357 for (pr = inetdomain.dom_protosw; 358 pr < inetdomain.dom_protoswNPROTOSW; pr++) 359 if (pr->pr_domain->dom_family == PF_INET && 360 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 361 ip_protox[pr->pr_protocol] = pr - inetsw; 362 LIST_INIT(&ipq); 363 ip_id = time.tv_sec & 0xffff; 364 ipintrq.ifq_maxlen = ipqmaxlen; 365 TAILQ_INIT(&in_ifaddr); 366 in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IFADDR, 367 M_WAITOK, &in_ifaddrhash); 368 in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IPMADDR, 369 M_WAITOK, &in_multihash); 370 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 371 #ifdef GATEWAY 372 ipflow_init(); 373 #endif 374 375 #ifdef PFIL_HOOKS 376 /* Register our Packet Filter hook. */ 377 inet_pfil_hook.ph_type = PFIL_TYPE_AF; 378 inet_pfil_hook.ph_af = AF_INET; 379 i = pfil_head_register(&inet_pfil_hook); 380 if (i != 0) 381 printf("ip_init: WARNING: unable to register pfil hook, " 382 "error %d\n", i); 383 #endif /* PFIL_HOOKS */ 384 385 #ifdef INET_CSUM_COUNTERS 386 evcnt_attach_static(&ip_hwcsum_bad); 387 evcnt_attach_static(&ip_hwcsum_ok); 388 evcnt_attach_static(&ip_swcsum); 389 #endif /* INET_CSUM_COUNTERS */ 390 391 #ifdef MBUFTRACE 392 MOWNER_ATTACH(&ip_tx_mowner); 393 MOWNER_ATTACH(&ip_rx_mowner); 394 #endif /* MBUFTRACE */ 395 } 396 397 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 398 struct route ipforward_rt; 399 400 /* 401 * IP software interrupt routine 402 */ 403 void 404 ipintr() 405 { 406 int s; 407 struct mbuf *m; 408 409 while (1) { 410 s = splnet(); 411 IF_DEQUEUE(&ipintrq, m); 412 splx(s); 413 if (m == 0) 414 return; 415 MCLAIM(m, &ip_rx_mowner); 416 ip_input(m); 417 } 418 } 419 420 /* 421 * Ip input routine. Checksum and byte swap header. If fragmented 422 * try to reassemble. Process options. Pass to next level. 423 */ 424 void 425 ip_input(struct mbuf *m) 426 { 427 struct ip *ip = NULL; 428 struct ipq *fp; 429 struct in_ifaddr *ia; 430 struct ifaddr *ifa; 431 struct ipqent *ipqe; 432 int hlen = 0, mff, len; 433 int downmatch; 434 int checkif; 435 int srcrt = 0; 436 #ifdef FAST_IPSEC 437 struct m_tag *mtag; 438 struct tdb_ident *tdbi; 439 struct secpolicy *sp; 440 int s, error; 441 #endif /* FAST_IPSEC */ 442 443 MCLAIM(m, &ip_rx_mowner); 444 #ifdef DIAGNOSTIC 445 if ((m->m_flags & M_PKTHDR) == 0) 446 panic("ipintr no HDR"); 447 #endif 448 #ifdef IPSEC 449 /* 450 * should the inner packet be considered authentic? 451 * see comment in ah4_input(). 452 */ 453 if (m) { 454 m->m_flags &= ~M_AUTHIPHDR; 455 m->m_flags &= ~M_AUTHIPDGM; 456 } 457 #endif 458 459 /* 460 * If no IP addresses have been set yet but the interfaces 461 * are receiving, can't do anything with incoming packets yet. 462 */ 463 if (TAILQ_FIRST(&in_ifaddr) == 0) 464 goto bad; 465 ipstat.ips_total++; 466 /* 467 * If the IP header is not aligned, slurp it up into a new 468 * mbuf with space for link headers, in the event we forward 469 * it. Otherwise, if it is aligned, make sure the entire 470 * base IP header is in the first mbuf of the chain. 471 */ 472 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 473 if ((m = m_copyup(m, sizeof(struct ip), 474 (max_linkhdr + 3) & ~3)) == NULL) { 475 /* XXXJRT new stat, please */ 476 ipstat.ips_toosmall++; 477 return; 478 } 479 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 480 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 481 ipstat.ips_toosmall++; 482 return; 483 } 484 } 485 ip = mtod(m, struct ip *); 486 if (ip->ip_v != IPVERSION) { 487 ipstat.ips_badvers++; 488 goto bad; 489 } 490 hlen = ip->ip_hl << 2; 491 if (hlen < sizeof(struct ip)) { /* minimum header length */ 492 ipstat.ips_badhlen++; 493 goto bad; 494 } 495 if (hlen > m->m_len) { 496 if ((m = m_pullup(m, hlen)) == 0) { 497 ipstat.ips_badhlen++; 498 return; 499 } 500 ip = mtod(m, struct ip *); 501 } 502 503 /* 504 * RFC1122: packets with a multicast source address are 505 * not allowed. 506 */ 507 if (IN_MULTICAST(ip->ip_src.s_addr)) { 508 ipstat.ips_badaddr++; 509 goto bad; 510 } 511 512 /* 127/8 must not appear on wire - RFC1122 */ 513 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 514 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 515 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 516 ipstat.ips_badaddr++; 517 goto bad; 518 } 519 } 520 521 switch (m->m_pkthdr.csum_flags & 522 ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) | 523 M_CSUM_IPv4_BAD)) { 524 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 525 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 526 goto badcsum; 527 528 case M_CSUM_IPv4: 529 /* Checksum was okay. */ 530 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 531 break; 532 533 default: 534 /* Must compute it ourselves. */ 535 INET_CSUM_COUNTER_INCR(&ip_swcsum); 536 if (in_cksum(m, hlen) != 0) 537 goto bad; 538 break; 539 } 540 541 /* Retrieve the packet length. */ 542 len = ntohs(ip->ip_len); 543 544 /* 545 * Check for additional length bogosity 546 */ 547 if (len < hlen) { 548 ipstat.ips_badlen++; 549 goto bad; 550 } 551 552 /* 553 * Check that the amount of data in the buffers 554 * is as at least much as the IP header would have us expect. 555 * Trim mbufs if longer than we expect. 556 * Drop packet if shorter than we expect. 557 */ 558 if (m->m_pkthdr.len < len) { 559 ipstat.ips_tooshort++; 560 goto bad; 561 } 562 if (m->m_pkthdr.len > len) { 563 if (m->m_len == m->m_pkthdr.len) { 564 m->m_len = len; 565 m->m_pkthdr.len = len; 566 } else 567 m_adj(m, len - m->m_pkthdr.len); 568 } 569 570 #ifdef IPSEC 571 /* ipflow (IP fast forwarding) is not compatible with IPsec. */ 572 m->m_flags &= ~M_CANFASTFWD; 573 #else 574 /* 575 * Assume that we can create a fast-forward IP flow entry 576 * based on this packet. 