ip_input.c revision 1.114.4.9
1 /* $NetBSD: ip_input.c,v 1.114.4.9 2002/11/10 19:58:05 itojun Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /*- 33 * Copyright (c) 1998 The NetBSD Foundation, Inc. 34 * All rights reserved. 35 * 36 * This code is derived from software contributed to The NetBSD Foundation 37 * by Public Access Networks Corporation ("Panix"). It was developed under 38 * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon. 39 * 40 * Redistribution and use in source and binary forms, with or without 41 * modification, are permitted provided that the following conditions 42 * are met: 43 * 1. Redistributions of source code must retain the above copyright 44 * notice, this list of conditions and the following disclaimer. 45 * 2. Redistributions in binary form must reproduce the above copyright 46 * notice, this list of conditions and the following disclaimer in the 47 * documentation and/or other materials provided with the distribution. 48 * 3. All advertising materials mentioning features or use of this software 49 * must display the following acknowledgement: 50 * This product includes software developed by the NetBSD 51 * Foundation, Inc. and its contributors. 52 * 4. Neither the name of The NetBSD Foundation nor the names of its 53 * contributors may be used to endorse or promote products derived 54 * from this software without specific prior written permission. 55 * 56 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 57 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 58 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 59 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 60 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 61 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 62 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 63 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 64 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 65 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 66 * POSSIBILITY OF SUCH DAMAGE. 67 */ 68 69 /* 70 * Copyright (c) 1982, 1986, 1988, 1993 71 * The Regents of the University of California. All rights reserved. 72 * 73 * Redistribution and use in source and binary forms, with or without 74 * modification, are permitted provided that the following conditions 75 * are met: 76 * 1. Redistributions of source code must retain the above copyright 77 * notice, this list of conditions and the following disclaimer. 78 * 2. Redistributions in binary form must reproduce the above copyright 79 * notice, this list of conditions and the following disclaimer in the 80 * documentation and/or other materials provided with the distribution. 81 * 3. 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 ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); 314 #ifdef GATEWAY 315 ipflow_init(); 316 #endif 317 } 318 319 struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; 320 struct route ipforward_rt; 321 322 /* 323 * IP software interrupt routine 324 */ 325 void 326 ipintr() 327 { 328 int s; 329 struct mbuf *m; 330 331 while (1) { 332 s = splimp(); 333 IF_DEQUEUE(&ipintrq, m); 334 splx(s); 335 if (m == 0) 336 return; 337 ip_input(m); 338 } 339 } 340 341 /* 342 * Ip input routine. Checksum and byte swap header. If fragmented 343 * try to reassemble. Process options. Pass to next level. 344 */ 345 void 346 ip_input(struct mbuf *m) 347 { 348 struct ip *ip = NULL; 349 struct ipq *fp; 350 struct in_ifaddr *ia; 351 struct ifaddr *ifa; 352 struct ipqent *ipqe; 353 int hlen = 0, mff, len; 354 int downmatch; 355 #ifdef PFIL_HOOKS 356 struct packet_filter_hook *pfh; 357 struct mbuf *m0; 358 int rv; 359 #endif /* PFIL_HOOKS */ 360 361 #ifdef DIAGNOSTIC 362 if ((m->m_flags & M_PKTHDR) == 0) 363 panic("ipintr no HDR"); 364 #endif 365 #ifdef IPSEC 366 /* 367 * should the inner packet be considered authentic? 368 * see comment in ah4_input(). 369 */ 370 if (m) { 371 m->m_flags &= ~M_AUTHIPHDR; 372 m->m_flags &= ~M_AUTHIPDGM; 373 } 374 #endif 375 /* 376 * If no IP addresses have been set yet but the interfaces 377 * are receiving, can't do anything with incoming packets yet. 378 */ 379 if (in_ifaddr.tqh_first == 0) 380 goto bad; 381 ipstat.ips_total++; 382 if (m->m_len < sizeof (struct ip) && 383 (m = m_pullup(m, sizeof (struct ip))) == 0) { 384 ipstat.ips_toosmall++; 385 return; 386 } 387 ip = mtod(m, struct ip *); 388 if (ip->ip_v != IPVERSION) { 389 ipstat.ips_badvers++; 390 goto bad; 391 } 392 hlen = ip->ip_hl << 2; 393 if (hlen < sizeof(struct ip)) { /* minimum header length */ 394 ipstat.ips_badhlen++; 395 goto bad; 396 } 397 if (hlen > m->m_len) { 398 if ((m = m_pullup(m, hlen)) == 0) { 399 ipstat.ips_badhlen++; 400 return; 401 } 402 ip = mtod(m, struct ip *); 403 } 404 405 /* 406 * RFC1122: packets with a multicast source address are 407 * not allowed. 408 */ 409 if (IN_MULTICAST(ip->ip_src.s_addr)) { 410 /* XXX stat */ 411 goto bad; 412 } 413 414 if (in_cksum(m, hlen) != 0) { 415 ipstat.ips_badsum++; 416 goto bad; 417 } 418 419 /* 420 * Convert fields to host representation. 421 */ 422 NTOHS(ip->ip_len); 423 NTOHS(ip->ip_off); 424 len = ip->ip_len; 425 426 /* 427 * Check for additional length bogosity 428 */ 429 if (len < hlen) { 430 ipstat.ips_badlen++; 431 goto bad; 432 } 433 434 /* 435 * Check that the amount of data in the buffers 436 * is as at least much as the IP header would have us expect. 