ip_input.c revision 1.114.4.8
1 /* $NetBSD: ip_input.c,v 1.114.4.8 2002/02/26 21:07:56 he 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. All advertising materials mentioning features or use of this software 82 * must display the following acknowledgement: 83 * This product includes software developed by the University of 84 * California, Berkeley and its contributors. 85 * 4. Neither the name of the University nor the names of its contributors 86 * may be used to endorse or promote products derived from this software 87 * without specific prior written permission. 88 * 89 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 90 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 91 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 92 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 93 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 94 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 95 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 96 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 97 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 98 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 99 * SUCH DAMAGE. 100 * 101 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 102 */ 103 104 #include "opt_gateway.h" 105 #include "opt_pfil_hooks.h" 106 #include "opt_ipsec.h" 107 #include "opt_mrouting.h" 108 109 #include <sys/param.h> 110 #include <sys/systm.h> 111 #include <sys/malloc.h> 112 #include <sys/mbuf.h> 113 #include <sys/domain.h> 114 #include <sys/protosw.h> 115 #include <sys/socket.h> 116 #include <sys/socketvar.h> 117 #include <sys/errno.h> 118 #include <sys/time.h> 119 #include <sys/kernel.h> 120 #include <sys/proc.h> 121 #include <sys/pool.h> 122 123 #include <vm/vm.h> 124 #include <sys/sysctl.h> 125 126 #include <net/if.h> 127 #include <net/if_dl.h> 128 #include <net/route.h> 129 #include <net/pfil.h> 130 131 #include <netinet/in.h> 132 #include <netinet/in_systm.h> 133 #include <netinet/ip.h> 134 #include <netinet/in_pcb.h> 135 #include <netinet/in_var.h> 136 #include <netinet/ip_var.h> 137 #include <netinet/ip_icmp.h> 138 /* just for gif_ttl */ 139 #include <netinet/in_gif.h> 140 #include "gif.h" 141 #include <net/if_gre.h> 142 #include "gre.h" 143 144 #ifdef MROUTING 145 #include <netinet/ip_mroute.h> 146 #endif 147 148 #ifdef IPSEC 149 #include <netinet6/ipsec.h> 150 #include <netkey/key.h> 151 #include <netkey/key_debug.h> 152 #endif 153 154 #ifndef IPFORWARDING 155 #ifdef GATEWAY 156 #define IPFORWARDING 1 /* forward IP packets not for us */ 157 #else /* GATEWAY */ 158 #define IPFORWARDING 0 /* don't forward IP packets not for us */ 159 #endif /* GATEWAY */ 160 #endif /* IPFORWARDING */ 161 #ifndef IPSENDREDIRECTS 162 #define IPSENDREDIRECTS 1 163 #endif 164 #ifndef IPFORWSRCRT 165 #define IPFORWSRCRT 1 /* forward source-routed packets */ 166 #endif 167 #ifndef IPALLOWSRCRT 168 #define IPALLOWSRCRT 1 /* allow source-routed packets */ 169 #endif 170 #ifndef IPMTUDISC 171 #define IPMTUDISC 0 172 #endif 173 #ifndef IPMTUDISCTIMEOUT 174 #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ 175 #endif 176 177 /* 178 * Note: DIRECTED_BROADCAST is handled this way so that previous 179 * configuration using this option will Just Work. 180 */ 181 #ifndef IPDIRECTEDBCAST 182 #ifdef DIRECTED_BROADCAST 183 #define IPDIRECTEDBCAST 1 184 #else 185 #define IPDIRECTEDBCAST 0 186 #endif /* DIRECTED_BROADCAST */ 187 #endif /* IPDIRECTEDBCAST */ 188 int ipforwarding = IPFORWARDING; 189 int ipsendredirects = IPSENDREDIRECTS; 190 int ip_defttl = IPDEFTTL; 191 int ip_forwsrcrt = IPFORWSRCRT; 192 int ip_directedbcast = IPDIRECTEDBCAST; 193 int ip_allowsrcrt = IPALLOWSRCRT; 194 int ip_mtudisc = IPMTUDISC; 195 u_int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; 196 #ifdef DIAGNOSTIC 197 int ipprintfs = 0; 198 #endif 199 200 struct rttimer_queue *ip_mtudisc_timeout_q = NULL; 201 202 extern struct domain inetdomain; 203 int ipqmaxlen = IFQ_MAXLEN; 204 struct in_ifaddrhead in_ifaddr; 205 struct in_ifaddrhashhead *in_ifaddrhashtbl; 206 struct ifqueue ipintrq; 207 struct ipstat ipstat; 208 u_int16_t ip_id; 209 210 struct ipqhead ipq; 211 int ipq_locked; 212 int ip_nfragpackets = 0; 213 int ip_maxfragpackets = 200; 214 215 static __inline int ipq_lock_try __P((void)); 216 static __inline void ipq_unlock __P((void)); 217 218 static __inline int 219 ipq_lock_try() 220 { 221 int s; 222 223 s = splimp(); 224 if (ipq_locked) { 225 splx(s); 226 return (0); 227 } 228 ipq_locked = 1; 229 splx(s); 230 return (1); 231 } 232 233 static __inline void 234 ipq_unlock() 235 { 236 int s; 237 238 s = splimp(); 239 ipq_locked = 0; 240 splx(s); 241 } 242 243 #ifdef DIAGNOSTIC 244 #define IPQ_LOCK() \ 245 do { \ 246 if (ipq_lock_try() == 0) { \ 247 printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \ 248 panic("ipq_lock"); \ 249 } \ 250 } while (0) 251 #define IPQ_LOCK_CHECK() \ 252 do { \ 253 if (ipq_locked == 0) { \ 254 printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \ 255 panic("ipq lock check"); \ 256 } \ 257 } while (0) 258 #else 259 #define IPQ_LOCK() (void) ipq_lock_try() 260 #define IPQ_LOCK_CHECK() /* nothing */ 261 #endif 262 263 #define IPQ_UNLOCK() ipq_unlock() 264 265 struct pool ipqent_pool; 266 267 /* 268 * We need to save the IP options in case a protocol wants to respond 269 * to an incoming packet over the same route if the packet got here 270 * using IP source routing. This allows connection establishment and 271 * maintenance when the remote end is on a network that is not known 272 * to us. 273 */ 274 int ip_nhops = 0; 275 static struct ip_srcrt { 276 struct in_addr dst; /* final destination */ 277 char nop; /* one NOP to align */ 278 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ 279 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; 280 } ip_srcrt; 281 282 static void save_rte __P((u_char *, struct in_addr)); 283 284 /* 285 * IP initialization: fill in IP protocol switch table. 