577 */ 578 m->m_flags |= M_CANFASTFWD; 579 #endif 580 581 #ifdef PFIL_HOOKS 582 /* 583 * Run through list of hooks for input packets. If there are any 584 * filters which require that additional packets in the flow are 585 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 586 * Note that filters must _never_ set this flag, as another filter 587 * in the list may have previously cleared it. 588 */ 589 /* 590 * let ipfilter look at packet on the wire, 591 * not the decapsulated packet. 592 */ 593 #ifdef IPSEC 594 if (!ipsec_getnhist(m)) 595 #else 596 if (1) 597 #endif 598 { 599 struct in_addr odst; 600 601 odst = ip->ip_dst; 602 if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif, 603 PFIL_IN) != 0) 604 return; 605 if (m == NULL) 606 return; 607 ip = mtod(m, struct ip *); 608 hlen = ip->ip_hl << 2; 609 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 610 } 611 #endif /* PFIL_HOOKS */ 612 613 #ifdef ALTQ 614 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 615 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) { 616 /* packet dropped by traffic conditioner */ 617 return; 618 } 619 #endif 620 621 /* 622 * Process options and, if not destined for us, 623 * ship it on. ip_dooptions returns 1 when an 624 * error was detected (causing an icmp message 625 * to be sent and the original packet to be freed). 626 */ 627 ip_nhops = 0; /* for source routed packets */ 628 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 629 return; 630 631 /* 632 * Enable a consistency check between the destination address 633 * and the arrival interface for a unicast packet (the RFC 1122 634 * strong ES model) if IP forwarding is disabled and the packet 635 * is not locally generated. 636 * 637 * XXX - Checking also should be disabled if the destination 638 * address is ipnat'ed to a different interface. 639 * 640 * XXX - Checking is incompatible with IP aliases added 641 * to the loopback interface instead of the interface where 642 * the packets are received. 643 * 644 * XXX - We need to add a per ifaddr flag for this so that 645 * we get finer grain control. 646 */ 647 checkif = ip_checkinterface && (ipforwarding == 0) && 648 (m->m_pkthdr.rcvif != NULL) && 649 ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0); 650 651 /* 652 * Check our list of addresses, to see if the packet is for us. 653 * 654 * Traditional 4.4BSD did not consult IFF_UP at all. 655 * The behavior here is to treat addresses on !IFF_UP interface 656 * as not mine. 657 */ 658 downmatch = 0; 659 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 660 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 661 if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif) 662 continue; 663 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 664 break; 665 else 666 downmatch++; 667 } 668 } 669 if (ia != NULL) 670 goto ours; 671 if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 672 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) { 673 if (ifa->ifa_addr->sa_family != AF_INET) 674 continue; 675 ia = ifatoia(ifa); 676 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 677 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 678 /* 679 * Look for all-0's host part (old broadcast addr), 680 * either for subnet or net. 681 */ 682 ip->ip_dst.s_addr == ia->ia_subnet || 683 ip->ip_dst.s_addr == ia->ia_net) 684 goto ours; 685 /* 686 * An interface with IP address zero accepts 687 * all packets that arrive on that interface. 688 */ 689 if (in_nullhost(ia->ia_addr.sin_addr)) 690 goto ours; 691 } 692 } 693 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 694 struct in_multi *inm; 695 #ifdef MROUTING 696 extern struct socket *ip_mrouter; 697 698 if (M_READONLY(m)) { 699 if ((m = m_pullup(m, hlen)) == 0) { 700 ipstat.ips_toosmall++; 701 return; 702 } 703 ip = mtod(m, struct ip *); 704 } 705 706 if (ip_mrouter) { 707 /* 708 * If we are acting as a multicast router, all 709 * incoming multicast packets are passed to the 710 * kernel-level multicast forwarding function. 711 * The packet is returned (relatively) intact; if 712 * ip_mforward() returns a non-zero value, the packet 713 * must be discarded, else it may be accepted below. 714 * 715 * (The IP ident field is put in the same byte order 716 * as expected when ip_mforward() is called from 717 * ip_output().) 718 */ 719 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 720 ipstat.ips_cantforward++; 721 m_freem(m); 722 return; 723 } 724 725 /* 726 * The process-level routing demon needs to receive 727 * all multicast IGMP packets, whether or not this 728 * host belongs to their destination groups. 729 */ 730 if (ip->ip_p == IPPROTO_IGMP) 731 goto ours; 732 ipstat.ips_forward++; 733 } 734 #endif 735 /* 736 * See if we belong to the destination multicast group on the 737 * arrival interface. 738 */ 739 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 740 if (inm == NULL) { 741 ipstat.ips_cantforward++; 742 m_freem(m); 743 return; 744 } 745 goto ours; 746 } 747 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 748 in_nullhost(ip->ip_dst)) 749 goto ours; 750 751 /* 752 * Not for us; forward if possible and desirable. 753 */ 754 if (ipforwarding == 0) { 755 ipstat.ips_cantforward++; 756 m_freem(m); 757 } else { 758 /* 759 * If ip_dst matched any of my address on !IFF_UP interface, 760 * and there's no IFF_UP interface that matches ip_dst, 761 * send icmp unreach. Forwarding it will result in in-kernel 762 * forwarding loop till TTL goes to 0. 763 */ 764 if (downmatch) { 765 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 766 ipstat.ips_cantforward++; 767 return; 768 } 769 #ifdef IPSEC 770 if (ipsec4_in_reject(m, NULL)) { 771 ipsecstat.in_polvio++; 772 goto bad; 773 } 774 #endif 775 #ifdef FAST_IPSEC 776 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 777 s = splsoftnet(); 778 if (mtag != NULL) { 779 tdbi = (struct tdb_ident *)(mtag + 1); 780 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 781 } else { 782 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 783 IP_FORWARDING, &error); 784 } 785 if (sp == NULL) { /* NB: can happen if error */ 786 splx(s); 787 /*XXX error stat???