ip_input.c revision 1.186
1 /* $NetBSD: ip_input.c,v 1.186 2003/11/24 20:54:59 scw 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 1.186 2003/11/24 20:54:59 scw Exp $"); 102 103 #include "opt_inet.h" 104 #include "opt_gateway.h" 105 #include "opt_pfil_hooks.h" 106 #include "opt_ipsec.h" 107 #include "opt_mrouting.h" 108 #include "opt_mbuftrace.h" 109 #include "opt_inet_csum.h" 110 111 #include <sys/param.h> 112 #include <sys/systm.h> 113 #include <sys/malloc.h> 114 #include <sys/mbuf.h> 115 #include <sys/domain.h> 116 #include <sys/protosw.h> 117 #include <sys/socket.h> 118 #include <sys/socketvar.h> 119 #include <sys/errno.h> 120 #include <sys/time.h> 121 #include <sys/kernel.h> 122 #include <sys/pool.h> 123 #include <sys/sysctl.h> 124 125 #include <net/if.h> 126 #include <net/if_dl.h> 127 #include <net/route.h> 128 #include <net/pfil.h> 129 130 #include <netinet/in.h> 131 #include <netinet/in_systm.h> 132 #include <netinet/ip.h> 133 #include <netinet/in_pcb.h> 134 #include <netinet/in_var.h> 135 #include <netinet/ip_var.h> 136 #include <netinet/ip_icmp.h> 137 /* just for gif_ttl */ 138 #include <netinet/in_gif.h> 139 #include "gif.h" 140 #include <net/if_gre.h> 141 #include "gre.h" 142 143 #ifdef MROUTING 144 #include <netinet/ip_mroute.h> 145 #endif 146 147 #ifdef IPSEC 148 #include <netinet6/ipsec.h> 149 #include <netkey/key.h> 150 #endif 151 #ifdef FAST_IPSEC 152 #include <netipsec/ipsec.h> 153 #include <netipsec/key.h> 154 #endif /* FAST_IPSEC*/ 155 156 #ifndef IPFORWARDING 157 #ifdef GATEWAY 158 #define IPFORWARDING 1 /* forward IP packets not for us */ 159 #else /* GATEWAY */ 160 #define IPFORWARDING 0 /* don't forward IP packets not for us */ 161 #endif /* GATEWAY */ 162 #endif /* IPFORWARDING */ 163 #ifndef IPSENDREDIRECTS 164 #define IPSENDREDIRECTS 1 165 #endif 166 #ifndef IPFORWSRCRT 167 #define IPFORWSRCRT 1 /* forward source-routed packets */ 168 #endif 169 #ifndef IPALLOWSRCRT 170 #define IPALLOWSRCRT 1 /* allow source-routed packets */ 171 #endif 172 #ifndef IPMTUDISC 173 #define IPMTUDISC 1 174 #endif 175 #ifndef IPMTUDISCTIMEOUT 176 #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 177 #endif 178 179 /* 180 * Note: DIRECTED_BROADCAST is handled this way so that previous 181 * configuration using this option will Just Work. 182 */ 183 #ifndef IPDIRECTEDBCAST 184 #ifdef DIRECTED_BROADCAST 185 #define IPDIRECTEDBCAST 1 186 #else 187 #define IPDIRECTEDBCAST 0 188 #endif /* DIRECTED_BROADCAST */ 189 #endif /* IPDIRECTEDBCAST */ 190 int ipforwarding = IPFORWARDING; 191 int ipsendredirects = IPSENDREDIRECTS; 192 int ip_defttl = IPDEFTTL; 193 int ip_forwsrcrt = IPFORWSRCRT; 194 int ip_directedbcast = IPDIRECTEDBCAST; 195 int ip_allowsrcrt = IPALLOWSRCRT; 196 int ip_mtudisc = IPMTUDISC; 197 int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 198 #ifdef DIAGNOSTIC 199 int ipprintfs = 0; 200 #endif 201 202 #ifdef RANDOM_IP_ID 203 int ip_do_randomid = 0; 204 #endif 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 extern struct domain inetdomain; 225 int ipqmaxlen = IFQ_MAXLEN; 226 u_long in_ifaddrhash; /* size of hash table - 1 */ 227 int in_ifaddrentries; /* total number of addrs */ 228 struct in_ifaddrhead in_ifaddrhead; 229 struct in_ifaddrhashhead *in_ifaddrhashtbl; 230 u_long in_multihash; /* size of hash table - 1 */ 231 int in_multientries; /* total number of addrs */ 232 struct in_multihashhead *in_multihashtbl; 233 struct ifqueue ipintrq; 234 struct ipstat ipstat; 235 uint16_t ip_id; 236 237 #ifdef PFIL_HOOKS 238 struct pfil_head inet_pfil_hook; 239 #endif 240 241 struct ipqhead ipq; 242 int ipq_locked; 243 int ip_nfragpackets = 0; 244 int ip_maxfragpackets = 200; 245 246 static __inline int ipq_lock_try __P((void)); 247 static __inline void ipq_unlock __P((void)); 248 249 static __inline int 250 ipq_lock_try() 251 { 252 int s; 253 254 /* 255 * Use splvm() -- we're blocking things that would cause 256 * mbuf allocation. 257 */ 258 s = splvm(); 259 if (ipq_locked) { 260 splx(s); 261 return (0); 262 } 263 ipq_locked = 1; 264 splx(s); 265 return (1); 266 } 267 268 static __inline void 269 ipq_unlock() 270 { 271 int s; 272 273 s = splvm(); 274 ipq_locked = 0; 275 splx(s); 276 } 277 278 #ifdef DIAGNOSTIC 279 #define IPQ_LOCK() \ 280 do { \ 281 if (ipq_lock_try() == 0) { \ 282 printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \ 283 panic("ipq_lock"); \ 284 } \ 285 } while (/*CONSTCOND*/ 0) 286 #define IPQ_LOCK_CHECK() \ 287 do { \ 288 if (ipq_locked == 0) { \ 289 printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \ 290 panic("ipq lock check"); \ 291 } \ 292 } while (/*CONSTCOND*/ 0) 293 #else 294 #define IPQ_LOCK() (void) ipq_lock_try() 295 #define IPQ_LOCK_CHECK() /* nothing */ 296 #endif 297 298 #define IPQ_UNLOCK() ipq_unlock() 299 300 struct pool inmulti_pool; 301 struct pool ipqent_pool; 302 303 #ifdef INET_CSUM_COUNTERS 304 #include <sys/device.h> 305 306 struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 307 NULL, "inet", "hwcsum bad"); 308 struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 309 NULL, "inet", "hwcsum ok"); 310 struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 311 NULL, "inet", "swcsum"); 312 313 #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 314 315 #else 316 317 #define INET_CSUM_COUNTER_INCR(ev) /* nothing */ 318 319 #endif /* INET_CSUM_COUNTERS */ 320 321 /* 322 * We need to save the IP options in case a protocol wants to respond 323 * to an incoming packet over the same route if the packet got here 324 * using IP source routing. This allows connection establishment and 325 * maintenance when the remote end is on a network that is not known 326 * to us. 327 */ 328 int ip_nhops = 0; 329 static struct ip_srcrt { 330 struct in_addr dst; /* final destination */ 331 char nop; /* one NOP to align */ 332 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 333 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 334 } ip_srcrt; 335 336 static void save_rte __P((u_char *, struct in_addr)); 337 338 #ifdef MBUFTRACE 339 struct mowner ip_rx_mowner = { "internet", "rx" }; 340 struct mowner ip_tx_mowner = { "internet", "tx" }; 341 #endif 342 343 /* 344 * IP initialization: fill in IP protocol switch table. 345 * All protocols not implemented in kernel go to raw IP protocol handler. 