437 * Trim mbufs if longer than we expect. 438 * Drop packet if shorter than we expect. 439 */ 440 if (m->m_pkthdr.len < len) { 441 ipstat.ips_tooshort++; 442 goto bad; 443 } 444 if (m->m_pkthdr.len > len) { 445 if (m->m_len == m->m_pkthdr.len) { 446 m->m_len = len; 447 m->m_pkthdr.len = len; 448 } else 449 m_adj(m, len - m->m_pkthdr.len); 450 } 451 452 #ifdef IPSEC 453 /* ipflow (IP fast fowarding) is not compatible with IPsec. */ 454 m->m_flags &= ~M_CANFASTFWD; 455 #else 456 /* 457 * Assume that we can create a fast-forward IP flow entry 458 * based on this packet. 459 */ 460 m->m_flags |= M_CANFASTFWD; 461 #endif 462 463 #ifdef PFIL_HOOKS 464 #ifdef IPSEC 465 /* 466 * let ipfilter look at packet on the wire, 467 * not the decapsulated packet. 468 */ 469 if (ipsec_gethist(m, NULL)) 470 goto nofilt; 471 #endif 472 /* 473 * Run through list of hooks for input packets. If there are any 474 * filters which require that additional packets in the flow are 475 * not fast-forwarded, they must clear the M_CANFASTFWD flag. 476 * Note that filters must _never_ set this flag, as another filter 477 * in the list may have previously cleared it. 478 */ 479 m0 = m; 480 pfh = pfil_hook_get(PFIL_IN, &inetsw[ip_protox[IPPROTO_IP]].pr_pfh); 481 for (; pfh; pfh = pfh->pfil_link.tqe_next) 482 if (pfh->pfil_func) { 483 rv = pfh->pfil_func(ip, hlen, 484 m->m_pkthdr.rcvif, 0, &m0); 485 if (rv) 486 return; 487 m = m0; 488 if (m == NULL) 489 return; 490 ip = mtod(m, struct ip *); 491 } 492 #ifdef IPSEC 493 nofilt:; 494 #endif 495 #endif /* PFIL_HOOKS */ 496 497 /* 498 * Process options and, if not destined for us, 499 * ship it on. ip_dooptions returns 1 when an 500 * error was detected (causing an icmp message 501 * to be sent and the original packet to be freed). 502 */ 503 ip_nhops = 0; /* for source routed packets */ 504 if (hlen > sizeof (struct ip) && ip_dooptions(m)) 505 return; 506 507 /* 508 * Check our list of addresses, to see if the packet is for us. 509 * 510 * Traditional 4.4BSD did not consult IFF_UP at all. 511 * The behavior here is to treat addresses on !IFF_UP interface 512 * as not mine. 513 */ 514 downmatch = 0; 515 for (ia = IN_IFADDR_HASH(ip->ip_dst.s_addr).lh_first; 516 ia != NULL; 517 ia = ia->ia_hash.le_next) { 518 if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { 519 if ((ia->ia_ifp->if_flags & IFF_UP) != 0) 520 break; 521 else 522 downmatch++; 523 } 524 } 525 if (ia != NULL) 526 goto ours; 527 if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { 528 for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first; 529 ifa != NULL; ifa = ifa->ifa_list.tqe_next) { 530 if (ifa->ifa_addr->sa_family != AF_INET) continue; 531 ia = ifatoia(ifa); 532 if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || 533 in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || 534 /* 535 * Look for all-0's host part (old broadcast addr), 536 * either for subnet or net. 537 */ 538 ip->ip_dst.s_addr == ia->ia_subnet || 539 ip->ip_dst.s_addr == ia->ia_net) 540 goto ours; 541 /* 542 * An interface with IP address zero accepts 543 * all packets that arrive on that interface. 544 */ 545 if (in_nullhost(ia->ia_addr.sin_addr)) 546 goto ours; 547 } 548 } 549 if (IN_MULTICAST(ip->ip_dst.s_addr)) { 550 struct in_multi *inm; 551 #ifdef MROUTING 552 extern struct socket *ip_mrouter; 553 554 if (m->m_flags & M_EXT) { 555 if ((m = m_pullup(m, hlen)) == 0) { 556 ipstat.ips_toosmall++; 557 return; 558 } 559 ip = mtod(m, struct ip *); 560 } 561 562 if (ip_mrouter) { 563 /* 564 * If we are acting as a multicast router, all 565 * incoming multicast packets are passed to the 566 * kernel-level multicast forwarding function. 567 * The packet is returned (relatively) intact; if 568 * ip_mforward() returns a non-zero value, the packet 569 * must be discarded, else it may be accepted below. 570 * 571 * (The IP ident field is put in the same byte order 572 * as expected when ip_mforward() is called from 573 * ip_output().) 574 */ 575 if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { 576 ipstat.ips_cantforward++; 577 m_freem(m); 578 return; 579 } 580 581 /* 582 * The process-level routing demon needs to receive 583 * all multicast IGMP packets, whether or not this 584 * host belongs to their destination groups. 585 */ 586 if (ip->ip_p == IPPROTO_IGMP) 587 goto ours; 588 ipstat.ips_forward++; 589 } 590 #endif 591 /* 592 * See if we belong to the destination multicast group on the 593 * arrival interface. 594 */ 595 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); 596 if (inm == NULL) { 597 ipstat.ips_cantforward++; 598 m_freem(m); 599 return; 600 } 601 goto ours; 602 } 603 if (ip->ip_dst.s_addr == INADDR_BROADCAST || 604 in_nullhost(ip->ip_dst)) 605 goto ours; 606 607 /* 608 * Not for us; forward if possible and desirable. 609 */ 610 if (ipforwarding == 0) { 611 ipstat.ips_cantforward++; 612 m_freem(m); 613 } else { 614 /* 615 * If ip_dst matched any of my address on !IFF_UP interface, 616 * and there's no IFF_UP interface that matches ip_dst, 617 * send icmp unreach. Forwarding it will result in in-kernel 618 * forwarding loop till TTL goes to 0. 