286 * All protocols not implemented in kernel go to raw IP protocol handler. 287 */ 288 void 289 ip_init() 290 { 291 struct protosw *pr; 292 int i; 293 294 pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", 295 0, NULL, NULL, M_IPQ); 296 297 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); 298 if (pr == 0) 299 panic("ip_init"); 300 for (i = 0; i < IPPROTO_MAX; i++) 301 ip_protox[i] = pr - inetsw; 302 for (pr = inetdomain.dom_protosw; 303 pr < inetdomain.dom_protoswNPROTOSW; pr++) 304 if (pr->pr_domain->dom_family == PF_INET && 305 pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) 306 ip_protox[pr->pr_protocol] = pr - inetsw; 307 LIST_INIT(&ipq); 308 ip_id = time.tv_sec & 0xffff; 309 ipintrq.ifq_maxlen = ipqmaxlen; 310 TAILQ_INIT(&in_ifaddr); 311 in_ifaddrhashtbl = 312 hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, M_WAITOK, &in_ifaddrhash); 313 if (ip_mtudisc != 0) 314 ip_mtudisc_timeout_q = 315 rt_timer_queue_create(ip_mtudisc_timeout); 316 #ifdef GATEWAY 317 ipflow_init(); 318 #endif 319 } 320 321 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 322 struct route ipforward_rt; 323 324 /* 325 * IP software interrupt routine 326 */ 327 void 328 ipintr() 329 { 330 int s; 331 struct mbuf *m; 332 333 while (1) { 334 s = splimp(); 335 IF_DEQUEUE(&ipintrq, m); 336 splx(s); 337 if (m == 0) 338 return; 339 ip_input(m); 340 } 341 } 342 343 /* 344 * Ip input routine. Checksum and byte swap header. If fragmented 345 * try to reassemble. Process options. Pass to next level. 346 */ 347 void 348 ip_input(struct mbuf *m) 349 { 350 struct ip *ip = NULL; 351 struct ipq *fp; 352 struct in_ifaddr *ia; 353 struct ifaddr *ifa; 354 struct ipqent *ipqe; 355 int hlen = 0, mff, len; 356 int downmatch; 357 #ifdef PFIL_HOOKS 358 struct packet_filter_hook *pfh; 359 struct mbuf *m0; 360 int rv; 361 #endif /* PFIL_HOOKS */ 362 363 #ifdef DIAGNOSTIC 364 if ((m->m_flags & M_PKTHDR) == 0) 365 panic("ipintr no HDR"); 366 #endif 367 #ifdef IPSEC 368 /* 369 * should the inner packet be considered authentic? 370 * see comment in ah4_input(). 371 */ 372 if (m) { 373 m->m_flags &= ~M_AUTHIPHDR; 374 m->m_flags &= ~M_AUTHIPDGM; 375 } 376 #endif 377 /* 378 * If no IP addresses have been set yet but the interfaces 379 * are receiving, can't do anything with incoming packets yet. 380 */ 381 if (in_ifaddr.tqh_first == 0) 382 goto bad; 383 ipstat.ips_total++; 384 if (m->m_len < sizeof (struct ip) && 385 (m = m_pullup(m, sizeof (struct ip))) == 0) { 386 ipstat.ips_toosmall++; 387 return; 388 } 389 ip = mtod(m, struct ip *); 390 if (ip->ip_v != IPVERSION) { 391 ipstat.ips_badvers++; 392 goto bad; 393 } 394 hlen = ip->ip_hl << 2; 395 if (hlen < sizeof(struct ip)) { /* minimum header length */ 396 ipstat.ips_badhlen++; 397 goto bad; 398 } 399 if (hlen > m->m_len) { 400 if ((m = m_pullup(m, hlen)) == 0) { 401 ipstat.ips_badhlen++; 402 return; 403 } 404 ip = mtod(m, struct ip *); 405 } 406 407 /* 408 * RFC1122: packets with a multicast source address are 409 * not allowed. 410 */ 411 if (IN_MULTICAST(ip->ip_src.s_addr)) { 412 /* XXX stat */ 413 goto bad; 414 } 415 416 if (in_cksum(m, hlen) != 0) { 417 ipstat.ips_badsum++; 418 goto bad; 419 } 420 421 /* 422 * Convert fields to host representation. 423 */ 424 NTOHS(ip->ip_len); 425 NTOHS(ip->ip_off); 426 len = ip->ip_len; 427 428 /* 429 * Check for additional length bogosity 430 */ 431 if (len < hlen) { 432 ipstat.ips_badlen++; 433 goto bad; 434 } 435 436 /* 437 * Check that the amount of data in the buffers 438 * is as at least much as the IP header would have us expect. 439 * Trim mbufs if longer than we expect. 440 * Drop packet if shorter than we expect. 441 */ 442 if (m->m_pkthdr.len < len) { 443 ipstat.ips_tooshort++; 444 goto bad; 445 } 446 if (m->m_pkthdr.len > len) { 447 if (m->m_len == m->m_pkthdr.len) { 448 m->m_len = len; 449 m->m_pkthdr.len = len; 450 } else 451 m_adj(m, len - m->m_pkthdr.len); 452 } 453 454 #ifdef IPSEC 455 /* ipflow (IP fast fowarding) is not compatible with IPsec. */ 456 m->m_flags &= ~M_CANFASTFWD; 457 #else 458 /* 459 * Assume that we can create a fast-forward IP flow entry 460 * based on this packet. 461 */ 462 m->m_flags |= M_CANFASTFWD; 463 #endif 464 465 #ifdef PFIL_HOOKS 466 #ifdef IPSEC 467 /* 468 * let ipfilter look at packet on the wire, 469 * not the decapsulated packet. 470 */ 471 if (ipsec_gethist(m, NULL)) 472 goto nofilt; 473 #endif 474 /* 475 * Run through list of hooks for input packets. If there are any 476 * filters which require that additional packets in the flow are 477 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 478 * Note that filters must _never_ set this flag, as another filter 479 * in the list may have previously cleared it. 480 */ 481 m0 = m; 482 pfh = pfil_hook_get(PFIL_IN, &inetsw[ip_protox[IPPROTO_IP]].pr_pfh); 483 for (; pfh; pfh = pfh->pfil_link.tqe_next) 484 if (pfh->pfil_func) { 485 rv = pfh->pfil_func(ip, hlen, 486 m->m_pkthdr.rcvif, 0, &m0); 487 if (rv) 488 return; 489 m = m0; 490 if (m == NULL) 491 return; 492 ip = mtod(m, struct ip *); 493 } 494 #ifdef IPSEC 495 nofilt:; 496 #endif 497 #endif /* PFIL_HOOKS */ 498 499 /* 500 * Process options and, if not destined for us, 501 * ship it on. ip_dooptions returns 1 when an 502 * error was detected (causing an icmp message 503 * to be sent and the original packet to be freed). 