*/ 788 DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/ 789 goto bad; 790 } 791 792 /* 793 * Check security policy against packet attributes. 794 */ 795 error = ipsec_in_reject(sp, m); 796 KEY_FREESP(&sp); 797 splx(s); 798 if (error) { 799 ipstat.ips_cantforward++; 800 goto bad; 801 } 802 #endif /* FAST_IPSEC */ 803 804 ip_forward(m, srcrt); 805 } 806 return; 807 808 ours: 809 /* 810 * If offset or IP_MF are set, must reassemble. 811 * Otherwise, nothing need be done. 812 * (We could look in the reassembly queue to see 813 * if the packet was previously fragmented, 814 * but it's not worth the time; just let them time out.) 815 */ 816 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 817 if (M_READONLY(m)) { 818 if ((m = m_pullup(m, hlen)) == NULL) { 819 ipstat.ips_toosmall++; 820 goto bad; 821 } 822 ip = mtod(m, struct ip *); 823 } 824 825 /* 826 * Look for queue of fragments 827 * of this datagram. 828 */ 829 IPQ_LOCK(); 830 LIST_FOREACH(fp, &ipq, ipq_q) 831 if (ip->ip_id == fp->ipq_id && 832 in_hosteq(ip->ip_src, fp->ipq_src) && 833 in_hosteq(ip->ip_dst, fp->ipq_dst) && 834 ip->ip_p == fp->ipq_p) 835 goto found; 836 fp = 0; 837 found: 838 839 /* 840 * Adjust ip_len to not reflect header, 841 * set ipqe_mff if more fragments are expected, 842 * convert offset of this to bytes. 843 */ 844 ip->ip_len = htons(ntohs(ip->ip_len) - hlen); 845 mff = (ip->ip_off & htons(IP_MF)) != 0; 846 if (mff) { 847 /* 848 * Make sure that fragments have a data length 849 * that's a non-zero multiple of 8 bytes. 850 */ 851 if (ntohs(ip->ip_len) == 0 || 852 (ntohs(ip->ip_len) & 0x7) != 0) { 853 ipstat.ips_badfrags++; 854 IPQ_UNLOCK(); 855 goto bad; 856 } 857 } 858 ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3); 859 860 /* 861 * If datagram marked as having more fragments 862 * or if this is not the first fragment, 863 * attempt reassembly; if it succeeds, proceed. 864 */ 865 if (mff || ip->ip_off != htons(0)) { 866 ipstat.ips_fragments++; 867 ipqe = pool_get(&ipqent_pool, PR_NOWAIT); 868 if (ipqe == NULL) { 869 ipstat.ips_rcvmemdrop++; 870 IPQ_UNLOCK(); 871 goto bad; 872 } 873 ipqe->ipqe_mff = mff; 874 ipqe->ipqe_m = m; 875 ipqe->ipqe_ip = ip; 876 m = ip_reass(ipqe, fp); 877 if (m == 0) { 878 IPQ_UNLOCK(); 879 return; 880 } 881 ipstat.ips_reassembled++; 882 ip = mtod(m, struct ip *); 883 hlen = ip->ip_hl << 2; 884 ip->ip_len = htons(ntohs(ip->ip_len) + hlen); 885 } else 886 if (fp) 887 ip_freef(fp); 888 IPQ_UNLOCK(); 889 } 890 891 #if defined(IPSEC) 892 /* 893 * enforce IPsec policy checking if we are seeing last header. 894 * note that we do not visit this with protocols with pcb layer 895 * code - like udp/tcp/raw ip. 896 */ 897 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 && 898 ipsec4_in_reject(m, NULL)) { 899 ipsecstat.in_polvio++; 900 goto bad; 901 } 902 #endif 903 #if FAST_IPSEC 904 /* 905 * enforce IPsec policy checking if we are seeing last header. 906 * note that we do not visit this with protocols with pcb layer 907 * code - like udp/tcp/raw ip. 908 */ 909 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 910 /* 911 * Check if the packet has already had IPsec processing 912 * done. If so, then just pass it along. This tag gets 913 * set during AH, ESP, etc. input handling, before the 914 * packet is returned to the ip input queue for delivery. 915 */ 916 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 917 s = splsoftnet(); 918 if (mtag != NULL) { 919 tdbi = (struct tdb_ident *)(mtag + 1); 920 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 921 } else { 922 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 923 IP_FORWARDING, &error); 924 } 925 if (sp != NULL) { 926 /* 927 * Check security policy against packet attributes. 928 */ 929 error = ipsec_in_reject(sp, m); 930 KEY_FREESP(&sp); 931 } else { 932 /* XXX error stat??? */ 933 error = EINVAL; 934 DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/ 935 goto bad; 936 } 937 splx(s); 938 if (error) 939 goto bad; 940 } 941 #endif /* FAST_IPSEC */ 942 943 /* 944 * Switch out to protocol's input routine. 945 */ 946 #if IFA_STATS 947 if (ia && ip) 948 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 949 #endif 950 ipstat.ips_delivered++; 951 { 952 int off = hlen, nh = ip->ip_p; 953 954 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 955 return; 956 } 957 bad: 958 m_freem(m); 959 return; 960 961 badcsum: 962 ipstat.ips_badsum++; 963 m_freem(m); 964 } 965 966 /* 967 * Take incoming datagram fragment and try to 968 * reassemble it into whole datagram. If a chain for 969 * reassembly of this datagram already exists, then it 970 * is given as fp; otherwise have to make a chain. 971 */ 972 struct mbuf * 973 ip_reass(ipqe, fp) 974 struct ipqent *ipqe; 975 struct ipq *fp; 976 { 977 struct mbuf *m = ipqe->ipqe_m; 978 struct ipqent *nq, *p, *q; 979 struct ip *ip; 980 struct mbuf *t; 981 int hlen = ipqe->ipqe_ip->ip_hl << 2; 982 int i, next; 983 984 IPQ_LOCK_CHECK(); 985 986 /* 987 * Presence of header sizes in mbufs 988 * would confuse code below. 989 */ 990 m->m_data += hlen; 991 m->m_len -= hlen; 992 993 /* 994 * If first fragment to arrive, create a reassembly queue. 995 */ 996 if (fp == 0) { 997 /* 998 * Enforce upper bound on number of fragmented packets 999 * for which we attempt reassembly; 1000 * If maxfrag is 0, never accept fragments. 1001 * If maxfrag is -1, accept all fragments without limitation. 