619 */ 620 if (downmatch) { 621 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); 622 ipstat.ips_cantforward++; 623 return; 624 } 625 #ifdef IPSEC 626 if (ipsec4_in_reject(m, NULL)) { 627 ipsecstat.in_polvio++; 628 goto bad; 629 } 630 #endif 631 632 ip_forward(m, 0); 633 } 634 return; 635 636 ours: 637 /* 638 * If offset or IP_MF are set, must reassemble. 639 * Otherwise, nothing need be done. 640 * (We could look in the reassembly queue to see 641 * if the packet was previously fragmented, 642 * but it's not worth the time; just let them time out.) 643 */ 644 if (ip->ip_off & ~(IP_DF|IP_RF)) { 645 /* 646 * Look for queue of fragments 647 * of this datagram. 648 */ 649 IPQ_LOCK(); 650 for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) 651 if (ip->ip_id == fp->ipq_id && 652 in_hosteq(ip->ip_src, fp->ipq_src) && 653 in_hosteq(ip->ip_dst, fp->ipq_dst) && 654 ip->ip_p == fp->ipq_p) 655 goto found; 656 fp = 0; 657 found: 658 659 /* 660 * Adjust ip_len to not reflect header, 661 * set ipqe_mff if more fragments are expected, 662 * convert offset of this to bytes. 663 */ 664 ip->ip_len -= hlen; 665 mff = (ip->ip_off & IP_MF) != 0; 666 if (mff) { 667 /* 668 * Make sure that fragments have a data length 669 * that's a non-zero multiple of 8 bytes. 670 */ 671 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { 672 ipstat.ips_badfrags++; 673 IPQ_UNLOCK(); 674 goto bad; 675 } 676 } 677 ip->ip_off <<= 3; 678 679 /* 680 * If datagram marked as having more fragments 681 * or if this is not the first fragment, 682 * attempt reassembly; if it succeeds, proceed. 683 */ 684 if (mff || ip->ip_off) { 685 ipstat.ips_fragments++; 686 ipqe = pool_get(&ipqent_pool, PR_NOWAIT); 687 if (ipqe == NULL) { 688 ipstat.ips_rcvmemdrop++; 689 IPQ_UNLOCK(); 690 goto bad; 691 } 692 ipqe->ipqe_mff = mff; 693 ipqe->ipqe_m = m; 694 ipqe->ipqe_ip = ip; 695 m = ip_reass(ipqe, fp); 696 if (m == 0) { 697 IPQ_UNLOCK(); 698 return; 699 } 700 ipstat.ips_reassembled++; 701 ip = mtod(m, struct ip *); 702 hlen = ip->ip_hl << 2; 703 ip->ip_len += hlen; 704 } else 705 if (fp) 706 ip_freef(fp); 707 IPQ_UNLOCK(); 708 } 709 710 #ifdef IPSEC 711 /* 712 * enforce IPsec policy checking if we are seeing last header. 713 * note that we do not visit this with protocols with pcb layer 714 * code - like udp/tcp/raw ip. 715 */ 716 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 && 717 ipsec4_in_reject(m, NULL)) { 718 ipsecstat.in_polvio++; 719 goto bad; 720 } 721 #endif 722 723 /* 724 * Switch out to protocol's input routine. 725 */ 726 #if IFA_STATS 727 ia->ia_ifa.ifa_data.ifad_inbytes += ip->ip_len; 728 #endif 729 ipstat.ips_delivered++; 730 { 731 int off = hlen, nh = ip->ip_p; 732 733 (*inetsw[ip_protox[nh]].pr_input)(m, off, nh); 734 return; 735 } 736 bad: 737 m_freem(m); 738 } 739 740 /* 741 * Take incoming datagram fragment and try to 742 * reassemble it into whole datagram. If a chain for 743 * reassembly of this datagram already exists, then it 744 * is given as fp; otherwise have to make a chain. 745 */ 746 struct mbuf * 747 ip_reass(ipqe, fp) 748 struct ipqent *ipqe; 749 struct ipq *fp; 750 { 751 struct mbuf *m = ipqe->ipqe_m; 752 struct ipqent *nq, *p, *q; 753 struct ip *ip; 754 struct mbuf *t; 755 int hlen = ipqe->ipqe_ip->ip_hl << 2; 756 int i, next; 757 758 IPQ_LOCK_CHECK(); 759 760 /* 761 * Presence of header sizes in mbufs 762 * would confuse code below. 763 */ 764 m->m_data += hlen; 765 m->m_len -= hlen; 766 767 /* 768 * If first fragment to arrive, create a reassembly queue. 769 */ 770 if (fp == 0) { 771 /* 772 * Enforce upper bound on number of fragmented packets 773 * for which we attempt reassembly; 774 * If maxfrag is 0, never accept fragments. 775 * If maxfrag is -1, accept all fragments without limitation. 776 */ 777 if (ip_maxfragpackets < 0) 778 ; 779 else if (ip_nfragpackets >= ip_maxfragpackets) 780 goto dropfrag; 781 ip_nfragpackets++; 782 MALLOC(fp, struct ipq *, sizeof (struct ipq), 783 M_FTABLE, M_NOWAIT); 784 if (fp == NULL) 785 goto dropfrag; 786 LIST_INSERT_HEAD(&ipq, fp, ipq_q); 787 fp->ipq_ttl = IPFRAGTTL; 788 fp->ipq_p = ipqe->ipqe_ip->ip_p; 789 fp->ipq_id = ipqe->ipqe_ip->ip_id; 790 LIST_INIT(&fp->ipq_fragq); 791 fp->ipq_src = ipqe->ipqe_ip->ip_src; 792 fp->ipq_dst = ipqe->ipqe_ip->ip_dst; 793 p = NULL; 794 goto insert; 795 } 796 797 /* 798 * Find a segment which begins after this one does. 799 */ 800 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 801 p = q, q = q->ipqe_q.le_next) 802 if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) 803 break; 804 805 /* 806 * If there is a preceding segment, it may provide some of 807 * our data already. If so, drop the data from the incoming 808 * segment. If it provides all of our data, drop us. 809 */ 810 if (p != NULL) { 811 i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - 812 ipqe->ipqe_ip->ip_off; 813 if (i > 0) { 814 if (i >= ipqe->ipqe_ip->ip_len) 815 goto dropfrag; 816 m_adj(ipqe->ipqe_m, i); 817 ipqe->ipqe_ip->ip_off += i; 818 ipqe->ipqe_ip->ip_len -= i; 819 } 820 } 821 822 /* 823 * While we overlap succeeding segments trim them or, 824 * if they are completely covered, dequeue them. 825 */ 826 for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > 827 q->ipqe_ip->ip_off; q = nq) { 828 i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - 829 q->ipqe_ip->ip_off; 830 if (i < q->ipqe_ip->ip_len) { 831 q->ipqe_ip->ip_len -= i; 832 q->ipqe_ip->ip_off += i; 833 m_adj(q->ipqe_m, i); 834 break; 835 } 836 nq = q->ipqe_q.le_next; 837 m_freem(q->ipqe_m); 838 LIST_REMOVE(q, ipqe_q); 839 pool_put(&ipqent_pool, q); 840 } 841 842 insert: 843 /* 844 * Stick new segment in its place; 845 * check for complete reassembly. 