504 */ 505 ip_nhops = 0; /* for source routed packets */ 506 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 507 return; 508 509 /* 510 * Check our list of addresses, to see if the packet is for us. 511 * 512 * Traditional 4.4BSD did not consult IFF_UP at all. 513 * The behavior here is to treat addresses on !IFF_UP interface 514 * as not mine. 515 */ 516 downmatch = 0; 517 for (ia = IN_IFADDR_HASH(ip->ip_dst.s_addr).lh_first; 518 ia != NULL; 519 ia = ia->ia_hash.le_next) { 520 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 521 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 522 break; 523 else 524 downmatch++; 525 } 526 } 527 if (ia != NULL) 528 goto ours; 529 if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 530 for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first; 531 ifa != NULL; ifa = ifa->ifa_list.tqe_next) { 532 if (ifa->ifa_addr->sa_family != AF_INET) continue; 533 ia = ifatoia(ifa); 534 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 535 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 536 /* 537 * Look for all-0's host part (old broadcast addr), 538 * either for subnet or net. 539 */ 540 ip->ip_dst.s_addr == ia->ia_subnet || 541 ip->ip_dst.s_addr == ia->ia_net) 542 goto ours; 543 /* 544 * An interface with IP address zero accepts 545 * all packets that arrive on that interface. 546 */ 547 if (in_nullhost(ia->ia_addr.sin_addr)) 548 goto ours; 549 } 550 } 551 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 552 struct in_multi *inm; 553 #ifdef MROUTING 554 extern struct socket *ip_mrouter; 555 556 if (m->m_flags & M_EXT) { 557 if ((m = m_pullup(m, hlen)) == 0) { 558 ipstat.ips_toosmall++; 559 return; 560 } 561 ip = mtod(m, struct ip *); 562 } 563 564 if (ip_mrouter) { 565 /* 566 * If we are acting as a multicast router, all 567 * incoming multicast packets are passed to the 568 * kernel-level multicast forwarding function. 569 * The packet is returned (relatively) intact; if 570 * ip_mforward() returns a non-zero value, the packet 571 * must be discarded, else it may be accepted below. 572 * 573 * (The IP ident field is put in the same byte order 574 * as expected when ip_mforward() is called from 575 * ip_output().) 576 */ 577 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 578 ipstat.ips_cantforward++; 579 m_freem(m); 580 return; 581 } 582 583 /* 584 * The process-level routing demon needs to receive 585 * all multicast IGMP packets, whether or not this 586 * host belongs to their destination groups. 587 */ 588 if (ip->ip_p == IPPROTO_IGMP) 589 goto ours; 590 ipstat.ips_forward++; 591 } 592 #endif 593 /* 594 * See if we belong to the destination multicast group on the 595 * arrival interface. 596 */ 597 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 598 if (inm == NULL) { 599 ipstat.ips_cantforward++; 600 m_freem(m); 601 return; 602 } 603 goto ours; 604 } 605 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 606 in_nullhost(ip->ip_dst)) 607 goto ours; 608 609 /* 610 * Not for us; forward if possible and desirable. 611 */ 612 if (ipforwarding == 0) { 613 ipstat.ips_cantforward++; 614 m_freem(m); 615 } else { 616 /* 617 * If ip_dst matched any of my address on !IFF_UP interface, 618 * and there's no IFF_UP interface that matches ip_dst, 619 * send icmp unreach. Forwarding it will result in in-kernel 620 * forwarding loop till TTL goes to 0. 621 */ 622 if (downmatch) { 623 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 624 ipstat.ips_cantforward++; 625 return; 626 } 627 #ifdef IPSEC 628 if (ipsec4_in_reject(m, NULL)) { 629 ipsecstat.in_polvio++; 630 goto bad; 631 } 632 #endif 633 634 ip_forward(m, 0); 635 } 636 return; 637 638 ours: 639 /* 640 * If offset or IP_MF are set, must reassemble. 641 * Otherwise, nothing need be done. 642 * (We could look in the reassembly queue to see 643 * if the packet was previously fragmented, 644 * but it's not worth the time; just let them time out.) 645 */ 646 if (ip->ip_off & ~(IP_DF|IP_RF)) { 647 /* 648 * Look for queue of fragments 649 * of this datagram. 650 */ 651 IPQ_LOCK(); 652 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 653 if (ip->ip_id == fp->ipq_id && 654 in_hosteq(ip->ip_src, fp->ipq_src) && 655 in_hosteq(ip->ip_dst, fp->ipq_dst) && 656 ip->ip_p == fp->ipq_p) 657 goto found; 658 fp = 0; 659 found: 660 661 /* 662 * Adjust ip_len to not reflect header, 663 * set ipqe_mff if more fragments are expected, 664 * convert offset of this to bytes. 665 */ 666 ip->ip_len -= hlen; 667 mff = (ip->ip_off & IP_MF) != 0; 668 if (mff) { 669 /* 670 * Make sure that fragments have a data length 671 * that's a non-zero multiple of 8 bytes. 672 */ 673 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 674 ipstat.ips_badfrags++; 675 IPQ_UNLOCK(); 676 goto bad; 677 } 678 } 679 ip->ip_off <<= 3; 680 681 /* 682 * If datagram marked as having more fragments 683 * or if this is not the first fragment, 684 * attempt reassembly; if it succeeds, proceed. 685 */ 686 if (mff || ip->ip_off) { 687 ipstat.ips_fragments++; 688 ipqe = pool_get(&ipqent_pool, PR_NOWAIT); 689 if (ipqe == NULL) { 690 ipstat.ips_rcvmemdrop++; 691 IPQ_UNLOCK(); 692 goto bad; 693 } 694 ipqe->ipqe_mff = mff; 695 ipqe->ipqe_m = m; 696 ipqe->ipqe_ip = ip; 697 m = ip_reass(ipqe, fp); 698 if (m == 0) { 699 IPQ_UNLOCK(); 700 return; 701 } 702 ipstat.ips_reassembled++; 703 ip = mtod(m, struct ip *); 704 hlen = ip->ip_hl << 2; 705 ip->ip_len += hlen; 706 } else 707 if (fp) 708 ip_freef(fp); 709 IPQ_UNLOCK(); 710 } 711 712 #ifdef IPSEC 713 /* 714 * enforce IPsec policy checking if we are seeing last header. 715 * note that we do not visit this with protocols with pcb layer 716 * code - like udp/tcp/raw ip. 717 */ 718 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 && 719 ipsec4_in_reject(m, NULL)) { 720 ipsecstat.in_polvio++; 721 goto bad; 722 } 723 #endif 724 725 /* 726 * Switch out to protocol's input routine. 727 */ 728 #if IFA_STATS 729 ia->ia_ifa.ifa_data.ifad_inbytes += ip->ip_len; 730 #endif 731 ipstat.ips_delivered++; 732 { 733 int off = hlen, nh = ip->ip_p; 734 735 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 736 return; 737 } 738 bad: 739 m_freem(m); 740 } 741 742 /* 743 * Take incoming datagram fragment and try to 744 * reassemble it into whole datagram. If a chain for 745 * reassembly of this datagram already exists, then it 746 * is given as fp; otherwise have to make a chain. 