1002 */ 1003 if (ip_maxfragpackets < 0) 1004 ; 1005 else if (ip_nfragpackets >= ip_maxfragpackets) 1006 goto dropfrag; 1007 ip_nfragpackets++; 1008 MALLOC(fp, struct ipq *, sizeof (struct ipq), 1009 M_FTABLE, M_NOWAIT); 1010 if (fp == NULL) 1011 goto dropfrag; 1012 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 1013 fp->ipq_ttl = IPFRAGTTL; 1014 fp->ipq_p = ipqe->ipqe_ip->ip_p; 1015 fp->ipq_id = ipqe->ipqe_ip->ip_id; 1016 TAILQ_INIT(&fp->ipq_fragq); 1017 fp->ipq_src = ipqe->ipqe_ip->ip_src; 1018 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 1019 p = NULL; 1020 goto insert; 1021 } 1022 1023 /* 1024 * Find a segment which begins after this one does. 1025 */ 1026 for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; 1027 p = q, q = TAILQ_NEXT(q, ipqe_q)) 1028 if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off)) 1029 break; 1030 1031 /* 1032 * If there is a preceding segment, it may provide some of 1033 * our data already. If so, drop the data from the incoming 1034 * segment. If it provides all of our data, drop us. 1035 */ 1036 if (p != NULL) { 1037 i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) - 1038 ntohs(ipqe->ipqe_ip->ip_off); 1039 if (i > 0) { 1040 if (i >= ntohs(ipqe->ipqe_ip->ip_len)) 1041 goto dropfrag; 1042 m_adj(ipqe->ipqe_m, i); 1043 ipqe->ipqe_ip->ip_off = 1044 htons(ntohs(ipqe->ipqe_ip->ip_off) + i); 1045 ipqe->ipqe_ip->ip_len = 1046 htons(ntohs(ipqe->ipqe_ip->ip_len) - i); 1047 } 1048 } 1049 1050 /* 1051 * While we overlap succeeding segments trim them or, 1052 * if they are completely covered, dequeue them. 1053 */ 1054 for (; q != NULL && 1055 ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) > 1056 ntohs(q->ipqe_ip->ip_off); q = nq) { 1057 i = (ntohs(ipqe->ipqe_ip->ip_off) + 1058 ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off); 1059 if (i < ntohs(q->ipqe_ip->ip_len)) { 1060 q->ipqe_ip->ip_len = 1061 htons(ntohs(q->ipqe_ip->ip_len) - i); 1062 q->ipqe_ip->ip_off = 1063 htons(ntohs(q->ipqe_ip->ip_off) + i); 1064 m_adj(q->ipqe_m, i); 1065 break; 1066 } 1067 nq = TAILQ_NEXT(q, ipqe_q); 1068 m_freem(q->ipqe_m); 1069 TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q); 1070 pool_put(&ipqent_pool, q); 1071 } 1072 1073 insert: 1074 /* 1075 * Stick new segment in its place; 1076 * check for complete reassembly. 1077 */ 1078 if (p == NULL) { 1079 TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 1080 } else { 1081 TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q); 1082 } 1083 next = 0; 1084 for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; 1085 p = q, q = TAILQ_NEXT(q, ipqe_q)) { 1086 if (ntohs(q->ipqe_ip->ip_off) != next) 1087 return (0); 1088 next += ntohs(q->ipqe_ip->ip_len); 1089 } 1090 if (p->ipqe_mff) 1091 return (0); 1092 1093 /* 1094 * Reassembly is complete. Check for a bogus message size and 1095 * concatenate fragments. 1096 */ 1097 q = TAILQ_FIRST(&fp->ipq_fragq); 1098 ip = q->ipqe_ip; 1099 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 1100 ipstat.ips_toolong++; 1101 ip_freef(fp); 1102 return (0); 1103 } 1104 m = q->ipqe_m; 1105 t = m->m_next; 1106 m->m_next = 0; 1107 m_cat(m, t); 1108 nq = TAILQ_NEXT(q, ipqe_q); 1109 pool_put(&ipqent_pool, q); 1110 for (q = nq; q != NULL; q = nq) { 1111 t = q->ipqe_m; 1112 nq = TAILQ_NEXT(q, ipqe_q); 1113 pool_put(&ipqent_pool, q); 1114 m_cat(m, t); 1115 } 1116 1117 /* 1118 * Create header for new ip packet by 1119 * modifying header of first packet; 1120 * dequeue and discard fragment reassembly header. 1121 * Make header visible. 1122 */ 1123 ip->ip_len = htons(next); 1124 ip->ip_src = fp->ipq_src; 1125 ip->ip_dst = fp->ipq_dst; 1126 LIST_REMOVE(fp, ipq_q); 1127 FREE(fp, M_FTABLE); 1128 ip_nfragpackets--; 1129 m->m_len += (ip->ip_hl << 2); 1130 m->m_data -= (ip->ip_hl << 2); 1131 /* some debugging cruft by sklower, below, will go away soon */ 1132 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 1133 int plen = 0; 1134 for (t = m; t; t = t->m_next) 1135 plen += t->m_len; 1136 m->m_pkthdr.len = plen; 1137 } 1138 return (m); 1139 1140 dropfrag: 1141 ipstat.ips_fragdropped++; 1142 m_freem(m); 1143 pool_put(&ipqent_pool, ipqe); 1144 return (0); 1145 } 1146 1147 /* 1148 * Free a fragment reassembly header and all 1149 * associated datagrams. 1150 */ 1151 void 1152 ip_freef(fp) 1153 struct ipq *fp; 1154 { 1155 struct ipqent *q, *p; 1156 1157 IPQ_LOCK_CHECK(); 1158 1159 for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) { 1160 p = TAILQ_NEXT(q, ipqe_q); 1161 m_freem(q->ipqe_m); 1162 TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q); 1163 pool_put(&ipqent_pool, q); 1164 } 1165 LIST_REMOVE(fp, ipq_q); 1166 FREE(fp, M_FTABLE); 1167 ip_nfragpackets--; 1168 } 1169 1170 /* 1171 * IP timer processing; 1172 * if a timer expires on a reassembly 1173 * queue, discard it. 1174 */ 1175 void 1176 ip_slowtimo() 1177 { 1178 struct ipq *fp, *nfp; 1179 int s = splsoftnet(); 1180 1181 IPQ_LOCK(); 1182 for (fp = LIST_FIRST(&ipq); fp != NULL; fp = nfp) { 1183 nfp = LIST_NEXT(fp, ipq_q); 1184 if (--fp->ipq_ttl == 0) { 1185 ipstat.ips_fragtimeout++; 1186 ip_freef(fp); 1187 } 1188 } 1189 /* 1190 * If we are over the maximum number of fragments 1191 * (due to the limit being lowered), drain off 1192 * enough to get down to the new limit. 1193 */ 1194 if (ip_maxfragpackets < 0) 1195 ; 1196 else { 1197 while (ip_nfragpackets > ip_maxfragpackets && LIST_FIRST(&ipq)) 1198 ip_freef(LIST_FIRST(&ipq)); 1199 } 1200 IPQ_UNLOCK(); 1201 #ifdef GATEWAY 1202 ipflow_slowtimo(); 1203 #endif 1204 splx(s); 1205 } 1206 1207 /* 1208 * Drain off all datagram fragments. 1209 */ 1210 void 1211 ip_drain() 1212 { 1213 1214 /* 1215 * We may be called from a device's interrupt context. If 1216 * the ipq is already busy, just bail out now. 1217 */ 1218 if (ipq_lock_try() == 0) 1219 return; 1220 1221 while (LIST_FIRST(&ipq) != NULL) { 1222 ipstat.ips_fragdropped++; 1223 ip_freef(LIST_FIRST(&ipq)); 1224 } 1225 1226 IPQ_UNLOCK(); 1227 } 1228 1229 /* 1230 * Do option processing on a datagram, 1231 * possibly discarding it if bad options are encountered, 1232 * or forwarding it if source-routed. 