846 */ 847 if (p == NULL) { 848 LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); 849 } else { 850 LIST_INSERT_AFTER(p, ipqe, ipqe_q); 851 } 852 next = 0; 853 for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; 854 p = q, q = q->ipqe_q.le_next) { 855 if (q->ipqe_ip->ip_off != next) 856 return (0); 857 next += q->ipqe_ip->ip_len; 858 } 859 if (p->ipqe_mff) 860 return (0); 861 862 /* 863 * Reassembly is complete. Check for a bogus message size and 864 * concatenate fragments. 865 */ 866 q = fp->ipq_fragq.lh_first; 867 ip = q->ipqe_ip; 868 if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { 869 ipstat.ips_toolong++; 870 ip_freef(fp); 871 return (0); 872 } 873 m = q->ipqe_m; 874 t = m->m_next; 875 m->m_next = 0; 876 m_cat(m, t); 877 nq = q->ipqe_q.le_next; 878 pool_put(&ipqent_pool, q); 879 for (q = nq; q != NULL; q = nq) { 880 t = q->ipqe_m; 881 nq = q->ipqe_q.le_next; 882 pool_put(&ipqent_pool, q); 883 m_cat(m, t); 884 } 885 886 /* 887 * Create header for new ip packet by 888 * modifying header of first packet; 889 * dequeue and discard fragment reassembly header. 890 * Make header visible. 891 */ 892 ip->ip_len = next; 893 ip->ip_src = fp->ipq_src; 894 ip->ip_dst = fp->ipq_dst; 895 LIST_REMOVE(fp, ipq_q); 896 FREE(fp, M_FTABLE); 897 ip_nfragpackets--; 898 m->m_len += (ip->ip_hl << 2); 899 m->m_data -= (ip->ip_hl << 2); 900 /* some debugging cruft by sklower, below, will go away soon */ 901 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ 902 int plen = 0; 903 for (t = m; t; t = t->m_next) 904 plen += t->m_len; 905 m->m_pkthdr.len = plen; 906 } 907 return (m); 908 909 dropfrag: 910 ipstat.ips_fragdropped++; 911 m_freem(m); 912 pool_put(&ipqent_pool, ipqe); 913 return (0); 914 } 915 916 /* 917 * Free a fragment reassembly header and all 918 * associated datagrams. 919 */ 920 void 921 ip_freef(fp) 922 struct ipq *fp; 923 { 924 struct ipqent *q, *p; 925 926 IPQ_LOCK_CHECK(); 927 928 for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { 929 p = q->ipqe_q.le_next; 930 m_freem(q->ipqe_m); 931 LIST_REMOVE(q, ipqe_q); 932 pool_put(&ipqent_pool, q); 933 } 934 LIST_REMOVE(fp, ipq_q); 935 FREE(fp, M_FTABLE); 936 ip_nfragpackets--; 937 } 938 939 /* 940 * IP timer processing; 941 * if a timer expires on a reassembly 942 * queue, discard it. 943 */ 944 void 945 ip_slowtimo() 946 { 947 struct ipq *fp, *nfp; 948 int s = splsoftnet(); 949 950 IPQ_LOCK(); 951 for (fp = ipq.lh_first; fp != NULL; fp = nfp) { 952 nfp = fp->ipq_q.le_next; 953 if (--fp->ipq_ttl == 0) { 954 ipstat.ips_fragtimeout++; 955 ip_freef(fp); 956 } 957 } 958 /* 959 * If we are over the maximum number of fragments 960 * (due to the limit being lowered), drain off 961 * enough to get down to the new limit. 962 */ 963 if (ip_maxfragpackets < 0) 964 ; 965 else { 966 while (ip_nfragpackets > ip_maxfragpackets && ipq.lh_first) 967 ip_freef(ipq.lh_first); 968 } 969 IPQ_UNLOCK(); 970 #ifdef GATEWAY 971 ipflow_slowtimo(); 972 #endif 973 splx(s); 974 } 975 976 /* 977 * Drain off all datagram fragments. 978 */ 979 void 980 ip_drain() 981 { 982 983 /* 984 * We may be called from a device's interrupt context. If 985 * the ipq is already busy, just bail out now. 986 */ 987 if (ipq_lock_try() == 0) 988 return; 989 990 while (ipq.lh_first != NULL) { 991 ipstat.ips_fragdropped++; 992 ip_freef(ipq.lh_first); 993 } 994 995 IPQ_UNLOCK(); 996 } 997 998 /* 999 * Do option processing on a datagram, 1000 * possibly discarding it if bad options are encountered, 1001 * or forwarding it if source-routed. 1002 * Returns 1 if packet has been forwarded/freed, 1003 * 0 if the packet should be processed further. 1004 */ 1005 int 1006 ip_dooptions(m) 1007 struct mbuf *m; 1008 { 1009 struct ip *ip = mtod(m, struct ip *); 1010 u_char *cp, *cp0; 1011 struct ip_timestamp *ipt; 1012 struct in_ifaddr *ia; 1013 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0; 1014 struct in_addr dst; 1015 n_time ntime; 1016 1017 dst = ip->ip_dst; 1018 cp = (u_char *)(ip + 1); 1019 cnt = (ip->ip_hl << 2) - sizeof (struct ip); 1020 for (; cnt > 0; cnt -= optlen, cp += optlen) { 1021 opt = cp[IPOPT_OPTVAL]; 1022 if (opt == IPOPT_EOL) 1023 break; 1024 if (opt == IPOPT_NOP) 1025 optlen = 1; 1026 else { 1027 if (cnt < IPOPT_OLEN + sizeof(*cp)) { 1028 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1029 goto bad; 1030 } 1031 optlen = cp[IPOPT_OLEN]; 1032 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { 1033 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1034 goto bad; 1035 } 1036 } 1037 switch (opt) { 1038 1039 default: 1040 break; 1041 1042 /* 1043 * Source routing with record. 1044 * Find interface with current destination address. 1045 * If none on this machine then drop if strictly routed, 1046 * or do nothing if loosely routed. 