346 */ 347 void 348 ip_init() 349 { 350 struct protosw *pr; 351 int i; 352 353 pool_init(&inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", 354 NULL); 355 pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", 356 NULL); 357 358 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 359 if (pr == 0) 360 panic("ip_init"); 361 for (i = 0; i < IPPROTO_MAX; i++) 362 ip_protox[i] = pr - inetsw; 363 for (pr = inetdomain.dom_protosw; 364 pr < inetdomain.dom_protoswNPROTOSW; pr++) 365 if (pr->pr_domain->dom_family == PF_INET && 366 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 367 ip_protox[pr->pr_protocol] = pr - inetsw; 368 LIST_INIT(&ipq); 369 ip_id = time.tv_sec & 0xfffff; 370 ipintrq.ifq_maxlen = ipqmaxlen; 371 TAILQ_INIT(&in_ifaddrhead); 372 in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IFADDR, 373 M_WAITOK, &in_ifaddrhash); 374 in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IPMADDR, 375 M_WAITOK, &in_multihash); 376 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 377 #ifdef GATEWAY 378 ipflow_init(); 379 #endif 380 381 #ifdef PFIL_HOOKS 382 /* Register our Packet Filter hook. */ 383 inet_pfil_hook.ph_type = PFIL_TYPE_AF; 384 inet_pfil_hook.ph_af = AF_INET; 385 i = pfil_head_register(&inet_pfil_hook); 386 if (i != 0) 387 printf("ip_init: WARNING: unable to register pfil hook, " 388 "error %d\n", i); 389 #endif /* PFIL_HOOKS */ 390 391 #ifdef INET_CSUM_COUNTERS 392 evcnt_attach_static(&ip_hwcsum_bad); 393 evcnt_attach_static(&ip_hwcsum_ok); 394 evcnt_attach_static(&ip_swcsum); 395 #endif /* INET_CSUM_COUNTERS */ 396 397 #ifdef MBUFTRACE 398 MOWNER_ATTACH(&ip_tx_mowner); 399 MOWNER_ATTACH(&ip_rx_mowner); 400 #endif /* MBUFTRACE */ 401 } 402 403 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 404 struct route ipforward_rt; 405 406 /* 407 * IP software interrupt routine 408 */ 409 void 410 ipintr() 411 { 412 int s; 413 struct mbuf *m; 414 415 while (1) { 416 s = splnet(); 417 IF_DEQUEUE(&ipintrq, m); 418 splx(s); 419 if (m == 0) 420 return; 421 MCLAIM(m, &ip_rx_mowner); 422 ip_input(m); 423 } 424 } 425 426 /* 427 * Ip input routine. Checksum and byte swap header. If fragmented 428 * try to reassemble. Process options. Pass to next level. 429 */ 430 void 431 ip_input(struct mbuf *m) 432 { 433 struct ip *ip = NULL; 434 struct ipq *fp; 435 struct in_ifaddr *ia; 436 struct ifaddr *ifa; 437 struct ipqent *ipqe; 438 int hlen = 0, mff, len; 439 int downmatch; 440 int checkif; 441 int srcrt = 0; 442 #ifdef FAST_IPSEC 443 struct m_tag *mtag; 444 struct tdb_ident *tdbi; 445 struct secpolicy *sp; 446 int s, error; 447 #endif /* FAST_IPSEC */ 448 449 MCLAIM(m, &ip_rx_mowner); 450 #ifdef DIAGNOSTIC 451 if ((m->m_flags & M_PKTHDR) == 0) 452 panic("ipintr no HDR"); 453 #endif 454 455 /* 456 * If no IP addresses have been set yet but the interfaces 457 * are receiving, can't do anything with incoming packets yet. 458 */ 459 if (TAILQ_FIRST(&in_ifaddrhead) == 0) 460 goto bad; 461 ipstat.ips_total++; 462 /* 463 * If the IP header is not aligned, slurp it up into a new 464 * mbuf with space for link headers, in the event we forward 465 * it. Otherwise, if it is aligned, make sure the entire 466 * base IP header is in the first mbuf of the chain. 467 */ 468 if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { 469 if ((m = m_copyup(m, sizeof(struct ip), 470 (max_linkhdr + 3) & ~3)) == NULL) { 471 /* XXXJRT new stat, please */ 472 ipstat.ips_toosmall++; 473 return; 474 } 475 } else if (__predict_false(m->m_len < sizeof (struct ip))) { 476 if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { 477 ipstat.ips_toosmall++; 478 return; 479 } 480 } 481 ip = mtod(m, struct ip *); 482 if (ip->ip_v != IPVERSION) { 483 ipstat.ips_badvers++; 484 goto bad; 485 } 486 hlen = ip->ip_hl << 2; 487 if (hlen < sizeof(struct ip)) { /* minimum header length */ 488 ipstat.ips_badhlen++; 489 goto bad; 490 } 491 if (hlen > m->m_len) { 492 if ((m = m_pullup(m, hlen)) == 0) { 493 ipstat.ips_badhlen++; 494 return; 495 } 496 ip = mtod(m, struct ip *); 497 } 498 499 /* 500 * RFC1122: packets with a multicast source address are 501 * not allowed. 502 */ 503 if (IN_MULTICAST(ip->ip_src.s_addr)) { 504 ipstat.ips_badaddr++; 505 goto bad; 506 } 507 508 /* 127/8 must not appear on wire - RFC1122 */ 509 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 510 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 511 if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { 512 ipstat.ips_badaddr++; 513 goto bad; 514 } 515 } 516 517 switch (m->m_pkthdr.csum_flags & 518 ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) | 519 M_CSUM_IPv4_BAD)) { 520 case M_CSUM_IPv4|M_CSUM_IPv4_BAD: 521 INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); 522 goto badcsum; 523 524 case M_CSUM_IPv4: 525 /* Checksum was okay. */ 526 INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); 527 break; 528 529 default: 530 /* Must compute it ourselves. */ 531 INET_CSUM_COUNTER_INCR(&ip_swcsum); 532 if (in_cksum(m, hlen) != 0) 533 goto bad; 534 break; 535 } 536 537 /* Retrieve the packet length. */ 538 len = ntohs(ip->ip_len); 539 540 /* 541 * Check for additional length bogosity 542 */ 543 if (len < hlen) { 544 ipstat.ips_badlen++; 545 goto bad; 546 } 547 548 /* 549 * Check that the amount of data in the buffers 550 * is as at least much as the IP header would have us expect. 551 * Trim mbufs if longer than we expect. 552 * Drop packet if shorter than we expect. 553 */ 554 if (m->m_pkthdr.len < len) { 555 ipstat.ips_tooshort++; 556 goto bad; 557 } 558 if (m->m_pkthdr.len > len) { 559 if (m->m_len == m->m_pkthdr.len) { 560 m->m_len = len; 561 m->m_pkthdr.len = len; 562 } else 563 m_adj(m, len - m->m_pkthdr.len); 564 } 565 566 #ifdef IPSEC 567 /* ipflow (IP fast forwarding) is not compatible with IPsec. */ 568 m->m_flags &= ~M_CANFASTFWD; 569 #else 570 /* 571 * Assume that we can create a fast-forward IP flow entry 572 * based on this packet. 573 */ 574 m->m_flags |= M_CANFASTFWD; 575 #endif 576 577 #ifdef PFIL_HOOKS 578 /* 579 * Run through list of hooks for input packets. If there are any 580 * filters which require that additional packets in the flow are 581 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 582 * Note that filters must _never_ set this flag, as another filter 583 * in the list may have previously cleared it. 584 */ 585 /* 586 * let ipfilter look at packet on the wire, 587 * not the decapsulated packet. 588 */ 589 #ifdef IPSEC 590 if (!ipsec_getnhist(m)) 591 #elif defined(FAST_IPSEC) 592 if (!ipsec_indone(m)) 593 #else 594 if (1) 595 #endif 596 { 597 struct in_addr odst; 598 599 odst = ip->ip_dst; 600 if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif, 601 PFIL_IN) != 0) 602 return; 603 if (m == NULL) 604 return; 605 ip = mtod(m, struct ip *); 606 hlen = ip->ip_hl << 2; 607 srcrt = (odst.s_addr != ip->ip_dst.s_addr); 608 } 609 #endif /* PFIL_HOOKS */ 610 611 #ifdef ALTQ 612 /* XXX Temporary until ALTQ is changed to use a pfil hook */ 613 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) { 614 /* packet dropped by traffic conditioner */ 615 return; 616 } 617 #endif 618 619 /* 620 * Process options and, if not destined for us, 621 * ship it on. ip_dooptions returns 1 when an 622 * error was detected (causing an icmp message 623 * to be sent and the original packet to be freed). 