1233 * Returns 1 if packet has been forwarded/freed, 1234 * 0 if the packet should be processed further. 1235 */ 1236 int 1237 ip_dooptions(m) 1238 struct mbuf *m; 1239 { 1240 struct ip *ip = mtod(m, struct ip *); 1241 u_char *cp, *cp0; 1242 struct ip_timestamp *ipt; 1243 struct in_ifaddr *ia; 1244 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 1245 struct in_addr dst; 1246 n_time ntime; 1247 1248 dst = ip->ip_dst; 1249 cp = (u_char *)(ip + 1); 1250 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 1251 for (; cnt > 0; cnt -= optlen, cp += optlen) { 1252 opt = cp[IPOPT_OPTVAL]; 1253 if (opt == IPOPT_EOL) 1254 break; 1255 if (opt == IPOPT_NOP) 1256 optlen = 1; 1257 else { 1258 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 1259 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1260 goto bad; 1261 } 1262 optlen = cp[IPOPT_OLEN]; 1263 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 1264 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1265 goto bad; 1266 } 1267 } 1268 switch (opt) { 1269 1270 default: 1271 break; 1272 1273 /* 1274 * Source routing with record. 1275 * Find interface with current destination address. 1276 * If none on this machine then drop if strictly routed, 1277 * or do nothing if loosely routed. 1278 * Record interface address and bring up next address 1279 * component. If strictly routed make sure next 1280 * address is on directly accessible net. 1281 */ 1282 case IPOPT_LSRR: 1283 case IPOPT_SSRR: 1284 if (ip_allowsrcrt == 0) { 1285 type = ICMP_UNREACH; 1286 code = ICMP_UNREACH_NET_PROHIB; 1287 goto bad; 1288 } 1289 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1290 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1291 goto bad; 1292 } 1293 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1294 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1295 goto bad; 1296 } 1297 ipaddr.sin_addr = ip->ip_dst; 1298 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1299 if (ia == 0) { 1300 if (opt == IPOPT_SSRR) { 1301 type = ICMP_UNREACH; 1302 code = ICMP_UNREACH_SRCFAIL; 1303 goto bad; 1304 } 1305 /* 1306 * Loose routing, and not at next destination 1307 * yet; nothing to do except forward. 1308 */ 1309 break; 1310 } 1311 off--; /* 0 origin */ 1312 if ((off + sizeof(struct in_addr)) > optlen) { 1313 /* 1314 * End of source route. Should be for us. 1315 */ 1316 save_rte(cp, ip->ip_src); 1317 break; 1318 } 1319 /* 1320 * locate outgoing interface 1321 */ 1322 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 1323 sizeof(ipaddr.sin_addr)); 1324 if (opt == IPOPT_SSRR) 1325 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1326 else 1327 ia = ip_rtaddr(ipaddr.sin_addr); 1328 if (ia == 0) { 1329 type = ICMP_UNREACH; 1330 code = ICMP_UNREACH_SRCFAIL; 1331 goto bad; 1332 } 1333 ip->ip_dst = ipaddr.sin_addr; 1334 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1335 (caddr_t)(cp + off), sizeof(struct in_addr)); 1336 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1337 /* 1338 * Let ip_intr's mcast routing check handle mcast pkts 1339 */ 1340 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1341 break; 1342 1343 case IPOPT_RR: 1344 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1345 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1346 goto bad; 1347 } 1348 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1349 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1350 goto bad; 1351 } 1352 /* 1353 * If no space remains, ignore. 1354 */ 1355 off--; /* 0 origin */ 1356 if ((off + sizeof(struct in_addr)) > optlen) 1357 break; 1358 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 1359 sizeof(ipaddr.sin_addr)); 1360 /* 1361 * locate outgoing interface; if we're the destination, 1362 * use the incoming interface (should be same). 1363 */ 1364 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 1365 == NULL && 1366 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 1367 type = ICMP_UNREACH; 1368 code = ICMP_UNREACH_HOST; 1369 goto bad; 1370 } 1371 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1372 (caddr_t)(cp + off), sizeof(struct in_addr)); 1373 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1374 break; 1375 1376 case IPOPT_TS: 1377 code = cp - (u_char *)ip; 1378 ipt = (struct ip_timestamp *)cp; 1379 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1380 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1381 goto bad; 1382 } 1383 if (ipt->ipt_ptr < 5) { 1384 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1385 goto bad; 1386 } 1387 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1388 if (++ipt->ipt_oflw == 0) { 1389 code = (u_char *)&ipt->ipt_ptr - 1390 (u_char *)ip; 1391 goto bad; 1392 } 1393 break; 1394 } 1395 cp0 = (cp + ipt->ipt_ptr - 1); 1396 switch (ipt->ipt_flg) { 1397 1398 case IPOPT_TS_TSONLY: 1399 break; 1400 1401 case IPOPT_TS_TSANDADDR: 1402 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1403 sizeof(struct in_addr) > ipt->ipt_len) { 1404 code = (u_char *)&ipt->ipt_ptr - 1405 (u_char *)ip; 1406 goto bad; 1407 } 1408 ipaddr.sin_addr = dst; 1409 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1410 m->m_pkthdr.rcvif)); 1411 if (ia == 0) 1412 continue; 1413 bcopy(&ia->ia_addr.sin_addr, 1414 cp0, sizeof(struct in_addr)); 1415 ipt->ipt_ptr += sizeof(struct in_addr); 1416 break; 1417 1418 case IPOPT_TS_PRESPEC: 1419 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1420 sizeof(struct in_addr) > ipt->ipt_len) { 1421 code = (u_char *)&ipt->ipt_ptr - 1422 (u_char *)ip; 1423 goto bad; 1424 } 1425 bcopy(cp0, &ipaddr.sin_addr, 1426 sizeof(struct in_addr)); 1427 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1428 == NULL) 1429 continue; 1430 ipt->ipt_ptr += sizeof(struct in_addr); 1431 break; 1432 1433 default: 1434 /* XXX can't take &ipt->ipt_flg */ 1435 code = (u_char *)&ipt->ipt_ptr - 1436 (u_char *)ip + 1; 1437 goto bad; 1438 } 1439 ntime = iptime(); 1440 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1441 bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1, 1442 sizeof(n_time)); 1443 ipt->ipt_ptr += sizeof(n_time); 1444 } 1445 } 1446 if (forward) { 1447 if (ip_forwsrcrt == 0) { 1448 type = ICMP_UNREACH; 1449 code = ICMP_UNREACH_SRCFAIL; 1450 goto bad; 1451 } 1452 ip_forward(m, 1); 1453 return (1); 1454 } 1455 return (0); 1456 bad: 1457 icmp_error(m, type, code, 0, 0); 1458 ipstat.ips_badoptions++; 1459 return (1); 1460 } 1461 1462 /* 1463 * Given address of next destination (final or next hop), 1464 * return internet address info of interface to be used to get there. 