1047 * Record interface address and bring up next address 1048 * component. If strictly routed make sure next 1049 * address is on directly accessible net. 1050 */ 1051 case IPOPT_LSRR: 1052 case IPOPT_SSRR: 1053 if (ip_allowsrcrt == 0) { 1054 type = ICMP_UNREACH; 1055 code = ICMP_UNREACH_NET_PROHIB; 1056 goto bad; 1057 } 1058 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1059 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1060 goto bad; 1061 } 1062 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1063 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1064 goto bad; 1065 } 1066 ipaddr.sin_addr = ip->ip_dst; 1067 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1068 if (ia == 0) { 1069 if (opt == IPOPT_SSRR) { 1070 type = ICMP_UNREACH; 1071 code = ICMP_UNREACH_SRCFAIL; 1072 goto bad; 1073 } 1074 /* 1075 * Loose routing, and not at next destination 1076 * yet; nothing to do except forward. 1077 */ 1078 break; 1079 } 1080 off--; /* 0 origin */ 1081 if ((off + sizeof(struct in_addr)) > optlen) { 1082 /* 1083 * End of source route. Should be for us. 1084 */ 1085 save_rte(cp, ip->ip_src); 1086 break; 1087 } 1088 /* 1089 * locate outgoing interface 1090 */ 1091 bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, 1092 sizeof(ipaddr.sin_addr)); 1093 if (opt == IPOPT_SSRR) 1094 ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); 1095 else 1096 ia = ip_rtaddr(ipaddr.sin_addr); 1097 if (ia == 0) { 1098 type = ICMP_UNREACH; 1099 code = ICMP_UNREACH_SRCFAIL; 1100 goto bad; 1101 } 1102 ip->ip_dst = ipaddr.sin_addr; 1103 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1104 (caddr_t)(cp + off), sizeof(struct in_addr)); 1105 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1106 /* 1107 * Let ip_intr's mcast routing check handle mcast pkts 1108 */ 1109 forward = !IN_MULTICAST(ip->ip_dst.s_addr); 1110 break; 1111 1112 case IPOPT_RR: 1113 if (optlen < IPOPT_OFFSET + sizeof(*cp)) { 1114 code = &cp[IPOPT_OLEN] - (u_char *)ip; 1115 goto bad; 1116 } 1117 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) { 1118 code = &cp[IPOPT_OFFSET] - (u_char *)ip; 1119 goto bad; 1120 } 1121 /* 1122 * If no space remains, ignore. 1123 */ 1124 off--; /* 0 origin */ 1125 if ((off + sizeof(struct in_addr)) > optlen) 1126 break; 1127 bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, 1128 sizeof(ipaddr.sin_addr)); 1129 /* 1130 * locate outgoing interface; if we're the destination, 1131 * use the incoming interface (should be same). 1132 */ 1133 if ((ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr)))) 1134 == NULL && 1135 (ia = ip_rtaddr(ipaddr.sin_addr)) == NULL) { 1136 type = ICMP_UNREACH; 1137 code = ICMP_UNREACH_HOST; 1138 goto bad; 1139 } 1140 bcopy((caddr_t)&ia->ia_addr.sin_addr, 1141 (caddr_t)(cp + off), sizeof(struct in_addr)); 1142 cp[IPOPT_OFFSET] += sizeof(struct in_addr); 1143 break; 1144 1145 case IPOPT_TS: 1146 code = cp - (u_char *)ip; 1147 ipt = (struct ip_timestamp *)cp; 1148 if (ipt->ipt_len < 4 || ipt->ipt_len > 40) { 1149 code = (u_char *)&ipt->ipt_len - (u_char *)ip; 1150 goto bad; 1151 } 1152 if (ipt->ipt_ptr < 5) { 1153 code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; 1154 goto bad; 1155 } 1156 if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) { 1157 if (++ipt->ipt_oflw == 0) { 1158 code = (u_char *)&ipt->ipt_ptr - 1159 (u_char *)ip; 1160 goto bad; 1161 } 1162 break; 1163 } 1164 cp0 = (cp + ipt->ipt_ptr - 1); 1165 switch (ipt->ipt_flg) { 1166 1167 case IPOPT_TS_TSONLY: 1168 break; 1169 1170 case IPOPT_TS_TSANDADDR: 1171 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1172 sizeof(struct in_addr) > ipt->ipt_len) { 1173 code = (u_char *)&ipt->ipt_ptr - 1174 (u_char *)ip; 1175 goto bad; 1176 } 1177 ipaddr.sin_addr = dst; 1178 ia = ifatoia(ifaof_ifpforaddr(sintosa(&ipaddr), 1179 m->m_pkthdr.rcvif)); 1180 if (ia == 0) 1181 continue; 1182 bcopy(&ia->ia_addr.sin_addr, 1183 cp0, sizeof(struct in_addr)); 1184 ipt->ipt_ptr += sizeof(struct in_addr); 1185 break; 1186 1187 case IPOPT_TS_PRESPEC: 1188 if (ipt->ipt_ptr - 1 + sizeof(n_time) + 1189 sizeof(struct in_addr) > ipt->ipt_len) { 1190 code = (u_char *)&ipt->ipt_ptr - 1191 (u_char *)ip; 1192 goto bad; 1193 } 1194 bcopy(cp0, &ipaddr.sin_addr, 1195 sizeof(struct in_addr)); 1196 if (ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))) 1197 == NULL) 1198 continue; 1199 ipt->ipt_ptr += sizeof(struct in_addr); 1200 break; 1201 1202 default: 1203 /* XXX can't take &ipt->ipt_flg */ 1204 code = (u_char *)&ipt->ipt_ptr - 1205 (u_char *)ip + 1; 1206 goto bad; 1207 } 1208 ntime = iptime(); 1209 cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ 1210 bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1, 1211 sizeof(n_time)); 1212 ipt->ipt_ptr += sizeof(n_time); 1213 } 1214 } 1215 if (forward) { 1216 if (ip_forwsrcrt == 0) { 1217 type = ICMP_UNREACH; 1218 code = ICMP_UNREACH_SRCFAIL; 1219 goto bad; 1220 } 1221 ip_forward(m, 1); 1222 return (1); 1223 } 1224 return (0); 1225 bad: 1226 icmp_error(m, type, code, 0, 0); 1227 ipstat.ips_badoptions++; 1228 return (1); 1229 } 1230 1231 /* 1232 * Given address of next destination (final or next hop), 1233 * return internet address info of interface to be used to get there. 