624 */ 625 ip_nhops = 0; /* for source routed packets */ 626 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 627 return; 628 629 /* 630 * Enable a consistency check between the destination address 631 * and the arrival interface for a unicast packet (the RFC 1122 632 * strong ES model) if IP forwarding is disabled and the packet 633 * is not locally generated. 634 * 635 * XXX - Checking also should be disabled if the destination 636 * address is ipnat'ed to a different interface. 637 * 638 * XXX - Checking is incompatible with IP aliases added 639 * to the loopback interface instead of the interface where 640 * the packets are received. 641 * 642 * XXX - We need to add a per ifaddr flag for this so that 643 * we get finer grain control. 644 */ 645 checkif = ip_checkinterface && (ipforwarding == 0) && 646 (m->m_pkthdr.rcvif != NULL) && 647 ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0); 648 649 /* 650 * Check our list of addresses, to see if the packet is for us. 651 * 652 * Traditional 4.4BSD did not consult IFF_UP at all. 653 * The behavior here is to treat addresses on !IFF_UP interface 654 * as not mine. 655 */ 656 downmatch = 0; 657 LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { 658 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 659 if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif) 660 continue; 661 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 662 break; 663 else 664 downmatch++; 665 } 666 } 667 if (ia != NULL) 668 goto ours; 669 if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 670 TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) { 671 if (ifa->ifa_addr->sa_family != AF_INET) 672 continue; 673 ia = ifatoia(ifa); 674 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 675 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 676 /* 677 * Look for all-0's host part (old broadcast addr), 678 * either for subnet or net. 679 */ 680 ip->ip_dst.s_addr == ia->ia_subnet || 681 ip->ip_dst.s_addr == ia->ia_net) 682 goto ours; 683 /* 684 * An interface with IP address zero accepts 685 * all packets that arrive on that interface. 686 */ 687 if (in_nullhost(ia->ia_addr.sin_addr)) 688 goto ours; 689 } 690 } 691 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 692 struct in_multi *inm; 693 #ifdef MROUTING 694 extern struct socket *ip_mrouter; 695 696 if (M_READONLY(m)) { 697 if ((m = m_pullup(m, hlen)) == 0) { 698 ipstat.ips_toosmall++; 699 return; 700 } 701 ip = mtod(m, struct ip *); 702 } 703 704 if (ip_mrouter) { 705 /* 706 * If we are acting as a multicast router, all 707 * incoming multicast packets are passed to the 708 * kernel-level multicast forwarding function. 709 * The packet is returned (relatively) intact; if 710 * ip_mforward() returns a non-zero value, the packet 711 * must be discarded, else it may be accepted below. 712 * 713 * (The IP ident field is put in the same byte order 714 * as expected when ip_mforward() is called from 715 * ip_output().) 716 */ 717 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 718 ipstat.ips_cantforward++; 719 m_freem(m); 720 return; 721 } 722 723 /* 724 * The process-level routing demon needs to receive 725 * all multicast IGMP packets, whether or not this 726 * host belongs to their destination groups. 727 */ 728 if (ip->ip_p == IPPROTO_IGMP) 729 goto ours; 730 ipstat.ips_forward++; 731 } 732 #endif 733 /* 734 * See if we belong to the destination multicast group on the 735 * arrival interface. 736 */ 737 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 738 if (inm == NULL) { 739 ipstat.ips_cantforward++; 740 m_freem(m); 741 return; 742 } 743 goto ours; 744 } 745 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 746 in_nullhost(ip->ip_dst)) 747 goto ours; 748 749 /* 750 * Not for us; forward if possible and desirable. 751 */ 752 if (ipforwarding == 0) { 753 ipstat.ips_cantforward++; 754 m_freem(m); 755 } else { 756 /* 757 * If ip_dst matched any of my address on !IFF_UP interface, 758 * and there's no IFF_UP interface that matches ip_dst, 759 * send icmp unreach. Forwarding it will result in in-kernel 760 * forwarding loop till TTL goes to 0. 761 */ 762 if (downmatch) { 763 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 764 ipstat.ips_cantforward++; 765 return; 766 } 767 #ifdef IPSEC 768 if (ipsec4_in_reject(m, NULL)) { 769 ipsecstat.in_polvio++; 770 goto bad; 771 } 772 #endif 773 #ifdef FAST_IPSEC 774 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 775 s = splsoftnet(); 776 if (mtag != NULL) { 777 tdbi = (struct tdb_ident *)(mtag + 1); 778 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 779 } else { 780 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 781 IP_FORWARDING, &error); 782 } 783 if (sp == NULL) { /* NB: can happen if error */ 784 splx(s); 785 /*XXX error stat???*/ 786 DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/ 787 goto bad; 788 } 789 790 /* 791 * Check security policy against packet attributes. 792 */ 793 error = ipsec_in_reject(sp, m); 794 KEY_FREESP(&sp); 795 splx(s); 796 if (error) { 797 ipstat.ips_cantforward++; 798 goto bad; 799 } 800 #endif /* FAST_IPSEC */ 801 802 ip_forward(m, srcrt); 803 } 804 return; 805 806 ours: 807 /* 808 * If offset or IP_MF are set, must reassemble. 809 * Otherwise, nothing need be done. 810 * (We could look in the reassembly queue to see 811 * if the packet was previously fragmented, 812 * but it's not worth the time; just let them time out.) 813 */ 814 if (ip->ip_off & ~htons(IP_DF|IP_RF)) { 815 if (M_READONLY(m)) { 816 if ((m = m_pullup(m, hlen)) == NULL) { 817 ipstat.ips_toosmall++; 818 goto bad; 819 } 820 ip = mtod(m, struct ip *); 821 } 822 823 /* 824 * Look for queue of fragments 825 * of this datagram. 826 */ 827 IPQ_LOCK(); 828 LIST_FOREACH(fp, &ipq, ipq_q) 829 if (ip->ip_id == fp->ipq_id && 830 in_hosteq(ip->ip_src, fp->ipq_src) && 831 in_hosteq(ip->ip_dst, fp->ipq_dst) && 832 ip->ip_p == fp->ipq_p) 833 goto found; 834 fp = 0; 835 found: 836 837 /* 838 * Adjust ip_len to not reflect header, 839 * set ipqe_mff if more fragments are expected, 840 * convert offset of this to bytes. 841 */ 842 ip->ip_len = htons(ntohs(ip->ip_len) - hlen); 843 mff = (ip->ip_off & htons(IP_MF)) != 0; 844 if (mff) { 845 /* 846 * Make sure that fragments have a data length 847 * that's a non-zero multiple of 8 bytes. 848 */ 849 if (ntohs(ip->ip_len) == 0 || 850 (ntohs(ip->ip_len) & 0x7) != 0) { 851 ipstat.ips_badfrags++; 852 IPQ_UNLOCK(); 853 goto bad; 854 } 855 } 856 ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3); 857 858 /* 859 * If datagram marked as having more fragments 860 * or if this is not the first fragment, 861 * attempt reassembly; if it succeeds, proceed. 