747 */ 748 struct mbuf * 749 ip_reass(ipqe, fp) 750 struct ipqent *ipqe; 751 struct ipq *fp; 752 { 753 struct mbuf *m = ipqe->ipqe_m; 754 struct ipqent *nq, *p, *q; 755 struct ip *ip; 756 struct mbuf *t; 757 int hlen = ipqe->ipqe_ip->ip_hl << 2; 758 int i, next; 759 760 IPQ_LOCK_CHECK(); 761 762 /* 763 * Presence of header sizes in mbufs 764 * would confuse code below. 765 */ 766 m->m_data += hlen; 767 m->m_len -= hlen; 768 769 /* 770 * If first fragment to arrive, create a reassembly queue. 771 */ 772 if (fp == 0) { 773 /* 774 * Enforce upper bound on number of fragmented packets 775 * for which we attempt reassembly; 776 * If maxfrag is 0, never accept fragments. 777 * If maxfrag is -1, accept all fragments without limitation. 778 */ 779 if (ip_maxfragpackets < 0) 780 ; 781 else if (ip_nfragpackets >= ip_maxfragpackets) 782 goto dropfrag; 783 ip_nfragpackets++; 784 MALLOC(fp, struct ipq *, sizeof (struct ipq), 785 M_FTABLE, M_NOWAIT); 786 if (fp == NULL) 787 goto dropfrag; 788 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 789 fp->ipq_ttl = IPFRAGTTL; 790 fp->ipq_p = ipqe->ipqe_ip->ip_p; 791 fp->ipq_id = ipqe->ipqe_ip->ip_id; 792 LIST_INIT(&fp->ipq_fragq); 793 fp->ipq_src = ipqe->ipqe_ip->ip_src; 794 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 795 p = NULL; 796 goto insert; 797 } 798 799 /* 800 * Find a segment which begins after this one does. 801 */ 802 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 803 p = q, q = q->ipqe_q.le_next) 804 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 805 break; 806 807 /* 808 * If there is a preceding segment, it may provide some of 809 * our data already. If so, drop the data from the incoming 810 * segment. If it provides all of our data, drop us. 811 */ 812 if (p != NULL) { 813 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 814 ipqe->ipqe_ip->ip_off; 815 if (i > 0) { 816 if (i >= ipqe->ipqe_ip->ip_len) 817 goto dropfrag; 818 m_adj(ipqe->ipqe_m, i); 819 ipqe->ipqe_ip->ip_off += i; 820 ipqe->ipqe_ip->ip_len -= i; 821 } 822 } 823 824 /* 825 * While we overlap succeeding segments trim them or, 826 * if they are completely covered, dequeue them. 827 */ 828 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 829 q->ipqe_ip->ip_off; q = nq) { 830 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 831 q->ipqe_ip->ip_off; 832 if (i < q->ipqe_ip->ip_len) { 833 q->ipqe_ip->ip_len -= i; 834 q->ipqe_ip->ip_off += i; 835 m_adj(q->ipqe_m, i); 836 break; 837 } 838 nq = q->ipqe_q.le_next; 839 m_freem(q->ipqe_m); 840 LIST_REMOVE(q, ipqe_q); 841 pool_put(&ipqent_pool, q); 842 } 843 844 insert: 845 /* 846 * Stick new segment in its place; 847 * check for complete reassembly. 848 */ 849 if (p == NULL) { 850 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 851 } else { 852 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 853 } 854 next = 0; 855 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 856 p = q, q = q->ipqe_q.le_next) { 857 if (q->ipqe_ip->ip_off != next) 858 return (0); 859 next += q->ipqe_ip->ip_len; 860 } 861 if (p->ipqe_mff) 862 return (0); 863 864 /* 865 * Reassembly is complete. Check for a bogus message size and 866 * concatenate fragments. 867 */ 868 q = fp->ipq_fragq.lh_first; 869 ip = q->ipqe_ip; 870 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 871 ipstat.ips_toolong++; 872 ip_freef(fp); 873 return (0); 874 } 875 m = q->ipqe_m; 876 t = m->m_next; 877 m->m_next = 0; 878 m_cat(m, t); 879 nq = q->ipqe_q.le_next; 880 pool_put(&ipqent_pool, q); 881 for (q = nq; q != NULL; q = nq) { 882 t = q->ipqe_m; 883 nq = q->ipqe_q.le_next; 884 pool_put(&ipqent_pool, q); 885 m_cat(m, t); 886 } 887 888 /* 889 * Create header for new ip packet by 890 * modifying header of first packet; 891 * dequeue and discard fragment reassembly header. 892 * Make header visible. 893 */ 894 ip->ip_len = next; 895 ip->ip_src = fp->ipq_src; 896 ip->ip_dst = fp->ipq_dst; 897 LIST_REMOVE(fp, ipq_q); 898 FREE(fp, M_FTABLE); 899 ip_nfragpackets--; 900 m->m_len += (ip->ip_hl << 2); 901 m->m_data -= (ip->ip_hl << 2); 902 /* some debugging cruft by sklower, below, will go away soon */ 903 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 904 int plen = 0; 905 for (t = m; t; t = t->m_next) 906 plen += t->m_len; 907 m->m_pkthdr.len = plen; 908 } 909 return (m); 910 911 dropfrag: 912 ipstat.ips_fragdropped++; 913 m_freem(m); 914 pool_put(&ipqent_pool, ipqe); 915 return (0); 916 } 917 918 /* 919 * Free a fragment reassembly header and all 920 * associated datagrams. 921 */ 922 void 923 ip_freef(fp) 924 struct ipq *fp; 925 { 926 struct ipqent *q, *p; 927 928 IPQ_LOCK_CHECK(); 929 930 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 931 p = q->ipqe_q.le_next; 932 m_freem(q->ipqe_m); 933 LIST_REMOVE(q, ipqe_q); 934 pool_put(&ipqent_pool, q); 935 } 936 LIST_REMOVE(fp, ipq_q); 937 FREE(fp, M_FTABLE); 938 ip_nfragpackets--; 939 } 940 941 /* 942 * IP timer processing; 943 * if a timer expires on a reassembly 944 * queue, discard it. 945 */ 946 void 947 ip_slowtimo() 948 { 949 struct ipq *fp, *nfp; 950 int s = splsoftnet(); 951 952 IPQ_LOCK(); 953 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 954 nfp = fp->ipq_q.le_next; 955 if (--fp->ipq_ttl == 0) { 956 ipstat.ips_fragtimeout++; 957 ip_freef(fp); 958 } 959 } 960 /* 961 * If we are over the maximum number of fragments 962 * (due to the limit being lowered), drain off 963 * enough to get down to the new limit. 964 */ 965 if (ip_maxfragpackets < 0) 966 ; 967 else { 968 while (ip_nfragpackets > ip_maxfragpackets && ipq.lh_first) 969 ip_freef(ipq.lh_first); 970 } 971 IPQ_UNLOCK(); 972 #ifdef GATEWAY 973 ipflow_slowtimo(); 974 #endif 975 splx(s); 976 } 977 978 /* 979 * Drain off all datagram fragments. 