1234 */ 1235 struct in_ifaddr * 1236 ip_rtaddr(dst) 1237 struct in_addr dst; 1238 { 1239 struct sockaddr_in *sin; 1240 1241 sin = satosin(&ipforward_rt.ro_dst); 1242 1243 if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { 1244 if (ipforward_rt.ro_rt) { 1245 RTFREE(ipforward_rt.ro_rt); 1246 ipforward_rt.ro_rt = 0; 1247 } 1248 sin->sin_family = AF_INET; 1249 sin->sin_len = sizeof(*sin); 1250 sin->sin_addr = dst; 1251 1252 rtalloc(&ipforward_rt); 1253 } 1254 if (ipforward_rt.ro_rt == 0) 1255 return ((struct in_ifaddr *)0); 1256 return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); 1257 } 1258 1259 /* 1260 * Save incoming source route for use in replies, 1261 * to be picked up later by ip_srcroute if the receiver is interested. 1262 */ 1263 void 1264 save_rte(option, dst) 1265 u_char *option; 1266 struct in_addr dst; 1267 { 1268 unsigned olen; 1269 1270 olen = option[IPOPT_OLEN]; 1271 #ifdef DIAGNOSTIC 1272 if (ipprintfs) 1273 printf("save_rte: olen %d\n", olen); 1274 #endif /* 0 */ 1275 if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) 1276 return; 1277 bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); 1278 ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); 1279 ip_srcrt.dst = dst; 1280 } 1281 1282 /* 1283 * Retrieve incoming source route for use in replies, 1284 * in the same form used by setsockopt. 1285 * The first hop is placed before the options, will be removed later. 1286 */ 1287 struct mbuf * 1288 ip_srcroute() 1289 { 1290 struct in_addr *p, *q; 1291 struct mbuf *m; 1292 1293 if (ip_nhops == 0) 1294 return ((struct mbuf *)0); 1295 m = m_get(M_DONTWAIT, MT_SOOPTS); 1296 if (m == 0) 1297 return ((struct mbuf *)0); 1298 1299 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) 1300 1301 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ 1302 m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) + 1303 OPTSIZ; 1304 #ifdef DIAGNOSTIC 1305 if (ipprintfs) 1306 printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); 1307 #endif 1308 1309 /* 1310 * First save first hop for return route 1311 */ 1312 p = &ip_srcrt.route[ip_nhops - 1]; 1313 *(mtod(m, struct in_addr *)) = *p--; 1314 #ifdef DIAGNOSTIC 1315 if (ipprintfs) 1316 printf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr)); 1317 #endif 1318 1319 /* 1320 * Copy option fields and padding (nop) to mbuf. 1321 */ 1322 ip_srcrt.nop = IPOPT_NOP; 1323 ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; 1324 bcopy((caddr_t)&ip_srcrt.nop, 1325 mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); 1326 q = (struct in_addr *)(mtod(m, caddr_t) + 1327 sizeof(struct in_addr) + OPTSIZ); 1328 #undef OPTSIZ 1329 /* 1330 * Record return path as an IP source route, 1331 * reversing the path (pointers are now aligned). 1332 */ 1333 while (p >= ip_srcrt.route) { 1334 #ifdef DIAGNOSTIC 1335 if (ipprintfs) 1336 printf(" %x", ntohl(q->s_addr)); 1337 #endif 1338 *q++ = *p--; 1339 } 1340 /* 1341 * Last hop goes to final destination. 1342 */ 1343 *q = ip_srcrt.dst; 1344 #ifdef DIAGNOSTIC 1345 if (ipprintfs) 1346 printf(" %x\n", ntohl(q->s_addr)); 1347 #endif 1348 return (m); 1349 } 1350 1351 /* 1352 * Strip out IP options, at higher 1353 * level protocol in the kernel. 1354 * Second argument is buffer to which options 1355 * will be moved, and return value is their length. 1356 * XXX should be deleted; last arg currently ignored. 1357 */ 1358 void 1359 ip_stripoptions(m, mopt) 1360 struct mbuf *m; 1361 struct mbuf *mopt; 1362 { 1363 int i; 1364 struct ip *ip = mtod(m, struct ip *); 1365 caddr_t opts; 1366 int olen; 1367 1368 olen = (ip->ip_hl << 2) - sizeof (struct ip); 1369 opts = (caddr_t)(ip + 1); 1370 i = m->m_len - (sizeof (struct ip) + olen); 1371 bcopy(opts + olen, opts, (unsigned)i); 1372 m->m_len -= olen; 1373 if (m->m_flags & M_PKTHDR) 1374 m->m_pkthdr.len -= olen; 1375 ip->ip_len -= olen; 1376 ip->ip_hl = sizeof (struct ip) >> 2; 1377 } 1378 1379 int inetctlerrmap[PRC_NCMDS] = { 1380 0, 0, 0, 0, 1381 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, 1382 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, 1383 EMSGSIZE, EHOSTUNREACH, 0, 0, 1384 0, 0, 0, 0, 1385 ENOPROTOOPT 1386 }; 1387 1388 /* 1389 * Forward a packet. If some error occurs return the sender 1390 * an icmp packet. Note we can't always generate a meaningful 1391 * icmp message because icmp doesn't have a large enough repertoire 1392 * of codes and types. 1393 * 1394 * If not forwarding, just drop the packet. This could be confusing 1395 * if ipforwarding was zero but some routing protocol was advancing 1396 * us as a gateway to somewhere. However, we must let the routing 1397 * protocol deal with that. 1398 * 1399 * The srcrt parameter indicates whether the packet is being forwarded 1400 * via a source route. 1401 */ 1402 void 1403 ip_forward(m, srcrt) 1404 struct mbuf *m; 1405 int srcrt; 1406 { 1407 struct ip *ip = mtod(m, struct ip *); 1408 struct sockaddr_in *sin; 1409 struct rtentry *rt; 1410 int error, type = 0, code = 0; 1411 struct mbuf *mcopy; 1412 n_long dest; 1413 struct ifnet *destifp; 1414 #ifdef IPSEC 1415 struct ifnet dummyifp; 1416 #endif 1417 1418 dest = 0; 1419 #ifdef DIAGNOSTIC 1420 if (ipprintfs) 1421 printf("forward: src %2.2x dst %2.2x ttl %x\n", 1422 ntohl(ip->ip_src.s_addr), 1423 ntohl(ip->ip_dst.s_addr), ip->ip_ttl); 1424 #endif 1425 if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { 1426 ipstat.ips_cantforward++; 1427 m_freem(m); 1428 return; 1429 } 1430 if (ip->ip_ttl <= IPTTLDEC) { 1431 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); 1432 return; 1433 } 1434 ip->ip_ttl -= IPTTLDEC; 1435 1436 sin = satosin(&ipforward_rt.ro_dst); 1437 if ((rt = ipforward_rt.ro_rt) == 0 || 1438 !in_hosteq(ip->ip_dst, sin->sin_addr)) { 1439 if (ipforward_rt.ro_rt) { 1440 RTFREE(ipforward_rt.ro_rt); 1441 ipforward_rt.ro_rt = 0; 1442 } 1443 sin->sin_family = AF_INET; 1444 sin->sin_len = sizeof(struct sockaddr_in); 1445 sin->sin_addr = ip->ip_dst; 1446 1447 rtalloc(&ipforward_rt); 1448 if (ipforward_rt.ro_rt == 0) { 1449 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); 1450 return; 1451 } 1452 rt = ipforward_rt.ro_rt; 1453 } 1454 1455 /* 1456 * Save at most 68 bytes of the packet in case 1457 * we need to generate an ICMP message to the src. 1458 * Pullup to avoid sharing mbuf cluster between m and mcopy. 