1465 */ 1466 struct in_ifaddr * 1467 ip_rtaddr(dst) 1468 struct in_addr dst; 1469 { 1470 struct sockaddr_in *sin; 1471 1472 sin = satosin(&ipforward_rt.ro_dst); 1473 1474 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 1475 if (ipforward_rt.ro_rt) { 1476 RTFREE(ipforward_rt.ro_rt); 1477 ipforward_rt.ro_rt = 0; 1478 } 1479 sin->sin_family = AF_INET; 1480 sin->sin_len = sizeof(*sin); 1481 sin->sin_addr = dst; 1482 1483 rtalloc(&ipforward_rt); 1484 } 1485 if (ipforward_rt.ro_rt == 0) 1486 return ((struct in_ifaddr *)0); 1487 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 1488 } 1489 1490 /* 1491 * Save incoming source route for use in replies, 1492 * to be picked up later by ip_srcroute if the receiver is interested. 1493 */ 1494 void 1495 save_rte(option, dst) 1496 u_char *option; 1497 struct in_addr dst; 1498 { 1499 unsigned olen; 1500 1501 olen = option[IPOPT_OLEN]; 1502 #ifdef DIAGNOSTIC 1503 if (ipprintfs) 1504 printf("save_rte: olen %d\n", olen); 1505 #endif /* 0 */ 1506 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1507 return; 1508 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 1509 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1510 ip_srcrt.dst = dst; 1511 } 1512 1513 /* 1514 * Retrieve incoming source route for use in replies, 1515 * in the same form used by setsockopt. 1516 * The first hop is placed before the options, will be removed later. 1517 */ 1518 struct mbuf * 1519 ip_srcroute() 1520 { 1521 struct in_addr *p, *q; 1522 struct mbuf *m; 1523 1524 if (ip_nhops == 0) 1525 return ((struct mbuf *)0); 1526 m = m_get(M_DONTWAIT, MT_SOOPTS); 1527 if (m == 0) 1528 return ((struct mbuf *)0); 1529 1530 MCLAIM(m, &inetdomain.dom_mowner); 1531 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1532 1533 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1534 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1535 OPTSIZ; 1536 #ifdef DIAGNOSTIC 1537 if (ipprintfs) 1538 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1539 #endif 1540 1541 /* 1542 * First save first hop for return route 1543 */ 1544 p = &ip_srcrt.route[ip_nhops - 1]; 1545 *(mtod(m, struct in_addr *)) = *p--; 1546 #ifdef DIAGNOSTIC 1547 if (ipprintfs) 1548 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1549 #endif 1550 1551 /* 1552 * Copy option fields and padding (nop) to mbuf. 1553 */ 1554 ip_srcrt.nop = IPOPT_NOP; 1555 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1556 bcopy((caddr_t)&ip_srcrt.nop, 1557 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 1558 q = (struct in_addr *)(mtod(m, caddr_t) + 1559 sizeof(struct in_addr) + OPTSIZ); 1560 #undef OPTSIZ 1561 /* 1562 * Record return path as an IP source route, 1563 * reversing the path (pointers are now aligned). 1564 */ 1565 while (p >= ip_srcrt.route) { 1566 #ifdef DIAGNOSTIC 1567 if (ipprintfs) 1568 printf(" %x", ntohl(q->s_addr)); 1569 #endif 1570 *q++ = *p--; 1571 } 1572 /* 1573 * Last hop goes to final destination. 1574 */ 1575 *q = ip_srcrt.dst; 1576 #ifdef DIAGNOSTIC 1577 if (ipprintfs) 1578 printf(" %x\n", ntohl(q->s_addr)); 1579 #endif 1580 return (m); 1581 } 1582 1583 /* 1584 * Strip out IP options, at higher 1585 * level protocol in the kernel. 1586 * Second argument is buffer to which options 1587 * will be moved, and return value is their length. 1588 * XXX should be deleted; last arg currently ignored. 1589 */ 1590 void 1591 ip_stripoptions(m, mopt) 1592 struct mbuf *m; 1593 struct mbuf *mopt; 1594 { 1595 int i; 1596 struct ip *ip = mtod(m, struct ip *); 1597 caddr_t opts; 1598 int olen; 1599 1600 olen = (ip->ip_hl << 2) - sizeof (struct ip); 1601 opts = (caddr_t)(ip + 1); 1602 i = m->m_len - (sizeof (struct ip) + olen); 1603 bcopy(opts + olen, opts, (unsigned)i); 1604 m->m_len -= olen; 1605 if (m->m_flags & M_PKTHDR) 1606 m->m_pkthdr.len -= olen; 1607 ip->ip_len = htons(ntohs(ip->ip_len) - olen); 1608 ip->ip_hl = sizeof (struct ip) >> 2; 1609 } 1610 1611 const int inetctlerrmap[PRC_NCMDS] = { 1612 0, 0, 0, 0, 1613 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1614 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1615 EMSGSIZE, EHOSTUNREACH, 0, 0, 1616 0, 0, 0, 0, 1617 ENOPROTOOPT 1618 }; 1619 1620 /* 1621 * Forward a packet. If some error occurs return the sender 1622 * an icmp packet. Note we can't always generate a meaningful 1623 * icmp message because icmp doesn't have a large enough repertoire 1624 * of codes and types. 1625 * 1626 * If not forwarding, just drop the packet. This could be confusing 1627 * if ipforwarding was zero but some routing protocol was advancing 1628 * us as a gateway to somewhere. However, we must let the routing 1629 * protocol deal with that. 1630 * 1631 * The srcrt parameter indicates whether the packet is being forwarded 1632 * via a source route. 1633 */ 1634 void 1635 ip_forward(m, srcrt) 1636 struct mbuf *m; 1637 int srcrt; 1638 { 1639 struct ip *ip = mtod(m, struct ip *); 1640 struct sockaddr_in *sin; 1641 struct rtentry *rt; 1642 int error, type = 0, code = 0; 1643 struct mbuf *mcopy; 1644 n_long dest; 1645 struct ifnet *destifp; 1646 #if defined(IPSEC) || defined(FAST_IPSEC) 1647 struct ifnet dummyifp; 1648 #endif 1649 1650 /* 1651 * We are now in the output path. 1652 */ 1653 MCLAIM(m, &ip_tx_mowner); 1654 1655 /* 1656 * Clear any in-bound checksum flags for this packet. 