980 */ 981 void 982 ip_drain() 983 { 984 985 /* 986 * We may be called from a device's interrupt context. If 987 * the ipq is already busy, just bail out now. 988 */ 989 if (ipq_lock_try() == 0) 990 return; 991 992 while (ipq.lh_first != NULL) { 993 ipstat.ips_fragdropped++; 994 ip_freef(ipq.lh_first); 995 } 996 997 IPQ_UNLOCK(); 998 } 999 1000 /* 1001 * Do option processing on a datagram, 1002 * possibly discarding it if bad options are encountered, 1003 * or forwarding it if source-routed. 1004 * Returns 1 if packet has been forwarded/freed, 1005 * 0 if the packet should be processed further. 1006 */ 1007 int 1008 ip_dooptions(m) 1009 struct mbuf *m; 1010 { 1011 struct ip *ip = mtod(m, struct ip *); 1012 u_char *cp, *cp0; 1013 struct ip_timestamp *ipt; 1014 struct in_ifaddr *ia; 1015 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 1016 struct in_addr dst; 1017 n_time ntime; 1018 1019 dst = ip->ip_dst; 1020 cp = (u_char *)(ip + 1); 1021 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 1022 for (; cnt > 0; cnt -= optlen, cp += optlen) { 1023 opt = cp[IPOPT_OPTVAL]; 1024 if (opt == IPOPT_EOL) 1025 break; 1026 if (opt == IPOPT_NOP) 1027 optlen = 1; 1028 else { 1029 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 1030 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1031 goto bad; 1032 } 1033 optlen = cp[IPOPT_OLEN]; 1034 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 1035 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1036 goto bad; 1037 } 1038 } 1039 switch (opt) { 1040 1041 default: 1042 break; 1043 1044 /* 1045 * Source routing with record. 1046 * Find interface with current destination address. 1047 * If none on this machine then drop if strictly routed, 1048 * or do nothing if loosely routed. 1049 * Record interface address and bring up next address 1050 * component. If strictly routed make sure next 1051 * address is on directly accessible net. 1052 */ 1053 case IPOPT_LSRR: 1054 case IPOPT_SSRR: 1055 if (ip_allowsrcrt == 0) { 1056 type = ICMP_UNREACH; 1057 code = ICMP_UNREACH_NET_PROHIB; 1058 goto bad; 1059 } 1060 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1061 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1062 goto bad; 1063 } 1064 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1065 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1066 goto bad; 1067 } 1068 ipaddr.sin_addr = ip->ip_dst; 1069 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1070 if (ia == 0) { 1071 if (opt == IPOPT_SSRR) { 1072 type = ICMP_UNREACH; 1073 code = ICMP_UNREACH_SRCFAIL; 1074 goto bad; 1075 } 1076 /* 1077 * Loose routing, and not at next destination 1078 * yet; nothing to do except forward. 1079 */ 1080 break; 1081 } 1082 off--; /* 0 origin */ 1083 if ((off + sizeof(struct in_addr)) > optlen) { 1084 /* 1085 * End of source route. Should be for us. 1086 */ 1087 save_rte(cp, ip->ip_src); 1088 break; 1089 } 1090 /* 1091 * locate outgoing interface 1092 */ 1093 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 1094 sizeof(ipaddr.sin_addr)); 1095 if (opt == IPOPT_SSRR) 1096 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1097 else 1098 ia = ip_rtaddr(ipaddr.sin_addr); 1099 if (ia == 0) { 1100 type = ICMP_UNREACH; 1101 code = ICMP_UNREACH_SRCFAIL; 1102 goto bad; 1103 } 1104 ip->ip_dst = ipaddr.sin_addr; 1105 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1106 (caddr_t)(cp + off), sizeof(struct in_addr)); 1107 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1108 /* 1109 * Let ip_intr's mcast routing check handle mcast pkts 1110 */ 1111 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1112 break; 1113 1114 case IPOPT_RR: 1115 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1116 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1117 goto bad; 1118 } 1119 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1120 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1121 goto bad; 1122 } 1123 /* 1124 * If no space remains, ignore. 1125 */ 1126 off--; /* 0 origin */ 1127 if ((off + sizeof(struct in_addr)) > optlen) 1128 break; 1129 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 1130 sizeof(ipaddr.sin_addr)); 1131 /* 1132 * locate outgoing interface; if we're the destination, 1133 * use the incoming interface (should be same). 1134 */ 1135 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 1136 == NULL && 1137 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 1138 type = ICMP_UNREACH; 1139 code = ICMP_UNREACH_HOST; 1140 goto bad; 1141 } 1142 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1143 (caddr_t)(cp + off), sizeof(struct in_addr)); 1144 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1145 break; 1146 1147 case IPOPT_TS: 1148 code = cp - (u_char *)ip; 1149 ipt = (struct ip_timestamp *)cp; 1150 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1151 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1152 goto bad; 1153 } 1154 if (ipt->ipt_ptr < 5) { 1155 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1156 goto bad; 1157 } 1158 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1159 if (++ipt->ipt_oflw == 0) { 1160 code = (u_char *)&ipt->ipt_ptr - 1161 (u_char *)ip; 1162 goto bad; 1163 } 1164 break; 1165 } 1166 cp0 = (cp + ipt->ipt_ptr - 1); 1167 switch (ipt->ipt_flg) { 1168 1169 case IPOPT_TS_TSONLY: 1170 break; 1171 1172 case IPOPT_TS_TSANDADDR: 1173 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1174 sizeof(struct in_addr) > ipt->ipt_len) { 1175 code = (u_char *)&ipt->ipt_ptr - 1176 (u_char *)ip; 1177 goto bad; 1178 } 1179 ipaddr.sin_addr = dst; 1180 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1181 m->m_pkthdr.rcvif)); 1182 if (ia == 0) 1183 continue; 1184 bcopy(&ia->ia_addr.sin_addr, 1185 