862 */ 863 if (mff || ip->ip_off != htons(0)) { 864 ipstat.ips_fragments++; 865 ipqe = pool_get(&ipqent_pool, PR_NOWAIT); 866 if (ipqe == NULL) { 867 ipstat.ips_rcvmemdrop++; 868 IPQ_UNLOCK(); 869 goto bad; 870 } 871 ipqe->ipqe_mff = mff; 872 ipqe->ipqe_m = m; 873 ipqe->ipqe_ip = ip; 874 m = ip_reass(ipqe, fp); 875 if (m == 0) { 876 IPQ_UNLOCK(); 877 return; 878 } 879 ipstat.ips_reassembled++; 880 ip = mtod(m, struct ip *); 881 hlen = ip->ip_hl << 2; 882 ip->ip_len = htons(ntohs(ip->ip_len) + hlen); 883 } else 884 if (fp) 885 ip_freef(fp); 886 IPQ_UNLOCK(); 887 } 888 889 #if defined(IPSEC) 890 /* 891 * enforce IPsec policy checking if we are seeing last header. 892 * note that we do not visit this with protocols with pcb layer 893 * code - like udp/tcp/raw ip. 894 */ 895 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 && 896 ipsec4_in_reject(m, NULL)) { 897 ipsecstat.in_polvio++; 898 goto bad; 899 } 900 #endif 901 #if FAST_IPSEC 902 /* 903 * enforce IPsec policy checking if we are seeing last header. 904 * note that we do not visit this with protocols with pcb layer 905 * code - like udp/tcp/raw ip. 906 */ 907 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { 908 /* 909 * Check if the packet has already had IPsec processing 910 * done. If so, then just pass it along. This tag gets 911 * set during AH, ESP, etc. input handling, before the 912 * packet is returned to the ip input queue for delivery. 913 */ 914 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); 915 s = splsoftnet(); 916 if (mtag != NULL) { 917 tdbi = (struct tdb_ident *)(mtag + 1); 918 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); 919 } else { 920 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, 921 IP_FORWARDING, &error); 922 } 923 if (sp != NULL) { 924 /* 925 * Check security policy against packet attributes. 926 */ 927 error = ipsec_in_reject(sp, m); 928 KEY_FREESP(&sp); 929 } else { 930 /* XXX error stat??? */ 931 error = EINVAL; 932 DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/ 933 goto bad; 934 } 935 splx(s); 936 if (error) 937 goto bad; 938 } 939 #endif /* FAST_IPSEC */ 940 941 /* 942 * Switch out to protocol's input routine. 943 */ 944 #if IFA_STATS 945 if (ia && ip) 946 ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); 947 #endif 948 ipstat.ips_delivered++; 949 { 950 int off = hlen, nh = ip->ip_p; 951 952 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 953 return; 954 } 955 bad: 956 m_freem(m); 957 return; 958 959 badcsum: 960 ipstat.ips_badsum++; 961 m_freem(m); 962 } 963 964 /* 965 * Take incoming datagram fragment and try to 966 * reassemble it into whole datagram. If a chain for 967 * reassembly of this datagram already exists, then it 968 * is given as fp; otherwise have to make a chain. 969 */ 970 struct mbuf * 971 ip_reass(ipqe, fp) 972 struct ipqent *ipqe; 973 struct ipq *fp; 974 { 975 struct mbuf *m = ipqe->ipqe_m; 976 struct ipqent *nq, *p, *q; 977 struct ip *ip; 978 struct mbuf *t; 979 int hlen = ipqe->ipqe_ip->ip_hl << 2; 980 int i, next; 981 982 IPQ_LOCK_CHECK(); 983 984 /* 985 * Presence of header sizes in mbufs 986 * would confuse code below. 987 */ 988 m->m_data += hlen; 989 m->m_len -= hlen; 990 991 /* 992 * If first fragment to arrive, create a reassembly queue. 993 */ 994 if (fp == 0) { 995 /* 996 * Enforce upper bound on number of fragmented packets 997 * for which we attempt reassembly; 998 * If maxfrag is 0, never accept fragments. 999 * If maxfrag is -1, accept all fragments without limitation. 1000 */ 1001 if (ip_maxfragpackets < 0) 1002 ; 1003 else if (ip_nfragpackets >= ip_maxfragpackets) 1004 goto dropfrag; 1005 ip_nfragpackets++; 1006 MALLOC(fp, struct ipq *, sizeof (struct ipq), 1007 M_FTABLE, M_NOWAIT); 1008 if (fp == NULL) 1009 goto dropfrag; 1010 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 1011 fp->ipq_ttl = IPFRAGTTL; 1012 fp->ipq_p = ipqe->ipqe_ip->ip_p; 1013 fp->ipq_id = ipqe->ipqe_ip->ip_id; 1014 TAILQ_INIT(&fp->ipq_fragq); 1015 fp->ipq_src = ipqe->ipqe_ip->ip_src; 1016 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 1017 p = NULL; 1018 goto insert; 1019 } 1020 1021 /* 1022 * Find a segment which begins after this one does. 1023 */ 1024 for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; 1025 p = q, q = TAILQ_NEXT(q, ipqe_q)) 1026 if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off)) 1027 break; 1028 1029 /* 1030 * If there is a preceding segment, it may provide some of 1031 * our data already. If so, drop the data from the incoming 1032 * segment. If it provides all of our data, drop us. 1033 */ 1034 if (p != NULL) { 1035 i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) - 1036 ntohs(ipqe->ipqe_ip->ip_off); 1037 if (i > 0) { 1038 if (i >= ntohs(ipqe->ipqe_ip->ip_len)) 1039 goto dropfrag; 1040 m_adj(ipqe->ipqe_m, i); 1041 ipqe->ipqe_ip->ip_off = 1042 htons(ntohs(ipqe->ipqe_ip->ip_off) + i); 1043 ipqe->ipqe_ip->ip_len = 1044 htons(ntohs(ipqe->ipqe_ip->ip_len) - i); 1045 } 1046 } 1047 1048 /* 1049 * While we overlap succeeding segments trim them or, 1050 * if they are completely covered, dequeue them. 1051 */ 1052 for (; q != NULL && 1053 ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) > 1054 ntohs(q->ipqe_ip->ip_off); q = nq) { 1055 i = (ntohs(ipqe->ipqe_ip->ip_off) + 1056 ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off); 1057 if (i < ntohs(q->ipqe_ip->ip_len)) { 1058 q->ipqe_ip->ip_len = 1059 htons(ntohs(q->ipqe_ip->ip_len) - i); 1060 q->ipqe_ip->ip_off = 1061 htons(ntohs(q->ipqe_ip->ip_off) + i); 1062 m_adj(q->ipqe_m, i); 1063 break; 1064 } 1065 nq = TAILQ_NEXT(q, ipqe_q); 1066 m_freem(q->ipqe_m); 1067 TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q); 1068 pool_put(&ipqent_pool, q); 1069 } 1070 1071 insert: 1072 /* 1073 * Stick new segment in its place; 1074 * check for complete reassembly. 1075 */ 1076 if (p == NULL) { 1077 TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 1078 } else { 1079 TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q); 1080 } 1081 next = 0; 1082 for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; 1083 p = q, q = TAILQ_NEXT(q, ipqe_q)) { 1084 if (ntohs(q->ipqe_ip->ip_off) != next) 1085 return (0); 1086 next += ntohs(q->ipqe_ip->ip_len); 1087 } 1088 if (p->ipqe_mff) 1089 return (0); 1090 1091 /* 1092 * Reassembly is complete. Check for a bogus message size and 1093 * concatenate fragments. 