1459 */ 1460 mcopy = m_copym(m, 0, imin((int)ip->ip_len, 68), M_DONTWAIT); 1461 if (mcopy) 1462 mcopy = m_pullup(mcopy, ip->ip_hl << 2); 1463 1464 /* 1465 * If forwarding packet using same interface that it came in on, 1466 * perhaps should send a redirect to sender to shortcut a hop. 1467 * Only send redirect if source is sending directly to us, 1468 * and if packet was not source routed (or has any options). 1469 * Also, don't send redirect if forwarding using a default route 1470 * or a route modified by a redirect. 1471 */ 1472 if (rt->rt_ifp == m->m_pkthdr.rcvif && 1473 (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && 1474 !in_nullhost(satosin(rt_key(rt))->sin_addr) && 1475 ipsendredirects && !srcrt) { 1476 if (rt->rt_ifa && 1477 (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == 1478 ifatoia(rt->rt_ifa)->ia_subnet) { 1479 if (rt->rt_flags & RTF_GATEWAY) 1480 dest = satosin(rt->rt_gateway)->sin_addr.s_addr; 1481 else 1482 dest = ip->ip_dst.s_addr; 1483 /* 1484 * Router requirements says to only send host 1485 * redirects. 1486 */ 1487 type = ICMP_REDIRECT; 1488 code = ICMP_REDIRECT_HOST; 1489 #ifdef DIAGNOSTIC 1490 if (ipprintfs) 1491 printf("redirect (%d) to %x\n", code, 1492 (u_int32_t)dest); 1493 #endif 1494 } 1495 } 1496 1497 #ifdef IPSEC 1498 /* Don't lookup socket in forwading case */ 1499 (void)ipsec_setsocket(m, NULL); 1500 #endif 1501 error = ip_output(m, (struct mbuf *)0, &ipforward_rt, 1502 (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); 1503 if (error) 1504 ipstat.ips_cantforward++; 1505 else { 1506 ipstat.ips_forward++; 1507 if (type) 1508 ipstat.ips_redirectsent++; 1509 else { 1510 if (mcopy) { 1511 #ifdef GATEWAY 1512 if (mcopy->m_flags & M_CANFASTFWD) 1513 ipflow_create(&ipforward_rt, mcopy); 1514 #endif 1515 m_freem(mcopy); 1516 } 1517 return; 1518 } 1519 } 1520 if (mcopy == NULL) 1521 return; 1522 destifp = NULL; 1523 1524 switch (error) { 1525 1526 case 0: /* forwarded, but need redirect */ 1527 /* type, code set above */ 1528 break; 1529 1530 case ENETUNREACH: /* shouldn't happen, checked above */ 1531 case EHOSTUNREACH: 1532 case ENETDOWN: 1533 case EHOSTDOWN: 1534 default: 1535 type = ICMP_UNREACH; 1536 code = ICMP_UNREACH_HOST; 1537 break; 1538 1539 case EMSGSIZE: 1540 type = ICMP_UNREACH; 1541 code = ICMP_UNREACH_NEEDFRAG; 1542 #ifndef IPSEC 1543 if (ipforward_rt.ro_rt) 1544 destifp = ipforward_rt.ro_rt->rt_ifp; 1545 #else 1546 /* 1547 * If the packet is routed over IPsec tunnel, tell the 1548 * originator the tunnel MTU. 1549 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz 1550 * XXX quickhack!!! 1551 */ 1552 if (ipforward_rt.ro_rt) { 1553 struct secpolicy *sp; 1554 int ipsecerror; 1555 size_t ipsechdr; 1556 struct route *ro; 1557 1558 sp = ipsec4_getpolicybyaddr(mcopy, 1559 IPSEC_DIR_OUTBOUND, 1560 IP_FORWARDING, 1561 &ipsecerror); 1562 1563 if (sp == NULL) 1564 destifp = ipforward_rt.ro_rt->rt_ifp; 1565 else { 1566 /* count IPsec header size */ 1567 ipsechdr = ipsec4_hdrsiz(mcopy, 1568 IPSEC_DIR_OUTBOUND, 1569 NULL); 1570 1571 /* 1572 * find the correct route for outer IPv4 1573 * header, compute tunnel MTU. 1574 * 1575 * XXX BUG ALERT 1576 * The "dummyifp" code relies upon the fact 1577 * that icmp_error() touches only ifp->if_mtu. 1578 */ 1579 /*XXX*/ 1580 destifp = NULL; 1581 if (sp->req != NULL 1582 && sp->req->sav != NULL 1583 && sp->req->sav->sah != NULL) { 1584 ro = &sp->req->sav->sah->sa_route; 1585 if (ro->ro_rt && ro->ro_rt->rt_ifp) { 1586 dummyifp.if_mtu = 1587 ro->ro_rt->rt_ifp->if_mtu; 1588 dummyifp.if_mtu -= ipsechdr; 1589 destifp = &dummyifp; 1590 } 1591 } 1592 1593 key_freesp(sp); 1594 } 1595 } 1596 #endif /*IPSEC*/ 1597 ipstat.ips_cantfrag++; 1598 break; 1599 1600 case ENOBUFS: 1601 type = ICMP_SOURCEQUENCH; 1602 code = 0; 1603 break; 1604 } 1605 icmp_error(mcopy, type, code, dest, destifp); 1606 } 1607 1608 void 1609 ip_savecontrol(inp, mp, ip, m) 1610 struct inpcb *inp; 1611 struct mbuf **mp; 1612 struct ip *ip; 1613 struct mbuf *m; 1614 { 1615 1616 if (inp->inp_socket->so_options & SO_TIMESTAMP) { 1617 struct timeval tv; 1618 1619 microtime(&tv); 1620 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), 1621 SCM_TIMESTAMP, SOL_SOCKET); 1622 if (*mp) 1623 mp = &(*mp)->m_next; 1624 } 1625 if (inp->inp_flags & INP_RECVDSTADDR) { 1626 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, 1627 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); 1628 if (*mp) 1629 mp = &(*mp)->m_next; 1630 } 1631 #ifdef notyet 1632 /* 1633 * XXX 1634 * Moving these out of udp_input() made them even more broken 1635 * than they already were. 1636 * - fenner (at) parc.xerox.com 1637 */ 1638 /* options were tossed already */ 1639 if (inp->inp_flags & INP_RECVOPTS) { 1640 *mp = sbcreatecontrol((caddr_t) opts_deleted_above, 1641 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); 1642 if (*mp) 1643 mp = &(*mp)->m_next; 1644 } 1645 /* ip_srcroute doesn't do what we want here, need to fix */ 1646 if (inp->inp_flags & INP_RECVRETOPTS) { 1647 *mp = sbcreatecontrol((caddr_t) ip_srcroute(), 1648 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); 1649 if (*mp) 1650 mp = &(*mp)->m_next; 1651 } 1652 #endif 1653 if (inp->inp_flags & INP_RECVIF) { 1654 struct sockaddr_dl sdl; 1655 1656 sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); 1657 sdl.sdl_family = AF_LINK; 1658 sdl.sdl_index = m->m_pkthdr.rcvif ? 