cp0, sizeof(struct in_addr)); 1186 ipt->ipt_ptr += sizeof(struct in_addr); 1187 break; 1188 1189 case IPOPT_TS_PRESPEC: 1190 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1191 sizeof(struct in_addr) > ipt->ipt_len) { 1192 code = (u_char *)&ipt->ipt_ptr - 1193 (u_char *)ip; 1194 goto bad; 1195 } 1196 bcopy(cp0, &ipaddr.sin_addr, 1197 sizeof(struct in_addr)); 1198 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1199 == NULL) 1200 continue; 1201 ipt->ipt_ptr += sizeof(struct in_addr); 1202 break; 1203 1204 default: 1205 /* XXX can't take &ipt->ipt_flg */ 1206 code = (u_char *)&ipt->ipt_ptr - 1207 (u_char *)ip + 1; 1208 goto bad; 1209 } 1210 ntime = iptime(); 1211 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1212 bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1, 1213 sizeof(n_time)); 1214 ipt->ipt_ptr += sizeof(n_time); 1215 } 1216 } 1217 if (forward) { 1218 if (ip_forwsrcrt == 0) { 1219 type = ICMP_UNREACH; 1220 code = ICMP_UNREACH_SRCFAIL; 1221 goto bad; 1222 } 1223 ip_forward(m, 1); 1224 return (1); 1225 } 1226 return (0); 1227 bad: 1228 icmp_error(m, type, code, 0, 0); 1229 ipstat.ips_badoptions++; 1230 return (1); 1231 } 1232 1233 /* 1234 * Given address of next destination (final or next hop), 1235 * return internet address info of interface to be used to get there. 1236 */ 1237 struct in_ifaddr * 1238 ip_rtaddr(dst) 1239 struct in_addr dst; 1240 { 1241 struct sockaddr_in *sin; 1242 1243 sin = satosin(&ipforward_rt.ro_dst); 1244 1245 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 1246 if (ipforward_rt.ro_rt) { 1247 RTFREE(ipforward_rt.ro_rt); 1248 ipforward_rt.ro_rt = 0; 1249 } 1250 sin->sin_family = AF_INET; 1251 sin->sin_len = sizeof(*sin); 1252 sin->sin_addr = dst; 1253 1254 rtalloc(&ipforward_rt); 1255 } 1256 if (ipforward_rt.ro_rt == 0) 1257 return ((struct in_ifaddr *)0); 1258 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 1259 } 1260 1261 /* 1262 * Save incoming source route for use in replies, 1263 * to be picked up later by ip_srcroute if the receiver is interested. 1264 */ 1265 void 1266 save_rte(option, dst) 1267 u_char *option; 1268 struct in_addr dst; 1269 { 1270 unsigned olen; 1271 1272 olen = option[IPOPT_OLEN]; 1273 #ifdef DIAGNOSTIC 1274 if (ipprintfs) 1275 printf("save_rte: olen %d\n", olen); 1276 #endif /* 0 */ 1277 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1278 return; 1279 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 1280 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1281 ip_srcrt.dst = dst; 1282 } 1283 1284 /* 1285 * Retrieve incoming source route for use in replies, 1286 * in the same form used by setsockopt. 1287 * The first hop is placed before the options, will be removed later. 1288 */ 1289 struct mbuf * 1290 ip_srcroute() 1291 { 1292 struct in_addr *p, *q; 1293 struct mbuf *m; 1294 1295 if (ip_nhops == 0) 1296 return ((struct mbuf *)0); 1297 m = m_get(M_DONTWAIT, MT_SOOPTS); 1298 if (m == 0) 1299 return ((struct mbuf *)0); 1300 1301 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1302 1303 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1304 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1305 OPTSIZ; 1306 #ifdef DIAGNOSTIC 1307 if (ipprintfs) 1308 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1309 #endif 1310 1311 /* 1312 * First save first hop for return route 1313 */ 1314 p = &ip_srcrt.route[ip_nhops - 1]; 1315 *(mtod(m, struct in_addr *)) = *p--; 1316 #ifdef DIAGNOSTIC 1317 if (ipprintfs) 1318 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1319 #endif 1320 1321 /* 1322 * Copy option fields and padding (nop) to mbuf. 1323 */ 1324 ip_srcrt.nop = IPOPT_NOP; 1325 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1326 bcopy((caddr_t)&ip_srcrt.nop, 1327 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 1328 q = (struct in_addr *)(mtod(m, caddr_t) + 1329 sizeof(struct in_addr) + OPTSIZ); 1330 #undef OPTSIZ 1331 /* 1332 * Record return path as an IP source route, 1333 * reversing the path (pointers are now aligned). 1334 */ 1335 while (p >= ip_srcrt.route) { 1336 #ifdef DIAGNOSTIC 1337 if (ipprintfs) 1338 printf(" %x", ntohl(q->s_addr)); 1339 #endif 1340 *q++ = *p--; 1341 } 1342 /* 1343 * Last hop goes to final destination. 1344 */ 1345 *q = ip_srcrt.dst; 1346 #ifdef DIAGNOSTIC 1347 if (ipprintfs) 1348 printf(" %x\n", ntohl(q->s_addr)); 1349 #endif 1350 return (m); 1351 } 1352 1353 /* 1354 * Strip out IP options, at higher 1355 * level protocol in the kernel. 1356 * Second argument is buffer to which options 1357 * will be moved, and return value is their length. 1358 * XXX should be deleted; last arg currently ignored. 1359 */ 1360 void 1361 ip_stripoptions(m, mopt) 1362 struct mbuf *m; 1363 struct mbuf *mopt; 1364 { 1365 int i; 1366 struct ip *ip = mtod(m, struct ip *); 1367 caddr_t opts; 1368 int olen; 1369 1370 olen = (ip->ip_hl << 2) - sizeof (struct ip); 1371 opts = (caddr_t)(ip + 1); 1372 i = m->m_len - (sizeof (struct ip) + olen); 1373 bcopy(opts + olen, opts, (unsigned)i); 1374 m->m_len -= olen; 1375 if (m->m_flags & M_PKTHDR) 1376 m->m_pkthdr.len -= olen; 1377 ip->ip_len -= olen; 1378 ip->ip_hl = sizeof (struct ip) >> 2; 1379 } 1380 1381 int inetctlerrmap[PRC_NCMDS] = { 1382 0, 0, 0, 0, 1383 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1384 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1385 EMSGSIZE, EHOSTUNREACH, 0, 0, 1386 0, 0, 0, 0, 1387 ENOPROTOOPT 1388 }; 1389 1390 /* 1391 * Forward a packet. If some error occurs return the sender 1392 * an icmp packet. Note we can't always generate a meaningful 1393 * icmp message because icmp doesn't have a large enough repertoire 1394 * of codes and types. 