1094 */ 1095 q = TAILQ_FIRST(&fp->ipq_fragq); 1096 ip = q->ipqe_ip; 1097 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 1098 ipstat.ips_toolong++; 1099 ip_freef(fp); 1100 return (0); 1101 } 1102 m = q->ipqe_m; 1103 t = m->m_next; 1104 m->m_next = 0; 1105 m_cat(m, t); 1106 nq = TAILQ_NEXT(q, ipqe_q); 1107 pool_put(&ipqent_pool, q); 1108 for (q = nq; q != NULL; q = nq) { 1109 t = q->ipqe_m; 1110 nq = TAILQ_NEXT(q, ipqe_q); 1111 pool_put(&ipqent_pool, q); 1112 m_cat(m, t); 1113 } 1114 1115 /* 1116 * Create header for new ip packet by 1117 * modifying header of first packet; 1118 * dequeue and discard fragment reassembly header. 1119 * Make header visible. 1120 */ 1121 ip->ip_len = htons(next); 1122 ip->ip_src = fp->ipq_src; 1123 ip->ip_dst = fp->ipq_dst; 1124 LIST_REMOVE(fp, ipq_q); 1125 FREE(fp, M_FTABLE); 1126 ip_nfragpackets--; 1127 m->m_len += (ip->ip_hl << 2); 1128 m->m_data -= (ip->ip_hl << 2); 1129 /* some debugging cruft by sklower, below, will go away soon */ 1130 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 1131 int plen = 0; 1132 for (t = m; t; t = t->m_next) 1133 plen += t->m_len; 1134 m->m_pkthdr.len = plen; 1135 } 1136 return (m); 1137 1138 dropfrag: 1139 ipstat.ips_fragdropped++; 1140 m_freem(m); 1141 pool_put(&ipqent_pool, ipqe); 1142 return (0); 1143 } 1144 1145 /* 1146 * Free a fragment reassembly header and all 1147 * associated datagrams. 1148 */ 1149 void 1150 ip_freef(fp) 1151 struct ipq *fp; 1152 { 1153 struct ipqent *q, *p; 1154 1155 IPQ_LOCK_CHECK(); 1156 1157 for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) { 1158 p = TAILQ_NEXT(q, ipqe_q); 1159 m_freem(q->ipqe_m); 1160 TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q); 1161 pool_put(&ipqent_pool, q); 1162 } 1163 LIST_REMOVE(fp, ipq_q); 1164 FREE(fp, M_FTABLE); 1165 ip_nfragpackets--; 1166 } 1167 1168 /* 1169 * IP timer processing; 1170 * if a timer expires on a reassembly 1171 * queue, discard it. 1172 */ 1173 void 1174 ip_slowtimo() 1175 { 1176 struct ipq *fp, *nfp; 1177 int s = splsoftnet(); 1178 1179 IPQ_LOCK(); 1180 for (fp = LIST_FIRST(&ipq); fp != NULL; fp = nfp) { 1181 nfp = LIST_NEXT(fp, ipq_q); 1182 if (--fp->ipq_ttl == 0) { 1183 ipstat.ips_fragtimeout++; 1184 ip_freef(fp); 1185 } 1186 } 1187 /* 1188 * If we are over the maximum number of fragments 1189 * (due to the limit being lowered), drain off 1190 * enough to get down to the new limit. 1191 */ 1192 if (ip_maxfragpackets < 0) 1193 ; 1194 else { 1195 while (ip_nfragpackets > ip_maxfragpackets && LIST_FIRST(&ipq)) 1196 ip_freef(LIST_FIRST(&ipq)); 1197 } 1198 IPQ_UNLOCK(); 1199 #ifdef GATEWAY 1200 ipflow_slowtimo(); 1201 #endif 1202 splx(s); 1203 } 1204 1205 /* 1206 * Drain off all datagram fragments. 1207 */ 1208 void 1209 ip_drain() 1210 { 1211 1212 /* 1213 * We may be called from a device's interrupt context. If 1214 * the ipq is already busy, just bail out now. 1215 */ 1216 if (ipq_lock_try() == 0) 1217 return; 1218 1219 while (LIST_FIRST(&ipq) != NULL) { 1220 ipstat.ips_fragdropped++; 1221 ip_freef(LIST_FIRST(&ipq)); 1222 } 1223 1224 IPQ_UNLOCK(); 1225 } 1226 1227 /* 1228 * Do option processing on a datagram, 1229 * possibly discarding it if bad options are encountered, 1230 * or forwarding it if source-routed. 1231 * Returns 1 if packet has been forwarded/freed, 1232 * 0 if the packet should be processed further. 1233 */ 1234 int 1235 ip_dooptions(m) 1236 struct mbuf *m; 1237 { 1238 struct ip *ip = mtod(m, struct ip *); 1239 u_char *cp, *cp0; 1240 struct ip_timestamp *ipt; 1241 struct in_ifaddr *ia; 1242 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 1243 struct in_addr dst; 1244 n_time ntime; 1245 1246 dst = ip->ip_dst; 1247 cp = (u_char *)(ip + 1); 1248 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 1249 for (; cnt > 0; cnt -= optlen, cp += optlen) { 1250 opt = cp[IPOPT_OPTVAL]; 1251 if (opt == IPOPT_EOL) 1252 break; 1253 if (opt == IPOPT_NOP) 1254 optlen = 1; 1255 else { 1256 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 1257 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1258 goto bad; 1259 } 1260 optlen = cp[IPOPT_OLEN]; 1261 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 1262 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1263 goto bad; 1264 } 1265 } 1266 switch (opt) { 1267 1268 default: 1269 break; 1270 1271 /* 1272 * Source routing with record. 1273 * Find interface with current destination address. 1274 * If none on this machine then drop if strictly routed, 1275 * or do nothing if loosely routed. 1276 * Record interface address and bring up next address 1277 * component. If strictly routed make sure next 1278 * address is on directly accessible net. 1279 */ 1280 case IPOPT_LSRR: 1281 case IPOPT_SSRR: 1282 if (ip_allowsrcrt == 0) { 1283 type = ICMP_UNREACH; 1284 code = ICMP_UNREACH_NET_PROHIB; 1285 goto bad; 1286 } 1287 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1288 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1289 goto bad; 1290 } 1291 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1292 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1293 goto bad; 1294 } 1295 ipaddr.sin_addr = ip->ip_dst; 1296 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1297 if (ia == 0) { 1298 if (opt == IPOPT_SSRR) { 1299 type = ICMP_UNREACH; 1300 code = ICMP_UNREACH_SRCFAIL; 1301 goto bad; 1302 } 1303 /* 1304 * Loose routing, and not at next destination 1305 * yet; nothing to do except forward. 1306 */ 1307 break; 1308 } 1309 off--; /* 0 origin */ 1310 if ((off + sizeof(struct in_addr)) > optlen) { 1311 /* 1312 * End of source route. Should be for us. 1313 */ 1314 save_rte(cp, ip->ip_src); 1315 break; 1316 } 1317 /* 1318 * locate outgoing interface 1319 */ 1320 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 1321 sizeof(ipaddr.sin_addr)); 1322 if (opt == IPOPT_SSRR) 1323 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1324 else 1325 ia = ip_rtaddr(ipaddr.sin_addr); 1326 if (ia == 0) { 1327 type = ICMP_UNREACH; 1328 code = ICMP_UNREACH_SRCFAIL; 1329 goto bad; 1330 } 1331 ip->ip_dst = ipaddr.sin_addr; 1332 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1333 (caddr_t)(cp + off), sizeof(struct in_addr)); 1334 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1335 /* 1336 * Let ip_intr's mcast routing check handle mcast pkts 1337 */ 1338 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1339 break; 1340 1341 case IPOPT_RR: 1342 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1343 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1344 goto bad; 1345 } 1346 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1347 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1348 goto bad; 1349 } 1350 /* 1351 * If no space remains, ignore. 1352 */ 1353 off--; /* 0 origin */ 1354 if ((off + sizeof(struct in_addr)) > optlen) 1355 break; 1356 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 1357 sizeof(ipaddr.sin_addr)); 1358 /* 1359 * locate outgoing interface; if we're the destination, 1360 * use the incoming interface (should be same). 