1657 */ 1658 m->m_pkthdr.csum_flags = 0; 1659 1660 dest = 0; 1661 #ifdef DIAGNOSTIC 1662 if (ipprintfs) 1663 printf("forward: src %2.2x dst %2.2x ttl %x\n", 1664 ntohl(ip->ip_src.s_addr), 1665 ntohl(ip->ip_dst.s_addr), ip->ip_ttl); 1666 #endif 1667 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1668 ipstat.ips_cantforward++; 1669 m_freem(m); 1670 return; 1671 } 1672 if (ip->ip_ttl <= IPTTLDEC) { 1673 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1674 return; 1675 } 1676 ip->ip_ttl -= IPTTLDEC; 1677 1678 sin = satosin(&ipforward_rt.ro_dst); 1679 if ((rt = ipforward_rt.ro_rt) == 0 || 1680 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1681 if (ipforward_rt.ro_rt) { 1682 RTFREE(ipforward_rt.ro_rt); 1683 ipforward_rt.ro_rt = 0; 1684 } 1685 sin->sin_family = AF_INET; 1686 sin->sin_len = sizeof(struct sockaddr_in); 1687 sin->sin_addr = ip->ip_dst; 1688 1689 rtalloc(&ipforward_rt); 1690 if (ipforward_rt.ro_rt == 0) { 1691 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1692 return; 1693 } 1694 rt = ipforward_rt.ro_rt; 1695 } 1696 1697 /* 1698 * Save at most 68 bytes of the packet in case 1699 * we need to generate an ICMP message to the src. 1700 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1701 */ 1702 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1703 if (mcopy) 1704 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1705 1706 /* 1707 * If forwarding packet using same interface that it came in on, 1708 * perhaps should send a redirect to sender to shortcut a hop. 1709 * Only send redirect if source is sending directly to us, 1710 * and if packet was not source routed (or has any options). 1711 * Also, don't send redirect if forwarding using a default route 1712 * or a route modified by a redirect. 1713 */ 1714 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1715 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1716 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1717 ipsendredirects && !srcrt) { 1718 if (rt->rt_ifa && 1719 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1720 ifatoia(rt->rt_ifa)->ia_subnet) { 1721 if (rt->rt_flags & RTF_GATEWAY) 1722 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1723 else 1724 dest = ip->ip_dst.s_addr; 1725 /* 1726 * Router requirements says to only send host 1727 * redirects. 1728 */ 1729 type = ICMP_REDIRECT; 1730 code = ICMP_REDIRECT_HOST; 1731 #ifdef DIAGNOSTIC 1732 if (ipprintfs) 1733 printf("redirect (%d) to %x\n", code, 1734 (u_int32_t)dest); 1735 #endif 1736 } 1737 } 1738 1739 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1740 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1741 (struct ip_moptions *)NULL, (struct socket *)NULL); 1742 1743 if (error) 1744 ipstat.ips_cantforward++; 1745 else { 1746 ipstat.ips_forward++; 1747 if (type) 1748 ipstat.ips_redirectsent++; 1749 else { 1750 if (mcopy) { 1751 #ifdef GATEWAY 1752 if (mcopy->m_flags & M_CANFASTFWD) 1753 ipflow_create(&ipforward_rt, mcopy); 1754 #endif 1755 m_freem(mcopy); 1756 } 1757 return; 1758 } 1759 } 1760 if (mcopy == NULL) 1761 return; 1762 destifp = NULL; 1763 1764 switch (error) { 1765 1766 case 0: /* forwarded, but need redirect */ 1767 /* type, code set above */ 1768 break; 1769 1770 case ENETUNREACH: /* shouldn't happen, checked above */ 1771 case EHOSTUNREACH: 1772 case ENETDOWN: 1773 case EHOSTDOWN: 1774 default: 1775 type = ICMP_UNREACH; 1776 code = ICMP_UNREACH_HOST; 1777 break; 1778 1779 case EMSGSIZE: 1780 type = ICMP_UNREACH; 1781 code = ICMP_UNREACH_NEEDFRAG; 1782 #if !defined(IPSEC) && !defined(FAST_IPSEC) 1783 if (ipforward_rt.ro_rt) 1784 destifp = ipforward_rt.ro_rt->rt_ifp; 1785 #else 1786 /* 1787 * If the packet is routed over IPsec tunnel, tell the 1788 * originator the tunnel MTU. 1789 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz 1790 * XXX quickhack!!! 1791 */ 1792 if (ipforward_rt.ro_rt) { 1793 struct secpolicy *sp; 1794 int ipsecerror; 1795 size_t ipsechdr; 1796 struct route *ro; 1797 1798 sp = ipsec4_getpolicybyaddr(mcopy, 1799 IPSEC_DIR_OUTBOUND, IP_FORWARDING, 1800 &ipsecerror); 1801 1802 if (sp == NULL) 1803 destifp = ipforward_rt.ro_rt->rt_ifp; 1804 else { 1805 /* count IPsec header size */ 1806 ipsechdr = ipsec4_hdrsiz(mcopy, 1807 IPSEC_DIR_OUTBOUND, NULL); 1808 1809 /* 1810 * find the correct route for outer IPv4 1811 * header, compute tunnel MTU. 1812 * 1813 * XXX BUG ALERT 1814 * The "dummyifp" code relies upon the fact 1815 * that icmp_error() touches only ifp->if_mtu. 1816 */ 1817 /*XXX*/ 1818 destifp = NULL; 1819 if (sp->req != NULL 1820 && sp->req->sav != NULL 1821 && sp->req->sav->sah != NULL) { 1822 ro = &sp->req->sav->sah->sa_route; 1823 if (ro->ro_rt && ro->ro_rt->rt_ifp) { 1824 dummyifp.if_mtu = 1825 ro->ro_rt->rt_rmx.rmx_mtu ? 1826 ro->ro_rt->rt_rmx.rmx_mtu : 1827 ro->ro_rt->rt_ifp->if_mtu; 1828 dummyifp.if_mtu -= ipsechdr; 1829 destifp = &dummyifp; 1830 } 1831 } 1832 1833 #ifdef IPSEC 1834 key_freesp(sp); 1835 #else 1836 KEY_FREESP(&sp); 1837 #endif 1838 } 1839 } 1840 #endif /*IPSEC*/ 1841 ipstat.ips_cantfrag++; 1842 break; 1843 1844 case ENOBUFS: 1845 #if 1 1846 /* 1847 * a router should not generate ICMP_SOURCEQUENCH as 1848 * required in RFC1812 Requirements for IP Version 4 Routers. 1849 * source quench could be a big problem under DoS attacks, 1850 * or if the underlying interface is rate-limited. 