1395 * 1396 * If not forwarding, just drop the packet. This could be confusing 1397 * if ipforwarding was zero but some routing protocol was advancing 1398 * us as a gateway to somewhere. However, we must let the routing 1399 * protocol deal with that. 1400 * 1401 * The srcrt parameter indicates whether the packet is being forwarded 1402 * via a source route. 1403 */ 1404 void 1405 ip_forward(m, srcrt) 1406 struct mbuf *m; 1407 int srcrt; 1408 { 1409 struct ip *ip = mtod(m, struct ip *); 1410 struct sockaddr_in *sin; 1411 struct rtentry *rt; 1412 int error, type = 0, code = 0; 1413 struct mbuf *mcopy; 1414 n_long dest; 1415 struct ifnet *destifp; 1416 #ifdef IPSEC 1417 struct ifnet dummyifp; 1418 #endif 1419 1420 dest = 0; 1421 #ifdef DIAGNOSTIC 1422 if (ipprintfs) 1423 printf("forward: src %2.2x dst %2.2x ttl %x\n", 1424 ntohl(ip->ip_src.s_addr), 1425 ntohl(ip->ip_dst.s_addr), ip->ip_ttl); 1426 #endif 1427 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1428 ipstat.ips_cantforward++; 1429 m_freem(m); 1430 return; 1431 } 1432 if (ip->ip_ttl <= IPTTLDEC) { 1433 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1434 return; 1435 } 1436 ip->ip_ttl -= IPTTLDEC; 1437 1438 sin = satosin(&ipforward_rt.ro_dst); 1439 if ((rt = ipforward_rt.ro_rt) == 0 || 1440 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1441 if (ipforward_rt.ro_rt) { 1442 RTFREE(ipforward_rt.ro_rt); 1443 ipforward_rt.ro_rt = 0; 1444 } 1445 sin->sin_family = AF_INET; 1446 sin->sin_len = sizeof(struct sockaddr_in); 1447 sin->sin_addr = ip->ip_dst; 1448 1449 rtalloc(&ipforward_rt); 1450 if (ipforward_rt.ro_rt == 0) { 1451 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1452 return; 1453 } 1454 rt = ipforward_rt.ro_rt; 1455 } 1456 1457 /* 1458 * Save at most 68 bytes of the packet in case 1459 * we need to generate an ICMP message to the src. 1460 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1461 */ 1462 mcopy = m_copym(m, 0, imin((int)ip->ip_len, 68), M_DONTWAIT); 1463 if (mcopy) 1464 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1465 1466 /* 1467 * If forwarding packet using same interface that it came in on, 1468 * perhaps should send a redirect to sender to shortcut a hop. 1469 * Only send redirect if source is sending directly to us, 1470 * and if packet was not source routed (or has any options). 1471 * Also, don't send redirect if forwarding using a default route 1472 * or a route modified by a redirect. 1473 */ 1474 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1475 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1476 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1477 ipsendredirects && !srcrt) { 1478 if (rt->rt_ifa && 1479 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1480 ifatoia(rt->rt_ifa)->ia_subnet) { 1481 if (rt->rt_flags & RTF_GATEWAY) 1482 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1483 else 1484 dest = ip->ip_dst.s_addr; 1485 /* 1486 * Router requirements says to only send host 1487 * redirects. 1488 */ 1489 type = ICMP_REDIRECT; 1490 code = ICMP_REDIRECT_HOST; 1491 #ifdef DIAGNOSTIC 1492 if (ipprintfs) 1493 printf("redirect (%d) to %x\n", code, 1494 (u_int32_t)dest); 1495 #endif 1496 } 1497 } 1498 1499 #ifdef IPSEC 1500 /* Don't lookup socket in forwading case */ 1501 (void)ipsec_setsocket(m, NULL); 1502 #endif 1503 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1504 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1505 if (error) 1506 ipstat.ips_cantforward++; 1507 else { 1508 ipstat.ips_forward++; 1509 if (type) 1510 ipstat.ips_redirectsent++; 1511 else { 1512 if (mcopy) { 1513 #ifdef GATEWAY 1514 if (mcopy->m_flags & M_CANFASTFWD) 1515 ipflow_create(&ipforward_rt, mcopy); 1516 #endif 1517 m_freem(mcopy); 1518 } 1519 return; 1520 } 1521 } 1522 if (mcopy == NULL) 1523 return; 1524 destifp = NULL; 1525 1526 switch (error) { 1527 1528 case 0: /* forwarded, but need redirect */ 1529 /* type, code set above */ 1530 break; 1531 1532 case ENETUNREACH: /* shouldn't happen, checked above */ 1533 case EHOSTUNREACH: 1534 case ENETDOWN: 1535 case EHOSTDOWN: 1536 default: 1537 type = ICMP_UNREACH; 1538 code = ICMP_UNREACH_HOST; 1539 break; 1540 1541 case EMSGSIZE: 1542 type = ICMP_UNREACH; 1543 code = ICMP_UNREACH_NEEDFRAG; 1544 #ifndef IPSEC 1545 if (ipforward_rt.ro_rt) 1546 destifp = ipforward_rt.ro_rt->rt_ifp; 1547 #else 1548 /* 1549 * If the packet is routed over IPsec tunnel, tell the 1550 * originator the tunnel MTU. 1551 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz 1552 * XXX quickhack!!! 1553 */ 1554 if (ipforward_rt.ro_rt) { 1555 struct secpolicy *sp; 1556 int ipsecerror; 1557 size_t ipsechdr; 1558 struct route *ro; 1559 1560 sp = ipsec4_getpolicybyaddr(mcopy, 1561 IPSEC_DIR_OUTBOUND, 1562 IP_FORWARDING, 1563 &ipsecerror); 1564 1565 if (sp == NULL) 1566 destifp = ipforward_rt.ro_rt->rt_ifp; 1567 else { 1568 /* count IPsec header size */ 1569 ipsechdr = ipsec4_hdrsiz(mcopy, 1570 IPSEC_DIR_OUTBOUND, 1571 NULL); 1572 1573 /* 1574 * find the correct route for outer IPv4 1575 * header, compute tunnel MTU. 1576 * 1577 * XXX BUG ALERT 1578 * The "dummyifp" code relies upon the fact 1579 * that icmp_error() touches only ifp->if_mtu. 1580 */ 1581 /*XXX*/ 1582 destifp = NULL; 1583 if (sp->req != NULL 1584 && sp->req->sav != NULL 1585 && sp->req->sav->sah != NULL) { 1586 ro = &sp->req->sav->sah->sa_route; 1587 if (ro->ro_rt && ro->ro_rt->rt_ifp) { 1588 dummyifp.if_mtu = 1589 ro->ro_rt->rt_ifp->if_mtu; 1590 dummyifp.if_mtu -= ipsechdr; 1591 destifp = &dummyifp; 1592 } 1593 } 1594 1595 key_freesp(sp); 1596 } 1597 } 1598 #endif /*IPSEC*/ 1599 ipstat.ips_cantfrag++; 1600 break; 1601 1602 case ENOBUFS: 1603 type = ICMP_SOURCEQUENCH; 1604 code = 0; 1605 break; 1606 } 1607 icmp_error(mcopy, type, code, dest, destifp); 1608 } 1609 1610 void 1611 ip_savecontrol(inp, mp, ip, m) 1612 struct inpcb *inp; 1613 struct mbuf **mp; 1614 struct ip *ip; 1615 struct mbuf *m; 1616 { 1617 1618 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1619 struct timeval