1361 */ 1362 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 1363 == NULL && 1364 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 1365 type = ICMP_UNREACH; 1366 code = ICMP_UNREACH_HOST; 1367 goto bad; 1368 } 1369 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1370 (caddr_t)(cp + off), sizeof(struct in_addr)); 1371 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1372 break; 1373 1374 case IPOPT_TS: 1375 code = cp - (u_char *)ip; 1376 ipt = (struct ip_timestamp *)cp; 1377 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1378 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1379 goto bad; 1380 } 1381 if (ipt->ipt_ptr < 5) { 1382 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1383 goto bad; 1384 } 1385 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1386 if (++ipt->ipt_oflw == 0) { 1387 code = (u_char *)&ipt->ipt_ptr - 1388 (u_char *)ip; 1389 goto bad; 1390 } 1391 break; 1392 } 1393 cp0 = (cp + ipt->ipt_ptr - 1); 1394 switch (ipt->ipt_flg) { 1395 1396 case IPOPT_TS_TSONLY: 1397 break; 1398 1399 case IPOPT_TS_TSANDADDR: 1400 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1401 sizeof(struct in_addr) > ipt->ipt_len) { 1402 code = (u_char *)&ipt->ipt_ptr - 1403 (u_char *)ip; 1404 goto bad; 1405 } 1406 ipaddr.sin_addr = dst; 1407 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1408 m->m_pkthdr.rcvif)); 1409 if (ia == 0) 1410 continue; 1411 bcopy(&ia->ia_addr.sin_addr, 1412 cp0, sizeof(struct in_addr)); 1413 ipt->ipt_ptr += sizeof(struct in_addr); 1414 break; 1415 1416 case IPOPT_TS_PRESPEC: 1417 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1418 sizeof(struct in_addr) > ipt->ipt_len) { 1419 code = (u_char *)&ipt->ipt_ptr - 1420 (u_char *)ip; 1421 goto bad; 1422 } 1423 bcopy(cp0, &ipaddr.sin_addr, 1424 sizeof(struct in_addr)); 1425 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1426 == NULL) 1427 continue; 1428 ipt->ipt_ptr += sizeof(struct in_addr); 1429 break; 1430 1431 default: 1432 /* XXX can't take &ipt->ipt_flg */ 1433 code = (u_char *)&ipt->ipt_ptr - 1434 (u_char *)ip + 1; 1435 goto bad; 1436 } 1437 ntime = iptime(); 1438 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1439 bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1, 1440 sizeof(n_time)); 1441 ipt->ipt_ptr += sizeof(n_time); 1442 } 1443 } 1444 if (forward) { 1445 if (ip_forwsrcrt == 0) { 1446 type = ICMP_UNREACH; 1447 code = ICMP_UNREACH_SRCFAIL; 1448 goto bad; 1449 } 1450 ip_forward(m, 1); 1451 return (1); 1452 } 1453 return (0); 1454 bad: 1455 icmp_error(m, type, code, 0, 0); 1456 ipstat.ips_badoptions++; 1457 return (1); 1458 } 1459 1460 /* 1461 * Given address of next destination (final or next hop), 1462 * return internet address info of interface to be used to get there. 1463 */ 1464 struct in_ifaddr * 1465 ip_rtaddr(dst) 1466 struct in_addr dst; 1467 { 1468 struct sockaddr_in *sin; 1469 1470 sin = satosin(&ipforward_rt.ro_dst); 1471 1472 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 1473 if (ipforward_rt.ro_rt) { 1474 RTFREE(ipforward_rt.ro_rt); 1475 ipforward_rt.ro_rt = 0; 1476 } 1477 sin->sin_family = AF_INET; 1478 sin->sin_len = sizeof(*sin); 1479 sin->sin_addr = dst; 1480 1481 rtalloc(&ipforward_rt); 1482 } 1483 if (ipforward_rt.ro_rt == 0) 1484 return ((struct in_ifaddr *)0); 1485 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 1486 } 1487 1488 /* 1489 * Save incoming source route for use in replies, 1490 * to be picked up later by ip_srcroute if the receiver is interested. 1491 */ 1492 void 1493 save_rte(option, dst) 1494 u_char *option; 1495 struct in_addr dst; 1496 { 1497 unsigned olen; 1498 1499 olen = option[IPOPT_OLEN]; 1500 #ifdef DIAGNOSTIC 1501 if (ipprintfs) 1502 printf("save_rte: olen %d\n", olen); 1503 #endif /* 0 */ 1504 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1505 return; 1506 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 1507 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1508 ip_srcrt.dst = dst; 1509 } 1510 1511 /* 1512 * Retrieve incoming source route for use in replies, 1513 * in the same form used by setsockopt. 1514 * The first hop is placed before the options, will be removed later. 1515 */ 1516 struct mbuf * 1517 ip_srcroute() 1518 { 1519 struct in_addr *p, *q; 1520 struct mbuf *m; 1521 1522 if (ip_nhops == 0) 1523 return ((struct mbuf *)0); 1524 m = m_get(M_DONTWAIT, MT_SOOPTS); 1525 if (m == 0) 1526 return ((struct mbuf *)0); 1527 1528 MCLAIM(m, &inetdomain.dom_mowner); 1529 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1530 1531 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1532 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1533 OPTSIZ; 1534 #ifdef DIAGNOSTIC 1535 if (ipprintfs) 1536 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1537 #endif 1538 1539 /* 1540 * First save first hop for return route 1541 */ 1542 p = &ip_srcrt.route[ip_nhops - 1]; 1543 *(mtod(m, struct in_addr *)) = *p--; 1544 #ifdef DIAGNOSTIC 1545 if (ipprintfs) 1546 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1547 #endif 1548 1549 /* 1550 * Copy option fields and padding (nop) to mbuf. 1551 */ 1552 ip_srcrt.nop = IPOPT_NOP; 1553 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1554 bcopy((caddr_t)&ip_srcrt.nop, 1555 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 1556 q = (struct in_addr *)(mtod(m, caddr_t) + 1557 sizeof(struct in_addr) + OPTSIZ); 1558 #undef OPTSIZ 1559 /* 1560 * Record return path as an IP source route, 1561 * reversing the path (pointers are now aligned). 1562 */ 1563 while (p >= ip_srcrt.route) { 1564 #ifdef DIAGNOSTIC 1565 if (ipprintfs) 1566 printf(" %x", ntohl(q->s_addr)); 1567 #endif 1568 *q++ = *p--; 1569 } 1570 /* 1571 * Last hop goes to final destination. 1572 */ 1573 *q = ip_srcrt.dst; 1574 #ifdef DIAGNOSTIC 1575 if (ipprintfs) 1576 printf(" %x\n", ntohl(q->s_addr)); 1577 #endif 1578 return (m); 1579 } 1580 1581 /* 1582 * Strip out IP options, at higher 1583 * level protocol in the kernel. 1584 * Second argument is buffer to which options 1585 * will be moved, and return value is their length. 1586 * XXX should be deleted; last arg currently ignored. 