1659 m->m_pkthdr.rcvif->if_index : 0; 1660 sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; 1661 *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, 1662 IP_RECVIF, IPPROTO_IP); 1663 if (*mp) 1664 mp = &(*mp)->m_next; 1665 } 1666 } 1667 1668 int 1669 ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 1670 int *name; 1671 u_int namelen; 1672 void *oldp; 1673 size_t *oldlenp; 1674 void *newp; 1675 size_t newlen; 1676 { 1677 extern int subnetsarelocal, hostzeroisbroadcast; 1678 1679 int error, old; 1680 1681 /* All sysctl names at this level are terminal. */ 1682 if (namelen != 1) 1683 return (ENOTDIR); 1684 1685 switch (name[0]) { 1686 case IPCTL_FORWARDING: 1687 return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); 1688 case IPCTL_SENDREDIRECTS: 1689 return (sysctl_int(oldp, oldlenp, newp, newlen, 1690 &ipsendredirects)); 1691 case IPCTL_DEFTTL: 1692 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); 1693 #ifdef notyet 1694 case IPCTL_DEFMTU: 1695 return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); 1696 #endif 1697 case IPCTL_FORWSRCRT: 1698 /* Don't allow this to change in a secure environment. */ 1699 if (securelevel > 0) 1700 return (sysctl_rdint(oldp, oldlenp, newp, 1701 ip_forwsrcrt)); 1702 else 1703 return (sysctl_int(oldp, oldlenp, newp, newlen, 1704 &ip_forwsrcrt)); 1705 case IPCTL_DIRECTEDBCAST: 1706 return (sysctl_int(oldp, oldlenp, newp, newlen, 1707 &ip_directedbcast)); 1708 case IPCTL_ALLOWSRCRT: 1709 return (sysctl_int(oldp, oldlenp, newp, newlen, 1710 &ip_allowsrcrt)); 1711 case IPCTL_SUBNETSARELOCAL: 1712 return (sysctl_int(oldp, oldlenp, newp, newlen, 1713 &subnetsarelocal)); 1714 case IPCTL_MTUDISC: 1715 error = sysctl_int(oldp, oldlenp, newp, newlen, 1716 &ip_mtudisc); 1717 if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) { 1718 ip_mtudisc_timeout_q = 1719 rt_timer_queue_create(ip_mtudisc_timeout); 1720 } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) { 1721 rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE); 1722 ip_mtudisc_timeout_q = NULL; 1723 } 1724 return error; 1725 case IPCTL_ANONPORTMIN: 1726 old = anonportmin; 1727 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); 1728 if (anonportmin >= anonportmax || anonportmin < 0 1729 || anonportmin > 65535 1730 #ifndef IPNOPRIVPORTS 1731 || anonportmin < IPPORT_RESERVED 1732 #endif 1733 ) { 1734 anonportmin = old; 1735 return (EINVAL); 1736 } 1737 return (error); 1738 case IPCTL_ANONPORTMAX: 1739 old = anonportmax; 1740 error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); 1741 if (anonportmin >= anonportmax || anonportmax < 0 1742 || anonportmax > 65535 1743 #ifndef IPNOPRIVPORTS 1744 || anonportmax < IPPORT_RESERVED 1745 #endif 1746 ) { 1747 anonportmax = old; 1748 return (EINVAL); 1749 } 1750 return (error); 1751 case IPCTL_MTUDISCTIMEOUT: 1752 error = sysctl_int(oldp, oldlenp, newp, newlen, 1753 &ip_mtudisc_timeout); 1754 if (ip_mtudisc_timeout_q != NULL) 1755 rt_timer_queue_change(ip_mtudisc_timeout_q, 1756 ip_mtudisc_timeout); 1757 return (error); 1758 #ifdef GATEWAY 1759 case IPCTL_MAXFLOWS: 1760 { 1761 int s; 1762 1763 error = sysctl_int(oldp, oldlenp, newp, newlen, 1764 &ip_maxflows); 1765 s = splsoftnet(); 1766 ipflow_reap(0); 1767 splx(s); 1768 return (error); 1769 } 1770 #endif 1771 case IPCTL_HOSTZEROBROADCAST: 1772 return (sysctl_int(oldp, oldlenp, newp, newlen, 1773 &hostzeroisbroadcast)); 1774 #if NGIF > 0 1775 case IPCTL_GIF_TTL: 1776 return(sysctl_int(oldp, oldlenp, newp, newlen, 1777 &ip_gif_ttl)); 1778 #endif 1779 1780 #if NGRE > 0 1781 case IPCTL_GRE_TTL: 1782 return(sysctl_int(oldp, oldlenp, newp, newlen, 1783 &ip_gre_ttl)); 1784 #endif 1785 1786 #ifndef IPNOPRIVPORTS 1787 case IPCTL_LOWPORTMIN: 1788 old = lowportmin; 1789 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin); 1790 if (lowportmin >= lowportmax 1791 || lowportmin > IPPORT_RESERVEDMAX 1792 || lowportmin < IPPORT_RESERVEDMIN 1793 ) { 1794 lowportmin = old; 1795 return (EINVAL); 1796 } 1797 return (error); 1798 case IPCTL_LOWPORTMAX: 1799 old = lowportmax; 1800 error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax); 1801 if (lowportmin >= lowportmax 1802 || lowportmax > IPPORT_RESERVEDMAX 1803 || lowportmax < IPPORT_RESERVEDMIN 1804 ) { 1805 lowportmax = old; 1806 return (EINVAL); 1807 } 1808 return (error); 1809 #endif 1810 1811 case IPCTL_MAXFRAGPACKETS: 1812 return (sysctl_int(oldp, oldlenp, newp, newlen, 1813 &ip_maxfragpackets)); 1814 1815 default: 1816 return (EOPNOTSUPP); 1817 } 1818 /* NOTREACHED */ 1819 } 1820
Indexes created Wed Oct 15 16:09:53 GMT 2025