tv; 1620 1621 microtime(&tv); 1622 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1623 SCM_TIMESTAMP, SOL_SOCKET); 1624 if (*mp) 1625 mp = &(*mp)->m_next; 1626 } 1627 if (inp->inp_flags & INP_RECVDSTADDR) { 1628 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1629 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1630 if (*mp) 1631 mp = &(*mp)->m_next; 1632 } 1633 #ifdef notyet 1634 /* 1635 * XXX 1636 * Moving these out of udp_input() made them even more broken 1637 * than they already were. 1638 * - fenner (at) parc.xerox.com 1639 */ 1640 /* options were tossed already */ 1641 if (inp->inp_flags & INP_RECVOPTS) { 1642 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1643 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1644 if (*mp) 1645 mp = &(*mp)->m_next; 1646 } 1647 /* ip_srcroute doesn't do what we want here, need to fix */ 1648 if (inp->inp_flags & INP_RECVRETOPTS) { 1649 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1650 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1651 if (*mp) 1652 mp = &(*mp)->m_next; 1653 } 1654 #endif 1655 if (inp->inp_flags & INP_RECVIF) { 1656 struct sockaddr_dl sdl; 1657 1658 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1659 sdl.sdl_family = AF_LINK; 1660 sdl.sdl_index = m->m_pkthdr.rcvif ? 1661 m->m_pkthdr.rcvif->if_index : 0; 1662 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1663 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1664 IP_RECVIF, IPPROTO_IP); 1665 if (*mp) 1666 mp = &(*mp)->m_next; 1667 } 1668 } 1669 1670 int 1671 ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1672 int *name; 1673 u_int namelen; 1674 void *oldp; 1675 size_t *oldlenp; 1676 void *newp; 1677 size_t newlen; 1678 { 1679 extern int subnetsarelocal, hostzeroisbroadcast; 1680 1681 int error, old; 1682 1683 /* All sysctl names at this level are terminal. */ 1684 if (namelen != 1) 1685 return (ENOTDIR); 1686 1687 switch (name[0]) { 1688 case IPCTL_FORWARDING: 1689 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1690 case IPCTL_SENDREDIRECTS: 1691 return (sysctl_int(oldp, oldlenp, newp, newlen, 1692 &ipsendredirects)); 1693 case IPCTL_DEFTTL: 1694 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1695 #ifdef notyet 1696 case IPCTL_DEFMTU: 1697 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1698 #endif 1699 case IPCTL_FORWSRCRT: 1700 /* Don't allow this to change in a secure environment. */ 1701 if (securelevel > 0) 1702 return (sysctl_rdint(oldp, oldlenp, newp, 1703 ip_forwsrcrt)); 1704 else 1705 return (sysctl_int(oldp, oldlenp, newp, newlen, 1706 &ip_forwsrcrt)); 1707 case IPCTL_DIRECTEDBCAST: 1708 return (sysctl_int(oldp, oldlenp, newp, newlen, 1709 &ip_directedbcast)); 1710 case IPCTL_ALLOWSRCRT: 1711 return (sysctl_int(oldp, oldlenp, newp, newlen, 1712 &ip_allowsrcrt)); 1713 case IPCTL_SUBNETSARELOCAL: 1714 return (sysctl_int(oldp, oldlenp, newp, newlen, 1715 &subnetsarelocal)); 1716 case IPCTL_MTUDISC: 1717 error = sysctl_int(oldp, oldlenp, newp, newlen, 1718 &ip_mtudisc); 1719 if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) { 1720 ip_mtudisc_timeout_q = 1721 rt_timer_queue_create(ip_mtudisc_timeout); 1722 } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) { 1723 rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE); 1724 ip_mtudisc_timeout_q = NULL; 1725 } 1726 return error; 1727 case IPCTL_ANONPORTMIN: 1728 old = anonportmin; 1729 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1730 if (anonportmin >= anonportmax || anonportmin < 0 1731 || anonportmin > 65535 1732 #ifndef IPNOPRIVPORTS 1733 || anonportmin < IPPORT_RESERVED 1734 #endif 1735 ) { 1736 anonportmin = old; 1737 return (EINVAL); 1738 } 1739 return (error); 1740 case IPCTL_ANONPORTMAX: 1741 old = anonportmax; 1742 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1743 if (anonportmin >= anonportmax || anonportmax < 0 1744 || anonportmax > 65535 1745 #ifndef IPNOPRIVPORTS 1746 || anonportmax < IPPORT_RESERVED 1747 #endif 1748 ) { 1749 anonportmax = old; 1750 return (EINVAL); 1751 } 1752 return (error); 1753 case IPCTL_MTUDISCTIMEOUT: 1754 error = sysctl_int(oldp, oldlenp, newp, newlen, 1755 &ip_mtudisc_timeout); 1756 if (ip_mtudisc_timeout_q != NULL) 1757 rt_timer_queue_change(ip_mtudisc_timeout_q, 1758 ip_mtudisc_timeout); 1759 return (error); 1760 #ifdef GATEWAY 1761 case IPCTL_MAXFLOWS: 1762 { 1763 int s; 1764 1765 error = sysctl_int(oldp, oldlenp, newp, newlen, 1766 &ip_maxflows); 1767 s = splsoftnet(); 1768 ipflow_reap(0); 1769 splx(s); 1770 return (error); 1771 } 1772 #endif 1773 case IPCTL_HOSTZEROBROADCAST: 1774 return (sysctl_int(oldp, oldlenp, newp, newlen, 1775 &hostzeroisbroadcast)); 1776 #if NGIF > 0 1777 case IPCTL_GIF_TTL: 1778 return(sysctl_int(oldp, oldlenp, newp, newlen, 1779 &ip_gif_ttl)); 1780 #endif 1781 1782 #if NGRE > 0 1783 case IPCTL_GRE_TTL: 1784 return(sysctl_int(oldp, oldlenp, newp, newlen, 1785 &ip_gre_ttl)); 1786 #endif 1787 1788 #ifndef IPNOPRIVPORTS 1789 case IPCTL_LOWPORTMIN: 1790 old = lowportmin; 1791 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin); 1792 if (lowportmin >= lowportmax 1793 || lowportmin > IPPORT_RESERVEDMAX 1794 || lowportmin < IPPORT_RESERVEDMIN 1795 ) { 1796 lowportmin = old; 1797 return (EINVAL); 1798 } 1799 return (error); 1800 case IPCTL_LOWPORTMAX: 1801 old = lowportmax; 1802 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax); 1803 if (lowportmin >= lowportmax 1804 || lowportmax > IPPORT_RESERVEDMAX 1805 || lowportmax < IPPORT_RESERVEDMIN 1806 ) { 1807 lowportmax = old; 1808 return (EINVAL); 1809 } 1810 return (error); 1811 #endif 1812 1813 case IPCTL_MAXFRAGPACKETS: 1814 return (sysctl_int(oldp, oldlenp, newp, newlen, 1815 &ip_maxfragpackets)); 1816 1817 default: 1818 return (EOPNOTSUPP); 1819 } 1820 /* NOTREACHED */ 1821 } 1822
Indexes created Wed Oct 15 16:09:53 GMT 2025