1587 */ 1588 void 1589 ip_stripoptions(m, mopt) 1590 struct mbuf *m; 1591 struct mbuf *mopt; 1592 { 1593 int i; 1594 struct ip *ip = mtod(m, struct ip *); 1595 caddr_t opts; 1596 int olen; 1597 1598 olen = (ip->ip_hl << 2) - sizeof (struct ip); 1599 opts = (caddr_t)(ip + 1); 1600 i = m->m_len - (sizeof (struct ip) + olen); 1601 bcopy(opts + olen, opts, (unsigned)i); 1602 m->m_len -= olen; 1603 if (m->m_flags & M_PKTHDR) 1604 m->m_pkthdr.len -= olen; 1605 ip->ip_len = htons(ntohs(ip->ip_len) - olen); 1606 ip->ip_hl = sizeof (struct ip) >> 2; 1607 } 1608 1609 const int inetctlerrmap[PRC_NCMDS] = { 1610 0, 0, 0, 0, 1611 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1612 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1613 EMSGSIZE, EHOSTUNREACH, 0, 0, 1614 0, 0, 0, 0, 1615 ENOPROTOOPT 1616 }; 1617 1618 /* 1619 * Forward a packet. If some error occurs return the sender 1620 * an icmp packet. Note we can't always generate a meaningful 1621 * icmp message because icmp doesn't have a large enough repertoire 1622 * of codes and types. 1623 * 1624 * If not forwarding, just drop the packet. This could be confusing 1625 * if ipforwarding was zero but some routing protocol was advancing 1626 * us as a gateway to somewhere. However, we must let the routing 1627 * protocol deal with that. 1628 * 1629 * The srcrt parameter indicates whether the packet is being forwarded 1630 * via a source route. 1631 */ 1632 void 1633 ip_forward(m, srcrt) 1634 struct mbuf *m; 1635 int srcrt; 1636 { 1637 struct ip *ip = mtod(m, struct ip *); 1638 struct sockaddr_in *sin; 1639 struct rtentry *rt; 1640 int error, type = 0, code = 0; 1641 struct mbuf *mcopy; 1642 n_long dest; 1643 struct ifnet *destifp; 1644 #if defined(IPSEC) || defined(FAST_IPSEC) 1645 struct ifnet dummyifp; 1646 #endif 1647 1648 /* 1649 * We are now in the output path. 1650 */ 1651 MCLAIM(m, &ip_tx_mowner); 1652 1653 /* 1654 * Clear any in-bound checksum flags for this packet. 1655 */ 1656 m->m_pkthdr.csum_flags = 0; 1657 1658 dest = 0; 1659 #ifdef DIAGNOSTIC 1660 if (ipprintfs) 1661 printf("forward: src %2.2x dst %2.2x ttl %x\n", 1662 ntohl(ip->ip_src.s_addr), 1663 ntohl(ip->ip_dst.s_addr), ip->ip_ttl); 1664 #endif 1665 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1666 ipstat.ips_cantforward++; 1667 m_freem(m); 1668 return; 1669 } 1670 if (ip->ip_ttl <= IPTTLDEC) { 1671 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1672 return; 1673 } 1674 ip->ip_ttl -= IPTTLDEC; 1675 1676 sin = satosin(&ipforward_rt.ro_dst); 1677 if ((rt = ipforward_rt.ro_rt) == 0 || 1678 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1679 if (ipforward_rt.ro_rt) { 1680 RTFREE(ipforward_rt.ro_rt); 1681 ipforward_rt.ro_rt = 0; 1682 } 1683 sin->sin_family = AF_INET; 1684 sin->sin_len = sizeof(struct sockaddr_in); 1685 sin->sin_addr = ip->ip_dst; 1686 1687 rtalloc(&ipforward_rt); 1688 if (ipforward_rt.ro_rt == 0) { 1689 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1690 return; 1691 } 1692 rt = ipforward_rt.ro_rt; 1693 } 1694 1695 /* 1696 * Save at most 68 bytes of the packet in case 1697 * we need to generate an ICMP message to the src. 1698 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1699 */ 1700 mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); 1701 if (mcopy) 1702 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1703 1704 /* 1705 * If forwarding packet using same interface that it came in on, 1706 * perhaps should send a redirect to sender to shortcut a hop. 1707 * Only send redirect if source is sending directly to us, 1708 * and if packet was not source routed (or has any options). 1709 * Also, don't send redirect if forwarding using a default route 1710 * or a route modified by a redirect. 1711 */ 1712 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1713 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1714 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1715 ipsendredirects && !srcrt) { 1716 if (rt->rt_ifa && 1717 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1718 ifatoia(rt->rt_ifa)->ia_subnet) { 1719 if (rt->rt_flags & RTF_GATEWAY) 1720 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1721 else 1722 dest = ip->ip_dst.s_addr; 1723 /* 1724 * Router requirements says to only send host 1725 * redirects. 1726 */ 1727 type = ICMP_REDIRECT; 1728 code = ICMP_REDIRECT_HOST; 1729 #ifdef DIAGNOSTIC 1730 if (ipprintfs) 1731 printf("redirect (%d) to %x\n", code, 1732 (u_int32_t)dest); 1733 #endif 1734 } 1735 } 1736 1737 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1738 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 1739 (struct ip_moptions *)NULL, (struct socket *)NULL); 1740 1741 if (error) 1742 ipstat.ips_cantforward++; 1743 else { 1744 ipstat.ips_forward++; 1745 if (type) 1746 ipstat.ips_redirectsent++; 1747 else { 1748 if (mcopy) { 1749 #ifdef GATEWAY 1750 if (mcopy->m_flags & M_CANFASTFWD) 1751 ipflow_create(&ipforward_rt, mcopy); 1752 #endif 1753 m_freem(mcopy); 1754 } 1755 return; 1756 } 1757 } 1758 if (mcopy == NULL) 1759 return; 1760 destifp = NULL; 1761 1762 switch (error) { 1763 1764 case 0: /* forwarded, but need redirect */ 1765 /* type, code set above */ 1766 break; 1767 1768 case ENETUNREACH: /* shouldn't happen, checked above */ 1769 case EHOSTUNREACH: 1770 case ENETDOWN: 1771 case EHOSTDOWN: 1772 default: 1773 type = ICMP_UNREACH; 1774 code = ICMP_UNREACH_HOST; 1775 break; 1776 1777 case EMSGSIZE: 1778 type = ICMP_UNREACH; 1779 code = ICMP_UNREACH_NEEDFRAG; 1780 #if !defined(IPSEC) && !defined(FAST_IPSEC) 1781 if (ipforward_rt.ro_rt) 1782 destifp = ipforward_rt.ro_rt->rt_ifp; 1783 #else 1784 /* 1785 * If the packet is routed over IPsec tunnel, tell the 1786 * originator the tunnel MTU. 1787 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz 1788 * XXX quickhack!!! 1789 */ 1790 if (ipforward_rt.ro_rt) { 1791 struct secpolicy *sp; 1792 int ipsecerror; 1793 size_t ipsechdr; 1794 struct route *ro; 1795 1796 sp = ipsec4_getpolicybyaddr(mcopy, 1797 IPSEC_DIR_OUTBOUND, IP_FORWARDING, 1798 &ipsecerror); 1799 1800 if (sp == NULL) 1801 destifp = ipforward_rt.ro_rt->rt_ifp; 1802 else { 1803 /* count IPsec header size */ 1804 ipsechdr = ipsec4_hdrsiz(mcopy, 1805 IPSEC_DIR_OUTBOUND, NULL); 1806 1807 /* 1808 * find the correct route for outer IPv4 1809 * header, compute tunnel MTU. 1810 * 1811 * XXX BUG ALERT 1812 * The "dummyifp" code relies upon the fact 1813 * that icmp_error() touches only ifp->if_mtu. 1814 */ 1815 /*XXX*/ 1816 destifp = NULL; 1817 if (sp->req != NULL 1818 && sp->req->sav != NULL 1819 && sp->req->sav->sah != NULL) { 1820 ro = &sp->req->sav->sah->sa_route; 1821 if (ro->ro_rt && ro->ro_rt->rt_ifp) { 1822 dummyifp.if_mtu = 1823 ro->ro_rt->rt_rmx.rmx_mtu ? 1824 ro->ro_rt->rt_rmx.rmx_mtu : 1825 ro->ro_rt->rt_ifp->if_mtu; 1826 dummyifp.if_mtu -= ipsechdr; 1827 destifp = &dummyifp; 1828 } 1829 } 1830 1831 #ifdef IPSEC 1832 key_freesp(sp); 1833 #else 1834 KEY_FREESP(&sp); 1835 #endif 1836 } 1837 } 1838 #endif /*IPSEC*/ 1839 ipstat.ips_cantfrag++; 1840 break; 1841 1842 case ENOBUFS: 1843 #if 1 1844 /* 1845 * a router should not generate ICMP_SOURCEQUENCH as 1846 * required in RFC1812 Requirements for IP Version 4 Routers. 1847 * source quench could be a big problem under DoS attacks, 1848 * or if the underlying interface is rate-limited. 