1851 */ 1852 if (mcopy) 1853 m_freem(mcopy); 1854 return; 1855 #else 1856 type = ICMP_SOURCEQUENCH; 1857 code = 0; 1858 break; 1859 #endif 1860 } 1861 icmp_error(mcopy, type, code, dest, destifp); 1862 } 1863 1864 void 1865 ip_savecontrol(inp, mp, ip, m) 1866 struct inpcb *inp; 1867 struct mbuf **mp; 1868 struct ip *ip; 1869 struct mbuf *m; 1870 { 1871 1872 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1873 struct timeval tv; 1874 1875 microtime(&tv); 1876 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1877 SCM_TIMESTAMP, SOL_SOCKET); 1878 if (*mp) 1879 mp = &(*mp)->m_next; 1880 } 1881 if (inp->inp_flags & INP_RECVDSTADDR) { 1882 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1883 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1884 if (*mp) 1885 mp = &(*mp)->m_next; 1886 } 1887 #ifdef notyet 1888 /* 1889 * XXX 1890 * Moving these out of udp_input() made them even more broken 1891 * than they already were. 1892 * - fenner@parc.xerox.com 1893 */ 1894 /* options were tossed already */ 1895 if (inp->inp_flags & INP_RECVOPTS) { 1896 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1897 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1898 if (*mp) 1899 mp = &(*mp)->m_next; 1900 } 1901 /* ip_srcroute doesn't do what we want here, need to fix */ 1902 if (inp->inp_flags & INP_RECVRETOPTS) { 1903 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1904 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1905 if (*mp) 1906 mp = &(*mp)->m_next; 1907 } 1908 #endif 1909 if (inp->inp_flags & INP_RECVIF) { 1910 struct sockaddr_dl sdl; 1911 1912 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1913 sdl.sdl_family = AF_LINK; 1914 sdl.sdl_index = m->m_pkthdr.rcvif ? 1915 m->m_pkthdr.rcvif->if_index : 0; 1916 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1917 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1918 IP_RECVIF, IPPROTO_IP); 1919 if (*mp) 1920 mp = &(*mp)->m_next; 1921 } 1922 } 1923 1924 int 1925 ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1926 int *name; 1927 u_int namelen; 1928 void *oldp; 1929 size_t *oldlenp; 1930 void *newp; 1931 size_t newlen; 1932 { 1933 extern int subnetsarelocal, hostzeroisbroadcast; 1934 1935 int error, old; 1936 1937 /* All sysctl names at this level are terminal. */ 1938 if (namelen != 1) 1939 return (ENOTDIR); 1940 1941 switch (name[0]) { 1942 case IPCTL_FORWARDING: 1943 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1944 case IPCTL_SENDREDIRECTS: 1945 return (sysctl_int(oldp, oldlenp, newp, newlen, 1946 &ipsendredirects)); 1947 case IPCTL_DEFTTL: 1948 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1949 #ifdef notyet 1950 case IPCTL_DEFMTU: 1951 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1952 #endif 1953 case IPCTL_FORWSRCRT: 1954 /* Don't allow this to change in a secure environment. */ 1955 if (securelevel > 0) 1956 return (sysctl_rdint(oldp, oldlenp, newp, 1957 ip_forwsrcrt)); 1958 else 1959 return (sysctl_int(oldp, oldlenp, newp, newlen, 1960 &ip_forwsrcrt)); 1961 case IPCTL_DIRECTEDBCAST: 1962 return (sysctl_int(oldp, oldlenp, newp, newlen, 1963 &ip_directedbcast)); 1964 case IPCTL_ALLOWSRCRT: 1965 return (sysctl_int(oldp, oldlenp, newp, newlen, 1966 &ip_allowsrcrt)); 1967 case IPCTL_SUBNETSARELOCAL: 1968 return (sysctl_int(oldp, oldlenp, newp, newlen, 1969 &subnetsarelocal)); 1970 case IPCTL_MTUDISC: 1971 error = sysctl_int(oldp, oldlenp, newp, newlen, 1972 &ip_mtudisc); 1973 if (error == 0 && ip_mtudisc == 0) 1974 rt_timer_queue_remove_all(ip_mtudisc_timeout_q, TRUE); 1975 return error; 1976 case IPCTL_ANONPORTMIN: 1977 old = anonportmin; 1978 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1979 if (anonportmin >= anonportmax || anonportmin < 0 1980 || anonportmin > 65535 1981 #ifndef IPNOPRIVPORTS 1982 || anonportmin < IPPORT_RESERVED 1983 #endif 1984 ) { 1985 anonportmin = old; 1986 return (EINVAL); 1987 } 1988 return (error); 1989 case IPCTL_ANONPORTMAX: 1990 old = anonportmax; 1991 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1992 if (anonportmin >= anonportmax || anonportmax < 0 1993 || anonportmax > 65535 1994 #ifndef IPNOPRIVPORTS 1995 || anonportmax < IPPORT_RESERVED 1996 #endif 1997 ) { 1998 anonportmax = old; 1999 return (EINVAL); 2000 } 2001 return (error); 2002 case IPCTL_MTUDISCTIMEOUT: 2003 old = ip_mtudisc_timeout; 2004 error = sysctl_int(oldp, oldlenp, newp, newlen, 2005 &ip_mtudisc_timeout); 2006 if (ip_mtudisc_timeout < 0) { 2007 ip_mtudisc_timeout = old; 2008 return (EINVAL); 2009 } 2010 if (error == 0) 2011 rt_timer_queue_change(ip_mtudisc_timeout_q, 2012 ip_mtudisc_timeout); 2013 return (error); 2014 #ifdef GATEWAY 2015 case IPCTL_MAXFLOWS: 2016 { 2017 int s; 2018 2019 error = sysctl_int(oldp, oldlenp, newp, newlen, 2020 &ip_maxflows); 2021 s = splsoftnet(); 2022 ipflow_reap(0); 2023 splx(s); 2024 return (error); 2025 } 2026 #endif 2027 case IPCTL_HOSTZEROBROADCAST: 2028 return (sysctl_int(oldp, oldlenp, newp, newlen, 2029 &hostzeroisbroadcast)); 2030 #if NGIF > 0 2031 case IPCTL_GIF_TTL: 2032 return (sysctl_int(oldp, oldlenp, newp, newlen, 2033 &ip_gif_ttl)); 2034 #endif 2035 2036 #if NGRE > 0 2037 case IPCTL_GRE_TTL: 2038 return (sysctl_int(oldp, oldlenp, newp, newlen, 2039 &ip_gre_ttl)); 2040 #endif 2041 2042 #ifndef IPNOPRIVPORTS 2043 case IPCTL_LOWPORTMIN: 2044 old = lowportmin; 2045 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin); 2046 if (lowportmin >= lowportmax 2047 || lowportmin > IPPORT_RESERVEDMAX 2048 || lowportmin < IPPORT_RESERVEDMIN 2049 ) { 2050 lowportmin = old; 2051 return (EINVAL); 2052 } 2053 return (error); 2054 case IPCTL_LOWPORTMAX: 2055 old = lowportmax; 2056 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax); 2057 if (lowportmin >= lowportmax 2058 || lowportmax > IPPORT_RESERVEDMAX 2059 || lowportmax < IPPORT_RESERVEDMIN 2060 ) { 2061 lowportmax = old; 2062 return (EINVAL); 2063 } 2064 return (error); 2065 #endif 2066 2067 case IPCTL_MAXFRAGPACKETS: 2068 return (sysctl_int(oldp, oldlenp, newp, newlen, 2069 &ip_maxfragpackets)); 2070 2071 case IPCTL_CHECKINTERFACE: 2072 return (sysctl_int(oldp, oldlenp, newp, newlen, 2073 &ip_checkinterface)); 2074 default: 2075 return (EOPNOTSUPP); 2076 } 2077 /* NOTREACHED */ 2078 } 2079