1849 */ 1850 if (mcopy) 1851 m_freem(mcopy); 1852 return; 1853 #else 1854 type = ICMP_SOURCEQUENCH; 1855 code = 0; 1856 break; 1857 #endif 1858 } 1859 icmp_error(mcopy, type, code, dest, destifp); 1860 } 1861 1862 void 1863 ip_savecontrol(inp, mp, ip, m) 1864 struct inpcb *inp; 1865 struct mbuf **mp; 1866 struct ip *ip; 1867 struct mbuf *m; 1868 { 1869 1870 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1871 struct timeval tv; 1872 1873 microtime(&tv); 1874 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1875 SCM_TIMESTAMP, SOL_SOCKET); 1876 if (*mp) 1877 mp = &(*mp)->m_next; 1878 } 1879 if (inp->inp_flags & INP_RECVDSTADDR) { 1880 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1881 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1882 if (*mp) 1883 mp = &(*mp)->m_next; 1884 } 1885 #ifdef notyet 1886 /* 1887 * XXX 1888 * Moving these out of udp_input() made them even more broken 1889 * than they already were. 1890 * - fenner (at) parc.xerox.com 1891 */ 1892 /* options were tossed already */ 1893 if (inp->inp_flags & INP_RECVOPTS) { 1894 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1895 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1896 if (*mp) 1897 mp = &(*mp)->m_next; 1898 } 1899 /* ip_srcroute doesn't do what we want here, need to fix */ 1900 if (inp->inp_flags & INP_RECVRETOPTS) { 1901 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1902 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1903 if (*mp) 1904 mp = &(*mp)->m_next; 1905 } 1906 #endif 1907 if (inp->inp_flags & INP_RECVIF) { 1908 struct sockaddr_dl sdl; 1909 1910 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1911 sdl.sdl_family = AF_LINK; 1912 sdl.sdl_index = m->m_pkthdr.rcvif ? 1913 m->m_pkthdr.rcvif->if_index : 0; 1914 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1915 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1916 IP_RECVIF, IPPROTO_IP); 1917 if (*mp) 1918 mp = &(*mp)->m_next; 1919 } 1920 } 1921 1922 int 1923 ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1924 int *name; 1925 u_int namelen; 1926 void *oldp; 1927 size_t *oldlenp; 1928 void *newp; 1929 size_t newlen; 1930 { 1931 extern int subnetsarelocal, hostzeroisbroadcast; 1932 1933 int error, old; 1934 1935 /* All sysctl names (except ifq.*) at this level are terminal. */ 1936 if ((namelen != 1) && !(namelen == 2 && name[0] == IPCTL_IFQ)) 1937 return (ENOTDIR); 1938 1939 switch (name[0]) { 1940 case IPCTL_FORWARDING: 1941 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1942 case IPCTL_SENDREDIRECTS: 1943 return (sysctl_int(oldp, oldlenp, newp, newlen, 1944 &ipsendredirects)); 1945 case IPCTL_DEFTTL: 1946 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1947 #ifdef notyet 1948 case IPCTL_DEFMTU: 1949 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1950 #endif 1951 case IPCTL_FORWSRCRT: 1952 /* Don't allow this to change in a secure environment. */ 1953 if (securelevel > 0) 1954 return (sysctl_rdint(oldp, oldlenp, newp, 1955 ip_forwsrcrt)); 1956 else 1957 return (sysctl_int(oldp, oldlenp, newp, newlen, 1958 &ip_forwsrcrt)); 1959 case IPCTL_DIRECTEDBCAST: 1960 return (sysctl_int(oldp, oldlenp, newp, newlen, 1961 &ip_directedbcast)); 1962 case IPCTL_ALLOWSRCRT: 1963 return (sysctl_int(oldp, oldlenp, newp, newlen, 1964 &ip_allowsrcrt)); 1965 case IPCTL_SUBNETSARELOCAL: 1966 return (sysctl_int(oldp, oldlenp, newp, newlen, 1967 &subnetsarelocal)); 1968 case IPCTL_MTUDISC: 1969 error = sysctl_int(oldp, oldlenp, newp, newlen, 1970 &ip_mtudisc); 1971 if (error == 0 && ip_mtudisc == 0) 1972 rt_timer_queue_remove_all(ip_mtudisc_timeout_q, TRUE); 1973 return error; 1974 case IPCTL_ANONPORTMIN: 1975 old = anonportmin; 1976 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1977 if (anonportmin >= anonportmax || anonportmin < 0 1978 || anonportmin > 65535 1979 #ifndef IPNOPRIVPORTS 1980 || anonportmin < IPPORT_RESERVED 1981 #endif 1982 ) { 1983 anonportmin = old; 1984 return (EINVAL); 1985 } 1986 return (error); 1987 case IPCTL_ANONPORTMAX: 1988 old = anonportmax; 1989 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1990 if (anonportmin >= anonportmax || anonportmax < 0 1991 || anonportmax > 65535 1992 #ifndef IPNOPRIVPORTS 1993 || anonportmax < IPPORT_RESERVED 1994 #endif 1995 ) { 1996 anonportmax = old; 1997 return (EINVAL); 1998 } 1999 return (error); 2000 case IPCTL_MTUDISCTIMEOUT: 2001 old = ip_mtudisc_timeout; 2002 error = sysctl_int(oldp, oldlenp, newp, newlen, 2003 &ip_mtudisc_timeout); 2004 if (ip_mtudisc_timeout < 0) { 2005 ip_mtudisc_timeout = old; 2006 return (EINVAL); 2007 } 2008 if (error == 0) 2009 rt_timer_queue_change(ip_mtudisc_timeout_q, 2010 ip_mtudisc_timeout); 2011 return (error); 2012 #ifdef GATEWAY 2013 case IPCTL_MAXFLOWS: 2014 { 2015 int s; 2016 2017 error = sysctl_int(oldp, oldlenp, newp, newlen, 2018 &ip_maxflows); 2019 s = splsoftnet(); 2020 ipflow_reap(0); 2021 splx(s); 2022 return (error); 2023 } 2024 #endif 2025 case IPCTL_HOSTZEROBROADCAST: 2026 return (sysctl_int(oldp, oldlenp, newp, newlen, 2027 &hostzeroisbroadcast)); 2028 #if NGIF > 0 2029 case IPCTL_GIF_TTL: 2030 return (sysctl_int(oldp, oldlenp, newp, newlen, 2031 &ip_gif_ttl)); 2032 #endif 2033 2034 #if NGRE > 0 2035 case IPCTL_GRE_TTL: 2036 return (sysctl_int(oldp, oldlenp, newp, newlen, 2037 &ip_gre_ttl)); 2038 #endif 2039 2040 #ifndef IPNOPRIVPORTS 2041 case IPCTL_LOWPORTMIN: 2042 old = lowportmin; 2043 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin); 2044 if (lowportmin >= lowportmax 2045 || lowportmin > IPPORT_RESERVEDMAX 2046 || lowportmin < IPPORT_RESERVEDMIN 2047 ) { 2048 lowportmin = old; 2049 return (EINVAL); 2050 } 2051 return (error); 2052 case IPCTL_LOWPORTMAX: 2053 old = lowportmax; 2054 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax); 2055 if (lowportmin >= lowportmax 2056 || lowportmax > IPPORT_RESERVEDMAX 2057 || lowportmax < IPPORT_RESERVEDMIN 2058 ) { 2059 lowportmax = old; 2060 return (EINVAL); 2061 } 2062 return (error); 2063 #endif 2064 2065 case IPCTL_MAXFRAGPACKETS: 2066 return (sysctl_int(oldp, oldlenp, newp, newlen, 2067 &ip_maxfragpackets)); 2068 2069 case IPCTL_CHECKINTERFACE: 2070 return (sysctl_int(oldp, oldlenp, newp, newlen, 2071 &ip_checkinterface)); 2072 2073 case IPCTL_IFQ: 2074 return (sysctl_ifq(name + 1, namelen - 1, oldp, oldlenp, 2075 newp, newlen, &ipintrq)); 2076 2077 case IPCTL_RANDOMID: 2078 #ifdef RANDOM_IP_ID 2079 return (sysctl_int(oldp, oldlenp, newp, newlen, 2080 &ip_do_randomid)); 2081 #else 2082 return (sysctl_rdint(oldp, oldlenp, newp, 0)); 2083 #endif 2084 2085 default: 2086 return (EOPNOTSUPP); 2087 } 2088 /* NOTREACHED */ 2089 } 2090
Indexes created Wed Sep 24 14:09:57 GMT 2025